paperID
stringlengths 36
36
| pwc_id
stringlengths 8
47
| arxiv_id
stringlengths 6
16
⌀ | nips_id
float64 | url_abs
stringlengths 18
329
| url_pdf
stringlengths 18
742
| title
stringlengths 8
325
| abstract
stringlengths 1
7.27k
⌀ | authors
stringlengths 2
7.06k
| published
stringlengths 10
10
⌀ | conference
stringlengths 12
47
⌀ | conference_url_abs
stringlengths 16
198
⌀ | conference_url_pdf
stringlengths 27
199
⌀ | proceeding
stringlengths 6
47
⌀ | taskID
stringlengths 7
1.44k
| areaID
stringclasses 688
values | embedding
stringlengths 9.26k
12.5k
| umap_embedding
stringlengths 29
44
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
9ed33008-50ec-413d-bca3-e2469d702595
|
deep-learning-representation-using
|
1409.7164
| null |
http://arxiv.org/abs/1409.7164v1
|
http://arxiv.org/pdf/1409.7164v1.pdf
|
Deep Learning Representation using Autoencoder for 3D Shape Retrieval
|
We study the problem of how to build a deep learning representation for 3D
shape. Deep learning has shown to be very effective in variety of visual
applications, such as image classification and object detection. However, it
has not been successfully applied to 3D shape recognition. This is because 3D
shape has complex structure in 3D space and there are limited number of 3D
shapes for feature learning. To address these problems, we project 3D shapes
into 2D space and use autoencoder for feature learning on the 2D images. High
accuracy 3D shape retrieval performance is obtained by aggregating the features
learned on 2D images. In addition, we show the proposed deep learning feature
is complementary to conventional local image descriptors. By combing the global
deep learning representation and the local descriptor representation, our
method can obtain the state-of-the-art performance on 3D shape retrieval
benchmarks.
|
['Song Bai', 'Zhuotun Zhu', 'Xinggang Wang', 'Cong Yao', 'Xiang Bai']
|
2014-09-25
| null | null | null | null |
['3d-shape-retrieval', '3d-shape-recognition']
|
['computer-vision', 'computer-vision']
|
[-4.65285420e-01 -7.22081721e-01 -1.38298824e-01 -4.00615841e-01
-5.98659515e-01 -5.19092023e-01 8.44276130e-01 7.28919953e-02
-1.24752954e-01 -9.63500664e-02 1.88638091e-01 -1.60679609e-01
-3.68946940e-01 -1.07227337e+00 -4.24140006e-01 -7.71175683e-01
1.98411699e-02 7.19120681e-01 4.85014319e-02 2.11172774e-02
2.59203196e-01 1.35076261e+00 -1.32942879e+00 8.62257108e-02
-3.32052931e-02 1.38061976e+00 -9.06569883e-02 2.34946042e-01
-5.24168968e-01 -1.81524921e-02 -5.59125960e-01 9.90264565e-02
6.06928647e-01 1.03493221e-01 -6.52311563e-01 1.00039817e-01
5.61701715e-01 -6.61208630e-01 -6.84091866e-01 8.01335812e-01
9.94287133e-01 3.83924786e-03 1.20852876e+00 -1.06928420e+00
-1.31561673e+00 -3.29268366e-01 -5.19670367e-01 8.14737007e-02
2.63818115e-01 -2.18627319e-01 1.02717078e+00 -1.38033319e+00
5.20637155e-01 1.63772416e+00 3.93347800e-01 2.27226600e-01
-8.21310699e-01 -4.35061574e-01 -2.65483677e-01 2.58466471e-02
-1.58600497e+00 -5.85121512e-02 1.11368954e+00 -4.01813567e-01
1.49299002e+00 -1.25743106e-01 7.71836579e-01 6.54933333e-01
2.54439205e-01 1.08475089e+00 6.93123221e-01 -1.94374889e-01
-8.91323388e-02 -3.93063694e-01 -1.64037392e-01 7.53681421e-01
1.58584535e-01 2.91236490e-01 2.98337382e-03 -1.66614383e-01
1.22522163e+00 4.92252737e-01 2.27221176e-01 -7.30795085e-01
-9.65952575e-01 8.61819685e-01 9.24208105e-01 6.20897174e-01
-4.09626752e-01 8.97064805e-02 3.60177308e-01 6.22668207e-01
6.11915946e-01 7.67763481e-02 -4.56801474e-01 6.11969419e-02
-5.56408942e-01 2.74171442e-01 4.85906959e-01 1.07559013e+00
6.66436851e-01 3.49860996e-01 -1.14717349e-01 1.13788676e+00
5.45583785e-01 1.12920737e+00 3.85446906e-01 -6.15993679e-01
-2.59292535e-02 1.10538936e+00 -2.94963330e-01 -1.29420483e+00
-4.68678057e-01 -1.97023124e-01 -1.12802446e+00 4.41584766e-01
1.40200198e-01 5.46176910e-01 -9.62614894e-01 1.11641717e+00
3.19797754e-01 -1.21209010e-01 -6.04797862e-02 1.08806872e+00
1.27863038e+00 7.27817595e-01 -3.38076144e-01 4.18290019e-01
9.93332982e-01 -6.28857195e-01 -2.92140156e-01 2.75199413e-01
4.48480755e-01 -9.83866751e-01 5.92742443e-01 -2.69205153e-01
-1.04468298e+00 -8.01719964e-01 -1.02109134e+00 -3.37602913e-01
-7.45405853e-01 9.66513827e-02 6.43245816e-01 4.01535541e-01
-9.53736007e-01 3.67694676e-01 -7.32168436e-01 -5.37375867e-01
7.76297867e-01 5.54568052e-01 -6.80038035e-01 -2.56853908e-01
-6.41621888e-01 8.31024647e-01 4.66573201e-02 -1.12349287e-01
-7.97522128e-01 -2.46488839e-01 -9.82990384e-01 1.97543755e-01
-2.26235703e-01 -6.39273822e-01 8.10444474e-01 -1.13806374e-01
-1.33427036e+00 1.32316685e+00 4.57747057e-02 6.52248189e-02
-8.38823244e-02 7.11216778e-02 -2.51993150e-01 2.67177187e-02
-1.84674188e-01 6.71360791e-01 9.75173533e-01 -1.03145099e+00
-1.18658647e-01 -6.90458834e-01 -1.06998958e-01 1.76388964e-01
-3.17146122e-01 -7.06710219e-02 -4.46995676e-01 -5.41849136e-01
5.83228767e-01 -7.66256154e-01 1.16449513e-01 3.77000153e-01
5.30135557e-02 -7.31761277e-01 1.32686126e+00 2.98638791e-02
5.23446620e-01 -2.28544950e+00 -2.08894201e-02 3.81277561e-01
2.84638464e-01 5.40681183e-01 -6.51922822e-01 3.78721803e-01
1.13304920e-01 1.35546774e-01 1.45946383e-01 -1.51906729e-01
2.78373122e-01 3.20479423e-01 -2.44323060e-01 5.29852450e-01
4.75035071e-01 1.53640425e+00 -5.88373840e-01 -2.83545852e-01
5.10263503e-01 7.83392131e-01 -3.01339120e-01 3.05179685e-01
3.29472482e-01 2.29848754e-02 -9.00876224e-01 1.03761661e+00
9.43236113e-01 -4.78602707e-01 -3.85428131e-01 -2.42915034e-01
8.30506235e-02 1.33665934e-01 -8.28153074e-01 1.68931448e+00
-3.49162340e-01 4.19016629e-01 -2.20070601e-01 -1.19015503e+00
1.67258286e+00 1.43265873e-01 5.94000459e-01 -8.84856701e-01
2.71393716e-01 3.92806321e-01 -2.14061856e-01 -2.22134441e-01
9.33231339e-02 -3.12580280e-02 -8.85073543e-02 7.59207129e-01
4.01628941e-01 -7.08274186e-01 -4.75894988e-01 -3.55361432e-01
8.40503693e-01 -1.33224279e-01 3.36883187e-01 -8.61678421e-02
6.78722084e-01 -4.41262513e-01 7.07426667e-02 7.06882477e-01
-2.39842534e-01 6.88409984e-01 6.95063770e-02 -1.03298426e+00
-1.23909509e+00 -1.20525646e+00 -3.67126644e-01 6.04659617e-01
1.87690899e-01 -4.05719802e-02 -1.27197564e-01 -6.73513293e-01
5.33100426e-01 -2.03640640e-01 -4.60125029e-01 -4.14645702e-01
-5.64094663e-01 -1.79789513e-01 5.26080251e-01 5.61278522e-01
6.18892074e-01 -1.15172374e+00 -4.28711534e-01 2.95146927e-03
7.07527459e-01 -8.83441567e-01 -6.34274960e-01 -5.41463718e-02
-1.03195953e+00 -1.06783390e+00 -1.00792265e+00 -1.31270278e+00
6.38035417e-01 6.86469555e-01 1.02146971e+00 3.32895607e-01
-4.85917240e-01 9.60561097e-01 -3.25716376e-01 -4.41663384e-01
-1.54801264e-01 -1.52725726e-02 6.31517619e-02 -1.17886961e-01
8.90800953e-01 -5.82005441e-01 -4.96890992e-01 2.01658517e-01
-8.86710048e-01 -6.51254952e-01 7.89056540e-01 1.18778563e+00
7.95804858e-01 3.58403707e-03 5.62985837e-01 -1.26550660e-01
6.92309141e-01 -1.08655542e-01 -6.58062220e-01 1.65822357e-01
-3.26999217e-01 1.56557128e-01 2.63201267e-01 -2.93803811e-01
-5.39807737e-01 7.42325485e-02 -3.59014273e-01 -7.07655787e-01
-4.68299568e-01 3.18602115e-01 5.54358913e-03 -5.31465590e-01
3.51083696e-01 5.47102094e-01 2.73800194e-01 -7.91563272e-01
1.46739900e-01 8.71902883e-01 -5.94802201e-02 -5.16635120e-01
9.22016799e-01 4.06463444e-01 4.33929622e-01 -9.91592467e-01
-5.14706671e-01 -4.91344571e-01 -9.92329895e-01 1.61566153e-01
6.93139732e-01 -8.53209019e-01 -8.76895607e-01 6.18725419e-01
-1.28936887e+00 1.27209485e-01 -2.53449887e-01 3.37561667e-01
-6.87631845e-01 3.40507984e-01 -4.98641431e-01 -5.74361563e-01
-6.26284003e-01 -1.12840593e+00 1.74274457e+00 -1.68567859e-02
2.44992077e-01 -9.68791962e-01 1.80393204e-01 -1.83919415e-01
7.12975919e-01 1.11702979e-01 1.38447142e+00 -7.59820700e-01
-6.65427804e-01 -6.98523164e-01 -6.18499100e-01 3.75219047e-01
4.24286216e-01 -4.17240322e-01 -8.50686014e-01 -5.18126190e-01
-1.28334507e-01 -5.61376750e-01 8.56355369e-01 4.43953842e-01
1.23725855e+00 1.34229243e-01 -1.39962733e-01 5.19330084e-01
1.37754691e+00 2.53581911e-01 3.57335210e-01 -1.39437750e-01
4.96594638e-01 2.23308578e-01 1.87910095e-01 4.98080850e-01
2.00021207e-01 7.66220033e-01 3.28284740e-01 -1.01654530e-01
-3.56928527e-01 -2.23920390e-01 -2.11493164e-01 1.08396113e+00
-1.08574323e-01 4.21077088e-02 -9.39071596e-01 4.70672607e-01
-1.63350630e+00 -6.22039318e-01 4.59468961e-01 1.99734795e+00
3.21127027e-01 -7.66809955e-02 -8.93614367e-02 -7.28897601e-02
3.60768110e-01 4.03993338e-01 -4.91734207e-01 -2.90341109e-01
-4.13353175e-01 4.92819220e-01 -7.91205019e-02 1.74002871e-01
-1.20869732e+00 1.01168823e+00 6.60017443e+00 8.74001503e-01
-1.26244187e+00 -3.10646873e-02 3.21073681e-01 2.94770390e-01
-3.04205328e-01 -4.30597872e-01 -6.31093979e-01 -1.10902525e-01
1.51700988e-01 1.24596087e-02 2.26116166e-01 7.29592681e-01
-2.85802990e-01 4.74954218e-01 -1.07053185e+00 1.59760427e+00
2.19425097e-01 -1.33676791e+00 6.72221899e-01 2.48678476e-01
7.22830653e-01 1.84574187e-01 1.66947037e-01 4.33874398e-01
-2.31128752e-01 -1.28056419e+00 2.10223600e-01 5.86524069e-01
8.69235694e-01 -7.25726962e-01 8.47301006e-01 1.80146828e-01
-1.37472439e+00 1.21424235e-01 -9.41483498e-01 2.35934909e-02
-1.59150034e-01 4.70606774e-01 -5.91230094e-01 4.21368599e-01
5.57888687e-01 1.05617070e+00 -3.54403287e-01 1.34507656e+00
7.96286985e-02 -1.35598391e-01 -5.49858749e-01 -2.93447167e-01
4.55134094e-01 -1.75882891e-01 6.13489807e-01 9.58744466e-01
6.75920844e-01 1.44968256e-01 3.58878970e-01 1.11158657e+00
-2.53865451e-01 2.49110684e-01 -1.49924695e+00 -2.95256019e-01
3.48847926e-01 1.00188935e+00 -4.73735541e-01 -1.61342382e-01
-6.30432129e-01 8.37079287e-01 1.50025263e-01 3.85019004e-01
9.64296237e-03 -5.09438157e-01 8.30039859e-01 -1.71886027e-01
6.17600620e-01 -6.67729855e-01 -9.87528544e-03 -8.65578175e-01
-1.69263929e-01 -4.76444066e-01 2.09777087e-01 -7.53326416e-01
-1.73603773e+00 5.26875436e-01 -2.57250190e-01 -1.37387586e+00
-1.33721605e-01 -1.15628815e+00 -5.08100927e-01 7.95860648e-01
-1.58992970e+00 -1.38121200e+00 -1.69376522e-01 8.41413379e-01
2.93145478e-01 -7.05417812e-01 1.21853578e+00 3.39978307e-01
2.02491641e-01 5.83675683e-01 1.41917914e-01 4.58612084e-01
5.63973427e-01 -1.02311969e+00 5.62808394e-01 -1.77981239e-02
6.16769254e-01 4.00300264e-01 -3.45680475e-01 -3.98989022e-01
-1.86685276e+00 -8.53377879e-01 8.46909344e-01 -4.39577043e-01
2.65118003e-01 -1.63781166e-01 -8.13887954e-01 2.57510185e-01
-1.26514779e-02 5.64385831e-01 6.32226944e-01 -2.91555654e-02
-6.12351120e-01 -1.82065085e-01 -1.12295783e+00 1.78360537e-01
1.20014036e+00 -8.41224015e-01 -7.13083148e-01 2.70367444e-01
5.08778393e-01 -4.38430458e-01 -1.21827853e+00 5.61047316e-01
6.69839025e-01 -6.13942504e-01 1.46008742e+00 -8.77405643e-01
1.99744001e-01 -2.01712504e-01 -6.15990341e-01 -1.10607851e+00
-6.92649484e-01 8.03747997e-02 -1.63575336e-01 7.96152949e-01
-5.50014898e-02 -4.53726500e-01 6.45164728e-01 1.27551287e-01
-1.32703289e-01 -8.77174616e-01 -1.15114307e+00 -9.84629571e-01
4.98287469e-01 -1.30867764e-01 9.58360016e-01 7.25804806e-01
-5.89021206e-01 2.59204119e-01 -1.06008120e-01 -9.20588225e-02
5.25025070e-01 9.74093080e-01 8.08624446e-01 -1.49505198e+00
2.13200390e-01 -9.66297030e-01 -1.14487612e+00 -1.64441049e+00
4.41020906e-01 -1.43992472e+00 -4.53114629e-01 -1.47410274e+00
3.37749302e-01 -5.19373238e-01 -6.53123915e-01 7.44553387e-01
4.21990246e-01 4.58825469e-01 3.16533923e-01 6.48050606e-02
-4.99434620e-01 1.00211656e+00 1.75946248e+00 -7.25309610e-01
8.75752978e-03 -9.89600345e-02 -2.57777959e-01 3.49126667e-01
5.05731225e-01 -2.58157432e-01 -6.17481880e-02 -6.55673027e-01
3.56910890e-03 -1.70446426e-01 4.02784765e-01 -6.18861079e-01
2.82192260e-01 1.36100631e-02 9.19720232e-01 -1.15757477e+00
5.79885840e-01 -1.11002350e+00 -3.69896740e-01 1.71507001e-01
5.46624735e-02 -6.41388223e-02 2.58184642e-01 4.43550318e-01
-7.86754906e-01 -1.74903274e-01 5.85622907e-01 -2.43270069e-01
-7.53500998e-01 8.46526444e-01 2.68016309e-02 -2.07562059e-01
5.57268322e-01 -3.84105533e-01 -1.16455078e-01 -3.70547146e-01
-5.28369963e-01 1.45727387e-02 2.62770712e-01 7.49309957e-01
1.17122602e+00 -2.28124046e+00 -6.36834979e-01 8.29060614e-01
2.18912929e-01 4.82034957e-04 1.14337228e-01 3.36799651e-01
-4.22788262e-01 7.68725812e-01 -4.94840592e-01 -9.56951678e-01
-9.76509571e-01 5.80999196e-01 3.12974542e-01 7.18773603e-02
-7.32224584e-01 6.10953152e-01 1.61506698e-01 -9.75301266e-01
2.59877235e-01 -2.55923450e-01 -5.44355512e-02 -1.29333660e-01
4.11211938e-01 7.00755715e-02 2.10074514e-01 -7.71861196e-01
-5.71740568e-01 1.59284222e+00 9.14704353e-02 5.11815131e-01
1.84204638e+00 1.86834604e-01 -2.76295751e-01 1.48327380e-01
1.95378947e+00 -3.32030922e-01 -8.68773580e-01 -5.54068625e-01
-6.07991517e-02 -6.76962018e-01 1.58006504e-01 -4.17402029e-01
-1.27292609e+00 1.24010766e+00 8.07139635e-01 2.45511815e-01
9.13214564e-01 2.88738787e-01 9.87574935e-01 1.04739535e+00
4.29349035e-01 -7.79748738e-01 2.98696578e-01 1.22628260e+00
1.33137321e+00 -1.50052774e+00 -4.77933511e-02 9.92968306e-02
1.29178516e-03 1.37433112e+00 2.58358568e-01 -6.13845468e-01
1.26365709e+00 1.42825134e-02 6.54720888e-02 -5.54224253e-01
-5.24538219e-01 -3.67889076e-01 8.86659801e-01 5.40008485e-01
2.03179970e-01 5.82217649e-02 2.03119472e-01 3.22695136e-01
2.49138847e-01 -4.62772757e-01 -2.05772161e-01 8.21246922e-01
-4.65316564e-01 -1.30107212e+00 -2.39280537e-01 4.31979775e-01
-1.00966752e-01 2.12077111e-01 -6.31344140e-01 7.27903247e-01
-2.52020538e-01 3.72173876e-01 2.04480484e-01 -2.28971407e-01
4.73771334e-01 2.27788478e-01 8.78649652e-01 -4.65130717e-01
-3.88714485e-02 3.35706174e-01 -5.61940730e-01 -4.46148068e-01
-4.32243735e-01 -2.88927734e-01 -1.05395341e+00 -2.09385782e-01
-1.61639184e-01 -3.13329667e-01 6.82746172e-01 8.50804806e-01
6.78281486e-01 1.89917870e-02 8.28711092e-01 -1.14027739e+00
-7.62940049e-01 -7.33784378e-01 -8.73552382e-01 5.79630971e-01
4.51298922e-01 -1.01807976e+00 -1.35896102e-01 -5.29987216e-01]
|
[8.1653413772583, -3.9036552906036377]
|
72ea91b6-1932-4970-88b9-87ef29ab08ed
|
deepmask-an-algorithm-for-cloud-and-cloud
|
1911.03607
| null |
https://arxiv.org/abs/1911.03607v1
|
https://arxiv.org/pdf/1911.03607v1.pdf
|
DeepMask: an algorithm for cloud and cloud shadow detection in optical satellite remote sensing images using deep residual network
|
Detecting and masking cloud and cloud shadow from satellite remote sensing images is a pervasive problem in the remote sensing community. Accurate and efficient detection of cloud and cloud shadow is an essential step to harness the value of remotely sensed data for almost all downstream analysis. DeepMask, a new algorithm for cloud and cloud shadow detection in optical satellite remote sensing imagery, is proposed in this study. DeepMask utilizes ResNet, a deep convolutional neural network, for pixel-level cloud mask generation. The algorithm is trained and evaluated on the Landsat 8 Cloud Cover Assessment Validation Dataset distributed across 8 different land types. Compared with CFMask, the most widely used cloud detection algorithm, land-type-specific DeepMask models achieve higher accuracy across all land types. The average accuracy is 93.56%, compared with 85.36% from CFMask. DeepMask also achieves 91.02% accuracy on all-land-type dataset. Compared with other CNN-based cloud mask algorithms, DeepMask benefits from the parsimonious architecture and the residual connection of ResNet. It is compatible with input of any size and shape. DeepMask still maintains high performance when using only red, green, blue, and NIR bands, indicating its potential to be applied to other satellite platforms that only have limited optical bands.
|
['Ke Xu', 'Kaiyu Guan', 'Yunan Luo', 'Sibo Wang', 'Jian Peng']
|
2019-11-09
| null | null | null | null |
['shadow-detection', 'cloud-detection']
|
['computer-vision', 'computer-vision']
|
[ 2.42018655e-01 -5.90726376e-01 2.76814289e-02 -5.84861450e-02
-5.09223342e-01 -8.48342657e-01 5.04653573e-01 -3.31192166e-01
-1.99513346e-01 6.18647397e-01 -3.65506440e-01 -7.88057864e-01
1.04954675e-01 -1.04649436e+00 -2.54916161e-01 -8.92560601e-01
-2.69739598e-01 1.99122094e-02 4.11329716e-02 -2.49942347e-01
-1.65910646e-01 9.43048835e-01 -1.51099050e+00 2.80509531e-01
1.04678142e+00 9.14936304e-01 6.24869943e-01 6.75319254e-01
-1.22995876e-01 4.41412657e-01 -3.28225136e-01 3.49063069e-01
6.97096169e-01 -3.80861372e-01 -6.12541735e-01 -2.24405117e-02
7.20848024e-01 -5.16453087e-01 -1.91506475e-01 1.12894142e+00
4.90552962e-01 -1.72448546e-01 2.96901733e-01 -9.46677208e-01
-3.34827513e-01 3.65646631e-01 -6.58307672e-01 4.77119923e-01
-6.23668373e-01 2.88194299e-01 7.60820329e-01 -1.13322055e+00
1.95804238e-01 7.13066757e-01 8.94579887e-01 1.69683665e-01
-1.02627110e+00 -1.05967605e+00 5.30038290e-02 -3.40820700e-02
-1.87493443e+00 -1.59581304e-01 -2.31764819e-02 -7.77498186e-01
1.07279563e+00 7.35854208e-01 1.08022666e+00 -3.29976231e-02
1.67251989e-01 3.02793086e-01 1.63343501e+00 -1.80209115e-01
1.45245478e-01 -3.33564490e-01 -2.14191526e-01 1.97317168e-01
4.72990066e-01 3.48222286e-01 4.16257344e-02 -6.64719567e-02
6.93028629e-01 3.91911030e-01 -2.62283266e-01 4.22253817e-01
-7.83437967e-01 8.66388440e-01 9.46155906e-01 1.87385872e-01
-5.93093514e-01 4.37580377e-01 -2.07129478e-01 1.89917892e-01
9.52217281e-01 4.56094682e-01 -3.77531469e-01 3.75006527e-01
-1.49470389e+00 3.56669784e-01 1.67894900e-01 6.48510516e-01
9.17116225e-01 7.37629354e-01 9.88179967e-02 6.68649197e-01
2.41672739e-01 1.12600207e+00 6.38236701e-02 -7.09560633e-01
1.95400640e-02 7.14387476e-01 2.52754718e-01 -8.48955333e-01
-4.49874490e-01 -7.46362686e-01 -9.97666657e-01 7.42084861e-01
-3.17333907e-01 -1.36650279e-01 -1.33942568e+00 9.14112568e-01
1.11064598e-01 3.78840506e-01 1.14345461e-01 1.13456118e+00
9.71423745e-01 8.06075156e-01 1.76517636e-01 1.98947564e-01
1.41210294e+00 -5.54039776e-01 -3.94936025e-01 -6.61774158e-01
1.69503644e-01 -6.62603915e-01 6.50673330e-01 -7.13909268e-02
-2.44673133e-01 -2.57073969e-01 -1.01396036e+00 3.24176490e-01
-8.18482220e-01 3.40661496e-01 9.17437196e-01 5.36276579e-01
-1.37760186e+00 3.43351781e-01 -6.55207336e-01 -2.30542406e-01
6.26877666e-01 1.70846269e-01 1.24282695e-01 -4.47080508e-02
-1.11804974e+00 1.02021897e+00 4.34868425e-01 9.17979062e-01
-8.72543335e-01 -6.44773006e-01 -5.88122010e-01 3.00484091e-01
-2.77314931e-02 -2.35147163e-01 9.25382435e-01 -1.46098459e+00
-8.90703261e-01 1.02099955e+00 -1.30233122e-02 -5.91542363e-01
2.69571215e-01 -5.77616580e-02 -5.76461554e-01 1.85650960e-01
9.05972645e-02 7.26899564e-01 9.15185213e-01 -1.37004769e+00
-7.67048836e-01 -1.70123383e-01 -1.08052015e-01 1.87199563e-01
7.34995902e-02 1.31928578e-01 1.79596826e-01 -6.24026299e-01
3.40493232e-01 -1.24498403e+00 -3.55405599e-01 5.33728264e-02
-9.88860242e-03 4.06676501e-01 1.21206617e+00 -8.42399478e-01
8.88157785e-01 -2.03587866e+00 -5.46177447e-01 2.31977761e-01
3.65865767e-01 9.03661728e-01 -3.15072298e-01 4.18377101e-01
-2.25047812e-01 4.84104007e-01 -7.67706871e-01 2.43066311e-01
-3.88208091e-01 3.53939474e-01 -6.24573708e-01 5.29417455e-01
5.65651238e-01 9.00532663e-01 -7.40233362e-01 -5.79788797e-02
5.28278589e-01 5.24730504e-01 9.73355547e-02 3.23958918e-02
-3.27085942e-01 1.44716099e-01 -3.47611941e-02 8.94137263e-01
1.56345367e+00 -1.03685707e-01 2.06916839e-01 3.79365236e-01
-6.66076779e-01 1.68964844e-02 -8.97215128e-01 8.62511754e-01
-4.20110941e-01 1.29335880e+00 4.27645266e-01 -3.42301846e-01
1.01649928e+00 1.83123708e-01 1.53837964e-01 -6.77587926e-01
-1.89869151e-01 3.86326969e-01 1.98651433e-01 -4.64368463e-01
7.62383580e-01 -3.52159023e-01 4.25454736e-01 2.01382652e-01
-6.48615241e-01 -3.86955887e-01 -3.73748779e-01 3.04188691e-02
5.54414749e-01 -4.19352651e-02 -1.03521444e-01 -5.37451506e-01
9.28970228e-04 5.16637087e-01 3.04708153e-01 7.25677013e-01
3.46515328e-02 7.52734959e-01 -8.03664327e-02 -8.29436243e-01
-8.36305380e-01 -6.80843651e-01 -3.34226638e-01 1.01683104e+00
-5.19126887e-04 -8.92885681e-03 -4.41945106e-01 -1.52504236e-01
3.11334670e-01 4.01565850e-01 -5.16583562e-01 3.80199015e-01
-1.56869859e-01 -1.27796555e+00 5.44532001e-01 3.36202532e-01
8.58985543e-01 -1.14154816e+00 -7.24437416e-01 1.51887819e-01
-2.95526743e-01 -1.11399329e+00 2.28062645e-01 8.14706460e-02
-9.26083803e-01 -9.92952704e-01 -4.54773128e-01 -4.05121386e-01
5.06623983e-01 1.05978394e+00 1.16395378e+00 5.22218227e-01
-5.71784675e-01 -1.79514989e-01 -4.94606465e-01 -6.55402303e-01
-3.05869937e-01 1.72072873e-01 -4.37879235e-01 -1.77442562e-02
4.74992782e-01 -4.62802142e-01 -8.40138495e-01 6.53176159e-02
-1.20985150e+00 1.38614342e-01 6.30201221e-01 4.60068941e-01
5.58620155e-01 2.78242439e-01 1.21863060e-01 -4.72501665e-01
-1.13829061e-01 -5.82954586e-01 -1.10005510e+00 -2.03527421e-01
-6.97737932e-01 -6.70001984e-01 5.16966395e-02 2.26328313e-01
-7.72778034e-01 3.49296242e-01 1.46368861e-01 -2.32236192e-01
-9.69386175e-02 9.57666278e-01 2.33900115e-01 -3.60136569e-01
6.39216840e-01 3.61870021e-01 4.37127016e-02 -4.17893559e-01
8.37288722e-02 9.85059917e-01 5.07495224e-01 2.15026766e-01
1.12641752e+00 1.02436996e+00 -1.35195374e-01 -1.31962752e+00
-6.12865806e-01 -7.11416662e-01 -5.68089604e-01 -4.06835139e-01
8.18289042e-01 -1.43421626e+00 -3.02531809e-01 9.85553503e-01
-8.93690169e-01 -7.07060099e-01 1.63455382e-01 3.48328143e-01
3.80115479e-01 6.70012534e-02 -3.52548137e-02 -1.07831895e+00
-8.38075697e-01 -6.55812204e-01 9.47348058e-01 2.25971997e-01
3.00056905e-01 -6.45016313e-01 -1.77806318e-01 -4.93262298e-02
9.77761984e-01 6.92115486e-01 2.45275229e-01 2.75464598e-02
-1.06986415e+00 -2.92508960e-01 -8.18897247e-01 5.99491417e-01
5.63607931e-01 5.62103689e-01 -1.41320038e+00 -4.27378535e-01
-4.35628116e-01 1.17928416e-01 1.43386722e+00 5.23504376e-01
1.15965605e+00 -3.88927609e-01 -6.85618594e-02 1.00187945e+00
2.01480317e+00 -3.71607654e-02 9.43264186e-01 7.13434637e-01
7.59476483e-01 3.08598131e-01 4.01512325e-01 3.45453113e-01
2.04044491e-01 1.01997130e-01 1.30886245e+00 -8.35377097e-01
-9.15688500e-02 4.73022491e-01 4.52637672e-03 3.07612985e-01
-3.16149652e-01 -1.10770704e-03 -1.41044664e+00 6.67213261e-01
-1.54707575e+00 -1.37368000e+00 -6.31615102e-01 1.90597486e+00
4.77721125e-01 -2.24535316e-01 -2.75268674e-01 -1.54488489e-01
8.28908026e-01 6.21109784e-01 -5.31365216e-01 -1.65454492e-01
-6.37207389e-01 5.93308091e-01 1.27849007e+00 5.02231181e-01
-1.36437774e+00 1.29426193e+00 6.38891697e+00 4.20185566e-01
-1.72773552e+00 1.65354788e-01 2.96173215e-01 9.52622220e-02
-3.01107228e-01 1.82883322e-01 -5.27008533e-01 3.67112756e-01
7.36811042e-01 2.66424984e-01 5.26043653e-01 7.71691024e-01
7.08565295e-01 -5.05017042e-01 -3.96990255e-02 6.53275430e-01
-3.53976190e-01 -1.74223864e+00 -1.28784776e-01 8.93892422e-02
9.39773381e-01 1.06953323e+00 5.36911236e-03 -4.57744189e-02
5.65756917e-01 -1.45302975e+00 6.61785960e-01 2.81422317e-01
1.38694918e+00 -5.26991308e-01 8.97399843e-01 1.16602287e-01
-1.40853953e+00 -1.18096955e-01 -5.45029581e-01 -5.49819827e-01
-4.78242368e-01 6.99030757e-01 -8.66209090e-01 4.15008396e-01
9.89389777e-01 7.00649977e-01 -5.66845775e-01 1.17404819e+00
-1.66104928e-01 9.57516670e-01 -4.26484615e-01 3.29853594e-01
6.65790081e-01 -3.49180281e-01 2.00127393e-01 1.49911892e+00
3.68420243e-01 4.49392974e-01 7.63159245e-02 8.21407557e-01
1.24695197e-01 -2.92780846e-01 -5.66226602e-01 -1.47499904e-01
8.31630230e-01 1.57723999e+00 -7.72784352e-01 -3.25438827e-01
-1.15129195e-01 6.58457220e-01 -5.08036315e-01 3.94913733e-01
-7.09565759e-01 -3.19252074e-01 1.19032359e+00 1.93585113e-01
3.94225568e-01 -4.87752587e-01 -4.10416871e-01 -7.83695221e-01
-1.32937074e-01 -8.72672379e-01 8.03611800e-02 -1.23254132e+00
-7.57121503e-01 7.16115057e-01 -1.60017252e-01 -1.65580261e+00
1.77303374e-01 -6.56848311e-01 -9.09046173e-01 1.49863768e+00
-2.40210795e+00 -1.37215662e+00 -1.32865417e+00 2.08550677e-01
1.56179458e-01 8.79655257e-02 1.03497851e+00 -8.95759016e-02
-2.79673398e-01 -1.63667142e-01 5.50463259e-01 2.72220612e-01
2.22984880e-01 -1.12506950e+00 7.85222530e-01 1.22751439e+00
-3.41194987e-01 2.04970971e-01 4.17370051e-01 -6.15075350e-01
-1.22009170e+00 -1.85130525e+00 6.06441915e-01 2.32898947e-02
5.54126322e-01 -1.45445451e-01 -9.83863056e-01 6.13102496e-01
1.81515574e-01 1.99772969e-01 5.55032730e-01 -4.16213632e-01
-4.01941389e-01 -6.37870014e-01 -1.07632923e+00 2.47237444e-01
5.57617545e-01 -6.65464103e-01 1.16223106e-02 6.65896535e-01
5.40897846e-01 -5.49033940e-01 -5.48929811e-01 4.05302823e-01
7.58675933e-01 -9.65930104e-01 6.64513469e-01 -2.54354119e-01
5.61992526e-01 -6.72305822e-01 -3.50240201e-01 -1.33081365e+00
-8.18720996e-01 -3.23530972e-01 5.27049959e-01 8.26184869e-01
3.30719978e-01 -8.78693521e-01 5.82883894e-01 1.08146526e-01
-2.99895108e-01 -3.33429053e-02 -9.54903066e-01 -8.82335663e-01
2.70206600e-01 -2.29410067e-01 1.09374356e+00 1.21675706e+00
-9.56679344e-01 -4.03549910e-01 -1.39096722e-01 1.02604914e+00
3.77693772e-01 6.34933412e-01 7.38337934e-01 -1.47588658e+00
3.19687277e-01 -6.01333201e-01 -1.82104692e-01 -1.44767761e-01
-8.13688114e-02 -8.68649244e-01 4.03357036e-02 -1.76001954e+00
1.15045018e-01 -8.48649979e-01 -1.29185781e-01 9.48745072e-01
-2.04522878e-01 8.43600392e-01 3.05145621e-01 5.55471122e-01
4.60997462e-01 2.80778170e-01 8.71965408e-01 -4.04576868e-01
-4.67866510e-02 -2.36700289e-02 -3.80183756e-01 4.86872911e-01
1.27989304e+00 -6.57597125e-01 1.97278947e-01 -8.37352455e-01
2.92359591e-01 -3.76483679e-01 8.84350777e-01 -1.05620384e+00
-3.60634893e-01 -6.50021434e-01 4.88510817e-01 -1.02338016e+00
1.98313221e-01 -9.21819448e-01 6.68695092e-01 7.16735005e-01
3.96733165e-01 1.16711790e-02 6.25863433e-01 1.29963616e-02
-1.80889726e-01 7.58484527e-02 9.81804550e-01 -4.24265742e-01
-1.18396521e+00 5.39642572e-01 -7.99614310e-01 -4.77162451e-01
6.78525209e-01 -3.46775234e-01 -6.78807616e-01 -1.21374004e-01
-3.84278744e-01 2.86418706e-01 5.80195010e-01 3.56724441e-01
6.05640948e-01 -9.66071308e-01 -1.25659966e+00 2.90706962e-01
3.10251355e-01 1.39019534e-01 2.45266393e-01 4.84294504e-01
-1.16945636e+00 1.96880817e-01 -1.24754027e-01 -8.12290847e-01
-1.25183105e+00 -8.69204998e-02 9.34699595e-01 3.22209477e-01
-5.31450510e-01 6.53774142e-01 -2.26193905e-01 -4.18270200e-01
-5.98259807e-01 -3.62116426e-01 6.00095838e-02 2.22923592e-01
5.62425852e-01 1.57326818e-01 4.77151930e-01 -6.18434906e-01
-5.98180473e-01 3.57725650e-01 5.08091509e-01 1.00199006e-01
1.53126097e+00 3.99348065e-02 -6.84016705e-01 1.64510220e-01
6.45721555e-01 -1.05163176e-02 -1.20745659e+00 -2.23855183e-01
-2.23813459e-01 -7.75747240e-01 5.79307139e-01 -9.82310832e-01
-1.45224845e+00 8.29555333e-01 1.04892874e+00 3.10530871e-01
1.25185657e+00 -3.17960650e-01 3.28422964e-01 1.83988437e-01
-8.67616907e-02 -8.46323192e-01 -6.26697958e-01 5.95326543e-01
9.80792403e-01 -1.38961518e+00 3.41973990e-01 -1.89847544e-01
-3.14725876e-01 1.01933527e+00 4.95232821e-01 -1.40522838e-01
6.09039664e-01 3.16122413e-01 5.94218314e-01 -3.99549454e-01
-2.99400508e-01 -7.71197438e-01 -1.09101497e-01 6.62887156e-01
2.09464967e-01 7.75683820e-01 1.33503959e-01 -1.40403077e-01
-1.74400389e-01 4.31875214e-02 7.30412066e-01 9.01869357e-01
-8.42071295e-01 -3.42181414e-01 -7.18465447e-01 7.70429552e-01
-2.91899443e-01 -8.08928430e-01 -5.10302901e-01 8.26252699e-01
1.77689433e-01 9.71832514e-01 3.84634495e-01 -2.95664400e-01
-3.32706541e-01 -2.67645746e-01 -1.99258372e-01 -7.09380507e-01
-6.71341658e-01 1.51252281e-02 -9.50279832e-02 -2.70095885e-01
-6.82082772e-01 -5.16344428e-01 -9.64442790e-01 -7.24308074e-01
-6.19242489e-01 -3.54216434e-02 1.01040375e+00 6.74509525e-01
4.96606112e-01 2.66133159e-01 9.07669485e-01 -1.11716127e+00
-9.65192541e-02 -1.14806831e+00 -9.69493866e-01 -4.33285952e-01
8.77191782e-01 -4.35161948e-01 -4.08723146e-01 -1.58156887e-01]
|
[9.75967025756836, -1.7043261528015137]
|
86e4b810-115e-4803-b914-356dc3f5c550
|
pointclm-a-contrastive-learning-based
|
2209.00219
| null |
https://arxiv.org/abs/2209.00219v1
|
https://arxiv.org/pdf/2209.00219v1.pdf
|
PointCLM: A Contrastive Learning-based Framework for Multi-instance Point Cloud Registration
|
Multi-instance point cloud registration is the problem of estimating multiple poses of source point cloud instances within a target point cloud. Solving this problem is challenging since inlier correspondences of one instance constitute outliers of all the other instances. Existing methods often rely on time-consuming hypothesis sampling or features leveraging spatial consistency, resulting in limited performance. In this paper, we propose PointCLM, a contrastive learning-based framework for mutli-instance point cloud registration. We first utilize contrastive learning to learn well-distributed deep representations for the input putative correspondences. Then based on these representations, we propose a outlier pruning strategy and a clustering strategy to efficiently remove outliers and assign the remaining correspondences to correct instances. Our method outperforms the state-of-the-art methods on both synthetic and real datasets by a large margin.
|
['Manning Wang', 'Xinrong Chen', 'Qiuye Jin', 'Zhihao LI', 'Mingzhi Yuan']
|
2022-09-01
| null | null | null | null |
['point-cloud-registration']
|
['computer-vision']
|
[-1.23023249e-01 -2.78926432e-01 1.39766455e-01 -2.63451695e-01
-1.41268873e+00 -3.49543393e-01 5.56007504e-01 4.04923886e-01
-1.04209691e-01 4.65656072e-01 -3.87789488e-01 2.28633344e-01
-2.22543448e-01 -4.86676753e-01 -1.21594715e+00 -5.27013481e-01
-1.59395292e-01 1.01122761e+00 4.15390611e-01 1.32699355e-01
5.47616661e-01 1.01186264e+00 -1.61635113e+00 1.72351345e-01
9.29759264e-01 9.75230277e-01 5.46060910e-04 1.96629211e-01
-1.17447883e-01 2.32000947e-01 -6.59685731e-01 -1.56071141e-01
5.70815921e-01 -3.62522714e-02 -4.02060539e-01 9.40924734e-02
1.00449610e+00 -1.69068351e-01 -2.40709707e-01 9.52017844e-01
3.46975893e-01 2.82726318e-01 7.31923401e-01 -1.59771216e+00
-3.13327551e-01 -3.85935269e-02 -1.00396943e+00 3.22596356e-02
2.81635284e-01 -4.33113798e-02 9.14657891e-01 -1.41480279e+00
3.42817873e-01 9.86059129e-01 1.04629266e+00 -5.05239479e-02
-1.10253811e+00 -8.84070039e-01 2.04559371e-01 2.06875652e-01
-1.73517036e+00 -4.19920683e-01 8.92846763e-01 -5.23499250e-01
9.44392264e-01 2.87558079e-01 5.96851528e-01 7.38418937e-01
1.18054502e-01 3.68139893e-01 4.96507168e-01 -5.48259690e-02
2.24776685e-01 -5.14455974e-01 -1.83837593e-01 5.94144881e-01
5.91871798e-01 1.70996577e-01 -5.42950332e-01 -7.06149638e-01
7.97172785e-01 4.86443669e-01 -3.77543494e-02 -7.45307207e-01
-1.29271364e+00 7.69766867e-01 6.21842384e-01 3.39026861e-02
-5.71706414e-01 3.93775374e-01 2.22099811e-01 1.51789680e-01
6.25572741e-01 4.37569261e-01 -5.64705193e-01 1.32070437e-01
-1.34787273e+00 5.44547498e-01 4.61135119e-01 1.17553151e+00
1.13733375e+00 -2.13512391e-01 1.96426004e-01 6.88144684e-01
4.72658664e-01 3.79730999e-01 3.46684694e-01 -7.60078549e-01
5.35542071e-01 7.36521006e-01 1.84912369e-01 -1.21885431e+00
-3.11910450e-01 -2.53543735e-01 -6.65622532e-01 4.08434182e-01
1.66241184e-01 3.20933461e-01 -1.07291424e+00 1.19978809e+00
4.86794800e-01 1.18857825e+00 -3.36065382e-01 8.47914457e-01
5.39484501e-01 5.12797832e-01 -3.67825031e-01 7.29613304e-02
6.25987828e-01 -7.34443545e-01 -1.57170773e-01 -2.04362631e-01
3.86941791e-01 -1.00381207e+00 7.78278887e-01 1.93166807e-01
-9.69191790e-01 -3.28301728e-01 -1.06141114e+00 -9.57037322e-03
-1.90078497e-01 4.07489017e-02 2.28350550e-01 -2.42946502e-02
-6.79218531e-01 9.93088901e-01 -1.09306860e+00 -8.57479870e-02
7.44034648e-01 6.02809846e-01 -4.72327530e-01 -1.85363278e-01
-3.16588730e-01 5.87754190e-01 1.07105359e-01 2.09494278e-01
-7.31437266e-01 -1.01018488e+00 -8.69724989e-01 -1.64511353e-01
2.05006018e-01 -6.47698760e-01 8.58218968e-01 -5.77161491e-01
-1.00076556e+00 8.09278846e-01 -3.93196106e-01 -2.63202548e-01
7.55349636e-01 -5.55163324e-01 -1.64370418e-01 -1.63952008e-01
5.20743549e-01 4.17027324e-01 1.03553760e+00 -1.71420419e+00
-6.41238213e-01 -4.84287620e-01 -4.44034457e-01 -4.36393768e-02
4.22085583e-01 -2.31436163e-01 -5.88902950e-01 -6.42752349e-01
1.02613318e+00 -1.01069891e+00 -4.56602991e-01 1.60104364e-01
-6.93337023e-01 -1.28647089e-01 1.03398371e+00 -4.02095258e-01
5.90492547e-01 -2.22928214e+00 2.49742586e-02 6.87470675e-01
3.69529873e-01 -6.28047660e-02 -2.11127102e-01 3.48772675e-01
-2.43884951e-01 -1.58531785e-01 -2.09376752e-01 -7.04034388e-01
-3.15639116e-02 2.02214375e-01 -6.19886816e-01 1.11745417e+00
1.69289559e-01 5.36055326e-01 -8.42335284e-01 -2.73719639e-01
5.65004945e-01 3.06615472e-01 -5.73021889e-01 3.47584844e-01
-1.37864918e-01 6.38738394e-01 -1.44724876e-01 8.14110339e-01
1.07788551e+00 -1.93617478e-01 -5.60212553e-01 -3.51784497e-01
2.91094556e-02 7.12991729e-02 -1.42360377e+00 1.96603358e+00
-3.20723772e-01 3.31476897e-01 -3.32091361e-01 -6.58756614e-01
9.33287859e-01 -2.57656034e-02 1.03456128e+00 -1.97777122e-01
-8.08264762e-02 5.82929313e-01 -1.62321582e-01 -4.19876203e-02
5.72837949e-01 1.55211210e-01 3.10036335e-02 1.95674464e-01
-8.46627951e-02 -2.59967834e-01 -3.49454463e-01 -8.14161673e-02
1.05387497e+00 2.56580204e-01 3.56138051e-01 -2.98794061e-02
3.24311674e-01 -6.36260863e-03 7.87078023e-01 7.72039235e-01
-2.20921002e-02 1.19701552e+00 5.67988828e-02 -7.37300813e-01
-1.05924714e+00 -1.30080032e+00 -4.07802984e-02 5.41309595e-01
5.25359988e-01 -4.10602361e-01 -2.67952323e-01 -6.77533031e-01
4.63821888e-01 4.14189667e-01 -3.17914009e-01 -5.80088757e-02
-8.38045418e-01 -3.88646394e-01 2.09090695e-01 4.15731996e-01
-1.40086496e-02 -7.75709391e-01 -4.66412961e-01 1.40582487e-01
8.29673186e-02 -9.92283404e-01 -3.58198732e-01 4.16593477e-02
-1.01553941e+00 -1.45208526e+00 -3.81733298e-01 -5.38504243e-01
9.64530826e-01 3.91583949e-01 1.29510522e+00 3.95536929e-01
-3.50537837e-01 1.54964656e-01 -1.39192775e-01 -4.16105688e-01
-1.61267698e-01 1.64132696e-02 3.07909936e-01 -7.62602836e-02
3.55770826e-01 -6.91026092e-01 -5.70337296e-01 3.50955606e-01
-7.17532337e-01 -4.59562808e-01 4.57463294e-01 7.43637443e-01
1.20020556e+00 -1.00613691e-01 1.66698590e-01 -6.62824988e-01
1.12760663e-01 -6.50048256e-01 -8.94772232e-01 3.13451253e-02
-1.47573873e-01 -2.13990733e-01 4.35710937e-01 -2.18718305e-01
-2.45549157e-01 4.68236923e-01 -2.43530702e-03 -1.23965609e+00
-2.71472275e-01 1.92581967e-01 -2.17293948e-01 -4.38337535e-01
5.36979377e-01 -6.20419607e-02 -2.64238238e-01 -5.24336934e-01
2.50184953e-01 2.29222924e-01 7.21814632e-01 -7.55380034e-01
1.37409711e+00 7.55172908e-01 7.89076611e-02 -7.61062741e-01
-7.10895538e-01 -8.50077152e-01 -9.96437967e-01 -5.88633046e-02
3.78053278e-01 -1.02691054e+00 -3.24279696e-01 3.36789131e-01
-1.33738399e+00 1.12481736e-01 -4.14440095e-01 5.04951835e-01
-7.68719494e-01 4.57030892e-01 -2.73783177e-01 -4.64458674e-01
3.23489569e-02 -1.31647503e+00 1.76334631e+00 -1.84724733e-01
-2.14363471e-01 -5.58350027e-01 2.85846353e-01 -1.28476098e-02
-1.09584048e-01 5.41548491e-01 5.34509897e-01 -7.88870513e-01
-1.11500204e+00 -5.97644389e-01 -1.46599948e-01 -2.17849031e-01
2.46024877e-01 1.80036411e-01 -8.93745184e-01 -5.33297062e-01
-9.72526819e-02 -2.53637750e-02 7.69018114e-01 4.09831643e-01
1.54470801e+00 -2.08394781e-01 -6.67894661e-01 1.02352703e+00
1.49459112e+00 -3.24999094e-01 5.86910069e-01 4.73605156e-01
1.06401241e+00 1.79875121e-01 8.73015702e-01 6.03327811e-01
2.59343803e-01 7.45491326e-01 9.11790907e-01 -1.36617303e-01
2.84239054e-01 -2.62018919e-01 -1.07548140e-01 6.50700033e-01
7.42662624e-02 1.32066384e-01 -1.10410678e+00 6.78475618e-01
-2.19910669e+00 -7.90513813e-01 -2.20657557e-01 2.50248146e+00
2.90927619e-01 -4.76358868e-02 -4.94759530e-02 -1.78811848e-02
8.65265012e-01 1.34194195e-01 -6.99625671e-01 2.70994514e-01
1.17500626e-01 2.01189846e-01 6.42167091e-01 3.48087400e-01
-1.31186068e+00 9.38582897e-01 5.71766472e+00 5.23437798e-01
-9.60828722e-01 4.25743982e-02 1.82987183e-01 -2.64285773e-01
-1.95547998e-01 1.45129219e-01 -5.60580909e-01 4.58557487e-01
4.95985359e-01 -1.67951733e-02 1.44871756e-01 1.01911712e+00
1.44114932e-02 1.12512089e-01 -1.36117148e+00 1.36009169e+00
1.89526111e-01 -1.52038181e+00 8.65843669e-02 2.08585843e-01
8.30754697e-01 4.87839937e-01 -1.26558155e-01 7.45528415e-02
2.40851194e-01 -9.51821804e-01 7.16155469e-01 6.23480856e-01
4.95877802e-01 -9.31170821e-01 6.17244124e-01 2.45023489e-01
-1.21462607e+00 1.93704367e-01 -7.08900154e-01 2.13931128e-01
1.18316069e-01 8.32066000e-01 -6.93927348e-01 8.21810901e-01
8.52835894e-01 9.41613674e-01 -6.12068474e-01 1.72033322e+00
8.15637633e-02 1.34471506e-01 -6.79044545e-01 7.89225280e-01
8.36835057e-02 -3.21293741e-01 7.71961033e-01 7.11730361e-01
7.77030230e-01 -3.44145864e-01 5.99862278e-01 9.17706311e-01
-1.31996319e-01 -1.97124342e-03 -9.67363954e-01 6.04334116e-01
8.79259169e-01 1.02428997e+00 -5.99623322e-01 -9.39714983e-02
-4.45827842e-01 1.12177753e+00 5.22357941e-01 1.54333651e-01
-9.92886841e-01 -9.23615098e-02 1.14401317e+00 1.85021132e-01
3.73819858e-01 -4.65378165e-01 -3.92566293e-01 -1.14456451e+00
3.93249512e-01 -6.24089897e-01 2.47926280e-01 -4.83669132e-01
-1.63923860e+00 4.98276919e-01 -1.69160038e-01 -1.85806882e+00
-2.20067278e-01 -2.26640925e-01 -8.45218539e-01 7.04672039e-01
-1.62529337e+00 -1.33449900e+00 -5.89229107e-01 6.61166131e-01
4.01812285e-01 -2.45671406e-01 6.43100977e-01 4.00415421e-01
-4.60201293e-01 5.54250836e-01 3.06017607e-01 5.73956817e-02
8.41158569e-01 -1.09769285e+00 7.26160824e-01 9.84445930e-01
4.98532474e-01 6.50691867e-01 5.56194007e-01 -8.27785075e-01
-1.14564848e+00 -1.55052376e+00 4.79950041e-01 -5.61660290e-01
5.16291320e-01 -2.42032051e-01 -1.18372524e+00 9.61297989e-01
-2.83432484e-01 5.84621072e-01 5.22693992e-01 -1.08541045e-02
-4.50377941e-01 -1.23818725e-01 -1.22568095e+00 5.34256279e-01
9.56531525e-01 -3.85398746e-01 -6.50681257e-01 6.41567469e-01
5.72188616e-01 -9.44599986e-01 -8.69891644e-01 5.72410524e-01
1.69136748e-01 -9.87258434e-01 1.24766195e+00 -4.93915945e-01
2.41160378e-01 -7.14808822e-01 -2.77776808e-01 -1.28365314e+00
-1.90510884e-01 -4.85973388e-01 -2.06384137e-01 1.02922130e+00
1.69382274e-01 -5.53145349e-01 9.93916452e-01 3.55260640e-01
-5.26500106e-01 -7.07122207e-01 -1.34926975e+00 -9.98904526e-01
2.91475374e-03 -5.58764935e-01 1.03734100e+00 1.06731021e+00
-5.94825625e-01 -3.04130375e-01 -1.45947665e-01 8.56566370e-01
9.44322407e-01 3.88102204e-01 1.21847224e+00 -1.59866500e+00
1.02353655e-01 -2.61118591e-01 -8.55688274e-01 -7.48082757e-01
4.95119601e-01 -7.24053025e-01 2.76296526e-01 -1.31613970e+00
-2.48188093e-01 -8.33114088e-01 -1.55856520e-01 2.82127529e-01
-2.56084561e-01 2.95890570e-01 1.41649246e-02 8.01317811e-01
-6.49985552e-01 5.54445326e-01 4.58462149e-01 -9.13429037e-02
-1.90372378e-01 1.78898200e-01 -1.28387243e-01 9.59995866e-01
7.46428072e-01 -8.15008938e-01 -3.40267532e-02 -5.39513409e-01
-4.36821673e-03 -2.50800997e-01 7.34392405e-01 -1.46281624e+00
3.71746868e-01 -1.04814149e-01 6.39258921e-01 -1.21265590e+00
6.46042168e-01 -1.23024273e+00 3.03297132e-01 1.60124779e-01
5.23738153e-02 5.36972165e-01 1.59050718e-01 9.42486286e-01
-2.60046899e-01 1.25798164e-02 8.34073544e-01 -1.43465862e-01
-6.41652048e-01 8.99215698e-01 2.59847343e-01 -9.08721685e-02
1.27262449e+00 -2.85613894e-01 -1.30057916e-01 -2.88036522e-02
-4.61259037e-01 9.27449539e-02 9.34135735e-01 5.08668125e-01
1.03000927e+00 -1.52949178e+00 -6.86083913e-01 5.06926179e-01
4.59240526e-01 6.96221113e-01 7.02218665e-03 7.80774653e-01
-6.38059556e-01 -9.81953591e-02 1.06963608e-03 -1.16131079e+00
-1.11094475e+00 3.10679972e-01 3.92815888e-01 1.10125065e-01
-9.60913658e-01 7.74801373e-01 -3.75444815e-02 -6.79023445e-01
1.78608268e-01 -3.71947974e-01 3.84450555e-01 -2.58311689e-01
1.01921961e-01 4.01019841e-01 4.69093651e-01 -1.00186992e+00
-5.98376572e-01 8.47459078e-01 -1.64052293e-01 2.22123116e-01
1.33758450e+00 2.27525890e-01 -2.39959583e-01 5.07409096e-01
1.34496093e+00 1.92724004e-01 -1.18773425e+00 -1.35141522e-01
3.41336668e-01 -1.11990499e+00 -1.79399669e-01 -7.96157494e-02
-1.19067514e+00 4.00573999e-01 5.75676024e-01 -3.42389584e-01
7.04105198e-01 1.79383129e-01 8.51826310e-01 4.48416144e-01
5.70148647e-01 -8.77832949e-01 1.31176282e-02 4.16388720e-01
1.02582264e+00 -1.44383466e+00 3.24159473e-01 -4.94755238e-01
-2.13000149e-01 1.02319503e+00 7.51005292e-01 -8.02634478e-01
6.95045233e-01 1.45620540e-01 8.33742023e-02 -5.14462471e-01
-2.84012347e-01 -2.49869078e-02 3.80918324e-01 5.85715532e-01
1.42129570e-01 -1.36628404e-01 3.62282813e-01 1.47104159e-01
-2.42251500e-01 -2.98615754e-01 2.24845812e-01 9.04975593e-01
-3.88593584e-01 -1.04563844e+00 -7.52956152e-01 7.12303877e-01
-8.58622715e-02 5.38479425e-02 -2.99201757e-01 9.05546069e-01
1.62367508e-01 6.35682344e-01 4.11166489e-01 -3.53896499e-01
5.84584832e-01 -2.97112703e-01 3.53907734e-01 -7.08895683e-01
-3.99177074e-01 7.30047897e-02 -5.31280220e-01 -1.07607865e+00
-3.09977710e-01 -9.30197358e-01 -1.21839881e+00 -2.33856797e-01
-4.26336855e-01 -7.39289373e-02 5.66354096e-01 8.94249737e-01
7.35395670e-01 2.88136095e-01 6.90050244e-01 -1.48740232e+00
-4.63414282e-01 -6.21251523e-01 -4.52395737e-01 6.39455259e-01
6.40427947e-01 -8.57940197e-01 -4.96304512e-01 -4.43754196e-01]
|
[7.7149553298950195, -3.0300865173339844]
|
93739500-759d-4fe8-88f5-fc77d05cdf38
|
on-exploring-and-improving-robustness-of
|
2110.057
| null |
https://arxiv.org/abs/2110.05700v1
|
https://arxiv.org/pdf/2110.05700v1.pdf
|
On Exploring and Improving Robustness of Scene Text Detection Models
|
It is crucial to understand the robustness of text detection models with regard to extensive corruptions, since scene text detection techniques have many practical applications. For systematically exploring this problem, we propose two datasets from which to evaluate scene text detection models: ICDAR2015-C (IC15-C) and CTW1500-C (CTW-C). Our study extends the investigation of the performance and robustness of the proposed region proposal, regression and segmentation-based scene text detection frameworks. Furthermore, we perform a robustness analysis of six key components: pre-training data, backbone, feature fusion module, multi-scale predictions, representation of text instances and loss function. Finally, we present a simple yet effective data-based method to destroy the smoothness of text regions by merging background and foreground, which can significantly increase the robustness of different text detection networks. We hope that this study will provide valid data points as well as experience for future research. Benchmark, code and data will be made available at \url{https://github.com/wushilian/robust-scene-text-detection-benchmark}.
|
['Zengfu Wang', 'Kewei Wang', 'Yongrui Li', 'Wei Zhai', 'Shilian Wu']
|
2021-10-12
| null | null | null | null |
['scene-text-detection']
|
['computer-vision']
|
[ 4.27621692e-01 -4.29031193e-01 1.86111644e-01 -3.10295284e-01
-6.53305471e-01 -4.39711362e-01 8.86469305e-01 1.01262592e-01
-2.33098850e-01 3.50368261e-01 2.38572627e-01 -2.57938296e-01
2.93932855e-01 -7.36585617e-01 -5.18734574e-01 -7.99921870e-01
3.69950056e-01 1.04871228e-01 7.83735216e-01 1.31633468e-02
5.13821244e-01 4.75236475e-01 -1.49925578e+00 5.00928879e-01
8.48367572e-01 7.79088914e-01 2.86248267e-01 8.72605383e-01
-2.08058441e-03 6.73275530e-01 -5.40657461e-01 -4.43327546e-01
3.23120534e-01 -1.53152943e-01 -4.97070223e-01 1.98642910e-01
4.90289032e-01 -3.11998516e-01 -5.69146752e-01 1.01333916e+00
7.69279838e-01 1.43434361e-01 8.24049115e-01 -1.12273347e+00
-3.33286911e-01 4.74497646e-01 -8.55966389e-01 2.75442183e-01
3.30266625e-01 3.15389067e-01 6.51196063e-01 -9.70114708e-01
6.01088345e-01 1.29532838e+00 7.98308671e-01 4.09017444e-01
-6.80100918e-01 -7.49002218e-01 2.07170963e-01 2.31184751e-01
-1.34867382e+00 -4.96887326e-01 6.06942058e-01 -3.89030635e-01
5.39728343e-01 6.11724317e-01 1.73206314e-01 1.36153662e+00
2.09404692e-01 1.18341482e+00 9.97096479e-01 -6.00191832e-01
1.28203835e-02 2.03299224e-01 2.44773224e-01 7.31381118e-01
5.05019665e-01 -8.67494717e-02 -5.65987706e-01 7.18122870e-02
5.37651122e-01 -6.15026616e-02 -3.00586641e-01 2.22931895e-03
-1.19367385e+00 6.88632309e-01 2.47660056e-01 3.17746222e-01
1.19998112e-01 4.82355431e-02 4.95865673e-01 -7.48620629e-02
7.26613224e-01 -1.93771720e-01 -2.12353095e-01 2.08444402e-01
-1.22327828e+00 9.58492383e-02 4.67721552e-01 1.01785743e+00
2.61246622e-01 1.74428359e-01 -5.60421884e-01 9.04107511e-01
6.15125120e-01 9.50555861e-01 4.05961663e-01 -4.62648869e-01
7.72384882e-01 6.35153830e-01 -1.91497386e-01 -1.10437930e+00
-4.16732490e-01 -9.19892564e-02 -1.09078217e+00 -1.84798062e-01
2.98063040e-01 -2.08535030e-01 -1.04058015e+00 9.48674977e-01
4.95429039e-01 3.37254584e-01 -1.59769893e-01 7.53376961e-01
1.18480074e+00 5.83546698e-01 4.05566022e-02 5.48959486e-02
1.33969581e+00 -1.02351904e+00 -7.05787003e-01 -1.26698881e-01
7.18595326e-01 -1.16718507e+00 1.01204085e+00 3.43178511e-01
-8.86852920e-01 -5.49308360e-01 -8.23857307e-01 -1.23689100e-01
-5.61835945e-01 6.42039120e-01 2.90068388e-01 8.94147933e-01
-8.56711268e-01 1.37120768e-01 -7.91363358e-01 -9.62744296e-01
4.55541760e-01 1.12395339e-01 -1.93846244e-02 -8.28922987e-02
-8.61283123e-01 6.26122832e-01 5.04425883e-01 2.10686952e-01
-9.37227905e-01 -3.16261083e-01 -6.55228794e-01 -2.70605505e-01
3.60115230e-01 -4.42441404e-01 7.74535537e-01 -6.56788588e-01
-1.06441224e+00 1.04048610e+00 -2.50257075e-01 -4.74293858e-01
8.70396852e-01 -2.25380436e-01 -4.16262031e-01 1.87811121e-01
1.16837934e-01 5.84267020e-01 1.03558993e+00 -1.34852827e+00
-7.40249038e-01 -4.10941392e-01 -6.02802336e-01 1.55695498e-01
-3.07203293e-01 4.17524457e-01 -9.32545543e-01 -1.06911790e+00
1.23658597e-01 -7.28408575e-01 4.98470366e-02 2.63295360e-02
-1.02525806e+00 -9.06111449e-02 1.15531003e+00 -8.23513985e-01
1.20665240e+00 -2.08926105e+00 -4.34397906e-01 8.26629549e-02
6.43028095e-02 3.02311748e-01 -2.05566525e-01 4.34381604e-01
4.62864228e-02 2.98785359e-01 -2.02158153e-01 -5.28741658e-01
-1.03264287e-01 -2.93264478e-01 -4.60668713e-01 8.51857841e-01
6.90856725e-02 8.00994754e-01 -2.04743594e-01 -9.14211154e-01
8.69099915e-01 5.10376751e-01 -1.46111622e-01 -8.85452852e-02
-1.49524212e-01 9.84053016e-02 -6.41802371e-01 9.69283879e-01
9.17792797e-01 -1.26818955e-01 -2.42771298e-01 -2.79822111e-01
-2.57243235e-02 -2.43995234e-01 -1.48056161e+00 1.10640144e+00
1.35240123e-01 1.04655051e+00 5.48416562e-02 -7.61358440e-01
8.74165237e-01 1.08847037e-01 3.13796520e-01 -7.75145173e-01
4.63984966e-01 -2.25807443e-01 -4.79577333e-01 -5.16847730e-01
8.29881489e-01 3.70470673e-01 2.19471827e-01 1.66802764e-01
-2.77567178e-01 -1.70561045e-01 3.17383945e-01 3.64185810e-01
9.53439057e-01 6.80421218e-02 6.19078949e-02 -2.25529730e-01
6.97590172e-01 3.15953046e-02 3.20019931e-01 1.01774263e+00
-4.70135838e-01 8.90042007e-01 2.63925374e-01 -2.06855536e-01
-9.57804084e-01 -9.17364359e-01 -4.83356386e-01 1.05798340e+00
3.47305626e-01 -2.88716674e-01 -9.00271475e-01 -7.11870015e-01
2.32813656e-02 7.86455154e-01 -7.08113253e-01 6.13616668e-02
-4.97668862e-01 -1.17009795e+00 9.81037378e-01 4.81633842e-01
9.15678859e-01 -9.67158794e-01 -2.45820716e-01 -3.23938936e-01
-4.17873800e-01 -1.43358803e+00 -3.52293015e-01 1.35336936e-01
-9.47323263e-01 -1.25533783e+00 -7.28343070e-01 -5.97988844e-01
6.31840587e-01 6.87166154e-01 7.71926939e-01 4.07215983e-01
-6.41311765e-01 5.97430885e-01 -6.48953378e-01 -4.46378559e-01
-4.53215301e-01 -1.94148213e-01 -2.06398174e-01 -3.46732251e-02
3.45454425e-01 2.45969996e-01 -5.97483456e-01 6.58393204e-01
-9.66488719e-01 1.52151451e-01 4.41915333e-01 3.71820003e-01
4.53942776e-01 2.86382943e-01 -5.62581532e-02 -7.98048913e-01
4.46581036e-01 -2.96913534e-01 -5.79235256e-01 4.20313656e-01
-3.44838351e-01 -5.23345411e-01 3.54255438e-01 -1.85596019e-01
-1.32127345e+00 2.45274261e-01 -4.13732938e-02 -2.87329406e-01
-5.76427877e-01 -7.89844990e-02 -1.30529493e-01 5.23033785e-03
7.79541492e-01 5.29230356e-01 -5.79495668e-01 -3.64599168e-01
3.22549254e-01 9.31967735e-01 4.76528674e-01 -4.00288433e-01
1.06998646e+00 8.71152163e-01 -1.30060300e-01 -1.43550014e+00
-5.44328451e-01 -9.85103726e-01 -8.00370276e-01 -2.89592683e-01
9.83607590e-01 -1.10127199e+00 -2.50598729e-01 9.26787496e-01
-1.06984246e+00 -5.06467938e-01 1.54581472e-01 1.07633814e-01
-2.31449470e-01 9.52301443e-01 -5.80002069e-01 -9.78830576e-01
-5.84753811e-01 -9.06589389e-01 1.58214271e+00 1.34906724e-01
2.47933671e-01 -1.10796225e+00 -1.75132558e-01 7.37164080e-01
8.07461590e-02 1.80843994e-01 5.14346063e-01 -7.15718567e-01
-5.13823628e-01 -2.49699578e-01 -6.00093842e-01 9.96819958e-02
-1.40421763e-01 5.14833868e-01 -1.27086747e+00 -3.39008480e-01
-1.89137325e-01 -8.12584534e-02 1.30159402e+00 6.28222227e-01
1.20218515e+00 8.01561847e-02 -5.03622055e-01 6.46284103e-01
1.43293512e+00 2.60270573e-02 8.79249692e-01 4.74099696e-01
8.94474506e-01 5.21878719e-01 7.01835513e-01 5.89860857e-01
2.48649999e-01 4.64110374e-01 4.26919013e-01 -3.25361043e-01
-4.40273196e-01 5.14137112e-02 3.73820424e-01 4.93226737e-01
8.17590132e-02 -8.77230585e-01 -1.01756859e+00 2.59168774e-01
-1.79594231e+00 -8.58770072e-01 -8.36513579e-01 2.04186869e+00
3.44389617e-01 5.19296974e-02 1.11468613e-01 3.64169747e-01
1.05815279e+00 1.72802344e-01 -3.71759266e-01 1.58707649e-01
-4.74869400e-01 -2.27802128e-01 6.44372165e-01 2.42304191e-01
-1.58617771e+00 1.36749434e+00 5.95687914e+00 1.23120344e+00
-1.06757617e+00 1.10760987e-01 7.63437092e-01 -4.80485687e-05
2.61425078e-01 -3.14046979e-01 -1.14866233e+00 5.99703729e-01
6.60238683e-01 2.20922515e-01 9.02852044e-02 5.95283866e-01
6.52074575e-01 -4.84601498e-01 -6.94830298e-01 9.08903420e-01
3.98623049e-01 -1.13361919e+00 1.34117424e-01 -2.50871480e-01
7.56694317e-01 3.14263046e-01 -5.38063794e-02 2.02765897e-01
2.39855319e-01 -7.85058320e-01 6.83843374e-01 4.31904644e-01
6.16213799e-01 -2.58847833e-01 7.29632795e-01 2.36405239e-01
-1.37897539e+00 -5.04953507e-03 -5.15965462e-01 5.22274077e-01
-1.28893629e-01 7.64975131e-01 -7.87542820e-01 7.07141817e-01
9.49824333e-01 8.21161985e-01 -1.23721933e+00 1.19916904e+00
-1.02732755e-01 9.41769183e-01 -3.73852819e-01 -2.13867828e-01
-6.14220351e-02 1.31693287e-02 5.06729782e-01 1.76652813e+00
1.84823975e-01 -2.38773629e-01 1.29548430e-01 7.58535922e-01
5.61027462e-03 3.21412116e-01 -6.15935624e-01 2.83812821e-01
3.15896094e-01 1.32474244e+00 -1.46738374e+00 -3.50192994e-01
-4.30956006e-01 9.80791390e-01 -2.81066835e-01 4.71184611e-01
-1.04763401e+00 -2.13621721e-01 -2.59017739e-02 1.02733910e-01
1.29026532e-01 -1.34551555e-01 -7.27754295e-01 -1.30025172e+00
-2.28187982e-02 -8.84641469e-01 4.34636354e-01 -1.00512171e+00
-1.02894783e+00 1.05427757e-01 -1.14032373e-01 -1.13496339e+00
4.79476452e-01 -7.88814425e-01 -8.81105125e-01 4.62841749e-01
-1.47683084e+00 -1.45029557e+00 -7.18319118e-01 8.91576052e-01
1.06887865e+00 -2.11670652e-01 2.01564997e-01 2.54281372e-01
-1.16920209e+00 6.48642421e-01 3.71820331e-01 4.48271126e-01
8.96433711e-01 -9.74856317e-01 4.99704480e-01 1.36658108e+00
1.03271835e-01 1.42117947e-01 6.44043565e-01 -9.14607882e-01
-1.35578811e+00 -1.47078371e+00 3.08717877e-01 -6.32063687e-01
5.11163354e-01 -4.64053720e-01 -7.69182265e-01 5.82768559e-01
8.70851800e-02 -1.42882273e-01 1.86176121e-01 -2.67426103e-01
4.56421264e-02 1.06707126e-01 -1.18458033e+00 8.16822171e-01
6.92038119e-01 -2.01286569e-01 -2.26638556e-01 5.80846727e-01
4.74994600e-01 -2.92368829e-01 -5.13598263e-01 4.09519017e-01
2.74933159e-01 -1.11632884e+00 9.31559801e-01 1.38825223e-01
2.83424705e-01 -2.36768618e-01 -4.58865970e-01 -5.13487458e-01
1.78381167e-02 -2.99921930e-01 -1.23353750e-02 1.49234819e+00
1.89617768e-01 -5.22054315e-01 9.44532394e-01 1.87775537e-01
-4.35434990e-02 -3.40786189e-01 -8.34669650e-01 -5.96913278e-01
1.73564315e-01 -8.23587418e-01 1.59210637e-01 9.43335652e-01
-4.62879807e-01 -7.99485669e-03 -3.74572754e-01 4.37291592e-01
6.66329980e-01 -3.08715343e-01 1.01872444e+00 -1.05233681e+00
2.11056069e-01 -5.54248393e-01 -2.58946091e-01 -9.27697778e-01
-1.41809210e-01 -7.19051421e-01 1.14292830e-01 -1.51789391e+00
4.44537669e-01 -2.44171098e-01 4.40498465e-04 4.06529486e-01
-4.10317510e-01 4.09440339e-01 3.27896506e-01 2.97751486e-01
-9.25552905e-01 5.51755667e-01 1.19305003e+00 -1.96294382e-01
-1.54955098e-02 1.71111003e-01 -2.88693577e-01 8.36540461e-01
1.07292533e+00 -4.40262258e-01 -5.37650436e-02 -2.39733890e-01
-1.26229003e-01 -2.96202183e-01 5.90595365e-01 -1.08104765e+00
3.97058845e-01 -1.25425994e-01 7.41893232e-01 -1.17627454e+00
1.81165323e-01 -7.45437145e-01 -3.57990623e-01 4.55178857e-01
-1.21372446e-01 -2.38589659e-01 3.29851031e-01 6.37655258e-01
1.97578788e-01 -4.24900025e-01 9.69082296e-01 1.27522305e-01
-7.88651764e-01 -1.25277718e-03 -5.31412184e-01 7.93770924e-02
1.11865747e+00 -5.33479452e-01 -6.97197735e-01 -1.31734297e-01
-2.01284170e-01 3.68103325e-01 5.42726636e-01 4.75145280e-01
7.14609265e-01 -7.22177744e-01 -8.81555974e-01 2.19583347e-01
1.43444583e-01 -5.10800704e-02 2.25917220e-01 9.49380457e-01
-8.27048838e-01 4.79668677e-01 1.77028686e-01 -9.45940793e-01
-1.70254755e+00 4.92511243e-01 2.90564835e-01 -1.67715639e-01
-8.08401048e-01 6.29507482e-01 3.00172180e-01 -4.98325735e-01
6.21930122e-01 -4.22372520e-01 -3.89165394e-02 -1.48793429e-01
2.78972208e-01 7.39512801e-01 2.87042353e-02 -6.18251324e-01
-3.45595330e-01 8.37115824e-01 -4.64234427e-02 1.66767836e-01
8.20933640e-01 -3.63925129e-01 1.45921260e-01 2.11802661e-01
6.16028786e-01 3.75026353e-02 -1.06045413e+00 -5.68289571e-02
-2.66322494e-02 -3.58582199e-01 1.31900877e-01 -9.47168648e-01
-1.00381422e+00 9.34723914e-01 9.63639140e-01 2.45881811e-01
1.19689417e+00 -1.14653237e-01 4.49905396e-01 4.30064142e-01
-1.13899760e-01 -1.26597726e+00 3.81904423e-01 4.86140400e-01
7.91314662e-01 -1.48239386e+00 4.06801105e-01 -6.55810535e-01
-6.38101816e-01 1.03531420e+00 6.34669065e-01 1.07204981e-01
5.78429759e-01 4.63852257e-01 1.07476048e-01 -1.35865599e-01
-5.81035912e-01 -4.06328529e-01 3.65318060e-01 7.03023195e-01
4.33579087e-01 -5.77975884e-02 5.63434735e-02 1.39467776e-01
5.52015193e-02 -3.78101110e-01 5.17001152e-01 7.26018071e-01
-6.86928868e-01 -7.08451152e-01 -8.49183977e-01 6.27671301e-01
-5.74613512e-01 -3.20538014e-01 -7.38740921e-01 1.02260387e+00
1.20327482e-02 1.23596561e+00 -2.70535022e-01 -1.89531833e-01
2.29632407e-01 -1.10648975e-01 1.13590337e-01 -4.22703862e-01
-5.28910935e-01 4.67589140e-01 -4.80760336e-02 -3.63283366e-01
-3.36837023e-01 -7.51931548e-01 -1.13936937e+00 -5.59306026e-01
-8.21664631e-01 -4.15461779e-01 8.30046117e-01 6.81288600e-01
9.75440219e-02 5.59443951e-01 4.92383361e-01 -7.86590159e-01
-1.18747286e-01 -1.14210093e+00 -5.79920590e-01 2.12309659e-01
-2.62847636e-02 -4.06485170e-01 -5.28651834e-01 4.43188131e-01]
|
[12.04699420928955, 2.2847037315368652]
|
52cf92f3-246a-4279-b168-6a08bb629f05
|
multi-modal-classifiers-for-open-vocabulary
|
2306.05493
| null |
https://arxiv.org/abs/2306.05493v1
|
https://arxiv.org/pdf/2306.05493v1.pdf
|
Multi-Modal Classifiers for Open-Vocabulary Object Detection
|
The goal of this paper is open-vocabulary object detection (OVOD) $\unicode{x2013}$ building a model that can detect objects beyond the set of categories seen at training, thus enabling the user to specify categories of interest at inference without the need for model retraining. We adopt a standard two-stage object detector architecture, and explore three ways for specifying novel categories: via language descriptions, via image exemplars, or via a combination of the two. We make three contributions: first, we prompt a large language model (LLM) to generate informative language descriptions for object classes, and construct powerful text-based classifiers; second, we employ a visual aggregator on image exemplars that can ingest any number of images as input, forming vision-based classifiers; and third, we provide a simple method to fuse information from language descriptions and image exemplars, yielding a multi-modal classifier. When evaluating on the challenging LVIS open-vocabulary benchmark we demonstrate that: (i) our text-based classifiers outperform all previous OVOD works; (ii) our vision-based classifiers perform as well as text-based classifiers in prior work; (iii) using multi-modal classifiers perform better than either modality alone; and finally, (iv) our text-based and multi-modal classifiers yield better performance than a fully-supervised detector.
|
['Andrew Zisserman', 'Weidi Xie', 'Prannay Kaul']
|
2023-06-08
| null | null | null | null |
['open-vocabulary-object-detection']
|
['computer-vision']
|
[ 1.19329147e-01 1.28356308e-01 -2.66138762e-01 -2.77297914e-01
-9.95845854e-01 -8.26904655e-01 9.48579431e-01 2.95191497e-01
-4.47426647e-01 2.88918465e-01 2.31134370e-01 -1.25511110e-01
1.70277759e-01 -5.48456073e-01 -8.75387490e-01 -3.71942729e-01
1.09754868e-01 5.45897424e-01 5.05068541e-01 2.10281555e-02
-1.12688681e-02 5.33810556e-01 -2.07403970e+00 5.72883248e-01
5.36872387e-01 1.31517196e+00 2.75508702e-01 6.07896924e-01
-2.79602945e-01 7.07212448e-01 -3.42000991e-01 -4.06234115e-01
2.29753286e-01 -3.63074355e-02 -7.88728237e-01 5.15368700e-01
9.09404337e-01 -3.32486540e-01 -3.77611220e-01 9.03769612e-01
2.65524328e-01 4.17846814e-02 1.12000501e+00 -1.27832997e+00
-9.58551943e-01 5.64202785e-01 -3.11846614e-01 1.21227935e-01
3.00545126e-01 4.95135754e-01 1.22194672e+00 -1.41127002e+00
8.88962328e-01 1.50184441e+00 5.15991986e-01 5.84746838e-01
-1.45860493e+00 -5.05426288e-01 3.95537615e-01 1.60109460e-01
-1.66515934e+00 -5.77803791e-01 5.47504306e-01 -7.86691308e-01
1.16615784e+00 9.15719718e-02 5.62484503e-01 1.07157135e+00
-7.15379491e-02 1.23313785e+00 9.88240421e-01 -6.83636427e-01
1.11796327e-01 5.50201058e-01 3.89192164e-01 9.02639389e-01
3.50206494e-01 9.57173854e-02 -4.01915282e-01 -1.68786287e-01
4.03371692e-01 3.38359028e-02 -7.59574696e-02 -5.39629400e-01
-1.39317644e+00 1.03246760e+00 6.78631485e-01 1.73055708e-01
-3.96699876e-01 1.88522235e-01 4.68360633e-01 1.06237374e-01
4.79366481e-01 2.97795027e-01 -2.33956620e-01 5.85177004e-01
-9.92959797e-01 2.22868919e-01 5.23974001e-01 1.14522529e+00
8.95936966e-01 -3.46440747e-02 -4.40294027e-01 9.90014136e-01
6.06059670e-01 7.72839904e-01 3.97172064e-01 -5.91244042e-01
2.86893219e-01 7.40182340e-01 -2.24560261e-01 -4.65447187e-01
-4.11499321e-01 -1.68809801e-01 -4.58685666e-01 9.72906649e-02
2.07213894e-01 1.52462944e-01 -1.31307483e+00 1.60640347e+00
1.58497959e-01 -2.15556756e-01 2.32235283e-01 5.98321557e-01
1.52394724e+00 5.72066605e-01 4.38713342e-01 -7.37591237e-02
1.67373395e+00 -9.21320081e-01 -3.64415497e-01 -4.63351130e-01
5.42559206e-01 -6.57711625e-01 1.03398550e+00 2.93576747e-01
-8.57868314e-01 -7.77581215e-01 -1.12685263e+00 -9.22586098e-02
-7.60954142e-01 2.10748717e-01 6.08793199e-01 4.16901380e-01
-1.20312154e+00 -2.26000145e-01 -4.86385822e-01 -5.72966516e-01
6.55130267e-01 1.74483672e-01 -3.84620488e-01 -2.42542654e-01
-8.57504308e-01 9.07750428e-01 8.00278544e-01 -3.80056620e-01
-1.21727419e+00 -3.45635295e-01 -1.18215489e+00 -2.87521213e-01
4.25890923e-01 -7.84314692e-01 1.06132603e+00 -1.07220805e+00
-6.49255574e-01 1.30132461e+00 -1.47813022e-01 -4.10032839e-01
2.97521442e-01 2.83811182e-01 -3.00144821e-01 5.02194941e-01
2.34814927e-01 1.53131723e+00 1.08106482e+00 -1.65794718e+00
-9.11744475e-01 -2.86258936e-01 1.13359682e-01 2.54130185e-01
-4.66912150e-01 -9.80724469e-02 -7.62977898e-01 -5.90584517e-01
1.62436157e-01 -9.04753625e-01 1.05757922e-01 3.34799886e-01
-6.49108350e-01 -8.25198710e-01 9.45412815e-01 -2.94771969e-01
8.59871209e-01 -2.22097230e+00 5.83753316e-03 -1.06307723e-01
5.21585643e-01 2.81899184e-01 -2.50072092e-01 9.71186906e-02
1.55098751e-01 8.79777893e-02 -1.89597346e-02 -5.05959630e-01
1.80031657e-01 2.46547148e-01 -6.18623555e-01 1.67573914e-01
3.01145166e-01 1.22437048e+00 -5.52919567e-01 -9.71795499e-01
3.63810301e-01 2.68724829e-01 -4.03570771e-01 -9.78522524e-02
-6.14112496e-01 -1.13374628e-01 -3.54827464e-01 9.21427131e-01
3.85182828e-01 -3.58857363e-01 -2.73847997e-01 -2.99414366e-01
8.77311379e-02 -1.98352024e-01 -1.18681908e+00 1.26199055e+00
-4.04604822e-01 7.73774803e-01 -9.54800770e-02 -9.94605720e-01
7.32364535e-01 2.14564696e-01 1.27238974e-01 -2.80714065e-01
9.85722989e-02 1.04671918e-01 -4.67094749e-01 -5.85219562e-01
2.25638911e-01 -7.85375535e-02 -2.40394011e-01 3.39547515e-01
6.04411602e-01 -3.44758630e-01 3.85035187e-01 4.68047976e-01
8.55987132e-01 -2.29207277e-02 3.99225056e-01 -3.28293107e-02
4.33904797e-01 2.28625312e-01 1.20344751e-01 1.19615889e+00
-2.57145166e-01 4.79102433e-01 2.01300964e-01 -4.22492862e-01
-9.28939581e-01 -1.30643392e+00 -5.68772376e-01 1.45906305e+00
1.19917750e-01 -2.32926071e-01 -3.61812919e-01 -8.17316711e-01
3.23295921e-01 7.71237254e-01 -5.84835768e-01 -8.58231634e-03
3.79961431e-02 -4.47045088e-01 7.03879952e-01 6.66237891e-01
4.03353095e-01 -1.10499609e+00 -2.92166799e-01 -1.37220755e-01
-3.63418370e-01 -1.27551877e+00 -4.65784252e-01 4.39733744e-01
-5.71925104e-01 -8.57179284e-01 -6.28116906e-01 -1.14169228e+00
6.40370548e-01 3.22850883e-01 1.10654593e+00 1.25716180e-01
-5.49209356e-01 1.01153040e+00 -3.50555450e-01 -7.61860192e-01
-5.18553138e-01 -2.82890826e-01 1.09315529e-01 1.06064588e-01
6.89096034e-01 7.20964894e-02 -2.64864326e-01 -2.65302928e-03
-8.51572514e-01 1.10003449e-01 7.85564721e-01 7.45952725e-01
6.81959152e-01 -1.98698580e-01 4.49748755e-01 -5.21542788e-01
2.89050579e-01 -3.87198448e-01 -5.26145160e-01 3.36713970e-01
-4.99405771e-01 5.01268767e-02 3.46512437e-01 -7.16320217e-01
-8.73981297e-01 4.89582628e-01 3.70534584e-02 -6.88468456e-01
-4.93853062e-01 3.24177206e-01 -9.28361863e-02 -5.26657030e-02
9.08723772e-01 5.64266682e-01 2.98930425e-03 -2.90569603e-01
7.58962870e-01 9.60543931e-01 5.09169221e-01 -3.56874913e-01
8.15859914e-01 6.90292716e-01 -2.75434852e-01 -1.14026570e+00
-1.02655196e+00 -8.10373008e-01 -9.27332580e-01 -2.61899263e-01
1.07985878e+00 -1.35222638e+00 -4.42494571e-01 2.15299115e-01
-1.08065903e+00 -9.01216567e-02 -2.55492717e-01 4.01208937e-01
-4.80749220e-01 4.01386023e-01 -4.65171158e-01 -7.65539825e-01
-2.49333054e-01 -1.17017090e+00 1.38902164e+00 1.58595264e-01
-9.21523347e-02 -1.05986691e+00 -4.12814230e-01 5.68066835e-01
-7.89524913e-02 -1.65302120e-03 7.64332712e-01 -1.06785059e+00
-7.86615133e-01 -3.82561088e-01 -4.82740819e-01 3.27354938e-01
-3.07932824e-01 -5.61179779e-02 -1.28268099e+00 -3.17569971e-01
-4.61021364e-01 -9.53509212e-01 1.35655344e+00 3.95440429e-01
9.13127780e-01 -1.63304284e-01 -7.06560552e-01 4.87196654e-01
1.29738939e+00 -2.02903315e-01 2.28350684e-01 2.04374939e-01
7.20888436e-01 5.14500022e-01 4.40389931e-01 1.85215265e-01
6.68126047e-01 6.60384774e-01 3.48239034e-01 -2.29457110e-01
-4.93299633e-01 -2.55990058e-01 3.60933989e-01 1.98158830e-01
3.00327986e-01 -1.95995688e-01 -9.40299392e-01 8.61056745e-01
-1.71863842e+00 -9.94738579e-01 -1.27275288e-01 1.89234364e+00
7.32374251e-01 2.53104001e-01 2.78327107e-01 -1.72602624e-01
5.81055760e-01 -9.84094217e-02 -6.41240835e-01 -9.18834880e-02
-3.30222487e-01 2.08710656e-02 4.29954767e-01 2.32990324e-01
-1.59426284e+00 1.05642211e+00 6.59570312e+00 8.63995671e-01
-9.54821885e-01 2.72596002e-01 3.62456262e-01 -1.02660380e-01
-2.33159944e-01 -1.00454956e-01 -1.22391963e+00 1.45787382e-02
4.72829461e-01 -4.23974879e-02 1.89018741e-01 1.17028391e+00
-2.43022531e-01 -1.28122553e-01 -1.34739041e+00 1.16338980e+00
5.93037009e-01 -1.33944249e+00 5.24475873e-01 1.09315805e-01
5.82282960e-01 3.82944465e-01 5.90951480e-02 5.88658154e-01
4.76299137e-01 -9.12506759e-01 1.15926099e+00 3.96052986e-01
8.51982236e-01 -2.11264387e-01 4.30572331e-01 4.65974599e-01
-1.28689086e+00 -4.33523685e-01 -3.45405817e-01 2.89419562e-01
-1.48465142e-01 3.05451363e-01 -8.98312211e-01 2.74931043e-01
6.78686678e-01 6.13440633e-01 -1.04423952e+00 9.19179380e-01
-2.04018340e-01 6.49996221e-01 -4.67151791e-01 -1.70367241e-01
2.97150582e-01 5.15929937e-01 5.87208152e-01 1.47779572e+00
-1.44189164e-01 -1.28258495e-02 6.70205355e-01 1.05803919e+00
-8.33969563e-02 8.17716569e-02 -7.71361768e-01 5.60204983e-02
5.22968709e-01 1.33844960e+00 -8.77513647e-01 -6.44227087e-01
-7.95213699e-01 7.91260540e-01 2.95716107e-01 4.21936214e-01
-5.05748391e-01 -2.09835723e-01 2.64256120e-01 -9.73549262e-02
6.08457923e-01 -4.02857475e-02 -2.63198346e-01 -1.23099768e+00
-1.98284648e-02 -6.31172776e-01 7.23201513e-01 -1.01750457e+00
-1.40882480e+00 5.72744489e-01 2.85557896e-01 -1.19929230e+00
-1.14579998e-01 -8.85853946e-01 -4.24379371e-02 6.79574251e-01
-1.40627098e+00 -1.82253563e+00 -2.98914135e-01 6.78452730e-01
8.01043093e-01 -4.34793472e-01 6.94328070e-01 1.47203505e-01
-1.42939627e-01 4.50529277e-01 -1.33310661e-01 3.74755323e-01
5.65084875e-01 -1.11506426e+00 1.13513656e-01 8.28958511e-01
6.56491041e-01 4.77781832e-01 5.22455573e-01 -6.79444015e-01
-1.36995280e+00 -1.32116437e+00 7.79642165e-01 -1.01362956e+00
5.71207583e-01 -6.25048935e-01 -7.37436771e-01 8.78138781e-01
-9.81675610e-02 2.86963791e-01 4.68278229e-01 4.51778732e-02
-6.75134778e-01 5.86851463e-02 -9.93446648e-01 4.93032336e-01
1.03648782e+00 -6.77151680e-01 -9.47284043e-01 5.83040893e-01
7.36097574e-01 -1.37609512e-01 -5.76173723e-01 3.49391490e-01
3.31588745e-01 -5.12162745e-01 1.25608957e+00 -4.98068213e-01
1.89443603e-01 -3.64031583e-01 -4.27942663e-01 -9.04465735e-01
-4.71167535e-01 9.19540077e-02 -6.97052255e-02 1.28164744e+00
4.98278052e-01 -5.09613633e-01 2.26117283e-01 3.23126882e-01
-1.42199859e-01 -4.02976215e-01 -9.20227945e-01 -7.42885590e-01
6.43671155e-02 -7.81495988e-01 8.42061937e-02 6.10697269e-01
-1.74930766e-01 6.43490911e-01 -4.41420153e-02 9.63304415e-02
7.02972293e-01 1.06510878e-01 7.57887661e-01 -1.23098993e+00
-1.29676312e-01 -6.28036082e-01 -5.58529615e-01 -1.16573477e+00
2.40618810e-01 -1.32439804e+00 1.91687286e-01 -1.76113272e+00
6.62862241e-01 -4.68432575e-01 -1.03247173e-01 9.45285916e-01
-1.46993428e-01 7.50108421e-01 4.16843235e-01 4.75170106e-01
-1.01982117e+00 3.04357857e-01 9.95182633e-01 -5.88348746e-01
-7.06229582e-02 -1.82979628e-01 -8.52755010e-01 8.04211020e-01
-1.26191555e-02 -2.00272813e-01 -3.21666241e-01 -2.25458056e-01
-7.37100169e-02 -2.52901226e-01 8.48918796e-01 -8.29877853e-01
2.15705261e-01 8.08506310e-02 6.75885618e-01 -8.61785650e-01
4.27105814e-01 -6.75269485e-01 -3.55408788e-01 3.66824389e-01
-3.30530286e-01 -6.14981115e-01 2.61504143e-01 7.11315155e-01
-3.42377461e-03 -2.93632865e-01 8.85584593e-01 -1.85005799e-01
-1.38461828e+00 2.95498222e-01 -6.41999066e-01 5.44822812e-02
1.13303316e+00 -4.12452817e-01 -2.86047727e-01 -2.12329075e-01
-9.04580951e-01 4.00343508e-01 4.59320009e-01 8.30832124e-01
6.29285872e-01 -1.32112372e+00 -7.35943556e-01 1.90132141e-01
9.70947921e-01 -1.95670173e-01 -3.55957216e-03 5.95006585e-01
-1.89456880e-01 5.69156229e-01 3.31440151e-01 -1.26613772e+00
-1.30858612e+00 9.48663294e-01 2.57011145e-01 1.58096224e-01
-7.09732413e-01 9.83886838e-01 6.22010350e-01 -4.75859046e-01
5.05832076e-01 -1.14526570e-01 -3.32888663e-01 4.06584144e-01
5.55716991e-01 -2.71612853e-01 -6.24584779e-02 -1.02542996e+00
-5.36893427e-01 5.54405272e-01 -9.12870020e-02 -1.17498770e-01
1.05223179e+00 -2.64030278e-01 1.83764771e-01 6.69310510e-01
1.00627542e+00 -4.79704976e-01 -1.08040857e+00 -6.57717109e-01
-6.48889318e-02 -7.69737288e-02 1.55079782e-01 -9.07439172e-01
-4.80236232e-01 8.01771224e-01 7.70413816e-01 2.32426405e-01
1.02869654e+00 7.74995983e-01 3.50406736e-01 5.03816247e-01
2.12881014e-01 -1.07186770e+00 3.34467947e-01 5.29274762e-01
8.63367975e-01 -1.61426651e+00 -5.92286773e-02 -2.58917391e-01
-8.80882204e-01 1.06169820e+00 5.40257215e-01 2.14164719e-01
6.92740798e-01 2.49754451e-02 1.49475500e-01 -4.36110556e-01
-9.15195584e-01 -7.90900588e-01 8.92902434e-01 7.53214777e-01
1.21636890e-01 3.18119451e-02 1.03625111e-01 5.60287774e-01
5.72501794e-02 -1.88042819e-01 1.71603814e-01 7.55899966e-01
-7.51946509e-01 -5.35457134e-01 -6.15523636e-01 7.43917823e-01
-6.52945712e-02 -1.97504848e-01 -5.18671036e-01 8.35897863e-01
3.86385649e-01 1.17206109e+00 2.86921203e-01 -8.54535252e-02
2.60985583e-01 4.04866874e-01 4.39859033e-01 -1.03384590e+00
-1.45512730e-01 1.76245540e-01 3.91548574e-02 -2.71831602e-01
-2.41607934e-01 -8.23281646e-01 -1.06650472e+00 1.88822210e-01
-5.15602171e-01 -2.80311465e-01 6.60261452e-01 1.25385380e+00
1.34529680e-01 3.14016849e-01 3.02667469e-01 -1.09239495e+00
-4.40698743e-01 -1.00806451e+00 -4.53583270e-01 6.52976513e-01
5.32980204e-01 -8.98897171e-01 -3.73763651e-01 5.02727270e-01]
|
[9.900038719177246, 1.6387935876846313]
|
e94d31d4-e6b6-4ed7-aae5-3062f63a79bc
|
a-theoretical-justification-for-image
|
2302.01217
| null |
https://arxiv.org/abs/2302.01217v1
|
https://arxiv.org/pdf/2302.01217v1.pdf
|
A Theoretical Justification for Image Inpainting using Denoising Diffusion Probabilistic Models
|
We provide a theoretical justification for sample recovery using diffusion based image inpainting in a linear model setting. While most inpainting algorithms require retraining with each new mask, we prove that diffusion based inpainting generalizes well to unseen masks without retraining. We analyze a recently proposed popular diffusion based inpainting algorithm called RePaint (Lugmayr et al., 2022), and show that it has a bias due to misalignment that hampers sample recovery even in a two-state diffusion process. Motivated by our analysis, we propose a modified RePaint algorithm we call RePaint$^+$ that provably recovers the underlying true sample and enjoys a linear rate of convergence. It achieves this by rectifying the misalignment error present in drift and dispersion of the reverse process. To the best of our knowledge, this is the first linear convergence result for a diffusion based image inpainting algorithm.
|
['Sanjay Shakkottai', 'Constantine Caramanis', 'Advait Parulekar', 'Litu Rout']
|
2023-02-02
| null | null | null | null |
['image-inpainting']
|
['computer-vision']
|
[ 3.04971427e-01 1.49602056e-01 -4.09752637e-01 9.81127936e-03
-9.95531321e-01 -5.75820923e-01 4.87048477e-01 -2.43916973e-01
-5.09936333e-01 9.38576877e-01 1.06581107e-01 -6.52956516e-02
-7.90488049e-02 -3.18027943e-01 -1.01265681e+00 -8.27635586e-01
1.19073102e-02 4.41921443e-01 -1.07905986e-02 1.60202980e-01
3.49297166e-01 5.02595603e-01 -7.39930332e-01 6.75019398e-02
9.17455196e-01 6.92753673e-01 2.41978750e-01 1.16787851e+00
7.86947384e-02 1.04800141e+00 -6.34383142e-01 -2.38425285e-01
6.22082114e-01 -9.14693713e-01 -8.47980738e-01 5.90319276e-01
6.08665407e-01 -7.30534196e-01 -7.77156949e-01 1.05971313e+00
4.23692346e-01 4.70693745e-02 6.47915125e-01 -9.46011186e-01
-1.14842725e+00 5.33507407e-01 -9.83997762e-01 6.51623309e-01
1.48636624e-01 2.69788086e-01 5.98838627e-01 -7.46326208e-01
1.04804647e+00 1.08637321e+00 6.95584834e-01 9.10382032e-01
-1.43011141e+00 -3.94906968e-01 3.94131765e-02 -1.04940087e-01
-1.15643084e+00 -7.21189976e-01 8.34078193e-01 -3.75379324e-01
6.83762968e-01 4.77793902e-01 5.57860374e-01 9.68144178e-01
2.22726032e-01 1.18240499e+00 1.39479649e+00 -5.41878283e-01
2.96142191e-01 -1.60358801e-01 -1.89933419e-01 7.78235614e-01
1.63282782e-01 1.16337352e-01 -7.20454693e-01 -2.92510539e-01
1.07604730e+00 3.44574377e-02 -5.16091466e-01 -1.40538245e-01
-1.14049995e+00 7.18250930e-01 2.42309645e-01 1.87114879e-01
-4.25371468e-01 5.58633089e-01 -3.32062989e-02 7.78394461e-01
1.11357713e+00 2.64757156e-01 -1.34044755e-02 -1.50719061e-01
-1.57690620e+00 3.02958876e-01 7.33266354e-01 1.05195272e+00
6.85636401e-01 1.14946894e-01 -1.14308864e-01 7.23169804e-01
-1.22651260e-03 8.71591508e-01 7.09252894e-01 -1.54740477e+00
2.70006061e-01 -2.81301260e-01 3.69705439e-01 -7.05403924e-01
2.15401039e-01 -2.69029498e-01 -8.83782506e-01 2.47208893e-01
3.99260074e-01 -3.15022558e-01 -7.19848812e-01 1.83570361e+00
1.74339175e-01 5.45757890e-01 7.02262074e-02 7.42082775e-01
-4.34375294e-02 7.51469314e-01 -6.57578290e-01 -9.08658147e-01
6.00922287e-01 -1.15416276e+00 -9.80194330e-01 -3.04123312e-01
2.71482944e-01 -9.11889613e-01 5.68403184e-01 5.44696212e-01
-1.59005749e+00 -1.56353399e-01 -1.10317266e+00 -1.47089601e-01
4.99901831e-01 -3.13773304e-01 2.63337940e-01 5.51716924e-01
-1.29892468e+00 1.02205276e+00 -1.08227682e+00 -3.33188534e-01
4.77073401e-01 1.91791877e-01 -3.39169890e-01 -3.11184466e-01
-4.71432745e-01 7.65796423e-01 -1.17703684e-01 -3.07135861e-02
-1.01875460e+00 -6.63383543e-01 -3.65419120e-01 -5.40467739e-01
3.68875861e-01 -8.69192958e-01 1.32521582e+00 -1.24208832e+00
-1.58586907e+00 8.71680439e-01 -7.07270622e-01 -9.56433773e-01
1.10231352e+00 -5.17223477e-01 -5.08688018e-02 4.18994099e-01
2.27644816e-01 5.87068677e-01 1.66604471e+00 -1.32493579e+00
-2.13761181e-01 -3.02520126e-01 -4.43652093e-01 1.31846234e-01
-1.14449866e-01 -1.88518867e-01 -2.45577201e-01 -1.02667463e+00
1.09782770e-01 -1.00916696e+00 -3.11085552e-01 4.47449446e-01
-2.67347932e-01 3.58130336e-01 9.28617060e-01 -8.37137938e-01
1.07283938e+00 -2.07614326e+00 4.62520927e-01 -1.81477293e-01
5.47419906e-01 1.39006793e-01 -1.92452773e-01 6.45549774e-01
1.56824186e-01 2.11167261e-02 -8.62460017e-01 -9.14807618e-01
-3.48742306e-01 5.69582820e-01 -6.84558988e-01 9.88423109e-01
-4.25486034e-03 9.65140939e-01 -1.03276384e+00 -4.97762233e-01
-7.38123730e-02 1.79710031e-01 -5.74509084e-01 1.04809679e-01
-2.00713396e-01 5.48785329e-01 -3.49312313e-02 3.85108024e-01
1.04833472e+00 -1.64225638e-01 -5.89491650e-02 1.85647890e-01
4.83924672e-02 -5.38541734e-01 -1.01669061e+00 1.89985132e+00
-3.41767907e-01 1.11074555e+00 6.71976328e-01 -8.85198474e-01
6.11308455e-01 2.08243012e-01 5.87184668e-01 -2.05837071e-01
-1.16090156e-01 5.08429408e-01 -3.55889529e-01 -5.16864717e-01
7.74098814e-01 -6.65734410e-01 4.56673741e-01 8.38353872e-01
-1.51881456e-01 -5.42040229e-01 -2.06708778e-02 4.70836610e-01
1.22717655e+00 -9.19771940e-02 -1.50275975e-01 -2.41077602e-01
-5.45777194e-02 -1.60216466e-01 3.23907852e-01 1.13569283e+00
-3.90884250e-01 8.89042377e-01 1.68903515e-01 -1.03031546e-01
-1.33094561e+00 -1.05595696e+00 1.78632393e-01 3.93096536e-01
1.12110995e-01 -1.75607398e-01 -9.62838590e-01 -4.20469224e-01
2.95486469e-02 5.09430587e-01 -8.03832471e-01 -1.16041467e-01
-6.27107561e-01 -8.71342361e-01 5.68487704e-01 8.74731168e-02
7.16852129e-01 -8.22354794e-01 -3.73565346e-01 4.07960594e-01
-1.94163382e-01 -1.01828229e+00 -1.11717570e+00 -1.90057948e-01
-1.47158563e+00 -9.11191761e-01 -1.20871222e+00 -6.72267079e-01
6.91540599e-01 5.39026201e-01 7.95782149e-01 3.37193571e-02
3.62267084e-02 6.80857062e-01 -1.87809095e-01 -1.35340109e-01
-7.71275997e-01 -1.72785036e-02 -1.21544814e-02 1.89464688e-01
-2.47481644e-01 -6.76263630e-01 -7.77131081e-01 -7.35527789e-03
-1.33908367e+00 -1.32942215e-01 5.54756701e-01 8.66861939e-01
6.95098579e-01 7.26951510e-02 3.29623967e-01 -8.18249762e-01
1.02204263e+00 -4.13049817e-01 -4.53654349e-01 -1.25071466e-01
-8.46943021e-01 3.06820303e-01 3.83989900e-01 -6.98566198e-01
-8.27732265e-01 -1.33714050e-01 -2.55710874e-02 -8.29312623e-01
3.89419883e-01 1.33110970e-01 5.35532475e-01 -2.13959619e-01
7.44370878e-01 5.19732714e-01 5.04571497e-01 -5.06159484e-01
6.41998589e-01 4.47470874e-01 6.59473538e-01 -6.51757479e-01
9.37070191e-01 1.20867968e+00 9.96496435e-03 -1.00418663e+00
-8.31229270e-01 -2.89951533e-01 -3.36182684e-01 -2.26054519e-01
5.94564915e-01 -6.83377385e-01 -2.68163562e-01 8.05955648e-01
-1.21770084e+00 -7.38954902e-01 -9.21662509e-01 2.90182769e-01
-1.03022969e+00 7.25268006e-01 -1.02685797e+00 -9.71671164e-01
-3.43660533e-01 -8.68229270e-01 8.01594257e-01 -1.67399660e-01
2.77525950e-02 -1.05659199e+00 3.15326363e-01 4.55169491e-02
4.54638541e-01 2.03005731e-01 1.08284414e-01 4.43555340e-02
-6.97568059e-01 -9.92050581e-03 -1.84073206e-02 6.45044148e-01
2.49578670e-01 -1.42389446e-01 -6.49486840e-01 -5.74364781e-01
8.04666758e-01 -3.15161422e-02 1.25047290e+00 7.57926404e-01
7.25453258e-01 -7.19041407e-01 -2.86306143e-01 6.35059476e-01
1.39971280e+00 -5.56719005e-02 6.56143844e-01 2.70171285e-01
5.25941849e-01 2.13842779e-01 3.66468310e-01 4.09438610e-01
-2.62121916e-01 3.38107884e-01 1.81398317e-01 1.07075259e-01
-3.77644449e-01 -1.65085301e-01 4.70736116e-01 9.90847051e-01
4.21581268e-02 -2.46403053e-01 -2.61271447e-01 7.34945416e-01
-1.90883636e+00 -1.23147571e+00 -1.20545655e-01 2.19170403e+00
1.33149028e+00 2.59613874e-03 1.31483540e-01 -1.40047614e-02
7.38423467e-01 4.26977783e-01 -8.77447069e-01 -3.04795593e-01
-3.90591204e-01 3.85213047e-01 1.03764999e+00 1.14711249e+00
-7.21959710e-01 1.07245350e+00 7.22807217e+00 9.69441891e-01
-8.18288565e-01 5.55843830e-01 6.53360188e-01 -4.56178427e-01
-2.95900762e-01 2.05262899e-02 -3.73221427e-01 4.20492291e-01
8.34608972e-01 -1.95484862e-01 8.35223615e-01 4.06028956e-01
2.91864455e-01 -3.07590127e-01 -8.71097744e-01 1.14913046e+00
2.70359844e-01 -1.54959476e+00 -1.83939546e-01 2.17319921e-01
1.20184004e+00 5.29686026e-02 2.97889709e-01 -5.31459868e-01
3.79729480e-01 -8.14156830e-01 9.21619594e-01 6.72698498e-01
7.15712130e-01 -5.69134951e-01 6.93114027e-02 4.81562018e-01
-5.17591715e-01 1.10507207e-02 -3.73975664e-01 -1.87310148e-02
6.52292848e-01 1.03829205e+00 -4.43983465e-01 2.87146538e-01
1.03432558e-01 9.37797904e-01 -4.89461869e-02 9.80917811e-01
-2.53016129e-02 7.89059162e-01 -3.88208747e-01 5.57599902e-01
8.83578658e-02 -3.95292222e-01 1.08779299e+00 9.21504021e-01
5.83195329e-01 8.20524693e-02 -1.88800395e-01 8.36399496e-01
-3.25392216e-01 -3.69308919e-01 -6.81960523e-01 -6.01779893e-02
2.78427571e-01 7.48225927e-01 -6.35580242e-01 -5.47562420e-01
-1.14038624e-02 1.92104578e+00 1.67979315e-01 5.62157691e-01
-6.51464581e-01 1.56283844e-02 4.51672286e-01 1.72997758e-01
5.69900215e-01 -6.93441093e-01 -1.67696670e-01 -1.46497512e+00
1.67760119e-01 -7.46969342e-01 5.37734898e-03 -7.25354075e-01
-1.37188995e+00 3.74118119e-01 -5.08289747e-02 -9.38263834e-01
-1.17674127e-01 -8.45397711e-02 -2.50658751e-01 8.16328347e-01
-1.39476383e+00 -7.56462514e-01 2.30760500e-01 4.60100859e-01
8.53645980e-01 2.35173911e-01 1.72150135e-01 6.90115690e-02
-3.49315584e-01 3.97039086e-01 5.73951364e-01 -3.17907035e-01
8.73210907e-01 -1.29393411e+00 4.48626816e-01 1.24564278e+00
1.93241641e-01 5.32053113e-01 1.17190754e+00 -7.95870006e-01
-1.55241823e+00 -9.51502800e-01 5.98859191e-01 -3.19636017e-01
6.65178537e-01 5.86588793e-02 -8.26896012e-01 1.08270681e+00
5.38904011e-01 -1.13732800e-01 1.56113133e-01 -6.40306175e-01
-2.19652101e-01 -1.00974515e-02 -1.39040160e+00 5.24805307e-01
1.15000641e+00 -6.48304582e-01 -3.55936646e-01 6.53999507e-01
6.20689332e-01 -5.86122334e-01 -6.99857473e-01 -1.74174681e-01
2.98665345e-01 -1.02260029e+00 9.91708994e-01 -2.37297416e-01
4.25398409e-01 -1.02166928e-01 -6.52715042e-02 -1.18651462e+00
-1.01752117e-01 -1.45382047e+00 -6.61627114e-01 9.51526105e-01
-3.33518013e-02 -7.14886785e-01 8.54455411e-01 4.53897208e-01
1.27868295e-01 -4.63027000e-01 -9.99890089e-01 -1.26253223e+00
4.87610817e-01 -3.72067273e-01 1.06766038e-01 8.42326403e-01
-3.35901797e-01 6.19047275e-03 -9.15088534e-01 2.25377455e-02
9.54657257e-01 3.00689377e-02 5.94481707e-01 -6.78590775e-01
-7.63380289e-01 -3.38535249e-01 1.48799300e-01 -1.83029068e+00
8.65718648e-02 -7.45149434e-01 7.58937150e-02 -1.26199102e+00
1.96837947e-01 -3.60337168e-01 3.30478162e-01 -1.60040706e-01
-1.14381053e-01 3.33651721e-01 2.61228204e-01 9.15214360e-01
-3.62714082e-01 5.53398669e-01 1.61680162e+00 -2.21886918e-01
-2.71603554e-01 -8.02579597e-02 -6.88371599e-01 3.70784193e-01
7.54390121e-01 -7.88343608e-01 -4.27010745e-01 -7.93266237e-01
1.06147200e-01 3.04461181e-01 2.85535336e-01 -8.11547637e-01
2.80230522e-01 -2.45419428e-01 1.94751546e-02 -2.39646569e-01
4.09461051e-01 -4.65827316e-01 1.94293693e-01 7.31262147e-01
-4.00229573e-01 2.32530087e-01 2.07890719e-02 9.90389705e-01
-5.09441532e-02 -5.03518879e-01 1.01480186e+00 -1.72684193e-01
-1.08072594e-01 5.11809945e-01 -5.42240381e-01 3.32086235e-01
8.52637231e-01 -1.30518153e-01 -2.33676091e-01 -7.83206701e-01
-7.99947381e-01 -3.16309422e-01 8.35478604e-01 -1.45866081e-01
6.70229852e-01 -1.15044367e+00 -7.63044775e-01 1.15006484e-01
-6.61511838e-01 1.67512879e-01 2.07846150e-01 1.00895488e+00
-7.60749280e-01 -2.52895683e-01 2.82848626e-01 -4.27400827e-01
-1.10436809e+00 7.49107301e-01 3.79805088e-01 -2.50305623e-01
-8.70317876e-01 1.00182211e+00 -2.55225480e-01 2.97308087e-01
-2.16054901e-01 -1.94443271e-01 8.29591274e-01 -2.79167682e-01
7.05379307e-01 5.48181891e-01 -2.99954563e-01 -4.06310230e-01
1.01496875e-02 5.51538646e-01 -3.48996103e-01 -9.97252762e-01
1.14188802e+00 -5.20735741e-01 -3.09230536e-01 4.70419765e-01
1.21309376e+00 4.27106529e-01 -1.81781423e+00 -2.89746016e-01
-3.02901149e-01 -7.15353310e-01 8.59979820e-03 -2.07861021e-01
-1.25866425e+00 6.05042458e-01 3.94817144e-01 5.01645207e-01
1.13028300e+00 5.10310195e-02 1.27639878e+00 1.96554646e-01
-5.12486286e-02 -1.05133355e+00 2.09076285e-01 2.49023125e-01
1.03217113e+00 -8.62656355e-01 1.96039468e-01 -2.06071258e-01
-2.73806304e-01 9.20782626e-01 -1.16520897e-01 -6.27989471e-01
6.45230532e-01 3.91982675e-01 -1.02065623e-01 4.23630029e-02
-6.90855682e-01 1.88408494e-01 -3.01153719e-01 6.90733850e-01
-1.86088588e-02 -1.58359200e-01 -4.89410639e-01 -2.94889539e-01
1.01211801e-01 3.13296407e-01 7.88233757e-01 1.26899743e+00
-5.08444011e-01 -1.31018162e+00 -5.37143469e-01 2.73764938e-01
-3.95516008e-01 -2.60189503e-01 -3.53530169e-01 5.05771756e-01
-3.42324108e-01 9.65294242e-01 -7.80670121e-02 8.71840492e-02
-1.67325795e-01 -1.61309093e-01 1.00903654e+00 -1.79813668e-01
-2.74319381e-01 3.20989639e-01 -2.93872923e-01 -3.52024466e-01
-8.17518950e-01 -8.49997461e-01 -8.92931402e-01 -7.43709087e-01
-2.06621066e-01 4.98098098e-02 3.14510316e-01 8.77632499e-01
5.51449060e-01 1.23086900e-01 6.32100523e-01 -7.64233172e-01
-7.03477383e-01 -9.12826240e-01 -8.52928221e-01 2.76158571e-01
8.97352874e-01 -1.06448807e-01 -6.87109172e-01 3.63914579e-01]
|
[11.71802043914795, -2.3448965549468994]
|
d350a39f-35d2-46c0-8e4a-af8d508678ca
|
a-review-of-machine-learning-approaches
|
2206.01728
| null |
https://arxiv.org/abs/2206.01728v1
|
https://arxiv.org/pdf/2206.01728v1.pdf
|
A review of machine learning approaches, challenges and prospects for computational tumor pathology
|
Computational pathology is part of precision oncology medicine. The integration of high-throughput data including genomics, transcriptomics, proteomics, metabolomics, pathomics, and radiomics into clinical practice improves cancer treatment plans, treatment cycles, and cure rates, and helps doctors open up innovative approaches to patient prognosis. In the past decade, rapid advances in artificial intelligence, chip design and manufacturing, and mobile computing have facilitated research in computational pathology and have the potential to provide better-integrated solutions for whole-slide images, multi-omics data, and clinical informatics. However, tumor computational pathology now brings some challenges to the application of tumour screening, diagnosis and prognosis in terms of data integration, hardware processing, network sharing bandwidth and machine learning technology. This review investigates image preprocessing methods in computational pathology from a pathological and technical perspective, machine learning-based methods, and applications of computational pathology in breast, colon, prostate, lung, and various tumour disease scenarios. Finally, the challenges and prospects of machine learning in computational pathology applications are discussed.
|
['Shaoliang Peng', 'Zhichao Feng', 'Liangrui Pan']
|
2022-05-31
| null | null | null | null |
['data-integration']
|
['knowledge-base']
|
[ 4.25796568e-01 -3.12271863e-01 -8.34981441e-01 1.05045021e-01
-7.67146587e-01 -2.24814117e-01 -2.85685109e-03 1.01881289e+00
-1.34050071e-01 5.27458549e-01 1.43888354e-01 -5.49750566e-01
-3.66715938e-01 -6.89024091e-01 2.39497751e-01 -1.11583686e+00
-4.37062867e-02 7.51493573e-01 -3.20299000e-01 2.49598294e-01
-1.15468621e-01 9.74320292e-01 -6.69045687e-01 4.90791351e-01
5.61876357e-01 6.64822936e-01 1.59256652e-01 1.13072443e+00
-3.81437212e-01 5.20832360e-01 -7.88694993e-02 -1.99030519e-01
-3.29433322e-01 -1.75990924e-01 -5.48459411e-01 7.28794560e-03
-3.95707607e-01 2.61934549e-01 -1.61122948e-01 8.92033219e-01
7.92524457e-01 -7.09238708e-01 3.01387161e-01 -9.13607240e-01
-3.03287834e-01 2.91859955e-01 -6.62705123e-01 1.00882433e-01
1.08588427e-01 3.01111281e-01 1.87905997e-01 -4.77449566e-01
8.36525381e-01 8.20744932e-01 8.27161551e-01 5.23019075e-01
-9.08735812e-01 -8.94296989e-02 -7.22777724e-01 8.97094086e-02
-9.71595943e-01 -3.68741393e-01 1.31622821e-01 -3.93981874e-01
9.55923498e-01 5.91773093e-01 6.67580068e-01 4.06448215e-01
1.07895517e+00 4.43154305e-01 7.26216495e-01 -6.09513104e-01
3.38573128e-01 1.87766105e-01 5.26624583e-02 6.39644802e-01
4.88037467e-01 -1.89580753e-01 -5.41540384e-01 -4.00252253e-01
3.32909614e-01 6.62629068e-01 -7.97664225e-02 -4.79594395e-02
-1.43071449e+00 6.20977402e-01 2.17431918e-01 6.99173808e-01
-7.89083466e-02 6.94285482e-02 9.25858676e-01 -4.42572758e-02
2.33906731e-01 2.64916211e-01 -5.32172024e-01 1.42092288e-01
-5.50910354e-01 -3.08863550e-01 5.72802901e-01 3.82466346e-01
1.14436373e-01 -4.75157648e-01 5.64016476e-02 5.76040566e-01
3.32291335e-01 4.43634182e-01 1.10547054e+00 -4.73704100e-01
-2.96400845e-01 7.01470792e-01 -3.88673902e-01 -6.85163498e-01
-1.28894627e+00 -3.19869399e-01 -1.10287774e+00 -4.17115510e-01
2.67323524e-01 1.49610773e-01 -7.44165659e-01 6.43143356e-01
6.73616290e-01 1.25596076e-01 1.68950543e-01 7.38915920e-01
7.51389503e-01 -4.55885977e-02 4.37217504e-01 -5.61011970e-01
1.85690761e+00 -6.73328757e-01 -7.51880765e-01 1.33283809e-01
1.38945949e+00 -8.33893239e-01 4.29175913e-01 1.81070358e-01
-7.12736726e-01 5.13610372e-04 -5.44901431e-01 -1.13588504e-01
-6.35926187e-01 4.84181583e-01 1.22368944e+00 1.00108659e+00
-9.16062772e-01 4.67903465e-01 -1.30098033e+00 -1.13415265e+00
8.16315889e-01 5.34674466e-01 -3.81048083e-01 -5.66762805e-01
-3.82801145e-01 7.11254179e-01 -2.54705157e-02 -1.34905562e-01
-3.79168630e-01 -1.25756216e+00 -5.40786624e-01 -6.89086840e-02
-3.93956065e-01 -1.05535793e+00 1.08119750e+00 -2.71436125e-01
-1.71248639e+00 1.06995678e+00 -3.19134325e-01 -2.64564872e-01
-1.09922372e-01 6.55215740e-01 -4.75999266e-01 2.60516405e-01
-1.03106685e-01 4.26341057e-01 -2.32858226e-01 6.17842562e-02
-5.94741464e-01 -8.57127905e-01 -1.10425365e+00 -9.75054055e-02
-2.89220899e-01 2.67054945e-01 -4.08822179e-01 5.48166186e-02
1.34275958e-01 -8.40097487e-01 -9.50293779e-01 4.42647040e-01
-1.78979516e-01 3.61889124e-01 7.19531357e-01 -4.40774769e-01
5.33541501e-01 -2.35263777e+00 -1.20811842e-01 3.31716001e-01
1.65191308e-01 2.65904278e-01 7.05796778e-02 2.75862187e-01
2.28004269e-02 4.17847663e-01 6.05083168e-01 1.32998079e-01
-5.03492713e-01 1.76819451e-02 6.60958350e-01 8.60208094e-01
-9.30822268e-02 1.15218318e+00 -8.62902045e-01 -5.54885328e-01
3.93594176e-01 5.25935233e-01 -1.47927940e-01 -3.46357644e-01
-7.56054595e-02 5.49812078e-01 -5.59535265e-01 1.39219427e+00
4.63723451e-01 -7.64045477e-01 1.00472212e+00 -1.03833772e-01
7.81864375e-02 -8.28552917e-02 -2.13943571e-01 1.64311993e+00
-1.98317498e-01 4.15392160e-01 3.65154862e-01 -7.38748014e-01
5.90537369e-01 4.47884440e-01 1.05812275e+00 -7.67181993e-01
3.70683104e-01 2.11961851e-01 2.27436081e-01 -7.63589144e-01
-1.15663096e-01 -2.13089541e-01 3.44004273e-01 1.62169799e-01
-3.58043432e-01 2.48867720e-02 1.24891274e-01 -4.96686064e-02
1.51622856e+00 -6.85003459e-01 4.96005952e-01 -4.54806276e-02
5.33841729e-01 5.35998344e-01 4.52426344e-01 4.35333997e-02
-7.37175107e-01 7.26180971e-02 5.10975659e-01 -4.16642696e-01
-9.24026549e-01 -7.28090882e-01 -1.51148051e-01 6.54629767e-01
-6.61738962e-02 -8.91485736e-02 -2.20423624e-01 -1.73165694e-01
3.63781333e-01 -3.62867564e-01 -3.02443713e-01 -1.24028184e-01
-2.29764342e-01 -1.56341505e+00 6.42727673e-01 3.96145135e-01
-1.52012020e-01 -3.78086597e-01 -2.70286322e-01 4.33307678e-01
3.23134482e-01 -1.10052502e+00 -6.85113072e-02 1.50718465e-01
-1.29419696e+00 -1.44877887e+00 -3.64842653e-01 -8.05577695e-01
7.71719694e-01 2.25567341e-01 5.57159543e-01 1.89862087e-01
-1.45770705e+00 1.94117978e-01 -1.92469954e-01 -9.44000304e-01
-5.31170607e-01 -2.32929334e-01 -7.10194139e-03 -3.35887820e-01
5.59934080e-01 -2.83382177e-01 -6.37893319e-01 -1.03565879e-01
-5.56612790e-01 -1.43677786e-01 1.23765385e+00 8.76005173e-01
1.08870697e+00 3.06134850e-01 5.62467217e-01 -1.11933291e+00
1.48016497e-01 -5.58006108e-01 -2.15658322e-01 2.69197553e-01
-5.19279659e-01 -3.53110194e-01 4.60083902e-01 -2.68601887e-02
-9.57951665e-01 4.10088569e-01 -2.44507045e-02 9.23511013e-02
-1.64295614e-01 6.77278459e-01 -1.41325831e-01 -5.44513583e-01
8.06112826e-01 -1.63276300e-01 8.33180726e-01 -1.41651675e-01
-1.06856860e-02 4.95954275e-01 3.22465986e-01 2.17263043e-01
-3.25172730e-02 8.44499469e-01 6.20899022e-01 -8.87440920e-01
-3.36117804e-01 -8.69655609e-01 -3.12125087e-01 -6.47916198e-02
6.99081004e-01 -9.03370261e-01 -9.64045286e-01 3.91591102e-01
-6.18061185e-01 -1.21985346e-01 -1.46742269e-01 8.03933859e-01
-1.80226132e-01 1.41905576e-01 -1.28953195e+00 -3.23901772e-01
-8.94368052e-01 -9.99688923e-01 1.02381492e+00 3.94766480e-01
-3.42244864e-01 -1.15492296e+00 2.34347984e-01 3.91398340e-01
4.67630357e-01 6.02304339e-01 1.39074767e+00 -4.05164301e-01
-2.90045261e-01 -6.76616490e-01 -2.16005430e-01 -5.47692120e-01
3.48708361e-01 3.96005571e-01 -8.12150002e-01 2.46021487e-02
-2.59886175e-01 -2.44450453e-03 3.83197337e-01 7.95791984e-01
1.24230933e+00 1.70100063e-01 -1.51237476e+00 5.69458067e-01
1.59212244e+00 1.69854760e-01 5.56785583e-01 -8.49262923e-02
2.54373759e-01 4.38460529e-01 6.43665254e-01 3.39121401e-01
-1.13709122e-01 -4.18955721e-02 1.94094539e-01 -2.67345965e-01
-1.78566903e-01 3.69487643e-01 -2.13585049e-01 7.10190535e-01
4.69665021e-01 -2.32886225e-01 -1.24762666e+00 6.48184836e-01
-1.42835009e+00 -6.47801638e-01 -4.43241984e-01 1.70500088e+00
8.28256190e-01 -3.48427415e-01 -1.59082308e-01 1.06138721e-01
7.14981318e-01 -8.48098040e-01 -5.36769271e-01 -4.43894863e-01
1.06717817e-01 2.41651967e-01 7.70949066e-01 -5.36770150e-02
-5.95575154e-01 4.01040673e-01 7.35510349e+00 7.65044272e-01
-1.46535492e+00 2.86894262e-01 1.02814865e+00 -2.79676288e-01
1.24059115e-02 -3.89860660e-01 -6.79401636e-01 2.90134102e-01
1.28030849e+00 -1.55015513e-01 -1.50489718e-01 8.03891182e-01
6.22940898e-01 -5.75306118e-01 -1.18213379e+00 9.67482865e-01
-3.20989966e-01 -2.02903223e+00 -3.70751798e-01 4.53545600e-01
4.54010665e-01 2.70142019e-01 6.97720721e-02 -3.98945212e-01
2.11946517e-02 -9.72059608e-01 -4.79920387e-01 5.59125841e-01
9.30817664e-01 -4.82939571e-01 1.26165295e+00 1.53193250e-01
-6.45318806e-01 -6.59944043e-02 -3.12314928e-01 -1.52898962e-02
-1.82229206e-01 1.28248572e+00 -1.56700981e+00 6.08461261e-01
4.60442007e-01 6.01762056e-01 -4.94722068e-01 1.12600482e+00
5.57480514e-01 2.56236225e-01 -1.40464261e-01 -3.64551216e-01
-4.65267062e-01 2.50030220e-01 -1.29513383e-01 1.34381104e+00
2.69023001e-01 1.63073301e-01 1.71221778e-01 2.36668978e-02
7.19744340e-02 2.11825281e-01 -8.03654268e-02 -5.49017966e-01
4.19793874e-01 1.87464237e+00 -1.28849840e+00 -1.61208183e-01
-4.57650959e-01 3.71115118e-01 -1.37315348e-01 -2.08522230e-01
-4.10785913e-01 -4.05238606e-02 7.92811692e-01 2.21651196e-01
-4.88128453e-01 2.02165563e-02 -8.29998076e-01 -5.53862214e-01
-7.71104991e-01 -7.77670324e-01 6.93026721e-01 -3.16200525e-01
-1.18090570e+00 -3.19369882e-01 -8.49418223e-01 -7.69151151e-01
1.92406729e-01 -1.01841307e+00 -8.10220003e-01 5.60905635e-01
-1.49486053e+00 -8.85462344e-01 -3.51296544e-01 1.22974567e-01
7.50474110e-02 -2.68885661e-02 1.27572489e+00 3.82717341e-01
-8.43034387e-01 3.97112489e-01 6.41674221e-01 -1.22203670e-01
8.48321676e-01 -7.25005507e-01 -2.63473123e-01 6.82325512e-02
-7.70730674e-01 8.50669742e-02 2.93686002e-01 -6.66201651e-01
-2.23027921e+00 -1.31179428e+00 7.74323404e-01 -2.46835873e-01
6.17312014e-01 -2.71108598e-02 -3.01780522e-01 3.12235206e-01
-2.52686948e-01 3.21293920e-01 1.73257172e+00 -1.59132376e-01
3.08513612e-01 -2.48942643e-01 -1.67988980e+00 4.84857678e-01
2.20033050e-01 -3.58392805e-01 6.79357290e-01 1.05598092e+00
3.29509944e-01 -5.91206491e-01 -1.32501364e+00 2.47823611e-01
7.52538562e-01 -3.02782416e-01 7.53787220e-01 -4.81399983e-01
1.23341649e-03 -2.63193578e-01 3.58687252e-01 -9.03542936e-01
-4.01678026e-01 -3.26935321e-01 2.07694948e-01 7.07959652e-01
3.95034492e-01 -4.88026530e-01 1.35386980e+00 6.43150508e-01
-2.49414340e-01 -1.05186439e+00 -7.27318287e-01 -2.74759203e-01
-9.83440280e-02 -1.11888260e-01 2.89743036e-01 7.11166799e-01
9.69778001e-01 -1.40737042e-01 4.67436850e-01 -3.15545034e-03
3.43952000e-01 -1.24877103e-01 5.73693514e-01 -1.06537247e+00
-2.69204617e-01 -4.89713520e-01 -9.28281784e-01 2.01843560e-01
-4.20179814e-01 -8.60281587e-01 -3.98474216e-01 -1.51043093e+00
3.96798998e-01 -4.54532206e-01 -4.10705842e-02 3.17445427e-01
-4.00545001e-02 2.79516757e-01 -3.90967965e-01 1.29280522e-01
-4.36292648e-01 -3.55988592e-01 1.27314103e+00 -3.71882826e-01
-7.98998848e-02 -3.45900893e-01 -8.01936924e-01 5.03271341e-01
9.41150546e-01 -3.44458550e-01 7.35946819e-02 -9.91401002e-02
2.09228337e-01 1.44915983e-01 9.05554593e-02 -9.59055305e-01
8.22717249e-01 -3.64618391e-01 1.04194510e+00 -4.14273977e-01
9.17775780e-02 -7.77168751e-01 6.54774964e-01 1.33734930e+00
1.26241103e-01 -3.67456913e-01 2.13239849e-01 5.06355405e-01
-1.48722112e-01 3.00650686e-01 8.76396894e-01 -3.67501974e-01
-3.43162507e-01 2.87230343e-01 -8.41284990e-01 -7.31634915e-01
1.50353003e+00 -3.26704234e-01 -8.02053869e-01 4.91372883e-01
-8.64679217e-01 5.04975855e-01 2.70334303e-01 -3.04120183e-01
3.16066325e-01 -9.69172835e-01 -5.67186773e-01 3.93905789e-02
7.57558271e-02 -1.20321669e-01 8.47627223e-01 1.67404199e+00
-1.04297173e+00 6.96168184e-01 -8.77497122e-02 -8.45775366e-01
-1.50169408e+00 7.52355456e-01 3.55190486e-01 -7.04784453e-01
-1.64528668e-01 6.81258142e-01 -1.84393436e-01 -3.27132910e-01
1.75694786e-02 -2.68845230e-01 1.08607948e-01 -3.26535791e-01
8.30041885e-01 5.34698069e-01 5.58227181e-01 2.03462802e-02
-5.86134315e-01 4.07545447e-01 -2.93441117e-01 5.95938742e-01
1.03922486e+00 -2.03459382e-01 -5.70230961e-01 1.28455669e-01
9.88322079e-01 -2.93945432e-01 -1.25505865e-01 7.44061470e-02
1.60847560e-01 -1.23746753e-01 8.82357806e-02 -9.51460660e-01
-8.88893962e-01 5.79880714e-01 6.27767146e-01 -8.36589485e-02
1.32743943e+00 5.72224259e-02 7.60388076e-01 1.68432444e-01
1.35703117e-01 -1.02405608e+00 -5.17816693e-02 1.31897688e-01
5.62326331e-03 -1.00236976e+00 4.60315794e-01 -1.03813159e+00
-1.04464874e-01 1.48920536e+00 3.41793627e-01 4.45082814e-01
9.16689932e-01 8.49075019e-01 2.46747762e-01 -2.49418378e-01
-1.02519023e+00 1.57488391e-01 -5.52662313e-01 7.20435262e-01
8.24954391e-01 3.40352595e-01 -3.88279140e-01 6.82154775e-01
1.06594861e-01 4.79027420e-01 5.52430093e-01 1.09605992e+00
-5.05225837e-01 -1.25734651e+00 -5.07996738e-01 7.33032525e-01
-9.15799499e-01 1.33756801e-01 -5.29931903e-01 5.74860513e-01
2.71317903e-02 6.87567592e-01 1.10092126e-01 2.57428885e-02
-6.68191388e-02 1.26226529e-01 3.65168810e-01 -6.21504903e-01
-5.41390121e-01 3.71407598e-01 1.39951017e-02 -4.14354861e-01
-3.92959446e-01 -6.02675498e-01 -1.58201575e+00 -5.73376954e-01
-5.02900541e-01 5.67371249e-02 1.31591237e+00 8.33657205e-01
8.49516034e-01 9.56848025e-01 3.88216257e-01 -3.61120105e-01
5.90177625e-02 -6.36348426e-01 -7.48792410e-01 -4.89189893e-01
-5.54407276e-02 2.04263642e-01 1.60366684e-01 2.83068776e-01]
|
[15.23124885559082, -3.026981830596924]
|
c361c5a7-f604-40b8-9dff-d94e8579aa76
|
self-relation-attention-and-temporal
|
2209.07629
| null |
https://arxiv.org/abs/2209.07629v2
|
https://arxiv.org/pdf/2209.07629v2.pdf
|
Self-Relation Attention and Temporal Awareness for Emotion Recognition via Vocal Burst
|
The technical report presents our emotion recognition pipeline for high-dimensional emotion task (A-VB High) in The ACII Affective Vocal Bursts (A-VB) 2022 Workshop \& Competition. Our proposed method contains three stages. Firstly, we extract the latent features from the raw audio signal and its Mel-spectrogram by self-supervised learning methods. Then, the features from the raw signal are fed to the self-relation attention and temporal awareness (SA-TA) module for learning the valuable information between these latent features. Finally, we concatenate all the features and utilize a fully-connected layer to predict each emotion's score. By empirical experiments, our proposed method achieves a mean concordance correlation coefficient (CCC) of 0.7295 on the test set, compared to 0.5686 on the baseline model. The code of our method is available at https://github.com/linhtd812/A-VB2022.
|
['Guee-Sang Lee', 'Minh-Cong Vo', 'Dang-Linh Trinh']
|
2022-09-15
| null | null | null | null |
['a-vb-high']
|
['speech']
|
[-1.12761199e-01 -7.73906931e-02 3.74384709e-02 -5.28723776e-01
-1.11192381e+00 -4.94234115e-01 1.16691709e-01 -6.64824545e-02
-2.15347052e-01 3.91077220e-01 3.86732787e-01 3.25945854e-01
1.79069713e-01 -3.80781777e-02 -1.40331432e-01 -6.37558818e-01
-3.36548865e-01 -3.27401310e-01 -2.87086189e-01 1.21982828e-01
-7.28258565e-02 1.69734895e-01 -1.62063324e+00 5.75556755e-01
5.46732962e-01 1.71383178e+00 -7.49536157e-02 9.53841150e-01
1.15523405e-01 7.12790430e-01 -6.00941241e-01 -1.82739332e-01
-1.40554294e-01 -6.26684308e-01 -7.96127915e-01 -8.61671343e-02
-2.19183296e-01 -1.53254140e-02 -1.67188928e-01 7.91802466e-01
6.84822142e-01 3.23212028e-01 4.75326389e-01 -1.33576381e+00
1.05191879e-01 4.95315135e-01 -3.90166998e-01 3.32110286e-01
3.94493759e-01 1.61477894e-01 1.26693892e+00 -1.24403334e+00
3.15776736e-01 8.59330893e-01 7.90640473e-01 5.93670070e-01
-9.10105050e-01 -9.55392957e-01 3.42192203e-02 2.94504881e-01
-1.26280963e+00 -7.25168169e-01 9.81888056e-01 -3.60526681e-01
1.14632821e+00 3.74098510e-01 8.69137526e-01 1.31736434e+00
1.48528740e-02 9.94432449e-01 1.17696476e+00 -1.51478738e-01
1.52839079e-01 2.19318777e-01 3.09621572e-01 7.22716510e-01
-7.89121747e-01 -2.52242163e-02 -1.01937115e+00 -2.80629527e-02
1.43178612e-01 -4.24279690e-01 -1.01091504e-01 3.99960667e-01
-8.40532243e-01 5.17470717e-01 1.65501878e-01 4.63710546e-01
-7.96907663e-01 1.42274112e-01 7.58687675e-01 3.93922359e-01
8.32106948e-01 2.94304997e-01 -6.11687243e-01 -8.36651325e-01
-7.54801571e-01 -2.16760427e-01 6.71856761e-01 4.74770427e-01
4.24093336e-01 3.16547871e-01 -1.11976460e-01 1.08133125e+00
2.17741787e-01 4.74120528e-02 6.59398317e-01 -9.41734552e-01
1.12052880e-01 1.13914423e-01 -1.00607179e-01 -1.03290427e+00
-6.04094923e-01 -5.03463030e-01 -7.32192516e-01 -2.70910949e-01
-1.25349820e-01 -7.46498942e-01 -2.81875312e-01 1.76285374e+00
2.04515919e-01 4.39772129e-01 1.28533527e-01 8.60582471e-01
1.02526867e+00 9.25675988e-01 1.56373501e-01 -7.05569923e-01
1.17253113e+00 -1.01813078e+00 -1.02638805e+00 -1.03833437e-01
4.66904521e-01 -8.05992723e-01 1.20217144e+00 6.66722000e-01
-1.14544940e+00 -6.73102975e-01 -9.56547260e-01 2.11870775e-01
-6.45162091e-02 4.37151939e-01 5.76583922e-01 3.05281013e-01
-1.01958168e+00 5.84442496e-01 -8.27011824e-01 -1.78882465e-01
2.45308191e-01 4.25711662e-01 -2.46717900e-01 6.84330583e-01
-1.42901087e+00 4.86100554e-01 4.89903502e-02 2.56116897e-01
-8.99199665e-01 -6.18144691e-01 -6.95836067e-01 4.24254350e-02
-8.25152323e-02 -7.20515996e-02 1.52076828e+00 -1.19917357e+00
-2.07553434e+00 6.89858854e-01 -3.27870607e-01 -2.03004852e-01
4.13427800e-02 -5.88587105e-01 -6.66925609e-01 3.15807492e-01
-1.37049586e-01 5.58474541e-01 9.14387107e-01 -9.85387146e-01
-5.67986786e-01 8.58233590e-03 -4.40525711e-01 2.69724637e-01
-5.57289600e-01 3.45496833e-01 -3.97140026e-01 -5.71409523e-01
-6.91126958e-02 -9.19463336e-01 2.38396861e-02 -5.05706489e-01
-3.37736458e-01 -5.26382208e-01 7.54149437e-01 -7.59677172e-01
1.55324209e+00 -2.71185851e+00 8.27059522e-02 2.50747979e-01
7.12858438e-02 1.27306461e-01 -3.29117209e-01 5.28080821e-01
-4.50717896e-01 -1.83864355e-01 8.18776991e-03 -6.91829026e-01
7.78482854e-02 -2.75171459e-01 -4.38713998e-01 3.52026701e-01
3.55306804e-01 8.39476168e-01 -8.51701677e-01 -3.17411751e-01
5.47991320e-02 4.52654600e-01 -4.59814161e-01 6.94443703e-01
-3.37667987e-02 5.05769730e-01 -2.28617296e-01 3.08241457e-01
1.08090892e-01 -4.71702255e-02 1.78898707e-01 -2.47507095e-01
-1.88538864e-01 5.66271067e-01 -8.72399092e-01 1.68776321e+00
-7.45847583e-01 8.71103525e-01 1.82080850e-01 -6.73380017e-01
1.06765306e+00 8.27756882e-01 8.54542732e-01 -4.18016672e-01
4.73404825e-01 3.76675315e-02 -1.23217747e-01 -8.52675617e-01
1.86987400e-01 -3.46134424e-01 -3.97621125e-01 4.84868318e-01
4.64486450e-01 -9.23573077e-02 -1.50784865e-01 1.40979752e-01
1.11000967e+00 -1.53655848e-02 1.44563258e-01 3.81153747e-02
4.29112673e-01 -4.14228290e-01 6.56639159e-01 6.71818778e-02
-4.68560308e-01 3.58047873e-01 6.47354126e-01 -8.00692365e-02
-5.93941867e-01 -7.67785668e-01 1.31857172e-01 1.07084990e+00
-2.39318609e-01 -8.15204680e-01 -6.61885738e-01 -6.26218379e-01
-2.75016248e-01 7.98707306e-01 -6.91515684e-01 -5.16608000e-01
-8.07323828e-02 -3.39979619e-01 5.75536609e-01 6.09290719e-01
2.25133374e-01 -1.40878546e+00 -5.65767884e-01 1.69231936e-01
-5.59271932e-01 -9.18699324e-01 -5.28389931e-01 4.86206830e-01
-4.76969928e-01 -4.56897974e-01 -1.60397291e-01 -6.15015030e-01
1.04321510e-01 -2.61836857e-01 7.16958582e-01 -2.07656384e-01
-2.95207322e-01 3.24417651e-01 -4.72841829e-01 -2.97712415e-01
-1.08770624e-01 -7.85194710e-03 3.46554607e-01 4.04961556e-01
4.89419997e-01 -7.85898983e-01 -4.92026359e-01 -1.96164604e-02
-2.79210597e-01 2.92073237e-03 3.14461172e-01 8.20224464e-01
5.87080240e-01 3.29301059e-02 9.72207487e-01 -3.74247223e-01
9.59244251e-01 -6.57414436e-01 -2.96463907e-01 -3.18107784e-01
-3.71517450e-01 -3.70086491e-01 4.79326725e-01 -5.66469610e-01
-1.05056691e+00 2.55404681e-01 -5.10116756e-01 -7.29998529e-01
-2.30729535e-01 5.62078536e-01 -1.22372635e-01 6.14123344e-01
3.10160160e-01 2.04345118e-02 -1.93430141e-01 -3.50240380e-01
3.89508456e-01 1.07834351e+00 5.52360237e-01 -3.36848915e-01
3.27317625e-01 -5.47790304e-02 -5.27879119e-01 -7.19916463e-01
-1.22480094e+00 -7.40935504e-01 -6.16387784e-01 -6.75836682e-01
1.06953514e+00 -9.98251319e-01 -8.69365871e-01 4.53146487e-01
-1.04156935e+00 -5.19865334e-01 -4.26688075e-01 8.62292528e-01
-8.01150024e-01 -5.46917878e-03 -9.75750685e-01 -1.12758994e+00
-6.25333548e-01 -7.29441941e-01 9.51547921e-01 2.56460488e-01
-8.15544963e-01 -7.30158329e-01 3.66963178e-01 2.28404284e-01
3.07257891e-01 1.42855272e-01 5.04315078e-01 -8.15913975e-01
3.62036496e-01 -1.75271615e-01 1.84797689e-01 6.57797158e-01
1.01096623e-01 -1.02334164e-01 -1.52664518e+00 -2.41857246e-02
2.39156589e-01 -8.18618059e-01 6.89481199e-01 3.09277773e-01
1.39973235e+00 -2.90321440e-01 3.27911153e-02 5.20979643e-01
9.13921237e-01 2.64009655e-01 1.89852431e-01 -2.77173311e-01
2.96457857e-01 6.14201963e-01 8.64226341e-01 8.87000263e-01
1.56941250e-01 4.90594715e-01 2.46895611e-01 -5.36176153e-02
3.29546742e-02 -5.22240140e-02 7.57904112e-01 1.75659716e+00
8.91102776e-02 4.30526547e-02 -7.06499398e-01 5.94343722e-01
-1.70699978e+00 -1.01278543e+00 -5.27370051e-02 1.64723766e+00
1.09721076e+00 4.77817953e-02 2.84393996e-01 3.10366392e-01
4.60363895e-01 2.69944280e-01 -4.94958818e-01 -9.46891427e-01
1.65338472e-01 3.12932193e-01 -4.51943517e-01 5.16451180e-01
-1.12722230e+00 9.74139273e-01 5.77388954e+00 7.96573520e-01
-1.15581369e+00 2.89725453e-01 6.29286945e-01 -5.47308862e-01
-2.51868926e-02 -4.53916073e-01 -3.32706571e-01 3.96673024e-01
1.54870987e+00 -1.93716794e-01 4.55501050e-01 7.09005117e-01
3.03305477e-01 1.28009543e-01 -8.61527801e-01 1.13992798e+00
5.34510799e-02 -7.83943236e-01 -6.80583894e-01 -2.55952746e-01
4.69618410e-01 -4.30178680e-02 2.54720837e-01 6.57868505e-01
-6.26279339e-02 -8.49073410e-01 5.71675181e-01 6.54532075e-01
9.40917015e-01 -1.10933590e+00 6.05969250e-01 1.49596199e-01
-1.22737586e+00 -6.84544519e-02 -5.76269906e-03 -2.51202248e-02
1.18028797e-01 7.31657624e-01 -7.00997531e-01 3.97727042e-01
9.17174518e-01 8.32235456e-01 -7.37119168e-02 5.51873744e-01
-4.32681352e-01 1.17704570e+00 -1.13807775e-01 -1.82856973e-02
1.57848030e-01 8.07222910e-03 5.20365834e-01 1.55033100e+00
2.02430591e-01 4.37627465e-01 -2.24439487e-01 4.79455501e-01
-1.59392506e-01 2.73507416e-01 -3.03816915e-01 -3.71293873e-01
3.95587951e-01 1.81364751e+00 -4.06280249e-01 -2.93880045e-01
-2.32375771e-01 1.19455147e+00 3.73892516e-01 2.41956890e-01
-1.05495965e+00 -7.11664677e-01 7.36930311e-01 -4.97613609e-01
2.51339883e-01 -7.43729174e-02 -8.15164894e-02 -1.03011549e+00
-1.85294300e-01 -9.85116363e-01 3.56738806e-01 -1.04512465e+00
-1.44904029e+00 1.10665238e+00 -4.96351957e-01 -1.31784701e+00
-3.86922747e-01 -2.41883099e-01 -7.45461226e-01 7.03186333e-01
-9.13572729e-01 -7.86425948e-01 -2.46002510e-01 5.90354085e-01
7.40765870e-01 -1.47875339e-01 1.10209477e+00 3.61644864e-01
-7.03688443e-01 7.35275030e-01 -1.70823351e-01 6.04366474e-02
7.20166266e-01 -1.08715439e+00 9.89156961e-02 5.89374065e-01
1.69676304e-01 3.39018434e-01 4.86106038e-01 -3.80647242e-01
-1.12344527e+00 -1.04420984e+00 1.03925931e+00 -2.01521948e-01
8.75924647e-01 -5.67296028e-01 -7.82467306e-01 4.22799110e-01
5.85902214e-01 3.95268425e-02 1.26374173e+00 2.68564314e-01
-3.25164884e-01 -1.79424047e-01 -7.29064882e-01 3.22654963e-01
7.28107333e-01 -8.92266572e-01 -4.84359831e-01 9.92316157e-02
9.40514445e-01 -1.36608794e-01 -1.14950657e+00 3.69955391e-01
7.94959962e-01 -8.25963736e-01 5.39964855e-01 -5.69518387e-01
4.32655603e-01 5.11514768e-02 -3.04788113e-01 -1.46284008e+00
-1.82852983e-01 -8.72788787e-01 -1.09720998e-01 1.43039155e+00
4.50527847e-01 -2.22548991e-01 2.83484906e-01 1.23613872e-01
-3.70724648e-01 -1.16083539e+00 -8.85860562e-01 -4.39167559e-01
-2.77357370e-01 -9.69736636e-01 2.06920996e-01 1.13295102e+00
6.93348229e-01 4.35152680e-01 -4.88995165e-01 1.15265854e-01
2.28465572e-01 3.40199709e-01 3.56693894e-01 -8.74299467e-01
-3.90120059e-01 -3.35149735e-01 1.28190275e-02 -7.29074121e-01
4.17377293e-01 -1.00139105e+00 3.62512022e-01 -1.24612415e+00
5.58123216e-02 -1.24115394e-02 -8.80564034e-01 6.74834490e-01
-9.78245363e-02 3.49101663e-01 1.51269361e-01 1.95310600e-02
-8.32859695e-01 9.38851774e-01 8.49188805e-01 2.58362710e-01
-4.85444814e-01 1.47040859e-01 -4.22351688e-01 7.67551541e-01
1.26346660e+00 -5.78402519e-01 -2.11335540e-01 7.29792640e-02
-1.59434438e-01 3.12389672e-01 2.11382881e-02 -1.04020631e+00
3.67325991e-02 7.26156384e-02 3.76840144e-01 -6.48646295e-01
8.49042177e-01 -5.14709353e-01 -9.33085904e-02 3.93145770e-01
-6.56147420e-01 -6.14160821e-02 3.82500321e-01 1.88865066e-01
-4.94011462e-01 2.11946413e-01 6.39813662e-01 3.23689550e-01
-3.70176613e-01 2.82456756e-01 -6.74207747e-01 -8.45052898e-02
7.84659803e-01 3.84115249e-01 -7.04143196e-02 -6.52890444e-01
-1.06531131e+00 2.23754168e-01 -2.95703501e-01 6.01521492e-01
6.86837792e-01 -1.47195375e+00 -5.35524130e-01 3.79900575e-01
8.94133449e-02 -6.08775973e-01 4.98524934e-01 1.19400191e+00
2.01046988e-01 2.59173870e-01 -1.93418950e-01 -4.44058150e-01
-1.45868397e+00 2.84119934e-01 3.32208425e-01 -1.31156102e-01
-3.03223759e-01 1.20623660e+00 -5.10290936e-02 -1.80366904e-01
3.65910232e-01 -6.38081133e-02 -2.77017146e-01 4.69829440e-01
3.21417451e-01 2.80505389e-01 -1.05918154e-01 -8.15259814e-01
-6.26087308e-01 3.51808578e-01 2.29271740e-01 -6.15956068e-01
1.63097155e+00 -1.24790877e-01 -7.78695047e-02 9.53433812e-01
1.53852701e+00 2.65891671e-01 -1.10380328e+00 -1.78819701e-01
3.60674858e-02 5.33831492e-02 9.11159813e-02 -7.64028072e-01
-1.07739532e+00 1.16183507e+00 7.29290307e-01 1.89381257e-01
1.47119236e+00 1.24572873e-01 7.71590889e-01 1.33403257e-01
-2.15824455e-01 -1.19319487e+00 5.09319901e-01 5.58640301e-01
1.07294440e+00 -8.11829090e-01 -3.35089952e-01 -9.67866927e-02
-1.16801596e+00 9.02305365e-01 6.52814448e-01 -8.35151076e-02
8.65156054e-01 4.08322871e-01 5.77601552e-01 -2.95704186e-01
-1.44890583e+00 -2.69838840e-01 4.22186911e-01 2.05811203e-01
6.25219941e-01 1.58761516e-02 -2.11187676e-01 1.34338951e+00
-4.40981358e-01 -1.56964451e-01 1.34048671e-01 5.07470250e-01
-2.65522659e-01 -8.94697547e-01 2.04246372e-01 3.58958811e-01
-5.95392406e-01 1.14435563e-02 -5.47379494e-01 1.39900416e-01
-2.72422191e-02 1.33423114e+00 7.80367628e-02 -1.03527296e+00
3.59536618e-01 6.31291032e-01 9.61209908e-02 -6.15297616e-01
-1.00519443e+00 6.38069868e-01 3.43611211e-01 -9.34809983e-01
-4.16039199e-01 -7.71274090e-01 -1.49692786e+00 2.86873758e-01
-2.50113517e-01 4.56467062e-01 7.32881486e-01 6.66807830e-01
4.46000069e-01 7.68585563e-01 1.22371233e+00 -7.24468529e-01
-2.18140066e-01 -1.11865902e+00 -5.08277953e-01 3.17945659e-01
3.82045329e-01 -3.80403817e-01 -5.22395909e-01 1.12164699e-01]
|
[13.523615837097168, 5.691315174102783]
|
54079074-e42a-42f7-81bc-953889e83d10
|
utahbmi-at-semeval-2016-task-12-extracting
| null | null |
https://aclanthology.org/S16-1195
|
https://aclanthology.org/S16-1195.pdf
|
UtahBMI at SemEval-2016 Task 12: Extracting Temporal Information from Clinical Text
| null |
['Stephane Meystre', 'Abdulrahman Khalifa', 'Sumithra Velupillai']
|
2016-06-01
| null | null | null |
semeval-2016-6
|
['temporal-information-extraction']
|
['natural-language-processing']
|
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
|
[-7.417088985443115, 3.7920775413513184]
|
242c7858-6b9f-4058-b390-17ef5223efbd
|
rgbd1k-a-large-scale-dataset-and-benchmark
|
2208.09787
| null |
https://arxiv.org/abs/2208.09787v3
|
https://arxiv.org/pdf/2208.09787v3.pdf
|
RGBD1K: A Large-scale Dataset and Benchmark for RGB-D Object Tracking
|
RGB-D object tracking has attracted considerable attention recently, achieving promising performance thanks to the symbiosis between visual and depth channels. However, given a limited amount of annotated RGB-D tracking data, most state-of-the-art RGB-D trackers are simple extensions of high-performance RGB-only trackers, without fully exploiting the underlying potential of the depth channel in the offline training stage. To address the dataset deficiency issue, a new RGB-D dataset named RGBD1K is released in this paper. The RGBD1K contains 1,050 sequences with about 2.5M frames in total. To demonstrate the benefits of training on a larger RGB-D data set in general, and RGBD1K in particular, we develop a transformer-based RGB-D tracker, named SPT, as a baseline for future visual object tracking studies using the new dataset. The results, of extensive experiments using the SPT tracker emonstrate the potential of the RGBD1K dataset to improve the performance of RGB-D tracking, inspiring future developments of effective tracker designs. The dataset and codes will be available on the project homepage: https://github.com/xuefeng-zhu5/RGBD1K.
|
['Josef Kittler', 'Xiao-Jun Wu', 'Xiao Yang', 'Haodong Liu', 'Zucheng Wu', 'Zhangyong Tang', 'Tianyang Xu', 'Xue-Feng Zhu']
|
2022-08-21
| null | null | null | null |
['visual-object-tracking']
|
['computer-vision']
|
[-4.27719295e-01 -2.54934877e-01 -3.35760713e-01 -1.22501276e-01
-5.35246372e-01 -7.90679693e-01 3.83592337e-01 -4.55904454e-01
-2.56090611e-01 2.40074545e-01 -8.36962238e-02 -2.43754506e-01
3.75069588e-01 -3.37717444e-01 -6.43545151e-01 -6.33212686e-01
-4.00525220e-02 3.19535136e-02 6.30887628e-01 3.07578240e-02
-2.12216884e-01 7.71784842e-01 -1.62862527e+00 -2.27380663e-01
3.25985640e-01 1.33361793e+00 1.48844734e-01 6.28684103e-01
1.40268579e-01 2.97664523e-01 -5.36003292e-01 -2.92439431e-01
6.91614687e-01 -2.19634697e-01 -1.97422713e-01 1.22926615e-01
7.12808907e-01 -5.51731169e-01 -8.64577651e-01 8.28386009e-01
7.24759161e-01 -1.84979975e-01 4.86278608e-02 -1.69928038e+00
-5.98922968e-01 -1.75456166e-01 -5.59151530e-01 3.79794329e-01
4.37845260e-01 6.64756954e-01 5.84627450e-01 -7.82059610e-01
5.67136586e-01 1.01130557e+00 9.15828824e-01 9.29653823e-01
-8.20317984e-01 -9.58025515e-01 -6.53559268e-02 -1.05784997e-01
-1.20958829e+00 -2.51410782e-01 8.08147490e-01 -3.28063518e-01
8.41753066e-01 5.58752418e-02 1.17830384e+00 1.15271354e+00
2.18630061e-02 1.11509979e+00 1.01217031e+00 -1.35277271e-01
-4.56969030e-02 -1.37869820e-01 -1.22082271e-01 6.75294697e-01
5.16511977e-01 7.43182778e-01 -8.15777063e-01 1.06055036e-01
9.99193192e-01 -2.76134592e-02 -2.57813483e-01 -8.92246783e-01
-1.34539962e+00 3.87986302e-01 8.63525867e-01 2.99127735e-02
2.76696403e-02 6.03778303e-01 3.07086855e-01 -5.34737762e-03
3.52981448e-01 2.00526211e-02 -3.62843424e-01 -5.50538182e-01
-6.49212539e-01 1.72466218e-01 2.98084289e-01 1.33369446e+00
4.72727597e-01 1.07826360e-01 2.09068414e-02 1.29583389e-01
7.96932042e-01 1.05180860e+00 1.52812958e-01 -1.03314519e+00
4.09317523e-01 6.96963608e-01 3.66627902e-01 -5.43847978e-01
-5.49010396e-01 -1.59711614e-01 -2.89285243e-01 4.90917295e-01
7.33250320e-01 -1.01442643e-01 -9.01317120e-01 1.44789898e+00
7.51001477e-01 1.20964155e-01 -8.33103955e-02 1.46161222e+00
1.05123234e+00 2.63994813e-01 -7.49873668e-02 2.13056698e-01
1.41851544e+00 -5.62059462e-01 -5.88310182e-01 -1.21373683e-01
5.46065986e-01 -7.01099038e-01 8.91320229e-01 5.34333736e-02
-7.84068406e-01 -8.08543622e-01 -1.00423837e+00 -8.16692505e-03
-2.83948809e-01 3.92594278e-01 9.87644732e-01 1.03783679e+00
-1.03734505e+00 6.72525018e-02 -1.26464248e+00 -6.15807235e-01
4.80336457e-01 4.47136402e-01 -3.19525450e-01 -2.17554951e-03
-1.04025722e+00 8.82268250e-01 2.17252627e-01 2.93558627e-01
-9.67552066e-01 -7.97107875e-01 -6.48991227e-01 -6.59391940e-01
3.43219906e-01 -3.85721773e-01 1.30038142e+00 -1.38198107e-01
-1.38255417e+00 1.12215710e+00 1.53147981e-01 -1.18271232e-01
5.23558736e-01 -3.42062533e-01 -3.35360736e-01 1.93197027e-01
-1.47042692e-01 7.54154980e-01 5.96520662e-01 -1.23643923e+00
-8.21456730e-01 -5.85839331e-01 -4.34730873e-02 2.55499370e-02
-8.53306055e-02 -1.68190271e-01 -1.08836138e+00 -4.86166984e-01
1.45244166e-01 -1.24933243e+00 1.00352973e-01 6.90930426e-01
-3.21991086e-01 -2.38016129e-01 1.16066635e+00 -2.70785481e-01
8.01870286e-01 -2.29631710e+00 -1.76391453e-01 -1.87810928e-01
2.96762973e-01 6.60042465e-01 4.81622219e-02 7.52877295e-02
2.34854251e-01 -3.81478220e-01 3.64654452e-01 -4.40119654e-01
3.47119793e-02 2.05152586e-01 -1.10266000e-01 9.14854407e-01
-5.86326746e-03 1.12313318e+00 -1.02886438e+00 -4.71259445e-01
7.09503293e-01 5.95740139e-01 -1.58727095e-01 1.95514977e-01
-1.42704815e-01 7.19931245e-01 -6.57917798e-01 1.27022302e+00
6.61759853e-01 -3.15912426e-01 -9.63012055e-02 -4.89672512e-01
-4.14489686e-01 7.76557028e-02 -8.58352542e-01 1.92954850e+00
1.23616308e-01 1.02464199e+00 -1.02490775e-01 -3.70249987e-01
9.34932351e-01 2.09016189e-01 9.61071610e-01 -9.70637262e-01
3.68780255e-01 2.05953613e-01 -1.82079703e-01 -2.84996182e-01
6.23217940e-01 9.13578868e-02 -6.51415214e-02 2.79913783e-01
-2.62218770e-02 -1.15909152e-01 2.20233388e-02 2.22903714e-01
1.08207691e+00 6.59999847e-01 -1.50500759e-01 1.74103260e-01
1.19930431e-01 3.71652812e-01 5.15466988e-01 6.86929643e-01
-8.33643377e-01 5.70069671e-01 -2.87688011e-03 -4.15801853e-01
-1.10747278e+00 -1.23678064e+00 -2.80812114e-01 4.96200562e-01
7.70147502e-01 -3.52502525e-01 -2.61138320e-01 -6.36599422e-01
6.86248481e-01 1.72812581e-01 -6.38130903e-01 -1.65182963e-01
-4.33839202e-01 -5.19567609e-01 9.42725658e-01 7.22625971e-01
7.32380211e-01 -6.52964056e-01 -1.25772941e+00 -5.23781329e-02
1.15374988e-02 -1.35870957e+00 -3.18886101e-01 2.51365483e-01
-1.08635712e+00 -1.23132193e+00 -8.58120441e-01 -2.67365456e-01
2.82271564e-01 7.69296885e-01 9.25800025e-01 -2.74217576e-02
-2.98907310e-01 9.80236530e-01 -4.89931881e-01 -4.25646037e-01
-1.14735872e-01 -1.47510529e-01 3.24728072e-01 -5.44719398e-01
6.43428862e-01 -1.41160341e-03 -7.13668823e-01 7.06163228e-01
-4.25006181e-01 7.87482932e-02 4.74732101e-01 1.76698461e-01
6.05127275e-01 -3.42732698e-01 -1.63377658e-01 3.54369506e-02
-2.11746484e-01 -7.07284808e-02 -1.16798246e+00 -3.19524505e-03
-4.32393461e-01 -2.14012861e-01 -1.59355905e-02 -5.32303989e-01
-7.26572454e-01 3.59898418e-01 3.34568694e-02 -9.77379858e-01
1.33523002e-01 -2.50734329e-01 -1.71955600e-01 -5.03861785e-01
3.69378567e-01 2.22216323e-01 2.81409025e-01 -5.55554152e-01
3.56358528e-01 4.40617055e-01 7.38650620e-01 -3.38126391e-01
1.13677204e+00 6.82294607e-01 -8.69147480e-02 -7.00147450e-01
-7.96279907e-01 -6.89538658e-01 -5.05346000e-01 -8.52567434e-01
8.71112466e-01 -1.18216908e+00 -1.21989715e+00 7.60564625e-01
-8.44858229e-01 -5.34239888e-01 -3.21274757e-01 7.89780378e-01
-4.64643866e-01 2.04299867e-01 -4.53062505e-01 -8.54672790e-01
2.41257045e-02 -1.08108127e+00 1.34559906e+00 6.56417131e-01
2.49187559e-01 -7.77718604e-01 1.66165128e-01 3.05538148e-01
3.17920238e-01 3.43481570e-01 -1.44847156e-02 1.01982029e-02
-1.04738092e+00 -4.01004344e-01 -3.77390057e-01 9.96989533e-02
2.25425929e-01 -1.74294151e-02 -9.43043530e-01 -4.29621309e-01
-3.31551105e-01 -3.16735029e-01 5.73479652e-01 4.13475543e-01
7.10962713e-01 4.73847657e-01 -5.57358503e-01 8.55497241e-01
1.35089970e+00 1.08339019e-01 5.10331035e-01 6.82634175e-01
9.92117584e-01 -1.98191091e-01 9.08360064e-01 4.92623389e-01
5.41767657e-01 1.10046744e+00 7.68451512e-01 -2.32287735e-01
-6.34655118e-01 -3.02647561e-01 4.45984274e-01 6.14424169e-01
-1.94448456e-01 -1.17547356e-01 -1.00103533e+00 3.26165080e-01
-1.55285037e+00 -7.25478590e-01 -3.58457744e-01 2.11917329e+00
7.32688367e-01 2.17245743e-01 4.27587181e-01 -5.48604429e-02
5.39521933e-01 1.03825308e-01 -8.77323568e-01 4.41061199e-01
-2.02080309e-01 -1.26900539e-01 1.02127874e+00 -4.44670133e-02
-1.16458106e+00 8.57091188e-01 6.20400047e+00 3.89577538e-01
-1.15640211e+00 -6.76726550e-02 -2.08262771e-01 -3.83561075e-01
1.87890887e-01 6.57112673e-02 -1.45126677e+00 6.60659432e-01
7.13862002e-01 -6.96700737e-02 1.02804743e-01 9.92062688e-01
1.92542374e-01 -3.57459605e-01 -9.20400679e-01 1.30806804e+00
-1.52196273e-01 -1.15205491e+00 -5.23817420e-01 3.58532459e-01
3.19725454e-01 4.42714095e-01 9.81556028e-02 1.47222415e-01
2.94627607e-01 -4.38629985e-01 9.00588334e-01 3.38693351e-01
6.79177225e-01 -3.45611900e-01 4.67796504e-01 1.57421231e-02
-1.64635515e+00 2.57902056e-01 -5.08435011e-01 1.58885270e-01
2.06664070e-01 3.91636044e-01 -5.23224831e-01 5.36538124e-01
1.22975659e+00 1.28361058e+00 -9.25099492e-01 1.42843676e+00
-3.13704729e-01 4.43033248e-01 -6.25016212e-01 -4.61606830e-02
1.10943325e-01 1.38827205e-01 4.91634160e-01 8.18131089e-01
3.04810762e-01 2.05729574e-01 4.37796414e-02 6.12249076e-01
-5.25421463e-02 -7.21128643e-01 -5.26585340e-01 -5.33353090e-02
5.71890056e-01 1.17354608e+00 -7.86899805e-01 -8.96606892e-02
-6.22550488e-01 6.22956634e-01 -1.04227372e-01 2.36796558e-01
-1.31942797e+00 -7.71239996e-02 1.03992474e+00 -3.98490876e-02
6.56881869e-01 -6.67644739e-01 -8.77668113e-02 -1.11793566e+00
-4.53706458e-02 -6.51076794e-01 2.34407455e-01 -1.11686563e+00
-1.05034935e+00 2.25520462e-01 -4.65512611e-02 -1.88707221e+00
2.34219998e-01 -9.58888829e-01 3.28049399e-02 6.68377876e-01
-1.51539266e+00 -1.26343954e+00 -8.61739099e-01 6.95531547e-01
2.15136018e-02 4.77015972e-02 2.80558378e-01 4.39491689e-01
-4.62142467e-01 5.73059142e-01 1.73527300e-01 3.70547652e-01
7.73258507e-01 -1.08265579e+00 5.75485289e-01 8.95521045e-01
6.73592612e-02 4.36399907e-01 6.60814106e-01 -7.36475289e-01
-2.28204107e+00 -8.73598754e-01 2.01976255e-01 -1.00703621e+00
5.56033254e-01 -5.39102972e-01 -5.07001162e-01 6.89284503e-01
-1.18050493e-01 5.45142412e-01 4.99624103e-01 -2.91016519e-01
-1.66581661e-01 -1.57026932e-01 -9.82491136e-01 3.03120703e-01
1.28802061e+00 -5.15181720e-01 -4.20086563e-01 1.41424373e-01
7.36181796e-01 -1.10815883e+00 -1.23130095e+00 2.54606515e-01
8.96621764e-01 -1.00330651e+00 1.33399880e+00 -8.07850510e-02
-2.39941269e-01 -8.82222950e-01 -1.90483004e-01 -8.24551702e-01
1.87328216e-02 -3.41090292e-01 -5.92774034e-01 1.14299917e+00
-1.98995993e-01 -3.57700914e-01 1.18233025e+00 6.34936929e-01
-1.67935103e-01 -4.19249296e-01 -1.16551208e+00 -1.16933107e+00
-3.91391069e-01 -7.40893006e-01 4.29003090e-01 4.64630663e-01
-3.68079305e-01 -2.39687219e-01 -4.99966294e-01 2.24643350e-01
1.26752162e+00 3.06340784e-01 1.25993001e+00 -9.51147258e-01
-1.01278424e-01 -1.47323310e-01 -7.52313912e-01 -1.62698388e+00
-3.66263360e-01 -6.62882984e-01 8.49158540e-02 -1.24314404e+00
-1.10678479e-01 -8.21814954e-01 -3.61852758e-02 5.39550185e-01
-1.50874108e-01 7.72469997e-01 5.84984481e-01 5.36297739e-01
-8.73728454e-01 7.57608235e-01 1.57342935e+00 2.04035584e-02
-5.40758632e-02 1.18008845e-01 -2.75464386e-01 2.70714045e-01
4.98636991e-01 -4.77956265e-01 -2.40303278e-02 -4.22862858e-01
-2.06800327e-01 -3.25391814e-02 6.78638875e-01 -1.16392076e+00
4.11667526e-01 1.21213809e-01 8.60081315e-01 -1.11761510e+00
5.70169151e-01 -1.02221620e+00 4.16553348e-01 7.24283040e-01
2.22675830e-01 2.62986198e-02 5.86926758e-01 4.36292052e-01
-6.58481801e-03 4.29625779e-01 6.85500920e-01 5.24366423e-02
-1.31812644e+00 6.26512945e-01 -3.81704643e-02 2.40209296e-01
1.19609606e+00 -8.01880121e-01 -4.65326279e-01 -8.72460427e-04
-3.88601303e-01 3.15995514e-01 9.81466234e-01 7.98636794e-01
5.69868803e-01 -1.71987152e+00 -1.86036572e-01 1.85504407e-01
3.56574178e-01 -9.71440598e-02 2.78753191e-02 9.13420081e-01
-4.71688479e-01 6.21681333e-01 -5.29485106e-01 -1.20564914e+00
-1.15713203e+00 4.91768241e-01 4.26077425e-01 1.85973957e-01
-9.73248243e-01 7.91722059e-01 -1.10933676e-01 -1.58412993e-01
4.61936533e-01 -4.72272724e-01 3.55922222e-01 -1.75479174e-01
4.18213308e-01 3.53976488e-01 -8.11434835e-02 -7.13027716e-01
-8.71940494e-01 6.86389565e-01 4.06860381e-01 8.28825086e-02
1.12662089e+00 -3.56586367e-01 4.70946789e-01 2.99571902e-01
9.87141430e-01 -3.53193820e-01 -1.97507286e+00 -2.07504392e-01
2.88979989e-03 -9.41110313e-01 9.52071883e-03 -4.43318546e-01
-1.51411211e+00 5.72021544e-01 1.15201056e+00 -3.71215865e-02
1.01632166e+00 3.43566597e-01 7.37446904e-01 -6.55526519e-02
7.79569030e-01 -6.41608238e-01 3.25185448e-01 3.44373792e-01
2.76910901e-01 -1.43073833e+00 2.85001725e-01 -1.14962853e-01
-4.65671986e-01 1.03355134e+00 7.55009353e-01 1.75577626e-02
3.62324327e-01 5.08193970e-01 3.67920429e-01 -3.19106936e-01
-3.02187055e-01 -5.29029667e-01 1.26060069e-01 9.14936781e-01
3.59309763e-01 -1.88403666e-01 2.88340688e-01 7.35421525e-03
-4.22279947e-02 2.12089509e-01 1.48068398e-01 1.24558091e+00
-1.93622842e-01 -1.09641445e+00 -6.56455219e-01 6.21110797e-02
3.65413539e-02 4.40453321e-01 -2.74747163e-01 1.28347015e+00
5.67845739e-02 7.11600423e-01 -8.56479108e-02 -4.45405722e-01
6.17924750e-01 -3.72420907e-01 8.98025215e-01 -5.51885106e-02
-5.85369110e-01 -2.47140378e-02 -2.32375681e-01 -1.06123567e+00
-9.43405807e-01 -8.71838927e-01 -1.16120493e+00 -6.21075869e-01
-4.55806136e-01 -2.79977441e-01 8.85219872e-01 6.27031922e-01
3.32158417e-01 3.81784528e-01 2.33015463e-01 -1.24553096e+00
-2.49783561e-01 -4.84974712e-01 -6.03922784e-01 2.55721152e-01
4.33242112e-01 -1.08879673e+00 -1.18474506e-01 -3.99021171e-02]
|
[6.5484747886657715, -2.1562163829803467]
|
59991fc5-92e9-4378-80c5-cc1e44e1d929
|
vectormapnet-end-to-end-vectorized-hd-map
|
2206.0892
| null |
https://arxiv.org/abs/2206.08920v6
|
https://arxiv.org/pdf/2206.08920v6.pdf
|
VectorMapNet: End-to-end Vectorized HD Map Learning
|
Autonomous driving systems require High-Definition (HD) semantic maps to navigate around urban roads. Existing solutions approach the semantic mapping problem by offline manual annotation, which suffers from serious scalability issues. Recent learning-based methods produce dense rasterized segmentation predictions to construct maps. However, these predictions do not include instance information of individual map elements and require heuristic post-processing to obtain vectorized maps. To tackle these challenges, we introduce an end-to-end vectorized HD map learning pipeline, termed VectorMapNet. VectorMapNet takes onboard sensor observations and predicts a sparse set of polylines in the bird's-eye view. This pipeline can explicitly model the spatial relation between map elements and generate vectorized maps that are friendly to downstream autonomous driving tasks. Extensive experiments show that VectorMapNet achieve strong map learning performance on both nuScenes and Argoverse2 dataset, surpassing previous state-of-the-art methods by 14.2 mAP and 14.6mAP. Qualitatively, VectorMapNet is capable of generating comprehensive maps and capturing fine-grained details of road geometry. To the best of our knowledge, VectorMapNet is the first work designed towards end-to-end vectorized map learning from onboard observations. Our project website is available at \url{https://tsinghua-mars-lab.github.io/vectormapnet/}.
|
['Tianyuan Yuan', 'Hang Zhao', 'Yilun Wang', 'Yue Wang', 'Yicheng Liu']
|
2022-06-17
| null | null | null | null |
['3d-lane-detection', 'hd-semantic-map-learning']
|
['computer-vision', 'computer-vision']
|
[-1.11492701e-01 3.49693567e-01 -2.73655444e-01 -8.36683571e-01
-8.30527663e-01 -6.21226311e-01 7.73552001e-01 6.00212812e-02
-3.04325551e-01 6.37496769e-01 4.46966439e-02 -5.10661125e-01
8.93424004e-02 -1.28575206e+00 -1.01509571e+00 -5.06716259e-02
-1.49378389e-01 8.55709553e-01 5.37736177e-01 -7.17454195e-01
1.20201416e-01 3.13744664e-01 -1.83322930e+00 8.95039588e-02
1.07993829e+00 8.48960817e-01 5.20396113e-01 7.89457977e-01
-2.60588318e-01 6.42555416e-01 9.55766663e-02 -1.79582566e-01
3.89329612e-01 1.88774481e-01 -6.61255360e-01 -3.13333005e-01
7.00406313e-01 -5.09656250e-01 -6.00047708e-01 8.23728085e-01
1.14682019e-01 1.32568553e-01 5.21637559e-01 -1.52404511e+00
-2.02769279e-01 4.76040393e-01 -3.37625891e-01 1.68556705e-01
-6.21454455e-02 3.36576998e-01 1.03216445e+00 -8.80075753e-01
5.76465547e-01 9.24051583e-01 8.56031835e-01 -6.48531318e-02
-1.07848716e+00 -7.87763894e-01 3.10372561e-01 3.12391251e-01
-1.72773516e+00 -2.94993043e-01 6.96185887e-01 -8.20096076e-01
1.03126156e+00 2.42019966e-01 5.81763506e-01 7.09235787e-01
6.81378692e-02 7.72918165e-01 1.08635294e+00 2.23678738e-01
2.58400440e-01 8.77558962e-02 1.87353179e-01 7.59263873e-01
-8.31663460e-02 3.79335850e-01 -5.12761295e-01 3.52134436e-01
7.28388190e-01 -1.21532403e-01 2.20486388e-01 -5.49943089e-01
-1.21555996e+00 8.80697370e-01 1.00303745e+00 -3.15769851e-01
-5.08316398e-01 2.97444344e-01 1.86115354e-01 -2.95920204e-02
6.25419855e-01 2.15254515e-01 -4.40731555e-01 -3.59478146e-01
-9.25403059e-01 6.37025058e-01 4.90825415e-01 1.19120133e+00
1.52237737e+00 4.50299941e-02 3.32366943e-01 7.76731730e-01
4.74147536e-02 8.97322118e-01 -8.44151825e-02 -1.07538807e+00
8.08987916e-01 6.21349752e-01 1.89903691e-01 -9.93949652e-01
-7.51805842e-01 -2.42380396e-01 -6.09897852e-01 4.49022621e-01
1.92566827e-01 -2.70822644e-01 -1.18564379e+00 1.35613537e+00
2.59909242e-01 2.92258829e-01 -1.22455582e-01 9.96399462e-01
9.57607150e-01 9.34770703e-01 9.58354250e-02 7.82741904e-01
9.10567462e-01 -1.19032741e+00 -4.31417406e-01 -7.55174577e-01
7.83930898e-01 -3.05396825e-01 1.07296920e+00 -8.14747140e-02
-5.32846153e-01 -7.64650881e-01 -1.20311129e+00 -1.74783766e-01
-9.20233548e-01 1.66138411e-01 8.66090953e-01 1.87312514e-01
-1.25688434e+00 1.62927896e-01 -9.56000686e-01 -4.30891365e-01
7.11682498e-01 4.24090214e-02 -3.80627930e-01 -8.74422640e-02
-1.15074933e+00 1.10202682e+00 5.66108644e-01 1.05535716e-01
-9.46537495e-01 -1.01365125e+00 -1.25966549e+00 -2.58333474e-01
1.36231542e-01 -3.57842177e-01 1.35247886e+00 -3.12933058e-01
-1.11010206e+00 7.23903835e-01 -2.62638927e-01 -5.41559637e-01
5.62989533e-01 -3.30671877e-01 -4.83944505e-01 -2.26661369e-01
6.37247741e-01 1.36159301e+00 3.09490532e-01 -1.23098600e+00
-1.14636147e+00 -2.44247541e-01 -1.82475206e-02 3.06021899e-01
2.06050858e-01 -5.64154506e-01 -6.39964283e-01 -1.98660716e-01
1.36060521e-01 -1.06880629e+00 -5.55355489e-01 -1.86593793e-02
-5.50234616e-01 -1.34241149e-01 8.25751543e-01 -6.54651880e-01
9.55044925e-01 -1.98512530e+00 -2.48853356e-01 2.44793415e-01
4.21316445e-01 1.86862707e-01 -6.25082180e-02 4.43628490e-01
3.08516324e-01 -1.18418343e-01 -5.71268380e-01 -1.41111732e-01
2.07659736e-01 3.12677890e-01 -5.01042724e-01 4.11237001e-01
1.59617767e-01 1.23107100e+00 -9.47773337e-01 -4.66983676e-01
8.74706030e-01 4.53481615e-01 -3.91998351e-01 1.15808643e-01
-4.09972012e-01 3.70980978e-01 -5.40323913e-01 5.47158480e-01
9.95835125e-01 8.93962942e-03 -2.08261013e-01 -6.09377213e-02
-6.37163520e-01 3.42980832e-01 -9.44567502e-01 1.80413902e+00
-5.42878389e-01 1.18184292e+00 -1.87752590e-01 -9.09309745e-01
1.21303916e+00 -4.11188900e-01 4.15304124e-01 -1.19618678e+00
-1.83919936e-01 1.87345549e-01 -4.95504111e-01 -2.40930647e-01
8.41398895e-01 2.48766974e-01 -3.54506046e-01 1.54002666e-01
-1.51134431e-01 -5.08309841e-01 -6.10437011e-03 1.82266250e-01
8.31359744e-01 3.55604172e-01 -2.20006495e-03 -2.54845560e-01
9.49439704e-02 9.49777663e-01 3.40988964e-01 5.99564552e-01
-9.79106352e-02 6.52565062e-01 2.14166611e-01 -8.76183093e-01
-1.35640204e+00 -1.27150321e+00 -3.85594994e-01 1.15008903e+00
4.63481665e-01 -4.65510547e-01 -6.11336112e-01 -5.50962985e-01
3.50019127e-01 8.92158210e-01 -5.03251791e-01 2.92674124e-01
-5.74870527e-01 -6.02802396e-01 6.32076323e-01 8.16271424e-01
6.32973015e-01 -8.48102450e-01 -6.54898822e-01 3.18542808e-01
-2.24355683e-01 -1.23722804e+00 8.65488499e-02 1.43109918e-01
-4.65059072e-01 -9.27466989e-01 -1.78221837e-01 -6.44479215e-01
5.31598270e-01 4.06179190e-01 1.11529899e+00 -4.21410888e-01
-1.37689352e-01 -1.27220497e-01 -1.22615598e-01 -6.36521995e-01
-2.18961276e-02 5.26313961e-01 -2.14434862e-01 -3.78175706e-01
5.20993352e-01 -5.80331922e-01 -4.93405163e-01 5.88994026e-01
-3.15806448e-01 6.99254394e-01 4.85977799e-01 2.67823160e-01
9.98985469e-01 1.01527557e-01 4.25794572e-01 -7.64545977e-01
-4.56616543e-02 -7.86857843e-01 -9.97115791e-01 -2.33387813e-01
-6.34250343e-01 4.70895041e-03 4.16762322e-01 3.80100667e-01
-1.02568007e+00 5.67276657e-01 -4.81144339e-01 -1.27099231e-01
-4.98013526e-01 4.96920437e-01 5.48953414e-02 5.97928911e-02
7.80381978e-01 1.75876990e-01 -8.20683464e-02 -3.90788406e-01
7.89968610e-01 7.11167634e-01 1.03059649e+00 -9.24321711e-02
1.11828148e+00 6.94987655e-01 -1.17572226e-01 -8.79561901e-01
-9.21787083e-01 -6.68349683e-01 -9.32642698e-01 -4.28639859e-01
9.80539382e-01 -1.36052239e+00 -2.40580097e-01 3.97957712e-01
-7.54251122e-01 -9.61846828e-01 -2.41617840e-02 3.34846705e-01
-8.51309836e-01 -4.56686527e-01 -1.17008358e-01 -4.49936539e-01
-9.36718062e-02 -9.54948902e-01 1.25764489e+00 1.37322545e-01
-1.87739179e-01 -9.68942702e-01 3.03369820e-01 3.89462799e-01
5.38497865e-01 4.18065012e-01 3.00271422e-01 -1.73612654e-01
-8.41730714e-01 -3.63299310e-01 -4.89315867e-01 -1.23436667e-01
-1.58095330e-01 -2.77467072e-01 -1.10223651e+00 2.12613523e-01
-8.43803883e-01 -1.66964814e-01 1.16422892e+00 4.43303049e-01
1.16107750e+00 -1.13419034e-01 -6.15285158e-01 1.00327349e+00
1.45843530e+00 -1.47409841e-01 6.02741063e-01 6.44804120e-01
1.20390689e+00 7.47781754e-01 1.10425878e+00 2.79526234e-01
1.22080839e+00 8.22785437e-01 8.01437974e-01 -4.89882618e-01
-1.88593209e-01 -8.28772783e-01 -3.38429995e-02 2.70839810e-01
2.48515308e-01 -1.31949931e-01 -1.51413894e+00 8.46097529e-01
-2.14847350e+00 -7.62400568e-01 -5.89591444e-01 1.89386737e+00
1.97130546e-01 3.56562108e-01 8.57319459e-02 -3.50604773e-01
4.96640712e-01 4.40791845e-01 -8.69450033e-01 -1.47696942e-01
-1.44816265e-01 -8.87153819e-02 1.26411951e+00 8.91924858e-01
-1.45929623e+00 1.69006050e+00 5.70206118e+00 6.15439713e-01
-1.13610053e+00 2.13531539e-01 4.37118590e-01 -3.06269974e-02
-5.08528471e-01 9.75135937e-02 -1.00954366e+00 3.04190934e-01
1.06473732e+00 -1.08959724e-03 3.68564367e-01 1.20154035e+00
3.29000980e-01 -1.76458806e-01 -6.54201686e-01 9.06814277e-01
-2.85382092e-01 -1.80673623e+00 -3.65242958e-01 1.25899523e-01
8.19665253e-01 1.11235094e+00 -2.84897331e-02 5.32742083e-01
8.69497359e-01 -1.23629248e+00 8.38138163e-01 4.97881919e-01
9.93665457e-01 -7.73750901e-01 6.30980372e-01 3.95374566e-01
-1.54093480e+00 8.38857442e-02 -3.99467081e-01 -3.13114196e-01
6.02665842e-01 4.84402001e-01 -1.24582160e+00 3.24741513e-01
7.80347705e-01 1.15099907e+00 -7.60857642e-01 9.09395337e-01
-3.06060612e-01 6.49432063e-01 -5.38173199e-01 2.16638431e-01
6.82693958e-01 -6.13477863e-02 2.94806063e-01 1.27284956e+00
2.22856835e-01 -1.05169564e-01 3.88938308e-01 9.31373000e-01
2.56719530e-01 -7.06393272e-02 -1.04313946e+00 2.77363092e-01
7.42667496e-01 1.24562764e+00 -5.83247125e-01 -3.89099896e-01
-2.09622651e-01 5.91631532e-01 6.02742255e-01 3.29722017e-01
-1.00368023e+00 -4.87170726e-01 1.03517270e+00 4.51844335e-01
3.76548648e-01 -6.19063914e-01 -8.20443511e-01 -7.02801764e-01
-6.64181039e-02 -5.11174165e-02 -2.80247834e-02 -9.83671904e-01
-6.32946670e-01 5.02653778e-01 1.61101401e-01 -1.24452925e+00
-3.76084238e-01 -3.47350329e-01 -6.40929103e-01 8.67168963e-01
-1.83550489e+00 -1.55364716e+00 -8.93431365e-01 2.54883707e-01
6.77566767e-01 -1.59506276e-01 5.19637883e-01 1.01469539e-01
-3.65298480e-01 1.45239785e-01 1.37902454e-01 4.26941104e-02
3.68516296e-01 -1.35305643e+00 1.28398883e+00 7.79287398e-01
1.11390397e-01 -1.53034076e-01 6.48779750e-01 -7.73744047e-01
-1.24159718e+00 -1.87038946e+00 1.00525391e+00 -7.63780296e-01
5.88345945e-01 -5.67965508e-01 -7.84344852e-01 9.81073737e-01
-2.25223243e-01 4.24792916e-02 1.48169428e-01 1.52866155e-01
-2.60110080e-01 -3.33217531e-01 -7.69178271e-01 4.86482203e-01
1.29410839e+00 -6.66593373e-01 -1.83791399e-01 4.74192113e-01
8.11550438e-01 -5.76611459e-01 -7.83061206e-01 4.83036935e-01
4.45661485e-01 -9.90663528e-01 9.70337868e-01 -7.71644264e-02
4.89684522e-01 -5.65476239e-01 -2.81317383e-01 -1.56718302e+00
-5.55145144e-01 -4.32382524e-02 1.63898870e-01 8.81043732e-01
7.98917353e-01 -6.33419812e-01 9.70096529e-01 3.83740723e-01
-7.17062652e-01 -5.14493465e-01 -6.35966063e-01 -5.99156678e-01
1.74210027e-01 -1.06263077e+00 9.87185180e-01 7.42903054e-01
-1.69276714e-01 2.15948418e-01 -2.44007617e-01 5.23504138e-01
6.80346847e-01 9.83578712e-02 1.29199409e+00 -1.22608793e+00
4.24617112e-01 -5.56126952e-01 -6.73735619e-01 -1.10177732e+00
4.53114748e-01 -9.80568171e-01 2.99487680e-01 -2.08610153e+00
-3.98088157e-01 -9.95543242e-01 -8.80669430e-03 8.05526316e-01
-5.44769801e-02 4.74946558e-01 -1.85064733e-01 2.14921385e-01
-6.80587649e-01 6.23114944e-01 8.27132523e-01 -2.38573521e-01
-3.25633347e-01 -1.52316719e-01 -4.36147511e-01 4.23355132e-01
1.37803090e+00 -2.05270156e-01 -6.16979957e-01 -6.41805351e-01
2.73263901e-01 -1.29737094e-01 5.83143651e-01 -1.25762618e+00
3.47133875e-01 -3.42522562e-01 1.06170438e-01 -1.33193696e+00
3.66802096e-01 -4.64961648e-01 3.18634510e-01 8.05462673e-02
-9.83092859e-02 -3.52977403e-02 4.45973307e-01 4.69278932e-01
-2.35764936e-01 4.59329873e-01 4.91279334e-01 2.75719110e-02
-1.59652054e+00 5.73534906e-01 -4.94941652e-01 1.71340797e-02
1.13315475e+00 -2.35768348e-01 -4.89037037e-01 -3.33281398e-01
-2.46811345e-01 8.52911472e-01 5.26632726e-01 7.31320977e-01
5.17109752e-01 -1.23490322e+00 -8.88684869e-01 5.07901728e-01
5.88849127e-01 5.98320782e-01 4.81596828e-01 5.61719000e-01
-9.07404304e-01 7.29173005e-01 -4.34577316e-01 -8.56804907e-01
-7.27533817e-01 2.45997563e-01 3.14328969e-01 2.62824863e-01
-1.16179264e+00 8.23223352e-01 3.31218988e-01 -1.13229835e+00
-2.38032386e-01 -2.22446457e-01 -2.07642689e-01 -4.73254062e-02
3.99405032e-01 4.09668028e-01 1.13094464e-01 -1.02539122e+00
-3.21452379e-01 5.57499588e-01 2.59979367e-01 -2.60352969e-01
1.39420938e+00 -2.61694223e-01 2.88462222e-01 4.32172507e-01
1.18738127e+00 -4.86276269e-01 -1.75120533e+00 -2.26693854e-01
-2.85730362e-02 -4.53604102e-01 4.86125171e-01 -7.46076643e-01
-9.87761378e-01 8.34538043e-01 6.13172531e-01 -2.66239882e-01
5.57307661e-01 3.31394911e-01 9.90580738e-01 4.14012343e-01
5.30914545e-01 -1.23118341e+00 -5.30056775e-01 7.21409500e-01
7.31747270e-01 -1.68942118e+00 -2.17774019e-01 -4.38657016e-01
-9.28274274e-01 6.93955958e-01 9.35981929e-01 -1.42306507e-01
9.17069674e-01 2.93303251e-01 3.13230306e-01 -5.35419703e-01
-5.09989798e-01 -5.70607007e-01 3.32324147e-01 8.54928851e-01
-2.01801404e-01 8.15995574e-01 5.68621576e-01 1.53911218e-01
-1.03786266e+00 -1.06817566e-01 3.00088316e-01 7.33956873e-01
-8.30257177e-01 -5.50164044e-01 -1.88924789e-01 6.44084036e-01
5.42082250e-01 2.18504071e-02 -1.25391886e-01 7.02262104e-01
1.41387790e-01 7.71406114e-01 5.95957100e-01 -7.96120286e-01
4.64325368e-01 -2.49016240e-01 -1.78422987e-01 -4.06337976e-01
1.88185126e-01 -4.70077515e-01 3.91468406e-01 -9.65128660e-01
1.26009062e-01 -6.74000382e-01 -1.47157586e+00 -4.83716995e-01
3.63661081e-01 -1.16465881e-01 1.17503881e+00 8.25958848e-01
5.33345401e-01 4.09416616e-01 5.73339283e-01 -1.29068339e+00
1.45564228e-01 -8.33768308e-01 -4.30171251e-01 9.33794379e-02
4.40718710e-01 -7.80120552e-01 4.18862738e-02 -1.24911994e-01]
|
[7.994074821472168, -1.8607659339904785]
|
dce863b5-eb24-4091-8492-954df6bbd54e
|
unsupervised-learning-of-full-waveform-1
|
2110.07584
| null |
https://arxiv.org/abs/2110.07584v2
|
https://arxiv.org/pdf/2110.07584v2.pdf
|
Unsupervised Learning of Full-Waveform Inversion: Connecting CNN and Partial Differential Equation in a Loop
|
This paper investigates unsupervised learning of Full-Waveform Inversion (FWI), which has been widely used in geophysics to estimate subsurface velocity maps from seismic data. This problem is mathematically formulated by a second order partial differential equation (PDE), but is hard to solve. Moreover, acquiring velocity map is extremely expensive, making it impractical to scale up a supervised approach to train the mapping from seismic data to velocity maps with convolutional neural networks (CNN). We address these difficulties by integrating PDE and CNN in a loop, thus shifting the paradigm to unsupervised learning that only requires seismic data. In particular, we use finite difference to approximate the forward modeling of PDE as a differentiable operator (from velocity map to seismic data) and model its inversion by CNN (from seismic data to velocity map). Hence, we transform the supervised inversion task into an unsupervised seismic data reconstruction task. We also introduce a new large-scale dataset OpenFWI, to establish a more challenging benchmark for the community. Experiment results show that our model (using seismic data alone) yields comparable accuracy to the supervised counterpart (using both seismic data and velocity map). Furthermore, it outperforms the supervised model when involving more seismic data.
|
['Youzuo Lin', 'Zicheng Liu', 'Sharon Xiaolei Huang', 'Yinpeng Chen', 'Xitong Zhang', 'Peng Jin']
|
2021-10-14
|
unsupervised-learning-of-full-waveform
|
https://openreview.net/forum?id=izvwgBic9q
|
https://openreview.net/pdf?id=izvwgBic9q
|
iclr-2022-4
|
['geophysics']
|
['miscellaneous']
|
[ 1.89409107e-01 6.90238327e-02 3.97590220e-01 -2.49887198e-01
-1.08020604e+00 -4.71722424e-01 4.62502480e-01 -6.28538579e-02
-7.75819123e-01 5.47622323e-01 1.37705177e-01 -5.77802300e-01
-1.43393829e-01 -1.15400624e+00 -1.06404638e+00 -9.53966022e-01
-3.64284575e-01 6.36858225e-01 3.40369791e-01 -2.90065259e-01
4.16736513e-01 3.94221455e-01 -1.27368093e+00 -1.40307873e-01
1.01077986e+00 1.01959074e+00 1.31276801e-01 7.35164881e-01
-1.38194665e-01 9.55119073e-01 -2.63063699e-01 1.71694279e-01
8.99168476e-02 -3.19941521e-01 -1.15684390e+00 -3.35186511e-01
1.90769002e-01 -6.18669212e-01 -4.64623123e-01 7.81443954e-01
4.71026808e-01 2.84101754e-01 8.97963524e-01 -1.08009410e+00
-3.63336235e-01 4.22252953e-01 -5.99748433e-01 1.74605340e-01
5.24014123e-02 -2.48567417e-01 1.00152457e+00 -9.23566461e-01
3.55160683e-01 8.39523733e-01 1.13629293e+00 1.27636015e-01
-1.25973666e+00 -3.90700459e-01 -3.87240410e-01 5.81559092e-02
-1.28368545e+00 -3.97076696e-01 1.01944780e+00 -6.63483441e-01
8.55382800e-01 8.59540626e-02 5.59941113e-01 6.83278501e-01
-2.75223225e-01 9.49201584e-01 5.73614240e-01 -2.27537453e-02
4.63140637e-01 -5.20102501e-01 2.94502676e-02 4.80113566e-01
1.64363924e-02 1.45126358e-01 -5.55954874e-01 -4.46600206e-02
9.85758185e-01 -3.58463973e-01 -4.30945396e-01 -1.01132737e-03
-1.29872704e+00 1.03189790e+00 3.78468692e-01 5.67173511e-02
-4.61149335e-01 4.54082221e-01 3.81573081e-01 4.62826908e-01
6.25620186e-01 5.13242662e-01 -3.80693972e-01 -3.85429651e-01
-1.31842124e+00 6.14732683e-01 9.80411410e-01 5.16375363e-01
1.17137933e+00 5.14334917e-01 5.94129145e-01 6.51357770e-01
4.19011205e-01 8.76261353e-01 3.45561504e-01 -1.19616354e+00
5.21520436e-01 -2.73284074e-02 2.76714385e-01 -1.11605501e+00
-6.79974139e-01 -1.44179359e-01 -1.24262607e+00 6.06574491e-02
6.40648127e-01 -4.15654212e-01 -4.77068126e-01 1.40258288e+00
-5.23233637e-02 9.06948626e-01 2.50872254e-01 1.12408435e+00
8.26410174e-01 1.08786011e+00 -2.68231332e-01 3.88279767e-03
9.38171625e-01 -6.54916883e-01 -6.09768093e-01 -3.46514374e-01
1.03161275e+00 -2.35242501e-01 8.53111148e-01 3.39163214e-01
-1.12120414e+00 -4.26175237e-01 -1.18500066e+00 -6.97576627e-02
-2.76596338e-01 -1.46974802e-01 6.46157026e-01 7.61083812e-02
-1.19613719e+00 9.56434250e-01 -1.19728398e+00 1.96795076e-01
2.25767434e-01 2.19958916e-01 -3.83655936e-01 2.88790137e-01
-1.68068242e+00 5.19015074e-01 2.65735209e-01 5.96623421e-01
-7.80454755e-01 -1.13173115e+00 -1.33358765e+00 1.43404864e-02
-7.65783265e-02 -1.74613789e-01 1.23214066e+00 -4.63296056e-01
-1.49962080e+00 3.43235582e-01 -3.35853510e-02 -5.83466589e-01
6.75156891e-01 -7.21179694e-02 3.73228975e-02 1.58203691e-01
2.35508710e-01 2.67333150e-01 6.96678281e-01 -1.07623816e+00
-3.98071498e-01 7.61598861e-03 1.49848517e-02 -3.18312705e-01
-3.98194849e-01 -4.25025046e-01 -1.10227011e-01 -4.65784281e-01
4.92253155e-01 -5.68848193e-01 -1.70206502e-01 -1.16753273e-01
-2.65651107e-01 7.81114176e-02 6.83356702e-01 -1.14596736e+00
9.85808969e-01 -1.94670129e+00 3.44962627e-01 3.14585507e-01
1.42047629e-01 2.15001833e-02 -3.87362391e-02 3.72378409e-01
-6.29255176e-02 2.42245831e-02 -1.21001828e+00 -4.20287430e-01
-1.56770170e-01 5.28019965e-01 -4.09114778e-01 6.20444715e-01
4.97809529e-01 9.39281821e-01 -1.08159828e+00 -3.05836231e-01
-1.03311032e-01 4.77268726e-01 -9.40447092e-01 3.10681134e-01
5.26363542e-03 1.07686841e+00 -4.50127929e-01 2.90093362e-01
1.31571090e+00 -1.66666120e-01 -2.31844690e-02 -2.61222422e-01
-5.65719843e-01 1.49435803e-01 -1.37027562e+00 1.82744944e+00
-8.16900671e-01 7.69321859e-01 -1.48922345e-02 -1.95381975e+00
1.06686044e+00 3.77440631e-01 7.41240919e-01 -7.54683435e-01
-1.76128194e-01 6.71832979e-01 -1.22974940e-01 -1.02695858e+00
3.24641168e-01 -2.65896410e-01 -1.67229146e-01 6.66355491e-01
1.80387363e-01 -3.87378186e-01 1.25507519e-01 1.30740562e-02
1.01123893e+00 3.64739299e-01 -5.50087929e-01 -3.69092107e-01
5.37801981e-01 1.57437995e-01 4.23164278e-01 6.82326019e-01
3.87037098e-01 1.00037611e+00 6.85537696e-01 -4.41049784e-01
-1.13659501e+00 -1.04056585e+00 -1.62122801e-01 7.75447249e-01
-6.61233589e-02 1.75767139e-01 -4.77478415e-01 -7.74248913e-02
5.12098894e-02 2.36439198e-01 -6.83411896e-01 -3.71624343e-02
-1.02221489e+00 -9.74108934e-01 9.52437222e-01 1.02399933e+00
8.70742917e-01 -9.05993521e-01 -6.86253965e-01 5.03635645e-01
-6.62331402e-01 -1.05162549e+00 -7.12478161e-02 2.96807557e-01
-8.49623561e-01 -7.22340286e-01 -1.01784980e+00 -9.31178391e-01
4.13382471e-01 -1.88200861e-01 8.79235327e-01 3.44557948e-02
9.36254188e-02 1.17063008e-01 -3.34015101e-01 -2.78237611e-01
-3.08496296e-01 2.25651205e-01 -2.67551690e-01 2.91094899e-01
-2.41184205e-01 -8.46433103e-01 -5.07474124e-01 5.31464331e-02
-1.10575533e+00 -7.50735849e-02 1.84078380e-01 8.86300445e-01
3.34561974e-01 -7.46612102e-02 5.70970833e-01 -8.18867087e-01
3.93207937e-01 -6.99923158e-01 -8.29197645e-01 -1.85440049e-01
2.29252726e-01 1.76254898e-01 4.46878225e-01 -2.17623025e-01
-1.26869154e+00 -8.31839293e-02 -8.43944430e-01 -1.29637897e-01
1.93307683e-01 1.36947322e+00 2.54828423e-01 -2.48653069e-01
5.33683896e-01 5.28535724e-01 2.35391304e-01 -7.27921307e-01
-9.51248705e-02 6.95578814e-01 1.02305722e+00 -5.31723797e-01
9.52732325e-01 8.99550378e-01 1.73912376e-01 -1.20419049e+00
-7.37748325e-01 -5.86770952e-01 -9.62882876e-01 -1.05183281e-01
9.32612121e-01 -8.10948074e-01 -7.96868443e-01 7.92618930e-01
-1.39783323e+00 -8.05794775e-01 -6.15236834e-02 6.84056222e-01
-5.60023725e-01 5.23509443e-01 -6.92918181e-01 -7.36844242e-01
-1.09417215e-01 -9.62274492e-01 1.29546690e+00 -1.47858337e-01
-9.08560958e-03 -1.48577142e+00 2.72546798e-01 -1.24098815e-01
5.73363125e-01 5.87224185e-01 5.91648757e-01 -2.73708224e-01
-3.81464958e-01 -2.16897354e-01 -3.17471087e-01 3.34866881e-01
-2.48747766e-01 -8.74508396e-02 -1.12872887e+00 -8.36191885e-03
2.65533149e-01 -3.32247853e-01 1.19967008e+00 5.37403762e-01
1.11901629e+00 3.75364274e-02 1.95220053e-01 1.23280847e+00
1.29993391e+00 -1.35967642e-01 7.64623046e-01 4.20228630e-01
8.19378674e-01 6.26761377e-01 3.31566483e-01 6.16986692e-01
4.49585766e-01 1.20352350e-01 4.06257689e-01 -2.43296608e-01
3.07302594e-01 -3.59148346e-02 1.50703639e-03 1.28888655e+00
-7.70021796e-01 6.21047989e-02 -1.48665893e+00 6.73585117e-01
-1.86697185e+00 -7.25153327e-01 -5.39083064e-01 2.04221892e+00
6.42299116e-01 6.10326650e-03 -2.28473559e-01 6.99038208e-01
5.39113730e-02 1.84543535e-01 -1.97189763e-01 -1.04584135e-01
-1.24075130e-01 4.57383603e-01 6.31454706e-01 8.20797861e-01
-1.30976713e+00 5.72313428e-01 6.01177645e+00 2.01490983e-01
-1.51186216e+00 1.74680740e-01 4.30002272e-01 6.35450244e-01
-5.64074457e-01 -2.06838220e-01 -2.32694566e-01 2.42572978e-01
1.13276398e+00 1.69725224e-01 2.72974521e-01 2.53182739e-01
3.09900820e-01 -6.85210153e-02 -1.04379702e+00 7.61850476e-01
-3.48066598e-01 -1.74921048e+00 -2.45774657e-01 -1.27962187e-01
5.39334059e-01 2.31267914e-01 -2.81200588e-01 3.06373477e-01
-4.62722033e-02 -9.70263422e-01 8.12488496e-01 7.07676589e-01
8.37206185e-01 -6.51812673e-01 9.63563263e-01 5.71891665e-01
-1.38290155e+00 2.11097077e-01 -4.05051410e-01 -4.16278720e-01
5.61262369e-01 4.03083116e-01 -2.91606456e-01 7.20193624e-01
7.86175370e-01 1.06854224e+00 8.18232447e-02 1.01914752e+00
-7.86877945e-02 8.86884570e-01 -3.80771130e-01 4.40961421e-01
6.37866676e-01 -4.49949682e-01 2.28330031e-01 1.13916421e+00
6.43121481e-01 1.27171380e-02 -7.26574957e-02 8.52610171e-01
-1.11239282e-02 -1.65682346e-01 -4.90675598e-01 8.30647361e-04
1.81887180e-01 8.65330577e-01 -5.74400604e-01 -2.95031101e-01
-5.60126364e-01 6.98848844e-01 1.35045126e-01 6.74082637e-01
-9.29380059e-01 -5.24015486e-01 4.53800142e-01 1.79417923e-01
2.86030710e-01 -5.18113852e-01 -2.50566155e-01 -1.06166160e+00
-1.36826098e-01 -7.53843263e-02 2.12123811e-01 -7.21648037e-01
-1.00370455e+00 2.71908373e-01 1.32394835e-01 -1.41783369e+00
-3.72322083e-01 -7.26925433e-01 -9.43787634e-01 9.70587373e-01
-2.17016244e+00 -7.09966600e-01 -6.97767615e-01 4.01479632e-01
1.09909937e-01 3.90878081e-01 6.79457545e-01 7.27355421e-01
-1.83957353e-01 2.34696329e-01 3.26943427e-01 6.75467074e-01
3.58091056e-01 -1.14475322e+00 7.69160986e-01 5.19358099e-01
-2.92933643e-01 1.11752793e-01 7.12220550e-01 -5.33321798e-01
-1.75273180e+00 -1.05822325e+00 9.45947111e-01 -1.50849624e-02
1.13579869e+00 -3.25117618e-01 -1.72224450e+00 6.73054516e-01
-2.69531637e-01 5.04872620e-01 2.97730595e-01 -5.00897825e-01
2.83317138e-02 1.14404358e-01 -8.29303801e-01 8.72292370e-02
8.91942978e-01 -5.51254570e-01 -2.95311838e-01 1.37253314e-01
6.10328496e-01 -5.41185260e-01 -1.10187197e+00 5.68378508e-01
4.12798882e-01 -6.16113603e-01 1.17577183e+00 -2.59601206e-01
9.55025852e-01 -1.17695302e-01 4.96745817e-02 -1.49799955e+00
2.33986005e-02 -5.85848749e-01 -4.75877896e-02 8.36027920e-01
5.27816355e-01 -9.96691346e-01 7.13716209e-01 2.10522696e-01
-3.57128263e-01 -5.47917902e-01 -8.62649083e-01 -7.24581718e-01
5.78956604e-01 -8.38064790e-01 5.07274568e-01 1.06478822e+00
-1.26380071e-01 -2.78522700e-01 -3.48879963e-01 5.43375671e-01
6.40588343e-01 3.77758406e-02 4.64983344e-01 -1.40737081e+00
-2.37764537e-01 -2.37894505e-01 -2.48878419e-01 -1.33460951e+00
2.15037659e-01 -8.44738424e-01 6.45789385e-01 -1.57095754e+00
-5.10789096e-01 -4.25304353e-01 -1.02583960e-01 2.92615443e-01
5.17717116e-02 5.53995132e-01 -3.54791909e-01 3.75874639e-01
1.11148722e-01 6.10329449e-01 1.31710207e+00 -3.51538539e-01
-2.01770626e-02 -1.46117836e-01 -8.11292157e-02 1.03022003e+00
6.52884126e-01 -4.42179829e-01 -3.35618287e-01 -1.04663396e+00
5.70041418e-01 6.50206745e-01 5.05228102e-01 -1.00723648e+00
5.62276542e-01 -6.08318113e-02 9.09997057e-03 -6.40650213e-01
7.88066015e-02 -3.92927945e-01 -2.18115717e-01 4.23779517e-01
-1.08276621e-01 -1.22122362e-01 1.98418677e-01 2.51664340e-01
-7.29533732e-01 -4.25696045e-01 5.20351231e-01 -4.44775298e-02
-9.46823657e-01 4.61901337e-01 -5.29026568e-01 3.84003013e-01
2.45227754e-01 -6.16944842e-02 3.21274072e-01 -4.10914689e-01
-7.45120287e-01 1.78364530e-01 -9.44831297e-02 -2.31916919e-01
8.43982935e-01 -1.09469271e+00 -1.08808911e+00 6.37325168e-01
-1.72307715e-02 7.42535472e-01 3.76701772e-01 1.26736116e+00
-1.29519475e+00 -6.47652373e-02 7.27976253e-03 -8.87376666e-01
-1.98409051e-01 -8.88646320e-02 4.31936949e-01 -1.53013498e-01
-8.49269807e-01 1.18689573e+00 7.73732644e-03 -8.64753604e-01
1.07169032e-01 -4.77190405e-01 -2.57447213e-01 2.94685751e-01
5.30206025e-01 5.01346469e-01 2.36108005e-01 -5.30445397e-01
-1.71388701e-01 7.27859676e-01 6.04110956e-01 -2.73155004e-01
1.83502507e+00 1.42816767e-01 -2.07277551e-01 5.24915278e-01
1.72056091e+00 -4.89880711e-01 -1.45845568e+00 -3.42506230e-01
1.34703651e-01 -1.86013088e-01 2.75831044e-01 1.16580792e-01
-1.36830771e+00 1.32089615e+00 -5.45941181e-02 2.57087886e-01
8.25092316e-01 -2.10777104e-01 1.06549513e+00 6.35679483e-01
1.09792203e-01 -9.70161498e-01 1.32751375e-01 1.00531840e+00
8.81992400e-01 -1.34553373e+00 -3.14783990e-01 -2.84323812e-01
-3.40378642e-01 1.33339846e+00 1.56726643e-01 -5.31972587e-01
1.07843375e+00 8.29708338e-01 1.26094550e-01 -1.75173119e-01
-1.08045667e-01 1.09690696e-01 9.99426693e-02 4.71780747e-01
3.95931304e-01 -3.13914210e-01 1.73460677e-01 3.57156426e-01
-3.59523147e-01 1.35779947e-01 5.25264978e-01 1.17857754e+00
-3.25131267e-01 -6.27038896e-01 -3.90049309e-01 3.56748402e-01
-4.44746278e-02 -7.34640732e-02 4.48787570e-01 6.97119534e-01
-8.87489468e-02 5.21706998e-01 6.24515951e-01 -6.23237295e-03
2.33466998e-01 -3.39714736e-02 1.52484745e-01 -2.55274087e-01
3.55946571e-02 2.74308957e-02 -8.61915872e-02 -3.88137817e-01
-5.25789678e-01 -6.77570343e-01 -1.72675931e+00 -3.54007632e-01
5.83371148e-02 3.02123845e-01 6.07717872e-01 1.12771642e+00
-3.66687894e-01 8.53519380e-01 5.56356132e-01 -1.53258359e+00
-4.33523506e-01 -1.01569557e+00 -8.36072981e-01 3.70209634e-01
7.28927553e-01 -7.72002757e-01 -6.09007001e-01 2.73329169e-01]
|
[6.861297130584717, 2.513056755065918]
|
4ff04f5e-37f2-497c-9d20-d92b03fbf056
|
photo-pre-training-but-for-sketch
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Li_Photo_Pre-Training_but_for_Sketch_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Li_Photo_Pre-Training_but_for_Sketch_CVPR_2023_paper.pdf
|
Photo Pre-Training, but for Sketch
|
The sketch community has faced up to its unique challenges over the years, that of data scarcity however still remains the most significant to date. This lack of sketch data has imposed on the community a few "peculiar" design choices -- the most representative of them all is perhaps the coerced utilisation of photo-based pre-training (i.e., no sketch), for many core tasks that otherwise dictates specific sketch understanding. In this paper, we ask just the one question -- can we make such photo-based pre-training, to actually benefit sketch? Our answer lies in cultivating the topology of photo data learned at pre-training, and use that as a "free" source of supervision for downstream sketch tasks. In particular, we use fine-grained sketch-based image retrieval (FG-SBIR), one of the most studied and data-hungry sketch tasks, to showcase our new perspective on pre-training. In this context, the topology-informed supervision learned from photos act as a constraint that take effect at every fine-tuning step -- neighbouring photos in the pre-trained model remain neighbours under each FG-SBIR updates. We further portray this neighbourhood consistency constraint as a photo ranking problem and formulate it into a neat cross-modal triplet loss. We also show how this target is better leveraged as a meta objective rather than optimised in parallel with the main FG-SBIR objective. With just this change on pre-training, we beat all previously published results on all five product-level FG-SBIR benchmarks with significant margins (sometimes >10%). And the most beautiful thing, as we note, is such gigantic leap is made possible with just a few extra lines of code! Our implementation is available at https://github.com/KeLi-SketchX/Photo-Pre-Training-But-for-Sketch
|
['Yi-Zhe Song', 'Kaiyue Pang', 'Ke Li']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['sketch-based-image-retrieval']
|
['computer-vision']
|
[ 2.95027405e-01 1.84651375e-01 -2.86066324e-01 -3.19764584e-01
-8.06413054e-01 -9.24010158e-01 1.00985396e+00 -3.33396643e-01
-2.13396579e-01 5.52359641e-01 3.61367851e-01 -2.32343167e-01
-3.86097670e-01 -5.99074125e-01 -8.59207809e-01 -7.57026374e-01
1.01625487e-01 5.29400587e-01 3.44165079e-02 -4.98918861e-01
4.05868262e-01 7.88785636e-01 -1.45428205e+00 3.75963539e-01
3.00351471e-01 1.02984071e+00 1.28958553e-01 5.29920280e-01
-8.40018317e-02 2.90389359e-01 -6.59310520e-02 -9.18985426e-01
6.94578528e-01 -1.75184831e-01 -8.15991163e-01 9.14146900e-02
9.41089511e-01 -2.57800132e-01 -3.55871409e-01 5.70496440e-01
5.46540499e-01 -1.25800585e-02 7.00038731e-01 -1.19866574e+00
-7.16945767e-01 4.03479069e-01 -4.80029166e-01 -2.21750885e-01
1.81763142e-01 3.98316830e-01 1.48842871e+00 -1.11173105e+00
9.29218888e-01 1.15390098e+00 7.48921514e-01 5.28425395e-01
-1.54693592e+00 -4.60995764e-01 1.71445221e-01 -1.12973131e-01
-1.49283504e+00 -6.60409629e-01 1.00298452e+00 -4.47322130e-01
6.60784423e-01 5.05277097e-01 6.65069103e-01 1.17535484e+00
-2.90497929e-01 8.56844842e-01 1.00633299e+00 -4.37767088e-01
-6.14577085e-02 1.25700295e-01 -2.04151243e-01 7.26587176e-01
2.92183049e-02 2.67018497e-01 -4.85393345e-01 -3.25096212e-02
1.05302048e+00 7.79288784e-02 -2.63195336e-01 -7.43602753e-01
-1.22153282e+00 6.10329568e-01 6.36445165e-01 4.08183455e-01
3.53195854e-02 2.83579648e-01 3.38845342e-01 5.64258754e-01
3.12998384e-01 5.24373293e-01 -3.96479368e-01 2.40324736e-02
-1.02856731e+00 2.45567754e-01 6.88671708e-01 8.14946711e-01
9.51576471e-01 -2.46588394e-01 -9.86320432e-03 1.01416731e+00
3.07020843e-02 3.11214715e-01 -5.50186671e-02 -8.36486220e-01
2.96225846e-01 3.03520232e-01 -2.03937367e-01 -1.07443404e+00
2.20897440e-02 -3.60377043e-01 -8.98296297e-01 3.68590176e-01
6.47696137e-01 3.06495577e-01 -9.05947804e-01 1.82973659e+00
2.24281505e-01 -6.20871335e-02 -5.33926129e-01 1.09135139e+00
4.75657433e-01 3.73342723e-01 1.11722900e-02 2.62864202e-01
1.28466463e+00 -8.96603942e-01 -1.03585951e-01 -9.17291567e-02
2.09486544e-01 -9.67933893e-01 1.31116390e+00 5.56796312e-01
-1.02453935e+00 -4.85624164e-01 -1.07506514e+00 -3.93664747e-01
-6.70308232e-01 1.19432203e-01 8.94389689e-01 3.57841522e-01
-1.30203974e+00 9.47809160e-01 -2.35788643e-01 -7.51160979e-01
4.98953670e-01 3.30840945e-01 -7.33905792e-01 -3.82891804e-01
-8.67824852e-01 8.36249292e-01 -2.08107561e-01 9.42845270e-02
-6.13058746e-01 -9.22195256e-01 -4.40114170e-01 -9.83167514e-02
6.75070822e-01 -9.87908185e-01 8.68171096e-01 -1.27196109e+00
-1.38595045e+00 1.25200820e+00 2.84967832e-02 -3.91330086e-02
9.25865531e-01 9.29246768e-02 -7.36123025e-02 -2.58326624e-02
-8.00004750e-02 1.13456059e+00 1.20541871e+00 -1.56078196e+00
-1.24660529e-01 -2.70179480e-01 1.75431922e-01 6.79561421e-02
-5.84953837e-02 -1.95941404e-01 -7.79294550e-01 -8.06379795e-01
-1.68162346e-01 -1.05064690e+00 6.38496876e-02 5.64290702e-01
-2.81612635e-01 -3.66190523e-01 5.99658549e-01 -2.92640597e-01
1.02664840e+00 -2.08999085e+00 2.77540445e-01 3.03980321e-01
7.90721104e-02 2.59137273e-01 -4.93396252e-01 9.29208457e-01
-2.62430459e-01 3.60552043e-01 -3.91316652e-01 -6.09540284e-01
3.73551071e-01 2.91011870e-01 -6.57181442e-01 3.24611276e-01
5.79228640e-01 1.21368206e+00 -7.26589739e-01 -3.78658205e-01
4.26338464e-01 7.08644807e-01 -4.18285102e-01 -4.24124440e-03
-3.03213716e-01 3.47127736e-01 -2.41728649e-01 6.37015045e-01
7.01903820e-01 -3.39679301e-01 1.64846435e-01 -4.76465881e-01
-1.48965389e-01 4.11159433e-02 -1.23027992e+00 2.22766328e+00
-3.09718579e-01 4.38269079e-01 1.69201896e-01 -9.54020083e-01
6.69389784e-01 -4.88738436e-03 5.58098078e-01 -8.14138591e-01
-1.92234084e-01 2.86150873e-01 -2.57969111e-01 -1.77768499e-01
4.75776076e-01 -3.22197944e-01 5.79654947e-02 3.86065394e-01
1.29281089e-01 -3.26177835e-01 2.66146045e-02 3.14062744e-01
9.21512663e-01 5.62915683e-01 4.10728268e-02 -3.13122869e-01
3.90817791e-01 -3.66114490e-02 1.85859740e-01 6.22159004e-01
1.40664175e-01 1.16850233e+00 4.99331653e-01 -5.41469812e-01
-1.54826021e+00 -1.07297742e+00 -2.99032539e-01 1.19103587e+00
-1.50228525e-02 -4.34345692e-01 -3.14260006e-01 -7.43431628e-01
3.82152349e-01 1.38237253e-01 -7.33957291e-01 2.30261877e-01
-5.72072983e-01 -2.44740441e-01 4.68952984e-01 2.31736749e-01
1.44449860e-01 -1.06329036e+00 -1.31257415e-01 -9.07253623e-02
3.69229496e-01 -8.16129982e-01 -6.03347301e-01 3.62128690e-02
-6.50279522e-01 -8.87400150e-01 -1.12605262e+00 -5.62169015e-01
6.43507600e-01 3.55536729e-01 1.41648579e+00 5.38914502e-01
-4.95491058e-01 6.22973561e-01 -2.77208924e-01 -3.74848396e-02
-3.33806872e-02 1.47950321e-01 -3.02808434e-01 1.12716556e-01
-1.11322679e-01 -1.01626921e+00 -9.07792866e-01 4.64050651e-01
-1.09260011e+00 1.23815194e-01 9.88487363e-01 8.44352007e-01
5.14086783e-01 -4.53221887e-01 4.96558934e-01 -8.71980727e-01
3.00268143e-01 -2.06861436e-01 -4.17475790e-01 5.30513227e-01
-5.86273193e-01 1.23040408e-01 5.76238871e-01 -3.67335320e-01
-6.86619103e-01 1.54005215e-01 -2.53998399e-01 -6.76503062e-01
-7.38122761e-02 1.16800644e-01 -2.34477475e-01 -2.69669592e-01
5.10006428e-01 1.24520957e-01 7.95223340e-02 -7.24924088e-01
8.00676525e-01 2.84434646e-01 5.05678713e-01 -1.04136050e+00
1.20700967e+00 7.75766075e-01 2.69899130e-01 -9.10055459e-01
-5.61722875e-01 -4.51725721e-01 -6.60657585e-01 -9.80895460e-02
5.63753009e-01 -7.65259027e-01 -8.01563442e-01 2.08318412e-01
-9.26121652e-01 -5.53420722e-01 -4.19248611e-01 -2.53393203e-01
-5.82629561e-01 4.49143440e-01 -3.15557599e-01 -7.94426560e-01
-2.42468521e-01 -9.83632207e-01 1.31536627e+00 -1.64938375e-01
1.47138983e-02 -9.20580149e-01 9.47648585e-02 2.90597200e-01
5.18256247e-01 2.23946229e-01 8.48471880e-01 -1.75918341e-01
-1.00828314e+00 -1.73369963e-02 -6.42476141e-01 3.40771854e-01
-2.68510785e-02 1.41074911e-01 -1.18863857e+00 -3.09705049e-01
-4.13187534e-01 -5.09376347e-01 1.00409055e+00 -3.06566674e-02
1.18387187e+00 -2.81958044e-01 -1.05693653e-01 7.16470182e-01
1.80948400e+00 -5.73438346e-01 7.61077464e-01 2.13785302e-02
8.26041996e-01 7.20737159e-01 4.87858176e-01 1.68089688e-01
3.46796691e-01 1.03441691e+00 3.09826136e-01 -2.43385762e-01
-6.02191746e-01 -6.70968771e-01 1.00270651e-01 5.56091845e-01
-2.41017923e-01 -7.53445551e-02 -6.15728617e-01 5.02853453e-01
-1.84890258e+00 -1.05740237e+00 9.16488692e-02 2.36699677e+00
7.80287802e-01 -1.64770588e-01 4.19786304e-01 1.22565441e-02
3.61291260e-01 4.49732542e-01 -3.92495543e-01 -2.28688315e-01
-2.53217012e-01 4.87071157e-01 4.48691905e-01 6.09325349e-01
-1.01195514e+00 9.16855812e-01 5.27369356e+00 1.07667744e+00
-1.34712362e+00 -1.28293112e-01 7.71892667e-01 -9.31861550e-02
-6.30490184e-01 2.86171049e-01 -4.65969622e-01 4.26686704e-01
2.48501360e-01 2.79074103e-01 7.65617490e-01 6.46992803e-01
-2.42378693e-02 -6.99292347e-02 -1.38695967e+00 1.14341021e+00
-4.39287163e-02 -1.59266329e+00 2.39693612e-01 2.19259381e-01
5.63326538e-01 2.79599335e-02 2.67182797e-01 1.33175701e-01
1.52742833e-01 -1.17519450e+00 9.16684985e-01 5.79965651e-01
1.27803314e+00 -4.21044558e-01 9.00559649e-02 5.97404242e-02
-1.24574685e+00 7.96719715e-02 -4.03992414e-01 1.41268419e-02
1.22573011e-01 6.09963357e-01 -4.14747894e-01 8.07261646e-01
3.79776090e-01 7.47646809e-01 -5.43942988e-01 8.74702632e-01
-1.25359371e-01 1.82815671e-01 -4.76706415e-01 2.34657869e-01
3.62149715e-01 -3.41439724e-01 4.97313529e-01 1.27782798e+00
1.41955093e-01 3.12580392e-02 5.97425774e-02 8.77004087e-01
-1.76950261e-01 -1.81148089e-02 -8.33367527e-01 1.59580410e-02
2.91428983e-01 1.53159618e+00 -5.69500268e-01 -7.54420012e-02
-2.85441399e-01 1.09535468e+00 5.42781711e-01 2.28659526e-01
-5.51214397e-01 -1.02637112e-02 6.71964288e-01 4.64775532e-01
4.90974605e-01 -1.48011461e-01 -4.46532041e-01 -1.24003625e+00
2.67310083e-01 -8.20164263e-01 2.61972368e-01 -8.79863143e-01
-1.73052835e+00 2.58520097e-01 -2.04127401e-01 -1.11784184e+00
-6.97138458e-02 -6.89782321e-01 -7.02384293e-01 8.08820248e-01
-1.85561752e+00 -1.73532903e+00 -1.24726705e-01 5.27972162e-01
4.54629153e-01 3.73503536e-01 5.97528517e-01 5.77312410e-01
-2.89994895e-01 8.17303956e-01 -7.28935674e-02 9.56364572e-02
1.05911160e+00 -1.23083031e+00 4.28256303e-01 4.97018725e-01
5.68721354e-01 6.95168197e-01 5.39357424e-01 -4.51366484e-01
-1.87818551e+00 -7.04392791e-01 8.63877535e-01 -8.42034638e-01
7.49560654e-01 -6.87763989e-01 -7.42130220e-01 4.91630763e-01
1.36770517e-01 1.79661006e-01 1.03362083e-01 2.35969141e-01
-7.79875875e-01 -2.66808808e-01 -9.33213949e-01 5.77792287e-01
1.59318876e+00 -7.27424800e-01 -3.03424776e-01 2.37808689e-01
5.20599127e-01 -1.45345286e-01 -9.48604941e-01 2.45841205e-01
9.97858047e-01 -1.13995266e+00 1.42870188e+00 -5.51458180e-01
5.82989931e-01 -3.08371902e-01 -1.92888886e-01 -9.73399758e-01
-3.05316687e-01 -8.47897291e-01 4.89731759e-01 1.50674331e+00
3.51127982e-01 -4.00923461e-01 1.00823939e+00 6.38009846e-01
-4.68045063e-02 -1.00355160e+00 -7.26896346e-01 -8.25493217e-01
2.32248142e-01 -3.89016479e-01 4.99921173e-01 1.09719002e+00
-1.74976736e-01 4.04740602e-01 -4.91343617e-01 -2.83595085e-01
6.99494898e-01 6.87827840e-02 1.21123719e+00 -1.10178781e+00
-4.20366496e-01 -7.10625291e-01 -2.73477554e-01 -1.17271554e+00
-1.61966741e-01 -9.87395465e-01 -2.18381628e-01 -1.28009689e+00
2.29240492e-01 -9.45033550e-01 -8.55959505e-02 6.21942401e-01
6.77068830e-02 5.36105335e-01 6.07156873e-01 4.33435500e-01
-4.74437088e-01 3.14407587e-01 1.34285116e+00 -1.37341782e-01
1.85700282e-01 -2.02722788e-01 -6.22294009e-01 2.22825289e-01
3.38411570e-01 -2.44150400e-01 -4.17880714e-01 -4.58069086e-01
4.90904838e-01 -1.81045607e-01 8.09744835e-01 -7.08756447e-01
2.31912464e-01 -4.29012552e-02 2.80340910e-01 -2.64713466e-01
5.64379692e-01 -9.25193846e-01 5.20063221e-01 -3.47992629e-02
-3.12443763e-01 -1.78210437e-01 3.15382518e-02 4.81164694e-01
2.21929960e-02 -3.69546958e-03 6.85304284e-01 -3.15620720e-01
-7.94634044e-01 5.35846114e-01 4.99671638e-01 -4.35194336e-02
5.96656740e-01 -4.45141345e-01 -2.48718843e-01 -3.03405464e-01
-6.00058258e-01 -7.16004595e-02 9.73072410e-01 4.33072031e-01
3.28227133e-01 -1.34629536e+00 -6.81672096e-01 2.01446578e-01
2.73385465e-01 7.43943974e-02 3.96136999e-01 8.30993712e-01
-9.73367095e-02 4.45442259e-01 -2.25023508e-01 -5.92709601e-01
-1.12629974e+00 6.66006744e-01 1.38811395e-01 -3.09379548e-01
-6.91982031e-01 7.44213343e-01 3.20929408e-01 -4.50615227e-01
2.06397861e-01 -8.61018896e-02 4.88788813e-01 6.73931912e-02
1.96505785e-01 4.80947383e-02 1.99072585e-01 -3.81545156e-01
-2.83225149e-01 9.44979072e-01 8.18388835e-02 -1.85917795e-01
1.63644958e+00 5.49608618e-02 -2.69973055e-02 2.64622450e-01
1.32974589e+00 1.46656141e-01 -1.49000633e+00 -1.72090620e-01
-7.94701278e-02 -6.48657382e-01 -2.40426078e-01 -1.22570801e+00
-1.07841253e+00 9.92304564e-01 2.83353657e-01 2.15162694e-01
8.71804774e-01 1.68549255e-01 4.47560579e-01 2.30153292e-01
4.94580060e-01 -8.58833492e-01 8.94876570e-02 1.04176827e-01
1.37405264e+00 -1.24671090e+00 2.97524214e-01 -2.61034846e-01
-5.50081134e-01 1.01211262e+00 8.65036175e-02 -4.78944659e-01
4.63129371e-01 -8.67114812e-02 -1.78510129e-01 -2.90045351e-01
-6.43755019e-01 -3.14458013e-01 5.20872414e-01 4.06124413e-01
3.53018999e-01 -8.75263140e-02 -9.51875672e-02 2.86935680e-02
-1.39153181e-02 1.09220430e-01 -7.50303194e-02 6.99752450e-01
-1.71482861e-01 -1.64651310e+00 -2.21846521e-01 3.93076390e-01
-8.35938007e-02 -1.44764245e-01 -7.83686638e-01 1.06147945e+00
1.20162174e-01 4.38294172e-01 -1.36445552e-01 -3.67061555e-01
3.31910908e-01 7.35134780e-02 8.16179454e-01 -2.99255461e-01
-7.01315463e-01 -1.16002485e-01 1.27205476e-01 -7.12652802e-01
-3.58950257e-01 -5.47285378e-01 -6.41937971e-01 -7.66659796e-01
-1.42520830e-01 -1.46049947e-01 8.40972483e-01 6.28647268e-01
6.11390233e-01 -2.48682890e-02 6.23183489e-01 -1.31431842e+00
-5.41947365e-01 -6.28443897e-01 -2.52863199e-01 6.80822194e-01
3.12327713e-01 -5.97111762e-01 -4.92450535e-01 -2.19630599e-01]
|
[11.625414848327637, 0.5611799359321594]
|
45281a06-2eee-4fb8-bb49-3638940db557
|
dronet-efficient-convolutional-neural-network
|
1807.06789
| null |
http://arxiv.org/abs/1807.06789v1
|
http://arxiv.org/pdf/1807.06789v1.pdf
|
DroNet: Efficient convolutional neural network detector for real-time UAV applications
|
Unmanned Aerial Vehicles (drones) are emerging as a promising technology for
both environmental and infrastructure monitoring, with broad use in a plethora
of applications. Many such applications require the use of computer vision
algorithms in order to analyse the information captured from an on-board
camera. Such applications include detecting vehicles for emergency response and
traffic monitoring. This paper therefore, explores the trade-offs involved in
the development of a single-shot object detector based on deep convolutional
neural networks (CNNs) that can enable UAVs to perform vehicle detection under
a resource constrained environment such as in a UAV. The paper presents a
holistic approach for designing such systems; the data collection and training
stages, the CNN architecture, and the optimizations necessary to efficiently
map such a CNN on a lightweight embedded processing platform suitable for
deployment on UAVs. Through the analysis we propose a CNN architecture that is
capable of detecting vehicles from aerial UAV images and can operate between
5-18 frames-per-second for a variety of platforms with an overall accuracy of
~95%. Overall, the proposed architecture is suitable for UAV applications,
utilizing low-power embedded processors that can be deployed on commercial
UAVs.
|
['Christos-Savvas Bouganis', 'Christos Kyrkou', 'Theocharis Theocharides', 'George Plastiras', 'Stylianos Venieris']
|
2018-07-18
| null | null | null | null |
['one-shot-object-detection', 'object-detection-in-aerial-images']
|
['computer-vision', 'computer-vision']
|
[ 4.05096859e-01 -2.98992962e-01 3.84680182e-01 -9.93541032e-02
1.77929059e-01 -6.33507788e-01 3.26869845e-01 2.31868491e-01
-7.87979305e-01 4.05513644e-02 -9.02234554e-01 -4.17672873e-01
-1.09813318e-01 -8.87087286e-01 -5.10233045e-01 -6.84554279e-01
-5.02032757e-01 -2.49410182e-01 4.13328260e-01 -2.46787086e-01
-3.06833953e-01 9.39227581e-01 -2.06154537e+00 -8.73654634e-02
2.39835665e-01 1.60795951e+00 5.12804747e-01 1.08075213e+00
7.88502038e-01 3.33769262e-01 -7.08496332e-01 -5.93352737e-03
6.88021958e-01 2.16181517e-01 -1.18563928e-01 1.51889101e-01
5.06227970e-01 -9.14449453e-01 -2.47922510e-01 9.22165334e-01
3.02401215e-01 -8.52966458e-02 3.99174958e-01 -1.37554967e+00
2.81147063e-01 -2.18239665e-01 -2.08523154e-01 5.26827335e-01
-2.90650010e-01 5.36551535e-01 5.39832234e-01 -4.84846801e-01
9.25126299e-02 8.23361933e-01 5.74214101e-01 1.85775921e-01
-5.44823766e-01 -6.17982924e-01 -2.21423909e-01 8.57102796e-02
-1.25495887e+00 -3.61721307e-01 2.82559246e-01 -5.68448663e-01
1.12618673e+00 1.17995068e-01 9.71081734e-01 4.34496731e-01
7.06901729e-01 1.89683884e-01 6.35011017e-01 -2.43808001e-01
3.29348356e-01 -3.66242230e-02 -2.16992721e-01 7.53065407e-01
9.58997071e-01 3.42895955e-01 -1.36161959e-02 1.81994215e-01
6.69009864e-01 2.53017783e-01 -2.37971973e-02 -4.85614240e-02
-9.86098230e-01 7.37417400e-01 7.05685377e-01 2.62363181e-02
-6.69402063e-01 4.87186581e-01 5.24768651e-01 2.78575033e-01
2.50365645e-01 4.52287555e-01 -5.32230377e-01 3.19417924e-01
-8.88389289e-01 2.65705109e-01 4.54554319e-01 1.05540609e+00
7.18959212e-01 4.18763250e-01 2.97610819e-01 1.55861173e-02
4.32052493e-01 9.08991158e-01 2.50292476e-03 -5.87716341e-01
1.04185894e-01 8.48250031e-01 1.65469795e-01 -1.06519938e+00
-5.97036541e-01 -3.97246897e-01 -7.65552819e-01 5.80630302e-01
-1.42721578e-01 -5.80754817e-01 -6.74860716e-01 1.03826690e+00
6.27212763e-01 2.14068163e-02 1.75781652e-01 1.04409933e+00
7.55313575e-01 7.99726725e-01 -2.45125145e-02 7.85588622e-02
1.74824977e+00 -5.36388874e-01 -2.49548569e-01 -4.93189216e-01
3.51537645e-01 -4.99137938e-01 3.37522745e-01 3.70647162e-01
-4.67197865e-01 -7.61860073e-01 -1.64337647e+00 1.85540408e-01
-5.90029716e-01 7.70849109e-01 2.16569319e-01 5.92988253e-01
-1.11762559e+00 2.49941036e-01 -1.10301197e+00 -7.49646902e-01
2.42932633e-01 8.53410065e-01 -2.20085338e-01 1.69127300e-01
-7.85304487e-01 7.77377367e-01 5.88503838e-01 4.36179847e-01
-1.41766512e+00 -4.10018206e-01 -8.64302397e-01 2.53120065e-01
2.63540775e-01 -4.71945256e-01 1.28833830e+00 -9.30761099e-01
-1.20828164e+00 4.83525932e-01 5.12487173e-01 -8.94173265e-01
7.19013214e-02 -3.56116623e-01 -3.67180377e-01 3.67057592e-01
-3.23384166e-01 7.76274621e-01 1.20036268e+00 -4.73833203e-01
-1.24682832e+00 -2.73540378e-01 3.70016336e-01 5.88920936e-02
-5.85642397e-01 2.61644095e-01 1.46594033e-01 -6.85212165e-02
-4.14614916e-01 -1.15256715e+00 -1.60134494e-01 3.14751476e-01
-3.46716074e-03 2.88691465e-02 1.62069964e+00 -2.10044891e-01
7.56237984e-01 -1.96534860e+00 -2.13345259e-01 -2.15791315e-02
1.50587961e-01 1.01504529e+00 -1.85953118e-02 3.40902001e-01
2.29958102e-01 -4.77652818e-01 2.87530129e-03 1.42814130e-01
-4.39034224e-01 -1.34235427e-01 -1.42177776e-01 7.28375971e-01
4.95513231e-01 4.45113778e-01 -6.48931265e-01 3.35943885e-02
7.60637522e-01 6.87225342e-01 -1.73646897e-01 5.15886903e-01
-1.62086785e-01 1.48026705e-01 -3.59501332e-01 8.61114681e-01
5.18285573e-01 2.35433280e-01 1.79835632e-01 -2.52459168e-01
-8.16374719e-01 -2.41929904e-01 -9.16760862e-01 8.77630889e-01
-4.47091132e-01 1.30271089e+00 7.52147079e-01 -7.95713544e-01
1.07563424e+00 3.08691949e-01 2.60905117e-01 -1.12837464e-01
1.00194979e+00 1.63081586e-01 7.78752267e-02 -4.95125949e-01
7.25810826e-01 1.44206807e-01 4.84928451e-02 -4.14827727e-02
1.97110683e-01 -1.00014932e-01 3.95599842e-01 -3.12941849e-01
1.28069353e+00 -2.54339188e-01 4.33238924e-01 -4.47807521e-01
6.28264070e-01 3.57828259e-01 2.44953260e-01 2.06092089e-01
-3.00158530e-01 -2.84526706e-01 -4.37374264e-02 -1.06916118e+00
-9.20356810e-01 -2.86648214e-01 4.22746204e-02 7.09489584e-01
1.86698779e-01 -2.63893455e-01 -8.47019732e-01 -8.16465169e-02
-1.80154741e-01 -5.38450330e-02 -1.15423419e-01 -1.82930157e-01
-2.55297333e-01 -7.52898872e-01 5.72705805e-01 4.56395239e-01
8.64158452e-01 -7.75173366e-01 -2.14872098e+00 4.02696699e-01
5.60577810e-01 -1.53941321e+00 4.46223438e-01 3.32315177e-01
-7.14248598e-01 -1.37130094e+00 -2.24040955e-01 -6.96797013e-01
5.34964859e-01 9.46791172e-01 6.66135371e-01 8.84541571e-02
-7.67712474e-01 6.82949245e-01 -4.07676011e-01 -1.05026805e+00
-8.46007764e-02 1.54764593e-01 4.50605899e-01 3.75339054e-02
4.54003572e-01 -2.52330333e-01 -7.89622188e-01 1.66040227e-01
-1.04555464e+00 -3.13447952e-01 8.74046087e-01 3.69361430e-01
3.67898226e-01 4.97042358e-01 1.22343805e-02 -5.04263379e-02
2.48455212e-01 -5.19885480e-01 -1.59182453e+00 7.65175670e-02
-2.62181997e-01 -5.75476050e-01 9.22024310e-01 -1.36587843e-01
-3.25525522e-01 6.31100416e-01 1.42500743e-01 -5.19030511e-01
-4.53381091e-01 3.00989836e-01 2.41631456e-02 -6.14123106e-01
3.97717029e-01 -1.71559379e-01 1.25140488e-01 -1.59802176e-02
-1.85597628e-01 9.17075396e-01 5.06881714e-01 1.31490737e-01
8.32323968e-01 5.99602938e-01 3.81813854e-01 -1.56348026e+00
-3.38618249e-01 -6.04465604e-01 -6.69066012e-01 -8.17673206e-01
1.01199353e+00 -1.46776640e+00 -9.03272152e-01 5.56838989e-01
-1.29588270e+00 -5.30845346e-03 2.33660564e-01 6.67559505e-01
-8.61832872e-02 -9.90726203e-02 -1.48982227e-01 -9.16215718e-01
-7.79209137e-01 -1.34106421e+00 1.12180567e+00 6.13167465e-01
9.25177857e-02 -6.00334048e-01 -1.60872832e-01 -1.34112954e-01
4.94418591e-01 4.75847244e-01 2.81674147e-01 -2.76675731e-01
-1.01663268e+00 -7.13605165e-01 -2.25950629e-01 4.91428256e-01
-9.80141312e-02 4.24305290e-01 -8.80099356e-01 -8.07644963e-01
-2.84639657e-01 -2.50332326e-01 6.84325218e-01 4.54715401e-01
6.51409149e-01 -7.17797950e-02 -2.95227498e-01 7.65331447e-01
1.85925460e+00 4.22913671e-01 3.33305568e-01 3.23263258e-01
3.76304299e-01 3.90076458e-01 8.74170363e-01 6.33211076e-01
-8.28147028e-03 5.14417291e-01 1.28138530e+00 -1.27218783e-01
3.80491227e-01 3.19738775e-01 5.89051485e-01 2.92914897e-01
-1.59907907e-01 -5.65995932e-01 -8.22553754e-01 5.53823292e-01
-1.55510187e+00 -5.20973742e-01 -8.58247057e-02 1.94649637e+00
-4.49778348e-01 -9.53171402e-02 2.19917465e-02 1.58011317e-01
7.16855824e-01 1.19711727e-01 -4.52474117e-01 -4.45263118e-01
4.78713781e-01 1.76389143e-01 1.04446852e+00 -1.08502693e-01
-1.65642881e+00 7.75643945e-01 5.82187891e+00 1.48572817e-01
-1.50563776e+00 -1.53345317e-01 3.10365651e-02 -4.97675575e-02
8.94716501e-01 -2.83413440e-01 -1.01295710e+00 2.05732271e-01
1.36515653e+00 2.80718923e-01 7.02823251e-02 1.32307851e+00
2.99508542e-01 -3.14881295e-01 -7.50168622e-01 7.85276115e-01
-8.06043074e-02 -1.08152878e+00 8.94359425e-02 2.71272808e-01
3.30580086e-01 2.31032744e-01 8.44402462e-02 -2.45336786e-01
-5.06585501e-02 -7.95805097e-01 6.97511315e-01 5.14017269e-02
9.54779327e-01 -9.61258113e-01 1.15960300e+00 3.12775582e-01
-1.64796090e+00 -4.35176462e-01 -7.56647170e-01 -5.70229471e-01
-1.15539670e-01 4.10229772e-01 -1.11572087e+00 2.90719062e-01
9.58356321e-01 4.51748163e-01 -3.92283380e-01 1.00804055e+00
9.04576245e-06 3.22847754e-01 -5.11761904e-01 -4.24232423e-01
5.99672496e-01 -8.66482630e-02 6.52481318e-01 1.09257662e+00
9.38758731e-01 2.31097952e-01 2.80983239e-01 3.87061417e-01
1.46677256e-01 -3.11984152e-01 -1.03111398e+00 -1.92420855e-01
3.81469220e-01 1.80664110e+00 -9.70683575e-01 -1.23603657e-01
-4.64227051e-01 4.35627431e-01 -1.22772016e-01 -1.93240821e-01
-8.14436316e-01 -7.10913837e-01 1.10953331e+00 1.50629684e-01
5.64481318e-01 -6.51578546e-01 3.98056090e-01 -6.27771676e-01
-1.13470547e-01 -5.51361859e-01 -6.20292313e-03 -6.46504223e-01
-3.76380861e-01 9.99411464e-01 -3.55257816e-03 -1.67474031e+00
-2.34225709e-02 -1.33964872e+00 -6.24538362e-01 3.95374686e-01
-1.60871804e+00 -1.26736546e+00 -9.32682872e-01 3.11379969e-01
5.87622702e-01 -4.95017081e-01 7.84004748e-01 9.72858369e-02
-8.06212366e-01 -7.64056072e-02 -2.56666243e-01 1.81386262e-01
2.29720071e-01 -6.16140246e-01 3.49347979e-01 1.48323298e+00
-1.77845299e-01 2.95542955e-01 6.00428283e-01 -4.37616050e-01
-1.88932025e+00 -1.63919306e+00 2.21700132e-01 -2.06381809e-02
3.62859935e-01 -7.33110607e-02 -1.02120191e-01 6.25726283e-01
4.52477217e-01 2.41416767e-01 5.45504153e-01 -7.26666927e-01
2.22414285e-01 -6.05650127e-01 -1.10859859e+00 2.46894136e-01
3.00611943e-01 -8.26508105e-02 -5.95054701e-02 2.03339785e-01
4.89737719e-01 -4.17824149e-01 -5.63267171e-01 5.49027383e-01
5.81798375e-01 -9.27030921e-01 7.63032436e-01 -2.71996468e-01
2.26216689e-01 -8.19011748e-01 -1.06763646e-01 -1.18034935e+00
-3.40257287e-01 -6.26535952e-01 -7.20315203e-02 7.02242732e-01
-8.92099440e-02 -4.77500468e-01 3.96988451e-01 2.43223920e-01
-3.37611854e-01 -4.03481632e-01 -9.06700134e-01 -7.18842924e-01
-8.87537301e-01 -2.97286093e-01 5.37853956e-01 1.18264034e-01
-5.07803202e-01 1.07059218e-01 -3.01554263e-01 9.08315241e-01
5.04479110e-01 -2.62995869e-01 8.54432762e-01 -1.42911124e+00
3.52534026e-01 1.76143676e-01 -1.14444590e+00 -6.33352399e-01
-1.35811150e-01 -2.84534156e-01 1.74226627e-01 -1.20361006e+00
-4.94460046e-01 -2.78326217e-02 5.04158773e-02 2.68394828e-01
3.79483014e-01 4.55960810e-01 1.98067024e-01 -8.55131149e-02
-5.07932544e-01 1.15469933e-01 4.35904086e-01 -1.98424190e-01
5.80922440e-02 8.50706303e-04 -2.53876805e-01 8.44409227e-01
7.86844194e-01 -3.26093137e-01 -3.88648391e-01 -5.58760345e-01
1.26137212e-01 -2.34614819e-01 8.24656487e-01 -1.86513817e+00
5.73804140e-01 2.28965893e-01 5.61516583e-01 -6.09613895e-01
4.22785044e-01 -1.46699524e+00 2.50207931e-01 9.84372258e-01
2.77002692e-01 5.07538676e-01 5.23695529e-01 5.44371903e-01
-3.30427796e-01 -1.56550646e-01 9.55780685e-01 -2.24443734e-01
-1.28722215e+00 3.23802054e-01 -1.06595290e+00 -6.51948750e-01
1.70771456e+00 -1.00460835e-01 -2.70338684e-01 4.95047979e-02
-6.00177199e-02 6.84773326e-02 3.23184639e-01 2.34299898e-01
7.94987977e-01 -8.29652905e-01 -3.73545110e-01 5.12243807e-01
1.18824355e-01 1.48390248e-01 1.61126912e-01 3.96879226e-01
-1.34235620e+00 8.85894060e-01 -7.15193272e-01 -7.36517549e-01
-1.54439545e+00 2.95394838e-01 3.83399099e-01 1.67016402e-01
-3.26273233e-01 7.13868558e-01 -2.39890128e-01 1.61599547e-01
1.32646486e-01 -6.44673645e-01 -2.64688224e-01 1.30725294e-01
9.09551740e-01 5.52512228e-01 4.14226919e-01 -5.14222205e-01
-4.81417984e-01 5.76762676e-01 2.74075657e-01 2.71038026e-01
1.31295657e+00 2.24214289e-02 -1.17358409e-01 -7.83408061e-02
7.93729067e-01 -7.07641363e-01 -1.48179138e+00 3.35172951e-01
-3.54427338e-01 -1.97513357e-01 6.87998474e-01 -2.13011503e-01
-1.10076463e+00 9.26532984e-01 9.63267982e-01 3.99799794e-01
1.35306108e+00 -5.26540041e-01 8.18266809e-01 8.12048852e-01
6.93278015e-01 -9.90289450e-01 -2.51631409e-01 5.16583979e-01
4.81673926e-01 -1.14141297e+00 2.17213839e-01 -3.26692998e-01
-2.59485006e-01 1.64784503e+00 7.13675857e-01 -4.08064574e-01
5.74909866e-01 4.65146393e-01 -8.88839513e-02 -3.88947755e-01
-8.56912494e-01 -6.05424523e-01 8.07745978e-02 7.09129214e-01
-1.02985628e-01 3.09098452e-01 1.79276586e-01 1.46846890e-01
1.36784807e-01 -1.03012890e-01 6.67226136e-01 1.18581486e+00
-8.76108646e-01 -4.53572214e-01 -4.01119351e-01 3.22226763e-01
-3.66575897e-01 8.64756256e-02 -4.46227431e-01 8.70887518e-01
6.05646789e-01 1.08227491e+00 3.69902790e-01 -8.32572520e-01
3.99860114e-01 -4.67301041e-01 9.21368524e-02 -5.68033934e-01
-6.50649369e-01 -2.76930988e-01 -1.22884652e-02 -4.85690683e-01
-7.63673246e-01 -4.85177994e-01 -4.88502204e-01 -1.65314823e-01
-2.91006565e-01 -2.59205818e-01 1.05240667e+00 8.00994694e-01
4.88331050e-01 6.11009896e-01 6.53578877e-01 -1.13982487e+00
-1.64161235e-01 -6.95823669e-01 -6.19789600e-01 -7.70021319e-01
6.36623859e-01 -8.02543759e-01 -2.83146352e-01 1.54744327e-01]
|
[8.442584991455078, -1.0307365655899048]
|
54636fca-015d-4923-aad1-25ab4b591c4a
|
the-effectiveness-of-pre-trained-code
| null | null |
https://openreview.net/forum?id=H1glKiCqtm
|
https://openreview.net/pdf?id=H1glKiCqtm
|
The Effectiveness of Pre-Trained Code Embeddings
|
Word embeddings are widely used in machine learning based natural language processing systems. It is common to use pre-trained word embeddings which provide benefits such as reduced training time and improved overall performance. There has been a recent interest in applying natural language processing techniques to programming languages. However, none of this recent work uses pre-trained embeddings on code tokens. Using extreme summarization as the downstream task, we show that using pre-trained embeddings on code tokens provides the same benefits as it does to natural languages, achieving: over 1.9x speedup, 5\% improvement in test loss, 4\% improvement in F1 scores, and resistance to over-fitting. We also show that the choice of language used for the embeddings does not have to match that of the task to achieve these benefits and that even embeddings pre-trained on human languages provide these benefits to programming languages.
|
['Ben Trevett', 'N. K. Taylor', 'Donald Reay']
|
2019-05-01
| null | null | null |
iclr-2019-5
|
['extreme-summarization']
|
['natural-language-processing']
|
[-5.83788678e-02 -9.36170481e-03 -2.91268557e-01 -5.20159602e-01
-5.52851677e-01 -4.70712870e-01 4.02849287e-01 9.83275354e-01
-9.16256309e-01 5.06516732e-02 3.69740754e-01 -6.33259356e-01
2.46644095e-01 -7.67464995e-01 -4.35264021e-01 -1.18345171e-01
-4.16799515e-01 -7.75611550e-02 3.67170244e-01 -1.70005083e-01
5.32394409e-01 2.51905680e-01 -1.36900687e+00 7.07326531e-02
8.27575684e-01 4.25730735e-01 8.55756775e-02 8.85383070e-01
-6.98648870e-01 6.53499901e-01 -4.08988595e-01 -4.66492563e-01
2.65652299e-01 1.02452487e-01 -8.12158763e-01 -1.83395132e-01
2.47705832e-01 -2.09739149e-01 -1.15039103e-01 1.05845320e+00
3.56605619e-01 9.05196965e-02 4.79094684e-01 -8.10868442e-01
-9.77296531e-01 6.30093396e-01 -6.99949265e-01 2.56007403e-01
4.44104373e-01 2.04944108e-02 1.42425227e+00 -8.95633936e-01
4.22696769e-01 1.03827477e+00 7.75852799e-01 5.22636950e-01
-1.38771200e+00 -2.42608085e-01 -8.98960009e-02 -3.42667967e-01
-1.29982173e+00 -1.83150038e-01 3.71616900e-01 -5.72299659e-01
1.56032550e+00 -1.64747640e-01 2.05084682e-01 4.21359271e-01
5.60740173e-01 5.58263123e-01 6.07894599e-01 -9.85032737e-01
1.23909257e-01 5.05249023e-01 6.44691706e-01 9.94393826e-01
4.90429133e-01 -2.42535368e-01 -8.86425376e-02 -4.16190028e-01
3.42423201e-01 1.11594677e-01 1.39194960e-03 -3.52953613e-01
-1.07762754e+00 1.16590714e+00 3.96795899e-01 5.53302646e-01
-8.20849016e-02 5.67399204e-01 9.60462868e-01 4.48735297e-01
3.97699893e-01 9.41061616e-01 -5.15918016e-01 -4.90241528e-01
-8.04988861e-01 1.53150871e-01 7.49115169e-01 1.02783179e+00
8.27808261e-01 1.71266526e-01 2.09419787e-01 9.82471287e-01
3.35356474e-01 9.94172916e-02 8.25865924e-01 -6.08336270e-01
5.15473127e-01 7.56570876e-01 -2.01984778e-01 -8.35008264e-01
-1.33375302e-01 -1.96736872e-01 -1.77276656e-01 2.54817456e-01
3.16424280e-01 -2.51373976e-01 -8.07354867e-01 1.72737837e+00
-1.48356080e-01 -3.75741899e-01 2.32045904e-01 3.12724620e-01
2.62299865e-01 8.79192829e-01 1.14946865e-01 2.03432381e-01
1.48919523e+00 -7.97738433e-01 -2.96535820e-01 -5.99331915e-01
1.31770849e+00 -1.04442036e+00 1.36388993e+00 1.94390640e-01
-9.38772023e-01 -4.63608980e-01 -1.29352796e+00 -2.67973363e-01
-4.86113906e-01 -1.15513310e-01 6.71030521e-01 9.06425178e-01
-1.02668571e+00 4.38677520e-01 -8.35601032e-01 -5.66379547e-01
1.76214427e-01 3.52022558e-01 -4.91196275e-01 -3.52397263e-01
-7.83408403e-01 1.10756397e+00 3.87552381e-01 -5.87168992e-01
-3.87471229e-01 -8.91546726e-01 -1.21326983e+00 3.32804441e-01
1.54984728e-01 -1.78489253e-01 1.22395372e+00 -9.91607487e-01
-1.08335710e+00 8.99491966e-01 -3.09198141e-01 -5.17131627e-01
-3.25821340e-02 -4.69517171e-01 -2.23183781e-01 -2.27577165e-01
5.69709130e-02 5.53797901e-01 3.67705524e-01 -7.89597094e-01
-3.87457281e-01 -4.17576283e-02 2.39274189e-01 -1.58519104e-01
-9.68554914e-01 3.43511224e-01 -4.63272005e-01 -5.16629398e-01
-3.03758234e-01 -8.27991486e-01 -4.72822815e-01 1.79586157e-01
1.95392162e-01 -5.97160697e-01 4.84354615e-01 -4.88667130e-01
1.34852767e+00 -2.38955522e+00 -2.90481061e-01 7.97809958e-02
2.49979287e-01 6.18466318e-01 -5.28529048e-01 8.13210070e-01
-5.44472709e-02 5.62243164e-01 -3.44740689e-01 -2.16281265e-01
1.69809878e-01 2.71820039e-01 -4.39201832e-01 3.61208230e-01
7.68379271e-01 6.34885609e-01 -9.72245216e-01 -3.40125620e-01
8.39013010e-02 3.73244643e-01 -1.06647682e+00 2.73490429e-01
-6.08310811e-02 -5.57493806e-01 -2.37801358e-01 5.41949421e-02
3.91001910e-01 8.47337097e-02 2.25605875e-01 4.68456894e-01
-2.77430147e-01 5.99520385e-01 -9.68485415e-01 1.71437025e+00
-1.05899560e+00 8.31190467e-01 -1.86655268e-01 -9.29480195e-01
9.53999162e-01 2.44741276e-01 1.59579098e-01 -5.60892522e-01
-3.22737731e-02 2.31798664e-01 3.29274714e-01 -6.71572328e-01
6.83206141e-01 -2.41597667e-01 -3.10740530e-01 7.52274334e-01
2.24229068e-01 -2.33233407e-01 4.07234401e-01 2.91370124e-01
1.33533895e+00 -2.21959993e-01 4.68828529e-01 -6.08802378e-01
3.75079662e-01 2.62941748e-01 5.53591192e-01 4.19354320e-01
4.75984849e-02 4.50039953e-01 7.28496253e-01 -4.00613725e-01
-1.26572597e+00 -8.25556755e-01 -1.22615494e-01 1.48834085e+00
-2.32460663e-01 -9.58706081e-01 -5.22848010e-01 -3.84404421e-01
6.53097183e-02 8.69960189e-01 -5.21694183e-01 -1.42495662e-01
-6.58923030e-01 -5.25277436e-01 8.38884354e-01 7.44645536e-01
-1.40065864e-01 -7.21947253e-01 -6.75207198e-01 4.28706378e-01
5.73391199e-01 -9.32473898e-01 -5.79064548e-01 3.84270221e-01
-9.75425661e-01 -7.90498734e-01 -5.38497627e-01 -1.16533160e+00
8.85013223e-01 2.54126817e-01 1.14199007e+00 4.37256366e-01
-5.44682145e-01 3.20273280e-01 -5.17617822e-01 -4.93504405e-01
-5.15428185e-01 2.06588343e-01 3.57910208e-02 -5.34833312e-01
7.49425769e-01 -3.13565820e-01 -1.08360052e-01 -2.74222761e-01
-1.15631640e+00 -4.48460907e-01 4.53894138e-01 9.31976676e-01
-1.84510257e-02 -6.45937696e-02 3.94742489e-01 -1.14755666e+00
9.42416966e-01 -4.87217367e-01 -5.01587272e-01 4.36030850e-02
-8.82472396e-01 5.99690080e-01 1.05516291e+00 -5.47257423e-01
-6.64349139e-01 -3.72688286e-02 -4.09555256e-01 3.37312482e-02
4.05200534e-02 7.47737646e-01 4.03481305e-01 7.11333007e-02
8.64134729e-01 -5.10308612e-03 1.03361858e-02 -2.72460163e-01
4.81893331e-01 7.43013084e-01 2.04132181e-02 -7.59499013e-01
7.68763721e-01 2.35651340e-02 -5.24587691e-01 -9.94622886e-01
-3.98310125e-01 -7.05431640e-01 -4.93565619e-01 6.28057837e-01
9.07211304e-01 -7.85415232e-01 -1.34094208e-01 -2.96222925e-01
-1.29828703e+00 -1.58847660e-01 -2.78040558e-01 4.79421228e-01
-2.54348040e-01 5.97665489e-01 -5.90027392e-01 -7.16700137e-01
-4.35343802e-01 -1.23938000e+00 5.39664567e-01 1.73346877e-01
-5.86546659e-01 -1.16531515e+00 3.00894231e-01 -1.26629710e-01
7.48199105e-01 -8.91399458e-02 1.43038857e+00 -8.46126080e-01
-2.12495700e-01 -4.65830058e-01 -2.34159410e-01 6.80178642e-01
1.78516492e-01 3.07939589e-01 -7.75064230e-01 -3.65131021e-01
-2.29154289e-01 -3.66456032e-01 7.33507037e-01 -7.52925873e-02
8.04541290e-01 -9.57481563e-02 2.54214741e-03 2.95979530e-01
1.87654376e+00 9.04837549e-02 5.37481666e-01 1.72689229e-01
5.71043909e-01 6.11884654e-01 2.09397435e-01 2.50254065e-01
1.92995951e-01 2.52816617e-01 5.84136061e-02 6.66020885e-02
1.97410285e-01 -2.03044698e-01 7.37998426e-01 1.08151674e+00
3.29293281e-01 4.52552736e-02 -1.35586405e+00 1.19765532e+00
-1.51298535e+00 -7.00586200e-01 -1.30883992e-01 2.33065319e+00
9.55978632e-01 4.74787831e-01 -7.93954581e-02 1.13827050e-01
3.37743729e-01 1.69575959e-01 -1.26946643e-01 -1.46708167e+00
4.91505802e-01 7.85561502e-01 6.20281041e-01 5.53968847e-01
-7.62396336e-01 9.38811481e-01 6.17075348e+00 4.92652684e-01
-1.20771456e+00 2.41504133e-01 9.11391750e-02 2.13579386e-01
-4.33103472e-01 3.22131991e-01 -5.32393813e-01 1.41933039e-01
1.23076320e+00 -6.88144147e-01 3.06174278e-01 1.08065629e+00
-1.85895208e-02 -4.82669212e-02 -1.22528481e+00 6.63618207e-01
2.15029955e-01 -1.07585955e+00 -1.01932846e-01 2.81172022e-02
7.00338125e-01 3.21344674e-01 -2.28895158e-01 6.49638176e-01
5.76096356e-01 -1.08098245e+00 3.98110002e-01 -1.64077669e-01
7.24506557e-01 -1.01466465e+00 8.95547152e-01 3.72059882e-01
-1.21131504e+00 -1.03523515e-01 -7.17922926e-01 -2.80963838e-01
-1.16100885e-01 6.06655061e-01 -1.10114348e+00 1.49984807e-01
3.02492917e-01 3.71170402e-01 -6.56696796e-01 9.17856872e-01
-2.04248011e-01 6.70143187e-01 -2.04868913e-01 -4.49276477e-01
5.37461936e-01 1.20790027e-01 8.01533833e-02 1.69099069e+00
2.00803638e-01 -2.99693078e-01 3.02949190e-01 7.54873395e-01
-1.69392213e-01 4.34567600e-01 -9.98232722e-01 -6.15263104e-01
2.77876616e-01 1.01946199e+00 -4.79172170e-01 -4.57230121e-01
-9.98564661e-01 7.78476715e-01 3.60402107e-01 1.02343168e-02
-5.78075230e-01 -1.16422069e+00 1.06184018e+00 1.79389298e-01
3.69817108e-01 -7.29420304e-01 -3.01833749e-01 -1.18385768e+00
5.89457490e-02 -7.81828284e-01 7.79387802e-02 -3.09309751e-01
-1.29159355e+00 6.17179334e-01 -1.23271026e-01 -9.22053993e-01
-2.88109452e-01 -9.36261475e-01 -9.37367499e-01 1.06994498e+00
-1.40783572e+00 -5.79984426e-01 2.14157403e-01 -6.72279149e-02
7.53500581e-01 -2.06581950e-02 1.05528510e+00 3.36045265e-01
-2.83137739e-01 7.83870220e-01 1.29238769e-01 4.66187090e-01
6.62190676e-01 -1.38994229e+00 6.52163446e-01 1.11500692e+00
2.68804073e-01 1.42660010e+00 7.61543512e-01 -1.18048124e-01
-1.71366978e+00 -9.73875761e-01 1.36408269e+00 -4.17462617e-01
9.48788404e-01 -4.76135522e-01 -1.07253230e+00 6.08550191e-01
5.03270924e-01 2.29126774e-02 8.99831653e-01 4.61314410e-01
-7.90022671e-01 2.12762374e-02 -9.76416469e-01 5.88998675e-01
5.88957906e-01 -8.29337239e-01 -9.44488585e-01 9.06469375e-02
8.87315333e-01 -1.03518620e-01 -9.76728737e-01 -8.04667175e-02
3.82737190e-01 -4.21482712e-01 7.78361022e-01 -8.72849464e-01
7.90899754e-01 -1.81965932e-01 -2.59765893e-01 -1.26630652e+00
-2.54303634e-01 -2.57740736e-01 4.57354963e-01 1.40152848e+00
6.40224576e-01 -7.43200779e-01 4.61248696e-01 7.69553423e-01
-8.90896395e-02 -7.80700505e-01 -5.25244176e-01 -8.34382117e-01
6.66022658e-01 -7.25299835e-01 2.02061892e-01 9.77101445e-01
4.11482304e-01 3.91913623e-01 1.72177047e-01 6.04452044e-02
2.87678510e-01 8.81241634e-03 8.58148456e-01 -1.09681666e+00
-2.88577825e-01 -4.37439740e-01 -6.85846984e-01 -1.00959444e+00
2.35302061e-01 -1.22423518e+00 1.14246026e-01 -1.58953238e+00
4.87041324e-02 -5.37755251e-01 -2.90568024e-01 6.73262239e-01
-7.39172101e-02 -8.07749778e-02 2.01467916e-01 -2.12999552e-01
-2.54929394e-01 2.95868188e-01 4.31649595e-01 -9.74710658e-02
-1.23149179e-01 -4.70994174e-01 -8.89120340e-01 5.47178984e-01
7.96829820e-01 -7.94809699e-01 -4.17842269e-01 -1.04004323e+00
4.09146249e-01 -4.51038867e-01 -2.69447088e-01 -9.54135597e-01
1.66411385e-01 -1.35461288e-02 -2.29524210e-01 2.12594956e-01
-2.94605512e-02 -6.03738487e-01 -5.66681981e-01 6.73037946e-01
-5.04455149e-01 5.31848729e-01 5.73794842e-01 4.41314667e-01
-2.25633785e-01 -1.01587927e+00 8.27884138e-01 -1.58180386e-01
-8.12496901e-01 -1.26357228e-01 -7.41941750e-01 2.89664179e-01
8.17382455e-01 -1.15480781e-01 1.10634165e-02 1.45138621e-01
-9.96713787e-02 2.88598537e-01 7.56731033e-01 5.64122558e-01
4.48060632e-01 -1.03330076e+00 -5.47820270e-01 2.69510984e-01
5.62270403e-01 -2.04894111e-01 -3.55243176e-01 3.65619779e-01
-9.82671380e-01 3.84466171e-01 -1.09035209e-01 -3.30122977e-01
-1.44532919e+00 4.73233640e-01 -1.03578851e-01 -3.73861849e-01
-4.74566787e-01 8.26913416e-01 -1.02116793e-01 -6.54052317e-01
7.61260316e-02 -7.45447218e-01 1.49717778e-01 -9.90096629e-02
7.11983740e-01 -1.13766246e-01 5.73767014e-02 -1.35653183e-01
-5.15375495e-01 5.26046813e-01 -3.77290964e-01 -1.22933403e-01
1.74167871e+00 3.91569763e-01 -1.22270316e-01 6.15181506e-01
1.50956213e+00 4.44452673e-01 -7.25628018e-01 -4.46208030e-01
4.55275685e-01 -4.97113019e-01 7.87407160e-02 -1.73885211e-01
-7.52904177e-01 1.31440985e+00 3.82207632e-01 4.04513091e-01
7.22395599e-01 -2.09243327e-01 7.93903947e-01 5.78931630e-01
4.91890907e-01 -1.03373206e+00 -1.61956009e-02 8.33003223e-01
4.65809017e-01 -1.12961221e+00 -9.28173773e-03 -1.00132212e-01
-5.27841568e-01 1.41704214e+00 4.49090034e-01 -7.81166136e-01
5.80437243e-01 4.09055769e-01 -1.44815780e-02 -4.33834456e-02
-9.55315351e-01 -1.01641767e-01 -3.87047976e-02 3.93720120e-01
1.07965481e+00 -9.60508659e-02 -5.93340755e-01 2.43567452e-01
-4.33441512e-02 -1.07731014e-01 7.72603035e-01 1.36008048e+00
-6.20025814e-01 -1.58778012e+00 -1.01259425e-01 6.63220227e-01
-6.54421389e-01 -4.65161413e-01 -1.04188308e-01 7.67085195e-01
3.65628675e-02 8.07785273e-01 1.95207924e-01 -3.28308046e-01
3.97667706e-01 3.43514621e-01 3.39659512e-01 -1.41402030e+00
-8.21774065e-01 -5.38455844e-01 3.46413702e-02 -2.77270854e-01
-4.03833110e-03 -2.91551679e-01 -1.55422544e+00 -5.83046377e-01
-5.43032348e-01 2.86098272e-01 7.33394742e-01 6.62613809e-01
5.28355062e-01 4.31075186e-01 4.08941269e-01 -4.29335505e-01
-7.27078021e-01 -8.06501985e-01 -2.33743966e-01 3.21519494e-01
2.91184723e-01 -1.40372336e-01 -3.50880355e-01 1.31038100e-01]
|
[7.65729284286499, 7.917394161224365]
|
4e6e48fa-66a7-4d9c-8d61-b00bcbdea0b0
|
improving-autoregressive-nlp-tasks-via
|
2304.08453
| null |
https://arxiv.org/abs/2304.08453v3
|
https://arxiv.org/pdf/2304.08453v3.pdf
|
Improving Autoregressive NLP Tasks via Modular Linearized Attention
|
Various natural language processing (NLP) tasks necessitate models that are efficient and small based on their ultimate application at the edge or in other resource-constrained environments. While prior research has reduced the size of these models, increasing computational efficiency without considerable performance impacts remains difficult, especially for autoregressive tasks. This paper proposes modular linearized attention (MLA), which combines multiple efficient attention mechanisms, including cosFormer, to maximize inference quality while achieving notable speedups. We validate this approach on several autoregressive NLP tasks, including speech-to-text neural machine translation (S2T NMT), speech-to-text simultaneous translation (SimulST), and autoregressive text-to-spectrogram, noting efficiency gains on TTS and competitive performance for NMT and SimulST during training and inference.
|
['Lizhong Chen', 'Victor Agostinelli']
|
2023-04-17
| null | null | null | null |
['nmt']
|
['computer-code']
|
[ 4.26431417e-01 2.47409999e-01 -1.81267411e-01 -4.06940997e-01
-1.39699638e+00 -5.25571465e-01 7.29680002e-01 -4.40693974e-01
-3.21581841e-01 6.16953433e-01 4.90305245e-01 -1.12421024e+00
1.27382681e-01 -1.95013434e-01 -7.37475693e-01 -3.91469091e-01
2.42248207e-01 8.24680507e-01 -4.88596648e-01 -1.05607929e-02
-1.31130442e-01 4.39586192e-01 -8.02306175e-01 1.95733592e-01
1.03574872e+00 6.37963414e-01 5.40881932e-01 1.10852981e+00
-1.66531920e-01 8.04545283e-01 -4.32712078e-01 -5.75036526e-01
1.14236519e-01 -3.30113262e-01 -5.05026698e-01 -3.86795163e-01
4.79502439e-01 -4.10268515e-01 -2.91486830e-01 7.15629876e-01
7.30790377e-01 3.62800598e-01 6.97966576e-01 -9.69998360e-01
-1.04119432e+00 8.94940734e-01 -5.19129336e-01 4.32282060e-01
1.19555458e-01 3.61830920e-01 1.17995405e+00 -1.21807194e+00
2.93911517e-01 1.81499505e+00 6.87860787e-01 3.45745355e-01
-1.28255880e+00 -5.65599144e-01 8.11246559e-02 -5.14163524e-02
-1.03236210e+00 -1.11375105e+00 2.91225582e-01 -5.14259785e-02
1.92166233e+00 2.55790621e-01 1.34976342e-01 1.45001781e+00
4.97387409e-01 9.82632458e-01 7.80070662e-01 -5.73337555e-01
-9.86416787e-02 -1.54084623e-01 -3.22242230e-02 4.13480669e-01
-1.51869610e-01 -2.59366576e-02 -6.74456298e-01 1.05926298e-01
8.24698150e-01 -5.14327288e-01 -8.56877118e-03 4.69657123e-01
-1.25849485e+00 7.48318136e-01 -6.85852244e-02 1.58906132e-01
-7.19858170e-01 3.77632231e-01 3.36140871e-01 4.73859429e-01
7.94766605e-01 6.58349633e-01 -8.53073418e-01 -6.47843719e-01
-1.08239174e+00 5.38227446e-02 8.19846034e-01 1.15051210e+00
2.53630728e-01 9.04333591e-01 -3.28812838e-01 1.03660190e+00
3.14557254e-01 1.04895020e+00 6.71828747e-01 -9.51367080e-01
8.39866519e-01 -1.77656755e-01 -9.09956917e-02 -6.14477336e-01
-2.94601083e-01 -6.19111955e-01 -9.51656520e-01 -3.52838457e-01
1.51503205e-01 -5.66693842e-01 -1.02276516e+00 1.84138072e+00
-2.31137063e-04 1.71445534e-01 1.27684385e-01 7.44421184e-01
5.24633765e-01 1.34437811e+00 1.14851899e-01 -4.63054508e-01
1.17498624e+00 -1.16960371e+00 -9.91950035e-01 -8.04080963e-01
5.68332851e-01 -1.15404522e+00 1.32975614e+00 2.05432937e-01
-1.58574903e+00 -6.78024411e-01 -6.16288781e-01 -7.30457008e-01
-6.02664240e-02 2.50120580e-01 5.47421992e-01 4.72360045e-01
-1.26225495e+00 3.42124671e-01 -1.17908394e+00 -3.73520762e-01
-1.50125269e-02 5.67393363e-01 3.23509425e-02 8.62683654e-02
-1.31293464e+00 1.27972066e+00 -6.67336807e-02 2.80118525e-01
-7.20606029e-01 -8.82023871e-01 -9.21349168e-01 4.64420944e-01
9.08567458e-02 -1.09134638e+00 1.71411645e+00 -9.92126107e-01
-2.11479592e+00 2.40571991e-01 -7.69074678e-01 -8.60823214e-01
3.36689621e-01 -6.33379579e-01 -2.26065502e-01 -1.41272172e-01
-1.02411978e-01 7.82758653e-01 1.08620739e+00 -5.69656551e-01
-2.40143105e-01 -1.09315120e-01 -3.62237215e-01 5.72997332e-01
-3.50614250e-01 5.02734661e-01 -3.76661956e-01 -6.79529071e-01
-8.19922462e-02 -1.02158356e+00 -1.59985334e-01 -6.21797502e-01
-2.65995830e-01 -2.96416402e-01 7.32797742e-01 -1.17291689e+00
1.07826293e+00 -1.82020986e+00 3.39183748e-01 -3.02461147e-01
-2.14349970e-01 3.54838520e-01 -4.56909657e-01 4.22376692e-01
-4.40405980e-02 1.82681695e-01 -5.87530732e-02 -9.29954231e-01
1.20692812e-01 3.48801345e-01 -7.56864309e-01 1.43315837e-01
6.26764596e-01 1.38306606e+00 -6.29632115e-01 -5.44043422e-01
3.64111245e-01 6.18205309e-01 -5.83372712e-01 2.26606697e-01
-3.19442004e-01 4.62100536e-01 -2.26161882e-01 4.36252952e-01
2.32918531e-01 -2.07545996e-01 1.40613005e-01 1.58291921e-01
-1.60605729e-01 9.06829119e-01 -4.28988010e-01 1.38352966e+00
-1.14443719e+00 1.12888443e+00 1.93725690e-01 -7.64444172e-01
6.71625614e-01 5.44514418e-01 2.86946952e-01 -6.77206218e-01
1.86625659e-01 1.06518582e-01 2.96295434e-01 -4.33605999e-01
9.72787380e-01 -1.41288608e-01 1.93336129e-01 4.88199502e-01
2.02776447e-01 -4.06273127e-01 -3.44778113e-02 1.30105242e-01
7.72106290e-01 2.00067371e-01 3.02399099e-01 -2.77176082e-01
-8.69332403e-02 -1.18178941e-01 2.65569568e-01 9.19288218e-01
1.50503263e-01 6.16361678e-01 3.75839293e-01 -4.43688817e-02
-1.39949155e+00 -1.03245640e+00 1.05964959e-01 1.56389201e+00
-6.74126208e-01 -2.01307401e-01 -7.94823825e-01 -4.03061770e-02
-2.99699694e-01 1.29557276e+00 -5.45718335e-02 5.64888231e-02
-8.85323822e-01 -7.95945764e-01 6.90381885e-01 6.03724062e-01
7.32778758e-02 -1.05563748e+00 -1.39212146e-01 3.26660335e-01
-4.97616112e-01 -1.41576385e+00 -7.40310371e-01 1.40562966e-01
-9.83630657e-01 -1.13196567e-01 -6.81016505e-01 -6.97963715e-01
2.38066211e-01 2.25179389e-01 1.20357800e+00 -6.12072289e-01
2.48691306e-01 1.96296901e-01 -4.20240760e-02 -5.86144626e-01
-6.29115641e-01 3.84070426e-01 3.11858624e-01 -2.45095089e-01
9.78862643e-02 -5.96478105e-01 5.75943217e-02 -1.55324653e-01
-4.61166382e-01 3.74780655e-01 1.00740695e+00 9.82839525e-01
2.91796833e-01 -3.55252683e-01 6.33182228e-01 -6.67495191e-01
1.00528288e+00 -4.70433265e-01 -4.57935393e-01 1.76035151e-01
-6.05467081e-01 7.55994841e-02 9.84394908e-01 -7.01070726e-01
-1.32801259e+00 -3.02741170e-01 -3.56923223e-01 -7.06634521e-01
2.26190373e-01 8.59930634e-01 8.80208910e-02 4.14736301e-01
5.59523106e-01 5.25250137e-01 -2.48468146e-02 -2.87328959e-01
5.48387825e-01 7.77781010e-01 5.39887846e-01 -6.06429875e-01
6.60356581e-01 -2.29798779e-01 -3.46102327e-01 -1.19089305e+00
-6.21098161e-01 -1.03134759e-01 -3.52485180e-01 2.53217071e-01
7.35155106e-01 -1.07919574e+00 -5.04592180e-01 2.12525994e-01
-1.53251386e+00 -6.56965852e-01 3.81557308e-02 8.89207661e-01
-6.93797827e-01 2.32882604e-01 -8.80086243e-01 -1.14797592e+00
-8.85250211e-01 -1.22017550e+00 1.32354915e+00 -2.29907811e-01
-6.20577693e-01 -1.09764755e+00 -2.06732661e-01 5.21613955e-01
7.92469859e-01 -6.41382456e-01 1.18619049e+00 -6.38877630e-01
-6.95310712e-01 -1.02161013e-01 -2.75849968e-01 3.44465047e-01
-2.45405272e-01 2.25267276e-01 -9.52556372e-01 -3.15292507e-01
3.17040086e-02 -9.36266258e-02 5.19810140e-01 1.02484798e+00
7.75422692e-01 -7.49845922e-01 1.39019638e-02 6.28521800e-01
7.95070469e-01 1.72617137e-01 4.58154947e-01 -2.48241097e-01
6.23711407e-01 1.94468737e-01 2.14749217e-01 1.18583821e-01
3.31143796e-01 6.76310062e-01 -1.08372673e-01 -1.76996410e-01
-3.10072750e-01 -3.27627689e-01 9.17267203e-01 1.65838897e+00
-8.56691692e-03 -7.53791332e-01 -9.58988786e-01 5.05541444e-01
-1.73464322e+00 -7.40611434e-01 -3.28555554e-02 1.93786955e+00
8.67655456e-01 1.77388579e-01 -4.36857015e-01 -3.34606856e-01
4.30231541e-01 2.10208610e-01 -6.19237006e-01 -9.74722207e-01
-1.23888165e-01 5.28115034e-01 3.29391152e-01 8.81573141e-01
-7.01874912e-01 1.39328873e+00 7.06478071e+00 9.01174307e-01
-1.08642519e+00 3.78605008e-01 7.36424387e-01 -2.39559546e-01
-2.84624845e-01 -1.34242103e-01 -8.09061885e-01 2.56822258e-01
1.68696034e+00 -3.95394355e-01 9.80808318e-01 8.04861844e-01
8.03490102e-01 3.03549588e-01 -1.16793525e+00 8.39218020e-01
-7.64726549e-02 -1.06916678e+00 2.20129296e-01 -1.82049051e-01
7.63501167e-01 5.30217767e-01 4.92322057e-01 7.77518332e-01
7.21747160e-01 -1.10873079e+00 5.88296711e-01 1.00424007e-01
8.79451871e-01 -5.02704978e-01 4.28931922e-01 4.18603152e-01
-8.92968953e-01 3.65542201e-03 -3.66284460e-01 -1.62532344e-01
2.89866686e-01 3.82511497e-01 -1.28985322e+00 3.12564373e-01
1.99538082e-01 5.05951226e-01 7.73626380e-04 3.74256760e-01
-3.00730109e-01 1.22152650e+00 -4.80546623e-01 -2.96029478e-01
4.99707580e-01 -2.65289783e-01 7.99609184e-01 1.45716202e+00
5.99606156e-01 -7.03938827e-02 -8.27667862e-02 7.05118537e-01
-3.32961798e-01 2.21356109e-01 -5.70371866e-01 -4.52124387e-01
6.63937449e-01 1.03729796e+00 -2.43943840e-01 -5.44301033e-01
-4.31755304e-01 9.35598493e-01 3.74597043e-01 8.07791948e-01
-1.00280738e+00 4.85407095e-03 8.29024911e-01 -1.74062297e-01
-5.89518715e-03 -7.83362806e-01 -6.08942688e-01 -1.19440067e+00
-1.23080142e-01 -1.14699435e+00 -1.34312799e-02 -1.22893369e+00
-9.77291584e-01 6.72186792e-01 -2.17088372e-01 -6.86768591e-01
-8.77986372e-01 -3.91881764e-01 -4.01819497e-01 1.35687017e+00
-1.33794403e+00 -1.28133917e+00 4.11828607e-01 2.36573324e-01
1.30741143e+00 -1.34111762e-01 8.37376893e-01 2.72522569e-01
-6.94995880e-01 7.93843985e-01 4.69243884e-01 -1.76317751e-01
6.79159105e-01 -1.10701871e+00 1.10069156e+00 1.17560053e+00
3.71372461e-01 9.71419930e-01 8.11659753e-01 -4.69195753e-01
-1.88980246e+00 -1.15955019e+00 1.38409233e+00 -5.25987744e-01
9.18073356e-01 -4.35172945e-01 -7.33936548e-01 1.04939783e+00
5.77283442e-01 -6.51430070e-01 2.16372535e-01 5.02736211e-01
-1.76127598e-01 1.73583686e-01 -4.65890050e-01 1.13364625e+00
7.84401655e-01 -8.59765708e-01 -5.11989713e-01 5.85658669e-01
1.20424652e+00 -6.38164461e-01 -8.29902053e-01 1.68418467e-01
5.93682289e-01 -1.75849840e-01 9.30614889e-01 -8.82764995e-01
6.96187139e-01 1.80476353e-01 -1.41978532e-01 -1.60155439e+00
-3.84357303e-01 -1.22134244e+00 -2.34475598e-01 1.00359714e+00
8.22669208e-01 -7.43446052e-01 3.40811163e-01 7.17136800e-01
-7.00516582e-01 -4.65222061e-01 -1.05681217e+00 -7.44833350e-01
3.50653917e-01 -7.52567232e-01 3.39422137e-01 8.61132562e-01
-1.27879247e-01 8.55726600e-01 -7.83002734e-01 2.71263570e-01
1.22319818e-01 -2.06466377e-01 7.65008748e-01 -6.05178893e-01
-5.32619894e-01 -6.64536178e-01 2.28584826e-01 -1.55209911e+00
3.95657361e-01 -9.49392140e-01 2.81892985e-01 -1.56218874e+00
-1.11274235e-01 1.20747872e-01 1.17344029e-01 5.28788090e-01
-3.68331134e-01 -2.06371516e-01 2.15068638e-01 1.46937326e-01
-1.51322380e-01 6.86574817e-01 1.18401873e+00 -8.75869989e-02
-1.95126384e-01 1.96678527e-02 -5.00344932e-01 4.44425076e-01
7.44751334e-01 -2.92110473e-01 -4.89767373e-01 -1.14478576e+00
-4.46596742e-02 4.51741725e-01 -8.92848223e-02 -6.43537283e-01
2.59552151e-01 -2.06548050e-01 2.22599402e-01 -5.08543313e-01
8.05319488e-01 -4.13734108e-01 -4.49650362e-02 1.20975405e-01
-7.28548408e-01 4.70551550e-01 4.22429413e-01 2.19037369e-01
-1.12661533e-01 8.75945911e-02 7.31957078e-01 7.59925544e-02
-1.08766958e-01 3.30416441e-01 -7.83761144e-01 1.05050720e-01
3.22735220e-01 1.39357194e-01 -3.50474656e-01 -9.00032163e-01
-5.08488774e-01 1.10005595e-01 -1.23984851e-01 5.92577934e-01
4.69856918e-01 -9.79604900e-01 -1.13652706e+00 2.28982702e-01
-5.21913648e-01 -1.09031945e-01 1.05268076e-01 1.09669387e+00
-2.16489568e-01 1.05444515e+00 2.66362876e-01 -3.77197742e-01
-1.23044598e+00 3.27026010e-01 1.63103089e-01 -4.47926700e-01
-4.23236459e-01 8.33633065e-01 1.94210216e-01 -5.58087230e-01
5.24285287e-02 -6.10190570e-01 3.91083002e-01 -2.48059526e-01
2.41577655e-01 3.74969274e-01 1.11145206e-01 -4.15090054e-01
5.38258068e-02 1.32755399e-01 -1.87772483e-01 -4.11772341e-01
1.13297522e+00 -3.05844247e-01 -6.55364618e-02 4.55886334e-01
1.06261158e+00 -1.62551641e-01 -1.06843591e+00 -4.28536326e-01
-2.11888954e-01 -1.89470112e-01 3.15655828e-01 -6.82732165e-01
-6.20672405e-01 1.29597569e+00 7.83823058e-02 1.17337555e-01
8.95393968e-01 -4.36130792e-01 1.08343661e+00 7.44354248e-01
-1.04366533e-01 -9.87397909e-01 -2.11980507e-01 1.17595232e+00
9.39460337e-01 -1.09000969e+00 -2.11097479e-01 -1.47927970e-01
-7.05930769e-01 9.61795866e-01 2.57609218e-01 2.73151129e-01
2.99826086e-01 6.93104744e-01 3.62168215e-02 3.77772659e-01
-1.33621454e+00 7.63391778e-02 2.45872587e-01 4.66085345e-01
8.55748773e-01 2.03695118e-01 5.72974458e-02 2.28982225e-01
-5.74158370e-01 -2.36734420e-01 2.58905321e-01 3.73808682e-01
-2.19057739e-01 -7.47408152e-01 -2.98665702e-01 5.25398076e-01
-6.45533085e-01 -9.96555209e-01 -2.72776276e-01 6.52883410e-01
-6.79530025e-01 1.16524386e+00 2.08859116e-01 -8.92730802e-02
-1.98665857e-02 4.43470657e-01 3.08162898e-01 -5.19800007e-01
-7.37729371e-01 5.15159011e-01 4.65265453e-01 -2.71941543e-01
1.19350523e-01 -7.40878463e-01 -9.82002795e-01 -5.88928521e-01
-3.66265297e-01 -8.12266469e-02 9.54429090e-01 9.81162429e-01
6.14695668e-01 7.78512120e-01 4.16129887e-01 -8.60705197e-01
-8.85060608e-01 -1.49744189e+00 4.07006964e-02 -1.64003983e-01
3.49794358e-01 -1.26288250e-01 -3.19189131e-02 2.11860895e-01]
|
[14.461668968200684, 7.239919662475586]
|
22cbf84c-5299-46ff-b748-acfaf933018b
|
generative-pre-trained-transformer-for
|
2110.04071
| null |
https://arxiv.org/abs/2110.04071v1
|
https://arxiv.org/pdf/2110.04071v1.pdf
|
Generative Pre-Trained Transformer for Cardiac Abnormality Detection
|
ECG heartbeat classification plays a vital role in diagnosis of cardiac arrhythmia. The goal of the Physionet/CinC 2021 challenge was to accurately classify clinical diagnosis based on 12, 6, 4, 3 or 2-lead ECG recordings in order to aid doctors in the diagnoses of different heart conditions. Transformers have had great success in the field of natural language processing in the past years. Our team, CinCSEM, proposes to draw the parallel between text and periodic time series signals by viewing the repeated period as words and the whole signal as a sequence of such words. In this way, the attention mechanisms of the transformers can be applied to periodic time series signals. In our implementation, we follow the Transformer Encoder architecture, which combines several encoder layers followed by a dense layer with linear or sigmoid activation for generative pre-training or classification, respectively. The use case presented here is multi-label classification of heartbeat abnormalities of ECG recordings shared by the challenge. Our best entry, not exceeding the challenge's hardware limitations, achieved a score of 0.12, 0.07, 0.10, 0.10 and 0.07 on 12-lead, 6-lead, 4-lead, 3-lead and 2-lead test set respectively. Unfortunately, our team was unable to be ranked because of a missing pre-print.
|
['Ricard Delgado-Gonzalo', 'Mathieu Lemay', 'Jérôme Van Zaen', 'Clémentine Aguet', 'Halla Sigurthorsdottir', 'Pierre Louis Gaudilliere']
|
2021-10-07
| null | null | null | null |
['heartbeat-classification']
|
['medical']
|
[ 2.72279769e-01 8.39511976e-02 3.76357317e-01 -4.35725838e-01
-8.86221468e-01 -6.71258628e-01 2.40591615e-02 1.83762312e-01
-2.27349717e-02 7.82980144e-01 1.70903414e-01 -6.21217966e-01
-2.66860157e-01 -3.61815453e-01 -2.06037387e-01 -4.76662785e-01
-3.40336919e-01 4.92340893e-01 -2.63463438e-01 -1.71397198e-02
2.62821876e-02 1.74733505e-01 -9.62442279e-01 6.24072909e-01
2.35140741e-01 1.08324730e+00 -2.59351730e-01 1.19261110e+00
3.11085224e-01 8.13630760e-01 -8.70173275e-01 -2.04239801e-01
1.00734733e-01 -9.38350022e-01 -8.56562078e-01 7.67091960e-02
8.69074389e-02 -5.28315865e-02 7.34227970e-02 5.05230963e-01
1.25415349e+00 -3.83669883e-01 5.89626610e-01 -9.46826279e-01
-5.68867214e-02 8.65090311e-01 -3.33288908e-01 5.84560335e-01
3.96361589e-01 -4.71348539e-02 9.02878523e-01 -6.00225747e-01
3.01344663e-01 6.58223748e-01 1.14554214e+00 3.52492601e-01
-1.02006292e+00 -5.44999778e-01 -4.75220174e-01 2.11719468e-01
-1.36748028e+00 -3.32578957e-01 7.83919871e-01 -5.75932503e-01
1.20825469e+00 3.88157487e-01 8.70735347e-01 1.11850369e+00
5.90217292e-01 2.36176670e-01 1.16554677e+00 -3.43914270e-01
-4.86483537e-02 -1.02012150e-01 1.25452653e-01 4.55752343e-01
-1.61643460e-01 -5.79239689e-02 -3.80711347e-01 -2.86002368e-01
4.63079453e-01 -7.56120235e-02 -2.23226219e-01 4.61957544e-01
-1.35716486e+00 6.64397717e-01 1.54341400e-01 4.12731290e-01
-5.18429220e-01 1.59536660e-01 9.03021932e-01 6.94493651e-01
3.11188251e-01 5.67611754e-01 -5.73385179e-01 -3.92291933e-01
-1.07464933e+00 6.43529892e-02 6.92948103e-01 6.01910114e-01
-1.25276167e-02 1.20053329e-01 -4.54989225e-01 6.43816054e-01
2.02337027e-01 1.82838097e-01 7.73081481e-01 -5.89122832e-01
2.39404812e-01 7.54946023e-02 -9.53635350e-02 -7.41927326e-01
-7.52673209e-01 -9.75569844e-01 -1.18787909e+00 -1.20533034e-01
1.42472669e-01 -3.86303365e-01 -9.00463462e-01 1.41626012e+00
-1.18238561e-01 3.98356140e-01 2.06115857e-01 8.15924525e-01
8.92362595e-01 5.44894755e-01 -7.09394291e-02 -4.98564780e-01
1.66524601e+00 -4.32989001e-01 -7.36984849e-01 2.86694672e-02
5.63687563e-01 -8.64378154e-01 7.37353325e-01 7.32667387e-01
-1.14426148e+00 -6.18632495e-01 -9.96898234e-01 2.62717426e-01
8.80467426e-03 2.89710134e-01 2.28330538e-01 6.50956035e-01
-1.05265296e+00 6.50075972e-01 -6.98964179e-01 -3.14221978e-01
5.17078519e-01 2.65041530e-01 -5.35719618e-02 4.69131827e-01
-1.48617816e+00 8.20376337e-01 2.59360760e-01 7.35822925e-03
-8.38073075e-01 -7.67993033e-01 -5.83022118e-01 5.73083088e-02
-3.56655121e-01 -6.74127400e-01 1.15978372e+00 -6.06379151e-01
-1.25248528e+00 1.15844727e+00 1.41534805e-01 -8.12555790e-01
5.58033407e-01 -6.86707497e-02 -6.96849167e-01 4.49194945e-03
1.13307148e-01 2.05801710e-01 7.69168854e-01 -5.43115497e-01
-5.82539856e-01 -3.74671817e-01 -2.33659625e-01 7.84833208e-02
1.26686350e-01 1.15154535e-01 1.75645560e-01 -8.04299951e-01
1.07444346e-01 -8.46824944e-01 1.21775148e-02 -6.45659626e-01
-5.02217114e-01 -7.58680105e-02 4.83314723e-01 -8.23886096e-01
1.39388192e+00 -2.37222457e+00 7.10356086e-02 2.23715469e-01
3.38001102e-01 5.05325682e-02 2.95742244e-01 5.89569867e-01
-4.78256166e-01 1.45822689e-01 -2.97604203e-01 -1.55834779e-01
-2.08897755e-01 7.11004734e-02 -3.88725460e-01 5.98812282e-01
3.38660657e-01 8.49195719e-01 -7.50663996e-01 -3.49175215e-01
2.39310823e-02 6.20733678e-01 -1.41812474e-01 1.77353621e-01
2.00823963e-01 5.49982488e-01 -2.79854890e-02 3.10878158e-01
2.63399005e-01 -4.81494099e-01 2.04054117e-01 -2.70957053e-01
3.79923768e-02 5.50567806e-01 -9.23656583e-01 1.81452441e+00
-1.91451713e-01 7.30718911e-01 -4.24142957e-01 -1.08081627e+00
9.43183064e-01 1.16203570e+00 6.71628356e-01 -4.34147656e-01
1.70714080e-01 2.22832695e-01 4.60185081e-01 -6.85058713e-01
-1.74561650e-01 -6.20061815e-01 -6.84570149e-02 5.61002851e-01
-6.28439113e-02 3.34661566e-02 8.61248747e-03 -9.81403887e-02
1.29864323e+00 -1.71497703e-01 4.59260553e-01 -1.41599819e-01
3.84672433e-01 -2.92547882e-01 3.81503522e-01 5.45216203e-01
-1.59937859e-01 1.16890204e+00 7.16389716e-01 -6.75254643e-01
-9.94185984e-01 -9.83103216e-01 -3.46711040e-01 5.01963556e-01
-6.03866577e-01 -7.07113266e-01 -4.80244249e-01 -4.23684269e-01
-4.01845306e-01 2.83146471e-01 -5.35574853e-01 -1.26232237e-01
-5.75885177e-01 -1.02929413e+00 1.15874887e+00 5.88245869e-01
1.38374671e-01 -1.33262324e+00 -1.35093701e+00 5.80400646e-01
-3.48990232e-01 -7.75435567e-01 -2.08963647e-01 6.04872942e-01
-8.74416709e-01 -9.32730079e-01 -7.47523963e-01 -6.78766668e-01
1.13629699e-01 -5.15350521e-01 1.38272440e+00 -1.26658380e-01
-6.37596607e-01 1.64510787e-01 -5.18251300e-01 -6.16165221e-01
-2.88715929e-01 1.19694635e-01 -1.39428169e-01 1.10316902e-01
4.58494201e-02 -7.47851610e-01 -7.33988047e-01 -8.96715075e-02
-5.99158049e-01 -6.11775070e-02 4.07103777e-01 9.44135725e-01
3.95614624e-01 -4.72820252e-02 8.51061106e-01 -1.06943679e+00
9.54983056e-01 -5.32750249e-01 1.68852322e-02 -2.52062688e-03
-8.25277448e-01 -3.37529570e-01 6.07357621e-01 -2.13497832e-01
-1.95095524e-01 -1.88666508e-02 -4.53868568e-01 -4.71551687e-01
-1.70237824e-01 7.84681380e-01 2.97263563e-01 4.69184458e-01
8.39584470e-01 3.37647885e-01 -2.81624496e-01 -2.17178926e-01
-1.11955270e-01 8.65657091e-01 6.14442289e-01 -1.84520453e-01
2.30727211e-01 3.80374976e-02 1.20335193e-02 -4.99121070e-01
-6.08174980e-01 -5.24510622e-01 -3.91839266e-01 -1.05334580e-01
9.80259120e-01 -9.00290251e-01 -5.04028559e-01 4.12041754e-01
-1.26881874e+00 6.70690611e-02 -4.25409377e-01 2.98262656e-01
-4.80045319e-01 1.81200221e-01 -7.56557345e-01 -7.63982296e-01
-9.62542593e-01 -7.15088487e-01 1.14797103e+00 -1.94273397e-01
-8.36323559e-01 -8.96878958e-01 7.47112185e-02 3.15202549e-02
4.75029498e-01 6.70661390e-01 1.02762723e+00 -6.76735163e-01
1.00740075e-01 -4.11160439e-01 2.08024532e-01 5.08193552e-01
2.91108936e-01 -2.76016235e-01 -1.08759642e+00 -4.52084333e-01
4.12121475e-01 -1.94819525e-01 3.82656544e-01 4.51370746e-01
1.14213622e+00 -7.26959929e-02 -6.27492592e-02 5.66156805e-01
1.22856641e+00 4.59156007e-01 7.73869812e-01 -1.76754639e-01
4.96448845e-01 1.95016697e-01 8.38241875e-02 6.65040076e-01
3.00368905e-01 3.59735370e-01 8.72865841e-02 -1.86008096e-01
-5.77167012e-02 1.51524618e-01 1.80843204e-01 1.03260565e+00
-1.06456857e-02 -2.82365412e-01 -1.23898995e+00 5.86503148e-01
-1.40112996e+00 -9.77423251e-01 -4.90982354e-01 2.21162963e+00
9.05106306e-01 3.11501026e-01 1.63481668e-01 9.00131881e-01
4.00854170e-01 -1.33429796e-01 -3.14421058e-01 -6.41640246e-01
-5.67814223e-02 6.28054678e-01 -1.98091418e-02 2.56459564e-01
-1.05724549e+00 2.22580075e-01 6.41698790e+00 2.75686115e-01
-1.59619367e+00 2.33810678e-01 8.57614934e-01 3.66479158e-02
1.61078110e-01 -2.83271581e-01 -3.72424901e-01 7.21551180e-01
1.46381092e+00 -1.13648139e-01 1.93605676e-01 1.29466057e-01
2.29211077e-01 3.77165645e-01 -1.33385873e+00 1.38617527e+00
1.08300997e-02 -1.22644663e+00 -3.39717656e-01 -2.67431557e-01
2.45669797e-01 1.89511791e-01 -5.52337393e-02 1.71421885e-01
-4.54803973e-01 -1.46105278e+00 4.77227032e-01 6.06628418e-01
1.23131824e+00 -4.71616149e-01 9.88577783e-01 2.07088605e-01
-9.57883120e-01 1.11754853e-02 1.12753235e-01 -1.17952578e-01
8.94183069e-02 7.03152776e-01 -1.14486217e+00 5.86064875e-01
6.73873365e-01 1.00712895e+00 -2.53610164e-01 8.98270607e-01
8.86975899e-02 1.13483953e+00 -2.32934639e-01 4.13475811e-01
-6.24114973e-03 2.66533673e-01 5.05752325e-01 1.38597572e+00
3.25264692e-01 3.17857832e-01 1.55475467e-01 4.76633340e-01
1.92744613e-01 -2.69320025e-03 -5.14403522e-01 5.34641817e-02
9.13757458e-02 1.11613512e+00 -7.18250990e-01 -5.09345055e-01
-8.08817148e-02 9.10995901e-01 -2.44198084e-01 1.86812252e-01
-1.00823772e+00 -7.35197544e-01 6.94697574e-02 3.23444039e-01
6.89629018e-02 2.76215881e-01 -5.16489148e-01 -7.89606929e-01
9.35320482e-02 -1.21410406e+00 5.36255360e-01 -9.47533906e-01
-1.31459177e+00 1.30347359e+00 -3.14079762e-01 -1.39119041e+00
-6.21570945e-01 -3.36612731e-01 -5.81365407e-01 1.37794960e+00
-8.77090693e-01 -7.50415027e-01 -1.73168883e-01 4.42538798e-01
4.47527677e-01 -2.04708844e-01 1.34923291e+00 7.66868412e-01
-1.90459207e-01 5.52984536e-01 -5.21746457e-01 2.12261334e-01
6.93815053e-01 -1.49451077e+00 2.97477126e-01 4.60398704e-01
2.56371677e-01 4.29925174e-01 6.84366643e-01 -2.77803838e-01
-8.96744609e-01 -1.15909004e+00 1.51227832e+00 -5.61118305e-01
2.51723588e-01 -1.20477006e-01 -7.48104990e-01 3.85320276e-01
4.21371847e-01 3.61918882e-02 9.57747996e-01 1.50380172e-02
2.07220204e-02 -2.59714931e-01 -9.87571537e-01 -1.67289190e-02
6.03371203e-01 -7.92514861e-01 -6.27978384e-01 4.68018264e-01
2.68344551e-01 -8.05531621e-01 -1.21572614e+00 4.98194575e-01
6.37638807e-01 -7.59448886e-01 7.72403061e-01 -7.73130596e-01
5.98523200e-01 -3.26980352e-01 7.82098845e-02 -1.34944296e+00
-4.30588096e-01 -9.05112326e-01 1.41614210e-02 8.37486386e-01
5.22628486e-01 -5.27868330e-01 5.24916887e-01 -2.21231028e-01
-3.92284691e-01 -9.19887960e-01 -7.39506245e-01 -2.00067118e-01
-1.64140359e-01 -6.04873121e-01 2.17504725e-01 9.80564296e-01
9.47400630e-02 8.60907793e-01 -6.55906916e-01 7.24426424e-03
2.78851539e-01 1.32078394e-01 1.14511475e-01 -1.31423008e+00
-6.21821880e-01 -2.74313599e-01 -6.81519449e-01 -3.59467626e-01
-3.71591747e-01 -1.10757399e+00 -2.17326194e-01 -1.49175906e+00
-1.59357525e-02 -3.89835864e-01 -7.80722380e-01 6.92773521e-01
-9.11548175e-03 7.06459820e-01 4.09109099e-03 1.04029603e-01
-2.02554360e-01 -2.44540557e-01 7.89543092e-01 -3.74492817e-02
-2.50827782e-02 3.01487803e-01 -6.90253258e-01 2.23553002e-01
1.07454622e+00 -6.37021959e-01 -6.49225175e-01 -2.74474025e-01
5.31553388e-01 5.31182528e-01 3.87201369e-01 -1.21654880e+00
4.12683412e-02 6.37681961e-01 5.44821918e-01 -4.68739182e-01
1.79737449e-01 -6.80357456e-01 6.98887289e-01 6.61554337e-01
-4.80580211e-01 6.47464216e-01 1.57885984e-01 1.69179842e-01
-4.38723981e-01 1.00726061e-01 5.94946623e-01 -2.91570395e-01
-6.51608780e-02 1.77315444e-01 -4.46893573e-01 3.17475259e-01
7.32347071e-01 -1.34799108e-01 1.27111718e-01 -2.56495982e-01
-1.33857882e+00 7.95254484e-02 -3.41267973e-01 3.65473628e-01
7.41681635e-01 -1.20453024e+00 -1.24369800e+00 4.08125520e-01
9.40516964e-02 -3.90183479e-02 1.35071307e-01 1.28477407e+00
-7.04007447e-01 4.11198854e-01 -4.15965468e-01 -8.84001315e-01
-1.27764714e+00 7.57894665e-02 7.19681323e-01 -3.77270848e-01
-9.94983315e-01 7.33899295e-01 -3.74523163e-01 1.97823361e-01
2.98974931e-01 -6.33779883e-01 -1.44458354e-01 1.96953610e-01
4.09363300e-01 1.41496390e-01 5.35351634e-01 -2.29865342e-01
-5.29786587e-01 4.75778282e-01 1.42171249e-01 -2.16533050e-01
1.22872555e+00 9.14442241e-02 -1.32269487e-01 9.81909454e-01
1.11980462e+00 -3.37131202e-01 -4.77017432e-01 1.81561351e-01
-1.90847050e-02 1.35540560e-01 -9.80613753e-02 -1.24635947e+00
-1.02660811e+00 1.11109757e+00 1.08213854e+00 6.50319934e-01
1.45712018e+00 -2.96512485e-01 7.07406759e-01 -6.79780766e-02
2.30325699e-01 -5.11864305e-01 -2.92596370e-01 4.16630447e-01
8.60445619e-01 -8.01004827e-01 -2.63434082e-01 5.55204637e-02
-6.58321202e-01 1.10097659e+00 -1.36210442e-01 -8.41580629e-02
7.00101137e-01 4.47692156e-01 4.16197985e-01 -3.21215481e-01
-1.06624293e+00 2.81132072e-01 2.58049548e-01 3.39045137e-01
1.20543408e+00 2.82792181e-01 -4.89713639e-01 5.83596110e-01
-3.45496565e-01 3.04805666e-01 5.41126072e-01 7.23599076e-01
6.62331581e-02 -1.10429680e+00 -2.06569344e-01 7.25846529e-01
-1.14252818e+00 -3.13221604e-01 -1.43764958e-01 2.09637985e-01
3.61185640e-01 8.82640839e-01 -1.09133191e-01 -4.75885332e-01
4.03508306e-01 5.61785400e-01 4.29519475e-01 -6.98376417e-01
-1.17197347e+00 3.11933219e-01 1.86693475e-01 -1.98078007e-01
-2.44374931e-01 -6.53932214e-01 -1.27781594e+00 1.22532248e-01
1.16510063e-01 4.47458953e-01 4.06364143e-01 7.00477123e-01
4.77105916e-01 1.13676453e+00 4.66240853e-01 -2.13150650e-01
-6.95220470e-01 -1.36237395e+00 -7.00611293e-01 3.94846469e-01
7.20727980e-01 1.42557276e-02 -3.03373691e-02 6.16566956e-01]
|
[14.365626335144043, 3.347059726715088]
|
a3ef6a23-2691-41e7-b4dd-43a95c763dae
|
objects-can-move-3d-change-detection-by
|
2208.0987
| null |
https://arxiv.org/abs/2208.09870v1
|
https://arxiv.org/pdf/2208.09870v1.pdf
|
Objects Can Move: 3D Change Detection by Geometric Transformation Constistency
|
AR/VR applications and robots need to know when the scene has changed. An example is when objects are moved, added, or removed from the scene. We propose a 3D object discovery method that is based only on scene changes. Our method does not need to encode any assumptions about what is an object, but rather discovers objects by exploiting their coherent move. Changes are initially detected as differences in the depth maps and segmented as objects if they undergo rigid motions. A graph cut optimization propagates the changing labels to geometrically consistent regions. Experiments show that our method achieves state-of-the-art performance on the 3RScan dataset against competitive baselines. The source code of our method can be found at https://github.com/katadam/ObjectsCanMove.
|
['Tomas Pajdla', 'Konstantinos Karantzalos', 'Torsten Sattler', 'Aikaterini Adam']
|
2022-08-21
| null | null | null | null |
['object-discovery']
|
['computer-vision']
|
[ 1.93409950e-01 2.36878358e-02 -7.00783283e-02 -3.51737887e-01
-4.50163186e-01 -9.00524914e-01 5.95065355e-01 2.34583929e-01
-2.21146584e-01 2.65673339e-01 -9.32704508e-02 7.19477981e-02
6.73235729e-02 -8.26273561e-01 -9.48470592e-01 -4.69543636e-01
-1.32799745e-01 7.82543600e-01 9.78475153e-01 -1.23827480e-01
3.20331842e-01 8.75803888e-01 -1.35609686e+00 1.01681218e-01
3.22775275e-01 7.18929887e-01 5.49375355e-01 8.41084361e-01
-4.88827229e-02 6.14543796e-01 -1.89535573e-01 -9.62787196e-02
5.35025418e-01 -2.91545182e-01 -1.00734103e+00 3.77925247e-01
4.99990582e-01 -3.52847189e-01 -5.78024328e-01 1.19776034e+00
4.14741226e-02 3.82571697e-01 4.69484419e-01 -1.12996256e+00
-4.43667680e-01 4.14707005e-01 -7.47871280e-01 3.58091921e-01
4.65996116e-01 1.54210240e-01 8.25643957e-01 -9.38989460e-01
1.12543368e+00 1.36538303e+00 1.80253193e-01 3.36536139e-01
-1.14679718e+00 -3.92915070e-01 6.58388495e-01 4.26311314e-01
-1.35919392e+00 -5.58378637e-01 8.46841753e-01 -6.36315405e-01
7.44187057e-01 1.54392928e-01 6.51966393e-01 4.49735284e-01
7.85858780e-02 8.09886098e-01 4.60143179e-01 -2.49462858e-01
4.26617920e-01 -2.29567185e-01 1.77176282e-01 8.94802690e-01
3.92285734e-01 -2.41439968e-01 -2.67027885e-01 1.17637731e-01
9.06635761e-01 1.45363092e-01 -2.90196568e-01 -9.80308414e-01
-1.34465528e+00 4.90295112e-01 7.52009332e-01 1.96603164e-01
-3.44533652e-01 5.59779406e-01 8.02746788e-02 3.45773133e-03
3.14032525e-01 3.80362660e-01 -6.07789040e-01 -1.07842915e-01
-3.93414140e-01 1.88827023e-01 3.03726494e-01 1.22120738e+00
9.70496893e-01 -3.22291404e-01 3.54457021e-01 3.23819846e-01
1.96924761e-01 3.20446044e-01 6.95305541e-02 -1.35656297e+00
2.07557350e-01 7.12975621e-01 2.79207736e-01 -1.15033841e+00
-4.79535013e-01 -2.63863616e-02 -1.89524561e-01 3.60158265e-01
1.53154716e-01 2.50858366e-01 -1.29825211e+00 1.26376212e+00
9.40751374e-01 5.01052976e-01 -1.94174930e-01 9.42877829e-01
8.87693405e-01 4.56560135e-01 -4.08589423e-01 9.50565487e-02
1.10011303e+00 -1.08086562e+00 -5.01452684e-01 -4.11681235e-01
6.17671907e-01 -5.07737935e-01 6.95241511e-01 1.94832176e-01
-1.09369898e+00 -4.86157477e-01 -7.52653718e-01 -2.87737578e-01
-3.75218064e-01 -3.52129936e-01 6.06708229e-01 7.42304400e-02
-9.71641600e-01 3.42975676e-01 -1.29890168e+00 -5.38666964e-01
4.92607921e-01 1.92867652e-01 -4.90742952e-01 -2.55122721e-01
-3.90741348e-01 7.78802991e-01 6.20120287e-01 7.26543069e-02
-9.71530914e-01 -3.87194127e-01 -1.00235713e+00 -3.07755798e-01
7.53191471e-01 -7.01866210e-01 1.58479476e+00 -7.12484360e-01
-1.33006632e+00 9.29291070e-01 -4.57230479e-01 -2.01073408e-01
5.38233578e-01 -3.34018201e-01 -1.78638268e-02 2.38921806e-01
1.38633177e-01 6.94809437e-01 4.62725192e-01 -1.68208456e+00
-1.03070784e+00 -4.88227993e-01 1.91517100e-01 3.81446421e-01
5.61468065e-01 -1.23317637e-01 -1.12108386e+00 -1.14677168e-01
8.49655449e-01 -1.21292663e+00 -4.17013407e-01 1.76384643e-01
-5.58601618e-01 -1.35606164e-02 1.05833662e+00 -2.33274594e-01
5.66093266e-01 -2.14039493e+00 2.45309338e-01 1.41184494e-01
2.59986937e-01 -2.88953751e-01 -1.75294042e-01 3.10603917e-01
2.42782801e-01 7.80930445e-02 -3.35643888e-01 -3.78738225e-01
-1.81570500e-01 1.65153667e-01 -6.17922321e-02 7.50174224e-01
-2.05060974e-01 7.67573178e-01 -1.25042987e+00 -2.14779139e-01
5.01897037e-01 4.25175548e-01 -4.11671460e-01 -1.10020511e-01
-4.56693947e-01 7.39453971e-01 -5.83271027e-01 6.09182596e-01
7.64997005e-01 -2.82709032e-01 -3.04746274e-02 -1.96744353e-02
-1.72654331e-01 4.13039118e-01 -1.29233706e+00 2.07819796e+00
2.18608156e-02 7.54861653e-01 -1.65158704e-01 -6.24354959e-01
5.72408855e-01 -1.08712070e-01 5.41942537e-01 -3.87461156e-01
1.66608632e-01 -9.69428197e-02 -7.79541284e-02 -4.53654140e-01
6.37477517e-01 4.11183029e-01 -5.58308959e-02 1.66053548e-01
-3.86651039e-01 -5.67993045e-01 3.04199100e-01 3.23245019e-01
1.42528057e+00 2.53982812e-01 2.57477909e-01 9.70456973e-02
8.86682048e-02 4.05444115e-01 7.62025535e-01 6.58624887e-01
-3.26280780e-02 6.82200730e-01 1.08193725e-01 -3.79204363e-01
-6.20977879e-01 -1.30798233e+00 -1.08045032e-02 7.87779033e-01
9.59143221e-01 -2.33687118e-01 -2.52244473e-01 -6.79971755e-01
4.84462120e-02 7.26939619e-01 -6.54628754e-01 3.74699049e-02
-6.29279673e-01 -5.34600914e-02 -2.52674609e-01 4.42583263e-01
3.09257001e-01 -9.94518578e-01 -1.07533610e+00 4.01708484e-02
-1.66137338e-01 -1.28450215e+00 -4.64790583e-01 5.94203025e-02
-1.04836786e+00 -1.26068199e+00 -5.52465320e-02 -7.78849483e-01
1.02608681e+00 7.17182040e-01 9.25341427e-01 4.04289924e-02
-4.28446978e-01 6.50876164e-01 -5.57058513e-01 -3.81710798e-01
-1.01630501e-01 2.83080339e-02 -8.76352936e-02 -2.14862138e-01
1.10027999e-01 -3.81418705e-01 -6.91135287e-01 3.86803001e-01
-5.73366702e-01 1.27189204e-01 3.10858697e-01 -1.10216200e-01
1.18839526e+00 3.16802621e-01 -9.35798660e-02 -9.12326097e-01
-1.40447766e-01 -4.57536787e-01 -7.98353016e-01 -9.48390886e-02
-1.05305016e-01 -8.27926770e-02 -8.96586701e-02 -3.00938547e-01
-9.14251983e-01 7.08069444e-01 4.76649493e-01 -6.65259719e-01
-5.09549975e-01 2.54331350e-01 -1.37133479e-01 1.38434291e-01
3.58123481e-01 -4.36672010e-02 -5.52385151e-01 -5.77036798e-01
7.32998967e-01 1.26120359e-01 7.31683373e-01 -3.12417984e-01
9.50016499e-01 1.05971742e+00 -1.02585740e-02 -6.28305495e-01
-5.43377817e-01 -8.74178708e-01 -1.20257986e+00 -3.41599137e-01
8.67398024e-01 -7.61019409e-01 -5.20138025e-01 2.70725787e-01
-1.20295656e+00 -7.66310871e-01 -3.85492414e-01 5.03558755e-01
-4.85164672e-01 1.71867624e-01 -5.41647792e-01 -5.74591517e-01
1.61520332e-01 -9.72071469e-01 1.04439938e+00 2.94279307e-01
-1.73314542e-01 -8.52361619e-01 1.56623036e-01 1.57917917e-01
-1.01577960e-01 4.99749780e-01 5.26169777e-01 -3.47987384e-01
-1.18183565e+00 -1.81011647e-01 1.86884701e-02 -3.39175314e-01
6.76177204e-01 1.93944991e-01 -6.37000740e-01 -2.33037964e-01
-3.17928821e-01 1.85805798e-01 8.05379093e-01 5.41196346e-01
1.11358321e+00 -2.38735750e-01 -8.43098819e-01 5.78829825e-01
1.35445142e+00 5.73497951e-01 4.97959882e-01 4.19267267e-01
9.99171376e-01 5.45436800e-01 7.32532144e-01 3.55663925e-01
6.64421678e-01 7.67729461e-01 9.37218904e-01 7.29185194e-02
-2.71766961e-01 -1.57968864e-01 9.64921266e-02 5.56114197e-01
-1.76317114e-02 -3.34749520e-01 -1.06346798e+00 6.87604666e-01
-2.05682135e+00 -8.21527719e-01 -5.96549451e-01 2.09850883e+00
3.68312389e-01 2.26443529e-01 -7.22917840e-02 -2.84743547e-01
8.01445901e-01 4.00584862e-02 -1.01437736e+00 -5.39200120e-02
1.96024403e-01 -2.96142250e-01 6.85035527e-01 7.97245383e-01
-1.05471122e+00 1.37293828e+00 5.52151489e+00 2.22151298e-02
-8.31398547e-01 4.55532297e-02 2.74423063e-01 -3.72844160e-01
-9.51669440e-02 1.59524813e-01 -7.27875352e-01 1.19790405e-01
1.85930222e-01 7.72987772e-03 3.10055763e-01 8.03639114e-01
3.35964471e-01 -4.96887207e-01 -1.40676498e+00 8.70526433e-01
5.57918437e-02 -1.14405954e+00 -1.75806105e-01 1.00321442e-01
8.10260534e-01 4.35622692e-01 -1.03571236e-01 -1.29783601e-01
8.17808628e-01 -5.19111454e-01 1.06951165e+00 5.69381475e-01
3.60981792e-01 -3.96353602e-01 3.42065334e-01 2.08633572e-01
-1.46246099e+00 3.02150905e-01 -2.91004568e-01 -8.78286585e-02
3.36265355e-01 5.17041147e-01 -1.21812654e+00 4.06663120e-01
9.55732644e-01 8.58650267e-01 -6.75423563e-01 1.37437952e+00
-4.69377875e-01 2.11299345e-01 -3.97278458e-01 2.98781246e-01
-5.45472577e-02 -1.72688037e-01 8.19628954e-01 9.90426958e-01
2.92086810e-01 4.08443570e-01 4.63099986e-01 8.00838411e-01
-1.64209351e-01 -2.91766107e-01 -6.74400389e-01 2.85222888e-01
7.77739465e-01 1.20448101e+00 -1.43457937e+00 -2.59954810e-01
-1.69427842e-01 1.28594077e+00 1.30310267e-01 4.29850608e-01
-6.94504023e-01 -2.48739466e-01 7.66925931e-01 2.24999994e-01
6.93370461e-01 -6.16603255e-01 -1.12155214e-01 -9.57356334e-01
1.12443738e-01 -1.93017349e-01 3.05509716e-01 -1.04920101e+00
-8.37018490e-01 4.58715171e-01 1.07948825e-01 -1.21138906e+00
7.99634121e-03 -2.67852932e-01 -3.37324589e-01 1.45126179e-01
-1.39689064e+00 -7.52675831e-01 -7.30764925e-01 5.68074882e-01
8.88743162e-01 4.91205722e-01 3.21739525e-01 -6.71264529e-02
-2.95386702e-01 -3.96206677e-02 -2.75579467e-02 1.55641705e-01
3.48376483e-01 -1.08948612e+00 6.12966955e-01 1.07202351e+00
3.21999371e-01 3.41591209e-01 7.59151101e-01 -9.12782550e-01
-1.34146130e+00 -1.13679624e+00 4.08892512e-01 -8.95817518e-01
4.25650924e-01 -5.37090123e-01 -9.10501301e-01 1.15803671e+00
-2.59325467e-02 3.54406774e-01 9.31638479e-02 -1.02738947e-01
-2.84813076e-01 1.38969943e-01 -1.00169611e+00 6.18895352e-01
1.62079084e+00 -1.91154867e-01 -4.58432496e-01 4.34477240e-01
1.04996681e+00 -9.93515790e-01 -4.82462734e-01 4.14251953e-01
3.14176083e-01 -7.20729411e-01 9.64241147e-01 -4.83995140e-01
5.37517853e-02 -8.65704060e-01 -2.77250528e-01 -1.24817908e+00
-5.17997324e-01 -3.78164738e-01 -2.02246010e-01 7.68311501e-01
2.92657644e-01 -5.05144715e-01 8.26952755e-01 5.80225646e-01
-4.24456775e-01 -4.90877181e-01 -8.15214574e-01 -9.43895817e-01
-4.53456044e-01 -4.11019892e-01 6.01875246e-01 1.15056908e+00
-2.59892195e-01 2.27324307e-01 3.01680416e-01 7.82780349e-01
5.46286881e-01 3.42466593e-01 1.11243713e+00 -1.40641201e+00
-6.44670278e-02 -5.46293080e-01 -7.53249884e-01 -1.21842659e+00
4.64725215e-03 -8.59691560e-01 5.08216739e-01 -2.20885634e+00
1.53779998e-01 -6.64032936e-01 -8.03366750e-02 7.38542438e-01
1.15313746e-01 4.79026213e-02 3.03155422e-01 3.98757935e-01
-9.93275642e-01 4.36490536e-01 1.14672208e+00 -1.86780274e-01
-6.74219847e-01 -1.31880179e-01 -3.32159907e-01 9.89412308e-01
1.00881708e+00 -7.09136188e-01 -4.61465955e-01 -7.00113118e-01
1.40724078e-01 -6.70360103e-02 3.30517918e-01 -1.01472747e+00
3.54872286e-01 -4.13706422e-01 2.15493202e-01 -7.08812118e-01
6.53484881e-01 -9.59717155e-01 4.35360968e-01 4.69485641e-01
-1.50835842e-01 2.17619568e-01 3.32044572e-01 6.84348047e-01
2.55563200e-01 -2.47738481e-01 6.43761337e-01 -3.40542287e-01
-1.20216095e+00 4.97825533e-01 -2.82439768e-01 -1.36249736e-01
1.26524007e+00 -4.60503250e-01 -3.33715022e-01 -2.59841025e-01
-7.31194317e-01 3.35660428e-01 9.70860660e-01 7.32022047e-01
8.42468381e-01 -1.06212580e+00 -5.24519384e-01 -2.79831886e-01
1.97596893e-01 7.59481966e-01 1.66942105e-01 5.80365837e-01
-7.22821057e-01 2.49096751e-01 1.55720130e-01 -8.02997470e-01
-1.48181915e+00 5.89539886e-01 3.82240474e-01 3.95080507e-01
-1.08242500e+00 1.05700624e+00 6.12758934e-01 -4.61373836e-01
2.10010692e-01 -6.97131395e-01 4.48902622e-02 -3.13912958e-01
3.39517117e-01 4.55009609e-01 -8.99204984e-03 -7.43695199e-01
-6.21663511e-01 9.49451149e-01 -1.83904871e-01 -1.07318752e-01
1.28893840e+00 -4.35814768e-01 -9.55152735e-02 8.39706540e-01
1.14630568e+00 1.13715202e-01 -1.52618670e+00 -3.76158148e-01
-7.26954192e-02 -9.16062236e-01 9.49873254e-02 -5.78819454e-01
-1.16998136e+00 3.95875812e-01 6.08350158e-01 -3.15959752e-02
9.61121917e-01 7.13952720e-01 6.26027346e-01 5.98167956e-01
8.36490870e-01 -7.24900961e-01 2.90410966e-01 5.69627643e-01
9.23861682e-01 -1.14817572e+00 1.30985260e-01 -6.42424107e-01
-5.02412617e-01 8.14718664e-01 6.53764665e-01 -1.66869447e-01
5.45061767e-01 1.78272486e-01 -1.05878383e-01 -6.17949486e-01
-4.77785468e-01 -4.54615712e-01 4.39925417e-02 5.85404396e-01
-1.75713539e-01 5.93524240e-02 2.67876565e-01 -1.26903385e-01
-2.85052389e-01 -3.07343543e-01 8.01629663e-01 1.18504167e+00
-7.43754327e-01 -6.21410847e-01 -3.02342325e-01 1.45266637e-01
8.20815489e-02 1.95816353e-01 -6.54329777e-01 7.45042622e-01
1.99372426e-01 8.75856221e-01 4.04573619e-01 -4.07665789e-01
5.49782097e-01 -3.13544989e-01 6.19297385e-01 -9.38660979e-01
1.56441376e-01 1.76060665e-02 -1.74241707e-01 -8.39304924e-01
-5.56841195e-01 -1.07720208e+00 -2.06234145e+00 -3.64822112e-02
-3.05404693e-01 -1.33061469e-01 7.22606361e-01 5.60468793e-01
5.70966542e-01 6.89242661e-01 4.89280701e-01 -1.06317687e+00
5.66847324e-01 -4.42493021e-01 -3.21008772e-01 3.07431936e-01
5.04334986e-01 -8.54676247e-01 -2.97815084e-01 4.43069309e-01]
|
[7.830582618713379, -2.341735601425171]
|
f332dd69-2594-40ff-b173-4395e668c25d
|
real-time-facial-expression-recognition-using
|
2202.00102
| null |
https://arxiv.org/abs/2202.00102v1
|
https://arxiv.org/pdf/2202.00102v1.pdf
|
Real-Time Facial Expression Recognition using Facial Landmarks and Neural Networks
|
This paper presents a lightweight algorithm for feature extraction, classification of seven different emotions, and facial expression recognition in a real-time manner based on static images of the human face. In this regard, a Multi-Layer Perceptron (MLP) neural network is trained based on the foregoing algorithm. In order to classify human faces, first, some pre-processing is applied to the input image, which can localize and cut out faces from it. In the next step, a facial landmark detection library is used, which can detect the landmarks of each face. Then, the human face is split into upper and lower faces, which enables the extraction of the desired features from each part. In the proposed model, both geometric and texture-based feature types are taken into account. After the feature extraction phase, a normalized vector of features is created. A 3-layer MLP is trained using these feature vectors, leading to 96% accuracy on the test set.
|
['Ahmad Kalhor', 'Mehdi Tale Masouleh', 'Ehsan Saeedizade', 'Mohammad Amin Haghpanah']
|
2022-01-31
| null | null | null | null |
['facial-expression-recognition', 'facial-landmark-detection']
|
['computer-vision', 'computer-vision']
|
[ 3.61941725e-01 1.34204095e-02 -1.72698125e-01 -6.38331592e-01
-1.12126261e-01 2.86402199e-02 3.25871170e-01 1.83582440e-01
-6.07157052e-01 2.99305707e-01 -2.38257408e-01 1.50185540e-01
8.86017829e-03 -9.26222324e-01 -2.38585994e-01 -9.69595909e-01
-1.58575639e-01 1.63226768e-01 -1.75831556e-01 2.51401544e-01
2.26969674e-01 1.28046179e+00 -1.99800682e+00 2.75901526e-01
2.58739799e-01 1.44875252e+00 -2.25571021e-02 3.44249964e-01
-2.46775433e-01 5.49943507e-01 -4.18909043e-01 -6.26845732e-02
3.94432954e-02 -2.18091011e-01 -5.30956149e-01 4.35366780e-01
2.11748421e-01 -3.10101241e-01 2.00121716e-01 9.19599235e-01
3.94569010e-01 1.48749605e-01 8.73582840e-01 -1.02330434e+00
-5.21479314e-03 7.41220936e-02 -6.01377666e-01 -3.51603061e-01
3.93322676e-01 -2.14543685e-01 6.68100357e-01 -1.37119114e+00
3.99608016e-01 1.19625652e+00 3.49313647e-01 4.03407902e-01
-1.08405721e+00 -7.93353558e-01 -2.85014331e-01 2.61870176e-01
-1.64174175e+00 -7.06153095e-01 1.19500506e+00 -4.24139470e-01
5.12260020e-01 1.30927280e-01 7.00077772e-01 5.14631212e-01
3.27658474e-01 4.12822127e-01 1.18121517e+00 -7.25330591e-01
2.91835159e-01 4.30654854e-01 4.00569253e-02 7.87921429e-01
-1.17344044e-01 8.35255533e-03 -2.48409629e-01 -6.62865564e-02
5.28804839e-01 8.83513689e-02 5.85991181e-02 -1.81391820e-01
-3.64117175e-01 5.52339017e-01 3.84678215e-01 6.32482111e-01
-8.10636222e-01 -3.96063596e-01 3.83522630e-01 -1.07743762e-01
2.50759035e-01 -1.97249830e-01 2.14621741e-02 2.42031574e-01
-1.08680356e+00 -4.66931015e-01 7.95307338e-01 3.40689898e-01
1.14208138e+00 -1.19846784e-01 1.33618757e-01 6.89497769e-01
6.18775070e-01 4.90434200e-01 3.86270434e-01 -4.19089347e-01
-4.79656048e-02 1.04141724e+00 -1.36541843e-01 -1.45086157e+00
-6.04187191e-01 -4.71259616e-02 -1.02464044e+00 6.62469923e-01
2.32315525e-01 -8.32408667e-02 -7.27019131e-01 1.20238590e+00
6.98753655e-01 -6.94985539e-02 1.81184709e-01 7.51309335e-01
8.70993435e-01 6.07467234e-01 1.28452420e-01 -4.11331147e-01
1.58833706e+00 -3.51634622e-01 -4.89005417e-01 -8.22421536e-02
1.17108077e-01 -8.33632052e-01 6.48864090e-01 3.75606149e-01
-6.53161824e-01 -7.46146619e-01 -1.00860703e+00 1.41337097e-01
-4.97042537e-01 8.30761254e-01 1.75287768e-01 6.77248716e-01
-7.79444039e-01 2.79176861e-01 -5.51846266e-01 -2.26370707e-01
3.06472689e-01 5.61163247e-01 -7.76967227e-01 9.48841497e-02
-6.94957078e-01 9.06922340e-01 4.59996790e-01 7.25318491e-01
-5.03051937e-01 -1.82977933e-02 -1.01075757e+00 2.51553476e-01
1.55301755e-02 -1.48216793e-02 5.97145498e-01 -1.56324327e+00
-1.91156185e+00 1.09122407e+00 -5.65849304e-01 1.71600983e-01
1.12024225e-01 2.47066334e-01 -4.07841921e-01 3.82257104e-01
-1.86134323e-01 3.93248320e-01 1.35716307e+00 -1.14361167e+00
-5.98268449e-01 -5.71821868e-01 -2.23470211e-01 -3.99149908e-03
-3.68567079e-01 3.28663021e-01 -4.09574538e-01 -1.50192454e-01
1.78684473e-01 -5.69823027e-01 1.60880685e-01 -2.91929040e-02
-2.67489165e-01 -3.44191402e-01 8.84536028e-01 -7.36855686e-01
8.91356945e-01 -2.71246123e+00 7.76932240e-02 8.58602822e-01
7.15731755e-02 1.65243998e-01 -6.58945441e-02 6.87125996e-02
-1.86336756e-01 -4.31165487e-01 1.00451997e-02 -4.07440305e-01
-2.82381147e-01 2.85554621e-02 2.55079150e-01 6.16219044e-01
3.67669225e-01 3.33111733e-01 -2.64305383e-01 -8.04643929e-01
4.99401808e-01 7.25043833e-01 -1.79426625e-01 1.00973509e-01
2.19045579e-02 3.70861202e-01 -5.28171539e-01 8.14500988e-01
9.36044395e-01 3.40076685e-01 3.14981937e-01 -2.09134921e-01
-2.37301901e-01 -3.56513530e-01 -1.42509079e+00 8.86209309e-01
-7.27302492e-01 3.98648500e-01 4.52775687e-01 -9.71949279e-01
1.36811888e+00 5.38706660e-01 6.23608649e-01 -6.15314901e-01
6.01407588e-01 3.06687802e-01 -2.56554008e-01 -5.91718137e-01
1.08667195e-01 -5.24073951e-02 2.19820857e-01 4.89015013e-01
1.04277581e-01 3.88301015e-01 2.28084579e-01 -5.72910428e-01
4.57151800e-01 -1.13129104e-02 4.62803364e-01 -2.22571809e-02
1.27287805e+00 -3.01506370e-01 3.68226528e-01 -2.63743903e-02
-7.21231923e-02 1.42397776e-01 3.23798686e-01 -7.31916249e-01
-6.01981878e-01 -5.83842576e-01 -1.57865390e-01 9.80171859e-01
-1.06649391e-01 -6.97862580e-02 -8.97996724e-01 -4.84016567e-01
-7.38307759e-02 2.15217516e-01 -5.32724917e-01 5.61100105e-03
-4.63111728e-01 -6.01802766e-01 2.00762108e-01 1.23672150e-01
5.61478436e-01 -1.44860566e+00 -1.09907830e+00 1.34839281e-01
2.73539633e-01 -8.04075181e-01 1.77070856e-01 7.83380121e-03
-6.67608440e-01 -1.19201601e+00 -5.78622103e-01 -1.05641007e+00
1.08134472e+00 -1.62121296e-01 3.39865327e-01 -2.93205678e-03
-3.72833222e-01 6.58578277e-02 -1.81900382e-01 -4.03727859e-01
-2.74606377e-01 -1.91432938e-01 9.24181864e-02 9.30248260e-01
6.82401657e-01 -4.11746025e-01 -2.45482132e-01 1.80542231e-01
-5.77705383e-01 -2.89740056e-01 1.02933979e+00 6.71342194e-01
5.40521264e-01 2.57898211e-01 3.62709522e-01 -3.58173639e-01
3.43523711e-01 -1.43107161e-01 -8.38213146e-01 3.23225766e-01
-5.20875938e-02 -1.67948231e-01 8.73593509e-01 -3.49504232e-01
-1.07440233e+00 5.67228854e-01 -4.08312768e-01 -1.82056859e-01
-6.17603958e-01 5.41386902e-01 -4.69890356e-01 -4.42179590e-01
4.05449033e-01 3.46789449e-01 3.92981797e-01 -3.62202734e-01
1.59306765e-01 1.08606017e+00 3.04654092e-01 -1.09751068e-01
6.08983099e-01 4.13065523e-01 2.25684598e-01 -1.31987059e+00
-3.56769800e-01 -3.95969629e-01 -1.14044178e+00 -6.62608922e-01
8.01711261e-01 -3.91360670e-01 -9.61302578e-01 6.59179151e-01
-1.23469782e+00 2.16478035e-01 -3.07496786e-02 6.58777893e-01
-3.35558414e-01 9.54760239e-03 -4.22316432e-01 -1.20577538e+00
-5.87529004e-01 -1.09382617e+00 7.32198536e-01 5.74241698e-01
-7.93101862e-02 -6.37775660e-01 -2.48413429e-01 3.41925584e-02
2.45411918e-01 2.60279149e-01 9.20819461e-01 -6.16574466e-01
-1.85992077e-01 -6.50593698e-01 -2.38766745e-01 5.67970455e-01
4.08758730e-01 3.05053502e-01 -1.13999522e+00 -9.65043306e-02
2.77651995e-01 -1.87111616e-01 4.66834933e-01 3.48194510e-01
8.31355095e-01 -1.79045811e-01 -3.48545402e-01 4.64857578e-01
1.38413727e+00 5.70777357e-01 3.33768845e-01 1.59591377e-01
2.67697036e-01 7.25537837e-01 5.81309617e-01 5.16439021e-01
1.77283779e-01 2.26513043e-01 3.14669341e-01 -3.08864743e-01
2.59288579e-01 2.74836272e-01 3.13533723e-01 2.77919203e-01
-1.52813196e-01 3.35129499e-01 -5.60068429e-01 1.26083866e-02
-1.19486380e+00 -7.58846283e-01 3.79865915e-01 2.20017910e+00
4.86798733e-01 -3.32820043e-02 1.38914794e-01 6.09253883e-01
8.36795568e-01 8.94391164e-02 -2.22570404e-01 -5.53368032e-01
1.12084962e-01 3.83922577e-01 -2.35341657e-02 3.88313621e-01
-1.18904054e+00 8.26156080e-01 5.65103960e+00 5.21977365e-01
-1.92365408e+00 -3.59003037e-01 5.60904086e-01 9.94837657e-02
4.69297737e-01 -4.07155514e-01 -8.37150276e-01 2.88272113e-01
6.94379330e-01 3.11800446e-02 4.02996212e-01 1.04744768e+00
2.98371941e-01 -3.62619489e-01 -8.25793743e-01 1.07976258e+00
3.06999803e-01 -7.12969005e-01 -5.60935289e-02 -8.87867436e-02
5.79316653e-02 -5.61231911e-01 -6.26133308e-02 -2.91828346e-02
-6.64750874e-01 -1.03764617e+00 4.87434298e-01 8.20436478e-01
7.95214236e-01 -1.20925796e+00 9.56980884e-01 3.73733044e-01
-1.25184119e+00 -2.58536279e-01 -2.99323827e-01 -1.38795646e-02
-8.86416212e-02 7.69813061e-01 -8.60963821e-01 5.83209395e-01
4.46070045e-01 2.97170043e-01 -3.09206873e-01 8.85262012e-01
-2.03868434e-01 1.90798879e-01 -5.57211161e-01 -1.50261506e-01
-2.21367162e-02 -4.36304718e-01 2.19171926e-01 1.17472446e+00
3.22344929e-01 4.64860350e-02 1.13507397e-01 5.34916818e-01
1.10556430e-03 6.59693062e-01 -5.10753691e-01 1.42988667e-01
2.02543750e-01 1.85234392e+00 -9.22928989e-01 -2.80591577e-01
-4.30959344e-01 7.82501638e-01 2.19444349e-01 1.39750168e-01
-4.79967088e-01 -7.11775839e-01 1.96302533e-01 1.03502823e-02
1.61898836e-01 -1.43420482e-02 -2.71840785e-02 -5.95705330e-01
1.93877414e-01 -6.62312090e-01 2.72408605e-01 -4.03198153e-01
-7.51487970e-01 8.69553566e-01 -2.24471569e-01 -9.30893004e-01
-4.50035810e-01 -8.74347448e-01 -7.65505970e-01 1.08112049e+00
-1.32037795e+00 -1.26469600e+00 -3.60351771e-01 5.75920343e-01
9.38248038e-02 -2.05684707e-01 1.04732955e+00 4.28132623e-01
-8.74528885e-01 4.24549580e-01 -2.32499093e-02 3.99790138e-01
4.67219681e-01 -6.78705990e-01 -4.08577770e-01 7.47500896e-01
-6.15495294e-02 4.47342724e-01 2.22526550e-01 -3.83625031e-01
-1.23125541e+00 -9.79341924e-01 1.21279716e+00 3.01911354e-01
1.34082273e-01 -3.62551063e-01 -7.32797623e-01 5.57369113e-01
-3.05173457e-01 1.16710879e-01 6.99306667e-01 -1.52308390e-01
3.05563863e-02 -4.64743763e-01 -1.38105738e+00 2.97586292e-01
1.39820397e-01 -6.23943150e-01 -4.16715533e-01 -1.09768808e-01
-2.86280304e-01 -1.71752378e-01 -9.78798151e-01 3.00758213e-01
1.02961791e+00 -1.00111735e+00 6.65992320e-01 -1.40866324e-01
1.42717540e-01 -2.47077033e-01 1.50936782e-01 -9.73708212e-01
-2.24307656e-01 -5.69375902e-02 2.89579540e-01 1.19857311e+00
2.89696336e-01 -7.23594606e-01 7.77976811e-01 3.42015743e-01
3.36149961e-01 -8.22694480e-01 -1.07669342e+00 -3.09638292e-01
-4.85529810e-01 -1.19196117e-01 4.76308823e-01 5.33686876e-01
-1.60078313e-02 2.54663378e-01 -6.85469285e-02 2.47854427e-01
5.54241359e-01 3.66163999e-01 5.83238661e-01 -1.47317541e+00
3.29814732e-01 -3.88503164e-01 -6.06116593e-01 -4.19331789e-01
4.31599945e-01 -7.47142792e-01 -5.62918298e-02 -1.15525782e+00
-7.65951425e-02 -2.30464980e-01 -2.37059876e-01 5.84315479e-01
2.42920995e-01 5.15236676e-01 2.07662266e-02 -9.95873436e-02
-4.99091856e-02 4.16287333e-01 7.62051105e-01 4.13963059e-03
-3.58672053e-01 3.25704932e-01 -1.61985114e-01 9.36853170e-01
8.02745640e-01 -2.20742390e-01 9.44472179e-02 1.10582948e-01
-4.74281639e-01 -1.10259317e-02 3.32764596e-01 -1.16367376e+00
2.71062344e-01 -8.91521573e-02 9.38323736e-01 -5.36344349e-01
6.91237628e-01 -1.12741756e+00 9.31412801e-02 6.57617092e-01
1.73170522e-01 -2.85279959e-01 2.07536712e-01 -1.35370001e-01
-4.62910533e-01 -4.75659460e-01 1.04810154e+00 1.32075831e-01
-6.60194695e-01 1.28429756e-01 -4.69329208e-01 -9.11775410e-01
1.31461358e+00 -4.10137802e-01 3.53419542e-01 1.08331658e-01
-9.19799984e-01 -1.94770515e-01 1.63960978e-01 6.33364683e-03
8.66594493e-01 -1.14426267e+00 -4.99813437e-01 8.83642972e-01
7.41933510e-02 -2.24531293e-01 3.21351439e-01 8.38516414e-01
-5.80984652e-01 1.41748682e-01 -6.47167206e-01 -4.72244769e-01
-1.78403151e+00 6.25898182e-01 4.00872707e-01 2.22242549e-01
-2.84902900e-01 4.43998009e-01 -3.60965341e-01 -6.10994138e-02
2.05700621e-01 -1.04292773e-01 -8.66938293e-01 5.76654911e-01
7.48507202e-01 2.89809078e-01 3.70543450e-02 -1.33986604e+00
-5.65400779e-01 1.04446101e+00 1.81746960e-01 -8.38241260e-03
1.19605184e+00 1.47840723e-01 -5.44187725e-01 2.13514969e-01
1.34028804e+00 2.41262302e-01 -7.62784958e-01 -2.09208176e-01
4.63434635e-03 -3.75227064e-01 -9.94861871e-03 -3.34944516e-01
-1.00382888e+00 8.89052212e-01 6.61439359e-01 1.07577972e-01
1.55043888e+00 -2.65648842e-01 2.47515038e-01 2.22230181e-01
3.18825305e-01 -1.03148556e+00 -2.74822116e-01 3.18898231e-01
8.43673766e-01 -9.75651085e-01 -2.03263044e-01 -4.34013277e-01
-3.20639074e-01 1.60542023e+00 4.07726437e-01 -2.25266311e-02
9.07935560e-01 2.09935725e-01 1.19968258e-01 -1.45789489e-01
-2.12789834e-01 -2.50016540e-01 4.18625206e-01 2.64309198e-01
2.91487843e-01 -4.45181243e-02 -3.28273416e-01 4.99922842e-01
-1.48827389e-01 2.22001031e-01 5.50469011e-02 8.91867340e-01
-6.56201661e-01 -7.85181224e-01 -7.80801952e-01 2.95716494e-01
-4.66040283e-01 2.95803457e-01 -3.30889225e-01 6.99305236e-01
4.49721009e-01 9.14342701e-01 3.40659261e-01 -4.19657856e-01
3.98640066e-01 3.56540293e-01 4.59777594e-01 -4.55744535e-01
-4.63303059e-01 9.35123861e-02 -2.70160705e-01 -4.49424654e-01
-2.82756507e-01 -3.77902716e-01 -1.35344744e+00 -2.62566004e-02
-1.18714489e-01 2.77546853e-01 1.00493777e+00 9.17149842e-01
1.44535586e-01 1.38011143e-01 1.05469179e+00 -1.35521066e+00
-2.54748464e-01 -9.77521420e-01 -6.42405450e-01 1.99849501e-01
3.47259402e-01 -7.74139702e-01 -4.84182447e-01 4.14610794e-03]
|
[13.14210319519043, 0.7767835855484009]
|
b5dfda52-4047-441a-8740-8db9f6a3166a
|
generative-question-answering-learning-to
| null | null |
https://openreview.net/forum?id=Bkx0RjA9tX
|
https://openreview.net/pdf?id=Bkx0RjA9tX
|
Generative Question Answering: Learning to Answer the Whole Question
|
Discriminative question answering models can overfit to superficial biases in datasets, because their loss function saturates when any clue makes the answer likely. We introduce generative models of the joint distribution of questions and answers, which are trained to explain the whole question, not just to answer it.Our question answering (QA) model is implemented by learning a prior over answers, and a conditional language model to generate the question given the answer—allowing scalable and interpretable many-hop reasoning as the question is generated word-by-word. Our model achieves competitive performance with specialised discriminative models on the SQUAD and CLEVR benchmarks, indicating that it is a more general architecture for language understanding and reasoning than previous work. The model greatly improves generalisation both from biased training data and to adversarial testing data, achieving a new state-of-the-art on ADVERSARIAL SQUAD. We will release our code.
|
['Mike Lewis', 'Angela Fan']
|
2019-05-01
| null | null | null |
iclr-2019-5
|
['generative-question-answering']
|
['natural-language-processing']
|
[ 3.64978194e-01 7.96950638e-01 1.29333898e-01 -6.33599639e-01
-1.47736084e+00 -1.04339159e+00 7.84106672e-01 -1.44676894e-01
-3.62887442e-01 6.92927659e-01 6.55166268e-01 -6.51078522e-01
-2.83610448e-02 -1.13103795e+00 -9.97846901e-01 -1.50287569e-01
4.89129394e-01 1.05492532e+00 2.91985184e-01 -6.97524726e-01
-7.77980909e-02 -1.34432033e-01 -1.03099573e+00 7.77768493e-01
5.52214086e-01 7.09213853e-01 -2.38211557e-01 1.24824071e+00
-4.66646343e-01 1.73792660e+00 -9.29720581e-01 -1.30659783e+00
1.78901330e-02 -7.39861310e-01 -1.58144128e+00 -4.21747416e-01
9.79460239e-01 -4.93016630e-01 -5.92712343e-01 6.60049438e-01
3.76201451e-01 4.80517671e-02 8.09693575e-01 -1.07521296e+00
-1.41178858e+00 7.67022491e-01 3.22633713e-01 3.81109893e-01
7.63515115e-01 4.45998549e-01 1.52743649e+00 -8.52788568e-01
6.94459498e-01 1.43216980e+00 5.64510167e-01 1.23906839e+00
-1.36893010e+00 -4.14501101e-01 1.62413001e-01 3.28810781e-01
-5.23501158e-01 -1.59737378e-01 5.52472591e-01 -2.74924278e-01
1.08800304e+00 5.56215346e-01 2.26034056e-02 1.63688517e+00
2.85115033e-01 9.14517701e-01 9.99076486e-01 -8.90245885e-02
1.56495035e-01 6.78070784e-02 3.18603903e-01 4.67038929e-01
-2.45614409e-01 -1.91853046e-01 -2.42156684e-01 -3.83510321e-01
1.64890200e-01 -2.15336457e-01 -1.49721369e-01 -1.87285274e-01
-9.35147464e-01 1.29106033e+00 8.81259084e-01 -1.54093817e-01
-1.85948670e-01 5.54227591e-01 1.65444508e-01 7.02645957e-01
3.03781509e-01 8.94950271e-01 -5.47584236e-01 1.04767554e-01
-7.61476517e-01 1.06368589e+00 1.33557761e+00 7.94509113e-01
5.72147489e-01 -1.37579054e-01 -7.86350429e-01 5.01417994e-01
1.74812436e-01 6.53928220e-01 3.17263544e-01 -1.22857654e+00
6.11012697e-01 5.56741238e-01 7.40168840e-02 -7.48336732e-01
-8.67094398e-02 -2.95487046e-01 -5.56701779e-01 -1.34542286e-01
7.08512425e-01 -1.84630409e-01 -1.02167010e+00 1.91728091e+00
1.28061548e-01 -1.77736476e-01 2.77151138e-01 6.56403184e-01
1.42781270e+00 6.10981405e-01 2.89297998e-01 6.29451871e-01
1.47776115e+00 -1.15654767e+00 -5.92559338e-01 -1.04655480e+00
3.63849282e-01 -6.42463505e-01 1.34621632e+00 6.42261207e-02
-1.23955762e+00 -5.70938945e-01 -5.98510981e-01 -7.01930404e-01
-4.33277190e-01 -3.66631925e-01 4.63577896e-01 3.97054940e-01
-1.20924199e+00 1.89836398e-02 -1.70263097e-01 -4.61870730e-02
4.90215898e-01 2.50606257e-02 -1.26906708e-01 -4.88336027e-01
-1.61791646e+00 1.13964891e+00 6.35609627e-02 -2.13137969e-01
-1.25760460e+00 -8.48193169e-01 -9.95769143e-01 -1.27390614e-02
2.51356572e-01 -1.31682861e+00 1.71875012e+00 -5.82402587e-01
-1.14585865e+00 1.28391492e+00 -4.26895797e-01 -8.83634448e-01
4.08376545e-01 -4.83949959e-01 -2.45496318e-01 1.54285684e-01
2.44377300e-01 9.60863292e-01 8.91963601e-01 -1.15349674e+00
-5.97861260e-02 -1.83487028e-01 8.43009889e-01 -2.21276596e-01
1.86078101e-01 -1.47965506e-01 -7.20848292e-02 -5.15212834e-01
-6.79111257e-02 -7.15579569e-01 -2.62049288e-01 -1.38138920e-01
-4.31495249e-01 -5.46853304e-01 4.88526821e-01 -1.09130144e+00
1.00775969e+00 -1.59751737e+00 2.79255301e-01 -2.26120993e-01
3.02064002e-01 2.58770585e-01 -3.69731098e-01 4.95004624e-01
-1.48150548e-01 1.47933632e-01 -4.46292728e-01 -3.00698698e-01
5.10968745e-01 5.80807924e-01 -1.33018756e+00 -1.52736321e-01
8.12399089e-01 1.52048647e+00 -9.86476541e-01 -1.99525014e-01
-2.86026716e-01 2.04363048e-01 -1.03411686e+00 7.12793589e-01
-9.63777304e-01 1.91290021e-01 -3.28669012e-01 2.14055002e-01
5.33160031e-01 -4.87684160e-01 -2.78026313e-01 1.76856160e-01
9.67034996e-01 9.05327141e-01 -5.16179502e-01 1.61569178e+00
-6.68616891e-01 5.33354640e-01 -2.22714841e-02 -6.99177206e-01
9.73910153e-01 2.89523989e-01 -6.01520956e-01 -5.91155767e-01
-1.15986601e-01 1.87567040e-01 -1.78660318e-01 -7.55433440e-01
4.90101963e-01 -4.62658077e-01 -4.45315182e-01 4.54097807e-01
3.89201611e-01 -7.83348143e-01 6.83657452e-02 7.26803303e-01
1.33675826e+00 3.73422474e-01 -1.34739250e-01 -1.04636461e-01
7.21352339e-01 1.57818094e-01 -1.71855792e-01 1.08536768e+00
-3.15559953e-02 8.72978985e-01 7.17214942e-01 -5.07069409e-01
-8.56485248e-01 -1.58520961e+00 1.65431246e-01 1.36638069e+00
-1.14376277e-01 -2.72619665e-01 -8.62973750e-01 -1.23272741e+00
1.96162954e-01 1.45327306e+00 -1.01157463e+00 -3.84623796e-01
-7.05488384e-01 -1.76301628e-01 7.69617796e-01 6.09022379e-01
3.95308793e-01 -1.34817374e+00 -1.76472291e-01 1.25175118e-01
-5.01039147e-01 -1.06967962e+00 -4.00880992e-01 1.77635521e-01
-5.02571225e-01 -1.06594813e+00 -4.81186152e-01 -7.73213387e-01
4.77324188e-01 -3.12175602e-01 2.08208704e+00 1.24288552e-01
1.31587777e-03 7.15558589e-01 -7.55617544e-02 -3.56530368e-01
-8.77525628e-01 2.50714779e-01 -6.89735949e-01 -3.18116158e-01
6.61379278e-01 -3.18230361e-01 -8.26787055e-01 1.35371968e-01
-1.18057323e+00 -3.55586320e-01 1.37846485e-01 8.85966718e-01
3.25768113e-01 -7.08555937e-01 9.40626323e-01 -1.29904151e+00
8.80223691e-01 -9.50588942e-01 -1.21533051e-02 2.30743289e-01
-1.17238216e-01 4.41189617e-01 8.10139477e-01 -1.27858609e-01
-1.05877841e+00 -5.69232225e-01 -9.08952653e-01 7.31276944e-02
-3.29770178e-01 -6.01356290e-02 -1.49250925e-01 4.03992206e-01
1.23453856e+00 1.98706582e-01 -1.03608832e-01 -2.85149634e-01
1.00642955e+00 3.63354743e-01 8.77930760e-01 -7.81531990e-01
1.10874128e+00 4.44130450e-01 -2.09476426e-01 -1.61971077e-02
-1.72867560e+00 -2.34711483e-01 -1.16374768e-01 1.50097668e-01
1.14601421e+00 -7.77193308e-01 -5.18911660e-01 -1.00286700e-01
-1.54904258e+00 -3.64491284e-01 -6.92992330e-01 -2.91759938e-01
-6.14060938e-01 -3.59748350e-03 -7.23002672e-01 -6.04679286e-01
-4.66049433e-01 -8.18671584e-01 1.05109036e+00 6.70320839e-02
-8.29406917e-01 -1.37282085e+00 3.04160684e-01 1.09765756e+00
6.84735358e-01 2.63901740e-01 1.16646242e+00 -1.10057831e+00
-6.62405312e-01 -3.71427804e-01 -1.63092941e-01 7.51684785e-01
-3.49409252e-01 -6.44819796e-01 -1.22891867e+00 2.30883043e-02
3.28898042e-01 -1.11634564e+00 1.18563449e+00 -2.28100553e-01
1.15738833e+00 -6.12621665e-01 2.09244519e-01 2.49900937e-01
1.09376061e+00 -6.66030288e-01 9.64601934e-01 2.30896994e-01
5.16341329e-01 8.08272719e-01 2.04785749e-01 -3.83706778e-01
7.77511358e-01 1.74439147e-01 8.33037972e-01 1.56430483e-01
-5.36111772e-01 -7.59781718e-01 4.15291965e-01 2.87106961e-01
7.29225099e-01 -4.26464647e-01 -8.67442131e-01 9.05173123e-01
-1.49239230e+00 -1.06799483e+00 -4.45576638e-01 1.68375707e+00
1.20040226e+00 1.90163583e-01 -2.25819364e-01 -1.58568040e-01
1.86172858e-01 4.25662428e-01 -6.61600649e-01 -8.64315152e-01
-1.86662063e-01 9.18016255e-01 -5.80352359e-02 9.59793687e-01
-8.87457907e-01 9.03098285e-01 7.36323404e+00 6.99802697e-01
-4.54785049e-01 2.63755262e-01 7.13470280e-01 3.94883342e-02
-1.17425334e+00 8.89685452e-02 -7.24553347e-01 9.68092531e-02
1.04808414e+00 1.85959697e-01 3.98883253e-01 8.70943308e-01
-5.91194928e-01 2.38632366e-01 -1.32754481e+00 3.73099655e-01
4.16437030e-01 -1.28685319e+00 5.26267409e-01 -4.92213100e-01
6.87601805e-01 5.63030876e-02 2.97939211e-01 8.32361281e-01
9.54373956e-01 -1.62776291e+00 7.17750907e-01 6.18456483e-01
4.14517760e-01 -5.09927452e-01 8.87595534e-01 5.13574302e-01
-4.09011424e-01 -1.22071899e-01 -3.25576693e-01 -2.86908537e-01
7.83420429e-02 2.08763286e-01 -9.17099237e-01 3.76696736e-01
5.60631156e-01 1.00889154e-01 -9.31780934e-01 2.60847241e-01
-1.08752728e+00 9.58560884e-01 -3.39285913e-03 -1.91155463e-01
4.60658997e-01 3.72813582e-01 3.87499899e-01 1.13506818e+00
-1.21952817e-01 -1.30889371e-01 -1.79043129e-01 1.39394474e+00
-3.30788463e-01 -2.20102072e-01 -4.86731350e-01 2.98305713e-02
4.10233438e-02 1.18356323e+00 1.19141504e-01 -4.50680643e-01
-2.93633997e-01 1.09804523e+00 6.57522023e-01 5.09238720e-01
-8.47275674e-01 -2.81238377e-01 5.21588504e-01 1.90657720e-01
3.84198010e-01 1.37549773e-01 -4.67218965e-01 -1.10678875e+00
4.11535427e-02 -1.37995505e+00 6.98393762e-01 -1.14354372e+00
-1.66876602e+00 6.87624693e-01 -1.28835335e-01 -3.81138951e-01
-7.43744373e-01 -7.73913026e-01 -7.87143290e-01 1.29690874e+00
-1.66550672e+00 -1.26085806e+00 -1.50453195e-01 5.02750278e-01
4.50063407e-01 -2.87444051e-02 1.14301991e+00 -1.95145056e-01
8.40205774e-02 6.36013985e-01 -4.71648425e-01 3.22123826e-01
9.05287504e-01 -1.74284291e+00 1.01162171e+00 7.15811074e-01
4.54520643e-01 8.08052003e-01 9.91042018e-01 -2.02510431e-01
-1.10412693e+00 -1.09787607e+00 1.33179867e+00 -1.74175394e+00
8.49265873e-01 -4.94052410e-01 -1.11158955e+00 1.05824375e+00
6.36609972e-01 -1.48574635e-01 8.94120514e-01 1.33618966e-01
-1.04134977e+00 1.22370966e-01 -1.25427878e+00 6.54686451e-01
7.02812314e-01 -9.58692014e-01 -1.45613134e+00 5.83000720e-01
1.15850508e+00 -4.91799265e-01 -5.25077760e-01 2.27064669e-01
1.17731459e-01 -9.79835391e-01 1.16847467e+00 -1.29679739e+00
1.19636595e+00 -1.24151289e-01 -2.17192575e-01 -1.12930179e+00
-2.76451826e-01 -6.19750142e-01 -3.99839252e-01 1.11509943e+00
6.98867321e-01 -5.86941481e-01 7.44810283e-01 7.80816078e-01
1.41578555e-01 -8.40110302e-01 -8.11306059e-01 -2.75072694e-01
8.54867160e-01 -5.28449416e-01 6.83874428e-01 3.59756559e-01
-4.08390403e-01 8.70033979e-01 -8.36384371e-02 9.39171687e-02
4.94706124e-01 1.94496989e-01 9.33669627e-01 -6.18945956e-01
-8.19309175e-01 -1.29833639e-01 -2.40370948e-02 -1.57408702e+00
4.65712160e-01 -1.20889568e+00 7.19222277e-02 -1.89194942e+00
6.40902901e-03 2.87929356e-01 1.84794396e-01 3.06861490e-01
-8.71835351e-01 4.46935058e-01 2.75087859e-02 -2.61728346e-01
-7.46283889e-01 4.08679932e-01 1.45539129e+00 -2.79465795e-01
7.35552192e-01 -9.09083337e-03 -1.20740509e+00 7.20203817e-01
6.81595027e-01 -5.41419208e-01 -6.26461208e-01 -9.71369922e-01
8.12020898e-01 -1.09136589e-01 1.07530320e+00 -5.96193194e-01
1.81687742e-01 2.49986991e-01 2.70671070e-01 -3.77984166e-01
4.03613478e-01 -4.14099932e-01 -4.91352469e-01 2.93111324e-01
-1.06332529e+00 -5.71925454e-02 2.49978095e-01 7.36185730e-01
-2.81499684e-01 -4.57546800e-01 5.96362174e-01 -3.66317749e-01
-3.27836871e-01 5.89813590e-02 -2.86154114e-02 1.18973100e+00
3.83495480e-01 2.99594074e-01 -6.58096492e-01 -1.07394624e+00
-7.61545479e-01 5.32708824e-01 -2.67926417e-02 5.87374628e-01
5.47684789e-01 -1.31897318e+00 -1.16936636e+00 -5.14076315e-02
1.82688832e-01 2.52367079e-01 3.00115913e-01 8.52033729e-04
-5.95734775e-01 5.25043070e-01 4.28702801e-01 -2.07079425e-01
-8.50673139e-01 5.85602701e-01 5.56397498e-01 -7.02701569e-01
-2.60968119e-01 1.61438870e+00 3.60247970e-01 -1.12051523e+00
5.88833019e-02 -3.24625522e-01 -1.55574709e-01 -2.96578288e-01
7.42437124e-01 -3.35063599e-02 -1.32959768e-01 -2.85931438e-01
-2.87835747e-01 2.73716658e-01 -1.66506678e-01 -1.52113631e-01
8.74604106e-01 1.63897574e-01 -3.47322643e-01 2.26630375e-01
1.16875923e+00 2.42106300e-02 -1.05934739e+00 -3.86892766e-01
-6.53659925e-02 -2.09559649e-01 -4.95011032e-01 -1.23611772e+00
-4.15857404e-01 1.18946326e+00 -1.50812715e-01 4.83477145e-01
6.23671710e-01 6.11985445e-01 1.11707127e+00 5.38312793e-01
-3.24145034e-02 -4.95274365e-01 5.14203131e-01 8.56591582e-01
1.55261409e+00 -1.21000934e+00 -4.01089370e-01 -1.54137537e-01
-6.00462258e-01 8.54280591e-01 7.19724059e-01 -6.35347486e-01
2.33144611e-01 8.46467540e-02 3.51356477e-01 -3.13338697e-01
-1.05764937e+00 -2.71399476e-04 5.41100979e-01 7.50026703e-01
5.54723382e-01 -1.26940832e-01 2.38933340e-01 8.98330629e-01
-8.51320386e-01 -2.62155116e-01 2.37651706e-01 4.82437253e-01
-2.60433823e-01 -1.12886202e+00 -2.90351003e-01 2.88468808e-01
-7.16146767e-01 -4.16655749e-01 -8.75890493e-01 4.67468232e-01
-3.77736278e-02 1.32265711e+00 -8.51709545e-02 -1.59559682e-01
4.66363281e-01 6.42037213e-01 5.22396743e-01 -8.27905715e-01
-1.00936913e+00 -9.84062791e-01 4.39716190e-01 -5.87299943e-01
1.94954336e-01 -1.74036056e-01 -1.16905653e+00 -2.60189772e-01
1.34040624e-01 3.29914600e-01 2.72313654e-01 1.09529161e+00
3.05203974e-01 7.88742483e-01 2.26296216e-01 -1.73731130e-02
-1.31991279e+00 -9.84923840e-01 1.73082992e-01 8.21670473e-01
5.03201246e-01 2.17107624e-01 -7.05992579e-01 1.22163678e-02]
|
[11.113917350769043, 8.040690422058105]
|
f85773a1-d373-4ca9-80bd-db5fff518973
|
learning-sparse-analytic-filters-for-piano
|
2108.10382
| null |
https://arxiv.org/abs/2108.10382v3
|
https://arxiv.org/pdf/2108.10382v3.pdf
|
Learning Sparse Analytic Filters for Piano Transcription
|
In recent years, filterbank learning has become an increasingly popular strategy for various audio-related machine learning tasks. This is partly due to its ability to discover task-specific audio characteristics which can be leveraged in downstream processing. It is also a natural extension of the nearly ubiquitous deep learning methods employed to tackle a diverse array of audio applications. In this work, several variations of a frontend filterbank learning module are investigated for piano transcription, a challenging low-level music information retrieval task. We build upon a standard piano transcription model, modifying only the feature extraction stage. The filterbank module is designed such that its complex filters are unconstrained 1D convolutional kernels with long receptive fields. Additional variations employ the Hilbert transform to render the filters intrinsically analytic and apply variational dropout to promote filterbank sparsity. Transcription results are compared across all experiments, and we offer visualization and analysis of the filterbanks.
|
['Zhiyao Duan', 'Mojtaba Heydari', 'Frank Cwitkowitz']
|
2021-08-23
| null | null | null | null |
['music-information-retrieval']
|
['music']
|
[ 3.63176495e-01 -1.28971174e-01 -1.48868397e-01 -7.82941282e-02
-1.13438094e+00 -6.92021668e-01 5.10101259e-01 -7.36501217e-02
-2.84480989e-01 5.79048216e-01 6.51378691e-01 1.22922726e-01
-3.49124074e-01 -2.81231225e-01 -4.59018171e-01 -7.91418195e-01
-2.47374475e-01 -1.88905761e-01 -1.14236720e-01 -8.56977403e-02
1.58667445e-01 5.43524504e-01 -1.63845003e+00 7.05168426e-01
3.60110104e-01 1.23535025e+00 1.01517707e-01 8.52709413e-01
2.79950708e-01 8.02497268e-01 -6.90378904e-01 -1.72494382e-01
1.91926748e-01 -3.70956182e-01 -6.32037401e-01 -5.12966849e-02
4.26465303e-01 -6.70376346e-02 -4.33098257e-01 7.14703679e-01
8.60699415e-01 5.94013572e-01 5.19410431e-01 -8.90302956e-01
-1.75887868e-01 7.59362340e-01 -2.77066857e-01 4.33971941e-01
3.80344003e-01 2.99667027e-02 1.45204520e+00 -9.73141670e-01
2.67753869e-01 1.24028337e+00 7.73495495e-01 1.89022064e-01
-1.36633587e+00 -7.72084773e-01 1.67603884e-02 2.91335136e-01
-1.30505419e+00 -7.65339792e-01 9.95424211e-01 -4.29710686e-01
1.00584984e+00 2.81032592e-01 6.09513998e-01 1.19100630e+00
1.92199618e-01 9.85698998e-01 7.28755891e-01 -4.86296088e-01
1.44845948e-01 -2.96227813e-01 -1.87866837e-01 3.59828413e-01
-5.17101347e-01 3.50198448e-02 -1.19433630e+00 -2.41027325e-01
9.00304198e-01 -1.26000240e-01 -5.42655945e-01 -7.87851214e-02
-1.14854157e+00 7.53163755e-01 3.07584882e-01 3.13909799e-01
-3.96885306e-01 4.45428044e-01 5.71585894e-01 6.22865260e-01
4.53417420e-01 6.08041286e-01 -4.29702520e-01 -4.41600919e-01
-1.11322916e+00 4.43489760e-01 7.58234024e-01 6.33953393e-01
4.31669712e-01 4.48570073e-01 -4.92187500e-01 1.01316786e+00
1.82350367e-01 -1.09671308e-02 6.63178444e-01 -1.07822037e+00
2.61201829e-01 -4.77703474e-03 -1.57985583e-01 -5.52759051e-01
-2.87691444e-01 -9.94726717e-01 -7.92575717e-01 8.42295960e-02
3.56630176e-01 -8.99118707e-02 -6.37919605e-01 1.46915543e+00
6.55232444e-02 5.40303230e-01 -1.87783584e-01 8.95210445e-01
7.03614652e-01 6.95402384e-01 -8.27229097e-02 -1.89971309e-02
1.29423571e+00 -8.64857495e-01 -7.34106541e-01 8.05981383e-02
2.60027051e-01 -1.14455259e+00 1.25946522e+00 9.05363798e-01
-1.28019309e+00 -9.28090215e-01 -1.06823349e+00 -2.98005879e-01
-1.27640739e-01 3.74277711e-01 6.47787750e-01 3.85476828e-01
-8.65239322e-01 9.28180754e-01 -7.16031194e-01 -2.09253784e-02
4.35995668e-01 3.68523866e-01 -6.18291460e-02 3.47816527e-01
-1.15818071e+00 3.33731085e-01 2.05605119e-01 7.62077272e-02
-9.18175519e-01 -1.11449897e+00 -7.58988798e-01 5.20504355e-01
1.59421653e-01 -7.46094108e-01 1.63164842e+00 -8.31530571e-01
-1.90598881e+00 6.41057909e-01 1.70505848e-02 -6.65117741e-01
2.12757841e-01 -6.71980619e-01 -3.73307019e-01 2.10103214e-01
-1.96893990e-01 5.10434151e-01 1.66623533e+00 -5.47587633e-01
-6.21195376e-01 4.99066263e-02 -1.70490295e-02 2.13580966e-01
-4.23637062e-01 -7.54549056e-02 -2.08307371e-01 -1.29110217e+00
-1.06912374e-01 -7.10159242e-01 -2.50914805e-02 -1.51734740e-01
-2.16765434e-01 -1.63299680e-01 6.86515808e-01 -4.99140233e-01
1.48346364e+00 -2.47943282e+00 4.71014977e-01 1.21786505e-01
1.64323390e-01 1.86701715e-01 -1.89475030e-01 6.59071386e-01
-1.01676404e-01 -1.58509940e-01 -6.40077889e-02 -4.36497897e-01
1.57635927e-01 -1.39067426e-01 -7.43359089e-01 3.45933318e-01
4.75764185e-01 7.23613679e-01 -6.43293858e-01 -3.88462096e-02
1.24858774e-01 6.88818157e-01 -7.94249833e-01 2.04203472e-01
-2.12887481e-01 7.46818364e-01 -2.86822587e-01 4.29678947e-01
2.65490115e-01 -2.81166714e-02 -1.88544154e-01 -3.43067199e-01
-2.11796209e-01 6.35147870e-01 -1.15510941e+00 2.13300157e+00
-7.46635616e-01 7.86521018e-01 3.51175278e-01 -9.95283842e-01
8.53518307e-01 7.55405307e-01 5.68611383e-01 -2.80823201e-01
9.51159000e-02 2.25976259e-01 2.40435272e-01 -5.07971525e-01
3.31964612e-01 -3.36104810e-01 -5.62407114e-02 1.83096334e-01
5.30301869e-01 -3.54978442e-01 3.59222032e-02 -1.92024499e-01
1.01282203e+00 2.58347899e-01 2.33342931e-01 -2.54413426e-01
6.23588502e-01 -5.16015887e-01 5.35098493e-01 6.93818629e-01
6.94088489e-02 8.25031400e-01 3.54865015e-01 -3.08303922e-01
-8.02838683e-01 -1.12252629e+00 -1.28498510e-01 1.43858540e+00
-6.03853822e-01 -7.87020981e-01 -4.81803477e-01 -7.83819258e-02
-1.28159329e-01 2.48321533e-01 -3.45684886e-01 -1.98652938e-01
-5.07618785e-01 -9.22667310e-02 8.33858550e-01 5.86506784e-01
3.05346280e-01 -1.28725231e+00 -4.54369277e-01 4.82004613e-01
-2.95496117e-02 -8.96102071e-01 -5.12806892e-01 4.79912817e-01
-8.85478437e-01 -8.28910768e-01 -8.28510046e-01 -8.33384156e-01
-3.35112363e-01 -7.30708614e-02 1.07453203e+00 -3.45906764e-01
-2.95638651e-01 4.88387674e-01 -5.02180934e-01 -5.18251777e-01
-1.59181669e-01 5.26853204e-01 1.25418141e-01 2.71995157e-01
2.43748441e-01 -1.06099439e+00 -5.38974643e-01 -1.39795244e-01
-9.18282151e-01 -2.57799327e-01 5.04136384e-01 1.03156173e+00
6.22156560e-01 6.88871518e-02 9.33609188e-01 -6.22350574e-01
9.91983414e-01 -3.61643374e-01 -4.33101684e-01 -2.80895263e-01
1.08174220e-01 -1.12458929e-01 8.11872900e-01 -6.02802396e-01
-9.20340240e-01 4.12121452e-02 -4.27935660e-01 -6.87229633e-01
-2.53708810e-01 6.92165911e-01 -9.36399102e-02 2.42132872e-01
5.76202571e-01 -2.66228933e-02 -2.63514251e-01 -8.62570047e-01
2.46491879e-01 6.05885029e-01 5.66024721e-01 -6.07141554e-01
6.75413966e-01 3.42449516e-01 -3.27290818e-02 -1.23667943e+00
-1.05680180e+00 -6.78075910e-01 -6.02374852e-01 -1.96380705e-01
7.97532797e-01 -9.46864963e-01 -9.51887667e-01 2.43934184e-01
-9.99966860e-01 -5.12347221e-01 -5.32944143e-01 8.83238554e-01
-7.91887224e-01 -1.17485516e-01 -8.12576652e-01 -7.67674387e-01
-3.52952152e-01 -1.04057598e+00 1.33898938e+00 1.98582262e-01
-5.37929595e-01 -9.57670450e-01 1.31547987e-01 -1.23801135e-01
3.82061809e-01 -1.03408374e-01 1.00094700e+00 -4.74802613e-01
-4.08999771e-01 6.01719879e-03 2.14890704e-01 5.62622488e-01
8.53290632e-02 -2.08567947e-01 -1.73926926e+00 -5.35071075e-01
1.59376264e-01 -4.15114462e-01 1.16009867e+00 6.91374838e-01
1.46118164e+00 -4.25630510e-02 1.19716100e-01 9.51954782e-01
9.68058825e-01 3.44824381e-02 4.04225022e-01 4.21824716e-02
4.60748285e-01 4.19566661e-01 3.26065809e-01 5.35331666e-01
-3.50899279e-01 6.91981792e-01 1.76687762e-01 -1.02042124e-01
-2.70870596e-01 -3.68162304e-01 4.21226829e-01 1.18989420e+00
-5.96658364e-02 -4.02488410e-02 -5.71665823e-01 4.33403432e-01
-1.78847659e+00 -9.41637456e-01 2.42676362e-01 2.11139774e+00
6.48314238e-01 1.38840899e-01 2.49333814e-01 6.41239643e-01
2.04804808e-01 3.91714901e-01 -4.92739886e-01 -3.16567838e-01
-2.45605092e-02 1.11717749e+00 -2.67377328e-02 3.67389292e-01
-1.26850653e+00 7.96428800e-01 6.32765770e+00 1.01350367e+00
-1.24580204e+00 1.09265022e-01 1.95070282e-01 -3.06513458e-01
-6.86665811e-03 -1.83519304e-01 -6.62927568e-01 1.66685525e-02
8.40270996e-01 1.11563124e-01 4.70773339e-01 5.07954359e-01
4.84534770e-01 2.21623689e-01 -1.27524257e+00 1.25252211e+00
-3.33504051e-01 -1.49606860e+00 -2.38981515e-01 -1.58528239e-01
4.99969661e-01 -6.43884987e-02 4.43544924e-01 3.99423450e-01
-1.80645287e-01 -1.17244530e+00 7.03803599e-01 6.19223773e-01
9.44277227e-01 -9.61120248e-01 3.65152687e-01 1.12903282e-01
-1.50411379e+00 -3.81330878e-01 -1.80630356e-01 -5.26414871e-01
7.77495801e-02 4.92439032e-01 -7.59245813e-01 3.13877136e-01
9.24762785e-01 1.03532958e+00 -1.52142897e-01 1.23641145e+00
-2.07012966e-01 1.12483656e+00 -2.58005828e-01 2.28681758e-01
2.99330920e-01 -5.57745956e-02 7.38519192e-01 1.43479645e+00
3.54497969e-01 -2.40756422e-01 1.59886867e-01 6.98332846e-01
-3.42570007e-01 9.58100408e-02 -5.87157786e-01 -1.07185394e-01
3.74354631e-01 1.25252390e+00 -3.65575194e-01 -2.82359514e-02
-3.30477595e-01 7.73317456e-01 6.21165708e-02 6.25233650e-01
-5.89659631e-01 -6.21420562e-01 1.15529406e+00 9.57107022e-02
6.01178110e-01 -3.15796137e-01 1.97745152e-02 -1.04306710e+00
-2.20601540e-02 -1.01700735e+00 2.90654182e-01 -6.46289110e-01
-1.33225238e+00 4.59215522e-01 -2.92092115e-01 -1.40629423e+00
-4.56498027e-01 -5.53529024e-01 -7.73032486e-01 9.53732312e-01
-1.60583627e+00 -9.81030166e-01 2.10978135e-01 7.84625709e-01
7.82309890e-01 -3.57244730e-01 1.03850782e+00 5.50813138e-01
-2.15291038e-01 4.48323637e-01 1.75373018e-01 1.07682869e-01
7.92406023e-01 -1.14181876e+00 2.32796937e-01 5.05011439e-01
5.86928546e-01 5.95678389e-01 5.95540643e-01 -1.67762324e-01
-1.31475246e+00 -1.07263255e+00 5.84462106e-01 -8.18202738e-03
7.83925414e-01 -7.68496335e-01 -9.53348994e-01 5.66342473e-01
3.20143133e-01 -8.31296369e-02 8.92644346e-01 2.38890141e-01
-2.35862643e-01 -3.00177664e-01 -5.16137838e-01 4.55757052e-01
9.35095549e-01 -8.47917080e-01 -4.42827582e-01 1.64169759e-01
6.91818058e-01 -2.86395401e-01 -1.09272075e+00 1.58541128e-01
7.02125788e-01 -8.07964444e-01 1.13606620e+00 -7.05382288e-01
2.71884084e-01 -2.24515930e-01 -1.24506466e-01 -1.34197986e+00
-4.27805305e-01 -1.37088394e+00 -3.50154340e-01 1.36501276e+00
1.64677829e-01 -2.24110693e-01 7.16332257e-01 -2.76284933e-01
-4.50007170e-01 -6.22348249e-01 -9.93816495e-01 -5.85263133e-01
-4.28952612e-02 -6.95020080e-01 3.24835271e-01 6.75842643e-01
1.53743615e-02 5.55483222e-01 -5.08069277e-01 -3.79802249e-02
3.64115417e-01 1.94302276e-01 7.23152637e-01 -1.35261667e+00
-8.62064302e-01 -5.86152434e-01 -2.83218086e-01 -1.35698831e+00
2.16207147e-01 -9.87730443e-01 4.74461354e-03 -9.37520862e-01
-5.12234867e-01 -1.37334481e-01 -5.26296437e-01 2.60091603e-01
2.38830104e-01 4.28331971e-01 2.07030624e-01 1.39380231e-01
-1.09248400e-01 6.15201175e-01 1.17912841e+00 -9.90030095e-02
-4.24685121e-01 5.17685413e-01 -3.37132633e-01 9.46834028e-01
8.33020568e-01 -3.65823776e-01 -6.30784869e-01 -2.74126649e-01
2.04085797e-01 1.86372086e-01 3.38435024e-01 -1.29764295e+00
1.44590929e-01 2.77517080e-01 4.47396845e-01 -5.00870764e-01
7.73713648e-01 -5.20819008e-01 -1.51676061e-02 1.99957505e-01
-7.46763766e-01 -2.46416494e-01 2.86128759e-01 5.28304577e-01
-6.25933945e-01 -1.57183602e-01 6.16258562e-01 -4.63843718e-02
-4.32493150e-01 2.99057454e-01 -7.64223039e-01 6.56497292e-03
2.01253235e-01 -2.37627458e-02 3.47771436e-01 -6.21933460e-01
-1.12736952e+00 -2.87558973e-01 -1.52313039e-01 3.53409946e-01
5.37693381e-01 -1.28891933e+00 -6.95884466e-01 4.07411695e-01
-4.38959524e-02 -7.54580200e-02 1.82843804e-01 8.77637267e-01
-7.51500577e-02 4.49353725e-01 -1.63563460e-01 -6.61896825e-01
-1.05163062e+00 2.34048471e-01 2.22721398e-01 -2.76688993e-01
-8.40708613e-01 1.07160068e+00 3.14198226e-01 2.37135682e-02
7.74724007e-01 -5.73534191e-01 -2.23674968e-01 2.25212216e-01
6.66822851e-01 2.87454247e-01 2.70235121e-01 -3.30825388e-01
-6.57717809e-02 3.63030225e-01 1.39880210e-01 -2.79388219e-01
1.55172014e+00 6.24355376e-02 2.46093631e-01 8.18344355e-01
1.14582837e+00 3.09279561e-01 -1.47489107e+00 -2.23198354e-01
1.01522654e-01 -2.40386143e-01 2.61728317e-01 -4.07924652e-01
-9.77459431e-01 1.42488956e+00 4.24391985e-01 2.76520818e-01
1.40803576e+00 -2.06108555e-01 8.00448716e-01 3.41784745e-01
3.11245713e-02 -9.44057226e-01 2.86022425e-01 5.93083620e-01
1.05298960e+00 -6.90667093e-01 -1.44087344e-01 -1.81501016e-01
-2.16779456e-01 1.43110871e+00 1.95096415e-02 -5.04206657e-01
8.46241236e-01 4.82394576e-01 1.79809593e-02 -1.91901233e-02
-7.52817690e-01 -3.01737368e-01 5.17542303e-01 6.49020076e-01
9.37163293e-01 -2.90235072e-01 -8.24831799e-02 8.08420360e-01
-3.94188255e-01 2.46739537e-02 1.51230738e-01 7.11017668e-01
-3.82841885e-01 -1.18076754e+00 -3.40362579e-01 3.68515730e-01
-7.96053469e-01 -1.90637231e-01 -2.55554080e-01 4.67842311e-01
-6.58056438e-02 9.50620770e-01 1.81028731e-02 -2.53034472e-01
2.47230023e-01 2.95893252e-01 5.23990571e-01 -7.75443852e-01
-1.07354975e+00 7.43000150e-01 -1.43722430e-01 -5.02115309e-01
-4.40765202e-01 -5.12303472e-01 -1.04372001e+00 2.01270342e-01
-2.44922593e-01 2.00033382e-01 5.13358653e-01 7.52181113e-01
3.56528550e-01 1.00735009e+00 3.33584994e-01 -1.32268810e+00
-6.60495877e-01 -1.07787323e+00 -7.50983596e-01 2.86678791e-01
7.84960747e-01 -6.80685043e-01 -1.03757702e-01 3.76033902e-01]
|
[15.568767547607422, 5.475106716156006]
|
2e660c66-7c41-4da6-8cb4-530209948d83
|
noise-and-edge-based-dual-branch-image
|
2207.00724
| null |
https://arxiv.org/abs/2207.00724v1
|
https://arxiv.org/pdf/2207.00724v1.pdf
|
Noise and Edge Based Dual Branch Image Manipulation Detection
|
Unlike ordinary computer vision tasks that focus more on the semantic content of images, the image manipulation detection task pays more attention to the subtle information of image manipulation. In this paper, the noise image extracted by the improved constrained convolution is used as the input of the model instead of the original image to obtain more subtle traces of manipulation. Meanwhile, the dual-branch network, consisting of a high-resolution branch and a context branch, is used to capture the traces of artifacts as much as possible. In general, most manipulation leaves manipulation artifacts on the manipulation edge. A specially designed manipulation edge detection module is constructed based on the dual-branch network to identify these artifacts better. The correlation between pixels in an image is closely related to their distance. The farther the two pixels are, the weaker the correlation. We add a distance factor to the self-attention module to better describe the correlation between pixels. Experimental results on four publicly available image manipulation datasets demonstrate the effectiveness of our model.
|
['Jinjin Wang', 'Lin Zhu', 'Yanxiang Zhao', 'Yi Qian', 'Zhongyuan Zhang']
|
2022-07-02
| null | null | null | null |
['image-manipulation-detection', 'edge-detection', 'image-manipulation']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[ 4.17741567e-01 -3.00834119e-01 5.55125698e-02 -2.25214437e-01
9.28981155e-02 -1.13885783e-01 2.63637125e-01 2.51546223e-02
-2.96607345e-01 7.15598390e-02 1.26963422e-01 6.98696971e-02
6.60326611e-03 -8.09731066e-01 -9.08965886e-01 -6.77306175e-01
1.77641541e-01 -6.48864388e-01 5.67012250e-01 -2.48789608e-01
5.00397980e-01 1.76234692e-01 -1.27211726e+00 3.70994419e-01
8.08345973e-01 1.17500985e+00 6.35618269e-01 1.82662219e-01
-1.19409822e-01 8.12483668e-01 -6.79566205e-01 -7.53117204e-02
5.23530483e-01 -4.40813571e-01 -1.91767707e-01 3.10702026e-01
4.55263942e-01 -5.68333626e-01 -4.92539704e-01 1.76419222e+00
1.79769948e-01 1.62644073e-01 2.74502248e-01 -1.29517758e+00
-9.39075530e-01 7.00910270e-01 -1.06448781e+00 6.43702030e-01
1.10824130e-01 5.60401320e-01 4.77720380e-01 -7.01818526e-01
5.87362885e-01 1.35880423e+00 5.36584437e-01 1.46893099e-01
-1.05248022e+00 -9.54633117e-01 5.13399303e-01 4.37361807e-01
-1.10698020e+00 -5.60806394e-02 1.26814628e+00 -3.21710944e-01
3.01038384e-01 1.20152719e-01 5.00472307e-01 1.15855145e+00
-4.80606034e-02 7.58453846e-01 7.69243419e-01 -1.70235634e-01
-9.23557505e-02 3.69159728e-02 3.66053611e-01 6.71589136e-01
4.56426352e-01 2.74882894e-02 -3.14780653e-01 2.94657499e-01
1.08658195e+00 5.03510594e-01 -7.65526593e-01 -1.86739907e-01
-1.16385806e+00 3.48342001e-01 8.05789948e-01 4.47660476e-01
-4.13546801e-01 2.60922730e-01 4.35246497e-01 7.00997412e-02
1.55120313e-01 3.40346575e-01 -3.83716047e-01 1.38165364e-02
-6.00663543e-01 -1.81225315e-02 1.98489502e-01 1.20280015e+00
9.15267467e-01 -1.42111838e-01 -4.21208441e-01 6.12481654e-01
1.15913302e-02 9.42922011e-02 2.84105331e-01 -8.69886756e-01
6.11258030e-01 9.85494316e-01 4.26446721e-02 -1.44466662e+00
-9.68732461e-02 -4.48918551e-01 -9.56410408e-01 3.02840710e-01
5.07635772e-01 5.57380319e-02 -8.53420496e-01 1.58470500e+00
1.46105394e-01 4.74644810e-01 -5.86554825e-01 1.19871104e+00
6.46961510e-01 3.35610688e-01 6.95478842e-02 -1.73768904e-02
1.54571581e+00 -1.11274040e+00 -9.49564397e-01 -3.81865829e-01
2.93913037e-01 -6.28628731e-01 1.29678643e+00 1.87274471e-01
-9.50953126e-01 -7.42449760e-01 -1.15119159e+00 -4.47260946e-01
-5.37757397e-01 1.97771192e-01 4.54807997e-01 2.26036966e-01
-4.65807378e-01 8.40058625e-01 -6.52761281e-01 -8.96129664e-03
6.07680082e-01 -9.78881121e-02 -2.09866390e-01 -9.58177671e-02
-1.09059918e+00 6.91247284e-01 3.79633307e-01 3.74908060e-01
-6.32182539e-01 -7.41566122e-01 -8.58621418e-01 3.24333698e-01
7.52770245e-01 -2.35339060e-01 7.13321924e-01 -1.23728299e+00
-1.08203185e+00 6.95690453e-01 -7.08430037e-02 1.21851921e-01
6.49304867e-01 -2.05912217e-01 -2.55514055e-01 1.10261798e-01
1.15464054e-01 3.10796738e-01 1.06730592e+00 -1.53120005e+00
-4.58485663e-01 -3.92675161e-01 2.82639921e-01 -4.84490283e-02
-2.88834661e-01 1.33946225e-01 -9.16897357e-01 -9.65101540e-01
-4.82876264e-02 -5.34026921e-01 4.31963466e-02 4.10601556e-01
-3.63147885e-01 3.33468094e-02 1.11213720e+00 -8.18806231e-01
1.51601577e+00 -2.52151322e+00 1.47049263e-01 1.64072618e-01
2.79714435e-01 3.69649291e-01 -4.56659436e-01 -3.05168685e-02
-3.93389851e-01 2.36183941e-01 -1.76639438e-01 -1.85969844e-02
-3.77710462e-01 -3.09534464e-02 -9.10266265e-02 3.15210223e-01
4.82673019e-01 9.22064900e-01 -1.18035948e+00 -2.96506524e-01
1.71237424e-01 1.26638189e-01 -3.12459648e-01 8.61191005e-02
-1.51257217e-01 2.94287980e-01 -5.29954970e-01 6.60947144e-01
1.02735102e+00 -1.28627330e-01 -3.60059619e-01 -7.13528097e-01
-1.40449628e-01 1.50747104e-02 -1.03160357e+00 1.85108221e+00
-7.18927011e-02 4.98467267e-01 2.07169428e-01 -7.14132369e-01
6.96828246e-01 -1.20152190e-01 1.20696984e-01 -7.50744581e-01
3.64775121e-01 -3.00288856e-01 2.64136672e-01 -1.02330267e+00
2.70974874e-01 4.79315698e-01 2.97650814e-01 1.79364026e-01
-3.93048346e-01 2.37093031e-01 1.76482648e-01 1.56136334e-01
9.72449780e-01 2.42394045e-01 3.30548249e-02 -1.61186501e-01
4.92629081e-01 -2.78902948e-01 8.05924237e-01 7.93452680e-01
-6.27976954e-01 6.01602674e-01 6.81860149e-01 -2.85081387e-01
-8.46031487e-01 -6.97437823e-01 6.58964068e-02 9.38271523e-01
9.34707403e-01 -4.26817209e-01 -9.45963264e-01 -8.44645739e-01
2.67988071e-03 5.20074069e-01 -8.43520641e-01 -5.44111550e-01
-7.60449231e-01 -3.30165356e-01 1.99448138e-01 8.04580152e-01
9.93384361e-01 -1.23831642e+00 -6.15422487e-01 -1.11096472e-01
-3.19710314e-01 -9.42771316e-01 -9.60804999e-01 -1.29999239e-02
-6.25298738e-01 -1.29238677e+00 -4.69909310e-01 -9.11661446e-01
9.12871063e-01 8.15454781e-01 5.86307108e-01 5.98801672e-01
-5.29646277e-01 -2.29839981e-01 -2.91269451e-01 -2.91888833e-01
9.99120325e-02 -3.91868979e-01 -4.69156981e-01 1.12614378e-01
4.41881448e-01 -4.23443586e-01 -7.56437004e-01 3.32997173e-01
-1.12653291e+00 3.87468636e-02 6.56170666e-01 8.27906549e-01
1.97446942e-01 5.08920550e-01 3.29110324e-02 -6.40371025e-01
5.48067808e-01 -3.89810979e-01 -4.38592613e-01 2.16733381e-01
-1.97121024e-01 1.14623733e-01 5.17447531e-01 -9.90962446e-01
-1.07944727e+00 -1.35532692e-01 4.10385162e-01 -7.84053206e-01
-1.11635171e-01 2.92130202e-01 -6.21981323e-01 -1.41384706e-01
4.10820156e-01 1.21605344e-01 -9.28656533e-02 -5.74610591e-01
2.64355570e-01 5.11071742e-01 5.31018019e-01 -2.34445632e-01
8.60691488e-01 4.92819965e-01 -2.03722104e-01 -6.76832855e-01
-7.57434249e-01 -3.01262110e-01 -4.73005801e-01 -2.64031351e-01
9.65115130e-01 -5.37679732e-01 -6.07082903e-01 7.36620128e-01
-1.28710639e+00 -3.12914819e-01 -2.49470063e-02 2.17012838e-01
-5.68902604e-02 6.28664911e-01 -6.34767115e-01 -6.79154456e-01
-6.12895600e-02 -1.29569840e+00 1.07161546e+00 4.32112843e-01
9.38659608e-02 -4.45609897e-01 -7.34091640e-01 1.90565497e-01
3.39589179e-01 1.74640104e-01 1.24983251e+00 -8.06523412e-02
-8.88649166e-01 -9.54871476e-02 -8.41613710e-01 4.89184707e-01
3.33228976e-01 -3.57839689e-02 -8.25977743e-01 1.22806042e-01
3.96123767e-01 1.44347787e-01 1.08082521e+00 1.78402543e-01
1.80086100e+00 -5.52873015e-02 -4.14101124e-01 6.97007179e-01
1.35863280e+00 3.09779555e-01 1.04775333e+00 4.36396450e-01
9.06203270e-01 5.33397377e-01 4.94090647e-01 1.13898017e-01
4.13530134e-02 3.11195642e-01 6.83822632e-01 -1.64839789e-01
-4.38046008e-01 -2.67851084e-01 2.55263299e-01 3.53244662e-01
-1.27953485e-01 -3.73814441e-02 -6.48263097e-01 2.86024243e-01
-1.82710946e+00 -1.01960278e+00 -4.25016314e-01 2.15893960e+00
7.31404960e-01 3.50829840e-01 -4.18047726e-01 1.83485344e-01
1.06310892e+00 2.63859034e-01 -6.64628327e-01 2.21204400e-01
-5.98192513e-02 -1.72849774e-01 4.58456695e-01 1.39619976e-01
-1.06709301e+00 8.71530950e-01 5.61530924e+00 9.40360844e-01
-1.02521348e+00 -5.43715023e-02 4.38253820e-01 5.57107888e-02
7.91927353e-02 4.13716175e-02 -4.78909791e-01 9.94527698e-01
-8.83636251e-02 2.05725223e-01 6.83247030e-01 7.33348131e-01
4.54712540e-01 -4.26656365e-01 -1.03144979e+00 1.04174137e+00
2.34193429e-01 -8.35536718e-01 6.63128868e-02 -1.94525331e-01
4.84337777e-01 -4.76473451e-01 6.22376278e-02 9.91068222e-03
-2.09162727e-01 -6.57913446e-01 8.57767284e-01 6.91675305e-01
5.11061907e-01 -5.84503591e-01 5.63444555e-01 4.53962833e-01
-1.14210224e+00 -3.29136193e-01 -3.75717640e-01 -1.22103810e-01
-4.37294319e-02 7.31752515e-01 8.57377574e-02 2.50296235e-01
8.46484423e-01 8.76600444e-01 -7.01847672e-01 1.05767524e+00
-5.16767442e-01 2.64625609e-01 -7.18750879e-02 2.45312810e-01
3.17476057e-02 -6.03971183e-01 5.31866550e-01 1.01358581e+00
-5.55391982e-02 1.00445651e-01 1.08105063e-01 1.29275250e+00
-3.37509423e-01 -2.90133983e-01 -5.01558363e-01 -1.27969473e-03
4.07780468e-01 1.27928889e+00 -6.80217683e-01 -3.94791752e-01
-6.29274845e-01 1.24505293e+00 4.11754191e-01 6.03400886e-01
-1.08665001e+00 -7.88089693e-01 6.65717006e-01 2.02822283e-01
4.65502709e-01 -1.28594279e-01 -4.48600352e-01 -1.26483309e+00
4.54276800e-01 -8.46341014e-01 -4.64725494e-02 -1.10654330e+00
-1.31589150e+00 2.26499096e-01 -2.14737758e-01 -1.11255229e+00
6.38573468e-01 -5.78354955e-01 -8.61265123e-01 1.02461588e+00
-1.61311853e+00 -8.15529644e-01 -9.70841408e-01 5.85261881e-01
5.03564835e-01 4.62674290e-01 2.77190357e-01 4.45987552e-01
-1.01885498e+00 5.01310229e-01 -3.63999277e-01 6.08009934e-01
6.80981100e-01 -9.09193456e-01 1.22328684e-01 1.16399527e+00
-4.87292320e-01 8.79484653e-01 4.56610590e-01 -9.72364902e-01
-1.19627714e+00 -1.12453902e+00 3.26966882e-01 -3.50287676e-01
7.60672092e-01 -4.20266718e-01 -1.33444929e+00 6.41840041e-01
1.08003743e-01 2.18249753e-01 5.40797831e-04 -2.39911482e-01
-6.60224497e-01 -1.85722977e-01 -1.00002754e+00 7.58592308e-01
1.47696233e+00 -6.14906251e-01 -7.27549732e-01 9.65055749e-02
6.73297822e-01 -1.70438811e-01 -4.18557912e-01 4.74257410e-01
3.63258570e-01 -9.16166961e-01 9.79951978e-01 -3.00275058e-01
7.91275799e-01 -5.68005800e-01 3.20451975e-01 -1.16994429e+00
-6.60224319e-01 -3.80432516e-01 5.43586072e-03 1.22673368e+00
-4.37132418e-02 -2.17052206e-01 5.17050982e-01 4.73004162e-01
3.64186950e-02 -6.23895824e-01 -2.46707141e-01 -8.20637286e-01
-3.65950108e-01 -2.82148868e-01 6.41429126e-01 9.53730226e-01
-1.45853922e-01 1.79272607e-01 -1.88409403e-01 3.80368650e-01
5.75000763e-01 1.22740239e-01 5.28053105e-01 -9.21227753e-01
-1.68022394e-01 -7.63823032e-01 -5.06175160e-01 -1.31593788e+00
-5.70831560e-02 -6.29316032e-01 2.12745711e-01 -1.11207771e+00
5.02214491e-01 -1.62028924e-01 -3.68916273e-01 4.11282837e-01
-8.63873243e-01 8.74293596e-02 2.18498677e-01 1.68343335e-01
-3.70282143e-01 4.01643544e-01 1.51052487e+00 -2.52148300e-01
-1.60422996e-01 -4.33707952e-01 -7.43004680e-01 1.03204751e+00
6.39407277e-01 -3.89268070e-01 -2.80773818e-01 -7.89161980e-01
-1.52628899e-01 -2.64959544e-01 4.21557009e-01 -8.71979833e-01
2.37987518e-01 -1.97257370e-01 6.27925456e-01 -5.24837971e-01
-5.32124527e-02 -1.08183718e+00 -1.27517074e-01 4.84300196e-01
-4.07594711e-01 -1.72824740e-01 7.41443112e-02 7.73046076e-01
-6.24546520e-02 -4.24583733e-01 9.50298548e-01 -3.91961664e-01
-7.91117609e-01 3.50471765e-01 9.26227681e-03 -2.41290808e-01
1.14609134e+00 -2.59568393e-01 -5.40573061e-01 -8.70597437e-02
-4.45387691e-01 3.79750341e-01 5.85020542e-01 6.74354851e-01
6.10997558e-01 -1.21091163e+00 -1.95282504e-01 4.43680614e-01
2.77248234e-01 1.47939727e-01 2.46596664e-01 8.32454264e-01
-3.33485067e-01 -1.94436401e-01 -2.08948717e-01 -3.11869532e-01
-1.17654133e+00 1.09671664e+00 3.87595594e-01 1.91472888e-01
-8.78978670e-01 7.41889358e-01 5.38206398e-01 3.20803106e-01
5.60796797e-01 -6.91522241e-01 -3.41796130e-02 -8.95674154e-02
9.11643505e-01 4.89150286e-01 -3.87045443e-01 -3.55344981e-01
-1.52778909e-01 5.78526258e-01 -1.71012983e-01 3.51406395e-01
1.02738714e+00 -2.45624080e-01 -4.57340151e-01 3.11288089e-01
1.14235592e+00 -1.09913185e-01 -1.58733153e+00 -3.60386282e-01
3.27290744e-02 -8.10845375e-01 3.82400095e-01 -8.40753853e-01
-1.48689675e+00 8.78470778e-01 4.78289783e-01 5.01806401e-02
1.40559173e+00 -2.36449033e-01 7.20006049e-01 1.23865612e-01
1.87834486e-01 -9.63134170e-01 3.43561590e-01 3.38187814e-01
1.01643789e+00 -1.34652543e+00 -1.28753930e-01 -8.37899983e-01
-3.96198064e-01 1.03607225e+00 1.00107515e+00 -4.08563972e-01
4.88915503e-01 4.61194247e-01 -1.53570384e-01 -1.92768902e-01
-6.51891679e-02 -2.80245423e-01 4.73834313e-02 5.32313883e-01
-7.44245648e-02 -3.54186743e-01 -2.98281640e-01 7.39841878e-01
2.99175352e-01 2.09250763e-01 3.79904687e-01 9.58427191e-01
-4.97771978e-01 -5.92478335e-01 -2.81455219e-01 3.29104513e-01
-2.04217479e-01 -3.00318688e-01 -3.57812911e-01 5.66122651e-01
7.19441533e-01 8.90869975e-01 2.19433412e-01 -2.93077558e-01
4.74834204e-01 -1.37932733e-01 4.02619183e-01 -3.89151365e-01
-7.11900353e-01 3.94043811e-02 -4.23950940e-01 -1.03186262e+00
-2.52922982e-01 -4.08700287e-01 -1.23541260e+00 -6.09069057e-02
-6.99571550e-01 -3.89248908e-01 6.69290364e-01 6.36168182e-01
4.11928684e-01 8.27010274e-01 5.40829897e-01 -8.28816950e-01
-5.23676336e-01 -1.13232493e+00 -7.46715128e-01 1.04187655e+00
3.83367956e-01 -6.71950758e-01 -4.14975733e-01 1.52950764e-01]
|
[12.140385627746582, 0.8312212228775024]
|
64ae7d90-5662-446e-9fc4-cafee1f0bf43
|
chatgpt-is-a-knowledgeable-but-inexperienced
|
2303.16421
| null |
https://arxiv.org/abs/2303.16421v1
|
https://arxiv.org/pdf/2303.16421v1.pdf
|
ChatGPT is a Knowledgeable but Inexperienced Solver: An Investigation of Commonsense Problem in Large Language Models
|
Large language models (LLMs) such as ChatGPT and GPT-4 have made significant progress in NLP. However, their ability to memorize, represent, and leverage commonsense knowledge has been a well-known pain point for LLMs. It remains unclear that: (1) Can GPTs effectively answer commonsense questions? (2) Are GPTs knowledgeable in commonsense? (3) Are GPTs aware of the underlying commonsense knowledge for answering a specific question? (4) Can GPTs effectively leverage commonsense for answering questions? To evaluate the above commonsense problems, we conduct a series of experiments to evaluate ChatGPT's commonsense abilities, and the experimental results show that: (1) GPTs can achieve good QA accuracy in commonsense tasks, while they still struggle with certain types of knowledge. (2) ChatGPT is knowledgeable, and can accurately generate most of the commonsense knowledge using knowledge prompts. (3) Despite its knowledge, ChatGPT is an inexperienced commonsense problem solver, which cannot precisely identify the needed commonsense knowledge for answering a specific question, i.e., ChatGPT does not precisely know what commonsense knowledge is required to answer a question. The above findings raise the need to investigate better mechanisms for utilizing commonsense knowledge in LLMs, such as instruction following, better commonsense guidance, etc.
|
['Ben He', 'Yaojie Lu', 'Hongyu Lin', 'Le Sun', 'Xianpei Han', 'Ning Bian']
|
2023-03-29
| null | null | null | null |
['instruction-following']
|
['natural-language-processing']
|
[ 1.66554078e-01 4.01316464e-01 3.33290137e-02 1.27829671e-01
-6.11799479e-01 -8.23773146e-01 3.03069085e-01 2.39122331e-01
-8.34807158e-02 9.38401639e-01 3.66711617e-01 -8.89566541e-01
-1.19073495e-01 -1.24422932e+00 -7.61947989e-01 -4.98413518e-02
5.50066710e-01 4.86194462e-01 5.61578035e-01 -8.87966394e-01
5.31404555e-01 1.14670485e-01 -1.28658104e+00 3.11086595e-01
1.90953529e+00 5.90065718e-01 4.72533435e-01 6.27931654e-01
-6.52395725e-01 1.64001250e+00 -9.77533996e-01 -6.52927816e-01
-2.48328850e-01 -7.65974283e-01 -1.67449617e+00 -5.38951159e-01
3.11429650e-01 -3.32807869e-01 -3.85131419e-01 1.34144294e+00
8.50432292e-02 3.71836871e-01 4.54115868e-01 -1.22390699e+00
-1.29643440e+00 1.07952392e+00 2.66037405e-01 5.16499758e-01
7.55252182e-01 7.39784479e-01 1.03993523e+00 -2.76177730e-02
5.06801486e-01 1.47767210e+00 4.68364388e-01 7.60836065e-01
-9.02565956e-01 -7.91149855e-01 -1.20250173e-01 6.74234986e-01
-1.14204097e+00 -1.91390246e-01 6.59817636e-01 -2.06456050e-01
1.37680686e+00 4.10064667e-01 8.20308924e-01 9.86514807e-01
2.51525372e-01 9.11661863e-01 1.59283280e+00 -5.72697222e-01
1.21180840e-01 1.04599334e-01 4.28095341e-01 8.12093735e-01
3.19960386e-01 5.89000061e-02 -6.86712623e-01 -8.27623904e-02
8.27237725e-01 -4.47593480e-01 -3.51765424e-01 5.02740085e-01
-1.22514153e+00 1.00285649e+00 4.99845505e-01 6.56236589e-01
-1.45369530e-01 3.15578669e-01 4.60509241e-01 6.57254457e-01
-1.94404975e-01 1.20423710e+00 -3.65591228e-01 -7.83577979e-01
-5.63483238e-01 4.12210971e-01 1.05701685e+00 1.01530683e+00
6.36315763e-01 3.20523679e-01 -1.04637422e-01 5.49577832e-01
3.76804061e-02 7.88043499e-01 7.71777272e-01 -1.21550500e+00
6.46803558e-01 7.81104207e-01 1.03164613e-02 -1.11757886e+00
-7.34581277e-02 -2.75499672e-02 -2.49214515e-01 -3.47588897e-01
6.71511650e-01 -1.58960849e-01 -5.45602262e-01 2.04240894e+00
8.94360468e-02 1.62366673e-01 2.17210650e-01 6.31498814e-01
1.09439743e+00 6.77950740e-01 2.38719255e-01 9.82575566e-02
1.41196418e+00 -4.99237388e-01 -8.20869446e-01 -6.98098242e-01
1.02746010e+00 -4.49888647e-01 1.65849841e+00 2.58142836e-02
-1.12853754e+00 -2.13966846e-01 -8.71723533e-01 -5.72904050e-01
-6.22740626e-01 -5.61915994e-01 8.77859950e-01 5.42957008e-01
-9.14826512e-01 5.75044215e-01 -4.83235508e-01 -3.21221083e-01
3.47924620e-01 -1.59030020e-01 2.53369994e-02 -4.47485715e-01
-1.99636841e+00 1.58453834e+00 6.58386528e-01 -1.64907560e-01
-8.72431636e-01 -6.55917406e-01 -1.12939143e+00 2.63343662e-01
6.70982480e-01 -9.26339030e-01 1.38574278e+00 -8.40974212e-01
-1.52161634e+00 8.99438620e-01 -3.37104470e-01 -3.28187734e-01
7.30031580e-02 -2.87563860e-01 -2.96143830e-01 2.22233295e-01
3.14648062e-01 3.79535645e-01 3.41645539e-01 -1.03889894e+00
-3.16657275e-01 -1.16412982e-01 6.46074593e-01 2.02829599e-01
-2.13722140e-03 1.74302422e-02 3.23510081e-01 -4.16426778e-01
4.97482531e-03 -7.15099871e-01 3.56025100e-01 -3.89339238e-01
-6.52866483e-01 -4.56872672e-01 4.96157646e-01 -8.29176068e-01
1.29646218e+00 -1.79780328e+00 -1.53397536e-02 -6.39600679e-02
3.29427660e-01 4.52117562e-01 -8.72022659e-02 5.01531005e-01
1.99035779e-01 5.16904235e-01 -2.37076413e-02 4.65581357e-01
3.07383180e-01 5.18448710e-01 -7.03868568e-01 -3.94362301e-01
1.35737687e-01 1.42287683e+00 -1.44282198e+00 -7.08840728e-01
1.51627064e-01 -1.49345607e-01 -4.88988280e-01 2.55144328e-01
-5.72202384e-01 -8.08396116e-02 -4.92703646e-01 5.26198387e-01
2.45241582e-01 -5.37393630e-01 3.85305774e-03 1.12588018e-01
2.76510090e-01 9.86282349e-01 -8.56810808e-01 1.14084005e+00
-5.58631659e-01 8.97609949e-01 -3.66870761e-01 -6.73125148e-01
5.35052896e-01 1.43633530e-01 -5.63865006e-01 -5.41119039e-01
2.31947899e-01 2.43104085e-01 3.87573659e-01 -8.78384411e-01
5.38551092e-01 -9.17257011e-01 -8.65715295e-02 7.29776144e-01
-8.35126266e-02 -8.72625589e-01 1.66940004e-01 5.62720716e-01
1.21314251e+00 -3.66371632e-01 4.07725632e-01 -1.81636408e-01
3.34389061e-01 6.98522985e-01 2.87493527e-01 8.87836039e-01
-4.61857051e-01 -7.47359693e-02 6.12712324e-01 5.17812669e-02
-3.19275975e-01 -1.05940866e+00 3.44586968e-01 1.24864519e+00
3.34954202e-01 -1.37925327e-01 -7.95256853e-01 -4.92189288e-01
-2.09864795e-01 1.83526278e+00 -3.39564204e-01 -5.51982462e-01
-4.89390254e-01 5.73231652e-03 1.05425990e+00 5.60795486e-01
8.95302892e-01 -1.52984810e+00 -7.90994406e-01 1.47363082e-01
-1.02802801e+00 -1.23776650e+00 -3.91558319e-01 1.44437373e-01
-9.97105479e-01 -1.42287207e+00 3.96207981e-02 -7.35579491e-01
3.67462695e-01 4.98477519e-01 1.34399068e+00 6.41348720e-01
1.65458873e-01 4.55621988e-01 -4.22612041e-01 -4.89315629e-01
-8.42256486e-01 -8.89374912e-02 -4.10475641e-01 -1.06558907e+00
8.21427763e-01 -6.11152947e-01 -3.01870182e-02 1.09088868e-01
-8.55892897e-01 -5.68056554e-02 2.71685928e-01 7.80407667e-01
-2.25512777e-02 4.27335829e-01 6.68527246e-01 -8.40857506e-01
1.38672650e+00 -5.88841140e-01 -2.66542822e-01 4.43787426e-01
-2.51769602e-01 1.53136894e-01 6.22501671e-01 -4.97906744e-01
-1.12564600e+00 -9.34736371e-01 -8.81605148e-02 -1.93705618e-01
-1.38227537e-01 8.29835534e-01 -1.04068913e-01 4.13523242e-02
9.91349876e-01 6.63338423e-01 5.13502471e-02 9.24410596e-02
4.88254070e-01 4.24730420e-01 4.88202214e-01 -1.35070169e+00
9.18433189e-01 -6.66262582e-02 -4.83460218e-01 -6.59289300e-01
-1.29944336e+00 -2.55301893e-01 -3.93111072e-03 1.83251575e-01
8.95794988e-01 -6.98899031e-01 -8.45398903e-01 3.26186389e-01
-1.36101842e+00 -8.21267366e-01 -3.91564757e-01 2.48042107e-01
-7.08672523e-01 5.29232144e-01 -9.97880101e-01 -6.16047025e-01
-3.08939725e-01 -9.88172114e-01 4.38099533e-01 4.54737127e-01
-7.98070192e-01 -1.28885496e+00 -1.53226346e-01 8.50638092e-01
6.37998939e-01 -7.17763603e-02 1.74965429e+00 -8.02488983e-01
-5.73980451e-01 1.09033093e-01 -3.06273997e-01 6.03819728e-01
1.45578757e-01 -1.17818326e-01 -6.73624873e-01 5.41624784e-01
2.64459044e-01 -9.59680319e-01 3.42198730e-01 6.29843846e-02
9.69050288e-01 -5.69586635e-01 2.11928785e-02 6.93689138e-02
1.05860078e+00 3.99070159e-02 9.19977427e-01 4.12852705e-01
3.88053417e-01 4.06427592e-01 4.47983742e-01 -8.56677741e-02
9.53988075e-01 1.12577043e-01 2.48445407e-01 5.95303833e-01
-1.71404645e-01 -6.98054194e-01 5.20632088e-01 8.28131318e-01
-4.11446951e-02 1.11989388e-02 -1.14438725e+00 7.74298072e-01
-1.69232643e+00 -1.23945749e+00 -2.67169654e-01 1.53533757e+00
1.69028711e+00 1.63961686e-02 -3.57712120e-01 1.02693453e-01
5.94567955e-01 1.81952536e-01 -6.60276353e-01 -7.96517491e-01
-1.23778000e-01 3.32578331e-01 -1.28084019e-01 5.90082884e-01
-2.94015765e-01 1.42615438e+00 6.33436728e+00 1.05702233e+00
-8.02061319e-01 9.16514546e-02 -1.50207980e-02 1.87888399e-01
-5.90514958e-01 1.40969813e-01 -5.26058197e-01 4.53260213e-01
7.44915366e-01 -6.51965678e-01 7.85051048e-01 7.77306318e-01
-1.74935773e-01 -2.70780027e-01 -9.78371382e-01 7.27203250e-01
7.93119445e-02 -1.38049650e+00 3.71867836e-01 -2.84656912e-01
7.61520386e-01 -3.11222643e-01 -1.77406445e-01 1.10210502e+00
9.58601832e-01 -1.27809119e+00 6.17727041e-01 2.47677326e-01
2.84559995e-01 -6.37638688e-01 7.01743364e-01 1.00365710e+00
-5.16838431e-01 6.40647113e-02 -4.91481900e-01 -5.34210384e-01
3.38961780e-02 4.59931582e-01 -8.17679703e-01 1.65210381e-01
1.82294562e-01 8.54390487e-02 -5.41286528e-01 6.85836792e-01
-1.19723976e+00 8.86013687e-01 -2.62817115e-01 -4.31634426e-01
2.99540579e-01 1.85387909e-01 4.54516083e-01 9.05367553e-01
2.89413452e-01 7.31500089e-01 -2.17691734e-02 1.45545626e+00
-2.24320099e-01 -3.05509448e-01 -3.71960402e-01 -6.12729669e-01
8.99644136e-01 6.58642769e-01 -3.53243768e-01 -5.86732507e-01
1.93762593e-02 7.52384245e-01 5.45264840e-01 3.23428273e-01
-7.25383759e-01 -5.25430024e-01 6.16082489e-01 4.52214293e-02
-1.88753501e-01 -7.29207844e-02 -5.37347376e-01 -1.26233852e+00
-1.22627653e-01 -1.13265038e+00 3.54409009e-01 -1.50373590e+00
-1.41912341e+00 1.41110808e-01 2.12781638e-01 -4.17152077e-01
-3.94589573e-01 -6.42473221e-01 -9.90301251e-01 8.96814346e-01
-1.63117218e+00 -8.82798433e-01 -2.85160810e-01 8.01899791e-01
4.06777501e-01 2.53147960e-01 8.07751238e-01 -4.93506223e-01
-1.67956337e-01 4.44815308e-01 -4.47700083e-01 3.18402886e-01
3.18445206e-01 -1.37983644e+00 1.15865976e-01 6.02936745e-01
-2.07786188e-01 1.32222795e+00 9.31385040e-01 -8.79915476e-01
-1.50495279e+00 -8.53388369e-01 1.16783202e+00 -8.35586131e-01
1.01307654e+00 3.07817400e-01 -1.32419002e+00 1.04004371e+00
2.46496886e-01 -4.39138204e-01 9.54317629e-01 4.00735736e-01
-6.88253939e-01 5.89096367e-01 -1.29372489e+00 7.80120313e-01
1.06637335e+00 -9.98665929e-01 -1.76094842e+00 5.79969466e-01
1.13315916e+00 -6.69477940e-01 -6.21988714e-01 -2.83743620e-01
8.52379501e-02 -7.93755651e-01 6.98007703e-01 -8.93113256e-01
1.03132820e+00 -2.54867822e-01 -1.45545810e-01 -1.72570324e+00
-1.20261222e-01 -4.97128904e-01 -3.17702085e-01 1.10675311e+00
2.43616402e-01 -8.56664777e-01 2.68723786e-01 1.06875396e+00
-2.01563776e-01 -6.18915260e-01 -9.35491979e-01 -1.13933611e+00
5.66254079e-01 -8.01471293e-01 7.85422027e-01 1.34408808e+00
6.56795979e-01 6.15005732e-01 1.15656532e-01 -1.41162528e-02
2.36107320e-01 3.05894136e-01 5.59253573e-01 -9.38124120e-01
-3.17815572e-01 -5.65160632e-01 -5.90796722e-03 -1.02723622e+00
6.61147356e-01 -1.10118508e+00 1.66171193e-01 -1.84226465e+00
3.35165620e-01 -3.78726423e-01 1.43946692e-01 7.63187170e-01
-6.97120547e-01 -4.22173172e-01 3.70611876e-01 -1.20873347e-01
-5.44811845e-01 2.99100131e-01 1.75804663e+00 -1.06094569e-01
-2.18699537e-02 -4.90218490e-01 -1.35169840e+00 7.65602529e-01
7.91722536e-01 -2.33839244e-01 -6.51014984e-01 -5.12215316e-01
6.08897626e-01 1.87340885e-01 6.29706979e-01 -5.89354157e-01
3.33379954e-01 -9.63836670e-01 -1.16102614e-01 -1.20199956e-01
-6.49434282e-03 -4.94195342e-01 -3.21440011e-01 4.75950986e-01
-2.61660457e-01 -1.67232648e-01 6.33299530e-01 8.95349830e-02
-2.21859112e-01 -7.13033557e-01 5.24895012e-01 -5.61187804e-01
-9.87373769e-01 -6.05040848e-01 -4.63003755e-01 1.01292241e+00
7.53769577e-01 -3.85839939e-01 -8.60483408e-01 -5.55581748e-01
-2.62537271e-01 2.94985086e-01 5.69393694e-01 1.33568347e-01
6.44381881e-01 -1.07719374e+00 -4.61503774e-01 -4.05503809e-01
8.94158781e-02 1.27540722e-01 2.85441279e-01 5.00863373e-01
-5.43277800e-01 6.10975921e-01 -6.35209456e-02 -6.66075572e-02
-9.79084849e-01 4.53066558e-01 4.80947345e-01 -4.11408484e-01
-3.30519676e-01 1.10253847e+00 6.54487386e-02 -4.98266339e-01
-3.27594340e-01 -7.11678684e-01 7.83263743e-02 -2.65360653e-01
7.33207405e-01 3.95809829e-01 -2.54634321e-01 -2.61734217e-01
-3.35778773e-01 2.16196328e-01 8.79421923e-03 1.95896178e-01
9.28394616e-01 4.32764217e-02 -2.95286685e-01 4.18704242e-01
7.08124399e-01 -7.79372901e-02 -3.87380481e-01 -1.91723153e-01
-2.48152837e-01 -3.41326237e-01 -1.10771924e-01 -1.34606564e+00
-4.48863238e-01 1.14043128e+00 -4.44185048e-01 1.82300419e-01
7.68501937e-01 1.62734374e-01 1.29738259e+00 7.91298151e-01
6.72017574e-01 -1.07103479e+00 2.59329498e-01 1.22433174e+00
9.26326215e-01 -1.11914313e+00 -2.52535462e-01 -5.42812049e-01
-7.97410727e-01 9.62913096e-01 9.09247398e-01 1.85652673e-01
2.60044578e-02 -1.16070181e-01 -5.93849234e-02 -5.90416789e-01
-8.44229519e-01 -3.65993917e-01 1.53165264e-02 7.47184038e-01
3.68755937e-01 2.29782611e-01 8.50379001e-03 8.16758871e-01
-1.02784836e+00 2.38502428e-01 6.11165285e-01 1.01770055e+00
-8.72854531e-01 -4.14593518e-01 -6.35255635e-01 5.84399521e-01
-7.30599882e-03 -5.46828687e-01 -7.71621943e-01 7.83488989e-01
-6.98121712e-02 1.40201569e+00 -5.26308775e-01 -2.50148416e-01
4.27656263e-01 4.68736857e-01 7.18088806e-01 -1.04879045e+00
-7.93287396e-01 -1.00513148e+00 3.21969867e-01 -4.73211020e-01
9.61362869e-02 -1.73902288e-01 -1.72543621e+00 -8.90191197e-01
-5.61556220e-01 2.85104334e-01 -1.46098305e-02 1.59710455e+00
1.15054026e-01 4.45101142e-01 -2.31793404e-01 1.11387610e-01
-1.02496958e+00 -9.67481315e-01 -3.04291606e-01 4.71068054e-01
2.86949202e-02 -5.96051693e-01 -6.84137225e-01 -5.21236062e-02]
|
[10.117500305175781, 7.961483955383301]
|
1b774326-e171-4866-be28-712b8986cd3b
|
place-recognition-in-gardens-by-learning
|
1906.12151
| null |
https://arxiv.org/abs/1906.12151v1
|
https://arxiv.org/pdf/1906.12151v1.pdf
|
Place recognition in gardens by learning visual representations: data set and benchmark analysis
|
Visual place recognition is an important component of systems for camera localization and loop closure detection. It concerns the recognition of a previously visited place based on visual cues only. Although it is a widely studied problem for indoor and urban environments, the recent use of robots for automation of agricultural and gardening tasks has created new problems, due to the challenging appearance of garden-like environments. Garden scenes predominantly contain green colors, as well as repetitive patterns and textures. The lack of available data recorded in gardens and natural environments makes the improvement of visual localization algorithms difficult. In this paper we propose an extended version of the TB-Places data set, which is designed for testing algorithms for visual place recognition. It contains images with ground truth camera pose recorded in real gardens in different seasons, with varying light conditions. We constructed and released a ground truth for all possible pairs of images, indicating whether they depict the same place or not. We present the results of a benchmark analysis of methods based on convolutional neural networks for holistic image description and place recognition. We train existing networks (i.e. ResNet, DenseNet and VGG NetVLAD) as backbone of a two-way architecture with a contrastive loss function. The results that we obtained demonstrate that learning garden-tailored representations contribute to an improvement of performance, although the generalization capabilities are limited.
|
['Maria Leyva-Vallina', 'Nicolai Petkov', 'Nicola Strisciuglio']
|
2019-06-28
| null | null | null | null |
['camera-localization', 'loop-closure-detection']
|
['computer-vision', 'computer-vision']
|
[ 1.74616441e-01 -3.92658025e-01 3.28604728e-02 -3.25460404e-01
-2.73453048e-03 -7.66382992e-01 6.85744762e-01 5.43911159e-01
-6.12815320e-01 6.66278481e-01 -1.63631245e-01 -1.58680052e-01
-5.95968887e-02 -1.02917063e+00 -9.71516907e-01 -7.93532073e-01
-3.77541333e-01 1.49971426e-01 2.40737066e-01 -4.91532624e-01
6.87182741e-03 8.43786299e-01 -1.86988151e+00 4.69810553e-02
5.00688851e-01 8.18600118e-01 9.42351341e-01 7.12411046e-01
1.71342447e-01 8.17926884e-01 -3.61046046e-01 2.51472443e-01
3.06757599e-01 -1.90831065e-01 -5.00429988e-01 1.60100982e-01
3.80043089e-01 -1.78125158e-01 -4.17442441e-01 1.04764736e+00
3.50571573e-01 3.75479311e-01 5.43130577e-01 -1.36189377e+00
-8.95122707e-01 1.04037501e-01 -1.75252169e-01 7.44765326e-02
5.66310287e-01 2.53290892e-01 7.58917034e-01 -7.42834032e-01
7.68972695e-01 8.93652797e-01 7.84259319e-01 5.65835759e-02
-1.31550968e+00 -2.37732559e-01 3.71467203e-01 5.73823452e-01
-1.72001755e+00 -3.54722589e-01 4.92304265e-01 -2.36426726e-01
1.03611839e+00 9.34613645e-02 9.02197957e-01 1.24977636e+00
2.02066600e-01 6.79575145e-01 1.24402893e+00 -6.20490491e-01
3.83537531e-01 3.08799744e-02 -2.42036328e-01 7.90720224e-01
2.88515151e-01 1.76562577e-01 -3.97625238e-01 1.35229409e-01
9.15232003e-01 4.25401837e-01 -5.61584473e-01 -1.03865814e+00
-1.56009305e+00 6.64590001e-01 1.32098567e+00 6.12703025e-01
-5.41333914e-01 1.72007531e-01 2.29533296e-02 3.55436057e-01
7.15219006e-02 4.65349495e-01 -2.66405433e-01 2.78437078e-01
-7.69864142e-01 3.94532770e-01 7.82524407e-01 1.25418437e+00
9.86700892e-01 -1.52067170e-01 5.27467057e-02 6.70358479e-01
3.25492173e-01 7.75726140e-01 5.07019579e-01 -5.48848927e-01
1.18170135e-01 5.47834933e-01 3.30671310e-01 -1.50335431e+00
-7.48312175e-01 -2.16519594e-01 -9.33650255e-01 3.07529926e-01
5.69719136e-01 3.21670145e-01 -1.22719812e+00 1.70034814e+00
3.63619588e-02 -6.21222379e-03 1.50977328e-01 1.03109205e+00
9.26554084e-01 6.29912972e-01 1.03663914e-02 4.11569148e-01
1.24532413e+00 -9.52934206e-01 -6.51395500e-01 -5.01403153e-01
4.60449815e-01 -6.69782460e-01 1.00461674e+00 2.06579670e-01
-3.06826174e-01 -5.18373489e-01 -1.20354724e+00 -7.73452073e-02
-1.11160254e+00 3.57539237e-01 5.88443041e-01 4.01323855e-01
-1.40261519e+00 6.02711797e-01 -6.39979959e-01 -1.16875184e+00
1.69896021e-01 1.17967278e-01 -9.35034394e-01 -2.91128844e-01
-8.36459935e-01 9.66167331e-01 5.00428498e-01 5.02921402e-01
-1.09054780e+00 -3.29265743e-02 -1.23151183e+00 6.94713145e-02
1.06663577e-01 -2.00807512e-01 9.41425085e-01 -9.74336982e-01
-1.16706645e+00 1.10916889e+00 3.78864221e-02 -5.08412302e-01
5.10877967e-01 2.25113239e-02 -3.92533302e-01 -8.67748111e-02
4.09266770e-01 8.09769213e-01 4.67507333e-01 -1.38712704e+00
-6.97343588e-01 -3.59295756e-01 2.38657802e-01 1.87000200e-01
2.28689492e-01 -4.64482307e-01 -9.21409428e-02 -4.38956618e-01
5.98202944e-01 -1.10935664e+00 -4.74057764e-01 3.20074409e-01
-4.17979836e-01 2.85452276e-01 6.30681157e-01 -4.79086995e-01
4.84842598e-01 -2.12528992e+00 -8.46790224e-02 3.31059903e-01
-5.26453145e-02 6.65818155e-02 -2.51938194e-01 6.58437312e-01
-4.93401289e-02 -1.25162959e-01 -3.22521627e-01 -1.48759663e-01
-4.65767384e-02 4.89292622e-01 -2.28726909e-01 8.88461411e-01
-2.72749439e-02 6.21367157e-01 -1.11974120e+00 -1.53369874e-01
5.41675389e-01 3.70039701e-01 -1.21143796e-01 2.02897102e-01
-1.42570406e-01 4.23877627e-01 -8.63899514e-02 8.64295423e-01
7.22692251e-01 -1.51639909e-01 1.89967826e-01 1.25920847e-01
-3.59194726e-01 -1.12807356e-01 -1.39931750e+00 2.09515095e+00
-5.52501023e-01 9.64224577e-01 -1.11305170e-01 -6.43089235e-01
1.10194194e+00 5.08224452e-03 4.51771207e-02 -9.14293408e-01
3.37124355e-02 3.23077708e-01 -3.52117807e-01 -3.50262552e-01
7.87872791e-01 4.01154310e-01 -3.07784602e-02 -1.49218842e-01
2.06423149e-01 -1.42464444e-01 1.60853744e-01 -3.66472423e-01
1.05063617e+00 3.82139236e-01 5.46331465e-01 -3.53503197e-01
3.47645372e-01 2.93650091e-01 1.84682578e-01 9.68638539e-01
-3.38194758e-01 8.64783168e-01 6.66840002e-02 -8.60932291e-01
-1.08714128e+00 -1.11394572e+00 -2.32579466e-02 1.06884277e+00
6.26321971e-01 -1.97153658e-01 -3.16371948e-01 -4.09373939e-01
-4.91105951e-02 4.71285611e-01 -7.91348636e-01 -4.91221100e-02
-4.78253067e-01 -5.41408181e-01 4.91346806e-01 5.10900557e-01
8.46394420e-01 -1.38526058e+00 -1.16943979e+00 9.93493423e-02
-2.77458310e-01 -1.18071139e+00 1.36033028e-01 8.88672650e-01
-3.91772628e-01 -1.22397435e+00 -8.73844624e-01 -1.04352415e+00
6.28463745e-01 6.75513148e-01 1.09851825e+00 -1.11292571e-01
-3.49650830e-01 2.91001618e-01 -6.40663981e-01 -3.95007491e-01
-1.16381804e-02 3.16873014e-01 -8.67651924e-02 -9.60705206e-02
3.61469984e-01 -5.28081298e-01 -5.72761297e-01 3.90555441e-01
-7.77330697e-01 -1.56443790e-01 4.88109797e-01 9.61702824e-01
7.04206765e-01 -1.86111957e-01 -2.09177390e-01 -2.85449713e-01
1.52587846e-01 -5.36745727e-01 -8.48586679e-01 3.48630607e-01
-2.63156354e-01 1.48247592e-02 5.94403088e-01 -1.47541612e-01
-4.94589120e-01 4.43923503e-01 -2.32074093e-02 -3.65294963e-01
-8.06788266e-01 3.58298331e-01 -1.09734245e-01 -4.67204034e-01
9.05479193e-01 3.90661448e-01 -3.82203281e-01 -2.69345373e-01
4.03189778e-01 2.90408224e-01 5.38080633e-01 -1.32854849e-01
7.76614726e-01 5.54597080e-01 9.92838219e-02 -1.26193702e+00
-3.44458908e-01 -7.30969667e-01 -8.06428194e-01 -1.68974102e-01
1.04543471e+00 -1.00997162e+00 -5.71032286e-01 4.76884812e-01
-1.17385614e+00 -5.10070145e-01 -2.65325129e-01 5.54058969e-01
-7.01539099e-01 8.54211953e-03 -7.55559728e-02 -6.40302896e-01
1.38741627e-01 -1.13511395e+00 1.06301749e+00 2.61024296e-01
9.68241915e-02 -8.57194602e-01 2.76586860e-01 -4.45610374e-01
3.76137406e-01 6.22649252e-01 5.50308883e-01 -3.29319775e-01
-7.91224301e-01 -1.97219610e-01 -2.93342143e-01 1.35393767e-02
2.35815600e-01 -3.11609507e-01 -1.32847440e+00 -3.54945272e-01
-3.60911876e-01 -3.33037138e-01 9.14261580e-01 2.86389828e-01
8.65231216e-01 -2.89336126e-03 -4.63311493e-01 9.18682516e-01
1.87115788e+00 2.42523700e-01 6.59886658e-01 9.41826403e-01
6.50069535e-01 4.77648139e-01 5.37923217e-01 4.51022834e-01
3.71328264e-01 7.37911522e-01 1.09921718e+00 -4.34359640e-01
6.93999380e-02 -4.00659412e-01 2.08077040e-02 -4.73771505e-02
-2.40475714e-01 -4.49740648e-01 -1.04003859e+00 8.82135987e-01
-1.98106873e+00 -9.32176590e-01 3.13175619e-02 2.17842269e+00
8.73024091e-02 -3.89820188e-01 -1.52682126e-01 2.11487517e-01
5.32419860e-01 4.41980153e-01 -3.00182343e-01 -1.49478644e-01
-5.24952829e-01 6.13120049e-02 1.16670418e+00 2.27271780e-01
-1.61701643e+00 1.06990242e+00 6.00761461e+00 3.52418363e-01
-1.28608191e+00 -1.71907514e-01 1.73975721e-01 2.99550593e-01
1.70740440e-01 -1.85782731e-01 -5.24961591e-01 1.80801824e-02
6.04820967e-01 4.81064528e-01 6.37413859e-01 1.15670681e+00
-1.53584499e-02 -4.19113129e-01 -1.03997767e+00 1.20748997e+00
1.64781719e-01 -1.15191054e+00 -2.56231815e-01 -1.18817054e-01
6.66863978e-01 4.64798152e-01 1.28210843e-01 1.80772975e-01
3.84210736e-01 -1.20417118e+00 1.19298458e+00 3.58961195e-01
5.59052587e-01 -5.16897023e-01 8.08394551e-01 2.96184182e-01
-1.40949869e+00 -2.07990006e-01 -6.44343078e-01 -1.88655734e-01
-2.66459554e-01 1.27353311e-01 -1.02599740e+00 6.11246943e-01
1.13627958e+00 1.00738513e+00 -9.28873897e-01 1.40053034e+00
-4.92556661e-01 4.25509773e-02 -5.31561732e-01 -2.41002351e-01
6.97931707e-01 -1.30515099e-01 3.81533027e-01 1.06511557e+00
4.99703079e-01 -3.43949169e-01 3.72991234e-01 9.64406550e-01
2.66599596e-01 4.16690409e-02 -1.44572031e+00 2.34644964e-01
2.64202386e-01 1.16233230e+00 -1.13047016e+00 2.61025727e-01
-9.67722833e-02 1.23700428e+00 3.50100875e-01 4.92151380e-01
-6.84118390e-01 -5.27019501e-01 6.17080152e-01 2.08194014e-02
4.43551987e-01 -5.98645806e-01 1.22858033e-01 -1.08418095e+00
6.70825392e-02 -6.22166932e-01 -3.60347480e-02 -9.72345173e-01
-8.34605992e-01 7.13706434e-01 -7.04049170e-02 -1.18164062e+00
-1.78938687e-01 -8.53199124e-01 -3.42494816e-01 8.60398233e-01
-1.82199383e+00 -1.37050653e+00 -9.60360706e-01 7.14972377e-01
3.72321784e-01 1.08271651e-02 1.23903215e+00 1.95286825e-01
-9.20106098e-02 3.01609606e-01 5.46783566e-01 1.93022355e-01
5.80644310e-01 -1.37560642e+00 4.51053828e-01 8.66852820e-01
5.12855351e-01 3.68193775e-01 7.53103554e-01 -2.33493611e-01
-1.19569576e+00 -1.29948485e+00 9.50921416e-01 -2.88344711e-01
3.22108090e-01 -6.33930683e-01 -6.55489624e-01 7.43275762e-01
1.53841034e-01 2.75344282e-01 1.47680759e-01 -7.26460591e-02
-2.01563358e-01 -1.15733735e-01 -1.20015895e+00 7.45458782e-01
1.16832697e+00 -5.30036271e-01 -3.31476420e-01 4.69325215e-01
4.02359813e-01 -4.99790907e-01 -2.90830076e-01 2.40701124e-01
4.35279399e-01 -1.17128956e+00 1.02882648e+00 -1.34756565e-01
5.38700968e-02 -5.93915403e-01 -7.39048481e-01 -1.47600830e+00
-5.02471030e-01 -7.77378380e-02 6.23106182e-01 7.92190552e-01
1.25534788e-01 -5.49672544e-01 6.76648974e-01 -3.02407593e-01
-3.66029814e-02 -2.59088069e-01 -9.10250843e-01 -9.17593062e-01
-3.98221493e-01 -1.75455168e-01 7.08777428e-01 9.12397861e-01
-4.64599341e-01 -8.58444348e-02 -2.13655531e-01 5.85388720e-01
3.12185585e-01 1.98797360e-01 7.94921219e-01 -1.16773903e+00
1.49600253e-01 -2.66797662e-01 -1.11330700e+00 -7.16656268e-01
2.04139501e-01 -8.69651377e-01 5.75339079e-01 -1.77369881e+00
-2.89498717e-01 -5.33358455e-01 -1.53944150e-01 7.14922190e-01
3.78009528e-01 4.05041784e-01 2.88557351e-01 1.01897314e-01
-4.90500331e-01 5.77669382e-01 7.56433427e-01 -4.44846362e-01
-2.15339795e-01 -1.68612495e-01 1.35946557e-01 6.52897120e-01
9.03421998e-01 -2.78246582e-01 -3.78765672e-01 -5.62638938e-01
2.02423915e-01 -4.69196558e-01 8.47102702e-01 -1.62840629e+00
1.35354474e-01 -1.31776761e-02 6.79947495e-01 -5.89691162e-01
4.23753947e-01 -1.29324818e+00 2.01671362e-01 6.76592052e-01
-1.89352572e-01 4.12941933e-01 3.46632659e-01 7.71953583e-01
-2.96508849e-01 -1.39537500e-03 6.62250102e-01 -5.16381860e-01
-1.45465338e+00 3.04249413e-02 -4.14112329e-01 -5.07170141e-01
1.11655843e+00 -4.35014278e-01 -2.87533045e-01 -2.35476956e-01
-7.67937541e-01 2.72787828e-02 8.30633521e-01 6.35821640e-01
7.44714141e-01 -1.34315920e+00 -4.40027475e-01 5.84872305e-01
6.77382052e-01 -7.56338099e-03 7.26873800e-02 5.46128094e-01
-1.20717168e+00 4.65776026e-01 -8.21053267e-01 -9.12203550e-01
-1.02034760e+00 7.00918257e-01 4.87202346e-01 -8.49874318e-02
-7.72521734e-01 6.81418538e-01 2.80292690e-01 -6.65368617e-01
3.08247030e-01 -8.42639267e-01 -4.98652548e-01 -6.63632229e-02
3.18455964e-01 1.67576224e-01 2.67681092e-01 -8.56990576e-01
-5.28661668e-01 4.78345245e-01 4.70087051e-01 1.70349717e-01
1.33675969e+00 -1.35699898e-01 -4.80283499e-02 6.57313406e-01
8.31027448e-01 -4.69699949e-01 -1.02408254e+00 -1.30456716e-01
3.17916833e-02 -5.46227753e-01 -7.10257143e-02 -7.55543590e-01
-8.84390116e-01 7.33373702e-01 1.27358449e+00 8.60017389e-02
1.02637959e+00 -2.33494967e-01 1.75218225e-01 9.44984674e-01
1.03415275e+00 -6.93040669e-01 -2.18568891e-01 7.66246080e-01
9.26963389e-01 -1.55536509e+00 -4.04145718e-01 -9.54705402e-02
-3.00734639e-01 1.02549958e+00 4.22183365e-01 -3.57178003e-01
5.32818377e-01 -1.60943061e-01 1.99584544e-01 -1.73121452e-01
-8.33063722e-02 -6.25317991e-01 -2.69217063e-02 1.11633039e+00
1.62258178e-01 2.22282410e-01 3.31715524e-01 1.08636469e-02
-1.78965256e-01 -2.01161936e-01 4.57821816e-01 1.29039097e+00
-3.59808236e-01 -5.96863151e-01 -7.07006931e-01 -1.21528737e-01
2.03952506e-01 -2.77308729e-02 -4.07812327e-01 1.03236353e+00
4.30815548e-01 9.03520703e-01 -3.23184915e-02 -2.76141018e-01
5.84806859e-01 -1.67628869e-01 5.62373638e-01 -4.61984903e-01
-5.57991624e-01 -4.08156544e-01 -1.74559027e-01 -8.04947019e-01
-5.00582039e-01 -4.91020381e-01 -8.38123977e-01 -2.95582592e-01
-2.15693638e-01 -1.32802948e-01 9.06332374e-01 7.22221076e-01
5.91950379e-02 6.50629580e-01 3.91773194e-01 -1.37109983e+00
-1.40192330e-01 -7.91553736e-01 -1.05913532e+00 1.88335285e-01
6.80313826e-01 -9.07490134e-01 -4.32968326e-02 4.45451401e-02]
|
[7.608855247497559, -1.8551018238067627]
|
78e15dcf-a91a-4930-9e96-0fb633417d64
|
argoverse-3d-tracking-and-forecasting-with-1
|
1911.0262
| null |
https://arxiv.org/abs/1911.02620v1
|
https://arxiv.org/pdf/1911.02620v1.pdf
|
Argoverse: 3D Tracking and Forecasting with Rich Maps
|
We present Argoverse -- two datasets designed to support autonomous vehicle machine learning tasks such as 3D tracking and motion forecasting. Argoverse was collected by a fleet of autonomous vehicles in Pittsburgh and Miami. The Argoverse 3D Tracking dataset includes 360 degree images from 7 cameras with overlapping fields of view, 3D point clouds from long range LiDAR, 6-DOF pose, and 3D track annotations. Notably, it is the only modern AV dataset that provides forward-facing stereo imagery. The Argoverse Motion Forecasting dataset includes more than 300,000 5-second tracked scenarios with a particular vehicle identified for trajectory forecasting. Argoverse is the first autonomous vehicle dataset to include "HD maps" with 290 km of mapped lanes with geometric and semantic metadata. All data is released under a Creative Commons license at www.argoverse.org. In our baseline experiments, we illustrate how detailed map information such as lane direction, driveable area, and ground height improves the accuracy of 3D object tracking and motion forecasting. Our tracking and forecasting experiments represent only an initial exploration of the use of rich maps in robotic perception. We hope that Argoverse will enable the research community to explore these problems in greater depth.
|
['Slawomir Bak', 'Patsorn Sangkloy', 'Ming-Fang Chang', 'John Lambert', 'Jagjeet Singh', 'Deva Ramanan', 'James Hays', 'Andrew Hartnett', 'Peter Carr', 'Simon Lucey', 'De Wang']
|
2019-11-06
|
argoverse-3d-tracking-and-forecasting-with
|
http://openaccess.thecvf.com/content_CVPR_2019/html/Chang_Argoverse_3D_Tracking_and_Forecasting_With_Rich_Maps_CVPR_2019_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2019/papers/Chang_Argoverse_3D_Tracking_and_Forecasting_With_Rich_Maps_CVPR_2019_paper.pdf
|
cvpr-2019-6
|
['3d-object-tracking']
|
['computer-vision']
|
[-4.46070969e-01 6.31727129e-02 -4.73303139e-01 -7.54492939e-01
-4.13665354e-01 -1.04954553e+00 1.02575600e+00 -1.96393952e-01
-3.26580554e-01 3.71028066e-01 -1.47970524e-02 -6.29888594e-01
6.04010224e-02 -8.39203477e-01 -9.77363169e-01 -3.00608754e-01
-3.63753974e-01 8.65934074e-01 6.24812901e-01 -5.09009123e-01
2.17066571e-01 1.19956052e+00 -1.76198125e+00 -1.54775783e-01
4.10152704e-01 1.08724511e+00 4.31678206e-01 6.41857028e-01
7.06128404e-02 4.37936038e-01 2.61188596e-01 -1.50982350e-01
8.40814888e-01 8.07763636e-01 -2.81603813e-01 7.71545991e-02
1.11726046e+00 -7.58889854e-01 -6.27749503e-01 6.35287285e-01
-2.24973168e-02 -2.18089763e-03 5.95062077e-01 -2.10722637e+00
-9.84478220e-02 -1.44548649e-02 -1.84830680e-01 2.99720943e-01
1.59631148e-01 6.33646965e-01 5.63672721e-01 -9.78333831e-01
9.65423286e-01 1.30184126e+00 1.03265059e+00 8.31681192e-02
-7.70536840e-01 -9.08316016e-01 2.45500609e-01 2.68395513e-01
-1.48332930e+00 -6.27152324e-01 3.44490051e-01 -1.03105783e+00
1.06454003e+00 -1.27706707e-01 6.26971185e-01 8.73314321e-01
4.35397059e-01 5.40062129e-01 6.78091228e-01 3.14010203e-01
2.28534006e-02 4.73670475e-02 2.12203607e-01 9.90750253e-01
3.64531100e-01 6.31910503e-01 -4.42073852e-01 3.97935174e-02
4.60585773e-01 2.63338894e-01 1.70369133e-01 -1.11376297e+00
-1.44673073e+00 8.39331150e-01 4.71845567e-01 -6.79649770e-01
-2.56189615e-01 3.83063465e-01 2.18657345e-01 2.38029629e-01
3.71814340e-01 1.36010572e-02 -4.19656366e-01 -9.05056447e-02
-5.32571554e-01 7.09453642e-01 5.92069745e-01 1.73651516e+00
1.12133014e+00 1.28758267e-01 5.17999411e-01 2.02641353e-01
5.85633099e-01 1.32980394e+00 -4.10638958e-01 -1.84345198e+00
6.46839619e-01 4.86288041e-01 7.05223262e-01 -9.71157372e-01
-7.81306982e-01 -2.46160794e-02 -1.84281468e-01 7.15375960e-01
2.14634672e-01 -1.76855892e-01 -7.84414530e-01 1.15619874e+00
3.42319399e-01 -3.20086218e-02 1.52764767e-01 1.01171136e+00
7.40132928e-01 5.87468505e-01 -1.46143362e-01 2.28450596e-01
9.12028909e-01 -6.88383162e-01 -4.22404736e-01 -5.35000920e-01
8.57095599e-01 -4.99900937e-01 2.86631465e-01 -7.41329044e-02
-5.31321645e-01 -5.04643083e-01 -1.03350711e+00 1.21670790e-01
-7.06228018e-01 6.37078658e-02 6.42508686e-01 2.19674617e-01
-1.23039699e+00 5.33119589e-02 -9.20423090e-01 -7.18124509e-01
5.07982850e-01 1.25734553e-01 -6.78668559e-01 -2.22026721e-01
-9.69143093e-01 1.47664320e+00 2.51756132e-01 -1.67646129e-02
-1.03118396e+00 -8.20411026e-01 -1.17698002e+00 -6.23681426e-01
2.41034806e-01 -4.34972286e-01 1.21549976e+00 -6.46482268e-03
-6.91360235e-01 9.92426276e-01 -2.19427601e-01 -7.16951430e-01
6.40403450e-01 -3.22718084e-01 -6.30338550e-01 -1.94075868e-01
4.97209460e-01 1.31571817e+00 4.14296806e-01 -1.34682620e+00
-1.14313424e+00 -6.61567330e-01 -1.47579715e-01 2.98857480e-01
5.39221108e-01 -3.05183083e-01 -3.42027217e-01 -5.75196072e-02
3.79059762e-01 -1.38981915e+00 -1.74684078e-01 4.09876883e-01
-5.18577769e-02 -1.36018649e-01 1.21908295e+00 -2.24005491e-01
2.47394860e-01 -2.12064981e+00 -3.99867624e-01 -6.89978674e-02
9.34128687e-02 -8.58996287e-02 -6.79002032e-02 5.51049292e-01
5.49494505e-01 -4.05425206e-02 2.57543653e-01 -9.75824669e-02
2.16740415e-01 3.45244497e-01 -6.31881356e-01 8.20581973e-01
-1.26968756e-01 9.32228506e-01 -8.08268070e-01 -1.91026941e-01
6.38316214e-01 2.09594175e-01 -1.30768269e-01 -3.26048344e-01
-2.19358787e-01 3.10518444e-01 -4.69209045e-01 9.06957030e-01
1.09848440e+00 2.38798425e-01 -4.64991421e-01 -1.43006071e-01
-8.79555285e-01 1.05153257e-02 -9.33827996e-01 1.43842864e+00
1.07939050e-01 1.34327197e+00 6.88650161e-02 -2.05727488e-01
1.16082597e+00 -2.12432623e-01 6.00683033e-01 -6.33911967e-01
-1.91128477e-01 2.35301644e-01 -6.44393325e-01 -4.46145326e-01
1.08512843e+00 2.85294026e-01 -2.77738214e-01 -1.29991367e-01
-4.56072688e-01 -5.02041757e-01 -4.82849125e-03 2.69397348e-01
9.83048975e-01 2.38586426e-01 2.29444001e-02 -2.31849149e-01
-1.83583274e-02 1.30500650e+00 5.24673998e-01 7.25345850e-01
-4.50544298e-01 2.84014374e-01 -1.52959898e-01 -9.04904485e-01
-1.47379243e+00 -1.33999431e+00 -5.43680429e-01 5.81639349e-01
7.39644408e-01 7.60563016e-02 1.07982591e-01 -4.68883067e-01
9.75565672e-01 7.89692342e-01 -3.50490838e-01 3.48165303e-01
-5.13572454e-01 -5.29093407e-02 5.68819702e-01 6.73022449e-01
6.35642231e-01 -4.75436598e-01 -7.81920075e-01 8.00031051e-02
3.48706245e-02 -1.42268980e+00 -2.46466488e-01 -8.49723518e-02
-7.45315015e-01 -1.04434097e+00 -4.57286313e-02 -5.85150361e-01
3.53638977e-01 1.03895950e+00 7.88614929e-01 -3.36858630e-01
1.21213563e-01 5.96624970e-01 -1.96978480e-01 -7.94746935e-01
-3.91583741e-01 -1.88379824e-01 4.49578613e-01 -5.26208103e-01
7.40649939e-01 -2.36940101e-01 -3.25847834e-01 8.38624895e-01
1.70245662e-01 3.08369935e-01 4.57756847e-01 1.72030136e-01
5.95019042e-01 -2.63619751e-01 1.86790958e-01 -2.44180024e-01
-3.62455130e-01 -7.45579660e-01 -1.30025661e+00 -3.09197068e-01
-5.80527842e-01 -3.98014814e-01 -7.14166239e-02 1.28930792e-01
-8.90968680e-01 6.94560945e-01 3.13836843e-01 -7.64109731e-01
-6.07022583e-01 2.39022523e-01 1.81653589e-01 -1.91044524e-01
5.69992304e-01 1.14182055e-01 4.41118032e-01 -2.15606287e-01
4.80380863e-01 6.43485904e-01 1.03094172e+00 9.99428257e-02
1.17975760e+00 9.44891214e-01 2.66377389e-01 -8.18664670e-01
-4.79022473e-01 -7.61304021e-01 -9.51080322e-01 -6.81055069e-01
9.38167274e-01 -1.56080616e+00 -7.18663454e-01 5.12170017e-01
-1.06158960e+00 -5.44421852e-01 1.43732056e-01 6.71718776e-01
-8.00924003e-01 -1.07367322e-01 -2.89765429e-02 -8.13717961e-01
2.14749575e-01 -9.87311900e-01 1.24749422e+00 -6.21637180e-02
8.69828183e-03 -6.21241927e-01 7.49132559e-02 4.11878943e-01
2.34338537e-01 4.19170976e-01 1.66884825e-01 -1.93674922e-01
-1.33675969e+00 -3.98268998e-01 -2.95031756e-01 -3.75176907e-01
-2.05121949e-01 -6.59129722e-03 -7.82859147e-01 -2.55313933e-01
-6.54852629e-01 -8.66186470e-02 1.02566528e+00 4.18883860e-01
3.04482818e-01 3.19575705e-02 -9.69514847e-01 6.91012204e-01
1.29308450e+00 4.29508865e-01 2.52224147e-01 7.13179648e-01
7.97861934e-01 6.90725625e-01 1.29840112e+00 2.37734899e-01
1.04135942e+00 9.33685839e-01 1.05218053e+00 1.80206150e-01
5.68119697e-02 -4.77301329e-01 3.24180007e-01 1.31537512e-01
5.19197024e-02 -1.23061955e-01 -1.48106730e+00 6.12576187e-01
-1.89358687e+00 -1.24309862e+00 -6.57462418e-01 1.96590698e+00
-1.75531149e-01 2.51331627e-01 -3.63097563e-02 -5.71447670e-01
6.77729487e-01 2.38987014e-01 -7.84344971e-01 4.14910875e-02
-1.07591845e-01 -1.19742310e+00 1.65744114e+00 6.62192345e-01
-1.35840750e+00 1.06396878e+00 5.89238214e+00 1.60169467e-01
-8.40280414e-01 2.49754675e-02 -1.44078970e-01 1.60762127e-02
-3.76787782e-01 3.02077323e-01 -1.61486423e+00 2.58698434e-01
1.03839719e+00 -2.37252042e-02 1.68411717e-01 1.03435326e+00
5.50785422e-01 -2.90109843e-01 -8.39406013e-01 6.65201485e-01
-1.66940764e-01 -1.90185845e+00 -3.47023427e-01 5.08752108e-01
6.21671855e-01 1.06995237e+00 -5.56322932e-02 3.10347795e-01
7.09344566e-01 -7.43175328e-01 1.26484275e+00 7.33509123e-01
7.58900881e-01 -7.08915114e-01 6.11009121e-01 7.15516329e-01
-1.38695061e+00 -2.77261853e-01 -3.06328982e-01 -9.48256701e-02
6.17913187e-01 9.18483883e-02 -1.17931688e+00 3.72672945e-01
9.99952018e-01 1.08189797e+00 -4.44759518e-01 1.12027204e+00
3.57143641e-01 1.87835544e-01 -5.33002794e-01 1.58299595e-01
5.17745256e-01 -7.40943849e-02 1.01170552e+00 8.33143592e-01
4.72983837e-01 1.14965364e-01 5.46498299e-01 5.76952636e-01
3.08703154e-01 -6.57003760e-01 -1.58107567e+00 3.16613227e-01
1.10797465e+00 1.03538406e+00 -4.03605849e-01 -1.86520666e-01
-5.83786428e-01 1.43530130e-01 -1.03477649e-01 2.74496019e-01
-8.70166063e-01 9.68107209e-02 1.20074487e+00 5.61817229e-01
4.03602421e-01 -9.13469076e-01 -3.05415869e-01 -5.88371813e-01
-3.20986688e-01 1.18102893e-01 -1.82119966e-01 -1.34359717e+00
-9.41617668e-01 3.46514702e-01 5.30078053e-01 -1.81570041e+00
-4.88850921e-01 -7.68378437e-01 -1.63706601e-01 7.26256728e-01
-1.68776119e+00 -1.31530607e+00 -8.38894427e-01 3.01449746e-01
6.75832748e-01 -5.67272127e-01 3.13751101e-01 -2.86172051e-02
2.15631202e-02 -2.56806225e-01 2.37354562e-01 -1.51244119e-01
6.25242472e-01 -8.91822577e-01 1.02100575e+00 5.74339330e-01
-1.92692041e-01 6.62573986e-03 8.59813988e-01 -1.11220360e+00
-1.87020600e+00 -1.63621116e+00 8.39224935e-01 -1.11824512e+00
6.95603848e-01 -2.97801316e-01 -5.43084323e-01 1.39354908e+00
-7.37116113e-02 2.41466671e-01 -2.08377719e-01 -3.12053889e-01
-8.77228528e-02 -3.15630972e-01 -1.04647660e+00 4.35121149e-01
1.46123409e+00 -2.45184869e-01 -3.19280446e-01 3.37172151e-01
7.33861864e-01 -9.46484387e-01 -8.30009222e-01 7.43504882e-01
7.50136733e-01 -7.77714789e-01 9.58217502e-01 -7.55476505e-02
-2.18047798e-01 -7.99511969e-01 -6.08409524e-01 -1.08262527e+00
-4.04636115e-01 -2.43435800e-01 8.69639392e-04 8.54195833e-01
3.88087869e-01 -8.06230605e-01 9.00875866e-01 4.67287421e-01
-8.43619645e-01 -1.28014460e-01 -8.66212606e-01 -1.13793707e+00
-1.14052378e-01 -8.35057259e-01 6.44436836e-01 6.49301052e-01
-6.57592475e-01 -1.51865050e-01 -1.75780460e-01 8.08146954e-01
8.62847865e-01 2.19154835e-01 1.52973056e+00 -1.68900859e+00
7.19185054e-01 -2.70497292e-01 -9.52775478e-01 -1.21000874e+00
3.62355769e-01 -8.70901346e-01 2.29241401e-01 -1.51327682e+00
-2.03220606e-01 -7.78675675e-01 5.92529416e-01 5.24688482e-01
6.31854534e-01 1.09375291e-01 7.92535692e-02 5.86504519e-01
-4.84413326e-01 5.00524461e-01 8.27663839e-01 -1.97540894e-01
5.87679818e-02 8.55796710e-02 -5.80102056e-02 7.10423589e-01
8.20239007e-01 -3.01027894e-01 -4.01878595e-01 -6.99903965e-01
-1.14073880e-01 2.24079549e-01 8.63446236e-01 -1.07652223e+00
6.29493237e-01 -4.63508964e-01 4.05929983e-01 -1.80015707e+00
7.42511153e-01 -1.04551828e+00 5.42980433e-01 2.97405332e-01
8.75750184e-02 4.70086694e-01 6.07071042e-01 7.32860208e-01
1.39016688e-01 2.50111789e-01 4.31791186e-01 -5.06485812e-02
-1.92614627e+00 5.42024434e-01 -7.20274091e-01 -2.56246507e-01
1.42689991e+00 -6.96183264e-01 -8.02936673e-01 -3.24870408e-01
-3.69038343e-01 9.38634336e-01 9.83822167e-01 8.48754942e-01
6.52015984e-01 -1.46277094e+00 -7.49177098e-01 4.17189956e-01
5.26630759e-01 7.06999153e-02 1.71019256e-01 7.81493127e-01
-6.84275210e-01 8.04424703e-01 -5.11243522e-01 -1.24809182e+00
-1.06915843e+00 3.94011289e-01 2.52995849e-01 7.73137748e-01
-1.09285402e+00 3.29033375e-01 -5.01158908e-02 -8.66611600e-01
-5.63764945e-03 -1.38687463e-02 1.43583352e-02 -1.32800758e-01
3.85394841e-01 5.70142984e-01 -8.48340765e-02 -1.30794704e+00
-5.78309715e-01 6.53245211e-01 2.97439039e-01 -1.83478251e-01
1.03821719e+00 -7.75438011e-01 4.51601952e-01 5.16255140e-01
9.48233664e-01 -1.63481593e-01 -2.03773880e+00 1.27830893e-01
1.31432459e-01 -5.46089292e-01 3.11626703e-01 -4.36550051e-01
-8.13091636e-01 4.53350097e-01 8.51388574e-01 -1.09562226e-01
3.33759606e-01 3.01938772e-01 5.99264920e-01 7.36852407e-01
1.09150362e+00 -7.02908754e-01 -6.18797719e-01 9.97553885e-01
9.05394971e-01 -1.64956784e+00 2.50986665e-02 -3.56180072e-01
-8.08430195e-01 9.26308811e-01 8.16777527e-01 2.48047262e-02
4.52984631e-01 3.85332257e-01 1.85716748e-01 -4.02469426e-01
-8.04011464e-01 -3.40426683e-01 4.19223495e-02 9.14632559e-01
-4.72745925e-01 1.83563173e-01 5.26950061e-01 -8.42993259e-02
-4.24114913e-01 -7.93874711e-02 5.64109027e-01 9.81935501e-01
-1.04746139e+00 -3.58312428e-01 -4.93229896e-01 2.81214774e-01
6.77946150e-01 3.72048914e-01 -8.01112652e-02 1.27323830e+00
1.48408353e-01 9.15588498e-01 6.22065306e-01 -5.72894990e-01
4.40232724e-01 -1.66015327e-01 2.47939602e-01 -2.57231891e-01
3.29058021e-01 -4.94900018e-01 5.22649527e-01 -8.71296942e-01
-1.54283881e-01 -1.11914265e+00 -1.49063838e+00 -8.36398602e-01
1.30105346e-01 -2.87809342e-01 1.15976763e+00 7.87773550e-01
5.90317726e-01 -1.68058112e-01 6.15885496e-01 -1.34009218e+00
-1.85060441e-01 -5.58513820e-01 -5.63646615e-01 -2.04164460e-01
6.81531489e-01 -1.18351293e+00 -1.59672201e-01 -7.98135996e-02]
|
[7.719656467437744, -1.9704999923706055]
|
8660ef49-3e8f-4bbc-9a37-0400e35903d6
|
learning-robust-agents-for-visual-navigation
|
2109.10493
| null |
https://arxiv.org/abs/2109.10493v2
|
https://arxiv.org/pdf/2109.10493v2.pdf
|
Benchmarking Augmentation Methods for Learning Robust Navigation Agents: the Winning Entry of the 2021 iGibson Challenge
|
Recent advances in deep reinforcement learning and scalable photorealistic simulation have led to increasingly mature embodied AI for various visual tasks, including navigation. However, while impressive progress has been made for teaching embodied agents to navigate static environments, much less progress has been made on more dynamic environments that may include moving pedestrians or movable obstacles. In this study, we aim to benchmark different augmentation techniques for improving the agent's performance in these challenging environments. We show that adding several dynamic obstacles into the scene during training confers significant improvements in test-time generalization, achieving much higher success rates than baseline agents. We find that this approach can also be combined with image augmentation methods to achieve even higher success rates. Additionally, we show that this approach is also more robust to sim-to-sim transfer than image augmentation methods. Finally, we demonstrate the effectiveness of this dynamic obstacle augmentation approach by using it to train an agent for the 2021 iGibson Challenge at CVPR, where it achieved 1st place for Interactive Navigation. Video link: https://www.youtube.com/watch?v=HxUX2HeOSE4
|
['Sehoon Ha', 'Dhruv Batra', 'Qian Luo', 'Naoki Yokoyama']
|
2021-09-22
| null | null | null | null |
['image-augmentation', 'pointgoal-navigation']
|
['computer-vision', 'robots']
|
[ 1.35626295e-03 4.68018539e-02 1.80486575e-01 -2.29071304e-01
-5.26638687e-01 -6.49282932e-01 8.90745461e-01 -9.69596729e-02
-9.36975837e-01 9.18155015e-01 2.36783117e-01 -3.94031793e-01
3.60483944e-01 -6.31830156e-01 -9.49523509e-01 -4.42844033e-01
-4.06043470e-01 4.44744080e-01 3.62742454e-01 -7.56580889e-01
8.83283168e-02 3.96789551e-01 -1.67721164e+00 1.26494676e-01
6.73395991e-01 4.37710345e-01 4.26116019e-01 8.54315937e-01
3.95079553e-01 9.51824427e-01 -5.54339707e-01 -1.00621149e-01
5.38605809e-01 -4.40156281e-01 -8.05904806e-01 -2.49600902e-01
4.43832725e-01 -5.22632480e-01 -7.63687909e-01 8.95751417e-01
5.64940691e-01 6.66130483e-01 4.42328244e-01 -1.43722415e+00
-5.92653751e-01 2.96536535e-01 -4.27413017e-01 3.46317828e-01
5.43393373e-01 5.49193740e-01 5.82294941e-01 -5.63422620e-01
7.97710478e-01 1.17584729e+00 5.56048572e-01 8.56633365e-01
-1.00350034e+00 -6.36751831e-01 2.72050411e-01 4.09571141e-01
-8.49028468e-01 -4.58208084e-01 3.99115324e-01 -9.87750292e-02
1.20787776e+00 9.05685872e-02 9.22962904e-01 1.45371068e+00
-2.13461872e-02 9.60775256e-01 1.09691155e+00 -3.05988967e-01
2.98730612e-01 -1.47281975e-01 -3.58056813e-01 8.94564033e-01
-1.92742832e-02 6.10155046e-01 -2.36520901e-01 2.56440490e-01
9.52281356e-01 -2.73553789e-01 -3.02132785e-01 -5.49372494e-01
-1.30596364e+00 7.17675567e-01 9.66542423e-01 1.62286788e-01
-2.42554858e-01 6.32239401e-01 3.84193540e-01 2.38089204e-01
1.39998853e-01 7.88142681e-01 -2.48955444e-01 -3.23954135e-01
-2.79098600e-01 6.48082674e-01 6.51124775e-01 8.34973395e-01
4.57478404e-01 4.01623368e-01 3.00278142e-02 5.99901676e-01
1.07251532e-01 4.11934912e-01 3.31300616e-01 -1.40394461e+00
2.91984200e-01 2.35810041e-01 2.97968209e-01 -7.42589593e-01
-7.21702933e-01 -2.84151971e-01 -4.49956864e-01 1.05989969e+00
3.91567439e-01 -4.16704088e-01 -1.27371848e+00 1.95465446e+00
3.45842928e-01 2.18719646e-01 2.73294151e-01 1.06488168e+00
8.99196982e-01 6.06747031e-01 3.51832151e-01 5.94397902e-01
1.08921635e+00 -1.36295569e+00 -3.67197782e-01 -5.25579274e-01
7.46079564e-01 -4.76166695e-01 1.19800103e+00 2.05378145e-01
-1.04071069e+00 -4.64319348e-01 -1.18235779e+00 -2.94129979e-02
-5.06530881e-01 -3.18400204e-01 8.99165094e-01 4.41166967e-01
-1.19137216e+00 5.18178403e-01 -1.00516152e+00 -5.98383307e-01
4.61366743e-01 4.61100638e-01 -6.69813514e-01 -2.73962729e-02
-1.11548901e+00 1.23820543e+00 2.53463149e-01 -1.39414296e-01
-1.32415342e+00 -6.09966278e-01 -1.07629621e+00 -2.83583939e-01
3.14333498e-01 -7.52457798e-01 1.40488839e+00 -9.11313236e-01
-1.56639445e+00 6.83364213e-01 5.19163385e-02 -6.36272371e-01
6.46454811e-01 -3.61283660e-01 -1.24654591e-01 3.40118371e-02
-1.89516880e-02 1.33965254e+00 3.67216974e-01 -1.55912304e+00
-7.97170401e-01 -1.65281981e-01 7.52470136e-01 7.39352584e-01
-1.76851880e-02 -1.13775276e-01 -4.29023206e-01 -4.26163197e-01
-3.34416091e-01 -1.24172175e+00 -5.51222384e-01 5.83017915e-02
-3.50831859e-02 5.62655926e-03 8.34031224e-01 -5.93303084e-01
2.07310185e-01 -1.93016779e+00 3.93790603e-01 -1.56811193e-01
9.66211632e-02 2.51537740e-01 -4.05297607e-01 4.50916290e-01
1.47147357e-01 -6.74564913e-02 -1.80465922e-01 -4.98828858e-01
9.47399139e-02 1.77548528e-01 -8.23048577e-02 2.66221434e-01
3.24539877e-02 1.27144122e+00 -1.03204644e+00 -1.05631329e-01
4.46771383e-01 7.26965547e-01 -7.20100284e-01 2.64212652e-03
-2.93545932e-01 8.89523029e-01 -9.91695970e-02 3.61648619e-01
2.66781539e-01 6.70113936e-02 -2.09557682e-01 2.24515557e-01
-2.18007848e-01 2.43599955e-02 -8.16703141e-01 1.95424092e+00
-4.41700965e-01 1.10299814e+00 1.25738859e-01 -7.71329701e-01
5.29188454e-01 2.49036159e-02 2.59957790e-01 -1.24130785e+00
2.38444239e-01 1.29675940e-02 2.20861897e-01 -4.03148651e-01
6.63268626e-01 -5.23203462e-02 6.73397481e-02 8.19401219e-02
-7.92157352e-02 -3.93245190e-01 3.00321102e-01 2.83478647e-01
1.27832294e+00 6.90737009e-01 -1.17131017e-01 -1.31808668e-01
1.14847280e-01 4.34561968e-01 1.21592551e-01 8.88502777e-01
-3.72090250e-01 5.32142162e-01 1.41157368e-02 -4.36971545e-01
-1.32714260e+00 -9.59946752e-01 2.40188763e-01 1.14884090e+00
4.65628237e-01 -1.73592344e-01 -8.84510100e-01 -5.35090744e-01
-1.34090498e-01 7.43000507e-01 -8.49071562e-01 -2.69878060e-01
-7.04928339e-01 -5.10318398e-01 7.68510520e-01 8.01879644e-01
9.27931607e-01 -1.54846382e+00 -1.10507631e+00 -9.81387962e-03
-1.09734565e-01 -1.15269518e+00 6.94626709e-03 1.51294380e-01
-4.55469161e-01 -8.84188294e-01 -8.98121059e-01 -9.64274645e-01
4.88936186e-01 3.60617787e-01 9.92010891e-01 2.26989016e-01
-3.36940944e-01 7.34159946e-01 -5.21140814e-01 -4.50982779e-01
-3.95342588e-01 -3.54667380e-02 1.42135859e-01 -8.38348389e-01
2.38982961e-02 -3.52919489e-01 -8.56831849e-01 3.02901983e-01
-6.60012960e-01 2.53117859e-01 4.17198181e-01 7.59275317e-01
1.02624349e-01 -3.19333404e-01 3.20612788e-01 -5.03130138e-01
5.50224721e-01 -2.50639677e-01 -4.99691308e-01 -1.93915412e-01
-2.10856453e-01 -1.07851453e-01 4.16991442e-01 -4.78603572e-01
-8.86834085e-01 -1.01674512e-01 -5.86791396e-01 -7.86417797e-02
-4.41681832e-01 2.30453521e-01 1.40298709e-01 -5.03986299e-01
8.58461738e-01 8.42688307e-02 2.08142214e-02 6.51437044e-02
4.48020607e-01 1.94810003e-01 6.58641756e-01 -4.64276999e-01
8.68613660e-01 6.75602019e-01 -1.13249667e-01 -9.12317157e-01
-3.23234886e-01 1.52035384e-02 -3.07274312e-01 -3.55329692e-01
1.04331088e+00 -9.99669850e-01 -9.10023391e-01 3.54864091e-01
-7.84932077e-01 -1.24257159e+00 -3.97358745e-01 5.51512122e-01
-7.87841797e-01 1.46570206e-01 -5.39190650e-01 -4.69937623e-01
8.56276900e-02 -1.23912013e+00 7.72405803e-01 5.25916517e-01
-1.83292419e-01 -9.45455432e-01 3.33789438e-01 2.26946652e-01
5.53848922e-01 3.31637949e-01 5.13279438e-01 -4.22959149e-01
-5.88281751e-01 -1.12384029e-01 -1.87334001e-01 -1.58479512e-02
-1.60831213e-01 -2.97825992e-01 -7.89004087e-01 -5.76456904e-01
-5.72448850e-01 -6.98713601e-01 8.30527306e-01 3.00720453e-01
7.36317635e-01 2.36570001e-01 -3.48530293e-01 7.01436341e-01
1.13665402e+00 6.09080613e-01 9.55185533e-01 1.01829898e+00
7.04951227e-01 4.92707700e-01 5.85575104e-01 2.88335290e-02
5.17752945e-01 8.77015471e-01 8.36491644e-01 -4.65446949e-01
-4.09851402e-01 -1.87989652e-01 2.85708547e-01 1.34130031e-01
-5.00354648e-01 -3.63182664e-01 -1.05487239e+00 3.87523293e-01
-1.68405282e+00 -1.06715262e+00 1.83338225e-01 1.91653097e+00
3.62355918e-01 1.29207894e-01 2.57739842e-01 -1.99090943e-01
2.83556968e-01 6.52415305e-02 -7.03615904e-01 -3.22492033e-01
-1.93387926e-01 1.36042550e-01 4.53943819e-01 6.76731050e-01
-1.11436009e+00 1.22935033e+00 6.57329798e+00 2.75872052e-01
-9.10082221e-01 5.34694158e-02 3.42985332e-01 -8.27408805e-02
-2.14892372e-01 -7.19028488e-02 -4.51643169e-01 7.84775913e-02
5.77097774e-01 9.81104076e-02 6.54753685e-01 7.66097903e-01
2.10199673e-02 -3.59994918e-01 -8.64993572e-01 7.37510264e-01
1.48605481e-01 -1.29389310e+00 -2.45668143e-01 1.42333657e-01
8.12910020e-01 4.30566221e-01 3.27932984e-01 7.37421453e-01
8.25556040e-01 -1.18059039e+00 6.15602851e-01 8.39591101e-02
5.23889959e-01 -8.30636203e-01 5.02002537e-01 2.39988968e-01
-9.38581586e-01 -1.86240584e-01 -3.41937095e-01 -3.42404276e-01
1.07601553e-01 -5.48333883e-01 -7.83484817e-01 3.01555514e-01
8.98697138e-01 4.49985027e-01 -4.08553213e-01 1.36843634e+00
-2.86842823e-01 1.05794497e-01 -3.15115184e-01 -2.70121157e-01
5.48250973e-01 -1.21796809e-01 6.03225529e-01 9.69614506e-01
1.84146881e-01 2.35549420e-01 7.75661469e-02 4.32686657e-01
-9.09999236e-02 -9.90262777e-02 -9.51793075e-01 2.05851302e-01
1.96846709e-01 1.03871417e+00 -8.81716192e-01 -2.27441728e-01
-4.04468477e-01 1.29236698e+00 6.28016293e-01 6.30386889e-01
-1.09298110e+00 -2.77239710e-01 8.56963456e-01 -7.33732432e-02
3.19471538e-01 -6.58111036e-01 1.38464749e-01 -9.25059557e-01
-2.65374422e-01 -1.03707778e+00 -5.74653782e-02 -9.96061683e-01
-6.21818364e-01 9.03348207e-01 -8.74900296e-02 -9.68053997e-01
-1.83260664e-01 -7.12930620e-01 -5.48261404e-01 5.47729611e-01
-1.23337054e+00 -1.18597674e+00 -6.26551330e-01 5.59193790e-01
7.54896224e-01 -2.83171207e-01 9.75711882e-01 2.20336810e-01
-3.42856973e-01 4.10227329e-01 1.64696917e-01 1.48350760e-01
6.34339690e-01 -1.29281938e+00 8.17327142e-01 8.55781019e-01
2.33292460e-01 2.97887564e-01 1.11354160e+00 -6.30625427e-01
-1.26910424e+00 -7.52950728e-01 -1.62847981e-01 -6.76569879e-01
5.22416353e-01 -3.95751685e-01 -5.84402204e-01 9.30156767e-01
6.55486882e-01 -1.59842670e-01 2.63842344e-01 1.42544329e-01
-2.79071301e-01 4.53631580e-01 -1.04434013e+00 1.21143675e+00
1.22942853e+00 -3.40407193e-01 -4.79458332e-01 2.07549348e-01
6.62768841e-01 -8.22266281e-01 -4.14242566e-01 4.99925792e-01
6.91915810e-01 -9.73055124e-01 1.13168752e+00 -9.05593455e-01
6.96981192e-01 -3.57954085e-01 -1.90943971e-01 -1.70820808e+00
-2.35200688e-01 -2.69171417e-01 1.22813754e-01 7.40193844e-01
3.92093003e-01 -5.59348047e-01 1.06533921e+00 5.61257005e-01
-3.37246120e-01 -5.88054121e-01 -8.11550200e-01 -7.17537165e-01
2.80744523e-01 -2.74705380e-01 2.06408054e-01 7.01591313e-01
1.21726664e-02 1.61076665e-01 -4.24955666e-01 1.84837237e-01
5.08982420e-01 -2.74456829e-01 1.05200553e+00 -7.73201644e-01
-2.78530896e-01 -4.74173695e-01 -6.68002963e-01 -1.14792430e+00
2.08344713e-01 -7.61168480e-01 2.49287754e-01 -1.87709928e+00
9.10725519e-02 -4.65168208e-01 -5.06460555e-02 4.12487924e-01
-1.15483344e-01 6.54474616e-01 5.01754045e-01 -5.32195270e-02
-7.30755866e-01 7.19094634e-01 1.51532269e+00 -4.45869714e-02
-1.92634702e-01 -2.77552962e-01 -5.13045132e-01 6.75234199e-01
1.05526114e+00 -1.76855940e-02 -4.78440136e-01 -7.29858756e-01
1.26535475e-01 -2.74503231e-01 6.95086837e-01 -1.44998157e+00
2.49886125e-01 9.75340158e-02 6.42391026e-01 -2.16209054e-01
7.43075907e-01 -6.61560833e-01 9.04029422e-03 6.87999010e-01
-2.52469361e-01 3.70292455e-01 8.40468824e-01 4.70712483e-01
1.26940832e-01 -1.28798634e-01 5.91554224e-01 -2.52800852e-01
-1.24502993e+00 1.43685311e-01 -6.32263601e-01 1.33840308e-01
1.22860181e+00 -2.70480037e-01 -6.23774886e-01 -6.28772616e-01
-8.67452204e-01 5.26160479e-01 7.46266425e-01 5.64366639e-01
7.11430490e-01 -1.21969628e+00 -6.48374856e-01 -6.05128845e-03
-3.80554199e-02 -2.11054116e-01 3.66571873e-01 5.47474027e-01
-8.56242418e-01 1.83544487e-01 -8.05348992e-01 -4.99368012e-01
-1.29508245e+00 4.88640726e-01 4.69300061e-01 2.03028724e-01
-8.87183011e-01 1.10047674e+00 3.52709711e-01 -6.16392016e-01
3.91016036e-01 5.64293638e-02 -2.39492983e-01 -3.62603784e-01
5.41178882e-01 1.54263496e-01 -2.90399194e-01 -6.40856504e-01
-2.89356887e-01 4.52414185e-01 -1.35475397e-01 -5.62522352e-01
1.37777758e+00 -7.46297240e-02 5.57765365e-01 2.49475345e-01
8.15462172e-01 -1.25230461e-01 -1.73138618e+00 1.92661345e-01
-3.97386640e-01 -2.90985376e-01 -2.03275278e-01 -8.86998892e-01
-8.70867729e-01 7.52763450e-01 7.96432674e-01 4.26024795e-02
8.44945014e-01 -8.63539577e-02 5.24422705e-01 7.36443639e-01
6.84643805e-01 -7.71560252e-01 2.88554341e-01 6.64230883e-01
1.05383384e+00 -1.51450491e+00 -1.48777157e-01 -1.32211775e-01
-9.69750762e-01 6.70914888e-01 8.76032054e-01 -3.67659420e-01
1.93518683e-01 4.37862009e-01 2.41210029e-01 -1.03099085e-01
-5.51187992e-01 -6.03573620e-01 1.34298652e-02 1.05340755e+00
2.36890092e-01 -1.37625277e-01 2.21065685e-01 -2.57369071e-01
-3.86820018e-01 -2.66440898e-01 5.51573932e-01 1.07427776e+00
-4.08471495e-01 -9.22947288e-01 -1.11608356e-01 7.52599761e-02
-1.28678367e-01 -5.21236472e-02 -2.86965162e-01 1.37487769e+00
-5.32283522e-02 7.30612695e-01 -1.63311977e-02 -3.69458109e-01
5.41050851e-01 -1.62655711e-01 9.66057181e-01 -4.12145972e-01
-5.71347177e-01 -3.29163939e-01 2.54486382e-01 -7.13984728e-01
-4.18185264e-01 -6.74540401e-01 -1.55373263e+00 -3.99572402e-01
1.15470603e-01 6.61422685e-02 8.72514129e-01 7.59202123e-01
1.70411736e-01 9.56160367e-01 -5.27920350e-02 -1.45923686e+00
-2.51527186e-02 -7.27305472e-01 -5.76539524e-02 3.38836938e-01
4.14566487e-01 -9.73642290e-01 -2.53431886e-01 -1.51843384e-01]
|
[4.459345817565918, 0.8035050630569458]
|
347afeb9-9325-459a-99a0-bb84d054280d
|
transformer-based-vulnerability-detection-in
|
2306.01754
| null |
https://arxiv.org/abs/2306.01754v1
|
https://arxiv.org/pdf/2306.01754v1.pdf
|
Transformer-based Vulnerability Detection in Code at EditTime: Zero-shot, Few-shot, or Fine-tuning?
|
Software vulnerabilities bear enterprises significant costs. Despite extensive efforts in research and development of software vulnerability detection methods, uncaught vulnerabilities continue to put software owners and users at risk. Many current vulnerability detection methods require that code snippets can compile and build before attempting detection. This, unfortunately, introduces a long latency between the time a vulnerability is injected to the time it is removed, which can substantially increases the cost of fixing a vulnerability. We recognize that the current advances in machine learning can be used to detect vulnerable code patterns on syntactically incomplete code snippets as the developer is writing the code at EditTime. In this paper we present a practical system that leverages deep learning on a large-scale data set of vulnerable code patterns to learn complex manifestations of more than 250 vulnerability types and detect vulnerable code patterns at EditTime. We discuss zero-shot, few-shot, and fine-tuning approaches on state of the art pre-trained Large Language Models (LLMs). We show that in comparison with state of the art vulnerability detection models our approach improves the state of the art by 10%. We also evaluate our approach to detect vulnerability in auto-generated code by code LLMs. Evaluation on a benchmark of high-risk code scenarios shows a reduction of up to 90% vulnerability reduction.
|
['Neel Sundaresan', 'Mohamed Elkamhawy', 'Eslam Kamal', 'Alec Helyar', 'Yevhen Mohylevskyy', 'Roshanak Zilouchian Moghaddam', 'Anant Kharkar', 'Aaron Chan']
|
2023-05-23
| null | null | null | null |
['vulnerability-detection']
|
['miscellaneous']
|
[ 3.41084227e-02 -9.81758349e-04 -2.45852292e-01 -8.50499719e-02
-1.20956707e+00 -1.08997440e+00 1.66167811e-01 6.51059091e-01
1.37783453e-01 -7.35442489e-02 -5.07693999e-02 -9.86844838e-01
2.13949367e-01 -9.44974184e-01 -6.88549638e-01 1.32372186e-01
-4.72093880e-01 -3.36748123e-01 5.81389785e-01 -4.55372423e-01
6.68165743e-01 9.53541547e-02 -1.41227221e+00 5.93619525e-01
7.78321028e-01 9.47496071e-02 -1.29844353e-01 9.15125072e-01
-5.58205307e-01 8.45385969e-01 -8.37308049e-01 -6.69729948e-01
3.76707673e-01 2.15582922e-01 -9.59138513e-01 -6.15657985e-01
6.83422744e-01 -3.51988286e-01 6.75778463e-03 1.50947607e+00
2.05709040e-01 -5.79965115e-01 1.41530871e-01 -1.21835232e+00
-7.49947846e-01 1.13483775e+00 -1.02521777e+00 5.23924291e-01
5.74039578e-01 1.25281677e-01 8.48564923e-01 -7.09833741e-01
5.62656581e-01 1.09974015e+00 1.00566661e+00 6.32208765e-01
-1.01863575e+00 -4.79712307e-01 6.23184964e-02 -1.22923553e-01
-1.16792786e+00 -1.82189971e-01 5.28395772e-01 -1.09627092e+00
1.97281909e+00 1.70188397e-01 1.44844614e-02 9.85571921e-01
7.31917143e-01 1.51916698e-01 4.16373223e-01 -7.44837821e-01
1.90962240e-01 1.57742798e-01 6.87886357e-01 9.15929675e-01
4.61490989e-01 1.14998706e-01 3.06274086e-01 -9.52720463e-01
-8.05903301e-02 2.35682756e-01 2.12367862e-01 1.75584733e-01
-4.65332180e-01 1.04343367e+00 4.24950160e-02 4.51407403e-01
2.03273922e-01 2.51644701e-01 9.53030765e-01 6.89383209e-01
3.70296627e-01 6.76034033e-01 -6.23613417e-01 -4.37429875e-01
-1.12200630e+00 1.16167657e-01 1.02665746e+00 8.06405008e-01
9.40026581e-01 1.60610080e-01 6.72223568e-02 5.28569341e-01
3.17956805e-01 2.54224807e-01 3.45604330e-01 -3.84214014e-01
8.53343844e-01 1.15129340e+00 -1.89765394e-01 -9.84472275e-01
-6.83160126e-02 -1.11308813e-01 -4.32560220e-02 8.42741549e-01
-7.75400996e-02 -2.27181688e-01 -7.37821460e-01 1.35169113e+00
-2.84812953e-02 -1.17844641e-01 2.64920816e-02 -9.72829759e-02
2.33337089e-01 3.94606650e-01 7.60619566e-02 1.86234102e-01
1.01943624e+00 -7.03582406e-01 -4.31115814e-02 -5.34010589e-01
1.12500679e+00 -8.24292600e-01 1.07852519e+00 2.74684846e-01
-7.48330772e-01 -1.18088625e-01 -1.18815637e+00 3.18785757e-01
-6.51144683e-01 -3.79611343e-01 5.48117876e-01 1.40522730e+00
-1.20073712e+00 6.31079257e-01 -8.66075933e-01 -2.28547886e-01
2.79542834e-01 8.36270079e-02 -2.61715263e-01 2.40998656e-01
-9.47205663e-01 7.74469972e-01 1.63685247e-01 -4.58074510e-01
-1.37822294e+00 -1.17651975e+00 -9.37588751e-01 2.51862377e-01
5.53791821e-01 2.57913649e-01 1.44799459e+00 -8.58267307e-01
-5.46707988e-01 8.04094255e-01 3.94163504e-02 -2.59657055e-01
1.56917542e-01 -5.31906724e-01 -6.47363901e-01 -3.43854696e-01
2.34278053e-01 -2.74127454e-01 8.61243546e-01 -7.54243314e-01
-5.84874511e-01 8.28371942e-02 6.95088863e-01 -8.40442240e-01
-7.89153934e-01 9.54300761e-01 3.64974141e-02 -4.54776078e-01
-8.33436847e-01 -7.11238623e-01 -4.25020456e-01 -4.25250232e-01
-4.49844986e-01 3.64324450e-02 7.44507313e-01 -7.97845066e-01
1.98860669e+00 -2.10366368e+00 -1.06811017e-01 2.69762397e-01
4.01115805e-01 6.64069474e-01 -6.47229910e-01 8.05256248e-01
-5.58170557e-01 8.83434176e-01 -6.83274508e-01 -2.52615027e-02
-3.23647447e-02 -3.26604813e-01 -8.51437867e-01 2.82682508e-01
1.77334309e-01 9.09127116e-01 -9.29216743e-01 -1.09832227e-01
-2.70911038e-01 1.80070341e-01 -7.75173128e-01 1.67247102e-01
-3.31119031e-01 -6.77294910e-01 -3.00159663e-01 9.85827982e-01
5.51994979e-01 5.19196801e-02 -1.07927680e-01 6.62509501e-01
-3.71099859e-01 3.51704478e-01 -6.82560325e-01 1.42069757e+00
-7.77854502e-01 7.13961303e-01 -1.96924463e-01 -5.16064465e-01
8.68322074e-01 3.84086490e-01 -1.84862733e-01 -4.42503631e-01
-3.69409353e-01 3.05889279e-01 1.44929498e-01 -7.69299030e-01
2.58773535e-01 4.41082209e-01 -6.61586165e-01 9.89467263e-01
-2.33606875e-01 1.81400612e-01 3.49646628e-01 4.17803168e-01
1.93414664e+00 -1.65489405e-01 3.94685477e-01 -2.67160356e-01
5.92336893e-01 -2.78093871e-02 4.37999636e-01 8.49643290e-01
9.76773053e-02 3.75973582e-02 1.18101835e+00 -7.59197235e-01
-1.16636181e+00 -9.75048244e-01 1.49860144e-01 1.25482035e+00
-5.86108148e-01 -9.88063574e-01 -1.22082150e+00 -1.29030693e+00
-1.32730871e-01 7.80782878e-01 -8.70496631e-01 -6.03461504e-01
-7.76898563e-01 -6.83367431e-01 9.49891269e-01 8.11217606e-01
-1.95903957e-01 -1.13183689e+00 -1.07841265e+00 2.53389806e-01
4.02848214e-01 -4.45353895e-01 -4.67534482e-01 -1.01319820e-01
-3.99200231e-01 -1.19912481e+00 -1.64147496e-01 -5.39598644e-01
6.12545550e-01 5.19445874e-02 1.37016320e+00 7.50669301e-01
-1.11498797e+00 2.72225022e-01 -5.70683718e-01 -4.05521989e-01
-1.12850630e+00 2.57388353e-01 -3.17403197e-01 -7.16572523e-01
6.38366640e-01 -5.58483899e-01 3.27344425e-02 -1.68433696e-01
-9.43524122e-01 -8.37188601e-01 4.90363091e-01 2.64229745e-01
-2.12940220e-02 1.62713915e-01 4.92744863e-01 -1.06847417e+00
8.74882579e-01 -9.12040770e-01 -1.13575757e+00 6.45206511e-01
-5.35405695e-01 4.10512574e-02 6.55939460e-01 -3.57281178e-01
-1.02926826e+00 -6.13521412e-02 -4.05192435e-01 -2.55120471e-02
-3.78041714e-02 8.36872220e-01 1.07161574e-01 -5.63989639e-01
1.10123956e+00 -1.27272859e-01 -3.10082793e-01 -5.25075912e-01
2.97461659e-01 4.49357778e-01 1.75634950e-01 -5.91956556e-01
1.29181719e+00 -8.17370489e-02 -4.15001661e-01 -4.82638031e-01
-5.44448495e-01 -1.39097288e-01 -5.81949830e-01 1.14164233e-01
4.86000031e-01 -5.76665521e-01 -1.55338794e-01 4.44018304e-01
-1.18657136e+00 -3.66038680e-01 7.21490532e-02 -6.13717020e-01
1.06090494e-01 8.25093746e-01 -4.92686957e-01 -7.62709796e-01
-5.95767677e-01 -1.35593438e+00 7.69742668e-01 -4.87711169e-02
-4.66703236e-01 -9.44414377e-01 8.45850170e-01 -3.63154083e-01
7.58660018e-01 4.98505622e-01 1.53983665e+00 -6.04667604e-01
-6.14030063e-01 -7.27917492e-01 -1.00809094e-02 2.51955122e-01
1.87634125e-01 7.13873565e-01 -7.80881345e-01 -4.28638190e-01
-3.59165221e-01 -1.95288658e-01 6.08761787e-01 -1.69359431e-01
8.54013562e-01 -3.39069068e-01 -6.29634023e-01 4.43859190e-01
1.79802442e+00 3.56091261e-01 7.07536519e-01 5.89368939e-01
8.16014051e-01 6.80184305e-01 3.50065470e-01 3.93720508e-01
9.35806409e-02 3.23183239e-01 6.55770838e-01 3.13784122e-01
2.84911424e-01 1.48464143e-01 7.93645084e-01 3.45657587e-01
3.03853989e-01 1.22757539e-01 -1.72926259e+00 8.82712483e-01
-1.56360281e+00 -9.67954457e-01 -1.77986115e-01 2.30708885e+00
8.43928933e-01 4.32345569e-01 1.14895277e-01 3.90504301e-02
4.79592264e-01 2.79244017e-02 -3.90069902e-01 -1.16182756e+00
5.03978074e-01 2.13726178e-01 2.73591816e-01 6.91102803e-01
-1.16877747e+00 9.01361585e-01 6.85024357e+00 5.77694952e-01
-1.06271088e+00 2.36549556e-01 1.28621265e-01 -1.09073341e-01
-6.60863042e-01 3.49533528e-01 -9.93974984e-01 2.99704880e-01
1.46155035e+00 -5.85614741e-01 2.99699217e-01 1.47410452e+00
-4.32219476e-01 1.63685307e-01 -1.26212692e+00 2.76203334e-01
1.25705525e-01 -1.30803716e+00 -1.35310426e-01 -3.51997055e-02
6.54726982e-01 1.92002997e-01 1.25364870e-01 5.91603279e-01
4.58426207e-01 -1.01985729e+00 8.12877297e-01 2.95693964e-01
6.71840131e-01 -1.01433909e+00 6.71483874e-01 1.57979980e-01
-1.44508553e+00 -7.12592959e-01 -4.84774202e-01 -4.47478108e-02
-6.99611828e-02 5.43651581e-01 -7.11242795e-01 1.30667806e-01
8.56470048e-01 4.96438533e-01 -1.27756405e+00 9.79243279e-01
-2.87291944e-01 6.01505280e-01 1.77404433e-01 8.58791322e-02
1.79280996e-01 5.69985092e-01 5.11754155e-01 1.58008540e+00
3.07328790e-01 -4.53514338e-01 1.71669826e-01 1.30237114e+00
3.08903128e-01 -1.48223564e-01 -7.97063589e-01 -4.32232797e-01
5.38525045e-01 1.26759660e+00 -6.15959108e-01 -2.24739596e-01
-8.85178685e-01 4.29281265e-01 5.68048477e-01 1.03943728e-01
-7.77208626e-01 -1.02394509e+00 9.62931991e-01 4.39612716e-02
2.42019817e-01 -1.56846926e-01 -6.79040700e-02 -1.10728490e+00
4.25611377e-01 -1.09690630e+00 3.80329162e-01 -5.42584546e-02
-1.14142442e+00 9.29229617e-01 8.01846385e-02 -1.00197923e+00
-3.83102775e-01 -4.66343373e-01 -1.45560575e+00 1.00408816e+00
-1.24615729e+00 -9.42094266e-01 1.09013334e-01 4.22860645e-02
5.79870105e-01 -6.10193372e-01 9.51686978e-01 3.60694200e-01
-7.18693972e-01 1.08431911e+00 -7.42788464e-02 4.18521583e-01
4.29496199e-01 -1.35103059e+00 1.72676980e+00 1.57290602e+00
-1.55308068e-01 1.31132185e+00 4.64115858e-01 -1.20651197e+00
-1.15935349e+00 -1.20423782e+00 8.29075754e-01 -1.11055601e+00
1.22517681e+00 -8.14355075e-01 -1.35244751e+00 8.38388860e-01
1.45328179e-01 -1.45438448e-01 7.69153476e-01 1.35018766e-01
-1.18015337e+00 3.36669743e-01 -1.04607129e+00 5.81358671e-01
5.87167084e-01 -9.69570160e-01 -6.35799706e-01 1.97376773e-01
9.98947620e-01 3.30976099e-02 -7.07770884e-01 1.03525311e-01
4.48736459e-01 -1.16963673e+00 9.35936689e-01 -5.97759247e-01
7.69719839e-01 -1.03358336e-01 1.20303137e-02 -9.35353518e-01
-3.15830439e-01 -8.84427130e-01 -2.62972951e-01 1.44577074e+00
7.54916191e-01 -6.12334132e-01 3.06946903e-01 7.97426164e-01
-1.17822982e-01 -5.96975267e-01 -6.74378097e-01 -8.20986867e-01
6.15831912e-01 -5.62961936e-01 7.89300919e-01 8.99545133e-01
5.49051702e-01 -4.34462041e-01 -1.44583220e-02 3.72329235e-01
6.57858551e-01 -7.59965880e-03 4.05241191e-01 -1.08394122e+00
-5.95167756e-01 -6.74897254e-01 -4.49296653e-01 1.62671506e-01
4.65615153e-01 -7.76700318e-01 -9.37518105e-02 -9.49527562e-01
5.30679703e-01 -2.75596082e-02 -2.75119454e-01 9.58925188e-01
-4.94617671e-01 -2.14327171e-01 -1.67454109e-01 -1.42134577e-01
-3.96692425e-01 -4.11073446e-01 -1.87344775e-01 -4.14805412e-01
-9.98423323e-02 -6.31462634e-02 -7.93833137e-01 8.93837035e-01
7.65821397e-01 -9.24375474e-01 -4.31765288e-01 -4.44676399e-01
1.01194704e+00 -9.80992764e-02 1.65043175e-01 -8.69396150e-01
-6.51394650e-02 -4.50272448e-02 -3.48354906e-01 -9.35241207e-02
-4.20187414e-01 -3.61643195e-01 -2.30093136e-01 7.50475645e-01
-4.83102590e-01 5.27716517e-01 7.59492815e-01 3.35069567e-01
1.58339664e-01 -9.09765303e-01 6.65269136e-01 -4.37703311e-01
-9.43174183e-01 3.24877322e-01 -8.15111935e-01 2.07070708e-01
1.28176093e+00 6.40827045e-02 -7.40027130e-01 3.71102601e-01
-2.36691877e-01 -5.86225055e-02 8.80954266e-01 1.12859917e+00
6.03361666e-01 -7.49859273e-01 -5.76099992e-01 2.80657291e-01
5.07411301e-01 -8.18653345e-01 3.17255318e-01 3.23702633e-01
-5.11538625e-01 2.21498162e-01 -2.70920277e-01 -4.16638749e-03
-1.67449808e+00 1.20868945e+00 2.95748115e-01 -2.07344502e-01
-7.70635426e-01 9.72427189e-01 -3.04712411e-02 -4.20801044e-01
1.26929641e-01 -2.07922906e-01 -1.99590907e-01 -1.86245456e-01
1.22269022e+00 3.98973376e-01 1.19209446e-01 -2.93895572e-01
-7.23044574e-01 7.75496304e-01 -7.46670842e-01 2.38757685e-01
1.31349146e+00 4.73189414e-01 -7.30390370e-01 2.27178439e-01
1.30855370e+00 3.59363884e-01 -8.34525287e-01 2.03133672e-01
6.14621043e-01 -6.12895191e-01 -3.44943196e-01 -9.60296929e-01
-8.80648673e-01 1.12433624e+00 5.76089501e-01 4.67550278e-01
7.86462545e-01 7.21282884e-02 6.26689196e-01 6.51331604e-01
6.47372067e-01 -5.39216578e-01 1.91743985e-01 5.78238964e-01
6.96441054e-01 -1.13474786e+00 -2.03319177e-01 -1.87307149e-01
-1.47193596e-01 1.29670191e+00 8.18674088e-01 -3.38564545e-01
7.31412649e-01 1.05949199e+00 1.01434156e-01 -2.98631340e-01
-8.40855896e-01 4.19245780e-01 1.70370676e-02 7.26469576e-01
6.37696683e-01 2.58145090e-02 1.37957528e-01 5.32035768e-01
1.14679314e-01 -5.07996678e-01 1.15850735e+00 1.39335501e+00
-7.14275777e-01 -1.52110302e+00 -3.92430246e-01 4.01558191e-01
-9.06031966e-01 -7.47803211e-01 -6.22995496e-01 4.65122849e-01
6.19650185e-02 9.76414800e-01 -4.66119260e-01 -4.87087905e-01
2.44035363e-01 -2.30698343e-02 8.17660391e-02 -1.28638697e+00
-1.15396309e+00 -4.46589857e-01 8.72958452e-02 -8.79275739e-01
4.56177384e-01 -6.22293472e-01 -1.00928688e+00 -8.69121253e-02
-5.72560392e-02 5.79129867e-02 3.56365234e-01 7.27191269e-01
4.92785752e-01 5.78476906e-01 4.64639395e-01 -5.57763159e-01
-7.68906713e-01 -5.31955421e-01 -1.40622020e-01 1.01982512e-01
6.82125747e-01 -3.95888001e-01 -5.12137115e-01 1.67748198e-01]
|
[7.112658500671387, 7.785399913787842]
|
85bf0592-6b6b-44d5-8746-40f82c28321e
|
docbank-a-benchmark-dataset-for-document
|
2006.01038
| null |
https://arxiv.org/abs/2006.01038v3
|
https://arxiv.org/pdf/2006.01038v3.pdf
|
DocBank: A Benchmark Dataset for Document Layout Analysis
|
Document layout analysis usually relies on computer vision models to understand documents while ignoring textual information that is vital to capture. Meanwhile, high quality labeled datasets with both visual and textual information are still insufficient. In this paper, we present \textbf{DocBank}, a benchmark dataset that contains 500K document pages with fine-grained token-level annotations for document layout analysis. DocBank is constructed using a simple yet effective way with weak supervision from the \LaTeX{} documents available on the arXiv.com. With DocBank, models from different modalities can be compared fairly and multi-modal approaches will be further investigated and boost the performance of document layout analysis. We build several strong baselines and manually split train/dev/test sets for evaluation. Experiment results show that models trained on DocBank accurately recognize the layout information for a variety of documents. The DocBank dataset is publicly available at \url{https://github.com/doc-analysis/DocBank}.
|
['Furu Wei', 'Yiheng Xu', 'Zhoujun Li', 'Shaohan Huang', 'Ming Zhou', 'Minghao Li', 'Lei Cui']
|
2020-06-01
| null |
https://aclanthology.org/2020.coling-main.82
|
https://aclanthology.org/2020.coling-main.82.pdf
|
coling-2020-8
|
['document-layout-analysis']
|
['computer-vision']
|
[-2.13357046e-01 -1.08633690e-01 -2.67683089e-01 -3.44394058e-01
-1.18250453e+00 -1.22313070e+00 7.59040654e-01 2.27363870e-01
2.73392219e-02 3.09793621e-01 5.35247326e-01 -3.84882867e-01
4.68727425e-02 -4.42520380e-01 -6.82181418e-01 -6.31486356e-01
2.42150024e-01 5.47777116e-01 7.91411498e-04 2.66341269e-01
5.87227881e-01 2.83750206e-01 -1.03702497e+00 1.14747620e+00
6.45722508e-01 9.24533963e-01 4.89009947e-01 1.11799800e+00
-5.82780361e-01 6.80934310e-01 -7.31569409e-01 -5.44116378e-01
8.71095508e-02 -1.97526053e-01 -9.54533219e-01 3.86034310e-01
1.01033056e+00 -3.44429702e-01 -4.97347027e-01 1.11373937e+00
5.06926656e-01 -2.58978903e-01 8.13134789e-01 -8.78491759e-01
-1.29993784e+00 9.07397926e-01 -8.93082976e-01 1.29293963e-01
4.35136139e-01 3.22383903e-02 1.27511704e+00 -1.21159732e+00
8.04030538e-01 1.54013407e+00 4.28088933e-01 2.35804409e-01
-8.33779156e-01 -9.72155854e-02 1.56801969e-01 2.86723822e-01
-1.06444025e+00 -5.30538142e-01 8.99325550e-01 -5.50782740e-01
7.58537412e-01 2.65991211e-01 1.23448811e-01 1.37811291e+00
9.34404582e-02 1.24870145e+00 1.28599036e+00 -1.08921754e+00
-1.31786749e-01 9.43086371e-02 6.38318241e-01 1.13105249e+00
1.94377765e-01 -5.31650662e-01 -5.77907681e-01 3.67953718e-01
4.33784336e-01 -1.20062925e-01 -3.65996063e-01 -3.63764167e-01
-1.28449750e+00 4.33301300e-01 3.28328669e-01 3.13662976e-01
2.33639032e-01 -2.77627800e-02 8.03662121e-01 -5.66562153e-02
1.91171557e-01 3.29829305e-01 -3.11026037e-01 -2.00091958e-01
-9.99459922e-01 2.91401427e-02 5.76118231e-01 1.30043769e+00
3.98897916e-01 -1.60179138e-01 -4.42031741e-01 1.11505914e+00
5.43721497e-01 8.37518513e-01 2.83129901e-01 -9.20818269e-01
1.25345743e+00 7.52762318e-01 -1.16244875e-01 -8.10975730e-01
-3.29202771e-01 -9.92543623e-02 -8.23141456e-01 -8.03629011e-02
7.38001406e-01 7.81704709e-02 -1.02649105e+00 7.94863701e-01
4.14577406e-03 -6.79854751e-01 -5.58939390e-02 5.86313665e-01
1.11382210e+00 5.92385948e-01 -2.79159307e-01 1.97439656e-01
1.68214774e+00 -1.48710918e+00 -9.85696793e-01 -1.85902044e-01
9.63411272e-01 -1.26703632e+00 1.69148159e+00 7.81401098e-01
-1.10076153e+00 -5.91173589e-01 -1.11416471e+00 -4.83384609e-01
-7.76793420e-01 8.17242682e-01 3.03100199e-01 7.27666020e-01
-7.87274718e-01 1.04385301e-01 -7.17085660e-01 -5.06312907e-01
8.85853767e-01 -4.37718481e-01 -4.18676704e-01 -5.89255810e-01
-6.68542802e-01 5.17898381e-01 3.34118217e-01 2.05422848e-01
-8.79624009e-01 -4.35644478e-01 -7.82171726e-01 6.42983988e-02
3.44825357e-01 -2.33065054e-01 1.33497131e+00 -4.64393884e-01
-1.09360766e+00 1.02612603e+00 -2.64231302e-02 -9.20447558e-02
5.72615087e-01 -4.45957839e-01 -3.69081706e-01 5.41345417e-01
5.65253906e-02 3.99438769e-01 6.97664261e-01 -1.38168931e+00
-2.15645790e-01 -6.22091413e-01 6.78574592e-02 -1.26239747e-01
-7.57257640e-01 1.00222394e-01 -1.12208259e+00 -6.94041669e-01
-1.28202617e-01 -8.45586777e-01 4.91180658e-01 2.37719771e-02
-1.15662861e+00 -1.99572116e-01 1.13967717e+00 -9.71418977e-01
1.26227641e+00 -1.85724211e+00 -7.01021999e-02 -3.51997316e-02
-5.86865842e-02 2.06009045e-01 -3.79703969e-01 7.35722780e-01
1.52980193e-01 2.58376956e-01 -2.88574733e-02 -5.89673162e-01
3.94371003e-01 -1.67079031e-01 -4.37758327e-01 4.13117826e-01
-3.63083273e-01 1.22935939e+00 -3.68659556e-01 -8.87594223e-01
1.70657322e-01 3.61081213e-01 -3.10076505e-01 4.29717563e-02
-5.03610313e-01 2.01992299e-02 -4.32109326e-01 1.09726298e+00
7.86555052e-01 -4.43130940e-01 4.58099097e-01 -5.16910553e-01
1.88760378e-03 -6.97590858e-02 -1.02176201e+00 2.09736228e+00
-2.89399147e-01 1.13803136e+00 8.25279802e-02 -9.28265333e-01
8.24293315e-01 2.97749359e-02 -1.06237769e-01 -9.86944497e-01
1.53336182e-01 -1.70126975e-01 -3.38871628e-01 -8.56327891e-01
7.62425542e-01 7.59201825e-01 -2.52311826e-01 4.57944155e-01
-1.00225039e-01 -2.11513937e-02 7.07254589e-01 6.89004183e-01
1.02002394e+00 3.89378488e-01 -2.34437406e-01 -1.62920266e-01
4.85643506e-01 2.28162959e-01 -7.45704025e-02 8.07654440e-01
-4.63929921e-02 9.09632921e-01 7.32456207e-01 -1.24697946e-01
-1.42007351e+00 -9.95490789e-01 -2.12641776e-01 1.19314981e+00
-1.17770940e-01 -8.83265376e-01 -1.04415023e+00 -8.43058825e-01
-7.84060806e-02 5.21495819e-01 -6.33786023e-01 5.72243512e-01
-5.33511639e-01 -5.11288643e-01 9.61187780e-01 9.03394282e-01
5.19034684e-01 -1.12110186e+00 -1.48612574e-01 -4.31403995e-01
-2.58898795e-01 -1.31389749e+00 -3.60437185e-01 3.26173822e-03
-5.12177467e-01 -1.17012489e+00 -8.54653180e-01 -7.22162962e-01
8.72075319e-01 4.20755178e-01 1.06022978e+00 1.21460110e-01
-5.72491348e-01 6.18990183e-01 -5.45093358e-01 -4.85301226e-01
-2.02704772e-01 6.80407360e-02 -3.90952229e-01 -4.85046774e-01
2.75148332e-01 1.74806893e-01 -6.00467801e-01 -4.91168089e-02
-9.04473603e-01 -4.30213399e-02 6.77973211e-01 6.71688855e-01
4.77340698e-01 1.13087751e-01 4.67872620e-02 -1.18057096e+00
6.31391704e-01 2.69173700e-02 -5.28243542e-01 7.42554307e-01
-3.59114379e-01 -9.64510739e-02 6.97298586e-01 7.75714219e-02
-1.39086425e+00 -3.21264416e-01 2.48615101e-01 -3.46943647e-01
-4.46237713e-01 6.79494202e-01 -4.17636395e-01 5.41575193e-01
5.05193830e-01 9.68081728e-02 -3.80584717e-01 -1.05288243e+00
7.62970448e-01 9.90398288e-01 6.98135138e-01 -8.64602149e-01
6.73960686e-01 6.04686081e-01 -3.18153262e-01 -9.44067121e-01
-1.30021620e+00 -4.45047498e-01 -1.14619398e+00 -2.83596545e-01
6.51963651e-01 -6.30538583e-01 -5.17084956e-01 6.48022115e-01
-1.15560341e+00 -4.45709676e-01 2.36567691e-01 9.29077864e-02
-2.80507773e-01 7.75745451e-01 -6.94654763e-01 -5.27539253e-01
-4.36927199e-01 -8.85179758e-01 1.36533403e+00 9.60374996e-02
-1.30284615e-02 -1.09998202e+00 8.29356387e-02 1.05283713e+00
-2.43821591e-01 -1.42189404e-02 1.13595366e+00 -3.49395216e-01
-6.68974102e-01 -4.72648233e-01 -7.04307616e-01 2.53661871e-01
-4.26093303e-02 6.62918746e-01 -1.20001233e+00 -3.47787470e-01
-8.66885483e-01 -5.89275956e-01 1.12388957e+00 4.18834060e-01
1.61840665e+00 -1.82683080e-01 -5.48642814e-01 2.42650375e-01
1.49151790e+00 -4.04123068e-02 8.10250223e-01 6.58567071e-01
1.18662262e+00 7.20839798e-01 6.32823884e-01 4.70015258e-01
4.12166923e-01 4.27303106e-01 2.55249143e-01 8.02239403e-02
-6.26346469e-01 -1.30705342e-01 3.60549241e-01 1.00101805e+00
1.93327218e-01 -9.60846126e-01 -1.43772900e+00 5.25937259e-01
-1.79764116e+00 -1.02500653e+00 -3.18905741e-01 1.53069830e+00
7.31705010e-01 3.14842403e-01 -1.23492472e-01 2.91637868e-01
5.56908607e-01 4.16989833e-01 -7.86230564e-02 -4.94581193e-01
-4.00754631e-01 -8.80601332e-02 2.89239526e-01 3.30828398e-01
-1.38281143e+00 8.59013438e-01 5.47050190e+00 9.04054105e-01
-7.50970781e-01 -3.38354707e-02 4.66153741e-01 -1.12180702e-01
-3.43428433e-01 -1.11818708e-01 -1.08349812e+00 4.90967393e-01
8.33620369e-01 3.01202089e-01 1.67999938e-02 7.83828080e-01
2.07807705e-01 -2.03087077e-01 -1.09501088e+00 8.45603824e-01
4.18639302e-01 -1.67018044e+00 6.70161098e-02 1.12539545e-01
7.07823873e-01 -1.41185150e-01 3.44193041e-01 1.88490793e-01
1.50678381e-01 -1.02678311e+00 9.85173166e-01 7.28683114e-01
7.73019671e-01 -4.86537904e-01 4.59597737e-01 1.22693509e-01
-1.24703515e+00 2.24047694e-02 -4.31666672e-01 3.50015014e-01
-7.95161277e-02 6.78612828e-01 -9.12323833e-01 6.49788857e-01
9.81011868e-01 9.28202629e-01 -1.26397228e+00 1.09556627e+00
-3.81510109e-01 5.56773663e-01 2.38211811e-01 -8.49188045e-02
2.75709957e-01 -1.23109356e-01 1.62800536e-01 1.71708691e+00
2.39435598e-01 -3.71783584e-01 2.46083349e-01 5.17495394e-01
-2.17026711e-01 1.62651330e-01 -5.16612470e-01 -5.96081913e-01
4.26495910e-01 1.54689288e+00 -9.95025694e-01 -4.15296495e-01
-6.67985857e-01 8.08587074e-01 5.02337754e-01 5.56249380e-01
-6.86936557e-01 -5.71600437e-01 -1.37568638e-01 -7.93439820e-02
4.70443308e-01 -5.98225772e-01 -4.66674656e-01 -1.17470467e+00
2.40670651e-01 -9.19896007e-01 5.12785852e-01 -1.23896241e+00
-1.18577754e+00 3.66624206e-01 -2.23029241e-01 -1.21145475e+00
2.78020322e-01 -1.16196358e+00 -3.73248607e-01 5.09014785e-01
-1.44336998e+00 -1.66483200e+00 -4.73875433e-01 2.54563510e-01
9.43453670e-01 -2.45921239e-01 7.90608287e-01 2.39079416e-01
-1.06255639e+00 5.88144422e-01 7.02169836e-01 7.26191521e-01
1.08950841e+00 -1.64781177e+00 2.52992660e-01 7.76279271e-01
4.67396230e-01 7.03425467e-01 4.77866709e-01 -5.73359311e-01
-1.63505018e+00 -1.04955769e+00 6.30261540e-01 -9.29613769e-01
6.66784644e-01 -7.45797813e-01 -9.28580523e-01 8.84623230e-01
9.48403716e-01 -1.61211312e-01 6.72279835e-01 2.16138169e-01
-7.91952312e-01 2.00156078e-01 -6.69499278e-01 5.36918879e-01
7.32472956e-01 -8.01295161e-01 -6.82580531e-01 7.13138700e-01
2.52735794e-01 -5.45625448e-01 -8.15004289e-01 -1.36554733e-01
6.57025933e-01 -8.19828093e-01 1.06629479e+00 -4.82340187e-01
9.96780694e-01 -2.62811661e-01 -2.50086963e-01 -9.39083278e-01
-4.99142595e-02 7.69406382e-04 -3.89058650e-01 1.72190452e+00
3.87063622e-01 3.91746387e-02 7.84123182e-01 1.59445763e-01
-2.84103066e-01 -5.00275254e-01 -3.10271293e-01 -8.33229661e-01
2.46869445e-01 -4.87494260e-01 2.55946666e-01 9.60295439e-01
2.42975414e-01 3.30875516e-01 -2.05057263e-01 2.13041469e-01
7.53782749e-01 4.77584511e-01 8.09647739e-01 -9.47911024e-01
-3.11728790e-02 -5.13900995e-01 1.29621521e-01 -1.08493078e+00
2.52030224e-01 -1.16292894e+00 -1.71762362e-01 -2.01565599e+00
5.46789706e-01 -3.16003338e-02 -1.14033505e-01 5.98082721e-01
2.04097897e-01 3.43607992e-01 1.96327984e-01 3.06687683e-01
-1.09255719e+00 3.78192723e-01 1.46397579e+00 -8.24638307e-01
3.23521435e-01 -7.20228672e-01 -3.60660851e-01 8.56174052e-01
7.57425725e-01 4.03363109e-02 -2.53637373e-01 -7.56182015e-01
7.84983635e-02 -1.96785167e-01 2.90647835e-01 -6.49281442e-01
1.88992426e-01 1.53363580e-02 9.87670302e-01 -1.36178362e+00
2.74437606e-01 -3.53637785e-01 -7.45734811e-01 -1.29330726e-02
-5.50819755e-01 6.93244189e-02 4.13860202e-01 5.14307976e-01
-1.19954139e-01 -4.99548376e-01 5.79443455e-01 -2.17687145e-01
-8.91221762e-01 -1.16197849e-02 -1.98042706e-01 1.36980012e-01
6.82877898e-01 -1.49823695e-01 -1.22675598e+00 7.02386349e-02
-6.34705365e-01 2.07803622e-01 4.98276353e-01 6.51591897e-01
5.71883082e-01 -1.25010395e+00 -4.18195397e-01 -2.34690428e-01
4.67977405e-01 -2.09712192e-01 4.47748482e-01 6.95672691e-01
-8.64833117e-01 1.15934134e+00 -2.00327396e-01 -5.73817432e-01
-1.56258559e+00 6.08703911e-01 -1.57629058e-01 -1.64793521e-01
-7.17658341e-01 8.57154191e-01 1.86813831e-01 -4.82944846e-01
4.29973930e-01 -3.26706856e-01 -4.21319842e-01 4.59750056e-01
7.71461725e-01 4.26679939e-01 2.19592199e-01 -4.52432066e-01
-3.85516673e-01 6.70120299e-01 -4.76738095e-01 -8.80630165e-02
1.21343935e+00 -3.62130165e-01 -1.41893268e-01 3.32038075e-01
1.29004037e+00 1.87549591e-01 -1.01857424e+00 -2.57568896e-01
2.95327902e-01 -6.65335894e-01 -2.41524279e-02 -1.03042877e+00
-9.07776952e-01 1.29249704e+00 4.90905583e-01 4.46753234e-01
7.54414558e-01 5.42637371e-02 5.94310999e-01 6.33880317e-01
-1.55748695e-01 -1.52496004e+00 7.42135584e-01 5.55301309e-01
1.20118403e+00 -1.29204249e+00 2.86875427e-01 -3.22154969e-01
-6.33915424e-01 1.42898965e+00 6.40988052e-01 2.41521195e-01
3.18586081e-01 4.79587674e-01 3.50885808e-01 -4.21709746e-01
-6.15991473e-01 2.91479360e-02 5.30542016e-01 4.91929650e-01
8.38371336e-01 -1.20482527e-01 5.79728074e-02 4.86322969e-01
-2.11303174e-01 -4.34735388e-01 6.82779670e-01 1.07564294e+00
-4.66824651e-01 -1.22030509e+00 -7.46889353e-01 3.15379292e-01
-3.68328065e-01 -2.01287702e-01 -6.57092631e-01 7.47806966e-01
-3.71059716e-01 8.34405065e-01 8.84599239e-02 3.22177649e-01
1.44858032e-01 2.55988777e-01 7.99084842e-01 -5.36804676e-01
2.59707868e-03 3.40762854e-01 2.84962714e-01 -3.54202688e-01
-2.88267344e-01 -7.64149427e-01 -1.32023752e+00 -2.27571547e-01
-3.84697020e-02 -1.22706361e-01 6.77002549e-01 7.50738859e-01
1.95786923e-01 8.35691869e-01 1.47683486e-01 -6.86995685e-01
-2.27816790e-01 -9.88136530e-01 -6.52820706e-01 3.57431173e-01
2.25713089e-01 -3.89521509e-01 -5.71501926e-02 6.89536810e-01]
|
[11.637399673461914, 2.407153606414795]
|
af7ef0d8-21de-4add-90b8-a6e3bee67521
|
enhanced-low-resolution-lidar-camera
|
2211.03932
| null |
https://arxiv.org/abs/2211.03932v1
|
https://arxiv.org/pdf/2211.03932v1.pdf
|
Enhanced Low-resolution LiDAR-Camera Calibration Via Depth Interpolation and Supervised Contrastive Learning
|
Motivated by the increasing application of low-resolution LiDAR recently, we target the problem of low-resolution LiDAR-camera calibration in this work. The main challenges are two-fold: sparsity and noise in point clouds. To address the problem, we propose to apply depth interpolation to increase the point density and supervised contrastive learning to learn noise-resistant features. The experiments on RELLIS-3D demonstrate that our approach achieves an average mean absolute rotation/translation errors of 0.15cm/0.33\textdegree on 32-channel LiDAR point cloud data, which significantly outperforms all reference methods.
|
['Fengbo Ren', 'Sanjeev Agarwal', 'Raghuveer Rao', 'Suya You', 'Zifan Yu', 'Zhikang Zhang']
|
2022-11-08
| null | null | null | null |
['camera-calibration']
|
['computer-vision']
|
[ 1.56349361e-01 -3.83123994e-01 -3.63711774e-01 -4.27385807e-01
-1.12840116e+00 -1.02511607e-01 3.42809170e-01 -2.40313441e-01
-4.78977680e-01 8.62891793e-01 -2.86993504e-01 -1.28045946e-01
-3.65667343e-02 -6.74271047e-01 -7.11470306e-01 -4.57194090e-01
-7.23091885e-02 6.94756150e-01 2.92347759e-01 1.66245908e-01
3.93162459e-01 1.07040954e+00 -1.61667538e+00 -2.99534440e-01
1.00406575e+00 9.98090804e-01 3.09753269e-01 2.26275980e-01
-1.99782923e-01 2.38584742e-01 -5.98477535e-02 -4.17214222e-02
6.49589360e-01 3.65799665e-01 -2.94271559e-01 3.64146940e-02
1.10496593e+00 -4.64394420e-01 -3.32185656e-01 1.12139821e+00
5.05637944e-01 -1.32981703e-01 6.48134768e-01 -1.24345851e+00
-3.18947941e-01 1.05113670e-01 -1.13374925e+00 1.57600835e-01
6.47488190e-03 -6.21414967e-02 6.23932958e-01 -1.53871441e+00
3.10197085e-01 1.39884496e+00 9.37045336e-01 3.04312021e-01
-1.25738955e+00 -1.29048395e+00 -8.85556489e-02 -4.98834662e-02
-2.07653666e+00 -4.19064701e-01 7.81624436e-01 -5.72601855e-01
6.23832524e-01 -1.86845496e-01 4.72122878e-01 6.78788781e-01
5.47901168e-02 1.48789123e-01 1.15616262e+00 -2.78244257e-01
1.33432791e-01 -7.43616894e-02 -2.44704038e-01 5.46569228e-01
8.58787239e-01 4.08216000e-01 -5.19981802e-01 -2.28274718e-01
1.16526949e+00 1.40334934e-01 -2.79915091e-02 -5.02386272e-01
-9.80087698e-01 1.00723255e+00 7.10848689e-01 -1.46841899e-01
-1.50526971e-01 2.59917289e-01 -6.89191918e-04 -2.85063654e-01
3.73671561e-01 -2.37231497e-02 -3.29936385e-01 4.99084257e-02
-9.66072381e-01 2.28504553e-01 2.30867580e-01 1.58449888e+00
1.36798751e+00 2.37236649e-01 5.85313618e-01 7.62771666e-01
6.69706941e-01 1.17754114e+00 -6.02888204e-02 -1.16168678e+00
8.28806758e-01 4.62716520e-01 5.56801260e-02 -8.47485662e-01
-1.79920107e-01 -2.01524585e-01 -1.19461763e+00 4.07223821e-01
-8.24499577e-02 5.89320995e-02 -9.61905658e-01 1.04670513e+00
3.62647504e-01 5.36572933e-01 -1.57826453e-01 9.67380822e-01
7.55634010e-01 3.79954606e-01 -1.23533033e-01 -3.03525537e-01
7.08531201e-01 -4.47671652e-01 -5.27740359e-01 -5.32035112e-01
2.48020858e-01 -8.88587832e-01 1.00914550e+00 2.06287831e-01
-9.01320457e-01 -7.66363919e-01 -1.19071662e+00 -1.12580672e-01
7.57253394e-02 8.34558383e-02 4.06098783e-01 3.81413221e-01
-5.56479990e-01 5.11869490e-01 -7.40479946e-01 3.44230607e-03
6.33599699e-01 4.56504583e-01 -2.24681213e-01 -3.65971893e-01
-6.35815382e-01 7.12701023e-01 9.33815986e-02 1.73358038e-01
-3.42038661e-01 -8.36767852e-01 -9.12763357e-01 -3.05333734e-01
3.62463556e-02 -5.88897765e-01 9.95450377e-01 -5.02135307e-02
-1.22732246e+00 8.04558158e-01 -4.23741937e-01 -3.25388551e-01
5.34183681e-01 -6.21224463e-01 -2.37983987e-01 -1.99374352e-02
4.37021613e-01 7.49175668e-01 6.92802489e-01 -1.57635021e+00
-8.55083227e-01 -6.66904747e-01 -6.65770531e-01 6.28789514e-02
1.30538136e-01 -3.47037882e-01 -3.92953157e-01 -3.29861701e-01
8.04403067e-01 -1.09408474e+00 -5.07753432e-01 2.66705930e-01
-1.15732975e-01 1.55628353e-01 1.28430068e+00 5.67282364e-02
7.34634876e-01 -2.09204292e+00 -3.17868143e-01 2.73977578e-01
4.72218059e-02 2.07503382e-02 8.22970495e-02 4.12956811e-03
3.25967483e-02 1.30861670e-01 -4.33295906e-01 -5.06332457e-01
-4.68746424e-01 5.33272743e-01 -5.67408204e-01 6.90749884e-01
3.61809969e-01 6.34317636e-01 -7.26495504e-01 -7.81770408e-01
6.09660268e-01 7.27381110e-01 -2.81114668e-01 4.69223484e-02
2.31885925e-01 6.16291881e-01 -4.72009927e-01 1.05164576e+00
1.49916828e+00 2.06593666e-02 -3.77158999e-01 -2.03914389e-01
-4.03110623e-01 1.81211457e-01 -1.57269192e+00 1.75789189e+00
-3.39584768e-01 5.24221241e-01 7.07326531e-02 -4.03442115e-01
1.33855879e+00 -2.44377673e-01 7.04379320e-01 -4.86542553e-01
-6.93339333e-02 3.64409149e-01 -3.49124640e-01 -6.44415915e-02
6.44541502e-01 -3.24224681e-01 7.30700120e-02 -2.12436765e-01
-2.56975055e-01 -8.33134174e-01 -5.06157756e-01 -1.93565756e-01
6.19236290e-01 7.52797201e-02 2.68515736e-01 -2.68299967e-01
5.29154181e-01 8.45820382e-02 8.64911914e-01 5.00198841e-01
-4.21299599e-03 8.89698923e-01 -2.79985368e-01 -3.40265185e-01
-1.02409124e+00 -1.36402464e+00 -5.91908216e-01 1.32339507e-01
3.15376729e-01 -3.54588598e-01 -9.62478444e-02 -3.04840207e-01
4.84142333e-01 2.40816042e-01 -2.86505297e-02 2.91057229e-01
-9.90134776e-01 -3.68733525e-01 2.20878333e-01 8.36555123e-01
5.98926783e-01 -2.70718485e-01 -5.07496297e-01 -2.48199403e-02
1.30544022e-01 -1.63136411e+00 -1.84152976e-01 1.73477054e-01
-1.51106608e+00 -8.52368474e-01 -2.55353838e-01 -6.19355142e-01
6.84786916e-01 5.39598227e-01 1.05666792e+00 -1.57604530e-01
-3.20472479e-01 1.78519771e-01 5.79332151e-02 -4.82708991e-01
3.08411926e-01 6.11855797e-02 4.29555684e-01 -4.55802381e-01
4.91320014e-01 -8.20393741e-01 -4.10848767e-01 3.10528368e-01
-3.96004915e-01 -3.69849950e-01 7.31147885e-01 6.55790508e-01
1.24197686e+00 -1.46466702e-01 -6.16791211e-02 -6.53405845e-01
1.00941829e-01 -1.58274069e-01 -1.24924409e+00 -3.79062712e-01
-9.29082274e-01 -1.25358537e-01 1.85650289e-01 -4.39969927e-01
-5.81583977e-01 8.25487375e-01 4.55612410e-03 -1.07775855e+00
-1.26995161e-01 2.23089993e-01 -1.56687036e-01 -5.83455205e-01
5.65280676e-01 1.09760799e-01 -3.69702607e-01 -4.29797500e-01
3.22116017e-01 6.84766650e-01 7.93195188e-01 -7.26591408e-01
1.58355188e+00 7.57009923e-01 5.71614921e-01 -1.26186335e+00
-6.18043959e-01 -6.76719546e-01 -1.13068235e+00 7.48000294e-02
5.60257971e-01 -1.59796655e+00 -5.88042438e-01 6.86114505e-02
-1.20195913e+00 1.01103954e-01 -4.79900241e-01 7.03437030e-01
-5.03628671e-01 5.38750231e-01 -9.16102082e-02 -9.06012893e-01
-2.09814444e-01 -1.10187888e+00 1.39328229e+00 1.53157875e-01
1.79961547e-01 -4.29525465e-01 1.68012261e-01 1.17073461e-01
1.69012651e-01 1.26374125e-01 4.46863949e-01 2.36312985e-01
-1.31599951e+00 -7.13941976e-02 -4.84634191e-01 3.80201399e-01
8.91776979e-02 8.75375196e-02 -1.01158154e+00 -3.56111616e-01
-4.46240194e-02 -2.57804543e-01 5.96448362e-01 4.61151809e-01
1.14255786e+00 1.03914432e-01 -4.66058314e-01 1.08971322e+00
1.82046008e+00 -1.37161657e-01 6.74697638e-01 2.15297550e-01
7.57330060e-01 9.11206082e-02 1.05353534e+00 5.07159412e-01
3.96674663e-01 6.97100222e-01 6.41632080e-01 4.02222462e-02
8.90768394e-02 -6.18904710e-01 -1.53999135e-01 8.26873779e-01
-1.83418155e-01 6.44056499e-01 -1.15239906e+00 3.69191915e-01
-1.61546004e+00 -6.79453254e-01 -5.27688503e-01 2.25799966e+00
4.37211335e-01 2.22871840e-01 -1.91603735e-01 -1.76231489e-02
5.38930774e-01 1.96084544e-01 -5.12386262e-01 4.61755171e-02
-5.08088805e-02 4.61901635e-01 1.10855079e+00 7.26501703e-01
-1.04774404e+00 1.01246202e+00 7.03043938e+00 4.59172308e-01
-1.16790748e+00 -2.53681429e-02 3.04126069e-02 6.66649938e-02
-8.20275098e-02 1.15883246e-01 -1.49900520e+00 4.15588260e-01
8.00339520e-01 -1.38206214e-01 1.10251710e-01 1.11436534e+00
8.86368230e-02 -2.18717769e-01 -7.86733210e-01 1.45434201e+00
-1.60161570e-01 -1.25808966e+00 5.02728634e-02 3.16888779e-01
7.11201191e-01 5.17194390e-01 3.24698724e-02 2.03309894e-01
2.11538404e-01 -1.05411577e+00 4.05978262e-01 3.25698912e-01
1.34117973e+00 -7.93481946e-01 5.53650022e-01 5.38814366e-01
-1.70002568e+00 1.72597170e-01 -9.86034453e-01 -2.25200400e-01
1.93201751e-01 8.79148424e-01 -8.35500062e-01 4.64357853e-01
9.15067852e-01 8.57602298e-01 -3.16915393e-01 1.15740430e+00
-3.27165484e-01 2.72854507e-01 -7.63291180e-01 3.95147830e-01
-5.16394414e-02 -4.81902272e-01 2.62766421e-01 1.11619842e+00
6.13111913e-01 2.90844411e-01 4.37605321e-01 8.70027661e-01
-1.58908635e-01 -1.72504991e-01 -9.51228976e-01 5.66856742e-01
1.05199397e+00 1.08053613e+00 -1.18812688e-01 -2.35604797e-03
-4.89951313e-01 5.66387892e-01 1.63803205e-01 6.66879937e-02
-7.18058586e-01 -2.12265909e-01 9.28083062e-01 4.91785526e-01
4.64428574e-01 -8.87287199e-01 -8.19970548e-01 -1.05039525e+00
1.60566926e-01 -2.93199122e-01 -1.14756122e-01 -7.90073693e-01
-1.34344459e+00 4.06009734e-01 4.19262499e-02 -1.69498491e+00
-7.79417902e-02 -5.54259002e-01 -3.18306357e-01 1.14687884e+00
-2.13552260e+00 -1.28153968e+00 -7.54094601e-01 6.06505036e-01
5.00144422e-01 -7.14746937e-02 6.17882848e-01 3.62549514e-01
-1.48532987e-01 2.10300863e-01 -3.61742489e-02 6.62207752e-02
6.54025972e-01 -8.95280063e-01 6.27843142e-01 7.16283619e-01
2.53098533e-02 6.56494141e-01 4.39325929e-01 -9.11860466e-01
-1.50521743e+00 -1.28377950e+00 8.34781468e-01 -4.11171347e-01
6.32603109e-01 -3.36073339e-01 -9.46576357e-01 7.53676355e-01
-2.58557737e-01 7.55440831e-01 3.34692717e-01 -8.39072317e-02
-4.13573086e-01 -3.83419544e-01 -1.31630278e+00 1.56720653e-01
1.28002119e+00 -4.64699239e-01 -6.48547411e-01 1.20933652e-01
5.51053405e-01 -7.52649605e-01 -8.51983011e-01 1.04923332e+00
3.38284969e-01 -7.32847989e-01 1.35621905e+00 -6.19569048e-02
6.93636686e-02 -5.31918764e-01 -5.43936670e-01 -6.87195539e-01
-4.75380570e-01 -2.96430528e-01 -4.64075096e-02 1.07946980e+00
4.52607647e-02 -4.99856800e-01 1.27755964e+00 1.34134546e-01
-5.08896559e-02 -3.80896658e-01 -1.44281769e+00 -9.53287721e-01
2.28130072e-01 -5.05991042e-01 3.75789136e-01 8.13129723e-01
-6.93070352e-01 5.18405497e-01 -2.90807635e-01 6.58816099e-01
1.19760549e+00 1.06169514e-01 1.15895259e+00 -1.80801642e+00
3.73326391e-01 1.12281829e-01 -5.45385540e-01 -1.44949663e+00
1.96913272e-01 -5.31450152e-01 1.18429720e-01 -1.30292296e+00
-2.13115346e-02 -7.55532920e-01 1.06491260e-01 5.78103401e-02
1.14444509e-01 1.73002213e-01 6.67736754e-02 4.77692306e-01
-7.37875625e-02 6.60859168e-01 7.86760807e-01 -5.47052585e-02
3.80558781e-02 2.34235108e-01 -3.58250588e-01 9.10441279e-01
5.80801308e-01 -6.24104619e-01 -2.23070368e-01 -6.77502573e-01
-1.15254074e-01 -1.19623169e-01 2.68242747e-01 -1.40580940e+00
3.39941263e-01 -5.39786756e-01 6.10654950e-01 -1.55812573e+00
8.26559246e-01 -1.38773274e+00 9.56551284e-02 4.65286881e-01
3.21842432e-01 3.41261059e-01 2.35126466e-01 5.34164429e-01
-2.23546281e-01 -3.30107957e-02 1.17640615e+00 -3.23866494e-02
-6.64391577e-01 7.78702319e-01 3.49599160e-02 -8.70142281e-02
8.44630957e-01 -4.75812256e-01 -9.67588052e-02 -3.59437242e-02
-1.35790378e-01 1.88783646e-01 6.63554788e-01 1.70923546e-01
1.13240004e+00 -1.61684358e+00 -7.44929314e-01 5.35841405e-01
3.08317035e-01 7.46196687e-01 -1.64532840e-01 4.81219977e-01
-4.96501774e-01 4.37537163e-01 -4.36808392e-02 -1.30388963e+00
-1.21179628e+00 2.00185433e-01 1.91072926e-01 2.59625793e-01
-6.99610770e-01 7.52506137e-01 -4.34966356e-01 -5.50597548e-01
1.53858066e-01 -3.90442163e-01 3.11051786e-01 -3.07067513e-01
2.60410070e-01 4.31551129e-01 3.23495902e-02 -7.86717653e-01
-6.34104609e-01 1.43891811e+00 1.63275346e-01 4.13860306e-02
1.36756492e+00 -6.31143823e-02 5.37405796e-02 5.28550804e-01
1.10268891e+00 2.50599116e-01 -1.30783451e+00 -4.52962846e-01
2.64383852e-01 -1.17196858e+00 2.12350130e-01 -1.42830223e-01
-9.76800144e-01 1.17163944e+00 1.01483369e+00 -4.49098736e-01
7.83815622e-01 -1.44515276e-01 5.63675284e-01 5.71077406e-01
6.80033565e-01 -9.21095073e-01 -1.47278115e-01 8.08886230e-01
7.85357952e-01 -1.57631660e+00 6.69512153e-01 -8.02290618e-01
-2.47816771e-01 1.06342280e+00 7.01638281e-01 -4.02712792e-01
7.35952020e-01 7.91572213e-01 1.10004991e-01 9.00886878e-02
-2.50485897e-01 -3.56643975e-01 4.05967198e-02 8.64353299e-01
2.61764437e-01 -1.19248956e-01 -6.81578135e-03 -4.95233350e-02
-2.43160620e-01 2.50367045e-01 2.86632091e-01 9.92008090e-01
-8.47387731e-01 -1.09207392e+00 -6.60008192e-01 2.04692647e-01
1.51156530e-01 4.10999246e-02 -1.61762357e-01 1.01870251e+00
9.11945105e-02 6.17481768e-01 3.21565658e-01 -5.12815893e-01
6.49855673e-01 -2.32519269e-01 4.78054732e-01 -7.11555481e-01
1.34369433e-01 2.84726888e-01 -3.64273667e-01 -5.18712521e-01
-4.59109545e-01 -9.32653189e-01 -1.36548555e+00 -3.45592558e-01
-4.24574941e-01 2.09808536e-02 9.11048889e-01 4.11554217e-01
3.88777882e-01 -5.58643648e-03 8.25430572e-01 -1.03652406e+00
-7.11595356e-01 -8.55083048e-01 -5.49634337e-01 -9.08307657e-02
5.06047666e-01 -8.58693719e-01 -6.22123659e-01 -2.48267099e-01]
|
[7.987765789031982, -2.72666597366333]
|
43a3b42b-bafc-47a0-ab19-1d48f85be669
|
emergence-of-selective-invariance-in
|
1701.08837
| null |
http://arxiv.org/abs/1701.08837v1
|
http://arxiv.org/pdf/1701.08837v1.pdf
|
Emergence of Selective Invariance in Hierarchical Feed Forward Networks
|
Many theories have emerged which investigate how in- variance is generated in
hierarchical networks through sim- ple schemes such as max and mean pooling.
The restriction to max/mean pooling in theoretical and empirical studies has
diverted attention away from a more general way of generating invariance to
nuisance transformations. We con- jecture that hierarchically building
selective invariance (i.e. carefully choosing the range of the transformation
to be in- variant to at each layer of a hierarchical network) is im- portant
for pattern recognition. We utilize a novel pooling layer called adaptive
pooling to find linear pooling weights within networks. These networks with the
learnt pooling weights have performances on object categorization tasks that
are comparable to max/mean pooling networks. In- terestingly, adaptive pooling
can converge to mean pooling (when initialized with random pooling weights),
find more general linear pooling schemes or even decide not to pool at all. We
illustrate the general notion of selective invari- ance through object
categorization experiments on large- scale datasets such as SVHN and ILSVRC
2012.
|
['Dipan K. Pal', 'Vishnu Boddeti', 'Marios Savvides']
|
2017-01-30
| null | null | null | null |
['object-categorization']
|
['computer-vision']
|
[ 4.33056772e-01 1.24694861e-01 -6.65929243e-02 -5.92322111e-01
3.17579173e-02 -9.78521645e-01 6.72215641e-01 -2.40963653e-01
-7.42998838e-01 4.00089473e-01 4.34493959e-01 -9.98069271e-02
-6.74861133e-01 -9.88109887e-01 -6.75050378e-01 -7.35022664e-01
-3.61067802e-01 1.27390563e-01 3.38111937e-01 -1.93243921e-01
4.21136916e-01 8.01020861e-01 -1.62213135e+00 5.90137064e-01
5.00905514e-01 9.24346387e-01 -1.90776721e-01 6.37514114e-01
2.34032258e-01 5.71621239e-01 -4.79526579e-01 -4.09412444e-01
6.90527201e-01 -3.32463652e-01 -1.02043140e+00 8.56052153e-03
9.72244263e-01 3.40961926e-02 -3.03445756e-01 1.27548158e+00
3.84288639e-01 4.60894257e-01 9.02215958e-01 -1.37469625e+00
-1.11549211e+00 1.21004820e+00 -2.48745546e-01 6.33752942e-01
-1.76292956e-01 3.46073620e-02 1.41365075e+00 -7.38595188e-01
5.09039581e-01 1.62504148e+00 6.67388678e-01 6.14982843e-01
-1.76735997e+00 -7.17321277e-01 3.18062961e-01 8.00440237e-02
-1.43746185e+00 -5.60337603e-01 3.98467958e-01 -4.29061711e-01
8.87645841e-01 5.65825403e-01 1.75869584e-01 9.31779385e-01
4.20425862e-01 6.10742927e-01 1.23392856e+00 -1.91318542e-01
-5.45658218e-03 1.36349112e-01 4.24355745e-01 5.14551163e-01
5.34730792e-01 -1.90464929e-02 -4.85598445e-01 4.36134040e-02
9.68150258e-01 -1.14297405e-01 -1.48608521e-01 -2.94031560e-01
-1.43413842e+00 6.95480049e-01 1.10398543e+00 4.96720284e-01
-3.53108257e-01 -3.00470460e-02 3.63356471e-01 7.51922250e-01
-2.32333615e-02 1.33901751e+00 -4.17377651e-01 7.03904271e-01
-8.15814137e-01 2.54900366e-01 5.63022435e-01 9.13409472e-01
8.89565408e-01 2.24178303e-02 -5.92185199e-01 8.73174012e-01
-9.20665562e-02 7.87677392e-02 7.66303301e-01 -1.14438534e+00
4.63527143e-01 7.41090298e-01 -2.14844286e-01 -1.15455306e+00
-7.75736094e-01 -4.19997096e-01 -1.21458387e+00 8.55292603e-02
5.02214968e-01 -4.62495498e-02 -9.31501687e-01 2.18171883e+00
-2.49410272e-01 -2.43483648e-01 2.51283068e-02 5.75677216e-01
6.28165066e-01 2.68454999e-01 2.27968693e-01 5.87353036e-02
1.38243353e+00 -7.62548804e-01 -3.41360152e-01 -3.27545315e-01
4.75397617e-01 -3.67201984e-01 1.14856398e+00 1.83237284e-01
-9.76404250e-01 -7.17681170e-01 -1.09831500e+00 1.39283426e-02
-8.17576706e-01 -1.70112252e-01 6.69961214e-01 7.09742785e-01
-1.31223404e+00 8.14995587e-01 -5.05999863e-01 -4.96216416e-01
6.39459908e-01 7.02809632e-01 -6.77037299e-01 1.21479794e-01
-1.22617400e+00 9.33675051e-01 9.57828939e-01 1.40786961e-01
-5.87994874e-01 -9.70912814e-01 -8.27297270e-01 2.80960828e-01
1.13812074e-01 -6.63651586e-01 8.98391366e-01 -1.21651411e+00
-1.26844764e+00 7.77693689e-01 1.75790172e-02 -5.68382978e-01
3.90438855e-01 1.05759241e-01 -9.18197110e-02 -8.14911202e-02
5.73065281e-02 1.26073432e+00 9.42504466e-01 -9.94396925e-01
-3.96966159e-01 -3.54626626e-01 2.02987000e-01 1.20768405e-01
-7.72881329e-01 2.48215005e-01 2.95250475e-01 -7.11803675e-01
6.65937722e-01 -8.84725928e-01 -2.07540706e-01 -1.34972990e-01
-6.65667057e-01 -4.74023879e-01 6.80600047e-01 -1.56340599e-01
1.03134215e+00 -2.01446414e+00 1.67330146e-01 3.75601381e-01
2.00945169e-01 -4.88780774e-02 -4.92806673e-01 3.83562036e-02
-5.09788036e-01 5.87026536e-01 -1.09151840e-01 6.19260743e-02
1.92300662e-01 2.31419668e-01 -4.22363073e-01 5.14470994e-01
3.07055861e-01 8.98530066e-01 -5.56136787e-01 -3.70969713e-01
8.13594311e-02 2.02898562e-01 -1.00422263e+00 -7.47287720e-02
3.10510129e-01 -1.51873520e-02 -8.43669400e-02 3.58304411e-01
5.70944488e-01 -1.52711570e-01 7.80042931e-02 -2.56140143e-01
-2.44160354e-01 2.01577276e-01 -1.36508727e+00 9.50629115e-01
-1.55373469e-01 8.08634162e-01 -2.34153017e-01 -8.45396340e-01
7.66020834e-01 1.63043201e-01 -3.56479101e-02 -4.26637858e-01
3.06823343e-01 -1.28629148e-01 7.53540516e-01 3.82613987e-02
6.95389628e-01 -1.64015874e-01 -1.77366138e-01 3.97767574e-01
5.58117568e-01 -1.22843422e-01 2.97620714e-01 2.54483540e-02
8.70389700e-01 -3.86297345e-01 3.56426984e-01 -1.00867987e+00
5.78794837e-01 -5.35318315e-01 4.60279673e-01 1.44602275e+00
-3.88836414e-01 6.77695751e-01 7.49441981e-01 -3.18643898e-01
-1.09621787e+00 -1.06439388e+00 -3.82103980e-01 1.76429236e+00
-2.10700244e-01 -5.36795631e-02 -7.15341508e-01 -5.31412303e-01
-2.42462028e-02 6.36518061e-01 -1.08970499e+00 -4.23606515e-01
-5.23704946e-01 -7.87892401e-01 7.33536899e-01 6.57473266e-01
9.07867730e-01 -1.51604557e+00 -4.57614511e-01 -1.04997672e-01
1.55256853e-01 -9.02555227e-01 -4.66985136e-01 5.61011016e-01
-9.29064691e-01 -7.23220587e-01 -4.36098844e-01 -6.66476727e-01
9.42349315e-01 1.64963260e-01 9.46382880e-01 2.98722554e-02
-9.25789177e-02 2.18910009e-01 1.27224792e-02 -3.67698878e-01
-2.69701928e-01 6.84554935e-01 3.93128425e-01 1.58876240e-01
2.30572447e-01 -7.27571726e-01 -5.33002317e-01 7.95979261e-01
-1.13931525e+00 -2.35759988e-01 6.42308533e-01 4.59794611e-01
1.58450902e-01 2.59624243e-01 6.64657891e-01 -7.98252225e-01
8.93302500e-01 -2.89101601e-01 -4.21840519e-01 2.93089032e-01
-2.73907691e-01 3.35318834e-01 6.31284416e-01 -6.87088728e-01
-9.03394938e-01 -7.85747394e-02 1.87843680e-01 -1.23602867e-01
-2.14011192e-01 3.06232721e-02 -2.15622842e-01 -1.10746481e-01
1.02050090e+00 1.06773071e-01 -3.11908960e-01 -1.32707059e-01
4.54462469e-01 3.54042798e-01 3.66768509e-01 -6.99461997e-01
9.30515289e-01 3.91582936e-01 5.91908442e-03 -8.99283588e-01
-1.03113830e+00 -1.73893347e-01 -1.15430510e+00 2.18423586e-02
9.25469100e-01 -4.74174231e-01 -7.69113600e-01 5.03768265e-01
-8.35760117e-01 -3.50957155e-01 -4.60670590e-01 2.63450682e-01
-5.28480589e-01 -4.59804349e-02 -5.11207044e-01 -3.54607165e-01
-3.22221145e-02 -9.82424319e-01 4.94147480e-01 4.04827952e-01
-4.76929337e-01 -9.75342929e-01 -3.17160726e-01 -2.49038696e-01
7.40586877e-01 -1.70381889e-01 1.00519168e+00 -1.00146425e+00
-3.69975418e-01 -1.71746030e-01 -5.20591259e-01 6.54135048e-01
2.50441283e-01 -1.12324752e-01 -1.12009645e+00 -2.42243484e-01
-2.51568913e-01 -2.33135283e-01 1.21695483e+00 3.64697695e-01
1.32877803e+00 -6.02786541e-01 3.99947353e-02 5.58175921e-01
1.08187652e+00 -8.91192555e-02 7.40129948e-01 4.21066344e-01
5.09772062e-01 7.66710937e-01 -2.62219876e-01 -1.13219656e-01
-5.45038432e-02 4.11627740e-01 8.07836726e-02 2.33404145e-01
-1.13897864e-02 -2.23461077e-01 1.63274512e-01 4.56232190e-01
-3.68376046e-01 -3.48490924e-02 -6.54225111e-01 5.15364707e-01
-1.44477832e+00 -1.27657413e+00 2.28084743e-01 2.19502354e+00
9.77746248e-01 3.37915689e-01 -3.77613939e-02 -3.22441384e-02
8.75822127e-01 3.28162700e-01 -3.64692509e-01 -5.75768232e-01
-4.26933557e-01 1.05082445e-01 7.18831182e-01 4.85244364e-01
-1.42729473e+00 1.22370827e+00 7.14883232e+00 5.09756088e-01
-9.73437488e-01 -1.42613769e-01 3.71657729e-01 1.81143954e-01
4.30890135e-02 -3.52851182e-01 -9.97509778e-01 2.57138535e-02
8.63179147e-01 -1.88118637e-01 4.94344264e-01 7.79235065e-01
-2.28919625e-01 3.16111118e-01 -1.40830004e+00 4.70135570e-01
-2.21816581e-02 -1.12153327e+00 4.43470985e-01 4.54242304e-02
9.58364725e-01 2.74323896e-02 3.27309936e-01 5.61956704e-01
6.91728890e-01 -1.09999549e+00 6.59149170e-01 4.49358195e-01
2.91121930e-01 -6.07536077e-01 5.92349887e-01 2.03669623e-01
-1.00487125e+00 -4.33817267e-01 -9.77893054e-01 -1.58231750e-01
-3.42715770e-01 6.93010837e-02 -6.22604907e-01 1.87220499e-01
8.30103934e-01 4.70598012e-01 -1.06285763e+00 1.07193804e+00
-3.45629513e-01 4.03401345e-01 -2.93861389e-01 1.88918505e-02
2.49847531e-01 2.75682807e-01 4.10739571e-01 1.45986235e+00
7.87456334e-02 -1.23127088e-01 -3.85474600e-02 9.08466995e-01
-4.14983720e-01 4.91057821e-02 -6.78389132e-01 6.45833230e-03
4.16292697e-01 1.38414025e+00 -1.12222242e+00 -3.05445999e-01
-2.37556845e-02 7.32377887e-01 4.83338803e-01 3.89001817e-01
-4.96277422e-01 -5.93694925e-01 7.84066916e-01 -7.75744542e-02
4.33042794e-01 -2.66846836e-01 -5.23393154e-01 -1.00793982e+00
-4.46094096e-01 -6.18954957e-01 6.48905993e-01 -5.97846448e-01
-1.55726314e+00 6.42780840e-01 4.09664661e-01 -7.85022020e-01
1.85609907e-01 -9.58018422e-01 -6.24162555e-01 9.23471868e-01
-1.07765627e+00 -7.48150110e-01 -1.21396169e-01 6.68102026e-01
3.40385824e-01 -1.25031784e-01 6.65970325e-01 -1.74089506e-01
-6.66423321e-01 8.30034018e-01 -4.15134400e-01 4.75064427e-01
6.13351524e-01 -1.26005471e+00 2.80939549e-01 9.19555366e-01
1.00607269e-01 1.33463967e+00 7.01133192e-01 -2.94040233e-01
-9.15749431e-01 -1.13499928e+00 7.99898148e-01 -7.47776687e-01
8.19184601e-01 -6.95185840e-01 -1.11193967e+00 1.15061092e+00
2.20789090e-01 3.12048718e-02 4.54702646e-01 3.69114846e-01
-8.57678890e-01 -1.36422440e-01 -1.22686028e+00 9.91510630e-01
1.27959359e+00 -4.73202527e-01 -9.57655370e-01 1.06737331e-01
8.49205673e-01 1.52642392e-02 -7.36795843e-01 4.84629750e-01
7.37024009e-01 -9.33473468e-01 1.06099987e+00 -1.15332210e+00
3.18727255e-01 -1.37077078e-01 -3.43696594e-01 -1.32840586e+00
-9.99176323e-01 -2.94373393e-01 6.19887531e-01 1.17219436e+00
6.62566066e-01 -1.11678219e+00 4.75027651e-01 7.46723890e-01
-3.12025230e-02 -1.77500263e-01 -8.70366573e-01 -1.05892384e+00
4.80736226e-01 -2.46848032e-01 5.99864185e-01 1.13181162e+00
-2.54279763e-01 1.61778361e-01 1.03750132e-01 3.83672327e-01
5.18303752e-01 -2.62350649e-01 4.56740946e-01 -1.48652732e+00
7.73199052e-02 -9.50065494e-01 -7.28049457e-01 -8.18959177e-01
4.44189042e-01 -1.19110358e+00 1.84805200e-01 -1.21948242e+00
3.30049723e-01 -1.28976762e-01 -8.68252516e-01 8.31482768e-01
-1.01500928e-01 5.03613889e-01 3.52707982e-01 2.90675372e-01
-4.95737135e-01 2.31017709e-01 1.26613677e+00 -9.84114632e-02
-3.74029726e-01 -4.98011149e-02 -1.21320355e+00 7.47897387e-01
1.07647538e+00 -5.05654454e-01 -4.09183919e-01 -1.19139060e-01
2.27975249e-01 -7.67880738e-01 3.84548426e-01 -9.33764279e-01
2.94118643e-01 -2.67002255e-01 7.59027958e-01 7.60285370e-03
-8.99237543e-02 -6.22964144e-01 -3.07518929e-01 3.18156511e-01
-9.22437549e-01 1.64863318e-01 4.19876933e-01 2.31129304e-02
-4.54479270e-02 -4.46482062e-01 9.73023474e-01 -3.86018991e-01
-7.02433050e-01 2.46660486e-01 -4.92478907e-01 5.53176068e-02
6.48927927e-01 -3.09357435e-01 -6.51498437e-01 -7.08497688e-02
-7.88687825e-01 1.18763924e-01 3.86348888e-02 7.34203994e-01
2.34587327e-01 -1.31687593e+00 -6.77588284e-01 4.05360192e-01
7.97162950e-02 -4.68163043e-01 2.99357027e-01 7.02930510e-01
5.91036910e-03 5.53710878e-01 -6.80193424e-01 -4.27841455e-01
-1.10392094e+00 5.74908137e-01 6.76085949e-01 -6.69009238e-02
-2.96898156e-01 1.06118917e+00 6.18757725e-01 -8.46774518e-01
7.58926794e-02 -5.49341977e-01 -2.42968842e-01 3.10926169e-01
6.11126184e-01 2.85874695e-01 -4.11816426e-02 -5.48417032e-01
-2.41420954e-01 3.78931463e-01 -3.31846982e-01 -3.75452749e-02
1.25237191e+00 2.62440741e-01 -5.56898475e-01 4.63668466e-01
1.24154103e+00 -3.03955495e-01 -1.06308258e+00 -2.53515244e-01
1.69557929e-01 -3.22441101e-01 -1.22531146e-01 -6.01371884e-01
-9.51732397e-01 7.56550550e-01 4.20849204e-01 6.92683935e-01
8.99407983e-01 1.49996191e-01 -4.46037799e-01 1.01542568e+00
2.44937152e-01 -1.04327714e+00 1.32894295e-03 8.52636695e-01
1.36498153e+00 -9.55723226e-01 -2.28899289e-02 -1.81588560e-01
-5.25406241e-01 1.09147966e+00 8.63527298e-01 -6.10444129e-01
7.74638057e-01 2.67624613e-02 -2.39786088e-01 -7.86031857e-02
-5.64745843e-01 -1.71111286e-01 6.71642542e-01 5.39049029e-01
4.56459373e-01 1.43628463e-01 -2.83221900e-03 4.32540715e-01
-7.90384948e-01 -4.51805323e-01 5.98189056e-01 6.39764309e-01
-6.69872820e-01 -7.09604740e-01 -3.22129220e-01 5.96083879e-01
-3.48910570e-01 -2.13096201e-01 -5.55844545e-01 1.03248918e+00
2.07679436e-01 7.18467176e-01 4.52898949e-01 -3.09906065e-01
5.90899527e-01 4.55253363e-01 5.84325075e-01 -7.56151974e-01
-8.33086669e-01 -5.53897917e-01 -3.18346441e-01 -5.12684166e-01
-3.80277634e-01 -7.61209846e-01 -8.93044949e-01 -2.70760685e-01
-2.29991212e-01 -2.43273467e-01 3.12385410e-01 7.64973104e-01
1.57001317e-01 6.35262430e-01 4.32229161e-01 -1.09514177e+00
-7.07972944e-01 -1.26305687e+00 -6.86891198e-01 3.13934356e-01
3.40101838e-01 -6.47570729e-01 -8.01468313e-01 -1.31250665e-01]
|
[9.487236022949219, 2.4299702644348145]
|
a7ac2fc3-6c90-4a1e-9786-cc05118acd54
|
can-your-face-detector-do-anti-spoofing-face
|
2006.16836
| null |
https://arxiv.org/abs/2006.16836v2
|
https://arxiv.org/pdf/2006.16836v2.pdf
|
Can Your Face Detector Do Anti-spoofing? Face Presentation Attack Detection with a Multi-Channel Face Detector
|
In a typical face recognition pipeline, the task of the face detector is to localize the face region. However, the face detector localizes regions that look like a face, irrespective of the liveliness of the face, which makes the entire system susceptible to presentation attacks. In this work, we try to reformulate the task of the face detector to detect real faces, thus eliminating the threat of presentation attacks. While this task could be challenging with visible spectrum images alone, we leverage the multi-channel information available from off the shelf devices (such as color, depth, and infrared channels) to design a multi-channel face detector. The proposed system can be used as a live-face detector obviating the need for a separate presentation attack detection module, making the system reliable in practice without any additional computational overhead. The main idea is to leverage a single-stage object detection framework, with a joint representation obtained from different channels for the PAD task. We have evaluated our approach in the multi-channel WMCA dataset containing a wide variety of attacks to show the effectiveness of the proposed framework.
|
['Sebastien Marcel', 'Anjith George']
|
2020-06-30
| null | null | null | null |
['face-presentation-attack-detection']
|
['computer-vision']
|
[ 2.49230117e-01 4.07036804e-02 3.32299560e-01 -8.28303397e-02
-5.97519279e-01 -8.57211351e-01 4.45578605e-01 -1.46502331e-01
-3.50231647e-01 3.36808199e-03 -3.04731131e-01 -3.14884990e-01
4.77119505e-01 -6.17864847e-01 -7.04436123e-01 -9.68162596e-01
1.71011418e-01 -2.90567040e-01 3.80097330e-01 1.04016766e-01
7.62424916e-02 9.21593428e-01 -1.73845816e+00 4.63773847e-01
1.31679311e-01 1.39183962e+00 1.34641379e-01 7.41650939e-01
1.09375104e-01 4.00264233e-01 -6.81634247e-01 -4.55826223e-01
5.53556323e-01 -3.60267907e-01 -8.24229866e-02 2.57147998e-01
6.31075919e-01 -7.61358500e-01 -1.03243813e-01 1.10368621e+00
6.24535203e-01 -3.66949081e-01 4.21973467e-01 -1.23372161e+00
-8.24210700e-03 1.83832631e-01 -1.04204285e+00 -2.74823681e-02
4.51932400e-01 1.90751836e-01 6.25013530e-01 -8.61559153e-01
4.22581524e-01 1.36518085e+00 4.42381084e-01 4.64599818e-01
-1.02055120e+00 -1.10323060e+00 -1.19879022e-02 1.37468398e-01
-1.38941944e+00 -9.15624142e-01 8.53930593e-01 -2.56830931e-01
3.88354480e-01 9.72691625e-02 3.39667261e-01 1.02541924e+00
1.45604834e-01 5.17167628e-01 1.05718386e+00 -5.06304383e-01
3.79776210e-01 4.43339974e-01 -9.22325253e-02 7.63786614e-01
4.74418730e-01 -6.03937125e-03 -3.18429291e-01 -2.36587867e-01
5.53292334e-01 1.83296934e-01 -1.83238134e-01 -4.68643606e-01
-4.02227640e-01 5.83277345e-01 2.01159552e-01 1.37487397e-01
-2.69063175e-01 -8.66840407e-03 2.71547109e-01 1.86224997e-01
2.16593146e-01 -1.84437111e-01 2.07422096e-02 2.82533467e-01
-9.64538574e-01 -1.39687613e-01 6.77001536e-01 5.88996649e-01
7.24090636e-01 -7.20541775e-02 1.26195112e-02 4.45754766e-01
6.39907897e-01 6.22862518e-01 -8.79171193e-02 -3.96188796e-01
4.12647367e-01 4.72084641e-01 2.19674665e-03 -1.06363940e+00
-3.08733851e-01 -2.69322127e-01 -4.88906384e-01 6.97150230e-01
6.40343070e-01 -3.90594274e-01 -8.46266389e-01 1.53593647e+00
6.31554306e-01 3.48445863e-01 -1.33679181e-01 1.06415761e+00
4.09765422e-01 5.06081164e-01 5.54631092e-02 -2.34016012e-02
1.69696486e+00 -4.73118752e-01 -5.09260833e-01 -1.59789622e-01
3.76115412e-01 -1.05129516e+00 6.34993196e-01 5.32272637e-01
-6.91068888e-01 -4.89550769e-01 -1.40432358e+00 1.99216336e-01
-2.15174660e-01 3.43130618e-01 3.89178395e-01 1.40670323e+00
-8.86215925e-01 1.26374170e-01 -6.81650817e-01 -3.79874676e-01
5.23917139e-01 4.56484914e-01 -6.09390557e-01 -1.62069023e-01
-6.62913740e-01 6.69947028e-01 1.36775777e-01 3.34143192e-01
-9.02644038e-01 -2.72505909e-01 -6.82362080e-01 2.34994739e-01
3.98524702e-01 4.27161939e-02 9.22934413e-01 -1.08413768e+00
-1.43601799e+00 5.96939027e-01 -2.72617519e-01 -1.39486581e-01
5.51263690e-01 -1.15186371e-01 -3.44538718e-01 5.28149784e-01
-4.27121162e-01 3.41389298e-01 1.46335769e+00 -1.21824157e+00
-5.37910163e-01 -5.93043685e-01 2.43833616e-01 -1.92120120e-01
-6.82193518e-01 4.05762225e-01 -5.31180561e-01 -3.80957395e-01
-9.93583649e-02 -9.85898376e-01 2.08808973e-01 3.06413263e-01
-4.91004109e-01 1.97185397e-01 1.19753468e+00 -6.75422192e-01
7.64686346e-01 -2.49395752e+00 -4.62294430e-01 3.31709772e-01
-9.15560648e-02 5.50684273e-01 -2.08417386e-01 2.28261113e-01
-2.55162448e-01 -2.22341463e-01 1.28701866e-01 -5.44308126e-01
-2.20101520e-01 -2.93091655e-01 -2.89153755e-01 9.53335464e-01
4.62099791e-01 4.32110667e-01 -2.74554372e-01 -3.15472126e-01
2.47444704e-01 6.57828629e-01 -3.86600345e-01 2.12150201e-01
1.47320405e-01 3.11071962e-01 -3.68344367e-01 8.32080245e-01
1.21772408e+00 1.49767160e-01 2.68881857e-01 -1.86535493e-01
8.81011784e-02 7.09253922e-02 -1.66796207e+00 1.32367170e+00
-3.99078965e-01 4.83065635e-01 7.02238142e-01 -7.41551578e-01
8.89638364e-01 4.87681538e-01 1.58211410e-01 -6.55453146e-01
3.23641956e-01 5.96955530e-02 1.39064834e-01 -2.71680057e-01
-6.52007312e-02 -8.39561038e-03 1.54417619e-01 6.89244866e-01
-6.24890365e-02 5.01458883e-01 -1.97430342e-01 -6.68724552e-02
1.22310042e+00 1.46946311e-02 9.85710025e-02 2.05887016e-02
6.96751356e-01 -5.10980666e-01 3.14603925e-01 6.02043688e-01
-2.03810304e-01 4.47859585e-01 5.19588470e-01 -2.40269244e-01
-8.02640140e-01 -9.75294292e-01 1.06527144e-03 9.50762987e-01
9.43432450e-02 -3.16761047e-01 -9.78964686e-01 -7.38529623e-01
-4.42511439e-02 -2.67667044e-02 -3.61416489e-01 -2.15151217e-02
-5.20204902e-01 -6.47574127e-01 7.06659794e-01 3.17819625e-01
3.34887117e-01 -7.23907351e-01 -1.13341546e+00 -8.37889910e-02
3.32148284e-01 -1.32044005e+00 -2.85287052e-01 1.29920870e-01
-4.61016148e-01 -1.20111001e+00 -5.26908040e-01 -5.63603759e-01
7.18062341e-01 4.84855741e-01 3.63112748e-01 1.30084425e-01
-6.24671757e-01 2.88783163e-01 -2.94248372e-01 -4.23974901e-01
-2.33689487e-01 -3.16672802e-01 -3.47672626e-02 6.12907112e-01
2.78040767e-01 -3.63355339e-01 -6.96081161e-01 2.01156154e-01
-8.25912833e-01 -4.27626938e-01 7.37767279e-01 5.87051332e-01
6.32811710e-02 3.34196121e-01 4.88857299e-01 -6.67036653e-01
1.81091681e-01 -1.93693131e-01 -9.63530242e-01 1.78634465e-01
-1.41762033e-01 -2.18206346e-01 7.66392827e-01 -6.61824584e-01
-9.48656976e-01 7.25673497e-01 -1.32003099e-01 -4.33978647e-01
-3.40471923e-01 -1.69375524e-01 -7.12965488e-01 -5.13875246e-01
3.45804453e-01 8.19022954e-02 1.22842126e-01 -5.84233463e-01
4.28789854e-01 9.39271390e-01 2.93639779e-01 -1.79577887e-01
1.05160081e+00 7.32607782e-01 2.11368412e-01 -9.94978249e-01
-2.25413322e-01 -4.54997540e-01 -5.05200565e-01 -2.72960067e-01
5.22986233e-01 -1.05572617e+00 -1.11068141e+00 5.34096718e-01
-1.31278956e+00 2.12524757e-01 3.34313154e-01 1.44848943e-01
6.03260286e-02 6.32636428e-01 -5.43011904e-01 -1.17086720e+00
-3.41309488e-01 -1.37449145e+00 1.26666451e+00 2.17954442e-01
2.71215767e-01 -5.19699812e-01 -4.75921363e-01 8.21697563e-02
4.28311437e-01 1.56020388e-01 5.65389872e-01 -4.64479893e-01
-6.78673804e-01 -5.30124307e-01 -3.77682298e-01 3.87719959e-01
1.16929770e-01 -3.69798020e-02 -1.70072639e+00 -4.13116902e-01
2.70470232e-01 -2.60193825e-01 8.81959140e-01 6.59979284e-02
1.08489144e+00 -1.13816760e-01 -2.23434001e-01 5.01679957e-01
1.46897316e+00 2.17770994e-01 5.87088466e-01 8.08757171e-02
4.95201468e-01 8.29714060e-01 3.31571400e-01 4.76412386e-01
-6.01693839e-02 8.23855877e-01 5.50614834e-01 -2.77780652e-01
-2.39987418e-01 1.24967203e-01 7.29394436e-01 -1.75820351e-01
4.52578127e-01 -2.56363660e-01 -7.81551361e-01 1.22492716e-01
-1.37840772e+00 -8.51545930e-01 1.95153013e-01 2.24195576e+00
2.66600102e-01 -8.47095065e-03 1.67925656e-01 3.42767984e-01
8.33088458e-01 -2.08111070e-02 -3.54621053e-01 -4.09437090e-01
1.41840413e-01 3.68699342e-01 5.88634670e-01 2.08005846e-01
-1.27543890e+00 5.63969076e-01 6.06090260e+00 7.84915030e-01
-1.58118200e+00 1.06829509e-01 5.46669364e-01 -2.40409836e-01
2.73379475e-01 -1.51929036e-01 -9.79448497e-01 4.38649803e-01
8.29470575e-01 4.24285203e-01 3.28363210e-01 8.87015283e-01
1.42193571e-01 -3.86645436e-01 -1.14529133e+00 1.11098385e+00
2.97929168e-01 -6.99181795e-01 -2.66050339e-01 1.82840824e-01
3.76280993e-02 -4.54773158e-01 2.63612241e-01 -1.51471928e-01
-3.90703171e-01 -9.38533664e-01 7.23573923e-01 9.82601847e-03
7.08631396e-01 -8.83986115e-01 6.16097689e-01 2.90631533e-01
-1.33572924e+00 -3.59557509e-01 -3.17225456e-01 8.78388900e-03
-2.31737979e-02 4.75857556e-01 -1.01161468e+00 3.41800272e-01
5.40933609e-01 8.12809691e-02 -7.16220498e-01 9.90978658e-01
-4.04899903e-02 3.88774246e-01 -5.92889726e-01 3.59422773e-01
-7.03812391e-02 2.36695856e-01 2.98536003e-01 1.23651147e+00
2.87750870e-01 -1.66932061e-01 2.01595351e-01 6.15164518e-01
-8.74181911e-02 2.63070799e-02 -6.57354832e-01 7.65540544e-03
4.38291550e-01 1.64669740e+00 -9.81540859e-01 1.15175508e-01
-7.52899230e-01 1.06131136e+00 7.47280046e-02 1.53695092e-01
-7.55071044e-01 -3.02562445e-01 6.22718096e-01 2.43044585e-01
5.62600851e-01 -1.13504045e-01 -1.33083180e-01 -8.26941609e-01
3.31699401e-01 -9.58481252e-01 2.94875175e-01 -2.68681407e-01
-9.57655609e-01 4.96600062e-01 -3.32241476e-01 -1.09314167e+00
-1.36741161e-01 -8.01661611e-01 -6.50127828e-01 1.01538122e+00
-1.60710382e+00 -1.25600946e+00 -3.90639305e-01 7.35075116e-01
7.43371472e-02 -7.81135708e-02 6.83214784e-01 5.32746136e-01
-8.24166000e-01 1.00143003e+00 -2.59877950e-01 2.88593233e-01
8.52412045e-01 -6.39507890e-01 3.77490819e-01 1.24919116e+00
9.39430147e-02 5.89654863e-01 4.36015129e-01 -4.83536124e-01
-1.88092756e+00 -8.98936391e-01 2.39627272e-01 -9.73484516e-02
3.32066745e-01 -8.77118170e-01 -6.91040218e-01 1.90658152e-01
-6.56443276e-03 1.78564772e-01 4.75916594e-01 -1.16908357e-01
-6.33831203e-01 -3.37330937e-01 -1.23779488e+00 3.33837658e-01
5.66870213e-01 -6.68147326e-01 -1.03278130e-01 5.21610416e-02
2.84623094e-02 -6.37794361e-02 -2.50162154e-01 1.53817013e-01
9.31234777e-01 -9.77805316e-01 9.78285432e-01 2.29355637e-02
-7.15234950e-02 -5.29470325e-01 -8.76876116e-02 -6.37789071e-01
4.47378159e-02 -4.50060636e-01 -1.51234880e-01 1.47449362e+00
1.36152968e-01 -6.73771977e-01 8.01124334e-01 5.35124481e-01
4.57389176e-01 -2.42310897e-01 -1.16322351e+00 -7.11995959e-01
-5.70952177e-01 -4.27283734e-01 4.48888153e-01 4.48406577e-01
-6.58138096e-02 1.45047426e-01 -4.46965486e-01 7.24090993e-01
6.13195896e-01 2.61083618e-02 8.35676849e-01 -1.14551878e+00
-5.92024148e-01 -6.23634942e-02 -6.05076849e-01 -7.75343895e-01
6.21524379e-02 -5.13405025e-01 1.14858024e-01 -7.26141751e-01
1.88172951e-01 -3.01325053e-01 -2.15551525e-01 6.23466730e-01
-5.87978363e-02 6.34710729e-01 4.90087420e-01 -7.44828731e-02
-2.85504997e-01 3.65087688e-02 5.69388092e-01 6.32820185e-03
-1.40346497e-01 -2.78741241e-01 -6.58569038e-01 7.06038058e-01
5.03918946e-01 -5.01964569e-01 -4.22128588e-01 -3.78993243e-01
-2.65849680e-01 -9.99592021e-02 5.12238383e-01 -1.09886026e+00
4.60005611e-01 3.13145697e-01 9.02739942e-01 -3.74717236e-01
6.42635405e-01 -1.21059120e+00 1.04653545e-01 4.59254473e-01
1.88827172e-01 -4.92178276e-02 4.29171771e-01 5.24786055e-01
3.53783853e-02 -1.62346512e-01 1.12358403e+00 2.06699282e-01
-5.64652920e-01 3.63326445e-02 -2.57277280e-01 -6.89593434e-01
1.35222483e+00 -2.69444078e-01 -3.67455035e-01 -1.38384759e-01
-4.04433310e-01 -2.00178221e-01 6.76131606e-01 3.51066530e-01
5.23645341e-01 -8.29336405e-01 -4.25726235e-01 7.24620223e-01
-4.81980965e-02 -3.57722759e-01 3.32608491e-01 6.33507907e-01
-3.52214068e-01 1.79636687e-01 -3.09545606e-01 -5.52531600e-01
-1.50182235e+00 6.56192541e-01 2.71585107e-01 2.60957301e-01
-6.58064485e-01 6.96374834e-01 4.14187849e-01 3.74795288e-01
4.11947340e-01 1.57304615e-01 -1.04199834e-01 2.30569527e-01
1.11062729e+00 1.32207081e-01 3.40894938e-01 -5.52607119e-01
-6.40382349e-01 5.63550174e-01 -2.38542899e-01 -1.89388350e-01
1.04971421e+00 -1.30984843e-01 9.71934351e-04 -5.19202724e-02
1.08015382e+00 2.01375693e-01 -1.19508207e+00 6.35314733e-02
-1.58874586e-01 -5.99281549e-01 2.60432720e-01 -5.13969660e-01
-1.15481579e+00 1.12427056e+00 1.09378767e+00 -6.23263195e-02
1.40325177e+00 -3.59138012e-01 6.66273594e-01 1.50162160e-01
4.81804281e-01 -7.57162929e-01 7.58397579e-02 -5.65257706e-02
4.63555753e-01 -1.10288799e+00 -9.93630067e-02 -6.62648380e-01
-3.39545965e-01 1.28719866e+00 3.62128407e-01 -5.80563508e-02
6.73156321e-01 5.77418387e-01 1.29583001e-01 -2.81081777e-02
-5.03332973e-01 -1.96539998e-01 6.05630428e-02 5.71556926e-01
1.85195580e-01 -2.25309014e-01 8.20416957e-02 3.98228735e-01
2.69693077e-01 -2.37627119e-01 5.01131833e-01 9.70048726e-01
-3.88198167e-01 -1.20643258e+00 -7.61245668e-01 8.78289267e-02
-6.88221037e-01 1.72809094e-01 -3.94080728e-01 4.54691619e-01
1.58677414e-01 1.31268120e+00 -2.76053529e-02 -3.62507015e-01
1.60998970e-01 3.35576057e-01 3.82301986e-01 -5.57768583e-01
-5.28365314e-01 3.81024003e-01 -1.64983362e-01 -7.24885941e-01
-1.58155233e-01 -5.99439919e-01 -8.42160940e-01 -2.68213660e-01
-4.61320847e-01 -4.05549556e-01 9.25961375e-01 8.31831217e-01
3.72262329e-01 9.41319019e-02 8.79945517e-01 -9.03005064e-01
-5.41143537e-01 -7.66109765e-01 -6.38519704e-01 2.29976997e-01
6.04838789e-01 -6.75150335e-01 -1.19087361e-01 -7.10646808e-02]
|
[13.199747085571289, 0.924435555934906]
|
b75c7e23-99e4-44c8-8073-64c217655712
|
calculating-question-similarity-is-enough-a
|
2111.07658
| null |
https://arxiv.org/abs/2111.07658v4
|
https://arxiv.org/pdf/2111.07658v4.pdf
|
Calculating Question Similarity is Enough: A New Method for KBQA Tasks
|
Knowledge Base Question Answering (KBQA) aims to answer natural language questions with the help of an external knowledge base. The core idea is to find the link between the internal knowledge behind questions and known triples of the knowledge base. Traditional KBQA task pipelines contain several steps, including entity recognition, entity linking, answering selection, etc. In this kind of pipeline methods, errors in any procedure will inevitably propagate to the final prediction. To address this challenge, this paper proposes a Corpus Generation - Retrieve Method (CGRM) with Pre-training Language Model (PLM) for the KBQA task. The major novelty lies in the design of the new method, wherein our approach, the knowledge enhanced T5 (kT5) model aims to generate natural language QA pairs based on Knowledge Graph triples and directly solve the QA by retrieving the synthetic dataset. The new method can extract more information about the entities from PLM to improve accuracy and simplify the processes. We test our method on NLPCC-ICCPOL 2016 KBQA dataset, and the results show that our method improves the performance of KBQA and the out straight-forward method is competitive with the state-of-the-art.
|
['Jie Tang', 'Ledell Wu', 'Guoqiang Wang', 'Xiang Pan', 'Jiahong Leng', 'Sha Yuan', 'Hanyu Zhao']
|
2021-11-15
| null | null | null | null |
['knowledge-base-question-answering', 'question-similarity']
|
['natural-language-processing', 'natural-language-processing']
|
[-2.74952829e-01 5.53023636e-01 2.88519651e-01 -1.02572985e-01
-1.34800541e+00 -6.70696497e-01 5.04144251e-01 4.62129936e-02
-2.53042668e-01 1.15509355e+00 2.74492174e-01 -3.53836536e-01
-1.05524331e-01 -1.24014807e+00 -1.00294340e+00 -4.16534990e-02
5.42067945e-01 1.14547420e+00 8.59126031e-01 -8.21495831e-01
1.36000216e-01 -1.63754895e-01 -1.18301630e+00 8.55426252e-01
1.57335973e+00 9.61122394e-01 2.08516747e-01 4.06202883e-01
-8.93969357e-01 1.45696712e+00 -6.69606626e-01 -1.19175506e+00
-2.00805217e-01 -5.57425439e-01 -1.68005753e+00 -7.28577375e-01
2.90555894e-01 4.29370776e-02 -1.31392181e-01 8.29240084e-01
6.21048868e-01 1.70355774e-02 4.15581852e-01 -1.25243306e+00
-9.24094617e-01 7.93535113e-01 9.03022736e-02 -1.21530622e-01
8.95880044e-01 -1.22540042e-01 1.11341596e+00 -1.02286613e+00
9.04287398e-01 1.37133920e+00 6.50458395e-01 6.11493528e-01
-5.52439272e-01 -2.05699354e-01 -3.21074039e-01 9.35604095e-01
-1.41788554e+00 -2.78460622e-01 5.31308591e-01 -6.94686696e-02
1.15235531e+00 2.99504757e-01 2.96647519e-01 6.17750704e-01
-2.49862000e-01 1.00401688e+00 7.93243408e-01 -7.21742213e-01
2.53636204e-02 2.48709112e-01 4.51172709e-01 9.84132528e-01
1.87405989e-01 -6.36947334e-01 -6.06270075e-01 -2.59462416e-01
2.00535804e-01 -7.63156891e-01 -3.23963970e-01 -1.29270837e-01
-1.22496569e+00 6.95278525e-01 5.02483130e-01 8.77870396e-02
-3.80922735e-01 -9.82521847e-02 2.52645195e-01 4.55769658e-01
1.15379086e-02 8.42879236e-01 -8.26827466e-01 9.31992009e-02
-4.62394446e-01 6.77235723e-01 1.22232974e+00 1.16632152e+00
9.52171087e-01 -5.88138700e-01 -6.57502294e-01 5.57424664e-01
4.47512567e-01 6.36408687e-01 4.15167302e-01 -7.07920313e-01
1.01195109e+00 1.23931575e+00 3.23094934e-01 -8.60864639e-01
-2.04723373e-01 -2.97942106e-02 -3.26614767e-01 -6.95892692e-01
5.79169452e-01 -2.31649876e-01 -8.48478317e-01 1.38907945e+00
7.54060864e-01 7.54094124e-03 6.26004577e-01 7.16199696e-01
1.67205477e+00 9.35952008e-01 1.82680666e-01 1.35112857e-03
1.66651309e+00 -1.29329944e+00 -9.75705743e-01 -1.08273789e-01
9.74454463e-01 -6.18724585e-01 1.22222245e+00 -1.72065608e-02
-1.02615297e+00 -4.27007884e-01 -6.17497981e-01 -5.77504098e-01
-6.08092248e-01 2.68163323e-01 3.95465463e-01 3.39366078e-01
-8.86121154e-01 -5.36476038e-02 -2.42441997e-01 -2.69720674e-01
6.88355491e-02 2.11383626e-01 -3.10782254e-01 -4.59261417e-01
-1.90301752e+00 1.29619575e+00 8.76161337e-01 1.57764405e-01
-6.44070983e-01 -9.28006589e-01 -8.34320486e-01 6.54812679e-02
7.21462607e-01 -1.33514583e+00 1.34633648e+00 -4.94998068e-01
-1.47510934e+00 5.87700248e-01 -2.73087233e-01 -6.07602417e-01
2.24613950e-01 -3.78234982e-01 -5.43151557e-01 2.63870507e-01
3.99417967e-01 5.76947451e-01 3.32190901e-01 -1.13029623e+00
-5.15548348e-01 -1.97945058e-01 3.34248543e-01 3.05493027e-01
2.40882173e-01 -4.05345298e-02 -7.34621823e-01 7.83340111e-02
-7.98071846e-02 -6.44526422e-01 2.29944964e-03 -8.06487381e-01
-4.12221968e-01 -7.86144972e-01 5.41843116e-01 -1.18981373e+00
1.32232654e+00 -1.56756043e+00 1.22926600e-01 9.34698433e-02
-5.38672730e-02 5.78062892e-01 -9.00383443e-02 8.16935778e-01
4.15201634e-01 -6.13703541e-02 -2.27547675e-01 1.51969180e-01
2.06514448e-01 1.88172117e-01 -6.74162805e-01 -4.06974375e-01
5.79933286e-01 1.62148452e+00 -1.08571291e+00 -9.70452249e-01
-3.47885132e-01 2.50002313e-02 -4.39659923e-01 4.23580617e-01
-9.34172809e-01 3.08640093e-01 -6.88915610e-01 6.91083550e-01
5.60135245e-01 -4.10711169e-01 1.44929036e-01 -4.98244792e-01
2.99844116e-01 8.21337521e-01 -1.13989258e+00 1.87474906e+00
-3.01532626e-01 1.17109232e-01 -3.87341589e-01 -5.86253822e-01
9.13440645e-01 4.57092226e-01 -2.51473635e-01 -9.32361722e-01
-2.55849987e-01 5.62489688e-01 -1.87269345e-01 -1.04225516e+00
6.44317865e-01 9.81354043e-02 -1.15373395e-01 9.53366086e-02
4.28563386e-01 -3.47112715e-01 4.20922786e-01 6.29644632e-01
1.16776323e+00 3.11244309e-01 2.57946163e-01 3.80995423e-02
8.90192449e-01 7.63200760e-01 5.12592196e-01 6.11762106e-01
2.61045814e-01 1.28320664e-01 4.79392946e-01 -1.48861229e-01
-5.92210054e-01 -9.39019918e-01 4.33831364e-01 6.33001447e-01
2.06154346e-01 -5.29143631e-01 -8.33411515e-01 -1.24036586e+00
-1.36703327e-01 1.05837119e+00 -2.93051690e-01 -1.23936243e-01
-8.00115585e-01 -4.33799058e-01 8.84975493e-01 2.15880498e-01
8.09497714e-01 -1.44068074e+00 -1.82500873e-02 1.99595168e-01
-9.27781940e-01 -1.24823487e+00 3.35114449e-02 -6.43772900e-01
-5.61395526e-01 -1.31309962e+00 -3.39170128e-01 -9.34345961e-01
6.16098583e-01 -8.07906613e-02 1.63500428e+00 5.37900627e-03
1.42162457e-01 4.34980214e-01 -6.40429139e-01 -4.18682218e-01
-4.73037541e-01 3.41369748e-01 -4.98052537e-01 -7.70015195e-02
6.01740897e-01 -3.78482491e-02 -5.09775341e-01 1.09485693e-01
-8.02427530e-01 2.23286569e-01 6.80582821e-01 5.69962084e-01
6.07101619e-01 -1.72556788e-01 1.10920715e+00 -1.11758935e+00
9.15664613e-01 -5.88080168e-01 -4.68800247e-01 1.17092466e+00
-4.56761420e-01 4.05550450e-01 5.37207782e-01 1.06008925e-01
-1.56244993e+00 -1.03957169e-01 -2.69106030e-01 5.93405925e-02
2.55503684e-01 9.31774020e-01 -7.01566458e-01 1.08129725e-01
7.67246604e-01 2.76157469e-01 -4.53704685e-01 -4.80974734e-01
9.06829715e-01 4.88875329e-01 5.35881579e-01 -7.91736484e-01
6.61455452e-01 1.10307619e-01 -1.58635452e-01 -1.69273898e-01
-1.20603871e+00 -4.52571988e-01 -5.06129384e-01 -9.71900672e-02
8.06774735e-01 -9.46486056e-01 -8.18438292e-01 1.25659421e-01
-1.47360373e+00 -1.95466548e-01 -3.56520742e-01 2.51381099e-01
-3.48058432e-01 3.62645477e-01 -4.71144646e-01 -7.12348342e-01
-7.81096280e-01 -8.40787470e-01 9.39992547e-01 3.87674898e-01
1.18891984e-01 -9.15476739e-01 5.05482137e-01 1.13056159e+00
2.31474563e-01 -9.37443003e-02 1.19327652e+00 -8.59468341e-01
-9.38962698e-01 -5.09759821e-02 -2.78913975e-01 1.26883700e-01
-1.98959962e-01 -2.53940463e-01 -7.13557541e-01 3.21194679e-01
-4.13017690e-01 -6.89787447e-01 7.86360145e-01 -3.31659883e-01
6.59842372e-01 -3.71207982e-01 -1.78434908e-01 -1.45418510e-01
1.43980670e+00 -9.64407921e-02 1.11889088e+00 4.98340666e-01
7.79841363e-01 8.19721580e-01 8.85666251e-01 -3.43367159e-01
1.16274691e+00 5.87835014e-01 2.10278496e-01 2.85781741e-01
-3.20755839e-01 -6.82444274e-01 4.16369110e-01 1.21050525e+00
1.05999880e-01 -3.03037912e-01 -1.21924293e+00 9.64235723e-01
-2.10456061e+00 -7.65422463e-01 -5.26428223e-01 1.85371375e+00
1.34168518e+00 -3.14695358e-01 -2.77427346e-01 -2.18256041e-01
3.72434676e-01 -4.48839754e-01 -3.30911130e-01 -5.02066053e-02
-2.52706587e-01 4.40420181e-01 7.39168599e-02 6.73210502e-01
-6.50883615e-01 1.42266297e+00 5.05819368e+00 1.09870899e+00
-4.62157875e-01 1.31597281e-01 1.86824035e-02 4.75665033e-01
-5.84382653e-01 2.93319076e-01 -1.16325176e+00 1.44917801e-01
1.04970694e+00 -2.71069020e-01 2.97251582e-01 5.81249475e-01
-3.30126762e-01 -7.66957924e-02 -7.04292715e-01 6.75503135e-01
2.70279884e-01 -1.55182564e+00 6.12276852e-01 -6.22724175e-01
6.36012495e-01 -1.03421785e-01 -4.87348169e-01 9.43285763e-01
5.81604779e-01 -9.43523765e-01 4.31290895e-01 1.04734039e+00
4.58056182e-01 -6.31504834e-01 1.06449413e+00 5.60808182e-01
-1.06715798e+00 7.04153925e-02 -5.46319366e-01 2.91133374e-01
4.85647470e-01 6.57109320e-01 -1.33125699e+00 1.35835445e+00
7.14587808e-01 9.81758162e-02 -7.68562913e-01 9.54515994e-01
-7.97300875e-01 6.91332817e-01 -8.99492055e-02 -2.37832844e-01
1.13021627e-01 -1.90197185e-01 3.25398773e-01 1.03200126e+00
3.25889647e-01 2.46638641e-01 -2.23304883e-01 9.42834020e-01
-5.40038884e-01 4.63144362e-01 -3.38556409e-01 -8.03277642e-02
5.04896283e-01 1.40784621e+00 -2.59769671e-02 -5.48294842e-01
-3.68232876e-01 8.20047021e-01 8.80014420e-01 2.15950266e-01
-6.26086593e-01 -6.49415851e-01 -1.13099724e-01 -1.93561524e-01
1.62986979e-01 7.11083636e-02 1.65320605e-01 -1.31849670e+00
4.58852917e-01 -1.14365637e+00 7.78210223e-01 -1.14731038e+00
-1.44506586e+00 5.96765816e-01 -1.41798541e-01 -8.12854171e-01
-4.16300088e-01 -3.61297488e-01 -3.16590756e-01 1.02243769e+00
-1.90428340e+00 -1.53655958e+00 -3.78524661e-01 8.24262321e-01
2.70024985e-01 -7.54805878e-02 8.86772096e-01 5.17588556e-01
-3.78297716e-01 4.35127765e-01 -2.04424411e-01 3.62958461e-01
9.56394851e-01 -1.37828171e+00 4.19945031e-01 7.27957487e-01
3.55163246e-01 6.45905435e-01 2.92652160e-01 -9.33324277e-01
-1.63709939e+00 -1.31709898e+00 1.49048769e+00 -1.04909730e+00
7.07607806e-01 -1.39287785e-01 -1.10003054e+00 6.41150892e-01
3.97878587e-01 -2.82439291e-01 6.54564023e-01 -2.88826954e-02
-3.82961452e-01 -7.67239034e-02 -1.11351705e+00 4.53529418e-01
7.64177084e-01 -5.47731757e-01 -1.18990505e+00 5.00562370e-01
1.29495609e+00 -6.62314177e-01 -9.26799893e-01 6.35203004e-01
1.48352653e-01 -4.37879980e-01 9.15446579e-01 -1.03926611e+00
3.99202853e-01 -9.25853193e-01 -5.05008884e-02 -1.24054074e+00
1.40801728e-01 -4.55077946e-01 -5.95328927e-01 1.59152138e+00
1.04873121e+00 -4.95937735e-01 5.16796052e-01 6.89940870e-01
-2.83214688e-01 -6.22477770e-01 -9.89140332e-01 -4.56474364e-01
-1.04005210e-01 -1.53962225e-01 9.72624838e-01 1.02382421e+00
1.14746928e-01 9.47618484e-01 -5.31878397e-02 4.50192809e-01
1.67612359e-01 3.61084819e-01 9.58684862e-01 -8.74813139e-01
-2.20850185e-01 2.55174965e-01 5.79419769e-02 -1.14326632e+00
1.00611582e-01 -9.14988577e-01 -6.45736456e-02 -2.28003311e+00
4.24239039e-02 -4.94758964e-01 6.29802719e-02 4.67544645e-01
-6.64967597e-01 -1.29990369e-01 6.36027902e-02 9.23913643e-02
-1.21729624e+00 7.49378741e-01 1.43753994e+00 -1.70460448e-01
-2.09853232e-01 -2.35718340e-01 -7.80441821e-01 3.03187817e-01
5.77165604e-01 -3.89378190e-01 -6.15736067e-01 -6.82096183e-01
1.01866519e+00 2.10900977e-01 3.43108237e-01 -7.43477941e-01
6.31131709e-01 1.12792933e-02 -1.36790201e-01 -8.30465674e-01
1.86441928e-01 -5.86047947e-01 -2.94512566e-02 1.90893993e-01
-1.52958602e-01 1.19778305e-01 2.17948973e-01 3.97107273e-01
-5.86515546e-01 -4.66742456e-01 -5.18399477e-03 -3.06955963e-01
-8.47448111e-01 -1.13358104e-03 1.73293337e-01 7.20041394e-01
7.08775759e-01 4.43860352e-01 -9.35896456e-01 -2.56286442e-01
-3.99552077e-01 7.27140427e-01 -1.23135418e-01 2.30112866e-01
5.02907097e-01 -1.31357956e+00 -1.05155671e+00 -3.75714272e-01
4.18703556e-01 4.34263676e-01 3.26226145e-01 7.98664987e-01
-7.56286740e-01 7.83045590e-01 1.52128160e-01 -6.74691424e-02
-9.11855459e-01 3.98881525e-01 5.03259003e-01 -8.35983276e-01
-5.71044050e-02 8.80301893e-01 -1.97565556e-01 -1.06583881e+00
-2.17810631e-01 -4.29940820e-02 -8.23032916e-01 1.57845560e-02
4.99832213e-01 3.91117394e-01 4.15345132e-01 -4.06613618e-01
-3.19496274e-01 1.24997057e-01 -1.30034491e-01 -1.05869733e-01
1.04683745e+00 8.51828754e-02 -6.94259405e-01 8.13644081e-02
6.18093133e-01 2.62854785e-01 -2.50673681e-01 -4.97790128e-01
3.13370377e-01 -1.16141446e-01 -3.78733039e-01 -1.27458334e+00
-6.74631655e-01 7.37492621e-01 1.16054654e-01 1.33653671e-01
8.77163708e-01 3.82864505e-01 1.23676109e+00 8.91580284e-01
4.60164696e-01 -9.86176074e-01 -3.26712839e-02 8.43237281e-01
1.06009698e+00 -1.13447309e+00 -2.94272989e-01 -7.08641708e-01
-9.16701734e-01 8.56990278e-01 9.43507493e-01 3.35688978e-01
1.96286589e-01 -4.67604518e-01 1.10873856e-01 -5.93193591e-01
-8.86545062e-01 -5.14547169e-01 6.12063706e-01 5.39998233e-01
1.74634904e-01 -1.22194737e-01 -4.91425723e-01 1.00926542e+00
-4.25720006e-01 1.03948310e-01 1.17296189e-01 7.19272554e-01
-4.11519378e-01 -1.33627903e+00 -1.98425323e-01 3.44603807e-01
-2.48308107e-01 -4.44414228e-01 -8.49367917e-01 6.74287140e-01
1.39593065e-01 1.13151753e+00 -5.11481225e-01 -2.29737580e-01
7.80539930e-01 5.30734718e-01 3.98775548e-01 -6.09104931e-01
-7.86821067e-01 -8.70970607e-01 6.78772628e-01 -3.03933024e-01
-4.88738894e-01 -2.19521344e-01 -1.49626565e+00 -7.35735074e-02
-6.63874805e-01 8.94748390e-01 4.09769565e-01 1.14983034e+00
7.91142166e-01 3.62143904e-01 1.22666821e-01 4.60698158e-01
-3.17923605e-01 -1.05401945e+00 2.01461315e-02 4.17132705e-01
-1.43316522e-01 -3.20124805e-01 1.20372422e-01 1.32188469e-01]
|
[10.671053886413574, 7.948182582855225]
|
8484baae-f1d3-4888-9018-4cdfb24ccc12
|
a-general-purpose-algorithm-for-constrained
| null | null |
https://aclanthology.org/K19-1045
|
https://aclanthology.org/K19-1045.pdf
|
A General-Purpose Algorithm for Constrained Sequential Inference
|
Inference in structured prediction involves finding the best output structure for an input, subject to certain constraints. Many current approaches use sequential inference, which constructs the output in a left-to-right manner. However, there is no general framework to specify constraints in these approaches. We present a principled approach for incorporating constraints into sequential inference algorithms. Our approach expresses constraints using an automaton, which is traversed in lock-step during inference, guiding the search to valid outputs. We show that automata can express commonly used constraints and are easily incorporated into sequential inference. When it is more natural to represent constraints as a set of automata, our algorithm uses an active set method for demonstrably fast and efficient inference. We experimentally show the benefits of our algorithm on constituency parsing and semantic role labeling. For parsing, unlike unconstrained approaches, our algorithm always generates valid output, incurring only a small drop in performance. For semantic role labeling, imposing constraints using our algorithm corrects common errors, improving F1 by 1.5 points. These benefits increase in low-resource settings. Our active set method achieves a 5.2x relative speed-up over a naive approach.
|
['Dan Roth', 'Daniel Deutsch', 'Shyam Upadhyay']
|
2019-11-01
| null | null | null |
conll-2019-11
|
['constituency-parsing']
|
['natural-language-processing']
|
[ 8.08387220e-01 6.29566252e-01 -6.03384435e-01 -8.02278578e-01
-1.02787161e+00 -1.03780890e+00 2.50627100e-01 1.68647602e-01
-3.12770039e-01 6.89050019e-01 3.71488303e-01 -8.10645580e-01
2.81093508e-01 -1.00824118e+00 -7.86622703e-01 -1.75673977e-01
2.92623311e-01 6.06264710e-01 5.82466066e-01 -5.01997657e-02
2.90051550e-01 2.14943275e-01 -1.62368298e+00 6.02250636e-01
5.78825593e-01 5.81800163e-01 2.58399427e-01 9.32439983e-01
-4.53397870e-01 9.59251046e-01 -4.91299599e-01 -5.77312052e-01
1.30182594e-01 -5.63935459e-01 -1.16478240e+00 -2.66513973e-01
4.64635313e-01 -2.75402188e-01 3.61865044e-01 8.69983196e-01
9.08688232e-02 6.28762916e-02 2.51134455e-01 -9.10719872e-01
-1.02692291e-01 1.19087505e+00 1.72543272e-01 4.32721749e-02
6.78432465e-01 -4.74808402e-02 1.76899970e+00 -5.38694263e-01
7.18408167e-01 1.33669484e+00 3.99429977e-01 7.48119295e-01
-1.57801294e+00 -4.31097388e-01 5.23200512e-01 -3.01216692e-01
-8.36098075e-01 -7.86213517e-01 3.43978316e-01 -1.18854113e-01
1.43535984e+00 7.61290073e-01 4.96731162e-01 5.71746945e-01
-4.77797911e-02 9.30007100e-01 8.45263541e-01 -9.61980522e-01
3.47945035e-01 -7.17097223e-02 5.03431141e-01 9.62009311e-01
7.50122070e-02 -1.06984396e-02 -7.61009336e-01 -5.61455190e-01
4.99709010e-01 -4.36975032e-01 2.25922346e-01 2.31939122e-01
-9.01585400e-01 8.96133959e-01 -2.03833252e-01 -7.27020875e-02
3.04227740e-01 3.26063544e-01 3.08888614e-01 4.99433070e-01
2.81605214e-01 6.98949933e-01 -8.17044258e-01 -1.64810017e-01
-7.81470716e-01 3.92545998e-01 1.36271298e+00 1.08468151e+00
5.36158204e-01 -3.35594922e-01 -8.62454697e-02 7.14196742e-01
5.55526912e-01 2.43891373e-01 -2.09403001e-02 -1.45552778e+00
4.39175248e-01 6.05367422e-01 2.11484760e-01 -5.89803159e-01
-2.37253383e-01 1.04526497e-01 4.41284776e-02 1.01105981e-01
7.14228928e-01 -1.62788749e-01 -6.63933754e-01 1.91475129e+00
4.08442020e-01 -2.48510927e-01 9.30995867e-02 4.06049281e-01
3.49742174e-01 6.71012044e-01 2.08483025e-01 -5.60742974e-01
1.32985222e+00 -8.68822336e-01 -5.89737117e-01 -3.79959643e-01
1.05919385e+00 -7.56524801e-01 1.11249852e+00 4.65772122e-01
-1.35954785e+00 -1.93888977e-01 -9.35301423e-01 -2.51438886e-01
-4.46280511e-03 -2.49136344e-01 9.86293375e-01 7.64194727e-01
-1.02096784e+00 6.60225153e-01 -9.54348981e-01 -6.69715134e-03
-3.46438773e-02 7.54227281e-01 -1.02943920e-01 2.26957887e-01
-9.21454847e-01 7.45105386e-01 5.42211533e-01 -1.85148820e-01
-3.04102421e-01 -5.59200048e-01 -9.52797711e-01 3.64040621e-02
7.28054047e-01 -6.16596162e-01 1.96714580e+00 -9.69070494e-01
-1.78554332e+00 7.80690730e-01 -7.07383752e-01 -4.24171120e-01
1.98947042e-01 -3.06730658e-01 -1.24086075e-01 -1.97779313e-01
8.39798898e-02 6.41791165e-01 2.33124048e-01 -8.99393022e-01
-7.78380811e-01 -2.33782664e-01 6.09139502e-01 3.48545611e-01
3.16727422e-02 6.50044441e-01 -7.42376685e-01 -2.86497176e-01
4.68658447e-01 -1.11881876e+00 -4.67309594e-01 3.42858471e-02
-4.05829281e-01 -4.39045131e-01 3.00202727e-01 -3.64980251e-01
1.60892904e+00 -1.85459304e+00 1.26446947e-01 2.27442324e-01
6.73648790e-02 -9.88168493e-02 2.94303358e-01 2.07314491e-01
2.43780836e-01 4.51328993e-01 -5.25344014e-01 -2.71451026e-01
1.62860647e-01 7.21239090e-01 -4.53219771e-01 -8.98889005e-02
2.27821246e-01 8.42683434e-01 -8.76895130e-01 -6.88699841e-01
-1.40756145e-01 -2.88870066e-01 -1.11625397e+00 3.29780191e-01
-9.21541333e-01 2.58729588e-02 -4.77657109e-01 6.68463171e-01
1.73174500e-01 -2.79752463e-01 9.93479431e-01 4.16770846e-01
-2.84217536e-01 1.03315997e+00 -1.28514886e+00 1.57799590e+00
-5.60416818e-01 3.52620929e-01 -2.40948927e-02 -7.88023412e-01
7.73410857e-01 1.59783721e-01 -3.10595497e-03 -4.97076988e-01
-2.40496933e-01 3.01913291e-01 1.11166976e-01 -1.87623963e-01
5.34695566e-01 -1.05817266e-01 -4.59746063e-01 4.85369891e-01
-2.17965260e-01 -1.05562069e-01 2.96490461e-01 1.83227181e-01
1.20492065e+00 4.44868416e-01 5.61096609e-01 -3.45377386e-01
3.01066816e-01 2.79112995e-01 9.68018532e-01 9.55310583e-01
3.10312748e-01 2.21319005e-01 8.07968438e-01 -4.93009090e-01
-9.20105934e-01 -8.28230560e-01 -1.22880258e-01 1.64965081e+00
-1.56283811e-01 -8.15323174e-01 -6.61426067e-01 -9.15894330e-01
-1.65171281e-01 8.31634521e-01 -2.97371924e-01 3.39370757e-01
-1.13162267e+00 -7.15018392e-01 8.40025902e-01 6.82269454e-01
1.53921857e-01 -1.03085244e+00 -8.04246128e-01 5.21449149e-01
2.00248230e-02 -9.67003584e-01 -2.63623297e-01 5.91785133e-01
-1.22404325e+00 -8.32021415e-01 3.68750513e-01 -8.18262041e-01
7.64918983e-01 -4.33298975e-01 1.36632955e+00 4.52850938e-01
1.31992042e-01 -1.65909514e-01 -2.33472213e-01 -2.72232831e-01
-8.49548101e-01 7.61931911e-02 -1.94900632e-01 -6.73771203e-01
2.91237026e-01 -3.38477671e-01 4.19973843e-02 1.71144918e-01
-4.36248928e-01 3.73235196e-01 3.62087935e-01 9.44028854e-01
7.04969406e-01 -3.07314515e-01 8.06499198e-02 -1.73087716e+00
2.42618799e-01 -2.16868192e-01 -8.10848892e-01 3.15341353e-01
-7.60498762e-01 6.18625045e-01 8.43148172e-01 -1.13655291e-01
-1.15893161e+00 6.20685041e-01 -3.19169998e-01 3.12573254e-01
-2.13518739e-01 4.30170923e-01 -4.19739693e-01 3.41403157e-01
7.11657107e-01 -1.20016532e-02 -6.03777915e-02 -4.06621784e-01
3.28739852e-01 4.80384141e-01 3.99731547e-01 -1.23635638e+00
4.61063951e-01 5.05701639e-02 -7.54312947e-02 -3.16031843e-01
-1.08867085e+00 -1.86704651e-01 -5.99403679e-01 1.95781425e-01
4.96572793e-01 -6.04116499e-01 -7.71926999e-01 -1.33161768e-02
-1.18607736e+00 -5.83758831e-01 -1.33187696e-01 -5.50595112e-02
-5.92407823e-01 3.67742032e-01 -6.01608992e-01 -9.73320782e-01
-4.13230956e-01 -1.17940223e+00 9.79163885e-01 -3.27560566e-02
-8.53632569e-01 -8.18567991e-01 -6.50679171e-02 4.26388919e-01
2.79321764e-02 -1.17994048e-01 1.12321925e+00 -8.96680951e-01
-5.58035016e-01 2.34883279e-01 1.63126796e-01 3.07006128e-02
-1.65511861e-01 1.54142156e-01 -8.10183525e-01 1.84247553e-01
-4.36335921e-01 -2.26307526e-01 5.07223964e-01 1.59945842e-02
1.23747087e+00 -6.64443135e-01 -4.47869658e-01 2.65882850e-01
1.38598812e+00 4.47390586e-01 4.96231496e-01 5.09032421e-02
4.07955766e-01 5.01434743e-01 7.53671527e-01 1.45682275e-01
2.92002738e-01 7.45512426e-01 1.88435242e-02 2.36225635e-01
1.87756091e-01 -4.34539199e-01 4.63242948e-01 6.67916536e-01
2.23602746e-02 -3.38594913e-01 -9.88364697e-01 3.95084888e-01
-1.93405378e+00 -1.02156699e+00 -1.06762610e-01 2.19578266e+00
1.45733464e+00 7.03962445e-01 6.03550635e-02 2.86064237e-01
5.27400851e-01 -8.37983638e-02 -4.48240221e-01 -9.98922169e-01
2.02910349e-01 6.19493365e-01 4.97190475e-01 1.09079778e+00
-8.81164610e-01 1.24001110e+00 7.49014425e+00 4.48991954e-01
-6.75704241e-01 -2.73639932e-02 4.87693876e-01 -7.77672380e-02
-7.04501331e-01 5.48631787e-01 -1.30912662e+00 3.88841927e-01
1.10326397e+00 1.73603460e-01 4.14388955e-01 9.23910379e-01
-2.58985646e-02 -1.30488247e-01 -1.42435646e+00 3.21336538e-01
-2.65041560e-01 -1.55392170e+00 -1.94423661e-01 -8.96521658e-02
4.08918381e-01 -3.77151698e-01 -4.10927355e-01 2.96095222e-01
8.17328215e-01 -8.83632541e-01 9.01883125e-01 1.99335992e-01
7.99087107e-01 -6.95392251e-01 2.83993542e-01 5.76333106e-01
-1.04549909e+00 -1.52103573e-01 -9.04654041e-02 -6.55545712e-01
4.29382809e-02 3.48519444e-01 -9.35346901e-01 5.69251366e-03
3.69028032e-01 2.72971690e-01 -1.00252844e-01 4.11775291e-01
-5.53476691e-01 1.14975464e+00 -6.71375573e-01 -4.08822030e-01
1.63899027e-02 -2.07175147e-02 4.38525409e-01 1.61957145e+00
-2.48960376e-01 4.93832052e-01 5.03078461e-01 7.19834507e-01
1.45640597e-03 5.32057509e-02 -4.82824445e-01 -6.61405846e-02
8.84156585e-01 8.00925970e-01 -9.00677681e-01 -6.02243960e-01
-4.20985758e-01 7.87471533e-01 4.56269622e-01 -1.74100339e-01
-7.33793080e-01 -2.46492997e-01 5.25473416e-01 -5.25960848e-02
1.44345701e-01 -2.48693183e-01 -7.21042395e-01 -1.06753325e+00
8.23670253e-02 -1.18231857e+00 6.44465506e-01 -2.98951089e-01
-7.20227063e-01 2.03182876e-01 2.79743701e-01 -5.88247657e-01
-5.48132896e-01 -7.08221614e-01 -4.28952783e-01 7.65707910e-01
-8.74139369e-01 -7.56332099e-01 2.55083233e-01 -1.33111654e-02
8.49855483e-01 2.40542471e-01 1.34392309e+00 -1.01193719e-01
-4.84161437e-01 8.73561263e-01 -4.14763361e-01 1.42333433e-01
2.44417548e-01 -1.49566495e+00 8.05453181e-01 1.25727189e+00
3.49596351e-01 1.11965120e+00 6.99147344e-01 -5.59435129e-01
-1.58284700e+00 -7.32080579e-01 1.41464281e+00 -6.01667404e-01
5.30548871e-01 -5.30011356e-01 -6.38212919e-01 1.11225438e+00
-3.24306153e-02 -1.84228435e-01 9.04767096e-01 5.84840715e-01
-5.88222623e-01 2.69007564e-01 -9.77813482e-01 6.96494937e-01
1.52854860e+00 -4.38032478e-01 -7.98239112e-01 2.50229686e-01
1.03575552e+00 -9.07371819e-01 -9.12592947e-01 2.24241778e-01
8.04784477e-01 -6.22349560e-01 6.67580426e-01 -8.55910778e-01
5.46376109e-01 -3.16825390e-01 -5.22105098e-01 -7.02680051e-01
-5.02416849e-01 -9.27433908e-01 -3.83014292e-01 9.47755754e-01
1.14448333e+00 -5.44297993e-01 9.22897935e-01 1.10823309e+00
-2.68821448e-01 -8.57776642e-01 -4.99443650e-01 -3.79578888e-01
-1.83550399e-02 -7.76688516e-01 5.32529473e-01 7.47482061e-01
2.69452095e-01 6.29487514e-01 -2.67621372e-02 3.19925576e-01
4.38081145e-01 5.75694382e-01 4.50033039e-01 -1.19494414e+00
-9.96107757e-01 -2.32684419e-01 -7.99782807e-04 -1.34282815e+00
3.77317429e-01 -1.11455989e+00 2.69904077e-01 -1.26889265e+00
1.17396951e-01 -9.36213136e-01 1.93343703e-02 1.04187763e+00
-2.13842124e-01 -1.09321363e-02 1.90084204e-01 -1.21106068e-02
-5.60870230e-01 -4.00056124e-01 7.93490708e-01 9.24808979e-02
-2.27841079e-01 -3.77192115e-03 -1.02160549e+00 8.57535958e-01
9.67015326e-01 -7.15901256e-01 -4.70989168e-01 -5.22300899e-01
6.82681382e-01 3.41255277e-01 -1.27148524e-01 -6.09568059e-01
2.34214291e-01 -5.38824499e-01 1.90077648e-01 -3.40832829e-01
2.38707289e-01 -4.37805831e-01 3.27983499e-01 4.23150480e-01
-8.45797300e-01 5.54025061e-02 2.29587927e-01 2.92272240e-01
9.64361876e-02 -6.70356274e-01 5.88353992e-01 -3.55851531e-01
-7.18080640e-01 -1.66571453e-01 -5.49485445e-01 1.96938351e-01
5.44383466e-01 -2.59111643e-01 -2.82355007e-02 1.08322605e-01
-8.96350622e-01 1.10696368e-01 4.39307183e-01 5.05176932e-02
3.10548455e-01 -6.81089580e-01 -4.05227721e-01 1.90829411e-01
-2.61254702e-02 2.03143284e-01 -5.52536070e-01 2.66877413e-01
-5.16989648e-01 2.59587139e-01 1.87809944e-01 -4.20745105e-01
-1.64714611e+00 1.31943256e-01 1.49999991e-01 -3.79379332e-01
-5.63443840e-01 1.11284769e+00 -2.96378106e-01 -6.92374229e-01
1.34826317e-01 -5.49639404e-01 6.10905848e-02 -1.43287152e-01
4.62868929e-01 -7.73792341e-02 1.09086685e-01 -5.34693673e-02
-4.53849435e-01 2.78315008e-01 -1.64461911e-01 -4.39321667e-01
1.07788908e+00 3.05148482e-01 -2.66571283e-01 5.19824564e-01
8.40561092e-01 4.60414588e-01 -9.55457807e-01 -1.49445981e-01
3.53730559e-01 -3.19657356e-01 -5.53359203e-02 -1.01604843e+00
-4.30288553e-01 2.54963189e-01 -1.16719447e-01 2.43218392e-01
9.24759090e-01 6.43558279e-02 5.65555632e-01 7.94237137e-01
6.60105646e-01 -9.07397628e-01 -3.79872501e-01 7.05769658e-01
3.68864328e-01 -1.04255629e+00 5.07611260e-02 -8.97397697e-01
-4.46891040e-01 1.11057794e+00 5.55606484e-01 7.21481442e-02
1.58844486e-01 1.03247881e+00 -1.12595960e-01 1.04052141e-01
-1.38638079e+00 -1.23913670e-02 -7.28991404e-02 7.02116638e-02
7.82473624e-01 4.22481686e-01 -5.34473717e-01 5.27638495e-01
-5.78387499e-01 -2.12223083e-01 4.06506449e-01 1.35886002e+00
-7.37555742e-01 -1.72935140e+00 -1.51323825e-01 5.36205590e-01
-7.27173924e-01 -3.44343215e-01 -5.11579454e-01 6.77225947e-01
7.08281696e-02 1.09530699e+00 1.96238711e-01 -2.80156970e-01
1.76644877e-01 5.32322466e-01 8.08799326e-01 -1.17824733e+00
-8.28198254e-01 -1.33202389e-01 9.61387515e-01 -6.25262320e-01
-2.24288136e-01 -8.68432343e-01 -1.84997344e+00 -1.68166652e-01
-5.49498022e-01 1.73084959e-01 3.60297590e-01 9.36313987e-01
3.45483989e-01 9.30349827e-02 4.36063111e-01 -3.31423692e-02
-6.42684698e-01 -4.91419911e-01 1.10225707e-01 -5.60546108e-02
-1.05896145e-01 -3.46692979e-01 -1.74367249e-01 2.72000939e-01]
|
[10.435199737548828, 9.454289436340332]
|
91cee885-4eae-4bbf-a06c-32d1b5341ff9
|
a-lip-sync-expert-is-all-you-need-for-speech
|
2008.1001
| null |
https://arxiv.org/abs/2008.10010v1
|
https://arxiv.org/pdf/2008.10010v1.pdf
|
A Lip Sync Expert Is All You Need for Speech to Lip Generation In The Wild
|
In this work, we investigate the problem of lip-syncing a talking face video of an arbitrary identity to match a target speech segment. Current works excel at producing accurate lip movements on a static image or videos of specific people seen during the training phase. However, they fail to accurately morph the lip movements of arbitrary identities in dynamic, unconstrained talking face videos, resulting in significant parts of the video being out-of-sync with the new audio. We identify key reasons pertaining to this and hence resolve them by learning from a powerful lip-sync discriminator. Next, we propose new, rigorous evaluation benchmarks and metrics to accurately measure lip synchronization in unconstrained videos. Extensive quantitative evaluations on our challenging benchmarks show that the lip-sync accuracy of the videos generated by our Wav2Lip model is almost as good as real synced videos. We provide a demo video clearly showing the substantial impact of our Wav2Lip model and evaluation benchmarks on our website: \url{cvit.iiit.ac.in/research/projects/cvit-projects/a-lip-sync-expert-is-all-you-need-for-speech-to-lip-generation-in-the-wild}. The code and models are released at this GitHub repository: \url{github.com/Rudrabha/Wav2Lip}. You can also try out the interactive demo at this link: \url{bhaasha.iiit.ac.in/lipsync}.
|
['C. V. Jawahar', 'Vinay Namboodiri', 'Rudrabha Mukhopadhyay', 'K R Prajwal']
|
2020-08-23
| null | null | null | null |
['talking-head-generation', 'lip-sync', 'talking-face-generation']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[-4.76680361e-02 -6.82248082e-03 -4.37790543e-01 -1.43156737e-01
-1.31654882e+00 -6.53105915e-01 3.13324958e-01 -5.74463069e-01
1.93493426e-01 5.84776461e-01 3.79322350e-01 -6.06014244e-02
2.81106710e-01 -1.24645434e-01 -6.67374015e-01 -6.86771452e-01
6.35551885e-02 2.22179711e-01 9.97163579e-02 1.54419318e-01
1.37791753e-01 2.82259434e-01 -1.75101817e+00 4.09865886e-01
4.53062087e-01 8.70203733e-01 2.17029661e-01 1.10288382e+00
-5.43081500e-02 3.54751915e-01 -4.96131539e-01 -6.27934813e-01
3.39860708e-01 -7.87248552e-01 -7.48445213e-01 9.54877436e-02
7.59538770e-01 -3.00840348e-01 -3.17464352e-01 1.29013240e+00
9.74159002e-01 -1.94658637e-01 3.19108635e-01 -1.80301666e+00
-2.61863410e-01 4.08903480e-01 -5.02880037e-01 1.09131172e-01
8.82064462e-01 4.68234777e-01 6.32781982e-01 -9.61065531e-01
7.54626930e-01 1.27379334e+00 7.49641895e-01 9.06167567e-01
-1.05653608e+00 -1.27425969e+00 -3.16744357e-01 3.20639193e-01
-1.70096779e+00 -1.41140854e+00 7.92492867e-01 -3.18632215e-01
2.28961334e-01 4.31873828e-01 5.43986320e-01 1.48536801e+00
-2.42762908e-01 8.57744575e-01 8.76464665e-01 -3.99379730e-01
-1.91428736e-01 4.76527847e-02 -3.47870409e-01 5.59819579e-01
-1.30212083e-01 1.19803034e-01 -1.00283587e+00 1.28688589e-01
6.68822289e-01 -5.04506052e-01 -6.49073958e-01 -8.46022833e-03
-1.20093739e+00 6.07770681e-01 -2.30595037e-01 2.56405145e-01
-1.50850955e-02 2.92541176e-01 4.51034069e-01 2.57988483e-01
3.71419996e-01 -1.21257097e-01 -2.27952778e-01 -5.95056474e-01
-1.14961839e+00 2.42295370e-01 6.95185781e-01 1.24073231e+00
4.75294769e-01 1.17596164e-01 7.40951002e-02 9.26170588e-01
1.85388461e-01 6.68942928e-01 5.18767655e-01 -1.60426748e+00
3.40160728e-01 -1.18011378e-01 -1.41486958e-01 -7.36657262e-01
-3.18633616e-02 2.98740596e-01 -4.22821522e-01 2.61998296e-01
6.60567045e-01 -3.46901625e-01 -6.12573445e-01 1.85586536e+00
3.03616524e-01 4.70587492e-01 2.20630392e-02 7.35568345e-01
1.09787655e+00 5.15428305e-01 -6.71210587e-02 -6.74550593e-01
1.32754970e+00 -7.98799217e-01 -1.04916584e+00 9.33031887e-02
3.50621015e-01 -1.26657331e+00 1.06059670e+00 2.03357130e-01
-1.48265481e+00 -6.60230696e-01 -7.12086856e-01 1.52536318e-01
-5.05753513e-03 1.80779755e-01 3.03598046e-01 7.98173964e-01
-1.43158841e+00 2.44187504e-01 -4.07873243e-01 -6.77302480e-01
4.23719436e-01 2.80113041e-01 -5.69503307e-01 1.12020811e-02
-9.36923921e-01 3.57151270e-01 -3.51731107e-02 -3.34466934e-01
-7.82153487e-01 -8.11800957e-01 -7.95998454e-01 -4.71648157e-01
3.90859306e-01 -2.26972997e-01 1.47574532e+00 -1.23142278e+00
-1.70670354e+00 1.14815366e+00 -6.54447138e-01 -1.42747134e-01
7.77162254e-01 -1.33241238e-02 -5.73963225e-01 5.96266866e-01
2.31572598e-01 1.08593130e+00 9.47488487e-01 -1.54134893e+00
-5.16395450e-01 -3.29199359e-02 -2.52070040e-01 1.00927636e-01
4.97457273e-02 4.46372420e-01 -8.42638254e-01 -7.74988115e-01
-2.13337854e-01 -1.11906111e+00 6.77501380e-01 1.32329255e-01
-5.08120835e-01 -1.63744941e-01 1.10779154e+00 -7.87109017e-01
1.07884693e+00 -2.19205475e+00 -3.68520886e-01 -3.62787843e-01
-1.06029436e-01 3.74770969e-01 -1.73404291e-01 4.29165274e-01
-2.47572646e-01 4.77116644e-01 1.09807856e-01 -5.68667948e-01
-7.31808171e-02 -1.83420554e-01 -1.94079876e-01 5.74954033e-01
-1.71274737e-01 7.98502266e-01 -8.55791271e-01 -9.85747993e-01
5.28099351e-02 7.91562796e-01 -4.42232728e-01 2.81315923e-01
1.04574934e-01 5.03202438e-01 9.06434841e-03 9.87978578e-01
7.14193523e-01 1.84194282e-01 1.53795883e-01 -4.14288104e-01
1.71059798e-02 5.33424765e-02 -1.19415665e+00 1.68395126e+00
-3.33408594e-01 1.08816910e+00 3.66441548e-01 -5.62142432e-01
6.57356679e-01 8.22900832e-01 7.36734033e-01 -4.27323192e-01
1.64489165e-01 3.45910162e-01 -2.51066953e-01 -7.21259296e-01
1.63342059e-01 1.21605210e-01 2.29866564e-01 4.05795276e-01
2.19218060e-01 -2.80883789e-01 3.55283827e-01 -1.05815595e-02
6.73652172e-01 1.52717859e-01 9.05867219e-02 -7.50094056e-02
6.63623333e-01 -5.34073234e-01 4.44256514e-01 2.35422850e-01
-7.49111891e-01 1.04239035e+00 5.26907444e-01 1.30081385e-01
-8.41475904e-01 -1.21664453e+00 -2.13767648e-01 9.08417046e-01
1.35119557e-01 -6.37589991e-01 -1.13193476e+00 -5.06980360e-01
-2.64683962e-01 5.02516031e-01 -4.26451474e-01 2.16599286e-01
-4.14752990e-01 1.02675125e-01 9.91548359e-01 1.41392604e-01
4.98278409e-01 -1.18411839e+00 -3.01780760e-01 -2.71429539e-01
-7.87995756e-01 -1.34707320e+00 -1.06922293e+00 -4.69540089e-01
-3.36625397e-01 -1.14159477e+00 -1.03727448e+00 -8.38693738e-01
5.01954854e-01 2.53290355e-01 7.84848869e-01 1.83434010e-01
-2.29951948e-01 6.10836387e-01 -3.06511104e-01 -3.61442506e-01
-7.92415619e-01 -2.65410125e-01 3.73257548e-01 2.61944205e-01
2.59533733e-01 -6.59708500e-01 -6.21567667e-01 6.91757321e-01
-5.31708717e-01 9.13880989e-02 -1.43385157e-02 4.18719947e-01
3.73216540e-01 -2.17699841e-01 6.18626833e-01 -1.46866754e-01
4.46479976e-01 -2.93604434e-01 -4.64306861e-01 1.22357802e-02
-1.84897289e-01 -5.32207429e-01 1.74400523e-01 -7.78731823e-01
-6.38406873e-01 2.23993603e-02 -3.69837582e-01 -9.17211890e-01
-3.63374740e-01 -2.09052190e-01 -4.76054102e-01 -4.68649641e-02
2.36947939e-01 2.85417467e-01 4.14104193e-01 -3.04146230e-01
1.34121656e-01 9.88089859e-01 7.94598997e-01 -5.26474357e-01
7.29108393e-01 5.66310763e-01 -3.86718631e-01 -1.12169039e+00
-3.61762166e-01 -3.64564061e-01 -4.01575863e-01 -7.83042133e-01
8.74583900e-01 -8.58866811e-01 -1.20306039e+00 8.34614754e-01
-1.15220976e+00 -6.87083721e-01 -2.72814512e-01 4.48942930e-01
-9.81578946e-01 4.28979158e-01 -4.12386626e-01 -8.09928358e-01
-8.31769966e-03 -1.40378916e+00 1.03283036e+00 4.74607825e-01
-4.49629664e-01 -8.56958449e-01 9.23143998e-02 7.13596880e-01
2.50798941e-01 1.19423335e-02 2.60400563e-01 -5.49805820e-01
-3.62830311e-01 -6.99561834e-02 3.18711065e-02 4.31374639e-01
3.11026454e-01 5.61937094e-01 -1.34698713e+00 -3.38833094e-01
-3.27809155e-01 -4.35599715e-01 3.68211180e-01 5.30635893e-01
1.00391507e+00 -5.09069681e-01 -2.47347668e-01 7.32258022e-01
1.11693954e+00 1.92651376e-01 7.11918712e-01 -1.07131355e-01
3.05345118e-01 6.98804259e-01 5.86049378e-01 3.43459398e-01
3.05407733e-01 1.16070235e+00 2.52133429e-01 4.13915664e-02
-8.45717788e-01 -4.08712983e-01 8.39014471e-01 6.19686782e-01
-1.36258855e-01 -3.53843182e-01 -7.56262243e-01 6.64176285e-01
-1.45096266e+00 -1.25202513e+00 -4.74048890e-02 2.25637031e+00
1.11106312e+00 -1.40327603e-01 5.20868540e-01 2.62432694e-01
1.09245694e+00 2.83213049e-01 -1.89839527e-01 -2.23055005e-01
-1.03743836e-01 -9.65331402e-03 1.89747900e-01 8.96080017e-01
-8.79177332e-01 1.12193704e+00 5.21891594e+00 1.04408813e+00
-1.48393297e+00 3.74321580e-01 5.54260433e-01 -3.36784333e-01
-2.17418000e-02 -9.84472409e-02 -1.01217031e+00 7.90002406e-01
1.12483191e+00 -4.64292556e-01 4.38937932e-01 5.38974643e-01
7.59784877e-01 -1.25296727e-01 -9.66518104e-01 1.40747654e+00
3.16229224e-01 -1.38168812e+00 -1.25355080e-01 2.03992054e-01
5.03647685e-01 -1.75343409e-01 2.52125621e-01 -9.92046446e-02
-2.43160799e-02 -9.47571278e-01 1.03745258e+00 4.20131177e-01
1.55471468e+00 -5.10246038e-01 3.36072922e-01 -1.98877648e-01
-1.35689139e+00 1.41420037e-01 -4.58000489e-02 6.28416777e-01
3.21772337e-01 4.42148820e-02 -8.02623510e-01 2.90776700e-01
8.96414101e-01 5.39825916e-01 -3.53135347e-01 9.93252993e-01
-6.23063035e-02 7.26566613e-01 -2.30541676e-01 3.56857628e-01
-3.90772939e-01 1.76373497e-01 8.31355453e-01 1.31418324e+00
4.55356002e-01 -1.14078343e-01 -1.93074390e-01 6.53041303e-01
-2.17376828e-01 1.82582498e-01 -7.97641337e-01 5.07264473e-02
6.34133339e-01 1.15361452e+00 -5.47187865e-01 -4.04214002e-02
-2.14053765e-01 7.53664732e-01 -3.91005844e-01 4.99962747e-01
-1.04122317e+00 -2.15230569e-01 1.02267826e+00 4.45940852e-01
1.57243505e-01 -1.73439719e-02 1.83509469e-01 -8.66752744e-01
5.09184934e-02 -1.14037359e+00 7.85227120e-02 -1.14315152e+00
-7.87487209e-01 5.26839972e-01 1.92083552e-01 -1.27925909e+00
-5.11326909e-01 -1.94394559e-01 -7.65429974e-01 7.07183778e-01
-1.31496382e+00 -1.30491841e+00 -3.81370544e-01 1.03489637e+00
9.69625771e-01 -2.97499120e-01 7.04644918e-01 4.38766629e-01
-4.15999174e-01 1.08588457e+00 -1.78573295e-01 3.62997562e-01
1.27666521e+00 -7.06986845e-01 6.89052716e-02 7.61922657e-01
1.04605108e-01 1.52558088e-01 8.71694088e-01 -5.28438449e-01
-1.13610625e+00 -6.71061695e-01 9.50747728e-01 -4.30623621e-01
6.51649296e-01 -5.04693151e-01 -7.15319335e-01 6.26400173e-01
5.41567445e-01 1.17595784e-01 6.48152053e-01 -5.95758259e-01
-2.41441160e-01 -2.72860587e-01 -1.08757615e+00 6.62680447e-01
1.02976418e+00 -6.78183973e-01 -2.55107731e-01 3.49907100e-01
5.73588252e-01 -2.19953641e-01 -7.20357120e-01 2.50927299e-01
9.55019236e-01 -1.31514561e+00 8.66771519e-01 -1.63583323e-01
2.33390272e-01 -1.71091452e-01 -2.55305916e-01 -7.82721758e-01
4.76394981e-01 -1.38808751e+00 -7.42746815e-02 1.87467015e+00
1.63058773e-01 -5.25870204e-01 8.24588835e-01 5.62371984e-02
1.26599267e-01 -3.84351492e-01 -1.15233088e+00 -8.80926013e-01
1.11298850e-02 -7.58978546e-01 2.81904936e-01 8.20345819e-01
8.30979347e-02 -3.38600464e-02 -3.66646498e-01 8.97731557e-02
7.33877122e-01 -4.66381282e-01 1.16135764e+00 -8.30866396e-01
5.79832196e-02 -4.76467341e-01 -3.62435371e-01 -8.27244341e-01
3.62721890e-01 -6.94718897e-01 1.91823542e-02 -1.17151809e+00
-2.41244473e-02 -1.78927541e-01 3.65977973e-01 3.59245032e-01
6.53646514e-02 5.87578893e-01 5.76430380e-01 5.05439341e-01
-4.13467795e-01 2.20322713e-01 1.13947558e+00 -2.62337942e-02
-1.48491517e-01 3.28570902e-01 -5.02525628e-01 7.15638578e-01
8.31224024e-01 -4.30223733e-01 -4.12772566e-01 3.45426388e-02
-3.28598082e-01 4.08722222e-01 5.25112987e-01 -1.19167876e+00
3.22643071e-01 -4.83885184e-02 -9.29867923e-02 -4.22849059e-01
7.49466062e-01 -4.83823746e-01 3.51356894e-01 2.59678364e-01
-1.47102028e-01 3.94507758e-02 3.19046229e-01 5.97927626e-03
-2.01980054e-01 -1.84717253e-01 1.11567354e+00 1.00921066e-02
-5.37497282e-01 3.13844323e-01 -3.14260930e-01 5.05742610e-01
1.03752005e+00 -5.30816853e-01 -3.52900147e-01 -1.11159599e+00
-7.23131478e-01 -9.96574759e-02 7.65671253e-01 5.29787362e-01
5.31841755e-01 -1.32603884e+00 -8.41271460e-01 2.89594740e-01
-1.08505048e-01 -4.62860048e-01 2.52267152e-01 1.18482971e+00
-4.67472404e-01 2.80608088e-01 -3.68095785e-01 -7.66188741e-01
-1.93404782e+00 2.99512953e-01 5.43561697e-01 2.52298385e-01
-4.87419635e-01 8.31646740e-01 2.17244044e-01 1.88505929e-02
4.77801889e-01 9.14540663e-02 2.07915276e-01 3.03928375e-01
4.71708655e-01 2.89021552e-01 -3.56281966e-01 -1.07829356e+00
-4.32735264e-01 8.57770383e-01 4.06250775e-01 -3.67529660e-01
9.20852900e-01 -4.61906552e-01 3.27373326e-01 4.06747520e-01
1.43432510e+00 6.53675139e-01 -1.32494140e+00 2.69989431e-01
-4.93600249e-01 -6.10368490e-01 -4.12323624e-01 -5.33096075e-01
-1.25105321e+00 9.21796739e-01 7.54227579e-01 1.41881660e-01
1.02588379e+00 2.40045786e-01 6.42921865e-01 -2.82348216e-01
1.12467960e-01 -8.00028861e-01 4.01250184e-01 1.84897780e-01
1.20998836e+00 -1.15861523e+00 -2.43885994e-01 -4.23719257e-01
-8.73351872e-01 1.23298550e+00 2.78403819e-01 4.07105118e-01
7.28320122e-01 4.40242767e-01 6.26490355e-01 2.53444761e-01
-6.45806611e-01 -3.51721495e-01 -1.44353852e-01 1.09891868e+00
3.75893861e-01 -1.93521231e-01 -1.91192217e-02 1.16534658e-01
-5.31930029e-01 2.14423344e-01 7.09692180e-01 5.09574413e-01
2.69371420e-02 -1.05849826e+00 -5.61275721e-01 -2.07281366e-01
-6.80984676e-01 -1.17618248e-01 -3.62136543e-01 9.98031676e-01
1.11726806e-01 1.19414079e+00 -2.53236927e-02 -3.99935365e-01
2.68458333e-02 3.48461449e-01 3.46814007e-01 -2.33344004e-01
-2.38614351e-01 4.37730193e-01 2.13817745e-01 -6.03509545e-01
-4.63151723e-01 -9.54783261e-01 -1.04235578e+00 -6.97952390e-01
4.91477437e-02 2.65134394e-01 7.69164205e-01 3.73104781e-01
3.11447114e-01 4.91808169e-02 6.85792923e-01 -1.29581296e+00
-9.06630158e-02 -9.42509890e-01 -3.89715821e-01 3.40687901e-01
4.72691774e-01 -6.21740580e-01 -8.83808017e-01 5.62229395e-01]
|
[13.312664031982422, -0.3519928455352783]
|
d2dce1d5-735b-4d15-812f-e288fc02cb90
|
interactive-video-stylization-using-few-shot
|
2004.14489
| null |
https://arxiv.org/abs/2004.14489v1
|
https://arxiv.org/pdf/2004.14489v1.pdf
|
Interactive Video Stylization Using Few-Shot Patch-Based Training
|
In this paper, we present a learning-based method to the keyframe-based video stylization that allows an artist to propagate the style from a few selected keyframes to the rest of the sequence. Its key advantage is that the resulting stylization is semantically meaningful, i.e., specific parts of moving objects are stylized according to the artist's intention. In contrast to previous style transfer techniques, our approach does not require any lengthy pre-training process nor a large training dataset. We demonstrate how to train an appearance translation network from scratch using only a few stylized exemplars while implicitly preserving temporal consistency. This leads to a video stylization framework that supports real-time inference, parallel processing, and random access to an arbitrary output frame. It can also merge the content from multiple keyframes without the need to perform an explicit blending operation. We demonstrate its practical utility in various interactive scenarios, where the user paints over a selected keyframe and sees her style transferred to an existing recorded sequence or a live video stream.
|
['Daniel Sýkora', 'Sergey Tulyakov', 'Šárka Sochorová', 'Ondřej Jamriška', 'Michal Kučera', 'Ondřej Texler', 'Menglei Chai', 'David Futschik']
|
2020-04-29
| null | null | null | null |
['video-propagation']
|
['computer-vision']
|
[ 4.78370726e-01 -2.53410488e-02 2.85928603e-03 -2.09484279e-01
-4.57826704e-01 -6.84600353e-01 7.39730120e-01 -3.07718758e-02
-4.21396643e-01 6.07365191e-01 -2.11081401e-01 -1.39152631e-01
2.75619894e-01 -6.80684328e-01 -9.51016128e-01 -5.40624261e-01
2.84521431e-01 4.70486969e-01 5.11325538e-01 -2.44570985e-01
1.31923094e-01 8.86621416e-01 -1.34313476e+00 2.27612257e-01
4.98996884e-01 8.75073850e-01 3.92544508e-01 1.00344539e+00
-2.70277411e-01 7.45470822e-01 -6.59480453e-01 -5.43546617e-01
2.95405477e-01 -6.76045358e-01 -6.93723142e-01 4.78542536e-01
8.44213247e-01 -4.97972578e-01 -1.75108537e-01 8.10472548e-01
1.88319772e-01 3.30450535e-01 3.82991284e-01 -1.34416354e+00
-3.05586725e-01 2.90436447e-01 -5.97270131e-01 -6.56424984e-02
4.63530838e-01 1.06130607e-01 7.40842760e-01 -6.60092652e-01
1.18493462e+00 1.17376077e+00 5.66394985e-01 6.17204964e-01
-1.44887698e+00 -5.34406960e-01 3.42416078e-01 5.94581962e-02
-1.04373384e+00 -4.08243924e-01 8.98470521e-01 -3.54265273e-01
3.29676509e-01 3.34998101e-01 1.15860951e+00 9.41070139e-01
1.58396646e-01 8.53285372e-01 8.29097271e-01 -5.43490529e-01
2.12679818e-01 -1.45300552e-01 -3.02402884e-01 8.43820095e-01
-2.49864161e-01 -9.16187987e-02 -6.55525982e-01 -5.35651669e-02
1.42778003e+00 1.06344759e-01 -2.81712502e-01 -6.66164637e-01
-1.42874813e+00 4.20574158e-01 1.32068932e-01 1.22143231e-01
-3.79461735e-01 6.32062972e-01 4.56914961e-01 5.04733086e-01
4.16597068e-01 3.78159076e-01 -3.35404515e-01 -2.64308751e-01
-1.44612539e+00 3.72025490e-01 5.79407990e-01 1.18776286e+00
8.78077149e-01 1.50105968e-01 -1.06344208e-01 3.28237742e-01
-8.92739668e-02 5.41937053e-01 2.01123029e-01 -1.41399074e+00
1.58264190e-01 7.36183450e-02 3.13791573e-01 -8.08732808e-01
1.03145011e-01 2.14860007e-01 -6.19481862e-01 7.90037334e-01
5.46388745e-01 -1.33478642e-01 -8.69271100e-01 1.65217757e+00
3.99625987e-01 4.93072331e-01 -3.63490105e-01 6.19013727e-01
4.38796639e-01 6.47532105e-01 -6.68974444e-02 -3.53072733e-01
1.18824100e+00 -1.04021537e+00 -7.37458944e-01 1.34001389e-01
1.08839661e-01 -1.02625740e+00 1.43003261e+00 5.52521706e-01
-1.52047670e+00 -7.83531547e-01 -8.85519981e-01 -2.32115671e-01
6.13958808e-03 -2.34213501e-01 6.03344083e-01 3.22419345e-01
-1.21647346e+00 8.17497730e-01 -9.42186296e-01 -5.25866508e-01
1.83367059e-01 3.20166230e-01 -4.27111238e-01 3.19814384e-01
-6.65267110e-01 5.83333313e-01 3.53179216e-01 -2.41924137e-01
-7.57632673e-01 -8.05347025e-01 -7.19280064e-01 -1.05008699e-01
4.57062840e-01 -1.05327034e+00 1.59842765e+00 -1.74580538e+00
-2.03739953e+00 6.89901054e-01 -4.86877501e-01 -2.19917029e-01
9.27061558e-01 -6.31389678e-01 -2.99013555e-02 3.95988762e-01
-2.08216775e-02 9.05397654e-01 1.29011440e+00 -1.46835911e+00
-8.96661997e-01 1.30531847e-01 3.55093509e-01 3.27706724e-01
-3.33880410e-02 2.00907234e-02 -9.60722566e-01 -1.26678038e+00
1.21814851e-03 -1.07788122e+00 -2.99601424e-02 4.00675267e-01
-2.40375444e-01 1.36850417e-01 1.14207840e+00 -4.64596093e-01
1.05612063e+00 -2.17475939e+00 6.50992692e-01 3.05054784e-01
1.19550146e-01 1.35123089e-01 -4.52640839e-02 4.77015346e-01
-9.56293419e-02 -1.64492637e-01 -1.57285869e-01 -6.28487706e-01
-1.70237392e-01 2.26310372e-01 -4.29936767e-01 2.22149149e-01
-5.95861003e-02 8.86891723e-01 -1.08731782e+00 -6.18712246e-01
3.88575286e-01 3.81229639e-01 -6.97321236e-01 4.69929636e-01
-5.18001854e-01 6.73621893e-01 -8.73051211e-02 2.95757264e-01
3.49386424e-01 -3.05719338e-02 1.19841784e-01 -3.69048476e-01
-8.85698274e-02 -3.90774496e-02 -1.35595357e+00 2.33216262e+00
-6.10519111e-01 7.92100489e-01 -6.18978925e-02 -5.91030657e-01
5.66866457e-01 4.01603281e-01 3.96326184e-01 -2.53732085e-01
8.97158962e-03 -1.41211422e-02 -5.41833937e-01 -3.22678775e-01
7.17642248e-01 -2.52122611e-01 4.52140160e-02 9.92762983e-01
-6.58918545e-03 -5.01927316e-01 2.96269625e-01 3.32607865e-01
7.20409930e-01 8.33599627e-01 4.39477473e-01 -4.10786383e-02
3.85710925e-01 -1.68342069e-01 4.27276313e-01 5.10389745e-01
4.02446747e-01 6.79257393e-01 5.01357913e-01 -5.76270878e-01
-1.25925457e+00 -1.17961442e+00 4.25256252e-01 1.22638369e+00
3.19593400e-01 -4.90920871e-01 -9.66576755e-01 -6.53153181e-01
-3.69117439e-01 6.59531653e-01 -5.80448449e-01 2.03685760e-01
-1.06373692e+00 1.83678880e-01 2.77126074e-01 4.99374747e-01
3.56240422e-01 -1.24841583e+00 -9.34317887e-01 1.97988018e-01
-1.91112667e-01 -9.00471985e-01 -1.12093902e+00 -1.64727643e-01
-1.14220476e+00 -7.46919036e-01 -9.29704964e-01 -8.16793859e-01
8.51892650e-01 2.64802098e-01 1.15067303e+00 1.23784021e-01
9.17288754e-03 4.75999773e-01 -2.17037529e-01 -5.71151711e-02
-6.68710709e-01 -3.53579558e-02 -6.56282827e-02 3.97387668e-02
-1.58843011e-01 -7.85225987e-01 -3.82999808e-01 1.57017395e-01
-1.20317221e+00 6.55331314e-01 1.86963499e-01 7.97383368e-01
8.19595635e-01 -1.70149013e-01 1.67038605e-01 -9.90457416e-01
4.01618987e-01 2.68306583e-01 -5.62620044e-01 2.54823208e-01
-1.68839931e-01 1.74045488e-01 7.20696211e-01 -7.26266086e-01
-1.21187055e+00 2.95881510e-01 1.89812556e-01 -6.66608095e-01
-1.56419516e-01 7.51141161e-02 -1.50549784e-01 1.24729194e-01
3.31058383e-01 1.47676514e-02 -4.47704978e-02 -4.29921299e-01
9.28050160e-01 1.71527222e-01 8.61697018e-01 -6.44839108e-01
1.08163428e+00 6.08625054e-01 -1.07392035e-01 -8.29031408e-01
-3.12523961e-01 -9.25735682e-02 -1.10351288e+00 -3.48214686e-01
8.26324940e-01 -5.94964385e-01 -6.62505269e-01 4.42485332e-01
-1.40212381e+00 -7.57808208e-01 -6.80352032e-01 2.51421183e-01
-1.02129138e+00 5.43398261e-01 -5.88032722e-01 -3.39190125e-01
-2.74865687e-01 -1.17832446e+00 1.13015902e+00 8.17695335e-02
-8.27943861e-01 -9.74435449e-01 -1.50180250e-01 -2.20621198e-01
2.75279526e-02 4.36107606e-01 8.64225984e-01 1.46943063e-01
-7.03784406e-01 -2.05098419e-03 -6.07676283e-02 -8.78731895e-04
5.16682267e-01 4.62272882e-01 -6.82944477e-01 -3.09063792e-01
-1.42936245e-01 -1.02929533e-01 4.46039140e-01 2.35528037e-01
1.07118320e+00 -3.01079869e-01 -1.91171929e-01 8.04311454e-01
1.19954705e+00 4.25817758e-01 4.65996146e-01 3.78669620e-01
1.00651574e+00 4.18192416e-01 5.91613472e-01 2.55057633e-01
3.09929829e-02 8.42872381e-01 1.17245629e-01 -3.65707010e-01
-2.49636069e-01 -3.49807888e-01 3.52414489e-01 5.03283620e-01
-4.56649750e-01 -2.04589382e-01 -2.83062220e-01 1.86459675e-01
-1.84126794e+00 -1.13867807e+00 3.30083609e-01 2.33532619e+00
9.64644670e-01 1.21591836e-01 2.74035454e-01 1.23069994e-01
6.39352024e-01 1.65503815e-01 -5.21254301e-01 -4.03072566e-01
2.29190260e-01 4.44338679e-01 3.56440961e-01 7.33325720e-01
-1.02751195e+00 1.27730477e+00 6.31204844e+00 7.81520784e-01
-1.43185925e+00 -7.07997307e-02 5.03546119e-01 -3.50855887e-01
-4.95798677e-01 -6.16497956e-02 -2.66716003e-01 2.82242239e-01
4.26210344e-01 -1.65175468e-01 6.31400228e-01 5.88114440e-01
4.34424311e-01 -2.28930593e-01 -1.47593236e+00 1.04936695e+00
3.39421742e-02 -1.49970675e+00 3.38857144e-01 -2.57012784e-01
6.70773208e-01 -6.05927706e-01 -5.50758466e-02 -2.79052615e-01
2.96180934e-01 -8.51587951e-01 1.16001737e+00 6.58968627e-01
1.25890148e+00 -8.21226597e-01 7.03558624e-02 8.78423378e-02
-1.23033929e+00 4.66609240e-01 6.55043423e-02 6.68000802e-02
6.07668221e-01 1.33001516e-02 -6.13363743e-01 3.63620073e-01
5.53453743e-01 5.55967033e-01 -2.94324487e-01 6.69974804e-01
-3.20282131e-01 3.39597762e-01 -3.03433001e-01 3.19064319e-01
1.43197820e-01 -3.13186705e-01 5.01032889e-01 1.26725137e+00
5.04928529e-01 9.98803824e-02 1.58258706e-01 4.76473600e-01
-5.84209301e-02 5.87299690e-02 -5.13711870e-01 1.61679074e-01
3.56134981e-01 1.13295281e+00 -1.02235281e+00 -7.38853514e-01
-4.25269663e-01 1.83942711e+00 8.94563794e-02 6.08780205e-01
-9.35119927e-01 -4.88782227e-01 5.67123771e-01 4.11923379e-01
4.33236033e-01 -4.05207545e-01 -1.81726396e-01 -1.21439624e+00
-6.45008460e-02 -1.07635581e+00 5.97096682e-02 -9.97603238e-01
-8.24588835e-01 7.80348003e-01 1.53734222e-01 -1.31229782e+00
-5.10643244e-01 -2.52727062e-01 -6.78247273e-01 7.33316839e-01
-1.04745364e+00 -1.34927487e+00 -3.64323974e-01 7.03315139e-01
8.58037949e-01 -1.05978318e-01 7.71246850e-01 1.50293186e-01
-1.56277373e-01 5.02734840e-01 -1.47960663e-01 3.13163246e-03
9.93516862e-01 -1.36106503e+00 6.68613195e-01 9.44512844e-01
3.86426687e-01 5.35065770e-01 1.00533962e+00 -5.86283863e-01
-1.16203952e+00 -8.02145958e-01 8.39722812e-01 -2.47664019e-01
6.02188110e-01 -2.65626103e-01 -6.81796014e-01 9.59881961e-01
6.02020204e-01 -2.02717453e-01 4.75407630e-01 -2.70596683e-01
-8.17172676e-02 -2.25810975e-01 -7.67248571e-01 9.89300907e-01
1.16947234e+00 -5.00438213e-01 -6.08669937e-01 -6.05580620e-02
6.75053537e-01 -7.01430678e-01 -6.37880862e-01 -3.65918353e-02
8.14105093e-01 -9.75199342e-01 9.24901307e-01 -6.84721470e-01
5.00037909e-01 -6.02261126e-01 5.62934801e-02 -1.19367015e+00
-2.68588036e-01 -1.25685251e+00 -2.25896403e-01 1.13786840e+00
-9.36859753e-03 -1.03416726e-01 8.07370245e-01 6.21210396e-01
4.87056673e-02 -3.03688109e-01 -5.92999816e-01 -5.31823754e-01
-3.39130372e-01 -4.37090993e-01 6.96343899e-01 7.63807893e-01
-1.18535660e-01 2.01330110e-01 -6.70615375e-01 -4.29399312e-03
5.04420280e-01 4.36835736e-01 1.27320123e+00 -9.96294677e-01
-5.93080342e-01 -2.83461511e-01 -1.05733208e-01 -1.36829913e+00
8.90251845e-02 -5.21609604e-01 -4.85502034e-02 -1.32762921e+00
-8.91258046e-02 -4.22305882e-01 9.46998410e-03 2.73379683e-01
-2.29759425e-01 4.61392432e-01 4.82160330e-01 2.88327962e-01
-3.90057117e-01 2.62092739e-01 1.65079796e+00 -4.48101275e-02
-4.84236419e-01 6.57218471e-02 -3.16754043e-01 1.03595889e+00
5.64802229e-01 -2.80290753e-01 -7.02623725e-01 -5.07921398e-01
4.48054783e-02 3.55572969e-01 3.77620667e-01 -8.14723372e-01
4.48478721e-02 -4.36968297e-01 4.52861190e-01 -4.07921046e-01
3.84324372e-01 -8.81588042e-01 6.29084945e-01 3.33204687e-01
-3.14263254e-01 3.90995681e-01 1.84986964e-01 4.49661970e-01
-6.60432503e-02 -2.78217077e-01 8.48086298e-01 -2.41352171e-01
-8.44742954e-01 4.39618260e-01 -4.77635831e-01 -2.43153065e-01
1.21278954e+00 -4.62803870e-01 4.36700583e-01 -5.12314975e-01
-9.37477171e-01 -3.84862781e-01 1.05830109e+00 1.97139680e-01
4.03947920e-01 -1.51043439e+00 -2.57575035e-01 1.71302512e-01
-2.31935754e-01 1.09940134e-01 1.59839832e-03 3.11845511e-01
-1.05862057e+00 -3.06167640e-02 -4.01026160e-01 -5.43448031e-01
-1.59826267e+00 7.70208061e-01 8.64163190e-02 3.05924448e-04
-1.05346680e+00 7.30383754e-01 4.90704417e-01 2.08975106e-01
2.32238889e-01 -4.92779672e-01 1.82885289e-01 1.22563601e-01
6.12322509e-01 3.00373316e-01 -2.54887193e-01 -5.77086508e-01
5.15236445e-02 7.72894323e-01 5.86787565e-03 -6.71921074e-01
1.07974708e+00 -2.41487384e-01 -1.26534536e-01 8.07648659e-01
9.58051145e-01 2.38417760e-01 -1.80195963e+00 -2.88054019e-01
-1.77881002e-01 -6.50664806e-01 -2.45621607e-01 -2.85569519e-01
-1.05257571e+00 5.99797785e-01 2.69055694e-01 2.10572947e-02
1.24959683e+00 -1.88419342e-01 1.01677406e+00 4.25557494e-01
4.15277988e-01 -1.06250012e+00 3.91524881e-01 2.15953752e-01
8.68710816e-01 -6.92270637e-01 4.15144041e-02 -3.62101614e-01
-7.00487077e-01 1.38437033e+00 3.60896945e-01 -2.27238029e-01
3.64196986e-01 4.18300569e-01 4.30171728e-01 6.48553073e-02
-4.42099482e-01 1.44131452e-01 2.06941798e-01 5.55598378e-01
5.07607818e-01 -1.21406026e-01 -1.76118746e-01 -1.27906337e-01
-2.99501717e-01 2.24425450e-01 5.59108853e-01 9.37546790e-01
-2.56124407e-01 -1.54223132e+00 -3.20374489e-01 7.19171250e-03
-3.77230287e-01 -1.43455386e-01 -1.50356263e-01 8.39519978e-01
1.72069982e-01 3.34891319e-01 2.54746139e-01 7.82074872e-03
3.12248886e-01 1.80825174e-01 7.94542193e-01 -6.51579797e-01
-4.51219350e-01 6.35233521e-01 -2.16556817e-01 -7.52042115e-01
-7.20169187e-01 -7.61291802e-01 -1.21277595e+00 -5.61995327e-01
1.11987583e-01 3.18855755e-02 9.89377722e-02 7.74266779e-01
1.39510572e-01 6.21522725e-01 3.99603695e-01 -1.40707457e+00
2.48082459e-01 -6.59889996e-01 -4.28683549e-01 6.22115076e-01
4.16798741e-01 -3.36257011e-01 2.56324738e-01 1.04690802e+00]
|
[11.059039115905762, -0.707206666469574]
|
664d4699-c60b-4d86-89eb-d99bca25e810
|
actor-critic-approach-for-temporal-predictive
| null | null |
https://openreview.net/forum?id=r1ln504YvH
|
https://openreview.net/pdf?id=r1ln504YvH
|
Actor-Critic Approach for Temporal Predictive Clustering
|
Due to the wider availability of modern electronic health records (EHR), patient care data is often being stored in the form of time-series. Clustering such time-series data is crucial for patient phenotyping, anticipating patients’ prognoses by identifying “similar” patients, and designing treatment guidelines that are tailored to homogeneous patient subgroups. In this paper, we develop a deep learning approach for clustering time-series data, where each cluster comprises patients who share similar future outcomes of interest (e.g., adverse events, the onset of comorbidities, etc.). The clustering is carried out by using our novel loss functions that encourage each cluster to have homogeneous future outcomes. We adopt actor-critic models to allow “back-propagation” through the sampling process that is required for assigning clusters to time-series inputs. Experiments on two real-world datasets show that our model achieves superior clustering performance over state-of-the-art benchmarks and identifies meaningful clusters that can be translated into actionable information for clinical decision-making.
|
['Mihaela van der Schaar', 'Changhee Lee']
|
2019-09-25
| null | null | null | null |
['patient-phenotyping', 'time-series-clustering']
|
['medical', 'time-series']
|
[-1.43077761e-01 5.13081811e-02 -3.80518585e-01 -6.74203992e-01
-1.07076824e+00 -1.27012253e-01 6.33723885e-02 9.58583415e-01
-1.23823218e-01 5.42093694e-01 6.08597457e-01 -3.46515238e-01
-5.38764358e-01 -5.41281939e-01 -3.45149785e-01 -8.28306198e-01
-5.80269456e-01 9.07423139e-01 -5.92060626e-01 4.79781151e-01
-3.16051543e-01 2.38889217e-01 -9.55615401e-01 3.90364349e-01
8.90393794e-01 9.65032995e-01 -9.43164974e-02 3.83257836e-01
1.21015020e-01 8.74495268e-01 -4.55903560e-01 -1.13364197e-02
1.55821770e-01 -5.37723899e-01 -4.65332329e-01 1.20706543e-01
-4.02836204e-01 1.46184102e-01 -3.24783534e-01 7.46321142e-01
6.61807716e-01 1.01624131e-02 4.06710833e-01 -1.04604506e+00
-4.09732223e-01 8.91258001e-01 -4.40622836e-01 -6.91697747e-02
-1.47628888e-01 2.48192489e-01 7.84748971e-01 -2.19742134e-02
3.83580476e-01 9.54185188e-01 7.87049830e-01 7.02645957e-01
-1.35141778e+00 -3.16683978e-01 3.59252274e-01 2.58207470e-01
-1.35972679e+00 -1.98869228e-01 8.85581732e-01 -5.67372978e-01
4.06670362e-01 4.79738235e-01 7.37711787e-01 1.18657386e+00
3.83899838e-01 6.27222359e-01 5.19238174e-01 1.02273956e-01
6.71971977e-01 -3.01311761e-01 1.82932198e-01 1.83895603e-01
-1.12567432e-01 -2.14263827e-01 1.74708396e-01 -4.80692804e-01
3.78992140e-01 9.99603152e-01 -1.68328449e-01 -2.15773493e-01
-1.45664322e+00 6.87021613e-01 5.90320230e-01 1.83507189e-01
-9.60107446e-01 1.15107134e-01 7.44342089e-01 9.09704864e-02
4.79872257e-01 5.27301669e-01 -6.18679881e-01 6.74831569e-02
-8.40405822e-01 -1.59361288e-01 3.83051574e-01 7.32390761e-01
1.20405950e-01 -3.63267720e-01 -7.25450933e-01 6.22440338e-01
1.49787739e-01 2.53849208e-01 9.14715409e-01 -9.62936938e-01
4.24735487e-01 9.14349675e-01 1.99523523e-01 -6.20783150e-01
-9.60158587e-01 -4.99816269e-01 -1.40569866e+00 -5.25971353e-01
2.50596941e-01 -5.90701461e-01 -8.79637301e-01 1.80940986e+00
2.93374896e-01 6.26623154e-01 2.12412491e-01 7.53561020e-01
3.54402602e-01 6.85923696e-01 4.48606491e-01 -6.10385656e-01
1.33794487e+00 -3.12944144e-01 -6.23646855e-01 2.99411714e-01
8.03174853e-01 -1.26047537e-01 7.11911201e-01 3.20059925e-01
-9.01398718e-01 -2.40221262e-01 -3.22275758e-01 3.36919010e-01
-3.82964872e-02 1.62054032e-01 7.30848610e-01 2.02594787e-01
-6.32233560e-01 7.36722827e-01 -1.39682817e+00 -3.97549033e-01
7.91926146e-01 4.74372685e-01 5.38947210e-02 2.05668136e-01
-1.06595254e+00 2.07695812e-01 2.41937622e-01 1.40196502e-01
-8.75865698e-01 -1.19380665e+00 -5.03276885e-01 2.27777869e-01
1.12743609e-01 -1.00734305e+00 8.94527733e-01 -8.53199542e-01
-1.19672728e+00 6.20309949e-01 -2.04730749e-01 -8.01826239e-01
5.01256108e-01 -6.83167726e-02 -6.28675818e-01 -7.36634582e-02
-4.49743792e-02 3.85460049e-01 5.76527476e-01 -8.05748403e-01
-6.82239115e-01 -6.42395556e-01 -6.32377028e-01 -7.65512213e-02
-4.05391276e-01 1.28418868e-02 -2.94107199e-01 -6.79547250e-01
-2.10315302e-01 -9.69920874e-01 -9.60815728e-01 -1.79757431e-01
-7.92812645e-01 -3.39880824e-01 6.58951759e-01 -6.89658582e-01
1.67989242e+00 -2.17775464e+00 3.37471068e-01 2.04432830e-01
6.27166986e-01 1.51432216e-01 1.77910075e-01 4.54495788e-01
-2.44182602e-01 7.05107823e-02 -3.02082747e-01 -3.97918284e-01
-2.81980932e-01 3.65534029e-03 -2.05404609e-01 5.89972615e-01
2.38993421e-01 8.04418445e-01 -9.64168429e-01 -3.44714671e-01
3.14505726e-01 5.09390175e-01 -6.49104774e-01 4.47999060e-01
-5.05679250e-01 1.03607476e+00 -8.43400121e-01 4.13669944e-01
1.92319766e-01 -9.56407845e-01 3.86901855e-01 -9.37210321e-02
1.62358463e-01 -7.85757825e-02 -5.74696779e-01 1.49888635e+00
-4.09466445e-01 2.83293307e-01 -3.63384038e-01 -1.26615465e+00
7.35278189e-01 5.56621015e-01 1.32969260e+00 -2.88878262e-01
2.90389329e-01 -2.08405763e-01 1.62769541e-01 -6.46481574e-01
-1.62404910e-01 -5.47812060e-02 -3.03269535e-01 4.29939628e-01
-5.64226031e-01 5.67518175e-01 -6.93627000e-02 -1.09865300e-01
1.26728797e+00 -5.85535407e-01 1.62555069e-01 3.93713340e-02
4.49856818e-01 2.78088376e-02 1.04594302e+00 2.74879098e-01
-2.08678424e-01 5.10682404e-01 6.38438761e-01 -7.46618986e-01
-1.22894871e+00 -9.52356219e-01 -2.19728217e-01 4.28377450e-01
-3.17584544e-01 -1.84668452e-01 -3.55452180e-01 -5.15631735e-01
1.62109867e-01 6.70338690e-01 -8.46233428e-01 -5.77658355e-01
-4.79995251e-01 -1.25613022e+00 2.11638153e-01 7.10955262e-01
-2.09182248e-01 -1.12819314e+00 -8.43273163e-01 6.88067853e-01
-1.60639107e-01 -6.26210868e-01 -5.34268498e-01 2.75025427e-01
-9.88341272e-01 -1.05783141e+00 -8.55215788e-01 -7.50194967e-01
7.99487829e-01 -3.58242095e-01 1.08801913e+00 -1.97485596e-01
-4.52545971e-01 1.05629332e-01 -3.29526544e-01 -4.42048728e-01
-2.95720398e-01 2.39012428e-02 -3.98873463e-02 6.17717803e-01
5.66139281e-01 -4.77863908e-01 -1.24046183e+00 2.78148465e-02
-6.75714433e-01 -6.98641911e-02 2.85293102e-01 8.31439674e-01
1.03766537e+00 1.99407456e-03 8.87222230e-01 -1.15275741e+00
5.24301231e-01 -1.06312692e+00 -3.66795242e-01 4.18340385e-01
-6.77726209e-01 -5.98170012e-02 1.19356287e+00 -5.15067339e-01
-6.42692268e-01 3.43790650e-01 2.01037869e-01 -8.05897474e-01
-2.89734334e-01 6.16529465e-01 -4.18126360e-02 8.16965520e-01
2.71392435e-01 8.49746391e-02 6.78116754e-02 -4.15907502e-01
1.82726696e-01 7.75085032e-01 3.90360087e-01 -2.87586510e-01
1.12960190e-01 6.74210548e-01 5.48433699e-03 -3.02359372e-01
-9.52449560e-01 -7.11941659e-01 -3.97338688e-01 2.80344021e-02
1.15175498e+00 -1.06285369e+00 -1.12460256e+00 1.01552330e-01
-7.09466696e-01 -2.59470254e-01 -5.02549648e-01 6.10519469e-01
-6.60750270e-01 3.24629731e-02 -6.08686864e-01 -5.69684386e-01
-5.35695195e-01 -1.02305865e+00 9.64851737e-01 1.48988202e-01
-5.83048284e-01 -1.24474645e+00 4.60399210e-01 4.56478223e-02
1.03214227e-01 8.17940950e-01 1.40886855e+00 -8.37200582e-01
-1.31387338e-01 -3.38709623e-01 6.49643242e-02 -7.88846314e-02
4.11795557e-01 9.37049910e-02 -5.35458446e-01 -3.81669551e-01
-9.89744663e-02 1.25596583e-01 6.99123979e-01 1.07052124e+00
2.01338696e+00 -4.87083435e-01 -7.99218714e-01 9.95562196e-01
1.21476769e+00 6.88841879e-01 4.88880843e-01 -8.00117627e-02
8.97377133e-01 7.07896948e-01 3.39343846e-01 9.45912421e-01
7.02851057e-01 4.57595706e-01 4.24760759e-01 -2.66419947e-01
4.25308406e-01 -1.05134353e-01 -2.86210217e-02 7.85077035e-01
2.08280534e-01 -2.47827053e-01 -1.20891476e+00 7.45865703e-01
-2.25402546e+00 -1.06511462e+00 -2.46560499e-01 2.31869698e+00
7.79209733e-01 -2.71618694e-01 4.21912193e-01 -1.41273886e-01
9.74051833e-01 -2.13018343e-01 -1.18105400e+00 1.76980030e-02
3.42234001e-02 -9.97791663e-02 4.14342225e-01 3.43194865e-02
-1.27242541e+00 4.80857491e-01 5.60445833e+00 2.49964789e-01
-1.49970901e+00 -9.25243348e-02 1.45628870e+00 -3.27788860e-01
-5.75407185e-02 -3.68739456e-01 -2.84132212e-01 9.57816839e-01
1.23015046e+00 -4.36419934e-01 2.58821368e-01 3.71371597e-01
8.22116137e-01 5.98085523e-01 -1.33899045e+00 9.17343795e-01
-5.76569915e-01 -1.32708180e+00 -1.26790091e-01 -7.09136128e-02
7.89646983e-01 1.34814978e-01 2.13314220e-01 1.54482657e-02
7.32847869e-01 -9.37805593e-01 2.11945295e-01 8.88358235e-01
7.74777949e-01 -8.53272200e-01 6.28242731e-01 4.43095416e-02
-9.38419223e-01 -6.20075822e-01 -1.53282523e-01 3.71289909e-01
2.87488103e-01 9.19246078e-01 -9.32430863e-01 4.28020537e-01
5.02338767e-01 1.24968648e+00 -3.11776310e-01 1.22286868e+00
3.18883419e-01 9.41815674e-01 -1.36580914e-01 1.41943872e-01
1.72753185e-01 -3.14734608e-01 1.13383584e-01 8.02535415e-01
3.99247497e-01 2.06488460e-01 4.26536918e-01 6.00003183e-01
-7.29146525e-02 1.33618742e-01 -1.26784891e-01 -1.65959507e-01
5.58844805e-01 1.18762445e+00 -7.80100286e-01 -5.24898469e-01
-1.86344951e-01 6.25116348e-01 1.92871466e-01 2.80019701e-01
-8.68579268e-01 -1.84322357e-01 6.82931900e-01 1.97597682e-01
1.30752861e-01 2.77073205e-01 -5.28645217e-01 -1.13198328e+00
-1.98523700e-01 -7.08893478e-01 8.62309992e-01 -2.07804933e-01
-1.78977156e+00 5.94238997e-01 -4.94984925e-01 -1.63131011e+00
-2.74234772e-01 8.25024024e-03 -9.01891172e-01 6.76254988e-01
-1.09436965e+00 -7.97129869e-01 -1.95431858e-01 7.50089049e-01
2.19953448e-01 -1.00606337e-01 8.22978854e-01 4.39456731e-01
-1.06942475e+00 5.36304355e-01 5.61149180e-01 2.58632869e-01
7.30672300e-01 -1.23490703e+00 2.69466072e-01 3.17060471e-01
-3.75535935e-01 4.98780996e-01 3.39749515e-01 -7.27142334e-01
-1.16851544e+00 -1.85138083e+00 7.71284759e-01 -3.32131594e-01
6.65921569e-01 -8.47154185e-02 -1.05931425e+00 5.90288281e-01
-2.56247997e-01 5.26596382e-02 1.13715196e+00 1.21527366e-01
-3.65075059e-02 -4.36664134e-01 -1.13189292e+00 5.73811173e-01
5.55820465e-01 -1.61966100e-01 -1.36643052e-01 7.20526755e-01
7.85182059e-01 -5.62243834e-02 -1.34307277e+00 3.50369781e-01
2.40521729e-01 -5.00540674e-01 8.32117677e-01 -1.22999275e+00
4.75707799e-01 -2.12197617e-01 3.04505408e-01 -1.55501413e+00
-6.08080149e-01 -9.55577493e-01 -2.86869287e-01 8.42230678e-01
4.56292003e-01 -5.14522672e-01 8.04626167e-01 8.72864604e-01
-1.14535809e-01 -9.18229759e-01 -7.90822446e-01 -4.01102215e-01
3.04265004e-02 -2.32923813e-02 1.04445779e+00 1.34849930e+00
3.21542382e-01 1.49830312e-01 -3.86076450e-01 2.08755910e-01
6.21041954e-01 3.67938191e-01 3.59935164e-01 -1.40724361e+00
-1.96437806e-01 -6.62116289e-01 -2.82687724e-01 -3.04502994e-01
4.38047647e-02 -9.53740656e-01 -1.64803818e-01 -1.49454832e+00
3.70536208e-01 -7.81680226e-01 -8.63693476e-01 4.41469848e-01
-5.62495708e-01 -4.30282891e-01 -9.21275094e-02 3.57378304e-01
-9.25193071e-01 5.27816296e-01 9.32161331e-01 -1.21884875e-01
-6.56285226e-01 4.26709682e-01 -7.10632861e-01 6.02860153e-01
9.58511472e-01 -5.47335446e-01 -2.84151524e-01 -2.55249083e-01
3.11344583e-02 6.70090437e-01 2.16743559e-01 -9.62364256e-01
3.32392752e-01 -3.30365360e-01 4.40519512e-01 -5.05213976e-01
-6.76597357e-02 -1.03295743e+00 5.00306964e-01 8.43606830e-01
-8.64596725e-01 3.11763942e-01 -1.72843426e-01 9.37296212e-01
-2.09294319e-01 6.67901576e-01 7.52149940e-01 1.18450016e-01
-1.04541264e-01 8.06306243e-01 -3.64641339e-01 1.72617212e-01
1.51554930e+00 3.51442933e-01 3.37263197e-02 -4.93683875e-01
-1.05865979e+00 8.66923690e-01 2.23945513e-01 3.78854543e-01
3.80437553e-01 -1.37134349e+00 -9.25060391e-01 -2.37264559e-02
2.35842839e-01 1.94264725e-01 5.17563283e-01 9.90322530e-01
-3.12474310e-01 2.16319114e-01 6.22612163e-02 -7.94461012e-01
-1.10737681e+00 1.14240384e+00 3.15740108e-01 -4.21454161e-01
-8.38495791e-01 3.21451128e-01 1.10646792e-01 -1.71935141e-01
4.54269707e-01 -4.89132881e-01 -2.52803415e-01 1.38178766e-01
5.13664305e-01 4.30651605e-01 -8.25547799e-02 -2.97589600e-01
-3.42507571e-01 3.56868923e-01 -2.49305800e-01 4.94732022e-01
1.74133050e+00 -7.30959512e-03 -5.76932840e-02 6.60194993e-01
1.27711189e+00 -6.03136361e-01 -1.41534019e+00 -1.59494296e-01
2.02965677e-01 -1.65115315e-02 -1.44564509e-01 -6.60610199e-01
-1.39293468e+00 8.88100386e-01 7.51058698e-01 3.24673980e-01
1.32514930e+00 1.97588056e-01 1.00907195e+00 1.70673847e-01
5.08471578e-02 -8.95029962e-01 -2.98706561e-01 6.71124533e-02
3.85182232e-01 -1.26663375e+00 -2.71597266e-01 9.90957469e-02
-8.16753328e-01 7.59869635e-01 2.07466841e-01 -2.58364379e-01
9.37492669e-01 1.47109758e-02 3.16007167e-01 -2.21332103e-01
-1.12612867e+00 1.97230026e-01 6.93199635e-02 4.45443124e-01
5.00143170e-01 6.79626107e-01 -1.13142729e-01 8.05265546e-01
2.97916103e-02 6.91243075e-03 2.54718661e-01 4.82322812e-01
9.38979834e-02 -9.11630929e-01 -3.33809853e-01 9.87230062e-01
-6.22628510e-01 1.26470387e-01 -1.81368321e-01 1.21609569e-01
-1.19027290e-02 7.80120492e-01 3.70792270e-01 -2.08485708e-01
2.10623100e-01 2.23231148e-02 -5.57778515e-02 -5.23671150e-01
-8.01085174e-01 2.43612632e-01 -4.08998936e-01 -5.97604215e-01
-1.91372901e-01 -8.34729612e-01 -1.50723767e+00 -1.33156434e-01
2.34207287e-01 3.37908238e-01 2.68762589e-01 6.96602881e-01
1.05782449e+00 1.10047257e+00 1.18842947e+00 -4.36787158e-01
-5.70488870e-01 -6.05289340e-01 -4.01365429e-01 7.41060674e-01
5.20448565e-01 -3.76331322e-02 1.18957809e-03 3.73204023e-01]
|
[7.878611087799072, 6.178296089172363]
|
423d389b-8368-49ae-8c59-72424faceac7
|
rocnet-3d-robust-registration-of-point-clouds
|
2303.07963
| null |
https://arxiv.org/abs/2303.07963v1
|
https://arxiv.org/pdf/2303.07963v1.pdf
|
RoCNet: 3D Robust Registration of Point-Clouds using Deep Learning
|
This paper introduces a new method for 3D point cloud registration based on deep learning. The architecture is composed of three distinct blocs: (i) an encoder composed of a convolutional graph-based descriptor that encodes the immediate neighbourhood of each point and an attention mechanism that encodes the variations of the surface normals. Such descriptors are refined by highlighting attention between the points of the same set and then between the points of the two sets. (ii) a matching process that estimates a matrix of correspondences using the Sinkhorn algorithm. (iii) Finally, the rigid transformation between the two point clouds is calculated by RANSAC using the Kc best scores from the correspondence matrix. We conduct experiments on the ModelNet40 dataset, and our proposed architecture shows very promising results, outperforming state-of-the-art methods in most of the simulated configurations, including partial overlap and data augmentation with Gaussian noise.
|
['Catherine Achard', 'Brahim Tamadazte', 'Karim Slimani']
|
2023-03-14
| null | null | null | null |
['point-cloud-registration']
|
['computer-vision']
|
[-1.33518308e-01 -1.05629072e-01 2.89405525e-01 -2.12687910e-01
-6.54026389e-01 -2.55661994e-01 8.18227291e-01 3.58950049e-01
-3.27056468e-01 1.71761448e-03 -2.09798649e-01 1.84177086e-02
-2.69842982e-01 -7.24623442e-01 -1.06126571e+00 -6.29360795e-01
-3.02162200e-01 1.02466989e+00 4.51147616e-01 -4.58355784e-01
5.71584642e-01 1.09164357e+00 -1.63741469e+00 5.56232128e-03
5.29308677e-01 1.37308002e+00 4.47614826e-02 3.16638350e-01
-1.04719155e-01 1.70194492e-01 -2.41113529e-01 -1.88160285e-01
6.93918884e-01 1.01799689e-01 -6.72377884e-01 1.02404252e-01
8.26839626e-01 9.59097967e-02 -2.45953307e-01 9.89774227e-01
2.21846789e-01 1.58232242e-01 5.29042661e-01 -1.08657503e+00
-4.52118635e-01 -7.52827823e-02 -6.30815268e-01 -1.07696004e-01
4.46646839e-01 -9.71399918e-02 8.18848848e-01 -1.06479371e+00
7.43097603e-01 1.17870903e+00 8.92359674e-01 6.07943945e-02
-1.15951467e+00 -4.46622133e-01 -2.64848173e-01 2.23092020e-01
-1.69208813e+00 -2.50954986e-01 8.18586707e-01 -7.28216231e-01
1.14312291e+00 1.95580125e-01 8.18509519e-01 4.11846638e-01
2.19820410e-01 -1.26500232e-02 5.73277056e-01 -3.93970698e-01
2.05710664e-01 -8.60937759e-02 2.25006007e-02 7.76711285e-01
7.55613223e-02 1.65430486e-01 -2.41908923e-01 -3.48353982e-01
9.98254061e-01 1.80762485e-01 -1.40187830e-01 -1.04067230e+00
-1.25535953e+00 6.57853723e-01 1.09405828e+00 3.77688736e-01
-6.36458516e-01 1.71539783e-01 -3.47879678e-02 4.90562655e-02
3.66079450e-01 3.28620493e-01 -2.20783994e-01 3.85794908e-01
-8.71221066e-01 2.14462340e-01 7.08068728e-01 1.14567173e+00
1.20870221e+00 -3.54786336e-01 3.04679632e-01 4.33807880e-01
6.52034760e-01 3.98344249e-01 1.75097004e-01 -7.67798543e-01
4.65403229e-01 9.57211196e-01 8.36651921e-02 -1.42911482e+00
-5.01315892e-01 -2.84299046e-01 -8.20576131e-01 6.02937937e-01
1.00456618e-01 1.81765959e-01 -8.44096303e-01 1.28163898e+00
4.69338149e-01 7.23387420e-01 -2.75498569e-01 1.01618624e+00
8.70840549e-01 3.91253054e-01 -3.40832353e-01 2.67168134e-01
1.22698605e+00 -8.15403759e-01 -2.35485286e-01 3.22355353e-03
3.01693201e-01 -8.78740609e-01 3.49480778e-01 -1.01657547e-01
-1.27343619e+00 -8.61557841e-01 -1.25077713e+00 -2.23004445e-01
-4.40250903e-01 -1.53583502e-02 3.35574031e-01 4.73331362e-02
-1.50569344e+00 9.66759562e-01 -9.05170381e-01 -5.73033452e-01
2.92343616e-01 7.14521766e-01 -6.81619406e-01 2.19567701e-01
-6.20391786e-01 8.17269206e-01 1.98556036e-01 3.80248606e-01
-4.67947841e-01 -6.66922271e-01 -1.05784714e+00 9.47820693e-02
-1.91086113e-01 -1.01917815e+00 6.12127304e-01 -7.21844494e-01
-1.35837913e+00 1.22537351e+00 -6.28444329e-02 -3.37595493e-01
4.84163314e-01 -2.21372262e-01 4.24582278e-03 -5.28105488e-03
1.55231088e-01 7.84983754e-01 8.29937756e-01 -1.42702043e+00
-5.47261298e-01 -6.54440045e-01 -2.32800752e-01 2.21405193e-01
1.95923701e-01 -4.60497960e-02 -8.66712689e-01 -1.90219864e-01
7.64522135e-01 -1.15462482e+00 -3.41794282e-01 1.84283361e-01
-4.41800356e-01 -2.47345373e-01 7.72515118e-01 -4.61303443e-01
6.25211954e-01 -2.26352429e+00 3.23110908e-01 6.02005839e-01
3.02692682e-01 1.25144124e-01 -2.44007900e-01 4.47809815e-01
-4.44432825e-01 3.92261930e-02 -9.68006030e-02 -6.50046289e-01
-4.41597514e-02 -1.28514424e-01 -1.05611019e-01 7.99429655e-01
2.74664223e-01 8.43246698e-01 -7.26462364e-01 -9.74087417e-02
6.03127956e-01 7.61775732e-01 -3.14648539e-01 4.13200706e-01
8.52391347e-02 4.06174242e-01 -3.02702367e-01 2.61222154e-01
9.91544247e-01 -3.05894613e-01 -2.95130104e-01 -3.61480892e-01
-2.32250452e-01 3.27466697e-01 -1.45642722e+00 1.90393436e+00
-1.29442856e-01 3.32692713e-01 -4.69665639e-02 -6.45898104e-01
1.32780302e+00 4.37191911e-02 8.52508187e-01 -2.39595547e-01
2.15756565e-01 1.18632466e-01 -1.06202200e-01 -1.35284215e-01
4.86827433e-01 3.59226853e-01 2.50148445e-01 1.60227180e-01
1.25900507e-01 -5.64162970e-01 -1.89295784e-01 5.63122157e-04
9.77926970e-01 8.32122564e-02 7.56945685e-02 -3.10805857e-01
7.94823706e-01 -7.91069642e-02 8.49194601e-02 3.36917967e-01
1.27582088e-01 9.48089659e-01 2.79711545e-01 -7.90523708e-01
-1.14643288e+00 -9.59094465e-01 -1.01174906e-01 3.25708598e-01
6.17694080e-01 -5.13680935e-01 -6.26629531e-01 -3.63281071e-01
3.57146800e-01 2.78977096e-01 -7.74109483e-01 -6.73976317e-02
-4.38339531e-01 -1.56615853e-01 -1.58144403e-02 5.14989495e-01
4.21842754e-01 -8.75350773e-01 -5.39400041e-01 -2.03829363e-01
1.24946475e-01 -1.16329181e+00 -2.31161132e-01 2.40395308e-01
-1.10107076e+00 -1.24809778e+00 -3.38099748e-01 -9.78852153e-01
8.62561226e-01 2.88011253e-01 1.11069214e+00 3.24187547e-01
1.61797062e-01 1.49542511e-01 -1.17654629e-01 -3.45513374e-01
-2.18458921e-01 -3.16388486e-03 -6.89928159e-02 2.38173351e-01
3.94387722e-01 -5.42116225e-01 -4.49710637e-01 4.58038241e-01
-4.77415502e-01 -5.63541129e-02 3.19932133e-01 5.15469134e-01
9.59225655e-01 -2.88887560e-01 -4.70218569e-01 -3.98925275e-01
2.17019171e-01 -3.49563122e-01 -8.86780322e-01 -4.42117676e-02
-8.65155980e-02 3.58165912e-02 2.18486324e-01 -8.84389579e-02
-3.61038625e-01 5.09735644e-01 -4.92555723e-02 -9.58821535e-01
-3.96180630e-01 9.89421457e-02 7.90616423e-02 -6.04804933e-01
5.38802564e-01 -8.37178230e-02 5.35872355e-02 -5.57188451e-01
3.63063097e-01 5.28480351e-01 6.95225179e-01 -2.57236063e-01
1.19896543e+00 8.27642381e-01 3.31617624e-01 -9.19769287e-01
-2.68081188e-01 -7.55495727e-01 -1.34247696e+00 -1.19214088e-01
1.12508428e+00 -7.60774374e-01 -8.45950902e-01 5.26037753e-01
-1.57849002e+00 -1.24435704e-02 -3.73146892e-01 6.18538857e-01
-7.63865232e-01 2.53058165e-01 -4.28525299e-01 -3.29639405e-01
-3.43135357e-01 -1.44248688e+00 1.54823303e+00 2.11310282e-01
-5.94396256e-02 -7.66831636e-01 4.18730855e-01 1.36822671e-01
7.86867067e-02 4.65509623e-01 8.12579691e-01 -7.09568381e-01
-9.92035866e-01 -4.68688190e-01 -2.24020153e-01 1.84361041e-01
2.18455400e-02 3.00604850e-01 -9.40548956e-01 -4.36107904e-01
-5.33379279e-02 1.67008340e-01 5.62801361e-01 3.46802235e-01
8.39764237e-01 1.87947631e-01 -5.30829966e-01 1.13694632e+00
1.53725886e+00 1.03155382e-01 7.73221254e-01 5.81600666e-01
9.46352839e-01 3.06408197e-01 2.01162681e-01 2.55069196e-01
3.33994448e-01 8.99977624e-01 1.02629042e+00 -2.40461141e-01
-4.47853506e-02 -1.85988878e-03 -1.73523083e-01 9.60664511e-01
-4.91970897e-01 1.68554693e-01 -1.07633877e+00 3.92057627e-01
-1.89225650e+00 -5.99561334e-01 -3.99160922e-01 2.43881488e+00
-1.40237743e-02 4.59549911e-02 -2.50424922e-01 -1.03448361e-01
7.77845442e-01 2.28968054e-01 -3.37491602e-01 -4.00456309e-01
9.77835730e-02 5.38841546e-01 3.87536913e-01 4.06861782e-01
-1.34506977e+00 1.00737774e+00 6.10341883e+00 2.22465813e-01
-1.03835118e+00 -1.92362472e-01 1.57242358e-01 3.20129514e-01
4.42485996e-02 2.63029039e-02 -5.90354741e-01 2.95781970e-01
5.35365880e-01 3.43539923e-01 2.92214155e-01 7.41197705e-01
-2.74548799e-01 1.10517696e-01 -1.28945529e+00 1.08571804e+00
2.71970928e-01 -1.45405185e+00 7.44855553e-02 2.29268685e-01
8.31286550e-01 6.69649780e-01 -1.19765490e-01 -2.41118371e-01
1.28335565e-01 -9.31864440e-01 7.70698488e-01 9.22460616e-01
6.19681537e-01 -7.17846155e-01 9.20919955e-01 1.99717864e-01
-1.34641623e+00 2.61311978e-01 -6.51387095e-01 1.87583417e-01
-1.75935507e-01 2.51290739e-01 -7.45358765e-01 7.85802364e-01
9.72628117e-01 8.92914355e-01 -6.01193488e-01 1.24870348e+00
-2.59333849e-01 -1.33663327e-01 -4.99584705e-01 2.59640217e-01
2.64567047e-01 -5.78547359e-01 7.16600180e-01 8.91262829e-01
4.71377015e-01 -1.38754085e-01 9.81707126e-02 9.99847472e-01
-8.48022327e-02 4.27283384e-02 -7.60920882e-01 6.10384285e-01
3.76579612e-01 1.34487271e+00 -6.55546725e-01 -1.34357870e-01
-3.91619265e-01 9.30231392e-01 4.51817453e-01 2.43090734e-01
-5.28614998e-01 -3.50801557e-01 1.01096046e+00 2.06592858e-01
6.19261980e-01 -4.94361371e-01 -2.62805730e-01 -8.06725085e-01
-6.79355173e-04 -4.37837720e-01 9.33815092e-02 -1.15021122e+00
-1.11802340e+00 9.16466773e-01 -2.97747344e-01 -1.41703677e+00
-1.46152988e-01 -5.96494555e-01 -6.55820131e-01 1.18233681e+00
-1.36886263e+00 -1.12714410e+00 -7.52284765e-01 5.94330907e-01
7.17842951e-02 -1.54177383e-01 8.05261731e-01 1.83797583e-01
-2.01153159e-01 2.48248115e-01 1.38591394e-01 2.41402641e-01
3.16324323e-01 -1.07717407e+00 1.00777626e+00 5.33577144e-01
4.48731542e-01 6.24459684e-01 1.85251340e-01 -6.12646163e-01
-1.29950905e+00 -8.58673096e-01 8.70290279e-01 -5.78425825e-01
5.10613441e-01 -4.18993920e-01 -1.07167339e+00 6.87680840e-01
1.01656988e-01 2.19252884e-01 1.87212244e-01 -1.69886529e-01
-2.20017239e-01 -7.96963722e-02 -1.00493300e+00 3.23041022e-01
1.01190364e+00 -5.61730802e-01 -5.25687993e-01 3.57586533e-01
5.90945005e-01 -9.83198702e-01 -9.64739144e-01 4.24540669e-01
3.71179312e-01 -1.13014090e+00 1.23889077e+00 -5.92576146e-01
2.31704712e-01 -5.67456663e-01 -1.03379920e-01 -1.26973724e+00
-7.57098079e-01 -5.06381750e-01 1.27073929e-01 8.83472264e-01
4.86295819e-02 -3.29122454e-01 7.08031356e-01 4.02480364e-01
-2.68756270e-01 -7.24455774e-01 -9.70500171e-01 -5.47085643e-01
-2.78518677e-01 -3.36254090e-01 9.57461894e-01 9.66244280e-01
-4.61285830e-01 2.46426836e-01 2.03681394e-01 6.60926461e-01
5.45959413e-01 1.05680220e-01 1.02752900e+00 -1.51559174e+00
1.77846551e-01 -4.97507811e-01 -1.21949756e+00 -9.44722235e-01
1.88297659e-01 -1.06030583e+00 9.87310708e-02 -1.42294109e+00
-1.01407707e-01 -5.40745020e-01 -2.04908967e-01 3.26342940e-01
1.47814333e-01 2.08891809e-01 2.99248278e-01 4.22419995e-01
-3.92728597e-01 4.50047553e-01 1.06591332e+00 5.92354406e-03
-2.50386506e-01 -8.65250453e-02 -1.67668343e-01 1.01222026e+00
4.24189687e-01 -4.39847559e-01 1.95339903e-01 -6.58122659e-01
6.55782502e-03 -1.46696180e-01 5.71704328e-01 -1.48779976e+00
5.29999435e-01 2.68883765e-01 4.99547660e-01 -9.35016036e-01
4.64699179e-01 -1.33522201e+00 4.98689383e-01 3.76342565e-01
-7.82289803e-02 5.17460883e-01 2.63600826e-01 4.93595600e-01
-1.18967786e-01 -1.14669695e-01 8.68196130e-01 9.65359583e-02
-5.90048909e-01 5.70033193e-01 3.14067572e-01 -4.03312683e-01
1.06520367e+00 -4.32091296e-01 -2.49914522e-03 -2.06943169e-01
-5.46461344e-01 4.21937443e-02 7.98934281e-01 4.87731755e-01
7.37025678e-01 -1.58957124e+00 -6.71001494e-01 8.10953259e-01
1.10952884e-01 4.76856261e-01 -1.37886301e-01 7.24128067e-01
-8.81276429e-01 2.48979181e-01 -3.96922857e-01 -1.12970304e+00
-1.13915849e+00 4.77717251e-01 6.96095824e-01 -2.79753376e-02
-6.06685817e-01 7.55717039e-01 3.69888656e-02 -6.57878578e-01
1.95351496e-01 -5.63455343e-01 -2.57026643e-01 -5.70606664e-02
2.55369097e-01 4.35224116e-01 6.47651613e-01 -1.33226800e+00
-6.80646122e-01 1.37958336e+00 2.46249646e-01 1.46109313e-01
1.54476881e+00 1.88349932e-01 -4.79564071e-01 1.55982986e-01
1.44783425e+00 -2.45831087e-01 -1.15492737e+00 -2.71313697e-01
1.26374781e-01 -6.11806571e-01 4.53563482e-02 -2.56020665e-01
-1.30900156e+00 8.64466548e-01 7.65802264e-01 1.36788309e-01
8.66774917e-01 1.85175404e-01 6.09572411e-01 3.28938544e-01
4.99549091e-01 -5.23756742e-01 -1.55670077e-01 7.93292761e-01
1.06343389e+00 -1.12928224e+00 3.28364857e-02 -3.44853163e-01
-2.34486520e-01 1.21525383e+00 3.29116076e-01 -7.52300322e-01
9.71109152e-01 2.70661563e-02 1.63569838e-01 -6.71268404e-01
-3.56466264e-01 -4.40589309e-01 7.38564193e-01 6.73339725e-01
1.22498035e-01 -3.18613231e-01 1.15891233e-01 -1.77301541e-02
-2.48187110e-01 -2.00850785e-01 -2.66724303e-02 4.94325668e-01
-1.80101439e-01 -7.81795681e-01 -5.93063891e-01 1.55421585e-01
1.13840811e-01 1.01466484e-01 -6.13871455e-01 9.46468115e-01
2.43643716e-01 6.50370598e-01 8.01912367e-01 -7.10820377e-01
6.93488300e-01 -1.27296105e-01 3.52539331e-01 -5.45077026e-01
-8.87700379e-01 5.87103255e-02 -3.94314587e-01 -9.81114626e-01
-4.76181328e-01 -6.03494167e-01 -1.28387237e+00 -3.27987075e-01
-3.97974998e-01 3.16556059e-02 9.83711600e-01 6.32158041e-01
6.39256895e-01 2.38997310e-01 7.82688916e-01 -1.43555903e+00
-2.39346206e-01 -7.18925416e-01 -5.48277557e-01 8.57313275e-01
2.68617809e-01 -6.49414539e-01 -2.38760591e-01 -1.56168818e-01]
|
[7.731423854827881, -2.994391918182373]
|
ab9d34b7-e258-4eb7-9bc0-779da61a60a8
|
learning-roi-transformer-for-oriented-object
| null | null |
http://openaccess.thecvf.com/content_CVPR_2019/html/Ding_Learning_RoI_Transformer_for_Oriented_Object_Detection_in_Aerial_Images_CVPR_2019_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2019/papers/Ding_Learning_RoI_Transformer_for_Oriented_Object_Detection_in_Aerial_Images_CVPR_2019_paper.pdf
|
Learning RoI Transformer for Oriented Object Detection in Aerial Images
|
Object detection in aerial images is an active yet challenging task in computer vision because of the bird's-eye view perspective, the highly complex backgrounds, and the variant appearances of objects. Especially when detecting densely packed objects in aerial images, methods relying on horizontal proposals for common object detection often introduce mismatches between the Region of Interests (RoIs) and objects. This leads to the common misalignment between the final object classification confidence and localization accuracy. In this paper, we propose a RoI Transformer to address these problems. The core idea of RoI Transformer is to apply spatial transformations on RoIs and learn the transformation parameters under the supervision of oriented bounding box (OBB) annotations. RoI Transformer is with lightweight and can be easily embedded into detectors for oriented object detection. Simply apply the RoI Transformer to light head RCNN has achieved state-of-the-art performances on two common and challenging aerial datasets, i.e., DOTA and HRSC2016, with a neglectable reduction to detection speed. Our RoI Transformer exceeds the deformable Position Sensitive RoI pooling when oriented bounding-box annotations are available. Extensive experiments have also validated the flexibility and effectiveness of our RoI Transformer.
|
[' Qikai Lu', ' Gui-Song Xia', ' Yang Long', ' Nan Xue', 'Jian Ding']
|
2019-06-01
| null | null | null |
cvpr-2019-6
|
['object-detection-in-aerial-images']
|
['computer-vision']
|
[ 3.14864993e-01 -1.73051998e-01 3.57167423e-01 -3.29952538e-01
-3.11272442e-01 -6.48411870e-01 3.49415690e-01 -3.59472305e-01
-7.03640640e-01 3.99531215e-01 -3.38165402e-01 3.12507758e-03
4.67785262e-02 -6.42629683e-01 -7.45957851e-01 -8.10311258e-01
1.60994649e-01 -9.53702778e-02 9.80627775e-01 -2.18066648e-01
-6.09523095e-02 7.07141578e-01 -1.70565808e+00 4.66406733e-01
6.44016087e-01 1.35611165e+00 5.76622307e-01 5.05707502e-01
2.54278213e-01 6.55178249e-01 -5.69285154e-01 -4.04484898e-01
8.17204773e-01 -3.75247672e-02 -5.36645234e-01 1.38601661e-01
7.96804845e-01 -6.21362209e-01 -2.59629190e-01 1.13026297e+00
5.21297395e-01 1.05897173e-01 5.64412177e-01 -1.22680676e+00
-6.56214058e-01 4.35014248e-01 -9.04571533e-01 6.47905290e-01
2.38952916e-02 3.44630063e-01 8.74806702e-01 -1.15680575e+00
2.86203921e-01 1.11087286e+00 7.80952632e-01 4.49505955e-01
-1.08597136e+00 -5.99189460e-01 5.46807110e-01 1.52489498e-01
-1.66268301e+00 -2.38509431e-01 3.69326323e-01 -4.76144880e-01
7.18485534e-01 5.43299556e-01 4.54320133e-01 8.64561975e-01
2.18458120e-02 7.30410159e-01 1.12245512e+00 -2.46982947e-01
1.82170365e-02 3.93763661e-01 4.01668511e-02 7.19976783e-01
3.88631582e-01 1.01691946e-01 -5.51211350e-02 2.17468217e-01
6.85825288e-01 5.07900000e-01 -5.19596517e-01 -3.45764875e-01
-1.23315620e+00 5.29265165e-01 9.48301017e-01 5.84834777e-02
-1.67674407e-01 2.19619721e-02 9.63032469e-02 -8.11847895e-02
4.45772022e-01 4.04067904e-01 -5.16124249e-01 7.22807109e-01
-8.96200120e-01 2.19213933e-01 1.93366036e-01 1.17179608e+00
4.43789721e-01 -1.26446486e-01 -5.81119835e-01 5.20711303e-01
3.14686596e-01 7.25090683e-01 3.46033335e-01 -3.19708556e-01
6.10111177e-01 9.32044089e-01 2.56769210e-01 -1.07881570e+00
-6.04139268e-01 -5.26634157e-01 -7.19199717e-01 3.90718132e-01
7.22359657e-01 -7.09229633e-02 -9.58270967e-01 1.28054535e+00
4.23575133e-01 -2.35238895e-01 -1.88154176e-01 1.30190468e+00
1.13862276e+00 5.97073317e-01 -7.03679696e-02 2.54143268e-01
1.50617433e+00 -1.04274988e+00 -3.66249532e-01 -5.75039864e-01
4.47866082e-01 -6.77448988e-01 1.07975972e+00 1.59569718e-02
-9.58682775e-01 -6.94809914e-01 -9.82931852e-01 -1.84691846e-01
-5.91705322e-01 9.03696358e-01 4.26967889e-01 5.53377211e-01
-7.65501857e-01 3.70365113e-01 -6.35978580e-01 -4.93025213e-01
6.71380579e-01 4.91258830e-01 -2.56603301e-01 2.05049384e-03
-6.70682192e-01 9.09468055e-01 4.76524949e-01 4.52918679e-01
-8.18043709e-01 -6.26310050e-01 -6.49697244e-01 2.21865550e-02
7.06799448e-01 -4.01522368e-01 9.20116127e-01 -7.87533879e-01
-1.18886828e+00 8.21439683e-01 3.07109445e-01 -5.10023236e-01
7.30898798e-01 -2.67060757e-01 -1.32893011e-01 -7.43431002e-02
-8.68866220e-03 9.15134549e-01 1.14745891e+00 -8.40085030e-01
-1.06299639e+00 -5.53928852e-01 2.90890008e-01 1.26312435e-01
-3.07431966e-01 3.16804916e-01 -2.06429824e-01 -6.13462985e-01
1.58573672e-01 -9.17252064e-01 -1.71986893e-01 5.98333061e-01
-3.62646461e-01 -2.79594630e-01 1.18757606e+00 -5.36283255e-01
1.01203918e+00 -2.10063338e+00 1.61790654e-01 -2.82934815e-01
9.11976397e-02 4.25889671e-01 -1.07730158e-01 -1.53699711e-01
-3.85418907e-02 -7.72145689e-02 -2.62482882e-01 7.26987198e-02
-1.37293577e-01 -2.63010710e-01 -6.92052186e-01 7.41064787e-01
5.47779918e-01 9.64522421e-01 -6.42257154e-01 -4.80499357e-01
4.09252048e-01 3.63992274e-01 -5.02107143e-01 2.83563733e-01
-4.53627966e-02 1.73623249e-01 -3.31522644e-01 9.81579542e-01
8.33889961e-01 -1.16753116e-01 -3.67769629e-01 -7.18223691e-01
-3.51513803e-01 -1.57203868e-01 -1.23703206e+00 1.15387869e+00
-1.18682057e-01 7.62804389e-01 -1.59601960e-02 -5.96781433e-01
8.47826540e-01 -1.18390769e-01 -6.11751378e-02 -3.57520342e-01
3.92549068e-01 3.27497423e-02 1.96574524e-01 -4.13209230e-01
4.31843907e-01 1.05448797e-01 1.51078269e-01 -1.04279583e-02
1.53393641e-01 -2.71879643e-01 2.35742152e-01 -2.90132761e-01
7.63463736e-01 3.54786634e-01 5.47143817e-01 -4.08564448e-01
5.10327101e-01 -2.64649063e-01 4.16942149e-01 5.71838915e-01
-2.24428698e-01 8.28702569e-01 1.81959957e-01 -7.57928193e-01
-6.81643844e-01 -8.60048473e-01 -5.12539744e-01 1.23922682e+00
3.83866847e-01 -6.95853904e-02 -7.78293133e-01 -1.12504315e+00
-1.35072649e-01 3.92832190e-01 -8.06564033e-01 -2.05697641e-02
-4.26911294e-01 -9.62743819e-01 4.96722966e-01 7.39801288e-01
8.73783588e-01 -9.93650734e-01 -1.30546463e+00 -1.84933618e-01
-1.42922714e-01 -1.50078344e+00 -4.96854693e-01 3.01899731e-01
-6.12243950e-01 -1.12145448e+00 -9.16320384e-01 -4.95396554e-01
9.17651653e-01 8.02969635e-01 7.26496279e-01 -1.72171161e-01
-9.11157250e-01 3.57727185e-02 -3.78941566e-01 -7.79439867e-01
2.65495628e-01 -3.02269254e-02 6.37115464e-02 1.49339065e-01
1.30763799e-01 7.43482681e-03 -7.98503697e-01 1.00132990e+00
-9.59609926e-01 -1.76482648e-01 7.06712782e-01 7.19122827e-01
4.59269702e-01 6.68241903e-02 -3.28973494e-02 -4.20605004e-01
-4.45420779e-02 -6.18797131e-02 -1.00824440e+00 3.70552778e-01
-9.08078849e-02 -2.81633526e-01 5.22001982e-01 -4.95885998e-01
-8.19235146e-01 3.41111034e-01 3.13771904e-01 -3.50254267e-01
-2.46451274e-01 -2.93965369e-01 -1.97675452e-01 -4.48369861e-01
8.07196856e-01 1.16663445e-02 -4.50207442e-01 -1.98721647e-01
1.55787334e-01 5.87671638e-01 3.60989779e-01 2.87988596e-02
1.04271805e+00 7.53247738e-01 -1.23344004e-01 -8.89981151e-01
-1.19401443e+00 -5.05379617e-01 -9.91601467e-01 -2.90480018e-01
1.00751543e+00 -1.02563047e+00 -5.28586745e-01 4.07881618e-01
-1.17516673e+00 -2.51135916e-01 -4.53580797e-01 3.73702228e-01
-1.78334638e-01 6.57292455e-02 -5.66302165e-02 -7.36818731e-01
-3.01847041e-01 -1.30370080e+00 1.36666524e+00 4.92373973e-01
1.89953297e-01 -4.55803424e-01 -6.31513655e-01 2.32436106e-01
3.43644619e-01 2.57645428e-01 4.48985666e-01 -4.49533015e-01
-7.71735132e-01 -4.32897538e-01 -8.40686977e-01 5.37409484e-01
-1.21661667e-02 1.44610927e-01 -1.34731901e+00 -3.74688476e-01
-1.54312909e-01 -2.37503991e-01 1.07718301e+00 4.29327309e-01
1.30766582e+00 -1.01628073e-01 -4.80957419e-01 8.30843270e-01
1.27164865e+00 -9.28858586e-04 4.87479627e-01 1.01336844e-01
6.85559094e-01 7.58716285e-01 9.14573550e-01 4.21009541e-01
-8.24125260e-02 1.00592482e+00 7.51284063e-01 -3.94256294e-01
-1.61741450e-01 1.33882895e-01 5.15927017e-01 -6.04815520e-02
-5.19685507e-01 -3.15869212e-01 -7.26438284e-01 2.75096953e-01
-1.63371730e+00 -9.29815948e-01 -2.23302871e-01 2.03277969e+00
4.83397394e-01 4.34132256e-02 1.35236993e-01 -4.44713905e-02
7.86649048e-01 1.03139840e-01 -5.53348243e-01 2.86863625e-01
-1.63751706e-01 -1.81911781e-01 1.11233854e+00 -2.03105733e-01
-1.64792573e+00 1.04389691e+00 5.86363649e+00 7.71830797e-01
-1.22069490e+00 1.09141283e-01 4.17514741e-01 -1.96462795e-01
7.60055959e-01 -4.09933120e-01 -1.36806309e+00 5.06203473e-01
1.33597255e-01 2.27477968e-01 2.61329591e-01 1.06505847e+00
-1.28450796e-01 -4.74178977e-02 -1.04192209e+00 1.01936901e+00
2.37996176e-01 -9.63060141e-01 -6.14333004e-02 -1.47799281e-02
5.22264063e-01 3.70038301e-02 3.12129587e-01 2.10492015e-01
4.66337129e-02 -9.36119020e-01 1.04522800e+00 2.02699959e-01
9.53923225e-01 -4.27034676e-01 9.33663726e-01 2.28773832e-01
-1.43150413e+00 -5.96608579e-01 -9.30670798e-01 -6.16602553e-03
-1.49812505e-01 1.35547340e-01 -8.03654253e-01 2.21005291e-01
1.28122318e+00 6.85765326e-01 -1.09823275e+00 1.21866548e+00
-2.03151196e-01 2.17321888e-01 -3.30518961e-01 -4.49128076e-02
3.43575507e-01 -1.39335513e-01 5.60158670e-01 1.18490267e+00
3.43211651e-01 1.32351220e-01 1.45572394e-01 1.25638986e+00
-5.87834828e-02 -7.98373148e-02 -6.84875727e-01 2.39600971e-01
7.00796694e-02 1.70640874e+00 -1.02087736e+00 -7.33116036e-03
-3.17494899e-01 9.93393600e-01 1.97686762e-01 1.86059460e-01
-1.21318543e+00 -4.94052052e-01 4.14456427e-01 4.14166898e-01
7.85479844e-01 7.95915350e-02 1.69648796e-01 -1.00248766e+00
2.50665069e-01 -5.86192608e-01 2.29862213e-01 -7.40164638e-01
-1.18373287e+00 8.50108504e-01 2.32093230e-01 -1.48307276e+00
5.71225524e-01 -1.16556096e+00 -5.72105467e-01 4.85459268e-01
-1.54790819e+00 -1.44436395e+00 -9.18954015e-01 4.37795341e-01
7.50760078e-01 -1.49168387e-01 5.13320148e-01 3.19505066e-01
-8.69678974e-01 5.01412809e-01 -2.67420918e-01 2.85826266e-01
6.72016084e-01 -1.07548380e+00 3.36350143e-01 1.07953382e+00
1.18130662e-01 2.47874305e-01 5.12643397e-01 -2.07896963e-01
-1.16385281e+00 -1.61078429e+00 1.33556679e-01 -8.12576175e-01
4.32267904e-01 -6.02434695e-01 -7.74007738e-01 6.59371257e-01
-1.25181273e-01 7.71563113e-01 2.64635623e-01 -4.92381006e-01
-3.03805768e-01 -2.88300961e-01 -1.10331130e+00 5.17307162e-01
1.13205838e+00 -7.27154687e-02 -5.16951382e-01 5.21642983e-01
6.97099626e-01 -4.71706122e-01 -3.27036709e-01 6.76255941e-01
3.84264022e-01 -1.11909151e+00 1.27585232e+00 -4.04606611e-01
1.14722341e-01 -6.94060445e-01 -3.24258834e-01 -9.86880660e-01
-4.53157336e-01 -2.75208920e-01 2.29173601e-01 1.08342826e+00
2.62827188e-01 -5.06693900e-01 3.00016940e-01 4.23425525e-01
-3.30839418e-02 -6.57620490e-01 -8.33809137e-01 -8.36508155e-01
-3.83244008e-01 9.55664739e-03 5.07167935e-01 4.63033766e-01
-5.46146333e-01 1.22631639e-01 -2.35788733e-01 4.93812919e-01
5.50434351e-01 7.78339282e-02 6.91250384e-01 -1.25877380e+00
-1.38529763e-01 -2.91771382e-01 -6.75605953e-01 -1.03516912e+00
-2.23175034e-01 -7.54004776e-01 2.58664936e-01 -1.25313902e+00
1.60429955e-01 -4.00930732e-01 -2.08707362e-01 5.49576461e-01
-2.51925409e-01 7.98910022e-01 3.98906291e-01 6.27065077e-02
-8.28681648e-01 4.78802949e-01 1.12336528e+00 -2.51301169e-01
-1.58888891e-01 2.04522923e-01 -3.17223310e-01 1.15272832e+00
6.93262637e-01 -4.65980262e-01 -1.74140573e-01 -4.85526353e-01
-2.54001096e-02 -5.89567423e-01 9.15349007e-01 -1.31365347e+00
2.42647052e-01 1.58772327e-03 7.67960429e-01 -7.71129727e-01
4.06735092e-01 -1.09852421e+00 -4.36854422e-01 4.91397589e-01
-1.56948194e-01 6.14659488e-02 4.88090307e-01 6.81943476e-01
1.36275306e-01 -2.66167611e-01 1.14469194e+00 -1.72901124e-01
-7.94045210e-01 2.90530384e-01 -5.27754240e-02 5.97074404e-02
1.53599668e+00 -4.11416590e-01 -3.54087442e-01 4.25163954e-01
-3.20661068e-01 1.79333147e-02 4.14319456e-01 5.13154805e-01
7.84501076e-01 -9.60101306e-01 -6.23186588e-01 4.93770272e-01
2.83581823e-01 3.26410204e-01 1.80796161e-01 1.16402578e+00
-3.87690067e-01 5.35639763e-01 -3.58501822e-01 -7.47745574e-01
-1.52479923e+00 5.69451034e-01 6.08903229e-01 1.36752471e-01
-8.07985365e-01 1.16372991e+00 8.33799720e-01 -3.35477024e-01
3.28045338e-01 -9.68976438e-01 -2.46787682e-01 2.02722237e-01
8.42215419e-01 4.43816006e-01 1.07634991e-01 -5.29930234e-01
-4.75526065e-01 8.22279811e-01 -1.91843525e-01 5.34838438e-01
1.17377675e+00 -5.70616983e-02 2.75327638e-02 1.83532640e-01
6.17582262e-01 -2.28277087e-01 -1.51084948e+00 -1.73462659e-01
-3.15478683e-01 -7.61244059e-01 2.32443631e-01 -7.21528411e-01
-1.17667210e+00 1.02753997e+00 9.02207673e-01 1.03524916e-01
8.90462935e-01 8.28387067e-02 4.47008967e-01 5.35185456e-01
4.01742190e-01 -8.38818669e-01 1.17215797e-01 1.02132767e-01
1.17193270e+00 -1.49001682e+00 2.93132693e-01 -5.42100728e-01
-6.71306074e-01 1.08717346e+00 9.63327825e-01 -1.58918679e-01
3.37570846e-01 2.46084109e-01 -8.78630280e-02 -1.41048104e-01
-3.61582816e-01 -4.71046865e-01 7.26498604e-01 5.84346235e-01
1.07483111e-01 -6.90721199e-02 2.43550450e-01 6.16396785e-01
1.69010967e-01 -3.53964180e-01 3.79128158e-01 6.52036607e-01
-5.45719147e-01 -4.01323557e-01 -5.70977032e-01 3.51109833e-01
-4.39437360e-01 -5.88650815e-02 -4.82673615e-01 1.03596818e+00
4.94297802e-01 7.75591552e-01 1.14324100e-01 -1.83797345e-01
5.73522925e-01 -3.40068609e-01 5.71002483e-01 -6.78884864e-01
-5.83431482e-01 -2.00822830e-01 -5.57613850e-01 -6.93024099e-01
-2.94512510e-01 -5.15870392e-01 -9.42052543e-01 2.85860449e-01
-9.12426293e-01 -4.05212611e-01 5.23195982e-01 6.76565766e-01
1.66959003e-01 7.81926453e-01 3.66261214e-01 -1.10591114e+00
-8.64333570e-01 -9.00529206e-01 -5.44295251e-01 1.31084308e-01
2.68627197e-01 -9.26768661e-01 -3.82387251e-01 7.45850801e-02]
|
[8.705233573913574, -0.7603228688240051]
|
26bf56b3-f7b4-4578-9a25-34118f726d0e
|
fast-video-shot-transition-localization-with
|
1808.04234
| null |
http://arxiv.org/abs/1808.04234v1
|
http://arxiv.org/pdf/1808.04234v1.pdf
|
Fast Video Shot Transition Localization with Deep Structured Models
|
Detection of video shot transition is a crucial pre-processing step in video
analysis. Previous studies are restricted on detecting sudden content changes
between frames through similarity measurement and multi-scale operations are
widely utilized to deal with transitions of various lengths. However,
localization of gradual transitions are still under-explored due to the high
visual similarity between adjacent frames. Cut shot transitions are abrupt
semantic breaks while gradual shot transitions contain low-level
spatial-temporal patterns caused by video effects in addition to the gradual
semantic breaks, e.g. dissolve. In order to address the problem, we propose a
structured network which is able to detect these two shot transitions using
targeted models separately. Considering speed performance trade-offs, we design
a smart framework. With one TITAN GPU, the proposed method can achieve a
30\(\times\) real-time speed. Experiments on public TRECVID07 and RAI databases
show that our method outperforms the state-of-the-art methods. In order to
train a high-performance shot transition detector, we contribute a new database
ClipShots, which contains 128636 cut transitions and 38120 gradual transitions
from 4039 online videos. ClipShots intentionally collect short videos for more
hard cases caused by hand-held camera vibrations, large object motions, and
occlusion.
|
['Wei zhang', 'Zhangkui Kuang', 'Yimin Chen', 'Shitao Tang', 'Litong Feng']
|
2018-08-13
| null | null | null | null |
['camera-shot-boundary-detection']
|
['computer-vision']
|
[ 2.28313386e-01 -6.25425220e-01 -1.83495760e-01 -1.26019955e-01
-4.44636613e-01 -3.16200495e-01 2.80053467e-01 8.02067295e-02
-4.78095502e-01 2.61006176e-01 1.66860834e-01 8.21933448e-02
1.12038508e-01 -6.11332715e-01 -7.36807287e-01 -5.01912415e-01
-3.29447687e-01 -1.82238027e-01 1.01655269e+00 -5.01899123e-02
3.77595812e-01 1.79717883e-01 -1.76075780e+00 5.06584764e-01
5.17604828e-01 1.07876372e+00 2.38096401e-01 5.66951871e-01
-1.29106209e-01 7.63062596e-01 -3.75710338e-01 -1.02935225e-01
2.95832902e-01 -5.03451407e-01 -3.62315238e-01 2.90746540e-01
4.47680920e-01 -4.33155090e-01 -5.73499799e-01 1.31812596e+00
4.34366792e-01 3.82861823e-01 3.16648990e-01 -1.09230018e+00
-1.54555604e-01 3.88754278e-01 -9.79794443e-01 8.47775400e-01
7.61056423e-01 2.80688971e-01 6.73157096e-01 -1.04453933e+00
8.27765286e-01 1.25581574e+00 6.97646976e-01 2.67543107e-01
-6.15128994e-01 -7.18908668e-01 2.15178251e-01 8.65723252e-01
-1.40783298e+00 -5.38456082e-01 9.59003448e-01 -3.99389416e-01
9.37070429e-01 1.86401993e-01 6.37977660e-01 1.09130418e+00
3.69373590e-01 7.19702184e-01 5.26693344e-01 -2.33906209e-01
1.98065802e-01 -3.45416546e-01 1.79578498e-01 6.26399040e-01
-1.98778622e-02 -1.74425691e-01 -5.70031703e-01 1.13009527e-01
6.54850960e-01 4.49736327e-01 -4.79769260e-01 -2.70634323e-01
-1.15473390e+00 5.13029754e-01 -1.26561131e-02 2.70388901e-01
-2.66610265e-01 -1.42688632e-01 8.18973124e-01 3.63831997e-01
2.29000464e-01 -3.06482077e-01 -1.46673217e-01 -5.46308100e-01
-1.02289772e+00 1.52806953e-01 4.73540336e-01 1.00668502e+00
6.04297757e-01 1.96462333e-01 -1.56680077e-01 6.64871573e-01
4.27033491e-02 1.99620783e-01 6.49148703e-01 -7.80425847e-01
6.84608996e-01 4.31977123e-01 -3.37447459e-03 -1.49461162e+00
-3.40782225e-01 1.75543483e-02 -9.10199761e-01 -5.90135492e-02
3.78243834e-01 6.24960251e-02 -7.70925701e-01 1.24219191e+00
4.56858933e-01 5.47694683e-01 -2.77449846e-01 1.00407588e+00
9.25871551e-01 9.11555052e-01 -2.11058155e-01 -7.05449820e-01
1.63848066e+00 -9.71221864e-01 -8.80301356e-01 -3.42627466e-02
4.61253107e-01 -8.50237429e-01 1.13912010e+00 4.23011899e-01
-1.11018896e+00 -8.84891450e-01 -1.16933870e+00 2.55120635e-01
-1.34999767e-01 -5.90684563e-02 1.04000919e-01 3.81951720e-01
-6.64919972e-01 6.45080090e-01 -8.81791890e-01 -4.71699476e-01
2.29239047e-01 -5.25176637e-02 -2.65569955e-01 -1.59840584e-01
-1.30842388e+00 4.93831843e-01 4.23044711e-01 2.16593459e-01
-7.87181616e-01 -4.24486756e-01 -9.61644053e-01 1.11142389e-01
1.00884867e+00 2.51669846e-02 9.02779222e-01 -9.92943347e-01
-1.49227440e+00 5.54130733e-01 -2.58360356e-01 -4.31682706e-01
6.50196016e-01 -2.30996862e-01 -7.38370717e-01 5.40969789e-01
-1.10348299e-01 1.69325560e-01 9.81854141e-01 -6.08299673e-01
-7.04047978e-01 -1.89078704e-01 -3.79516110e-02 3.07315856e-01
-4.54520553e-01 6.05783224e-01 -9.21994984e-01 -6.68336451e-01
5.34025431e-02 -8.69477570e-01 5.28469868e-02 1.33733274e-02
-2.31829152e-01 -1.15683578e-01 1.19790459e+00 -6.27107739e-01
1.44361186e+00 -2.33335257e+00 -1.88104749e-01 -6.81527257e-02
1.18645020e-01 5.00966787e-01 -1.08895727e-01 3.72521043e-01
-4.73001935e-02 -2.11925209e-01 9.20227170e-02 -1.39499962e-01
-3.54288369e-01 -3.47837619e-02 -2.11582705e-01 6.05603099e-01
-6.81036562e-02 4.08309877e-01 -7.13883042e-01 -6.48813426e-01
3.38371664e-01 2.75450319e-01 -4.91631299e-01 1.59947455e-01
1.31784931e-01 1.32274568e-01 -2.56958544e-01 7.25117981e-01
6.97113395e-01 1.43051773e-01 -1.28934577e-01 -3.58570158e-01
-3.39053869e-01 -4.44897823e-02 -1.59430313e+00 1.76551485e+00
5.87373562e-02 6.84297740e-01 -5.44284247e-02 -1.09076202e+00
7.71240950e-01 2.78771669e-01 3.47302437e-01 -6.27175272e-01
2.62137741e-01 -7.24932253e-02 -6.69487715e-02 -1.00297570e+00
6.08418226e-01 1.46407589e-01 1.08301118e-01 8.45514759e-02
-1.51010528e-01 3.75112504e-01 5.06694853e-01 5.67249842e-02
1.26870847e+00 2.27047831e-01 3.58854026e-01 -8.09909031e-02
5.00462651e-01 -3.23662907e-01 1.01791084e+00 6.66431010e-01
-5.54788172e-01 8.01243186e-01 5.23697615e-01 -6.79140985e-01
-8.31971943e-01 -8.79424453e-01 2.93528467e-01 1.05544090e+00
6.44196808e-01 -6.03392184e-01 -6.97717309e-01 -3.87314647e-01
-5.71330070e-01 1.97513670e-01 -3.99213105e-01 -3.09852600e-01
-8.42262924e-01 -5.47997832e-01 3.83762330e-01 5.03733039e-01
7.15379655e-01 -1.17187941e+00 -1.00335610e+00 4.04578149e-01
-3.47118437e-01 -1.44584405e+00 -7.97761798e-01 -2.82645464e-01
-7.58624077e-01 -1.23428071e+00 -6.32465065e-01 -8.83600116e-01
1.96548209e-01 7.77812839e-01 6.72621608e-01 -1.18578114e-01
-6.35302722e-01 -4.11704406e-02 -5.63426793e-01 1.05157547e-01
-2.07621291e-01 -3.15449566e-01 2.98790395e-01 6.66189864e-02
4.95615363e-01 -5.06752193e-01 -6.55255318e-01 5.62032282e-01
-1.02172983e+00 8.26547518e-02 2.04145044e-01 4.85029548e-01
4.87219661e-01 3.09692889e-01 1.34256616e-01 -2.62725204e-01
4.29578751e-01 -3.09389502e-01 -4.62624133e-01 2.45717227e-01
4.91432399e-02 -4.35522288e-01 8.62392485e-01 -8.05226266e-01
-1.15161562e+00 -2.79703122e-02 3.60206142e-02 -8.53967369e-01
-3.88314515e-01 4.49039876e-01 -1.16808623e-01 1.30452141e-01
4.95460451e-01 4.06561941e-01 -1.62481532e-01 -3.69131953e-01
-1.53536759e-02 5.79879224e-01 7.16903806e-01 -3.44731987e-01
6.07508063e-01 4.48917300e-01 -2.17553318e-01 -1.13252723e+00
-5.56435704e-01 -5.82860827e-01 -3.80674064e-01 -5.38407922e-01
9.74413931e-01 -1.06050634e+00 -6.27697468e-01 6.24566674e-01
-1.13893962e+00 -8.45651329e-03 2.35987511e-02 6.40894830e-01
-3.60892385e-01 9.65398550e-01 -9.39985812e-01 -5.92673957e-01
-2.88755864e-01 -1.21062160e+00 7.07215369e-01 4.41993654e-01
1.69825442e-02 -5.14022827e-01 8.63309503e-02 9.76271480e-02
1.37924746e-01 3.17749113e-01 3.54817748e-01 -3.59838843e-01
-4.58766222e-01 -2.49207154e-01 -8.45608935e-02 1.30837590e-01
1.50730655e-01 5.40060401e-01 -5.25590479e-01 -3.64860773e-01
2.16165781e-01 -9.69776884e-02 7.90364861e-01 4.14255917e-01
1.27785921e+00 -1.36232704e-01 -2.29206115e-01 6.32150769e-01
1.23257232e+00 6.44694567e-01 8.92590225e-01 3.12482923e-01
7.10171223e-01 4.03744310e-01 9.39105570e-01 7.42557883e-01
1.51781812e-01 8.16603184e-01 3.36777329e-01 1.27873659e-01
2.58369334e-02 4.73464513e-03 6.52184248e-01 8.89349878e-01
-4.30322230e-01 -3.25520515e-01 -8.42830658e-01 3.31962943e-01
-1.92956591e+00 -1.38109362e+00 -3.03412616e-01 2.28505754e+00
5.97883046e-01 6.28492057e-01 1.62958398e-01 1.72334254e-01
1.16402781e+00 4.64046657e-01 -4.40265089e-01 -9.73905995e-02
-6.48775250e-02 -1.77108243e-01 1.62939891e-01 5.47299497e-02
-1.27573264e+00 8.46367896e-01 5.46385002e+00 1.23377800e+00
-1.09274542e+00 7.04548834e-03 3.75978410e-01 -3.81843209e-01
3.46357197e-01 -1.15618765e-01 -7.50119150e-01 7.97242045e-01
7.20340014e-01 1.07431181e-01 1.91992298e-01 7.30338216e-01
5.73316276e-01 -3.18247378e-01 -8.89222980e-01 1.40074706e+00
3.43447030e-01 -1.32175076e+00 -1.17830023e-01 -2.73179442e-01
5.43439448e-01 -8.50458369e-02 -2.88639039e-01 3.17094386e-01
-3.30386221e-01 -4.20718193e-01 6.81695342e-01 4.39415723e-01
7.02107191e-01 -7.73222327e-01 6.93135440e-01 2.38589138e-01
-1.75816989e+00 -6.75367482e-04 -5.77646375e-01 -2.90983111e-01
3.42178553e-01 5.97945929e-01 -1.30984873e-01 4.60703731e-01
9.69394922e-01 9.88703132e-01 -4.06967580e-01 1.15985250e+00
1.05612889e-01 5.71957529e-01 -3.90964240e-01 1.92816421e-01
2.01671034e-01 -2.80778885e-01 6.89651370e-01 1.32961452e+00
5.39761484e-01 2.25484341e-01 3.27581167e-01 3.39238435e-01
1.01617523e-01 9.56753567e-02 -5.61936259e-01 1.54409662e-01
3.70672852e-01 1.20974827e+00 -1.17170203e+00 -5.50595582e-01
-6.48615122e-01 1.14196301e+00 -7.61904940e-02 3.34840119e-01
-1.28160822e+00 -6.29463255e-01 6.46326184e-01 2.09855270e-02
4.92033571e-01 -2.72939086e-01 2.27730811e-01 -1.56481194e+00
3.29482853e-01 -9.28869486e-01 3.60976309e-01 -6.21465564e-01
-7.42523193e-01 4.78881299e-01 -1.14721144e-02 -1.76219916e+00
-1.58597127e-01 -1.41473442e-01 -9.09771860e-01 2.15233520e-01
-1.00695217e+00 -5.82318723e-01 -6.35863125e-01 7.38058329e-01
1.16262722e+00 -2.11229846e-02 3.09797734e-01 5.04275262e-01
-8.85732293e-01 4.28073168e-01 1.18532628e-01 1.33237019e-01
7.96886623e-01 -4.92119133e-01 4.63583916e-01 1.25922239e+00
1.42041564e-01 3.82693231e-01 9.68441844e-01 -7.07069814e-01
-1.40484846e+00 -8.66945207e-01 4.83066171e-01 2.19972283e-01
7.22895801e-01 -3.79805773e-01 -1.16598320e+00 3.22114497e-01
1.70115024e-01 2.70205945e-01 4.13001537e-01 -4.28020149e-01
-1.70106098e-01 -1.11369200e-01 -7.81828463e-01 6.69689476e-01
1.31900394e+00 -4.36068535e-01 -6.71441376e-01 3.91244031e-02
5.46370089e-01 -5.72810829e-01 -4.59585667e-01 4.24973816e-01
4.78136241e-01 -1.34848070e+00 9.42491829e-01 -2.25532517e-01
5.66847146e-01 -4.07545298e-01 1.44924060e-01 -8.85453463e-01
-1.53979778e-01 -6.76228762e-01 -3.41998428e-01 1.22772086e+00
-4.17306513e-01 -2.34295905e-01 6.45486236e-01 3.25497180e-01
-1.27229422e-01 -6.33486032e-01 -1.03355289e+00 -9.59354401e-01
-6.89531386e-01 -5.13438284e-01 2.01552674e-01 9.13146257e-01
2.63946414e-01 1.45621151e-01 -7.56377220e-01 -2.22033467e-02
5.11388183e-01 6.87948167e-02 7.06656694e-01 -9.76970553e-01
-2.19896212e-01 -2.90224791e-01 -7.29780376e-01 -1.25478721e+00
-1.76965475e-01 -1.11763909e-01 2.62053758e-01 -1.01258981e+00
3.78745556e-01 1.93232656e-01 -1.70574456e-01 1.86254948e-01
-1.49317145e-01 3.73582006e-01 3.12885076e-01 2.77680874e-01
-9.25790608e-01 5.92022896e-01 9.68544781e-01 1.26172267e-02
-3.46461594e-01 -1.44142523e-01 5.77173494e-02 1.12025487e+00
6.23109639e-01 -3.80688161e-01 -4.17647690e-01 -1.73439458e-01
7.17628300e-02 4.23350632e-01 2.17440248e-01 -1.52975881e+00
3.68220776e-01 -1.97780639e-01 2.11418733e-01 -8.72506440e-01
3.46355647e-01 -6.96216822e-01 2.38851786e-01 5.01718283e-01
-1.68042496e-01 1.74444079e-01 1.13292471e-01 7.65885592e-01
-4.51074868e-01 -2.14781970e-01 7.41916358e-01 -6.92255571e-02
-1.09902835e+00 3.62858713e-01 -8.29162538e-01 1.06399842e-01
1.30735910e+00 -6.49715841e-01 -1.56296089e-01 -2.90306300e-01
-6.33370996e-01 5.66837983e-03 3.85585159e-01 7.65375853e-01
8.58252168e-01 -1.18515599e+00 -3.62184823e-01 1.22325592e-01
3.06831505e-02 -2.45257616e-02 7.40716159e-01 9.17016983e-01
-6.58369720e-01 3.77158038e-02 -4.29614365e-01 -5.97258806e-01
-1.81984007e+00 7.54019558e-01 1.17324702e-02 4.93834838e-02
-1.05058503e+00 6.57186151e-01 1.83775619e-01 4.81580198e-01
3.38776916e-01 -5.28742254e-01 -3.92690986e-01 3.54144514e-01
8.74296069e-01 5.39346933e-01 -1.03466205e-01 -7.01242089e-01
-4.20550406e-01 7.72846580e-01 -2.44992182e-01 2.66133338e-01
1.01980317e+00 -3.43932390e-01 3.02580029e-01 5.17460287e-01
1.27473664e+00 -2.66128063e-01 -1.48892069e+00 -2.83617258e-01
-2.31951371e-01 -6.37448490e-01 -2.12916538e-01 3.31478491e-02
-1.08348954e+00 8.59715283e-01 7.76198149e-01 1.24855667e-01
1.19672978e+00 -2.96910286e-01 1.14761424e+00 3.59364182e-01
4.38098133e-01 -1.44688988e+00 2.70597398e-01 4.66530710e-01
5.11716545e-01 -1.19144952e+00 -4.57783416e-02 -5.76403737e-01
-5.75247049e-01 1.28751564e+00 7.13669896e-01 -1.81139126e-01
3.78513962e-01 1.56479582e-01 -8.95590037e-02 -6.60710782e-02
-5.67587256e-01 -4.38131653e-02 1.17419727e-01 2.89495885e-01
8.33864883e-02 -3.66966367e-01 -4.23041254e-01 2.72645205e-01
2.73041636e-01 1.07269302e-01 7.89313853e-01 9.83100593e-01
-6.10573351e-01 -4.93369073e-01 -5.77644825e-01 2.96889931e-01
-5.56183100e-01 4.88399304e-02 2.17833921e-01 6.26145244e-01
1.85075566e-01 9.39169466e-01 3.18949163e-01 -5.49297333e-01
2.63479203e-01 -1.28609657e-01 1.69548705e-01 -1.79037675e-01
-4.24311578e-01 3.70221257e-01 -9.73606259e-02 -9.94188249e-01
-6.34028256e-01 -7.30009615e-01 -1.18746781e+00 -1.87122837e-01
-2.24500969e-01 -1.14559360e-01 2.40820795e-01 9.35057878e-01
2.06037134e-01 5.78222871e-01 4.86494720e-01 -1.01719844e+00
-3.24863911e-01 -9.56450582e-01 -5.62024653e-01 6.85951591e-01
1.88505650e-01 -7.17063248e-01 -4.63173240e-01 2.86927909e-01]
|
[8.695755004882812, 0.24842388927936554]
|
755d3be0-c4ec-4834-b615-ad592707db21
|
smoothing-matters-momentum-transformer-for
|
2203.07988
| null |
https://arxiv.org/abs/2203.07988v1
|
https://arxiv.org/pdf/2203.07988v1.pdf
|
Smoothing Matters: Momentum Transformer for Domain Adaptive Semantic Segmentation
|
After the great success of Vision Transformer variants (ViTs) in computer vision, it has also demonstrated great potential in domain adaptive semantic segmentation. Unfortunately, straightforwardly applying local ViTs in domain adaptive semantic segmentation does not bring in expected improvement. We find that the pitfall of local ViTs is due to the severe high-frequency components generated during both the pseudo-label construction and features alignment for target domains. These high-frequency components make the training of local ViTs very unsmooth and hurt their transferability. In this paper, we introduce a low-pass filtering mechanism, momentum network, to smooth the learning dynamics of target domain features and pseudo labels. Furthermore, we propose a dynamic of discrepancy measurement to align the distributions in the source and target domains via dynamic weights to evaluate the importance of the samples. After tackling the above issues, extensive experiments on sim2real benchmarks show that the proposed method outperforms the state-of-the-art methods. Our codes are available at https://github.com/alpc91/TransDA
|
['Wenbing Huang', 'Tingyang Xu', 'Fuchun Sun', 'Jiaqi Han', 'Shangmin Guo', 'Yu Rong', 'Runfa Chen']
|
2022-03-15
| null | null | null | null |
['synthetic-to-real-translation']
|
['computer-vision']
|
[ 1.79025635e-01 -5.36124595e-02 -2.32118621e-01 -5.00118852e-01
-8.90151918e-01 -4.52359736e-01 6.54278576e-01 -1.87557787e-01
-3.99903297e-01 5.83074391e-01 -9.30031613e-02 -2.52725370e-02
2.78997906e-02 -4.87044424e-01 -6.12861991e-01 -8.22420597e-01
2.78858751e-01 6.28122330e-01 7.06579506e-01 -4.15758267e-02
3.02799523e-01 6.09004907e-02 -1.37856126e+00 7.97189251e-02
1.13355255e+00 1.01737022e+00 3.04515541e-01 2.32257336e-01
-3.01927060e-01 5.01487613e-01 -4.99635756e-01 -3.36775243e-01
4.17531848e-01 -4.90427047e-01 -8.49716902e-01 3.59182388e-01
4.65819091e-01 -9.89109948e-02 -1.33249402e-01 1.42706609e+00
4.48906094e-01 1.22488938e-01 6.45807624e-01 -1.38448131e+00
-4.75832552e-01 5.25847793e-01 -8.51235926e-01 1.50392950e-01
-1.00227758e-01 2.16217071e-01 1.07048619e+00 -9.76186275e-01
6.75210834e-01 1.24899220e+00 6.27021730e-01 5.41535258e-01
-1.01573277e+00 -5.65799713e-01 3.66979867e-01 3.94375384e-01
-1.11968064e+00 -4.12701160e-01 8.89978528e-01 -5.66153228e-01
4.87756789e-01 -1.32824138e-01 4.34780777e-01 1.08131850e+00
-1.09607011e-01 1.07197893e+00 1.18918884e+00 -3.69673312e-01
3.28004897e-01 2.08487555e-01 3.37086588e-01 6.67320907e-01
-4.49136123e-02 -1.56913083e-02 -4.50762570e-01 -1.60764426e-01
7.39743710e-01 -2.57565886e-01 -1.10229291e-01 -5.57269216e-01
-9.83500302e-01 9.03283417e-01 3.55405122e-01 1.87052533e-01
-2.39932820e-01 1.31803587e-01 5.41115165e-01 1.87052190e-01
5.15434325e-01 1.65710732e-01 -6.02813601e-01 -1.45419851e-01
-7.75081396e-01 1.46915749e-01 3.05629194e-01 9.40508008e-01
6.33880973e-01 -3.66466083e-02 -1.91102237e-01 1.12962234e+00
2.31462300e-01 2.99413443e-01 5.56582451e-01 -1.06402993e+00
5.15534461e-01 5.16472995e-01 4.89705615e-02 -6.36538327e-01
-2.09681869e-01 -3.54551882e-01 -5.95230401e-01 4.23561148e-02
8.69533956e-01 -1.30148456e-01 -1.23693001e+00 1.66386509e+00
5.44123173e-01 4.85862672e-01 -9.95384157e-02 1.04579031e+00
4.97258663e-01 4.93983388e-01 1.84872851e-01 4.52649556e-02
1.15593803e+00 -1.27337956e+00 -5.47053456e-01 -4.41464007e-01
3.28803837e-01 -8.53017092e-01 1.32294428e+00 1.66455284e-01
-1.04428363e+00 -6.31316423e-01 -8.21435750e-01 -2.21535489e-02
-5.95476776e-02 1.43535003e-01 5.39593697e-01 3.36800039e-01
-6.28777444e-01 6.77960813e-01 -9.46560323e-01 -3.69511783e-01
5.92602730e-01 1.52950034e-01 6.13918751e-02 1.42476242e-02
-1.00425971e+00 6.26446009e-01 2.99529850e-01 -4.16909046e-02
-7.26049542e-01 -6.99599147e-01 -5.59131384e-01 -1.36118859e-01
6.18740797e-01 -5.86335361e-01 1.38433897e+00 -1.20169866e+00
-1.60258007e+00 8.86773765e-01 -9.86123681e-02 -5.18604875e-01
6.85389519e-01 -3.12959582e-01 -4.72516529e-02 1.00133359e-01
2.67999530e-01 8.53533089e-01 1.03534186e+00 -1.11130297e+00
-8.60838592e-01 -3.19697708e-01 -1.99348629e-01 3.69364291e-01
-2.67833352e-01 -1.10285975e-01 -8.92363369e-01 -6.98061109e-01
1.38085276e-01 -1.04666579e+00 -4.13210422e-01 -2.43171137e-02
-3.01452190e-01 -3.16343188e-01 5.62857985e-01 -4.31104958e-01
8.07826877e-01 -2.29546666e+00 7.91812614e-02 7.61872232e-02
3.41468677e-02 3.28108400e-01 -2.29032382e-01 -2.60566454e-02
2.34243259e-01 -1.95836395e-01 -4.96427149e-01 -2.44398177e-01
3.44508886e-02 1.81674540e-01 -2.80512124e-01 4.00241315e-01
2.44209155e-01 7.64373243e-01 -9.47053432e-01 -4.56749350e-01
2.78795451e-01 2.95896828e-01 -4.52785939e-01 -2.97229383e-02
-3.48298818e-01 5.02144754e-01 -7.48420954e-01 5.90982676e-01
8.28134239e-01 -3.72770607e-01 1.50068328e-01 -8.20114017e-02
1.82089180e-01 4.32603270e-01 -1.05746377e+00 1.69763863e+00
-5.95632643e-02 3.88612032e-01 1.05140775e-01 -1.24817574e+00
8.15853655e-01 -6.40945602e-03 4.46521521e-01 -9.12485421e-01
1.88868403e-01 3.07227850e-01 -1.08526260e-01 -2.41078228e-01
3.47718626e-01 -1.34242058e-01 9.78420079e-02 -5.44939488e-02
1.05992720e-01 -2.18038291e-01 1.64543360e-01 6.13045394e-02
7.09437609e-01 3.07336479e-01 5.06128147e-02 -3.53372544e-01
5.92060804e-01 1.95275694e-01 8.97654712e-01 6.06498778e-01
-4.49799716e-01 7.88617373e-01 5.10502875e-01 -1.57355949e-01
-1.00234568e+00 -1.15604293e+00 -9.34859142e-02 1.04278493e+00
4.79447812e-01 -1.35353938e-01 -9.11564589e-01 -9.49606597e-01
1.24397073e-02 5.81328213e-01 -3.37349892e-01 -1.41799614e-01
-4.99531209e-01 -7.96849489e-01 3.16163450e-01 5.70239723e-01
5.87399840e-01 -9.83849049e-01 -3.91391337e-01 1.91181675e-01
-2.70405263e-01 -1.34848070e+00 -6.22228682e-01 2.07106590e-01
-1.00679922e+00 -9.97502565e-01 -9.89987254e-01 -9.39827621e-01
7.30530977e-01 3.19828689e-01 1.00311863e+00 -2.58367717e-01
-1.64941996e-01 1.47542253e-01 -3.73431027e-01 -2.89767802e-01
-2.50253201e-01 2.75609136e-01 -1.19918175e-01 -1.44461721e-01
4.42062318e-01 -4.85244215e-01 -6.47226512e-01 5.49795926e-01
-5.72010279e-01 6.66289777e-02 3.76816839e-01 1.02153945e+00
8.19937527e-01 -2.90230736e-02 5.31795681e-01 -1.05887377e+00
3.95474821e-01 -4.48795289e-01 -7.97148943e-01 1.42736420e-01
-6.36643529e-01 -1.41458381e-02 6.00888014e-01 -5.21898806e-01
-1.12129545e+00 1.04173109e-01 -1.46168515e-01 -4.44395870e-01
-1.22969754e-01 1.79348707e-01 8.06034952e-02 1.51901111e-01
5.85613430e-01 6.18685968e-02 -4.53096814e-02 -5.45934498e-01
3.05786282e-01 4.71032500e-01 4.54534143e-01 -6.88668668e-01
8.52013230e-01 5.08770287e-01 -3.14103931e-01 -7.27011979e-01
-1.07774055e+00 -6.70911312e-01 -4.01778966e-01 9.10068229e-02
8.01819146e-01 -1.00222647e+00 -1.65026933e-01 8.33179653e-01
-8.16508055e-01 -5.73060393e-01 -4.40631628e-01 2.56341994e-01
-6.44586384e-01 4.07795012e-01 -5.25067389e-01 -4.81653720e-01
-2.37324342e-01 -1.30730188e+00 9.58785951e-01 5.25432050e-01
1.00665605e-02 -9.08953011e-01 -1.70222521e-01 5.69541991e-01
2.57416248e-01 -1.32130701e-02 6.58317208e-01 -5.10032713e-01
-5.13552606e-01 2.49116331e-01 -4.40469265e-01 5.34341633e-01
1.45729661e-01 -7.02027604e-02 -9.99917030e-01 -2.30318934e-01
7.22770169e-02 -3.46641719e-01 9.68127847e-01 6.90204978e-01
1.09883404e+00 1.38936639e-01 -2.69091815e-01 7.08550930e-01
1.39180183e+00 2.61061311e-01 4.56095546e-01 4.25838709e-01
8.21686089e-01 5.45922697e-01 1.07154632e+00 3.25676918e-01
4.96056378e-01 7.87577212e-01 2.27784440e-01 -2.67064780e-01
-2.36271009e-01 -1.44598588e-01 3.99022371e-01 6.59551919e-01
1.53999999e-01 -3.57304811e-02 -9.11881983e-01 7.96584308e-01
-2.02391338e+00 -6.04011893e-01 -1.67130977e-01 2.03640246e+00
9.00377452e-01 4.70292687e-01 2.34275550e-01 -2.66676426e-01
7.44686007e-01 2.33208183e-02 -7.68211424e-01 -3.00756007e-01
-7.00988397e-02 1.02736160e-01 6.91882014e-01 5.24194539e-01
-1.20889974e+00 1.40714502e+00 5.48933554e+00 1.16441631e+00
-1.09656489e+00 2.00415432e-01 7.46824205e-01 6.04810156e-02
-1.37962148e-01 8.34158733e-02 -8.88953805e-01 6.13491893e-01
6.62724674e-01 -1.10354684e-02 3.27478856e-01 9.86550450e-01
2.35017389e-02 -2.13629499e-01 -8.13643098e-01 7.74725497e-01
-2.60763675e-01 -1.09245884e+00 -1.54460296e-01 -1.75120711e-01
8.91144633e-01 4.55964684e-01 2.06147745e-01 2.17436433e-01
3.35621059e-01 -5.62356174e-01 8.91668379e-01 1.04759842e-01
5.60447097e-01 -6.12756073e-01 6.62029326e-01 1.24635808e-01
-8.89755368e-01 -9.06561241e-02 -6.15027249e-01 2.30890974e-01
1.09150827e-01 7.56715357e-01 -8.56175482e-01 4.18409616e-01
6.98952377e-01 7.21305490e-01 -4.66207117e-01 8.82286668e-01
-3.20817798e-01 7.76535690e-01 -3.82455349e-01 3.18327457e-01
4.01955187e-01 -3.61745358e-01 6.33533657e-01 1.01451325e+00
1.51690334e-01 -1.25672147e-01 2.99253404e-01 8.07236612e-01
3.34102288e-02 -2.83404719e-02 -2.45465636e-01 -2.56329719e-02
4.31279480e-01 1.10640121e+00 -9.67736363e-01 -3.25486213e-01
-4.59248841e-01 1.05105245e+00 1.88757315e-01 4.47346747e-01
-9.28014398e-01 -8.27852786e-02 8.24537516e-01 8.86851400e-02
5.02173305e-01 -1.97271198e-01 -5.20042181e-01 -1.18134797e+00
2.20426083e-01 -8.61028194e-01 3.66879076e-01 -3.61974299e-01
-1.43856502e+00 3.72920692e-01 -7.11545795e-02 -1.15931046e+00
-1.28920108e-01 -3.96560520e-01 -4.36522275e-01 6.93621814e-01
-1.83073652e+00 -1.03891623e+00 -1.55163005e-01 5.52774429e-01
8.98313999e-01 -1.02094702e-01 3.45751852e-01 3.85217756e-01
-5.25002658e-01 6.95549726e-01 2.84187436e-01 -3.28884050e-02
9.42628920e-01 -1.24572587e+00 7.65089214e-01 8.12274694e-01
9.38850790e-02 3.45018089e-01 8.09171736e-01 -4.72020417e-01
-9.17624116e-01 -1.02348363e+00 4.66913402e-01 -2.75367379e-01
7.00835466e-01 -3.31285089e-01 -9.64100599e-01 5.21428466e-01
-2.86784023e-02 -2.96996329e-02 1.96340680e-01 -1.66154858e-02
-2.60766685e-01 -1.00848876e-01 -1.14035249e+00 5.71098864e-01
1.02188468e+00 -1.87070221e-01 -4.82150763e-01 3.46725971e-01
6.85019791e-01 -5.02052009e-01 -5.23473918e-01 4.53849763e-01
2.97099799e-01 -1.00109673e+00 9.74723220e-01 -1.62451789e-01
4.15022105e-01 -2.79576451e-01 4.47129970e-03 -1.22191906e+00
-2.05301836e-01 -4.08772916e-01 2.60713488e-01 1.43315327e+00
4.22442675e-01 -7.57779896e-01 9.79780197e-01 2.91920036e-01
-1.21594340e-01 -6.92731559e-01 -9.37389314e-01 -8.43422055e-01
2.11765707e-01 -2.14753017e-01 2.78636009e-01 9.07814741e-01
-3.82768542e-01 3.49530160e-01 -1.99849367e-01 4.28072028e-02
7.59012163e-01 9.96694863e-02 5.71031153e-01 -1.12194800e+00
-3.64232451e-01 -4.83240426e-01 -1.63644046e-01 -1.18222845e+00
1.99671779e-02 -7.65261054e-01 1.44265085e-01 -1.37334609e+00
2.23511234e-01 -7.59736001e-01 -4.23738956e-01 5.05255580e-01
-3.76820952e-01 2.32495949e-01 2.77570575e-01 2.19800308e-01
-6.14609480e-01 7.27276504e-01 1.43774891e+00 -8.18428695e-02
-3.60570341e-01 6.36068508e-02 -6.27007842e-01 1.04227841e+00
9.89026189e-01 -6.05274022e-01 -6.11469805e-01 -5.22191525e-01
-9.63593572e-02 -2.31783777e-01 3.11053932e-01 -9.56813812e-01
8.69834125e-02 -3.31413329e-01 6.52620792e-02 -5.42372048e-01
3.16349059e-01 -6.53289437e-01 -2.73714960e-01 2.88789064e-01
-2.08492503e-01 -3.35796446e-01 1.65589288e-01 4.70784843e-01
-3.35048139e-01 -2.24289104e-01 1.19999969e+00 -3.22041810e-02
-9.11270916e-01 2.47360528e-01 -1.38342485e-01 3.97668868e-01
9.22580302e-01 -3.16154152e-01 -2.25618839e-01 -1.12187698e-01
-5.08901298e-01 5.27230561e-01 6.18769467e-01 5.23677826e-01
3.70683461e-01 -1.09941280e+00 -5.22249281e-01 9.90789235e-02
-1.96987875e-02 3.29467356e-01 3.08724254e-01 8.89853954e-01
-2.69683093e-01 8.90782624e-02 -1.98819771e-01 -7.84657001e-01
-1.21075940e+00 1.70799509e-01 2.40292177e-01 -2.13372648e-01
-7.68854380e-01 1.14673996e+00 5.35180211e-01 -5.35261273e-01
3.38888317e-01 -2.62611747e-01 -7.67230839e-02 1.28890246e-01
1.56184971e-01 2.63812780e-01 -1.37735501e-01 -3.39131355e-01
-5.20825386e-01 7.86592007e-01 -3.67341101e-01 -1.47802323e-01
1.32464981e+00 -3.02124679e-01 2.00861335e-01 3.71267915e-01
8.97588491e-01 -1.92982316e-01 -1.66813362e+00 -3.32129389e-01
2.04819888e-01 -3.53732049e-01 -6.05385527e-02 -7.54159093e-01
-1.13411736e+00 8.90312731e-01 6.71689630e-01 -3.94663634e-03
1.20689142e+00 6.89882711e-02 1.11675894e+00 -1.95320770e-02
2.94145495e-01 -1.49637258e+00 1.17038764e-01 7.23830700e-01
4.58660096e-01 -1.35242295e+00 -1.40528291e-01 -6.43459141e-01
-1.08174479e+00 7.41669536e-01 6.68108702e-01 -3.24067503e-01
4.68255728e-01 1.33390993e-01 3.73791873e-01 -8.13696757e-02
-6.03128910e-01 -2.94367701e-01 1.50377512e-01 6.27923191e-01
1.23141184e-01 -4.12286120e-03 -4.14639086e-01 3.79325062e-01
1.09580858e-03 -3.08926310e-03 3.87871891e-01 6.41808569e-01
-4.15021837e-01 -1.28550661e+00 -2.83193260e-01 2.03865826e-01
-5.65458477e-01 2.57575209e-03 -2.94756740e-01 7.15141177e-01
1.99114323e-01 6.81408167e-01 -1.76745262e-02 -1.08739018e-01
2.51268119e-01 6.98992312e-02 5.02019405e-01 -4.99980658e-01
-3.76529425e-01 3.91075253e-01 -1.91931461e-03 -5.78273952e-01
-3.90895337e-01 -7.88278282e-01 -1.58282578e+00 2.11770274e-02
-3.39830607e-01 3.38291824e-02 6.96831882e-01 8.98533821e-01
3.12450588e-01 5.33685088e-01 5.57205737e-01 -6.43786311e-01
-8.35799277e-01 -8.09711158e-01 -5.02225935e-01 5.39030969e-01
1.66850924e-01 -7.59948611e-01 -3.26470792e-01 4.15357463e-02]
|
[9.59254264831543, 1.3822647333145142]
|
70e3a4a5-cbae-4212-a9ba-0d988c611c30
|
on-sequential-bayesian-inference-for
|
2301.01828
| null |
https://arxiv.org/abs/2301.01828v2
|
https://arxiv.org/pdf/2301.01828v2.pdf
|
On Sequential Bayesian Inference for Continual Learning
|
Sequential Bayesian inference can be used for continual learning to prevent catastrophic forgetting of past tasks and provide an informative prior when learning new tasks. We revisit sequential Bayesian inference and test whether having access to the true posterior is guaranteed to prevent catastrophic forgetting in Bayesian neural networks. To do this we perform sequential Bayesian inference using Hamiltonian Monte Carlo. We propagate the posterior as a prior for new tasks by fitting a density estimator on Hamiltonian Monte Carlo samples. We find that this approach fails to prevent catastrophic forgetting demonstrating the difficulty in performing sequential Bayesian inference in neural networks. From there we study simple analytical examples of sequential Bayesian inference and CL and highlight the issue of model misspecification which can lead to sub-optimal continual learning performance despite exact inference. Furthermore, we discuss how task data imbalances can cause forgetting. From these limitations, we argue that we need probabilistic models of the continual learning generative process rather than relying on sequential Bayesian inference over Bayesian neural network weights. In this vein, we also propose a simple baseline called Prototypical Bayesian Continual Learning, which is competitive with state-of-the-art Bayesian continual learning methods on class incremental continual learning vision benchmarks.
|
['Stephen J. Roberts', 'Stefan Zohren', 'Tim G. J. Rudner', 'Adam Cobb', 'Samuel Kessler']
|
2023-01-04
| null | null | null | null |
['sequential-bayesian-inference']
|
['time-series']
|
[ 3.86672407e-01 9.41928253e-02 -3.79259139e-02 -4.73199457e-01
-4.46114093e-01 -1.93816409e-01 7.92473376e-01 3.48023325e-02
-8.92501295e-01 1.18258607e+00 -5.02852462e-02 -5.21944165e-01
-5.51649868e-01 -5.28871596e-01 -1.22027946e+00 -7.06394315e-01
8.31169784e-02 6.06389344e-01 5.29299796e-01 2.55951017e-01
3.10976028e-01 3.73599827e-01 -1.66476679e+00 1.51781395e-01
7.30868220e-01 3.92544806e-01 2.92052358e-01 7.65186131e-01
1.30500540e-01 7.72505999e-01 -6.36544168e-01 -4.75181133e-01
-6.95560053e-02 -2.40898892e-01 -4.23721403e-01 -2.66740531e-01
8.19493890e-01 -6.05635345e-01 -3.48303169e-01 1.06050086e+00
4.35387969e-01 4.59687620e-01 1.08189034e+00 -1.19672775e+00
-4.64016229e-01 9.88061786e-01 -7.02979863e-01 6.70141041e-01
-2.04283223e-01 3.30400199e-01 7.14505911e-01 -8.35468829e-01
3.87874156e-01 1.32752204e+00 1.07911456e+00 5.33580244e-01
-1.57169294e+00 -7.85033822e-01 5.90097368e-01 3.93853664e-01
-1.35551703e+00 -6.23452604e-01 6.23518527e-01 -4.05006111e-01
1.15301824e+00 -2.33942688e-01 8.23883891e-01 1.56500649e+00
9.90767777e-01 7.58888423e-01 1.27757490e+00 -5.46801865e-01
6.89768076e-01 -1.00564845e-01 8.31065357e-01 6.36807978e-01
7.02070236e-01 5.32682598e-01 -1.17256403e+00 -3.31619889e-01
5.50167978e-01 1.80718377e-01 -1.38523594e-01 -2.61913687e-01
-7.30336428e-01 6.71558738e-01 -5.47449999e-02 -1.37126431e-01
-2.96716541e-01 7.20710158e-01 3.29049885e-01 2.91682124e-01
3.12422276e-01 -1.91239230e-02 -5.12412369e-01 1.65658668e-02
-1.12451518e+00 5.67608774e-01 7.17504561e-01 5.67688167e-01
5.95238090e-01 1.18064933e-01 -5.61318159e-01 5.23295224e-01
3.01796973e-01 5.76384604e-01 1.17395163e-01 -1.14191759e+00
-1.25281006e-01 -4.62159604e-01 2.12048396e-01 -3.98388356e-01
-2.91028857e-01 -6.57675922e-01 -7.90849447e-01 3.44845146e-01
6.39255941e-01 -5.78424111e-02 -9.75544989e-01 1.98890150e+00
-1.09815724e-01 4.17756915e-01 -2.84791112e-01 1.81138009e-01
-1.31291434e-01 3.57277215e-01 3.06217670e-01 -7.00909078e-01
1.11724865e+00 -5.30234456e-01 -1.08845091e+00 -3.34415942e-01
2.56875549e-02 -4.83832300e-01 1.22980547e+00 9.97646570e-01
-9.49763119e-01 -4.67323512e-01 -1.23903060e+00 1.52986139e-01
-1.14520721e-01 -2.99363196e-01 8.63529086e-01 1.00401437e+00
-1.07752120e+00 8.84653926e-01 -1.42852890e+00 -6.06160238e-02
5.93696773e-01 1.92292690e-01 3.46228927e-01 -3.71868104e-01
-1.03139448e+00 1.00424230e+00 7.71190107e-01 -2.44046390e-01
-1.46881461e+00 -9.27638173e-01 -2.79528797e-01 -6.55087233e-02
5.97745776e-01 -1.08037412e+00 1.47970521e+00 -4.17719781e-01
-1.33968401e+00 1.87617809e-01 -3.91156971e-01 -1.05610776e+00
6.56921506e-01 -6.62268043e-01 2.44080015e-02 -2.48376235e-01
-3.92239004e-01 7.55945623e-01 1.40279114e+00 -1.02740610e+00
-4.51529950e-01 -3.03063661e-01 -1.28247708e-01 2.51740273e-02
-3.46978068e-01 -5.92557728e-01 -2.24125892e-01 -5.36086977e-01
-4.24693152e-02 -9.70460773e-01 9.95330960e-02 -1.13303818e-01
-2.45309457e-01 -3.99510086e-01 5.99153221e-01 -4.29906338e-01
1.04076803e+00 -1.97515213e+00 -1.30658850e-01 -1.10459644e-02
2.95126885e-01 -2.03605846e-01 6.38426468e-02 -2.85212249e-01
1.32845074e-01 -1.83519110e-01 -2.14922279e-01 -6.26393080e-01
1.41362567e-02 4.47296113e-01 -7.37684250e-01 5.39036334e-01
-8.99163038e-02 7.63336718e-01 -7.58471310e-01 -1.74551144e-01
1.25774033e-02 6.49946392e-01 -7.42632151e-01 -2.69184619e-01
-4.19405669e-01 -7.03018829e-02 2.12010011e-01 2.34919831e-01
7.41397202e-01 -1.90191552e-01 2.41914526e-01 -7.79636130e-02
2.13847458e-01 1.32597417e-01 -1.09239376e+00 1.61961758e+00
-1.93493500e-01 6.56273007e-01 -4.58981961e-01 -7.72965610e-01
4.30854291e-01 5.39322235e-02 -1.14112221e-01 -4.14454103e-01
-1.61561623e-01 8.28441989e-04 -2.35550161e-02 6.88048527e-02
4.40550745e-01 -4.08599555e-01 1.53580055e-01 7.92742193e-01
4.94580597e-01 -1.91117376e-01 5.94068281e-02 3.64262193e-01
1.14907467e+00 2.65235633e-01 1.27207831e-01 -4.57297593e-01
-2.79408723e-01 -3.58564764e-01 5.46559572e-01 1.89849842e+00
-1.44333497e-01 3.27846915e-01 4.26430434e-01 -5.40318549e-01
-1.06579638e+00 -1.65076602e+00 -3.22759360e-01 1.29310453e+00
-4.21115398e-01 -2.37000883e-01 -5.43974161e-01 -5.88349044e-01
7.80110359e-02 1.41567731e+00 -7.07251132e-01 -7.12194443e-01
-2.73649037e-01 -1.47882819e+00 3.62503797e-01 4.86406505e-01
3.52465063e-01 -6.65015042e-01 -8.69227588e-01 3.22670847e-01
7.85744190e-02 -5.75744033e-01 -2.51474679e-01 8.90892863e-01
-1.26079941e+00 -8.01445365e-01 -6.23914480e-01 1.51992515e-01
3.51788461e-01 2.26449594e-01 1.30362821e+00 -2.45479524e-01
-3.02646667e-01 7.29186654e-01 2.73091614e-01 -6.89469278e-01
-3.07660609e-01 2.66641498e-01 6.09034002e-01 -4.84946728e-01
4.24655795e-01 -7.95910120e-01 -4.07697737e-01 -4.11748476e-02
-8.87667477e-01 1.29587157e-02 6.29993439e-01 9.94378805e-01
6.16242707e-01 3.84535968e-01 6.38894498e-01 -1.01355708e+00
8.16640496e-01 -3.65967065e-01 -7.47992694e-01 4.80679065e-01
-1.09493995e+00 3.79043788e-01 1.26237705e-01 -9.10588980e-01
-1.56019068e+00 -1.70555755e-01 8.35812837e-02 -3.98861468e-01
-2.60749925e-02 4.49296355e-01 2.96045572e-01 2.65526086e-01
9.82273102e-01 1.91360608e-01 -9.52381268e-02 -4.22679007e-01
3.10671866e-01 -7.78988423e-03 5.66894293e-01 -9.11173522e-01
3.10087204e-01 5.28046787e-01 9.30061340e-02 -5.25122166e-01
-1.47053301e+00 1.85922623e-01 -5.53574920e-01 -3.62456501e-01
5.39129615e-01 -8.93531859e-01 -8.07538986e-01 7.62574732e-01
-1.29849899e+00 -7.02732146e-01 -5.23575783e-01 5.39397180e-01
-8.20203364e-01 1.57944590e-01 -7.35017657e-01 -1.12603164e+00
-2.73733959e-02 -8.26718986e-01 4.65156138e-01 1.55264437e-01
-2.23330960e-01 -1.08610499e+00 1.20799534e-01 2.41215397e-02
5.56010664e-01 -3.91341358e-01 1.04899573e+00 -4.20370102e-01
-6.08810186e-01 9.95840356e-02 2.31064484e-02 3.36627781e-01
-2.15319216e-01 -9.55428835e-03 -1.21080863e+00 -6.44336343e-01
2.33648136e-01 -4.68650639e-01 1.69249034e+00 7.95604825e-01
1.08481288e+00 -2.16649473e-01 -3.61931026e-01 3.12181145e-01
1.32661378e+00 -3.10317078e-03 6.16541088e-01 -5.96583895e-02
4.36692744e-01 2.64346629e-01 2.50401050e-01 5.49432755e-01
1.26932710e-01 2.02368930e-01 1.00521728e-01 7.27426887e-01
-3.67557228e-01 -3.98464650e-01 5.07028222e-01 5.61325431e-01
3.77553813e-02 -9.25317332e-02 -1.13429081e+00 3.59366506e-01
-1.81050313e+00 -1.00662804e+00 4.73110192e-02 2.42941737e+00
1.39231312e+00 8.65165532e-01 -1.99431390e-01 6.82705194e-02
5.42179227e-01 -1.29576698e-01 -9.29345787e-01 1.04968920e-01
1.59705896e-02 1.41877040e-01 6.21008933e-01 8.54983270e-01
-8.45318198e-01 8.56390655e-01 7.45482779e+00 1.29736757e+00
-5.68296432e-01 6.99911773e-01 8.59107733e-01 -5.36534488e-01
-3.32772881e-01 5.06572500e-02 -1.60596979e+00 3.13933313e-01
1.27630091e+00 -6.34718090e-02 6.03868186e-01 5.48928559e-01
-2.08599404e-01 -6.97867632e-01 -1.24306738e+00 8.24174464e-01
-2.27455534e-02 -1.26170671e+00 1.41132444e-01 -1.32945120e-01
9.95188594e-01 2.03885645e-01 3.20322812e-01 5.99005401e-01
1.00695550e+00 -9.40503716e-01 9.18977499e-01 1.24981737e+00
4.75477993e-01 -8.67759347e-01 4.71340865e-01 6.54229581e-01
-3.80457699e-01 -1.27905130e-01 -6.25757217e-01 -2.22995192e-01
1.09867766e-01 1.40902913e+00 -8.34495246e-01 -2.91509312e-02
7.81730711e-01 4.47069317e-01 -7.67768919e-01 1.12343776e+00
-3.38801801e-01 9.30574059e-01 -5.84477246e-01 5.48361950e-02
-3.01426113e-01 2.81717628e-01 5.35133243e-01 9.68594611e-01
1.88065365e-01 -4.31158751e-01 -1.49592265e-01 1.11799550e+00
1.50974587e-01 -6.57191217e-01 -5.23153901e-01 2.47065723e-01
7.01845407e-01 4.74719375e-01 -1.00812542e+00 -4.35933679e-01
1.70751847e-02 8.49564791e-01 5.86543441e-01 6.20167732e-01
-6.92375541e-01 -2.48635970e-02 7.36499801e-02 -4.52710949e-02
3.35360110e-01 -4.10119444e-01 -4.85188037e-01 -1.04632771e+00
-2.66751379e-01 -6.05403900e-01 3.40853125e-01 -7.15760529e-01
-1.47015989e+00 -3.04278061e-02 5.87162554e-01 -1.37502328e-01
-2.24876598e-01 -5.11413276e-01 -4.59526807e-01 6.01479828e-01
-1.55966794e+00 -5.29563665e-01 -2.20013596e-02 5.66135943e-01
5.66620827e-01 -9.04697180e-02 6.45006299e-01 -9.26426724e-02
-4.03537363e-01 6.06015146e-01 1.65618405e-01 -7.21799970e-01
9.93217230e-01 -1.16011250e+00 3.19560170e-01 7.57985950e-01
5.27905077e-02 9.63569283e-01 1.08485365e+00 -1.10735631e+00
-1.16653490e+00 -8.22204053e-01 6.07607007e-01 -7.01997459e-01
4.32112813e-01 -4.76163745e-01 -1.16406584e+00 9.08361375e-01
9.28886086e-02 -1.55981705e-01 4.13965166e-01 5.43206215e-01
-5.36692441e-01 -1.33636579e-01 -9.21260834e-01 5.93386590e-01
1.19036126e+00 -4.50158060e-01 -8.39317977e-01 3.82347196e-01
7.83910692e-01 6.42026484e-04 -3.41214627e-01 3.12007368e-01
6.81289494e-01 -1.11236370e+00 1.05533516e+00 -3.68743122e-01
6.91662356e-02 -1.03515290e-01 -1.05151027e-01 -1.19410670e+00
-2.53675014e-01 -6.71076179e-01 -6.82394981e-01 1.00596797e+00
1.32622868e-01 -6.28210604e-01 6.27997577e-01 4.36188966e-01
4.37119417e-02 -2.39858314e-01 -1.18250585e+00 -1.13432586e+00
3.58360112e-01 -9.97595608e-01 8.40504989e-02 5.78494251e-01
-6.17849886e-01 3.64958644e-01 -4.09825534e-01 -1.64410874e-01
1.44897306e+00 -6.60170615e-01 3.36776167e-01 -1.54609644e+00
-6.61691129e-01 -3.15521806e-01 3.14816743e-01 -1.04291940e+00
4.02454495e-01 -5.50618410e-01 1.77871063e-01 -1.06592989e+00
7.53568292e-01 -2.25457147e-01 -5.83368182e-01 6.11025572e-01
-2.91762501e-01 -3.19659077e-02 -8.38525072e-02 2.66737759e-01
-8.73748899e-01 7.40265489e-01 9.09397781e-01 -2.19984964e-01
-8.53732079e-02 -9.52455029e-02 -5.03170311e-01 8.57130289e-01
6.28588259e-01 -9.80742276e-01 -7.04138815e-01 -2.61628538e-01
8.07940662e-01 -1.48944482e-01 7.18388200e-01 -1.14458573e+00
7.20603228e-01 -8.35879520e-02 5.43828368e-01 -8.89291227e-01
2.48829216e-01 -5.10838747e-01 8.37473720e-02 7.23105192e-01
-4.36480522e-01 -2.89444417e-01 4.36117142e-01 1.21212566e+00
5.93895853e-01 -5.19655645e-01 8.19646180e-01 -2.25326881e-01
-3.21386009e-01 -2.15781704e-02 -8.48732293e-01 7.91476220e-02
5.85223913e-01 9.42167416e-02 -3.89476657e-01 -1.06265441e-01
-1.11352158e+00 -2.70269699e-02 2.74188161e-01 7.35978335e-02
6.55908823e-01 -1.04480016e+00 -5.74466348e-01 -1.68645494e-02
-3.48484457e-01 -1.59209982e-01 4.89131421e-01 8.85913610e-01
1.20206632e-01 2.42081478e-01 -1.93201780e-01 -6.90720618e-01
-8.75538647e-01 4.37014073e-01 3.93626213e-01 -3.67423832e-01
-5.18480897e-01 1.00739241e+00 -1.04377478e-01 -8.28953609e-02
6.45369709e-01 -1.70532614e-01 2.39625856e-01 2.74462163e-01
6.92228615e-01 4.88357693e-01 1.30998924e-01 6.91442251e-01
-1.40126228e-01 -7.66445026e-02 -7.17030406e-01 -6.88992202e-01
1.25679886e+00 -2.80946463e-01 -3.53064090e-02 1.16638803e+00
4.50921357e-01 -4.06405538e-01 -2.10941052e+00 -4.46872622e-01
2.15272773e-02 -2.14570776e-01 4.93376315e-01 -9.60277438e-01
-6.22209549e-01 1.07905507e+00 8.59631717e-01 -2.49592498e-01
6.74931884e-01 -2.57387191e-01 2.59340346e-01 1.02614963e+00
6.39002442e-01 -1.36290956e+00 4.53809291e-01 9.23513949e-01
6.23653173e-01 -9.79788244e-01 4.57381457e-01 2.61929154e-01
-1.20580547e-01 1.02180636e+00 3.61675352e-01 -8.66969079e-02
1.07186389e+00 5.75348258e-01 -6.48254037e-01 -7.63670430e-02
-1.11510897e+00 3.10468286e-01 9.32682306e-02 6.24789059e-01
1.67981818e-01 -9.31198448e-02 1.32861082e-02 7.14406848e-01
-1.37628749e-01 4.99218971e-01 6.15907013e-01 1.04477620e+00
-6.01303756e-01 -6.16541922e-01 -3.58175576e-01 6.93298042e-01
-4.28390175e-01 -4.80444133e-01 1.68620586e-01 4.72901613e-01
-5.54421395e-02 5.86698174e-01 2.92166591e-01 -4.39885110e-02
-2.03525111e-01 7.26475537e-01 1.10080361e+00 -5.70385516e-01
4.69638221e-02 8.89165699e-02 -2.13964880e-01 -1.03004359e-01
-1.34478480e-01 -1.11620331e+00 -7.71147311e-01 -5.27590513e-01
-3.60495687e-01 -3.54873925e-01 5.75971186e-01 1.06748760e+00
1.66726366e-01 8.86860430e-01 -2.46037409e-01 -7.79649734e-01
-1.05600715e+00 -1.04269290e+00 -5.68538904e-01 -1.42890826e-01
3.22327107e-01 -9.88239527e-01 -6.77392423e-01 2.08417386e-01]
|
[7.187438488006592, 3.8544070720672607]
|
56cd9bae-eb62-497e-b2b7-9ab53856546f
|
improving-mitosis-detection-via-unet-based
|
2209.09193
| null |
https://arxiv.org/abs/2209.09193v1
|
https://arxiv.org/pdf/2209.09193v1.pdf
|
Improving Mitosis Detection Via UNet-based Adversarial Domain Homogenizer
|
The effective localization of mitosis is a critical precursory task for deciding tumor prognosis and grade. Automated mitosis detection through deep learning-oriented image analysis often fails on unseen patient data due to inherent domain biases. This paper proposes a domain homogenizer for mitosis detection that attempts to alleviate domain differences in histology images via adversarial reconstruction of input images. The proposed homogenizer is based on a U-Net architecture and can effectively reduce domain differences commonly seen with histology imaging data. We demonstrate our domain homogenizer's effectiveness by observing the reduction in domain differences between the preprocessed images. Using this homogenizer, along with a subsequent retina-net object detector, we were able to outperform the baselines of the 2021 MIDOG challenge in terms of average precision of the detected mitotic figures.
|
['Amit Sethi', 'Nikhil Cherian Kurian', 'Sahar Almahfouz Nasser', 'Tirupati Saketh Chandr']
|
2022-09-15
| null | null | null | null |
['mitosis-detection']
|
['medical']
|
[ 4.48816210e-01 3.05932432e-01 -1.94050491e-01 -3.99737433e-02
-1.27731729e+00 -7.02769876e-01 4.62527126e-01 4.15538400e-01
-8.31659019e-01 1.10406506e+00 1.54474691e-01 -3.31848681e-01
1.75730988e-01 -5.96562445e-01 -7.01426387e-01 -1.10162103e+00
3.34543139e-01 6.37944996e-01 1.07208073e-01 1.38390586e-01
1.90655753e-01 4.02546406e-01 -7.88522840e-01 3.95893246e-01
6.38888955e-01 7.69930005e-01 -1.72778755e-01 1.09636700e+00
2.32815340e-01 8.04354191e-01 -6.22034073e-01 -1.78251103e-01
2.77747929e-01 -2.84700662e-01 -7.61055112e-01 6.58320785e-02
6.41337812e-01 -4.36744779e-01 -5.91561019e-01 1.23625207e+00
7.82501042e-01 -4.49388802e-01 1.06983733e+00 -1.12095237e+00
-2.79249191e-01 3.14568639e-01 -5.03619373e-01 7.89089143e-01
-2.29889944e-01 4.56591040e-01 6.09229088e-01 -3.59910429e-01
1.04422259e+00 6.93907559e-01 8.30101669e-01 7.42188096e-01
-1.62504601e+00 -5.33122778e-01 -6.52398467e-01 -8.48696455e-02
-1.37083697e+00 -3.40577692e-01 5.09786189e-01 -5.66348732e-01
8.61418724e-01 5.95763922e-02 4.63591397e-01 1.17357755e+00
7.99863815e-01 6.41006708e-01 1.16444159e+00 -2.57438034e-01
3.74860883e-01 5.71498275e-02 2.94073761e-01 6.60873771e-01
3.84365886e-01 6.46904856e-02 -2.88229108e-01 -7.45491236e-02
1.04622364e+00 -2.10809276e-01 -4.42282736e-01 -1.31530419e-01
-1.37543261e+00 5.94084740e-01 6.60113037e-01 9.70172808e-02
1.24173306e-01 2.76982009e-01 6.47356987e-01 1.73815846e-01
4.79299307e-01 6.15801334e-01 -6.98951930e-02 3.17361891e-01
-1.16738319e+00 1.40779108e-01 3.93816531e-01 7.67927170e-01
1.46283597e-01 -3.19446206e-01 -5.45748353e-01 2.65625179e-01
-2.10402422e-02 2.92326629e-01 7.66981423e-01 -8.48819554e-01
-7.98090026e-02 8.75989616e-01 -1.30651504e-01 -5.92715025e-01
-6.90989316e-01 -6.45450115e-01 -1.11583567e+00 3.38355631e-01
9.91803169e-01 -1.19077727e-01 -1.47327065e+00 1.61385763e+00
2.10989878e-01 1.92128047e-01 2.04716966e-01 9.72506642e-01
9.17704225e-01 6.13438413e-02 2.90268153e-01 3.27618085e-02
1.50521779e+00 -5.19536138e-01 -6.17366433e-01 -2.04790786e-01
8.17905307e-01 -5.45181632e-01 8.22702706e-01 2.49241859e-01
-9.52072799e-01 2.22467966e-02 -1.29752672e+00 -4.25868124e-01
-4.32048261e-01 1.87627539e-01 3.56429994e-01 5.02749085e-01
-1.40183806e+00 4.28986371e-01 -9.32612598e-01 -5.98218918e-01
1.00231016e+00 6.97568893e-01 -4.97192472e-01 9.52607542e-02
-7.49005139e-01 8.09514403e-01 4.01886284e-01 -3.93255234e-01
-1.11471868e+00 -1.28042924e+00 -7.42737830e-01 -5.89423776e-02
-9.81441364e-02 -1.10727727e+00 1.34325624e+00 -7.39652574e-01
-8.96520317e-01 1.51142275e+00 -1.46491051e-01 -9.84710336e-01
7.95362473e-01 4.29034710e-01 2.89020012e-03 5.06504178e-01
5.75072765e-02 1.20299482e+00 5.68308532e-01 -7.53533483e-01
-6.99534178e-01 -3.63008082e-01 -6.15007043e-01 9.97501612e-02
-3.27576429e-01 -4.92951006e-01 -3.19632113e-01 -8.54461968e-01
-2.76484583e-02 -8.00803006e-01 -4.98126417e-01 3.64312023e-01
-4.07003105e-01 1.73004061e-01 7.29911923e-01 -7.32089520e-01
6.82324111e-01 -2.09615827e+00 -1.19654439e-01 1.95266634e-01
7.34797120e-01 2.67020576e-02 -3.33903208e-02 -4.87114340e-01
-6.84006810e-02 1.37568653e-01 -1.54719859e-01 -2.23978356e-01
-2.41590619e-01 -1.30132530e-02 2.89123598e-02 1.05257297e+00
3.64073455e-01 1.12867975e+00 -9.04364586e-01 -7.54265845e-01
-4.04199585e-02 4.56313431e-01 -6.22410059e-01 -8.86436366e-03
-2.97972977e-01 3.17237258e-01 2.13297203e-01 8.89017761e-01
6.10021591e-01 -4.53606039e-01 1.49763793e-01 -2.48730093e-01
3.83895606e-01 -2.97958218e-02 -3.69765759e-01 1.47813249e+00
1.40975222e-01 1.12704241e+00 1.91273257e-01 -6.27833366e-01
5.97014964e-01 3.59766304e-01 5.81173241e-01 -5.79126954e-01
4.68511641e-01 3.05557489e-01 2.03875512e-01 -7.12734908e-02
3.18107545e-01 -5.04403889e-01 -6.50492683e-02 2.26388887e-01
1.43059611e-01 -3.28066915e-01 8.33992511e-02 2.92737007e-01
1.62223184e+00 -4.91142660e-01 3.23303849e-01 -6.89690471e-01
3.89096469e-01 4.17786598e-01 7.02118039e-01 4.63585287e-01
-9.01497185e-01 8.28116059e-01 9.62359250e-01 -5.73164582e-01
-1.37453258e+00 -1.31945074e+00 -3.39031160e-01 3.53294134e-01
5.82309365e-02 2.19377562e-01 -7.69253492e-01 -9.43746269e-01
1.76717326e-01 1.82259709e-01 -1.06898046e+00 -3.42591226e-01
-2.83222854e-01 -1.01300263e+00 1.12002373e+00 6.83823466e-01
4.50759977e-01 -7.10864186e-01 -5.05527973e-01 7.23218024e-02
-3.03790681e-02 -1.05782008e+00 -6.89858496e-01 6.29829228e-01
-9.32779491e-01 -1.27535152e+00 -1.13643932e+00 -1.17615581e+00
1.22211015e+00 1.84861682e-02 1.23239613e+00 -5.52952066e-02
-7.35321045e-01 2.02518841e-03 1.67458862e-01 -5.84252000e-01
-7.92686462e-01 1.69774979e-01 -9.42633897e-02 -5.30694246e-01
5.76175392e-01 -2.44372219e-01 -1.00625777e+00 2.07796901e-01
-1.03400779e+00 -1.13766670e-01 8.08136463e-01 1.16789889e+00
1.03570676e+00 1.97429746e-01 3.88304383e-01 -1.01830411e+00
3.81408691e-01 -1.74702138e-01 -5.39860249e-01 -1.83732212e-02
-6.09727979e-01 2.69553848e-02 3.39456916e-01 -5.40012717e-01
-7.32687950e-01 2.21341655e-01 1.73564404e-01 -4.61249113e-01
-7.73098990e-02 1.71157029e-02 2.50798315e-01 -5.02096713e-01
1.06134951e+00 1.41600326e-01 4.19349521e-01 1.54015943e-01
-2.19407886e-01 3.46781343e-01 1.14777541e+00 -8.87568444e-02
4.83055502e-01 8.41513991e-01 2.84415007e-01 -6.60415530e-01
-5.30258954e-01 -5.51940084e-01 -1.79240853e-01 -5.95058277e-02
8.14805686e-01 -1.06969452e+00 -6.34051144e-01 6.53280735e-01
-1.01922321e+00 -5.97435474e-01 -3.10978711e-01 2.89836228e-01
-6.06113017e-01 1.71754792e-01 -1.11673415e+00 3.82210799e-02
-6.45895600e-01 -1.15762532e+00 1.02398801e+00 5.44655323e-01
-3.96662056e-01 -1.04953301e+00 2.28526279e-01 4.14221913e-01
1.13989137e-01 4.91864681e-01 1.20713186e+00 -6.08676434e-01
-3.59450251e-01 -1.66010886e-01 -4.73516792e-01 7.37854326e-03
6.32510986e-03 1.26878187e-01 -1.09278738e+00 -5.00009537e-01
-2.83660829e-01 -2.32292920e-01 1.06362283e+00 9.90790188e-01
8.92969131e-01 -1.96699977e-01 -7.23521233e-01 8.88880134e-01
1.53235412e+00 -5.01064174e-02 9.15117145e-01 4.80359465e-01
2.57270783e-01 3.70434135e-01 2.92054564e-01 2.23660027e-03
8.55800658e-02 1.09740399e-01 4.56531703e-01 -5.72634995e-01
-5.76583385e-01 -5.49618490e-02 -8.79819766e-02 -1.12700239e-01
4.66293275e-01 -3.46610546e-01 -1.27086449e+00 1.13294017e+00
-1.44398701e+00 -6.17761612e-01 4.05783579e-02 1.90733504e+00
9.72878277e-01 4.09998477e-01 1.32068858e-01 -2.74529364e-02
7.87032247e-01 -4.69264507e-01 -7.97572911e-01 -1.11708023e-01
-3.08458269e-01 1.08800113e-01 1.08236670e+00 2.97574222e-01
-1.27126336e+00 7.22085118e-01 7.41024876e+00 7.24104643e-01
-1.10192394e+00 -4.35317829e-02 1.26492143e+00 -2.61142939e-01
3.46617848e-02 -4.82807070e-01 -5.78814626e-01 3.56157809e-01
7.12588906e-01 -2.42563635e-01 -1.31087989e-01 5.70084572e-01
7.18007237e-02 -3.36555481e-01 -1.43397772e+00 8.56177509e-01
-1.32367685e-01 -1.85978222e+00 -2.14096382e-01 4.40978706e-01
8.80695343e-01 -1.06807135e-01 5.04274249e-01 1.41268790e-01
4.99953032e-01 -1.30756402e+00 7.94249773e-02 3.34403127e-01
1.01024592e+00 -7.91428745e-01 1.22758961e+00 7.72226825e-02
-4.27952468e-01 2.28375643e-01 -3.24145228e-01 2.26220220e-01
-4.96512324e-01 6.82788968e-01 -1.87448847e+00 -1.89147130e-01
4.03583825e-01 5.17803431e-01 -6.70999646e-01 1.28474998e+00
2.53403425e-01 5.20924687e-01 -3.13828647e-01 7.95195252e-02
5.43225780e-02 4.24999326e-01 6.05203331e-01 1.28407550e+00
1.53273016e-01 -1.74068630e-01 -3.18979472e-01 7.39796937e-01
-5.14093280e-01 -2.60748208e-01 -5.29309630e-01 -6.87199757e-02
2.51421213e-01 1.23739529e+00 -1.27429044e+00 -2.65469313e-01
5.96947297e-02 7.52609074e-01 1.01217762e-01 1.78720713e-01
-7.08581328e-01 -2.08498195e-01 7.69272029e-01 2.94655800e-01
9.43602175e-02 2.55400479e-01 -9.84736264e-01 -8.10832977e-01
-4.42537606e-01 -9.68474746e-01 6.58887744e-01 -6.01107478e-01
-1.26957250e+00 2.89664030e-01 -6.09427631e-01 -1.23049724e+00
1.13433100e-01 -7.33402073e-01 -6.25915647e-01 7.33943701e-01
-1.44534457e+00 -1.04821169e+00 -3.17842871e-01 5.05598843e-01
3.25767934e-01 -1.34985894e-01 5.89397609e-01 2.36976340e-01
-4.83411878e-01 1.09254062e+00 6.32888079e-02 4.12990957e-01
1.21638966e+00 -1.57807791e+00 1.02374718e-01 8.23898435e-01
-4.42843139e-01 3.75301510e-01 9.46768105e-01 -7.73085296e-01
-9.22425449e-01 -1.28637326e+00 5.86820066e-01 -5.46066046e-01
6.30148768e-01 -1.12139083e-01 -6.28084004e-01 6.51260376e-01
1.85222819e-01 3.14484537e-01 7.38749921e-01 -4.50944722e-01
-1.92573160e-01 -9.50683095e-03 -1.82086527e+00 7.10502326e-01
4.99544531e-01 -4.56316561e-01 -2.94738501e-01 3.69157553e-01
5.45119762e-01 -7.18325019e-01 -1.03294718e+00 4.29591984e-01
3.75343621e-01 -5.66596389e-01 8.96822155e-01 -6.02240205e-01
6.76261067e-01 -1.91495836e-01 1.40961260e-01 -1.18820965e+00
-3.56959343e-01 -3.60353529e-01 1.39962524e-01 1.01332259e+00
3.92395735e-01 -3.61096948e-01 1.60282838e+00 3.88096809e-01
-3.58308777e-02 -5.93570411e-01 -1.14306474e+00 -4.75596905e-01
5.66772640e-01 2.16849893e-01 9.28516835e-02 6.90720737e-01
1.82557389e-01 1.45748630e-02 2.60124445e-01 2.37238809e-01
8.33026111e-01 -5.37941158e-01 4.54507828e-01 -9.05546546e-01
-1.24953486e-01 -6.69067919e-01 -1.02167106e+00 -2.99954802e-01
4.21542302e-02 -9.02816117e-01 1.80285290e-01 -1.07800901e+00
4.81216133e-01 -9.66637507e-02 -4.07728434e-01 3.64545166e-01
-1.75064236e-01 7.79661477e-01 -2.06212237e-01 1.79935887e-01
-4.55711186e-01 -1.57426879e-01 1.37234306e+00 -6.82091355e-01
-9.17030722e-02 -3.42131019e-01 -8.14859867e-01 7.06040442e-01
8.18561196e-01 -6.03307247e-01 -1.76257879e-01 -1.31286904e-01
5.09333750e-03 -2.40315422e-02 6.55252457e-01 -1.31178570e+00
5.50074279e-01 1.75784066e-01 9.23276544e-01 -6.24593616e-01
-4.66991477e-02 -4.87879276e-01 -1.14509106e-01 8.73910248e-01
-5.07953465e-01 -3.02739710e-01 6.11683488e-01 6.06571078e-01
1.50554255e-03 2.21841767e-01 1.21343005e+00 -4.52730916e-02
-4.76775825e-01 1.72161311e-01 -6.59736216e-01 1.82663828e-01
1.33570564e+00 -3.52505863e-01 -9.17815983e-01 -3.00488137e-02
-7.10005939e-01 3.07799160e-01 9.53900516e-01 -3.56134027e-01
5.01495481e-01 -1.13446450e+00 -7.18346477e-01 7.92967677e-02
1.11682102e-01 2.85245836e-01 3.09150159e-01 9.82354999e-01
-1.10224700e+00 4.70027834e-01 -5.21218479e-01 -8.68957698e-01
-1.36777079e+00 4.77398485e-01 7.03574836e-01 -9.13964450e-01
-4.38834369e-01 1.21166551e+00 3.93123806e-01 -1.34435102e-01
3.49151552e-01 -5.96730828e-01 1.21191695e-01 -1.41991660e-01
4.96469975e-01 2.51614898e-01 3.62837791e-01 -7.27274790e-02
-5.50869107e-01 1.01775423e-01 -6.12970471e-01 -2.50122696e-02
9.92672741e-01 2.20740765e-01 1.54487699e-01 -2.27726936e-01
1.24179232e+00 -4.12860960e-01 -1.59199274e+00 -1.65854879e-02
-7.04387575e-02 -2.66959630e-02 4.67244446e-01 -1.02741110e+00
-9.86847758e-01 6.81356668e-01 9.70772564e-01 -2.36081123e-01
1.19881225e+00 -1.11124061e-01 8.64396274e-01 1.92732438e-01
3.48519757e-02 -9.98631477e-01 9.04608592e-02 9.64721292e-02
2.70334482e-01 -1.56085646e+00 4.01518308e-02 -3.31443787e-01
-2.92438030e-01 9.60098088e-01 6.79568112e-01 -1.96216136e-01
1.81539446e-01 9.68737960e-01 4.53467518e-01 -1.46617442e-01
-8.21595192e-01 1.08287215e-01 2.53625605e-02 1.03523374e+00
1.82364851e-01 -4.06371765e-02 -2.32541472e-01 5.91290295e-01
1.88857652e-02 3.08581978e-01 8.66789818e-01 7.88131356e-01
-2.46319130e-01 -5.76046467e-01 -6.15474522e-01 5.98711967e-01
-6.43613458e-01 -4.91580665e-02 -6.02316260e-01 8.65028203e-01
-2.50251200e-02 3.22253078e-01 3.84399593e-01 -1.54207081e-01
-8.13109130e-02 -1.22259855e-01 5.41578531e-01 -4.43678558e-01
-5.86818099e-01 1.49668619e-01 -2.78062522e-01 -2.45348483e-01
-4.50092629e-02 -5.80006957e-01 -1.35327876e+00 -4.86380249e-01
2.43413731e-01 -3.43625098e-01 1.38376594e-01 5.68153262e-01
2.17666477e-01 9.31866467e-01 1.46472365e-01 -4.67561483e-01
-3.74085993e-01 -6.30738795e-01 -8.01517308e-01 2.20160440e-01
7.27975249e-01 -4.25242335e-01 -4.56711203e-01 4.82159078e-01]
|
[15.10588264465332, -3.139479160308838]
|
2ad127c4-3cef-421f-9776-643d72231581
|
self-supervised-learning-for-time-series
|
2306.10125
| null |
https://arxiv.org/abs/2306.10125v1
|
https://arxiv.org/pdf/2306.10125v1.pdf
|
Self-Supervised Learning for Time Series Analysis: Taxonomy, Progress, and Prospects
|
Self-supervised learning (SSL) has recently achieved impressive performance on various time series tasks. The most prominent advantage of SSL is that it reduces the dependence on labeled data. Based on the pre-training and fine-tuning strategy, even a small amount of labeled data can achieve high performance. Compared with many published self-supervised surveys on computer vision and natural language processing, a comprehensive survey for time series SSL is still missing. To fill this gap, we review current state-of-the-art SSL methods for time series data in this article. To this end, we first comprehensively review existing surveys related to SSL and time series, and then provide a new taxonomy of existing time series SSL methods. We summarize these methods into three categories: generative-based, contrastive-based, and adversarial-based. All methods can be further divided into ten subcategories. To facilitate the experiments and validation of time series SSL methods, we also summarize datasets commonly used in time series forecasting, classification, anomaly detection, and clustering tasks. Finally, we present the future directions of SSL for time series analysis.
|
['Shirui Pan', 'Dongjin Song', 'Guansong Pang', 'Yuxuan Liang', 'James Zhang', 'Yong liu', 'Ming Jin', 'Rongyao Cai', 'Chaoli Zhang', 'Qingsong Wen', 'Kexin Zhang']
|
2023-06-16
| null | null | null | null |
['anomaly-detection', 'time-series']
|
['methodology', 'time-series']
|
[-2.90328432e-02 -5.59263706e-01 -1.80878505e-01 -5.06773949e-01
-4.12374020e-01 -8.04782212e-01 5.80089748e-01 3.68700773e-01
-9.03070346e-03 3.98643434e-01 -2.77455062e-01 -2.59778142e-01
-1.84847713e-01 -7.07874298e-01 -2.50804514e-01 -8.83601308e-01
-8.50515902e-01 2.18317926e-01 -1.43800676e-01 -2.58874953e-01
1.99039936e-01 6.35864913e-01 -1.53382540e+00 7.46887326e-02
9.10133123e-01 1.40607548e+00 -5.97355425e-01 5.13212025e-01
-3.33724439e-01 8.13365340e-01 -9.58898365e-01 1.81125462e-01
1.96716756e-01 -4.64911461e-01 -4.90856051e-01 7.31724501e-02
-7.17223659e-02 -1.61475632e-02 -4.50877160e-01 9.36493039e-01
3.22259337e-01 3.28402221e-01 8.75868797e-01 -2.03855896e+00
-6.18710697e-01 5.58996558e-01 -3.08793962e-01 6.71877980e-01
2.35624716e-01 -3.53522156e-03 7.05726504e-01 -6.52131855e-01
2.42729455e-01 8.80831659e-01 1.06935620e+00 3.55145842e-01
-1.13535559e+00 -7.15970397e-01 3.54886591e-01 5.11599302e-01
-1.10563219e+00 -1.40266389e-01 1.54979622e+00 -5.97587943e-01
1.06127608e+00 2.89781988e-01 8.50111246e-01 1.08677161e+00
3.31331164e-01 9.88500834e-01 1.14150620e+00 -5.20573318e-01
6.40306175e-01 -5.35803854e-01 2.63081342e-01 4.16002512e-01
-3.88424963e-01 1.27264395e-01 -1.86896697e-01 -5.08379102e-01
4.73158628e-01 4.49065566e-01 2.64863551e-01 -2.53014624e-01
-1.18043566e+00 9.28135335e-01 3.77317257e-02 7.26544917e-01
-2.41230950e-01 -2.79686987e-01 1.12895966e+00 9.27506626e-01
1.10665429e+00 3.98604751e-01 -6.24777973e-01 -2.28838563e-01
-1.10644460e+00 -8.15638714e-03 6.85414493e-01 6.20875657e-01
3.14448863e-01 9.35968697e-01 -1.43314721e-02 9.17193413e-01
-1.04058184e-01 6.75751090e-01 1.07846558e+00 -5.42536438e-01
1.15807451e-01 3.63158703e-01 -3.94741178e-01 -1.06048441e+00
-4.68955189e-01 -3.29671264e-01 -1.22365057e+00 -1.52567457e-02
1.30674988e-01 -1.94161847e-01 -9.99849796e-01 1.53225780e+00
-6.91733658e-02 5.40120304e-01 -2.29149358e-03 3.72233003e-01
7.07618356e-01 9.95969296e-01 -5.59215099e-02 -8.98404062e-01
6.49890304e-01 -9.60185289e-01 -1.10889411e+00 -6.80882558e-02
4.04331028e-01 -8.10106575e-01 8.64540815e-01 1.47714823e-01
-5.79694092e-01 -5.64126492e-01 -8.00551772e-01 5.13160229e-01
-6.99669957e-01 -5.08727133e-02 7.08401918e-01 4.10379916e-01
-7.21823573e-01 8.99246693e-01 -1.31242096e+00 -5.21688223e-01
3.14144790e-01 3.43000107e-02 1.00657336e-01 5.56894958e-01
-1.38422477e+00 5.70906222e-01 1.89557910e-01 -9.56198797e-02
-7.48239517e-01 -7.57204175e-01 -9.95577276e-01 -4.70172971e-01
-3.24931485e-03 -7.81821907e-02 1.49344146e+00 -7.99870610e-01
-1.25513279e+00 8.55658233e-01 -1.18708618e-01 -8.71150911e-01
2.73021281e-01 -1.16511218e-01 -1.27521431e+00 8.07074010e-02
1.99762404e-01 1.00645401e-01 1.24730599e+00 -6.22083724e-01
-3.93521339e-01 -3.46895248e-01 -5.75479925e-01 -3.99091274e-01
-5.18100739e-01 1.57431796e-01 2.25726724e-01 -1.58056414e+00
1.53558373e-01 -8.47907126e-01 -2.58082181e-01 -6.30740449e-02
-3.30657572e-01 -6.36189044e-01 1.46089876e+00 -4.99564588e-01
1.67077100e+00 -2.39942074e+00 -3.09856057e-01 2.94962436e-01
-9.39120501e-02 3.01716924e-01 7.67291151e-03 1.09169483e+00
-6.56185031e-01 -2.28675276e-01 -4.69118357e-01 -4.16063190e-01
2.40003373e-02 3.36638302e-01 -1.23715281e+00 7.00758219e-01
7.31946081e-02 9.95383382e-01 -1.06892788e+00 -1.74480230e-01
6.23084009e-01 8.23941976e-02 7.29867294e-02 1.51560798e-01
-2.52162248e-01 7.98127830e-01 -3.21065575e-01 8.92730057e-01
2.38736987e-01 -1.09776430e-01 -3.56496572e-01 -1.93300113e-01
-2.98527896e-01 1.11367725e-01 -7.27178395e-01 1.40033591e+00
-2.14264885e-01 7.89087296e-01 -7.64813602e-01 -1.50973010e+00
1.26571858e+00 4.53333646e-01 1.27076435e+00 -5.79093754e-01
-5.42781781e-03 4.28678155e-01 -2.98924327e-01 -5.47284186e-01
-7.57721588e-02 -5.59439398e-02 -3.45452368e-01 7.40974605e-01
8.18122402e-02 -2.73989230e-01 2.21311390e-01 -1.36894122e-01
9.58617508e-01 -1.31954908e-01 5.38198173e-01 -7.19091576e-03
7.47875214e-01 1.31388560e-01 5.30524135e-01 5.33512950e-01
-5.80471098e-01 3.28924209e-01 1.40444502e-01 -1.02021039e+00
-1.09145010e+00 -1.10449934e+00 -5.40634915e-02 1.10619485e+00
-3.52310449e-01 -4.66446042e-01 -3.29980701e-01 -7.13938773e-01
2.62202993e-02 8.41844559e-01 -6.93022013e-01 -3.70902270e-01
-5.37099719e-01 -9.30695176e-01 6.68922305e-01 6.35322332e-01
4.38376158e-01 -1.40437269e+00 -2.51601040e-01 2.78638482e-01
-2.02460349e-01 -8.55210364e-01 -4.51329827e-01 1.07968397e-01
-1.31531346e+00 -8.18711758e-01 -6.88174427e-01 -8.86143982e-01
5.97258866e-01 1.89892396e-01 1.05138266e+00 -4.84851003e-01
-3.01677585e-01 7.79919922e-01 -7.19697893e-01 -8.72266471e-01
-4.04568493e-01 -9.97500271e-02 4.15386528e-01 1.44997597e-01
7.85347044e-01 -9.88473594e-01 -4.12718505e-01 2.99445540e-01
-8.01877975e-01 -5.74077785e-01 -2.28092864e-01 6.64655387e-01
8.47073138e-01 5.72659671e-01 9.65623200e-01 -5.24487793e-01
7.11589754e-01 -6.58058226e-01 -4.47043955e-01 1.76221803e-01
-8.42864335e-01 -3.44008297e-01 1.13476646e+00 -7.80166686e-01
-5.91632605e-01 1.27120679e-02 -1.64893016e-01 -8.51920605e-01
-3.28786075e-01 5.36701739e-01 6.45957768e-01 4.43908051e-02
6.21962190e-01 7.50048935e-01 2.23575726e-01 -5.84355712e-01
2.03878298e-01 5.98770738e-01 5.40234149e-01 -3.30118537e-01
1.01707745e+00 8.30700099e-01 -2.75864184e-01 -7.94453323e-01
-1.08668792e+00 -5.03485560e-01 -7.39246249e-01 -5.31238854e-01
4.90306735e-01 -4.89014894e-01 -1.20775983e-01 1.05843508e+00
-6.28997028e-01 -3.35253507e-01 -8.44056845e-01 4.87604916e-01
-7.91014910e-01 4.25092101e-01 -7.83314645e-01 -9.36685324e-01
-5.72159469e-01 -5.07865846e-01 9.49421823e-01 -9.33611244e-02
-3.55673492e-01 -1.60244179e+00 4.62720364e-01 -2.64968425e-01
6.10388815e-01 7.40563095e-01 8.48816931e-01 -1.08797932e+00
3.18359107e-01 -6.59697771e-01 4.94898707e-01 4.91120577e-01
5.76859117e-01 4.34643812e-02 -1.03591990e+00 -5.63259661e-01
5.24168551e-01 -1.99817210e-01 4.59417433e-01 5.61124861e-01
1.60249579e+00 -2.38680437e-01 -2.89918005e-01 5.71148336e-01
1.03319275e+00 7.34794855e-01 2.16928199e-01 3.56486946e-01
5.96329153e-01 4.63108808e-01 8.77202392e-01 6.76146269e-01
1.19772233e-01 1.82866186e-01 7.56984204e-02 -1.82615533e-01
3.07966560e-01 -2.32657164e-01 4.04729128e-01 1.48902059e+00
-1.05067365e-01 -3.66085097e-02 -1.00567663e+00 4.77319092e-01
-1.83273089e+00 -1.25026608e+00 -4.33550365e-02 2.00705266e+00
5.95315933e-01 4.21053171e-02 4.76381630e-01 1.00855744e+00
7.23058522e-01 5.33630908e-01 -9.99481857e-01 -1.50472417e-01
-2.40006506e-01 1.41581282e-01 7.19434693e-02 -4.29694764e-02
-1.71697342e+00 6.36906981e-01 7.59469318e+00 9.05325472e-01
-1.48517680e+00 -3.45023796e-02 4.73412037e-01 1.37470052e-01
5.48677891e-02 -4.18533117e-01 -1.92188695e-01 6.33883178e-01
9.67728317e-01 -6.56668007e-01 4.00071681e-01 9.53220427e-01
3.92013907e-01 6.38094723e-01 -1.15986598e+00 1.39069831e+00
8.46298933e-02 -1.02144027e+00 -1.23946145e-01 -5.24544835e-01
8.37110102e-01 2.02001587e-01 1.39684543e-01 4.28452283e-01
-1.37121662e-01 -6.77791655e-01 5.84024906e-01 3.15871656e-01
6.68970346e-01 -6.22212052e-01 5.69163144e-01 2.87517786e-01
-1.57917559e+00 -2.89528072e-01 1.96489152e-02 -2.69963294e-01
3.16882998e-01 9.05859947e-01 -7.33577162e-02 6.46236181e-01
8.12169075e-01 1.49520302e+00 -3.70702714e-01 1.02418780e+00
3.89584005e-02 1.04929805e+00 -3.30878705e-01 2.17738166e-01
2.32037440e-01 -1.65110141e-01 8.41564715e-01 1.00172520e+00
3.45189810e-01 -4.82677557e-02 5.59139907e-01 4.26496714e-01
5.89288890e-01 1.66972995e-01 -7.47191966e-01 -4.91679639e-01
6.02936327e-01 7.29319274e-01 -7.72813141e-01 -4.02671129e-01
-6.46285057e-01 8.47524285e-01 -1.96943834e-01 4.51536477e-01
-7.91241050e-01 -5.63050926e-01 3.24024618e-01 -2.37055153e-01
-6.75679296e-02 -5.58228076e-01 -1.16850965e-01 -1.34978640e+00
3.46147418e-02 -8.72688353e-01 9.71419930e-01 -6.43982530e-01
-2.16277027e+00 8.03941667e-01 2.57622153e-01 -2.12078524e+00
-6.48558497e-01 -3.82730067e-01 -7.99563587e-01 3.43114197e-01
-1.17813921e+00 -8.95451248e-01 -1.91233248e-01 8.22611868e-01
9.37939346e-01 -7.39085138e-01 1.02866578e+00 4.32617158e-01
-6.58006132e-01 3.08376163e-01 3.71639311e-01 2.78814197e-01
6.69947267e-01 -1.18935597e+00 8.36710274e-01 9.02236104e-01
9.79277864e-02 4.37537044e-01 5.98434508e-01 -5.93009412e-01
-1.14663541e+00 -1.47709084e+00 8.15594733e-01 -1.72949538e-01
1.17049718e+00 -1.70477271e-01 -1.17889380e+00 8.07202816e-01
-3.74222212e-02 1.99119508e-01 9.03204083e-01 -1.88448578e-01
-2.89100587e-01 -4.24861431e-01 -1.07038534e+00 4.44590509e-01
8.19945514e-01 -7.31361032e-01 -9.45942402e-01 6.19994998e-01
3.49493206e-01 -1.32135600e-01 -9.79480028e-01 5.07560253e-01
1.85268223e-01 -7.18915761e-01 1.04855561e+00 -5.72493374e-01
4.69486713e-02 -1.76779449e-01 4.40345928e-02 -1.49008906e+00
-3.73985857e-01 -9.48753178e-01 -5.42762995e-01 1.16933155e+00
-1.33305937e-01 -1.01866853e+00 4.38490301e-01 8.02578405e-02
-2.94079274e-01 -5.62877834e-01 -7.24906623e-01 -1.49061120e+00
-4.03412208e-02 -8.60072494e-01 6.62073374e-01 1.49392343e+00
1.99196950e-01 -6.47451915e-03 -4.56956565e-01 -1.14392221e-01
9.59243357e-01 5.75834334e-01 4.33384806e-01 -1.40127265e+00
2.38092735e-01 -5.16143739e-01 -5.15214264e-01 -6.65056825e-01
4.25324649e-01 -9.37969267e-01 -1.84762672e-01 -1.12271774e+00
-4.19704646e-01 -4.39219475e-01 -5.53670108e-01 5.83857596e-01
2.09481746e-01 2.72413462e-01 -2.46937215e-01 7.45457530e-01
-4.11708355e-01 6.55764878e-01 8.75749111e-01 -1.82167470e-01
-3.35164964e-01 4.01986182e-01 1.19717762e-01 8.68224025e-01
1.37739003e+00 -3.38051140e-01 -7.29455829e-01 -9.46103781e-02
-2.92506516e-01 2.32627895e-02 2.51165062e-01 -9.06761646e-01
2.57087648e-01 -2.68270642e-01 1.36009589e-01 -1.09174454e+00
2.64576846e-03 -8.44089150e-01 -2.91119497e-02 6.77038550e-01
-2.89972275e-01 4.02858824e-01 3.21670771e-01 6.25660777e-01
-7.45464623e-01 1.46415513e-02 7.58168697e-01 8.61862823e-02
-1.09140396e+00 6.42141759e-01 -7.42831886e-01 1.21090591e-01
1.20892465e+00 -7.47050419e-02 -5.16211949e-02 -7.26186514e-01
-8.94960701e-01 1.67280674e-01 -8.86013284e-02 8.01932633e-01
4.10628378e-01 -1.75521159e+00 -4.86508995e-01 6.12902284e-01
3.82071853e-01 -4.70006377e-01 2.78688937e-01 7.12061226e-01
-2.72197247e-01 5.68934858e-01 -3.62792909e-01 -6.83622599e-01
-9.56367254e-01 1.01577425e+00 2.83796519e-01 -1.08319446e-01
-8.41003776e-01 2.93609798e-01 -2.47174874e-01 -3.48803580e-01
4.16052669e-01 -3.49285901e-01 -2.95996130e-01 5.92842139e-03
5.34197867e-01 6.45079136e-01 7.10044950e-02 -4.44125056e-01
-5.10101199e-01 7.93913722e-01 1.57678619e-01 1.18168637e-01
1.53439605e+00 -9.90596339e-02 -3.01233113e-01 1.34809911e+00
1.15370870e+00 -2.70547926e-01 -8.11192691e-01 -4.21936095e-01
1.38348462e-02 -1.08608015e-01 -2.37890601e-01 -3.03048402e-01
-1.09136677e+00 6.81962848e-01 6.88596249e-01 1.09225416e+00
1.68513691e+00 -3.03481948e-02 9.44195747e-01 2.63176829e-01
5.05951166e-01 -1.11014640e+00 2.28990167e-01 6.24136269e-01
1.03546810e+00 -1.15942109e+00 -1.51360378e-01 -3.11525941e-01
-3.78037900e-01 1.18439436e+00 3.47011805e-01 -3.95488024e-01
1.13444197e+00 3.90346467e-01 2.38671735e-01 -1.16553139e-02
-6.11364305e-01 -1.10441092e-02 5.24530828e-01 6.99439764e-01
3.86042058e-01 -3.68209854e-02 -5.09124584e-02 4.24882025e-01
-3.71929735e-01 -7.77925625e-02 -7.91851208e-02 1.09849989e+00
-1.37326464e-01 -1.17756665e+00 -4.46856797e-01 6.51667237e-01
-5.28082311e-01 2.22410232e-01 -1.42978355e-01 4.64072019e-01
-1.90566108e-01 1.08093822e+00 4.77581352e-01 -3.52053791e-01
4.28061575e-01 3.03098142e-01 1.75627932e-01 -3.12015623e-01
-4.35196131e-01 1.93456054e-01 -3.68798912e-01 -4.68903154e-01
-8.26453686e-01 -7.64904320e-01 -1.06973219e+00 -2.37757355e-01
-1.64702665e-02 9.34108421e-02 3.51517558e-01 8.69663775e-01
1.97300702e-01 5.00013292e-01 1.15104222e+00 -7.34255254e-01
-4.23852772e-01 -8.78825545e-01 -6.26407743e-01 6.11716509e-01
6.31035626e-01 -5.78226864e-01 -7.11662710e-01 3.89023423e-01]
|
[7.194713592529297, 2.897273540496826]
|
de1764b7-47df-4df4-bba5-d641ea183937
|
using-pre-trained-language-models-for
|
2204.0144
| null |
https://arxiv.org/abs/2204.01440v1
|
https://arxiv.org/pdf/2204.01440v1.pdf
|
Using Pre-Trained Language Models for Producing Counter Narratives Against Hate Speech: a Comparative Study
|
In this work, we present an extensive study on the use of pre-trained language models for the task of automatic Counter Narrative (CN) generation to fight online hate speech in English. We first present a comparative study to determine whether there is a particular Language Model (or class of LMs) and a particular decoding mechanism that are the most appropriate to generate CNs. Findings show that autoregressive models combined with stochastic decodings are the most promising. We then investigate how an LM performs in generating a CN with regard to an unseen target of hate. We find out that a key element for successful `out of target' experiments is not an overall similarity with the training data but the presence of a specific subset of training data, i.e. a target that shares some commonalities with the test target that can be defined a-priori. We finally introduce the idea of a pipeline based on the addition of an automatic post-editing step to refine generated CNs.
|
['Marco Guerini', 'Margherita Fanton', 'Helena Bonaldi', 'Serra Sinem Tekiroglu']
|
2022-04-04
| null |
https://aclanthology.org/2022.findings-acl.245
|
https://aclanthology.org/2022.findings-acl.245.pdf
|
findings-acl-2022-5
|
['automatic-post-editing', 'automatic-post-editing']
|
['computer-vision', 'natural-language-processing']
|
[ 2.47576624e-01 1.48421660e-01 4.86339554e-02 -1.22938149e-01
-8.23863029e-01 -7.01379836e-01 1.16837430e+00 1.48476347e-01
-5.92581153e-01 7.17546761e-01 5.50766468e-01 -2.43806258e-01
5.79222366e-02 -3.76385957e-01 -6.11936510e-01 -3.26620758e-01
2.06008971e-01 6.08730078e-01 3.10395241e-01 -6.35943472e-01
4.71456200e-01 4.46566939e-01 -1.22183359e+00 5.97824633e-01
5.29584885e-01 1.72863647e-01 2.90792793e-01 8.20077896e-01
-5.28679136e-03 1.26000762e+00 -1.46006632e+00 -6.21360600e-01
-1.48940980e-01 -9.31254685e-01 -9.74209905e-01 -1.99845731e-01
1.66767418e-01 -1.68419778e-01 2.18550395e-02 8.55221331e-01
7.28234470e-01 5.13850488e-02 9.50786889e-01 -9.32417154e-01
-3.08479130e-01 9.79696214e-01 -9.96567076e-04 4.12634343e-01
4.59138066e-01 2.05762073e-01 5.51289618e-01 -5.79672456e-01
1.07540858e+00 1.04720688e+00 5.34582913e-01 7.09337890e-01
-1.24980879e+00 -4.97172892e-01 -2.41357028e-01 3.23647410e-02
-1.31142092e+00 -5.81510484e-01 8.93791080e-01 -6.37319207e-01
9.24255848e-01 1.66524082e-01 4.09176081e-01 1.82441890e+00
2.93284148e-01 4.28896010e-01 1.19316959e+00 -6.83790326e-01
2.86410749e-01 5.54092765e-01 -7.79680237e-02 1.51575893e-01
5.16689643e-02 8.98253247e-02 -6.66021407e-01 -1.98308840e-01
4.00865138e-01 -1.01146042e+00 -1.14651576e-01 4.29231636e-02
-9.75225389e-01 1.16656053e+00 -7.75927305e-02 1.06337392e+00
-3.91320884e-01 4.23845239e-02 8.16598892e-01 8.40026587e-02
4.11666870e-01 9.38517034e-01 -6.70008808e-02 -4.30737942e-01
-1.36732411e+00 4.31208640e-01 8.99652898e-01 5.53567946e-01
8.38180631e-02 3.75630379e-01 -3.05336833e-01 6.68225825e-01
-8.14869329e-02 2.03041181e-01 5.13451636e-01 -5.27619839e-01
4.38075751e-01 2.67239213e-01 6.82585537e-02 -8.65866005e-01
-2.95356542e-01 -6.17149055e-01 -1.72783837e-01 9.69775245e-02
4.86988336e-01 -4.70161647e-01 -5.86748958e-01 1.78010678e+00
4.60514203e-02 -6.58667646e-03 2.11781234e-01 6.67987227e-01
6.11748815e-01 5.29381692e-01 1.99516729e-01 -3.44396025e-01
1.03187418e+00 -6.17001355e-01 -8.06035221e-01 -3.82624418e-01
9.39940333e-01 -1.08795214e+00 8.41716886e-01 3.91099632e-01
-8.54728043e-01 -3.98908675e-01 -1.07769811e+00 1.87654465e-01
-3.47422510e-01 1.19635150e-01 2.30956540e-01 8.95353317e-01
-5.99387884e-01 6.27795815e-01 -2.90542364e-01 -4.15425450e-01
-2.33470932e-01 -1.14574768e-01 -2.62093812e-01 4.19557303e-01
-1.30716860e+00 1.45013976e+00 7.07313716e-01 -4.73942384e-02
-1.13742292e+00 -3.24456245e-01 -8.21586728e-01 -4.64900821e-01
3.39560181e-01 -1.63908988e-01 1.20496821e+00 -1.34214771e+00
-1.45269120e+00 1.01138198e+00 8.63649845e-02 -5.67457914e-01
6.41048074e-01 -2.38792554e-01 -4.09438729e-01 1.04521483e-01
2.00418010e-01 2.51327604e-01 8.62438023e-01 -1.40203774e+00
-3.78672034e-01 7.58157223e-02 3.90063450e-02 -1.70297846e-02
-1.44023880e-01 8.55159521e-01 6.45392388e-02 -9.06993747e-01
-6.89174473e-01 -9.71148670e-01 7.59598315e-02 -9.30974483e-01
-6.59824848e-01 -1.78929448e-01 4.79918212e-01 -1.02674985e+00
1.71781135e+00 -1.99052048e+00 3.14577818e-01 2.39625275e-01
-2.29181468e-01 4.21669006e-01 -1.91036072e-02 9.38798308e-01
-3.64725322e-01 3.11218888e-01 -1.45648167e-01 -3.56415063e-01
-5.53201139e-02 -5.30730411e-02 -3.71529162e-01 4.75283056e-01
3.76362294e-01 5.10980844e-01 -7.88008630e-01 -4.55466717e-01
-9.08872187e-02 4.75062370e-01 -3.57845515e-01 3.32731873e-01
-2.51292914e-01 6.37864947e-01 -2.37289414e-01 8.71296972e-02
2.13918507e-01 4.72403616e-01 1.59073979e-01 1.30835518e-01
-3.74210209e-01 4.82615829e-01 -1.05995977e+00 1.31536174e+00
-1.97983995e-01 7.16077805e-01 -1.20395795e-01 -6.66215479e-01
1.03783941e+00 4.43024576e-01 -9.89063531e-02 -4.12403792e-01
5.15967369e-01 4.52797920e-01 3.46764565e-01 -5.71885645e-01
8.02801728e-01 -3.95965308e-01 -4.57721531e-01 4.91169631e-01
3.80824000e-01 -1.52565747e-01 4.58318472e-01 4.08078618e-02
9.48137343e-01 2.07888767e-01 2.72022754e-01 -3.25975865e-01
5.41808784e-01 4.30042595e-01 3.05986792e-01 9.07842219e-01
8.52659903e-03 7.47458577e-01 8.49198222e-01 -3.43502080e-03
-1.18635368e+00 -5.22355556e-01 1.91901803e-01 8.56168330e-01
-4.18942839e-01 -6.74532712e-01 -1.06982934e+00 -6.24025881e-01
-8.14498901e-01 1.68429983e+00 -6.94504797e-01 -1.53047040e-01
-6.63518190e-01 -3.74359846e-01 1.03562069e+00 -4.80960943e-02
-4.48658466e-02 -1.36464441e+00 -9.69275117e-01 8.15918148e-02
-2.98194647e-01 -1.05060458e+00 -2.35477611e-01 2.58877456e-01
-2.88264275e-01 -8.93370390e-01 -5.41867316e-01 -4.87203032e-01
3.54870319e-01 -1.87187299e-01 8.17624152e-01 1.73651829e-01
6.22564070e-02 1.47904456e-01 -7.94746816e-01 -7.39327431e-01
-1.32165325e+00 2.12058201e-01 -1.28771275e-01 -2.52865165e-01
2.18353078e-01 -7.39898980e-02 8.46234411e-02 -1.30099148e-01
-1.06672359e+00 1.30785257e-01 3.26714903e-01 7.07285225e-01
-7.36489818e-02 2.93650687e-01 1.38681337e-01 -1.02200603e+00
1.11411428e+00 -5.18341660e-01 -1.79848164e-01 1.09665073e-01
4.50543612e-02 3.45978290e-02 7.20193803e-01 -6.48212075e-01
-8.92797828e-01 -1.00848176e-01 -4.50889468e-01 -3.44105542e-01
-4.06301647e-01 7.37785459e-01 1.30346149e-01 3.49073678e-01
9.99375463e-01 7.30508640e-02 -1.42368600e-02 -4.46162730e-01
1.72393411e-01 4.72099066e-01 2.25985676e-01 -6.22991681e-01
9.70383346e-01 -3.63707580e-02 -3.17768633e-01 -8.74402821e-01
-5.35917759e-01 -1.41782045e-01 -7.23576367e-01 -4.87142175e-01
9.85247552e-01 -5.48495352e-01 -1.81714222e-01 4.91063982e-01
-1.57433438e+00 -6.31663799e-01 -1.83319762e-01 5.41653931e-01
-5.32232523e-01 1.35785043e-01 -4.65648383e-01 -1.05513108e+00
-2.34896496e-01 -1.18986750e+00 8.08681846e-01 5.33779152e-02
-8.83665085e-01 -9.02100861e-01 4.71738100e-01 3.18492979e-01
1.46639392e-01 3.43868464e-01 9.28947091e-01 -1.26117826e+00
1.59569055e-01 -3.69311333e-01 3.08957011e-01 3.70701909e-01
-2.43126959e-01 2.53322244e-01 -7.46369243e-01 -1.14975601e-01
1.34332791e-01 -3.33472580e-01 3.42128277e-01 4.15795296e-02
2.63082176e-01 -3.29717696e-01 -8.78499374e-02 6.67286525e-03
1.36653447e+00 2.95758426e-01 8.59751940e-01 6.08420491e-01
4.05638874e-01 1.07333708e+00 6.00982666e-01 2.82550871e-01
3.60761434e-02 9.80963290e-01 -2.57105809e-02 1.29833654e-01
-1.08599015e-01 -4.76497799e-01 9.46898639e-01 6.48391962e-01
3.39082293e-02 -2.68955261e-01 -1.11420667e+00 5.74699521e-01
-1.43844950e+00 -1.27992606e+00 -3.22583824e-01 2.16221666e+00
8.76873314e-01 5.04388213e-01 5.78709245e-01 2.24825859e-01
7.40801156e-01 1.09701887e-01 4.30223197e-01 -9.86612082e-01
-8.74109790e-02 4.51257408e-01 3.73568207e-01 6.88910067e-01
-9.08425808e-01 1.19674921e+00 6.54504728e+00 1.19426060e+00
-1.21477067e+00 1.96302146e-01 3.22820723e-01 1.74570251e-02
-2.00341642e-01 1.72409534e-01 -8.76059651e-01 6.26638651e-01
1.36124849e+00 -1.43157512e-01 1.96572021e-01 6.92700565e-01
4.43288237e-01 -2.00765565e-01 -8.25780213e-01 2.92389125e-01
6.15725338e-01 -1.19095910e+00 8.16820115e-02 2.10859012e-02
2.64307827e-01 -2.79304117e-01 -2.64831632e-01 2.40044221e-01
2.30602086e-01 -1.15416086e+00 1.44379997e+00 4.84854817e-01
4.21824217e-01 -7.88620949e-01 9.82007146e-01 7.37367928e-01
-4.81242776e-01 1.22374430e-01 2.49476563e-02 7.97162801e-02
1.31791621e-01 -1.56345665e-02 -1.06535947e+00 4.90284383e-01
2.95955807e-01 2.49699075e-02 -8.60066235e-01 7.76622415e-01
-5.75876474e-01 9.14541960e-01 -5.11394888e-02 -2.83440620e-01
3.86344254e-01 2.04845428e-01 1.02301860e+00 1.70959604e+00
3.47975433e-01 -1.41089648e-01 -3.17675583e-02 8.08747530e-01
6.48312867e-01 5.44184327e-01 -7.51399755e-01 -3.80347818e-01
2.79949278e-01 1.06071281e+00 -7.43382394e-01 -1.11564629e-01
-1.47327989e-01 9.72056687e-01 2.06143618e-01 -3.14461440e-02
-9.91319060e-01 -2.00202376e-01 9.94327068e-02 2.89301276e-01
2.73368955e-01 -3.43653291e-01 -7.22358301e-02 -7.89875031e-01
-3.90189230e-01 -1.33920741e+00 2.36650422e-01 -9.31264162e-01
-1.01959276e+00 1.00052691e+00 2.89043009e-01 -1.03877687e+00
-7.92676926e-01 -3.58180463e-01 -8.87465000e-01 8.88922930e-01
-8.71698618e-01 -1.36321318e+00 3.10961008e-01 2.89953798e-01
4.70336795e-01 -3.47135484e-01 9.16128516e-01 5.98034412e-02
-6.91514850e-01 3.83056164e-01 -3.11669409e-01 3.48223448e-01
6.27396107e-01 -1.01060092e+00 1.23235518e-02 1.24647164e+00
2.45702073e-01 6.20963871e-01 1.56567037e+00 -9.91136193e-01
-6.74796999e-01 -7.30784774e-01 1.34013546e+00 -5.45494676e-01
9.08670187e-01 -4.43695217e-01 -8.09519053e-01 3.82257044e-01
5.70288777e-01 -9.53355074e-01 7.17601895e-01 4.03537862e-02
-2.08125144e-01 5.52428484e-01 -8.79706264e-01 7.04898953e-01
6.19525313e-01 -5.57949901e-01 -9.19170976e-01 1.93477929e-01
3.85494322e-01 -3.95131737e-01 -4.83875126e-01 5.10555543e-02
1.78281769e-01 -9.89092946e-01 4.41854507e-01 -7.44224966e-01
8.10065329e-01 -2.19072551e-01 1.53665647e-01 -1.38872874e+00
-1.07598841e-01 -8.39347184e-01 4.61509347e-01 1.63657880e+00
5.48450470e-01 -1.54545814e-01 5.34120500e-01 4.57888782e-01
4.03358936e-02 -2.54875600e-01 -8.76277030e-01 -6.69556141e-01
4.13589716e-01 -7.05269933e-01 4.90247495e-02 9.37948227e-01
-5.18532582e-02 6.89698696e-01 -7.07329929e-01 -1.74396217e-01
1.93150088e-01 -4.46084350e-01 1.05905545e+00 -8.80567968e-01
-3.26171339e-01 -4.45267081e-01 -3.53757650e-01 -2.76453584e-01
4.49942261e-01 -8.36740375e-01 1.11404635e-01 -1.18780756e+00
1.40858933e-01 2.51467749e-02 1.11166596e-01 1.46670774e-01
-9.37545970e-02 4.41220356e-03 5.36204875e-01 1.80092424e-01
-8.86324197e-02 1.53468400e-01 7.14230597e-01 3.45664442e-01
-3.94805908e-01 -6.05410747e-02 -6.15378976e-01 5.94576657e-01
8.91506433e-01 -9.73246574e-01 -2.94825047e-01 -1.70125604e-01
2.71301180e-01 -1.21663548e-01 3.21137935e-01 -8.89793098e-01
1.51138082e-01 -1.90270424e-01 1.04282103e-01 -2.73830533e-01
2.95867801e-01 -5.01934826e-01 4.06698614e-01 4.98781949e-01
-5.57383120e-01 5.89220189e-02 4.17856187e-01 1.37544304e-01
-1.23621695e-01 -1.14580679e+00 6.72766328e-01 -1.43566638e-01
-5.50950468e-01 -6.63995028e-01 -1.09792507e+00 1.67599246e-01
9.46006656e-01 -3.34167093e-01 -2.19939888e-01 -6.83261991e-01
-5.10174453e-01 -3.60901415e-01 4.59296763e-01 5.28425157e-01
1.60090789e-01 -9.24719095e-01 -9.88285005e-01 -2.50631750e-01
3.13904397e-02 -8.91476989e-01 1.29525075e-02 7.37787902e-01
-5.82869470e-01 4.00412619e-01 -3.03401709e-01 1.17478885e-01
-1.42736316e+00 4.39870209e-01 2.71888226e-01 -5.90495110e-01
-3.18730146e-01 6.02659225e-01 -2.47840941e-01 -1.33302897e-01
-1.00258462e-01 4.53274131e-01 -5.79088569e-01 2.31572255e-01
5.63800931e-01 -1.53308809e-02 -1.29160449e-01 -1.39916337e+00
-2.40114942e-01 -6.47529587e-02 -4.64827903e-02 -5.63443542e-01
1.40441382e+00 3.28761578e-01 -1.24912309e-02 7.24460304e-01
8.29118609e-01 6.04481101e-01 -6.49591863e-01 1.80241182e-01
3.42239708e-01 -3.27863306e-01 -1.40107557e-01 -8.05613339e-01
-5.23342967e-01 5.77655554e-01 1.93930566e-01 3.13899010e-01
6.90279126e-01 2.71168221e-02 4.15097445e-01 -3.63572799e-02
3.16802084e-01 -1.47134387e+00 1.51109630e-02 7.19992280e-01
1.31816411e+00 -6.74511790e-01 1.47740450e-02 -3.68041217e-01
-1.05910003e+00 1.30556762e+00 4.50203598e-01 -2.48323739e-01
2.62425631e-01 2.86117584e-01 6.04225211e-02 -2.64861703e-01
-7.25013018e-01 -4.19595569e-01 2.58429855e-01 6.45902753e-01
7.28221714e-01 -1.24292478e-01 -7.71769583e-01 8.18038404e-01
-6.97616041e-01 -1.04253195e-01 7.13669538e-01 8.85383427e-01
-2.21688017e-01 -1.16062117e+00 -5.68308413e-01 9.95944589e-02
-7.10576952e-01 -8.39621499e-02 -1.15273380e+00 1.04785848e+00
3.19989651e-01 1.05632794e+00 -1.76470041e-01 -6.49370074e-01
3.99073780e-01 2.48683721e-01 4.43070680e-01 -1.00062323e+00
-1.19898570e+00 1.36651829e-01 8.09581876e-01 -4.70697582e-02
-3.27128291e-01 -7.58633018e-01 -7.02931941e-01 -4.00646985e-01
-4.80705857e-01 2.52224743e-01 6.95547462e-01 1.25965059e+00
-1.17815204e-01 3.22903305e-01 2.53114372e-01 -7.45679557e-01
-3.22295457e-01 -1.08842826e+00 -2.54787177e-01 5.26083291e-01
-1.89349458e-01 -5.90830982e-01 -3.48089278e-01 1.24016434e-01]
|
[11.640914916992188, 8.864846229553223]
|
d3fb28e3-db57-4b55-b127-05aaf9d2d15b
|
hurricane-forecasting-a-novel-multimodal
|
2011.06125
| null |
https://arxiv.org/abs/2011.06125v4
|
https://arxiv.org/pdf/2011.06125v4.pdf
|
Hurricane Forecasting: A Novel Multimodal Machine Learning Framework
|
This paper describes a novel machine learning (ML) framework for tropical cyclone intensity and track forecasting, combining multiple ML techniques and utilizing diverse data sources. Our multimodal framework, called Hurricast, efficiently combines spatial-temporal data with statistical data by extracting features with deep-learning encoder-decoder architectures and predicting with gradient-boosted trees. We evaluate our models in the North Atlantic and Eastern Pacific basins on 2016-2019 for 24-hour lead time track and intensity forecasts and show they achieve comparable mean absolute error and skill to current operational forecast models while computing in seconds. Furthermore, the inclusion of Hurricast into an operational forecast consensus model could improve over the National Hurricane Center's official forecast, thus highlighting the complementary properties with existing approaches. In summary, our work demonstrates that utilizing machine learning techniques to combine different data sources can lead to new opportunities in tropical cyclone forecasting.
|
['Dimitris Bertsimas', 'Théo Guénais', 'Cynthia Zeng', 'Léonard Boussioux']
|
2020-11-11
| null | null | null | null |
['hurricane-forecasting', 'tropical-cyclone-intensity-forecasting', 'tropical-cyclone-track-forecasting']
|
['computer-vision', 'time-series', 'time-series']
|
[-3.01940918e-01 -3.83263677e-01 -6.28853023e-01 -8.63330960e-01
-1.17629218e+00 -7.12506950e-01 1.08688033e+00 2.47976389e-02
-2.63453782e-01 8.44098270e-01 7.57335186e-01 -8.97216439e-01
1.29416427e-02 -6.99654520e-01 -3.70715618e-01 -6.31863952e-01
-5.18206000e-01 3.57746601e-01 -6.94284737e-01 -6.45837009e-01
1.71610564e-01 4.48931098e-01 -1.15516353e+00 3.36012870e-01
5.31471729e-01 1.13139498e+00 -2.26148695e-01 9.40035641e-01
2.02886254e-01 1.14128983e+00 -3.01329136e-01 -2.87133269e-02
5.07094681e-01 -6.88136294e-02 -2.83553034e-01 -5.63712120e-01
5.12857497e-01 -5.10858953e-01 -4.07342732e-01 4.82135743e-01
6.00088537e-01 5.73299043e-02 3.42336804e-01 -8.86241794e-01
1.93833243e-02 5.38237989e-01 -3.99661034e-01 6.85305178e-01
3.42606381e-02 3.60610366e-01 9.92030084e-01 -1.04934788e+00
2.38368630e-01 1.27178311e+00 1.10486317e+00 -7.12428568e-03
-1.35162079e+00 -9.91736054e-01 -5.44363111e-02 -1.98903650e-01
-1.12995207e+00 -7.10116386e-01 3.23408693e-01 -8.85474443e-01
1.41219127e+00 3.35032165e-01 5.06469250e-01 8.13737810e-01
9.20333803e-01 4.48753893e-01 1.41688514e+00 4.86676656e-02
2.12592594e-02 -1.33064345e-01 9.64502022e-02 4.37934369e-01
-2.51027822e-01 9.65050519e-01 -8.15069318e-01 -2.43005425e-01
6.23118207e-02 1.96263283e-01 -1.18846685e-01 5.34991682e-01
-1.43743289e+00 1.05173326e+00 4.19581383e-01 9.05812830e-02
-4.50696111e-01 3.35142106e-01 6.48035228e-01 5.57080030e-01
1.23001254e+00 5.73315442e-01 -9.49758649e-01 -3.23605776e-01
-1.81823981e+00 4.39020187e-01 8.50468099e-01 5.13670862e-01
7.81804860e-01 7.57109106e-01 -8.64976551e-03 2.93736577e-01
4.08848077e-01 1.51965141e+00 8.87274146e-02 -8.08617353e-01
5.32562613e-01 -1.29622007e-02 1.79636031e-01 -1.10672784e+00
-9.08306241e-01 -8.48799706e-01 -1.11870599e+00 2.14197531e-01
-3.27546075e-02 -1.02563977e+00 -6.83294535e-01 1.47010779e+00
-7.34951571e-02 3.08676124e-01 2.40830690e-01 1.00798190e+00
7.00848699e-01 1.20497000e+00 2.23172665e-01 -2.42871821e-01
1.10378182e+00 -6.76930606e-01 -9.38896060e-01 -3.80996108e-01
6.96809351e-01 -8.46084177e-01 1.96859390e-01 1.60262868e-01
-6.59822762e-01 -7.45439231e-01 -8.80113900e-01 2.33360037e-01
-6.51755512e-01 3.46486032e-01 7.67781138e-01 2.34141603e-01
-1.19403291e+00 4.71993089e-01 -9.26217258e-01 1.55592710e-01
-9.79986638e-02 1.79828525e-01 -2.26844326e-01 4.12768930e-01
-1.74204314e+00 1.34121382e+00 1.22131243e-01 3.94542128e-01
-1.20610893e+00 -9.70369518e-01 -1.20959401e+00 1.27999172e-01
-3.22472632e-01 -5.42229176e-01 1.42268121e+00 -6.55656099e-01
-1.37877142e+00 2.34032035e-01 -5.18093824e-01 -8.68223071e-01
1.19090170e-01 -4.54780370e-01 -8.53797972e-01 -3.70389938e-01
7.90600199e-03 6.84989274e-01 7.38520503e-01 -7.73981690e-01
-9.54121113e-01 -1.76464990e-01 -3.12464923e-01 8.44940096e-02
2.29361773e-01 1.61930993e-01 3.02534580e-01 -7.09374845e-01
-2.83556223e-01 -8.72719109e-01 -5.34229279e-01 -6.62356913e-01
-2.20128205e-02 8.35711360e-02 7.72011042e-01 -9.81320381e-01
1.54584074e+00 -1.97478080e+00 8.26219991e-02 2.41251305e-01
2.37277314e-01 2.14444306e-02 -1.20306745e-01 9.54783797e-01
7.46812746e-02 3.26844119e-02 -3.03676575e-01 -5.01132905e-01
7.19680265e-02 2.71143734e-01 -1.29927170e+00 6.49253607e-01
3.02371740e-01 8.46880674e-01 -9.37897146e-01 3.12900282e-02
5.95421612e-01 5.74647129e-01 -2.94526756e-01 2.41941214e-01
-1.36537626e-01 8.60153615e-01 3.39671820e-02 5.66394389e-01
8.36310863e-01 -6.87600374e-02 1.63343981e-01 2.50444748e-02
-6.52574420e-01 6.35969400e-01 -4.95015949e-01 1.19850838e+00
-9.82058167e-01 1.27455521e+00 1.22046158e-01 -4.82669294e-01
1.08178496e+00 4.17776525e-01 3.83545011e-01 -9.27153885e-01
-2.07593650e-01 3.14652145e-01 -2.24949226e-01 -4.83623028e-01
7.64414907e-01 -4.31629419e-01 -3.41651052e-01 4.63340014e-01
4.56408411e-02 -4.55538601e-01 -1.57127470e-01 1.57616865e-02
5.42940080e-01 -3.64035517e-02 4.20098715e-02 -5.57434618e-01
1.22537695e-01 1.44512519e-01 5.75868487e-01 7.32158363e-01
8.30211211e-03 2.61227399e-01 3.32328290e-01 -1.12798131e+00
-1.08680940e+00 -6.99137330e-01 -5.03643870e-01 1.35282695e+00
-6.48661315e-01 -6.16758049e-01 1.60205349e-01 -3.72842610e-01
2.67111808e-01 8.91489506e-01 -6.71776474e-01 2.53276706e-01
-5.25668979e-01 -1.14332080e+00 8.93350482e-01 4.73584890e-01
3.87713015e-01 -3.89314085e-01 -5.95117867e-01 2.92090148e-01
-8.75794739e-02 -9.34624076e-01 -1.28580928e-01 5.43190658e-01
-7.80080736e-01 -5.24869621e-01 -6.57701433e-01 -1.97639335e-02
-7.69269988e-02 -4.31413911e-02 1.40990877e+00 -4.99720633e-01
2.99061894e-01 -3.69275957e-02 -7.84006529e-03 -4.57873106e-01
-2.51643151e-01 2.40973800e-01 3.52937847e-01 -1.44139066e-01
4.59036797e-01 -3.69052440e-01 -5.97382843e-01 -1.58785284e-01
-5.41220367e-01 2.26417586e-01 3.83296996e-01 7.44147360e-01
1.53552413e-01 -4.95765209e-01 7.32285321e-01 -4.78196919e-01
5.38493216e-01 -1.00825918e+00 -1.17174733e+00 -1.33990869e-01
-8.65464509e-01 1.15193270e-01 5.79672873e-01 4.96218026e-01
-1.11198699e+00 -4.75599634e-04 -1.94842637e-01 -3.27452302e-01
-2.57732011e-02 1.17542720e+00 9.45859313e-01 1.33820206e-01
4.11671162e-01 4.79095340e-01 -1.06303729e-01 -6.56348526e-01
5.28213322e-01 7.86617398e-01 4.73502219e-01 -1.24270990e-01
6.40545130e-01 4.61607993e-01 2.93779988e-02 -6.46216810e-01
-1.08923745e+00 -4.75028366e-01 -6.38924360e-01 -3.28299552e-01
8.11621308e-01 -1.81874955e+00 -4.90022361e-01 4.33328569e-01
-1.22108901e+00 -4.10290629e-01 2.87260771e-01 1.03572881e+00
-7.65702799e-02 -1.68326735e-01 -7.76231945e-01 -9.62728262e-01
-6.22437179e-01 -1.04088092e+00 1.63092768e+00 -9.58774015e-02
-2.39488259e-01 -1.40462470e+00 9.28906202e-01 -8.24931264e-02
1.16234493e+00 3.35006535e-01 5.35256684e-01 -4.70776469e-01
-3.03757876e-01 -1.44484803e-01 -3.82121384e-01 1.86020657e-01
1.32432964e-04 9.06032994e-02 -1.19581509e+00 -5.71111381e-01
-2.45115727e-01 -3.69137466e-01 1.28627121e+00 6.04304552e-01
5.81362724e-01 -6.70715690e-01 -1.69343248e-01 9.82080281e-01
1.38441217e+00 -5.97740756e-03 1.74780245e-04 1.88770235e-01
2.36427814e-01 3.66307050e-01 4.75480318e-01 9.98309374e-01
6.17487907e-01 4.08565551e-01 1.25660300e-01 -5.41233301e-01
2.53425807e-01 -9.04974863e-02 6.13014519e-01 1.12512314e+00
-6.69666231e-02 -3.67960706e-02 -1.51825452e+00 4.67883795e-01
-1.66636884e+00 -1.18741667e+00 -8.91874433e-02 2.00354505e+00
6.56996846e-01 -3.10195506e-01 -3.34266990e-01 -4.17231321e-01
2.70491783e-02 9.08730268e-01 -7.31985718e-02 -8.58282924e-01
-3.64355057e-01 3.64590138e-01 1.21352661e+00 9.95383680e-01
-1.54892540e+00 9.57587957e-01 7.80856323e+00 5.54085076e-01
-1.82368731e+00 2.75229782e-01 7.83685029e-01 -1.99729294e-01
-3.37497681e-01 8.58808011e-02 -1.10410166e+00 3.44453126e-01
1.74181592e+00 -2.13042900e-01 3.27975333e-01 6.12619460e-01
9.18471098e-01 7.99940452e-02 -8.36885631e-01 6.07705116e-01
-1.95160002e-01 -1.82656980e+00 -2.07994118e-01 -1.10303111e-01
1.04622388e+00 1.06152618e+00 3.03899676e-01 6.61238849e-01
6.46435499e-01 -1.30028617e+00 4.03472215e-01 9.10171926e-01
9.64110970e-01 -7.94168949e-01 1.17739367e+00 2.50554860e-01
-1.33102858e+00 -8.27651564e-03 -1.58963636e-01 -4.77933377e-01
3.98886144e-01 9.57642198e-01 -7.85644174e-01 6.09538615e-01
6.55840039e-01 1.33474398e+00 -2.47941241e-01 6.10520899e-01
1.07606709e-01 1.06818402e+00 -4.56789851e-01 3.61950696e-01
8.62990260e-01 2.81596109e-02 5.29628515e-01 1.74737537e+00
5.08041501e-01 -7.47307017e-02 3.35989356e-01 3.57629418e-01
1.65656969e-01 -1.07795589e-01 -1.06924558e+00 2.15847179e-01
3.41008961e-01 9.48836505e-01 2.76045114e-01 -7.11531520e-01
-5.18508077e-01 1.71209604e-01 -1.26146019e-01 5.22353888e-01
-9.16325271e-01 -2.32132271e-01 7.39094615e-01 -2.64994383e-01
1.03472590e-01 -7.47354925e-01 -5.07171273e-01 -1.20244861e+00
-6.57093287e-01 -1.04379606e+00 2.14201406e-01 -7.15469122e-01
-1.05443561e+00 8.74897897e-01 1.09387495e-01 -1.55815589e+00
-6.42305732e-01 -3.06219041e-01 -6.23448491e-01 1.22014642e+00
-2.01290941e+00 -1.10323238e+00 2.36735791e-01 1.75243363e-01
5.10774374e-01 -4.51045156e-01 8.18420768e-01 2.76268899e-01
-3.84613514e-01 8.25563222e-02 7.60492384e-01 5.70592359e-02
7.26873577e-01 -1.16674995e+00 6.74606979e-01 6.35225177e-01
2.44693123e-02 5.38579106e-01 7.43274271e-01 -5.74545145e-01
-1.53762376e+00 -1.46294916e+00 1.53565133e+00 -4.16373402e-01
8.84227335e-01 -3.08296621e-01 -6.03893638e-01 9.12469864e-01
8.31830025e-01 -2.37260893e-01 7.24419355e-01 4.22285229e-01
-3.79073501e-01 -3.54086816e-01 -5.31950414e-01 2.46046498e-01
-8.31686929e-02 -8.96479666e-01 -5.36095738e-01 6.66072786e-01
4.45464581e-01 -4.43756789e-01 -1.11603081e+00 7.36938477e-01
5.79428434e-01 -6.23586774e-01 5.32638490e-01 -9.38869059e-01
6.18437946e-01 -2.42059365e-01 -4.45437044e-01 -1.82602227e+00
-3.18536013e-01 -7.43472636e-01 -1.53713018e-01 3.91853154e-01
7.33277798e-01 -7.14862466e-01 1.42861798e-01 1.76538229e-01
-1.82470903e-01 -5.38605213e-01 -1.15387666e+00 -4.57875788e-01
4.38096106e-01 -8.43229353e-01 5.86089551e-01 1.08998322e+00
6.07932620e-02 2.61846513e-01 -1.16166079e+00 5.92583835e-01
3.61324430e-01 5.67019165e-01 6.11866474e-01 -8.91345978e-01
-3.39608304e-02 -3.76088351e-01 5.77937551e-02 -1.26402485e+00
3.56972456e-01 -9.12732005e-01 -1.17106907e-01 -1.03187203e+00
-1.46349728e-01 -3.84481996e-01 -3.55597138e-01 7.34565496e-01
1.81783259e-01 2.83166379e-01 1.23774208e-01 2.17188507e-01
-3.47771734e-01 7.89907694e-01 7.66946852e-01 -3.43172222e-01
6.26470670e-02 -4.61268574e-02 7.97978044e-02 4.33168650e-01
7.97876179e-01 -4.49410528e-01 2.12544560e-01 -8.72921765e-01
5.24745703e-01 6.56277120e-01 2.18210846e-01 -9.82545316e-01
2.32552797e-01 -2.42385402e-01 4.96485502e-01 -8.60141635e-01
2.25074649e-01 -5.23729503e-01 1.22991621e-01 6.43092513e-01
-4.45958734e-01 7.96714723e-01 8.64407361e-01 3.75150472e-01
-7.42685199e-01 7.66862929e-01 6.01937652e-01 1.23360120e-01
-8.46175313e-01 1.95244357e-01 -1.15518236e+00 -2.12149903e-01
5.39844394e-01 6.33239567e-01 -4.89393890e-01 -8.34326923e-01
-6.51091456e-01 7.13316500e-01 -1.21352598e-01 6.03608489e-01
4.37985539e-01 -1.01077807e+00 -1.38935721e+00 5.04754424e-01
1.50855992e-03 -7.61383414e-01 1.58121839e-01 1.13840127e+00
-6.63782001e-01 1.18851197e+00 8.59272033e-02 -6.79839909e-01
-7.48594522e-01 1.48478031e-01 6.35725975e-01 -4.01629329e-01
-2.45303795e-01 4.97573435e-01 -1.99453846e-01 -8.60156059e-01
-1.24488629e-01 -5.35327196e-01 -1.15435876e-01 3.04968596e-01
5.47422826e-01 1.04982652e-01 1.36042014e-01 -6.99113429e-01
-3.97793353e-01 5.62801957e-01 4.04133677e-01 -3.81810129e-01
1.24948490e+00 -1.73073709e-01 -3.17791961e-02 6.66666090e-01
1.29911041e+00 3.97201404e-02 -1.00882733e+00 -2.18292654e-01
2.77048890e-02 -2.76803404e-01 7.60448873e-01 -1.08226633e+00
-1.21830070e+00 1.31318402e+00 7.99927413e-01 -1.78868040e-01
8.58692646e-01 -5.30249953e-01 8.84118855e-01 7.09977806e-01
7.91206062e-02 -7.26994157e-01 -9.09409821e-01 1.09734213e+00
7.99243331e-01 -1.79098272e+00 -3.04226857e-03 7.26228893e-01
-7.44110346e-01 1.37506354e+00 9.11589712e-03 5.52952923e-02
1.17469323e+00 6.55284107e-01 6.09576464e-01 -3.03649187e-01
-1.55164170e+00 -6.88471273e-02 4.36550826e-01 -5.60940541e-02
7.60860145e-01 4.77206916e-01 -1.20181993e-01 1.16226628e-01
-2.58529991e-01 4.06393707e-02 1.48770258e-01 6.38352692e-01
-5.11803448e-01 -5.64448714e-01 -3.96210849e-01 6.03857398e-01
-5.50719380e-01 -8.38959396e-01 1.78639412e-01 6.11181498e-01
-1.60079058e-02 1.00595462e+00 3.47494930e-01 -5.89205444e-01
-2.53059179e-01 1.32650957e-01 -4.53212380e-01 -1.95748791e-01
-9.02217269e-01 6.36272952e-02 3.84144932e-01 -6.76177680e-01
-2.43540809e-01 -4.63717848e-01 -5.88727474e-01 -8.51298451e-01
1.31260529e-01 4.74250317e-01 8.48025322e-01 1.00120187e+00
5.87972999e-01 3.75906050e-01 1.12551546e+00 -1.25038087e+00
-6.92921281e-01 -1.18693483e+00 -2.62569726e-01 -2.24837899e-01
1.14745963e+00 -2.95565218e-01 -5.84739089e-01 -7.08711818e-02]
|
[6.583719253540039, 2.919562816619873]
|
68d5527e-0e6b-4a8a-aabd-ecb8ddc245df
|
sensor-fault-detection-and-isolation-in
|
2304.08837
| null |
https://arxiv.org/abs/2304.08837v1
|
https://arxiv.org/pdf/2304.08837v1.pdf
|
Sensor Fault Detection and Isolation in Autonomous Nonlinear Systems Using Neural Network-Based Observers
|
This paper presents a new observer-based approach to detect and isolate faulty sensors in industrial systems. Two types of sensor faults are considered: complete failure and sensor deterioration. The proposed method is applicable to general autonomous nonlinear systems without making any assumptions about its triangular and/or normal form, which is usually considered in the observer design literature. The key aspect of our approach is a learning-based design of the Luenberger observer, which involves using a neural network to approximate the injective map that transforms the nonlinear system into a stable linear system with output injection. This learning-based Luenberger observer accurately estimates the system's state, allowing for the detection of sensor faults through residual generation. The residual is computed as the norm of the difference between the system's measured output and the observer's predicted output vectors. Fault isolation is achieved by comparing each sensor's measurement with its corresponding predicted value. We demonstrate the effectiveness of our approach in capturing and isolating sensor faults while remaining robust in the presence of measurement noise and system uncertainty. We validate our method through numerical simulations of sensor faults in a network of Kuramoto oscillators.
|
['Karl Henrik Johansson', 'Muhammad Umar B. Niazi', 'John Cao']
|
2023-04-18
| null | null | null | null |
['fault-detection']
|
['miscellaneous']
|
[ 4.74243850e-01 5.95133185e-01 1.16449157e-02 3.43403339e-01
-3.50950271e-01 -5.69799066e-01 1.94708630e-01 3.29109222e-01
1.09165214e-01 6.14228845e-01 -5.70591569e-01 -3.53983879e-01
-3.55106205e-01 -2.43361145e-01 -1.06100285e+00 -9.28171217e-01
-4.26847488e-02 3.32839936e-02 2.01204345e-01 -1.34825855e-01
-1.20872566e-02 4.23070818e-01 -1.08829486e+00 -6.44336045e-01
6.04683220e-01 1.18976545e+00 -2.21569240e-01 9.43972945e-01
1.09164691e+00 6.84079528e-01 -9.44625914e-01 7.41520286e-01
4.25004810e-01 -6.19469047e-01 -2.42641971e-01 4.03380066e-01
2.81372778e-02 -2.85834759e-01 -5.49574316e-01 1.40076780e+00
3.56941402e-01 -1.67786181e-01 6.57257497e-01 -1.31121719e+00
-3.92856807e-01 4.34284180e-01 8.80301893e-02 -1.02524422e-01
4.04893577e-01 1.87567517e-01 2.70952046e-01 -5.92782676e-01
2.08834767e-01 8.64162266e-01 6.71101332e-01 3.07958066e-01
-1.41922247e+00 -2.81392246e-01 -1.50268286e-01 -1.90382034e-01
-1.33589768e+00 -3.47662508e-01 7.61063218e-01 -7.22251475e-01
6.71594322e-01 2.12539688e-01 4.35120404e-01 6.36368155e-01
8.53325129e-01 2.75914371e-01 8.61593664e-01 -3.38029891e-01
5.71083248e-01 6.04511313e-02 1.94906309e-01 5.60101330e-01
7.22401381e-01 5.05783617e-01 5.05476678e-03 -2.50067741e-01
1.12674642e+00 7.52056539e-02 -7.01896787e-01 -5.87161839e-01
-1.04461658e+00 4.45995897e-01 4.04839337e-01 3.09976459e-01
-5.43580472e-01 2.68100262e-01 1.47924349e-01 8.43613625e-01
5.36953025e-02 6.85709298e-01 -1.82094857e-01 1.72873169e-01
-2.63008326e-01 9.45519879e-02 1.12990677e+00 8.70318592e-01
3.39876294e-01 9.80482876e-01 3.58694941e-01 1.98267356e-01
3.56363118e-01 9.53065455e-01 6.40066803e-01 -9.36528504e-01
3.26258913e-02 6.73558712e-01 5.31995833e-01 -8.44757140e-01
-2.66089916e-01 -3.31082135e-01 -1.02150154e+00 5.92011154e-01
4.74525362e-01 -6.66709304e-01 -9.31380808e-01 1.46899486e+00
2.08418250e-01 3.98172557e-01 3.50773007e-01 7.68354118e-01
-1.27991512e-01 6.94771290e-01 -8.49624991e-01 -6.71777010e-01
6.70384526e-01 -4.01413292e-01 -1.08576846e+00 -1.64436236e-01
3.08722705e-01 -3.03904593e-01 3.91005248e-01 4.79215235e-01
-1.05928671e+00 -4.20810997e-01 -1.51429141e+00 8.09508324e-01
-1.51870891e-01 1.82767913e-01 -7.56594479e-01 2.72954762e-01
-8.58319640e-01 7.87705123e-01 -1.00553071e+00 -8.96472856e-02
-6.55723989e-01 5.66128731e-01 -3.21204513e-01 5.18718302e-01
-1.27892148e+00 1.23555064e+00 4.43031013e-01 8.21691573e-01
-1.01908207e+00 -2.93341428e-01 -7.67260790e-01 4.89977784e-02
5.69728255e-01 -2.88365424e-01 1.35427308e+00 -1.04364705e+00
-1.71041763e+00 -3.38698238e-01 -5.21102510e-02 -5.05835772e-01
2.80674100e-01 2.80027259e-02 -4.28088635e-01 2.29389649e-02
-3.71447712e-01 -3.98895890e-01 1.21214068e+00 -1.18501794e+00
-1.95621803e-01 -2.23956347e-01 -4.09037352e-01 -1.19792499e-01
-2.63895124e-01 -7.02897370e-01 6.07801676e-01 -2.27557212e-01
4.89145160e-01 -9.75723982e-01 -3.35060745e-01 -1.76120177e-01
-4.49259490e-01 1.10498011e-01 9.45313752e-01 -4.67320502e-01
1.10585749e+00 -1.95708406e+00 3.08035254e-01 6.82086349e-01
3.14586074e-03 4.46483314e-01 1.61092639e-01 8.21888804e-01
-1.87377647e-01 -4.10061330e-01 -3.63577813e-01 2.50204325e-01
-2.92744543e-02 4.14654315e-02 -4.92539495e-01 9.06441033e-01
5.00082195e-01 3.18826944e-01 -8.09243500e-01 3.67643148e-01
3.45239341e-01 5.30740261e-01 1.25561059e-01 3.57458293e-01
4.39245403e-02 4.97157216e-01 -2.92344421e-01 3.14790428e-01
8.68824497e-02 1.28702939e-01 9.28649083e-02 2.43380498e-02
-1.69612855e-01 -4.01619449e-02 -1.62865067e+00 5.05210042e-01
-3.41885746e-01 6.01390123e-01 5.57341158e-01 -1.25033641e+00
1.21157873e+00 1.07592845e+00 1.82861939e-01 -2.78050840e-01
5.84511936e-01 5.79847038e-01 1.62489086e-01 -4.15320545e-01
-2.30330110e-01 -1.74708422e-02 -1.12887412e-01 1.98079333e-01
-7.45899528e-02 -4.97099549e-01 -3.68990600e-02 -2.70951480e-01
1.30233634e+00 -3.17250729e-01 5.08173883e-01 -3.43642920e-01
7.69130886e-01 -2.17518985e-01 7.44150162e-01 5.90391397e-01
-6.92799017e-02 6.48179799e-02 8.85929286e-01 -1.25910148e-01
-1.10269916e+00 -8.22940707e-01 -9.94217675e-03 -5.75788394e-02
2.30383456e-01 4.28713381e-01 -6.52124107e-01 -1.40395924e-01
3.30867678e-01 3.94652039e-01 -5.43739378e-01 -9.84638214e-01
-5.93356371e-01 -1.48874134e-01 4.91198778e-01 4.45060313e-01
-1.84573792e-02 -5.27727306e-01 -7.16781318e-01 5.12377441e-01
4.52907205e-01 -5.54124355e-01 -1.48575097e-01 5.66187799e-01
-9.14967299e-01 -1.22898483e+00 -4.85498697e-01 -8.84784698e-01
1.18242705e+00 -4.02175844e-01 1.73399031e-01 -1.91127688e-01
-1.51558429e-01 5.14449120e-01 6.29849657e-02 -3.55273396e-01
-1.01239896e+00 -3.08129728e-01 6.39009833e-01 8.11785832e-02
-4.20037389e-01 -1.11648761e-01 -1.14569031e-01 5.25208712e-01
-1.04203868e+00 -7.39424407e-01 3.98776829e-01 1.08646131e+00
2.14209035e-01 5.80580473e-01 9.68601465e-01 -6.02790177e-01
1.02821898e+00 -4.70661342e-01 -1.32682395e+00 6.71502799e-02
-8.94966245e-01 3.82221825e-02 1.05981326e+00 -7.69009233e-01
-5.26303411e-01 5.70332885e-01 2.67538846e-01 -5.91828763e-01
7.16515109e-02 4.76649016e-01 -8.92050788e-02 -2.89491594e-01
6.24352992e-01 2.10497275e-01 6.72697783e-01 -1.73071265e-01
3.29524763e-02 7.96396911e-01 7.65502930e-01 -1.50969606e-02
8.66982400e-01 -1.77462175e-01 3.32552433e-01 -9.02567744e-01
-4.91353422e-02 -3.51212293e-01 -4.44993198e-01 -3.39539707e-01
4.81328964e-02 -6.79349303e-01 -1.22042704e+00 9.29923713e-01
-1.07418442e+00 -2.57762253e-01 -5.47717333e-01 5.31958818e-01
-6.48325920e-01 7.10997730e-02 -8.04897070e-01 -1.39364755e+00
2.17700331e-03 -1.23034048e+00 7.60582983e-01 1.87340587e-01
-1.74559638e-01 -1.13235354e+00 3.15036744e-01 -5.31436145e-01
3.32455069e-01 5.37170231e-01 5.57365835e-01 -5.18326700e-01
-3.53351623e-01 -1.06088877e+00 4.86438364e-01 9.09583151e-01
1.46674082e-01 -2.86233928e-02 -6.90188527e-01 -8.34976912e-01
7.81694472e-01 4.84798886e-02 3.29098254e-01 2.94992864e-01
-1.96232125e-01 -5.72286487e-01 -2.83751220e-01 -1.00807525e-01
1.56300282e+00 6.66912198e-01 8.42691809e-02 4.63485718e-04
6.65660143e-01 2.73251504e-01 3.19422305e-01 1.43025666e-01
-2.23703936e-01 3.46961677e-01 6.48055315e-01 6.35873750e-02
5.45240641e-01 1.44007593e-01 8.59415948e-01 8.15316617e-01
3.41692895e-01 -2.44838998e-01 -7.38471210e-01 6.52012050e-01
-1.91659617e+00 -3.46744239e-01 -1.86789230e-01 2.52061272e+00
5.00358105e-01 2.06415415e-01 -1.28920302e-01 8.22187126e-01
9.47546184e-01 -5.20265818e-01 -9.60468054e-01 -6.15800083e-01
1.61256358e-01 -1.25213176e-01 8.89876604e-01 9.39069033e-01
-8.92927468e-01 -1.13744559e-02 6.28840113e+00 -1.48613423e-01
-1.21220076e+00 -4.21498358e-01 6.81538954e-02 4.06916112e-01
4.50724542e-01 -1.98895767e-01 -4.30387199e-01 3.96732301e-01
1.35300589e+00 -3.26645106e-01 3.87900710e-01 7.08967388e-01
2.94496566e-01 -2.27183804e-01 -1.16020656e+00 3.14197749e-01
1.31628260e-01 -6.30356789e-01 -6.75011694e-01 -1.61333129e-01
8.90224934e-01 -6.63867176e-01 1.46362334e-01 -2.95125935e-02
1.52955547e-01 -5.57381511e-01 6.55585885e-01 7.98616827e-01
2.74480999e-01 -5.25464356e-01 1.18890834e+00 4.41479325e-01
-8.75433564e-01 -6.01924121e-01 -1.71645626e-01 -5.28541148e-01
1.85608685e-01 7.49636233e-01 -1.18588483e+00 2.50514895e-01
-6.00189269e-02 6.01688981e-01 -1.74915731e-01 1.05886769e+00
-3.14194947e-01 7.20819771e-01 -4.87252563e-01 -2.36890599e-01
-1.09770253e-01 -1.41280562e-01 9.39408243e-01 4.80590671e-01
4.77682292e-01 -2.46652633e-01 3.28708470e-01 7.31971920e-01
3.60205650e-01 -5.29780626e-01 -8.81501079e-01 7.00030550e-02
6.79202437e-01 6.80094063e-01 -4.20049220e-01 -4.23745036e-01
1.56062037e-01 6.60377681e-01 -1.77349612e-01 4.88722026e-01
-3.70910794e-01 -9.46328580e-01 3.35162878e-01 -2.65839882e-02
1.35367125e-01 -1.00414552e-01 -2.15104058e-01 -7.48853385e-01
3.12982053e-01 -6.50786519e-01 -1.14676446e-01 -5.11101604e-01
-9.23689127e-01 6.16410196e-01 -1.26717985e-01 -1.45761502e+00
-1.08171678e+00 -5.40010452e-01 -5.12916982e-01 9.09122944e-01
-7.92145729e-01 -3.30302119e-01 8.55315477e-02 3.01246822e-01
-5.26272580e-02 1.05171636e-01 6.95294499e-01 -3.44379470e-02
-8.89425457e-01 -3.23637458e-03 6.82002246e-01 -5.20768017e-02
6.13357961e-01 -1.52190244e+00 7.74521157e-02 1.06065047e+00
-6.74869418e-01 4.74906474e-01 1.28999817e+00 -6.96800351e-01
-1.68282807e+00 -1.11137748e+00 6.68823838e-01 -1.48442745e-01
1.03269565e+00 -1.57177135e-01 -9.52492535e-01 1.02316463e+00
8.80027488e-02 1.98892355e-01 -5.16790450e-01 -7.76542008e-01
1.05488665e-01 -5.57201028e-01 -1.18124592e+00 3.23795110e-01
-5.86475842e-02 -4.64091629e-01 -6.57163978e-01 1.07364759e-01
4.94140416e-01 -6.42149508e-01 -1.00113404e+00 4.32891518e-01
2.92038769e-01 -3.18124652e-01 5.23645461e-01 1.45803988e-02
-1.30217820e-01 -6.93218768e-01 3.04245234e-01 -1.85328424e+00
-1.75511062e-01 -9.07056272e-01 -4.31841165e-01 7.65784919e-01
3.02592218e-01 -1.24396503e+00 2.98053771e-01 3.01505059e-01
-4.43490893e-02 -5.99665582e-01 -1.01910090e+00 -1.13970327e+00
-5.25857024e-02 3.31354260e-01 -2.95035280e-02 5.96967876e-01
6.00540578e-01 1.98170066e-01 -3.02913755e-01 8.80803466e-01
5.02836704e-01 -4.27812308e-01 4.32678401e-01 -9.73265946e-01
-2.02211171e-01 -1.27650931e-01 -7.24980474e-01 -5.72597802e-01
1.25121593e-01 -2.45905757e-01 8.65987182e-01 -1.21451950e+00
-8.18784058e-01 8.71834084e-02 -4.35909241e-01 2.50028878e-01
-4.90004495e-02 1.19131900e-01 -1.38298661e-01 9.28872004e-02
-3.90739962e-02 3.17280233e-01 6.99768960e-01 -9.03238654e-02
-3.74027073e-01 4.18515503e-01 1.43872760e-02 6.29948199e-01
8.15305352e-01 -3.72332036e-01 -5.28961778e-01 -1.19127974e-01
-1.46161914e-01 7.28721201e-01 5.07939816e-01 -1.54189920e+00
4.80571389e-01 1.42239094e-01 1.99464843e-01 -1.31729230e-01
1.26237750e-01 -1.34364057e+00 5.68430603e-01 1.20244801e+00
-2.41642550e-01 2.57617146e-01 -1.17344871e-01 6.97390258e-01
-4.62039828e-01 -3.34813118e-01 7.83195853e-01 3.60186130e-01
-1.62429884e-02 -4.40158397e-01 -8.97670686e-01 -5.70764899e-01
1.10326755e+00 2.57358467e-03 -2.94689506e-01 -5.67981482e-01
-8.97059917e-01 3.43879730e-01 3.31216007e-01 3.15804295e-02
5.32459259e-01 -1.06250560e+00 -4.44419026e-01 7.58971870e-01
-2.05531791e-01 -1.52822286e-01 -1.39013320e-01 8.76531661e-01
-1.51933968e-01 5.76497197e-01 4.11613733e-02 -6.90288007e-01
-9.63723361e-01 5.77172220e-01 8.04764807e-01 1.70486271e-01
-1.07705697e-01 1.65576547e-01 -4.04128045e-01 -3.14388573e-01
2.28232712e-01 -8.05046439e-01 1.38007253e-01 -3.97808552e-01
1.93206698e-01 5.64583659e-01 1.49688050e-01 -3.40410143e-01
-4.11931723e-02 4.40780044e-01 4.16588634e-01 -1.03782386e-01
8.59541893e-01 -7.69562349e-02 -2.17499584e-01 1.15398979e+00
8.68263006e-01 -4.75349367e-01 -1.49623621e+00 1.22601623e-02
4.36496064e-02 1.40629098e-01 1.20899817e-02 -5.98892689e-01
-9.31945682e-01 3.25808764e-01 7.87760973e-01 8.93119514e-01
1.05121374e+00 -4.48506087e-01 4.04043585e-01 6.31725669e-01
2.44373962e-01 -8.44592452e-01 -1.78671390e-01 5.16196489e-01
8.00368905e-01 -7.24463940e-01 -1.01036616e-01 -2.63956040e-01
-9.91572067e-02 1.26085067e+00 4.84102100e-01 -1.07921541e+00
7.72476792e-01 6.26457274e-01 2.67885953e-01 3.29158425e-01
-7.84354746e-01 3.13239068e-01 5.42596392e-02 5.06339550e-01
5.19138835e-02 -9.64893922e-02 -2.13816464e-01 1.98602807e-02
4.29437250e-01 -2.84508150e-03 1.20056903e+00 1.21277201e+00
-6.40388966e-01 -6.66899502e-01 -8.64224434e-01 1.97076619e-01
-1.50465846e-01 5.90816498e-01 -1.58616796e-01 8.16964447e-01
-2.42085859e-01 9.91220832e-01 1.34681299e-01 -1.16704769e-01
9.40659940e-01 1.39435217e-01 2.04894468e-01 -5.11311114e-01
-4.38513279e-01 3.22649390e-01 -3.27182055e-01 -3.92613739e-01
-8.71868730e-02 -5.53455710e-01 -1.05821800e+00 2.79273570e-01
-8.58404756e-01 3.56707603e-01 6.64998472e-01 8.03794980e-01
9.37130302e-02 9.26769376e-01 8.83774996e-01 -7.19006658e-01
-1.22247326e+00 -9.87301707e-01 -7.50408530e-01 -8.40250105e-02
1.19721937e+00 -6.54706359e-01 -9.35824335e-01 1.66276991e-01]
|
[5.37402868270874, 2.5767807960510254]
|
1a73de4c-d28b-4b37-9bab-3c504e76ebbd
|
darkvision-a-benchmark-for-low-light-image
|
2301.06269
| null |
https://arxiv.org/abs/2301.06269v1
|
https://arxiv.org/pdf/2301.06269v1.pdf
|
DarkVision: A Benchmark for Low-light Image/Video Perception
|
Imaging and perception in photon-limited scenarios is necessary for various applications, e.g., night surveillance or photography, high-speed photography, and autonomous driving. In these cases, cameras suffer from low signal-to-noise ratio, which degrades the image quality severely and poses challenges for downstream high-level vision tasks like object detection and recognition. Data-driven methods have achieved enormous success in both image restoration and high-level vision tasks. However, the lack of high-quality benchmark dataset with task-specific accurate annotations for photon-limited images/videos delays the research progress heavily. In this paper, we contribute the first multi-illuminance, multi-camera, and low-light dataset, named DarkVision, serving for both image enhancement and object detection. We provide bright and dark pairs with pixel-wise registration, in which the bright counterpart provides reliable reference for restoration and annotation. The dataset consists of bright-dark pairs of 900 static scenes with objects from 15 categories, and 32 dynamic scenes with 4-category objects. For each scene, images/videos were captured at 5 illuminance levels using three cameras of different grades, and average photons can be reliably estimated from the calibration data for quantitative studies. The static-scene images and dynamic videos respectively contain around 7,344 and 320,667 instances in total. With DarkVision, we established baselines for image/video enhancement and object detection by representative algorithms. To demonstrate an exemplary application of DarkVision, we propose two simple yet effective approaches for improving performance in video enhancement and object detection respectively. We believe DarkVision would advance the state-of-the-arts in both imaging and related computer vision tasks in low-light environment.
|
['Qionghai Dai', 'Jinli Suo', 'Jiayi Xie', 'Zhihong Zhang', 'Runzhao Yang', 'Yuchen Guo', 'Bo Zhang']
|
2023-01-16
| null | null | null | null |
['video-enhancement']
|
['computer-vision']
|
[ 6.01707816e-01 -7.73142636e-01 4.86127436e-02 -4.63757962e-01
-9.73607302e-01 -5.62669337e-01 5.94196856e-01 -1.57894388e-01
-6.71346128e-01 5.18400967e-01 -2.15161629e-02 -1.03114687e-01
1.83472529e-01 -5.74593306e-01 -7.46560633e-01 -1.13520670e+00
2.43811399e-01 -3.25275689e-01 5.57869017e-01 -3.60365883e-02
2.60814428e-01 3.07128429e-01 -1.91207266e+00 3.45220745e-01
6.10998094e-01 1.17540240e+00 5.59007883e-01 9.45200920e-01
2.35731944e-01 7.41299927e-01 -3.43702853e-01 -4.77148414e-01
6.41415238e-01 -1.79214418e-01 -1.12300426e-01 4.15140748e-01
1.13459945e+00 -7.76794255e-01 -5.56257367e-01 1.24643743e+00
7.81684518e-01 8.43139589e-02 4.67146873e-01 -1.29906666e+00
-5.63464820e-01 -2.21723184e-01 -9.24680233e-01 5.98276317e-01
2.09195137e-01 8.77597213e-01 5.34331322e-01 -1.07817471e+00
1.65698618e-01 9.57718730e-01 5.49495935e-01 3.96335185e-01
-9.49056625e-01 -8.91298831e-01 -3.07038158e-01 3.74325216e-01
-1.12324142e+00 -8.76361728e-01 3.15094173e-01 -4.90993083e-01
7.49020815e-01 3.15428227e-02 4.43910033e-01 8.62756610e-01
1.58815131e-01 3.16255361e-01 1.51078784e+00 -1.99535191e-01
-1.22370692e-02 2.31483188e-02 7.80103654e-02 4.52499777e-01
5.99838376e-01 6.37004316e-01 -5.87053001e-01 2.64943331e-01
5.43355525e-01 1.48487881e-01 -5.82111895e-01 1.63241386e-01
-1.17554092e+00 2.76609719e-01 3.74078453e-01 -8.85596946e-02
-1.60178542e-01 1.48826689e-01 1.68428823e-01 1.42746106e-01
2.08544776e-01 5.68618663e-02 -2.74120122e-01 1.75019473e-01
-7.19070673e-01 -2.64095981e-02 1.03934959e-01 1.00483108e+00
7.86867440e-01 8.94618109e-02 -3.88945758e-01 9.63609040e-01
1.71949312e-01 8.86342406e-01 1.23642206e-01 -1.22587252e+00
3.90061766e-01 1.77563220e-01 3.49422216e-01 -8.05319190e-01
-1.93163991e-01 -4.36020255e-01 -9.04305875e-01 5.26265264e-01
4.29415017e-01 1.06186122e-01 -9.93743539e-01 1.35436499e+00
2.95655489e-01 4.03475702e-01 2.45997697e-01 1.39150786e+00
1.06360126e+00 7.63610303e-01 3.47295515e-02 -4.62913096e-01
1.67121446e+00 -9.56911862e-01 -4.91145968e-01 -4.42569971e-01
8.07712451e-02 -9.79818702e-01 1.06191707e+00 5.60912728e-01
-1.23930478e+00 -8.60947728e-01 -1.01784611e+00 -2.02104107e-01
-2.37273052e-02 2.18469396e-01 5.04716814e-01 8.22509050e-01
-1.02159381e+00 7.88012296e-02 -4.04243797e-01 -2.55652189e-01
5.16052127e-01 -2.15833951e-02 -4.16360855e-01 -8.11798096e-01
-9.10702407e-01 8.42284977e-01 3.10440421e-01 1.01243183e-01
-1.56777632e+00 -6.68830514e-01 -6.25404358e-01 -2.58178979e-01
3.84622395e-01 -6.24167621e-01 8.86619389e-01 -6.64382458e-01
-1.08539116e+00 1.18754864e+00 -2.76696216e-02 -4.31869596e-01
2.49682799e-01 4.51426655e-02 -6.34922504e-01 2.58958220e-01
1.23830050e-01 6.04415834e-01 9.47844505e-01 -1.38230073e+00
-1.05099654e+00 -2.58896291e-01 1.82285100e-01 3.37837100e-01
-3.26491028e-01 4.35303837e-01 -9.54043329e-01 -2.30833083e-01
-1.62582681e-01 -5.86990595e-01 -5.25678247e-02 2.95749754e-01
-1.14619479e-01 2.61739016e-01 7.72971928e-01 -4.88843501e-01
6.28358901e-01 -2.36284447e+00 -5.56294918e-01 -6.46484375e-01
1.48860261e-01 4.03988630e-01 -3.95598143e-01 -7.43477717e-02
1.33713931e-01 -3.60657215e-01 -2.63246626e-01 -3.64888370e-01
-5.19969106e-01 2.00753942e-01 -1.31704792e-01 8.47921193e-01
1.69082791e-01 5.85685134e-01 -1.01076376e+00 -4.97061223e-01
7.28141010e-01 4.10922587e-01 -3.91184419e-01 3.38714123e-01
1.50730148e-01 5.64760923e-01 1.73753500e-02 1.09626663e+00
1.03655255e+00 -1.31985052e-02 -3.01513374e-01 -7.60197878e-01
-3.69587988e-01 -2.41567314e-01 -1.12114918e+00 1.33583748e+00
-5.90734959e-01 1.12665176e+00 2.76431441e-01 -5.30790746e-01
6.98266745e-01 -8.95363018e-02 3.00958723e-01 -1.02806520e+00
3.79574820e-02 1.93704814e-01 -8.35914612e-02 -7.60370910e-01
7.40305007e-01 -6.23759301e-03 3.42404723e-01 -5.66163771e-02
-6.49213791e-02 -3.14234704e-01 4.05899316e-01 4.11380194e-02
1.06595194e+00 -1.92044497e-01 1.56284496e-01 9.32910889e-02
5.29933810e-01 -1.05996795e-01 5.69762290e-01 8.98916185e-01
-4.86567765e-01 9.08979356e-01 -2.34220088e-01 -1.66602090e-01
-1.24450064e+00 -1.29679310e+00 -5.56419730e-01 9.42887187e-01
5.53183794e-01 -1.17580533e-01 -3.66085291e-01 7.37069454e-03
-2.52389669e-01 4.88723457e-01 -5.72656095e-03 -3.12451310e-02
-3.30617011e-01 -1.21402884e+00 4.60112125e-01 2.58541912e-01
1.06166792e+00 -6.45842612e-01 -4.95348424e-01 -8.44140574e-02
-2.65089840e-01 -1.72606850e+00 -3.84619057e-01 4.18202169e-02
-4.29560632e-01 -1.42019677e+00 -4.70456749e-01 -6.68667614e-01
5.77603996e-01 1.24669135e+00 1.17685378e+00 -5.56478053e-02
-7.34833539e-01 4.26768303e-01 -2.03122795e-01 -2.39328533e-01
-2.70928770e-01 -8.41939211e-01 2.10322514e-01 2.33260140e-01
1.52391270e-01 -2.43989483e-01 -1.14302397e+00 7.03198612e-01
-1.11169970e+00 -9.86929517e-03 7.78487265e-01 9.62290227e-01
5.54992139e-01 1.67967543e-01 2.49010056e-01 -2.75367945e-01
1.12488143e-01 -2.36922085e-01 -1.00897861e+00 6.01998568e-02
-6.63549542e-01 -6.21344030e-01 4.97380465e-01 -1.74863249e-01
-1.25418341e+00 7.45293275e-02 1.15617514e-01 -4.42761213e-01
-3.43798876e-01 -1.44874722e-01 -2.91268826e-01 -4.42057699e-01
8.51344883e-01 4.23153818e-01 -2.54359454e-01 -1.74875051e-01
3.26756984e-01 1.04076838e+00 1.23453927e+00 -2.46695921e-01
8.17600965e-01 6.71928763e-01 1.41096607e-01 -1.01909792e+00
-8.89529586e-01 -8.94018829e-01 -1.12502113e-01 -5.51374793e-01
9.71099675e-01 -1.67356718e+00 -7.58797824e-01 8.64859939e-01
-1.12120354e+00 -2.76725322e-01 -4.90431488e-02 5.46684802e-01
-1.82866767e-01 5.89045584e-01 -5.63343585e-01 -8.75478983e-01
-1.99777991e-01 -1.37952960e+00 1.29021764e+00 4.82315809e-01
8.08071911e-01 -4.78664964e-01 -3.93367946e-01 8.29256594e-01
4.34793204e-01 -8.13879445e-02 2.97956705e-01 3.45326215e-01
-8.81380558e-01 -7.44361207e-02 -8.42205405e-01 8.61882865e-01
-2.52167825e-02 -1.68038644e-02 -1.49310219e+00 -3.74953300e-01
3.06910239e-02 -3.97379965e-01 1.15499353e+00 5.70608914e-01
1.41999447e+00 1.74887508e-01 -7.06941783e-02 1.06136501e+00
1.59959114e+00 1.14203379e-01 9.62105751e-01 2.34474599e-01
7.29867578e-01 5.21750748e-01 9.13305521e-01 4.15686458e-01
3.43650132e-01 8.89852643e-01 8.01765501e-01 -4.19972599e-01
-7.10991383e-01 3.35952133e-01 6.74640656e-01 3.07276994e-01
-1.35086849e-01 -3.91544372e-01 -6.75559402e-01 4.98903632e-01
-1.26692510e+00 -1.06936252e+00 -7.02445924e-01 2.35476446e+00
7.97932982e-01 -4.70213853e-02 -2.59762794e-01 -4.81086113e-02
9.08634901e-01 1.36445835e-01 -5.83667934e-01 3.06136429e-01
-7.40090609e-01 -1.46290511e-01 1.09541547e+00 2.18588576e-01
-1.16679478e+00 5.69130540e-01 6.00889254e+00 9.18140829e-01
-9.16274428e-01 3.10498118e-01 7.89615750e-01 -3.26797009e-01
1.98150590e-01 -2.73006782e-02 -1.15258515e+00 7.14014769e-01
8.63911808e-01 1.68897793e-01 4.43489671e-01 6.28307164e-01
5.91536582e-01 -5.73838174e-01 -8.54456484e-01 1.60518837e+00
3.85316730e-01 -1.27005386e+00 -2.54038781e-01 -2.00404637e-02
1.06054854e+00 3.66424114e-01 1.67217270e-01 -1.29687697e-01
2.75031012e-02 -7.31179714e-01 5.52008629e-01 4.28696334e-01
1.24865675e+00 -3.09230834e-01 6.14247382e-01 1.85537919e-01
-1.16280460e+00 -3.04286987e-01 -6.27538800e-01 1.79077387e-01
3.19949359e-01 7.89073706e-01 -3.97234559e-01 4.78364229e-01
1.08683050e+00 9.81956959e-01 -7.34673083e-01 1.40149379e+00
-1.15599550e-01 5.73381305e-01 -1.58947140e-01 5.23638189e-01
-2.28705518e-02 -3.79291087e-01 5.38664401e-01 1.29509640e+00
2.34131739e-01 3.73719573e-01 2.68863261e-01 6.38356924e-01
-1.54391885e-01 -5.30992806e-01 -5.49594045e-01 3.84741932e-01
4.70809907e-01 1.70513010e+00 -4.49026942e-01 -3.00640672e-01
-6.60138011e-01 9.23136532e-01 -5.09987235e-01 4.30578947e-01
-9.27340508e-01 -2.21336693e-01 8.85988235e-01 1.76244751e-01
-9.93483588e-02 -3.56891125e-01 -6.57197610e-02 -1.16478252e+00
4.84816544e-02 -8.03957522e-01 1.77830979e-01 -1.29708123e+00
-1.30817640e+00 4.58433002e-01 -1.10563273e-02 -1.35893941e+00
2.71633804e-01 -9.40401554e-01 -5.56992829e-01 7.73811579e-01
-2.08659744e+00 -1.05910718e+00 -1.18674183e+00 7.72583425e-01
7.83493936e-01 -1.07265525e-01 1.68067500e-01 9.25854981e-01
-6.86232269e-01 3.08313757e-01 4.24828023e-01 -4.29790318e-02
1.08524406e+00 -8.31024885e-01 1.67406127e-01 1.39890146e+00
-1.15614355e-01 6.84706271e-02 6.73151374e-01 -2.44324937e-01
-1.74589014e+00 -1.47891557e+00 9.84373540e-02 -4.02588248e-01
4.17775631e-01 -2.79237241e-01 -6.77349806e-01 8.58369023e-02
1.05066784e-01 6.46691024e-01 2.42058322e-01 -5.73236883e-01
-4.24448960e-02 -4.79085624e-01 -1.06217849e+00 4.45696592e-01
1.02315116e+00 -5.46791553e-01 -1.56969614e-02 6.37892127e-01
5.80981731e-01 -3.70937884e-01 -6.15272880e-01 4.20260817e-01
4.08510834e-01 -1.32891393e+00 1.41318810e+00 7.49285817e-02
3.23769242e-01 -8.46968234e-01 -5.33322752e-01 -9.30938959e-01
-2.56044813e-03 -3.38166952e-01 9.97196808e-02 1.40762281e+00
-1.05562404e-01 -5.73967993e-01 4.84402537e-01 2.00780660e-01
-5.41519344e-01 -2.12305069e-01 -7.41569996e-01 -8.53223026e-01
-5.31471252e-01 -6.52640939e-01 2.25265637e-01 6.20852470e-01
-1.00311196e+00 5.04370630e-02 -6.86190784e-01 3.93334508e-01
1.19324827e+00 -5.01406984e-03 1.01403451e+00 -7.79242992e-01
-3.81861210e-01 -1.51945680e-01 -6.31632924e-01 -1.12108362e+00
-1.82338253e-01 -6.02312386e-01 2.76034892e-01 -1.50351977e+00
5.76007366e-01 -2.58906335e-01 -1.33455858e-01 1.15842193e-01
-2.12020740e-01 8.86841416e-01 2.13807207e-02 4.39509004e-01
-7.88070261e-01 4.16534215e-01 1.02013004e+00 -3.33578408e-01
2.28809386e-01 -1.77029163e-01 -5.26051700e-01 6.10575676e-01
5.05923867e-01 -2.51053900e-01 -1.75001919e-01 -5.48090637e-01
-1.25087991e-01 -1.20069847e-01 9.28508341e-01 -1.24203253e+00
4.60881025e-01 -2.79883653e-01 5.77903569e-01 -4.93389189e-01
6.46390319e-01 -7.26382196e-01 1.27992779e-01 3.25786114e-01
-6.86657280e-02 -2.74813384e-01 1.86631884e-02 7.42726326e-01
-2.43466526e-01 -1.28846705e-01 1.31758463e+00 -5.28949574e-02
-1.35108328e+00 4.51541811e-01 -8.38271156e-02 7.34466463e-02
1.08010793e+00 -5.02817571e-01 -1.11879849e+00 -1.17891259e-01
9.63968597e-03 1.46971256e-01 6.50013268e-01 2.49650672e-01
9.52584326e-01 -1.03456950e+00 -9.25679803e-01 2.45153919e-01
5.92824042e-01 -1.21036477e-01 5.35850048e-01 9.98997390e-01
-4.57169324e-01 1.06422797e-01 -4.98344839e-01 -9.27139878e-01
-1.40037739e+00 4.28706467e-01 5.07049561e-01 5.50097466e-01
-4.94959354e-01 6.79770410e-01 4.25243616e-01 3.18036258e-01
1.16624959e-01 -3.25147003e-01 -2.25777645e-02 -2.68156767e-01
9.27315891e-01 6.45829499e-01 1.95932999e-01 -7.27912903e-01
-2.43973657e-01 7.04497695e-01 6.91376328e-02 2.43136033e-01
1.06626403e+00 -5.72851241e-01 4.09932174e-02 9.06608105e-02
9.91246641e-01 -1.89861178e-01 -1.67727375e+00 -1.56232074e-01
-5.30468762e-01 -1.09586704e+00 5.87595344e-01 -7.36978292e-01
-1.18252254e+00 8.86150360e-01 1.12595642e+00 -1.50394648e-01
1.58188939e+00 -6.45433515e-02 6.61433578e-01 1.27222210e-01
5.08031666e-01 -8.82312238e-01 2.45429650e-01 2.32127428e-01
5.48243046e-01 -1.97110128e+00 9.27128270e-02 -3.92905980e-01
-5.59045255e-01 9.81561422e-01 7.68929482e-01 3.51568669e-01
5.86604625e-02 3.88964117e-01 4.56184782e-02 -5.99830225e-02
-7.20321894e-01 -5.55354059e-01 2.10931778e-01 9.85636294e-01
1.94533080e-01 -2.22507372e-01 1.91369563e-01 1.05830297e-01
4.11321878e-01 -2.30397314e-01 8.54677558e-01 4.86630768e-01
-8.62806022e-01 -6.20383978e-01 -6.48561239e-01 4.53947753e-01
-3.82224143e-01 -3.79106581e-01 1.81434453e-01 4.65420872e-01
4.31973130e-01 1.55244339e+00 3.93138789e-02 -2.56221354e-01
2.86660254e-01 -7.21853137e-01 4.62962031e-01 -4.00641948e-01
-3.55728060e-01 -5.69353029e-02 8.50824416e-02 -7.05118656e-01
-5.98247826e-01 -5.90869486e-01 -8.32598567e-01 -2.79687136e-01
-4.11459416e-01 -4.16974604e-01 9.08951283e-01 4.54000920e-01
1.92701146e-01 4.19465065e-01 7.33817875e-01 -1.09854698e+00
-4.04220194e-01 -7.66240597e-01 -7.98248529e-01 4.46516454e-01
4.98071700e-01 -8.06476593e-01 -6.60375834e-01 4.45713878e-01]
|
[10.732760429382324, -2.3827462196350098]
|
ffef19df-03b6-44b9-9670-746c2ce13dbc
|
exploring-topic-metadata-relationships-with-1
| null | null |
https://openreview.net/forum?id=5zmfwLi_mzB
|
https://openreview.net/pdf?id=5zmfwLi_mzB
|
Exploring Topic-Metadata Relationships with the STM: A Bayesian Approach
|
The initial purpose of topic models was to identify latent topical clusters within unstructured text. Meanwhile, the focus of advanced studies has changed primarily to estimating the relationship between the discovered topical structure and theoretically relevant metadata. Methods used to estimate such relationships must take into account that the topical structure is not directly observed, but instead being estimated itself in an unsupervised fashion. In the Structural Topic Model (STM;Roberts et al., 2016), for instance, multiple repeated linear regressions of sampled topic proportions on metadata covariates are performed. This is done by using a Monte Carlo sampling technique known as the \textit{method of composition}. In this paper, we propose two modifications of this approach: First, we implement a substantial correction to the model by replacing linear regression with the more appropriate Beta regression. Second, we provide a fundamental enhancement of the entire estimation framework by substituting the current blending of frequentist and Bayesian methods with a fully Bayesian approach instead. This allows for a more appropriate quantification of uncertainty. We illustrate our improved methodology by investigating relationships between Twitter posts by German parliamentarians and different metadata covariates related to their electoral districts.
|
['Anonymous']
|
2021-11-16
| null | null | null |
acl-arr-november-2021-11
|
['topic-models']
|
['natural-language-processing']
|
[ 4.36955243e-02 2.81625688e-01 -4.32374775e-01 -5.12052953e-01
-8.67840052e-01 -5.00840127e-01 1.33090758e+00 4.54995364e-01
-4.04984593e-01 9.45314050e-01 7.12641537e-01 -2.51871139e-01
-3.23243380e-01 -8.76512647e-01 -7.65496731e-01 -8.04060280e-01
1.72351629e-01 5.87688208e-01 1.65576637e-01 3.62691134e-01
4.69009340e-01 -2.56898031e-02 -1.65800548e+00 -2.16351785e-02
1.09275079e+00 4.74389553e-01 1.16287991e-01 3.06845844e-01
-3.84368241e-01 5.82567990e-01 -4.75300252e-01 -4.59610730e-01
-1.62347704e-01 -1.61998093e-01 -6.56092525e-01 1.40831739e-01
4.74340379e-01 -1.82079807e-01 1.49000674e-01 9.75902975e-01
-5.15696332e-02 4.98016104e-02 1.08837354e+00 -8.25690329e-01
1.87770873e-01 1.19101703e+00 -6.78572118e-01 -3.76195833e-02
2.40983173e-01 -2.94959664e-01 1.05930245e+00 -6.42577469e-01
6.29153073e-01 1.15244853e+00 5.60436845e-01 -2.85135210e-01
-1.63638067e+00 -5.62717676e-01 2.22588778e-01 -4.02132981e-02
-1.53409040e+00 -5.07425249e-01 6.79258525e-01 -8.18531871e-01
3.67742747e-01 3.78283262e-01 5.62405944e-01 1.16436601e+00
1.36898085e-01 5.94446242e-01 1.52768040e+00 -6.22551024e-01
4.43816394e-01 5.90254247e-01 3.84886414e-01 5.81825227e-02
5.76828539e-01 -2.36776635e-01 -4.38492626e-01 -6.59600198e-01
2.20752448e-01 -1.99763447e-01 1.48551511e-02 -4.86846507e-01
-1.16880763e+00 1.12601662e+00 -3.23504955e-01 4.51146781e-01
-4.79821742e-01 9.63314101e-02 2.46268585e-01 -1.37995452e-01
1.22107470e+00 2.96716750e-01 -2.71782011e-01 -2.20798880e-01
-1.64394569e+00 4.39279944e-01 1.15576065e+00 5.45646787e-01
8.97946239e-01 -4.03783411e-01 -1.79865599e-01 6.56882763e-01
7.22497344e-01 2.98354685e-01 3.29757631e-01 -8.28793287e-01
2.54887342e-01 2.55736947e-01 3.09571028e-01 -9.53806937e-01
-1.16855815e-01 -4.25589472e-01 -5.08110404e-01 -3.18186611e-01
6.69690132e-01 -7.61912540e-02 -7.23681152e-01 1.64335084e+00
6.79896414e-01 1.79433465e-01 -4.10988837e-01 2.47186482e-01
2.77893543e-01 6.62380099e-01 4.30137306e-01 -5.10865629e-01
1.53509712e+00 -3.06107223e-01 -1.00831187e+00 2.77182788e-01
3.29152584e-01 -7.32430100e-01 8.46411705e-01 5.56861401e-01
-8.40964139e-01 -7.54064396e-02 -7.70435810e-01 5.37216142e-02
-3.47439826e-01 3.11608966e-02 6.75145268e-01 8.40245545e-01
-9.05760109e-01 5.77465773e-01 -8.72943461e-01 -5.19732296e-01
3.20097283e-02 1.07307017e-01 1.17254771e-01 2.28108749e-01
-1.19520080e+00 7.19066262e-01 3.09926152e-01 -1.37280807e-01
-5.15340149e-01 -7.28995740e-01 -7.08960891e-01 9.27647576e-03
6.20288372e-01 -5.37918687e-01 1.26971376e+00 -5.32155156e-01
-1.59348893e+00 7.32869625e-01 -4.06906933e-01 -4.43746597e-01
6.80595398e-01 -1.75147623e-01 8.30816478e-02 -8.51003751e-02
2.52802342e-01 1.87591448e-01 8.36897314e-01 -1.43527973e+00
-4.25339192e-01 -4.68742400e-01 -2.08611324e-01 8.83731712e-03
-3.72843027e-01 1.99698254e-01 -3.82068932e-01 -6.72160447e-01
2.87210792e-01 -8.02142382e-01 -1.82659686e-01 -6.97364092e-01
-4.76042002e-01 -4.88027155e-01 3.23992878e-01 -7.35544920e-01
1.57126176e+00 -1.91121507e+00 2.50492617e-02 4.93589282e-01
3.02648634e-01 -4.60334450e-01 6.18108034e-01 8.43962014e-01
2.19513729e-01 6.01464212e-02 -3.77761513e-01 -6.21769845e-01
2.59632349e-01 -3.50587592e-02 -6.72653019e-01 8.11548531e-01
-1.75103143e-01 5.07019401e-01 -6.54354274e-01 -7.75608540e-01
1.78900927e-01 4.28211749e-01 -5.94264507e-01 -2.21939355e-01
-4.02051866e-01 4.65880990e-01 -4.19151038e-01 2.73024261e-01
8.16908598e-01 -2.32210994e-01 6.69952035e-01 -2.18192115e-02
-8.52733433e-01 8.08573246e-01 -1.33221138e+00 1.45714903e+00
-2.50939131e-01 6.66760385e-01 1.61398813e-01 -9.73463953e-01
7.58276761e-01 2.64639765e-01 6.95299566e-01 -2.11105347e-02
9.30584073e-02 9.49963108e-02 -3.17376196e-01 -2.08628103e-01
9.70789075e-01 -2.20211744e-01 -1.77486032e-01 7.02546299e-01
-1.73671082e-01 -2.16571197e-01 1.09880500e-01 3.31356138e-01
6.12232447e-01 3.67521882e-01 6.30668700e-01 -8.34062159e-01
1.33423328e-01 8.97303224e-02 3.54926795e-01 1.01969123e+00
2.36818865e-01 3.52779865e-01 7.14471221e-01 7.53482338e-03
-1.10419190e+00 -9.40749705e-01 -9.85219359e-01 8.54413033e-01
-4.41220313e-01 -7.57182956e-01 -8.34929883e-01 -3.23419422e-01
1.15285233e-01 9.80600536e-01 -7.64310181e-01 3.51758271e-01
-3.48095745e-01 -1.18223512e+00 3.47390234e-01 -5.79299107e-02
2.07114115e-01 -5.26790977e-01 -5.10963023e-01 2.44358331e-01
-4.53056157e-01 -7.59583354e-01 4.38985713e-02 1.17983520e-01
-1.11771798e+00 -6.64151311e-01 -5.77695370e-01 6.72372207e-02
4.15532529e-01 -7.98734725e-02 9.35105979e-01 -4.14992601e-01
6.03475906e-02 5.07314861e-01 -3.16640049e-01 -4.04030055e-01
-3.60923171e-01 3.91849428e-01 6.36980534e-02 2.30160698e-01
5.08130789e-01 -7.54798770e-01 -3.88795912e-01 6.38214573e-02
-9.87808526e-01 7.71876145e-03 4.68064219e-01 4.91842687e-01
2.08077252e-01 4.09721434e-02 3.17796767e-01 -1.31560218e+00
5.24484873e-01 -8.81670773e-01 -6.63126826e-01 2.77248900e-02
-8.68014216e-01 6.02221861e-02 -7.33554959e-02 -3.53806108e-01
-1.31599975e+00 -1.91885918e-01 1.57160774e-01 1.57410011e-01
-2.68567979e-01 9.60176468e-01 -6.25366643e-02 5.14476955e-01
3.08935791e-01 1.36819407e-01 -7.28586391e-02 -6.85100198e-01
3.05854857e-01 9.44726467e-01 2.21123487e-01 -8.69303524e-01
7.10489094e-01 6.54662907e-01 -2.61313647e-01 -1.02085090e+00
-8.61872256e-01 -6.47183836e-01 -7.95934975e-01 -3.22763622e-01
6.18972480e-01 -9.06055629e-01 -6.07606888e-01 3.79216701e-01
-1.02556753e+00 -3.50067943e-01 -3.12275529e-01 1.01020694e+00
-4.60155994e-01 6.55632377e-01 -3.12568128e-01 -1.19043338e+00
2.23193526e-01 -9.75005984e-01 1.10730231e+00 -1.72251821e-01
-3.66589844e-01 -1.11387980e+00 4.79027599e-01 3.61645252e-01
3.49910438e-01 1.65533274e-01 8.61153245e-01 -7.29411185e-01
-4.83101368e-01 -7.53191262e-02 -1.11333556e-01 -5.81705198e-02
1.68711945e-01 3.67215693e-01 -1.15159190e+00 -1.29953042e-01
4.15453553e-01 2.96110809e-01 9.40281272e-01 8.83939624e-01
8.46139014e-01 -4.82545167e-01 -6.02853358e-01 3.39912564e-01
1.33540070e+00 -3.07833642e-01 4.67811406e-01 4.67076570e-01
3.53830099e-01 1.07364023e+00 4.76668268e-01 7.10560739e-01
7.56381273e-01 9.96814549e-01 2.62928847e-02 3.73997957e-01
2.85713285e-01 -3.15012276e-01 3.29979211e-01 9.75815654e-01
-2.03068241e-01 1.20285908e-02 -1.09324920e+00 5.78092873e-01
-1.88362908e+00 -9.80430603e-01 -5.07064223e-01 2.70117664e+00
1.20043218e+00 4.75502051e-02 2.53209800e-01 -3.02854870e-02
7.89620876e-01 2.14885816e-01 1.23309030e-03 2.66742110e-02
2.04066172e-01 -1.07192941e-01 6.09875560e-01 7.09323287e-01
-1.20988536e+00 7.00604975e-01 6.53403807e+00 8.73856306e-01
-7.69888401e-01 1.27988741e-01 4.76009518e-01 4.03912514e-02
-5.53049028e-01 5.20894766e-01 -1.07065880e+00 7.59880602e-01
1.24623370e+00 -1.79498076e-01 7.25605264e-02 6.18500412e-01
5.31214476e-01 -8.29396963e-01 -7.81945288e-01 2.27183402e-01
1.06277885e-02 -9.73544240e-01 -1.83648705e-01 7.98250616e-01
7.90324569e-01 -2.18920350e-01 -9.32748616e-02 2.24057764e-01
5.27793944e-01 -6.30343139e-01 8.79175305e-01 8.38233352e-01
5.62852442e-01 -3.83133858e-01 4.86399502e-01 4.02145624e-01
-7.82335997e-01 1.20679639e-01 -2.88095087e-01 -1.24036789e-01
2.71223426e-01 1.23085499e+00 -1.01058125e+00 5.51937819e-01
6.43287182e-01 5.44367135e-01 -3.47350210e-01 1.11392236e+00
-7.20916167e-02 1.16341460e+00 -5.72809219e-01 -3.44346650e-02
9.29819867e-02 -5.16445816e-01 8.31491232e-01 1.21325374e+00
3.42320293e-01 -3.77995312e-01 -7.91449621e-02 9.43937600e-01
2.81220913e-01 3.36873561e-01 -6.00424588e-01 1.31810874e-01
5.51182866e-01 1.24649274e+00 -1.01166904e+00 -4.48322058e-01
-3.13072711e-01 1.61081210e-01 2.10194767e-01 3.16178918e-01
-6.35415435e-01 1.15018390e-01 2.15793818e-01 5.04268944e-01
4.56932902e-01 -4.20214683e-01 -3.70122582e-01 -1.27901411e+00
-1.16282165e-01 -6.92745388e-01 1.98365197e-01 -3.15686971e-01
-1.12712371e+00 -4.86790240e-02 8.64266872e-01 -8.26594532e-01
-3.99492443e-01 -9.27485377e-02 -5.12818813e-01 8.49895775e-01
-1.35654986e+00 -1.08377850e+00 3.46914195e-02 2.34311566e-01
3.28689247e-01 3.38194937e-01 6.20340288e-01 4.13664579e-02
-3.88112634e-01 1.08244620e-01 5.74537694e-01 -4.65197563e-01
8.31269383e-01 -1.51134825e+00 8.69812891e-02 5.69965065e-01
1.09189823e-02 1.01299167e+00 1.18770564e+00 -1.03627968e+00
-9.25995946e-01 -7.88550854e-01 1.22543740e+00 -6.56557739e-01
9.45245385e-01 -6.83152437e-01 -7.17282712e-01 7.81877637e-01
2.86933303e-01 -7.73617268e-01 8.94543052e-01 7.11676359e-01
-1.27022892e-01 2.35628858e-02 -8.55405688e-01 4.29748714e-01
4.06182706e-01 -2.72608340e-01 -6.84358895e-01 2.51646906e-01
5.46397984e-01 1.04297571e-01 -1.12553883e+00 3.13710749e-01
7.14741588e-01 -8.49794984e-01 6.66012645e-01 -2.35042945e-01
2.97505528e-01 -2.11277142e-01 -2.23186791e-01 -1.01886511e+00
-2.83386130e-02 -5.67379951e-01 -1.92798480e-01 1.60202563e+00
2.55687684e-01 -7.52200246e-01 6.09302700e-01 6.75482810e-01
6.52910327e-04 -3.99937123e-01 -1.00220406e+00 -3.47458482e-01
2.30525076e-01 -4.81539696e-01 5.04915655e-01 1.11999655e+00
1.46897480e-01 4.61640880e-02 -4.25006092e-01 2.02160738e-02
9.30339515e-01 5.90331331e-02 9.35025394e-01 -1.79628837e+00
-2.70467997e-01 -2.49472901e-01 1.30733307e-02 -1.01105261e+00
1.05600476e-01 -5.32329679e-01 1.21081084e-01 -1.20312607e+00
6.58333182e-01 -4.80381787e-01 9.08628926e-02 -1.45219892e-01
-8.92040282e-02 -5.81017844e-02 -2.23976642e-01 6.23646736e-01
-3.65735888e-01 5.60609281e-01 6.57754540e-01 1.01599731e-01
-2.55371153e-01 3.52259666e-01 -7.04418480e-01 7.18126118e-01
6.24008477e-01 -7.27770269e-01 -1.88266039e-01 -4.47348021e-02
6.10822439e-01 -1.58529803e-01 5.30599535e-01 -6.70072556e-01
4.23762590e-01 -1.68283135e-01 -3.31178270e-02 -9.64297950e-01
1.84357941e-01 -5.78510404e-01 3.65895420e-01 4.44820039e-02
-3.81632924e-01 -1.85421854e-01 6.56538084e-03 7.85365164e-01
-2.30780154e-01 -5.29004931e-01 1.68202937e-01 -9.53388065e-02
9.24271494e-02 -2.57590204e-01 -7.56124318e-01 -2.91429579e-01
7.13568747e-01 -1.54630437e-01 -1.75222471e-01 -4.58378017e-01
-7.41629004e-01 -1.16405249e-01 4.57977563e-01 -1.50137484e-01
-7.37614930e-02 -9.38327432e-01 -8.30439627e-01 -2.25782067e-01
-1.40073031e-01 -3.21752578e-02 1.17141657e-01 1.43559909e+00
-1.29680429e-02 7.36180663e-01 5.09380758e-01 -6.19224668e-01
-9.13520694e-01 2.16773987e-01 -8.59984308e-02 -5.24035335e-01
-5.63996553e-01 1.47366181e-01 2.40930736e-01 -3.82485598e-01
1.74610481e-01 -4.03244436e-01 -4.74091411e-01 6.27504170e-01
1.25705481e-01 5.45162618e-01 -6.81416839e-02 -4.70150828e-01
-3.97669114e-02 2.66014159e-01 -2.14422688e-01 -5.43414652e-01
1.47322667e+00 -5.02432287e-01 -4.24852550e-01 1.06949580e+00
6.65016234e-01 5.96199632e-01 -1.07412744e+00 -2.83509165e-01
4.29831952e-01 -2.57900059e-01 2.23905414e-01 -4.75826234e-01
-3.36940467e-01 4.32812452e-01 7.15367720e-02 7.69265175e-01
5.32905638e-01 9.58253145e-02 -1.99444741e-02 5.98708875e-02
2.47863591e-01 -1.04099774e+00 -4.65585470e-01 1.36614397e-01
5.47213376e-01 -1.06777859e+00 5.96429586e-01 -4.75186110e-01
-1.60657898e-01 9.23392534e-01 -9.53368843e-03 1.28415599e-01
1.07781613e+00 2.41751615e-02 -3.03300500e-01 -4.48852271e-01
-6.07503057e-01 1.21629853e-02 4.28074628e-01 -2.53167190e-02
7.75508285e-01 1.06444940e-01 -9.72815156e-01 5.48052371e-01
-4.09803927e-01 -1.79982111e-01 6.26164913e-01 6.93198442e-01
-4.71030682e-01 -1.14158368e+00 -5.17335653e-01 6.70831203e-01
-7.37154961e-01 -3.44420522e-01 -8.13324973e-02 9.39495265e-01
-3.37442815e-01 9.33129370e-01 2.57614732e-01 -3.61260958e-02
-2.31247783e-01 3.39464128e-01 1.64145768e-01 -7.59795845e-01
-3.62018228e-01 5.14377594e-01 1.65515691e-01 -1.50259301e-01
-8.14002514e-01 -1.45915806e+00 -3.46107036e-01 -1.30972788e-01
-5.84871709e-01 5.44994712e-01 1.19351208e+00 1.02537346e+00
9.03788134e-02 3.01591963e-01 6.15916729e-01 -9.02950585e-01
-6.96298778e-01 -1.45330572e+00 -7.28909671e-01 -1.38526196e-02
3.52670848e-02 -8.87708724e-01 -6.81103587e-01 4.07169051e-02]
|
[10.334623336791992, 6.998668670654297]
|
4d08612e-4c6c-4693-b55b-a06ed097b129
|
taxonomy-and-evolution-predicting-using-deep
|
2206.14011
| null |
https://arxiv.org/abs/2206.14011v1
|
https://arxiv.org/pdf/2206.14011v1.pdf
|
Taxonomy and evolution predicting using deep learning in images
|
Molecular and morphological characters, as important parts of biological taxonomy, are contradictory but need to be integrated. Organism's image recognition and bioinformatics are emerging and hot problems nowadays but with a gap between them. In this work, a multi-branching recognition framework mediated by genetic information bridges this barrier, which establishes the link between macro-morphology and micro-molecular information of mushrooms. The novel multi-perspective structure is proposed to fuse the feature images from three branching models, which significantly improves the accuracy of recognition by about 10% and up to more than 90%. Further, genetic information is implemented to the mushroom image recognition task by using genetic distance embeddings as the representation space for predicting image distance and species identification. Semantic overfitting of traditional classification tasks and the granularity of fine-grained image recognition are also discussed in depth for the first time. The generalizability of the model was investigated in fine-grained scenarios using zero-shot learning tasks, which could predict the taxonomic and evolutionary information of unseen samples. We presented the first method to map images to DNA, namely used an encoder mapping image to genetic distances, and then decoded DNA through a pre-trained decoder, where the total test accuracy on 37 species for DNA prediction is 87.45%. This study creates a novel recognition framework by systematically studying the mushroom image recognition problem, bridging the gap between macroscopic biological information and microscopic molecular information, which will provide a new reference for intelligent biometrics in the future.
|
['Yihua Yang', 'Jianxin Wang', 'Jing Wang', 'Ming Zhang', 'Wenbin Liao', 'Jiewen Xiao']
|
2022-06-28
| null | null | null | null |
['fine-grained-image-recognition']
|
['computer-vision']
|
[ 6.29075706e-01 -2.66704112e-01 -9.25824940e-02 -1.97986305e-01
-2.46992037e-01 -5.10109305e-01 4.22509134e-01 3.19616884e-01
-4.39121544e-01 6.05271339e-01 -1.52761459e-01 4.25792672e-02
-3.68029356e-01 -1.16374278e+00 -6.23699546e-01 -1.33537471e+00
-5.11889486e-03 2.62351662e-01 1.00332521e-01 1.49453897e-02
5.44447124e-01 5.48624694e-01 -2.06513977e+00 4.10205096e-01
8.15957665e-01 1.12156606e+00 5.12391269e-01 6.94018245e-01
-4.20141399e-01 2.54688829e-01 -6.16408110e-01 -6.78543687e-01
6.82093799e-02 -1.46136358e-01 -5.43238103e-01 1.91758871e-01
2.06698999e-01 -7.04165250e-02 8.29237849e-02 1.05346692e+00
5.97419679e-01 -1.90440521e-01 9.67463732e-01 -8.46957445e-01
-1.56005001e+00 4.94691908e-01 -3.14314544e-01 -1.99836995e-02
1.32494688e-01 2.55519092e-01 8.28159153e-01 -8.37812603e-01
5.88221312e-01 1.14443552e+00 6.61511540e-01 4.30449367e-01
-1.06383574e+00 -4.26834196e-01 -1.17877714e-01 8.34454894e-01
-1.53630149e+00 -3.72120552e-02 4.20043558e-01 -8.60062122e-01
8.65237236e-01 3.35951388e-01 8.18259597e-01 9.65628803e-01
3.21259767e-01 1.83910742e-01 1.16019058e+00 -3.83945465e-01
1.95968881e-01 -3.22592780e-02 2.61170805e-01 7.84090281e-01
5.55210233e-01 1.54220089e-01 -4.95910794e-01 2.31322318e-01
5.23835301e-01 3.69086415e-01 -2.65223235e-01 -1.73249498e-01
-1.05195689e+00 8.31699371e-01 5.14160335e-01 6.55978203e-01
1.60206284e-03 -5.33970118e-01 7.56798163e-02 -1.13587983e-01
2.25593984e-01 5.08932412e-01 -1.22456640e-01 1.41970158e-01
-6.62810445e-01 -4.06922758e-01 7.55995631e-01 3.94174844e-01
7.64461279e-01 1.05496570e-01 -3.39920633e-02 1.08291876e+00
1.32669568e-01 7.17103958e-01 9.17928398e-01 -4.63962555e-01
-1.59783721e-01 9.75042760e-01 -3.52190912e-01 -1.18920398e+00
-4.11936522e-01 -2.87203997e-01 -1.05131984e+00 1.24832131e-01
5.30362964e-01 5.60730934e-01 -8.16651165e-01 1.40431178e+00
4.35133904e-01 3.06503594e-01 2.35510990e-04 9.60858583e-01
8.89078259e-01 7.56842375e-01 -1.60477698e-01 -1.50378838e-01
1.68996751e+00 -6.89545572e-01 -3.97945195e-01 1.21083334e-01
4.45462823e-01 -6.02208912e-01 9.31804836e-01 2.78450489e-01
-4.46707875e-01 -6.44504309e-01 -1.45983970e+00 7.02892914e-02
-9.83261347e-01 3.79432482e-03 6.06607437e-01 8.62370014e-01
-6.71576738e-01 7.20501781e-01 -6.91674054e-01 -6.97656870e-01
3.37953448e-01 3.59443516e-01 -5.42569399e-01 -2.67930329e-01
-8.59899223e-01 9.61141288e-01 6.28954470e-01 1.48431957e-01
-7.09070146e-01 -5.64571261e-01 -7.14968383e-01 1.73055127e-01
-7.46043995e-02 -5.94172895e-01 3.08173180e-01 -6.33609354e-01
-1.62432671e+00 1.13552535e+00 7.18752667e-02 -2.88174450e-01
-1.80184603e-01 2.57708669e-01 -5.14424980e-01 1.79914162e-01
-1.66730121e-01 5.38590789e-01 5.58036327e-01 -1.04690063e+00
-3.26252759e-01 -9.40779150e-01 -4.13992733e-01 -2.44708031e-01
-5.67116618e-01 -3.64057809e-01 -4.85537127e-02 -6.31771207e-01
-9.40637812e-02 -6.13272548e-01 1.35018185e-01 -1.89639740e-02
-9.92937759e-03 2.67160982e-02 5.52424967e-01 -7.51236081e-01
9.53713775e-01 -2.05696130e+00 5.26360214e-01 -7.82065168e-02
1.34584695e-01 4.68987882e-01 -5.03688395e-01 4.05053943e-01
9.88615081e-02 9.15165246e-02 -5.79783142e-01 3.50747168e-01
-3.23126763e-02 2.58206517e-01 -1.03293523e-01 6.62272334e-01
3.85855973e-01 9.52413559e-01 -6.45592451e-01 -2.82604426e-01
4.10762638e-01 6.92764878e-01 -3.21371943e-01 3.39637101e-01
1.25047654e-01 1.67931840e-01 -7.81737044e-02 1.06672108e+00
8.20212543e-01 -2.16912270e-01 1.77550539e-01 -2.48909503e-01
-1.00878030e-01 -5.59817135e-01 -8.40757132e-01 1.66176760e+00
-2.03608349e-01 2.02883318e-01 -9.09693539e-02 -1.30682719e+00
1.39067268e+00 -1.90125868e-01 3.13169062e-01 -7.10612416e-01
2.69616276e-01 1.70731381e-01 1.73854932e-01 -7.38989651e-01
-9.74478200e-03 -3.92014235e-01 8.35368857e-02 -1.16947174e-01
4.32151973e-01 -1.37657955e-01 1.08015798e-01 -5.12943864e-01
6.80042684e-01 1.36029914e-01 5.83064198e-01 -1.57006204e-01
7.68833876e-01 1.42392600e-02 4.98163611e-01 4.91692990e-01
-1.50301307e-01 4.85208780e-01 -1.59658603e-02 -6.54609799e-01
-9.49117720e-01 -9.74308908e-01 -6.38093352e-01 8.98110032e-01
4.28909093e-01 -1.51440591e-01 -9.61029470e-01 -2.25284204e-01
1.24015689e-01 3.93322200e-01 -9.18643415e-01 -5.41094482e-01
2.77190376e-02 -1.30345464e+00 7.09699869e-01 2.62589395e-01
3.73397768e-01 -7.34202266e-01 -4.40649360e-01 1.85833141e-01
9.83955637e-02 -9.63183165e-01 8.59543532e-02 4.43633199e-01
-7.00527608e-01 -1.31874382e+00 -7.41439700e-01 -7.85325766e-01
3.71033698e-01 1.62375927e-01 3.91112983e-01 2.82352101e-02
-1.01592910e+00 -9.83258486e-02 -5.37598491e-01 -1.04396969e-01
-1.59809902e-01 -8.12941119e-02 9.31462646e-02 2.47499019e-01
7.04153478e-01 -5.48918426e-01 -5.62286556e-01 4.56318647e-01
-9.41222847e-01 -1.22265235e-01 7.95196831e-01 1.30925751e+00
7.06748366e-01 -2.22834840e-01 5.24790049e-01 -3.70401055e-01
2.04633102e-01 -5.55688679e-01 -5.42254388e-01 6.68869615e-01
-5.15578032e-01 2.69472420e-01 9.02455151e-01 -4.65267301e-01
-7.57575333e-01 -4.41533513e-02 -3.65046144e-01 -6.64416701e-02
-5.11559963e-01 3.41268539e-01 -4.41387355e-01 -3.31100404e-01
5.99974692e-01 5.89673519e-01 2.03254521e-01 -5.87855041e-01
2.91307926e-01 1.04338443e+00 6.38382018e-01 -5.87903261e-01
3.22517127e-01 4.69878942e-01 1.48816615e-01 -1.15965033e+00
-5.74772358e-01 -4.66613770e-01 -1.10541177e+00 -8.10328051e-02
1.10964668e+00 -5.58084011e-01 -9.57583666e-01 6.37617528e-01
-9.95374501e-01 5.76691218e-02 -6.88587204e-02 5.01114964e-01
-3.77761751e-01 7.40507722e-01 -6.80163264e-01 -6.49492085e-01
-2.81614095e-01 -1.16507685e+00 1.12890208e+00 1.23580806e-01
2.31015861e-01 -7.77724564e-01 1.98103368e-01 5.87135255e-01
1.87776610e-01 2.41162032e-01 1.09137797e+00 -7.04247415e-01
-4.91250843e-01 -1.83115780e-01 -3.04472148e-01 2.44263262e-01
3.90430480e-01 1.78715333e-01 -1.08295929e+00 -1.52544215e-01
5.79712167e-02 -2.85821825e-01 9.41898227e-01 1.51648045e-01
1.26147950e+00 -3.10447775e-02 -2.43962914e-01 1.04990768e+00
1.62320650e+00 5.53822100e-01 7.02691197e-01 5.21733046e-01
5.59373856e-01 7.72535920e-01 3.54488730e-01 5.51828384e-01
2.10438341e-01 6.07702315e-01 5.60035586e-01 2.55252689e-01
-2.75775552e-01 -2.43788108e-01 2.72332996e-01 1.15467215e+00
-3.03656548e-01 -1.38346672e-01 -9.16134179e-01 1.87594399e-01
-1.46693027e+00 -1.14910972e+00 1.31695047e-02 2.23119712e+00
5.88814378e-01 -5.28752208e-01 -1.90729752e-01 1.63408458e-01
8.46074462e-01 -1.84301481e-01 -7.50432372e-01 -5.05902827e-01
-5.82561135e-01 3.49398822e-01 4.12089795e-01 3.26465398e-01
-1.00516307e+00 8.76419008e-01 6.32381201e+00 1.07159364e+00
-1.43707252e+00 -3.63187790e-02 6.09600544e-01 7.50500187e-02
7.94245005e-02 -2.30599508e-01 -8.25903893e-01 7.07354426e-01
9.53595221e-01 -4.09914926e-02 5.25647402e-01 7.35305548e-01
-2.05758125e-01 1.12200096e-01 -9.04262781e-01 1.22903180e+00
5.48229873e-01 -1.48800468e+00 4.51026171e-01 3.40975225e-01
3.17129314e-01 -1.88462690e-01 -5.65671586e-02 1.08424518e-02
6.91460669e-02 -1.41561329e+00 3.31116855e-01 6.30195439e-01
8.51992249e-01 -5.60029030e-01 9.30197179e-01 3.41775358e-01
-1.20322573e+00 -2.22127125e-01 -9.69655991e-01 -1.90593958e-01
-5.89820147e-02 5.78703701e-01 -8.12679768e-01 8.09079468e-01
6.93116963e-01 7.50953376e-01 -8.58874321e-01 1.04328287e+00
1.65682301e-01 3.42104614e-01 -7.56409094e-02 -3.48062694e-01
-5.94768189e-02 -5.92160225e-01 6.73170686e-02 1.24680793e+00
8.45312178e-01 -1.11384038e-02 -2.93392222e-02 8.58545661e-01
2.72735298e-01 3.30563724e-01 -6.71643198e-01 -3.37661386e-01
4.26700711e-01 1.30125189e+00 -8.33618283e-01 -1.98342159e-01
-3.44859570e-01 1.08403838e+00 3.84407222e-01 -8.12177211e-02
-7.79918551e-01 -5.44085562e-01 6.39962852e-01 -2.03185290e-01
4.62562531e-01 -1.04006909e-01 -4.13299441e-01 -1.32474113e+00
-3.41594875e-01 -5.63703954e-01 4.32743996e-01 -5.01326442e-01
-1.55972755e+00 3.87354493e-01 -4.47507977e-01 -9.76747334e-01
8.40319619e-02 -1.25889266e+00 -2.17293173e-01 5.99257946e-01
-1.34675503e+00 -1.24082565e+00 -4.99460280e-01 3.75759333e-01
3.20257068e-01 -3.38675499e-01 1.21794665e+00 1.50618613e-01
-7.74150133e-01 5.93141496e-01 6.01259470e-01 -1.00945286e-01
5.69224775e-01 -8.94483387e-01 -1.04084097e-01 5.29461086e-01
2.32249454e-01 5.26525378e-01 2.01373965e-01 -4.81390238e-01
-1.71758497e+00 -9.41658199e-01 5.75158954e-01 -2.57895410e-01
7.52918661e-01 -4.69371080e-01 -1.07689869e+00 -7.32930750e-02
-7.87057951e-02 -8.67216960e-02 1.40563631e+00 -1.77188694e-01
-6.90966189e-01 -1.70196667e-01 -1.27254725e+00 3.17406923e-01
9.44111705e-01 -4.75361228e-01 -8.01489234e-01 7.43915662e-02
5.65792501e-01 2.78510571e-01 -1.31047213e+00 3.16883236e-01
8.96942019e-01 -7.53195763e-01 1.19827247e+00 -4.80230004e-01
4.63402539e-01 -4.81795073e-01 -7.89645016e-01 -1.17360461e+00
-6.76455438e-01 2.48543024e-01 7.26668313e-02 1.29517543e+00
-3.37806866e-02 -4.96552229e-01 5.66294432e-01 -1.77869797e-01
-1.03546023e-01 -7.02791572e-01 -1.03674245e+00 -9.25664961e-01
2.96818435e-01 6.27262220e-02 8.35703492e-01 9.77929831e-01
2.46877238e-01 2.06192151e-01 -2.37398654e-01 1.45433038e-01
5.98985314e-01 4.12191361e-01 4.44012254e-01 -1.37656188e+00
-3.11731398e-01 -6.60074532e-01 -1.18386507e+00 -5.93062401e-01
1.85094252e-01 -1.31150222e+00 -4.31909002e-02 -1.30668163e+00
4.59266335e-01 8.80672336e-02 -3.17118257e-01 2.46070534e-01
-1.52025372e-01 4.65859264e-01 9.47726890e-02 1.89687833e-01
-9.83359409e-04 6.32678926e-01 1.12789726e+00 -4.88326460e-01
3.59344393e-01 -5.81000328e-01 -6.55654788e-01 4.94593143e-01
7.76912868e-01 -6.71331212e-02 1.80823300e-02 -3.34573209e-01
-2.54117548e-01 -1.79714888e-01 3.60088617e-01 -1.14584458e+00
2.64155388e-01 -2.97366619e-01 6.02608502e-01 -1.70399964e-01
4.73114729e-01 -7.79572666e-01 4.06125903e-01 7.71711886e-01
4.37695970e-04 -4.23320472e-01 -1.25830844e-02 5.61000705e-01
-3.46395731e-01 -3.72142971e-01 9.58474696e-01 -9.33505818e-02
-1.00087607e+00 3.02885205e-01 -3.90653163e-01 -4.97707546e-01
1.27387941e+00 -8.34519386e-01 -5.04582047e-01 5.08003891e-01
-6.61411881e-01 -3.91965181e-01 6.57200158e-01 3.08449894e-01
7.28048742e-01 -1.09828627e+00 -6.08720899e-01 4.23920870e-01
4.41135854e-01 -5.25767744e-01 5.41958869e-01 6.67457521e-01
-6.23196423e-01 5.31413198e-01 -7.69875288e-01 -8.01832318e-01
-1.22524309e+00 1.08140385e+00 1.61152497e-01 1.92594796e-01
-2.81412303e-01 7.95476615e-01 2.68526405e-01 -5.22823751e-01
1.52213588e-01 -1.06779248e-01 -4.07078892e-01 3.41157943e-01
5.91247976e-01 2.46125549e-01 1.81749821e-01 -8.98132145e-01
-3.85071486e-01 1.23021615e+00 2.07754046e-01 4.62154716e-01
1.54817724e+00 -2.52707936e-02 -4.51322496e-01 5.83073616e-01
1.18434286e+00 -4.42543298e-01 -9.46668863e-01 3.66543606e-02
7.77431577e-02 -6.87007070e-01 -3.49080503e-01 -8.88429046e-01
-8.62999916e-01 1.34162343e+00 9.11558747e-01 5.63723519e-02
9.83135998e-01 -4.05327566e-02 5.46233833e-01 4.48884815e-01
5.85753441e-01 -8.93042445e-01 1.00374641e-02 2.74725020e-01
7.16011643e-01 -1.17793274e+00 -2.98860610e-01 -4.19427663e-01
-2.43014321e-01 1.55967343e+00 5.22194266e-01 1.21242449e-01
3.73308629e-01 4.05910611e-01 -3.48565459e-01 -5.43387383e-02
-5.30874133e-01 -3.64447504e-01 1.54577032e-01 1.12116647e+00
5.09821534e-01 2.99374998e-01 -4.22642410e-01 1.01989746e+00
-2.79024214e-01 7.84475803e-02 1.09508604e-01 6.28046870e-01
-9.09630179e-01 -1.21822059e+00 -5.81391573e-01 3.07684600e-01
-2.70797104e-01 1.27826445e-02 -6.71661198e-01 3.85420501e-01
5.47240436e-01 8.44111443e-01 1.57104537e-01 -8.51310015e-01
1.33783519e-02 2.14415878e-01 6.40890002e-01 -4.04334933e-01
-3.07319254e-01 -2.37224236e-01 -2.97037870e-01 -2.10797191e-01
-3.49970341e-01 -3.99291039e-01 -9.07904088e-01 -4.40838337e-01
-3.39777291e-01 6.49360344e-02 7.92613149e-01 8.26622903e-01
4.76408482e-01 3.82618695e-01 5.21729827e-01 -8.28625441e-01
-4.45290297e-01 -8.40564072e-01 -7.35002935e-01 3.16082239e-01
-1.92436218e-01 -7.56404579e-01 -3.41352582e-01 2.83998430e-01]
|
[9.546222686767578, 2.17950701713562]
|
ce408803-d54e-4f02-9f59-c56b8952b85e
|
label-semantics-for-few-shot-named-entity
|
2203.08985
| null |
https://arxiv.org/abs/2203.08985v1
|
https://arxiv.org/pdf/2203.08985v1.pdf
|
Label Semantics for Few Shot Named Entity Recognition
|
We study the problem of few shot learning for named entity recognition. Specifically, we leverage the semantic information in the names of the labels as a way of giving the model additional signal and enriched priors. We propose a neural architecture that consists of two BERT encoders, one to encode the document and its tokens and another one to encode each of the labels in natural language format. Our model learns to match the representations of named entities computed by the first encoder with label representations computed by the second encoder. The label semantics signal is shown to support improved state-of-the-art results in multiple few shot NER benchmarks and on-par performance in standard benchmarks. Our model is especially effective in low resource settings.
|
['Dan Roth', 'Yaser Al-Onaizan', 'Sunil Mallya', 'Rishita Anubhai', 'Srikanth Doss', 'Miguel Ballesteros', 'Jie Ma']
|
2022-03-16
| null |
https://aclanthology.org/2022.findings-acl.155
|
https://aclanthology.org/2022.findings-acl.155.pdf
|
findings-acl-2022-5
|
['few-shot-ner']
|
['natural-language-processing']
|
[ 1.42371789e-01 3.35379690e-01 -3.70709211e-01 -6.57486200e-01
-9.54657555e-01 -5.59908867e-01 8.83826137e-01 2.07125321e-01
-7.39927351e-01 6.92535996e-01 8.08363676e-01 2.88491517e-01
2.64259934e-01 -9.27706957e-01 -6.81493282e-01 -3.91677052e-01
5.96975256e-03 6.99947596e-01 3.11201513e-01 -5.96627370e-02
4.11001630e-02 2.15903237e-01 -1.26923573e+00 4.08524871e-01
3.06053162e-01 9.69932675e-01 2.23035008e-01 4.41145360e-01
-6.53671503e-01 1.35492671e+00 -3.60874832e-01 -7.00777769e-01
1.99307323e-01 -2.35184878e-01 -1.09963000e+00 -2.98872534e-02
2.75266618e-01 -3.84794325e-01 -6.93389773e-01 1.10143590e+00
4.90557700e-01 6.16848588e-01 7.07566202e-01 -7.68794656e-01
-8.67970526e-01 9.63796258e-01 -2.37384602e-01 2.92778403e-01
4.91920672e-02 8.79626162e-03 1.61084843e+00 -9.55345571e-01
1.15521371e+00 1.00473642e+00 4.30666238e-01 7.75693536e-01
-1.08971798e+00 -3.71199727e-01 4.48072106e-02 2.55796671e-01
-1.25566244e+00 -8.15585077e-01 2.74627417e-01 -4.56898123e-01
1.30671203e+00 -5.12459636e-01 1.13845959e-01 1.20669770e+00
-3.85994226e-01 7.45210052e-01 5.36613762e-01 -4.15487885e-01
6.08368456e-01 9.22144875e-02 7.75104403e-01 7.82434940e-01
1.06599040e-01 8.74837562e-02 -5.25681078e-01 -1.40002906e-01
2.84241259e-01 1.61362454e-01 5.91970086e-02 -2.58977622e-01
-8.53414178e-01 1.05797267e+00 4.89974588e-01 4.30002183e-01
-6.29376888e-01 5.42009532e-01 5.45186043e-01 -4.59859557e-02
4.56889093e-01 7.35443413e-01 -5.54358482e-01 -2.65550107e-01
-8.38025153e-01 -2.51382202e-01 9.91707087e-01 1.24739933e+00
8.56387258e-01 -5.58797121e-02 -6.97906792e-01 9.69451368e-01
1.94252700e-01 2.01550722e-01 4.90571350e-01 -8.71894360e-01
4.58515167e-01 3.17012519e-01 3.26669395e-01 -3.45171422e-01
-2.99093723e-01 -1.27685949e-01 -3.39619637e-01 -3.98026645e-01
1.31661803e-01 -3.16090286e-01 -1.38262999e+00 1.93073452e+00
8.93829092e-02 7.85752714e-01 1.45023853e-01 4.99658138e-01
8.81760895e-01 9.49905634e-01 4.91832137e-01 1.19849272e-01
1.69733930e+00 -1.13789475e+00 -7.45871127e-01 -5.30057251e-01
9.04117405e-01 -1.73512340e-01 5.52117229e-01 -4.55558896e-01
-9.03448641e-01 -3.80742460e-01 -8.78215075e-01 -2.91048229e-01
-6.65930510e-01 4.86569144e-02 4.06228989e-01 4.52527225e-01
-7.78926551e-01 7.91706979e-01 -9.36751544e-01 -4.38231617e-01
6.30786777e-01 -2.15007458e-02 -3.47834736e-01 -3.15630138e-01
-1.59234452e+00 1.09285831e+00 6.09693408e-01 -4.44802076e-01
-1.18191159e+00 -7.78431475e-01 -1.34123850e+00 6.56564534e-01
4.22803521e-01 -4.48613673e-01 1.58614755e+00 -3.19616228e-01
-1.23791265e+00 1.02360642e+00 -1.95357829e-01 -7.12966144e-01
-1.02475613e-01 -1.67433158e-01 -5.03959775e-01 1.84835956e-01
4.84440804e-01 7.44700909e-01 3.86560261e-01 -9.46978271e-01
-6.89520299e-01 -5.98893315e-02 7.36966282e-02 -1.36578873e-01
-3.37722927e-01 2.91716188e-01 -5.20051122e-01 -4.55933690e-01
-4.49092567e-01 -5.82070470e-01 -2.85471588e-01 -3.18288028e-01
-4.53205436e-01 -3.54832500e-01 2.49198750e-01 -5.23435056e-01
9.55124021e-01 -2.31751609e+00 -4.37761843e-02 -9.60276425e-02
9.55641717e-02 1.82419211e-01 -3.54233682e-01 5.69300115e-01
-2.19827611e-02 1.52368471e-01 -2.50102997e-01 -6.08247042e-01
4.23946142e-01 3.59380066e-01 -6.51998401e-01 4.41574484e-01
5.12496829e-01 1.10343874e+00 -1.18216753e+00 -2.61614919e-01
-8.64211842e-02 4.27301466e-01 -2.82052964e-01 6.01099968e-01
-2.03544572e-01 -3.21229361e-02 -3.27032924e-01 3.28851610e-01
1.62221640e-01 -6.20879054e-01 6.22050464e-02 -1.11341171e-01
4.62096483e-02 7.80931413e-01 -1.16092706e+00 1.89384627e+00
-4.25601453e-01 5.45129240e-01 -2.93136030e-01 -7.37950385e-01
8.84381056e-01 7.30800867e-01 2.54769981e-01 -6.42231584e-01
2.27811173e-01 7.04902187e-02 -4.80170012e-01 -4.54076469e-01
3.97886574e-01 -6.43659890e-01 -3.46178681e-01 7.24549174e-01
8.97052050e-01 2.22246066e-01 5.27903616e-01 3.99637938e-01
1.44521189e+00 -1.34221628e-01 4.82109278e-01 1.57742292e-01
-1.57450035e-03 -2.38753527e-01 6.60968781e-01 1.05579615e+00
-2.13361949e-01 3.99420291e-01 4.15398836e-01 -4.75471020e-01
-1.27510190e+00 -9.89325941e-01 -1.04255751e-02 1.65992737e+00
-8.68299138e-03 -5.17624736e-01 -4.10759538e-01 -8.17889094e-01
-3.84636559e-02 1.27136612e+00 -9.01695132e-01 -3.78135204e-01
-3.84926677e-01 -2.14463294e-01 8.42274785e-01 1.03078282e+00
2.85461038e-01 -1.32894444e+00 -6.26753926e-01 3.02118450e-01
6.25172034e-02 -1.35825729e+00 -5.56019902e-01 7.98357010e-01
-4.29610342e-01 -7.94326901e-01 -7.58587956e-01 -8.27197671e-01
3.62649828e-01 -7.16769099e-02 1.22173250e+00 -3.18587363e-01
-2.83827245e-01 3.30265909e-01 -5.24887323e-01 -3.25173885e-01
-1.82880193e-01 2.56199598e-01 -1.66300222e-01 1.36906683e-01
6.16346359e-01 -2.93291777e-01 -1.28638536e-01 -7.55091757e-02
-9.96171951e-01 -3.52940291e-01 3.73432010e-01 1.00178361e+00
2.93176174e-01 -2.80555129e-01 4.02082771e-01 -1.34647536e+00
1.66802645e-01 -9.24009502e-01 -3.60805035e-01 3.81536841e-01
-2.04336911e-01 6.00594163e-01 7.07114100e-01 -1.70711040e-01
-1.46095634e+00 1.64948344e-01 -4.87502664e-02 -2.94433355e-01
-5.04187703e-01 4.53314453e-01 -4.58132327e-02 6.06003881e-01
6.84751928e-01 -4.65670228e-02 -7.91059077e-01 -8.73971343e-01
8.97419333e-01 6.34404540e-01 7.23197818e-01 -6.13693118e-01
5.17249763e-01 3.69943768e-01 -4.04213279e-01 -6.89605057e-01
-1.59644806e+00 -9.29535866e-01 -7.58053422e-01 2.54424334e-01
1.16810179e+00 -1.05057347e+00 4.82854694e-02 1.67672768e-01
-1.42887580e+00 -9.11511928e-02 -8.40256512e-01 4.96765345e-01
-5.36734879e-01 -5.56194074e-02 -1.13200676e+00 -5.72550058e-01
-3.01166743e-01 -8.59443367e-01 1.13552272e+00 4.10853326e-01
-7.27556646e-02 -9.74762619e-01 5.71074128e-01 -9.21011269e-02
3.49639028e-01 -2.17156678e-01 1.04812562e+00 -1.55654335e+00
-1.87939271e-01 -3.66527498e-01 -4.46082711e-01 1.05087787e-01
-3.94681692e-02 -5.11822760e-01 -1.39134371e+00 -1.46235883e-01
-1.79562569e-01 -6.01179957e-01 1.30563796e+00 1.62243307e-01
5.32886982e-01 -1.59528658e-01 -4.05119300e-01 6.02744997e-01
1.49832284e+00 7.75508210e-02 4.83849108e-01 1.77833349e-01
6.64510310e-01 5.20348966e-01 1.87181875e-01 6.79664016e-01
1.83017120e-01 4.69171554e-01 8.85315761e-02 2.05267355e-01
-1.91260159e-01 -4.70504612e-01 2.24336118e-01 6.66586518e-01
3.30628455e-01 -1.97681218e-01 -1.07302260e+00 8.15424442e-01
-1.81435001e+00 -1.30346942e+00 4.72667217e-01 1.74136603e+00
8.98477793e-01 3.57089639e-02 -8.55093971e-02 -6.35277510e-01
1.12840462e+00 6.04575455e-01 -5.76132476e-01 -2.17714369e-01
2.01079533e-01 4.95126963e-01 5.16382992e-01 2.64058709e-01
-1.33764160e+00 1.19953716e+00 6.53511047e+00 4.34806436e-01
-5.00843167e-01 4.76574481e-01 2.59561509e-01 -1.23350583e-01
-1.31799802e-01 2.67322153e-01 -1.06552982e+00 3.77071261e-01
1.57903171e+00 -3.70190829e-01 2.00195581e-01 1.10756004e+00
-3.88333559e-01 2.95623243e-01 -1.38171530e+00 5.40220976e-01
6.93743676e-02 -1.55045617e+00 2.72944253e-02 -6.83651194e-02
9.43699360e-01 5.03584743e-01 -4.88621980e-01 7.17963517e-01
8.91817629e-01 -8.37516487e-01 7.00848758e-01 6.95640266e-01
8.61057639e-01 -6.21287942e-01 9.95528102e-01 2.26706058e-01
-1.22540033e+00 -2.29279935e-01 -7.65946090e-01 -7.79481512e-03
5.63886881e-01 3.35874200e-01 -9.79438365e-01 1.95527703e-01
2.61845589e-01 9.15749431e-01 -2.95144886e-01 1.25957215e+00
-5.28440356e-01 7.57341921e-01 -1.14147350e-01 -7.06688389e-02
5.88845253e-01 2.84979552e-01 2.93707669e-01 1.67535019e+00
6.26229271e-02 2.94001937e-01 2.67324895e-01 8.99188101e-01
-7.14179575e-01 1.22943290e-01 -8.80229950e-01 -5.29389024e-01
8.94793093e-01 1.10184538e+00 -6.57720029e-01 -9.50411081e-01
-6.10641479e-01 9.29731667e-01 8.55226994e-01 3.51589948e-01
-6.56602561e-01 -8.18380892e-01 6.77916944e-01 -2.95987904e-01
9.28095222e-01 7.83713236e-02 -6.35082051e-02 -1.38085890e+00
-4.92849052e-01 -2.29096264e-01 7.51841843e-01 -7.09133625e-01
-1.58023691e+00 4.26860392e-01 -1.67407990e-01 -6.75889373e-01
-3.80420536e-01 -6.24747574e-01 -6.87610745e-01 7.31077909e-01
-1.54177964e+00 -8.25931191e-01 -4.63595726e-02 2.89444447e-01
6.36146665e-01 -2.00413838e-01 1.22145402e+00 3.78321826e-01
-6.32794261e-01 3.56591016e-01 3.77873540e-01 7.84512997e-01
7.51191556e-01 -1.37099791e+00 9.03443694e-01 8.78531873e-01
7.20419347e-01 6.26842558e-01 3.80538940e-01 -5.59997976e-01
-9.54275429e-01 -1.10921180e+00 1.19989920e+00 -5.14148355e-01
7.56657124e-01 -3.52026045e-01 -8.80464375e-01 1.04990268e+00
2.28685677e-01 2.68453300e-01 1.01257598e+00 2.60077000e-01
-1.00253296e+00 3.43243778e-01 -9.23739135e-01 1.69906959e-01
8.20398569e-01 -9.62302625e-01 -1.29014003e+00 1.76030740e-01
1.08238590e+00 -1.20384969e-01 -7.50933886e-01 -1.56004995e-01
3.22996974e-01 -3.13361466e-01 9.16933060e-01 -1.32386529e+00
5.77322960e-01 7.70288035e-02 -5.49932122e-01 -1.41198158e+00
-6.21623158e-01 -1.32633746e-01 -3.17816943e-01 1.44506240e+00
5.66675544e-01 -1.58744484e-01 4.77795780e-01 9.35854733e-01
-1.44255251e-01 -2.87525326e-01 -9.08199668e-01 -8.27755809e-01
-1.68962240e-01 -3.99915576e-01 5.94349802e-01 1.16123700e+00
2.27167323e-01 8.48322272e-01 -4.81571317e-01 1.91841591e-02
5.54243922e-01 -2.85308901e-02 1.06115676e-01 -1.29580665e+00
-2.95373082e-01 -2.38459408e-01 -4.38685447e-01 -9.19993758e-01
6.48411989e-01 -1.24788702e+00 5.77390432e-01 -1.68767512e+00
5.16450882e-01 -3.24587747e-02 -7.61778474e-01 1.00120306e+00
-2.56212682e-01 4.95157056e-02 3.57050300e-01 3.05108149e-02
-1.01376796e+00 5.68821728e-01 4.09506798e-01 -3.13990802e-01
2.73768473e-02 -3.53557408e-01 -7.20185697e-01 5.58533132e-01
3.88087273e-01 -9.56146538e-01 -5.12101240e-02 -5.97922087e-01
1.06717765e-01 -1.11376025e-01 -1.65695380e-02 -9.38715756e-01
5.38105011e-01 -3.16136442e-02 2.36719668e-01 -3.92843336e-02
3.24419349e-01 -6.48164868e-01 -3.71248603e-01 1.81958720e-01
-1.02844775e+00 -4.24679250e-01 -2.55006701e-01 7.81434596e-01
-1.25739187e-01 -8.30366254e-01 9.31769669e-01 -3.89732510e-01
-1.25362206e+00 3.70870739e-01 -1.12211965e-01 7.33859062e-01
7.67533183e-01 3.99851203e-01 -5.08182585e-01 -3.03262472e-01
-8.47950578e-01 2.21812308e-01 3.06057930e-01 3.31068993e-01
2.81596184e-01 -1.51397943e+00 -6.18691683e-01 -5.91038167e-02
5.16223967e-01 -3.60878021e-01 4.68526892e-02 3.02815765e-01
-5.53547479e-02 4.71009165e-01 -1.98795959e-01 4.75443341e-02
-5.23857117e-01 6.76818848e-01 2.38864303e-01 -4.37877715e-01
-8.17514241e-01 1.09498918e+00 1.23451047e-01 -3.93107623e-01
4.12788779e-01 -1.02802301e-02 -2.08865955e-01 5.15602231e-01
9.45737839e-01 4.61855382e-01 -7.13996142e-02 -6.85035646e-01
-3.09408635e-01 1.30025536e-01 -3.40406269e-01 -3.42405736e-01
1.75639081e+00 6.63801953e-02 1.56371146e-01 8.06029558e-01
1.53374124e+00 -3.03308189e-01 -1.22624373e+00 -7.41513014e-01
5.94181538e-01 -2.47312948e-01 3.58452760e-02 -7.09981561e-01
-7.27684438e-01 9.88786638e-01 1.41524330e-01 -1.97768025e-02
4.83883023e-01 3.72019172e-01 9.00748968e-01 9.34997678e-01
3.89395207e-01 -1.26801121e+00 6.37686700e-02 1.03689706e+00
5.70144132e-02 -1.17389643e+00 -4.52348560e-01 1.89402059e-01
-9.43043113e-01 8.70381832e-01 2.88017452e-01 -1.69548512e-01
5.72520792e-01 3.66113752e-01 -1.17681772e-01 -5.26784658e-01
-1.04939830e+00 -6.49141371e-01 2.04396740e-01 4.28839564e-01
3.93297672e-01 -4.89439853e-02 7.50724450e-02 9.29590762e-01
3.20850521e-01 -4.22629826e-02 4.06532317e-01 1.02221167e+00
-8.85247350e-01 -7.91742384e-01 1.58571914e-01 6.09539866e-01
-4.71334308e-01 -4.64033097e-01 -2.16953173e-01 2.68027067e-01
-4.45834100e-02 8.30694914e-01 2.91110724e-01 -1.02296419e-01
4.61003363e-01 9.55893695e-01 -2.57047080e-02 -1.34142268e+00
-5.23617148e-01 -3.46698582e-01 4.63237226e-01 -6.83061838e-01
-1.72063246e-01 -3.40910017e-01 -1.55204904e+00 1.58266380e-01
-2.60354072e-01 4.40664321e-01 4.84634995e-01 1.19106007e+00
4.77430433e-01 4.05983716e-01 5.65871179e-01 -6.55264974e-01
-8.09530854e-01 -1.01149118e+00 -1.02607691e+00 7.77717471e-01
1.40785843e-01 -7.57408440e-01 -3.05544406e-01 1.51119366e-01]
|
[9.693373680114746, 9.329890251159668]
|
56f003a3-c666-4fef-b8e0-e8ca4cc0ee99
|
graph-structure-learning-from-unlabeled-data
|
1701.0147
| null |
http://arxiv.org/abs/1701.01470v1
|
http://arxiv.org/pdf/1701.01470v1.pdf
|
Graph Structure Learning from Unlabeled Data for Event Detection
|
Processes such as disease propagation and information diffusion often spread
over some latent network structure which must be learned from observation.
Given a set of unlabeled training examples representing occurrences of an event
type of interest (e.g., a disease outbreak), our goal is to learn a graph
structure that can be used to accurately detect future events of that type.
Motivated by new theoretical results on the consistency of constrained and
unconstrained subset scans, we propose a novel framework for learning graph
structure from unlabeled data by comparing the most anomalous subsets detected
with and without the graph constraints. Our framework uses the mean normalized
log-likelihood ratio score to measure the quality of a graph structure, and
efficiently searches for the highest-scoring graph structure. Using simulated
disease outbreaks injected into real-world Emergency Department data from
Allegheny County, we show that our method learns a structure similar to the
true underlying graph, but enables faster and more accurate detection.
|
['Sriram Somanchi', 'Daniel B. Neill']
|
2017-01-05
| null | null | null | null |
['graph-structure-learning']
|
['graphs']
|
[ 4.67005312e-01 5.09270966e-01 -2.94279695e-01 -2.61922091e-01
-3.15386087e-01 -5.74639857e-01 4.12480950e-01 8.41659963e-01
-4.01750579e-02 5.81922889e-01 -9.45156533e-03 -6.19309068e-01
-4.33252692e-01 -9.72718775e-01 -5.96619844e-01 -5.94064295e-01
-1.14602470e+00 9.91100550e-01 4.05003726e-01 4.23571706e-01
-7.19743744e-02 6.10094249e-01 -3.99606228e-01 3.35281566e-02
5.78572214e-01 3.24327916e-01 -1.65396795e-01 8.83328021e-01
1.67536199e-01 8.17495465e-01 -6.63466156e-01 -1.34666681e-01
2.17624307e-01 -7.05947638e-01 -6.47507548e-01 6.30255222e-01
8.62885546e-03 -1.62842616e-01 -3.43410403e-01 1.12917650e+00
7.43137747e-02 -8.48012641e-02 1.12693393e+00 -1.37209797e+00
-5.98269701e-02 5.25005102e-01 -6.19054139e-01 7.81296670e-01
3.06712508e-01 1.49826705e-01 9.89204109e-01 -2.11537093e-01
1.23122084e+00 1.16729009e+00 6.71619892e-01 2.72764295e-01
-1.56868505e+00 -5.52120090e-01 3.01847577e-01 -3.42957795e-01
-1.25342119e+00 -5.14986813e-02 6.33231163e-01 -6.40419900e-01
7.20910668e-01 6.33817241e-02 4.92461324e-01 9.84195590e-01
3.03162783e-01 3.48069280e-01 7.04935253e-01 -1.36275813e-01
4.67356175e-01 -6.27607778e-02 2.82402754e-01 1.30212057e+00
8.69923830e-01 1.26321569e-01 -2.39305571e-01 -1.07113636e+00
7.73016274e-01 3.45340699e-01 -3.88374627e-01 -5.48320651e-01
-1.11659980e+00 1.08700550e+00 2.77233034e-01 4.64473478e-02
-5.01693130e-01 -1.06470101e-01 7.81091079e-02 3.87519926e-01
7.84350812e-01 3.40061098e-01 -3.94965827e-01 4.42645252e-01
-8.54346216e-01 -1.82979688e-01 1.14546490e+00 5.87285161e-01
7.71617770e-01 -1.86843291e-01 -3.39950137e-02 1.86799213e-01
4.47732538e-01 7.17578948e-01 -2.96985120e-01 -4.47539777e-01
3.75393510e-01 8.91197383e-01 -6.05036542e-02 -1.03619778e+00
-6.31124675e-01 -4.70955938e-01 -8.50963533e-01 -4.83634397e-02
5.19026816e-01 -6.27785742e-01 -9.40338314e-01 1.91754913e+00
5.40487587e-01 7.23692179e-01 -3.41486596e-02 4.89565462e-01
1.45270765e-01 6.29076421e-01 -8.52392539e-02 -7.18467176e-01
9.47513819e-01 -4.10552531e-01 -3.87551516e-01 -2.54369199e-01
8.27738523e-01 -2.29384750e-01 1.97387576e-01 3.43203135e-02
-6.65813148e-01 1.79805070e-01 -6.07087016e-01 9.50177848e-01
-6.50959760e-02 -5.44265032e-01 3.71379048e-01 4.38849777e-01
-9.32626903e-01 4.11332279e-01 -1.17306435e+00 -8.69135678e-01
2.45083123e-01 1.10844411e-01 -2.92010635e-01 -1.64108112e-01
-9.10356104e-01 3.84535491e-01 4.95459110e-01 -3.07388186e-01
-1.16246355e+00 -2.45160535e-01 -6.48678005e-01 7.10005686e-02
8.38450849e-01 -6.19667351e-01 8.88544381e-01 -6.64586306e-01
-4.18280840e-01 6.83053493e-01 -2.99543917e-01 -7.67809093e-01
2.76041001e-01 2.59062618e-01 -6.16304874e-01 5.76435089e-01
2.21330509e-01 -1.58482775e-01 1.03467691e+00 -1.20794809e+00
-5.30782521e-01 -5.29973269e-01 -2.31155947e-01 -3.82330343e-02
-1.47385404e-01 -9.46915988e-03 -1.31316006e-01 -4.97763813e-01
2.58975089e-01 -7.80405402e-01 -6.22656405e-01 -1.41097158e-01
-8.46068680e-01 -4.88855466e-02 8.98980379e-01 -5.91288269e-01
1.15275192e+00 -1.80988288e+00 -1.05955303e-01 1.00582385e+00
7.53243864e-01 -1.56341463e-01 -1.80944130e-01 7.62845755e-01
1.43244818e-01 2.08325028e-01 -5.53477049e-01 2.37134155e-02
-5.56204975e-01 3.55290234e-01 -2.81954288e-01 8.94930780e-01
3.82749826e-01 5.01772344e-01 -1.33334875e+00 -3.92337441e-01
-2.91766673e-01 6.55005202e-02 -3.80941570e-01 3.48271221e-01
-3.45879138e-01 6.27843678e-01 -7.21552908e-01 4.84112680e-01
2.73142308e-01 -1.01809359e+00 7.63031483e-01 4.19553101e-01
5.38082361e-01 -4.42660190e-02 -1.15880489e+00 6.78370476e-01
2.03468904e-01 4.06851023e-01 3.84583831e-01 -1.05080330e+00
7.75689840e-01 3.01787078e-01 5.94584346e-01 1.05517142e-01
-1.44158646e-01 -1.53314635e-01 4.29339521e-02 -4.14900154e-01
-2.88759589e-01 2.29097437e-02 8.68885070e-02 1.16328561e+00
-3.21919359e-02 3.06460828e-01 2.67584682e-01 8.46855700e-01
1.86869621e+00 -8.59823883e-01 5.33995569e-01 -1.32301450e-01
-2.87524164e-02 3.33627015e-01 5.16326308e-01 1.15129697e+00
-5.88817373e-02 1.03501767e-01 9.19020951e-01 -4.26882178e-01
-9.25348699e-01 -1.45766830e+00 1.12668030e-01 5.88570714e-01
1.13814324e-02 -2.66281158e-01 -5.66953540e-01 -1.04510212e+00
3.94351073e-02 2.76498407e-01 -5.32843649e-01 -1.62312955e-01
-4.88841504e-01 -1.14154708e+00 3.43513250e-01 -3.12593244e-02
6.23339377e-02 -8.32930088e-01 -3.31447035e-01 4.00562376e-01
2.88661588e-02 -1.07240224e+00 -5.28138459e-01 3.13493535e-02
-1.24986053e+00 -1.88910437e+00 -2.52799869e-01 -7.41051674e-01
1.33606076e+00 6.59328401e-02 1.16442227e+00 2.68401355e-01
-5.93111217e-01 7.45579243e-01 -2.63202310e-01 -1.11105658e-01
-8.75698268e-01 -5.28751388e-02 1.51415795e-01 2.30573356e-01
3.20759505e-01 -2.44945347e-01 -4.13163543e-01 1.10124126e-01
-9.88096714e-01 -5.98958671e-01 4.72653568e-01 6.90372527e-01
4.39613163e-01 4.78905231e-01 5.21624267e-01 -1.28652537e+00
9.54127133e-01 -9.89032507e-01 -7.86991119e-01 4.89791960e-01
-7.94723690e-01 1.36484072e-01 3.15661967e-01 -4.54108745e-01
-7.39714265e-01 9.40555483e-02 6.31190598e-01 -1.74386665e-01
-3.27668726e-01 7.96392262e-01 4.76051539e-01 1.38954252e-01
7.75347412e-01 8.22356865e-02 2.74060685e-02 -2.19824091e-01
1.08773112e-01 2.97129273e-01 2.15745851e-01 -4.82194237e-02
9.07391787e-01 7.37283468e-01 2.70939797e-01 -1.13216102e+00
-8.02147806e-01 -7.74105489e-01 -4.81193006e-01 -2.11901322e-01
6.49607897e-01 -5.71589947e-01 -4.50754523e-01 1.89746469e-01
-1.06742513e+00 -4.55586910e-01 -4.91545871e-02 6.37713611e-01
-2.32244320e-02 5.60859621e-01 -7.55814016e-01 -1.12003124e+00
-1.33974433e-01 -3.32874149e-01 8.64104748e-01 -3.06848139e-01
-3.61646771e-01 -1.60847330e+00 7.16715217e-01 -1.95040151e-01
-2.65041050e-02 6.21114433e-01 1.24097359e+00 -1.18551600e+00
-6.91726625e-01 -4.19153035e-01 -1.67579055e-01 -2.30167285e-01
5.44581473e-01 1.76347762e-01 -5.00607550e-01 -7.68700659e-01
1.09968752e-01 1.16827525e-01 8.84285748e-01 6.62088037e-01
3.78269166e-01 -7.89869130e-01 -1.11857176e+00 2.81134188e-01
1.17092776e+00 1.85758188e-01 -8.01369920e-02 -2.75526255e-01
4.83926237e-01 6.18285418e-01 2.11653054e-01 4.58375186e-01
-3.75445634e-02 5.51614054e-02 2.43184552e-01 -2.97571808e-01
3.58535349e-01 -4.18669790e-01 3.12759101e-01 5.19564629e-01
2.26276875e-01 -7.24445105e-01 -1.20911813e+00 5.14309525e-01
-1.78931320e+00 -9.90452349e-01 -6.62209243e-02 2.39922500e+00
4.34916735e-01 5.00363886e-01 3.38726074e-01 -1.65439755e-01
1.18795347e+00 1.18946590e-01 -8.24905813e-01 1.59890339e-01
-1.50564164e-02 -1.15554415e-01 7.06891358e-01 8.19584668e-01
-9.78502035e-01 4.20817316e-01 6.93402529e+00 2.51914561e-01
-8.10907125e-01 -3.61295231e-02 6.46071613e-01 2.21404612e-01
-5.21893948e-02 1.13200042e-02 -4.87278491e-01 1.15693808e-01
1.10509503e+00 -3.97197217e-01 3.69515657e-01 4.70718086e-01
2.80057937e-01 -2.37890147e-02 -9.86528039e-01 3.91121268e-01
-1.34790242e-01 -1.30792594e+00 -4.52664904e-02 3.24919879e-01
8.74345303e-01 3.70120853e-01 -3.38479459e-01 -2.52271682e-01
9.92447972e-01 -6.64595962e-01 -9.59519222e-02 2.99141824e-01
5.78668892e-01 -5.35145164e-01 5.26071191e-01 7.50738382e-01
-1.21329391e+00 9.47410893e-03 -4.26946860e-03 1.35273531e-01
3.99501741e-01 1.00356770e+00 -1.77053702e+00 1.95083097e-01
2.51145273e-01 7.83015132e-01 -3.04796159e-01 1.05880404e+00
-3.22137237e-01 1.11870658e+00 -5.32112718e-01 5.57496361e-02
1.56639874e-01 -1.22832634e-01 1.08113861e+00 1.05921543e+00
1.11815602e-01 2.17548385e-01 7.44677246e-01 7.45807886e-01
-3.51108573e-02 -1.23297431e-01 -1.22857070e+00 -5.07499754e-01
4.83836234e-01 9.99195099e-01 -1.39422643e+00 -5.15056789e-01
-2.27596864e-01 7.84858346e-01 3.00472140e-01 6.41721845e-01
-3.78749013e-01 3.63309048e-02 2.60800660e-01 4.17253554e-01
1.68461472e-01 -3.13485324e-01 4.42618459e-01 -1.07870817e+00
-3.16376239e-01 -5.31354964e-01 9.56797481e-01 -1.70844749e-01
-1.68752408e+00 7.25150287e-01 7.39428252e-02 -9.62970734e-01
-5.19955575e-01 -3.07191491e-01 -9.39604998e-01 4.88585919e-01
-1.07595086e+00 -4.08246517e-01 -3.58852521e-02 6.53506458e-01
2.20876291e-01 3.58747728e-02 8.02448809e-01 -3.06779861e-01
-4.82498050e-01 -1.10185347e-01 2.62267351e-01 4.29414004e-01
2.51621217e-01 -1.11714613e+00 6.98610723e-01 9.84563410e-01
3.73976171e-01 4.64152396e-01 7.31627762e-01 -1.49507320e+00
-9.81251299e-01 -1.45025563e+00 6.84814692e-01 -3.92444700e-01
9.71521437e-01 -3.83677006e-01 -1.02770102e+00 9.93741810e-01
-1.54444292e-01 -1.73038784e-02 4.50707018e-01 -9.28881019e-03
-2.41558507e-01 4.37590241e-01 -1.37488389e+00 3.78843427e-01
1.15177846e+00 -4.06657159e-01 -4.57864493e-01 8.82477045e-01
5.04125476e-01 1.30497307e-01 -4.51675951e-01 2.82910228e-01
-1.48618877e-01 -5.73234260e-01 5.49880266e-01 -1.15027547e+00
-2.93348312e-01 -7.21868351e-02 3.68695647e-01 -1.37352538e+00
-3.72766495e-01 -7.85686076e-01 -3.48908454e-01 5.17756104e-01
6.58353329e-01 -9.70812678e-01 9.44438517e-01 2.23838329e-01
7.01659560e-01 -4.13716912e-01 -7.43402183e-01 -1.05764389e+00
-6.89033210e-01 3.59406509e-02 1.21258140e-01 1.22022092e+00
-1.30901104e-02 6.72147691e-01 -2.41282597e-01 1.04241395e+00
1.17997420e+00 1.35586694e-01 5.17765999e-01 -1.72757435e+00
-4.18895960e-01 1.32536054e-01 -4.50835228e-01 -6.94296658e-01
-3.80636305e-02 -8.78142238e-01 -1.28159866e-01 -1.55343616e+00
4.78650063e-01 -2.84953743e-01 -8.74417871e-02 2.83187628e-01
-1.15989275e-01 -2.32380494e-01 -3.98786962e-01 4.46847111e-01
-6.38240099e-01 -6.08648509e-02 7.48863757e-01 -1.66912973e-01
-3.46789718e-01 3.51866841e-01 -1.04527086e-01 9.05208766e-01
9.03725445e-01 -9.65254009e-01 -6.18609607e-01 2.01330297e-02
1.82165161e-01 5.66039801e-01 3.99493724e-01 -4.61671561e-01
3.00447613e-01 -2.02988043e-01 1.06035307e-01 -4.83915657e-01
-4.08208035e-02 -8.36661696e-01 2.44587958e-01 1.11250961e+00
-4.45968062e-01 1.91829473e-01 -1.92112148e-01 1.39830029e+00
2.79411059e-02 -5.94067015e-02 4.78542924e-01 -2.15219527e-01
-2.73231417e-01 6.76785946e-01 -5.67459345e-01 4.24285144e-01
1.18015552e+00 1.78801239e-01 -5.66174030e-01 -8.37526798e-01
-1.05313671e+00 4.06585693e-01 5.28653502e-01 2.13274569e-03
7.45765507e-01 -6.76715732e-01 -8.75695407e-01 2.78393269e-01
6.73173219e-02 -1.88032806e-01 -2.59452254e-01 7.76586235e-01
-3.49093556e-01 2.50147432e-02 3.03081214e-01 -7.11239815e-01
-1.30023241e+00 6.80315673e-01 2.69141763e-01 -7.44832158e-01
-8.21892679e-01 6.02125347e-01 2.31910974e-01 -1.64790154e-01
1.62827536e-01 -1.49526864e-01 1.23542108e-01 -2.21933976e-01
5.12751877e-01 3.89867991e-01 -2.91741580e-01 -3.05006564e-01
-4.94659096e-01 1.78186864e-01 -8.99244696e-02 -1.02471642e-01
1.22835910e+00 1.65704247e-02 -1.42565846e-01 2.27669582e-01
9.44770038e-01 7.98505098e-02 -1.21162176e+00 -5.34682512e-01
5.48065782e-01 -2.70501494e-01 -2.74827629e-01 -4.05090898e-01
-1.11032605e+00 4.08569574e-01 4.05916214e-01 1.00822461e+00
7.52102494e-01 4.73760337e-01 3.95492852e-01 6.89822257e-01
4.46655929e-01 -4.95093316e-01 1.08201958e-01 1.17871970e-01
3.08472008e-01 -1.14330196e+00 -1.23944087e-02 -6.31882966e-01
-3.81739557e-01 8.90422761e-01 6.12503663e-02 -3.20430040e-01
9.17366326e-01 3.51450026e-01 -1.53826252e-01 -9.85900283e-01
-8.66314709e-01 -5.01801148e-02 6.36574998e-02 7.12149322e-01
-1.71945497e-01 1.80048272e-01 1.56258717e-01 -2.29999557e-01
5.49160540e-01 -1.56456932e-01 4.90761340e-01 9.35410976e-01
-7.48092234e-01 -6.50863111e-01 -4.09059048e-01 1.06924295e+00
-4.84520525e-01 -1.27124891e-01 -7.83275068e-01 5.55762589e-01
-4.18547511e-01 9.92600739e-01 1.84486702e-01 -5.47173284e-02
1.16358988e-01 -5.01071103e-04 2.25215435e-01 -1.04267168e+00
-1.18957832e-01 1.26780748e-01 2.33157352e-01 -3.56200606e-01
-3.75076473e-01 -6.22366309e-01 -1.14084125e+00 -1.20793335e-01
-4.67160523e-01 4.24311787e-01 6.02066405e-02 8.15487027e-01
3.47551167e-01 2.32303873e-01 7.74136662e-01 -1.37433350e-01
-3.54142964e-01 -7.18156815e-01 -9.67448771e-01 4.93304700e-01
6.54718280e-01 -4.35563445e-01 -7.96576083e-01 1.08098894e-01]
|
[6.792013645172119, 5.185769081115723]
|
4bb8ada2-7027-43e5-a663-7ec09a6a0a27
|
safe-reinforcement-learning-for-probabilistic
|
2002.10126
| null |
https://arxiv.org/abs/2002.10126v1
|
https://arxiv.org/pdf/2002.10126v1.pdf
|
Safe reinforcement learning for probabilistic reachability and safety specifications: A Lyapunov-based approach
|
Emerging applications in robotics and autonomous systems, such as autonomous driving and robotic surgery, often involve critical safety constraints that must be satisfied even when information about system models is limited. In this regard, we propose a model-free safety specification method that learns the maximal probability of safe operation by carefully combining probabilistic reachability analysis and safe reinforcement learning (RL). Our approach constructs a Lyapunov function with respect to a safe policy to restrain each policy improvement stage. As a result, it yields a sequence of safe policies that determine the range of safe operation, called the safe set, which monotonically expands and gradually converges. We also develop an efficient safe exploration scheme that accelerates the process of identifying the safety of unexamined states. Exploiting the Lyapunov shielding, our method regulates the exploratory policy to avoid dangerous states with high confidence. To handle high-dimensional systems, we further extend our approach to deep RL by introducing a Lagrangian relaxation technique to establish a tractable actor-critic algorithm. The empirical performance of our method is demonstrated through continuous control benchmark problems, such as a reaching task on a planar robot arm.
|
['Subin Huh', 'Insoon Yang']
|
2020-02-24
| null | null | null | null |
['safe-exploration']
|
['robots']
|
[ 4.00826670e-02 4.68487680e-01 -6.43137038e-01 4.62053828e-02
-1.01714933e+00 -6.54462337e-01 4.20923740e-01 1.02612182e-01
-5.14678836e-01 8.96955729e-01 -5.62512241e-02 -6.66989326e-01
-4.50872213e-01 -5.56419015e-01 -9.06177700e-01 -9.87009943e-01
-2.05030456e-01 1.80908054e-01 1.78064689e-01 -2.61247069e-01
2.46300712e-01 3.91961753e-01 -1.12295341e+00 -5.61315477e-01
1.02918482e+00 8.70517135e-01 1.12451881e-01 2.81231999e-01
6.21795952e-01 7.24547625e-01 -3.75925787e-02 4.58134413e-01
3.64517838e-01 -1.40548453e-01 -7.57437646e-01 3.80388014e-02
-4.46890950e-01 -3.97320271e-01 -2.28518203e-01 1.28351927e+00
1.49469659e-01 4.57006484e-01 3.29326212e-01 -1.41707492e+00
5.49153127e-02 7.64864683e-01 -6.33998513e-01 -1.41937017e-01
5.54217584e-02 5.91505289e-01 8.15445781e-01 -3.33187521e-01
3.19638908e-01 1.23714876e+00 1.82265401e-01 5.88159382e-01
-1.26590967e+00 -6.38482392e-01 5.68463087e-01 -2.15904877e-01
-1.39234686e+00 -1.85037181e-01 7.74622142e-01 -3.36843401e-01
6.12810493e-01 -7.59668499e-02 6.43411517e-01 1.02075815e+00
7.71636903e-01 6.59844697e-01 8.67431641e-01 -2.63791978e-01
6.15301788e-01 -2.88166493e-01 -6.60599470e-02 8.52273166e-01
2.51028329e-01 6.40668988e-01 4.85812277e-02 -4.11382705e-01
5.12733757e-01 -3.76562625e-02 -1.72671601e-01 -7.10575104e-01
-1.19237149e+00 9.72510993e-01 2.75824845e-01 -2.58650720e-01
-3.65987629e-01 3.90424699e-01 3.93272161e-01 1.75730497e-01
-4.36894186e-02 6.65230811e-01 -3.44835788e-01 -1.07608564e-01
-3.43835801e-01 3.78443748e-01 7.93627679e-01 7.80964673e-01
6.36861265e-01 1.85956255e-01 -2.25736365e-01 3.77139062e-01
4.28781420e-01 3.21733296e-01 -1.28192708e-01 -1.34764659e+00
2.58626074e-01 5.55959702e-01 5.05470872e-01 -5.61482370e-01
-3.08115840e-01 -3.64268541e-01 -5.81595063e-01 7.09392488e-01
1.86494902e-01 -3.66491914e-01 -6.99439466e-01 1.99675190e+00
7.07722902e-01 2.62763321e-01 2.97931254e-01 8.95211220e-01
-4.85595018e-01 7.60619998e-01 4.16331878e-03 -4.60052907e-01
1.00935245e+00 -6.13233387e-01 -5.67101657e-01 -2.92864293e-01
6.25512540e-01 4.16164473e-02 1.05542850e+00 6.96408212e-01
-9.78516877e-01 5.17500564e-03 -1.14051592e+00 4.70809430e-01
2.69878834e-01 -1.04344897e-01 3.58184516e-01 1.05755217e-01
-7.48246133e-01 6.15647316e-01 -1.27858210e+00 2.06245352e-02
3.04812402e-01 6.45238161e-01 -8.26655626e-02 1.76409081e-01
-1.09348559e+00 1.11805010e+00 8.91782820e-01 1.91542879e-01
-1.84739184e+00 -4.99532431e-01 -1.03954279e+00 -1.52749911e-01
1.28492939e+00 -3.17941695e-01 1.36030805e+00 -3.59722346e-01
-1.87841976e+00 1.09504804e-01 2.41371378e-01 -7.02778876e-01
3.99816483e-01 -4.63289171e-01 5.44363409e-02 1.00241311e-01
2.57869381e-02 4.64992076e-01 9.97375190e-01 -1.19052911e+00
-6.26127064e-01 -8.66957679e-02 3.55372190e-01 1.68771729e-01
-2.00545296e-01 -3.63926977e-01 -1.96544960e-01 -1.98857605e-01
-2.22260833e-01 -1.36204338e+00 -9.75016594e-01 4.91212755e-02
-5.22904217e-01 -3.07411253e-01 5.44492662e-01 -1.54819667e-01
1.22565043e+00 -2.07433915e+00 6.02577627e-01 5.55478632e-01
1.10604405e-01 5.50588593e-02 -2.55522542e-02 2.79736519e-01
3.15896839e-01 -1.55781522e-01 -4.26399678e-01 1.64790586e-01
-9.27657858e-02 4.70386267e-01 -7.49719083e-01 8.26954365e-01
3.16070348e-01 6.07688367e-01 -1.13423145e+00 -4.48357314e-01
3.29149276e-01 1.19963646e-01 -6.87955081e-01 1.46047875e-01
-5.41670859e-01 7.81653404e-01 -1.12968874e+00 2.56235331e-01
2.48224750e-01 4.66682613e-02 2.51982361e-01 3.47278178e-01
-3.86773437e-01 -2.90843751e-02 -1.11683667e+00 1.49574685e+00
-6.42064333e-01 -9.36790481e-02 3.30943406e-01 -9.21712339e-01
8.93480659e-01 5.66767007e-02 5.64579308e-01 -3.03093672e-01
4.98665959e-01 3.63079943e-02 -1.12578519e-01 -1.71942070e-01
2.81577289e-01 -1.22285776e-01 -4.97966826e-01 3.31886351e-01
-3.67588758e-01 -2.94472814e-01 -2.04643637e-01 5.80759309e-02
1.12151122e+00 3.49015057e-01 4.51925069e-01 -5.83372056e-01
8.10083330e-01 2.06213593e-01 8.24063361e-01 6.54842198e-01
-3.81107807e-01 -3.19436103e-01 1.02851224e+00 -6.68516308e-02
-8.14317763e-01 -9.40279305e-01 1.01845339e-01 6.56158566e-01
5.31456590e-01 -2.44663164e-01 -6.59164429e-01 -7.82358706e-01
4.54048105e-02 9.46820319e-01 -7.07345843e-01 -8.67914796e-01
-7.23709524e-01 -1.88641399e-01 1.87387496e-01 3.78671825e-01
-7.20831901e-02 -8.44088078e-01 -1.03552985e+00 2.21102789e-01
2.95231968e-01 -8.67179751e-01 -5.95467985e-01 4.71426100e-01
-7.30551660e-01 -1.05730712e+00 -1.13630697e-01 -7.18563557e-01
7.74938345e-01 -2.27674186e-01 3.83111775e-01 -1.34614393e-01
3.35929357e-02 2.14978307e-01 7.90813938e-02 -1.83938459e-01
-7.51821756e-01 -3.94466519e-03 3.87180239e-01 -2.42040336e-01
-4.11376774e-01 -2.77778417e-01 -5.38992047e-01 4.01727766e-01
-6.10676289e-01 -9.49208066e-02 4.82088149e-01 8.68159890e-01
9.86055434e-01 2.51975298e-01 6.67198122e-01 -4.14364070e-01
6.94120049e-01 -6.14212990e-01 -1.45672834e+00 2.12471738e-01
-7.18162000e-01 7.86228299e-01 1.04058290e+00 -7.13152409e-01
-8.63262653e-01 5.44340789e-01 1.59119993e-01 -7.00914145e-01
2.45855853e-01 3.75602871e-01 -1.81512311e-01 -4.84625958e-02
5.11334479e-01 1.23283066e-01 4.46531236e-01 -5.08663766e-02
4.03802663e-01 2.29257509e-01 6.00078762e-01 -1.06509984e+00
9.27598715e-01 3.41911256e-01 3.16994727e-01 -3.53169799e-01
-9.03453708e-01 -2.59117663e-01 -1.06554307e-01 -3.35705370e-01
3.73147994e-01 -7.90537179e-01 -1.56358278e+00 -9.70962867e-02
-7.42033243e-01 -5.54516852e-01 -3.47621769e-01 3.62918466e-01
-9.39722121e-01 3.62412602e-01 -3.49501938e-01 -1.36905408e+00
-1.17346451e-01 -1.43313146e+00 1.03242564e+00 1.32029518e-01
-1.68321759e-01 -5.87612927e-01 3.40129793e-01 -2.62556911e-01
-1.27186533e-02 6.29545331e-01 7.86565065e-01 -2.17950329e-01
-4.72044706e-01 -4.77568284e-02 4.37132329e-01 1.41212031e-01
-5.19774482e-03 -1.08810104e-01 -3.03642869e-01 -6.30256772e-01
1.22562185e-01 -5.23981273e-01 6.74664021e-01 3.84230286e-01
1.24687314e+00 -7.49107420e-01 -6.10860944e-01 4.18784946e-01
1.18057752e+00 5.00487804e-01 2.67856956e-01 4.19548512e-01
4.56855923e-01 6.90498888e-01 1.17962146e+00 6.26965880e-01
2.02579945e-01 4.87842083e-01 9.27019060e-01 4.64716822e-01
7.43289888e-01 -4.87929195e-01 6.09254539e-01 1.15977935e-01
3.12779576e-01 7.32219070e-02 -8.63553405e-01 4.99499023e-01
-2.07729888e+00 -5.30807674e-01 5.07335067e-01 2.55070090e+00
1.14982343e+00 4.78821427e-01 2.15866849e-01 7.48529807e-02
5.67008197e-01 -9.21497717e-02 -1.12043202e+00 -3.71061981e-01
6.44922137e-01 -1.06925979e-01 7.16857314e-01 8.21377277e-01
-1.06304514e+00 9.77893472e-01 5.74609041e+00 8.83825421e-01
-1.03788877e+00 -2.99009919e-01 5.02501428e-01 -1.83835812e-02
-2.41653055e-01 2.25757465e-01 -8.09638023e-01 3.59233379e-01
1.04434431e+00 -3.88241112e-01 7.38048613e-01 1.21491301e+00
5.35275757e-01 -2.90362686e-01 -9.73388314e-01 3.90008062e-01
-6.70784533e-01 -1.03293288e+00 -3.93415213e-01 1.54544652e-01
5.31030416e-01 -1.63920254e-01 1.19544461e-01 2.50018597e-01
8.51174951e-01 -8.06946397e-01 9.12216485e-01 2.97738522e-01
6.35682046e-01 -1.30907118e+00 9.37058553e-02 7.30481923e-01
-9.63184357e-01 -5.60719430e-01 -1.09917127e-01 6.17077649e-02
3.91770303e-01 2.56957501e-01 -8.01270783e-01 3.06368113e-01
2.54489720e-01 6.96125448e-01 4.15452532e-02 7.60972202e-01
-5.61061740e-01 4.05762911e-01 -4.55150515e-01 -2.85831869e-01
4.91748244e-01 -3.03656191e-01 8.05204630e-01 5.07380843e-01
6.51618838e-02 1.29573748e-01 5.63343942e-01 9.00533617e-01
4.42490906e-01 -3.91218811e-01 -8.07242095e-01 -2.64664995e-03
4.75106180e-01 1.10753679e+00 -7.63295114e-01 1.59342304e-01
1.53297469e-01 3.39782029e-01 5.00775695e-01 2.62956351e-01
-1.12529790e+00 -2.31372446e-01 7.91065276e-01 -3.34835291e-01
1.31409094e-01 -5.49950123e-01 -1.14952818e-01 -8.03636312e-01
-1.19748063e-01 -7.34664083e-01 3.51932049e-01 -2.18994901e-01
-7.82188416e-01 4.98647243e-01 2.64415920e-01 -1.33821607e+00
-5.86362362e-01 -3.58934492e-01 -6.02242768e-01 5.07449567e-01
-1.35885727e+00 -5.56981087e-01 1.87756121e-01 6.70682371e-01
3.90910536e-01 1.35459632e-01 4.00833279e-01 -3.53692740e-01
-9.25048411e-01 2.78511643e-01 -1.13679850e-02 -3.09118688e-01
3.57671022e-01 -9.48151886e-01 2.48586573e-02 9.94393587e-01
-6.16043806e-01 5.62412918e-01 9.13587630e-01 -7.88163662e-01
-1.78761780e+00 -1.23370111e+00 -1.03162467e-01 -2.23607048e-01
1.02086616e+00 -2.14894697e-01 -8.77742946e-01 6.72210872e-01
-2.26529479e-01 1.82198305e-02 -1.96412414e-01 -2.21906662e-01
-8.55880007e-02 -1.66768178e-01 -1.04430282e+00 1.04365432e+00
7.20123112e-01 1.01606604e-02 -4.70363706e-01 1.89407423e-01
1.10287118e+00 -4.94435668e-01 -6.87854707e-01 5.60470760e-01
3.24578166e-01 -2.00665027e-01 8.23379695e-01 -7.02800691e-01
9.98465419e-02 -4.73168671e-01 1.52696148e-01 -1.21510041e+00
-1.45733431e-01 -1.29717958e+00 -6.14072025e-01 6.19912982e-01
3.04917067e-01 -7.07751930e-01 5.56477308e-01 5.13864875e-01
-3.71427834e-01 -1.25043309e+00 -1.23869765e+00 -1.09468400e+00
2.56253392e-01 -3.83795410e-01 3.37844998e-01 3.82412851e-01
3.67464513e-01 -2.85534039e-02 -5.38605809e-01 5.06445169e-01
7.22726762e-01 1.07441880e-02 3.72351825e-01 -7.86401272e-01
-3.01936328e-01 -3.15090030e-01 1.10810071e-01 -8.55220318e-01
8.26441526e-01 -4.74466532e-01 9.07235861e-01 -1.04096234e+00
1.86634436e-02 -5.64846456e-01 -4.44384396e-01 6.41638458e-01
-5.51133789e-02 -5.44745982e-01 -1.33293599e-01 6.54200688e-02
-8.40772986e-01 9.08545971e-01 1.30617535e+00 -5.36839850e-02
-5.70545375e-01 1.60659134e-01 -7.18653500e-01 6.99470162e-01
9.22933698e-01 -4.80240703e-01 -6.59740806e-01 1.48771033e-02
2.03939945e-01 5.27952969e-01 2.67947674e-01 -7.37015903e-01
6.04918115e-02 -8.57823610e-01 -3.74866754e-01 -3.84033799e-01
1.54118106e-01 -8.81105661e-01 -5.38478494e-02 1.13522029e+00
-7.31029093e-01 -1.91810578e-01 2.75707304e-01 1.04975712e+00
1.94994122e-01 -7.39234239e-02 1.09146428e+00 3.47137660e-01
-6.64306521e-01 5.15049636e-01 -5.72191715e-01 2.03687102e-02
1.58153820e+00 2.33396441e-01 1.32401297e-02 -2.25493625e-01
-5.30604899e-01 1.03852296e+00 3.78883094e-01 4.88157958e-01
6.85734987e-01 -1.05492556e+00 -2.78821588e-01 1.46001101e-01
9.56705809e-02 1.85954183e-01 -4.21188101e-02 8.64093423e-01
-1.45699427e-01 3.11739266e-01 -3.98023911e-02 -5.93429446e-01
-6.68113172e-01 1.00156450e+00 3.78695160e-01 -2.42305100e-01
-8.46932650e-01 3.34880203e-01 2.58289427e-01 -1.05043553e-01
3.50255519e-01 -6.09878659e-01 -1.48193285e-01 -5.55517554e-01
3.96463364e-01 3.42604280e-01 -3.68160516e-01 -2.19571084e-01
-5.88065386e-01 3.47411305e-01 -4.16123979e-02 -3.48716587e-01
1.12314796e+00 -8.80988017e-02 1.48148254e-01 2.11755663e-01
9.24902201e-01 -2.79379576e-01 -2.04733610e+00 1.07898064e-01
2.60969788e-01 -2.22688079e-01 1.97084084e-01 -5.00609994e-01
-8.25527549e-01 3.67848486e-01 2.22722441e-01 -1.50141045e-02
1.09412444e+00 5.79896830e-02 6.89095080e-01 6.68057919e-01
7.00817287e-01 -1.02394617e+00 5.14153056e-02 6.91761971e-01
8.11661065e-01 -1.10264575e+00 -2.08423898e-01 -1.52344927e-01
-8.36808383e-01 9.44371521e-01 8.19483876e-01 -4.04746234e-01
5.31070411e-01 6.08957112e-01 -4.95816171e-01 1.97832078e-01
-9.85051572e-01 1.67220347e-02 3.13692056e-02 2.75136411e-01
-3.61759245e-01 3.10161896e-02 -2.84792632e-01 4.72574800e-01
1.97916850e-01 -8.62546116e-02 5.50449610e-01 1.24240208e+00
-6.61593914e-01 -9.00703549e-01 -3.24509710e-01 1.14415638e-01
-2.30221629e-01 3.20283651e-01 1.40376285e-01 8.31483305e-01
-4.16473657e-01 8.33294928e-01 -1.79107174e-01 -3.37783396e-01
2.09859997e-01 -2.30397031e-01 3.69010717e-01 -7.18085766e-01
-1.23703003e-01 1.22976758e-01 -5.53901903e-02 -9.52231050e-01
4.21028659e-02 -5.84461272e-01 -1.67479050e+00 -2.82703061e-02
-4.73740250e-01 3.55578125e-01 3.74599099e-01 1.03122389e+00
3.26486081e-01 4.93444622e-01 9.30693984e-01 -6.07841194e-01
-1.24480677e+00 -3.04752976e-01 -3.74490649e-01 -3.53665233e-01
8.17691863e-01 -9.41450775e-01 -3.06285232e-01 -3.72856110e-01]
|
[4.631158828735352, 2.214484930038452]
|
d6a27429-09ed-4793-b919-6797bd599376
|
feature-normalisation-for-robust-speech
|
1507.04019
| null |
http://arxiv.org/abs/1507.04019v1
|
http://arxiv.org/pdf/1507.04019v1.pdf
|
Feature Normalisation for Robust Speech Recognition
|
Speech recognition system performance degrades in noisy environments. If the
acoustic models are built using features of clean utterances, the features of a
noisy test utterance would be acoustically mismatched with the trained model.
This gives poor likelihoods and poor recognition accuracy. Model adaptation and
feature normalisation are two broad areas that address this problem. While the
former often gives better performance, the latter involves estimation of lesser
number of parameters, making the system feasible for practical implementations.
This research focuses on the efficacies of various subspace, statistical and
stereo based feature normalisation techniques. A subspace projection based
method has been investigated as a standalone and adjunct technique involving
reconstruction of noisy speech features from a precomputed set of clean speech
building-blocks. The building blocks are learned using non-negative matrix
factorisation (NMF) on log-Mel filter bank coefficients, which form a basis for
the clean speech subspace. The work provides a detailed study on how the method
can be incorporated into the extraction process of Mel-frequency cepstral
coefficients. Experimental results show that the new features are robust to
noise, and achieve better results when combined with the existing techniques.
The work also proposes a modification to the training process of SPLICE
algorithm for noise robust speech recognition. It is based on feature
correlations, and enables this stereo-based algorithm to improve the
performance in all noise conditions, especially in unseen cases. Further, the
modified framework is extended to work for non-stereo datasets where clean and
noisy training utterances, but not stereo counterparts, are required. An
MLLR-based computationally efficient run-time noise adaptation method in SPLICE
framework has been proposed.
|
['D. S. Pavan Kumar']
|
2015-07-14
| null | null | null | null |
['robust-speech-recognition']
|
['speech']
|
[ 3.99716288e-01 -3.11383784e-01 5.08940160e-01 -5.00050902e-01
-9.38945711e-01 -3.95702809e-01 5.74648917e-01 -5.77832878e-01
-4.17674452e-01 4.76630956e-01 5.18121779e-01 -3.48182738e-01
-1.34643450e-01 -3.73329282e-01 -2.70019650e-01 -1.10029840e+00
2.66575843e-01 -3.89728583e-02 7.41531104e-02 -2.61527032e-01
1.32992133e-01 6.83638990e-01 -1.85392165e+00 2.69517094e-01
4.07122046e-01 6.21397913e-01 6.03814960e-01 9.86456573e-01
-7.87515640e-02 9.12480280e-02 -6.31025970e-01 3.96624953e-02
4.67936456e-01 -3.79915595e-01 -3.49624574e-01 3.55642736e-01
2.93802947e-01 -1.52779892e-01 -3.39989483e-01 1.01817083e+00
9.51842129e-01 5.74675381e-01 5.46336412e-01 -6.36008203e-01
-5.88335767e-02 9.19363946e-02 -8.66615325e-02 2.75001019e-01
6.19125247e-01 -1.62312388e-01 3.92986774e-01 -1.23875368e+00
8.62203762e-02 1.36387086e+00 7.12522984e-01 5.46411276e-01
-1.22255242e+00 -6.50383890e-01 -3.02694499e-01 3.45580220e-01
-1.28297162e+00 -1.06843138e+00 7.90351272e-01 -1.53549775e-01
1.59412146e+00 8.30332756e-01 4.08490568e-01 9.96378481e-01
-9.54545140e-02 4.93748695e-01 1.42244315e+00 -1.02827632e+00
2.19963923e-01 3.19676012e-01 1.87120408e-01 1.49668291e-01
-8.64270404e-02 5.52366972e-01 -5.87737739e-01 -2.78272867e-01
5.11939585e-01 -2.85469770e-01 -4.59367901e-01 -3.43542635e-01
-8.15066099e-01 5.65825880e-01 -1.29081085e-01 6.66915536e-01
-3.00613582e-01 -4.98630226e-01 3.95541519e-01 5.05245566e-01
2.38755330e-01 6.72012791e-02 -4.67865825e-01 -4.03302759e-01
-1.17908216e+00 -2.10610926e-02 1.06262445e+00 7.96037972e-01
5.79901457e-01 5.62011838e-01 1.84594885e-01 1.30475330e+00
6.52020633e-01 7.24308968e-01 7.41967499e-01 -6.18345499e-01
2.63302475e-01 1.89226605e-02 -1.20684989e-01 -7.75014281e-01
-2.52288103e-01 -4.39232886e-01 -5.08841932e-01 1.99707806e-01
1.60509318e-01 -6.50537945e-03 -1.19014275e+00 1.40932202e+00
5.05745590e-01 2.96623230e-01 4.05864865e-01 7.01848269e-01
5.77615619e-01 8.49047482e-01 -3.40989023e-01 -6.86730087e-01
9.41665411e-01 -8.27980280e-01 -9.94257271e-01 2.92099286e-02
3.32990378e-01 -1.57289112e+00 7.81496286e-01 7.62974679e-01
-8.02124023e-01 -8.19678485e-01 -1.06158340e+00 3.50137144e-01
-4.17019039e-01 1.35383161e-03 9.81402025e-02 1.29256880e+00
-1.17218733e+00 3.30817431e-01 -7.11575449e-01 -6.45432830e-01
-5.09288013e-01 6.41373038e-01 -5.03359377e-01 9.78258532e-03
-8.87107015e-01 1.08172548e+00 3.72992247e-01 4.73680049e-01
-4.54342693e-01 -1.69681553e-02 -9.33097839e-01 -1.27361000e-01
-1.51377711e-02 -3.04103166e-01 1.22952545e+00 -8.30691755e-01
-1.97018123e+00 2.22804189e-01 -6.36378944e-01 -1.78992718e-01
1.04405098e-01 -2.73067623e-01 -1.04690838e+00 -1.98213588e-02
-2.03679055e-01 -4.99833673e-02 1.17044854e+00 -1.12968266e+00
-4.85429972e-01 -3.27021003e-01 -6.08747303e-01 4.66649413e-01
-2.33018607e-01 2.87258685e-01 -2.24928826e-01 -5.73586166e-01
5.40828228e-01 -7.55230784e-01 -5.35162054e-02 -7.95872688e-01
-4.31009904e-02 1.32114664e-01 1.30447555e+00 -9.41749156e-01
1.24608994e+00 -2.43073916e+00 -2.02226236e-01 5.56869566e-01
-7.75023282e-01 7.88207829e-01 -7.83779696e-02 8.11657071e-01
-4.63836730e-01 -3.98471594e-01 -9.16821733e-02 -2.45129257e-01
-1.43682286e-01 4.87961620e-01 -2.36385524e-01 4.33848917e-01
-1.44822925e-01 1.81943536e-01 -4.79494482e-01 -1.51160970e-01
8.23830545e-01 7.97594070e-01 -3.22491348e-01 2.55379617e-01
7.30071664e-01 3.78106683e-01 -5.72381839e-02 3.27675283e-01
8.67624700e-01 9.83934999e-01 -1.68169454e-01 -4.08543319e-01
-2.42879659e-01 4.94699478e-01 -1.86917102e+00 1.37101388e+00
-5.97844541e-01 5.28360963e-01 5.61406136e-01 -1.18652415e+00
1.25727046e+00 9.48018730e-01 3.77738178e-02 -2.70913303e-01
-6.91426173e-02 6.16429269e-01 8.43123347e-02 -7.77985573e-01
3.24527323e-01 -4.62286949e-01 5.01304328e-01 -1.00350611e-01
3.18652213e-01 -4.59355056e-01 -1.23247998e-02 -2.36003622e-01
7.48639941e-01 2.44533837e-01 7.28500068e-01 -1.84469983e-01
1.13469803e+00 -5.29961765e-01 4.53001052e-01 5.76497495e-01
-1.92525148e-01 6.22203708e-01 -4.30288285e-01 -2.02752929e-02
-1.00732219e+00 -9.44028497e-01 -4.61219162e-01 7.58433640e-01
-3.70254666e-01 -3.21095258e-01 -7.61671007e-01 -2.71542996e-01
-4.12373573e-01 8.92814100e-01 9.98278111e-02 6.26080185e-02
-6.11096740e-01 -6.75743520e-01 4.37694550e-01 3.07966143e-01
2.93760389e-01 -8.84413242e-01 -1.22350760e-01 4.56835270e-01
-5.17998673e-02 -9.29922223e-01 -2.36064643e-01 5.73771060e-01
-9.48741257e-01 -5.86709797e-01 -5.42877972e-01 -1.00834692e+00
6.04741096e-01 5.78560650e-01 3.09947312e-01 -1.76160932e-01
-5.14805093e-02 6.94963634e-01 -6.28498375e-01 -4.55769897e-01
-6.66036189e-01 -4.78521764e-01 5.56055307e-01 1.43773481e-01
8.15555513e-01 -7.48770714e-01 -8.02182630e-02 4.81214136e-01
-7.79691637e-01 -4.01701838e-01 6.96874499e-01 1.33441424e+00
4.20383543e-01 2.13783190e-01 5.98767459e-01 -5.10114431e-01
6.77794993e-01 5.83668798e-03 -3.15845847e-01 -5.58146946e-02
-4.73897696e-01 -9.46065709e-02 6.73272431e-01 -2.85798818e-01
-1.54089987e+00 4.09417540e-01 -7.17186093e-01 -2.03778267e-01
-7.96012223e-01 2.86956877e-01 -4.72291023e-01 -2.05455109e-01
7.56022692e-01 6.12379968e-01 7.89240673e-02 -9.68293071e-01
1.85231969e-01 1.45973647e+00 5.08374751e-01 -2.23481357e-01
9.33555365e-01 1.03272706e-01 -2.75746077e-01 -1.64300442e+00
-1.90930352e-01 -1.33370280e+00 -7.60685921e-01 -4.64562215e-02
2.41825700e-01 -8.02817643e-01 7.08459392e-02 4.74644542e-01
-9.56887305e-01 3.10480595e-01 -1.74336851e-01 1.06527066e+00
-5.48952281e-01 8.31511438e-01 -2.71712035e-01 -1.41234183e+00
-3.02560180e-01 -1.27908480e+00 9.08355057e-01 5.76735921e-02
-2.03573003e-01 -8.30571353e-01 6.95811957e-02 5.55151701e-01
5.16164005e-01 -4.81847405e-01 5.77411294e-01 -1.06121588e+00
6.98237792e-02 -3.01200211e-01 3.33059460e-01 9.81836140e-01
4.09276545e-01 -1.79004654e-01 -1.53526819e+00 -5.25047779e-01
8.00254464e-01 1.59905523e-01 6.18527889e-01 2.90123791e-01
3.45108986e-01 -2.27770135e-01 -1.28111064e-01 4.96326894e-01
1.25836837e+00 6.64385796e-01 7.13545501e-01 2.70965755e-01
3.38391006e-01 5.35891175e-01 5.73589265e-01 1.08589135e-01
-3.69985491e-01 8.05543005e-01 -7.17102289e-02 -2.37256638e-03
-2.74375707e-01 4.99780923e-02 7.33784556e-01 1.63864768e+00
4.37211134e-02 1.46104619e-01 -5.60325205e-01 5.18922508e-01
-1.47175753e+00 -1.13266599e+00 -1.80706918e-01 2.40920043e+00
5.48184812e-01 -1.30015165e-01 -1.11236759e-01 8.30985487e-01
7.05322742e-01 -9.11421776e-02 -5.55352308e-02 -9.30676818e-01
-9.70566943e-02 6.05311096e-01 2.71038711e-01 8.84689748e-01
-1.11247766e+00 6.31356418e-01 6.35381317e+00 1.06006646e+00
-1.19169128e+00 1.12092711e-01 -5.90183251e-02 1.26818165e-01
7.45039955e-02 4.67765480e-02 -6.82588100e-01 1.65715143e-01
1.25366926e+00 1.99812710e-01 6.06446385e-01 7.64576018e-01
6.59362853e-01 -2.86204845e-01 -9.14115071e-01 1.16774714e+00
2.40586400e-01 -5.15566885e-01 -2.12406710e-01 3.63272019e-02
4.88666981e-01 -3.21295150e-02 -9.11277607e-02 2.82803029e-01
-4.29219127e-01 -8.25552225e-01 4.74117726e-01 2.64963806e-01
4.09419745e-01 -7.59405077e-01 1.02827454e+00 5.50150871e-01
-1.12169468e+00 -8.04291442e-02 -4.70687270e-01 -2.47415587e-01
2.16172487e-01 4.43965465e-01 -1.32952905e+00 6.89822555e-01
6.06083810e-01 1.42993048e-01 -1.77855968e-01 1.23041332e+00
-1.91372670e-02 8.65569115e-01 -5.92071235e-01 2.41530198e-03
5.82964867e-02 -3.32109213e-01 1.02559304e+00 1.70617902e+00
6.51789367e-01 -4.16277498e-02 -1.44891083e-01 1.31001532e-01
8.56483817e-01 5.04155099e-01 -9.25463974e-01 2.69696891e-01
4.11434978e-01 1.28272557e+00 -3.32460642e-01 -1.76995784e-01
-5.74506342e-01 8.37768197e-01 -2.70731717e-01 5.61524272e-01
-1.02459304e-01 -4.60411251e-01 4.95315135e-01 -1.75940722e-01
4.64826196e-01 -4.04076159e-01 -1.54952556e-01 -8.68565559e-01
1.14066094e-01 -1.27854979e+00 -1.80344507e-02 -6.31264031e-01
-1.11705244e+00 6.84122622e-01 5.04004359e-02 -1.24305642e+00
-7.43333220e-01 -7.14583933e-01 -6.06811106e-01 1.38259816e+00
-1.22125661e+00 -9.67725337e-01 2.16592878e-01 6.54776990e-01
9.73507762e-01 -5.79993963e-01 1.31759119e+00 3.01987261e-01
-3.50005537e-01 4.83001620e-01 6.06244385e-01 -2.58379132e-01
5.91732025e-01 -1.10148573e+00 2.12770507e-01 1.32279932e+00
4.62270379e-01 8.70357752e-01 1.02087104e+00 -5.93676031e-01
-1.25388598e+00 -7.04939604e-01 1.06003666e+00 -3.47503908e-02
3.44964802e-01 -4.28913802e-01 -8.90475690e-01 2.05230951e-01
2.54108220e-01 -3.77746075e-01 9.57429945e-01 3.44251171e-02
-2.01837216e-02 -1.61450565e-01 -1.08457184e+00 3.36145580e-01
6.24631524e-01 -8.23676646e-01 -1.12084854e+00 5.73969558e-02
2.26292953e-01 -3.01558077e-01 -6.09121919e-01 3.21448922e-01
5.87205827e-01 -1.08102214e+00 9.14059758e-01 -1.28193423e-01
-6.78781390e-01 -6.37156129e-01 -6.24529362e-01 -1.46740890e+00
-2.79392451e-01 -9.52510595e-01 1.25445083e-01 1.47155333e+00
4.92766112e-01 -7.81420231e-01 4.57400471e-01 4.32766020e-01
-3.55397880e-01 -3.18344623e-01 -1.31307638e+00 -9.85537767e-01
-4.66266245e-01 -7.12961793e-01 2.17040315e-01 4.79963481e-01
1.02337204e-01 4.55989152e-01 -4.71029192e-01 3.06199163e-01
3.21785212e-01 -4.25833941e-01 7.83531427e-01 -7.55318582e-01
-4.90631104e-01 5.14715388e-02 -7.91037977e-01 -9.70178008e-01
-1.48385763e-01 -5.11903286e-01 2.67725408e-01 -1.18456388e+00
-1.74699798e-01 -1.14197105e-01 -1.90228552e-01 6.88808933e-02
-6.86879829e-02 -4.79203388e-02 1.72261909e-01 4.11798619e-03
2.39510328e-01 3.82758796e-01 7.90502667e-01 1.81120202e-01
-3.65103990e-01 4.87327963e-01 -1.58958182e-01 7.23231673e-01
8.25063765e-01 -3.20402443e-01 -5.80076993e-01 -6.17214218e-02
-6.20953262e-01 -1.38427213e-01 5.66451922e-02 -1.20671380e+00
3.36644709e-01 1.57818034e-01 3.92212182e-01 -6.61690772e-01
8.87315094e-01 -1.22013164e+00 4.45118219e-01 2.39749715e-01
5.79038784e-02 -2.38215700e-02 2.03552336e-01 6.26299739e-01
-5.29616237e-01 -7.20484734e-01 7.67147243e-01 -5.81093170e-02
-7.31155217e-01 -5.08999884e-01 -6.82539701e-01 -6.91085160e-01
7.21502066e-01 -9.51618373e-01 3.48729491e-01 -5.30594409e-01
-9.77033794e-01 -5.82702935e-01 1.33650720e-01 3.16217095e-01
6.91197813e-01 -1.10124457e+00 -5.08903325e-01 8.63622010e-01
-2.89494812e-01 -3.09932441e-01 2.16398239e-01 9.13758457e-01
-3.71794522e-01 7.71487772e-01 -1.78812593e-02 -5.58217883e-01
-1.74883747e+00 4.52251732e-01 2.61338264e-01 1.89775825e-01
-3.60381812e-01 8.52846146e-01 9.25877225e-03 -5.97124577e-01
4.27629918e-01 -1.36159271e-01 -4.01585639e-01 -2.42364556e-01
5.16237676e-01 5.19368649e-01 6.02663875e-01 -1.33646405e+00
-2.07528591e-01 6.11950159e-01 7.84427673e-02 -6.24236822e-01
1.35564423e+00 -4.41882551e-01 2.07154397e-02 5.19113064e-01
1.33053243e+00 3.90924215e-01 -7.38416374e-01 -3.73802900e-01
9.51303449e-03 -6.18442237e-01 3.38762015e-01 -7.88685977e-01
-4.52398956e-01 9.16216016e-01 1.12775505e+00 1.83305502e-01
1.47643447e+00 -5.63972175e-01 4.02048707e-01 4.33450937e-01
3.10440004e-01 -1.30012441e+00 -5.27365327e-01 5.77767134e-01
9.29774106e-01 -9.37523544e-01 -2.48419523e-01 -6.35272205e-01
-2.86507934e-01 1.42010176e+00 2.98309386e-01 1.75510123e-01
7.45418429e-01 4.32754248e-01 4.06263143e-01 2.96242267e-01
-4.66207147e-01 -3.00736010e-01 4.33734447e-01 1.06728947e+00
5.88564336e-01 6.86963350e-02 -5.23410261e-01 3.99943203e-01
-4.45147842e-01 -4.02735829e-01 2.81676441e-01 1.03453505e+00
-7.00898945e-01 -1.54770756e+00 -1.10722184e+00 4.38960820e-01
-5.12347817e-01 -2.28654623e-01 -1.98797226e-01 4.54847425e-01
8.53647962e-02 1.54245663e+00 -4.32984114e-01 -5.76459587e-01
5.33434093e-01 5.27328551e-01 2.94149727e-01 -6.72988713e-01
-6.24114692e-01 8.75658453e-01 3.61319244e-01 -2.46940315e-01
-5.48057616e-01 -9.61951911e-01 -8.58877540e-01 2.05080286e-01
-9.48093653e-01 4.02739584e-01 1.08536541e+00 1.04280317e+00
2.46691257e-02 1.73137784e-01 7.94330478e-01 -1.01903772e+00
-8.82167399e-01 -1.32260239e+00 -6.41562104e-01 2.88634360e-01
3.55130076e-01 -4.47394550e-01 -5.36670566e-01 2.55286008e-01]
|
[14.879878044128418, 5.865756034851074]
|
02a05199-436f-4ff1-add9-c2a550013ed3
|
learn-to-combine-linguistic-and-symbolic
| null | null |
https://aclanthology.org/2020.coling-main.466
|
https://aclanthology.org/2020.coling-main.466.pdf
|
Learn to Combine Linguistic and Symbolic Information for Table-based Fact Verification
|
Table-based fact verification is expected to perform both linguistic reasoning and symbolic reasoning. Existing methods lack attention to take advantage of the combination of linguistic information and symbolic information. In this work, we propose HeterTFV, a graph-based reasoning approach, that learns to combine linguistic information and symbolic information effectively. We first construct a program graph to encode programs, a kind of LISP-like logical form, to learn the semantic compositionality of the programs. Then we construct a heterogeneous graph to incorporate both linguistic information and symbolic information by introducing program nodes into the heterogeneous graph. Finally, we propose a graph-based reasoning approach to reason over the multiple types of nodes to make an effective combination of both types of information. Experimental results on a large-scale benchmark dataset TABFACT illustrate the effect of our approach.
|
['Ting Liu', 'Qingyu Yin', 'Yu Zhang', 'Qi Shi']
|
2020-12-01
| null | null | null |
coling-2020-8
|
['table-based-fact-verification']
|
['natural-language-processing']
|
[-9.14796367e-02 3.38909417e-01 -7.51880586e-01 -5.89881361e-01
-5.65014839e-01 -6.82241082e-01 6.10184371e-01 4.83290404e-01
2.69355476e-01 5.04082680e-01 1.78476393e-01 -6.49548292e-01
1.59643404e-02 -1.53541756e+00 -1.19552219e+00 1.73884571e-01
-7.78997540e-02 2.07104594e-01 5.98985374e-01 -3.65630597e-01
8.06002170e-02 1.26575351e-01 -1.47696507e+00 6.91814184e-01
9.99322414e-01 7.21361578e-01 -2.94287503e-01 2.94393986e-01
-6.25983179e-01 1.90213084e+00 -1.48340747e-01 -6.77642643e-01
1.33729264e-01 -3.62643659e-01 -8.56715918e-01 -2.19735265e-01
1.83634207e-01 -3.08378726e-01 -3.16947341e-01 1.29457080e+00
-1.64425686e-01 -1.74936369e-01 1.37470931e-01 -1.59360731e+00
-4.00452971e-01 1.39431298e+00 -1.94550961e-01 -3.42703372e-01
7.73514926e-01 2.91713834e-01 1.29056156e+00 -4.88037109e-01
7.77136326e-01 1.54572582e+00 7.69261062e-01 2.69216061e-01
-1.04579937e+00 -3.75006557e-01 1.11533627e-01 2.76983052e-01
-1.24566066e+00 -2.55973279e-01 1.05581450e+00 -5.50581634e-01
1.18234134e+00 2.97989219e-01 6.42368495e-01 4.14347321e-01
1.55588120e-01 7.61725783e-01 9.55980897e-01 -5.82366467e-01
1.92196831e-01 2.59492546e-01 6.46408021e-01 1.37745607e+00
3.63056690e-01 8.63948688e-02 -4.20650780e-01 -5.07236540e-01
3.53705227e-01 1.09614283e-02 -8.87894332e-02 -6.13479793e-01
-1.06832159e+00 8.95915747e-01 5.61542571e-01 3.28616798e-01
1.04307279e-01 5.89853108e-01 8.71941149e-01 4.97217387e-01
-7.39410147e-02 3.67036700e-01 -5.67377925e-01 1.13714263e-01
-6.31497979e-01 1.95494264e-01 1.31424081e+00 1.21440315e+00
1.00435090e+00 -1.68820359e-02 -9.47558060e-02 3.84072393e-01
6.35590434e-01 4.24825042e-01 1.94927141e-01 -9.79549825e-01
5.42684138e-01 1.47772992e+00 -2.22178176e-01 -1.19253898e+00
4.97791264e-03 2.68395860e-02 -2.44351253e-01 2.20213652e-01
2.70872623e-01 2.36965343e-01 -4.49223816e-01 1.77265322e+00
2.97872335e-01 2.39100024e-01 2.68694639e-01 5.22539139e-01
8.22795868e-01 3.65509361e-01 -2.15306893e-01 1.01076066e-01
1.23572433e+00 -9.66200173e-01 -6.15480959e-01 -1.35712535e-03
1.13572884e+00 3.56937610e-02 1.03577662e+00 4.86162677e-02
-1.12151814e+00 -2.94218600e-01 -1.32556295e+00 -1.78676337e-01
-6.15118146e-01 -5.34089580e-02 9.29064393e-01 5.21615386e-01
-9.58789825e-01 4.52284127e-01 -7.96483159e-01 1.91536129e-01
3.41566443e-01 3.85100186e-01 -2.79327393e-01 -3.00218314e-01
-1.25903296e+00 5.81099987e-01 9.89295363e-01 -6.88460618e-02
-7.37743437e-01 -7.66259134e-01 -1.58900011e+00 7.16838464e-02
1.04728556e+00 -9.43481624e-01 1.15053308e+00 -1.01235855e+00
-1.33944333e+00 6.80461109e-01 -2.11585701e-01 -5.82549632e-01
2.03490853e-01 3.93061727e-01 -3.89494270e-01 1.35560304e-01
-1.26015335e-01 1.70773894e-01 2.39582524e-01 -1.25100684e+00
-3.12009454e-01 -4.58701849e-01 9.14278567e-01 -3.23748082e-01
2.30536833e-02 6.98713288e-02 -3.31447124e-01 -5.81300259e-02
-8.66812244e-02 -6.83187664e-01 -1.60904154e-02 1.89201124e-02
-4.77393925e-01 -2.62314826e-01 6.31121099e-01 -5.34182727e-01
1.33439529e+00 -1.95817482e+00 2.41293505e-01 5.00866592e-01
5.96843719e-01 8.90974104e-02 1.87798500e-01 6.33765221e-01
1.88634947e-01 4.20494020e-01 -2.46550113e-01 2.91991800e-01
6.17945969e-01 3.38829368e-01 -3.14080000e-01 8.64343345e-02
3.23018581e-01 1.37637138e+00 -1.09911847e+00 -8.31200719e-01
9.71135125e-02 2.03055535e-02 -9.33553874e-01 -4.46648151e-02
-9.72878039e-01 -1.33932799e-01 -8.31060886e-01 7.98475027e-01
4.95582998e-01 -2.40914091e-01 7.02725232e-01 -1.81693241e-01
1.98299646e-01 4.98535424e-01 -1.08069742e+00 1.87882543e+00
-6.85411334e-01 1.96986988e-01 -1.02983639e-01 -8.56912553e-01
8.28485191e-01 1.64803788e-01 4.76101739e-03 -3.98453653e-01
8.03159848e-02 2.56617278e-01 -5.31931259e-02 -6.08321667e-01
4.92964387e-02 -1.70625046e-01 -4.69024748e-01 5.15892982e-01
-4.01200503e-02 -3.60247910e-01 3.30238819e-01 5.63388646e-01
1.23725784e+00 4.65340078e-01 5.51297426e-01 -2.00573653e-01
1.06378853e+00 3.35589319e-01 6.93999290e-01 6.58761382e-01
2.19684467e-01 -2.22345814e-01 1.12896562e+00 -6.62670195e-01
-6.92688584e-01 -5.18206358e-01 3.97598088e-01 7.83933878e-01
6.89638034e-02 -9.97967958e-01 -7.95349300e-01 -1.11369216e+00
1.59273788e-01 7.75095701e-01 -3.45345259e-01 -5.12424052e-01
-7.48658776e-01 -1.08231224e-01 8.52524817e-01 7.89805591e-01
4.87306148e-01 -8.83575797e-01 -5.36938608e-01 -6.16865791e-02
-2.18848243e-01 -1.08837557e+00 -1.97695538e-01 1.67477094e-02
-7.05972791e-01 -1.39316547e+00 5.76175630e-01 -6.69568539e-01
7.17333913e-01 -2.61802137e-01 1.36526215e+00 7.28392422e-01
1.68260932e-01 2.00382262e-01 -1.81921035e-01 2.14174669e-02
-8.34509194e-01 -1.34961784e-01 -4.80944663e-01 -3.06103289e-01
2.81164199e-01 -5.72512925e-01 1.98299080e-01 -3.21564712e-02
-1.10896778e+00 2.92771757e-01 2.35134736e-01 8.51823390e-01
4.21382785e-01 6.18129134e-01 1.00127809e-01 -1.39586830e+00
3.98718059e-01 -3.08814675e-01 -1.03107953e+00 7.58545220e-01
-2.92042315e-01 7.10849345e-01 9.60552633e-01 -1.78336501e-01
-9.20364082e-01 2.10360717e-02 1.77452654e-01 -4.29238170e-01
2.32342675e-01 1.28958762e+00 -7.98143804e-01 -3.05314690e-01
4.46325839e-01 1.38177305e-01 -9.76735279e-02 1.02165103e-01
4.84313041e-01 3.16783547e-01 5.00784099e-01 -1.46919060e+00
1.01838291e+00 2.84259081e-01 2.29145914e-01 3.03420722e-02
-5.58239818e-01 1.75201982e-01 -6.22427344e-01 7.05895796e-02
5.03526390e-01 -7.36726701e-01 -1.10924220e+00 1.07264422e-01
-1.12623620e+00 -7.01022863e-01 -3.32809359e-01 1.43993631e-01
-7.24728942e-01 3.39018732e-01 -6.11846745e-01 -7.31275797e-01
-1.15691312e-02 -1.40196598e+00 1.00114453e+00 -3.08356732e-01
-7.99680725e-02 -1.21532047e+00 -2.68250834e-02 2.08120704e-01
3.18536011e-04 4.90770251e-01 1.62261569e+00 -7.95129955e-01
-9.87559497e-01 -3.54886353e-01 -4.81712580e-01 3.08561027e-02
-4.15330706e-03 1.71694636e-01 -5.39633334e-01 1.47746474e-01
-1.46370590e-01 -4.08357829e-01 3.62335831e-01 -2.05887631e-01
9.61133361e-01 -6.22299790e-01 -2.92525053e-01 4.39469308e-01
1.77215123e+00 2.42608756e-01 6.09657168e-01 2.27599815e-01
8.46262336e-01 4.41772550e-01 4.22424585e-01 1.69825748e-01
8.30474496e-01 6.78460360e-01 3.62642497e-01 3.76101196e-01
-2.41757780e-01 -9.03570235e-01 3.81890655e-01 8.90621424e-01
1.12697370e-01 3.24548155e-01 -1.42944646e+00 2.15590969e-01
-2.04125381e+00 -9.65602279e-01 -1.44715175e-01 1.89549434e+00
1.15467358e+00 1.21707097e-01 -1.48600154e-03 2.98508823e-01
4.93123233e-01 -8.94847959e-02 -2.80249357e-01 -3.39436620e-01
5.86845540e-02 -1.77108884e-01 2.67123729e-01 8.17655385e-01
-7.47442961e-01 1.07459903e+00 5.82199240e+00 7.51469135e-01
-8.50444317e-01 -5.58786206e-02 -1.42730623e-01 5.35130143e-01
-1.02408624e+00 5.11037469e-01 -5.70616722e-01 2.26858497e-01
1.08123779e+00 -5.07850111e-01 8.52842391e-01 1.02559614e+00
-3.68587226e-01 1.02242805e-01 -1.43113339e+00 5.58508158e-01
-2.67401673e-02 -1.46443856e+00 2.54519731e-01 -2.48947233e-01
4.40138847e-01 -4.03347641e-01 -4.63358581e-01 7.69519806e-01
9.08323526e-01 -8.48091662e-01 1.16631353e+00 6.45676613e-01
4.60971564e-01 -6.56127393e-01 6.77536905e-01 4.62331951e-01
-1.57871592e+00 -2.84189224e-01 -1.60170555e-01 -2.44811684e-01
-3.38599443e-01 5.39246619e-01 -6.15276754e-01 1.24830365e+00
1.43540278e-01 9.20054078e-01 -6.19160593e-01 6.48524106e-01
-7.14114010e-01 1.49795726e-01 -2.80122366e-03 -7.76512548e-02
5.73551506e-02 -1.23690598e-01 1.99551314e-01 9.63354290e-01
-3.61910015e-02 1.77653566e-01 5.65687954e-01 1.30134439e+00
-1.79979131e-01 -2.32235685e-01 -1.00518012e+00 -3.54052395e-01
4.10816640e-01 8.38026524e-01 -5.68129599e-01 -7.82724202e-01
-7.50191450e-01 3.06613147e-01 5.26524246e-01 2.81008333e-01
-1.00479078e+00 -3.81992191e-01 1.06355295e-01 -1.06900714e-01
8.01611096e-02 -3.07666898e-01 -2.47511178e-01 -1.49815333e+00
3.58431160e-01 -1.22314084e+00 4.18293208e-01 -8.92992854e-01
-7.81645596e-01 4.95256305e-01 3.77686143e-01 -8.43157649e-01
-2.91787565e-01 -5.42139649e-01 -4.81012583e-01 7.54173279e-01
-1.64780378e+00 -1.40707374e+00 -1.62870005e-01 7.30063617e-01
2.44254507e-02 7.47734495e-03 7.82108784e-01 2.66651362e-01
-4.90318030e-01 5.73503673e-01 -7.95679033e-01 3.90574366e-01
1.35920987e-01 -1.23740041e+00 7.23174289e-02 9.49131668e-01
1.65397003e-02 1.12465441e+00 4.74386662e-01 -7.70380199e-01
-2.28566599e+00 -1.24201870e+00 9.61178899e-01 -5.29625714e-01
1.12211931e+00 -2.91806847e-01 -9.44761097e-01 1.22798574e+00
-1.33692399e-01 2.21752405e-01 3.98280799e-01 6.58844411e-02
-1.24272633e+00 -1.65862769e-01 -1.16535044e+00 5.30302346e-01
1.21516764e+00 -1.12679911e+00 -8.36176753e-01 1.64789841e-01
1.26636744e+00 -6.49414122e-01 -9.97465730e-01 4.91447866e-01
5.23410261e-01 -8.00106704e-01 8.64810288e-01 -6.88939691e-01
7.75143623e-01 -8.18192363e-01 -4.76136953e-01 -9.45248663e-01
1.24237485e-01 -3.43190193e-01 -3.79559696e-01 1.32751894e+00
4.69431251e-01 -8.34861040e-01 4.19338852e-01 7.71857679e-01
-2.16937393e-01 -6.60312235e-01 -3.84224743e-01 -7.47292221e-01
-4.36647329e-03 -6.60071135e-01 1.15394282e+00 9.94035304e-01
7.43887782e-01 1.41949058e-01 9.79815796e-02 2.25447416e-01
4.22743708e-01 7.67464221e-01 7.64972031e-01 -1.24535346e+00
-5.76834559e-01 -4.99620676e-01 -5.45195997e-01 -3.62126321e-01
8.38692904e-01 -1.44371641e+00 -1.59818023e-01 -1.54684365e+00
3.67019892e-01 -3.66746396e-01 -4.87825647e-02 1.09047043e+00
8.32917355e-03 -4.22315925e-01 1.55335248e-01 -5.88366613e-02
-7.08121061e-01 2.43689373e-01 1.01656675e+00 -6.30407512e-01
1.96685009e-02 -4.13060933e-01 -7.24711776e-01 7.48836756e-01
5.34715414e-01 -5.06421506e-01 -7.39198744e-01 -4.01800334e-01
7.95344532e-01 6.84609592e-01 5.71458519e-01 -8.31153154e-01
5.10820568e-01 -4.57370490e-01 -4.45122838e-01 -1.46825820e-01
9.09891911e-03 -1.10791516e+00 4.20094639e-01 6.84933484e-01
-5.46323299e-01 -1.45531490e-01 2.71861702e-01 3.64799827e-01
-5.99046946e-01 -2.29255393e-01 3.87432873e-01 -2.77167320e-01
-7.84171999e-01 1.54076219e-01 1.32196262e-01 1.02243014e-01
9.83323574e-01 2.30478629e-01 -6.16274297e-01 -1.11694977e-01
-4.26230729e-01 3.69423568e-01 7.80981064e-01 9.97044593e-02
3.87666881e-01 -1.40837133e+00 -2.43371278e-01 3.04522753e-01
2.78964311e-01 -2.27905035e-01 -1.69534281e-01 9.55940306e-01
-6.89453483e-01 2.92694420e-01 -1.28922686e-01 -1.58094063e-01
-1.00996578e+00 1.10002232e+00 4.45234984e-01 -5.89955986e-01
-4.91342545e-01 1.73522368e-01 -1.70291010e-02 -8.79949749e-01
6.72934875e-02 -9.33300674e-01 3.00155766e-02 -4.08414304e-01
4.50639039e-01 -5.35412058e-02 3.65647525e-02 -3.62039715e-01
-5.22571266e-01 6.18895769e-01 4.38417763e-01 -5.31027876e-02
1.00548685e+00 3.20461750e-01 -8.19809794e-01 4.94458258e-01
1.05316448e+00 2.92613834e-01 -5.52614927e-01 -5.59150279e-01
5.16074359e-01 -4.36740756e-01 -1.04549289e-01 -8.91105711e-01
-1.07645929e+00 9.22163486e-01 -5.47702253e-01 2.74703950e-01
1.09765470e+00 2.81739403e-02 6.08690202e-01 6.26219630e-01
8.70606780e-01 -3.60663533e-01 -1.50020927e-01 5.20478904e-01
5.74449122e-01 -9.27997589e-01 -1.89529881e-01 -9.00161862e-01
-4.60129917e-01 1.23980653e+00 5.11854291e-01 -1.92512646e-02
2.77718931e-01 8.21315944e-01 -2.92161554e-01 -3.44280183e-01
-9.15408432e-01 -1.91778421e-01 2.10359916e-01 4.22226995e-01
1.32856831e-01 5.22732586e-02 -1.65502563e-01 9.85778451e-01
-7.51370415e-02 5.57089627e-01 4.92419600e-01 1.35658062e+00
-2.66126752e-01 -1.51209807e+00 -2.32135370e-01 5.64004518e-02
-2.43963271e-01 -1.39581829e-01 -4.85386610e-01 9.76551831e-01
1.62554145e-01 7.27331400e-01 -5.76158702e-01 -5.89543521e-01
3.73237222e-01 3.20989639e-01 6.38409853e-01 -7.35334575e-01
-6.66380346e-01 -4.52081144e-01 3.93374145e-01 -7.91219056e-01
-3.83591622e-01 -2.13126972e-01 -1.63855934e+00 -5.63196778e-01
-5.03270663e-02 1.48116410e-01 3.98842931e-01 8.96675706e-01
1.09771520e-01 7.11273074e-01 3.67897570e-01 -1.56663388e-01
-4.21564311e-01 -3.33787017e-02 -5.22867978e-01 2.81467587e-01
1.89234629e-01 -5.25945544e-01 -3.54203731e-01 9.38262884e-03]
|
[9.262487411499023, 7.573309898376465]
|
1bae6445-b0cc-4bf6-9c31-6e38d1dccc71
|
revisiting-random-forests-in-a-comparative
|
2305.19292
| null |
https://arxiv.org/abs/2305.19292v1
|
https://arxiv.org/pdf/2305.19292v1.pdf
|
Revisiting Random Forests in a Comparative Evaluation of Graph Convolutional Neural Network Variants for Traffic Prediction
|
Traffic prediction is a spatiotemporal predictive task that plays an essential role in intelligent transportation systems. Today, graph convolutional neural networks (GCNNs) have become the prevailing models in the traffic prediction literature since they excel at extracting spatial correlations. In this work, we classify the components of successful GCNN prediction models and analyze the effects of matrix factorization, attention mechanism, and weight sharing on their performance. Furthermore, we compare these variations against random forests, a traditional regression method that predates GCNNs by over 15 years. We evaluated these methods using simulated data of two regions in Toronto as well as real-world sensor data from selected California highways. We found that incorporating matrix factorization, attention, and location-specific model weights either individually or collectively into GCNNs can result in a better overall performance. Moreover, although random forest regression is a less compact model, it matches or exceeds the performance of all variations of GCNNs in our experiments. This suggests that the current graph convolutional methods may not be the best approach to traffic prediction and there is still room for improvement. Finally, our findings also suggest that for future research on GCNN for traffic prediction to be credible, researchers must include performance comparison to random forests.
|
['Baher Abdulhai', 'Scott Sanner', 'Xiaocan Li', 'Ta Jiun Ting']
|
2023-05-30
| null | null | null | null |
['traffic-prediction']
|
['time-series']
|
[-8.35757628e-02 -3.46021466e-02 -6.06495142e-01 -4.72909838e-01
-2.33478919e-01 -1.33540764e-01 6.02322876e-01 -6.16349056e-02
-2.17179403e-01 7.87071288e-01 4.91896898e-01 -1.07379031e+00
-2.84837306e-01 -1.14845145e+00 -7.18677402e-01 -2.24992305e-01
-3.22058916e-01 3.85862857e-01 5.20540178e-01 -5.28407276e-01
1.01542108e-01 7.53405750e-01 -1.55029488e+00 1.61559120e-01
1.07079589e+00 9.45322871e-01 -7.43007511e-02 6.48527801e-01
-1.20274387e-01 1.07133186e+00 -1.67160645e-01 -5.94152093e-01
3.59136432e-01 1.13946572e-01 -6.66698456e-01 -3.67541432e-01
8.44980538e-01 -8.61664936e-02 -9.43925023e-01 4.97282952e-01
1.27634108e-01 5.42102039e-01 5.38757622e-01 -1.60756612e+00
-6.10342622e-01 5.37763298e-01 -4.98004049e-01 6.56135380e-01
-1.61904097e-01 3.48049998e-01 1.03765392e+00 -4.14982438e-01
4.47930485e-01 1.24600315e+00 1.15360177e+00 2.20489904e-01
-1.18347764e+00 -9.95889246e-01 4.38511193e-01 5.15162706e-01
-1.14616334e+00 -5.55099547e-01 5.81884027e-01 -4.97002929e-01
1.22444034e+00 2.38214448e-01 8.09850693e-01 8.28232706e-01
5.61830640e-01 5.90313494e-01 8.66398573e-01 -6.22107536e-02
-1.43042028e-01 -2.63902187e-01 5.01533270e-01 7.90894032e-01
4.16836888e-01 4.56310719e-01 -4.93126988e-01 -2.02754773e-02
5.24796605e-01 3.23040001e-02 8.46732110e-02 -8.75841007e-02
-8.54376733e-01 9.15855706e-01 8.92776906e-01 -8.86068419e-02
-4.72422570e-01 7.28307486e-01 1.77725807e-01 3.98824923e-02
7.23970532e-01 2.29207277e-01 -2.26797953e-01 -3.43788117e-01
-9.50557530e-01 3.90829146e-01 5.29076874e-01 7.81592667e-01
9.42725003e-01 2.56977499e-01 -1.55593470e-01 5.82816958e-01
1.35740131e-01 4.13195252e-01 -1.31002545e-01 -9.08557713e-01
7.31504381e-01 7.95306861e-01 -4.44489606e-02 -1.45311081e+00
-7.90309250e-01 -5.23657918e-01 -5.86643100e-01 8.65966901e-02
5.76152980e-01 -4.33511168e-01 -1.12452769e+00 1.57736087e+00
-1.11482747e-01 5.02000451e-01 -4.41030949e-01 8.14344823e-01
6.53439045e-01 5.65885723e-01 4.15787667e-01 3.38185698e-01
8.78157914e-01 -1.12083685e+00 -5.52210987e-01 -4.96213585e-01
7.03517020e-01 -4.69879836e-01 8.92801225e-01 -5.59925660e-02
-6.91749215e-01 -6.93959117e-01 -7.24620819e-01 1.75680429e-01
-7.65573084e-01 -1.65499941e-01 1.16754532e+00 9.02281821e-01
-1.24045515e+00 5.31921208e-01 -9.15515780e-01 -7.54603446e-01
6.38394356e-01 4.45331365e-01 -1.74737453e-01 -1.62074447e-01
-1.31059599e+00 1.30160904e+00 -8.27281475e-02 2.29196697e-01
-3.92836094e-01 -6.51969016e-01 -5.43913901e-01 1.05169073e-01
2.09625795e-01 -7.60259330e-01 8.85556102e-01 -7.13392615e-01
-7.87319958e-01 3.04140504e-02 -5.86362004e-01 -8.94261301e-01
2.79834509e-01 -1.80583194e-01 -8.96475792e-01 -4.07834798e-01
2.59463519e-01 8.75575244e-01 3.11494470e-01 -9.65478063e-01
-8.69160891e-01 -2.25961432e-01 1.24784537e-01 -2.13261694e-02
2.97195110e-02 -1.88086078e-01 -3.56582195e-01 -3.25986713e-01
-1.29297808e-01 -1.22395205e+00 -6.56897664e-01 -3.24531347e-01
-2.92545497e-01 -3.04056793e-01 7.22090006e-01 -6.76212370e-01
1.44792187e+00 -1.83243811e+00 -5.13067424e-01 5.12147665e-01
5.04486859e-01 2.28164420e-01 -1.74871981e-01 4.37365919e-01
5.33404984e-02 4.76720393e-01 2.15420321e-01 -6.98916774e-05
-2.17376590e-01 3.27045053e-01 -1.80389747e-01 2.83794105e-01
2.31611609e-01 1.25643981e+00 -6.87990904e-01 -2.65516758e-01
3.52870375e-01 3.60747784e-01 -5.74345767e-01 -4.28776145e-01
-1.54481649e-01 1.85990527e-01 -2.84346670e-01 4.93433744e-01
5.69751501e-01 -1.53563917e-01 -1.80220976e-02 -1.14248052e-01
-2.84339398e-01 4.62294936e-01 -7.73244679e-01 9.22814965e-01
-3.38892460e-01 1.29502809e+00 -3.85032326e-01 -9.28822160e-01
8.70739579e-01 -1.70271516e-01 5.00463903e-01 -9.45308983e-01
-1.78817943e-01 -1.12379894e-01 4.41080928e-01 -3.78434509e-01
9.37165678e-01 5.92050813e-02 2.10989565e-01 4.08097923e-01
-3.26257765e-01 3.85954499e-01 3.28171253e-01 2.29546025e-01
1.28903580e+00 -1.41177401e-01 -2.00512439e-01 -1.51519209e-01
3.45432125e-02 4.14592952e-01 4.33646262e-01 9.02836919e-01
-3.59524339e-01 3.37913632e-01 4.57552850e-01 -8.85659575e-01
-7.55832493e-01 -8.11198711e-01 1.29772276e-01 1.23401165e+00
-2.14986416e-04 -5.60204327e-01 -3.74458909e-01 -7.35836864e-01
3.09333920e-01 9.01555538e-01 -8.12384903e-01 -1.26929700e-01
-5.95946014e-01 -6.37431502e-01 7.09220409e-01 8.37585568e-01
5.32692969e-01 -7.46213973e-01 -1.29612282e-01 1.88826770e-01
-2.85922021e-01 -1.15172672e+00 -1.20907865e-01 1.34150371e-01
-8.50039363e-01 -1.08593512e+00 -1.20711282e-01 -2.92603225e-01
2.83337027e-01 8.91422033e-01 1.13491774e+00 2.69835413e-01
3.19733590e-01 3.39312762e-01 -1.74535915e-01 -5.19881010e-01
-7.60579035e-02 6.83040261e-01 -7.29184598e-02 -1.27018675e-01
6.34865046e-01 -6.37646198e-01 -5.22237539e-01 5.24667799e-01
-3.07905495e-01 2.11779937e-01 5.00728190e-01 3.81372452e-01
1.97570637e-01 2.99744546e-01 6.03334665e-01 -9.87606645e-01
8.37133706e-01 -8.08107615e-01 -2.06814423e-01 1.82083454e-02
-9.03175831e-01 -1.72781497e-01 3.79799724e-01 -5.34203425e-02
-8.42758715e-01 -7.11034313e-02 -1.24948941e-01 -3.09657127e-01
-2.86530793e-01 8.49083066e-01 3.19351166e-01 -2.07961828e-01
8.41404259e-01 -5.87932467e-02 1.42902821e-01 -1.42114386e-01
3.69875729e-01 3.03586513e-01 6.09997958e-02 -3.51725101e-01
7.34234035e-01 4.17589337e-01 6.30255416e-02 -9.21537519e-01
-5.92658937e-01 -2.36133650e-01 -5.74302137e-01 -6.94060147e-01
9.53152299e-01 -8.84869695e-01 -5.38997650e-01 1.29767274e-02
-8.33125770e-01 -6.71349168e-01 -8.31603818e-03 4.69765931e-01
-2.98490822e-01 -4.93120681e-03 -2.78097421e-01 -7.55826533e-01
1.67557046e-01 -1.05236340e+00 5.81502974e-01 2.23884746e-01
-2.99419045e-01 -1.01976192e+00 -9.17130783e-02 6.55011833e-01
9.95850623e-01 1.34231567e-01 1.09213591e+00 -4.37618405e-01
-7.82102525e-01 -1.83009624e-01 -5.22302508e-01 -1.75798282e-01
1.85337678e-01 3.16860825e-01 -7.17477083e-01 1.55260265e-01
-1.04095948e+00 2.52469808e-01 1.29470098e+00 7.26093709e-01
1.08505249e+00 -1.70338526e-01 -7.09736466e-01 4.21338290e-01
1.10257602e+00 2.51293659e-01 8.86117518e-01 4.99262035e-01
9.33507562e-01 6.48795068e-01 4.28436160e-01 -1.64943546e-01
1.00556004e+00 6.54491961e-01 4.94183570e-01 -3.30185741e-01
-4.73072171e-01 -3.95024329e-01 2.07119748e-01 3.60866517e-01
-6.14657760e-01 -3.42309058e-01 -1.40166926e+00 4.94709820e-01
-2.09980011e+00 -1.35801435e+00 -6.91099823e-01 1.86357045e+00
-2.08195895e-01 4.43950742e-01 6.93549037e-01 -2.95192655e-02
6.39800429e-01 3.39987546e-01 -4.55926448e-01 -5.76698422e-01
-7.80069903e-02 1.09443091e-01 1.05020523e+00 3.60240817e-01
-1.03602719e+00 1.29886627e+00 7.27029371e+00 4.99498636e-01
-1.26307333e+00 -2.38067042e-02 7.51749456e-01 -8.17519203e-02
-3.76693010e-01 1.43790022e-01 -7.83772469e-01 4.78030294e-01
1.46186197e+00 3.35728377e-02 5.64782143e-01 7.28191912e-01
6.07122898e-01 -2.41773397e-01 -5.85472286e-01 4.78415728e-01
-3.09296787e-01 -1.73808920e+00 8.21634382e-02 3.04401755e-01
7.15165257e-01 6.74914598e-01 1.76844433e-01 6.02803826e-01
7.93442726e-01 -1.34195781e+00 5.00518560e-01 6.75725937e-01
3.46664131e-01 -8.59369874e-01 7.06124842e-01 1.14134945e-01
-1.31058800e+00 -2.36602291e-01 -3.98819089e-01 -5.33118725e-01
4.15618509e-01 3.56062353e-01 -8.26315999e-01 4.05906409e-01
7.64940917e-01 1.02219594e+00 -1.07782757e+00 1.16483879e+00
1.35773510e-01 1.30541623e+00 -2.47026488e-01 -1.65092781e-01
4.87407267e-01 -1.08767189e-01 2.94694334e-01 1.31865716e+00
1.51042953e-01 -9.69908908e-02 1.33728370e-01 4.59064692e-01
2.65300870e-01 -7.43277222e-02 -9.70779896e-01 9.86030698e-03
4.94391382e-01 8.34977269e-01 -7.96202123e-01 -5.80243059e-02
-8.17164600e-01 1.28927514e-01 5.58682144e-01 7.25787342e-01
-1.16432703e+00 -1.21726163e-01 1.11913538e+00 6.77024186e-01
3.48714173e-01 -6.27735019e-01 -7.37845302e-01 -7.23677635e-01
-3.73483896e-01 -4.32328612e-01 2.28323415e-01 -8.29926848e-01
-1.14543653e+00 5.41481912e-01 -4.26714681e-03 -1.11342347e+00
-2.20327564e-02 -4.02307242e-01 -7.98494458e-01 7.78167248e-01
-1.55559802e+00 -1.33451033e+00 -4.68557805e-01 5.66675067e-01
3.77635866e-01 -1.66048303e-01 4.06020850e-01 4.40276414e-01
-8.49178076e-01 4.94918674e-01 -1.84163913e-01 1.26952305e-01
3.70510697e-01 -7.72820175e-01 9.62922156e-01 9.51388597e-01
1.32354349e-01 6.12767398e-01 5.90746582e-01 -9.56567764e-01
-1.03897798e+00 -1.51511562e+00 9.54848468e-01 -4.98677582e-01
7.81380594e-01 -7.91630372e-02 -6.32046938e-01 1.13725460e+00
1.55351907e-01 -8.56209323e-02 7.20314801e-01 6.90941989e-01
-1.43203929e-01 -3.28435004e-01 -7.17727900e-01 8.28647137e-01
1.43056965e+00 -3.04696888e-01 2.89731473e-02 1.21642865e-01
5.95157146e-01 -1.69345722e-01 -7.20004857e-01 3.00350159e-01
8.21862400e-01 -1.00379384e+00 7.86065400e-01 -9.40331995e-01
4.59756851e-01 -3.14598620e-01 -2.70921826e-01 -1.39282250e+00
-9.36967731e-01 -2.41825074e-01 -6.85979053e-02 9.32936311e-01
1.00068915e+00 -8.75196040e-01 1.27681816e+00 9.75667179e-01
-2.43377969e-01 -8.20984304e-01 -8.73347044e-01 -7.08208978e-01
1.93457901e-01 -1.20132458e+00 9.44133878e-01 7.38574028e-01
-3.41454953e-01 4.33238894e-01 -4.77946818e-01 4.96484414e-02
2.80172050e-01 -1.49082899e-01 1.20791066e+00 -1.37755752e+00
3.88360679e-01 -6.92312658e-01 -8.50859165e-01 -1.00012469e+00
3.81077945e-01 -9.55613315e-01 -4.27024931e-01 -1.94858778e+00
-2.18905076e-01 -9.81351733e-01 -3.37968618e-01 5.80804646e-01
-1.30919755e-01 1.90597251e-01 2.39384621e-01 2.27647778e-02
-4.53572124e-01 3.21588308e-01 1.03303659e+00 -2.59950012e-01
-2.29682520e-01 4.61300135e-01 -1.03433025e+00 3.41493964e-01
1.04732430e+00 -3.20490301e-01 -5.53627074e-01 -4.77348208e-01
1.78470582e-01 -4.31368500e-02 2.76504576e-01 -1.30114126e+00
4.70208377e-01 -4.65519786e-01 4.90488410e-01 -7.76469588e-01
2.34233037e-01 -9.24846172e-01 3.46166998e-01 2.29152545e-01
-2.38589257e-01 5.66702962e-01 4.37129080e-01 8.12640488e-01
5.18304147e-02 5.91421425e-01 2.32003212e-01 1.18531987e-01
-1.04430926e+00 4.40728456e-01 -1.00453210e+00 -2.08464608e-01
8.87411237e-01 -7.23514259e-01 -5.89803994e-01 -7.66979635e-01
-5.91402471e-01 4.19889867e-01 2.47979194e-01 7.20979035e-01
4.43184137e-01 -1.38059425e+00 -5.15114725e-01 9.63837430e-02
1.58160385e-02 -4.83427346e-01 2.03275800e-01 1.02869403e+00
-3.98588449e-01 6.59140348e-01 -2.33334973e-01 -5.00317514e-01
-9.56920385e-01 3.37581962e-01 3.28796446e-01 -1.78854734e-01
-4.64028269e-01 5.34797907e-01 -3.19444180e-01 -6.62183642e-01
-1.00773172e-02 -4.81686831e-01 -3.86526376e-01 -4.52863909e-02
3.86980884e-02 7.71125376e-01 1.34177893e-01 -1.02246428e+00
-3.78230393e-01 3.63726825e-01 1.10382393e-01 1.01936758e-01
1.35008037e+00 -1.71657324e-01 3.26901972e-01 3.00631881e-01
7.25706995e-01 -2.92301718e-02 -1.04110193e+00 3.94520797e-02
3.16931270e-02 -4.67121482e-01 7.27470443e-02 -7.48630345e-01
-1.53829443e+00 7.47292876e-01 4.90685701e-01 3.10201228e-01
8.08104098e-01 -2.82374173e-01 9.20684457e-01 2.29236335e-01
4.01970685e-01 -8.60741496e-01 -4.84193504e-01 7.45040834e-01
5.40795326e-01 -1.30789387e+00 1.16287395e-01 -4.03923750e-01
-7.31885791e-01 1.01919830e+00 1.04032958e+00 -2.41729140e-01
1.20264864e+00 -8.90403017e-02 -7.30880648e-02 -3.56218308e-01
-1.01099622e+00 -6.26756430e-01 3.66510630e-01 8.76465619e-01
5.34460068e-01 4.69583899e-01 -9.88499448e-03 3.47965568e-01
-5.70269406e-01 -3.99865583e-03 3.52341563e-01 5.57050645e-01
-4.82266665e-01 -8.38260829e-01 -7.44161457e-02 1.26761746e+00
-2.92082001e-02 -1.34820029e-01 -3.92461687e-01 1.08910847e+00
1.24064513e-01 1.30975389e+00 2.72299707e-01 -1.16399825e+00
5.00452042e-01 1.81208774e-02 4.60932516e-02 -3.07747275e-01
-6.34888589e-01 -7.02355444e-01 6.53234482e-01 -7.67012537e-01
-2.66141981e-01 -6.76120222e-01 -9.68171358e-01 -1.01256537e+00
-4.39542145e-01 7.63593465e-02 5.92232227e-01 8.99543107e-01
7.20147014e-01 6.57950521e-01 2.63854176e-01 -7.61985362e-01
3.84422392e-01 -9.23969746e-01 -2.67768025e-01 -8.92207995e-02
1.96237311e-01 -1.04286683e+00 -5.38496263e-02 -3.91445875e-01]
|
[6.4677958488464355, 2.024811267852783]
|
3345622d-b95b-47a5-a0f6-b47b204734cc
|
deepmeshflow-content-adaptive-mesh
|
1912.05131
| null |
https://arxiv.org/abs/1912.05131v1
|
https://arxiv.org/pdf/1912.05131v1.pdf
|
DeepMeshFlow: Content Adaptive Mesh Deformation for Robust Image Registration
|
Image alignment by mesh warps, such as meshflow, is a fundamental task which has been widely applied in various vision applications(e.g., multi-frame HDR/denoising, video stabilization). Traditional mesh warp methods detect and match image features, where the quality of alignment highly depends on the quality of image features. However, the image features are not robust in occurrence of low-texture and low-light scenes. Deep homography methods, on the other hand, are free from such problem by learning deep features for robust performance. However, a homography is limited to plane motions. In this work, we present a deep meshflow motion model, which takes two images as input and output a sparse motion field with motions located at mesh vertexes. The deep meshflow enjoys the merics of meshflow that can describe nonlinear motions while also shares advantage of deep homography that is robust against challenging textureless scenarios. In particular, a new unsupervised network structure is presented with content-adaptive capability. On one hand, the image content that cannot be aligned under mesh representation are rejected by our learned mask, similar to the RANSAC procedure. On the other hand, we learn multiple mesh resolutions, combining to a non-uniform mesh division. Moreover, a comprehensive dataset is presented, covering various scenes for training and testing. The comparison between both traditional mesh warp methods and deep based methods show the effectiveness of our deep meshflow motion model.
|
['Lanpeng Jia', 'Shuaicheng Liu', 'Chuan Wang', 'Yongqing Cui', 'Jue Wang', 'Nianjin Ye']
|
2019-12-11
| null | null | null | null |
['video-stabilization', 'homography-estimation']
|
['computer-vision', 'computer-vision']
|
[ 5.02425358e-02 -4.20914352e-01 -8.55861008e-02 -8.77891928e-02
-2.09969252e-01 -3.12514216e-01 4.28025186e-01 -2.90964723e-01
-1.53839335e-01 5.70395470e-01 1.21854106e-02 4.47916001e-01
-2.11988732e-01 -1.00502455e+00 -8.64659607e-01 -8.99527550e-01
3.75176698e-01 4.92837667e-01 3.55908245e-01 -4.67665017e-01
1.69508845e-01 5.29324830e-01 -1.50536954e+00 -2.15824731e-02
7.93722153e-01 9.09919143e-01 2.75733620e-01 2.14377388e-01
-1.35940567e-01 3.61322105e-01 -9.02585220e-03 -1.96234047e-01
4.25528735e-01 -4.65731025e-01 -6.22087419e-01 3.53628844e-01
8.07361126e-01 -5.68771720e-01 -7.28202581e-01 1.25906587e+00
3.23472887e-01 1.83161303e-01 5.78050911e-01 -1.08202982e+00
-5.01148462e-01 1.03718914e-01 -7.93641984e-01 -2.80300956e-02
3.53287965e-01 3.26879531e-01 6.58622384e-01 -8.94717574e-01
1.05521679e+00 1.32991254e+00 6.59186184e-01 3.62234354e-01
-1.22615361e+00 -5.99189103e-01 -1.36056334e-01 2.07259744e-01
-1.30891931e+00 -1.96323767e-01 1.14364123e+00 -5.16699791e-01
4.67963785e-01 3.44166398e-01 7.75669217e-01 1.03478575e+00
5.42934895e-01 4.00401324e-01 8.79881561e-01 -5.40684909e-02
-6.67158067e-02 -4.18440968e-01 -2.93184578e-01 7.43728936e-01
1.28551260e-01 2.41241767e-03 -3.70588511e-01 7.59181082e-02
1.39097881e+00 2.61013895e-01 -5.90909362e-01 -6.18226051e-01
-1.57816577e+00 5.24892807e-01 3.66773546e-01 4.17478144e-01
-2.77299106e-01 7.86988884e-02 3.50294948e-01 2.71783292e-01
3.41559321e-01 1.31570697e-01 -4.95637208e-02 1.84176952e-01
-1.07063925e+00 3.62887949e-01 5.12702942e-01 8.29221368e-01
9.65261579e-01 3.61044407e-01 -2.65698414e-02 6.96692824e-01
2.31925786e-01 6.84338033e-01 5.19414306e-01 -1.15949988e+00
3.37033629e-01 5.89168191e-01 -8.08692202e-02 -1.83819497e+00
-4.00814801e-01 -3.45577449e-01 -1.58198988e+00 3.21706623e-01
1.79859102e-01 2.78180093e-01 -6.76129520e-01 1.51009393e+00
5.38160086e-01 5.05865395e-01 -1.82602614e-01 1.13124382e+00
8.72904658e-01 6.93614006e-01 -3.85145247e-01 -4.66986179e-01
1.24512923e+00 -9.13087904e-01 -1.02030337e+00 1.91124529e-01
6.86300769e-02 -1.13859034e+00 7.95384526e-01 4.31394041e-01
-1.02179933e+00 -7.43607461e-01 -9.63959873e-01 -2.85769761e-01
6.31362498e-02 -1.72611475e-01 3.83189350e-01 1.42264351e-01
-9.30408537e-01 9.02865469e-01 -7.78629184e-01 -3.78794193e-01
1.30058765e-01 3.23939770e-01 -5.98113835e-01 -8.91597718e-02
-1.01167798e+00 5.66345811e-01 2.79261768e-01 3.14560026e-01
-7.94521213e-01 -3.77922744e-01 -8.75447929e-01 2.82245707e-02
2.72431254e-01 -9.57283378e-01 6.09370410e-01 -1.11717153e+00
-1.74767709e+00 7.81090081e-01 3.82374413e-02 -1.10617489e-01
8.85150850e-01 -9.56146494e-02 -2.42740855e-01 2.29921296e-01
5.67389913e-02 4.23282772e-01 1.14713299e+00 -1.23182404e+00
-3.14210802e-01 -4.03967276e-02 -1.54921040e-01 3.48853767e-01
-3.60398233e-01 -2.35809073e-01 -5.51609874e-01 -9.67590153e-01
4.72476363e-01 -8.30941141e-01 -1.42290071e-01 3.27901870e-01
-4.29708123e-01 1.20185345e-01 1.10335445e+00 -5.89558542e-01
1.17115080e+00 -2.12243366e+00 5.74838698e-01 2.80514687e-01
3.49480122e-01 6.32915273e-02 -1.61886573e-01 4.40582544e-01
-4.50046100e-02 -1.47907957e-01 -4.02510881e-01 -1.75035030e-01
-2.82963693e-01 3.49968374e-01 -1.90468013e-01 8.47967446e-01
1.18945286e-01 6.72778428e-01 -7.51347780e-01 -6.45000041e-01
6.00396395e-01 4.73790526e-01 -5.80867469e-01 2.50970840e-01
-5.90586960e-02 8.70355129e-01 -3.74352127e-01 4.63807613e-01
1.06713271e+00 -1.64909601e-01 -7.84429163e-02 -8.32063615e-01
-4.26802695e-01 -5.09067595e-01 -1.59516752e+00 1.94916749e+00
-2.61517346e-01 4.59834218e-01 3.31584513e-01 -9.60711241e-01
9.02350903e-01 2.48390973e-01 8.83648038e-01 -5.22442818e-01
4.12129492e-01 3.77217591e-01 -2.37395048e-01 -7.22941935e-01
3.85520250e-01 2.08341077e-01 3.55666101e-01 2.27843486e-02
-1.02713488e-01 -2.85685897e-01 8.53681937e-03 -1.87526524e-01
7.13299155e-01 2.49113023e-01 1.24487437e-01 -3.78604680e-01
8.98380578e-01 -7.99861774e-02 7.75517225e-01 1.94082573e-01
1.58374801e-01 1.12322628e+00 2.31957704e-01 -7.77357519e-01
-1.29487538e+00 -6.95236385e-01 -3.09520513e-01 2.24256277e-01
7.85170674e-01 -2.09692016e-01 -7.19848990e-01 -1.17639273e-01
-1.96247652e-01 -7.69949481e-02 -4.23795998e-01 3.16745378e-02
-1.06364870e+00 -4.63152409e-01 1.79315761e-01 1.20368339e-01
9.65689540e-01 -1.02172112e+00 -4.48764592e-01 2.39805683e-01
-3.27661365e-01 -1.20538998e+00 -6.22544289e-01 -5.34577549e-01
-8.88137639e-01 -1.22684407e+00 -9.73639250e-01 -1.02236211e+00
6.07170880e-01 4.50147629e-01 9.27725911e-01 3.76221478e-01
-2.07708746e-01 1.28581405e-01 -1.92211553e-01 4.09025878e-01
-2.77047753e-01 -7.12867156e-02 2.11113706e-01 4.04853821e-01
-3.25477481e-01 -8.14647198e-01 -8.95577312e-01 6.40719831e-01
-1.20743382e+00 2.99012512e-01 5.26109993e-01 9.42450464e-01
9.73935723e-01 1.05770260e-01 3.07430979e-02 -5.38004935e-01
1.49983376e-01 -3.43818188e-01 -6.25545084e-01 1.17977850e-01
-1.74399137e-01 -1.69021368e-01 6.66550457e-01 -4.24483418e-01
-8.37938190e-01 2.99998492e-01 -2.63958514e-01 -9.07168031e-01
-3.25983413e-03 1.54065281e-01 -3.47680658e-01 -5.11342168e-01
3.37559432e-01 3.19497734e-01 3.43556911e-01 -4.10790414e-01
1.97592974e-01 1.63099542e-01 6.41034842e-01 -5.10947704e-01
1.20956910e+00 7.99922526e-01 4.38923478e-01 -9.87997591e-01
-2.61895329e-01 -2.47850016e-01 -7.24152803e-01 -3.79416406e-01
1.13441634e+00 -6.95020676e-01 -8.39869082e-01 7.85566926e-01
-1.37816596e+00 -9.10345986e-02 -4.25482653e-02 6.44113421e-01
-7.10892797e-01 8.82273316e-01 -7.07383394e-01 -2.19655395e-01
-3.64584565e-01 -1.42983496e+00 1.01088727e+00 2.56406248e-01
9.66496393e-02 -9.39899802e-01 1.16419885e-02 1.66244969e-01
3.35356206e-01 6.90024137e-01 8.59591722e-01 2.69085844e-03
-8.99826527e-01 1.69266760e-01 -9.02682841e-02 3.61025780e-01
2.10173637e-01 3.01904619e-01 -6.50492728e-01 -3.69650394e-01
3.02022010e-01 -1.82737842e-01 5.15874505e-01 4.76250142e-01
1.23667562e+00 -2.49562457e-01 -1.43139884e-01 1.08893621e+00
1.62814224e+00 -1.43929869e-01 6.79462612e-01 5.47240615e-01
1.17216778e+00 5.25831819e-01 5.54152012e-01 2.07511485e-01
1.47261262e-01 9.64648306e-01 7.18497694e-01 -4.62847382e-01
-1.59871861e-01 3.28172222e-02 4.37188409e-02 1.10764873e+00
-3.14562052e-01 -1.28822058e-01 -7.31353879e-01 2.26929873e-01
-2.00266480e+00 -8.53003860e-01 -4.62986410e-01 2.28606725e+00
5.70656180e-01 -1.14032239e-01 -1.88375860e-01 5.12654558e-02
7.80825555e-01 3.44495326e-01 -5.15233576e-01 5.61586879e-02
-4.83161539e-01 -6.79307245e-03 4.05354351e-01 4.27978456e-01
-1.07611918e+00 8.73599529e-01 4.86724806e+00 9.44675267e-01
-1.24666822e+00 8.99644289e-03 3.99806559e-01 2.76915342e-01
-2.57299662e-01 -6.10138401e-02 -3.77489388e-01 5.13639450e-01
2.05438491e-03 5.89286014e-02 4.82374758e-01 5.58870256e-01
3.03653359e-01 1.22929201e-01 -8.91019464e-01 1.40311646e+00
1.29199579e-01 -1.40400696e+00 3.50875795e-01 8.41285586e-02
9.82391357e-01 -2.43051633e-01 7.15801492e-03 -2.14080632e-01
-2.05085203e-01 -9.57675457e-01 6.62445247e-01 7.06513882e-01
8.99219275e-01 -7.60044336e-01 8.45186651e-01 1.98912248e-01
-1.39511633e+00 2.32946977e-01 -5.58319211e-01 1.56719595e-01
3.95419359e-01 5.42793274e-01 1.59555137e-01 1.09188473e+00
6.47572041e-01 1.05830240e+00 -1.75756246e-01 1.01500332e+00
1.28500119e-01 3.05818263e-02 -1.51123285e-01 6.56819880e-01
2.08625570e-01 -6.66838884e-01 7.00396597e-01 7.10836232e-01
5.18083692e-01 2.52315938e-01 4.85554785e-01 7.55809009e-01
-9.34284776e-02 4.61331785e-01 -7.25604117e-01 4.72330481e-01
2.21715048e-01 1.31592953e+00 -8.18343699e-01 -2.62933284e-01
-4.37319547e-01 1.11504698e+00 -6.73384815e-02 3.42685729e-01
-7.59280562e-01 -9.30771753e-02 6.83280408e-01 2.27616414e-01
-1.68989804e-02 -4.72961634e-01 -6.28022775e-02 -1.41287076e+00
3.93953174e-03 -1.05588150e+00 -6.67285770e-02 -5.64234078e-01
-1.30192065e+00 6.06355071e-01 -5.97760603e-02 -1.75582528e+00
1.68765694e-01 -5.52717626e-01 -7.20783591e-01 6.91448450e-01
-1.40238297e+00 -1.20501566e+00 -8.70756090e-01 8.88480723e-01
7.94838727e-01 -2.24680342e-02 4.27411944e-01 5.41479290e-01
-6.00230575e-01 2.41562203e-01 3.05239409e-01 2.35549465e-01
8.51644933e-01 -7.56373227e-01 2.37008259e-01 7.63869166e-01
-2.00253069e-01 5.41868031e-01 6.41062796e-01 -8.34509671e-01
-1.69862497e+00 -1.03399563e+00 3.45308632e-01 3.66914198e-02
4.71317917e-01 -3.37972231e-02 -1.24790239e+00 3.85733604e-01
3.66151154e-01 1.95250675e-01 3.31021962e-03 -6.50959253e-01
4.56386581e-02 -1.67820320e-01 -1.05397165e+00 5.31950355e-01
1.10836875e+00 -2.35469803e-01 -2.59397089e-01 4.27981168e-01
4.58240151e-01 -8.71422529e-01 -9.59129393e-01 6.40852690e-01
5.56104064e-01 -1.16477966e+00 1.03077865e+00 -2.20722258e-01
6.71184897e-01 -5.32808840e-01 -9.27237421e-02 -9.94977534e-01
-5.66353202e-01 -6.96570218e-01 -1.84121612e-03 1.25972211e+00
-4.53498095e-01 -4.35368180e-01 6.35443330e-01 1.41387761e-01
-1.00622736e-01 -7.35310793e-01 -9.46263194e-01 -7.18426108e-01
-2.77022049e-02 1.51102751e-01 5.65557778e-01 1.34981477e+00
-7.25848258e-01 5.41440174e-02 -6.26554132e-01 2.24257395e-01
6.99991465e-01 1.89116195e-01 8.83964956e-01 -1.21437228e+00
-1.16942637e-01 -4.91315782e-01 -6.62480772e-01 -1.03190482e+00
1.99912533e-01 -7.28481472e-01 -1.52254179e-01 -1.38451922e+00
4.14718986e-02 -2.91666836e-01 8.09773058e-02 1.26498386e-01
1.02513246e-01 3.93320411e-01 1.39602423e-01 6.23209357e-01
-1.43814385e-01 8.06894064e-01 1.63736987e+00 -2.23459661e-01
-1.10443167e-01 -3.57018590e-01 2.19732925e-01 9.13764656e-01
5.78891635e-01 -2.63039291e-01 -3.48669916e-01 -7.48937666e-01
2.54898965e-01 1.55475140e-01 4.15841669e-01 -1.21452332e+00
3.90704185e-01 -3.91143888e-01 3.77157748e-01 -4.02072757e-01
1.74698010e-01 -1.00629306e+00 5.69285035e-01 4.91448671e-01
1.18187122e-01 4.23394263e-01 -7.52392262e-02 6.24513328e-01
-4.31051165e-01 -3.59378681e-02 9.99896646e-01 -2.91485488e-01
-6.77596390e-01 7.43412733e-01 -4.28894609e-02 -7.43091032e-02
9.97327685e-01 -2.90927351e-01 -8.46344605e-02 -1.60675615e-01
-5.34874141e-01 1.74725190e-01 9.26285028e-01 3.91252100e-01
7.48876631e-01 -1.52723444e+00 -6.34505749e-01 4.42094564e-01
-1.53368995e-01 5.27044654e-01 5.19983947e-01 9.35021162e-01
-1.14380705e+00 -1.60019487e-01 -6.34648085e-01 -9.06427801e-01
-1.15562725e+00 4.87134606e-01 4.00802195e-01 1.13442568e-02
-9.64671791e-01 3.42451841e-01 5.03827393e-01 -1.91907212e-01
1.06553644e-01 -3.44587982e-01 -3.22432458e-01 -7.75627196e-02
3.57113242e-01 4.14877564e-01 -4.26756628e-02 -1.05037963e+00
-9.18496549e-02 1.43200743e+00 1.81899071e-01 1.98072344e-01
1.17551827e+00 -2.45519787e-01 -5.23438036e-01 1.92984670e-01
1.13386774e+00 5.43632135e-02 -1.26426816e+00 -1.43792227e-01
-4.96109486e-01 -6.70829117e-01 -2.64170505e-02 7.89997280e-02
-1.42131686e+00 8.18060219e-01 7.51272798e-01 -2.96412483e-02
1.03886962e+00 -4.88545567e-01 1.15382695e+00 1.32105380e-01
3.38777304e-01 -8.09766889e-01 1.97462127e-01 4.50419396e-01
1.01838672e+00 -1.25216067e+00 6.56274855e-02 -5.93886733e-01
-2.19721809e-01 1.62851071e+00 8.18184257e-01 -4.16370511e-01
4.70927805e-01 4.14307527e-02 8.76205266e-02 -3.19426179e-01
-1.11067913e-01 3.20480093e-02 2.46814072e-01 3.52802038e-01
9.35668424e-02 -3.73662561e-01 -5.21022320e-01 -8.34560990e-02
-4.52600569e-02 -1.67772710e-01 3.24988037e-01 5.76130629e-01
-2.40121022e-01 -1.02177525e+00 -5.75084627e-01 4.90274951e-02
-1.96977451e-01 1.94588125e-01 -3.14153545e-03 8.49746525e-01
3.41045380e-01 5.40191829e-01 1.10896796e-01 -4.35472131e-01
4.08429086e-01 -4.59443599e-01 3.16925526e-01 -2.43046448e-01
-4.40545291e-01 2.82540441e-01 -4.45120692e-01 -8.26099694e-01
-6.88493073e-01 -3.57052922e-01 -1.04857802e+00 -4.85546350e-01
2.66298093e-02 -2.62442559e-01 3.80824357e-01 8.25978756e-01
1.67335480e-01 4.04305398e-01 7.65982449e-01 -1.19935524e+00
-2.05471411e-01 -5.42623580e-01 -5.02722681e-01 7.76214302e-01
3.55126083e-01 -9.00358558e-01 -2.81559736e-01 1.08502716e-01]
|
[9.208930969238281, -2.3349547386169434]
|
c0708912-c4a8-487f-96c6-33a3fa119306
|
learning-fine-grained-visual-understanding
|
2210.03941
| null |
https://arxiv.org/abs/2210.03941v1
|
https://arxiv.org/pdf/2210.03941v1.pdf
|
Learning Fine-Grained Visual Understanding for Video Question Answering via Decoupling Spatial-Temporal Modeling
|
While recent large-scale video-language pre-training made great progress in video question answering, the design of spatial modeling of video-language models is less fine-grained than that of image-language models; existing practices of temporal modeling also suffer from weak and noisy alignment between modalities. To learn fine-grained visual understanding, we decouple spatial-temporal modeling and propose a hybrid pipeline, Decoupled Spatial-Temporal Encoders, integrating an image- and a video-language encoder. The former encodes spatial semantics from larger but sparsely sampled frames independently of time, while the latter models temporal dynamics at lower spatial but higher temporal resolution. To help the video-language model learn temporal relations for video QA, we propose a novel pre-training objective, Temporal Referring Modeling, which requires the model to identify temporal positions of events in video sequences. Extensive experiments demonstrate that our model outperforms previous work pre-trained on orders of magnitude larger datasets.
|
['Winston H. Hsu', 'Jia-Fong Yeh', 'Tsung-Han Wu', 'Bing-Chen Tsai', 'Hung-Ting Su', 'Hsin-Ying Lee']
|
2022-10-08
| null | null | null | null |
['video-question-answering']
|
['computer-vision']
|
[-4.29664627e-02 -1.61730006e-01 -4.94854510e-01 -5.38308740e-01
-8.24256003e-01 -7.15711474e-01 8.25406373e-01 -1.34525821e-01
-2.76586741e-01 3.94119114e-01 5.86745024e-01 -4.32811767e-01
2.52982259e-01 -5.60980976e-01 -1.19374275e+00 -2.11719275e-01
-8.87820795e-02 1.67267606e-01 6.09843075e-01 2.08384991e-01
1.01519838e-01 9.90160480e-02 -1.49821794e+00 1.02121341e+00
1.79916039e-01 1.00369263e+00 3.52239400e-01 1.05162084e+00
-4.31878679e-02 1.83090115e+00 -5.47275133e-02 -9.07736714e-04
-1.03321662e-02 -4.47511137e-01 -1.10397184e+00 3.49319369e-01
8.14100266e-01 -7.99355268e-01 -1.03372133e+00 4.60535467e-01
-1.78820461e-01 3.59263808e-01 2.84131348e-01 -1.26772547e+00
-1.00163829e+00 2.35993072e-01 -4.87648666e-01 6.35388255e-01
6.43457055e-01 2.46648610e-01 1.01407194e+00 -7.01721907e-01
8.45715344e-01 1.44200504e+00 4.14731324e-01 6.51514351e-01
-9.60971415e-01 -3.42501670e-01 6.56187177e-01 6.12796903e-01
-1.21648526e+00 -5.86864531e-01 5.30717611e-01 -5.97278953e-01
1.23251045e+00 -1.43629229e-02 6.02620184e-01 1.23135042e+00
-1.11473985e-01 1.01749134e+00 8.05838466e-01 -7.38808960e-02
-3.86530012e-02 -2.97827542e-01 -1.63358703e-01 9.37509298e-01
-5.69293141e-01 1.90058440e-01 -9.95733500e-01 1.17229886e-01
1.24431229e+00 1.85392216e-01 -1.70492709e-01 -4.14359391e-01
-1.43645465e+00 5.24469137e-01 2.18522638e-01 3.75145763e-01
-2.60950059e-01 7.95668781e-01 3.95449281e-01 2.72865206e-01
2.21906438e-01 -1.73143193e-01 -4.80865628e-01 -4.85674500e-01
-1.22610629e+00 1.67306140e-01 4.23188061e-01 1.04807663e+00
6.28427207e-01 -8.65957290e-02 -4.35337067e-01 2.32659951e-01
4.90451694e-01 2.39201069e-01 2.32071623e-01 -1.50868821e+00
4.99262989e-01 3.37420732e-01 8.27160552e-02 -9.48679686e-01
2.56039277e-02 1.37615606e-01 -4.41521078e-01 -3.39460880e-01
5.20054281e-01 3.34165871e-01 -1.14138985e+00 1.75397408e+00
1.58963978e-01 9.33720171e-01 -2.00929627e-01 1.13297927e+00
6.47188067e-01 1.06335771e+00 4.61663544e-01 -1.09173015e-01
1.42371118e+00 -1.33442509e+00 -6.41247571e-01 -2.47160822e-01
5.53597569e-01 -4.17737663e-01 1.12109017e+00 6.18823804e-02
-1.50465631e+00 -6.54882133e-01 -6.60400569e-01 -5.73337436e-01
-2.05577746e-01 -6.22251965e-02 4.27935421e-01 1.02502361e-01
-1.36392903e+00 2.18613133e-01 -1.10433090e+00 -3.91375661e-01
3.73728633e-01 -6.20232224e-02 -5.08771718e-01 -3.19857180e-01
-1.22736526e+00 6.42612755e-01 2.39064232e-01 -1.47334456e-01
-1.51838577e+00 -8.22391212e-01 -1.18354023e+00 -4.70857024e-02
4.58865225e-01 -1.00178361e+00 1.60697377e+00 -1.21238184e+00
-1.26547718e+00 9.14759278e-01 -8.38216662e-01 -6.59365058e-01
3.28668118e-01 -2.77563602e-01 -3.25254738e-01 8.43978941e-01
1.73862278e-01 9.27698314e-01 1.13445914e+00 -1.28735805e+00
-7.75004327e-01 -1.67906255e-01 5.60850382e-01 2.36433357e-01
-1.60241574e-01 1.32217273e-01 -1.18748069e+00 -6.23360455e-01
-1.78922668e-01 -4.85882014e-01 -9.36028436e-02 3.68221432e-01
2.37139896e-01 -8.66344124e-02 1.20750391e+00 -8.19425404e-01
1.32362723e+00 -2.07883883e+00 1.52468607e-01 -3.38984519e-01
2.41987363e-01 -4.94322926e-02 -4.60905850e-01 3.92319202e-01
-2.17342943e-01 5.48180118e-02 1.56566069e-01 -4.91036981e-01
-1.92090347e-01 4.63667452e-01 -5.86061597e-01 3.28337282e-01
3.53183776e-01 1.17997777e+00 -1.28912091e+00 -7.37479627e-01
4.33521599e-01 5.92551172e-01 -7.83320069e-01 3.89779896e-01
-7.52593815e-01 4.66840714e-01 -4.97677982e-01 7.08370030e-01
2.25367576e-01 -9.38258350e-01 1.11218482e-01 -4.88553494e-01
-6.19254075e-02 3.50690186e-01 -6.67863727e-01 2.25268722e+00
-3.09902668e-01 8.53361607e-01 -1.52410462e-03 -9.18707013e-01
5.42390831e-02 7.16000378e-01 7.74953604e-01 -1.09396148e+00
-3.21405560e-01 -3.34123373e-01 -5.34419119e-01 -9.23442721e-01
6.12393081e-01 -2.82963552e-02 1.44973472e-01 4.71584439e-01
2.65129179e-01 1.55197531e-01 2.82457113e-01 6.81118190e-01
1.09270406e+00 7.61788309e-01 -5.63294366e-02 3.20594579e-01
4.14896518e-01 1.82559729e-01 2.59416729e-01 6.50315106e-01
-4.32822138e-01 7.93004274e-01 4.21881437e-01 -5.59308112e-01
-1.14317858e+00 -1.15726602e+00 3.53401989e-01 1.53073037e+00
4.04383779e-01 -8.77070308e-01 -5.33696711e-01 -6.57148361e-01
-1.48679882e-01 4.96135473e-01 -6.99101150e-01 6.14226982e-02
-8.92995119e-01 -2.88681476e-03 6.20377898e-01 7.97227800e-01
2.59332418e-01 -7.92221487e-01 -6.67375445e-01 7.19248131e-02
-6.05845034e-01 -1.67883945e+00 -7.53072798e-01 -3.48228902e-01
-7.68694043e-01 -1.11485338e+00 -6.01618350e-01 -7.69074142e-01
3.08666497e-01 6.77322209e-01 1.64030242e+00 1.65686861e-01
-7.77210370e-02 1.03180444e+00 -4.54530865e-01 2.23850772e-01
-9.29067880e-02 -4.81153041e-01 -2.77715415e-01 -4.90532331e-02
3.37399185e-01 -4.64342386e-01 -7.62755215e-01 3.66543770e-01
-1.19011140e+00 3.72627705e-01 2.90191919e-01 7.08009481e-01
6.11609221e-01 -2.74548888e-01 1.73282385e-01 -3.73154163e-01
2.52059158e-02 -7.14952350e-01 -3.98337930e-01 4.73370612e-01
2.01034788e-02 1.27584860e-02 3.26965541e-01 -4.92736906e-01
-1.16099918e+00 -3.04665565e-02 6.37546852e-02 -1.14370823e+00
-3.09130549e-01 4.65206861e-01 1.41669407e-01 2.37502262e-01
2.45710388e-01 4.65316862e-01 -1.80020139e-01 -2.49701634e-01
7.38600016e-01 -7.32120425e-02 7.83519864e-01 -8.18099976e-01
6.25887871e-01 8.71360779e-01 -3.36474717e-01 -7.78311789e-01
-9.28107798e-01 -7.25225806e-01 -7.53447533e-01 -3.19414735e-01
1.31809354e+00 -1.39795566e+00 -7.38234997e-01 1.39000595e-01
-1.36051416e+00 -7.07063317e-01 -2.40291938e-01 2.78075606e-01
-8.80408287e-01 5.63063145e-01 -8.43314886e-01 -5.32343268e-01
2.97599256e-01 -8.38983059e-01 1.61030495e+00 -5.57866432e-02
-1.68515965e-01 -1.17396450e+00 -3.17165852e-02 7.04520285e-01
1.38543993e-01 -8.08817297e-02 7.39264131e-01 7.88425356e-02
-1.31248951e+00 2.40878537e-01 -5.33033133e-01 -4.38855439e-02
-1.72784537e-01 -1.41397323e-02 -8.64586949e-01 -1.65925682e-01
1.96319576e-02 -6.43523395e-01 9.49060917e-01 5.10516822e-01
1.48087966e+00 -2.13192791e-01 -2.62542635e-01 6.08977497e-01
1.32097244e+00 9.42279026e-02 7.35663474e-01 2.85331964e-01
8.93288434e-01 3.50231528e-01 6.90229714e-01 4.15185750e-01
1.01739120e+00 6.50961399e-01 4.67663497e-01 -2.20204629e-02
-2.44221702e-01 -5.37936151e-01 7.16689348e-01 5.30955970e-01
-1.40058607e-01 -3.22863936e-01 -8.63265336e-01 8.92522156e-01
-2.22562385e+00 -1.57243943e+00 1.90645114e-01 1.67808485e+00
7.24847496e-01 -2.11968794e-01 2.13581771e-01 -3.20515007e-01
1.71389371e-01 6.41937017e-01 -4.32408780e-01 7.92758614e-02
-9.81202796e-02 -2.26122662e-01 2.85012454e-01 6.10602558e-01
-1.18886817e+00 1.13760233e+00 6.98144007e+00 5.34664571e-01
-1.05348074e+00 3.26425880e-01 6.48495018e-01 -4.62557942e-01
-5.47136068e-01 1.73946843e-01 -3.64811867e-01 2.04377383e-01
1.11626422e+00 1.80625454e-01 3.68641019e-01 4.70577329e-01
6.09744132e-01 -9.31268185e-02 -1.40505064e+00 1.12850273e+00
2.13562742e-01 -1.81232381e+00 2.43796319e-01 -1.99971706e-01
6.69268906e-01 1.16897672e-01 -6.17579818e-02 1.87173858e-01
2.09202170e-01 -1.30595577e+00 1.20901501e+00 7.76495695e-01
7.42500305e-01 -1.32886305e-01 -1.97464637e-02 3.99011940e-01
-1.67891872e+00 -1.45118907e-01 9.28522274e-02 -1.50295153e-01
7.18256414e-01 8.02113563e-02 -2.51405120e-01 4.13074046e-01
9.93354976e-01 1.25173807e+00 -4.30676281e-01 4.95858431e-01
2.73635183e-02 6.20552301e-01 -1.51994914e-01 6.08203471e-01
5.48325777e-01 7.67413825e-02 2.45737597e-01 1.21554303e+00
1.48836464e-01 3.96351129e-01 3.09677720e-01 7.26373136e-01
-2.91214082e-02 -3.99161488e-01 -5.47512829e-01 -4.41219062e-01
2.09271416e-01 8.20938349e-01 -3.02252769e-01 -6.40022457e-01
-1.10860777e+00 1.23794949e+00 2.56052941e-01 7.76384950e-01
-1.17973959e+00 5.52975535e-01 8.36446881e-01 3.54790539e-01
4.68871027e-01 -6.85888171e-01 1.20176136e-01 -1.65305662e+00
7.38451332e-02 -8.97924304e-01 8.05016279e-01 -1.30626273e+00
-1.10443079e+00 3.79088670e-01 9.06117707e-02 -1.13552809e+00
-5.86080194e-01 -6.56270444e-01 -2.56448150e-01 7.01251924e-01
-1.66173065e+00 -1.55065095e+00 -3.52931410e-01 1.12591636e+00
9.95802402e-01 2.36886472e-01 5.52287340e-01 4.86879051e-01
-1.83109403e-01 1.66275442e-01 -3.48675668e-01 2.15647623e-01
7.23389626e-01 -1.08534610e+00 2.41150722e-01 9.76113081e-01
4.86106277e-01 6.02400184e-01 4.74649012e-01 -5.04089057e-01
-1.70371866e+00 -1.15140212e+00 1.02643692e+00 -8.58372390e-01
9.59834993e-01 -2.55045474e-01 -8.79934788e-01 1.12476432e+00
4.69389290e-01 3.12473565e-01 4.37476724e-01 -1.44136786e-01
-8.03435981e-01 2.01402307e-01 -5.63422084e-01 6.28229141e-01
1.06514263e+00 -1.42906749e+00 -6.96485996e-01 2.20402867e-01
9.62917864e-01 -4.19848233e-01 -8.65474641e-01 2.81173795e-01
6.96152210e-01 -9.61676180e-01 1.36603117e+00 -9.11388993e-01
6.93995535e-01 -5.26177406e-01 -3.77665758e-01 -5.91274559e-01
-2.76802212e-01 -4.09230918e-01 -7.53755450e-01 7.83965111e-01
1.28431441e-02 3.04290891e-01 1.03657877e+00 5.68728566e-01
2.98257284e-02 -5.79665065e-01 -5.93877137e-01 -5.16746819e-01
-9.72273946e-02 -7.76927292e-01 1.68109424e-02 9.64634836e-01
1.12937978e-02 3.59506786e-01 -6.49837434e-01 3.23676407e-01
3.20183814e-01 1.58004239e-01 5.35529613e-01 -5.61813533e-01
-4.87881631e-01 -2.41989702e-01 -2.66480505e-01 -1.88669050e+00
2.54102767e-01 -3.12457025e-01 5.49305677e-02 -1.62457371e+00
3.46791089e-01 2.82018811e-01 -1.74735829e-01 5.23944557e-01
1.48755973e-02 3.03369105e-01 1.44239992e-01 2.04702884e-01
-1.23980105e+00 4.25462216e-01 1.38938594e+00 -1.80482790e-01
1.73947871e-01 -6.68067634e-01 -2.04645619e-01 7.32029557e-01
1.21893853e-01 -1.05272390e-01 -9.24148560e-01 -1.09313560e+00
8.54467526e-02 5.85515618e-01 8.18402648e-01 -7.48661816e-01
3.87866735e-01 -4.74705517e-01 4.13566738e-01 -7.55812705e-01
6.86366498e-01 -7.85734057e-01 4.89541441e-02 -5.95919341e-02
-5.05362272e-01 3.21442246e-01 1.61281347e-01 9.04522121e-01
-6.68460190e-01 4.04282272e-01 5.19980371e-01 -3.77395481e-01
-1.37351584e+00 7.59229660e-01 -5.17024994e-01 7.93228447e-02
1.02269137e+00 -3.15840065e-01 -3.62298012e-01 -8.21008801e-01
-9.58129406e-01 3.63771111e-01 6.16013348e-01 7.66339242e-01
7.97057807e-01 -1.27627158e+00 -3.01408708e-01 -7.61888400e-02
2.09927410e-01 -1.58267155e-01 6.83535695e-01 9.56444860e-01
-4.54034090e-01 7.24971831e-01 3.68661694e-02 -9.25753474e-01
-1.16016996e+00 7.90704012e-01 4.25019026e-01 -2.68515080e-01
-6.18217289e-01 9.25154686e-01 7.25657761e-01 7.75609761e-02
4.40491557e-01 -4.45880681e-01 -6.19867332e-02 -1.11320816e-01
7.23718882e-01 -7.51672834e-02 -3.29017609e-01 -9.74211037e-01
-3.02000254e-01 5.54588795e-01 1.42664880e-01 -3.84207785e-01
1.02712524e+00 -6.20977759e-01 -3.17539722e-02 6.46971881e-01
1.39557755e+00 -4.33901250e-01 -1.75904465e+00 -4.57048118e-01
4.36404869e-02 -6.03385627e-01 -9.37030017e-02 -5.20316541e-01
-8.91340792e-01 1.18945551e+00 2.68969774e-01 5.83996885e-02
1.33285046e+00 2.66403049e-01 8.85989368e-01 1.41380414e-01
2.99998790e-01 -8.66142333e-01 7.15786159e-01 8.42193127e-01
6.48333848e-01 -1.27232909e+00 -3.68279785e-01 -2.52870589e-01
-7.18196929e-01 1.04741871e+00 7.63752162e-01 8.53065699e-02
6.85655355e-01 1.44786239e-01 9.34192687e-02 -3.01888973e-01
-1.39057100e+00 -2.38758132e-01 6.03610575e-01 4.56332952e-01
5.87499917e-01 -3.84319514e-01 3.38710010e-01 4.01400268e-01
4.54512358e-01 4.13924217e-01 8.24178010e-02 9.84023094e-01
-1.12816140e-01 -1.04182339e+00 -1.19084768e-01 -5.75134382e-02
-3.67812306e-01 -2.49718592e-01 -8.89575947e-03 6.96537614e-01
7.35192224e-02 9.18426991e-01 5.21763921e-01 -3.07678074e-01
-1.76975895e-02 -1.24280602e-02 7.56285667e-01 -4.10131961e-01
-1.23887986e-01 2.32725278e-01 1.45186692e-01 -1.32060587e+00
-9.41589653e-01 -6.22409105e-01 -1.30231738e+00 -2.29642451e-01
4.11792487e-01 1.36102542e-01 1.60623685e-01 1.19936419e+00
4.38856632e-01 5.22982001e-01 1.56156242e-01 -8.71283829e-01
-2.00994238e-02 -4.74925548e-01 -3.16963643e-01 6.58147693e-01
6.80220246e-01 -4.21591818e-01 2.56738663e-02 9.29301441e-01]
|
[10.023728370666504, 0.8418449759483337]
|
7b8d9fb0-87bf-4b67-a004-12cf073127f0
|
dummy-prototypical-networks-for-few-shot-open
|
2206.13691
| null |
https://arxiv.org/abs/2206.13691v1
|
https://arxiv.org/pdf/2206.13691v1.pdf
|
Dummy Prototypical Networks for Few-Shot Open-Set Keyword Spotting
|
Keyword spotting is the task of detecting a keyword in streaming audio. Conventional keyword spotting targets predefined keywords classification, but there is growing attention in few-shot (query-by-example) keyword spotting, e.g., N-way classification given M-shot support samples. Moreover, in real-world scenarios, there can be utterances from unexpected categories (open-set) which need to be rejected rather than classified as one of the N classes. Combining the two needs, we tackle few-shot open-set keyword spotting with a new benchmark setting, named splitGSC. We propose episode-known dummy prototypes based on metric learning to detect an open-set better and introduce a simple and powerful approach, Dummy Prototypical Networks (D-ProtoNets). Our D-ProtoNets shows clear margins compared to recent few-shot open-set recognition (FSOSR) approaches in the suggested splitGSC. We also verify our method on a standard benchmark, miniImageNet, and D-ProtoNets shows the state-of-the-art open-set detection rate in FSOSR.
|
['Simyung Chang', 'Inseop Chung', 'Seunghan Yang', 'Byeonggeun Kim']
|
2022-06-28
| null | null | null | null |
['open-set-learning', 'keyword-spotting']
|
['miscellaneous', 'speech']
|
[ 5.65063596e-01 1.11420423e-01 -8.05531889e-02 -1.84375376e-01
-1.25883925e+00 -4.04119283e-01 4.62026924e-01 3.01821589e-01
-4.24173772e-01 4.30144995e-01 1.33381197e-02 -3.94283384e-02
-4.18336689e-01 -3.86563599e-01 -7.76640236e-01 -5.71132123e-01
-3.83387893e-01 5.74933529e-01 8.47470641e-01 -3.56399804e-01
3.88701707e-01 2.08603233e-01 -1.98771834e+00 5.52378714e-01
4.26992536e-01 1.13504219e+00 3.44514310e-01 7.02688336e-01
-1.99815333e-01 2.19605520e-01 -8.03276896e-01 -1.53142750e-01
2.47072101e-01 -2.52107054e-01 -6.85090423e-01 9.26124752e-02
2.55907893e-01 1.94563344e-01 -2.41881728e-01 9.97621059e-01
8.41788054e-01 2.71189988e-01 6.25081718e-01 -1.43331623e+00
-1.13108650e-01 1.16238332e+00 -6.03500605e-01 4.50980246e-01
7.29891837e-01 1.45781398e-01 1.26666844e+00 -1.51075816e+00
6.18507504e-01 1.23969913e+00 6.55194759e-01 4.26348537e-01
-8.90497208e-01 -8.59086752e-01 2.23667249e-01 6.77931547e-01
-1.92439723e+00 -5.22551179e-01 4.89457875e-01 -2.35784382e-01
9.54941392e-01 5.81699431e-01 4.44415748e-01 1.39496338e+00
-3.12490523e-01 8.50109339e-01 8.74141932e-01 -5.56528270e-01
5.28292120e-01 2.67761141e-01 2.76221305e-01 1.95053980e-01
-5.44621162e-02 9.52920094e-02 -1.01353133e+00 -3.69961679e-01
-1.18995748e-01 -8.54301229e-02 -7.75723636e-01 -1.11969039e-01
-1.30534220e+00 5.87838352e-01 2.93957423e-02 3.93015981e-01
-1.52520582e-01 -2.78353453e-01 6.58046007e-01 5.01827478e-01
3.89032423e-01 5.60970008e-01 -4.38758820e-01 -4.09793735e-01
-1.42378938e+00 2.30245322e-01 9.72340047e-01 1.08925307e+00
4.41881120e-01 -4.92155738e-02 -6.29949391e-01 1.08735800e+00
-7.15071112e-02 2.05171376e-01 7.74210632e-01 -2.56529063e-01
4.64610338e-01 3.20417732e-01 4.33473252e-02 -8.72129083e-01
-3.67964208e-01 -5.41517913e-01 -5.00095189e-01 -3.80514115e-01
-5.91118075e-02 2.70690262e-01 -1.02814686e+00 1.24998915e+00
2.71828830e-01 4.74305362e-01 5.78471040e-03 9.87044632e-01
1.17701471e+00 9.08154190e-01 -5.78610361e-01 -5.34529328e-01
1.43766952e+00 -8.34456980e-01 -6.94692731e-01 -2.16885302e-02
6.22466862e-01 -5.68224609e-01 1.22627127e+00 8.30714881e-01
-4.21743929e-01 -2.08784953e-01 -1.08563924e+00 6.24389172e-01
-7.26372480e-01 -2.13901684e-01 -1.98324900e-02 6.19726360e-01
-6.63457572e-01 5.09209275e-01 -1.72536135e-01 -6.12154841e-01
3.53010565e-01 1.16771743e-01 -2.30159596e-01 3.90158482e-02
-1.39831460e+00 3.44126672e-01 6.45672321e-01 -3.49647194e-01
-1.32038605e+00 -6.56743050e-01 -5.35796642e-01 2.16684788e-01
1.27431428e+00 1.59777373e-01 1.33236790e+00 -4.49736416e-01
-1.12813759e+00 7.29232371e-01 1.88633054e-02 -7.97470152e-01
4.17653501e-01 -1.16383359e-01 -8.52232575e-01 5.19111335e-01
1.00074902e-01 6.58204317e-01 1.15186608e+00 -1.17172587e+00
-7.77458429e-01 -5.86051531e-02 -1.39668643e-01 5.09151705e-02
-5.61057448e-01 3.83662850e-01 -5.14903069e-01 -9.28966463e-01
2.15815321e-01 -5.33586860e-01 1.28447652e-01 -8.32088739e-02
-1.05831659e+00 -6.83688104e-01 1.09983599e+00 -2.35578477e-01
1.76307988e+00 -2.26271105e+00 -3.07389200e-01 2.68493950e-01
3.62191796e-01 3.96092296e-01 -4.81413305e-02 3.86464536e-01
-2.71201223e-01 1.36297509e-01 -2.36207798e-01 -1.85790256e-01
1.06274299e-01 1.41737893e-01 -6.05588675e-01 4.92429107e-01
-6.72389120e-02 4.89113331e-01 -8.35983276e-01 -6.51158273e-01
-2.03843161e-01 -1.21318594e-01 -3.49376798e-01 -4.44025286e-02
-2.66717374e-01 -1.29024327e-01 -1.23055829e-02 9.42640245e-01
4.25168306e-01 -8.49094540e-02 -3.81298751e-01 -9.68144089e-03
-5.93158649e-03 1.06039636e-01 -1.55060053e+00 1.35225761e+00
4.52162661e-02 4.85819042e-01 -2.13097841e-01 -1.28177118e+00
9.56670344e-01 5.64592183e-01 8.89521241e-02 -4.26003695e-01
3.25053209e-03 5.36418736e-01 6.57073706e-02 -4.36853439e-01
6.64322555e-01 6.74991459e-02 -1.71696603e-01 2.68636376e-01
4.49083686e-01 4.91456501e-02 2.31526211e-01 4.04658645e-01
1.29053986e+00 -4.48216647e-01 4.37735528e-01 -1.51657805e-01
2.01695338e-01 -2.92770624e-01 5.05419850e-01 1.50510812e+00
-2.57055998e-01 1.06524968e+00 3.37731659e-01 -6.87865317e-02
-6.05726719e-01 -5.91094792e-01 -2.04536840e-01 1.21903503e+00
1.40397117e-01 -7.94459939e-01 -5.06573260e-01 -7.37659216e-01
5.49280979e-02 7.72294164e-01 -3.70876998e-01 -2.07963988e-01
-1.25836208e-01 -3.34791511e-01 1.02547193e+00 -6.05115145e-02
-2.14284450e-01 -1.17394507e+00 -9.03418720e-01 2.28917167e-01
-1.31272689e-01 -1.26446211e+00 -5.66544831e-01 8.35287213e-01
-1.92297399e-01 -1.03771281e+00 -1.07861650e+00 -6.74113572e-01
8.93915743e-02 4.11459833e-01 9.51437056e-01 -3.18150073e-01
-5.52439690e-01 5.04661083e-01 -9.84014332e-01 -6.20298028e-01
-3.10974985e-01 -8.59017819e-02 5.66917777e-01 4.39004838e-01
3.34166616e-01 -3.58845413e-01 -4.44032848e-01 7.69225836e-01
-7.77035534e-01 -3.04406285e-01 3.33445847e-01 8.45757663e-01
6.83868587e-01 1.60673127e-01 8.32349002e-01 -6.68876410e-01
6.88750148e-01 -6.79081500e-01 -2.31896967e-01 3.43979865e-01
-7.64537752e-01 -3.21303755e-01 4.48786139e-01 -1.03276670e+00
-3.00760806e-01 -2.61695772e-01 4.05938514e-02 -9.79927838e-01
-2.31883332e-01 4.20193076e-01 -2.95609515e-02 -5.78845963e-02
7.90706336e-01 5.41306913e-01 -3.32534909e-01 -6.44078255e-01
2.06197396e-01 1.18399549e+00 6.97233796e-01 -1.04690582e-01
7.34452546e-01 3.34403783e-01 -5.57559609e-01 -1.18687415e+00
-8.06347668e-01 -1.22366571e+00 -1.71975404e-01 -4.22348142e-01
3.12678516e-01 -9.32132244e-01 -2.75748700e-01 2.78194338e-01
-1.06553805e+00 1.72630131e-01 -6.68000698e-01 3.55356544e-01
-5.03972888e-01 3.82051289e-01 -1.13430023e-01 -1.06788886e+00
-5.11585295e-01 -1.00097573e+00 1.49570715e+00 5.45776524e-02
-5.26150763e-01 -1.09463833e-01 -1.52589664e-01 -1.28232315e-01
1.01068921e-01 -2.17966661e-01 5.46835244e-01 -1.78084791e+00
-2.28195325e-01 -4.41483200e-01 9.77901146e-02 2.10996240e-01
-3.07671517e-01 -3.06590736e-01 -1.02998459e+00 -2.54048288e-01
-4.82186116e-02 -5.25709808e-01 1.12647474e+00 6.46817759e-02
1.58144009e+00 -4.49608535e-01 -5.73041439e-01 4.48713332e-01
9.52165127e-01 2.76556373e-01 4.94014502e-01 8.06754231e-02
3.24860543e-01 2.19434157e-01 1.13741195e+00 9.15303409e-01
-1.82726324e-01 9.28890765e-01 2.43668988e-01 1.44488096e-01
1.05533727e-01 -1.54509872e-01 2.93616474e-01 8.40461016e-01
5.59091568e-01 -6.87989235e-01 -9.75864470e-01 7.10578442e-01
-1.68746412e+00 -9.42174554e-01 1.26547828e-01 2.09102845e+00
9.09105361e-01 6.41835511e-01 1.31808147e-01 9.71846223e-01
9.94619608e-01 2.32923508e-01 -4.66201127e-01 -1.31925598e-01
-4.32313949e-01 3.14226925e-01 2.29341179e-01 -2.13744953e-01
-1.11337471e+00 7.00832307e-01 5.22201633e+00 1.72446287e+00
-1.03315759e+00 4.03305560e-01 4.79315162e-01 -4.72745478e-01
-6.81702495e-02 -3.40168029e-02 -1.25829685e+00 5.39489865e-01
9.58887279e-01 -1.74978837e-01 1.56399403e-02 8.05765331e-01
1.42432243e-01 -2.11593360e-01 -1.29525363e+00 1.61243188e+00
5.44345975e-01 -1.33136332e+00 -9.95655265e-03 -1.96522653e-01
3.19313854e-01 6.11939318e-02 -2.74254799e-01 7.61967957e-01
-3.88216048e-01 -6.77213967e-01 1.08092272e+00 4.84445006e-01
1.09318364e+00 -6.18572116e-01 5.55519879e-01 6.41601562e-01
-1.06332850e+00 -3.36587578e-01 -3.53666961e-01 2.67067730e-01
2.86185239e-02 8.07143807e-01 -1.17817008e+00 4.07882571e-01
1.03575170e+00 5.66521823e-01 -5.45788050e-01 1.63197529e+00
3.05297583e-01 1.06072986e+00 -6.82146430e-01 -4.55385685e-01
1.95275709e-01 4.73347396e-01 1.37002492e+00 1.38086367e+00
3.53592485e-01 -2.31585167e-02 3.98230314e-01 6.24748886e-01
-5.43729588e-02 2.61072576e-01 -6.46669865e-01 -1.90121695e-01
6.60541415e-01 1.06871855e+00 -1.12186623e+00 -4.78766352e-01
-1.70805499e-01 9.05584872e-01 -2.94274837e-01 9.08793733e-02
-8.17452729e-01 -1.03685462e+00 1.31810859e-01 2.05057427e-01
7.99664915e-01 5.56925356e-01 2.92030364e-01 -9.06925023e-01
1.20735914e-01 -1.12316978e+00 6.06440246e-01 -6.03217661e-01
-1.19148636e+00 5.36672473e-01 1.53534472e-01 -1.73644030e+00
5.44107854e-02 -2.64761686e-01 -6.46285653e-01 2.46511936e-01
-1.11340225e+00 -2.62536943e-01 -1.11239426e-01 4.94628787e-01
1.11251163e+00 -3.89216244e-01 7.51885176e-01 4.09276217e-01
-2.97221273e-01 1.05059063e+00 8.52298588e-02 -2.16827303e-01
7.75577605e-01 -9.75843489e-01 4.63960588e-01 7.04386532e-01
6.78453684e-01 1.19680606e-01 9.14060950e-01 -5.51054001e-01
-1.21151650e+00 -1.18833792e+00 7.35222697e-01 -1.45710744e-02
8.95482421e-01 -9.27207768e-01 -1.04884529e+00 -1.23955190e-01
-3.07501405e-01 3.53593439e-01 5.48698366e-01 -1.59736991e-01
-4.33467299e-01 -1.07554778e-01 -9.42413568e-01 3.25266421e-01
1.12658501e+00 -6.55312836e-01 -6.86010897e-01 6.83015049e-01
1.23832214e+00 -2.14037716e-01 -3.14246655e-01 6.77986085e-01
3.43912512e-01 -6.30738378e-01 8.83458257e-01 -4.91099119e-01
-3.86044309e-02 -2.39153206e-01 -5.45525074e-01 -1.08801687e+00
2.59362847e-01 -1.00858343e+00 -4.80857074e-01 1.22317398e+00
5.87763131e-01 -2.57642508e-01 4.75280762e-01 -4.75637555e-01
-2.67250597e-01 -1.11633801e+00 -1.45853877e+00 -1.28944588e+00
-6.61024034e-01 -8.56807649e-01 5.14926076e-01 8.12731802e-01
2.64373690e-01 2.73903102e-01 -6.31312251e-01 1.51019990e-01
4.17374343e-01 -1.05759449e-01 5.29452443e-01 -1.32414389e+00
-5.36583781e-01 -2.26713344e-01 -6.47739947e-01 -1.08205247e+00
-2.56701589e-01 -9.10644948e-01 5.60603797e-01 -1.06399786e+00
1.46117032e-01 -2.58755773e-01 -4.62247580e-01 3.96712780e-01
2.19955713e-01 2.63102919e-01 1.52875081e-01 1.26349926e-01
-1.32658160e+00 6.50377154e-01 5.36746621e-01 -3.37791622e-01
-2.83662736e-01 3.18423182e-01 -4.26409245e-01 6.51789129e-01
4.15496111e-01 -7.69770622e-01 -1.71179965e-01 5.39303064e-01
2.69381046e-01 1.96299240e-01 3.53080690e-01 -1.30853307e+00
3.38550329e-01 9.40940157e-02 -2.48879984e-01 -1.19400656e+00
3.20902318e-01 -5.90810001e-01 -3.62971723e-01 2.60580540e-01
-4.75626856e-01 -4.14568067e-01 -1.23012923e-01 9.30855751e-01
-2.06349447e-01 -6.45370483e-01 5.96149087e-01 -2.38949303e-02
-9.73902225e-01 1.39984772e-01 -3.50748599e-01 3.47348541e-01
1.12323499e+00 -6.31256878e-01 -7.81649575e-02 -3.69554609e-01
-8.77880156e-01 3.69036824e-01 -4.01835114e-01 6.30787373e-01
1.14432895e+00 -1.03180492e+00 -5.35219848e-01 3.05007603e-02
8.78544867e-01 8.05794075e-03 -2.62147579e-02 9.91117716e-01
1.13879524e-01 4.32700455e-01 6.96893692e-01 -1.00194407e+00
-1.55456209e+00 4.97101635e-01 1.66792244e-01 1.39466688e-01
-9.39637303e-01 1.30058646e+00 1.22213952e-01 -2.85856783e-01
1.13215780e+00 -3.81600142e-01 -2.65191495e-01 4.38625038e-01
7.58245945e-01 2.14930013e-01 4.00734872e-01 -3.91135573e-01
-5.33634603e-01 1.33706331e-01 -1.88554674e-01 5.33623807e-02
1.15274405e+00 3.43883550e-03 3.35736960e-01 1.08516490e+00
1.12162077e+00 -3.64513814e-01 -6.32158875e-01 -6.24142170e-01
2.03024834e-01 -2.68061399e-01 -1.35834813e-01 -8.03310215e-01
-4.73272234e-01 6.61419094e-01 1.09682834e+00 4.02471721e-01
9.08279300e-01 5.26453495e-01 6.99739456e-01 8.47620249e-01
6.46741629e-01 -1.35405695e+00 2.53608316e-01 6.34016991e-01
1.08016443e+00 -1.09286296e+00 -1.18402965e-01 -3.65743428e-01
-4.89961058e-01 8.57519686e-01 3.50962698e-01 2.52615005e-01
9.22678173e-01 2.45535776e-01 -3.76381516e-01 -4.33697432e-01
-1.06689823e+00 -2.43523344e-01 2.75679827e-01 4.04115111e-01
-3.65778625e-01 1.92211449e-01 -1.59575447e-01 9.91914809e-01
-2.86303639e-01 -2.24586174e-01 5.26171148e-01 9.26543713e-01
-8.42035770e-01 -3.87830913e-01 -5.22881746e-01 1.04013455e+00
-3.73737752e-01 -3.24300349e-01 -4.00763214e-01 5.26170909e-01
6.16818815e-02 8.85669053e-01 4.34999950e-02 -7.33132660e-01
3.46132249e-01 3.88686359e-01 -3.12861875e-02 -9.36067462e-01
-4.08991069e-01 2.71779150e-01 9.90608484e-02 -4.83749717e-01
-1.60223603e-01 -4.74721283e-01 -9.98026192e-01 5.49219489e-01
-1.08444035e+00 3.09418708e-01 3.93330902e-01 1.00691330e+00
2.85162985e-01 5.11725962e-01 7.92051792e-01 -7.34359384e-01
-8.96471620e-01 -1.21854174e+00 -8.19584966e-01 2.75573313e-01
2.88916558e-01 -9.17760849e-01 -7.76894808e-01 -3.05643827e-01]
|
[14.17634391784668, 6.199252128601074]
|
bc002dbe-f286-4515-ab29-e71028c26395
|
head2headfs-video-based-head-reenactment-with
|
2103.16229
| null |
https://arxiv.org/abs/2103.16229v1
|
https://arxiv.org/pdf/2103.16229v1.pdf
|
Head2HeadFS: Video-based Head Reenactment with Few-shot Learning
|
Over the past years, a substantial amount of work has been done on the problem of facial reenactment, with the solutions coming mainly from the graphics community. Head reenactment is an even more challenging task, which aims at transferring not only the facial expression, but also the entire head pose from a source person to a target. Current approaches either train person-specific systems, or use facial landmarks to model human heads, a representation that might transfer unwanted identity attributes from the source to the target. We propose head2headFS, a novel easily adaptable pipeline for head reenactment. We condition synthesis of the target person on dense 3D face shape information from the source, which enables high quality expression and pose transfer. Our video-based rendering network is fine-tuned under a few-shot learning strategy, using only a few samples. This allows for fast adaptation of a generic generator trained on a multiple-person dataset, into a person-specific one.
|
['Stefanos Zafeiriou', 'Viktoriia Sharmanska', 'Mohammad Rami Koujan', 'Michail Christos Doukas']
|
2021-03-30
| null | null | null | null |
['pose-transfer']
|
['computer-vision']
|
[ 2.27610067e-01 4.39396381e-01 2.23496944e-01 -7.65000820e-01
-8.94715488e-01 -3.90214235e-01 6.99711084e-01 -6.21297419e-01
-8.48586857e-02 5.63134611e-01 3.36891681e-01 5.65036535e-01
6.70050561e-01 -4.71632928e-01 -6.14677250e-01 -6.29961967e-01
2.24518493e-01 8.57723057e-01 3.24036032e-02 -4.18131799e-01
-2.87797987e-01 7.36947358e-01 -2.02935386e+00 1.18354104e-01
3.71061921e-01 8.56633544e-01 -2.14513958e-01 7.04515636e-01
1.67057533e-02 3.34735811e-01 -6.84351027e-01 -8.96574378e-01
2.33089402e-01 -6.34960055e-01 -5.98622024e-01 3.54067713e-01
1.02358305e+00 -2.96519011e-01 -6.59373328e-02 7.08215594e-01
8.81154895e-01 1.72913551e-01 4.68352199e-01 -1.59243846e+00
-2.78173000e-01 1.96867213e-01 -5.99969864e-01 -6.17031753e-01
7.46341586e-01 1.80239201e-01 5.76866627e-01 -1.02019441e+00
1.02210379e+00 1.57988715e+00 7.54707932e-01 1.18031430e+00
-1.12349319e+00 -9.52417552e-01 -7.95294866e-02 -4.19426076e-02
-1.46199608e+00 -1.00549364e+00 7.84012377e-01 -4.31189001e-01
4.18678433e-01 2.26601988e-01 1.06465828e+00 1.50985730e+00
-1.69609874e-01 5.46804070e-01 1.10887158e+00 -3.63014251e-01
5.50414696e-02 2.90540963e-01 -3.90679568e-01 7.40302026e-01
-2.76530862e-01 3.50536108e-02 -7.56194890e-01 -1.88882753e-01
9.39043283e-01 -3.04179579e-01 -5.02695918e-01 -5.81818104e-01
-7.34024286e-01 6.14392698e-01 2.41560280e-01 7.31005967e-02
-1.44925177e-01 1.19531393e-01 3.91154617e-01 1.85000211e-01
6.12896502e-01 1.38020232e-01 -2.77961761e-01 -9.40481052e-02
-1.14167213e+00 6.41041219e-01 8.19845974e-01 1.23662126e+00
8.32305908e-01 2.93246470e-02 -4.84617442e-01 7.17259228e-01
1.72934338e-01 4.36396450e-01 2.84478664e-01 -1.09240091e+00
2.43745781e-02 2.65356004e-01 1.11920282e-01 -7.01066136e-01
-2.42144942e-01 -3.99919413e-02 -5.76277316e-01 5.15425503e-01
4.46618557e-01 -3.18907648e-01 -1.03123832e+00 2.10206652e+00
8.29448462e-01 3.14110368e-01 -2.98639566e-01 9.58321095e-01
8.42222035e-01 4.07897383e-01 -1.96834374e-03 4.54410091e-02
1.43362093e+00 -8.61726761e-01 -5.96794844e-01 -2.46531278e-01
3.79821748e-01 -8.07650089e-01 1.09875739e+00 1.84132501e-01
-1.29580748e+00 -5.34281909e-01 -7.37242520e-01 -4.05741543e-01
-1.57416210e-01 6.29729182e-02 4.82700437e-01 9.19645846e-01
-1.26604998e+00 5.76730847e-01 -5.39355636e-01 -7.22149193e-01
5.38813770e-01 5.71559489e-01 -8.24728191e-01 -6.47780597e-02
-9.39068198e-01 9.57321882e-01 -1.72057435e-01 5.87271303e-02
-8.33221734e-01 -8.15831184e-01 -1.10793543e+00 -1.68351486e-01
5.37285581e-02 -1.10242224e+00 1.36138022e+00 -1.52190948e+00
-2.14837575e+00 1.47783303e+00 -2.79086560e-01 -5.23085520e-03
9.00232732e-01 -1.38466567e-01 -2.02778757e-01 -1.02549516e-01
-7.86861312e-03 1.01185429e+00 1.44944596e+00 -1.25008035e+00
-2.78572381e-01 -6.76682353e-01 -2.74025321e-01 1.87335700e-01
-1.37262851e-01 3.77278775e-01 -8.38417411e-01 -6.70954287e-01
-4.64661032e-01 -1.16362965e+00 9.37661380e-02 5.70380211e-01
-3.83852243e-01 -1.68527156e-01 9.11712646e-01 -7.19865680e-01
5.10514140e-01 -2.07844806e+00 4.28257138e-01 1.60283089e-01
-3.06153949e-02 1.53156757e-01 -2.47792050e-01 1.23464361e-01
-5.20365715e-01 -4.05960172e-01 -2.18690723e-01 -1.10456574e+00
1.20977880e-02 -5.72558492e-03 -2.93703556e-01 6.79264545e-01
2.18218029e-01 8.01339149e-01 -7.61510491e-01 -5.48138201e-01
-1.92503110e-02 8.84888530e-01 -6.59508049e-01 7.07777917e-01
-9.59404409e-02 9.29924607e-01 -3.28331217e-02 4.82982606e-01
8.05259347e-01 2.11150512e-01 -9.23958421e-02 -2.32309833e-01
8.61129090e-02 -2.45461449e-01 -1.02534425e+00 2.08575201e+00
-5.91361344e-01 4.08760041e-01 4.36377972e-01 -1.34031415e-01
1.00280523e+00 5.32109141e-01 3.56232315e-01 -1.88846990e-01
4.82140839e-01 1.39341369e-01 -2.99701154e-01 -2.90663987e-01
3.99185807e-01 -5.65912068e-01 -1.41877115e-01 5.22008061e-01
4.25072670e-01 -4.25041318e-01 -2.53174871e-01 -3.56993359e-03
6.41578376e-01 6.70083344e-01 1.43512478e-02 -1.04320729e-02
4.94219601e-01 -4.90934819e-01 4.54643428e-01 4.78354618e-02
-4.25191037e-03 1.24319232e+00 2.33570769e-01 -2.63761669e-01
-1.13549972e+00 -9.78884578e-01 2.25871176e-01 1.34618711e+00
-3.20692658e-01 -4.16240096e-01 -1.32181180e+00 -4.94670779e-01
-7.87028149e-02 6.70679927e-01 -9.18124259e-01 -2.03854144e-01
-7.06284344e-01 -3.45407844e-01 7.03800976e-01 3.62459540e-01
2.56815165e-01 -1.01213145e+00 -5.88097513e-01 -1.00667849e-02
-4.93519641e-02 -1.02199745e+00 -8.43823135e-01 -4.12658572e-01
-3.59753698e-01 -8.36223185e-01 -1.35779786e+00 -6.95739150e-01
8.37561846e-01 -2.52490938e-01 1.14346135e+00 1.86288089e-01
-4.20509338e-01 5.11271119e-01 -9.59364697e-02 -5.54465950e-01
-5.09640455e-01 3.51015069e-02 1.76562011e-01 6.81672394e-01
1.52291760e-01 -6.09514534e-01 -3.84194344e-01 2.84869552e-01
-4.34784323e-01 1.57463014e-01 5.58193438e-02 4.97331411e-01
2.29997382e-01 -7.29531348e-01 2.96168625e-01 -9.84582007e-01
4.50605720e-01 -7.11655840e-02 -3.93974334e-01 1.75597612e-02
-1.43419281e-02 -1.76339597e-01 5.76139033e-01 -5.03906071e-01
-1.28122103e+00 4.25527602e-01 -4.72497195e-01 -7.82998264e-01
-3.27712357e-01 -4.89595085e-01 -6.75940454e-01 -1.85432017e-01
7.61581779e-01 -3.38518210e-02 2.87616700e-01 -5.02328396e-01
7.29548037e-01 5.22572637e-01 9.31116760e-01 -7.27068901e-01
9.30425763e-01 5.85636675e-01 -9.04230699e-02 -8.36650908e-01
-6.03811443e-01 -1.46673620e-01 -1.01224220e+00 -3.71113688e-01
8.03471088e-01 -8.93540561e-01 -6.11648560e-01 8.55296373e-01
-1.40486908e+00 -5.18287301e-01 -5.24375260e-01 -6.33955235e-03
-7.89525032e-01 -7.28275031e-02 -3.04338068e-01 -4.95220542e-01
-5.01972377e-01 -9.28869188e-01 1.65884674e+00 2.73443162e-01
-6.33308232e-01 -6.90547824e-01 2.89541274e-01 8.34378749e-02
2.88085699e-01 4.58464772e-01 5.09749115e-01 -1.96702197e-01
-2.05191642e-01 -3.92755121e-01 5.87231538e-04 6.84179887e-02
1.23852856e-01 2.10636288e-01 -1.41604435e+00 -4.72299039e-01
-8.25711712e-02 -4.79742587e-01 4.30055082e-01 1.14647120e-01
8.72405827e-01 -2.41223052e-01 -4.06248957e-01 9.46343482e-01
7.44920909e-01 -4.25050527e-01 7.64901221e-01 -1.04414627e-01
9.33508635e-01 1.15443039e+00 3.36227119e-01 5.35868585e-01
5.54066062e-01 1.08680999e+00 1.28420934e-01 -3.54579866e-01
-4.53292996e-01 -5.18476248e-01 4.97364342e-01 3.13208550e-01
-2.96166927e-01 3.89486440e-02 -5.58351338e-01 3.21040004e-01
-1.44459915e+00 -7.87001431e-01 3.41969013e-01 2.27633953e+00
9.04470265e-01 -4.16277885e-01 4.80307281e-01 -1.33434117e-01
7.68665969e-01 4.19268832e-02 -4.94430155e-01 -2.95798749e-01
4.97908965e-02 4.28663969e-01 -1.71878599e-02 4.77036983e-01
-7.91942477e-01 1.25065851e+00 6.28177977e+00 4.64722723e-01
-1.47183669e+00 1.31433845e-01 5.49896121e-01 -2.23352656e-01
-1.66310519e-01 -1.46182746e-01 -8.72975767e-01 1.97024941e-01
7.91884363e-01 -4.24462944e-01 1.64679691e-01 8.56927633e-01
4.74842414e-02 2.67778546e-01 -1.43910742e+00 1.30964005e+00
6.15346372e-01 -8.97562742e-01 1.16383076e-01 7.45324744e-03
4.75626111e-01 -3.95035714e-01 1.57839075e-01 2.96177626e-01
1.74028784e-01 -1.18343282e+00 9.44521785e-01 7.24009633e-01
1.31519961e+00 -8.27094078e-01 3.17202091e-01 1.28328040e-01
-1.09204769e+00 3.52277219e-01 -2.01257750e-01 2.61429846e-01
4.70126271e-01 1.44347921e-01 -9.80287254e-01 2.52937049e-01
6.79226816e-01 4.55729276e-01 -5.65524280e-01 9.16764319e-01
-3.82148862e-01 8.55366886e-02 -3.74584258e-01 4.51298624e-01
-4.48628098e-01 -1.26678273e-01 5.54698408e-01 1.14662743e+00
5.81247747e-01 6.85140416e-02 3.41387652e-02 7.70670772e-01
-2.28868604e-01 2.46135846e-01 -6.20988309e-01 4.55845118e-01
1.01443499e-01 1.63463712e+00 -2.34834567e-01 -2.76671976e-01
-7.09380955e-02 1.41324997e+00 3.72598231e-01 7.51943737e-02
-7.28384793e-01 -2.52727687e-01 8.66113126e-01 4.63572174e-01
4.81372997e-02 4.99570183e-02 2.23874003e-01 -1.12148345e+00
-1.93842441e-01 -1.05195653e+00 1.12342052e-01 -9.90705013e-01
-1.12014818e+00 9.52169001e-01 -2.33902093e-02 -1.04869390e+00
-6.98408127e-01 -3.93983454e-01 -8.02451611e-01 1.02503300e+00
-9.97544169e-01 -1.77789056e+00 -6.40851915e-01 8.97022665e-01
3.76958907e-01 -1.06486432e-01 1.11294734e+00 3.06653738e-01
-5.61670482e-01 9.50661063e-01 -7.87931025e-01 8.28497782e-02
1.15733564e+00 -9.44503725e-01 7.67122805e-01 4.77098018e-01
-5.07213734e-02 4.59355414e-01 9.98871326e-01 -3.46601784e-01
-1.21987951e+00 -1.16017973e+00 8.06564391e-01 -6.76190197e-01
1.31913438e-01 -8.97932053e-01 -9.75496590e-01 9.04342294e-01
1.85320750e-01 1.65778384e-01 6.25982523e-01 1.16249351e-02
-5.11890292e-01 -3.72653082e-02 -1.35470223e+00 8.69643331e-01
1.12562990e+00 -6.24843478e-01 -2.94115216e-01 1.43769100e-01
2.67803192e-01 -6.55207038e-01 -7.38146007e-01 1.08527556e-01
8.29055846e-01 -1.03815126e+00 7.82093883e-01 -5.83112121e-01
2.42674500e-02 -2.32586965e-01 1.85312063e-01 -1.52035415e+00
-1.24152027e-01 -1.06932640e+00 1.28551126e-01 1.47491181e+00
-3.85569520e-02 -3.46178919e-01 1.06053138e+00 9.81127799e-01
6.32502809e-02 -4.05405462e-01 -8.56989980e-01 -4.83089089e-01
1.20927855e-01 -1.38367832e-01 9.64841545e-01 8.40396881e-01
-2.82296658e-01 4.92568970e-01 -7.68987894e-01 -1.30599841e-01
6.30618751e-01 6.56638816e-02 1.51469445e+00 -1.31936657e+00
-2.80940980e-01 -1.93198979e-01 -4.12777394e-01 -8.79098177e-01
6.96636260e-01 -9.77256358e-01 1.31843612e-01 -1.00907588e+00
1.10959890e-03 -1.38271481e-01 4.83961701e-01 6.74211383e-01
-8.61054007e-03 4.71664727e-01 3.37600738e-01 1.99343101e-03
-1.91126630e-04 8.74362588e-01 1.29415989e+00 1.79014519e-01
-1.51845217e-01 1.12704121e-01 -5.33656895e-01 9.72470224e-01
3.73515129e-01 -3.30901474e-01 -2.62524694e-01 -3.04790348e-01
-1.63238883e-01 1.86359715e-02 5.36480546e-01 -9.52439666e-01
8.75109285e-02 2.01122612e-01 6.44520640e-01 -1.21673279e-01
9.43960190e-01 -6.29570425e-01 5.64198256e-01 -3.05814785e-03
3.78576946e-03 1.70389612e-04 1.75789401e-01 1.87581122e-01
2.74792574e-02 7.74845248e-03 1.19320416e+00 -1.02802143e-01
-5.72404861e-01 7.00993896e-01 5.53156808e-02 -1.02709141e-02
9.35321212e-01 -1.21521190e-01 1.45136535e-01 -7.12420464e-01
-1.00054967e+00 -1.68861851e-01 9.68814492e-01 6.04419112e-01
4.96277541e-01 -1.47504687e+00 -9.79668677e-01 5.91865420e-01
1.56756461e-01 -6.75946251e-02 1.54959545e-01 5.72984397e-01
-2.87894547e-01 -2.69468635e-01 -4.52138394e-01 -5.77376783e-01
-1.60929132e+00 2.66923070e-01 6.54220223e-01 3.25653464e-01
-7.94301867e-01 1.16851819e+00 3.90453786e-01 -5.59366047e-01
-2.06040610e-02 2.92954743e-01 -6.34432733e-02 2.85976559e-01
7.78580546e-01 2.51553059e-01 -1.24302367e-02 -1.43008840e+00
-2.91839957e-01 1.00471294e+00 9.80030820e-02 -4.14564043e-01
1.16633379e+00 1.78597437e-03 -9.14443284e-02 3.10434103e-01
1.44037151e+00 6.88991547e-02 -1.41412210e+00 -3.31876874e-02
-4.18279827e-01 -7.21321166e-01 -3.28914464e-01 -4.55807149e-01
-1.19641888e+00 9.44158792e-01 4.97853965e-01 -5.51107049e-01
9.70533669e-01 1.15004934e-01 8.92788827e-01 -9.57590193e-02
8.17990124e-01 -9.38375652e-01 5.72351813e-02 3.86282802e-01
1.38987851e+00 -8.48647714e-01 -2.46077269e-01 -5.72247088e-01
-6.80600584e-01 9.85093474e-01 6.43256843e-01 -1.18760571e-01
5.10890722e-01 1.89727172e-01 3.49452525e-01 -7.75793940e-02
-4.52266216e-01 -3.61771621e-02 2.30969027e-01 9.51047301e-01
4.63312358e-01 -1.25002831e-01 4.45485383e-01 5.25591671e-01
-7.96398938e-01 2.98081756e-01 3.21050793e-01 5.56580544e-01
-6.61954656e-02 -1.41729248e+00 -5.49843192e-01 -1.34304073e-02
-2.83545017e-01 2.38810688e-01 -5.75184584e-01 6.46215081e-01
2.24969223e-01 5.49475729e-01 1.86515808e-01 -6.33790120e-02
7.22031355e-01 2.07095116e-01 9.40702498e-01 -9.23998654e-01
-6.66769683e-01 3.36648524e-02 -6.20795153e-02 -6.58123076e-01
-2.66075820e-01 -8.31089795e-01 -1.13158929e+00 -3.99595678e-01
3.82475089e-03 -1.42154872e-01 4.31540012e-01 7.08456695e-01
4.24879104e-01 1.40366346e-01 6.41445577e-01 -1.72837639e+00
-9.99729410e-02 -7.54431307e-01 -7.69231439e-01 6.94720566e-01
2.17435524e-01 -6.98226571e-01 -1.73499405e-01 2.19653338e-01]
|
[12.887380599975586, -0.3275938332080841]
|
c5f45edc-ab11-4d2f-b205-c6646362db58
|
mmdf-mobile-microscopy-deep-framework
|
2007.13701
| null |
https://arxiv.org/abs/2007.13701v3
|
https://arxiv.org/pdf/2007.13701v3.pdf
|
Deep learning Framework for Mobile Microscopy
|
Mobile microscopy is a promising technology to assist and to accelerate disease diagnostics, with its widespread adoption being hindered by the mediocre quality of acquired images. Although some paired image translation and super-resolution approaches for mobile microscopy have emerged, a set of essential challenges, necessary for automating it in a high-throughput setting, still await to be addressed. The issues like in-focus/out-of-focus classification, fast scanning deblurring, focus-stacking, etc. -- all have specific peculiarities when the data are recorded using a mobile device. In this work, we aspire to create a comprehensive pipeline by connecting a set of methods purposely tuned to mobile microscopy: (1) a CNN model for stable in-focus / out-of-focus classification, (2) modified DeblurGAN architecture for image deblurring, (3) FuseGAN model for combining in-focus parts from multiple images to boost the detail. We discuss the limitations of the existing solutions developed for professional clinical microscopes, propose corresponding improvements, and compare to the other state-of-the-art mobile analytics solutions.
|
['Dmitry V. Dylov', 'Valeriya Pronina', 'Olga Novitskaya', 'Egor Sevriugov', 'Mikhail Salnikov', 'Kirill Shcherbakov', 'Maria Begicheva', 'Anatasiia Kornilova']
|
2020-07-27
| null | null | null | null |
['multi-focus-microscopical-images-fusion']
|
['medical']
|
[ 5.30494750e-01 -4.29871321e-01 3.84225458e-01 -2.08446667e-01
-8.02725434e-01 -3.42968851e-01 3.37323546e-01 -7.19709694e-02
-6.62329376e-01 6.47005558e-01 -9.06369910e-02 -2.54350245e-01
-2.84356385e-01 -2.00025454e-01 -5.02453983e-01 -1.16563380e+00
1.54154673e-02 4.09052014e-01 3.48860294e-01 -4.27592024e-02
4.75677848e-01 6.04697466e-01 -1.56014633e+00 3.84613812e-01
7.23671079e-01 6.03813887e-01 6.32589936e-01 1.32046640e+00
-2.97169816e-02 8.97858083e-01 -6.35252595e-01 5.57590425e-02
-2.64454991e-01 -5.03587782e-01 -1.09639966e+00 -2.41142467e-01
4.76590991e-01 -4.73226726e-01 2.65230183e-02 1.01025677e+00
9.27204847e-01 -4.08790857e-01 4.91582334e-01 -5.92900217e-01
-6.63580000e-01 -6.47299886e-02 -4.56701845e-01 9.72054601e-01
4.09364760e-01 4.50932205e-01 1.03797741e-01 -6.34408057e-01
1.04703963e+00 9.54663098e-01 8.05596828e-01 7.93933928e-01
-1.13182271e+00 -1.96039736e-01 -3.73357356e-01 3.66865605e-01
-1.03564191e+00 -5.88752091e-01 3.06988090e-01 -6.64608657e-01
1.02435970e+00 3.07217777e-01 6.87612295e-01 9.65885282e-01
7.71268308e-01 2.91020721e-01 1.36831689e+00 -2.31211454e-01
-5.50212432e-03 7.44778812e-02 -8.28038156e-02 6.75654590e-01
-6.92536980e-02 -2.89397955e-01 -6.92076981e-01 1.15920022e-01
8.93065095e-01 2.21416607e-01 -7.15810418e-01 -1.47774611e-02
-1.65411413e+00 2.38086969e-01 1.72039032e-01 7.45636642e-01
-2.38726676e-01 -8.09948519e-02 3.51009458e-01 3.81167650e-01
7.94852138e-01 5.17210722e-01 -4.20027107e-01 -2.53015071e-01
-1.34757125e+00 7.47853145e-02 4.06722754e-01 5.59240580e-01
5.98873615e-01 -4.76392776e-01 -1.20127089e-01 3.75989258e-01
9.27258059e-02 4.87203002e-01 7.28672683e-01 -8.11521947e-01
-4.60933801e-03 4.35776681e-01 3.12907323e-02 -8.61205459e-01
-7.79401720e-01 -1.59155712e-01 -1.06421018e+00 -1.24190617e-02
4.69953120e-01 -1.23334378e-01 -7.57605493e-01 1.26994348e+00
5.48852682e-01 5.83869159e-01 -1.27017647e-01 9.64588702e-01
1.06653249e+00 3.35948855e-01 -4.25438315e-01 -3.58309567e-01
1.49780858e+00 -9.94296968e-01 -9.36576426e-01 2.74366409e-01
6.83743179e-01 -1.03624117e+00 8.84509385e-01 2.06979454e-01
-1.07940507e+00 -3.40269953e-01 -7.40621209e-01 -4.33254629e-01
-6.11874878e-01 -3.25261131e-02 2.70525098e-01 4.51019377e-01
-1.57947910e+00 6.54710948e-01 -1.20279241e+00 -7.53546357e-01
6.03537321e-01 7.28689253e-01 -6.65918410e-01 4.50251587e-02
-6.11196458e-01 7.74890482e-01 5.39980270e-03 6.36521205e-02
-5.13856471e-01 -1.22518396e+00 -4.35159087e-01 -2.37746298e-01
-1.87591195e-01 -1.02230680e+00 9.77544606e-01 -3.45883936e-01
-1.64314878e+00 1.37255383e+00 -4.01333719e-01 -2.63133675e-01
4.20227766e-01 2.17576046e-02 -2.44112998e-01 5.24907708e-01
-1.53068826e-01 6.35575593e-01 5.78367352e-01 -7.61477232e-01
-6.66889727e-01 -6.08430088e-01 -2.72542417e-01 3.39334458e-02
-1.96612820e-01 3.31119001e-01 -5.30444443e-01 -2.36111253e-01
-1.96666256e-01 -7.76450694e-01 -7.16539100e-02 -3.16424780e-02
-3.50669682e-01 2.71089524e-01 1.35642624e+00 -8.11719298e-01
1.02782309e+00 -1.83949327e+00 4.03004974e-01 -5.82387805e-01
5.55206060e-01 7.41550684e-01 1.22692391e-01 2.70038307e-01
-2.19433345e-02 -3.97366621e-02 -1.66481326e-03 -7.31099725e-01
-6.67229772e-01 -2.40608573e-01 8.65556970e-02 8.45390558e-01
8.04660246e-02 1.09113586e+00 -1.10608721e+00 -5.85184693e-01
5.84154189e-01 1.04762352e+00 -3.53242427e-01 4.49374914e-01
2.44621769e-01 1.19472456e+00 5.67027479e-02 7.19572783e-01
8.33962858e-01 -7.56397665e-01 1.95271689e-02 -4.59377289e-01
-3.31381708e-01 8.86335969e-02 -8.53532255e-01 1.63783491e+00
-1.39064446e-01 7.91847229e-01 5.41327775e-01 -8.83968115e-01
2.62080520e-01 2.94594347e-01 8.66621733e-01 -2.74287164e-01
1.28093302e-01 3.54660958e-01 -2.13666856e-01 -1.09574926e+00
5.61133862e-01 1.15529764e-02 6.42617464e-01 3.17814738e-01
3.50831151e-01 -4.99383733e-02 1.19543500e-01 -7.77657032e-02
1.12645185e+00 1.69474650e-02 1.81646869e-01 -4.27063078e-01
7.52060652e-01 3.45746018e-02 4.43988200e-03 5.34539402e-01
-5.00174761e-01 1.14252996e+00 2.91953355e-01 -5.95470428e-01
-9.52504873e-01 -3.59833628e-01 -2.14105874e-01 5.00192225e-01
1.03683293e-01 -2.23444372e-01 -1.20899248e+00 -5.26445270e-01
-4.60861921e-01 -4.37441856e-01 -6.25943601e-01 3.10636133e-01
-4.91469026e-01 -1.13592458e+00 4.91708040e-01 -1.17848724e-01
4.22230810e-01 -8.52260530e-01 -9.51929510e-01 9.64746475e-02
-3.78630042e-01 -1.09800589e+00 -4.75362241e-01 1.90644637e-01
-6.64915562e-01 -1.30052865e+00 -1.16456819e+00 -6.66510344e-01
6.21207178e-01 7.68738329e-01 8.75865936e-01 1.61751136e-01
-5.34940481e-01 2.85652101e-01 -4.63799722e-02 -2.38024816e-01
-3.85302126e-01 1.65891081e-01 1.05366990e-01 -1.12455159e-01
4.11591172e-01 -4.92615670e-01 -1.14756548e+00 2.66012132e-01
-1.19342017e+00 1.16734013e-01 3.00215364e-01 9.64073181e-01
8.02086353e-01 -1.54500008e-01 1.01713493e-01 -1.17687309e+00
3.96025956e-01 -2.96089709e-01 -3.26890349e-01 1.89484537e-01
-4.76081759e-01 -4.06217277e-01 5.29779971e-01 -3.68839383e-01
-7.49996841e-01 -1.58228740e-01 -4.34954524e-01 -5.74183345e-01
-2.46578306e-01 2.82203674e-01 3.04461181e-01 -5.41184247e-01
6.70424998e-01 4.09028292e-01 5.87301373e-01 -3.67265642e-01
6.57834858e-02 8.72504950e-01 6.06058657e-01 1.88771114e-01
2.03383446e-01 1.04666317e+00 7.13078082e-02 -9.54907715e-01
-6.10827327e-01 -9.61761057e-01 -6.93011761e-01 -2.37287849e-01
1.10841072e+00 -8.42557490e-01 -8.78647506e-01 1.05884814e+00
-1.13516247e+00 -3.03122908e-01 -4.86309901e-02 2.50110090e-01
-5.27654171e-01 5.51167667e-01 -9.43949163e-01 -4.78995413e-01
-7.48537421e-01 -1.61951017e+00 1.46087158e+00 4.26147342e-01
-2.07551926e-01 -1.22163391e+00 3.21299911e-01 6.46630406e-01
8.87238443e-01 1.75790355e-01 4.20890599e-01 -2.90700972e-01
-5.67944944e-01 8.68142098e-02 -2.04370409e-01 8.40927660e-02
4.95031029e-01 1.61506265e-01 -1.11224627e+00 -6.24194980e-01
3.89316678e-01 -1.08190440e-01 7.66620398e-01 8.28857660e-01
1.05804849e+00 -1.92932159e-01 -6.31874621e-01 1.13709033e+00
1.35194612e+00 -9.90493074e-02 8.27898920e-01 3.32145423e-01
7.42160261e-01 4.82272089e-01 4.22754824e-01 6.68382719e-02
3.56249928e-01 7.47104287e-01 3.77915800e-01 -3.22239459e-01
-4.12077725e-01 3.71373504e-01 1.47999868e-01 1.02855468e+00
-2.58576512e-01 -1.85807616e-01 -7.68209696e-01 4.51224536e-01
-1.64399874e+00 -1.01286519e+00 -4.31495935e-01 2.01954961e+00
7.25856602e-01 -5.81113458e-01 1.74536392e-01 -6.67179227e-02
6.76406741e-01 -4.89378050e-02 -4.92521286e-01 1.40070096e-01
-2.99009055e-01 6.29559085e-02 2.94362128e-01 5.54975510e-01
-1.13506007e+00 7.17016935e-01 6.92905092e+00 7.88636029e-01
-1.83849478e+00 2.50430048e-01 8.75010431e-01 -1.96525693e-01
1.17531210e-01 -5.33387363e-01 -9.49793041e-01 8.12706351e-01
1.06463575e+00 5.29322207e-01 3.51917684e-01 3.92271936e-01
3.17044765e-01 -2.73282945e-01 -8.67693007e-01 1.32090783e+00
1.03749804e-01 -1.89828646e+00 -2.63478965e-01 2.45658740e-01
6.83044553e-01 4.85872000e-01 2.86105245e-01 -3.67546827e-01
-4.16803360e-01 -9.79571879e-01 2.52438307e-01 6.57282054e-01
1.19557488e+00 -2.53849417e-01 1.05151474e+00 1.67192876e-01
-7.59794354e-01 4.30309027e-01 -2.71032125e-01 3.60487431e-01
1.01510555e-01 8.88961375e-01 -8.53447080e-01 6.44544423e-01
1.06086349e+00 9.02860641e-01 -5.53469181e-01 9.04604018e-01
4.66067672e-01 2.36202538e-01 -1.15875885e-01 6.89887926e-02
-1.32014990e-01 -1.33895367e-01 4.49495852e-01 1.53734303e+00
6.33592010e-01 -2.58265585e-01 -6.66957438e-01 6.37825012e-01
3.28021795e-01 -1.15099020e-01 -3.43523413e-01 -4.26605456e-02
2.31631905e-01 1.61701560e+00 -9.61207807e-01 -2.34170392e-01
-3.43617499e-01 1.26570606e+00 2.20006809e-01 3.77183072e-02
-4.86187875e-01 -2.04846278e-01 6.99333966e-01 4.19120908e-01
2.62345254e-01 3.74768861e-02 3.69777009e-02 -1.50694168e+00
-1.46571636e-01 -1.03797984e+00 2.75122702e-01 -9.55574274e-01
-1.14573383e+00 6.00599885e-01 -4.51486617e-01 -1.13699961e+00
-1.71661317e-01 -7.60571420e-01 -4.51212645e-01 9.80127394e-01
-1.79903698e+00 -1.25840282e+00 -5.08759081e-01 6.71819746e-01
3.41973394e-01 3.42476107e-02 9.16756511e-01 5.24776995e-01
-4.78432357e-01 2.08215505e-01 4.68162298e-01 -3.56493652e-01
1.13570750e+00 -1.12928855e+00 1.62274256e-01 7.52650976e-01
-3.61279517e-01 9.11083639e-01 7.47193754e-01 -3.31392020e-01
-1.62395811e+00 -8.97286534e-01 9.34190333e-01 -7.85770357e-01
4.91371036e-01 -6.02983013e-02 -9.77222860e-01 4.16092604e-01
2.94178277e-01 2.05613196e-01 8.36165905e-01 -2.45905757e-01
2.20574945e-01 -8.84246528e-02 -1.45602524e+00 4.82116431e-01
7.08491623e-01 -5.33098698e-01 1.47407567e-02 5.45012712e-01
4.84940320e-01 -8.98296833e-01 -8.68043065e-01 8.65650028e-02
5.64799666e-01 -1.42036295e+00 1.01301277e+00 -2.36369327e-01
6.83730006e-01 -5.43839693e-01 3.64379197e-01 -1.10674059e+00
-3.62239122e-01 -8.76754284e-01 -3.00839841e-01 9.38295960e-01
-1.49135947e-01 -5.62588632e-01 8.75661075e-01 2.72270236e-02
-1.83741927e-01 -1.12577808e+00 -1.15592742e+00 -2.55843818e-01
-2.39957348e-01 1.62343994e-01 4.85406131e-01 9.32786703e-01
1.43063262e-01 3.46910357e-01 -4.81501520e-01 1.27512300e-02
4.15134847e-01 -2.20285475e-01 7.90208459e-01 -9.42784369e-01
-1.58742577e-01 -3.64878774e-01 -5.16875505e-01 -1.07042789e+00
-4.19315964e-01 -4.71444547e-01 -2.19695196e-01 -1.44935215e+00
4.63115036e-01 -7.06077963e-02 -1.03645772e-01 2.75163036e-02
-3.84826332e-01 6.40490949e-01 -1.38487786e-01 6.30120993e-01
-8.43163073e-01 -2.17893988e-01 1.37688601e+00 -4.44499962e-02
-1.09060593e-01 -2.29583547e-01 -6.96624100e-01 3.21607292e-01
5.66415310e-01 -2.33915985e-01 -1.67174935e-01 -4.09918040e-01
1.72076419e-01 9.05445442e-02 3.58128339e-01 -1.13428390e+00
6.69798613e-01 1.81002319e-01 2.66881377e-01 -6.11634374e-01
2.78276682e-01 -7.49505520e-01 5.21245837e-01 6.10645175e-01
8.07940215e-02 5.10800779e-02 -9.75379422e-02 3.86917174e-01
-2.53985226e-01 2.64186896e-02 1.11791098e+00 -3.17997485e-01
-2.19810754e-01 2.95543134e-01 -6.00476563e-01 -2.54064322e-01
9.41224456e-01 -4.52782333e-01 -7.01594770e-01 -7.29364380e-02
-4.15504754e-01 -1.42705858e-01 8.65766883e-01 8.80290717e-02
4.01628256e-01 -7.71912158e-01 -4.40508932e-01 3.48713785e-01
-1.04310684e-01 3.50771368e-01 9.08850551e-01 1.67574573e+00
-1.00278902e+00 7.27106690e-01 -2.62261182e-01 -1.08574772e+00
-1.69985163e+00 5.57277083e-01 6.31428123e-01 -7.59545386e-01
-4.53579485e-01 1.06334865e+00 5.19284280e-03 -4.14218038e-01
-2.86384910e-01 -4.07451957e-01 -3.41036707e-01 -8.33739117e-02
1.04222429e+00 4.63899612e-01 5.55170000e-01 -6.30077839e-01
-4.64205563e-01 7.98018098e-01 -3.23196143e-01 3.50653410e-01
1.50056672e+00 -6.76517725e-01 -5.82906365e-01 4.02167648e-01
1.17702460e+00 -1.74980894e-01 -1.05770159e+00 -1.28261745e-02
-6.07214987e-01 -3.70656580e-01 8.62064213e-02 -5.27137160e-01
-1.00539756e+00 9.85789597e-01 9.69954371e-01 3.24952513e-01
1.18946934e+00 -9.90148485e-02 8.12061965e-01 5.58227673e-02
2.79753476e-01 -8.20360959e-01 -1.80103246e-03 5.40756464e-01
3.19961160e-01 -1.37656450e+00 -5.92037067e-02 -4.26973999e-01
-1.54734090e-01 1.26418912e+00 1.99585155e-01 2.88484544e-01
7.28535950e-01 5.97854257e-01 2.51007289e-01 -4.56796378e-01
-5.68608761e-01 1.30080789e-01 1.23621568e-01 9.46851015e-01
7.57365108e-01 -4.75289643e-01 -1.50392847e-02 -6.63774787e-03
6.84917122e-02 3.87521476e-01 5.37450850e-01 8.75428259e-01
-2.02514902e-01 -7.76329219e-01 -5.21329522e-01 5.42922199e-01
-9.61098135e-01 -4.51716296e-02 -6.00928925e-02 4.03348207e-01
3.61663908e-01 8.54405761e-01 1.24840610e-01 -3.41595411e-01
2.36118045e-02 -1.98657498e-01 5.55564761e-01 -4.24071342e-01
-8.17017436e-01 2.02539235e-01 -5.15433669e-01 -6.34551227e-01
-9.83041108e-01 -5.78603864e-01 -6.77891552e-01 -7.05411553e-01
-2.71885902e-01 -1.70258313e-01 5.50766051e-01 8.87023807e-01
9.84511495e-01 6.06888592e-01 5.37238382e-02 -1.29467666e+00
9.41346139e-02 -1.09935582e+00 -5.77731490e-01 2.63693184e-01
8.25573325e-01 -1.73817039e-01 -5.76425731e-01 6.24067128e-01]
|
[12.940359115600586, -2.735344171524048]
|
edf5112b-fcaa-4c4d-96cc-08cb8a549281
|
deep-probabilistic-time-series-forecasting
|
2106.05848
| null |
https://arxiv.org/abs/2106.05848v2
|
https://arxiv.org/pdf/2106.05848v2.pdf
|
Deep Probabilistic Time Series Forecasting using Augmented Recurrent Input for Dynamic Systems
|
The demand of probabilistic time series forecasting has been recently raised in various dynamic system scenarios, for example, system identification and prognostic and health management of machines. To this end, we combine the advances in both deep generative models and state space model (SSM) to come up with a novel, data-driven deep probabilistic sequence model. Specifically, we follow the popular encoder-decoder generative structure to build the recurrent neural networks (RNN) assisted variational sequence model on an augmented recurrent input space, which could induce rich stochastic sequence dependency. Besides, in order to alleviate the inconsistency issue of the posterior between training and predicting as well as improving the mining of dynamic patterns, we (i) propose using a lagged hybrid output as input for the posterior at next time step, which brings training and predicting into alignment; and (ii) further devise a generalized auto-regressive strategy that encodes all the historical dependencies for the posterior. Thereafter, we first investigate the methodological characteristics of the proposed deep probabilistic sequence model on toy cases, and then comprehensively demonstrate the superiority of our model against existing deep probabilistic SSM models through extensive numerical experiments on eight system identification benchmarks from various dynamic systems. Finally, we apply our sequence model to a real-world centrifugal compressor forecasting problem, and again verify its outstanding performance by quantifying the time series predictive distribution.
|
['Xiaofang Wang', 'Shuhua Yang', 'Xudong Chen', 'Xiaomo Jiang', 'Changjun Liu', 'Haitao Liu']
|
2021-06-03
| null | null | null | null |
['probabilistic-time-series-forecasting']
|
['time-series']
|
[ 2.14135274e-02 -2.85249233e-01 3.29661340e-01 -8.31986070e-02
-6.00827932e-01 -2.13208050e-01 5.23137093e-01 -3.73126417e-01
1.88022465e-01 7.07636356e-01 3.37031186e-01 -5.35254061e-01
-3.91727418e-01 -5.42146266e-01 -7.25693882e-01 -1.23486400e+00
2.70830765e-02 4.45696920e-01 -2.47010320e-01 -5.37610538e-02
1.14494348e-02 1.60814479e-01 -1.11286199e+00 -1.94857031e-01
7.46992111e-01 9.27582026e-01 4.49004441e-01 5.62449515e-01
1.00548841e-01 9.21215892e-01 -4.92302090e-01 -2.08146602e-01
1.49987184e-03 -5.52447438e-01 -2.06433773e-01 9.64608118e-02
-9.99765754e-01 -2.55511433e-01 -7.49037087e-01 8.54780316e-01
5.46489418e-01 3.07606310e-01 9.50586319e-01 -1.05313194e+00
-7.10236847e-01 9.08293426e-01 -5.33544362e-01 3.58693719e-01
-1.45043224e-01 4.06334221e-01 7.25609958e-01 -7.38145530e-01
-5.77404834e-02 1.09763241e+00 7.06570327e-01 3.63887399e-01
-9.90189910e-01 -5.42853475e-01 2.14781955e-01 2.71850288e-01
-1.32716608e+00 -2.80458331e-01 1.25067651e+00 -6.53846323e-01
8.00123274e-01 -1.99246913e-01 5.50542831e-01 1.58258975e+00
7.82658339e-01 9.80220497e-01 7.95546234e-01 7.57431537e-02
8.62352327e-02 -3.15275341e-02 9.75913778e-02 3.79261166e-01
-2.50955313e-01 3.52132678e-01 -3.10057580e-01 -1.85991660e-01
6.61888123e-01 4.60490376e-01 -2.79623508e-01 1.96009595e-02
-1.08189166e+00 5.97685516e-01 -2.15699181e-01 2.32714891e-01
-9.50185418e-01 2.15866968e-01 4.00210232e-01 -1.31089045e-02
5.39507985e-01 7.53450990e-02 -4.18451905e-01 -3.47321361e-01
-9.96806681e-01 1.35984272e-01 7.53028035e-01 9.30301368e-01
2.31198967e-01 6.81752264e-01 -3.12165439e-01 6.33722901e-01
5.64279497e-01 6.21998966e-01 9.32174265e-01 -5.67692578e-01
3.67847532e-01 1.06709577e-01 1.09830581e-01 -9.74562168e-01
-3.25304180e-01 -8.18684399e-01 -1.33519638e+00 -4.93741274e-01
-1.82940140e-01 -5.70670307e-01 -7.88902223e-01 1.80262339e+00
1.23103201e-01 5.95164418e-01 3.39432716e-01 6.72340155e-01
2.24475667e-01 1.36126912e+00 2.91255862e-02 -5.84334910e-01
1.14757168e+00 -6.99136376e-01 -9.36395109e-01 2.75522292e-01
3.97827595e-01 -6.36523366e-01 6.19490385e-01 3.10969859e-01
-9.93144751e-01 -6.84188426e-01 -9.23376381e-01 3.27767313e-01
-4.06331802e-03 3.04714859e-01 1.87318206e-01 2.04792827e-01
-6.75725400e-01 7.71719396e-01 -1.36003435e+00 1.62111342e-01
-1.44864889e-02 -2.87542194e-02 2.30937004e-01 4.11999017e-01
-1.47141373e+00 8.84766221e-01 6.75480068e-01 8.31474245e-01
-1.32835448e+00 -8.34764123e-01 -6.40960932e-01 2.81555384e-01
2.97423393e-01 -9.61435080e-01 1.16441143e+00 -5.35419106e-01
-1.93150854e+00 -2.90669143e-01 -2.98050821e-01 -6.78800225e-01
3.19253892e-01 -2.63020813e-01 -5.10013163e-01 -3.50244612e-01
-3.17924052e-01 -6.86145648e-02 8.29830289e-01 -1.07388115e+00
-4.10289168e-01 -4.69378307e-02 -5.72986007e-01 2.19619393e-01
-1.91239774e-01 -2.15756327e-01 -1.58221275e-02 -8.51786077e-01
7.59797636e-03 -8.45726848e-01 -4.25883859e-01 -1.00701547e+00
-6.31438076e-01 -4.18119639e-01 7.65166700e-01 -1.07964683e+00
1.44130278e+00 -2.18877649e+00 4.54945356e-01 1.63885102e-01
-5.26983216e-02 1.85409904e-01 2.89288938e-01 9.22239780e-01
-2.32202187e-01 -7.10049197e-02 -4.69781131e-01 -7.28389442e-01
1.65627435e-01 4.74089801e-01 -9.99080062e-01 3.97937834e-01
3.74533445e-01 9.77020919e-01 -6.86387181e-01 -2.82253355e-01
3.58249187e-01 5.77694893e-01 -1.82341844e-01 3.64160240e-01
-1.85501546e-01 5.58108509e-01 -6.79676652e-01 2.73042083e-01
4.21150982e-01 -4.31143045e-01 9.49901193e-02 -6.03880323e-02
-4.87003177e-02 9.77403373e-02 -1.09052575e+00 1.26468015e+00
-4.64576960e-01 2.06515312e-01 -4.45406854e-01 -1.23486602e+00
9.37121093e-01 6.46780789e-01 7.23463535e-01 -1.87935174e-01
2.06876710e-01 2.58608982e-02 -3.08846254e-02 -5.60346663e-01
5.43197036e-01 -4.77775723e-01 -3.72289382e-02 3.95752281e-01
1.69332877e-01 1.38472185e-01 -1.85784549e-01 3.40746976e-02
6.91003323e-01 3.30361128e-01 -9.10019204e-02 -1.73488915e-01
6.06430471e-01 -4.66814071e-01 7.10064232e-01 3.88570279e-01
1.41036659e-01 4.28734034e-01 6.97230995e-01 -6.48999140e-02
-1.20544779e+00 -9.55054522e-01 -5.79975080e-03 5.31000137e-01
4.58949730e-02 -2.22404435e-01 -3.15456301e-01 -1.92301273e-01
-1.24209933e-01 1.02511978e+00 -5.13299763e-01 -4.08563584e-01
-5.08387804e-01 -1.33901358e+00 4.24333483e-01 6.93167746e-01
9.24218073e-02 -9.61276352e-01 -1.40713304e-01 7.31132627e-01
-1.34245962e-01 -7.20667720e-01 -4.00705665e-01 2.62851357e-01
-6.50498509e-01 -5.45976937e-01 -8.50471795e-01 -4.68504995e-01
2.38456979e-01 -3.17422330e-01 7.35771537e-01 -4.44627047e-01
2.84796506e-01 2.85515696e-01 -2.46325985e-01 -4.11390126e-01
-6.86211765e-01 1.55379713e-01 3.41012537e-01 2.58537441e-01
-5.02326228e-02 -9.43706095e-01 -6.08792663e-01 1.66044697e-01
-9.48450148e-01 2.93940693e-01 5.42810678e-01 1.08470690e+00
6.46588743e-01 2.82980591e-01 9.28068042e-01 -5.34041584e-01
7.38632023e-01 -1.10041213e+00 -7.35716522e-01 3.15851599e-01
-9.29667830e-01 2.59915024e-01 8.38592291e-01 -4.97645825e-01
-1.15786481e+00 -1.20017782e-01 -4.19647962e-01 -9.96065617e-01
2.89876431e-01 9.53533649e-01 1.57701045e-01 7.59047866e-01
-4.25190181e-02 1.00644803e+00 6.83704317e-02 -5.99098861e-01
2.90676713e-01 5.09075701e-01 6.60599172e-01 -6.10867560e-01
7.66290247e-01 7.98400119e-02 1.08811967e-01 -4.46191192e-01
-6.95160151e-01 -1.85072601e-01 -2.40974590e-01 -1.33664772e-01
6.10047638e-01 -9.57832754e-01 -1.03812182e+00 6.47810221e-01
-1.22567558e+00 -9.71972272e-02 -1.95287853e-01 7.63211608e-01
-6.05961144e-01 4.16327953e-01 -1.08393848e+00 -1.36695242e+00
-4.35779154e-01 -1.22034740e+00 1.09158063e+00 2.39416957e-01
2.22025722e-01 -1.16907322e+00 2.80086249e-01 -3.00540388e-01
4.19257820e-01 3.18526596e-01 9.18395042e-01 -7.16502368e-01
-3.54192585e-01 -1.32195905e-01 1.80681050e-01 6.66938722e-01
-5.65085094e-03 2.49293387e-01 -7.43005335e-01 -2.04879016e-01
5.99898458e-01 3.04329038e-01 7.07298398e-01 5.57699203e-01
1.20214844e+00 -3.80375236e-01 -4.28914279e-01 5.93600273e-01
1.18466830e+00 6.26902163e-01 4.50002909e-01 -1.03841521e-01
7.29144812e-01 4.79219228e-01 1.98699772e-01 7.26662040e-01
5.14682233e-01 1.14905618e-01 2.25753531e-01 3.28283668e-01
5.43460190e-01 -7.46850252e-01 6.04373515e-01 1.79746771e+00
3.12415194e-02 -5.41561067e-01 -7.92272151e-01 4.63586479e-01
-1.91713941e+00 -1.02031791e+00 1.14548486e-02 1.85130978e+00
6.86526716e-01 2.49908399e-02 -1.26404375e-01 -1.84249841e-02
6.50814474e-01 1.88654363e-01 -6.17387533e-01 1.01928599e-01
-1.44243941e-01 -9.31773707e-02 3.23898345e-01 1.81959987e-01
-8.95237029e-01 5.06326437e-01 5.68971252e+00 1.10178208e+00
-1.26640320e+00 7.97965303e-02 8.96222174e-01 4.81563061e-02
-3.35703343e-01 -5.79055287e-02 -9.13592517e-01 1.04712439e+00
1.53568888e+00 -5.20884335e-01 2.61395156e-01 7.44693577e-01
5.59960783e-01 4.24305111e-01 -9.71288860e-01 9.61741507e-01
-2.13186160e-01 -1.20093656e+00 1.19764209e-01 3.96124125e-02
7.64840603e-01 1.00092739e-01 2.01496795e-01 7.09954500e-01
3.09435248e-01 -9.44876432e-01 7.74856329e-01 1.21831751e+00
1.85867503e-01 -8.18674624e-01 1.00338161e+00 7.31744349e-01
-9.75187719e-01 -1.17652953e-01 -3.22702467e-01 1.19157076e-01
7.14844882e-01 7.27706373e-01 -6.30544424e-01 1.06836140e+00
4.76037920e-01 9.57544267e-01 3.11670657e-02 8.13959420e-01
-4.04627286e-02 1.08750284e+00 -3.98440212e-01 -1.65269300e-01
2.53930300e-01 -5.52366495e-01 7.41100192e-01 8.24313223e-01
6.32404506e-01 1.43537045e-01 4.89920238e-03 1.03305542e+00
2.68746883e-01 -2.66673446e-01 -4.77058381e-01 -3.79660726e-01
3.69751126e-01 1.01289070e+00 -3.50525975e-01 -3.63746941e-01
-7.26441592e-02 7.46969998e-01 -1.66838214e-01 6.21086359e-01
-1.25380480e+00 -9.77297276e-02 3.39373857e-01 -3.69235992e-01
4.19868112e-01 -6.16500318e-01 -1.31351445e-02 -1.19046354e+00
4.92490269e-02 -5.40349603e-01 1.18916340e-01 -8.94453168e-01
-1.44017267e+00 6.87310994e-01 5.16286790e-02 -1.33301282e+00
-8.13739657e-01 -2.90866613e-01 -7.18403399e-01 1.34383142e+00
-1.34656835e+00 -9.24108922e-01 4.48010117e-01 4.86458361e-01
6.24703705e-01 -5.38245700e-02 4.47109491e-01 3.01722348e-01
-1.14199948e+00 1.93654194e-01 7.90169060e-01 -1.52543008e-01
1.63470611e-01 -1.03166115e+00 4.23391163e-01 9.67064500e-01
-9.77541432e-02 7.79179215e-01 8.68722081e-01 -6.05110347e-01
-1.63156557e+00 -1.20415282e+00 4.60130602e-01 -5.60536802e-01
1.00609672e+00 -1.54624388e-01 -1.25920248e+00 7.03207791e-01
1.67993277e-01 -4.75245416e-01 3.73542488e-01 -2.58461237e-01
3.35894942e-01 5.74226631e-03 -5.56042194e-01 3.45319033e-01
3.92498195e-01 -5.12647092e-01 -7.17962444e-01 4.17419553e-01
9.54734385e-01 -5.50563216e-01 -1.07192349e+00 5.51923454e-01
3.85171592e-01 -4.76435900e-01 6.46166265e-01 -5.90701282e-01
6.34793222e-01 -4.46769446e-01 -8.43970478e-02 -1.48430908e+00
-3.28456938e-01 -1.07433128e+00 -5.96964777e-01 1.47194433e+00
3.87453258e-01 -8.09250653e-01 3.79369408e-01 3.12067240e-01
-6.17945969e-01 -1.17721891e+00 -8.88274729e-01 -6.56826437e-01
8.93036053e-02 -6.33973002e-01 7.37785995e-01 6.69854462e-01
-1.74771294e-01 1.76857799e-01 -8.95865858e-01 5.75237393e-01
2.30073646e-01 2.16937199e-01 4.71088231e-01 -6.99589074e-01
-6.05402112e-01 -4.74571049e-01 -2.59186029e-02 -1.35203087e+00
1.22454591e-01 -5.45585573e-01 3.68177056e-01 -1.03303838e+00
7.49773607e-02 -1.88470617e-01 -7.77331829e-01 8.00131038e-02
-4.24476892e-01 -4.60992008e-01 -4.91477214e-02 4.86485928e-01
-2.75873750e-01 1.40945494e+00 9.65245605e-01 2.35536665e-01
6.44327998e-02 4.09386456e-01 -4.84131545e-01 3.84468406e-01
5.04048407e-01 -3.72551054e-01 -5.54748833e-01 -9.71761048e-02
1.50507376e-01 7.32140541e-01 3.21963519e-01 -8.37355196e-01
2.76212335e-01 -1.60261974e-01 2.14422822e-01 -9.34270501e-01
3.25611889e-01 -7.02006280e-01 4.82016921e-01 5.32978594e-01
-3.57974917e-02 3.38962942e-01 8.70113373e-02 1.01982617e+00
-3.96309972e-01 3.89730074e-02 3.70391935e-01 3.07930648e-01
-4.83164728e-01 6.53639376e-01 -4.67419893e-01 -3.69511068e-01
8.26136112e-01 2.93457985e-01 -1.50946015e-02 -3.78058344e-01
-7.23647892e-01 6.26908183e-01 -1.60927609e-01 4.88929480e-01
3.99516106e-01 -1.18061006e+00 -7.00513601e-01 2.19552219e-01
-3.65499109e-01 -1.58740785e-02 8.22934747e-01 1.08937252e+00
-9.48737860e-02 5.59341431e-01 4.12623197e-01 -7.99943030e-01
-3.88170332e-01 8.12946916e-01 4.23653811e-01 -4.04469669e-01
-7.20107436e-01 5.18448830e-01 2.64018446e-01 -6.31776005e-02
-8.50868374e-02 -4.63271201e-01 -1.38548031e-01 -1.82604119e-01
1.05324909e-01 2.68313438e-01 -1.87718451e-01 -4.58174765e-01
-1.24304108e-01 2.52594441e-01 1.76267967e-01 -9.39386487e-02
1.43830717e+00 -3.56648952e-01 4.03789096e-02 9.77417111e-01
1.05475843e+00 -5.56363821e-01 -1.49620116e+00 -2.63675690e-01
-6.70652613e-02 2.82307774e-01 -1.03414347e-02 -3.09532285e-01
-1.01108778e+00 9.43725049e-01 3.46091688e-01 5.42492032e-01
1.16365075e+00 -2.26460353e-01 1.13619173e+00 1.22143447e-01
1.31021246e-01 -6.81990266e-01 -4.42081690e-01 6.34984016e-01
7.17321813e-01 -9.16162550e-01 -3.07761997e-01 2.85400361e-01
-8.10284913e-01 1.09244096e+00 5.94128706e-02 -3.36121947e-01
1.11400366e+00 2.88252980e-01 -3.17483455e-01 -7.41950050e-02
-1.21731186e+00 2.83030421e-01 3.11527401e-01 5.74224889e-02
2.57228762e-01 1.98281318e-01 -2.98578411e-01 1.28009164e+00
-3.42933297e-01 3.10672037e-02 3.53808045e-01 6.66070879e-01
1.37263676e-02 -7.23048806e-01 -5.70324436e-02 3.88802052e-01
-4.72190410e-01 -1.94505170e-01 5.31031072e-01 4.25095886e-01
-4.20596212e-01 5.90311825e-01 -2.29343423e-03 -5.87464154e-01
1.57801673e-01 3.34971070e-01 -1.78975478e-01 -2.52695501e-01
-3.57919365e-01 3.25516850e-01 -3.78218234e-01 -1.03738243e-02
-1.52264997e-01 -8.31466079e-01 -9.27758455e-01 -2.01068267e-01
-5.04817128e-01 4.46704686e-01 6.70441568e-01 1.24859345e+00
5.78649819e-01 1.09008360e+00 9.83448803e-01 -8.77263129e-01
-1.11904347e+00 -1.36154222e+00 -7.11535394e-01 2.59469319e-02
3.67159605e-01 -6.53775573e-01 -4.49300259e-01 1.57089740e-01]
|
[6.962170600891113, 3.1276347637176514]
|
4ca025ab-548e-4c43-b08b-da8d2115300a
|
disentangling-visual-embeddings-for
|
2205.08536
| null |
https://arxiv.org/abs/2205.08536v1
|
https://arxiv.org/pdf/2205.08536v1.pdf
|
Disentangling Visual Embeddings for Attributes and Objects
|
We study the problem of compositional zero-shot learning for object-attribute recognition. Prior works use visual features extracted with a backbone network, pre-trained for object classification and thus do not capture the subtly distinct features associated with attributes. To overcome this challenge, these studies employ supervision from the linguistic space, and use pre-trained word embeddings to better separate and compose attribute-object pairs for recognition. Analogous to linguistic embedding space, which already has unique and agnostic embeddings for object and attribute, we shift the focus back to the visual space and propose a novel architecture that can disentangle attribute and object features in the visual space. We use visual decomposed features to hallucinate embeddings that are representative for the seen and novel compositions to better regularize the learning of our model. Extensive experiments show that our method outperforms existing work with significant margin on three datasets: MIT-States, UT-Zappos, and a new benchmark created based on VAW. The code, models, and dataset splits are publicly available at https://github.com/nirat1606/OADis.
|
['Abhinav Shrivastava', 'Khoi Pham', 'Nirat Saini']
|
2022-05-17
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Saini_Disentangling_Visual_Embeddings_for_Attributes_and_Objects_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Saini_Disentangling_Visual_Embeddings_for_Attributes_and_Objects_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['compositional-zero-shot-learning']
|
['computer-vision']
|
[ 1.63741544e-01 -1.98336497e-01 -5.09679735e-01 -5.25883853e-01
-8.48494112e-01 -5.97147465e-01 8.62808526e-01 2.00429782e-01
-3.73258114e-01 3.19081992e-01 6.81094587e-01 8.75158608e-03
6.52141273e-02 -7.75176287e-01 -3.48956853e-01 -6.49095595e-01
2.72051662e-01 5.21447957e-01 -2.01312482e-01 9.03845429e-02
1.74021833e-02 2.53597945e-01 -1.81735671e+00 5.63215256e-01
6.06418252e-01 1.29452753e+00 6.52851677e-03 5.30976117e-01
-2.40715206e-01 7.54384100e-01 -1.95774034e-01 -3.83013129e-01
2.95591295e-01 -3.93111765e-01 -8.70728612e-01 3.12246114e-01
8.02433074e-01 -1.64857611e-01 -5.48268318e-01 9.15637016e-01
5.43656111e-01 2.89104313e-01 1.09874785e+00 -1.61058509e+00
-1.58468068e+00 3.91910315e-01 -3.03921193e-01 -1.87310297e-02
2.86494017e-01 3.26692849e-01 1.75574172e+00 -1.31255209e+00
6.55675948e-01 1.31483722e+00 4.74743307e-01 7.33952761e-01
-1.66244614e+00 -5.35659611e-01 1.51143387e-01 6.85223162e-01
-1.54608405e+00 -6.58055902e-01 7.49807417e-01 -6.04840457e-01
8.62280607e-01 3.18949521e-01 6.42636061e-01 1.73301542e+00
-4.17202502e-01 1.00950170e+00 1.01897645e+00 -3.92714351e-01
2.01603651e-01 1.98622391e-01 3.90947282e-01 5.75758994e-01
1.07633464e-01 -3.36535051e-02 -6.26724660e-01 -1.43981725e-01
3.91581833e-01 2.68685937e-01 -3.49418700e-01 -8.84202003e-01
-1.34480786e+00 1.06022608e+00 5.96696734e-01 -6.77736700e-02
-1.29571229e-01 -3.57399806e-02 3.12705308e-01 3.50132883e-01
5.12931228e-01 5.36410332e-01 -2.50875324e-01 1.86604988e-02
-4.83375967e-01 3.30105349e-02 6.12331748e-01 9.48601961e-01
9.89608228e-01 1.01673074e-01 -6.30591273e-01 1.11381733e+00
2.29795247e-01 3.76062036e-01 5.31595945e-01 -9.35830355e-01
2.37704992e-01 7.18750834e-01 -1.88736275e-01 -6.13759041e-01
-1.88345298e-01 -2.20334858e-01 -5.03575563e-01 2.92888194e-01
4.50577587e-01 3.37789744e-01 -1.30488861e+00 1.75524652e+00
1.75571471e-01 2.17499644e-01 1.48447230e-01 1.09960997e+00
1.21267343e+00 4.59784091e-01 3.32837582e-01 4.90468919e-01
1.64058363e+00 -1.11398482e+00 -6.04250193e-01 -5.03784120e-01
5.46693385e-01 -5.16823888e-01 1.65389407e+00 -5.40953875e-02
-7.81677783e-01 -5.35217464e-01 -1.31481493e+00 -6.25293076e-01
-8.00621390e-01 9.84987542e-02 6.18648410e-01 3.92516673e-01
-6.93653882e-01 4.12827492e-01 -5.95716178e-01 -4.53281373e-01
8.25607300e-01 2.06177123e-02 -6.08278453e-01 -2.14062408e-01
-1.15193832e+00 9.84479606e-01 2.63088375e-01 -3.32316190e-01
-9.20495808e-01 -8.71231437e-01 -1.28100729e+00 1.51218653e-01
3.83333415e-01 -6.82218254e-01 9.64190006e-01 -9.62893486e-01
-1.11428416e+00 1.19183016e+00 -2.68519372e-01 -1.91368133e-01
1.15202725e-01 2.83043087e-02 -2.61889607e-01 4.36273925e-02
1.56105936e-01 7.61801064e-01 8.09081316e-01 -1.33267486e+00
-1.42657220e-01 -4.78501260e-01 5.41653521e-02 7.11392984e-02
-7.39529967e-01 -1.21849321e-01 -2.46603221e-01 -6.12792134e-01
-7.34486133e-02 -8.34790349e-01 1.74734369e-01 3.52092654e-01
-4.40260768e-01 -3.29445928e-01 7.43826389e-01 -3.04914683e-01
6.95666969e-01 -2.42169666e+00 2.90464103e-01 -1.95862576e-01
5.06738663e-01 1.64440185e-01 -5.87673545e-01 4.09659743e-01
-1.93341658e-01 6.40367195e-02 -3.13239366e-01 -7.58724570e-01
2.32628971e-01 4.27943110e-01 -4.68184531e-01 5.13462722e-01
4.63914216e-01 1.18365014e+00 -9.50746596e-01 -4.69925404e-01
1.11328892e-01 6.13167703e-01 -4.44168478e-01 3.43979061e-01
-3.15433927e-02 1.45759046e-01 -2.14388594e-01 8.81411195e-01
4.95427251e-01 -4.34376210e-01 3.20286825e-02 -4.05394644e-01
3.03409070e-01 3.45206171e-01 -8.01422119e-01 1.81327450e+00
-4.97035444e-01 7.02275753e-01 -3.32963169e-01 -9.51344252e-01
8.88519466e-01 2.11910039e-01 1.72125563e-01 -6.52605236e-01
2.01273307e-01 -9.67865884e-02 -1.55689940e-01 -6.05561554e-01
4.26319510e-01 -3.02119136e-01 -1.73959762e-01 4.03255522e-01
6.76457286e-01 1.53322533e-01 6.26232103e-02 2.52968669e-01
8.76215875e-01 2.80660272e-01 3.90805423e-01 -1.35411188e-01
2.71766484e-01 -1.60388350e-01 6.27264082e-01 5.01020432e-01
-4.48276103e-01 9.58097696e-01 5.38521349e-01 -4.35943067e-01
-1.13729751e+00 -1.58940935e+00 -3.48900378e-01 1.45012903e+00
2.41284549e-01 -5.16963899e-01 -1.85291171e-01 -7.38580763e-01
3.66554201e-01 1.02272201e+00 -1.00790596e+00 -4.01498854e-01
-2.69610640e-02 -4.85170156e-01 3.39060873e-01 8.39862704e-01
3.93390432e-02 -1.09481180e+00 -2.96430320e-01 -1.47238418e-01
-3.12431037e-01 -1.15092134e+00 -4.49711204e-01 2.27039307e-01
-4.50641245e-01 -1.03113925e+00 -5.33972681e-01 -7.69063294e-01
4.91096228e-01 1.31275654e-01 1.21912801e+00 -1.05098493e-01
-6.39661849e-01 5.72377920e-01 -5.00380099e-01 -4.29172903e-01
-6.51298612e-02 -2.51129698e-02 -9.28629597e-05 5.08825541e-01
8.60924840e-01 -5.28657436e-01 -4.07818437e-01 3.22652906e-02
-6.17399216e-01 1.45401787e-02 4.67225552e-01 1.08937871e+00
5.23932517e-01 -9.91253316e-01 4.99460399e-01 -7.52722919e-01
3.52061242e-01 -7.31732607e-01 3.25774550e-02 3.04081321e-01
-5.11400461e-01 3.31968009e-01 5.79661250e-01 -7.06267834e-01
-8.53137732e-01 -5.10796197e-02 1.69566050e-01 -8.59034896e-01
-2.96246022e-01 1.45991474e-01 -3.95171851e-01 2.81672835e-01
6.48446321e-01 1.96806863e-01 1.24337770e-01 -6.34248793e-01
7.43525624e-01 7.28487968e-01 4.43869352e-01 -5.38586318e-01
7.27261782e-01 7.04063475e-01 -2.00070634e-01 -6.64739370e-01
-1.20191717e+00 -5.70172191e-01 -7.56910324e-01 1.24907263e-01
1.10986257e+00 -1.02331853e+00 -5.38235188e-01 1.32147893e-01
-1.04251206e+00 -9.01947543e-02 -6.09487712e-01 4.65250522e-01
-7.02362955e-01 1.36730954e-01 -5.56312323e-01 -5.36357999e-01
-1.80097461e-01 -1.03739893e+00 1.08096826e+00 -1.54995658e-02
-5.46506464e-01 -8.84494543e-01 7.88096040e-02 3.09034377e-01
3.39409083e-01 2.02247128e-01 1.13571560e+00 -9.36830819e-01
-4.88429934e-01 -7.31560439e-02 -4.90373671e-01 3.21957141e-01
1.36536837e-01 -2.73357034e-01 -1.44949615e+00 -2.00152665e-01
-1.76196009e-01 -1.02091825e+00 1.47534788e+00 -1.24163561e-01
1.25687706e+00 -2.41920710e-01 -2.60345221e-01 9.89032507e-01
1.29613483e+00 -3.42318207e-01 6.21286333e-01 2.60354638e-01
9.73824739e-01 6.00575149e-01 2.93654531e-01 3.98407936e-01
5.93996644e-01 7.55070567e-01 4.27669138e-01 1.05858803e-01
-6.42042994e-01 -3.66036981e-01 2.80321062e-01 6.00269198e-01
-2.38962434e-02 -1.30988359e-01 -7.93524802e-01 8.28436196e-01
-1.64696181e+00 -1.09263098e+00 2.53662020e-01 2.05002832e+00
8.94527972e-01 -3.30314368e-01 2.76604947e-02 -1.03424881e-02
4.48602647e-01 5.89834511e-01 -5.07083952e-01 -3.32751304e-01
-1.94734052e-01 3.44418764e-01 1.50658369e-01 4.01068687e-01
-1.19745445e+00 9.01262999e-01 5.44556141e+00 6.68184280e-01
-9.39830720e-01 3.00800085e-01 3.84984553e-01 -4.39668179e-01
-5.46062052e-01 5.02837673e-02 -6.34567082e-01 2.73105770e-01
8.61590683e-01 -1.42086551e-01 3.82812709e-01 7.49348879e-01
-5.41701555e-01 4.57494527e-01 -1.54740024e+00 1.04527557e+00
5.45957804e-01 -1.15735817e+00 3.00040096e-01 1.10911897e-04
3.51798385e-01 1.07646376e-01 4.37681526e-01 5.53856015e-01
3.17203432e-01 -1.29456115e+00 7.65403390e-01 5.53616405e-01
1.01232529e+00 -3.47218275e-01 2.95866519e-01 -1.49222568e-01
-1.22468936e+00 -2.62792975e-01 -6.14548981e-01 -1.78406447e-01
-2.03190431e-01 -1.44053195e-02 -6.12784445e-01 3.26871514e-01
5.92002988e-01 1.05452204e+00 -8.10762823e-01 8.29056382e-01
-3.70024264e-01 4.12421703e-01 2.55921692e-01 1.94014199e-02
1.40495151e-01 -6.61542341e-02 5.36709130e-01 9.88553703e-01
1.83202118e-01 -4.89215329e-02 3.60216171e-01 1.23574090e+00
-2.33702511e-01 6.37707440e-03 -7.02978253e-01 -2.85594195e-01
5.41467011e-01 1.37143397e+00 -2.48108625e-01 -5.41994333e-01
-7.69375503e-01 1.09975469e+00 7.24840879e-01 5.38608491e-01
-7.72978246e-01 -4.89119679e-01 1.21881032e+00 -4.40748874e-03
4.85483468e-01 -2.54786313e-01 -3.50115806e-01 -1.51252806e+00
-1.23449437e-01 -6.68021739e-01 4.91849273e-01 -6.95297778e-01
-1.74903762e+00 5.55242479e-01 -1.48711100e-01 -1.41035175e+00
-1.05255038e-01 -9.17853415e-01 -5.63812613e-01 1.01014054e+00
-1.50604248e+00 -1.51389098e+00 -4.46363419e-01 5.51534593e-01
6.14257514e-01 -2.59845257e-01 1.21672809e+00 1.38911575e-01
-4.42842543e-01 8.49914551e-01 -6.38636649e-02 3.91489059e-01
1.01859045e+00 -1.31307685e+00 3.52164119e-01 4.70538110e-01
6.05772018e-01 5.38127244e-01 4.94987041e-01 -4.33298409e-01
-1.52967763e+00 -1.01567543e+00 9.51820672e-01 -1.04385209e+00
8.96253228e-01 -8.51394951e-01 -1.09286034e+00 8.57460558e-01
3.40969115e-01 5.79385817e-01 1.14581704e+00 4.64238048e-01
-1.28187549e+00 4.16257046e-02 -8.06984246e-01 5.16889095e-01
1.31139243e+00 -9.64347720e-01 -7.56864130e-01 1.40268162e-01
8.19733500e-01 5.92568889e-02 -7.61690557e-01 2.07096860e-01
6.90250516e-01 -6.57671392e-01 1.22771120e+00 -1.04378426e+00
5.86452365e-01 -7.55776167e-02 -5.28228819e-01 -1.53996837e+00
-6.38355672e-01 8.83430764e-02 -1.92137167e-01 1.20803201e+00
5.49028575e-01 -5.83623648e-01 4.77950633e-01 4.24201548e-01
-8.19984898e-02 -7.66627431e-01 -8.40597868e-01 -1.00610435e+00
5.78114167e-02 -3.38135630e-01 6.26634121e-01 1.15600467e+00
-8.01177397e-02 7.46196151e-01 -4.54398036e-01 -1.18580468e-01
8.56893837e-01 3.71222794e-01 4.70045209e-01 -1.36190879e+00
-2.95904219e-01 -4.98267889e-01 -7.20866144e-01 -5.74768186e-01
4.97007191e-01 -1.51127529e+00 -1.35575712e-01 -1.55331993e+00
5.17493010e-01 -2.34713987e-01 -6.17421269e-01 9.53525126e-01
-2.99246132e-01 5.38866758e-01 4.36694801e-01 2.82798022e-01
-5.43728292e-01 1.05444658e+00 1.00555992e+00 -5.57296336e-01
2.66080439e-01 -5.94907165e-01 -7.99420834e-01 5.12301087e-01
6.63072348e-01 -3.39669913e-01 -4.45115179e-01 -6.68675482e-01
-1.80843472e-01 -5.34379184e-01 7.66192853e-01 -8.07146668e-01
1.07801370e-02 -1.18721463e-01 7.46664703e-01 -4.00004089e-02
9.61756825e-01 -7.01210380e-01 -4.93706584e-01 5.61359636e-02
-6.27532542e-01 -1.79231212e-01 -5.00077270e-02 6.96611762e-01
-2.49235898e-01 -1.21013351e-01 6.77351534e-01 2.64187157e-02
-1.00759006e+00 5.30970037e-01 5.63449338e-02 3.81137669e-01
1.03282285e+00 -2.66353041e-01 -6.54787719e-01 -2.83319145e-01
-8.34782183e-01 2.62813300e-01 7.26490736e-01 7.89146781e-01
8.31318617e-01 -1.84911776e+00 -6.64882600e-01 4.26952749e-01
9.37456310e-01 -6.91859365e-01 2.02512667e-01 7.04667509e-01
1.85121804e-01 -5.51939458e-02 -5.79921007e-01 -4.03166890e-01
-1.09518731e+00 9.39677358e-01 4.57605086e-02 3.43001485e-01
-6.85530365e-01 9.09552753e-01 3.73273283e-01 -5.69358349e-01
1.56296745e-01 1.10968933e-01 -1.87401131e-01 5.43273866e-01
6.24800563e-01 1.97781265e-01 -2.84691423e-01 -9.12748277e-01
-4.90037858e-01 5.36153197e-01 -9.86033753e-02 -5.97017594e-02
1.30442607e+00 -1.02508001e-01 2.23993673e-03 1.01799035e+00
1.60662735e+00 -1.62775397e-01 -1.19298875e+00 -7.49472976e-01
1.23054590e-02 -7.01585591e-01 -1.83571681e-01 -5.97555459e-01
-8.96082401e-01 1.27913153e+00 4.94868159e-01 5.42061776e-02
7.66961217e-01 3.91563892e-01 5.77558517e-01 2.46604234e-01
-4.96986024e-02 -9.15404499e-01 2.19300523e-01 4.98192042e-01
9.35311258e-01 -1.41251981e+00 -1.77449718e-01 -2.29956105e-01
-1.01199317e+00 1.00816655e+00 8.08051527e-01 -1.21432364e-01
4.31602567e-01 -1.21351266e-02 9.53922719e-02 -7.72157907e-02
-9.01193559e-01 -4.96550202e-01 7.30130553e-01 6.50842369e-01
4.96176928e-01 2.66316384e-01 2.31788874e-01 7.20793068e-01
5.15511679e-03 -4.06556547e-01 3.77177954e-01 7.76605785e-01
-3.34744632e-01 -1.16770566e+00 -5.48245870e-02 5.49166977e-01
-5.64038381e-02 -2.54631937e-01 -5.17191529e-01 5.62261522e-01
9.14937258e-02 7.12030828e-01 4.19381082e-01 -4.18092847e-01
2.32961223e-01 6.41954839e-01 5.16866088e-01 -8.12271059e-01
-4.31593060e-02 -5.19417822e-01 2.47845456e-01 -6.84189022e-01
-2.03948259e-01 -5.23409545e-01 -1.02708459e+00 -2.24154368e-02
1.15741350e-01 -2.07454711e-01 3.55459332e-01 6.87071562e-01
4.63600904e-01 3.74653667e-01 4.37620372e-01 -8.11039269e-01
-7.77731001e-01 -9.64943409e-01 -5.17422915e-01 9.95122552e-01
5.57314932e-01 -1.07523394e+00 -4.23800349e-01 -9.17035416e-02]
|
[10.159984588623047, 2.226302146911621]
|
c8cec394-75cd-449b-9fc2-6984581ac00f
|
privacy-preserving-in-non-intrusive-load
|
2011.06205
| null |
https://arxiv.org/abs/2011.06205v1
|
https://arxiv.org/pdf/2011.06205v1.pdf
|
Privacy Preserving in Non-Intrusive Load Monitoring: A Differential Privacy Perspective
|
Smart meter devices enable a better understanding of the demand at the potential risk of private information leakage. One promising solution to mitigating such risk is to inject noises into the meter data to achieve a certain level of differential privacy. In this paper, we cast one-shot non-intrusive load monitoring (NILM) in the compressive sensing framework, and bridge the gap between theoretical accuracy of NILM inference and differential privacy's parameters. We then derive the valid theoretical bounds to offer insights on how the differential privacy parameters affect the NILM performance. Moreover, we generalize our conclusions by proposing the hierarchical framework to solve the multi-shot NILM problem. Numerical experiments verify our analytical results and offer better physical insights of differential privacy in various practical scenarios. This also demonstrates the significance of our work for the general privacy preserving mechanism design.
|
['Chenye Wu', 'Chenbei Lu', 'Jiasheng Zhang', 'Haoxiang Wang']
|
2020-11-12
| null | null | null | null |
['non-intrusive-load-monitoring', 'non-intrusive-load-monitoring', 'non-intrusive-load-monitoring']
|
['knowledge-base', 'miscellaneous', 'time-series']
|
[ 3.85958493e-01 -2.30141133e-02 -2.81757116e-01 -3.38025153e-01
-1.04801607e+00 -7.79813111e-01 1.13731317e-01 9.48302522e-02
9.60367173e-03 7.07996666e-01 5.28135896e-01 -3.46263885e-01
-3.21447581e-01 -9.89687681e-01 -4.72844392e-01 -1.00165021e+00
-1.53682679e-01 -1.81018993e-01 -3.90992492e-01 -4.16564941e-02
2.16589823e-01 2.60140955e-01 -5.90766132e-01 -1.80263132e-01
1.07178652e+00 1.14200759e+00 -1.16809025e-01 1.89902157e-01
5.81296027e-01 6.29041553e-01 -6.75168693e-01 -5.80386698e-01
8.07580650e-01 -4.39553618e-01 -3.62950891e-01 4.89179417e-02
-4.06029522e-01 -8.38292301e-01 -6.78930700e-01 1.47279751e+00
6.36626780e-01 -9.90532488e-02 2.21073225e-01 -1.78932869e+00
-7.14521706e-01 1.16970718e+00 -1.00353360e+00 3.07160795e-01
4.25079703e-01 -1.41283825e-01 7.54862010e-01 -2.10306272e-01
-7.41232485e-02 7.31638670e-01 8.85298669e-01 1.03368245e-01
-1.29248357e+00 -9.29804981e-01 -1.85422868e-01 5.27391374e-01
-1.78438210e+00 -6.47811174e-01 8.71977866e-01 8.74459520e-02
4.90390807e-01 7.60475993e-01 2.39391297e-01 6.68434083e-01
7.79310241e-02 8.95446897e-01 1.24258411e+00 -3.23638558e-01
6.89615965e-01 9.44429711e-02 2.27210775e-01 -1.87329546e-01
1.01187515e+00 1.86382562e-01 -5.26127636e-01 -7.88794398e-01
4.29734141e-01 2.50632018e-01 -6.87921286e-01 -2.51184195e-01
-4.95406181e-01 9.27922249e-01 -3.70192807e-03 1.83566853e-01
-4.10890907e-01 1.61842987e-01 3.24817985e-01 3.52952003e-01
1.56330585e-01 -3.49357538e-03 -2.42326379e-01 -2.88738720e-02
-1.05287015e+00 -1.11865014e-01 1.08216953e+00 1.21810782e+00
5.17521918e-01 5.42107150e-02 -1.63546950e-01 1.70211866e-01
2.70580351e-01 5.74520886e-01 8.18690658e-02 -1.04345405e+00
7.75001049e-01 -4.00388166e-02 4.10328060e-01 -9.93844211e-01
-1.02290995e-01 -2.89535791e-01 -1.12734413e+00 -2.12107897e-01
1.72106802e-01 -3.99745196e-01 1.62968159e-01 1.73412907e+00
2.28896607e-02 3.88733178e-01 3.51723105e-01 8.17529202e-01
4.16494571e-02 5.37480354e-01 -1.32353529e-01 -5.65153837e-01
1.31848991e+00 -8.55638459e-02 -1.06356132e+00 7.90926889e-02
6.65097535e-01 -2.94697851e-01 5.68087101e-01 1.01632409e-01
-1.04830241e+00 4.37686682e-01 -1.29596460e+00 3.34285706e-01
1.32882014e-01 -4.54838336e-01 5.91358066e-01 1.41423488e+00
-7.54171789e-01 3.07711124e-01 -6.95306361e-01 -2.22936258e-01
7.06408858e-01 2.96497494e-01 1.38341799e-01 -1.53514937e-01
-1.29605126e+00 4.80966359e-01 2.50158221e-01 -2.08691824e-02
-4.91413921e-01 -9.78075385e-01 -7.36971915e-01 4.23138827e-01
4.78362411e-01 -5.13532758e-01 1.12766230e+00 -1.66270398e-02
-1.11048758e+00 3.65216464e-01 -2.35125516e-02 -1.24523783e+00
7.69036949e-01 1.67575423e-02 -4.66177195e-01 2.56799430e-01
1.75890535e-01 -2.03187034e-01 3.99120659e-01 -1.36127579e+00
-3.70452881e-01 -7.20705390e-01 -4.67435345e-02 -1.03903539e-01
-3.87826920e-01 -9.67227444e-02 2.92078704e-01 -6.85663104e-01
1.05858430e-01 -7.03130841e-01 -5.15484035e-01 -2.51604408e-01
-6.94202006e-01 4.70648795e-01 1.14794779e+00 -6.51602447e-01
1.19103348e+00 -2.30673742e+00 -7.19939709e-01 4.93011534e-01
1.17880210e-01 1.11282222e-01 2.33792588e-01 8.26080501e-01
2.94766366e-01 6.79369550e-03 -4.24472779e-01 -4.01381999e-01
3.86749536e-01 3.69428724e-01 -6.56038404e-01 9.58156765e-01
-8.72621298e-01 1.18367684e+00 -7.69061625e-01 1.30865471e-02
5.05792618e-01 4.60097253e-01 -3.36032242e-01 -2.80889217e-02
3.26132476e-01 5.13685644e-01 -5.94033360e-01 5.60124695e-01
1.47755647e+00 -1.85044542e-01 6.21265590e-01 -2.41482168e-01
1.55824348e-01 5.60907163e-02 -1.31086397e+00 1.28205061e+00
-3.46898884e-01 2.79206783e-01 5.36142290e-01 -1.06654763e+00
6.68147922e-01 3.77829641e-01 6.84460700e-01 -8.43204498e-01
4.11552370e-01 -1.09582946e-01 -5.36895096e-01 -2.20075637e-01
5.11392415e-01 -3.30571055e-01 -4.72021401e-01 7.99549043e-01
-6.29399538e-01 4.06993151e-01 -5.54328620e-01 2.40150481e-01
1.00521219e+00 -7.98374176e-01 5.70328057e-01 -6.60927773e-01
1.94600731e-01 -7.82102108e-01 8.42150033e-01 1.12682939e+00
-5.92244029e-01 8.66513774e-02 1.19256303e-01 1.35161236e-01
-6.25710249e-01 -1.12315702e+00 -2.97161669e-01 4.45282161e-01
4.82119530e-01 -5.18984497e-01 -7.00981617e-01 -2.33466282e-01
3.92807752e-01 1.37856722e+00 -3.19118708e-01 -3.32485735e-01
-1.29120901e-01 -1.26230860e+00 8.39741707e-01 4.88885850e-01
8.88553381e-01 -2.24671792e-02 -6.72585130e-01 1.47589728e-01
-3.51178557e-01 -1.25231290e+00 -4.93417472e-01 2.62211859e-01
-6.82409525e-01 -8.08552206e-01 -8.43255967e-02 -2.56851494e-01
4.37823921e-01 9.54996645e-01 6.32193029e-01 -2.37898454e-01
5.39352857e-02 7.44965315e-01 -2.86421418e-01 -4.36170936e-01
-2.86532521e-01 -2.17727393e-01 1.10741504e-01 2.53969580e-01
6.79426432e-01 -1.29540324e+00 -8.68813515e-01 3.13367248e-01
-9.23186362e-01 -5.11873722e-01 1.17823526e-01 1.90367252e-01
4.49369043e-01 6.54503822e-01 8.52757275e-01 -6.25076711e-01
8.05295646e-01 -8.03049147e-01 -7.29036510e-01 8.65123644e-02
-1.05784881e+00 -1.89697132e-01 6.53071582e-01 2.28510499e-02
-1.00958431e+00 -1.01792909e-01 -8.57075378e-02 7.18694702e-02
1.40618682e-01 2.41058599e-03 -9.20450747e-01 -3.34012598e-01
-6.99516805e-03 4.20296222e-01 -2.83425242e-01 -5.20323336e-01
7.53535450e-01 7.44327724e-01 6.80160701e-01 -5.53004146e-01
1.02235293e+00 9.66579974e-01 7.83735365e-02 -8.38925660e-01
-6.53777540e-01 -3.54464918e-01 -2.56987244e-01 3.13664228e-01
4.33697402e-01 -1.07979536e+00 -1.12681317e+00 4.99306947e-01
-6.29351616e-01 -8.85204598e-03 -6.82706714e-01 3.99791181e-01
-6.67006373e-01 1.05355227e+00 -7.91707873e-01 -1.30429494e+00
-5.07173717e-01 -8.60814929e-01 7.60476887e-01 8.75655785e-02
-4.52313274e-02 -9.86267626e-01 -1.19117267e-01 4.45352197e-01
4.69541728e-01 2.77891517e-01 5.92202306e-01 -6.58580363e-01
-7.89989889e-01 -3.28647673e-01 -1.57783434e-01 8.38846341e-02
1.70317695e-01 -1.05858719e+00 -1.04613614e+00 -8.23154926e-01
8.99013102e-01 2.24713922e-01 4.85794932e-01 4.92014140e-01
1.35847485e+00 -8.82514060e-01 -2.21148103e-01 8.87637854e-01
1.81921315e+00 1.46317244e-01 1.13493133e+00 7.61525482e-02
3.97573680e-01 -1.54685779e-02 2.72766769e-01 1.37144208e+00
7.15329468e-01 3.81408691e-01 4.83900696e-01 2.54577130e-01
6.70476913e-01 -5.33837676e-01 2.30527416e-01 7.99372673e-01
5.91507554e-01 -5.04972637e-01 1.30538698e-02 5.66236913e-01
-1.85879683e+00 -1.22355461e+00 -2.92110056e-01 2.52227044e+00
5.86899161e-01 -3.14083248e-01 1.85178787e-01 6.68848991e-01
7.91223407e-01 4.10142630e-01 -5.96490622e-01 -2.27010354e-01
-5.00214219e-01 -4.21365835e-02 1.39520156e+00 6.09127641e-01
-7.77434051e-01 1.19175069e-01 6.71920967e+00 7.10167408e-01
-3.67983252e-01 4.44558293e-01 6.16493762e-01 -7.22132251e-02
-6.57674670e-01 2.00421721e-01 -5.40943921e-01 7.95372903e-01
1.02625060e+00 -8.94342840e-01 5.30387640e-01 4.67281222e-01
5.34196198e-01 -3.46996695e-01 -1.00012183e+00 1.23377192e+00
1.72073022e-02 -1.21252370e+00 -4.46542650e-01 6.99975669e-01
7.21324027e-01 -1.94451630e-01 -1.68099895e-01 -2.13147148e-01
6.44614875e-01 -5.93094230e-01 3.02634090e-01 4.43907559e-01
4.39602226e-01 -1.02103412e+00 4.19977516e-01 5.90213895e-01
-1.29277265e+00 -3.99039000e-01 -4.74119902e-01 -3.13128531e-01
7.07411826e-01 1.02182186e+00 -3.85992259e-01 8.54001164e-01
5.07471383e-01 2.18009919e-01 -9.75050554e-02 9.63512659e-01
-1.60010919e-01 7.42110908e-01 -6.91702902e-01 2.81130522e-01
-1.75648198e-01 -4.61694092e-01 6.10980392e-01 9.28176045e-01
5.12509286e-01 4.90965694e-01 1.00640260e-01 9.89508450e-01
-1.91290408e-01 -1.26817048e-01 -7.01026678e-01 4.61239785e-01
1.26869917e+00 8.03620636e-01 -3.68831784e-01 -6.72820359e-02
-5.48584998e-01 1.23484862e+00 -3.98007780e-01 3.47692758e-01
-6.64422333e-01 -1.69641435e-01 9.45830941e-01 5.23922406e-02
3.98899198e-01 -6.02356754e-02 -1.02773499e+00 -1.28602505e+00
2.91222543e-01 -5.27002454e-01 6.11412823e-01 -2.96686083e-01
-1.60621178e+00 -6.38493061e-01 1.42825879e-02 -1.12822819e+00
9.24935862e-02 3.23073685e-01 -8.75412822e-01 6.65912807e-01
-1.42627144e+00 -6.87948048e-01 1.45024329e-01 7.86046743e-01
-2.67845094e-01 2.63538539e-01 8.52993906e-01 5.06485522e-01
-4.23239112e-01 1.01836550e+00 8.14114034e-01 6.75055534e-02
-6.94614695e-03 -8.96289945e-01 4.34680223e-01 1.15272903e+00
-1.34830102e-01 4.06444073e-01 7.92112529e-01 -5.73606312e-01
-1.76855218e+00 -1.03615165e+00 6.16249144e-01 -8.30493569e-02
6.78965688e-01 -5.59430957e-01 -8.65167737e-01 8.54741991e-01
2.47180372e-01 -1.14187971e-01 1.31502342e+00 -4.88988638e-01
-3.12649906e-01 -2.40848243e-01 -2.04227734e+00 2.49298170e-01
9.22116756e-01 -7.43153274e-01 -3.66993159e-01 2.25047410e-01
6.09679759e-01 1.41978860e-01 -1.15339446e+00 1.05418032e-02
3.15843731e-01 -1.01733005e+00 8.94288838e-01 3.61258499e-02
-6.37884676e-01 -3.20897788e-01 -1.07721925e+00 -1.00960505e+00
-1.48572147e-01 -1.29732800e+00 -3.83724660e-01 1.60890901e+00
-1.75571248e-01 -1.09917140e+00 8.60427201e-01 1.01156092e+00
5.08899868e-01 -1.82307884e-01 -1.25823772e+00 -9.10837233e-01
2.07522318e-01 -7.84468234e-01 1.33742118e+00 1.15098572e+00
3.53910118e-01 -2.08222970e-01 -8.55771184e-01 6.88922703e-01
1.37788844e+00 1.87862992e-01 5.42668343e-01 -9.79732335e-01
-3.64732146e-01 2.08091691e-01 -4.70591098e-01 -1.35337102e+00
-1.03052901e-02 -9.08060610e-01 -4.93956417e-01 -1.07508361e+00
4.67453182e-01 -3.10083866e-01 -4.98845488e-01 1.13535695e-01
1.79634601e-01 2.07053870e-01 4.23085749e-01 1.19635127e-01
-5.23600757e-01 7.34112024e-01 5.61529279e-01 -1.02070989e-02
7.58692995e-02 2.77925730e-01 -1.53942215e+00 3.70900631e-01
1.05388486e+00 -4.65147823e-01 -7.23521769e-01 -2.59969711e-01
1.26917837e-02 4.42469478e-01 4.18410987e-01 -6.64164484e-01
1.26801729e-01 -1.83036074e-01 8.05990845e-02 -8.29416335e-01
1.11812465e-01 -1.29885268e+00 2.74198651e-01 6.21132016e-01
-1.09989166e-01 -2.11698160e-01 -1.80915609e-01 8.72107983e-01
4.79516506e-01 -1.98039368e-01 7.37807572e-01 8.47444683e-02
-2.97163129e-01 2.96832323e-01 -3.65084499e-01 2.27743775e-01
1.15516591e+00 -2.30871618e-01 -3.70241761e-01 -1.01053965e+00
-6.69749022e-01 5.31786680e-01 6.67836607e-01 -1.95393693e-02
2.23557234e-01 -1.51337445e+00 -6.45803928e-01 3.32978904e-01
-1.54105633e-01 -6.81212485e-01 4.19152915e-01 8.39826465e-01
2.65694141e-01 2.64258206e-01 2.93536425e-01 -3.04421037e-01
-7.81163096e-01 7.31890559e-01 2.69817114e-01 -7.06269741e-02
-8.63466740e-01 1.43229827e-01 2.48936098e-02 2.08371267e-01
3.08367074e-01 -2.67509669e-02 6.85186207e-01 -2.43428648e-01
9.38845217e-01 9.21197116e-01 5.37882037e-02 -6.47270799e-01
-3.65640551e-01 1.17282562e-01 2.31380165e-01 -6.05367161e-02
1.23705697e+00 -1.06111872e+00 -6.41708151e-02 2.56273877e-02
1.38756418e+00 3.41307133e-01 -1.19399095e+00 -4.11898911e-01
-3.11991991e-03 -8.90595973e-01 7.78511316e-02 -4.94909406e-01
-1.34792244e+00 3.81820142e-01 5.68981946e-01 5.61415434e-01
1.31644595e+00 -1.58695608e-01 1.28986204e+00 1.35296077e-01
9.93371725e-01 -1.02270341e+00 -7.61290431e-01 -2.68869013e-01
2.00222582e-01 -8.67722571e-01 1.78504556e-01 -4.92826909e-01
-3.94263864e-01 4.38048691e-01 -2.94531405e-01 -9.00608599e-02
9.50492084e-01 8.04339468e-01 -2.18024433e-01 5.66918999e-02
-2.65054703e-01 2.66705573e-01 -6.28356397e-01 9.81229842e-01
-1.66647270e-01 7.46202826e-01 -6.46958768e-01 1.16114426e+00
-3.62980455e-01 -5.41928560e-02 9.43822086e-01 1.23308575e+00
-1.73775330e-01 -1.26535690e+00 -6.23093545e-01 4.20041382e-01
-7.10059106e-01 6.52024755e-03 -1.32903293e-01 2.29878932e-01
-3.45787644e-01 1.51950407e+00 -2.55404502e-01 -2.52846241e-01
2.08882034e-01 -1.48036957e-01 4.57315236e-01 2.03885332e-01
-1.01841331e-01 -1.29121512e-01 -1.72126636e-01 -3.86202961e-01
-1.75310686e-01 -9.96784508e-01 -1.08127725e+00 -1.00264907e+00
-3.71188343e-01 4.53200191e-01 2.46828780e-01 8.98588836e-01
5.20752251e-01 -5.18480986e-02 1.20912898e+00 -2.99272448e-01
-1.25081921e+00 -3.76699924e-01 -1.24343228e+00 4.19082910e-01
2.28526756e-01 -1.06700711e-01 -5.31279027e-01 -5.03431439e-01]
|
[5.906966209411621, 6.53985071182251]
|
df491114-b13d-4c9d-b165-7191318f136f
|
kernel-metric-learning-for-clustering-mixed
|
2306.0189
| null |
https://arxiv.org/abs/2306.01890v1
|
https://arxiv.org/pdf/2306.01890v1.pdf
|
Kernel Metric Learning for Clustering Mixed-type Data
|
Distance-based clustering and classification are widely used in various fields to group mixed numeric and categorical data. A predefined distance measurement is used to cluster data points based on their dissimilarity. While there exist numerous distance-based measures for data with pure numerical attributes and several ordered and unordered categorical metrics, an optimal distance for mixed-type data is an open problem. Many metrics convert numerical attributes to categorical ones or vice versa. They handle the data points as a single attribute type or calculate a distance between each attribute separately and add them up. We propose a metric that uses mixed kernels to measure dissimilarity, with cross-validated optimal kernel bandwidths. Our approach improves clustering accuracy when utilized for existing distance-based clustering algorithms on simulated and real-world datasets containing pure continuous, categorical, and mixed-type data.
|
['John R. J. Thompson', 'Jesse S. Ghashti']
|
2023-06-02
| null | null | null | null |
['metric-learning', 'metric-learning']
|
['computer-vision', 'methodology']
|
[-2.89238602e-01 -6.38994217e-01 -3.18096101e-01 -8.72570574e-01
-6.30665243e-01 -7.96962678e-01 3.19235772e-01 1.07122958e+00
-6.49285018e-01 4.92679864e-01 1.07777759e-01 -4.08232093e-01
-9.52613175e-01 -1.28110230e+00 1.49254575e-01 -7.24950790e-01
-6.61526501e-01 8.49130690e-01 1.82060122e-01 1.88126251e-01
5.12084663e-01 5.27506948e-01 -1.81525755e+00 1.45327166e-01
1.30957031e+00 8.30370188e-01 -2.63271600e-01 6.10241055e-01
-3.18273246e-01 1.36425778e-01 -5.90926111e-01 -1.82237312e-01
3.37120324e-01 -2.76045680e-01 -6.97034717e-01 -2.65457124e-01
-4.29691784e-02 9.38905999e-02 2.27368489e-01 9.70569968e-01
4.45440501e-01 3.40369225e-01 1.46729517e+00 -1.77927959e+00
-1.03671288e+00 5.29614270e-01 -7.16260016e-01 -1.06197380e-01
2.76612461e-01 -3.07805419e-01 8.86369705e-01 -6.56783521e-01
7.12439120e-02 1.15727341e+00 8.31693888e-01 6.23259433e-02
-1.57571518e+00 -4.80019361e-01 -3.15413773e-01 4.68712687e-01
-1.87156677e+00 1.34163111e-01 5.90010762e-01 -6.63162112e-01
5.77825010e-01 5.29906631e-01 3.82799864e-01 1.83470652e-01
-1.29954875e-01 5.07037304e-02 1.30143940e+00 -2.85042197e-01
6.70844078e-01 1.66425869e-01 4.76699829e-01 1.12814620e-01
4.65373009e-01 -1.39604613e-01 3.77297580e-01 -5.46317101e-01
2.00053439e-01 3.76058102e-01 1.67319432e-01 -7.72161245e-01
-1.32880116e+00 1.11098099e+00 2.38533258e-01 3.40594023e-01
-3.73909891e-01 -3.31470251e-01 5.43009818e-01 4.61193562e-01
3.02193195e-01 2.93489069e-01 -2.83745021e-01 -2.51929700e-01
-9.86859918e-01 9.01614279e-02 8.40770006e-01 9.11251485e-01
9.64343250e-01 -4.19505656e-01 -2.45060585e-02 1.30318236e+00
1.47403330e-01 2.16268331e-01 5.39506793e-01 -9.09889042e-01
2.21361518e-01 9.72739160e-01 6.27392083e-02 -1.27106988e+00
-5.73628247e-01 1.82989761e-01 -1.12442636e+00 3.67305040e-01
6.70302093e-01 1.64185986e-01 -5.82867920e-01 1.61655796e+00
5.38194120e-01 -5.57310414e-04 1.15297280e-01 7.02998579e-01
5.12857795e-01 6.83812857e-01 1.53401271e-01 -3.28140378e-01
1.12717009e+00 -1.80270642e-01 -5.18332660e-01 7.18502462e-01
7.94924021e-01 -6.28621578e-01 1.00615764e+00 3.36021274e-01
-8.60145688e-01 -5.05962908e-01 -7.99634039e-01 1.36197895e-01
-9.35226440e-01 -1.83713838e-01 4.36041713e-01 9.06769931e-01
-1.15571797e+00 5.02046943e-01 -5.85626841e-01 -4.03439552e-01
9.60153043e-02 4.82268780e-01 -3.59545946e-01 1.20795533e-01
-1.09015501e+00 5.38068354e-01 4.59070385e-01 -4.41321433e-01
1.24730663e-02 -7.97259688e-01 -5.68355024e-01 1.43789575e-01
-5.95017970e-02 -3.61644596e-01 4.13401008e-01 -3.71389687e-01
-9.00935531e-01 7.74510443e-01 1.11910291e-01 -1.85388714e-01
2.94305593e-01 6.85176909e-01 -8.21769834e-01 3.62338610e-02
2.11123317e-01 3.15946162e-01 2.40181655e-01 -1.09834647e+00
-8.00004959e-01 -4.76228535e-01 -3.03095639e-01 6.94737956e-02
-5.05327582e-01 -1.38422713e-01 1.48560613e-01 -7.95178056e-01
3.14430743e-01 -5.66356063e-01 -1.66213855e-01 -1.01480752e-01
-1.36486501e-01 -5.62816739e-01 5.87190509e-01 -3.62821549e-01
1.57592368e+00 -2.14976358e+00 9.18706059e-02 7.34152436e-01
2.55075365e-01 -3.53891514e-02 1.33766294e-01 6.07708335e-01
-1.80434078e-01 1.61852151e-01 -6.01501107e-01 2.79416442e-01
8.95118937e-02 8.85552391e-02 3.58290374e-01 6.41027272e-01
-1.38452187e-01 2.76978165e-01 -1.00364172e+00 -8.14824879e-01
3.39266717e-01 4.35591429e-01 -4.54987139e-01 1.06195875e-01
5.01552224e-01 -1.75179899e-01 -9.07745510e-02 6.66129827e-01
9.56750274e-01 1.92488924e-01 -2.29274705e-02 -2.75506020e-01
-1.89330831e-01 -3.13633054e-01 -1.69532168e+00 1.14338136e+00
-2.67270088e-01 3.24118942e-01 -1.34564742e-01 -1.49064958e+00
1.37135053e+00 7.23180696e-02 1.14962101e+00 -4.80096303e-02
1.41736299e-01 2.99726158e-01 2.88406104e-01 -3.43950003e-01
5.01233995e-01 -1.26960635e-01 -3.24318498e-01 5.09958744e-01
-2.76303619e-01 1.54297411e-01 4.26148862e-01 8.78195241e-02
1.16654706e+00 -5.94535351e-01 4.18452799e-01 -6.29210770e-01
7.04940975e-01 2.06133425e-01 3.78929973e-01 5.23712456e-01
-2.32458711e-01 7.92920530e-01 4.69412446e-01 -1.91259980e-01
-1.12738645e+00 -1.51647377e+00 -4.27924871e-01 8.77962589e-01
1.39782980e-01 -2.25221753e-01 -5.76713443e-01 -4.53361660e-01
4.39251930e-01 5.19538701e-01 -7.29919374e-01 -2.09260538e-01
-4.88872454e-02 -9.68573511e-01 5.30644000e-01 4.42164660e-01
3.57616305e-01 -6.39873743e-01 -2.36090109e-01 3.77657741e-01
-6.20394535e-02 -3.73993576e-01 -5.07648408e-01 2.94720411e-01
-8.32502007e-01 -1.24471724e+00 -4.90382195e-01 -9.04245257e-01
6.60401821e-01 2.26948485e-01 8.13972175e-01 -6.32527545e-02
-5.11004627e-01 2.30288804e-01 -5.06999195e-01 1.84794571e-02
-3.07405859e-01 3.01074237e-03 3.22658300e-01 2.87877023e-01
9.18847263e-01 -5.06267130e-01 -6.48262322e-01 7.23341525e-01
-9.98061180e-01 -6.11762166e-01 3.71564001e-01 7.11040854e-01
4.36203152e-01 4.46819454e-01 7.65119672e-01 -5.49145401e-01
8.36043358e-01 -9.49847400e-01 -3.60160947e-01 2.77373463e-01
-8.36731493e-01 6.24719681e-03 9.03792024e-01 -6.91114187e-01
-5.70980847e-01 -1.34551555e-01 2.67487854e-01 -1.28858283e-01
-4.59334850e-01 5.62218547e-01 -3.17634106e-01 2.13007301e-01
6.28289700e-01 1.39033929e-01 3.38717043e-01 -3.69009852e-01
3.52538794e-01 1.29307401e+00 5.58115542e-01 -8.28951716e-01
6.81090176e-01 5.36486864e-01 8.85200649e-02 -6.65778041e-01
5.91272675e-02 -9.14445937e-01 -1.13892889e+00 5.20101525e-02
9.38011825e-01 -3.34859431e-01 -1.11556005e+00 3.63363773e-01
-5.55035174e-01 4.31059673e-02 -1.89996108e-01 7.36996412e-01
-5.83667815e-01 4.30277377e-01 -2.76967496e-01 -7.40011752e-01
7.43460208e-02 -8.34683836e-01 5.39109647e-01 -1.44916372e-02
-3.61222774e-01 -1.31632626e+00 1.84822947e-01 -1.64642513e-01
3.25475514e-01 3.97446513e-01 1.25729573e+00 -1.02519786e+00
3.14599067e-01 -3.69515419e-01 -4.43156153e-01 2.12269649e-01
6.90713644e-01 2.97395349e-01 -2.53465772e-01 -3.39635342e-01
-2.85878628e-01 4.67473902e-02 4.44273591e-01 3.90474707e-01
1.47676837e+00 -3.04907650e-01 -4.00360674e-01 5.87319076e-01
1.55423188e+00 5.58795214e-01 4.42927569e-01 2.25766689e-01
6.00725889e-01 8.83733690e-01 8.37394416e-01 5.25522470e-01
6.03705108e-01 4.79609132e-01 9.56383348e-02 -1.23509422e-01
4.63201970e-01 3.14216554e-01 -1.87187672e-01 9.12609994e-01
-1.35938212e-01 2.02331729e-02 -1.18461359e+00 6.83535516e-01
-1.86974263e+00 -1.29001057e+00 -6.41620517e-01 2.69898319e+00
9.30586278e-01 -1.91480845e-01 8.11944544e-01 7.41151571e-01
1.36177135e+00 -5.77067077e-01 -4.04512376e-01 -7.02314854e-01
2.49996223e-02 -4.15998511e-02 4.99082774e-01 4.79144335e-01
-9.79748011e-01 2.06822485e-01 6.42841244e+00 8.29982221e-01
-7.56374478e-01 -1.98212281e-01 5.77087581e-01 1.34191185e-01
-6.09101020e-02 -1.15136042e-01 -3.86959076e-01 7.48218894e-01
9.88233805e-01 -5.61238408e-01 2.26983503e-01 7.56341100e-01
2.17696697e-01 -2.56074846e-01 -1.19198096e+00 1.08618057e+00
-3.24611604e-01 -8.53296518e-01 -1.10263400e-01 3.95262361e-01
4.81768847e-01 -6.79749846e-01 1.24568373e-01 1.92834914e-01
6.80047810e-01 -1.13090992e+00 1.01153351e-01 4.71706778e-01
8.98505509e-01 -1.12622273e+00 6.71352446e-01 1.46094993e-01
-1.40947473e+00 -1.63553748e-02 -6.41882539e-01 -1.37561843e-01
-2.55152136e-01 7.05232024e-01 -5.05366743e-01 6.33311450e-01
7.59836376e-01 6.74533486e-01 -6.94881082e-01 1.33861220e+00
8.89815927e-01 4.68060046e-01 -3.70744556e-01 -2.42390528e-01
1.86162069e-01 -8.18388224e-01 3.83074731e-02 1.11914802e+00
5.86093783e-01 2.44054675e-01 1.79149821e-01 6.27869368e-01
4.88810509e-01 3.27770501e-01 -4.53382373e-01 3.02100629e-01
1.18542087e+00 1.19241440e+00 -1.07037377e+00 -4.30934906e-01
-4.74459469e-01 6.47577345e-01 1.97333843e-02 2.27217823e-01
-9.48814392e-01 -1.16182470e+00 9.15928662e-01 3.16812456e-01
-1.15070552e-01 -3.96159172e-01 -5.28387666e-01 -7.22032905e-01
-1.45157561e-01 -4.92980242e-01 1.03782594e+00 -2.38987610e-01
-1.80438566e+00 1.87604651e-01 5.49097121e-01 -1.67300737e+00
-8.83971527e-02 -4.72150445e-01 -4.67799664e-01 9.03844357e-01
-7.80084848e-01 -5.88392735e-01 -4.44693774e-01 8.34076107e-01
-1.17759302e-01 -1.37273476e-01 8.56305242e-01 4.28143680e-01
-1.55252680e-01 6.47479057e-01 8.66216183e-01 3.95639390e-01
9.56996560e-01 -1.60933638e+00 -1.45303860e-01 1.90773726e-01
-2.86161035e-01 8.45803559e-01 5.48905194e-01 -4.88172084e-01
-1.08269393e+00 -1.03721368e+00 6.23810649e-01 -3.34396780e-01
6.35029197e-01 -3.69621575e-01 -1.33362341e+00 6.70035481e-02
-1.54851392e-01 8.03290755e-02 1.24207222e+00 1.10825382e-01
-4.33731884e-01 -3.71857971e-01 -1.76986802e+00 2.61423796e-01
8.34934771e-01 -1.98918730e-01 -5.65442801e-01 -3.49564143e-02
1.67916849e-01 4.57507193e-01 -1.70401430e+00 4.25499439e-01
4.88284349e-01 -1.22115672e+00 1.14544475e+00 -2.42741525e-01
8.05271491e-02 -6.97251320e-01 -3.17504048e-01 -1.59035468e+00
-5.86388171e-01 -5.48808649e-02 4.34745044e-01 1.53017747e+00
1.38632357e-01 -8.21426213e-01 5.34405768e-01 4.93294656e-01
1.54445469e-01 -5.47416389e-01 -8.44989657e-01 -1.14649534e+00
3.49951148e-01 -2.27094010e-01 9.99983609e-01 1.40607190e+00
3.74733120e-01 -7.36594349e-02 2.17020676e-01 1.61023103e-02
1.07135403e+00 1.56305730e-01 5.56391120e-01 -1.79219675e+00
1.29024252e-01 -8.97375584e-01 -1.04094553e+00 2.33206786e-02
-1.05729066e-01 -9.74559903e-01 -2.63420612e-01 -1.46616650e+00
-5.63657507e-02 -1.30979788e+00 -2.77346939e-01 3.15660089e-01
-9.02580544e-02 2.86264360e-01 -3.60310018e-01 3.52108419e-01
-1.97809815e-01 9.78256464e-02 3.60260040e-01 -8.84542391e-02
-4.46162760e-01 -3.11695859e-02 -4.03903633e-01 3.75472844e-01
9.58301306e-01 -4.31805313e-01 -5.71888506e-01 1.80134892e-01
-7.40203187e-02 -8.86772051e-02 1.67388827e-01 -1.25423741e+00
3.41420949e-01 -5.79237938e-01 3.04546028e-01 -7.67941117e-01
-5.53472303e-02 -1.14681613e+00 5.05687118e-01 3.91402572e-01
-4.19550538e-01 2.05216765e-01 -2.86007673e-01 4.16340679e-01
-3.25025052e-01 1.04095116e-02 1.15992033e+00 2.16997683e-01
-5.15180349e-01 2.97351897e-01 -3.06551009e-01 -5.00100106e-02
1.56863606e+00 -7.48539865e-01 -2.22743616e-01 -1.36068493e-01
-9.63025153e-01 2.44477525e-01 8.29587400e-01 2.47191861e-01
5.98806322e-01 -1.85903966e+00 -7.78196573e-01 1.83746479e-02
4.66736555e-01 -3.36048454e-01 -4.72353809e-02 8.06180954e-01
-4.98021215e-01 7.29422048e-02 -6.62533522e-01 -7.90030599e-01
-1.40498805e+00 1.00606287e+00 7.60087594e-02 3.59638840e-01
-1.30674511e-01 2.44548663e-01 -3.09113175e-01 -6.85940683e-01
2.26691753e-01 -3.07400703e-01 -3.41360718e-01 6.28935754e-01
2.81433821e-01 1.03572178e+00 1.42885923e-01 -6.52440488e-01
-6.09538972e-01 8.26093972e-01 3.99736226e-01 -1.37074247e-01
1.14181089e+00 -2.24179685e-01 -4.20991272e-01 9.06122386e-01
1.74884999e+00 -2.63266534e-01 -5.51429629e-01 -1.09698109e-01
3.40013146e-01 -6.67087197e-01 -4.95291203e-01 -3.74949902e-01
-5.18810689e-01 6.56782150e-01 6.80866063e-01 8.87512028e-01
1.25477874e+00 -2.31364444e-02 2.29868665e-01 1.05917282e-01
3.70013475e-01 -1.31832731e+00 -1.61975160e-01 1.37647703e-01
2.31241480e-01 -1.14861965e+00 -7.58324638e-02 -3.98373127e-01
-5.76333761e-01 1.07602811e+00 3.70336354e-01 -2.26233751e-01
9.59249616e-01 2.72439688e-01 -1.99590884e-02 1.64634347e-01
-2.88914263e-01 -2.38422021e-01 2.74961650e-01 1.07145607e+00
6.20597422e-01 3.22420180e-01 -8.48334432e-01 3.17298323e-01
-2.58917391e-01 -3.52161258e-01 4.59679842e-01 8.64971876e-01
-6.33706093e-01 -1.13590682e+00 -7.77644396e-01 8.94137919e-01
-1.57521904e-01 2.44823873e-01 -2.52576500e-01 7.36255050e-01
6.96472675e-02 1.19171500e+00 5.71649134e-01 -5.42966664e-01
1.98031455e-01 8.56513605e-02 1.47315949e-01 -3.92400950e-01
-5.58943272e-01 -4.51809347e-01 -3.05661976e-01 -1.34017050e-01
-4.33468908e-01 -7.96026230e-01 -1.45154727e+00 -9.10269916e-01
-8.58510733e-02 6.16712689e-01 7.24074423e-01 3.38134348e-01
2.10047454e-01 1.36969477e-01 1.10383856e+00 -2.90049732e-01
-4.39895242e-01 -7.52482116e-01 -9.11951125e-01 7.09705353e-01
1.93490118e-01 -8.11553597e-01 -5.96704841e-01 -6.03803471e-02]
|
[7.583483695983887, 4.5744099617004395]
|
ee5ffc2a-5198-4129-8b97-2f2cf680eff8
|
unobtrusive-pain-monitoring-in-older-adults
|
2101.03251
| null |
https://arxiv.org/abs/2101.03251v1
|
https://arxiv.org/pdf/2101.03251v1.pdf
|
Unobtrusive Pain Monitoring in Older Adults with Dementia using Pairwise and Contrastive Training
|
Although pain is frequent in old age, older adults are often undertreated for pain. This is especially the case for long-term care residents with moderate to severe dementia who cannot report their pain because of cognitive impairments that accompany dementia. Nursing staff acknowledge the challenges of effectively recognizing and managing pain in long-term care facilities due to lack of human resources and, sometimes, expertise to use validated pain assessment approaches on a regular basis. Vision-based ambient monitoring will allow for frequent automated assessments so care staff could be automatically notified when signs of pain are displayed. However, existing computer vision techniques for pain detection are not validated on faces of older adults or people with dementia, and this population is not represented in existing facial expression datasets of pain. We present the first fully automated vision-based technique validated on a dementia cohort. Our contributions are threefold. First, we develop a deep learning-based computer vision system for detecting painful facial expressions on a video dataset that is collected unobtrusively from older adult participants with and without dementia. Second, we introduce a pairwise comparative inference method that calibrates to each person and is sensitive to changes in facial expression while using training data more efficiently than sequence models. Third, we introduce a fast contrastive training method that improves cross-dataset performance. Our pain estimation model outperforms baselines by a wide margin, especially when evaluated on faces of people with dementia. Pre-trained model and demo code available at https://github.com/TaatiTeam/pain_detection_demo
|
['Babak Taati', 'Thomas Hadjistavropoulos', 'Kenneth M. Prkachin', 'Shun Zhao', 'Abhishek Moturu', 'Siavash Rezaei']
|
2021-01-08
| null | null | null | null |
['pain-intensity-regression']
|
['medical']
|
[ 6.98926523e-02 -2.45973185e-01 -3.35747272e-01 -4.74571079e-01
-1.22249663e+00 -1.83235839e-01 -2.12758467e-01 -6.53477237e-02
-1.16840053e+00 1.03690827e+00 6.43255353e-01 2.57905573e-01
2.20615417e-01 -4.30878282e-01 -1.92001089e-01 -1.60961673e-01
-3.95833224e-01 4.63935494e-01 -4.91674423e-01 -8.46485943e-02
-2.31845558e-01 3.22138816e-01 -1.44291234e+00 5.23441076e-01
4.41083997e-01 1.10266733e+00 2.14523986e-01 4.81444955e-01
3.66123587e-01 6.26475334e-01 -4.29681510e-01 3.10983043e-02
-1.30383093e-02 -2.67430276e-01 -7.42430866e-01 -1.22477524e-01
6.03033364e-01 -1.41277742e+00 -1.59062833e-01 6.44873559e-01
1.02628350e+00 -1.25134081e-01 5.51442444e-01 -1.08900976e+00
-1.92387119e-01 -2.94844508e-01 -6.07175946e-01 4.40887064e-01
7.53764272e-01 2.36596137e-01 5.50733447e-01 -7.35166073e-01
5.48184216e-01 1.23970294e+00 8.28305364e-01 1.04594898e+00
-1.38312876e+00 -9.49467838e-01 6.58619925e-02 5.38397908e-01
-1.17856085e+00 -5.74409127e-01 4.90068853e-01 -4.96448785e-01
1.12839055e+00 1.76799539e-02 1.42589772e+00 1.80984282e+00
2.11472586e-01 4.66161460e-01 1.53223419e+00 -1.30518779e-01
6.15571141e-01 -4.92372781e-01 1.07269064e-01 5.67027450e-01
1.75282285e-01 -1.32977024e-01 -5.86624444e-01 -7.35820353e-01
5.99132240e-01 2.81984746e-01 -6.22403145e-01 2.05261391e-02
-7.94443846e-01 8.96373749e-01 5.07669568e-01 9.52254906e-02
-6.14545703e-01 5.11303306e-01 6.25410736e-01 1.83430478e-01
2.21464053e-01 -1.66927189e-01 -3.38894993e-01 -7.43059993e-01
-8.93748403e-01 3.52416426e-01 5.84242642e-01 3.65002334e-01
2.98853636e-01 -5.61344981e-01 -2.54053883e-02 1.18662262e+00
2.52138257e-01 4.98697639e-01 5.10984600e-01 -1.77975523e+00
-5.06603569e-02 2.71370828e-01 1.66548982e-01 -5.90225399e-01
-8.49239051e-01 1.10121697e-01 -6.10706151e-01 9.98003185e-01
1.20427802e-01 -4.77845311e-01 -6.78813100e-01 2.01068282e+00
-2.10823361e-02 -5.83645821e-01 -2.73192614e-01 1.00984406e+00
3.23496044e-01 -2.03037187e-01 4.26977038e-01 -6.17996156e-01
1.97909725e+00 -4.18147326e-01 -7.74655640e-01 -7.62162268e-01
5.75841427e-01 -2.85455883e-01 1.14551938e+00 7.85982072e-01
-7.27638721e-01 7.16397830e-04 -9.14484739e-01 -8.32796544e-02
8.10173452e-02 -1.27768978e-01 6.13711357e-01 7.41563261e-01
-1.23272347e+00 3.54204088e-01 -1.02949929e+00 -8.08272600e-01
1.11327374e+00 4.78017300e-01 -8.71195555e-01 -3.10317069e-01
-9.14836466e-01 1.31006503e+00 -8.22724625e-02 2.37757891e-01
-3.96637648e-01 -5.33184469e-01 -7.49991179e-01 -4.58686590e-01
-2.80013829e-01 -1.07128513e+00 1.78646600e+00 -1.36304069e+00
-9.69316840e-01 1.12117290e+00 -2.91346312e-01 -2.51114577e-01
8.23561907e-01 -9.16653752e-01 -1.16191842e-01 7.01493382e-01
4.18247551e-01 1.22795594e+00 5.78210175e-01 -9.70780313e-01
-2.07717523e-01 -8.59937489e-01 -5.14720492e-02 4.40224826e-01
-2.63228595e-01 1.49932399e-01 1.06607042e-01 -4.11690593e-01
-4.16400701e-01 -9.14054334e-01 -1.01248130e-01 1.14463687e+00
2.74793983e-01 2.54049301e-01 1.07794344e+00 -9.24644113e-01
7.33186960e-01 -2.17799497e+00 -2.52072453e-01 -3.36305089e-02
1.75394684e-01 -1.70607522e-01 3.71548057e-01 9.84218568e-02
-7.04130381e-02 -4.03232463e-02 -4.91768032e-01 -5.78778744e-01
-2.04659954e-01 5.17622411e-01 3.47068012e-01 5.03500998e-01
1.19788535e-01 6.97055995e-01 -8.15129757e-01 -5.52750230e-01
1.38558879e-01 7.34530568e-01 -5.37754774e-01 -5.71995266e-02
1.10363521e-01 4.87185270e-01 -7.34460652e-02 8.17243934e-01
4.16540980e-01 1.37981802e-01 1.22667767e-01 -8.91933888e-02
2.06043810e-01 -1.78026617e-01 -6.55632555e-01 2.05424118e+00
-3.11378717e-01 2.41206408e-01 6.28303409e-01 -5.94050527e-01
5.19695580e-01 3.09808612e-01 6.21141791e-01 -7.16404796e-01
4.14784551e-01 3.31329137e-01 -1.79761007e-01 -8.60760808e-01
-2.07323089e-01 -8.38478327e-01 1.53440565e-01 4.31483865e-01
-5.07011175e-01 2.16523170e-01 -2.15817973e-01 -3.82791720e-02
1.61751914e+00 5.26337959e-02 4.13638562e-01 -9.09299869e-03
-1.81697071e-01 -1.17064618e-01 4.10656035e-01 4.78520840e-01
-8.51104558e-01 8.84391904e-01 3.55064452e-01 -3.18677992e-01
-7.68485606e-01 -1.01019537e+00 -3.54407519e-01 8.74323189e-01
-5.98838389e-01 -2.37343177e-01 -5.01554370e-01 -5.91916107e-02
1.97440356e-01 4.33447748e-01 -8.08464408e-01 -3.40137362e-01
-1.65468633e-01 -6.49942875e-01 5.63671768e-01 1.00572920e+00
5.96201837e-01 -1.08630085e+00 -1.38614821e+00 1.45545751e-01
-5.15784383e-01 -6.34892702e-01 -1.35329977e-01 3.91673714e-01
-8.19911063e-01 -1.05625534e+00 -1.26557350e+00 -8.37430716e-01
4.45089430e-01 -8.99201035e-02 5.42033732e-01 -2.14758843e-01
-7.90230870e-01 1.34758139e+00 -2.22116783e-01 -4.19539869e-01
1.94174394e-01 -5.62805891e-01 3.27985942e-01 -5.10102212e-01
7.46781588e-01 -9.13912654e-01 -1.16366041e+00 -1.54080957e-01
-4.57819819e-01 -3.06049794e-01 6.49164975e-01 6.95488393e-01
6.42369270e-01 -7.45951533e-01 5.27002871e-01 -9.93995890e-02
1.01452410e+00 -5.44217467e-01 3.57646674e-01 -2.48103783e-01
-3.61007005e-01 -1.26163900e-01 -1.26028031e-01 -3.91829103e-01
-7.90445447e-01 1.60993844e-01 -2.93800592e-01 -4.53608751e-01
-3.24378222e-01 4.90464449e-01 1.51948631e-01 2.00108007e-01
8.43533814e-01 -5.12004137e-01 6.66539431e-01 -2.78913826e-01
-1.55289948e-01 9.19472396e-01 9.29348946e-01 -6.89447999e-01
-1.44087344e-01 9.36595321e-01 -1.89424738e-01 -6.78792477e-01
-5.59493244e-01 -5.01428127e-01 -5.26926994e-01 -5.36613345e-01
1.17373908e+00 -1.29132509e+00 -7.83506095e-01 6.79250360e-01
-9.68372166e-01 -5.51015615e-01 -7.83514231e-04 8.16898286e-01
-8.04688752e-01 2.79010445e-01 -6.14190400e-01 -9.61604774e-01
-9.02256250e-01 -9.39958811e-01 1.18336654e+00 -1.98404565e-01
-1.26324666e+00 -4.98385668e-01 4.93657619e-01 4.47867602e-01
2.82147855e-01 8.09460342e-01 7.76106238e-01 1.48886114e-01
4.05051231e-01 -2.16587827e-01 -1.13638967e-01 4.36313748e-01
5.66506982e-01 -3.79469067e-01 -1.04922235e+00 -1.95034072e-01
2.66258448e-01 -1.15373528e+00 6.98530555e-01 6.30129218e-01
8.02735925e-01 -2.01019675e-01 -2.93206781e-01 1.46646470e-01
1.04740989e+00 1.34579450e-01 9.00474787e-01 9.93730009e-01
1.49324641e-01 4.72157925e-01 2.71733552e-01 6.06569886e-01
3.70210946e-01 7.13878989e-01 4.17748272e-01 1.39273494e-01
2.15314060e-01 6.06835425e-01 5.97929418e-01 -1.14765130e-01
-3.97339433e-01 5.85379541e-01 -7.63105214e-01 1.68967366e-01
-1.78568316e+00 -1.11047888e+00 1.62934661e-01 1.92198133e+00
9.14441526e-01 -1.18353795e-02 4.77683485e-01 2.42871359e-01
4.18797165e-01 -2.90942341e-01 -7.75519907e-01 -4.79953498e-01
6.58655027e-03 7.38377333e-01 1.88126430e-01 2.66560495e-01
-1.03052902e+00 3.81097086e-02 5.98671389e+00 -4.35541905e-02
-9.97436523e-01 3.21679682e-01 6.82339072e-01 -8.33187401e-01
9.65661854e-02 -4.40659463e-01 -9.15728137e-03 3.54213506e-01
7.80316651e-01 2.62475848e-01 3.44809830e-01 1.11403012e+00
7.46363282e-01 -7.48070478e-01 -1.13248074e+00 1.60748506e+00
1.04356140e-01 -5.20351708e-01 -5.71722448e-01 1.86267570e-02
-1.24265157e-01 1.79233238e-01 -1.87565386e-01 5.13678044e-02
-2.10503802e-01 -1.20916486e+00 9.12294745e-01 8.10497403e-01
1.07294583e+00 -7.07955658e-01 6.96900666e-01 -1.74406514e-01
-6.74374282e-01 -1.98190525e-01 2.38809109e-01 -6.41561210e-01
2.65064508e-01 1.74637124e-01 -4.69176263e-01 -3.17681491e-01
1.35948277e+00 4.54748929e-01 -2.77227104e-01 9.49706256e-01
-1.49724975e-01 2.57499576e-01 -6.48803711e-01 2.63175368e-01
-5.81291644e-03 1.00199863e-01 1.06889777e-01 1.02139831e+00
4.15535808e-01 3.68972719e-01 -3.28373462e-02 5.64392567e-01
2.17583537e-01 -2.16202006e-01 -5.97426593e-01 2.89809644e-01
1.36782393e-01 1.26234186e+00 -4.44344103e-01 1.90251455e-01
-7.21034169e-01 1.44442379e+00 3.99920315e-01 2.15441376e-01
-4.06898499e-01 2.88325399e-01 8.25598300e-01 2.81784058e-01
-1.28372863e-01 -3.31058413e-01 -2.61753440e-01 -7.29558825e-01
4.88875449e-01 -1.03455019e+00 5.53302646e-01 -1.02273977e+00
-1.33860874e+00 5.35531864e-02 -1.48839764e-02 -1.04514337e+00
-3.45307142e-01 -7.88698316e-01 -4.51249093e-01 6.53824806e-01
-8.21041167e-01 -8.12195539e-01 -8.68136287e-01 8.56862247e-01
7.42233992e-02 3.94812673e-01 1.54792607e+00 1.44616157e-01
-3.17662925e-01 3.40937793e-01 -2.22731590e-01 1.54981852e-01
8.88906360e-01 -8.24640691e-01 -4.53476697e-01 2.74205729e-02
-7.57464707e-01 6.71666801e-01 5.05556881e-01 -7.31575787e-01
-1.08840048e+00 -7.64899373e-01 1.41836435e-01 -2.18515992e-01
4.91179228e-01 -2.53075153e-01 -7.23487973e-01 6.53412104e-01
3.75667587e-02 -2.56303608e-01 1.05568445e+00 1.47438524e-02
-4.37102079e-01 1.09478220e-01 -1.46300828e+00 5.65944970e-01
1.29872012e+00 -8.29486072e-01 -8.16750705e-01 3.74449015e-01
-4.37629931e-02 -4.26294208e-02 -7.99831033e-01 3.68592203e-01
1.22909224e+00 -9.36386108e-01 6.52445138e-01 -1.21232927e-01
4.88057584e-01 8.17706436e-02 -3.38996053e-01 -8.29975545e-01
-1.88923076e-01 -2.94535935e-01 3.83112766e-02 6.63492143e-01
-1.41679943e-01 -5.18005371e-01 7.25250959e-01 1.18755567e+00
1.50296971e-01 -8.44790339e-01 -1.42647886e+00 -6.06846690e-01
-5.86022288e-02 -6.86655164e-01 -4.94055331e-01 5.60341895e-01
4.99108851e-01 1.95402369e-01 -2.16987759e-01 -3.54311973e-01
3.03153366e-01 -5.83709180e-01 2.76104987e-01 -1.21744061e+00
-2.84778535e-01 -3.94777298e-01 -8.99597466e-01 3.78275886e-02
6.25563189e-02 -4.47569877e-01 -8.23819339e-02 -1.93192291e+00
6.22202337e-01 4.73307185e-02 -1.38063103e-01 1.01578903e+00
4.81459498e-03 1.72880605e-01 -2.25825578e-01 3.36157888e-01
-1.13313481e-01 7.13962972e-01 6.70373023e-01 -1.26706049e-01
-2.25677714e-01 -3.31613779e-01 -9.83500600e-01 7.04521954e-01
9.84530568e-01 -3.42172354e-01 -4.35960442e-01 -2.39327922e-01
1.96502581e-02 -3.33437949e-01 1.00087869e+00 -1.36177039e+00
-3.75480741e-01 1.95238903e-01 7.77788639e-01 -1.49138764e-01
1.04853261e+00 -8.18167984e-01 3.03777426e-01 8.53266835e-01
1.97693706e-01 1.16964661e-01 4.75196242e-01 4.74525213e-01
2.12563992e-01 -5.32207005e-02 8.95414472e-01 -6.00250185e-01
-7.16824710e-01 -5.15514892e-03 -9.42741513e-01 2.88905110e-02
9.40982521e-01 -2.72809476e-01 -4.86433893e-01 -9.05186892e-01
-9.87695515e-01 9.12313610e-02 7.62527645e-01 1.08956531e-01
6.19309664e-01 -1.38734794e+00 -3.28620344e-01 -3.51978391e-01
2.62654513e-01 5.83781227e-02 1.86757326e-01 1.32755029e+00
-6.30491316e-01 -2.21013367e-01 -7.42035151e-01 -2.90034622e-01
-1.40357101e+00 2.22376153e-01 4.04488802e-01 2.79149026e-01
-1.03433597e+00 6.81768596e-01 -6.75572753e-02 1.62129119e-01
4.79137093e-01 -3.47424507e-01 2.06604272e-01 3.42387527e-01
8.73341799e-01 1.52675495e-01 6.67793378e-02 -5.82343102e-01
-7.64128625e-01 3.01469326e-01 -2.83552021e-01 -6.53648615e-01
1.49631953e+00 1.10309899e-01 -7.89997727e-02 5.57783365e-01
1.26826859e+00 -4.76445735e-01 -1.20655763e+00 4.32449639e-01
-3.16613466e-01 8.75809342e-02 7.65640214e-02 -7.03035891e-01
-6.43756211e-01 5.50921559e-01 1.43882561e+00 -6.78960800e-01
1.33615530e+00 1.58087328e-01 8.18009675e-01 7.66193926e-01
5.50810218e-01 -1.26388121e+00 3.44916523e-01 -4.36885357e-02
1.37785649e+00 -9.96398807e-01 1.60834461e-01 1.04197249e-01
-6.34219110e-01 1.16960609e+00 4.05110657e-01 -7.90861435e-03
5.63649237e-01 5.03371119e-01 3.95325780e-01 -3.59140366e-01
-2.85082817e-01 -1.91836774e-01 -4.70735133e-01 8.80058825e-01
2.79367119e-01 1.54666856e-01 -5.64968646e-01 7.74108410e-01
-3.74951601e-01 6.97487295e-01 2.96315312e-01 1.79550111e+00
-5.87757409e-01 -9.38796401e-01 -6.31722391e-01 9.36791182e-01
-5.11185527e-01 4.30733077e-02 -6.16042852e-01 6.16998792e-01
1.45006284e-01 7.79604912e-01 6.76399320e-02 -1.47689521e-01
3.27618062e-01 5.82211792e-01 8.25760007e-01 -5.15037358e-01
-3.59944887e-02 7.73448423e-02 5.14621794e-01 -6.91453159e-01
-8.03456068e-01 -1.12187994e+00 -1.39553154e+00 5.51039614e-02
5.30788302e-01 -6.60180449e-01 2.93504864e-01 9.27667856e-01
3.13793361e-01 2.65865885e-02 -2.08762586e-01 -1.09637928e+00
1.83432512e-02 -1.06290209e+00 -7.24696994e-01 3.83896053e-01
3.84302437e-01 -1.05291307e+00 -3.22004795e-01 -3.48807499e-02]
|
[13.588364601135254, 2.1585919857025146]
|
f0a2dee7-dde3-44c6-bec9-a884f0ef7841
|
scene-aware-egocentric-3d-human-pose
|
2212.11684
| null |
https://arxiv.org/abs/2212.11684v2
|
https://arxiv.org/pdf/2212.11684v2.pdf
|
Scene-aware Egocentric 3D Human Pose Estimation
|
Egocentric 3D human pose estimation with a single head-mounted fisheye camera has recently attracted attention due to its numerous applications in virtual and augmented reality. Existing methods still struggle in challenging poses where the human body is highly occluded or is closely interacting with the scene. To address this issue, we propose a scene-aware egocentric pose estimation method that guides the prediction of the egocentric pose with scene constraints. To this end, we propose an egocentric depth estimation network to predict the scene depth map from a wide-view egocentric fisheye camera while mitigating the occlusion of the human body with a depth-inpainting network. Next, we propose a scene-aware pose estimation network that projects the 2D image features and estimated depth map of the scene into a voxel space and regresses the 3D pose with a V2V network. The voxel-based feature representation provides the direct geometric connection between 2D image features and scene geometry, and further facilitates the V2V network to constrain the predicted pose based on the estimated scene geometry. To enable the training of the aforementioned networks, we also generated a synthetic dataset, called EgoGTA, and an in-the-wild dataset based on EgoPW, called EgoPW-Scene. The experimental results of our new evaluation sequences show that the predicted 3D egocentric poses are accurate and physically plausible in terms of human-scene interaction, demonstrating that our method outperforms the state-of-the-art methods both quantitatively and qualitatively.
|
['Christian Theobalt', 'Diogo Luvizon', 'Kripasindhu Sarkar', 'Weipeng Xu', 'Lingjie Liu', 'Jian Wang']
|
2022-12-20
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Wang_Scene-Aware_Egocentric_3D_Human_Pose_Estimation_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Wang_Scene-Aware_Egocentric_3D_Human_Pose_Estimation_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['3d-human-pose-estimation', 'egocentric-pose-estimation']
|
['computer-vision', 'computer-vision']
|
[-1.71728864e-01 2.29099303e-01 2.45664507e-01 -4.94750887e-01
-1.78475335e-01 -7.97202587e-02 1.61181718e-01 -6.43931985e-01
-2.50582963e-01 3.75515103e-01 4.30423856e-01 3.72527540e-01
1.58929408e-01 -6.25835180e-01 -8.16222548e-01 -2.48106092e-01
1.30639583e-01 4.76618290e-01 1.48567721e-01 -2.49251187e-01
3.24102258e-03 5.20880461e-01 -1.43911731e+00 -1.52111754e-01
6.03772104e-01 9.24177587e-01 3.99563730e-01 6.08549297e-01
4.57375318e-01 6.40585184e-01 -2.11917341e-01 -1.70232117e-01
5.30854106e-01 -1.90959200e-01 -3.42446476e-01 4.99089122e-01
7.78273404e-01 -8.44121575e-01 -9.06915545e-01 6.82188630e-01
6.20661974e-01 4.52852666e-01 2.39595130e-01 -1.19185090e+00
-1.35568738e-01 -1.87218770e-01 -7.62964845e-01 -1.99758247e-01
9.67957377e-01 9.02281627e-02 4.78828490e-01 -1.07368553e+00
9.94498968e-01 1.38738918e+00 5.57526946e-01 6.09866023e-01
-8.98248434e-01 -5.66622615e-01 3.98475200e-01 1.57016575e-01
-1.74383831e+00 -2.24858716e-01 1.32780004e+00 -5.06910384e-01
7.53192902e-01 -1.98210012e-02 1.16447353e+00 1.02011240e+00
4.89161521e-01 8.46317232e-01 5.76707244e-01 -3.29522580e-01
9.82201025e-02 -1.34604663e-01 -4.31435943e-01 9.33519363e-01
-1.45484611e-01 1.30277321e-01 -6.51458263e-01 9.78098586e-02
1.34360242e+00 4.13769692e-01 -4.11651850e-01 -1.41219831e+00
-1.05010927e+00 6.71300709e-01 6.46108031e-01 -3.70178699e-01
-3.50262612e-01 3.66056263e-01 1.99873760e-01 -3.51050586e-01
7.17719853e-01 2.65390128e-01 -6.45247102e-01 6.73538074e-02
-4.53060716e-01 7.22597361e-01 3.19669962e-01 1.20005584e+00
7.25309312e-01 3.29502597e-02 1.90770859e-03 5.57280362e-01
4.13456470e-01 4.45291847e-01 5.54147810e-02 -1.27443147e+00
6.29566014e-01 6.77537143e-01 3.07546288e-01 -1.19081247e+00
-6.55145288e-01 -3.71368974e-01 -3.79728138e-01 8.71334299e-02
2.80076474e-01 -1.35893077e-01 -7.50659645e-01 1.80632532e+00
9.39963341e-01 2.37697929e-01 -2.90999651e-01 1.37924421e+00
6.91245496e-01 3.15578401e-01 -1.20630555e-01 7.54816160e-02
1.32652235e+00 -1.11448491e+00 -6.52394176e-01 -5.12270153e-01
5.23086965e-01 -4.04953420e-01 1.11849678e+00 2.06874818e-01
-1.11741602e+00 -6.04460180e-01 -8.55744720e-01 -3.04309785e-01
5.91083355e-02 1.23263367e-01 5.51690757e-01 3.24942827e-01
-7.03889012e-01 2.76811659e-01 -9.80945408e-01 -4.05052662e-01
1.10056557e-01 2.34077230e-01 -8.10880184e-01 -2.66277015e-01
-1.03041220e+00 8.77317727e-01 2.22344562e-01 2.70828962e-01
-9.04142439e-01 -8.02697897e-01 -1.55777156e+00 1.03741754e-02
6.21443152e-01 -1.29076588e+00 1.05592763e+00 -3.13227057e-01
-1.58449042e+00 9.68823791e-01 -2.16401704e-02 -1.05671346e-01
8.07955146e-01 -8.09314966e-01 5.13407849e-02 2.87257016e-01
8.68620425e-02 8.27114046e-01 5.38106561e-01 -1.24838495e+00
-3.43850136e-01 -9.47401226e-01 2.94290304e-01 8.59980524e-01
9.47651640e-02 -2.56201893e-01 -8.96861672e-01 -6.48358583e-01
5.61239302e-01 -9.63698447e-01 -3.60061228e-01 4.34795767e-01
-5.86715519e-01 1.19804256e-01 9.43212450e-01 -6.60202742e-01
7.03372955e-01 -1.94536948e+00 5.02150297e-01 -5.16003966e-02
3.94568831e-01 -2.61301577e-01 4.21849489e-02 1.70459673e-01
-1.78391501e-01 -6.01752162e-01 8.78715813e-02 -8.79820764e-01
-2.50009358e-01 1.52439401e-01 -8.20304900e-02 9.26480830e-01
-2.13580355e-01 9.64205027e-01 -9.70526874e-01 -4.70022589e-01
8.22673738e-01 8.55189741e-01 -1.00915360e+00 7.17004240e-01
-5.32126538e-02 8.76826286e-01 -6.00759029e-01 4.60679203e-01
8.58427346e-01 6.16324991e-02 3.61725269e-03 -2.62153894e-01
4.22222242e-02 -2.63339281e-02 -1.04077339e+00 2.47844362e+00
-4.39658761e-01 3.76884460e-01 2.64573507e-02 -5.28792322e-01
7.21889555e-01 1.13499776e-01 7.22840548e-01 -6.02088690e-01
3.09700251e-01 -3.45570236e-01 -6.34060264e-01 -6.62286282e-01
5.14263928e-01 -7.11141899e-02 -1.09611839e-01 -7.54456148e-02
-3.06011438e-02 -4.54305530e-01 -4.20899749e-01 1.30850449e-01
7.07092285e-01 7.90193558e-01 2.01296136e-01 2.51735151e-01
4.66285259e-01 -1.90945908e-01 6.95531726e-01 1.39906511e-01
-2.07187518e-01 9.66939986e-01 2.32920036e-01 -6.29488587e-01
-1.12122595e+00 -1.27845693e+00 1.40506476e-01 6.69740736e-01
5.92449188e-01 -4.96409982e-01 -9.29638803e-01 -5.77068686e-01
-1.06439427e-01 6.28434896e-01 -6.92799032e-01 -1.70124814e-01
-6.90954864e-01 1.64405137e-01 -2.01683454e-02 6.70559645e-01
6.21150196e-01 -8.29630971e-01 -9.46370304e-01 -1.35259330e-01
-4.38458055e-01 -1.46530855e+00 -6.38051391e-01 -3.27174574e-01
-6.88090265e-01 -1.22400498e+00 -9.24758792e-01 -5.12234926e-01
8.61154139e-01 4.11027551e-01 8.59993815e-01 -4.33299690e-01
-2.97961920e-01 6.79810047e-01 -2.28838488e-01 -7.86168650e-02
3.50150466e-01 -2.60237902e-01 3.27087104e-01 -1.11032449e-01
1.46395802e-01 -6.83009386e-01 -1.13267827e+00 4.81417567e-01
-3.47844273e-01 4.63228434e-01 4.49863672e-02 6.35634959e-01
6.27035081e-01 -2.09098592e-01 -2.12847859e-01 -7.51057684e-01
-9.35039297e-02 -2.67500281e-01 -5.62993348e-01 -1.95882261e-01
7.83384070e-02 -4.87548739e-01 3.76021922e-01 -2.69707143e-01
-1.16886508e+00 7.03082383e-01 -1.32270992e-01 -1.05569363e+00
-1.13190532e-01 1.09868407e-01 -6.20039642e-01 -6.49757609e-02
4.64604914e-01 9.45391804e-02 -6.47284091e-02 -3.59514654e-01
4.98527467e-01 1.30453691e-01 8.08097780e-01 -3.56088877e-01
8.24147344e-01 8.24430704e-01 -5.43200560e-02 -6.64783239e-01
-1.01308763e+00 -4.57673937e-01 -1.05254018e+00 -4.18155640e-01
1.16870379e+00 -1.37397075e+00 -8.26991379e-01 3.55941206e-01
-1.37597167e+00 -3.46168369e-01 -2.69091368e-01 7.84408152e-01
-1.14486110e+00 5.66082895e-01 -3.74433041e-01 -7.30775476e-01
-8.72512609e-02 -1.28887117e+00 1.59609520e+00 1.83327179e-02
-3.62018257e-01 -9.01563227e-01 3.53218988e-02 6.45523727e-01
-3.17915022e-01 6.69547677e-01 4.24084187e-01 2.07156420e-01
-6.65367305e-01 -2.98786283e-01 1.07641136e-02 5.65481111e-02
-4.88383323e-02 -5.00704110e-01 -9.79373872e-01 -2.54874110e-01
1.80205107e-01 -1.49285227e-01 2.36946106e-01 7.98891425e-01
1.22742462e+00 -3.02972291e-02 -3.29141110e-01 1.20368731e+00
1.04594636e+00 -7.65618235e-02 7.24992752e-01 2.22083494e-01
1.17217636e+00 8.08123648e-01 9.58839357e-01 7.00902820e-01
7.34808028e-01 1.22027934e+00 7.28972197e-01 -9.27792713e-02
-5.06291073e-03 -9.58822906e-01 1.01544298e-01 4.49547559e-01
1.46419136e-02 -1.14531122e-01 -6.83469832e-01 3.06739151e-01
-1.86731517e+00 -4.67637718e-01 5.84371854e-03 2.24063778e+00
1.64625302e-01 -2.00778127e-01 -4.48643118e-02 -2.96843737e-01
5.00436366e-01 1.54188231e-01 -7.72085965e-01 -1.36104051e-03
3.43920082e-01 -3.55374217e-01 2.34668911e-01 5.33075094e-01
-1.07567430e+00 1.23239088e+00 6.01893473e+00 2.60733038e-01
-9.27343905e-01 -6.81021530e-03 2.73273557e-01 -3.01136762e-01
-1.92864500e-02 -1.44872248e-01 -6.43023610e-01 2.07541704e-01
2.14640304e-01 1.41864002e-01 1.31183118e-01 1.28649521e+00
4.45018739e-01 -2.30082661e-01 -1.24993491e+00 1.42768586e+00
3.17234129e-01 -9.70909595e-01 -1.15642875e-01 3.05715650e-01
6.66453958e-01 -7.37675205e-02 -1.64365098e-02 1.51920006e-01
-5.30654266e-02 -6.76699460e-01 8.64718199e-01 5.44004321e-01
9.06743407e-01 -7.70946264e-01 4.74203497e-01 7.11871266e-01
-1.27188897e+00 -6.91964477e-02 -4.99204665e-01 -4.11585391e-01
4.91075784e-01 2.02415511e-01 -5.43783724e-01 3.38815600e-01
7.83440411e-01 7.17342913e-01 -2.97074944e-01 9.08027411e-01
-4.10542935e-01 -3.48162591e-01 -1.79623693e-01 6.04922414e-01
-6.77831694e-02 -2.17045337e-01 6.85622752e-01 5.53560197e-01
2.71893382e-01 2.83960313e-01 3.36800188e-01 9.12638485e-01
1.73102811e-01 1.18973993e-01 -9.91970420e-01 7.07839787e-01
9.38971862e-02 1.11638570e+00 -5.27690768e-01 -2.23303080e-01
-1.27064213e-01 1.30928791e+00 4.47645336e-01 4.55983549e-01
-9.60911334e-01 -2.66068969e-02 8.77764761e-01 4.33423728e-01
2.06888691e-01 -5.01226068e-01 -1.69982966e-02 -1.56745923e+00
2.27431849e-01 -3.54652733e-01 7.35533386e-02 -1.45736337e+00
-6.54844940e-01 5.51801443e-01 2.92118192e-01 -1.45144355e+00
-5.21591365e-01 -5.13610601e-01 -3.78998458e-01 7.99930096e-01
-1.02846789e+00 -1.21896684e+00 -9.31468546e-01 7.38193333e-01
7.25241065e-01 9.72650945e-02 5.43029070e-01 -6.35226211e-03
-4.40821260e-01 5.81312895e-01 -4.55919385e-01 5.55825531e-02
6.92461967e-01 -8.75079632e-01 4.54897940e-01 6.17345214e-01
-2.32948661e-01 6.62779927e-01 7.46827841e-01 -6.71905160e-01
-1.70035195e+00 -1.13237572e+00 4.79510874e-01 -9.22949612e-01
9.37160030e-02 -7.70353079e-01 -4.47593272e-01 8.88675094e-01
-4.50425416e-01 4.43486184e-01 2.20463529e-01 -1.02348685e-01
-1.60038620e-01 -4.45616618e-03 -1.29879510e+00 8.37969720e-01
1.47958708e+00 -5.42418599e-01 -4.85301554e-01 3.99522722e-01
9.02838826e-01 -1.11076260e+00 -6.91910923e-01 3.45605612e-01
8.78701091e-01 -1.01675057e+00 1.26616740e+00 -3.81393135e-01
6.68049514e-01 -3.70020926e-01 -3.46500814e-01 -1.36909270e+00
-2.39167169e-01 -4.16144609e-01 -3.44513595e-01 6.55297458e-01
-4.65612441e-01 -2.49071106e-01 1.42407393e+00 8.48538637e-01
-2.80860335e-01 -8.48780036e-01 -9.37724531e-01 -4.50827479e-01
-3.10908198e-01 -6.13747835e-01 3.55051130e-01 6.65553927e-01
-4.92161997e-02 1.80826679e-01 -6.28432930e-01 2.28762642e-01
6.72095776e-01 -2.23473147e-01 1.51885009e+00 -1.00343132e+00
-1.36832491e-01 2.41485193e-01 -7.85783291e-01 -1.64575100e+00
2.77417958e-01 -2.98968643e-01 2.06584707e-01 -1.28797126e+00
2.59090900e-01 -3.65178809e-02 3.74293923e-01 -2.50585619e-02
-1.32595181e-01 3.42901319e-01 1.52537182e-01 4.83101234e-02
-3.88063550e-01 1.04845059e+00 1.79680836e+00 2.47755259e-01
-3.24114174e-01 -7.95975849e-02 -3.27833474e-01 1.14446759e+00
2.26275891e-01 -1.97348610e-01 -7.30800867e-01 -5.16084731e-01
1.32872254e-01 5.05844831e-01 5.79713225e-01 -1.02867436e+00
1.52891785e-01 -2.09794536e-01 5.95926642e-01 -1.09466374e+00
1.10692060e+00 -9.64456916e-01 1.68895885e-01 3.86523873e-01
-8.18140656e-02 -4.76863086e-02 -5.05957752e-04 6.97084129e-01
-5.56466449e-03 3.98600519e-01 6.43542945e-01 -2.97248751e-01
-7.91587651e-01 8.36170733e-01 1.29678756e-01 4.47985046e-02
1.12928295e+00 -6.20517671e-01 2.43472174e-01 -7.10330248e-01
-6.85349703e-01 3.34963679e-01 9.02766347e-01 4.89697069e-01
1.15929008e+00 -1.30032527e+00 -4.52980399e-01 6.02303326e-01
3.64791363e-01 6.14154339e-01 8.16616356e-01 6.99855566e-01
-8.70562613e-01 5.11191547e-01 -2.28792727e-01 -9.50046957e-01
-1.24308741e+00 6.28110945e-01 5.94572842e-01 2.55768001e-02
-1.16505265e+00 7.75468647e-01 1.33075023e+00 -8.30434918e-01
2.60891974e-01 -2.18704507e-01 -9.24454853e-02 -6.40480578e-01
4.38886642e-01 4.24881488e-01 -3.79202992e-01 -1.10268545e+00
-2.73656636e-01 1.02621317e+00 1.01212353e-01 -6.73391595e-02
1.28015172e+00 -4.59244579e-01 2.08971918e-01 7.90118277e-02
1.32490063e+00 -1.39378523e-02 -1.88457239e+00 -1.56360000e-01
-8.63680720e-01 -9.67137635e-01 -1.57068968e-02 -2.03392014e-01
-1.16967201e+00 9.45229411e-01 5.66464126e-01 -7.50439823e-01
8.66625547e-01 -6.78022802e-02 8.33970845e-01 2.46648163e-01
6.92128241e-01 -9.99429762e-01 3.72645140e-01 5.17727911e-01
1.17763114e+00 -9.77571428e-01 2.53092140e-01 -1.01808262e+00
-6.49196684e-01 8.65022063e-01 1.28913724e+00 -3.90850157e-01
6.16438091e-01 -1.88777298e-01 1.01363100e-02 -5.10974169e-01
-2.03707859e-01 2.68803418e-01 3.78533512e-01 8.40824068e-01
1.20859869e-01 -3.40840407e-02 7.49454051e-02 3.48638028e-01
-4.61462200e-01 -6.06621616e-02 3.40845436e-01 6.98915303e-01
-5.71153462e-02 -5.94730675e-01 -4.91918504e-01 -1.95034653e-01
-3.30076031e-02 2.79095262e-01 -2.09870070e-01 8.85350108e-01
3.07242662e-01 5.12078762e-01 2.76564837e-01 -6.06891513e-01
7.35235572e-01 -3.39859813e-01 6.91838324e-01 -8.70411634e-01
-5.07726222e-02 2.04665601e-01 -1.68052778e-01 -1.14676225e+00
-8.20997581e-02 -3.78019840e-01 -1.17508292e+00 -1.99666217e-01
-3.21874797e-01 -2.83133060e-01 7.20514894e-01 9.12263811e-01
3.22209060e-01 4.84474838e-01 6.64341986e-01 -1.64744663e+00
-1.57438725e-01 -6.69092476e-01 -7.59639382e-01 4.96432483e-01
2.25431070e-01 -1.15342724e+00 -7.68592283e-02 -2.41821751e-01]
|
[7.056766033172607, -0.9641082882881165]
|
c735b8ac-d476-45b7-b351-27461487eecd
|
topic-aware-encoding-for-extractive
|
2112.09572
| null |
https://arxiv.org/abs/2112.09572v3
|
https://arxiv.org/pdf/2112.09572v3.pdf
|
Topic-Aware Encoding for Extractive Summarization
|
Document summarization provides an instrument for faster understanding the collection of text documents and has several real-life applications. With the growth of online text data, numerous summarization models have been proposed recently. The Sequence-to-Sequence (Seq2Seq) based neural summarization model is the most widely used in the summarization field due to its high performance. This is because semantic information and structure information in the text is adequately considered when encoding. However, the existing extractive summarization models pay little attention to and use the central topic information to assist the generation of summaries, which leads to models not ensuring the generated summary under the primary topic. A lengthy document can span several topics, and a single summary cannot do justice to all the topics. Therefore, the key to generating a high-quality summary is determining the central topic and building a summary based on it, especially for a long document. We propose a topic-aware encoding for document summarization to deal with this issue. This model effectively combines syntactic-level and topic-level information to build a comprehensive sentence representation. Specifically, a neural topic model is added in the neural-based sentence-level representation learning to adequately consider the central topic information for capturing the critical content in the original document. The experimental results on three public datasets show that our model outperforms the state-of-the-art models.
|
['Liping Jing', 'Mingyang Song']
|
2021-12-17
| null | null | null | null |
['extractive-summarization']
|
['natural-language-processing']
|
[ 3.74205381e-01 -3.54285575e-02 -3.95997941e-01 -2.78663427e-01
-9.11895335e-01 -1.65349007e-01 3.51726294e-01 5.81968546e-01
-2.50469595e-01 9.28044915e-01 1.09403396e+00 1.16631456e-01
5.33343889e-02 -8.08509350e-01 -4.85096961e-01 -5.38153112e-01
3.15602213e-01 1.89944535e-01 3.62577230e-01 -3.02136302e-01
7.60531902e-01 -2.52724409e-01 -1.25124025e+00 4.14409518e-01
1.65752995e+00 6.15158856e-01 7.94509590e-01 5.53231359e-01
-7.90161371e-01 4.07799095e-01 -1.13357675e+00 -7.83777237e-03
-3.06732774e-01 -9.45967853e-01 -6.80791497e-01 1.43863056e-02
2.20130116e-01 -4.91982788e-01 -1.80990532e-01 1.15088904e+00
6.66488230e-01 3.21614355e-01 6.07391775e-01 -6.37719393e-01
-5.28072357e-01 9.61574793e-01 -6.39549673e-01 3.51996839e-01
3.96329284e-01 -1.67014912e-01 1.19554436e+00 -4.50320065e-01
5.09507656e-01 1.21359038e+00 2.11657286e-01 4.24643606e-01
-6.13723397e-01 -3.33304137e-01 4.45720702e-01 1.60264403e-01
-8.18963349e-01 -2.89302021e-01 8.18738937e-01 -8.15360993e-02
9.67925012e-01 3.58142227e-01 7.65735686e-01 9.21591401e-01
6.15604162e-01 1.14039183e+00 3.53850037e-01 -2.32935190e-01
2.75790870e-01 -2.15127543e-01 5.59995294e-01 2.26784259e-01
6.44354165e-01 -8.36184382e-01 -5.31036615e-01 -6.85665980e-02
2.49719664e-01 1.60250381e-01 -4.47195470e-01 4.09141749e-01
-1.11961281e+00 7.87928581e-01 2.64590085e-01 3.50752592e-01
-7.68956184e-01 -5.07495850e-02 9.40287590e-01 -1.82251558e-01
6.67715609e-01 5.68143964e-01 -1.96662039e-01 -3.48373979e-01
-1.41565609e+00 4.79231894e-01 7.91123450e-01 7.80098259e-01
4.90825504e-01 2.71601647e-01 -7.24271774e-01 8.65679920e-01
3.13733816e-02 3.49821955e-01 1.07659626e+00 -7.41774261e-01
9.07359660e-01 8.97395670e-01 -1.38315693e-01 -1.14008319e+00
-1.95887744e-01 -6.14531934e-01 -1.32168460e+00 -7.59010375e-01
-3.83780241e-01 -3.30463201e-01 -6.83614552e-01 1.51665139e+00
-1.23637961e-03 -5.39860390e-02 3.55980963e-01 6.39986277e-01
1.22278726e+00 1.36731458e+00 -4.55190502e-02 -7.20501781e-01
1.39610946e+00 -1.03074729e+00 -1.09969151e+00 -4.32568222e-01
4.69380051e-01 -5.49316585e-01 7.06674933e-01 2.62369096e-01
-1.10390925e+00 -5.15842736e-01 -1.31552708e+00 -1.69123098e-01
-6.74096942e-02 2.01645821e-01 3.17630947e-01 1.52410209e-01
-7.88083673e-01 5.12212396e-01 -5.86646616e-01 -4.23287272e-01
4.04514015e-01 -1.44927381e-02 4.21807356e-02 -1.88419893e-01
-1.41300845e+00 6.57618582e-01 1.04101098e+00 9.73096564e-02
-4.17944640e-01 -4.02459353e-01 -7.61818886e-01 5.10545433e-01
4.38674062e-01 -9.88803864e-01 1.28876734e+00 -6.54412508e-01
-1.33198357e+00 -2.83668060e-02 -6.33577764e-01 -4.89985436e-01
1.98926166e-01 -2.32788399e-01 -1.74416304e-01 3.43122572e-01
3.07261735e-01 5.03070295e-01 6.00499332e-01 -9.23654377e-01
-8.10007215e-01 -3.03736836e-01 -1.86514243e-01 5.71545899e-01
-5.33678412e-01 4.26204875e-02 -4.36931640e-01 -7.63194859e-01
7.07774758e-02 -3.12388837e-01 -1.48277953e-01 -8.87862325e-01
-9.34202433e-01 -5.34085453e-01 7.29980052e-01 -1.14178216e+00
1.89687729e+00 -1.78703654e+00 2.84635544e-01 -4.04045582e-01
1.20402396e-01 4.97185826e-01 -1.36334777e-01 9.25293624e-01
3.37375104e-01 2.29146436e-01 -4.67676371e-01 -3.31446111e-01
-1.62772775e-01 -1.27864286e-01 -5.97680271e-01 -2.21010476e-01
3.87565568e-02 9.03489470e-01 -9.61835861e-01 -7.00850189e-01
-1.97467253e-01 1.64539590e-01 -4.11987096e-01 3.47872853e-01
-4.72636551e-01 2.41150990e-01 -8.93572569e-01 1.40240029e-01
5.84319293e-01 -8.76222104e-02 -1.31820038e-01 -2.62902696e-02
-1.78143457e-01 5.95626831e-01 -7.97550499e-01 1.95613754e+00
-1.35882169e-01 4.23336208e-01 -2.69531429e-01 -1.04347014e+00
9.85602260e-01 3.12402993e-01 3.33443403e-01 -5.81918597e-01
2.91939825e-02 2.50884950e-01 -6.96412697e-02 -5.91100991e-01
1.15351808e+00 1.38302715e-02 -2.62291998e-01 6.34961903e-01
-1.25231355e-01 -2.03719467e-01 6.75247014e-01 6.51666284e-01
8.97460699e-01 -1.64669141e-01 5.77239275e-01 -1.84177041e-01
5.81487000e-01 4.44995947e-02 8.70429933e-01 6.39166713e-01
1.40755117e-01 8.02730024e-01 6.72966301e-01 -1.02066241e-01
-9.77097869e-01 -3.93980205e-01 4.09034073e-01 7.93016315e-01
1.48457721e-01 -7.18316376e-01 -1.07109404e+00 -5.44513106e-01
-3.37699711e-01 1.12481487e+00 -4.39203352e-01 -5.42934358e-01
-5.78107417e-01 -6.36995256e-01 3.49405110e-01 4.41650599e-01
8.36817026e-01 -1.30966079e+00 -5.36665320e-01 5.66372514e-01
-7.26360440e-01 -6.21817112e-01 -8.52097332e-01 -1.99426085e-01
-1.10757625e+00 -7.10128009e-01 -1.04962134e+00 -6.12841487e-01
5.34624875e-01 5.15576899e-01 6.62683427e-01 -3.80923273e-03
1.76759511e-01 -1.56375080e-01 -5.58598578e-01 -6.43955410e-01
-5.48255861e-01 7.27452159e-01 -1.75619423e-01 -1.97750375e-01
1.07728645e-01 -3.82091433e-01 -5.59534013e-01 -2.56967813e-01
-1.17467546e+00 4.14291173e-01 7.89219975e-01 6.63978875e-01
2.94725627e-01 2.55611748e-01 1.25560844e+00 -8.25910509e-01
1.32671404e+00 -5.96774817e-01 1.86382923e-02 3.66000056e-01
-3.64446968e-01 6.67622909e-02 9.56027091e-01 -1.81169763e-01
-1.34096777e+00 -6.68130517e-01 -1.39883786e-01 1.89515546e-01
7.25295814e-03 1.07692659e+00 -3.18546236e-01 9.47752357e-01
2.24640936e-01 1.01225317e+00 -5.30088134e-02 -4.45637226e-01
2.95398626e-02 9.00960803e-01 3.75451565e-01 -4.34269249e-01
2.50450224e-01 6.68080011e-03 -3.17072093e-01 -1.05833113e+00
-1.10342872e+00 -5.67304611e-01 -4.26490009e-01 -1.01326453e-02
7.33521998e-01 -6.73898041e-01 -2.23471433e-01 5.02526700e-01
-1.68786800e+00 2.86725789e-01 -2.82331198e-01 3.81643921e-01
-3.36793274e-01 8.84001374e-01 -3.78735960e-01 -7.45981991e-01
-1.15406871e+00 -9.91717577e-01 1.02825189e+00 7.88326383e-01
-3.65672886e-01 -7.49041378e-01 1.44655168e-01 1.15345277e-01
4.58597779e-01 -1.49255944e-02 1.04390049e+00 -9.03826416e-01
-4.17914689e-01 -1.92510352e-01 -2.26627052e-01 5.21799743e-01
2.97902703e-01 -8.72554481e-02 -5.35393834e-01 -1.76623717e-01
1.50413945e-01 -3.21869068e-02 1.35195494e+00 6.57880723e-01
1.07230532e+00 -7.54527509e-01 -2.18572825e-01 1.76422328e-01
1.06118178e+00 2.64063090e-01 6.47747099e-01 2.69269019e-01
7.28295088e-01 7.05663323e-01 6.60298824e-01 5.42702675e-01
5.89313507e-01 1.95671514e-01 1.48702562e-01 2.48958260e-01
3.48941199e-02 -5.20366549e-01 4.01573837e-01 1.52411604e+00
2.77009487e-01 -7.18700528e-01 -5.27096927e-01 5.99500954e-01
-2.10262346e+00 -1.19399595e+00 -2.54847389e-02 1.99201512e+00
9.87460613e-01 9.09976512e-02 -8.54101107e-02 3.57739925e-02
8.99754167e-01 5.66643775e-01 -7.53344595e-01 -5.28064966e-01
-1.28053620e-01 -4.41992700e-01 -7.62788504e-02 7.15960190e-02
-6.83361888e-01 8.37985635e-01 5.51172209e+00 1.11516488e+00
-9.73003507e-01 -3.04357648e-01 5.19252121e-01 -7.90823810e-03
-5.37684321e-01 -9.69027653e-02 -1.05962002e+00 8.96949589e-01
7.87930131e-01 -8.47020388e-01 -1.06651723e-01 5.83093882e-01
7.69801617e-01 -4.32410568e-01 -7.14724004e-01 5.71373463e-01
4.72410649e-01 -1.41406763e+00 6.66408300e-01 -3.98790054e-02
9.21403110e-01 -2.73521245e-01 -3.64856809e-01 4.23333645e-01
-1.54365718e-01 -6.31530881e-01 4.96229082e-01 6.47065341e-01
4.55961645e-01 -9.14660096e-01 1.02775753e+00 9.06307340e-01
-1.01451707e+00 9.66676846e-02 -6.99801505e-01 6.95099905e-02
4.59661543e-01 8.23244512e-01 -7.90585339e-01 9.79742110e-01
1.69500217e-01 1.00556910e+00 -4.51251864e-01 1.25404823e+00
-2.61354685e-01 8.15425038e-01 -7.05295876e-02 -3.70617151e-01
4.14309084e-01 -2.84509659e-01 8.44136238e-01 1.30211735e+00
5.36074758e-01 1.46017358e-01 3.17076147e-01 6.24709070e-01
-1.96477905e-01 5.03414869e-01 -3.81871194e-01 -2.21103311e-01
5.00656664e-01 1.00458348e+00 -6.39810920e-01 -6.25873446e-01
-9.55387056e-02 9.70116615e-01 2.30469167e-01 2.77264208e-01
-3.17322999e-01 -8.00732732e-01 1.87489703e-01 -9.46164355e-02
2.76773155e-01 -1.34255797e-01 -3.56610298e-01 -1.15041625e+00
1.65420771e-01 -7.78517127e-01 3.91191661e-01 -7.33499467e-01
-9.13938224e-01 4.71146137e-01 1.24125041e-01 -1.03248596e+00
-4.04952049e-01 3.32244188e-01 -1.25898063e+00 9.12502885e-01
-1.54491937e+00 -8.45487475e-01 -2.02897370e-01 -2.22943112e-01
1.30754924e+00 -1.53839752e-01 5.71469188e-01 -2.20407322e-01
-9.35636103e-01 2.09859505e-01 3.58948529e-01 -1.15628399e-01
6.80220127e-01 -1.13650322e+00 4.62436676e-01 1.09212589e+00
-4.43525940e-01 8.67263854e-01 7.66008198e-01 -1.15394580e+00
-1.27688730e+00 -1.15347111e+00 1.14610672e+00 8.94877017e-02
1.69347048e-01 1.31056547e-01 -1.30277777e+00 3.47188324e-01
5.96489906e-01 -1.04296112e+00 5.65611124e-01 -6.43799528e-02
1.43998921e-01 -1.24347240e-01 -6.84515178e-01 5.27262092e-01
6.47856772e-01 -6.93509169e-03 -1.16928446e+00 1.14382371e-01
1.18664682e+00 -2.87580937e-01 -3.48750204e-01 1.85448751e-01
3.12465876e-01 -8.04141402e-01 3.86446118e-01 -4.24582362e-01
1.00244939e+00 -2.00861722e-01 2.60394692e-01 -1.81026375e+00
-2.09579885e-01 -5.17569244e-01 -2.89005309e-01 1.56512880e+00
8.55424106e-02 -5.71808875e-01 6.66200638e-01 1.72681466e-01
-5.42823970e-01 -8.63537788e-01 -5.93366385e-01 -3.80245209e-01
-4.19546291e-02 -9.49020125e-03 7.66684055e-01 5.07968187e-01
2.19686091e-01 7.91865110e-01 -3.17041218e-01 -3.33856136e-01
3.44074428e-01 3.31492394e-01 6.81614697e-01 -1.25913525e+00
2.03601360e-01 -7.47624874e-01 -9.42727737e-03 -1.43192673e+00
1.88780725e-01 -8.13076913e-01 2.57796705e-01 -2.47758150e+00
7.54198194e-01 8.75917301e-02 6.56467602e-02 6.84332997e-02
-8.67035151e-01 -6.47525311e-01 1.13256440e-01 3.54050577e-01
-8.78138721e-01 1.05418718e+00 1.45363402e+00 -2.48712331e-01
-2.69065231e-01 1.13588557e-01 -1.26501000e+00 5.27303457e-01
8.48035574e-01 -3.67262006e-01 -4.48782563e-01 -4.09546018e-01
1.77654147e-01 2.38108218e-01 -3.35802943e-01 -8.78872156e-01
6.16557300e-01 -2.15317205e-01 1.72333896e-01 -1.27134025e+00
4.74555716e-02 -2.83684075e-01 -4.05533046e-01 4.85721618e-01
-6.58802330e-01 -1.89117379e-02 1.38001904e-01 7.34648764e-01
-4.35298324e-01 -5.31764448e-01 3.66469175e-01 -3.06587845e-01
-5.27783930e-01 3.00891578e-01 -3.51001352e-01 3.17439556e-01
5.86061895e-01 -2.66115963e-01 -4.66552854e-01 -5.90620756e-01
1.27289981e-01 7.13198423e-01 2.44870871e-01 5.08320749e-01
7.62738943e-01 -1.06040454e+00 -1.26847267e+00 -6.79209828e-02
-2.82280054e-02 4.90804285e-01 6.72020137e-01 4.18184876e-01
-3.37787032e-01 8.48256469e-01 -9.33649892e-04 -3.34324509e-01
-1.05778217e+00 2.03595847e-01 -2.30093762e-01 -6.30964458e-01
-6.24289274e-01 3.94091070e-01 2.72298723e-01 4.40180674e-02
1.27720416e-01 -3.73315573e-01 -7.35163271e-01 3.55846316e-01
9.81806159e-01 5.32305777e-01 -5.23001812e-02 -4.73981529e-01
1.01920553e-01 3.29809904e-01 -4.53738868e-01 -1.12916324e-02
1.35296845e+00 -2.07606524e-01 -3.86330873e-01 5.77967823e-01
1.11800122e+00 -6.39278591e-02 -8.41712117e-01 -2.90409982e-01
-1.16855204e-01 2.36525647e-02 1.31379679e-01 -6.09113455e-01
-6.37927532e-01 1.03801978e+00 -4.72581565e-01 3.69084775e-01
1.05200994e+00 -3.34767967e-01 1.36727166e+00 5.16626179e-01
-1.96438320e-02 -1.23425746e+00 9.80222076e-02 8.30933452e-01
9.61923420e-01 -9.35177088e-01 3.81111652e-01 -2.39962667e-01
-8.18376541e-01 1.09802830e+00 6.51056767e-01 -1.06619699e-02
-3.08702569e-02 -2.86301285e-01 -3.66145045e-01 -1.00222817e-02
-7.30920136e-01 1.91408366e-01 3.52343082e-01 2.69703716e-01
3.54289919e-01 -1.48173124e-01 -7.24388003e-01 1.01347959e+00
-3.26538891e-01 -1.85158357e-01 7.22624600e-01 7.60193586e-01
-1.06653953e+00 -9.26277578e-01 -2.48082966e-01 9.54420328e-01
-5.03203213e-01 -2.04510346e-01 -4.08281863e-01 1.32286206e-01
-4.24904883e-01 9.89844799e-01 2.40481142e-02 -1.03313059e-01
2.70763934e-01 9.78307873e-02 -4.73520793e-02 -1.04580557e+00
-5.10397971e-01 2.27533057e-02 2.34061554e-02 -8.43719989e-02
-1.94620818e-01 -7.03348577e-01 -1.41824090e+00 -2.00515434e-01
-4.10040677e-01 6.76283598e-01 7.81219065e-01 1.02930129e+00
6.18436515e-01 9.87680376e-01 7.58209646e-01 -7.77165413e-01
-7.27711976e-01 -1.48751211e+00 -3.94594550e-01 2.12720148e-02
3.32937211e-01 -1.11551687e-01 -1.81215391e-01 -1.65505320e-01]
|
[12.560626029968262, 9.465726852416992]
|
6bb8277e-1bdf-43bc-908b-c3b1b8aa4432
|
optimal-energy-management-in-autonomous-power
|
2208.08953
| null |
https://arxiv.org/abs/2208.08953v1
|
https://arxiv.org/pdf/2208.08953v1.pdf
|
Optimal Energy Management in Autonomous Power Systems with Probabilistic Security Constraints and Adaptive Frequency Control
|
The decarbonization of many heavy power-consuming industries is dependent on the integration of renewable energy sources and energy storage systems in isolated autonomous power systems. The optimal energy management in such schemes becomes harder due to the increased complexity and stability requirements, the rapidly varying operating conditions and uncertainty of renewable sources, the conflicting objectives across different timescales, the limited amount of reliable power sources and energy storage. The state of charge management when energy storage is used for multiple services, such as optimal scheduling and frequency support, is one of the most notorious problems in this context. To address this issue, an optimal energy management system is proposed in this paper. It co-optimizes the primary frequency control layer and the dispatch schedule of conventional generators and energy storage by taking advantage of an algorithm that provides adaptive active power demand uncertainty quantification, theoretical guarantees for frequency stability, and bounds for the reserves for frequency support assigned to the energy storage system. A convex reformulation is derived enabling the efficient solution of the involved optimization problem, being a test case of an isolated offshore oil and gas platform presented for validation.
|
['Elisabetta Tedeschi', 'Vincenzo Trovato', 'Erick Alves', 'Spyridon Chapaloglou']
|
2022-08-18
| null | null | null | null |
['energy-management']
|
['time-series']
|
[-3.14647228e-01 3.20336908e-01 -2.43466720e-01 3.42659056e-01
-2.79659927e-01 -8.57169330e-01 4.15036052e-01 3.82667810e-01
-5.39316460e-02 1.34953415e+00 -2.79595435e-01 -8.35164413e-02
-9.28090036e-01 -6.23697996e-01 -2.15484500e-01 -1.26427662e+00
-3.03590983e-01 4.08902019e-01 -3.95801157e-01 -1.92282081e-01
1.22389041e-01 7.36904740e-01 -1.73201966e+00 -6.99573398e-01
1.01069474e+00 1.52613902e+00 4.46430922e-01 2.05896467e-01
3.21982384e-01 -1.49975851e-01 -6.76462650e-01 5.09572268e-01
3.90065521e-01 -9.10352394e-02 -6.94698617e-02 -1.14809185e-01
-9.51162934e-01 -1.10123962e-01 4.01042223e-01 1.27348566e+00
4.89486217e-01 5.92405677e-01 4.71709967e-01 -1.46056271e+00
-2.18507737e-01 6.21434271e-01 -1.94558054e-01 4.63737458e-01
-2.96251029e-01 -1.39086977e-01 7.75615275e-01 -7.13754714e-01
6.12386465e-02 4.43518013e-01 -2.46412963e-01 2.99524188e-01
-9.21797454e-01 -1.06923714e-01 -5.62929474e-02 2.39708662e-01
-1.25197697e+00 -3.93709064e-01 5.35036504e-01 -4.70880449e-01
1.04569721e+00 3.81798863e-01 1.04683483e+00 5.47594130e-02
2.41743147e-01 1.06862709e-01 9.10474896e-01 -4.85205382e-01
8.19875002e-01 2.33845994e-01 -2.36348599e-01 -2.36063913e-01
8.91247869e-01 2.14801908e-01 -8.06711167e-02 3.65308113e-02
6.01971522e-02 -4.14815754e-01 -4.79543597e-01 -4.54099089e-01
-5.47933102e-01 5.94108701e-01 8.70541558e-02 7.94851780e-01
-4.33964938e-01 -2.05492809e-01 2.09001273e-01 1.58327013e-01
4.44907606e-01 4.63056773e-01 -3.70153964e-01 -9.77425054e-02
-1.00009823e+00 -8.60109180e-02 9.56490040e-01 7.82655537e-01
1.59792274e-01 7.27809250e-01 1.67229950e-01 2.25172684e-01
3.03120166e-01 7.25097775e-01 6.29322231e-01 -5.83873034e-01
1.12382405e-01 3.10513675e-01 6.67686641e-01 -4.56118762e-01
-3.68413001e-01 -7.35611260e-01 -7.79157519e-01 6.04448020e-01
5.74403964e-02 -5.58592379e-01 -1.51729852e-01 1.48133075e+00
4.73329574e-01 -4.59626228e-01 1.85758621e-01 8.11855555e-01
-2.98076332e-01 9.37300861e-01 -1.88127160e-01 -1.14293861e+00
1.26451993e+00 -2.16239333e-01 -1.20533931e+00 1.07701398e-01
1.02779977e-01 -4.30971533e-01 8.36800784e-02 3.44586283e-01
-1.65611994e+00 -9.86543074e-02 -1.42432463e+00 5.24177790e-01
-7.22653747e-01 3.64308059e-01 -2.70149916e-01 3.61435711e-01
-8.42994511e-01 6.27948523e-01 -7.37063110e-01 2.38852203e-01
-1.74454957e-01 3.33807111e-01 -1.05009060e-02 6.18564248e-01
-1.39551282e+00 1.42012477e+00 1.00308967e+00 7.99532413e-01
-1.09336480e-01 -8.17375004e-01 -7.39089847e-01 6.41343892e-01
6.02885365e-01 -3.78356159e-01 9.07875717e-01 -5.95243275e-01
-1.72539270e+00 -9.00716558e-02 3.41850162e-01 -5.88374257e-01
5.08064985e-01 5.66649735e-02 -3.58711571e-01 2.20579103e-01
-1.67559788e-01 -6.26041830e-01 7.07710624e-01 -7.70594835e-01
-5.93531966e-01 -3.67858261e-01 -5.45964956e-01 2.42238402e-01
-3.55211854e-01 -2.43724242e-01 8.66798460e-01 -5.32081842e-01
-4.31548744e-01 -7.73575485e-01 -1.55117050e-01 -6.41318381e-01
-6.20170310e-02 -4.30586159e-01 7.66318917e-01 -7.34381199e-01
1.23457873e+00 -1.98535681e+00 8.58541846e-01 4.95334148e-01
-5.74202716e-01 1.33370191e-01 7.35686004e-01 7.43337691e-01
-3.69985789e-01 -1.90603491e-02 -2.35874817e-01 3.82245779e-02
4.12253052e-01 4.11979258e-01 -2.53690630e-01 4.99182045e-01
1.20606966e-01 3.18735480e-01 -8.54723275e-01 8.07165056e-02
4.41734552e-01 1.81019619e-01 2.93079257e-01 6.68385327e-02
-4.72058028e-01 1.24794707e-01 -4.23289269e-01 4.68752652e-01
5.04309177e-01 -6.52263388e-02 3.34953189e-01 -1.38689518e-01
-8.43368113e-01 -3.69224429e-01 -1.62495542e+00 1.42403412e+00
-8.32620144e-01 2.15119466e-01 7.96775222e-01 -1.54396093e+00
5.79496920e-01 4.10290241e-01 9.73202705e-01 -6.39660001e-01
1.32346585e-01 7.61165261e-01 7.19894916e-02 -3.93143594e-01
5.24085462e-01 -2.83042699e-01 1.64312288e-01 4.31165695e-01
-3.32941152e-02 -2.09530100e-01 7.11448491e-01 -1.76084608e-01
4.14824665e-01 -1.26112208e-01 6.23496413e-01 -1.11044323e+00
8.18751216e-01 -3.01148683e-01 7.03926563e-01 -4.57560271e-01
3.24130625e-01 -3.13807249e-01 3.82062078e-01 1.72847673e-01
-9.26783323e-01 -6.48517489e-01 -4.60221648e-01 3.15418631e-01
9.97653157e-02 1.52394682e-01 -2.15400696e-01 -8.08464140e-02
3.88169914e-01 1.17168248e+00 -3.79221559e-01 -1.52338177e-01
-4.61834699e-01 -9.72093582e-01 -6.39550984e-01 -6.66157603e-02
-3.45320739e-02 -3.79020810e-01 -1.30580735e+00 4.71028984e-01
3.56920034e-01 -8.84909511e-01 -8.84919167e-02 6.10431552e-01
-4.68488455e-01 -1.01812506e+00 -8.90793622e-01 -8.59962106e-02
6.84295714e-01 -3.53090525e-01 1.05150676e+00 9.87804905e-02
-2.64212370e-01 2.57324338e-01 -1.60416961e-01 -6.99221313e-01
-3.75095397e-01 -6.50290325e-02 5.56759655e-01 -8.38770047e-02
-5.55792451e-01 -6.36789978e-01 -5.97881734e-01 2.19655871e-01
-9.56093967e-01 -3.60033929e-01 4.33346719e-01 8.24375510e-01
5.55584311e-01 9.79093492e-01 1.30174553e+00 -3.46856415e-02
6.79458976e-01 -8.35297585e-01 -1.54344893e+00 5.40566981e-01
-1.03600109e+00 7.72350207e-02 8.49375606e-01 -6.23140596e-02
-9.65262234e-01 -9.66499895e-02 5.31546652e-01 -1.64259851e-01
4.17278677e-01 5.20045817e-01 -4.88936394e-01 -9.70709696e-02
-1.54057771e-01 2.86383897e-01 2.18472376e-01 -4.02435690e-01
6.67987540e-02 3.78148615e-01 4.15785015e-01 -5.14451921e-01
1.07282758e+00 -3.36132199e-02 8.43508899e-01 -7.08732426e-01
-3.43099177e-01 -7.68274963e-02 -3.26744407e-01 -3.79837573e-01
4.48513299e-01 -5.94368458e-01 -9.52427804e-01 1.02529220e-01
-8.36698949e-01 -5.08603230e-02 -7.71813452e-01 4.85755950e-01
-5.08225441e-01 3.15062612e-01 2.75235891e-01 -1.40560365e+00
-4.93823439e-01 -1.04236734e+00 3.40396911e-01 5.05817652e-01
2.80425996e-01 -1.02867174e+00 1.81754716e-02 -3.41370255e-01
7.60240734e-01 8.47068608e-01 7.48600781e-01 -1.44357502e-01
-5.31840265e-01 4.86106612e-02 4.25732255e-01 7.20478714e-01
4.15560454e-01 8.21823403e-02 -2.78637111e-01 -7.20244884e-01
3.43079388e-01 8.68472084e-02 4.39067371e-02 3.54830742e-01
5.87718546e-01 -6.69706523e-01 -1.57651559e-01 8.60831589e-02
2.04979300e+00 3.92252296e-01 3.51866409e-02 3.17259938e-01
-3.19712371e-01 6.99971795e-01 8.14762890e-01 8.55695963e-01
7.57870916e-03 8.11507285e-01 8.75455141e-01 6.02176189e-01
6.89602375e-01 6.94771409e-01 2.14320764e-01 6.55194581e-01
-9.71588865e-02 -3.80579084e-01 -4.87094909e-01 7.62872279e-01
-1.81265604e+00 -9.00965989e-01 4.81566519e-01 2.49415278e+00
6.76322758e-01 -1.42363831e-01 7.83892050e-02 5.64038932e-01
6.02047563e-01 -1.54487714e-01 -7.29941130e-01 -4.84649658e-01
-2.11140424e-01 -1.96469948e-02 6.42793000e-01 3.64457458e-01
-3.87123168e-01 -3.95505220e-01 4.63528395e+00 6.79972589e-01
-9.06414211e-01 -2.42484678e-02 4.45049256e-01 -5.88926673e-01
-2.68762052e-01 -2.57293463e-01 -5.77058077e-01 1.34031868e+00
1.25830925e+00 -1.28257263e+00 8.31470490e-01 7.56040871e-01
6.96766675e-01 -6.57934308e-01 -8.11133564e-01 5.10392785e-01
-3.96043330e-01 -1.16253710e+00 -6.59972012e-01 3.12036723e-01
8.81918967e-01 -1.13758974e-01 -2.90215582e-01 -2.46786475e-01
-1.82695970e-01 -6.38785183e-01 9.60161269e-01 7.86634743e-01
4.33969736e-01 -1.23571861e+00 8.03009868e-01 4.74647135e-01
-1.15697467e+00 -6.72211528e-01 -7.19839260e-02 -3.11801676e-02
8.89574766e-01 1.25600839e+00 -1.41172633e-01 1.12854981e+00
6.67665362e-01 3.00438136e-01 2.18446776e-01 8.70561481e-01
-8.36843029e-02 -1.05936982e-01 -8.09414446e-01 -2.28818819e-01
-1.50272340e-01 -6.78233087e-01 5.91369748e-01 4.61007804e-01
8.66788447e-01 3.37997109e-01 -8.32752883e-02 8.06173503e-01
1.89091370e-01 -4.74385917e-02 -4.39470619e-01 -1.79927379e-01
8.13789666e-01 1.56672406e+00 -6.96796298e-01 -1.89331263e-01
-1.34179533e-01 4.85557923e-03 -3.00290048e-01 1.76955238e-01
-5.43868899e-01 -4.19524133e-01 7.55486608e-01 -5.39024398e-02
7.50233755e-02 -4.67581004e-01 -2.85562664e-01 -8.75872970e-01
4.06969011e-01 -9.90765393e-02 4.53240812e-01 -3.51512104e-01
-1.00849247e+00 3.45467985e-01 4.27969843e-01 -1.31291997e+00
-1.02849424e+00 -5.03470480e-01 -6.80639267e-01 1.25223160e+00
-2.11583614e+00 -2.94173181e-01 1.89777911e-01 3.51436436e-01
4.10428196e-01 -3.44736218e-01 5.44870853e-01 3.47485125e-01
-9.93940949e-01 -1.36131063e-01 8.70877206e-01 -6.03980243e-01
-2.86242992e-01 -1.54866874e+00 -6.92621946e-01 1.18608129e+00
-6.34837389e-01 9.11672562e-02 9.16884184e-01 -3.70849758e-01
-1.86946297e+00 -5.16196966e-01 4.83000875e-01 5.15946686e-01
1.13907015e+00 -8.17672983e-02 -7.59463370e-01 1.40465409e-01
7.30248094e-01 8.90108049e-02 5.76745927e-01 -8.03880930e-01
7.61936963e-01 -1.00338869e-01 -1.21378195e+00 1.40820235e-01
2.15424955e-01 1.25874183e-03 -2.27863505e-01 4.56498504e-01
2.05824092e-01 -4.64569241e-01 -1.37730050e+00 3.32255036e-01
3.48759830e-01 -4.69535410e-01 6.41291380e-01 -1.79741293e-01
-1.70372114e-01 -3.21962774e-01 -8.86020362e-02 -1.84862053e+00
-6.97336905e-03 -1.00189328e+00 -7.99088001e-01 1.28834188e+00
1.83864400e-01 -9.27942693e-01 1.86321482e-01 7.02615142e-01
-1.89617127e-01 -7.83313394e-01 -1.88103271e+00 -1.00390565e+00
1.34581989e-02 5.08847773e-01 6.68872178e-01 7.40227938e-01
6.54940367e-01 -6.88810572e-02 1.07844204e-01 6.72639072e-01
9.24138904e-01 3.70690823e-01 -1.76926926e-01 -1.17139387e+00
-2.82281548e-01 -5.72279572e-01 8.45650025e-03 2.32989773e-01
3.07720482e-01 -6.05708778e-01 -3.20360139e-02 -1.59022033e+00
-8.60943913e-01 -2.47817665e-01 -3.54102284e-01 8.56226757e-02
3.36506397e-01 -3.27920049e-01 4.02431667e-01 4.30113077e-02
2.16843188e-01 9.12932515e-01 6.35163248e-01 -7.09954947e-02
4.06889021e-02 2.94984996e-01 -1.25587031e-01 3.37233603e-01
9.47049677e-01 5.88484146e-02 -8.38676214e-01 -1.76607013e-01
6.91170394e-01 5.48108041e-01 -1.69726595e-01 -9.69975352e-01
2.21155450e-01 -3.46234143e-01 -2.77567003e-02 -4.64601338e-01
1.20970644e-01 -1.61216784e+00 8.46162438e-01 7.57442653e-01
2.36538574e-01 3.75614256e-01 7.57621750e-02 4.44922596e-01
-1.93177477e-01 -6.23764694e-01 7.28897989e-01 -5.91870695e-02
-5.83337843e-01 -1.13870487e-01 -2.88182110e-01 -2.85028458e-01
1.68807459e+00 5.74195459e-02 -4.72583592e-01 5.51824924e-03
-1.09667110e+00 9.41811621e-01 1.89502344e-01 2.85135001e-01
1.02703407e-01 -1.10380852e+00 -4.95386690e-01 5.73907718e-02
-5.10663092e-01 7.44528770e-02 2.38144770e-01 7.77810574e-01
-3.29768062e-02 4.73209321e-01 -3.54376882e-01 -2.81439424e-01
-6.71168506e-01 5.84679961e-01 7.21907735e-01 -2.92150944e-01
-2.15741336e-01 -8.68170410e-02 -8.20493221e-01 7.29254901e-01
-2.33266264e-01 -5.02905190e-01 -2.43011922e-01 7.53497124e-01
4.25618082e-01 9.96140480e-01 5.02141953e-01 -2.64009178e-01
-2.67713338e-01 3.81393075e-01 7.78230369e-01 6.35863692e-02
1.44800150e+00 -4.29442108e-01 -4.87884283e-01 3.08780760e-01
8.14141333e-01 -1.28596872e-01 -9.10818160e-01 1.64522320e-01
1.59542352e-01 -1.37017548e-01 3.09302241e-01 -9.20831978e-01
-1.32236850e+00 2.22164735e-01 3.15039992e-01 1.19648874e+00
1.33094490e+00 -6.60371542e-01 1.87009662e-01 2.10528359e-01
5.42420268e-01 -1.65276885e+00 -5.81759095e-01 7.64655992e-02
1.06434202e+00 -6.11634433e-01 3.32641006e-01 -4.79857251e-02
1.24232635e-01 1.34598744e+00 -4.48260363e-03 -2.59250998e-02
9.69986618e-01 6.43951118e-01 -6.45476580e-01 3.22382271e-01
-7.15105057e-01 -8.22038297e-03 -2.75465287e-03 -9.28057656e-02
-1.02678820e-01 1.97067708e-01 -9.64520335e-01 5.19172132e-01
-6.21235110e-02 4.00356911e-02 6.75074816e-01 1.05118537e+00
-3.57403547e-01 -9.08061802e-01 -4.69839841e-01 3.30979407e-01
-5.60950696e-01 4.02308106e-01 5.73548079e-01 8.26662600e-01
1.88374206e-01 7.40105629e-01 1.61997601e-01 5.40678680e-01
5.53829968e-01 1.49397254e-01 1.53022259e-01 -4.14677590e-01
-2.52924711e-01 6.24495558e-04 2.95717176e-02 -1.21189311e-01
-5.10516107e-01 -8.88414383e-01 -1.55285788e+00 1.52038172e-01
-7.85430849e-01 1.14200222e+00 1.37502289e+00 1.08108830e+00
3.66752416e-01 6.72821701e-01 1.14681542e+00 -9.85510707e-01
-1.25403512e+00 -7.14973092e-01 -1.31443000e+00 -3.97204608e-01
3.03148478e-01 -8.45631242e-01 -9.02237713e-01 -3.99373859e-01]
|
[5.63739013671875, 2.5111825466156006]
|
da500857-6e29-4d2a-8560-c1d88930f310
|
complex-program-induction-for-querying
| null | null |
https://aclanthology.org/Q19-1012
|
https://aclanthology.org/Q19-1012.pdf
|
Complex Program Induction for Querying Knowledge Bases in the Absence of Gold Programs
|
Recent years have seen increasingly complex question-answering on knowledge bases (KBQA) involving logical, quantitative, and comparative reasoning over KB subgraphs. Neural Program Induction (NPI) is a pragmatic approach toward modularizing the reasoning process by translating a complex natural language query into a multi-step executable program. While NPI has been commonly trained with the {`}{`}gold{'}{'} program or its sketch, for realistic KBQA applications such gold programs are expensive to obtain. There, practically only natural language queries and the corresponding answers can be provided for training. The resulting combinatorial explosion in program space, along with extremely sparse rewards, makes NPI for KBQA ambitious and challenging. We present Complex Imperative Program Induction from Terminal Rewards (CIPITR), an advanced neural programmer that mitigates reward sparsity with auxiliary rewards, and restricts the program space to semantically correct programs using high-level constraints, KB schema, and inferred answer type. CIPITR solves complex KBQA considerably more accurately than key-value memory networks and neural symbolic machines (NSM). For moderately complex queries requiring 2- to 5-step programs, CIPITR scores at least 3{\mbox{$\times$}} higher F1 than the competing systems. On one of the hardest class of programs (comparative reasoning) with 5{--}10 steps, CIPITR outperforms NSM by a factor of 89 and memory networks by 9 times.
|
['Abhishek Laddha', 'Ghulam Ahmed Ansari', 'Karthik Sankaranarayanan', 'Soumen Chakrabarti', 'Amrita Saha']
|
2019-03-01
| null | null | null |
tacl-2019-3
|
['program-induction']
|
['computer-code']
|
[ 1.34623244e-01 5.98595440e-01 -5.33984601e-01 -4.30926502e-01
-1.29708207e+00 -5.80835044e-01 -1.03149628e-02 3.29609245e-01
-2.69396156e-01 7.77007043e-01 -7.47098261e-03 -1.19351375e+00
-1.68773696e-01 -1.27497554e+00 -1.21816134e+00 1.23975866e-01
-1.31052300e-01 7.61878073e-01 4.41192418e-01 -3.91508669e-01
1.97414562e-01 5.05177937e-02 -1.33543766e+00 6.21795237e-01
1.05610478e+00 9.49568093e-01 3.99528295e-02 9.44513023e-01
-4.86165911e-01 1.57686162e+00 -6.03239894e-01 -7.36964345e-01
-1.31192848e-01 -1.73989952e-01 -1.45281625e+00 -1.00991917e+00
4.18231040e-01 -5.28127909e-01 -2.95482457e-01 1.13384283e+00
-4.99246866e-02 1.59723371e-01 2.97306538e-01 -1.29856443e+00
-9.50581670e-01 1.08307874e+00 -1.44292593e-01 9.27541405e-02
6.91054702e-01 3.83560926e-01 1.57868028e+00 -7.05737710e-01
3.57637942e-01 1.26482332e+00 7.97808528e-01 5.14082670e-01
-1.45359421e+00 -3.07169586e-01 -1.91656902e-01 5.01684658e-02
-1.00963092e+00 -5.44536337e-02 3.19373608e-01 -4.19931561e-01
1.82403874e+00 4.59090620e-01 3.68574947e-01 4.77491111e-01
8.33844319e-02 8.71834219e-01 6.92870855e-01 -4.12102222e-01
1.10837027e-01 2.06182137e-01 7.15608537e-01 1.37358522e+00
2.00738475e-01 2.82325171e-04 -2.30366901e-01 -6.08362436e-01
5.86643338e-01 -2.23332673e-01 -5.27359843e-02 -2.23619625e-01
-1.15307105e+00 1.21479595e+00 5.10666072e-01 -2.80898102e-02
-3.95843619e-03 7.59203792e-01 5.64621329e-01 6.67394578e-01
-2.42301300e-01 1.06428206e+00 -7.25195289e-01 -3.05822760e-01
-7.93782771e-01 5.73967576e-01 1.33652163e+00 1.15832472e+00
1.04156494e+00 3.47565770e-01 -2.59025902e-01 4.37271118e-01
1.21458396e-01 7.63965011e-01 7.54732266e-02 -1.24815917e+00
8.30779314e-01 9.86366093e-01 -6.31367788e-03 -9.39565480e-01
-2.79337227e-01 6.63756207e-02 -3.90222162e-01 2.15567593e-02
6.64304674e-01 -6.71142712e-02 -6.29722357e-01 1.70238328e+00
-1.15383521e-01 -5.64410031e-01 2.85628587e-01 5.52322805e-01
1.13820636e+00 1.16788661e+00 1.66160539e-01 1.45349741e-01
1.31905329e+00 -1.04081523e+00 -7.80607015e-03 -4.95306343e-01
1.16637754e+00 -1.07976161e-01 1.55015016e+00 3.94287407e-01
-1.44492459e+00 -3.85752678e-01 -9.16259587e-01 -3.76962185e-01
-2.67844915e-01 -8.67509618e-02 1.27743804e+00 5.76283157e-01
-1.36065698e+00 3.28464687e-01 -4.22287345e-01 1.39205471e-01
3.16693991e-01 7.82966852e-01 -1.92332178e-01 -3.91834617e-01
-1.29047298e+00 1.08586490e+00 4.41697240e-01 -1.45978585e-01
-9.62523401e-01 -9.72482681e-01 -1.25561011e+00 3.59001696e-01
6.83538616e-01 -7.01172590e-01 1.64967561e+00 -7.17172265e-01
-1.41214955e+00 7.56819308e-01 -8.05315152e-02 -6.18134558e-01
-2.07608491e-01 -5.97296953e-02 -1.69648021e-01 2.56364383e-02
1.35349914e-01 7.62306929e-01 3.91354620e-01 -7.96800077e-01
-2.24579066e-01 -1.26748547e-01 9.52968180e-01 -1.66318104e-01
-2.17975322e-02 1.91034809e-01 -2.46032819e-01 -1.15999490e-01
-2.00274140e-01 -8.67839336e-01 -2.77415365e-01 -2.34064057e-01
-1.71133623e-01 -5.44462860e-01 1.55823931e-01 -8.83782029e-01
1.27317953e+00 -2.02431726e+00 3.66869301e-01 2.31033638e-01
3.40501606e-01 1.62951112e-01 -1.62693903e-01 8.63345116e-02
1.28243297e-01 2.52249092e-01 -2.79300034e-01 4.40429479e-01
5.41240811e-01 4.94740427e-01 -5.39786935e-01 -1.48048446e-01
5.10045290e-01 1.47252333e+00 -8.71775091e-01 -6.02427840e-01
-4.10660356e-01 -3.44945669e-01 -1.25592780e+00 3.05171430e-01
-1.29741001e+00 -3.04198712e-01 -4.36270505e-01 9.11645710e-01
1.67652428e-01 -6.60329819e-01 1.25237584e-01 9.46957693e-02
2.59619862e-01 5.50572038e-01 -8.30233932e-01 1.75182486e+00
-5.76819181e-01 4.80954021e-01 4.10657786e-02 -1.09759414e+00
8.21762145e-01 8.28071609e-02 -7.55029544e-02 -8.26406658e-01
-3.26841980e-01 4.05203730e-01 6.49937987e-02 -6.01672471e-01
7.22338855e-01 -1.96913406e-01 -8.58076274e-01 4.21439290e-01
1.28870890e-01 -7.09373355e-01 1.60775483e-01 3.54446322e-01
1.48720181e+00 1.57872975e-01 9.70873311e-02 -1.94977045e-01
4.87833947e-01 6.95016742e-01 5.41003823e-01 1.14544713e+00
1.04360230e-01 -7.22796051e-03 1.02290475e+00 -5.54371059e-01
-9.62822139e-01 -1.00669944e+00 4.20328081e-01 1.65721118e+00
-2.11996570e-01 -2.42212430e-01 -5.51624954e-01 -5.52170813e-01
2.10403338e-01 9.40871716e-01 -1.95159003e-01 -2.25909010e-01
-9.11170244e-01 -4.07262832e-01 1.17268634e+00 9.05042529e-01
4.42284793e-01 -1.45952189e+00 -5.94453335e-01 1.16628319e-01
-3.38492870e-01 -6.81146264e-01 -2.04275042e-01 5.78994513e-01
-9.31458771e-01 -9.61572051e-01 -1.58436924e-01 -8.73654723e-01
6.06916428e-01 -3.10394675e-01 1.81013966e+00 2.81009525e-01
-2.24648297e-01 4.41988915e-01 -1.16597051e-02 -1.17668718e-01
-6.92651033e-01 5.91240823e-02 -4.64029074e-01 -9.90917683e-01
4.00634289e-01 -3.49510759e-01 -6.33942932e-02 -2.09083617e-01
-8.38225067e-01 -3.03865820e-02 8.14907968e-01 1.00317895e+00
2.61132240e-01 -3.71629260e-02 5.00622690e-01 -1.22531116e+00
5.78600526e-01 -5.89335382e-01 -9.96179640e-01 4.83899236e-01
-4.49785173e-01 5.25003493e-01 7.67438054e-01 -2.84490049e-01
-9.76490319e-01 -1.85739517e-01 -1.05389506e-01 -1.85355827e-01
1.03089996e-01 1.03353322e+00 -8.23933035e-02 -3.30219865e-02
1.18968666e+00 1.98594302e-01 -9.52169374e-02 2.33595222e-01
5.41726291e-01 3.19504365e-02 8.79629135e-01 -1.59065163e+00
7.78837562e-01 -2.54663855e-01 -1.60490885e-01 -1.37553290e-01
-6.73515081e-01 -8.17794576e-02 -1.59072146e-01 4.25717622e-01
7.38179266e-01 -7.76830614e-01 -1.16864896e+00 -1.69342130e-01
-1.33230472e+00 -9.67074215e-01 -3.07097256e-01 -6.66721687e-02
-7.08044350e-01 2.24160463e-01 -8.60289156e-01 -6.66933298e-01
-6.23915672e-01 -1.34077668e+00 7.72416234e-01 -2.74685360e-02
-7.01795161e-01 -7.59103954e-01 -8.76727998e-02 6.60064101e-01
5.58648705e-01 7.68619925e-02 2.01502442e+00 -6.11728370e-01
-1.07509339e+00 -1.47144169e-01 -4.57834333e-01 3.82382721e-01
-5.02137184e-01 -1.75961688e-01 -6.12284958e-01 1.33090466e-01
-2.94936150e-01 -9.55057740e-01 5.56503236e-01 7.92853907e-02
1.16258645e+00 -6.25253379e-01 -1.89785715e-02 2.65318543e-01
1.42873728e+00 3.74922574e-01 7.12949038e-01 2.04721540e-01
5.52553296e-01 3.96206260e-01 4.90524232e-01 8.76452550e-02
6.66906536e-01 2.77630072e-02 4.46665317e-01 3.23633164e-01
2.86304742e-01 -2.56189853e-01 6.35456741e-01 5.07904053e-01
1.12774700e-01 2.04609588e-01 -1.45977473e+00 6.80138528e-01
-1.64340448e+00 -8.98735762e-01 -5.38959913e-02 2.01275682e+00
1.38305497e+00 3.92385721e-01 2.33662855e-02 -1.26497090e-01
1.10918581e-01 -1.44242838e-01 -7.14618087e-01 -9.84075010e-01
2.01134682e-01 8.42230737e-01 2.49087214e-01 7.14142621e-01
-6.66792750e-01 1.08192754e+00 6.00926971e+00 4.79500145e-01
-8.35124373e-01 -7.69591928e-02 4.98127669e-01 1.69704817e-02
-8.04053128e-01 2.70615906e-01 -7.55010903e-01 1.00082885e-02
1.48144221e+00 -1.17137596e-01 9.04866040e-01 1.34146750e+00
-6.12622499e-01 -3.85460645e-01 -1.59366548e+00 6.74882948e-01
-1.66302279e-01 -1.73729372e+00 -3.82889025e-02 -3.55358899e-01
5.14654756e-01 -2.02237274e-02 1.85902074e-01 1.42889178e+00
9.76806045e-01 -1.45918119e+00 6.26380265e-01 5.13615727e-01
9.56158876e-01 -8.12170625e-01 5.39148331e-01 4.68322039e-01
-9.29645896e-01 -5.02362907e-01 -3.38614941e-01 -1.67630166e-01
-4.95878994e-01 2.94154882e-01 -8.55346084e-01 2.93489754e-01
7.36120760e-01 3.06721311e-02 -5.74858904e-01 3.42361063e-01
-2.30637819e-01 5.78341067e-01 -3.40156734e-01 -5.51243961e-01
3.64011079e-01 1.04299225e-01 6.63838014e-02 1.18441820e+00
1.58550456e-01 5.86290419e-01 1.07900105e-01 1.58291185e+00
-2.76196539e-01 -2.29023412e-01 -6.95115387e-01 -4.34817106e-01
3.34545344e-01 8.38498175e-01 -1.97124347e-01 -6.71979547e-01
-5.85305810e-01 5.78310370e-01 6.96549058e-01 2.85697848e-01
-9.50729728e-01 -5.91478050e-01 2.13677540e-01 -2.00050220e-01
2.65376717e-01 -1.04699850e-01 -3.99199396e-01 -9.97978091e-01
2.16412053e-01 -1.30201554e+00 6.35186672e-01 -9.41950202e-01
-8.52965236e-01 4.53994602e-01 1.36233017e-01 -2.59653300e-01
-5.86968899e-01 -8.06799650e-01 -3.69121343e-01 1.03833044e+00
-1.28583825e+00 -9.95392799e-01 -4.99937795e-02 5.84497631e-01
1.35809496e-01 -3.23073305e-02 1.20706689e+00 1.50217474e-01
-3.39527309e-01 7.34157145e-01 -3.94862592e-01 2.86072254e-01
1.29292756e-01 -1.38984191e+00 2.90968537e-01 4.45729285e-01
-2.20217690e-01 1.03152061e+00 4.01028067e-01 -4.60725158e-01
-2.21768498e+00 -9.92546380e-01 8.15098882e-01 -8.09230924e-01
9.15919542e-01 -1.04249045e-01 -1.01361358e+00 9.55159664e-01
-8.90620053e-02 4.35823984e-02 6.45573497e-01 2.81934172e-01
-8.82715762e-01 -1.12376608e-01 -1.02779138e+00 6.33188844e-01
7.42093921e-01 -1.02638376e+00 -8.56389403e-01 3.28018546e-01
1.18094790e+00 -6.27063215e-01 -1.27778840e+00 4.88968164e-01
3.18568438e-01 -7.01437533e-01 8.88062775e-01 -9.59469438e-01
9.08935368e-01 -2.20787585e-01 -5.17117560e-01 -7.21027493e-01
-1.24234200e-01 -4.57841873e-01 -3.12238157e-01 7.40557015e-01
7.64885783e-01 -4.87813860e-01 1.06403148e+00 1.13562429e+00
-2.21057802e-01 -9.40539122e-01 -5.61681926e-01 -3.63021255e-01
4.79806870e-01 -6.50059998e-01 5.75916290e-01 6.62050605e-01
5.15126467e-01 4.48724627e-01 3.69721800e-02 2.43447095e-01
1.54132366e-01 6.94459736e-01 8.27890158e-01 -1.03375661e+00
-1.03579199e+00 -5.93774259e-01 4.99565825e-02 -1.05924201e+00
4.61647898e-01 -1.20892954e+00 2.24217325e-01 -1.41713810e+00
2.62188733e-01 -6.59350574e-01 9.79814678e-02 9.65590477e-01
-3.15158702e-02 -3.72307748e-01 -1.13980882e-01 -1.29063621e-01
-7.78791308e-01 3.44108492e-02 8.37720335e-01 -5.81086159e-01
-5.04290685e-02 -1.12665392e-01 -1.00122035e+00 5.13992906e-01
5.30421257e-01 -2.96251416e-01 -5.73879361e-01 -5.16189396e-01
1.15885055e+00 1.03677464e+00 5.26234210e-01 -8.24078977e-01
4.20560688e-01 -3.72170806e-01 -1.68147981e-01 -4.00884807e-01
2.65527725e-01 -3.69448751e-01 -2.82559283e-02 5.63835680e-01
-6.10194743e-01 2.79874712e-01 6.08318090e-01 3.37822258e-01
-3.06414783e-01 -7.02670217e-01 4.90603805e-01 -6.37193739e-01
-8.87739718e-01 -2.74463110e-02 -3.41021985e-01 4.45431560e-01
6.06913447e-01 3.42036411e-02 -6.94153249e-01 -3.71675462e-01
-5.83780587e-01 4.46978152e-01 1.38823450e-01 -1.27250955e-01
6.39193356e-01 -9.25220430e-01 -2.86947340e-01 2.10872497e-02
9.56399143e-02 6.47017300e-01 1.03607737e-02 6.72046542e-01
-9.31949317e-01 7.42809057e-01 -4.80338037e-02 -2.83580393e-01
-9.42000151e-01 6.39993489e-01 3.59466821e-01 -7.58347869e-01
-1.27586290e-01 1.18287790e+00 8.54822546e-02 -1.08061051e+00
2.65970767e-01 -1.01061583e+00 3.30365032e-01 -5.23274362e-01
1.45855367e-01 1.71082318e-01 -6.46027774e-02 1.71229661e-01
-2.58640170e-01 -8.44717771e-02 -9.69917923e-02 1.26811266e-01
1.39744747e+00 9.13332045e-01 -8.44065011e-01 1.60524666e-01
1.03361464e+00 -2.52843052e-01 -5.86464703e-01 -3.84141415e-01
4.21502799e-01 1.36159910e-02 -4.35218841e-01 -9.82096612e-01
-5.74682534e-01 8.86106730e-01 -2.10654974e-01 9.76323187e-02
8.41159940e-01 1.49344280e-01 8.01143944e-01 1.30861771e+00
6.65619552e-01 -6.62189186e-01 4.75217551e-01 8.97862136e-01
7.57713199e-01 -1.17403376e+00 -8.26462209e-02 -1.16822645e-01
-4.24301654e-01 1.11538827e+00 9.57855105e-01 -1.64988730e-02
1.54759586e-01 4.22557384e-01 -6.04286969e-01 -3.42134297e-01
-7.64801562e-01 2.96155751e-01 7.12607205e-02 3.60766292e-01
2.14549720e-01 5.52081279e-02 4.07810479e-01 9.77929056e-01
-3.98583591e-01 8.76730606e-02 4.05757308e-01 1.09852755e+00
-5.65601230e-01 -8.46393108e-01 -4.71245706e-01 7.59034634e-01
-3.32090557e-01 -5.24320483e-01 -6.98808432e-02 9.79316950e-01
-2.00625688e-01 6.07243001e-01 -1.62489861e-01 -2.91513324e-01
2.22023144e-01 4.03897136e-01 6.28936470e-01 -9.06701267e-01
-8.11344266e-01 -7.55563140e-01 5.45624197e-01 -5.55526555e-01
1.28295302e-01 -2.15018824e-01 -1.83794880e+00 -6.65836990e-01
-1.63315073e-01 2.08948731e-01 2.50909120e-01 7.69237339e-01
2.12438688e-01 4.57206696e-01 -9.68826264e-02 -1.92844614e-01
-1.00497842e+00 -5.73219121e-01 -6.03377596e-02 -5.03995083e-02
2.03820199e-01 -1.79023460e-01 -1.29592434e-01 -5.06793000e-02]
|
[9.396053314208984, 7.495462894439697]
|
e11bbaf2-e7fe-406c-8333-284006997903
|
tenet-triple-excitation-network-for-video
|
2007.09943
| null |
https://arxiv.org/abs/2007.09943v2
|
https://arxiv.org/pdf/2007.09943v2.pdf
|
TENet: Triple Excitation Network for Video Salient Object Detection
|
In this paper, we propose a simple yet effective approach, named Triple Excitation Network, to reinforce the training of video salient object detection (VSOD) from three aspects, spatial, temporal, and online excitations. These excitation mechanisms are designed following the spirit of curriculum learning and aim to reduce learning ambiguities at the beginning of training by selectively exciting feature activations using ground truth. Then we gradually reduce the weight of ground truth excitations by a curriculum rate and replace it by a curriculum complementary map for better and faster convergence. In particular, the spatial excitation strengthens feature activations for clear object boundaries, while the temporal excitation imposes motions to emphasize spatio-temporal salient regions. Spatial and temporal excitations can combat the saliency shifting problem and conflict between spatial and temporal features of VSOD. Furthermore, our semi-curriculum learning design enables the first online refinement strategy for VSOD, which allows exciting and boosting saliency responses during testing without re-training. The proposed triple excitations can easily plug in different VSOD methods. Extensive experiments show the effectiveness of all three excitation methods and the proposed method outperforms state-of-the-art image and video salient object detection methods.
|
['Xin Yang', 'Guoqiang Han', 'Chu Han', 'Sucheng Ren', 'Shengfeng He']
|
2020-07-20
| null |
https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/3089_ECCV_2020_paper.php
|
https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123500205.pdf
|
eccv-2020-8
|
['video-salient-object-detection']
|
['computer-vision']
|
[ 2.34117314e-01 7.57909715e-02 -2.14063957e-01 -4.85119708e-02
-3.36747795e-01 -1.65253624e-01 3.94604683e-01 -5.91408163e-02
-3.75350773e-01 6.62887156e-01 3.72873425e-01 8.24500695e-02
-5.71751408e-02 -5.89920402e-01 -8.74949515e-01 -7.62629509e-01
1.85853288e-01 -9.92488638e-02 1.00002253e+00 -2.88756192e-01
3.76834184e-01 2.32816190e-01 -1.75807452e+00 2.12546870e-01
1.23921287e+00 8.99605572e-01 8.16766739e-01 2.62542397e-01
-1.40452862e-01 9.12844718e-01 -3.52720737e-01 8.14242363e-02
9.40771624e-02 -3.91602308e-01 -8.44368100e-01 2.36461163e-01
4.15804654e-01 -4.84043598e-01 -3.17689657e-01 1.20966160e+00
5.10014772e-01 4.08327401e-01 2.95019895e-01 -1.28433347e+00
-8.33347023e-01 6.48363292e-01 -1.15383649e+00 6.34866536e-01
5.73139787e-02 1.57960415e-01 8.17458093e-01 -1.04996908e+00
4.21389461e-01 1.21915078e+00 4.82499003e-01 6.17738962e-01
-1.05465090e+00 -6.72565997e-01 7.27375567e-01 4.30347353e-01
-1.19657552e+00 -1.42385498e-01 1.08454514e+00 -4.60742563e-01
5.91528535e-01 1.75782129e-01 8.99381697e-01 9.96286869e-01
-1.06237881e-01 1.26994371e+00 8.70871127e-01 -3.38591367e-01
1.25238299e-01 2.07781091e-01 2.47971248e-02 8.00730586e-01
1.41577601e-01 3.93776596e-01 -6.38320506e-01 -3.26187955e-03
1.20451307e+00 1.79286167e-01 -4.91255254e-01 -7.95948207e-01
-1.31133842e+00 7.62999713e-01 7.38852680e-01 2.50355482e-01
-4.80758548e-01 7.10086897e-02 3.64213139e-01 -1.27113074e-01
1.84691682e-01 4.69462097e-01 -3.54022771e-01 2.09113047e-01
-9.82477963e-01 1.21634007e-01 -1.55466199e-01 1.00707459e+00
8.61368835e-01 3.86482686e-01 -6.39635384e-01 5.93062937e-01
1.07264861e-01 2.58974671e-01 5.33591866e-01 -6.60160661e-01
2.70081908e-01 5.70500731e-01 1.98302776e-01 -1.07931828e+00
-3.37969452e-01 -6.58760011e-01 -6.68445706e-01 1.15256049e-01
1.13110974e-01 -2.44606435e-01 -1.03606057e+00 1.79392850e+00
5.93428016e-01 7.97850072e-01 -1.67845517e-01 1.42258954e+00
9.92678285e-01 8.68327022e-01 3.85984719e-01 -2.19233766e-01
1.20439219e+00 -1.24597728e+00 -7.20425665e-01 -2.72389233e-01
2.83313364e-01 -6.31416380e-01 1.27168667e+00 2.37866547e-02
-1.23549104e+00 -9.39794421e-01 -9.00026917e-01 5.33185303e-02
9.93425027e-03 2.53636986e-01 5.78175485e-01 -1.02753006e-02
-8.95928264e-01 4.45256889e-01 -7.06583202e-01 -6.32630289e-02
5.55698872e-01 2.87560821e-01 1.87622577e-01 1.30579636e-01
-1.31476974e+00 6.90846384e-01 6.68935180e-01 -3.54656540e-02
-1.24179006e+00 -9.90548015e-01 -9.22570348e-01 2.45180562e-01
5.93258440e-01 -4.85202461e-01 1.09037733e+00 -1.50065732e+00
-1.29516685e+00 4.42576379e-01 -4.98681553e-02 -2.04143241e-01
3.17048341e-01 -3.46944422e-01 -2.86299467e-01 2.72286832e-01
7.62845799e-02 1.21270967e+00 1.29249656e+00 -1.36946845e+00
-9.61994886e-01 1.55634522e-01 6.90489188e-02 3.83208692e-01
-6.27456486e-01 4.61424589e-02 -3.57822537e-01 -8.63771915e-01
-5.16249314e-02 -5.05510628e-01 -2.64579177e-01 1.18426979e-01
-1.91947520e-01 -1.92166567e-01 1.18581343e+00 -5.08835733e-01
1.14761591e+00 -2.34994888e+00 3.97251785e-01 3.62955295e-02
3.11465055e-01 4.69287753e-01 -3.99852365e-01 -5.87447137e-02
-1.83532625e-01 -2.78921008e-01 -1.63434684e-01 1.41786402e-02
-2.72271812e-01 6.25901148e-02 -4.73549932e-01 1.25810668e-01
7.20681965e-01 1.04048109e+00 -1.28996193e+00 -6.65888429e-01
3.61942589e-01 5.46789229e-01 -7.00505257e-01 2.09414423e-01
-2.81193465e-01 3.83464903e-01 -5.06940007e-01 4.15352702e-01
7.00209916e-01 -4.03993160e-01 -1.55786410e-01 -1.76215500e-01
-4.34616983e-01 2.26031438e-01 -1.33639193e+00 1.61468291e+00
6.94648456e-03 5.92576206e-01 -9.68165174e-02 -1.05357206e+00
9.76111531e-01 8.85822996e-02 2.51937032e-01 -9.21254814e-01
1.29803523e-01 -1.44458666e-01 -1.21677972e-01 -4.39993650e-01
7.19148755e-01 1.57214433e-01 4.05511379e-01 1.56767100e-01
1.61769912e-01 3.00104737e-01 -3.90684232e-02 2.57298142e-01
5.25903881e-01 3.19619030e-01 4.78728190e-02 -6.78420007e-01
7.44907022e-01 -2.21580435e-02 8.83225501e-01 4.98639435e-01
-3.26391786e-01 5.09470999e-01 4.15367693e-01 -3.82327050e-01
-9.49630797e-01 -1.05599451e+00 4.32953760e-02 1.52377701e+00
8.36584508e-01 -1.02604896e-01 -7.38121629e-01 -6.68424487e-01
-3.68833318e-02 3.59992445e-01 -9.71011639e-01 -3.94339412e-01
-7.14911163e-01 -3.29464376e-01 -5.96035756e-02 8.31655324e-01
6.78686619e-01 -1.47260690e+00 -8.78926396e-01 1.17701799e-01
-2.54941851e-01 -6.74532175e-01 -8.71161819e-01 8.30323324e-02
-7.79718399e-01 -8.51697505e-01 -1.07735503e+00 -1.35202003e+00
9.23520029e-01 8.29206169e-01 8.92407298e-01 3.60970378e-01
-2.14847147e-01 1.55032098e-01 -3.93518031e-01 -3.85541439e-01
9.79596376e-02 2.42292583e-02 -7.73217008e-02 7.45796934e-02
-9.15233977e-03 -4.34302360e-01 -8.60093057e-01 5.19830644e-01
-8.80883276e-01 4.08897012e-01 6.26623333e-01 1.02662003e+00
6.53781831e-01 -2.53312111e-01 5.89801371e-01 -4.97252375e-01
4.02536452e-01 -4.63793039e-01 -6.70852721e-01 2.22104341e-01
-2.11507246e-01 1.02140509e-01 4.51819450e-01 -8.78908634e-01
-1.10911059e+00 2.04938669e-02 2.06109792e-01 -8.04418862e-01
5.52444793e-02 2.45985970e-01 1.10511295e-01 -1.42166689e-01
5.67224920e-01 4.60232526e-01 -2.90018469e-01 -1.80769593e-01
1.95269674e-01 7.47103468e-02 7.25776792e-01 -6.27956629e-01
9.09857094e-01 3.55395943e-01 -4.38128978e-01 -6.11645162e-01
-1.03590047e+00 -4.90513235e-01 -4.88242745e-01 -3.98090541e-01
8.43413413e-01 -9.55834270e-01 -4.26298052e-01 3.68950248e-01
-9.86723363e-01 -4.47336257e-01 -4.11896378e-01 3.45721424e-01
-3.27173948e-01 3.46353799e-01 -4.61695582e-01 -5.75171351e-01
-3.32273930e-01 -1.07466412e+00 1.07930136e+00 7.19956577e-01
2.36377846e-02 -7.86516190e-01 -2.21879989e-01 -1.39579251e-01
3.75225931e-01 2.06423596e-01 5.89567602e-01 -9.17387288e-03
-7.26226568e-01 5.86991727e-01 -5.02631187e-01 4.89573665e-02
2.13698328e-01 8.53156522e-02 -7.67316401e-01 -5.59033394e-01
-3.02246630e-01 -4.13292438e-01 8.86870801e-01 6.32523298e-01
1.25542557e+00 -2.98356175e-01 -3.37131679e-01 5.34827113e-01
1.27098227e+00 1.45546302e-01 5.48136771e-01 4.68437433e-01
7.12238312e-01 6.21897221e-01 1.01462710e+00 5.18526673e-01
1.04441084e-01 6.08370960e-01 5.48809528e-01 -6.93018079e-01
-8.06180611e-02 -4.61172432e-01 2.36447364e-01 3.81409436e-01
4.97507751e-02 2.68449962e-01 -6.28928483e-01 8.92253220e-01
-2.08292532e+00 -1.08154893e+00 -6.22929819e-02 2.03936958e+00
9.28849518e-01 3.25232506e-01 4.21533883e-01 1.14821121e-01
9.66460645e-01 1.45287529e-01 -5.21792889e-01 -2.87541132e-02
-4.23226915e-02 -6.02548793e-02 1.23708330e-01 3.92554343e-01
-1.12234771e+00 1.06540906e+00 5.92044115e+00 7.51787663e-01
-1.29822230e+00 1.10535614e-01 5.25093496e-01 -6.59565851e-02
-5.20981073e-01 -1.17020190e-01 -7.54924774e-01 5.69238126e-01
9.29760486e-02 -1.19909912e-01 3.37651730e-01 1.04510272e+00
1.94500417e-01 2.65176944e-03 -7.71179974e-01 8.39990497e-01
-1.04408979e-01 -1.65807986e+00 2.16670513e-01 -4.61416155e-01
1.09108889e+00 -2.14049861e-01 2.17553288e-01 3.96995455e-01
1.81256592e-01 -6.42877817e-01 1.07475650e+00 2.69000590e-01
4.56953377e-01 -9.28940713e-01 5.08252800e-01 2.80959606e-01
-1.39354539e+00 -3.69282424e-01 -5.10968983e-01 6.50731623e-02
-5.72408130e-03 2.17371076e-01 -3.71777594e-01 2.26748884e-01
9.48991477e-01 6.80899858e-01 -5.93212008e-01 1.21028960e+00
-2.74607182e-01 3.94779444e-01 -3.65740918e-02 -1.86908215e-01
6.20050609e-01 2.80287233e-03 5.79003751e-01 1.18743575e+00
-1.60945524e-02 3.40325624e-01 3.43575746e-01 9.81176376e-01
3.04469198e-01 -6.62051067e-02 -1.91684440e-01 3.84547740e-01
5.80511510e-01 1.22446430e+00 -6.96417332e-01 -4.28017825e-01
-3.32883537e-01 7.85204351e-01 4.65634435e-01 4.91108179e-01
-1.24268043e+00 -2.69440174e-01 4.47920501e-01 1.82920128e-01
7.55504310e-01 1.21774055e-01 -9.29971188e-02 -9.02265012e-01
-1.22602433e-01 -7.82367408e-01 3.55700225e-01 -9.94593680e-01
-8.06026399e-01 5.04563272e-01 5.67892492e-02 -1.32087541e+00
2.24996388e-01 -2.58151054e-01 -8.95873427e-01 7.04662204e-01
-1.81524158e+00 -9.70274031e-01 -6.39380217e-01 8.79156768e-01
6.80539906e-01 1.05164573e-01 1.71481058e-01 2.25782782e-01
-6.14232898e-01 5.54032624e-01 -1.95482045e-01 7.86003545e-02
4.92715925e-01 -1.13664961e+00 1.34055346e-01 1.14283335e+00
-3.88991088e-02 4.32307780e-01 7.25014091e-01 -5.48196197e-01
-8.72046351e-01 -1.18727720e+00 3.94073814e-01 1.56292304e-01
5.46195686e-01 -2.88370550e-01 -1.25361836e+00 5.11873901e-01
2.28322640e-01 6.86316788e-02 1.46247745e-01 1.70017704e-02
-1.03744507e-01 -1.01123795e-01 -9.00951326e-01 6.99676871e-01
1.03062415e+00 -2.28171736e-01 -8.43909144e-01 4.98678759e-02
1.08040047e+00 -5.42794287e-01 -3.46356720e-01 5.05250990e-01
1.45223290e-01 -7.92243540e-01 9.72526670e-01 -6.69369102e-01
6.55346811e-01 -6.72893941e-01 3.40802789e-01 -1.19324291e+00
-9.08043504e-01 -6.27634883e-01 -2.30207711e-01 1.31993973e+00
1.07045487e-01 -2.48121291e-01 8.08886349e-01 8.42080917e-03
-4.88264859e-01 -8.10232759e-01 -6.36821628e-01 -4.53107953e-01
-2.41363153e-01 1.75043032e-01 5.55227637e-01 1.17147148e+00
9.33938175e-02 1.52344033e-01 -4.48797435e-01 3.23833972e-01
5.62723994e-01 3.18974525e-01 4.61818725e-01 -1.05075300e+00
-2.58279473e-01 -5.21024406e-01 -2.18464926e-01 -1.39409173e+00
-1.37277305e-01 -4.06777650e-01 2.55707651e-01 -1.24799156e+00
4.83333349e-01 -3.43213797e-01 -7.40300894e-01 6.64507210e-01
-8.01507592e-01 1.18976727e-01 2.19114244e-01 3.38093750e-02
-8.32365453e-01 9.78326142e-01 1.80757964e+00 -1.90638706e-01
-3.90370607e-01 -4.31055725e-01 -8.02437305e-01 6.01317823e-01
6.78417265e-01 -2.81590134e-01 -7.90902913e-01 -5.17156482e-01
-2.47191682e-01 -1.81962904e-02 6.11471891e-01 -9.49359477e-01
1.94081664e-01 -4.25088078e-01 6.32966161e-01 -7.52109289e-01
-3.05930413e-02 -6.15654588e-01 -2.55372673e-01 5.86789489e-01
-4.03027117e-01 -6.05491474e-02 5.49887300e-01 5.39488018e-01
-2.34947637e-01 -5.60177714e-02 1.11594045e+00 8.08811784e-02
-1.20765805e+00 3.31678599e-01 -2.01426864e-01 1.36644378e-01
1.21839523e+00 -4.12315875e-01 -3.81931096e-01 -5.95012233e-02
-6.01982951e-01 6.37882352e-01 2.68552482e-01 7.21796751e-01
9.89987075e-01 -1.46425307e+00 -6.22885704e-01 2.35302821e-01
5.25547676e-02 1.19625956e-01 4.95423466e-01 8.61521721e-01
-1.87242493e-01 6.90106899e-02 -6.28122747e-01 -8.63453805e-01
-1.13944876e+00 9.87397730e-01 3.91743362e-01 2.48558111e-02
-6.46934807e-01 1.11026466e+00 7.90190995e-01 -5.60381263e-02
5.07753909e-01 -2.13257730e-01 -4.66474295e-01 -1.38776109e-01
6.91287756e-01 2.08975613e-01 -4.09428656e-01 -3.31813753e-01
-3.10609639e-01 5.52697062e-01 -2.69844502e-01 2.07333788e-01
1.18453741e+00 -7.80956820e-02 2.05943212e-01 3.75003740e-02
7.53964126e-01 -2.29756594e-01 -1.88333702e+00 -3.97414833e-01
-3.06816071e-01 -6.89535737e-01 1.45083934e-01 -5.67941308e-01
-1.21276653e+00 7.99315870e-01 6.47249043e-01 6.62128329e-02
1.38290513e+00 1.55279832e-02 6.81605816e-01 -1.20871579e-02
6.08723052e-02 -1.11870480e+00 6.48498774e-01 3.40502888e-01
8.88797522e-01 -1.13011551e+00 -1.36457786e-01 -5.95708668e-01
-8.84660721e-01 7.89089024e-01 1.23621035e+00 -3.05863678e-01
3.04937243e-01 4.24452275e-02 -2.45873868e-01 -2.06838265e-01
-6.16709054e-01 -3.33869845e-01 5.08570850e-01 7.47328401e-01
1.84631869e-01 -2.44707569e-01 -3.37602198e-01 3.47266912e-01
2.21369311e-01 -1.90930352e-01 3.14578116e-01 8.18058908e-01
-7.36036897e-01 -5.53608656e-01 -3.35672915e-01 2.21039250e-01
-3.08897980e-02 -2.11773798e-01 -1.41723961e-01 8.89015675e-01
2.94708580e-01 5.66718936e-01 3.06745827e-01 -3.15516561e-01
3.31387609e-01 -4.34415549e-01 3.27470958e-01 -4.39410955e-01
-4.92614090e-01 1.94932878e-01 -3.29334557e-01 -4.38921005e-01
-4.05643165e-01 -7.20921040e-01 -1.49592352e+00 -1.96441561e-01
-5.94163597e-01 3.89067650e-01 -9.17996243e-02 8.67717266e-01
3.30513507e-01 9.26062047e-01 7.61253834e-01 -1.04557514e+00
-3.43816102e-01 -7.67301857e-01 -3.42012137e-01 3.93387914e-01
5.28002083e-01 -1.08357155e+00 -5.84792346e-02 1.47677198e-01]
|
[9.68899917602539, -0.33926454186439514]
|
324782fa-0de2-4606-87eb-d63fd6ec1824
|
talking-face-generation-with-multilingual-tts
|
2205.06421
| null |
https://arxiv.org/abs/2205.06421v1
|
https://arxiv.org/pdf/2205.06421v1.pdf
|
Talking Face Generation with Multilingual TTS
|
In this work, we propose a joint system combining a talking face generation system with a text-to-speech system that can generate multilingual talking face videos from only the text input. Our system can synthesize natural multilingual speeches while maintaining the vocal identity of the speaker, as well as lip movements synchronized to the synthesized speech. We demonstrate the generalization capabilities of our system by selecting four languages (Korean, English, Japanese, and Chinese) each from a different language family. We also compare the outputs of our talking face generation model to outputs of a prior work that claims multilingual support. For our demo, we add a translation API to the preprocessing stage and present it in the form of a neural dubber so that users can utilize the multilingual property of our system more easily.
|
['Kang-wook Kim', 'Dongho Choi', 'Youseong Lee', 'Hyunjae Cho', 'Seungmin Yang', 'Junhyeok Lee', 'Sang Hoon Woo', 'Hyoung-Kyu Song']
|
2022-05-13
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Song_Talking_Face_Generation_With_Multilingual_TTS_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Song_Talking_Face_Generation_With_Multilingual_TTS_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['talking-face-generation']
|
['computer-vision']
|
[-1.60715193e-01 4.15191829e-01 3.36853378e-02 -4.28223759e-01
-9.76963639e-01 -8.89635086e-01 7.29120255e-01 -8.40002894e-01
3.06684338e-02 6.89272106e-01 4.38743532e-01 -4.42210048e-01
7.88227379e-01 -3.73920262e-01 -6.45161510e-01 -2.85909355e-01
5.41834116e-01 3.85189384e-01 -1.10070512e-01 -3.66149306e-01
-2.09572703e-01 5.29816926e-01 -1.58039796e+00 4.57953751e-01
7.06630230e-01 5.36333919e-01 2.17250735e-01 7.76315331e-01
-8.59496742e-02 4.89893526e-01 -7.44349539e-01 -5.40708721e-01
3.44644070e-01 -6.05737388e-01 -6.18571877e-01 2.14714363e-01
7.92840183e-01 -4.96547610e-01 -3.13144922e-01 6.45453393e-01
6.36691868e-01 -2.72406250e-01 3.22674543e-01 -1.38930762e+00
-6.81599259e-01 7.90740311e-01 -1.91663235e-01 -3.77346098e-01
7.07636237e-01 3.80957931e-01 3.96699041e-01 -9.54223275e-01
1.10902452e+00 1.42893648e+00 2.85209417e-01 9.03813958e-01
-1.03543115e+00 -1.07345617e+00 4.47968729e-02 -2.73169667e-01
-1.58534241e+00 -1.46552980e+00 4.80924994e-01 -2.56377995e-01
9.61334705e-01 8.59677047e-02 3.69075596e-01 1.42729592e+00
1.29522377e-04 3.61795247e-01 7.08245397e-01 -5.88473439e-01
-2.54854530e-01 5.57655215e-01 -5.09526610e-01 8.33829701e-01
-4.15130794e-01 -1.87557280e-01 -6.91055954e-01 -1.56853460e-02
6.07312322e-01 -7.80323505e-01 -3.04555446e-01 1.20217260e-02
-1.29364336e+00 4.59112883e-01 -1.92913219e-01 2.63284355e-01
8.23820084e-02 6.21681381e-03 1.54812694e-01 4.45633888e-01
4.49778467e-01 2.51791980e-02 -7.71026015e-02 9.52007808e-03
-1.00656915e+00 -1.57333054e-02 9.16052043e-01 1.41866767e+00
5.92096329e-01 3.87133956e-01 -1.41778663e-01 7.94352472e-01
3.50446641e-01 8.81313145e-01 4.65278447e-01 -1.18491995e+00
4.94258493e-01 9.01659057e-02 1.22703530e-01 -4.03948396e-01
-2.03281567e-02 7.16924816e-02 -4.59996611e-02 -1.06629804e-01
3.09352577e-01 -5.58009446e-01 -6.84948564e-01 1.95392394e+00
2.36591861e-01 1.23274848e-01 4.33573425e-01 5.97168267e-01
8.16743255e-01 7.55839765e-01 -2.02117875e-01 -3.42275560e-01
1.09614658e+00 -1.05174899e+00 -7.83395767e-01 -6.42795190e-02
2.62933075e-01 -1.18121612e+00 1.10313106e+00 9.38250422e-02
-1.19803977e+00 -6.11778140e-01 -8.77994895e-01 -2.62232959e-01
-1.96437687e-01 4.73981291e-01 1.80128902e-01 7.75807083e-01
-1.59376502e+00 -3.72152552e-02 -6.64245963e-01 -6.74765229e-01
-1.72225624e-01 2.88752377e-01 -6.38048172e-01 4.84320521e-02
-9.62154269e-01 8.32660079e-01 -3.74470539e-02 -2.14368746e-01
-7.51195073e-01 -7.04445317e-02 -1.08862054e+00 -1.76199883e-01
1.40812367e-01 -7.22072005e-01 1.63925934e+00 -1.30392420e+00
-2.08249927e+00 8.44042182e-01 -5.92915416e-01 7.23400433e-03
4.19541538e-01 7.97364712e-02 -7.01253533e-01 4.18703735e-01
1.72716156e-01 1.19235384e+00 8.88987243e-01 -1.10595298e+00
-4.28944647e-01 -1.13426127e-01 1.91557761e-02 2.16219664e-01
-2.82082140e-01 4.13876444e-01 -8.24025512e-01 -5.70752025e-01
-3.77582729e-01 -1.11750329e+00 4.00440067e-01 -1.00857187e-02
-3.87164414e-01 8.23822320e-02 1.01795328e+00 -8.77482593e-01
9.13215756e-01 -2.36108351e+00 3.86471711e-02 -3.12315337e-02
-4.16947544e-01 2.80734338e-02 -3.75042915e-01 4.35866177e-01
-8.27424005e-02 1.94824010e-01 4.22874130e-02 -7.24696636e-01
-1.01530358e-01 -1.09451842e-02 -4.02235985e-01 3.39923322e-01
2.20513746e-01 7.36526310e-01 -6.17998481e-01 -6.45140052e-01
-1.22318648e-01 5.77811480e-01 -6.03373885e-01 2.40104243e-01
-2.87554324e-01 5.93431652e-01 8.91022310e-02 6.92519963e-01
3.63826275e-01 1.85212344e-01 5.86582303e-01 -6.84514046e-02
-4.15723383e-01 4.42462862e-01 -9.68454838e-01 1.88402939e+00
-7.94071257e-01 7.21755147e-01 5.82935333e-01 -9.10997167e-02
7.77921796e-01 8.76206696e-01 1.18879154e-01 -2.96297967e-01
5.19026630e-02 2.39528656e-01 -7.69290179e-02 -6.06371284e-01
5.71511269e-01 -1.72483809e-02 -2.39721835e-02 7.95460224e-01
3.62752199e-01 -3.67351919e-01 3.47552270e-01 2.39674523e-01
2.52487510e-01 5.31242549e-01 -3.94730046e-02 -9.53348652e-02
4.21326101e-01 -3.09065878e-01 2.11679652e-01 2.49012545e-01
7.19717890e-02 5.91215909e-01 3.83277386e-01 1.80689365e-01
-1.10684037e+00 -1.21735907e+00 2.80357711e-02 1.14809072e+00
-3.47467184e-01 -5.97563148e-01 -1.06715465e+00 -5.64886928e-01
-3.68992984e-01 8.15375984e-01 -5.04769087e-02 1.84607893e-01
-6.60174727e-01 3.24450433e-02 8.93679440e-01 2.94121087e-01
2.05332950e-01 -9.70172882e-01 -6.60871863e-02 -1.10222101e-01
-6.18351579e-01 -1.42671835e+00 -1.18972743e+00 -6.00462615e-01
-2.27584794e-01 -7.06767797e-01 -6.93112850e-01 -1.14318204e+00
5.41248679e-01 1.14992142e-01 7.24467158e-01 -3.42807442e-01
4.60217968e-02 5.74796200e-01 -1.10138915e-01 -3.11901331e-01
-9.95160460e-01 1.50494263e-01 3.95243675e-01 1.29964292e-01
-1.28439635e-01 -4.21288103e-01 -7.29956999e-02 3.66371393e-01
-7.99425423e-01 3.89536798e-01 2.93633454e-02 3.44668478e-01
6.99136257e-02 -5.37331939e-01 8.53908539e-01 -3.79860461e-01
6.75783038e-01 -2.79487431e-01 -6.77934945e-01 3.33577007e-01
-1.04418933e-01 -9.77998301e-02 6.77081943e-01 -4.78577614e-01
-1.18668306e+00 3.11128914e-01 -2.54622996e-01 -4.55016226e-01
-2.15612650e-01 1.56731129e-01 -5.95354795e-01 1.92774385e-01
3.69796783e-01 2.89059043e-01 4.02097613e-01 -2.55260289e-01
7.76688516e-01 1.34336305e+00 9.03979778e-01 -6.15130007e-01
6.06368363e-01 2.55576432e-01 -5.49558818e-01 -1.05480158e+00
-1.01878658e-01 7.34336823e-02 -3.97848934e-01 -3.43563497e-01
6.57290816e-01 -1.22275400e+00 -8.09520900e-01 5.35217226e-01
-1.38986540e+00 -3.23860168e-01 -3.34006548e-02 4.16391939e-01
-5.28101325e-01 2.53690630e-01 -6.78562343e-01 -6.09562755e-01
-2.94381231e-01 -1.44591641e+00 1.33050001e+00 2.06333399e-01
-2.94294834e-01 -5.84550977e-01 -6.34284019e-02 3.90100181e-01
4.60448086e-01 -1.43857151e-01 6.88808560e-01 -3.68957341e-01
-3.76964390e-01 1.28832728e-01 1.63888577e-02 2.96860814e-01
3.95277917e-01 5.96655369e-01 -1.22826469e+00 -3.34014744e-01
-1.55762061e-01 -3.76803398e-01 3.69086891e-01 9.13692564e-02
3.91553253e-01 -5.37492037e-01 -2.86368549e-01 6.21178448e-01
6.92965031e-01 2.56284326e-01 4.82864559e-01 -2.86763579e-01
5.46023846e-01 8.49181533e-01 2.87722945e-02 -1.48551015e-03
7.83452511e-01 7.70701766e-01 -1.12921789e-01 -1.37316152e-01
-3.36027503e-01 -3.83282632e-01 1.05317307e+00 9.58352447e-01
2.30346844e-01 -3.69070619e-01 -6.60030246e-01 4.88728523e-01
-1.39946723e+00 -1.05984294e+00 3.28243524e-01 2.04197741e+00
1.06819654e+00 -3.44407231e-01 2.28815466e-01 -5.23302495e-01
8.81629467e-01 -3.41463797e-02 -3.02089334e-01 -5.53627670e-01
-5.36654592e-02 2.01339439e-01 9.65333804e-02 9.41553116e-01
-7.88441837e-01 1.32368445e+00 7.24794006e+00 4.20368612e-01
-1.83125532e+00 1.02336155e-02 3.47325593e-01 -3.77395928e-01
-4.18266058e-01 -1.74963009e-02 -8.08143139e-01 2.01569274e-01
1.20135987e+00 -4.09103334e-01 7.67506719e-01 4.98628557e-01
4.57598537e-01 1.58651441e-01 -1.32313502e+00 8.28126609e-01
5.43272078e-01 -1.19185579e+00 2.30752304e-01 4.15498018e-02
5.77299774e-01 -8.72042682e-03 1.50905848e-01 1.52395755e-01
2.03757316e-01 -1.04642367e+00 1.04034495e+00 3.84202838e-01
1.40535665e+00 -5.90218306e-01 1.35950118e-01 2.95134395e-01
-1.15757871e+00 1.64273262e-01 1.94404334e-01 2.72015184e-01
4.73645687e-01 -5.18621132e-02 -1.27714884e+00 6.83204532e-01
2.64836252e-01 4.43270594e-01 -4.48074102e-01 3.25582594e-01
-2.62631744e-01 3.89335603e-01 -3.63771021e-01 4.46633875e-01
-2.90491134e-01 -1.97643302e-02 5.20583034e-01 1.11173797e+00
7.12274253e-01 -2.79274046e-01 3.68385941e-01 7.70327508e-01
-3.64886612e-01 3.31311136e-01 -8.89157295e-01 -3.10740650e-01
4.95400816e-01 1.26184118e+00 -3.67799282e-01 -3.49489570e-01
-4.98610824e-01 1.05364621e+00 -6.35441542e-02 4.75261033e-01
-6.60666347e-01 -2.88252056e-01 5.47093868e-01 1.19440123e-01
6.49518892e-02 -4.30017769e-01 3.03101063e-01 -1.53684235e+00
6.12275079e-02 -1.34885728e+00 -1.01105429e-01 -1.29060864e+00
-8.82614732e-01 1.08570600e+00 2.96755936e-02 -9.84481633e-01
-9.95868862e-01 -3.24188083e-01 -3.53503406e-01 1.22004986e+00
-1.17241704e+00 -1.57019258e+00 3.10686119e-02 7.60655284e-01
5.59412718e-01 -4.34041202e-01 1.01933157e+00 3.95952493e-01
-4.99981016e-01 6.74490154e-01 -4.37387466e-01 3.03498417e-01
1.35405552e+00 -7.30750561e-01 6.75789237e-01 8.70253563e-01
6.73785806e-02 8.92832458e-01 5.31319499e-01 -6.23725235e-01
-1.30081141e+00 -8.50856781e-01 1.22113037e+00 -3.62998307e-01
6.35693550e-01 -6.88229263e-01 -3.56915265e-01 1.11853850e+00
8.95840168e-01 -4.57488447e-01 6.65333390e-01 -2.35721275e-01
-3.27139050e-01 -1.19274236e-01 -1.13295484e+00 8.93792033e-01
8.50362420e-01 -9.11996245e-01 -3.31340015e-01 1.68117315e-01
8.68687272e-01 -4.92560089e-01 -5.54847538e-01 3.32265869e-02
8.32424939e-01 -7.28020251e-01 5.83282232e-01 -1.57036066e-01
2.27491185e-01 -5.15362859e-01 -2.55827010e-01 -1.34011400e+00
3.50458473e-01 -9.43367600e-01 4.90945816e-01 1.67941093e+00
5.99329174e-01 -7.34963119e-01 2.23425865e-01 3.29105079e-01
-1.58260018e-01 7.69303888e-02 -8.56537461e-01 -5.98346233e-01
1.44878596e-01 -2.00264066e-01 7.90853977e-01 9.58823085e-01
3.01983893e-01 6.58530474e-01 -4.33463752e-01 3.43615979e-01
1.16370678e-01 8.54376107e-02 1.02476096e+00 -7.62645900e-01
-2.58166611e-01 -2.20969722e-01 2.72973448e-01 -8.88922811e-01
6.40322804e-01 -1.16700864e+00 1.91178873e-01 -1.34086740e+00
-1.04360603e-01 5.09382188e-02 5.15011430e-01 7.06204057e-01
3.24790776e-01 2.95561522e-01 3.94465297e-01 1.38086155e-01
-6.37799874e-03 3.91773880e-01 1.08990049e+00 -4.30785865e-02
-3.01809460e-01 -2.22524166e-01 -7.69737720e-01 6.73641860e-01
6.53812766e-01 -2.16097027e-01 -4.93903548e-01 -7.69698024e-01
-2.08954930e-01 4.49708879e-01 1.20239537e-02 -8.68224144e-01
2.78527319e-01 -1.80658456e-02 1.25264347e-01 -1.05463006e-01
4.16456878e-01 -5.74630857e-01 3.44646662e-01 -5.04163206e-02
-2.81530678e-01 2.80416429e-01 4.05184925e-01 -2.51506746e-01
-2.13656142e-01 1.03308186e-01 6.94015801e-01 -6.78948080e-03
-1.86017960e-01 -8.64119851e-04 -6.13186836e-01 -3.18761766e-01
9.15759861e-01 8.39364678e-02 -6.68831706e-01 -8.95997226e-01
-7.86719680e-01 1.27751738e-01 9.00445759e-01 7.23905504e-01
3.43761981e-01 -1.15241683e+00 -8.08881581e-01 5.45065284e-01
1.59642458e-01 -6.19911015e-01 -1.43129990e-01 5.10065198e-01
-4.84930515e-01 4.18691635e-01 -2.32385561e-01 -4.09708798e-01
-1.31276095e+00 4.53943908e-01 4.73893732e-01 4.34079736e-01
-2.12475926e-01 3.09106380e-01 7.10079679e-03 -6.97624207e-01
1.54815838e-01 -7.38593340e-02 4.76941355e-02 4.72776778e-02
4.84213352e-01 -1.81290302e-02 -4.49491367e-02 -1.08667922e+00
-4.18687433e-01 4.56522644e-01 3.29593241e-01 -9.67518210e-01
9.24891472e-01 -3.21487457e-01 -2.72164822e-01 4.32885349e-01
9.17359352e-01 9.25706863e-01 -9.56160367e-01 1.45463437e-01
-5.77086687e-01 -1.20405294e-01 -4.21669960e-01 -7.46042073e-01
-9.42303181e-01 6.78580642e-01 3.82822692e-01 -1.01302959e-01
9.78866756e-01 1.47585720e-01 6.94029450e-01 2.60228068e-01
3.48580748e-01 -9.40708935e-01 -2.33650804e-01 4.94998366e-01
9.03179526e-01 -8.63870919e-01 -3.58367890e-01 -4.13435042e-01
-6.93919539e-01 1.17360890e+00 5.36737859e-01 4.23643827e-01
3.65053833e-01 7.14628518e-01 6.49714470e-01 3.95184457e-01
-9.02435124e-01 -7.59972781e-02 6.83818460e-02 6.33460104e-01
7.04803467e-01 -5.07923476e-02 -2.20839351e-01 4.37165558e-01
-6.41466320e-01 1.67509198e-01 6.84605718e-01 6.88193202e-01
-1.46167055e-01 -1.30711043e+00 -4.60854292e-01 -3.53504330e-01
-4.92485434e-01 -3.55561346e-01 -6.19400501e-01 6.21676028e-01
1.54614657e-01 1.34426439e+00 1.90849915e-01 -2.95281976e-01
1.32250577e-01 6.58291221e-01 5.19569755e-01 -7.12396801e-01
-5.89573562e-01 5.05566597e-01 4.17058617e-01 -4.30533886e-01
-3.28948945e-01 -6.62478924e-01 -1.12708986e+00 -2.89913982e-01
-9.17984545e-03 -5.42976931e-02 8.49561095e-01 8.72164726e-01
6.10799491e-01 1.85573056e-01 6.62967563e-01 -9.40272629e-01
8.17049742e-02 -1.06494212e+00 -2.50513375e-01 -2.66615208e-02
2.69958913e-01 6.84001073e-02 -2.06094384e-01 6.58509731e-01]
|
[13.262125968933105, -0.3623529076576233]
|
83019ec3-8a70-4317-bff9-63497edda49f
|
lipformer-high-fidelity-and-generalizable
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Wang_LipFormer_High-Fidelity_and_Generalizable_Talking_Face_Generation_With_a_Pre-Learned_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Wang_LipFormer_High-Fidelity_and_Generalizable_Talking_Face_Generation_With_a_Pre-Learned_CVPR_2023_paper.pdf
|
LipFormer: High-Fidelity and Generalizable Talking Face Generation With a Pre-Learned Facial Codebook
|
Generating a talking face video from the input audio sequence is a practical yet challenging task. Most existing methods either fail to capture fine facial details or need to train a specific model for each identity. We argue that a codebook pre-learned on high-quality face images can serve as a useful prior that facilitates high-fidelity and generalizable talking head synthesis. Thanks to the strong capability of the codebook in representing face textures, we simplify the talking face generation task as finding proper lip-codes to characterize the variation of lips during a portrait talking. To this end, we propose LipFormer, a transformer-based framework, to model the audio-visual coherence and predict the lip-codes sequence based on the input audio features. We further introduce an adaptive face warping module, which helps warp the reference face to the target pose in the feature space, to alleviate the difficulty of lip-code prediction under different poses. By this means, LipFormer can make better use of the pre-learned priors in images and is robust to posture change. Extensive experiments show that LipFormer can produce more realistic talking face videos compared to previous methods and faithfully generalize to unseen identities.
|
['Jingren Zhou', 'Deli Zhao', 'Yujun Shen', 'Yingya Zhang', 'Shiwei Zhang', 'Kang Zhao', 'Jiayu Wang']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['talking-face-generation', 'face-generation']
|
['computer-vision', 'computer-vision']
|
[ 1.68115422e-01 -1.75830930e-01 -1.46946654e-01 -5.44610679e-01
-7.86926150e-01 -4.88447756e-01 4.11270052e-01 -1.00254130e+00
4.30839390e-01 4.37913030e-01 7.23436773e-01 4.48094666e-01
1.63349092e-01 -3.53812605e-01 -6.91653967e-01 -9.62458611e-01
4.35945392e-01 1.80621706e-02 -1.70426697e-01 -9.20400694e-02
1.36421457e-01 4.29696172e-01 -1.91731524e+00 5.50734103e-01
4.93651330e-01 1.17092884e+00 4.46736902e-01 3.49016905e-01
2.16961782e-02 4.26619142e-01 -3.80545229e-01 -5.05703270e-01
1.56919479e-01 -5.82753420e-01 -4.15153384e-01 5.24446011e-01
5.19659996e-01 -3.57822984e-01 -4.83097404e-01 1.14134181e+00
7.31877327e-01 -7.17093842e-03 6.11699522e-01 -1.36164796e+00
-4.78190184e-01 3.39602500e-01 -5.03474474e-01 -2.20466375e-01
6.68497562e-01 2.06467807e-01 7.02858865e-01 -1.15959847e+00
5.60664654e-01 1.52488661e+00 6.56156540e-01 8.23242009e-01
-1.06466460e+00 -1.18993449e+00 2.35024057e-02 4.27600503e-01
-1.70594287e+00 -1.32365727e+00 1.13243115e+00 -2.92212218e-01
1.07276253e-01 2.78956950e-01 6.04973793e-01 1.48591805e+00
1.06454417e-02 6.44430339e-01 7.80202389e-01 -2.14948833e-01
-9.15799960e-02 1.08488806e-01 -6.92332447e-01 5.49186885e-01
-3.68691653e-01 -5.99078499e-02 -9.93240118e-01 -3.32139828e-03
7.63969481e-01 -1.25401855e-01 -8.97159934e-01 -4.19386983e-01
-1.07159889e+00 6.13820374e-01 1.33051679e-01 1.81901634e-01
-3.48195791e-01 2.05222443e-02 2.80334800e-01 -1.12134084e-01
2.92906314e-01 -1.40905172e-01 -8.63715559e-02 -8.74266401e-02
-1.07876372e+00 -7.04492107e-02 5.02266943e-01 9.26250815e-01
5.55550337e-01 1.47498578e-01 -2.81923115e-01 1.11398947e+00
5.65772474e-01 6.63022399e-01 6.03701055e-01 -1.06574070e+00
1.95138276e-01 -5.13955690e-02 -3.19323651e-02 -1.11369991e+00
-6.53912732e-03 -2.47937799e-01 -9.07169878e-01 -2.29419351e-01
4.85354401e-02 1.95651948e-02 -6.24218225e-01 2.02012610e+00
3.93434912e-01 5.33406019e-01 -8.21872652e-02 9.46168125e-01
7.49916732e-01 7.35060751e-01 -3.09992999e-01 -6.70775831e-01
1.20745349e+00 -6.33061886e-01 -1.11842966e+00 -1.03145540e-01
-2.80113459e-01 -1.10559702e+00 1.03467607e+00 4.36440617e-01
-1.01485777e+00 -8.96896720e-01 -6.22445822e-01 1.53405517e-01
3.05426270e-01 3.61176282e-01 3.25148761e-01 6.66789532e-01
-1.03731322e+00 5.74160852e-02 -4.26150382e-01 -2.52604783e-01
3.98522019e-01 2.60083646e-01 -5.59389234e-01 -2.14823112e-01
-9.79435384e-01 4.31905150e-01 6.03230707e-02 2.93597400e-01
-1.07422078e+00 -5.48173189e-01 -9.97623026e-01 1.03346355e-01
2.12548882e-01 -5.88012815e-01 1.12966168e+00 -1.32557487e+00
-2.05386162e+00 7.53247321e-01 -7.43492723e-01 4.78767902e-02
2.46601075e-01 -4.06818204e-02 -4.13099557e-01 1.67552665e-01
-1.41551733e-01 8.21838856e-01 1.50590348e+00 -1.39870238e+00
-1.62985474e-01 -1.73694029e-01 -4.53274518e-01 2.54163355e-01
-4.66421932e-01 7.58598894e-02 -8.38654578e-01 -8.56358230e-01
1.57832757e-01 -9.37325835e-01 5.25469840e-01 2.16068208e-01
-3.12652260e-01 -8.53198245e-02 9.88248110e-01 -6.89611912e-01
9.73795593e-01 -2.38735008e+00 1.83845937e-01 6.66543320e-02
-2.27254465e-01 1.75299138e-01 -3.15601766e-01 2.36190856e-01
-1.64530486e-01 -2.73430049e-01 1.06502548e-01 -4.67455298e-01
-4.17851582e-02 2.34068185e-01 -5.36862493e-01 5.43761253e-01
1.14475638e-01 5.42365551e-01 -6.42535031e-01 -6.30675197e-01
5.00232242e-02 1.00982022e+00 -8.26300621e-01 4.68819648e-01
1.23382181e-01 5.51010251e-01 -2.04912364e-01 6.87796175e-01
8.42773080e-01 1.21022344e-01 2.49395385e-01 -6.41852021e-01
2.03730568e-01 3.21217515e-02 -1.23216522e+00 1.91589081e+00
-5.94707131e-01 6.71325326e-01 5.80894768e-01 -8.04101110e-01
1.19124019e+00 6.25964642e-01 4.27938312e-01 -3.79052103e-01
1.45698264e-01 -8.80243480e-02 -1.59900874e-01 -8.09439003e-01
-4.39120382e-02 -3.31854343e-01 2.54766822e-01 1.35858715e-01
3.89313638e-01 -4.03161496e-01 -3.16175371e-01 -3.40721935e-01
4.44580585e-01 1.62048131e-01 -4.69007976e-02 -1.25084937e-01
9.53366935e-01 -9.66571867e-01 7.68258989e-01 2.68252268e-02
-9.07156244e-02 9.32408273e-01 1.49474382e-01 -2.04463243e-01
-7.09303498e-01 -9.87937212e-01 -2.08024502e-01 1.09936464e+00
3.24638411e-02 -4.20120507e-01 -1.02430487e+00 -3.17658633e-01
-2.85356998e-01 3.55718106e-01 -4.99322265e-01 -2.75667936e-01
-3.44280124e-01 -3.78783643e-01 3.85786802e-01 2.67595798e-01
6.30835295e-01 -8.97823870e-01 8.92706290e-02 5.89111559e-02
-7.88986921e-01 -1.09299171e+00 -1.01029503e+00 -5.25142789e-01
-2.42181823e-01 -8.60052645e-01 -9.04790342e-01 -1.03560901e+00
6.41886592e-01 3.31704229e-01 5.60924888e-01 -2.27239594e-01
-1.27402440e-01 3.67773741e-01 -1.08829834e-01 -2.77538776e-01
-3.47725034e-01 -4.19076562e-01 4.30106819e-01 9.17385042e-01
6.84381556e-03 -7.26325274e-01 -5.66666782e-01 5.57357669e-01
-6.46966398e-01 1.86567694e-01 3.41441423e-01 1.10226679e+00
4.48463380e-01 1.05423793e-01 6.16572618e-01 -1.49679407e-01
6.25157773e-01 -2.98456430e-01 -3.51183027e-01 3.22022825e-01
3.63476537e-02 -9.15219262e-02 6.22191846e-01 -8.12200606e-01
-1.38332272e+00 3.15614790e-01 -4.17624861e-01 -8.89127612e-01
-4.55350727e-02 -1.14841117e-02 -8.03126633e-01 -1.62019536e-01
3.16745579e-01 5.56482971e-01 3.20116669e-01 -4.72880006e-01
2.87289977e-01 9.18679833e-01 8.20990562e-01 -6.30088210e-01
7.42656052e-01 5.70255816e-01 -9.97346640e-02 -9.62622404e-01
-6.60751462e-01 -2.57683665e-01 -5.55811048e-01 -5.46408296e-01
6.92843497e-01 -1.00833249e+00 -1.17938638e+00 4.86392528e-01
-1.15500724e+00 -1.41257539e-01 8.82299542e-02 4.17724341e-01
-8.36622715e-01 4.63999510e-01 -3.61241817e-01 -8.57113779e-01
-2.50798374e-01 -1.33779347e+00 1.29299116e+00 2.31501654e-01
-7.26159662e-02 -5.95117569e-01 -3.71044688e-02 5.10849297e-01
4.09922689e-01 -5.40751852e-02 4.67204332e-01 -2.03587431e-02
-3.91354531e-01 2.60517038e-02 4.89818864e-02 4.91597891e-01
4.72058654e-01 1.41501531e-01 -1.34555447e+00 -3.26951653e-01
3.21094185e-01 -3.65846902e-01 5.71742356e-01 3.99459153e-01
1.40134895e+00 -6.03245258e-01 -2.40133017e-01 8.05965483e-01
9.99369919e-01 9.21044052e-02 5.92124462e-01 -4.56382930e-01
4.90950674e-01 8.25991273e-01 4.63032037e-01 6.01472914e-01
3.15441966e-01 1.01293051e+00 2.98454285e-01 1.06872775e-01
-3.51098716e-01 -6.27926469e-01 6.42283440e-01 8.90351653e-01
1.52645148e-02 -1.25079855e-01 -6.45762920e-01 4.07239348e-01
-1.48384380e+00 -1.09498715e+00 6.38510942e-01 1.98615611e+00
1.05503583e+00 -2.46784851e-01 2.25832686e-02 1.88527629e-01
9.06979680e-01 6.98294118e-02 -4.45210665e-01 2.95574237e-02
-6.50908351e-02 7.14377761e-02 -1.95705131e-01 5.08458972e-01
-7.92010128e-01 8.15534592e-01 6.32430744e+00 1.07707691e+00
-1.57659245e+00 1.64626073e-02 5.16528964e-01 -1.05547816e-01
-1.80175781e-01 -3.08052897e-01 -6.59629583e-01 6.70481265e-01
7.06485808e-01 -2.41695374e-01 5.98496556e-01 7.14970589e-01
4.74518299e-01 2.86188841e-01 -1.08374310e+00 1.51125646e+00
5.44190884e-01 -1.07028127e+00 2.12915748e-01 -1.07563719e-01
5.23936749e-01 -5.24032056e-01 3.44541609e-01 -3.29903923e-02
-1.76453561e-01 -1.11167073e+00 8.26218307e-01 9.15329814e-01
1.08527744e+00 -7.86582768e-01 4.99207497e-01 4.13099200e-01
-1.29183090e+00 -1.45664558e-01 -3.49511862e-01 3.42922300e-01
5.35177663e-02 5.35327271e-02 -8.99624467e-01 2.65631080e-01
7.35924125e-01 6.79981172e-01 -1.21347763e-01 8.61497283e-01
-1.18274368e-01 5.89602709e-01 -2.21831307e-01 4.68331665e-01
-3.82083535e-01 8.72399807e-02 4.71594095e-01 1.03305972e+00
5.80177784e-01 2.05641165e-01 8.41859281e-02 7.30179965e-01
-1.56411827e-01 1.77484244e-01 -5.64576089e-01 1.98781371e-01
6.39143586e-01 1.20662892e+00 -1.73432961e-01 5.42918444e-02
-3.45410138e-01 9.19095695e-01 -2.90193677e-01 4.07850862e-01
-8.24859262e-01 -1.19645476e-01 9.42991555e-01 2.28129640e-01
3.83541375e-01 7.02345073e-02 2.83975482e-01 -1.22627831e+00
3.53383012e-02 -1.17814147e+00 -1.22647464e-01 -1.18825233e+00
-1.08368838e+00 6.82100296e-01 -1.97967619e-01 -1.28252041e+00
-4.23903108e-01 -3.83307397e-01 -5.86828411e-01 9.48846042e-01
-1.39838696e+00 -1.41106176e+00 -5.31802177e-01 1.17148829e+00
8.46939862e-01 -2.16250926e-01 8.57609987e-01 1.64345711e-01
-4.93797958e-01 9.33866084e-01 -4.75688577e-02 1.11603200e-01
1.12094831e+00 -3.75130981e-01 -1.64211571e-01 5.47294319e-01
1.04602829e-01 6.36341631e-01 7.67289758e-01 -2.97953039e-01
-1.43794012e+00 -9.32399631e-01 6.13011718e-01 -1.07424855e-01
3.55532646e-01 -5.80901444e-01 -8.61488342e-01 4.14099753e-01
2.33608633e-01 1.68835551e-01 7.90655494e-01 -2.45853901e-01
-3.13464195e-01 -6.56384766e-01 -1.08641112e+00 3.98188293e-01
1.05491912e+00 -8.49298418e-01 -4.83582735e-01 2.32448086e-01
2.96549052e-01 -3.24160397e-01 -7.61855543e-01 4.13370311e-01
7.93632209e-01 -9.73544180e-01 1.00681341e+00 -2.24311762e-02
1.39879152e-01 -2.54492372e-01 -2.29089692e-01 -1.15477908e+00
-2.62978077e-01 -1.08680272e+00 9.92031544e-02 1.73051572e+00
-1.73009440e-01 -3.47272068e-01 5.57944059e-01 1.83331981e-01
6.49490952e-02 -4.57660109e-01 -1.04599452e+00 -5.02095997e-01
-3.10569793e-01 -2.77379930e-01 8.12088370e-01 8.58823657e-01
9.45906416e-02 2.69410729e-01 -8.10130000e-01 1.95901036e-01
5.94835162e-01 3.01602110e-03 9.58138287e-01 -1.15391541e+00
-1.51600584e-01 -1.28893375e-01 -3.90638858e-01 -1.18683040e+00
5.85794151e-01 -6.05982721e-01 3.24067414e-01 -1.00720799e+00
1.85463503e-01 -1.69117644e-01 1.11757986e-01 3.77648711e-01
4.34350744e-02 4.07537073e-01 2.36414641e-01 3.61200094e-01
-1.99737459e-01 9.17258263e-01 1.49845648e+00 -1.62606701e-01
1.02637187e-01 1.33927524e-01 -6.59508705e-01 8.20319116e-01
4.92152363e-01 -1.48852155e-01 -6.06965244e-01 -2.46201143e-01
-2.38764182e-01 4.74918902e-01 3.02504569e-01 -9.68260348e-01
3.48530740e-01 -2.26329297e-01 4.88918066e-01 -3.55930418e-01
8.76474977e-01 -8.50854278e-01 3.63600791e-01 4.87551093e-02
-3.35808635e-01 -3.59796464e-01 1.73275322e-01 4.00295019e-01
-4.93928730e-01 1.00421257e-01 9.13543582e-01 5.34280464e-02
-3.42204601e-01 4.90748107e-01 -1.84596509e-01 -1.80899382e-01
8.75348389e-01 -3.30162615e-01 2.28194520e-01 -8.93414557e-01
-8.02798331e-01 -4.78705242e-02 3.94282371e-01 5.52476048e-01
7.71952033e-01 -1.59371066e+00 -6.77337527e-01 8.01989079e-01
-9.29832533e-02 -2.80606508e-01 4.97934818e-01 5.62013030e-01
-2.31196117e-02 2.62917280e-01 -4.91894245e-01 -8.22863579e-01
-1.51608479e+00 6.21271968e-01 4.74301606e-01 4.66846615e-01
-4.09644634e-01 1.02942693e+00 6.52783573e-01 4.38868813e-02
4.76354986e-01 -3.82963941e-02 -2.68433481e-01 1.17280796e-01
7.49016762e-01 -9.78049487e-02 -1.37250274e-01 -1.19553924e+00
-2.52044737e-01 9.48228002e-01 2.64054924e-01 -1.06710568e-01
1.19710445e+00 -4.60610032e-01 4.41552810e-02 3.16547483e-01
1.26764584e+00 3.83110315e-01 -1.55911922e+00 -2.56502390e-01
-6.24853313e-01 -7.26997972e-01 -2.47398987e-02 -3.73746216e-01
-1.26506793e+00 1.12478971e+00 5.78080773e-01 -2.72908062e-01
1.51258540e+00 -8.17870796e-02 5.40565550e-01 5.90502061e-02
4.03490394e-01 -8.68708909e-01 4.80177939e-01 1.51656002e-01
1.36399281e+00 -9.73433852e-01 -3.96019936e-01 -6.62962258e-01
-6.97501302e-01 1.42738640e+00 4.17658478e-01 2.64373094e-01
7.08190918e-01 2.59742320e-01 1.02013372e-01 2.43866786e-01
-6.23278081e-01 9.56509728e-03 4.69023019e-01 8.81253064e-01
2.52296895e-01 -2.49509424e-01 3.76236469e-01 6.82764649e-01
-5.46297193e-01 1.46922633e-01 1.94463268e-01 2.51262605e-01
-3.69252294e-01 -8.72561276e-01 -7.64270067e-01 -1.73645705e-01
-3.60772908e-01 6.29696995e-02 -2.35727280e-01 4.09222215e-01
3.96082580e-01 9.55657601e-01 3.24794874e-02 -5.06995022e-01
1.88899964e-01 3.34985673e-01 5.91618657e-01 -4.40026700e-01
-6.01047231e-03 5.32003403e-01 -2.93812245e-01 -5.43804646e-01
-4.72926676e-01 -6.98861003e-01 -7.39956021e-01 -3.10795516e-01
-3.13352227e-01 4.99357097e-02 4.32044238e-01 6.19678915e-01
3.50646645e-01 1.76497638e-01 9.52966750e-01 -1.25619209e+00
-4.28763956e-01 -9.77195561e-01 -5.46518147e-01 4.95721251e-01
4.61665511e-01 -8.87323916e-01 -3.22695047e-01 5.55822611e-01]
|
[13.231817245483398, -0.3607860803604126]
|
7731df50-09d7-4789-9d9f-2bb2a97c2b36
|
covid-19-misinformation-on-twitter
| null | null |
https://openreview.net/forum?id=aDCizGE1HR2
|
https://openreview.net/pdf?id=aDCizGE1HR2
|
COVID-19 Misinformation on Twitter: Multilingual Analysis
|
In the current scenario of the coronavirus disease pandemic (COVID-19), the Internet has become an important source of health information for users worldwide. During pandemic situations, myths, sensationalism, rumours and misinformation, generated intentionally or unintentionally, spread rapidly through social networks. Twitter is one of these popular social networks people use to share COVID-19related news, information, and thoughts that reflect their perception and opinion about the pandemic. Analysis of tweets for identifying misinformation can generate valuable insight to evaluate the quality and readability of online information about the COVID-19. This paper presents a multilingual COVID-19 related tweet analysis method, CMTA, that usesBERT, a deep learning model for multilingual tweet misinformation detection and classification.CMTA extracts features from multilingual textual data, which is then categorized into specific information classes. Classification is done by a Dense-CNN model trained on tweets manually annotated into information classes (i.e., ’false’, ’partly false’, ’misleading’). The paper assesses CMTA experimenting an analysis of multilingual tweets from February to June, showing the distribution type of information spread across different languages.
|
['Genoveva Vargas-Solar', 'Ambesh Shekhar', 'Mehrdad Farokhenajd', 'Raj Ratn Pranesh']
|
2021-01-06
| null | null | null | null |
['rumour-detection']
|
['natural-language-processing']
|
[-2.66763479e-01 1.94657207e-01 -2.26345018e-01 8.62420276e-02
-5.33584356e-01 -6.72188938e-01 1.11784852e+00 1.12955689e+00
-4.46775854e-01 7.70030499e-01 7.94168353e-01 -4.68040794e-01
5.54540336e-01 -9.30246711e-01 -5.34747541e-01 -3.66071343e-01
-2.11125612e-01 8.41651499e-01 -3.34262729e-01 -6.75532937e-01
3.26593548e-01 1.75191015e-01 -8.61162305e-01 6.28134429e-01
5.53175688e-01 7.10490167e-01 -1.38744891e-01 6.43015504e-01
-7.03752995e-01 1.17682326e+00 -1.13707805e+00 -2.45330721e-01
-4.57019061e-01 -2.37817213e-01 -9.40809548e-01 -4.15856987e-01
-1.60061985e-01 -2.15443969e-01 1.45807594e-01 1.09513557e+00
4.68400538e-01 -7.42180705e-01 7.02098250e-01 -1.12398136e+00
-5.20373940e-01 8.27517688e-01 -4.77891564e-01 8.22014928e-01
6.02700531e-01 -5.41484170e-02 2.85324574e-01 -6.76999032e-01
1.23906255e+00 1.46626163e+00 1.11448228e+00 3.22893038e-02
-6.74040496e-01 -7.85768449e-01 -4.32591230e-01 -1.13928914e-01
-1.32185769e+00 1.11314997e-01 5.47699988e-01 -1.37642634e+00
7.89441943e-01 2.98115909e-01 8.20563793e-01 1.54022491e+00
8.23440731e-01 4.59093899e-01 1.16312218e+00 2.64130682e-01
-2.69489378e-01 6.43267632e-01 5.72171435e-02 6.08668327e-01
3.19901884e-01 -4.69070151e-02 -2.42356956e-01 -6.35233998e-01
-1.88020945e-01 1.28813103e-01 -2.72508264e-01 7.77692556e-01
-1.37255609e+00 1.48944879e+00 6.09467983e-01 5.88217199e-01
-7.30690360e-01 -6.42286420e-01 9.24149215e-01 5.72987318e-01
1.26845193e+00 4.26365286e-01 -3.12760264e-01 1.18318908e-01
-7.06292927e-01 3.25578153e-01 9.75718617e-01 3.39854687e-01
6.28452301e-01 -2.27741629e-01 -1.27068684e-01 5.90519488e-01
3.24850649e-01 1.24132085e+00 5.80987453e-01 2.07997590e-01
4.65870351e-01 5.38209915e-01 1.89382643e-01 -2.18326020e+00
-1.15507817e+00 -6.50877416e-01 -1.10519266e+00 -5.69055855e-01
-1.56079680e-01 -7.68121719e-01 -4.08455342e-01 1.40819573e+00
3.79588038e-01 -1.62133634e-01 -5.56503013e-02 7.39861012e-01
1.38431668e+00 1.17667127e+00 1.05730832e-01 -3.69393975e-01
1.58146465e+00 -4.51590836e-01 -1.17273569e+00 9.92197692e-02
6.29111469e-01 -1.20953822e+00 3.31923634e-01 1.17951497e-01
-5.71795881e-01 2.89339405e-02 -7.88440824e-01 9.35626030e-02
-1.31557000e+00 -5.64766943e-01 -1.81153230e-03 4.87509072e-01
-9.55025733e-01 2.45964855e-01 -1.34815440e-01 -5.47330856e-01
5.15299499e-01 -3.36136550e-01 -1.44976825e-01 2.99133658e-01
-1.68301404e+00 1.08370507e+00 5.30595303e-01 -1.64538860e-01
-9.36296344e-01 -6.35019779e-01 -7.52716660e-01 -7.22139180e-01
-1.48163587e-01 -3.32339019e-01 9.95871186e-01 -8.36593151e-01
-7.18190908e-01 1.25398755e+00 7.20321909e-02 -5.69826245e-01
5.04752338e-01 -1.44901993e-02 -1.09274745e+00 -1.50632276e-03
5.55464149e-01 3.66050780e-01 7.51625299e-01 -9.76069212e-01
-4.04591173e-01 -2.66795099e-01 -2.89874315e-01 -3.97656649e-01
-4.87919301e-02 5.72125614e-01 5.11085570e-01 -9.12133574e-01
-5.38864076e-01 -7.71066010e-01 1.08412512e-01 -6.83052480e-01
-9.53280568e-01 -2.19139710e-01 1.13213146e+00 -1.05602670e+00
1.26238739e+00 -1.82298744e+00 -4.04708833e-01 1.41318962e-01
7.60895669e-01 4.60785091e-01 1.15179516e-01 1.09516692e+00
3.62971455e-01 6.11767530e-01 3.27390283e-02 1.42519385e-01
-1.95453227e-01 2.74368897e-02 -4.61187124e-01 6.82756186e-01
2.27092341e-01 8.20297003e-01 -1.40779245e+00 -2.43764743e-01
-1.72925502e-01 8.02457809e-01 -3.75706136e-01 1.51078627e-01
-4.98504758e-01 8.87086093e-01 -6.91953599e-01 2.74039596e-01
9.69306767e-01 -5.87184191e-01 1.74557585e-02 -3.37439060e-01
-7.01151729e-01 5.67808449e-01 -3.78303267e-02 6.30535543e-01
-4.23281610e-01 1.19686329e+00 -5.40540293e-02 -4.76295173e-01
6.53397560e-01 4.95725662e-01 4.12928045e-01 -6.02657557e-01
7.93362498e-01 5.73958941e-02 -5.25083184e-01 -8.72498930e-01
6.09655678e-01 -1.61321145e-02 -5.01727283e-01 7.85468817e-01
-4.94811922e-01 3.93635601e-01 -1.85296014e-01 3.85526150e-01
6.18089974e-01 -1.05518878e+00 6.20784402e-01 -4.11933988e-01
3.98175031e-01 5.80105186e-01 5.04938699e-03 5.67777574e-01
-1.61789894e-01 1.66299865e-01 3.56755376e-01 -8.18374276e-01
-8.19776535e-01 -7.57786572e-01 -2.70878017e-01 1.00443923e+00
-2.11029634e-01 -1.23480082e-01 -7.59360492e-01 -4.09668654e-01
-5.43508045e-02 6.32941246e-01 -9.24789786e-01 2.32942760e-01
-3.54089022e-01 -1.14039421e+00 8.34778070e-01 -3.97560865e-01
5.97361922e-01 -1.29341578e+00 -6.83578551e-01 3.78556162e-01
-9.68704283e-01 -8.79617929e-01 -4.20038015e-01 -3.39291424e-01
7.52349123e-02 -1.33930624e+00 -5.58789074e-01 -6.74676716e-01
4.68342811e-01 2.52474010e-01 1.13324034e+00 -1.75633039e-02
-1.93712607e-01 5.31226881e-02 -4.44084316e-01 -9.13914859e-01
-1.13167584e+00 -3.77950184e-02 -3.11063044e-02 3.38071957e-02
4.94534999e-01 9.78026316e-02 -4.60236490e-01 -2.13466585e-01
-1.10402298e+00 6.08192384e-03 1.87152445e-01 4.26191926e-01
-6.39405921e-02 -1.12012655e-01 6.90039814e-01 -1.35999894e+00
1.00933266e+00 -1.68423092e+00 -3.54077108e-02 -4.34430569e-01
-2.14180276e-01 -3.82531285e-01 4.29304004e-01 -2.01381117e-01
-6.95765972e-01 -1.03621185e+00 -3.75747412e-01 3.31024826e-01
-2.03045592e-01 1.25357282e+00 7.95986414e-01 2.87372291e-01
8.05865943e-01 1.51973814e-01 -4.39484511e-03 -4.69910145e-01
1.64953187e-01 1.43805420e+00 9.10345390e-02 6.16094649e-01
5.68589926e-01 7.42838323e-01 -6.88801348e-01 -1.28277254e+00
-1.25860274e+00 -7.03784406e-01 -5.97300828e-02 -4.74680245e-01
1.16720796e+00 -9.19622183e-01 -7.98016906e-01 7.65840948e-01
-1.70679605e+00 1.98733047e-01 5.06067753e-01 8.33246484e-02
2.37687409e-01 6.31621256e-02 -1.06485248e+00 -4.86724555e-01
-7.35676467e-01 -7.68058717e-01 6.36150539e-01 -2.93561757e-01
-7.16818929e-01 -1.44673336e+00 6.39513612e-01 4.68288630e-01
1.13254344e+00 1.02333963e+00 8.27830553e-01 -1.28834164e+00
2.09414825e-01 -2.83427238e-01 -5.43314934e-01 -2.53604800e-01
4.54360396e-01 -3.47985089e-01 -9.11983192e-01 -4.04967129e-01
4.11343016e-02 -5.75343728e-01 6.58176303e-01 1.27392724e-01
3.75151753e-01 -1.56533015e+00 -5.05925238e-01 -5.88479750e-02
1.24802506e+00 -1.89209096e-02 1.47230983e-01 4.50572997e-01
5.52975178e-01 5.41885436e-01 7.82718435e-02 9.49020684e-01
7.45485365e-01 2.74371773e-01 5.56564033e-01 -1.36832297e-01
2.23249644e-01 -3.76244843e-01 4.84247625e-01 1.28946078e+00
4.27405328e-01 -5.22549510e-01 -1.14492321e+00 5.08895516e-01
-1.13789856e+00 -1.14086318e+00 -3.22769463e-01 1.66200817e+00
9.60772753e-01 -1.28575027e-01 2.46427566e-01 -2.31402338e-01
9.13508236e-01 6.08296275e-01 -7.23339543e-02 -7.52358794e-01
-2.94042796e-01 -5.19402385e-01 4.96619225e-01 6.73077166e-01
-1.24504066e+00 5.65634847e-01 5.94487476e+00 4.52837616e-01
-1.69361246e+00 5.38127959e-01 5.64777493e-01 5.97015738e-01
-4.14616317e-01 -9.08664584e-01 -5.69647670e-01 7.43713856e-01
1.18884027e+00 -2.23976895e-01 1.78631529e-01 3.85511398e-01
7.06468403e-01 2.12033823e-01 -1.24740869e-01 6.54572606e-01
6.09908044e-01 -1.87194252e+00 3.70720625e-01 -9.51689109e-02
8.72330904e-01 9.58115280e-01 2.43750960e-01 -3.09073534e-02
2.51220554e-01 -8.10384750e-01 6.25455916e-01 4.28335488e-01
6.29443944e-01 -6.68335319e-01 1.27441955e+00 4.73801523e-01
-6.41430497e-01 2.99382448e-01 -2.67002404e-01 1.31337151e-01
3.90116483e-01 8.99483263e-01 -1.54751599e+00 1.26720667e-01
5.79320014e-01 1.02670085e+00 -4.58845317e-01 6.27144158e-01
2.27066740e-01 6.38771594e-01 -4.71313717e-03 -3.88792396e-01
5.84815502e-01 2.34318152e-01 1.06349254e+00 2.08564496e+00
1.76409170e-01 -3.41167569e-01 3.31182539e-01 7.32761502e-01
-8.55889171e-03 4.54644710e-01 -1.09754682e+00 -7.42821455e-01
2.84914285e-01 9.69311416e-01 -5.47698855e-01 -4.76738334e-01
3.32435146e-02 6.00232124e-01 1.33365929e-01 8.66553709e-02
-7.12642789e-01 -9.44438055e-02 2.06825167e-01 3.46961588e-01
-1.29063800e-01 2.27775708e-01 2.85228163e-01 -8.78912628e-01
-6.31651819e-01 -9.99317646e-01 5.28317094e-01 -5.12937307e-01
-1.57994163e+00 1.01232135e+00 -2.88571894e-01 -1.03210187e+00
-3.47612202e-01 -2.20195010e-01 -5.63724518e-01 4.63323981e-01
-1.44250119e+00 -1.02816749e+00 -2.08197862e-01 4.10617381e-01
4.66834813e-01 -1.87369123e-01 8.92548621e-01 4.29888219e-01
-1.78797737e-01 1.48627922e-01 7.28625432e-02 4.67001349e-01
4.18083966e-01 -6.97284818e-01 2.30213463e-01 -8.93737152e-02
-5.39812207e-01 3.39034230e-01 1.05852985e+00 -1.08126867e+00
-8.31294477e-01 -1.78822637e+00 1.75530803e+00 -3.90352935e-01
1.14373505e+00 -7.66874552e-02 -6.93029344e-01 7.03726709e-01
7.81905472e-01 -1.02558279e+00 1.08053970e+00 -2.63615400e-01
-5.62173307e-01 5.42836368e-01 -1.53252685e+00 4.27283853e-01
2.63453156e-01 -5.55135906e-01 -6.81061864e-01 1.16396153e+00
9.14639235e-01 -1.09616235e-01 -6.88924134e-01 -1.27957821e-01
2.87919223e-01 -7.95722008e-01 6.65432096e-01 -8.78685236e-01
5.41182101e-01 2.62052827e-02 2.08664939e-01 -1.71174777e+00
-1.59954205e-01 -6.11249745e-01 5.71891010e-01 8.25900972e-01
6.12106740e-01 -1.04097378e+00 -9.70906243e-02 -6.74302161e-01
3.47711369e-02 -6.63899407e-02 -6.61331236e-01 -1.29158854e-01
5.20644411e-02 -3.63584757e-01 4.84301567e-01 1.72360659e+00
3.46731156e-01 5.34621835e-01 -7.08347082e-01 1.35202095e-01
3.56365830e-01 -6.43005744e-02 4.65069294e-01 -1.16997135e+00
4.64017212e-01 -5.25719404e-01 -2.07529083e-01 -3.83190364e-01
-2.91695476e-01 -1.07884681e+00 -1.67078391e-01 -1.26821840e+00
2.83179909e-01 -2.28942752e-01 1.10125475e-01 7.91110918e-02
3.66898656e-01 5.45065045e-01 -4.71928194e-02 2.84514576e-01
-5.80917895e-01 -7.98021629e-02 1.08825767e+00 -5.29894292e-01
-1.52121363e-02 -1.69859827e-01 -4.14905757e-01 8.93106520e-01
1.01597357e+00 -6.89745188e-01 6.50615394e-02 -1.60891309e-01
1.21652067e+00 3.79973687e-02 2.70241767e-01 -6.03000522e-01
-1.13565579e-01 6.43338189e-02 -1.59127861e-02 -1.10729921e+00
-4.02572975e-02 -5.14105439e-01 3.17952931e-01 1.04109251e+00
-6.24008954e-01 4.28032398e-01 1.70970783e-01 6.25844359e-01
-3.69966209e-01 3.33697677e-01 6.87336922e-01 -1.29525334e-01
1.00710429e-02 2.38013506e-01 -1.41458273e+00 5.70216894e-01
5.25918186e-01 6.03998244e-01 -9.32404637e-01 -7.81197250e-01
-6.80178523e-01 7.15828389e-02 3.44222179e-03 6.32174611e-01
5.60911834e-01 -1.18052125e+00 -1.40920293e+00 1.13492094e-01
3.36828321e-01 -5.88045359e-01 2.25576490e-01 1.18630481e+00
-9.12953556e-01 8.23097646e-01 -1.49011865e-01 -5.05048037e-01
-8.51904094e-01 8.24076772e-01 9.13088918e-02 -2.62179285e-01
-4.27177757e-01 3.18083823e-01 -4.77493182e-02 -6.55529976e-01
-1.92655876e-01 -1.57516107e-01 -9.30203438e-01 7.26037920e-01
1.08935726e+00 4.71752673e-01 -1.46215662e-01 -1.43404651e+00
-3.10072631e-01 1.53381810e-01 -1.23235054e-01 3.14886510e-01
1.00546873e+00 -4.08832431e-01 -6.83384418e-01 7.11305022e-01
1.76321101e+00 2.02491492e-01 1.16646804e-01 -2.45320797e-01
8.20507854e-02 -7.06262290e-02 1.86473783e-02 -1.10854840e+00
-8.65935922e-01 6.97763324e-01 3.77711326e-01 1.02276003e+00
3.95328462e-01 2.08721831e-01 1.33895195e+00 2.50400096e-01
8.07404742e-02 -7.10331321e-01 -1.12583600e-01 8.29949737e-01
1.36777163e+00 -1.44896352e+00 -4.89080012e-01 -8.13335627e-02
-6.79939687e-01 9.23256636e-01 -3.08527410e-01 2.98717022e-01
1.33458614e+00 1.11820757e-01 6.81635320e-01 -8.17036390e-01
-4.95116264e-01 2.24536210e-01 1.89978287e-01 6.87716842e-01
4.71798062e-01 5.59477448e-01 -4.31415945e-01 2.27768436e-01
-4.91517425e-01 -1.56519040e-01 7.72669256e-01 5.19334733e-01
-7.53637671e-01 -1.41895175e-01 -5.13104677e-01 5.27351677e-01
-1.10386729e+00 -2.05668867e-01 -5.96928775e-01 4.93434638e-01
4.03375477e-01 1.27762413e+00 2.77101696e-01 -3.81621301e-01
-3.28683257e-01 -3.41236025e-01 -5.45221567e-01 -3.75805676e-01
-8.95917714e-01 -1.45264149e-01 5.32399654e-01 -1.93470374e-01
-5.41951835e-01 -3.79125118e-01 -1.08809614e+00 -8.26445460e-01
1.66164428e-01 3.74823630e-01 7.61221588e-01 1.16750562e+00
3.39603454e-01 1.62916943e-01 8.76563370e-01 -5.10207891e-01
7.87426904e-02 -1.03306282e+00 -8.69049430e-02 6.10511661e-01
1.05367684e+00 -1.75018862e-01 -4.63005245e-01 -2.10159481e-01]
|
[8.42109489440918, 9.80932331085205]
|
0e39f87f-a9ea-4d2d-b127-9f7fc546ea35
|
enhancing-low-light-videos-by-exploring-high
| null | null |
http://openaccess.thecvf.com/content_ICCV_2019/html/Wang_Enhancing_Low_Light_Videos_by_Exploring_High_Sensitivity_Camera_Noise_ICCV_2019_paper.html
|
http://openaccess.thecvf.com/content_ICCV_2019/papers/Wang_Enhancing_Low_Light_Videos_by_Exploring_High_Sensitivity_Camera_Noise_ICCV_2019_paper.pdf
|
Enhancing Low Light Videos by Exploring High Sensitivity Camera Noise
|
Enhancing low light videos, which consists of denoising and brightness adjustment, is an intriguing but knotty problem. Under low light condition, due to high sensitivity camera setting, commonly negligible noises become obvious and severely deteriorate the captured videos. To recover high quality videos, a mass of image/video denoising/enhancing algorithms are proposed, most of which follow a set of simple assumptions about the statistic characters of camera noise, e.g., independent and identically distributed(i.i.d.), white, additive, Gaussian, Poisson or mixture noises. However, the practical noise under high sensitivity setting in real captured videos is complex and inaccurate to model with these assumptions. In this paper, we explore the physical origins of the practical high sensitivity noise in digital cameras, model them mathematically, and propose to enhance the low light videos based on the noise model by using an LSTM-based neural network. Specifically, we generate the training data with the proposed noise model and train the network with the dark noisy video as input and clear-bright video as output. Extensive comparisons on both synthetic and real captured low light videos with the state-of-the-art methods are conducted to demonstrate the effectiveness of the proposed method.
|
[' Tao Yue', ' Xuemei Hu', ' Xiang Li', ' Cheng Yang', ' Xin Chen', 'Wei Wang']
|
2019-10-01
| null | null | null |
iccv-2019-10
|
['video-denoising']
|
['computer-vision']
|
[ 4.26052898e-01 -9.16882455e-01 4.43353146e-01 -5.52930161e-02
-2.64045566e-01 -3.80646557e-01 2.50866890e-01 -7.93305457e-01
-5.00466824e-01 7.56949365e-01 -1.88238584e-02 -2.64456570e-02
-3.83895934e-02 -6.37569845e-01 -9.42125559e-01 -1.29693425e+00
3.72515470e-01 -5.97620845e-01 9.26013365e-02 -1.21155172e-03
1.10526741e-01 9.18655563e-03 -1.73277736e+00 5.41024357e-02
7.49810159e-01 1.03944409e+00 5.16017497e-01 9.05580461e-01
-4.13600763e-04 1.02313066e+00 -7.56469131e-01 -2.94451475e-01
3.99587393e-01 -3.85707706e-01 1.29508451e-01 3.52179021e-01
3.03148031e-01 -8.74256313e-01 -8.83698165e-01 1.76987123e+00
6.26066744e-01 2.78966725e-01 3.39819610e-01 -1.12609160e+00
-9.99644339e-01 1.84602648e-01 -8.75325680e-01 4.81450856e-01
1.71322510e-01 5.20533025e-01 -6.38101995e-02 -7.37073064e-01
2.60714710e-01 1.23418701e+00 6.11982882e-01 6.95050597e-01
-7.45261192e-01 -8.01876605e-01 -1.01077836e-02 3.51192981e-01
-1.31961501e+00 -6.15366101e-01 8.26090634e-01 -1.60385504e-01
3.84840369e-01 3.89068313e-02 5.35954177e-01 1.42909741e+00
4.56733912e-01 3.84002686e-01 1.36002481e+00 -2.01405048e-01
-1.57629177e-02 -1.37961041e-02 -1.25045434e-01 3.40282589e-01
5.28830230e-01 1.97818622e-01 -1.90904230e-01 4.70594577e-02
8.35114002e-01 3.54265362e-01 -6.44354284e-01 2.92477727e-01
-8.97974849e-01 1.68312013e-01 4.16222215e-02 1.79863721e-01
-4.18683648e-01 2.79856384e-01 3.11698109e-01 1.12257861e-01
3.33656251e-01 -3.44392270e-01 -2.45265931e-01 -1.06103562e-01
-6.97651029e-01 -1.20341696e-01 4.44425493e-01 1.07961452e+00
6.26808584e-01 5.79779088e-01 -1.21790422e-02 8.26779246e-01
3.59074980e-01 8.90116274e-01 3.35971475e-01 -1.22277212e+00
4.00299937e-01 -2.73759186e-01 2.70753562e-01 -1.21111965e+00
-2.10605301e-02 -5.45002460e-01 -1.46258724e+00 1.33719653e-01
5.05419821e-02 -4.27502275e-01 -8.53112519e-01 1.73279154e+00
2.14056447e-01 6.51696265e-01 2.01083750e-01 1.10251653e+00
9.59546566e-01 1.12442279e+00 -1.78920012e-02 -9.64840710e-01
8.84042740e-01 -6.28563821e-01 -1.31789422e+00 1.03871882e-01
-5.78809604e-02 -1.07596087e+00 8.86084855e-01 8.09567869e-01
-1.16327000e+00 -8.77998054e-01 -7.93523431e-01 1.14231311e-01
4.55450304e-02 1.22430697e-01 2.40801275e-01 7.26662993e-01
-8.71058881e-01 3.92897874e-01 -5.14134705e-01 -1.03019930e-01
3.00950974e-01 -6.15558028e-03 -1.38443217e-01 -5.21393418e-01
-1.31564426e+00 4.38430369e-01 1.65665209e-01 8.26442540e-01
-1.14215040e+00 -3.95880073e-01 -5.74310422e-01 -2.34258726e-01
6.99514747e-01 -7.67104268e-01 9.88576829e-01 -1.03400910e+00
-1.54677641e+00 4.23386395e-01 -9.12415385e-02 1.80736724e-02
4.79287148e-01 -2.66997844e-01 -6.68185711e-01 1.88531622e-01
-1.34023145e-01 9.92629975e-02 1.10989964e+00 -1.60289955e+00
-5.48576534e-01 -9.83710587e-02 3.38640958e-02 1.60542101e-01
-1.89510718e-01 2.01887459e-01 -4.30741608e-01 -6.21311665e-01
-1.60474658e-01 -6.39001191e-01 -3.32564078e-02 -9.24113095e-02
-3.09727192e-01 2.63137609e-01 1.05902004e+00 -8.11913192e-01
1.10520315e+00 -2.39041400e+00 -3.25227082e-01 -1.55146122e-01
1.69915691e-01 3.86459887e-01 -1.41860405e-02 1.31131575e-01
4.58790995e-02 5.80393188e-02 -1.14031248e-01 -3.98801193e-02
-3.17934960e-01 4.66227144e-01 -1.73593745e-01 7.08842099e-01
-7.65867829e-02 3.69900703e-01 -8.47417355e-01 -4.22610402e-01
5.82701147e-01 7.40769625e-01 -1.30999282e-01 3.98487329e-01
1.08171143e-01 3.60161334e-01 -3.46793801e-01 6.82432115e-01
1.23720324e+00 1.68920651e-01 -3.07247996e-01 -6.99242175e-01
-2.71281153e-01 -5.83241165e-01 -1.58063424e+00 1.13649797e+00
-3.73481899e-01 6.86123013e-01 6.06601000e-01 -8.08199942e-01
8.55491757e-01 2.11538851e-01 2.18832955e-01 -6.50269687e-01
5.43546677e-01 8.28892663e-02 -4.36940007e-02 -1.40216064e+00
3.68160695e-01 -4.67088193e-01 5.92174530e-01 4.47807871e-02
-1.39414623e-01 7.21793249e-02 1.05162129e-01 -2.32026749e-03
9.05792654e-01 2.65938282e-01 -3.78762111e-02 1.62278071e-01
5.82629859e-01 -6.40798390e-01 9.61869240e-01 7.93471634e-01
-4.96011525e-01 7.37269700e-01 1.04661763e-01 -1.62899345e-01
-1.04718864e+00 -8.58629763e-01 9.96156558e-02 5.96999228e-01
6.57929003e-01 1.59465373e-01 -8.66960645e-01 -3.98761779e-02
-5.40061235e-01 4.92126971e-01 -2.31849864e-01 -2.98852116e-01
-4.94396031e-01 -1.15087664e+00 3.61499399e-01 1.13420181e-01
1.02859843e+00 -7.31081843e-01 -2.08676651e-01 7.01290742e-02
-3.53882581e-01 -1.49335814e+00 -2.76465952e-01 -1.03880592e-01
-4.87785131e-01 -1.11791837e+00 -6.12308860e-01 -7.39189804e-01
4.74930018e-01 8.93148601e-01 8.01709354e-01 1.79611772e-01
-1.83919787e-01 4.95588094e-01 -2.90773243e-01 -2.84230709e-01
-3.66484731e-01 -8.61920297e-01 3.17685544e-01 4.57427591e-01
1.94269806e-01 -4.98361498e-01 -9.18032527e-01 1.99946940e-01
-1.48453867e+00 -9.17462334e-02 6.47946835e-01 7.55724609e-01
4.68980104e-01 9.00330067e-01 1.75290480e-01 -4.09861982e-01
6.18436396e-01 -4.27334964e-01 -5.08072078e-01 8.75801891e-02
-9.63245183e-02 -4.93610114e-01 1.02205360e+00 -7.85410583e-01
-1.56325161e+00 -2.77923822e-01 -7.12620765e-02 -9.11847174e-01
-5.07165551e-01 1.48653239e-01 -5.62865913e-01 -2.61532754e-01
2.75292397e-01 5.68274796e-01 -1.46284953e-01 -3.76737058e-01
1.65731534e-01 8.58925462e-01 7.70110846e-01 -3.58103007e-01
9.87851560e-01 5.88044107e-01 1.13480188e-01 -1.13104975e+00
-6.61530733e-01 -8.38392451e-02 -1.02984026e-01 -6.80849195e-01
8.24465394e-01 -9.84749913e-01 -8.63730431e-01 1.16209388e+00
-1.25120056e+00 -1.57634288e-01 9.49622020e-02 7.76834607e-01
-2.45529845e-01 7.03101337e-01 -1.02432990e+00 -1.07137132e+00
-1.37471035e-01 -1.34698784e+00 7.70495653e-01 6.56509817e-01
6.85086489e-01 -7.83803940e-01 -3.13088268e-01 2.43187755e-01
5.15712619e-01 1.78818449e-01 7.04145312e-01 2.94466436e-01
-8.42112958e-01 3.22254486e-02 -4.69438165e-01 1.01735258e+00
3.04743201e-01 4.54500169e-01 -9.57795441e-01 -1.23145282e-01
8.29746127e-01 -6.41631782e-02 7.50973880e-01 8.54877830e-01
1.38768530e+00 -3.29977423e-01 1.11660115e-01 9.81512070e-01
1.84744227e+00 4.43151534e-01 9.01295424e-01 2.61020899e-01
6.87312782e-01 3.16569686e-01 4.75537241e-01 6.63094163e-01
-7.28958100e-03 5.25692962e-02 7.91173935e-01 -2.39888981e-01
-7.89686665e-02 3.41710374e-02 4.89275932e-01 1.15808141e+00
-2.39234522e-01 -6.83337748e-01 -1.78182214e-01 2.60335028e-01
-1.40998209e+00 -1.24619865e+00 -6.60374880e-01 1.99629569e+00
7.22295702e-01 6.61537349e-02 -4.29329753e-01 9.00294632e-02
1.22621930e+00 3.10874969e-01 -4.62982655e-01 -1.96955152e-04
-7.00100422e-01 -1.89355910e-01 7.28766263e-01 2.18080133e-01
-9.93048072e-01 4.76302981e-01 5.68421030e+00 1.01513851e+00
-1.20876527e+00 1.51667967e-01 7.97385514e-01 -1.53090298e-01
-1.45577192e-01 -1.65589556e-01 -5.68202853e-01 1.02553642e+00
5.19531906e-01 5.86103797e-02 7.46758223e-01 3.45708281e-01
9.86160696e-01 -3.08463484e-01 -4.34276283e-01 1.46345305e+00
2.81998813e-01 -7.76927888e-01 2.69040346e-01 -1.54146999e-01
8.24723244e-01 -2.40834236e-01 2.57670581e-01 1.41800493e-02
-9.66949314e-02 -7.06194222e-01 5.54070711e-01 1.07510996e+00
7.03084409e-01 -5.09236991e-01 9.61482644e-01 4.01784897e-01
-7.95956731e-01 -1.93702430e-01 -7.67477453e-01 -1.63477391e-01
3.91687989e-01 8.55422795e-01 2.69875020e-01 4.99197066e-01
1.00785244e+00 9.20487761e-01 -2.86585659e-01 1.11575949e+00
-6.67648911e-02 8.04391146e-01 -2.83392787e-01 1.68094650e-01
1.86613113e-01 -6.75326169e-01 6.03153527e-01 1.10770941e+00
6.53564870e-01 4.61139590e-01 -2.11691521e-02 7.71685660e-01
-7.36384243e-02 -2.64953792e-01 -6.36099517e-01 2.79651403e-01
1.50466353e-01 1.43762791e+00 -3.76835287e-01 -3.65198970e-01
-7.10933745e-01 7.12834179e-01 -5.98677576e-01 9.58788574e-01
-1.07374394e+00 -4.61465418e-01 4.88786608e-01 -2.08362658e-02
3.12281922e-02 -2.79521227e-01 -6.00311756e-02 -1.26037967e+00
2.32186452e-01 -9.48790610e-01 -6.93622530e-02 -1.52664793e+00
-1.45691133e+00 4.29134071e-01 -2.21602768e-01 -1.37449157e+00
4.25172091e-01 -6.12513423e-01 -7.70293832e-01 8.08338046e-01
-1.68774068e+00 -8.31486404e-01 -8.25165153e-01 7.81970501e-01
6.43357277e-01 -4.52045985e-02 -1.26775563e-01 7.81807065e-01
-9.92745638e-01 7.42817968e-02 6.63458467e-01 -3.38493325e-02
7.68863201e-01 -6.75570488e-01 -2.42912278e-01 1.25235510e+00
-5.64056039e-01 4.67623293e-01 1.01064670e+00 -6.66814327e-01
-1.71270871e+00 -1.19057035e+00 6.11104891e-02 1.75052434e-01
6.65665507e-01 -2.03150719e-01 -1.06014681e+00 3.67490977e-01
5.86147070e-01 2.04190597e-01 1.48010835e-01 -1.01353705e+00
2.12082669e-01 -4.26732928e-01 -1.13991797e+00 5.41780233e-01
1.00502026e+00 -2.32363299e-01 -3.49107176e-01 2.34009594e-01
8.21529448e-01 -2.25634396e-01 -6.60350144e-01 4.97933418e-01
3.16616148e-01 -1.09078920e+00 9.32926476e-01 -1.41737923e-01
4.95314002e-01 -5.70752680e-01 -3.74672651e-01 -1.18557429e+00
-3.12831730e-01 -8.05758417e-01 1.18207201e-01 1.46258509e+00
-2.75207818e-01 -5.08851171e-01 4.35829371e-01 3.47292215e-01
-1.22569427e-01 -4.01513994e-01 -5.56265712e-01 -6.68330312e-01
-4.46598470e-01 -4.63551134e-01 2.21958131e-01 7.85763741e-01
-8.44900846e-01 7.68563002e-02 -8.29995275e-01 4.38910484e-01
1.05815005e+00 -4.94224966e-01 7.15604961e-01 -7.24317551e-01
-8.61843228e-02 -6.87356479e-03 -1.97508723e-01 -1.10421383e+00
4.22264747e-02 -1.71509366e-02 3.28497678e-01 -1.48313522e+00
3.14863235e-01 3.72723606e-03 -2.77018398e-01 -2.88726330e-01
-3.65602016e-01 3.72749597e-01 -1.60012972e-02 1.94948778e-01
-5.37195086e-01 7.67190099e-01 1.31697202e+00 -1.27808005e-01
8.20209607e-02 -8.12190324e-02 -3.49028856e-01 8.90361547e-01
6.02076173e-01 -3.14125597e-01 -2.76973426e-01 -8.05693686e-01
3.07630092e-01 4.99553122e-02 5.58586657e-01 -9.80141401e-01
2.57566482e-01 -4.29013669e-01 5.20044684e-01 -5.01214564e-01
4.83903468e-01 -1.03677523e+00 1.82250917e-01 3.21537673e-01
-3.07718255e-02 1.30348904e-02 2.09904998e-03 8.42941344e-01
-3.40676695e-01 -4.26622033e-01 9.87381816e-01 -4.44572121e-01
-6.70449436e-01 3.31434071e-01 -5.51320434e-01 1.26754478e-01
8.99004281e-01 -4.43917245e-01 -6.95748389e-01 -6.67430162e-01
-4.16009098e-01 -1.63861498e-01 3.30115169e-01 1.07428163e-01
8.19001019e-01 -1.23953116e+00 -6.69380248e-01 2.25559741e-01
-3.70960027e-01 -1.57324940e-01 8.90419245e-01 9.46100354e-01
-5.81637919e-01 -2.14678615e-01 -9.87417400e-02 -5.91103673e-01
-1.24627841e+00 8.10253680e-01 4.86576051e-01 4.53497738e-01
-3.91504824e-01 4.57312942e-01 3.72609526e-01 3.23891006e-02
1.63414285e-01 -3.58498901e-01 -1.67132393e-01 -4.34609264e-01
5.63949406e-01 7.21402884e-01 -2.62826949e-01 -6.35740817e-01
8.97797048e-02 8.82185817e-01 3.88166159e-01 2.74926692e-01
1.32097518e+00 -6.89101577e-01 -3.14811409e-01 3.58654320e-01
1.15321565e+00 -8.74196552e-03 -1.32020211e+00 -2.51462817e-01
-8.05357099e-01 -8.44953239e-01 2.72021949e-01 -3.75728816e-01
-1.37818658e+00 8.96420956e-01 8.16955268e-01 4.46151733e-01
1.47616529e+00 -5.27183652e-01 1.05214453e+00 2.37847596e-01
1.64071247e-01 -1.25077403e+00 -8.50712997e-04 2.26639286e-01
4.52973783e-01 -1.16721988e+00 -1.43580317e-01 -4.00919020e-01
-3.32902312e-01 1.15918040e+00 7.85896957e-01 -1.28116876e-01
5.39120018e-01 6.05511487e-01 2.37328172e-01 2.49515191e-01
-6.67254984e-01 -1.06285401e-01 -4.67605174e-01 7.43711114e-01
3.80372815e-02 -4.88872617e-01 -1.74848750e-01 4.13827449e-01
3.37047517e-01 3.69196564e-01 1.16725588e+00 5.45871794e-01
-5.49348176e-01 -3.58936667e-01 -9.03023183e-01 4.37434614e-01
-8.65235507e-01 -1.53959930e-01 2.62952834e-01 5.76291203e-01
5.47142982e-01 1.40576780e+00 -7.29636401e-02 -4.01961565e-01
3.85989755e-01 -4.72646415e-01 3.04277211e-01 6.48952648e-02
-9.83868092e-02 4.29273397e-01 -3.90453428e-01 -2.68430561e-01
-8.04672539e-01 -3.80059689e-01 -7.25050926e-01 -5.16278088e-01
-3.96998584e-01 -1.45116329e-01 6.20519042e-01 9.44269836e-01
6.69837892e-02 6.14079833e-01 8.31802964e-01 -8.25829983e-01
-5.41146278e-01 -1.01241732e+00 -8.77326131e-01 6.21449232e-01
4.67152238e-01 -6.13112450e-01 -8.94253254e-01 3.70113283e-01]
|
[11.148478507995605, -2.3816940784454346]
|
45679c5d-0d32-449c-a623-9f4c2abead7e
|
moments-in-time-dataset-one-million-videos
|
1801.0315
| null |
http://arxiv.org/abs/1801.03150v3
|
http://arxiv.org/pdf/1801.03150v3.pdf
|
Moments in Time Dataset: one million videos for event understanding
|
We present the Moments in Time Dataset, a large-scale human-annotated
collection of one million short videos corresponding to dynamic events
unfolding within three seconds. Modeling the spatial-audio-temporal dynamics
even for actions occurring in 3 second videos poses many challenges: meaningful
events do not include only people, but also objects, animals, and natural
phenomena; visual and auditory events can be symmetrical in time ("opening" is
"closing" in reverse), and either transient or sustained. We describe the
annotation process of our dataset (each video is tagged with one action or
activity label among 339 different classes), analyze its scale and diversity in
comparison to other large-scale video datasets for action recognition, and
report results of several baseline models addressing separately, and jointly,
three modalities: spatial, temporal and auditory. The Moments in Time dataset,
designed to have a large coverage and diversity of events in both visual and
auditory modalities, can serve as a new challenge to develop models that scale
to the level of complexity and abstract reasoning that a human processes on a
daily basis.
|
['Dan Gutfruend', 'Sarah Adel Bargal', 'Lisa Brown', 'Kandan Ramakrishnan', 'Aude Oliva', 'Tom Yan', 'Quanfu Fan', 'Carl Vondrick', 'Bolei Zhou', 'Mathew Monfort', 'Alex Andonian']
|
2018-01-09
| null | null | null | null |
['multimodal-activity-recognition']
|
['computer-vision']
|
[ 2.28952363e-01 -4.12548453e-01 -9.98173207e-02 -1.79605797e-01
-5.71611345e-01 -9.30930674e-01 9.25927222e-01 2.15093404e-01
-4.90924329e-01 5.11118054e-01 9.06936407e-01 2.49999523e-01
-4.71048942e-03 -4.51273054e-01 -5.57284713e-01 -4.79137510e-01
-4.17331040e-01 2.65135914e-02 4.88892525e-01 6.07610792e-02
9.23057552e-03 6.11092031e-01 -1.95716548e+00 8.90383363e-01
4.61688731e-03 1.08266604e+00 -1.03452154e-01 9.42562759e-01
1.93870991e-01 1.72092688e+00 -5.90042412e-01 -9.97878388e-02
2.40087640e-02 -4.29762304e-01 -9.42178309e-01 1.94366306e-01
6.37785435e-01 -4.16580796e-01 -7.43766248e-01 6.50623739e-01
4.86809611e-01 5.07258296e-01 5.59933901e-01 -1.60027277e+00
-4.47503895e-01 4.47226405e-01 -3.55332673e-01 7.59555280e-01
1.06844509e+00 3.00751597e-01 9.28892910e-01 -6.58727586e-01
1.04520357e+00 1.35401130e+00 6.63177371e-01 3.64576072e-01
-8.42071533e-01 -4.06102777e-01 2.65958875e-01 7.48969197e-01
-1.06832123e+00 -5.01324236e-01 5.63340425e-01 -8.87910128e-01
1.34134102e+00 3.38703066e-01 1.08693039e+00 1.80589414e+00
-3.63583677e-02 1.15116417e+00 6.48444355e-01 -8.17121789e-02
1.49462059e-01 -5.56168854e-01 -9.78787169e-02 2.78180838e-01
-4.42914069e-01 -1.71550736e-01 -9.93478060e-01 -4.21556970e-03
6.57460153e-01 2.24825919e-01 -2.36675560e-01 -1.10803351e-01
-1.99356031e+00 3.68448526e-01 -1.31252766e-01 2.47129142e-01
-4.97555286e-01 2.94383526e-01 8.23222756e-01 2.09036037e-01
2.49003485e-01 1.93134636e-01 -5.36928833e-01 -1.07418036e+00
-8.51262927e-01 4.10256863e-01 5.93494117e-01 8.10263813e-01
2.58950651e-01 4.73951101e-02 -5.31334043e-01 6.80943012e-01
-2.28729904e-01 4.65142906e-01 5.06447375e-01 -1.44774461e+00
4.05518293e-01 5.06889641e-01 2.87127614e-01 -1.01958525e+00
-6.49231791e-01 1.18404381e-01 -5.11431336e-01 -2.20713794e-01
6.52794003e-01 -3.87932286e-02 -7.17172980e-01 1.86704743e+00
4.94139254e-01 4.83878553e-01 -3.59727234e-01 6.87230945e-01
1.16228867e+00 8.21312070e-01 3.87694240e-01 -4.57602352e-01
1.67248380e+00 -8.69705677e-01 -7.98149407e-01 -8.00438821e-02
4.87222850e-01 -5.64272225e-01 1.07102001e+00 4.61374432e-01
-1.19233084e+00 -6.46375477e-01 -4.66820866e-01 -2.29807839e-01
-5.19647241e-01 1.93065312e-02 5.38291693e-01 1.24184042e-01
-8.36087525e-01 3.75969172e-01 -8.41002405e-01 -7.47962475e-01
2.53423154e-01 -3.10976088e-01 -6.88708723e-01 1.35300964e-01
-1.12680519e+00 7.35115707e-01 3.37661058e-01 -3.21167141e-01
-1.04638445e+00 -6.48162544e-01 -1.01009583e+00 -1.19989872e-01
4.77708399e-01 -2.82800913e-01 1.49997783e+00 -9.17082310e-01
-9.55647707e-01 9.16647315e-01 -3.75345588e-01 -4.16699320e-01
5.73205054e-01 -2.81167656e-01 -7.79162705e-01 6.67009830e-01
3.02317828e-01 8.23741436e-01 5.89111924e-01 -5.37220538e-01
-9.03572142e-01 -1.82259485e-01 2.11935550e-01 2.00859934e-01
-3.92182678e-01 5.76873481e-01 -3.76816005e-01 -9.27090943e-01
-1.43031567e-01 -7.72522449e-01 2.19816014e-01 3.29168402e-02
-5.25697246e-02 -1.83042184e-01 7.90109336e-01 -7.42943108e-01
1.47794223e+00 -2.51713920e+00 2.26383045e-01 -4.67060566e-01
-7.02236081e-03 -3.98385704e-01 -1.21218339e-01 7.69436538e-01
-3.33273530e-01 -7.00721145e-02 2.35631064e-01 -2.63981164e-01
7.15277195e-02 1.59902439e-01 -4.80803341e-01 4.77174997e-01
-6.56289905e-02 8.37069511e-01 -1.29814696e+00 -7.54878819e-01
2.58432925e-01 4.36162025e-01 -4.40391064e-01 2.68486775e-02
-6.22395752e-03 5.46503186e-01 -1.75114021e-01 8.86472821e-01
-8.40320885e-02 -3.92504781e-01 -1.37525186e-01 -3.88707697e-01
-3.10977817e-01 9.78549868e-02 -1.26876640e+00 1.54525971e+00
7.42457584e-02 1.12753534e+00 -3.16978782e-01 -6.30656660e-01
8.30717757e-02 7.68588185e-01 1.13163400e+00 -7.11931169e-01
-1.09023280e-01 -2.97903836e-01 -3.08708161e-01 -9.15830731e-01
5.79630792e-01 1.63649336e-01 -3.64820749e-01 4.76289779e-01
2.71966070e-01 2.04545900e-01 8.20362628e-01 3.70869994e-01
1.50919950e+00 2.47982457e-01 4.45028961e-01 3.05509835e-01
1.41922921e-01 1.07367411e-02 6.11666381e-01 6.95362628e-01
-7.11041093e-01 7.28513837e-01 6.70163870e-01 -6.79409266e-01
-9.10493374e-01 -1.15832460e+00 1.11995786e-01 1.55360401e+00
-9.09026265e-02 -6.76179409e-01 -4.58133250e-01 -3.05368334e-01
-1.40520796e-01 3.91851276e-01 -7.53798246e-01 -3.96821238e-02
-5.58629215e-01 -2.90951729e-01 7.97220886e-01 9.57535505e-01
4.04130101e-01 -1.53398728e+00 -9.48540747e-01 2.35345110e-01
-9.06682849e-01 -1.51317418e+00 -5.52135050e-01 -6.90447614e-02
-5.09233654e-01 -1.31660342e+00 -5.41017354e-01 -3.99262875e-01
1.30258977e-01 2.27570266e-01 1.23430240e+00 -5.45171559e-01
-5.81592858e-01 1.08606923e+00 -7.02020764e-01 -3.94674122e-01
2.88972282e-03 -6.42928004e-01 1.93939924e-01 2.55031914e-01
3.95557016e-01 -5.56418061e-01 -6.51029229e-01 4.72055346e-01
-8.85773778e-01 -8.03783834e-02 -4.99755144e-02 3.35496426e-01
4.38371807e-01 2.62558851e-02 2.43749857e-01 -1.16912007e-01
5.04270531e-02 -7.24991202e-01 4.28044386e-02 1.90679938e-01
4.60646331e-01 -7.19384611e-01 2.63448298e-01 -9.44924057e-01
-8.65404725e-01 2.82964975e-01 2.07755834e-01 -5.25182962e-01
-4.98005062e-01 4.77292061e-01 1.90427005e-01 4.28003818e-01
5.96188188e-01 3.45822304e-01 -4.05924082e-01 -3.17818075e-01
4.44588542e-01 2.60707289e-01 8.36909890e-01 -4.81363326e-01
3.01725060e-01 9.08004642e-01 -1.05688177e-01 -1.00884950e+00
-7.98774302e-01 -7.27984548e-01 -8.97118211e-01 -8.64581943e-01
1.15870082e+00 -1.25828564e+00 -9.43305373e-01 6.80810750e-01
-1.01625550e+00 -4.35448289e-01 -6.90515697e-01 7.12388396e-01
-7.47315764e-01 4.21081245e-01 -8.71675968e-01 -6.67365372e-01
3.67612392e-01 -5.76456726e-01 1.20768321e+00 -2.27809530e-02
-7.87545383e-01 -8.18374515e-01 2.80458294e-02 4.17755187e-01
1.25112105e-03 7.25339234e-01 4.23937827e-01 -5.30015886e-01
-3.08907151e-01 -2.36617133e-01 -7.42500797e-02 -2.70647742e-02
1.37782872e-01 4.37700212e-01 -8.13705921e-01 -1.40747160e-01
-8.04649591e-02 -6.97471678e-01 9.24473703e-01 5.46396434e-01
1.19170761e+00 -2.26831898e-01 -1.58676967e-01 2.09563881e-01
6.98083460e-01 2.54491270e-01 6.06572688e-01 2.75977522e-01
6.14452720e-01 5.34704745e-01 6.67369962e-01 9.66982424e-01
5.79763472e-01 4.91204083e-01 3.43379557e-01 3.40054423e-01
-1.11009695e-01 -1.80174097e-01 7.34584808e-01 5.39359927e-01
-5.50793588e-01 -3.58011186e-01 -8.96171510e-01 8.72338831e-01
-1.99274719e+00 -2.06068873e+00 -1.65978298e-01 1.93842208e+00
7.47193754e-01 -6.94813356e-02 6.17354274e-01 3.31617028e-01
6.08167470e-01 5.81609786e-01 -3.43305647e-01 4.61467505e-02
-3.23050916e-01 -3.29179615e-01 9.12574772e-03 -8.63934159e-02
-1.45855832e+00 5.63378334e-01 7.68588257e+00 6.38035476e-01
-1.14087820e+00 1.35316908e-01 1.52147591e-01 -8.32158387e-01
7.06335381e-02 -2.79920906e-01 -3.83074373e-01 5.60300112e-01
1.12220061e+00 2.32011303e-02 4.18300927e-01 5.99275172e-01
5.27382553e-01 -2.20755711e-01 -1.45237505e+00 1.25191748e+00
2.32650325e-01 -1.22359431e+00 6.27902001e-02 -2.06832439e-01
6.11431599e-01 5.36012724e-02 -1.41516864e-01 4.14124995e-01
7.80835599e-02 -9.30671453e-01 1.43495345e+00 6.63196385e-01
7.93119609e-01 -2.35632092e-01 9.94186029e-02 4.04040009e-01
-1.73303151e+00 -3.86676341e-01 2.61360615e-01 -3.96807313e-01
5.67201257e-01 2.43145794e-01 -6.69909641e-02 1.65479630e-01
1.22877169e+00 1.33816218e+00 -6.34417117e-01 9.18063462e-01
3.91503461e-02 5.65051734e-01 -3.31857830e-01 1.76160336e-01
1.57294661e-01 2.82995582e-01 6.41883254e-01 1.55866003e+00
5.02540886e-01 3.42911273e-01 3.45082283e-01 2.68344432e-01
1.84460104e-01 -2.09288225e-01 -6.67885482e-01 -5.20612836e-01
5.60891032e-01 9.49387372e-01 -9.88058627e-01 -4.63287652e-01
-6.47974432e-01 1.00234759e+00 1.46599021e-02 5.36564708e-01
-1.25759876e+00 -1.38446063e-01 5.99246383e-01 1.56043828e-01
3.80076021e-01 -3.34831655e-01 3.82046968e-01 -1.41187179e+00
1.52764171e-01 -8.50924194e-01 8.84031773e-01 -1.38100302e+00
-1.12800455e+00 2.11204916e-01 3.78939480e-01 -1.54001677e+00
-3.37728411e-01 -3.79169017e-01 -3.16590637e-01 8.98088440e-02
-9.31266844e-01 -1.14916801e+00 -6.47272825e-01 1.08050478e+00
8.40875030e-01 5.55283986e-02 5.66533566e-01 6.36282206e-01
-3.02512735e-01 2.62075718e-02 -2.80893356e-01 3.75840604e-01
8.43059778e-01 -1.07637966e+00 1.53314754e-01 6.32487595e-01
3.71236056e-01 3.02292436e-01 5.73754728e-01 -6.37413323e-01
-1.29653084e+00 -1.04681563e+00 1.00176954e+00 -9.59691584e-01
9.17135417e-01 -1.82955965e-01 -5.48874557e-01 1.19353509e+00
3.90639566e-02 2.35388801e-01 7.73588359e-01 -1.42140258e-02
-5.83955586e-01 1.87166736e-01 -8.61957371e-01 6.28873169e-01
1.41241109e+00 -9.86803234e-01 -7.83643007e-01 3.92805755e-01
6.02743685e-01 -4.69321668e-01 -8.37251782e-01 3.01138073e-01
8.32540751e-01 -1.09229219e+00 1.24158609e+00 -9.14125681e-01
6.04546010e-01 -2.79942960e-01 -2.44663119e-01 -8.88395190e-01
-4.50652778e-01 -6.61011755e-01 -5.28169394e-01 1.11429060e+00
-8.00134689e-02 -1.60204038e-01 3.64155203e-01 3.01262021e-01
7.21700210e-03 -2.83649564e-01 -1.12703443e+00 -9.04025555e-01
-5.15714824e-01 -9.58572268e-01 3.75194639e-01 1.08450186e+00
2.88849264e-01 -8.50432087e-03 -5.37232280e-01 -9.90959704e-02
2.25574061e-01 -1.20355859e-01 5.44278800e-01 -1.25528741e+00
-5.11540920e-02 -4.61314172e-01 -7.43541837e-01 -9.47552264e-01
-1.15579860e-02 -3.62406820e-01 -1.33471414e-01 -1.48407269e+00
3.95694792e-01 5.46569705e-01 -2.88256168e-01 6.29215300e-01
2.18424186e-01 3.42023522e-01 1.44846126e-01 2.73483187e-01
-1.08692801e+00 3.40054035e-01 9.92035210e-01 -1.50996253e-01
1.21959008e-01 -3.71145189e-01 -1.06354117e-01 1.08462155e+00
1.07546240e-01 -1.63987175e-01 -3.79131705e-01 -3.12228292e-01
5.28739214e-01 3.15770507e-01 7.68020570e-01 -1.27111721e+00
2.46230394e-01 -6.21752679e-01 4.34150666e-01 -6.50882125e-01
7.19071269e-01 -7.77556360e-01 3.73725533e-01 2.05417842e-01
-5.10831892e-01 1.69391558e-01 1.19501509e-01 8.34340274e-01
-5.11194348e-01 3.34236264e-01 5.25625110e-01 -3.36252570e-01
-1.24206734e+00 2.60419577e-01 -7.97057748e-01 3.34597945e-01
1.26548302e+00 -4.92559373e-01 -3.82631898e-01 -8.68745446e-01
-1.23819697e+00 1.60786301e-01 2.08273694e-01 7.33741581e-01
3.64696413e-01 -1.63430798e+00 -5.27592897e-01 -2.24236891e-01
4.52777982e-01 -4.96227205e-01 7.27506399e-01 9.51811671e-01
-3.89548182e-01 2.73126066e-01 -4.11770344e-01 -7.07489729e-01
-1.34305751e+00 6.44046128e-01 -4.55048829e-02 5.34696970e-03
-7.11449325e-01 7.67906368e-01 -1.79890320e-02 5.99237867e-02
6.23846591e-01 -5.57031214e-01 -5.64911902e-01 8.06024134e-01
8.48645627e-01 6.89592540e-01 -4.21093374e-01 -1.04735458e+00
-4.64601457e-01 4.65111971e-01 5.87514699e-01 -1.35163248e-01
1.27870226e+00 -2.88638145e-01 -2.47991588e-02 1.12420952e+00
1.03505564e+00 -1.23688690e-01 -1.29071474e+00 -1.90471813e-01
-1.91375524e-01 -3.87380958e-01 -4.64341074e-01 -6.50773287e-01
-7.75567234e-01 8.44501019e-01 3.50129187e-01 4.98741478e-01
1.16308522e+00 2.02058211e-01 9.01186585e-01 2.93906033e-01
4.35493648e-01 -1.24697053e+00 6.36470437e-01 7.48669565e-01
1.03836179e+00 -1.14228022e+00 -4.34059277e-02 -1.49963975e-01
-9.72038329e-01 1.06327224e+00 5.48854828e-01 2.88596541e-01
6.45770550e-01 1.19875655e-01 -2.02195212e-01 -2.55788356e-01
-9.91152763e-01 -3.41957122e-01 3.27597737e-01 4.29216474e-01
4.43699926e-01 -8.96895379e-02 6.65428266e-02 5.09671152e-01
1.53751597e-02 1.28768042e-01 5.74059069e-01 1.22273564e+00
-2.92634159e-01 -3.74577254e-01 -2.60209531e-01 2.02642649e-01
-4.87001598e-01 2.77684808e-01 -3.63024503e-01 7.52700150e-01
2.14455113e-01 1.00715518e+00 3.97533923e-01 -2.89786905e-01
4.22236741e-01 2.18908027e-01 4.90617305e-01 -4.76806045e-01
-4.46270823e-01 9.71995220e-02 3.09373677e-01 -1.02838898e+00
-9.10367429e-01 -1.11011887e+00 -1.25221503e+00 -2.59151220e-01
4.65282172e-01 -3.73768479e-01 1.58454359e-01 9.54516053e-01
2.85130888e-01 5.64974010e-01 1.87960356e-01 -1.10877979e+00
-2.56864503e-02 -9.43229973e-01 -6.50599658e-01 9.71800327e-01
4.70952421e-01 -7.28152156e-01 -5.89488626e-01 8.06807518e-01]
|
[8.33129596710205, 0.5621653199195862]
|
9849d113-f419-47bb-bcba-33490c0ea89e
|
vidimu-multimodal-video-and-imu-kinematic
|
2303.1615
| null |
https://arxiv.org/abs/2303.16150v1
|
https://arxiv.org/pdf/2303.16150v1.pdf
|
VIDIMU. Multimodal video and IMU kinematic dataset on daily life activities using affordable devices
|
Human activity recognition and clinical biomechanics are challenging problems in physical telerehabilitation medicine. However, most publicly available datasets on human body movements cannot be used to study both problems in an out-of-the-lab movement acquisition setting. The objective of the VIDIMU dataset is to pave the way towards affordable patient tracking solutions for remote daily life activities recognition and kinematic analysis. The dataset includes 13 activities registered using a commodity camera and five inertial sensors. The video recordings were acquired in 54 subjects, of which 16 also had simultaneous recordings of inertial sensors. The novelty of VIDIMU lies in: i) the clinical relevance of the chosen movements, ii) the combined utilization of affordable video and custom sensors, and iii) the implementation of state-of-the-art tools for multimodal data processing of 3D body pose tracking and motion reconstruction in a musculoskeletal model from inertial data. The validation confirms that a minimally disturbing acquisition protocol, performed according to real-life conditions can provide a comprehensive picture of human joint angles during daily life activities.
|
['Cristina Simón-Martínez', 'Henning Müller', 'Francisco J. Díaz-Pernas', 'Míriam Antón-Rodríguez', 'Javier González-Alonso', 'Mario Martínez-Zarzuela']
|
2023-03-27
| null | null | null | null |
['pose-tracking', 'human-activity-recognition', 'human-activity-recognition']
|
['computer-vision', 'computer-vision', 'time-series']
|
[ 1.78737998e-01 -1.22379668e-01 -2.33357579e-01 1.58488408e-01
-5.56174994e-01 -8.81437287e-02 1.88467860e-01 -1.13827832e-01
-7.81822801e-01 6.67523980e-01 4.07773167e-01 4.31641228e-02
-4.20944929e-01 -1.84191838e-01 -4.99737740e-01 -4.44253087e-01
-4.17287797e-01 8.54277194e-01 1.41992152e-01 -3.63317400e-01
-2.64181197e-01 5.21747887e-01 -1.66708207e+00 2.47778352e-02
3.88117522e-01 6.82671249e-01 1.05815284e-01 8.21464896e-01
6.20376468e-01 2.09887818e-01 -2.99231499e-01 6.07930385e-02
1.25123009e-01 -5.23898661e-01 -5.56095362e-01 2.41999686e-01
3.64009529e-01 -3.37924272e-01 -2.52976775e-01 3.14358234e-01
1.05696785e+00 1.32346079e-01 1.37749970e-01 -8.00354004e-01
1.73064470e-01 1.24871284e-01 -9.52972174e-02 1.91671848e-01
1.25362790e+00 3.01191956e-01 1.55065417e-01 -7.23365843e-01
1.11375844e+00 5.11869431e-01 9.80957866e-01 5.26476502e-01
-1.02675676e+00 -1.58357099e-01 -5.88405311e-01 2.03856066e-01
-1.39449644e+00 -4.75722790e-01 5.81547260e-01 -7.22277045e-01
8.62668276e-01 7.22725570e-01 1.54311788e+00 1.48855102e+00
6.27592564e-01 2.92380363e-01 1.10596955e+00 -4.90010202e-01
2.43526027e-01 -2.21447602e-01 -1.34704873e-01 3.89102817e-01
6.62177384e-01 1.39472866e-03 -8.89649749e-01 -1.32414410e-02
1.03523636e+00 2.46802390e-01 -4.88601536e-01 -5.56866586e-01
-1.75478208e+00 3.03418972e-02 2.16353405e-02 6.98046505e-01
-7.70368934e-01 1.47119090e-01 4.33889598e-01 1.84384316e-01
1.73559546e-01 9.38185304e-02 -5.11351228e-01 -9.87770200e-01
-8.84347439e-01 2.01255724e-01 6.72497332e-01 6.50098979e-01
3.32443858e-03 -1.28550112e-01 8.02504644e-02 2.42103145e-01
4.72496867e-01 6.89691186e-01 8.44350874e-01 -7.13200927e-01
4.61188316e-01 5.33329189e-01 -4.33961861e-03 -8.48613143e-01
-9.69447613e-01 -2.93805033e-01 -3.81608844e-01 1.00701936e-01
6.08674467e-01 -3.41708452e-01 -4.35441554e-01 1.21192563e+00
8.18736613e-01 -1.19597606e-01 -4.45208818e-01 1.34712481e+00
4.81799096e-01 -2.62973905e-01 -6.72465190e-02 -3.55914742e-01
1.62143648e+00 -1.46789402e-01 -8.46613526e-01 -1.72670074e-02
9.94637311e-01 -8.13035190e-01 1.01634514e+00 6.26964390e-01
-1.11403382e+00 -5.43084025e-01 -9.16324079e-01 2.05880418e-01
-4.13982011e-02 3.37931752e-01 4.94171560e-01 9.12844718e-01
-6.21040702e-01 6.77749574e-01 -1.45657468e+00 -7.81816304e-01
-1.48060098e-01 7.23738194e-01 -9.34853971e-01 2.46908665e-01
-9.69384313e-01 1.17664766e+00 1.60085917e-01 5.21254539e-01
-3.65820229e-01 -5.91468751e-01 -8.51020932e-01 -5.62107384e-01
4.74015921e-01 -1.05118680e+00 8.83673847e-01 -4.85475034e-01
-1.83735704e+00 1.09837151e+00 2.34140769e-01 4.84886207e-02
1.15337932e+00 -9.67354178e-01 -4.33422089e-01 3.81561279e-01
-2.47571126e-01 -1.78408399e-01 5.51648676e-01 -4.60331917e-01
9.81381685e-02 -7.80066311e-01 -6.29743338e-01 4.20033216e-01
-1.48079157e-01 -2.01284349e-01 -3.62274736e-01 -6.23065710e-01
3.62748116e-01 -1.29702592e+00 8.03789124e-03 1.38855323e-01
-1.37440160e-01 6.10999107e-01 5.05591273e-01 -9.30121183e-01
1.04970360e+00 -1.84325254e+00 5.76252997e-01 3.58778417e-01
-1.24952262e-02 2.47576565e-01 6.76987410e-01 6.54836237e-01
-1.86922550e-01 -5.49935222e-01 3.01803529e-01 -6.46944940e-02
-2.18796656e-01 2.89099485e-01 3.80542934e-01 1.16420054e+00
-7.74799228e-01 6.97958946e-01 -6.94449425e-01 -6.08162403e-01
8.28619778e-01 5.46909809e-01 -2.76373565e-01 2.00566888e-01
3.04512948e-01 1.18764925e+00 -4.11606133e-01 7.58898377e-01
-1.30263912e-02 1.84101999e-01 4.69734758e-01 -4.37845975e-01
-1.59184143e-01 1.77399404e-02 -1.60486937e+00 2.35309744e+00
-2.14575484e-01 1.15409367e-01 7.36805797e-02 -5.17524540e-01
4.25020844e-01 7.94787943e-01 1.15478408e+00 -5.68778872e-01
6.45063519e-01 4.88966763e-01 1.19129010e-02 -1.02727997e+00
2.28306904e-01 -1.76637813e-01 3.31636705e-02 1.99272096e-01
7.72189051e-02 4.52079140e-02 -3.06690261e-02 -4.80283827e-01
9.81737792e-01 6.00676417e-01 5.93402803e-01 -1.93325847e-01
4.53823566e-01 -9.92032886e-02 6.57662824e-02 3.32953364e-01
-3.03602070e-01 7.87784874e-01 -2.30080619e-01 -4.01823610e-01
-7.37666190e-01 -1.09443378e+00 -1.22989120e-03 6.20585263e-01
-2.30438799e-01 -4.28899378e-01 -7.58130670e-01 8.68491828e-03
1.28293201e-01 -1.69975385e-01 -6.13505125e-01 -8.03008676e-02
-7.08088338e-01 -4.13179278e-01 5.59504986e-01 5.15420496e-01
-1.05174169e-01 -6.87272668e-01 -1.30135453e+00 3.98422241e-01
-1.73834682e-01 -1.05009162e+00 -1.80655405e-01 -2.38750979e-01
-1.21924901e+00 -1.42373884e+00 -8.80042791e-01 -2.47073337e-01
3.78652185e-01 -1.67606160e-01 7.35026419e-01 -1.37037873e-01
-5.34009218e-01 9.71193373e-01 -4.52703744e-01 -1.53585270e-01
-1.16626151e-01 -1.40807582e-02 6.60894096e-01 -1.52244568e-01
-7.08045587e-02 -7.87957907e-01 -7.44970500e-01 5.64954102e-01
-5.79387486e-01 -9.69287902e-02 5.30131280e-01 3.65616262e-01
7.50015676e-01 -9.06205416e-01 -1.84210747e-01 -3.75923812e-01
4.23033535e-01 -1.58597842e-01 -1.23883739e-01 1.27754817e-02
-3.09259802e-01 -5.99229038e-01 8.67723313e-04 -5.11238217e-01
-6.06031895e-01 2.31179103e-01 -4.55398709e-01 -2.24715650e-01
-4.05621141e-01 4.88094091e-01 4.44619507e-02 -1.25811100e-01
7.29361832e-01 -2.74239834e-02 5.76960504e-01 -6.19026780e-01
1.46197453e-01 3.71744365e-01 8.90585661e-01 -5.70352912e-01
2.45260730e-01 6.21822298e-01 4.03728813e-01 -1.18422830e+00
-2.09315959e-02 -7.47512817e-01 -1.09184706e+00 -8.95137310e-01
8.73283863e-01 -8.35357368e-01 -1.08745933e+00 5.81573129e-01
-5.92000544e-01 -1.97150365e-01 -5.49676180e-01 1.39890671e+00
-9.90048826e-01 6.69564962e-01 -5.24341702e-01 -9.09772992e-01
-4.53213811e-01 -1.00870359e+00 1.25098765e+00 -1.10726461e-01
-9.11231220e-01 -7.64129996e-01 5.81349432e-01 5.56266665e-01
2.21946165e-01 9.79009628e-01 -1.24504559e-01 -6.39164969e-02
1.86508358e-01 -9.10329938e-01 7.87566543e-01 -8.90219137e-02
1.73452362e-01 -1.77743003e-01 -6.26435280e-01 -4.08614457e-01
1.87434539e-01 -2.12882161e-01 -2.04104826e-01 5.35330892e-01
5.27192891e-01 7.02092350e-02 -2.46985897e-01 3.88243407e-01
1.10823393e+00 -2.45151520e-01 9.17570174e-01 4.86309588e-01
8.69274259e-01 6.55772984e-01 7.95971096e-01 5.62153459e-01
-7.63772056e-02 1.16222334e+00 3.72967303e-01 -1.66170287e-03
-2.22988776e-03 1.27264410e-02 4.11801070e-01 1.04413080e+00
-9.85314488e-01 3.82219255e-01 -1.16603065e+00 2.71409661e-01
-1.77241218e+00 -7.59336412e-01 -5.78707099e-01 2.58442140e+00
4.55443054e-01 9.55036432e-02 4.33052450e-01 5.60689688e-01
3.58210206e-01 -2.00598568e-01 -5.58657423e-02 5.28009376e-03
2.96422869e-01 2.56399542e-01 5.62106788e-01 1.83221966e-01
-8.07384610e-01 -2.09330959e-04 6.07316160e+00 1.06390730e-01
-1.21740448e+00 3.19177687e-01 -3.71785730e-01 -5.00867069e-01
4.37747061e-01 -2.10913315e-01 -3.75421911e-01 5.18264949e-01
1.36555052e+00 2.90358484e-01 -2.30492577e-01 8.95322561e-01
6.21696174e-01 -4.50576276e-01 -1.07476604e+00 9.85504150e-01
1.70800179e-01 -9.80621219e-01 -4.52633739e-01 3.66516322e-01
1.96174294e-01 1.43002972e-01 -3.36375147e-01 -3.07185590e-01
-8.69043648e-01 -6.78930104e-01 7.40474105e-01 1.14442265e+00
6.96725845e-01 -2.27130651e-01 6.60412669e-01 5.18424332e-01
-9.47661042e-01 2.99120456e-01 2.95508027e-01 -3.79130661e-01
6.66813493e-01 4.75014448e-01 -5.08038402e-01 9.86817420e-01
5.08775592e-01 7.10352302e-01 -3.88964564e-01 1.01040781e+00
-2.24818606e-02 5.27799845e-01 -6.50975764e-01 1.99336544e-01
-1.87780797e-01 -2.61070579e-01 6.98717356e-01 9.75461185e-01
5.25876999e-01 -1.10201091e-01 1.23750024e-01 6.61362931e-02
6.25146091e-01 2.22884193e-01 -7.49864399e-01 6.12420067e-02
-3.34449828e-01 1.05482686e+00 -6.20819747e-01 7.13384375e-02
-1.07831940e-01 8.20530176e-01 -4.11846012e-01 -1.81716487e-01
-6.66550756e-01 1.13773651e-01 5.86378932e-01 8.11612427e-01
-1.79155767e-01 -7.45374322e-01 -2.10161239e-01 -1.20268679e+00
4.93408531e-01 -9.68389213e-01 4.72905725e-01 -8.92610073e-01
-5.61296761e-01 4.12910022e-02 2.90477395e-01 -1.76440775e+00
-6.64246380e-01 -7.11741507e-01 -2.20995978e-01 5.92637777e-01
-3.32515895e-01 -1.13268781e+00 -5.68234801e-01 7.95835733e-01
2.20105425e-01 3.46145689e-01 1.05966723e+00 5.89501441e-01
-4.70494896e-01 -7.07327295e-03 -1.17547825e-01 -2.34220356e-01
6.85206056e-01 -8.37687254e-01 -2.05054507e-01 6.26917779e-01
4.09860611e-02 8.04394305e-01 9.34247434e-01 -7.34388709e-01
-2.07469344e+00 -1.62736997e-01 7.43123770e-01 -9.58520472e-01
5.45410156e-01 -1.63829893e-01 -5.38272142e-01 7.51519918e-01
-2.98769504e-01 7.48053864e-02 9.55242276e-01 -1.98233679e-01
5.58959723e-01 1.44801447e-02 -9.78873074e-01 4.07694250e-01
1.13334382e+00 -4.10124481e-01 -9.50902700e-01 4.78153348e-01
-1.52431741e-01 -1.21923983e+00 -1.39292085e+00 3.95637929e-01
1.49966824e+00 -7.41979718e-01 1.03749478e+00 -4.23810929e-01
1.39283985e-01 -1.66322723e-01 7.98873380e-02 -7.77677178e-01
1.76769748e-01 -7.64540076e-01 -3.22041035e-01 6.82283580e-01
-3.36112827e-01 -5.65273285e-01 8.85036170e-01 5.28513670e-01
-5.23726456e-02 -6.78887546e-01 -1.49135351e+00 -6.98058069e-01
-5.56273580e-01 -8.32664967e-01 7.21815834e-03 5.57802856e-01
4.30493325e-01 -1.82879403e-01 -6.66624844e-01 -7.32074156e-02
4.42107171e-01 -3.10705096e-01 1.24065936e+00 -1.15479219e+00
-4.51342672e-01 7.06681684e-02 -1.06850421e+00 -3.18775713e-01
-5.48089266e-01 -1.59233242e-01 -2.85154969e-01 -1.37890124e+00
-3.71862829e-01 2.80029327e-01 6.76898137e-02 5.95455319e-02
2.96026617e-01 4.71197754e-01 -5.00564799e-02 4.88926888e-01
-3.67631435e-01 2.14511260e-01 1.23165584e+00 5.34940958e-01
-1.70334145e-01 1.10369280e-01 1.13625966e-01 6.66680574e-01
2.78421789e-01 -5.43586791e-01 -3.09203893e-01 -6.16100542e-02
3.79682869e-01 2.61055231e-01 5.09687662e-01 -1.47112584e+00
2.51057178e-01 1.94096819e-01 5.16725063e-01 -4.84368801e-01
4.04589027e-01 -1.11807394e+00 1.17898190e+00 1.14583266e+00
4.00174946e-01 1.48830622e-01 3.45429257e-02 3.06618661e-01
-7.17382133e-02 2.14419574e-01 3.28150094e-01 -2.44777888e-01
-5.67584395e-01 3.77310179e-02 -5.88657916e-01 3.08343992e-02
1.14987123e+00 -1.02595651e+00 4.24995124e-02 -1.25309750e-01
-1.39161742e+00 -2.71341413e-01 6.48154020e-01 2.52543807e-01
3.56529176e-01 -1.21969008e+00 -3.79837573e-01 3.33658159e-01
2.35823199e-01 -2.29120731e-01 6.11596167e-01 1.84254146e+00
-9.09941494e-01 4.86728907e-01 -7.17293382e-01 -8.87206972e-01
-1.47862768e+00 3.89112085e-02 4.33062822e-01 -2.35978171e-01
-1.04429233e+00 1.84914395e-01 -7.59183526e-01 -1.70513004e-01
-3.20390239e-02 -3.49204153e-01 -1.46581024e-01 1.70929983e-01
2.70467132e-01 7.11156905e-01 6.95588291e-01 -9.30688858e-01
-6.97346926e-01 7.55290389e-01 4.53765690e-01 -3.60971212e-01
1.12910068e+00 -3.41959864e-01 2.38760620e-01 6.96278930e-01
7.68360734e-01 5.36718629e-02 -8.35606098e-01 2.52731204e-01
-6.67333901e-02 -5.73133886e-01 -1.17891721e-01 -5.83736897e-01
-6.41533673e-01 4.81002599e-01 1.04039371e+00 -2.00837791e-01
1.09169412e+00 -2.52843410e-01 7.11465418e-01 1.77531317e-01
8.24301660e-01 -1.27568996e+00 -1.70254081e-01 4.19062711e-02
1.04375219e+00 -6.71376944e-01 4.27994072e-01 -4.76498157e-01
-4.42486703e-01 1.07850218e+00 8.93453285e-02 -8.18599090e-02
6.57380879e-01 2.31528699e-01 1.57642975e-01 -3.89677793e-01
-6.45042211e-02 -2.50841111e-01 5.80179214e-01 6.95857823e-01
7.01465786e-01 4.76647764e-02 -1.04790998e+00 3.21816742e-01
-2.06454217e-01 6.11807406e-01 3.45449209e-01 1.68123531e+00
7.18540326e-02 -9.20346558e-01 -7.54997611e-01 2.14824468e-01
-7.36235738e-01 6.83682323e-01 -2.98933506e-01 1.35632098e+00
1.30211949e-01 5.72679102e-01 -3.78166944e-01 -3.94282579e-01
1.07597983e+00 1.59492299e-01 9.72492158e-01 -5.86197674e-01
-9.75560546e-01 1.37614712e-01 2.26067588e-01 -1.15333724e+00
-6.42119884e-01 -1.09057438e+00 -9.60832775e-01 -1.85643673e-01
-1.26833403e-02 -1.67708367e-01 1.07319379e+00 9.80457067e-01
3.78536254e-01 6.28253102e-01 -1.46095261e-01 -1.41554606e+00
-3.52477998e-01 -1.11532259e+00 -9.12509859e-01 6.60568595e-01
3.15978415e-02 -1.05979717e+00 -2.25119013e-02 2.12991461e-01]
|
[7.0790839195251465, 0.10524610430002213]
|
2de5b422-e4cb-44e6-a6b3-add20b97836c
|
contextual-augmentation-data-augmentation-by
|
1805.06201
| null |
http://arxiv.org/abs/1805.06201v1
|
http://arxiv.org/pdf/1805.06201v1.pdf
|
Contextual Augmentation: Data Augmentation by Words with Paradigmatic Relations
|
We propose a novel data augmentation for labeled sentences called contextual
augmentation. We assume an invariance that sentences are natural even if the
words in the sentences are replaced with other words with paradigmatic
relations. We stochastically replace words with other words that are predicted
by a bi-directional language model at the word positions. Words predicted
according to a context are numerous but appropriate for the augmentation of the
original words. Furthermore, we retrofit a language model with a
label-conditional architecture, which allows the model to augment sentences
without breaking the label-compatibility. Through the experiments for six
various different text classification tasks, we demonstrate that the proposed
method improves classifiers based on the convolutional or recurrent neural
networks.
|
['Sosuke Kobayashi']
|
2018-05-16
|
contextual-augmentation-data-augmentation-by-1
|
https://aclanthology.org/N18-2072
|
https://aclanthology.org/N18-2072.pdf
|
naacl-2018-6
|
['text-augmentation']
|
['natural-language-processing']
|
[ 6.53708100e-01 4.54833269e-01 -2.04385936e-01 -8.21934938e-01
-1.68258220e-01 -3.59709024e-01 7.78535903e-01 -1.24432512e-01
-7.03604639e-01 8.46769154e-01 7.46716797e-01 -4.74269688e-01
5.41506767e-01 -6.59596026e-01 -7.57863045e-01 -5.09337723e-01
4.31666046e-01 3.19582164e-01 -2.24880740e-01 -5.84619820e-01
1.03846252e-01 2.34419033e-01 -1.34105933e+00 6.22957289e-01
8.55707228e-01 4.58796144e-01 1.85402498e-01 5.01786113e-01
-7.43501782e-01 9.00460482e-01 -7.07777143e-01 -3.83742005e-01
8.13676342e-02 -4.80557978e-01 -8.57004583e-01 1.44890353e-01
3.75245810e-01 -4.82141785e-02 -4.46268767e-01 9.54495072e-01
1.08312204e-01 3.21453393e-01 6.18043602e-01 -9.43450928e-01
-1.26490855e+00 1.24719203e+00 -3.20363790e-01 1.46980220e-02
2.68283576e-01 -2.94122130e-01 1.01909494e+00 -1.18543506e+00
3.38617772e-01 1.46457863e+00 5.77116787e-01 8.82294357e-01
-1.29661417e+00 -6.75027490e-01 9.07034099e-01 5.84361181e-02
-1.15800071e+00 -4.58959311e-01 1.00679851e+00 -1.82840973e-01
1.10608113e+00 3.28071922e-01 4.08242971e-01 1.48916340e+00
3.37664396e-01 5.12997746e-01 1.07682812e+00 -9.54506278e-01
2.25621276e-02 3.08507204e-01 8.46380234e-01 3.07007909e-01
1.34718001e-01 -1.47099808e-01 -5.15609443e-01 -7.04484060e-02
1.98315963e-01 1.14366211e-01 8.62080529e-02 7.45702684e-02
-1.18110573e+00 8.64008248e-01 3.81753832e-01 5.61851323e-01
-3.42852622e-01 1.33905366e-01 5.37625849e-01 1.66935742e-01
8.84307206e-01 5.76462924e-01 -8.05755377e-01 5.31666100e-01
-4.31233823e-01 5.61231263e-02 4.55629110e-01 1.16363919e+00
6.78755164e-01 3.44212085e-01 -5.73474050e-01 8.98993909e-01
3.19342256e-01 5.39684474e-01 9.60968912e-01 -2.95683146e-01
3.12265545e-01 6.89295471e-01 -7.26716593e-02 -8.12603593e-01
-5.95593035e-01 -7.03888178e-01 -1.10651565e+00 -3.70070815e-01
-1.88605487e-01 -1.48639172e-01 -1.10635245e+00 2.22842860e+00
-1.29776120e-01 2.69723296e-01 3.99168849e-01 3.53807867e-01
1.10880554e+00 7.69247234e-01 5.60750127e-01 -6.20370924e-01
1.44907975e+00 -1.33650827e+00 -1.47390723e+00 -6.87700510e-01
1.11731339e+00 -5.99445105e-01 1.59068286e+00 2.20804475e-02
-6.80718124e-01 -8.82946968e-01 -9.68514800e-01 -2.10070297e-01
-7.51003206e-01 8.44651386e-02 5.31227469e-01 7.01215327e-01
-8.63757014e-01 2.33869115e-03 -3.66215587e-01 -1.32406518e-01
-6.77906647e-02 2.05609098e-01 -4.04971480e-01 3.62030566e-01
-1.82380748e+00 1.13868928e+00 6.04775608e-01 1.70364723e-01
-4.14044410e-01 -3.90883744e-01 -1.08765137e+00 -3.78456861e-02
8.79314169e-02 -6.21739507e-01 1.05259168e+00 -1.19195294e+00
-1.25008225e+00 9.25466001e-01 -6.44378185e-01 -5.38823426e-01
3.74502279e-02 -4.21613783e-01 -6.58975244e-01 -4.66899157e-01
5.61570786e-02 7.24457681e-01 6.85706317e-01 -1.38706529e+00
-2.14572921e-01 -1.88759983e-01 8.03652704e-02 4.44988042e-01
-8.17629576e-01 4.87276576e-02 5.62538281e-02 -1.18750882e+00
1.84537724e-01 -1.01627052e+00 -4.41868782e-01 -6.56559467e-01
-7.48323739e-01 -3.16285610e-01 9.83874142e-01 -4.99742270e-01
1.32581007e+00 -2.02966261e+00 4.86564115e-02 -1.25862643e-01
-5.93900904e-02 2.65486270e-01 -3.89910638e-01 3.25582117e-01
-6.27498806e-01 5.51241875e-01 -3.12772334e-01 -8.06881130e-01
-1.83816850e-01 5.82668960e-01 -7.86166549e-01 -1.01192094e-01
1.85790285e-01 1.14893270e+00 -6.31124616e-01 -3.45683396e-01
1.37085184e-01 2.92288363e-01 -4.07165647e-01 2.28761896e-01
-3.99249762e-01 5.88032007e-01 -2.43091285e-01 -7.81821162e-02
5.81299484e-01 -6.86010718e-02 7.61232674e-02 -7.23508447e-02
2.28345200e-01 7.55510807e-01 -9.00638759e-01 1.63510752e+00
-6.57625377e-01 3.15363109e-01 -5.49419999e-01 -1.00686228e+00
1.12137806e+00 3.46156746e-01 -1.38199419e-01 -2.91341603e-01
1.26034856e-01 -1.37843013e-01 1.51676014e-01 -5.14235139e-01
8.20918441e-01 -4.08846945e-01 -2.51081228e-01 5.55662990e-01
-4.73388433e-02 -1.83066249e-01 3.82778682e-02 3.62754464e-01
6.72614515e-01 6.30556494e-02 2.58461744e-01 -2.68128097e-01
8.31224382e-01 -3.81529480e-01 5.57012975e-01 9.18439031e-01
1.66251093e-01 2.91950703e-01 2.13859230e-01 -4.70160335e-01
-1.00677693e+00 -6.98283792e-01 -3.94849300e-01 1.66737878e+00
-2.16109917e-01 -3.84398282e-01 -4.85191554e-01 -1.02838981e+00
-3.13908309e-01 1.46055245e+00 -9.62648928e-01 -5.04690349e-01
-4.79914904e-01 -7.77650654e-01 4.53095317e-01 6.84545755e-01
4.04962867e-01 -1.35587013e+00 3.95090729e-01 8.20856020e-02
-2.12835237e-01 -1.25379920e+00 -6.06308103e-01 6.38650477e-01
-6.74422860e-01 -3.33459824e-01 -1.29059955e-01 -1.24593377e+00
1.06238699e+00 9.29528251e-02 9.51002896e-01 1.03751391e-01
5.39782226e-01 -5.12127252e-03 -7.19922960e-01 -4.60219264e-01
-8.09873879e-01 3.14651340e-01 4.11299765e-01 9.38702151e-02
6.76289737e-01 -4.93259698e-01 7.17427162e-03 -2.90647801e-02
-1.24039865e+00 4.07365799e-01 3.28506380e-01 1.12770426e+00
3.92803371e-01 -4.24807280e-01 9.67821002e-01 -1.44210649e+00
7.81779408e-01 -2.93897539e-01 9.56871733e-03 3.34841549e-01
-5.70425749e-01 3.24968368e-01 7.03470707e-01 -6.90843642e-01
-1.26593852e+00 9.83929709e-02 -3.69189084e-01 4.35366966e-02
-2.86131203e-01 6.00130916e-01 -3.10351938e-01 3.72245699e-01
8.44839096e-01 3.40498537e-01 -3.30958813e-01 -3.31256479e-01
9.65285897e-01 7.26530433e-01 3.80369991e-01 -4.43710595e-01
7.89556324e-01 9.35482532e-02 -1.31775364e-01 -6.90715313e-01
-1.35774744e+00 5.11523336e-02 -9.56798494e-01 1.29152695e-02
7.59742856e-01 -7.83236206e-01 -8.40065330e-02 4.03429747e-01
-1.70039809e+00 1.62868917e-01 -5.44929087e-01 4.62049067e-01
-1.22714914e-01 3.31587374e-01 -5.81364393e-01 -9.04250443e-01
-3.01876843e-01 -1.00867975e+00 8.37717354e-01 2.46433616e-02
-4.72811162e-01 -1.17735374e+00 -9.88134742e-02 1.95508122e-01
3.83069634e-01 -3.07035655e-01 1.21113241e+00 -1.58327186e+00
1.43124118e-01 -3.80519181e-01 1.80971816e-01 8.68012667e-01
6.17214918e-01 -1.31062582e-01 -1.24828315e+00 -3.47783901e-02
1.63696289e-01 -3.32312584e-01 9.57392514e-01 1.96539506e-01
1.28103483e+00 -4.64637756e-01 -1.85000271e-01 2.70869434e-01
6.50712073e-01 3.69563490e-01 5.24854362e-01 1.41275644e-01
8.89570236e-01 7.65667260e-01 3.46990317e-01 1.20683974e-02
1.15801863e-01 4.03802246e-01 1.15164950e-01 -2.11633921e-01
2.60223318e-02 -4.41006809e-01 3.28164309e-01 1.38484907e+00
5.68205535e-01 -4.73656595e-01 -8.74648154e-01 3.59335303e-01
-1.73216271e+00 -6.18125737e-01 -5.45316637e-01 1.80134654e+00
1.18658447e+00 4.92123157e-01 -5.96880913e-01 2.84199174e-02
9.22267139e-01 3.36123079e-01 -3.29027474e-01 -6.47351265e-01
-7.11028516e-01 2.49805897e-01 1.53242663e-01 9.17832613e-01
-1.13773310e+00 1.41351128e+00 7.51142693e+00 5.86399257e-01
-8.76153946e-01 1.56914696e-01 9.52010214e-01 1.62261561e-01
-7.21231222e-01 -1.35031557e-02 -9.42175150e-01 2.84013391e-01
8.83572102e-01 -2.45730191e-01 7.71696568e-02 8.29312861e-01
2.80930549e-01 2.55974412e-01 -1.22695971e+00 5.25042772e-01
4.00813878e-01 -1.10209930e+00 7.88741112e-01 -4.46945846e-01
9.64776516e-01 -4.03282344e-01 2.30477512e-01 6.39660239e-01
2.34410942e-01 -1.14601934e+00 4.65932876e-01 6.72014475e-01
5.43993056e-01 -5.97053289e-01 1.04745162e+00 3.81868005e-01
-7.18887150e-01 1.69393376e-01 -2.70991802e-01 -5.76073110e-01
5.03682941e-02 3.96298587e-01 -7.70929694e-01 2.50781506e-01
1.40258729e-01 5.00163734e-01 -8.06753099e-01 7.93371871e-02
-6.79188609e-01 6.98263705e-01 1.17975980e-01 -2.11079687e-01
2.99722832e-02 -1.97786555e-01 4.82220531e-01 1.31548083e+00
-9.43436921e-02 9.21573490e-03 1.39909908e-01 6.95818067e-01
-4.38195229e-01 6.56815708e-01 -9.54025269e-01 -1.51070030e-02
5.22100031e-01 1.10535622e+00 -4.84082431e-01 -6.77634001e-01
-5.44776142e-01 9.65486050e-01 3.75613570e-01 5.99841177e-01
-6.25600398e-01 -1.60870641e-01 3.32610250e-01 -2.40929455e-01
-3.05766284e-01 -8.79707038e-02 -8.36776137e-01 -1.08877993e+00
1.41712919e-01 -6.54353619e-01 9.18313190e-02 -1.04281092e+00
-1.64235783e+00 1.02263427e+00 -8.71000960e-02 -9.35611129e-01
-2.36426041e-01 -6.06605649e-01 -4.89543259e-01 1.09757602e+00
-1.18765771e+00 -1.24242055e+00 2.27379594e-02 3.29723567e-01
6.63391352e-01 -4.37027156e-01 1.25610268e+00 -3.74252275e-02
-5.77051163e-01 6.49455070e-01 -1.65968865e-01 1.85471937e-01
8.59419584e-01 -1.08489215e+00 6.94622397e-01 9.48815823e-01
2.59319335e-01 1.01150250e+00 7.89965510e-01 -8.94915581e-01
-7.09899962e-01 -1.25899506e+00 1.54637718e+00 -5.42011023e-01
7.14601874e-01 -7.92361259e-01 -1.30852509e+00 1.28546321e+00
6.46891236e-01 6.10105135e-02 9.74802852e-01 3.47838432e-01
-5.33652723e-01 1.79061890e-02 -7.79055417e-01 8.53924870e-01
1.08572185e+00 -8.23302209e-01 -1.25259376e+00 7.08398104e-01
1.64446640e+00 -8.69640708e-02 -4.21668291e-02 6.55574203e-01
2.78760605e-02 -1.58039480e-01 5.19037545e-01 -1.50347042e+00
4.37576622e-01 -2.24171847e-01 -3.65457296e-01 -1.47172213e+00
-4.56941128e-01 -3.19753289e-01 8.27479884e-02 1.45114934e+00
8.37436736e-01 -7.99723983e-01 6.38749063e-01 7.52730846e-01
-3.23129445e-01 -2.69253105e-01 -9.24728036e-01 -7.03932524e-01
3.94275665e-01 -6.65283203e-01 5.57863235e-01 1.28618681e+00
1.33150578e-01 1.10229158e+00 -5.88976324e-01 -7.74086267e-02
-9.19584781e-02 -3.84353906e-01 3.74004513e-01 -1.09032941e+00
-2.24494729e-02 -2.88822919e-01 -1.06677197e-01 -1.27906120e+00
9.35413301e-01 -1.13644564e+00 1.74704820e-01 -1.34927368e+00
3.71608943e-01 -2.66689867e-01 -4.88523185e-01 7.07042098e-01
-6.91903591e-01 6.78417906e-02 -1.23500906e-01 -1.32793024e-01
-2.92013288e-01 1.05140114e+00 1.13118958e+00 -2.64312863e-01
-2.72522926e-01 -1.29742175e-01 -8.38288665e-01 9.33050632e-01
7.74679899e-01 -4.86483246e-01 -5.45985579e-01 -4.45660084e-01
4.29438382e-01 -3.60170186e-01 -1.02381818e-01 -5.19598424e-01
8.94635171e-03 -2.55252495e-02 2.35484466e-01 -6.45651102e-01
2.56514341e-01 -9.35788155e-01 -2.53711700e-01 2.90000081e-01
-1.29349959e+00 2.76839644e-01 3.92437369e-01 4.92165595e-01
-9.70247313e-02 -5.69211721e-01 6.63865447e-01 1.90556020e-01
-3.56023371e-01 -1.34401903e-01 -7.22362936e-01 -9.82594043e-02
9.36732650e-01 2.26337194e-01 -4.06423658e-01 -4.93303865e-01
-1.11918771e+00 3.63028422e-02 -7.75697231e-02 7.17995703e-01
5.74534535e-01 -1.67570662e+00 -9.04860497e-01 3.86026621e-01
3.35445851e-01 -2.31779262e-01 1.55191943e-01 2.77276576e-01
-2.60273218e-02 5.69207847e-01 1.89363554e-01 -3.03285241e-01
-1.17865157e+00 9.93932843e-01 3.51566583e-01 -2.36134127e-01
-2.82947987e-01 9.12666321e-01 6.05404437e-01 -8.46564472e-01
3.48743707e-01 -4.45484102e-01 -5.75257599e-01 7.98414554e-03
4.10122365e-01 -2.62995899e-01 1.81133941e-01 -8.26713264e-01
-2.84176469e-01 1.19089097e-01 -5.39664030e-01 -1.22914024e-01
1.04997921e+00 -2.16829389e-01 -3.90837997e-01 1.11519194e+00
9.01307166e-01 1.81898415e-01 -4.26280260e-01 -7.84390509e-01
1.09847218e-01 -3.07596978e-02 -3.61114778e-02 -8.29985440e-01
-6.99798703e-01 6.85268164e-01 3.81720334e-01 3.42321873e-01
9.19261575e-01 -1.64207071e-01 2.89779365e-01 6.54391289e-01
-7.22198263e-02 -8.88235033e-01 1.72324166e-01 1.06577420e+00
1.28698242e+00 -1.05141664e+00 -2.13979751e-01 -6.07384801e-01
-7.25923598e-01 9.25014079e-01 1.01545870e+00 1.98292173e-03
7.99472749e-01 2.88908035e-01 2.64570504e-01 2.45698407e-01
-9.04974997e-01 -6.22888654e-03 2.38010630e-01 4.52910960e-01
8.66390824e-01 1.79318916e-02 -7.36335754e-01 9.25158560e-01
-3.52576762e-01 -6.72513723e-01 5.17207384e-01 5.84593475e-01
-3.29805464e-01 -1.42453825e+00 -3.61454457e-01 4.13609743e-01
-2.29415402e-01 -7.44921446e-01 -6.41144454e-01 4.71760064e-01
1.70840979e-01 1.17361677e+00 1.91279814e-01 -5.08404255e-01
3.27254802e-01 9.47459340e-01 -2.16175944e-01 -1.09161484e+00
-5.70184767e-01 -8.88884217e-02 2.66955554e-01 2.40569904e-01
-3.62386167e-01 -3.86032760e-01 -1.39884841e+00 1.36792496e-01
-3.65437955e-01 3.40328217e-01 6.28895938e-01 1.27129793e+00
2.25620508e-01 9.62525725e-01 8.05487871e-01 -2.94596463e-01
-2.78355181e-01 -1.68624806e+00 -4.22024101e-01 5.90941727e-01
1.72933847e-01 -4.60475653e-01 -3.72236669e-01 1.88123614e-01]
|
[11.257574081420898, 8.895630836486816]
|
ae57b0b0-bee9-4304-a201-d9c07f18c1f6
|
using-search-queries-to-understand-health
|
1806.0574
| null |
http://arxiv.org/abs/1806.05740v2
|
http://arxiv.org/pdf/1806.05740v2.pdf
|
Using Search Queries to Understand Health Information Needs in Africa
|
The lack of comprehensive, high-quality health data in developing nations
creates a roadblock for combating the impacts of disease. One key challenge is
understanding the health information needs of people in these nations. Without
understanding people's everyday needs, concerns, and misconceptions, health
organizations and policymakers lack the ability to effectively target education
and programming efforts. In this paper, we propose a bottom-up approach that
uses search data from individuals to uncover and gain insight into health
information needs in Africa. We analyze Bing searches related to HIV/AIDS,
malaria, and tuberculosis from all 54 African nations. For each disease, we
automatically derive a set of common search themes or topics, revealing a
wide-spread interest in various types of information, including disease
symptoms, drugs, concerns about breastfeeding, as well as stigma, beliefs in
natural cures, and other topics that may be hard to uncover through traditional
surveys. We expose the different patterns that emerge in health information
needs by demographic groups (age and sex) and country. We also uncover
discrepancies in the quality of content returned by search engines to users by
topic. Combined, our results suggest that search data can help illuminate
health information needs in Africa and inform discussions on health policy and
targeted education efforts both on- and offline.
|
['H. Andrew Schwartz', 'Jennifer Wortman Vaughan', 'Rediet Abebe', 'Shawndra Hill', 'Peter M. Small']
|
2018-06-14
| null | null | null | null |
['misconceptions']
|
['miscellaneous']
|
[-7.19975978e-02 6.83865622e-02 -8.28636587e-01 6.03568107e-02
-7.09520042e-01 -7.47039020e-01 4.64565784e-01 1.24207580e+00
-5.06521940e-01 4.16407198e-01 1.29785180e+00 -9.97022331e-01
-3.64138454e-01 -8.24316502e-01 -2.98764467e-01 -3.40220690e-01
1.16024971e-01 5.31720579e-01 -2.33036295e-01 -4.78267610e-01
4.38803911e-01 2.01731473e-01 -9.02441561e-01 2.00517163e-01
1.21240687e+00 7.32157081e-02 1.78848743e-01 2.80845672e-01
-4.41543698e-01 3.00822198e-01 -4.58712548e-01 -4.11355972e-01
-1.97059304e-01 -4.65599418e-01 -6.37098670e-01 -5.40258348e-01
2.44132698e-01 -8.35944712e-01 -2.70848960e-01 9.73672271e-01
6.52040839e-01 -6.72102094e-01 4.05642807e-01 -4.86387044e-01
-5.76524913e-01 3.41476679e-01 -4.49126244e-01 5.44294596e-01
8.72777820e-01 1.73324317e-01 3.55954081e-01 -3.57860833e-01
9.06477094e-01 1.41908360e+00 9.35114384e-01 3.81649941e-01
-1.10575223e+00 -6.84430897e-01 -2.81712234e-01 -2.57900685e-01
-7.52808869e-01 -5.76465666e-01 -2.60443743e-02 -9.69741881e-01
6.06601298e-01 5.80900550e-01 8.31361771e-01 1.10149193e+00
7.78981745e-02 4.58837636e-02 9.96301115e-01 -2.69840717e-01
-3.21771681e-01 3.98967505e-01 -6.70870766e-02 6.08116329e-01
7.93245196e-01 -3.26835185e-01 -3.56606901e-01 -9.64086294e-01
4.68579203e-01 7.47849405e-01 -5.63072145e-01 2.81854063e-01
-1.11689293e+00 1.09366930e+00 2.43926153e-01 4.15056974e-01
-4.84636992e-01 -8.00958514e-01 3.73887748e-01 9.49043036e-02
7.97506809e-01 4.32257384e-01 -5.96071899e-01 -1.63925767e-01
-6.30739689e-01 1.48493677e-01 1.26922345e+00 1.58366665e-01
5.35430908e-01 -5.66127479e-01 -3.09368409e-02 8.67179275e-01
4.76386845e-01 8.40510130e-01 -1.94233563e-02 -5.11673987e-01
4.44849849e-01 7.33385384e-01 2.05828488e-01 -1.60324085e+00
-5.44575930e-01 -4.11847651e-01 -6.59332037e-01 -7.64746964e-01
5.14502585e-01 -5.99398196e-01 -7.14561224e-01 1.51927185e+00
7.06260562e-01 -6.52147710e-01 -2.47143283e-01 6.59192979e-01
8.94653201e-01 6.75147355e-01 4.40056592e-01 -2.08633900e-01
1.64214087e+00 -1.35395139e-01 -7.99561560e-01 -2.19041482e-01
4.56887245e-01 -8.84073853e-01 5.50141037e-01 -1.46142080e-01
-1.08223855e+00 2.43044063e-01 -1.65115967e-01 6.09133057e-02
-7.62589633e-01 -5.08753359e-01 3.66583616e-01 8.34427893e-01
-1.19326496e+00 3.26351553e-01 -8.98745596e-01 -1.28777444e+00
3.42736691e-01 -3.63789685e-02 -1.35895044e-01 -4.89454091e-01
-1.05250084e+00 1.06252420e+00 4.41171825e-02 -4.11152005e-01
-5.73155105e-01 -1.00368702e+00 -8.36534917e-01 -3.88496183e-02
5.87646008e-01 -6.64978921e-01 8.06383073e-01 -1.53962389e-01
-3.09423000e-01 5.81493437e-01 -3.03240657e-01 1.10092580e-01
1.36191413e-01 -1.95145711e-01 -5.67870378e-01 3.89126033e-01
6.92419887e-01 4.75659966e-01 1.94958478e-01 -5.89795232e-01
-6.86458170e-01 -7.73872495e-01 -3.79542947e-01 7.85582885e-02
-7.52106428e-01 6.75226152e-01 -4.96189475e-01 -5.55255651e-01
-8.49752948e-02 -2.15606600e-01 -4.01433170e-01 -2.28451174e-02
-1.39504403e-01 -6.40850887e-02 5.60918868e-01 -1.51029623e+00
1.49373078e+00 -1.67655778e+00 -1.34426892e-01 5.73448874e-02
3.94988179e-01 1.16694868e-01 1.10227756e-01 1.29894292e+00
7.24239647e-01 7.66864240e-01 1.52531505e-01 6.03696465e-01
-2.12995052e-01 2.08021551e-01 -1.81035995e-01 5.47962844e-01
2.67607599e-01 6.61783814e-01 -1.45577204e+00 -5.22505105e-01
8.52513686e-02 1.06793463e+00 -8.13703060e-01 -6.75145686e-02
-9.45563316e-02 5.58876753e-01 -7.91000009e-01 1.01845849e+00
4.39487219e-01 -6.26445889e-01 5.40898085e-01 5.20318449e-02
-5.87167323e-01 1.08213258e+00 -2.97981590e-01 9.12161112e-01
-2.21685052e-01 3.88918191e-01 5.94150484e-01 -4.29679126e-01
2.48943046e-01 4.34569478e-01 5.79222679e-01 -8.81084085e-01
-1.49745151e-01 9.29875150e-02 5.82139567e-03 -8.21631551e-01
2.20534548e-01 2.22257182e-01 -8.48209560e-02 6.77808166e-01
-3.18976969e-01 3.81955892e-01 -1.20366834e-01 1.96434930e-01
9.31264997e-01 -4.16260839e-01 3.28517556e-01 -8.54330301e-01
-1.52638271e-01 5.16902268e-01 4.44513559e-01 8.18656385e-01
2.02591553e-01 -5.30662686e-02 5.53737342e-01 -4.81946200e-01
-8.05480242e-01 -6.61056757e-01 -4.34857398e-01 1.24995029e+00
-4.19260979e-01 -4.78518277e-01 -4.61745977e-01 -2.21379310e-01
-4.29255404e-02 4.74288464e-02 -7.03440964e-01 5.04381180e-01
-2.77565151e-01 -1.05209100e+00 2.31334493e-01 -3.40269655e-01
5.80130816e-01 -6.57376885e-01 -1.02569413e+00 3.67452741e-01
-4.83067602e-01 -6.77780211e-01 -4.11918402e-01 -3.75525892e-01
-7.37635076e-01 -1.36867893e+00 -1.22517240e+00 -5.27317703e-01
7.51149118e-01 2.22117975e-01 1.08984935e+00 3.08636159e-01
-5.45336425e-01 4.23414141e-01 -1.56369090e-01 -4.03265178e-01
-5.82620084e-01 4.54817116e-02 -1.37853310e-01 -7.74000823e-01
3.98790568e-01 2.21854135e-01 -1.01683521e+00 5.48700243e-02
-1.02077198e+00 -1.97158903e-01 5.21493673e-01 4.09872949e-01
-6.43565357e-02 -2.27719307e-01 5.02844751e-01 -1.03662074e+00
7.84562051e-01 -1.65511739e+00 -3.81064057e-01 2.33789816e-01
-5.27054727e-01 -3.76731873e-01 -3.05964202e-02 -3.21884453e-01
-6.97382629e-01 -8.73597503e-01 -3.62156093e-01 9.54066217e-01
-4.63922583e-02 1.12487423e+00 3.48607987e-01 2.47334726e-02
5.52544773e-01 -8.47508479e-03 2.24345773e-01 -9.86857295e-01
-2.56768614e-01 8.75702143e-01 -3.97870056e-02 -2.76299268e-01
1.56514496e-01 5.43675244e-01 -6.92398369e-01 -1.30033803e+00
-6.28292859e-01 -7.59967983e-01 5.55470467e-01 1.20362103e-01
7.78863966e-01 -8.91233027e-01 -7.67305613e-01 3.35621744e-01
-8.24392557e-01 -2.25281432e-01 4.61255401e-01 4.62684333e-01
7.31932163e-01 2.25120515e-01 -9.31609631e-01 -7.10598528e-01
-5.66129863e-01 -9.25057769e-01 6.89837158e-01 2.30410427e-01
-6.06794417e-01 -1.36141717e+00 5.12117028e-01 4.77839142e-01
1.20253849e+00 3.89637679e-01 1.14838123e+00 -6.79169178e-01
-4.30166088e-02 2.25955859e-01 -5.75275302e-01 -6.39756858e-01
7.91049540e-01 -2.98590124e-01 -2.10710108e-01 -4.41736579e-01
4.74923253e-02 -2.88149863e-01 5.43849051e-01 3.92683536e-01
4.88150001e-01 -1.48591328e+00 -7.11666882e-01 1.35684714e-01
1.58604574e+00 2.04692736e-01 1.65705591e-01 3.09373230e-01
3.35090727e-01 1.08088362e+00 -1.55371174e-01 6.33483708e-01
7.97248960e-01 4.37498689e-01 -1.25922794e-02 -2.78669268e-01
2.88077623e-01 -5.22687137e-01 2.04227403e-01 5.37938416e-01
1.20317183e-01 1.55405020e-02 -1.68670833e+00 1.03636909e+00
-1.11611474e+00 -8.14526558e-01 1.54621989e-01 2.07825375e+00
1.24537349e+00 -1.60666049e-01 3.42138052e-01 -5.46418905e-01
5.18922687e-01 -1.21058211e-01 -3.35282654e-01 -2.87135363e-01
1.72096878e-01 2.81445354e-01 6.11273825e-01 6.35810673e-01
-5.80781698e-01 4.96925145e-01 7.16577053e+00 2.76502579e-01
-1.26884854e+00 1.11277387e-01 8.43555987e-01 3.58474821e-01
-9.19744074e-01 -2.78804779e-01 -6.34076118e-01 4.34161127e-01
8.48071218e-01 -2.94740926e-02 3.55972111e-01 3.21641117e-01
5.09808302e-01 -1.82780698e-01 -4.90192533e-01 3.03009778e-01
-1.45328194e-02 -1.55003142e+00 -5.19773886e-02 4.11918283e-01
8.46324801e-01 4.88680094e-01 2.31734142e-01 -2.64099002e-01
4.70838726e-01 -1.04760909e+00 -1.32776694e-02 1.67627081e-01
8.03412616e-01 -2.80638635e-01 4.64529037e-01 3.58016193e-01
-6.94191813e-01 1.12719342e-01 -1.19356543e-01 -2.00405195e-01
-6.90227076e-02 4.81200457e-01 -9.71401811e-01 1.19224511e-01
7.95575559e-01 5.06325960e-01 -5.48131108e-01 1.17718649e+00
4.31161255e-01 6.06104672e-01 -3.28636676e-01 -7.09329918e-02
3.99061233e-01 -9.80071723e-02 2.17788890e-01 1.38132393e+00
4.42941040e-01 4.18593585e-01 -6.70981407e-02 6.50413036e-01
1.53384268e-01 2.85164833e-01 -8.21617305e-01 -8.34757507e-01
5.87547719e-01 8.89663160e-01 -7.21393585e-01 -2.20496222e-01
-5.83959877e-01 3.25356275e-01 -5.69497831e-02 5.27249634e-01
1.64439306e-02 -1.07309014e-01 6.58023894e-01 5.46459734e-01
-1.61762848e-01 8.16085562e-02 2.76519537e-01 -8.38063717e-01
-5.51777422e-01 -1.44860387e+00 7.44510770e-01 1.64002050e-02
-9.44619536e-01 1.25776371e-02 1.17540173e-01 -4.72001731e-02
-2.91384727e-01 -4.88548577e-02 -2.42467731e-01 9.45885897e-01
-1.36236548e+00 -6.99935496e-01 1.57516092e-01 2.34625503e-01
3.59274864e-01 5.18631160e-01 7.41020381e-01 4.11322057e-01
-3.67790669e-01 3.59332636e-02 1.24422394e-01 1.37575015e-01
4.66405243e-01 -4.74087328e-01 3.78651381e-01 1.05275966e-01
-5.77775419e-01 1.03090262e+00 7.44384766e-01 -1.28303468e+00
-1.52445340e+00 -7.04943359e-01 1.49361503e+00 -3.98204267e-01
7.42907405e-01 -7.02468157e-02 -9.95017648e-01 5.19696236e-01
5.14739335e-01 -1.11045372e+00 1.01046157e+00 3.90977025e-01
-2.45293766e-01 4.51276302e-01 -1.54071486e+00 6.99933708e-01
7.15152442e-01 -6.13917649e-01 -5.42823374e-01 7.26441383e-01
5.63707948e-01 -8.94820616e-02 -8.32215190e-01 2.07031280e-01
8.02624762e-01 -7.71103203e-01 1.00615954e+00 -6.75185382e-01
3.05121213e-01 7.13557541e-01 1.23511694e-01 -9.48427975e-01
-6.40408471e-02 -7.06300616e-01 4.06147063e-01 8.53817046e-01
3.11856508e-01 -9.14445579e-01 5.55982590e-01 7.89008915e-01
3.14165980e-01 -6.06720686e-01 -4.80801940e-01 1.51131272e-01
9.73459780e-02 1.38333946e-01 5.91477990e-01 1.49334931e+00
2.34705836e-01 -1.74118176e-01 1.19820621e-03 1.83511451e-01
5.51846862e-01 -9.38262343e-02 2.78292567e-01 -9.93686378e-01
-1.24972634e-01 -3.84332448e-01 4.22102243e-01 -5.79848826e-01
-7.73962796e-01 -1.85513973e-01 -5.84876120e-01 -1.98781657e+00
6.90415382e-01 -5.05972683e-01 1.91839442e-01 4.96446580e-01
-1.58770964e-01 -9.16334093e-02 -1.69816434e-01 3.14507842e-01
-2.78326780e-01 -2.05148354e-01 9.81024086e-01 -2.46468652e-02
-4.17760432e-01 -3.11467826e-01 -1.28788710e+00 6.65781617e-01
8.56404901e-01 -5.46611130e-01 9.14633870e-02 -8.21890473e-01
9.02149916e-01 3.98310959e-01 2.87182719e-01 -5.21812320e-01
2.70825684e-01 -6.18822038e-01 3.03381830e-01 -7.68053114e-01
-5.91122322e-02 -7.18081236e-01 5.06197691e-01 1.42794383e+00
-7.67504275e-02 2.86505550e-01 3.01668525e-01 9.37421694e-02
1.64967626e-01 1.17623024e-01 -6.17282763e-02 -4.81023788e-01
2.25401342e-01 1.89572990e-01 -4.48031396e-01 5.42739630e-01
2.25036830e-01 2.48688504e-01 -9.43259835e-01 -6.03745759e-01
-2.80620903e-01 5.00493586e-01 5.77284157e-01 7.27585033e-02
5.09682775e-01 -7.19498217e-01 -9.97489810e-01 2.08430812e-01
-1.15314543e-01 -3.42618138e-01 2.49646276e-01 8.11886370e-01
-6.98766887e-01 8.21823776e-01 -2.93882817e-01 -2.54816823e-02
-8.75399172e-01 4.01846468e-01 -1.26505554e-01 -1.29930839e-01
-1.64716244e-01 2.61666596e-01 1.20158754e-01 -2.43310779e-01
3.36720765e-01 -1.22525014e-01 -2.56057411e-01 4.49832827e-01
8.63774955e-01 8.21902454e-01 -4.58313346e-01 -4.45444018e-01
-5.18025875e-01 3.39197695e-01 -1.65659070e-01 -7.51994848e-02
1.57229698e+00 -2.78456300e-01 -5.01804829e-01 -5.89634404e-02
1.09373450e+00 2.70930171e-01 -5.90904415e-01 -4.35098857e-02
6.50597587e-02 -6.51002645e-01 -2.58199930e-01 -9.98888195e-01
-6.64086163e-01 8.03878844e-01 3.06672782e-01 7.20158517e-01
9.06507254e-01 4.01504427e-01 7.94680893e-01 1.75407320e-01
-1.14314733e-02 -7.14921534e-01 -1.28232837e-01 6.77369116e-03
8.04058552e-01 -1.20815182e+00 -3.71418446e-02 -1.41066670e-01
5.28779738e-02 7.38277018e-01 5.88179007e-02 8.76991510e-01
7.13046968e-01 -1.36252224e-01 2.42260814e-01 -5.50635040e-01
-6.40104294e-01 -1.04187898e-01 8.24062377e-02 6.69530272e-01
4.04056787e-01 -1.53443769e-01 -7.21776843e-01 5.28548583e-02
3.30748379e-01 -1.92951083e-01 2.60867864e-01 7.21738577e-01
-7.13021338e-01 -1.09960842e+00 -6.30254328e-01 7.99462020e-01
-1.38390028e+00 -5.32613277e-01 -6.15993619e-01 7.07959294e-01
1.51567519e-01 9.03430104e-01 1.70728877e-01 8.51968750e-02
-1.82121843e-01 -2.97550112e-01 -1.02209888e-01 -6.73227847e-01
-8.24184358e-01 3.10186446e-01 3.43962252e-01 -1.18337624e-01
-3.38865608e-01 -6.03186548e-01 -5.25822222e-01 -5.94169617e-01
-3.11115710e-03 3.88453603e-01 9.78279591e-01 5.78280687e-01
6.58010542e-01 -1.89174980e-01 2.49405265e-01 -1.19781524e-01
-1.81473181e-01 -6.80991709e-01 2.96070158e-01 2.04656065e-01
6.94405437e-01 5.45578136e-04 -8.55896026e-02 -4.74135816e-01]
|
[8.460418701171875, 9.515969276428223]
|
19522628-b863-4e6a-a13d-6b3cf912b5fe
|
deep-residual-3d-u-net-for-joint-segmentation
|
2006.14215
| null |
https://arxiv.org/abs/2006.14215v2
|
https://arxiv.org/pdf/2006.14215v2.pdf
|
Deep Residual 3D U-Net for Joint Segmentation and Texture Classification of Nodules in Lung
|
In this work we present a method for lung nodules segmentation, their texture classification and subsequent follow-up recommendation from the CT image of lung. Our method consists of neural network model based on popular U-Net architecture family but modified for the joint nodule segmentation and its texture classification tasks and an ensemble-based model for the follow-up recommendation. This solution was evaluated within the LNDb medical imaging challenge and produced the best nodule segmentation result on the final leaderboard.
|
['Alexandr G. Rassadin']
|
2020-06-25
| null | null | null | null |
['texture-classification']
|
['computer-vision']
|
[ 1.32317409e-01 6.34526193e-01 -7.23895311e-01 -6.01426184e-01
-8.49951088e-01 -2.27468222e-01 2.85831958e-01 -1.51746944e-01
4.36341390e-02 4.81224507e-01 6.85057521e-01 -9.46704209e-01
-6.35939479e-01 -1.00372326e+00 -2.67871320e-01 -6.12489760e-01
1.92470830e-02 1.22735870e+00 8.66820157e-01 4.32116678e-03
9.74240676e-02 4.45756942e-01 -9.03273880e-01 9.35708702e-01
2.86163747e-01 1.49977827e+00 2.65001893e-01 1.11015677e+00
-1.23904511e-01 1.21018100e+00 -1.22542074e-02 1.51350990e-01
3.29995871e-01 -2.53401160e-01 -1.51295149e+00 1.10218517e-01
6.90974355e-01 -4.30292726e-01 -4.52437341e-01 4.52592254e-01
6.45311236e-01 1.34477690e-01 1.21328568e+00 -4.83039528e-01
-3.24082792e-01 9.23975945e-01 2.82993726e-02 6.00684643e-01
-5.52420735e-01 -4.05137211e-01 8.93249214e-01 -2.35308066e-01
9.33137178e-01 5.93268156e-01 1.00548208e+00 8.37606847e-01
-6.45310402e-01 -7.96690211e-02 -3.78080010e-01 3.96541320e-02
-6.46768272e-01 1.22394644e-01 -6.53112829e-02 -5.11591911e-01
9.70282853e-01 7.98422158e-01 6.68307304e-01 1.06797731e+00
5.55286705e-01 7.62521565e-01 9.94846284e-01 6.39418215e-02
-1.56995118e-01 5.68652824e-02 5.97668767e-01 1.18337679e+00
9.09835026e-02 1.14600666e-01 3.68742585e-01 -2.44156569e-01
1.32939923e+00 2.63447404e-01 1.02110557e-01 -9.10062790e-02
-1.32724631e+00 7.88417161e-01 9.73447740e-01 6.84941947e-01
-3.05846870e-01 3.82329643e-01 4.78819191e-01 2.78026491e-01
8.21906149e-01 1.30576342e-01 -2.51971990e-01 6.66569054e-01
-1.14931226e+00 -1.89146444e-01 1.07272506e+00 5.26971459e-01
1.00073896e-01 -4.90043789e-01 -1.36314094e+00 6.33093417e-01
6.36534929e-01 -3.29361372e-02 1.08645606e+00 -6.27109587e-01
-2.76483387e-01 2.03864381e-01 -2.71773964e-01 -3.86663318e-01
-7.11581945e-01 -8.16973269e-01 -1.07006085e+00 -1.67454481e-02
3.00947219e-01 -4.11400080e-01 -1.57891166e+00 5.73564112e-01
2.74346620e-01 6.31643116e-01 -1.97945446e-01 7.62172699e-01
1.66942644e+00 6.79789782e-02 5.64880818e-02 2.40874350e-01
1.49730766e+00 -1.89510167e+00 -3.13060790e-01 4.45229381e-01
7.42248535e-01 -6.61038995e-01 1.61610872e-01 1.72761112e-01
-7.03634679e-01 -7.44306266e-01 -6.80267632e-01 2.58791327e-01
-9.11502466e-02 3.63238275e-01 7.24825263e-01 8.95424545e-01
-1.62731993e+00 1.10268104e+00 -1.01891541e+00 -8.45393181e-01
7.77363539e-01 8.29877019e-01 -5.90798482e-02 2.24410161e-01
-5.17961979e-01 8.69122922e-01 4.04439956e-01 -2.69028932e-01
-7.73041368e-01 -5.94936430e-01 1.16999283e-01 -1.50141433e-01
3.77975166e-01 -1.32041836e+00 1.81805813e+00 -7.92263746e-01
-1.68624163e+00 8.88599694e-01 1.54527336e-01 -1.01131976e+00
6.60436034e-01 4.27965045e-01 -2.65503019e-01 -1.17560588e-01
-5.33777699e-02 7.44192541e-01 8.86922419e-01 -5.91901064e-01
-1.18327391e+00 1.06149837e-01 -6.13147497e-01 1.30363956e-01
2.78156865e-02 -5.04452586e-01 -5.60875893e-01 -8.37330282e-01
2.59279191e-01 -1.03158081e+00 -8.24196815e-01 -4.52771157e-01
-4.31352586e-01 -5.68852544e-01 8.61662030e-01 -7.12591469e-01
1.05810511e+00 -1.50530159e+00 -1.90744624e-01 7.74456620e-01
5.72616935e-01 -1.40392035e-01 7.63133839e-02 -4.71616179e-01
-1.33592889e-01 5.18429101e-01 4.78546232e-01 1.00769967e-01
-2.40099907e-01 4.17219311e-01 3.72915387e-01 1.03385709e-01
-2.32948318e-01 1.29007900e+00 -4.60938245e-01 -7.55458772e-01
1.32448778e-01 -1.51625285e-02 -6.64597690e-01 4.23177272e-01
-3.89609635e-01 6.91778481e-01 -9.28370059e-01 6.81799054e-01
2.27231771e-01 -5.21519482e-01 1.84281737e-01 -6.19916379e-01
1.67787835e-01 1.24883711e-01 -6.13222420e-01 1.40845573e+00
-1.56719536e-01 1.15508467e-01 -1.26054645e-01 -7.69085169e-01
7.75006890e-01 7.07942069e-01 1.11152208e+00 -2.43002012e-01
4.56259221e-01 4.63043332e-01 2.84231722e-01 -6.42581642e-01
1.49947822e-01 7.53354132e-02 4.08535242e-01 7.12225795e-01
6.77231327e-02 -1.03225686e-01 2.25551888e-01 -2.03263357e-01
1.60312247e+00 -3.47830355e-02 2.00690493e-01 -6.11152232e-01
4.75955606e-01 4.09913540e-01 -1.44888405e-02 1.25474763e+00
-4.05012578e-01 7.98056901e-01 3.47315401e-01 -7.07250535e-01
-9.35747981e-01 -8.35042834e-01 -3.25853646e-01 1.25053263e+00
-3.94772738e-01 1.29764289e-01 -6.28141344e-01 -1.71733999e+00
-1.82787571e-02 1.39941975e-01 -1.16752589e+00 4.63690668e-01
-5.08479893e-01 -7.50036359e-01 2.28077948e-01 5.99543154e-01
4.25912827e-01 -1.22642207e+00 -2.66624004e-01 3.91540170e-01
-1.33725896e-01 -3.69379044e-01 -6.03068411e-01 7.20396996e-01
-1.42189384e+00 -1.24370027e+00 -1.07226729e+00 -1.21993029e+00
7.38477528e-01 -1.05531581e-01 1.42314959e+00 3.54941338e-01
-3.15355420e-01 4.51493680e-01 -2.87509173e-01 -4.68395561e-01
-6.54431283e-01 4.22656208e-01 -6.80172801e-01 -1.52409732e-01
-4.58063744e-03 -1.25113860e-01 -8.47443044e-01 4.24596429e-01
-7.85988450e-01 6.47540437e-04 8.42116237e-01 1.11163247e+00
1.10369289e+00 2.57423967e-02 2.70731688e-01 -1.68890190e+00
4.53969657e-01 -7.64115870e-01 -6.94729015e-02 4.14984614e-01
-4.30229485e-01 -1.44410089e-01 -6.58052042e-02 -6.77579939e-02
-1.05539560e+00 2.15634450e-01 -4.23492193e-01 -1.33347780e-01
-3.46778095e-01 4.37916607e-01 1.02351844e+00 -5.09060085e-01
9.36750531e-01 -4.30962145e-01 3.76739763e-02 -4.23489302e-01
6.92000315e-02 7.39036858e-01 3.14637274e-01 -6.14563301e-02
2.78017938e-01 5.12218118e-01 2.36945704e-01 -2.87908137e-01
-1.27186942e+00 -9.00910974e-01 -7.13850439e-01 -3.19875717e-01
1.13068914e+00 -4.10903931e-01 -4.30465005e-02 -2.19555255e-02
-5.84793091e-01 -3.84443581e-01 -4.97309595e-01 5.19832730e-01
-5.19633174e-01 -4.78290915e-02 -7.91070819e-01 4.32503968e-02
-9.90991294e-01 -1.18681467e+00 6.81948543e-01 2.28723764e-01
-1.49084941e-01 -9.93373692e-01 2.84971207e-01 4.89161849e-01
1.15191650e+00 3.16778049e-02 9.13295686e-01 -1.36295640e+00
-4.78162020e-01 -2.72481084e-01 -4.92224663e-01 2.53942847e-01
2.18313381e-01 -5.12215905e-02 -6.77686930e-01 -1.58019771e-03
-8.76981318e-02 -2.21068621e-01 1.34336722e+00 1.21182203e+00
1.56484044e+00 1.13398448e-01 -8.44173849e-01 6.27522230e-01
1.38249791e+00 6.60889298e-02 2.66245812e-01 6.53549731e-02
6.83399618e-01 9.85042527e-02 9.57807973e-02 -5.40844649e-02
-1.09015323e-01 -1.11445732e-01 6.62619948e-01 -2.37900123e-01
-6.40245378e-01 3.95217478e-01 -4.46587086e-01 7.01161027e-01
-4.17181551e-01 -4.61093575e-01 -1.05944896e+00 3.49893451e-01
-1.91013837e+00 -5.70174038e-01 -2.83455908e-01 1.63083053e+00
9.07774493e-02 1.60478473e-01 1.87931433e-01 -6.61510408e-01
5.34284174e-01 -2.70030826e-01 -3.06955695e-01 -2.09032834e-01
3.86564851e-01 6.81263685e-01 9.01041746e-01 7.07448572e-02
-1.76983547e+00 4.70631301e-01 7.58888817e+00 1.24845004e+00
-1.11686504e+00 6.72400832e-01 1.47584891e+00 1.35746166e-01
-9.10784230e-02 -5.62966883e-01 -5.41552007e-01 -2.07211643e-01
9.30321455e-01 3.66461158e-01 -1.44782946e-01 7.03463137e-01
-1.13757759e-01 -2.63949960e-01 -8.50921631e-01 3.76003593e-01
-1.61782622e-01 -2.04545808e+00 -1.21289261e-01 7.54748238e-03
1.22979283e+00 1.18424678e+00 4.41875570e-02 3.73465836e-01
5.15273809e-01 -1.12822914e+00 -1.17353253e-01 8.12887907e-01
6.99906230e-01 -3.14600289e-01 1.30824494e+00 -1.45179639e-02
-9.10899699e-01 4.59848754e-02 -3.84930402e-01 7.49902844e-01
-5.39178014e-01 3.09847564e-01 -1.59127998e+00 7.43397474e-01
6.80592954e-01 8.44288111e-01 -8.70637119e-01 1.35882461e+00
5.72045863e-01 1.17875266e+00 -4.43921179e-01 -3.93684834e-01
6.50084615e-01 2.77977902e-02 3.22545707e-01 1.23507130e+00
7.36360252e-01 -8.14991631e-03 5.95429957e-01 3.10833186e-01
-1.32582158e-01 5.47457039e-01 -2.12936237e-01 9.51095000e-02
-3.68619233e-01 1.86318970e+00 -1.59855711e+00 -8.33387733e-01
-3.20956737e-01 6.83636427e-01 -1.65591575e-02 -9.84949544e-02
-6.37601018e-01 5.65115988e-01 -5.10601938e-01 2.86115736e-01
6.99302495e-01 8.59009922e-01 -3.32982898e-01 -6.89386427e-01
-9.20199037e-01 -5.00006676e-01 8.28089952e-01 -3.70699048e-01
-1.68276405e+00 1.08860636e+00 -3.95325929e-01 -1.22867548e+00
-8.10883492e-02 -9.36526775e-01 -1.09153533e+00 5.88228941e-01
-9.96062636e-01 -1.38312495e+00 -3.97361904e-01 5.07834792e-01
3.19877714e-01 -5.52315056e-01 8.95974040e-01 1.30594432e-01
-3.58024031e-01 1.99955150e-01 5.14197826e-01 -1.19794859e-02
8.08680296e-01 -1.75964189e+00 2.84064701e-03 -1.78004444e-01
-3.47363949e-02 -3.74793857e-01 1.25331387e-01 -9.78277981e-01
-8.27087402e-01 -1.71434200e+00 5.84077597e-01 -6.15230501e-01
7.64884949e-01 6.97062850e-01 -3.12028587e-01 7.76999772e-01
4.64939415e-01 3.97874802e-01 5.83245456e-01 -4.96047139e-02
5.08536875e-01 2.15133429e-01 -1.39521849e+00 3.46713543e-01
5.11567712e-01 5.44892028e-02 -2.47049734e-01 9.24901247e-01
2.60686636e-01 -7.36898541e-01 -1.34641850e+00 8.16061437e-01
5.41247487e-01 -8.71112823e-01 7.89282918e-01 -7.08002687e-01
4.83224988e-01 1.57252789e-01 -4.34321351e-02 -9.28045571e-01
-1.21085548e+00 -1.58961445e-01 1.52178645e-01 4.17407513e-01
7.54712343e-01 6.87799184e-03 1.68534362e+00 -1.41056469e-02
-3.24286550e-01 -1.29482234e+00 -6.71627998e-01 -1.66776136e-01
1.19088657e-01 -2.96358764e-01 1.20411650e-03 4.20635790e-01
-6.48336709e-01 2.00821817e-01 -6.48120418e-02 -3.05598140e-01
4.47613299e-01 1.15932524e-01 4.10970207e-03 -1.27250063e+00
-7.24767864e-01 -9.83054817e-01 -1.11323558e-01 -5.78374386e-01
-4.16948855e-01 -1.56179941e+00 9.09458697e-02 -1.96442068e+00
5.30416131e-01 -4.35298681e-01 -8.23631704e-01 2.63013154e-01
1.23603515e-01 6.81211293e-01 -3.55866015e-01 1.73488244e-01
-6.62210107e-01 -4.64703828e-01 1.67868412e+00 -3.91273081e-01
1.27600431e-01 9.01145279e-01 -4.10820782e-01 7.29723036e-01
7.72102654e-01 -4.95482802e-01 -2.38180682e-01 -1.01189926e-01
-2.93844193e-01 4.56710935e-01 2.00909942e-01 -1.22005904e+00
1.92467570e-01 6.90251067e-02 6.89463198e-01 -1.23759973e+00
-3.20249707e-01 -1.03142929e+00 2.61547506e-01 1.06684673e+00
-4.82441187e-01 -4.60855514e-01 -1.13115191e-01 7.56285071e-01
3.53727788e-02 -6.45543456e-01 7.42443025e-01 -6.01646125e-01
-4.31521386e-01 8.86273146e-01 -7.86363780e-01 -6.02888823e-01
9.81381595e-01 -1.24339633e-01 3.77728269e-02 -9.25580934e-02
-1.71462381e+00 2.56704479e-01 -3.80415499e-01 -5.89114940e-03
1.71706468e-01 -1.41830349e+00 -1.09910464e+00 1.73750296e-02
-2.61006385e-01 3.97987803e-03 2.65215516e-01 1.41160905e+00
-9.45558727e-01 6.83685541e-01 -2.64374465e-01 -6.98173702e-01
-1.49394929e+00 -2.58520581e-02 1.09170496e+00 -1.37098241e+00
-4.36811417e-01 1.38446689e+00 -7.18174875e-02 -6.40499294e-01
4.41170633e-01 -6.68105185e-01 -7.71304846e-01 1.59026869e-02
1.47393560e-02 3.95844251e-01 5.21687508e-01 -1.60565421e-01
1.51536793e-01 1.79921776e-01 -3.09449464e-01 2.99576253e-01
1.24577999e+00 1.01157293e-01 -5.16107261e-01 1.69908702e-01
9.97915208e-01 -4.53242689e-01 -4.93125677e-01 -4.81679887e-01
1.67857692e-01 5.00867903e-01 7.48275757e-01 -1.27238464e+00
-1.51498723e+00 2.60574311e-01 1.25098777e+00 3.64468783e-01
8.00859034e-01 7.12812617e-02 8.34492326e-01 8.68551433e-01
-2.54494399e-01 -7.93694913e-01 -1.92011986e-02 7.55474031e-01
5.48100471e-01 -1.61717820e+00 4.10039388e-02 -4.40566659e-01
-3.53511214e-01 1.57433891e+00 7.07180738e-01 -6.55431211e-01
1.48425078e+00 2.75740534e-01 3.08289021e-01 -3.92324001e-01
-1.00649285e+00 -3.29747766e-01 1.07190800e+00 4.70172882e-01
9.70958948e-01 4.39322203e-01 -7.69272670e-02 7.87662923e-01
1.43141404e-01 2.45030746e-01 -1.51706403e-02 7.00690031e-01
-7.95916021e-01 -1.09454882e+00 -3.51455331e-01 1.96560800e+00
-9.40176785e-01 -1.01996116e-01 -4.24256504e-01 5.62270045e-01
3.13069969e-01 4.73301470e-01 8.52527693e-02 -3.98169905e-01
3.70846950e-02 1.63684800e-01 5.50212741e-01 -8.70352328e-01
-1.46056998e+00 8.08949709e-01 4.55993235e-01 -5.98958015e-01
-5.00500560e-01 -3.95739168e-01 -9.27659214e-01 -1.08958691e-01
-5.32762110e-01 1.28303543e-01 3.52525353e-01 8.73295009e-01
-1.04883440e-01 1.33945429e+00 5.77356398e-01 -7.92124867e-01
-4.99362469e-01 -1.43112779e+00 -6.82641923e-01 1.57917872e-01
4.28356938e-02 -3.45151685e-02 1.68879658e-01 -1.72355488e-01]
|
[15.383230209350586, -2.145216703414917]
|
9bc79ded-bfc0-4727-87ef-8ce1011f2470
|
a-reference-less-quality-metric-for-automatic
|
2306.13114
| null |
https://arxiv.org/abs/2306.13114v1
|
https://arxiv.org/pdf/2306.13114v1.pdf
|
A Reference-less Quality Metric for Automatic Speech Recognition via Contrastive-Learning of a Multi-Language Model with Self-Supervision
|
The common standard for quality evaluation of automatic speech recognition (ASR) systems is reference-based metrics such as the Word Error Rate (WER), computed using manual ground-truth transcriptions that are time-consuming and expensive to obtain. This work proposes a multi-language referenceless quality metric, which allows comparing the performance of different ASR models on a speech dataset without ground truth transcriptions. To estimate the quality of ASR hypotheses, a pre-trained language model (LM) is fine-tuned with contrastive learning in a self-supervised learning manner. In experiments conducted on several unseen test datasets consisting of outputs from top commercial ASR engines in various languages, the proposed referenceless metric obtains a much higher correlation with WER scores and their ranks than the perplexity metric from the state-of-art multi-lingual LM in all experiments, and also reduces WER by more than $7\%$ when used for ensembling hypotheses. The fine-tuned model and experiments are made available for the reproducibility: https://github.com/aixplain/NoRefER
|
['Golara Javadi', 'Mohamed Al-Badrashiny', 'Ahmet Gunduz', 'Thiago Ferreira', 'Kamer Ali Yuksel']
|
2023-06-21
| null | null | null | null |
['contrastive-learning', 'self-supervised-learning', 'learning-to-rank', 'contrastive-learning', 'learning-to-rank', 'automatic-speech-recognition']
|
['computer-vision', 'computer-vision', 'graphs', 'methodology', 'miscellaneous', 'speech']
|
[-7.10207671e-02 -9.43816975e-02 8.45807837e-04 -4.84339863e-01
-1.77667952e+00 -5.12138903e-01 4.33927387e-01 4.42182198e-02
-5.80226302e-01 7.05607474e-01 3.71839166e-01 -4.66965675e-01
2.44503036e-01 -2.30425775e-01 -5.49559712e-01 -4.74157721e-01
3.43088508e-01 5.90123832e-01 5.84057830e-02 -3.07807237e-01
3.98474075e-02 2.24646911e-01 -1.29161286e+00 1.96650848e-01
8.67879689e-01 8.72125804e-01 4.37940955e-01 7.87104487e-01
1.69207770e-02 4.39599514e-01 -1.08102286e+00 -7.42871165e-01
-2.37267073e-02 -4.99639273e-01 -8.01004171e-01 6.73942715e-02
4.50534374e-01 1.20148577e-01 -5.75143576e-01 1.09621727e+00
7.92653024e-01 1.98709026e-01 4.46615458e-01 -7.33379245e-01
-6.16083622e-01 1.02082813e+00 1.79522350e-01 4.38573241e-01
3.53206426e-01 3.06368291e-01 1.34660172e+00 -1.20576906e+00
1.71259448e-01 1.35387325e+00 2.14460611e-01 7.43085623e-01
-9.89732444e-01 -5.11838257e-01 -1.68872252e-01 2.59742200e-01
-1.74814534e+00 -1.26466775e+00 4.31912184e-01 -8.63803029e-02
1.29166913e+00 3.65250975e-01 -4.38731946e-02 1.19335949e+00
-1.57688156e-01 7.32727885e-01 9.87920702e-01 -6.37378097e-01
2.36628070e-01 1.71817347e-01 -3.81911397e-02 7.37484932e-01
-4.46178839e-02 -3.40742469e-02 -9.51097965e-01 1.11895777e-01
2.94973999e-01 -6.65189743e-01 -6.63400829e-01 3.78004223e-01
-1.30373514e+00 5.63164949e-01 -7.75750354e-02 6.27415061e-01
-2.37492830e-01 -8.60937089e-02 3.79670203e-01 5.66714644e-01
6.35942996e-01 5.17346263e-01 -6.96704626e-01 -5.98354936e-01
-1.15189409e+00 -3.43635559e-01 5.88069081e-01 9.03636694e-01
5.14809668e-01 3.93701077e-01 -4.44816321e-01 1.51480937e+00
4.92417544e-01 8.22067916e-01 9.60698783e-01 -6.64380550e-01
8.53497267e-01 3.03891957e-01 2.37876587e-02 -2.93885142e-01
1.62582144e-01 -5.30999184e-01 -5.97621024e-01 -3.73910338e-01
2.34850019e-01 2.28773262e-02 -1.00135720e+00 1.56407583e+00
-1.12666793e-01 3.22743095e-02 4.17920351e-01 7.00385571e-01
8.74858141e-01 1.03149045e+00 -7.27039799e-02 -5.06514192e-01
1.11850154e+00 -1.20191669e+00 -9.43437815e-01 -2.00618982e-01
7.91940689e-01 -1.18039882e+00 1.68937075e+00 4.53897238e-01
-1.25614250e+00 -6.65009081e-01 -1.17531121e+00 9.54081416e-02
-2.08467439e-01 6.05482101e-01 -2.53366292e-01 8.96848142e-01
-1.22198486e+00 6.32124782e-01 -7.75095284e-01 -1.55499786e-01
-1.36440575e-01 -4.22554463e-03 -4.37787890e-01 -1.78379435e-02
-1.24293125e+00 1.03490126e+00 2.65597761e-01 3.47773619e-02
-1.10948694e+00 -3.44816029e-01 -8.15894544e-01 -4.71884310e-02
1.62150294e-01 1.32899620e-02 1.59916830e+00 -6.77944899e-01
-2.14933681e+00 8.38167727e-01 -4.40263569e-01 -3.29471022e-01
2.66739070e-01 -3.38135600e-01 -9.48562264e-01 -6.19970374e-02
-2.58231908e-01 2.09523603e-01 6.60479307e-01 -9.30528939e-01
-3.90938014e-01 -7.73934051e-02 -3.15538466e-01 1.78023219e-01
-3.52619678e-01 2.61648268e-01 -5.02922416e-01 -6.06528759e-01
6.97426572e-02 -7.65970886e-01 1.70597479e-01 -7.77974308e-01
-4.40521091e-01 -5.72476268e-01 3.83503109e-01 -1.02343833e+00
1.74533808e+00 -2.11013889e+00 3.15742777e-03 -3.95651609e-02
-3.18977058e-01 7.53762722e-01 -4.30481017e-01 4.08418447e-01
7.44212344e-02 3.40311438e-01 -1.99318781e-01 -4.31948513e-01
1.09259941e-01 3.76837812e-02 -1.24292299e-01 2.45192260e-01
1.75805584e-01 6.09361470e-01 -9.16800857e-01 -3.58745903e-01
4.10284519e-01 4.13351148e-01 -1.24960737e-02 6.03214622e-01
-1.06917813e-01 1.80580243e-01 6.69724941e-02 4.79835451e-01
1.52747646e-01 -5.57495318e-02 1.97718307e-01 -8.31615329e-02
-3.30191255e-02 1.23167884e+00 -1.07217014e+00 1.74684513e+00
-9.44684863e-01 6.76863015e-01 -2.10665852e-01 -6.28667235e-01
1.22947168e+00 9.30788517e-01 -2.38245502e-01 -8.46779346e-01
-6.89525550e-05 5.68171263e-01 1.52841926e-01 -9.11948606e-02
4.78602260e-01 9.59453434e-02 7.02749938e-02 2.42459089e-01
4.75957125e-01 -4.67130005e-01 3.01356286e-01 -1.52595043e-02
1.18375444e+00 -2.89480507e-01 2.38034040e-01 -1.92408219e-01
8.92951369e-01 -2.86487073e-01 4.66695040e-01 4.49741215e-01
-1.67289734e-01 8.01186740e-01 -9.72753018e-02 1.44795343e-01
-1.21491909e+00 -1.12691081e+00 -1.59970775e-01 8.77938747e-01
-3.21693689e-01 -6.36009395e-01 -9.75792646e-01 -5.18093824e-01
-4.37546939e-01 1.09285212e+00 1.03287250e-01 -1.06192097e-01
-5.96256673e-01 -4.05382514e-01 8.95366848e-01 2.60843456e-01
3.42478037e-01 -1.09311306e+00 3.91816258e-01 2.99733460e-01
-5.78004956e-01 -1.27729297e+00 -7.38195777e-01 1.07385203e-01
-6.65847957e-01 -5.37427664e-01 -7.35604048e-01 -5.83689690e-01
3.56315285e-01 1.19199693e-01 1.18050599e+00 1.92553625e-01
2.27123469e-01 1.05547987e-01 -6.48609042e-01 -4.78102155e-02
-1.09086227e+00 2.07449928e-01 3.94382477e-01 3.99731584e-02
3.31707716e-01 -2.06442490e-01 -2.99120128e-01 6.49901152e-01
-5.44004440e-01 -3.15919131e-01 6.41288579e-01 8.39078426e-01
7.33930469e-01 -1.98926330e-01 7.54225194e-01 -4.04379934e-01
7.77106285e-01 -1.06162779e-01 -6.43196642e-01 5.87020338e-01
-7.71577299e-01 3.61954927e-01 6.12391770e-01 -2.37233505e-01
-1.01521659e+00 -3.75151306e-01 -7.25129247e-01 -4.02349055e-01
-1.44319519e-01 3.83846343e-01 -4.68813509e-01 3.60122174e-01
7.72914529e-01 3.35906506e-01 -4.42943603e-01 -7.35206425e-01
3.39538336e-01 1.43016064e+00 3.80697608e-01 -4.82357442e-01
7.85682082e-01 -5.27605295e-01 -7.15514481e-01 -1.08211350e+00
-8.84200394e-01 -6.44860625e-01 -4.75641429e-01 -1.35201409e-01
5.23224533e-01 -1.07265651e+00 -1.51944786e-01 3.77464205e-01
-1.21925092e+00 -4.60291088e-01 -3.68021093e-02 8.13236475e-01
-4.38931227e-01 3.75069737e-01 -6.59392893e-01 -8.89430642e-01
-6.64424658e-01 -1.28674638e+00 1.15580869e+00 -7.90190175e-02
-2.75001049e-01 -7.66985118e-01 1.79927751e-01 7.78649092e-01
3.71628046e-01 -7.39913881e-01 6.33556664e-01 -7.58182943e-01
-2.33611971e-01 -2.91704059e-01 1.43484756e-01 1.10085785e+00
2.29102671e-01 9.28252116e-02 -1.16453838e+00 -3.26292366e-01
-2.94704467e-01 -4.10628706e-01 5.63046217e-01 4.83752899e-02
1.08087623e+00 -4.18366998e-01 1.58344135e-01 1.81866437e-01
1.15132344e+00 3.46177965e-01 7.96046436e-01 -7.93138891e-02
4.11733210e-01 1.30893111e-01 6.60171390e-01 1.07236929e-01
-3.26642804e-02 9.23280239e-01 -1.60785034e-01 2.58793473e-01
-4.43097085e-01 -4.54420388e-01 8.61666143e-01 1.91245151e+00
3.99174504e-02 -8.32763135e-01 -9.94730532e-01 5.84348917e-01
-1.39005446e+00 -8.43977392e-01 1.06214657e-01 2.60265350e+00
1.20767963e+00 2.40538239e-01 -2.41740607e-03 3.98618996e-01
8.54592502e-01 1.83341429e-01 -6.07979447e-02 -5.63136578e-01
-1.89546064e-01 3.61535728e-01 3.04913551e-01 9.23796654e-01
-5.43392539e-01 1.31146657e+00 6.23191881e+00 1.27131069e+00
-1.19467783e+00 5.12157798e-01 6.32007420e-01 -9.72461849e-02
-3.23460937e-01 -2.22111925e-01 -9.34044242e-01 3.68708730e-01
1.83274424e+00 -3.26207310e-01 5.77564716e-01 7.82565773e-01
5.29210806e-01 3.14562559e-01 -1.06976008e+00 1.20101285e+00
5.17874770e-02 -9.66983914e-01 7.03140255e-03 -2.06440046e-01
6.15316212e-01 4.94029820e-01 -9.88513976e-02 4.07275885e-01
3.34050089e-01 -9.37063575e-01 8.24060023e-01 1.43230572e-01
1.17527544e+00 -6.20395482e-01 7.87436903e-01 2.41172418e-01
-1.03252387e+00 2.61074513e-01 -4.87246186e-01 3.51593941e-01
2.12900758e-01 7.01726079e-01 -1.18801129e+00 4.19717729e-01
4.46700722e-01 1.72172949e-01 -3.74316216e-01 9.25738811e-01
-4.84680444e-01 1.30624676e+00 -1.52033731e-01 -1.94023564e-01
1.72513738e-01 -2.10400894e-02 6.32068872e-01 1.53634179e+00
5.34068704e-01 -1.16081655e-01 -3.03286791e-01 4.88437206e-01
-4.08075958e-01 5.74147999e-01 -1.68750659e-01 -4.90981191e-01
8.66625488e-01 1.14322114e+00 -2.07288787e-01 -3.45811814e-01
-2.60399252e-01 1.04965174e+00 3.71591359e-01 2.24810332e-01
-6.44681931e-01 -4.11322474e-01 6.57384217e-01 5.84810823e-02
-9.22772754e-03 -4.62431252e-01 5.12841232e-02 -1.25381160e+00
2.53511667e-01 -1.25514364e+00 -1.93192475e-02 -7.37689078e-01
-1.19310176e+00 1.33785164e+00 -3.53703916e-01 -1.36102915e+00
-3.94126058e-01 -6.06737792e-01 -3.01007777e-01 1.06209183e+00
-1.43874717e+00 -6.27581179e-01 -2.14935690e-02 3.24234486e-01
1.09517598e+00 -5.41908920e-01 1.00333750e+00 5.87505639e-01
-8.10539126e-01 1.11074650e+00 7.73212537e-02 2.27456555e-01
6.86541855e-01 -1.13066602e+00 6.66284204e-01 1.10831296e+00
6.17498636e-01 4.49494183e-01 6.12187326e-01 -3.20705414e-01
-1.19970727e+00 -1.09663904e+00 1.30672193e+00 -5.11852980e-01
6.37826025e-01 -2.58830518e-01 -9.98417795e-01 4.01790649e-01
2.72963881e-01 -1.45292133e-01 6.83408737e-01 1.66551977e-01
-4.23189074e-01 -3.45232964e-01 -7.37266600e-01 4.46551830e-01
1.03545189e+00 -8.15122604e-01 -6.30113602e-01 4.08224583e-01
9.93388355e-01 -2.57421643e-01 -1.01984572e+00 3.37901622e-01
2.31139958e-01 -5.64124465e-01 5.08730590e-01 -3.50943565e-01
-2.60075890e-02 -4.22631174e-01 -6.23874724e-01 -1.65679646e+00
-9.93240103e-02 -7.81675637e-01 1.65856585e-01 1.54295576e+00
1.09530199e+00 -5.05287349e-01 1.16125301e-01 1.69148251e-01
-6.59537435e-01 -5.05158901e-01 -1.20492494e+00 -1.20064223e+00
-9.92681757e-02 -6.04830921e-01 5.52418828e-01 6.79391563e-01
-7.47590214e-02 5.64536989e-01 -2.10073143e-01 3.27235430e-01
2.92852819e-01 -6.72116458e-01 4.66581494e-01 -7.81138003e-01
-3.09192747e-01 -3.96087706e-01 -3.38588357e-01 -1.02903211e+00
2.46863127e-01 -1.02149451e+00 4.88844812e-01 -1.31185675e+00
-2.43156672e-01 -4.74192530e-01 -4.70731020e-01 4.87307966e-01
-1.21802539e-01 5.68519048e-02 2.23554876e-02 1.82303056e-01
-5.85827708e-01 7.74216890e-01 8.75732303e-01 -2.34486923e-01
-2.18585938e-01 7.09953671e-03 -2.43932232e-01 3.73522133e-01
9.46782470e-01 -6.45040214e-01 -2.96346098e-01 -6.28973365e-01
-2.10013703e-01 2.08013549e-01 -2.28488490e-01 -1.23932326e+00
-6.43294528e-02 -4.65010852e-02 -1.30520284e-01 -3.51999313e-01
4.69320983e-01 -2.67889619e-01 3.33365873e-02 1.29495829e-01
-5.48106790e-01 2.59285450e-01 4.98385802e-02 -2.47115120e-02
-6.09178662e-01 -4.44322348e-01 1.00249445e+00 -8.01043492e-03
-4.83398974e-01 1.26785070e-01 -9.10710394e-02 2.16670901e-01
3.18716884e-01 1.44586429e-01 -3.55992019e-01 -4.26511139e-01
-5.00242352e-01 -2.86840469e-01 1.39969230e-01 6.10879898e-01
7.03025758e-01 -1.29099834e+00 -1.14237964e+00 3.72965932e-02
3.39435518e-01 -5.20513475e-01 -4.09665108e-02 5.11560023e-01
-3.25452268e-01 6.75376415e-01 3.04456890e-01 -4.73575264e-01
-1.30436134e+00 3.27818841e-02 5.70003808e-01 -1.32451579e-01
-1.40645504e-01 9.63578820e-01 -3.57643306e-01 -4.94769871e-01
3.68666947e-01 -2.68618166e-01 1.62572354e-01 -3.27012837e-01
7.19469726e-01 4.39002961e-01 7.71409988e-01 -9.14599240e-01
-3.96218330e-01 1.79185465e-01 9.15452763e-02 -6.97170556e-01
9.93691981e-01 -3.15012693e-01 2.49172479e-01 6.76496387e-01
1.16176796e+00 1.58958107e-01 -5.81673801e-01 -4.40505445e-01
1.57170475e-01 -2.98927456e-01 3.59974563e-01 -9.19366598e-01
-8.67721677e-01 9.82780099e-01 6.75606132e-01 1.02880731e-01
8.44602466e-01 6.74644262e-02 8.56266797e-01 6.00161552e-01
2.90063322e-01 -1.41273916e+00 1.46803066e-01 6.58502877e-01
1.19939375e+00 -1.23111033e+00 -5.63388944e-01 -1.96987182e-01
-6.65052831e-01 9.53433692e-01 4.44601148e-01 3.82373452e-01
4.42762792e-01 1.36247292e-01 5.22130549e-01 2.97956944e-01
-9.59975123e-01 -3.72276455e-01 6.13088548e-01 3.93709153e-01
9.67129052e-01 4.32816714e-01 -4.34624523e-01 3.79153877e-01
-6.13242567e-01 -4.46461797e-01 3.32407117e-01 2.85924703e-01
-6.38632715e-01 -1.39402103e+00 -1.66164324e-01 1.95789099e-01
-5.52110851e-01 -4.64485317e-01 -2.22368225e-01 3.01270097e-01
-3.74332726e-01 1.51739967e+00 -1.93262950e-01 -5.71714818e-01
4.91868049e-01 4.10144955e-01 3.17312390e-01 -9.26378906e-01
-3.13411921e-01 7.39162937e-02 5.37140667e-01 -4.35384095e-01
-8.39363784e-02 -5.48008978e-01 -1.13033330e+00 -4.55434248e-02
-6.74406588e-01 5.07285476e-01 1.07786059e+00 9.85239446e-01
2.11746991e-01 3.73929918e-01 8.90735149e-01 -3.28472406e-01
-7.64205813e-01 -1.51153195e+00 -3.31693113e-01 3.39278668e-01
1.78737879e-01 -3.83161575e-01 -6.83198512e-01 1.89337209e-02]
|
[14.412086486816406, 6.84608793258667]
|
19dd5b7e-2666-43d8-b3be-d5f173c7035c
|
covir-a-virtual-rendering-of-a-novel-nn
| null | null |
https://www.sciencedirect.com/science/article/pii/S0097849322000358
|
https://www.sciencedirect.com/science/article/pii/S0097849322000358/pdfft?md5=f5810213e6df1df1a6258b3d72776484&pid=1-s2.0-S0097849322000358-main.pdf
|
COVIR: A virtual rendering of a novel NN architecture O-Net for COVID-19 Ct-scan automatic lung lesions segmentation
|
With the Coronavirus disease 2019 (COVID-19) spread, causing a world pandemic, and recently, the virus new variants continue to appear, making the situation more challenging and threatening, the visual assessment and quantification by expert radiologists have become costly and error-prone. Hence, there is a need to propose a model to predict the COVID-19 cases at the earliest possible to control the disease spread. In order to assist the medical professionals and reduce workload and the time the COVID-19 diagnosis cycle takes, this paper proposes a novel neural network architecture termed as O-Net to automatically segment chest Computerised Tomography Ct-scans infected by COVID-19 with optimised computing power and memory occupation. The O-Net consists of two convolutional autoencoders with an upsampling channel and a downsampling channel. Experimental tests show our proposal’s effectiveness and potential, with a dice score of 0.86, pixel accuracy, precision, specificity of 0.99, 0.99, 0.98, respectively. Performance on the external dataset illustrates generalisation and scalability capabilities of the O-Net model to Ct-scan obtained from different scanners with different sizes. The second objective of this work is to introduce our virtual reality platform, COVIR, that visualises and manipulates 3D reconstructed lungs and segmented infected lesions caused by COVID-19. COVIR platform acts as a reading and visualisation support for medical practitioners to diagnose COVID-19 lung infection. The COVIR platform could be used for medical education professional practice and training. It was tested by Thirteen participants (medical staff, researchers, and collaborators), they conclude that the 3D VR visualisation of segmented Ct-Scan provides an aid diagnosis tool for better interpretation.
|
['Hoceine Kennouche', 'Kahina Amara', 'Ali Aouf']
|
2022-05-15
| null | null | null |
computers-and-graphics-2022-2022-5
|
['2d-semantic-segmentation', 'covid-19-detection']
|
['computer-vision', 'medical']
|
[ 1.47127017e-01 7.83575028e-02 3.49322677e-01 -3.54628544e-03
1.63508996e-01 -4.47138280e-01 1.45986691e-01 7.82407448e-02
-5.32573938e-01 4.94370162e-01 -1.00660972e-01 -7.32009828e-01
-3.16923380e-01 -6.88037157e-01 -2.34263435e-01 -5.39396644e-01
-2.93746144e-01 7.54194260e-01 2.91551560e-01 1.54245481e-01
-2.48891935e-01 1.05937243e+00 -1.21811664e+00 5.02829552e-01
4.63102549e-01 8.07357430e-01 6.53736651e-01 1.28213739e+00
2.12522283e-01 8.04041922e-01 -5.09994805e-01 -2.17043073e-03
3.10127676e-01 -1.91529527e-01 -5.79303563e-01 -2.30524927e-01
6.14639074e-02 -5.53910136e-01 -1.81590497e-01 4.94203478e-01
8.71550560e-01 -9.58074257e-02 8.44396591e-01 -9.85563338e-01
-4.66942966e-01 -1.90241088e-04 -3.50666314e-01 7.60590613e-01
1.01843690e-02 2.66290039e-01 1.50656467e-02 -3.82060319e-01
8.59838307e-01 9.97402728e-01 1.15205944e+00 5.60627580e-01
-6.78111672e-01 -5.25936663e-01 -5.86142480e-01 2.47397333e-01
-1.13754654e+00 4.85224903e-01 1.94700241e-01 -8.97561312e-01
1.28226066e+00 5.83875835e-01 9.39864933e-01 1.11394870e+00
7.30339229e-01 7.32830614e-02 1.03826094e+00 -1.75863117e-01
5.93151934e-02 6.03411555e-01 -2.32441407e-02 7.79464900e-01
5.11242151e-01 4.02389169e-01 3.84161502e-01 -1.27745733e-01
1.17530715e+00 4.90223229e-01 -5.08709729e-01 -3.05536866e-01
-1.11760056e+00 1.00012255e+00 6.48922741e-01 6.71076179e-01
-7.09857881e-01 5.03062755e-02 8.06802392e-01 1.09902360e-01
-1.40140215e-02 1.75175324e-01 -5.35116434e-01 1.26315042e-01
-7.62957811e-01 -9.31019485e-02 5.30462444e-01 4.10517603e-01
-1.87931001e-01 1.34696469e-01 -2.37957776e-01 5.08079112e-01
5.64474344e-01 7.22183287e-01 9.18645620e-01 -5.20615160e-01
-7.15078637e-02 5.10355711e-01 -1.58762217e-01 -9.37608898e-01
-8.64774108e-01 -4.20048833e-01 -1.35230196e+00 3.96957487e-01
-1.32765889e-01 -3.32187921e-01 -1.18016386e+00 1.02799630e+00
4.74723637e-01 2.72158265e-01 4.81036752e-02 1.13641882e+00
1.14247167e+00 6.61808252e-01 1.22091942e-01 -2.74777591e-01
1.74877405e+00 -6.52411103e-01 -5.82209110e-01 3.87673587e-01
5.84399819e-01 -7.12826192e-01 5.20968199e-01 3.61420453e-01
-9.00072813e-01 -7.29332864e-01 -1.05887866e+00 4.18964982e-01
-3.96991700e-01 1.73695579e-01 3.37999314e-01 8.99525464e-01
-1.23541331e+00 5.47663093e-01 -9.20871496e-01 -4.32592899e-01
5.74766517e-01 3.36353034e-01 -3.09247047e-01 1.89638227e-01
-1.09309030e+00 1.23890221e+00 4.79365259e-01 2.38371938e-01
-9.23178434e-01 -7.34412968e-01 -3.32563758e-01 1.54765034e-02
-1.36204660e-01 -1.23715794e+00 1.03462768e+00 -7.80199051e-01
-9.92354453e-01 9.06235874e-01 4.55392957e-01 -6.07226908e-01
8.19908857e-01 1.24178246e-01 -5.35450995e-01 4.10365462e-01
-1.82709932e-01 4.50373381e-01 6.16042554e-01 -1.12842453e+00
-4.69283134e-01 -2.57626891e-01 -4.49529439e-01 -1.13439490e-03
2.47332573e-01 1.29863560e-01 -1.63145512e-01 -4.92999822e-01
-3.77092093e-01 -9.73599851e-01 -4.25616294e-01 1.18247360e-01
-1.09313466e-01 1.49614401e-02 1.13730919e+00 -8.96080613e-01
8.16515326e-01 -1.99662125e+00 -4.18966174e-01 4.02978450e-01
5.90405345e-01 1.07072818e+00 3.64711672e-01 -2.54279058e-02
-3.59147370e-01 1.30327404e-01 -4.00632888e-01 3.60469878e-01
-4.44922537e-01 2.90635228e-01 2.33256906e-01 5.78447938e-01
-2.77607925e-02 8.91612291e-01 -5.46655118e-01 -7.79309332e-01
5.93671381e-01 1.17333829e+00 -1.79525807e-01 3.54244977e-01
1.55673400e-01 5.63496590e-01 -4.47707534e-01 2.45458424e-01
8.11305642e-01 -5.86805105e-01 6.84810281e-02 -6.98650777e-02
4.01771925e-02 -5.62383711e-01 -1.09118462e+00 8.64163637e-01
-4.33306694e-01 8.40984166e-01 1.20570734e-01 -8.98318350e-01
6.47583902e-01 8.60883355e-01 4.40401942e-01 -6.08615935e-01
6.85594320e-01 8.14783499e-02 6.05572611e-02 -1.24329042e+00
4.33118790e-02 -3.49104285e-01 6.35402262e-01 4.93748933e-01
-3.60099614e-01 3.20379883e-02 -1.96808904e-01 -5.73870093e-02
9.93527055e-01 -2.84409314e-01 4.97262955e-01 5.91173731e-02
7.45912731e-01 3.85223664e-02 -7.69392774e-02 5.12604237e-01
-5.33390760e-01 5.24113119e-01 3.25045474e-02 -7.72754431e-01
-1.14944983e+00 -1.10811210e+00 -2.99232304e-01 4.64692593e-01
-2.35790014e-01 5.23503304e-01 -7.95536637e-01 -7.36670434e-01
-2.03975737e-01 3.77080828e-01 -5.91369510e-01 1.46394208e-01
-7.95267642e-01 -3.83757979e-01 6.90566897e-01 7.26093471e-01
2.90007144e-01 -1.60842097e+00 -1.72159576e+00 1.70852169e-01
1.63416475e-01 -7.47767508e-01 9.84623209e-02 1.30483449e-01
-1.04493666e+00 -1.08985281e+00 -1.13121045e+00 -9.53455210e-01
4.88125533e-01 6.48795068e-02 1.03253829e+00 3.64308834e-01
-9.51598704e-01 4.23362732e-01 -2.07057223e-01 -6.39559984e-01
-6.85657680e-01 -3.05845439e-01 -1.63826481e-01 -5.72915971e-01
2.29660440e-02 -4.00683910e-01 -8.61459553e-01 1.59589529e-01
-9.57787573e-01 -4.00827033e-03 7.40269721e-01 6.13562644e-01
6.52092993e-01 -1.82906151e-01 2.76080698e-01 -1.00517440e+00
8.24111879e-01 -5.07884741e-01 -4.40835357e-01 1.00456737e-01
-6.58860028e-01 -4.27177787e-01 6.96552992e-01 -2.56053120e-01
-9.15278852e-01 -1.72111750e-01 -1.99022964e-01 -8.65309298e-01
-4.16370213e-01 6.48622587e-02 5.59070289e-01 1.37052879e-01
9.22275543e-01 6.15944117e-02 8.39098394e-02 -2.00784594e-01
1.34401336e-01 8.98663461e-01 4.84602898e-01 4.86217707e-01
5.48276484e-01 4.35215592e-01 1.63557500e-01 -9.34259057e-01
1.00056280e-03 -6.31130457e-01 -5.21075189e-01 -4.87645924e-01
1.52208817e+00 -7.85074651e-01 -8.58502984e-01 1.15146272e-01
-1.32548094e+00 -2.22415794e-02 -2.79800147e-01 8.16532493e-01
-3.29150051e-01 4.17536825e-01 -5.57468057e-01 -6.97485209e-01
-1.05409825e+00 -1.14834285e+00 5.99052668e-01 2.87901938e-01
-2.57359356e-01 -1.09373903e+00 3.49466085e-01 3.66855055e-01
7.46372998e-01 5.83202422e-01 8.27235162e-01 -8.61351550e-01
-5.28442800e-01 -2.02729464e-01 -5.79931557e-01 5.03515661e-01
-1.54499337e-01 -4.69468273e-02 -9.91990030e-01 -2.42910057e-01
2.57814497e-01 2.55982708e-02 4.81398374e-01 7.87477493e-01
1.16713214e+00 -1.83819190e-01 -5.17261446e-01 5.60857534e-01
1.55837560e+00 7.42490053e-01 4.36235636e-01 2.60455981e-02
7.35585809e-01 3.13593984e-01 3.28549653e-01 3.25616568e-01
-7.63619021e-02 1.94332853e-01 6.84235156e-01 -4.15282041e-01
-1.51450589e-01 4.57408905e-01 -1.62588611e-01 8.73678386e-01
-5.25648177e-01 -4.14932042e-01 -1.02759790e+00 4.40281093e-01
-1.20954740e+00 -8.61083090e-01 -6.33755386e-01 1.72223461e+00
2.33894527e-01 -2.22551107e-01 7.05575868e-02 1.77678913e-01
7.72626460e-01 -2.00005621e-01 -2.59713769e-01 -1.11753726e+00
5.27489245e-01 6.01761401e-01 4.73776639e-01 2.33588830e-01
-8.92497063e-01 1.10695980e-01 5.91286278e+00 3.83231848e-01
-1.62482762e+00 4.87069726e-01 5.40705144e-01 1.17115594e-01
7.80284405e-03 -7.21847773e-01 -1.70151919e-01 5.06957352e-01
1.12966013e+00 3.58000755e-01 9.98328403e-02 9.90857899e-01
3.38763416e-01 2.56426066e-01 -6.59944952e-01 1.06690836e+00
1.64786801e-01 -1.35340250e+00 -1.95709944e-01 1.85612626e-02
4.73253250e-01 5.42242587e-01 -8.54096487e-02 2.37629667e-01
-1.49662385e-03 -1.36958671e+00 2.48688459e-02 6.29160941e-01
1.11804056e+00 -6.67823911e-01 1.26454008e+00 3.52129966e-01
-1.06476796e+00 1.00516625e-01 -3.18510890e-01 2.75041550e-01
1.78240269e-01 2.31246054e-01 -1.73011196e+00 3.98688525e-01
8.03785861e-01 -2.54019443e-02 -4.12176609e-01 1.04966414e+00
1.96658179e-01 4.20789897e-01 -2.16283754e-01 -2.96482742e-01
2.70856827e-01 8.04164335e-02 5.03217638e-01 1.72084248e+00
3.56886983e-01 1.79487079e-01 -1.51985466e-01 6.69852018e-01
3.28320205e-01 2.90262580e-01 -8.19473088e-01 2.14949280e-01
2.87155621e-02 1.36707425e+00 -1.04346752e+00 -5.52810133e-01
-4.23889533e-02 9.28730905e-01 -3.01056266e-01 6.41645193e-02
-1.14662385e+00 -2.11754784e-01 6.47370052e-03 3.16965610e-01
6.22185349e-01 3.13745558e-01 -2.54482865e-01 -3.54221404e-01
-2.75079757e-01 -8.06537032e-01 3.76049697e-01 -1.06704116e+00
-9.99288917e-01 1.00486183e+00 -1.00931227e-02 -1.20571721e+00
-3.94357890e-01 -1.00214195e+00 -6.10572159e-01 9.42654371e-01
-1.11686671e+00 -9.09492850e-01 -5.59551954e-01 4.76636887e-01
2.24998802e-01 -2.66347438e-01 9.35436249e-01 2.39886954e-01
-1.57761529e-01 1.53863728e-01 1.13559075e-01 -8.01524594e-02
1.45965338e-01 -1.13987803e+00 4.82278503e-03 3.96312982e-01
-2.48609453e-01 4.10548091e-01 3.90139312e-01 -6.33265913e-01
-7.27436125e-01 -1.20393336e+00 7.07908273e-01 -3.53428364e-01
9.44271088e-02 1.83568284e-01 -8.55763972e-01 4.86916065e-01
3.86816710e-01 2.25240380e-01 7.40080893e-01 -8.34328413e-01
9.48909819e-02 2.53607363e-01 -1.68650258e+00 1.01266287e-01
5.66559434e-01 -3.27697068e-01 -6.36671722e-01 4.12590504e-01
6.91915631e-01 -4.65358645e-01 -9.61637259e-01 3.66860092e-01
6.83924496e-01 -1.22599626e+00 1.21090400e+00 -3.61019641e-01
3.58371168e-01 -2.32047781e-01 2.16627344e-01 -9.46273267e-01
-2.85804361e-01 3.88892926e-02 4.49675843e-02 2.65468776e-01
2.13638648e-01 -6.71374023e-01 6.83700204e-01 -4.77439910e-02
-5.19100055e-02 -9.99079466e-01 -9.56798196e-01 -2.83895135e-01
-2.07029253e-01 -4.61359292e-01 3.08643162e-01 1.05714154e+00
-6.74693525e-01 2.19832119e-02 -4.71630841e-02 1.84422195e-01
1.27762690e-01 -3.49905759e-01 4.53217775e-01 -1.42338252e+00
-4.45778728e-01 -3.00860792e-01 -7.36616850e-01 -3.78578782e-01
-7.08337963e-01 -8.49379778e-01 -3.17424357e-01 -1.92616653e+00
1.13700941e-01 -2.64825135e-01 -2.86082953e-01 1.84429452e-01
7.68386126e-02 3.47381860e-01 2.96567887e-01 8.80986601e-02
-1.07283778e-01 -2.20501289e-01 1.66467750e+00 9.40683857e-02
-1.55152902e-01 9.89439934e-02 -4.96880747e-02 7.68454909e-01
9.17760134e-01 -5.53821802e-01 -6.42021716e-01 3.32493335e-02
1.28079891e-01 3.54334086e-01 5.91533363e-01 -1.21514142e+00
-8.13610852e-02 1.92403018e-01 8.28657329e-01 -1.01035285e+00
-1.32997608e-04 -1.41735268e+00 6.89253986e-01 1.30132139e+00
1.63670853e-01 3.51920336e-01 2.90210634e-01 4.56026375e-01
1.23479605e-01 -3.81427944e-01 9.53214824e-01 -4.06382114e-01
-3.45627367e-01 1.48455396e-01 -7.55291164e-01 -1.03915565e-01
1.57759094e+00 -5.62699497e-01 -6.52709082e-02 -2.08886079e-02
-6.97098911e-01 -1.45578995e-01 1.15810826e-01 9.50372145e-02
7.58990705e-01 -8.06301653e-01 -6.40480757e-01 4.36612129e-01
-2.72737861e-01 3.41824055e-01 7.27122307e-01 1.02232194e+00
-1.60321438e+00 5.86596668e-01 -4.61422652e-01 -9.09563661e-01
-1.72261667e+00 8.45846534e-01 6.54206753e-01 -6.08940065e-01
-8.35566461e-01 9.52687323e-01 8.85922462e-02 -4.96557236e-01
1.01010039e-01 -6.04346514e-01 -8.15053999e-01 -8.72689858e-02
4.14471447e-01 5.39701700e-01 1.89192966e-01 -6.81852698e-01
-4.08053219e-01 5.65254629e-01 2.71423515e-02 2.63178498e-01
1.21815312e+00 1.32729292e-01 1.81945711e-01 1.82422027e-01
1.13607454e+00 -3.01742554e-01 -5.59909880e-01 2.87976861e-01
-4.74948287e-01 -1.00752473e-01 1.58631373e-02 -1.15446877e+00
-1.19627082e+00 9.72158372e-01 1.63327932e+00 3.44426751e-01
1.09994543e+00 -1.33058056e-01 7.59360254e-01 1.94684252e-01
-2.48481944e-01 -6.92233026e-01 -1.24472290e-01 1.90365553e-01
6.89294815e-01 -1.10338235e+00 1.44602796e-02 -1.29949003e-01
-9.16936100e-01 1.12313187e+00 2.85508454e-01 -1.14727564e-01
6.98037863e-01 3.66453528e-01 5.31339884e-01 -6.19708657e-01
-4.64757532e-01 3.78365964e-01 1.60779670e-01 1.00054014e+00
5.14971673e-01 5.20122349e-01 -3.28773677e-01 3.97142678e-01
-1.30467519e-01 3.84330928e-01 4.45065886e-01 9.08949852e-01
-4.37266946e-01 -4.17001277e-01 -7.41955698e-01 7.52279282e-01
-7.52737224e-01 1.68656722e-01 -5.56144118e-02 1.08598399e+00
7.58163273e-01 4.68789279e-01 1.77937120e-01 -9.79277343e-02
2.95727581e-01 -6.35602847e-02 1.84900701e-01 -2.30842158e-01
-9.85363007e-01 -2.08345592e-01 -1.51335493e-01 -3.22747231e-01
-4.14861023e-01 -2.23631948e-01 -1.42650962e+00 -1.55486211e-01
-1.66628420e-01 -6.14282154e-02 1.08648372e+00 5.10433257e-01
2.67224818e-01 1.13885224e+00 3.15696985e-01 -4.77214545e-01
-4.57616299e-01 -1.04971993e+00 -4.04427975e-01 2.44036376e-01
4.91531372e-01 -3.68058681e-01 -2.18140781e-01 1.05586007e-01]
|
[15.564326286315918, -1.713329792022705]
|
51cd3843-c7d1-49d9-aeab-8c616dbf84d0
|
iql-td-mpc-implicit-q-learning-for
|
2306.00867
| null |
https://arxiv.org/abs/2306.00867v1
|
https://arxiv.org/pdf/2306.00867v1.pdf
|
IQL-TD-MPC: Implicit Q-Learning for Hierarchical Model Predictive Control
|
Model-based reinforcement learning (RL) has shown great promise due to its sample efficiency, but still struggles with long-horizon sparse-reward tasks, especially in offline settings where the agent learns from a fixed dataset. We hypothesize that model-based RL agents struggle in these environments due to a lack of long-term planning capabilities, and that planning in a temporally abstract model of the environment can alleviate this issue. In this paper, we make two key contributions: 1) we introduce an offline model-based RL algorithm, IQL-TD-MPC, that extends the state-of-the-art Temporal Difference Learning for Model Predictive Control (TD-MPC) with Implicit Q-Learning (IQL); 2) we propose to use IQL-TD-MPC as a Manager in a hierarchical setting with any off-the-shelf offline RL algorithm as a Worker. More specifically, we pre-train a temporally abstract IQL-TD-MPC Manager to predict "intent embeddings", which roughly correspond to subgoals, via planning. We empirically show that augmenting state representations with intent embeddings generated by an IQL-TD-MPC manager significantly improves off-the-shelf offline RL agents' performance on some of the most challenging D4RL benchmark tasks. For instance, the offline RL algorithms AWAC, TD3-BC, DT, and CQL all get zero or near-zero normalized evaluation scores on the medium and large antmaze tasks, while our modification gives an average score over 40.
|
['Olivier Delalleau', 'Zheqing Zhu', 'Urun Dogan', 'Lucas Lehnert', 'Bobak Hashemi', 'Yingchen Xu', 'Rohan Chitnis']
|
2023-06-01
| null | null | null | null |
['q-learning', 'offline-rl', 'model-based-reinforcement-learning', 'd4rl']
|
['methodology', 'playing-games', 'reasoning', 'robots']
|
[-2.79763281e-01 3.01488549e-01 -4.39415097e-01 5.44618517e-02
-1.03788066e+00 -5.71205437e-01 8.74994874e-01 1.02843270e-02
-6.75659359e-01 8.66098166e-01 4.67798412e-01 -2.94523656e-01
-2.31997147e-01 -5.89155555e-01 -8.43666732e-01 -6.32243812e-01
-5.86472690e-01 1.00071502e+00 1.52706265e-01 -4.74220455e-01
1.99295938e-01 2.61838287e-01 -1.36682832e+00 5.13815209e-02
8.15987468e-01 8.98804724e-01 3.86588156e-01 7.32113540e-01
3.29244912e-01 1.25744724e+00 -3.12511444e-01 2.56145060e-01
5.13265610e-01 -2.61817306e-01 -9.11494255e-01 -2.34250307e-01
-2.76560515e-01 -7.03026533e-01 -6.50025964e-01 4.04913157e-01
5.51420629e-01 6.75254822e-01 3.78952712e-01 -1.46329188e+00
-3.18534642e-01 5.14807820e-01 -2.06276685e-01 -1.46909133e-01
3.45306039e-01 1.10270917e+00 9.87512767e-01 -3.60577077e-01
6.32850409e-01 1.44722486e+00 4.63174582e-01 7.34823108e-01
-1.50437105e+00 -4.37499791e-01 5.65539420e-01 2.75193661e-01
-9.23843503e-01 -2.90880680e-01 5.30256748e-01 -2.48076424e-01
1.56572902e+00 -3.02534819e-01 9.82536256e-01 1.36620378e+00
5.03298998e-01 1.17435062e+00 1.35560775e+00 -7.84731358e-02
8.51014555e-01 -4.25629854e-01 -3.86388987e-01 8.10029447e-01
-3.87187362e-01 1.03066874e+00 -5.91843188e-01 -1.50592774e-01
7.67046988e-01 -1.39775127e-01 1.63136169e-01 -6.54966116e-01
-1.27164817e+00 1.01640439e+00 4.57185060e-01 -1.27683550e-01
-5.62335670e-01 7.56765723e-01 5.35630882e-01 5.61728954e-01
3.43652159e-01 1.08497930e+00 -7.86272764e-01 -5.33981860e-01
-4.98305142e-01 7.78466880e-01 8.32811296e-01 1.01022267e+00
8.19544137e-01 2.15548575e-01 -4.50447410e-01 2.77796060e-01
7.13364109e-02 2.21109331e-01 7.06159592e-01 -1.65596461e+00
4.18739051e-01 1.63761795e-01 6.85562551e-01 -2.15559676e-01
-7.14852631e-01 -2.74992257e-01 -2.31313437e-01 5.22346556e-01
1.23519234e-01 -5.58147848e-01 -8.64712417e-01 1.92495894e+00
3.51236880e-01 4.71638501e-01 4.89123195e-01 8.03623259e-01
1.08894669e-01 8.64550829e-01 7.33638778e-02 -2.15622693e-01
7.92809606e-01 -1.63487577e+00 -3.10328931e-01 -5.97814083e-01
8.76373708e-01 3.81487124e-02 1.23007488e+00 5.71908772e-01
-9.30246115e-01 -5.86856008e-01 -8.67885053e-01 1.80176333e-01
-1.26073048e-01 -2.85341591e-01 7.66667366e-01 -1.01871490e-01
-1.12837553e+00 9.38942611e-01 -1.27524638e+00 -2.99868435e-01
2.64363378e-01 4.39308345e-01 -2.13957801e-01 -1.51520178e-01
-1.04375672e+00 1.25711298e+00 5.53744316e-01 -2.70219117e-01
-1.92509091e+00 -7.37331927e-01 -8.83193851e-01 -1.11745149e-01
8.48597527e-01 -5.69710851e-01 2.00054765e+00 -7.17907965e-01
-2.07362509e+00 1.85865223e-01 1.26922488e-01 -8.39470923e-01
4.07316327e-01 -4.14493591e-01 -1.12076364e-01 8.98700133e-02
2.02380627e-01 1.00548351e+00 8.47586811e-01 -1.21081603e+00
-5.72000206e-01 -1.91349149e-01 4.41445053e-01 5.08976221e-01
2.30328411e-01 -2.66620338e-01 -8.06125924e-02 -3.38580817e-01
-4.74338382e-01 -1.21186411e+00 -6.99698627e-01 -1.81426972e-01
1.53647631e-01 -6.15114152e-01 7.10719466e-01 -3.95580143e-01
7.81523526e-01 -1.85758722e+00 3.68966699e-01 -3.11948061e-01
-6.84576333e-02 1.60264552e-01 -7.21909702e-01 8.44744027e-01
2.23336577e-01 -2.16300786e-01 -5.13501875e-02 -4.22596157e-01
3.58875424e-01 7.45003760e-01 -5.23320615e-01 2.54939169e-01
2.98359632e-01 1.07406759e+00 -1.39961445e+00 -3.75889450e-01
2.12990567e-01 -1.68788373e-01 -7.77990758e-01 5.72278500e-01
-1.07671571e+00 7.53810346e-01 -5.54766297e-01 3.19033444e-01
4.40080129e-02 6.37852103e-02 3.48256618e-01 4.87054020e-01
-4.21435446e-01 5.63893020e-01 -8.23143005e-01 2.30289221e+00
-6.22679651e-01 3.44976574e-01 1.64356432e-03 -9.19088900e-01
5.27770579e-01 4.20888752e-01 7.20201612e-01 -1.06465089e+00
-2.26751909e-01 3.22272480e-02 2.03265604e-02 -3.92466128e-01
5.10954738e-01 -9.74954367e-02 -3.87292504e-01 7.69323647e-01
5.13418317e-02 -4.60945487e-01 2.57222503e-01 5.46413213e-02
1.54009736e+00 8.63176346e-01 2.24976033e-01 -6.15957454e-02
-6.31680042e-02 4.87618208e-01 7.96618223e-01 9.20943558e-01
-5.26671112e-01 4.02364954e-02 7.54102528e-01 -6.68748200e-01
-8.51035595e-01 -9.99050498e-01 5.27264476e-01 1.54195011e+00
-1.87600944e-02 -5.11455953e-01 -3.47326249e-01 -8.80287170e-01
1.71482414e-01 1.16966879e+00 -7.06964731e-01 -3.08582872e-01
-8.49386990e-01 -3.69626522e-01 3.86323154e-01 6.88020706e-01
2.91942298e-01 -1.42935109e+00 -1.13525999e+00 6.92471385e-01
2.14471444e-01 -7.68137693e-01 -4.09075916e-01 6.12820745e-01
-8.68876755e-01 -9.26818430e-01 -2.23188609e-01 -5.15371501e-01
2.62448311e-01 6.70581460e-02 1.19643009e+00 -2.89570332e-01
5.11378944e-02 5.86163402e-01 -4.69770402e-01 -2.52887160e-01
-3.30064088e-01 2.40151361e-02 3.27072054e-01 -6.04714274e-01
-1.43964633e-01 -5.24613380e-01 -6.09584391e-01 1.93248481e-01
-5.68158865e-01 7.70709515e-02 6.45851433e-01 1.02992821e+00
6.02308512e-01 3.16463970e-02 6.27127111e-01 -4.28583473e-01
8.83803606e-01 -4.53540295e-01 -8.81719410e-01 -2.10559200e-02
-8.57816160e-01 6.48883343e-01 9.23090756e-01 -6.10457242e-01
-9.71560597e-01 8.13407302e-02 1.14924327e-01 -6.12182200e-01
3.86926811e-03 5.00482380e-01 2.90347993e-01 2.59243578e-01
5.81974506e-01 3.59860003e-01 3.09572488e-01 -3.30960572e-01
5.75657725e-01 3.63506675e-02 3.32834721e-01 -1.11285341e+00
6.50079966e-01 2.05607757e-01 -1.41946031e-02 -2.14431688e-01
-9.30701673e-01 -1.79824725e-01 -3.09718549e-01 5.80789931e-02
7.76362956e-01 -1.07374692e+00 -1.01990700e+00 2.22853974e-01
-9.04491246e-01 -1.62075365e+00 -6.57377303e-01 4.38522995e-01
-1.45695364e+00 -9.15288106e-02 -6.72707677e-01 -8.38716686e-01
-1.83674082e-01 -1.27450573e+00 8.76461565e-01 7.99711496e-02
-1.93555802e-02 -8.72529149e-01 7.04783320e-01 -3.39721795e-03
4.28175837e-01 1.76584065e-01 9.15674210e-01 -5.94490111e-01
-4.97036546e-01 3.92255604e-01 1.87309086e-01 1.27370447e-01
-2.75392622e-01 -4.85979974e-01 -7.62125194e-01 -6.89250469e-01
-3.01037490e-01 -1.10295379e+00 7.01323569e-01 3.75285685e-01
1.01920331e+00 -5.96994162e-01 -2.63460815e-01 3.42831999e-01
1.28538013e+00 4.71230209e-01 3.66429806e-01 5.32579541e-01
2.53045112e-01 4.42683011e-01 1.23571002e+00 8.12399685e-01
5.95721662e-01 7.20963955e-01 8.94003093e-01 2.84154803e-01
2.25767016e-01 -7.09569931e-01 8.86633635e-01 3.61668497e-01
1.38045996e-01 5.63544296e-02 -7.86800444e-01 5.67070782e-01
-2.36671472e+00 -9.92412865e-01 5.69693387e-01 2.03901911e+00
1.00906634e+00 2.67986748e-02 4.56213504e-01 -4.97217357e-01
-7.12639466e-02 4.41821158e-01 -1.19603395e+00 -5.34135818e-01
4.43121552e-01 3.99981171e-01 3.91889304e-01 5.65778553e-01
-1.08253884e+00 1.38794124e+00 5.97998762e+00 6.58775866e-01
-9.10309136e-01 2.72599250e-01 3.06129783e-01 -4.46220279e-01
1.14591986e-01 2.90439844e-01 -5.88620782e-01 2.18593851e-01
1.23429382e+00 -1.70265540e-01 1.08342731e+00 1.14280057e+00
4.75459218e-01 -2.45476738e-01 -1.50750923e+00 5.85957527e-01
-4.69534367e-01 -1.17485094e+00 -5.48276722e-01 2.26724565e-01
9.00398314e-01 5.36870539e-01 1.70744583e-01 1.34707177e+00
1.09258378e+00 -1.18912148e+00 7.93376565e-01 3.97199422e-01
4.83863890e-01 -7.85014808e-01 3.78948420e-01 7.89117992e-01
-1.12805820e+00 -5.99517405e-01 -3.27883512e-01 -4.77364749e-01
5.22423675e-03 -2.42925599e-01 -9.95763361e-01 2.97102869e-01
4.26289439e-01 7.32479274e-01 -2.27538079e-01 7.18427360e-01
-4.63979781e-01 4.02387232e-01 -1.63396850e-01 -4.56069633e-02
8.74394953e-01 -1.08666025e-01 3.74351710e-01 6.38966143e-01
5.49567528e-02 9.84976217e-02 7.64845669e-01 9.06130910e-01
1.34432703e-01 -4.70983475e-01 -7.19449520e-01 -2.14480057e-01
4.30290341e-01 9.42178845e-01 -1.31990939e-01 -2.56038994e-01
-1.13528870e-01 8.26318443e-01 6.95471704e-01 3.10972333e-01
-9.15734529e-01 7.27536902e-02 1.03680885e+00 -2.47390941e-01
3.53250504e-01 -5.45060575e-01 2.66012460e-01 -9.54097509e-01
-3.75155061e-01 -1.14679027e+00 2.77505487e-01 -7.55993307e-01
-1.00647700e+00 3.06989849e-01 1.23556390e-01 -1.12527835e+00
-9.47697222e-01 -5.28661609e-01 -5.26333332e-01 5.00669301e-01
-1.66755140e+00 -8.65295529e-01 6.65876418e-02 6.44571424e-01
8.72308552e-01 -1.09630749e-01 1.04695404e+00 -4.26839650e-01
-5.24725735e-01 1.33142635e-01 2.07546130e-01 -1.33670434e-01
6.14932656e-01 -1.47656584e+00 4.42428499e-01 4.71764028e-01
-1.27000928e-01 1.92069426e-01 7.96399832e-01 -4.90692735e-01
-1.91410518e+00 -1.29115462e+00 1.49518996e-01 -4.76501256e-01
8.31731200e-01 -2.04126582e-01 -4.19644833e-01 1.04106343e+00
4.01351511e-01 1.64487988e-01 9.60514694e-02 7.14225620e-02
-2.17984095e-01 -1.07698075e-01 -1.05898380e+00 8.23125660e-01
1.06591344e+00 -4.16628152e-01 -7.85258770e-01 5.92426360e-01
1.24295306e+00 -4.93561864e-01 -9.79599535e-01 3.19297127e-02
4.92341310e-01 -8.91519010e-01 8.13866496e-01 -9.63192821e-01
3.43028247e-01 -1.81667134e-01 -9.21826884e-02 -1.85894358e+00
-3.04116458e-01 -1.06484663e+00 -4.44606751e-01 4.87478226e-01
1.86659709e-01 -7.12605476e-01 7.85279989e-01 5.94722092e-01
-4.27833229e-01 -1.00784731e+00 -9.83966887e-01 -1.11822355e+00
2.79585034e-01 -5.49142897e-01 5.59681714e-01 4.52039301e-01
1.73428237e-01 2.54212856e-01 -4.48469132e-01 5.00775203e-02
4.57902133e-01 2.88539708e-01 9.50957775e-01 -9.33445334e-01
-7.56256402e-01 -2.03993246e-01 2.89880097e-01 -1.29213417e+00
6.64196789e-01 -5.33939779e-01 4.08134401e-01 -1.63924789e+00
-1.38331532e-01 -7.64535844e-01 -2.71688253e-01 8.52335095e-01
3.12606454e-01 -6.17485344e-01 4.09346879e-01 1.96348906e-01
-1.09020674e+00 1.24817407e+00 1.45911610e+00 -8.06811228e-02
-5.59207439e-01 -9.61790830e-02 -4.05525476e-01 5.27909696e-01
8.99123728e-01 -5.71406782e-01 -7.34019995e-01 -5.19969165e-01
7.62070939e-02 6.79994702e-01 3.12625647e-01 -1.12202191e+00
1.66756645e-01 -7.60567844e-01 -8.78664926e-02 -3.91011238e-01
7.02157080e-01 -6.42776608e-01 -2.30188385e-01 9.46129978e-01
-6.51952744e-01 3.22968334e-01 2.41496980e-01 8.79738450e-01
1.31339952e-01 -7.42164925e-02 5.88287234e-01 -4.99921381e-01
-1.05890286e+00 3.84485245e-01 -6.54342115e-01 4.07925695e-01
1.18105948e+00 2.18327403e-01 -5.00543714e-01 -4.70062196e-01
-6.02966905e-01 7.02990472e-01 4.32812691e-01 3.25048000e-01
5.48617482e-01 -1.08737040e+00 -4.22980160e-01 2.42913850e-02
6.72828183e-02 1.78929657e-01 -3.86221055e-03 8.91566396e-01
-2.55462706e-01 6.65988982e-01 -3.46110225e-01 -2.17196614e-01
-4.86472011e-01 8.57058823e-01 3.48814577e-01 -9.50098217e-01
-7.27467597e-01 6.46536946e-01 8.26236978e-02 -8.71168435e-01
2.86473036e-01 -6.16339505e-01 1.63306698e-01 -2.10067645e-01
2.34134540e-01 3.58911902e-01 -2.84598649e-01 2.24339757e-02
-1.58137158e-01 8.67806375e-02 -1.98556390e-02 -6.09353006e-01
1.57600284e+00 3.28266084e-01 3.14531982e-01 3.15994710e-01
8.36421549e-01 -6.67691410e-01 -2.16476989e+00 -1.38614133e-01
9.64626744e-02 -1.08900592e-01 1.31614476e-01 -9.88170147e-01
-6.21067286e-01 6.38886392e-01 4.27820086e-01 -9.49162990e-02
8.61479640e-01 -1.72028214e-01 7.36634433e-01 8.64832401e-01
1.00454795e+00 -1.47085345e+00 6.40274882e-01 9.27866340e-01
9.98826087e-01 -1.24650788e+00 -4.32190970e-02 7.73964763e-01
-1.12075377e+00 7.57205188e-01 9.14242148e-01 -4.85390067e-01
1.31845891e-01 2.51751803e-02 -3.61373901e-01 2.39351671e-02
-1.76518488e+00 -4.53243613e-01 -2.76342154e-01 8.40134978e-01
-1.36842549e-01 2.02962384e-01 2.12605640e-01 3.94938827e-01
-8.85363221e-02 1.06758326e-01 4.71010864e-01 1.24991155e+00
-4.86448109e-01 -1.12456644e+00 -3.37017439e-02 2.04606116e-01
1.09954931e-01 1.08768895e-01 -8.28065500e-02 1.03342164e+00
-1.77573264e-01 1.01296794e+00 1.36520281e-01 -4.57791865e-01
2.29845881e-01 -4.95416811e-03 6.89787507e-01 -7.46728241e-01
-7.42232203e-01 -9.19644162e-02 2.81511486e-01 -1.32436848e+00
-8.28502402e-02 -7.26885676e-01 -1.57230949e+00 -1.63823798e-01
3.14778209e-01 1.67695567e-01 5.04208207e-01 9.95411098e-01
6.08516395e-01 4.91995156e-01 6.46379709e-01 -1.31396961e+00
-1.28886414e+00 -1.01038969e+00 -4.44111556e-01 -7.58641511e-02
4.44133431e-01 -8.85942161e-01 -2.14800641e-01 -5.26749372e-01]
|
[4.118249416351318, 1.6126997470855713]
|
7542cad9-a9e3-43a6-b766-10089370f33d
|
convolutional-neural-network-models-for
|
1906.07794
| null |
https://arxiv.org/abs/1906.07794v1
|
https://arxiv.org/pdf/1906.07794v1.pdf
|
Convolutional neural network models for cancer type prediction based on gene expression
|
Background Precise prediction of cancer types is vital for cancer diagnosis and therapy. Important cancer marker genes can be inferred through predictive model. Several studies have attempted to build machine learning models for this task however none has taken into consideration the effects of tissue of origin that can potentially bias the identification of cancer markers. Results In this paper, we introduced several Convolutional Neural Network (CNN) models that take unstructured gene expression inputs to classify tumor and non-tumor samples into their designated cancer types or as normal. Based on different designs of gene embeddings and convolution schemes, we implemented three CNN models: 1D-CNN, 2D-Vanilla-CNN, and 2D-Hybrid-CNN. The models were trained and tested on combined 10,340 samples of 33 cancer types and 731 matched normal tissues of The Cancer Genome Atlas (TCGA). Our models achieved excellent prediction accuracies (93.9-95.0%) among 34 classes (33 cancers and normal). Furthermore, we interpreted one of the models, known as 1D-CNN model, with a guided saliency technique and identified a total of 2,090 cancer markers (108 per class). The concordance of differential expression of these markers between the cancer type they represent and others is confirmed. In breast cancer, for instance, our model identified well-known markers, such as GATA3 and ESR1. Finally, we extended the 1D-CNN model for prediction of breast cancer subtypes and achieved an average accuracy of 88.42% among 5 subtypes. The codes can be found at https://github.com/chenlabgccri/CancerTypePrediction.
|
['Yu-Chiao Chiu', 'Yidong Chen', 'Yufei Huang', 'Milad Mostavi']
|
2019-06-18
| null | null | null | null |
['type-prediction']
|
['computer-code']
|
[ 1.32662645e-02 3.31736475e-01 -7.52468348e-01 -3.09599340e-01
-9.46429133e-01 -8.85308981e-02 4.71393675e-01 4.42313492e-01
-2.84298539e-01 7.56627440e-01 2.78153330e-01 -6.77938342e-01
-2.80702021e-02 -9.00171757e-01 -5.03915966e-01 -1.04553866e+00
-2.04652444e-01 5.09419441e-01 -1.14410594e-01 -1.32169873e-01
5.44452444e-02 5.90241313e-01 -1.18071222e+00 3.67084980e-01
6.87374234e-01 9.39430118e-01 2.04905737e-02 8.80199730e-01
-2.97695309e-01 5.17420471e-01 -2.95828581e-01 -1.44785956e-01
-2.00674817e-01 -2.13928044e-01 -6.86660945e-01 -6.38919711e-01
1.58016175e-01 1.39694124e-01 -3.73741567e-01 1.13983428e+00
5.61925471e-01 -5.60238302e-01 7.81687558e-01 -1.22084320e+00
-8.31535220e-01 4.14380372e-01 -4.47686493e-01 8.53544101e-02
-1.33965507e-01 -3.30644734e-02 9.88098145e-01 -1.02851915e+00
6.86457694e-01 7.97162712e-01 1.05397713e+00 1.02116776e+00
-1.08729279e+00 -7.58152425e-01 -4.82892781e-01 -1.25756217e-02
-1.65924180e+00 -4.19453174e-01 3.08406562e-01 -4.48631138e-01
8.31847966e-01 6.08982801e-01 8.24739397e-01 1.07631481e+00
6.28902793e-01 6.86498344e-01 9.02921200e-01 -4.44478452e-01
1.66823804e-01 2.62909293e-01 3.85167509e-01 8.67331326e-01
3.61922860e-01 1.44719690e-01 -6.81037754e-02 -4.42806840e-01
4.99589384e-01 2.62399197e-01 -4.23197538e-01 1.73852593e-01
-1.27109504e+00 9.54414666e-01 5.77522635e-01 5.46008110e-01
-6.37942627e-02 5.94840571e-02 6.60334885e-01 -1.27403975e-01
6.78641319e-01 3.85171920e-01 -7.04496145e-01 2.22318724e-01
-7.49534309e-01 -2.31659353e-01 5.91082215e-01 7.79951692e-01
5.49715638e-01 -2.65327513e-01 -3.21892440e-01 9.95169103e-01
2.42102608e-01 1.75460219e-01 1.01329803e+00 -2.68909723e-01
-3.30778390e-01 7.98191667e-01 -2.28985801e-01 -9.79840040e-01
-9.35301423e-01 -5.23167074e-01 -1.30238605e+00 -1.21799754e-02
3.34181488e-01 -1.08478852e-02 -1.06672359e+00 1.55069351e+00
1.57344311e-01 3.64535332e-01 2.47669339e-01 7.56606698e-01
1.30758190e+00 2.52380937e-01 3.21240336e-01 2.28334397e-01
1.54964542e+00 -8.16758871e-01 -6.25277877e-01 2.16900125e-01
1.45892000e+00 -3.32237542e-01 6.80754066e-01 -1.06107175e-01
-5.46815038e-01 -3.16197902e-01 -8.28677416e-01 -2.86939442e-01
-8.76057744e-01 3.72723907e-01 9.06430364e-01 5.39666712e-01
-1.26248479e+00 3.98449630e-01 -8.96704257e-01 -7.38002002e-01
8.88877213e-01 5.23567080e-01 -5.19649625e-01 -1.16538452e-02
-1.20676005e+00 8.92586768e-01 2.85248250e-01 7.44666159e-02
-1.08193004e+00 -1.00880992e+00 -5.75776517e-01 2.92940792e-02
-4.51370567e-01 -7.94842839e-01 9.09300208e-01 -9.92988706e-01
-1.10145712e+00 1.46226835e+00 -2.54483670e-01 -2.51147181e-01
-3.31734158e-02 1.36691824e-01 -5.16716897e-01 -2.01147363e-01
2.39017792e-03 7.96479583e-01 -2.10059389e-01 -8.16552877e-01
-6.07596040e-01 -3.77039015e-01 -3.91279012e-01 -4.17832024e-02
-3.90713334e-01 -2.24595293e-01 -3.56396079e-01 -3.29421699e-01
-6.11885153e-02 -9.31837678e-01 -4.21494365e-01 4.26811635e-01
-7.86709309e-01 -2.24301860e-01 5.63535869e-01 -6.07185364e-01
8.98167729e-01 -2.13540745e+00 -2.41838530e-01 1.07197747e-01
5.05113780e-01 2.62085676e-01 -1.24207459e-01 4.56361920e-02
-4.63232398e-01 6.89288497e-01 8.94153342e-02 6.87173754e-02
-1.97461754e-01 1.61720276e-01 2.87499398e-01 8.03069174e-01
4.37134475e-01 1.27595925e+00 -1.00793636e+00 -4.69094008e-01
-5.46383448e-02 7.30307698e-01 -2.16959432e-01 1.37575492e-01
-2.29140390e-02 1.05042167e-01 -5.08212805e-01 1.17788756e+00
7.11622059e-01 -2.05475137e-01 1.40831441e-01 -3.15218836e-01
1.46823660e-01 4.12487946e-02 -1.67435721e-01 1.33481979e+00
-2.00542077e-01 8.12221467e-01 -7.85314739e-02 -9.46371853e-01
1.09742069e+00 2.73075044e-01 5.42237639e-01 -3.51012796e-01
2.27651715e-01 2.39255279e-01 3.12043756e-01 -5.01541495e-01
1.00841522e-01 -1.87995568e-01 6.42846897e-02 -2.05511227e-01
1.53619843e-02 3.04024220e-01 -2.78451294e-01 -5.81672452e-02
1.33842731e+00 -4.11673844e-01 5.13637304e-01 -6.06604934e-01
3.57354581e-01 4.03798193e-01 8.98078263e-01 4.70362931e-01
-5.58160007e-01 6.97498262e-01 9.17436540e-01 -7.63467371e-01
-7.57899702e-01 -7.73292243e-01 -5.38236976e-01 7.64978349e-01
-2.77311206e-01 -1.22091293e-01 -2.43731022e-01 -6.37295902e-01
1.03380986e-01 6.04503036e-01 -1.32031703e+00 -5.02203643e-01
-1.22560434e-01 -1.41760635e+00 1.24169731e+00 4.43350196e-01
2.38669962e-01 -5.93244016e-01 -6.72987476e-02 -1.87172294e-02
1.48326278e-01 -4.69213635e-01 -1.90538406e-01 7.19617784e-01
-7.87352383e-01 -1.55320597e+00 -6.51601374e-01 -1.05793166e+00
9.56571639e-01 -1.28237575e-01 1.14132559e+00 3.85844171e-01
-6.16385758e-01 -2.37166628e-01 -1.73764721e-01 -7.77208805e-01
-4.96950001e-01 1.11392170e-01 -9.94415507e-02 -2.36788511e-01
8.53799403e-01 -2.61676721e-02 -5.17409265e-01 4.17810045e-02
-5.49466550e-01 3.23787689e-01 8.85631979e-01 1.23072875e+00
8.43245149e-01 -2.17371330e-01 5.68724394e-01 -1.13732958e+00
3.60380650e-01 -9.03670728e-01 -9.00029987e-02 1.22579657e-01
-4.74395663e-01 -3.13644171e-01 5.34280181e-01 -2.45730251e-01
-6.51974618e-01 2.22843856e-01 -6.64621055e-01 -1.96785435e-01
-4.77779120e-01 5.40879667e-01 -1.32036075e-01 -6.99690655e-02
5.64303637e-01 2.33353674e-01 1.81992650e-01 -7.25178868e-02
-5.25408946e-02 7.91620851e-01 3.11182469e-01 -1.07428253e-01
9.90690887e-02 4.61978495e-01 2.51682043e-01 -6.43413424e-01
-8.77633989e-01 -4.05966252e-01 -5.67679167e-01 -3.44974883e-02
9.55616415e-01 -8.70793343e-01 -6.45144641e-01 5.99637568e-01
-7.31384397e-01 -4.70367610e-01 1.33012787e-01 4.65547115e-01
-7.08706975e-02 -1.59261480e-01 -9.17709887e-01 -3.08110714e-01
-6.31210029e-01 -9.48190451e-01 1.10500491e+00 3.76392603e-01
-3.28618228e-01 -1.40997040e+00 3.25985700e-01 -2.38031253e-01
5.55951297e-01 6.13533497e-01 1.22080302e+00 -1.29718602e+00
-3.69568504e-02 -5.26223123e-01 -4.55109477e-01 -9.29089114e-02
3.41946572e-01 4.41098630e-01 -1.18417549e+00 -6.29540011e-02
-7.59216011e-01 -1.51834801e-01 9.44669724e-01 6.23195171e-01
1.47800267e+00 -9.32951421e-02 -1.24337494e+00 1.11598337e+00
1.57132876e+00 3.85337651e-01 7.72860587e-01 3.76480639e-01
5.49089372e-01 1.35638133e-01 4.27508652e-01 6.96714385e-04
1.46857440e-01 3.65030736e-01 5.29194951e-01 -6.16937101e-01
-3.92609648e-02 -3.43957804e-02 -1.04403071e-01 5.59699714e-01
1.14057839e-01 -3.82655591e-01 -1.12620246e+00 7.98466861e-01
-1.27977335e+00 -6.54795766e-01 -4.85153526e-01 1.83807373e+00
8.44824553e-01 -1.83460396e-02 -3.54275405e-01 -1.32334277e-01
7.47946024e-01 -1.14819460e-01 -6.06260121e-01 -5.69316745e-01
-3.49712431e-01 1.57445326e-01 5.37856936e-01 2.09708482e-01
-1.13701892e+00 5.81239402e-01 6.62377930e+00 8.03483188e-01
-1.33403242e+00 2.55782008e-02 1.47387755e+00 6.19579628e-02
-2.43745282e-01 -3.55103791e-01 -8.60669971e-01 3.54248703e-01
1.22831798e+00 -1.48318172e-01 -4.93499160e-01 8.45662475e-01
1.24896683e-01 -1.42990956e-02 -1.21911800e+00 5.68718433e-01
-1.87853709e-01 -1.70880902e+00 -2.13837504e-01 1.67638406e-01
4.62711662e-01 3.72016996e-01 -1.18423820e-01 4.39298511e-01
3.30391884e-01 -1.45364141e+00 -1.96883768e-01 8.50286663e-01
1.18444192e+00 -6.33162558e-01 1.35416174e+00 1.45965114e-01
-6.00318015e-01 2.52751648e-01 -5.69679677e-01 2.05667019e-01
-4.67993408e-01 9.85706747e-01 -1.43223298e+00 3.11254561e-01
8.06100309e-01 8.74061882e-01 -6.71008587e-01 9.35896933e-01
3.27562839e-01 9.47801888e-01 -1.13739654e-01 -4.57028121e-01
4.86091264e-02 2.99125433e-01 1.11998394e-02 1.51484716e+00
5.40829480e-01 7.46697783e-02 -1.94274321e-01 8.05317521e-01
1.00453626e-02 4.21503596e-02 -4.02876884e-01 -1.61099583e-01
4.47200447e-01 1.61352193e+00 -6.64798141e-01 -3.41566265e-01
-5.08253157e-01 6.33950293e-01 3.94662410e-01 1.04607418e-01
-8.24014604e-01 -5.37499189e-01 1.06212234e+00 -8.47159252e-02
-1.67265028e-01 4.24050033e-01 -5.04463851e-01 -7.87332475e-01
-7.51281857e-01 -4.25114453e-01 5.46602845e-01 -6.62476301e-01
-1.52174509e+00 5.48399270e-01 -5.54577529e-01 -1.02430546e+00
2.25458354e-01 -9.73951221e-01 -8.36022973e-01 9.21921074e-01
-1.48526573e+00 -1.05745900e+00 -4.42280561e-01 1.58738330e-01
2.07128048e-01 -2.55719960e-01 1.48545086e+00 1.80511057e-01
-9.64874268e-01 7.98960686e-01 3.64014477e-01 5.92032254e-01
6.85764432e-01 -1.25278056e+00 5.96231595e-03 5.16535752e-02
-4.93083358e-01 4.88000304e-01 3.06613863e-01 -4.59982157e-01
-1.27030027e+00 -1.71619701e+00 1.14668810e+00 -2.41276413e-01
6.03217065e-01 -1.88602373e-01 -9.57235754e-01 7.42044866e-01
1.12193719e-01 4.82210785e-01 1.37917519e+00 7.95035660e-02
-1.17022850e-01 -7.38110915e-02 -1.28852570e+00 5.05401373e-01
6.43526673e-01 -2.31495589e-01 1.06047191e-01 5.63777447e-01
5.27444661e-01 -4.80850816e-01 -1.28001440e+00 4.44021672e-01
6.48502469e-01 -8.62589717e-01 1.00225163e+00 -6.98773026e-01
6.92336321e-01 -1.86309978e-01 -8.36623535e-02 -1.44643939e+00
-8.59429061e-01 3.64614785e-01 2.36103192e-01 1.00250852e+00
6.99819505e-01 -8.07392001e-01 9.59051847e-01 5.20235062e-01
-5.57981789e-01 -1.55616891e+00 -9.86449361e-01 -2.57278502e-01
3.62577081e-01 -1.88468203e-01 5.69190443e-01 1.20248735e+00
1.22965179e-01 -8.47101361e-02 5.75260706e-02 1.60078973e-01
1.08338572e-01 -2.38420621e-01 4.06021535e-01 -1.02058387e+00
1.22072414e-01 -8.27835262e-01 -8.42673719e-01 -3.50150704e-01
2.00965494e-01 -1.38197637e+00 -1.87093914e-01 -1.36364841e+00
5.07201433e-01 -7.75643945e-01 -7.72339404e-01 8.99274170e-01
-1.82438821e-01 2.92801917e-01 -6.60215080e-01 -1.68445893e-02
-1.62565440e-01 2.63058066e-01 1.09976292e+00 -4.24610704e-01
2.34538943e-01 -1.75291523e-01 -1.10286975e+00 5.31553507e-01
1.24981904e+00 -3.06524873e-01 1.89176440e-01 -8.76279250e-02
-9.42466110e-02 -3.16483565e-02 5.33648193e-01 -8.13516796e-01
1.20764218e-01 -2.82394171e-01 9.58970785e-01 -6.43310368e-01
1.12861954e-01 -5.32135069e-01 4.33839560e-01 8.38349700e-01
-5.87840915e-01 -1.19253598e-01 4.10458207e-01 3.26876491e-01
-2.22737789e-01 -1.35653928e-01 7.10587561e-01 -1.61206350e-01
-7.04612076e-01 5.08802176e-01 -5.94898999e-01 -3.76521617e-01
1.07342911e+00 -3.07580709e-01 -5.80767453e-01 4.82984781e-02
-6.14453971e-01 4.06898409e-02 3.38650018e-01 1.66128129e-01
6.59949541e-01 -1.52576435e+00 -8.22853744e-01 2.16523111e-01
5.32537520e-01 -5.22920378e-02 1.81638151e-01 1.12300956e+00
-6.69916749e-01 6.16347671e-01 -2.03564242e-01 -5.69300890e-01
-1.19662583e+00 4.45453137e-01 8.44262660e-01 -2.53856301e-01
-2.41097927e-01 1.15950680e+00 3.25121552e-01 -6.30713820e-01
3.68973091e-02 -3.05195868e-01 -4.14705664e-01 -2.10147440e-01
4.06930566e-01 9.57650021e-02 1.48214296e-01 -5.12724161e-01
-6.23670757e-01 3.02487344e-01 -1.84326530e-01 7.85115600e-01
1.31450105e+00 4.02193964e-01 -5.07939517e-01 4.75906730e-01
1.67972624e+00 -2.87389100e-01 -4.53227311e-01 1.63883120e-01
9.63873938e-02 -1.65646881e-01 1.68252274e-01 -9.23014283e-01
-1.32813168e+00 6.63656652e-01 9.04276490e-01 -1.22525595e-01
1.13676560e+00 1.06234876e-02 6.39543951e-01 -3.78232673e-02
-1.27463952e-01 -5.74139595e-01 -6.23633802e-01 4.25502628e-01
5.12530327e-01 -1.35777581e+00 -2.20412239e-01 -3.30355912e-01
-1.73040733e-01 1.42101634e+00 8.37794483e-01 -1.47203505e-01
9.27567720e-01 5.06956279e-01 3.95019352e-01 -4.04678583e-01
-1.23345470e+00 9.69895497e-02 2.04083875e-01 7.44539082e-01
9.99140739e-01 5.03338337e-01 -3.17743063e-01 1.05727398e+00
-1.48474589e-01 2.40979120e-01 4.35714483e-01 6.14232242e-01
-3.23816985e-01 -6.17050469e-01 -1.47859648e-01 1.02481461e+00
-6.63998008e-01 -3.12047869e-01 -4.57683504e-01 1.00934899e+00
2.38226458e-01 4.05631185e-01 3.63866687e-01 -7.11637259e-01
-1.13321304e-01 1.39248952e-01 -1.65321141e-01 -4.84603614e-01
-4.37372059e-01 -9.21646953e-02 2.20850378e-01 -1.25221848e-01
-3.12667519e-01 -4.94228214e-01 -1.26795375e+00 -5.35559580e-02
-3.90064508e-01 1.38889998e-01 4.91844416e-01 4.99161422e-01
5.48230886e-01 8.08721066e-01 2.28768632e-01 -5.17528653e-01
-3.77883986e-02 -1.11371791e+00 -7.34630466e-01 1.30722374e-01
3.11391622e-01 -5.55224180e-01 -3.08269858e-01 2.06045043e-02]
|
[15.136741638183594, -2.9698455333709717]
|
b4f5fb1a-f41d-4caa-86a0-48e74ccb1133
|
priorlane-a-prior-knowledge-enhanced-lane
|
2209.06994
| null |
https://arxiv.org/abs/2209.06994v3
|
https://arxiv.org/pdf/2209.06994v3.pdf
|
PriorLane: A Prior Knowledge Enhanced Lane Detection Approach Based on Transformer
|
Lane detection is one of the fundamental modules in self-driving. In this paper we employ a transformer-only method for lane detection, thus it could benefit from the blooming development of fully vision transformer and achieve the state-of-the-art (SOTA) performance on both CULane and TuSimple benchmarks, by fine-tuning the weight fully pre-trained on large datasets. More importantly, this paper proposes a novel and general framework called PriorLane, which is used to enhance the segmentation performance of the fully vision transformer by introducing the low-cost local prior knowledge. Specifically, PriorLane utilizes an encoder-only transformer to fuse the feature extracted by a pre-trained segmentation model with prior knowledge embeddings. Note that a Knowledge Embedding Alignment (KEA) module is adapted to enhance the fusion performance by aligning the knowledge embedding. Extensive experiments on our Zjlab dataset show that PriorLane outperforms SOTA lane detection methods by a 2.82% mIoU when prior knowledge is employed, and the code will be released at: https://github.com/vincentqqb/PriorLane.
|
['Xiaofei He', 'Gang Huang', 'Wei Hua', 'Haiming Gao', 'Qibo Qiu']
|
2022-09-15
| null | null | null | null |
['lane-detection']
|
['computer-vision']
|
[-1.91436484e-01 -1.13128863e-01 -2.18026340e-01 -3.20360243e-01
-7.69735157e-01 -2.73950845e-01 5.27308643e-01 -3.67744386e-01
-4.97648925e-01 3.31695586e-01 1.67902514e-01 -3.20528448e-01
1.51734129e-01 -7.95888424e-01 -8.06846261e-01 -6.82766676e-01
3.79251450e-01 -6.46572933e-02 8.28258157e-01 -3.49906921e-01
3.10357511e-01 1.90042824e-01 -1.71706021e+00 -8.00387785e-02
1.19283533e+00 1.02942288e+00 3.68796557e-01 4.57114488e-01
-2.80234087e-02 6.60254359e-01 -7.57549405e-02 -4.30410862e-01
4.50409681e-01 5.75549230e-02 -3.59461069e-01 3.24499793e-02
7.00679123e-01 -3.72898400e-01 -5.37028372e-01 1.07797146e+00
4.90053147e-01 -3.98085825e-03 3.88066292e-01 -1.44067574e+00
-2.80742049e-01 2.98959404e-01 -7.86743164e-01 2.31407776e-01
-2.05744863e-01 5.35592258e-01 8.91287565e-01 -1.11909556e+00
3.66669387e-01 1.07223189e+00 6.90214992e-01 -1.46993082e-02
-7.32624531e-01 -7.52123356e-01 6.42933249e-02 8.22000742e-01
-1.56064975e+00 -5.57145536e-01 1.00314116e+00 -4.70281065e-01
7.90964544e-01 -7.31118023e-02 5.71443677e-01 9.40777242e-01
3.69568527e-01 9.46295798e-01 1.13646519e+00 -2.16922998e-01
-1.02123894e-01 2.28370115e-01 4.24205065e-01 1.27751708e+00
3.34219903e-01 2.70184815e-01 -4.44069147e-01 3.13578606e-01
6.63709164e-01 -1.37232453e-01 -2.19643787e-01 -7.96627402e-01
-1.43853354e+00 8.37570071e-01 4.62447137e-01 -9.14090499e-02
-8.41490328e-02 1.07741684e-01 5.47635317e-01 -9.37454626e-02
1.15571491e-01 4.60405387e-02 -9.77675915e-02 -1.37537226e-01
-8.24836135e-01 6.67949468e-02 4.87248212e-01 1.23143148e+00
1.24862254e+00 4.45128828e-02 -2.13716835e-01 6.98906660e-01
4.35440749e-01 6.13717020e-01 3.97109926e-01 -1.13683939e+00
5.22997975e-01 6.59783304e-01 -1.73627988e-01 -8.26890886e-01
-1.96787879e-01 -6.27193332e-01 -5.47139287e-01 3.18895429e-01
1.51563361e-01 -4.96958680e-02 -1.02090788e+00 1.42998946e+00
3.09519231e-01 6.91614091e-01 1.60987124e-01 9.35332298e-01
6.73770189e-01 5.54614544e-01 -1.62085518e-01 2.15166926e-01
1.31826746e+00 -1.76665556e+00 -7.27424085e-01 -5.64671934e-01
6.17548823e-01 -8.18442702e-01 1.11692119e+00 2.80370355e-01
-6.83377802e-01 -9.17153299e-01 -1.56049705e+00 -4.63774145e-01
-6.41455054e-01 5.36693037e-01 5.04253685e-01 4.50519800e-01
-8.27589214e-01 6.60552830e-03 -6.48826957e-01 -3.79150242e-01
3.44077915e-01 -3.58022712e-02 -3.05542260e-01 -2.00480476e-01
-1.21538353e+00 9.57345605e-01 5.48438311e-01 3.05531502e-01
-8.75155687e-01 -5.55962503e-01 -1.07244861e+00 -1.56862751e-01
6.53418481e-01 -6.39313340e-01 1.00521207e+00 -2.36500055e-01
-1.47455180e+00 6.27294481e-01 -2.00116083e-01 -5.04535258e-01
4.95677769e-01 -4.23826545e-01 -3.90327126e-01 1.39435083e-01
4.33223754e-01 8.45295072e-01 8.42124343e-01 -1.21985161e+00
-1.03095710e+00 -1.52105674e-01 3.61249335e-02 1.52166292e-01
-2.71705091e-01 -5.31753361e-01 -9.93593037e-01 -4.62711334e-01
-1.85356885e-01 -9.38367367e-01 -1.27333403e-01 -9.33545902e-02
-3.20869029e-01 -2.37372130e-01 1.20583904e+00 -6.48975611e-01
1.37834108e+00 -2.33070564e+00 -1.28372118e-01 3.80430892e-02
3.32615763e-01 4.19185162e-01 -1.01612002e-01 2.28771925e-01
4.33633655e-01 -3.59962523e-01 -3.71370226e-01 -3.21287900e-01
1.58326074e-01 2.72286922e-01 -3.26419085e-01 3.98809314e-01
1.25229239e-01 1.01556325e+00 -8.24941099e-01 -7.21242428e-01
6.85476005e-01 3.51558328e-01 -3.78128737e-01 7.10900798e-02
1.61521912e-01 1.39563102e-02 -3.64856213e-01 7.21505523e-01
9.23378348e-01 -4.17761505e-02 -2.43466184e-01 -5.35184145e-01
-5.35186172e-01 -4.00127135e-02 -1.09273803e+00 1.96012747e+00
-3.59998018e-01 7.31080055e-01 -3.50804883e-03 -7.78167605e-01
1.01768935e+00 -2.64667988e-01 1.61256567e-01 -7.61699915e-01
3.40152711e-01 2.05823421e-01 -1.46569207e-01 -4.70551580e-01
5.92686772e-01 3.64336073e-01 -2.73107231e-01 -1.97176099e-01
-7.50748292e-02 5.59493378e-02 2.37214714e-01 2.46034771e-01
9.71277952e-01 2.63444930e-01 5.32032996e-02 -3.09628218e-01
8.63384426e-01 2.88425356e-01 8.55319262e-01 5.56940854e-01
-7.19562232e-01 3.68547767e-01 2.10313171e-01 -2.26254910e-02
-8.48609209e-01 -1.23397529e+00 -1.60326228e-01 8.44630539e-01
6.76254272e-01 -3.83509219e-01 -7.32993603e-01 -9.07435656e-01
2.25928038e-01 7.29730248e-01 -3.97902906e-01 -3.57811034e-01
-5.79725802e-01 -3.57217103e-01 6.61376655e-01 8.26490462e-01
1.19847989e+00 -5.42136371e-01 -6.44956172e-01 1.41392797e-01
-1.05455965e-01 -1.43011940e+00 -6.50948584e-01 2.81267129e-02
-4.81261671e-01 -1.07122076e+00 -5.01987398e-01 -8.69511843e-01
3.59536558e-01 7.59162545e-01 5.25486052e-01 -2.48425901e-01
-1.74433649e-01 3.46765429e-01 -2.86132365e-01 -3.24592352e-01
1.07325137e-01 2.13824764e-01 -1.73480719e-01 1.08952351e-01
6.06451869e-01 -2.32772335e-01 -6.07627392e-01 4.15084690e-01
-4.68035549e-01 5.08018970e-01 1.03528512e+00 9.11146343e-01
5.37865639e-01 -5.09030409e-02 3.86375695e-01 -5.17655969e-01
2.99538821e-01 -3.37553054e-01 -7.25046754e-01 8.37571472e-02
-8.41557741e-01 5.37315384e-02 3.98077846e-01 -3.25295962e-02
-1.12908363e+00 2.26514712e-01 -1.98938757e-01 -8.13490093e-01
-1.79606646e-01 3.75540733e-01 -3.84143561e-01 -1.89779416e-01
2.56616414e-01 3.38495612e-01 1.50528163e-01 -4.76983488e-01
5.93909860e-01 8.13299417e-01 8.45017910e-01 -1.64767176e-01
9.26389575e-01 6.28058374e-01 -1.15436219e-01 -5.93727946e-01
-8.33712757e-01 -7.65968502e-01 -7.22592890e-01 -2.45282993e-01
9.96169925e-01 -1.17365503e+00 -4.83719468e-01 6.75622463e-01
-7.83914208e-01 -3.46756458e-01 1.09681524e-01 6.26011133e-01
-5.55745423e-01 5.12780547e-01 -2.34002843e-01 -2.83845216e-01
-1.62301257e-01 -1.37751102e+00 9.76847827e-01 3.91021430e-01
6.26190901e-01 -7.90901244e-01 3.75428759e-02 7.53866255e-01
4.13104177e-01 -2.23581474e-02 4.50428873e-01 -1.93166614e-01
-8.81464481e-01 -4.48478200e-02 -6.38354838e-01 6.09068394e-01
5.14063723e-02 -7.08505362e-02 -1.12848651e+00 -2.11188495e-01
-4.39099401e-01 1.66811757e-02 1.45195043e+00 5.95592894e-02
1.00109839e+00 2.13470221e-01 -4.77218539e-01 7.59115636e-01
1.46769726e+00 5.29722199e-02 6.57186747e-01 6.18082583e-01
1.21866024e+00 3.70484173e-01 9.94146705e-01 7.49222264e-02
1.16901040e+00 6.56077862e-01 4.08812553e-01 -1.77066296e-01
-4.96614039e-01 -3.42348397e-01 6.37674868e-01 9.92377639e-01
2.73179412e-01 3.86536978e-02 -9.57699537e-01 7.57844627e-01
-2.20185494e+00 -7.90405393e-01 -2.19184443e-01 1.89614189e+00
6.09229147e-01 5.84858179e-01 -1.76061671e-02 1.32908329e-01
5.92797041e-01 3.43962252e-01 -6.55692160e-01 -3.17748964e-01
3.84151638e-02 -2.80644089e-01 1.06483996e+00 7.39072680e-01
-1.41775501e+00 1.31450152e+00 5.28179979e+00 1.14142358e+00
-1.24979830e+00 2.15657309e-01 8.30446035e-02 3.09876978e-01
-9.28455293e-02 1.83412060e-01 -1.19257021e+00 5.15372872e-01
7.82232702e-01 -2.31643155e-01 1.32625729e-01 9.26431954e-01
3.00626040e-01 -3.32144976e-01 -5.91595888e-01 9.29195702e-01
2.67333478e-01 -1.25779796e+00 -1.33508638e-01 1.15245037e-01
5.44857502e-01 3.07662100e-01 4.26985659e-02 7.96293139e-01
-1.59637909e-03 -4.48819876e-01 7.35365629e-01 7.05348670e-01
6.94777787e-01 -8.13139260e-01 9.71225142e-01 4.04940099e-01
-1.56401360e+00 -8.97340253e-02 -3.12759399e-01 1.05628528e-01
2.77520061e-01 6.08316898e-01 -8.10837865e-01 7.58369505e-01
6.81956887e-01 1.07112849e+00 -9.68575060e-01 1.15916765e+00
-3.62908483e-01 7.38457322e-01 -2.67344892e-01 3.73557806e-01
4.99398559e-01 -1.83191791e-01 4.88408178e-01 1.39573383e+00
1.43513009e-01 -3.71933639e-01 4.45424050e-01 6.88596606e-01
2.41935804e-01 -1.25348195e-01 -5.53467214e-01 3.05076659e-01
5.13706684e-01 1.47933102e+00 -2.81797558e-01 -4.63944346e-01
-7.85899401e-01 8.18234801e-01 1.80557519e-01 2.61438727e-01
-1.27033329e+00 -8.78097057e-01 6.14924788e-01 3.52009460e-02
6.90044582e-01 -3.73206407e-01 -2.84507215e-01 -1.08891988e+00
3.84314707e-03 -3.90180171e-01 1.10174634e-01 -8.60106111e-01
-1.00624943e+00 3.16252589e-01 7.02481344e-02 -1.42922938e+00
8.44954327e-02 -7.43022084e-01 -6.35466218e-01 7.23782957e-01
-2.28541660e+00 -1.54765439e+00 -7.97956884e-01 5.45404673e-01
6.67633533e-01 -1.87131017e-01 2.34380111e-01 5.07602394e-01
-1.00953770e+00 7.86811650e-01 -8.19856301e-03 1.39417529e-01
9.56557274e-01 -1.19345379e+00 2.21650690e-01 1.10552394e+00
-2.58285314e-01 2.96510309e-01 5.38584709e-01 -4.71744299e-01
-1.52711201e+00 -1.41958201e+00 6.25938058e-01 -3.73507768e-01
7.43949831e-01 -3.07004005e-01 -8.09083402e-01 6.64487541e-01
3.18552822e-01 1.73140198e-01 3.75836611e-01 -3.35969836e-01
-3.49679410e-01 -4.86545831e-01 -9.21672344e-01 5.02435029e-01
1.07976532e+00 -4.40473676e-01 -7.57908046e-01 -2.99519539e-01
7.02552617e-01 -3.82760018e-01 -8.22205782e-01 7.05221415e-01
4.91177112e-01 -1.04279113e+00 8.39183688e-01 2.71475732e-01
1.81173816e-01 -9.61058080e-01 -3.46430950e-02 -9.58028316e-01
-4.67084914e-01 -2.03512773e-01 -2.88899958e-01 1.19764078e+00
1.49955213e-01 -7.87948251e-01 6.50182962e-01 -1.29473954e-01
-6.92294240e-01 -8.42944801e-01 -9.26099837e-01 -8.41462791e-01
-8.46226811e-02 -4.10406828e-01 5.13367534e-01 6.14504933e-01
-1.92041010e-01 3.88833880e-01 -2.31941238e-01 4.30514783e-01
8.84388328e-01 8.37253127e-03 9.83640909e-01 -9.22740817e-01
8.00948665e-02 -3.83008838e-01 -5.75267076e-01 -1.52444863e+00
1.48713678e-01 -1.00607955e+00 2.12920412e-01 -1.66338718e+00
1.49948627e-01 -2.78336704e-01 -5.48771322e-01 6.40085697e-01
-1.65734634e-01 1.95697054e-01 2.00509027e-01 1.59909144e-01
-9.19286609e-01 8.80494356e-01 1.33496618e+00 -1.98077008e-01
3.06538735e-02 -3.96753013e-01 -7.41958916e-01 8.12737226e-01
9.64668632e-01 -1.40299788e-02 -6.08788073e-01 -5.32122076e-01
-2.75582224e-01 -3.83423626e-01 6.98651671e-01 -1.38237596e+00
5.98846555e-01 -3.01791579e-02 8.52445662e-02 -1.16598117e+00
3.60877663e-01 -7.05657542e-01 -3.58952671e-01 3.97772700e-01
2.05880940e-01 7.34997028e-03 4.75634724e-01 7.15840816e-01
-2.81245023e-01 1.08790487e-01 6.80222631e-01 2.12873757e-01
-1.68829465e+00 1.48693547e-01 -2.65962809e-01 -8.09771791e-02
1.24757600e+00 -5.14239967e-01 -7.25412488e-01 5.61369509e-02
-1.87039778e-01 7.70746410e-01 4.84128356e-01 6.87058628e-01
7.65471458e-01 -1.48328447e+00 -3.81212115e-01 4.38917905e-01
5.77628076e-01 -6.46529272e-02 2.35202059e-01 1.19524062e+00
-4.72179085e-01 5.44014454e-01 -3.95481080e-01 -7.04662740e-01
-8.92004371e-01 4.57765818e-01 1.57910109e-01 -8.35642889e-02
-6.97780848e-01 5.40861845e-01 1.27913147e-01 -3.64710987e-01
1.70044720e-01 -2.82276511e-01 -2.75855780e-01 -1.42237708e-01
3.33891958e-01 6.51909590e-01 -7.09213465e-02 -8.17281187e-01
-4.90663648e-01 8.17251682e-01 -2.27184623e-01 -5.34644164e-03
8.33102465e-01 -4.17748690e-01 2.43795946e-01 3.69612426e-01
1.06925786e+00 1.97384760e-01 -1.59813881e+00 -2.59655833e-01
-1.46089032e-01 -4.27918583e-01 5.88166177e-01 -6.31790698e-01
-1.09458661e+00 9.93076146e-01 8.13228846e-01 -3.55212063e-01
1.03622651e+00 -2.54779398e-01 1.17898691e+00 2.87376881e-01
4.71139878e-01 -1.20544243e+00 -4.54847738e-02 6.82011545e-01
5.90010047e-01 -1.40304565e+00 -1.96450308e-01 -6.21128321e-01
-7.36822486e-01 9.41739261e-01 9.36300874e-01 -3.45617682e-01
7.25073874e-01 2.42556885e-01 2.04380915e-01 -1.04973197e-01
-6.16452932e-01 -6.80396438e-01 2.50885457e-01 4.00700629e-01
-1.13271780e-01 5.63235469e-02 -5.15349030e-01 5.41316628e-01
-5.46337962e-02 7.34294727e-02 4.42329109e-01 8.98079991e-01
-7.88488686e-01 -9.39056098e-01 -1.21706694e-01 3.53528321e-01
3.00161034e-01 -1.33864228e-02 4.42449152e-02 7.37358034e-01
6.06267631e-01 8.04183245e-01 -7.90314823e-02 -8.20093572e-01
4.54804331e-01 -9.37359482e-02 1.07262313e-01 -3.58880699e-01
3.64200212e-03 9.32483748e-02 6.10892884e-02 -8.06017578e-01
-1.60730734e-01 -5.66667080e-01 -1.33032119e+00 -3.45909148e-01
-3.30826074e-01 -1.51471302e-01 5.32217920e-01 7.67173946e-01
6.84734225e-01 6.17592454e-01 6.74180806e-01 -7.73638070e-01
-3.31532627e-01 -7.63360143e-01 -3.47959638e-01 -3.88555527e-02
2.91741967e-01 -1.12946713e+00 -4.19360399e-01 -7.57373646e-02]
|
[8.082282066345215, -1.4608144760131836]
|
4d80eee8-e797-4599-8e45-540065c8cc71
|
temporal-spatial-feature-pyramid-for-video
|
2105.04213
| null |
https://arxiv.org/abs/2105.04213v2
|
https://arxiv.org/pdf/2105.04213v2.pdf
|
Temporal-Spatial Feature Pyramid for Video Saliency Detection
|
Multi-level features are important for saliency detection. Better combination and use of multi-level features with time information can greatly improve the accuracy of the video saliency model. In order to fully combine multi-level features and make it serve the video saliency model, we propose a 3D fully convolutional encoder-decoder architecture for video saliency detection, which combines scale, space and time information for video saliency modeling. The encoder extracts multi-scale temporal-spatial features from the input continuous video frames, and then constructs temporal-spatial feature pyramid through temporal-spatial convolution and top-down feature integration. The decoder performs hierarchical decoding of temporal-spatial features from different scales, and finally produces a saliency map from the integration of multiple video frames. Our model is simple yet effective, and can run in real time. We perform abundant experiments, and the results indicate that the well-designed structure can improve the precision of video saliency detection significantly. Experimental results on three purely visual video saliency benchmarks and six audio-video saliency benchmarks demonstrate that our method outperforms the existing state-of-the-art methods.
|
['Shiping Zhu', 'Qinyao Chang']
|
2021-05-10
| null | null | null | null |
['video-saliency-detection']
|
['computer-vision']
|
[ 2.45320588e-01 -6.05274379e-01 -5.38782299e-01 -1.71694696e-01
-6.52127206e-01 5.40192202e-02 1.50862515e-01 -1.01195388e-01
-2.93915808e-01 3.40203613e-01 5.64142883e-01 9.29100737e-02
3.70821357e-01 -4.03443605e-01 -8.31549704e-01 -3.84455949e-01
-2.98237622e-01 -5.84852397e-01 1.28277647e+00 -2.47540265e-01
4.95729148e-01 -2.51929667e-02 -2.08005786e+00 5.60511708e-01
9.06150460e-01 1.37187135e+00 8.17260623e-01 6.56424820e-01
2.27870066e-02 1.05625665e+00 -1.95814073e-01 3.18702519e-01
8.91038328e-02 -3.47304106e-01 -6.25910699e-01 1.07363939e-01
3.28411937e-01 -5.68636954e-01 -5.98096371e-01 1.02963066e+00
4.53741938e-01 3.35388519e-02 1.65862143e-01 -1.49075377e+00
-7.22701311e-01 3.60259205e-01 -7.18066573e-01 6.69114411e-01
6.01587653e-01 6.50925376e-03 8.56486917e-01 -1.16310692e+00
4.05249983e-01 1.34902048e+00 5.53426087e-01 3.00424546e-01
-6.26019239e-01 -6.37611449e-01 4.08772290e-01 7.46889114e-01
-1.33134973e+00 -3.03433657e-01 8.74538362e-01 -2.91515678e-01
7.97587097e-01 1.56238541e-01 9.34163630e-01 6.00985348e-01
2.22180083e-01 1.46296430e+00 7.51328468e-01 -1.08043879e-01
-1.02109402e-01 -2.80697227e-01 -8.99403095e-02 7.53761530e-01
-1.21316023e-01 8.72844383e-02 -1.05537653e+00 7.37013668e-02
1.20030594e+00 4.53974187e-01 -2.53337115e-01 -4.98241782e-01
-1.61348796e+00 6.14108324e-01 9.52297568e-01 2.57266521e-01
-4.02163863e-01 2.71568626e-01 4.81799453e-01 3.58643010e-02
4.28208888e-01 -9.13933590e-02 -3.57999563e-01 -1.60500348e-01
-1.41019547e+00 2.32642248e-01 4.24300246e-02 1.20572889e+00
8.76494825e-01 3.23568791e-01 -4.74890530e-01 5.58342636e-01
2.57621229e-01 6.89125001e-01 6.79699957e-01 -1.07227921e+00
2.41451189e-01 6.34518027e-01 1.18365042e-01 -1.15148962e+00
-2.96736538e-01 -4.38633673e-02 -4.83001590e-01 7.63884140e-03
-1.23500407e-01 1.83377951e-01 -9.97236073e-01 1.44221759e+00
2.00612813e-01 7.66091466e-01 -1.91812247e-01 1.49591362e+00
1.28345597e+00 7.06848681e-01 1.20848507e-01 -2.22171322e-01
1.35634148e+00 -1.27826583e+00 -6.76263094e-01 -1.83490053e-01
4.77733612e-02 -7.49194801e-01 1.00066364e+00 -2.38912106e-01
-1.25557256e+00 -9.01660621e-01 -9.90794301e-01 -4.19893324e-01
-2.05288500e-01 2.38363668e-01 6.97725236e-01 -8.04331824e-02
-1.25954545e+00 2.75587857e-01 -6.99055672e-01 -1.12489693e-01
4.25443560e-01 4.01090473e-01 -5.49293831e-02 1.56831726e-01
-1.45592546e+00 6.32321358e-01 3.44697297e-01 -3.28418687e-02
-1.15322828e+00 -5.41132927e-01 -1.22149205e+00 1.76759377e-01
3.45934153e-01 -4.35081691e-01 1.37631202e+00 -1.28474450e+00
-1.07198584e+00 5.15680671e-01 -8.90671849e-01 -3.39111388e-01
9.29228067e-02 -2.36747816e-01 -4.67886478e-01 6.42041504e-01
5.24152696e-01 1.25397563e+00 1.34142983e+00 -8.20333779e-01
-1.20307994e+00 1.40572727e-01 8.44688863e-02 3.71330529e-01
-2.42501348e-01 4.95975882e-01 -7.09765077e-01 -6.93344891e-01
2.80770153e-01 -5.15513420e-01 -1.47454783e-01 6.99365214e-02
-3.16009969e-02 -2.07205620e-02 1.18476689e+00 -7.51508713e-01
1.36747503e+00 -2.32702756e+00 2.15361044e-01 -3.47104609e-01
2.66287923e-01 1.53405055e-01 -1.10219218e-01 -1.88452408e-01
6.14630543e-02 -2.13807687e-01 -1.90353692e-02 2.97102816e-02
-3.70328069e-01 -3.86111766e-01 -3.79597694e-01 1.61610067e-01
4.31814969e-01 1.29356563e+00 -1.15049374e+00 -9.67266202e-01
3.74046803e-01 4.02790070e-01 -6.56053245e-01 1.79672062e-01
-8.72858912e-02 1.12982802e-01 -6.14215136e-01 1.04219508e+00
4.50256139e-01 -2.99368441e-01 -4.90885228e-01 -2.09335566e-01
-4.01320457e-01 3.89315516e-01 -8.47333729e-01 1.84876072e+00
1.29746571e-01 9.90297556e-01 -2.85604596e-01 -6.52857661e-01
8.13137770e-01 1.63048416e-01 4.88087088e-01 -9.06820059e-01
-1.26928724e-02 2.55327523e-01 -3.92354250e-01 -4.98660445e-01
9.53844070e-01 3.20294291e-01 -1.67128295e-01 -1.33689418e-01
1.87690958e-01 3.12141096e-03 1.09700717e-01 3.40830535e-01
7.65910923e-01 2.73051023e-01 2.74697512e-01 -3.11959565e-01
6.45583689e-01 5.72393164e-02 8.56414258e-01 4.47471142e-01
-7.00770378e-01 8.75793457e-01 3.53452533e-01 -4.62300092e-01
-9.71235096e-01 -1.10946596e+00 3.94423813e-01 1.39962363e+00
9.19655323e-01 -4.66341555e-01 -6.24504983e-01 -5.57281554e-01
-1.16001688e-01 1.03849776e-01 -6.71762884e-01 -3.28341067e-01
-5.92313230e-01 -9.95290205e-02 5.17235398e-02 7.11088836e-01
7.42695093e-01 -1.31643522e+00 -9.61305737e-01 2.58434266e-01
-5.06393492e-01 -1.02088296e+00 -9.86967027e-01 -2.48663276e-01
-9.56921935e-01 -9.91740823e-01 -9.01766062e-01 -1.38079703e+00
2.96914309e-01 1.25041270e+00 8.45572770e-01 1.89988762e-01
-2.87302434e-01 8.96086767e-02 -4.52741474e-01 -2.72956699e-01
2.57632017e-01 -9.22330469e-02 -5.45503721e-02 -8.97331610e-02
5.12425959e-01 -3.72558445e-01 -7.51874447e-01 4.60937232e-01
-8.58602107e-01 5.47277391e-01 4.80327159e-01 7.17381954e-01
5.75461507e-01 -3.72494131e-01 6.88353658e-01 4.58785519e-02
1.95943102e-01 -3.33029866e-01 -4.61074352e-01 1.78463325e-01
4.46522087e-02 -2.39593238e-01 1.39237642e-01 -4.74739820e-01
-7.67020822e-01 5.51125668e-02 2.53117174e-01 -7.78515160e-01
1.06599040e-01 3.25562328e-01 -1.69938728e-02 -2.57317036e-01
2.72567987e-01 7.90884614e-01 -7.65353665e-02 -2.53203601e-01
2.24912658e-01 6.28911138e-01 6.41896963e-01 -4.53589670e-03
5.02445042e-01 4.39412177e-01 -2.21525908e-01 -8.02558064e-01
-8.55316997e-01 -5.22714555e-01 -7.41457582e-01 -4.98562634e-01
8.62857342e-01 -1.54091954e+00 -3.67932528e-01 6.19201243e-01
-1.08271384e+00 -1.62144050e-01 -9.21145007e-02 4.41334128e-01
-6.77107871e-01 4.13814723e-01 -7.02255547e-01 -3.58008713e-01
-2.12536454e-01 -1.48140836e+00 1.43315113e+00 6.69767201e-01
7.58625790e-02 -6.45186424e-01 -4.59877014e-01 -5.81974816e-03
5.98582387e-01 -6.11003488e-02 2.87519336e-01 8.19888487e-02
-1.08637857e+00 1.35006905e-01 -6.33133352e-01 3.28672081e-02
1.38615593e-01 7.77816102e-02 -7.81860352e-01 -7.63352439e-02
-5.97638115e-02 -2.64490873e-01 1.28350163e+00 8.41361582e-01
1.24097908e+00 1.96676534e-02 -4.41780567e-01 4.78155583e-01
1.13867772e+00 6.93511590e-02 5.87673366e-01 4.43328381e-01
8.38353574e-01 1.50421351e-01 1.26447487e+00 3.27160001e-01
8.71689975e-01 6.67712450e-01 5.55311143e-01 -2.55044103e-01
-2.56511331e-01 -3.72380883e-01 7.02999711e-01 9.56979871e-01
5.29698329e-03 4.97362703e-01 -4.22041178e-01 9.91961420e-01
-2.11110616e+00 -1.34233689e+00 -1.05209857e-01 1.79785669e+00
9.34496105e-01 2.23222792e-01 3.93366992e-01 5.57101984e-03
9.76240337e-01 4.21745986e-01 -5.70764244e-01 -1.61008134e-01
-2.74376661e-01 -2.59452939e-01 3.84369045e-01 2.84012467e-01
-1.50183117e+00 1.33010674e+00 7.46194601e+00 9.02524471e-01
-1.31635833e+00 1.65640980e-01 4.06971812e-01 -3.63874197e-01
-3.65004867e-01 -9.50336307e-02 -7.22533882e-01 7.43238091e-01
6.68417692e-01 -3.63662034e-01 2.47940198e-01 1.05134594e+00
3.52543741e-01 -3.15567493e-01 -8.13927770e-01 1.14281845e+00
2.02237710e-01 -1.59543180e+00 5.00770332e-03 -4.24636006e-01
9.03844178e-01 2.67498463e-01 2.00071946e-01 3.37983608e-01
-1.52817398e-01 -7.05335855e-01 1.16764057e+00 5.40202022e-01
6.96766615e-01 -5.78433573e-01 4.69094247e-01 5.28968051e-02
-1.85663235e+00 -1.48148492e-01 -4.59788203e-01 -2.16529787e-01
2.32448339e-01 4.86954361e-01 -4.06223536e-01 2.08272427e-01
9.87088978e-01 1.40329230e+00 -7.67420769e-01 1.44035959e+00
-1.25185236e-01 3.70768547e-01 -1.52233183e-01 -2.16801658e-01
3.02287817e-01 3.09480608e-01 5.12206733e-01 1.36388695e+00
2.65543133e-01 -3.37557048e-02 3.78266752e-01 5.70750177e-01
2.55698770e-01 -1.43417448e-01 -3.44690889e-01 2.10215926e-01
4.90294367e-01 1.06240571e+00 -6.13930523e-01 -5.93793094e-01
-7.28624523e-01 9.73406315e-01 1.31955758e-01 3.00159782e-01
-1.22315395e+00 -4.85962152e-01 7.65477061e-01 -1.36677802e-01
7.44177401e-01 -2.07587302e-01 -2.52804309e-01 -1.33240402e+00
1.59594715e-01 -5.65803766e-01 1.45442814e-01 -1.19821787e+00
-5.52213252e-01 4.22821730e-01 -2.08999529e-01 -1.63461637e+00
-1.37819752e-01 -1.93183213e-01 -5.71739197e-01 7.09413707e-01
-1.92882550e+00 -1.16185617e+00 -4.25794482e-01 8.76831532e-01
9.84463096e-01 -2.22423971e-01 3.12712282e-01 2.00165167e-01
-2.52304316e-01 2.79082924e-01 -2.77849525e-01 3.64549346e-02
5.46581507e-01 -9.06070650e-01 5.05096138e-01 1.05870724e+00
6.35823526e-04 1.49051100e-01 5.33261418e-01 -7.10433424e-01
-1.31308627e+00 -1.09371686e+00 1.01366484e+00 -1.16608523e-01
6.15880072e-01 -1.54920518e-01 -8.91870081e-01 4.18811828e-01
9.76242200e-02 9.07115638e-02 2.58994371e-01 -4.50593263e-01
-2.26829156e-01 -9.05630887e-02 -8.15468371e-01 7.60952592e-01
1.13914871e+00 -7.80456245e-01 -9.63482082e-01 -7.16186240e-02
1.43851268e+00 -4.86431003e-01 -3.40581417e-01 5.10078549e-01
4.71869975e-01 -1.05148649e+00 1.11220443e+00 -3.15732419e-01
7.85548210e-01 -7.12400496e-01 -1.60932630e-01 -9.65854824e-01
-6.80990398e-01 -4.02712405e-01 -3.49965215e-01 8.29202116e-01
1.16411790e-01 1.31085664e-02 5.09014666e-01 1.12451166e-01
-3.54659975e-01 -7.79874027e-01 -9.55744684e-01 -4.23838109e-01
-7.38438725e-01 -3.40558261e-01 4.93585616e-01 3.92938524e-01
2.99984843e-01 5.99561967e-02 -5.59865832e-01 1.69169739e-01
4.94519264e-01 5.69591284e-01 4.33796585e-01 -8.95624816e-01
2.73471624e-01 -5.98070681e-01 -7.72122741e-01 -1.47242439e+00
8.42004418e-02 -4.62766021e-01 2.38482893e-01 -1.45707726e+00
5.32301664e-01 7.97377974e-02 -6.51884258e-01 5.20561874e-01
-6.30411625e-01 5.29481828e-01 2.42742777e-01 2.65561014e-01
-1.28894126e+00 8.43203247e-01 1.51793623e+00 -1.37863696e-01
-1.64171532e-01 -3.25667590e-01 -7.17612863e-01 6.85715020e-01
5.79167247e-01 -2.17222705e-01 -2.48904750e-01 -3.68821740e-01
-2.74117351e-01 1.30440071e-01 5.82065761e-01 -1.19818592e+00
3.77062827e-01 -4.68089193e-01 6.31011307e-01 -9.65855539e-01
4.67833221e-01 -6.57805979e-01 -4.31455046e-01 3.83410424e-01
-2.35795796e-01 -5.25169037e-02 2.65199542e-01 4.08487052e-01
-6.91950500e-01 2.44267941e-01 6.33348227e-01 -1.25597194e-02
-1.54264176e+00 4.41425204e-01 -3.10810626e-01 -2.43612472e-02
1.07976270e+00 -4.35307711e-01 -6.29924014e-02 -4.04979676e-01
-5.07361948e-01 4.34787214e-01 6.14021778e-01 9.71328616e-01
1.15277719e+00 -1.65330589e+00 -4.88007247e-01 3.96077067e-01
6.21692725e-02 -2.55014330e-01 3.74521762e-01 9.06177998e-01
-4.49032217e-01 4.84385878e-01 -6.01442575e-01 -1.00178766e+00
-1.33732343e+00 7.96507478e-01 1.39459699e-01 4.28678751e-01
-4.26313847e-01 9.18195486e-01 4.59126979e-01 2.82145500e-01
2.98733413e-01 -5.68725467e-01 -3.18696916e-01 -1.08831689e-01
9.33233202e-01 1.65762872e-01 -4.93824333e-01 -1.02069175e+00
-5.47004402e-01 6.33954346e-01 -9.46673378e-03 2.12796237e-02
1.19839323e+00 -4.57820654e-01 -9.00096744e-02 5.84630430e-01
1.13479209e+00 -3.17066729e-01 -1.74242771e+00 -4.33154941e-01
-2.16883272e-01 -7.64475524e-01 5.64486049e-02 -3.02826852e-01
-1.10807574e+00 9.08914387e-01 4.02536541e-01 9.12735984e-02
1.53526211e+00 -4.51654159e-02 1.02125573e+00 -1.85018748e-01
5.86361051e-01 -1.09147370e+00 4.31582034e-01 7.15066075e-01
8.46738577e-01 -1.44113708e+00 -5.49297519e-02 -6.04521930e-01
-8.58634055e-01 8.75473738e-01 8.05157721e-01 -4.07045275e-01
6.47539675e-01 -4.79907580e-02 -1.39696598e-01 8.66358727e-02
-7.22985029e-01 -4.89062279e-01 6.45367563e-01 4.46370453e-01
2.00749174e-01 6.38478547e-02 -1.59178764e-01 6.49824917e-01
1.14321716e-01 2.80434251e-01 4.05865252e-01 1.01757133e+00
-1.08495998e+00 -6.06103063e-01 -2.31659070e-01 3.15135628e-01
-3.16157103e-01 -3.59998852e-01 -9.49132890e-02 3.43640119e-01
9.17076245e-02 8.86355042e-01 1.39548123e-01 -5.80215275e-01
8.66181329e-02 -2.55272657e-01 1.71097741e-01 -3.51788729e-01
-2.95969486e-01 3.72264266e-01 -3.37271303e-01 -9.91377711e-01
-7.70868003e-01 -7.08044231e-01 -1.43284953e+00 -8.55809450e-02
-1.83904767e-01 3.30968052e-02 4.18099910e-01 8.58542502e-01
5.72908759e-01 6.22958243e-01 7.45495081e-01 -1.45684505e+00
-9.02860686e-02 -8.04026484e-01 -4.55728084e-01 2.43828461e-01
7.86511838e-01 -9.07640815e-01 -1.20513327e-01 2.74649084e-01]
|
[9.72631549835205, -0.3171845078468323]
|
49eb060f-2813-4ec7-9dd5-b5e4a6f7735b
|
perspective-transformer-nets-learning-single
|
1612.00814
| null |
http://arxiv.org/abs/1612.00814v3
|
http://arxiv.org/pdf/1612.00814v3.pdf
|
Perspective Transformer Nets: Learning Single-View 3D Object Reconstruction without 3D Supervision
|
Understanding the 3D world is a fundamental problem in computer vision.
However, learning a good representation of 3D objects is still an open problem
due to the high dimensionality of the data and many factors of variation
involved. In this work, we investigate the task of single-view 3D object
reconstruction from a learning agent's perspective. We formulate the learning
process as an interaction between 3D and 2D representations and propose an
encoder-decoder network with a novel projection loss defined by the perspective
transformation. More importantly, the projection loss enables the unsupervised
learning using 2D observation without explicit 3D supervision. We demonstrate
the ability of the model in generating 3D volume from a single 2D image with
three sets of experiments: (1) learning from single-class objects; (2) learning
from multi-class objects and (3) testing on novel object classes. Results show
superior performance and better generalization ability for 3D object
reconstruction when the projection loss is involved.
|
['Ersin Yumer', 'Xinchen Yan', 'Yijie Guo', 'Honglak Lee', 'Jimei Yang']
|
2016-12-01
|
perspective-transformer-nets-learning-single-1
|
http://papers.nips.cc/paper/6206-perspective-transformer-nets-learning-single-view-3d-object-reconstruction-without-3d-supervision
|
http://papers.nips.cc/paper/6206-perspective-transformer-nets-learning-single-view-3d-object-reconstruction-without-3d-supervision.pdf
|
neurips-2016-12
|
['3d-object-reconstruction']
|
['computer-vision']
|
[ 2.87032813e-01 3.03125829e-01 -8.59875306e-02 -7.01416492e-01
-6.09374166e-01 -2.88535923e-01 8.20262969e-01 -4.27678555e-01
-1.33579522e-01 3.62941891e-01 1.26715049e-01 -3.21379327e-03
-1.36530166e-02 -7.02009916e-01 -1.01837695e+00 -6.99888587e-01
3.97517197e-02 7.62068391e-01 1.65723264e-01 2.62947410e-01
2.01785669e-01 8.93225431e-01 -1.51805794e+00 1.68785900e-01
4.65505362e-01 1.02071595e+00 4.07818526e-01 5.32867491e-01
-3.30714062e-02 4.35650527e-01 -3.34046453e-01 -1.85749575e-01
8.13219488e-01 -4.16089177e-01 -5.35664082e-01 7.56648004e-01
5.16352117e-01 -5.05845845e-01 -1.67425424e-01 9.35147166e-01
4.62904781e-01 7.79941082e-02 1.03083849e+00 -1.15633285e+00
-9.33344603e-01 4.20852378e-02 -5.76256573e-01 -7.69168735e-02
3.73427004e-01 9.42232013e-02 8.41761947e-01 -1.19068682e+00
7.32504487e-01 1.51536775e+00 4.04695958e-01 7.89999068e-01
-1.42197108e+00 -2.48966858e-01 2.38040447e-01 1.53992951e-01
-1.16981351e+00 -1.42070636e-01 1.18143606e+00 -7.85357177e-01
8.78891289e-01 -1.08885698e-01 7.08119452e-01 1.16524076e+00
1.53243914e-01 9.58318710e-01 1.31861567e+00 -3.09152216e-01
2.59335637e-01 3.72850090e-01 -1.20059595e-01 6.67724311e-01
7.24619329e-02 5.14612496e-01 -4.47142780e-01 2.03209907e-01
1.02386487e+00 2.17435122e-01 -1.60568506e-01 -1.03567374e+00
-1.07777762e+00 9.89414215e-01 5.55384994e-01 -3.12270612e-01
-3.51418436e-01 -1.32314846e-01 -3.13292220e-02 2.09746629e-01
6.68015718e-01 4.15618896e-01 -4.23983693e-01 2.41325930e-01
-4.41708684e-01 2.03786299e-01 5.95236838e-01 1.20143116e+00
5.60758233e-01 1.83919504e-01 3.02543104e-01 5.99202096e-01
5.61548948e-01 6.62321210e-01 3.13419670e-01 -9.35367525e-01
5.50422370e-01 7.21834600e-01 -1.15469113e-01 -5.91020226e-01
-3.76794398e-01 -3.73218745e-01 -8.33316565e-01 7.18483031e-01
2.59743005e-01 1.88811675e-01 -1.08565378e+00 1.69897258e+00
4.47231352e-01 -1.30077124e-01 2.27631196e-01 9.24435019e-01
7.73959219e-01 4.78376657e-01 -5.49622178e-01 -1.75970599e-01
9.04377341e-01 -9.18095052e-01 -3.91716957e-01 -1.84607789e-01
2.31963679e-01 -4.59582359e-01 9.77879286e-01 2.68800855e-01
-1.28201163e+00 -6.95114553e-01 -9.98680234e-01 -3.12791407e-01
-3.03627342e-01 -7.30968863e-02 4.00104702e-01 3.52534950e-01
-5.96863508e-01 4.01712507e-01 -9.31096137e-01 -2.82205552e-01
7.48322189e-01 3.46159399e-01 -5.97705662e-01 -2.03987867e-01
-6.06078625e-01 1.06702626e+00 4.27316219e-01 -1.70298547e-01
-1.04364872e+00 -8.08523655e-01 -9.59219337e-01 -1.32672071e-01
2.97263920e-01 -1.06360197e+00 1.03109550e+00 -6.19556785e-01
-1.61619687e+00 1.28600216e+00 1.00578032e-01 -2.73477793e-01
8.04366529e-01 -2.10150719e-01 2.50819884e-02 1.80857375e-01
1.10092297e-01 6.74211621e-01 1.06175888e+00 -1.61023831e+00
-3.94654661e-01 -9.80946064e-01 9.77508649e-02 5.78031480e-01
1.44486740e-01 -6.20901465e-01 -9.26964954e-02 -5.03964603e-01
6.05239570e-01 -1.00381887e+00 -2.79894415e-02 4.63892430e-01
-4.95604247e-01 -4.48679514e-02 8.95567834e-01 -4.88349825e-01
2.38679916e-01 -2.15113139e+00 4.57268268e-01 -6.02861978e-02
2.09788352e-01 -1.05786145e-01 -4.59936000e-02 3.57441828e-02
-1.12826355e-01 -1.30795330e-01 -3.48932356e-01 -3.49343002e-01
-6.49691969e-02 1.97169498e-01 -3.72006983e-01 5.16133845e-01
3.83036137e-01 9.76797819e-01 -7.91706979e-01 -1.94238827e-01
2.85947233e-01 4.34375435e-01 -7.13766217e-01 5.18070817e-01
-2.59711415e-01 7.28563368e-01 -6.17482603e-01 5.37864983e-01
6.19281530e-01 -5.12148321e-01 -2.61939943e-01 -3.28962356e-01
1.95858985e-01 1.06720477e-01 -1.08564663e+00 1.92367613e+00
-4.23494905e-01 3.07792991e-01 -1.76214799e-01 -1.10579002e+00
1.16325974e+00 1.52157664e-01 4.38318640e-01 -5.00863135e-01
-1.14195719e-01 3.41983102e-02 -2.36361288e-02 -7.36285567e-01
2.08819807e-02 -6.61955357e-01 1.61972374e-01 6.67494059e-01
3.93590868e-01 -7.33084738e-01 -3.64710569e-01 -7.69240931e-02
7.82797933e-01 4.26678240e-01 3.58818829e-01 9.26780179e-02
2.99311608e-01 -3.55088621e-01 3.48916143e-01 4.27682251e-01
-3.46792862e-02 8.84605706e-01 3.18037927e-01 -7.65343606e-01
-1.23692858e+00 -1.54604638e+00 -2.47279555e-01 2.47237891e-01
1.78993836e-01 2.42688179e-01 -1.59834743e-01 -1.08774340e+00
2.74041355e-01 8.61831248e-01 -6.52185440e-01 -3.40667963e-01
-4.22975987e-01 -6.29156351e-01 -2.80114450e-03 6.10935807e-01
4.45399940e-01 -9.61378098e-01 -8.23980570e-01 -2.60947585e-01
1.13396846e-01 -1.18898666e+00 -4.02518183e-01 2.26462513e-01
-1.26943374e+00 -1.00087237e+00 -7.33969986e-01 -6.94106281e-01
1.05973792e+00 4.34235841e-01 9.46739316e-01 -4.94567782e-01
-2.34403282e-01 7.99709558e-01 -1.12069994e-01 -6.66928351e-01
-4.37097788e-01 -2.34502643e-01 2.40753487e-01 2.90755834e-02
3.15405965e-01 -8.48086655e-01 -2.79058427e-01 3.58164966e-01
-7.18438148e-01 3.50222349e-01 7.85018682e-01 9.12633657e-01
9.01485503e-01 -1.65645257e-02 6.23815894e-01 -9.04288292e-01
1.31776482e-01 -2.17624277e-01 -7.90660083e-01 1.25285223e-01
-4.82146978e-01 1.17217146e-01 4.24824834e-01 -4.40812767e-01
-1.17247474e+00 5.09292185e-01 9.91804004e-02 -8.15378189e-01
-3.56226206e-01 3.80542167e-02 -5.27402580e-01 9.94267762e-02
5.98304987e-01 4.57209259e-01 3.69017303e-01 -7.22236335e-01
4.28091973e-01 3.37934554e-01 4.49692905e-02 -4.16735440e-01
9.97069538e-01 7.63789415e-01 2.22757921e-01 -8.03419948e-01
-9.93747115e-01 -2.02577725e-01 -1.15373766e+00 -2.23987341e-01
9.68144357e-01 -9.91722226e-01 -5.04289150e-01 2.91034132e-01
-1.16545773e+00 -1.67357698e-01 -7.61295915e-01 8.52614045e-01
-1.11889255e+00 2.04710662e-01 -1.56510875e-01 -6.32908583e-01
-1.28137976e-01 -1.26029539e+00 1.32826841e+00 3.57359312e-02
1.37335420e-01 -9.45308685e-01 2.26183347e-02 5.80463588e-01
-1.36442229e-01 3.14295381e-01 1.18877292e+00 -5.65904498e-01
-9.75227654e-01 -1.21410124e-01 -1.31615579e-01 6.60983503e-01
3.36870492e-01 -4.46183354e-01 -1.20720172e+00 -3.41581881e-01
6.00516200e-01 -5.11321366e-01 6.95421100e-01 5.44383347e-01
1.16475880e+00 -2.09766194e-01 -9.26893204e-02 7.57342398e-01
1.31070888e+00 2.90240288e-01 2.89665759e-01 6.36753365e-02
8.22879732e-01 8.77495527e-01 3.38991553e-01 3.17561626e-01
2.83268183e-01 6.66418135e-01 6.72918141e-01 2.43117854e-01
-2.06778660e-01 -5.98360956e-01 2.23895833e-01 8.17126691e-01
-6.29195422e-02 -8.23520496e-02 -8.53269219e-01 3.03039968e-01
-1.49210238e+00 -9.15196240e-01 4.11870211e-01 2.33604121e+00
3.60211968e-01 2.30582774e-01 -2.90796328e-02 -3.33047099e-02
4.85364079e-01 1.12789065e-01 -1.09237254e+00 -6.11412153e-02
1.07634375e-02 -3.40639114e-01 5.28986380e-02 4.26871657e-01
-9.21582520e-01 5.53703010e-01 5.81823778e+00 2.24189386e-01
-1.12723064e+00 -1.85050458e-01 6.17160022e-01 -1.41430125e-01
-4.07497138e-01 -2.28586644e-01 -9.42145705e-01 1.94819674e-01
2.28764102e-01 -2.08015647e-02 2.38456622e-01 9.07584310e-01
-7.17535466e-02 1.86878577e-01 -1.66770375e+00 1.25051749e+00
5.36130548e-01 -1.17173827e+00 5.42654753e-01 3.73616695e-01
8.10159206e-01 -1.30403368e-02 7.86616951e-02 1.56800687e-01
-7.91716296e-03 -9.23362374e-01 7.64505923e-01 6.57204032e-01
6.40145123e-01 -3.76433492e-01 3.03649455e-01 7.91877031e-01
-6.06567144e-01 8.00560508e-03 -3.50673378e-01 1.53152451e-01
3.94142151e-01 4.82195586e-01 -9.05209243e-01 5.03868997e-01
5.60877860e-01 1.08512044e+00 -4.07764286e-01 9.39116180e-01
-2.48633787e-01 7.27120116e-02 -3.34072053e-01 2.40155324e-01
4.68564965e-03 -4.47141051e-01 8.63211989e-01 5.44363022e-01
3.55248958e-01 1.69835880e-01 2.42613494e-01 1.13248193e+00
-2.16952920e-01 -2.65243322e-01 -1.15789258e+00 3.04909140e-01
-1.20651303e-02 8.99022877e-01 -6.38085604e-01 -1.50121478e-02
-4.64299083e-01 7.34221578e-01 4.14153963e-01 3.06628793e-01
-4.97656435e-01 1.20184518e-01 2.38926247e-01 2.03892887e-01
5.10912657e-01 -2.57476568e-01 -3.32375199e-01 -1.22746992e+00
3.06788921e-01 -4.84485090e-01 1.29466325e-01 -9.30760086e-01
-1.70152462e+00 5.04624128e-01 2.41074860e-01 -1.66383946e+00
-2.44354233e-01 -8.51779580e-01 -3.18449706e-01 6.57617509e-01
-1.55857158e+00 -1.08733940e+00 -2.56822824e-01 5.17826021e-01
7.20968544e-01 -4.83676732e-01 7.28562236e-01 1.06283508e-01
1.04532484e-02 3.13691407e-01 -5.12262434e-02 -5.57379797e-02
4.76383150e-01 -1.33478093e+00 1.66072160e-01 3.47994298e-01
4.47196841e-01 1.33220747e-01 2.52134860e-01 -4.93697643e-01
-1.58879697e+00 -1.08243191e+00 6.17980361e-01 -7.46946812e-01
1.24584235e-01 -6.00463092e-01 -8.90934944e-01 9.38405097e-01
-2.36278743e-01 3.11615855e-01 6.92346990e-01 -1.46757767e-01
-4.49550658e-01 -1.41640287e-02 -1.36640596e+00 3.38279516e-01
1.17183232e+00 -5.53105175e-01 -8.30549657e-01 3.97460788e-01
8.02216649e-01 -6.59644186e-01 -9.42407310e-01 4.53576893e-01
6.70700312e-01 -8.73646140e-01 1.20311928e+00 -8.70623171e-01
7.46489227e-01 -1.66491702e-01 -4.98273104e-01 -1.39259398e+00
-1.42621174e-01 -7.72273168e-02 -3.42274874e-01 8.13936532e-01
3.00974280e-01 -4.88113761e-01 8.70249569e-01 5.21323144e-01
-1.64555207e-01 -9.98065174e-01 -9.50894415e-01 -9.47297156e-01
1.33303702e-01 -2.76824415e-01 2.60936290e-01 7.30283141e-01
-5.51834881e-01 7.62756288e-01 -4.03241277e-01 2.74635822e-01
9.87325370e-01 5.83966553e-01 9.36292112e-01 -1.34662962e+00
-4.34052467e-01 -2.12436914e-01 -6.81942523e-01 -1.34627330e+00
2.31449768e-01 -1.17154264e+00 -8.77771005e-02 -1.37650931e+00
3.54557067e-01 -3.51498693e-01 8.40380937e-02 -2.75195371e-02
1.99748918e-01 -7.89654255e-02 4.57241118e-01 3.02278280e-01
-3.87187064e-01 1.17514646e+00 1.83655035e+00 -1.74084514e-01
-1.84127882e-01 4.69135940e-01 -4.33476925e-01 9.73637879e-01
5.89037478e-01 -4.71194386e-01 -6.95661366e-01 -8.41818571e-01
-6.41954988e-02 1.79510251e-01 6.27773881e-01 -6.33331478e-01
-5.37580550e-02 -1.40315190e-01 6.43377542e-01 -9.29941833e-01
8.11595261e-01 -1.14321411e+00 -8.26360285e-02 2.99076796e-01
-4.31533426e-01 -8.92349258e-02 4.62180153e-02 9.31204617e-01
-1.24333702e-01 -1.15945980e-01 9.73138452e-01 -3.27641428e-01
-5.69702566e-01 6.60612822e-01 1.78987950e-01 1.86556011e-01
1.01940274e+00 -4.19730753e-01 4.41092774e-02 -2.66626596e-01
-9.85129774e-01 6.48638532e-02 4.49014425e-01 6.14536524e-01
1.18922937e+00 -1.55396295e+00 -7.57497311e-01 6.89924479e-01
2.30803683e-01 4.43081200e-01 1.58533394e-01 4.14700508e-01
-2.62299478e-01 2.71652430e-01 -4.29720759e-01 -1.03483558e+00
-1.06537116e+00 7.25013196e-01 4.89086181e-01 -5.09655587e-02
-1.09130013e+00 7.84819722e-01 8.09267044e-01 -9.27160800e-01
3.26673269e-01 -1.99878007e-01 -1.12834960e-01 -1.95505172e-01
2.88507789e-01 3.08541298e-01 -2.04509601e-01 -7.77861357e-01
-7.73775652e-02 9.42706048e-01 -1.62333682e-01 -3.80470813e-03
1.58877921e+00 -8.46552923e-02 2.84942120e-01 8.52334380e-01
1.50225961e+00 -5.51957726e-01 -1.63883436e+00 -5.69841683e-01
-1.58272788e-01 -7.29945719e-01 -1.37546748e-01 -6.10378087e-01
-1.06856668e+00 1.23035872e+00 8.15111339e-01 -1.58301204e-01
7.54157424e-01 2.74101973e-01 5.40608048e-01 5.96106946e-01
4.65755880e-01 -7.21038342e-01 5.36662936e-01 5.13742864e-01
1.28139043e+00 -1.64922535e+00 2.48056024e-01 -2.64035225e-01
-5.60580969e-01 9.86375868e-01 7.45608926e-01 -2.34881356e-01
1.07414734e+00 -2.03452364e-01 -5.84551878e-02 -3.84455919e-01
-7.29294658e-01 1.30077809e-01 5.08496344e-01 6.80714905e-01
-7.28195757e-02 4.39991988e-02 3.39540303e-01 3.24075192e-01
-1.35215521e-01 -2.03638628e-01 3.92430156e-01 7.47935474e-01
-1.01574972e-01 -8.25861216e-01 -1.06375083e-01 5.52142501e-01
-1.11058410e-02 4.12186146e-01 -2.77890861e-01 8.57783556e-01
8.25814456e-02 2.37311214e-01 4.40303311e-02 -1.46407038e-01
6.51295066e-01 9.59778801e-02 9.42823827e-01 -9.62479472e-01
1.20764740e-01 5.19112498e-02 -4.62067366e-01 -3.94627571e-01
-6.19866133e-01 -7.41862059e-01 -1.08215094e+00 2.44150996e-01
-2.19256237e-01 -2.52641737e-01 7.64049351e-01 8.71629298e-01
3.42928916e-01 3.50924045e-01 1.01737714e+00 -1.14458847e+00
-9.45673048e-01 -6.21807754e-01 -9.29697037e-01 5.82920551e-01
5.20762205e-01 -1.04447663e+00 -5.11364400e-01 2.93469369e-01]
|
[8.363031387329102, -3.2367196083068848]
|
807106e8-6d55-4e5e-86d8-1b81f0a0c36a
|
align-yourself-self-supervised-pre-training
|
2106.15788
| null |
https://arxiv.org/abs/2106.15788v4
|
https://arxiv.org/pdf/2106.15788v4.pdf
|
Exploring Localization for Self-supervised Fine-grained Contrastive Learning
|
Self-supervised contrastive learning has demonstrated great potential in learning visual representations. Despite their success in various downstream tasks such as image classification and object detection, self-supervised pre-training for fine-grained scenarios is not fully explored. We point out that current contrastive methods are prone to memorizing background/foreground texture and therefore have a limitation in localizing the foreground object. Analysis suggests that learning to extract discriminative texture information and localization are equally crucial for fine-grained self-supervised pre-training. Based on our findings, we introduce cross-view saliency alignment (CVSA), a contrastive learning framework that first crops and swaps saliency regions of images as a novel view generation and then guides the model to localize on foreground objects via a cross-view alignment loss. Extensive experiments on both small- and large-scale fine-grained classification benchmarks show that CVSA significantly improves the learned representation.
|
['Stan Z. Li', 'Zelin Zang', 'Siyuan Li', 'Di wu']
|
2021-06-30
| null | null | null | null |
['fine-grained-image-recognition']
|
['computer-vision']
|
[ 8.60950530e-01 1.41971046e-03 -3.95143241e-01 -4.89823103e-01
-6.83162391e-01 -5.56328595e-01 8.47252846e-01 1.50517836e-01
4.96032238e-02 7.48389065e-01 2.37342015e-01 -8.91848356e-02
8.36236775e-02 -6.10515773e-01 -1.07281971e+00 -7.40917802e-01
1.52520508e-01 1.52842268e-01 7.69575596e-01 1.60099007e-02
5.55880010e-01 4.66043591e-01 -1.83299792e+00 7.53161550e-01
7.47343838e-01 9.60545659e-01 6.27895176e-01 6.34989977e-01
-4.16214466e-02 9.50800359e-01 -6.00942016e-01 -1.46021411e-01
2.33337492e-01 -4.45793360e-01 -1.01496124e+00 3.98307085e-01
1.19587970e+00 -2.00351700e-01 9.11023766e-02 8.74459267e-01
3.28997433e-01 5.40963337e-02 5.99058628e-01 -1.25042653e+00
-7.22139239e-01 4.81402338e-01 -8.83616447e-01 8.75820398e-01
-9.36064571e-02 1.63120136e-01 1.06507349e+00 -1.07697666e+00
6.34770095e-01 1.24831152e+00 5.12036443e-01 5.98201871e-01
-1.41733706e+00 -5.23746490e-01 4.75774556e-01 4.78281751e-02
-9.77359354e-01 -4.41338092e-01 9.78860080e-01 -2.44153664e-01
8.42555523e-01 3.02798390e-01 6.30127966e-01 1.10881436e+00
4.62744892e-01 1.14526629e+00 1.73621821e+00 -4.03445572e-01
2.25439101e-01 2.63360031e-02 -1.31505309e-03 8.64440441e-01
4.10648257e-01 2.13367209e-01 -9.60870028e-01 9.75890681e-02
1.08785784e+00 -2.20458768e-03 5.81626371e-02 -9.18235183e-01
-1.37451410e+00 7.25579679e-01 7.78677285e-01 3.28117125e-02
-1.79698199e-01 3.19567263e-01 2.47498810e-01 6.09733127e-02
7.20353723e-01 5.77327967e-01 -5.09489775e-01 3.59646350e-01
-1.30804133e+00 1.01797096e-01 1.50728852e-01 7.10137308e-01
1.02569962e+00 4.07139599e-01 -3.83208990e-01 6.48362875e-01
-1.20742582e-02 7.40490973e-01 3.11590970e-01 -8.69444013e-01
2.14427635e-01 7.11673856e-01 -1.66537717e-01 -1.04660785e+00
-2.01951027e-01 -6.94160402e-01 -7.04966187e-01 6.13001227e-01
4.62260067e-01 2.57017136e-01 -1.24454045e+00 1.69148564e+00
1.99146450e-01 3.39441985e-01 -3.39325797e-03 9.55974936e-01
8.85804534e-01 2.30886072e-01 3.09544742e-01 1.12325080e-01
1.14861083e+00 -1.30205619e+00 -2.15410218e-01 -7.59109914e-01
-2.07507759e-02 -8.72460783e-01 1.28897333e+00 1.49249077e-01
-1.27900708e+00 -7.94147491e-01 -9.47167039e-01 -2.02985004e-01
-4.46739763e-01 -1.95014514e-02 1.08247685e+00 4.76303667e-01
-1.06462514e+00 4.58879739e-01 -6.80174112e-01 -1.49442986e-01
1.14347267e+00 1.00454509e-01 -3.34129751e-01 -4.55028703e-03
-7.21662045e-01 7.21379220e-01 3.41763318e-01 -1.81801125e-01
-1.34733927e+00 -8.84687603e-01 -9.02658582e-01 1.44098908e-01
2.86543518e-01 -9.53068554e-01 9.38386977e-01 -1.43802822e+00
-1.03501940e+00 1.23574829e+00 -3.97407353e-01 -6.00921929e-01
2.94332236e-01 -2.12732196e-01 -1.47155911e-01 3.13054293e-01
6.66210532e-01 1.15111804e+00 1.54695547e+00 -1.60878026e+00
-9.48709190e-01 -2.48228908e-01 7.94212520e-02 3.90225142e-01
1.04485363e-01 -2.25992724e-01 -8.80591720e-02 -1.06493068e+00
2.16102958e-01 -6.83358252e-01 -1.41408086e-01 1.85508206e-02
-3.77699524e-01 -7.66036808e-02 1.02511859e+00 -2.44368643e-01
7.21554458e-01 -1.86943686e+00 -1.35071501e-02 5.79946265e-02
3.68834555e-01 1.51777789e-01 -2.00078070e-01 -1.92021400e-01
-2.06936136e-01 -1.72142074e-01 -1.25110134e-01 -1.98703110e-01
-3.81541967e-01 2.87261028e-02 -6.79820895e-01 1.99845821e-01
8.53887856e-01 1.30710030e+00 -1.24606860e+00 -6.22652292e-01
4.92767006e-01 2.41224602e-01 -5.12575150e-01 4.58415486e-02
-4.37298268e-01 4.34371978e-01 -3.07917088e-01 9.14183438e-01
7.42127478e-01 -8.24679971e-01 -2.29030922e-02 -4.92470652e-01
8.65194052e-02 1.69208273e-01 -8.22069585e-01 1.69087112e+00
-2.95308620e-01 6.69560969e-01 -2.74265677e-01 -9.83188391e-01
7.12146819e-01 -4.39507484e-01 4.91321757e-02 -1.01753092e+00
-1.54360399e-01 4.51607518e-02 -2.65102953e-01 1.47140929e-02
5.72538853e-01 5.47859520e-02 6.54868111e-02 5.18760204e-01
2.16093585e-01 -2.34691769e-01 -2.78798696e-02 4.72009897e-01
7.22694099e-01 3.47963154e-01 3.12768281e-01 -5.54955363e-01
3.21531206e-01 2.21924424e-01 2.15703517e-01 9.30002809e-01
-2.65484333e-01 8.72923791e-01 1.79188952e-01 -3.83616686e-01
-8.96287143e-01 -1.48508596e+00 -6.74010348e-03 1.75673437e+00
5.46595573e-01 -9.24867168e-02 -7.20631361e-01 -1.07793581e+00
1.20567858e-01 4.96321976e-01 -9.61900890e-01 -2.38718480e-01
-5.27571380e-01 -5.29257178e-01 1.64686382e-01 8.15369785e-01
6.76106930e-01 -1.35055852e+00 -9.87025440e-01 3.17542776e-02
-1.29789487e-01 -9.51127350e-01 -5.02569973e-01 6.49022579e-01
-9.68832493e-01 -1.12084889e+00 -7.33957350e-01 -1.11237824e+00
8.97596598e-01 1.20628071e+00 1.71971631e+00 4.70567532e-02
-5.42836308e-01 4.67337400e-01 -7.22450912e-02 -4.26087826e-01
-2.04602778e-01 -1.46036390e-02 -2.39293829e-01 -5.47190383e-02
2.09627062e-01 -3.29408824e-01 -9.13982809e-01 2.77058750e-01
-7.86515832e-01 3.38946015e-01 7.95779526e-01 1.02329385e+00
9.41678822e-01 -1.75855398e-01 6.84171796e-01 -1.16428185e+00
1.25824466e-01 -2.55333990e-01 -4.10822600e-01 3.66411030e-01
-4.79429632e-01 1.43558010e-01 2.50219822e-01 -2.47088134e-01
-1.23784542e+00 1.02161519e-01 2.30235890e-01 -4.88364130e-01
-3.51555169e-01 -9.04336870e-02 3.58822197e-02 -2.90594757e-01
6.64060235e-01 4.27969694e-01 -1.98198661e-01 -1.30601749e-01
5.70915699e-01 -2.75272294e-03 6.31137371e-01 -4.99677926e-01
7.73821354e-01 9.10939395e-01 5.94426915e-02 -6.22182012e-01
-1.53694296e+00 -4.15632546e-01 -7.11298108e-01 -2.20435914e-02
8.06223869e-01 -1.16939414e+00 -3.10829192e-01 4.25502509e-01
-6.36866450e-01 -6.62858784e-01 -6.15749419e-01 -2.26229981e-01
-8.27321410e-01 1.38344824e-01 -1.97260797e-01 -3.72640014e-01
-2.93560475e-01 -8.72705758e-01 1.63597035e+00 4.08296406e-01
-1.96473569e-01 -1.07857299e+00 -3.60330194e-01 4.09154862e-01
5.62411189e-01 3.49483527e-02 7.68970072e-01 -1.49301708e-01
-9.48338866e-01 6.23957813e-01 -7.29922831e-01 2.35409170e-01
3.36089373e-01 -3.00390512e-01 -1.15473092e+00 -3.99696678e-01
-3.31108540e-01 -6.41045928e-01 1.32867789e+00 6.34589612e-01
1.41057026e+00 -8.95895883e-02 -4.91817653e-01 6.46385014e-01
1.37577641e+00 -3.55075598e-01 3.80855590e-01 4.11376834e-01
8.36406946e-01 5.20755529e-01 8.67485106e-01 1.45200238e-01
1.87911764e-01 4.21246171e-01 4.91355866e-01 -6.88367665e-01
-7.86282182e-01 -4.08320755e-01 -2.75306236e-02 -4.02157791e-02
9.81944352e-02 2.91477982e-02 -6.17163539e-01 7.01339066e-01
-1.55686879e+00 -1.20990968e+00 1.34397089e-01 1.79790926e+00
9.84542668e-01 4.18217808e-01 2.16969773e-01 -2.49457229e-02
6.03014827e-01 4.08859968e-01 -7.89738655e-01 -1.56866029e-01
-6.42443836e-01 5.87606668e-01 5.87418318e-01 3.47530633e-01
-1.52113628e+00 1.56996989e+00 6.63018560e+00 9.33673680e-01
-1.44557130e+00 1.46708116e-01 1.09896243e+00 2.67185010e-02
-2.97710478e-01 -1.14464648e-01 -7.73970425e-01 4.42168623e-01
1.38951614e-01 2.46153012e-01 2.91293263e-01 9.49677348e-01
-1.25267416e-01 -4.60815609e-01 -9.39068913e-01 8.47759843e-01
2.74028689e-01 -1.76780176e+00 3.71739328e-01 -3.83450180e-01
1.36183453e+00 3.59640345e-02 5.76076627e-01 3.03559192e-02
3.50900769e-01 -1.03817427e+00 9.07160521e-01 3.09900641e-01
9.24333096e-01 -5.73308706e-01 3.34714621e-01 -6.13549016e-02
-1.20990849e+00 -2.54193563e-02 -5.20217717e-01 5.78292497e-02
-1.10788964e-01 5.99390864e-01 -8.17442060e-01 1.58529162e-01
9.20280516e-01 8.01326334e-01 -1.18866813e+00 7.10009575e-01
-1.90441847e-01 6.75464392e-01 1.97378814e-01 2.79279888e-01
2.82669485e-01 3.87413919e-01 2.96543747e-01 1.29925501e+00
-2.68090069e-01 -3.67052644e-01 2.13722497e-01 1.00250053e+00
2.22584575e-01 -3.44436586e-01 -4.46673036e-01 8.78406838e-02
2.03617841e-01 1.24535799e+00 -1.30080044e+00 -5.18703163e-01
-2.99352974e-01 1.30953300e+00 5.41920900e-01 3.93617630e-01
-7.14345098e-01 5.20667322e-02 6.86352253e-01 3.06231946e-01
7.68251300e-01 1.08857527e-02 -7.13398755e-01 -1.07204413e+00
-2.87806630e-01 -9.55112755e-01 3.15467268e-01 -8.84423196e-01
-1.36349058e+00 4.28904891e-01 -8.82927328e-02 -9.93576705e-01
-1.17571175e-01 -5.87695003e-01 -4.68544543e-01 6.28641725e-01
-1.80104089e+00 -1.71237278e+00 -5.99666178e-01 7.10814953e-01
8.46515477e-01 -2.02533275e-01 6.00316465e-01 -2.03794137e-01
-1.12174325e-01 5.76432407e-01 -1.71727374e-01 -1.80746064e-01
8.30663383e-01 -1.54297042e+00 5.21194339e-01 9.46072280e-01
3.37146789e-01 6.41609013e-01 6.71919703e-01 -7.79949546e-01
-1.08151209e+00 -1.37147152e+00 5.85450470e-01 -7.11371601e-01
4.63455319e-01 -5.08384645e-01 -8.45644593e-01 4.25991833e-01
2.11603314e-01 4.99070913e-01 3.04453522e-01 2.44159568e-02
-5.78579724e-01 -1.00738496e-01 -1.03758597e+00 5.60319424e-01
1.36238956e+00 -6.91684306e-01 -5.85401356e-01 2.75000900e-01
5.60258210e-01 -3.81542444e-01 -2.92605042e-01 5.63177764e-01
4.14588571e-01 -1.23747623e+00 1.41481721e+00 -7.38639772e-01
6.47346973e-01 -3.86775076e-01 -8.06423128e-02 -1.30168867e+00
-6.13812268e-01 -1.80190340e-01 -1.75553873e-01 1.08125687e+00
1.59415174e-02 -4.01938677e-01 1.15162480e+00 -1.00656331e-01
-7.16944411e-02 -6.29607558e-01 -5.27327240e-01 -5.43293297e-01
-4.31066789e-02 7.50276521e-02 2.57122397e-01 8.00192356e-01
-6.15467489e-01 5.99451482e-01 -2.63995379e-01 1.75073013e-01
9.53656673e-01 7.85587311e-01 8.06839049e-01 -1.21700191e+00
-2.35632956e-01 -6.03706717e-01 -3.44588906e-01 -1.08294141e+00
2.01717228e-01 -7.46505380e-01 6.58037737e-02 -1.45778954e+00
5.99596977e-01 -6.62841022e-01 -5.24266481e-01 4.20178682e-01
-6.64001286e-01 9.03456569e-01 8.68661969e-04 -8.49380810e-03
-1.03638887e+00 2.56211072e-01 1.52172089e+00 -3.29777628e-01
1.38543129e-01 -1.35338381e-01 -1.02823162e+00 6.75337255e-01
6.72154665e-01 -1.80941939e-01 -5.65486372e-01 -2.53814250e-01
-8.13138634e-02 -3.97842079e-01 8.97046089e-01 -9.27590549e-01
-1.63763627e-01 -3.92058223e-01 9.64940071e-01 -7.30960906e-01
2.45463312e-01 -4.86691028e-01 -4.71950620e-01 3.08571041e-01
-4.48075086e-01 -1.05618730e-01 3.86645794e-01 7.11012006e-01
-2.22244889e-01 1.71893954e-01 1.01947200e+00 -2.53569901e-01
-1.32048690e+00 9.61192995e-02 -1.80235088e-01 4.60279405e-01
9.11969304e-01 -5.46223819e-01 -5.96032917e-01 -1.59016564e-01
-5.91634333e-01 -1.58390701e-01 6.52224004e-01 5.24410009e-01
7.43471861e-01 -1.13455629e+00 -5.61796308e-01 5.90714157e-01
2.92760253e-01 1.34370867e-02 4.24607754e-01 4.71157670e-01
-2.68028259e-01 4.46102232e-01 -6.78580761e-01 -9.77639675e-01
-1.39520729e+00 6.29118860e-01 1.25729829e-01 -1.41806662e-01
-4.56156880e-01 1.24690664e+00 1.03188074e+00 -4.82337410e-03
6.31046295e-02 -3.24918658e-01 -3.70990559e-02 -1.35795549e-01
4.84374672e-01 1.64792672e-01 1.17395325e-02 -6.80538058e-01
-4.17199045e-01 6.06050253e-01 -2.79453605e-01 2.15280816e-01
1.24206150e+00 -3.79222721e-01 5.29935062e-02 3.45204353e-01
9.43534076e-01 1.35000199e-01 -1.78996217e+00 -3.95772666e-01
-3.11683659e-02 -7.23629355e-01 3.67545225e-02 -9.28539932e-01
-1.00925076e+00 9.16356444e-01 7.38354087e-01 -5.66506721e-02
1.19349992e+00 3.29001904e-01 4.39593166e-01 1.55873850e-01
4.49253708e-01 -9.50140953e-01 9.67577815e-01 5.86692035e-01
7.85651147e-01 -1.71350443e+00 1.01311348e-01 -6.17115021e-01
-8.46477747e-01 7.82076299e-01 9.91981387e-01 -5.10595560e-01
5.05753338e-01 3.41849953e-01 2.10972980e-01 -2.17596248e-01
-7.03151226e-01 -5.91021240e-01 7.21891820e-01 9.37034786e-01
3.82008642e-01 4.29479182e-02 3.65292490e-01 2.31729209e-01
-9.40093771e-02 -3.74550223e-01 2.53127337e-01 9.60395634e-01
-6.43776298e-01 -7.28834748e-01 -2.40639523e-01 5.94247878e-01
-4.23779756e-01 -3.24827731e-01 -5.21666884e-01 5.74627459e-01
2.81106919e-01 6.68604851e-01 3.64803284e-01 -3.69439758e-02
-1.75271422e-01 -1.91491351e-01 8.23956549e-01 -8.89818013e-01
-4.21281874e-01 2.85278298e-02 -2.84866542e-01 -7.43373632e-01
-7.03884065e-01 -5.21511674e-01 -1.06127536e+00 6.86384588e-02
-2.89985031e-01 -3.11758578e-01 2.85913289e-01 8.94504011e-01
4.05388236e-01 8.15073609e-01 4.98684943e-01 -1.21171451e+00
-2.64903098e-01 -4.82340962e-01 -4.08539385e-01 6.53849661e-01
4.87031996e-01 -1.05919397e+00 -2.56168991e-01 2.09201112e-01]
|
[9.725367546081543, 1.4944113492965698]
|
261b49f0-f34c-45b4-be87-527875cb3da8
|
gandiffface-controllable-generation-of
|
2305.19962
| null |
https://arxiv.org/abs/2305.19962v1
|
https://arxiv.org/pdf/2305.19962v1.pdf
|
GANDiffFace: Controllable Generation of Synthetic Datasets for Face Recognition with Realistic Variations
|
Face recognition systems have significantly advanced in recent years, driven by the availability of large-scale datasets. However, several issues have recently came up, including privacy concerns that have led to the discontinuation of well-established public datasets. Synthetic datasets have emerged as a solution, even though current synthesis methods present other drawbacks such as limited intra-class variations, lack of realism, and unfair representation of demographic groups. This study introduces GANDiffFace, a novel framework for the generation of synthetic datasets for face recognition that combines the power of Generative Adversarial Networks (GANs) and Diffusion models to overcome the limitations of existing synthetic datasets. In GANDiffFace, we first propose the use of GANs to synthesize highly realistic identities and meet target demographic distributions. Subsequently, we fine-tune Diffusion models with the images generated with GANs, synthesizing multiple images of the same identity with a variety of accessories, poses, expressions, and contexts. We generate multiple synthetic datasets by changing GANDiffFace settings, and compare their mated and non-mated score distributions with the distributions provided by popular real-world datasets for face recognition, i.e. VGG2 and IJB-C. Our results show the feasibility of the proposed GANDiffFace, in particular the use of Diffusion models to enhance the (limited) intra-class variations provided by GANs towards the level of real-world datasets.
|
['Maxim Schaubert', 'Florian Domin', 'Dominik Lawatsch', 'Ruben Vera-Rodriguez', 'Ruben Tolosana', 'Christian Rathgeb', 'Pietro Melzi']
|
2023-05-31
| null | null | null | null |
['face-recognition']
|
['computer-vision']
|
[ 2.81233221e-01 6.73895003e-03 3.09928566e-01 -5.62900722e-01
-5.08337975e-01 -6.76426113e-01 8.06664050e-01 -7.37825871e-01
-6.79812729e-02 9.11886036e-01 1.86562553e-01 2.35873073e-01
1.23810261e-01 -8.92332673e-01 -4.82508779e-01 -6.30838275e-01
3.27090293e-01 4.90434468e-01 -5.57792187e-01 -2.82867551e-01
-1.71060115e-01 6.48589909e-01 -1.71124029e+00 1.66280016e-01
8.75958204e-01 9.32242513e-01 -5.43021083e-01 3.29418480e-01
1.57594343e-03 3.39437544e-01 -1.11584163e+00 -1.06513453e+00
7.16130257e-01 -8.80453050e-01 -1.24357454e-01 1.10386826e-01
7.95874953e-01 -5.10241032e-01 -4.55956846e-01 9.74487066e-01
8.90693605e-01 -1.52302712e-01 5.64894140e-01 -1.74804771e+00
-1.06867814e+00 2.87908107e-01 -6.00935578e-01 -3.79837096e-01
3.91590357e-01 4.87192780e-01 4.42741841e-01 -6.65372372e-01
7.24527061e-01 1.51291740e+00 7.07651675e-01 1.27747750e+00
-1.37362731e+00 -1.16742861e+00 -2.46415377e-01 -1.33391753e-01
-1.40343022e+00 -7.67621458e-01 7.89901972e-01 -2.55662262e-01
2.63031930e-01 3.35456222e-01 5.30686319e-01 2.00838518e+00
-1.50919646e-01 4.02441144e-01 1.41295612e+00 -2.25104436e-01
2.79069006e-01 2.90693432e-01 -5.93293965e-01 3.47598195e-01
2.31347218e-01 2.28415936e-01 -5.98211825e-01 -3.53635639e-01
8.59082580e-01 -1.67040676e-01 -2.11689070e-01 -3.88956159e-01
-9.21214879e-01 8.35466623e-01 2.04878692e-02 -4.62734737e-02
-3.68679285e-01 -8.11026916e-02 2.85068870e-01 2.35415742e-01
4.68310744e-01 3.12319160e-01 2.28768215e-02 9.80637968e-04
-8.26296449e-01 4.68583941e-01 7.35280275e-01 1.01868308e+00
4.75675583e-01 3.78007382e-01 -4.65083569e-01 9.60561454e-01
1.56257227e-01 7.94668078e-01 5.57747722e-01 -8.76302361e-01
3.91002297e-01 4.61158365e-01 9.33013558e-02 -1.12356043e+00
2.84573257e-01 -3.20914626e-01 -1.02938259e+00 3.31695616e-01
6.04083240e-01 -3.95946652e-01 -1.08745813e+00 2.21411657e+00
2.35921443e-01 1.06784157e-01 1.46805301e-01 5.63542604e-01
7.34152138e-01 3.49968106e-01 7.92122260e-02 9.45396051e-02
9.82098937e-01 -6.39689326e-01 -7.08281457e-01 -1.18341215e-01
-5.33503890e-02 -6.18125379e-01 1.06737626e+00 1.49353296e-01
-9.98664379e-01 -6.43177748e-01 -8.15721273e-01 3.61809760e-01
-3.43781263e-01 9.30537358e-02 5.31401038e-01 1.42821574e+00
-1.31244266e+00 2.81001717e-01 -4.46160018e-01 -3.42607200e-01
9.26191092e-01 4.45342153e-01 -5.86289227e-01 -2.83083409e-01
-1.05294323e+00 4.82059121e-01 -7.22517967e-02 2.53815860e-01
-1.06592667e+00 -7.97013640e-01 -6.73249781e-01 -1.61232725e-01
4.88331653e-02 -7.15255201e-01 7.48604178e-01 -1.50419891e+00
-1.60660410e+00 1.05608761e+00 1.70192242e-01 -1.81958541e-01
1.01232719e+00 5.10569885e-02 -6.60313666e-01 -2.75507927e-01
-5.50354831e-02 8.67605388e-01 9.92775142e-01 -1.32701969e+00
5.09440042e-02 -5.20344853e-01 -9.62168947e-02 -1.24786228e-01
-5.54651558e-01 3.83685231e-02 -3.98243815e-01 -8.51853192e-01
-4.15043026e-01 -1.02438533e+00 1.14999758e-02 1.09293878e-01
-5.51053047e-01 3.91051114e-01 7.70720303e-01 -7.15917468e-01
6.91785455e-01 -2.21967602e+00 9.51016098e-02 3.31531823e-01
5.99561371e-02 3.99322152e-01 -5.30827403e-01 2.73222655e-01
-1.89019665e-01 2.56747574e-01 -1.72438011e-01 -6.80548131e-01
-4.48708311e-02 1.96945116e-01 -1.54478580e-01 1.98063195e-01
2.91087657e-01 8.65654290e-01 -6.72191024e-01 -8.91354457e-02
-1.42884552e-01 7.71110594e-01 -6.33217752e-01 4.23037946e-01
-1.42329276e-01 8.13447833e-01 -1.86644852e-01 7.86757052e-01
9.96353686e-01 2.30751872e-01 1.62268206e-01 -1.21619418e-01
5.20933449e-01 -4.12558526e-01 -1.05033493e+00 1.39166713e+00
-1.81466043e-01 3.77934754e-01 1.29515797e-01 -5.14983714e-01
1.15777063e+00 2.15872139e-01 3.57307166e-01 -6.83605969e-01
2.07854867e-01 2.07500950e-01 2.45167799e-02 -1.51216164e-01
1.74141929e-01 -2.52366602e-01 3.75588797e-02 3.72844189e-01
1.35839716e-01 -1.03403501e-01 -3.18148322e-02 -1.07056051e-02
7.86096454e-01 1.14290632e-01 -2.28837520e-01 8.72190371e-02
4.23643321e-01 -5.04347146e-01 5.56586564e-01 5.51266849e-01
-2.20832661e-01 1.11283386e+00 5.19013226e-01 -3.23886693e-01
-1.12024057e+00 -1.08886862e+00 2.29896918e-01 5.81036747e-01
-1.69278651e-01 -4.80485149e-02 -1.15781677e+00 -7.95583189e-01
1.05647080e-01 5.65893352e-01 -1.05575669e+00 -2.95746624e-01
-3.02662849e-01 -9.69199538e-01 1.06867266e+00 3.20912093e-01
7.79155850e-01 -1.08621955e+00 -9.45883691e-02 -1.53281465e-01
-2.96808407e-02 -1.12671828e+00 -4.73740488e-01 -6.13142848e-01
-3.13227803e-01 -9.42247987e-01 -9.80978787e-01 -4.32440907e-01
9.08301473e-01 -2.91044146e-01 1.16997862e+00 -9.64839682e-02
-3.43024313e-01 3.89919192e-01 -1.98905706e-01 -4.00132358e-01
-7.48016119e-01 -7.80548528e-02 1.73676595e-01 6.36283576e-01
1.41781226e-01 -6.84961081e-01 -8.04059267e-01 5.85261464e-01
-1.03879261e+00 1.20855747e-02 3.38964254e-01 8.59339297e-01
2.62897789e-01 -1.34721577e-01 9.35638011e-01 -1.08969796e+00
7.48635411e-01 -4.64110464e-01 -4.30341750e-01 2.64171451e-01
-5.51826000e-01 -2.77960658e-01 6.33913934e-01 -6.67998075e-01
-1.27029443e+00 -2.82523304e-01 -2.38897443e-01 -6.80360854e-01
-2.62241155e-01 -4.22075652e-02 -7.51776755e-01 -2.30443135e-01
6.85411811e-01 1.42480209e-01 4.46643203e-01 -2.25599170e-01
4.03812051e-01 6.85495317e-01 5.36778629e-01 -6.91923022e-01
1.03183627e+00 3.98137420e-01 -2.52357602e-01 -4.91229981e-01
-3.59768808e-01 4.98523474e-01 -2.99125761e-01 -2.33060792e-01
5.13971448e-01 -9.66862023e-01 -5.37568986e-01 1.22755730e+00
-7.55683541e-01 -3.52871954e-01 -4.77153152e-01 1.18377283e-01
-3.91842753e-01 8.11823364e-03 -5.15144348e-01 -7.33895361e-01
-4.59447265e-01 -1.17551088e+00 8.89415622e-01 4.59061056e-01
-2.14767694e-01 -8.04022193e-01 -1.29730463e-01 6.06432319e-01
8.47055197e-01 7.95388639e-01 6.51735425e-01 -4.94564742e-01
-3.83365840e-01 -2.54252225e-01 -1.01613402e-01 6.69991136e-01
5.95240176e-01 1.01530738e-01 -1.15783489e+00 -5.10447741e-01
-2.19120115e-01 -5.44701099e-01 2.26253316e-01 1.36823962e-02
1.14603162e+00 -4.10199493e-01 -5.02409413e-02 9.02071714e-01
1.20920157e+00 3.22451353e-01 9.90581095e-01 -1.72456484e-02
6.81734264e-01 6.46091402e-01 2.33615950e-01 5.36638916e-01
8.63258615e-02 6.92257583e-01 3.68688971e-01 -2.01421261e-01
-3.23696285e-01 -5.73114038e-01 2.66739070e-01 2.67803282e-01
-9.17744339e-02 -3.88819695e-01 -7.46942937e-01 3.74063402e-01
-1.26576829e+00 -1.01269031e+00 4.69768941e-01 2.19943261e+00
9.16758120e-01 -1.29792660e-01 3.41515452e-01 -2.11458206e-01
7.71093190e-01 2.86413450e-02 -8.74733031e-01 -3.18142354e-01
-4.52347845e-01 5.23900807e-01 2.89278835e-01 1.07666418e-01
-6.00870788e-01 8.02662492e-01 6.36455297e+00 7.55205691e-01
-1.37288690e+00 -2.64173467e-02 1.23762870e+00 -2.92623043e-01
-5.76238036e-01 -4.58151758e-01 -6.79285765e-01 7.14277983e-01
7.74357677e-01 -2.22475842e-01 7.65978694e-01 7.56564975e-01
-4.00976613e-02 5.13110757e-01 -1.09261870e+00 1.16562474e+00
4.47204441e-01 -1.12717426e+00 4.00105000e-01 2.49936253e-01
1.05965090e+00 -3.07611465e-01 6.36130154e-01 2.45652542e-01
4.00940865e-01 -1.47780228e+00 6.25539064e-01 4.03040797e-01
1.36742985e+00 -7.52927065e-01 7.45840311e-01 -5.04372157e-02
-5.45104980e-01 7.88366515e-03 -1.17152222e-01 3.25173646e-01
-1.11973271e-01 2.96283394e-01 -7.33400166e-01 4.58914191e-01
5.19516289e-01 3.00176680e-01 -7.48085380e-01 5.97426951e-01
2.87064835e-02 5.01630545e-01 -2.49485210e-01 2.29291841e-01
-1.80952355e-01 -4.46240008e-01 1.93712786e-01 6.09819114e-01
6.61812007e-01 -2.04783499e-01 -4.00887609e-01 1.12776124e+00
-5.58786750e-01 3.45745794e-02 -7.32629418e-01 -2.18846008e-01
6.34624541e-01 1.09604597e+00 -2.59546697e-01 -4.73352224e-02
-2.85422832e-01 1.15846741e+00 4.94084731e-02 4.71428931e-01
-9.37695205e-01 3.61289456e-02 1.10153222e+00 1.79935485e-01
1.10921971e-01 1.07696734e-01 -2.56563723e-01 -1.07184744e+00
-3.51580903e-02 -1.72044349e+00 2.04661861e-01 -5.75416803e-01
-1.55318177e+00 9.14548516e-01 -6.97015822e-02 -9.28988397e-01
-4.02043134e-01 -3.48745942e-01 -6.15393639e-01 1.21816385e+00
-9.91261363e-01 -1.49987614e+00 -5.67155778e-01 7.20541358e-01
2.66597331e-01 -6.69549286e-01 1.05942988e+00 4.93558019e-01
-6.44205630e-01 1.28183281e+00 1.45520210e-01 2.60632545e-01
9.06861842e-01 -7.48619735e-01 6.62757397e-01 6.23359919e-01
-2.29063500e-02 6.22824371e-01 5.53891778e-01 -5.26339889e-01
-1.26343226e+00 -1.09881449e+00 1.94442153e-01 -4.62649792e-01
1.21126704e-01 -7.36814976e-01 -6.50154054e-01 5.67892969e-01
2.04482660e-01 8.54585171e-02 8.55025649e-01 -3.01529408e-01
-4.59298849e-01 -3.66053373e-01 -1.84244728e+00 7.73719013e-01
1.22012389e+00 -5.24563491e-01 1.78296670e-01 -2.19785906e-02
2.37109840e-01 -3.17154050e-01 -7.48615801e-01 5.01463056e-01
7.73020208e-01 -1.07212901e+00 9.92166281e-01 -5.03406823e-01
3.63454521e-01 -1.33668512e-01 -1.39745265e-01 -1.53424895e+00
4.73661833e-02 -8.47561955e-01 2.07024246e-01 1.90612638e+00
2.60097653e-01 -8.34727168e-01 1.18584073e+00 9.61068690e-01
5.47449589e-01 -5.73117673e-01 -7.99049973e-01 -6.21088684e-01
7.67988265e-02 2.17319448e-02 1.18221164e+00 1.00889885e+00
-7.15278029e-01 -6.27466366e-02 -6.86694741e-01 -1.61773026e-01
7.98008323e-01 -1.02180079e-01 1.23068440e+00 -8.98660839e-01
-3.17292064e-01 -3.39559019e-01 -5.01701236e-01 -4.26667601e-01
2.78857291e-01 -6.16170526e-01 -3.21939021e-01 -8.32670927e-01
9.22285989e-02 -7.01647103e-01 -3.97007912e-02 5.08435726e-01
-6.96531385e-02 7.40374386e-01 2.52425641e-01 -8.15652087e-02
2.63395488e-01 7.95405805e-01 1.42534554e+00 -1.97645053e-01
4.97709587e-02 -7.89229274e-02 -9.72121656e-01 3.66146863e-01
8.53409708e-01 -1.67115450e-01 -6.86252117e-01 -3.53636682e-01
-1.53675884e-01 -1.19404830e-01 2.34707385e-01 -1.16842413e+00
-2.35355020e-01 -7.50992522e-02 7.20861316e-01 1.30203411e-01
4.63413328e-01 -6.90007865e-01 8.02942812e-01 2.23502889e-01
-3.14018399e-01 1.39464065e-01 3.96601617e-01 3.51131201e-01
-1.59454495e-01 2.95297384e-01 7.97699809e-01 -1.06167095e-02
-3.21791857e-01 5.89970708e-01 1.59587920e-01 2.31566578e-01
1.11430085e+00 -4.25318658e-01 -2.28681788e-01 -7.46524572e-01
-4.72222239e-01 -1.87572077e-01 8.44089627e-01 7.12001503e-01
5.55014014e-01 -1.44721258e+00 -9.88988698e-01 7.71148384e-01
1.11083783e-01 -2.73351759e-01 3.05430889e-01 2.50182360e-01
-4.67745841e-01 -2.74151295e-01 -8.22409570e-01 -2.63250530e-01
-1.30710280e+00 3.16457719e-01 4.57363576e-01 3.45446989e-02
-1.90236881e-01 9.07283068e-01 3.93666059e-01 -5.95450103e-01
7.57216737e-02 3.36947083e-01 5.19782640e-02 -2.24279845e-03
4.26355720e-01 1.69073895e-01 -8.27695727e-02 -8.04185569e-01
-2.72368580e-01 3.04795921e-01 5.33283800e-02 -1.16327323e-01
1.24660993e+00 9.01720151e-02 1.99938193e-02 -1.96983486e-01
9.74052727e-01 2.71058530e-01 -1.42929637e+00 1.39958620e-01
-5.24203360e-01 -8.60554814e-01 -5.80864966e-01 -1.07964575e+00
-1.58219874e+00 5.28120458e-01 8.81658971e-01 -1.79239050e-01
1.14822865e+00 -3.61208558e-01 7.16144204e-01 -3.02976996e-01
6.08045995e-01 -6.21859193e-01 1.47495106e-01 3.71492282e-02
1.04608679e+00 -1.01490796e+00 -4.20744985e-01 -4.73993778e-01
-6.60060108e-01 6.82187319e-01 8.52612317e-01 1.86841473e-01
3.79339248e-01 2.00721800e-01 4.75054085e-01 1.81134373e-01
-3.45625132e-01 3.09324712e-01 3.88286216e-03 1.13966107e+00
2.49303609e-01 8.19153041e-02 1.29157990e-01 4.65958714e-01
-4.21874255e-01 3.32599431e-02 3.18128288e-01 5.62726498e-01
6.30760491e-01 -1.59601367e+00 -4.91872311e-01 3.64235044e-01
-5.04557908e-01 1.71130642e-01 -6.81711078e-01 7.16122031e-01
3.97548735e-01 8.28756750e-01 5.17452601e-03 -5.27905524e-01
3.23270917e-01 1.71520151e-02 6.25734925e-01 -3.35122764e-01
-6.81196451e-01 -5.32238066e-01 8.58964846e-02 -3.27089071e-01
-1.74561992e-01 -6.99344277e-01 -4.00244832e-01 -6.79425180e-01
1.07115522e-01 5.16098086e-03 6.71695471e-01 3.94137293e-01
7.49497652e-01 2.20147893e-01 7.36664593e-01 -7.05277979e-01
-7.28389025e-01 -9.66431558e-01 -5.57335734e-01 1.04391360e+00
-1.02260567e-01 -5.93699574e-01 -3.59862179e-01 5.94011061e-02]
|
[12.808968544006348, 0.5515887141227722]
|
0e77c9db-30ea-49a1-b832-73dbbbcdac27
|
semanticstylegan-learning-compositional
|
2112.02236
| null |
https://arxiv.org/abs/2112.02236v3
|
https://arxiv.org/pdf/2112.02236v3.pdf
|
SemanticStyleGAN: Learning Compositional Generative Priors for Controllable Image Synthesis and Editing
|
Recent studies have shown that StyleGANs provide promising prior models for downstream tasks on image synthesis and editing. However, since the latent codes of StyleGANs are designed to control global styles, it is hard to achieve a fine-grained control over synthesized images. We present SemanticStyleGAN, where a generator is trained to model local semantic parts separately and synthesizes images in a compositional way. The structure and texture of different local parts are controlled by corresponding latent codes. Experimental results demonstrate that our model provides a strong disentanglement between different spatial areas. When combined with editing methods designed for StyleGANs, it can achieve a more fine-grained control to edit synthesized or real images. The model can also be extended to other domains via transfer learning. Thus, as a generic prior model with built-in disentanglement, it could facilitate the development of GAN-based applications and enable more potential downstream tasks.
|
['Xiaohui Shen', 'Yangyue Wan', 'Xiao Yang', 'Yichun Shi']
|
2021-12-04
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Shi_SemanticStyleGAN_Learning_Compositional_Generative_Priors_for_Controllable_Image_Synthesis_and_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Shi_SemanticStyleGAN_Learning_Compositional_Generative_Priors_for_Controllable_Image_Synthesis_and_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['facial-editing']
|
['computer-vision']
|
[ 5.67536354e-01 3.89245898e-01 -1.99204117e-01 -4.36895102e-01
-4.19330806e-01 -8.94024074e-01 9.36171055e-01 -6.99139297e-01
2.92049557e-01 6.31690621e-01 5.13757885e-01 -4.25041728e-02
5.59199691e-01 -1.16431439e+00 -8.56122136e-01 -8.01088512e-01
4.88768131e-01 2.52877444e-01 4.07681018e-02 -4.44111824e-01
-2.25799307e-02 6.36030316e-01 -1.19604540e+00 5.38221896e-01
7.60894775e-01 6.52328908e-01 4.47536498e-01 6.16831064e-01
-3.66891287e-02 8.64744663e-01 -6.63400471e-01 -2.11711675e-01
3.52331549e-01 -9.62986052e-01 -3.71428311e-01 2.94717789e-01
5.95672429e-01 -3.41242224e-01 -3.76483321e-01 9.95822132e-01
2.65126824e-01 -3.69167686e-01 7.85379410e-01 -1.14939761e+00
-1.12662923e+00 4.04322982e-01 -5.03506839e-01 -5.66415370e-01
1.50439665e-01 4.34394956e-01 8.79174948e-01 -4.60644603e-01
6.39681756e-01 1.47846413e+00 2.00254962e-01 1.07083571e+00
-1.59174371e+00 -8.97584796e-01 2.04783291e-01 -3.41638803e-01
-8.83239806e-01 -4.89613295e-01 7.14271009e-01 -4.03608948e-01
3.64263535e-01 2.92431653e-01 6.73074126e-01 1.44516313e+00
2.55610764e-01 7.77847469e-01 1.30983460e+00 -3.62849981e-01
9.02099237e-02 -4.37017679e-02 -1.02340484e+00 7.61566997e-01
-6.03182539e-02 3.94494206e-01 -5.98949790e-01 3.15890729e-01
1.42411363e+00 -1.54475078e-01 -2.57027149e-01 -5.26257694e-01
-1.53700233e+00 9.60274577e-01 6.30606711e-01 -2.42783148e-02
-4.63704206e-02 7.06422210e-01 3.57116461e-02 3.55897248e-01
3.10038596e-01 8.93114448e-01 -2.47643605e-01 1.73845112e-01
-9.44189668e-01 1.78511813e-01 5.54974616e-01 1.40356803e+00
7.78468847e-01 5.86018562e-01 -7.59426355e-01 6.10219538e-01
8.58296156e-02 6.47186339e-01 2.08312541e-01 -1.22905958e+00
2.66780019e-01 3.34844410e-01 4.24978212e-02 -7.96937346e-01
2.81721234e-01 -2.09603891e-01 -1.01119530e+00 7.03241110e-01
1.53689578e-01 -3.36647749e-01 -1.25659418e+00 1.96638644e+00
-7.46888071e-02 -1.80495996e-02 -3.04464459e-01 7.23020792e-01
3.48660380e-01 7.49048769e-01 1.80422142e-02 3.93743366e-01
1.40039015e+00 -1.28856635e+00 -6.51472211e-01 -4.69440192e-01
1.71044707e-01 -8.66972089e-01 1.19653738e+00 2.02771544e-01
-1.18344700e+00 -6.67904496e-01 -1.17617023e+00 -2.26104975e-01
-1.86908782e-01 2.11761758e-01 5.54459989e-01 5.62354624e-01
-1.30708849e+00 2.77438343e-01 -7.57574677e-01 -1.84915796e-01
5.86566210e-01 2.28019193e-01 -3.22995394e-01 7.57904649e-02
-1.02170241e+00 8.45298827e-01 7.58230463e-02 -1.89286634e-01
-1.31046891e+00 -7.39884317e-01 -1.01214385e+00 5.78009374e-02
3.35986242e-02 -1.15877604e+00 1.06152296e+00 -1.22205281e+00
-2.03527808e+00 1.01440334e+00 -4.30248416e-04 -1.82127297e-01
6.78562462e-01 6.02520518e-02 -1.70176476e-01 2.13917028e-02
2.17871889e-01 1.20074475e+00 1.28960156e+00 -1.36760092e+00
-4.17396337e-01 4.92375493e-02 2.19932765e-01 1.86554581e-01
-2.81854182e-01 -6.84084594e-02 -4.69950914e-01 -1.32965624e+00
-3.97436887e-01 -1.11215723e+00 -2.03564107e-01 4.98848230e-01
-3.73518765e-01 4.50001061e-01 1.02220690e+00 -4.45614278e-01
7.91528940e-01 -2.15174460e+00 7.45075226e-01 -5.51033616e-02
2.83701122e-01 2.34400511e-01 -5.38294017e-01 4.85220522e-01
-3.02481875e-02 2.15086430e-01 -2.43798673e-01 -3.13608855e-01
6.19681738e-03 1.14060804e-01 -4.67864662e-01 9.29167792e-02
5.03499329e-01 1.29414976e+00 -8.76935422e-01 -1.47934750e-01
2.68707603e-01 5.49993098e-01 -7.39886343e-01 4.34411138e-01
-5.31217813e-01 1.04117405e+00 -4.83125359e-01 2.98907816e-01
4.71884906e-01 -1.47382945e-01 2.61221439e-01 -2.76723176e-01
4.35827151e-02 1.33131012e-01 -6.25067770e-01 1.88798499e+00
-9.23623562e-01 8.50595474e-01 2.93978691e-01 -6.57230735e-01
9.48129892e-01 2.43815184e-01 4.91550565e-03 -6.52859926e-01
-5.72927995e-03 -2.79741567e-02 -1.81190565e-01 -2.24003065e-02
4.07237142e-01 -3.21680695e-01 -4.03160065e-01 4.66990471e-01
4.64131590e-03 -9.20864046e-01 -9.12606791e-02 1.49902701e-01
8.59060943e-01 4.23151642e-01 8.32779184e-02 -3.45349908e-01
2.63168305e-01 -3.20920259e-01 3.52515072e-01 6.14704370e-01
3.09212774e-01 1.04464507e+00 4.62111264e-01 -2.38409743e-01
-1.45744741e+00 -1.24321270e+00 2.49123380e-01 9.95791137e-01
3.34032509e-03 -3.69013906e-01 -8.69562685e-01 -6.12869501e-01
-1.11474045e-01 6.64973497e-01 -7.29287028e-01 -4.73526895e-01
-5.90059757e-01 -9.38448459e-02 6.37618780e-01 6.00538075e-01
7.98906446e-01 -1.04503310e+00 -3.64071369e-01 -6.33298606e-02
1.64030269e-02 -1.10021889e+00 -1.06088138e+00 -4.66238521e-02
-6.79355383e-01 -6.04774833e-01 -9.73589838e-01 -1.03008592e+00
8.43386292e-01 1.22785307e-01 1.19500744e+00 2.91311811e-03
4.83191349e-02 1.43138990e-01 -2.07179904e-01 -3.19095761e-01
-6.90369606e-01 1.67874455e-01 -3.08176816e-01 9.63013247e-02
-3.65160495e-01 -8.11555803e-01 -8.58248174e-01 4.76928145e-01
-1.11504900e+00 6.05967343e-01 5.20584524e-01 9.87066865e-01
2.88383901e-01 -1.99428633e-01 4.44323540e-01 -1.21311426e+00
5.62017679e-01 -7.44699165e-02 -7.17542052e-01 2.98773736e-01
-3.85292649e-01 3.67898434e-01 7.15781391e-01 -4.29627240e-01
-1.33853614e+00 -2.33742651e-02 1.39676899e-01 -2.58825153e-01
-1.09402634e-01 5.62078282e-02 -6.05876803e-01 -2.17230618e-01
5.67373216e-01 2.61269420e-01 1.43314183e-01 -2.59667635e-01
8.90613198e-01 3.58672172e-01 5.25652945e-01 -8.76651645e-01
1.08007360e+00 6.01599872e-01 -7.40702206e-04 -5.37419498e-01
-5.95463037e-01 4.88215655e-01 -3.62099707e-01 1.61993399e-01
1.26105475e+00 -1.27176416e+00 -2.47068137e-01 6.56973779e-01
-1.02034128e+00 -9.48179841e-01 -5.76536536e-01 -1.01240836e-01
-9.13067341e-01 -5.29621765e-02 -8.18527162e-01 -2.30018366e-02
-4.99615585e-03 -1.28368831e+00 1.42210042e+00 2.21651465e-01
-4.76284087e-01 -1.06972182e+00 -1.40979856e-01 2.43539095e-01
6.42568111e-01 3.34734410e-01 9.79954004e-01 2.57138073e-01
-9.18760657e-01 1.28610283e-01 -2.28809223e-01 3.14690292e-01
5.44410646e-01 6.58387467e-02 -7.57130206e-01 -3.19869965e-01
-1.53746918e-01 -3.81963670e-01 8.90597701e-01 2.24930778e-01
1.36437297e+00 -4.54308480e-01 -2.81959027e-01 9.95235205e-01
1.15396631e+00 2.39136785e-01 9.39928472e-01 9.32368338e-02
1.08110559e+00 3.59459519e-01 2.14963719e-01 8.56156722e-02
2.53326118e-01 8.18453848e-01 2.66829878e-01 -4.76088911e-01
-8.28583419e-01 -7.52135813e-01 3.98950964e-01 4.94370729e-01
6.87475875e-02 -3.66675109e-01 -3.61835331e-01 3.44306916e-01
-1.60826910e+00 -8.52529883e-01 3.45565945e-01 1.70743811e+00
1.12168419e+00 -1.35347396e-01 -1.61007375e-01 -3.41921896e-01
7.61414409e-01 5.05011737e-01 -5.97916901e-01 -4.64651912e-01
-1.95064366e-01 6.31494641e-01 4.93809372e-01 5.08294106e-01
-8.19317281e-01 1.34014583e+00 7.32972145e+00 9.25303221e-01
-1.35037482e+00 1.45664170e-01 7.43297100e-01 -1.31928250e-01
-9.00674403e-01 1.57789335e-01 -5.34546316e-01 5.17405689e-01
4.17915225e-01 1.56409055e-01 7.14785933e-01 4.35185581e-01
1.77294135e-01 3.06640178e-01 -1.22097421e+00 7.41870046e-01
-1.33010000e-01 -1.76604283e+00 4.21645790e-01 1.66662291e-01
1.47862852e+00 -2.81364471e-01 4.55921113e-01 9.09064263e-02
8.17967832e-01 -1.23622215e+00 1.07991815e+00 4.02865887e-01
1.44305563e+00 -5.49683034e-01 1.77902672e-02 5.35609126e-02
-9.90131617e-01 1.79487094e-01 -1.03030369e-01 -9.37115476e-02
2.49893114e-01 4.46853548e-01 -2.90341973e-01 1.92305148e-01
2.23811939e-01 8.47561300e-01 -3.55796546e-01 2.25882307e-01
-8.71972680e-01 2.85686523e-01 1.69104323e-01 1.44711062e-01
2.07186746e-03 -3.22042435e-01 2.88396329e-01 1.02883470e+00
5.27423084e-01 -5.94570078e-02 -2.70688049e-02 1.36492872e+00
-3.30003202e-01 -4.90142792e-01 -7.92681813e-01 -2.93788970e-01
4.43731874e-01 1.13371539e+00 -5.89327514e-01 -3.57457399e-01
-2.27001369e-01 1.55542314e+00 2.59503603e-01 6.71215653e-01
-9.98864233e-01 -4.06150162e-01 9.70787585e-01 2.10484222e-01
5.22793412e-01 -4.73631531e-01 -6.37487650e-01 -1.62511337e+00
-4.09737527e-01 -1.24470472e+00 -2.76688099e-01 -1.04636407e+00
-1.17369175e+00 6.12656593e-01 -1.50726840e-01 -1.06997442e+00
-1.79250345e-01 -6.63044631e-01 -8.01522851e-01 9.76143658e-01
-1.25899351e+00 -1.65515876e+00 -2.25899950e-01 5.69998741e-01
5.39495051e-01 -3.72374684e-01 1.03409803e+00 5.91613576e-02
-1.79064319e-01 7.06765652e-01 -1.20873526e-01 1.84735864e-01
8.49947810e-01 -1.20915639e+00 6.59865141e-01 8.84891570e-01
1.02116883e-01 5.15305758e-01 6.40799165e-01 -5.19017994e-01
-1.20121408e+00 -1.27478838e+00 3.51122975e-01 -6.04992807e-01
3.44416976e-01 -7.29158938e-01 -2.93905348e-01 9.17195737e-01
7.82732844e-01 -7.11378381e-02 4.24809724e-01 -2.22503111e-01
-5.90910316e-01 -2.55446404e-01 -9.77505982e-01 8.45545828e-01
1.30176520e+00 -9.40289855e-01 -1.29306421e-01 1.14731200e-01
8.43840897e-01 -5.16550601e-01 -6.39874816e-01 2.47927934e-01
5.14555156e-01 -9.48478878e-01 8.54249001e-01 -4.38772380e-01
1.01853740e+00 -4.87817585e-01 -3.58518437e-02 -1.76101375e+00
-5.77956676e-01 -7.06599712e-01 8.38094726e-02 1.43737829e+00
3.26727569e-01 -5.38938642e-01 7.69308567e-01 4.86658096e-01
-5.57818040e-02 -3.29335779e-01 -2.82894522e-01 -5.39050281e-01
2.93392956e-01 2.69530304e-02 9.60982800e-01 8.61223161e-01
-4.87470239e-01 3.71257305e-01 -7.43777752e-01 -6.54249564e-02
3.85042071e-01 2.59043753e-01 8.89887094e-01 -6.78768575e-01
-4.97976273e-01 -5.97320080e-01 -2.91785419e-01 -1.32183743e+00
3.04863244e-01 -8.12934339e-01 8.33877921e-02 -1.30739939e+00
-4.10289243e-02 -5.23324072e-01 1.25084534e-01 5.37505269e-01
8.89250413e-02 6.97178841e-01 4.27772313e-01 -3.87623459e-02
-1.72175989e-02 8.16518128e-01 2.06670427e+00 -3.27885956e-01
1.60368443e-01 -2.75769562e-01 -1.10792089e+00 4.01010424e-01
6.98801935e-01 -1.92609400e-01 -7.33901680e-01 -8.72645319e-01
2.00444803e-01 -1.26516595e-01 3.56055051e-01 -6.87458158e-01
-7.71304816e-02 -5.17868340e-01 4.57171738e-01 2.24994674e-01
2.75004745e-01 -7.13742197e-01 4.87414777e-01 3.13595653e-01
-5.10094762e-01 -1.80722952e-01 3.45981643e-02 4.90174562e-01
-3.65438074e-01 4.19811666e-01 9.81384099e-01 -4.92893934e-01
-4.02030289e-01 4.41564202e-01 -4.30482447e-01 -4.00844552e-02
1.07218790e+00 -1.47937268e-01 -1.86257169e-01 -7.89291918e-01
-6.39004230e-01 1.82702169e-02 1.02514660e+00 5.92705250e-01
4.97124732e-01 -1.81710982e+00 -5.89806497e-01 6.43216908e-01
9.04749110e-02 1.92815542e-01 -4.97969911e-02 1.13751359e-01
-7.80970991e-01 2.53896028e-01 -6.59600377e-01 -5.38663566e-01
-7.35253632e-01 4.85369653e-01 2.78344035e-01 -4.87808585e-02
-2.91763842e-01 8.74521911e-01 9.61513937e-01 -5.11279821e-01
-2.23256573e-01 -1.86254427e-01 3.64302546e-01 -2.19962493e-01
3.75254065e-01 -1.71435565e-01 -3.50962818e-01 -3.70962560e-01
1.73000455e-01 7.16824174e-01 1.54215202e-01 -2.62538195e-01
1.41019309e+00 -2.45544642e-01 -3.28872442e-01 5.18404655e-02
1.06400800e+00 2.84265429e-01 -1.93411887e+00 1.29967257e-01
-5.86683333e-01 -5.79563498e-01 -1.02187745e-01 -7.34351933e-01
-1.50622332e+00 9.61506188e-01 2.76724160e-01 -1.03776053e-01
1.31438255e+00 -7.21837729e-02 6.42482281e-01 -2.72651792e-01
4.87170398e-01 -7.56376863e-01 4.12153363e-01 3.37191612e-01
1.10657823e+00 -8.58155966e-01 -4.50675994e-01 -4.41994727e-01
-7.39303410e-01 9.66141582e-01 7.14745641e-01 -4.14073318e-01
4.51615304e-01 5.48576772e-01 1.10276341e-01 3.29485908e-03
-6.81112111e-01 1.75448984e-01 4.04449612e-01 8.07235301e-01
6.66576385e-01 2.81221867e-01 -3.73338573e-02 1.75911505e-02
-3.41427237e-01 -1.53982069e-03 4.44619447e-01 5.83832562e-01
6.17663115e-02 -1.70423007e+00 -1.41012445e-01 7.25982487e-02
-2.09807947e-01 -1.50473937e-01 -3.57281625e-01 5.30596673e-01
2.67242342e-01 6.79859638e-01 6.20717444e-02 -4.26229030e-01
3.90667021e-02 -2.86190599e-01 8.65020335e-01 -9.56810355e-01
-3.12022984e-01 3.68877761e-02 6.75980991e-04 -6.81622446e-01
-3.40411484e-01 -2.72430331e-01 -7.29528904e-01 -5.08951604e-01
4.98083495e-02 -2.01291800e-01 3.24260205e-01 6.63404882e-01
6.50847971e-01 8.08656871e-01 6.29614115e-01 -9.38496888e-01
-2.38409460e-01 -6.76419556e-01 -6.09608650e-01 4.07835096e-01
3.09124202e-01 -3.31074476e-01 1.28879901e-02 4.65354383e-01]
|
[11.655485153198242, -0.4387718439102173]
|
49b21da2-1727-4f30-ac5c-7d1c1f90f6b3
|
instance-segmentation-in-3d-scenes-using
|
2108.07478
| null |
https://arxiv.org/abs/2108.07478v1
|
https://arxiv.org/pdf/2108.07478v1.pdf
|
Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks
|
Instance segmentation in 3D scenes is fundamental in many applications of scene understanding. It is yet challenging due to the compound factors of data irregularity and uncertainty in the numbers of instances. State-of-the-art methods largely rely on a general pipeline that first learns point-wise features discriminative at semantic and instance levels, followed by a separate step of point grouping for proposing object instances. While promising, they have the shortcomings that (1) the second step is not supervised by the main objective of instance segmentation, and (2) their point-wise feature learning and grouping are less effective to deal with data irregularities, possibly resulting in fragmented segmentations. To address these issues, we propose in this work an end-to-end solution of Semantic Superpoint Tree Network (SSTNet) for proposing object instances from scene points. Key in SSTNet is an intermediate, semantic superpoint tree (SST), which is constructed based on the learned semantic features of superpoints, and which will be traversed and split at intermediate tree nodes for proposals of object instances. We also design in SSTNet a refinement module, termed CliqueNet, to prune superpoints that may be wrongly grouped into instance proposals. Experiments on the benchmarks of ScanNet and S3DIS show the efficacy of our proposed method. At the time of submission, SSTNet ranks top on the ScanNet (V2) leaderboard, with 2% higher of mAP than the second best method. The source code in PyTorch is available at https://github.com/Gorilla-Lab-SCUT/SSTNet.
|
['Kui Jia', 'Mingkui Tan', 'Songcen Xu', 'Zhihao LI', 'Zhihao Liang']
|
2021-08-17
| null |
http://openaccess.thecvf.com//content/ICCV2021/html/Liang_Instance_Segmentation_in_3D_Scenes_Using_Semantic_Superpoint_Tree_Networks_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Liang_Instance_Segmentation_in_3D_Scenes_Using_Semantic_Superpoint_Tree_Networks_ICCV_2021_paper.pdf
|
iccv-2021-1
|
['3d-instance-segmentation-1']
|
['computer-vision']
|
[ 1.73848465e-01 4.17217284e-01 -9.69371423e-02 -5.73579729e-01
-7.18625247e-01 -4.71528769e-01 4.73315507e-01 2.87517428e-01
-1.46782130e-01 3.68799865e-01 -2.24565327e-01 -9.58850160e-02
-2.24175498e-01 -7.56173670e-01 -8.66447628e-01 -4.32593346e-01
-1.22208469e-01 8.62042189e-01 9.53707635e-01 -1.18876761e-02
5.47965705e-01 4.89291072e-01 -1.75706661e+00 3.24950337e-01
9.69292402e-01 1.18531513e+00 5.15742123e-01 1.93097964e-01
-6.63837969e-01 1.57056570e-01 -3.31582576e-01 -1.59688830e-01
5.56188762e-01 -2.11304456e-01 -1.13499188e+00 3.15363377e-01
6.10480011e-01 -3.08983475e-02 1.47697747e-01 1.01809835e+00
1.16471723e-01 1.08404689e-01 4.30327564e-01 -1.31353188e+00
7.52454847e-02 5.53769946e-01 -8.21910322e-01 4.97970693e-02
-3.10267117e-02 2.24769294e-01 1.12974060e+00 -1.16640818e+00
6.34771526e-01 1.27205813e+00 6.02545619e-01 3.35945159e-01
-1.15294826e+00 -4.85060155e-01 5.72582364e-01 2.92902410e-01
-1.44231594e+00 -2.30098695e-01 7.50771165e-01 -2.78339446e-01
7.94382632e-01 2.08922237e-01 6.96480572e-01 5.92359006e-01
-2.74971038e-01 1.05411303e+00 8.56255889e-01 1.04585096e-01
5.57217121e-01 2.43471581e-02 2.62899935e-01 5.49361765e-01
9.21006948e-02 -2.16723934e-01 -2.90702969e-01 -3.52116488e-02
8.31411779e-01 1.03385948e-01 1.70115270e-02 -8.25837374e-01
-1.16503835e+00 8.06164861e-01 1.01896930e+00 1.76861033e-01
-5.86240470e-01 5.04993834e-02 2.70769626e-01 -9.18008760e-02
5.03089726e-01 3.95956904e-01 -8.87304664e-01 9.17737633e-02
-1.27595615e+00 4.29654747e-01 6.62343800e-01 1.04030621e+00
1.09244490e+00 -4.68861818e-01 -6.60498813e-02 8.27243626e-01
2.80630082e-01 1.21319622e-01 -9.87962727e-03 -7.70275831e-01
4.62585300e-01 1.04379761e+00 -1.93427742e-01 -6.90400302e-01
-5.12786686e-01 -6.80216372e-01 -4.89461213e-01 1.29570782e-01
2.70244598e-01 3.14848810e-01 -1.41860080e+00 1.17410386e+00
8.27546656e-01 6.41932845e-01 -3.22725922e-01 1.16408300e+00
1.12652791e+00 5.45683384e-01 1.81443706e-01 2.82434642e-01
1.30063379e+00 -1.16454327e+00 -7.30144151e-04 -3.92835289e-01
4.59117770e-01 -6.72398925e-01 9.16717231e-01 2.82256275e-01
-1.10749733e+00 -6.46097064e-01 -6.93378806e-01 -1.17744811e-01
-3.66433799e-01 -4.81613129e-02 6.02187514e-01 2.88717151e-01
-9.19121504e-01 5.04529536e-01 -8.46746504e-01 -4.86781120e-01
9.35206413e-01 4.13802952e-01 -2.32567027e-01 -9.43104327e-02
-5.97826600e-01 4.73776847e-01 6.40229642e-01 3.01962525e-01
-8.56979072e-01 -1.03348517e+00 -7.33334243e-01 7.97393769e-02
7.98115671e-01 -6.95268452e-01 1.10867298e+00 -7.82119095e-01
-1.15124905e+00 8.42495501e-01 -2.15314627e-01 -3.47828865e-01
5.84801733e-01 -2.45135695e-01 1.46526694e-01 7.54148513e-02
4.99670267e-01 1.14622700e+00 7.47356713e-01 -1.54838431e+00
-1.19044447e+00 -5.95546544e-01 6.30545169e-02 2.57239372e-01
4.29318488e-01 -4.97149348e-01 -1.00414550e+00 -2.80532420e-01
8.62365723e-01 -7.74371028e-01 -6.23061895e-01 -1.37592673e-01
-6.33000910e-01 -5.59536695e-01 9.77322161e-01 -2.49754742e-01
8.19447756e-01 -2.14440727e+00 1.28138945e-01 4.38797504e-01
1.98588446e-01 2.48859301e-01 -2.62669206e-01 2.34671772e-01
6.25529289e-02 1.35443866e-01 -5.18702447e-01 -4.90137607e-01
-1.07841477e-01 2.76394099e-01 -2.32808828e-01 3.73962611e-01
4.90051359e-01 9.54158485e-01 -9.94829476e-01 -3.71651947e-01
5.81349432e-01 1.85590178e-01 -6.41441762e-01 -1.48298657e-02
-7.09693909e-01 6.09590352e-01 -7.25234151e-01 8.65851641e-01
1.04716146e+00 -3.86080623e-01 -2.16531277e-01 -2.70230412e-01
-2.75673926e-01 5.92504561e-01 -1.62978220e+00 2.10733128e+00
1.18675586e-02 5.45240529e-02 -1.85622927e-02 -1.02352047e+00
9.06900287e-01 -1.59845054e-01 4.94050831e-01 -5.56912184e-01
-1.26465961e-01 3.07857126e-01 -3.21248293e-01 -2.48537794e-01
6.03408873e-01 2.48655424e-01 -1.85749419e-02 7.00268894e-02
1.42706886e-01 -5.29797971e-01 2.07318932e-01 3.11162829e-01
1.01684272e+00 3.41483116e-01 1.49872806e-02 -3.15623879e-01
4.71470624e-01 3.83915126e-01 7.21988499e-01 8.12750399e-01
-1.02593519e-01 7.89340794e-01 5.68243146e-01 -6.33741736e-01
-9.60785031e-01 -1.16816366e+00 -4.16411757e-01 8.04461658e-01
7.54880965e-01 -4.48663741e-01 -6.25376582e-01 -9.70157743e-01
5.71676381e-02 6.58529043e-01 -4.40975696e-01 2.91494727e-01
-4.98655587e-01 -3.78013223e-01 1.20845139e-01 3.91848981e-01
4.95594352e-01 -1.16426694e+00 -6.38980806e-01 3.09883505e-01
-3.46083790e-02 -1.21455574e+00 -1.15817174e-01 2.91646510e-01
-1.11207032e+00 -1.26293278e+00 -3.91385019e-01 -5.38404346e-01
9.01892602e-01 5.37444234e-01 1.21047091e+00 2.77305931e-01
-2.08805934e-01 1.55487671e-01 -5.55462182e-01 -3.85998785e-01
7.07098693e-02 5.71453512e-01 -4.52053428e-01 -9.74537581e-02
3.71720523e-01 -3.98598671e-01 -8.56134355e-01 5.99351883e-01
-9.09563422e-01 2.02581868e-01 6.13109827e-01 5.24298906e-01
1.06762004e+00 8.27696770e-02 3.06710392e-01 -1.12079012e+00
-1.22945011e-01 -5.88570714e-01 -8.51719260e-01 3.75512689e-02
-2.29442611e-01 -2.01367393e-01 2.19154194e-01 1.03090003e-01
-7.98496842e-01 3.61807972e-01 -1.82109848e-01 -4.40713853e-01
-6.14512801e-01 2.97004759e-01 -7.92415887e-02 -2.65505724e-02
4.92804557e-01 6.10755421e-02 -2.58401960e-01 -5.98705411e-01
4.88500893e-01 2.37417236e-01 3.00665945e-01 -5.31530619e-01
8.61358941e-01 5.85831106e-01 5.09351818e-03 -7.89467812e-01
-9.77095962e-01 -8.54431927e-01 -8.27113092e-01 -1.02013454e-01
7.67750025e-01 -9.63053405e-01 -3.38277370e-01 2.72748202e-01
-1.09162211e+00 -2.71482587e-01 -4.59421128e-01 1.79897651e-01
-4.78107035e-01 1.22194305e-01 -1.73492745e-01 -5.18222511e-01
-8.06111470e-02 -1.23660874e+00 1.50671911e+00 3.74372512e-01
5.49189597e-02 -6.50088131e-01 -2.57785857e-01 5.28125942e-01
7.90900439e-02 2.28202075e-01 7.19969690e-01 -6.35693669e-01
-1.18270469e+00 -2.15429123e-02 -3.94129306e-01 1.32875100e-01
-8.34387317e-02 -5.45194671e-02 -8.97055328e-01 -1.59141839e-01
-1.87405795e-01 -2.39920288e-01 1.01272857e+00 6.26977682e-01
1.42915094e+00 1.03272498e-01 -6.05995953e-01 8.03361893e-01
1.54892373e+00 -2.15711817e-01 6.85113490e-01 2.66731590e-01
8.01117241e-01 7.42767215e-01 9.07727003e-01 4.10999775e-01
5.85320592e-01 6.38441920e-01 9.22194719e-01 -3.86081576e-01
-1.14394277e-01 -3.13641459e-01 -1.51888967e-01 3.38507682e-01
1.06269605e-01 -4.62222658e-02 -1.00073218e+00 7.87273526e-01
-1.99365175e+00 -6.34468675e-01 -6.14365041e-01 2.06792641e+00
4.27604556e-01 3.28137308e-01 3.01203728e-01 -2.68944679e-03
6.55523062e-01 6.66000172e-02 -6.32185936e-01 -1.03620142e-01
1.32040486e-01 3.66086721e-01 6.56114578e-01 3.40489984e-01
-1.17463088e+00 1.37667608e+00 4.64497757e+00 9.05525506e-01
-9.57452536e-01 2.62647886e-02 6.60753429e-01 3.29430476e-02
-1.00232363e-01 3.67107600e-01 -1.00433373e+00 4.07664359e-01
1.85872465e-01 3.24088126e-01 1.11158520e-01 9.10167813e-01
2.02193797e-01 -3.97349596e-01 -1.10729480e+00 6.93135858e-01
-2.85743594e-01 -1.33432913e+00 3.73284593e-02 -5.58042601e-02
8.14459860e-01 5.58962941e-01 -1.47923484e-01 2.52016693e-01
9.68228132e-02 -7.65784204e-01 8.18278909e-01 9.84314755e-02
3.75156015e-01 -7.38736272e-01 6.18741274e-01 4.58747387e-01
-1.35005140e+00 5.06204478e-02 -6.18942678e-01 2.16613770e-01
1.86457038e-01 8.16459835e-01 -1.11326468e+00 8.05260897e-01
9.26163614e-01 8.23419333e-01 -5.96142232e-01 1.53996909e+00
-4.13736135e-01 5.81458390e-01 -6.57935262e-01 2.56638497e-01
6.72451735e-01 -1.06452197e-01 6.70824587e-01 9.36352074e-01
2.03502595e-01 1.63082197e-01 5.45436263e-01 1.16853869e+00
1.44822299e-01 -2.94436384e-02 -2.33087942e-01 4.79541749e-01
5.37090957e-01 1.51420534e+00 -1.30762184e+00 -2.67788589e-01
-2.15170011e-01 7.39582121e-01 3.19114536e-01 1.60344392e-01
-7.82652974e-01 -6.30090088e-02 5.52945197e-01 3.16912442e-01
7.25322247e-01 -1.95335865e-01 -5.93666375e-01 -9.22430873e-01
2.18872070e-01 -4.44924206e-01 3.61245751e-01 -4.53402519e-01
-1.27598202e+00 5.02498448e-01 2.24419553e-02 -1.14794135e+00
1.63985372e-01 -3.21066678e-01 -5.53763092e-01 7.02017963e-01
-1.62262452e+00 -1.19591057e+00 -4.49667037e-01 6.30897820e-01
1.04558277e+00 3.37365806e-01 4.07168567e-01 1.73968390e-01
-3.04685593e-01 1.99456736e-01 -2.96863019e-01 -1.64486662e-01
1.37926549e-01 -1.31664979e+00 6.50790036e-01 7.26544917e-01
2.85989404e-01 2.80459911e-01 5.34160078e-01 -7.47565210e-01
-1.02005875e+00 -1.30544853e+00 6.96378291e-01 -5.51834226e-01
3.63406211e-01 -5.29476643e-01 -1.07822692e+00 3.74701530e-01
-3.08666021e-01 2.61747032e-01 2.50337809e-01 2.07067341e-01
-9.06605348e-02 -6.13926686e-02 -1.22259414e+00 3.25623870e-01
1.25696123e+00 2.34241392e-02 -6.06672645e-01 4.69671607e-01
8.10493290e-01 -6.63183212e-01 -6.19904339e-01 6.29035711e-01
1.12828091e-01 -1.06340146e+00 9.91882205e-01 -2.06316471e-01
4.07392830e-01 -8.25056136e-01 3.00320350e-02 -1.03054070e+00
-3.42631876e-01 -2.11798996e-01 2.16286704e-01 1.11205268e+00
5.25159001e-01 -4.93106961e-01 1.12249982e+00 3.75077426e-01
-4.54697996e-01 -9.03834343e-01 -1.08561921e+00 -7.39938676e-01
-2.63818473e-01 -6.01950645e-01 8.22388649e-01 7.15601206e-01
-5.26470006e-01 2.45785117e-01 2.01707512e-01 5.17732620e-01
8.25110018e-01 2.29683384e-01 1.16215742e+00 -1.31870043e+00
1.25565603e-02 -4.83246595e-01 -5.28907359e-01 -1.28154016e+00
-1.82557151e-01 -9.60329592e-01 2.70334214e-01 -1.78035223e+00
-3.86168920e-02 -1.05727315e+00 -2.53278434e-01 5.16295016e-01
-3.02162081e-01 1.09021455e-01 3.01842093e-01 3.23545694e-01
-9.29994345e-01 5.42554557e-01 1.20463133e+00 1.56765487e-02
-6.37733757e-01 2.51819491e-01 -4.35869128e-01 7.55700350e-01
6.92340553e-01 -6.44515038e-01 -4.47421610e-01 -5.21420062e-01
-4.42300066e-02 -2.26343811e-01 7.00022280e-01 -1.07562983e+00
2.84349233e-01 -1.85003415e-01 2.37115726e-01 -1.31246626e+00
3.36464792e-01 -9.49015558e-01 1.35498926e-01 2.75136650e-01
-7.99390376e-02 -2.52909154e-01 1.29667014e-01 5.48466742e-01
-1.71332583e-01 -1.75700113e-01 6.56628370e-01 -3.65340054e-01
-1.09705210e+00 6.06529593e-01 2.51311481e-01 -5.74031435e-02
1.16881180e+00 -5.96147537e-01 -1.34072050e-01 3.28085840e-01
-5.71056664e-01 6.67611778e-01 5.84105134e-01 5.31001508e-01
5.96139252e-01 -8.61392975e-01 -6.82749510e-01 2.69337684e-01
2.92032450e-01 1.00111866e+00 3.82756680e-01 8.50822449e-01
-5.30870974e-01 2.12792724e-01 2.00125366e-01 -1.29612935e+00
-8.84608269e-01 2.62335628e-01 1.17000893e-01 -6.92345649e-02
-8.59757006e-01 1.08843768e+00 4.92813915e-01 -6.41020834e-01
2.23087385e-01 -4.92499709e-01 -6.77172914e-02 -1.42647073e-01
1.20968468e-01 2.32960686e-01 2.19410986e-01 -5.97625256e-01
-5.49455345e-01 7.27278173e-01 -2.29909286e-01 2.02834055e-01
1.56362247e+00 3.14125679e-02 -2.43268609e-01 1.63988009e-01
8.62842500e-01 -3.90846491e-01 -1.32376730e+00 -2.78773069e-01
2.97418207e-01 -7.42510319e-01 1.66005120e-02 -8.08406353e-01
-1.16565335e+00 6.37375295e-01 4.52078015e-01 2.38473997e-01
9.69527304e-01 4.80049014e-01 7.79662549e-01 8.15514997e-02
5.54039299e-01 -9.69020665e-01 -2.40786612e-01 4.94685799e-01
7.59610772e-01 -1.21799707e+00 8.65139905e-03 -8.55653465e-01
-4.30582076e-01 9.23718572e-01 8.80008519e-01 -2.34830409e-01
6.23947442e-01 -1.33505002e-01 -1.45252392e-01 -5.97936332e-01
-6.50828183e-01 -3.49911422e-01 3.47828537e-01 4.01566714e-01
5.74584119e-03 1.22889727e-01 -2.60635942e-01 3.86263371e-01
-1.26462072e-01 -7.40977302e-02 1.18427590e-01 7.74830222e-01
-6.63188219e-01 -1.05482781e+00 -3.51686984e-01 5.83670020e-01
-1.83404654e-01 5.15685938e-02 -2.48591080e-01 7.52633691e-01
5.71801007e-01 7.92808235e-01 2.04568610e-01 -1.19567707e-01
5.46222925e-01 -3.44251662e-01 1.60507664e-01 -9.43889856e-01
-6.52251065e-01 7.77638853e-02 -1.68438479e-01 -9.74896371e-01
-5.12176216e-01 -7.02976406e-01 -1.54956055e+00 9.21825413e-03
-4.68111038e-01 1.21745244e-01 9.14115489e-01 9.08225536e-01
5.96076429e-01 3.60132337e-01 6.08437181e-01 -1.10267341e+00
-1.32182345e-01 -6.59501731e-01 -4.16051090e-01 4.42393541e-01
4.76847701e-02 -6.21624351e-01 -2.07268342e-01 -1.60685033e-01]
|
[7.999940395355225, -3.0723824501037598]
|
f75e2900-4a2b-4f19-bd78-ec7913f3676e
|
hedging-against-complexity-distributionally
|
2212.01518
| null |
https://arxiv.org/abs/2212.01518v1
|
https://arxiv.org/pdf/2212.01518v1.pdf
|
Hedging against Complexity: Distributionally Robust Optimization with Parametric Approximation
|
Empirical risk minimization (ERM) and distributionally robust optimization (DRO) are popular approaches for solving stochastic optimization problems that appear in operations management and machine learning. Existing generalization error bounds for these methods depend on either the complexity of the cost function or dimension of the uncertain parameters; consequently, the performance of these methods is poor for high-dimensional problems with objective functions under high complexity. We propose a simple approach in which the distribution of uncertain parameters is approximated using a parametric family of distributions. This mitigates both sources of complexity; however, it introduces a model misspecification error. We show that this new source of error can be controlled by suitable DRO formulations. Our proposed parametric DRO approach has significantly improved generalization bounds over existing ERM / DRO methods and parametric ERM for a wide variety of settings. Our method is particularly effective under distribution shifts. We also illustrate the superior performance of our approach on both synthetic and real-data portfolio optimization and regression tasks.
|
['Tianyu Wang', 'Henry Lam', 'Garud Iyengar']
|
2022-12-03
| null | null | null | null |
['portfolio-optimization']
|
['time-series']
|
[ 1.79105103e-02 -1.58574298e-01 -1.34990335e-01 -4.17706847e-01
-1.05389988e+00 -6.20669007e-01 3.23574603e-01 3.74610513e-01
-3.67588639e-01 1.04355001e+00 -2.15595976e-01 -3.59690309e-01
-7.96472907e-01 -6.07488751e-01 -6.53353751e-01 -8.79662037e-01
-1.11854590e-01 4.68223244e-01 -1.47361085e-01 -1.53595269e-01
3.75386685e-01 4.95586723e-01 -1.26479375e+00 -4.73839849e-01
1.25119340e+00 1.59419894e+00 -7.26488680e-02 2.85398990e-01
1.57572389e-01 1.65706441e-01 -7.84079969e-01 -4.91476685e-01
6.93250716e-01 1.17548615e-01 -1.34831697e-01 1.74998313e-01
-1.01168081e-01 6.01074882e-02 1.55491203e-01 1.30743062e+00
6.34163201e-01 5.44661522e-01 9.72139955e-01 -1.28508651e+00
-3.97752613e-01 5.31438947e-01 -7.77749002e-01 2.39455670e-01
-1.59417048e-01 -1.42904639e-01 9.62002099e-01 -9.81772184e-01
-1.21494152e-01 1.14265382e+00 6.77367449e-01 1.80786088e-01
-1.10463297e+00 -5.41000426e-01 3.41248959e-01 -4.10050005e-01
-1.36576188e+00 -2.14929283e-01 6.07435763e-01 -6.45223975e-01
4.97309864e-01 1.34460464e-01 3.25834565e-02 6.52904212e-01
3.35111231e-01 4.44166899e-01 1.08257997e+00 -2.35603929e-01
5.65641761e-01 4.78507549e-01 -2.99310628e-02 2.33625799e-01
6.59458280e-01 3.71180654e-01 -1.22613594e-01 -5.80529034e-01
5.97410440e-01 6.76698759e-02 -3.35675210e-01 -4.24769163e-01
-8.87809515e-01 1.12438321e+00 -6.43620566e-02 -1.83143944e-01
-4.56054777e-01 2.58276127e-02 1.66947201e-01 3.92836481e-01
7.67384231e-01 8.11754704e-01 -5.94846606e-01 -2.28246693e-02
-7.10264921e-01 4.61660802e-01 9.97558773e-01 1.10509253e+00
1.80793688e-01 3.10246438e-01 -2.94267237e-01 8.85539174e-01
3.46228033e-01 5.27132511e-01 1.82807177e-01 -8.34702969e-01
9.88020539e-01 1.64379448e-01 7.43433475e-01 -1.19893515e+00
-3.69707614e-01 -8.74981284e-01 -4.87336367e-01 2.30587170e-01
6.04542553e-01 -4.67349410e-01 -5.79088688e-01 1.71865392e+00
3.60986620e-01 -3.10463384e-02 9.09674391e-02 8.03782821e-01
-2.19792664e-01 6.65333450e-01 -2.23099962e-01 -6.01467788e-01
9.67754662e-01 -7.84642935e-01 -9.88641500e-01 -1.58227280e-01
3.84414434e-01 -6.26490414e-01 1.07384253e+00 5.38503528e-01
-1.15143859e+00 6.69677630e-02 -1.15148199e+00 7.35679865e-01
-1.02997229e-01 7.53525831e-03 4.04351056e-01 8.75131369e-01
-4.56166595e-01 7.21224904e-01 -5.23241162e-01 3.35600108e-01
3.49630743e-01 4.44387108e-01 2.10876182e-01 2.30098903e-01
-1.11312878e+00 8.18089426e-01 3.81087124e-01 4.17231351e-01
-7.13609576e-01 -9.94813561e-01 -7.55705476e-01 1.67007551e-01
9.21634734e-01 -4.90637153e-01 1.12282133e+00 -6.14114940e-01
-1.62816679e+00 6.14282824e-02 3.66614252e-01 -5.68653762e-01
6.87012196e-01 -5.24072945e-01 -3.34643245e-01 -1.55665159e-01
-2.81593442e-01 -4.40123945e-01 1.09262347e+00 -1.08171022e+00
-5.71457684e-01 -3.72959495e-01 2.06690691e-02 1.49012476e-01
-2.82265306e-01 1.43860251e-01 2.41985098e-01 -1.13742328e+00
-1.35819212e-01 -8.26594710e-01 -5.97640932e-01 -2.95983583e-01
-2.68694490e-01 6.29951954e-02 2.11679935e-01 -4.88180548e-01
1.53334594e+00 -1.96988010e+00 1.84944317e-01 4.98610646e-01
-3.23415756e-01 1.84057876e-02 3.95045549e-01 4.57769215e-01
-6.47254959e-02 4.05411720e-01 -7.43095875e-01 -3.82884949e-01
3.96640092e-01 7.20305070e-02 -5.40194452e-01 6.06144786e-01
1.84110403e-01 4.34874803e-01 -6.54696584e-01 -2.67740823e-02
-5.43164015e-02 8.13376084e-02 -4.94928330e-01 2.17587441e-01
-2.93853670e-01 3.95708345e-02 -5.82403362e-01 6.77385271e-01
6.33347273e-01 -5.67546226e-02 -1.53646454e-01 8.24313536e-02
6.74419329e-02 -8.59427378e-02 -1.71768236e+00 9.97850299e-01
-6.12166047e-01 1.73308719e-02 1.72387108e-01 -1.39040804e+00
9.21629727e-01 6.81488812e-02 2.98216760e-01 4.60824221e-02
1.63825050e-01 3.46515685e-01 -1.58217117e-01 -3.00568104e-01
3.82034391e-01 -7.41740763e-01 -3.10125172e-01 3.85035634e-01
-3.36414486e-01 -1.94850922e-01 -4.40856162e-03 -4.32352662e-01
5.57101130e-01 -1.12643883e-01 4.03088391e-01 -7.10069299e-01
4.05507058e-01 -4.26500142e-01 1.00541437e+00 6.32291734e-01
-2.28841808e-02 4.63229686e-01 8.11770201e-01 1.29716635e-01
-7.25771070e-01 -8.95637333e-01 -4.16935086e-01 7.34928071e-01
3.36313876e-03 1.72610860e-02 -3.72768641e-01 -6.07530117e-01
5.88748097e-01 1.01549530e+00 -5.81542790e-01 -1.84724942e-01
-3.35511088e-01 -1.38933873e+00 1.01551518e-01 5.75847208e-01
-8.76074657e-02 -3.83679211e-01 -2.75865942e-01 3.39117378e-01
3.24044317e-01 -1.05052543e+00 -6.99377418e-01 9.19075981e-02
-8.34637463e-01 -9.31306720e-01 -7.60887623e-01 -1.48439392e-01
7.11788356e-01 -2.40525261e-01 9.07095432e-01 -5.89008272e-01
5.20624220e-03 3.02904904e-01 2.45842021e-02 -8.76825452e-01
-1.35623470e-01 -3.28571349e-01 4.63821441e-01 4.35775071e-01
-2.82142848e-01 -3.18816453e-01 -5.34235716e-01 5.54374218e-01
-9.93960261e-01 -8.16224694e-01 3.35536182e-01 1.09434044e+00
7.48734593e-01 5.54093838e-01 1.06201601e+00 -9.94462252e-01
1.11283767e+00 -8.54450166e-01 -1.36935282e+00 4.43624228e-01
-1.03748798e+00 3.24280232e-01 6.53430820e-01 -6.36723340e-01
-1.19725430e+00 -3.87230009e-01 5.22175133e-01 -5.54790437e-01
4.75631326e-01 9.18586016e-01 -1.61120251e-01 -6.01732656e-02
3.09890300e-01 -2.37776950e-01 -5.61397448e-02 -6.21080279e-01
-2.19009116e-01 5.33211410e-01 2.97096997e-01 -1.03808308e+00
8.92937243e-01 1.51351735e-01 4.17960048e-01 -3.13878298e-01
-1.05440128e+00 -1.66068166e-01 -3.32059368e-04 1.28395066e-01
3.09685022e-01 -6.80090785e-01 -6.21478140e-01 3.18804324e-01
-5.39487422e-01 -3.54517624e-02 -5.29874325e-01 7.53133774e-01
-7.12772906e-01 3.96196872e-01 -1.63698480e-01 -1.45757735e+00
-1.24398097e-01 -1.30789506e+00 7.85466015e-01 1.39737502e-01
2.23206252e-01 -1.32534921e+00 -8.77320766e-02 1.48258373e-01
5.01202404e-01 8.97886038e-01 1.00800014e+00 -8.41441572e-01
-1.58713028e-01 -4.83481973e-01 1.28438652e-01 6.77929938e-01
8.52475092e-02 2.66687348e-02 -4.42949921e-01 -4.76739585e-01
6.73384130e-01 -3.68756242e-02 5.40633678e-01 5.38616121e-01
1.40239584e+00 -5.95136881e-01 -8.12062807e-03 7.05029726e-01
1.56623185e+00 4.51026917e-01 -1.06295561e-02 5.02494395e-01
2.61557490e-01 9.23081875e-01 1.18350410e+00 9.43915009e-01
2.29207635e-01 5.64512134e-01 4.81352806e-01 2.64653832e-01
9.58869934e-01 4.42213565e-02 3.60560715e-01 4.51504022e-01
1.22522376e-02 -3.59218389e-01 -9.19767201e-01 4.70635861e-01
-2.11051917e+00 -6.60163462e-01 3.33632827e-01 2.84182668e+00
7.43735790e-01 2.55606979e-01 3.14246178e-01 5.62575236e-02
9.25198853e-01 -1.10189535e-01 -7.86510110e-01 -5.57271957e-01
-2.86074072e-01 4.33281511e-02 8.65425289e-01 5.67053974e-01
-1.01230645e+00 2.25559682e-01 6.33274460e+00 9.45047081e-01
-5.60224712e-01 -1.47855625e-01 7.17868686e-01 -4.05714691e-01
-3.19606543e-01 -1.61096856e-01 -9.48133469e-01 7.39955366e-01
9.93307829e-01 -6.86940908e-01 3.10565323e-01 9.39242899e-01
2.53017038e-01 -7.50368508e-03 -1.00688887e+00 9.38619971e-01
-1.92690805e-01 -1.07668221e+00 -2.39545405e-01 2.94242591e-01
9.78620529e-01 -4.37164217e-01 5.11611879e-01 2.71329641e-01
1.16929427e-01 -1.22268784e+00 6.29011154e-01 7.70148993e-01
4.04771268e-01 -1.35412014e+00 1.14913559e+00 4.05730426e-01
-6.75537765e-01 -6.49316430e-01 -4.82484221e-01 1.01324335e-01
3.69641423e-01 9.49978709e-01 -3.06231141e-01 6.46163523e-01
4.73451942e-01 2.26911306e-01 -1.65210307e-01 1.27195323e+00
6.23399243e-02 5.29344082e-01 -6.65772200e-01 -3.13038416e-02
1.82755038e-01 -6.40629828e-01 8.72531116e-01 8.64302456e-01
4.73609924e-01 -6.76774532e-02 1.88743711e-01 8.51321399e-01
1.49853036e-01 1.78533778e-01 -4.51961726e-01 -2.08736375e-01
6.21288300e-01 8.65199685e-01 -2.83300042e-01 1.20235346e-01
-3.90253484e-01 9.62392315e-02 3.73928249e-02 5.96613348e-01
-9.41019773e-01 -7.31128573e-01 7.29207754e-01 -7.84353539e-02
3.71879965e-01 -2.70926178e-01 -3.68926287e-01 -1.18092382e+00
4.93573487e-01 -9.26608443e-01 7.40552247e-01 6.10992275e-02
-1.52450228e+00 3.27522397e-01 3.20825905e-01 -1.24922657e+00
-4.40044045e-01 -6.76552355e-01 -4.25852418e-01 8.14167678e-01
-1.58018410e+00 -3.50230724e-01 2.94985622e-01 3.29476982e-01
3.60822260e-01 -4.12534624e-01 3.96271586e-01 1.40689120e-01
-9.85988915e-01 7.31206000e-01 7.68445134e-01 -4.66988683e-01
5.69756627e-01 -1.55695498e+00 -1.39162689e-01 8.36331606e-01
-4.90009904e-01 5.02721190e-01 1.05602074e+00 -3.84543091e-01
-1.30421972e+00 -1.07187545e+00 2.81559944e-01 -3.39182049e-01
1.02895892e+00 -3.94916207e-01 -8.39416504e-01 4.89601851e-01
-3.58191669e-01 -3.43111046e-02 8.17841291e-01 -6.55503199e-02
9.21885110e-03 -2.59051323e-01 -1.49430346e+00 4.73699242e-01
5.27730048e-01 -7.93920010e-02 -4.74738836e-01 6.16265416e-01
8.19620490e-01 -3.57693881e-01 -1.47585475e+00 7.17616022e-01
2.71119297e-01 -6.78290904e-01 8.52609396e-01 -8.11422110e-01
4.16465215e-02 -2.76026487e-01 -4.86183077e-01 -1.48345590e+00
1.43071502e-01 -1.30139446e+00 -4.61273670e-01 1.24373293e+00
5.55851996e-01 -1.28535616e+00 1.93429321e-01 1.00555968e+00
2.84785517e-02 -1.23140121e+00 -9.51412737e-01 -1.28119016e+00
3.46205652e-01 -4.66175973e-01 7.81800926e-01 6.97915673e-01
-1.42775029e-01 -1.57722756e-01 -3.89982045e-01 4.79982883e-01
1.03476703e+00 2.24983409e-01 4.76518184e-01 -1.08728313e+00
-6.15917742e-01 -4.95630771e-01 -1.28584146e-01 -5.95674813e-01
2.20101088e-01 -4.45805669e-01 1.34146065e-02 -7.24179089e-01
-3.71015996e-01 -6.13055587e-01 -6.06916010e-01 -8.22662115e-02
-3.83769751e-01 -4.62101132e-01 8.11558068e-02 -5.56964912e-02
-2.33589053e-01 1.00532615e+00 9.52340841e-01 1.83605537e-01
-3.73834401e-01 7.51990199e-01 -9.06539381e-01 8.84266496e-01
8.07902217e-01 -6.06905580e-01 -5.76648474e-01 -7.27971047e-02
4.82204169e-01 3.05417299e-01 4.92609441e-02 -5.11851013e-01
1.61999203e-02 -6.14509881e-01 7.15989992e-02 -4.19837236e-01
2.16171607e-01 -8.30660522e-01 1.03284635e-01 2.67599434e-01
-2.62464583e-01 4.00835901e-01 1.67750791e-01 1.01504290e+00
-3.66430014e-01 -3.78340214e-01 9.35345948e-01 2.43351862e-01
6.58747032e-02 4.57942516e-01 -7.65363425e-02 3.43876809e-01
1.14513755e+00 9.26311314e-03 -2.01837182e-01 -5.98498523e-01
-7.75992811e-01 7.07079947e-01 2.32801706e-01 3.24417949e-01
5.26263773e-01 -1.27796590e+00 -6.83274269e-01 -1.03026398e-01
-1.22559778e-01 1.27187267e-01 -8.93705264e-02 1.18891108e+00
-2.08647594e-01 3.60399961e-01 4.41542178e-01 -2.76482344e-01
-6.09687865e-01 7.17849493e-01 5.80795348e-01 -3.51679832e-01
-2.20660120e-01 7.74238288e-01 4.04096097e-01 -3.00186008e-01
3.37117553e-01 -4.82471198e-01 7.79949687e-03 9.76943150e-02
4.13840920e-01 7.99300015e-01 -1.55933108e-02 -3.35684419e-01
-3.07692647e-01 5.32250106e-01 6.66850209e-02 -1.72861084e-01
1.41757679e+00 -2.82911032e-01 1.18560724e-01 6.60322726e-01
9.88494694e-01 5.63542284e-02 -1.37602937e+00 -2.75553942e-01
4.41218495e-01 -6.44346893e-01 8.96677300e-02 -5.84228575e-01
-1.01929116e+00 6.33143246e-01 2.86621302e-01 2.64197141e-01
1.20939493e+00 -4.66291636e-01 4.82377708e-01 3.88156354e-01
3.49869668e-01 -1.37987125e+00 -2.25183532e-01 2.46584058e-01
1.15151882e+00 -1.22372282e+00 1.67255297e-01 -4.36739713e-01
-8.26096773e-01 1.00955582e+00 2.77980208e-01 -3.74940038e-01
1.03382134e+00 2.95181602e-01 -2.42724732e-01 1.55103639e-01
-7.88990378e-01 1.34875506e-01 5.78909039e-01 3.13573271e-01
-5.65613583e-02 5.97319342e-02 -5.45189440e-01 1.20583320e+00
-2.65174598e-01 -5.29861748e-01 6.38435364e-01 1.04135180e+00
-2.78790593e-01 -7.53884673e-01 -5.46427846e-01 3.93702447e-01
-9.94483054e-01 1.92687381e-03 9.45296437e-02 8.72623384e-01
-4.81238931e-01 9.41854477e-01 -1.60030723e-01 1.08002022e-01
6.02006912e-01 -1.58535823e-01 1.94568977e-01 -5.84481716e-01
-3.81003857e-01 1.94029957e-01 1.46084219e-01 -4.20769304e-01
-1.03096448e-01 -9.10164535e-01 -8.52887273e-01 9.36640054e-02
-7.88447320e-01 3.64305288e-01 7.27264345e-01 9.52853501e-01
3.25266838e-01 3.31992328e-01 1.21654582e+00 -4.22080666e-01
-1.82093859e+00 -6.83162272e-01 -1.13688743e+00 6.72838697e-03
4.53471124e-01 -1.19283772e+00 -9.47442770e-01 -5.24631798e-01]
|
[5.34758186340332, 3.784514904022217]
|
6941b6a6-e82b-4b07-aae2-24527b141fd1
|
sigtyp-2021-shared-task-robust-spoken
|
2106.03895
| null |
https://arxiv.org/abs/2106.03895v1
|
https://arxiv.org/pdf/2106.03895v1.pdf
|
SIGTYP 2021 Shared Task: Robust Spoken Language Identification
|
While language identification is a fundamental speech and language processing task, for many languages and language families it remains a challenging task. For many low-resource and endangered languages this is in part due to resource availability: where larger datasets exist, they may be single-speaker or have different domains than desired application scenarios, demanding a need for domain and speaker-invariant language identification systems. This year's shared task on robust spoken language identification sought to investigate just this scenario: systems were to be trained on largely single-speaker speech from one domain, but evaluated on data in other domains recorded from speakers under different recording circumstances, mimicking realistic low-resource scenarios. We see that domain and speaker mismatch proves very challenging for current methods which can perform above 95% accuracy in-domain, which domain adaptation can address to some degree, but that these conditions merit further investigation to make spoken language identification accessible in many scenarios.
|
['Ekaterina Vylomova', 'Ryan Cotterell', 'Ritesh Kumar', 'Edoardo Ponti', 'Oleg Serikov', 'Elena Klyachko', 'Sabrina J. Mielke', 'Badr M. Abdullah', 'Elizabeth Salesky']
|
2021-06-07
| null |
https://aclanthology.org/2021.sigtyp-1.11
|
https://aclanthology.org/2021.sigtyp-1.11.pdf
|
naacl-sigtyp-2021-6
|
['spoken-language-identification']
|
['speech']
|
[ 3.96038145e-01 -1.93886802e-01 -1.10120848e-01 -6.56222641e-01
-1.17841041e+00 -9.71102297e-01 6.59758985e-01 -2.40559459e-01
-5.65983117e-01 7.53204823e-01 3.20651084e-01 -6.13858163e-01
4.20054607e-02 8.98799151e-02 -9.83146057e-02 -5.66305041e-01
1.02256626e-01 8.67618978e-01 1.58062562e-01 -2.62193054e-01
3.11048375e-03 6.08643174e-01 -1.53548574e+00 -5.18593490e-02
5.89448273e-01 3.02221417e-01 1.07910596e-01 9.17537987e-01
-3.98021758e-01 3.02839905e-01 -8.19512129e-01 -2.13118047e-01
1.89048141e-01 -4.28258926e-01 -9.29497540e-01 9.06651095e-02
4.09278154e-01 -3.65149587e-01 7.54853524e-03 9.68549073e-01
9.79748487e-01 -5.76242048e-04 6.80312514e-01 -1.02065992e+00
-2.92804629e-01 5.69232345e-01 -4.41962741e-02 4.21836823e-01
7.62923300e-01 2.95152575e-01 5.12358308e-01 -6.85094297e-01
2.83155382e-01 1.53863275e+00 7.02881336e-01 7.06013858e-01
-1.43146014e+00 -8.88575971e-01 9.48215351e-02 -3.83668005e-01
-1.54674590e+00 -1.35286808e+00 3.88268143e-01 -5.00048637e-01
1.13014066e+00 6.59119040e-02 1.28121480e-01 1.14847147e+00
-4.93844956e-01 3.49784404e-01 9.61116374e-01 -4.38339144e-01
7.24813193e-02 6.79691732e-01 -1.12170540e-01 4.25223894e-02
8.76273289e-02 1.02761149e-01 -6.63159966e-01 -3.92951965e-01
4.11743462e-01 -7.29570329e-01 -3.12814415e-01 -6.46350905e-02
-1.37438142e+00 6.97078407e-01 -3.89377296e-01 6.25171840e-01
-2.15841532e-01 -7.32841015e-01 6.20080948e-01 6.58031762e-01
3.89701039e-01 2.88546830e-01 -6.03195190e-01 -5.85110247e-01
-1.15391362e+00 2.52229214e-01 1.31112516e+00 8.44588816e-01
4.12453741e-01 4.86541033e-01 3.14810783e-01 1.41901302e+00
9.85558927e-02 7.14874625e-01 6.90803826e-01 -5.37177563e-01
3.72364908e-01 1.85524181e-01 5.28133698e-02 -5.62454462e-01
-4.60691482e-01 -1.18766569e-01 -6.34550452e-01 7.52346814e-02
9.92044389e-01 -1.20596588e-01 -7.28097618e-01 1.93762243e+00
2.68119425e-01 -3.48377451e-02 3.57589513e-01 9.15597737e-01
6.98628306e-01 4.65088218e-01 3.10408831e-01 -3.59986305e-01
1.37183630e+00 -1.10679761e-01 -5.15053570e-01 -7.11168826e-01
5.23420870e-01 -1.14400494e+00 1.05923879e+00 2.44900987e-01
-9.69503999e-01 -3.67405027e-01 -7.86255658e-01 2.80160546e-01
-2.54811764e-01 -1.54087499e-01 2.02461675e-01 1.25348771e+00
-1.20527041e+00 -1.21541381e-01 -4.79266047e-01 -1.06312299e+00
-1.39057279e-01 5.15426874e-01 -6.82712436e-01 -6.85812831e-02
-1.35259247e+00 1.22474098e+00 2.46867776e-01 -1.10499471e-01
-6.77517772e-01 -5.64324975e-01 -9.03991878e-01 -2.08317652e-01
3.57204899e-02 -2.14939207e-01 1.41102564e+00 -1.10462415e+00
-1.44656384e+00 1.46218884e+00 -4.86582786e-01 -3.92383188e-01
5.17298639e-01 2.97964990e-01 -1.04731536e+00 -1.24753438e-01
2.28618816e-01 3.93561095e-01 8.67521524e-01 -1.10679007e+00
-5.82887411e-01 -4.94584203e-01 -4.50950176e-01 4.26335663e-01
-2.89627165e-01 8.16190958e-01 -1.78040326e-01 -3.73617917e-01
4.27862406e-02 -8.13302279e-01 2.44174689e-01 -2.97104537e-01
-7.00095221e-02 -9.55437198e-02 8.83136511e-01 -1.08546531e+00
9.36141968e-01 -2.32067943e+00 -3.25952262e-01 -1.62574276e-02
-3.89857054e-01 7.40965307e-01 -1.54514328e-01 4.46922183e-01
-9.29927304e-02 2.34372377e-01 -3.08813453e-01 -3.74519736e-01
1.89569630e-02 2.16520756e-01 -2.41904214e-01 5.97908258e-01
3.54012251e-01 1.64180741e-01 -6.38611674e-01 -4.35102105e-01
1.72537312e-01 5.02440095e-01 -1.98884472e-01 1.77355409e-01
2.21506819e-01 6.30526662e-01 -6.22775368e-02 7.33616710e-01
6.19054079e-01 4.38027799e-01 2.18697011e-01 4.21092540e-01
-2.42538288e-01 6.10986650e-01 -1.48258102e+00 1.31266773e+00
-4.05071944e-01 7.07061112e-01 9.03557539e-01 -1.05166030e+00
1.03119612e+00 5.67076862e-01 1.65221289e-01 -5.14511943e-01
1.20424516e-02 5.90465784e-01 5.39135277e-01 -4.26382929e-01
4.39355642e-01 -5.71528792e-01 -1.15481280e-01 5.44256330e-01
-8.01290050e-02 -5.31307995e-01 -1.27674252e-01 -6.01960272e-02
8.71233165e-01 -3.60517621e-01 2.51627356e-01 -5.32995284e-01
6.52144730e-01 1.06553257e-01 5.19257784e-01 7.63793290e-01
-7.90117681e-01 5.71374178e-01 -5.02175326e-03 3.83319741e-04
-9.32470918e-01 -1.09093034e+00 -5.28711259e-01 1.39709210e+00
-3.40421170e-01 3.20496142e-01 -6.36029303e-01 -1.08751081e-01
-6.32361248e-02 7.95732915e-01 3.38891774e-01 -1.05150789e-01
-3.75252068e-01 -5.23254216e-01 1.07025635e+00 1.21438138e-01
1.70522660e-01 -1.00005054e+00 2.74684876e-02 4.60581422e-01
-1.29168583e-02 -1.30168772e+00 -6.47824049e-01 2.74466842e-01
-1.60081193e-01 -7.85008311e-01 -1.10158849e+00 -1.22542214e+00
2.02537581e-01 1.24075718e-01 1.28172839e+00 -2.98284233e-01
-1.59130469e-01 7.25381136e-01 -1.46779194e-01 -6.32761657e-01
-1.15393448e+00 2.65526563e-01 7.56191075e-01 6.86443374e-02
8.63225877e-01 -3.82407695e-01 -1.65579900e-01 6.37750030e-01
-6.26281261e-01 -6.04906738e-01 2.96355069e-01 7.66861379e-01
-1.21803641e-01 9.53527912e-02 1.06703699e+00 -4.74435031e-01
8.47242951e-01 -3.62755954e-01 -5.58435380e-01 2.81024784e-01
-1.69039235e-01 -2.86712050e-01 4.62320477e-01 -7.20457673e-01
-9.68396187e-01 2.77136266e-01 -3.74918044e-01 -2.34083105e-02
-6.30571961e-01 1.62403613e-01 -4.35638249e-01 -8.58541764e-03
7.78654575e-01 5.91226816e-01 4.19470906e-01 -4.02831852e-01
-1.73290446e-01 1.45740247e+00 6.57403469e-01 -5.78084469e-01
8.22927654e-01 8.40259045e-02 -7.31922150e-01 -1.60388184e+00
-7.95185119e-02 -7.29898751e-01 -7.37714589e-01 1.09715819e-01
5.87209046e-01 -1.24503684e+00 -4.80828404e-01 6.60530210e-01
-1.04976380e+00 -3.75007659e-01 1.68950379e-01 6.05565608e-01
-2.24672675e-01 3.21949929e-01 9.02191643e-03 -1.29787266e+00
-1.41283795e-01 -1.24311292e+00 1.08858478e+00 7.62685901e-03
-7.57018209e-01 -1.17541099e+00 3.49002443e-02 3.02789837e-01
7.27061272e-01 -3.63377899e-01 7.19376504e-01 -1.35610068e+00
8.13587382e-02 -3.58873367e-01 -1.01500610e-02 5.98311901e-01
5.10118067e-01 -9.52113122e-02 -1.22806609e+00 -5.93539059e-01
-7.64137879e-02 -6.01381481e-01 2.32672274e-01 1.61739796e-01
2.98420817e-01 3.14830728e-02 -1.70959696e-01 1.32624850e-01
8.30304921e-01 4.35720980e-01 2.87339203e-02 -7.59623870e-02
3.13406736e-01 1.02911747e+00 1.71515688e-01 2.51522094e-01
5.77620268e-01 7.48000681e-01 -4.43373382e-01 -9.85865444e-02
-3.69136706e-02 8.24781507e-03 6.33656263e-01 7.84425020e-01
7.03722656e-01 -1.61376476e-01 -1.45187366e+00 9.85064924e-01
-1.24647248e+00 -1.16590941e+00 1.71781942e-01 2.62267876e+00
9.06004250e-01 1.95818797e-01 7.36186147e-01 2.98944622e-01
1.01099753e+00 -4.02510241e-02 -7.26976395e-01 -7.37618625e-01
-4.97072905e-01 -1.39100596e-01 3.08607727e-01 6.27467036e-01
-8.63012433e-01 8.74675453e-01 7.12535334e+00 4.76291955e-01
-1.45185459e+00 -7.97888562e-02 5.26176929e-01 7.15387762e-02
-7.29516000e-02 -1.70660660e-01 -1.02674019e+00 4.69066113e-01
1.29583609e+00 -4.99079973e-01 5.70629120e-01 6.67816520e-01
4.88057941e-01 -1.07170604e-01 -1.26552248e+00 1.15656531e+00
2.17484757e-01 -5.16980231e-01 -3.21218699e-01 4.09632968e-03
1.48407266e-01 2.31328458e-01 1.86533090e-02 4.39189315e-01
4.89689589e-01 -1.24429119e+00 4.14639443e-01 -2.37734273e-01
1.19785154e+00 -6.24552727e-01 4.15081948e-01 8.06797147e-01
-9.06126678e-01 1.66110277e-01 -2.24560365e-01 -4.70596626e-02
2.55379587e-01 2.32278123e-01 -1.41762710e+00 -8.90436843e-02
5.51134646e-01 1.92911863e-01 -2.46345520e-01 8.59574139e-01
3.65882248e-01 9.32751596e-01 -7.16748536e-01 -8.98195729e-02
3.47753912e-02 2.00996071e-01 6.20269239e-01 1.45185626e+00
3.65735650e-01 -7.49663562e-02 2.58979797e-01 2.45805666e-01
3.06799710e-01 1.32853374e-01 -1.13385463e+00 -2.33860284e-01
8.05900872e-01 1.00526440e+00 -4.19813216e-01 -1.23152006e-02
-5.85658967e-01 8.99950206e-01 3.48643810e-02 3.01830679e-01
-2.31322005e-01 -1.82783201e-01 1.15249836e+00 1.87693000e-01
-2.17168093e-01 -2.83656061e-01 -1.68367743e-01 -1.18097544e+00
-1.24975353e-01 -1.34655130e+00 3.99086952e-01 -2.57345408e-01
-1.50515234e+00 6.46825016e-01 -3.12953559e-03 -1.09403706e+00
-1.00693250e+00 -6.88953102e-01 -4.29385841e-01 1.39510953e+00
-1.28772593e+00 -1.14701188e+00 2.07834154e-01 7.71096289e-01
5.83909929e-01 -5.99435925e-01 1.13735044e+00 4.22061622e-01
-3.42289567e-01 9.03960407e-01 1.96745381e-01 1.50533661e-01
1.19159186e+00 -9.37441289e-01 4.45264012e-01 7.08875299e-01
1.33053750e-01 6.82206750e-01 7.91520894e-01 -2.91241616e-01
-1.39924932e+00 -7.33457744e-01 1.22124958e+00 -6.19955242e-01
6.17111802e-01 -7.85116792e-01 -1.19206429e+00 5.96586049e-01
9.31781679e-02 -4.07453269e-01 9.06706452e-01 2.94383079e-01
-3.65734577e-01 7.23151043e-02 -1.38094485e+00 4.96395767e-01
9.19228792e-01 -1.08334720e+00 -6.08034968e-01 3.22380811e-01
3.28844011e-01 -1.57671750e-01 -7.19557106e-01 8.17749798e-02
4.42688823e-01 -7.14548349e-01 8.27281654e-01 -4.53880399e-01
-4.44711149e-01 -3.77841562e-01 -2.87206143e-01 -1.43813825e+00
-1.72953382e-02 -7.89694011e-01 7.90487230e-01 1.90977752e+00
5.51364601e-01 -1.06499887e+00 5.13764918e-01 1.09401381e+00
7.41910189e-02 2.97586381e-01 -1.38204694e+00 -1.01173878e+00
4.71364081e-01 -6.22225881e-01 6.49149001e-01 1.03763270e+00
-1.48798339e-03 5.08012772e-01 -2.83555418e-01 3.67732555e-01
3.55355650e-01 -4.48264629e-01 8.96830738e-01 -1.09240103e+00
-6.99111298e-02 -5.56948483e-01 -4.21455055e-01 -6.14022732e-01
7.12550282e-01 -7.18792498e-01 3.04487616e-01 -9.24572408e-01
-2.13561177e-01 -5.18426061e-01 8.70820358e-02 3.36799175e-01
-1.06176876e-01 -4.21709940e-02 -5.27952909e-02 1.35879412e-01
8.98825526e-02 1.51912794e-01 4.45428312e-01 -2.89668530e-01
-3.22631091e-01 2.97994882e-01 -8.86808872e-01 5.64436257e-01
5.63731134e-01 -1.27112925e-01 -5.25031388e-01 -4.51317221e-01
-5.12147486e-01 1.08015902e-01 1.31119505e-01 -9.52825844e-01
3.77747923e-01 -1.40236124e-01 1.50525004e-01 -2.02399120e-01
4.14025277e-01 -8.24594319e-01 9.99096856e-02 2.49093518e-01
-1.94521815e-01 -8.57363343e-02 6.21570885e-01 8.12786222e-02
-3.55401129e-01 -1.45101801e-01 1.07497096e+00 -5.96075840e-02
-9.26846683e-01 2.94941179e-02 -6.11512661e-01 4.32143122e-01
7.79987931e-01 -4.30339873e-01 -7.03166053e-02 -7.27999806e-01
-5.25138140e-01 1.77459285e-01 6.86124861e-01 6.09416962e-01
2.78063238e-01 -9.43394244e-01 -1.11250377e+00 6.34065211e-01
3.50726485e-01 -1.31863311e-01 1.40015841e-01 4.81239617e-01
-1.61009714e-01 5.01951098e-01 5.76840267e-02 -7.04891682e-01
-1.52031338e+00 2.44128987e-01 4.85750765e-01 3.56415451e-01
-2.22097844e-01 9.12177086e-01 1.64283037e-01 -8.65772009e-01
4.05666262e-01 2.39676267e-01 5.74521674e-03 1.21826142e-01
7.43378699e-01 -1.37881964e-01 1.23161025e-01 -1.27467668e+00
-8.32720160e-01 2.71517247e-01 -1.67029038e-01 -4.51018035e-01
8.11055243e-01 -2.85689712e-01 3.11826825e-01 5.10658026e-01
1.09768498e+00 4.27164175e-02 -9.38115001e-01 -3.90134037e-01
1.18430674e-01 -3.30954671e-01 -1.25972345e-01 -8.38855505e-01
-2.71604657e-01 8.54663730e-01 6.81354284e-01 3.60031098e-01
9.48480129e-01 8.19864422e-02 4.23782051e-01 2.60034114e-01
3.20014089e-01 -1.06831145e+00 -5.60839295e-01 6.21510804e-01
9.28501666e-01 -1.52015281e+00 -3.09311002e-01 -1.67324066e-01
-8.15707624e-01 7.19714820e-01 3.47758591e-01 5.25758028e-01
7.19306469e-01 3.26125026e-01 6.35907531e-01 3.04041743e-01
-4.60549980e-01 -3.11802328e-01 6.45808429e-02 1.08516049e+00
7.29514480e-01 6.20551966e-02 1.94785316e-02 3.54857385e-01
-4.93759364e-01 -2.57466584e-01 1.88386664e-01 7.69470632e-01
-2.88334608e-01 -1.21910882e+00 -8.13884437e-01 3.56729180e-01
-4.70388025e-01 -1.23217180e-01 -7.21273303e-01 7.71485865e-01
-2.48134151e-01 1.36949646e+00 2.86952376e-01 -8.56642425e-02
4.57048684e-01 5.92720032e-01 -1.21364174e-02 -8.96900952e-01
-6.57820582e-01 7.08000502e-03 4.66761023e-01 1.56706944e-01
-2.20140919e-01 -1.21241820e+00 -9.69956219e-01 -4.34027880e-01
-1.65331557e-01 2.67877392e-02 9.06621456e-01 8.52616131e-01
2.53965110e-01 -7.96869397e-02 5.19940615e-01 -9.10482585e-01
-9.33526397e-01 -9.94481504e-01 -8.13853621e-01 3.94791365e-01
6.30341887e-01 -3.56246561e-01 -6.12628043e-01 7.33142942e-02]
|
[14.207488059997559, 6.6061930656433105]
|
d3670137-74be-49db-820c-d57a4ea91dcf
|
dtw-at-qur-an-qa-2022-utilising-transfer
|
2205.06025
| null |
https://arxiv.org/abs/2205.06025v1
|
https://arxiv.org/pdf/2205.06025v1.pdf
|
DTW at Qur'an QA 2022: Utilising Transfer Learning with Transformers for Question Answering in a Low-resource Domain
|
The task of machine reading comprehension (MRC) is a useful benchmark to evaluate the natural language understanding of machines. It has gained popularity in the natural language processing (NLP) field mainly due to the large number of datasets released for many languages. However, the research in MRC has been understudied in several domains, including religious texts. The goal of the Qur'an QA 2022 shared task is to fill this gap by producing state-of-the-art question answering and reading comprehension research on Qur'an. This paper describes the DTW entry to the Quran QA 2022 shared task. Our methodology uses transfer learning to take advantage of available Arabic MRC data. We further improve the results using various ensemble learning strategies. Our approach provided a partial Reciprocal Rank (pRR) score of 0.49 on the test set, proving its strong performance on the task.
|
['Ruslan Mitkov', 'Wajdi Zaghouani', 'Tharindu Ranasinghe', 'Damith Premasiri']
|
2022-05-12
| null | null | null | null |
['machine-reading-comprehension']
|
['natural-language-processing']
|
[ 4.48334426e-01 4.02284920e-01 1.57501310e-01 -3.46688747e-01
-1.30928922e+00 -6.55555189e-01 7.32410192e-01 3.91438663e-01
-4.67572957e-01 9.62022305e-01 6.51187539e-01 -6.35913908e-01
-1.04958758e-01 -8.61895740e-01 -5.50621748e-01 -4.44011480e-01
1.10486433e-01 7.27837801e-01 2.65803516e-01 -9.29619849e-01
7.58417487e-01 -1.88830674e-01 -1.38680220e+00 8.15669239e-01
1.46242416e+00 8.49610388e-01 3.00949723e-01 7.56561637e-01
4.01027091e-02 1.43470550e+00 -7.16151536e-01 -5.61011672e-01
-2.44881317e-01 -9.48461592e-01 -1.64876592e+00 -6.84143424e-01
4.36247379e-01 -6.28248751e-02 1.01198331e-01 6.59118593e-01
5.22355676e-01 1.66531369e-01 7.03812242e-01 -8.32780540e-01
-9.32127535e-01 6.61527395e-01 -8.19782540e-02 3.23293686e-01
9.15213466e-01 -3.34570557e-01 1.48409271e+00 -8.65782738e-01
5.11072814e-01 1.09471810e+00 2.27039754e-01 4.56626981e-01
-6.58466280e-01 -2.60007679e-01 -6.02570474e-01 9.42665756e-01
-9.51888502e-01 -2.91162610e-01 5.84753990e-01 -2.69937724e-01
1.01008654e+00 1.89005598e-01 4.84973416e-02 8.48656714e-01
2.83806175e-01 9.29534912e-01 1.56429183e+00 -9.30735350e-01
2.10295618e-01 -2.68366963e-01 2.74365067e-01 5.70874393e-01
-3.93431485e-01 -3.90650511e-01 -8.11020911e-01 1.69571355e-01
1.33480832e-01 -8.77181172e-01 -3.47486079e-01 2.62116104e-01
-1.16226912e+00 1.24172139e+00 3.27473462e-01 5.12388885e-01
-2.78011650e-01 -2.10916117e-01 3.50180686e-01 9.80925500e-01
5.60751379e-01 8.01198483e-01 -8.79533410e-01 -5.54621100e-01
-6.24546707e-01 3.52985680e-01 1.15726161e+00 2.07190827e-01
4.25910145e-01 -5.42404115e-01 -1.95240438e-01 1.14958668e+00
2.33086929e-01 6.00153923e-01 5.11482418e-01 -8.92195463e-01
8.87929559e-01 6.66071594e-01 1.47680603e-02 -9.30957437e-01
-3.29631478e-01 -5.22933640e-02 -6.67811155e-01 -1.64545715e-01
8.53329480e-01 -1.71250761e-01 -5.64067960e-01 1.54979014e+00
3.27533334e-02 -4.51573730e-01 4.93617475e-01 6.70849025e-01
1.10715115e+00 9.92788672e-01 7.58114457e-03 -2.97550783e-02
1.58177757e+00 -9.73491311e-01 -6.80852592e-01 -1.31740302e-01
8.50060523e-01 -8.37384343e-01 1.10885131e+00 5.57919919e-01
-8.91301811e-01 -5.28696418e-01 -9.49312329e-01 -4.30342853e-01
-4.01216030e-01 5.90000190e-02 3.42040926e-01 7.18710601e-01
-8.52095723e-01 2.61546999e-01 -4.45822984e-01 -1.62710249e-01
2.95905143e-01 7.14232326e-02 -3.62730652e-01 -5.30654371e-01
-1.69193077e+00 1.55032551e+00 5.31372607e-01 -5.48193790e-02
-6.94306314e-01 -5.07186890e-01 -8.51887763e-01 -3.03721875e-02
4.54691648e-01 -3.80146444e-01 1.43247950e+00 -1.05129075e+00
-1.68820274e+00 1.06815696e+00 -9.83441025e-02 -6.36658490e-01
3.61215621e-01 -5.36008596e-01 -3.10096473e-01 2.85313666e-01
7.68383965e-02 3.73367131e-01 4.57702428e-01 -8.62244308e-01
-5.69736481e-01 -4.73760903e-01 2.59859383e-01 4.40461785e-01
1.32697657e-01 2.43083358e-01 2.55075186e-01 -3.87639821e-01
-1.43853799e-01 -5.96454978e-01 3.02626938e-01 -6.51001751e-01
1.32082447e-01 -9.20865476e-01 4.15846080e-01 -1.37581193e+00
1.22496319e+00 -1.70662534e+00 4.40620244e-01 -1.86528906e-01
-5.08697070e-02 2.81556368e-01 -3.41534287e-01 9.24017191e-01
8.90569240e-02 -1.28701165e-01 -5.59350669e-01 1.48168504e-01
-1.53678551e-01 2.29370520e-01 -4.28457737e-01 2.13804409e-01
3.07580560e-01 1.12398839e+00 -1.02539539e+00 -3.15744102e-01
-8.83614644e-02 -7.58976163e-03 -2.39927799e-01 3.73051107e-01
-5.08355498e-01 4.47335809e-01 -4.58430707e-01 5.05859613e-01
2.43534878e-01 -1.66674107e-01 -4.45369817e-02 2.32079610e-01
-3.21566202e-02 7.72247970e-01 -5.16498566e-01 1.78586531e+00
-4.83115613e-01 6.74964249e-01 -3.06416690e-01 -1.20967960e+00
1.02645540e+00 3.70048255e-01 1.47091508e-01 -1.22810030e+00
9.38883275e-02 4.65991169e-01 4.70866829e-01 -7.72276819e-01
5.96095681e-01 -2.87512153e-01 -2.48209238e-01 6.76195681e-01
2.28012800e-02 -4.36048776e-01 1.59053266e-01 2.30360165e-01
1.05069721e+00 2.89980710e-01 6.75548494e-01 -5.60983121e-01
9.66784596e-01 2.42429644e-01 -4.60217483e-02 5.75072289e-01
-1.04347102e-01 5.13548732e-01 4.91540074e-01 -1.63071722e-01
-8.39952946e-01 -8.09107959e-01 -2.22499058e-01 1.46097136e+00
-1.24954768e-01 -1.88468084e-01 -8.35766613e-01 -7.16582239e-01
-4.37561035e-01 1.15885973e+00 -7.72243917e-01 1.33593008e-01
-8.02692354e-01 -7.53387630e-01 7.93689132e-01 1.80681944e-01
7.46163070e-01 -1.44751346e+00 -7.66632438e-01 3.42481375e-01
-8.82049739e-01 -8.24221492e-01 1.53118297e-01 -1.51320189e-01
-5.85836053e-01 -1.43433857e+00 -8.08399796e-01 -9.65036154e-01
-1.76429171e-02 -1.87395394e-01 1.52004230e+00 1.26721948e-01
2.57599056e-01 3.27655971e-01 -1.16975534e+00 -6.35742605e-01
-9.74866748e-01 4.70511198e-01 -4.68726188e-01 -1.58792511e-01
5.45932412e-01 1.63457189e-02 -1.55632704e-01 2.69406289e-01
-8.57182324e-01 2.20055833e-01 3.70416433e-01 9.86118376e-01
-8.71516317e-02 -2.56578654e-01 1.26864624e+00 -6.71856761e-01
1.01127172e+00 -5.82464159e-01 -2.50449330e-01 5.88634133e-01
-2.87026048e-01 8.69916677e-02 4.01722938e-01 1.74666256e-01
-1.21074200e+00 -6.39698565e-01 -5.30482054e-01 8.70130897e-01
-1.31827369e-01 1.05580747e+00 -1.06129020e-01 2.47893885e-01
8.68025064e-01 2.25513607e-01 1.06117278e-01 -4.37319398e-01
3.53395730e-01 1.07648301e+00 4.52758521e-01 -7.03987598e-01
3.73240650e-01 -1.53116718e-01 -1.12378843e-01 -1.08254862e+00
-1.37558591e+00 -4.04032230e-01 -6.41727984e-01 -1.23506971e-01
1.03870857e+00 -8.04295361e-01 -5.75066209e-01 5.54279983e-01
-1.18312621e+00 -4.14513320e-01 8.25389400e-02 2.60068387e-01
-6.07837856e-01 4.10873324e-01 -5.04993260e-01 -7.34631181e-01
-5.09677231e-01 -8.68149459e-01 7.47626066e-01 1.35119602e-01
-3.61276358e-01 -1.07218182e+00 5.92452347e-01 1.18188989e+00
5.78218699e-01 1.84481025e-01 1.32657075e+00 -9.98545647e-01
-1.54793441e-01 1.95172772e-01 -1.30375996e-01 6.00296080e-01
8.82262141e-02 -5.09475172e-01 -9.30603445e-01 -1.13773592e-01
2.02754840e-01 -9.70291555e-01 8.38799715e-01 -8.80185217e-02
6.82388544e-01 -6.56230748e-02 5.22192359e-01 -4.92868930e-01
1.21386147e+00 1.16255417e-01 1.03415442e+00 5.92616439e-01
2.86346763e-01 9.95602190e-01 8.60985339e-01 2.90873237e-02
7.51640558e-01 4.94875073e-01 3.93144518e-01 3.39825392e-01
-1.11391090e-01 -1.96724147e-01 5.27086973e-01 1.18370914e+00
-3.24454546e-01 -3.25042099e-01 -1.51171219e+00 7.54023731e-01
-1.72884202e+00 -8.35953832e-01 -3.84999216e-01 1.85708368e+00
1.07099795e+00 -2.04262123e-01 -1.83015078e-01 3.11488003e-01
1.68963775e-01 7.96861649e-02 -8.32977612e-03 -6.40772104e-01
-3.47445190e-01 6.28087163e-01 -2.57422328e-01 7.16925561e-01
-8.68532538e-01 9.99055326e-01 5.75910044e+00 9.07341838e-01
-4.45213348e-01 1.28291562e-01 5.69658339e-01 8.29900086e-01
-2.35836163e-01 -1.15460418e-01 -3.24699193e-01 1.19563028e-01
1.12599587e+00 -3.98740880e-02 3.13673049e-01 2.78044373e-01
-6.79140538e-03 -6.29086792e-01 -1.06206751e+00 6.01055384e-01
6.55176222e-01 -1.13125575e+00 1.52294055e-01 -1.94849968e-01
7.18761861e-01 4.30714667e-01 -2.60144830e-01 6.48146629e-01
1.74934819e-01 -1.63125241e+00 3.12311143e-01 3.93714637e-01
5.71428657e-01 -7.25154459e-01 1.31622875e+00 8.04198802e-01
-4.60160613e-01 7.22569507e-03 -2.67393291e-01 -6.05662525e-01
2.01276541e-01 2.04090580e-01 -9.54071879e-01 8.57287288e-01
4.89904016e-01 4.08307344e-01 -9.49567556e-01 6.31632388e-01
-6.31674170e-01 1.14235735e+00 -1.07713424e-01 -4.50344115e-01
3.20872277e-01 -2.93717533e-01 2.51882017e-01 1.00044680e+00
-3.63668352e-02 3.04628372e-01 -8.36551264e-02 2.81091809e-01
-2.72448003e-01 6.97160244e-01 -2.59899914e-01 4.02889811e-02
7.65772387e-02 7.70134807e-01 -2.08637536e-01 -2.78406680e-01
-3.44041944e-01 8.70067000e-01 4.34034020e-01 -9.34739634e-02
-4.88544434e-01 -4.38544452e-01 -1.30374447e-01 -3.08370441e-01
1.37108369e-02 -2.44878933e-01 -2.38557830e-01 -1.24200726e+00
-8.46494064e-02 -1.47292078e+00 5.65244198e-01 -7.96592712e-01
-1.49565768e+00 5.42491853e-01 8.52008983e-02 -6.80641234e-01
-3.56487364e-01 -8.48430872e-01 -3.39096308e-01 9.51040626e-01
-1.62881601e+00 -1.19410217e+00 -2.08321080e-01 4.52036530e-01
7.49979258e-01 -3.73301625e-01 1.23255706e+00 -1.25598740e-02
9.19024646e-02 2.97176659e-01 7.57906139e-02 3.37957531e-01
8.24072242e-01 -1.44721389e+00 2.02907294e-01 6.25860751e-01
3.04911852e-01 2.70956248e-01 6.48703337e-01 -3.17617148e-01
-1.21820521e+00 -3.93210769e-01 1.48146951e+00 -9.80975866e-01
8.32336247e-01 4.63017747e-02 -1.19005239e+00 5.35655856e-01
8.54983628e-01 -8.99003088e-01 1.05150032e+00 2.38870755e-01
-4.24454331e-01 2.31265262e-01 -1.03748369e+00 4.13426578e-01
3.60864431e-01 -6.26972020e-01 -1.65800369e+00 4.07782286e-01
4.81235802e-01 -2.74018258e-01 -1.13870990e+00 3.78444403e-01
4.28442895e-01 -8.65551829e-01 4.73543197e-01 -6.39958858e-01
1.03939450e+00 -2.18200088e-01 -5.33914030e-01 -1.53929889e+00
1.84705794e-01 -2.63039559e-01 8.37525725e-02 1.01754713e+00
6.59192979e-01 -5.54967046e-01 4.44201231e-01 2.38718927e-01
-7.22347647e-02 -6.17805898e-01 -1.20664406e+00 -2.01247111e-01
7.88236856e-01 -4.72597659e-01 1.86325386e-01 9.98008966e-01
2.55516142e-01 1.05997109e+00 -2.03352690e-01 -1.86161995e-01
2.52559781e-01 2.87571996e-01 6.74560487e-01 -1.24672127e+00
-3.90208364e-01 -1.67303920e-01 -3.61796431e-02 -1.20840538e+00
1.76807120e-01 -1.10209489e+00 1.25311702e-01 -1.85253751e+00
1.19566061e-01 -9.41384211e-02 -1.35461120e-02 5.75425863e-01
-3.61015022e-01 2.31068254e-01 2.06293061e-01 7.75880739e-02
-6.39163613e-01 6.57051444e-01 1.37890303e+00 -9.37546864e-02
1.40302941e-01 -2.30857342e-01 -6.25700116e-01 4.97989684e-01
1.24634218e+00 -1.90517217e-01 -2.49831393e-01 -6.16233408e-01
5.56413949e-01 2.00427622e-01 7.22960979e-02 -6.45167768e-01
-6.94135427e-02 -1.51517922e-02 4.34741862e-02 -6.22817159e-01
9.19284597e-02 -3.36780131e-01 -4.89769369e-01 5.67330003e-01
-5.51204741e-01 2.72321582e-01 -3.90468957e-03 9.08648521e-02
-5.61390579e-01 -4.15900171e-01 6.67114019e-01 -2.53400832e-01
-6.84044838e-01 -4.05459672e-01 -5.33337116e-01 6.32200360e-01
7.12404311e-01 1.41617015e-01 -5.91168225e-01 -5.99791646e-01
-3.19199979e-01 5.07447839e-01 9.15662572e-02 7.32073545e-01
5.45355201e-01 -7.60251462e-01 -1.60505474e+00 -1.90784112e-01
3.92774791e-01 -6.35371059e-02 1.07916430e-01 6.14265442e-01
-8.95236313e-01 6.59121037e-01 -3.72373164e-01 -4.19720471e-01
-1.24204659e+00 -1.48628671e-02 2.12861255e-01 -6.44101143e-01
-2.64973223e-01 6.04942918e-01 -4.02385920e-01 -7.21356392e-01
-2.67040133e-01 1.79712385e-01 -8.82750690e-01 2.55972683e-01
8.23827267e-01 7.07346797e-01 2.97938198e-01 -7.84005761e-01
-1.45676225e-01 3.80164385e-01 -1.04997875e-02 -4.09664780e-01
1.33115351e+00 -5.11455759e-02 -5.39754748e-01 5.86228311e-01
9.60227430e-01 1.83689386e-01 -3.38377148e-01 -1.50218546e-01
5.13358831e-01 -5.97582459e-02 -2.50153452e-01 -1.49097872e+00
-1.85834885e-01 9.22415853e-01 2.20652312e-01 3.52373451e-01
8.91843140e-01 5.36728874e-02 7.23835051e-01 8.71856451e-01
5.28484285e-01 -1.08626044e+00 1.27031580e-01 1.30883837e+00
1.31474495e+00 -1.44404233e+00 -9.73124206e-02 -1.89153075e-01
-7.90494621e-01 1.17255270e+00 3.32796872e-01 -4.68614064e-02
2.10844457e-01 -4.34090555e-01 4.73626494e-01 -2.30962262e-01
-6.25861049e-01 -2.38716286e-02 4.65744615e-01 4.77512866e-01
7.91142285e-01 2.15282708e-01 -8.78307879e-01 4.56208855e-01
-7.15387106e-01 -1.31631374e-01 6.66706681e-01 7.69912362e-01
-6.57900214e-01 -1.22599065e+00 -3.13256443e-01 2.86465168e-01
-7.99283862e-01 -4.13709164e-01 -6.05959594e-01 7.54474580e-01
-2.09656268e-01 1.51736534e+00 -3.53878140e-01 3.65512408e-02
1.85823277e-01 2.73094207e-01 7.20133901e-01 -6.78895652e-01
-8.42032611e-01 -6.44502163e-01 5.33363163e-01 -2.10918888e-01
-7.91343510e-01 -5.55737972e-01 -1.12369931e+00 -8.35258607e-03
-2.21158385e-01 6.19944215e-01 5.75694621e-01 1.42923629e+00
-8.13856050e-02 3.91735405e-01 3.03008705e-01 -2.54490793e-01
-7.67414868e-01 -1.51914501e+00 -1.34532377e-01 4.27965343e-01
2.22087517e-01 -2.64718443e-01 -1.82823375e-01 3.74973789e-02]
|
[11.370680809020996, 8.233490943908691]
|
efc2f144-0dad-4862-aa97-b06047b60a54
|
learning-to-attend-on-essential-terms-an
|
1808.09492
| null |
https://arxiv.org/abs/1808.09492v5
|
https://arxiv.org/pdf/1808.09492v5.pdf
|
Learning to Attend On Essential Terms: An Enhanced Retriever-Reader Model for Open-domain Question Answering
|
Open-domain question answering remains a challenging task as it requires models that are capable of understanding questions and answers, collecting useful information, and reasoning over evidence. Previous work typically formulates this task as a reading comprehension or entailment problem given evidence retrieved from search engines. However, existing techniques struggle to retrieve indirectly related evidence when no directly related evidence is provided, especially for complex questions where it is hard to parse precisely what the question asks. In this paper we propose a retriever-reader model that learns to attend on essential terms during the question answering process. We build (1) an essential term selector which first identifies the most important words in a question, then reformulates the query and searches for related evidence; and (2) an enhanced reader that distinguishes between essential terms and distracting words to predict the answer. We evaluate our model on multiple open-domain multiple-choice QA datasets, notably performing at the level of the state-of-the-art on the AI2 Reasoning Challenge (ARC) dataset.
|
['Weizhu Chen', 'Jianmo Ni', 'Chenguang Zhu', 'Julian McAuley']
|
2018-08-28
|
learning-to-attend-on-essential-terms-an-1
|
https://aclanthology.org/N19-1030
|
https://aclanthology.org/N19-1030.pdf
|
naacl-2019-6
|
['multiple-choice-qa']
|
['natural-language-processing']
|
[ 3.80529761e-01 6.26918912e-01 -1.15355834e-01 -3.39179546e-01
-1.61596966e+00 -9.07144666e-01 6.90735161e-01 8.19698274e-01
-6.12532020e-01 6.38170421e-01 6.03641093e-01 -7.32337236e-01
-5.58167815e-01 -8.07474017e-01 -7.98335671e-01 8.77785385e-02
3.50098282e-01 1.12005365e+00 7.19995618e-01 -7.06330538e-01
5.46141446e-01 -8.64432827e-02 -1.50286722e+00 5.21888673e-01
1.28107369e+00 1.09627187e+00 2.58449137e-01 8.45381379e-01
-6.18293524e-01 1.40891421e+00 -6.54869318e-01 -7.46661603e-01
-2.07062453e-01 -4.67068017e-01 -1.82846606e+00 -5.27802169e-01
7.52045274e-01 -3.32087964e-01 -7.64982253e-02 9.64116395e-01
1.92532212e-01 2.01616004e-01 7.50940263e-01 -8.35919738e-01
-1.02728701e+00 7.69245505e-01 1.13070823e-01 7.81708062e-01
8.31262290e-01 3.37644331e-02 1.56019819e+00 -8.04800630e-01
5.70571184e-01 1.25701928e+00 -4.17966880e-02 5.47826171e-01
-9.46137965e-01 -4.76286002e-02 2.18282491e-01 8.23516488e-01
-8.37851882e-01 -2.81002015e-01 5.65863252e-01 -2.28094116e-01
1.13869762e+00 4.46537077e-01 1.32043123e-01 8.92226279e-01
-1.73522547e-01 9.78704810e-01 1.06648004e+00 -7.66453803e-01
1.71204135e-01 -8.80081952e-02 9.31357205e-01 5.76891482e-01
8.01047757e-02 -2.52379924e-01 -4.54481006e-01 -2.59952307e-01
-9.04732868e-02 -4.92645472e-01 -4.62368876e-01 4.29514721e-02
-1.05620170e+00 9.01097715e-01 4.35309559e-01 2.94824898e-01
-6.29331350e-01 -4.02492508e-02 -4.63909358e-02 5.93343139e-01
5.38606010e-02 1.02116168e+00 -7.65199363e-01 -8.82692263e-02
-6.65159643e-01 7.51402199e-01 1.32515383e+00 7.70274341e-01
6.17816687e-01 -9.51866448e-01 -6.83503032e-01 9.17798817e-01
4.40454870e-01 5.72601616e-01 2.67460555e-01 -1.28392458e+00
6.94798887e-01 8.17290366e-01 2.98231870e-01 -7.58770704e-01
-1.17844872e-01 -3.59849602e-01 5.36694080e-02 -3.71587813e-01
5.98108828e-01 7.38035813e-02 -6.67002559e-01 1.77104735e+00
3.61762971e-01 -4.00348246e-01 2.62913346e-01 8.71985257e-01
1.27502418e+00 6.41382873e-01 2.06726462e-01 1.72839716e-01
2.03123856e+00 -1.27059770e+00 -9.33278561e-01 -7.65962601e-01
5.37812889e-01 -6.51927173e-01 1.32287538e+00 2.50841558e-01
-1.36896014e+00 -1.74298629e-01 -8.69439662e-01 -9.77168381e-01
-4.84726757e-01 -3.82782429e-01 6.68000756e-03 3.93775627e-02
-8.90482783e-01 1.13055699e-01 1.83662213e-02 -1.78992927e-01
2.02166840e-01 -1.13485605e-01 7.25626349e-02 -9.11658704e-01
-1.76445138e+00 1.49945426e+00 1.70097873e-01 -2.40899950e-01
-9.71684575e-01 -7.74008989e-01 -8.80284429e-01 4.57773864e-01
1.03290617e+00 -1.16848731e+00 1.87717330e+00 -4.35032159e-01
-1.02540946e+00 1.06266499e+00 -4.31689501e-01 -4.06708419e-01
-7.23428950e-02 -5.53402364e-01 -4.76473331e-01 5.86768568e-01
4.07579690e-01 6.31309628e-01 6.24896049e-01 -1.10358453e+00
-6.01392567e-01 -5.85997224e-01 8.24703872e-01 3.60679179e-01
1.96756870e-01 3.93783953e-03 -4.47131246e-01 -4.19101238e-01
9.10488442e-02 -5.29255152e-01 1.33153543e-01 -6.93503395e-02
-2.81327814e-01 -9.23250079e-01 2.80547023e-01 -9.86417592e-01
1.32039571e+00 -1.65428412e+00 2.97283143e-01 1.25233397e-01
4.52963352e-01 4.13391367e-02 -4.47144598e-01 5.50201714e-01
2.06176847e-01 -2.24924553e-03 -1.78126678e-01 2.10307226e-01
3.27160507e-01 2.91934133e-01 -5.98819554e-01 -3.93048197e-01
3.18892092e-01 1.27567041e+00 -1.14186108e+00 -6.91679120e-01
-4.00170237e-01 -1.81474939e-01 -5.02097726e-01 5.00502288e-01
-1.07204473e+00 -7.02587515e-02 -6.90388501e-01 7.19646096e-01
2.60221452e-01 -6.52652204e-01 -2.84347653e-01 -1.02870621e-01
4.38357860e-01 1.11471260e+00 -8.02036643e-01 1.72464848e+00
-4.87532914e-01 4.52388585e-01 3.60534899e-02 -7.74663508e-01
6.55286193e-01 1.45442605e-01 -2.78998762e-01 -1.21521306e+00
4.58622910e-02 4.05511141e-01 4.82615903e-02 -1.19865489e+00
4.78994727e-01 -1.72660127e-01 -9.65451002e-02 5.58812082e-01
1.37095898e-01 -3.82667094e-01 4.32671040e-01 6.81904018e-01
1.41855216e+00 -1.68278575e-01 1.25872999e-01 -3.52880895e-01
8.24742019e-01 4.51824963e-01 -3.10398251e-01 1.00815368e+00
6.89448193e-02 1.16498537e-01 4.98996705e-01 -4.01317365e-02
-4.59545940e-01 -1.15918219e+00 1.68372437e-01 1.48572791e+00
3.36220384e-01 -3.95983577e-01 -6.34529173e-01 -7.17148542e-01
-2.99985707e-02 1.38017821e+00 -5.63954532e-01 -2.61640429e-01
-5.47495425e-01 1.58514485e-01 5.95769048e-01 3.59133393e-01
3.60338479e-01 -1.09942055e+00 -6.67015910e-01 1.65807888e-01
-9.69960093e-01 -1.13198435e+00 -2.67800868e-01 1.82736501e-01
-6.00547433e-01 -1.45978653e+00 -4.69446659e-01 -8.89073074e-01
3.26273441e-01 1.52510270e-01 1.95867717e+00 6.34775400e-01
2.64932755e-02 7.68477380e-01 -4.60452765e-01 -6.07661963e-01
-3.54687989e-01 1.71996087e-01 -7.56147385e-01 -5.42751729e-01
7.53612459e-01 -9.38477293e-02 -7.42793798e-01 1.92576990e-01
-1.02850461e+00 -2.52007693e-01 6.01293445e-01 6.37004793e-01
4.70869780e-01 -3.71966153e-01 7.40280271e-01 -6.49165928e-01
1.25474000e+00 -8.35547984e-01 -3.07271272e-01 8.86916399e-01
-5.33192575e-01 5.66092372e-01 2.76490986e-01 -5.14785796e-02
-1.11879468e+00 -8.61527979e-01 -3.26386154e-01 1.48190767e-01
-1.08743630e-01 8.92462552e-01 -2.08334737e-02 3.84488225e-01
8.76582682e-01 5.74282520e-02 -2.57010698e-01 -4.57548887e-01
8.03438127e-01 4.76771355e-01 7.06097960e-01 -1.14194763e+00
7.22079754e-01 9.79747921e-02 -1.63429424e-01 -4.13734287e-01
-1.75298810e+00 -7.60977209e-01 -1.07441708e-01 -9.87619460e-02
9.71975803e-01 -5.76839685e-01 -8.05536509e-01 -2.03291103e-01
-1.17079699e+00 -1.87527537e-01 -4.75342005e-01 2.59040389e-03
-4.20838207e-01 2.29168013e-01 -5.11846900e-01 -7.18115926e-01
-4.70390290e-01 -7.78814435e-01 1.08208025e+00 2.56212324e-01
-4.90407705e-01 -8.85501623e-01 3.34266812e-01 1.20179522e+00
5.98103464e-01 -2.91796893e-01 1.69862080e+00 -1.17132437e+00
-8.50247502e-01 -7.25595355e-02 -1.80768490e-01 5.43468166e-03
-4.39161956e-01 -5.97456098e-01 -8.45387816e-01 3.12771410e-01
2.11656630e-01 -1.14300895e+00 9.03011560e-01 -7.35022575e-02
1.13431501e+00 -4.03291911e-01 -8.17506388e-02 -3.12969625e-01
9.83796358e-01 -2.02901810e-01 6.09875679e-01 4.86494392e-01
-8.89709871e-03 1.19459963e+00 7.26297021e-01 -2.17719853e-01
1.02464890e+00 2.86478370e-01 4.77083832e-01 3.92604619e-01
-1.35929644e-01 -4.93316352e-01 -1.72024071e-01 5.54447293e-01
3.36040765e-01 -4.17557091e-01 -1.07238603e+00 9.54928994e-01
-1.61879158e+00 -1.00329268e+00 -3.18736166e-01 1.68932903e+00
1.33183062e+00 5.49910553e-02 -3.57766837e-01 1.18764860e-04
1.18214317e-01 1.21547237e-01 -6.44365013e-01 -2.67074853e-01
-1.52693868e-01 7.02811062e-01 -2.92732209e-01 9.06815290e-01
-5.61907232e-01 7.48570919e-01 5.69331312e+00 7.69517004e-01
-7.28878602e-02 -2.59769857e-02 4.39408898e-01 1.73389882e-01
-8.50677729e-01 2.94439644e-01 -6.47924602e-01 -1.10514134e-01
8.12176466e-01 -1.62532628e-01 3.91568899e-01 5.34756958e-01
-5.21540582e-01 -5.51766396e-01 -1.42650008e+00 4.67353314e-01
4.11324173e-01 -1.07282507e+00 3.82591128e-01 -4.39900160e-01
3.40245217e-01 -1.16412282e-01 -1.42001599e-01 7.80261040e-01
3.07252705e-01 -1.35853112e+00 5.06800354e-01 7.91860044e-01
1.08419955e-01 -4.05842423e-01 6.53081417e-01 9.28870261e-01
-6.91161990e-01 -1.37517408e-01 -3.74273509e-02 -8.05554986e-02
2.37843260e-01 2.05994576e-01 -5.99826217e-01 4.72777694e-01
6.16333306e-01 8.61631855e-02 -9.43963349e-01 8.30318809e-01
-9.39098120e-01 6.37189567e-01 -2.67257333e-01 -4.83241886e-01
3.38852644e-01 1.30049571e-01 5.02011478e-01 7.81316102e-01
-5.02553023e-03 6.76394284e-01 -1.64473504e-01 9.69227016e-01
-4.16968375e-01 1.42473206e-02 -5.39988354e-02 -3.98136787e-02
5.70537388e-01 9.03235734e-01 -6.27372116e-02 -5.32839715e-01
-4.58768010e-01 7.66524971e-01 7.24270701e-01 2.97715396e-01
-4.24880177e-01 -3.15567970e-01 8.77150819e-02 -4.86204848e-02
2.39826053e-01 2.27285624e-01 5.31130657e-02 -1.15019917e+00
3.70773762e-01 -1.39606726e+00 1.16470218e+00 -1.37502432e+00
-1.52303958e+00 4.23013568e-01 1.72305316e-01 -4.07773942e-01
-4.18389171e-01 -6.64060175e-01 -2.63869375e-01 1.09611487e+00
-2.19409394e+00 -7.51515806e-01 -4.32698101e-01 6.87271237e-01
7.85634458e-01 4.54645246e-01 8.36288452e-01 -1.31508906e-03
7.02746287e-02 1.49426863e-01 -4.25121933e-01 -1.12839323e-02
6.61423087e-01 -1.38279033e+00 1.16486594e-01 5.61225712e-01
3.62770349e-01 8.85233164e-01 8.48002136e-01 -4.88896191e-01
-1.57164073e+00 -3.63796741e-01 1.60002375e+00 -1.09368324e+00
8.35157812e-01 5.41737601e-02 -1.37127244e+00 5.75820386e-01
7.25427151e-01 -3.67253572e-01 8.01405728e-01 2.73624688e-01
-7.59938657e-01 2.10968167e-01 -9.47718084e-01 6.58390403e-01
8.07905495e-01 -7.51008689e-01 -1.92997718e+00 5.18990397e-01
9.88497615e-01 -3.93038929e-01 -6.41324699e-01 2.28522927e-01
2.45362848e-01 -5.00201285e-01 1.04873145e+00 -1.11996806e+00
7.83797860e-01 -2.44498879e-01 -3.92572045e-01 -1.13343871e+00
7.70442327e-03 -1.96792275e-01 -5.95501184e-01 8.63787293e-01
8.96495640e-01 -1.57916248e-01 2.09520876e-01 9.34087217e-01
5.22687882e-02 -8.80355418e-01 -9.06385839e-01 -3.83089960e-01
2.39156634e-01 -3.48246962e-01 4.05630678e-01 7.15628326e-01
2.37661049e-01 1.04921651e+00 3.92727613e-01 2.36532047e-01
3.93252045e-01 3.63659322e-01 4.00523335e-01 -1.28905535e+00
-2.20727518e-01 -4.29723084e-01 4.13029790e-01 -1.69306231e+00
2.93290287e-01 -9.24002409e-01 2.07881927e-01 -2.08780861e+00
2.42482081e-01 -1.11304894e-01 -1.16599701e-01 2.33281776e-01
-7.23734558e-01 -4.53312695e-01 2.24429350e-02 1.09984959e-02
-1.07796347e+00 4.67213154e-01 1.30304158e+00 -2.97841549e-01
4.81607229e-01 -2.17658892e-01 -1.07181287e+00 6.03920400e-01
3.59512180e-01 -3.94150734e-01 -6.14381790e-01 -8.43739033e-01
1.01914036e+00 4.89711016e-01 6.54426992e-01 -5.82835734e-01
6.16118371e-01 -1.65976882e-01 -7.40462765e-02 -6.57822847e-01
3.65310848e-01 -8.51852655e-01 -8.20355535e-01 2.57051378e-01
-1.11904144e+00 2.93587804e-01 1.35280743e-01 5.92949808e-01
-2.92752057e-01 -8.58925700e-01 2.30845556e-01 -2.45071709e-01
-6.45879209e-01 -4.83035222e-02 -2.20709980e-01 1.19072831e+00
4.73849326e-01 4.76243705e-01 -8.24497938e-01 -6.80985928e-01
-5.61236084e-01 8.48894894e-01 -4.71222177e-02 5.16935170e-01
8.31160903e-01 -1.00519097e+00 -9.20889020e-01 -4.88716424e-01
6.21335745e-01 7.53141716e-02 2.00228855e-01 5.43409228e-01
-2.58956373e-01 8.16274226e-01 3.53538334e-01 -2.00289354e-01
-1.04574454e+00 7.32866466e-01 3.08500469e-01 -7.02465177e-01
-1.17170647e-01 9.92226481e-01 -1.65867925e-01 -4.54388082e-01
2.74566293e-01 -4.73143876e-01 -7.23322749e-01 7.69777521e-02
7.90485144e-01 2.04831690e-01 2.63255298e-01 -2.33562797e-01
-2.33610317e-01 4.66898799e-01 -1.52530357e-01 -4.20418441e-01
9.54765439e-01 -3.73865098e-01 -3.78384411e-01 2.89979309e-01
8.67583334e-01 -2.82763075e-02 -5.07950246e-01 -7.20950067e-01
5.17256558e-01 -2.29697585e-01 -7.96623304e-02 -1.46349049e+00
-1.37985706e-01 9.15453494e-01 8.13934859e-03 3.95878792e-01
8.80412400e-01 7.34305501e-01 1.01494896e+00 1.13870704e+00
9.74554792e-02 -8.60770464e-01 5.39167404e-01 9.11737919e-01
1.20317245e+00 -1.29336989e+00 -2.40754753e-01 -2.41705731e-01
-4.47510242e-01 1.02069938e+00 7.69706786e-01 2.82943755e-01
2.01898471e-01 -2.32910186e-01 1.37058347e-01 -9.61760819e-01
-1.20786202e+00 -4.74008799e-01 8.09499502e-01 2.04063267e-01
2.63232678e-01 -4.16077316e-01 -3.13630879e-01 7.72351027e-01
-3.34897816e-01 -2.33987778e-01 1.22128166e-01 1.22818255e+00
-8.62813294e-01 -8.03962409e-01 -3.32541674e-01 5.12294888e-01
-4.44356531e-01 -5.66503584e-01 -7.78112888e-01 4.81246740e-01
-2.56509304e-01 1.48676944e+00 -1.96755454e-01 3.10525507e-01
6.01720393e-01 5.30555606e-01 5.83932400e-01 -5.46572804e-01
-6.96825147e-01 -7.63318419e-01 5.06912887e-01 -5.25981665e-01
-4.07367408e-01 -2.04336062e-01 -1.09385300e+00 1.25905365e-01
-3.39786410e-01 6.50480926e-01 4.46817726e-01 1.47752702e+00
4.06312883e-01 4.97011274e-01 -1.32680228e-02 4.19529110e-01
-9.64828789e-01 -9.80620205e-01 8.04974735e-02 6.53500915e-01
5.49705923e-01 -4.59639102e-01 -3.33085507e-01 -1.01102673e-01]
|
[11.194342613220215, 7.987008571624756]
|
ae23c1c9-3476-499c-85b4-5daadd5c7549
|
zen-pre-training-chinese-text-encoder
|
1911.0072
| null |
https://arxiv.org/abs/1911.00720v1
|
https://arxiv.org/pdf/1911.00720v1.pdf
|
ZEN: Pre-training Chinese Text Encoder Enhanced by N-gram Representations
|
The pre-training of text encoders normally processes text as a sequence of tokens corresponding to small text units, such as word pieces in English and characters in Chinese. It omits information carried by larger text granularity, and thus the encoders cannot easily adapt to certain combinations of characters. This leads to a loss of important semantic information, which is especially problematic for Chinese because the language does not have explicit word boundaries. In this paper, we propose ZEN, a BERT-based Chinese (Z) text encoder Enhanced by N-gram representations, where different combinations of characters are considered during training. As a result, potential word or phase boundaries are explicitly pre-trained and fine-tuned with the character encoder (BERT). Therefore ZEN incorporates the comprehensive information of both the character sequence and words or phrases it contains. Experimental results illustrated the effectiveness of ZEN on a series of Chinese NLP tasks. We show that ZEN, using less resource than other published encoders, can achieve state-of-the-art performance on most tasks. Moreover, it is shown that reasonable performance can be obtained when ZEN is trained on a small corpus, which is important for applying pre-training techniques to scenarios with limited data. The code and pre-trained models of ZEN are available at https://github.com/sinovation/zen.
|
['Jiaxin Bai', 'Yan Song', 'Shizhe Diao', 'Yonggang Wang', 'Tong Zhang']
|
2019-11-02
| null |
https://aclanthology.org/2020.findings-emnlp.425
|
https://aclanthology.org/2020.findings-emnlp.425.pdf
|
findings-of-the-association-for-computational
|
['chinese-named-entity-recognition', 'sentence-pair-modeling']
|
['natural-language-processing', 'natural-language-processing']
|
[ 2.38645688e-01 4.60601598e-02 -3.05805326e-01 -2.63561040e-01
-6.91941381e-01 -4.31576490e-01 3.65094513e-01 1.63567126e-01
-6.22325599e-01 8.97243142e-01 2.58260071e-01 -5.22655666e-01
2.90522456e-01 -9.72778201e-01 -7.01349139e-01 -5.34683287e-01
4.70893830e-02 4.57937092e-01 3.45281631e-01 -1.02361485e-01
1.36174366e-01 1.32310227e-01 -1.21116865e+00 5.59696198e-01
1.08344948e+00 5.97671032e-01 8.36178601e-01 7.01985598e-01
-6.39250636e-01 5.93405306e-01 -7.63488591e-01 -5.27689278e-01
-4.82171141e-02 -4.78236854e-01 -8.33168447e-01 1.25547107e-02
-5.56633361e-02 -3.41117889e-01 -2.81377524e-01 1.14231265e+00
3.47190917e-01 -4.24936600e-02 5.98599017e-01 -5.66422641e-01
-9.21044469e-01 1.20999253e+00 -2.41303623e-01 -2.48986222e-02
-7.12584928e-02 -1.88339293e-01 1.16061485e+00 -9.54629719e-01
4.63758588e-01 1.08667493e+00 4.12104249e-01 8.10063660e-01
-7.59461761e-01 -7.03653574e-01 2.82385796e-01 1.95545807e-01
-1.37960649e+00 -2.45909005e-01 4.90929186e-01 -1.81212693e-01
1.34149539e+00 3.25557254e-02 2.85539478e-01 1.02737820e+00
3.12091947e-01 1.10194051e+00 3.90520215e-01 -7.90674627e-01
1.06844500e-01 1.65326029e-01 5.85277602e-02 4.16611552e-01
3.01444918e-01 -2.09108680e-01 -2.77456194e-01 3.44462305e-01
6.83902383e-01 -5.99823967e-02 -2.95843869e-01 2.39682361e-01
-1.09061968e+00 9.04002070e-01 2.45472223e-01 6.84236884e-01
-2.60646760e-01 1.61350578e-01 5.96004367e-01 1.58198982e-01
4.95450169e-01 4.34675336e-01 -7.95326948e-01 -5.10743737e-01
-8.65941823e-01 -1.54718727e-01 6.36891723e-01 1.53705990e+00
8.49556088e-01 1.37521133e-01 -1.32418796e-01 8.81216168e-01
3.99004715e-03 3.10183018e-01 8.88896704e-01 -2.87059367e-01
8.05871725e-01 4.71189231e-01 -7.77978450e-02 -3.68630528e-01
-1.44516379e-01 -3.59535575e-01 -9.08469141e-01 -4.15704101e-01
2.48289332e-01 -5.09131312e-01 -1.19233716e+00 1.45466220e+00
-1.31503925e-01 -8.14784244e-02 3.30012411e-01 4.73151565e-01
5.84056675e-01 1.17658782e+00 -6.53022453e-02 -7.19072297e-02
1.45310831e+00 -1.20382142e+00 -1.16614330e+00 -5.45753419e-01
8.70545566e-01 -8.54330301e-01 1.14133358e+00 1.89259261e-01
-9.75269973e-01 -6.03748500e-01 -8.71304691e-01 -1.24411412e-01
-7.16735780e-01 4.70452368e-01 4.64318603e-01 6.29970610e-01
-8.36430311e-01 5.61442494e-01 -8.87560546e-01 -1.58225000e-01
2.72219211e-01 2.25671545e-01 5.84327690e-02 -1.51098073e-01
-1.71736383e+00 7.39082694e-01 8.15834165e-01 1.90616846e-01
-5.07293701e-01 -4.68262911e-01 -9.89016175e-01 4.23766971e-01
4.96070385e-01 -1.51902825e-01 1.47102749e+00 -1.00818849e+00
-1.61810660e+00 2.52320439e-01 -4.71922696e-01 -5.17253697e-01
3.46111238e-01 -5.05590260e-01 -6.55904710e-01 3.22479874e-01
3.51197980e-02 6.22822762e-01 7.19264209e-01 -8.56242478e-01
-6.73293889e-01 1.76821202e-01 -7.12973103e-02 2.08263263e-01
-5.34420788e-01 2.32811525e-01 -1.00808275e+00 -8.16251874e-01
-4.90980536e-01 -6.43011928e-01 -2.37744212e-01 -3.40792656e-01
-4.17582721e-01 -2.55216420e-01 6.35458350e-01 -7.52891541e-01
1.66609168e+00 -2.26005650e+00 -1.56080857e-01 -5.93884103e-02
-3.33491534e-01 6.56453967e-01 -8.33666623e-02 8.81290495e-01
1.91767275e-01 5.82656920e-01 -3.47944379e-01 -3.62215281e-01
-4.63800207e-02 4.78900880e-01 -3.05491596e-01 -2.26992797e-02
5.15461147e-01 8.78044665e-01 -9.22903836e-01 -5.83011091e-01
2.94834971e-01 5.08187234e-01 -4.18525726e-01 1.48698166e-01
-4.53676164e-01 1.25122070e-01 -6.01972938e-01 4.30603415e-01
5.47186315e-01 -4.10394490e-01 1.60188273e-01 3.29194516e-01
-2.60282844e-01 8.28471363e-01 -8.52218747e-01 1.47338665e+00
-6.83956444e-01 7.95510769e-01 -3.27772766e-01 -8.28765392e-01
9.55581248e-01 6.21687889e-01 6.74085617e-02 -5.36231816e-01
1.75306320e-01 2.73261040e-01 3.95779237e-02 -3.19695741e-01
7.24806726e-01 1.41962841e-01 -9.72470343e-02 3.25299323e-01
1.98911473e-01 -1.09552547e-01 6.39934480e-01 6.62368536e-02
8.56525600e-01 2.46814117e-02 3.99609268e-01 -1.60536140e-01
5.75267911e-01 2.52494011e-02 4.97685283e-01 6.23160005e-01
2.20520332e-01 6.89807355e-01 5.51362574e-01 -1.39628917e-01
-1.24090207e+00 -7.22522140e-01 -5.29620051e-01 9.21647727e-01
1.74686238e-02 -8.33583117e-01 -8.92695963e-01 -6.35724366e-01
-3.28334302e-01 9.35576737e-01 -5.25925100e-01 -1.61474273e-01
-7.89863825e-01 -5.25949359e-01 6.21584773e-01 6.78663909e-01
4.33058113e-01 -1.43644726e+00 -3.97102773e-01 6.11894011e-01
-2.15602592e-01 -1.26192367e+00 -6.07509911e-01 4.89760965e-01
-7.94241905e-01 -8.39405477e-01 -8.75016809e-01 -1.12106133e+00
7.81628311e-01 4.10633385e-02 8.66821170e-01 1.23678371e-01
6.58117905e-02 -2.12845787e-01 -1.08873069e+00 -5.06188750e-01
-6.03715003e-01 4.64842319e-01 -3.96639496e-01 -1.14173137e-01
6.58421040e-01 -4.40535471e-02 -2.04184800e-01 2.03630984e-01
-1.15294671e+00 3.16743851e-01 9.10289109e-01 9.71417725e-01
4.29662913e-01 4.75540996e-01 6.56680048e-01 -1.17300594e+00
5.57860434e-01 -3.01040113e-01 -6.19073331e-01 3.54629308e-01
-4.91621673e-01 2.78617233e-01 1.08950567e+00 -5.78108013e-01
-1.22605705e+00 -1.77519798e-01 -5.13379931e-01 -1.00964516e-01
-1.01695478e-01 9.35644150e-01 -2.46937260e-01 5.23983896e-01
3.19739282e-01 4.25159007e-01 -2.99540311e-01 -6.95865095e-01
2.15264261e-01 1.08914423e+00 1.92603543e-01 -5.21494389e-01
5.41514456e-01 -8.43800721e-04 -6.95465624e-01 -9.80481148e-01
-5.35512984e-01 -4.58074957e-01 -8.12760830e-01 2.22992361e-01
7.51749992e-01 -1.02110124e+00 -5.70072085e-02 4.61566895e-01
-1.33833349e+00 -5.92609465e-01 -1.91085458e-01 6.45094514e-01
-1.67891786e-01 5.13538778e-01 -1.04061472e+00 -6.39620304e-01
-4.94622946e-01 -1.03928125e+00 9.24246609e-01 1.28316790e-01
-1.92408115e-01 -1.17320788e+00 -3.28165084e-01 -2.34982252e-01
3.13309968e-01 -4.41293746e-01 9.54832196e-01 -6.99235320e-01
-4.09056932e-01 -2.49262676e-01 -1.21291041e-01 6.10357404e-01
4.04183835e-01 2.31245279e-01 -7.28810012e-01 -2.77428806e-01
-2.72865087e-01 -2.14988813e-01 9.27709162e-01 2.45102376e-01
1.41152203e+00 -4.36451942e-01 -2.76465118e-01 4.75585997e-01
1.42168832e+00 4.50278968e-01 7.73735166e-01 3.38919789e-01
6.80512846e-01 4.02304798e-01 6.74392104e-01 4.51886922e-01
1.66542560e-01 3.34440559e-01 2.67125785e-01 -1.62068568e-02
4.18993831e-02 -4.18012887e-01 5.63056290e-01 1.21997702e+00
1.83357328e-01 -6.43927217e-01 -9.24851656e-01 7.08262980e-01
-1.67160463e+00 -6.69113040e-01 -5.17476201e-02 2.03605580e+00
1.37293196e+00 3.32400560e-01 -4.69644874e-01 1.51384592e-01
9.15682435e-01 1.46146506e-01 -3.61486286e-01 -5.72455823e-01
1.93579346e-02 4.98827487e-01 7.96969593e-01 5.61255991e-01
-1.00894737e+00 1.45730019e+00 5.86587858e+00 1.19035327e+00
-1.20350158e+00 -5.82407638e-02 2.91345835e-01 2.00363368e-01
-3.56732011e-01 -1.94057953e-02 -1.22813690e+00 7.32637465e-01
1.13729572e+00 -3.77420306e-01 2.22187698e-01 8.55889797e-01
2.41150066e-01 9.83561054e-02 -9.44864511e-01 5.65266192e-01
-1.17569365e-01 -1.33442080e+00 1.82537884e-01 -1.42940983e-01
8.02090883e-01 1.42150238e-01 -1.77834630e-01 4.27078545e-01
3.48230243e-01 -9.53664422e-01 8.12928736e-01 -6.38068765e-02
1.15663350e+00 -7.77667761e-01 1.01483655e+00 4.97225612e-01
-1.23516738e+00 1.02713540e-01 -9.16312695e-01 -1.42333098e-02
3.03371638e-01 8.27486515e-01 -9.51502621e-01 7.12920666e-01
4.97930884e-01 9.73689675e-01 -3.62712771e-01 8.61399829e-01
-7.18662083e-01 8.81871283e-01 -1.46511942e-01 -4.58068311e-01
5.46555996e-01 -3.15983184e-02 1.14298062e-02 1.69972479e+00
5.52498162e-01 5.68435080e-02 -1.49392605e-01 6.07296109e-01
-1.06410392e-01 2.64326274e-01 -3.04282844e-01 -3.88874918e-01
5.94483733e-01 8.75364542e-01 -6.12465799e-01 -5.90069950e-01
-7.07649589e-01 1.08659542e+00 3.43834609e-01 6.16017759e-01
-7.19214618e-01 -8.23636889e-01 5.33645153e-01 -2.04180107e-01
7.79426932e-01 -3.12446445e-01 -2.16795981e-01 -1.38771689e+00
-8.51645251e-04 -6.50796711e-01 1.88186198e-01 -5.90830147e-01
-1.02692866e+00 9.24667299e-01 -9.88673940e-02 -1.31642067e+00
-5.02366126e-01 -8.48753095e-01 -5.87568939e-01 1.01237011e+00
-1.81036508e+00 -8.18396509e-01 1.76302716e-01 3.98898363e-01
9.79924619e-01 -4.09469120e-02 8.16932619e-01 3.54375958e-01
-6.47719562e-01 7.48964190e-01 5.30661225e-01 5.76869786e-01
6.81696892e-01 -1.26706851e+00 7.34933794e-01 1.10935378e+00
1.56096458e-01 5.21030784e-01 4.54340190e-01 -7.64276922e-01
-1.07148862e+00 -1.34965193e+00 1.50961018e+00 -3.53017496e-03
7.53491879e-01 -7.04742610e-01 -1.18047237e+00 8.88545752e-01
3.70821387e-01 -3.06114793e-01 6.32519007e-01 3.42231952e-02
7.07617626e-02 2.18407229e-01 -4.99424100e-01 6.87650740e-01
6.78844869e-01 -5.98997891e-01 -6.81463838e-01 2.11298928e-01
1.09265566e+00 -5.22927463e-01 -6.12528205e-01 1.04588404e-01
2.12422371e-01 -5.69773734e-01 5.15507221e-01 -2.97875613e-01
5.96027970e-01 -4.16223630e-02 1.27388626e-01 -1.49956357e+00
-3.28285694e-01 -4.21548277e-01 -3.40800285e-02 1.36852551e+00
8.19789648e-01 -5.47385693e-01 6.33914292e-01 2.78344750e-01
-5.52531421e-01 -6.55578911e-01 -6.98768616e-01 -8.85207713e-01
3.91508818e-01 -5.30915976e-01 8.06675315e-01 8.54400277e-01
2.76360095e-01 3.99956018e-01 -4.34922606e-01 4.47167531e-02
3.80541682e-02 -1.14006110e-01 3.22215706e-01 -7.83966601e-01
-2.03143895e-01 -4.26664442e-01 5.78511171e-02 -1.57933748e+00
2.35843554e-01 -7.73198247e-01 3.69910330e-01 -1.48722005e+00
-2.00792834e-01 -5.57837725e-01 -1.94689512e-01 6.30249619e-01
-5.40534973e-01 -3.09084177e-01 1.14410833e-01 1.26209408e-01
-2.47499079e-01 8.50910127e-01 1.42602897e+00 -1.80154160e-01
-1.87014222e-01 8.00443292e-02 -5.49950898e-01 3.39072973e-01
1.26844573e+00 -5.16202569e-01 -5.00288725e-01 -8.28215837e-01
1.59049064e-01 -1.80861071e-01 -3.56315404e-01 -7.87996411e-01
3.08715641e-01 -7.07267150e-02 2.54532844e-01 -7.36813486e-01
1.32653579e-01 -8.49259496e-01 -2.86557436e-01 3.68062049e-01
-4.15498614e-01 1.93344224e-02 3.85587245e-01 2.57571965e-01
-6.19505286e-01 -7.85684347e-01 4.98660594e-01 -2.43703410e-01
-8.19439352e-01 3.80159765e-01 -7.20714748e-01 3.29064801e-02
7.47637689e-01 -1.87880516e-01 -2.81675220e-01 -3.56069654e-01
-2.42883384e-01 3.68272573e-01 4.16581482e-01 4.73871142e-01
6.07226074e-01 -1.10867703e+00 -6.55304611e-01 4.51651067e-01
1.41183347e-01 4.01979834e-01 1.35543406e-01 2.54032552e-01
-8.44381154e-01 7.62873948e-01 2.28038207e-02 -2.13217139e-01
-9.28673804e-01 7.21874595e-01 9.68663895e-04 -4.73478526e-01
-8.92964780e-01 6.92930102e-01 3.49652171e-01 -2.61315972e-01
2.64092177e-01 -7.64586687e-01 -4.23290759e-01 -1.15881100e-01
7.71382630e-01 -1.67240977e-01 3.04756630e-02 -3.79661143e-01
-1.15097746e-01 3.65091920e-01 -3.48250896e-01 -6.73261732e-02
1.16662073e+00 -2.36336395e-01 -3.37573737e-02 4.77162153e-01
1.26251531e+00 5.06539159e-02 -1.34423947e+00 -5.52258790e-01
6.73473850e-02 -3.32559824e-01 -1.71808362e-01 -6.23578370e-01
-9.27864492e-01 1.23040736e+00 -1.71100810e-01 -4.98085879e-02
1.15462685e+00 -8.42400640e-02 9.58733916e-01 4.65950429e-01
1.90065056e-01 -1.33692074e+00 -7.80722126e-02 1.06479347e+00
5.09662449e-01 -9.68469799e-01 -3.62092912e-01 -5.31522453e-01
-7.92449713e-01 1.40490115e+00 5.69804311e-01 5.77139407e-02
6.66940272e-01 4.15468931e-01 5.65598905e-02 2.27827713e-01
-9.52112377e-01 -2.31043816e-01 8.72278810e-02 4.08190578e-01
5.95173240e-01 2.05625162e-01 -4.14842874e-01 6.92230284e-01
-3.63478869e-01 -2.03383982e-01 7.56267190e-01 9.97621536e-01
-6.58338964e-01 -1.45613778e+00 -1.52528912e-01 4.84706402e-01
-7.28455305e-01 -6.43860698e-01 -1.09965108e-01 8.34960878e-01
-6.50771856e-02 7.99616992e-01 3.07827264e-01 -1.35949790e-01
-4.07818593e-02 1.80618703e-01 1.10736273e-01 -9.45018589e-01
-4.11488652e-01 3.36377442e-01 1.80474713e-01 -7.93895796e-02
-7.31563568e-02 -5.53887486e-01 -1.49916267e+00 -2.71315038e-01
-4.29497749e-01 5.28440893e-01 5.27791202e-01 8.27135146e-01
1.30625740e-01 8.09947729e-01 4.71182972e-01 -4.67965007e-01
-3.81793737e-01 -1.28090990e+00 -6.10795379e-01 1.87033728e-01
3.34122807e-01 -2.26590827e-01 -2.08114818e-01 2.15959758e-01]
|
[10.189912796020508, 10.127735137939453]
|
c81947e1-3ed8-4fd7-a38f-7df27ac11922
|
adaptive-rotated-convolution-for-rotated
|
2303.0782
| null |
https://arxiv.org/abs/2303.07820v1
|
https://arxiv.org/pdf/2303.07820v1.pdf
|
Adaptive Rotated Convolution for Rotated Object Detection
|
Rotated object detection aims to identify and locate objects in images with arbitrary orientation. In this scenario, the oriented directions of objects vary considerably across different images, while multiple orientations of objects exist within an image. This intrinsic characteristic makes it challenging for standard backbone networks to extract high-quality features of these arbitrarily orientated objects. In this paper, we present Adaptive Rotated Convolution (ARC) module to handle the aforementioned challenges. In our ARC module, the convolution kernels rotate adaptively to extract object features with varying orientations in different images, and an efficient conditional computation mechanism is introduced to accommodate the large orientation variations of objects within an image. The two designs work seamlessly in rotated object detection problem. Moreover, ARC can conveniently serve as a plug-and-play module in various vision backbones to boost their representation ability to detect oriented objects accurately. Experiments on commonly used benchmarks (DOTA and HRSC2016) demonstrate that equipped with our proposed ARC module in the backbone network, the performance of multiple popular oriented object detectors is significantly improved (e.g. +3.03% mAP on Rotated RetinaNet and +4.16% on CFA). Combined with the highly competitive method Oriented R-CNN, the proposed approach achieves state-of-the-art performance on the DOTA dataset with 81.77% mAP.
|
['Gao Huang', 'Shiji Song', 'Zidong Wang', 'Weihao Gan', 'Yulin Wang', 'Yizeng Han', 'Zhuofan Xia', 'Yiru Wang', 'Yifan Pu']
|
2023-03-14
| null | null | null | null |
['object-detection-in-aerial-images']
|
['computer-vision']
|
[-1.50164932e-01 -3.36795956e-01 -3.50303482e-04 -3.67989153e-01
-1.80569082e-01 -4.80618089e-01 2.74454087e-01 -4.96408612e-01
-6.80792332e-01 7.12725148e-02 -3.72869998e-01 -1.92814171e-01
-1.38342187e-01 -6.76777303e-01 -8.00208569e-01 -8.42857540e-01
-1.29267350e-01 -7.19137192e-02 6.86293602e-01 -2.29457989e-01
2.55984277e-01 8.70730102e-01 -1.56473839e+00 3.05814236e-01
5.86920798e-01 1.17003632e+00 5.77763021e-01 4.29275304e-01
1.03291154e-01 6.39492750e-01 -4.95242029e-01 -2.36813188e-01
3.26683044e-01 2.40428299e-01 -4.40319598e-01 2.37352222e-01
6.59974217e-01 -4.70235616e-01 -5.61277151e-01 1.09585285e+00
4.69851285e-01 -4.61986475e-02 6.05130136e-01 -9.04670000e-01
-1.00364351e+00 6.41987145e-01 -9.01879132e-01 7.22011864e-01
-1.96960509e-01 3.14812034e-01 7.18801439e-01 -1.29315853e+00
2.27918282e-01 1.28220212e+00 1.82704136e-01 3.45737994e-01
-8.96754682e-01 -6.58881009e-01 6.30816519e-01 3.18414122e-01
-1.57917786e+00 -4.15806025e-01 6.61346793e-01 -4.00489777e-01
8.91077816e-01 1.37644649e-01 5.89787960e-01 9.18056071e-01
3.26227099e-02 8.92079234e-01 7.88807392e-01 -8.68903846e-02
-6.16765674e-03 5.02890460e-02 4.71046358e-01 6.27077043e-01
7.65967727e-01 -1.76129818e-01 -2.91073978e-01 4.32337761e-01
8.54625463e-01 3.07797939e-01 -4.33628589e-01 -3.32973599e-01
-1.42894518e+00 2.30205119e-01 1.17624342e+00 7.18083531e-02
-3.81017476e-01 1.62212387e-01 1.42185107e-01 -1.14575706e-01
6.64922893e-02 1.38458759e-01 -4.18913662e-01 4.77711529e-01
-3.81829679e-01 2.80969650e-01 1.03547610e-01 1.22961819e+00
5.89069963e-01 2.39196792e-01 -3.79462272e-01 9.54745650e-01
5.50682425e-01 7.59583414e-01 5.74465871e-01 -4.11454618e-01
6.06804729e-01 6.22376740e-01 1.15272805e-01 -1.20846784e+00
-6.17967248e-01 -9.30062830e-01 -1.05111003e+00 -3.94323887e-03
5.49496293e-01 2.36352473e-01 -1.12658775e+00 1.40439236e+00
5.02499104e-01 6.82126433e-02 -2.21714582e-02 1.28751755e+00
8.94459844e-01 4.53888237e-01 -3.94946560e-02 2.31161579e-01
1.93956065e+00 -9.87674952e-01 -3.42371792e-01 -4.01022196e-01
2.24677399e-01 -8.24248135e-01 8.72364104e-01 4.02447701e-01
-1.00229907e+00 -9.83007371e-01 -1.24936235e+00 -6.82129432e-03
-2.47930422e-01 6.84903383e-01 6.65466070e-01 5.41911483e-01
-9.15342391e-01 9.65876877e-02 -7.31132209e-01 -2.04703286e-02
8.19596112e-01 4.88969207e-01 -2.00191870e-01 -4.32868570e-01
-6.53744578e-01 4.61178005e-01 4.46593791e-01 6.97977424e-01
-8.09296250e-01 -5.59713840e-01 -5.37206352e-01 2.75892802e-02
3.36269438e-01 -6.26535237e-01 1.03745866e+00 -7.96262324e-01
-1.10663879e+00 6.49245679e-01 -1.65309943e-02 -3.46881062e-01
5.53508639e-01 -3.13244700e-01 -5.58488607e-01 1.19727753e-01
-2.25985004e-03 8.11581790e-01 1.14702880e+00 -9.15703475e-01
-9.44490552e-01 -6.91632569e-01 1.71610996e-01 1.83801532e-01
-4.15180713e-01 2.23343924e-01 -8.99596572e-01 -5.45295537e-01
5.17017841e-01 -9.68737304e-01 -1.41019702e-01 1.67495921e-01
-4.15275395e-01 -4.34270710e-01 9.14938629e-01 -2.19531134e-01
8.44316065e-01 -2.25734210e+00 -1.00259870e-01 -1.19923852e-01
3.70643914e-01 4.47125912e-01 -1.44207299e-01 -4.42324191e-01
-1.85032278e-01 -1.73484683e-01 2.68340528e-01 2.73368862e-02
-2.73981452e-01 -1.83376551e-01 -2.90131807e-01 6.91020012e-01
7.13769197e-01 9.15877104e-01 -7.07935929e-01 -3.24366033e-01
2.29012489e-01 6.22660220e-01 -5.05811810e-01 1.50034521e-02
2.26783291e-01 1.91379160e-01 -6.55892193e-01 8.60275269e-01
1.09741211e+00 -3.95412296e-01 3.29387262e-02 -5.42721868e-01
-2.85958797e-01 4.63568904e-02 -1.30662501e+00 1.41385269e+00
-2.27765441e-01 6.37434185e-01 -2.15375394e-01 -8.97697747e-01
1.14863396e+00 -5.32411076e-02 4.41960879e-02 -7.51364410e-01
1.57529458e-01 1.62612021e-01 5.71302891e-01 -5.20013869e-01
6.38510466e-01 6.06653631e-01 1.83639154e-01 -1.27960831e-01
9.07909684e-03 4.41622883e-01 5.32471947e-02 -7.97972456e-02
8.38183343e-01 -1.30724117e-01 5.66778556e-02 -2.80045182e-01
7.02552855e-01 -4.47206378e-01 6.09581470e-01 7.31686354e-01
-3.12848926e-01 7.79544413e-01 1.33029535e-01 -7.63521314e-01
-9.11541224e-01 -9.17813599e-01 -5.06010413e-01 1.04959965e+00
5.81583858e-01 7.52590820e-02 -4.45320994e-01 -7.17537224e-01
-1.47199899e-01 1.02598235e-01 -5.65545201e-01 -6.29308820e-02
-7.92276859e-01 -1.10569942e+00 3.72280508e-01 9.59159017e-01
1.02042031e+00 -9.55472231e-01 -6.76326931e-01 1.05000518e-01
4.42964919e-02 -1.28090763e+00 -4.63884234e-01 1.06639847e-01
-6.78652346e-01 -9.67708588e-01 -7.69272983e-01 -1.22938645e+00
9.69634175e-01 1.23318815e+00 8.10792148e-01 -2.25886166e-01
-5.52976787e-01 -1.56857294e-03 -3.44928533e-01 -2.89573580e-01
3.37448746e-01 8.41664448e-02 -1.07297972e-02 5.68741083e-01
2.96120584e-01 -3.54827255e-01 -1.18748546e+00 9.01583314e-01
-9.57988739e-01 2.30229143e-02 8.80542159e-01 7.42306828e-01
4.40100908e-01 -1.15053067e-02 6.04882181e-01 -4.14899081e-01
1.28241956e-01 -4.29082036e-01 -9.63464141e-01 1.84062153e-01
-3.32709372e-01 -6.23807497e-02 5.60042620e-01 -7.99367726e-01
-8.60576034e-01 1.94935992e-01 1.62225604e-01 -5.01552701e-01
-9.72697884e-02 1.94418747e-02 -3.24029505e-01 -2.21145570e-01
6.75601840e-01 2.77184606e-01 -4.65061724e-01 -5.10107517e-01
2.85991281e-01 8.30127776e-01 5.68914354e-01 -3.68189991e-01
8.93814385e-01 6.99209094e-01 -9.26258191e-02 -6.93083525e-01
-6.87515378e-01 -6.37704372e-01 -4.47032630e-01 -2.45493740e-01
7.13379323e-01 -1.34426856e+00 -8.20644438e-01 8.78268361e-01
-1.17101264e+00 -3.09590064e-02 1.50401309e-01 5.89229345e-01
5.49243838e-02 5.61129600e-02 -2.70994127e-01 -6.58831120e-01
-3.88973862e-01 -1.53501773e+00 1.12065375e+00 7.14824378e-01
4.97807801e-01 -3.43808860e-01 -7.07621336e-01 3.15759033e-01
5.41321397e-01 -2.75309712e-01 5.59390426e-01 -4.45765525e-01
-1.06932724e+00 -3.24969202e-01 -8.43432426e-01 3.76671970e-01
2.64701229e-02 1.15754128e-01 -1.11838722e+00 -4.60534483e-01
-2.83847719e-01 -9.50520784e-02 1.01077998e+00 3.18136275e-01
1.49162948e+00 -1.38245687e-01 -2.77692884e-01 8.16398919e-01
1.24094927e+00 2.63768077e-01 6.62476301e-01 3.35639238e-01
9.05261815e-01 3.28020424e-01 7.41618693e-01 3.37167561e-01
3.12272280e-01 8.70865881e-01 7.38856912e-01 1.17570870e-01
-2.97825068e-01 1.09630749e-01 2.39653766e-01 6.53420985e-01
-3.66331816e-01 -1.51680112e-01 -6.41763687e-01 6.29126072e-01
-1.69757807e+00 -6.81197584e-01 -2.88347930e-01 2.10439229e+00
3.82417560e-01 2.40434989e-01 4.63184863e-02 -5.89641370e-03
9.92850840e-01 2.97687352e-01 -7.09456742e-01 1.96834862e-01
-2.99435645e-01 -1.37258740e-03 7.93400645e-01 -1.89592734e-01
-1.38024652e+00 7.56136596e-01 4.98397779e+00 8.22801769e-01
-1.35332930e+00 -1.17277764e-01 6.14189625e-01 -2.23477301e-03
4.05575335e-01 -2.34730214e-01 -1.26711094e+00 4.71208960e-01
3.81983757e-01 1.97744146e-01 3.42196047e-01 1.24926269e+00
2.48973425e-02 3.42812300e-01 -8.00826728e-01 1.02100456e+00
1.69398971e-02 -1.17227733e+00 -1.24445081e-01 4.54354882e-02
7.36429572e-01 2.65727848e-01 6.84251785e-01 2.79844582e-01
-1.13672957e-01 -9.20065701e-01 8.97321284e-01 2.28600442e-01
6.86273932e-01 -7.04437673e-01 6.70618474e-01 8.13727081e-02
-1.42409503e+00 -3.42934072e-01 -8.81872416e-01 1.38029486e-01
-3.96988958e-01 4.80932444e-01 -8.16228092e-01 5.69095373e-01
1.17046976e+00 7.50941396e-01 -8.16592932e-01 1.40093684e+00
-1.26875624e-01 3.41963857e-01 -1.28272533e-01 -9.29498151e-02
2.84100920e-01 -3.40026654e-02 4.27755445e-01 1.03148103e+00
2.16033801e-01 -1.56204566e-01 -1.55861512e-01 8.50910783e-01
-2.38703087e-01 3.32803316e-02 -3.46975863e-01 3.27571899e-01
5.02175868e-01 1.72203112e+00 -8.67412090e-01 -2.27851793e-01
-4.93168503e-01 6.96658492e-01 5.69025517e-01 4.27645117e-01
-1.14021468e+00 -5.16136944e-01 7.28836179e-01 6.82563186e-02
1.01913047e+00 -2.38216147e-01 1.81765720e-01 -1.24441218e+00
3.64886433e-01 -1.03368998e+00 1.11769006e-01 -8.00604582e-01
-1.17924762e+00 8.33720982e-01 -3.50604087e-01 -1.40883803e+00
4.23653215e-01 -1.15082061e+00 -4.79977190e-01 7.63262451e-01
-1.68977618e+00 -1.14908624e+00 -8.15791905e-01 5.73410928e-01
6.22804046e-01 -1.93447784e-01 1.55976802e-01 5.25530219e-01
-1.11537552e+00 7.53849030e-01 4.16472517e-02 4.77113664e-01
6.70679808e-01 -9.92644370e-01 5.18773735e-01 1.02054286e+00
2.16724157e-01 9.31874633e-01 1.08626392e-02 -2.60005802e-01
-1.77893460e+00 -1.67922032e+00 9.89919603e-02 -2.20915854e-01
4.47889656e-01 -2.77074784e-01 -9.88529623e-01 6.41704261e-01
-1.44008100e-01 8.07040334e-01 4.74964790e-02 -8.81683901e-02
-8.13463569e-01 -5.73319912e-01 -7.65333593e-01 6.78356707e-01
1.12447011e+00 -1.58327386e-01 -2.07224682e-01 3.86652231e-01
7.57186353e-01 -4.93401766e-01 -5.19216776e-01 7.02778697e-01
5.90402484e-01 -9.95646238e-01 1.40385580e+00 -7.44081914e-01
1.04940139e-01 -8.76099110e-01 -2.39319399e-01 -7.72355855e-01
-5.58206260e-01 -4.06048149e-01 -1.73107773e-01 1.07562053e+00
1.27112344e-01 -8.99023473e-01 5.00163317e-01 2.02183276e-01
-2.26925462e-01 -7.96638370e-01 -7.46154070e-01 -7.51075327e-01
-3.60350072e-01 -4.69800502e-01 6.14387691e-01 5.99935532e-01
-6.42769992e-01 1.20993227e-01 -4.31684516e-02 8.21735442e-01
6.10574782e-01 1.76053718e-01 7.32815504e-01 -7.70320415e-01
-3.19062710e-01 -6.17841482e-01 -1.00894678e+00 -1.51541650e+00
-1.77591234e-01 -7.16075480e-01 -3.30247614e-03 -9.59041059e-01
1.96055293e-01 -7.48779058e-01 -6.62085533e-01 2.96197623e-01
-4.53818172e-01 5.45609295e-01 2.68410265e-01 3.12211961e-01
-8.19713295e-01 4.23201650e-01 1.34193623e+00 -3.54853511e-01
-8.39232877e-02 2.21179068e-01 -7.74629891e-01 8.39375734e-01
7.30349243e-01 -3.42984706e-01 -3.19542080e-01 -7.68328846e-01
-7.34651387e-02 -5.32058537e-01 7.99314201e-01 -1.11269498e+00
2.52165198e-01 1.00547165e-01 6.55458689e-01 -8.22696924e-01
2.30364755e-01 -7.63604164e-01 -1.89978614e-01 3.64982009e-01
-7.85050988e-02 9.33700241e-03 2.61938512e-01 7.58205891e-01
-6.70925230e-02 -3.06515954e-03 9.57011640e-01 2.47986853e-01
-9.01866078e-01 4.51412201e-01 3.04750558e-02 6.45122898e-04
1.07592452e+00 -1.07045956e-01 -7.18590438e-01 2.66993910e-01
-3.25421363e-01 9.39037278e-02 1.39157653e-01 6.61978304e-01
8.82212579e-01 -1.25393414e+00 -5.53145230e-01 3.00098419e-01
3.05789262e-01 6.77518487e-01 6.02920473e-01 9.06511128e-01
-7.44781911e-01 5.09395540e-01 -2.79326230e-01 -9.42945004e-01
-9.94669676e-01 7.13844359e-01 5.60528100e-01 1.83372095e-01
-5.87774456e-01 1.03294313e+00 6.36559725e-01 -1.02289103e-01
2.64450818e-01 -5.56915879e-01 -2.83842951e-01 -4.66308817e-02
8.01811934e-01 2.29519054e-01 1.50011316e-01 -6.19271994e-01
-4.62992460e-01 7.92683840e-01 -5.40500700e-01 3.67389441e-01
1.27349412e+00 -1.68630853e-02 1.20872639e-01 1.71823665e-01
1.10751462e+00 -1.60320923e-01 -1.53581977e+00 -5.68078637e-01
-2.74707317e-01 -6.52841806e-01 1.38445154e-01 -3.72298509e-01
-1.49031997e+00 9.02419984e-01 7.85080791e-01 8.55510123e-03
1.05852485e+00 -5.75239696e-02 3.19805801e-01 5.35060108e-01
3.23481172e-01 -5.73956907e-01 2.97134966e-01 1.63512543e-01
8.23260128e-01 -1.30670333e+00 -2.19567977e-02 -4.62529421e-01
-6.59838974e-01 1.10716355e+00 1.02766275e+00 -3.14555675e-01
3.37671608e-01 -2.64107864e-02 7.21479356e-02 -1.16132438e-01
-3.98991227e-01 -2.76547849e-01 6.05632186e-01 6.21029019e-01
2.38857538e-01 -7.04729035e-02 2.79637333e-02 4.51064497e-01
1.68930024e-01 -4.77342188e-01 3.97101969e-01 6.82321310e-01
-3.76231700e-01 -5.08448362e-01 -4.94922906e-01 1.92793667e-01
-4.38855380e-01 -6.56226091e-03 1.90078765e-01 9.21384454e-01
2.10299313e-01 8.68143201e-01 2.59036779e-01 -3.39827120e-01
3.76894504e-01 -3.83100688e-01 3.76145244e-01 -3.26790333e-01
-2.49045804e-01 8.45779032e-02 -3.07455182e-01 -6.23373032e-01
-4.27838892e-01 -4.97312427e-01 -1.11506093e+00 1.03424728e-01
-5.74523389e-01 -3.97513628e-01 8.38014722e-01 5.57416260e-01
4.62874860e-01 1.00733984e+00 8.78010809e-01 -8.99144232e-01
-6.84288502e-01 -9.00408089e-01 -6.34917080e-01 8.77360329e-02
2.72080153e-01 -7.96553373e-01 -1.01011746e-01 -1.78069055e-01]
|
[8.781402587890625, -0.6965448260307312]
|
fa377334-4a93-4f0e-bad6-a2dcc1d6d08c
|
st-hoi-a-spatial-temporal-baseline-for-human
|
2105.11731
| null |
https://arxiv.org/abs/2105.11731v2
|
https://arxiv.org/pdf/2105.11731v2.pdf
|
ST-HOI: A Spatial-Temporal Baseline for Human-Object Interaction Detection in Videos
|
Detecting human-object interactions (HOI) is an important step toward a comprehensive visual understanding of machines. While detecting non-temporal HOIs (e.g., sitting on a chair) from static images is feasible, it is unlikely even for humans to guess temporal-related HOIs (e.g., opening/closing a door) from a single video frame, where the neighboring frames play an essential role. However, conventional HOI methods operating on only static images have been used to predict temporal-related interactions, which is essentially guessing without temporal contexts and may lead to sub-optimal performance. In this paper, we bridge this gap by detecting video-based HOIs with explicit temporal information. We first show that a naive temporal-aware variant of a common action detection baseline does not work on video-based HOIs due to a feature-inconsistency issue. We then propose a simple yet effective architecture named Spatial-Temporal HOI Detection (ST-HOI) utilizing temporal information such as human and object trajectories, correctly-localized visual features, and spatial-temporal masking pose features. We construct a new video HOI benchmark dubbed VidHOI where our proposed approach serves as a solid baseline.
|
['Jiashi Feng', 'Roger Zimmermann', 'Li-Wei Wang', 'Chun-Yu Liao', 'Meng-Jiun Chiou']
|
2021-05-25
| null | null | null | null |
['spatio-temporal-action-localization']
|
['computer-vision']
|
[ 3.12119067e-01 -2.98647135e-01 -6.51997179e-02 -1.54951721e-01
-3.88768941e-01 -5.16270280e-01 7.58789361e-01 1.42613053e-01
-2.70871997e-01 3.66621882e-01 5.69413044e-02 -2.88349450e-01
-6.09412454e-02 -2.86323875e-01 -8.07450235e-01 -5.63736379e-01
-3.41235220e-01 3.23157459e-02 8.03624451e-01 5.71190529e-02
2.89254397e-01 4.52474713e-01 -1.74366331e+00 6.15980566e-01
3.62824500e-01 8.66194189e-01 2.03703254e-01 9.57799077e-01
4.69631761e-01 1.02531171e+00 -5.69011927e-01 1.33384407e-01
1.87144846e-01 -7.36377597e-01 -8.01542997e-01 3.80654007e-01
5.09076059e-01 -6.09940469e-01 -4.50494885e-01 6.82597220e-01
-8.73887260e-03 4.87962306e-01 4.15571332e-01 -1.70097363e+00
-2.80810386e-01 1.66409910e-01 -7.20688581e-01 5.34741044e-01
8.74253213e-01 5.14477134e-01 9.67482150e-01 -8.74871492e-01
8.11097801e-01 1.26724100e+00 5.08623302e-01 4.99816835e-01
-1.04692042e+00 -3.17430139e-01 3.48759860e-01 5.97674966e-01
-1.35179126e+00 -5.07580101e-01 6.17367744e-01 -5.09973049e-01
1.17964482e+00 4.88289297e-01 7.58725822e-01 1.20937395e+00
1.92078814e-01 1.25604725e+00 1.03000617e+00 -3.98167461e-01
1.35823935e-01 -2.31656194e-01 2.88382679e-01 6.67181015e-01
4.06744964e-02 4.24414575e-01 -7.47625709e-01 2.02996098e-02
5.62164426e-01 1.90340862e-01 -4.64705050e-01 -4.72044945e-01
-1.79972088e+00 4.64326322e-01 2.73039609e-01 3.63309771e-01
-3.40633959e-01 6.54194579e-02 3.28168929e-01 2.47610852e-01
-8.81311148e-02 3.12667161e-01 -9.27006528e-02 -2.37856016e-01
-1.02623355e+00 4.47830498e-01 5.25849760e-01 8.91642034e-01
6.20201528e-01 -1.93537757e-01 -5.06632268e-01 1.55796692e-01
-7.97367394e-02 6.73456490e-02 2.46064305e-01 -1.06615639e+00
3.93811166e-01 4.62753415e-01 5.78375876e-01 -1.20247805e+00
-5.40136933e-01 9.71058607e-02 -5.80657244e-01 2.53985494e-01
8.28567207e-01 1.52629331e-01 -9.02542591e-01 1.56172585e+00
4.72718924e-01 5.05966783e-01 -1.93384245e-01 1.23832524e+00
5.95710516e-01 7.12200820e-01 -3.15012001e-02 -4.75208163e-01
1.30928075e+00 -1.20597231e+00 -8.19175303e-01 -2.42814690e-01
5.14964223e-01 -6.07571483e-01 9.92176294e-01 3.70505482e-01
-9.38357055e-01 -5.65316975e-01 -1.01933992e+00 8.97269174e-02
-2.98205376e-01 4.93471418e-03 5.57270110e-01 3.75904679e-01
-9.15505528e-01 4.28331763e-01 -1.05587888e+00 -7.30825424e-01
3.38447727e-02 1.28064990e-01 -5.83840787e-01 -6.36566505e-02
-8.98217440e-01 6.40188396e-01 4.57474053e-01 3.43780398e-01
-1.23875356e+00 -2.64193028e-01 -9.42490578e-01 -1.60332009e-01
8.61450732e-01 -3.75438899e-01 1.28250182e+00 -9.56289470e-01
-1.14652181e+00 6.91339135e-01 -7.48988748e-01 -6.61961555e-01
8.07902157e-01 -2.90426165e-01 -3.29567283e-01 4.04045314e-01
-6.34039938e-02 6.37084484e-01 1.10229421e+00 -1.31017041e+00
-7.65167594e-01 -2.52658933e-01 4.49409068e-01 1.30748972e-01
-8.29703659e-02 3.71523976e-01 -6.41795218e-01 -4.72771883e-01
7.90738165e-02 -1.20002711e+00 -3.25044878e-02 9.86491889e-02
-4.76331860e-01 -2.73893893e-01 1.21139646e+00 -6.58539355e-01
1.50647008e+00 -2.08632994e+00 -1.64339039e-02 4.89285262e-03
2.30581656e-01 3.91710639e-01 -6.34404793e-02 5.46755314e-01
-1.17996559e-01 -1.83678284e-01 -9.65831578e-02 -1.98781967e-01
-2.61659145e-01 4.03300039e-02 -8.97237957e-02 6.67651176e-01
1.40321091e-01 1.02411306e+00 -1.12498164e+00 -4.90253121e-01
5.40766656e-01 2.77381778e-01 -5.71773887e-01 1.86537117e-01
-1.83052614e-01 6.96011305e-01 -1.05860531e-01 7.63463676e-01
3.52166951e-01 -4.23550844e-01 1.69721022e-01 -1.35702834e-01
-2.19644666e-01 1.69170853e-02 -1.23367739e+00 1.32021427e+00
-5.69636486e-02 8.62260342e-01 -2.30170384e-01 -1.09555936e+00
3.60267699e-01 4.70455498e-01 6.23500228e-01 -5.95001459e-01
-1.25635192e-01 -1.09615132e-01 2.79343218e-01 -5.96595824e-01
4.07992095e-01 3.75660360e-01 1.59562901e-01 2.66984582e-01
-1.81848481e-01 4.82842922e-01 4.56628501e-01 1.75830111e-01
1.43291962e+00 2.49896869e-01 7.56782651e-01 1.06578961e-01
4.63902980e-01 7.07495734e-02 6.20251656e-01 1.11146045e+00
-8.19255173e-01 9.25912023e-01 3.41701508e-01 -6.56866908e-01
-9.44917977e-01 -7.98020780e-01 2.71686882e-01 1.08983421e+00
4.49850559e-01 -7.46264338e-01 -6.35480106e-01 -8.29640448e-01
-2.39489049e-01 5.38026214e-01 -8.06425393e-01 -1.73286851e-02
-7.84030557e-01 -3.33460063e-01 -4.06901129e-02 5.88132560e-01
5.62052906e-01 -1.15088177e+00 -1.30137086e+00 1.57679364e-01
-5.33451796e-01 -1.45293653e+00 -7.85278738e-01 1.31268259e-02
-5.16790926e-01 -1.23023975e+00 -5.98266244e-01 -5.34968436e-01
4.56841499e-01 9.40531194e-01 9.81053889e-01 1.60348132e-01
-4.78425682e-01 4.65738624e-01 -4.93659347e-01 -1.45826444e-01
-1.45431235e-01 -4.90665168e-01 -3.20380204e-03 3.00287694e-01
4.55632597e-01 -3.18441659e-01 -9.16412055e-01 7.70325303e-01
-6.75906658e-01 2.29249045e-01 2.96735376e-01 8.61021459e-01
3.04332614e-01 1.03464290e-01 -4.51866677e-03 -4.39026773e-01
8.38475749e-02 -3.07562441e-01 -3.75566691e-01 5.03173530e-01
-4.88584116e-02 -2.87455618e-01 5.02321184e-01 -7.35545754e-01
-9.62881684e-01 3.72527897e-01 3.58928949e-01 -7.60196447e-01
-3.89736027e-01 2.91816026e-01 5.80882728e-02 1.12867700e-02
6.02553725e-01 2.80067265e-01 -3.83405030e-01 -3.13348360e-02
5.33390902e-02 3.00693840e-01 5.61986685e-01 -3.14431161e-01
6.17734075e-01 6.87337637e-01 -1.16366118e-01 -1.04879940e+00
-7.89015710e-01 -1.01997006e+00 -7.91524470e-01 -4.09497648e-01
9.76175606e-01 -6.77353263e-01 -9.22318757e-01 4.37951505e-01
-1.35271311e+00 -3.76967669e-01 2.02677269e-02 4.36373085e-01
-6.82228148e-01 6.83365703e-01 -3.71635616e-01 -1.15795970e+00
1.57200292e-01 -1.01942730e+00 1.22121406e+00 -5.95240034e-02
-4.01269615e-01 -5.43598711e-01 -2.33667329e-01 3.31428349e-01
6.36902684e-03 4.09318298e-01 3.99159312e-01 -3.98264021e-01
-8.06692839e-01 -2.17904165e-01 -2.32436016e-01 2.42438465e-02
-3.11231185e-02 1.17603861e-01 -8.78848553e-01 -3.33243310e-01
-3.12527269e-02 -2.49875605e-01 7.60026336e-01 4.07333195e-01
1.02123940e+00 -3.22571307e-01 -5.40776551e-01 3.68326008e-01
1.11713731e+00 5.83790064e-01 6.52411163e-01 1.67340189e-01
6.90087199e-01 6.45330191e-01 9.69693542e-01 6.68681920e-01
3.17076832e-01 1.10538661e+00 2.93075383e-01 8.64968970e-02
-4.23556864e-02 -2.11832881e-01 5.61129510e-01 3.00182700e-01
-4.63092148e-01 -4.48826730e-01 -1.17104924e+00 6.41354680e-01
-2.20855165e+00 -1.32422018e+00 -1.45819560e-01 2.38765049e+00
3.05332035e-01 1.21623561e-01 4.44387347e-01 3.57377976e-01
7.51370668e-01 1.92030489e-01 -4.50856835e-01 1.44352943e-01
-1.08847041e-02 -3.59622598e-01 2.18817219e-01 3.72090966e-01
-1.51122499e+00 8.15093756e-01 6.24348450e+00 3.73698831e-01
-8.68007898e-01 2.25979194e-01 2.71629214e-01 -5.40068783e-02
2.42830694e-01 1.73868179e-01 -7.91468740e-01 3.46950799e-01
7.30076373e-01 1.28404228e-02 3.88092101e-01 6.32546604e-01
5.21334529e-01 -5.25679708e-01 -1.45675588e+00 1.26859927e+00
2.51912743e-01 -1.03294230e+00 -8.80089924e-02 -1.50415394e-02
4.81953740e-01 -3.55713695e-01 -1.40886024e-01 1.97696626e-01
-3.78094725e-02 -9.28212702e-01 9.07716453e-01 3.22368473e-01
4.18178588e-01 -2.76723981e-01 4.31234181e-01 5.62060654e-01
-1.59852219e+00 -3.33112150e-01 4.81243990e-03 -2.08345532e-01
2.11403921e-01 8.62438604e-02 -7.42590129e-01 4.11603242e-01
8.79099905e-01 7.97542393e-01 -5.76133966e-01 1.14703751e+00
-1.00354388e-01 4.76468831e-01 -4.27669287e-01 1.64908171e-01
3.83885801e-01 1.07491173e-01 7.25185096e-01 1.07878625e+00
2.61308640e-01 3.77655238e-01 5.72321892e-01 5.42710423e-01
5.25110006e-01 -2.21277148e-01 -9.53757703e-01 -1.47880256e-01
1.74620807e-01 8.45527232e-01 -9.51349378e-01 -3.04759651e-01
-9.41400647e-01 1.45624769e+00 1.93536744e-01 6.01582587e-01
-1.08477104e+00 -1.02057248e-01 4.77169722e-01 3.23713094e-01
3.76996547e-01 -5.95541239e-01 3.43223989e-01 -1.38362396e+00
2.26917028e-01 -9.54015017e-01 6.14551723e-01 -7.73670375e-01
-7.79444695e-01 3.27530831e-01 2.50121325e-01 -1.54661596e+00
-4.82800007e-01 -5.45682490e-01 -6.19756043e-01 3.09068382e-01
-1.25495231e+00 -1.13933754e+00 -6.16053641e-01 8.55591953e-01
9.10702825e-01 4.38529909e-01 4.82937932e-01 6.90483898e-02
-5.47871888e-01 4.80527967e-01 -3.36968273e-01 5.60951792e-02
5.25944471e-01 -1.06029880e+00 2.89689511e-01 1.19907331e+00
4.11171705e-01 7.68002868e-01 9.40095127e-01 -5.51511467e-01
-1.54526126e+00 -8.81482601e-01 9.95254695e-01 -6.36738122e-01
5.59262037e-01 -4.80949223e-01 -9.34661984e-01 9.68074918e-01
1.40686095e-01 4.50125694e-01 1.03743978e-01 -1.16199479e-01
-1.81546509e-01 1.95707157e-01 -6.79346979e-01 7.48090327e-01
1.46969259e+00 -5.36207378e-01 -5.60429871e-01 4.69571114e-01
4.12964135e-01 -3.47265810e-01 -3.00134182e-01 4.43283379e-01
7.85396218e-01 -1.31283152e+00 1.05134916e+00 -7.01822698e-01
2.16641560e-01 -5.90005994e-01 -1.68792251e-02 -7.99446225e-01
-2.33690307e-01 -9.25472975e-01 -4.48918819e-01 7.77662337e-01
-7.47437552e-02 -1.33370265e-01 6.72522962e-01 4.89737213e-01
6.34880736e-02 -4.09916013e-01 -7.71353126e-01 -1.19225442e+00
-6.76148772e-01 -6.22029185e-01 1.42010733e-01 8.04223120e-01
2.43322611e-01 9.48987715e-03 -8.31860781e-01 3.23764741e-01
4.72921282e-01 2.42198199e-01 1.05193377e+00 -8.55191469e-01
-3.93169075e-01 -2.11560979e-01 -6.56829953e-01 -1.23301530e+00
-1.29469000e-02 -3.37835491e-01 3.63940239e-01 -1.30722964e+00
5.51220655e-01 1.56540319e-01 -4.13741022e-01 4.85662520e-01
-2.25621328e-01 2.01321885e-01 3.25415760e-01 3.70049387e-01
-1.00518453e+00 2.64879078e-01 1.17828894e+00 -1.20428644e-01
-1.82905883e-01 -1.12758547e-01 1.54992729e-01 8.64551425e-01
4.70447630e-01 -1.54945374e-01 -5.53117573e-01 4.41011973e-03
-1.67789713e-01 4.89575982e-01 8.18868577e-01 -1.30542862e+00
4.34522688e-01 -4.21135008e-01 2.63350189e-01 -7.38742769e-01
3.77937078e-01 -7.45447397e-01 1.42289177e-01 4.15611029e-01
-2.21241057e-01 1.59360990e-01 -8.94604772e-02 8.52825880e-01
-4.08271551e-01 -3.65038887e-02 6.13817692e-01 -2.60493159e-01
-1.33816755e+00 2.47978032e-01 -5.50637186e-01 -1.56886816e-01
1.51831055e+00 -5.65569401e-01 -2.81016827e-01 -5.41144609e-01
-8.44755113e-01 1.20694384e-01 5.18924773e-01 5.53165317e-01
7.71644950e-01 -9.99419272e-01 -3.56152654e-01 1.64579257e-01
3.96545649e-01 -4.95002955e-01 4.28582460e-01 1.17065728e+00
-2.85704583e-01 6.43756270e-01 -8.42259079e-02 -8.60434175e-01
-1.67480266e+00 1.01574433e+00 2.27092534e-01 -1.50642246e-01
-9.11859453e-01 6.26907945e-01 6.71632707e-01 1.98113665e-01
5.27326047e-01 -4.03698832e-01 -1.67587832e-01 -6.93188533e-02
7.70997286e-01 2.14700818e-01 -1.47625089e-01 -7.69258320e-01
-5.44023454e-01 3.37172806e-01 4.34280969e-02 -1.24105960e-01
7.02428043e-01 -3.34083468e-01 2.55128980e-01 7.48869836e-01
1.00320828e+00 -3.26910734e-01 -1.49837923e+00 -3.03531826e-01
1.63241804e-01 -7.60033488e-01 -2.13323385e-01 -4.60893750e-01
-6.60143316e-01 1.08704627e+00 4.67559487e-01 1.34859040e-01
1.04022646e+00 -4.30336259e-02 7.89641082e-01 4.49077636e-01
6.97330832e-01 -8.50351334e-01 4.69815642e-01 3.32487226e-01
9.26177979e-01 -1.54937768e+00 -1.45024121e-01 -4.90327120e-01
-8.00601542e-01 9.90944445e-01 8.49235475e-01 2.31906548e-01
4.74814415e-01 3.54466699e-02 -3.34867686e-01 -1.17204137e-01
-1.01179218e+00 -5.15048265e-01 3.97976518e-01 4.96039867e-01
3.04998130e-01 -6.50796741e-02 -2.16846198e-01 6.44460917e-02
4.14899319e-01 8.60357806e-02 3.55342388e-01 1.26280200e+00
-2.18381345e-01 -5.30668318e-01 -4.42135930e-01 1.47117183e-01
-2.25874156e-01 1.74287438e-01 -2.66541928e-01 9.99642611e-01
9.34679881e-02 1.07044899e+00 3.62802148e-02 -4.46739823e-01
2.75971711e-01 1.00951716e-01 5.66002071e-01 -4.99481499e-01
-2.36124277e-01 7.55201653e-02 4.15968103e-03 -1.19240141e+00
-7.62299895e-01 -9.76494491e-01 -9.06155288e-01 -1.46549284e-01
-1.30156159e-01 -2.44310796e-01 1.42380789e-01 9.66540992e-01
1.55109525e-01 2.51527131e-01 3.87635112e-01 -1.09398818e+00
-2.29839161e-01 -7.52329528e-01 -4.09645259e-01 8.33513439e-01
6.40546560e-01 -8.91392291e-01 -4.25191432e-01 3.93576890e-01]
|
[8.315956115722656, 0.46128374338150024]
|
a3b206f6-defd-4fde-ab49-e0ab54dd0441
|
playing-go-without-game-tree-search-using
|
1907.04658
| null |
https://arxiv.org/abs/1907.04658v1
|
https://arxiv.org/pdf/1907.04658v1.pdf
|
Playing Go without Game Tree Search Using Convolutional Neural Networks
|
The game of Go has a long history in East Asian countries, but the field of Computer Go has yet to catch up to humans until the past couple of years. While the rules of Go are simple, the strategy and combinatorics of the game are immensely complex. Even within the past couple of years, new programs that rely on neural networks to evaluate board positions still explore many orders of magnitude more board positions per second than a professional can. We attempt to mimic human intuition in the game by creating a convolutional neural policy network which, without any sort of tree search, should play the game at or above the level of most humans. We introduce three structures and training methods that aim to create a strong Go player: non-rectangular convolutions, which will better learn the shapes on the board, supervised learning, training on a data set of 53,000 professional games, and reinforcement learning, training on games played between different versions of the network. Our network has already surpassed the skill level of intermediate amateurs simply using supervised learning. Further training and implementation of non-rectangular convolutions and reinforcement learning will likely increase this skill level much further.
|
['Jeffrey Barratt', 'Chuanbo Pan']
|
2019-07-02
| null | null | null | null |
['game-of-go']
|
['playing-games']
|
[-1.35862932e-01 1.18822441e-01 4.20960099e-01 -1.79059997e-01
-3.13512951e-01 -7.81346381e-01 3.49334508e-01 -1.87900990e-01
-6.95592225e-01 5.54237843e-01 -1.34443611e-01 -1.04062176e+00
7.21933469e-02 -1.30250800e+00 -6.28196955e-01 -2.50029981e-01
-3.44266772e-01 4.94106442e-01 5.01804054e-01 -9.41522717e-01
3.33295017e-01 1.02393597e-01 -1.48433304e+00 3.87235403e-01
8.34014714e-01 1.13298774e+00 1.82137504e-01 1.18386507e+00
1.78801008e-02 1.25749600e+00 -6.63090527e-01 -5.29480398e-01
6.24937832e-01 -3.36321652e-01 -9.44492757e-01 -4.29348886e-01
1.90071285e-01 -3.89460385e-01 -2.40227014e-01 1.09532571e+00
3.89991790e-01 1.37851715e-01 1.58089340e-01 -7.75521994e-01
-4.74329501e-01 6.79819465e-01 -2.36294493e-01 4.00484860e-01
2.55659014e-01 5.77772856e-01 1.04404557e+00 -4.43781137e-01
2.29538023e-01 8.89035463e-01 1.05166256e+00 3.31225067e-01
-9.25494969e-01 -6.71467721e-01 -2.37174690e-01 -1.09260984e-01
-1.24206352e+00 -1.01051256e-02 3.45040679e-01 -5.06562233e-01
1.15226114e+00 -1.15081649e-02 1.27803314e+00 5.12127101e-01
1.40274286e-01 4.77927595e-01 1.05348051e+00 -5.52375495e-01
1.09396435e-01 -3.14425141e-01 -4.70360696e-01 9.67024922e-01
3.81288771e-03 5.68563461e-01 -1.69986188e-01 5.34837842e-02
1.39218998e+00 -3.12608689e-01 9.87235680e-02 -1.92776278e-01
-9.97647405e-01 1.13014662e+00 4.49987113e-01 4.18059051e-01
-1.69184521e-01 4.35438335e-01 2.65680194e-01 4.44279790e-01
-1.32240718e-02 1.14041245e+00 -6.73850417e-01 -6.08065009e-01
-1.05581975e+00 6.01413667e-01 1.13813949e+00 5.85044086e-01
6.73921525e-01 2.11329535e-01 4.03789610e-01 5.02857149e-01
-2.89835472e-04 1.06257372e-01 4.17366803e-01 -1.19608843e+00
2.54723161e-01 6.00907922e-01 1.33250970e-02 -1.08482671e+00
-5.11130452e-01 -4.06868696e-01 -5.28846145e-01 8.95254433e-01
1.20276129e+00 -7.57148027e-01 -7.19280779e-01 1.28208816e+00
-8.08052644e-02 -1.45263687e-01 -3.24619770e-01 7.63220489e-01
4.97764677e-01 7.87654042e-01 -2.28052363e-01 5.61174989e-01
1.19851840e+00 -8.77287388e-01 8.38549659e-02 -4.19553220e-01
8.41274023e-01 -6.72582626e-01 1.26233757e+00 9.30591941e-01
-1.48674273e+00 -6.52782381e-01 -1.23248327e+00 1.97158262e-01
-4.12773132e-01 -2.23931208e-01 1.22838604e+00 1.13485563e+00
-1.27774262e+00 1.04135358e+00 -8.67454708e-01 -2.04826757e-01
3.51465940e-01 7.73732781e-01 -9.78097990e-02 2.39025056e-01
-1.33176446e+00 1.10919619e+00 6.48333549e-01 3.79652977e-02
-6.77428603e-01 -6.23705745e-01 -8.28663647e-01 2.00541466e-01
6.33887529e-01 -3.91126513e-01 1.61269867e+00 -1.34898317e+00
-1.77711809e+00 9.35715973e-01 7.09005654e-01 -5.01295567e-01
5.20505786e-01 -8.52173716e-02 -1.65302426e-01 -1.25063568e-01
-1.87767684e-01 6.87922478e-01 2.96772778e-01 -5.87221026e-01
-9.79711592e-01 -5.17132394e-02 6.86152160e-01 3.17584842e-01
-3.71253453e-02 1.93700626e-01 -4.08211410e-01 -5.73368430e-01
-1.83246046e-01 -8.55001330e-01 -6.00920439e-01 -2.05036372e-01
3.08171064e-01 -2.74520576e-01 -6.48772940e-02 -5.75298071e-01
1.27064323e+00 -1.86179245e+00 -1.64529935e-01 4.42392737e-01
1.82527855e-01 3.09585571e-01 6.58336356e-02 1.84045777e-01
-2.37716675e-01 1.01749517e-01 2.06386372e-01 6.46794140e-01
-3.32823247e-02 1.86061293e-01 1.46863952e-01 2.11150929e-01
-1.39469430e-02 8.97905946e-01 -1.19470942e+00 -9.51570421e-02
7.07210228e-02 -1.83729127e-01 -1.21090293e+00 -9.93188191e-03
-3.28780293e-01 1.18813440e-01 -2.39465311e-01 3.17537099e-01
2.24405274e-01 -2.49317631e-01 2.64301956e-01 5.77678978e-01
-2.20935300e-01 5.85860670e-01 -1.22894275e+00 1.68141913e+00
-3.75282049e-01 5.84137917e-01 8.82671848e-02 -1.06376338e+00
8.32671404e-01 3.65093984e-02 3.27921778e-01 -9.07402456e-01
3.20545822e-01 3.91573012e-01 7.64920771e-01 -3.51787299e-01
7.36890495e-01 -3.79987091e-01 -2.93962151e-01 4.72380459e-01
-1.08239517e-01 -6.20882750e-01 4.20232713e-01 -1.11777768e-01
1.25124621e+00 2.98665106e-01 1.51314989e-01 -3.36141020e-01
1.11708403e-01 4.24177706e-01 4.08582956e-01 1.08180571e+00
-1.34287298e-01 3.00693363e-01 7.93292046e-01 -1.06208980e+00
-1.14200902e+00 -8.16218436e-01 4.71105099e-01 1.76523566e+00
-2.63761610e-01 -3.02790076e-01 -9.69293654e-01 -2.94022650e-01
-2.85100907e-01 4.56622481e-01 -4.13614124e-01 -4.73832600e-02
-7.50154555e-01 -5.74451208e-01 7.60307312e-01 5.85966706e-01
7.85555840e-01 -1.48953843e+00 -9.08417642e-01 4.79351372e-01
3.34416896e-01 -7.07486689e-01 -3.30188394e-01 5.45203567e-01
-7.35402942e-01 -1.06379938e+00 -6.44467294e-01 -9.84836161e-01
2.85308957e-01 -2.18573123e-01 1.51056683e+00 3.90764475e-01
-2.29690239e-01 -1.02646269e-01 -2.57725805e-01 -5.39526224e-01
-4.13764566e-01 3.26361030e-01 -1.34085342e-01 -8.16993654e-01
4.10277665e-01 -6.94223762e-01 -5.79903841e-01 2.16530427e-01
-6.08538449e-01 2.19419435e-01 6.05019271e-01 9.23747361e-01
-2.57277131e-01 4.23157066e-01 5.56369387e-02 -9.25080597e-01
7.92468190e-01 -7.43509904e-02 -9.13677156e-01 -1.46308079e-01
-2.93194920e-01 1.07679248e-01 6.23599529e-01 -4.13866431e-01
-6.13287508e-01 1.23625975e-02 -3.74835670e-01 1.59907624e-01
2.07581386e-01 6.91707551e-01 1.85796678e-01 -2.05665514e-01
1.15540910e+00 -1.64002970e-01 8.92758183e-03 -1.22973897e-01
3.67132425e-01 3.94494921e-01 7.07395494e-01 -9.61044133e-01
7.41226196e-01 7.58586749e-02 -4.37960148e-01 -5.70094287e-01
-4.55246001e-01 -7.39100948e-02 -3.48622382e-01 -1.89631835e-01
8.39670956e-01 -7.31579661e-01 -1.14441395e+00 7.57163823e-01
-9.56846297e-01 -1.08540213e+00 -2.94852108e-01 2.28861168e-01
-5.85269272e-01 -8.97298083e-02 -8.07530880e-01 -4.40475672e-01
5.85101582e-02 -1.03063643e+00 2.67335206e-01 5.18341959e-01
-3.42403889e-01 -9.53137398e-01 -4.43123339e-04 1.74706504e-01
6.27517164e-01 1.02509312e-01 7.77907252e-01 -4.36708152e-01
-4.65273201e-01 -3.04354727e-01 -5.79052381e-02 3.90485436e-01
-1.82144791e-01 -1.32576779e-01 -4.72729057e-01 3.58075947e-02
-1.87566638e-01 -3.99678767e-01 5.62302291e-01 4.47525114e-01
1.18146145e+00 -5.77218458e-02 9.29706693e-02 5.86880803e-01
1.17791629e+00 6.60951316e-01 1.01869667e+00 6.63822711e-01
4.25724506e-01 3.65585893e-01 1.43256783e-01 1.95172161e-01
3.67545426e-01 3.91012669e-01 4.41434890e-01 -1.38308987e-01
3.24847490e-01 -3.89356941e-01 2.61586368e-01 6.62731528e-01
-7.16093898e-01 2.24774152e-01 -1.29486549e+00 3.92169595e-01
-1.52887511e+00 -9.88219678e-01 2.55704165e-01 2.08920884e+00
9.16324258e-01 1.09002388e+00 5.34006476e-01 1.37679040e-01
4.94098067e-01 -8.06919262e-02 -2.34014556e-01 -9.94602084e-01
2.76454300e-01 9.06362236e-01 7.95012891e-01 4.53224540e-01
-1.07874334e+00 1.29696298e+00 7.61636305e+00 8.90649974e-01
-1.17930388e+00 -3.66943777e-01 7.66855121e-01 1.35387599e-01
7.43152574e-02 1.02448173e-01 -4.69627649e-01 2.46755153e-01
8.28719318e-01 1.27241267e-02 1.10193110e+00 1.05031967e+00
-8.03861488e-03 -2.70470113e-01 -8.51050794e-01 8.90390873e-01
-3.10997277e-01 -1.68047583e+00 -3.92363310e-01 2.55751580e-01
8.06178451e-01 1.07805431e-01 4.92978655e-02 8.74925554e-01
1.33280909e+00 -1.74189997e+00 7.50000834e-01 2.46680796e-01
6.37100399e-01 -8.65836143e-01 5.75752556e-01 5.21007419e-01
-9.38509166e-01 -2.59669244e-01 -4.08619672e-01 -1.09629548e+00
-4.80243802e-01 -1.56053342e-02 -6.33672476e-01 5.57083264e-02
8.09403718e-01 2.70473629e-01 -5.23407400e-01 1.10272789e+00
-3.57927829e-01 5.20421684e-01 -5.53322673e-01 -5.71401358e-01
7.75568128e-01 -1.22869700e-01 -1.07556984e-01 9.04525816e-01
2.13100657e-01 4.58806366e-01 1.22539610e-01 5.99359453e-01
2.01497730e-02 -1.23428173e-01 -5.17727971e-01 -3.33386064e-01
1.46364883e-01 1.05665267e+00 -9.29446578e-01 -1.65425077e-01
-5.66402078e-01 6.23162270e-01 3.27759415e-01 -2.37882859e-03
-7.38660634e-01 -6.75989747e-01 4.32974607e-01 3.95901710e-01
5.78769088e-01 -4.55357492e-01 -4.72927392e-01 -7.93469489e-01
-3.74032378e-01 -1.50911593e+00 3.31836373e-01 -8.11128020e-01
-9.19131875e-01 3.48950237e-01 -2.98491031e-01 -8.25609803e-01
-4.74655628e-01 -1.30560815e+00 -1.06361473e+00 8.79316807e-01
-7.51325548e-01 -5.80782294e-01 2.80194711e-02 4.23252523e-01
2.69447029e-01 -3.22957516e-01 7.57697999e-01 1.86269835e-01
-4.17780317e-03 5.37038863e-01 -1.21848099e-01 7.19810009e-01
1.40410319e-01 -1.40739489e+00 7.26873696e-01 4.47493523e-01
2.25730315e-01 4.92381066e-01 5.87386131e-01 -1.58342957e-01
-1.15022981e+00 -2.73651838e-01 4.78586614e-01 -3.67995113e-01
9.74615633e-01 -3.49685580e-01 -4.17103767e-01 5.78042328e-01
8.69683474e-02 -1.47864372e-01 3.05858135e-01 4.77390707e-01
-1.94952920e-01 1.00609679e-02 -6.49394393e-01 8.08818698e-01
8.63684773e-01 -4.63762611e-01 -6.44505382e-01 -3.17154303e-02
2.04705685e-01 -8.84241462e-01 -4.60765719e-01 1.45139098e-01
8.44089329e-01 -1.23489082e+00 9.66543853e-01 -8.01312506e-01
7.98695743e-01 -7.89068937e-02 9.80467871e-02 -1.43594110e+00
-3.09070289e-01 -9.90905881e-01 5.52482545e-01 3.98174644e-01
5.51874161e-01 -3.82656187e-01 1.38622713e+00 4.89119709e-01
-6.12697080e-02 -8.61350715e-01 -6.32460237e-01 -5.92999518e-01
8.37990105e-01 -7.08663762e-01 5.26357532e-01 7.93794513e-01
4.23632443e-01 2.19570190e-01 -1.75631538e-01 -2.53070384e-01
4.85073254e-02 -9.08649117e-02 8.79201651e-01 -1.09137201e+00
-9.35975313e-01 -1.08561683e+00 -4.93652135e-01 -1.18919528e+00
-4.25069928e-01 -9.04861987e-01 1.13263510e-01 -1.27979517e+00
-2.85336822e-01 -4.65733111e-01 2.29256530e-03 5.12519121e-01
-4.20752540e-02 3.07083726e-01 2.00280741e-01 -3.13751012e-01
-5.23004651e-01 -1.71723530e-01 1.39760756e+00 -3.42007130e-02
-3.41032982e-01 5.76787535e-03 -1.15348983e+00 1.32786202e+00
7.82707155e-01 -2.88226128e-01 -1.35256901e-01 -4.01169777e-01
9.86536920e-01 1.21064104e-01 2.51277387e-01 -1.37883174e+00
4.14002717e-01 -1.80091590e-01 6.19570851e-01 -7.74998367e-02
-7.85410926e-02 -5.17692506e-01 1.54239926e-02 7.12929606e-01
-1.32022783e-01 2.13088989e-01 5.07992566e-01 -4.59619612e-02
4.62810397e-02 -4.22534734e-01 7.17892945e-01 -7.80656338e-01
-7.94738114e-01 6.07260596e-03 -8.46229017e-01 3.86152685e-01
7.66771078e-01 -6.40275598e-01 1.21367306e-01 -7.05044866e-01
-9.34487522e-01 3.52107584e-01 4.34471458e-01 9.66156945e-02
8.30619037e-02 -9.73457873e-01 -4.74773169e-01 6.81752190e-02
-3.57399255e-01 1.95154414e-01 -1.02550410e-01 4.07539964e-01
-1.36993563e+00 1.35722309e-01 -6.37105227e-01 -1.36564016e-01
-9.00979459e-01 1.97010517e-01 7.76296318e-01 -7.68339157e-01
-5.32221556e-01 1.04866064e+00 3.14002782e-02 -7.18173087e-01
7.33211711e-02 -4.39706177e-01 -8.88365954e-02 -2.20783859e-01
5.13292253e-01 1.52863756e-01 2.35290546e-02 4.91209924e-02
-1.78489480e-02 2.44862363e-01 5.81003577e-02 -2.09920645e-01
1.55938685e+00 6.01379812e-01 1.19024895e-01 1.71407908e-01
5.36996543e-01 -4.64317501e-02 -1.44661045e+00 1.25039741e-01
4.09022607e-02 -2.05004796e-01 -3.70015055e-02 -8.85448754e-01
-9.40132022e-01 1.03126681e+00 2.72377133e-01 5.12974977e-01
9.09298897e-01 -4.60425705e-01 6.53286159e-01 6.29084766e-01
6.09224379e-01 -1.38183224e+00 4.30490226e-01 1.07991779e+00
4.41408336e-01 -9.63076890e-01 -9.86630395e-02 -1.16922654e-01
-4.81489927e-01 1.36098909e+00 6.07251823e-01 -7.09431648e-01
7.23442376e-01 5.76145828e-01 -1.13911234e-01 -3.07076544e-01
-5.14758170e-01 -3.44390094e-01 -1.02561802e-01 6.78261936e-01
5.09290278e-01 1.68604761e-01 -1.35573655e-01 7.20024168e-01
-9.34820592e-01 2.91938245e-01 5.75906336e-01 9.35634971e-01
-7.75195897e-01 -9.97501850e-01 -4.92407411e-01 3.91770720e-01
-6.69030428e-01 -4.14759248e-01 -7.75718540e-02 1.01198494e+00
4.70719486e-01 7.91524827e-01 2.94782132e-01 -5.65791309e-01
3.87224525e-01 -2.57784035e-02 7.37550080e-01 -7.36311555e-01
-9.98327851e-01 -1.18747175e-01 2.35423967e-01 -4.16736811e-01
8.18819478e-02 -3.59594643e-01 -1.20995104e+00 -1.09315264e+00
-1.24669760e-01 1.67095214e-01 2.78893799e-01 9.10063982e-01
-3.20317656e-01 4.36182261e-01 2.04977840e-01 -1.05308235e+00
-5.91047764e-01 -9.15242910e-01 -7.37396657e-01 -1.22163862e-01
-3.25420409e-01 -3.18315923e-01 -7.92688057e-02 -1.51714653e-01]
|
[3.4598724842071533, 1.437034010887146]
|
e8096404-9dbe-47f2-b8ac-e9f8acfe79e5
|
revisiting-end-to-end-speech-to-text
|
2206.04571
| null |
https://arxiv.org/abs/2206.04571v1
|
https://arxiv.org/pdf/2206.04571v1.pdf
|
Revisiting End-to-End Speech-to-Text Translation From Scratch
|
End-to-end (E2E) speech-to-text translation (ST) often depends on pretraining its encoder and/or decoder using source transcripts via speech recognition or text translation tasks, without which translation performance drops substantially. However, transcripts are not always available, and how significant such pretraining is for E2E ST has rarely been studied in the literature. In this paper, we revisit this question and explore the extent to which the quality of E2E ST trained on speech-translation pairs alone can be improved. We reexamine several techniques proven beneficial to ST previously, and offer a set of best practices that biases a Transformer-based E2E ST system toward training from scratch. Besides, we propose parameterized distance penalty to facilitate the modeling of locality in the self-attention model for speech. On four benchmarks covering 23 languages, our experiments show that, without using any transcripts or pretraining, the proposed system reaches and even outperforms previous studies adopting pretraining, although the gap remains in (extremely) low-resource settings. Finally, we discuss neural acoustic feature modeling, where a neural model is designed to extract acoustic features from raw speech signals directly, with the goal to simplify inductive biases and add freedom to the model in describing speech. For the first time, we demonstrate its feasibility and show encouraging results on ST tasks.
|
['Rico Sennrich', 'Barry Haddow', 'Biao Zhang']
|
2022-06-09
| null | null | null | null |
['speech-to-text-translation']
|
['natural-language-processing']
|
[ 5.81387043e-01 3.72235090e-01 -1.35112286e-01 -5.16205728e-01
-1.35572112e+00 -6.19775593e-01 6.19044363e-01 -5.44268906e-01
-3.51033926e-01 7.39683986e-01 5.55792928e-01 -7.77018785e-01
4.91987675e-01 -1.89112484e-01 -1.11098731e+00 -5.74789047e-01
3.16240817e-01 4.80038702e-01 -1.67369276e-01 -2.85508811e-01
-2.90608495e-01 1.81197733e-01 -1.10581458e+00 4.49364513e-01
7.71035194e-01 8.16479027e-01 3.92818063e-01 6.77335262e-01
-2.28910521e-01 6.05958104e-01 -6.67284310e-01 -5.93514442e-01
3.17772537e-01 -6.73234522e-01 -8.81424546e-01 6.01494350e-02
3.97724539e-01 -3.04434568e-01 -2.92107493e-01 7.91900575e-01
6.75343692e-01 2.43435577e-02 5.93412817e-01 -9.13300037e-01
-9.52125072e-01 1.07787526e+00 -1.89606354e-01 2.28453085e-01
7.55483434e-02 1.57699287e-01 1.11265945e+00 -1.32636476e+00
5.59583366e-01 1.27083814e+00 4.90319490e-01 6.22102261e-01
-1.26594019e+00 -4.92822498e-01 9.42305923e-02 6.11871146e-02
-1.18657279e+00 -1.27603090e+00 5.63780844e-01 -9.18404385e-02
1.63134694e+00 1.99572310e-01 2.84476072e-01 1.62290025e+00
1.16538681e-01 1.11907470e+00 1.01450944e+00 -6.27593935e-01
-1.04266526e-02 1.98743910e-01 -2.37883940e-01 4.95776892e-01
-3.54058981e-01 2.23868310e-01 -7.83142149e-01 1.22386664e-01
5.07478714e-01 -5.00181794e-01 -3.38094532e-01 1.12045817e-02
-1.33938658e+00 6.66051328e-01 3.80836964e-01 4.07222271e-01
-1.90566644e-01 1.53334260e-01 5.46674550e-01 7.74954319e-01
6.77117884e-01 3.29473704e-01 -6.37054920e-01 -4.54422176e-01
-9.76135015e-01 -4.83222663e-01 5.94161510e-01 1.17807865e+00
5.90760469e-01 3.98162037e-01 -2.04440281e-01 1.02873158e+00
9.96952802e-02 9.09424305e-01 6.44492090e-01 -6.16767645e-01
9.41247046e-01 -1.36883080e-01 -2.19238415e-01 -3.96772742e-01
-9.16654803e-03 -8.15505922e-01 -8.01154256e-01 -3.60379308e-01
1.26058921e-01 -3.79721284e-01 -8.95622373e-01 1.85759056e+00
-1.19485207e-01 1.36972100e-01 3.01509142e-01 8.16883802e-01
5.85022748e-01 8.66011918e-01 -2.78575897e-01 -2.98757344e-01
9.17236626e-01 -1.46655703e+00 -8.83655429e-01 -6.08476043e-01
8.74615014e-01 -9.84510958e-01 1.38016295e+00 1.46466821e-01
-1.33010876e+00 -5.67105770e-01 -7.71997333e-01 -2.83220947e-01
-4.05540131e-02 4.10696208e-01 2.68802315e-01 6.12814069e-01
-1.41236854e+00 4.71043736e-01 -1.12358296e+00 -5.86936414e-01
1.06474966e-01 5.32485902e-01 -2.45653704e-01 1.34174362e-01
-1.39634883e+00 1.08940327e+00 2.36695129e-02 2.07727119e-01
-8.38013887e-01 -4.20263052e-01 -8.30977082e-01 1.08572848e-01
2.42887273e-01 -8.67164612e-01 1.53469098e+00 -1.40622032e+00
-2.13511610e+00 4.63904142e-01 -6.88241780e-01 -5.82908452e-01
3.48898113e-01 -2.70241767e-01 -4.17741448e-01 -6.68279901e-02
-7.48789757e-02 7.93040514e-01 9.00747240e-01 -9.81393874e-01
-3.22745860e-01 -3.42052206e-02 -1.30535066e-01 3.87016833e-01
-4.97337043e-01 2.50790060e-01 -5.36864221e-01 -6.63946331e-01
-1.19137563e-01 -1.07918894e+00 6.29158691e-03 -3.99146557e-01
-3.63410711e-01 -8.11047927e-02 5.65191507e-01 -7.66707122e-01
1.06417656e+00 -2.17951274e+00 3.47763926e-01 -7.28266537e-02
-1.54235825e-01 2.74904579e-01 -5.13791919e-01 7.01205432e-01
-4.01819013e-02 1.72435269e-01 -3.43091458e-01 -8.13293099e-01
8.65366906e-02 3.30161750e-01 -5.98521829e-01 2.05357760e-01
4.68929589e-01 1.29796517e+00 -6.05141103e-01 -2.56574929e-01
-9.62018315e-03 5.92832208e-01 -5.02345383e-01 1.23736970e-01
-7.59429634e-02 5.77331543e-01 -1.71421722e-01 3.90566379e-01
2.25885659e-01 -3.21858197e-01 1.51856139e-01 3.45141366e-02
-4.30505089e-02 1.18681681e+00 -3.93160164e-01 1.79728210e+00
-9.92285192e-01 9.47581768e-01 5.63102402e-02 -1.00244021e+00
8.55672002e-01 5.39521754e-01 1.03373520e-01 -8.22706103e-01
1.17296852e-01 5.17372131e-01 1.95698798e-01 -2.85484195e-01
4.50280279e-01 -3.04893404e-01 1.23341434e-01 5.03136277e-01
1.66771501e-01 -7.09482804e-02 -2.47784480e-01 -1.83112975e-02
1.01365912e+00 1.21942222e-01 6.39725523e-03 -1.40076289e-02
2.97100902e-01 -1.21264227e-01 3.18401873e-01 6.49047494e-01
3.65829691e-02 6.43143177e-01 1.31357208e-01 2.42355019e-02
-1.22367167e+00 -9.49293435e-01 4.09776829e-02 1.30762887e+00
-3.07851225e-01 -2.67084420e-01 -8.36699963e-01 -8.12877774e-01
-4.19198573e-01 8.39998901e-01 -2.91760892e-01 -1.83308765e-01
-7.94895291e-01 -4.83041495e-01 8.45052183e-01 6.67669475e-01
2.55257964e-01 -9.78055835e-01 -2.78846044e-02 3.84798080e-01
-6.42991602e-01 -1.38108635e+00 -8.01346004e-01 4.86944288e-01
-1.00408626e+00 -1.17043309e-01 -8.70120466e-01 -8.50679934e-01
5.05540013e-01 3.12299728e-01 1.14722908e+00 -1.92438632e-01
5.41501701e-01 3.79821807e-02 -4.35332298e-01 -1.10517398e-01
-7.29142189e-01 6.23009145e-01 1.90788835e-01 -6.06100000e-02
3.24230015e-01 -5.82491159e-01 -1.84860840e-01 3.61706585e-01
-5.27470767e-01 2.25724250e-01 1.01609921e+00 1.03057277e+00
3.33003998e-01 -5.56288600e-01 6.74572170e-01 -6.12649381e-01
4.51001018e-01 -3.25218499e-01 -2.43609920e-01 2.08735123e-01
-4.40874636e-01 3.01685095e-01 8.78059149e-01 -3.65342468e-01
-8.79410267e-01 -1.08986475e-01 -6.21545672e-01 -5.82571328e-01
1.16045825e-01 5.65635681e-01 -1.57841697e-01 1.21081121e-01
5.45876384e-01 5.87524951e-01 -5.35096377e-02 -5.67034245e-01
3.57952535e-01 1.07143497e+00 2.86171347e-01 -5.93075573e-01
7.19721079e-01 8.32547471e-02 -4.92974699e-01 -7.77191281e-01
-7.76454270e-01 -2.30776846e-01 -5.89823544e-01 2.11353213e-01
5.52953482e-01 -1.03588283e+00 -2.16491818e-01 4.22197841e-02
-1.38474095e+00 -6.66986585e-01 -2.13690877e-01 6.50735557e-01
-7.12869942e-01 2.51639843e-01 -8.26895714e-01 -7.08932817e-01
-4.29151714e-01 -1.36410964e+00 1.31992388e+00 -5.53734839e-01
-1.95263147e-01 -9.39257622e-01 -8.59037712e-02 3.99670959e-01
6.98348701e-01 -5.60399592e-01 8.36989105e-01 -7.01568544e-01
-6.03191137e-01 3.56910706e-01 -1.30448109e-02 6.33379817e-01
1.57835796e-01 -2.86427081e-01 -1.17428446e+00 -4.35348243e-01
1.63248405e-02 -2.30224043e-01 8.60633612e-01 2.80696273e-01
8.19790900e-01 -6.04606569e-01 -1.48870051e-01 6.39577091e-01
9.73311067e-01 3.86187322e-02 4.72044051e-01 6.18405268e-02
5.75846791e-01 5.77259004e-01 3.21199775e-01 -7.35119283e-02
3.95681530e-01 9.34163272e-01 5.91189824e-02 -3.18823010e-01
-5.08840144e-01 -3.98611814e-01 1.07258224e+00 1.72827399e+00
1.53570130e-01 -5.86585104e-01 -8.45694304e-01 6.53542459e-01
-1.59480131e+00 -6.84386313e-01 1.34613648e-01 2.05979705e+00
9.84152555e-01 1.87338293e-01 1.67101268e-02 -7.01821819e-02
6.17056727e-01 1.04611747e-01 -5.17047822e-01 -6.90297008e-01
-2.59210408e-01 3.26976150e-01 4.99111354e-01 7.10847437e-01
-7.62592435e-01 1.42700648e+00 6.77639341e+00 9.16440904e-01
-1.47749949e+00 5.30893743e-01 5.74493766e-01 -2.60936737e-01
-4.54354525e-01 -1.69560071e-02 -7.59273410e-01 2.20267341e-01
1.50977266e+00 -4.74178381e-02 7.14837134e-01 6.38551533e-01
3.29161942e-01 6.04976356e-01 -1.43211782e+00 8.31341445e-01
5.70266210e-02 -1.09883714e+00 1.06447183e-01 8.51176083e-02
7.16585398e-01 6.81255996e-01 2.88052857e-01 5.05279541e-01
4.07075405e-01 -1.01287615e+00 9.04569387e-01 2.93584056e-02
1.07757437e+00 -5.10897934e-01 6.48164213e-01 4.28746283e-01
-1.00391448e+00 3.51864137e-02 -3.82781744e-01 -6.09246269e-02
2.68466085e-01 4.65578437e-01 -1.23245716e+00 6.27616405e-01
3.53972435e-01 6.88969195e-01 -2.42424652e-01 4.35551673e-01
-3.22436571e-01 1.12368989e+00 -3.78681391e-01 -2.12326020e-01
5.61064303e-01 6.90867659e-03 5.59369743e-01 1.47029221e+00
7.03133702e-01 -2.27208599e-01 -5.19595779e-02 7.16586709e-01
-3.09220701e-01 2.57033229e-01 -7.11703062e-01 -2.06185058e-01
4.53447878e-01 7.04025805e-01 -2.30713814e-01 -4.11314040e-01
-6.25933886e-01 1.25971985e+00 4.62077677e-01 6.79857433e-01
-8.02378833e-01 -8.32784399e-02 8.19290578e-01 1.08976625e-02
4.57124382e-01 -4.59840953e-01 -3.31231892e-01 -1.37314057e+00
2.36299619e-01 -1.14521432e+00 -1.97084650e-01 -7.09470391e-01
-1.05321455e+00 1.10233998e+00 -4.38556939e-01 -1.05313373e+00
-6.58393383e-01 -3.92932445e-01 -2.50319302e-01 1.00975382e+00
-1.59968829e+00 -1.17219162e+00 4.38762575e-01 4.11552459e-01
9.97443259e-01 -2.06735969e-01 7.59215534e-01 4.69117105e-01
-5.73360205e-01 1.00706291e+00 4.10823822e-01 1.49999171e-01
8.06553721e-01 -9.61082578e-01 1.05204058e+00 9.17269051e-01
5.38326383e-01 7.63063312e-01 6.30359113e-01 -4.25388902e-01
-1.54359460e+00 -1.19489217e+00 1.35877490e+00 -6.07511044e-01
6.76872075e-01 -8.19229543e-01 -8.19108248e-01 9.87060547e-01
4.89765167e-01 -1.17639996e-01 4.67597187e-01 2.99425781e-01
-2.99994111e-01 6.14987276e-02 -6.25322521e-01 8.15956652e-01
1.36448026e+00 -9.16993141e-01 -5.88079333e-01 2.35427663e-01
1.07971501e+00 -3.67063850e-01 -6.11623943e-01 2.83239752e-01
3.90100449e-01 -5.84005177e-01 8.14854383e-01 -5.45511246e-01
4.69434410e-01 -2.41972413e-02 -3.89271438e-01 -1.62862420e+00
-1.80968940e-01 -9.76897955e-01 9.43370461e-02 1.20113993e+00
9.09290254e-01 -7.35232711e-01 5.40623188e-01 2.06142858e-01
-7.07940400e-01 -7.43647516e-01 -1.15004253e+00 -1.14554560e+00
4.41933274e-01 -5.96259832e-01 6.97191179e-01 8.57576966e-01
-6.44567385e-02 7.25099206e-01 -5.47574043e-01 1.55117676e-01
9.06684399e-02 -6.65288791e-02 7.57712543e-01 -5.53168118e-01
-5.99758208e-01 -5.06323278e-01 4.02415358e-02 -1.84363019e+00
4.62301970e-01 -1.26630652e+00 2.14744255e-01 -1.42989492e+00
-7.82048609e-03 -3.18414599e-01 -1.18067674e-01 5.97779334e-01
-9.67124701e-02 2.51060247e-01 1.96185231e-01 3.26130092e-01
-2.73722798e-01 9.87978995e-01 1.20387733e+00 -1.35720581e-01
-9.20595750e-02 -1.89415544e-01 -5.16539633e-01 3.09188962e-01
6.88165367e-01 -4.99571800e-01 -4.30008501e-01 -1.07581556e+00
5.45591265e-02 2.09037468e-01 5.58621138e-02 -6.73812449e-01
1.85495645e-01 1.62915185e-01 -8.52090344e-02 -3.56540442e-01
5.52371144e-01 -7.28573442e-01 -4.31539342e-02 1.36291951e-01
-6.07340276e-01 9.95729566e-02 1.38097674e-01 2.58407027e-01
-3.95055085e-01 -1.77396908e-02 5.05691946e-01 2.02649802e-01
-2.16213480e-01 1.23473898e-01 -4.93995249e-01 7.92607069e-02
3.09117824e-01 8.86960700e-03 -1.64899155e-01 -6.90988362e-01
-6.25531912e-01 -9.37720835e-02 3.11343253e-01 5.93598902e-01
4.41792071e-01 -1.33136177e+00 -9.12858188e-01 2.93807447e-01
9.12383348e-02 -3.58411759e-01 -1.65847525e-01 1.06220162e+00
9.21773091e-02 7.60484219e-01 2.53625154e-01 -6.96675897e-01
-1.10148370e+00 4.64509130e-01 3.36977452e-01 -1.72926396e-01
-6.46283805e-01 8.92463565e-01 3.24533463e-01 -7.20192134e-01
2.21034884e-01 -6.02416515e-01 3.71249110e-01 -3.52502406e-01
1.70158744e-01 -1.50167942e-01 4.39669222e-01 -8.05908740e-01
-3.37877274e-01 4.10620958e-01 -1.32294685e-01 -5.42386591e-01
1.21034348e+00 -5.68086565e-01 2.37173975e-01 5.94443619e-01
1.36397433e+00 1.02685936e-01 -1.10656965e+00 -5.75017214e-01
-4.03654901e-03 -1.97781637e-01 8.16908330e-02 -8.70778501e-01
-1.05272830e+00 1.34929430e+00 2.39283025e-01 -1.28077775e-01
1.09745753e+00 4.15201252e-03 1.22717154e+00 7.23948479e-01
3.51698279e-01 -9.56029534e-01 -1.00586832e-01 9.60288107e-01
1.00483799e+00 -1.19353735e+00 -7.65159309e-01 -3.63965720e-01
-8.61102402e-01 9.15657938e-01 1.49892360e-01 1.93360746e-01
3.29230309e-01 5.80082715e-01 2.74233609e-01 3.48382175e-01
-1.07288873e+00 -2.73183465e-01 3.01818699e-01 4.23656493e-01
7.73753583e-01 1.69431437e-02 -1.54394461e-02 4.23108876e-01
-6.21055901e-01 -1.82901964e-01 2.59015471e-01 6.11022353e-01
-2.85187453e-01 -1.30618727e+00 -2.61803001e-01 1.90695122e-01
-5.89552939e-01 -7.19410002e-01 -6.17583871e-01 4.58242774e-01
-3.53193879e-01 1.16457224e+00 -5.08601256e-02 -4.91185129e-01
2.27883220e-01 3.72987747e-01 3.41379136e-01 -6.60941780e-01
-7.73457646e-01 3.72100592e-01 4.55620468e-01 -4.14638132e-01
-1.22988425e-01 -5.97769618e-01 -1.07653093e+00 -2.40235656e-01
-4.80300963e-01 2.39502400e-01 9.51404393e-01 1.13854170e+00
6.35238469e-01 6.24378026e-01 7.31894791e-01 -6.55272901e-01
-7.25523233e-01 -1.10269558e+00 -4.48935628e-02 5.51522058e-03
6.22306526e-01 -2.98493773e-01 -3.54221672e-01 1.33764550e-01]
|
[14.461381912231445, 7.140631675720215]
|
f6e21503-e4bf-4f03-80d3-5e86c5ba400c
|
mopo-lsi-a-user-guide
|
2307.01719
| null |
https://arxiv.org/abs/2307.01719v1
|
https://arxiv.org/pdf/2307.01719v1.pdf
|
MOPO-LSI: A User Guide
|
MOPO-LSI is an open-source Multi-Objective Portfolio Optimization Library for Sustainable Investments. This document provides a user guide for MOPO-LSI version 1.0, including problem setup, workflow and the hyper-parameters in configurations.
|
["Michael O'Leary", 'Wang', 'David', 'Jasmine Xu', 'Kumar Neelotpal Shukla', 'Yong Zheng']
|
2023-07-04
| null | null | null | null |
['portfolio-optimization']
|
['time-series']
|
[-4.56487000e-01 -4.62545037e-01 -3.66112769e-01 -1.81688011e-01
-1.08933799e-01 -6.57409668e-01 -1.85745716e-01 -3.15565109e-01
1.17771104e-01 1.11281610e+00 -1.18230343e-01 -5.68838537e-01
-1.19553375e+00 -8.84570003e-01 -1.26347482e-01 -8.34708214e-01
-2.52836674e-01 9.98131752e-01 -3.61058831e-01 -4.47005332e-01
6.82324648e-01 4.19536859e-01 -1.18901992e+00 1.85132533e-01
6.72446012e-01 1.21925759e+00 3.77183974e-01 9.05497789e-01
-7.33948126e-02 5.79247057e-01 -6.40448511e-01 -6.50286436e-01
5.18716753e-01 7.78765678e-02 -8.70225310e-01 -2.57126391e-01
-5.54550767e-01 2.52470374e-02 3.19464579e-02 7.61858881e-01
9.12773430e-01 2.52860129e-01 5.75797498e-01 -1.77685082e+00
-4.19595420e-01 6.91635549e-01 -2.48312578e-01 4.96262193e-01
1.98568448e-01 6.01255178e-01 1.53264880e+00 -6.81828141e-01
6.02906466e-01 1.27837074e+00 5.25300086e-01 -9.73619670e-02
-9.55211043e-01 -2.08478302e-01 -9.59617719e-02 1.49912432e-01
-1.14343274e+00 -1.26154751e-01 6.12621546e-01 -2.96485484e-01
1.88177657e+00 1.03076065e+00 8.15793872e-01 3.95501018e-01
5.65517306e-01 7.12388337e-01 9.80485499e-01 -3.32385093e-01
1.47362918e-01 3.17315936e-01 -2.49866359e-02 2.05207124e-01
7.10167289e-01 2.55565941e-01 -6.57041788e-01 -2.19174862e-01
5.52529633e-01 -3.61368388e-01 1.85364351e-01 -1.18067697e-01
-9.79524255e-01 1.01148307e+00 -1.61375791e-01 4.21968728e-01
-4.35759753e-01 2.06053883e-01 5.93209192e-02 4.80899125e-01
4.77387428e-01 1.27323914e+00 -4.83069569e-01 -4.07777667e-01
-6.95617080e-01 8.74423146e-01 9.02065396e-01 1.13347375e+00
3.00792038e-01 3.43693942e-01 -5.70514262e-01 7.99713850e-01
5.03087044e-01 3.57265621e-01 5.71465939e-02 -1.45847583e+00
8.79695594e-01 4.44742739e-01 7.18543589e-01 -1.01106668e+00
-5.70182741e-01 -9.99182761e-01 1.61418349e-01 3.15153152e-01
-2.26154234e-02 -4.79605734e-01 6.11959174e-02 7.32870996e-01
-6.47316650e-02 -5.35110295e-01 2.89894931e-04 7.16184378e-01
7.54351914e-01 6.25518203e-01 -4.74469125e-01 -3.75437140e-01
8.03183913e-01 -1.54436624e+00 -9.45579767e-01 -3.57347935e-01
4.11001474e-01 -1.08163226e+00 7.16558397e-01 3.07745516e-01
-1.66575158e+00 3.54392558e-01 -1.04330242e+00 6.92829728e-01
-7.86812425e-01 -1.29058689e-01 1.15314102e+00 1.12245071e+00
-9.85294104e-01 9.98169363e-01 -9.23143625e-02 1.81223258e-01
4.10077125e-01 4.44192469e-01 3.10478568e-01 2.76067276e-02
-9.11752105e-01 1.26412559e+00 9.38062966e-01 -4.65584770e-02
-4.43665445e-01 -1.21073747e+00 -5.70695937e-01 4.49925870e-01
8.62617552e-01 -4.90408957e-01 1.16330659e+00 -3.93579304e-01
-2.08475351e+00 6.12840414e-01 1.26420319e-01 7.92536661e-02
3.90985161e-01 2.98292547e-01 -5.54111600e-01 -2.26907805e-01
-4.89841886e-02 -1.17181331e-01 1.78552300e-01 -7.34158933e-01
-8.30348074e-01 2.76485622e-01 1.78271875e-01 4.10903633e-01
-2.65348405e-01 7.83769727e-01 -2.69153148e-01 -7.81197786e-01
-5.43588698e-01 -4.44569230e-01 -6.60086095e-01 -8.10048401e-01
-2.81489015e-01 -2.72038598e-02 -5.99448420e-02 -4.31222886e-01
1.85395300e+00 -1.17790914e+00 -6.81216316e-03 3.27855915e-01
-3.04525524e-01 4.48334441e-02 -4.81563359e-01 5.96153378e-01
-3.27458739e-01 2.64190257e-01 -7.75171071e-02 -4.07352090e-01
7.77166963e-01 6.85651451e-02 3.93856704e-01 1.19333871e-01
1.87923178e-01 1.11197472e+00 -1.07680547e+00 -1.64928854e-01
2.94349343e-01 -5.65835118e-01 -4.93021756e-01 2.97844708e-01
-2.34955817e-01 -1.03967432e-02 -2.68122017e-01 1.45319605e+00
8.06750238e-01 -1.84987366e-01 1.21758096e-01 3.12552869e-01
-7.00670183e-01 1.75019994e-01 -1.64880311e+00 1.26054716e+00
-4.63198841e-01 5.00450671e-01 1.93569481e-01 -1.18841410e+00
5.79184532e-01 2.33702078e-01 1.05599988e+00 -5.63407362e-01
1.09489635e-01 4.66993123e-01 -3.92982513e-01 -4.88142014e-01
5.95991015e-01 5.86769953e-02 -6.22365475e-02 6.72014534e-01
2.36944526e-01 -2.39617646e-01 1.26653540e+00 -4.87511992e-01
1.00492728e+00 8.92772675e-02 2.02643335e-01 -8.34761202e-01
7.33380735e-01 2.74787575e-01 9.06665921e-01 5.31005740e-01
7.75077492e-02 3.38341981e-01 7.72132933e-01 -3.72743279e-01
-6.81605756e-01 -7.67730832e-01 -3.82038176e-01 1.23870349e+00
-2.89712131e-01 -2.20094100e-01 -2.46132478e-01 -1.62963212e-01
7.30838835e-01 1.28007901e+00 -2.85213053e-01 5.90997696e-01
-4.94884044e-01 -1.50670862e+00 -1.96637705e-01 1.34826556e-01
-8.91827047e-03 -1.18442619e+00 -4.18238044e-01 6.79256678e-01
-1.54181093e-01 -8.56721878e-01 -4.05762225e-01 -3.74971703e-02
-7.47107148e-01 -1.00827134e+00 -8.38370562e-01 -2.60168344e-01
4.66120243e-01 6.82809427e-02 1.87309825e+00 -3.61738428e-02
-5.61139464e-01 2.04404458e-01 1.05886103e-03 -6.73821509e-01
2.71315128e-01 -1.16964325e-01 -3.56764287e-01 -2.95049310e-01
-1.90230515e-02 -2.64612019e-01 -2.63968080e-01 7.08461761e-01
-4.84523475e-01 -3.34513634e-01 2.07795531e-01 7.44105995e-01
6.40835464e-01 5.20552874e-01 6.08868361e-01 -5.55723906e-01
9.12024200e-01 -7.13640630e-01 -1.61529970e+00 1.07280481e+00
-8.95273507e-01 1.90884814e-01 1.18737593e-01 9.97536704e-02
-1.12970817e+00 -6.69602454e-01 -1.14078775e-01 -4.13711928e-02
6.53446198e-01 7.97334552e-01 -6.24806762e-01 -5.52203894e-01
-1.71519101e-01 -6.16052389e-01 -4.27019536e-01 -6.62455618e-01
-1.01372153e-01 2.22663477e-01 2.16904655e-03 -7.73164392e-01
5.11799991e-01 -1.76839486e-01 3.68818283e-01 -9.77003202e-02
-9.41039145e-01 -4.27070141e-01 -6.11940725e-03 -6.14391744e-01
3.60137731e-01 -1.68060601e-01 -1.02567506e+00 6.59418225e-01
-6.92246497e-01 -4.66038525e-01 -2.11311489e-01 2.91384995e-01
-5.81355453e-01 -3.97818983e-01 4.35748734e-02 -8.43446076e-01
-5.46748698e-01 -1.67553627e+00 1.46215886e-01 4.75022167e-01
2.00321376e-01 -1.69359565e+00 3.99241596e-01 8.55113745e-01
1.12402141e+00 6.75052285e-01 4.31559443e-01 -4.11735028e-01
-3.63828033e-01 -2.50388443e-01 4.63042893e-02 3.03119868e-01
-1.87624633e-01 1.21585794e-01 -3.90094161e-01 -3.09620500e-01
-1.18091092e-01 6.96548913e-03 1.12007588e-01 7.85507143e-01
1.22854638e+00 -7.47345805e-01 -1.04033068e-01 1.10508251e+00
1.96863961e+00 3.19698632e-01 2.45222390e-01 1.38537657e+00
2.48021744e-02 7.79157758e-01 1.36721098e+00 9.22373652e-01
5.85724302e-02 6.01847708e-01 6.97493672e-01 4.14661914e-01
7.99631953e-01 6.40427053e-01 8.46374854e-02 5.55431724e-01
-4.62797523e-01 -7.18271255e-01 -1.02873659e+00 4.88499761e-01
-1.79089820e+00 -9.83790636e-01 4.87003848e-02 2.07990885e+00
4.62514192e-01 2.33771384e-01 4.03738767e-01 -1.60630345e-02
5.16043723e-01 2.43763283e-01 -4.93184090e-01 -9.81457949e-01
-5.15308321e-01 4.42265332e-01 1.30283475e+00 6.72151864e-01
-7.90695846e-01 4.57843035e-01 8.29273033e+00 6.91845357e-01
-7.83643126e-01 1.95758790e-01 6.60630584e-01 -1.02312040e+00
-6.54011369e-01 1.86566040e-02 -9.28275049e-01 6.70309961e-01
1.20447659e+00 -1.17317343e+00 8.90549421e-01 8.19113970e-01
2.97697365e-01 -1.59769922e-01 -6.66592240e-01 4.91548598e-01
-3.95929873e-01 -2.03978825e+00 -6.51635408e-01 2.96119094e-01
9.91106033e-01 -1.70232505e-01 2.38904148e-01 2.65377373e-01
3.98108959e-01 -1.11379611e+00 8.63072097e-01 7.86315203e-01
4.64521617e-01 -1.32702744e+00 1.25330293e+00 -3.71410668e-01
-1.03561020e+00 -4.97428596e-01 -5.12462854e-01 -8.90197903e-02
4.92712796e-01 6.85492635e-01 -3.44585389e-01 1.15901554e+00
1.00036275e+00 6.25473738e-01 -1.84143230e-01 1.60371780e+00
-1.00314364e-01 4.27834958e-01 -3.78805220e-01 -1.30923688e-01
3.26253533e-01 -7.31922030e-01 8.53593290e-01 1.08189237e+00
4.71434087e-01 4.50703092e-02 -2.75638223e-01 8.89657378e-01
4.45241868e-01 -8.61301050e-02 -8.17736834e-02 -3.90718579e-01
6.14921808e-01 1.17878556e+00 -1.38399825e-01 1.68916076e-01
-1.81438312e-01 1.27932400e-01 -9.20983255e-02 2.41268411e-01
-9.61754858e-01 -1.03314221e+00 1.19321084e+00 -2.42866963e-01
2.02208787e-01 -2.38842051e-02 -8.41368437e-01 -6.39416993e-01
5.17619289e-02 -7.56497681e-01 5.93495965e-01 -7.32218862e-01
-7.51414955e-01 -1.44057363e-01 3.55608761e-01 -1.08662558e+00
-5.07863685e-02 -1.29999483e+00 -1.21027684e+00 1.13074052e+00
-1.93307865e+00 -6.19143724e-01 -4.61224951e-02 -3.01188350e-01
2.95312196e-01 -6.92039609e-01 5.50240755e-01 4.31322366e-01
-1.49088025e+00 8.46622884e-01 8.86405349e-01 -7.37619042e-01
2.53027707e-01 -1.40925789e+00 2.83519566e-01 7.96700239e-01
-1.08255744e+00 2.93523192e-01 8.94203424e-01 -3.23581904e-01
-1.56116509e+00 -6.98434114e-01 9.28094566e-01 -1.22368164e-01
1.12100232e+00 2.56031632e-01 -1.69266254e-01 4.58386421e-01
7.08129287e-01 -4.89984721e-01 8.78621697e-01 -2.45119885e-01
9.02721763e-01 -4.64141905e-01 -1.45912755e+00 3.36043149e-01
8.40422571e-01 5.08899450e-01 1.00429289e-01 7.81589568e-01
5.61088145e-01 -4.77828175e-01 -1.75732219e+00 8.85801762e-02
3.22875768e-01 -8.32967460e-01 1.60124326e+00 -4.64203715e-01
2.52544373e-01 1.44199893e-01 -8.00881088e-02 -1.37713289e+00
-7.26755977e-01 -1.46313119e+00 -3.26504916e-01 1.40275335e+00
8.62842143e-01 -1.35596287e+00 4.17034268e-01 8.76383126e-01
-2.65679300e-01 -1.19432104e+00 -1.12017465e+00 -1.18886268e+00
2.19168186e-01 -2.70880640e-01 1.53942943e+00 4.72608209e-01
7.77656585e-02 -4.61998940e-01 -3.29291940e-01 3.75683531e-02
9.49838579e-01 2.07539767e-01 2.96968102e-01 -6.94334149e-01
-5.13962388e-01 -1.30492580e+00 2.79169619e-01 4.50703725e-02
-2.24203989e-01 -6.20242119e-01 -4.66322452e-01 -1.48808670e+00
-1.21104643e-01 -7.41420150e-01 -9.39761758e-01 3.95912915e-01
-3.17274451e-01 -2.87522584e-01 3.16411078e-01 8.76166075e-02
-7.51013994e-01 4.26112056e-01 1.12726378e+00 -2.62510926e-01
-2.03629121e-01 4.78909522e-01 -8.91887367e-01 4.10600513e-01
1.22468960e+00 -9.07605529e-01 -1.14947334e-01 -5.91683030e-01
5.72265089e-01 2.71146953e-01 -4.52477813e-01 -6.59434915e-01
-3.68761420e-01 -1.39219928e+00 -7.80758932e-02 -9.39268708e-01
3.28142762e-01 -5.02372503e-01 6.32266700e-01 3.38645995e-01
-1.13557912e-02 5.64511955e-01 2.92699158e-01 -2.69330472e-01
-1.74880493e-02 -1.10009551e+00 4.94578212e-01 -2.57672846e-01
-4.34756339e-01 5.63236661e-02 -4.52169955e-01 1.96384683e-01
1.23303843e+00 -1.46914542e-01 -5.87685823e-01 2.89104849e-01
-4.35392827e-01 1.19767654e+00 5.00443220e-01 1.28681198e-01
3.82391304e-01 -1.47122562e+00 -9.43978548e-01 -3.60585958e-01
-1.98871940e-01 -4.16969180e-01 -2.19246745e-02 8.82596314e-01
-1.13382459e+00 1.29116416e+00 -6.39580965e-01 2.35183164e-01
-9.69938159e-01 1.81593239e-01 7.70484090e-01 -1.25856638e+00
1.98105499e-01 1.21609414e+00 -5.25570929e-01 -4.85625088e-01
1.07550927e-01 -7.78381824e-02 -3.87231678e-01 -1.03609785e-01
5.46712935e-01 1.38607478e+00 1.24634758e-01 -9.23560932e-02
-6.88295603e-01 3.77114475e-01 5.65413773e-01 -2.50859141e-01
2.05440497e+00 -5.55137452e-03 -6.06106102e-01 3.50169875e-02
1.00923610e+00 -4.32015926e-01 -8.46315682e-01 1.15889162e-01
5.48695147e-01 -8.81770253e-01 4.06278819e-01 -1.05447853e+00
-1.46247566e+00 2.72330493e-01 2.27261215e-01 -1.92573309e-01
1.24357462e+00 -6.33550227e-01 6.57753646e-01 2.95353085e-01
3.11011493e-01 -1.64943838e+00 4.08502249e-03 4.99706864e-01
1.46610272e+00 -8.93908918e-01 7.45956659e-01 -5.83168827e-02
-9.05773699e-01 1.19547153e+00 2.60841280e-01 2.08888024e-01
7.56304026e-01 2.67171293e-01 -3.50079745e-01 -1.64459959e-01
-8.59808922e-01 -8.66811797e-02 3.14404786e-01 5.94080567e-01
2.51863956e-01 3.20576161e-01 -3.67938250e-01 1.05598819e+00
-2.53999352e-01 -3.62415761e-01 4.29938495e-01 1.39048088e+00
-1.50628477e-01 -1.42428458e+00 -8.65502119e-01 4.54447478e-01
-9.36819673e-01 4.48521301e-02 -1.15715630e-01 3.67477268e-01
-2.33486950e-01 1.27748680e+00 -3.15189995e-02 -2.71771457e-02
5.19587398e-01 -5.13625979e-01 2.50650436e-01 -3.06317210e-01
-1.04671741e+00 -2.46023208e-01 7.90247917e-01 -7.85962343e-01
-7.95493275e-02 -1.03798521e+00 -6.97273016e-01 -7.14665055e-01
-5.31092703e-01 2.29991645e-01 1.06185949e+00 5.13172388e-01
1.86568514e-01 1.00816333e+00 1.29100144e+00 -1.16314685e+00
-4.69949424e-01 -6.58912778e-01 -5.26822507e-01 -6.29601359e-01
-3.42104495e-01 -9.71401095e-01 -5.94237804e-01 -8.35444689e-01]
|
[5.818556785583496, 3.6512136459350586]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.