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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
54c7bcf0-2c61-4fe1-aedc-2e3d13e69aa3
|
hinet-half-instance-normalization-network-for
|
2105.06086
| null |
https://arxiv.org/abs/2105.06086v2
|
https://arxiv.org/pdf/2105.06086v2.pdf
|
HINet: Half Instance Normalization Network for Image Restoration
|
In this paper, we explore the role of Instance Normalization in low-level vision tasks. Specifically, we present a novel block: Half Instance Normalization Block (HIN Block), to boost the performance of image restoration networks. Based on HIN Block, we design a simple and powerful multi-stage network named HINet, which consists of two subnetworks. With the help of HIN Block, HINet surpasses the state-of-the-art (SOTA) on various image restoration tasks. For image denoising, we exceed it 0.11dB and 0.28 dB in PSNR on SIDD dataset, with only 7.5% and 30% of its multiplier-accumulator operations (MACs), 6.8 times and 2.9 times speedup respectively. For image deblurring, we get comparable performance with 22.5% of its MACs and 3.3 times speedup on REDS and GoPro datasets. For image deraining, we exceed it by 0.3 dB in PSNR on the average result of multiple datasets with 1.4 times speedup. With HINet, we won 1st place on the NTIRE 2021 Image Deblurring Challenge - Track2. JPEG Artifacts, with a PSNR of 29.70. The code is available at https://github.com/megvii-model/HINet.
|
['Chengpeng Chen', 'Xiaojie Chu', 'Jie Zhang', 'Xin Lu', 'Liangyu Chen']
|
2021-05-13
| null | null | null | null |
['single-image-deraining']
|
['computer-vision']
|
[ 1.66625515e-01 -3.36544693e-01 6.29221722e-02 -6.76502958e-02
-7.30835855e-01 -1.49660558e-01 4.64346647e-01 -2.59052604e-01
-8.52002740e-01 4.99261260e-01 3.10106933e-01 -5.24159431e-01
2.03905478e-01 -5.06349325e-01 -9.18427408e-01 -7.85971820e-01
-1.21854648e-01 -5.16524851e-01 3.68419975e-01 -3.48018527e-01
2.67785877e-01 3.88063490e-01 -1.24973714e+00 3.30750108e-01
6.68258667e-01 1.15636992e+00 2.50837505e-01 8.85030687e-01
3.40934366e-01 9.90294516e-01 -6.23026013e-01 -3.29720259e-01
4.48245317e-01 -1.45072550e-01 -5.73274612e-01 -7.56597444e-02
8.48172665e-01 -9.87840712e-01 -9.09701824e-01 1.22539866e+00
9.27323818e-01 8.43426026e-03 2.30206564e-01 -1.19718361e+00
-6.86143756e-01 6.62262499e-01 -1.05180156e+00 6.59409404e-01
-2.21656337e-01 4.48995233e-01 5.30113876e-01 -9.33362603e-01
4.78960037e-01 1.23706996e+00 6.77156687e-01 4.81693208e-01
-1.06328273e+00 -8.49427044e-01 -1.49084479e-01 7.55786598e-01
-1.23428667e+00 -7.72157192e-01 2.58467555e-01 6.45842031e-02
1.04688549e+00 2.74973035e-01 1.63921371e-01 9.11876142e-01
3.50540549e-01 6.93051696e-01 1.38964212e+00 -4.93647903e-02
1.11626975e-01 -5.87697983e-01 3.35121930e-01 2.92242378e-01
3.18877846e-01 1.51209831e-01 -5.15377641e-01 3.84300768e-01
8.87267768e-01 -1.89064592e-01 -6.26180172e-01 4.03553933e-01
-1.25290596e+00 3.37646812e-01 8.42580497e-01 1.77265003e-01
-3.92391354e-01 5.90944886e-01 6.39014304e-01 5.32438576e-01
4.01306748e-01 -6.59758747e-02 -3.69641006e-01 -1.07790016e-01
-1.02500582e+00 1.17848694e-01 4.44732517e-01 7.54950404e-01
4.55810398e-01 1.97479978e-01 -2.57758617e-01 9.63228703e-01
2.10768487e-02 5.55670440e-01 4.14111108e-01 -1.28444993e+00
3.81016582e-01 6.31843135e-02 -1.10788144e-01 -9.15071309e-01
-3.20821583e-01 -7.45285332e-01 -1.72035170e+00 4.88544136e-01
3.64131808e-01 -2.55979295e-03 -1.23838329e+00 1.59519660e+00
-2.22320929e-02 5.30658841e-01 5.76098785e-02 1.14067781e+00
9.10431981e-01 8.10815454e-01 -9.69360098e-02 -2.44465336e-01
1.60808778e+00 -1.26068306e+00 -7.28351295e-01 -3.60469908e-01
3.09916198e-01 -9.82593775e-01 1.06572628e+00 7.37040818e-01
-1.35551977e+00 -7.43743420e-01 -1.25159478e+00 -3.65689248e-01
7.71334097e-02 9.96011570e-02 3.21113586e-01 5.03600657e-01
-1.44202232e+00 7.99695194e-01 -9.18734074e-01 -1.65406317e-01
6.17746234e-01 2.84617662e-01 -2.81821758e-01 -5.23040056e-01
-9.57120955e-01 6.96634412e-01 6.81856275e-02 2.35526606e-01
-1.16504586e+00 -8.62354457e-01 -5.79883993e-01 3.97218354e-02
2.36542299e-01 -6.66515052e-01 1.12664104e+00 -5.77352345e-01
-1.32666612e+00 6.40561223e-01 -1.14477919e-02 -8.12808394e-01
6.25861943e-01 -5.25598288e-01 -5.25603592e-01 2.86434025e-01
-2.11871956e-02 8.20502818e-01 9.11489069e-01 -9.88335967e-01
-3.59085083e-01 -3.33937138e-01 1.56290829e-01 -1.04127545e-02
-3.82346988e-01 2.75264740e-01 -8.90666068e-01 -8.84125948e-01
7.64058307e-02 -6.76863730e-01 -2.92095184e-01 2.90694118e-01
-4.39867377e-01 2.09676072e-01 6.50383532e-01 -1.14794481e+00
1.22799683e+00 -2.38425040e+00 -7.81971961e-02 -3.25746536e-01
5.46402872e-01 6.02416635e-01 -4.85764861e-01 7.92790577e-02
-1.52511850e-01 9.35565382e-02 -3.73407394e-01 -5.41471779e-01
-1.28202736e-01 4.46723700e-02 -1.97911203e-01 7.71361887e-01
-2.21215896e-02 7.07629740e-01 -5.84621012e-01 -9.91221517e-03
2.62975156e-01 6.73901379e-01 -5.56534588e-01 8.87255222e-02
3.27629924e-01 2.07661629e-01 8.66252333e-02 5.61687827e-01
1.43708920e+00 -2.99469501e-01 -2.97841337e-02 -8.08226109e-01
-1.75091088e-01 1.83201566e-01 -1.17140782e+00 1.75371945e+00
-5.31926394e-01 9.98919010e-01 3.90198380e-01 -7.22157061e-01
4.20896083e-01 1.77297205e-01 1.22464426e-01 -1.05464399e+00
2.15859547e-01 2.65059382e-01 -1.87879261e-02 -2.93306023e-01
3.51149917e-01 3.37686062e-01 3.47880930e-01 2.38316834e-01
4.50350111e-03 3.49839896e-01 3.65232319e-01 4.51235682e-01
1.55068386e+00 -2.76344836e-01 -3.76097597e-02 -3.04104060e-01
5.57421327e-01 -3.95090371e-01 4.10597652e-01 8.70197058e-01
-3.05548549e-01 7.47949660e-01 4.30310071e-01 -3.90246123e-01
-1.04917228e+00 -1.23456228e+00 -2.20950037e-01 7.90852368e-01
3.71689647e-01 -6.23058200e-01 -8.54016244e-01 -1.78258255e-01
-4.13057059e-01 4.37349141e-01 -2.90433019e-01 -1.94303095e-02
-7.49691308e-01 -1.23065448e+00 7.61708021e-01 3.24158579e-01
1.32282722e+00 -8.35640073e-01 -3.47628146e-01 1.27223149e-01
-2.90759832e-01 -1.39132464e+00 -6.89558625e-01 -1.58494815e-01
-9.11865115e-01 -8.98924768e-01 -8.36400151e-01 -6.25720978e-01
4.77655679e-01 7.47860909e-01 1.13776875e+00 3.03971767e-01
-1.95942149e-01 -1.56338558e-01 -2.51913309e-01 2.54853994e-01
-2.79888809e-01 -4.42783087e-02 1.46069359e-02 -2.41338462e-01
-2.48807356e-01 -1.04097736e+00 -1.03189039e+00 3.70597839e-01
-1.18960261e+00 1.79619491e-01 8.08347404e-01 9.18133318e-01
3.50572914e-01 1.16659999e-01 3.87212098e-01 -2.93048739e-01
4.53200549e-01 -2.80589700e-01 -5.70423424e-01 -2.82930247e-02
-7.80031681e-01 -1.10238358e-01 5.39971352e-01 -4.54568267e-01
-7.14842319e-01 -3.50545883e-01 -3.62309545e-01 -5.27145624e-01
-1.10227309e-01 2.50561088e-01 1.79574043e-02 -2.14341074e-01
6.56444013e-01 3.22146714e-01 1.16696984e-01 -7.64330864e-01
3.92222494e-01 7.07821250e-01 9.71289277e-01 -1.92385435e-01
8.54675710e-01 4.46450323e-01 -9.22685396e-03 -6.14241242e-01
-5.91170013e-01 -2.71802127e-01 -2.39219237e-02 -6.32295161e-02
8.15006793e-01 -1.41356659e+00 -8.88249457e-01 1.24447870e+00
-1.17982423e+00 -5.41954696e-01 1.85335234e-01 2.85016328e-01
-1.32096812e-01 7.82796264e-01 -1.24270844e+00 -1.70599073e-01
-8.78643155e-01 -1.34814000e+00 6.95524275e-01 1.25307262e-01
4.36443746e-01 -4.09039885e-01 -5.34575820e-01 5.61957359e-01
8.13070655e-01 1.80180781e-02 3.79746914e-01 2.03234497e-02
-7.62754142e-01 1.85494199e-01 -9.39099252e-01 9.72755969e-01
-2.51112524e-02 -4.75539178e-01 -8.98468137e-01 -8.33572268e-01
1.03207044e-01 -2.04145163e-01 1.31943834e+00 4.68488753e-01
1.38727760e+00 -3.31438869e-01 2.14698985e-01 1.06957448e+00
1.54587352e+00 -1.49869055e-01 1.18854868e+00 5.83765626e-01
6.93996668e-01 -3.31877656e-02 2.19360620e-01 3.71409655e-01
3.05446923e-01 7.31083393e-01 7.84595549e-01 -3.57138723e-01
-8.51044595e-01 2.65394509e-01 7.71820307e-01 8.83247674e-01
-1.79391071e-01 -2.35576868e-01 -7.48449326e-01 4.21869755e-01
-1.53140783e+00 -7.07133353e-01 -5.18252373e-01 1.95742226e+00
1.02693164e+00 2.28291854e-01 -1.76635846e-01 1.37927920e-01
7.21374929e-01 4.05966252e-01 -6.45205379e-01 -9.61827114e-02
-3.33296537e-01 3.77736479e-01 1.01076901e+00 6.46547377e-01
-1.02007532e+00 8.83624613e-01 4.87520027e+00 1.25479245e+00
-8.74400795e-01 3.30542088e-01 9.79112267e-01 -1.39979944e-01
2.89044619e-01 -2.07177326e-01 -6.36683464e-01 6.10951781e-01
1.00485480e+00 6.36254475e-02 7.98135459e-01 3.63515466e-01
4.06987369e-01 -1.32445112e-01 -6.94395661e-01 1.19361150e+00
1.13918267e-01 -1.50261080e+00 -3.63570615e-03 1.51427791e-01
6.67412877e-01 3.97389770e-01 6.20329715e-02 5.16406894e-02
6.23293184e-02 -1.03137386e+00 6.09249353e-01 1.31898955e-01
1.01499593e+00 -7.27176607e-01 1.00329506e+00 4.53287475e-02
-9.68921363e-01 1.17589489e-01 -4.65112358e-01 -4.48146947e-02
3.26723754e-01 9.93393362e-01 -1.94889903e-01 5.84994793e-01
1.17332017e+00 9.04458523e-01 -6.94555879e-01 1.18742514e+00
-3.47883075e-01 8.34016621e-01 -3.71837735e-01 7.09821105e-01
1.81871578e-01 1.05520440e-02 5.62377930e-01 1.20240581e+00
3.58613878e-01 1.33540019e-01 -3.70660961e-01 3.93708706e-01
-4.86772150e-01 -3.79259169e-01 1.94908723e-01 6.04132295e-01
3.35363269e-01 1.27128816e+00 -4.31729287e-01 -5.19008756e-01
-2.06769809e-01 1.44168913e+00 -7.20252767e-02 3.39986384e-01
-1.10041988e+00 -4.51099664e-01 1.05436528e+00 -1.45956367e-01
4.71044153e-01 -2.96018481e-01 -2.97977954e-01 -1.22339821e+00
2.13803053e-01 -1.23839939e+00 -4.96634402e-06 -8.67887855e-01
-1.27684903e+00 7.84149230e-01 -3.37336242e-01 -1.08003008e+00
4.36185628e-01 -6.07064426e-01 -5.02219081e-01 9.22393560e-01
-1.89793527e+00 -7.75027573e-01 -6.88817143e-01 7.55664945e-01
5.30805647e-01 7.68276975e-02 2.72635520e-01 9.28126097e-01
-8.61695290e-01 7.40768731e-01 2.08148107e-01 1.99649245e-01
1.00812185e+00 -9.28135872e-01 1.04947937e+00 1.31045854e+00
-3.23336303e-01 4.25791174e-01 7.58706570e-01 -3.90519917e-01
-1.64342511e+00 -1.14384699e+00 7.18899310e-01 7.64328241e-02
7.85561085e-01 -1.50710002e-01 -1.03561378e+00 2.83796400e-01
5.83374918e-01 4.36633587e-01 -2.07712930e-02 -4.24286366e-01
-6.30887270e-01 -3.86800736e-01 -1.13571799e+00 7.38622606e-01
1.19867361e+00 -2.16645166e-01 9.00107063e-03 4.00093824e-01
8.70606542e-01 -6.39554381e-01 -9.03772950e-01 4.27221775e-01
2.67422855e-01 -1.21698499e+00 1.46472132e+00 -9.33654159e-02
7.91013896e-01 -5.02778172e-01 -3.45774710e-01 -1.07486689e+00
-4.59962457e-01 -7.18074560e-01 -1.30323932e-01 1.02126157e+00
2.27072742e-02 -5.65614879e-01 4.56248254e-01 -4.43993770e-02
-3.43575358e-01 -6.66023493e-01 -1.06928742e+00 -9.15733039e-01
7.94680975e-03 -5.23052096e-01 2.84156501e-01 7.14961827e-01
-7.51029730e-01 1.98673084e-01 -7.55813241e-01 3.58807951e-01
1.10618675e+00 -3.27956110e-01 5.74609101e-01 -4.64179546e-01
-3.90272260e-01 -4.12010044e-01 -3.73059481e-01 -1.49179232e+00
-2.84526676e-01 -7.23646998e-01 -1.73413798e-01 -1.45018339e+00
2.12624118e-01 2.39648484e-02 -5.04713237e-01 6.89595878e-01
-2.35003993e-01 9.24892128e-01 5.84145546e-01 2.56106824e-01
-4.24562097e-01 4.53430444e-01 1.26273239e+00 -2.57036239e-01
7.99273029e-02 -2.81039804e-01 -7.61130869e-01 5.21299064e-01
1.23646784e+00 -4.60959524e-01 -1.44382298e-01 -9.65344429e-01
-4.28333916e-02 -5.88597555e-04 8.08957875e-01 -1.28661251e+00
3.20265710e-01 3.31652015e-01 3.53902876e-01 -5.04047692e-01
3.45461339e-01 -5.10287702e-01 1.40332788e-01 7.38598466e-01
-7.48451650e-02 1.93443894e-01 4.68915671e-01 3.09750468e-01
-1.94376141e-01 2.12322205e-01 1.05246592e+00 1.92732185e-01
-8.08693349e-01 2.70196587e-01 -3.69038761e-01 -5.37787238e-03
4.53619987e-01 -7.40899965e-02 -8.83241832e-01 -3.30914557e-01
-5.41206479e-01 1.11222319e-01 2.57983238e-01 1.89650148e-01
7.81133711e-01 -1.21051073e+00 -1.10086596e+00 5.95448986e-02
-4.33140904e-01 -2.45206773e-01 7.32682586e-01 1.16308379e+00
-6.70106232e-01 -4.09311578e-02 -4.96177673e-01 -4.46767211e-01
-1.36460531e+00 2.69001275e-01 3.40007216e-01 -3.79019678e-01
-8.76212776e-01 8.00586998e-01 1.63670078e-01 1.74740866e-01
2.34280393e-01 -3.39347988e-01 8.15689117e-02 -2.35714152e-01
9.97865736e-01 7.57751465e-01 3.15164566e-01 -2.97921002e-01
-3.62169623e-01 3.50572318e-01 -2.64977783e-01 6.46977052e-02
1.49996126e+00 -2.64534950e-01 -5.62383711e-01 -3.92756760e-01
1.30713201e+00 -1.83718756e-01 -1.50854814e+00 -2.50904292e-01
-4.07656908e-01 -5.94522715e-01 7.41978586e-01 -1.13865876e+00
-1.67075431e+00 6.57508671e-01 1.11592913e+00 -1.69123441e-01
1.82235527e+00 -1.68550953e-01 1.22264779e+00 7.81734660e-02
1.72596931e-01 -5.09380281e-01 -1.11283623e-02 4.56210375e-01
1.22518182e+00 -1.18753779e+00 2.66290814e-01 -3.97169173e-01
-1.80524901e-01 9.46764350e-01 4.95790333e-01 -2.93313563e-01
5.15298903e-01 5.45011938e-01 7.02282116e-02 1.68995142e-01
-6.35289967e-01 1.69836923e-01 -2.74601970e-02 2.88846314e-01
3.00872684e-01 -1.24631718e-01 -4.29561973e-01 2.86640048e-01
-3.92085500e-02 7.20476434e-02 7.51201153e-01 5.26377320e-01
-3.25956553e-01 -1.09870994e+00 -4.09759969e-01 2.24037856e-01
-6.97858870e-01 -7.63538837e-01 3.21045637e-01 4.99497712e-01
9.81534738e-03 1.16638494e+00 6.87621087e-02 -5.15163660e-01
3.57850581e-01 -8.05272996e-01 2.90861130e-01 2.43576169e-02
-6.97930157e-01 4.65405099e-02 7.44439662e-02 -9.70034242e-01
-3.74360055e-01 -3.67438972e-01 -9.00593758e-01 -9.84438300e-01
9.98961329e-02 -3.70731741e-01 6.92161202e-01 4.25331205e-01
5.68182170e-01 6.82791889e-01 5.25622487e-01 -1.01483226e+00
-6.50199771e-01 -1.01594698e+00 -3.80162239e-01 6.40978143e-02
6.22866035e-01 -1.11240484e-01 -6.98981106e-01 1.34559229e-01]
|
[11.15610122680664, -2.195042133331299]
|
d4220fbe-d5bc-4b46-99dd-4d9e5dce26b5
|
dense-steerable-filter-cnns-for-exploiting
|
2004.03037
| null |
https://arxiv.org/abs/2004.03037v2
|
https://arxiv.org/pdf/2004.03037v2.pdf
|
Dense Steerable Filter CNNs for Exploiting Rotational Symmetry in Histology Images
|
Histology images are inherently symmetric under rotation, where each orientation is equally as likely to appear. However, this rotational symmetry is not widely utilised as prior knowledge in modern Convolutional Neural Networks (CNNs), resulting in data hungry models that learn independent features at each orientation. Allowing CNNs to be rotation-equivariant removes the necessity to learn this set of transformations from the data and instead frees up model capacity, allowing more discriminative features to be learned. This reduction in the number of required parameters also reduces the risk of overfitting. In this paper, we propose Dense Steerable Filter CNNs (DSF-CNNs) that use group convolutions with multiple rotated copies of each filter in a densely connected framework. Each filter is defined as a linear combination of steerable basis filters, enabling exact rotation and decreasing the number of trainable parameters compared to standard filters. We also provide the first in-depth comparison of different rotation-equivariant CNNs for histology image analysis and demonstrate the advantage of encoding rotational symmetry into modern architectures. We show that DSF-CNNs achieve state-of-the-art performance, with significantly fewer parameters, when applied to three different tasks in the area of computational pathology: breast tumour classification, colon gland segmentation and multi-tissue nuclear segmentation.
|
['Simon Graham', 'Nasir Rajpoot', 'David Epstein']
|
2020-04-06
| null | null | null | null |
['colorectal-gland-segmentation', 'breast-tumour-classification', 'tumour-classification', 'multi-tissue-nucleus-segmentation', 'nuclear-segmentation']
|
['medical', 'medical', 'medical', 'medical', 'medical']
|
[ 3.22695971e-01 2.72466123e-01 -1.16761336e-02 -4.36225235e-01
-3.27940196e-01 -5.96050680e-01 5.79576552e-01 -4.47379723e-02
-7.95868158e-01 4.37313944e-01 1.39100641e-01 -4.63370442e-01
-1.85101882e-01 -6.23993754e-01 -9.20483351e-01 -1.05563021e+00
4.54403721e-02 2.48854339e-01 3.46105754e-01 -2.85246581e-01
-3.09712023e-01 1.07137489e+00 -1.02347684e+00 4.02234197e-01
1.25470728e-01 6.95334494e-01 2.35437065e-01 1.03924096e+00
5.33683777e-01 5.17689645e-01 -9.55598876e-02 -1.94106147e-01
2.18146980e-01 3.17407511e-02 -1.07299173e+00 2.60006972e-02
6.59626603e-01 -2.00664341e-01 -6.92870378e-01 6.95767403e-01
4.57831472e-01 4.77691218e-02 5.55194259e-01 -5.73533118e-01
-3.97766888e-01 4.60313708e-01 -2.77634174e-01 4.41124737e-01
-3.82314831e-01 1.73625443e-02 8.34461927e-01 -5.48506320e-01
9.50724781e-01 1.05602288e+00 9.75863814e-01 6.61507130e-01
-1.54634678e+00 -2.45884895e-01 -3.26772295e-02 -4.38046176e-04
-1.14827752e+00 -5.02766311e-01 2.78425604e-01 -3.02658081e-01
1.21414924e+00 4.86822605e-01 8.12270761e-01 8.31784368e-01
4.49098200e-01 3.89223874e-01 8.36998343e-01 -3.16607535e-01
2.67922739e-03 -2.27866352e-01 1.16489865e-01 1.03350413e+00
3.41124982e-01 5.28011844e-02 -3.42054814e-01 7.98840746e-02
1.31026256e+00 2.28633672e-01 -4.83244687e-01 -7.18531549e-01
-1.40628958e+00 7.85530567e-01 9.14891124e-01 3.54927510e-01
5.66005930e-02 4.24651802e-01 3.20528656e-01 3.39518368e-01
1.85412645e-01 3.63371402e-01 -5.57829082e-01 3.76220763e-01
-7.50391781e-01 6.28307387e-02 5.15949488e-01 7.00218201e-01
6.55983031e-01 -7.24912211e-02 3.48276533e-02 5.83266139e-01
2.44962513e-01 1.05065428e-01 6.03999317e-01 -6.58279359e-01
-2.28590742e-01 6.47647202e-01 -4.72037315e-01 -8.39048982e-01
-1.06721389e+00 -8.89710665e-01 -1.29376996e+00 -1.83051208e-03
5.24407089e-01 1.52425990e-01 -1.36305475e+00 1.65290833e+00
2.27365479e-01 1.87864453e-01 -1.60433158e-01 7.14171410e-01
8.58364224e-01 1.00410461e-01 -2.11848527e-01 2.22447529e-01
1.55511761e+00 -7.33020544e-01 -2.54086912e-01 -1.58363745e-01
9.45424557e-01 -7.58257747e-01 5.73592484e-01 1.52842760e-01
-8.52734447e-01 -1.86066926e-01 -1.09562171e+00 -4.73702222e-01
-4.06247884e-01 1.05500504e-01 9.77172732e-01 5.92445254e-01
-1.38794184e+00 6.95273757e-01 -1.43666530e+00 -4.38557595e-01
6.96747720e-01 8.37327957e-01 -8.98063600e-01 -1.31137103e-01
-7.63456106e-01 9.33813214e-01 1.96124196e-01 2.55192488e-01
-8.47108245e-01 -1.02184689e+00 -9.41461623e-01 1.36083022e-01
-2.25513175e-01 -9.65736389e-01 1.24590921e+00 -8.59239221e-01
-1.32108510e+00 1.04220974e+00 -8.65482464e-02 -6.85551226e-01
6.02181137e-01 3.15190226e-01 -5.74986488e-02 1.98526084e-01
-3.64660859e-01 1.01290250e+00 8.08995485e-01 -7.55560160e-01
-2.65435100e-01 -3.16558897e-01 4.40630578e-02 6.92344978e-02
-4.52383399e-01 -1.46807030e-01 -3.86216819e-01 -5.58731735e-01
3.82945657e-01 -1.28180647e+00 -6.28726661e-01 3.56711596e-01
-3.31140757e-01 2.66574562e-01 7.73216605e-01 -3.28697830e-01
6.32005453e-01 -2.26157880e+00 2.80150324e-01 2.75182337e-01
3.65824848e-01 1.29541591e-01 -2.10595414e-01 -8.18140283e-02
-5.53250492e-01 -5.70976920e-02 -1.50997952e-01 -1.47712037e-01
-4.83712673e-01 5.75349987e-01 1.19258456e-01 9.81345713e-01
2.62327880e-01 9.87980723e-01 -8.84435773e-01 -2.35291272e-01
2.58488864e-01 1.01793277e+00 -7.87009895e-01 -3.85913104e-01
3.52834553e-01 4.60038900e-01 -6.43380284e-02 3.17352533e-01
6.31287336e-01 -4.70409214e-01 3.75394851e-01 -5.31395674e-01
3.35960276e-02 2.17995346e-01 -9.25850272e-01 1.68903029e+00
-6.83333874e-01 8.13126326e-01 7.53985941e-02 -1.17417467e+00
5.13143003e-01 3.77282292e-01 6.09149337e-01 -2.23560050e-01
3.00607979e-01 2.21005201e-01 4.50326085e-01 -1.28528342e-01
6.88173547e-02 -1.70570701e-01 2.61632532e-01 6.16507344e-02
6.35515630e-01 -2.32246593e-01 1.94883257e-01 -8.05247761e-03
1.35108829e+00 -2.67122388e-01 2.66413748e-01 -7.21032500e-01
4.54038888e-01 -1.39674097e-01 3.89553130e-01 5.50308526e-01
2.41875365e-01 9.00540769e-01 4.70754564e-01 -1.03962243e+00
-1.02232683e+00 -1.01894355e+00 -5.75258553e-01 7.50171781e-01
-2.81607836e-01 -5.83195537e-02 -5.86662471e-01 -6.01618767e-01
-3.38337794e-02 7.66067728e-02 -1.01167357e+00 -1.28658861e-01
-8.42523515e-01 -1.18885410e+00 6.35440290e-01 6.38743341e-01
1.47319883e-01 -7.06763983e-01 -8.25669885e-01 8.14617798e-02
2.98300743e-01 -1.02764881e+00 -3.35466832e-01 8.44469190e-01
-1.13590598e+00 -1.27824378e+00 -7.58660197e-01 -9.84135747e-01
1.35407734e+00 4.00272459e-01 9.08816218e-01 1.26140252e-01
-9.12603140e-01 1.61706924e-01 2.81456634e-02 -3.42363417e-01
-2.48607129e-01 3.83506984e-01 -2.81499624e-01 -1.33165255e-01
5.80046326e-02 -5.54072380e-01 -6.98211551e-01 3.52227062e-01
-1.16182244e+00 2.54844636e-01 7.66979814e-01 1.23407936e+00
6.51108980e-01 -2.67973393e-01 1.95615785e-03 -1.04787433e+00
1.17745437e-01 -1.34620955e-02 -3.43132794e-01 -5.54260835e-02
-4.02128650e-03 2.00324520e-01 5.81905782e-01 -3.69478941e-01
-5.78923047e-01 6.38643503e-01 -4.32850756e-02 -1.82217658e-01
-6.46244884e-02 3.89574021e-01 5.89272499e-01 -7.23879158e-01
8.66229475e-01 4.14503030e-02 4.42311943e-01 -2.21389949e-01
2.90772259e-01 9.88266841e-02 6.17644250e-01 9.81678022e-04
8.39816868e-01 1.07740116e+00 5.84040344e-01 -9.78952467e-01
-7.01138198e-01 -6.89922333e-01 -8.92500579e-01 1.47646703e-02
8.86802375e-01 -8.24092269e-01 -7.48669982e-01 6.33529723e-01
-9.40680027e-01 -2.94584721e-01 -3.04271430e-01 5.66051006e-01
-5.85891962e-01 1.12529576e-01 -7.81557918e-01 2.21236199e-01
-2.50448346e-01 -1.53516841e+00 8.74069214e-01 1.51153952e-01
-2.69673616e-01 -1.21756637e+00 -1.63832530e-01 -6.53354451e-02
6.41688466e-01 1.75828010e-01 9.84290123e-01 -5.43112278e-01
-4.63567197e-01 -3.45921516e-01 -2.20807284e-01 3.08827758e-01
1.20338500e-01 -2.34411489e-02 -9.34381485e-01 -7.13373661e-01
-1.67549014e-01 -2.21818298e-01 1.29978085e+00 5.76942980e-01
1.34765506e+00 -1.80608749e-01 -4.54311132e-01 1.10802031e+00
1.27378106e+00 -4.38530743e-01 6.44838631e-01 3.84955078e-01
7.18311012e-01 2.85455674e-01 -3.09793055e-01 -1.72207355e-01
-6.69635227e-03 4.81627494e-01 5.14304698e-01 -7.36727297e-01
-4.14140970e-01 2.21244454e-01 2.18938310e-02 6.52618885e-01
-3.41343492e-01 9.50428173e-02 -9.50855017e-01 5.42352080e-01
-1.59267962e+00 -6.75727129e-01 -1.54230431e-01 2.00455284e+00
7.52013445e-01 4.21711802e-02 -2.22744867e-01 1.44269839e-01
4.98736501e-01 -4.00735438e-02 -1.81917667e-01 -2.99426019e-01
5.82302595e-03 7.20326245e-01 1.14903176e+00 5.42044938e-01
-1.38719749e+00 5.28696835e-01 6.48642921e+00 5.49804688e-01
-1.55688524e+00 -4.11955342e-02 6.89590633e-01 -1.93635345e-01
-1.16301529e-01 -8.71062744e-03 -6.34609938e-01 -1.77678689e-01
6.78601146e-01 2.97542125e-01 2.25540161e-01 6.45311117e-01
1.74465477e-02 1.22041404e-01 -1.23771536e+00 7.18492091e-01
-2.94165965e-02 -1.67058575e+00 4.21583541e-02 2.19762817e-01
6.98604405e-01 4.50831205e-01 2.50942409e-01 -2.96296567e-01
2.15151161e-01 -1.43476856e+00 4.11746085e-01 3.62236738e-01
7.28218675e-01 -7.52783298e-01 7.67761827e-01 -6.96419179e-02
-8.33514214e-01 8.30234513e-02 -6.03566706e-01 2.67158747e-01
-2.12716088e-01 5.96254289e-01 -1.30886352e+00 4.62816954e-01
6.78356647e-01 6.65729761e-01 -6.03865325e-01 1.02304375e+00
1.58183679e-01 2.61482835e-01 -5.26392400e-01 9.02496129e-02
4.00910169e-01 2.55170435e-01 3.60040545e-01 1.45201683e+00
2.48716444e-01 -2.53058016e-01 -1.46982893e-01 3.59723479e-01
-1.60101265e-01 -2.11953342e-01 -3.13737601e-01 3.20372045e-01
-5.73324971e-02 1.68294203e+00 -1.20751774e+00 3.49513590e-02
-3.38000000e-01 6.31907165e-01 3.38457644e-01 8.52042735e-02
-4.25415099e-01 -2.76925504e-01 6.22330427e-01 4.43588942e-01
5.42447150e-01 -2.95488834e-01 -1.05754800e-01 -1.07476223e+00
-3.56776357e-01 -7.30735004e-01 2.03912199e-01 -2.88130850e-01
-8.92995834e-01 4.41318601e-01 -2.61282772e-01 -1.01875281e+00
7.32809156e-02 -1.18693829e+00 -3.93738210e-01 7.04152465e-01
-1.76707184e+00 -1.34978402e+00 -3.46052289e-01 6.01690292e-01
1.89753264e-01 9.30999070e-02 1.10899663e+00 1.99687302e-01
-2.21343651e-01 5.62411726e-01 8.24861377e-02 4.46352363e-01
8.26458037e-01 -1.27354860e+00 2.90925473e-01 6.00549996e-01
1.17744133e-02 9.50131655e-01 5.34619689e-01 -1.99181456e-02
-1.49931812e+00 -1.10474062e+00 6.87852502e-01 -2.03622729e-01
7.40410686e-01 -4.16572988e-01 -6.31024122e-01 8.62700641e-01
-1.87167451e-02 8.11963618e-01 8.67457509e-01 4.38693464e-02
-4.47503388e-01 -1.74364835e-01 -1.07443345e+00 5.54559350e-01
8.72431755e-01 -2.91039735e-01 -2.46253256e-02 4.66932327e-01
3.72591734e-01 -7.25563943e-01 -9.53222692e-01 5.55391371e-01
7.27395415e-01 -9.15832460e-01 1.20773745e+00 -5.18068373e-01
2.97349125e-01 -2.92861938e-01 2.02934161e-01 -1.31517649e+00
-5.81967056e-01 -4.77935791e-01 2.29065925e-01 2.80773818e-01
4.84140396e-01 -8.03662181e-01 8.30410838e-01 -3.38081606e-02
-5.52579105e-01 -9.83598173e-01 -1.27352035e+00 -5.76478362e-01
1.95100695e-01 6.28612787e-02 2.05519989e-01 1.01878643e+00
-2.49207705e-01 -2.94949878e-02 1.80016845e-01 7.92370066e-02
2.52482504e-01 -2.99832463e-01 4.95014340e-01 -1.12590313e+00
-4.42338884e-01 -5.89040995e-01 -1.01129460e+00 -8.48307908e-01
-1.32319868e-01 -1.33201063e+00 -1.23733260e-01 -1.20670581e+00
1.97230324e-01 -3.68742049e-01 -2.80168533e-01 9.06691074e-01
2.51401186e-01 6.50701761e-01 -6.10287897e-02 -6.09977394e-02
-1.39014229e-01 -1.78356066e-01 1.51906693e+00 -1.58719063e-01
1.31733611e-01 -7.00103641e-02 -5.47443151e-01 8.96446347e-01
5.87180793e-01 -3.96931976e-01 -2.88473547e-01 -5.42396426e-01
2.86661744e-01 -3.44113290e-01 6.89809740e-01 -9.89083827e-01
3.32224071e-01 1.59747019e-01 8.91913652e-01 -9.76428315e-02
2.23325849e-01 -6.26769662e-01 1.78177729e-01 8.42145801e-01
-2.22456396e-01 -6.37676716e-02 2.78698921e-01 4.17926490e-01
-1.07676506e-01 -1.76155120e-01 1.25106633e+00 -3.34007382e-01
-1.35287851e-01 4.02644038e-01 -4.24238414e-01 -5.14112830e-01
8.84035289e-01 -2.20888570e-01 -2.73552328e-01 1.55037284e-01
-9.87159610e-01 -5.90579137e-02 5.15379310e-01 7.49146715e-02
3.96258056e-01 -9.97077823e-01 -5.53092837e-01 4.37542021e-01
-8.93956348e-02 6.20742798e-01 5.32532394e-01 1.09608424e+00
-1.13355410e+00 4.86323148e-01 -3.33090305e-01 -7.51918137e-01
-1.26644874e+00 1.76147476e-01 7.38702536e-01 -6.33035600e-01
-8.00518751e-01 1.08418405e+00 2.31183454e-01 -5.80697715e-01
-4.30959724e-02 -4.70498025e-01 -2.46331990e-01 -8.03049579e-02
2.45939553e-01 4.52442653e-03 7.80263007e-01 -5.30911624e-01
-3.51509869e-01 4.43651855e-01 -5.59745252e-01 3.14945728e-01
1.73864734e+00 4.49029356e-01 -2.92107463e-01 -1.01444677e-01
1.37011206e+00 -3.57319444e-01 -1.13478565e+00 -1.00463443e-01
-3.17782581e-01 -1.09923005e-01 4.86969590e-01 -6.30116761e-01
-1.23565912e+00 8.14045191e-01 5.22344232e-01 1.55866286e-02
9.71780837e-01 -1.12953179e-01 4.61085498e-01 6.28274500e-01
-4.81295101e-02 -7.11269498e-01 -7.92770088e-02 6.69980109e-01
7.44627774e-01 -8.01292539e-01 1.63620859e-01 -6.40313327e-01
2.36416142e-02 1.65443730e+00 3.13991100e-01 -5.80576956e-01
8.37415755e-01 5.73783517e-01 3.44155617e-02 -2.76003957e-01
-5.06946325e-01 4.04683361e-03 3.54427278e-01 6.07821465e-01
6.20630443e-01 6.66519031e-02 -1.42786786e-01 2.13762727e-02
-3.70605350e-01 -1.63162500e-01 6.69081032e-01 1.05808890e+00
-2.57394642e-01 -1.07017457e+00 -3.98427807e-02 5.50487101e-01
-6.37123108e-01 1.87420174e-02 -6.02048412e-02 8.65014315e-01
-1.59864306e-01 1.09961405e-01 3.98358047e-01 1.16796322e-01
1.03335164e-01 -1.91397220e-01 9.43119884e-01 -7.53546715e-01
-5.51078737e-01 9.93797481e-02 -2.53895581e-01 -3.77059013e-01
-4.87787336e-01 -5.27378678e-01 -1.16838646e+00 -1.88768476e-01
-2.82983989e-01 -4.52238381e-01 7.35316038e-01 7.57641554e-01
1.50731757e-01 1.00290716e+00 3.45524102e-01 -9.91181970e-01
-6.62693441e-01 -7.29638815e-01 -3.26974094e-01 1.79649353e-01
5.46522021e-01 -5.33629715e-01 -1.33782774e-01 1.45548239e-01]
|
[15.030021667480469, -2.8232991695404053]
|
e1d84e5c-61e1-4e20-9295-7e5ad2439291
|
deepmatching-hierarchical-deformable-dense
|
1506.07656
| null |
http://arxiv.org/abs/1506.07656v2
|
http://arxiv.org/pdf/1506.07656v2.pdf
|
DeepMatching: Hierarchical Deformable Dense Matching
|
We introduce a novel matching algorithm, called DeepMatching, to compute
dense correspondences between images. DeepMatching relies on a hierarchical,
multi-layer, correlational architecture designed for matching images and was
inspired by deep convolutional approaches. The proposed matching algorithm can
handle non-rigid deformations and repetitive textures and efficiently
determines dense correspondences in the presence of significant changes between
images. We evaluate the performance of DeepMatching, in comparison with
state-of-the-art matching algorithms, on the Mikolajczyk (Mikolajczyk et al
2005), the MPI-Sintel (Butler et al 2012) and the Kitti (Geiger et al 2013)
datasets. DeepMatching outperforms the state-of-the-art algorithms and shows
excellent results in particular for repetitive textures.We also propose a
method for estimating optical flow, called DeepFlow, by integrating
DeepMatching in the large displacement optical flow (LDOF) approach of Brox and
Malik (2011). Compared to existing matching algorithms, additional robustness
to large displacements and complex motion is obtained thanks to our matching
approach. DeepFlow obtains competitive performance on public benchmarks for
optical flow estimation.
|
['Cordelia Schmid', 'Zaid Harchaoui', 'Jerome Revaud', 'Philippe Weinzaepfel']
|
2015-06-25
| null | null | null | null |
['dense-pixel-correspondence-estimation']
|
['computer-vision']
|
[-3.17462504e-01 -5.90048492e-01 -1.15122125e-01 -8.46610218e-02
-3.53272468e-01 -5.44382572e-01 6.59022629e-01 -1.05359234e-01
-3.48636031e-01 4.22486156e-01 3.98754537e-01 1.48574427e-01
-2.71497905e-01 -8.79927456e-01 -6.46298110e-01 -3.75367552e-01
-3.71194303e-01 4.52622294e-01 3.34689856e-01 -2.63502032e-01
5.97682774e-01 8.80834043e-01 -1.65360606e+00 3.44249696e-01
5.05279005e-01 1.03317511e+00 4.38256301e-02 1.01255357e+00
8.13813694e-03 9.77249801e-01 -1.19842432e-01 -5.12915790e-01
7.10496902e-01 -2.76022762e-01 -1.31785083e+00 -6.27548844e-02
1.50223029e+00 -6.83284223e-01 -7.67935991e-01 7.63584554e-01
5.90460300e-01 3.78996372e-01 2.55866528e-01 -1.08773077e+00
-6.19027317e-01 1.90352559e-01 -5.28045058e-01 5.95893323e-01
5.76827645e-01 4.53199565e-01 9.68734741e-01 -9.78902578e-01
1.07519674e+00 1.48943102e+00 1.00661910e+00 3.11796069e-01
-1.23007441e+00 -4.69267339e-01 -2.79966325e-01 6.31245732e-01
-1.12372386e+00 -5.18699288e-01 4.64057297e-01 -7.43742049e-01
1.01908588e+00 1.00301839e-01 6.20908856e-01 6.64211392e-01
2.74627030e-01 6.47045791e-01 8.13276470e-01 -2.53932863e-01
-2.75888771e-01 -7.93580830e-01 -1.69081643e-01 1.00425065e+00
4.32660279e-04 5.41449904e-01 -5.57261050e-01 -6.79794252e-02
1.17214620e+00 -1.73524600e-02 -3.53907079e-01 -2.41201013e-01
-1.80545259e+00 5.20509183e-01 7.35817432e-01 3.70043427e-01
-4.03965026e-01 5.27914524e-01 5.53385258e-01 4.86816823e-01
4.18485016e-01 3.39046299e-01 -8.51010531e-02 -2.28259474e-01
-1.13826454e+00 5.36077738e-01 7.26798534e-01 8.38693500e-01
1.19718313e+00 -1.74388394e-01 -4.92042661e-01 5.04691124e-01
1.26173854e-01 3.28142673e-01 6.44593894e-01 -1.68940306e+00
5.36290169e-01 2.50975072e-01 3.28721516e-02 -1.47670424e+00
-4.79083806e-01 -1.75629988e-01 -1.13470459e+00 2.39092067e-01
6.88349307e-01 1.64596081e-01 -6.58874035e-01 1.43920875e+00
2.66240120e-01 9.42163944e-01 -1.45487057e-03 1.19539428e+00
9.93033707e-01 3.44261527e-01 -3.00457239e-01 2.58946538e-01
9.85856771e-01 -1.44573152e+00 -4.45514262e-01 1.04911953e-01
4.93118167e-01 -1.38799620e+00 6.33492649e-01 7.76527897e-02
-1.37885559e+00 -9.26456153e-01 -6.37191892e-01 -5.21340549e-01
-6.42011613e-02 -1.51133627e-01 8.56809318e-01 2.28302747e-01
-1.46043813e+00 1.20009363e+00 -7.89180577e-01 -5.65916181e-01
4.36702311e-01 4.27923411e-01 -7.73839951e-01 -1.57524109e-01
-9.20915961e-01 7.84218967e-01 -1.01762861e-01 2.94379115e-01
-6.16100669e-01 -1.10060740e+00 -1.01360369e+00 -7.01597631e-02
-1.94463477e-01 -1.23130143e+00 9.32334065e-01 -8.11037898e-01
-1.69584560e+00 1.22171819e+00 -3.18784028e-01 -4.94389087e-01
1.00394642e+00 -2.21958518e-01 5.93556948e-02 4.36319530e-01
3.38002965e-02 1.02112377e+00 6.80326581e-01 -7.70001531e-01
-4.59193379e-01 -4.28373367e-02 1.65372580e-01 -1.39130369e-01
1.80083692e-01 5.66376261e-02 -4.07996356e-01 -8.27138901e-01
-2.46941112e-03 -1.03548026e+00 -2.53194660e-01 3.66634071e-01
-2.15585455e-01 1.14557019e-03 7.79371619e-01 -4.84670401e-01
8.35436761e-01 -1.83672380e+00 3.23268771e-01 1.34427413e-01
4.35779959e-01 4.38148022e-01 -4.63559985e-01 4.17013824e-01
-9.95758399e-02 -2.91097760e-01 -1.73538044e-01 -6.67078435e-01
3.46984416e-02 1.55070633e-01 -2.20944077e-01 9.29869413e-01
4.89326492e-02 1.01341665e+00 -9.95178401e-01 -4.54182893e-01
7.02931285e-01 5.56882679e-01 -7.84847081e-01 3.93044740e-01
2.34650090e-01 7.09710956e-01 7.63914958e-02 6.64444625e-01
9.23998535e-01 -2.47597739e-01 -3.20625484e-01 -5.59932947e-01
-3.77769709e-01 1.67396218e-01 -1.49194360e+00 2.02849364e+00
-4.65190232e-01 9.46872056e-01 1.07020684e-01 -8.91431034e-01
9.12694097e-01 2.05504447e-01 8.75054359e-01 -6.84469700e-01
1.04949869e-01 3.74543637e-01 -7.69076943e-02 -5.68857551e-01
6.45302594e-01 2.61869907e-01 4.45313483e-01 5.81095159e-01
7.78001919e-02 -4.30210605e-02 6.46091759e-01 6.11583032e-02
1.26602972e+00 3.24765950e-01 5.86134605e-02 -5.83909333e-01
1.03935993e+00 -1.42946869e-01 5.19895792e-01 8.19883704e-01
-6.46346629e-01 9.93389189e-01 1.67527720e-01 -1.08117068e+00
-9.40831959e-01 -9.49380100e-01 -1.40949160e-01 6.96484506e-01
3.72796893e-01 -2.99278527e-01 -5.53193569e-01 -2.10055172e-01
4.09838796e-01 -5.56843638e-01 -7.87707984e-01 1.78557426e-01
-1.06547451e+00 -2.78810561e-01 6.35175884e-01 4.30668205e-01
9.31820691e-01 -1.32480919e+00 -6.45895362e-01 3.50194782e-01
-3.11764270e-01 -1.48830545e+00 -7.75210381e-01 -7.07947731e-01
-8.22283208e-01 -1.41778076e+00 -7.88230836e-01 -8.32984746e-01
3.87522221e-01 2.28325278e-01 1.50790894e+00 4.55061674e-01
-6.11929834e-01 4.73006845e-01 -1.97132736e-01 3.80725771e-01
-3.90169680e-01 1.27005458e-01 -1.08917885e-01 2.34021083e-01
-6.82481751e-03 -6.59181416e-01 -1.04551125e+00 5.72428703e-01
-8.50967646e-01 -1.54546574e-01 2.85133064e-01 7.52419770e-01
5.43628514e-01 -6.63071275e-01 2.08854564e-02 -5.38108408e-01
2.23499388e-01 -2.08576009e-01 -7.90559053e-01 -8.89299214e-02
-2.62719184e-01 1.62122414e-01 5.65523386e-01 -1.95784852e-01
-6.70364738e-01 6.83491975e-02 -2.69509941e-01 -8.30264688e-01
-1.54578120e-01 -8.66948739e-02 4.62226480e-01 -8.22583973e-01
4.25906897e-01 -9.78249088e-02 1.19332388e-01 -4.64547902e-01
3.77580523e-01 1.80107400e-01 9.77575660e-01 -6.55136347e-01
8.38104129e-01 1.03929043e+00 5.38747251e-01 -4.49008703e-01
-5.67093849e-01 -8.27259898e-01 -1.16824126e+00 -2.17257828e-01
7.81898141e-01 -8.09855521e-01 -1.05228257e+00 9.93340790e-01
-1.40929365e+00 -5.54615736e-01 -2.30619684e-01 6.54200613e-01
-9.59946334e-01 7.24152923e-01 -1.02658617e+00 4.19874676e-03
-5.32335699e-01 -1.34396970e+00 1.26797998e+00 3.90240587e-02
-3.09704244e-01 -1.46709538e+00 6.04406476e-01 2.44051740e-01
7.94425607e-01 6.19278312e-01 2.03485236e-01 1.52770102e-01
-9.27542508e-01 3.51011455e-02 -4.95839328e-01 1.67213708e-01
1.31231025e-01 3.48697960e-01 -8.16292346e-01 -4.19946492e-01
-5.52634597e-01 -1.93156615e-01 1.08821559e+00 4.00949806e-01
9.86744165e-01 -6.94629923e-02 2.33099721e-02 1.24533534e+00
1.47274566e+00 -4.83911544e-01 1.03704965e+00 6.67465270e-01
9.62105215e-01 6.17434680e-01 3.63900691e-01 4.02374268e-01
5.86107850e-01 9.32331622e-01 6.09211147e-01 -3.12135309e-01
-6.02215707e-01 2.68233359e-01 1.79148763e-01 7.73078740e-01
-4.19250876e-01 1.72096148e-01 -6.52332366e-01 6.35110319e-01
-2.14067364e+00 -1.33825839e+00 -5.66604376e-01 1.99710214e+00
6.99799001e-01 -3.88513654e-01 6.67911246e-02 -8.39346871e-02
7.49345601e-01 3.81836832e-01 -1.26901940e-01 -5.62103391e-01
-3.34410727e-01 5.54286480e-01 5.33000052e-01 8.22164118e-01
-1.25167823e+00 1.07645619e+00 6.03051138e+00 2.00935304e-01
-1.14925826e+00 5.83894961e-02 3.51737440e-01 9.06379223e-02
1.57221168e-01 -2.91095469e-02 -5.39029837e-01 4.67293680e-01
5.72215617e-01 -2.36523468e-02 5.50726354e-01 1.69400916e-01
2.31886104e-01 4.61015329e-02 -1.12613451e+00 1.14823186e+00
1.33039013e-01 -2.00033021e+00 -1.15758546e-01 -7.00470805e-02
1.02137744e+00 4.60658997e-01 -9.80767310e-02 -2.49986768e-01
2.59053856e-01 -8.53473961e-01 5.49941063e-01 7.68573225e-01
5.35940051e-01 -4.10853624e-01 8.57681394e-01 -3.47186178e-01
-1.49794793e+00 2.36770734e-01 -4.91218656e-01 -1.37096152e-01
8.96512717e-02 5.95890582e-01 5.62289031e-03 5.96079409e-01
1.02880061e+00 1.43723750e+00 -4.82950956e-01 1.38032866e+00
1.19029440e-01 6.16002716e-02 -2.18729496e-01 6.42059803e-01
5.52907288e-01 -3.22279423e-01 4.11519289e-01 1.52653503e+00
2.39264414e-01 -3.59556347e-01 1.80146724e-01 8.79583597e-01
-2.12001354e-01 1.02113806e-01 -4.73140150e-01 4.85270947e-01
1.95551634e-01 1.50656760e+00 -4.05401051e-01 -3.88827652e-01
-3.83711427e-01 1.09661126e+00 3.75681400e-01 1.60020992e-01
-4.09496933e-01 -3.14072192e-01 1.14708650e+00 3.29476483e-02
4.60281074e-01 -3.53253365e-01 6.10572696e-02 -1.29417288e+00
4.98127639e-02 -5.38132727e-01 5.09352982e-01 -6.36032164e-01
-1.43009591e+00 4.59506571e-01 -2.48938650e-01 -1.41793787e+00
-3.60840321e-01 -5.14062047e-01 -8.05635154e-01 9.79690790e-01
-2.07552552e+00 -1.10912764e+00 -9.60500181e-01 1.07506299e+00
2.42472410e-01 -4.66219150e-03 6.51658952e-01 7.49245346e-01
-3.40569466e-01 4.74109560e-01 -1.04983372e-03 5.72227836e-01
1.17166913e+00 -1.18911552e+00 9.06498790e-01 9.38202441e-01
1.12134598e-01 3.78551006e-01 2.46605113e-01 -2.71567762e-01
-1.49011707e+00 -1.13369584e+00 1.14063609e+00 -3.49866837e-01
8.62481415e-01 4.77432236e-02 -9.48546052e-01 7.30396986e-01
4.50136870e-01 8.61046612e-01 2.81465352e-01 -5.75493038e-01
-5.73233783e-01 -2.15918794e-01 -1.15463114e+00 3.16879153e-01
1.32498705e+00 -4.79722530e-01 -2.37449452e-01 3.35036933e-01
2.72402465e-01 -8.67584348e-01 -1.21569896e+00 3.42598438e-01
7.13944614e-01 -1.60098767e+00 1.22781825e+00 -5.53969681e-01
6.36183083e-01 -3.63667578e-01 -2.03897413e-02 -1.14653969e+00
-6.28455460e-01 -1.22328067e+00 -1.71292365e-01 8.91936600e-01
-3.18035781e-01 -5.59751809e-01 7.12727368e-01 6.73039034e-02
-1.68983698e-01 -2.76057690e-01 -9.07494068e-01 -8.32742989e-01
7.87255354e-03 -1.29883841e-01 7.19209671e-01 1.14940321e+00
-3.64680469e-01 -3.34116250e-01 -3.23375732e-01 2.73428392e-02
7.87764788e-01 4.14484531e-01 1.15115917e+00 -1.11032927e+00
-2.12200135e-01 -7.71192074e-01 -1.12711060e+00 -9.00248587e-01
5.95595181e-01 -9.97025192e-01 -2.80760378e-01 -1.45162451e+00
-1.48533598e-01 -4.25078183e-01 -9.38192084e-02 2.77947068e-01
-7.38700405e-02 7.98004031e-01 4.83428836e-01 4.89688307e-01
-5.25029063e-01 1.81874692e-01 1.64196682e+00 -1.06440201e-01
5.03557771e-02 -2.32003748e-01 5.96824475e-03 4.32057589e-01
4.76114929e-01 -8.97259787e-02 1.76476285e-01 -7.51145363e-01
-8.46859738e-02 -1.17351763e-01 6.94866896e-01 -1.15252149e+00
4.77334470e-01 1.70930792e-02 8.85211080e-02 -3.62354845e-01
-8.35605115e-02 -5.83922923e-01 1.82475641e-01 7.42004514e-01
-2.89308518e-01 7.16974258e-01 1.56113848e-01 7.14625567e-02
-4.70546305e-01 8.03570300e-02 1.02902913e+00 -1.25155926e-01
-9.68179345e-01 7.42796183e-01 -9.99683421e-03 3.24897557e-01
7.16539443e-01 -1.46092445e-01 -5.80171287e-01 -2.59120941e-01
-4.73020613e-01 7.63659626e-02 4.69788581e-01 5.85467458e-01
5.09205878e-01 -1.65979695e+00 -9.90140021e-01 4.69337672e-01
-1.76162925e-02 1.29648326e-02 9.52962488e-02 1.29140389e+00
-1.40181327e+00 2.65396923e-01 -6.55251801e-01 -7.85407603e-01
-1.07740211e+00 9.74557102e-02 7.39046335e-01 -3.22558850e-01
-8.10296237e-01 6.57375157e-01 -6.92498311e-02 -4.28167403e-01
8.76398310e-02 -4.15951014e-01 8.37755799e-02 -6.40864894e-02
7.40234792e-01 7.63047040e-01 3.20175499e-01 -9.47122872e-01
-5.38219154e-01 1.18176067e+00 2.65321821e-01 3.49939555e-01
1.15648448e+00 -1.44778058e-01 -5.16663134e-01 -1.31197155e-01
1.63154125e+00 -1.80907667e-01 -1.37130928e+00 -3.73176634e-01
-1.71468347e-01 -9.62855160e-01 5.80874234e-02 -1.98708415e-01
-1.69092250e+00 9.61497545e-01 5.91374159e-01 -2.97661033e-02
8.33828866e-01 -3.30316961e-01 1.29297924e+00 2.68534988e-01
1.55423179e-01 -7.66290426e-01 7.14331120e-02 7.52021968e-01
7.49897540e-01 -1.36982274e+00 2.32459232e-02 -2.81821221e-01
-4.35272492e-02 1.65918100e+00 4.69138443e-01 -6.29310250e-01
6.79924488e-01 3.09677601e-01 2.77560592e-01 -6.21078201e-02
-5.88709176e-01 -4.75313991e-01 5.57026148e-01 5.60255527e-01
4.64581758e-01 -3.30538332e-01 -2.72579733e-02 -6.87470496e-01
-1.15757748e-01 1.15643449e-01 5.11264563e-01 8.09055448e-01
1.38034821e-02 -1.23049808e+00 -4.44986761e-01 7.12947920e-02
-3.39924455e-01 -1.04379892e-01 -2.44381428e-01 9.27691579e-01
3.62066850e-02 6.43272042e-01 6.91681027e-01 -2.24661052e-01
4.78739321e-01 -5.76115787e-01 8.13003004e-01 -1.81655139e-01
-1.18896925e+00 -2.15316817e-01 -2.19214395e-01 -1.40155590e+00
-1.13297498e+00 -6.58793151e-01 -1.07855213e+00 -9.93018627e-01
2.82031387e-01 -2.51954794e-01 2.67394453e-01 8.89626563e-01
5.72769344e-01 1.86052144e-01 6.92193747e-01 -1.37815344e+00
-2.36876775e-02 -6.12760484e-01 -2.13051140e-01 9.15572166e-01
6.46878064e-01 -5.95372796e-01 -3.93913031e-01 1.84281603e-01]
|
[8.827816009521484, -1.895565390586853]
|
05020b4b-826d-4f82-8ead-d1f4e8716f3e
|
dreambooth3d-subject-driven-text-to-3d
|
2303.13508
| null |
https://arxiv.org/abs/2303.13508v2
|
https://arxiv.org/pdf/2303.13508v2.pdf
|
DreamBooth3D: Subject-Driven Text-to-3D Generation
|
We present DreamBooth3D, an approach to personalize text-to-3D generative models from as few as 3-6 casually captured images of a subject. Our approach combines recent advances in personalizing text-to-image models (DreamBooth) with text-to-3D generation (DreamFusion). We find that naively combining these methods fails to yield satisfactory subject-specific 3D assets due to personalized text-to-image models overfitting to the input viewpoints of the subject. We overcome this through a 3-stage optimization strategy where we jointly leverage the 3D consistency of neural radiance fields together with the personalization capability of text-to-image models. Our method can produce high-quality, subject-specific 3D assets with text-driven modifications such as novel poses, colors and attributes that are not seen in any of the input images of the subject.
|
['Varun Jampani', 'Yuanzhen Li', 'Jonathan Barron', 'Michael Rubinstein', 'Kfir Aberman', 'Shiran Zada', 'Ben Mildenhall', 'Nataniel Ruiz', 'Michael Niemeyer', 'Ben Poole', 'Srinivas Kaza', 'Amit Raj']
|
2023-03-23
| null | null | null | null |
['text-to-3d']
|
['computer-vision']
|
[ 1.14813946e-01 9.91170704e-02 5.25137067e-01 -6.63099647e-01
-8.15336168e-01 -7.44378030e-01 8.94428730e-01 -5.88994503e-01
3.72387096e-02 4.60011512e-01 2.39885896e-01 2.09654853e-01
4.66223657e-02 -4.33095813e-01 -8.08644831e-01 -5.67823768e-01
2.09128693e-01 7.62421370e-01 7.00259656e-02 -1.44556895e-01
-8.11712146e-02 7.89473891e-01 -1.55832827e+00 2.46784732e-01
6.11870766e-01 7.90755808e-01 2.97521830e-01 8.46423686e-01
2.02350721e-01 2.17784703e-01 -4.35602307e-01 -3.78531337e-01
7.39461362e-01 -4.61564839e-01 -2.62724459e-01 6.66727364e-01
1.05464721e+00 -3.78589183e-01 -4.07991111e-01 6.38241172e-01
5.78645885e-01 4.88721251e-01 6.71337724e-01 -1.07610559e+00
-1.02629066e+00 -4.82841432e-02 -6.21206522e-01 -8.86619911e-02
1.62114963e-01 4.68773186e-01 6.02831841e-01 -9.20603395e-01
9.83336568e-01 1.38563097e+00 5.67312717e-01 5.93918025e-01
-1.63099134e+00 -3.21229219e-01 1.88893780e-01 -1.59475744e-01
-1.37561691e+00 -5.25323689e-01 8.50036263e-01 -4.65290695e-01
8.09452534e-01 3.81579250e-01 9.59952176e-01 1.25346303e+00
5.28537273e-01 3.01200658e-01 1.28442717e+00 -6.94246590e-02
2.65437901e-01 5.26316702e-01 -5.93192875e-01 5.73879480e-01
-2.40512639e-02 -3.00853476e-02 -9.68318403e-01 -7.32946694e-02
9.80995476e-01 -1.52865097e-01 -1.09330505e-01 -7.07043946e-01
-1.11190593e+00 4.97421056e-01 4.94735807e-01 -2.35873416e-01
-5.35729170e-01 1.85603067e-01 -3.65443945e-01 -1.20763645e-01
8.11474085e-01 6.18825793e-01 -4.63928521e-01 2.20049024e-01
-1.02555919e+00 6.32878721e-01 1.69752166e-01 1.22761869e+00
9.10007477e-01 3.47249866e-01 -3.36654633e-01 4.27869022e-01
3.93147200e-01 9.39289331e-01 1.25985801e-01 -1.13310516e+00
4.26446497e-02 2.82750905e-01 2.42668152e-01 -6.78458989e-01
-3.38234276e-01 -5.62120378e-01 -4.40846950e-01 4.77556884e-01
-9.05648544e-02 -2.47821167e-01 -1.47834408e+00 1.67448997e+00
6.92924142e-01 -2.65093327e-01 -1.29356846e-01 9.53549445e-01
8.39758158e-01 6.27664745e-01 2.16876566e-02 1.34677544e-01
1.28414345e+00 -8.07304740e-01 -4.19844747e-01 -5.77351928e-01
-1.37876377e-01 -9.64374602e-01 1.08208632e+00 1.71324402e-01
-1.50087821e+00 -6.05741620e-01 -7.98671603e-01 -2.95430630e-01
-3.10723066e-01 -2.21364528e-01 6.84544683e-01 6.96826577e-01
-1.23482835e+00 4.55630988e-01 -5.89672685e-01 -5.28824449e-01
5.77562928e-01 8.54392499e-02 -3.83343518e-01 -1.31533384e-01
-6.63994730e-01 1.04073834e+00 9.80092362e-02 -3.21674258e-01
-1.23159075e+00 -1.20199978e+00 -7.02178299e-01 -1.99189216e-01
2.98938572e-01 -1.48781240e+00 1.17018592e+00 -7.40361392e-01
-1.44057512e+00 1.09887516e+00 -2.52824575e-01 1.19155459e-02
5.44324279e-01 -1.02308802e-01 -1.76341265e-01 1.79151490e-01
-3.74567881e-02 1.12310541e+00 1.28901482e+00 -1.68240356e+00
-2.44716153e-01 -4.69995439e-01 -1.13575496e-01 7.14031339e-01
-1.02524161e-01 -3.12092811e-01 -8.68557632e-01 -7.58937180e-01
2.05762208e-01 -9.59817111e-01 -3.51018220e-01 3.97244513e-01
-4.14909005e-01 3.81041318e-01 8.70197833e-01 -5.00187159e-01
3.33709538e-01 -2.09724784e+00 2.95291215e-01 4.86975089e-02
2.78736144e-01 -1.39247030e-01 -2.52550691e-01 2.28286043e-01
-1.03628479e-01 -6.39506653e-02 1.09656760e-02 -9.87806737e-01
2.31844887e-01 1.82985887e-01 -2.10440934e-01 3.30693394e-01
1.18299372e-01 1.12502623e+00 -8.09100509e-01 -4.11033750e-01
5.18289864e-01 8.13619256e-01 -7.45879114e-01 3.99260402e-01
-7.06497133e-01 7.62066305e-01 -3.50249350e-01 4.48794276e-01
9.25595164e-01 -3.48271549e-01 -1.57852948e-01 -5.26022613e-01
1.82466373e-01 -1.44454762e-01 -7.49987721e-01 1.99637473e+00
-3.94549072e-01 4.48465943e-01 1.01446316e-01 -5.58577962e-02
6.64699316e-01 8.01174492e-02 7.19492078e-01 -4.62620378e-01
-1.01999283e-01 -2.66235530e-01 -4.44186449e-01 -4.74956274e-01
7.80487955e-01 -4.65941966e-01 -1.60491858e-02 6.00156546e-01
4.16489631e-01 -8.98598254e-01 -2.74742097e-01 4.32822138e-01
6.15552783e-01 6.16090655e-01 -1.48116231e-01 -1.90065846e-01
-3.05045277e-01 -5.91754615e-02 2.20208570e-01 8.23077381e-01
2.53907293e-01 1.26025391e+00 -9.98145416e-02 -4.31537360e-01
-1.41840839e+00 -1.56853604e+00 -6.69263825e-02 8.05889368e-01
-1.31586850e-01 -4.46915925e-01 -6.49842322e-01 -6.08917713e-01
-2.63845399e-02 9.31107819e-01 -8.48450184e-01 -9.04669091e-02
-5.07074185e-02 -9.43054378e-01 6.12850077e-02 2.29441047e-01
4.50096756e-01 -6.71464801e-01 -6.11980319e-01 -7.07224905e-02
-8.54685828e-02 -1.16369951e+00 -8.82344007e-01 1.83626667e-01
-8.33359420e-01 -4.91845250e-01 -7.67782092e-01 -1.44977927e-01
1.01616228e+00 4.83072877e-01 1.35743570e+00 -2.47949347e-01
-4.37436312e-01 7.34184384e-01 -2.25879848e-01 -5.44317484e-01
-3.65364403e-01 -3.05293351e-01 -3.56558827e-03 1.98847666e-01
-1.81460798e-01 -7.90787816e-01 -7.37445295e-01 2.26731598e-01
-1.06624830e+00 5.06416142e-01 5.56790471e-01 4.31995898e-01
6.87901080e-01 6.54158294e-02 -1.01473399e-01 -8.24976385e-01
4.28761765e-02 -1.18544213e-01 -5.94478190e-01 2.42086127e-01
-6.13923788e-01 -9.85334534e-03 3.54843229e-01 -5.07384896e-01
-1.44913328e+00 2.98996508e-01 1.39647320e-01 -8.27926636e-01
-2.18504190e-01 -1.18090689e-01 -4.27100092e-01 -3.19951892e-01
9.19839799e-01 2.00430483e-01 -1.15783736e-01 -4.79505718e-01
8.04200649e-01 1.06287912e-01 6.61033750e-01 -6.07318759e-01
1.31489849e+00 6.62085116e-01 2.71011531e-01 -6.71144128e-01
-9.15154219e-01 -1.00807570e-01 -6.81496799e-01 -2.95069873e-01
1.06506681e+00 -1.01457071e+00 -3.38970512e-01 6.13944888e-01
-9.34373498e-01 -5.89974284e-01 -6.56245232e-01 2.01436177e-01
-7.02680409e-01 1.21671319e-01 -9.98025462e-02 -5.34935713e-01
-2.45227009e-01 -1.00283635e+00 1.47362339e+00 4.30520356e-01
-1.14223167e-01 -1.01418972e+00 -5.90858236e-02 6.53742969e-01
5.12102425e-01 2.64823943e-01 8.67331505e-01 7.71393552e-02
-9.87111628e-01 1.59718487e-02 2.41528079e-03 2.18284190e-01
1.58126950e-01 -5.76707274e-02 -1.30643845e+00 -2.31426254e-01
-2.34849676e-02 -6.65125325e-02 5.40512919e-01 5.65159380e-01
1.09751475e+00 -2.91835755e-01 -1.90556824e-01 1.13651454e+00
1.28773642e+00 -3.45883742e-02 7.57852614e-01 -1.49342148e-02
9.03977036e-01 6.11220300e-01 2.15706632e-01 6.11568451e-01
4.72679496e-01 9.91590202e-01 3.78396899e-01 -4.51250404e-01
-4.99740720e-01 -5.80708206e-01 2.25293055e-01 7.03434274e-02
-8.82785693e-02 -5.46536148e-01 -4.06577080e-01 4.10160840e-01
-1.39066124e+00 -9.57368910e-01 9.29166228e-02 2.21087933e+00
7.19214499e-01 -1.28077164e-01 -1.28872022e-01 -8.27143312e-01
2.46092394e-01 2.65556872e-01 -8.79164219e-01 -6.72487766e-02
-3.29528779e-01 2.00746417e-01 6.07303977e-01 5.21404147e-01
-7.39486754e-01 1.12800908e+00 7.29912615e+00 6.60083175e-01
-9.04861629e-01 2.84729302e-01 7.39433169e-01 -5.71606994e-01
-9.83069003e-01 1.64064094e-01 -6.81020677e-01 1.84999660e-01
5.04319549e-01 -6.82622790e-02 6.17793679e-01 6.38767123e-01
3.14228743e-01 -2.14787707e-01 -1.11520648e+00 1.08369625e+00
4.97457892e-01 -1.36981022e+00 4.23467934e-01 2.48970523e-01
1.24309194e+00 -1.29373502e-02 5.84235847e-01 -1.56195804e-01
4.68537211e-01 -9.15352345e-01 1.01698387e+00 9.80984926e-01
8.81216943e-01 -4.78253245e-01 -7.14074001e-02 1.44890836e-02
-6.47223294e-01 3.38894457e-01 -2.99458325e-01 5.49686432e-01
5.31749725e-01 5.80284297e-01 -1.01502812e+00 5.75982928e-01
1.17962170e+00 5.33543527e-01 -9.30937886e-01 7.02008724e-01
-7.80546740e-02 2.65460741e-02 -3.81970882e-01 4.20944393e-01
6.89820200e-03 -1.58008531e-01 8.13381076e-01 6.45756066e-01
5.58359444e-01 1.57100946e-01 -1.30259842e-01 1.31462002e+00
1.32632116e-02 -3.34582061e-01 -6.60807967e-01 1.56427354e-01
1.56927064e-01 1.44376290e+00 -4.64142412e-01 -2.78299034e-01
-8.14702064e-02 1.36925352e+00 -8.96058977e-02 5.32033503e-01
-5.80389142e-01 1.62908912e-01 7.86329269e-01 3.54641050e-01
5.38191319e-01 -2.98095375e-01 -3.90852600e-01 -1.15697801e+00
-6.56759366e-02 -7.84607947e-01 5.00045419e-02 -1.99795115e+00
-1.53759420e+00 6.99697316e-01 4.02999580e-01 -9.82453942e-01
-1.97209626e-01 -3.94945532e-01 -4.75850791e-01 1.09112179e+00
-9.98133898e-01 -1.76527071e+00 -5.59640348e-01 7.23516345e-01
4.45459962e-01 -1.07549243e-01 6.85270250e-01 1.06289063e-03
-3.10361013e-02 4.30530161e-01 1.05097834e-02 -6.03021324e-01
1.04154360e+00 -1.26584518e+00 9.42724288e-01 9.26363409e-01
2.70516217e-01 3.29050571e-01 9.41705644e-01 -7.59079814e-01
-1.41439605e+00 -1.05446863e+00 6.06057405e-01 -1.19384181e+00
1.10362098e-01 -5.92908084e-01 -5.09256244e-01 9.55915928e-01
4.67355520e-01 4.82754363e-03 3.29718173e-01 2.49712858e-02
-1.82291150e-01 -3.40126604e-01 -1.24114525e+00 9.19457853e-01
1.16788304e+00 -3.89745057e-01 -1.89028278e-01 5.55901110e-01
6.62697315e-01 -8.70756567e-01 -8.73848259e-01 4.72193658e-02
3.90696883e-01 -1.07965875e+00 1.25927401e+00 -4.50817943e-01
1.51921302e-01 -4.68508989e-01 -2.68873870e-01 -1.69517124e+00
-5.03064394e-01 -9.38866198e-01 2.71856606e-01 1.00635278e+00
2.57146627e-01 -3.63115042e-01 8.52151811e-01 9.30975020e-01
-3.90287250e-01 -3.01050723e-01 -6.99805319e-01 -4.84855413e-01
-4.61093038e-02 -3.16635698e-01 9.33694601e-01 6.59410954e-01
-6.53533936e-01 2.86782473e-01 -7.89675832e-01 3.03874552e-01
9.91300285e-01 1.59704328e-01 1.24931991e+00 -8.94874215e-01
-4.90375429e-01 -1.07323952e-01 -1.91416621e-01 -1.12933123e+00
-2.24850520e-01 -8.01248074e-01 1.32299721e-01 -1.36913228e+00
3.21227193e-01 -4.37468410e-01 3.21526349e-01 5.25879145e-01
-1.04343794e-01 5.76608181e-01 3.13276589e-01 8.05547461e-02
-4.12388533e-01 8.91908705e-01 1.73723364e+00 3.46020097e-03
-1.85139671e-01 -3.51770341e-01 -8.69566560e-01 4.65835541e-01
4.83894795e-01 -4.10474092e-01 -6.31932378e-01 -7.46708691e-01
3.50631922e-01 -6.36635274e-02 8.47634494e-01 -9.52601433e-01
-1.54530287e-01 -4.06529814e-01 1.15747356e+00 -9.01321352e-01
9.18398380e-01 -8.47306848e-01 8.94660830e-01 -3.57252866e-01
-5.32242432e-02 2.47357246e-02 3.01869869e-01 4.66606647e-01
5.76200962e-01 5.87250888e-02 8.32095563e-01 -5.20147741e-01
-5.62516987e-01 6.41991377e-01 -2.74154633e-01 -6.79154396e-02
9.43689287e-01 -2.07396716e-01 -3.20574671e-01 -7.14808106e-01
-7.66757786e-01 1.09789006e-01 9.71888542e-01 5.04055798e-01
5.39974213e-01 -1.31677449e+00 -7.14933157e-01 4.35855299e-01
2.21913800e-01 2.23954484e-01 7.57305562e-01 4.35390711e-01
-3.15331101e-01 1.88686326e-01 -2.96528488e-01 -7.79601395e-01
-1.11837637e+00 3.72160226e-01 6.17407739e-01 1.39167070e-01
-8.84345114e-01 8.92643273e-01 6.65834010e-01 -4.66013610e-01
-1.81038603e-01 6.22307323e-02 5.04854083e-01 -3.96318913e-01
1.69093996e-01 -1.03126906e-01 9.81057063e-02 -7.08400071e-01
-1.85254872e-01 8.04848969e-01 -1.62385985e-01 -6.10875487e-01
1.47596729e+00 -4.29888517e-01 2.68402696e-01 2.25621611e-01
8.43883276e-01 1.56205043e-01 -2.02519655e+00 -2.53433406e-01
-1.04976916e+00 -9.64817047e-01 2.97051251e-01 -1.09926236e+00
-1.23251867e+00 5.69135427e-01 6.47083044e-01 -3.91491294e-01
1.04074597e+00 2.89624393e-01 6.93336785e-01 2.79490381e-01
3.81867230e-01 -9.88300204e-01 3.07889700e-01 -6.95396494e-03
1.00313604e+00 -1.04726636e+00 4.82493460e-01 -7.97363147e-02
-8.36245656e-01 6.29587889e-01 7.19181836e-01 1.97141230e-01
5.83598197e-01 5.54151647e-03 2.69124627e-01 -3.96844357e-01
-8.71795833e-01 -1.06006870e-02 7.68024981e-01 1.13502109e+00
-1.00559883e-01 -6.58462346e-02 6.54663086e-01 6.79622740e-02
-3.84645253e-01 -4.46688890e-01 5.18229961e-01 6.32004857e-01
-5.69001539e-03 -1.02383542e+00 -6.99243903e-01 2.67020136e-01
1.10831819e-01 -2.13943318e-01 -2.75016218e-01 5.77606797e-01
3.03033561e-01 7.07623303e-01 1.35977477e-01 -2.91863680e-01
3.42525661e-01 -2.10196860e-02 9.85806704e-01 -7.94404209e-01
-4.26655114e-01 4.50426996e-01 -1.75019130e-01 -6.39422178e-01
-4.27076012e-01 -9.63125706e-01 -7.12626040e-01 -4.38708335e-01
1.12089537e-01 -2.45845512e-01 7.47493923e-01 6.30924404e-01
6.80880725e-01 4.49736059e-01 6.72832429e-01 -1.54579484e+00
3.20937708e-02 -7.55831599e-01 -6.89865708e-01 5.21540821e-01
3.68624657e-01 -6.23208880e-01 -3.27185541e-01 4.86531019e-01]
|
[9.268279075622559, -3.128288745880127]
|
65690f63-72d0-4001-818e-a4ac450dae1d
|
where-does-the-stimulus-go-deep-generative
|
2101.0923
| null |
https://arxiv.org/abs/2101.09230v1
|
https://arxiv.org/pdf/2101.09230v1.pdf
|
Where does the Stimulus go? Deep Generative Model for Commercial Banking Deposits
|
This paper examines deposits of individuals ("retail") and large companies ("wholesale") in the U.S. banking industry, and how these deposit types are impacted by macroeconomic factors, such as quantitative easing (QE). Actual data for deposits by holder are unavailable. We use a dataset on banks' financial information and probabilistic generative model to predict industry retail-wholesale deposit split from 2000 to 2020. Our model assumes account balances arise from separate retail and wholesale lognormal distributions and fit parameters of distributions by minimizing error between actual bank metrics and simulated metrics using the model's generative process. We use time-series regression to forward predict retail-wholesale deposits as function of loans, retail loans, and reserve balances at Fed banks. We find increase in reserves (representing QE) increases wholesale but not retail deposits, and increase in loans increase both wholesale and retail deposits evenly. The result shows that QE following the 2008 financial crisis benefited large companies more than average individuals, a relevant finding for economic decision making. In addition, this work benefits bank management strategy by providing forecasting capability for retail-wholesale deposits.
|
['Ni Zhan']
|
2021-01-22
| null | null | null | null |
['time-series-regression']
|
['time-series']
|
[-1.03222191e+00 1.32543445e-01 -1.17367446e-01 -3.60781431e-01
-6.01416230e-01 -7.05286086e-01 7.14930296e-01 -1.78011626e-01
-2.37689331e-01 1.03526342e+00 9.21168387e-01 -8.09494436e-01
1.46304801e-01 -1.52683377e+00 -3.18460196e-01 -6.60434067e-01
3.85246813e-01 7.91592121e-01 -3.58419299e-01 -2.71157864e-02
5.48208117e-01 6.11491084e-01 -7.35161304e-01 5.21043912e-02
5.38353384e-01 7.70298183e-01 1.10898793e-01 1.46205008e-01
-2.35420719e-01 8.38535130e-01 -4.00578588e-01 -6.52110279e-01
5.88214219e-01 -2.15349481e-01 -1.78815305e-01 -2.66371280e-01
-5.40359378e-01 -5.35883486e-01 -1.12027779e-01 8.68234813e-01
4.54269230e-01 -2.93986082e-01 1.13407803e+00 -6.32395804e-01
-1.70643628e+00 1.56915414e+00 -6.99913383e-01 7.05276728e-01
-1.31048739e-01 2.17708409e-01 1.05314636e+00 -1.05844843e+00
2.03284733e-02 1.35505152e+00 7.50712872e-01 -1.04273550e-01
-1.08008218e+00 -7.83086061e-01 -9.99128371e-02 -3.89784247e-01
-1.03957057e+00 -2.84307897e-01 2.51720965e-01 -8.45675170e-01
1.42892170e+00 -2.56308150e-02 9.65846419e-01 3.42361867e-01
6.37794852e-01 -1.50651271e-02 1.08053136e+00 -6.01635695e-01
2.53051579e-01 5.60629725e-01 1.50895089e-01 -4.04139608e-01
9.48094308e-01 4.99936976e-02 -3.31842929e-01 -8.41110349e-02
1.11500180e+00 3.08413208e-01 1.68943077e-01 7.12052464e-01
-1.04325378e+00 1.02857888e+00 1.24702265e-03 1.60727531e-01
-7.96158552e-01 4.00234580e-01 -1.13839202e-01 4.06617492e-01
4.01801288e-01 -1.46142468e-01 -3.82537931e-01 -1.84626937e-01
-9.29982126e-01 2.21965611e-01 1.10343707e+00 8.01552236e-01
3.25074613e-01 3.20443839e-01 1.29705593e-01 6.05943620e-01
8.18240464e-01 1.34037864e+00 7.33622789e-01 -5.46637416e-01
7.87496984e-01 2.95455396e-01 5.79740644e-01 -6.52235210e-01
-1.45163894e-01 -1.22926876e-01 -7.62681246e-01 3.02056998e-01
7.86982238e-01 -2.45389417e-01 -3.23478639e-01 9.95463669e-01
-4.24536645e-01 -7.17369199e-01 1.34495139e-01 5.94611406e-01
-1.57624125e-01 7.72096157e-01 2.06497461e-01 -4.18596476e-01
1.20150769e+00 -4.69864666e-01 -6.07433438e-01 1.24468505e-01
-2.25706264e-01 -6.24612808e-01 9.77730811e-01 4.16526854e-01
-1.57110429e+00 -1.40308738e-01 -3.72174621e-01 4.69362497e-01
-4.41958010e-02 1.27537698e-01 3.51744056e-01 9.47059393e-01
-1.12404609e+00 7.39274204e-01 -1.18524349e+00 2.31306832e-02
1.82150841e-01 1.35941163e-01 4.50107843e-01 8.12181830e-01
-9.61052477e-01 1.02246952e+00 -2.18122154e-01 -4.98005748e-03
-4.95773137e-01 -5.48084080e-01 -3.98343682e-01 3.74222457e-01
-6.84609294e-01 -5.65388858e-01 1.29819798e+00 -4.97985303e-01
-1.07924318e+00 3.25789958e-01 2.63846844e-01 -8.35908711e-01
4.70441580e-01 3.52910198e-02 -3.80849451e-01 -1.57554209e-01
3.15388113e-01 -3.44532400e-01 -3.18160169e-02 -7.75774837e-01
-7.29632914e-01 -6.51644349e-01 -6.92801893e-01 -4.70705554e-02
-4.51264977e-01 3.47703665e-01 8.24082971e-01 -9.74932730e-01
2.91250587e-01 -4.61458206e-01 -1.91737503e-01 -9.41658199e-01
-2.81393349e-01 -1.64114565e-01 -2.77536452e-01 -1.32575333e+00
1.46547961e+00 -1.46405709e+00 -8.08780432e-01 4.98250633e-01
-3.44617069e-01 -7.04545140e-01 8.49602401e-01 8.69210362e-01
-1.98805153e-01 6.53360963e-01 2.80156787e-02 -2.10768983e-01
6.87728047e-01 -4.30104323e-02 -7.75185406e-01 5.02475202e-01
8.67726803e-02 8.38630140e-01 -3.61007720e-01 -4.62516174e-02
-7.65760168e-02 -1.38700977e-01 4.09866124e-02 -3.04811537e-01
3.82649183e-01 -1.61002383e-01 -2.30033487e-01 1.12917531e+00
8.65595400e-01 -2.22244948e-01 -1.17035918e-02 2.51297504e-01
-6.48420334e-01 6.15132630e-01 -9.65433359e-01 3.68195713e-01
-3.92230690e-01 9.93269980e-02 -3.52075070e-01 -6.47694468e-01
1.29706180e+00 1.40664846e-01 1.15699358e-01 -7.30979800e-01
-5.78724407e-02 5.03261209e-01 -1.26734883e-01 -1.81933433e-01
6.45823240e-01 -1.07079506e+00 -6.82008564e-02 1.22185266e+00
-5.67616165e-01 9.82846543e-02 1.53008789e-01 2.67990679e-02
9.37281668e-01 -2.32455730e-01 3.41705620e-01 -6.27771795e-01
-2.21435875e-01 -9.92467850e-02 3.36124808e-01 1.64283559e-01
-3.89496982e-02 3.07495952e-01 5.86064279e-01 -1.82844833e-01
-1.40370047e+00 -1.72555077e+00 -3.38133186e-01 5.32174647e-01
-3.84284854e-01 6.39443099e-01 -3.48427802e-01 9.09633934e-02
8.45480680e-01 1.15613067e+00 -4.61853623e-01 2.16612518e-01
-3.31665277e-01 -1.52240002e+00 3.03904533e-01 1.00724995e+00
3.18941981e-01 -1.21089756e+00 -5.44236958e-01 4.95297670e-01
1.09433651e-01 -1.94714263e-01 -3.73649716e-01 2.95067221e-01
-1.14599001e+00 -6.87717795e-01 -8.82234931e-01 -3.95991892e-01
4.55021769e-01 -2.64237702e-01 1.49204969e+00 -4.25705642e-01
1.47800088e-01 1.26498550e-01 -7.07157701e-02 -6.63356900e-01
-5.03350496e-01 -4.03135329e-01 1.47751465e-01 -3.53476256e-01
6.91119134e-01 -7.73304045e-01 -7.11457014e-01 2.44598955e-01
-3.48415375e-01 -7.32846618e-01 6.13363087e-01 5.62448084e-01
4.47659701e-01 4.88639086e-01 1.61207712e+00 -4.44650322e-01
9.35707510e-01 -9.27936435e-01 -4.91614372e-01 2.98079938e-01
-1.47847903e+00 -2.54355758e-01 1.76232621e-01 -1.59535632e-01
-1.54871452e+00 -4.16282892e-01 3.91272545e-01 2.40822792e-01
3.02257329e-01 7.41013885e-01 3.06899548e-02 9.60453928e-01
3.86520505e-01 1.84758857e-01 7.87069499e-02 -7.08187461e-01
-2.42402270e-01 1.02794385e+00 6.02643251e-01 -6.11774921e-01
6.18427694e-01 3.76798213e-01 -5.48772097e-01 -7.28245303e-02
-1.54435366e-01 -4.27719325e-01 -5.48824728e-01 -8.97469521e-02
7.68849015e-01 -1.46921563e+00 -6.55150831e-01 7.34522641e-01
-6.99874640e-01 -5.75059056e-01 -6.57652259e-01 7.86944985e-01
-5.24592161e-01 5.16747795e-02 -1.14707685e+00 -1.70260715e+00
-5.95133185e-01 -6.34084105e-01 4.07120973e-01 4.08022016e-01
-1.33693218e-01 -1.27074063e+00 1.94967210e-01 1.91989571e-01
4.97661591e-01 2.71663189e-01 8.16830695e-01 -7.10355282e-01
-3.74352008e-01 5.75087443e-02 -2.61395991e-01 6.12627745e-01
5.01133144e-01 2.26142392e-01 -3.52472961e-01 -6.10802993e-02
2.98507124e-01 8.47823769e-02 6.14354610e-01 7.74711609e-01
-6.29007816e-02 -6.43492401e-01 7.29932338e-02 -2.61925254e-02
1.77304208e+00 3.64753127e-01 7.47006536e-01 9.11621511e-01
-1.06255673e-01 4.73811686e-01 4.55032557e-01 1.15476429e+00
6.99169338e-01 -2.52415210e-01 2.54360378e-01 5.46505332e-01
3.03433120e-01 -3.20699066e-01 8.58764529e-01 8.30278039e-01
-5.10143697e-01 -1.04927216e-02 -1.16283643e+00 9.77099657e-01
-1.36072910e+00 -1.44967306e+00 -5.33854067e-01 1.98991787e+00
1.17529416e+00 2.95988679e-01 7.30574250e-01 -7.66655356e-02
7.97982693e-01 -6.03063464e-01 -4.30840731e-01 -4.12864298e-01
-1.34766206e-01 2.76879758e-01 1.32122254e+00 2.26071715e-01
-1.89524069e-01 3.69663537e-01 6.60735416e+00 2.67054617e-01
-6.24448955e-01 -3.21291178e-01 1.28592539e+00 -1.60342336e-01
-9.68309999e-01 4.85798940e-02 -1.38619566e+00 1.07668781e+00
1.38801455e+00 -9.36094165e-01 4.34047908e-01 8.02442729e-01
8.85890841e-01 -4.94512916e-01 -7.77348161e-01 5.25904894e-01
-4.12869841e-01 -1.22917032e+00 2.64716148e-02 8.13346088e-01
6.92398548e-01 -1.27570674e-01 2.35588729e-01 3.63787524e-02
1.24483705e+00 -1.01439595e+00 1.52344489e+00 1.32446253e+00
6.51745260e-01 -1.08292246e+00 1.04561579e+00 3.81635547e-01
-9.45001364e-01 -4.68849897e-01 -8.29312265e-01 -4.44798827e-01
7.72697568e-01 1.14349794e+00 -7.63515115e-01 1.76018253e-01
1.04824984e+00 4.94857430e-01 -3.13320696e-01 4.84953523e-01
5.76563254e-02 9.71215487e-01 -4.71869141e-01 1.15035698e-01
-9.59524065e-02 -9.49670613e-01 -3.00713480e-01 8.99430454e-01
1.04916179e+00 4.19985682e-01 -5.69061697e-01 1.24879825e+00
1.68113008e-01 -3.14794891e-02 -2.10988358e-01 -2.10004508e-01
1.09489501e+00 8.37878406e-01 -7.69823849e-01 -2.21216932e-01
-4.78664845e-01 1.52067572e-01 -2.74860501e-01 -3.27113308e-02
-6.05653703e-01 -2.11556628e-01 2.83253103e-01 7.71306932e-01
7.43432269e-02 -3.09585184e-01 -1.28040051e+00 -8.05808663e-01
-1.40960827e-01 -4.03382897e-01 1.20332912e-01 -7.89993405e-01
-1.72785318e+00 7.04943687e-02 1.02725804e-01 -8.03983450e-01
-4.66766894e-01 -4.61990297e-01 -9.90948796e-01 1.59609294e+00
-1.41255045e+00 -9.50642169e-01 2.01819777e-01 3.78934205e-01
2.85666645e-01 -5.75135529e-01 3.07701409e-01 -4.73615229e-02
-2.91104943e-01 2.58445591e-01 6.12654567e-01 1.73612073e-01
4.88071859e-01 -1.90124059e+00 8.76413286e-01 6.69786096e-01
-2.74906129e-01 9.16835904e-01 3.62706095e-01 -1.22316515e+00
-5.88077784e-01 -8.37621212e-01 1.13790381e+00 -6.76101565e-01
1.29170012e+00 2.87404358e-01 -4.03679401e-01 9.39854085e-01
3.18400949e-01 -6.71351731e-01 9.11974132e-01 -3.89103562e-01
2.25639179e-01 1.11981547e-02 -1.36446333e+00 1.17620058e-01
3.32282245e-01 -1.89798713e-01 -7.13739216e-01 6.25849469e-03
3.20075393e-01 5.62618792e-01 -1.56454420e+00 -4.74813163e-01
7.81950057e-01 -1.14762425e+00 1.08697939e+00 -2.70679332e-02
5.33682764e-01 3.43723565e-01 -4.10505176e-01 -1.24427664e+00
-5.58659196e-01 -3.04403216e-01 1.92861125e-01 1.73704553e+00
6.71077073e-01 -1.01124501e+00 6.99757814e-01 1.16140985e+00
-1.98860437e-01 -5.97061276e-01 -8.85517120e-01 -8.88782620e-01
9.91871536e-01 2.09494561e-01 1.35119152e+00 8.41367662e-01
3.10124815e-01 -1.72869325e-01 2.27256678e-02 6.70398679e-03
6.35779977e-01 5.26500121e-02 4.54952300e-01 -1.15272689e+00
-3.39248985e-01 -6.27861381e-01 2.28099376e-01 -6.19081974e-01
-2.09876344e-01 -7.24437594e-01 -3.19283545e-01 -1.79379582e+00
5.29816985e-01 -8.40234697e-01 -4.64357994e-02 3.24461848e-01
2.56937683e-01 -2.15007842e-01 1.95312962e-01 7.36879230e-01
7.56516516e-01 3.62527251e-01 9.04192448e-01 1.70396790e-01
-1.38108164e-01 3.51797611e-01 -9.68605399e-01 7.64266431e-01
1.14348602e+00 -1.93666503e-01 -1.66642368e-01 -1.77062258e-01
6.69834018e-01 4.08414900e-01 4.28214669e-01 -5.16175866e-01
-4.28879336e-02 -6.66495502e-01 6.68589890e-01 -1.10658002e+00
-4.38690394e-01 -5.21582782e-01 7.91713834e-01 4.61817503e-01
1.19584035e-02 7.43193507e-01 -3.26517612e-01 6.18820310e-01
5.08312508e-02 -6.15416110e-01 6.40975893e-01 -4.71236438e-01
4.52852368e-01 -2.17368782e-01 -7.00834394e-01 -4.03394610e-01
9.48871613e-01 -5.02815425e-01 -4.41165030e-01 -3.14509720e-01
-1.12796330e+00 1.06866971e-01 7.03730643e-01 -3.99549931e-01
2.08059669e-01 -1.43208444e+00 -9.20798242e-01 -1.30147308e-01
-7.34774113e-01 1.29359327e-02 -3.06038469e-01 7.32702911e-01
-8.28804016e-01 3.51370782e-01 -8.48121122e-02 2.20634043e-01
-2.18913928e-01 1.57875016e-01 3.97013575e-01 -7.09115863e-01
-4.18451577e-01 8.42637420e-01 -1.64693221e-01 1.22907981e-01
-3.37167203e-01 -7.43373871e-01 3.34413000e-03 6.42313778e-01
4.53309953e-01 1.02184188e+00 -1.08308783e-02 -4.30411309e-01
-8.16444829e-02 1.19088270e-01 1.67213574e-01 -6.02426171e-01
1.81029379e+00 -7.84947991e-01 -3.47789824e-01 9.32771623e-01
3.36033881e-01 2.95805633e-01 -1.33375525e+00 3.00045878e-01
3.82665366e-01 -3.47614020e-01 -4.20794249e-01 -9.91198063e-01
-1.17705905e+00 6.30901992e-01 9.42260474e-02 6.60518587e-01
7.89900959e-01 7.88253993e-02 9.08124983e-01 -4.50136736e-02
5.03244162e-01 -1.27052653e+00 4.12229821e-03 -5.22912219e-02
1.00769877e+00 -7.68091559e-01 1.32926404e-01 2.40057588e-01
-5.93476176e-01 1.11551499e+00 8.23679641e-02 -4.17566836e-01
1.26775646e+00 6.38071775e-01 -2.06973732e-01 1.46014858e-02
-6.54726386e-01 4.49253321e-01 -6.39059246e-01 5.39648712e-01
4.46358711e-01 4.40669626e-01 -4.93062437e-01 1.71762168e+00
-8.40130508e-01 -1.97802052e-01 1.16311252e+00 6.36265159e-01
-8.63566637e-01 -9.96775270e-01 -7.55865991e-01 9.01333928e-01
-9.19588566e-01 -5.43630242e-01 -3.91998030e-02 8.39357436e-01
-2.23952860e-01 7.95881271e-01 8.20427477e-01 1.64450064e-01
2.61555344e-01 8.26444924e-02 -3.49172838e-02 -6.60263240e-01
-7.43952811e-01 3.45623970e-01 7.65593797e-02 3.97780657e-01
-1.84171081e-01 -1.66723430e+00 -1.62977803e+00 -9.41554189e-01
-6.06760025e-01 2.65973836e-01 5.30829489e-01 4.12152052e-01
-2.39128947e-01 4.73207198e-02 1.10013497e+00 -6.97814405e-01
-1.75325620e+00 -1.39615762e+00 -1.73691082e+00 -1.63402021e-01
-2.60637909e-01 -4.57167983e-01 -1.02212358e+00 8.22070420e-01]
|
[4.840504169464111, 4.108068943023682]
|
834e05d7-619a-4039-b6cb-16147d29a8db
|
information-directed-exploration-for-deep
|
1812.07544
| null |
http://arxiv.org/abs/1812.07544v2
|
http://arxiv.org/pdf/1812.07544v2.pdf
|
Information-Directed Exploration for Deep Reinforcement Learning
|
Efficient exploration remains a major challenge for reinforcement learning.
One reason is that the variability of the returns often depends on the current
state and action, and is therefore heteroscedastic. Classical exploration
strategies such as upper confidence bound algorithms and Thompson sampling fail
to appropriately account for heteroscedasticity, even in the bandit setting.
Motivated by recent findings that address this issue in bandits, we propose to
use Information-Directed Sampling (IDS) for exploration in reinforcement
learning. As our main contribution, we build on recent advances in
distributional reinforcement learning and propose a novel, tractable
approximation of IDS for deep Q-learning. The resulting exploration strategy
explicitly accounts for both parametric uncertainty and heteroscedastic
observation noise. We evaluate our method on Atari games and demonstrate a
significant improvement over alternative approaches.
|
['Andreas Krause', 'Nikolay Nikolov', 'Johannes Kirschner', 'Felix Berkenkamp']
|
2018-12-18
|
information-directed-exploration-for-deep-1
|
https://openreview.net/forum?id=Byx83s09Km
|
https://openreview.net/pdf?id=Byx83s09Km
|
iclr-2019-5
|
['distributional-reinforcement-learning']
|
['methodology']
|
[-2.77045190e-01 -6.67437613e-02 -7.16196001e-01 -1.92992255e-01
-1.21069980e+00 -5.37434161e-01 5.65249622e-01 6.25866130e-02
-5.27138412e-01 1.41726601e+00 1.89502686e-01 -5.43129861e-01
-5.54313838e-01 -9.21879888e-01 -8.91761124e-01 -6.21566117e-01
-1.35241672e-01 8.08514655e-01 -1.25304654e-01 1.34785786e-01
2.26563320e-01 2.87752569e-01 -1.27697599e+00 -2.95459867e-01
1.09085381e+00 1.13781071e+00 -5.12036905e-02 5.08742988e-01
-1.38602331e-01 9.21425700e-01 -5.88487446e-01 -2.50338942e-01
2.74865806e-01 -5.36595523e-01 -4.93419766e-01 1.50607284e-02
-1.28066659e-01 -7.40783274e-01 -2.33634263e-01 1.25414550e+00
2.99463987e-01 4.16614741e-01 1.04692769e+00 -1.27169991e+00
-4.65758890e-01 1.04255056e+00 -7.56914735e-01 1.35098949e-01
-3.22118759e-01 3.83780114e-02 1.10816956e+00 -4.22517031e-01
2.19210148e-01 1.41345572e+00 3.21155280e-01 3.33958596e-01
-1.60950398e+00 -7.31075287e-01 4.05592650e-01 5.84694669e-02
-8.58384430e-01 -1.46364853e-01 5.00511348e-01 -3.65713000e-01
5.72222531e-01 -8.16626549e-02 7.97899067e-01 1.32231951e+00
2.10112482e-01 1.28843033e+00 1.53423262e+00 -4.44877088e-01
1.08221197e+00 4.37736548e-02 -6.01714179e-02 1.73926920e-01
5.02385914e-01 7.71034896e-01 -2.98908234e-01 -3.94960880e-01
1.04533553e+00 1.67039037e-01 2.21450239e-01 -9.61824656e-01
-4.85749274e-01 1.53601801e+00 -4.25878726e-02 -4.36929643e-01
-6.12517774e-01 6.96708798e-01 3.36005181e-01 3.72599691e-01
5.34685671e-01 2.83365279e-01 -3.18954021e-01 -7.06532419e-01
-9.02731419e-01 6.51718557e-01 8.39704752e-01 7.69422770e-01
7.14063823e-01 4.12979722e-01 -5.18455446e-01 7.42235780e-01
3.61639440e-01 7.86975503e-01 3.25317115e-01 -1.27009344e+00
5.60767293e-01 -2.01190654e-02 6.51996017e-01 -2.28681594e-01
-1.64033175e-01 -6.38228118e-01 -4.22327518e-01 5.89508832e-01
5.76878309e-01 -5.61152697e-01 -8.93953145e-01 1.90408266e+00
3.08827102e-01 -1.62822068e-01 2.45996974e-02 6.26129627e-01
-9.53499302e-02 3.36781830e-01 1.22550093e-01 -3.28873366e-01
8.88362229e-01 -6.10231161e-01 -7.77636588e-01 -1.86944261e-01
3.87639433e-01 -1.50227442e-01 9.58980978e-01 6.34112418e-01
-1.24506104e+00 -8.04172158e-02 -7.91286886e-01 4.95981008e-01
1.14220843e-01 -2.85746813e-01 7.61451840e-01 9.41464186e-01
-7.01802433e-01 7.69156992e-01 -1.07386827e+00 1.49001077e-01
7.27667153e-01 1.46188438e-01 3.40610862e-01 9.82500985e-02
-1.06634438e+00 6.53240740e-01 5.06909370e-01 -2.98511833e-01
-1.32233381e+00 -6.12813175e-01 -5.99984407e-01 2.85417408e-01
9.68543768e-01 -3.07254285e-01 1.90148747e+00 -8.39230299e-01
-2.02547026e+00 -6.18308745e-02 1.81124210e-01 -1.07239974e+00
8.89129519e-01 -2.46190891e-01 8.97539258e-02 1.76935252e-02
-5.45259081e-02 2.09898666e-01 9.03259873e-01 -8.61737311e-01
-7.41435468e-01 -5.46505213e-01 9.78269381e-04 2.65216857e-01
-1.62758976e-01 -3.33075613e-01 1.15987584e-01 -6.74763501e-01
-2.91655183e-01 -9.82380092e-01 -4.60145563e-01 -4.29995388e-01
-2.60927975e-01 -1.36681706e-01 6.26804978e-02 -2.06468090e-01
1.13379979e+00 -1.75268400e+00 -1.72163114e-01 5.53448975e-01
-2.70955265e-01 -1.48186520e-01 3.05853188e-02 6.45171225e-01
2.03637734e-01 -9.30115059e-02 -2.39156350e-01 -2.01373920e-01
5.74217021e-01 4.83814687e-01 -6.63497746e-01 4.38815951e-01
-6.72458708e-02 9.89745319e-01 -8.31118822e-01 -1.17732882e-01
6.46007359e-02 -9.11220163e-02 -6.23025835e-01 7.92841464e-02
-6.14007473e-01 2.61964411e-01 -6.95510447e-01 3.87307078e-01
6.46948397e-01 -2.86761284e-01 2.96929657e-01 6.83122575e-01
-6.52964115e-02 3.49951684e-01 -1.37198007e+00 1.35129988e+00
-3.61145109e-01 1.08133465e-01 -3.77292037e-02 -1.35595787e+00
7.44062603e-01 2.03477159e-01 4.71655637e-01 -8.06543648e-01
2.48663537e-02 3.42954099e-01 -8.15325230e-02 -1.05841033e-01
4.34728742e-01 -3.74927968e-01 -6.35851845e-02 7.20895052e-01
2.35220809e-02 -1.41147763e-01 1.67070135e-01 2.11389717e-02
8.91244113e-01 5.16135514e-01 4.00244266e-01 -4.81505871e-01
-2.71111608e-01 -8.22406486e-02 6.01084650e-01 1.40538323e+00
-1.24213561e-01 1.00702338e-01 1.15371788e+00 -3.98058034e-02
-9.76327896e-01 -1.36184514e+00 -2.86185533e-01 1.16961658e+00
-4.08542901e-01 2.34813876e-02 -5.96491933e-01 -8.10223639e-01
6.00410461e-01 9.32763040e-01 -8.97410572e-01 -7.68439695e-02
3.27616446e-02 -8.46252561e-01 1.50556192e-01 7.88488686e-01
1.64236695e-01 -9.29150999e-01 -7.18912125e-01 5.46674132e-01
3.67208600e-01 -3.75385642e-01 -1.55774370e-01 5.67698002e-01
-1.09167385e+00 -8.51649582e-01 -9.14809883e-01 1.66818872e-01
-6.19526440e-03 -2.43587464e-01 1.09771812e+00 -7.64528394e-01
7.09230080e-02 6.47472441e-01 -1.05709441e-01 -7.99980640e-01
-2.94338912e-01 -3.75840925e-02 -6.50772378e-02 -1.62444577e-01
3.17248702e-01 -3.96455497e-01 -7.33020186e-01 1.27419010e-02
-8.50114346e-01 -5.60586154e-01 4.63616729e-01 1.17324352e+00
5.88639557e-01 -7.45561793e-02 9.61383700e-01 -1.08808923e+00
8.69853497e-01 -6.61036193e-01 -1.33367205e+00 1.55508712e-01
-9.52579379e-01 5.09149790e-01 2.20601350e-01 -5.00759184e-01
-1.20162404e+00 -3.13474536e-01 2.20827192e-01 -4.38499302e-01
5.23230880e-02 7.85677850e-01 1.73771411e-01 2.61234492e-01
3.94060999e-01 8.14333856e-02 2.32762530e-01 -4.58421260e-01
2.40600541e-01 4.53555405e-01 6.62765056e-02 -1.01736999e+00
3.53557974e-01 4.39880669e-01 1.95648968e-01 -4.59751457e-01
-8.04577589e-01 -6.05298243e-02 6.83389157e-02 1.48946896e-01
4.80947316e-01 -1.06164980e+00 -1.03719604e+00 1.33405283e-01
-4.82985675e-01 -8.43450487e-01 -8.07151973e-01 9.16920662e-01
-1.34096551e+00 1.18227139e-01 -4.85668510e-01 -1.70620012e+00
1.83119290e-02 -1.12590683e+00 6.95039988e-01 2.53190309e-01
5.04899956e-02 -1.06353307e+00 5.09486377e-01 -6.37935475e-02
3.71757179e-01 -8.13151523e-02 8.21648955e-01 -8.14742208e-01
-6.19987428e-01 9.06833634e-02 -3.86981154e-03 2.03969866e-01
-9.42557976e-02 -2.13296726e-01 -6.38152301e-01 -4.16833192e-01
-1.21787891e-01 -8.15374792e-01 1.09263217e+00 1.02040231e+00
1.23040795e+00 -2.57639050e-01 1.45140272e-02 3.74912530e-01
1.22373879e+00 4.05202985e-01 3.58826667e-01 5.55085838e-01
-9.19407606e-02 4.30174232e-01 9.38442171e-01 1.23842227e+00
1.96764380e-01 4.28185523e-01 5.58902860e-01 4.19812620e-01
7.43719518e-01 -7.45440602e-01 2.55063236e-01 -5.31015135e-02
2.40712643e-01 1.23190181e-02 -6.08932436e-01 7.11782098e-01
-2.12813854e+00 -1.11886311e+00 3.54864478e-01 2.65826988e+00
1.11371982e+00 1.72176003e-01 7.14217722e-01 -2.63179302e-01
3.30002964e-01 -1.26221433e-01 -1.19239390e+00 -4.81350154e-01
2.15869434e-02 5.71899414e-01 9.82705057e-01 5.87994218e-01
-9.27899420e-01 6.25482082e-01 6.95854759e+00 1.05410028e+00
-4.73575145e-01 -9.01573449e-02 7.77534842e-01 -4.21340078e-01
-5.76059043e-01 -8.24693218e-03 -8.72231066e-01 4.98013735e-01
1.02943277e+00 -1.97406143e-01 5.78983963e-01 1.02736318e+00
1.43592089e-01 -3.98168832e-01 -9.47925329e-01 6.28418744e-01
-5.90299070e-01 -1.05701828e+00 -3.41992944e-01 4.76994812e-01
1.14053321e+00 8.40015933e-02 5.14664412e-01 7.15274096e-01
1.34316301e+00 -1.10329032e+00 7.32932150e-01 3.85894120e-01
5.20525634e-01 -1.26488304e+00 4.10566777e-01 4.51146990e-01
-5.22885442e-01 -5.68866789e-01 -5.72693348e-01 -1.63911521e-01
-1.85077503e-01 4.82987821e-01 -5.73221505e-01 5.16083464e-02
4.16444123e-01 3.67770940e-01 -1.37328450e-02 1.10032284e+00
-2.81348586e-01 9.96959507e-01 -3.93017560e-01 -2.91996330e-01
7.03550458e-01 -5.93483031e-01 3.10389519e-01 6.56385779e-01
5.29240489e-01 -2.53929049e-01 2.17557162e-01 1.16757596e+00
1.11269884e-01 -3.23915109e-03 -5.05464494e-01 -3.38398248e-01
5.61428785e-01 7.03146100e-01 -4.60327744e-01 -4.07956213e-01
-3.53488415e-01 3.47959340e-01 4.98048127e-01 7.24316657e-01
-6.51944101e-01 -1.68575287e-01 8.07219386e-01 -2.26666495e-01
9.53069031e-01 -1.09470807e-01 -1.58187419e-01 -1.09320199e+00
-8.70663598e-02 -9.00802016e-01 5.80523372e-01 -8.37182626e-02
-1.30152571e+00 -1.21993229e-01 2.82486409e-01 -7.62634397e-01
-9.39554572e-01 -5.51885009e-01 -2.84468293e-01 8.73340011e-01
-1.70848656e+00 -5.55714726e-01 6.42888069e-01 5.21284640e-01
4.01113123e-01 -1.99284449e-01 5.36727965e-01 -4.33917046e-01
-3.98330629e-01 4.14624691e-01 1.16769195e+00 -2.40608945e-01
4.73514527e-01 -1.85121810e+00 2.34646112e-01 3.35242003e-01
3.48828852e-01 4.08877730e-01 8.13059688e-01 -7.47747838e-01
-1.42924786e+00 -7.91565061e-01 -9.57527831e-02 -1.15264989e-01
1.08685946e+00 -2.49457344e-01 -7.49022126e-01 7.56818235e-01
4.33260724e-02 -8.63752887e-02 6.18901610e-01 4.30752516e-01
-3.81238937e-01 -2.77475677e-02 -8.98091137e-01 5.21234393e-01
5.97394407e-01 -2.89218217e-01 -3.88797313e-01 4.74637263e-02
4.26401526e-01 -2.48472765e-01 -7.30579138e-01 1.06263518e-01
5.25909364e-01 -1.11174381e+00 7.13476956e-01 -1.04575825e+00
3.00444752e-01 3.73146385e-01 -2.17859015e-01 -1.68312764e+00
-1.21489763e-01 -9.97657537e-01 -5.02577603e-01 8.36510539e-01
3.17661494e-01 -8.48845005e-01 1.15295911e+00 7.90284872e-01
2.66995430e-01 -5.78460634e-01 -1.12533760e+00 -1.25281525e+00
7.34732687e-01 -5.30288279e-01 6.98948681e-01 3.84750336e-01
2.40460023e-01 -1.70531407e-01 -5.91307282e-01 -3.45121682e-01
1.17636347e+00 4.85126674e-01 7.07160413e-01 -1.21224570e+00
-8.72584522e-01 -5.04843771e-01 2.24564955e-01 -1.19697118e+00
3.42127442e-01 -3.16984862e-01 1.58438638e-01 -1.03134286e+00
3.70577306e-01 -3.56956810e-01 -5.59170544e-01 1.37665942e-01
-8.65921304e-02 -8.20784867e-02 4.78289612e-02 -3.16564590e-01
-6.62551522e-01 1.11788607e+00 1.03200424e+00 1.07293509e-01
-2.51411885e-01 6.62983477e-01 -7.59962380e-01 5.07562995e-01
8.61625314e-01 -4.56382841e-01 -6.83462203e-01 -1.05188517e-02
3.49430740e-01 5.75806916e-01 2.58538216e-01 -5.67365766e-01
-1.35218933e-01 -6.61054790e-01 3.27110022e-01 -6.54878199e-01
3.26924086e-01 -6.90528214e-01 -9.29745734e-02 5.57069719e-01
-7.10164368e-01 -8.08608159e-02 1.56418577e-01 1.19746912e+00
-7.92862624e-02 -4.51647133e-01 6.62674844e-01 -2.05441341e-01
-9.22954679e-02 4.10146922e-01 -6.80761039e-01 4.85019892e-01
7.52466559e-01 1.21359102e-01 -5.99368922e-02 -8.92000020e-01
-6.35691285e-01 3.78320694e-01 2.19285473e-01 -1.05064198e-01
4.27731514e-01 -1.33364773e+00 -6.00901961e-01 1.06846780e-01
6.97829900e-03 -3.09623629e-01 1.64875656e-01 6.95736229e-01
2.02816233e-01 5.23944855e-01 -4.76623662e-02 -2.24308744e-01
-5.71786404e-01 5.68590760e-01 2.64201403e-01 -5.61571360e-01
-2.97937334e-01 4.86304641e-01 2.50497282e-01 -2.45883524e-01
4.25994664e-01 -4.74932879e-01 7.67321736e-02 1.40731633e-01
5.06936848e-01 5.97102225e-01 -2.33437002e-01 3.25428039e-01
2.07571432e-01 -1.88190863e-02 -1.88843161e-01 -7.35754192e-01
1.30326426e+00 -1.68744639e-01 4.35114145e-01 8.97616923e-01
6.85422540e-01 -2.86939412e-01 -1.99110556e+00 -5.61915874e-01
2.07865715e-01 -6.34215653e-01 1.11203335e-01 -7.56768942e-01
-7.38975942e-01 9.67748940e-01 3.94896477e-01 4.39982027e-01
5.73193669e-01 -2.67803788e-01 2.46771947e-01 3.94820660e-01
4.95054781e-01 -1.48151636e+00 3.63944024e-02 5.09308815e-01
5.02743781e-01 -1.27379012e+00 2.48582959e-02 5.52970707e-01
-9.46335435e-01 8.75037313e-01 3.24355870e-01 -2.68247753e-01
7.21692443e-01 3.05109411e-01 -2.85679340e-01 1.46705672e-01
-1.05046737e+00 -3.20737064e-01 -3.64449508e-02 6.54098392e-01
1.62389621e-01 2.18524367e-01 -1.63866878e-01 5.80537319e-01
-9.61744692e-03 9.64839906e-02 5.31530499e-01 8.60108852e-01
-6.32395387e-01 -1.17540872e+00 -3.98845762e-01 6.43063366e-01
-7.95371294e-01 -4.84841969e-03 1.50903150e-01 7.79935122e-01
-8.69897366e-01 7.01885164e-01 2.61426300e-01 4.40720201e-01
3.62095684e-02 -2.29303781e-02 6.37912273e-01 -3.07671964e-01
-1.34088337e-01 5.04795432e-01 -1.78971559e-01 -5.21424532e-01
-1.19801536e-02 -1.06146479e+00 -7.90330946e-01 -1.79003060e-01
-1.55035824e-01 5.92151880e-01 6.35538816e-01 8.64574194e-01
1.23186223e-01 2.43971810e-01 6.29216194e-01 -4.03476477e-01
-2.04434657e+00 -9.63734627e-01 -1.32457852e+00 8.08370411e-02
5.29817283e-01 -1.07141793e+00 -3.45255822e-01 -7.13305354e-01]
|
[4.211427688598633, 2.6523420810699463]
|
0371bc1f-4bdb-4712-a3a3-8e07559f2b9d
|
world-consistent-video-to-video-synthesis
|
2007.08509
| null |
https://arxiv.org/abs/2007.08509v1
|
https://arxiv.org/pdf/2007.08509v1.pdf
|
World-Consistent Video-to-Video Synthesis
|
Video-to-video synthesis (vid2vid) aims for converting high-level semantic inputs to photorealistic videos. While existing vid2vid methods can achieve short-term temporal consistency, they fail to ensure the long-term one. This is because they lack knowledge of the 3D world being rendered and generate each frame only based on the past few frames. To address the limitation, we introduce a novel vid2vid framework that efficiently and effectively utilizes all past generated frames during rendering. This is achieved by condensing the 3D world rendered so far into a physically-grounded estimate of the current frame, which we call the guidance image. We further propose a novel neural network architecture to take advantage of the information stored in the guidance images. Extensive experimental results on several challenging datasets verify the effectiveness of our approach in achieving world consistency - the output video is consistent within the entire rendered 3D world. https://nvlabs.github.io/wc-vid2vid/
|
['Ming-Yu Liu', 'Ting-Chun Wang', 'Karan Sapra', 'Arun Mallya']
|
2020-07-16
| null |
https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/587_ECCV_2020_paper.php
|
https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123530358.pdf
|
eccv-2020-8
|
['video-to-video-synthesis']
|
['computer-vision']
|
[ 7.69370794e-02 -5.75762205e-02 1.29657954e-01 -1.72476187e-01
-3.34306687e-01 -6.23602748e-01 6.56432986e-01 -4.52742338e-01
-5.06986640e-02 5.37731767e-01 2.78331906e-01 -1.03647865e-01
2.74234802e-01 -9.18451011e-01 -9.89930749e-01 -3.56055111e-01
2.32476011e-01 5.84689528e-02 5.00822306e-01 -1.04034692e-01
1.20731711e-01 5.03795981e-01 -1.70031381e+00 2.81553656e-01
5.38389981e-01 1.09591627e+00 3.39590907e-01 9.38233554e-01
-1.72586247e-01 9.63737547e-01 -5.17032385e-01 -1.23882569e-01
5.57905912e-01 -6.98574185e-01 -6.54629767e-01 2.47811213e-01
6.25781953e-01 -9.55895841e-01 -7.63395607e-01 9.09999967e-01
2.94349372e-01 3.91340494e-01 1.76126644e-01 -1.43131161e+00
-6.30936861e-01 -9.35729370e-02 -3.51862431e-01 1.66083038e-01
8.11759174e-01 3.14134300e-01 5.17284632e-01 -8.16650391e-01
1.24018526e+00 1.39869177e+00 3.63130361e-01 6.98158801e-01
-8.03404748e-01 -5.78073919e-01 4.41944659e-01 1.17009841e-01
-1.32716382e+00 -5.17848313e-01 1.02353883e+00 -3.01899344e-01
8.52218032e-01 2.94135779e-01 1.04286921e+00 9.71758366e-01
1.16193347e-01 6.37032747e-01 9.15538490e-01 -3.16323400e-01
2.75265872e-01 -3.33191186e-01 -3.37785393e-01 7.94796348e-01
-1.08285286e-01 3.04711074e-01 -8.30661476e-01 1.39357120e-01
1.30831552e+00 -9.12257954e-02 -4.91742015e-01 -4.45609778e-01
-1.40107584e+00 2.77625561e-01 3.44012797e-01 -3.06297420e-03
-3.84657353e-01 4.89829183e-01 1.57894149e-01 2.35171407e-01
6.21424854e-01 2.32507922e-02 -1.22432210e-01 -2.00256020e-01
-9.27989125e-01 2.26325512e-01 5.17922521e-01 1.09260905e+00
6.50413573e-01 3.81678045e-01 6.40784428e-02 3.10817510e-01
1.28864452e-01 4.83785361e-01 2.50071883e-01 -1.60535908e+00
4.84562159e-01 3.18359971e-01 3.99063110e-01 -1.05252445e+00
-1.74403843e-02 1.10411547e-01 -6.25503480e-01 5.87514043e-01
2.41320252e-01 1.24486061e-02 -9.71836448e-01 1.67186522e+00
7.25849271e-01 6.65719569e-01 1.42650642e-02 9.90031362e-01
9.38930154e-01 1.09954047e+00 -1.57588676e-01 -1.18575260e-01
7.25864649e-01 -1.05013049e+00 -7.26914942e-01 -1.23065650e-01
1.16408713e-01 -5.55399060e-01 8.35169554e-01 1.98534206e-01
-1.40793157e+00 -7.56846905e-01 -1.16753852e+00 -3.52490067e-01
-8.41549337e-02 -2.74913996e-01 5.08905411e-01 2.91814655e-01
-1.34025705e+00 5.71805596e-01 -8.77915382e-01 -1.02072604e-01
1.92276955e-01 -7.51301181e-03 -5.77035487e-01 4.23009507e-02
-9.78035748e-01 6.25097275e-01 4.60973918e-01 2.65982628e-01
-9.97881591e-01 -6.55392826e-01 -9.62955952e-01 -1.88087776e-01
3.06355685e-01 -1.02834129e+00 1.30229390e+00 -1.01789653e+00
-1.65613377e+00 7.24722028e-01 -3.27209115e-01 -1.64046839e-01
9.88754690e-01 -2.78363228e-01 -1.60743743e-01 4.83699918e-01
-5.63229546e-02 9.32550490e-01 8.43374372e-01 -1.52633929e+00
-6.86063111e-01 -1.54549152e-01 3.53781730e-01 4.73097801e-01
-6.89992979e-02 -2.16757968e-01 -9.92689431e-01 -5.98982692e-01
3.20739418e-01 -8.66362035e-01 5.63751608e-02 5.37338793e-01
-1.60363153e-01 6.51422665e-02 9.12484646e-01 -7.76409507e-01
8.77191246e-01 -2.06236553e+00 2.74379790e-01 -1.20914720e-01
1.94981232e-01 4.16721791e-01 -1.82649419e-01 4.23023194e-01
3.29838730e-02 -1.08825848e-01 9.38496441e-02 -5.82832813e-01
-2.24296406e-01 2.29962856e-01 -6.12532079e-01 3.71205568e-01
1.58286497e-01 9.42417622e-01 -1.12755811e+00 -5.41078269e-01
7.56732821e-01 8.67061019e-01 -5.89582443e-01 4.49454933e-01
-4.86224622e-01 7.22751200e-01 -4.49916154e-01 4.88753378e-01
7.89697886e-01 -1.94786400e-01 1.42857641e-01 -3.16413373e-01
-2.08741173e-01 8.66653398e-02 -1.28072786e+00 2.15625286e+00
-3.57231975e-01 7.49292493e-01 -1.67783290e-01 -5.70388377e-01
7.75189519e-01 3.22884738e-01 5.86565256e-01 -1.07018232e+00
3.71603929e-02 1.03566453e-01 -6.62445962e-01 -4.49143410e-01
6.42719924e-01 1.23449787e-01 2.62189806e-01 3.17874223e-01
-1.57308280e-01 -4.54094231e-01 1.19037502e-01 3.68937731e-01
6.98294878e-01 9.50920165e-01 4.71423753e-02 2.48933062e-01
3.69057536e-01 -9.25574377e-02 7.71388233e-01 5.48872650e-01
-2.90239573e-01 1.06961858e+00 3.32549751e-01 -6.77288234e-01
-1.35397732e+00 -1.24394357e+00 4.46935385e-01 5.02461195e-01
7.80872703e-01 -3.70862246e-01 -7.08884120e-01 -3.46297294e-01
-3.09421569e-01 7.14503407e-01 -6.98674083e-01 -8.24823305e-02
-7.62647450e-01 -1.89505331e-02 2.27104262e-01 5.39499283e-01
6.18288875e-01 -1.01380527e+00 -1.03382170e+00 2.43799642e-01
-4.53917265e-01 -1.41648901e+00 -5.03318429e-01 -4.85880315e-01
-9.00765240e-01 -9.35050428e-01 -6.78670526e-01 -6.73158586e-01
5.01489043e-01 5.19913495e-01 1.18212819e+00 2.96518326e-01
2.52067596e-02 4.28233564e-01 -2.77314663e-01 -5.93989082e-02
-5.67514062e-01 -6.08070433e-01 -3.40581983e-02 1.72165390e-02
-2.65908748e-01 -7.40733862e-01 -7.97514379e-01 2.00147837e-01
-1.01226211e+00 7.60220885e-01 -7.85175059e-03 4.91394997e-01
7.60159671e-01 -1.69697702e-02 1.48625225e-01 -4.70468938e-01
1.24134049e-01 -1.07773848e-01 -6.94302619e-01 2.32352436e-01
-1.29216641e-01 -9.98806134e-02 5.36310434e-01 -4.82098669e-01
-1.30975974e+00 1.44894645e-01 -1.24069959e-01 -9.33265567e-01
1.07396558e-01 5.83580974e-03 -1.59298286e-01 1.17271684e-01
1.65841252e-01 3.08986306e-01 -1.26923695e-01 -2.44705215e-01
4.59458441e-01 2.82566667e-01 1.00857532e+00 -4.93646443e-01
8.57261896e-01 9.09014404e-01 -4.26416891e-03 -5.29290080e-01
-7.31501639e-01 -8.81572217e-02 -7.10072517e-01 -6.55007899e-01
9.47432637e-01 -1.03983378e+00 -7.37670839e-01 7.37850308e-01
-1.35006332e+00 -7.24972665e-01 -3.77260894e-01 3.52730751e-01
-8.36172640e-01 4.10484195e-01 -5.06418467e-01 -8.27932298e-01
-2.79591054e-01 -1.09219015e+00 1.29206741e+00 2.56193221e-01
2.53665552e-04 -8.95753801e-01 1.40684051e-02 1.55793115e-01
7.14596659e-02 6.82857513e-01 4.74290699e-01 4.14942205e-01
-8.57743323e-01 -1.58075821e-02 -3.16767752e-01 1.86447620e-01
6.03479035e-02 3.37168753e-01 -9.29010093e-01 -5.54541387e-02
6.36684569e-03 -1.49155647e-01 6.25576675e-01 3.52723211e-01
1.11948359e+00 -6.73775077e-02 -1.08725794e-01 9.38614905e-01
1.58874536e+00 4.14883703e-01 8.42459261e-01 2.71729708e-01
9.47812855e-01 2.99519718e-01 7.05521584e-01 4.04547095e-01
5.38681269e-01 9.05657411e-01 7.41038203e-01 -8.95074680e-02
-7.77758420e-01 -7.13589966e-01 3.98837864e-01 8.18914711e-01
-1.94427177e-01 -4.29663479e-01 -5.86870134e-01 4.80093509e-01
-1.93979335e+00 -1.05264711e+00 -1.83215160e-02 2.14338422e+00
5.78205645e-01 4.50608730e-02 -1.15470275e-01 1.61663499e-02
6.63524389e-01 4.40185189e-01 -7.00688779e-01 -2.14695066e-01
-1.97584361e-01 -6.61187917e-02 1.62914112e-01 7.71417856e-01
-7.69619644e-01 9.72446322e-01 5.76117039e+00 5.62504947e-01
-1.21442568e+00 6.21671267e-02 5.47776222e-01 -2.40261719e-01
-5.40013015e-01 1.49216339e-01 -3.62025857e-01 4.76630539e-01
5.99232078e-01 -1.95000023e-01 4.33511883e-01 6.06037676e-01
4.81229186e-01 -2.49030605e-01 -1.05713069e+00 1.14363325e+00
6.44815415e-02 -1.75779927e+00 1.96428671e-01 -1.39050990e-01
9.87599671e-01 -2.16056019e-01 1.56825241e-02 -2.21074417e-01
2.11887524e-01 -6.23079896e-01 1.48851907e+00 7.57320523e-01
1.17854357e+00 -5.66373110e-01 1.51438132e-01 1.73786938e-01
-1.33815968e+00 1.15707800e-01 -2.25728273e-01 -1.31025508e-01
6.08463109e-01 3.59889120e-01 -5.11859894e-01 6.97155356e-01
7.69084811e-01 8.99560690e-01 -4.49562192e-01 7.54627347e-01
-2.77386546e-01 9.82363820e-02 -2.90808350e-01 3.40135366e-01
2.33483836e-01 -3.11790913e-01 5.52515388e-01 7.10353076e-01
5.34432173e-01 3.34590524e-01 1.74710378e-02 9.38533008e-01
-4.57119681e-02 -3.39735746e-01 -7.34510541e-01 1.74383506e-01
4.97458428e-01 8.19558859e-01 -7.74078965e-01 -5.25574327e-01
-2.47861773e-01 1.40240645e+00 2.50682145e-01 5.31081915e-01
-1.07522643e+00 -1.08302981e-01 7.08111286e-01 -4.31209952e-02
2.84491867e-01 -6.19544804e-01 -1.71770491e-02 -1.42610991e+00
3.70200485e-01 -4.86395895e-01 1.81327999e-01 -1.50118327e+00
-6.56120420e-01 7.28240967e-01 4.81278785e-02 -1.48321617e+00
-4.72546428e-01 -2.89356321e-01 -4.01359141e-01 6.28444493e-01
-1.50661421e+00 -1.29947853e+00 -7.86759257e-01 6.77979052e-01
7.21471071e-01 3.43986541e-01 4.47718740e-01 1.44727796e-01
-2.54721165e-01 1.54197618e-01 -3.37580182e-02 -7.71537481e-04
4.93060648e-01 -9.19892907e-01 8.55370283e-01 1.15602338e+00
1.78846270e-01 2.17725143e-01 6.69689834e-01 -7.68445611e-01
-1.63795567e+00 -1.07555592e+00 6.33808911e-01 -4.67563301e-01
2.91237324e-01 -2.43130073e-01 -7.39257932e-01 6.53268278e-01
1.13023102e-01 2.16218963e-01 -1.39497399e-01 -8.22731674e-01
-4.46853280e-01 -7.54550174e-02 -1.01826644e+00 8.52102816e-01
1.57135212e+00 -6.45981193e-01 -3.13235641e-01 1.65821284e-01
1.06922615e+00 -9.04400647e-01 -6.76598728e-01 2.64410675e-01
7.00941026e-01 -1.32930648e+00 1.24163628e+00 -4.00805831e-01
6.86299860e-01 -6.73716307e-01 -2.06350058e-01 -1.10271418e+00
3.45743373e-02 -8.05338979e-01 -4.48748887e-01 1.00368202e+00
-3.20273995e-01 -4.68670607e-01 7.02709079e-01 7.52258599e-01
-9.36603919e-02 -4.64788169e-01 -8.93702328e-01 -8.60514820e-01
-1.40277624e-01 -6.18137121e-01 7.72968829e-01 7.23941386e-01
-4.40977246e-01 -1.92977577e-01 -7.12429225e-01 2.76951998e-01
6.67452991e-01 3.76097739e-01 9.71745133e-01 -7.28845119e-01
-1.45930633e-01 -1.77541047e-01 -6.27492070e-01 -1.21329236e+00
1.83821782e-01 -4.61540550e-01 6.71028718e-02 -1.77002633e+00
-9.82376412e-02 -3.71955395e-01 9.96978208e-02 2.73060173e-01
-1.95145383e-01 6.65057659e-01 5.42701006e-01 2.00163841e-01
-5.87391078e-01 6.28077686e-01 1.58922768e+00 5.56458198e-02
-2.63819516e-01 -5.36075771e-01 -2.39192829e-01 5.88258624e-01
7.29511917e-01 -1.81196749e-01 -6.52707756e-01 -9.65788186e-01
1.86847523e-01 3.99557143e-01 7.07178533e-01 -1.14661205e+00
1.72903493e-01 -4.47810143e-01 4.68146801e-01 -7.08804488e-01
7.46464074e-01 -6.72077119e-01 8.13590288e-01 3.17005664e-01
-1.42682329e-01 2.08538815e-01 1.57574341e-01 6.66725338e-01
-1.05362363e-01 1.15946643e-01 6.15843534e-01 -5.81907146e-02
-1.17442977e+00 5.33387363e-01 -2.18605734e-02 8.82257614e-03
1.08642447e+00 -5.17886341e-01 -2.77256042e-01 -6.72044396e-01
-5.27105451e-01 1.90453529e-02 9.27764952e-01 6.62796259e-01
9.38910604e-01 -1.49223757e+00 -5.03209054e-01 2.62846857e-01
-3.29451002e-02 1.91415951e-01 5.05314946e-01 3.25146526e-01
-1.05647254e+00 1.78370118e-01 -3.05405378e-01 -6.74218953e-01
-1.15797436e+00 7.14639425e-01 3.98873657e-01 1.77748173e-01
-1.18644989e+00 6.98718667e-01 3.88074130e-01 -6.42458573e-02
1.23943090e-01 -3.45597774e-01 2.73271412e-01 -3.33404362e-01
7.17863381e-01 2.35597938e-01 -1.14915565e-01 -8.95495594e-01
-1.83972239e-01 7.47570038e-01 3.54466349e-01 -5.16265869e-01
1.35016370e+00 -5.21671951e-01 2.65847687e-02 4.04533952e-01
1.36990750e+00 -1.14548311e-01 -1.95593095e+00 -4.97170910e-02
-5.78240097e-01 -9.57228780e-01 2.65464410e-02 -4.13040459e-01
-1.36791849e+00 7.19014645e-01 4.01131898e-01 -8.27753022e-02
1.31298220e+00 -2.94951349e-01 1.12494874e+00 4.53328155e-02
6.09305561e-01 -9.32815552e-01 2.10764796e-01 3.31807971e-01
9.36347127e-01 -8.54876757e-01 -3.77552062e-02 -5.33854663e-01
-5.25462151e-01 1.19926906e+00 6.38773143e-01 -1.32938296e-01
3.95452738e-01 9.30354372e-02 2.58794099e-01 -2.81532835e-02
-7.66071677e-01 -4.69842665e-02 4.32009734e-02 7.31494546e-01
-6.54535741e-02 -2.16393009e-01 1.12485103e-01 -1.89935699e-01
-1.57802895e-01 2.89618611e-01 8.23398769e-01 9.65499043e-01
-1.90807823e-02 -9.02724266e-01 -2.68419951e-01 -2.05079436e-01
-2.60791685e-02 1.75368801e-01 -1.62149474e-01 6.68933690e-01
5.29468879e-02 9.62658048e-01 2.28332728e-01 -2.92816460e-01
2.17587516e-01 -1.59196168e-01 7.01353788e-01 -1.77916080e-01
-1.65124163e-01 -8.51157773e-03 -5.04728369e-02 -1.14709198e+00
-6.79501593e-01 -3.80067647e-01 -1.36126590e+00 -4.93341744e-01
6.87073320e-02 -1.19960397e-01 7.25182712e-01 5.81780612e-01
4.70219940e-01 5.35140932e-01 7.07504630e-01 -1.38898373e+00
3.55040319e-02 -2.37750709e-01 -2.13182673e-01 7.24809885e-01
5.93162358e-01 -5.93882203e-01 -2.26260155e-01 5.79264104e-01]
|
[9.757499694824219, -2.2445871829986572]
|
d33a0f6b-bd66-40ec-aaa8-34c3ee2cbddd
|
ris-gan-explore-residual-and-illumination
|
1911.09178
| null |
https://arxiv.org/abs/1911.09178v2
|
https://arxiv.org/pdf/1911.09178v2.pdf
|
RIS-GAN: Explore Residual and Illumination with Generative Adversarial Networks for Shadow Removal
|
Residual images and illumination estimation have been proved very helpful in image enhancement. In this paper, we propose a general and novel framework RIS-GAN which explores residual and illumination with Generative Adversarial Networks for shadow removal. Combined with the coarse shadow-removal image, the estimated negative residual images and inverse illumination maps can be used to generate indirect shadow-removal images to refine the coarse shadow-removal result to the fine shadow-free image in a coarse-to-fine fashion. Three discriminators are designed to distinguish whether the predicted negative residual images, shadow-removal images, and the inverse illumination maps are real or fake jointly compared with the corresponding ground-truth information. To our best knowledge, we are the first one to explore residual and illumination for shadow removal. We evaluate our proposed method on two benchmark datasets, i.e., SRD and ISTD, and the extensive experiments demonstrate that our proposed method achieves the superior performance to state-of-the-arts, although we have no particular shadow-aware components designed in our generators.
|
['Chunxia Xiao', 'Ling Zhang', 'Chengjiang Long', 'Xiaolong Zhang']
|
2019-11-20
| null | null | null | null |
['shadow-removal']
|
['computer-vision']
|
[ 8.26919675e-01 1.84433758e-01 3.94022495e-01 -6.21303134e-02
-5.60919583e-01 -4.85870570e-01 4.37956899e-01 -1.04235518e+00
1.05253175e-01 1.16206205e+00 -6.66849613e-02 -3.05343926e-01
4.55401719e-01 -7.58056402e-01 -6.27348840e-01 -1.09638083e+00
4.63824987e-01 -5.31363394e-03 3.34863901e-01 -3.56308937e-01
6.46827593e-02 3.49826545e-01 -1.34313154e+00 6.23961613e-02
1.18264842e+00 9.97017145e-01 4.06313717e-01 6.64247036e-01
1.54526487e-01 6.96035564e-01 -9.31215286e-01 -2.06979200e-01
5.23681283e-01 -7.99955666e-01 1.20937116e-02 8.94217938e-02
1.93432122e-01 -6.21426821e-01 -4.39188004e-01 9.34736252e-01
6.00660264e-01 -8.47764388e-02 5.64704299e-01 -1.29919970e+00
-9.42387283e-01 -1.53179660e-01 -7.49614060e-01 -1.23018108e-01
2.36964777e-01 4.40224499e-01 2.00487241e-01 -9.37321842e-01
5.01835585e-01 1.21764541e+00 5.31802833e-01 5.01400113e-01
-9.85532522e-01 -9.84713614e-01 -1.66113805e-02 -7.37408921e-02
-1.35335886e+00 -5.07156014e-01 1.13089168e+00 3.89399864e-02
2.05514118e-01 5.80795765e-01 5.08158326e-01 1.16149604e+00
3.87222588e-01 6.86068416e-01 1.97332358e+00 -3.94120842e-01
4.95028123e-02 4.15612429e-01 -5.50831497e-01 8.60844254e-01
2.45675549e-01 6.52869403e-01 -2.96950787e-01 1.12122521e-02
8.81170034e-01 6.83687329e-02 -8.61371577e-01 8.82925000e-03
-9.37427700e-01 3.22253436e-01 6.61585629e-01 -3.42337564e-02
-3.21579158e-01 9.46928784e-02 -2.85852492e-01 5.29346168e-02
4.83847886e-01 2.27301180e-01 -2.32094139e-01 5.67385435e-01
-8.05338621e-01 -1.12988323e-01 5.65498054e-01 8.87642980e-01
9.42632735e-01 4.87870753e-01 -4.99739438e-01 5.13598204e-01
1.65581778e-01 1.13154364e+00 2.57190317e-01 -8.28919113e-01
3.37709516e-01 3.63510013e-01 4.02964801e-01 -1.07331789e+00
9.92850885e-02 -5.73800862e-01 -1.15237725e+00 5.41091800e-01
-9.04354155e-02 -2.30573133e-01 -1.32161367e+00 1.57033741e+00
3.26301873e-01 7.20035911e-01 3.52973402e-01 8.67423117e-01
8.87200713e-01 7.73785770e-01 -4.02331173e-01 -4.54708993e-01
9.19264078e-01 -1.10048163e+00 -1.00583577e+00 -5.31583488e-01
-1.33209839e-01 -9.37368393e-01 1.04364240e+00 3.99837852e-01
-9.68823552e-01 -7.65401185e-01 -1.29716861e+00 2.87709028e-01
-1.36149421e-01 4.60219324e-01 4.64004159e-01 8.84141803e-01
-9.92776752e-01 3.30037475e-01 -3.32043618e-01 1.48424819e-01
5.10139406e-01 1.91874411e-02 5.18031903e-02 -1.84447333e-01
-1.25735176e+00 8.05902541e-01 1.15693778e-01 4.67387497e-01
-1.21089971e+00 -6.12530470e-01 -5.38198173e-01 -3.06479514e-01
5.06631434e-01 -5.70297122e-01 6.76628888e-01 -1.26745868e+00
-1.53890944e+00 6.72616839e-01 -2.19267890e-01 -4.09168638e-02
7.09111750e-01 9.52642784e-02 -4.54305649e-01 -1.72379926e-01
4.32882505e-03 2.08589002e-01 1.23777354e+00 -2.12453651e+00
-3.91398072e-01 -2.50202835e-01 -6.37860075e-02 4.28063124e-01
-4.69167456e-02 -3.75598192e-01 -6.11381054e-01 -9.52189028e-01
4.80024144e-02 -1.04024541e+00 -1.16807945e-01 3.69255580e-02
-8.78550470e-01 6.20349705e-01 1.32274318e+00 -8.92739117e-01
9.49748516e-01 -2.00940657e+00 -1.84352875e-01 1.88303813e-01
1.38445511e-01 4.24137056e-01 -1.93052679e-01 1.11067761e-02
1.45927742e-01 2.63836347e-02 -5.95188260e-01 -4.45375115e-01
-9.22920555e-02 3.25940818e-01 -6.38518512e-01 5.85078299e-01
6.12186827e-02 1.05053043e+00 -8.47979069e-01 -4.33478057e-01
4.26380903e-01 8.92952740e-01 2.35796884e-01 5.71912348e-01
3.41372639e-02 7.74869442e-01 -4.44254756e-01 8.00309539e-01
1.27129030e+00 1.32075697e-01 1.33765861e-01 -3.58673334e-01
2.03087017e-01 -2.65797615e-01 -1.03691614e+00 1.18546796e+00
-6.13704324e-01 7.03105569e-01 1.48419812e-01 -2.46270642e-01
1.08247542e+00 -2.06279941e-02 -1.18443839e-01 -1.01067245e+00
1.43631235e-01 1.86640307e-01 -3.83530676e-01 -1.52676702e-01
6.09181404e-01 -1.88931897e-01 1.59023842e-03 3.68504226e-01
-5.20797193e-01 -3.48954916e-01 -3.36882174e-01 8.33351687e-02
8.68458152e-01 3.99275899e-01 -7.39461854e-02 1.10719122e-01
6.83084369e-01 -4.39620912e-01 7.16072679e-01 6.71832502e-01
-1.12386113e-02 8.84860098e-01 2.18409717e-01 7.12118223e-02
-8.09653163e-01 -1.22481430e+00 1.61879256e-01 7.40696669e-01
7.20187247e-01 3.10820282e-01 -9.10034776e-01 -8.54374170e-01
-2.40834311e-01 8.09375048e-01 -6.63144052e-01 -9.47698280e-02
-5.35364091e-01 -6.86490297e-01 5.60017645e-01 2.66863644e-01
9.38603818e-01 -1.19083738e+00 -2.60798723e-01 -1.81063741e-01
-1.52310625e-01 -1.07382679e+00 -6.08410001e-01 -3.05478983e-02
-3.35779965e-01 -1.01389194e+00 -9.05500889e-01 -6.86980128e-01
9.20571446e-01 5.42464495e-01 1.01058829e+00 3.47923398e-01
-2.79158562e-01 -6.11595176e-02 -2.64159769e-01 -4.84227985e-01
-5.99081159e-01 -6.32237136e-01 -2.29408011e-01 3.28876793e-01
-6.60206556e-01 -4.79451776e-01 -1.11342454e+00 5.76830745e-01
-1.05111217e+00 4.98332530e-01 1.03012013e+00 8.52726340e-01
6.47781610e-01 3.02087218e-01 2.54065424e-01 -1.31170607e+00
4.93739337e-01 -1.12372488e-01 -6.00607038e-01 4.10499513e-01
-8.21942687e-01 -3.48945819e-02 8.57088506e-01 -3.71054471e-01
-1.74362922e+00 1.11977784e-02 1.36761963e-01 -5.72548211e-01
-1.37463585e-01 -2.85409898e-01 -5.97524464e-01 -5.21160841e-01
5.16669452e-01 6.50197685e-01 -4.52217817e-01 -1.51367173e-01
4.67431754e-01 4.67342585e-01 8.09711337e-01 -2.92858392e-01
1.42392504e+00 8.01029861e-01 1.10845305e-01 -3.72702479e-01
-8.90781939e-01 -2.65167337e-02 -1.86106429e-01 -3.52276266e-01
5.99455118e-01 -8.20293903e-01 -3.33895236e-01 9.18028474e-01
-1.13965833e+00 -6.55150712e-01 -2.54477143e-01 -2.04504684e-01
-3.33581835e-01 4.31355625e-01 -2.97836334e-01 -1.00729048e+00
-5.16607881e-01 -1.20334291e+00 1.25648189e+00 4.49671298e-01
6.20300710e-01 -7.85481513e-01 -1.18237741e-01 3.31809610e-01
5.42705417e-01 6.62124693e-01 4.64156717e-01 2.19821081e-01
-9.12791014e-01 1.00058708e-02 -5.35081089e-01 6.34652138e-01
3.61623496e-01 -2.74948239e-01 -1.21911037e+00 -3.02819937e-01
1.87296867e-01 -1.98895279e-02 9.66698945e-01 1.36752963e-01
1.26808083e+00 -3.96862298e-01 -3.31701815e-01 8.83047879e-01
1.68156946e+00 2.68091351e-01 1.26729465e+00 6.86399117e-02
9.77672994e-01 1.66269332e-01 1.05257690e+00 3.36814642e-01
1.23647630e-01 5.75491548e-01 5.08502603e-01 -8.34023833e-01
-8.11987281e-01 -2.78213620e-01 3.71005118e-01 4.21715081e-01
-1.80775568e-01 -8.01877558e-01 -2.86787450e-01 2.88394839e-01
-1.49659014e+00 -8.24765742e-01 -3.27886522e-01 2.08118200e+00
7.87587166e-01 -2.63388287e-02 -6.78489506e-01 4.18616906e-02
8.91558468e-01 7.21307814e-01 -6.62958801e-01 1.81855075e-02
-4.34458345e-01 4.87807423e-01 7.73793519e-01 5.31108260e-01
-7.97576189e-01 1.01876223e+00 5.72099304e+00 9.56936657e-01
-9.80325222e-01 1.51967049e-01 9.45650339e-01 3.99130970e-01
-7.16670096e-01 2.01123282e-01 -4.21578944e-01 7.09078372e-01
3.60657305e-01 3.56246233e-01 6.85099244e-01 4.32251632e-01
1.70394704e-01 -2.10226119e-01 -4.89269853e-01 8.91688347e-01
3.58881146e-01 -1.05808842e+00 -1.76983297e-01 4.14784066e-02
1.30304098e+00 -5.95187247e-01 4.60509628e-01 1.29907385e-01
3.07236582e-01 -1.04900765e+00 5.99611461e-01 9.54020917e-01
1.22748268e+00 -6.91510379e-01 8.31726372e-01 1.06494687e-01
-1.16092861e+00 8.71167779e-02 -2.20178246e-01 3.25807840e-01
2.04349428e-01 8.34874332e-01 -8.78936350e-01 6.95380330e-01
4.10702646e-01 3.02400321e-01 -4.87110406e-01 3.97970557e-01
-9.70957696e-01 4.82382327e-01 1.97439685e-01 2.61561304e-01
-2.11764932e-01 -3.22246671e-01 4.91524220e-01 9.32490945e-01
3.71344268e-01 3.30123544e-01 -4.02805060e-02 1.13837492e+00
-2.33858123e-01 -3.79714757e-01 -4.39369828e-01 3.47874075e-01
4.09361064e-01 1.37079585e+00 -6.62991226e-01 -3.46487254e-01
-4.53811809e-02 1.59421527e+00 -2.32977524e-01 7.89247930e-01
-1.29666471e+00 -4.08183664e-01 4.15099949e-01 2.54797544e-02
1.84048265e-01 1.50234669e-01 -3.20885181e-01 -1.04411542e+00
5.26859686e-02 -8.36231053e-01 -2.58045644e-01 -1.37987041e+00
-1.10771608e+00 6.91033185e-01 -5.05358636e-01 -1.09621704e+00
9.62651297e-02 -1.95086479e-01 -8.66967678e-01 1.13030398e+00
-1.92479920e+00 -1.47644734e+00 -1.05712402e+00 6.62613571e-01
3.56243908e-01 -1.02549881e-01 5.64204335e-01 -3.06671276e-03
-3.69203031e-01 6.47255719e-01 3.42949301e-01 -1.12969503e-02
8.49279225e-01 -1.00532293e+00 3.80884230e-01 1.16569602e+00
-3.78596872e-01 4.18489464e-02 7.15418100e-01 -9.02845025e-01
-1.52453935e+00 -1.37194359e+00 2.46075448e-02 -2.40464136e-01
1.86060756e-01 -2.85482854e-01 -7.56026089e-01 3.32995057e-01
2.82855600e-01 1.38059542e-01 4.37549315e-02 -7.53126323e-01
-1.00734115e-01 -4.73540604e-01 -1.53337944e+00 6.20930433e-01
1.09224367e+00 -3.85543466e-01 -1.09632023e-01 2.84664094e-01
8.73736143e-01 -7.28106737e-01 -3.39817405e-01 6.59498334e-01
4.22957242e-01 -1.29388750e+00 1.12254786e+00 3.22698116e-01
3.95994544e-01 -7.03975201e-01 -1.37290701e-01 -1.30633759e+00
-3.10968757e-02 -6.81941032e-01 -9.99883935e-02 1.47556400e+00
1.20693743e-01 -9.87475216e-01 8.45020175e-01 2.27654457e-01
-2.74664044e-01 -8.02404404e-01 -4.57460970e-01 -6.31332040e-01
-5.13774574e-01 -1.00291811e-01 8.55181873e-01 6.32381439e-01
-1.17578995e+00 -4.56006452e-02 -8.85684669e-01 4.68853742e-01
1.04664230e+00 4.68095899e-01 1.09863961e+00 -6.46757603e-01
-5.45452774e-01 7.94428065e-02 -6.06392138e-03 -9.55862045e-01
1.69219136e-01 -4.08083886e-01 5.75621068e-01 -1.47570014e+00
2.80221879e-01 -6.22487843e-01 -4.02206033e-01 3.95948499e-01
-4.86207515e-01 8.66299033e-01 1.84882358e-01 2.26668864e-01
-2.85999149e-01 7.11765647e-01 1.78537691e+00 -2.61529326e-01
-3.30892980e-01 -4.58299555e-02 -9.01247263e-01 5.26542723e-01
7.83560991e-01 -4.00809675e-01 -5.29882848e-01 -1.79738641e-01
-2.03572392e-01 1.53478429e-01 7.08082855e-01 -9.30501640e-01
-2.28157401e-01 -4.23583031e-01 7.54918098e-01 -5.40711820e-01
5.75679779e-01 -7.36254513e-01 4.79215860e-01 3.81991416e-01
3.85316424e-02 -5.85143328e-01 8.25922787e-02 8.00007582e-01
4.73534614e-02 2.26830915e-01 8.66650641e-01 -1.07512370e-01
-6.75758779e-01 3.20490003e-01 1.77827969e-01 -4.84290086e-02
1.07233667e+00 -3.15234363e-01 -8.57888997e-01 -6.89700603e-01
-1.94990650e-01 -1.59823015e-01 7.44596958e-01 1.77638158e-01
1.00380719e+00 -1.30260944e+00 -7.36073017e-01 4.21208829e-01
-1.51485369e-01 -9.68332402e-03 4.41228181e-01 6.59046173e-01
-4.50453043e-01 -1.42148230e-02 -3.26407701e-02 -2.21545070e-01
-1.31478631e+00 5.60730398e-01 2.34881744e-01 -2.96756953e-01
-4.09094632e-01 6.83128178e-01 9.22636986e-01 -2.65764922e-01
2.01159474e-02 2.67394297e-02 2.45706931e-01 -5.95431030e-01
3.14853609e-01 4.15913016e-01 -7.54918456e-02 -6.35729969e-01
-1.73039347e-01 5.20839155e-01 3.40525001e-01 1.15194591e-02
1.17666602e+00 -3.38272393e-01 -2.01122209e-01 -2.54213735e-02
9.79463756e-01 3.64082187e-01 -1.55613816e+00 -1.81139465e-02
-9.21147943e-01 -1.02180433e+00 9.70165804e-02 -1.14732933e+00
-1.56954813e+00 5.41769803e-01 1.04230881e+00 -3.53191718e-02
1.73004675e+00 -1.20304815e-01 1.13069129e+00 -1.61637440e-01
2.95221806e-01 -8.77628863e-01 2.62039483e-01 5.70156947e-02
1.08765185e+00 -1.10069811e+00 2.39290878e-01 -5.48800886e-01
-7.49182403e-01 6.15367174e-01 7.44794488e-01 -2.29911059e-01
1.70735419e-01 4.80606228e-01 1.53614953e-01 1.59797594e-01
-3.06989193e-01 -2.95181215e-01 3.89775574e-01 9.54732358e-01
-9.10480767e-02 1.89192846e-01 -8.75524729e-02 3.21579099e-01
2.71558878e-03 -1.49844706e-01 6.00130320e-01 6.08140171e-01
-3.32108200e-01 -1.08782876e+00 -8.28409731e-01 2.76339680e-01
-6.37107119e-02 -2.68994570e-01 -5.79239011e-01 7.32715964e-01
4.07016397e-01 1.08229184e+00 -3.43165606e-01 -5.38586199e-01
1.19461216e-01 -5.37112415e-01 5.18674970e-01 -1.69122323e-01
-1.64120987e-01 6.80210367e-02 2.82146540e-02 -5.74064732e-01
-2.37424135e-01 -1.07535698e-01 -1.05627942e+00 -2.97592968e-01
-5.64110994e-01 -2.14975908e-01 6.26274526e-01 6.87333822e-01
2.37680316e-01 8.21857989e-01 1.16781056e+00 -1.13688195e+00
-7.92844519e-02 -7.03061044e-01 -6.60202146e-01 2.79902101e-01
4.67790276e-01 -5.13363898e-01 -6.41695261e-01 -5.25094382e-02]
|
[10.848403930664062, -4.100964069366455]
|
936c73d0-a303-49fb-bf6d-d92f6c7648d9
|
the-effects-of-system-initiative-during
|
2202.09728
| null |
https://arxiv.org/abs/2202.09728v1
|
https://arxiv.org/pdf/2202.09728v1.pdf
|
The Effects of System Initiative during Conversational Collaborative Search
|
Our research in this paper lies at the intersection of collaborative and conversational search. We report on a Wizard of Oz lab study in which 27 pairs of participants collaborated on search tasks over the Slack messaging platform. To complete tasks, pairs of collaborators interacted with a so-called \emph{searchbot} with conversational capabilities. The role of the searchbot was played by a reference librarian. It is widely accepted that conversational search systems should be able to engage in \emph{mixed-initiative interaction} -- take and relinquish control of a multi-agent conversation as appropriate. Research in discourse analysis differentiates between dialog- and task-level initiative. Taking \emph{dialog-level} initiative involves leading a conversation for the sole purpose of establishing mutual belief between agents. Conversely, taking \emph{task-level} initiative involves leading a conversation with the intent to influence the goals of the other agent(s). Participants in our study experienced three \emph{searchbot conditions}, which varied based on the level of initiative the human searchbot was able to take: (1) no initiative, (2) only dialog-level initiative, and (3) both dialog- and task-level initiative. We investigate the effects of the searchbot condition on six different types of outcomes: (RQ1) perceptions of the searchbot's utility, (RQ2) perceptions of workload, (RQ3) perceptions of the collaboration, (RQ4) patterns of communication and collaboration, and perceived (RQ5) benefits and (RQ6) challenges from engaging with the searchbot.
|
['Jaime Arguello', 'Bogeum Choi', 'Sandeep Avula']
|
2022-02-20
| null | null | null | null |
['conversational-search']
|
['natural-language-processing']
|
[-4.15598676e-02 6.31591082e-01 -1.92149468e-02 -3.68557304e-01
-6.00447834e-01 -7.65171945e-01 8.70159507e-01 4.27243680e-01
-6.02309108e-01 5.51748395e-01 6.80920959e-01 -6.56286240e-01
-3.01850766e-01 -3.65800291e-01 9.87293720e-02 -2.93360114e-01
5.07353783e-01 3.81909013e-01 7.54505247e-02 -4.57440674e-01
6.71665907e-01 2.34959126e-01 -1.16529727e+00 1.63831443e-01
5.93255997e-01 4.51070338e-01 4.34053093e-01 7.79137909e-01
-1.51932716e-01 1.60546350e+00 -9.52547967e-01 -4.55325171e-02
3.93097922e-02 -5.39249957e-01 -1.42098224e+00 3.17686975e-01
-2.22537652e-01 -5.48556685e-01 -1.73894882e-01 6.57473445e-01
4.50327456e-01 5.36884606e-01 1.62392944e-01 -1.70580089e+00
-1.12579554e-01 5.55101812e-01 2.64735937e-01 2.81339109e-01
8.54420781e-01 6.36252999e-01 9.57654953e-01 -3.61211449e-01
9.17637229e-01 1.25315547e+00 4.03761476e-01 3.56596917e-01
-1.28386927e+00 -5.80933332e-01 3.87382582e-02 -2.79442102e-01
-1.07108605e+00 -7.79079437e-01 4.40635979e-01 -7.04625368e-01
1.04052806e+00 4.42016244e-01 3.47047091e-01 9.04140413e-01
2.26619486e-02 1.28352150e-01 1.32380044e+00 -4.16092753e-01
3.22665095e-01 7.39958227e-01 5.27519107e-01 1.24040946e-01
-8.71106982e-02 -2.34809950e-01 -8.29525650e-01 -7.02276051e-01
6.30033851e-01 -2.67602533e-01 -3.56841892e-01 3.35454434e-01
-1.31478441e+00 7.55646169e-01 3.44743468e-02 7.32879102e-01
-6.18397057e-01 5.86295091e-02 4.77123827e-01 6.22095644e-01
2.13467583e-01 9.69596982e-01 -6.51919693e-02 -1.04548621e+00
-2.75528021e-02 2.56040066e-01 1.45877683e+00 8.27647090e-01
6.64447606e-01 -4.26068813e-01 -3.06665242e-01 7.95084417e-01
3.35482895e-01 -3.88959646e-02 2.45798588e-01 -1.51304030e+00
5.36538422e-01 8.24244142e-01 8.83894086e-01 -8.88011992e-01
-4.94674176e-01 3.51283818e-01 6.01599030e-02 2.42065147e-01
7.42309093e-01 -6.93028390e-01 4.05916274e-02 1.64782321e+00
2.57772505e-01 -6.85423613e-01 -3.34717408e-02 8.91888916e-01
8.73999059e-01 3.52444381e-01 1.23526910e-02 -4.20648813e-01
1.33447170e+00 -8.84243965e-01 -1.09341204e+00 -3.44875336e-01
1.04252028e+00 -1.06260645e+00 1.20509350e+00 -1.71473056e-01
-1.06611526e+00 -3.42429996e-01 -8.19645464e-01 4.05873358e-03
-7.54986480e-02 -3.78369272e-01 3.03071827e-01 5.56980133e-01
-1.27906036e+00 1.65419072e-01 -4.75755721e-01 -6.33812070e-01
-5.84939063e-01 1.03012621e-01 -3.43626529e-01 3.14751863e-01
-1.08128691e+00 1.13791764e+00 -3.17426234e-01 1.45021126e-01
-4.13392961e-01 -1.74238786e-01 -6.46133065e-01 2.23740432e-02
4.73668516e-01 -4.36872452e-01 1.75275683e+00 -9.37362552e-01
-1.68506622e+00 7.61501253e-01 -1.18317775e-01 1.43679500e-01
4.11578715e-01 5.31223491e-02 2.60794044e-01 7.71114081e-02
5.48834622e-01 3.37419838e-01 -1.43457338e-01 -1.13265836e+00
-5.38770616e-01 -3.10926080e-01 8.27041686e-01 7.63156056e-01
1.01350039e-01 4.15212482e-01 1.47829756e-01 1.19527750e-01
1.22446865e-02 -1.06036353e+00 1.55229375e-01 -2.79229015e-01
-3.89164150e-01 -6.72109962e-01 5.65747023e-01 -3.47544163e-01
1.13039553e+00 -2.14058685e+00 -2.92240679e-01 -2.29404882e-01
4.53787118e-01 1.39147667e-02 2.38946885e-01 1.21717429e+00
4.02661175e-01 4.35633659e-01 7.69420385e-01 -3.68193567e-01
2.63799340e-01 -2.24645138e-01 2.72755355e-01 3.39590907e-01
-4.47727829e-01 4.48348314e-01 -9.09202337e-01 -3.02230060e-01
1.17708952e-03 4.64662649e-02 -1.41974911e-01 5.00754356e-01
-5.74290045e-02 5.18820167e-01 -5.45166314e-01 1.70354322e-01
7.97265097e-02 -3.35226119e-01 3.39340985e-01 4.13794070e-01
-8.32082629e-01 1.05734563e+00 -8.87949765e-01 8.58275056e-01
-5.95923603e-01 9.89359677e-01 9.53590155e-01 -3.32334310e-01
8.80814672e-01 8.31288815e-01 1.62009239e-01 -4.37662423e-01
1.30852014e-01 2.02835962e-01 5.08990347e-01 -5.37144065e-01
5.84999979e-01 2.01911792e-01 -1.55733615e-01 1.27163172e+00
-3.92911494e-01 -1.87400162e-01 -1.03839748e-01 4.10756767e-01
1.37802780e+00 -3.78758311e-01 2.39471316e-01 -2.68341362e-01
1.15435496e-01 2.47994512e-01 2.73004472e-01 1.03113604e+00
-6.93361580e-01 -1.43653542e-01 8.70313168e-01 -1.77257791e-01
-4.90070671e-01 -3.62026423e-01 4.07644242e-01 1.37901688e+00
5.07644355e-01 -3.96762639e-01 -5.46306431e-01 -1.85388833e-01
-2.91401833e-01 1.06541777e+00 -1.58334211e-01 -6.86720805e-03
-2.55871356e-01 1.00558899e-01 5.80648959e-01 5.05155064e-02
9.23106670e-01 -1.21737146e+00 -1.03961647e+00 1.93753019e-01
-9.99585271e-01 -1.01917613e+00 -9.55743492e-01 -1.04498670e-01
-3.31575960e-01 -1.12239528e+00 2.21586861e-02 -5.77601135e-01
4.17389184e-01 7.14341760e-01 7.14061022e-01 5.21114647e-01
1.65337682e-01 8.37234616e-01 -3.94930542e-01 -2.90117145e-01
-5.98830938e-01 -3.13496709e-01 -1.74543068e-01 -3.24228019e-01
3.51838499e-01 -2.87814140e-01 -5.40793896e-01 8.67452204e-01
-4.09810573e-01 2.56642222e-01 2.53129564e-02 5.21458566e-01
-4.74574119e-01 -1.37882218e-01 6.68423295e-01 -4.24778104e-01
1.50979948e+00 -4.02089030e-01 -8.12401026e-02 9.14564431e-02
-4.73252684e-01 -5.71403384e-01 -3.15204002e-02 -3.98381650e-01
-1.27776372e+00 -5.07809520e-01 4.15648907e-01 2.64907479e-01
-1.79252625e-01 6.41766250e-01 1.19145237e-01 4.03318107e-02
6.37814820e-01 -1.26411468e-01 4.50756907e-01 3.87498587e-02
-1.51598245e-01 1.19672239e+00 -1.17782429e-01 -6.52674735e-01
-6.70624338e-03 -2.77049728e-02 -6.90066099e-01 -1.09374022e+00
-1.17423132e-01 -5.84881783e-01 -1.52988732e-01 -6.45806730e-01
9.57081974e-01 -7.67402172e-01 -1.65422273e+00 5.45612872e-01
-1.34900796e+00 -9.95939672e-01 2.60052800e-01 6.04235411e-01
-4.23194498e-01 -1.12807341e-02 -6.42487943e-01 -1.41773808e+00
-7.25276545e-02 -1.43914735e+00 4.51439768e-01 5.63074946e-01
-1.12829947e+00 -8.83054018e-01 -1.56485885e-01 1.04325128e+00
5.53623021e-01 -1.10672377e-01 6.46469414e-01 -1.17072725e+00
-5.99460244e-01 -3.38773608e-01 -1.38980851e-01 -4.42472361e-02
6.07441247e-01 -2.18654498e-01 -6.20278418e-01 -7.97607657e-03
5.09809792e-01 -6.47333562e-01 -4.06972677e-01 2.02554643e-01
-1.72994837e-01 -5.69669485e-01 -4.40587282e-01 -5.70824325e-01
7.92121649e-01 8.27330232e-01 1.61039039e-01 4.86970395e-01
-2.81747198e-03 1.00021279e+00 8.48748922e-01 2.57219344e-01
7.97801673e-01 8.58840168e-01 2.25155260e-02 3.54964316e-01
2.17393294e-01 -1.07483260e-01 4.72451776e-01 1.71674639e-01
7.16061145e-02 -2.37228766e-01 -9.92633581e-01 5.42470396e-01
-1.99495542e+00 -8.45211387e-01 -9.85610038e-02 1.92477691e+00
7.65751958e-01 1.98617950e-01 2.27008581e-01 -2.90016592e-01
9.84052360e-01 2.10394189e-01 -2.68767565e-01 -7.30527401e-01
6.95196986e-01 -3.72434080e-01 -2.42548101e-02 1.02089500e+00
-3.46910477e-01 5.82493007e-01 5.16780710e+00 -6.61266921e-03
-7.93131709e-01 3.36812586e-01 5.89983165e-01 -7.65792802e-02
-1.76045999e-01 4.62250769e-01 -6.55856788e-01 2.91731030e-01
6.45344257e-01 -5.35651326e-01 4.50769573e-01 5.91186821e-01
6.54907167e-01 -9.33618426e-01 -1.33175004e+00 3.77927244e-01
-2.40758047e-01 -1.05649674e+00 -1.00412512e+00 2.39482701e-01
8.48689303e-02 -3.07832330e-01 -5.12214959e-01 3.01335335e-01
5.50800622e-01 -7.77504385e-01 5.99429309e-01 3.02132219e-01
3.26717228e-01 -1.44747108e-01 8.24191868e-01 9.63863730e-01
-8.33421707e-01 1.59168802e-02 4.41042215e-01 -9.19358909e-01
4.17897701e-01 -2.87838012e-01 -1.24999118e+00 4.78559965e-03
4.84769821e-01 -3.95602793e-01 1.87283620e-01 2.69345164e-01
-1.99418869e-02 2.57557720e-01 -2.64842957e-01 -6.69797361e-01
1.78634077e-01 -3.09543252e-01 6.83031142e-01 7.01090634e-01
-3.25747073e-01 5.55037677e-01 2.59710282e-01 9.13034260e-01
9.49407071e-02 -2.22280160e-01 -6.96794808e-01 -3.58665258e-01
1.31907713e+00 1.05712640e+00 -9.76197422e-01 -3.00316244e-01
-1.71816573e-01 4.14994776e-01 -2.33295802e-02 4.53330636e-01
-2.94284731e-01 -5.53772867e-01 8.40821445e-01 1.94379345e-01
-2.90144980e-01 -2.55667567e-01 -4.82171148e-01 -3.92053396e-01
5.99054620e-02 -1.06586015e+00 -2.39463046e-01 -1.00092554e+00
-7.06672132e-01 4.04137433e-01 4.65401597e-02 -3.77859294e-01
-4.13150579e-01 2.38901600e-01 -7.49660254e-01 1.40045488e+00
-5.34683406e-01 -5.52412570e-01 -5.56711972e-01 -4.84289713e-02
7.87140369e-01 2.39337683e-01 8.82400811e-01 -1.09948896e-01
-2.73274809e-01 1.26332089e-01 -4.83954996e-01 1.38279097e-02
9.94114637e-01 -7.04988956e-01 -2.63900906e-01 1.04559228e-01
-5.51339388e-01 1.16157460e+00 7.32798040e-01 -7.79142499e-01
-1.44693100e+00 -1.46144375e-01 1.49475896e+00 -4.05620456e-01
6.29263282e-01 -2.14721769e-01 -4.79322821e-01 9.52439249e-01
5.06564379e-01 -7.80910432e-01 6.91020608e-01 4.48601335e-01
2.56718665e-01 1.28853709e-01 -1.20649636e+00 8.37729573e-01
9.31391120e-01 -9.59494710e-01 -7.34725654e-01 5.80180407e-01
8.53432357e-01 -4.41501468e-01 -6.50418520e-01 -2.58639216e-01
4.99885529e-01 -1.21435690e+00 4.66239989e-01 -1.26514196e-01
3.44608039e-01 1.24873266e-01 7.16385320e-02 -1.04033124e+00
3.28301601e-02 -1.19683909e+00 7.28796661e-01 1.30729139e+00
2.76145458e-01 -1.02800441e+00 2.82096833e-01 1.75534511e+00
-2.44941249e-01 -3.76932383e-01 -7.68658280e-01 -2.66496062e-01
-2.89694935e-01 7.84165785e-03 2.68654555e-01 8.49430323e-01
1.12725818e+00 6.57038987e-01 1.58270478e-01 -2.67732497e-02
-1.63833752e-01 -2.41070822e-01 1.07252300e+00 -8.61296773e-01
-1.09306894e-01 -5.74819565e-01 2.40740001e-01 -1.12026191e+00
-1.75963104e-01 -3.29505026e-01 4.04865026e-01 -1.88426626e+00
-4.75271717e-02 -6.15872085e-01 6.07160389e-01 2.68927217e-01
-1.13931067e-01 -8.20826292e-01 5.25011063e-01 5.13380349e-01
-4.39246744e-01 1.01804852e-01 1.01221967e+00 2.22351208e-01
-8.85873914e-01 2.83857495e-01 -1.08512104e+00 4.46291208e-01
5.67208886e-01 -4.44674492e-02 -6.54582739e-01 -8.54076892e-02
3.49591762e-01 9.34245110e-01 1.94582537e-01 -2.83116549e-01
8.69577885e-01 -4.84157205e-01 -5.67402005e-01 -1.17995262e-01
5.78285277e-01 -5.20762622e-01 3.77259254e-01 4.13997233e-01
-7.27080286e-01 1.40844032e-01 6.10203966e-02 2.06403621e-02
-1.41058967e-01 -4.62296218e-01 2.60863900e-01 -3.55237842e-01
2.10764453e-01 -8.19207549e-01 -1.22830677e+00 7.33873770e-02
1.00536144e+00 -4.97545689e-01 -7.93242872e-01 -1.24914193e+00
-5.90880990e-01 5.18806815e-01 5.04287541e-01 2.85387009e-01
8.66693184e-02 -6.12274051e-01 -2.51464933e-01 -3.60390276e-01
-2.07639888e-01 2.40695942e-02 -1.44045755e-01 1.29554284e+00
-4.82947201e-01 7.19947040e-01 1.53674424e-01 -3.59521359e-01
-1.44300342e+00 -1.85885206e-01 3.34684789e-01 -1.42718002e-01
-1.99518368e-01 8.97938669e-01 1.72525749e-01 -4.58763063e-01
4.63424742e-01 -3.86009663e-02 -7.35442666e-03 2.31951550e-01
3.82724762e-01 7.56133080e-01 -2.31453225e-01 -5.34912765e-01
-3.19794118e-01 -2.68067747e-01 -1.29348272e-03 -7.20457017e-01
7.74342179e-01 -5.16716242e-01 -4.01663095e-01 9.64035809e-01
8.23893189e-01 -8.06430951e-02 -8.12096894e-01 -1.05431415e-01
6.37095794e-02 -5.19037843e-01 -2.90343165e-01 -9.39102471e-01
1.35226652e-01 1.41077369e-01 -1.67360559e-01 8.97432387e-01
2.66265124e-01 8.43992680e-02 3.00951332e-01 5.84783494e-01
5.84661484e-01 -1.31289136e+00 6.30135000e-01 5.11764467e-01
1.04346645e+00 -1.07250500e+00 -2.69573003e-01 -6.48625970e-01
-1.02705729e+00 8.52170229e-01 9.59946275e-01 4.79682148e-01
6.61063910e-01 4.35854569e-02 3.30689907e-01 -6.31412685e-01
-1.14344370e+00 1.57557175e-01 -5.26967645e-01 3.85928601e-01
7.12451279e-01 3.44038680e-02 -7.75428116e-01 6.07511461e-01
-1.30460307e-01 9.21902955e-02 7.85875380e-01 1.23881364e+00
-4.62812990e-01 -6.56328619e-01 -4.77720737e-01 3.90150726e-01
-3.05702556e-02 2.25024149e-01 -1.12376499e+00 8.96448195e-01
-2.47038171e-01 2.07524586e+00 1.67027444e-01 -2.59878844e-01
5.23441970e-01 3.72434765e-01 -2.65483439e-01 -9.06884074e-01
-1.26302695e+00 1.00093223e-02 9.92199302e-01 -4.78846997e-01
-3.64518940e-01 -9.20541883e-01 -1.14683318e+00 -4.73415017e-01
-6.87704742e-01 4.82854515e-01 6.22585416e-01 1.19897604e+00
5.03826737e-01 1.51516497e-01 7.10759521e-01 -6.32763445e-01
-6.67605877e-01 -1.56001449e+00 -1.97634161e-01 1.85579881e-01
3.49393129e-01 -5.47248006e-01 -8.39689374e-01 -3.09577316e-01]
|
[12.45665168762207, 7.849400043487549]
|
aa8a7be0-7e66-4d52-be5d-3b854d6d7847
|
attention-based-transformer-networks-for
|
2305.05433
| null |
https://arxiv.org/abs/2305.05433v1
|
https://arxiv.org/pdf/2305.05433v1.pdf
|
Attention-Based Transformer Networks for Quantum State Tomography
|
Neural networks have been actively explored for quantum state tomography (QST) due to their favorable expressibility. To further enhance the efficiency of reconstructing quantum states, we explore the similarity between language modeling and quantum state tomography and propose an attention-based QST method that utilizes the Transformer network to capture the correlations between measured results from different measurements. Our method directly retrieves the density matrices of quantum states from measured statistics, with the assistance of an integrated loss function that helps minimize the difference between the actual states and the retrieved states. Then, we systematically trace different impacts within a bag of common training strategies involving various parameter adjustments on the attention-based QST method. Combining these techniques, we establish a robust baseline that can efficiently reconstruct pure and mixed quantum states. Furthermore, by comparing the performance of three popular neural network architectures (FCNs, CNNs, and Transformer), we demonstrate the remarkable expressiveness of attention in learning density matrices from measured statistics.
|
['Herschel Rabitz', 'Chunlin Chen', 'Daoyi Dong', 'Zhenhong Sun', 'Hailan Ma']
|
2023-05-09
| null | null | null | null |
['quantum-state-tomography']
|
['medical']
|
[ 1.11005269e-01 -2.29879275e-01 1.03614010e-01 -2.13573456e-01
-1.24130106e+00 -3.31257343e-01 7.00653493e-01 -1.36721551e-01
-5.43477833e-01 7.38204658e-01 2.75148392e-01 -6.06163919e-01
-8.74290690e-02 -9.94803488e-01 -7.29070425e-01 -9.32835400e-01
3.65338683e-01 4.73952621e-01 -1.21953994e-01 -2.99442053e-01
3.42075348e-01 4.59910154e-01 -9.43783939e-01 -6.32463619e-02
9.86743510e-01 1.05537260e+00 -7.41755590e-02 3.54645967e-01
-1.56661823e-01 9.56975520e-01 -4.78529364e-01 -5.16262293e-01
1.48026958e-01 -6.07072711e-01 -7.69142628e-01 -6.17850840e-01
2.38360524e-01 -6.38248265e-01 -1.23512328e+00 1.43137169e+00
5.27862966e-01 2.75165677e-01 8.08262646e-01 -7.15834618e-01
-8.22182953e-01 6.80918396e-01 8.17875117e-02 3.13862830e-01
2.46844053e-01 4.91299301e-01 1.16319835e+00 -5.31380594e-01
3.46872091e-01 1.00148046e+00 4.04648453e-01 5.66772699e-01
-1.57590759e+00 -8.58371794e-01 -5.27033567e-01 5.55739462e-01
-1.56709993e+00 -7.31944382e-01 6.41339064e-01 -3.86153185e-03
1.38126218e+00 -1.86997607e-01 5.58456004e-01 1.24067056e+00
1.98836431e-01 5.58857858e-01 1.13639915e+00 -4.87063438e-01
2.03241274e-01 1.63928851e-01 4.29005139e-02 8.64931047e-01
-9.68209505e-02 4.57275301e-01 -6.68120146e-01 -1.16873235e-01
5.95562994e-01 -1.38132378e-01 -3.64381522e-01 -3.32407743e-01
-1.29852891e+00 7.26479530e-01 6.45418823e-01 1.42449424e-01
-2.27789938e-01 4.20243770e-01 3.59040260e-01 2.44000033e-01
1.39461026e-01 6.27838731e-01 -9.68136918e-03 -1.61254153e-01
-7.23616123e-01 -6.57102391e-02 6.47098660e-01 8.23378563e-01
1.17125714e+00 1.06159784e-01 -4.37259197e-01 2.28560612e-01
9.76037309e-02 8.95968139e-01 1.32755861e-01 -8.58103573e-01
3.63792062e-01 1.37620583e-01 1.87594429e-01 -4.21128631e-01
-6.78784624e-02 -2.32762575e-01 -9.23562527e-01 -4.56749707e-01
1.14646137e-01 5.55069260e-02 -8.09643507e-01 1.84860706e+00
7.78667554e-02 2.70133585e-01 2.38659576e-01 7.44752109e-01
6.27739370e-01 8.93671274e-01 -1.03129126e-01 -1.63537562e-02
8.46850395e-01 -6.22827530e-01 -6.85228288e-01 1.25018820e-01
6.81032836e-01 -2.36181334e-01 9.32121694e-01 5.85049242e-02
-9.83730137e-01 -3.84862423e-01 -1.13634539e+00 -5.87603629e-01
-3.20490032e-01 -1.11088477e-01 7.58333623e-01 5.61376631e-01
-9.84965265e-01 1.20069635e+00 -1.03823376e+00 1.71559099e-02
3.50617975e-01 4.68242079e-01 -3.41185421e-01 1.52539536e-02
-1.39278221e+00 1.06426084e+00 5.78277051e-01 2.34725714e-01
-1.12018085e+00 -6.59387827e-01 -7.12755680e-01 4.80781227e-01
2.96758860e-01 -7.98569143e-01 1.18163252e+00 -9.60242972e-02
-2.08265233e+00 4.85122442e-01 -2.76913822e-01 -4.90535736e-01
-1.08570211e-01 6.32000640e-02 -3.59565228e-01 2.80338883e-01
9.71361771e-02 3.40458781e-01 4.72631782e-01 -5.18278658e-01
7.23292008e-02 -1.13805585e-01 9.28679705e-02 -2.01198980e-01
-2.13797525e-01 -2.07324743e-01 -3.48572820e-01 2.77575999e-01
2.62731642e-01 -8.63120317e-01 -1.10152759e-01 -4.03269678e-01
-7.86204159e-01 -1.58241406e-01 1.54684663e-01 -2.37554431e-01
9.03839290e-01 -2.13321853e+00 3.50448489e-01 1.78471625e-01
3.04987222e-01 3.00068110e-01 -2.63529390e-01 5.80269754e-01
9.75000039e-02 1.61497444e-02 -2.10783169e-01 -5.43694437e-01
4.27974254e-01 2.31746957e-01 -4.97399569e-01 3.45231801e-01
5.16519964e-01 1.28907406e+00 -7.94080734e-01 -2.44217277e-01
2.98288792e-01 3.76105160e-01 -7.52565205e-01 3.37988138e-01
-1.78145066e-01 7.98524380e-01 -5.25270522e-01 2.79383391e-01
5.59485495e-01 -6.45280898e-01 3.07989359e-01 -7.12459147e-01
-2.28351787e-01 1.25336754e+00 -5.23811281e-01 1.93030703e+00
-4.80173856e-01 4.34021145e-01 -1.99732646e-01 -1.18013561e+00
5.76439619e-01 2.72682250e-01 2.49985293e-01 -1.16639185e+00
4.99168485e-01 1.95538893e-01 1.69510782e-01 -3.09826732e-01
5.30134201e-01 -7.36968517e-01 -2.08058879e-01 6.43622816e-01
8.41481924e-01 -4.80355173e-01 4.51997668e-03 5.74186742e-01
9.04210269e-01 -1.11360652e-02 1.76032737e-01 -1.42205372e-01
2.34742835e-01 -3.91328484e-01 1.83437660e-01 1.09463298e+00
-3.30942005e-01 2.13515833e-01 5.87708771e-01 -1.29135102e-01
-1.26295555e+00 -1.38294363e+00 -3.08423460e-01 6.33783340e-01
1.23489618e-01 -6.59545958e-01 -3.35105807e-01 -3.81979138e-01
-2.62024313e-01 9.05066013e-01 -5.60848296e-01 -8.54852736e-01
-2.11048082e-01 -7.95105159e-01 6.91234827e-01 3.44806939e-01
7.47997522e-01 -8.48292410e-01 1.03345573e-01 9.51590911e-02
-3.26375961e-01 -1.38264215e+00 -2.94296175e-01 4.72155482e-01
-6.23028398e-01 -6.30554914e-01 -1.75200507e-01 -3.47350091e-02
1.30479708e-01 2.41954550e-02 9.81055558e-01 -3.30075115e-01
6.68273121e-02 2.70772189e-01 9.25725773e-02 -5.23428060e-02
-6.67901516e-01 2.81403124e-01 2.78076231e-01 -9.77533963e-03
5.10317624e-01 -7.67977893e-01 -3.29639286e-01 -2.71541983e-01
-7.44717777e-01 6.59917742e-02 5.51172495e-01 8.16992462e-01
2.86312908e-01 -3.88791412e-01 4.06035595e-02 -6.78173244e-01
5.04801333e-01 -4.36839670e-01 -7.93904245e-01 3.05811048e-01
-3.88677001e-01 8.50973308e-01 7.28582084e-01 -3.34632061e-02
-9.70669448e-01 -5.43554246e-01 -3.60555798e-01 -5.05184233e-01
1.26496017e-01 6.16969109e-01 5.61759956e-02 -2.67458647e-01
5.76284587e-01 7.53928363e-01 -1.79425701e-01 -2.12952420e-01
4.73135561e-01 4.87707347e-01 5.41524410e-01 -8.28107178e-01
8.89016926e-01 4.88000363e-01 2.56811112e-01 -5.50400794e-01
-1.46190488e+00 -2.98780203e-01 -5.27556121e-01 4.79022115e-01
8.45405340e-01 -8.90239477e-01 -1.03850472e+00 3.76219034e-01
-1.20446050e+00 -1.44574896e-01 -2.68588394e-01 7.83962905e-01
-5.59240997e-01 6.36116743e-01 -1.04221880e+00 -9.53627050e-01
-4.70845789e-01 -1.27699816e+00 1.19248974e+00 1.05821259e-01
3.52916032e-01 -9.77580547e-01 2.19977781e-01 2.02699497e-01
6.04812562e-01 -2.49228895e-01 1.32158744e+00 -4.12183017e-01
-1.14733124e+00 -1.57227769e-01 -6.08096600e-01 4.64678288e-01
-1.51762366e-01 -4.26272303e-01 -1.17604923e+00 -2.18257904e-01
3.05528194e-02 -6.95495427e-01 1.18015254e+00 1.55205265e-01
9.55385268e-01 -4.02336866e-02 -9.66899395e-02 9.47911441e-01
1.24010658e+00 7.45027140e-03 1.00364411e+00 -1.94072619e-01
8.05667579e-01 -3.47646885e-02 -4.09377903e-01 2.82408506e-01
2.99130440e-01 4.90797549e-01 1.17809080e-01 3.50657552e-01
1.96041688e-01 -5.06519616e-01 5.12008429e-01 1.39593780e+00
-7.04238489e-02 -7.89502487e-02 -5.57939887e-01 1.26068980e-01
-1.37509096e+00 -1.04413974e+00 3.23763341e-01 2.20916939e+00
7.63283551e-01 1.08141564e-01 -2.83183426e-01 -1.73066452e-01
2.28689507e-01 2.73181647e-01 -6.88660443e-01 -1.89591303e-01
3.08661219e-02 9.17516053e-01 3.64164501e-01 3.13456267e-01
-7.89782286e-01 1.10394609e+00 6.93569613e+00 9.71568167e-01
-1.24342513e+00 2.34678313e-01 1.25103310e-01 -8.19536820e-02
-4.64487702e-01 1.52906582e-01 -6.11755908e-01 3.01455498e-01
1.50448477e+00 -1.68773159e-03 8.57934773e-01 2.96540737e-01
-9.72961262e-02 4.51500304e-02 -1.15141213e+00 1.02987862e+00
-2.72438675e-01 -1.63985980e+00 3.09476703e-01 1.74415827e-01
6.52499855e-01 6.98693097e-01 2.58849114e-01 7.89728343e-01
2.03622237e-01 -9.98091400e-01 7.36603081e-01 7.54920185e-01
1.10226178e+00 -4.52908784e-01 6.10247612e-01 2.52682537e-01
-8.29562664e-01 5.01208603e-02 -5.74816644e-01 -6.87128827e-02
2.70129859e-01 4.29249793e-01 -6.43367529e-01 7.38675892e-01
3.64029646e-01 7.01400936e-01 -1.79106280e-01 5.25457323e-01
-1.96634412e-01 7.58035839e-01 -3.72327596e-01 -2.07072198e-01
3.41704100e-01 -7.71878541e-01 3.19459319e-01 7.83001244e-01
4.40415621e-01 1.05727889e-01 -1.59285069e-01 1.84150302e+00
-2.00274348e-01 -3.46938163e-01 -5.62311649e-01 -7.26155221e-01
4.64858055e-01 1.02685440e+00 -6.42638132e-02 -3.51014853e-01
-3.70988965e-01 9.31906521e-01 7.47749507e-01 5.13168812e-01
-7.93268502e-01 -2.00344339e-01 6.01527154e-01 -1.61697567e-01
3.62964392e-01 -5.45704842e-01 2.72312462e-01 -1.71865022e+00
-2.07368493e-01 -4.85763639e-01 -2.86084801e-01 -7.23392069e-01
-1.34291124e+00 5.18935859e-01 2.19407175e-02 -7.38996029e-01
-9.93784443e-02 -7.58463204e-01 -6.90814316e-01 1.11778116e+00
-1.67988193e+00 -8.04143429e-01 1.16221189e-01 5.71842730e-01
-4.85373884e-01 -7.51160383e-02 1.05771601e+00 3.94936323e-01
-8.85407686e-01 4.30982500e-01 4.73870486e-01 2.72267967e-01
2.56238610e-01 -1.21006525e+00 6.12441957e-01 6.04006648e-01
3.34000826e-01 1.20808935e+00 4.05967146e-01 -2.00115085e-01
-1.64130914e+00 -7.56290615e-01 5.43472350e-01 -4.57420081e-01
9.58683491e-01 -5.68856657e-01 -9.06807661e-01 6.67155564e-01
2.57487353e-02 1.51686832e-01 5.73465168e-01 3.07022721e-01
-8.28309298e-01 9.60075296e-03 -7.17946708e-01 4.22690988e-01
8.86654794e-01 -1.56354415e+00 -1.47849455e-01 2.66832620e-01
7.08207548e-01 -2.28720188e-01 -8.06359529e-01 1.38214827e-01
4.10538167e-01 -1.15936339e+00 9.08821106e-01 -7.77837574e-01
2.85558790e-01 -2.59046108e-02 -3.88764828e-01 -1.22350419e+00
-4.07682419e-01 -7.13990033e-01 -1.66638136e-01 6.69342339e-01
3.85866135e-01 -8.58055592e-01 6.70512915e-01 5.02215326e-01
-3.63384008e-01 -4.53735560e-01 -1.20524538e+00 -6.34226561e-01
4.62594390e-01 -4.56282973e-01 6.31141126e-01 5.37319839e-01
9.74089503e-02 6.77649260e-01 -5.48080862e-01 1.36897534e-01
6.20086730e-01 7.81257078e-02 5.23992300e-01 -7.88768768e-01
-6.19301617e-01 -3.01679224e-01 -3.06045443e-01 -1.33126748e+00
4.19322580e-01 -1.26175201e+00 6.32427558e-02 -1.06413054e+00
6.31946802e-01 -1.88426033e-01 -6.44574702e-01 4.55364101e-02
-1.02323771e-01 9.65436623e-02 2.51469970e-01 1.88759729e-01
-9.01780367e-01 1.26640832e+00 1.24637091e+00 -1.81833819e-01
3.55935365e-01 -2.80019164e-01 -5.21549940e-01 1.95641086e-01
4.93443280e-01 -6.24565661e-01 -6.44552186e-02 -3.81420135e-01
5.46574891e-01 3.61758620e-01 6.76979363e-01 -1.23420238e+00
2.96742380e-01 1.20590873e-01 -2.56452560e-02 -3.59709829e-01
7.63132036e-01 -1.00867540e-01 -1.63983792e-01 2.73582160e-01
-3.24384481e-01 -4.41060513e-01 1.01203464e-01 7.15234995e-01
-5.75841032e-02 -2.91189045e-01 7.78904617e-01 -2.44943932e-01
-3.09777021e-01 6.40743852e-01 -2.02759668e-01 3.79624367e-02
1.40504375e-01 3.04374188e-01 -3.10119182e-01 -4.83660847e-01
-4.76585954e-01 -1.82246402e-01 1.06713437e-01 -3.13554972e-01
3.23242933e-01 -1.28011131e+00 -2.82581836e-01 2.95477599e-01
8.10501203e-02 -3.19998354e-01 4.54746813e-01 1.14230776e+00
-3.69871289e-01 8.38644087e-01 -3.86135988e-02 -4.13344771e-01
-1.56075850e-01 5.20726919e-01 7.14719713e-01 -3.09417248e-01
-3.97237688e-01 6.51876509e-01 1.86659232e-01 -8.13008606e-01
-3.17942411e-01 -5.86180449e-01 4.11419600e-01 -5.57826042e-01
3.50383759e-01 1.07133470e-01 1.16537474e-01 -4.94834572e-01
-6.24128841e-02 2.41916060e-01 -1.30108818e-01 -2.38362715e-01
1.04224265e+00 7.52733722e-02 -1.81300089e-01 4.89903569e-01
1.59615242e+00 -1.45521775e-01 -1.25864720e+00 -6.31282449e-01
-2.09842324e-01 8.42217915e-03 3.08075994e-01 -5.24499357e-01
-8.70387554e-01 1.24412811e+00 3.48250955e-01 2.56287992e-01
6.96655035e-01 1.31587014e-01 1.10194528e+00 9.40107048e-01
6.11904442e-01 -7.68932343e-01 1.09313525e-01 8.49211216e-01
1.94569960e-01 -1.36425495e+00 -2.52960950e-01 -1.99840851e-02
-1.90967679e-01 9.46763098e-01 8.78150985e-02 -1.91990495e-01
4.71269101e-01 -2.60411650e-02 -2.75738031e-01 -4.51842219e-01
-6.31700695e-01 -2.91968584e-01 3.49230647e-01 3.91685456e-01
2.27939650e-01 4.78519686e-02 4.27717686e-01 2.40677297e-01
-2.58073986e-01 -2.08286569e-01 5.20034969e-01 5.16855597e-01
-3.22829217e-01 -9.45165396e-01 1.54521227e-01 3.88780147e-01
-2.50330478e-01 -5.75601399e-01 -1.12125099e-01 4.78529781e-01
-2.80409485e-01 6.93539441e-01 -2.88248673e-04 -5.47995031e-01
1.38344854e-01 1.89318836e-01 9.94372368e-01 -6.86045885e-01
-1.33174881e-01 -2.14659408e-01 -1.52208343e-01 -6.98246896e-01
-3.94319743e-01 -3.30465496e-01 -9.96102989e-01 -6.92504227e-01
-6.87595129e-01 3.86695474e-01 4.65611607e-01 1.40766013e+00
4.72784430e-01 4.57966417e-01 4.09807205e-01 -7.64081776e-01
-1.16890299e+00 -1.13477230e+00 -7.85261333e-01 2.73117781e-01
4.35658664e-01 -6.45992100e-01 -2.70216554e-01 -7.56700337e-01]
|
[5.591667652130127, 4.965409278869629]
|
e1931e39-69fe-4604-bb36-474bb5deb6a7
|
self-supervised-learning-and-graph
|
2306.08469
| null |
https://arxiv.org/abs/2306.08469v1
|
https://arxiv.org/pdf/2306.08469v1.pdf
|
Self-supervised Learning and Graph Classification under Heterophily
|
Self-supervised learning has shown its promising capability in graph representation learning in recent work. Most existing pre-training strategies usually choose the popular Graph neural networks (GNNs), which can be seen as a special form of low-pass filter, fail to effectively capture heterophily. In this paper, we first present an experimental investigation exploring the performance of low-pass and high-pass filters in heterophily graph classification, where the results clearly show that high-frequency signal is important for learning heterophily graph representation. On the other hand, it is still unclear how to effectively capture the structural pattern of graphs and how to measure the capability of the self-supervised pre-training strategy in capturing graph structure. To address the problem, we first design a quantitative metric to Measure Graph Structure (MGS), which analyzes correlation between structural similarity and embedding similarity of graph pairs. Then, to enhance the graph structural information captured by self-supervised learning, we propose a novel self-supervised strategy for Pre-training GNNs based on the Metric (PGM). Extensive experiments validate our pre-training strategy achieves state-of-the-art performance for molecular property prediction and protein function prediction. In addition, we find choosing the suitable filter sometimes may be better than designing good pre-training strategies for heterophily graph classification.
|
['Hao Hao', 'Zhen Liu', 'Yilin Ding']
|
2023-06-14
| null | null | null | null |
['graph-classification', 'property-prediction', 'protein-function-prediction', 'classification-1', 'graph-representation-learning', 'molecular-property-prediction']
|
['graphs', 'medical', 'medical', 'methodology', 'methodology', 'miscellaneous']
|
[ 1.44003555e-01 3.78170982e-02 -3.19645166e-01 -7.07716346e-02
2.01630265e-01 -2.49271020e-01 4.40453082e-01 5.29508770e-01
-2.54037920e-02 6.58243835e-01 1.48154691e-01 -2.80419111e-01
-4.51203585e-01 -1.22783589e+00 -6.17385983e-01 -8.92302096e-01
-3.38088751e-01 3.15531611e-01 1.44120902e-01 -4.47730333e-01
2.14973241e-02 5.29661417e-01 -1.28008270e+00 1.93280280e-01
1.12235558e+00 7.32791841e-01 1.71246111e-01 4.23132896e-01
-2.66026974e-01 7.44065583e-01 -4.01704013e-01 -9.38076004e-02
5.38238287e-02 -8.02906513e-01 -7.48124540e-01 -1.11215815e-01
2.34559059e-01 4.16534424e-01 -5.23917735e-01 1.21011782e+00
5.89345992e-01 -3.20823900e-02 7.63376117e-01 -9.87501919e-01
-4.97054726e-01 6.85441971e-01 -2.11341754e-01 5.73119998e-01
4.55940694e-01 8.27048495e-02 1.17820406e+00 -4.82339352e-01
9.23673272e-01 1.08864689e+00 7.16117740e-01 2.32609287e-01
-1.16333795e+00 -6.38309538e-01 -4.61085998e-02 2.53893197e-01
-1.26213026e+00 -9.97435115e-03 1.14348125e+00 -4.04956222e-01
9.17360425e-01 1.95761159e-01 9.22897696e-01 9.46065068e-01
5.79742908e-01 4.85656172e-01 1.15858698e+00 -2.66988218e-01
-1.54501526e-02 -1.13252737e-01 3.62713516e-01 1.13220966e+00
4.10156161e-01 2.22539619e-01 -4.40574884e-01 -3.20631444e-01
8.18329513e-01 2.17436597e-01 -5.55254579e-01 -4.67725515e-01
-9.45185900e-01 8.63145053e-01 8.56628716e-01 7.67072678e-01
-2.99925953e-01 -2.83028752e-01 4.10085887e-01 7.91667640e-01
4.14958149e-01 7.43542433e-01 -6.92064241e-02 2.75492162e-01
-5.58888376e-01 -4.16219234e-01 8.27636719e-01 3.78376871e-01
8.65068436e-01 1.24760732e-01 8.59239474e-02 6.20064795e-01
1.09590851e-01 1.48922637e-01 8.05944324e-01 6.46004379e-02
7.46405572e-02 1.22710550e+00 -6.47375941e-01 -1.37743855e+00
-6.51392400e-01 -7.30931938e-01 -1.18560719e+00 -2.68215299e-01
3.43509108e-01 1.12111948e-01 -7.00899005e-01 1.33437777e+00
1.92016184e-01 5.72461069e-01 1.18237816e-01 7.79666662e-01
1.30390155e+00 6.83907628e-01 8.84907097e-02 -5.65812230e-01
1.03325224e+00 -6.85431719e-01 -6.43693686e-01 1.27807841e-01
8.00006330e-01 -5.37864268e-01 1.01081479e+00 8.21785256e-02
-5.67896843e-01 -6.30018532e-01 -1.25666261e+00 4.02577817e-01
-7.38833070e-01 -5.12769409e-02 9.24424171e-01 5.71562290e-01
-7.12905884e-01 1.03635013e+00 -5.86468518e-01 -5.72299957e-01
1.79533005e-01 4.97226864e-01 -5.61783075e-01 2.08565183e-02
-1.50731015e+00 5.95601976e-01 4.84426945e-01 -2.25975886e-01
-5.10961235e-01 -5.24774730e-01 -6.63759947e-01 2.47385323e-01
2.63497710e-01 -6.30501986e-01 1.50987774e-01 -9.78445768e-01
-1.33366299e+00 6.97368801e-01 2.07847178e-01 -6.51286125e-01
-4.72267568e-02 4.04802114e-01 -5.76569498e-01 5.68318844e-01
-2.60239929e-01 1.47185966e-01 7.25237072e-01 -1.02165830e+00
-1.22650512e-01 -2.69764870e-01 -2.06411093e-01 1.48345664e-01
-7.48492420e-01 -4.00652617e-01 -1.37997023e-03 -6.67491913e-01
3.79232764e-01 -7.06854582e-01 5.45580350e-02 -4.18460697e-01
-2.35315278e-01 -3.31883758e-01 8.53127301e-01 -3.32456976e-01
1.47905612e+00 -2.04691792e+00 -9.42983665e-03 7.22086787e-01
6.58084571e-01 6.19883478e-01 -2.50246048e-01 7.54722238e-01
-3.99288505e-01 5.77060021e-02 -8.20901021e-02 4.35931444e-01
-4.10284132e-01 1.41845167e-01 -8.17480609e-02 5.42428136e-01
1.57115147e-01 9.51369703e-01 -1.09056985e+00 -4.66549605e-01
2.36097127e-01 5.86782932e-01 -2.09685594e-01 3.02473158e-01
-2.80880500e-02 3.16944003e-01 -5.46204984e-01 6.14105403e-01
5.46122253e-01 -7.81494677e-01 6.34165645e-01 -5.28931439e-01
3.65697533e-01 1.31285340e-01 -8.81079793e-01 1.07878590e+00
-9.88240689e-02 5.52620769e-01 -3.31582814e-01 -1.50887454e+00
1.36866879e+00 9.69834775e-02 7.42137969e-01 -6.78332150e-01
2.89228529e-01 7.01482594e-02 3.04067105e-01 -3.73892099e-01
-1.71588242e-01 -6.49663359e-02 4.81049836e-01 2.10161850e-01
3.75177383e-01 4.12892312e-01 2.46398389e-01 2.58880854e-01
1.25813496e+00 -5.16310632e-01 5.34045160e-01 -4.10168767e-01
8.33928883e-01 -1.96681604e-01 2.83476532e-01 5.78603506e-01
-1.16863042e-01 2.47299597e-01 7.04889178e-01 -5.45330226e-01
-5.60313821e-01 -9.09499466e-01 -1.00071765e-02 1.01813889e+00
3.40401709e-01 -7.30540693e-01 -4.83424217e-01 -8.36536884e-01
9.65496227e-02 -1.17531784e-01 -5.36909819e-01 -6.34672165e-01
-4.31628734e-01 -1.10801888e+00 3.67612362e-01 2.72476554e-01
6.33529782e-01 -1.00084949e+00 6.24148957e-02 1.85918793e-01
1.23985067e-01 -7.51099586e-01 -2.99901366e-01 4.21666265e-01
-1.03324842e+00 -1.55898285e+00 -4.17584509e-01 -1.12548482e+00
8.81468713e-01 5.18284023e-01 1.13700223e+00 6.63631856e-01
-2.18816206e-01 1.69171512e-01 -4.91381824e-01 1.28113866e-01
-4.32027012e-01 1.76653594e-01 -1.00219384e-01 1.58915564e-01
4.42012697e-01 -9.28803027e-01 -6.15255475e-01 3.60025644e-01
-7.79965639e-01 -2.27354765e-01 7.69568622e-01 1.08343613e+00
6.16601646e-01 5.16912758e-01 4.83733624e-01 -1.28089488e+00
9.97240961e-01 -4.23572123e-01 -2.41318628e-01 4.36263084e-01
-1.02149522e+00 7.40817785e-02 1.16137540e+00 -4.78413224e-01
-6.20522857e-01 -4.69488800e-02 -5.96606843e-02 -4.00868595e-01
8.97784308e-02 7.84492493e-01 -9.57900211e-02 -7.54017115e-01
9.62738514e-01 4.98372912e-01 3.44611049e-01 -2.07753450e-01
5.89529946e-02 4.38474357e-01 1.47120014e-01 -3.19218129e-01
7.02102125e-01 4.53283310e-01 3.40164363e-01 -1.12678325e+00
-4.29224998e-01 -8.01951766e-01 -4.18074906e-01 -2.12358370e-01
6.20802820e-01 -7.06082284e-01 -8.68395805e-01 1.75948679e-01
-4.71835822e-01 -1.33473292e-01 7.30734766e-02 4.10172313e-01
-3.87856275e-01 7.68282473e-01 -7.09567249e-01 -4.35079157e-01
-5.86322665e-01 -7.01796353e-01 5.77443242e-01 1.88863546e-01
1.09148264e-01 -1.43223786e+00 2.10123271e-01 1.64105982e-01
3.77817959e-01 4.33422327e-01 1.08831310e+00 -8.71692657e-01
-3.41105133e-01 -1.90296143e-01 -2.31655240e-01 1.58349186e-01
1.89316019e-01 -1.30892009e-01 -5.24499178e-01 -4.72139657e-01
-9.20201614e-02 -1.39610589e-01 1.12064993e+00 3.60995144e-01
8.66505682e-01 -2.70928681e-01 -5.02723336e-01 8.23468029e-01
1.58682728e+00 1.08155653e-01 7.19973445e-01 2.51165256e-02
8.77909482e-01 4.48923618e-01 3.66969705e-01 4.76826988e-02
-5.66736236e-02 2.19657242e-01 1.41241312e-01 -3.95121604e-01
-1.42595425e-01 -5.83563626e-01 1.83525831e-01 1.10219657e+00
-2.50792384e-01 -2.82080412e-01 -7.88686752e-01 -1.25952542e-01
-1.92848265e+00 -9.12793875e-01 -8.64797756e-02 1.94474733e+00
5.69643617e-01 2.52410978e-01 2.98224807e-01 3.26722652e-01
7.51966834e-01 2.96481818e-01 -3.04954618e-01 -3.90468463e-02
-5.14605403e-01 2.83465981e-01 4.50108618e-01 3.24867189e-01
-1.01421189e+00 9.27255929e-01 5.90065384e+00 9.93192554e-01
-1.44024992e+00 -2.95564383e-01 4.58896726e-01 5.69145918e-01
-2.70107329e-01 2.64412118e-03 -4.20970112e-01 5.06054103e-01
8.37222993e-01 -1.28564999e-01 3.95921350e-01 8.28023612e-01
-3.29849101e-03 2.64084756e-01 -8.47984731e-01 1.07332814e+00
1.11781649e-01 -1.58039105e+00 2.63355523e-01 2.70207096e-02
5.82837701e-01 -1.53239861e-01 -1.87923655e-01 2.12587491e-01
2.41164211e-02 -1.00119066e+00 -4.24411476e-01 5.08676648e-01
3.45810831e-01 -6.27874613e-01 9.33752477e-01 1.77148551e-01
-1.51203370e+00 -7.69921094e-02 -6.59269571e-01 -1.42579386e-02
-3.12102109e-01 6.99482083e-01 -9.02777612e-01 6.83062732e-01
2.89821655e-01 1.09399557e+00 -8.08071017e-01 1.07173789e+00
-1.33430183e-01 7.70421326e-01 -7.39491731e-02 -4.68150496e-01
1.92657784e-01 -4.57221031e-01 5.09112418e-01 1.26195979e+00
-1.27783686e-01 1.41404107e-01 3.47447574e-01 5.58335841e-01
-3.24320011e-02 5.96731901e-01 -7.57824481e-01 -6.47501826e-01
1.79145217e-01 1.17713606e+00 -1.19390261e+00 -4.03606206e-01
-2.61229664e-01 6.24966502e-01 5.20416141e-01 3.32621604e-01
-4.30026650e-01 -5.18252611e-01 2.57497221e-01 3.96586210e-01
1.00093402e-01 -1.26452267e-01 -4.92401719e-02 -1.20381021e+00
-3.85414749e-01 -9.18042004e-01 7.55320013e-01 -4.25549924e-01
-1.63225579e+00 6.55853212e-01 -3.46864194e-01 -1.18675542e+00
1.92351416e-01 -7.72052348e-01 -8.13597858e-01 4.11000520e-01
-1.41815960e+00 -1.01533771e+00 -4.50886458e-01 7.51574039e-01
-6.26042113e-02 -4.57797825e-01 6.83993459e-01 2.38627970e-01
-5.02321720e-01 5.46740949e-01 1.55418813e-01 1.93116859e-01
5.59426546e-01 -1.17331612e+00 -7.06681311e-02 6.58816636e-01
3.15000474e-01 9.91087079e-01 6.15665436e-01 -9.13345814e-01
-1.55760777e+00 -9.15215969e-01 7.52836108e-01 5.65920174e-02
7.51009166e-01 -3.09023052e-01 -1.19178450e+00 1.07714579e-01
-7.72125274e-02 1.04665302e-01 7.74540961e-01 2.17475861e-01
-3.88980418e-01 -2.70985514e-01 -1.00803542e+00 4.05573130e-01
1.29886043e+00 -6.95611596e-01 -4.92682338e-01 6.31645381e-01
3.97386611e-01 4.70935404e-02 -1.04948211e+00 7.09483206e-01
3.70657414e-01 -1.17702055e+00 1.00720620e+00 -6.98338449e-01
2.47227192e-01 -2.30349272e-01 3.08465630e-01 -1.33966494e+00
-6.19809747e-01 -5.97552240e-01 -1.73578635e-01 7.43114889e-01
2.29356200e-01 -9.91646647e-01 1.15857863e+00 -2.87202567e-01
1.03849985e-01 -1.01567960e+00 -5.26040077e-01 -8.52696836e-01
-2.58439779e-01 4.03000563e-01 3.26679826e-01 1.31489182e+00
4.13049787e-01 7.10502565e-01 -2.22144395e-01 -1.01394370e-01
4.53800231e-01 3.60004842e-01 6.64802134e-01 -1.40180779e+00
-2.94646561e-01 -5.55423439e-01 -9.07064259e-01 -7.92525649e-01
3.64351898e-01 -1.17669928e+00 -4.61017877e-01 -1.44431746e+00
1.71715513e-01 -1.21336028e-01 -5.14365673e-01 1.54364005e-01
-2.73285925e-01 7.53011778e-02 -2.00694904e-01 2.58052021e-01
-5.20981491e-01 4.34507132e-01 1.50091279e+00 -2.66601831e-01
-4.91504699e-01 -1.28555372e-01 -5.79627454e-01 4.14751649e-01
9.10606802e-01 -1.52702108e-01 -6.98083639e-01 4.92729306e-01
2.34740913e-01 3.21294460e-03 9.55116376e-02 -1.13791740e+00
2.13006511e-01 -3.72783169e-02 4.09445733e-01 -2.67350942e-01
-6.11920394e-02 -6.80870533e-01 2.44942054e-01 8.46407652e-01
-1.25255689e-01 -6.51692301e-02 -2.05686867e-01 8.68783832e-01
-4.92319018e-01 -2.47113798e-02 8.80888879e-01 -2.50314832e-01
-6.15711570e-01 4.81343478e-01 -3.03375602e-01 -9.11741480e-02
6.89945936e-01 -4.15485799e-01 -4.76249307e-01 -3.68105114e-01
-7.22720444e-01 8.80831406e-02 3.35047960e-01 6.35745302e-02
7.22202063e-01 -1.19387197e+00 -2.97061920e-01 3.96746993e-01
1.53930098e-01 -7.19129026e-01 1.72500044e-01 8.39928389e-01
-6.75830245e-01 3.54457378e-01 -5.16135871e-01 -5.31564891e-01
-1.45912921e+00 7.14883983e-01 3.83823514e-01 -4.88352120e-01
-5.94356179e-01 7.22112477e-01 1.49443835e-01 -1.60791531e-01
4.46692146e-02 -7.14380443e-02 -6.35661304e-01 9.59873348e-02
4.83686142e-02 2.56078660e-01 6.38028607e-02 -5.89402914e-01
-2.94879973e-01 5.83231091e-01 1.13430068e-01 8.20663571e-01
1.28435016e+00 2.05201015e-01 -3.71160656e-01 1.70970917e-01
1.45095897e+00 -1.40859470e-01 -6.66002929e-01 -1.49373919e-01
-7.30331391e-02 -1.75414428e-01 -4.60146144e-02 -3.62874001e-01
-1.13056731e+00 7.19631195e-01 6.16274536e-01 4.90430474e-01
1.03057480e+00 -1.14955574e-01 7.96376348e-01 6.60838068e-01
1.74396351e-01 -9.94924903e-01 5.31391740e-01 4.71734732e-01
6.09197974e-01 -1.10592604e+00 1.71565682e-01 -9.08883512e-01
-2.54060656e-01 1.34437037e+00 7.25883663e-01 -4.91196424e-01
9.36355591e-01 -1.08080007e-01 -1.04636833e-01 -7.36538053e-01
-4.74386483e-01 -4.31770772e-01 5.48333585e-01 5.24084270e-01
5.86419225e-01 9.69566479e-02 -7.04135239e-01 2.27561042e-01
-2.64063179e-01 -3.39904338e-01 1.94970131e-01 7.02553570e-01
-7.37473547e-01 -9.88172531e-01 2.10671946e-02 8.39255214e-01
-1.39916122e-01 -1.09025411e-01 -8.74850631e-01 6.25891209e-01
-2.39767041e-02 6.87827647e-01 -1.51722640e-01 -8.46070528e-01
5.87191470e-02 1.01069193e-02 4.55917686e-01 -5.01150250e-01
-6.80611074e-01 9.48849693e-02 8.68258178e-02 -4.56513405e-01
-6.07050240e-01 1.42614454e-01 -1.13596690e+00 -3.27331781e-01
-5.13267100e-01 5.37323117e-01 4.64331023e-02 6.96156383e-01
3.63060385e-01 4.52134311e-01 7.11054444e-01 -2.87049413e-01
-2.02853650e-01 -8.97323847e-01 -8.07295740e-01 7.86702454e-01
7.13602528e-02 -8.14584851e-01 -6.91434383e-01 -2.87928671e-01]
|
[7.025129795074463, 6.196854591369629]
|
080df7ae-4d86-4e62-b470-ddacc2ad0b21
|
aunet-attention-guided-dense-upsampling
|
1810.10151
| null |
https://arxiv.org/abs/1810.10151v3
|
https://arxiv.org/pdf/1810.10151v3.pdf
|
AUNet: Attention-guided dense-upsampling networks for breast mass segmentation in whole mammograms
|
Mammography is one of the most commonly applied tools for early breast cancer screening. Automatic segmentation of breast masses in mammograms is essential but challenging due to the low signal-to-noise ratio and the wide variety of mass shapes and sizes. Existing methods deal with these challenges mainly by extracting mass-centered image patches manually or automatically. However, manual patch extraction is time-consuming and automatic patch extraction brings errors that could not be compensated in the following segmentation step. In this study, we propose a novel attention-guided dense-upsampling network (AUNet) for accurate breast mass segmentation in whole mammograms directly. In AUNet, we employ an asymmetrical encoder-decoder structure and propose an effective upsampling block, attention-guided dense-upsampling block (AU block). Especially, the AU block is designed to have three merits. Firstly, it compensates the information loss of bilinear upsampling by dense upsampling. Secondly, it designs a more effective method to fuse high- and low-level features. Thirdly, it includes a channel-attention function to highlight rich-information channels. We evaluated the proposed method on two publicly available datasets, CBIS-DDSM and INbreast. Compared to three state-of-the-art fully convolutional networks, AUNet achieved the best performances with an average Dice similarity coefficient of 81.8% for CBIS-DDSM and 79.1% for INbreast.
|
['Shan-Shan Wang', 'David Dagan Feng', 'Hui Sun', 'Boqiang Liu', 'Hairong Zheng', 'Cheng Li']
|
2018-10-24
| null | null | null | null |
['breast-mass-segmentation-in-whole-mammograms']
|
['medical']
|
[ 5.28813481e-01 1.72136232e-01 -2.90457636e-01 -4.55553472e-01
-9.43484068e-01 3.03020775e-01 2.17248932e-01 3.60450029e-01
-5.36849141e-01 5.14227271e-01 1.31345898e-01 -4.17519778e-01
3.51907574e-02 -8.23014677e-01 -6.95580602e-01 -8.25898826e-01
3.88853438e-02 2.72089660e-01 4.48451519e-01 7.69075826e-02
-9.26710945e-03 3.66159648e-01 -1.18791378e+00 5.92072487e-01
9.64078784e-01 1.30360317e+00 4.86600459e-01 5.35089791e-01
-2.13265136e-01 6.69263184e-01 -1.15203999e-01 -6.78778112e-01
1.13627426e-01 -7.16903090e-01 -6.59452796e-01 8.34006891e-02
1.47748128e-01 -5.79814196e-01 -3.97409856e-01 1.24276507e+00
7.29756713e-01 -5.54246128e-01 6.78748310e-01 -6.79796398e-01
-4.17230427e-01 8.22856128e-01 -1.17735553e+00 4.74283934e-01
-4.06216890e-01 -1.61579922e-01 5.73287010e-01 -6.92947447e-01
3.55738044e-01 1.05284154e+00 8.63745034e-01 4.74405259e-01
-1.04618192e+00 -7.61784673e-01 -2.37715930e-01 2.27906227e-01
-1.30725908e+00 -2.82678425e-01 5.35640240e-01 -2.92159706e-01
4.35682774e-01 3.75093549e-01 8.47056329e-01 5.96723974e-01
5.45811832e-01 1.06648457e+00 9.54803169e-01 -3.08013201e-01
7.66703486e-02 -1.03148796e-01 -4.67997007e-02 7.23757446e-01
5.30151486e-01 -7.19102612e-03 2.38306969e-02 -8.73883516e-02
9.54758823e-01 2.51289934e-01 -1.60688981e-01 -1.67543769e-01
-1.17525601e+00 8.58821332e-01 9.81076121e-01 5.66414058e-01
-5.56059837e-01 6.02437444e-02 5.17167449e-01 -2.46218458e-01
5.40553212e-01 1.18191600e-01 4.60513961e-03 2.59674013e-01
-1.23159921e+00 -4.57597375e-02 3.79206002e-01 6.47865117e-01
4.63555247e-01 -3.01260442e-01 -4.63100970e-01 8.42752218e-01
3.46982449e-01 5.28589606e-01 7.14661837e-01 -2.78811872e-01
3.00400347e-01 6.78816319e-01 -3.21149617e-01 -1.13504076e+00
-7.03025997e-01 -7.14802146e-01 -1.44217157e+00 -3.01835895e-01
2.19365686e-01 -2.13189628e-02 -1.31045783e+00 1.27377307e+00
4.26004946e-01 -1.66176990e-01 -8.82382914e-02 1.03613937e+00
1.20768452e+00 4.44419950e-01 3.56426001e-01 -3.23122323e-01
1.73432624e+00 -6.83876336e-01 -8.97083938e-01 -3.18416238e-01
6.12183034e-01 -5.33167541e-01 5.58218539e-01 6.82929624e-03
-1.08594382e+00 -5.12504876e-01 -1.16182041e+00 -5.28305843e-02
1.33722216e-01 5.61521828e-01 6.24041259e-01 6.04081750e-01
-7.19829500e-01 4.35639054e-01 -1.27132428e+00 -1.47043258e-01
1.09331369e+00 4.35896933e-01 -1.50541469e-01 -1.42713428e-01
-1.00726545e+00 5.56252241e-01 3.68922949e-01 1.93173364e-01
-6.44205570e-01 -7.33441055e-01 -8.46672595e-01 1.12598114e-01
2.38979772e-01 -4.19669300e-01 1.42703760e+00 -1.05230987e+00
-9.74209785e-01 8.65711749e-01 -6.01614527e-02 -6.61330581e-01
5.86292803e-01 1.13946080e-01 -2.80928344e-01 5.58326960e-01
2.20236644e-01 8.24874401e-01 5.69115102e-01 -8.71150196e-01
-6.33598268e-01 -5.53502679e-01 -5.69997907e-01 1.32792652e-01
-4.55904067e-01 -5.24055436e-02 -6.69087529e-01 -7.76864946e-01
6.44307435e-01 -6.05482399e-01 -4.56620067e-01 3.19870800e-01
-3.29933256e-01 3.73142332e-01 6.73127353e-01 -9.86361086e-01
1.35994971e+00 -2.21295524e+00 -2.76941597e-01 7.02943057e-02
4.31011349e-01 4.94975120e-01 1.45756632e-01 -1.87750027e-01
1.09756244e-02 6.60937876e-02 -3.20242643e-01 -3.06674931e-02
-4.17277515e-01 -8.17788169e-02 4.68789071e-01 5.12039900e-01
4.38260198e-01 1.15994120e+00 -8.53714168e-01 -9.22446370e-01
1.36954203e-01 5.92378795e-01 -4.77762401e-01 -1.18681267e-01
9.75450594e-03 2.32593819e-01 -4.72079366e-01 8.44421148e-01
8.75084698e-01 -5.14561296e-01 1.16893910e-01 -6.30566061e-01
1.20339826e-01 -8.64716396e-02 -8.09774935e-01 1.42657340e+00
-2.93970853e-01 5.93069971e-01 1.98988423e-01 -9.21487451e-01
8.40275764e-01 2.21951634e-01 6.89952612e-01 -9.25134659e-01
4.92619932e-01 6.21942878e-01 4.18619812e-01 -5.96729517e-01
2.47898191e-01 -3.21794361e-01 2.48026893e-01 4.71970811e-02
-1.03849456e-01 2.69654784e-02 1.71536356e-01 1.42332152e-01
1.08471692e+00 -4.45823073e-01 5.79730809e-01 -3.91756922e-01
4.27413225e-01 -2.18020484e-01 6.22326970e-01 4.79401052e-01
-2.85836071e-01 8.10861528e-01 4.99656647e-01 -3.01941007e-01
-9.43423569e-01 -6.06332183e-01 -4.96293426e-01 5.14322042e-01
2.03255147e-01 -2.09389225e-01 -6.42552614e-01 -7.08542764e-01
-2.25778129e-02 2.67768294e-01 -7.66556442e-01 -1.06146142e-01
-5.21209896e-01 -1.11341393e+00 4.27912980e-01 9.15133655e-01
1.16808832e+00 -7.74922371e-01 -7.92424381e-01 3.23332936e-01
-4.25700217e-01 -7.95481741e-01 -4.53581810e-01 1.57902226e-01
-9.54206109e-01 -1.05756080e+00 -1.36609459e+00 -7.53464878e-01
8.94353628e-01 2.21510991e-01 1.00237763e+00 1.12208247e-01
-7.64080167e-01 -4.65246379e-01 -3.69360596e-01 -5.61783850e-01
-4.64198560e-01 9.29511935e-02 -7.96336532e-01 7.12253600e-02
4.86773700e-01 -1.18986294e-01 -1.01153672e+00 3.89816731e-01
-1.00951850e+00 4.32076216e-01 1.27309597e+00 1.24673426e+00
8.49112570e-01 5.74345514e-02 5.49877465e-01 -9.85626936e-01
1.22912318e-01 -4.74422246e-01 -3.38047236e-01 -3.12120151e-02
-2.34036773e-01 -3.13960552e-01 2.35026851e-01 -2.10650712e-01
-9.75189745e-01 1.78673387e-01 -5.36813080e-01 -7.83455148e-02
8.63083452e-02 5.19358099e-01 -1.04819432e-01 -2.25772634e-01
5.70532024e-01 1.67997658e-01 1.18069239e-01 -3.83710027e-01
-1.67115048e-01 9.40025449e-01 4.71735448e-01 1.34977579e-01
1.56061113e-01 4.35028523e-01 8.47543627e-02 -6.92454278e-01
-7.04514980e-01 -4.56737697e-01 -4.07148957e-01 -2.53005028e-02
1.02945495e+00 -8.99844766e-01 -3.33597451e-01 5.79880953e-01
-7.85437644e-01 -7.11840913e-02 -2.21359804e-01 6.42993629e-01
-1.43891856e-01 3.24702412e-01 -8.64414394e-01 -6.06863081e-01
-6.94244087e-01 -1.35312510e+00 1.09895873e+00 3.43070090e-01
-1.41766258e-02 -5.96669972e-01 -6.34587586e-01 2.71461844e-01
6.82678342e-01 3.48901927e-01 8.48371565e-01 -5.70972264e-01
-1.80065766e-01 -3.07383358e-01 -9.23804462e-01 2.63646364e-01
3.93414497e-01 -3.36797684e-01 -7.10919738e-01 -1.92970634e-01
-7.11045563e-02 -1.68239996e-01 1.25023592e+00 9.77221012e-01
1.48861384e+00 -1.82068169e-01 -6.87949300e-01 7.51331031e-01
1.36622810e+00 3.70449424e-01 7.78263628e-01 1.79770336e-01
5.64572215e-01 1.62150353e-01 6.82659745e-01 3.36421072e-01
9.11777690e-02 3.35539937e-01 5.42249024e-01 -6.21984363e-01
-3.43789965e-01 -7.52297789e-02 -3.22430849e-01 5.27425826e-01
6.67312145e-02 -4.32005376e-02 -9.62424099e-01 7.35580504e-01
-1.52660823e+00 -6.64596319e-01 -2.43550465e-01 1.90178287e+00
1.06931973e+00 9.33461189e-02 -6.57291114e-02 2.55263418e-01
6.47117734e-01 -5.35524525e-02 -5.40570855e-01 1.44864172e-02
1.32530585e-01 3.16684186e-01 7.61808515e-01 1.19333930e-01
-1.41891897e+00 3.25078815e-01 5.12722349e+00 1.13035643e+00
-1.21056557e+00 2.45721728e-01 1.18673277e+00 5.23361266e-02
1.07342319e-03 -4.28500444e-01 -8.51567626e-01 6.44083083e-01
6.54251456e-01 1.64143384e-01 -5.01641154e-01 6.57705486e-01
-8.36793929e-02 -4.70974028e-01 -8.87300193e-01 8.77704144e-01
1.60813574e-02 -1.40494680e+00 -1.18969224e-01 1.07002221e-02
7.06564426e-01 -1.83339432e-01 -1.05394162e-01 -2.80444021e-03
-2.87247330e-01 -9.20517325e-01 4.28282142e-01 2.80128896e-01
1.13487852e+00 -7.89935648e-01 1.26877367e+00 2.20585749e-01
-1.13498199e+00 -5.62798120e-02 -2.78094202e-01 4.32958394e-01
1.67906471e-02 1.01557350e+00 -8.61725092e-01 5.88350773e-01
6.87610447e-01 5.53488553e-01 -6.81692600e-01 1.50343466e+00
2.55241036e-01 6.16754353e-01 -2.55309999e-01 -1.16727903e-01
2.28536487e-01 2.96770275e-01 1.39336914e-01 1.45865679e+00
5.56360483e-01 1.76266611e-01 5.74839227e-02 7.26824164e-01
-1.27069443e-01 2.26657137e-01 -5.63960560e-02 -1.18757613e-01
1.76983401e-01 1.25753963e+00 -1.08134556e+00 -4.26103413e-01
-3.53735864e-01 7.35945523e-01 -3.18682194e-01 -2.21854165e-01
-8.36889327e-01 -5.71771622e-01 -1.48481742e-01 4.05769676e-01
4.47803229e-01 3.62740099e-01 -1.81152120e-01 -6.71602964e-01
-1.11416295e-01 -9.12111938e-01 2.47499138e-01 -3.72467846e-01
-1.11246622e+00 6.72137737e-01 -3.61337811e-02 -1.14161646e+00
2.34879643e-01 -3.81229281e-01 -3.07567000e-01 7.18714416e-01
-1.54663229e+00 -1.16539013e+00 -7.22980499e-01 1.98168144e-01
5.02911627e-01 7.14677200e-02 6.63341582e-01 7.09478199e-01
-4.20387477e-01 7.21150935e-01 3.19297649e-02 3.19644928e-01
5.34256697e-01 -1.06566691e+00 1.75771788e-01 6.02932334e-01
-4.26212817e-01 4.06316854e-02 1.86349154e-01 -7.78930545e-01
-1.03770912e+00 -1.34410524e+00 6.62275851e-01 5.78424156e-01
2.46498823e-01 -1.08792186e-01 -9.39305484e-01 3.37924421e-01
-3.38680744e-02 3.73685896e-01 5.33905089e-01 -6.07583284e-01
1.58809558e-01 -2.59452522e-01 -1.32395029e+00 2.58186758e-01
6.07710183e-01 1.72699794e-01 -1.23870827e-03 2.71908611e-01
3.66965681e-01 -7.63214648e-01 -8.28768551e-01 8.31469774e-01
5.54683566e-01 -9.83842671e-01 7.61881769e-01 1.11866646e-01
8.56350303e-01 -2.98794024e-02 4.93080309e-03 -1.02113092e+00
-5.10110199e-01 -1.10205650e-01 2.55165529e-02 9.38475728e-01
3.99076790e-01 -4.90972936e-01 9.00168657e-01 1.61240339e-01
-2.10844204e-01 -1.28715670e+00 -8.33344638e-01 -2.52741843e-01
-4.30754498e-02 -9.64230150e-02 5.21217108e-01 5.16476870e-01
-2.35566825e-01 -6.11639321e-02 -5.02667800e-02 -7.74364471e-02
5.91348588e-01 -5.71474340e-03 2.84704983e-01 -1.03277266e+00
-2.78058171e-01 -4.09994692e-01 -6.48838460e-01 -1.08465469e+00
-5.72632253e-01 -8.12596500e-01 1.29196659e-01 -1.72757196e+00
6.84148371e-01 -5.10786474e-01 -1.73534989e-01 4.38694298e-01
-4.00970966e-01 5.49722612e-01 -1.29474223e-01 -1.80688985e-02
-3.24522942e-01 3.42247844e-01 1.72641313e+00 -3.92715722e-01
-6.37170300e-02 1.62616074e-01 -8.39547873e-01 7.58396745e-01
8.43815267e-01 -3.58289659e-01 -1.45466223e-01 -3.62819165e-01
-1.25033021e-01 2.26189524e-01 5.08452773e-01 -1.25712490e+00
1.17595829e-01 1.45730346e-01 9.15695548e-01 -1.05397320e+00
1.57752350e-01 -6.80472970e-01 3.08733210e-02 8.20516646e-01
-2.59595245e-01 -5.04479229e-01 4.38777566e-01 5.45445204e-01
-3.89520556e-01 -2.72646159e-01 1.07930362e+00 -2.89283633e-01
-4.66056377e-01 3.29679400e-01 -2.91758418e-01 -2.60582328e-01
1.11367345e+00 -1.47974029e-01 -8.74510780e-02 -2.48353574e-02
-5.39763570e-01 1.60826102e-01 3.83144468e-02 -4.92120767e-03
7.44234920e-01 -1.41907310e+00 -9.30543661e-01 4.79697824e-01
3.67891565e-02 2.40279675e-01 5.47070980e-01 1.48308730e+00
-7.24959016e-01 3.02216768e-01 -1.17438950e-01 -8.41903508e-01
-1.48456109e+00 1.74188808e-01 4.86824006e-01 -5.04222214e-01
-7.57754207e-01 1.08871615e+00 3.46153826e-01 2.38294140e-01
2.61530101e-01 -6.48318708e-01 -1.73986092e-01 4.56589134e-03
7.15322971e-01 1.13485597e-01 3.39872539e-01 -6.39407754e-01
-2.39694163e-01 4.08098489e-01 -4.48683649e-01 4.16396827e-01
1.18154669e+00 1.37486935e-01 -5.82411066e-02 -8.07042271e-02
1.31276166e+00 -4.84188527e-01 -9.85471308e-01 -3.15637887e-01
-3.50503862e-01 -5.89678824e-01 5.49326479e-01 -7.19078958e-01
-1.47863030e+00 1.07478154e+00 9.81001616e-01 1.22644208e-01
1.39606869e+00 -3.38619910e-02 1.00014806e+00 -5.75308800e-02
-1.05766319e-02 -8.06811035e-01 4.49799821e-02 -8.73933956e-02
8.97596955e-01 -1.50206125e+00 3.30054253e-01 -5.91915727e-01
-5.82488060e-01 1.00591159e+00 6.23460948e-01 2.03514606e-01
7.28404939e-01 3.74337733e-01 -1.81895673e-01 -3.17080468e-01
-2.99578428e-01 -1.19037598e-01 4.38125432e-01 3.90817434e-01
7.39906609e-01 8.11298192e-02 -6.91946566e-01 8.90303969e-01
1.43738404e-01 2.77602673e-01 1.08981691e-01 1.05486870e+00
-6.05695367e-01 -6.06452167e-01 -4.63429898e-01 1.14177704e+00
-9.40549672e-01 -1.66350920e-02 -3.12677138e-02 7.87570000e-01
3.06629330e-01 6.49973452e-01 1.88154966e-01 -1.31385505e-01
9.08264220e-02 -4.40523446e-01 3.52676064e-01 -5.09317279e-01
-4.68904585e-01 6.48047388e-01 9.03648138e-03 -4.18350905e-01
-4.63999242e-01 -5.46568990e-01 -1.29978228e+00 1.45579353e-01
-7.77482212e-01 -7.01705217e-02 6.96652889e-01 6.22094512e-01
2.72770166e-01 1.06740665e+00 3.83095980e-01 -8.87264907e-01
-5.88531137e-01 -1.15121138e+00 -6.37787402e-01 3.41557741e-01
3.89193952e-01 -5.43266654e-01 8.72287247e-03 -9.59676865e-04]
|
[15.085184097290039, -2.51940655708313]
|
47e73742-8c59-4d31-be36-6c320d7c3435
|
intent-generation-for-goal-oriented-dialogue
|
1807.01292
| null |
http://arxiv.org/abs/1807.01292v1
|
http://arxiv.org/pdf/1807.01292v1.pdf
|
Intent Generation for Goal-Oriented Dialogue Systems based on Schema.org Annotations
|
Goal-oriented dialogue systems typically communicate with a backend (e.g.
database, Web API) to complete certain tasks to reach a goal. The intents that
a dialogue system can recognize are mostly included to the system by the
developer statically. For an open dialogue system that can work on more than a
small set of well curated data and APIs, this manual intent creation will not
scalable. In this paper, we introduce a straightforward methodology for intent
creation based on semantic annotation of data and services on the web. With
this method, the Natural Language Understanding (NLU) module of a goal-oriented
dialogue system can adapt to newly introduced APIs without requiring heavy
developer involvement. We were able to extract intents and necessary slots to
be filled from schema.org annotations. We were also able to create a set of
initial training sentences for classifying user utterances into the generated
intents. We demonstrate our approach on the NLU module of a state-of-the art
dialogue system development framework.
|
['Umutcan Şimşek', 'Dieter Fensel']
|
2018-07-03
| null | null | null | null |
['goal-oriented-dialogue-systems']
|
['natural-language-processing']
|
[-1.43497474e-02 1.12924051e+00 4.58054274e-01 -8.28649998e-01
-6.04098260e-01 -9.21173573e-01 8.70036364e-01 2.53140271e-01
-1.68617755e-01 7.72836685e-01 4.87405896e-01 -3.55135769e-01
4.50704731e-02 -8.28181744e-01 -1.33055598e-01 3.38234067e-01
3.21093351e-01 8.37292492e-01 6.63445890e-01 -9.69655693e-01
2.76096910e-01 -2.12005705e-01 -1.68006170e+00 5.52272975e-01
7.47408926e-01 5.62411606e-01 5.85682452e-01 8.35644543e-01
-1.02891850e+00 8.36884439e-01 -7.52721369e-01 -1.16045415e-01
2.96414718e-02 -5.67089260e-01 -1.50075436e+00 4.86643091e-02
-1.28086403e-01 -2.72918731e-01 4.36571330e-01 8.13536823e-01
3.87533933e-01 5.79262376e-02 2.55772322e-01 -1.42124927e+00
2.96843499e-02 1.02311122e+00 4.55652833e-01 -5.08288026e-01
1.12445474e+00 1.96713388e-01 9.48566556e-01 -4.92432117e-01
1.06678116e+00 1.13737583e+00 2.90122598e-01 1.15369236e+00
-9.72417533e-01 2.26689950e-02 -4.62210551e-02 -1.64082259e-01
-9.29613233e-01 -6.87048256e-01 6.79712296e-01 -4.86144185e-01
1.28104603e+00 7.59346485e-01 2.85862505e-01 6.42272651e-01
-3.74786466e-01 2.64081985e-01 7.92686403e-01 -8.97596359e-01
4.07218695e-01 7.16407716e-01 6.16379023e-01 7.65864670e-01
-3.59839588e-01 -7.03378081e-01 -4.23998564e-01 -4.61408705e-01
3.43659043e-01 -6.31539166e-01 -9.06546935e-02 -2.37986103e-01
-6.54530168e-01 7.43368626e-01 -3.40076566e-01 5.76988518e-01
-2.46731699e-01 -3.71919185e-01 7.44218349e-01 5.46474278e-01
1.74044371e-01 9.22203302e-01 -7.42450893e-01 -7.10386097e-01
-3.80153179e-01 2.67807513e-01 1.99604571e+00 1.28812325e+00
8.62648249e-01 -6.58766925e-01 -4.76389043e-02 1.06748998e+00
4.80848402e-01 -1.49379894e-01 5.02088368e-01 -9.90167379e-01
1.67847693e-01 1.33542478e+00 4.96426195e-01 -3.29907417e-01
-5.63964009e-01 2.97617733e-01 1.08079508e-01 1.97557405e-01
6.12185657e-01 -5.89071989e-01 -4.26675409e-01 1.44204080e+00
6.50612533e-01 -8.62463474e-01 5.25882959e-01 6.28326058e-01
1.12855351e+00 4.98434991e-01 8.53998363e-02 -2.83890456e-01
1.65738881e+00 -8.36387277e-01 -9.97915268e-01 -1.74642429e-01
1.03686380e+00 -8.76123130e-01 1.36942673e+00 1.63665578e-01
-1.03091586e+00 -3.74504477e-01 -8.34008098e-01 -1.17706589e-01
-7.52989650e-01 -2.18609646e-01 7.20884502e-01 7.07375407e-01
-1.10126245e+00 3.00545782e-01 -4.61158156e-01 -1.01715219e+00
-5.21850824e-01 2.30172411e-01 -1.86787486e-01 4.31017578e-01
-1.42471492e+00 8.27941418e-01 6.05230749e-01 -5.05182683e-01
-4.50012177e-01 -3.75671536e-01 -9.97624159e-01 6.08084947e-02
6.81866944e-01 -6.83444858e-01 1.93806648e+00 -8.61674905e-01
-1.84323061e+00 7.79882848e-01 2.36070514e-01 -3.77095699e-01
4.93889689e-01 4.47240882e-02 -2.71604568e-01 -1.70847297e-01
4.05318975e-01 5.84037542e-01 2.59982497e-01 -1.18283927e+00
-1.10678160e+00 -4.08749908e-01 7.22064257e-01 4.93637681e-01
-2.71716863e-01 4.41415429e-01 -6.87580049e-01 1.91562161e-01
-1.92509487e-01 -6.46291554e-01 -1.90953732e-01 -5.50315022e-01
-2.66476661e-01 -5.33169270e-01 6.22853994e-01 -7.74582744e-01
1.44838059e+00 -1.56092668e+00 -1.39987856e-01 4.80431598e-03
-4.62616310e-02 1.31345302e-01 1.13281533e-01 1.03162205e+00
2.03400105e-01 2.67939657e-01 1.24749653e-01 -9.07213017e-02
6.56150281e-01 2.40217611e-01 2.77371518e-02 -4.31662619e-01
-2.33470201e-01 2.16434285e-01 -9.70243096e-01 -4.59463000e-01
2.06658021e-01 -7.50371143e-02 -5.68096340e-01 7.49711871e-01
-1.04267526e+00 3.23042154e-01 -5.83414912e-01 2.81987846e-01
2.54341334e-01 -6.83612376e-02 5.92331052e-01 8.78547281e-02
-5.20028353e-01 9.19134021e-01 -1.15599561e+00 2.09804249e+00
-1.04975331e+00 1.68533757e-01 3.01321328e-01 -5.71924150e-01
1.12392747e+00 8.02114487e-01 2.86727309e-01 -3.75472873e-01
-1.39856398e-01 3.33204925e-01 -4.14450094e-02 -9.04718220e-01
5.39800048e-01 2.86032110e-01 -4.88664091e-01 9.49337065e-01
3.97984535e-01 -2.45432600e-01 6.07786417e-01 1.46204546e-01
1.17665088e+00 2.95711637e-01 6.00348532e-01 -2.78107375e-01
9.18053210e-01 6.29805863e-01 1.11566149e-01 8.62364650e-01
1.74219474e-01 2.72302199e-02 7.44120657e-01 -5.58264017e-01
-1.16396904e+00 -2.86571801e-01 -7.54640996e-02 1.59446311e+00
-3.41140002e-01 -8.75941992e-01 -1.30993617e+00 -1.04542279e+00
-6.83669567e-01 1.08247054e+00 7.56868534e-03 3.21603954e-01
-4.17273939e-01 -7.24941790e-02 5.55940330e-01 -2.19240546e-01
6.03209376e-01 -1.17380965e+00 -7.67390490e-01 6.16623878e-01
-2.86916077e-01 -9.58503067e-01 -1.97414130e-01 2.47388966e-02
-2.56199569e-01 -1.16308713e+00 -6.86746538e-02 -7.22285986e-01
4.45454299e-01 -3.62649232e-01 1.19562125e+00 2.27369174e-01
-7.49945939e-02 7.11144030e-01 -6.63137615e-01 -5.18585980e-01
-1.27642262e+00 4.34795052e-01 -3.71028244e-01 -4.07987475e-01
6.89142942e-01 -3.37283671e-01 -1.45696402e-01 4.84090209e-01
-8.68413031e-01 5.97976625e-01 -1.64973080e-01 6.08756185e-01
-2.43790954e-01 -1.32630527e-01 5.15456796e-01 -1.24351668e+00
1.27535260e+00 -3.12785655e-01 -7.14965641e-01 4.21223968e-01
-5.70138156e-01 3.15763354e-01 6.98226035e-01 -5.39518753e-03
-1.46774387e+00 2.43504971e-01 -7.45523751e-01 7.51314223e-01
-7.87030518e-01 7.55108953e-01 -3.58908087e-01 1.38149485e-01
8.51703525e-01 7.45466650e-02 1.25839397e-01 -8.06264162e-01
5.84420204e-01 1.30012798e+00 1.21134758e-01 -8.88573349e-01
3.51363420e-01 -9.94422585e-02 -7.14279175e-01 -8.51076484e-01
-3.72211486e-01 -8.58167112e-01 -5.73692083e-01 -3.19442600e-01
6.63999259e-01 -4.72527921e-01 -6.77866280e-01 1.46268591e-01
-1.40130436e+00 -6.76042020e-01 -4.36241239e-01 -1.41510084e-01
-8.79806280e-01 3.15660179e-01 -1.75551429e-01 -1.04741478e+00
-5.89090526e-01 -9.78529096e-01 8.19091201e-01 3.47528160e-01
-8.82128179e-01 -1.11624277e+00 2.77900040e-01 5.69488585e-01
5.29791236e-01 -1.32894710e-01 9.39719021e-01 -1.46406031e+00
-1.53253889e-02 -1.13253944e-01 -2.36726888e-02 7.23984465e-02
3.73139799e-01 -9.88697857e-02 -8.29653919e-01 2.77330577e-01
1.97385877e-01 -3.55896741e-01 -2.69513726e-01 -4.18246478e-01
3.92998070e-01 -6.71510160e-01 -1.95191920e-01 -6.65949509e-02
1.13454676e+00 5.73176086e-01 4.59003329e-01 5.98795831e-01
1.49273559e-01 1.19829261e+00 9.47739780e-01 6.84750617e-01
6.34234011e-01 1.05859041e+00 1.22438923e-01 4.86515388e-02
-2.46732589e-02 -1.72189772e-01 2.03692526e-01 3.00104648e-01
1.89665392e-01 -2.78060548e-02 -1.21442091e+00 4.88328755e-01
-2.09874845e+00 -6.71140313e-01 -9.87980366e-02 2.03390527e+00
1.38100910e+00 7.74491355e-02 2.62923539e-01 -2.89684802e-01
6.02999151e-01 -1.23913683e-01 -1.86302185e-01 -7.26349354e-01
6.74643874e-01 1.66985206e-02 4.17080987e-03 9.53041971e-01
-7.37057686e-01 1.18104136e+00 5.50759363e+00 2.76763529e-01
-7.52545476e-01 1.73896298e-01 7.61073679e-02 4.50809896e-01
-3.18186969e-01 4.77512062e-01 -1.13154113e+00 2.51248717e-01
1.12380004e+00 -4.13655847e-01 5.68853557e-01 1.09855688e+00
3.96907419e-01 -2.15873942e-01 -1.37299430e+00 3.78786862e-01
-1.68202549e-01 -1.33005822e+00 -1.62468329e-01 -1.26927882e-01
1.42346369e-03 -2.41275281e-01 -8.05794716e-01 5.98534226e-01
5.22767901e-01 -5.32337964e-01 3.86068225e-01 4.12626624e-01
6.94959223e-01 -1.44554004e-01 6.75007045e-01 8.01065564e-01
-7.25940943e-01 1.36099935e-01 -5.15898354e-02 -2.30872497e-01
1.81395799e-01 4.41565597e-03 -1.58858299e+00 3.42544526e-01
3.45654577e-01 -1.24137767e-01 -4.40731376e-01 6.33098722e-01
-1.06152453e-01 1.17246099e-01 -5.88081598e-01 -5.46245396e-01
1.61667705e-01 -2.09166974e-01 6.50077999e-01 1.18368268e+00
4.94906828e-02 1.09592006e-01 2.40593717e-01 7.12016821e-01
1.94554359e-01 7.44697511e-01 -7.05410123e-01 -1.35121718e-01
2.08982259e-01 1.35352349e+00 -4.21764940e-01 -4.78554904e-01
-6.11748159e-01 9.06200290e-01 4.12541255e-02 5.93718775e-02
-4.20619696e-01 -7.23238528e-01 5.31682253e-01 2.57929027e-01
-2.41456777e-01 -1.72688235e-02 1.36561766e-01 -9.67441738e-01
7.37215728e-02 -1.29246759e+00 4.68145043e-01 -8.92812848e-01
-7.15616643e-01 7.16893017e-01 1.69943273e-01 -5.99330962e-01
-9.36652780e-01 -3.35874200e-01 -5.91804683e-01 1.02098215e+00
-9.25751507e-01 -1.01927233e+00 -4.41529453e-01 5.14504671e-01
1.00710821e+00 -2.62974590e-01 1.45659137e+00 1.18786633e-01
-1.42731875e-01 -9.04833600e-02 -4.53054488e-01 1.49494529e-01
7.62852907e-01 -1.60897708e+00 4.64336932e-01 5.51696718e-01
-1.40162736e-01 9.42519844e-01 1.12035072e+00 -6.65370822e-01
-1.35359550e+00 -5.82883298e-01 1.34346414e+00 -2.97264010e-01
7.78308690e-01 -7.16134429e-01 -7.55445421e-01 5.92447102e-01
6.94614291e-01 -7.49417543e-01 8.45764220e-01 3.31545830e-01
6.80197701e-02 1.37113407e-01 -1.20974112e+00 6.28552377e-01
8.28018725e-01 -4.14007306e-01 -1.03058469e+00 8.43912840e-01
6.45033419e-01 -6.28638864e-01 -1.04901588e+00 -1.37418985e-01
3.60611856e-01 -9.61971879e-01 5.18833220e-01 -7.96360910e-01
1.02969781e-01 -2.29369432e-01 1.20343469e-01 -1.07058477e+00
4.80635077e-01 -1.27654970e+00 2.44148865e-01 1.72517788e+00
8.14300776e-01 -7.41210759e-01 4.88711834e-01 1.62354004e+00
-2.26118371e-01 1.73614230e-02 -5.65995336e-01 -3.59781653e-01
-4.18624580e-01 -5.16447186e-01 8.71781290e-01 7.08791673e-01
1.13690341e+00 9.50537443e-01 2.36874283e-03 7.77222887e-02
3.60522978e-02 4.11311956e-03 1.27125454e+00 -1.27979755e+00
-4.16116476e-01 -1.52231455e-01 1.20798431e-01 -9.03618217e-01
-6.75739497e-02 -6.21567965e-01 2.49707595e-01 -1.92367065e+00
-2.77325332e-01 -6.71445549e-01 5.65992057e-01 8.49827409e-01
2.16384321e-01 -6.39407039e-01 -7.67277479e-02 2.30496064e-01
-7.30200112e-01 -4.96855341e-02 6.89693213e-01 1.51155517e-02
-8.45964015e-01 1.82079598e-01 -7.51123548e-01 8.89821410e-01
9.71966982e-01 -4.51279670e-01 -5.64538419e-01 6.20432049e-02
3.72372627e-01 4.86504644e-01 -2.05063283e-01 -9.83471572e-01
4.86194581e-01 -2.26399764e-01 -4.94856209e-01 -1.30887166e-01
3.31018455e-02 -9.73946929e-01 2.95269281e-01 1.85552359e-01
-7.03553319e-01 -5.38825035e-01 1.29304200e-01 -2.07237750e-01
-1.29744783e-01 -1.12714064e+00 2.61891782e-01 -5.53076267e-01
-9.25589442e-01 -3.45848948e-01 -7.85130143e-01 2.79064178e-02
1.03450108e+00 -3.25037576e-02 -5.11546254e-01 -6.09635830e-01
-9.87626433e-01 4.62662339e-01 6.76850319e-01 4.89570320e-01
1.04284085e-01 -5.12767315e-01 -1.85199738e-01 8.92312825e-02
3.94839674e-01 -1.68979838e-01 -9.90668014e-02 2.23501146e-01
-6.84756875e-01 6.99788213e-01 -3.52147311e-01 -1.97173551e-01
-1.42932928e+00 2.61305690e-01 4.88632113e-01 -4.33532536e-01
-3.89436185e-01 3.67269307e-01 -1.30661160e-01 -1.04438889e+00
2.60890067e-01 -5.26527055e-02 -7.00032592e-01 7.58350641e-02
7.84793258e-01 -2.63082478e-02 2.79757440e-01 -3.46025497e-01
-1.60508618e-01 -7.73284286e-02 -2.29909155e-03 -6.77140772e-01
1.42981589e+00 -3.77116144e-01 -3.20949674e-01 4.10463214e-01
6.89782321e-01 -3.69687304e-02 -8.92200112e-01 8.24357476e-03
3.23181838e-01 -5.12088537e-02 -2.33120620e-01 -1.12434137e+00
-1.17368594e-01 3.34513932e-01 3.26401800e-01 1.30984044e+00
7.23163784e-01 2.11358741e-01 4.99624878e-01 8.31547439e-01
6.67036474e-01 -1.41712964e+00 -3.22624236e-01 7.37908602e-01
1.11021233e+00 -1.27485144e+00 -4.01694506e-01 -5.38375497e-01
-7.28009999e-01 1.60791898e+00 6.96259618e-01 6.84940577e-01
3.56390536e-01 3.56036127e-01 3.83050561e-01 -6.04157567e-01
-9.41754162e-01 -4.26953316e-01 -2.00885147e-01 8.51099014e-01
7.19686806e-01 -3.01750690e-01 -1.05999207e+00 7.64471054e-01
-3.23802352e-01 6.49636686e-02 8.00871015e-01 1.17336130e+00
-7.07304239e-01 -1.72422171e+00 -1.72195584e-01 1.96633503e-01
-4.31679934e-01 -2.26143330e-01 -8.25649500e-01 7.63633251e-01
-1.34445474e-01 1.32700074e+00 3.44583578e-02 -6.74694330e-02
6.42457843e-01 7.84111559e-01 4.10554111e-02 -1.43006170e+00
-1.11282694e+00 2.73307059e-02 1.28226793e+00 -5.93452752e-01
-3.00619930e-01 -6.67227268e-01 -1.40977311e+00 2.57227212e-01
-2.09366903e-01 6.99709117e-01 1.11853015e+00 8.71616483e-01
3.27712595e-01 2.35283226e-01 4.17747110e-01 -2.42189810e-01
-4.59595770e-01 -1.19146848e+00 -2.23482624e-01 4.31124181e-01
-3.16870749e-01 -8.08801949e-02 -1.73209995e-01 4.53093737e-01]
|
[12.84169864654541, 7.908576011657715]
|
c1af86a5-2799-4376-abf9-725a5799514d
|
multi-modal-fusion-for-end-to-end-rgb-t
|
1908.11714
| null |
https://arxiv.org/abs/1908.11714v1
|
https://arxiv.org/pdf/1908.11714v1.pdf
|
Multi-Modal Fusion for End-to-End RGB-T Tracking
|
We propose an end-to-end tracking framework for fusing the RGB and TIR modalities in RGB-T tracking. Our baseline tracker is DiMP (Discriminative Model Prediction), which employs a carefully designed target prediction network trained end-to-end using a discriminative loss. We analyze the effectiveness of modality fusion in each of the main components in DiMP, i.e. feature extractor, target estimation network, and classifier. We consider several fusion mechanisms acting at different levels of the framework, including pixel-level, feature-level and response-level. Our tracker is trained in an end-to-end manner, enabling the components to learn how to fuse the information from both modalities. As data to train our model, we generate a large-scale RGB-T dataset by considering an annotated RGB tracking dataset (GOT-10k) and synthesizing paired TIR images using an image-to-image translation approach. We perform extensive experiments on VOT-RGBT2019 dataset and RGBT210 dataset, evaluating each type of modality fusing on each model component. The results show that the proposed fusion mechanisms improve the performance of the single modality counterparts. We obtain our best results when fusing at the feature-level on both the IoU-Net and the model predictor, obtaining an EAO score of 0.391 on VOT-RGBT2019 dataset. With this fusion mechanism we achieve the state-of-the-art performance on RGBT210 dataset.
|
['Joost Van de Weijer', 'Abel Gonzalez-Garcia', 'Martin Danelljan', 'Lichao Zhang', 'Fahad Shahbaz Khan']
|
2019-08-30
| null | null | null | null |
['rgb-t-tracking']
|
['computer-vision']
|
[ 2.75891930e-01 -3.44363600e-02 -2.28852749e-01 -2.43649855e-01
-1.30139041e+00 -6.61854029e-01 7.15768158e-01 -5.03777981e-01
-5.21905482e-01 3.23217750e-01 -1.77207440e-01 6.62352797e-03
2.32134297e-01 -1.66326374e-01 -1.04312706e+00 -7.95707405e-01
3.83138329e-01 2.05891192e-01 4.38831598e-01 6.01853244e-02
-2.58017361e-01 3.62971753e-01 -1.69142723e+00 2.33170018e-01
6.15310848e-01 1.86007571e+00 5.35535291e-02 9.38078940e-01
3.85730147e-01 6.95692003e-01 -1.58273131e-01 -5.09242892e-01
6.76220298e-01 -2.76590753e-02 -1.87854111e-01 1.51258120e-02
1.16838050e+00 -5.07307127e-02 -4.86244410e-01 7.72499204e-01
8.11349690e-01 -1.13806076e-01 3.96132141e-01 -1.59925091e+00
-1.71873778e-01 5.50683849e-02 -5.87242663e-01 -1.76786929e-01
2.74654984e-01 7.19219148e-01 7.27459311e-01 -1.13617945e+00
6.06364131e-01 1.26005578e+00 9.47301030e-01 6.62879586e-01
-1.15704405e+00 -7.85019755e-01 1.25447139e-01 2.00118452e-01
-1.06808293e+00 -5.62326610e-01 6.61228895e-01 -3.77928913e-01
7.28663445e-01 2.35310465e-01 6.54746830e-01 1.38009918e+00
3.54961753e-01 1.16502905e+00 1.42940736e+00 -2.57102370e-01
-5.94668724e-02 1.24587744e-01 -1.00732394e-01 4.72116709e-01
-1.64930388e-01 7.54840314e-01 -9.40025926e-01 1.17548905e-01
4.61740106e-01 -1.67517781e-01 -1.02653049e-01 -5.03652930e-01
-1.41259646e+00 2.27490231e-01 7.53628671e-01 -8.24056193e-02
-2.83303529e-01 7.00702667e-01 3.08794200e-01 1.49475321e-01
3.25593203e-01 -2.95043528e-01 -6.62252784e-01 -4.00521010e-02
-8.97742331e-01 1.13454841e-01 2.65499860e-01 1.11480570e+00
4.70968843e-01 -3.35251428e-02 -6.84220910e-01 3.00263047e-01
6.62930667e-01 1.20882392e+00 6.68618605e-02 -8.92101645e-01
5.27778983e-01 2.92629302e-01 2.01745510e-01 -3.08889866e-01
-4.65015560e-01 -6.16086245e-01 -4.24116910e-01 4.08543199e-01
4.63110149e-01 -1.51925325e-01 -1.33336878e+00 1.96310914e+00
6.85532749e-01 3.84905964e-01 1.34771377e-01 1.09183991e+00
9.12109971e-01 3.11211467e-01 4.31432456e-01 -7.11157843e-02
1.47824955e+00 -1.25050509e+00 -5.65754712e-01 -2.35004380e-01
5.14758825e-01 -8.91563535e-01 8.13418984e-01 2.00602472e-01
-8.68360996e-01 -1.15518701e+00 -6.89483821e-01 3.46349850e-02
-4.18472618e-01 7.62880921e-01 2.70423055e-01 6.89172089e-01
-1.04277694e+00 3.88391107e-01 -9.37923610e-01 -4.99278575e-01
2.88539320e-01 5.10865033e-01 -4.65557098e-01 8.72231834e-03
-9.24779952e-01 9.65480030e-01 4.27100420e-01 3.66685778e-01
-1.15284264e+00 -8.68290484e-01 -5.52450538e-01 -3.36271256e-01
4.15823430e-01 -9.21819746e-01 1.16557693e+00 -7.54640818e-01
-1.38606834e+00 7.80396819e-01 -2.02444226e-01 -4.62286383e-01
7.76790798e-01 -3.09306771e-01 -4.44308460e-01 -5.14061563e-02
-1.93759091e-02 1.04587710e+00 1.11647952e+00 -1.34192967e+00
-1.22047746e+00 -3.65861952e-01 -3.12174290e-01 7.37999976e-02
-1.59211252e-02 -1.44052967e-01 -9.17209148e-01 -3.97873014e-01
-1.22692719e-01 -1.48958516e+00 1.42454743e-01 3.44543546e-01
-4.25461948e-01 -2.25893751e-01 1.20881248e+00 -6.54494822e-01
6.63311958e-01 -2.11148143e+00 1.30804539e-01 -7.54360110e-02
8.66750479e-02 3.36767107e-01 -2.37218022e-01 1.11563340e-01
-4.00412083e-02 -4.66735393e-01 1.36152551e-01 -9.70255613e-01
1.79496542e-01 1.59599185e-02 -1.56683877e-01 5.63956082e-01
5.46390265e-02 1.28817236e+00 -8.21237922e-01 -5.24703860e-01
5.79799891e-01 6.94866240e-01 -1.60744444e-01 3.20420772e-01
-3.39262843e-01 8.18475127e-01 -5.30902505e-01 1.19097376e+00
7.49604583e-01 -3.57536711e-02 -1.22026861e-01 -8.89291584e-01
-2.27581754e-01 -2.24587396e-01 -8.71251702e-01 1.97468674e+00
-5.73684752e-01 8.19012105e-01 -9.42421630e-02 -1.33539945e-01
6.54715300e-01 2.44593248e-01 6.50886059e-01 -8.22596550e-01
3.08152705e-01 3.54781657e-01 -2.94166476e-01 -1.89149678e-01
6.09366417e-01 8.03275853e-02 -4.98535752e-01 -7.74087533e-02
3.96414697e-01 1.89541653e-01 -1.09881293e-02 9.20683146e-02
1.11729074e+00 8.62156093e-01 -3.75488013e-01 2.68168956e-01
4.59941596e-01 2.12369189e-01 5.11463583e-01 8.84473264e-01
-4.68805343e-01 5.33784866e-01 5.18514439e-02 -1.02146976e-01
-8.91995430e-01 -1.22745407e+00 -8.28500092e-03 1.08618414e+00
3.15234125e-01 -1.47129238e-01 -3.80224288e-01 -8.97075832e-01
3.15876931e-01 6.08910501e-01 -8.43672216e-01 -3.60429883e-01
-4.08595234e-01 -3.06902707e-01 7.44902372e-01 7.21049547e-01
6.69537246e-01 -7.28044450e-01 -7.94827998e-01 -3.27048264e-02
-2.30125397e-01 -1.66296554e+00 -4.98859048e-01 4.74286288e-01
-6.95146501e-01 -9.70461249e-01 -5.10038853e-01 -1.88766256e-01
2.17972010e-01 2.69486934e-01 8.02606583e-01 -4.24421698e-01
-1.29468188e-01 9.74268198e-01 -2.51856118e-01 -3.37008983e-01
-1.46141052e-01 -6.73225671e-02 2.64601767e-01 2.97986448e-01
8.72487500e-02 -7.75374845e-02 -5.84428012e-01 4.01822895e-01
-6.34919107e-01 4.11886498e-02 8.49380434e-01 5.78056872e-01
8.13451231e-01 -5.07243574e-01 -6.59795627e-02 -3.67409512e-02
-2.64811009e-01 1.36747502e-03 -8.66507947e-01 5.42500496e-01
-4.48803753e-01 -9.64960977e-02 3.22359532e-01 -5.53499222e-01
-1.07676435e+00 8.54170322e-01 5.72911575e-02 -1.32601941e+00
-1.18725181e-01 -1.73474271e-02 -3.41203302e-01 -7.21602559e-01
3.96808505e-01 3.33401799e-01 -2.07504958e-01 -5.06464124e-01
7.11522639e-01 4.64637965e-01 9.92752075e-01 -4.74796742e-01
1.32633626e+00 5.19597054e-01 2.23916247e-01 -4.54264939e-01
-1.01267135e+00 -6.19360209e-01 -5.94847739e-01 -8.25975776e-01
1.00047433e+00 -1.32812071e+00 -1.11658394e+00 6.51311219e-01
-9.51461256e-01 -3.37949574e-01 -4.53961283e-01 5.72800875e-01
-6.13797307e-01 -1.97749538e-03 -2.48242944e-01 -8.28034103e-01
-3.22573572e-01 -1.22622693e+00 1.85249186e+00 4.58334059e-01
5.79767942e-01 -7.18373239e-01 2.40299150e-01 4.94413257e-01
3.61941069e-01 4.20049071e-01 -2.63326392e-02 -3.63840371e-01
-7.86812663e-01 -3.45839739e-01 -3.59545469e-01 2.43434444e-01
-2.75458813e-01 -1.60747245e-01 -1.58319104e+00 -4.15937036e-01
-3.64418060e-01 -6.33988738e-01 1.22674906e+00 2.56589860e-01
7.76643395e-01 3.37156832e-01 -6.20374978e-01 8.17804158e-01
1.59200644e+00 -4.10481617e-02 5.53538918e-01 2.40215048e-01
1.26525664e+00 2.19385043e-01 1.26487947e+00 1.52989149e-01
3.84641320e-01 1.17913413e+00 7.61606753e-01 -1.40405074e-01
-5.01245737e-01 -2.63512731e-01 8.34057331e-01 6.90959767e-02
1.37963109e-02 -2.85775572e-01 -7.49250770e-01 3.46087575e-01
-2.16297531e+00 -8.10298979e-01 -1.89231828e-01 2.12650228e+00
5.22357881e-01 1.42578736e-01 4.19055432e-01 -2.91051567e-01
4.55683649e-01 2.17213146e-02 -6.19661331e-01 2.62510747e-01
-2.04174086e-01 -1.39681324e-01 1.03084254e+00 2.08749086e-01
-1.27178979e+00 1.24161398e+00 5.34174681e+00 1.09922373e+00
-1.48575544e+00 3.26799273e-01 8.53990018e-02 -3.41004223e-01
1.79493353e-01 1.54762089e-01 -1.17008436e+00 4.90228623e-01
1.11830592e+00 4.23646331e-01 1.38481289e-01 7.43888855e-01
2.19431058e-01 -1.75813094e-01 -1.18333232e+00 1.04202592e+00
-7.34804943e-02 -1.06203008e+00 -3.74541044e-01 1.31678641e-01
4.36767012e-01 4.81115609e-01 2.74896622e-01 5.00004530e-01
1.03001811e-01 -6.37732863e-01 1.36679864e+00 8.64751279e-01
8.47795904e-01 -2.51787633e-01 5.63647330e-01 1.11869000e-01
-1.69151127e+00 -9.42754596e-02 -7.74731934e-02 6.11998320e-01
2.50752628e-01 3.30914140e-01 -6.76404238e-01 1.03996861e+00
8.39877546e-01 9.13772583e-01 -9.83885109e-01 1.12211227e+00
-1.97012037e-01 5.01918018e-01 -7.08933890e-01 4.30051118e-01
1.48294806e-01 2.71687746e-01 6.13346457e-01 1.04280663e+00
3.51872206e-01 -3.82340133e-01 3.92687470e-01 5.40599346e-01
7.97273591e-03 -6.15796983e-01 -3.83050382e-01 2.44509175e-01
4.65288281e-01 1.61159039e+00 -4.97998863e-01 -2.91887909e-01
-3.51077229e-01 7.86806107e-01 -1.85740907e-02 2.54346073e-01
-1.54031301e+00 2.26862639e-01 5.44229507e-01 -2.00012967e-01
7.91803718e-01 6.74927980e-02 -3.11695505e-02 -1.05066824e+00
1.64326206e-01 -6.95249259e-01 2.58917183e-01 -1.28523910e+00
-1.06226659e+00 6.04038656e-01 5.82816303e-02 -1.84411907e+00
-1.84012771e-01 -7.83239543e-01 -2.14167863e-01 8.09545100e-01
-1.62417901e+00 -2.12409234e+00 -5.95120490e-01 8.51043344e-01
5.73391356e-02 1.30690441e-01 2.24523976e-01 4.27952468e-01
-5.85752845e-01 9.20100927e-01 -1.01468891e-01 -1.40391484e-01
9.02109265e-01 -1.04809797e+00 1.42488688e-01 1.04100418e+00
3.93608175e-02 2.00205803e-01 7.26022065e-01 -6.42863572e-01
-1.93899763e+00 -1.50490928e+00 5.65092444e-01 -9.17234182e-01
6.29213214e-01 -4.69359040e-01 -2.45000258e-01 7.03976572e-01
1.21595494e-01 6.16442740e-01 1.88354164e-01 -1.84450254e-01
-3.72576445e-01 -4.51925308e-01 -1.05229688e+00 2.44113564e-01
1.02626979e+00 -6.50856197e-01 -2.96634018e-01 2.40609109e-01
8.25433969e-01 -7.95290649e-01 -1.21766031e+00 6.08653009e-01
9.74353850e-01 -8.14891458e-01 1.07761776e+00 -9.25639197e-02
-1.33330777e-01 -8.38563025e-01 -4.17291909e-01 -1.04834759e+00
-8.22383240e-02 -5.27352035e-01 -4.11595494e-01 1.33025765e+00
2.94758499e-01 -3.33339840e-01 8.37100625e-01 4.44286495e-01
-3.65704775e-01 -6.38980329e-01 -1.25749111e+00 -1.04959452e+00
-4.53792214e-01 -8.51919234e-01 2.52812058e-01 3.63369375e-01
-6.73132360e-01 3.55721086e-01 -8.03476751e-01 2.03653827e-01
1.18200195e+00 1.93366542e-01 1.13369536e+00 -9.41536665e-01
-3.16374093e-01 -4.09083180e-02 -3.50699604e-01 -9.64276314e-01
-1.12845659e-01 -6.30189896e-01 3.07969928e-01 -1.16842091e+00
-6.66380720e-03 -4.95206922e-01 -4.81724143e-01 7.18481362e-01
-1.87243447e-01 5.84980190e-01 7.34432936e-01 4.19515729e-01
-1.02001691e+00 7.42248952e-01 1.19238555e+00 -1.92328036e-01
-4.75399606e-02 -1.07071035e-01 -2.00565130e-01 4.29858804e-01
3.35426301e-01 -6.52696967e-01 -2.17054859e-02 -2.35109046e-01
-1.49366602e-01 1.83334097e-01 1.05124903e+00 -1.50528955e+00
4.64466602e-01 1.41212329e-01 6.78110600e-01 -1.19618154e+00
7.48404860e-01 -1.30503225e+00 4.87025797e-01 5.65898955e-01
-1.28477976e-01 -3.58369201e-01 5.20468116e-01 6.93426609e-01
-2.02426240e-02 6.19004905e-01 7.90214241e-01 5.16660452e-01
-1.09674644e+00 4.36844885e-01 3.15196931e-01 -3.32227021e-01
1.11403012e+00 -4.22272146e-01 -5.26469946e-01 1.14684470e-01
-7.59366691e-01 4.93706822e-01 5.43622851e-01 6.37162685e-01
2.92976648e-01 -1.65267539e+00 -3.74716043e-01 -4.31051627e-02
6.21563971e-01 -4.35281068e-01 3.89893651e-01 1.47079027e+00
1.53931394e-01 4.75176305e-01 -3.25854123e-01 -1.21028686e+00
-1.26592588e+00 5.63124061e-01 5.42896509e-01 -5.09665668e-01
-2.64171690e-01 6.96953475e-01 5.89386001e-03 -4.51564014e-01
4.77941751e-01 -3.81605059e-01 1.01304114e-01 -3.36788818e-02
2.07093000e-01 1.91919431e-01 2.95257512e-02 -1.07192159e+00
-8.45178545e-01 1.04500055e+00 2.38981679e-01 -1.55808210e-01
9.63833630e-01 -1.64549366e-01 4.41507578e-01 1.86619058e-01
1.15982056e+00 -2.35346168e-01 -1.77612281e+00 -2.33506501e-01
-4.15008903e-01 -3.62165987e-01 3.46391052e-01 -1.11316848e+00
-1.28838837e+00 6.92453027e-01 1.46226585e+00 -2.46203646e-01
1.36887026e+00 9.10085514e-02 8.34998906e-01 1.59855276e-01
4.90825593e-01 -8.92302275e-01 1.40629327e-02 4.31673586e-01
5.24967968e-01 -1.43662524e+00 -7.27426559e-02 -2.37634510e-01
-5.99232554e-01 7.46715009e-01 7.70182550e-01 5.21261171e-02
3.27465504e-01 1.66667357e-01 1.46490782e-01 -1.89842284e-01
-8.37571204e-01 -8.27651560e-01 6.65915012e-01 6.66143835e-01
8.60138834e-02 -1.46623358e-01 2.26169944e-01 2.24417269e-01
3.62373799e-01 2.59724408e-01 -2.28189975e-01 9.72255468e-01
-1.36091560e-01 -1.06811786e+00 -7.50515163e-01 1.30491331e-01
-1.29656807e-01 1.64629340e-01 -4.96461630e-01 1.02681804e+00
4.72606748e-01 8.51658762e-01 -4.04099345e-01 -1.05908537e+00
6.56572998e-01 -1.44647390e-01 7.54916966e-01 7.70643502e-02
-9.32363272e-01 3.54746312e-01 1.94151551e-01 -1.14982259e+00
-8.12450945e-01 -8.07308853e-01 -8.56517017e-01 -2.48448178e-01
-4.53010648e-01 -2.43978679e-01 7.15453029e-01 1.02732158e+00
4.74791467e-01 7.39139080e-01 6.02194548e-01 -1.45619011e+00
-4.21714008e-01 -9.15248096e-01 -3.64819616e-01 1.89596131e-01
5.14057875e-01 -9.41945791e-01 -9.04373676e-02 7.59507120e-02]
|
[6.354137420654297, -2.189080238342285]
|
c2ccde7e-bc30-4583-8d45-0afac7cd0de5
|
mdd-eval-self-training-on-augmented-data-for
|
2112.07194
| null |
https://arxiv.org/abs/2112.07194v2
|
https://arxiv.org/pdf/2112.07194v2.pdf
|
MDD-Eval: Self-Training on Augmented Data for Multi-Domain Dialogue Evaluation
|
Chatbots are designed to carry out human-like conversations across different domains, such as general chit-chat, knowledge exchange, and persona-grounded conversations. To measure the quality of such conversational agents, a dialogue evaluator is expected to conduct assessment across domains as well. However, most of the state-of-the-art automatic dialogue evaluation metrics (ADMs) are not designed for multi-domain evaluation. We are motivated to design a general and robust framework, MDD-Eval, to address the problem. Specifically, we first train a teacher evaluator with human-annotated data to acquire a rating skill to tell good dialogue responses from bad ones in a particular domain and then, adopt a self-training strategy to train a new evaluator with teacher-annotated multi-domain data, that helps the new evaluator to generalize across multiple domains. MDD-Eval is extensively assessed on six dialogue evaluation benchmarks. Empirical results show that the MDD-Eval framework achieves a strong performance with an absolute improvement of 7% over the state-of-the-art ADMs in terms of mean Spearman correlation scores across all the evaluation benchmarks.
|
['Haizhou Li', 'Thomas Friedrichs', "Luis Fernando D'Haro", 'Chen Zhang']
|
2021-12-14
| null | null | null | null |
['dialogue-evaluation']
|
['natural-language-processing']
|
[-1.80019617e-01 2.82713473e-01 1.60208493e-01 -7.36608684e-01
-9.06301498e-01 -6.43943548e-01 8.92744482e-01 2.73783691e-02
-3.33759695e-01 1.07258046e+00 3.65335286e-01 1.74071435e-02
7.06852898e-02 -5.85376620e-01 1.19079009e-01 -3.76205951e-01
3.19610894e-01 1.09535193e+00 2.75043935e-01 -7.79634118e-01
2.14627266e-01 -1.41926005e-01 -7.51391649e-01 3.85606736e-01
1.39843500e+00 7.69160628e-01 -4.22001518e-02 7.91219294e-01
-2.21021488e-01 1.17062223e+00 -1.21135902e+00 -1.00370324e+00
-3.09418797e-01 -9.32577133e-01 -1.66107607e+00 5.34466244e-02
-8.47935081e-02 -6.99712873e-01 -7.77249411e-02 7.58705616e-01
7.12831557e-01 3.94646376e-01 8.31349850e-01 -1.29807889e+00
-5.29283822e-01 7.66634583e-01 1.27346829e-01 -1.42135084e-01
8.50836694e-01 5.73548496e-01 1.07053602e+00 -6.49732769e-01
5.70162833e-01 1.47345877e+00 4.79214787e-01 1.04036975e+00
-1.01281345e+00 -2.93471903e-01 -1.94467574e-01 4.69166785e-02
-4.99848455e-01 -2.95992851e-01 7.00595558e-01 -4.82586682e-01
7.13744879e-01 4.60016616e-02 1.15736976e-01 1.44403052e+00
-5.70765853e-01 1.01252902e+00 1.39814627e+00 -1.53965533e-01
1.26370013e-01 5.43596506e-01 2.13341653e-01 4.23088580e-01
-5.82425773e-01 -4.22287315e-01 -6.68988585e-01 -1.55976176e-01
4.11422074e-01 -4.99812335e-01 -1.75538585e-01 3.34422439e-02
-1.22805679e+00 1.04418373e+00 -6.04016520e-03 3.59263897e-01
-4.12389606e-01 -6.14732683e-01 9.71670866e-01 8.77378523e-01
6.66377723e-01 9.88848209e-01 -4.95803386e-01 -9.17881072e-01
-2.51574546e-01 3.57186377e-01 1.49657702e+00 7.53819883e-01
4.50861841e-01 -3.67319643e-01 -6.43937409e-01 1.55146670e+00
-3.70645430e-03 4.79557477e-02 5.04267335e-01 -1.21954763e+00
8.00666749e-01 1.03989100e+00 1.96962267e-01 -4.81128126e-01
-3.24828774e-01 1.58698514e-01 -9.26933467e-01 3.19853537e-02
7.11336672e-01 -7.32577801e-01 -2.17426252e-02 1.73384809e+00
5.27296841e-01 -3.27094406e-01 6.40151203e-01 9.12965119e-01
1.26682115e+00 5.90985000e-01 -8.30583498e-02 -1.60530895e-01
1.20025682e+00 -1.35346460e+00 -6.76039398e-01 -4.92036901e-03
7.82523096e-01 -5.88734388e-01 1.49388671e+00 4.34414655e-01
-1.17548752e+00 -4.58641917e-01 -7.58111715e-01 -9.09233280e-03
-3.88151780e-02 -1.78736508e-01 2.15760022e-01 4.27133173e-01
-7.34481335e-01 4.92752790e-01 -2.51082987e-01 -3.74260157e-01
-4.18248102e-02 -4.62170914e-02 -4.18434173e-01 1.05255939e-01
-1.60409141e+00 1.26669168e+00 1.56579629e-01 -1.71284050e-01
-1.32472444e+00 -5.63081861e-01 -6.14151537e-01 -1.31923646e-01
2.67660826e-01 -4.14862454e-01 1.92768455e+00 -8.28708589e-01
-2.30207562e+00 9.79571819e-01 4.44826812e-01 -3.07061672e-01
9.29071248e-01 -2.31417328e-01 -1.22444898e-01 1.52052492e-01
6.08387068e-02 3.38457972e-01 1.11525558e-01 -1.04361784e+00
-5.26733279e-01 -7.35824630e-02 6.07450902e-01 6.37724340e-01
-4.83422726e-01 1.62989154e-01 -1.05536021e-01 -1.17140800e-01
-5.81671000e-01 -7.43693888e-01 -7.62436492e-03 -4.00846183e-01
-4.23270315e-01 -1.00118482e+00 6.04421616e-01 -4.71149147e-01
1.07398522e+00 -1.77103162e+00 1.90014020e-01 -2.95791537e-01
3.83506924e-01 7.05255568e-01 -2.73808807e-01 8.30127120e-01
5.50488412e-01 -1.87239528e-01 -2.30645165e-01 -5.55876493e-01
2.90546536e-01 1.77826703e-01 1.98657289e-01 2.18793228e-02
4.07200456e-01 5.67002058e-01 -1.35281467e+00 -4.87831205e-01
6.30619675e-02 -7.48268366e-02 -4.62713242e-01 1.20095515e+00
-4.37975764e-01 7.45898962e-01 -6.64315462e-01 1.55501157e-01
3.48334700e-01 -3.55627447e-01 2.44548291e-01 3.91252935e-01
7.15088993e-02 8.77454579e-01 -7.31994689e-01 1.75000179e+00
-6.29637957e-01 6.59813404e-01 2.02652752e-01 -9.94419336e-01
1.42920506e+00 7.02666759e-01 1.69640243e-01 -6.25009179e-01
7.86455497e-02 2.17534259e-01 1.52432993e-01 -6.54598296e-01
6.28474951e-01 -9.69724432e-02 -3.32422167e-01 9.50787723e-01
3.46558273e-01 -4.02536154e-01 2.86100417e-01 3.58577609e-01
1.32661605e+00 -3.34298670e-01 1.01130508e-01 3.09710465e-02
1.00731528e+00 3.61701362e-02 3.60600650e-01 4.95005161e-01
-6.49237037e-01 3.25892031e-01 9.86030996e-01 -2.32463241e-01
-8.12989533e-01 -6.99020684e-01 1.74173936e-01 1.60858333e+00
8.28024000e-02 -2.08520919e-01 -1.08715332e+00 -1.20886779e+00
-2.20855549e-01 6.80567145e-01 -4.86083269e-01 -6.39472157e-02
-3.81694883e-01 -3.01575929e-01 8.77163351e-01 1.73179999e-01
1.13944829e+00 -1.38005662e+00 -3.00236434e-01 4.75788206e-01
-6.99510932e-01 -1.21580863e+00 -4.09113109e-01 -7.60492980e-02
-3.63370985e-01 -1.10222578e+00 -7.91266263e-01 -6.93762004e-01
1.72779009e-01 -1.32499173e-01 1.48502326e+00 1.66330740e-01
4.28195715e-01 3.09820682e-01 -7.54796445e-01 -7.37120286e-02
-1.21717501e+00 3.48255903e-01 -6.76925629e-02 -1.71066940e-01
5.53195477e-01 -3.36358219e-01 -4.98323470e-01 1.00922203e+00
-3.45984429e-01 4.71086875e-02 2.80527711e-01 1.27146018e+00
-2.75083125e-01 -3.91160071e-01 1.09738743e+00 -1.12343538e+00
1.67631161e+00 -4.82856035e-01 -1.27118424e-01 4.28593338e-01
-4.99785870e-01 -5.94919808e-02 7.44578660e-01 -5.33371627e-01
-1.38740659e+00 -6.94801450e-01 -3.54756057e-01 1.69732407e-01
-2.26330534e-01 4.69086409e-01 -1.27607748e-01 2.58573323e-01
9.31521773e-01 1.25355303e-01 2.09480032e-01 -3.44267488e-01
2.35092625e-01 1.26988959e+00 4.85063404e-01 -1.03372157e+00
2.74841726e-01 -3.28362375e-01 -6.45486236e-01 -6.36939168e-01
-9.22549963e-01 -6.05772913e-01 -6.01464987e-01 -5.28939068e-01
9.20233905e-01 -7.29179502e-01 -1.15533841e+00 7.30278552e-01
-1.51952684e+00 -9.07838345e-01 1.09921262e-01 3.00197214e-01
-7.09888875e-01 1.09126024e-01 -8.66140902e-01 -8.71947825e-01
-6.01366162e-01 -1.19808948e+00 7.09819376e-01 3.50051671e-01
-6.03930771e-01 -1.20312631e+00 5.49433112e-01 8.52328598e-01
4.78707969e-01 1.04399368e-01 6.76178515e-01 -1.47133398e+00
1.00793742e-01 -2.26081703e-02 -2.15591922e-01 8.97632599e-01
1.54708866e-02 -1.70028642e-01 -1.08991361e+00 4.90720011e-02
-8.62379267e-04 -1.39129174e+00 1.41805604e-01 -3.79994422e-01
6.28189266e-01 -4.39996660e-01 1.94878042e-01 -2.02572256e-01
4.41440403e-01 2.45335743e-01 4.81652498e-01 2.81364590e-01
1.87463418e-01 9.66226041e-01 9.23115909e-01 6.52409792e-01
9.05463815e-01 6.32487535e-01 1.43751487e-01 8.63715187e-02
1.40653640e-01 -2.78518230e-01 4.36437428e-01 1.11279714e+00
2.76086722e-02 -2.98445135e-01 -9.01817679e-01 6.07888997e-01
-1.84525526e+00 -8.73029113e-01 6.43360987e-02 1.72340024e+00
1.60108519e+00 1.10167079e-01 6.58910334e-01 -7.05735907e-02
7.03016281e-01 1.41918555e-01 -4.96376604e-01 -7.84889877e-01
1.64745316e-01 -1.03029512e-01 -4.48384166e-01 5.19571424e-01
-7.85005331e-01 9.55815613e-01 5.77714062e+00 6.07325733e-01
-6.72760129e-01 2.34177247e-01 7.56700993e-01 4.48492110e-01
-4.31388542e-02 -1.95412621e-01 -6.08896315e-01 3.68276536e-01
1.02019060e+00 -3.82830322e-01 3.12927127e-01 9.57028270e-01
4.48012874e-02 6.74113110e-02 -1.33730316e+00 6.18162930e-01
-1.38401613e-01 -8.45279813e-01 -3.28793377e-01 -2.35101089e-01
7.99374580e-01 -1.06341332e-01 -2.72252172e-01 9.39619422e-01
9.89587963e-01 -9.07357335e-01 5.62807135e-02 3.50006938e-01
4.93350208e-01 -5.15422344e-01 1.05671799e+00 7.11043358e-01
-4.87691045e-01 2.80921876e-01 -2.54154712e-01 -2.37270400e-01
1.34924427e-01 2.67169714e-01 -1.52163351e+00 4.26289141e-01
2.90370554e-01 4.71586376e-01 -1.01186112e-01 5.69611430e-01
-4.61486548e-01 6.24425352e-01 1.38557211e-01 -6.62135482e-01
5.15480936e-01 -3.83976549e-01 3.15430015e-01 1.32648516e+00
-1.65287510e-01 3.75080049e-01 4.32807684e-01 7.58822262e-01
-3.94115895e-01 1.39242262e-01 -4.41483706e-01 -1.80034816e-01
7.87317812e-01 1.40622008e+00 1.27216399e-01 -4.51699317e-01
-3.91263932e-01 1.00242090e+00 6.39733493e-01 1.83328521e-02
-5.00149608e-01 -5.30569851e-01 8.43828261e-01 -2.94066310e-01
-3.41315120e-01 3.10295261e-02 -1.13502480e-01 -7.26928592e-01
-1.11934967e-01 -1.52214742e+00 3.37632000e-01 -5.99376976e-01
-1.74421620e+00 9.61763144e-01 -1.90493807e-01 -1.23029113e+00
-7.99412429e-01 -3.82578164e-01 -9.13942158e-01 7.92449355e-01
-1.34676147e+00 -8.46195698e-01 -6.40213072e-01 7.96553850e-01
8.24013293e-01 -5.54855168e-01 1.03189290e+00 1.39520034e-01
-5.33565104e-01 7.42644846e-01 -1.79028690e-01 7.15360820e-01
1.10653210e+00 -1.55921268e+00 2.72396713e-01 1.28281593e-01
-4.06994551e-01 4.35307711e-01 7.57855892e-01 -2.77194142e-01
-9.62703466e-01 -6.44453049e-01 8.55085552e-01 -6.67376518e-01
8.15830171e-01 -1.10610403e-01 -1.17327189e+00 2.74274141e-01
7.73166537e-01 -4.66233611e-01 8.04278374e-01 4.58989680e-01
-2.97210187e-01 5.93893304e-02 -1.29428387e+00 3.23910594e-01
7.63411224e-01 -8.07873666e-01 -9.51941550e-01 5.62207878e-01
6.39509618e-01 -6.82569921e-01 -1.27022970e+00 1.86028570e-01
3.93433183e-01 -1.20544302e+00 5.02038717e-01 -8.68162811e-01
8.24444592e-01 2.54246503e-01 2.03049123e-01 -1.92282689e+00
3.37590814e-01 -1.02311373e+00 2.58083761e-01 1.69095182e+00
5.84325910e-01 -5.47658443e-01 5.48577726e-01 7.97889352e-01
-3.13153535e-01 -6.31850898e-01 -7.10649431e-01 -6.77283645e-01
3.12085956e-01 1.86841115e-02 6.98091567e-01 1.24694431e+00
7.32103467e-01 1.01208937e+00 -4.24943626e-01 -3.56456548e-01
1.73053741e-01 -3.44627202e-02 1.48368263e+00 -1.35302866e+00
-1.84268117e-01 -6.02056742e-01 -2.30374094e-02 -1.42577362e+00
3.39130998e-01 -4.65396136e-01 3.73760700e-01 -1.59825516e+00
1.27043709e-01 -4.26296055e-01 1.75538301e-01 2.26344526e-01
-3.63868684e-01 -3.47956479e-01 -6.29143696e-03 -8.44383240e-03
-1.18574297e+00 9.62040782e-01 1.70180583e+00 -1.11352682e-01
-2.57634550e-01 3.57216984e-01 -6.07302129e-01 5.80222607e-01
6.79524243e-01 -2.92983383e-01 -5.49298465e-01 -3.12184036e-01
-2.32367188e-01 5.74523926e-01 -2.72969715e-02 -7.58815408e-01
3.23597819e-01 -3.16967905e-01 -3.92865449e-01 -2.53683269e-01
4.30127412e-01 -2.88326949e-01 -5.35241723e-01 8.75161663e-02
-8.26319337e-01 6.06728159e-02 -3.04866970e-01 9.11151171e-02
-6.68807685e-01 -4.12545204e-01 7.79474378e-01 -1.37101650e-01
-4.14937377e-01 2.96505517e-03 -1.92953423e-01 8.18918109e-01
8.28378618e-01 2.01568708e-01 -9.14393842e-01 -9.07506287e-01
-3.59352350e-01 7.65417457e-01 1.22560441e-01 3.42499167e-01
3.88462245e-01 -1.18441534e+00 -1.22455561e+00 -3.27215254e-01
3.74695092e-01 1.12978943e-01 1.40041888e-01 6.79517627e-01
-4.11664069e-01 2.78665453e-01 -3.14981341e-01 -5.75022995e-01
-1.32878923e+00 -1.21706896e-01 4.09515142e-01 -8.02017987e-01
-2.88210899e-01 1.00661087e+00 -8.83627236e-02 -1.10455239e+00
4.23267394e-01 -2.28939187e-02 -3.65489542e-01 2.76682764e-01
5.84087014e-01 4.86066163e-01 -6.22488037e-02 -3.26470375e-01
1.73477996e-02 -4.43760939e-02 -1.38383403e-01 -2.69159466e-01
1.16843462e+00 -1.31694004e-01 4.71419469e-02 4.63898927e-01
1.08035529e+00 -3.90088767e-01 -1.26218259e+00 -6.20099902e-01
2.00069442e-01 -2.72733569e-01 -5.95954716e-01 -1.24919116e+00
-5.13541937e-01 9.91953492e-01 -2.87986230e-02 7.29398191e-01
7.07867980e-01 -4.20300104e-02 7.25952804e-01 8.15015018e-01
2.00862184e-01 -1.39146125e+00 9.02125478e-01 1.17178226e+00
1.09318376e+00 -1.63569462e+00 -4.27987576e-01 -5.03582582e-02
-1.59404790e+00 9.86809552e-01 1.17360389e+00 2.12862521e-01
1.18406393e-01 -6.57947063e-02 4.34039980e-01 -2.61150241e-01
-1.13958764e+00 -1.13307029e-01 9.11083221e-02 5.82486272e-01
7.71061301e-01 2.16912001e-01 -3.55827540e-01 7.02876449e-01
-3.20150197e-01 -1.02555782e-01 5.05263746e-01 4.28259254e-01
-4.75724310e-01 -1.19273579e+00 1.20815888e-01 4.41885442e-02
-1.16645269e-01 2.48461470e-01 -1.01760197e+00 7.65318632e-01
-6.92863047e-01 1.53516483e+00 -2.54535407e-01 -5.94132960e-01
7.70321906e-01 2.21645787e-01 1.19936438e-02 -7.29319990e-01
-1.25034463e+00 -5.25229633e-01 9.49091494e-01 -2.63037920e-01
-6.34464502e-01 -4.25169677e-01 -9.34787154e-01 -6.14305496e-01
-3.19090754e-01 7.56721139e-01 3.81825745e-01 1.13839781e+00
-7.98194706e-02 4.08459932e-01 1.16499019e+00 -2.96768934e-01
-1.04468489e+00 -1.79608548e+00 -2.75093317e-01 8.18783641e-01
1.41531110e-01 -6.18739605e-01 -3.30166757e-01 -3.74374330e-01]
|
[12.725916862487793, 8.114681243896484]
|
1b1e016e-823d-475e-bc1b-5e37efd5503c
|
mobile-microphone-array-speech-detection-and
|
2106.14787
| null |
https://arxiv.org/abs/2106.14787v1
|
https://arxiv.org/pdf/2106.14787v1.pdf
|
Mobile Microphone Array Speech Detection and Localization in Diverse Everyday Environments
|
Joint sound event localization and detection (SELD) is an integral part of developing context awareness into communication interfaces of mobile robots, smartphones, and home assistants. For example, an automatic audio focus for video capture on a mobile phone requires robust detection of relevant acoustic events around the device and their direction. Existing SELD approaches have been evaluated using material produced in controlled indoor environments, or the audio is simulated by mixing isolated sounds to different spatial locations. This paper studies SELD of speech in diverse everyday environments, where the audio corresponds to typical usage scenarios of handheld mobile devices. In order to allow weighting the relative importance of localization vs. detection, we will propose a two-stage hierarchical system, where the first stage is to detect the target events, and the second stage is to localize them. The proposed method utilizes convolutional recurrent neural network (CRNN) and is evaluated on a database of manually annotated microphone array recordings from various acoustic conditions. The array is embedded in a contemporary mobile phone form factor. The obtained results show good speech detection and localization accuracy of the proposed method in contrast to a non-hierarchical flat classification model.
|
['Antti Eronen', 'Archontis Politis', 'Tuomas Virtanen', 'Eemi Fagerlund', 'Aapo Hakala', 'Emre Cakir', 'Pasi Pertilä']
|
2021-06-28
| null | null | null | null |
['sound-event-localization-and-detection']
|
['audio']
|
[ 5.60717046e-01 -2.18839645e-01 4.81637955e-01 -7.79973492e-02
-1.09701943e+00 -3.77169549e-01 4.94167060e-01 3.04482132e-01
-5.86908638e-01 4.43796575e-01 3.28538269e-01 -1.12475686e-01
-7.78867155e-02 -3.78117830e-01 -5.44607341e-01 -7.55622923e-01
-1.09004378e-01 5.73181883e-02 5.38881242e-01 2.02731371e-01
2.58303076e-01 4.15836483e-01 -2.06293845e+00 3.31870258e-01
3.00445348e-01 1.16083753e+00 9.39451337e-01 1.20044041e+00
8.22269842e-02 6.45081937e-01 -1.06397104e+00 3.13693106e-01
-2.74413407e-01 -2.51923829e-01 -2.48811722e-01 -3.43370773e-02
2.40085483e-01 -2.55650599e-02 1.73348598e-02 1.01211369e+00
9.38128591e-01 2.38668159e-01 5.49842417e-01 -1.08552754e+00
1.34134978e-01 7.04730034e-01 4.30742418e-03 3.71733159e-01
6.50397003e-01 -3.67690802e-01 7.88963854e-01 -1.19024956e+00
1.20782331e-01 1.02229583e+00 8.21295917e-01 8.04891959e-02
-7.63232052e-01 -5.89596331e-01 -1.91360209e-02 3.70896131e-01
-1.54535079e+00 -8.14763904e-01 9.32006717e-01 -4.68043596e-01
1.11096466e+00 1.34315848e-01 2.91286796e-01 1.12354958e+00
1.72481731e-01 3.62889141e-01 6.71361446e-01 -8.20111394e-01
5.80530286e-01 4.90167171e-01 9.27607566e-02 2.67373711e-01
-4.96078245e-02 -1.39438108e-01 -6.77722275e-01 -1.73325539e-01
3.17688257e-01 -7.05081001e-02 -2.48273402e-01 -2.61026639e-02
-1.20185685e+00 3.70370775e-01 1.64864540e-01 7.68689573e-01
-6.35728359e-01 -1.19765534e-03 4.36605245e-01 -1.61993615e-02
1.93692088e-01 -7.38797057e-03 -2.05438584e-01 -3.27610701e-01
-9.30552959e-01 -1.54904038e-01 9.21740413e-01 8.71416867e-01
2.77309984e-01 1.70649871e-01 6.32092878e-02 1.09548390e+00
5.40021896e-01 3.91317368e-01 8.29174221e-01 -6.10918641e-01
5.13671458e-01 4.95645180e-02 3.40235651e-01 -1.07278073e+00
-4.98405665e-01 -5.21984696e-01 -4.96077985e-01 2.73412950e-02
2.39656001e-01 -3.89427125e-01 -4.90079850e-01 1.48437500e+00
3.01640511e-01 4.33632851e-01 9.51190144e-02 6.68850780e-01
6.36820495e-01 9.47062910e-01 6.36469498e-02 -3.83324951e-01
1.46896076e+00 -7.78806388e-01 -1.14166069e+00 -2.47559234e-01
1.05482884e-01 -9.42377985e-01 9.86904979e-01 6.68443680e-01
-8.25324714e-01 -9.03935850e-01 -1.31466019e+00 1.88090846e-01
-6.15400612e-01 6.82267189e-01 -2.80196518e-02 8.63151670e-01
-1.02886379e+00 6.71230778e-02 -7.44527817e-01 -5.58082938e-01
-1.57316074e-01 4.02658671e-01 -9.17281136e-02 3.59394282e-01
-1.13434148e+00 4.37268585e-01 -5.53421639e-02 3.70149106e-01
-9.62411463e-01 -4.90725599e-02 -7.83664167e-01 2.33725816e-01
6.04249053e-02 -2.07609516e-02 1.41120148e+00 -7.13417470e-01
-1.73381579e+00 3.47931951e-01 -4.67858076e-01 -6.05092168e-01
1.54460177e-01 -5.01637399e-01 -7.54635334e-01 3.02764595e-01
2.02946767e-01 1.36383638e-01 1.07096457e+00 -1.06910920e+00
-1.00834429e+00 -2.00830936e-01 -3.10123980e-01 3.69463950e-01
-3.96454126e-01 2.41914749e-01 -2.60763466e-01 -4.77672637e-01
1.39146924e-01 -6.52615130e-01 1.29739791e-01 -5.44581115e-01
-2.96560019e-01 -7.33230487e-02 1.02493775e+00 -7.17605889e-01
1.26229644e+00 -2.28836298e+00 -2.46824160e-01 2.16168508e-01
-3.07370126e-01 1.65726468e-01 2.85581440e-01 3.82064998e-01
-3.82742286e-03 -4.18171257e-01 6.26350939e-02 -7.20634460e-01
-1.58547498e-02 -2.66946018e-01 -3.52785259e-01 2.00619265e-01
-1.00662231e-01 -3.83733884e-02 -6.83810532e-01 -4.13704187e-01
4.95747089e-01 7.41368592e-01 -3.44124347e-01 3.30619603e-01
2.00576901e-01 2.91022569e-01 -1.85633212e-01 5.79318047e-01
2.24864244e-01 2.58612543e-01 1.68597445e-01 -1.46576405e-01
-3.00725788e-01 6.31461680e-01 -1.64391673e+00 1.32660997e+00
-9.12210464e-01 8.70626450e-01 4.96103346e-01 -8.03721130e-01
9.60540652e-01 7.76825249e-01 2.03374788e-01 -4.35320109e-01
1.81457296e-01 4.63546544e-01 -2.20429897e-01 -6.68140411e-01
7.44606555e-01 3.13743919e-01 2.73529887e-02 2.12172419e-01
2.45270487e-02 4.61982898e-02 -1.21869363e-01 -1.81403413e-01
1.19668090e+00 -1.27348378e-01 4.27881092e-01 1.54106645e-02
7.67954946e-01 -4.39006388e-01 2.93129951e-01 9.32505906e-01
-3.58781308e-01 6.38475120e-01 -3.49796861e-02 1.24073766e-01
-4.70238626e-01 -1.05680394e+00 3.83098088e-02 1.33971131e+00
5.99350594e-02 -2.68274754e-01 -8.81598592e-01 -2.19221786e-01
-5.90917706e-01 6.50856793e-01 -1.05813526e-01 4.53426503e-02
-6.54813886e-01 -4.60898310e-01 6.02093339e-01 2.96030343e-01
5.20068824e-01 -1.31608605e+00 -8.10390055e-01 4.95018750e-01
-3.11145037e-01 -1.34014487e+00 -3.43993723e-01 5.89932144e-01
-3.40221792e-01 -6.92823410e-01 -4.51059371e-01 -1.36551380e+00
2.71949977e-01 3.76540720e-01 7.05925167e-01 -4.87060785e-01
-2.11162001e-01 7.33395576e-01 -5.04076838e-01 -5.99890649e-01
-2.70881861e-01 5.65439165e-02 3.66760522e-01 3.48175377e-01
3.30904424e-01 -8.39972377e-01 -5.73020637e-01 5.58498085e-01
-5.96153140e-01 -6.25078559e-01 4.75730658e-01 5.96912265e-01
3.93819809e-01 4.82908547e-01 8.50985229e-01 -2.54545271e-01
8.19149435e-01 -5.83068967e-01 -3.64396602e-01 7.98995048e-02
5.87404445e-02 -6.10252798e-01 5.55331767e-01 -6.41168892e-01
-1.15457451e+00 4.00262296e-01 -4.99704778e-01 -3.67001034e-02
-5.87020397e-01 2.61920989e-01 -5.55764973e-01 2.26873130e-01
6.41705632e-01 1.40147194e-01 -5.97671211e-01 -3.47893059e-01
-1.27305552e-01 1.45570350e+00 6.37957752e-01 -2.90635824e-01
9.89147052e-02 1.73062637e-01 -4.22227770e-01 -1.47238457e+00
-1.34508505e-01 -7.45022893e-01 -3.28109473e-01 -3.85173053e-01
8.11057210e-01 -9.61097836e-01 -5.31213999e-01 5.01213133e-01
-1.17104399e+00 -1.55593619e-01 5.43810800e-02 9.29279745e-01
-4.79427159e-01 6.06185868e-02 -4.91013080e-01 -1.38051522e+00
-1.82613268e-01 -1.36300576e+00 1.41383779e+00 1.32250592e-01
-3.85109395e-01 -6.77900612e-01 -1.00949340e-01 1.37998208e-01
4.21925962e-01 -2.12378234e-01 4.55996454e-01 -7.79233396e-01
-2.80650109e-01 -5.08092701e-01 2.32166424e-01 3.01452428e-01
5.26910841e-01 -1.65751427e-01 -1.50603592e+00 1.04475141e-01
4.53169197e-01 -1.24917202e-01 3.59027386e-01 7.58999765e-01
8.04270208e-01 -1.86153516e-01 -4.50034887e-01 -2.59667728e-02
1.02735960e+00 9.23569381e-01 4.29594725e-01 1.32742897e-01
3.88366282e-01 7.03546941e-01 6.63921237e-01 4.05142635e-01
1.45168170e-01 8.66126597e-01 2.61569589e-01 2.24660441e-01
-1.34890243e-01 -1.06470481e-01 7.06343710e-01 1.04788995e+00
4.61801946e-01 -5.18644989e-01 -7.69712150e-01 6.38977468e-01
-1.56661236e+00 -8.20827782e-01 2.06612945e-01 2.25111914e+00
4.20996398e-01 3.05292636e-01 2.19172630e-02 7.19259262e-01
1.20848060e+00 -5.90066575e-02 -1.65997133e-01 -3.69468540e-01
1.49850041e-01 2.10011408e-01 8.36431310e-02 8.30578685e-01
-1.19595301e+00 5.34523249e-01 5.70173311e+00 8.38873744e-01
-1.33705330e+00 3.84782255e-01 2.33694240e-01 -3.34230848e-02
3.84341568e-01 -5.63692272e-01 -8.58814836e-01 4.92956489e-01
1.18654144e+00 5.97774148e-01 2.10280225e-01 9.25131917e-01
5.77424884e-01 -3.05006981e-01 -1.11647928e+00 1.19824278e+00
1.33227989e-01 -7.58489728e-01 -4.19050694e-01 -2.66577512e-01
1.31296024e-01 -6.05153851e-02 1.40458897e-01 1.87555894e-01
-3.40327173e-01 -7.03558624e-01 9.35811102e-01 5.14083982e-01
4.74507809e-01 -6.24811947e-01 6.44347250e-01 2.46031627e-01
-1.49250925e+00 -2.48861626e-01 -2.37929225e-02 -2.16242209e-01
3.05358201e-01 4.72337574e-01 -1.37356293e+00 8.51289406e-02
9.35913265e-01 2.56333798e-01 -3.04029465e-01 1.04483974e+00
2.21822504e-02 8.63785863e-01 -6.25143409e-01 -4.60805565e-01
-6.80675730e-02 1.85418934e-01 8.04693341e-01 1.60540986e+00
7.20363975e-01 -3.79452199e-01 -7.37787485e-02 3.86188507e-01
7.05144927e-02 1.43162534e-01 -8.16476226e-01 2.23243505e-01
8.64367187e-01 1.16249406e+00 -1.06849039e+00 -1.35010287e-01
-2.79622138e-01 9.46732879e-01 -1.96984977e-01 3.75539303e-01
-7.17270374e-01 -9.09042716e-01 3.08986038e-01 -9.93250087e-02
5.02186120e-01 -2.33800739e-01 -1.23533504e-02 -5.27144074e-01
2.46480569e-01 -5.86423337e-01 -3.07001118e-02 -1.03875935e+00
-7.42980421e-01 8.08104932e-01 -1.79034680e-01 -1.37993300e+00
-5.42103469e-01 -5.88017583e-01 -7.15845108e-01 6.78987086e-01
-9.44832146e-01 -7.67908454e-01 -1.89110115e-01 5.56012809e-01
9.96463358e-01 -3.15280110e-01 9.14283872e-01 7.22080052e-01
-4.15162355e-01 5.09848595e-01 1.11870833e-01 -1.47736520e-01
6.41216278e-01 -1.11265886e+00 8.61252695e-02 9.82187688e-01
2.91418344e-01 6.17173851e-01 9.16073859e-01 -5.88795245e-01
-9.81569290e-01 -9.59240258e-01 1.06826138e+00 -2.52714634e-01
5.10190129e-01 -7.62898147e-01 -5.87784708e-01 3.96305084e-01
2.10859179e-01 -1.33153245e-01 5.39775670e-01 -1.53279543e-01
9.16875452e-02 -4.14229214e-01 -1.08017707e+00 4.88096654e-01
7.48347461e-01 -8.37064624e-01 -6.07126117e-01 1.73389599e-01
6.75457120e-01 -9.81702954e-02 -3.98664653e-01 2.38472581e-01
7.54601598e-01 -9.57282186e-01 9.09125209e-01 3.90132427e-01
-1.99532762e-01 -6.60560608e-01 -6.48594320e-01 -1.11117864e+00
1.87525377e-01 -7.08509445e-01 -1.80812776e-02 1.56423092e+00
4.20233399e-01 -4.39603776e-01 4.69417155e-01 -1.38065204e-01
-2.90487856e-01 -2.47275561e-01 -1.10030162e+00 -3.16110551e-01
-8.34897697e-01 -9.59446371e-01 2.96264380e-01 3.83675307e-01
2.32692491e-02 5.65170228e-01 -2.51138836e-01 7.65642285e-01
2.29105100e-01 -5.19873321e-01 4.41148221e-01 -9.60350573e-01
-3.09109896e-01 -3.17615569e-01 -5.53808689e-01 -9.64499891e-01
-1.39151350e-01 -1.78905621e-01 4.93059307e-01 -1.25234592e+00
-4.99382019e-01 -2.88809359e-01 -3.32688332e-01 -1.68878779e-01
1.59623012e-01 2.98756987e-01 -3.20777446e-01 -5.66772884e-03
-6.01822555e-01 3.83709878e-01 2.28813201e-01 -9.83349755e-02
-5.73096097e-01 4.94801581e-01 -1.74435258e-01 9.76217806e-01
7.81076729e-01 -2.91669160e-01 -6.29291832e-01 -1.15219362e-01
1.17062286e-01 2.25555465e-01 3.43698353e-01 -1.54090714e+00
7.50781357e-01 6.46055818e-01 2.04259560e-01 -8.07470441e-01
6.85777009e-01 -1.20116544e+00 4.04473469e-02 2.18888626e-01
-4.60363299e-01 2.40183678e-02 1.05653644e-01 6.68488920e-01
-4.19856608e-01 -3.89448375e-01 4.65640396e-01 5.45559749e-02
-6.23938441e-01 -6.76914275e-01 -9.88893628e-01 -5.63661456e-01
7.62142539e-01 -2.83009589e-01 1.19192056e-01 -5.99477828e-01
-1.02406049e+00 -3.65983903e-01 -3.49188089e-01 4.64109153e-01
7.22672343e-01 -1.14306092e+00 -7.76852369e-02 2.91500419e-01
-5.97154312e-02 -8.64610225e-02 1.15594096e-01 6.28188908e-01
-4.31727827e-01 4.59509760e-01 1.35777310e-01 -7.38222420e-01
-1.44485354e+00 2.01318398e-01 4.45008039e-01 1.46935835e-01
-1.93254024e-01 9.34882462e-01 7.95577243e-02 -2.41856888e-01
1.06533384e+00 -6.86445355e-01 -6.89778626e-01 2.57329017e-01
7.25670457e-01 6.20973468e-01 4.94956017e-01 -7.36599863e-01
-4.18223083e-01 5.27359843e-01 9.56692025e-02 -5.76794505e-01
8.61365438e-01 -5.33757865e-01 1.99717626e-01 9.61562335e-01
1.10328019e+00 4.07767713e-01 -8.72640729e-01 -1.44534752e-01
1.45313814e-01 1.76220074e-01 3.68067063e-02 -6.95374548e-01
-4.73054111e-01 9.21057403e-01 1.20618498e+00 4.72720176e-01
1.24308586e+00 -2.12190837e-01 5.00489056e-01 6.36262894e-01
6.19837105e-01 -1.18184388e+00 8.34487304e-02 4.03975070e-01
8.04530442e-01 -9.31458056e-01 -4.80813652e-01 -2.47712970e-01
-4.28558379e-01 9.75892007e-01 3.19232792e-01 1.59412585e-02
9.00091827e-01 4.60890740e-01 3.66054773e-02 9.67147425e-02
-4.00961488e-01 -1.40084743e-01 -8.71796720e-03 7.39840567e-01
3.49326283e-01 -1.26469046e-01 1.21234171e-01 8.24046910e-01
-3.67254883e-01 -2.63349444e-01 3.18909019e-01 1.06469512e+00
-6.96248710e-01 -5.13522983e-01 -8.41354072e-01 1.41688988e-01
-8.09718490e-01 -9.58734229e-02 -2.62401909e-01 3.87125075e-01
3.88312280e-01 1.68809152e+00 1.39932424e-01 -5.80597222e-01
4.36855406e-01 3.04309070e-01 -1.89702492e-02 -5.30241847e-01
-5.72392106e-01 5.02704561e-01 1.30117714e-01 -3.62831801e-01
-5.11659205e-01 -8.43751609e-01 -1.11876976e+00 7.52065182e-01
-3.80551785e-01 4.77277547e-01 1.15657961e+00 9.24964130e-01
1.64977044e-01 1.00257993e+00 7.87660062e-01 -1.40474284e+00
-1.71675123e-02 -1.22170699e+00 -5.63798368e-01 -1.88643843e-01
6.98031008e-01 -7.01666415e-01 -6.38216972e-01 2.20322624e-01]
|
[14.915132522583008, 5.643947124481201]
|
361386fb-2e43-416f-a45f-54ae5587dbb6
|
hashing-on-nonlinear-manifolds
|
1412.0826
| null |
http://arxiv.org/abs/1412.0826v1
|
http://arxiv.org/pdf/1412.0826v1.pdf
|
Hashing on Nonlinear Manifolds
|
Learning based hashing methods have attracted considerable attention due to
their ability to greatly increase the scale at which existing algorithms may
operate. Most of these methods are designed to generate binary codes preserving
the Euclidean similarity in the original space. Manifold learning techniques,
in contrast, are better able to model the intrinsic structure embedded in the
original high-dimensional data. The complexities of these models, and the
problems with out-of-sample data, have previously rendered them unsuitable for
application to large-scale embedding, however. In this work, how to learn
compact binary embeddings on their intrinsic manifolds is considered. In order
to address the above-mentioned difficulties, an efficient, inductive solution
to the out-of-sample data problem, and a process by which non-parametric
manifold learning may be used as the basis of a hashing method is proposed. The
proposed approach thus allows the development of a range of new hashing
techniques exploiting the flexibility of the wide variety of manifold learning
approaches available. It is particularly shown that hashing on the basis of
t-SNE outperforms state-of-the-art hashing methods on large-scale benchmark
datasets, and is very effective for image classification with very short code
lengths. The proposed hashing framework is shown to be easily improved, for
example, by minimizing the quantization error with learned orthogonal
rotations. In addition, a supervised inductive manifold hashing framework is
developed by incorporating the label information, which is shown to greatly
advance the semantic retrieval performance.
|
['Anton Van Den Hengel', 'Qinfeng Shi', 'Chunhua Shen', 'Fumin Shen', 'Zhenmin Tang', 'Heng Tao Shen']
|
2014-12-02
| null | null | null | null |
['semantic-retrieval']
|
['natural-language-processing']
|
[-5.18637113e-02 -7.33189732e-02 -4.84759957e-01 -2.99473405e-01
-1.02377963e+00 -5.50004423e-01 6.86494648e-01 4.12648171e-01
-4.66730177e-01 5.07984102e-01 -2.82450188e-02 5.93952052e-02
-3.49562407e-01 -9.38588262e-01 -4.62090522e-01 -9.54580843e-01
-3.14457029e-01 5.08699596e-01 3.06791123e-02 -2.41414428e-01
6.65957391e-01 6.58989847e-01 -2.01968360e+00 -2.74301618e-02
3.95231009e-01 8.65134060e-01 6.85838386e-02 4.45544183e-01
-2.08190486e-01 2.43164331e-01 -2.44656354e-01 -1.62816241e-01
1.42133430e-01 -2.50659287e-01 -7.57462680e-01 4.32008430e-02
4.31170255e-01 -1.96831644e-01 -5.43843806e-01 1.04036927e+00
4.57286537e-01 1.67720944e-01 9.45077956e-01 -1.33644938e+00
-6.32350147e-01 1.92944691e-01 -1.40201241e-01 -1.05937205e-01
5.14074385e-01 -5.24323344e-01 1.33170283e+00 -1.19754708e+00
5.66717803e-01 1.15958202e+00 6.11627996e-01 3.22760552e-01
-1.33075404e+00 -2.59057134e-01 -6.57045722e-01 4.23869550e-01
-1.78855014e+00 -2.71316409e-01 1.05205584e+00 -4.26011235e-01
7.33681142e-01 1.83495432e-01 3.38656157e-01 4.68647897e-01
1.07683375e-01 6.11821890e-01 1.05207086e+00 -7.34242141e-01
3.56207132e-01 4.20258254e-01 3.26291844e-02 8.05927932e-01
3.05433452e-01 9.40592661e-02 -4.04492110e-01 -4.96024579e-01
5.61331868e-01 1.68290555e-01 -1.00169241e-01 -1.22299516e+00
-1.18026638e+00 1.34862745e+00 7.67674029e-01 5.87981105e-01
-2.62742843e-02 8.08674246e-02 5.14690578e-01 2.49826506e-01
2.69266844e-01 4.92280155e-01 2.81523857e-02 -4.24169898e-02
-1.00894022e+00 2.83410043e-01 7.81721234e-01 8.22905958e-01
1.21029627e+00 -3.08614612e-01 5.07306874e-01 7.28586853e-01
3.67688775e-01 4.07526940e-01 7.33395875e-01 -6.42980635e-01
3.01546395e-01 6.38316453e-01 -4.72640358e-02 -1.35501003e+00
-2.88852274e-01 1.17573932e-01 -6.71740890e-01 1.06778838e-01
2.68021315e-01 6.76564038e-01 -5.03333867e-01 1.73586798e+00
4.78963882e-01 -1.07415833e-01 2.81149536e-01 6.37536824e-01
2.95386612e-01 6.64585710e-01 -3.40897232e-01 -2.60681827e-02
1.33494842e+00 -5.93619287e-01 -5.55905640e-01 4.31981504e-01
1.17267966e+00 -7.91527808e-01 8.87740731e-01 2.00793117e-01
-8.30114782e-01 -5.37514687e-01 -1.27672303e+00 -1.83419436e-01
-8.46859753e-01 -1.11055158e-01 4.03502136e-01 8.96950185e-01
-1.18276751e+00 6.45196855e-01 -6.97521091e-01 -5.96430302e-01
1.18401289e-01 6.53607786e-01 -6.04480267e-01 -2.89959013e-01
-1.25601494e+00 9.82235193e-01 5.68943322e-01 -9.67003331e-02
-4.86669838e-01 -2.47951940e-01 -1.21285617e+00 2.18490027e-02
-7.81412199e-02 -3.11528474e-01 6.92894697e-01 -4.36215609e-01
-1.03734004e+00 9.02098536e-01 -2.35761609e-03 -4.62595165e-01
9.74354893e-02 1.76136494e-01 -2.77181715e-01 7.05876887e-01
1.10917322e-01 9.12505984e-01 1.12304389e+00 -9.57968295e-01
-2.32892200e-01 -5.95516562e-01 -9.58810374e-02 2.01236710e-01
-8.82240713e-01 -2.45945469e-01 -7.51028256e-03 -5.72230041e-01
1.63934782e-01 -1.18103075e+00 -5.49744908e-03 2.53637940e-01
3.81263122e-02 -2.26200402e-01 1.07998633e+00 -2.20221028e-01
1.18167889e+00 -2.33670855e+00 5.55394709e-01 4.32867855e-01
-7.22028464e-02 2.41472453e-01 -4.67079058e-02 8.21567178e-01
-1.74373984e-01 -6.28879145e-02 -3.34626526e-01 -3.60991418e-01
1.66533396e-01 4.43186253e-01 -3.77924263e-01 7.64514148e-01
1.85987264e-01 6.32417381e-01 -7.78962255e-01 -8.10148299e-01
4.03508633e-01 7.96111345e-01 -5.36697149e-01 2.15977848e-01
3.17646861e-01 -9.23946649e-02 -3.17721367e-01 4.23754692e-01
4.40552533e-01 -1.19040728e-01 -2.49437094e-02 -1.57006904e-01
1.41557790e-02 1.18528560e-01 -1.38584733e+00 1.88680875e+00
-2.93950915e-01 4.20130640e-01 -1.09335572e-01 -1.29501808e+00
1.17525351e+00 4.52781379e-01 7.13093281e-01 -3.70602250e-01
-1.05633959e-01 5.67433000e-01 -3.07685405e-01 -1.71242818e-01
7.79770672e-01 -4.10687089e-01 -2.33071461e-01 5.21363497e-01
9.43338349e-02 -2.25520864e-01 1.28224522e-01 2.30018377e-01
7.40389109e-01 -1.54341593e-01 4.82184798e-01 -3.51318687e-01
8.82918477e-01 -8.30798671e-02 -2.45037768e-02 2.19152004e-01
-1.47375241e-01 5.49247742e-01 1.09925054e-01 -4.52174366e-01
-1.30365133e+00 -1.01695383e+00 -6.22981369e-01 7.15248704e-01
2.81349331e-01 -6.62407994e-01 -8.91982019e-01 -4.44812596e-01
1.98227435e-01 1.89199567e-01 -7.27010071e-01 -3.32131535e-01
-3.63612026e-01 -6.44973338e-01 5.15177250e-01 3.07438612e-01
1.66100994e-01 -8.79574060e-01 -5.85404873e-01 2.92156130e-01
-1.85396880e-01 -8.27553391e-01 -3.54597837e-01 1.83750972e-01
-1.29215205e+00 -1.01445806e+00 -7.67955184e-01 -9.62658286e-01
6.64375961e-01 5.87326765e-01 4.95760381e-01 1.35431394e-01
-6.80805385e-01 8.17855120e-01 -4.81382072e-01 1.30103961e-01
-5.08087039e-01 2.69765615e-01 4.27437425e-01 9.72739086e-02
6.25907183e-01 -5.33536732e-01 -4.57246780e-01 5.12275159e-01
-1.44903338e+00 -6.57096267e-01 4.89281148e-01 1.19863844e+00
5.27548313e-01 1.56784743e-01 6.04398847e-01 -3.92142713e-01
2.52383858e-01 -3.74270707e-01 -3.98219943e-01 2.66289487e-02
-8.20823371e-01 5.32142937e-01 6.41444564e-01 -3.66052419e-01
-2.92413890e-01 4.18917537e-02 8.30378458e-02 -3.04912269e-01
-3.94915938e-02 2.99228609e-01 -6.38616085e-02 -6.00593030e-01
6.05675220e-01 4.39926058e-01 5.60696006e-01 -3.68026555e-01
5.87790489e-01 9.42440867e-01 2.34078661e-01 -5.42825162e-01
1.05514574e+00 6.20849013e-01 3.37738097e-01 -1.08473015e+00
-3.46165895e-01 -9.64279950e-01 -1.04160643e+00 1.63145646e-01
7.95128167e-01 -8.65426302e-01 -1.95686832e-01 2.07672641e-01
-6.92139268e-01 3.32333773e-01 -3.05345416e-01 5.33560693e-01
-1.10568202e+00 7.96121836e-01 -5.58198988e-01 -7.04668164e-01
-1.62270963e-01 -1.18756461e+00 1.16431940e+00 -1.46318853e-01
-2.00154871e-01 -1.16140711e+00 3.17284495e-01 2.45574549e-01
2.12862566e-01 -2.34729517e-02 1.08199346e+00 -6.39376700e-01
-5.40784597e-01 -4.72381741e-01 -1.93190742e-02 4.17755753e-01
2.60334790e-01 -2.22250223e-01 -8.92237484e-01 -7.90651739e-01
1.58155665e-01 -5.36552012e-01 5.28887272e-01 -3.23185064e-02
7.62850642e-01 -9.19823945e-02 -2.69109339e-01 4.07736987e-01
1.71044397e+00 -2.06301913e-01 7.18954086e-01 5.02945483e-01
5.10037839e-01 7.03433871e-01 8.58701289e-01 4.31395203e-01
2.45292217e-01 1.02947211e+00 4.73301649e-01 4.47880849e-02
1.17077656e-01 -2.36147612e-01 3.32988024e-01 1.14244473e+00
3.54244232e-01 3.84133428e-01 -6.31118059e-01 7.02988684e-01
-1.55575454e+00 -1.07986176e+00 1.46378592e-01 2.43574500e+00
6.51937008e-01 -9.97742265e-02 2.99048811e-01 7.14959800e-01
7.92454481e-01 1.60252437e-01 -1.42910898e-01 -5.19066393e-01
-1.60355587e-02 1.45165622e-01 2.58013338e-01 4.47090447e-01
-1.22052789e+00 5.26372373e-01 6.12391996e+00 8.37246299e-01
-9.11906123e-01 -1.49421141e-01 8.42722356e-02 3.97957891e-01
-2.11947829e-01 2.90448014e-02 -8.41571987e-01 4.05328691e-01
1.07387745e+00 -8.02034959e-02 4.70785975e-01 9.79488194e-01
-2.66327113e-01 4.06774022e-02 -1.28302193e+00 1.29640472e+00
4.87044722e-01 -1.11647820e+00 3.39372784e-01 4.56349522e-01
3.92733127e-01 -3.39162588e-01 3.46425265e-01 1.12664469e-01
-5.14210999e-01 -9.55905676e-01 4.07817423e-01 5.31241409e-02
8.59225571e-01 -8.16591024e-01 6.70068026e-01 1.59070835e-01
-1.44676602e+00 -1.61107644e-01 -7.04851806e-01 3.55397873e-02
-7.05361962e-02 2.35326171e-01 -8.79619002e-01 6.54133320e-01
3.93696606e-01 7.14822710e-01 -6.85462415e-01 1.12488711e+00
2.66959697e-01 6.33170977e-02 -4.03712064e-01 -1.40602604e-01
4.31048483e-01 -3.08498740e-01 4.52116877e-01 1.07899690e+00
5.00135541e-01 -2.49199703e-01 -1.85621642e-02 5.14617503e-01
1.74890593e-01 4.99674350e-01 -1.14558017e+00 -5.25777005e-02
4.04861331e-01 1.23535156e+00 -6.66179836e-01 -2.35753596e-01
-3.82696599e-01 8.40360820e-01 3.37965339e-01 -1.06903084e-01
-5.24623215e-01 -6.72048807e-01 5.80501378e-01 1.16740510e-01
6.11369610e-01 -4.62517858e-01 1.16084330e-01 -1.14870656e+00
7.06329150e-03 -7.50992239e-01 4.67426658e-01 -3.70864660e-01
-1.07952476e+00 4.40806657e-01 3.69363092e-02 -1.56408453e+00
-4.90774184e-01 -6.89451516e-01 -8.54508281e-02 5.92501163e-01
-1.62114525e+00 -9.19341922e-01 1.27164632e-01 8.44036818e-01
2.38190547e-01 -3.02106977e-01 1.29800701e+00 4.29446489e-01
5.86166792e-02 5.21589339e-01 6.58889532e-01 8.05999488e-02
8.39031160e-01 -1.35588396e+00 -9.40603465e-02 4.19995964e-01
6.05283201e-01 7.80166745e-01 5.90216815e-01 -1.07126124e-01
-1.74988317e+00 -8.58597934e-01 9.23710704e-01 -3.40835989e-01
8.24813426e-01 -4.40224707e-01 -1.13746154e+00 2.36888289e-01
-3.47390682e-01 1.11723267e-01 9.94553089e-01 -8.39924142e-02
-8.28282773e-01 -2.06814677e-01 -1.26933086e+00 2.64681518e-01
4.37371910e-01 -1.08198261e+00 -8.83036494e-01 3.10249567e-01
4.17244464e-01 1.36093706e-01 -1.22249973e+00 7.64222890e-02
4.60949957e-01 -8.50451589e-01 1.20443463e+00 -3.88354331e-01
6.89628348e-02 -3.73107344e-01 -5.58825314e-01 -1.00472701e+00
-1.76907137e-01 -5.07339001e-01 -1.36056036e-01 1.05392897e+00
6.13045879e-02 -7.25222886e-01 8.15534949e-01 2.82747388e-01
2.68492401e-01 -6.71499789e-01 -1.27322400e+00 -9.47990656e-01
1.74916610e-01 5.46940751e-02 3.52910221e-01 8.47962081e-01
4.16949689e-01 1.62623376e-01 -2.90317535e-01 2.88303699e-02
9.88283873e-01 2.80139595e-01 7.11399078e-01 -1.40758085e+00
-1.30276322e-01 -1.93880841e-01 -1.27440691e+00 -8.77543986e-01
2.12946638e-01 -1.06130290e+00 -8.48089308e-02 -7.88169563e-01
1.48108989e-01 -7.09655762e-01 -3.74525785e-01 1.17134593e-01
4.18789424e-02 5.60080111e-01 2.00142667e-01 6.00148141e-01
-4.52550769e-01 8.39392960e-01 7.97658503e-01 -7.70470276e-02
1.22331299e-01 -1.92892283e-01 -3.33402753e-01 2.63365835e-01
6.07107401e-01 -5.89728415e-01 -4.66319740e-01 8.56756195e-02
3.33156250e-02 3.04865818e-02 2.59605825e-01 -1.11063409e+00
2.08879799e-01 5.06031930e-01 7.27795660e-02 -4.12539840e-01
7.27205813e-01 -9.75531995e-01 -3.91633883e-02 5.97372413e-01
-3.69057566e-01 3.08034718e-01 -2.22857863e-01 8.57186973e-01
-6.48477316e-01 -6.23141289e-01 8.47856343e-01 1.23309180e-01
-5.61282754e-01 2.11403400e-01 -1.03819333e-01 -1.56463772e-01
1.12705076e+00 -5.10462463e-01 1.42733231e-01 -2.58539647e-01
-5.98526120e-01 -2.36431554e-01 8.72947097e-01 3.05787295e-01
7.59624481e-01 -1.71092427e+00 -2.60737836e-01 3.43336552e-01
5.56481779e-01 -3.06496590e-01 -1.38700986e-02 6.02081001e-01
-5.10123610e-01 6.96226895e-01 -4.29186434e-01 -7.46952832e-01
-1.30994713e+00 1.13490760e+00 -5.54317944e-02 -7.89989084e-02
-5.49732268e-01 3.26545358e-01 -9.96788815e-02 -4.72729206e-01
2.29596600e-01 -9.52792466e-02 -6.55982569e-02 2.52034396e-01
5.52784085e-01 3.39875877e-01 1.77522898e-01 -1.03344905e+00
-2.02711195e-01 1.02316678e+00 -6.06595390e-02 -2.19401672e-01
1.17894518e+00 -1.77935883e-01 -3.04382443e-01 5.62112212e-01
1.87987375e+00 -3.31136376e-01 -7.50516951e-01 -3.51248592e-01
3.01607400e-01 -6.29331231e-01 -1.24872796e-01 1.77823097e-01
-4.48801577e-01 1.19828713e+00 6.71144664e-01 4.95341152e-01
9.26924646e-01 -3.51696461e-02 8.52334082e-01 6.40064716e-01
8.02860558e-01 -9.11842227e-01 2.02340201e-01 8.79435539e-02
5.10896087e-01 -1.24695849e+00 3.59288156e-02 -3.74968588e-01
-1.34028554e-01 1.38902795e+00 -7.18547851e-02 -2.89438307e-01
5.48988223e-01 -1.32904246e-01 -1.59883410e-01 -1.07743442e-01
-3.02750945e-01 3.53108719e-02 1.77832127e-01 6.41073167e-01
2.00438425e-01 -5.26065007e-03 -3.68271917e-01 -3.52615505e-01
-2.30151601e-02 -3.81540179e-01 6.00985408e-01 1.05775213e+00
-6.52477264e-01 -1.58016980e+00 -7.76507080e-01 1.09624013e-01
-2.06782490e-01 9.58910584e-02 -2.00937375e-01 9.35323119e-01
-3.19245487e-01 7.02086985e-01 -1.82966024e-01 -1.76223442e-01
-1.29666314e-01 2.35758349e-01 5.65069795e-01 -5.49523771e-01
-1.42170280e-01 -1.78048760e-01 -3.83567214e-01 -5.53970754e-01
-6.51733041e-01 -8.21339309e-01 -1.11669099e+00 1.23059587e-03
-4.58270848e-01 5.68024755e-01 9.53752577e-01 6.27718925e-01
1.87233344e-01 -3.87867063e-01 9.00136888e-01 -1.05599201e+00
-1.10191119e+00 -5.96418440e-01 -9.45888102e-01 6.89956069e-01
4.06204760e-01 -1.12094474e+00 -6.16291583e-01 -9.84292179e-02]
|
[8.027703285217285, 4.038346767425537]
|
0d8be60f-0acf-4a61-aba3-c962709e662e
|
single-exposure-absorption-imaging-of
|
2003.01643
| null |
https://arxiv.org/abs/2003.01643v2
|
https://arxiv.org/pdf/2003.01643v2.pdf
|
Single-exposure absorption imaging of ultracold atoms using deep learning
|
Absorption imaging is the most common probing technique in experiments with ultracold atoms. The standard procedure involves the division of two frames acquired at successive exposures, one with the atomic absorption signal and one without. A well-known problem is the presence of residual structured noise in the final image, due to small differences between the imaging light in the two exposures. Here we solve this problem by performing absorption imaging with only a single exposure, where instead of a second exposure the reference frame is generated by an unsupervised image-completion autoencoder neural network. The network is trained on images without absorption signal such that it can infer the noise overlaying the atomic signal based only on the information in the region encircling the signal. We demonstrate our approach on data captured with a quantum degenerate Fermi gas. The average residual noise in the resulting images is below that of the standard double-shot technique. Our method simplifies the experimental sequence, reduces the hardware requirements, and can improve the accuracy of extracted physical observables. The trained network and its generating scripts are available as an open-source repository (http://absDL.github.io/).
|
['Gal Ness', 'Yoav Sagi', 'Yanay Florshaim', 'Constantine Shkedrov', 'Anastasiya Vainbaum']
|
2020-03-03
| null | null | null | null |
['physical-attribute-prediction']
|
['computer-vision']
|
[ 4.92795616e-01 -1.58496067e-01 7.28847444e-01 -2.65545368e-01
-6.06075764e-01 -2.20098197e-01 4.48735058e-01 -9.67747793e-02
-8.78101230e-01 8.71820211e-01 -3.10514957e-01 -9.07728449e-02
2.66035408e-01 -8.34842622e-01 -8.52566898e-01 -1.36529696e+00
2.21016347e-01 7.35918224e-01 2.08483726e-01 -1.88965835e-02
9.75931138e-02 3.28523397e-01 -1.34687078e+00 1.11927979e-01
5.20966947e-01 9.16373491e-01 2.89404243e-01 6.96197748e-01
1.63191319e-01 8.45971584e-01 -5.27218044e-01 -1.80335730e-01
4.12023574e-01 -9.62793291e-01 -6.79764569e-01 -5.62586077e-02
2.99225420e-01 -5.48775792e-01 -7.28424728e-01 1.30531228e+00
5.06973624e-01 3.76138121e-01 3.53063703e-01 -5.69545805e-01
-3.44748110e-01 4.44467127e-01 -3.93211842e-01 2.79811293e-01
1.80235505e-01 3.86520177e-01 6.86291516e-01 -6.77522659e-01
8.43038201e-01 7.15409577e-01 3.88185740e-01 6.28368556e-01
-1.66044867e+00 -3.02304447e-01 -8.48012924e-01 3.70986730e-01
-1.02052557e+00 -7.25355983e-01 8.86002541e-01 -5.17091453e-01
7.96336353e-01 1.18852081e-02 6.15271866e-01 1.02237630e+00
2.86329448e-01 -1.52803004e-01 1.49669266e+00 -9.82769489e-01
2.98038721e-01 6.14911355e-02 4.22182769e-01 6.80891812e-01
2.37750992e-01 3.27341706e-01 -2.59371519e-01 -2.54095912e-01
7.41977215e-01 1.39005082e-02 -5.99004745e-01 -1.66020915e-01
-9.78798091e-01 6.17731512e-01 3.69899392e-01 5.43646514e-01
-5.30987084e-01 7.86161274e-02 1.62171930e-01 3.79617900e-01
2.86736995e-01 5.01069427e-01 6.39573634e-02 1.43468142e-01
-1.02189970e+00 1.60973027e-01 7.77768612e-01 2.09707171e-01
1.12684536e+00 1.14459544e-02 2.33091906e-01 3.29718351e-01
-4.37756814e-02 3.76362532e-01 5.16587377e-01 -1.22806346e+00
-2.31765792e-01 9.60739553e-02 3.75968993e-01 -3.85693401e-01
-2.44433761e-01 -1.41919553e-02 -7.05817699e-01 6.28064215e-01
8.47423434e-01 -4.83638674e-01 -7.81237900e-01 1.56365263e+00
4.35948700e-01 1.61422729e-01 1.63743988e-01 1.00632334e+00
6.95733547e-01 7.25683868e-01 -5.83404899e-01 -6.51047170e-01
1.17361212e+00 -6.44294858e-01 -8.31036329e-01 5.93729457e-03
1.52311295e-01 -6.89596415e-01 6.28286183e-01 3.40658665e-01
-1.21155620e+00 -5.34672976e-01 -1.20295012e+00 -4.13583308e-01
-5.65955788e-02 -1.20618738e-01 3.45886707e-01 4.12758291e-01
-9.47408795e-01 1.23067558e+00 -9.36210990e-01 8.36558044e-02
-4.58979160e-02 4.07760233e-01 -4.08896476e-01 8.19942951e-02
-1.09673154e+00 7.63622940e-01 4.32513297e-01 -5.33506162e-02
-6.92428768e-01 -6.34825885e-01 -4.33880895e-01 -1.17624857e-01
2.51540601e-01 -4.08846915e-01 1.26408339e+00 -1.07105315e+00
-1.75816977e+00 9.26609337e-01 -3.33832055e-01 -5.21296501e-01
4.68753844e-01 4.58913147e-02 -1.43316269e-01 6.48475707e-01
-6.40335977e-02 6.29724637e-02 1.00706685e+00 -1.18883991e+00
-4.28149803e-03 -4.21760589e-01 -5.58007546e-02 -1.86513618e-01
3.90717387e-01 -7.68477246e-02 -2.26207793e-01 1.97573885e-01
2.25487724e-01 -9.72220123e-01 -5.88540435e-02 -5.83260536e-01
-1.59369528e-01 4.04183388e-01 7.18500972e-01 -7.92999983e-01
4.77465928e-01 -2.09568357e+00 3.85624558e-01 1.33564994e-02
3.05771977e-01 1.03618838e-01 3.74498755e-01 5.89237392e-01
-2.57789701e-01 -4.83600289e-01 -5.49442828e-01 -4.61618513e-01
-3.78494591e-01 1.94879994e-02 -2.02316180e-01 8.36310506e-01
-1.85187325e-01 4.25841153e-01 -7.72837758e-01 -6.76874742e-02
1.78616837e-01 5.02328813e-01 -3.00765663e-01 5.03616095e-01
-1.83743253e-01 1.22392857e+00 2.97378283e-02 -1.35533726e-02
8.26832592e-01 -2.62432814e-01 3.32366943e-01 -4.51272577e-01
-6.14168465e-01 2.44199052e-01 -9.39059317e-01 1.54013515e+00
-7.33826458e-02 7.02722549e-01 2.78944284e-01 -1.08868289e+00
6.03932142e-01 5.50014973e-01 5.42396784e-01 -7.30537355e-01
3.76556903e-01 3.45475107e-01 2.82448411e-01 -6.89194918e-01
3.40291500e-01 -6.60481334e-01 3.33203763e-01 5.24324119e-01
4.68946785e-01 -3.05208594e-01 2.69647270e-01 1.58081889e-01
9.93914247e-01 1.35737702e-01 2.00157061e-01 -3.24981272e-01
3.30048203e-01 -1.55639753e-01 3.07493657e-01 9.47427511e-01
-1.95758879e-01 6.71875954e-01 4.80303347e-01 -7.47439563e-01
-1.70336652e+00 -5.12381256e-01 -3.54255021e-01 3.52996826e-01
1.66475654e-01 -1.96030959e-01 -1.15487814e+00 -2.86007617e-02
-3.68310601e-01 4.36980128e-01 -6.05447233e-01 -6.88346103e-02
-6.21560752e-01 -9.86415505e-01 1.86655387e-01 -1.69444725e-01
6.45081937e-01 -1.23877406e+00 -8.11064065e-01 1.32791623e-01
-2.80314624e-01 -1.15869677e+00 6.56441748e-02 3.91332746e-01
-6.52758062e-01 -1.02669907e+00 -3.84318829e-01 -2.14991584e-01
6.13413453e-01 1.56289950e-01 9.72472608e-01 1.58855423e-01
-5.98311186e-01 2.51936942e-01 -1.17048420e-01 -5.44089787e-02
-5.44561386e-01 -5.39899528e-01 2.87085354e-01 1.77575126e-01
2.60198504e-01 -8.14604938e-01 -6.88745797e-01 -4.07452792e-01
-8.08895290e-01 -1.30011722e-01 1.59708932e-01 9.74338651e-01
5.91826200e-01 8.47791284e-02 -1.98912010e-01 -8.13686848e-01
1.98844314e-01 -2.17948005e-01 -9.22240794e-01 -2.95574218e-01
-2.05202684e-01 3.10443759e-01 8.09952319e-01 -7.91491121e-02
-1.17577350e+00 -8.52954462e-02 -2.36457691e-01 -4.22847331e-01
-3.84727210e-01 1.35946602e-01 2.85399735e-01 -3.97587717e-01
7.68122256e-01 3.15289915e-01 1.01064168e-01 -3.87994826e-01
-9.04901624e-02 4.20574874e-01 7.59843528e-01 -5.38465083e-01
6.59797728e-01 9.15407419e-01 2.73424506e-01 -1.36898375e+00
-7.71687329e-01 -4.62769091e-01 -8.29224408e-01 -2.76236504e-01
1.00708580e+00 -7.01057374e-01 -9.54348087e-01 6.43421471e-01
-1.07124114e+00 -4.36477631e-01 -4.55394387e-01 8.16920280e-01
-4.47008401e-01 6.54324234e-01 -1.01082349e+00 -9.11505640e-01
-3.27720970e-01 -1.31599987e+00 7.41721928e-01 4.55306470e-01
2.19126374e-01 -7.51987517e-01 4.63396400e-01 3.76914591e-01
1.62748948e-01 1.99394256e-01 6.03882015e-01 -1.94661781e-01
-6.72452152e-01 -1.53421581e-01 1.11605555e-01 3.51017773e-01
-2.33975910e-02 8.26776624e-02 -1.29366720e+00 -4.51769650e-01
9.16972101e-01 -4.85147297e-01 1.04627490e+00 5.11437595e-01
8.52503836e-01 1.97595254e-01 3.02618090e-02 5.78032076e-01
1.62381375e+00 4.06758398e-01 7.91230500e-01 1.27230451e-01
5.80431044e-01 4.81150746e-01 -1.18246414e-02 2.05679655e-01
-4.53569025e-01 6.90268755e-01 2.91321516e-01 5.02985567e-02
6.82803094e-02 4.57944959e-01 3.59865069e-01 6.82903469e-01
-7.05319583e-01 -1.13960929e-01 -6.05065227e-01 1.59459606e-01
-1.65380001e+00 -1.37354338e+00 -3.81202102e-01 2.44112468e+00
5.84969580e-01 -1.21629149e-01 -8.41412246e-02 1.26678810e-01
4.62935627e-01 1.13361485e-01 -3.64855886e-01 -2.05740973e-01
1.96485594e-02 6.14193857e-01 4.11713719e-01 9.50364172e-01
-8.07512283e-01 5.43229163e-01 6.22839451e+00 2.86965936e-01
-1.35739720e+00 4.19482559e-01 3.17667991e-01 -8.40269178e-02
7.48207187e-03 2.34470561e-01 -4.62118804e-01 6.35080755e-01
1.17811561e+00 1.38428673e-01 6.90024376e-01 4.23157573e-01
2.15185285e-01 -5.15068829e-01 -8.82948995e-01 8.62515450e-01
-8.05402845e-02 -1.31631422e+00 -4.62722242e-01 1.43706696e-02
4.26546961e-01 2.76395053e-01 -1.22755155e-01 -1.90251037e-01
-5.99191003e-02 -6.14765286e-01 6.34228408e-01 7.42203951e-01
8.27763021e-01 -4.62246895e-01 7.04642415e-01 2.69245267e-01
-5.27708471e-01 1.97251990e-01 -6.41983092e-01 -3.89870167e-01
2.59549260e-01 9.83450413e-01 -4.21525180e-01 4.78929281e-01
6.32457793e-01 1.52023271e-01 -3.19057368e-02 6.05398417e-01
-8.00403878e-02 7.14240968e-01 -3.13739181e-01 2.71224052e-01
2.36461218e-02 -1.16458237e+00 6.10345662e-01 9.13416862e-01
3.98293465e-01 4.30601299e-01 3.74694169e-02 1.01391470e+00
-2.60402143e-01 -4.33053017e-01 -6.45353258e-01 8.75058323e-02
-8.38945992e-03 1.33253729e+00 -6.95630670e-01 -5.37952006e-01
-5.97377896e-01 1.26958072e+00 3.91287893e-01 4.45935100e-01
-4.88994241e-01 -3.41377139e-01 3.17218274e-01 1.02381781e-01
3.19022268e-01 -3.23812068e-01 3.22775602e-01 -1.47507405e+00
-8.97721350e-02 -5.27382970e-01 2.34796689e-03 -7.37720549e-01
-9.75817740e-01 6.41490877e-01 -1.72827885e-01 -5.94270766e-01
-9.14207846e-02 -6.88992321e-01 -7.28398323e-01 9.32738483e-01
-1.12909603e+00 -4.56026107e-01 -5.01794994e-01 4.34986144e-01
6.14164770e-02 3.17415506e-01 9.37708497e-01 2.97821820e-01
-6.61496043e-01 -1.98242977e-01 7.05099285e-01 1.67449396e-02
6.98912919e-01 -1.48378217e+00 6.94519132e-02 9.66856301e-01
1.26100928e-01 5.87476909e-01 1.20216990e+00 -4.80015516e-01
-1.27969515e+00 -4.06065136e-01 4.57837850e-01 -2.20581189e-01
6.71951294e-01 -3.53333950e-01 -1.07605100e+00 8.15458357e-01
6.61872923e-01 1.81313321e-01 6.10822499e-01 -3.34564328e-01
-2.83496361e-02 3.44524682e-02 -1.20199788e+00 2.47319490e-01
5.16466379e-01 -8.33776116e-01 -4.67072338e-01 5.04414976e-01
3.62969428e-01 -4.83495444e-01 -7.67976284e-01 -4.15977314e-02
5.38788021e-01 -1.58881509e+00 7.63807774e-01 -2.35492855e-01
4.06989127e-01 -2.51726627e-01 1.81951374e-02 -1.17784023e+00
-3.56524050e-01 -8.01900506e-01 3.04482244e-02 5.43573081e-01
1.10732764e-01 -7.80927062e-01 5.84975541e-01 4.60638374e-01
-2.81904475e-03 -2.08690956e-01 -9.04059112e-01 -4.89012688e-01
-1.33015230e-01 1.81563661e-01 1.06326640e-01 6.86262250e-01
5.77385128e-02 3.64396006e-01 -4.81911838e-01 3.46934438e-01
1.08421993e+00 4.66564566e-01 5.39110541e-01 -1.11625576e+00
-7.40099847e-01 3.37168932e-01 -1.34681925e-01 -7.01976717e-01
3.67603451e-02 -5.02072692e-01 2.71284938e-01 -7.03424871e-01
4.02015150e-01 1.88971534e-01 9.90650523e-03 1.08936904e-02
-5.73903918e-02 4.67251182e-01 5.68220653e-02 5.13305128e-01
-2.01564312e-01 3.76212031e-01 1.01941085e+00 2.80955583e-01
-5.35671152e-02 -2.96238244e-01 6.02868982e-02 7.94610739e-01
9.58988965e-01 -6.25700593e-01 6.61170408e-02 -1.57957241e-01
-4.87501137e-02 2.11823240e-01 4.92705762e-01 -1.30540848e+00
2.16297269e-01 2.77107507e-01 3.36245388e-01 -1.72605529e-01
6.17652833e-01 -6.13454461e-01 5.73881626e-01 6.19252264e-01
7.56587461e-02 -2.73499221e-01 -1.59681365e-01 2.98598439e-01
-1.16916627e-01 -9.80910242e-01 1.22109044e+00 -7.15118408e-01
-2.17289835e-01 1.74659401e-01 -3.44307780e-01 -3.81996125e-01
6.36414528e-01 2.39641905e-01 -2.76483595e-01 -3.69620472e-01
-9.51327205e-01 -5.63072383e-01 8.50854754e-01 -6.78346336e-01
2.35485077e-01 -8.04461181e-01 -2.55796671e-01 3.19803864e-01
-5.13736188e-01 -8.65804479e-02 3.85331124e-01 1.11471319e+00
-8.93339097e-01 1.15204059e-01 -3.05987507e-01 -3.81265730e-01
-1.18447840e+00 6.02670670e-01 6.38755202e-01 -2.09435374e-01
-9.55720603e-01 6.69158280e-01 9.47996378e-02 -6.58043548e-02
-4.81718719e-01 7.52658024e-02 3.26171555e-02 -2.38314241e-01
8.26878846e-01 3.70708942e-01 1.70313403e-01 -8.03558528e-01
2.00811252e-01 4.75163281e-01 -2.04822734e-01 -3.60490173e-01
1.39560664e+00 -4.21067066e-02 -5.46362340e-01 7.88820446e-01
1.18835425e+00 3.02231044e-01 -1.36709130e+00 -1.61016449e-01
-3.99095625e-01 -2.83544272e-01 2.51031548e-01 -1.23437926e-01
-8.31874788e-01 9.20826495e-01 5.64421237e-01 6.25609219e-01
8.45447719e-01 -1.19561791e-01 7.17618585e-01 5.21954715e-01
4.38548654e-01 -9.28098917e-01 -8.47218633e-02 4.28807735e-01
5.56254089e-01 -1.30298245e+00 3.38365324e-02 -1.20128296e-01
-1.15047574e-01 1.42775941e+00 1.11130737e-01 -4.66487259e-01
2.48294145e-01 2.24608287e-01 5.27463257e-02 -5.80112934e-01
-3.48111093e-01 -1.27879128e-01 -3.49636376e-01 2.44885579e-01
3.74499112e-01 -5.71566671e-02 -1.86425492e-01 -1.97655056e-02
-4.48021367e-02 -9.36778858e-02 9.48596299e-01 8.00605953e-01
-4.81077284e-01 -1.13861763e+00 -4.35525179e-01 2.42500916e-01
-7.62163818e-01 -5.12825437e-02 -2.03297600e-01 4.85584497e-01
1.96874902e-01 5.20785689e-01 1.21860474e-01 7.70313740e-02
5.27837593e-03 4.97829348e-01 9.32061791e-01 -3.80691350e-01
-1.47120133e-01 1.12715565e-01 -1.37378439e-01 -5.49904227e-01
-7.84147203e-01 -6.62485182e-01 -1.30322397e+00 -3.60458046e-01
-4.60714400e-01 5.88069499e-01 5.49013138e-01 9.26070511e-01
7.77950808e-02 4.52872127e-01 5.35909712e-01 -1.42816854e+00
-4.00745779e-01 -9.48689342e-01 -8.45708847e-01 6.13551259e-01
8.07433128e-01 -3.87345701e-01 -7.88090050e-01 2.34787881e-01]
|
[12.251259803771973, -2.6500351428985596]
|
0894c76e-40ae-4d04-9686-d363751f16b2
|
scrib-set-classifier-with-class-specific-risk
|
2103.03945
| null |
https://arxiv.org/abs/2103.03945v1
|
https://arxiv.org/pdf/2103.03945v1.pdf
|
SCRIB: Set-classifier with Class-specific Risk Bounds for Blackbox Models
|
Despite deep learning (DL) success in classification problems, DL classifiers do not provide a sound mechanism to decide when to refrain from predicting. Recent works tried to control the overall prediction risk with classification with rejection options. However, existing works overlook the different significance of different classes. We introduce Set-classifier with Class-specific RIsk Bounds (SCRIB) to tackle this problem, assigning multiple labels to each example. Given the output of a black-box model on the validation set, SCRIB constructs a set-classifier that controls the class-specific prediction risks with a theoretical guarantee. The key idea is to reject when the set classifier returns more than one label. We validated SCRIB on several medical applications, including sleep staging on electroencephalogram (EEG) data, X-ray COVID image classification, and atrial fibrillation detection based on electrocardiogram (ECG) data. SCRIB obtained desirable class-specific risks, which are 35\%-88\% closer to the target risks than baseline methods.
|
['Jimeng Sun', 'M. Brandon Westover', 'Lucas Glass', 'Cao Xiao', 'Zhen Lin']
|
2021-03-05
| null | null | null | null |
['sleep-staging', 'atrial-fibrillation-detection']
|
['medical', 'medical']
|
[ 3.17172110e-01 2.38600314e-01 -4.21442717e-01 -8.02991271e-01
-6.58598483e-01 -2.18082711e-01 9.33327898e-02 3.65263909e-01
-4.44420040e-01 9.69612896e-01 -2.79785633e-01 -5.77234030e-01
-5.64306200e-01 -6.81771576e-01 -3.60735923e-01 -7.60446429e-01
-3.51321101e-01 5.43529987e-01 -6.48621023e-02 1.49889395e-01
1.36101350e-01 2.98683047e-01 -1.50624418e+00 7.62044907e-01
1.05956733e+00 1.26696718e+00 -4.06534046e-01 2.78185457e-01
2.48488396e-01 6.31653607e-01 -9.38758790e-01 -1.65378645e-01
1.45723358e-01 -7.93733597e-01 -4.10216451e-01 -3.87398809e-01
4.48751859e-02 -6.00370318e-02 4.73666787e-01 1.10332918e+00
5.34697950e-01 -3.31191838e-01 1.14448011e+00 -1.68876374e+00
-2.07869753e-01 8.89192343e-01 -4.26118135e-01 4.71372545e-01
-4.05327231e-02 -1.76799685e-01 9.42416191e-01 -4.92053717e-01
-1.25689402e-01 7.31331229e-01 7.82542288e-01 8.09497833e-01
-1.42101204e+00 -1.19242752e+00 7.16650337e-02 -7.20532164e-02
-1.45910168e+00 -5.32399341e-02 4.19478744e-01 -4.82798845e-01
7.19087780e-01 6.46534860e-01 6.07024372e-01 1.18742621e+00
6.93381488e-01 3.55081558e-01 1.42471659e+00 -3.71962816e-01
4.96142834e-01 4.26500738e-01 6.08059347e-01 3.12979519e-01
4.59083378e-01 3.43791127e-01 -5.40156722e-01 -4.96634752e-01
3.86506408e-01 -5.07661607e-03 -2.86331534e-01 -9.69284326e-02
-9.06421542e-01 7.94034004e-01 5.81494570e-02 2.09891886e-01
-1.64449230e-01 -1.87069967e-01 4.23806995e-01 6.14816129e-01
5.44355333e-01 5.39449513e-01 -8.06318104e-01 2.96330482e-01
-9.31637526e-01 6.60740063e-02 7.23502100e-01 4.94180322e-01
1.64158791e-01 -1.58721343e-01 -3.95740896e-01 7.44333506e-01
1.18557133e-01 3.20932239e-01 5.79380631e-01 -4.18217540e-01
7.68896490e-02 5.74564040e-01 -2.03799494e-02 -5.70025504e-01
-6.03928983e-01 -8.42974424e-01 -1.22556221e+00 6.54422581e-01
3.28231186e-01 -2.32115179e-01 -6.80953622e-01 1.71034503e+00
-2.06046864e-01 2.52264589e-02 -5.05326986e-02 6.87246859e-01
4.27011490e-01 2.25880966e-01 1.81074604e-01 -5.90730488e-01
1.14723659e+00 -2.14779258e-01 -5.12167692e-01 -1.57100618e-01
8.45216632e-01 1.13628417e-01 1.02817655e+00 1.06991971e+00
-5.82933784e-01 -3.79972786e-01 -1.23998213e+00 6.50111616e-01
-4.70132846e-03 1.06637850e-01 4.51793879e-01 1.20855737e+00
-8.53749514e-01 7.82590449e-01 -6.23005092e-01 1.61820799e-01
5.48937619e-01 7.92474985e-01 -7.29292706e-02 5.34114659e-01
-1.24369586e+00 7.53804445e-01 3.69663000e-01 -2.10973695e-01
-8.75425875e-01 -6.21961415e-01 -4.59596276e-01 1.20090023e-01
1.29019588e-01 -3.85533452e-01 7.91957617e-01 -1.31444180e+00
-1.02409601e+00 9.59564269e-01 3.37875754e-01 -8.74466598e-01
6.46554887e-01 -3.06867599e-01 -4.36056495e-01 -4.45525825e-01
-1.28599266e-02 3.81490350e-01 1.14113045e+00 -9.29944873e-01
-6.69869840e-01 -3.16209465e-01 -2.13424981e-01 -1.54136658e-01
-3.04676503e-01 1.20861076e-01 4.46192890e-01 -7.26529717e-01
-8.97837281e-02 -7.02120781e-01 -2.21478596e-01 -1.49454877e-01
-6.33250117e-01 -3.97194535e-01 3.45292598e-01 -1.08733691e-01
1.44696283e+00 -2.26766944e+00 -1.94326371e-01 2.82888710e-01
3.57573926e-01 2.96417117e-01 1.67928457e-01 -2.34027728e-01
-5.88699520e-01 3.92540514e-01 -3.50475758e-01 -1.66370630e-01
-2.67767161e-01 8.06026757e-02 -5.10900021e-01 6.41328692e-01
2.56514490e-01 2.88230270e-01 -4.82001960e-01 -2.71820098e-01
1.50213659e-01 1.31122231e-01 -6.33103549e-01 1.91303551e-01
-9.44178626e-02 3.21847260e-01 -4.56293702e-01 3.14024031e-01
5.28587580e-01 -1.38419837e-01 1.69050246e-01 1.25724792e-01
3.26011628e-01 3.01446825e-01 -1.10486948e+00 8.14169049e-01
-2.11349562e-01 3.38450730e-01 -4.12900507e-01 -1.37135792e+00
1.28537714e+00 2.53448039e-01 4.13305581e-01 -3.59986633e-01
3.39273870e-01 1.59637675e-01 4.08895850e-01 -1.30932704e-01
-4.75061476e-01 -4.68893975e-01 -2.10248694e-01 5.22802889e-01
-1.05784208e-01 1.60670727e-01 -4.48292315e-01 -3.31572354e-01
1.03044999e+00 -1.44225776e-01 6.44635558e-01 -6.83828890e-01
5.16420960e-01 -4.09821451e-01 9.06224132e-01 1.18488801e+00
-2.00069621e-01 6.63823605e-01 7.83868730e-01 -7.74703324e-01
-4.16831344e-01 -9.23323452e-01 -7.99626827e-01 6.73722863e-01
-5.56186847e-02 -2.75175005e-01 -7.16341794e-01 -1.05559337e+00
-4.82424982e-02 1.02049208e+00 -8.21788013e-01 -5.76653063e-01
-1.56798914e-01 -1.43883932e+00 6.19118512e-01 5.43178320e-01
1.41193485e-03 -1.12744141e+00 -1.00830758e+00 -3.83200236e-02
2.09735364e-01 -4.86869037e-01 -5.96291125e-02 9.51054513e-01
-9.92015839e-01 -1.15394461e+00 -3.55454087e-01 -2.85927802e-01
4.70639944e-01 -6.54427588e-01 1.16237020e+00 3.07980359e-01
-2.47633100e-01 -1.71259910e-01 -2.18362138e-01 -7.82001197e-01
-5.87725461e-01 3.09858024e-02 4.40552294e-01 1.85174406e-01
4.56594020e-01 -5.25413752e-01 -6.08029366e-01 5.15194952e-01
-4.45037305e-01 -1.22416832e-01 3.83441418e-01 9.10002291e-01
5.19881129e-01 1.14971787e-01 1.34541595e+00 -8.89953613e-01
7.66257584e-01 -4.86779690e-01 -3.80631536e-01 2.86873996e-01
-1.27754319e+00 -8.15598965e-02 7.38826334e-01 -6.55227900e-01
-3.40251148e-01 -2.05498800e-01 -2.52964832e-02 -2.91610807e-01
-1.85366213e-01 1.81896657e-01 -1.59821078e-01 4.96532649e-01
9.13140118e-01 -4.40039337e-02 -1.57210454e-01 -2.64744550e-01
-4.03604001e-01 8.22022080e-01 1.13263510e-01 -3.53331268e-01
1.83059320e-01 6.23434298e-02 -1.40671119e-01 -3.00089002e-01
-1.03517306e+00 1.52343154e-01 -5.51237464e-01 -1.59101367e-01
8.53981256e-01 -7.52890348e-01 -7.47964621e-01 1.68729052e-01
-7.26074219e-01 -4.37210470e-01 -2.81237513e-01 4.67194468e-01
-3.53545517e-01 2.22977586e-02 -1.96290597e-01 -1.18482292e+00
-6.76771820e-01 -1.01861608e+00 8.40863526e-01 -8.39714780e-02
-6.64013922e-01 -6.47455215e-01 -2.99465626e-01 -4.43617627e-02
5.86557854e-03 3.34750146e-01 1.27083468e+00 -1.32809401e+00
4.71249931e-02 -3.63064796e-01 1.84943289e-01 7.98200011e-01
1.64446548e-01 -1.32956713e-01 -1.08729661e+00 -3.36559057e-01
3.38577688e-01 -1.95601672e-01 7.65402555e-01 6.15377426e-01
1.74375820e+00 -1.95966631e-01 -4.84993160e-01 6.77867591e-01
9.85356450e-01 6.52501881e-01 6.00702226e-01 3.30447197e-01
1.58122424e-02 4.18419838e-01 5.59110045e-01 5.36821187e-01
-2.62586921e-01 4.56517935e-01 3.42685789e-01 -2.79862713e-02
2.96295702e-01 2.44754151e-01 4.51680362e-01 1.97836384e-01
2.75790989e-01 -4.03245479e-01 -9.28418934e-01 1.56649292e-01
-1.49817407e+00 -6.07129872e-01 -2.59395480e-01 2.56545806e+00
8.91116977e-01 8.88323724e-01 1.53916642e-01 6.38341963e-01
6.39128804e-01 -3.55674028e-01 -6.76198721e-01 -4.91143525e-01
-6.35382012e-02 4.43840802e-01 1.73541322e-01 2.88859516e-01
-1.13265204e+00 2.88418174e-01 6.74258661e+00 7.40417302e-01
-1.19693148e+00 1.29142990e-02 1.35711539e+00 -1.96370095e-01
-9.21896920e-02 -2.23621309e-01 -8.96697462e-01 8.01966190e-01
1.03324533e+00 -1.29662782e-01 -4.68152277e-02 9.17055190e-01
-3.27140354e-02 -8.72137323e-02 -1.51870883e+00 9.48254824e-01
-7.29048476e-02 -1.07575536e+00 -9.62202027e-02 -3.16144638e-02
1.80517495e-01 -1.74173817e-01 1.75697058e-02 5.58185220e-01
-4.53427024e-02 -1.42582369e+00 7.17148781e-01 6.26509428e-01
1.03246212e+00 -9.05168653e-01 8.61821353e-01 6.95696890e-01
-2.83241987e-01 -2.77954489e-01 -1.89436063e-01 -2.69612700e-01
-4.13401097e-01 9.41772819e-01 -7.15614498e-01 2.71014720e-01
1.01265371e+00 5.71990788e-01 -4.20593262e-01 6.77986801e-01
-7.04636946e-02 1.01700008e+00 -2.58913457e-01 -1.48123235e-01
-2.95357376e-01 -1.17805511e-01 3.33480597e-01 9.65920031e-01
9.21046659e-02 2.54821092e-01 9.90734398e-02 8.42024505e-01
2.07819507e-01 -2.47018542e-02 -3.86486560e-01 5.23154140e-01
2.52187282e-01 8.45896363e-01 -9.10799503e-01 -1.48487538e-01
2.22114921e-02 5.98885834e-01 1.10419586e-01 -8.59543905e-02
-9.67806339e-01 -3.68858606e-01 4.23837304e-01 1.63112208e-01
-2.93277830e-01 4.37728763e-01 -9.16565776e-01 -9.63718355e-01
-2.25368708e-01 -8.52885425e-01 6.87268615e-01 -6.51767433e-01
-1.42040145e+00 8.48304272e-01 1.46000072e-01 -1.43007672e+00
3.88394296e-02 -6.13278151e-01 -5.77570736e-01 8.38121831e-01
-9.82936442e-01 -3.11830670e-01 -1.23515669e-02 3.65323186e-01
3.87129158e-01 -2.76597381e-01 1.15665472e+00 2.60799199e-01
-7.06800461e-01 9.54594493e-01 -1.90206975e-01 9.31480229e-02
8.05883825e-01 -1.31098950e+00 -1.20085411e-01 3.49770904e-01
-4.40917946e-02 4.49061275e-01 6.72598720e-01 -5.38890064e-01
-4.89009678e-01 -1.02865398e+00 7.65587628e-01 -7.44449437e-01
1.94481418e-01 -3.92666131e-01 -8.85215282e-01 5.14251530e-01
-1.75717995e-01 2.50526249e-01 1.10018182e+00 3.26834023e-01
-1.07105300e-01 -4.44404900e-01 -1.35472548e+00 3.00600469e-01
7.46786356e-01 -5.59694842e-02 -7.10856616e-01 2.32409999e-01
2.19820157e-01 -1.25547960e-01 -7.51284122e-01 9.44059193e-01
6.27771676e-01 -1.21563578e+00 7.99519360e-01 -7.75078833e-01
2.58651078e-01 7.80025199e-02 3.31907384e-02 -1.22072375e+00
-9.07937586e-02 -4.92777377e-01 9.56862420e-02 9.48350966e-01
5.98622203e-01 -7.29388297e-01 6.04039431e-01 5.54858029e-01
-3.61971520e-02 -1.10872984e+00 -9.58948910e-01 -6.90696836e-01
1.97710291e-01 -7.57242084e-01 5.52204013e-01 1.01135516e+00
5.09711020e-02 3.92680228e-01 -4.89178389e-01 2.64217317e-01
6.97981238e-01 -2.69080177e-02 1.88916132e-01 -1.62233746e+00
-3.08372021e-01 -4.10712183e-01 -4.22483504e-01 -1.95160568e-01
2.44915664e-01 -1.19465017e+00 -1.47895113e-01 -1.05423272e+00
2.45428041e-01 -9.74851251e-01 -1.02179956e+00 8.23977470e-01
-2.27487534e-01 1.09745219e-01 -1.68209851e-01 1.29089788e-01
-3.22588086e-01 2.38242120e-01 6.29241288e-01 -5.31658195e-02
-2.65749305e-01 6.47694826e-01 -7.47173309e-01 8.93838465e-01
9.94031191e-01 -1.02755952e+00 -6.95933878e-01 5.11539318e-02
3.23288798e-01 2.26901621e-01 2.83624887e-01 -1.12833238e+00
-3.12418729e-01 -1.44066349e-01 6.96961224e-01 -4.45975780e-01
-6.68240413e-02 -7.45660663e-01 1.46107331e-01 7.54101396e-01
-8.73398900e-01 -1.29474193e-01 2.08845466e-01 5.51311076e-01
1.92298248e-01 -2.38866404e-01 9.60601866e-01 2.47908235e-01
-1.47368219e-02 2.18943253e-01 -6.00775957e-01 9.81117561e-02
1.13506174e+00 -1.07686237e-01 8.01056996e-02 -1.98026642e-01
-1.12737679e+00 1.23769462e-01 -1.69315860e-01 4.29576397e-01
6.25965059e-01 -8.62252057e-01 -7.17304111e-01 6.94162369e-01
1.85596943e-01 -1.40390351e-01 -1.91472441e-01 7.38790751e-01
-1.09119192e-01 2.62253881e-02 -1.45834580e-01 -8.63548040e-01
-1.34551847e+00 3.85637939e-01 9.31654334e-01 -7.95324147e-02
-8.70769501e-01 1.03557062e+00 3.99603456e-01 -2.08498955e-01
5.43355048e-01 -6.05804563e-01 -2.68501699e-01 3.13380733e-02
6.51500106e-01 5.34658954e-02 4.10560131e-01 2.00323939e-01
-6.42635763e-01 3.05149611e-02 -7.61529207e-02 1.98368728e-01
1.27103627e+00 3.16341728e-01 -1.01459853e-01 6.92644835e-01
7.69334614e-01 -3.09746385e-01 -9.54937339e-01 4.58981782e-01
3.12557697e-01 -2.10393846e-01 9.63327065e-02 -1.30422676e+00
-9.73807216e-01 8.52858961e-01 1.05684423e+00 3.60373348e-01
1.25592899e+00 -1.98185161e-01 8.63789693e-02 2.87466317e-01
5.72306216e-01 -7.20803320e-01 -1.64415374e-01 4.35609035e-02
8.25308263e-01 -1.07837307e+00 1.10341981e-01 -1.41558930e-01
-7.18127191e-01 9.91985738e-01 6.75564349e-01 -2.41212979e-01
1.10895979e+00 5.62487304e-01 4.28013839e-02 3.27771194e-02
-9.94686961e-01 4.73831356e-01 2.28912279e-01 6.30999088e-01
3.60306919e-01 2.66049296e-01 -4.84921038e-01 1.34950149e+00
-8.64821300e-03 1.89714253e-01 3.77034575e-01 4.33439255e-01
-3.31844091e-01 -1.07817757e+00 -2.99886286e-01 1.19517112e+00
-8.47766221e-01 -2.56887794e-01 -2.22532824e-01 5.63128114e-01
3.26217502e-01 9.29220557e-01 1.10922925e-01 -6.87528491e-01
2.31628865e-01 3.39389056e-01 2.30893180e-01 -7.89703906e-01
-8.46136451e-01 -8.88791233e-02 5.19872271e-02 -2.67739624e-01
-1.07022531e-01 -5.18762350e-01 -1.13819861e+00 2.59202242e-01
-5.61945736e-01 3.61334771e-01 1.99292481e-01 8.47454965e-01
1.68352515e-01 5.96633375e-01 7.27918029e-01 -2.59651273e-01
-8.57830644e-01 -9.01834071e-01 -9.37852144e-01 2.81017900e-01
4.80015635e-01 -7.70482123e-01 -7.62773156e-01 -2.40855426e-01]
|
[8.872971534729004, 4.034825325012207]
|
4c19f93a-53ee-485a-b8ab-93f4bb6e655e
|
the-past-mistake-is-the-future-wisdom-error-1
|
2203.00991
| null |
https://arxiv.org/abs/2203.00991v1
|
https://arxiv.org/pdf/2203.00991v1.pdf
|
The Past Mistake is the Future Wisdom: Error-driven Contrastive Probability Optimization for Chinese Spell Checking
|
Chinese Spell Checking (CSC) aims to detect and correct Chinese spelling errors, which are mainly caused by the phonological or visual similarity. Recently, pre-trained language models (PLMs) promote the progress of CSC task. However, there exists a gap between the learned knowledge of PLMs and the goal of CSC task. PLMs focus on the semantics in text and tend to correct the erroneous characters to semantically proper or commonly used ones, but these aren't the ground-truth corrections. To address this issue, we propose an Error-driven COntrastive Probability Optimization (ECOPO) framework for CSC task. ECOPO refines the knowledge representations of PLMs, and guides the model to avoid predicting these common characters through an error-driven way. Particularly, ECOPO is model-agnostic and it can be combined with existing CSC methods to achieve better performance. Extensive experiments and detailed analyses on SIGHAN datasets demonstrate that ECOPO is simple yet effective.
|
['Hai-Tao Zheng', 'Yunbo Cao', 'Chao Li', 'Zizhen Wang', 'Rongyi Sun', 'Ruiyang Liu', 'Zhongli Li', 'Yangning Li', 'Qingyu Zhou', 'Yinghui Li']
|
2022-03-02
| null |
https://aclanthology.org/2022.findings-acl.252
|
https://aclanthology.org/2022.findings-acl.252.pdf
|
findings-acl-2022-5
|
['chinese-spell-checking']
|
['natural-language-processing']
|
[ 2.21018642e-01 -3.63823503e-01 -3.84216616e-03 -5.78790382e-02
-4.57337111e-01 -1.87599942e-01 4.80034620e-01 2.12100863e-01
-5.76315165e-01 5.75887680e-01 5.41246891e-01 -2.62370110e-01
2.95256674e-01 -4.89734322e-01 -6.07931435e-01 -4.54519987e-01
6.38105035e-01 1.98427185e-01 5.76441526e-01 -1.25868499e-01
8.14775646e-01 1.22006938e-01 -1.34515762e+00 6.71595156e-01
1.63275456e+00 5.73810399e-01 8.76903236e-01 5.71232498e-01
-6.90437019e-01 9.59282219e-01 -6.87269986e-01 -4.41169262e-01
-3.65629464e-01 -5.49596131e-01 -8.12894702e-01 -1.66679293e-01
-6.95871785e-02 1.67567149e-01 -1.17866412e-01 1.47192812e+00
2.77536511e-01 9.96062160e-02 6.18715703e-01 -9.93465781e-01
-1.22151530e+00 7.40219653e-01 -3.09927642e-01 2.34172732e-01
2.61553764e-01 -1.62099674e-02 9.19899106e-01 -1.17670882e+00
2.97183216e-01 1.24280858e+00 5.93925178e-01 7.33632207e-01
-4.59838122e-01 -7.20428169e-01 4.97467905e-01 5.74725389e-01
-1.60144699e+00 -1.65286124e-01 5.86685836e-01 -2.39844322e-01
9.07463789e-01 3.31165791e-01 6.09526455e-01 8.90058875e-01
1.10472918e-01 1.34101546e+00 1.02527261e+00 -8.05541694e-01
8.43977109e-02 1.52224436e-01 1.42477974e-01 5.29991806e-01
3.49380881e-01 -1.74534187e-01 -7.10309386e-01 2.75979519e-01
5.40834546e-01 7.85603896e-02 -5.52202642e-01 1.31662548e-01
-1.21840215e+00 4.85280931e-01 1.74916297e-01 5.80135226e-01
-1.13097370e-01 -8.45741928e-02 9.98184010e-02 -1.70529842e-01
2.87852883e-01 6.72651589e-01 -5.54104984e-01 -4.56774712e-01
-8.25774848e-01 2.07808763e-02 4.41903353e-01 1.19893110e+00
6.57786131e-01 -6.30454347e-02 -4.53223526e-01 9.09439027e-01
4.38677341e-01 5.26667476e-01 8.05167794e-01 -3.36587638e-01
5.84268808e-01 9.01799560e-01 1.25438571e-01 -1.07889128e+00
-2.06025779e-01 -4.31388170e-01 -5.38649499e-01 -3.01911563e-01
2.62258172e-01 -7.55637512e-03 -7.63776124e-01 1.27515852e+00
-2.12840855e-01 5.38881004e-01 -1.26222178e-01 1.01991177e+00
6.35647714e-01 8.14703405e-01 2.81697690e-01 -1.72053859e-01
1.00360489e+00 -1.25527298e+00 -9.56700683e-01 -7.32107103e-01
9.72058475e-01 -9.08690333e-01 1.59602880e+00 4.58779454e-01
-7.52780020e-01 -4.57911849e-01 -9.63568747e-01 -1.60394050e-02
-2.65380442e-01 7.40911782e-01 2.93099046e-01 7.07134902e-01
-6.78284228e-01 6.46796405e-01 -6.48079515e-01 -6.29236996e-02
3.98275703e-01 -1.15902863e-01 8.59188139e-02 -1.51520878e-01
-1.13803935e+00 9.65883195e-01 8.06808829e-01 2.53033906e-01
-3.82259458e-01 -6.74968421e-01 -6.80091679e-01 1.47348285e-01
4.57322717e-01 -1.13681629e-01 1.23851752e+00 -1.35235250e+00
-1.54202223e+00 6.24157071e-01 -3.98268253e-01 -1.19025372e-01
3.25786054e-01 -6.28382742e-01 -8.48679900e-01 -8.50040242e-02
1.06134504e-01 4.25822288e-01 6.42548859e-01 -1.19071281e+00
-1.04529989e+00 6.32628873e-02 -4.98890162e-01 1.94959462e-01
-5.42613149e-01 4.33364868e-01 -9.91657078e-01 -9.48340476e-01
1.34420767e-01 -7.65361011e-01 -1.67026266e-01 -2.27795124e-01
-3.69830519e-01 -2.70735681e-01 4.20606971e-01 -9.36307490e-01
1.88775563e+00 -2.20160484e+00 1.88827924e-02 8.29538479e-02
1.71188656e-02 7.68021226e-01 -2.62701690e-01 1.61352068e-01
2.60295510e-01 3.24078232e-01 -2.77760893e-01 -4.60768968e-01
-1.33246899e-01 1.27579510e-01 -3.30514520e-01 2.38429867e-02
3.95120233e-01 9.09772098e-01 -1.03691137e+00 -6.75082028e-01
-1.05488151e-01 3.24371234e-02 -5.72335660e-01 2.31629252e-01
-4.67897058e-01 3.06321353e-01 -4.81299192e-01 5.04522741e-01
8.97420406e-01 -3.42999309e-01 1.76404223e-01 1.67323217e-01
-3.08743685e-01 5.29842615e-01 -1.30996811e+00 1.50932384e+00
-1.83855787e-01 6.10961735e-01 -5.48211217e-01 -6.31047964e-01
1.03864050e+00 -1.63669258e-01 -2.03973100e-01 -1.04903316e+00
9.09138024e-02 3.33546162e-01 -8.90295953e-02 -5.83563626e-01
7.66536474e-01 1.75631121e-01 1.29503727e-01 3.30028325e-01
-3.64597887e-01 1.93243682e-01 1.97910100e-01 1.13197789e-01
7.34707057e-01 3.66670758e-01 2.95414537e-01 -3.04653347e-01
8.76936376e-01 2.51921058e-01 1.13526201e+00 8.45140815e-01
-1.87805906e-01 8.73683393e-01 3.28184247e-01 -5.82940504e-02
-6.67141378e-01 -8.30798745e-01 2.41068229e-01 9.49959457e-01
5.09830296e-01 -8.53014648e-01 -8.14144909e-01 -9.45185483e-01
-4.53318089e-01 1.03566635e+00 -2.74289310e-01 -3.32624257e-01
-6.47185445e-01 -5.91625333e-01 4.98096228e-01 9.00514483e-01
6.09707057e-01 -1.26872075e+00 -1.04302883e-01 3.20319623e-01
-2.83806920e-01 -1.15245593e+00 -8.55953872e-01 -2.14162260e-01
-6.64456189e-01 -9.91387248e-01 -7.00386763e-01 -1.10334551e+00
8.65667403e-01 4.90680128e-01 8.94963384e-01 7.55698264e-01
1.84974238e-01 -1.35402232e-01 -8.73294711e-01 -8.16129565e-01
-5.21401405e-01 -1.24228850e-01 -1.01539828e-01 4.12192605e-02
8.79906178e-01 9.48943645e-02 -3.79364967e-01 1.99446425e-01
-6.72046125e-01 4.43596184e-01 5.65786839e-01 6.92323387e-01
5.78499615e-01 -4.20362800e-02 2.45457321e-01 -9.60758865e-01
8.44853222e-01 -6.42259866e-02 -4.91311401e-01 5.87807655e-01
-8.38189244e-01 9.71262977e-02 8.25413883e-01 -5.74886084e-01
-1.35553396e+00 -4.60197143e-02 -6.83510751e-02 -2.43340582e-01
-1.26391798e-01 7.43887365e-01 -3.46425414e-01 4.93796989e-02
5.31366646e-01 7.66879320e-01 -4.48362857e-01 -7.67959654e-01
-5.60715050e-02 8.62042129e-01 4.02002364e-01 -5.17862678e-01
4.26121891e-01 1.10022463e-01 -6.62261248e-01 -6.15211487e-01
-9.33708847e-01 -4.12006676e-01 -4.83305991e-01 -9.70520005e-02
6.49659932e-01 -8.79980087e-01 -3.61162245e-01 7.48840153e-01
-1.35983908e+00 -3.51505369e-01 1.59664527e-01 6.26654506e-01
-1.24901265e-01 8.71050537e-01 -6.35181725e-01 -8.22192669e-01
-2.22630963e-01 -9.79602933e-01 8.17158937e-01 6.06217086e-01
-1.13849498e-01 -8.46621931e-01 -8.43861029e-02 1.97632343e-01
1.31514445e-01 -7.09590077e-01 1.10862613e+00 -6.06336594e-01
-5.98395467e-01 4.81502563e-02 -4.76213008e-01 4.54766840e-01
-9.10063386e-02 1.96180359e-01 -7.33307481e-01 9.62002352e-02
-3.13913912e-01 1.82713956e-01 8.18712890e-01 1.23549692e-01
1.46449733e+00 -2.81648636e-01 -3.77356976e-01 5.30927658e-01
1.26110160e+00 3.28718394e-01 9.75485623e-01 5.78650713e-01
7.94481933e-01 2.39426091e-01 8.56353581e-01 4.48725432e-01
6.76076055e-01 5.10886073e-01 2.46342614e-01 3.02310467e-01
-2.16418579e-01 -7.19059229e-01 5.41063726e-01 1.18449748e+00
-4.16196249e-02 -3.17321628e-01 -1.11657131e+00 6.12835050e-01
-1.98757410e+00 -6.38552547e-01 -6.56031668e-01 2.04245424e+00
1.02409124e+00 2.22814411e-01 -4.04209822e-01 1.72060356e-01
8.70737970e-01 -2.56231427e-02 -2.50219405e-02 -2.93630451e-01
-4.61968124e-01 6.05237633e-02 1.93524197e-01 3.53269547e-01
-9.56860900e-01 1.49427569e+00 5.77849150e+00 1.23291790e+00
-1.13769567e+00 1.21357277e-01 2.16735914e-01 4.17041600e-01
-6.13903403e-01 3.11291605e-01 -1.08663332e+00 9.19458210e-01
4.59585577e-01 -4.57897410e-02 3.80397439e-01 7.00177789e-01
2.69076824e-01 -1.74037397e-01 -7.10024595e-01 1.16737843e+00
3.23951721e-01 -1.63708377e+00 2.38036588e-01 -4.81570959e-01
8.67610931e-01 -2.39994034e-01 -2.16046676e-01 4.92781073e-01
2.13891581e-01 -9.81060922e-01 9.88348186e-01 7.28347600e-01
2.76886076e-01 -5.31576037e-01 7.23177969e-01 7.77668774e-01
-1.09588671e+00 -9.77211371e-02 -5.80612004e-01 -2.43921459e-01
-7.76812956e-02 3.28720748e-01 -5.09760618e-01 3.81652862e-01
8.21525097e-01 9.25176322e-01 -1.01922154e+00 1.41037798e+00
-8.25416505e-01 8.94886136e-01 1.62232012e-01 -6.82165086e-01
2.08132774e-01 -2.20036149e-01 3.79924715e-01 1.54406548e+00
5.18933237e-01 4.66422215e-02 2.62114089e-02 9.11736190e-01
1.25998080e-01 4.78530258e-01 1.26764014e-01 -2.85939723e-01
6.84542418e-01 7.03266680e-01 -6.02043033e-01 -2.64766097e-01
-5.57128966e-01 1.19613290e+00 6.11894310e-01 3.67475390e-01
-7.11271644e-01 -4.39113170e-01 4.27183032e-01 -1.54345945e-01
3.09401661e-01 -2.05862746e-01 -8.02384377e-01 -1.42829084e+00
6.12986088e-02 -9.59192336e-01 3.21985573e-01 -1.08250570e+00
-1.11856437e+00 4.04552668e-01 -5.75623453e-01 -1.45711184e+00
4.13084000e-01 -7.91899264e-01 -8.60902905e-01 8.43286633e-01
-1.98602450e+00 -9.87670422e-01 -2.57197618e-01 4.14600015e-01
8.08514833e-01 -2.27389112e-01 6.08563542e-01 2.62144923e-01
-8.25839281e-01 7.42928624e-01 7.68829361e-02 2.94594824e-01
7.01964080e-01 -1.14169264e+00 1.84450135e-01 1.53441536e+00
2.05012128e-01 6.46874428e-01 5.92477739e-01 -9.99069989e-01
-9.39767957e-01 -1.16951108e+00 1.54514527e+00 -2.38432869e-01
3.81888956e-01 1.14154657e-02 -1.26048505e+00 4.12939250e-01
-1.99725349e-02 -1.90396845e-01 5.14644384e-01 -4.75741401e-02
-2.26520911e-01 2.29423612e-01 -4.89758998e-01 9.69832659e-01
1.09897149e+00 -4.42903847e-01 -8.44083548e-01 1.71517238e-01
6.05726004e-01 -4.32096690e-01 -1.12809546e-01 9.57724378e-02
3.81077081e-02 -7.81289220e-01 4.27517176e-01 -7.18641877e-01
5.45424879e-01 -6.78074539e-01 3.14794518e-02 -1.46333456e+00
-6.00332439e-01 -4.81118202e-01 -8.12963583e-03 1.31257713e+00
5.18500924e-01 -2.13623390e-01 6.55637622e-01 4.60712641e-01
-5.97264290e-01 -4.13909763e-01 -5.73064446e-01 -7.10707605e-01
1.52513549e-01 -7.62484193e-01 8.67007196e-01 8.86848867e-01
1.76561475e-01 -3.19085307e-02 -4.19291705e-01 4.57679987e-01
1.00037239e-01 -2.76163016e-02 4.58050907e-01 -1.10579109e+00
-2.23159522e-01 -7.03311205e-01 3.03257462e-02 -1.42374325e+00
1.98799402e-01 -9.15088952e-01 3.49987239e-01 -1.48930895e+00
3.18711936e-01 -5.87119460e-01 -2.46261731e-01 4.18900937e-01
-9.54743981e-01 -1.82852671e-01 3.91285866e-01 3.12636524e-01
-9.12202001e-01 6.17958188e-01 1.36037135e+00 -4.40601595e-02
-2.76477218e-01 -2.25285012e-02 -8.00642014e-01 9.21608031e-01
8.34484816e-01 -5.58061063e-01 -2.03315984e-03 -8.48927319e-01
6.25834107e-01 -3.84956867e-01 1.89805776e-01 -8.55258942e-01
5.82819223e-01 -6.07576370e-01 3.75685394e-01 -5.48320115e-01
-1.89298749e-01 -5.72799921e-01 -3.51604283e-01 4.73993212e-01
-3.79871339e-01 1.03649139e-01 1.26488894e-01 5.70950866e-01
-2.10091978e-01 -6.14247382e-01 6.75306141e-01 -6.67637959e-02
-1.25408208e+00 2.56704297e-02 -6.70631051e-01 2.82014102e-01
7.66019702e-01 -4.14237320e-01 -3.73349875e-01 -1.29183054e-01
-2.18410537e-01 3.28420252e-01 3.06099951e-01 6.63481712e-01
8.62176836e-01 -1.19961941e+00 -6.34867072e-01 2.99445957e-01
4.45533395e-01 -1.59852877e-01 1.81401551e-01 7.45652020e-01
-7.61341214e-01 3.59337360e-01 -8.76148790e-03 -2.79573381e-01
-1.19456780e+00 4.09455806e-01 3.34350288e-01 -1.06773920e-01
-5.32378137e-01 8.93294752e-01 -2.64422223e-02 -4.03393090e-01
3.10219646e-01 -3.27272594e-01 -6.53179526e-01 -4.41562116e-01
8.90798509e-01 3.13356817e-01 2.68978328e-02 -4.42115873e-01
-3.94313335e-01 4.41826284e-01 -2.44465634e-01 2.74283707e-01
1.09490824e+00 -3.21089476e-01 -6.12608828e-02 2.22858578e-01
6.10920131e-01 3.40204537e-01 -1.29191756e+00 -4.68270630e-01
4.81331557e-01 -6.99098825e-01 6.42956272e-02 -9.24867570e-01
-9.36365426e-01 9.98101890e-01 3.44616652e-01 -4.53689307e-01
9.70428467e-01 -2.16351956e-01 8.19896460e-01 3.60920578e-01
1.20992422e-01 -1.74652612e+00 1.55904949e-01 8.99709761e-01
7.08599091e-01 -1.16641557e+00 -2.14239135e-01 -7.06681430e-01
-1.10008359e+00 1.23653507e+00 1.01215315e+00 -2.86134854e-02
4.59666997e-01 6.93951398e-02 5.20435460e-02 2.89085567e-01
-5.61658740e-01 -3.20806444e-01 4.02501911e-01 7.55189717e-01
4.46465552e-01 -4.21218202e-02 -7.77964175e-01 1.30215406e+00
-2.36126985e-02 1.52433485e-01 5.19902468e-01 8.51208091e-01
-6.51295900e-01 -1.13236344e+00 -3.07185650e-01 2.45706826e-01
-2.57390678e-01 -5.53978622e-01 -3.61619860e-01 3.91790450e-01
2.97878951e-01 1.06409812e+00 1.73074007e-02 -4.07769054e-01
1.60854295e-01 -5.74687719e-02 9.59718972e-02 -7.32951760e-01
-5.51105559e-01 4.96189781e-02 1.11636790e-02 -3.34630668e-01
-2.14126587e-01 -6.18978143e-01 -1.55717874e+00 -1.27793714e-01
-5.90135276e-01 1.67993888e-01 3.73303175e-01 1.39646173e+00
5.24619579e-01 4.55240011e-01 4.47085083e-01 -6.82982504e-02
-4.21594679e-01 -8.26526284e-01 -4.26567107e-01 5.41433871e-01
-2.05658153e-01 -2.84881473e-01 -7.33067393e-02 2.80612875e-02]
|
[10.939549446105957, 10.83447551727295]
|
df528e26-2ec5-4e23-bfcf-f4eb2dc84370
|
few-shot-learning-with-siamese-networks-and-1
|
2203.14655
| null |
https://arxiv.org/abs/2203.14655v2
|
https://arxiv.org/pdf/2203.14655v2.pdf
|
Few-Shot Learning with Siamese Networks and Label Tuning
|
We study the problem of building text classifiers with little or no training data, commonly known as zero and few-shot text classification. In recent years, an approach based on neural textual entailment models has been found to give strong results on a diverse range of tasks. In this work, we show that with proper pre-training, Siamese Networks that embed texts and labels offer a competitive alternative. These models allow for a large reduction in inference cost: constant in the number of labels rather than linear. Furthermore, we introduce label tuning, a simple and computationally efficient approach that allows to adapt the models in a few-shot setup by only changing the label embeddings. While giving lower performance than model fine-tuning, this approach has the architectural advantage that a single encoder can be shared by many different tasks.
|
['Marc Franco-Salvador', 'Guillermo Pérez-Torró', 'Thomas Müller']
|
2022-03-28
| null |
https://aclanthology.org/2022.acl-long.584
|
https://aclanthology.org/2022.acl-long.584.pdf
|
acl-2022-5
|
['few-shot-text-classification']
|
['natural-language-processing']
|
[ 2.92846262e-01 5.71299121e-02 -3.81717116e-01 -6.70244396e-01
-8.01096976e-01 -4.80407864e-01 9.02029514e-01 4.97217894e-01
-9.19995546e-01 5.92413068e-01 1.27670079e-01 -1.67107522e-01
4.07500044e-02 -6.12367749e-01 -5.48415780e-01 -5.61900198e-01
2.85143971e-01 5.79792082e-01 3.30421031e-01 -1.44155100e-01
3.66992146e-01 8.45232978e-02 -1.67160034e+00 2.97444612e-01
4.84131098e-01 9.06359434e-01 -5.30355126e-02 8.68455946e-01
-3.19208831e-01 1.11661220e+00 -4.02764171e-01 -9.26689267e-01
6.80404678e-02 -3.81492019e-01 -1.02002835e+00 -1.10537745e-02
5.47794282e-01 -3.07244003e-01 -3.24422240e-01 8.90046120e-01
4.99110281e-01 4.93673295e-01 8.73970687e-01 -1.01351082e+00
-6.91457272e-01 9.31095660e-01 -3.33479941e-01 1.37262672e-01
1.56099666e-02 -1.81237161e-01 1.51285684e+00 -9.50426102e-01
6.07679129e-01 1.08605444e+00 8.53508115e-01 5.77819288e-01
-1.57262242e+00 -3.27821195e-01 8.81967098e-02 3.18702698e-01
-1.14450097e+00 -5.42785764e-01 5.37750900e-01 -3.27356219e-01
1.27765465e+00 9.86383483e-02 2.45449796e-01 1.26657832e+00
6.22398481e-02 7.05099583e-01 8.38953912e-01 -8.09569955e-01
4.14085388e-01 3.21980834e-01 4.15552258e-01 7.60346413e-01
1.96624383e-01 -3.20773363e-01 -4.27706063e-01 -1.70810670e-01
-2.10278425e-02 1.61295548e-01 2.81092376e-02 -4.91300553e-01
-8.66821170e-01 1.29120350e+00 3.15615714e-01 5.33930063e-01
1.08334742e-01 4.55151021e-01 8.98005068e-01 5.42646825e-01
8.90646935e-01 6.96235836e-01 -5.28109908e-01 -1.40659854e-01
-1.22985911e+00 9.35438126e-02 1.06562960e+00 7.74194062e-01
6.88782394e-01 -1.52729273e-01 -3.25145781e-01 8.69912446e-01
5.93377510e-04 2.38478966e-02 8.68036628e-01 -8.70089114e-01
4.26449865e-01 1.36006713e-01 5.38040213e-02 -6.55002236e-01
-4.73356366e-01 -2.82876372e-01 -5.67289829e-01 1.53818130e-01
4.18416858e-01 -1.86667860e-01 -9.52047765e-01 1.81917059e+00
1.05759785e-01 -8.28556437e-03 1.81055423e-02 3.25930804e-01
2.97178954e-01 6.54842913e-01 1.86511099e-01 -1.23786218e-01
1.35944653e+00 -1.27445650e+00 -8.01672101e-01 -4.70336437e-01
1.24997282e+00 -6.77487373e-01 1.28637731e+00 2.81099737e-01
-1.08832085e+00 -2.34010413e-01 -1.20116889e+00 -4.83961403e-01
-7.59286106e-01 -1.49561867e-01 4.86041784e-01 7.12384820e-01
-8.23676229e-01 9.72477078e-01 -7.30138481e-01 -5.56540966e-01
4.29991841e-01 2.25707531e-01 -2.53332585e-01 -2.94680476e-01
-1.29803801e+00 1.31762981e+00 5.01667142e-01 -4.78188127e-01
-7.31221855e-01 -5.76241553e-01 -9.57999706e-01 4.42496419e-01
6.38347924e-01 -5.03114283e-01 1.64538610e+00 -9.15964186e-01
-1.63041461e+00 9.92610812e-01 -1.90159410e-01 -6.89259946e-01
5.59630811e-01 -8.47179517e-02 4.50498424e-02 -2.20056716e-02
-7.28974640e-02 4.21240956e-01 8.01182389e-01 -6.83514535e-01
-5.87698519e-01 -2.55310684e-01 1.06634401e-01 1.57108828e-01
-7.96230257e-01 2.42893979e-01 -1.72511563e-01 -6.01393700e-01
-3.44815373e-01 -9.06605482e-01 -1.18908919e-01 1.90340132e-01
-1.13846883e-01 -6.19586587e-01 4.27680343e-01 -2.88789302e-01
1.21260464e+00 -2.00147963e+00 2.02931955e-01 -2.36924529e-01
1.08747415e-01 3.87866020e-01 -1.16886683e-01 5.91910303e-01
7.49299079e-02 3.72073859e-01 -3.36491764e-01 -8.16094935e-01
4.01026040e-01 2.84647912e-01 -1.26614466e-01 3.73495758e-01
2.20661297e-01 1.01356411e+00 -7.69164681e-01 -5.25922179e-01
1.83010008e-02 2.03668430e-01 -6.05029523e-01 7.90595040e-02
-3.78344178e-01 -3.52342725e-01 -5.71003556e-02 -3.42543907e-02
2.15983585e-01 -5.71387589e-01 2.08884582e-01 2.50207543e-01
1.44184083e-01 4.86198008e-01 -1.10870039e+00 1.92094672e+00
-7.00378478e-01 8.41409743e-01 -1.12654790e-01 -1.34928370e+00
5.03156841e-01 5.58336258e-01 9.37460214e-02 -3.86059463e-01
3.81716609e-01 1.60045967e-01 -2.13578582e-01 -6.14292681e-01
4.04053062e-01 -6.60311878e-01 -2.87968427e-01 8.81298959e-01
5.39854705e-01 2.63043214e-02 4.59086180e-01 2.01290905e-01
1.18123913e+00 -9.20245945e-02 5.51925659e-01 -8.21461380e-02
3.02206665e-01 -6.03349246e-02 9.32455733e-02 8.72912347e-01
-7.13156462e-02 3.82764459e-01 4.98748124e-01 -3.55644971e-01
-1.42137599e+00 -5.44000149e-01 -2.89112180e-01 1.84140086e+00
-2.22400054e-01 -5.33792734e-01 -7.65021920e-01 -6.06420338e-01
-7.13750497e-02 9.45879161e-01 -9.52143848e-01 -4.73251820e-01
-5.57510734e-01 -7.50854135e-01 6.50937378e-01 6.01507902e-01
-4.88581322e-03 -8.53864968e-01 -6.12964869e-01 2.40358427e-01
-1.59440585e-03 -1.03646564e+00 -4.84027445e-01 9.28854942e-01
-7.90853798e-01 -6.14542246e-01 -7.42853343e-01 -9.38146234e-01
4.17343765e-01 1.56509876e-01 1.15250647e+00 -3.80496792e-02
-3.95565033e-01 1.91223789e-02 -4.75891501e-01 -2.64911413e-01
-3.92516434e-01 6.47841334e-01 -1.89462975e-01 -1.10759921e-01
6.20366037e-01 -2.77283549e-01 -1.19796284e-01 -5.38041964e-02
-1.04953337e+00 -1.29854962e-01 3.54347944e-01 1.19444120e+00
1.06334262e-01 -7.04807490e-02 6.28053188e-01 -1.36169100e+00
7.21993387e-01 -5.28794944e-01 -2.80246049e-01 3.57987642e-01
-9.84990835e-01 4.91561055e-01 1.01259983e+00 -5.76211274e-01
-1.05417073e+00 -2.71545537e-02 -1.10305220e-01 -1.68546975e-01
-1.68088049e-01 5.53287268e-01 2.38434896e-01 4.60245125e-02
9.75309193e-01 -2.84939166e-02 -8.58712867e-02 -4.95706528e-01
7.93187261e-01 7.73238659e-01 1.64621100e-01 -3.67899209e-01
5.38901508e-01 3.90504450e-01 -1.49016470e-01 -5.63893378e-01
-1.46529460e+00 -5.51760435e-01 -8.38661075e-01 3.08204293e-01
8.79930079e-01 -6.83184147e-01 -2.44999528e-01 1.59496158e-01
-1.01531291e+00 -4.98665601e-01 -5.04303038e-01 4.20730770e-01
-6.43146634e-01 3.50098670e-01 -9.50021148e-01 -5.96760154e-01
-2.83556014e-01 -7.43148327e-01 7.52627790e-01 -7.66958669e-02
-3.92079085e-01 -1.22667158e+00 1.63393974e-01 -6.97952230e-03
5.46530724e-01 -1.96927413e-01 1.14345241e+00 -1.19968724e+00
-5.05850241e-02 -5.76972425e-01 -1.02123946e-01 4.46690351e-01
-1.03556044e-01 -1.56891704e-01 -1.28831589e+00 -2.94222236e-01
1.25547215e-01 -8.23521554e-01 1.05196130e+00 1.02744304e-01
9.44744825e-01 -2.63280690e-01 -1.55027047e-01 4.65660989e-01
1.48914957e+00 -8.80796239e-02 3.05437267e-01 2.24633306e-01
5.91851294e-01 6.34352028e-01 4.01538670e-01 4.43456382e-01
1.79820985e-01 5.93341172e-01 3.99825461e-02 1.06547229e-01
-4.35642451e-02 -1.85078084e-01 1.76220059e-01 8.07479382e-01
3.24740052e-01 -4.58821803e-01 -7.92734206e-01 4.90876764e-01
-1.97219396e+00 -1.05431366e+00 2.02272668e-01 2.13411164e+00
1.07522750e+00 3.10135692e-01 1.85083821e-02 2.95022279e-01
5.88305831e-01 1.20186649e-01 -3.56936455e-01 -6.70680225e-01
6.85340539e-02 5.96398056e-01 5.46232402e-01 6.61050797e-01
-1.16832757e+00 9.40147042e-01 7.00003004e+00 8.43883097e-01
-8.97893667e-01 5.17840981e-01 3.28098387e-01 -5.80800831e-01
-9.77162421e-02 2.16980302e-03 -8.16749096e-01 5.54451644e-01
1.42037177e+00 -3.68268698e-01 3.99714917e-01 8.96715522e-01
-2.70143896e-01 2.26695612e-02 -1.54994583e+00 6.76286817e-01
5.37208319e-01 -1.20226943e+00 -1.42625079e-01 -1.85047477e-01
8.31641376e-01 1.07353412e-01 -2.26086676e-01 5.80557764e-01
5.04368186e-01 -8.80391359e-01 6.18002295e-01 2.41518214e-01
7.03454673e-01 -6.88789725e-01 9.17857885e-01 5.97405612e-01
-7.87621498e-01 -2.68759489e-01 -5.97475231e-01 -2.10837394e-01
1.25272751e-01 3.09944570e-01 -6.24745190e-01 9.01257023e-02
3.67089927e-01 5.39365768e-01 -5.34401059e-01 8.78591478e-01
-2.25944355e-01 5.91148198e-01 -2.14835227e-01 -3.64983737e-01
4.61229950e-01 2.66472280e-01 5.66644184e-02 1.55448222e+00
2.00576305e-01 -2.04639316e-01 1.94474712e-01 4.73051786e-01
-3.67479444e-01 3.33782494e-01 -7.15725958e-01 2.97021065e-02
4.10149366e-01 1.13189137e+00 -6.33197665e-01 -6.64121568e-01
-6.41235769e-01 1.19921517e+00 8.20649624e-01 8.95217508e-02
-7.35555351e-01 -8.49296272e-01 2.08053067e-01 -5.41158067e-03
6.52401209e-01 -7.00581372e-02 -2.02683344e-01 -1.29526687e+00
-9.34414789e-02 -5.26623130e-01 4.54011708e-01 -4.78122562e-01
-1.46530330e+00 3.86384785e-01 -5.58247492e-02 -7.68634677e-01
-5.67404389e-01 -8.86305451e-01 -4.01994586e-01 6.13920450e-01
-1.54013062e+00 -8.98609519e-01 1.73444692e-02 2.34628424e-01
7.45002031e-01 2.80644093e-02 1.12427795e+00 3.74364972e-01
-5.86528718e-01 7.38722324e-01 5.68611443e-01 9.76145715e-02
9.43194151e-01 -1.42669666e+00 3.48923773e-01 5.78558087e-01
3.67648333e-01 4.38185990e-01 8.16955030e-01 -1.10324427e-01
-8.17753077e-01 -9.79253113e-01 1.41778672e+00 -4.83220875e-01
9.05212402e-01 -6.60135329e-01 -1.07571745e+00 8.87751102e-01
3.76958787e-01 6.69150203e-02 1.01742744e+00 4.21883762e-01
-6.53572798e-01 6.37163222e-02 -9.83758450e-01 4.93791074e-01
7.58998334e-01 -8.84159863e-01 -8.33036363e-01 5.75625777e-01
6.20307565e-01 6.15799986e-02 -7.23571002e-01 -1.13813214e-01
4.04961854e-01 -6.16222620e-01 6.25690579e-01 -9.07939970e-01
6.62642717e-01 2.07785994e-01 -1.29654050e-01 -1.41266429e+00
-4.23604786e-01 -3.70212734e-01 -3.18215817e-01 1.17070639e+00
5.81756592e-01 -4.99563187e-01 7.86628187e-01 7.25511789e-01
-3.07829026e-02 -6.60725176e-01 -8.74492526e-01 -9.24726248e-01
5.51183939e-01 -3.12326878e-01 8.38343129e-02 1.04807591e+00
4.61939305e-01 9.93647456e-01 -6.00214064e-01 -5.20373404e-01
4.02901441e-01 -1.47124469e-01 3.67617011e-01 -1.57575250e+00
-4.25865084e-01 -5.10847867e-01 -2.52684832e-01 -8.61350834e-01
4.99504030e-01 -1.35492003e+00 2.59014755e-01 -1.37482369e+00
4.37866688e-01 -2.48267144e-01 -4.22723949e-01 5.78825593e-01
-2.27303252e-01 3.05833608e-01 1.18010417e-01 1.32596239e-01
-7.42634833e-01 3.84190351e-01 6.71477914e-01 -1.46028712e-01
2.51054466e-01 -1.56809866e-01 -6.11419559e-01 6.99777603e-01
8.45077872e-01 -9.19152558e-01 -4.13075566e-01 -6.05492651e-01
4.52367187e-01 -2.89335638e-01 5.25220390e-03 -8.82062018e-01
4.18273091e-01 2.04334021e-01 1.78637415e-01 -1.27550229e-01
4.39285666e-01 -8.09286594e-01 -4.71227109e-01 3.99350405e-01
-1.12885702e+00 -1.81345642e-01 -1.09114587e-01 7.27166891e-01
-8.32585469e-02 -1.20732963e+00 1.11738908e+00 -3.36718053e-01
-4.25996184e-01 8.16731900e-02 -5.47519743e-01 2.38205686e-01
1.11788428e+00 2.37583872e-02 -2.18504027e-01 -2.87314266e-01
-7.40254164e-01 2.37655565e-02 4.20618892e-01 2.51544446e-01
2.37814505e-02 -1.22036815e+00 -5.92469096e-01 -1.11273304e-01
3.37983847e-01 -3.59986901e-01 1.56347230e-02 5.14340520e-01
-2.09422365e-01 5.39275289e-01 7.70613775e-02 -2.87744582e-01
-1.13651645e+00 9.21611547e-01 2.67453194e-01 -5.98652780e-01
-7.32989490e-01 8.93186808e-01 -8.13939869e-02 -5.67209065e-01
3.74795884e-01 -1.68340057e-01 -5.55837192e-02 5.08873284e-01
7.32719719e-01 2.74150968e-01 2.76491851e-01 -2.90648580e-01
-4.87896502e-02 4.09881592e-01 -3.88966888e-01 -1.93340212e-01
1.19327343e+00 -1.65116131e-01 1.75558537e-01 1.08213151e+00
1.52309954e+00 -4.22131300e-01 -1.13363135e+00 -6.56693161e-01
4.47516829e-01 -2.29105353e-01 1.98210016e-01 -6.49796307e-01
-5.42218447e-01 1.25641692e+00 3.78327966e-01 5.12244403e-01
5.76851726e-01 4.21039248e-03 7.01998413e-01 8.76159132e-01
1.00961804e-01 -1.46060240e+00 3.01387936e-01 7.55155683e-01
2.22681820e-01 -1.38917434e+00 -2.82615554e-02 2.09346544e-02
-6.62860453e-01 1.09773588e+00 2.73263395e-01 -1.45600170e-01
6.30787790e-01 4.04681355e-01 -1.74207196e-01 -5.80906458e-02
-1.20727861e+00 -2.40884468e-01 5.61552867e-02 2.33317152e-01
6.69130385e-01 -1.29017070e-01 -3.07229817e-01 3.87541831e-01
-2.43188310e-02 2.70129219e-02 4.96960759e-01 9.85010505e-01
-8.10498059e-01 -1.09207046e+00 1.95063889e-01 6.93925619e-01
-8.25253248e-01 -3.60452205e-01 -2.39519477e-01 6.40719831e-01
-7.24658668e-02 8.55496228e-01 2.51344144e-01 -7.77009353e-02
1.08390152e-01 9.55508649e-01 4.48374659e-01 -1.05188084e+00
-6.88251078e-01 -2.79559523e-01 2.27025539e-01 -1.06706917e-01
-2.44775981e-01 -3.76729101e-01 -1.05848086e+00 -3.38762462e-01
-5.95352054e-01 2.23554119e-01 6.32072091e-01 1.18294454e+00
9.65836272e-02 5.07842720e-01 5.32957315e-01 -7.48711765e-01
-1.24876690e+00 -1.15766180e+00 -7.25447655e-01 5.70158243e-01
2.79813439e-01 -4.82058585e-01 -6.39884055e-01 1.60947725e-01]
|
[10.689817428588867, 7.871866703033447]
|
5af36480-a753-4334-b885-2ed80c1ecebd
|
causal-temporal-graph-convolutional-neural
|
2303.09634
| null |
https://arxiv.org/abs/2303.09634v1
|
https://arxiv.org/pdf/2303.09634v1.pdf
|
Causal Temporal Graph Convolutional Neural Networks (CTGCN)
|
Many large-scale applications can be elegantly represented using graph structures. Their scalability, however, is often limited by the domain knowledge required to apply them. To address this problem, we propose a novel Causal Temporal Graph Convolutional Neural Network (CTGCN). Our CTGCN architecture is based on a causal discovery mechanism, and is capable of discovering the underlying causal processes. The major advantages of our approach stem from its ability to overcome computational scalability problems with a divide and conquer technique, and from the greater explainability of predictions made using a causal model. We evaluate the scalability of our CTGCN on two datasets to demonstrate that our method is applicable to large scale problems, and show that the integration of causality into the TGCN architecture improves prediction performance up to 40% over typical TGCN approach. Our results are obtained without requiring additional domain knowledge, making our approach adaptable to various domains, specifically when little contextual knowledge is available.
|
['Joern Ploennigs', 'Christopher Lohse', 'Fabio Lorenzi', 'Amadou Ba', "Fearghal O'Donncha", 'Abigail Langbridge']
|
2023-03-16
| null | null | null | null |
['causal-discovery']
|
['knowledge-base']
|
[ 1.85403705e-01 4.38265979e-01 -4.40132290e-01 -2.30572268e-01
-9.84268188e-02 -5.04566491e-01 9.31896150e-01 3.56578827e-01
1.93802908e-01 8.60323548e-01 5.66753328e-01 -7.56521583e-01
-6.04223490e-01 -1.11751759e+00 -8.94978642e-01 -1.17685318e-01
-5.12274146e-01 4.49555486e-01 4.81408358e-01 -2.21176073e-01
6.90937936e-02 5.34397542e-01 -1.18838131e+00 3.42819363e-01
7.74208128e-01 7.81886637e-01 -1.27977446e-01 5.81654370e-01
-1.16865270e-01 1.29022312e+00 -3.53249520e-01 -1.02459639e-01
1.17888348e-02 -5.01334071e-01 -9.74547386e-01 -2.92792141e-01
1.18439510e-01 -3.04704726e-01 -4.86627907e-01 4.11670029e-01
-4.91663367e-02 6.46591699e-03 4.76297438e-01 -1.55595958e+00
-6.46334171e-01 9.27080512e-01 -3.43930304e-01 2.81525791e-01
3.50986362e-01 -1.91320740e-02 1.47191167e+00 -3.35866392e-01
7.21561790e-01 1.46096885e+00 8.95877779e-01 2.89538950e-01
-1.65607524e+00 -6.15216315e-01 6.39507353e-01 2.36486018e-01
-1.00979841e+00 -1.89747855e-01 5.74171424e-01 -3.99483860e-01
1.29051661e+00 7.02207759e-02 5.93973398e-01 1.00971019e+00
2.15447322e-01 5.93144298e-01 8.34041476e-01 -2.24683017e-01
2.58119553e-01 -5.38398325e-01 8.38369280e-02 7.38004684e-01
3.72706980e-01 2.72119671e-01 -6.38957858e-01 -3.42892528e-01
9.08331156e-01 -5.47695160e-02 -1.01209149e-01 -4.46997195e-01
-1.08237183e+00 9.51768041e-01 7.58431554e-01 3.03832740e-01
-3.45874310e-01 8.66939485e-01 2.98716962e-01 2.77506232e-01
4.70958501e-01 5.05387187e-01 -5.78576624e-01 -2.96170916e-02
-8.15393209e-01 3.57017845e-01 1.04870141e+00 7.77397633e-01
4.58456904e-01 -9.00638383e-03 3.62822898e-02 3.31052065e-01
1.53746888e-01 1.80872500e-01 1.72941402e-01 -8.07489812e-01
1.79951742e-01 1.01509547e+00 -4.66096289e-02 -1.20355368e+00
-6.91792846e-01 -2.59388298e-01 -6.45392239e-01 -6.20309934e-02
3.32954198e-01 -1.97320879e-01 -9.41988349e-01 1.86302745e+00
1.28104195e-01 3.60497326e-01 -1.33383885e-01 6.01805031e-01
7.15971291e-01 4.92122442e-01 3.49568903e-01 -1.39903829e-01
1.01071036e+00 -5.91097057e-01 -4.89673734e-01 -3.71802568e-01
6.99694037e-01 -2.33453810e-01 8.36931586e-01 1.61866844e-01
-6.26108587e-01 -6.92730695e-02 -9.90424693e-01 -5.78687266e-02
-2.87888616e-01 -3.17240775e-01 1.24138629e+00 3.22778165e-01
-1.24729264e+00 9.34024751e-01 -1.05805421e+00 -7.02868521e-01
4.92553473e-01 5.50640583e-01 -4.30156827e-01 -6.67451322e-02
-1.26572227e+00 7.06850231e-01 6.05162382e-01 -1.11466259e-01
-7.67874002e-01 -9.06864345e-01 -6.54308796e-01 4.77721363e-01
5.88681221e-01 -9.57228959e-01 1.28150332e+00 -7.92891979e-01
-1.04993260e+00 7.79502392e-02 -2.03635246e-01 -6.78984165e-01
3.58395904e-01 -6.55718818e-02 -4.12894309e-01 4.90002744e-02
7.79048204e-02 3.30963939e-01 3.74137998e-01 -9.74961936e-01
-5.42023301e-01 -6.09018318e-02 2.68831193e-01 -2.34225988e-01
-3.14883918e-01 3.79454345e-02 -3.02887917e-01 -4.91500169e-01
1.93614718e-02 -8.43488693e-01 -5.12560129e-01 -2.44728148e-01
-4.57233399e-01 -3.40529114e-01 8.57470036e-01 -3.07523549e-01
1.35557222e+00 -1.84636974e+00 1.46319762e-01 1.72121376e-01
5.96648455e-01 -7.31403455e-02 -2.09012568e-01 8.42447639e-01
-3.10887694e-01 5.90359032e-01 -3.00855726e-01 1.70372233e-01
-1.57541752e-01 5.26561439e-01 -4.22334135e-01 -4.84009786e-03
6.33277178e-01 1.21892262e+00 -1.05527151e+00 -3.13101709e-01
-8.63329247e-02 2.35910520e-01 -6.49540722e-01 1.50520161e-01
-7.31792331e-01 1.84222534e-01 -5.56601286e-01 2.29316577e-01
8.17184523e-02 -7.94535518e-01 9.17905629e-01 8.67450908e-02
4.29764800e-02 6.94763184e-01 -1.00977314e+00 1.26049173e+00
-3.95416290e-01 6.64994955e-01 -2.59452820e-01 -1.06274831e+00
6.21083081e-01 4.49414939e-01 5.46369851e-01 -5.96386969e-01
-1.89418271e-01 3.20483409e-02 1.68444738e-01 -2.14710772e-01
1.67255878e-01 -2.47678578e-01 6.23486424e-03 7.52138317e-01
-9.24363062e-02 2.24432349e-01 4.58471358e-01 6.77788436e-01
1.74406707e+00 1.71055123e-01 5.77666819e-01 -3.13183904e-01
-3.56908888e-02 3.66985202e-01 7.96031296e-01 7.42689908e-01
1.58668637e-01 1.32200882e-01 1.33316147e+00 -8.56808722e-01
-8.57677519e-01 -9.08876300e-01 1.53690904e-01 8.75310957e-01
-2.57873058e-01 -7.38113344e-01 -2.44250283e-01 -9.27367270e-01
3.50115567e-01 6.77259743e-01 -8.98361862e-01 -8.71959254e-02
-6.74813211e-01 -8.51954818e-01 5.30203462e-01 8.87418509e-01
2.37412751e-01 -9.17340636e-01 -4.90197599e-01 5.02084851e-01
-1.63619667e-01 -1.18110871e+00 1.06479160e-01 1.20214142e-01
-1.14698350e+00 -1.31453645e+00 6.41679317e-02 -8.83028060e-02
2.73698002e-01 1.57862365e-01 1.39540601e+00 1.60262495e-01
-2.47675762e-01 2.49646842e-01 -1.90698162e-01 -3.57041985e-01
-5.27504861e-01 2.86335826e-01 -1.12008177e-01 -3.92618120e-01
2.54716158e-01 -1.12022805e+00 -4.97516036e-01 2.98094898e-01
-8.13294411e-01 1.20878965e-01 5.52485347e-01 7.19918847e-01
1.11268617e-01 2.55786687e-01 8.38826895e-01 -1.21939611e+00
7.25126743e-01 -6.67661488e-01 -6.74876153e-01 4.31348607e-02
-1.10059559e+00 2.95521855e-01 6.17772698e-01 -2.57574797e-01
-9.57023203e-01 -4.91381399e-02 3.16015959e-01 -2.18527466e-01
-2.64421385e-02 1.00869155e+00 1.18256018e-01 2.01441556e-01
7.70256579e-01 -3.96858871e-01 -2.97357868e-02 -3.04981828e-01
6.85021341e-01 -9.87150669e-02 3.08343828e-01 -6.39125228e-01
6.69616461e-01 4.37981367e-01 4.80234534e-01 -3.58718067e-01
-6.11809313e-01 -1.45483002e-01 -6.42489076e-01 -8.09769332e-03
6.65080667e-01 -7.92757034e-01 -9.63938117e-01 -2.19720647e-01
-1.12336493e+00 -6.24693513e-01 -8.68062675e-03 2.27019802e-01
-3.42665702e-01 1.58984169e-01 -7.75139093e-01 -6.14008367e-01
-9.67107862e-02 -6.40111208e-01 6.13243878e-01 -1.98002309e-01
-5.57011724e-01 -1.22808111e+00 -2.35044677e-02 -7.59503841e-02
3.97605360e-01 5.83076954e-01 1.49132466e+00 -6.99411809e-01
-8.64630878e-01 -2.34419629e-02 -5.78792512e-01 -2.69123644e-01
2.26235017e-01 3.95036638e-02 -6.76748276e-01 -2.05482244e-02
-7.48037875e-01 -2.58578807e-01 9.95387971e-01 3.29051793e-01
9.70377862e-01 -4.42169189e-01 -6.87991321e-01 1.50846750e-01
1.36130643e+00 -4.64206077e-02 5.27884007e-01 3.09623629e-01
8.93891335e-01 5.35843790e-01 1.38639346e-01 2.50979424e-01
5.58414161e-01 5.44740438e-01 7.29090631e-01 -1.81864426e-01
-2.54136622e-01 -4.96322185e-01 1.12418443e-01 3.39233518e-01
-3.32118571e-01 -4.22517568e-01 -1.43707430e+00 6.95132911e-01
-2.30298376e+00 -9.81775880e-01 -5.02884924e-01 1.71032500e+00
6.85870528e-01 1.90750048e-01 1.60408884e-01 3.50526832e-02
3.42868537e-01 1.43751591e-01 -4.84876484e-01 -4.40110415e-01
1.59264341e-01 2.61545181e-01 5.48688412e-01 3.16421926e-01
-1.04882205e+00 1.06108868e+00 7.20057201e+00 4.36530054e-01
-9.78252351e-01 -9.61011574e-02 3.96460891e-01 -1.39564127e-01
-4.33917642e-01 3.55116844e-01 -3.33187908e-01 -9.76900980e-02
1.20403159e+00 -4.36044514e-01 4.91133839e-01 6.04415774e-01
4.92821634e-01 1.72161371e-01 -1.42691028e+00 2.96668410e-01
-5.74340880e-01 -1.66440046e+00 1.24238707e-01 2.25067258e-01
6.24038219e-01 2.92382985e-02 -2.23711282e-01 1.66338667e-01
1.21247387e+00 -1.39645743e+00 3.77888173e-01 1.85448542e-01
5.46352983e-01 -6.48470461e-01 4.29933250e-01 1.14954054e-01
-1.08899748e+00 -3.13955754e-01 -1.66235715e-01 -6.09946430e-01
4.55175266e-02 7.42944241e-01 -1.29980743e+00 8.83367598e-01
6.72380567e-01 9.57827926e-01 -4.24112111e-01 6.79892600e-01
-5.45505166e-01 1.04581201e+00 -2.45288894e-01 -6.34062439e-02
2.16786712e-01 2.64645010e-01 2.88470209e-01 1.14920545e+00
1.36541516e-01 2.79331058e-01 1.99545816e-01 1.03505588e+00
-1.93017840e-01 -1.42841533e-01 -7.55101979e-01 -4.82284546e-01
3.79989892e-01 9.56761599e-01 -8.36966276e-01 -2.30141491e-01
-4.00240153e-01 5.70128679e-01 6.23490810e-01 4.76394802e-01
-7.06272304e-01 -5.52759878e-02 5.30653238e-01 2.74671704e-01
3.93139213e-01 -3.77243042e-01 -4.82967407e-01 -1.00371230e+00
-1.82729378e-01 -8.49297702e-01 6.78994179e-01 -7.73440778e-01
-1.24652183e+00 5.27349710e-01 1.50433511e-01 -6.38108373e-01
-5.09299636e-01 -5.04764497e-01 -5.58150172e-01 8.17156494e-01
-1.32982850e+00 -1.38164485e+00 -1.59697548e-01 3.95511031e-01
-4.79530469e-02 1.88700318e-01 8.74372482e-01 1.13135621e-01
-4.04465765e-01 1.19263969e-01 -4.18995112e-01 8.93943384e-02
4.94588464e-01 -1.39682221e+00 7.50718832e-01 1.00512660e+00
3.17687951e-02 8.43717217e-01 7.83098340e-01 -9.49838340e-01
-1.35961366e+00 -1.31861222e+00 1.05153155e+00 -5.32290578e-01
1.20809174e+00 -4.75615740e-01 -9.72478509e-01 1.11511314e+00
1.79072693e-01 3.70759480e-02 5.74897349e-01 1.04586709e+00
-8.32045376e-01 -7.41057843e-02 -6.55402660e-01 6.61043823e-01
1.39125025e+00 -3.00667286e-01 -5.49319386e-01 2.55136788e-01
1.04304814e+00 -2.92314887e-02 -8.97587717e-01 4.81123060e-01
4.56468642e-01 -8.22009623e-01 7.41389215e-01 -1.05438590e+00
1.12549722e+00 -1.76875666e-01 1.16677107e-02 -1.30826390e+00
-5.98731339e-01 -4.50401515e-01 -2.58257240e-01 1.09929502e+00
7.25475252e-01 -7.00047433e-01 6.60801828e-01 7.51824737e-01
1.86463192e-01 -5.32325625e-01 -6.97064698e-01 -7.77883351e-01
6.96278140e-02 -7.01643527e-01 8.21105957e-01 1.21703625e+00
1.99454412e-01 6.16989374e-01 -3.78344089e-01 3.71678978e-01
5.01955450e-01 4.13813710e-01 7.28212297e-01 -1.64527380e+00
-5.05345166e-01 -4.84128058e-01 -3.45519423e-01 -6.58319116e-01
1.22043028e-01 -8.97653699e-01 -3.12569618e-01 -1.93488228e+00
1.56220645e-01 -6.00206077e-01 -3.35545510e-01 1.17384493e+00
-2.17188999e-01 4.88015562e-02 1.87781930e-01 2.45068967e-01
-3.21433872e-01 3.88475627e-01 9.31424439e-01 1.91188268e-02
-2.12691218e-01 -3.19698781e-01 -8.33064437e-01 6.33963168e-01
7.46248722e-01 -6.20071471e-01 -7.58092463e-01 -4.20195639e-01
5.63474715e-01 2.67410636e-01 6.32843316e-01 -6.84937954e-01
2.89011896e-01 -4.38472152e-01 2.33707324e-01 -1.41135409e-01
-1.57961756e-01 -5.50897479e-01 4.37539458e-01 4.27837759e-01
-4.21268702e-01 1.30644292e-01 5.52693427e-01 1.04104674e+00
-1.23961844e-01 4.39002097e-01 3.08969319e-01 -5.30273654e-02
-7.83223450e-01 1.88645169e-01 -3.53785783e-01 -2.57868767e-01
8.36293995e-01 2.27073193e-01 -6.08343244e-01 -5.54811358e-01
-5.90425432e-01 3.31644952e-01 2.56457001e-01 4.95978832e-01
3.10921669e-01 -1.19183588e+00 -4.79550600e-01 -1.14377633e-01
1.92357108e-01 -2.16816023e-01 -1.98889568e-01 7.23969638e-01
-3.07699114e-01 6.44535959e-01 2.50671264e-02 -3.79069149e-01
-9.53206778e-01 8.49612892e-01 3.03806633e-01 -6.12899721e-01
-8.56473207e-01 4.71763700e-01 3.17573279e-01 -2.71695942e-01
-1.42258495e-01 -5.43722630e-01 -4.15997505e-02 -2.21227989e-01
2.65048385e-01 2.01894477e-01 2.23007146e-02 1.01215661e-01
-4.52415377e-01 -1.65690798e-02 -3.03497934e-03 -3.20402123e-02
1.58116043e+00 1.94437727e-01 -4.43174183e-01 3.53963524e-01
6.30059719e-01 -9.45249274e-02 -1.06646001e+00 -1.64911732e-01
4.80696172e-01 -2.26990432e-01 1.30415767e-01 -1.18010557e+00
-9.09784198e-01 6.92590415e-01 -7.80018046e-02 6.00644588e-01
1.08606696e+00 7.19411448e-02 5.11503518e-01 3.10378015e-01
2.23480508e-01 -4.17073816e-01 -7.33853057e-02 5.03566325e-01
9.00294244e-01 -9.53263104e-01 1.75838947e-01 -6.84740722e-01
-2.66357571e-01 1.19380987e+00 4.50930357e-01 -7.93863684e-02
4.84687001e-01 2.41533577e-01 -1.38430789e-01 -5.97554147e-01
-1.38868296e+00 -1.78952202e-01 2.03044161e-01 3.49372715e-01
5.82467794e-01 2.28591278e-01 -3.53194237e-01 4.94142443e-01
1.07333921e-02 2.30305687e-01 6.91096485e-01 7.65124977e-01
-4.53757532e-02 -1.33511841e+00 1.19959163e-02 4.85088885e-01
-3.82057041e-01 -3.18504214e-01 -6.49658382e-01 1.16949272e+00
-2.16052815e-01 1.20735967e+00 2.74460800e-02 -4.18556929e-01
3.30708772e-01 8.88535380e-02 1.90757394e-01 -6.21442974e-01
-3.98563176e-01 -3.42085492e-03 5.03835797e-01 -9.41493928e-01
-3.49044263e-01 -5.43465316e-01 -1.31686509e+00 -6.15040720e-01
7.75198713e-02 1.20988768e-02 3.04800183e-01 8.30899358e-01
7.56535470e-01 9.94548023e-01 1.40287399e-01 -2.60750711e-01
-1.45515516e-01 -6.56975925e-01 -3.72577071e-01 2.56232679e-01
2.35875160e-01 -7.46829152e-01 -1.05327345e-01 6.33422062e-02]
|
[7.664956092834473, 5.928220748901367]
|
fb0c66dc-4881-404b-a52b-6cdc370b8c03
|
implementing-facial-landmark-tracking-for
|
2211.12723
| null |
https://arxiv.org/abs/2211.12723v3
|
https://arxiv.org/pdf/2211.12723v3.pdf
|
A Classification Model Utilizing Facial Landmark Tracking to Determine Sentence Types for American Sign Language Recognition
|
The deaf and hard of hearing community relies on American Sign Language (ASL) as their primary mode of communication, but communication with others who do not know ASL can be difficult, especially during emergencies where no interpreter is available. As an effort to alleviate this problem, research in computer vision based real time ASL interpreting models is ongoing. However, most of these models are hand shape (gesture) based and lack the integration of facial cues, which are crucial in ASL to convey tone and distinguish sentence types. Thus, the integration of facial cues in computer vision based ASL interpreting models has the potential to improve performance and reliability. In this paper, we introduce a simple, computationally efficient facial expression based classification model that can be used to improve ASL interpreting models. This model utilizes the relative angles of facial landmarks with principal component analysis and a Random Forest Classification tree model to classify frames taken from videos of ASL users signing a complete sentence. The model classifies the frames as statements or assertions. The model was able to achieve an accuracy of 86.5%.
|
['Y. Curtis Wang', 'Janice Nguyen']
|
2022-11-23
| null | null | null | null |
['sign-language-recognition', 'landmark-tracking']
|
['computer-vision', 'computer-vision']
|
[ 1.89234689e-01 -7.76488632e-02 -2.64524817e-01 -6.42754734e-01
-9.35949564e-01 -5.96987069e-01 2.98494935e-01 -1.39141589e-01
-5.78923225e-01 4.18257028e-01 4.58527714e-01 -4.50397223e-01
1.09834418e-01 -2.63905674e-01 9.25566256e-03 -6.25520527e-01
2.64381349e-01 1.50645867e-01 -2.72649843e-02 -1.35666236e-01
3.76669586e-01 8.78441811e-01 -1.58518839e+00 5.56129158e-01
3.28405201e-01 8.33501756e-01 -1.19934998e-01 9.21090603e-01
-4.08845931e-01 8.04459095e-01 -7.42450297e-01 -7.89902583e-02
1.46929678e-02 -4.21413988e-01 -5.63819647e-01 1.88556328e-01
7.21598446e-01 -8.72750938e-01 -7.80695630e-03 8.20310712e-01
8.69123340e-01 -1.67924285e-01 4.86942738e-01 -1.15099704e+00
-2.22309440e-01 2.93686818e-02 -3.36612344e-01 -1.29920527e-01
1.10173965e+00 8.33398178e-02 6.50492728e-01 -8.40919077e-01
5.61937809e-01 1.50803089e+00 5.79833627e-01 8.59216452e-01
-1.13674724e+00 -9.34247673e-01 1.83293462e-01 2.55961895e-01
-1.32354295e+00 -9.17668641e-01 9.03151333e-01 -4.91761237e-01
8.16804349e-01 4.09298927e-01 1.03283811e+00 9.28276062e-01
-2.85846621e-01 7.45654702e-01 1.65970457e+00 -9.54600155e-01
1.65748909e-01 -9.49345157e-02 1.93600848e-01 7.64171243e-01
-1.06709480e-01 -1.30137667e-01 -9.60081995e-01 -1.41408101e-01
4.60871667e-01 -3.85929525e-01 -4.89196301e-01 3.01384181e-01
-8.35695386e-01 7.42678583e-01 1.29493177e-01 3.75772983e-01
-2.87812144e-01 7.15351105e-02 8.48933831e-02 1.64986506e-01
3.38938117e-01 -8.30747113e-02 1.75849150e-03 -5.73211789e-01
-9.94962990e-01 1.52133675e-02 7.07166791e-01 2.84894019e-01
1.98365882e-01 -8.96841735e-02 1.54349312e-01 9.75680172e-01
8.85889649e-01 6.19587839e-01 1.97486162e-01 -1.09793067e+00
4.69875410e-02 6.30936205e-01 -1.68717667e-01 -9.05424356e-01
-4.91458625e-01 3.46821606e-01 -1.09182261e-01 8.70403886e-01
6.15291774e-01 -1.31320193e-01 -1.12088203e+00 1.41147041e+00
2.06578508e-01 -2.77024209e-01 -2.23027781e-01 1.05135167e+00
9.08892334e-01 1.59505874e-01 1.86454013e-01 -2.31519520e-01
1.33158338e+00 -3.14569563e-01 -9.25335228e-01 -9.52794924e-02
6.05656028e-01 -1.03726220e+00 8.36602867e-01 5.36646307e-01
-8.22883070e-01 -2.77854148e-02 -7.93875635e-01 -7.28281438e-02
-4.33426686e-02 2.56397992e-01 6.45935774e-01 1.15915120e+00
-1.29934418e+00 -6.62865117e-02 -8.75986993e-01 -6.89632475e-01
4.11750019e-01 4.98959005e-01 -6.47667289e-01 -1.67860270e-01
-7.08429635e-01 8.95187259e-01 -7.75849894e-02 4.74253803e-01
8.13084394e-02 -2.18436401e-03 -1.02808607e+00 -6.33405209e-01
-2.37765953e-01 -9.77636129e-02 1.42994761e+00 -1.17331636e+00
-1.66840017e+00 1.17548299e+00 -6.96868062e-01 8.47071186e-02
6.74601734e-01 -2.21084096e-02 -3.16908896e-01 6.26347125e-01
2.58973818e-02 7.61198163e-01 9.00142312e-01 -1.25440013e+00
-5.86884737e-01 -6.45785511e-01 -1.37685165e-02 1.79215416e-01
8.01171586e-02 5.55068552e-01 -2.70391881e-01 -4.08424228e-01
4.69708413e-01 -9.79744375e-01 2.87721932e-01 8.42055559e-01
-4.07946333e-02 -1.80885866e-01 9.96116161e-01 -1.44652748e+00
1.04717636e+00 -2.23087382e+00 -3.19293797e-01 4.43848640e-01
1.44187063e-01 4.66019571e-01 9.96926129e-02 2.31293216e-01
2.05894604e-01 1.48642674e-01 -5.96708469e-02 -1.91282198e-01
-2.99642533e-01 5.54748118e-01 2.61174768e-01 4.60438550e-01
-5.51654696e-02 4.51789588e-01 -7.69720018e-01 -7.13654101e-01
4.51511115e-01 8.55712175e-01 -3.03204089e-01 -1.10876970e-01
2.31074482e-01 6.90002739e-01 -2.93827534e-01 1.12099051e+00
5.47638297e-01 3.56611162e-01 3.13584238e-01 -8.83870795e-02
-2.00249061e-01 8.73161480e-02 -9.68688607e-01 1.34671962e+00
-3.78363639e-01 1.12779093e+00 5.77097237e-01 -7.14432478e-01
9.27299500e-01 6.02299869e-01 2.70540535e-01 -5.21211982e-01
2.62059003e-01 5.24110734e-01 3.68511938e-02 -1.01388454e+00
-8.47540796e-02 -4.04743105e-01 4.71429676e-01 3.06821674e-01
-4.88895148e-01 -2.08014280e-01 -6.24155663e-02 -1.43259972e-01
7.15582430e-01 1.46480680e-01 3.51615906e-01 3.73359233e-01
5.89101493e-01 -3.33274573e-01 2.05656186e-01 3.28483731e-01
-5.20669580e-01 6.02071166e-01 3.09120536e-01 -2.24668354e-01
-3.86833996e-01 -8.15130353e-01 -2.11855303e-02 7.77423739e-01
-3.52233410e-01 -3.84157002e-02 -7.53524899e-01 -3.15941274e-01
-1.19329415e-01 7.27106750e-01 -1.71695188e-01 4.08061385e-01
-3.10526073e-01 -1.75522104e-01 5.91716826e-01 4.46582794e-01
4.72412348e-01 -1.02201915e+00 -7.00211763e-01 1.87844917e-01
-3.25137943e-01 -1.12018192e+00 -1.38217807e-01 -2.75113165e-01
-5.11571825e-01 -1.23739791e+00 -7.95727134e-01 -9.73354459e-01
9.43932593e-01 2.79424600e-02 2.54457712e-01 2.74622053e-01
-4.08225864e-01 9.07877564e-01 -5.73556244e-01 -5.84389687e-01
-4.45085257e-01 -6.21313334e-01 -1.19270198e-03 2.39974216e-01
8.08768392e-01 -5.38548708e-01 -3.67677838e-01 1.55724317e-01
-6.94823682e-01 2.10535713e-02 2.97310710e-01 6.86049402e-01
9.75189060e-02 -5.64777374e-01 2.20360965e-01 -1.51545078e-01
8.31244528e-01 1.41869009e-01 -2.66844064e-01 1.69821739e-01
-3.52092236e-01 -2.28939474e-01 -1.33860648e-01 -4.07700300e-01
-1.03243732e+00 1.35820717e-01 -4.22287077e-01 8.30641761e-02
-4.23789322e-01 5.43763161e-01 6.50354624e-02 -5.45220315e-01
4.50761348e-01 4.59418818e-02 5.74103296e-01 -4.88785565e-01
-2.93020010e-01 1.54329836e+00 4.86388117e-01 -4.24751073e-01
3.69908512e-01 6.86377048e-01 -8.88998657e-02 -1.48182011e+00
-5.86532295e-01 -5.55546522e-01 -7.47844517e-01 -9.63422418e-01
9.25051808e-01 -6.39942646e-01 -9.25851047e-01 8.82394016e-01
-1.33737957e+00 -1.81139961e-01 3.53984416e-01 8.74233723e-01
-3.25058609e-01 5.01384079e-01 -2.84428895e-01 -1.46647453e+00
-7.59402290e-02 -1.14577889e+00 1.22240055e+00 2.62360334e-01
-7.24187255e-01 -6.13158286e-01 -1.81107476e-01 9.35871661e-01
2.64252603e-01 3.94049615e-01 7.64526486e-01 -3.10174942e-01
-1.97910652e-01 -6.53265536e-01 -2.08190560e-01 5.90854585e-01
3.95237178e-01 2.27576867e-01 -1.23543751e+00 4.36963066e-02
-2.10732698e-01 -1.91437766e-01 4.53923404e-01 5.55213213e-01
7.48239696e-01 -2.05629379e-01 1.49413636e-02 2.09337130e-01
9.13270295e-01 7.78802156e-01 5.25881171e-01 2.31467128e-01
1.77371651e-01 9.23189640e-01 3.15496147e-01 2.57492900e-01
4.70805526e-01 7.46778309e-01 8.93599093e-02 7.53683522e-02
-3.62750381e-01 -5.27995266e-02 4.91553903e-01 6.65267825e-01
-6.67414844e-01 2.69362897e-01 -1.14473975e+00 3.17080885e-01
-1.45964396e+00 -1.08041298e+00 -1.63764760e-01 1.97876060e+00
6.50640011e-01 -8.32714811e-02 6.31773546e-02 7.84599781e-01
6.35833144e-01 -9.38888714e-02 -1.79303020e-01 -7.69813120e-01
-1.05659015e-01 3.65915596e-01 1.70421153e-01 8.64403784e-01
-8.23022068e-01 6.80740535e-01 5.93396807e+00 2.78155774e-01
-1.49310470e+00 -1.87071666e-01 4.09827560e-01 6.25550523e-02
-5.50895855e-02 -1.52831107e-01 -5.78445792e-01 1.86719373e-01
4.54015225e-01 3.23232025e-01 4.08520192e-01 6.13231719e-01
9.42916453e-01 -6.38976336e-01 -7.84372151e-01 1.24695659e+00
4.39363569e-01 -8.41847479e-01 -2.37760216e-01 -8.83736238e-02
5.67322150e-02 -3.78153324e-01 -2.54947066e-01 -1.80269241e-01
-3.20123076e-01 -1.12053680e+00 8.10684204e-01 4.93350625e-01
1.12788105e+00 -3.34190130e-01 7.50386834e-01 1.51971206e-01
-9.71819997e-01 -2.10969020e-02 5.61954975e-01 -5.94196916e-01
3.25428218e-01 -1.89385086e-01 -1.11893439e+00 -2.84222513e-01
7.00942934e-01 2.55114615e-01 -2.78531134e-01 9.54974174e-01
-6.15729988e-01 6.48672640e-01 -6.05010450e-01 -3.67779762e-01
8.44770223e-02 3.79214026e-02 6.04403436e-01 1.17615330e+00
3.87468368e-01 3.64943624e-01 1.51350021e-01 3.24892625e-02
4.85211641e-01 3.28812093e-01 -5.03724158e-01 -1.07987404e-01
2.85699278e-01 6.94981992e-01 -6.39624715e-01 1.24655575e-01
-7.00466692e-01 9.56836641e-01 -3.95838380e-01 4.64806944e-01
-1.74954623e-01 -1.77186340e-01 7.78751433e-01 4.13489729e-01
-2.78890014e-01 -5.76671958e-01 -2.31680304e-01 -8.58343840e-01
3.29386890e-01 -1.01620662e+00 9.27192867e-02 -9.81374800e-01
-7.96578169e-01 3.35528225e-01 -6.79011792e-02 -1.16840434e+00
-4.18882132e-01 -1.03030336e+00 -2.69068003e-01 8.01380157e-01
-1.57217717e+00 -1.58466911e+00 -5.62262237e-01 6.03319705e-01
4.98316914e-01 -1.85657777e-02 1.05157900e+00 1.39536858e-01
-1.18237302e-01 5.39746284e-01 -3.15809548e-01 6.01612866e-01
4.93786126e-01 -7.61700988e-01 -3.89273345e-01 5.62272191e-01
7.27011934e-02 6.00378931e-01 6.18233204e-01 -5.01061380e-01
-9.56214070e-01 -3.62084210e-01 1.29200971e+00 -1.03826120e-01
4.09267247e-01 2.65041217e-02 -4.37645882e-01 5.76198041e-01
-2.26405129e-01 -2.57722765e-01 9.93433714e-01 -3.03496391e-01
-3.37834388e-01 8.96509886e-02 -1.37135315e+00 6.99986398e-01
8.93854439e-01 -8.44451904e-01 -5.50499856e-01 1.26855567e-01
-1.19759940e-01 -2.62224644e-01 -4.43877310e-01 3.90248783e-02
1.22904241e+00 -9.49526429e-01 6.59046233e-01 -3.64108264e-01
-1.17529765e-01 -3.22842777e-01 -3.02023739e-01 -7.94231236e-01
4.19709563e-01 -5.72073996e-01 4.49580938e-01 1.14762485e+00
4.19980705e-01 -9.13463771e-01 7.98080087e-01 1.15406740e+00
3.38517338e-01 -4.06732857e-01 -1.42342365e+00 -5.05970359e-01
-5.91664493e-01 -9.79381204e-01 2.63048261e-02 5.46502352e-01
1.44590810e-01 -9.88273770e-02 -2.41002843e-01 3.45198847e-02
5.37964046e-01 -3.18823874e-01 6.98165894e-01 -1.43273604e+00
1.93359390e-01 -5.02817631e-01 -1.05327439e+00 -6.96186841e-01
2.80642420e-01 -6.65917158e-01 -1.25982314e-01 -1.83240390e+00
-3.13235611e-01 -2.47111127e-01 3.75233114e-01 8.75754476e-01
3.46226156e-01 2.00969175e-01 3.62515479e-01 1.01029597e-01
3.48484725e-01 5.97589724e-02 1.00769055e+00 -1.51071250e-01
-1.99868321e-01 2.77989000e-01 -3.72736216e-01 1.12423062e+00
7.65062809e-01 -2.12678149e-01 -1.25289679e-01 -2.60208547e-01
-8.07944536e-02 2.04374507e-01 5.24898350e-01 -6.68833971e-01
1.16129376e-01 -2.23738417e-01 3.26055795e-01 -3.69589925e-01
8.01641166e-01 -1.01733327e+00 -8.15967992e-02 5.14837384e-01
-2.06512794e-01 -2.40631297e-01 7.35338703e-02 1.26895294e-01
-4.44939733e-01 -3.21054876e-01 7.21273541e-01 -2.84223985e-02
-7.96307564e-01 -2.77747929e-01 -7.64655590e-01 -4.59764391e-01
8.71142387e-01 -6.26184762e-01 3.53312679e-02 -1.19778967e+00
-8.78423512e-01 -1.06622413e-01 4.04918104e-01 2.45069921e-01
9.23536658e-01 -1.09066975e+00 -6.00127339e-01 3.54966104e-01
6.84438599e-03 -2.20551193e-01 -1.11588478e-01 1.05832326e+00
-1.06902111e+00 3.71893704e-01 -1.53172821e-01 -5.27913868e-01
-2.13905311e+00 -2.65985399e-01 2.04617694e-01 4.80835736e-01
-4.43739325e-01 9.09090221e-01 -5.55186689e-01 -2.82933474e-01
4.17647630e-01 -2.51537472e-01 -3.23675513e-01 3.28157395e-01
7.41516948e-01 4.38696206e-01 -1.20924912e-01 -1.09282923e+00
-6.09767973e-01 9.61832106e-01 3.60247135e-01 -6.12551630e-01
1.19950092e+00 -1.02964781e-01 -1.99775591e-01 4.92162734e-01
1.14227378e+00 2.91041076e-01 -7.71914542e-01 2.38227081e-02
5.34084812e-02 -6.98969185e-01 9.66578498e-02 -7.89097965e-01
-6.93578184e-01 1.04832077e+00 7.30318129e-01 -2.06341326e-01
1.38291073e+00 -6.18424416e-02 4.67104465e-01 2.48149693e-01
3.97800177e-01 -1.22531021e+00 -1.86191574e-01 6.30356818e-02
1.10955131e+00 -1.28601110e+00 -2.27298036e-01 -3.97809535e-01
-4.91302997e-01 1.64667106e+00 1.12343751e-01 4.79611218e-01
7.12527454e-01 2.90735543e-01 8.36405098e-01 -6.45579398e-02
-1.98471509e-02 -3.55223209e-01 2.23587319e-01 9.01162207e-01
7.33689487e-01 2.26416737e-01 -7.85561860e-01 1.02099277e-01
-2.98166662e-01 3.20319772e-01 3.80710006e-01 1.32288325e+00
-3.79991144e-01 -1.10497522e+00 -8.90316486e-01 3.40328991e-01
-3.99343312e-01 1.16404846e-01 -5.96602440e-01 7.23967493e-01
-5.06722890e-02 1.35704994e+00 -9.82759148e-02 -1.77761048e-01
2.22221270e-01 5.89536846e-01 5.08771956e-01 -2.95915425e-01
-2.07753927e-01 1.68964192e-01 5.93346119e-01 -3.88262242e-01
-9.33071136e-01 -9.82017636e-01 -1.34976470e+00 -5.52781373e-02
-8.13219100e-02 -4.47140992e-01 1.18052948e+00 1.15665030e+00
-2.00684547e-01 -2.03105301e-01 2.56703854e-01 -9.29544270e-01
-8.74640271e-02 -8.81022513e-01 -5.14064729e-01 2.26969495e-01
6.06754363e-01 -5.75695395e-01 -5.56948185e-01 2.18118221e-01]
|
[9.069245338439941, -6.359238147735596]
|
0600dae0-2bca-4117-8898-7bc7bbc5af19
|
vitmatte-boosting-image-matting-with
|
2305.15272
| null |
https://arxiv.org/abs/2305.15272v2
|
https://arxiv.org/pdf/2305.15272v2.pdf
|
ViTMatte: Boosting Image Matting with Pretrained Plain Vision Transformers
|
Recently, plain vision Transformers (ViTs) have shown impressive performance on various computer vision tasks, thanks to their strong modeling capacity and large-scale pretraining. However, they have not yet conquered the problem of image matting. We hypothesize that image matting could also be boosted by ViTs and present a new efficient and robust ViT-based matting system, named ViTMatte. Our method utilizes (i) a hybrid attention mechanism combined with a convolution neck to help ViTs achieve an excellent performance-computation trade-off in matting tasks. (ii) Additionally, we introduce the detail capture module, which just consists of simple lightweight convolutions to complement the detailed information required by matting. To the best of our knowledge, ViTMatte is the first work to unleash the potential of ViT on image matting with concise adaptation. It inherits many superior properties from ViT to matting, including various pretraining strategies, concise architecture design, and flexible inference strategies. We evaluate ViTMatte on Composition-1k and Distinctions-646, the most commonly used benchmark for image matting, our method achieves state-of-the-art performance and outperforms prior matting works by a large margin.
|
['Baoyuan Wang', 'Shusheng Yang', 'Xinggang Wang', 'Jingfeng Yao']
|
2023-05-24
| null | null | null | null |
['image-matting']
|
['computer-vision']
|
[ 2.13069350e-01 2.77877022e-02 -6.49429560e-02 -4.26219195e-01
-6.24385655e-01 -1.04351699e-01 7.03532517e-01 -4.96832252e-01
-3.90933782e-01 3.33365083e-01 1.05459765e-01 -5.24924755e-01
4.62472558e-01 -6.69395268e-01 -1.31069803e+00 -5.81781507e-01
5.01869977e-01 3.37118834e-01 2.34282330e-01 -3.31762433e-01
7.91430846e-03 1.30326912e-01 -1.38326943e+00 6.67181611e-01
1.12971210e+00 1.13002646e+00 5.82473814e-01 4.90144819e-01
-3.76796156e-01 1.06465185e+00 -3.91092002e-01 -7.23812759e-01
3.30798984e-01 -2.83206463e-01 -7.34255791e-01 3.11789960e-01
9.80850041e-01 -6.65635169e-01 -7.45913088e-01 7.18982697e-01
2.62170017e-01 -2.13764995e-01 4.90262210e-01 -1.14421391e+00
-9.06579614e-01 9.82384980e-01 -8.56959999e-01 4.84361313e-02
-1.02350883e-01 6.56890690e-01 7.85943747e-01 -1.03517246e+00
1.86915696e-01 1.24654114e+00 6.71845794e-01 5.52594364e-01
-1.19265151e+00 -6.07607186e-01 3.87391329e-01 3.20332676e-01
-1.09089613e+00 -4.76456046e-01 5.08962095e-01 -2.38367364e-01
1.04051101e+00 9.67529267e-02 6.86576962e-01 1.11545563e+00
3.29764038e-01 1.24182808e+00 1.23501468e+00 -2.66646922e-01
5.73016070e-02 -3.87783541e-04 -5.07944189e-02 9.54969823e-01
7.12582543e-02 -1.21680111e-01 -5.83683431e-01 4.57806885e-01
8.49220097e-01 2.24945426e-01 -1.63121253e-01 -9.25615877e-02
-1.36373234e+00 5.88738799e-01 9.47707832e-01 -2.66694985e-02
-3.95114362e-01 8.04738939e-01 2.96055496e-01 4.35772181e-01
3.39441419e-01 3.30968887e-01 -2.81126678e-01 5.28145321e-02
-1.21273553e+00 -1.39209032e-01 3.66244286e-01 1.07130408e+00
8.81304860e-01 5.39666533e-01 -3.92663777e-01 7.52544343e-01
2.36657798e-01 7.61036038e-01 5.15695393e-01 -8.40385139e-01
6.89205170e-01 6.31589472e-01 -3.15914512e-01 -4.90613759e-01
6.11586682e-02 -4.43387061e-01 -1.22505546e+00 3.81359935e-01
2.44153634e-01 7.10962713e-02 -1.59622967e+00 1.56428194e+00
8.68330942e-04 4.73054379e-01 -1.08080223e-01 7.65307069e-01
1.12551892e+00 7.10825264e-01 -6.06380850e-02 2.33532608e-01
1.42586589e+00 -1.59690118e+00 -6.00911796e-01 -5.39177120e-01
1.93393454e-01 -8.07105780e-01 1.22191799e+00 3.62672418e-01
-1.43999302e+00 -8.66722643e-01 -1.33323216e+00 -2.78065383e-01
-1.67749867e-01 -1.86210033e-02 1.09247601e+00 6.78881884e-01
-1.29705596e+00 6.32728457e-01 -1.10561764e+00 -1.95553094e-01
7.51155019e-01 2.83785462e-01 -3.37111443e-01 -3.79797667e-01
-6.78932607e-01 9.49003935e-01 7.56976828e-02 2.48529837e-01
-1.34595048e+00 -8.50610018e-01 -1.08488131e+00 1.98963154e-02
5.23558736e-01 -1.21992719e+00 1.28059947e+00 -9.96081889e-01
-1.60209525e+00 6.37216151e-01 -2.81425565e-01 -7.28569269e-01
4.21345890e-01 -4.91863161e-01 -8.51448998e-02 9.82955098e-02
-9.43259522e-02 1.05648363e+00 1.25335395e+00 -1.19106519e+00
-1.40546188e-01 -7.09223598e-02 1.70679137e-01 1.19746417e-01
-3.92793894e-01 -3.14161539e-01 -8.35668325e-01 -9.15061891e-01
-1.01111066e-02 -6.87631428e-01 -2.47724131e-01 5.31449951e-02
-3.95008445e-01 2.39491835e-01 7.65226841e-01 -5.63562155e-01
9.54060972e-01 -2.03921461e+00 2.73566008e-01 -1.87661991e-01
6.33133411e-01 6.67605996e-01 -4.61299688e-01 4.23053861e-01
3.09754964e-02 -1.02199264e-01 -3.92641753e-01 -7.84164429e-01
-5.67129394e-03 4.68287200e-01 -5.59642911e-01 3.01062226e-01
4.76006180e-01 1.64002669e+00 -5.79887450e-01 -4.32912260e-01
4.71050322e-01 3.17445219e-01 -5.68612456e-01 3.36233348e-01
-4.62650180e-01 1.94088556e-02 -1.09500252e-01 7.32329488e-01
8.43115211e-01 -3.73983502e-01 -1.40510663e-01 -5.53318083e-01
2.17890926e-02 1.84658572e-01 -8.09860110e-01 2.06189632e+00
-4.99427259e-01 5.97777128e-01 1.31802306e-01 -9.90424454e-01
6.32192194e-01 1.76572755e-01 1.83055848e-01 -7.88780034e-01
1.19925693e-01 5.51537909e-02 -1.82206735e-01 -2.80534416e-01
6.62274182e-01 1.13178372e-01 2.50065148e-01 4.81753767e-01
3.52289796e-01 -3.33128661e-01 4.70559783e-02 7.12219477e-01
1.10670507e+00 2.20206112e-01 -1.34578362e-01 -1.62265405e-01
5.98882474e-02 -1.51168466e-01 2.25163296e-01 9.44716871e-01
2.02246100e-01 8.00609112e-01 9.39446837e-02 -4.69794273e-01
-1.13879538e+00 -1.13324118e+00 1.10765785e-01 8.83468509e-01
2.52535611e-01 -5.57260811e-01 -9.26423490e-01 -6.49617970e-01
-1.87147073e-02 4.54734296e-01 -8.90078127e-01 -1.63614526e-01
-5.62715828e-01 -6.92435026e-01 6.65167332e-01 8.44931901e-01
1.01142848e+00 -1.05178499e+00 -3.24603200e-01 1.14603676e-01
-1.30614176e-01 -1.13960254e+00 -8.39648247e-01 3.05196047e-01
-1.01901972e+00 -7.85462618e-01 -9.21927810e-01 -8.47991943e-01
6.24234259e-01 8.23212028e-01 1.47226548e+00 4.42910284e-01
-2.31580630e-01 2.16251537e-01 -3.11898410e-01 -4.37211663e-01
-3.26418668e-01 8.99261683e-02 -1.86756998e-01 -5.19848503e-02
-1.96720630e-01 -7.88485050e-01 -7.61527896e-01 4.23662066e-01
-1.17258918e+00 8.51333797e-01 1.24625933e+00 1.07614839e+00
4.92039233e-01 -3.39668959e-01 3.32488865e-01 -9.86400962e-01
5.38350381e-02 -3.32938075e-01 -3.12281460e-01 3.24153751e-01
-6.63433731e-01 2.74214655e-01 4.26987916e-01 -4.61614251e-01
-1.06411898e+00 2.99389791e-02 -3.80907029e-01 -6.19915903e-01
7.38841221e-02 4.47501063e-01 -2.29551151e-01 -2.69826144e-01
4.56776887e-01 4.90740240e-01 2.37080351e-01 -5.87743044e-01
7.13191748e-01 2.90576130e-01 9.31209922e-01 -7.51711309e-01
1.13375509e+00 5.31094730e-01 -1.41858622e-01 -4.10956323e-01
-1.00618565e+00 -1.80742115e-01 -4.11357284e-01 9.11133364e-02
7.82571077e-01 -1.29511678e+00 -5.37930131e-01 9.14266050e-01
-1.02441788e+00 -8.47901762e-01 -1.99879676e-01 8.62841606e-02
-4.82851595e-01 4.80992913e-01 -1.04137659e+00 -2.88322717e-01
-7.28916466e-01 -1.44195390e+00 1.05920243e+00 1.05585620e-01
3.18656921e-01 -6.58922434e-01 -3.45618516e-01 6.07556522e-01
8.95772696e-01 5.30762374e-02 6.07873440e-01 1.81747779e-01
-9.38030601e-01 3.22624594e-01 -6.16124213e-01 4.36150432e-01
7.36824870e-02 -1.98987126e-01 -1.19346178e+00 -3.90373141e-01
-2.05316097e-01 -4.88577664e-01 1.61008883e+00 4.04924482e-01
1.27720428e+00 -4.02902186e-01 -1.41892418e-01 1.19534147e+00
1.31604755e+00 -1.51983917e-01 1.15168500e+00 4.14449990e-01
1.14204609e+00 -1.42938271e-01 2.23222360e-01 1.28150165e-01
5.89093089e-01 6.71671450e-01 8.11147988e-01 -7.17753112e-01
-6.52774274e-01 -2.31793657e-01 6.29204750e-01 9.89493906e-01
4.45074476e-02 -1.52897030e-01 -4.91005093e-01 3.58138800e-01
-2.00892973e+00 -8.72615397e-01 -2.99099460e-02 1.74729621e+00
9.98936892e-01 3.14194024e-01 -3.99140753e-02 -1.06600255e-01
2.57986248e-01 2.08401933e-01 -7.05823421e-01 -3.14188689e-01
-2.34613210e-01 6.11904800e-01 3.80880654e-01 2.87219822e-01
-1.02453542e+00 1.16663623e+00 6.39697838e+00 1.06505275e+00
-1.07167232e+00 3.24859656e-02 8.49879682e-01 9.88522395e-02
-3.91256928e-01 -8.78949612e-02 -6.85993314e-01 4.49218065e-01
6.49955034e-01 2.49421075e-01 5.90096414e-01 7.45081604e-01
-2.04952046e-01 6.54140860e-02 -1.09772968e+00 9.09269929e-01
1.42459095e-01 -1.74504328e+00 4.30552393e-01 -4.89757843e-02
8.94080400e-01 2.73515850e-01 3.85970175e-01 5.25071144e-01
2.70079851e-01 -1.40660357e+00 1.03581011e+00 3.13482344e-01
1.00941658e+00 -4.00910228e-01 7.47346401e-01 4.74936962e-02
-1.33109581e+00 1.52115390e-01 -4.30885106e-01 -1.13825612e-01
1.33183882e-01 6.72432840e-01 -6.23927116e-01 6.70665503e-01
5.34582078e-01 8.85454774e-01 -8.50652695e-01 1.13660467e+00
-2.65241891e-01 9.32556868e-01 -1.58826366e-01 4.24133599e-01
5.57676494e-01 1.15083851e-01 8.36066455e-02 1.24634767e+00
1.83466017e-01 -1.00820340e-01 1.57749370e-01 9.63278651e-01
-3.85915339e-01 -2.95293868e-01 -2.57665604e-01 -4.10021050e-03
2.27523774e-01 1.30477118e+00 -6.00249410e-01 -6.62841141e-01
-5.25378287e-01 1.19108570e+00 3.70414346e-01 2.18181446e-01
-1.02378488e+00 -1.30857512e-01 5.98704100e-01 1.81804374e-01
5.77968478e-01 -3.09888810e-01 -5.16439974e-01 -1.37760234e+00
4.38826904e-03 -1.06658185e+00 2.34457716e-01 -9.93314087e-01
-1.13903749e+00 9.00977969e-01 -5.43797053e-02 -9.28268433e-01
6.31334186e-02 -6.88853741e-01 -8.51954579e-01 7.45002329e-01
-1.71147013e+00 -1.68797052e+00 -7.15139568e-01 7.22235680e-01
8.44828367e-01 -2.07007140e-01 5.60407639e-01 3.56037408e-01
-6.68021381e-01 9.43952382e-01 -1.42597273e-01 1.14678837e-01
7.54428148e-01 -1.36365652e+00 1.04762852e+00 1.12212968e+00
4.34067696e-01 6.62142098e-01 4.16245520e-01 -4.44774240e-01
-1.89401400e+00 -1.31080258e+00 1.81813702e-01 -4.32472616e-01
4.47673768e-01 -4.59296346e-01 -8.82315397e-01 8.58656347e-01
8.51977944e-01 -7.36162961e-02 4.55867909e-02 -1.59481764e-02
-8.75272810e-01 -3.13714921e-01 -7.67475069e-01 7.46066093e-01
9.99351799e-01 -2.72041649e-01 -5.40656388e-01 1.76004499e-01
1.12885749e+00 -7.19113290e-01 -5.70177615e-01 3.29806328e-01
4.92276907e-01 -9.55680370e-01 1.17793524e+00 -1.72832921e-01
8.43553782e-01 -3.99835050e-01 -1.35346368e-01 -1.28388679e+00
-4.97872084e-01 -7.20128179e-01 -4.22589064e-01 1.26103079e+00
3.31943899e-01 -4.18824583e-01 8.35982740e-01 3.54701340e-01
-7.61269808e-01 -1.03607321e+00 -5.44126570e-01 -6.37828231e-01
-1.76151637e-02 -4.65532660e-01 7.33199060e-01 6.77489996e-01
-4.15970087e-01 2.99080610e-01 -8.34910691e-01 -6.71246722e-02
7.01650739e-01 2.12719083e-01 1.06678391e+00 -5.43734908e-01
-7.91923225e-01 -5.29687941e-01 -2.02222764e-01 -1.58938456e+00
-1.09085143e-01 -7.85741448e-01 1.70026720e-01 -1.67224944e+00
5.87554574e-01 -4.19105113e-01 -1.78197980e-01 8.99422348e-01
-5.19278049e-01 6.84550583e-01 4.60205883e-01 2.17247143e-01
-7.75741637e-01 7.33321667e-01 1.53002226e+00 -5.71502864e-01
1.97458938e-01 -3.21199089e-01 -9.69354749e-01 4.12076324e-01
6.56530261e-01 -1.55475289e-01 -4.63733286e-01 -1.07647347e+00
-4.90395688e-02 -1.34296700e-01 4.25857544e-01 -1.08670318e+00
2.11842120e-01 -2.34186053e-01 4.04839516e-01 -6.03993595e-01
5.39471865e-01 -4.32942361e-01 2.84594804e-01 4.40573901e-01
-3.07327788e-02 2.12674260e-01 5.26347637e-01 4.13416624e-01
-1.45242274e-01 3.09657715e-02 8.20277750e-01 -2.61735409e-01
-1.00381625e+00 6.09277844e-01 -2.82564253e-01 -4.86585274e-02
7.06576586e-01 -2.16555223e-01 -5.92188954e-01 -3.51979524e-01
-2.16515809e-01 1.73965812e-01 4.95915651e-01 2.72179037e-01
8.20296347e-01 -1.23561394e+00 -8.88692200e-01 2.28744045e-01
1.61328986e-02 2.04421446e-01 2.74811804e-01 8.60468745e-01
-6.06811106e-01 6.79079369e-02 -3.20211947e-01 -6.45612121e-01
-1.05805540e+00 6.85515165e-01 1.24939103e-02 -2.50427395e-01
-1.09847987e+00 1.17788768e+00 6.76551342e-01 6.43435568e-02
2.68607944e-01 -3.64258170e-01 4.69195783e-01 -4.26292270e-01
7.85072684e-01 -1.31525651e-01 2.94918597e-01 -2.18056813e-01
-1.50987476e-01 3.23321521e-01 -5.41738093e-01 1.32274792e-01
1.39490783e+00 -1.49830848e-01 -7.48685673e-02 5.35700144e-03
7.34932542e-01 -3.71956319e-01 -1.50923669e+00 -4.78355676e-01
-5.54966927e-01 -5.35292149e-01 7.38672689e-02 -8.70976925e-01
-1.55692661e+00 1.03640723e+00 3.48776013e-01 -2.96064466e-01
1.18185604e+00 -2.49242745e-02 1.29819632e+00 2.75648326e-01
3.98450613e-01 -5.51210344e-01 6.53152764e-01 4.31688845e-01
9.54155803e-01 -1.36496341e+00 -7.75082735e-03 -3.02332789e-01
-7.90783167e-01 7.57735193e-01 9.03650641e-01 -1.88857153e-01
2.54257947e-01 6.60456717e-01 2.31080741e-01 -1.03410028e-01
-1.01221764e+00 -2.32331708e-01 4.45849538e-01 5.62916458e-01
3.34059626e-01 -4.61730957e-02 3.36046427e-01 3.89481306e-01
-1.32783934e-01 -1.26069903e-01 3.43242288e-01 6.70893312e-01
-4.10784781e-01 -1.02555978e+00 -3.54813188e-01 5.65944374e-01
-3.04603368e-01 -5.78382254e-01 -7.78438076e-02 5.29021025e-01
7.44509026e-02 8.17568362e-01 -1.23232819e-01 -6.41602635e-01
7.22572058e-02 -2.98533052e-01 7.09561288e-01 -4.76378262e-01
-7.49364614e-01 -1.11267962e-01 -1.95113301e-01 -7.34706223e-01
-2.70392537e-01 -2.03037545e-01 -7.61902153e-01 -8.10881376e-01
-3.23310822e-01 -1.25543356e-01 6.04170024e-01 1.01885664e+00
5.24300873e-01 8.94993663e-01 3.11290234e-01 -1.16276741e+00
-4.51302409e-01 -1.02259457e+00 -2.62321264e-01 4.02383924e-01
2.81070113e-01 -5.05839169e-01 1.07044742e-01 1.43458664e-01]
|
[10.635175704956055, -0.8332929015159607]
|
a5a7a498-c8ec-4e0c-b18a-abb0f534c911
|
liar-liar-pants-on-fire-a-new-benchmark
|
1705.00648
| null |
http://arxiv.org/abs/1705.00648v1
|
http://arxiv.org/pdf/1705.00648v1.pdf
|
"Liar, Liar Pants on Fire": A New Benchmark Dataset for Fake News Detection
|
Automatic fake news detection is a challenging problem in deception
detection, and it has tremendous real-world political and social impacts.
However, statistical approaches to combating fake news has been dramatically
limited by the lack of labeled benchmark datasets. In this paper, we present
liar: a new, publicly available dataset for fake news detection. We collected a
decade-long, 12.8K manually labeled short statements in various contexts from
PolitiFact.com, which provides detailed analysis report and links to source
documents for each case. This dataset can be used for fact-checking research as
well. Notably, this new dataset is an order of magnitude larger than previously
largest public fake news datasets of similar type. Empirically, we investigate
automatic fake news detection based on surface-level linguistic patterns. We
have designed a novel, hybrid convolutional neural network to integrate
meta-data with text. We show that this hybrid approach can improve a text-only
deep learning model.
|
['William Yang Wang']
|
2017-05-01
| null | null | null |
acl-2017-7
|
['deception-detection']
|
['miscellaneous']
|
[-1.56271011e-02 -9.55954101e-03 -7.92847276e-01 -3.11325908e-01
-1.02470446e+00 -8.15588415e-01 1.12642348e+00 3.75666827e-01
-1.67679951e-01 8.95206034e-01 7.73552775e-01 -5.62157631e-01
5.27305841e-01 -7.74241865e-01 -9.87186491e-01 -2.08909482e-01
2.91956007e-01 3.58134300e-01 1.11029841e-01 -7.10928738e-01
8.42698872e-01 2.71846205e-01 -9.93722677e-01 1.03913522e+00
8.56950164e-01 7.87361801e-01 -8.15320790e-01 2.67819017e-01
6.35860907e-03 1.31178093e+00 -1.07579315e+00 -9.24801648e-01
8.45423639e-02 -4.31975782e-01 -9.37674105e-01 -2.27134556e-01
9.50242460e-01 -6.05891049e-01 -6.41989052e-01 1.27224779e+00
3.05099666e-01 -4.78050888e-01 5.12763202e-01 -1.09664381e+00
-1.07592785e+00 9.31892037e-01 -4.88058954e-01 5.03169298e-01
3.65601450e-01 1.86604604e-01 1.00448406e+00 -5.82787931e-01
1.01109242e+00 1.30905759e+00 8.86502028e-01 5.27441680e-01
-6.49670541e-01 -8.24291408e-01 -3.48584026e-01 9.64246094e-02
-6.56209230e-01 -5.13897061e-01 9.36987698e-01 -6.58289433e-01
8.25452030e-01 1.96920291e-01 5.59305131e-01 2.02646112e+00
3.64606440e-01 9.69527364e-01 1.48287892e+00 -3.10721815e-01
-1.99294522e-01 3.03411454e-01 3.00238043e-01 8.80508184e-01
6.67579472e-01 1.71055987e-01 -6.70836806e-01 -6.69460833e-01
2.20918104e-01 -1.74764097e-01 -2.68703312e-01 3.15283686e-01
-1.20099485e+00 1.41571939e+00 5.74331582e-01 6.15609825e-01
-4.28321958e-03 5.06246805e-01 8.88859928e-01 6.11777902e-01
1.24776399e+00 8.88140738e-01 -2.56489784e-01 -1.84107587e-01
-9.41143394e-01 5.76759696e-01 1.01291800e+00 4.44973618e-01
7.83938766e-02 -9.73350741e-03 -2.30263382e-01 5.56922555e-01
-1.29647866e-01 6.15086138e-01 5.55774927e-01 -6.52026892e-01
8.75505149e-01 5.22195518e-01 5.14682353e-01 -1.95157421e+00
-4.04601932e-01 -4.45228428e-01 -7.14013755e-01 -3.66442770e-01
5.97747386e-01 2.24643499e-02 -5.86165667e-01 1.30430174e+00
1.24870859e-01 -3.72768939e-02 -1.58634722e-01 1.06758928e+00
1.12384188e+00 6.11669660e-01 -4.07518119e-01 2.46805474e-01
1.37104392e+00 -1.03045380e+00 -8.11914504e-01 -3.91829550e-01
9.81191337e-01 -8.42887282e-01 9.14742887e-01 4.24894303e-01
-6.18874013e-01 2.77395606e-01 -1.08808458e+00 -4.40170407e-01
-8.28547895e-01 2.23926917e-01 8.67887259e-01 8.48879993e-01
-4.74190027e-01 6.37927115e-01 -4.82360840e-01 -1.21429637e-01
9.89416003e-01 -3.99016708e-01 -3.44011247e-01 8.35329387e-03
-1.58339417e+00 1.24423099e+00 2.64942348e-01 -1.18008114e-01
-9.67834115e-01 -3.19232911e-01 -7.22206056e-01 -2.13438123e-01
4.67466414e-01 -1.47080764e-01 1.10937655e+00 -1.18584311e+00
-1.26261008e+00 1.34798121e+00 5.72939366e-02 -7.84530282e-01
9.42880213e-01 -3.10552984e-01 -6.93313837e-01 2.45812070e-02
3.12232018e-01 -1.24199830e-01 1.14065719e+00 -1.16326916e+00
-1.97535291e-01 -3.16068828e-01 4.59844172e-02 -4.36997086e-01
-4.68955100e-01 4.62460577e-01 3.91440064e-01 -9.42693412e-01
-1.40984729e-01 -7.68207967e-01 2.10928574e-01 -2.84666389e-01
-6.93586886e-01 -1.35669038e-01 8.96243930e-01 -1.02475631e+00
1.19501185e+00 -1.85792768e+00 -2.87741572e-01 -2.00992972e-01
6.62180543e-01 5.38028181e-01 -9.52836946e-02 5.30390143e-01
1.81009948e-01 4.92539108e-01 -1.78398058e-01 -2.62934148e-01
-6.65201098e-02 -3.18742543e-01 -9.52153802e-01 1.07484961e+00
1.92351982e-01 1.16435957e+00 -1.25467658e+00 -1.94651261e-01
-3.04637045e-01 1.62766222e-02 -5.54986775e-01 -3.94773602e-01
-3.51640522e-01 2.97187388e-01 -4.84352857e-01 9.42146778e-01
5.77584743e-01 -3.15527141e-01 7.20321462e-02 -4.28105965e-02
-4.17592600e-02 9.59767938e-01 -7.07262475e-03 1.09366620e+00
-2.64767826e-01 1.32788825e+00 -2.97400001e-02 -1.13620830e+00
7.72859991e-01 3.73830013e-02 1.47066638e-02 -6.66063190e-01
5.52795947e-01 6.86145902e-01 -2.63036728e-01 -4.77280915e-01
8.85830879e-01 -1.67144135e-01 -5.58100641e-01 6.59849286e-01
-2.47663543e-01 -3.00886542e-01 -1.87881351e-01 4.22669530e-01
1.22978795e+00 -3.21877211e-01 3.75988811e-01 -1.28950685e-01
2.62512803e-01 6.64390266e-01 2.95806319e-01 1.01081717e+00
-6.33097053e-01 2.58843631e-01 8.95976424e-01 -9.38561440e-01
-1.05989516e+00 -3.08602631e-01 -7.78051913e-02 7.93044865e-01
2.61787884e-02 -3.35462928e-01 -5.40563405e-01 -1.16308331e+00
3.83226573e-01 7.06870198e-01 -8.73253465e-01 -1.89805552e-01
-8.48505080e-01 -9.24831927e-01 1.21030140e+00 -2.15814233e-01
6.82938874e-01 -9.96330738e-01 -2.37230763e-01 1.66133791e-01
-5.80298483e-01 -1.19943893e+00 -3.18605542e-01 -3.52514058e-01
-3.83996576e-01 -1.26332629e+00 -2.54513383e-01 -5.06858468e-01
2.51563489e-01 4.47930276e-01 1.22196364e+00 4.44600761e-01
-5.16369529e-02 -3.81097019e-01 -5.28417230e-01 -3.95739108e-01
-1.02863121e+00 1.54817894e-01 6.74810112e-02 -1.18739687e-01
4.83653277e-01 2.70175636e-02 -1.65328711e-01 2.90620513e-02
-8.83462727e-01 -1.57738835e-01 2.52635926e-01 1.18676043e+00
-2.42569208e-01 -1.93529248e-01 8.09771836e-01 -1.26547515e+00
1.01600420e+00 -9.08412039e-01 -6.76061928e-01 -1.56146869e-01
-1.97892457e-01 -3.24828744e-01 7.23238885e-01 -3.85434985e-01
-7.28024423e-01 -7.49885380e-01 7.30344504e-02 1.15404902e-02
4.24760282e-02 7.44378030e-01 4.93147373e-01 -2.05460966e-01
1.08625627e+00 2.62499675e-02 -1.13878377e-01 -3.12634617e-01
3.85979503e-01 1.19141102e+00 3.47842366e-01 -2.98432469e-01
8.97912502e-01 8.41695249e-01 -4.96851563e-01 -6.73724353e-01
-1.66881669e+00 -3.10471356e-01 -1.91039145e-01 -8.58964492e-03
4.68675405e-01 -8.72370601e-01 -5.26597261e-01 8.74110162e-01
-1.59701586e+00 -2.49074519e-01 3.23070973e-01 1.16092496e-01
-1.43384770e-01 5.15273750e-01 -1.08717537e+00 -5.79687119e-01
-2.87863553e-01 -7.11758852e-01 9.18631852e-01 -5.03610611e-01
-5.75550981e-02 -9.20443833e-01 9.85818282e-02 1.01997304e+00
5.02111673e-01 7.60157347e-01 4.93790209e-01 -1.07036388e+00
-2.41346136e-01 -5.78937948e-01 -5.76312900e-01 3.09279978e-01
1.17776562e-02 -1.94358543e-01 -7.81711459e-01 -3.31421584e-01
1.55932158e-01 -9.03997123e-01 1.34838915e+00 -9.65219662e-02
1.01067352e+00 -1.16782284e+00 -3.38767081e-01 1.74260333e-01
1.11908519e+00 -4.35153097e-01 4.57736045e-01 7.37702787e-01
7.74748504e-01 5.33272743e-01 4.60348308e-01 3.55055094e-01
2.65200108e-01 5.34386575e-01 3.85557026e-01 1.82467505e-01
-5.32024391e-02 -3.86831790e-01 6.78336203e-01 7.58079231e-01
2.24969655e-01 -6.41015470e-01 -1.14919674e+00 8.06556046e-01
-1.66176331e+00 -1.57026708e+00 -4.96270120e-01 1.56389129e+00
8.72925580e-01 3.36345047e-01 1.37751959e-02 -9.28221457e-03
7.52648652e-01 5.88848829e-01 -1.77420795e-01 -6.25096560e-01
-5.24251401e-01 -1.46459103e-01 8.57696712e-01 5.25108516e-01
-1.56035507e+00 1.42669940e+00 6.43387270e+00 1.13634026e+00
-1.28888726e+00 7.32092321e-01 5.95398784e-01 -9.99660492e-02
-3.56053621e-01 -2.29197398e-01 -4.63645816e-01 7.83852994e-01
8.40633392e-01 1.08784959e-01 3.42151701e-01 8.48050714e-01
2.69655406e-01 5.65755032e-02 -6.18531287e-01 7.15513885e-01
5.39143801e-01 -2.08755612e+00 8.64928737e-02 1.44595727e-01
1.15248489e+00 4.85990375e-01 1.15157053e-01 3.99395078e-01
5.10408759e-01 -1.19410455e+00 9.90576744e-01 8.42741206e-02
6.84654295e-01 -6.03878438e-01 1.08718991e+00 5.97814262e-01
1.58750266e-02 -1.15309775e-01 -2.17721865e-01 -3.41794789e-01
9.88656953e-02 9.70302343e-01 -7.30266273e-01 2.83756167e-01
5.04878879e-01 9.45800304e-01 -5.81917286e-01 3.87147188e-01
-3.07015836e-01 1.02310014e+00 -9.78892446e-02 -4.10575747e-01
7.45669603e-01 1.71264559e-01 8.48315716e-01 1.41114056e+00
1.56914722e-02 -4.96522002e-02 1.39351100e-01 1.07019627e+00
-7.06393301e-01 -1.13090552e-01 -9.80462790e-01 -7.26309538e-01
3.18221301e-01 1.01636565e+00 -5.02933681e-01 -7.18269289e-01
-3.38031292e-01 1.07079220e+00 7.11386442e-01 -1.43705249e-01
-1.01992178e+00 -1.48824155e-01 4.42790419e-01 1.48943186e-01
-5.20158261e-02 -2.54055202e-01 -3.66749793e-01 -1.76208222e+00
-8.36457685e-03 -1.33747458e+00 2.14519233e-01 -3.70924443e-01
-1.81660020e+00 3.90995711e-01 -3.28249186e-01 -1.03170431e+00
1.36599824e-01 -7.53928423e-01 -4.42692995e-01 3.79911929e-01
-1.47776580e+00 -1.32897854e+00 -1.45888686e-01 1.70961261e-01
2.12409779e-01 -3.66910517e-01 5.01923740e-01 2.65104920e-01
-4.56191212e-01 5.97524941e-01 3.47101927e-01 7.19568372e-01
9.20769632e-01 -8.50788236e-01 7.14572370e-01 7.05319941e-01
3.17629911e-02 4.93629098e-01 8.02596807e-01 -1.04033160e+00
-1.15553939e+00 -1.12057352e+00 1.34586334e+00 -1.10009599e+00
1.38320625e+00 -5.22072673e-01 -8.46756041e-01 8.91419530e-01
5.73208220e-02 -6.50881976e-02 4.46808517e-01 -3.76943797e-02
-1.05096269e+00 5.65928698e-01 -1.55324244e+00 3.71799678e-01
9.48047638e-01 -7.43295610e-01 -7.71404207e-01 9.61352050e-01
7.83163786e-01 -6.15432262e-01 -3.70399237e-01 8.46552402e-02
5.10481417e-01 -1.04870713e+00 5.62110007e-01 -1.13042581e+00
1.28816092e+00 7.66682550e-02 3.84869613e-02 -1.41121805e+00
-1.15598522e-01 -4.21105236e-01 -2.07408383e-01 8.88649881e-01
4.18377876e-01 -1.00779498e+00 7.42744744e-01 -1.88874692e-01
-1.37600958e-01 -5.68431675e-01 -1.09325612e+00 -8.27934146e-01
4.76260126e-01 -3.21287751e-01 3.91603261e-01 1.95570827e+00
1.77201331e-01 1.72132969e-01 -8.72377634e-01 7.36371754e-03
4.16005552e-01 4.09819484e-01 7.47213423e-01 -9.14796472e-01
-1.77401572e-01 -6.93760931e-01 -4.41087663e-01 -9.12852347e-01
4.19330418e-01 -1.09607899e+00 -2.96030372e-01 -1.18822527e+00
3.91357601e-01 -2.71980852e-01 2.90595204e-01 4.37995017e-01
-7.94766247e-02 6.66296482e-01 -1.74556822e-01 3.96636277e-01
-3.20039898e-01 4.83305424e-01 1.26162577e+00 -6.06963396e-01
3.40580136e-01 -3.65992755e-01 -6.14084065e-01 8.09657753e-01
1.06474841e+00 -8.15067112e-01 5.56949615e-01 -5.58269083e-01
6.21119022e-01 -2.12143898e-01 8.04035008e-01 -3.99071008e-01
-2.44496956e-01 -2.95661420e-01 3.56640890e-02 -5.33676982e-01
1.53694242e-01 -1.71140060e-01 -5.41393459e-01 6.63819313e-01
-4.39273804e-01 -1.81275129e-01 4.69307788e-02 7.50615299e-01
-3.49341601e-01 -8.45669582e-02 8.69455636e-01 -3.99016649e-01
-1.45619139e-01 -8.22806656e-02 -5.76970816e-01 4.05877441e-01
5.49935758e-01 2.62709498e-01 -1.22468495e+00 -6.04411364e-01
-3.77557985e-02 -3.22053552e-01 5.07243514e-01 5.80414176e-01
3.56871486e-01 -1.36000454e+00 -1.22381306e+00 -3.56546521e-01
2.78122455e-01 -7.94386506e-01 -2.19849676e-01 8.89521897e-01
-8.12984109e-01 5.62247157e-01 -1.67252477e-02 5.84782623e-02
-8.85812700e-01 6.31268978e-01 2.92621255e-01 -3.86985391e-01
-5.19007027e-01 6.52621508e-01 -5.63002408e-01 -6.98460281e-01
-3.73042971e-01 -2.44453356e-01 -3.18093486e-02 7.09017739e-02
5.77862561e-01 3.69490534e-01 6.97689578e-02 -1.05049682e+00
-3.29342216e-01 -1.79088965e-01 -1.34126678e-01 6.02812394e-02
1.39578629e+00 1.43444434e-01 -5.79981208e-01 3.70044857e-01
1.28871226e+00 6.68674111e-01 -2.07478195e-01 -2.92890400e-01
2.22022042e-01 -8.29168379e-01 1.31590351e-01 -1.04305422e+00
-8.93547535e-01 5.25071681e-01 -2.51464933e-01 6.13081217e-01
2.51782894e-01 9.66098681e-02 1.21999860e+00 4.85261202e-01
5.10656059e-01 -1.02968287e+00 4.48860675e-01 8.71563137e-01
1.26720309e+00 -1.74048936e+00 1.32096648e-01 -4.54983383e-01
-3.76248538e-01 9.43032980e-01 3.91303867e-01 -4.53190476e-01
2.94067472e-01 -1.74611434e-01 7.47325048e-02 -7.44900286e-01
-4.50269133e-01 4.63105440e-01 2.49849543e-01 9.01418552e-02
5.14631450e-01 3.48441005e-01 -7.80310452e-01 5.69197476e-01
-6.35693491e-01 -2.71344960e-01 1.05614543e+00 7.82539666e-01
-5.36289334e-01 -6.52591705e-01 -4.58895773e-01 6.69942319e-01
-9.49300468e-01 -3.78173828e-01 -1.11721623e+00 7.47879207e-01
1.94435585e-02 1.04616749e+00 -2.51334339e-01 -2.94768810e-01
-2.31458947e-01 -2.72452891e-01 3.08572531e-01 -4.28519070e-01
-9.15148139e-01 -6.45061374e-01 8.82749915e-01 -6.84469402e-01
-3.22153568e-01 -4.02657479e-01 -5.88222504e-01 -1.01043725e+00
-5.09815097e-01 -1.23241683e-02 7.63013482e-01 1.03573263e+00
3.83170664e-01 -9.70134810e-02 5.12831092e-01 -5.60837269e-01
-7.62162387e-01 -1.20844686e+00 -1.53937638e-01 6.29256904e-01
7.88713157e-01 -6.38080060e-01 -8.51395607e-01 -3.81110013e-01]
|
[8.143072128295898, 10.207809448242188]
|
f365cd8f-f9dc-4fd2-95c5-b18394c42a9b
|
exploring-the-relationship-between-alignment
|
2306.0279
| null |
https://arxiv.org/abs/2306.02790v1
|
https://arxiv.org/pdf/2306.02790v1.pdf
|
Exploring the Relationship between Alignment and Cross-lingual Transfer in Multilingual Transformers
|
Without any explicit cross-lingual training data, multilingual language models can achieve cross-lingual transfer. One common way to improve this transfer is to perform realignment steps before fine-tuning, i.e., to train the model to build similar representations for pairs of words from translated sentences. But such realignment methods were found to not always improve results across languages and tasks, which raises the question of whether aligned representations are truly beneficial for cross-lingual transfer. We provide evidence that alignment is actually significantly correlated with cross-lingual transfer across languages, models and random seeds. We show that fine-tuning can have a significant impact on alignment, depending mainly on the downstream task and the model. Finally, we show that realignment can, in some instances, improve cross-lingual transfer, and we identify conditions in which realignment methods provide significant improvements. Namely, we find that realignment works better on tasks for which alignment is correlated with cross-lingual transfer when generalizing to a distant language and with smaller models, as well as when using a bilingual dictionary rather than FastAlign to extract realignment pairs. For example, for POS-tagging, between English and Arabic, realignment can bring a +15.8 accuracy improvement on distilmBERT, even outperforming XLM-R Large by 1.7. We thus advocate for further research on realignment methods for smaller multilingual models as an alternative to scaling.
|
['Yannick Toussaint', 'Parisa Rastin', 'Patricio Cerda', 'Félix Gaschi']
|
2023-06-05
| null | null | null | null |
['cross-lingual-transfer', 'xlm-r']
|
['natural-language-processing', 'natural-language-processing']
|
[ 6.78265393e-02 -6.09404333e-02 -3.10569286e-01 -4.88737255e-01
-1.38938391e+00 -9.80101705e-01 6.99709654e-01 2.08666891e-01
-9.34395790e-01 9.06003833e-01 3.20237964e-01 -7.20757544e-01
3.04411501e-01 -6.27939284e-01 -1.14896035e+00 -2.75700748e-01
2.65762180e-01 5.95913947e-01 4.93294410e-02 -4.80031401e-01
9.30597857e-02 2.20477954e-01 -8.64054084e-01 3.66829634e-01
1.11837220e+00 -3.48459668e-02 3.75364900e-01 3.58157426e-01
-1.67210057e-01 1.20536327e-01 -4.96845156e-01 -6.97709024e-01
2.42429778e-01 -6.60035491e-01 -9.42877233e-01 -3.02085787e-01
6.00495219e-01 1.36287913e-01 2.37314090e-01 9.25746083e-01
4.57431316e-01 -1.36734188e-01 7.78667748e-01 -7.69269705e-01
-8.07371318e-01 1.12898219e+00 -7.12715268e-01 2.14867190e-01
1.48498058e-01 1.57484949e-01 1.20058644e+00 -8.41701865e-01
6.45976901e-01 1.36860585e+00 8.63384604e-01 3.73907745e-01
-1.53897130e+00 -8.87265265e-01 1.62543088e-01 2.55415030e-03
-1.19614303e+00 -6.36781037e-01 2.27198794e-01 -3.41129184e-01
1.19483650e+00 6.67717382e-02 2.61889100e-01 8.69605482e-01
1.94052517e-01 4.94195342e-01 1.48217022e+00 -8.55217755e-01
-3.79677385e-01 3.25768381e-01 -6.39791191e-02 3.84664357e-01
3.96845490e-01 1.34237379e-01 -4.04473722e-01 2.57955819e-01
5.84830999e-01 -7.25766778e-01 -3.23442817e-01 -1.43763512e-01
-1.49094486e+00 8.73086393e-01 4.49177414e-01 7.33453393e-01
-1.02085389e-01 4.00796905e-02 5.98226428e-01 7.65490830e-01
5.37896216e-01 1.08370674e+00 -9.07291830e-01 -2.11081371e-01
-8.55958462e-01 -6.19806722e-02 5.09460151e-01 6.67943597e-01
8.89423192e-01 -4.31450456e-02 1.77496642e-01 1.16751182e+00
1.67447049e-02 6.19413078e-01 9.04503405e-01 -5.70575178e-01
8.66423368e-01 1.94215909e-01 -2.77023852e-01 -4.80136305e-01
-2.28195950e-01 -4.60181028e-01 -3.39670569e-01 -5.95456250e-02
7.22900867e-01 -3.27529192e-01 -6.83487713e-01 2.18970394e+00
-1.29725307e-01 -2.87168741e-01 3.55981171e-01 6.11424625e-01
2.37271920e-01 7.00364292e-01 2.54419833e-01 -1.06409274e-01
1.41748548e+00 -8.77556205e-01 -2.64083385e-01 -7.37316489e-01
1.43191433e+00 -1.31258810e+00 1.53792608e+00 1.54445216e-01
-1.03139818e+00 -4.95164365e-01 -1.04576159e+00 -1.46694586e-01
-3.74743670e-01 2.60951463e-02 5.10815918e-01 5.91609597e-01
-9.68218267e-01 7.29342759e-01 -7.88604736e-01 -6.79025710e-01
-1.37662798e-01 2.45971531e-01 -7.56727576e-01 -3.86532426e-01
-1.22825396e+00 1.44643414e+00 4.24636334e-01 -3.00743073e-01
-3.80125403e-01 -6.49385750e-01 -9.23729181e-01 -1.46147132e-01
-1.64786607e-01 -4.75180000e-01 1.09041059e+00 -1.46344650e+00
-1.24358928e+00 1.15564108e+00 -1.79046348e-01 -4.48985070e-01
3.06120813e-01 -2.48533413e-01 -3.35168391e-01 -5.66399097e-01
3.90417099e-01 8.41215134e-01 2.76668131e-01 -1.06454706e+00
-5.89361548e-01 -2.55166560e-01 -1.30398288e-01 5.79586923e-01
-3.83695424e-01 4.48331565e-01 -3.22755843e-01 -6.20301723e-01
-7.94159248e-02 -1.17623520e+00 -2.20049500e-01 -7.58481145e-01
6.53382242e-02 -2.77464986e-01 4.59458828e-02 -1.00319588e+00
8.85019779e-01 -2.00352955e+00 1.84168965e-02 9.58406478e-02
-5.07601619e-01 3.59338999e-01 -6.54463887e-01 4.52070922e-01
-3.03551733e-01 3.20707828e-01 -1.66314796e-01 -2.34578967e-01
-1.88859150e-01 3.69187444e-01 -1.57969669e-01 3.95787925e-01
4.82691526e-01 1.08160102e+00 -7.49419987e-01 -2.69640177e-01
-2.33056754e-01 4.10445571e-01 -6.62841499e-01 -1.58191368e-01
3.49198505e-02 5.99964738e-01 1.68730527e-01 -3.99671867e-02
4.65287209e-01 1.47208825e-01 6.04779541e-01 5.00171594e-02
-1.87319830e-01 1.08485258e+00 -8.69497418e-01 1.66922402e+00
-1.12259161e+00 7.56226242e-01 -2.12671235e-01 -1.05079401e+00
7.67367721e-01 2.77698845e-01 -2.40188502e-02 -9.57050025e-01
-1.05252534e-01 7.96160579e-01 6.30765736e-01 -9.08122882e-02
5.45095265e-01 -6.79270625e-01 -1.42698571e-01 6.39730513e-01
1.49381429e-01 -2.66832292e-01 2.33168602e-01 -7.74807110e-02
7.18283236e-01 3.33503962e-01 3.47468764e-01 -5.17551363e-01
3.04547608e-01 1.55205980e-01 5.38146377e-01 4.29557353e-01
1.91086397e-01 3.60978782e-01 2.35601261e-01 -2.55248807e-02
-1.01447606e+00 -9.21336412e-01 -2.52765328e-01 1.44768012e+00
-3.12344849e-01 -3.19281369e-01 -7.56547868e-01 -9.27354217e-01
-5.56523018e-02 9.59446788e-01 -2.98329115e-01 -1.35748029e-01
-1.03922701e+00 -8.86353970e-01 6.27154291e-01 5.77337682e-01
7.43189752e-02 -9.64390337e-01 1.11541115e-01 2.24007457e-01
-3.61770093e-01 -1.03805447e+00 -5.90215981e-01 5.12224138e-01
-1.03452027e+00 -5.41475534e-01 -7.02906489e-01 -1.10821271e+00
7.35452831e-01 3.33478272e-01 1.40134156e+00 1.17855020e-01
4.13192660e-01 -1.09906629e-01 -3.11665475e-01 -1.99343309e-01
-7.86133587e-01 7.44613647e-01 1.29566580e-01 -4.94736791e-01
4.46099430e-01 -5.49302399e-01 -5.09328209e-02 4.21317101e-01
-5.82024097e-01 -5.74911125e-02 8.17163646e-01 8.79787743e-01
3.39511305e-01 -4.58427876e-01 7.97357023e-01 -1.08145940e+00
7.34317422e-01 -3.33236337e-01 -4.59355086e-01 2.62265235e-01
-6.08341038e-01 4.51645643e-01 6.89943075e-01 -4.74653691e-01
-8.40732932e-01 -1.72023728e-01 -3.25464308e-01 2.15904444e-01
-1.17815398e-01 6.73725963e-01 -1.27008364e-01 2.18291488e-02
1.05772221e+00 6.35775477e-02 -3.07132136e-02 -5.24850070e-01
5.93303323e-01 6.94377482e-01 2.15984225e-01 -8.80489647e-01
8.12470496e-01 -1.04462691e-01 -4.85509604e-01 -5.44887841e-01
-6.92153513e-01 -2.05222934e-01 -7.59602249e-01 3.83622259e-01
8.04252625e-01 -1.24036944e+00 3.66912484e-02 1.09787226e-01
-1.15365279e+00 -7.97680914e-01 -4.14691456e-02 8.82433534e-01
-3.83247524e-01 1.04698569e-01 -6.17752314e-01 -1.07833957e-02
-8.26099887e-02 -1.27313364e+00 7.83480227e-01 -1.57759011e-01
-5.41850626e-01 -1.32257628e+00 2.71986544e-01 3.63864630e-01
4.10446584e-01 -3.53296459e-01 1.16713321e+00 -1.01319802e+00
-2.39128292e-01 -4.79434356e-02 -1.11862510e-01 6.18133903e-01
4.15133446e-01 -2.44166642e-01 -6.57051563e-01 -5.89967132e-01
-2.99767613e-01 -4.48543817e-01 6.79886103e-01 1.29796684e-01
3.43166351e-01 -2.67267078e-01 -2.86622912e-01 4.88764971e-01
1.26056147e+00 1.26485815e-02 4.00444984e-01 5.09695172e-01
6.72160566e-01 7.44425774e-01 6.25235379e-01 -4.85858589e-01
5.68121552e-01 9.34875786e-01 -2.81320155e-01 -2.23473758e-01
-3.84326518e-01 -2.99015462e-01 7.97852516e-01 1.46797740e+00
1.15387216e-01 1.24583870e-01 -1.14367759e+00 8.76337409e-01
-1.43733048e+00 -5.47742844e-01 -1.59047052e-01 2.47498155e+00
1.39554477e+00 1.68827906e-01 2.09637564e-02 -1.72080114e-01
6.28709793e-01 -2.39900604e-01 -5.19950017e-02 -7.17173517e-01
-1.98886901e-01 5.84730685e-01 6.59917772e-01 1.04602015e+00
-6.87905252e-01 1.67023599e+00 5.86878538e+00 8.60345304e-01
-1.51942229e+00 4.51771915e-01 4.65586364e-01 2.10994750e-01
-5.83611727e-01 2.49378935e-01 -1.04412091e+00 2.96943873e-01
1.13004434e+00 -2.33785242e-01 4.76216495e-01 5.23721278e-01
-5.96613847e-02 5.04555032e-02 -1.28317773e+00 4.59007829e-01
6.86526522e-02 -8.78994346e-01 2.53071003e-02 8.75502601e-02
9.65797663e-01 4.54769313e-01 -1.79135069e-01 5.57329893e-01
7.38959491e-01 -1.13786149e+00 4.90903348e-01 -3.16124290e-01
8.58463228e-01 -8.19175243e-01 7.31759071e-01 4.26771998e-01
-8.38509142e-01 5.52803874e-01 -3.74226809e-01 -6.81307092e-02
1.93550423e-01 2.86226124e-01 -1.23199761e+00 5.19681871e-01
2.05534369e-01 5.48530221e-01 -7.90429533e-01 7.46116102e-01
-5.56930184e-01 9.25712764e-01 -2.75888741e-01 1.25672400e-01
2.99828738e-01 -1.39452606e-01 2.00165108e-01 1.59000587e+00
5.27715385e-01 -5.98598123e-01 -6.79042414e-02 2.46038720e-01
4.60801087e-02 7.58466899e-01 -8.03301871e-01 -1.15463778e-01
4.08597976e-01 1.04897249e+00 -5.53203881e-01 -3.14141721e-01
-6.62202537e-01 1.09965682e+00 7.60262549e-01 2.19101861e-01
-7.36190736e-01 -4.16056007e-01 6.29071295e-01 3.20972681e-01
2.94218153e-01 -5.38765788e-01 -4.06676233e-01 -1.25307906e+00
-1.19742662e-01 -1.32849848e+00 1.78114966e-01 -5.27458489e-01
-1.24198377e+00 7.38442183e-01 -2.23675907e-01 -1.01491988e+00
-4.98364687e-01 -7.02717662e-01 -3.42629224e-01 1.32301295e+00
-1.60110760e+00 -1.33737397e+00 3.92459959e-01 5.15565991e-01
4.59749848e-01 -7.76162818e-02 8.46418858e-01 4.79246438e-01
-2.19742879e-01 9.48053956e-01 2.24830378e-02 2.90975004e-01
1.38625264e+00 -1.11445129e+00 7.69101202e-01 1.09046185e+00
8.66463482e-01 9.35421586e-01 5.83597124e-01 -4.52244133e-01
-8.87018681e-01 -1.11443663e+00 1.44162214e+00 -5.13266265e-01
8.32273245e-01 -4.26646143e-01 -1.04086339e+00 1.22000682e+00
5.79036355e-01 -3.11997712e-01 7.27491856e-01 7.21440315e-01
-6.54587328e-01 -1.56453595e-01 -6.37340009e-01 8.46827209e-01
1.02932966e+00 -6.38787806e-01 -7.03670859e-01 3.95030856e-01
7.87491143e-01 -1.62560701e-01 -9.62510645e-01 3.32914174e-01
3.52880359e-01 -5.44394195e-01 7.35221803e-01 -7.35892475e-01
4.26140189e-01 -4.60754707e-02 -2.70710617e-01 -2.02564526e+00
-4.29979295e-01 -3.64150763e-01 8.43105674e-01 1.50253391e+00
1.04854035e+00 -1.00179851e+00 1.71511471e-01 8.68523344e-02
-2.69904405e-01 -4.23703045e-01 -8.00297022e-01 -1.20995343e+00
1.04589713e+00 -5.37442863e-01 4.94561404e-01 1.37908638e+00
8.54240730e-02 9.00245726e-01 -7.14757219e-02 5.79822250e-02
1.47567242e-01 -7.96709657e-02 9.52339232e-01 -9.36420798e-01
-5.18194258e-01 -5.47545016e-01 -2.30566874e-01 -1.10754597e+00
6.03513777e-01 -1.43901563e+00 2.30097666e-01 -1.32578325e+00
-7.30094546e-03 -9.55909371e-01 -2.90791601e-01 6.85195088e-01
-4.71017301e-01 5.23241878e-01 3.26096177e-01 2.82654762e-01
-5.21703362e-02 6.01948351e-02 1.14414847e+00 2.86565363e-01
-2.46921912e-01 -1.97771639e-01 -1.12378538e+00 5.63973427e-01
9.25781131e-01 -6.30081177e-01 -2.35852495e-01 -9.72274303e-01
3.52284551e-01 -2.49487624e-01 -1.89438373e-01 -7.95671105e-01
-2.32830480e-01 -2.35409990e-01 1.38956964e-01 2.11620778e-01
7.85304159e-02 -4.88500834e-01 -1.29232198e-01 4.44938660e-01
-2.48532534e-01 4.74997550e-01 5.37974358e-01 -1.21249646e-01
-2.89540470e-01 -3.29905152e-01 8.34654331e-01 -9.30233598e-02
-4.01891470e-01 -2.50881821e-01 -2.93366075e-01 4.44046527e-01
6.21861577e-01 4.05047904e-04 -3.62445563e-01 -1.72998399e-01
-3.55913877e-01 -7.37212971e-02 7.11588800e-01 5.84670603e-01
-2.44471774e-01 -1.39964807e+00 -1.07765532e+00 3.08125883e-01
1.40917033e-01 -3.60346705e-01 -4.22517896e-01 9.57483888e-01
-2.57005125e-01 6.01690352e-01 -1.79055646e-01 -5.24345577e-01
-1.21856487e+00 2.38906860e-01 2.18546152e-01 -4.73164231e-01
-1.35025680e-01 8.76780748e-01 4.00690049e-01 -8.99975896e-01
-2.61975914e-01 -3.43250871e-01 1.04073092e-01 1.24956116e-01
5.22659235e-02 -2.48924613e-01 4.51555818e-01 -8.79486561e-01
-4.98829663e-01 7.98772812e-01 -3.82414013e-01 -5.11574924e-01
1.23654258e+00 -3.44009027e-02 -1.17880434e-01 5.25130749e-01
1.29206753e+00 6.82486117e-01 -9.50274825e-01 -3.66837710e-01
8.81371051e-02 -1.50154144e-01 -3.02563280e-01 -9.22678947e-01
-9.34709251e-01 9.48807061e-01 1.27423450e-01 -1.38322532e-01
8.27275753e-01 1.40062377e-01 6.98387682e-01 3.07375938e-01
5.40344357e-01 -8.35883319e-01 -1.85394913e-01 8.31401587e-01
8.61755192e-01 -1.34105539e+00 -1.65067807e-01 -2.35355660e-01
-7.28125811e-01 6.76901639e-01 6.16535068e-01 -2.02576667e-01
3.86487812e-01 1.21974021e-01 3.57445985e-01 2.75396645e-01
-6.27497554e-01 -2.11513475e-01 2.93506384e-01 5.07417023e-01
1.22998106e+00 3.69455248e-01 -7.13740110e-01 3.72440189e-01
-7.98958778e-01 -3.68078381e-01 2.00062633e-01 6.21667564e-01
-3.01935464e-01 -1.70825136e+00 -4.08141494e-01 1.05416693e-01
-5.87624073e-01 -6.82150662e-01 -3.07431906e-01 9.63722289e-01
1.86310723e-01 7.86740124e-01 2.48314992e-01 -1.61219329e-01
3.03661197e-01 4.31087524e-01 6.72024131e-01 -9.50093567e-01
-8.10501456e-01 -1.97075810e-02 3.28349173e-01 -1.39317781e-01
-4.05400544e-02 -7.44970560e-01 -1.08695734e+00 -1.47443831e-01
-4.49020177e-01 2.34081954e-01 8.07683349e-01 1.06481004e+00
2.91028559e-01 2.56253064e-01 4.23582971e-01 -5.90932906e-01
-4.71279532e-01 -1.23025918e+00 -4.17724997e-02 3.98981273e-01
1.60566103e-02 -3.40718746e-01 -2.91807026e-01 1.16748028e-01]
|
[10.958650588989258, 10.013030052185059]
|
10b93781-c6cc-43f1-b93e-bd12c4f0b212
|
bayesian-mixtures-of-spatial-spline
|
1508.00635
| null |
http://arxiv.org/abs/1508.00635v1
|
http://arxiv.org/pdf/1508.00635v1.pdf
|
Bayesian mixtures of spatial spline regressions
|
This work relates the framework of model-based clustering for spatial
functional data where the data are surfaces. We first introduce a Bayesian
spatial spline regression model with mixed-effects (BSSR) for modeling spatial
function data. The BSSR model is based on Nodal basis functions for spatial
regression and accommodates both common mean behavior for the data through a
fixed-effects part, and variability inter-individuals thanks to a
random-effects part. Then, in order to model populations of spatial functional
data issued from heterogeneous groups, we integrate the BSSR model into a
mixture framework. The resulting model is a Bayesian mixture of spatial spline
regressions with mixed-effects (BMSSR) used for density estimation and
model-based surface clustering. The models, through their Bayesian formulation,
allow to integrate possible prior knowledge on the data structure and
constitute a good alternative to recent mixture of spatial spline regressions
model estimated in a maximum likelihood framework via the
expectation-maximization (EM) algorithm. The Bayesian model inference is
performed by Markov Chain Monte Carlo (MCMC) sampling. We derive two Gibbs
sampler to infer the BSSR and the BMSSR models and apply them on simulated
surfaces and a real problem of handwritten digit recognition using the MNIST
data set. The obtained results highlight the potential benefit of the proposed
Bayesian approaches for modeling surfaces possibly dispersed in particular in
clusters.
|
['Faicel Chamroukhi']
|
2015-08-04
| null | null | null | null |
['handwritten-digit-recognition']
|
['computer-vision']
|
[ 8.39124545e-02 -1.08186319e-01 1.36454508e-01 -2.81457573e-01
-6.87187493e-01 -3.15859877e-02 8.26453507e-01 -5.55193834e-02
-3.69178891e-01 1.09027481e+00 5.83801530e-02 -2.81688511e-01
-7.10578561e-01 -1.10094011e+00 -8.53773892e-01 -1.15343654e+00
-1.98206186e-01 8.91858935e-01 4.45472836e-01 1.09721690e-01
2.38387913e-01 6.31929457e-01 -1.63078904e+00 -5.16933650e-02
1.21560645e+00 2.96835899e-01 6.44549668e-01 3.93988550e-01
-1.75890446e-01 4.03080851e-01 -4.65671510e-01 1.55188426e-01
-2.30205730e-01 -4.80509028e-02 -2.29893401e-01 8.14640522e-02
-1.42951652e-01 1.36296228e-01 2.24653408e-01 7.54882336e-01
1.12951592e-01 3.38600338e-01 1.54957879e+00 -8.65943789e-01
-4.90641922e-01 3.88284534e-01 -8.03846538e-01 -2.28045970e-01
-5.51838661e-03 -1.04785338e-01 3.09082031e-01 -9.60501850e-01
3.97895098e-01 1.44868541e+00 6.66026771e-01 -1.31333172e-01
-1.78872621e+00 -2.06967682e-01 -5.51592782e-02 -1.45836994e-01
-2.02543354e+00 -1.44871235e-01 5.41435897e-01 -1.01425803e+00
5.28616309e-01 2.59888887e-01 6.76106334e-01 7.32452393e-01
3.16438407e-01 3.45328540e-01 1.31705761e+00 -2.95483261e-01
6.76375747e-01 8.52729008e-02 4.23085272e-01 6.85790628e-02
4.60815787e-01 9.93287936e-02 -2.70462036e-01 -5.83369493e-01
1.07388806e+00 -8.01958442e-02 7.11160107e-03 -1.76332355e-01
-7.72361636e-01 9.99029815e-01 1.33367062e-01 5.13953567e-01
-7.76446521e-01 4.15008992e-01 -3.88078004e-01 -3.83685619e-01
8.35618436e-01 -3.44273984e-01 -1.12708464e-01 4.36277598e-01
-1.47597384e+00 3.38714749e-01 7.19833672e-01 6.94857240e-01
9.88063097e-01 1.49518907e-01 -9.99822989e-02 1.00430894e+00
1.33506370e+00 1.03043127e+00 -1.30155548e-01 -7.45148301e-01
-1.52725298e-02 2.78446257e-01 1.74360886e-01 -9.12037313e-01
-5.92135526e-02 -2.01276377e-01 -9.15079236e-01 4.88680363e-01
6.32912040e-01 -1.32667646e-02 -9.10712481e-01 1.52403641e+00
5.75028300e-01 5.53271830e-01 -2.95907050e-01 3.36927861e-01
5.58988273e-01 1.00873566e+00 2.35111713e-01 -4.72418636e-01
1.26289058e+00 -5.92882000e-02 -8.69723856e-01 4.82334882e-01
2.90783912e-01 -5.14846921e-01 5.58726549e-01 6.45266294e-01
-1.01231670e+00 -5.83490372e-01 -6.22292876e-01 5.25786698e-01
-2.71421403e-01 1.61264375e-01 3.63154970e-02 8.29237401e-01
-1.39513409e+00 6.01447463e-01 -1.16952145e+00 -4.51679707e-01
2.50550330e-01 3.45498830e-01 3.44322100e-02 1.41813993e-01
-9.13228273e-01 8.17567468e-01 -6.15691356e-02 3.03593189e-01
-8.67889643e-01 -5.38689017e-01 -5.48177063e-01 -1.91162676e-01
-1.38265803e-01 -5.91168284e-01 3.94082874e-01 -4.71758187e-01
-1.30652285e+00 5.26887119e-01 -6.48961008e-01 -1.08132750e-01
4.13336754e-01 2.92572021e-01 -1.75364733e-01 -4.04257923e-02
8.72432962e-02 2.75895268e-01 8.27059507e-01 -1.62892377e+00
9.17013660e-02 -4.68148053e-01 -9.03907180e-01 -1.85413554e-01
1.90023288e-01 1.42874811e-02 -2.30809614e-01 -5.91856897e-01
1.92159176e-01 -6.96976364e-01 -4.10808027e-01 -3.64703834e-01
-2.24144459e-01 -1.30101666e-01 3.54809135e-01 -1.20232177e+00
1.34014237e+00 -2.16963768e+00 3.61736923e-01 8.23150039e-01
-1.47606343e-01 -2.99087971e-01 2.56581753e-01 5.43988764e-01
1.43771842e-01 3.94718312e-02 -8.10789287e-01 -3.13365430e-01
-1.48735136e-01 2.32495293e-01 6.09675646e-02 8.71606827e-01
3.33780944e-02 2.63152033e-01 -3.43820840e-01 -5.70523143e-01
5.12725711e-01 9.23226953e-01 -3.95199090e-01 6.53870180e-02
-2.76977330e-01 1.02201486e+00 -5.51899493e-01 2.90815145e-01
1.37972128e+00 -2.75960211e-02 1.06750593e-01 9.74420831e-02
-6.06948614e-01 -4.07816708e-01 -1.70404530e+00 1.24787331e+00
-1.24285936e-01 1.47283366e-02 5.26149452e-01 -9.28725302e-01
1.41485894e+00 3.71765018e-01 5.16267300e-01 3.99805345e-02
-1.27934948e-01 4.35375899e-01 -1.60019413e-01 -3.46021265e-01
3.50829124e-01 -3.23877215e-01 2.29669884e-01 6.21896200e-02
-4.06891555e-02 -5.37233315e-02 -1.87047988e-01 -5.45749404e-02
6.88330412e-01 5.35436630e-01 2.70263076e-01 -1.09401441e+00
6.92744374e-01 1.19276764e-03 8.41569528e-02 6.69995904e-01
4.86046463e-01 4.85309780e-01 4.24859196e-01 1.42597571e-01
-9.72292066e-01 -1.32565510e+00 -8.43927860e-01 4.80054498e-01
-1.41080037e-01 2.63591260e-01 -7.12496996e-01 3.49401414e-01
2.26573929e-01 8.22795928e-01 -6.63205147e-01 4.00280893e-01
-2.60488302e-01 -1.50819564e+00 3.34581316e-01 8.52563903e-02
9.93754044e-02 -7.90966749e-01 -1.53784171e-01 3.10533673e-01
1.34842962e-01 -2.25754067e-01 4.62998062e-01 1.96524546e-01
-1.15328574e+00 -8.33325565e-01 -9.85687494e-01 -1.92750782e-01
3.55633408e-01 -3.14113975e-01 9.29282784e-01 -1.59479007e-01
-2.06973180e-01 4.57825541e-01 -1.65634379e-01 9.39607993e-02
-6.58034503e-01 -3.91939163e-01 -9.35829133e-02 4.73735332e-01
3.58949482e-01 -8.29252243e-01 -3.39547157e-01 5.47219396e-01
-8.72167647e-01 -2.90318608e-01 4.48068887e-01 5.01243472e-01
6.51116312e-01 -4.69143912e-02 6.70508206e-01 -6.76817834e-01
2.82665849e-01 -1.13728201e+00 -9.09544349e-01 2.79424340e-01
-3.38241935e-01 -4.41420898e-02 -7.28450492e-02 -2.14536861e-01
-1.52596879e+00 1.05171129e-01 -4.01558429e-01 -2.18192697e-01
-7.52432287e-01 7.00480044e-01 -4.76136953e-01 5.23404852e-02
6.10301673e-01 1.97893336e-01 2.13840529e-01 -7.84397066e-01
1.42041743e-01 8.30658793e-01 2.13659197e-01 -1.00011206e+00
4.30433184e-01 6.42463744e-01 3.47515970e-01 -1.41756558e+00
2.18266651e-01 -4.54775274e-01 -8.36979926e-01 -4.78280663e-01
1.46275294e+00 -9.52148557e-01 -7.03882992e-01 7.29407012e-01
-1.14568686e+00 -4.91689026e-01 -1.33521527e-01 6.28552139e-01
-6.20755672e-01 3.89302105e-01 -2.87847281e-01 -1.64897263e+00
4.27585632e-01 -1.19137871e+00 1.07807899e+00 -4.93389107e-02
4.46319841e-02 -1.48224008e+00 4.27223980e-01 1.41362652e-01
3.59782994e-01 4.61748481e-01 7.73180246e-01 -3.16647559e-01
-7.37765193e-01 2.47043073e-01 1.00445189e-01 8.95354599e-02
8.81779194e-02 5.33008158e-01 -9.86287236e-01 -2.34438595e-03
6.70102239e-02 5.54369688e-01 9.46573377e-01 1.35933018e+00
6.45174205e-01 9.66439769e-02 -5.80684364e-01 4.57580388e-01
1.85287750e+00 3.14811945e-01 9.91726995e-01 -1.54966325e-01
5.79381049e-01 9.73867059e-01 5.08120060e-01 6.97806537e-01
2.29564071e-01 8.18635285e-01 3.64023685e-01 -1.51107118e-01
1.96776688e-01 1.41023234e-01 1.32975340e-01 4.90980804e-01
-4.57649559e-01 -1.97405189e-01 -1.25234127e+00 6.29865408e-01
-1.96437967e+00 -1.13629436e+00 -1.45153606e+00 2.60193753e+00
4.98911619e-01 -4.21318740e-01 3.43955904e-01 -1.45080104e-01
1.23715460e+00 -2.03984186e-01 -1.49717107e-01 1.10913359e-01
-2.57172287e-01 2.54840761e-01 6.25804901e-01 1.01346302e+00
-8.62643719e-01 5.35113871e-01 6.11038733e+00 1.18303418e+00
-4.31560725e-01 3.78726780e-01 5.14819443e-01 2.86395699e-01
-4.45936620e-01 1.98132500e-01 -8.86808157e-01 7.95064807e-01
1.16998112e+00 4.26458597e-01 3.14981073e-01 3.86468291e-01
8.07301760e-01 -7.55364597e-01 -5.25411487e-01 5.17335713e-01
-3.86820883e-01 -9.78749394e-01 -1.74056605e-01 7.76520669e-01
7.29496062e-01 -4.71905097e-02 -6.59140721e-02 -1.69877425e-01
7.19322085e-01 -1.07930422e+00 7.35241354e-01 1.28523970e+00
7.05116451e-01 -6.02304399e-01 4.87848878e-01 5.92179298e-01
-1.22691369e+00 2.85860509e-01 -4.68234420e-01 -3.08344979e-03
4.80516076e-01 9.55061138e-01 -4.71862853e-01 6.52652562e-01
5.88398218e-01 5.75663447e-01 -3.41223717e-01 1.10341859e+00
1.05330266e-01 8.13909948e-01 -5.74671149e-01 -2.85542198e-02
-1.94276974e-01 -9.81190383e-01 6.18087173e-01 9.69266295e-01
7.23067105e-01 -2.14083344e-01 -2.45984256e-01 1.49543738e+00
9.27119911e-01 1.69922471e-01 -5.84197998e-01 4.21459138e-01
7.40705788e-01 9.48051929e-01 -8.86920273e-01 -1.69837788e-01
-2.25538537e-01 3.69484037e-01 -1.07667878e-01 6.41742170e-01
-7.07906604e-01 2.53027290e-01 4.15694863e-02 6.00378633e-01
4.80749279e-01 -3.52194309e-01 -4.24213797e-01 -8.82199347e-01
-3.61951441e-01 -3.78645420e-01 1.22619204e-01 -7.48501122e-01
-1.52873385e+00 1.71575829e-01 7.67933011e-01 -8.34682107e-01
-1.55115724e-01 -5.82226634e-01 -6.47095442e-01 1.52125609e+00
-1.07753420e+00 -1.23271787e+00 -6.35256693e-02 7.01413512e-01
2.95149144e-02 3.15371389e-03 5.93613327e-01 7.86434039e-02
-3.12304646e-01 -3.47693264e-01 8.29493463e-01 -5.65344036e-01
2.81461626e-01 -1.38135946e+00 -3.77205275e-02 6.21362269e-01
-4.26393867e-01 8.48162055e-01 8.67010415e-01 -1.22925460e+00
-8.89242530e-01 -8.91814411e-01 5.62020242e-01 -3.91437292e-01
5.07481873e-01 -3.22069645e-01 -1.27975237e+00 4.49613005e-01
2.01090015e-02 -4.66694564e-01 7.68975914e-01 9.49406847e-02
3.76748711e-01 2.09764332e-01 -1.38846290e+00 1.48416236e-01
3.08495373e-01 2.98410431e-02 -4.56629574e-01 8.28931630e-02
2.52344668e-01 4.41541314e-01 -1.32608533e+00 3.45742494e-01
3.13701123e-01 -9.87425148e-01 9.02507007e-01 1.76849753e-01
-3.45984325e-02 -7.19730914e-01 -4.56826329e-01 -1.10279369e+00
-4.14233297e-01 -2.03413576e-01 3.17537099e-01 1.43255246e+00
4.78096128e-01 -6.06206954e-01 4.88308817e-01 5.52706480e-01
-9.84925255e-02 -2.12319627e-01 -1.25212729e+00 -6.56680703e-01
4.09843266e-01 -4.56199169e-01 5.62463582e-01 8.50147784e-01
-4.16929394e-01 -2.02829495e-01 -4.22997177e-01 3.24501902e-01
1.14897478e+00 -5.25120854e-01 7.22756624e-01 -1.88837218e+00
-6.07372582e-01 -1.75826624e-01 -1.38456911e-01 -7.21282482e-01
1.20123819e-01 -4.96167541e-01 2.04553515e-01 -1.67975032e+00
2.08526447e-01 -6.36888325e-01 3.06108028e-01 -2.20536157e-01
2.17347071e-02 -1.58548132e-01 -3.22043687e-01 4.14068639e-01
1.00078598e-01 5.70863247e-01 7.41612673e-01 2.73136020e-01
-2.85513908e-01 2.29470149e-01 2.11722739e-02 6.51785374e-01
3.05617481e-01 -5.99034607e-01 -2.61963278e-01 1.89110234e-01
1.98825613e-01 5.54919243e-01 6.92130685e-01 -8.89470696e-01
1.41190410e-01 -3.72609049e-01 2.30568409e-01 -1.05170417e+00
5.10099173e-01 -1.01269877e+00 1.08807611e+00 4.42956448e-01
1.74499556e-01 -4.10301954e-01 1.12075852e-02 9.82591569e-01
-1.57472834e-01 -5.21200597e-01 8.86983037e-01 -1.52771771e-01
-8.81586894e-02 -5.92840388e-02 -1.06300950e+00 -6.47581875e-01
1.00536156e+00 -5.12644291e-01 -3.46664824e-02 -2.05659553e-01
-1.43225479e+00 -1.15238719e-01 6.27102733e-01 -5.27074218e-01
3.88060212e-01 -1.12067056e+00 -9.25482154e-01 8.58728513e-02
-3.74663711e-01 8.98586661e-02 7.24530935e-01 1.27101505e+00
-5.88416338e-01 -1.60398167e-02 -6.08597286e-02 -9.94481564e-01
-9.15462792e-01 5.01030922e-01 4.14520800e-01 -1.98160291e-01
-6.54700175e-02 4.57301974e-01 2.63640374e-01 -6.89672709e-01
-1.68010920e-01 -1.47892416e-01 -5.44545531e-01 1.91540048e-01
4.50797155e-02 8.89373124e-01 -2.31764615e-01 -9.02777731e-01
-3.83592725e-01 7.67930448e-01 8.51780832e-01 -5.65859139e-01
1.61174810e+00 -3.10345441e-01 -6.10051632e-01 1.04611158e+00
7.09429085e-01 8.58789161e-02 -1.26933885e+00 7.56464228e-02
1.87107891e-01 -1.55799747e-01 -6.82060570e-02 -5.32483935e-01
-4.73189116e-01 8.42026114e-01 5.38344562e-01 3.66548389e-01
7.76123822e-01 4.53931317e-02 -5.20218492e-01 -1.29983306e-01
3.62409562e-01 -8.67688775e-01 -4.34198618e-01 7.05678537e-02
8.61188054e-01 -6.52219534e-01 2.34261472e-02 -7.09678531e-01
-3.54525298e-01 9.60069656e-01 6.72927722e-02 -4.91608679e-01
1.37724507e+00 4.72122312e-01 -5.22469163e-01 -2.63622195e-01
-4.00434285e-01 -2.72268236e-01 3.28844041e-01 7.82901347e-01
6.32044196e-01 1.80817157e-01 -5.81569731e-01 6.28357172e-01
3.34503889e-01 2.72122528e-02 4.67345446e-01 4.71521556e-01
-6.94468379e-01 -1.00257242e+00 -1.15371549e+00 2.74386704e-01
-1.42008767e-01 -1.25951201e-01 -5.77085763e-02 7.50107586e-01
1.74302563e-01 1.10627937e+00 3.73380393e-01 1.04643330e-01
-4.26998176e-03 2.11914599e-01 2.36880720e-01 -4.81816083e-01
-3.65898103e-01 8.00431550e-01 -7.74834678e-02 7.75071159e-02
-6.64501846e-01 -1.30851090e+00 -9.30329263e-01 -2.20457330e-01
-5.89176714e-01 2.84030378e-01 1.12159133e+00 9.03509736e-01
1.03893653e-01 5.14523268e-01 3.48407269e-01 -1.12104142e+00
-3.46584767e-01 -1.38014567e+00 -1.30650377e+00 -9.64659825e-02
-1.15765870e-01 -8.93310606e-01 -4.82711673e-01 1.84953213e-01]
|
[6.752548694610596, 3.9409313201904297]
|
d6d53c65-b5c8-47c4-8396-f1eec03a2baa
|
a-new-class-of-explanations-for-classifiers
|
2304.1476
| null |
https://arxiv.org/abs/2304.14760v1
|
https://arxiv.org/pdf/2304.14760v1.pdf
|
A New Class of Explanations for Classifiers with Non-Binary Features
|
Two types of explanations have received significant attention in the literature recently when analyzing the decisions made by classifiers. The first type explains why a decision was made and is known as a sufficient reason for the decision, also an abductive or PI-explanation. The second type explains why some other decision was not made and is known as a necessary reason for the decision, also a contrastive or counterfactual explanation. These explanations were defined for classifiers with binary, discrete and, in some cases, continuous features. We show that these explanations can be significantly improved in the presence of non-binary features, leading to a new class of explanations that relay more information about decisions and the underlying classifiers. Necessary and sufficient reasons were also shown to be the prime implicates and implicants of the complete reason for a decision, which can be obtained using a quantification operator. We show that our improved notions of necessary and sufficient reasons are also prime implicates and implicants but for an improved notion of complete reason obtained by a new quantification operator that we define and study in this paper.
|
['Adnan Darwiche', 'Chunxi Ji']
|
2023-04-28
| null | null | null | null |
['counterfactual-explanation']
|
['miscellaneous']
|
[ 4.47505534e-01 7.68742263e-01 -3.91536772e-01 -5.90640962e-01
1.25442892e-01 -6.64403260e-01 9.97456074e-01 6.22265160e-01
-2.97377199e-01 9.54090059e-01 2.88956642e-01 -7.23984122e-01
-6.66041076e-01 -9.28021610e-01 -5.97094119e-01 -7.57390857e-01
1.17430449e-01 4.11686689e-01 2.37852171e-01 -3.68278086e-01
4.02522087e-01 5.96072555e-01 -2.15545869e+00 2.96484500e-01
6.28776014e-01 8.77539456e-01 -2.25065604e-01 6.96264982e-01
-3.27783018e-01 6.43769801e-01 -3.49057972e-01 -5.19987524e-01
1.96929723e-01 -6.21009171e-01 -1.04077172e+00 2.72673845e-01
1.22460864e-01 2.77711362e-01 4.40915376e-01 1.02340066e+00
-3.23712647e-01 -1.43125221e-01 9.67925191e-01 -1.89133227e+00
-5.37367761e-01 9.46715176e-01 -1.93599626e-01 7.68601447e-02
4.77072984e-01 -3.29797387e-01 1.22474742e+00 -1.58843756e-01
3.47834706e-01 1.33990753e+00 2.62044400e-01 7.92071164e-01
-1.15459836e+00 -4.96253490e-01 1.94776475e-01 3.77469182e-01
-7.56375253e-01 -3.09734605e-02 4.14392799e-01 -4.12997186e-01
5.32892287e-01 8.43310833e-01 6.16014659e-01 4.68948811e-01
3.49671453e-01 5.68814635e-01 1.31892908e+00 -8.71101677e-01
3.94277096e-01 4.90755498e-01 7.20650613e-01 6.90064192e-01
7.69217789e-01 3.88095111e-01 -6.12071902e-02 -3.00460219e-01
5.46620309e-01 8.86709467e-02 -4.89655375e-01 -1.47645235e-01
-9.91878271e-01 1.27553117e+00 3.32564056e-01 6.56511307e-01
-4.23996329e-01 6.84307069e-02 6.38440996e-03 3.95273089e-01
-1.92484617e-01 6.12750888e-01 -7.85166860e-01 1.50413454e-01
-2.27723747e-01 2.79028863e-01 1.11515713e+00 5.48363686e-01
9.16760743e-01 -3.90204251e-01 3.22182961e-02 -1.90819919e-01
6.68122247e-02 2.28847712e-01 5.45718431e-01 -9.47490513e-01
1.02302395e-01 1.02426958e+00 3.46572667e-01 -8.64259660e-01
-6.62394047e-01 -3.44222188e-01 -6.45306528e-01 3.85262460e-01
8.85582924e-01 1.40676916e-01 -4.13669854e-01 2.14285398e+00
2.20463187e-01 -1.42356053e-01 3.18491936e-01 5.85871935e-01
3.71089399e-01 1.27390981e-01 -4.97199930e-02 -6.50889099e-01
1.42087865e+00 -4.09133136e-01 -8.73518229e-01 -1.68740004e-02
7.11855531e-01 -5.38599968e-01 1.01709449e+00 4.77447093e-01
-6.10891104e-01 -2.11892143e-01 -1.10468853e+00 1.85220614e-01
-4.40388858e-01 -5.38636819e-02 1.11533821e+00 8.40261817e-01
-7.47997165e-01 7.49281645e-01 -3.67454678e-01 -3.75716567e-01
-2.83147879e-02 4.36440647e-01 -4.70526576e-01 6.72488883e-02
-1.05586684e+00 8.53013039e-01 3.29358518e-01 -1.55097231e-01
-2.74757028e-01 -2.35590652e-01 -6.73063636e-01 3.73281568e-01
5.85934520e-01 -4.20476764e-01 1.13277054e+00 -1.05140388e+00
-9.98948514e-01 9.20566440e-01 -2.62491137e-01 -7.46526241e-01
6.07695401e-01 2.30505913e-01 -4.38224107e-01 -3.95289175e-02
3.63056511e-01 2.44755208e-01 4.47762281e-01 -1.00198710e+00
-8.22620511e-01 -6.78905785e-01 6.84261560e-01 -1.10239677e-01
1.89090520e-01 -1.59254551e-01 4.17813987e-01 -1.67194277e-01
4.82969224e-01 -9.87164974e-01 -2.35106885e-01 -9.87715796e-02
-7.02811897e-01 -4.86815393e-01 4.72875625e-01 -8.66371468e-02
8.62731934e-01 -1.88690495e+00 3.61570641e-02 2.12904140e-01
3.42120528e-01 -1.66116774e-01 3.60238791e-01 9.78056267e-02
-6.06386364e-01 6.37672901e-01 -2.63160050e-01 1.62948802e-01
3.16447675e-01 5.74702263e-01 -4.58445132e-01 6.10230207e-01
3.23745787e-01 3.20930839e-01 -5.59197545e-01 -3.29712987e-01
2.59324253e-01 9.41236094e-02 -5.59579074e-01 -1.12832636e-01
-1.76147357e-01 1.96004361e-01 -5.20367026e-01 -1.14330221e-02
5.40626585e-01 -5.16046621e-02 6.62100390e-02 8.52969736e-02
-3.53550583e-01 3.81884068e-01 -1.46804833e+00 7.13845193e-01
-2.79310465e-01 3.55830908e-01 -9.46195647e-02 -1.30155373e+00
8.92285705e-01 3.62167239e-01 4.72626500e-02 -7.41135329e-02
6.50391877e-01 4.28779662e-01 1.80628046e-01 -3.80617976e-01
2.39369005e-01 -9.03867781e-01 -1.95442274e-01 5.50007880e-01
-4.73717123e-01 -1.21537670e-01 2.46538654e-01 1.49285376e-01
9.27661002e-01 -3.47652495e-01 9.62379217e-01 -6.01571739e-01
1.05671036e+00 -5.39843109e-04 8.79704773e-01 8.06083322e-01
2.91986689e-02 4.58519995e-01 1.17385614e+00 -5.84644258e-01
-3.92175138e-01 -8.38096142e-01 -2.88486034e-01 5.85324824e-01
2.39345789e-01 -3.19650531e-01 -3.74834359e-01 -9.42757905e-01
1.98870882e-01 1.18980670e+00 -1.00331306e+00 -3.52021813e-01
-1.46421552e-01 -5.98991156e-01 1.17963910e-01 4.35668170e-01
5.55493653e-01 -6.80929363e-01 -1.19550025e+00 -1.00335158e-01
-2.00444877e-01 -6.85101807e-01 -2.51980000e-05 7.32961059e-01
-9.91284966e-01 -1.71906638e+00 -4.43224870e-02 1.06465541e-01
5.87021291e-01 -6.25202060e-02 9.19253349e-01 6.42429531e-01
2.31457248e-01 4.18758839e-02 -2.25998715e-01 -6.59859836e-01
-7.91285098e-01 -4.88533854e-01 3.61789286e-01 8.88396353e-02
2.76775926e-01 -4.11421448e-01 -2.35471819e-02 4.35736477e-01
-8.61364007e-01 -1.59244444e-02 4.88998145e-01 6.32725298e-01
4.80611324e-01 4.20624346e-01 3.16966355e-01 -9.36010122e-01
2.55121201e-01 -2.17476502e-01 -6.63067102e-01 2.46682718e-01
-7.82233000e-01 9.73474085e-01 7.20411360e-01 -3.37662518e-01
-1.08166528e+00 -4.82017128e-03 -1.16056893e-02 1.89509049e-01
-4.50407743e-01 4.10251081e-01 -7.62766004e-01 2.21360236e-01
7.25226164e-01 -2.22295254e-01 -2.93600649e-01 -4.09335136e-01
3.75050187e-01 5.24320424e-01 3.61446708e-01 -5.80289423e-01
6.74319923e-01 6.18371248e-01 7.63007045e-01 -3.52406114e-01
-9.12100017e-01 -2.22962096e-01 -4.82304782e-01 1.82209611e-01
8.10780525e-01 -1.14086337e-01 -8.79584193e-01 -2.59663671e-01
-1.06728160e+00 2.87159741e-01 -6.36737347e-01 5.49700439e-01
-7.27563500e-01 1.92506269e-01 -4.17734422e-02 -1.11957693e+00
3.44458997e-01 -1.37958407e+00 4.46895003e-01 2.73828089e-01
-5.04073381e-01 -9.06529486e-01 -4.69343454e-01 3.69398738e-03
-3.09017412e-02 4.00142223e-01 1.34972000e+00 -1.17227244e+00
-3.73062044e-02 -5.20765305e-01 -2.24807069e-01 1.55105099e-01
3.51967245e-01 9.04763192e-02 -6.10351443e-01 3.80399019e-01
3.55741084e-01 3.80480051e-01 7.62113750e-01 4.74796474e-01
5.36046803e-01 -7.66224802e-01 -4.17304873e-01 8.34355038e-03
1.45631397e+00 2.39543617e-01 4.28936034e-01 2.61824638e-01
9.75944251e-02 8.16356122e-01 5.94928563e-01 1.53116331e-01
2.94321030e-01 7.69473851e-01 7.27336168e-01 1.05806120e-01
2.14514568e-01 6.44017532e-02 -1.08150952e-01 -6.68359101e-02
-2.42272362e-01 1.59097433e-01 -5.95688164e-01 4.17938292e-01
-1.93549490e+00 -1.03300560e+00 -9.92566288e-01 2.36960983e+00
7.40302742e-01 2.32690096e-01 1.24031939e-01 9.22545016e-01
9.87083077e-01 -4.82053906e-01 -2.39113495e-01 -9.71714914e-01
-2.56928861e-01 7.70952031e-02 4.67132539e-01 8.42415333e-01
-8.10963035e-01 2.92196721e-01 5.64375734e+00 4.16270792e-01
-7.81922698e-01 2.01808140e-02 1.80471137e-01 4.06284600e-01
-5.37061989e-01 5.25748610e-01 -6.43351674e-01 1.61363870e-01
6.41328871e-01 -4.58831042e-01 -2.54761763e-02 1.03291619e+00
-1.53514385e-01 -5.81159770e-01 -1.63893616e+00 5.17538428e-01
-3.41003776e-01 -1.17034924e+00 -5.81320487e-02 3.48070771e-01
5.08733392e-01 -9.81235147e-01 -3.02519590e-01 -7.42863305e-03
3.57554525e-01 -7.11996078e-01 1.01359320e+00 2.57400960e-01
3.18716705e-01 -7.86469340e-01 1.06990969e+00 5.64899087e-01
-9.63780105e-01 -4.10280526e-01 -2.68142670e-01 -9.90389824e-01
-2.80743569e-01 7.75625229e-01 -4.61534590e-01 6.03137493e-01
1.56653449e-01 2.53370136e-01 -3.79515648e-01 8.48115206e-01
-8.63940179e-01 3.65492135e-01 -3.77937734e-01 -1.15853682e-01
1.59088865e-01 6.07716814e-02 6.78946793e-01 8.67144644e-01
1.82445899e-01 3.26553464e-01 -4.02382582e-01 1.03255045e+00
4.10985678e-01 -1.05111264e-01 -4.06215101e-01 3.40746611e-01
1.24209344e-01 8.76185417e-01 -1.04720247e+00 -4.43685681e-01
5.28906845e-02 5.91592073e-01 -2.51332551e-01 7.16683567e-02
-7.93793440e-01 -2.53928334e-01 6.66230798e-01 1.22052707e-01
-1.51014507e-01 4.10481870e-01 -4.94508713e-01 -1.31349504e+00
-9.15663466e-02 -4.06213671e-01 7.13560820e-01 -3.86694938e-01
-9.98943150e-01 4.11771715e-01 4.46874917e-01 -8.68411303e-01
-5.24980366e-01 -7.75172532e-01 -6.89575970e-01 9.28308427e-01
-1.20780754e+00 -5.55699348e-01 -3.71944383e-02 5.98330975e-01
4.58979569e-02 2.12582827e-01 9.01863992e-01 -4.94785428e-01
-2.34215274e-01 6.49874806e-02 -5.54187536e-01 -3.24516863e-01
6.35220557e-02 -1.53277802e+00 -4.11972463e-01 9.82199550e-01
6.29073605e-02 6.45812571e-01 1.56545770e+00 -2.57758856e-01
-9.52665687e-01 -5.45718551e-01 1.67440176e+00 -2.02128381e-01
4.05243784e-01 1.58662215e-01 -7.30707288e-01 8.31180036e-01
-5.22098988e-02 -4.25148219e-01 4.51611519e-01 4.40246254e-01
-4.92976546e-01 -3.68829072e-02 -1.44216502e+00 4.90964144e-01
6.65720046e-01 -7.64745325e-02 -1.21294856e+00 -8.60047899e-03
6.15643203e-01 1.71309233e-01 -1.16814211e-01 3.49050611e-01
6.23574853e-01 -1.46093833e+00 6.45686865e-01 -6.60847008e-01
4.35714245e-01 -5.72034717e-01 -4.98014927e-01 -1.02677155e+00
-1.92801237e-01 -2.94059545e-01 2.57856518e-01 1.13677156e+00
6.44020736e-01 -1.15866864e+00 3.98990959e-01 1.10881519e+00
1.87214777e-01 -5.20351827e-01 -1.23580456e+00 -8.31336081e-01
8.88628811e-02 -6.95309699e-01 9.13705051e-01 9.69381750e-01
4.13622737e-01 3.13035339e-01 1.30088836e-01 -3.08449250e-02
4.74915802e-01 3.72679889e-01 3.09421480e-01 -1.72535276e+00
-3.84825826e-01 -7.49844730e-01 -7.04606712e-01 -4.41235960e-01
2.10979477e-01 -8.53463948e-01 -1.06041119e-01 -1.50589645e+00
5.47506437e-02 -4.50351387e-01 -1.13846837e-02 8.17250967e-01
6.58428222e-02 -1.86115995e-01 2.87846267e-01 8.45206976e-02
4.95740734e-02 1.70635328e-01 8.80019009e-01 6.06593378e-02
-1.87093675e-01 3.11317086e-01 -1.17378688e+00 1.10102475e+00
7.69843638e-01 -6.08606696e-01 -2.71529377e-01 2.80805707e-01
6.77698135e-01 3.99316639e-01 6.38298452e-01 -8.06281149e-01
-7.76388124e-02 -3.34630072e-01 -3.26498821e-02 -1.24709234e-01
2.32128017e-02 -1.03986573e+00 3.40796560e-01 9.15892184e-01
-6.89106822e-01 -1.80756792e-01 1.61577817e-02 3.87416959e-01
6.27746657e-02 -5.96722007e-01 6.87138140e-01 6.90264627e-02
-4.11642104e-01 -4.57832277e-01 -3.69716883e-01 -4.07352418e-01
8.63470495e-01 -2.42483243e-01 -2.73183644e-01 -5.70615172e-01
-1.04630649e+00 -8.82204548e-02 3.78623605e-01 6.43853545e-02
4.19479638e-01 -1.18431878e+00 -4.62977380e-01 -1.08824089e-01
4.61055636e-02 -4.78666157e-01 -1.55818030e-01 1.14803851e+00
8.46268833e-02 5.22314668e-01 -2.81834662e-01 -2.51719922e-01
-1.38839865e+00 7.46867955e-01 2.57492959e-01 -2.64840215e-01
-4.33884919e-01 5.90798497e-01 5.16029894e-01 -1.79148436e-01
-8.84244666e-02 -8.51697445e-01 -4.40461814e-01 3.92363779e-02
5.70043802e-01 3.98630828e-01 -1.35797873e-01 -7.37885058e-01
-6.76494002e-01 5.17019331e-01 5.08852065e-01 -1.48886919e-01
1.11941111e+00 -1.59760252e-01 -2.73442954e-01 6.13929272e-01
8.17951560e-01 5.22038303e-02 -8.25003088e-01 1.48468122e-01
3.08645889e-02 -2.75689512e-01 -2.92013735e-01 -1.02733219e+00
-8.89372647e-01 9.46576893e-01 7.81070441e-02 8.60956311e-01
1.32914758e+00 3.69372934e-01 -9.88856703e-02 2.73327500e-01
5.24789989e-01 -4.64636028e-01 -6.57731354e-01 1.51851669e-01
1.22423697e+00 -9.36777532e-01 -1.83139399e-01 -8.58718038e-01
-5.83882153e-01 1.47760642e+00 4.01084311e-02 -5.88001451e-03
5.80206454e-01 2.70195991e-01 -2.44013220e-01 9.96746961e-03
-7.44590104e-01 -5.42236447e-01 2.27514386e-01 4.26038772e-01
3.85295749e-01 4.88058746e-01 -9.92398798e-01 1.38605535e+00
-5.10335684e-01 2.36396119e-01 8.87019932e-01 6.34745061e-01
-3.81442219e-01 -1.10291040e+00 -6.72471642e-01 3.52724761e-01
-3.90189558e-01 1.02673747e-01 -6.81905091e-01 1.14993954e+00
6.22682512e-01 1.46347189e+00 -1.18973225e-01 -4.68945563e-01
2.30649322e-01 8.70454013e-02 4.72387195e-01 -4.82549995e-01
-3.44663262e-01 -4.10570174e-01 1.08784840e-01 -3.34653795e-01
-5.52538335e-01 -6.83228791e-01 -1.71218598e+00 -4.07031029e-01
-7.57388890e-01 7.30257452e-01 6.14769459e-01 1.57077301e+00
-1.73657686e-01 3.55281860e-01 4.25467402e-01 -3.85667026e-01
-6.12807512e-01 -6.77712679e-01 -7.81934798e-01 1.93506911e-01
4.03392971e-01 -9.20372665e-01 -1.15153039e+00 -3.51182595e-02]
|
[8.732662200927734, 5.792830467224121]
|
7017ba58-336d-4f41-beae-599af3d44ce4
|
back-to-the-drawing-board-a-critical
|
2108.10241
| null |
https://arxiv.org/abs/2108.10241v2
|
https://arxiv.org/pdf/2108.10241v2.pdf
|
Back to the Drawing Board: A Critical Evaluation of Poisoning Attacks on Production Federated Learning
|
While recent works have indicated that federated learning (FL) may be vulnerable to poisoning attacks by compromised clients, their real impact on production FL systems is not fully understood. In this work, we aim to develop a comprehensive systemization for poisoning attacks on FL by enumerating all possible threat models, variations of poisoning, and adversary capabilities. We specifically put our focus on untargeted poisoning attacks, as we argue that they are significantly relevant to production FL deployments. We present a critical analysis of untargeted poisoning attacks under practical, production FL environments by carefully characterizing the set of realistic threat models and adversarial capabilities. Our findings are rather surprising: contrary to the established belief, we show that FL is highly robust in practice even when using simple, low-cost defenses. We go even further and propose novel, state-of-the-art data and model poisoning attacks, and show via an extensive set of experiments across three benchmark datasets how (in)effective poisoning attacks are in the presence of simple defense mechanisms. We aim to correct previous misconceptions and offer concrete guidelines to conduct more accurate (and more realistic) research on this topic.
|
['Daniel Ramage', 'Peter Kairouz', 'Amir Houmansadr', 'Virat Shejwalkar']
|
2021-08-23
| null | null | null | null |
['misconceptions']
|
['miscellaneous']
|
[-1.30527884e-01 -2.67798096e-01 1.00166604e-01 8.71082842e-02
-7.37573922e-01 -1.17395878e+00 7.41536975e-01 -1.27769634e-02
-3.84388655e-01 5.79913437e-01 1.24043584e-01 -8.34369123e-01
-1.82200179e-01 -5.82215726e-01 -7.30333209e-01 -6.93393111e-01
-3.46812308e-01 2.66786605e-01 4.27395672e-01 -2.75122225e-01
2.12418124e-01 1.03541648e+00 -1.27875280e+00 2.21277967e-01
2.47158259e-01 6.67499244e-01 -8.46862376e-01 6.54714823e-01
1.77649468e-01 1.35246146e+00 -1.09913456e+00 -9.78534579e-01
6.44609511e-01 -1.49137571e-01 -1.04750431e+00 5.04193678e-02
4.73114371e-01 -6.67776108e-01 -5.21376431e-01 1.02857590e+00
5.78948557e-01 -4.33337152e-01 3.05370271e-01 -1.88396621e+00
-3.18504870e-01 1.13338411e+00 -2.58507311e-01 3.11515093e-01
1.37066618e-01 9.05376971e-01 6.72860503e-01 -1.56364053e-01
3.52656692e-01 1.34969521e+00 7.91029632e-01 8.64911318e-01
-1.09828925e+00 -1.04836702e+00 1.89933509e-01 -1.36466231e-02
-1.13999903e+00 -5.68399131e-01 4.68340546e-01 -1.56957895e-01
7.95218170e-01 5.72905004e-01 -1.04046911e-02 1.64930272e+00
8.82923901e-02 5.78774929e-01 1.24370503e+00 -4.43689376e-01
3.52873236e-01 3.19031924e-01 3.68148118e-01 5.14429033e-01
8.67521226e-01 2.13040963e-01 -3.40221435e-01 -1.04883540e+00
4.13410544e-01 -2.00910255e-01 -4.92385805e-01 -5.65529168e-01
-7.44578719e-01 1.03618014e+00 2.21562862e-01 2.52249181e-01
-2.76148468e-01 3.44511598e-01 5.68709552e-01 3.86848718e-01
-8.06501955e-02 5.71222365e-01 -5.65337360e-01 1.26149073e-01
-4.22584295e-01 3.59243423e-01 1.23235667e+00 7.08796442e-01
4.63683188e-01 2.90513724e-01 1.27770454e-01 5.06826229e-02
2.38435879e-01 3.71154845e-01 2.07230598e-01 -1.07007861e+00
2.28924602e-01 1.97034076e-01 2.22741380e-01 -8.18904281e-01
-1.90325513e-01 -4.10005003e-01 -2.19563097e-01 2.40042657e-01
6.23147964e-01 -5.20442903e-01 -2.33899921e-01 1.80780089e+00
3.82538915e-01 2.46732637e-01 2.21707240e-01 5.87821782e-01
1.03806943e-01 -1.47771183e-02 5.04226387e-01 -2.01799765e-01
1.25713301e+00 -6.32468045e-01 -3.75789344e-01 9.60564241e-02
8.94994617e-01 -3.70512843e-01 1.04293048e+00 6.34582937e-01
-9.06984508e-01 3.09197724e-01 -8.84894073e-01 6.35586083e-01
-3.80235851e-01 -9.37604487e-01 9.24159765e-01 1.46723151e+00
-8.56132388e-01 4.17341679e-01 -9.14962351e-01 -8.23916316e-01
5.61937034e-01 3.90551567e-01 -2.20166758e-01 -2.00005397e-02
-1.13403857e+00 9.77711082e-01 1.59106124e-02 -5.93392670e-01
-1.52339184e+00 -6.63238764e-01 -3.72760326e-01 -3.70941162e-02
6.83561683e-01 -7.22420871e-01 1.41303301e+00 -6.42141044e-01
-9.52534497e-01 5.23993850e-01 5.65263450e-01 -9.64335978e-01
6.35792792e-01 -7.63453245e-02 -4.22084898e-01 5.14989316e-01
-4.84072149e-01 1.03431925e-01 5.76202273e-01 -1.80660415e+00
-4.47084188e-01 -3.98861080e-01 5.92060208e-01 -2.73334235e-01
-1.07504952e+00 8.71524930e-01 2.01709732e-01 -3.03726822e-01
-6.60070181e-01 -6.59198999e-01 -3.77882391e-01 -2.57267579e-02
-5.20265341e-01 2.26243474e-02 1.08826435e+00 -7.89403319e-02
1.16151166e+00 -1.78669727e+00 -5.11305928e-01 4.61604036e-02
3.67968380e-01 6.07206881e-01 -1.58510715e-01 8.60912979e-01
2.23734766e-01 5.89962542e-01 -1.40644521e-01 -2.40304545e-01
3.99986863e-01 1.38771906e-01 -8.06885183e-01 8.53975892e-01
-7.55461901e-02 6.92956626e-01 -7.65828907e-01 -4.78654385e-01
2.54364647e-02 4.30109262e-01 -6.17691100e-01 2.19984591e-01
-3.32605928e-01 2.95010507e-01 -4.97608334e-01 8.94867837e-01
5.45408010e-01 1.33869108e-02 2.32252017e-01 -7.32038841e-02
-7.21369609e-02 5.06286249e-02 -8.82273495e-01 7.80413032e-01
-5.01056723e-02 3.18098404e-02 5.11027038e-01 -4.61575359e-01
4.44799691e-01 3.56328517e-01 4.96085167e-01 -2.01739684e-01
4.98444140e-01 4.08049971e-02 -2.58617178e-02 -3.79051208e-01
1.49515942e-02 -2.62309283e-01 -2.04211637e-01 9.77030277e-01
-1.22357957e-01 2.91225046e-01 -1.59138992e-01 4.48491573e-01
1.77368426e+00 -4.75384384e-01 5.81744984e-02 -3.81000221e-01
3.98652554e-01 1.88386470e-01 2.05447599e-01 1.20681882e+00
-6.90902233e-01 6.18330203e-02 6.16131008e-01 -3.91894013e-01
-6.58147097e-01 -9.54264522e-01 2.23148968e-02 1.04234469e+00
7.57378116e-02 -5.03747702e-01 -1.15569615e+00 -1.37569690e+00
2.02037424e-01 9.48844612e-01 -4.43467975e-01 -3.23237240e-01
-4.48534518e-01 -1.07248843e+00 1.62345052e+00 3.94129515e-01
4.33170348e-01 -9.11102772e-01 -9.02643681e-01 4.12954129e-02
1.24721117e-01 -1.02514374e+00 -8.28712583e-02 2.58651733e-01
-5.02971888e-01 -1.59201741e+00 -5.81180118e-02 -8.64501745e-02
3.01649749e-01 4.37067807e-01 1.05707800e+00 4.86450434e-01
-2.08408356e-01 7.64011085e-01 -3.82133812e-01 -4.15684819e-01
-8.53743196e-01 -1.14579216e-01 3.90625119e-01 -6.96066767e-02
6.10851705e-01 -4.91265118e-01 -4.28218067e-01 4.26532179e-01
-1.37853670e+00 -8.82057786e-01 3.60547423e-01 4.03561652e-01
-1.52592719e-01 1.72038868e-01 5.40643573e-01 -1.14990115e+00
8.16281915e-01 -7.41491616e-01 -6.06210947e-01 2.86724567e-01
-6.22028828e-01 -9.84267816e-02 9.34124470e-01 -7.58905113e-01
-7.50277936e-01 -3.63414705e-01 -2.86562145e-01 -7.20599473e-01
-4.65630144e-01 -8.81019805e-04 -4.81017947e-01 -5.97228110e-01
1.14590359e+00 6.12862818e-02 -1.95455328e-02 -6.08758271e-01
3.87594253e-01 6.70427620e-01 4.42158371e-01 -1.04943740e+00
1.18083835e+00 7.68403590e-01 -5.10659404e-02 -5.57268083e-01
-5.78413844e-01 -2.55060978e-02 -1.88526839e-01 2.49244217e-02
2.95734286e-01 -4.89128500e-01 -1.10823619e+00 5.57780087e-01
-8.27910721e-01 -5.94111919e-01 -6.33376539e-02 -8.78288075e-02
-3.79278362e-01 1.00497055e+00 -1.01845109e+00 -9.63692546e-01
-4.36604440e-01 -1.16097593e+00 3.64012718e-01 -1.35062456e-01
-1.20094761e-01 -1.02655160e+00 1.65978938e-01 4.79978114e-01
7.71910727e-01 3.47993255e-01 8.85865808e-01 -1.33022320e+00
-4.40918028e-01 -2.09989995e-01 7.33667985e-02 3.04111272e-01
-1.16587482e-01 2.11017027e-01 -9.88492012e-01 -7.58601904e-01
3.34213376e-01 -6.83636844e-01 5.00279009e-01 -4.94656861e-01
1.22204351e+00 -9.57192004e-01 -1.43439874e-01 5.32740295e-01
1.64702713e+00 -1.47431925e-01 5.40859103e-01 6.76513612e-01
4.65728790e-01 6.59427166e-01 3.02090824e-01 6.99493468e-01
1.79032549e-01 4.53639090e-01 9.65571344e-01 1.29318982e-01
6.93820044e-02 -2.20782757e-01 5.79392791e-01 -5.59866652e-02
4.50119972e-01 -5.59408128e-01 -8.94531012e-01 3.27724516e-01
-1.38973546e+00 -1.06899130e+00 -2.30944946e-01 2.29041457e+00
6.98800206e-01 2.17050999e-01 6.86380804e-01 2.75775224e-01
6.31257713e-01 1.39340293e-02 -4.34166700e-01 -4.14664865e-01
-1.70294121e-01 2.28945091e-01 9.70400810e-01 2.53491312e-01
-1.13480318e+00 9.31465983e-01 7.55695486e+00 5.69621205e-01
-1.11699867e+00 3.78648460e-01 3.87226999e-01 -2.95406640e-01
-1.67734280e-01 9.57579762e-02 -7.80682385e-01 3.03445190e-01
1.35641146e+00 -2.23741129e-01 4.64193761e-01 8.47934842e-01
-1.98584914e-01 4.38138515e-01 -1.07724178e+00 3.19027930e-01
1.03072494e-01 -1.04858077e+00 2.18685135e-01 2.82662868e-01
3.70135367e-01 2.28311524e-01 1.37678966e-01 3.79553258e-01
1.05095804e+00 -1.03782153e+00 6.04018271e-01 -1.08154193e-01
3.13453704e-01 -1.09230983e+00 5.10790825e-01 4.82040137e-01
-5.16921520e-01 -5.51272750e-01 -2.02839553e-01 -1.06519051e-02
-1.62368655e-01 2.25001238e-02 -4.39726859e-01 4.05216455e-01
5.79604447e-01 -2.36571446e-01 -9.02355552e-01 9.65394676e-01
-1.44090252e-02 1.00484252e+00 -4.42609668e-01 1.71020672e-01
4.23680931e-01 6.61803305e-01 3.74465168e-01 1.21226203e+00
-1.54858515e-01 -1.01569332e-02 1.29454717e-01 6.22666836e-01
-2.46686600e-02 -1.13660946e-01 -7.36509800e-01 4.28846814e-02
6.77751064e-01 1.30713689e+00 -5.49538195e-01 -7.01707676e-02
-3.98278892e-01 5.94416499e-01 3.35623980e-01 1.44933537e-01
-1.01075220e+00 4.83155102e-02 1.00110388e+00 1.89272419e-01
3.43064927e-02 2.68413089e-02 -1.66723043e-01 -1.33484435e+00
-5.01214981e-01 -1.64059210e+00 7.76831925e-01 -1.00353129e-01
-1.63917768e+00 7.51378775e-01 1.74336240e-01 -7.15589583e-01
3.55707072e-02 -5.77981532e-01 -7.19416142e-01 3.48170042e-01
-1.40412891e+00 -1.23444712e+00 1.11823566e-01 9.92043138e-01
2.31361911e-02 -3.65812890e-02 9.73837852e-01 1.83998376e-01
-8.91865075e-01 1.05384672e+00 -1.48269191e-01 9.16466936e-02
7.21923232e-01 -8.62717986e-01 3.58484566e-01 1.23159206e+00
4.86958772e-02 9.12754595e-01 1.05461669e+00 -6.11433625e-01
-1.80026197e+00 -1.25485456e+00 2.44747430e-01 -9.73908365e-01
9.46614444e-01 -5.11323512e-01 -8.40013623e-01 9.91888463e-01
2.35094994e-01 5.93346097e-02 7.63190269e-01 -2.75086254e-01
-8.94959629e-01 1.84179649e-01 -1.79108858e+00 7.99578428e-01
9.54533398e-01 -2.87994087e-01 -2.23051190e-01 5.40613234e-01
7.91931033e-01 2.03968674e-01 -7.30275869e-01 3.71115386e-01
3.02430242e-01 -1.36865652e+00 9.62471664e-01 -9.06579196e-01
-2.04313844e-01 -3.75357494e-02 -4.67747033e-01 -6.15061700e-01
-8.64842385e-02 -9.43624198e-01 -4.66886789e-01 1.41846585e+00
-9.50335637e-02 -9.32867050e-01 9.47521031e-01 8.83379817e-01
2.50631094e-01 -6.82585597e-01 -6.60928845e-01 -1.06589484e+00
6.03807449e-01 -4.87603366e-01 8.14669967e-01 1.17971587e+00
8.15747976e-02 -2.17526719e-01 -3.12902004e-01 6.49811685e-01
1.17362678e+00 -4.04072553e-01 1.05544448e+00 -9.33999658e-01
-5.11760473e-01 -3.52925628e-01 -2.25167349e-01 -4.11697119e-01
2.05564618e-01 -3.33786130e-01 -4.70897943e-01 -7.70927131e-01
2.68318027e-01 -3.56183589e-01 -4.85733569e-01 9.12663400e-01
5.46961501e-02 5.56430221e-01 4.22397166e-01 3.77557874e-01
-7.10548222e-01 1.19294561e-01 5.02214849e-01 4.25017215e-02
4.03566271e-01 -9.56092700e-02 -1.23954606e+00 6.32894933e-01
8.64371955e-01 -6.28603399e-01 -3.67932171e-01 -2.24627182e-01
-3.42683047e-01 -2.15345427e-01 5.53883672e-01 -1.01532495e+00
2.23606110e-01 -4.59516197e-01 -1.32786274e-01 -1.19312875e-01
2.90858820e-02 -9.98216033e-01 2.73004681e-01 8.56331050e-01
-3.07686836e-01 9.45375413e-02 1.89546764e-01 2.83242851e-01
4.45777476e-01 -2.89206833e-01 9.07562017e-01 -3.56590301e-01
-2.18781322e-01 3.79731327e-01 -4.96464342e-01 -5.95370904e-02
1.44033873e+00 7.54671842e-02 -8.56224895e-01 -5.35301805e-01
-2.08796397e-01 1.69950798e-01 1.26346493e+00 7.90393129e-02
2.69949973e-01 -7.82827795e-01 -6.61145806e-01 8.09471682e-02
5.41383494e-03 -7.85254300e-01 -3.55852321e-02 5.31195819e-01
-5.42010188e-01 3.23104233e-01 -1.70804605e-01 -2.54126452e-03
-1.46234787e+00 1.36844552e+00 5.83779573e-01 -2.66680926e-01
-4.98603046e-01 5.99906981e-01 1.48681089e-01 -2.43001699e-01
7.15236545e-01 3.76613081e-01 2.11900488e-01 -4.94946212e-01
7.75600195e-01 4.45105076e-01 -2.44418770e-04 -6.58821404e-01
-6.00904047e-01 -1.95950493e-01 -1.79078817e-01 2.71562263e-02
1.07992339e+00 1.29791722e-01 -1.08938754e-01 -1.53883472e-01
8.57540607e-01 2.66051471e-01 -1.14613378e+00 -4.41976190e-02
2.54636370e-02 -6.36269331e-01 -1.82828873e-01 -1.09111249e+00
-1.14964104e+00 6.81790709e-01 2.78370798e-01 6.92228556e-01
1.13292229e+00 -4.70418669e-02 7.73528576e-01 4.38642710e-01
8.62120688e-01 -1.45715326e-01 1.37896121e-01 1.02179453e-01
4.68348563e-01 -5.15536785e-01 -4.69134897e-02 -4.67136532e-01
-5.29614747e-01 8.03701401e-01 7.85783947e-01 -1.11240111e-01
3.22472692e-01 6.83982611e-01 3.61314714e-01 -3.12267184e-01
-1.04255319e+00 1.19955882e-01 -7.98245609e-01 9.14667368e-01
-1.01329789e-01 -3.98440689e-01 -1.11496020e-02 7.21086979e-01
-2.42620446e-02 -4.05105919e-01 8.85388672e-01 1.22549284e+00
-3.33865583e-01 -1.40567470e+00 -9.92087841e-01 4.56566401e-02
-1.10817099e+00 2.19516754e-01 -9.12647128e-01 9.36630785e-01
-8.82489681e-02 1.43824863e+00 -5.06122887e-01 -5.07972836e-01
1.13245681e-01 1.09649107e-01 4.79960442e-01 -3.89216363e-01
-1.32013595e+00 -2.56671131e-01 1.76759377e-01 -5.93179703e-01
-1.19082322e-02 -5.83344281e-01 -7.46706963e-01 -9.32434082e-01
-5.18907607e-01 2.69639909e-01 4.77932900e-01 6.17516816e-01
2.68869996e-01 4.39200737e-03 8.61508906e-01 -5.14110804e-01
-1.51391816e+00 -6.55501902e-01 -5.99970996e-01 5.95019341e-01
2.56170705e-02 -4.68934298e-01 -9.14002657e-01 -4.58146870e-01]
|
[5.774926662445068, 7.538160800933838]
|
28bfa73c-a471-44ec-8b49-ca6f3da3c87b
|
divide-and-conquer-based-large-scale-spectral
| null | null |
https://www.researchgate.net/publication/351270623_Divide-and-conquer_based_Large-Scale_Spectral_Clustering
|
https://www.researchgate.net/profile/Hongmin-Li-7/publication/351270623_Divide-and-conquer_based_Large-Scale_Spectral_Clustering/links/608eb38a458515d315efa6ec/Divide-and-conquer-based-Large-Scale-Spectral-Clustering.pdf?_sg%5B0%5D=afU-HPOj3kpiSK8_UmQ5jVrrRNuCh7_Q3GujgkM7XaWsiGsIi1kJ8pHQEcENHYg9XZ0SDLQY7Rd3x77HT2KPcA.nHhOI-GZhe624JDE5doTvQLo1As-oDHbDpTzp_NHuikCGKpMfsUWbbVWUoPI6qebXzJ77Rxo6LaT8lWNTtwoaA&_sg%5B1%5D=1QJ0XmAml55eiMaCP05QNz4ti9936Drq2SGatgnmF6wm1CxH7W-bft3yXs4GlzacAIvoxgu6V4DYKya__KBObmpZ6hhuxFvRvqyPKR2SqEKT.nHhOI-GZhe624JDE5doTvQLo1As-oDHbDpTzp_NHuikCGKpMfsUWbbVWUoPI6qebXzJ77Rxo6LaT8lWNTtwoaA&_iepl=
|
Divide-and-conquer based Large-Scale Spectral Clustering
|
Spectral clustering is one of the most popular clustering methods. However, how to balance the efficiency and effectiveness of the large-scale spectral clustering with limited computing resources has not been properly solved for a long time. In this paper, we propose a divide-and-conquer based large-scale spectral clustering method to strike a good balance between efficiency and effectiveness. In the proposed method, a divide-and-conquer based landmark selection algorithm and a novel approximate similarity matrix approach are designed to construct a sparse similarity matrix within low computational complexities. Then clustering results can be computed quickly through a bipartite graph partition process. The proposed method achieves a lower computational complexity than most existing large-scale spectral clustering. Experimental results on ten large-scale datasets have demonstrated the efficiency and effectiveness of the proposed methods. The MATLAB code of the proposed method and experimental datasets are available at https://github.com/Li- Hongmin/MyPaperWithCode.
|
['Tetsuya Sakurai', 'Akira Imakura', 'Xiucai Ye', 'Hongmin Li']
|
2021-05-02
| null | null | null | null |
['image-clustering', 'imagedocument-clustering']
|
['computer-vision', 'computer-vision']
|
[-3.12911958e-01 -4.59122539e-01 -1.72387898e-01 -2.31059507e-01
-8.35836112e-01 -5.17925739e-01 8.42472985e-02 1.01355016e-01
-2.46727347e-01 4.65711921e-01 -1.30289569e-02 -2.13728771e-01
-5.62833548e-01 -7.98951089e-01 -1.86868787e-01 -9.15235043e-01
7.65229613e-02 4.94987577e-01 6.05738282e-01 3.56105238e-01
2.73912340e-01 1.41713127e-01 -1.12049639e+00 -1.27186090e-01
1.13160074e+00 6.93260133e-01 4.68145847e-01 3.02263737e-01
-1.24312937e-01 2.17218935e-01 -2.47405931e-01 1.06836319e-01
4.54664856e-01 -6.44212544e-01 -6.43060982e-01 3.31542909e-01
-1.05313070e-01 1.61139369e-01 -2.25073859e-01 1.23088896e+00
7.78127909e-01 2.61232585e-01 3.49647373e-01 -1.40244389e+00
-3.04564506e-01 6.72454178e-01 -1.26942408e+00 1.30991712e-01
2.69933194e-01 -1.69684306e-01 7.70603478e-01 -7.72298038e-01
4.34219748e-01 1.06118333e+00 6.51036263e-01 -1.04728155e-01
-1.27703130e+00 -1.07988822e+00 3.02297752e-02 3.21672380e-01
-2.22398043e+00 -2.21153498e-01 8.58885586e-01 -1.86775923e-01
3.85060549e-01 3.11093330e-01 7.24859774e-01 1.15293503e-01
-4.35175091e-01 6.40959501e-01 1.16062939e+00 -1.83411181e-01
3.41597915e-01 -6.54042289e-02 3.49323787e-02 7.31151760e-01
4.33092266e-01 -3.73358220e-01 -7.72472471e-02 -4.86921191e-01
5.75257301e-01 3.98352027e-01 -2.75404334e-01 -6.16204083e-01
-1.16055357e+00 7.78462887e-01 5.18842995e-01 4.35118943e-01
-1.71514049e-01 1.10972770e-01 3.47828656e-01 7.33258054e-02
2.34665215e-01 -1.93658397e-01 -1.13413082e-02 4.06288868e-03
-1.28847158e+00 6.30951673e-03 5.47126114e-01 9.93084252e-01
1.07739913e+00 -2.13731125e-01 3.34708452e-01 1.02717650e+00
2.21637011e-01 6.16952360e-01 4.85070556e-01 -7.63408959e-01
2.48080567e-01 8.09674561e-01 -1.70035288e-01 -1.54458952e+00
-4.24947977e-01 -2.33883664e-01 -1.15987778e+00 -3.65676075e-01
2.94108540e-01 -1.24568112e-01 -4.38780546e-01 1.32567954e+00
8.18248570e-01 6.63975418e-01 -1.75988302e-01 1.03004646e+00
5.06149232e-01 9.02495086e-01 -1.75419018e-01 -6.31514847e-01
1.23580086e+00 -9.34727728e-01 -6.04261100e-01 1.11781657e-01
4.97789711e-01 -9.57606852e-01 7.91774750e-01 1.63633227e-01
-9.53565300e-01 -3.21702272e-01 -9.55532730e-01 3.31968188e-01
-6.75720954e-03 1.51259407e-01 5.84920943e-01 5.60190916e-01
-9.08547163e-01 2.42557108e-01 -1.00031364e+00 -7.38790512e-01
2.68359780e-01 4.77308989e-01 -1.83908060e-01 -9.59741473e-02
-8.37354004e-01 1.18659325e-01 6.70388699e-01 7.96364397e-02
-3.60344917e-01 -3.58391911e-01 -3.96461040e-01 6.57119229e-02
8.61459017e-01 -3.90306115e-01 9.16860402e-01 -7.05877841e-01
-1.08582699e+00 5.16813159e-01 -2.27684975e-01 3.82776745e-02
2.16448113e-01 1.96626961e-01 -2.38701865e-01 3.91292334e-01
3.15868318e-01 2.06906959e-01 2.70279139e-01 -1.28854811e+00
-7.89492309e-01 -4.41839695e-01 -4.97117966e-01 3.15318555e-01
-4.50132936e-01 3.70574556e-02 -8.31495285e-01 -5.32324851e-01
5.54812610e-01 -1.12215817e+00 -5.59524536e-01 -3.79303902e-01
-4.04184192e-01 -2.37406895e-01 6.99170530e-01 -2.29279384e-01
1.72279060e+00 -2.23927999e+00 5.05667515e-02 7.86085308e-01
1.94629773e-01 2.19693705e-01 1.26823381e-01 7.69519985e-01
-5.51233999e-02 7.98943862e-02 -3.16283464e-01 2.31695652e-01
-1.34040520e-01 -2.99626440e-01 2.11155251e-01 7.71893024e-01
-5.91434896e-01 5.00905573e-01 -9.58365023e-01 -8.45668018e-01
1.54946014e-01 1.83199361e-01 -2.65529424e-01 1.18092656e-01
2.81163067e-01 -8.17457493e-03 -6.94058418e-01 6.45678937e-01
9.59665716e-01 -4.54861045e-01 6.20292664e-01 -3.73884588e-01
-7.20136091e-02 -3.38045746e-01 -1.94784319e+00 1.63323593e+00
-1.78368799e-02 1.16641253e-01 3.33520979e-01 -1.25726378e+00
9.33317840e-01 2.41547123e-01 8.38528454e-01 -3.80041242e-01
2.96487421e-01 5.34902632e-01 2.37510223e-02 -2.38190129e-01
2.18066528e-01 -1.41048327e-01 -2.45442409e-02 7.64633834e-01
-5.41089118e-01 4.46942858e-02 6.20611370e-01 5.01444399e-01
1.10936594e+00 -2.40471616e-01 5.61523736e-01 -5.45449555e-01
7.00543165e-01 3.20850879e-01 8.63500297e-01 3.23088080e-01
-3.33959848e-01 4.90484148e-01 1.83621526e-01 -8.93138424e-02
-8.99771035e-01 -9.04292881e-01 1.75767943e-01 6.93111718e-01
5.79189479e-01 -6.94345653e-01 -8.81917000e-01 -3.19321513e-01
-5.14246710e-02 -2.04237807e-03 -2.07934424e-01 7.01545998e-02
-4.10841376e-01 -7.78343379e-01 4.06501144e-01 2.28257611e-01
6.79284096e-01 -6.92040741e-01 -3.54998499e-01 1.46729633e-01
-3.47827643e-01 -7.96624184e-01 -7.64539540e-01 -2.60069162e-01
-8.55280399e-01 -1.33731306e+00 -6.11030638e-01 -1.22253954e+00
9.70084608e-01 1.02651227e+00 4.80050892e-01 4.05071527e-01
-3.92755955e-01 -3.47443074e-02 -5.80613315e-01 1.39690414e-01
1.42038405e-01 2.50686407e-01 2.69815058e-01 2.00047150e-01
5.02631128e-01 -8.83323014e-01 -9.06582355e-01 7.51958847e-01
-1.01449251e+00 -1.65521670e-02 7.42800236e-01 6.02917433e-01
8.33665609e-01 6.61915421e-01 5.04971743e-01 -8.27271760e-01
8.07109296e-01 -5.78409731e-01 -7.48140037e-01 1.78732231e-01
-8.02499235e-01 -3.21153194e-01 7.69283295e-01 -3.66747707e-01
-7.34721601e-01 4.32013065e-01 2.38673404e-01 -3.24590236e-01
-3.84439379e-02 6.63246214e-01 -1.86507657e-01 -5.72886914e-02
2.33852625e-01 4.32612777e-01 -7.14507774e-02 -5.45155764e-01
4.20034409e-01 9.14148033e-01 2.80984640e-01 -6.15058482e-01
1.09427845e+00 6.52447224e-01 -2.69600917e-02 -6.41396642e-01
-3.99602890e-01 -9.81570065e-01 -6.59310400e-01 -8.04004967e-02
5.28794587e-01 -9.40471172e-01 -7.89348900e-01 3.81380469e-01
-4.90375161e-01 -2.46998575e-02 2.88079232e-01 5.78923106e-01
-3.48649502e-01 7.59454548e-01 -3.69575292e-01 -7.49476671e-01
-3.68082017e-01 -9.67489421e-01 6.26728356e-01 4.44933176e-01
4.73079365e-03 -6.84299290e-01 1.72467589e-01 3.89905840e-01
1.62817582e-01 2.61043757e-01 5.32340229e-01 -5.76153338e-01
-6.67487144e-01 -9.77389216e-02 -5.15913308e-01 -3.56735617e-01
3.36781979e-01 4.71253209e-02 -2.04463840e-01 -6.42683268e-01
-2.66112983e-01 -9.28390995e-02 4.99048710e-01 2.28220746e-01
1.04738832e+00 -2.65283942e-01 -6.95126235e-01 6.59911931e-01
1.69806385e+00 4.06484753e-01 3.62535328e-01 2.96071798e-01
7.18146026e-01 3.50148290e-01 9.07836616e-01 6.28887355e-01
3.95774454e-01 4.34353203e-01 -9.61489305e-02 -6.96856454e-02
1.71782926e-01 -2.76613027e-01 -1.77540537e-02 1.27779830e+00
8.70757475e-02 2.39858050e-02 -1.14453161e+00 7.80466318e-01
-2.23595929e+00 -1.05117750e+00 -4.59603339e-01 2.06678891e+00
8.23684037e-01 -1.94834515e-01 4.99278456e-01 3.42933565e-01
1.13026333e+00 4.57329415e-02 -3.90923023e-01 5.96194863e-02
1.39294058e-01 -2.58186430e-01 5.10922611e-01 4.41082805e-01
-9.10767257e-01 9.45606649e-01 5.00658035e+00 1.42059290e+00
-8.96665275e-01 1.43359944e-01 4.62342173e-01 -1.74713373e-01
3.28382780e-03 3.29297811e-01 -5.07319808e-01 6.21703386e-01
7.57358253e-01 -5.08720577e-01 5.73322713e-01 9.28377807e-01
4.86363322e-01 -3.60985488e-01 -4.74408001e-01 1.31907308e+00
-1.74248144e-01 -1.11019194e+00 -2.51598597e-01 1.04732722e-01
7.50279665e-01 -1.53008059e-01 -3.53178412e-01 -2.83438206e-01
3.52180123e-01 -5.55707872e-01 3.12603384e-01 5.17317057e-02
6.01196349e-01 -1.07981908e+00 5.55949032e-01 3.96198481e-01
-1.89104855e+00 -3.94606851e-02 -5.67421317e-01 1.44307926e-01
2.06509829e-01 8.58145654e-01 -4.04007077e-01 7.08296597e-01
8.76646340e-01 5.14876187e-01 -4.12049055e-01 1.38174546e+00
-1.02788582e-02 6.76030159e-01 -5.40136456e-01 -2.45780870e-01
3.15096468e-01 -8.62999499e-01 3.62789661e-01 1.15835130e+00
4.47480321e-01 5.70035934e-01 7.44525433e-01 5.06304860e-01
5.74978851e-02 6.88807547e-01 -3.95341128e-01 -1.00143172e-01
1.02231741e+00 1.58916843e+00 -1.41885030e+00 -3.18974555e-01
-4.85666275e-01 6.53011918e-01 2.86170870e-01 2.01814964e-01
-8.44406784e-01 -8.92229021e-01 1.62109807e-01 3.66976529e-01
4.09068823e-01 -3.73730659e-01 -1.39905676e-01 -1.01538754e+00
-3.61494683e-02 -9.07894671e-01 7.21324205e-01 -6.05070293e-01
-1.13889539e+00 3.83838177e-01 1.04606867e-01 -1.22770369e+00
4.46899414e-01 3.33266594e-02 -6.13934934e-01 5.29009342e-01
-9.76189852e-01 -1.10996878e+00 -4.88695860e-01 9.69742298e-01
1.74358010e-01 -1.05060041e-02 3.67055416e-01 4.80970263e-01
-6.81681454e-01 3.05835456e-01 6.88432872e-01 2.05821142e-01
7.25934029e-01 -9.88396466e-01 -1.96297526e-01 1.03928959e+00
-6.58721998e-02 9.50660408e-01 5.45289278e-01 -7.78634489e-01
-1.45757520e+00 -8.18238735e-01 4.50195074e-01 3.11427832e-01
8.18698943e-01 -2.50340462e-01 -8.47642243e-01 3.43064308e-01
3.02168697e-01 -1.05745852e-01 9.35048819e-01 5.83746321e-02
-1.57197490e-01 -5.21250308e-01 -1.09138525e+00 4.81155396e-01
1.02083635e+00 -2.91177601e-01 -2.67649978e-01 3.79234910e-01
2.93366700e-01 -1.28894880e-01 -8.57158184e-01 2.76548654e-01
2.76023299e-01 -9.94708776e-01 8.47178221e-01 1.62908420e-01
-1.23729803e-01 -9.97741878e-01 -5.88788204e-02 -1.12406218e+00
-5.25249004e-01 -7.31814563e-01 3.39710802e-01 1.39025080e+00
7.09380805e-02 -7.58337438e-01 8.24077785e-01 2.41352603e-01
3.37041646e-01 -8.67075861e-01 -6.09324634e-01 -8.66736591e-01
-2.34632656e-01 -2.04805285e-02 6.12383783e-01 1.12202990e+00
2.79834300e-01 3.96911532e-01 -2.08766416e-01 3.11668783e-01
7.78036296e-01 7.58426130e-01 8.23307812e-01 -1.04774082e+00
-1.28258273e-01 -3.36257964e-01 -3.73401523e-01 -8.06443512e-01
-1.08910330e-01 -9.45628583e-01 -8.58639106e-02 -1.76868904e+00
8.33011150e-01 -6.73908234e-01 -3.30120832e-01 4.45215374e-01
-4.00574803e-01 1.93297312e-01 1.59403026e-01 6.27030492e-01
-1.04677725e+00 4.19965655e-01 8.44939888e-01 1.41832516e-01
-2.77028173e-01 -2.60260165e-01 -7.20270216e-01 5.13959527e-01
9.48943913e-01 -7.03741610e-01 -7.52208889e-01 -4.96629365e-02
-1.48712834e-02 2.31228828e-01 -1.94897637e-01 -1.06693065e+00
6.89636409e-01 -4.56482142e-01 5.78674078e-02 -6.85162604e-01
-4.46364880e-02 -1.07622516e+00 6.57601357e-01 6.44642055e-01
7.74964765e-02 2.01826170e-01 -5.35924248e-02 7.65735388e-01
-3.35780233e-01 1.21454597e-01 1.01822329e+00 -5.21498360e-02
-5.91637850e-01 5.81097662e-01 -1.12392664e-01 6.99464604e-02
1.49076653e+00 -2.60321409e-01 -1.46903276e-01 -3.58288765e-01
-5.10265291e-01 6.20309949e-01 6.41382635e-01 1.76675558e-01
5.77668846e-01 -1.57141876e+00 -5.12548447e-01 -1.68928623e-01
-1.90744717e-02 -4.55199219e-02 3.57078701e-01 1.08408058e+00
-7.57924378e-01 3.78132433e-01 -1.09478079e-01 -4.76847738e-01
-1.67882109e+00 8.56324911e-01 5.86202294e-02 -1.07518561e-01
-4.42177266e-01 5.29207051e-01 -1.30172940e-02 -4.97521728e-01
-2.32938435e-02 3.24125022e-01 1.94073960e-01 3.83506604e-02
3.38654011e-01 6.91045105e-01 -4.06807095e-01 -7.89941788e-01
-6.32203162e-01 9.34040725e-01 2.55586267e-01 -6.47279024e-02
1.39958465e+00 -6.38869882e-01 -5.49269080e-01 1.73502669e-01
1.14322662e+00 3.04005027e-01 -6.76925838e-01 -3.33048195e-01
1.45503459e-02 -8.69179904e-01 -4.43441570e-02 -2.24442124e-01
-1.31122565e+00 6.11349046e-01 4.40635055e-01 4.25938368e-01
1.39428318e+00 -1.38704032e-01 1.06850111e+00 1.88078150e-01
4.49677855e-01 -1.25498295e+00 -1.11898385e-01 -9.53886509e-02
2.65977234e-01 -1.03042686e+00 3.28566909e-01 -7.49607086e-01
-5.09379804e-01 7.71621883e-01 7.01866925e-01 -3.00577283e-01
8.51457298e-01 1.06742539e-01 4.26143147e-02 -2.05575347e-01
-3.55580807e-01 -3.10324013e-01 -1.26453131e-01 2.55398303e-01
3.32516283e-01 1.94357589e-01 -8.41813028e-01 5.97670019e-01
3.98240574e-02 -1.44030258e-01 4.36753839e-01 9.81408119e-01
-7.24167347e-01 -1.20864475e+00 -5.74173987e-01 3.53243828e-01
-3.48712593e-01 3.05169132e-02 -3.70383769e-01 5.26724517e-01
-4.30668108e-02 1.18780041e+00 -2.61492133e-01 -5.15247107e-01
9.42397416e-02 -2.25689963e-01 -2.68019177e-03 -3.66939992e-01
-2.17758119e-01 4.64836359e-01 -2.67503887e-01 -6.37346804e-01
-6.98489845e-01 -5.37444174e-01 -1.66617262e+00 -6.78270519e-01
-4.15376663e-01 7.88180888e-01 4.44432259e-01 4.21235561e-01
5.53765893e-01 1.32904842e-01 8.99198353e-01 -4.63176012e-01
-2.09727332e-01 -6.07881367e-01 -8.97506177e-01 4.69510645e-01
-2.22920731e-01 -4.51140374e-01 -3.18196446e-01 -3.91299315e-02]
|
[7.5129876136779785, 4.771225452423096]
|
78a38dd2-ba8b-41ff-b1bb-108680a574a8
|
bidirectional-learning-for-domain-adaptation
|
1904.1062
| null |
http://arxiv.org/abs/1904.10620v1
|
http://arxiv.org/pdf/1904.10620v1.pdf
|
Bidirectional Learning for Domain Adaptation of Semantic Segmentation
|
Domain adaptation for semantic image segmentation is very necessary since
manually labeling large datasets with pixel-level labels is expensive and time
consuming. Existing domain adaptation techniques either work on limited
datasets, or yield not so good performance compared with supervised learning.
In this paper, we propose a novel bidirectional learning framework for domain
adaptation of segmentation. Using the bidirectional learning, the image
translation model and the segmentation adaptation model can be learned
alternatively and promote to each other. Furthermore, we propose a
self-supervised learning algorithm to learn a better segmentation adaptation
model and in return improve the image translation model. Experiments show that
our method is superior to the state-of-the-art methods in domain adaptation of
segmentation with a big margin. The source code is available at
https://github.com/liyunsheng13/BDL.
|
['Nuno Vasconcelos', 'Yunsheng Li', 'Lu Yuan']
|
2019-04-24
|
bidirectional-learning-for-domain-adaptation-1
|
http://openaccess.thecvf.com/content_CVPR_2019/html/Li_Bidirectional_Learning_for_Domain_Adaptation_of_Semantic_Segmentation_CVPR_2019_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2019/papers/Li_Bidirectional_Learning_for_Domain_Adaptation_of_Semantic_Segmentation_CVPR_2019_paper.pdf
|
cvpr-2019-6
|
['synthetic-to-real-translation']
|
['computer-vision']
|
[ 3.24737638e-01 3.70906144e-02 -7.53106833e-01 -6.01025224e-01
-8.83319676e-01 -6.65960729e-01 2.78984696e-01 -1.76186919e-01
-4.66810316e-01 8.15624237e-01 -3.71096060e-02 -3.40182453e-01
2.81104803e-01 -6.94828272e-01 -8.49377871e-01 -6.42976761e-01
6.45009220e-01 7.35217690e-01 6.13237143e-01 1.32994264e-01
1.42087772e-01 6.48332909e-02 -9.72287118e-01 1.48543596e-01
1.10207880e+00 8.54868650e-01 4.56653297e-01 6.10038936e-01
-5.05270660e-01 4.51793015e-01 -2.74570525e-01 -2.35801786e-01
2.51815736e-01 -7.00425744e-01 -1.19123030e+00 2.33892933e-01
3.61020923e-01 -2.59288579e-01 -8.19133420e-04 1.14329123e+00
4.05232549e-01 -7.19143227e-02 6.57720804e-01 -1.15066707e+00
-8.12211633e-01 5.86618960e-01 -6.49490833e-01 1.02983609e-01
-4.15024459e-02 -6.48788214e-02 6.63543642e-01 -9.01223004e-01
7.10802019e-01 9.09552872e-01 4.94841784e-01 8.64472508e-01
-1.33009982e+00 -8.26186061e-01 3.73759329e-01 3.76546174e-01
-1.25451219e+00 -1.73971444e-01 9.67148304e-01 -3.90535653e-01
4.43655401e-01 -1.40699461e-01 5.68701029e-01 1.06738091e+00
-2.89671928e-01 1.06517386e+00 1.26873839e+00 -6.34696722e-01
1.48875877e-01 3.89534771e-01 -6.01057187e-02 6.08516395e-01
2.86927111e-02 -4.38661501e-02 -2.64877498e-01 2.14063078e-01
8.82620096e-01 -1.74096033e-01 -1.01961538e-01 -8.22977006e-01
-1.06968164e+00 7.88901865e-01 4.57736433e-01 2.98797339e-01
-8.07117671e-02 3.24338749e-02 3.53220582e-01 1.86795712e-01
6.13282323e-01 1.49742946e-01 -8.39981794e-01 -1.47537775e-02
-7.70235360e-01 -1.18224816e-02 5.09188414e-01 1.15939271e+00
1.07695615e+00 -2.22441897e-01 1.59367666e-01 1.10138428e+00
2.54602730e-01 4.54463214e-01 5.91055334e-01 -1.09528005e+00
3.40463102e-01 5.95221817e-01 -4.50534336e-02 -4.91083652e-01
-4.40124512e-01 -2.05777198e-01 -6.02309346e-01 1.04529627e-01
7.77295589e-01 -2.14196786e-01 -1.05392754e+00 1.64146018e+00
5.16061962e-01 3.15013230e-01 9.08622742e-02 8.45780611e-01
7.59988427e-01 6.10518038e-01 3.32421392e-01 1.09502405e-01
1.04360437e+00 -1.32762647e+00 -6.90066516e-01 -5.19504547e-01
6.60063803e-01 -1.04226911e+00 1.42632675e+00 1.46369353e-01
-1.03482354e+00 -6.10720396e-01 -7.96256363e-01 -2.27641553e-01
-3.29622179e-01 3.65579218e-01 3.94307762e-01 4.17512178e-01
-7.05026925e-01 2.84783304e-01 -9.44788933e-01 -5.77380359e-01
6.85737908e-01 3.13625723e-01 -1.22746326e-01 -1.12400740e-01
-9.91055489e-01 8.29119861e-01 6.84431136e-01 -2.55946428e-01
-6.07507229e-01 -6.03755355e-01 -7.44829595e-01 -4.08996046e-01
5.22495747e-01 -7.10919321e-01 1.51763237e+00 -1.40286005e+00
-1.74685729e+00 1.05406392e+00 -2.49474034e-01 -3.23541433e-01
6.58759594e-01 -1.55828640e-01 -2.02282712e-01 1.79738194e-01
2.66916633e-01 1.09936190e+00 7.35071123e-01 -1.37912488e+00
-8.72604668e-01 -4.08628732e-01 -1.25510097e-01 2.75274992e-01
-5.40454388e-01 -2.51488060e-01 -8.28629911e-01 -6.21874571e-01
2.08647221e-01 -1.07493901e+00 -2.94681698e-01 1.04190484e-01
-2.29156882e-01 -1.97179377e-01 9.26495194e-01 -7.35035121e-01
1.01169467e+00 -2.05786848e+00 1.48197234e-01 3.65266167e-02
-2.86877126e-01 3.88502479e-01 -1.14437014e-01 -4.33436409e-02
7.46567175e-02 9.56729501e-02 -7.03665614e-01 -1.53121486e-01
-2.45982230e-01 4.29353625e-01 8.74596444e-05 3.58762741e-01
2.08430402e-02 9.98529553e-01 -9.99953210e-01 -8.33152711e-01
3.09442997e-01 3.67022038e-01 -4.48903263e-01 2.52915472e-01
-3.78346235e-01 9.54601705e-01 -5.48628747e-01 5.10843039e-01
6.86747730e-01 -3.72867137e-01 1.70681685e-01 -1.34341940e-01
1.19220391e-01 1.00776613e-01 -9.63182390e-01 2.17056990e+00
-5.37886500e-01 4.65545326e-01 -1.08047739e-01 -1.22895157e+00
9.47558403e-01 2.48348057e-01 4.69300330e-01 -7.04633772e-01
1.93920523e-01 3.64656448e-01 -3.01409364e-01 -5.33013463e-01
8.15433636e-02 -1.77581802e-01 -1.51870307e-02 2.67941296e-01
2.43305862e-01 -3.11381847e-01 1.49826899e-01 -6.77736327e-02
4.30514127e-01 7.18433022e-01 2.78877854e-01 -9.75846648e-02
7.81374335e-01 2.95122147e-01 9.28866386e-01 4.43036050e-01
-3.90534878e-01 7.05322564e-01 2.67630160e-01 -2.58962572e-01
-1.07117856e+00 -1.04335558e+00 -1.16536170e-01 1.18883860e+00
3.81661624e-01 1.93022825e-02 -1.07554972e+00 -1.06357062e+00
-1.73230454e-01 7.22387612e-01 -4.22686964e-01 -1.90625697e-01
-6.82444572e-01 -4.69863623e-01 3.08588535e-01 9.24455822e-01
8.91414225e-01 -9.60549712e-01 -1.66671589e-01 -5.22946529e-02
-4.36689824e-01 -1.32817030e+00 -7.38416851e-01 1.02895424e-02
-1.25456870e+00 -9.91558492e-01 -9.38302517e-01 -1.36793792e+00
9.60811615e-01 9.22381952e-02 1.02223074e+00 -3.02498639e-02
1.56322643e-01 1.90833345e-01 -4.04053748e-01 -4.48610395e-01
-5.33722579e-01 5.20868957e-01 -2.86756724e-01 -1.30105868e-01
6.10273123e-01 -4.08914268e-01 -5.59818566e-01 5.97640634e-01
-8.68857920e-01 1.77774161e-01 6.08439982e-01 9.16788518e-01
9.87482607e-01 -1.86778679e-01 5.93280137e-01 -1.38461888e+00
2.28845999e-01 -1.78873658e-01 -8.90237391e-01 2.15548530e-01
-8.56394231e-01 -6.45103604e-02 5.78992903e-01 -4.38819528e-01
-1.40430605e+00 7.00327814e-01 -1.58556029e-01 -3.74702275e-01
-5.78155696e-01 2.49374658e-01 -3.06598514e-01 -1.41887277e-01
7.04092264e-01 1.14043638e-01 -6.25092983e-02 -5.95231175e-01
6.39712691e-01 6.48054361e-01 5.40077507e-01 -4.96209145e-01
8.17142904e-01 5.26233077e-01 -3.46168935e-01 -4.68699843e-01
-8.28398466e-01 -6.52199209e-01 -1.22796309e+00 -4.47710343e-02
9.00531173e-01 -8.27474833e-01 6.60031512e-02 5.52743852e-01
-9.54402328e-01 -8.74936223e-01 -3.10497314e-01 5.10061681e-01
-7.44695723e-01 3.36850435e-01 -5.03978729e-01 -2.38110349e-01
-9.73144080e-03 -1.04755247e+00 8.61675322e-01 4.35157925e-01
-1.32883236e-01 -1.35839343e+00 2.65282782e-04 6.87799573e-01
1.40902132e-01 -1.18151613e-01 7.21783400e-01 -5.88470161e-01
-4.76984501e-01 4.13890481e-02 -2.94742703e-01 4.78021681e-01
2.69854665e-01 -1.81898385e-01 -7.50330687e-01 -1.28889963e-01
-3.00396651e-01 -2.13545293e-01 8.59268904e-01 6.53104007e-01
1.41945219e+00 -1.54821113e-01 -4.52337563e-01 8.17103207e-01
1.38710892e+00 1.93438247e-01 4.92361188e-01 6.78952694e-01
8.37344944e-01 6.42761350e-01 9.69000161e-01 8.05161670e-02
5.83638012e-01 7.09680319e-01 4.68524434e-02 -4.34971243e-01
-4.40029800e-01 -4.31426227e-01 3.22376825e-02 6.27362490e-01
1.50906593e-01 4.78847027e-02 -1.16684854e+00 7.99970448e-01
-2.08389544e+00 -5.13279080e-01 -1.30132601e-01 2.00273132e+00
1.04944575e+00 5.10607101e-02 3.61553431e-01 -2.26741821e-01
8.03682804e-01 -2.09058613e-01 -6.81419253e-01 -3.05139184e-01
-3.36739495e-02 -3.59268002e-02 8.91687095e-01 6.26006722e-01
-1.29926932e+00 1.55487359e+00 6.27035856e+00 7.51038194e-01
-1.11273372e+00 4.02244419e-01 6.64965272e-01 4.96371165e-02
-2.50775397e-01 4.63678688e-02 -7.07628846e-01 4.68180895e-01
6.86485648e-01 -4.36308281e-03 3.26535612e-01 8.34089100e-01
1.11018553e-01 -1.02362921e-02 -7.73697913e-01 8.58548641e-01
3.75053985e-03 -1.04779005e+00 -1.56452686e-01 -1.36263549e-01
1.15442848e+00 2.31564343e-02 4.54967096e-02 1.19266644e-01
3.63120109e-01 -4.91890728e-01 4.25642908e-01 1.80172607e-01
5.42131007e-01 -6.25568032e-01 5.89425445e-01 3.11676353e-01
-9.58360493e-01 -7.44190589e-02 -3.88254404e-01 3.04107726e-01
2.55803198e-01 4.74614441e-01 -8.61442685e-01 2.27733359e-01
8.71043861e-01 8.72910917e-01 -6.37290835e-01 9.80774939e-01
-7.66591966e-01 9.06172037e-01 -3.24108005e-02 2.00019956e-01
1.68636352e-01 -4.69798744e-01 3.57452571e-01 1.23194718e+00
1.57489836e-01 -5.65036945e-02 2.96699971e-01 7.77814686e-01
-2.28828043e-01 3.38625997e-01 -5.41513503e-01 1.49003819e-01
3.34066421e-01 1.19558108e+00 -9.99867499e-01 -4.53815430e-01
-6.15903318e-01 1.24640131e+00 2.15224847e-01 5.55360377e-01
-9.28597450e-01 -1.63367927e-01 4.10695523e-01 -7.88551103e-03
4.45718586e-01 -2.40432784e-01 -7.17792869e-01 -1.06905127e+00
3.31848003e-02 -6.04654551e-01 6.39232337e-01 -6.61079288e-01
-1.26838636e+00 3.22618842e-01 4.89761904e-02 -1.39228964e+00
-1.96278960e-01 -5.03589928e-01 -3.24970007e-01 5.40267467e-01
-1.90157294e+00 -1.46614671e+00 -3.90231252e-01 6.92620814e-01
6.86600804e-01 -1.49369970e-01 6.90048873e-01 3.41813356e-01
-4.95820910e-01 6.06422067e-01 2.35104203e-01 3.32240254e-01
1.10775721e+00 -1.35066271e+00 2.83061057e-01 8.94423962e-01
9.04036835e-02 1.95520818e-01 4.08280283e-01 -6.50155365e-01
-7.78888702e-01 -1.27615631e+00 7.42289722e-01 -2.94290274e-01
5.54605424e-01 -1.75581396e-01 -1.09522283e+00 7.97152221e-01
2.98999786e-01 7.74791092e-02 8.33864391e-01 3.71496640e-02
-2.37295523e-01 -2.88969427e-01 -1.14483726e+00 5.15917063e-01
1.03911781e+00 -4.30707365e-01 -3.89345139e-01 4.51434016e-01
6.33441031e-01 -5.89995027e-01 -7.95872509e-01 3.30426127e-01
3.65628958e-01 -6.14953220e-01 9.98637497e-01 -4.50902104e-01
3.75827610e-01 -4.78104502e-01 4.37079221e-02 -1.30294883e+00
-2.83927232e-01 -1.23992324e-01 1.42477736e-01 1.44408572e+00
7.43754923e-01 -6.99636519e-01 1.09886110e+00 5.40950298e-01
-2.74990741e-02 -5.47124922e-01 -7.59352982e-01 -9.81892288e-01
5.02715588e-01 -2.29956985e-01 4.48446393e-01 1.03251624e+00
-1.89025253e-01 3.72608364e-01 -2.24238187e-01 1.83036000e-01
5.64368784e-01 3.23343366e-01 7.90158212e-01 -9.33138371e-01
-1.31445050e-01 -4.78552818e-01 -1.94943681e-01 -1.21246612e+00
4.82488990e-01 -9.52177465e-01 1.82159647e-01 -1.67214477e+00
1.95910066e-01 -4.75920707e-01 -4.00483012e-01 7.36072123e-01
-2.74161845e-01 4.41393554e-01 4.27630246e-02 3.54208618e-01
-5.66531777e-01 3.32375884e-01 1.47104144e+00 -2.86550552e-01
-4.01894033e-01 1.83765382e-01 -6.53451085e-01 8.54397714e-01
1.35442674e+00 -5.47155619e-01 -4.69001919e-01 -7.03013361e-01
-2.26600394e-01 -2.50095755e-01 2.13657588e-01 -6.83808863e-01
2.57168651e-01 -1.76866993e-01 3.64172459e-01 -3.62733632e-01
-7.88510195e-04 -9.83872533e-01 -2.28641376e-01 3.12238127e-01
-4.04028744e-01 -3.91587764e-01 2.91770726e-01 4.77167696e-01
-3.38896453e-01 -3.85098398e-01 1.10315430e+00 -2.53808945e-01
-1.10074759e+00 3.25891733e-01 -2.15244025e-01 1.49274588e-01
1.06384730e+00 -2.64025331e-01 -5.48344851e-02 -2.47936413e-01
-7.86363065e-01 3.55759412e-01 6.45766199e-01 5.66953301e-01
3.93386394e-01 -1.31624520e+00 -5.72256744e-01 1.07659111e-02
2.44137883e-01 2.86527425e-01 5.28568625e-02 8.56106341e-01
-4.86355960e-01 2.72927165e-01 -3.50365251e-01 -7.97491133e-01
-1.32054961e+00 5.34715831e-01 3.64490271e-01 -5.08220457e-02
-3.90926808e-01 9.17981267e-01 2.42112502e-01 -9.26884830e-01
9.32380930e-02 -1.66764036e-01 -1.89702377e-01 -1.18522063e-01
1.29855976e-01 2.85733849e-01 -1.29421830e-01 -6.42681837e-01
-2.54483700e-01 1.08638322e+00 -1.82692502e-02 -6.29979670e-02
1.10115898e+00 -3.84327322e-01 2.96656527e-02 4.09186780e-01
1.28718269e+00 -3.93868089e-01 -1.55920005e+00 -5.80959022e-01
1.17407814e-01 -6.31983995e-01 3.77797894e-02 -8.87488067e-01
-1.22213221e+00 8.87633920e-01 8.30612481e-01 -2.65825927e-01
1.50602877e+00 2.14497164e-01 9.69072640e-01 1.37327269e-01
1.92014888e-01 -1.52707040e+00 7.92906731e-02 3.84104282e-01
5.38633525e-01 -1.72129488e+00 -3.86630893e-02 -6.99574530e-01
-8.90669942e-01 1.00751591e+00 8.82020533e-01 6.40719244e-03
6.12422526e-01 3.40758115e-02 6.43235743e-01 2.65806645e-01
-2.11615607e-01 -3.22555393e-01 2.75978327e-01 8.13216507e-01
5.04197121e-01 6.76460490e-02 -4.55092788e-01 2.96315551e-01
4.93893214e-02 2.24110767e-01 2.28137806e-01 7.56202996e-01
-2.88267463e-01 -1.74414241e+00 -2.71971524e-01 2.35165004e-02
-2.98790604e-01 1.82654247e-01 -4.59299386e-01 7.17856288e-01
1.95270628e-01 8.00940990e-01 -6.21505789e-02 -1.67628944e-01
3.35666507e-01 1.73170179e-01 5.24049878e-01 -5.49426854e-01
-1.09493613e-01 3.36927176e-01 -1.30255982e-01 -4.12011504e-01
-6.99343383e-01 -8.20204854e-01 -1.56462049e+00 9.29274261e-02
-2.95666307e-01 -5.41751869e-02 6.82138324e-01 8.73490751e-01
2.69472510e-01 4.51728046e-01 5.00760019e-01 -5.50058424e-01
-1.42150238e-01 -7.95810878e-01 -2.44578376e-01 5.47990322e-01
8.02721754e-02 -5.29423773e-01 3.08928229e-02 5.99113464e-01]
|
[9.639220237731934, 1.3979662656784058]
|
e96b8755-5cb4-4e8c-ab3b-a2431b4a16c7
|
robo3d-towards-robust-and-reliable-3d
|
2303.17597
| null |
https://arxiv.org/abs/2303.17597v3
|
https://arxiv.org/pdf/2303.17597v3.pdf
|
Robo3D: Towards Robust and Reliable 3D Perception against Corruptions
|
The robustness of 3D perception systems under natural corruptions from environments and sensors is pivotal for safety-critical applications. Existing large-scale 3D perception datasets often contain data that are meticulously cleaned. Such configurations, however, cannot reflect the reliability of perception models during the deployment stage. In this work, we present Robo3D, the first comprehensive benchmark heading toward probing the robustness of 3D detectors and segmentors under out-of-distribution scenarios against natural corruptions that occur in real-world environments. Specifically, we consider eight corruption types stemming from adversarial weather conditions, external disturbances, and internal sensor failure. We uncover that, although promising results have been progressively achieved on standard benchmarks, state-of-the-art 3D perception models are at risk of being vulnerable to corruptions. We draw key observations on the use of data representations, augmentation schemes, and training strategies, that could severely affect the model's performance. To pursue better robustness, we propose a density-insensitive training framework along with a simple flexible voxelization strategy to enhance the model resiliency. We hope our benchmark and approach could inspire future research in designing more robust and reliable 3D perception models. Our robustness benchmark suite is publicly available.
|
['Ziwei Liu', 'Kai Chen', 'Liang Pan', 'Jiawei Ren', 'Wenwei Zhang', 'Runnan Chen', 'Xin Li', 'Youquan Liu', 'Lingdong Kong']
|
2023-03-30
| null | null | null | null |
['robust-3d-semantic-segmentation', 'robust-3d-object-detection']
|
['computer-vision', 'computer-vision']
|
[ 2.01519415e-01 -4.55006883e-02 1.90551266e-01 -1.08403504e-01
-6.32008433e-01 -8.85442495e-01 7.46156454e-01 3.54763210e-01
-1.72156006e-01 2.90645480e-01 1.63587928e-01 -4.53900307e-01
-2.08561420e-02 -7.64497459e-01 -1.04824615e+00 -7.14491725e-01
-4.60437119e-01 1.03603482e-01 5.42208374e-01 -3.69716614e-01
2.69943058e-01 8.25348318e-01 -1.91764665e+00 -1.45004049e-01
6.62150145e-01 1.28592670e+00 -1.50051072e-01 7.53899276e-01
5.45136631e-01 3.52488935e-01 -1.08067763e+00 -1.37985855e-01
6.98383868e-01 4.69352424e-01 -2.64540881e-01 6.74870163e-02
4.40005243e-01 -3.73005301e-01 -4.22407657e-01 1.00287747e+00
8.27884436e-01 1.32195815e-01 7.12461412e-01 -1.44888461e+00
-6.27472997e-01 1.50230899e-01 -3.41279596e-01 2.74632096e-01
5.22283792e-01 7.27890730e-01 2.54974902e-01 -5.48857927e-01
3.75760823e-01 1.45525980e+00 7.73897111e-01 4.90513384e-01
-1.22781181e+00 -4.04375583e-01 5.14376104e-01 -6.68626139e-03
-1.22521877e+00 -5.42188346e-01 7.30367541e-01 -5.12256444e-01
1.19149387e+00 2.71507502e-01 1.56243593e-01 1.81973457e+00
5.38836002e-01 2.82907009e-01 1.12610722e+00 6.64225519e-02
6.27445817e-01 -3.53871621e-02 -2.13273689e-02 4.18393612e-01
5.47566295e-01 6.08427465e-01 -5.20549536e-01 -1.92500025e-01
3.27010512e-01 -4.48356777e-01 -1.46981806e-01 -6.95215583e-01
-9.63123262e-01 3.30940753e-01 6.21805787e-01 -3.53370160e-01
-2.63410538e-01 1.69340074e-01 5.77680647e-01 2.57336259e-01
4.67857569e-01 4.13244933e-01 -4.88223284e-01 -1.45599380e-01
-1.93957493e-01 4.65683460e-01 5.72423875e-01 1.01392627e+00
4.51386184e-01 1.83689609e-01 -1.34607792e-01 5.82092047e-01
7.74921998e-02 8.28177571e-01 -1.39948027e-02 -9.64292467e-01
5.54238498e-01 4.64810073e-01 2.99956471e-01 -1.23256266e+00
-6.38222814e-01 -3.17223012e-01 -8.96967649e-01 7.03922927e-01
2.47477457e-01 6.28189743e-02 -1.31578839e+00 1.64665675e+00
4.59400028e-01 1.39734417e-01 3.39343876e-01 9.83172059e-01
6.25627577e-01 1.89332455e-01 4.47870493e-02 2.56783038e-01
9.86545980e-01 -2.31703535e-01 -2.61497617e-01 -4.35640663e-01
1.84491575e-01 -5.57600737e-01 1.05802953e+00 4.97806877e-01
-8.33521724e-01 -6.61518574e-01 -1.24253201e+00 3.06228220e-01
-6.44714713e-01 -5.72066307e-01 4.38764781e-01 9.16964471e-01
-7.59805977e-01 2.88326919e-01 -9.33374882e-01 -3.54825079e-01
4.48171198e-01 8.22830051e-02 -6.13214433e-01 -2.14138329e-01
-1.02318168e+00 1.25640595e+00 1.97198596e-02 1.52918547e-01
-1.70675075e+00 -6.19468510e-01 -9.53976333e-01 -4.19137686e-01
6.05793595e-01 -6.31380975e-01 9.95793998e-01 2.38760374e-02
-9.53029692e-01 7.61060178e-01 2.21378326e-01 -6.87376857e-01
6.10399067e-01 -2.91183978e-01 -3.89470220e-01 -1.59810886e-01
1.14628607e-02 3.21027786e-01 9.16369379e-01 -1.97759688e+00
-2.89771974e-01 -6.16582274e-01 2.94076115e-01 8.11719671e-02
2.01201931e-01 -3.17208707e-01 -3.77425343e-01 -5.10638237e-01
1.90675080e-01 -9.42551494e-01 -5.34873068e-01 -3.95539105e-02
-7.99188733e-01 1.58355013e-01 4.74929690e-01 -2.59832233e-01
5.71400940e-01 -2.24352050e+00 -7.51737319e-03 3.42276663e-01
-3.79291140e-02 2.27220312e-01 -1.72393188e-01 3.61270398e-01
1.81265533e-01 3.66469949e-01 -2.35492036e-01 -5.52813590e-01
3.24755162e-01 4.54026610e-01 -6.07690334e-01 7.49375701e-01
2.78862357e-01 4.09373403e-01 -6.74791992e-01 3.97537798e-02
4.31766659e-01 3.41603249e-01 -5.46734214e-01 3.05923998e-01
-2.03810260e-01 4.16432917e-01 -3.40481758e-01 1.28121650e+00
9.94074404e-01 4.87918198e-01 -3.89146596e-01 -1.46086216e-01
9.46820080e-02 2.68353492e-01 -1.04637849e+00 1.64008915e+00
-4.49432611e-01 2.40602538e-01 2.42174953e-01 -8.66946638e-01
1.13398159e+00 -8.75366330e-02 2.20425323e-01 -1.00469422e+00
2.26082951e-01 -5.36796860e-02 -3.11954349e-01 -4.37430531e-01
7.99604952e-01 1.30575702e-01 -5.54971516e-01 -5.64701594e-02
-3.10991555e-01 -7.07408071e-01 -2.26838410e-01 4.42722626e-02
1.44930363e+00 -4.96960804e-02 -6.85507953e-02 3.51659842e-02
-1.00598134e-01 -1.92879662e-02 6.34366333e-01 1.16493857e+00
-7.10851908e-01 7.51713336e-01 5.63972652e-01 -3.42841417e-01
-8.32600057e-01 -1.61878788e+00 -3.36108714e-01 5.59757710e-01
6.04384780e-01 -8.32463503e-02 -4.56217378e-01 -6.53907895e-01
6.21281266e-01 7.47809589e-01 -7.80472398e-01 -6.49541795e-01
-9.93677527e-02 -8.68238389e-01 1.02989352e+00 5.22029221e-01
3.47201586e-01 -7.29222059e-01 -8.91982794e-01 7.79892579e-02
1.31517738e-01 -1.47016549e+00 2.80722409e-01 4.97071952e-01
-5.31683087e-01 -1.28578401e+00 -9.08379629e-02 -1.54677764e-01
5.83819926e-01 5.51098645e-01 1.11037588e+00 -5.43877743e-02
-1.60983890e-01 7.60477960e-01 -6.09335423e-01 -8.40912282e-01
-5.62669277e-01 -2.29499176e-01 7.96419621e-01 -5.95695019e-01
-3.76163609e-02 -7.08000958e-01 -4.84207332e-01 6.18582845e-01
-1.04928994e+00 -6.28742695e-01 2.89978504e-01 3.59437078e-01
6.62429094e-01 3.24408084e-01 3.58467460e-01 -3.65032971e-01
7.59885907e-01 -7.28048444e-01 -6.50042713e-01 -1.42639399e-01
-5.15560925e-01 -1.85668394e-01 5.53942800e-01 -4.40308720e-01
-6.64797485e-01 -1.70707881e-01 -2.15826392e-01 -5.90823233e-01
-7.03147292e-01 1.71329498e-01 -4.01932746e-01 -1.36016026e-01
1.12242353e+00 1.50070814e-02 -1.78390905e-01 -5.39891005e-01
3.79069328e-01 5.24250805e-01 8.66645813e-01 -8.34751010e-01
1.18601537e+00 6.34031653e-01 9.11623389e-02 -7.98448205e-01
-5.79710007e-01 -1.62632495e-01 -2.36623988e-01 -2.33321965e-01
4.84943032e-01 -1.01678658e+00 -6.84505999e-01 8.69924724e-01
-1.02498686e+00 -4.93167281e-01 -2.71858990e-01 -9.70993116e-02
-6.02214575e-01 3.73145312e-01 -5.56567982e-02 -1.00679636e+00
1.33679777e-01 -1.25687969e+00 1.24350417e+00 5.38322367e-02
3.32511179e-02 -5.63207567e-01 1.55883729e-01 4.40522701e-01
3.08038145e-01 8.43747795e-01 7.49262571e-01 -3.42655957e-01
-5.70399165e-01 -3.48705649e-01 2.69812234e-02 6.15982592e-01
8.75880942e-02 -9.99135431e-03 -1.24098003e+00 -4.26880985e-01
-1.53612465e-01 -5.55966794e-01 9.36654806e-01 8.02266821e-02
1.36793685e+00 -9.00901034e-02 -2.19308525e-01 6.99373424e-01
1.16546094e+00 -1.36582321e-02 6.69995129e-01 5.21527827e-01
5.77740848e-01 5.54035008e-01 7.46634007e-01 5.74056029e-01
5.37694335e-01 5.52021980e-01 1.35713911e+00 -5.71325310e-02
-5.79950586e-02 -3.90881062e-01 4.47072119e-01 1.49432063e-01
5.64695820e-02 -5.91099381e-01 -1.05134916e+00 6.49197459e-01
-1.47745991e+00 -7.30602026e-01 2.80995071e-01 2.30204940e+00
3.10597450e-01 7.64872789e-01 -1.32708810e-02 4.96748149e-01
2.00642541e-01 3.94202769e-01 -9.26624835e-01 -5.59370577e-01
-3.97645473e-01 -1.10112138e-01 8.76332521e-01 3.19175959e-01
-1.22427750e+00 7.22493947e-01 6.89756823e+00 2.72968709e-01
-8.88857186e-01 -1.86931744e-01 4.95620996e-01 -1.83615535e-01
-2.83844918e-01 -2.16053888e-01 -7.04021633e-01 4.51695263e-01
8.23492825e-01 2.83290148e-01 2.41495520e-01 1.02554250e+00
3.47140133e-01 -3.74430686e-01 -9.17468488e-01 6.13869131e-01
6.25434816e-02 -9.54055130e-01 -1.68783695e-01 1.54034823e-01
5.07392824e-01 3.59197259e-01 3.97814810e-01 3.05659950e-01
6.07883394e-01 -1.03224123e+00 1.02036190e+00 5.26549220e-01
6.05245471e-01 -7.58674324e-01 7.34906852e-01 3.72280061e-01
-8.44517350e-01 -3.15479279e-01 -5.27918398e-01 -1.36247009e-01
2.22561166e-01 7.22311497e-01 -8.80501509e-01 5.96608460e-01
1.18159378e+00 3.87786567e-01 -8.77994776e-01 1.02703118e+00
-1.91649869e-01 4.57149833e-01 -4.78009045e-01 1.56605259e-01
9.00524110e-02 5.36320448e-01 9.78148103e-01 9.78158474e-01
1.44199461e-01 -2.04034120e-01 2.75400341e-01 6.37162745e-01
1.68993071e-01 -6.42579019e-01 -1.18097675e+00 5.18505812e-01
7.09360242e-01 6.37076735e-01 -4.60461974e-01 1.79683000e-01
-1.77816957e-01 7.03504324e-01 2.30441809e-01 4.86741036e-01
-9.19733405e-01 1.16188683e-01 1.53716421e+00 2.11833864e-01
2.93980092e-01 -7.15477765e-01 -5.92819750e-01 -8.31300735e-01
2.23528937e-01 -9.76072252e-01 1.07926466e-01 -7.60942578e-01
-1.53269470e+00 4.81093287e-01 2.88803101e-01 -1.19772947e+00
1.10684700e-01 -6.43923104e-01 -4.57756042e-01 4.84980375e-01
-1.58126521e+00 -1.04853046e+00 -4.93837148e-01 5.90939462e-01
2.00183049e-01 -7.86013976e-02 7.80525148e-01 -1.60403047e-02
-6.21468365e-01 5.43271124e-01 -2.74764478e-01 -2.88430542e-01
5.98510146e-01 -1.16121924e+00 9.63198006e-01 1.21985877e+00
-2.30584532e-01 3.77914011e-01 1.15112662e+00 -6.81905806e-01
-1.56719780e+00 -1.23859429e+00 2.00973079e-02 -8.92821789e-01
5.76013029e-01 -6.48371935e-01 -7.97183931e-01 4.16929543e-01
-2.74562716e-01 7.94093609e-02 3.15830439e-01 1.83096513e-01
-6.65437639e-01 -2.70797133e-01 -1.42595458e+00 7.08958328e-01
1.37484038e+00 -4.58330035e-01 -5.80376685e-01 1.44630790e-01
1.01857007e+00 -7.55669653e-01 -8.15515935e-01 8.64698112e-01
3.08951646e-01 -1.26143825e+00 1.46286762e+00 -4.62474883e-01
3.46399657e-02 -5.29413462e-01 -7.82216191e-01 -1.44445181e+00
-1.53053164e-01 -3.86530936e-01 -2.42417842e-01 1.13870871e+00
2.28664309e-01 -7.64542162e-01 5.49108267e-01 6.27211928e-01
-5.82671046e-01 -4.81977761e-01 -1.41250217e+00 -1.17405272e+00
9.52039957e-02 -1.04418647e+00 7.76433706e-01 4.60888118e-01
-4.21451062e-01 -3.37299794e-01 -3.51978421e-01 9.03985262e-01
7.94055641e-01 -3.60930681e-01 1.25638652e+00 -9.61227953e-01
3.86281051e-02 -2.52834916e-01 -7.42184103e-01 -8.38339210e-01
2.23863963e-02 -2.61134148e-01 4.32437539e-01 -1.35969341e+00
-3.97900969e-01 -8.17202628e-01 -4.30037051e-01 5.75159431e-01
-9.80708301e-02 2.37200707e-01 1.17148794e-01 -7.31297061e-02
-4.48261946e-01 7.50798941e-01 1.02150631e+00 -4.00647938e-01
-1.22198388e-01 1.90540209e-01 -8.44153225e-01 5.85680425e-01
9.53175426e-01 -4.55973297e-01 -5.40766060e-01 -6.17469966e-01
1.14632495e-01 -4.36363280e-01 8.82829547e-01 -1.32080662e+00
6.43785112e-03 -2.85140723e-01 2.21758574e-01 -5.94939470e-01
6.65052950e-01 -9.13751483e-01 -5.60527444e-02 2.77717322e-01
1.58123404e-01 2.54143596e-01 7.03129351e-01 8.29440236e-01
7.90026560e-02 3.41403812e-01 6.35845065e-01 -1.16479889e-01
-8.77644300e-01 2.55556256e-01 -5.13992906e-01 1.55159021e-02
1.10258949e+00 -3.22803587e-01 -7.76701033e-01 -1.52721509e-01
-3.25058728e-01 2.02702478e-01 1.01669359e+00 8.78093660e-01
7.87722170e-01 -1.09658158e+00 -4.55670774e-01 4.33576345e-01
4.80221927e-01 3.84811282e-01 3.29653323e-01 2.91614860e-01
-2.39726141e-01 1.43561801e-02 -2.70108879e-01 -6.95067704e-01
-9.41363752e-01 7.73358226e-01 3.60172480e-01 1.65551543e-01
-4.55569208e-01 8.32073033e-01 -1.72788486e-01 -6.08317912e-01
6.70642078e-01 -7.33253777e-01 2.54283667e-01 -2.29500160e-01
2.25670904e-01 3.49122792e-01 3.35122198e-01 -6.16119325e-01
-7.97040522e-01 4.03176516e-01 3.20888668e-01 6.81037158e-02
1.05632317e+00 -1.73677072e-01 5.75176954e-01 2.86081433e-01
6.21317863e-01 -2.11458746e-02 -1.61906540e+00 1.15710296e-01
-4.13700938e-01 -6.01536632e-01 -4.76562493e-02 -9.87053931e-01
-7.59379685e-01 6.43040240e-01 7.44051874e-01 4.82625425e-01
1.21140897e+00 1.37940021e-02 5.22800505e-01 3.06760758e-01
6.77689552e-01 -9.77666020e-01 7.27044344e-02 5.53311825e-01
1.09843123e+00 -1.41635668e+00 2.06592847e-02 -2.27606282e-01
-5.72316945e-01 2.97748744e-01 7.45095253e-01 -1.21410795e-01
6.79162860e-01 4.95873600e-01 3.28336358e-01 -1.30946934e-01
-8.20634902e-01 -2.37225711e-01 -3.62992706e-03 1.44341922e+00
-4.92984504e-01 2.08029956e-01 6.38156354e-01 5.15247405e-01
-3.21760327e-01 -7.35737264e-01 5.30414879e-01 9.70718861e-01
-3.81521910e-01 -7.61938691e-01 -7.49689221e-01 3.07079330e-02
-1.02459993e-02 2.57877201e-01 -5.55199623e-01 8.37842524e-01
4.28189397e-01 1.32348406e+00 -4.85972613e-02 -8.94993663e-01
1.07025599e+00 -4.27646697e-01 3.71227264e-01 -4.43675727e-01
-3.69670689e-01 -5.22859514e-01 2.86855847e-01 -1.01081216e+00
-1.50702611e-01 -6.83578253e-01 -7.73179471e-01 -4.90269601e-01
1.91174988e-02 -5.40959120e-01 8.92347038e-01 6.51889086e-01
5.64420998e-01 6.94811106e-01 6.21225893e-01 -1.20006847e+00
-7.61617541e-01 -7.61805892e-01 -3.34113151e-01 3.97994071e-01
6.98106349e-01 -1.19782054e+00 -6.55843139e-01 -3.14368159e-01]
|
[7.953094005584717, -1.6790202856063843]
|
e8f27815-1bcd-41b8-abb0-60ea869d6e6b
|
semantic-models-for-the-first-stage-retrieval
|
2103.04831
| null |
https://arxiv.org/abs/2103.04831v4
|
https://arxiv.org/pdf/2103.04831v4.pdf
|
Semantic Models for the First-stage Retrieval: A Comprehensive Review
|
Multi-stage ranking pipelines have been a practical solution in modern search systems, where the first-stage retrieval is to return a subset of candidate documents, and latter stages attempt to re-rank those candidates. Unlike re-ranking stages going through quick technique shifts during past decades, the first-stage retrieval has long been dominated by classical term-based models. Unfortunately, these models suffer from the vocabulary mismatch problem, which may block re-ranking stages from relevant documents at the very beginning. Therefore, it has been a long-term desire to build semantic models for the first-stage retrieval that can achieve high recall efficiently. Recently, we have witnessed an explosive growth of research interests on the first-stage semantic retrieval models. We believe it is the right time to survey current status, learn from existing methods, and gain some insights for future development. In this paper, we describe the current landscape of the first-stage retrieval models under a unified framework to clarify the connection between classical term-based retrieval methods, early semantic retrieval methods and neural semantic retrieval methods. Moreover, we identify some open challenges and envision some future directions, with the hope of inspiring more researches on these important yet less investigated topics.
|
['Yixing Fan', 'Yinqiong Cai', 'Jiafeng Guo', 'Xueqi Cheng', 'Ruqing Zhang', 'Fei Sun']
|
2021-03-08
| null | null | null | null |
['semantic-retrieval']
|
['natural-language-processing']
|
[ 1.54343814e-01 -4.07143921e-01 -4.17672902e-01 -3.23813945e-01
-1.08722329e+00 -5.06887257e-01 9.40182149e-01 4.71518397e-01
-7.20816195e-01 3.14979076e-01 2.63175279e-01 7.34406263e-02
-7.00834453e-01 -7.84379065e-01 -5.75479716e-02 -1.99943542e-01
1.08061478e-01 9.92277205e-01 5.24786294e-01 -6.68067873e-01
9.05995727e-01 2.61184663e-01 -1.88156819e+00 5.54748118e-01
6.42129838e-01 1.00237560e+00 3.38719279e-01 2.88321018e-01
-5.35131633e-01 4.22491908e-01 -5.92390120e-01 -4.77894843e-01
-1.72294468e-01 -2.09376961e-01 -1.21956837e+00 -5.61149001e-01
2.33847290e-01 -1.75914049e-01 -4.53011632e-01 1.03817415e+00
5.59417963e-01 4.97299999e-01 6.16134107e-01 -9.17773902e-01
-9.83452618e-01 4.93832648e-01 -1.01214238e-01 2.43988350e-01
4.29748088e-01 -6.89351261e-01 1.50999749e+00 -1.13034713e+00
7.61065722e-01 1.28450382e+00 2.87290961e-01 6.20694637e-01
-4.98336494e-01 -4.49064165e-01 2.78720051e-01 6.91798866e-01
-1.47984648e+00 -2.08883047e-01 6.71416759e-01 -1.43750325e-01
1.05600703e+00 3.55681926e-01 5.58735192e-01 8.37116539e-01
-1.12300374e-01 1.04943240e+00 6.89170361e-01 -5.64925969e-01
4.90871109e-02 1.41641609e-02 7.01666772e-01 4.14431602e-01
1.90470517e-01 -5.61886877e-02 -6.61866486e-01 -2.77501553e-01
4.13314611e-01 3.17082286e-01 -1.59167111e-01 -2.30333269e-01
-9.76644814e-01 8.80021453e-01 5.04753530e-01 7.40186572e-01
-1.88048750e-01 -1.00197151e-01 4.56074208e-01 5.29378116e-01
4.57606047e-01 8.70416522e-01 -2.66569942e-01 1.55239969e-01
-1.35082769e+00 4.09356982e-01 6.04767621e-01 8.36822987e-01
7.17478156e-01 -7.73278594e-01 -3.75754625e-01 1.17731202e+00
5.80726266e-01 2.09426105e-01 8.09428096e-01 -5.09906590e-01
1.14539377e-01 6.50276721e-01 1.37208447e-01 -1.09902382e+00
-2.61035681e-01 -6.20674253e-01 -5.24517357e-01 -6.07779682e-01
-3.21294293e-02 6.35884523e-01 -1.01924944e+00 1.25716341e+00
-2.34936193e-01 -1.25497043e-01 6.36961907e-02 1.14452326e+00
1.01985240e+00 7.70229399e-01 2.26484537e-01 -1.13864772e-01
1.41122103e+00 -1.29318213e+00 -6.32819772e-01 -3.54433447e-01
4.83466923e-01 -1.11645734e+00 8.95359993e-01 1.63257167e-01
-9.36262608e-01 -5.50749362e-01 -9.23420250e-01 -3.39045107e-01
-7.38021791e-01 3.39289568e-02 8.41413915e-01 4.50254023e-01
-1.28393614e+00 7.22049773e-01 -4.77770656e-01 -7.92373061e-01
1.69177018e-02 2.25646973e-01 -1.57318953e-02 -4.11902964e-01
-1.93921900e+00 1.00873148e+00 2.77111679e-01 2.95495391e-01
-7.69022107e-01 -4.13087159e-01 -4.12981808e-01 3.09952080e-01
4.48669136e-01 -9.54323828e-01 1.13874090e+00 -5.17705619e-01
-1.19085038e+00 1.01509845e+00 -4.24500585e-01 -1.44224083e-02
7.44057074e-03 -4.45308775e-01 -4.72255439e-01 1.60121948e-01
1.20583609e-01 5.52405000e-01 6.40638053e-01 -1.09976017e+00
-7.97792614e-01 -3.76580358e-01 1.98378652e-01 7.58097291e-01
-6.44209504e-01 4.55717206e-01 -1.01212084e+00 -4.57086563e-01
2.76843160e-01 -7.51589894e-01 -1.14668235e-01 -3.07868570e-01
1.73608869e-01 -7.40883410e-01 2.68902391e-01 -2.38840252e-01
1.48939848e+00 -1.94187152e+00 1.11957051e-01 1.46349773e-01
9.32577923e-02 4.17969525e-01 -3.00323188e-01 8.42261493e-01
2.17082843e-01 2.19461769e-01 2.39890039e-01 -2.93889582e-01
4.17035483e-02 -2.15574726e-01 -7.57634759e-01 -1.77259669e-01
-1.28489852e-01 1.26819026e+00 -1.28346455e+00 -4.97312963e-01
-3.97700556e-02 4.26572621e-01 -2.78690457e-01 2.11864784e-01
-6.45296648e-02 -1.98598728e-01 -9.03032839e-01 8.73664975e-01
1.45431936e-01 -4.54074055e-01 -6.08807318e-02 -7.05944896e-02
8.43214542e-02 5.84307671e-01 -7.26136148e-01 1.95453632e+00
-3.27107280e-01 5.52877963e-01 -2.63548613e-01 -1.11918211e+00
7.28702128e-01 2.61002898e-01 5.78441262e-01 -1.07041562e+00
-2.71050353e-03 6.96036279e-01 -3.76196414e-01 -2.25188106e-01
1.16140544e+00 -2.15997145e-01 -2.08255276e-01 4.85581934e-01
-1.72946349e-01 -1.46065533e-01 4.37197953e-01 4.38700944e-01
9.96016741e-01 2.39910521e-02 -4.47801724e-02 -8.68089870e-03
5.82558692e-01 3.76129717e-01 -1.01421393e-01 9.68876600e-01
-2.44721733e-02 8.01775694e-01 -9.95152146e-02 -3.25334907e-01
-5.79887390e-01 -7.92674899e-01 -2.22487431e-02 1.43966615e+00
4.90579963e-01 -6.24449193e-01 -3.11838627e-01 -5.14984429e-01
-1.31856844e-01 2.14809880e-01 -3.92139494e-01 -4.88083571e-01
-5.10793686e-01 -8.40969920e-01 4.44261760e-01 3.30523014e-01
2.17036068e-01 -1.19605947e+00 -1.70865372e-01 2.93518692e-01
-2.99613953e-01 -6.59953713e-01 -2.48025134e-01 8.77503082e-02
-1.15022504e+00 -9.66490686e-01 -1.30171573e+00 -1.06484413e+00
6.38372004e-01 8.64390790e-01 1.37976837e+00 6.05893612e-01
-6.10897280e-02 4.78051096e-01 -8.07568371e-01 -1.68740600e-01
2.41835505e-01 4.41657335e-01 -1.23345844e-01 -4.10039186e-01
7.61124015e-01 -1.74683314e-02 -9.63981390e-01 3.82978529e-01
-1.22231472e+00 -2.65473068e-01 6.28796339e-01 8.23487520e-01
5.07832646e-01 5.85297719e-02 6.17757976e-01 -6.13244295e-01
1.00803053e+00 -5.43952167e-01 -4.23010737e-01 8.65427136e-01
-1.19295156e+00 8.62147883e-02 2.95953751e-01 -1.40280843e-01
-9.61068690e-01 -7.23457575e-01 -1.18312262e-01 -1.57361478e-01
3.68991584e-01 1.08700061e+00 3.12985152e-01 2.08030775e-01
4.56154883e-01 3.23759675e-01 -4.40798461e-01 -7.76442587e-01
4.45161879e-01 9.08495784e-01 1.94379152e-03 -6.38632476e-01
5.85171580e-01 4.93984222e-01 -3.42162699e-01 -5.15872836e-01
-1.21554220e+00 -1.14528823e+00 -3.96958530e-01 -1.64755926e-01
3.49207014e-01 -8.14429462e-01 -8.85775685e-02 3.68921548e-01
-1.12332702e+00 9.38233584e-02 -1.53571635e-01 2.75131941e-01
-1.42409438e-02 5.89712262e-01 -6.01015031e-01 -4.36803877e-01
-7.63626635e-01 -1.06318617e+00 1.30110872e+00 4.59906220e-01
-1.33538514e-01 -1.14595389e+00 3.10172856e-01 6.11119151e-01
8.01906884e-01 -9.16305244e-01 1.10882890e+00 -1.01254475e+00
-5.85514128e-01 -5.63518941e-01 -3.98759067e-01 8.57027695e-02
-4.63890359e-02 -2.63197750e-01 -7.95024335e-01 -4.08864558e-01
-1.09942265e-01 -3.47188681e-01 1.28554761e+00 1.35262042e-01
1.01148069e+00 2.60758042e-01 -6.50128663e-01 2.39320636e-01
1.29028821e+00 3.70724976e-01 5.73789656e-01 6.99470043e-01
3.68727356e-01 7.10378051e-01 1.04655170e+00 -2.15248480e-01
4.99340236e-01 6.90684378e-01 -9.90427751e-03 1.44070268e-01
-2.46504769e-01 -3.24205041e-01 1.45515636e-01 1.28830636e+00
1.90924238e-02 -5.06741881e-01 -9.10141766e-01 6.24514818e-01
-1.92111325e+00 -7.59145141e-01 2.93945134e-01 2.36355281e+00
7.38128781e-01 -3.74266803e-02 -3.71304542e-01 -1.08672783e-01
8.13256919e-01 2.60064542e-01 -2.94011980e-01 -2.22094227e-02
2.68311426e-02 4.70793664e-01 -8.08240399e-02 4.33558851e-01
-8.61681700e-01 1.39852536e+00 6.60562181e+00 1.09788465e+00
-1.04094326e+00 -8.50803554e-02 3.98046613e-01 -1.59286037e-02
-5.74062288e-01 4.61210668e-01 -1.07758892e+00 1.42064735e-01
8.64181876e-01 -5.01179934e-01 2.95430541e-01 8.75161529e-01
-3.73871088e-01 -1.23951426e-02 -1.01050639e+00 9.25650656e-01
4.59203988e-01 -1.08536839e+00 5.39328754e-01 -2.44702861e-01
5.30828834e-01 1.53330863e-01 2.44573206e-02 8.52814615e-01
-1.88168157e-02 -9.56385732e-01 2.38424584e-01 5.22429347e-01
5.69150209e-01 -4.88192946e-01 8.12431335e-01 1.59351021e-01
-1.27936137e+00 5.89193292e-02 -7.56532013e-01 1.18698463e-01
1.48952261e-01 5.69096506e-01 -3.11747640e-01 8.46720099e-01
5.72040379e-01 8.89912188e-01 -5.66557288e-01 1.36227393e+00
-9.64343920e-02 1.18636943e-01 -6.94837943e-02 -3.51088583e-01
4.65137333e-01 -2.64923573e-02 3.92148614e-01 1.10851574e+00
4.35727149e-01 -1.43675338e-02 1.83558930e-02 3.55928332e-01
-1.54966041e-01 3.97684634e-01 -2.01535299e-01 -3.35986733e-01
4.69212234e-01 1.24136591e+00 -9.03180599e-01 -4.56799835e-01
-3.69298726e-01 1.01921260e+00 3.11140954e-01 4.56733406e-01
-3.82689059e-01 -6.64185345e-01 3.10857177e-01 -1.19909391e-01
-2.42606878e-01 -5.81217073e-02 8.52900147e-02 -1.35788620e+00
8.51366296e-02 -6.06485486e-01 8.23917449e-01 -7.81378031e-01
-1.29777801e+00 6.28398836e-01 9.35431197e-02 -9.67188954e-01
-2.82089204e-01 -4.46867585e-01 -1.55866876e-01 8.12394202e-01
-2.11939359e+00 -8.08862925e-01 -8.67492780e-02 2.70036191e-01
8.93347919e-01 -1.50286883e-01 9.02423859e-01 5.88346779e-01
-1.27083585e-01 3.14060330e-01 5.02504528e-01 -2.72433665e-02
9.51667547e-01 -9.15785253e-01 2.65911549e-01 4.45430130e-01
5.65363050e-01 1.24096453e+00 3.02261502e-01 -6.33872688e-01
-1.33597815e+00 -5.99506378e-01 1.48178387e+00 -5.63370228e-01
6.37697697e-01 3.39636579e-02 -9.71262217e-01 2.33136922e-01
-1.54239051e-02 -3.88318121e-01 6.20723367e-01 4.33108598e-01
-3.45968843e-01 -1.74330577e-01 -4.55289483e-01 6.14540577e-01
9.06387746e-01 -8.18811357e-01 -9.94890273e-01 4.34185773e-01
4.16864425e-01 7.39267319e-02 -5.40805757e-01 5.86149275e-01
8.30904126e-01 -6.11273110e-01 1.29203475e+00 -4.31780428e-01
1.73862249e-01 -1.57363474e-01 2.07725875e-02 -1.16327953e+00
-3.30541223e-01 -3.94662410e-01 -6.71758354e-02 9.54918683e-01
3.39344829e-01 -4.60887730e-01 7.17750669e-01 5.24390936e-01
-3.25702392e-02 -7.67435253e-01 -5.97492635e-01 -7.86365688e-01
2.74275869e-01 -3.06020588e-01 3.70630533e-01 7.60949671e-01
1.85529411e-01 5.92665255e-01 7.24300817e-02 -3.26769590e-01
2.60888219e-01 4.92234081e-01 2.71037996e-01 -1.53553402e+00
1.50574401e-01 -8.78684402e-01 -1.16155781e-01 -1.65211272e+00
5.20902388e-02 -1.11627126e+00 7.73433447e-02 -1.98468113e+00
7.62447536e-01 -6.69732988e-01 -1.12982905e+00 2.45428428e-01
-4.85975385e-01 3.76901835e-01 -8.05923715e-02 7.96647847e-01
-1.13158977e+00 6.65281594e-01 1.19970942e+00 -2.31226504e-01
-5.66411167e-02 -9.33439881e-02 -1.08790934e+00 4.09366190e-01
5.13810754e-01 -6.41995668e-01 -7.99181819e-01 -8.49648774e-01
9.14903462e-01 -1.15656950e-01 1.71047505e-02 -7.01355994e-01
6.61559165e-01 -6.32730573e-02 6.30205572e-02 -6.41087115e-01
2.82695770e-01 -6.51372015e-01 -2.40560621e-01 1.52797267e-01
-6.88526452e-01 1.28788605e-01 -1.69011891e-01 5.46901643e-01
-7.96242654e-01 -6.36835575e-01 2.67482221e-01 -3.73225749e-01
-1.05660164e+00 3.25692207e-01 -2.70162940e-01 3.65499556e-02
3.61477613e-01 -3.02624032e-02 -3.02242786e-01 -3.27145129e-01
-5.02117217e-01 3.85356605e-01 3.38990331e-01 1.03192818e+00
7.37815082e-01 -1.06023145e+00 -5.64501584e-01 -3.25220644e-01
4.99760419e-01 -1.26721114e-01 2.05456764e-01 4.15701836e-01
-3.12145054e-01 1.14083254e+00 2.33094737e-01 -2.15343401e-01
-1.14793468e+00 5.16462743e-01 -7.83609822e-02 -7.42455363e-01
-3.90375227e-01 1.02214849e+00 6.40190244e-02 -9.33870077e-02
3.18068534e-01 2.49504909e-01 -8.30882728e-01 4.50943589e-01
6.88224554e-01 2.20606878e-01 3.60437840e-01 -5.48653066e-01
-5.86549282e-01 7.99295902e-01 -6.23575449e-01 -2.92709112e-01
1.21346772e+00 -2.79422641e-01 -3.95706147e-01 2.12792948e-01
1.28917515e+00 -3.67969394e-01 -1.37869835e-01 -5.55091500e-01
6.07956290e-01 -3.76561910e-01 3.56536359e-01 -9.11808848e-01
-8.79355371e-01 8.87367010e-01 4.74968612e-01 2.01717675e-01
1.22318196e+00 1.89072311e-01 1.05543399e+00 7.23725140e-01
7.00854480e-01 -1.35427260e+00 1.54079407e-01 8.81699443e-01
7.32776344e-01 -1.19216156e+00 2.03282028e-01 -1.47561252e-01
-2.18441263e-01 1.09564877e+00 2.55314708e-01 3.17853913e-02
6.62937105e-01 -5.76893985e-01 1.14428572e-01 -6.93037868e-01
-7.98628092e-01 -6.39304578e-01 8.63904774e-01 1.85259998e-01
7.70210087e-01 -4.17723686e-01 -7.67665565e-01 4.38059747e-01
5.32532781e-02 -5.74288592e-02 -2.35468045e-01 1.01511657e+00
-7.10511327e-01 -1.68744385e+00 -8.45509693e-02 4.09781933e-01
-5.27833283e-01 -5.13823628e-01 -6.83443308e-01 4.99027371e-01
-5.05811453e-01 1.18215549e+00 -1.07085772e-01 -2.68292457e-01
2.34421968e-01 4.19196516e-01 3.52968514e-01 -8.82340014e-01
-6.27499819e-01 -1.32330423e-02 -1.54513210e-01 -5.37770092e-01
-4.85595465e-01 -3.36678922e-01 -9.26672339e-01 1.70530304e-01
-6.47718728e-01 7.92911053e-01 8.53310227e-01 1.01286530e+00
4.97547984e-01 2.17280895e-01 4.20155168e-01 -5.13982296e-01
-5.25105655e-01 -9.22790468e-01 -3.07083935e-01 2.81839788e-01
-5.31116091e-02 -6.80710971e-01 -3.85149479e-01 -3.96073520e-01]
|
[11.46572208404541, 7.656174659729004]
|
39f7011e-a0f0-4c97-bf30-595a12967620
|
deep-clustering-and-conventional-networks-for
|
1611.06265
| null |
http://arxiv.org/abs/1611.06265v2
|
http://arxiv.org/pdf/1611.06265v2.pdf
|
Deep Clustering and Conventional Networks for Music Separation: Stronger Together
|
Deep clustering is the first method to handle general audio separation
scenarios with multiple sources of the same type and an arbitrary number of
sources, performing impressively in speaker-independent speech separation
tasks. However, little is known about its effectiveness in other challenging
situations such as music source separation. Contrary to conventional networks
that directly estimate the source signals, deep clustering generates an
embedding for each time-frequency bin, and separates sources by clustering the
bins in the embedding space. We show that deep clustering outperforms
conventional networks on a singing voice separation task, in both matched and
mismatched conditions, even though conventional networks have the advantage of
end-to-end training for best signal approximation, presumably because its more
flexible objective engenders better regularization. Since the strengths of deep
clustering and conventional network architectures appear complementary, we
explore combining them in a single hybrid network trained via an approach akin
to multi-task learning. Remarkably, the combination significantly outperforms
either of its components.
|
['Zhuo Chen', 'Nima Mesgarani', 'John R. Hershey', 'Yi Luo', 'Jonathan Le Roux']
|
2016-11-18
| null | null | null | null |
['music-source-separation']
|
['music']
|
[ 1.71266079e-01 -2.46827096e-01 5.42468540e-02 2.53505856e-02
-1.30679882e+00 -8.83618295e-01 5.35918891e-01 -2.68704176e-01
-1.88139379e-01 3.69187593e-01 6.34847283e-01 -2.36326419e-02
-5.87127268e-01 6.31304905e-02 -2.67163485e-01 -9.26062822e-01
-2.59558052e-01 4.95664626e-01 -1.18207209e-01 -8.91752914e-02
-6.54155314e-02 2.73409754e-01 -1.54964101e+00 1.48669779e-01
7.41682172e-01 8.19358289e-01 -6.13628402e-02 8.74148488e-01
1.29433900e-01 5.44854820e-01 -8.55119884e-01 -2.26080731e-01
3.96639287e-01 -7.08794057e-01 -4.68018383e-01 -1.99679080e-02
4.50303733e-01 -5.71369827e-02 -2.63464510e-01 8.89120638e-01
1.01561987e+00 4.73073810e-01 6.67410672e-01 -1.28379297e+00
-2.50114977e-01 9.70511436e-01 -3.99987727e-01 1.90853447e-01
2.43773654e-01 -1.43625259e-01 1.20075774e+00 -7.77686596e-01
-5.92559285e-04 1.18341374e+00 1.02593839e+00 4.42908853e-01
-1.58668053e+00 -6.70069396e-01 -1.05444320e-01 1.63442101e-02
-1.15463352e+00 -1.12702608e+00 1.04782534e+00 -3.98372978e-01
7.59004474e-01 3.35176259e-01 3.76082182e-01 1.26524782e+00
-5.75198531e-01 9.20567155e-01 8.23158205e-01 -3.63291383e-01
9.11922976e-02 -3.21209133e-02 -1.57395378e-01 -9.84340534e-02
-6.31531999e-02 3.58568132e-02 -8.10254574e-01 -4.74916250e-01
3.02573383e-01 -1.78009272e-01 -5.62070608e-01 -3.12945873e-01
-1.50148857e+00 5.68105578e-01 8.44382718e-02 6.49226964e-01
-2.18316466e-01 2.99276382e-01 4.25293088e-01 3.07174414e-01
3.26331228e-01 6.77083135e-01 -3.29187661e-01 -2.90476739e-01
-1.56605148e+00 2.96002090e-01 9.81884897e-01 5.76764047e-01
2.80420840e-01 6.54976070e-01 -2.37086192e-02 1.18443477e+00
1.01647705e-01 3.22080135e-01 7.88576603e-01 -1.30402768e+00
2.27612868e-01 -2.18613401e-01 -3.00054215e-02 -7.64083207e-01
-3.89914334e-01 -8.73388886e-01 -7.54807830e-01 1.50335431e-01
7.38148332e-01 -5.11839211e-01 -7.13114202e-01 2.09023309e+00
8.57088529e-03 6.63598657e-01 1.92722321e-01 1.07796979e+00
5.79769254e-01 6.02759004e-01 -3.78442496e-01 -2.60124028e-01
9.68743980e-01 -1.10364568e+00 -6.69456661e-01 -4.22207415e-01
-1.46603078e-01 -1.04337895e+00 7.32346117e-01 7.32785761e-01
-1.25744081e+00 -5.90622485e-01 -1.03509164e+00 -1.94981729e-03
-1.69238821e-01 3.92797552e-02 6.27630532e-01 6.94135725e-01
-1.15591109e+00 8.10780406e-01 -6.41091824e-01 9.64613333e-02
1.28129661e-01 3.60304266e-01 -1.81939542e-01 3.02697420e-01
-1.01102269e+00 4.19919640e-01 7.41050169e-02 1.37279794e-01
-1.04693520e+00 -8.11487675e-01 -7.52669334e-01 3.38532329e-01
2.17589691e-01 -5.89828968e-01 1.35235310e+00 -1.20042205e+00
-1.59988558e+00 5.28400540e-01 -2.88582236e-01 -4.55534250e-01
2.24435464e-01 -3.12851280e-01 -5.81296504e-01 2.46382266e-01
7.31592700e-02 5.32326281e-01 1.42180753e+00 -1.34832060e+00
-3.68032873e-01 -2.24436149e-01 -3.75233710e-01 3.54771912e-01
-4.23022389e-01 1.58618510e-01 -1.91879049e-01 -8.65612566e-01
1.53108969e-01 -8.22500348e-01 -1.20942324e-01 -4.56871688e-01
-5.32714427e-01 -3.63136791e-02 6.42054617e-01 -6.38148606e-01
1.01996648e+00 -2.52337718e+00 5.43050706e-01 5.55300526e-02
3.18536729e-01 3.33778977e-01 -2.50953138e-01 5.60753763e-01
-3.43424350e-01 -4.06163000e-02 -4.00731742e-01 -8.35595489e-01
2.59521842e-01 -2.45305747e-01 -5.11543989e-01 4.25910234e-01
4.13205437e-02 5.06365061e-01 -8.43590438e-01 -1.05183177e-01
1.54003771e-02 6.80235207e-01 -5.22242904e-01 1.34714231e-01
1.19530238e-01 5.20764470e-01 2.53653556e-01 3.31906557e-01
3.77143949e-01 9.50867683e-02 2.11055949e-01 2.50044595e-02
7.82866850e-02 6.06181502e-01 -1.49358797e+00 1.95353937e+00
-3.53290766e-01 1.00703084e+00 8.60583842e-01 -1.15479267e+00
5.42080939e-01 9.05573070e-01 6.81870818e-01 4.38221022e-02
-1.51422145e-02 5.13291478e-01 4.97082412e-01 -2.06996113e-01
2.62756646e-01 -5.94152868e-01 3.97351496e-02 4.85734552e-01
4.56779003e-01 -1.97632164e-01 1.09926460e-03 2.04628065e-01
1.00980854e+00 -1.67531461e-01 4.24460880e-03 -1.44941345e-01
1.78034738e-01 -3.90312344e-01 5.26462376e-01 7.42502093e-01
-3.43271762e-01 9.95582640e-01 3.90436083e-01 2.90558815e-01
-6.39161050e-01 -1.27043676e+00 -1.68300718e-01 1.30113351e+00
-1.70339644e-01 -4.67100322e-01 -6.06447399e-01 -2.58299649e-01
2.11623922e-01 5.40772617e-01 -1.55987605e-01 -1.03022024e-01
-5.42159498e-01 -7.60319114e-01 1.02562368e+00 4.66727942e-01
-4.76926984e-03 -8.90174627e-01 -2.05806226e-01 3.63953441e-01
-3.91135126e-01 -8.94523621e-01 -5.62725961e-01 7.26799965e-01
-6.08866274e-01 -8.08279812e-01 -8.97800982e-01 -7.90404618e-01
-1.10584348e-01 5.94876766e-01 9.76254582e-01 -2.29113936e-01
-2.56634708e-02 5.11591375e-01 -5.06713167e-02 -4.57269609e-01
-3.50373477e-01 2.40248535e-02 4.07378495e-01 1.71254054e-01
3.48555326e-01 -1.16872299e+00 -5.39850056e-01 7.21267909e-02
-7.48742223e-01 -4.75831389e-01 2.93317318e-01 8.16578984e-01
-7.77017176e-02 3.71616900e-01 1.09670711e+00 -3.31270993e-01
9.04784501e-01 -5.72515488e-01 -1.99001580e-02 -2.97455668e-01
-2.85536885e-01 -1.58383891e-01 8.00387681e-01 -5.33047020e-01
-8.15836787e-01 -7.76931122e-02 -2.76918411e-01 -7.09554732e-01
-4.71344531e-01 3.33056629e-01 -1.66228011e-01 3.32529008e-01
5.86394250e-01 1.08485088e-01 5.74620701e-02 -7.51970410e-01
3.88390958e-01 8.13907802e-01 8.75616610e-01 -3.63887221e-01
9.00070965e-01 3.68302107e-01 -4.06017542e-01 -9.82616365e-01
-6.76405072e-01 -8.84197176e-01 -5.86839616e-01 1.02220021e-01
7.48528421e-01 -1.17750239e+00 -6.20500743e-01 4.51609313e-01
-9.97533143e-01 -3.02722692e-01 -3.76289845e-01 7.95085073e-01
-5.67803741e-01 3.84357661e-01 -6.13219917e-01 -1.05610704e+00
1.68821365e-02 -1.09993124e+00 1.13171983e+00 -6.14616647e-02
-4.62530911e-01 -1.06306148e+00 2.71993488e-01 3.93761754e-01
4.61657405e-01 -1.30308429e-02 6.61861539e-01 -1.11660635e+00
-1.20425105e-01 -1.44899532e-01 2.69994110e-01 6.75938845e-01
2.88572967e-01 -1.37365222e-01 -1.64540410e+00 -3.53252679e-01
4.47365254e-01 -2.15120614e-01 1.22333050e+00 6.87159002e-01
9.17004943e-01 -1.89672917e-01 -7.46459067e-02 6.91907883e-01
1.02009821e+00 1.83925390e-01 3.24364543e-01 -1.35028094e-01
7.55545020e-01 6.49358809e-01 -1.83970124e-01 8.73806477e-02
2.90715997e-03 6.84488416e-01 8.70114267e-02 -1.74403608e-01
-4.06520754e-01 -1.21210078e-02 5.31534135e-01 1.15309405e+00
1.67114526e-01 -3.10274214e-01 -7.16564953e-01 8.58775258e-01
-1.74038041e+00 -1.40857995e+00 1.78379640e-01 2.33018422e+00
9.39912319e-01 1.34934783e-01 5.90238631e-01 7.93629885e-01
6.68058276e-01 3.73012573e-01 -5.11803091e-01 -1.79743558e-01
-2.45161861e-01 2.73974657e-01 -1.65012665e-02 5.33614099e-01
-1.22776067e+00 5.04830837e-01 7.13568830e+00 1.04034638e+00
-1.15087509e+00 1.87173516e-01 1.94810525e-01 -6.16192698e-01
-3.08308423e-01 -1.18275188e-01 -3.26699466e-01 4.84626114e-01
1.18506300e+00 1.01913013e-01 8.53306115e-01 5.24218798e-01
1.77260995e-01 1.40244409e-01 -1.26783097e+00 1.14198983e+00
2.42355466e-01 -1.01706886e+00 -3.74005944e-01 -5.78698851e-02
4.98618633e-01 8.94726813e-02 2.21770316e-01 2.04126030e-01
1.90924510e-01 -1.20464098e+00 9.53528941e-01 5.89669682e-02
5.57559311e-01 -7.05459416e-01 3.92445207e-01 4.45764333e-01
-1.06832790e+00 -3.00199091e-01 -3.59353013e-02 -1.28253892e-01
2.29417264e-01 5.82478225e-01 -7.48286605e-01 4.60372120e-01
5.66003084e-01 6.59837842e-01 -2.69151568e-01 1.14287448e+00
4.04532701e-02 1.02707374e+00 -3.52077216e-01 5.58385611e-01
1.53187543e-01 -1.96923554e-01 1.09239662e+00 1.48386455e+00
2.70322740e-01 -2.64621735e-01 -1.20190270e-02 8.15033257e-01
4.39356156e-02 -2.34265536e-01 -6.75187290e-01 -2.22421840e-01
6.86723888e-01 1.31283140e+00 -6.12581611e-01 -1.37692556e-01
-2.32522771e-01 8.60510588e-01 1.13890685e-01 5.99020660e-01
-6.20760679e-01 -6.13761127e-01 8.79079580e-01 -1.20635398e-01
4.82226551e-01 -3.81704271e-01 -4.01030600e-01 -1.09943438e+00
-1.14236943e-01 -1.24531531e+00 2.33828813e-01 -5.64678371e-01
-1.34236085e+00 5.83199680e-01 -1.92998126e-01 -1.29263759e+00
-5.19424796e-01 -4.91745889e-01 -8.96562457e-01 8.58842492e-01
-1.29744017e+00 -8.02070916e-01 2.28617683e-01 6.31555676e-01
5.31534195e-01 -3.25879097e-01 8.60821664e-01 5.52545547e-01
-6.42450571e-01 6.36961341e-01 4.91642356e-01 1.15891129e-01
1.11524796e+00 -1.69782710e+00 1.39924124e-01 9.92851317e-01
8.30909491e-01 7.67321110e-01 7.64878094e-01 -8.47004950e-02
-1.13470936e+00 -6.72661602e-01 6.02525055e-01 -5.21766841e-01
6.54350579e-01 -6.01940334e-01 -8.56491566e-01 4.09383059e-01
6.14386201e-01 -4.02747780e-01 1.17274928e+00 4.91866350e-01
-5.11363983e-01 -1.27648443e-01 -6.75407350e-01 4.59897012e-01
8.92119884e-01 -8.56537580e-01 -7.52204835e-01 1.61453545e-01
7.91306496e-01 -1.62948608e-01 -7.53597200e-01 1.18574895e-01
4.32943255e-01 -1.34426415e+00 1.29417336e+00 -5.46137333e-01
3.13051224e-01 -3.35870981e-01 -3.28229815e-01 -1.75738287e+00
-4.95754689e-01 -1.18857753e+00 -1.38913289e-01 1.53267336e+00
5.56965947e-01 -4.32001650e-01 4.01666671e-01 9.97789949e-02
-5.58462262e-01 -3.23073506e-01 -1.00164366e+00 -9.35822189e-01
1.87533110e-01 -6.21636748e-01 5.54093122e-01 9.85376000e-01
1.36698231e-01 6.47977889e-01 -4.81732130e-01 1.54879183e-01
6.28764391e-01 2.24792600e-01 6.45985305e-01 -1.37323797e+00
-7.55033493e-01 -9.22813475e-01 -1.41204149e-01 -9.27713335e-01
4.59163398e-01 -1.00503731e+00 2.98590571e-01 -1.22789061e+00
-1.27526551e-01 -3.40254545e-01 -4.24461693e-01 2.34259397e-01
-3.08618128e-01 4.26543057e-01 3.05321515e-01 2.88874626e-01
-4.36171025e-01 4.84175712e-01 6.79218590e-01 -1.65854469e-01
-4.01150197e-01 2.66735196e-01 -1.03983688e+00 8.73255491e-01
6.73918962e-01 -4.13261980e-01 -4.48957175e-01 -5.21467090e-01
-1.50086045e-01 2.03197420e-01 4.49056208e-01 -1.30781782e+00
2.79108644e-01 3.38896096e-01 3.28642160e-01 -2.90317059e-01
7.90507257e-01 -7.52644122e-01 1.89260826e-01 -3.89168635e-02
-5.55450737e-01 -4.60418552e-01 3.03988636e-01 6.29248679e-01
-5.87215424e-01 -2.44498640e-01 7.08237231e-01 1.49734482e-01
-1.64279550e-01 -9.23104957e-02 -4.65713888e-01 2.82437235e-01
4.59794611e-01 -3.06403190e-01 4.39273426e-03 -7.77955711e-01
-8.61186802e-01 -1.06277488e-01 9.20608342e-02 3.90541673e-01
3.16648632e-01 -1.37454462e+00 -7.40577638e-01 2.61500746e-01
-3.27565432e-01 -1.43279761e-01 1.25256926e-01 1.11640549e+00
1.48431972e-01 2.88946748e-01 1.99746296e-01 -6.26856923e-01
-9.53297079e-01 4.80809569e-01 4.17512357e-01 7.91199505e-02
-4.01090115e-01 1.02041304e+00 2.45728388e-01 -4.34320122e-01
6.61500514e-01 1.91894062e-02 -4.11984958e-02 4.32698935e-01
5.25816143e-01 5.84981024e-01 1.16016410e-01 -6.75820947e-01
-4.00600195e-01 3.79610807e-01 3.66855353e-01 -5.44537604e-01
1.28159964e+00 -1.46712273e-01 -7.06943125e-02 8.51049721e-01
1.23618066e+00 6.39781415e-01 -1.26688564e+00 -1.57152504e-01
-1.73121765e-02 -3.93485874e-01 4.89675030e-02 -8.00221503e-01
-9.67903316e-01 1.14755571e+00 2.63117254e-01 6.20495617e-01
1.19915855e+00 -5.38032427e-02 8.08455944e-01 1.23471081e-01
-1.50384337e-01 -8.47380817e-01 1.41380414e-01 3.25810850e-01
8.01455915e-01 -1.25182307e+00 -3.28101873e-01 3.45944501e-02
-6.46601737e-01 1.00839067e+00 1.72053292e-01 -1.24832816e-01
6.73202217e-01 4.22992170e-01 3.54799241e-01 -4.24320437e-02
-5.88521361e-01 -3.32469821e-01 5.61752856e-01 7.85180032e-01
6.46486521e-01 5.43944500e-02 3.76465499e-01 8.56698453e-01
-5.07460117e-01 -6.41597390e-01 2.87435919e-01 3.64606202e-01
-4.11553144e-01 -9.93981183e-01 -6.99282527e-01 3.29376519e-01
-7.29268849e-01 -2.45325267e-01 -5.72569907e-01 3.97820920e-01
1.00876510e-01 1.53299952e+00 8.49780568e-04 -4.29799616e-01
1.04500517e-01 5.08427978e-01 2.47155398e-01 -6.34095669e-01
-8.59664738e-01 8.42673182e-01 8.27037692e-02 -2.85650015e-01
-5.86135387e-01 -8.57010543e-01 -1.02874970e+00 -6.08472005e-02
-3.68889242e-01 4.39836234e-01 4.56549436e-01 8.02028954e-01
3.07820141e-01 6.63180292e-01 7.03630328e-01 -1.39614642e+00
-7.11607933e-01 -1.01159298e+00 -8.21416438e-01 4.72311735e-01
1.01582003e+00 -4.48589861e-01 -8.76343727e-01 2.99614538e-02]
|
[15.399054527282715, 5.592049598693848]
|
df25b4b0-e426-4167-b4fc-7afba06bf101
|
toward-a-neural-semantic-parsing-system-for
|
2211.04569
| null |
https://arxiv.org/abs/2211.04569v1
|
https://arxiv.org/pdf/2211.04569v1.pdf
|
Toward a Neural Semantic Parsing System for EHR Question Answering
|
Clinical semantic parsing (SP) is an important step toward identifying the exact information need (as a machine-understandable logical form) from a natural language query aimed at retrieving information from electronic health records (EHRs). Current approaches to clinical SP are largely based on traditional machine learning and require hand-building a lexicon. The recent advancements in neural SP show a promise for building a robust and flexible semantic parser without much human effort. Thus, in this paper, we aim to systematically assess the performance of two such neural SP models for EHR question answering (QA). We found that the performance of these advanced neural models on two clinical SP datasets is promising given their ease of application and generalizability. Our error analysis surfaces the common types of errors made by these models and has the potential to inform future research into improving the performance of neural SP models for EHR QA.
|
['Kirk Roberts', 'Sarvesh Soni']
|
2022-11-08
| null | null | null | null |
['semantic-parsing']
|
['natural-language-processing']
|
[ 4.42820013e-01 7.05219090e-01 -1.00821503e-01 -7.95263469e-01
-1.43633699e+00 -4.22845513e-01 -1.97452873e-01 7.85016894e-01
-4.39231664e-01 4.18139786e-01 4.75730240e-01 -9.62449193e-01
-2.33674452e-01 -6.72989607e-01 -6.75387383e-01 1.71768412e-01
5.42875305e-02 7.97233701e-01 6.25398606e-02 -9.04202685e-02
2.97733042e-02 9.62704271e-02 -7.52461791e-01 6.97611332e-01
8.73705328e-01 9.08157110e-01 9.54403877e-02 7.38536119e-01
-3.23734164e-01 1.27154601e+00 -5.06416917e-01 -7.25615084e-01
-2.62431979e-01 -5.34872949e-01 -1.43198431e+00 -5.80880225e-01
1.44276693e-01 -3.37791651e-01 1.02236785e-01 9.86955762e-01
6.67875707e-01 -3.21469426e-01 8.93517882e-02 -7.01447010e-01
-7.03861237e-01 6.80726826e-01 3.36182117e-01 3.50200295e-01
7.94906557e-01 7.79252052e-02 1.29617083e+00 -3.35000902e-01
9.03769612e-01 1.05498099e+00 1.18166888e+00 9.07148600e-01
-9.06381488e-01 -2.96773821e-01 -2.07100004e-01 1.90922633e-01
-1.09671283e+00 -5.17509758e-01 4.22649264e-01 -3.13153565e-01
1.52640069e+00 2.45835200e-01 3.57521683e-01 7.91521013e-01
3.51126045e-01 8.45656216e-01 7.26965845e-01 -5.36578476e-01
3.55102748e-01 7.87959844e-02 7.93921590e-01 8.73767078e-01
4.70877439e-01 -2.71245956e-01 -3.00527781e-01 -8.07426751e-01
4.62586373e-01 -2.71160185e-01 -1.50175080e-01 1.31316885e-01
-7.09235311e-01 8.00663650e-01 4.29146379e-01 3.00269037e-01
-5.86383522e-01 1.40628099e-01 5.98123372e-01 7.23963827e-02
1.41697451e-01 9.13852096e-01 -1.05187321e+00 -3.80108863e-01
-5.79803526e-01 2.48644844e-01 1.19473767e+00 9.87454832e-01
1.05071463e-01 -5.44595420e-01 -3.87841091e-02 6.26804173e-01
4.55369860e-01 9.67080444e-02 4.77690160e-01 -9.96273220e-01
4.98729765e-01 8.67975891e-01 1.46844655e-01 -8.16947281e-01
-8.37263823e-01 -1.21932670e-01 -1.36669055e-01 -7.09442079e-01
3.88021052e-01 -2.00604230e-01 -8.55702698e-01 1.64334857e+00
3.06937355e-03 -1.51988745e-01 5.14850795e-01 4.43897963e-01
1.19118643e+00 1.26996413e-01 9.13048029e-01 2.33930796e-01
1.84649813e+00 -4.81178939e-01 -7.72960007e-01 -4.03256923e-01
1.20352256e+00 -3.90491277e-01 1.00287497e+00 2.52592769e-02
-1.25000799e+00 -6.22294173e-02 -6.68150306e-01 -5.05840778e-01
-1.63244367e-01 -1.14460818e-01 6.97287261e-01 8.00358713e-01
-1.11836493e+00 4.20418084e-01 -1.28393173e+00 -5.17968774e-01
6.09774530e-01 3.86043698e-01 -1.73826665e-01 -3.45446944e-01
-1.41021168e+00 1.04059803e+00 3.16459566e-01 1.26211375e-01
-2.73817480e-01 -7.27705479e-01 -1.14229810e+00 2.37908632e-01
2.16580957e-01 -1.21346211e+00 1.71576369e+00 -6.54243231e-01
-1.06449044e+00 9.68918622e-01 -4.66776460e-01 -5.80712795e-01
2.53083631e-02 -1.10624343e-01 -3.04294616e-01 4.46567982e-01
1.33086652e-01 4.92857754e-01 -2.24136710e-02 -6.44587994e-01
-4.84789699e-01 -6.46663368e-01 -6.49414212e-03 1.50625650e-02
2.85043176e-02 4.06343311e-01 -1.74136594e-01 -2.92977870e-01
2.21462160e-01 -6.91575348e-01 -3.57561916e-01 -2.73923930e-02
-1.93663999e-01 -4.93248582e-01 -1.03858430e-02 -1.19381130e+00
1.21030176e+00 -1.89657092e+00 -2.82452732e-01 -1.11430444e-01
2.78695166e-01 2.52322078e-01 6.93671927e-02 3.51213306e-01
1.22587018e-01 5.54225624e-01 -3.48538548e-01 -4.50894721e-02
-1.75135568e-01 3.53083909e-01 -5.65901883e-02 -1.51978090e-01
5.20847380e-01 1.47923028e+00 -1.00297797e+00 -5.73669672e-01
-3.50774676e-01 3.82009983e-01 -8.87934208e-01 3.36697727e-01
-3.83969039e-01 2.05097675e-01 -8.53472054e-01 7.96419501e-01
2.08535597e-01 -7.55730093e-01 5.37878096e-01 -6.40688688e-02
6.05259299e-01 1.11098027e+00 -5.18231273e-01 1.62812686e+00
-1.08549267e-01 1.12958767e-01 2.95688920e-02 -8.85147691e-01
5.74671209e-01 4.39244747e-01 4.77516055e-01 -7.56174862e-01
-7.08004907e-02 4.17162031e-01 7.54466578e-02 -1.05269289e+00
1.97089806e-01 -4.48429555e-01 -1.78929433e-01 3.77947897e-01
-1.78522035e-01 1.20903082e-01 -3.36076617e-01 8.27627480e-02
1.48798048e+00 -1.78812407e-02 6.45361483e-01 -2.91607022e-01
3.32401335e-01 5.72694957e-01 7.20213830e-01 9.98205900e-01
-3.54187071e-01 3.98779958e-01 4.92639959e-01 -7.11155772e-01
-7.23261356e-01 -7.29022980e-01 -3.71935338e-01 7.29252338e-01
-2.25691766e-01 -3.66656840e-01 -8.04201245e-01 -8.78492892e-01
-5.04540019e-02 8.76816154e-01 -3.19337368e-01 -1.32148847e-01
-7.61464953e-01 -8.74167502e-01 1.01352715e+00 9.90214586e-01
1.38388723e-01 -1.32820153e+00 -9.55015540e-01 6.04599774e-01
-5.29745936e-01 -1.43531823e+00 2.51071784e-03 1.97343752e-01
-1.17246234e+00 -1.48333728e+00 -2.62126744e-01 -9.05382097e-01
5.68138361e-01 -3.10751766e-01 1.40815330e+00 3.06666166e-01
-2.53452599e-01 6.27970099e-01 -2.38126323e-01 -5.58160603e-01
-8.44353437e-01 1.21256396e-01 -4.43248123e-01 -8.65847051e-01
1.20798099e+00 1.90961331e-01 -6.10775590e-01 -1.10207759e-01
-9.58866298e-01 -1.62615538e-01 3.30947995e-01 8.40341270e-01
4.54776704e-01 -3.23883981e-01 8.59064758e-01 -1.38349926e+00
1.07944059e+00 -6.29158139e-01 -1.35079250e-01 5.15027583e-01
-8.88384163e-01 2.80357808e-01 3.74861449e-01 2.68112391e-01
-1.07660472e+00 -1.46561558e-03 -8.97525609e-01 4.50422883e-01
-3.49744260e-01 1.10081208e+00 1.35085791e-01 1.82018384e-01
8.03456426e-01 -7.64426962e-02 5.63948266e-02 -5.32836080e-01
-9.94857401e-04 7.51321614e-01 3.20202559e-01 -6.31555021e-01
-1.35948151e-01 5.67199104e-02 -3.49433690e-01 -4.26291853e-01
-9.53962922e-01 -5.19731104e-01 -1.08262569e-01 5.38371563e-01
1.19051874e+00 -8.41863692e-01 -7.08654761e-01 1.36291996e-01
-1.01680660e+00 -2.91340739e-01 -2.79257804e-01 2.63361633e-01
-4.33747828e-01 5.50175667e-01 -1.16360843e+00 -4.56010491e-01
-7.07711995e-01 -1.09516954e+00 1.24216032e+00 3.31058912e-03
-8.68600547e-01 -1.11751759e+00 6.02226593e-02 9.04764235e-01
3.29692811e-01 -2.07878411e-01 1.60755062e+00 -1.17889214e+00
-3.63944590e-01 -3.31378043e-01 -2.97891468e-01 1.04411274e-01
1.09396361e-01 -6.03972375e-01 -7.30478704e-01 5.94736598e-02
3.89059365e-01 -4.24684107e-01 4.14975315e-01 4.89267915e-01
1.08254588e+00 -3.85001004e-01 -4.44852829e-01 3.30882818e-01
1.48470950e+00 4.08202171e-01 5.56180656e-01 3.05982441e-01
4.78307545e-01 5.46154499e-01 2.72238612e-01 3.58961523e-03
9.73584712e-01 2.01581568e-01 1.24951145e-02 7.12628290e-02
1.92362741e-01 -3.28353316e-01 -9.30505171e-02 6.03850961e-01
2.96857357e-01 -2.04452083e-01 -1.54402554e+00 5.73488176e-01
-1.63233054e+00 -4.03627306e-01 1.08880594e-01 1.79030514e+00
1.07350254e+00 6.66024338e-04 -3.60925257e-01 -2.79091328e-01
2.29230255e-01 -3.11931252e-01 -6.74739003e-01 -5.15559018e-01
2.32813224e-01 4.59678441e-01 2.76567280e-01 4.99384284e-01
-7.10260212e-01 8.20712030e-01 7.43582153e+00 -7.69324526e-02
-6.51412725e-01 8.67318809e-02 7.28463948e-01 3.24091196e-01
-4.59263772e-01 -3.07487816e-01 -8.99191737e-01 2.15517774e-01
1.42696202e+00 1.20616041e-01 8.93872082e-02 7.79373169e-01
1.09554887e-01 -3.73268016e-02 -1.46118855e+00 6.18582428e-01
3.83763239e-02 -1.50560927e+00 1.46898301e-02 -2.69899696e-01
1.23907134e-01 2.18507826e-01 -4.00246769e-01 3.56883794e-01
5.44532716e-01 -1.13031042e+00 9.33519453e-02 4.43573684e-01
5.03986061e-01 -9.70287025e-02 1.25048101e+00 2.50240862e-01
-6.18439078e-01 -3.16116631e-01 -1.33491457e-01 4.85662669e-02
2.89064735e-01 1.66970819e-01 -1.19568098e+00 4.52519953e-01
7.83570707e-01 4.94855136e-01 -5.74679017e-01 8.27206671e-01
-5.56233600e-02 7.40289152e-01 -2.31191009e-01 -2.62365676e-03
1.18895829e-01 4.74412411e-01 1.86496630e-01 1.36420989e+00
3.17168161e-02 5.58441162e-01 2.94973925e-02 8.45464230e-01
-1.78422794e-01 9.73570123e-02 -3.92842680e-01 -4.19235498e-01
3.12636703e-01 5.99482954e-01 -5.73915660e-01 -4.30670261e-01
-5.15440643e-01 7.07671404e-01 3.83379817e-01 1.96670666e-01
-3.96142304e-01 -4.34694327e-02 4.29925978e-01 4.70971130e-02
-1.80661911e-03 1.91783160e-01 -6.10363245e-01 -1.09987640e+00
3.34982008e-01 -1.20533228e+00 1.02687025e+00 -7.78494537e-01
-1.42002249e+00 5.26782155e-01 -2.37985402e-01 -4.54784542e-01
-5.49435914e-01 -7.50830412e-01 1.96888354e-02 8.55302989e-01
-1.65414584e+00 -1.03597808e+00 -1.18902683e-01 3.36069226e-01
3.39496464e-01 1.87317282e-01 1.32637656e+00 3.16150665e-01
-4.38942432e-01 6.06598496e-01 -2.94647723e-01 3.37565392e-01
4.56814736e-01 -1.08757973e+00 7.04552889e-01 4.57974255e-01
-2.03314260e-01 9.86672819e-01 3.41217637e-01 -8.79923820e-01
-1.37671649e+00 -8.42775524e-01 1.62948155e+00 -9.60259855e-01
3.70442033e-01 3.86079103e-02 -1.10513866e+00 1.10312784e+00
-2.19163820e-01 -2.99137145e-01 1.13371921e+00 2.88399607e-01
-1.93156436e-01 2.87109166e-01 -1.49504435e+00 1.52534932e-01
8.92701685e-01 -8.00893188e-01 -1.18475342e+00 3.43162328e-01
9.72300828e-01 -4.56269324e-01 -1.23620796e+00 5.82389116e-01
3.62425894e-01 -4.93994057e-01 9.39459383e-01 -1.32077289e+00
5.27250886e-01 8.41362402e-02 -8.25062916e-02 -7.46719778e-01
-1.47894964e-01 -3.06788296e-01 6.65831147e-03 5.05295157e-01
7.05271244e-01 -8.05857122e-01 9.22198832e-01 1.54576957e+00
-2.41485596e-01 -9.19012606e-01 -7.84868240e-01 -1.88617378e-01
1.76340044e-01 -5.03454685e-01 6.39641821e-01 1.15279126e+00
3.81239355e-01 3.04128557e-01 4.40324634e-01 3.66581678e-01
2.23876625e-01 -2.52473146e-01 -2.14609224e-02 -1.27625728e+00
-4.34586346e-01 -1.87319353e-01 -4.49081779e-01 -9.11271274e-01
-4.30220598e-03 -9.82471049e-01 4.48758304e-02 -2.01031590e+00
4.60850775e-01 -7.32576907e-01 -4.90836263e-01 8.75085950e-01
-5.34428835e-01 -3.01239729e-01 -1.91476986e-01 1.38596803e-01
-5.86933434e-01 -1.27028212e-01 7.16188073e-01 -4.04663431e-03
-2.95403183e-01 -6.64062724e-02 -1.29521990e+00 7.04263628e-01
8.16738427e-01 -7.15109766e-01 -3.76022220e-01 -9.58474100e-01
5.89125335e-01 5.56398928e-01 2.74838865e-01 -5.19425273e-01
3.82623464e-01 2.46822581e-01 -2.42004339e-02 -1.96766898e-01
-1.02787279e-01 -7.43254840e-01 -1.21419057e-01 7.44392812e-01
-7.93368340e-01 3.12651426e-01 4.27691311e-01 4.33357805e-01
-2.39168480e-01 -5.23708045e-01 4.11704004e-01 -6.20834887e-01
-5.61357260e-01 -6.84163198e-02 -3.98824275e-01 5.03178537e-01
5.13153315e-01 -6.12391010e-02 -3.50108594e-01 -2.17857689e-01
-9.00692165e-01 2.77527273e-01 1.25033304e-01 3.26907665e-01
5.76186538e-01 -5.87286532e-01 -4.65824455e-01 3.65637569e-03
2.21217349e-01 2.39818901e-01 9.37756673e-02 5.55871129e-01
-8.41476202e-01 9.53847110e-01 5.35035655e-02 -5.04299283e-01
-1.10028100e+00 4.45386529e-01 5.14248371e-01 -6.61764205e-01
-7.48553634e-01 8.89110982e-01 -2.25175872e-01 -6.17995501e-01
2.43597329e-01 -6.77677989e-01 -1.36416197e-01 -5.54667354e-01
4.68731910e-01 3.90052758e-02 2.70209670e-01 -1.27619565e-01
-6.33723259e-01 1.26486003e-01 -1.73211217e-01 1.48994938e-01
1.49120688e+00 1.29362330e-01 -3.01861048e-01 -1.80831142e-02
1.02407300e+00 -3.67923230e-01 -4.57333297e-01 -2.05110803e-01
7.68685400e-01 1.56202927e-01 -8.69062915e-02 -1.15546584e+00
-5.23117304e-01 7.44818270e-01 4.19461876e-01 -1.58483780e-03
9.86157417e-01 2.68654466e-01 1.11113751e+00 7.82090366e-01
3.22523028e-01 -8.22287619e-01 -4.04720068e-01 3.64583492e-01
2.92258948e-01 -1.26726592e+00 -2.43091673e-01 -4.84312654e-01
-5.27731240e-01 1.01789927e+00 3.94738853e-01 3.89183968e-01
5.31182349e-01 2.90191412e-01 2.89864451e-01 -7.49168634e-01
-6.50443435e-01 2.15965793e-01 9.40890312e-02 5.07821023e-01
7.83386052e-01 6.71463087e-02 -3.63822639e-01 8.91904831e-01
-9.19345096e-02 5.94029307e-01 2.53234506e-01 1.21931851e+00
-2.12319732e-01 -1.22728693e+00 1.06476285e-01 6.84032142e-01
-1.05555415e+00 -5.44759929e-01 -4.41969067e-01 4.69320893e-01
-2.18137726e-01 9.57548976e-01 -1.65660754e-01 8.86008143e-02
5.09307206e-01 6.96248174e-01 3.16441745e-01 -9.14091468e-01
-9.24253106e-01 -4.25009757e-01 6.99764073e-01 -7.76000023e-01
-2.42218763e-01 -6.76430106e-01 -1.56904614e+00 4.11329091e-01
-6.56821672e-03 1.19174421e-01 5.36483705e-01 1.00790358e+00
8.47376704e-01 4.09297526e-01 -3.04344207e-01 4.95829463e-01
-7.50833154e-01 -5.41588247e-01 2.31210649e-01 5.22486806e-01
1.78863391e-01 9.90993083e-02 2.17941284e-01 1.61303952e-01]
|
[8.751130104064941, 8.538110733032227]
|
d50fbb9f-892d-4c2e-a23e-24226fb207d6
|
the-meccano-dataset-understanding-human
|
2010.05654
| null |
https://arxiv.org/abs/2010.05654v1
|
https://arxiv.org/pdf/2010.05654v1.pdf
|
The MECCANO Dataset: Understanding Human-Object Interactions from Egocentric Videos in an Industrial-like Domain
|
Wearable cameras allow to collect images and videos of humans interacting with the world. While human-object interactions have been thoroughly investigated in third person vision, the problem has been understudied in egocentric settings and in industrial scenarios. To fill this gap, we introduce MECCANO, the first dataset of egocentric videos to study human-object interactions in industrial-like settings. MECCANO has been acquired by 20 participants who were asked to build a motorbike model, for which they had to interact with tiny objects and tools. The dataset has been explicitly labeled for the task of recognizing human-object interactions from an egocentric perspective. Specifically, each interaction has been labeled both temporally (with action segments) and spatially (with active object bounding boxes). With the proposed dataset, we investigate four different tasks including 1) action recognition, 2) active object detection, 3) active object recognition and 4) egocentric human-object interaction detection, which is a revisited version of the standard human-object interaction detection task. Baseline results show that the MECCANO dataset is a challenging benchmark to study egocentric human-object interactions in industrial-like scenarios. We publicy release the dataset at https://iplab.dmi.unict.it/MECCANO.
|
['Giovanni Maria Farinella', 'Salvatore Livatino', 'Antonino Furnari', 'Francesco Ragusa']
|
2020-10-12
| null | null | null | null |
['active-object-detection']
|
['computer-vision']
|
[ 1.88522279e-01 -1.93309218e-01 -9.80427265e-02 -8.06964040e-02
7.91944191e-02 -5.68481743e-01 7.24943817e-01 -3.83006692e-01
-3.41577142e-01 2.99471706e-01 2.01734781e-01 3.11280578e-01
-8.53117108e-02 -1.95538253e-01 -7.51925528e-01 -7.12523818e-01
-1.28326237e-01 4.65515614e-01 3.01753283e-01 7.78514445e-02
2.17201099e-01 5.88435233e-01 -1.77656507e+00 4.38310772e-01
2.84779251e-01 8.52834702e-01 2.94455051e-01 1.02016222e+00
7.65103579e-01 1.02492559e+00 -5.25672317e-01 -1.59069762e-01
4.19333071e-01 -7.39385933e-02 -7.52380252e-01 5.76412499e-01
6.06126547e-01 -4.41852987e-01 -5.99430144e-01 9.25334454e-01
3.77056211e-01 5.82140148e-01 5.91401041e-01 -1.85710669e+00
-2.64437646e-01 2.21667543e-01 -5.45000911e-01 3.08281988e-01
7.75621474e-01 5.89871168e-01 4.39173698e-01 -1.05137217e+00
8.32536459e-01 1.52255225e+00 7.94190168e-02 4.76357281e-01
-8.34738255e-01 -4.51101661e-01 1.97339863e-01 7.37474144e-01
-1.14313447e+00 -5.64087510e-01 6.63170218e-01 -8.10136855e-01
1.00860500e+00 1.37984931e-01 9.08152401e-01 1.64066660e+00
1.58942327e-01 1.21910799e+00 4.32710528e-01 -4.03380662e-01
3.44853222e-01 -5.54654412e-02 3.78100455e-01 2.41682485e-01
3.23278874e-01 -2.86514964e-02 -6.38620317e-01 2.32535139e-01
8.53860855e-01 3.12662780e-01 -3.13834012e-01 -1.05145252e+00
-1.78283656e+00 3.24970603e-01 2.08495304e-01 1.11895315e-01
-6.52480245e-01 2.32868716e-01 6.53560281e-01 -1.90775096e-02
1.56758428e-01 1.46024883e-01 -2.26808399e-01 -4.85368997e-01
-1.33358300e-01 4.49813485e-01 7.47200370e-01 1.42021191e+00
2.23482743e-01 -2.77379304e-01 -1.81234106e-01 5.25702178e-01
8.43989924e-02 4.19667304e-01 1.73297063e-01 -1.19450641e+00
8.48555207e-01 6.85709596e-01 6.22860670e-01 -9.92326379e-01
-3.11098784e-01 2.02362850e-01 -4.59422171e-01 3.42119753e-01
7.53787041e-01 -3.70393917e-02 -3.86274725e-01 1.28606009e+00
6.35318100e-01 1.34981155e-01 -3.25397588e-02 1.21131349e+00
6.09277844e-01 2.94518501e-01 1.15849853e-01 -2.78771576e-03
1.72706318e+00 -1.34127522e+00 -9.01392758e-01 -2.80680388e-01
6.12703979e-01 -5.49584627e-01 9.82215881e-01 6.85769022e-01
-9.92220104e-01 -8.27177405e-01 -8.33492100e-01 6.71777455e-03
-4.80704308e-01 3.21806014e-01 5.76267600e-01 4.38129902e-01
-3.39554191e-01 1.50410026e-01 -9.51937675e-01 -6.59896731e-01
4.84469593e-01 2.11937994e-01 -8.39211345e-01 -2.28768010e-02
-7.27214754e-01 7.50380695e-01 5.33423901e-01 2.52500534e-01
-1.18938148e+00 -2.99377799e-01 -9.85561073e-01 -1.73791438e-01
9.50756133e-01 -7.42620170e-01 1.28210521e+00 -9.80858207e-01
-1.27989876e+00 9.55238700e-01 -9.43070929e-03 -3.91656309e-01
8.31278801e-01 -8.05211127e-01 -3.91422868e-01 2.74999350e-01
1.32965399e-02 4.35793877e-01 7.68450856e-01 -1.11151063e+00
-5.56287885e-01 -8.12591076e-01 4.91635054e-01 3.38954151e-01
2.50526750e-03 2.18712151e-01 -5.70214391e-01 -6.25321507e-01
-2.66821533e-01 -1.38067245e+00 1.83794186e-01 6.72536716e-02
-6.89768732e-01 -4.10496354e-01 1.26128304e+00 -3.38259071e-01
7.20865011e-01 -2.14584279e+00 3.67836416e-01 -3.61076415e-01
1.69984847e-01 3.63689393e-01 -4.08675000e-02 3.68895084e-01
-3.30367953e-01 -4.29940581e-01 5.24828397e-02 -3.61139208e-01
5.35954908e-02 -1.80580616e-01 4.83833104e-02 7.66849041e-01
-2.08411053e-01 1.04789948e+00 -1.15754247e+00 -4.24315065e-01
8.20906162e-01 3.72155160e-01 -4.26131487e-01 3.57973486e-01
-1.51698405e-05 8.31097782e-01 -3.97980630e-01 9.38769162e-01
5.88404775e-01 -1.29173715e-02 4.55222949e-02 -4.79688644e-01
-4.47611175e-02 -3.98078799e-01 -1.44898629e+00 1.73971629e+00
-3.46160889e-01 7.61614501e-01 6.55841082e-02 -9.05026913e-01
3.26381356e-01 4.68541145e-01 8.08595419e-01 -5.26199341e-01
2.94136971e-01 -2.64133841e-01 -1.10457882e-01 -1.18111694e+00
4.83084977e-01 5.70688128e-01 -2.04373762e-01 1.84354931e-01
5.30615561e-02 6.43684119e-02 5.81606567e-01 4.47059311e-02
1.30206299e+00 4.24189091e-01 5.07947028e-01 -3.57147236e-03
5.46871364e-01 8.74281973e-02 1.91881359e-01 6.27736092e-01
-7.44454622e-01 5.41296840e-01 2.99496621e-01 -4.30833668e-01
-8.01558733e-01 -1.23867524e+00 2.27110147e-01 1.04875720e+00
5.58982968e-01 -2.50629544e-01 -9.40051138e-01 -9.12035048e-01
-1.04820378e-01 6.69213176e-01 -5.92731297e-01 -8.72332677e-02
-7.95550287e-01 -1.96750373e-01 1.26627505e-01 7.85818934e-01
7.10036278e-01 -1.57130218e+00 -1.21864295e+00 -1.35640666e-01
-3.88970375e-01 -1.61912155e+00 -7.51691937e-01 -2.16190815e-01
-4.45528239e-01 -1.73718989e+00 -8.08851719e-01 -5.35562336e-01
5.58229625e-01 5.72562695e-01 1.07641685e+00 -7.09237039e-01
-7.65274704e-01 1.20339477e+00 -3.91375661e-01 -6.68551207e-01
2.82786608e-01 -3.71667892e-01 4.68324751e-01 5.07596791e-01
5.76634347e-01 -3.82918596e-01 -8.95314395e-01 8.14809501e-01
-6.74086630e-01 -1.15167545e-02 2.94366688e-01 3.09122235e-01
1.13264315e-01 -2.02503592e-01 -6.76066149e-03 -5.76702535e-01
1.52983800e-01 -4.18775380e-01 -3.57236773e-01 1.76599979e-01
4.60562795e-01 -7.30831265e-01 1.87754914e-01 -6.69153273e-01
-1.16045249e+00 5.82065761e-01 6.23180628e-01 -7.77237892e-01
-6.90933943e-01 -1.79235786e-01 -5.85759699e-01 2.77302504e-01
6.02738380e-01 -2.23055631e-01 -3.35267276e-01 -2.28149027e-01
4.07813728e-01 6.81755543e-01 7.73387969e-01 -3.05846721e-01
5.23302257e-01 7.32376099e-01 -1.63705245e-01 -1.15679240e+00
-5.75808764e-01 -8.81574452e-01 -1.27545202e+00 -7.46939063e-01
1.31423616e+00 -9.95196640e-01 -1.36671877e+00 8.36721182e-01
-1.38827026e+00 -4.22506452e-01 -4.51125145e-01 9.35601413e-01
-1.05918193e+00 4.69979912e-01 -1.68010160e-01 -1.00942814e+00
8.41960013e-02 -1.06705546e+00 1.36140513e+00 8.83824825e-02
-4.83049363e-01 -6.84787929e-01 1.61440894e-02 8.52110267e-01
-2.07178593e-01 5.37402809e-01 5.59996068e-02 -3.20287555e-01
-7.63922691e-01 -3.69572282e-01 2.34767888e-02 3.43698621e-01
2.27917254e-01 -3.52703154e-01 -9.36492443e-01 -1.76734060e-01
7.93225989e-02 -3.07730079e-01 4.04833853e-01 3.81417960e-01
1.23793423e+00 -1.08647585e-01 -4.78878796e-01 2.57640451e-01
6.65141165e-01 4.61441308e-01 7.34659076e-01 1.42816678e-01
9.16485369e-01 9.16326106e-01 1.09856045e+00 5.60269237e-01
1.63469985e-01 1.25758708e+00 6.34811521e-01 3.06225240e-01
-2.40701791e-02 1.74682304e-01 7.27257133e-01 3.70937169e-01
-5.15066624e-01 -5.04674375e-01 -9.67801750e-01 5.60716808e-01
-2.26964045e+00 -1.16359234e+00 -3.72184038e-01 1.91487706e+00
-2.60688942e-02 -1.21291392e-01 4.61131513e-01 3.87388319e-02
9.70737875e-01 6.06375076e-02 -7.41415024e-01 1.54763639e-01
2.67661691e-01 -3.69506180e-01 2.65831649e-01 -6.60048425e-02
-1.53562760e+00 6.92431390e-01 5.27276039e+00 2.43139789e-01
-6.98214889e-01 1.09206691e-01 -3.99771705e-02 -3.94773304e-01
1.07485449e+00 -3.65445912e-01 -6.54025972e-01 5.31712472e-01
5.46934843e-01 4.21071686e-02 1.49832532e-01 1.31465876e+00
4.76833344e-01 -4.02998596e-01 -1.65216899e+00 1.49953091e+00
2.96274215e-01 -6.39181376e-01 -2.37878218e-01 -5.38916234e-03
5.09856701e-01 -3.25707883e-01 -1.10097289e-01 2.93360144e-01
-5.63391447e-02 -7.83475578e-01 9.14218366e-01 8.26678932e-01
2.66947240e-01 -3.75971168e-01 5.05959511e-01 4.89447981e-01
-1.41713607e+00 -4.11183685e-01 -6.58391938e-02 -9.13274214e-02
5.69393456e-01 2.76047915e-01 -4.54951674e-01 4.12026316e-01
9.48765874e-01 1.05397451e+00 -4.45069045e-01 1.12461150e+00
-5.12250289e-02 3.41060758e-02 -2.86926446e-03 2.31837064e-01
2.42707804e-02 -3.88278991e-01 9.21070516e-01 1.03978550e+00
2.00343534e-01 8.41136798e-02 2.11808369e-01 7.16315806e-01
2.21222416e-01 -2.18174532e-01 -1.03681707e+00 -9.05655995e-02
-3.61754633e-02 1.16190863e+00 -7.09340036e-01 -3.60596359e-01
-2.35842973e-01 1.32792318e+00 -1.37897268e-01 4.66550171e-01
-1.12942445e+00 -4.31063026e-01 7.83611178e-01 2.97105819e-01
3.43692899e-01 -5.25383353e-01 4.09051925e-01 -1.44358790e+00
3.59626055e-01 -7.94943690e-01 2.80709118e-01 -1.26184297e+00
-8.68411005e-01 2.60326266e-01 5.73723376e-01 -1.50935364e+00
-3.63603085e-01 -9.64385390e-01 -6.09133899e-01 3.34462017e-01
-5.27728081e-01 -1.23820591e+00 -1.08277500e+00 8.93823624e-01
1.14542329e+00 -2.47792408e-01 3.04139763e-01 4.11519080e-01
-8.09659481e-01 2.55712448e-03 -3.82752776e-01 2.42511466e-01
7.24262297e-01 -1.05696833e+00 2.69738764e-01 6.57360792e-01
1.05240114e-01 6.16096914e-01 6.53318703e-01 -5.98180890e-01
-1.63843369e+00 -1.08084881e+00 4.04282421e-01 -1.31287110e+00
4.14106101e-01 -7.73511767e-01 -3.09276342e-01 1.13804114e+00
7.67874345e-02 4.53136027e-01 1.85475573e-01 -3.93312246e-01
7.25126639e-02 2.88068559e-02 -9.85303998e-01 6.80826008e-01
1.65332472e+00 -3.10034841e-01 -7.13871658e-01 6.97834313e-01
3.46748590e-01 -3.52859437e-01 -5.22761285e-01 4.12823379e-01
8.07413638e-01 -9.76252139e-01 1.25219262e+00 -6.35007560e-01
2.26923361e-01 -6.29142344e-01 -2.49337301e-01 -1.00495934e+00
-1.52859643e-01 -4.28237140e-01 -6.97482526e-01 9.45160031e-01
-5.38557947e-01 -3.24283779e-01 7.55420566e-01 3.73703331e-01
-1.04949251e-01 -3.33706677e-01 -7.43289948e-01 -1.07374835e+00
-6.64748073e-01 -5.31052351e-01 1.86625496e-01 5.33934832e-01
9.15221423e-02 1.32987931e-01 -4.26551282e-01 -6.50093630e-02
8.39775085e-01 -2.79302746e-01 1.44945955e+00 -1.10697222e+00
-8.94870982e-02 -1.00868464e-01 -7.50398517e-01 -1.06059432e+00
1.86158419e-01 -4.23863769e-01 4.38063405e-02 -1.24236381e+00
2.91753113e-01 2.31830016e-01 5.83753400e-02 4.85668778e-02
1.29592419e-01 4.10049945e-01 3.09903890e-01 9.70594883e-02
-1.04731047e+00 4.82715279e-01 1.10532200e+00 -3.45316559e-01
3.20484005e-02 1.37617588e-01 1.90003633e-01 9.15287495e-01
5.59411645e-01 -1.83383599e-01 -5.20706892e-01 -1.18171766e-01
-1.91517904e-01 7.61063630e-03 1.05398715e+00 -1.50997901e+00
3.73022616e-01 -2.40693897e-01 4.00405407e-01 -8.67539942e-01
7.94000804e-01 -9.61797714e-01 3.66627157e-01 5.47605217e-01
-1.58015281e-01 5.84432110e-02 4.44168076e-02 7.48082697e-01
-7.63520151e-02 -8.20085257e-02 6.56786621e-01 -3.42150748e-01
-1.06569195e+00 2.87113667e-01 -4.73715931e-01 -1.41130490e-02
1.88325000e+00 -4.31443810e-01 -2.71768183e-01 -1.65987968e-01
-1.04887283e+00 1.68900922e-01 3.35794330e-01 9.11359787e-01
6.21426284e-01 -1.33386242e+00 -4.29418176e-01 6.62609115e-02
5.13723731e-01 -1.18666269e-01 5.96493661e-01 1.17018735e+00
-5.51497340e-01 6.86420023e-01 -5.05204439e-01 -9.08610642e-01
-1.61756074e+00 9.59813654e-01 2.47986779e-01 1.14945486e-01
-8.47326636e-01 4.69928801e-01 8.60154510e-01 -3.47133607e-01
5.09829462e-01 -2.54443318e-01 -3.31549495e-01 1.02337547e-01
5.43417394e-01 1.03350532e+00 -1.63397148e-01 -8.43911052e-01
-5.11163354e-01 6.95949078e-01 2.58066386e-01 1.18725948e-01
9.96517479e-01 -1.54817000e-01 2.07533881e-01 6.51106596e-01
9.41831887e-01 -3.72126669e-01 -1.37980247e+00 -1.09900604e-03
-7.27270097e-02 -7.33398438e-01 -3.58281881e-01 -5.58267415e-01
-9.07021046e-01 8.48950028e-01 7.95351326e-01 1.69460177e-02
6.68438494e-01 1.22621596e-01 5.01742303e-01 7.39087760e-01
7.53569722e-01 -1.17943347e+00 8.64096522e-01 5.03131688e-01
1.33724642e+00 -1.30490088e+00 -4.88503091e-02 -6.36337101e-01
-9.24742162e-01 9.35274422e-01 9.70382571e-01 -1.53207049e-01
6.29258096e-01 1.35554897e-03 -2.48198986e-01 -4.46651578e-01
-5.07043004e-01 -1.82382211e-01 2.67255753e-01 6.46038711e-01
3.76990288e-02 1.03316896e-01 2.46262893e-01 4.39434558e-01
6.51506335e-02 7.52999708e-02 3.01268846e-01 9.91146028e-01
1.13137290e-01 -5.38516462e-01 -5.91818333e-01 1.68371294e-02
-1.24731258e-01 7.03636706e-01 -5.13677597e-01 9.53735828e-01
3.50812346e-01 1.13568640e+00 3.42227221e-01 -9.78972912e-02
8.62747550e-01 -7.82271102e-02 7.47270107e-01 -5.98366141e-01
-4.43818539e-01 -4.08174843e-01 -3.39027196e-02 -1.10383224e+00
-6.02344930e-01 -9.66717184e-01 -7.77216971e-01 -9.90033336e-03
-7.64262453e-02 -2.82002658e-01 6.53694153e-01 7.87509382e-01
2.59980500e-01 7.26349533e-01 2.47642577e-01 -1.69445467e+00
-1.68252766e-01 -1.04964805e+00 -6.18195772e-01 8.05810928e-01
1.61217198e-01 -1.16268039e+00 -3.03959608e-01 3.25146377e-01]
|
[8.027946472167969, 0.35835838317871094]
|
44a0b2b6-cb07-4b5e-9944-58cdae3f85b7
|
leveraging-gpt-4-for-automatic-translation
|
2305.14878
| null |
https://arxiv.org/abs/2305.14878v1
|
https://arxiv.org/pdf/2305.14878v1.pdf
|
Leveraging GPT-4 for Automatic Translation Post-Editing
|
While Neural Machine Translation (NMT) represents the leading approach to Machine Translation (MT), the outputs of NMT models still require translation post-editing to rectify errors and enhance quality, particularly under critical settings. In this work, we formalize the task of translation post-editing with Large Language Models (LLMs) and explore the use of GPT-4 to automatically post-edit NMT outputs across several language pairs. Our results demonstrate that GPT-4 is adept at translation post-editing and produces meaningful edits even when the target language is not English. Notably, we achieve state-of-the-art performance on WMT-22 English-Chinese, English-German, Chinese-English and German-English language pairs using GPT-4 based post-editing, as evaluated by state-of-the-art MT quality metrics.
|
['Arul Menezes', 'Hany Hassan Awadallah', 'Amr Sharaf', 'Vikas Raunak']
|
2023-05-24
| null | null | null | null |
['nmt']
|
['computer-code']
|
[ 6.01463079e-01 1.42443195e-01 -1.98447585e-01 -2.99445629e-01
-1.40787828e+00 -7.06492126e-01 7.49978125e-01 1.12883244e-02
-5.49100757e-01 1.00399125e+00 2.47369185e-01 -8.42681348e-01
4.15330172e-01 -4.36293960e-01 -1.05079341e+00 2.70152360e-01
3.88008296e-01 9.82568145e-01 -4.79693204e-01 -7.06635535e-01
1.48570284e-01 -6.24723434e-02 -5.60320079e-01 6.54378176e-01
1.60198164e+00 2.34704033e-01 1.98598653e-01 8.09893310e-01
-1.47333413e-01 2.88671315e-01 -5.41779876e-01 -1.32259107e+00
4.13395017e-01 -7.23525167e-01 -9.83406782e-01 -4.32963908e-01
5.97398221e-01 -5.72887324e-02 1.19648306e-02 9.97056007e-01
7.25987315e-01 -1.83730349e-01 5.83348930e-01 -9.87282336e-01
-1.43883455e+00 1.15404534e+00 -2.88646430e-01 3.19400169e-02
3.50881815e-01 2.65677154e-01 9.01119113e-01 -1.52735937e+00
9.06274021e-01 1.38541448e+00 9.36986327e-01 6.40025795e-01
-1.33977771e+00 -4.84265029e-01 -3.62529099e-01 -6.20870255e-02
-1.06885684e+00 -7.25656569e-01 -3.32872123e-02 -1.52097881e-01
1.78937602e+00 2.87713230e-01 3.00212234e-01 1.36973321e+00
7.49695361e-01 5.24831355e-01 1.10046375e+00 -8.53211462e-01
-3.22945267e-01 -3.80781665e-02 -4.29432899e-01 4.25021470e-01
1.34620264e-01 1.71617493e-01 -8.25109482e-01 1.81107633e-02
3.87104034e-01 -6.30684376e-01 -2.75251251e-02 4.98613775e-01
-1.86521041e+00 5.21904826e-01 -1.81899238e-02 3.89188111e-01
-4.21535671e-01 2.06129089e-01 5.57253718e-01 1.09051526e+00
8.45271409e-01 9.91394997e-01 -8.07878196e-01 -6.52519286e-01
-1.11658812e+00 5.56819998e-02 5.72893143e-01 1.41631889e+00
5.75161099e-01 1.11247394e-02 -5.65504014e-01 1.05873382e+00
-1.01474524e-01 8.72476876e-01 4.77915227e-01 -6.68812633e-01
1.30765545e+00 2.88085043e-01 3.19363624e-02 -3.48466784e-01
1.12228051e-01 -5.27859569e-01 -9.34782505e-01 -2.49232426e-01
9.24811214e-02 -3.21140707e-01 -7.57911682e-01 1.79415035e+00
-2.80057728e-01 -4.21113163e-01 2.90449411e-01 4.21011508e-01
4.34985697e-01 7.72522509e-01 -1.49506554e-01 -3.62440735e-01
8.76918077e-01 -1.45670581e+00 -8.67731273e-01 -5.38400352e-01
1.02836382e+00 -1.45999491e+00 1.38250077e+00 -5.81499636e-02
-1.71802890e+00 -6.50626659e-01 -7.78954864e-01 -4.11523461e-01
-2.43793309e-01 3.81499290e-01 1.85085181e-02 2.59950966e-01
-1.46306050e+00 9.17006612e-01 -6.76045001e-01 -6.15019202e-01
7.30484128e-02 3.03533822e-01 -6.35922909e-01 -6.80969730e-02
-1.37193859e+00 1.70530653e+00 2.51697332e-01 1.62450656e-01
-7.06638217e-01 -9.21791136e-01 -6.66026831e-01 -2.42446691e-01
-1.61102310e-01 -1.15516710e+00 1.79622555e+00 -1.15513754e+00
-1.67545247e+00 9.83590305e-01 -5.15625000e-01 -4.36408609e-01
1.03913772e+00 -5.06819546e-01 -5.82889020e-01 -4.34959620e-01
3.39160651e-01 8.94991517e-01 5.93360782e-01 -7.83537686e-01
-6.03008986e-01 1.25318328e-02 -4.76009876e-01 4.46391076e-01
-2.25579679e-01 5.78907669e-01 -3.30193132e-01 -8.99862826e-01
-2.37285405e-01 -1.08009887e+00 -1.80066273e-01 -4.24656391e-01
-6.56166136e-01 2.69683879e-02 4.02010113e-01 -1.34980309e+00
1.29235673e+00 -1.64489198e+00 5.50779045e-01 -2.27213249e-01
-2.05830842e-01 4.10493135e-01 -7.07328558e-01 9.09576178e-01
1.34396955e-01 4.79316622e-01 -5.27731121e-01 -9.48438585e-01
1.59857392e-01 1.73386008e-01 -2.16776729e-01 -5.70610678e-03
5.98579526e-01 1.74057472e+00 -8.61510992e-01 -3.59996587e-01
-1.87547162e-01 2.98668504e-01 -3.69735450e-01 1.09859370e-01
-3.34178925e-01 5.71717799e-01 3.07456672e-01 6.05137706e-01
3.91054392e-01 8.71006325e-02 7.46729597e-02 2.81197131e-01
-1.54613033e-01 7.54215002e-01 -4.20137525e-01 2.07693672e+00
-6.56355441e-01 9.57339168e-01 -1.96414948e-01 -2.07314029e-01
7.95648873e-01 4.83490765e-01 -1.55259922e-01 -9.15958345e-01
-2.51429025e-02 8.50740850e-01 1.73455194e-01 -1.56804949e-01
1.10039055e+00 -5.79615608e-02 2.66048908e-02 7.44229615e-01
8.37074369e-02 -3.71921957e-01 4.05065119e-01 4.77480441e-02
8.43708158e-01 2.54351735e-01 8.75916556e-02 -2.99549311e-01
9.09494013e-02 3.04844022e-01 4.31303263e-01 7.11125314e-01
2.45164692e-01 6.01275384e-01 -8.27805698e-02 -1.32401749e-01
-1.65685928e+00 -9.51720595e-01 2.49357119e-01 1.05938494e+00
-4.81891602e-01 -5.30483782e-01 -1.05533862e+00 -5.85397720e-01
-1.74422309e-01 1.15518141e+00 -3.80675405e-01 -3.05234313e-01
-1.03545940e+00 -5.00253439e-01 1.12536669e+00 2.87417114e-01
3.55807245e-01 -1.15089428e+00 5.04165143e-02 6.07293367e-01
-8.10131967e-01 -1.15038466e+00 -1.14095366e+00 1.52237359e-02
-1.20533323e+00 -3.78897518e-01 -8.58670890e-01 -1.09785330e+00
6.36765957e-01 -1.62268784e-02 1.63250971e+00 5.01169264e-02
3.45000565e-01 -1.46217212e-01 -3.18962455e-01 -4.07682776e-01
-1.46794593e+00 5.94792366e-01 3.35582137e-01 -5.29619157e-01
1.96682960e-01 -5.24365604e-01 -5.16227372e-02 2.44495183e-01
-6.21508479e-01 6.19898617e-01 8.66836727e-01 8.03717971e-01
5.28234839e-01 -6.47562087e-01 4.72840756e-01 -7.55515277e-01
1.17392945e+00 4.73539196e-02 -2.26983488e-01 7.44575739e-01
-8.79559100e-01 6.69736862e-02 7.28478253e-01 -5.49801350e-01
-8.95801604e-01 -6.11421585e-01 -1.12883560e-02 -7.68107250e-02
4.31142390e-01 7.31148958e-01 -2.91736796e-02 1.17019564e-02
8.81683707e-01 4.10951942e-01 -1.83705449e-01 -5.93223035e-01
5.73339403e-01 8.27206373e-01 6.96712554e-01 -6.15684152e-01
8.86585772e-01 -3.77177507e-01 -3.26477021e-01 -1.51076302e-01
-2.28188008e-01 3.27365071e-01 -9.60096538e-01 2.30846316e-01
6.16311729e-01 -9.83598113e-01 1.06533051e-01 5.06055295e-01
-1.71515489e+00 -5.92999041e-01 -1.14095084e-01 3.77947688e-01
-5.26696622e-01 2.91483939e-01 -1.23847830e+00 -1.62747517e-01
-1.27448618e+00 -1.38467681e+00 1.10860991e+00 -2.51321197e-01
-6.77329004e-01 -8.56206596e-01 2.05695897e-01 4.14089322e-01
7.49085844e-01 -2.37682536e-01 1.25679219e+00 -4.11270529e-01
-3.38328540e-01 2.14469973e-02 -1.44786447e-01 4.61156189e-01
1.97572127e-01 1.50707066e-01 -3.54785264e-01 -4.81042415e-01
-5.87362468e-01 -8.92611817e-02 4.54279184e-01 1.52116328e-01
2.64436185e-01 -4.70768183e-01 -5.39931767e-02 5.81803203e-01
1.03537667e+00 -6.34122789e-02 6.43951476e-01 4.63382840e-01
6.53555930e-01 4.16973591e-01 6.84746742e-01 -1.05313890e-01
5.02819479e-01 7.36852586e-01 -1.13765374e-01 -2.70875603e-01
-3.73764724e-01 -6.80027843e-01 9.11691725e-01 1.72379017e+00
-5.30951917e-02 -3.61549795e-01 -8.61425400e-01 6.40803576e-01
-1.80849159e+00 -4.80961651e-01 -4.23238099e-01 2.06187701e+00
1.42596793e+00 1.00308597e-01 -3.20452750e-01 -4.11788434e-01
7.63668120e-01 -4.45065230e-01 -2.99277604e-01 -1.11262202e+00
-5.13515949e-01 4.09170419e-01 5.45087218e-01 6.92080379e-01
-4.99143958e-01 1.58019638e+00 7.06803417e+00 7.04874873e-01
-9.69040275e-01 5.34403026e-01 5.22075295e-01 -2.52116509e-02
-5.23068488e-01 1.74496442e-01 -6.10037982e-01 3.90078783e-01
1.29103422e+00 -3.69069815e-01 9.25636649e-01 3.22647095e-01
4.43507731e-01 3.40563864e-01 -1.51220870e+00 7.33608305e-01
-1.60456493e-01 -1.45844126e+00 4.84397918e-01 1.09996065e-01
1.27404547e+00 4.84636724e-01 4.80569974e-02 4.77025568e-01
4.63598430e-01 -1.12620473e+00 1.05284977e+00 3.97252470e-01
1.39769423e+00 -6.43599868e-01 5.79407930e-01 3.91268492e-01
-6.93951905e-01 3.98284823e-01 -4.31225210e-01 -6.60093427e-02
4.26044345e-01 6.31594360e-01 -9.27648485e-01 7.73205519e-01
3.93010348e-01 7.12555766e-01 -6.47174597e-01 5.67166269e-01
-4.21798974e-01 6.07525885e-01 2.99842339e-02 1.99198157e-01
1.71110243e-01 -2.58597821e-01 6.97592378e-01 1.69517875e+00
9.11538780e-01 -5.88011146e-01 -1.99462414e-01 7.95404136e-01
-7.61369705e-01 3.67542565e-01 -5.19003749e-01 -4.98798937e-01
6.25392497e-01 8.81822884e-01 -4.84531792e-03 -6.28360808e-01
-5.81256673e-02 1.68036783e+00 5.90609610e-01 3.43112022e-01
-7.05021501e-01 -3.50909889e-01 8.70605826e-01 -2.08899811e-01
-1.13837294e-01 -4.75664556e-01 -5.60669959e-01 -1.34473073e+00
3.60384524e-01 -1.54320264e+00 -1.83867380e-01 -1.07526696e+00
-1.14736426e+00 9.84322429e-01 -3.82268846e-01 -1.20494497e+00
-5.23786068e-01 -3.38085085e-01 -5.25963604e-01 1.35247242e+00
-1.34344196e+00 -1.49654841e+00 3.88838530e-01 1.29182428e-01
8.03215861e-01 -3.46210063e-01 9.98297513e-01 4.91672516e-01
-2.93612659e-01 1.09512174e+00 4.79850799e-01 1.53433174e-01
1.24135518e+00 -1.09644485e+00 1.59852314e+00 1.25203240e+00
2.62371242e-01 8.35416138e-01 7.15587735e-01 -1.03934968e+00
-1.42335367e+00 -1.39094973e+00 2.02073336e+00 -9.14944947e-01
4.99233454e-01 -4.27650809e-01 -6.34094000e-01 8.34512115e-01
8.95241380e-01 -6.01658165e-01 4.02505845e-01 -1.64236501e-02
-2.92793810e-01 3.99357110e-01 -9.40609217e-01 1.00092399e+00
1.42169118e+00 -8.63175631e-01 -6.96976244e-01 4.18136984e-01
1.26772201e+00 -6.96558416e-01 -1.11960018e+00 4.18259382e-01
4.20294434e-01 -3.20445657e-01 6.49555922e-01 -9.47570443e-01
8.54114175e-01 -1.95435941e-01 -1.98302016e-01 -2.11539960e+00
-3.75543922e-01 -1.21819293e+00 1.58832803e-01 1.30915678e+00
1.15912700e+00 -5.74806929e-01 1.73179448e-01 1.76660806e-01
-5.94967425e-01 -4.63434160e-01 -1.11104739e+00 -9.12299514e-01
6.54805303e-01 -3.82356972e-01 8.08254302e-01 1.16410077e+00
-3.82231921e-02 5.78849852e-01 -7.12287962e-01 -2.46361077e-01
2.22990215e-01 -2.85125852e-01 8.65463078e-01 -7.94428647e-01
-4.99517947e-01 -5.87561786e-01 3.09624314e-01 -8.23515296e-01
1.47288650e-01 -1.35388386e+00 9.59674716e-02 -1.82591403e+00
2.62456477e-01 -1.17015049e-01 2.10114941e-01 6.20561600e-01
-4.38306987e-01 4.01250690e-01 3.55821252e-01 4.93111134e-01
-1.98050484e-01 5.84245741e-01 1.51179612e+00 -2.57231832e-01
-1.67548627e-01 -3.06119561e-01 -7.28957295e-01 -1.42823875e-01
8.34288836e-01 -7.29528725e-01 4.56860475e-02 -1.42974079e+00
5.57031870e-01 -5.91195673e-02 -1.87453330e-01 -5.96753180e-01
1.04264580e-01 -1.36343449e-01 1.90238595e-01 -3.91422153e-01
4.15985696e-02 -2.69901603e-01 3.92655015e-01 3.66151869e-01
-6.07968688e-01 1.16561389e+00 4.07938331e-01 -7.73200169e-02
-1.68279782e-01 2.82165017e-02 4.43803161e-01 -1.59433901e-01
-4.70822081e-02 3.00766025e-02 -4.38007325e-01 2.59825408e-01
2.41977483e-01 -1.06748976e-01 -6.21283591e-01 -2.98455536e-01
-1.90038502e-01 2.52231389e-01 7.84060955e-01 7.75221407e-01
3.39113802e-01 -1.52275097e+00 -1.56272542e+00 1.47264495e-01
2.77080774e-01 -4.08474445e-01 -3.09425563e-01 1.02847731e+00
-5.32869041e-01 5.17455757e-01 -2.82841235e-01 -3.87507915e-01
-1.23634994e+00 2.39806063e-02 3.08376610e-01 -5.71362436e-01
-3.17793012e-01 7.55493879e-01 -5.45703590e-01 -1.22594142e+00
-1.80642933e-01 -4.58866626e-01 6.48914993e-01 -4.81619805e-01
3.18360865e-01 2.47937113e-01 6.25007808e-01 -7.23705649e-01
-3.01400404e-02 3.49156916e-01 -2.66122848e-01 -6.43221557e-01
1.02228367e+00 -2.32624233e-01 -5.81095576e-01 1.78695157e-01
9.96560097e-01 1.85272042e-02 -5.43298125e-01 -4.67456698e-01
2.90777475e-01 -2.94744849e-01 -3.80752265e-01 -1.44226980e+00
-3.13594401e-01 1.04700899e+00 2.65104741e-01 -6.37203634e-01
8.26506555e-01 -5.12756526e-01 1.26941502e+00 6.88005149e-01
4.95822430e-01 -1.44779372e+00 -2.97170073e-01 1.22273397e+00
1.02784884e+00 -1.26325393e+00 -5.01064777e-01 -2.18497906e-02
-6.17477596e-01 9.96445894e-01 4.42203879e-01 5.14570177e-01
-1.47033751e-01 9.88746285e-02 3.98363620e-01 5.36388755e-01
-1.05542171e+00 2.80733377e-01 4.95416462e-01 5.04643857e-01
8.88205349e-01 4.23134536e-01 -4.49603558e-01 2.80460358e-01
-6.58928633e-01 5.53273559e-02 3.69287312e-01 8.60317945e-01
3.42846513e-02 -1.70346034e+00 -2.47346684e-01 3.30087841e-01
-5.96052229e-01 -1.01235735e+00 -9.34306979e-01 4.71748769e-01
-2.10350290e-01 1.15179479e+00 -2.19805464e-01 -6.44954324e-01
4.40844506e-01 2.67273843e-01 6.49671137e-01 -6.57371283e-01
-1.34221113e+00 -7.10065886e-02 4.50097203e-01 -2.90866017e-01
1.18974425e-01 -5.09162903e-01 -9.73657131e-01 -8.27814877e-01
-2.49426380e-01 2.15642691e-01 8.80426109e-01 9.16889668e-01
9.10236478e-01 3.29020172e-01 2.71886855e-01 -6.36952281e-01
-7.17187881e-01 -1.55481267e+00 2.88908631e-01 3.34072828e-01
4.79614176e-02 1.03666671e-01 9.41815004e-02 1.71038166e-01]
|
[11.684884071350098, 10.270073890686035]
|
3d061656-0c29-410f-92f0-b0f8b54d808c
|
a-health-monitoring-system-for-elder-and-sick
|
1304.4652
| null |
http://arxiv.org/abs/1304.4652v1
|
http://arxiv.org/pdf/1304.4652v1.pdf
|
A Health Monitoring System for Elder and Sick Persons
|
This paper discusses a vision based health monitoring system which would be
very easy in use and deployment. Elder and sick people who are not able to talk
or walk they are dependent on other human beings for their daily needs and need
continuous monitoring. The developed system provides facility to the sick or
elder person to describe his or her need to their caretaker in lingual
description by showing particular hand gesture with the developed system. This
system uses fingertip detection technique for gesture extraction and artificial
neural network for gesture classification and recognition. The system is able
to work in different light conditions and can be connected to different devices
to announce users need on a distant location.
|
['Jagdish L. Raheja', 'Ankit Chaudhary']
|
2013-04-17
| null | null | null | null |
['fingertip-detection']
|
['computer-vision']
|
[-2.53738463e-01 -4.12486255e-01 -2.91867852e-01 -5.85373640e-01
1.79715604e-01 -4.61774826e-01 6.26632422e-02 -4.99770567e-02
-8.41579497e-01 8.37862968e-01 1.74492896e-01 -2.34400243e-01
1.39381826e-01 -5.40166199e-01 5.62784910e-01 -5.05870104e-01
3.39722298e-02 6.92726493e-01 3.41184884e-01 -2.56790578e-01
-1.83117285e-01 7.44465351e-01 -1.37277246e+00 -3.08141997e-03
-3.66857159e-03 5.34965694e-01 4.10749137e-01 9.73929703e-01
1.90109789e-01 2.34487295e-01 -6.47643924e-01 4.73325521e-01
1.51257530e-01 -2.02972010e-01 -4.68116701e-01 1.17102936e-01
-6.49550259e-02 -1.25244308e+00 2.14890856e-02 6.14063859e-01
1.23032784e+00 4.11322676e-02 7.77562976e-01 -1.10784733e+00
-1.64294884e-01 -8.52762759e-02 -3.01076025e-01 3.33301097e-01
8.85882735e-01 1.54713586e-01 1.05795123e-01 -5.25687277e-01
2.08725005e-01 1.10108876e+00 6.33090973e-01 1.01057148e+00
-4.12146986e-01 -6.30156100e-01 -1.13949664e-01 3.79400045e-01
-1.29193389e+00 -4.51443642e-01 4.85432804e-01 -2.48671964e-01
1.15618527e+00 3.58805686e-01 6.59003615e-01 8.92518342e-01
-1.76684231e-01 4.50064927e-01 9.41832006e-01 -7.37574577e-01
2.54765749e-01 3.18971165e-02 9.43967760e-01 7.90528297e-01
6.15866721e-01 -2.33247891e-01 -2.04618365e-01 -1.53370380e-01
7.68365622e-01 6.12631202e-01 -5.44091165e-01 2.01684088e-01
-1.23201489e+00 4.59084362e-01 3.48651260e-01 9.96099114e-01
-5.92266083e-01 -2.90172622e-02 2.78842181e-01 2.46625304e-01
-5.04060209e-01 -4.17826533e-01 -7.99416661e-01 -4.08478022e-01
-7.05219388e-01 -1.30962551e-01 1.05448174e+00 8.78629088e-01
-3.52456003e-01 -4.03362602e-01 -5.25831617e-02 4.20833409e-01
7.93094456e-01 7.68010020e-01 8.16321731e-01 -5.27830303e-01
2.82265723e-01 6.10695362e-01 5.62426448e-01 -2.06161648e-01
-1.08105135e+00 6.50860250e-01 -6.85881615e-01 8.24246585e-01
6.20859742e-01 -5.87851405e-01 -1.08270788e+00 7.75766850e-01
3.85929585e-01 -5.93909562e-01 5.05382195e-02 7.26544380e-01
1.27573216e+00 2.09488302e-01 3.30057502e-01 -4.96803045e-01
2.03522062e+00 -4.28127557e-01 -1.02750075e+00 3.22399512e-02
2.21989527e-01 -5.62916517e-01 9.03306127e-01 4.33968723e-01
-4.97319579e-01 -3.75502884e-01 -1.10206771e+00 1.58681720e-01
-4.97334898e-01 4.97081369e-01 6.87307894e-01 1.23144507e+00
-1.03089356e+00 3.30573916e-01 -1.33696115e+00 -1.57530010e+00
3.55304837e-01 1.00521183e+00 -2.64272720e-01 2.32859224e-01
-6.39542222e-01 1.10036957e+00 3.22779477e-01 1.87122613e-01
9.90183353e-02 3.18586916e-01 -6.55387163e-01 -4.71027702e-01
-4.60029989e-01 -5.97136974e-01 1.09612739e+00 -5.31414568e-01
-1.39878392e+00 8.56995821e-01 -2.92284518e-01 9.77635682e-02
7.21993566e-01 -1.62633777e-01 -6.58821046e-01 1.55483007e-01
-6.17163479e-02 3.93247277e-01 4.63462353e-01 -6.52954221e-01
-6.86579466e-01 -1.17406344e+00 -3.15427959e-01 5.34146309e-01
-2.84362972e-01 4.80859250e-01 -1.13945305e-01 -3.62796396e-01
3.97773564e-01 -6.40095294e-01 5.69208860e-01 4.54345524e-01
-2.09793732e-01 -3.55504245e-01 1.50997078e+00 -9.72992778e-01
9.85818386e-01 -1.50195360e+00 -4.35883015e-01 1.35803390e-02
-1.00068279e-01 5.77331066e-01 5.77546358e-01 4.25514758e-01
3.80359352e-01 -1.91542447e-01 8.78627449e-02 -3.00038066e-02
-1.12804644e-01 5.40713608e-01 8.22208405e-01 4.25067812e-01
-6.29135609e-01 3.95732254e-01 -7.05156386e-01 -7.80552149e-01
5.40784895e-01 9.33111727e-01 3.43590379e-01 2.40499660e-01
5.77964008e-01 8.81949484e-01 -9.08953369e-01 1.26752174e+00
2.14405522e-01 1.55172035e-01 1.91322446e-01 -1.79077595e-01
4.62238193e-02 -3.51391464e-01 -1.13016200e+00 1.18739390e+00
-2.87545949e-01 7.49221921e-01 3.01404059e-01 -7.62825191e-01
7.97293723e-01 8.95129263e-01 1.89703971e-01 -1.57022700e-01
6.03004575e-01 1.28814191e-01 -3.85717809e-01 -1.62846804e+00
-3.61470044e-01 -1.82527557e-01 5.32416344e-01 7.24216819e-01
-5.96905768e-01 6.57540083e-01 -1.59167141e-01 -5.25930882e-01
1.11431718e+00 4.15648192e-01 1.00480020e+00 1.10334516e-01
5.15257359e-01 -3.14173222e-01 -2.95129698e-02 4.11550164e-01
-9.01031613e-01 8.90088975e-02 -5.55361092e-01 -5.97170770e-01
-6.55647278e-01 -9.20489728e-01 -1.85311347e-01 1.38760483e+00
-2.73407787e-01 3.86690974e-01 -4.88142997e-01 -5.96707284e-01
-2.22309038e-01 1.69492260e-01 -2.83304960e-01 5.78117430e-01
-6.96299314e-01 -5.83718479e-01 4.06457901e-01 8.88515592e-01
1.05772352e+00 -1.63241148e+00 -1.62935328e+00 2.08566830e-01
-4.45456197e-03 -6.86381400e-01 -7.24823773e-02 2.28141963e-01
-1.09194589e+00 -9.51467991e-01 -1.43020940e+00 -1.27311015e+00
8.55646431e-01 -1.72463469e-02 5.52597046e-01 3.28031927e-01
-3.45517486e-01 7.41687596e-01 -7.99224734e-01 -6.02478921e-01
1.66025996e-01 5.70204556e-02 1.73411414e-01 -5.02931893e-01
1.13243854e+00 -8.27906787e-01 -7.36360013e-01 4.03287053e-01
-1.67865574e-01 -6.13156736e-01 -1.79366078e-02 3.79758239e-01
-2.44494483e-01 -1.45796880e-01 7.61397108e-02 -1.59601599e-01
9.81080353e-01 -9.53434110e-02 -5.73862903e-02 5.53498149e-01
-4.10766482e-01 1.15027584e-01 -1.86124057e-01 -3.84749889e-01
-7.39700437e-01 6.35413766e-01 -1.22947402e-01 6.11512363e-01
-8.78803849e-01 -1.61381915e-01 -5.43614388e-01 -6.22172020e-02
4.10306811e-01 -6.25511855e-02 -2.27848932e-01 -8.89855981e-01
-2.42185354e-01 1.70421398e+00 7.96401560e-01 3.60524617e-02
3.69608492e-01 4.74210382e-01 -1.95759848e-01 -1.27868152e+00
1.91523716e-01 -6.16656065e-01 -1.50291193e+00 -5.00247419e-01
1.26406205e+00 -3.51702541e-01 -1.18491626e+00 6.95179045e-01
-1.25795949e+00 -1.79752141e-01 6.13531232e-01 6.33612394e-01
4.27936539e-02 2.94559687e-01 -1.92662999e-01 -1.44830954e+00
-8.33217204e-01 -5.94997048e-01 8.77100945e-01 7.12749779e-01
-7.68909335e-01 -1.02525854e+00 -3.45606506e-01 2.27274954e-01
5.13067842e-01 6.65842414e-01 4.59052861e-01 -4.97269303e-01
3.56142431e-01 -8.27688992e-01 -1.73265100e-01 -1.45579562e-01
8.85795891e-01 -2.50340968e-01 -8.94162893e-01 -1.52761564e-01
2.36439984e-02 3.12430188e-02 3.75169307e-01 7.10718751e-01
3.63763720e-01 -3.52158099e-01 -7.87313700e-01 1.02103598e-01
1.38391769e+00 7.94587791e-01 5.03427565e-01 5.14297009e-01
6.69570923e-01 4.89347160e-01 7.39888176e-02 5.66313863e-01
4.77566332e-01 8.51077735e-01 9.78956223e-02 2.20710173e-01
4.15159538e-02 4.89501327e-01 3.27931345e-01 7.21716210e-02
-9.33848262e-01 -1.77808344e-01 -9.34534669e-01 6.08689375e-02
-1.84666574e+00 -9.65627551e-01 -4.86407459e-01 2.22362900e+00
3.53711933e-01 -3.41232359e-01 6.98809564e-01 8.34953010e-01
9.17965472e-01 -6.00830793e-01 -5.54316163e-01 -2.14222997e-01
5.42392731e-01 4.20627505e-01 7.19832122e-01 4.88894701e-01
-1.18233299e+00 5.05100369e-01 6.23663807e+00 -4.48243231e-01
-1.13675630e+00 4.15450424e-01 7.76823312e-02 -5.44377156e-02
8.19829226e-01 -7.05447197e-01 -8.06314707e-01 4.98570204e-01
5.14581501e-01 4.79708135e-01 2.75334746e-01 7.50412107e-01
7.81087995e-01 -4.82131571e-01 -1.23791039e+00 1.50109696e+00
-2.21089367e-03 -2.38916874e-01 -5.50151110e-01 -1.86355352e-01
-1.59057528e-01 -2.31171981e-01 -6.76437616e-01 -1.63424432e-01
-2.14266777e-01 -1.02760816e+00 2.29189485e-01 6.29810035e-01
7.91235626e-01 -3.35437208e-01 8.43971372e-01 4.17240530e-01
-1.29194641e+00 -1.08100034e-01 1.33367598e-01 -6.29124522e-01
4.50277597e-01 -5.19019306e-01 -9.17669594e-01 -5.52285194e-01
9.39371884e-01 6.47649541e-02 -2.12720677e-01 9.37752962e-01
-1.78926706e-01 1.79941565e-01 -8.68168771e-01 -7.52737105e-01
-8.03967267e-02 1.98460847e-01 1.99773982e-01 1.22526193e+00
6.62982285e-01 6.62390471e-01 -1.42857715e-01 -2.84784976e-02
5.40004611e-01 1.29222170e-01 -6.07459188e-01 4.45569009e-01
4.55044240e-01 9.38188970e-01 -1.13343537e+00 -3.65597665e-01
-1.96820587e-01 1.67040670e+00 -6.98140442e-01 2.81921923e-01
-2.13622555e-01 -7.10704565e-01 1.98829532e-01 3.67760986e-01
9.50098857e-02 -3.52132797e-01 -2.99173474e-01 -7.76620209e-01
3.37661207e-01 -4.93103325e-01 5.74189067e-01 -8.76009762e-01
-7.80371308e-01 4.79633749e-01 6.35923594e-02 -1.15489447e+00
-2.59330720e-01 -1.00213182e+00 -7.98309147e-01 7.98641324e-01
-6.96596801e-01 -1.37270713e+00 -9.62346911e-01 1.17012060e+00
6.13750100e-01 -4.09888387e-01 1.41397715e+00 2.43688017e-01
-3.77261341e-01 3.42297971e-01 -2.79822379e-01 4.22963917e-01
5.79554200e-01 -1.02067029e+00 -3.45758647e-01 3.88678104e-01
-4.30570602e-01 8.14878643e-01 5.59988737e-01 -6.95078850e-01
-1.06013870e+00 -2.88910896e-01 1.17290163e+00 -5.42359471e-01
-1.74232647e-01 4.06103134e-02 -2.10406080e-01 7.08438218e-01
2.05818951e-01 -5.40729091e-02 5.89833736e-01 -3.14764291e-01
2.85087883e-01 -5.63251190e-02 -1.98792434e+00 2.58275360e-01
1.00810504e+00 -1.63061678e-01 -8.36873472e-01 4.28694218e-01
-3.17607373e-01 7.62106478e-02 -4.15558666e-01 -1.33163109e-01
1.37191737e+00 -8.55439901e-01 7.74271846e-01 -4.35479552e-01
-7.31312037e-01 -3.63495350e-01 -1.80410892e-01 -4.92219299e-01
1.37597276e-02 -6.02837980e-01 1.11637205e-01 1.13873589e+00
3.29668790e-01 -8.04737926e-01 7.04063058e-01 1.30748367e+00
6.75293684e-01 -8.03868938e-03 -1.02488554e+00 -5.84370136e-01
-6.91502750e-01 -2.04776794e-01 5.30021489e-01 6.59034312e-01
5.62413931e-01 4.28267330e-01 -2.59696335e-01 2.63664246e-01
5.25114536e-01 -7.83089280e-01 2.93344855e-01 -1.64428961e+00
-6.68048039e-02 -2.70161867e-01 -6.75237954e-01 -5.12459457e-01
-7.49919415e-01 -1.97900265e-01 -1.56869575e-01 -2.25213027e+00
-1.17325149e-01 -1.29378647e-01 4.16840659e-03 9.91848767e-01
2.63522595e-01 3.41728747e-01 -1.76499695e-01 3.02843213e-01
-1.00873731e-01 -3.85294914e-01 7.73747683e-01 -4.09246013e-02
-7.30907500e-01 5.27179301e-01 -1.36076808e-01 9.21178997e-01
9.38578606e-01 -1.92036867e-01 -1.54367104e-01 -1.09862804e-01
-8.56114775e-02 1.36142358e-01 3.73128891e-01 -1.26527739e+00
1.38387397e-01 1.67726234e-01 9.10130978e-01 -5.99845946e-01
2.52228826e-01 -1.48008037e+00 1.09164812e-01 1.03106320e+00
2.28710994e-01 1.90039396e-01 -2.99485922e-01 -5.19821048e-02
6.50073886e-01 -4.87027198e-01 5.03060162e-01 -5.34651220e-01
-7.17691839e-01 -1.17478527e-01 -6.80213153e-01 -8.62537205e-01
1.12905955e+00 -8.86871934e-01 -9.71247721e-03 -6.17911041e-01
-1.39063382e+00 1.41918883e-01 3.30433518e-01 2.75038362e-01
4.47828770e-01 -1.31935108e+00 -2.86805898e-01 9.07545090e-02
6.13731369e-02 -4.54291373e-01 -6.13861442e-01 6.13063216e-01
-1.15773833e+00 4.26847756e-01 -8.37332785e-01 5.15014539e-03
-2.09952950e+00 1.71643659e-01 3.08815628e-01 2.78538406e-01
-1.02831483e+00 5.88551044e-01 -9.44846213e-01 -4.08943482e-02
9.23653364e-01 -7.80200064e-01 -8.55095208e-01 1.41914070e-01
9.91569936e-01 7.81113744e-01 -2.46258885e-01 -8.48516822e-01
-9.09314454e-01 9.89085913e-01 6.75123274e-01 -4.05214965e-01
1.44960022e+00 -3.42581481e-01 1.53894261e-01 5.36793709e-01
9.33176339e-01 -5.41517913e-01 -7.55181432e-01 2.47752234e-01
-1.92600489e-02 1.21042341e-01 6.38732240e-02 -1.13478780e+00
-9.22887981e-01 7.78824568e-01 1.73731482e+00 2.83509523e-01
1.09716868e+00 -1.54570952e-01 7.41559446e-01 9.01979983e-01
6.79029286e-01 -1.25419545e+00 -5.30911982e-01 -6.32183626e-03
9.29526627e-01 -1.55040050e+00 3.46886516e-01 1.09412521e-01
-4.99221116e-01 1.63391602e+00 -1.03733484e-02 1.06925502e-01
1.18528152e+00 5.16712189e-01 5.20457089e-01 -3.48125637e-01
4.04297501e-01 -4.71062005e-01 5.84043153e-02 1.61171412e+00
7.91732907e-01 4.92868721e-01 -9.66769099e-01 5.23864508e-01
-5.92725798e-02 6.32693231e-01 1.37994856e-01 1.55900443e+00
-8.30315769e-01 -1.10629439e+00 -8.96861553e-01 4.91360277e-01
-6.24737024e-01 3.51824939e-01 -3.81255478e-01 7.88248301e-01
4.67449129e-01 1.31582689e+00 -1.45501137e-01 -1.80344686e-01
2.87381172e-01 6.42309844e-01 7.51050889e-01 -3.07339787e-01
-3.54253769e-01 3.20821181e-02 2.31588945e-01 -2.24578485e-01
-7.00130522e-01 -8.20654809e-01 -1.67098272e+00 -3.61602567e-02
2.00380489e-01 -4.98930901e-01 1.03679514e+00 1.10575330e+00
-4.68960911e-01 2.22596824e-02 -6.17317744e-02 -1.20506454e+00
-3.53628770e-02 -1.53960752e+00 -8.70307684e-01 -1.10189877e-02
5.42741120e-01 -5.87519050e-01 1.00019880e-01 5.04434943e-01]
|
[6.488065242767334, -0.2487478256225586]
|
a1a22a50-2360-4bf0-9c05-8c17a4c669f2
|
uob-at-semeval-2021-task-5-extending-pre-1
|
2110.0373
| null |
https://arxiv.org/abs/2110.03730v1
|
https://arxiv.org/pdf/2110.03730v1.pdf
|
UoB at SemEval-2021 Task 5: Extending Pre-Trained Language Models to Include Task and Domain-Specific Information for Toxic Span Prediction
|
Toxicity is pervasive in social media and poses a major threat to the health of online communities. The recent introduction of pre-trained language models, which have achieved state-of-the-art results in many NLP tasks, has transformed the way in which we approach natural language processing. However, the inherent nature of pre-training means that they are unlikely to capture task-specific statistical information or learn domain-specific knowledge. Additionally, most implementations of these models typically do not employ conditional random fields, a method for simultaneous token classification. We show that these modifications can improve model performance on the Toxic Spans Detection task at SemEval-2021 to achieve a score within 4 percentage points of the top performing team.
|
['Harish Tayyar Madabushi', 'Erik Yan']
|
2021-10-07
|
uob-at-semeval-2021-task-5-extending-pre
|
https://aclanthology.org/2021.semeval-1.28
|
https://aclanthology.org/2021.semeval-1.28.pdf
|
semeval-2021
|
['toxic-spans-detection']
|
['natural-language-processing']
|
[ 2.31733784e-01 -1.21885367e-01 -3.07107210e-01 -7.85920322e-02
-1.27661717e+00 -6.62044525e-01 8.93196762e-01 1.07073426e+00
-9.14499640e-01 9.13169265e-01 4.57750142e-01 -2.23601922e-01
1.72078852e-02 -7.06282616e-01 -6.54943943e-01 -3.47694546e-01
-2.08908662e-01 2.76094794e-01 3.47221911e-01 1.52965918e-01
2.71743149e-01 3.33587199e-01 -8.74506474e-01 7.43870080e-01
7.80940473e-01 4.76382554e-01 -2.85186041e-02 5.07184744e-01
-1.54712543e-01 1.07658422e+00 -5.22502840e-01 -4.74040478e-01
-1.73380405e-01 -5.21666603e-03 -8.33652675e-01 -3.30843031e-01
3.45004618e-01 1.37792334e-01 -2.24828437e-01 9.48837996e-01
4.12848890e-01 -8.34651589e-02 9.74058688e-01 -7.07733631e-01
-4.09166694e-01 6.01983309e-01 -4.71969813e-01 2.81277478e-01
4.82356697e-01 3.28939199e-01 1.06866884e+00 -5.78435004e-01
6.80505872e-01 1.15056849e+00 1.05833316e+00 3.77787203e-01
-1.36691642e+00 -7.28589714e-01 1.65877834e-01 -4.25642096e-02
-1.45412934e+00 -4.19306040e-01 9.76145864e-02 -8.05799603e-01
1.48742115e+00 -1.09963886e-01 2.22639024e-01 1.21627545e+00
6.32736683e-01 4.91660863e-01 1.21931899e+00 -3.36714417e-01
2.22432584e-01 1.31420106e-01 5.27418628e-02 4.28383291e-01
3.64890665e-01 -2.22338498e-01 -8.45156312e-01 -6.38478339e-01
1.53812155e-01 -5.47098406e-02 1.18182279e-01 3.41598570e-01
-1.06055975e+00 8.85077775e-01 2.13344678e-01 5.07394731e-01
-6.03153825e-01 3.37602168e-01 6.08504474e-01 -2.99250726e-02
8.45611572e-01 5.64159453e-01 -5.92674375e-01 -1.37381554e-01
-1.04531753e+00 5.31553626e-01 8.02179456e-01 4.52978820e-01
3.65581334e-01 -3.66575688e-01 -4.64631706e-01 7.56380677e-01
5.24241924e-01 2.96324998e-01 2.35993505e-01 -7.13439465e-01
6.38082385e-01 5.30601144e-01 -1.11627840e-02 -8.02177608e-01
-5.50018132e-01 -4.70634103e-01 -3.50191027e-01 -2.00186577e-02
6.88659012e-01 -2.82540679e-01 -8.29436421e-01 1.59514356e+00
-4.92086634e-02 1.01074600e-03 -2.70843089e-01 8.17758963e-02
3.23916584e-01 4.27035362e-01 1.06984353e+00 -1.48808822e-01
1.21318972e+00 -3.44942600e-01 -5.21242857e-01 -4.77827728e-01
9.06743109e-01 -7.45449960e-01 6.62364125e-01 7.41668940e-01
-8.50559950e-01 1.13714993e-01 -7.95546830e-01 1.84767172e-02
-6.64524972e-01 -6.22738361e-01 6.59643412e-01 8.90506268e-01
-6.21414304e-01 7.36599863e-01 -8.34200621e-01 -4.04631346e-01
6.86686635e-01 2.77165294e-01 -4.31655496e-01 -3.34342659e-01
-1.30632770e+00 1.14496708e+00 2.36504123e-01 -3.23183209e-01
-9.40068185e-01 -1.04255009e+00 -5.98423481e-01 -3.30380276e-02
3.28807741e-01 -5.29354990e-01 1.43353331e+00 -3.05662811e-01
-8.36937428e-01 6.52008951e-01 -1.95409477e-01 -4.79464978e-01
5.80297828e-01 -2.77628928e-01 -3.91459018e-01 4.66130488e-02
3.30611587e-01 3.34440976e-01 1.81301326e-01 -7.26405561e-01
-6.88268244e-01 -2.15696633e-01 -2.15693384e-01 7.77028799e-02
-6.19528949e-01 6.84550583e-01 -1.85952753e-01 -3.47669035e-01
-5.24859846e-01 -6.15844607e-01 -7.74013698e-01 -7.20171705e-02
-2.58483857e-01 -4.77925271e-01 2.41981670e-01 -8.51199627e-01
1.11660719e+00 -1.80860031e+00 -4.84969676e-01 2.27634639e-01
4.05808121e-01 1.58033952e-01 -1.60086472e-02 8.08381498e-01
-8.02096277e-02 5.60618341e-01 -2.99416631e-01 -4.84161049e-01
-9.72290710e-03 -8.19806755e-02 4.51879352e-02 6.40345573e-01
3.48223180e-01 5.77933192e-01 -1.20177364e+00 -4.32648838e-01
-2.43321434e-02 5.41087508e-01 -6.24817610e-01 -2.44282737e-01
-5.39822221e-01 1.77252591e-01 -4.67332155e-01 3.39558423e-01
2.12654591e-01 -1.14761956e-01 3.89805257e-01 3.54734123e-01
-4.02916312e-01 9.37415838e-01 -8.93751502e-01 1.46239603e+00
-3.85737091e-01 3.02118301e-01 -2.35468894e-01 -6.52132809e-01
3.02356958e-01 4.93777573e-01 7.24422455e-01 -6.21765018e-01
-7.69697502e-02 7.64753576e-03 1.41161159e-02 -5.51929653e-01
1.32327974e-01 -6.54875398e-01 -2.31513023e-01 4.68445152e-01
-5.45766838e-02 1.56821579e-01 3.91852528e-01 2.78975964e-01
1.64968181e+00 -2.48786926e-01 3.25194061e-01 -3.16617191e-01
1.30658299e-01 1.60412803e-01 7.55879641e-01 8.84883285e-01
-8.76828060e-02 5.85451961e-01 5.76464117e-01 -7.23142549e-02
-8.19072425e-01 -9.39656138e-01 -2.47292191e-01 1.13326573e+00
-7.60764301e-01 -7.47193515e-01 -6.20605230e-01 -7.85906196e-01
1.56551152e-02 8.48641992e-01 -4.00680333e-01 -6.48031896e-03
-2.54277378e-01 -1.27070844e+00 9.22821879e-01 3.41470271e-01
6.60242559e-03 -9.76315856e-01 -1.96958944e-01 5.80670893e-01
-1.56933904e-01 -1.23172927e+00 -3.16197902e-01 4.34154868e-01
-6.88694179e-01 -1.22472572e+00 -4.98311698e-01 -4.05173451e-01
4.29800034e-01 -1.67405546e-01 1.14095426e+00 -1.34405956e-01
-2.97457516e-01 2.53112078e-01 -2.49935329e-01 -8.16957057e-01
-5.70326686e-01 1.93128750e-01 3.76482084e-02 -2.02467948e-01
7.32043564e-01 -4.53362674e-01 -2.47999683e-01 -3.18574101e-01
-9.02920246e-01 -4.46841866e-01 6.04606271e-01 3.47994566e-01
3.34592581e-01 3.08873415e-01 7.56054461e-01 -1.40260351e+00
8.21229517e-01 -8.47262323e-01 -1.70016259e-01 1.39290169e-01
-8.34939003e-01 -6.60108328e-02 5.13997376e-01 -2.49121964e-01
-8.08699012e-01 2.15246707e-01 -2.03166977e-01 3.22708011e-01
-3.37491572e-01 1.04931331e+00 1.10806100e-01 1.23478353e-01
9.38375771e-01 -5.71306683e-02 -3.92252654e-01 -7.13923872e-01
-9.73170530e-03 6.23261988e-01 2.29853317e-01 -4.99307305e-01
7.57120788e-01 1.33071765e-01 -1.35558009e-01 -9.01674449e-01
-9.87562597e-01 -7.35125899e-01 -4.09075469e-01 1.00891821e-01
1.08090436e+00 -1.20149636e+00 -4.46321368e-01 6.04586840e-01
-1.33399725e+00 -4.54088628e-01 5.89972399e-02 4.43798184e-01
-2.24111434e-02 5.24402797e-01 -6.49472058e-01 -8.75619948e-01
-3.17616820e-01 -5.70935011e-01 6.86472714e-01 -9.81443152e-02
-4.74135190e-01 -1.15551090e+00 4.38282490e-01 4.63996828e-01
3.28937620e-01 2.48575881e-01 1.12546515e+00 -8.59505117e-01
-1.65698290e-01 -5.04621327e-01 -5.56250326e-02 2.57226199e-01
3.57182436e-02 -4.71957102e-02 -1.09441316e+00 5.57870371e-03
-2.40845859e-01 -4.01719213e-01 8.65637779e-01 2.47259423e-01
9.22345996e-01 -2.29794994e-01 -5.01946509e-01 -1.54657483e-01
1.31797695e+00 -1.36692703e-01 6.59338713e-01 2.10283920e-01
4.65875983e-01 5.29027343e-01 7.53555521e-02 3.75929952e-01
5.05761325e-01 5.07366717e-01 1.58047169e-01 1.42568752e-01
-6.95146993e-02 -4.54986453e-01 6.44075513e-01 4.77671593e-01
1.02742992e-01 -2.34510884e-01 -1.30982924e+00 8.81459713e-01
-1.64026356e+00 -1.08816874e+00 -5.13639152e-01 2.14316583e+00
1.21181309e+00 4.69911337e-01 2.08212689e-01 5.94292656e-02
5.84553957e-01 -4.85237874e-02 -1.60549343e-01 -4.26063150e-01
9.42736417e-02 4.13224310e-01 8.13860953e-01 4.61069137e-01
-1.05845988e+00 7.98606098e-01 7.58861542e+00 7.76869178e-01
-8.80234838e-01 3.88865441e-01 3.19649786e-01 5.70556186e-02
-1.51661292e-01 -5.52304946e-02 -7.69538522e-01 4.96196777e-01
1.47220802e+00 -2.22218961e-01 2.68776506e-01 3.86065841e-01
6.42519414e-01 -3.90942514e-01 -1.20319152e+00 5.45353472e-01
-9.40357223e-02 -1.33480299e+00 -1.95263416e-01 3.05576652e-01
6.62926376e-01 6.34701133e-01 -2.98141409e-02 3.70618850e-01
5.63951612e-01 -1.38503349e+00 6.80791497e-01 5.85232496e-01
5.52268744e-01 -5.87716877e-01 5.93531728e-01 6.59564316e-01
-7.19406009e-01 -3.60797867e-02 -2.89737672e-01 -3.20656836e-01
8.41127932e-02 1.11640751e+00 -1.19911432e+00 4.40931559e-01
5.44335306e-01 5.71317136e-01 -6.63776875e-01 1.55075383e+00
-3.00883055e-01 1.11080480e+00 -4.84334379e-01 -1.31610468e-01
2.56397814e-01 6.74034297e-01 2.87919879e-01 1.69138706e+00
-1.91457555e-01 -2.06668749e-01 3.42281938e-01 5.73961258e-01
-4.32838976e-01 3.24318260e-01 -6.56458914e-01 -5.42904794e-01
3.05222869e-01 9.62336242e-01 -6.04138017e-01 -1.23735547e-01
-7.37862408e-01 3.04933280e-01 3.13706696e-01 3.11442632e-02
-4.09274668e-01 -2.99120545e-01 5.25652051e-01 5.66672623e-01
2.46270578e-02 -4.22449708e-01 -4.60400194e-01 -8.41825604e-01
-3.07215154e-01 -9.93997514e-01 4.91582572e-01 -2.14005083e-01
-1.66993725e+00 1.06526867e-01 -6.40574098e-02 -7.57740259e-01
-7.63153508e-02 -7.35447645e-01 -7.42070258e-01 9.77923155e-01
-1.37362611e+00 -8.63171399e-01 3.28785926e-01 2.95263052e-01
3.12442869e-01 2.65185554e-02 1.09896541e+00 6.77565396e-01
-3.42970163e-01 4.37848061e-01 1.67831540e-01 1.79922938e-01
9.29991186e-01 -1.20418406e+00 3.32531035e-01 7.29794919e-01
7.82653466e-02 7.68846393e-01 8.82543504e-01 -8.40053082e-01
-1.03405213e+00 -1.18675482e+00 1.60842609e+00 -9.78146076e-01
8.53923082e-01 -5.01911581e-01 -7.65892684e-01 5.42220354e-01
3.05470396e-02 -2.49714956e-01 1.15771234e+00 5.84085703e-01
-6.32048726e-01 1.81749165e-01 -1.19244540e+00 2.76920080e-01
6.87833667e-01 -8.47830772e-01 -5.66412747e-01 8.32991898e-01
4.57019478e-01 1.54795572e-01 -9.52049077e-01 7.75416149e-04
3.31411868e-01 -4.56220686e-01 7.50749588e-01 -8.90311003e-01
4.90421116e-01 -5.37979715e-02 -7.67055750e-02 -1.02101326e+00
-3.78601402e-01 -3.77418637e-01 2.18142867e-01 1.36169350e+00
8.31624210e-01 -3.54582965e-01 7.16073394e-01 1.02662408e+00
4.69891168e-02 -4.24391568e-01 -7.44813859e-01 -7.39459276e-01
5.79371274e-01 -7.67430186e-01 4.05515619e-02 1.02495408e+00
2.96881467e-01 4.28021133e-01 -1.49164289e-01 1.03971586e-01
6.74888194e-01 -7.58601129e-01 2.15097591e-01 -1.32379377e+00
-3.90930116e-01 -3.17199796e-01 -3.28589529e-01 -3.48536789e-01
2.05136925e-01 -1.16857004e+00 1.13853417e-01 -1.75390625e+00
6.80173039e-01 -3.84468555e-01 -6.13132775e-01 8.71205270e-01
-7.82230124e-02 3.88264567e-01 3.26413140e-02 -1.16053611e-01
-7.88796008e-01 1.51115477e-01 5.38213491e-01 -2.59669453e-01
-5.10369278e-02 -8.83428603e-02 -1.13879406e+00 7.67438591e-01
7.41487443e-01 -1.30903411e+00 -9.32370201e-02 -4.82279390e-01
6.77043140e-01 -4.24733430e-01 2.39027381e-01 -1.05754840e+00
3.56219262e-01 -3.54062140e-01 4.12096709e-01 -1.46946609e-01
-2.31716186e-02 -4.06581700e-01 1.38853058e-01 6.38739944e-01
-6.13803387e-01 -1.51662886e-01 2.89060235e-01 6.42716944e-01
1.22294344e-01 -5.30984938e-01 6.09552503e-01 -3.84718120e-01
-1.22597948e-01 1.73470646e-01 -1.12345874e+00 1.75443038e-01
6.58517241e-01 2.85422325e-01 -2.79820144e-01 -1.89960271e-01
-4.31797475e-01 2.57749319e-01 2.92323679e-01 3.66763175e-02
-6.98652714e-02 -7.57096887e-01 -9.33017075e-01 -6.51257455e-01
1.02282219e-01 -4.32321727e-02 1.22248553e-01 8.12220454e-01
-3.74024212e-01 4.75287616e-01 1.05537720e-01 -4.81821187e-02
-1.06203318e+00 3.39379579e-01 3.14057291e-01 -8.12437594e-01
-3.72536808e-01 6.38658106e-01 9.46134925e-02 -3.47280741e-01
2.42997244e-01 -2.29533389e-02 -1.91689447e-01 4.05511213e-03
6.98747039e-01 2.42472976e-01 3.22760940e-01 -2.78963029e-01
-5.55767834e-01 3.52805816e-02 -3.87169570e-01 -1.65292040e-01
1.69195282e+00 3.36182207e-01 -2.13090762e-01 4.19640362e-01
9.80730057e-01 3.70002627e-01 -7.52104342e-01 -2.24680200e-01
5.71340382e-01 -8.94709155e-02 3.51418227e-01 -1.25124764e+00
-4.67498362e-01 9.56047118e-01 2.36594766e-01 3.93126830e-02
4.54115301e-01 -2.04421040e-02 6.44462943e-01 6.56467974e-01
2.38345340e-01 -1.13974166e+00 -8.58404413e-02 5.85529923e-01
4.92861629e-01 -8.07066381e-01 2.55241156e-01 -4.90607589e-01
-4.57359046e-01 7.90245712e-01 2.55537480e-01 -1.88023467e-02
6.94242299e-01 3.58562917e-01 -1.27847463e-01 -2.37294748e-01
-9.94978309e-01 -6.65098801e-02 -1.65833104e-02 7.51114786e-01
8.27112615e-01 5.34269074e-03 -5.36284745e-01 7.41212308e-01
1.96095005e-01 1.98976219e-01 6.51759267e-01 8.70634317e-01
-5.39606750e-01 -1.46706367e+00 -2.02469945e-01 5.42181849e-01
-1.18498492e+00 -3.53125095e-01 -5.13337731e-01 3.40531915e-01
2.28594661e-01 1.10243273e+00 -4.77783412e-01 -1.00531273e-01
3.11891019e-01 5.84984541e-01 4.19625908e-01 -1.27245772e+00
-1.01391912e+00 -5.39048947e-02 6.02145195e-01 -1.98018298e-01
-2.39981189e-01 -8.89343977e-01 -1.32589316e+00 -3.55708063e-01
-7.29246363e-02 2.43273184e-01 6.31285489e-01 1.24379826e+00
1.72231510e-01 3.41464221e-01 1.41053125e-01 -5.61805069e-01
-5.58339536e-01 -9.99405146e-01 -5.77181518e-01 1.03274643e-01
1.51149714e-02 -4.12698984e-01 2.98386607e-02 1.66911528e-01]
|
[8.980640411376953, 10.572006225585938]
|
d14861b3-b2f1-43f2-9d90-47b8ae80cacd
|
a-comparative-study-of-deep-learning-loss
|
2009.13935
| null |
https://arxiv.org/abs/2009.13935v1
|
https://arxiv.org/pdf/2009.13935v1.pdf
|
A Comparative Study of Deep Learning Loss Functions for Multi-Label Remote Sensing Image Classification
|
This paper analyzes and compares different deep learning loss functions in the framework of multi-label remote sensing (RS) image scene classification problems. We consider seven loss functions: 1) cross-entropy loss; 2) focal loss; 3) weighted cross-entropy loss; 4) Hamming loss; 5) Huber loss; 6) ranking loss; and 7) sparseMax loss. All the considered loss functions are analyzed for the first time in RS. After a theoretical analysis, an experimental analysis is carried out to compare the considered loss functions in terms of their: 1) overall accuracy; 2) class imbalance awareness (for which the number of samples associated to each class significantly varies); 3) convexibility and differentiability; and 4) learning efficiency (i.e., convergence speed). On the basis of our analysis, some guidelines are derived for a proper selection of a loss function in multi-label RS scene classification problems.
|
['Begüm Demir', 'Hichame Yessou', 'Gencer Sumbul']
|
2020-09-29
| null | null | null | null |
['remote-sensing-image-classification']
|
['miscellaneous']
|
[ 4.26684290e-01 -1.40283540e-01 -1.58903003e-01 -5.15862346e-01
-9.39898551e-01 -1.60352871e-01 2.86740661e-01 8.96937490e-01
-7.44876027e-01 7.58628428e-01 -2.94459045e-01 3.76550555e-02
-7.76451409e-01 -8.15333784e-01 -4.91106570e-01 -1.02009845e+00
-3.08765739e-01 2.86774635e-01 -1.91483706e-01 1.62500083e-01
2.72125781e-01 8.35208476e-01 -1.78240418e+00 -5.87430745e-02
1.04455411e+00 1.39871275e+00 -4.17203121e-02 2.21603230e-01
-3.27855162e-02 1.13444030e+00 -4.22243059e-01 -1.32400274e-01
2.52596796e-01 -3.57462347e-01 -1.02568614e+00 -7.02573508e-02
2.72001475e-01 -3.25031355e-02 3.30509275e-01 1.07817411e+00
5.69282830e-01 4.49097216e-01 1.09963608e+00 -1.07169354e+00
-1.95482239e-01 4.66197193e-01 -6.30972266e-01 1.61461323e-01
-3.90389487e-02 -2.60599375e-01 1.09110045e+00 -6.32387519e-01
3.25206488e-01 9.83132184e-01 8.21733832e-01 -2.69811582e-02
-1.06074953e+00 -3.60131621e-01 2.97837378e-03 4.14517611e-01
-1.62070990e+00 -5.58875725e-02 5.27757764e-01 -7.45827794e-01
3.45098227e-01 5.18076003e-01 3.12174380e-01 4.62605447e-01
2.13872373e-01 5.19629776e-01 1.21408045e+00 -3.59004378e-01
3.07160735e-01 3.79107416e-01 2.72315204e-01 6.96416259e-01
1.14648595e-01 -2.96409488e-01 8.68034884e-02 -1.16512015e-01
1.85377643e-01 -1.84212059e-01 -1.74296468e-01 -2.46228591e-01
-7.52215683e-01 1.18969762e+00 5.91316640e-01 3.89833778e-01
-3.24813664e-01 -2.02665672e-01 6.33923352e-01 3.34866703e-01
7.04788506e-01 3.60424399e-01 -2.57698387e-01 6.55755043e-01
-9.34723973e-01 1.69451386e-01 6.43653870e-01 4.17793661e-01
1.02349114e+00 -7.40388315e-03 -2.30160937e-01 1.20690942e+00
1.85205340e-01 4.52873379e-01 4.24942911e-01 -3.70960236e-01
3.93283159e-01 4.72522855e-01 6.69690920e-03 -1.08133888e+00
-8.99802208e-01 -7.44133234e-01 -9.02337015e-01 2.18792364e-01
9.91021171e-02 -1.25485763e-01 -3.92757535e-01 1.51934958e+00
2.34414697e-01 -1.35896191e-01 3.60352807e-02 7.85561144e-01
1.07599199e+00 6.44294679e-01 2.93273538e-01 -3.47498715e-01
1.11527920e+00 -6.14619672e-01 -5.39763629e-01 1.50612727e-01
4.48741555e-01 -6.19205356e-01 9.14357722e-01 2.17152610e-01
-7.73363769e-01 -4.02351230e-01 -9.67130959e-01 9.42407101e-02
-6.07958555e-01 4.30557907e-01 4.04422075e-01 5.52734613e-01
-6.89921319e-01 6.65042579e-01 -2.69713134e-01 -3.48973155e-01
3.47139150e-01 1.32139161e-01 -2.15142861e-01 4.80644330e-02
-1.13581944e+00 8.85941088e-01 3.65519166e-01 1.02091506e-01
-7.07769632e-01 -7.42881060e-01 -7.69508302e-01 1.74626157e-01
1.33357778e-01 -2.77581125e-01 6.21322215e-01 -1.24324417e+00
-1.17116797e+00 1.35356188e+00 4.35579479e-01 -3.17103386e-01
8.55504394e-01 -2.00202867e-01 -3.02856326e-01 1.61232352e-01
2.19816994e-02 3.81637335e-01 3.23617727e-01 -1.44746411e+00
-5.30800045e-01 -5.87962508e-01 5.28442189e-02 3.15248877e-01
-1.76483229e-01 5.92140201e-03 4.28339034e-01 -6.86872959e-01
1.23413689e-01 -5.87130964e-01 -1.97222903e-01 9.27890912e-02
-5.25548100e-01 -2.40426555e-01 5.33509433e-01 -7.51299143e-01
1.05762768e+00 -2.00700545e+00 7.24082962e-02 3.64402920e-01
3.57322185e-03 3.77376080e-01 -5.12401685e-02 4.01245773e-01
-2.34652653e-01 1.28459975e-01 -6.15718961e-01 -8.52573812e-02
-1.72555000e-01 -3.22788388e-01 1.25559136e-01 9.91384745e-01
3.29493918e-02 3.05328131e-01 -7.14606047e-01 -5.71711719e-01
3.75362515e-01 6.46911383e-01 -2.33878866e-01 2.10574910e-01
3.08062043e-02 2.69967020e-01 -4.89440203e-01 5.83428800e-01
8.20164144e-01 -2.79299647e-01 -8.58791471e-02 -5.10646641e-01
-2.39181846e-01 -1.77292600e-01 -1.33764327e+00 1.03508556e+00
-6.66969121e-01 4.38506156e-01 5.25427759e-02 -1.27362168e+00
1.38856375e+00 2.44390145e-02 1.00435829e+00 -5.44298649e-01
2.46656403e-01 2.18102783e-01 -6.41040087e-01 -5.26455820e-01
3.98514271e-01 -2.54873157e-01 1.13442503e-01 1.28603399e-01
2.27327482e-03 1.94945768e-01 2.51542330e-01 -5.19836366e-01
5.40469944e-01 -3.27334374e-01 5.14189780e-01 -6.07886851e-01
7.39127576e-01 -2.51743197e-01 5.51427007e-01 5.26499927e-01
-3.14793527e-01 3.87768984e-01 4.25044000e-01 -6.08119547e-01
-8.26424420e-01 -8.75448585e-01 -5.87668836e-01 8.76580894e-01
2.43477285e-01 2.79610395e-01 -4.64788824e-01 -5.20802617e-01
1.70629844e-01 7.39589036e-01 -5.73651671e-01 7.67457038e-02
-3.09456676e-01 -1.42002022e+00 5.52463233e-01 7.54316822e-02
8.87102962e-01 -8.66640925e-01 -8.26475143e-01 -1.92092419e-01
-2.75599509e-01 -7.69394100e-01 1.49831325e-01 4.17710006e-01
-7.40399003e-01 -1.42382312e+00 -7.17625737e-01 -7.18290150e-01
4.65438277e-01 4.07497771e-02 9.57754850e-01 -1.93891704e-01
-2.98226863e-01 3.62333596e-01 -5.44872344e-01 -3.36437911e-01
-3.78065407e-01 9.60282460e-02 -3.78522277e-01 2.50367671e-01
1.94221854e-01 -2.66434431e-01 -5.98156750e-01 2.21626103e-01
-1.01436734e+00 -4.00855929e-01 4.49435920e-01 5.84017634e-01
9.16472018e-01 3.43432546e-01 4.99466598e-01 -8.95344436e-01
3.68675828e-01 -6.67305708e-01 -8.92749488e-01 6.22850716e-01
-7.41280317e-01 -2.03041121e-01 5.52734613e-01 7.70663545e-02
-9.20900226e-01 4.46369275e-02 -2.65289456e-01 -8.81750509e-02
-1.59140497e-01 5.67283988e-01 3.90745178e-02 -3.61474395e-01
7.38766313e-01 3.10366213e-01 -9.48577672e-02 -5.47053099e-01
1.55395716e-01 6.67257965e-01 -2.94799898e-02 -2.36510426e-01
2.69806504e-01 4.83736306e-01 3.91557962e-01 -1.09506083e+00
-9.79606390e-01 -6.11676335e-01 -3.76552731e-01 -4.37180221e-01
1.07463169e+00 -9.66470897e-01 -8.37134600e-01 7.24471509e-01
-8.19422364e-01 -1.96781784e-01 -3.72688949e-01 6.04908645e-01
-6.26060367e-01 3.83837938e-01 -7.13590205e-01 -9.95788395e-01
-6.12507463e-01 -1.18943334e+00 1.02433336e+00 1.38737842e-01
3.93757910e-01 -1.11483324e+00 1.47835584e-03 4.57328916e-01
3.06401998e-01 9.60994303e-01 1.01890981e+00 -4.94886220e-01
-2.37572417e-02 -5.34638390e-02 -4.19048101e-01 8.42033803e-01
-1.76723287e-01 5.55865513e-03 -9.52758789e-01 -3.94827098e-01
3.64878587e-02 -4.72421736e-01 9.79183018e-01 7.21210659e-01
1.40402734e+00 -4.01526153e-01 -3.09039764e-02 6.52214170e-01
2.03641319e+00 1.50102720e-01 6.50326669e-01 3.32279980e-01
5.11881292e-01 9.45381820e-01 6.74074531e-01 6.85350418e-01
2.94311613e-01 6.65363491e-01 7.62728393e-01 -2.24796489e-01
1.71437636e-01 3.12343627e-01 5.62478416e-02 6.14089012e-01
-5.62108457e-02 -4.62090433e-01 -8.40210438e-01 3.42516482e-01
-1.66611874e+00 -9.64320242e-01 -2.85505712e-01 2.31971908e+00
2.33782083e-01 -3.08653593e-01 2.68223137e-01 3.97176921e-01
1.00017178e+00 3.99288327e-01 -5.45931995e-01 -4.59550112e-01
-5.64903438e-01 2.28810403e-02 7.45365679e-01 5.34791589e-01
-1.74691367e+00 3.66078436e-01 5.82068491e+00 9.74862278e-01
-1.20658839e+00 1.22220092e-01 8.52309883e-01 1.57225996e-01
-1.97088420e-01 -4.41972464e-01 -5.07242322e-01 4.97480541e-01
3.71887445e-01 1.19216494e-01 2.26338193e-01 8.50927591e-01
-9.08571929e-02 -1.27355069e-01 -5.39902389e-01 1.03153038e+00
1.54783875e-01 -1.03093934e+00 -5.63654630e-03 -1.63156554e-01
6.39418066e-01 2.78236330e-01 -1.04969189e-01 -2.85884254e-02
-1.18504092e-01 -1.06351018e+00 7.55707502e-01 6.63563609e-01
8.54541183e-01 -8.96708131e-01 9.44652200e-01 4.35786434e-02
-1.22254753e+00 -4.66561139e-01 -4.99697000e-01 3.33953083e-01
-1.11904949e-01 1.26493335e+00 -1.98505878e-01 9.28427517e-01
5.26127160e-01 5.89486122e-01 -5.38271129e-01 1.27730489e+00
1.84878856e-02 4.09440160e-01 -1.87288672e-01 -8.49069506e-02
3.01116735e-01 -3.47049326e-01 5.94762325e-01 1.30768478e+00
4.23192799e-01 -1.97942749e-01 1.64618760e-01 7.99845397e-01
5.13223745e-02 6.53158367e-01 -4.01845187e-01 2.40062073e-01
4.20276195e-01 1.15600896e+00 -7.28756368e-01 1.03386074e-01
-7.35857412e-02 6.72310531e-01 5.85754141e-02 2.11543068e-01
-6.56585693e-01 -5.20891547e-01 5.29043078e-01 -2.02239025e-02
1.50757015e-01 3.66608500e-01 -2.65652150e-01 -8.72702777e-01
-7.55251050e-02 -3.76170456e-01 8.36559117e-01 -5.64488232e-01
-1.41447854e+00 6.46524847e-01 7.12355897e-02 -1.24739909e+00
1.62407681e-01 -5.71862221e-01 -1.77911833e-01 7.28328347e-01
-1.78052390e+00 -8.43979061e-01 -3.85085613e-01 4.05751675e-01
1.15121514e-01 -2.73346037e-01 7.47594237e-01 9.44930077e-01
-6.25666142e-01 7.38380194e-01 4.67006952e-01 -1.40024349e-01
1.23570450e-01 -1.07372522e+00 -4.53562111e-01 3.52650166e-01
-3.78633291e-01 -1.87283933e-01 4.91107911e-01 -2.11099640e-01
-6.91468060e-01 -1.32463419e+00 9.29375470e-01 1.94039494e-01
1.43832237e-01 6.82489276e-02 -7.95449615e-01 4.02468711e-01
-4.24876392e-01 -9.67511460e-02 9.35656846e-01 -1.58521496e-02
-3.13059926e-01 -6.07135475e-01 -1.65446293e+00 -8.94961581e-02
2.61978179e-01 -3.57932806e-01 4.68041971e-02 7.36205339e-01
2.70336300e-01 -1.11498922e-01 -1.15034914e+00 7.19215631e-01
6.50112033e-01 -1.21061432e+00 1.03106248e+00 -3.28125000e-01
4.38277394e-01 -2.65693873e-01 -3.17655623e-01 -8.96994114e-01
-4.67897922e-01 2.01344579e-01 4.83074069e-01 1.00206470e+00
2.67000109e-01 -6.75829768e-01 3.02745312e-01 2.80181598e-03
-5.17053157e-02 -1.17888057e+00 -9.53691363e-01 -7.80943751e-01
1.55151784e-01 -1.48951858e-01 5.65616667e-01 1.20525658e+00
-5.70240438e-01 2.04674646e-01 -3.76584113e-01 -3.66304107e-02
6.89694524e-01 1.32735237e-01 2.19878554e-01 -1.51170707e+00
-1.86886434e-02 -6.71247005e-01 -6.06868863e-01 -2.61311948e-01
7.22514018e-02 -8.86252820e-01 -6.42283112e-02 -1.44440377e+00
3.76716465e-01 -6.79196954e-01 -6.88470483e-01 3.17369491e-01
3.88564244e-02 1.51035294e-01 2.99061418e-01 2.09070534e-01
-5.21453798e-01 5.13885796e-01 8.67738426e-01 -1.10819913e-01
1.38440788e-01 2.97470123e-01 -3.88993025e-01 6.27084017e-01
8.30216646e-01 -4.38396722e-01 -3.09322447e-01 -2.87696451e-01
4.48152721e-01 -4.11074143e-03 4.10928994e-01 -1.12117457e+00
-1.98418707e-01 -1.19427025e-01 6.32214919e-02 -4.02574241e-01
5.47696128e-02 -7.93482065e-01 2.33321637e-01 6.49854898e-01
-5.17630994e-01 -3.54888588e-01 -1.18425898e-01 2.75620341e-01
-2.31263697e-01 -5.26051223e-01 1.41132987e+00 9.79550183e-02
-6.39550507e-01 2.05270633e-01 -1.15423962e-01 -1.01444356e-01
1.03589642e+00 -1.27324492e-01 -3.09125394e-01 -1.68554962e-01
-7.82961726e-01 1.12564892e-01 2.02671453e-01 1.83386430e-02
2.65281349e-01 -1.24403906e+00 -1.04840827e+00 -2.21210986e-01
2.66165406e-01 -2.14263991e-01 5.78545034e-01 8.01321387e-01
-7.95293868e-01 1.60886467e-01 -2.36640364e-01 -5.17325580e-01
-1.23313665e+00 2.39435002e-01 7.21005082e-01 -5.70279837e-01
-2.42875427e-01 8.40765476e-01 1.86883748e-01 -5.35460353e-01
3.90901595e-01 -3.94960195e-02 -5.39956987e-01 6.43862486e-01
4.15594131e-01 1.12666142e+00 3.50504816e-01 -8.84294391e-01
-5.72106361e-01 7.25420177e-01 3.79131198e-01 4.59236681e-01
1.43572474e+00 2.85402201e-02 -3.47205073e-01 5.97438216e-01
1.51829803e+00 -5.46514273e-01 -9.33173716e-01 -1.99364573e-01
-1.05212003e-01 -1.99778646e-01 1.69215411e-01 -8.73635232e-01
-1.16087210e+00 7.05497503e-01 1.09367168e+00 5.24168313e-01
1.42418063e+00 -3.87111455e-01 5.83987176e-01 2.38155961e-01
1.34454355e-01 -1.25375235e+00 -2.29240298e-01 3.85340810e-01
1.05478096e+00 -1.32084370e+00 2.78983176e-01 -3.31349999e-01
-4.41622853e-01 1.18345654e+00 1.33026987e-01 1.47558562e-02
1.08507156e+00 -2.09417999e-01 -6.50107861e-02 -4.65228915e-01
-8.02849326e-03 -4.92248088e-01 3.70042831e-01 3.42218369e-01
4.71827894e-01 2.19174400e-01 -9.22625244e-01 2.77388841e-01
8.51236209e-02 -2.08941132e-01 1.49212629e-01 4.49951291e-01
-5.67832947e-01 -3.45421731e-01 -2.90208638e-01 5.08873522e-01
-5.53309321e-01 2.30354816e-01 -2.48486370e-01 6.51913106e-01
5.52797318e-01 1.02643013e+00 -1.16950087e-01 -1.49396092e-01
4.77485090e-01 -2.79316783e-01 3.20734054e-01 -2.62453675e-01
-7.35656083e-01 -2.92689025e-01 4.04694490e-02 -4.02519464e-01
-7.29443491e-01 -6.14115655e-01 -8.22897077e-01 -2.03044161e-01
-4.46013272e-01 1.52278110e-01 1.00955856e+00 8.02576959e-01
-2.96033416e-02 3.49402577e-01 8.34607899e-01 -4.00349349e-01
-6.34934187e-01 -8.82942915e-01 -1.02548110e+00 4.06396002e-01
2.76183873e-01 -4.54205155e-01 -5.33460498e-01 -2.15482891e-01]
|
[9.119155883789062, 3.842449188232422]
|
ab01b3b6-bac6-452b-8dce-f877932d3baf
|
clip4clip-an-empirical-study-of-clip-for-end
|
2104.0886
| null |
https://arxiv.org/abs/2104.08860v2
|
https://arxiv.org/pdf/2104.08860v2.pdf
|
CLIP4Clip: An Empirical Study of CLIP for End to End Video Clip Retrieval
|
Video-text retrieval plays an essential role in multi-modal research and has been widely used in many real-world web applications. The CLIP (Contrastive Language-Image Pre-training), an image-language pre-training model, has demonstrated the power of visual concepts learning from web collected image-text datasets. In this paper, we propose a CLIP4Clip model to transfer the knowledge of the CLIP model to video-language retrieval in an end-to-end manner. Several questions are investigated via empirical studies: 1) Whether image feature is enough for video-text retrieval? 2) How a post-pretraining on a large-scale video-text dataset based on the CLIP affect the performance? 3) What is the practical mechanism to model temporal dependency between video frames? And 4) The Hyper-parameters sensitivity of the model on video-text retrieval task. Extensive experimental results present that the CLIP4Clip model transferred from the CLIP can achieve SOTA results on various video-text retrieval datasets, including MSR-VTT, MSVC, LSMDC, ActivityNet, and DiDeMo. We release our code at https://github.com/ArrowLuo/CLIP4Clip.
|
['Tianrui Li', 'Nan Duan', 'Wen Lei', 'Yang Chen', 'Ming Zhong', 'Lei Ji', 'Huaishao Luo']
|
2021-04-18
| null | null | null | null |
['video-text-retrieval']
|
['computer-vision']
|
[-1.56228617e-01 -1.07143509e+00 -6.11479521e-01 -7.78534710e-02
-9.91343677e-01 -4.47248369e-01 5.85716963e-01 -1.22041747e-01
-5.77648103e-01 1.57743096e-01 2.64371425e-01 -1.42708912e-01
-9.36838165e-02 -2.52013832e-01 -7.15604007e-01 -5.47790468e-01
-9.72771868e-02 8.78690258e-02 4.60936129e-01 -6.15670718e-02
3.70296389e-01 -1.28258578e-02 -1.57280946e+00 9.91670430e-01
4.21203762e-01 1.05675769e+00 5.94336867e-01 8.78841758e-01
-1.85210258e-01 1.10189486e+00 -2.78916955e-01 -1.57704338e-01
3.36879432e-01 -3.47487390e-01 -6.17743969e-01 3.94425094e-02
6.19310498e-01 -6.71034455e-01 -7.86353588e-01 8.83388579e-01
5.83721459e-01 1.26504272e-01 5.72111845e-01 -1.29669285e+00
-9.35568869e-01 3.59488308e-01 -9.34104443e-01 6.73199177e-01
4.22558218e-01 2.01498732e-01 8.73218298e-01 -1.17286801e+00
6.54026091e-01 1.46645367e+00 1.65080145e-01 3.22464556e-01
-5.56754827e-01 -8.30744267e-01 7.94798061e-02 8.56393456e-01
-1.77909613e+00 -4.20439154e-01 5.55024028e-01 -4.23589110e-01
7.66284287e-01 1.32685885e-01 5.39911687e-01 1.13539970e+00
4.33064073e-01 1.30355203e+00 8.25804830e-01 -5.04616499e-01
-2.76427925e-01 1.24974430e-01 6.95195124e-02 6.38072312e-01
-2.95271605e-01 3.97103056e-02 -9.48809922e-01 -2.37444360e-02
7.50510216e-01 2.34622344e-01 -3.73305917e-01 -1.15900815e-01
-1.26571178e+00 7.54630804e-01 1.97943121e-01 4.18686956e-01
-1.48236692e-01 5.78879595e-01 6.70400858e-01 6.40655518e-01
3.84379894e-01 -3.77193749e-01 -3.31596464e-01 -1.31699845e-01
-1.14135754e+00 -1.37492537e-01 3.90815973e-01 1.20817149e+00
7.02163041e-01 9.07262340e-02 -2.66670227e-01 1.12386823e+00
4.39830840e-01 9.50642824e-01 8.44877720e-01 -6.41168654e-01
6.08654618e-01 2.20902190e-02 -4.08366740e-01 -9.64085400e-01
1.36288747e-01 1.90385699e-01 -5.61701536e-01 -4.55571681e-01
-1.19354203e-01 2.44400464e-02 -9.83561218e-01 1.16027486e+00
-6.27406314e-02 3.78653944e-01 -1.17690347e-01 1.09768987e+00
1.00053608e+00 1.19996297e+00 2.81337529e-01 -2.96868503e-01
1.35048211e+00 -1.28568423e+00 -7.56983042e-01 -1.91400781e-01
5.04045248e-01 -1.08737612e+00 1.16566861e+00 3.05520087e-01
-1.00098872e+00 -6.98973954e-01 -7.83205688e-01 -1.03687026e-01
-4.23568338e-01 3.42079341e-01 2.37301752e-01 3.26982528e-01
-1.17164564e+00 5.78874126e-02 -5.66456258e-01 -6.04765892e-01
2.13221833e-01 4.88607325e-02 -2.28881195e-01 -3.96458000e-01
-1.33161259e+00 5.39703429e-01 6.62986875e-01 -1.23341471e-01
-1.01356316e+00 -5.82544625e-01 -4.69774693e-01 1.75771698e-01
6.89173818e-01 -5.93922436e-01 9.67258513e-01 -1.52966118e+00
-1.27300346e+00 8.92003357e-01 -2.41131946e-01 -2.36273557e-01
4.17494416e-01 -3.38768572e-01 -6.35632813e-01 1.07998848e+00
-9.03833471e-03 9.68601167e-01 1.42675292e+00 -1.17568910e+00
-5.77262878e-01 -1.63566172e-01 -8.61521661e-02 4.72897321e-01
-6.96156740e-01 4.21874344e-01 -1.52511942e+00 -9.48473811e-01
-2.70337671e-01 -1.04720747e+00 2.95434743e-01 3.09352785e-01
6.31445199e-02 -2.76651412e-01 1.12708974e+00 -7.08695054e-01
1.45237660e+00 -2.43376040e+00 -6.29403442e-02 6.72525540e-02
-1.27065897e-01 4.18497175e-01 -5.04238844e-01 7.44888783e-01
-5.67912385e-02 2.50851691e-01 4.89168555e-01 1.52162284e-01
-3.39067429e-01 -1.12177521e-01 -3.14117908e-01 4.14746493e-01
-2.04330623e-01 1.09397817e+00 -5.27699411e-01 -1.08274293e+00
5.24778962e-01 4.62424517e-01 -5.27084708e-01 1.41381636e-01
-1.42745823e-01 -7.02258870e-02 -6.40072882e-01 8.60152364e-01
4.13759798e-01 -5.20118713e-01 -9.77019072e-02 -5.14909506e-01
7.92443603e-02 -5.97752035e-01 -9.32972193e-01 1.88637114e+00
-2.56615728e-01 1.15344381e+00 -9.69994292e-02 -8.35749805e-01
3.64307165e-01 4.07588691e-01 6.66396558e-01 -1.22338951e+00
1.69004127e-01 -1.23787215e-02 -3.96133572e-01 -1.05261230e+00
6.51266038e-01 2.83690542e-01 2.03362793e-01 2.91761547e-01
2.47539371e-01 2.28034526e-01 4.76706564e-01 5.31202972e-01
6.23691857e-01 -8.57100561e-02 6.36475161e-02 -2.01121494e-01
8.04369509e-01 -7.30499625e-02 -1.02760678e-03 7.55172133e-01
-3.53159189e-01 9.54405427e-01 5.04525974e-02 -2.59463668e-01
-8.31331909e-01 -8.52872074e-01 4.23490182e-02 1.53268456e+00
5.20303369e-01 -7.18951225e-01 -3.37569952e-01 -3.81835312e-01
-6.21050075e-02 2.14960098e-01 -3.54480684e-01 -1.52460366e-01
-4.52213705e-01 -4.72319454e-01 5.96615911e-01 2.75444865e-01
5.99597573e-01 -1.10334074e+00 -3.24080020e-01 -7.72247016e-02
-5.34021556e-01 -1.53320754e+00 -1.10279095e+00 -3.86162430e-01
-7.17744350e-01 -1.05827498e+00 -1.00450373e+00 -1.02855217e+00
4.55307484e-01 9.97233391e-01 8.15028787e-01 4.70858127e-01
-4.23417807e-01 1.20178783e+00 -7.92397618e-01 -1.18078664e-01
-1.63824722e-01 -2.19088584e-01 -1.35257810e-01 1.31730810e-01
4.73689198e-01 -2.07882956e-01 -7.12183177e-01 3.62139434e-01
-1.43868530e+00 6.92984834e-02 6.65652335e-01 6.01132870e-01
6.01545513e-01 5.82499709e-03 3.24047863e-01 -3.97190720e-01
7.03013480e-01 -5.75297177e-01 -3.66504312e-01 6.41703784e-01
-4.42054331e-01 -2.97127664e-01 2.81628758e-01 -8.04602265e-01
-8.21131647e-01 -2.63547480e-01 1.07440025e-01 -1.23328733e+00
1.67162091e-01 8.42730820e-01 2.84756273e-01 4.72284779e-02
3.00647736e-01 6.50930583e-01 1.42624164e-02 -2.92442620e-01
2.84124315e-01 7.10744798e-01 2.71549761e-01 -7.17102945e-01
7.26466775e-01 5.07615209e-01 -4.96074080e-01 -1.31288898e+00
-4.54422623e-01 -9.84290957e-01 -4.41285551e-01 -5.23394823e-01
1.01746881e+00 -1.46995878e+00 -5.46788573e-01 4.28392887e-01
-9.55369473e-01 -3.19130182e-01 3.84039462e-01 5.28123260e-01
-3.70864689e-01 7.78028309e-01 -7.24456906e-01 -4.41800445e-01
-6.71506047e-01 -1.04264235e+00 9.67974305e-01 2.39740014e-01
4.64072526e-01 -9.45798457e-01 -1.72470957e-01 5.53756952e-01
4.12917316e-01 -5.49775064e-01 8.02269101e-01 -4.70643967e-01
-7.46720970e-01 1.13425730e-02 -4.83359307e-01 3.98110420e-01
-2.81912148e-01 2.04537496e-01 -6.53095901e-01 -6.94235981e-01
-2.68167228e-01 -5.77986896e-01 9.37850654e-01 5.69764972e-01
1.51503325e+00 -2.02728078e-01 -2.39840418e-01 5.19481599e-01
1.70734119e+00 3.06363553e-01 7.91084766e-01 4.08333987e-01
6.26004994e-01 1.27345353e-01 7.06122160e-01 2.94624031e-01
3.35712850e-01 6.28187060e-01 6.50993511e-02 2.06317529e-01
-3.25979680e-01 -2.18424797e-01 7.67059505e-01 1.22643793e+00
2.05629036e-01 -6.70457244e-01 -9.40644801e-01 5.22961557e-01
-1.87268376e+00 -1.09123468e+00 6.86717853e-02 1.86632013e+00
6.27514839e-01 -9.42007378e-02 -1.33309796e-01 -3.57737422e-01
7.76719213e-01 4.43634868e-01 -4.30490106e-01 2.00789183e-01
-1.75171867e-01 -2.43916824e-01 3.60398740e-01 2.33657390e-01
-1.08258986e+00 1.10439432e+00 5.60028648e+00 1.34907842e+00
-1.49521828e+00 2.33626619e-01 3.96753281e-01 -3.54977250e-01
1.92776144e-01 -6.20223768e-02 -7.25943327e-01 7.10103154e-01
8.93673956e-01 -5.79044402e-01 4.07528907e-01 6.16387963e-01
3.72052938e-01 -3.04516286e-01 -1.08548164e+00 1.51279807e+00
6.79891944e-01 -1.30882502e+00 6.87930167e-01 -1.81436598e-01
5.15740931e-01 2.24314556e-01 2.74725080e-01 4.48160052e-01
-3.96470010e-01 -5.77943683e-01 7.81104922e-01 3.45813692e-01
9.72947061e-01 -3.82242709e-01 4.14522380e-01 1.99754164e-01
-1.46463811e+00 -1.04008190e-01 -2.95819849e-01 6.85475051e-01
1.27457827e-01 9.10557061e-02 -4.52597350e-01 6.38189077e-01
1.05535972e+00 1.12118840e+00 -1.00770688e+00 1.10990858e+00
2.56750852e-01 6.91073179e-01 -1.19718432e-01 7.66289830e-02
2.66151607e-01 -9.25321355e-02 4.48762625e-01 1.63848841e+00
3.27112049e-01 5.89364879e-02 3.04486334e-01 2.73574769e-01
-1.29809380e-01 3.52125019e-01 -3.10678214e-01 -3.38974714e-01
3.63909274e-01 1.08575535e+00 -6.52116299e-01 -4.31022853e-01
-6.70360327e-01 1.05377054e+00 -2.23360881e-01 7.84094095e-01
-9.38387871e-01 -1.93455338e-01 1.99787751e-01 8.41496363e-02
5.90584695e-01 -1.87073275e-01 6.27545416e-01 -1.41669035e+00
8.26273113e-02 -1.12616742e+00 7.78416336e-01 -1.43728173e+00
-1.31017351e+00 3.11797976e-01 3.14679295e-01 -1.58171189e+00
-6.54124245e-02 -5.23450792e-01 -3.15345347e-01 2.87249714e-01
-1.76233995e+00 -1.23075187e+00 -4.19584841e-01 1.27831042e+00
1.16267729e+00 -3.82137358e-01 2.61913538e-01 8.44121635e-01
-4.21741128e-01 5.98165631e-01 3.80436301e-01 3.55077922e-01
1.12750077e+00 -4.43463266e-01 -4.42191213e-01 6.95532382e-01
3.19291711e-01 4.47127193e-01 3.14858407e-01 -3.98103416e-01
-1.92784202e+00 -9.34419811e-01 2.28833511e-01 -1.06474183e-01
8.51404786e-01 -1.34947728e-02 -8.86990488e-01 7.10320711e-01
7.16443777e-01 1.13190427e-01 6.02368176e-01 -4.29983526e-01
-3.48273665e-01 -2.37535149e-01 -5.78559041e-01 4.89746779e-01
6.16194844e-01 -8.41658831e-01 -4.78660226e-01 4.32418287e-01
7.20608413e-01 -2.53439933e-01 -6.61029577e-01 1.05333440e-01
6.56759202e-01 -6.58720970e-01 1.24297523e+00 -3.81557107e-01
6.87014520e-01 -1.72988683e-01 -3.73255223e-01 -9.07466650e-01
-3.36962976e-02 -2.81574488e-01 -2.05040395e-01 1.11169600e+00
-4.80717234e-03 -8.39473903e-02 3.57235372e-01 1.05996542e-01
2.34347686e-01 -3.92489254e-01 -7.88476348e-01 -7.59639621e-01
1.47955120e-01 -4.41234291e-01 -2.16562003e-01 9.55657423e-01
-1.94750473e-01 3.19367349e-01 -7.10497320e-01 3.05502228e-02
5.17514706e-01 -8.10435191e-02 6.87119246e-01 -8.05260539e-01
-1.79401159e-01 -3.79100084e-01 -2.46908650e-01 -1.38325059e+00
2.16846123e-01 -1.00259984e+00 -2.50523299e-01 -1.27888775e+00
7.40627348e-01 2.37808954e-02 -3.69396746e-01 2.32810169e-01
-1.60542801e-01 1.66850120e-01 6.87545061e-01 6.41028821e-01
-1.29441965e+00 5.14143825e-01 1.28215921e+00 -3.64471972e-01
1.34853711e-02 -5.07709980e-01 -2.37468705e-01 3.97477835e-01
4.65189070e-01 -4.15297806e-01 -5.71675897e-01 -8.08259845e-01
2.06147939e-01 5.15643418e-01 3.20498586e-01 -9.11302686e-01
6.39552176e-01 -2.61846393e-01 3.15097690e-01 -8.60321760e-01
3.65900218e-01 -1.06112731e+00 -2.60819215e-02 3.11243385e-01
-4.68628496e-01 2.66607732e-01 4.05559361e-01 7.55125642e-01
-5.26938319e-01 -2.49882802e-01 5.73040485e-01 -1.03058442e-01
-1.21481431e+00 5.34968317e-01 -6.40778720e-01 1.37109712e-01
1.00806248e+00 6.12127222e-03 -4.02432710e-01 -7.24224389e-01
-3.54009032e-01 6.19674325e-01 1.16640128e-01 1.02292037e+00
8.86359155e-01 -1.29723287e+00 -7.68460751e-01 2.59209778e-02
4.40428138e-01 -7.23858297e-01 5.70704520e-01 7.48743117e-01
-6.24920845e-01 6.53907776e-01 -5.05323038e-02 -1.00211000e+00
-1.65817153e+00 7.13743806e-01 1.60940900e-01 -5.25677241e-02
-5.48608899e-01 3.55711073e-01 3.17834646e-01 3.49057883e-01
4.54424113e-01 5.57429008e-02 -1.77660678e-02 1.44487113e-01
7.24154890e-01 9.59740058e-02 -2.09538341e-01 -6.97178483e-01
-2.06594571e-01 7.66760707e-01 -5.02455950e-01 -1.10132180e-01
1.07605648e+00 -5.86123288e-01 6.02877252e-02 3.69428664e-01
1.71431673e+00 -4.11573082e-01 -8.42830718e-01 -4.29199368e-01
-1.86769769e-01 -6.33876503e-01 3.21290046e-01 -6.11444890e-01
-1.42140555e+00 8.14284205e-01 9.75627244e-01 -1.50604159e-01
1.15462661e+00 4.20868304e-03 8.80692244e-01 6.70090795e-01
3.04557174e-01 -1.37055671e+00 7.75968492e-01 5.21210790e-01
9.92510259e-01 -1.53826225e+00 2.10645929e-01 -7.32284337e-02
-7.85651326e-01 1.27316654e+00 5.71754396e-01 -1.16195418e-02
1.07040167e+00 -8.76596496e-02 2.35366687e-01 -2.31885791e-01
-1.05792940e+00 -1.45356908e-01 4.43144143e-01 2.76230633e-01
1.89842924e-01 -3.85546237e-01 -2.34496415e-01 7.40947872e-02
4.31630462e-01 3.30843240e-01 3.86326551e-01 1.17796135e+00
-4.24923807e-01 -7.55675912e-01 -3.89134616e-01 2.54846543e-01
-6.91225231e-01 -3.91801089e-01 -1.32403627e-01 8.26902211e-01
-4.29479748e-01 9.03221488e-01 -3.74628627e-03 -3.36774617e-01
7.26997778e-02 5.11069372e-02 3.57440919e-01 -2.54476994e-01
-4.00358945e-01 5.47506511e-01 -3.61306578e-01 -5.23223817e-01
-7.62101412e-01 -3.37556481e-01 -1.10366189e+00 -3.95786166e-01
-3.82493615e-01 1.20301202e-01 5.48967242e-01 7.51777172e-01
4.54765052e-01 2.42495090e-01 6.21517718e-01 -6.82199538e-01
-1.10235862e-01 -7.58110166e-01 -4.43686157e-01 4.76687849e-01
2.01576129e-01 -3.93778861e-01 -5.40099204e-01 6.50879383e-01]
|
[10.324419975280762, 0.9445454478263855]
|
1ffee44d-4b62-467e-b44d-6fac6aba5364
|
transformers-for-headline-selection-for
|
2106.10487
| null |
https://arxiv.org/abs/2106.10487v1
|
https://arxiv.org/pdf/2106.10487v1.pdf
|
Transformers for Headline Selection for Russian News Clusters
|
In this paper, we explore various multilingual and Russian pre-trained transformer-based models for the Dialogue Evaluation 2021 shared task on headline selection. Our experiments show that the combined approach is superior to individual multilingual and monolingual models. We present an analysis of a number of ways to obtain sentence embeddings and learn a ranking model on top of them. We achieve the result of 87.28% and 86.60% accuracy for the public and private test sets respectively.
|
['Olga Sopilnyak', 'Pavel Voropaev']
|
2021-06-19
| null | null | null | null |
['dialogue-evaluation']
|
['natural-language-processing']
|
[-3.69951129e-01 3.12721342e-01 -8.52166787e-02 -8.11263740e-01
-1.43023360e+00 -6.86558783e-01 1.06896091e+00 2.33241528e-01
-9.85306263e-01 1.23513985e+00 8.25090766e-01 -4.44283038e-01
1.70552045e-01 -2.77102500e-01 1.03035616e-02 -1.23572037e-01
1.13055460e-01 9.12545741e-01 2.68506378e-01 -9.07856703e-01
8.00532624e-02 6.82634786e-02 -9.50607002e-01 5.52265942e-01
8.57844591e-01 5.93160212e-01 3.66870239e-02 1.02576149e+00
4.75265533e-02 8.52424204e-01 -9.44820642e-01 -1.00956273e+00
-1.85588717e-01 -1.81257874e-01 -1.32953072e+00 -4.05481756e-01
5.18318594e-01 -3.01557899e-01 -5.10689855e-01 8.46938014e-01
1.07776964e+00 3.17723572e-01 6.93370581e-01 -6.36177659e-01
-6.94536150e-01 1.12195575e+00 4.08256091e-02 3.68294388e-01
1.00200868e+00 -2.80585259e-01 1.51907873e+00 -1.34745824e+00
7.71984637e-01 1.40785253e+00 3.87833118e-01 7.56001234e-01
-8.99399519e-01 -2.93377191e-01 -7.40639716e-02 3.01546574e-01
-1.15722013e+00 -6.92681432e-01 4.46522415e-01 -6.97238818e-02
1.32457411e+00 5.60572505e-01 3.43966901e-01 1.11502647e+00
1.41921744e-01 1.03590536e+00 1.34700453e+00 -6.88675702e-01
-5.92514932e-01 7.85385311e-01 5.19147158e-01 8.65621746e-01
-3.97335440e-01 -2.67945051e-01 -8.48656654e-01 -3.25966537e-01
6.45891502e-02 -8.88309062e-01 -4.69978094e-01 1.30038530e-01
-1.07317889e+00 1.00136173e+00 -3.46455164e-03 3.08984190e-01
2.14616179e-01 -6.57861650e-01 1.02836668e+00 7.94231832e-01
7.28171825e-01 8.06131959e-01 -8.70269835e-01 -2.41059437e-01
-5.50207973e-01 3.02092016e-01 9.74917114e-01 8.93907964e-01
2.20944285e-01 -1.53574392e-01 -5.81946254e-01 1.55632484e+00
2.46697724e-01 3.95727932e-01 6.93582237e-01 -5.31241119e-01
9.48736668e-01 1.57067940e-01 1.35833219e-01 -4.19842929e-01
-4.34582770e-01 -2.89087385e-01 -4.54115152e-01 -3.60863894e-01
5.60901582e-01 -3.38705957e-01 -2.47968152e-01 1.57899439e+00
1.79120563e-02 -7.07488954e-01 5.47631741e-01 4.78975177e-01
1.28653622e+00 7.69568443e-01 7.16490000e-02 -1.18727118e-01
1.39798653e+00 -1.26392019e+00 -1.07891810e+00 3.58574130e-02
1.04419577e+00 -1.24413586e+00 1.36377764e+00 2.16347247e-01
-1.27139068e+00 -4.97968882e-01 -9.71790850e-01 -4.44720566e-01
-5.22349894e-01 6.14519715e-01 3.47927481e-01 6.50615931e-01
-1.25929129e+00 4.07346696e-01 -1.74643192e-02 -4.47133034e-01
-4.46118385e-01 2.21698806e-01 -5.24942458e-01 9.68052913e-03
-1.86275196e+00 1.66251683e+00 1.09452061e-01 9.30693746e-02
-7.78589964e-01 -3.02194476e-01 -9.74709094e-01 -3.51988316e-01
-4.13259864e-01 -1.52475089e-01 1.61370122e+00 -2.42369607e-01
-1.87475026e+00 1.32449555e+00 -1.76154584e-01 -6.34401679e-01
4.05891776e-01 -5.64273000e-01 -5.96845567e-01 -1.03563659e-01
3.10286768e-02 4.37938333e-01 1.74756516e-02 -6.10416472e-01
-9.88805950e-01 -4.13780399e-02 2.67664343e-01 8.80335569e-01
-5.13799548e-01 6.91054523e-01 -1.61721349e-01 -2.45172322e-01
-4.52762187e-01 -6.66099608e-01 -4.60294671e-02 -8.20134699e-01
-6.72171056e-01 -8.94003570e-01 4.82966661e-01 -1.24348736e+00
1.39272249e+00 -1.53076327e+00 4.41190064e-01 -3.03287596e-01
-1.31458610e-01 5.05289257e-01 -1.12069733e-02 7.65428841e-01
7.31344223e-02 5.43223098e-02 3.50214124e-01 -7.04935372e-01
7.97828957e-02 -1.99957833e-01 -2.65319049e-01 3.54107380e-01
1.05386958e-01 7.07329810e-01 -9.04923201e-01 -5.60266018e-01
3.32481623e-01 3.27586979e-01 -1.97684854e-01 6.34975851e-01
2.44258389e-01 5.07035851e-01 -2.73822635e-01 3.44112486e-01
2.14861065e-01 5.85074604e-01 4.52834487e-01 -1.57304436e-01
-1.61648110e-01 1.41228747e+00 -5.11954963e-01 1.62824094e+00
-9.82041240e-01 8.65724981e-01 -1.20010495e-01 -3.44756156e-01
1.10578799e+00 8.48105907e-01 -8.97595510e-02 -6.62343383e-01
2.10218787e-01 5.49140930e-01 -2.15799958e-02 -3.48242074e-01
1.08623004e+00 2.54360419e-02 -6.80662632e-01 4.03323412e-01
5.99288166e-01 -3.30914631e-02 2.91514844e-01 4.35697168e-01
8.56286407e-01 -1.54623181e-01 4.03877795e-01 -5.35420775e-01
1.13771832e+00 -2.98535645e-01 2.52992392e-01 5.11153162e-01
-4.22754318e-01 4.57733542e-01 3.17385316e-01 -4.22472894e-01
-8.11098695e-01 -9.56614614e-01 -3.53612602e-01 1.45720625e+00
-3.65287066e-01 -7.47547925e-01 -6.13678157e-01 -1.16363764e+00
-3.34749490e-01 9.75444078e-01 -4.76002663e-01 -7.72899538e-02
-8.54643464e-01 -6.61417305e-01 8.28685701e-01 2.62274683e-01
2.29546443e-01 -8.43418539e-01 2.18414545e-01 1.70058265e-01
-5.63696325e-01 -1.20491445e+00 -6.99414313e-01 2.38817006e-01
-4.61562842e-01 -9.46178496e-01 -6.20762944e-01 -1.20286226e+00
1.94323331e-01 -2.08137617e-01 1.35701382e+00 -2.94852823e-01
3.48332942e-01 8.63029584e-02 -4.52996224e-01 -6.15609139e-02
-5.70803642e-01 6.50740266e-01 1.00872368e-01 -3.11863244e-01
5.59703887e-01 1.50863603e-01 1.52541071e-01 5.12317456e-02
2.30983440e-02 -1.95688099e-01 -6.89564124e-02 1.33984005e+00
-1.84621513e-02 -5.30674934e-01 7.79637396e-01 -1.05476427e+00
1.43886375e+00 -1.87288940e-01 -2.39518970e-01 5.86601079e-01
-5.50262392e-01 1.80438057e-01 6.73970282e-01 -1.41028836e-01
-1.18403053e+00 -2.36675665e-01 -6.77848339e-01 3.13475907e-01
2.49250263e-01 4.93331283e-01 -2.74649471e-01 1.57012671e-01
3.57813418e-01 3.29527296e-02 -2.78221548e-01 -8.48417461e-01
4.40177381e-01 1.17151463e+00 1.69517666e-01 -5.46602428e-01
3.05681080e-01 -5.19872785e-01 -7.98478484e-01 -8.50707233e-01
-1.10008442e+00 -4.70509499e-01 -7.11281598e-01 -3.43283504e-01
8.42716157e-01 -9.65096116e-01 -3.27359796e-01 2.61146575e-01
-1.36855662e+00 -4.44667526e-02 -1.75300747e-01 5.25405586e-01
-3.45709771e-01 1.50020748e-01 -1.07147717e+00 -7.93848455e-01
-6.03179812e-01 -1.36307025e+00 8.35048258e-01 1.03835300e-01
-5.21319330e-01 -1.36827147e+00 4.38029766e-01 4.46701288e-01
3.53635013e-01 -4.80642110e-01 7.02721715e-01 -1.31451464e+00
3.25966537e-01 -3.36730033e-01 1.46540269e-01 5.03679812e-01
1.25542283e-02 8.00259858e-02 -1.13564670e+00 -4.20564920e-01
-1.67526603e-01 -7.50016868e-01 6.08693123e-01 3.08926031e-02
5.09044707e-01 -1.27485096e-01 -1.55742673e-04 1.75034061e-01
9.29897308e-01 8.16203505e-02 4.73562211e-01 3.53741944e-01
5.71358263e-01 1.03081262e+00 9.65589702e-01 2.50040293e-01
8.55932713e-01 9.86762404e-01 -1.44116253e-01 -1.23004317e-01
-1.57124460e-01 -4.55429077e-01 6.01990998e-01 1.65688324e+00
3.56991112e-01 -2.92701006e-01 -7.84789443e-01 7.27295756e-01
-1.51043785e+00 -8.32567036e-01 -1.51541829e-01 2.18258595e+00
1.46323299e+00 1.27285525e-01 3.02734733e-01 1.44251613e-02
7.47906029e-01 3.30532342e-01 3.81614387e-01 -1.15866327e+00
-5.29915512e-01 3.99283886e-01 5.19074023e-01 1.18025935e+00
-1.26196051e+00 1.39834559e+00 7.24041367e+00 7.78106093e-01
-5.62144876e-01 4.60718393e-01 6.00823641e-01 2.43876055e-01
-2.60011166e-01 -1.07810721e-01 -1.47700357e+00 -2.07127742e-02
1.50485599e+00 -6.41720772e-01 8.52638762e-03 4.86415386e-01
-1.50450259e-01 2.38278404e-01 -1.01965165e+00 5.83347499e-01
2.74464905e-01 -9.71978009e-01 7.79365599e-02 -3.72764438e-01
5.03455937e-01 1.82494536e-01 -1.17077986e-02 6.73411310e-01
6.80815578e-01 -1.07044983e+00 4.60188329e-01 2.93266147e-01
8.15682650e-01 -1.01344895e+00 1.13750982e+00 5.08580022e-02
-1.05063903e+00 1.93214059e-01 -4.10206228e-01 1.27123907e-01
3.44546258e-01 7.10227937e-02 -1.05838525e+00 6.41024113e-01
4.44360226e-01 6.88053370e-01 -7.89026201e-01 8.25116277e-01
-6.19546592e-01 5.90428472e-01 6.10908912e-03 -4.87436801e-01
3.03248882e-01 -4.40094098e-02 5.74921131e-01 1.59561360e+00
-1.85854465e-01 -6.67940557e-01 1.64822757e-01 -1.77106813e-01
-2.16149390e-01 7.54919708e-01 -8.31513524e-01 3.00764740e-01
5.94422996e-01 1.19081664e+00 2.75058270e-01 -5.26719332e-01
-4.32432801e-01 8.64205360e-01 8.50022912e-01 8.62410963e-02
-6.55790508e-01 -8.04183245e-01 6.89597607e-01 -4.54205364e-01
-1.63969040e-01 -9.50415507e-02 7.16240332e-02 -1.29420865e+00
-6.29236624e-02 -1.19869673e+00 5.91172814e-01 -3.44883353e-01
-1.25769925e+00 1.37717247e+00 1.01985447e-02 -1.12744796e+00
-5.37606299e-01 -7.20609605e-01 -5.28741479e-01 1.27451265e+00
-1.59261966e+00 -1.07558727e+00 5.57219744e-01 4.04539376e-01
8.63882124e-01 -7.99927473e-01 1.46983600e+00 6.01490855e-01
-6.36568725e-01 1.12519836e+00 3.39538276e-01 3.59477222e-01
1.26937222e+00 -1.66201115e+00 4.26583529e-01 5.49031913e-01
1.80699989e-01 5.93998790e-01 7.92656898e-01 -1.79784894e-01
-8.67127419e-01 -8.65971029e-01 2.07608414e+00 -7.50122249e-01
9.03857350e-01 -5.50103068e-01 -5.99386334e-01 7.13153839e-01
1.02513969e+00 -5.60690641e-01 8.23932827e-01 7.34362602e-01
-2.33917505e-01 -5.95440492e-02 -1.03994465e+00 7.16579735e-01
4.72415388e-01 -1.02970314e+00 -8.33703041e-01 7.43442953e-01
7.14851201e-01 -5.30281901e-01 -1.25026178e+00 2.24910095e-01
4.19904202e-01 -7.24726439e-01 9.13993001e-01 -8.82755339e-01
1.64197683e-01 1.43746361e-01 -2.22726360e-01 -1.64817846e+00
-9.66688897e-03 -6.12424195e-01 2.35368609e-01 1.45143163e+00
1.03916287e+00 -7.81987429e-01 1.77942917e-01 3.55735689e-01
-3.39841038e-01 -7.99299896e-01 -1.19765222e+00 -5.10078013e-01
5.14403045e-01 -5.74613102e-02 3.31480384e-01 8.20383728e-01
5.09682775e-01 1.23921144e+00 -5.84934175e-01 -2.92171329e-01
2.13676348e-01 -2.31625512e-01 5.60376823e-01 -7.97448814e-01
-9.70038399e-02 -2.76135355e-01 -1.79114699e-01 -1.15867066e+00
6.86676741e-01 -9.57366467e-01 -8.40679482e-02 -1.49028111e+00
-4.35979217e-02 -3.41731489e-01 -4.40847933e-01 1.35135233e-01
-3.31603199e-01 4.97287922e-02 -4.00832780e-02 -1.45453334e-01
-7.04389989e-01 8.44621062e-01 8.64833772e-01 -2.89315820e-01
1.06662057e-01 9.83085334e-02 -4.63951021e-01 3.56621653e-01
1.17411137e+00 -4.68582779e-01 -2.61483610e-01 -6.90329909e-01
-1.16642624e-01 2.33949587e-01 -3.62559408e-01 -6.28941000e-01
-6.44157082e-02 3.62862796e-01 -1.17938258e-01 -6.97060406e-01
6.01437032e-01 -2.22424436e-02 -7.22705245e-01 3.08655262e-01
-1.06165409e+00 6.79439306e-01 6.27554208e-02 1.30810181e-03
-6.27257884e-01 -5.24273217e-01 6.44656956e-01 6.16345517e-02
-3.35463494e-01 -2.95413375e-01 -6.19821668e-01 3.98538500e-01
4.52896208e-01 5.40974081e-01 -4.89279568e-01 -5.53934753e-01
-7.39672899e-01 4.26510900e-01 6.03200942e-02 7.80906677e-01
5.82698405e-01 -1.39353287e+00 -1.17897582e+00 1.88651867e-02
2.76476949e-01 -1.02045906e+00 -2.31719464e-01 8.29504073e-01
-4.31576580e-01 9.21830297e-01 -1.03388414e-01 -2.94276346e-02
-1.80128479e+00 -1.68850139e-01 5.15344262e-01 -9.09889996e-01
-1.71201110e-01 1.24347699e+00 -3.24442059e-01 -1.26126814e+00
3.82350415e-01 2.36310259e-01 -9.53965664e-01 1.47536248e-01
5.21380126e-01 3.04672718e-01 2.18677789e-01 -1.24204147e+00
-3.91868204e-01 1.18342772e-01 -5.52615702e-01 -5.42302728e-01
8.83716643e-01 -3.00959378e-01 -2.49759912e-01 8.59148204e-01
1.49881947e+00 3.43932182e-01 -1.95100144e-01 -4.81664926e-01
1.94145441e-01 -2.92052984e-01 -7.13340417e-02 -8.96667540e-01
-6.28234625e-01 1.05221164e+00 4.35709715e-01 2.61072993e-01
6.91334605e-01 -1.24804094e-01 6.89056873e-01 5.45061886e-01
3.48198354e-01 -1.46726596e+00 -3.90578330e-01 1.14834023e+00
1.05175519e+00 -1.21862555e+00 -1.72658876e-01 -2.35907748e-01
-9.68611836e-01 1.11851704e+00 6.45791471e-01 4.27510543e-03
4.25501585e-01 -4.45310138e-02 4.27586317e-01 -5.51365502e-02
-1.29905868e+00 -1.43725350e-01 4.03034866e-01 5.31715333e-01
1.35850108e+00 2.09707022e-01 -9.29762602e-01 3.97809714e-01
-6.89967394e-01 -8.23614299e-01 4.73428935e-01 7.08878815e-01
-2.39480346e-01 -1.62999380e+00 1.30966425e-01 3.66168112e-01
-8.11003149e-01 -4.31365490e-01 -7.84837425e-01 7.00726986e-01
-6.32695019e-01 1.18170059e+00 -3.82762998e-01 -7.99713075e-01
6.37872994e-01 4.82260942e-01 4.73501474e-01 -8.77310455e-01
-1.23718011e+00 -3.32605779e-01 1.37735796e+00 -1.69278949e-01
-1.25028148e-01 -6.82024300e-01 -6.08506680e-01 -1.45144567e-01
-6.32064402e-01 7.80377984e-01 5.43630183e-01 8.09081018e-01
-1.25097230e-01 4.57015216e-01 9.90457892e-01 -3.73993188e-01
-1.02580869e+00 -1.51953733e+00 -4.82279927e-01 1.93944663e-01
6.85105845e-02 -4.48410839e-01 -4.44580078e-01 -4.21491027e-01]
|
[12.44701862335205, 8.382412910461426]
|
a7371b61-2d50-4069-bf51-2de2ca514e5f
|
machine-learning-models-for-dota-2-outcomes
|
2106.01782
| null |
https://arxiv.org/abs/2106.01782v1
|
https://arxiv.org/pdf/2106.01782v1.pdf
|
Machine learning models for DOTA 2 outcomes prediction
|
Prediction of the real-time multiplayer online battle arena (MOBA) games' match outcome is one of the most important and exciting tasks in Esports analytical research. This research paper predominantly focuses on building predictive machine and deep learning models to identify the outcome of the Dota 2 MOBA game using the new method of multi-forward steps predictions. Three models were investigated and compared: Linear Regression (LR), Neural Networks (NN), and a type of recurrent neural network Long Short-Term Memory (LSTM). In order to achieve the goals, we developed a data collecting python server using Game State Integration (GSI) to track the real-time data of the players. Once the exploratory feature analysis and tuning hyper-parameters were done, our models' experiments took place on different players with dissimilar backgrounds of playing experiences. The achieved accuracy scores depend on the multi-forward prediction parameters, which for the worse case in linear regression 69\% but on average 82\%, while in the deep learning models hit the utmost accuracy of prediction on average 88\% for NN, and 93\% for LSTM models.
|
['Anh Huy Phan', 'Kodirjon Akhmedov']
|
2021-06-03
| null | null | null | null |
['dota-2']
|
['playing-games']
|
[-3.65319222e-01 -1.21557005e-01 2.79531274e-02 -1.51222795e-01
-5.24789035e-01 -1.17497154e-01 4.68610108e-01 5.43673225e-02
-8.49565506e-01 6.80932820e-01 1.55277759e-01 -3.62633854e-01
-7.26675212e-01 -1.03355038e+00 -4.36382949e-01 -5.08995235e-01
-4.07096267e-01 8.12585711e-01 3.52769911e-01 -8.75571668e-01
4.46352214e-01 3.14028978e-01 -1.71688235e+00 8.37842286e-01
1.81076020e-01 1.24355662e+00 -6.27057701e-02 1.02903032e+00
-1.44675478e-01 1.69453490e+00 -5.53688407e-01 -7.27879643e-01
3.82824123e-01 -1.92893788e-01 -1.05877948e+00 -1.06539178e+00
-4.67252016e-01 1.27039090e-01 -4.51908082e-01 5.20592630e-01
7.00755060e-01 4.50367004e-01 2.84368008e-01 -1.01179051e+00
1.98281437e-01 8.24885190e-01 -2.13953823e-01 6.06917381e-01
2.60226578e-01 2.76649386e-01 7.09245026e-01 -3.00264597e-01
7.08204865e-01 7.26866901e-01 1.29677439e+00 4.19996589e-01
-7.79437423e-01 -8.39231372e-01 -3.25282484e-01 8.51514399e-01
-1.10726738e+00 -2.00417057e-01 6.00650012e-01 -6.06987417e-01
1.13318264e+00 2.98867971e-01 9.31724131e-01 1.19464326e+00
6.96641207e-01 3.75288993e-01 1.25099802e+00 -2.35829711e-01
-2.97574792e-03 1.81557223e-01 3.40259045e-01 3.68166715e-01
-4.54030216e-01 5.40293694e-01 -7.81390548e-01 8.84560682e-03
5.66166162e-01 -1.61106050e-01 5.89419425e-01 4.10089195e-01
-6.33196950e-01 8.84328365e-01 3.48517954e-01 6.88253939e-01
-7.90527105e-01 -3.59517559e-02 9.33217525e-01 5.57368815e-01
4.14176673e-01 6.04455829e-01 -5.09557843e-01 -1.04155385e+00
-8.36539388e-01 6.24201775e-01 8.73270452e-01 9.26064029e-02
2.87691921e-01 2.01943606e-01 -2.99508840e-01 9.27423716e-01
-2.59642363e-01 -3.71241063e-01 1.20790291e+00 -5.99956810e-01
3.38647813e-01 6.30912960e-01 -7.46968836e-02 -1.26131213e+00
-8.22030008e-01 -8.59518647e-01 -6.90247893e-01 5.36690533e-01
7.42690444e-01 -2.52724022e-01 -3.28601629e-01 1.50531054e+00
-1.56108841e-01 1.10201865e-01 8.31044316e-02 8.75414193e-01
1.01443827e+00 7.76084006e-01 4.01877105e-01 -1.45475399e-02
1.10519087e+00 -6.65545225e-01 -5.60452104e-01 -2.51073778e-01
8.51871848e-01 -6.96874559e-01 7.27504730e-01 6.30164146e-01
-1.24341142e+00 -8.85387063e-01 -8.39141488e-01 2.50852555e-01
-5.28506219e-01 -1.20587103e-01 7.53275573e-01 4.00876254e-01
-9.43335474e-01 1.22594047e+00 -6.13122344e-01 -2.03462884e-01
-9.09180418e-02 7.45115459e-01 -2.71177769e-01 6.93405688e-01
-1.43142664e+00 1.25830078e+00 5.91417611e-01 2.44174361e-01
-5.74796319e-01 -5.77489674e-01 -3.32931250e-01 -3.06767095e-02
1.79951623e-01 -2.37341449e-01 1.19029796e+00 -1.04640007e+00
-1.77883947e+00 1.12635660e+00 3.81326646e-01 -8.66741359e-01
7.45589077e-01 -1.77864820e-01 -3.42627466e-01 -7.91356206e-01
-9.69771668e-02 7.68851563e-02 -4.64774948e-03 -6.85540795e-01
-7.54815936e-01 -6.21475279e-01 9.60889552e-03 2.56055087e-01
-3.36406827e-02 4.07280475e-01 9.96962413e-02 -2.39252970e-01
-7.15231299e-02 -8.62065434e-01 -3.77376676e-01 -9.67299044e-01
-5.29627278e-02 -3.16846579e-01 2.16405898e-01 -1.07444072e+00
1.53937483e+00 -2.00416899e+00 7.43937045e-02 3.46468896e-01
2.44375430e-02 4.49180692e-01 2.90418804e-01 6.35536075e-01
-3.78587604e-01 -2.88110435e-01 7.05052614e-01 -1.42252803e-01
-9.54572260e-02 5.21754101e-03 -3.48730803e-01 -3.28505710e-02
-6.21589005e-01 8.19652855e-01 -3.69262218e-01 -4.67523545e-01
2.54305720e-01 6.69215340e-03 -3.79697144e-01 2.88211256e-01
1.04354367e-01 3.86226326e-01 -2.20405430e-01 1.00791469e-01
1.26395881e-01 3.78489017e-01 1.47635356e-01 8.20899159e-02
-5.32268107e-01 3.37257743e-01 -1.09753335e+00 1.60508192e+00
-5.68290949e-01 6.66384757e-01 -2.28925467e-01 -1.18927765e+00
1.36793125e+00 3.02209705e-01 6.15135968e-01 -1.14040923e+00
5.67015588e-01 3.94884139e-01 4.42715853e-01 -6.91974759e-01
8.24883819e-01 -3.51661772e-01 -2.55219460e-01 2.88843185e-01
1.85120508e-01 5.88992059e-01 9.08965245e-02 -2.35504121e-01
1.08387482e+00 3.79952639e-01 5.93203045e-02 8.35790019e-03
3.26825559e-01 2.12102994e-01 7.19624698e-01 9.57291663e-01
-2.47725129e-01 2.13219583e-01 6.38887465e-01 -1.04779935e+00
-9.43937838e-01 -4.07051742e-01 3.19631904e-01 1.68045425e+00
-2.28880897e-01 -5.07240713e-01 -5.62560499e-01 -2.19178409e-03
-4.98832524e-01 5.75732768e-01 -7.75663793e-01 -4.68049407e-01
-6.35568380e-01 -9.09685910e-01 7.84420252e-01 4.33485687e-01
4.19311851e-01 -1.55833888e+00 -6.62461579e-01 6.06683612e-01
-1.95639238e-01 -6.84305966e-01 6.02067411e-01 5.32411158e-01
-7.18554318e-01 -8.95377576e-01 -1.35766864e-01 -5.97672939e-01
-6.72117054e-01 -6.33962214e-01 1.10100985e+00 -2.33245447e-01
-2.96051562e-01 -3.83908927e-01 -2.30120555e-01 -5.76316357e-01
-4.07170922e-01 3.86233121e-01 2.15201396e-02 -1.25497639e-01
5.03262162e-01 -1.03844440e+00 -3.17609310e-01 2.03097180e-01
-2.66031027e-01 2.83625692e-01 5.59265316e-01 6.67481840e-01
2.10488349e-01 -1.22559564e-02 4.25035626e-01 -8.80838990e-01
9.36052203e-01 -6.83083296e-01 -2.67852038e-01 -1.06198557e-01
-4.86319542e-01 -1.24169946e-01 7.35975921e-01 -5.47848165e-01
-7.89083064e-01 -4.10618007e-01 -6.43863440e-01 -3.60019773e-01
-1.55726939e-01 7.86879718e-01 3.98732245e-01 4.02119458e-02
1.02483928e+00 2.18439326e-01 5.30299619e-02 -5.64876258e-01
-3.67238492e-01 5.01582086e-01 5.49674749e-01 -3.35512400e-01
2.39539966e-01 -1.12769634e-01 -3.01459357e-02 -3.50181878e-01
-5.40041804e-01 -1.57867998e-01 -6.19456947e-01 -9.99463081e-01
8.24707448e-01 -6.04938984e-01 -1.53659463e+00 6.20114744e-01
-7.90669560e-01 -5.09281993e-01 -4.52083945e-01 6.92727149e-01
-6.07818842e-01 -4.33525145e-01 -8.94891262e-01 -1.02399075e+00
-6.03492439e-01 -9.09640849e-01 1.11447424e-01 3.95664364e-01
-5.64621091e-01 -8.06442857e-01 5.64735591e-01 6.67041063e-01
6.00695968e-01 3.32648486e-01 8.85644436e-01 -1.14660013e+00
2.42093265e-01 -5.57006776e-01 1.66081935e-01 -4.71749417e-02
-6.14940107e-01 -1.81554690e-01 -9.25697625e-01 2.55005121e-01
-7.64105991e-02 -3.53011161e-01 3.84105206e-01 6.36217892e-01
9.78971541e-01 -4.26751003e-02 1.04668811e-01 5.32371581e-01
1.22001195e+00 6.80762768e-01 8.51244330e-01 1.06254768e+00
3.26533854e-01 6.62567914e-01 7.83266783e-01 6.04016721e-01
2.17963412e-01 9.48483229e-01 3.72879297e-01 1.00094981e-01
2.27458343e-01 -3.47288936e-01 5.48866808e-01 6.64914429e-01
-8.65818679e-01 2.83406109e-01 -9.54367876e-01 2.67878175e-01
-2.02635789e+00 -1.40979981e+00 -4.99154061e-01 2.25906253e+00
3.78787220e-01 7.10619867e-01 7.53784001e-01 5.00454605e-01
5.49006462e-01 8.78214538e-02 -2.40246400e-01 -1.23487091e+00
9.77701470e-02 6.17244422e-01 4.13464695e-01 2.52754539e-01
-9.40806031e-01 1.01728833e+00 5.96817732e+00 1.40842748e+00
-1.23519301e+00 3.40251446e-01 6.88610315e-01 -5.76037526e-01
3.80821109e-01 -1.26113921e-01 -5.16106486e-01 4.55002695e-01
1.67227817e+00 -7.12987483e-02 4.57635462e-01 9.79374051e-01
4.67733860e-01 -4.57573086e-02 -4.68435317e-01 1.10433996e+00
-3.16930056e-01 -1.43526340e+00 -3.45723152e-01 3.12534213e-01
9.29450467e-02 1.83116004e-01 3.25347222e-02 1.13164520e+00
3.89527231e-01 -1.11891341e+00 7.29951918e-01 1.00900376e+00
3.27170312e-01 -1.12011468e+00 1.11027145e+00 7.23670065e-01
-8.29404414e-01 -7.12037086e-01 -3.01586807e-01 -9.56436098e-01
-1.40720502e-01 8.74525309e-02 -6.86535239e-01 3.88822019e-01
1.08773685e+00 3.95491153e-01 -5.64730406e-01 8.21938932e-01
4.50859457e-01 8.06767702e-01 -3.20135921e-01 -4.01958108e-01
5.45789003e-01 -3.36527467e-01 5.20884395e-01 9.26410496e-01
2.01438338e-01 7.57701993e-02 -1.78604782e-01 6.02579832e-01
3.88166815e-01 4.12320614e-01 -5.12863040e-01 1.92117006e-01
-5.73444515e-02 1.22834611e+00 -4.64543670e-01 1.23233035e-01
2.90154785e-01 4.58481461e-01 4.89750445e-01 -2.36629635e-01
-9.01992559e-01 -1.71798497e-01 5.25270700e-01 4.75240260e-01
-2.49801919e-01 6.93345964e-02 -5.83507061e-01 -6.41949534e-01
-2.57609904e-01 -6.96796715e-01 7.04939425e-01 -7.25138962e-01
-9.59511101e-01 9.36699808e-01 -1.62117198e-01 -1.16869283e+00
-8.00297201e-01 -6.58876419e-01 -9.97139037e-01 8.83882344e-01
-6.20996118e-01 -1.19873917e+00 -2.01817185e-01 6.46106303e-01
3.80950809e-01 -6.18892014e-01 1.01037276e+00 4.03299838e-01
-7.53862381e-01 4.75898921e-01 2.10417837e-01 2.06803098e-01
2.99150616e-01 -8.41180682e-01 -2.00573336e-02 4.01527971e-01
-1.06881440e-01 3.64066847e-02 9.03178811e-01 -4.24003363e-01
-8.97004306e-01 -5.28092682e-01 8.47892702e-01 6.93983957e-03
6.95913851e-01 -1.27128199e-01 -6.35503352e-01 7.66293466e-01
-6.27966672e-02 -5.13973951e-01 9.18392062e-01 7.36291945e-01
2.65408993e-01 -2.96514004e-01 -7.26524293e-01 4.59840059e-01
7.89295793e-01 -2.67836064e-01 -3.83689791e-01 1.03772379e-01
-1.42274529e-01 -4.53492373e-01 -1.17223406e+00 5.12193561e-01
8.11245978e-01 -1.55302787e+00 9.18074310e-01 -1.12272739e+00
6.82622671e-01 3.90490621e-01 1.23258732e-01 -9.05705810e-01
-4.22285408e-01 -3.74799490e-01 2.50966400e-01 1.04127705e+00
4.15276110e-01 -3.39245826e-01 1.09305942e+00 8.80779266e-01
-1.07116371e-01 -1.12006867e+00 -1.18220603e+00 -3.03078711e-01
1.31473437e-01 -1.02840865e+00 2.11865678e-01 6.29250944e-01
1.60469919e-01 2.96311170e-01 -7.89524674e-01 -5.42609096e-01
-2.53961142e-02 4.83715050e-02 9.84505236e-01 -1.33986497e+00
-5.66787660e-01 -6.86418712e-01 -9.64641213e-01 -1.88657746e-01
3.44131887e-02 -7.35441267e-01 -4.39637035e-01 -1.11224282e+00
1.04269214e-01 -4.76610899e-01 -4.69628096e-01 4.06360358e-01
3.19480538e-01 3.69038910e-01 2.07497463e-01 1.54205054e-01
-5.93081117e-01 1.86553746e-01 6.42727792e-01 3.27797681e-01
-5.55121601e-01 6.23052597e-01 -4.03506398e-01 9.58668232e-01
8.17927480e-01 -4.30851698e-01 3.16435657e-02 -8.36707801e-02
7.29441881e-01 5.08937955e-01 2.74652541e-01 -1.48421538e+00
3.39763284e-01 -1.13787845e-01 4.16839331e-01 -3.88814628e-01
7.64465213e-01 -2.85891026e-01 5.99487722e-01 6.97492480e-01
-5.51342964e-01 2.22797483e-01 3.06860715e-01 6.16760291e-02
-3.62206191e-01 -2.96966076e-01 5.78526497e-01 -3.19299698e-01
-9.52946842e-01 5.11182472e-02 -7.39245236e-01 -1.94643259e-01
1.01991856e+00 -6.64392352e-01 1.71125293e-01 -7.88595796e-01
-1.41379511e+00 -1.73309490e-01 -3.51358652e-01 4.31990921e-01
3.75747353e-01 -9.88712728e-01 -8.07800174e-01 -2.98739448e-02
-3.57816696e-01 -6.48910403e-01 8.78247797e-01 9.72135365e-01
-8.14120352e-01 1.73858166e-01 -8.40613842e-01 -1.34572402e-01
-1.45238364e+00 2.28178039e-01 7.25741982e-01 -1.07022786e+00
-4.75098878e-01 7.77595699e-01 -6.23625398e-01 -6.82372928e-01
-1.00388157e-03 3.23244572e-01 -9.93158638e-01 2.09549651e-01
3.66200894e-01 7.01768756e-01 4.00468171e-01 -8.28850031e-01
3.19545679e-02 1.79854289e-01 -5.78840226e-02 -1.57668278e-01
1.70194972e+00 3.98871660e-01 1.20444953e-01 1.04167485e+00
8.36847484e-01 -3.34585577e-01 -4.72849399e-01 2.61833258e-02
2.48806714e-03 -1.67452961e-01 7.16581717e-02 -1.01720595e+00
-9.32675600e-01 8.83109093e-01 8.56637716e-01 3.13252032e-01
9.09826219e-01 -4.56310630e-01 7.21352994e-01 1.60105020e-01
4.65470821e-01 -1.18970203e+00 -3.91554981e-01 8.28486800e-01
5.69575429e-01 -8.11883748e-01 -3.48826826e-01 3.17412466e-01
-9.14430559e-01 1.24990988e+00 7.59623587e-01 -4.00874764e-01
7.09495246e-01 1.61343038e-01 1.58178374e-01 -2.97781557e-01
-1.00081933e+00 -8.66548643e-02 2.56718304e-02 1.83219805e-01
4.05864686e-01 3.56592191e-03 -5.25299013e-01 1.41783559e+00
-1.07687414e+00 2.48374775e-01 2.88205236e-01 4.79754925e-01
-4.03755978e-02 -8.38079035e-01 -4.54543203e-01 6.92527115e-01
-8.64539921e-01 1.58989355e-01 -3.34185719e-01 8.74448717e-01
4.24585193e-01 8.33438873e-01 1.35833263e-01 -1.27655721e+00
4.85027105e-01 1.98312640e-01 3.60075124e-02 -8.25898498e-02
-1.50682616e+00 -3.47493112e-01 3.00594360e-01 -7.26435304e-01
1.94113515e-02 -6.41062796e-01 -9.95383799e-01 -1.02316117e+00
-1.74148157e-01 4.46372569e-01 9.05517697e-01 1.26990700e+00
1.47105560e-01 7.01929390e-01 4.05289382e-01 -9.85988498e-01
-2.55680889e-01 -1.35165274e+00 -8.29313219e-01 5.32530367e-01
-5.55373728e-01 -6.47171378e-01 -4.56915237e-02 -5.19184887e-01]
|
[6.724055767059326, 0.3714500069618225]
|
ceed2b79-ed31-4db7-9ac4-39d3be820f36
|
cendernet-center-and-curvature
|
2208.09829
| null |
https://arxiv.org/abs/2208.09829v1
|
https://arxiv.org/pdf/2208.09829v1.pdf
|
CenDerNet: Center and Curvature Representations for Render-and-Compare 6D Pose Estimation
|
We introduce CenDerNet, a framework for 6D pose estimation from multi-view images based on center and curvature representations. Finding precise poses for reflective, textureless objects is a key challenge for industrial robotics. Our approach consists of three stages: First, a fully convolutional neural network predicts center and curvature heatmaps for each view; Second, center heatmaps are used to detect object instances and find their 3D centers; Third, 6D object poses are estimated using 3D centers and curvature heatmaps. By jointly optimizing poses across views using a render-and-compare approach, our method naturally handles occlusions and object symmetries. We show that CenDerNet outperforms previous methods on two industry-relevant datasets: DIMO and T-LESS.
|
['Francis wyffels', 'Joris de Hoog', 'Taoufik Bourgana', 'Jonathan Croenen', 'Rembert Daems', 'Peter De Roovere']
|
2022-08-21
| null | null | null | null |
['6d-pose-estimation-1']
|
['computer-vision']
|
[-1.10359170e-01 -6.18455857e-02 1.35466963e-01 -2.91303217e-01
-7.19288588e-01 -9.03517306e-01 5.70801556e-01 9.01944637e-02
2.17842132e-01 -5.21369539e-02 -1.26151040e-01 3.17502953e-02
-1.47944957e-01 -3.83641094e-01 -9.59197283e-01 -2.74378538e-01
6.07078783e-02 9.93692577e-01 1.85555324e-01 2.14811061e-02
7.05134213e-01 1.04924822e+00 -1.36991882e+00 3.75240780e-02
2.75218070e-01 1.27075517e+00 2.39688829e-01 6.17699146e-01
3.98418218e-01 2.19612762e-01 -3.35612953e-01 -2.38147363e-01
7.91391790e-01 3.35206628e-01 -6.55570567e-01 4.69043136e-01
9.14629877e-01 -5.84188044e-01 2.25726254e-02 8.09879422e-01
2.62516648e-01 -3.84384021e-02 8.39304626e-01 -1.12884760e+00
-2.75653094e-01 -1.37272812e-02 -8.66970122e-01 -4.43745703e-01
6.58296227e-01 2.05377042e-01 8.72187257e-01 -1.29213297e+00
1.06292510e+00 1.59065366e+00 6.45369768e-01 8.48160461e-02
-1.09221184e+00 -1.37968644e-01 2.64893472e-01 -2.03004435e-01
-1.05975246e+00 -3.58979739e-02 1.17558658e+00 -6.12009108e-01
1.04259419e+00 1.79510280e-01 6.15437090e-01 7.78544366e-01
5.82687080e-01 6.78382576e-01 7.69089818e-01 -1.25846341e-01
5.47794364e-02 -2.20842287e-01 -2.03778610e-01 8.40116918e-01
2.23697022e-01 -8.55325609e-02 -4.49707121e-01 -7.56574348e-02
1.43437850e+00 3.25351059e-01 1.12782352e-01 -1.48532426e+00
-1.70553792e+00 4.89263713e-01 4.88398671e-01 -5.15925825e-01
-3.48466396e-01 1.59914345e-01 1.95872724e-01 -1.60546675e-02
4.37584132e-01 8.42761755e-01 -7.66586721e-01 -1.53509215e-01
-5.64378127e-02 7.12461472e-01 7.41849720e-01 1.48322737e+00
9.26054001e-01 -4.46830958e-01 2.73257554e-01 6.61108851e-01
3.88333291e-01 5.27184784e-01 -2.65879691e-01 -1.49785519e+00
6.80251896e-01 9.04975593e-01 3.57739031e-01 -1.05459976e+00
-5.75051069e-01 -8.80200937e-02 -2.25442365e-01 6.25436366e-01
3.85655314e-01 4.86597940e-02 -1.03147924e+00 9.37073171e-01
6.39076710e-01 -4.39709455e-01 -3.41240138e-01 1.18072724e+00
7.33107209e-01 2.06122950e-01 -6.73761129e-01 4.12389696e-01
1.37993765e+00 -9.22500610e-01 -1.30775914e-01 -4.48075235e-01
2.92004287e-01 -1.14183724e+00 6.62986338e-01 5.96502006e-01
-1.25142217e+00 -4.23603803e-01 -1.04984677e+00 -4.22623485e-01
-2.38400921e-01 4.89348859e-01 6.12709343e-01 4.36573215e-02
-5.29411316e-01 6.89367235e-01 -9.78249073e-01 -2.08842039e-01
3.69846702e-01 3.15635443e-01 -3.24411899e-01 -7.81607777e-02
-1.71494082e-01 9.43549931e-01 3.50670278e-01 2.82315075e-01
-9.52608168e-01 -7.71432877e-01 -9.14791465e-01 -3.61205161e-01
6.71981394e-01 -8.08237016e-01 1.33431232e+00 -7.23924711e-02
-1.69033098e+00 1.05538392e+00 -4.06064130e-02 6.44089356e-02
8.50748241e-01 -7.07129419e-01 1.87466204e-01 1.29532516e-01
3.72729003e-02 5.55762649e-01 8.03436995e-01 -1.71009326e+00
-3.67085785e-01 -8.37552249e-01 1.68627650e-01 4.26903069e-01
4.97982770e-01 -2.99439907e-01 -6.75570190e-01 -2.51783162e-01
1.11702669e+00 -9.58129883e-01 -3.77350509e-01 3.66838694e-01
-9.82960105e-01 -3.02849799e-01 1.25791979e+00 -5.21510124e-01
1.49271265e-01 -1.96764898e+00 3.66827905e-01 2.94838607e-01
3.29106808e-01 -3.40087920e-01 1.43525824e-01 2.09896341e-01
3.20717953e-02 -3.14053625e-01 3.17058176e-01 -2.82224715e-01
2.07142428e-01 -2.27056831e-01 -1.09084785e-01 6.13494456e-01
3.64039421e-01 8.76132727e-01 -7.69792736e-01 -2.61038929e-01
8.17997694e-01 3.88444245e-01 -5.22607207e-01 1.62141696e-01
-3.80436480e-01 3.50795597e-01 -5.15604675e-01 9.71642137e-01
8.14142048e-01 -2.30483040e-01 1.23793215e-01 -6.95837379e-01
-2.61595160e-01 2.57564127e-01 -1.35203516e+00 1.93799543e+00
-4.48917687e-01 3.03096980e-01 1.02317050e-01 -5.83623886e-01
1.30272567e+00 9.34385955e-02 8.94036233e-01 -2.39710227e-01
2.58029878e-01 3.11892003e-01 -7.92172790e-01 -2.89193332e-01
5.37034154e-01 2.56462246e-01 -1.24573767e-01 3.53442252e-01
-1.24360397e-01 -9.31003511e-01 -9.83664542e-02 -3.24711315e-02
7.24559546e-01 7.85822809e-01 1.31950155e-01 -2.77351998e-02
7.06894398e-02 -7.14137359e-03 4.49151635e-01 2.43156388e-01
1.11911349e-01 1.02111280e+00 6.39968336e-01 -9.75352287e-01
-1.32025337e+00 -1.54601145e+00 -1.43757731e-01 3.48143786e-01
7.44135797e-01 -2.84447372e-01 -4.72822815e-01 -6.88298166e-01
7.06858516e-01 2.67146438e-01 -3.26028794e-01 9.90753025e-02
-8.88855278e-01 -1.05563626e-01 -6.23190463e-01 6.76688731e-01
1.80382743e-01 -5.86112559e-01 -1.04378223e+00 5.38861044e-02
1.37058824e-01 -1.23860133e+00 -4.84552115e-01 3.55306566e-01
-1.11727953e+00 -1.47414923e+00 -5.60278773e-01 -6.66518867e-01
8.50909531e-01 5.07839441e-01 1.31489599e+00 -4.82010394e-01
-5.86552680e-01 3.76752943e-01 -1.45299271e-01 -6.82662189e-01
-1.13972723e-01 -1.01730917e-02 7.17255995e-02 -3.40346038e-01
1.30261242e-01 -6.70048714e-01 -8.63585830e-01 7.42421508e-01
-1.14637338e-01 2.07506955e-01 5.60118198e-01 1.29660845e-01
8.67807090e-01 -6.20281756e-01 -9.66350511e-02 -5.33038378e-01
7.44920820e-02 -5.69718666e-02 -1.13835084e+00 2.11942624e-02
-2.85645217e-01 -1.82159334e-01 1.00952208e-01 -3.10165912e-01
-7.07553148e-01 7.48991311e-01 3.74342382e-01 -9.30570304e-01
-2.29801968e-01 -1.82817534e-01 -9.88612697e-02 -3.70902605e-02
5.18627584e-01 -3.65924895e-01 1.66703016e-01 -7.21359134e-01
4.14891034e-01 3.77858222e-01 5.13670266e-01 -6.53265595e-01
8.97714019e-01 6.15167916e-01 2.28872567e-01 -6.36623800e-01
-8.73702586e-01 -5.99025309e-01 -1.23960888e+00 -4.57527220e-01
1.06872797e+00 -9.83541608e-01 -1.40647948e+00 3.94290984e-01
-1.40239918e+00 2.52322108e-02 -2.57618159e-01 6.58670425e-01
-1.01310396e+00 1.11882418e-01 -5.16457319e-01 -6.33726835e-01
-1.63392425e-01 -1.48340046e+00 2.05901003e+00 -8.29944536e-02
-1.18207201e-01 -5.81876576e-01 -2.50246197e-01 5.56627750e-01
-1.69865400e-01 6.98488295e-01 7.50873506e-01 -1.93020850e-01
-1.34124255e+00 -2.98669606e-01 -2.14586526e-01 -1.01656519e-01
2.38264412e-01 3.16453688e-02 -8.81030917e-01 -3.29314113e-01
-1.77629068e-01 -4.82482821e-01 3.43567371e-01 5.99901855e-01
1.30381215e+00 -1.92007646e-02 -5.68132281e-01 8.53335321e-01
1.25881839e+00 -2.34100735e-03 2.67084301e-01 4.70239162e-01
1.07238269e+00 3.85431796e-01 7.50169516e-01 3.94772530e-01
3.38203341e-01 9.69062924e-01 1.15930510e+00 9.03038457e-02
1.66217998e-01 -2.01251611e-01 2.72500478e-02 6.96777225e-01
-2.78919071e-01 2.79337972e-01 -7.21773207e-01 2.35635057e-01
-1.69176352e+00 -3.89945865e-01 -1.72365263e-01 2.18112135e+00
2.57031977e-01 2.78426468e-01 1.18217446e-01 -4.10507560e-01
5.16537130e-01 1.02776587e-01 -1.03673863e+00 -3.09495896e-01
3.27436745e-01 -3.54204595e-01 6.00848019e-01 1.57904416e-01
-1.25405681e+00 7.32114196e-01 6.73973656e+00 9.28344503e-02
-9.10374701e-01 -2.92220861e-01 3.42424870e-01 -2.54631072e-01
-1.44053221e-01 -9.03880522e-02 -6.90362930e-01 -2.02847987e-01
-1.08542098e-02 4.58355606e-01 4.40481603e-01 1.24645150e+00
-2.36582130e-01 -4.76549342e-02 -1.37036061e+00 1.15815282e+00
2.47503996e-01 -1.24312329e+00 -1.99176133e-01 2.98380613e-01
9.47104037e-01 2.53441006e-01 1.72601230e-02 -2.52138853e-01
6.04957283e-01 -5.67665577e-01 1.08619595e+00 4.90967482e-01
5.52492440e-01 -7.37830043e-01 3.29549164e-01 2.07883507e-01
-1.23334169e+00 -1.23067014e-01 -3.65133643e-01 2.84495175e-01
3.70852470e-01 7.77726471e-01 -1.34243715e+00 7.38091707e-01
7.47132123e-01 9.10413265e-01 -4.06699985e-01 8.69531214e-01
6.33805152e-03 -3.72293681e-01 -3.23050320e-01 1.00754667e-02
-1.13696374e-01 -3.19221765e-01 8.94594550e-01 5.49488425e-01
4.19663072e-01 -4.36229467e-01 4.12371784e-01 1.33191991e+00
7.65761957e-02 -3.72431785e-01 -6.93084836e-01 3.21533948e-01
3.10709745e-01 1.55910957e+00 -9.01254237e-01 1.38537228e-01
-8.07701573e-02 9.74705994e-01 2.16735169e-01 2.96483129e-01
-5.73281229e-01 -2.48528510e-01 8.88657391e-01 1.00349985e-01
4.97676253e-01 -7.27160037e-01 -4.73918855e-01 -1.12981677e+00
4.79466438e-01 -4.72805262e-01 -1.09324880e-01 -1.19231939e+00
-1.20154834e+00 1.15180872e-01 1.35413006e-01 -1.66932821e+00
-3.03678781e-01 -1.31937838e+00 -2.15531290e-01 7.06556737e-01
-1.14980459e+00 -1.41204095e+00 -4.22647953e-01 1.10696569e-01
8.59029174e-01 7.39787221e-02 5.54832101e-01 -3.63399357e-01
-1.24419115e-01 -1.03289954e-01 -5.76821752e-02 -4.13691215e-02
5.62714636e-01 -1.45882511e+00 8.70096803e-01 2.35403925e-01
9.31671634e-02 5.98060310e-01 5.59490740e-01 -6.95044994e-01
-2.15581298e+00 -9.78366196e-01 2.66801953e-01 -1.03097057e+00
2.34315902e-01 -7.15912521e-01 -3.16550553e-01 9.52665865e-01
-2.56317139e-01 1.20746218e-01 -3.02567929e-01 2.99973160e-01
-3.51834029e-01 -1.48613453e-01 -1.00960279e+00 4.45686400e-01
1.26606798e+00 -4.57557768e-01 -3.66487145e-01 6.72795594e-01
6.65109336e-01 -1.29314673e+00 -1.10656881e+00 4.13819313e-01
8.61596823e-01 -9.13527668e-01 1.35747027e+00 -2.46537179e-01
4.99870360e-01 -3.81127506e-01 2.69546099e-02 -1.28855443e+00
-1.40347242e-01 -7.53784239e-01 -1.84510484e-01 5.35084367e-01
2.68874884e-01 -3.35687459e-01 8.61941934e-01 3.76729548e-01
-4.91019398e-01 -8.35927486e-01 -7.54598022e-01 -5.51789701e-01
-2.84765333e-01 -2.95829564e-01 7.50625908e-01 6.76522791e-01
-2.92050272e-01 1.99914396e-01 1.66355878e-01 3.00177425e-01
7.44131327e-01 7.02905953e-01 1.24353278e+00 -1.68648446e+00
2.17409775e-01 -2.34003961e-01 -4.49054062e-01 -1.30209816e+00
-2.27413222e-01 -7.16172159e-01 2.82106757e-01 -1.53052866e+00
1.47053882e-01 -3.13694626e-01 1.89008042e-01 2.77891874e-01
2.71732986e-01 7.54528567e-02 1.81697920e-01 1.44730166e-01
-6.27760768e-01 2.48523265e-01 1.66835475e+00 1.03269303e-02
-2.04050064e-01 4.68224809e-02 -3.00353318e-01 9.80031788e-01
4.89772737e-01 3.91352624e-02 -9.59146321e-02 -6.04083002e-01
3.68611038e-01 6.62099868e-02 8.41802120e-01 -9.47645903e-01
-9.48801115e-02 -2.69764423e-01 8.81289601e-01 -1.36452115e+00
8.56707871e-01 -1.00283635e+00 8.13719351e-03 1.66753531e-01
-4.10206802e-02 4.65556502e-01 -3.89557853e-02 5.46708882e-01
3.70576441e-01 5.51445633e-02 3.66266310e-01 -5.75092435e-01
-3.70541573e-01 4.33130056e-01 6.87729642e-02 -3.25503230e-01
1.14104438e+00 -5.70073545e-01 -1.66716948e-01 -9.30540413e-02
-5.91183245e-01 3.00152361e-01 7.62080491e-01 7.78914332e-01
8.47382784e-01 -1.47738230e+00 -3.46732765e-01 4.51675475e-01
3.77650440e-01 8.64794075e-01 -6.51179478e-02 6.58536971e-01
-8.69469762e-01 2.73326308e-01 -7.81798735e-02 -1.31619287e+00
-1.10002971e+00 4.73024547e-01 4.40644145e-01 1.59526929e-01
-8.49368632e-01 8.10349524e-01 2.44096681e-01 -1.14679122e+00
1.59135044e-01 -7.06628382e-01 1.61556274e-01 -4.11810279e-01
1.32525429e-01 6.49202943e-01 2.39563853e-01 -5.11009753e-01
-2.08746627e-01 9.78619039e-01 -5.91480508e-02 6.64828792e-02
1.62996173e+00 -2.04195343e-02 -8.29372033e-02 4.93073702e-01
1.35231757e+00 -2.04630360e-01 -1.84226418e+00 1.00199081e-01
-1.56674027e-01 -6.74803376e-01 -2.16485649e-01 -6.40091717e-01
-1.07293105e+00 8.00011754e-01 5.09580612e-01 -9.89104249e-03
5.14151514e-01 3.49185228e-01 6.60758078e-01 5.89902282e-01
4.47900295e-01 -1.23317635e+00 4.50008690e-01 5.88169873e-01
1.21307909e+00 -1.18343318e+00 2.66031086e-01 -8.71820211e-01
-3.84991497e-01 1.49670911e+00 7.58543432e-01 -2.19908163e-01
5.53799808e-01 2.93417603e-01 2.02924401e-01 -8.08448911e-01
-3.25997680e-01 4.42557752e-01 6.66862786e-01 4.76921469e-01
1.44696519e-01 7.30357692e-02 5.57403982e-01 -1.15953349e-01
-3.44043642e-01 -3.73934209e-01 1.03218719e-01 1.14061785e+00
-3.86770278e-01 -7.67755568e-01 -6.38414443e-01 3.54382336e-01
-7.69589171e-02 7.42063761e-01 -5.61846673e-01 6.20501935e-01
-1.57597545e-03 5.13546050e-01 4.23232049e-01 -5.07207692e-01
7.41373956e-01 -2.05048919e-01 8.55718017e-01 -6.91786706e-01
-3.30729276e-01 2.65738159e-01 2.98891007e-03 -1.02512407e+00
-2.49902323e-01 -8.09393346e-01 -1.10415685e+00 1.20825090e-01
-3.93146813e-01 -4.78978634e-01 1.06250572e+00 7.85158634e-01
5.34481704e-01 4.78908986e-01 8.01061332e-01 -1.70812786e+00
-7.39582300e-01 -6.97454393e-01 -5.75093091e-01 5.10311246e-01
3.03821027e-01 -9.91312206e-01 -1.77234322e-01 -1.83825269e-02]
|
[7.461577892303467, -2.6132845878601074]
|
a4149b1f-d833-47ba-99cb-fbcff11d341f
|
natural-language-processing-state-of-the-art
|
1708.05148
| null |
http://arxiv.org/abs/1708.05148v1
|
http://arxiv.org/pdf/1708.05148v1.pdf
|
Natural Language Processing: State of The Art, Current Trends and Challenges
|
Natural language processing (NLP) has recently gained much attention for
representing and analysing human language computationally. It has spread its
applications in various fields such as machine translation, email spam
detection, information extraction, summarization, medical, and question
answering etc. The paper distinguishes four phases by discussing different
levels of NLP and components of Natural Language Generation (NLG) followed by
presenting the history and evolution of NLP, state of the art presenting the
various applications of NLP and current trends and challenges.
|
['Sukhdev Singh', 'Kiran Khatter', 'Diksha Khurana', 'Aditya Koli']
|
2017-08-17
| null | null | null | null |
['spam-detection']
|
['natural-language-processing']
|
[ 6.13419235e-01 7.02209473e-01 -1.15321867e-01 -2.59095547e-03
-7.23220825e-01 -7.98240364e-01 1.19766092e+00 1.07153964e+00
-4.29155022e-01 1.18530226e+00 7.83803821e-01 -3.46072108e-01
-2.77871937e-02 -6.16534531e-01 -1.04859143e-01 -1.09515168e-01
-9.58718136e-02 6.83768332e-01 2.59963304e-01 -3.52049351e-01
7.47251630e-01 6.67093158e-01 -1.28610957e+00 5.16477823e-01
1.02303827e+00 4.67850626e-01 -4.86890897e-02 8.72966409e-01
-1.05527067e+00 1.00913250e+00 -9.87568557e-01 -2.57939875e-01
-2.78170943e-01 -8.16751719e-01 -1.47073877e+00 6.94449246e-02
-2.64745474e-01 5.11832416e-01 -1.74341276e-02 9.32097912e-01
5.27934372e-01 2.30571806e-01 5.54642320e-01 -1.07143664e+00
-4.84701484e-01 3.92286867e-01 -4.60053921e-01 3.17829221e-01
1.33108819e+00 -7.06975907e-02 4.14220095e-01 -5.58743477e-01
9.59046185e-01 1.76051736e+00 4.06130701e-01 6.94424927e-01
-6.90844953e-01 -6.92415535e-02 -1.11759223e-01 1.26103401e-01
-9.91382003e-01 -1.91323489e-01 4.08278883e-01 -2.80387312e-01
1.25486338e+00 2.55127877e-01 3.41693908e-01 6.47873878e-01
5.87544739e-01 1.26002562e+00 9.91112053e-01 -9.67286944e-01
1.62829086e-01 2.71893531e-01 5.39713979e-01 5.55239081e-01
1.80719107e-01 -4.81771737e-01 -4.81539488e-01 -6.28813922e-01
2.18474820e-01 -5.86625695e-01 1.45880478e-02 5.60314596e-01
-1.38265657e+00 9.22850132e-01 -3.15901041e-01 6.55341446e-01
-7.98682988e-01 -2.46704757e-01 7.24891722e-01 2.42789388e-01
4.42912906e-01 8.78150463e-01 -4.11455572e-01 -4.88141328e-01
-8.86936724e-01 5.04154265e-01 1.19043040e+00 1.00245857e+00
3.22744697e-01 -6.66175485e-02 -5.03305197e-01 8.72132123e-01
8.66621956e-02 4.15667564e-01 1.00862396e+00 -8.91399562e-01
4.42666143e-01 7.56831288e-01 2.41213754e-01 -1.05667710e+00
-6.55719101e-01 3.33198547e-01 -9.36548710e-01 -5.02809346e-01
2.17012651e-02 -6.15091801e-01 -5.91720939e-01 1.02087855e+00
1.10840730e-01 -2.21152470e-01 3.16994846e-01 3.30119640e-01
1.13892841e+00 1.35748470e+00 3.05486381e-01 -8.61879408e-01
1.75655973e+00 -6.32945061e-01 -9.80181813e-01 -3.55604410e-01
3.62556607e-01 -1.15617180e+00 5.62885106e-01 3.20332706e-01
-1.29927981e+00 -3.89555424e-01 -4.60809410e-01 -1.79373935e-01
-6.70210719e-01 -1.94233626e-01 5.30891180e-01 5.04046738e-01
-9.89505589e-01 4.60275769e-01 -5.47558308e-01 -8.99583101e-01
2.09548384e-01 3.03113848e-01 -2.21308023e-01 6.13527857e-02
-1.59537041e+00 1.02223575e+00 7.87517846e-01 -2.11615756e-01
3.72912362e-03 -3.31830114e-01 -8.92788291e-01 -1.25112668e-01
3.03553104e-01 -9.53331172e-01 1.41711640e+00 -5.10095656e-01
-1.54063511e+00 1.04784119e+00 -7.04472780e-01 -1.01249087e+00
2.81487435e-01 -2.59398639e-01 -4.75877941e-01 3.21805030e-01
1.51658982e-01 6.50468290e-01 4.39171851e-01 -7.00771093e-01
-6.88278675e-01 -1.37208149e-01 -3.82824838e-01 1.40166566e-01
2.77197868e-01 7.52292275e-01 -1.39560714e-01 -5.42503059e-01
-1.54018819e-01 -5.21272480e-01 -4.00465131e-01 -5.03907502e-01
-5.53759277e-01 -1.01399183e+00 6.48671389e-01 -7.19957232e-01
1.35103452e+00 -1.80048883e+00 -1.43033177e-01 -9.84823033e-02
-3.80274914e-02 6.62983775e-01 -1.32016093e-01 1.37081850e+00
1.79663915e-02 4.34361726e-01 -2.29107201e-01 2.07262829e-01
5.36769032e-02 1.76894501e-01 -6.40644848e-01 -1.11939132e-01
4.44411546e-01 1.26780212e+00 -1.22042549e+00 -1.06408024e+00
3.08931887e-01 2.26186693e-01 1.50386065e-01 -7.53631517e-02
-4.89302427e-01 1.38711512e-01 -7.11642921e-01 5.17440617e-01
2.61337519e-01 -9.74012464e-02 -9.55638010e-03 4.68526661e-01
-2.58047044e-01 4.87967908e-01 -8.08186114e-01 1.17320418e+00
-2.71694243e-01 1.06360435e+00 -1.81280628e-01 -1.00271416e+00
8.98573756e-01 6.31999433e-01 2.60533214e-01 -5.57814300e-01
-1.03927672e-01 8.02903175e-02 -2.34503150e-01 -8.58104765e-01
8.18437397e-01 -3.36106539e-01 -2.58690476e-01 7.90843606e-01
-6.96965754e-02 -5.21889627e-01 6.35496974e-01 5.30465901e-01
1.04346001e+00 -2.44415894e-01 1.24693191e+00 -1.78325236e-01
1.18605268e+00 4.48056400e-01 1.35075957e-01 8.97005439e-01
-3.55350196e-01 2.38750547e-01 6.00647926e-01 -2.97180533e-01
-8.54884267e-01 -7.83749878e-01 8.44965205e-02 7.03233063e-01
-2.40320638e-01 -4.38455909e-01 -7.93273985e-01 -4.50893819e-01
-1.58693999e-01 9.81506169e-01 -2.28997692e-01 1.35595933e-01
-8.20975661e-01 -8.39642644e-01 7.24338293e-01 -6.06052987e-02
5.53797364e-01 -1.96172082e+00 -3.20265085e-01 4.85156775e-01
-4.76314992e-01 -1.36997831e+00 8.91241357e-02 -3.90677661e-01
-1.05496061e+00 -1.01686966e+00 -7.36053288e-01 -1.04767954e+00
6.60080791e-01 1.81271639e-02 1.13908195e+00 -1.58886611e-01
-3.77788454e-01 5.19095898e-01 -5.16891539e-01 -8.80933702e-01
-1.00744712e+00 2.49309257e-01 7.23400116e-02 -4.31019515e-01
5.48117936e-01 -1.97687954e-01 -1.54514730e-01 -4.29707319e-01
-1.15599406e+00 -1.73452094e-01 6.50121152e-01 4.92132008e-01
1.11534677e-01 3.12950313e-01 1.00478375e+00 -1.18347776e+00
1.74799240e+00 -2.90058047e-01 -1.38520390e-01 4.66958880e-01
-2.54023314e-01 1.93503171e-01 6.93314970e-01 -1.22938134e-01
-1.35341299e+00 -3.10372055e-01 -4.09504920e-01 9.12994206e-01
-5.95694184e-01 4.42899376e-01 2.38387156e-02 1.46832511e-01
8.48789513e-01 6.20698750e-01 -9.36837215e-03 -4.97635275e-01
6.65256739e-01 9.08350766e-01 3.90478879e-01 -3.04360926e-01
5.84869683e-01 2.28936195e-01 -1.38412282e-01 -1.34191406e+00
-8.26043248e-01 -7.62608469e-01 -6.15994573e-01 -6.89079463e-02
6.49787068e-01 -1.88770026e-01 -6.15351439e-01 2.30461985e-01
-1.76295507e+00 3.53989482e-01 -6.40824020e-01 2.55030245e-01
-2.73867935e-01 9.60894406e-01 -7.90192902e-01 -1.08855534e+00
-1.02869904e+00 -5.75964153e-01 9.78658855e-01 4.33375388e-01
-7.75855184e-01 -1.11638212e+00 2.14332044e-01 3.57588410e-01
2.07305834e-01 4.44144219e-01 1.18233192e+00 -9.43594515e-01
4.02242085e-03 -6.65594280e-01 -1.20369241e-01 1.71368167e-01
1.74309880e-01 -2.26921529e-01 -3.93264323e-01 2.55279958e-01
3.35374847e-02 -1.98867738e-01 5.61737716e-01 3.38262856e-01
4.53262180e-01 -7.57838726e-01 -5.64023018e-01 -5.07497609e-01
1.10816801e+00 3.15766156e-01 6.27621651e-01 2.63404310e-01
1.93005547e-01 1.19985020e+00 5.99487960e-01 7.53415972e-02
-1.88687388e-02 -6.89707994e-02 -4.31243867e-01 3.96715194e-01
1.19504519e-01 -2.22489491e-01 3.14489245e-01 1.08704233e+00
2.66547471e-01 -3.96860212e-01 -1.16318929e+00 6.00009620e-01
-1.92880464e+00 -9.12515223e-01 -3.43470842e-01 1.72969639e+00
9.64839399e-01 9.77539942e-02 -4.54710796e-03 2.21785739e-01
7.97835529e-01 3.19147348e-01 -1.41920179e-01 -1.08923173e+00
-2.00307101e-01 1.91799253e-01 -4.61063758e-02 6.67191148e-01
-6.64727628e-01 1.20706689e+00 7.16022968e+00 8.04060698e-01
-6.84843540e-01 -2.20905945e-01 4.35945362e-01 6.40624821e-01
1.37809962e-01 -1.15584105e-01 -7.39777446e-01 3.03777397e-01
1.00518775e+00 -9.44052815e-01 6.76162392e-02 4.42738861e-01
7.25706577e-01 -4.92611766e-01 -6.52803242e-01 8.72493505e-01
3.21310759e-01 -1.39384365e+00 8.30659032e-01 -2.76308805e-01
8.14271867e-01 -2.04761088e-01 -7.87273467e-01 2.61116356e-01
9.98797864e-02 -7.66415775e-01 -2.46368442e-03 4.52282995e-01
2.56619781e-01 -8.01450253e-01 1.09655511e+00 8.87627542e-01
-6.90751910e-01 1.40603587e-01 -3.65992039e-01 -3.10848832e-01
8.04057717e-01 9.17632818e-01 -1.16459692e+00 8.77208769e-01
1.66638382e-02 2.17666775e-01 -3.44636053e-01 9.76853311e-01
-4.32342380e-01 5.34841239e-01 -2.62486815e-01 -6.08984292e-01
3.93077701e-01 -7.97346309e-02 8.01651537e-01 1.47522807e+00
-2.53864616e-01 5.50203145e-01 2.26697013e-01 3.57793480e-01
-9.30558443e-02 5.89478612e-01 -4.66142982e-01 -5.95467627e-01
3.23362529e-01 8.53023291e-01 -8.87319386e-01 -9.67050791e-01
5.75342216e-02 9.01964545e-01 -1.80610657e-01 2.22906187e-01
-2.11642653e-01 -7.82989323e-01 -7.29094520e-02 2.65652865e-01
-4.41867977e-01 -2.53585398e-01 -1.96819156e-01 -9.28050756e-01
-1.38805822e-01 -9.52023983e-01 5.50256133e-01 -7.58706212e-01
-1.04576623e+00 8.70983303e-01 3.43272775e-01 -8.64629865e-01
-7.23451078e-01 -4.64892268e-01 -6.38642550e-01 7.31503129e-01
-1.16583359e+00 -5.40587246e-01 1.25862435e-01 9.85795781e-02
8.46176326e-01 -1.98303819e-01 8.82149041e-01 -1.88696921e-01
-1.36117175e-01 -1.25329822e-01 -1.51588723e-01 2.52335146e-03
3.58579457e-01 -1.09818757e+00 7.73814201e-01 7.64535189e-01
-3.00650224e-02 6.53165042e-01 8.59951377e-01 -5.86596429e-01
-1.11716139e+00 -8.83610904e-01 2.06483912e+00 -4.06905077e-02
7.14337945e-01 -1.85238924e-02 -7.69087791e-01 2.21903965e-01
6.68538034e-01 -8.62514555e-01 5.53524554e-01 -4.20348585e-01
3.76129031e-01 1.68676823e-01 -1.25846744e+00 7.85247743e-01
4.77598816e-01 -3.97946745e-01 -1.12597048e+00 1.15026784e+00
8.66405845e-01 -7.79800862e-02 -3.30911160e-01 4.68059331e-02
1.29545078e-01 -4.33035791e-01 7.39579618e-01 -7.90608704e-01
2.67745525e-01 -1.99292555e-01 7.30692565e-01 -1.00566149e+00
-1.27691582e-01 -1.32681870e+00 -1.47180766e-01 1.25436115e+00
4.35982883e-01 -9.36030388e-01 6.44843876e-01 4.78704602e-01
2.21247748e-01 -6.38547719e-01 -6.64244413e-01 -2.35708982e-01
-2.40192469e-02 -4.47076112e-01 3.00505966e-01 8.38441312e-01
6.82869494e-01 9.45200324e-01 -1.11047238e-01 -4.92919624e-01
3.44723642e-01 -1.08553588e-01 5.73733687e-01 -1.33232272e+00
1.51374653e-01 -5.07442951e-01 -3.81819040e-01 -1.12640345e+00
1.32716462e-01 -6.88806295e-01 1.29698589e-01 -2.35112119e+00
2.62769815e-02 5.69289148e-01 4.16176319e-01 4.31815177e-01
-1.21108487e-01 -2.06283569e-01 2.38276273e-01 1.43511191e-01
-8.09682190e-01 2.08799422e-01 1.19151545e+00 -1.01524377e-02
-3.71241838e-01 1.99219137e-02 -7.60500133e-01 8.55579376e-01
1.08989596e+00 -3.69391263e-01 -1.33200496e-01 -4.85660695e-02
4.77246761e-01 1.75859660e-01 -1.99396625e-01 -5.99675775e-01
4.26451057e-01 -2.05114875e-02 9.13106501e-02 -8.28582764e-01
-1.46173760e-01 -2.79597193e-01 -4.31689054e-01 6.69567108e-01
-4.93155539e-01 3.77598554e-01 2.39882037e-01 3.39366466e-01
-6.48898780e-01 -6.37150764e-01 6.32261753e-01 -6.25818193e-01
-6.90492094e-01 -2.95831203e-01 -1.13292038e+00 3.88802111e-01
9.40378547e-01 -1.33952230e-01 -2.48860076e-01 -5.40959358e-01
-5.48994541e-01 5.56359470e-01 -1.15822017e-01 5.42279899e-01
7.13964522e-01 -7.70220578e-01 -1.00441611e+00 -2.89340973e-01
-1.63590014e-01 7.70646334e-02 -1.23428702e-01 5.09850085e-01
-1.04785490e+00 1.07689941e+00 2.86089629e-01 -1.19546644e-01
-1.30664277e+00 4.40548241e-01 -1.68372318e-01 -8.95437002e-01
-5.99604070e-01 4.57538992e-01 -1.66590020e-01 -2.73602039e-01
9.73033998e-03 -1.72329098e-02 -7.04508781e-01 -5.05580287e-03
9.66197133e-01 4.86892253e-01 -1.34258702e-01 -6.22486770e-01
-4.87811953e-01 3.30427885e-01 -1.77569896e-01 -2.80742407e-01
8.82009745e-01 -1.51109651e-01 -8.62659216e-01 5.68747342e-01
9.07637894e-01 -1.18989825e-01 1.82434916e-01 -2.95440912e-01
7.96546042e-01 2.39627466e-01 -4.42459077e-01 -9.15666461e-01
4.94938642e-02 7.50814855e-01 -1.25916123e-01 6.49328411e-01
9.26169276e-01 1.12493932e-01 1.36080766e+00 9.26530004e-01
3.95910174e-01 -1.35839772e+00 -2.43859529e-01 8.44930708e-01
9.99067724e-01 -9.95836318e-01 3.88844401e-01 -4.00492489e-01
-5.90327680e-01 1.17239094e+00 -1.30656421e-01 -1.53224645e-02
7.17052639e-01 9.42382887e-02 1.65236175e-01 -2.88696110e-01
-7.72720277e-01 -2.76940633e-02 3.55590373e-01 6.29359901e-01
7.48105526e-01 -1.45199761e-01 -1.23808503e+00 1.87264055e-01
-5.41859031e-01 1.06698856e-01 5.17186642e-01 1.28776503e+00
-7.59522378e-01 -1.33539736e+00 -7.01913714e-01 5.35229266e-01
-7.97962964e-01 -2.60965347e-01 -1.18532896e+00 5.76174617e-01
-2.42342442e-01 1.30034292e+00 -2.29743212e-01 2.81953663e-01
3.93632054e-01 4.03664023e-01 2.48532787e-01 -9.95479524e-01
-9.29221809e-01 -2.92234272e-01 4.58133966e-01 -3.15547399e-02
-6.10448360e-01 -6.18659854e-01 -1.44491220e+00 -2.11656272e-01
-1.76033661e-01 1.03370523e+00 5.15716851e-01 1.07351589e+00
3.97936642e-01 2.72007585e-01 2.72211164e-01 -5.45386612e-01
-1.54101968e-01 -1.20762146e+00 -2.91705817e-01 7.93522075e-02
2.41259292e-01 2.39862740e-01 -2.72101790e-01 2.83216804e-01]
|
[12.412054061889648, 9.450925827026367]
|
fa02bb91-0488-4866-8f84-7269a47242cf
|
group-gated-fusion-on-attention-based
|
2201.06309
| null |
https://arxiv.org/abs/2201.06309v1
|
https://arxiv.org/pdf/2201.06309v1.pdf
|
Group Gated Fusion on Attention-based Bidirectional Alignment for Multimodal Emotion Recognition
|
Emotion recognition is a challenging and actively-studied research area that plays a critical role in emotion-aware human-computer interaction systems. In a multimodal setting, temporal alignment between different modalities has not been well investigated yet. This paper presents a new model named as Gated Bidirectional Alignment Network (GBAN), which consists of an attention-based bidirectional alignment network over LSTM hidden states to explicitly capture the alignment relationship between speech and text, and a novel group gated fusion (GGF) layer to integrate the representations of different modalities. We empirically show that the attention-aligned representations outperform the last-hidden-states of LSTM significantly, and the proposed GBAN model outperforms existing state-of-the-art multimodal approaches on the IEMOCAP dataset.
|
['Helen Meng', 'Kun Li', 'PengFei Liu']
|
2022-01-17
| null | null | null | null |
['multimodal-emotion-recognition', 'multimodal-emotion-recognition']
|
['computer-vision', 'speech']
|
[ 2.69364029e-01 -3.06086093e-01 -1.99803293e-01 -6.90319836e-01
-7.95255542e-01 -1.55173406e-01 7.73613393e-01 -1.18139103e-01
-4.37615335e-01 4.03679073e-01 5.41893959e-01 -6.18983768e-02
2.67826408e-01 -1.56250194e-01 -5.62255085e-01 -7.57364452e-01
1.01035953e-01 2.30602741e-01 -2.43332982e-01 -3.39815348e-01
-1.74083903e-01 7.84053206e-02 -1.26656151e+00 6.47397995e-01
5.85014880e-01 1.41445255e+00 -2.43510440e-01 7.12864697e-01
-4.73126471e-01 9.99938786e-01 -9.58525166e-02 -4.41433966e-01
-4.64530855e-01 -8.22096467e-01 -9.62465763e-01 -6.69906065e-02
1.44080415e-01 1.45253222e-02 -4.05425072e-01 8.77236605e-01
6.14743292e-01 6.37341976e-01 3.04792404e-01 -1.40654135e+00
-5.10800004e-01 6.18490934e-01 -4.07289356e-01 1.28345683e-01
3.71999681e-01 -2.34303996e-01 9.80501771e-01 -7.28079736e-01
2.97883272e-01 1.37073231e+00 3.29655051e-01 8.29888701e-01
-1.05406630e+00 -5.61892629e-01 6.10414684e-01 6.18609250e-01
-1.00429118e+00 -6.07876003e-01 9.91768897e-01 -8.55180994e-02
1.41524494e+00 3.26306015e-01 5.66595197e-01 1.73401678e+00
2.85797685e-01 1.21211302e+00 8.85963440e-01 -3.10859293e-01
6.09053299e-02 -1.70196876e-01 3.37682605e-01 7.20041037e-01
-8.44186187e-01 -3.06479573e-01 -1.04121053e+00 -7.64951184e-02
3.56990248e-01 1.16365470e-01 -1.68294623e-01 -2.41803721e-01
-1.21772921e+00 5.35060585e-01 4.80906278e-01 6.53801620e-01
-7.24646628e-01 3.68334830e-01 7.61449218e-01 3.08700353e-01
6.05232477e-01 7.73572773e-02 -4.10672754e-01 -5.10417283e-01
-6.70977831e-01 -4.13677782e-01 6.00506663e-01 6.11954331e-01
4.07043546e-01 3.48806381e-01 -2.60885477e-01 9.39963102e-01
5.13272941e-01 4.63210464e-01 5.75148046e-01 -7.12318599e-01
5.29259861e-01 6.95653677e-01 -5.50198369e-02 -9.90558803e-01
-5.33575058e-01 -1.10832766e-01 -1.11476302e+00 -4.31173921e-01
-2.50267331e-02 -3.47054482e-01 -8.66374671e-01 2.21817088e+00
1.65572420e-01 3.85046721e-01 3.62862587e-01 9.37301040e-01
1.08183205e+00 9.39009726e-01 4.79817957e-01 -1.66474298e-01
1.34799445e+00 -1.43726003e+00 -1.30703926e+00 -4.67218995e-01
5.15377820e-01 -6.50406539e-01 6.07004464e-01 1.86781988e-01
-1.24061227e+00 -4.35976923e-01 -7.62889683e-01 -2.54096687e-01
-4.97989863e-01 1.28166825e-01 6.64314508e-01 1.33671209e-01
-1.05538356e+00 1.89204171e-01 -8.31795871e-01 -7.19016612e-01
9.86478031e-02 3.78814638e-01 -6.19131804e-01 2.11220771e-01
-1.44163597e+00 1.07910621e+00 2.65814483e-01 7.53808498e-01
-6.99604034e-01 1.46834612e-01 -1.16260982e+00 2.38985389e-01
3.82479101e-01 -7.96590447e-01 1.33720160e+00 -1.47252119e+00
-2.00446606e+00 5.86900532e-01 -7.61168838e-01 -4.21600342e-01
-3.97237651e-02 -3.80344927e-01 -7.52557456e-01 7.01317042e-02
-6.35747612e-01 7.07354128e-01 7.63012946e-01 -1.24786115e+00
-3.25424433e-01 -5.04945099e-01 -3.01079959e-01 6.35570943e-01
-4.99798626e-01 2.63521850e-01 -5.28950632e-01 -2.68693000e-01
7.77949691e-02 -9.45346653e-01 -1.62389949e-01 -6.16718113e-01
-5.12826025e-01 -2.50707865e-01 7.96094298e-01 -8.88254285e-01
1.37379098e+00 -2.05527115e+00 7.65361309e-01 4.09882516e-02
-1.89939216e-01 1.98736146e-01 -6.19046688e-01 5.97175479e-01
-2.10404947e-01 6.82598054e-02 -1.33416122e-02 -1.04399657e+00
2.44681165e-01 2.93805838e-01 -3.45099151e-01 1.38241485e-01
2.11605921e-01 1.22363651e+00 -7.66633153e-01 -4.33412164e-01
2.06824496e-01 7.30711579e-01 -7.24820495e-02 5.98405004e-01
-1.72807083e-01 6.30904377e-01 -2.18245611e-01 6.66655719e-01
2.31052697e-01 -2.23965243e-01 3.51315111e-01 -5.31090975e-01
-1.11570932e-01 1.93303153e-01 -5.19836843e-01 2.10323167e+00
-5.21027386e-01 6.62126720e-01 1.23375200e-01 -8.93883049e-01
8.79434168e-01 8.71244848e-01 4.31894898e-01 -9.49735343e-01
6.93451226e-01 -1.32755727e-01 -2.90697124e-02 -6.90221012e-01
6.82168961e-01 -2.13302329e-01 -2.38098249e-01 3.55237722e-01
4.43908393e-01 3.57906133e-01 -1.35727957e-01 1.48269594e-01
7.71926582e-01 1.40745610e-01 4.04029824e-02 4.62533355e-01
4.95883226e-01 -6.48960888e-01 4.15187895e-01 4.12418753e-01
-3.26276779e-01 3.38949919e-01 3.97740871e-01 -2.91637659e-01
-6.72838688e-01 -5.86090267e-01 4.82644826e-01 1.48499525e+00
5.63469306e-02 -3.17389995e-01 -6.54630899e-01 -6.38286591e-01
-5.90572774e-01 5.68128765e-01 -9.47407067e-01 -3.30051839e-01
-2.77979523e-01 -4.82801735e-01 5.70015669e-01 7.56976366e-01
6.73286021e-01 -1.35868478e+00 -3.65549564e-01 5.00682592e-02
-7.93734312e-01 -1.42227781e+00 -4.90610421e-01 3.69555324e-01
-6.84218109e-01 -6.51988328e-01 -6.70148194e-01 -6.77225947e-01
3.49720210e-01 -4.66758721e-02 8.66811037e-01 -2.76129454e-01
4.20467019e-01 6.39066458e-01 -4.34240937e-01 -1.98279589e-01
8.65838863e-03 2.97107816e-01 -2.92809784e-01 6.56363428e-01
4.20791328e-01 -2.67018467e-01 -2.66752064e-01 3.37291181e-01
-7.40709186e-01 3.55039239e-01 4.94614691e-01 1.08124721e+00
3.73991609e-01 -6.61775112e-01 4.87151861e-01 -3.85885030e-01
5.72770119e-01 -5.67862034e-01 -2.78123990e-02 5.30607998e-01
1.52424082e-01 1.05471276e-01 2.93792486e-01 -5.88213682e-01
-1.32063437e+00 1.15518691e-02 -2.04384252e-01 -6.29562199e-01
-3.13666731e-01 7.85241604e-01 -3.65607917e-01 6.72851801e-02
-2.02443391e-01 1.39213860e-01 -1.93172082e-01 -2.73873925e-01
5.18130660e-01 7.50450075e-01 5.89786232e-01 -5.24441063e-01
-5.90231046e-02 3.62931907e-01 -1.23455472e-01 -8.19115460e-01
-8.54244232e-01 -5.00436246e-01 -3.98015022e-01 -4.49415118e-01
1.32044733e+00 -6.92521393e-01 -9.09101009e-01 6.37963235e-01
-1.51151574e+00 -5.25569558e-01 1.29516572e-01 6.12349570e-01
-5.28123140e-01 1.85590252e-01 -8.56147885e-01 -1.23303759e+00
-4.92725372e-01 -1.17318189e+00 1.18829429e+00 4.59540009e-01
-2.42816120e-01 -1.23511708e+00 2.23037973e-01 5.55375457e-01
5.04366696e-01 1.03111990e-01 8.28532934e-01 -8.73289406e-01
-1.14098571e-01 -1.72276169e-01 -1.92005690e-02 3.93070430e-01
-1.12709723e-01 -1.99844949e-02 -1.17135906e+00 -2.50162240e-02
-9.65171978e-02 -5.04536569e-01 1.08847964e+00 2.69990146e-01
7.78632820e-01 -1.45318553e-01 -2.37590984e-01 1.76669076e-01
7.21015215e-01 4.72384900e-01 6.99087918e-01 6.82079494e-02
8.41242611e-01 5.93276203e-01 3.36181402e-01 4.38342899e-01
6.45791590e-01 6.14730597e-01 6.51246309e-01 -3.64933908e-01
3.96290272e-01 -3.87656763e-02 5.30551374e-01 1.36436248e+00
-1.21011458e-01 -7.55388737e-01 -7.15259075e-01 4.55666631e-01
-2.42313957e+00 -1.03438151e+00 1.00330137e-01 1.82699287e+00
5.11366785e-01 -2.52802223e-01 -1.37780211e-03 -1.49392977e-01
7.97749460e-01 3.69514465e-01 -4.34634060e-01 -9.14171338e-01
-3.13842386e-01 3.84407863e-02 -2.24119082e-01 5.57457864e-01
-1.22941291e+00 1.17547488e+00 6.00733185e+00 3.87224376e-01
-1.23398471e+00 2.41668403e-01 6.89958751e-01 -5.12818545e-02
-1.07319079e-01 -1.25941277e-01 -4.43418205e-01 3.25047851e-01
1.20993006e+00 1.57910407e-01 4.11552876e-01 4.28573102e-01
2.16629863e-01 -8.91992822e-02 -1.14314675e+00 1.10987961e+00
4.52530056e-01 -7.57294416e-01 1.81498304e-01 -1.61571294e-01
6.26885951e-01 -1.16032111e-02 3.14111471e-01 3.86854678e-01
8.09449255e-02 -9.47259009e-01 5.37050962e-01 1.01135731e+00
2.49631777e-01 -7.13301599e-01 1.11618006e+00 -1.00200595e-02
-1.33713138e+00 -1.07889175e-01 1.87766910e-01 1.24303915e-01
7.68890858e-01 2.20693415e-03 -2.46326745e-01 6.99865997e-01
7.46294677e-01 6.95455194e-01 -2.01178402e-01 6.17765605e-01
-8.98059309e-02 5.01020908e-01 -6.51498437e-02 -1.23036131e-01
5.60915947e-01 -1.64374158e-01 4.28429306e-01 1.29272830e+00
3.95939618e-01 5.42630590e-02 4.68942448e-02 5.18938184e-01
-3.60819101e-01 1.00990988e-01 -4.05059457e-01 -5.77011883e-01
3.36045958e-02 1.38580549e+00 -1.40828505e-01 -5.04849434e-01
-4.40257519e-01 1.34727693e+00 3.22834402e-01 6.70173824e-01
-1.04209614e+00 4.73898463e-02 8.10626507e-01 -7.93840230e-01
2.42422938e-01 -2.32168168e-01 1.40219644e-01 -1.22997570e+00
-2.58356929e-01 -7.86357403e-01 6.04027808e-01 -1.25181425e+00
-1.43287909e+00 1.09494305e+00 -3.20382446e-01 -9.10935163e-01
-3.71176928e-01 -3.27204645e-01 -5.41517138e-01 8.80458415e-01
-1.34961259e+00 -1.48630524e+00 -2.63965547e-01 7.91985154e-01
5.19992590e-01 1.93319414e-02 1.12048662e+00 2.89825499e-01
-1.04597068e+00 5.36691964e-01 -2.42247835e-01 1.43615916e-01
8.44032884e-01 -9.37573671e-01 -3.64202261e-03 7.05720425e-01
1.90844089e-01 7.77917325e-01 4.90628004e-01 -4.75771457e-01
-1.45655167e+00 -8.60111952e-01 1.11055398e+00 -7.46913925e-02
6.52830899e-01 -2.36623794e-01 -1.04899597e+00 9.34608042e-01
1.21279037e+00 -6.96019754e-02 9.77462351e-01 2.55852491e-01
-4.07278270e-01 -2.65100729e-02 -5.96959472e-01 6.43853426e-01
6.62720859e-01 -8.94196272e-01 -4.25591320e-01 -2.13292852e-01
6.05600595e-01 -4.46004063e-01 -8.24548542e-01 5.91617882e-01
5.78697026e-01 -8.30925763e-01 5.45371532e-01 -9.07477915e-01
4.21668649e-01 1.32687762e-01 -3.87460470e-01 -1.48487890e+00
1.33180991e-02 -7.07883060e-01 -4.42887872e-01 1.31236804e+00
3.51964504e-01 -4.50027257e-01 2.99234748e-01 7.69883394e-01
-3.43005896e-01 -9.65205729e-01 -1.05062699e+00 -3.32736731e-01
-3.10899198e-01 -4.52170312e-01 3.63627046e-01 1.07426417e+00
4.69490260e-01 7.88100660e-01 -8.29680204e-01 6.08657300e-02
-1.01099154e-02 1.13753583e-02 5.15993893e-01 -8.07248175e-01
3.22356820e-02 -5.88598847e-01 -3.18110377e-01 -1.09268260e+00
5.76780617e-01 -6.17477059e-01 1.90469116e-01 -1.68271601e+00
2.79241711e-01 3.03671390e-01 -8.57239783e-01 7.42623150e-01
-9.75348463e-04 2.91097641e-01 1.64969131e-01 -3.29954535e-01
-1.25031054e+00 1.18633187e+00 8.63597691e-01 -2.10011199e-01
-1.55444875e-01 -5.19372344e-01 -2.60319501e-01 7.34717131e-01
6.99758351e-01 -2.94451658e-02 -1.77635640e-01 -4.70073313e-01
2.75226831e-01 3.03437799e-01 2.77454644e-01 -7.78699100e-01
5.57367921e-01 -7.25858808e-02 1.91497847e-01 -5.94029605e-01
9.46099460e-01 -9.41653609e-01 6.18594140e-02 -9.18829218e-02
-6.31676435e-01 4.28683043e-01 4.08537716e-01 4.21167076e-01
-6.41043842e-01 3.17037523e-01 4.38380897e-01 1.59877479e-01
-7.57967293e-01 3.53025079e-01 -6.16889954e-01 -3.01018775e-01
7.65231490e-01 2.39804074e-01 -3.92167062e-01 -7.64728010e-01
-9.34195697e-01 4.03362185e-01 -6.42604828e-02 7.89716065e-01
7.27438569e-01 -1.60895073e+00 -2.93702215e-01 -1.45104617e-01
1.26469389e-01 -5.92897356e-01 8.25191975e-01 1.18990505e+00
3.31073999e-01 4.34050500e-01 -1.52184784e-01 -5.16132295e-01
-1.63548040e+00 3.54027092e-01 4.82977718e-01 -4.35112864e-01
-5.37292473e-02 8.23270559e-01 5.92225650e-03 -3.92407566e-01
5.45976162e-01 -1.25562221e-01 -3.85424376e-01 1.86691791e-01
3.63031745e-01 1.99010357e-01 -7.52155557e-02 -1.18311131e+00
-4.16551322e-01 3.70379686e-01 3.95904072e-02 -5.45719266e-01
1.11993647e+00 -4.11760867e-01 -2.93182284e-01 1.10740983e+00
1.15125561e+00 -7.49106169e-01 -9.18309271e-01 -4.91805732e-01
-1.65767714e-01 1.29032418e-01 1.23442158e-01 -9.79599178e-01
-1.15851271e+00 1.41455841e+00 6.56590939e-01 7.45838657e-02
1.22974300e+00 -1.15073714e-02 1.02328837e+00 5.27722239e-01
-1.10435247e-01 -1.16927767e+00 5.43400407e-01 8.76132369e-01
9.36916471e-01 -1.25098836e+00 -6.99007452e-01 1.79819331e-01
-1.10648704e+00 9.30312812e-01 7.83065677e-01 5.11899233e-01
3.00086439e-01 -7.09427744e-02 2.86953539e-01 -9.57684740e-02
-1.17222977e+00 -4.42098051e-01 5.79516053e-01 -3.52188796e-02
6.98559880e-01 -1.76095441e-01 6.32668510e-02 9.32405591e-01
4.81879830e-01 -8.84942859e-02 -2.15115592e-01 8.53444278e-01
-3.18853743e-02 -1.16693878e+00 -1.80217296e-01 6.38904274e-02
-3.23511213e-01 -2.43055820e-01 -6.23052299e-01 4.01597530e-01
-1.08025700e-01 1.15807199e+00 1.90422967e-01 -8.08175266e-01
2.74012953e-01 8.04666400e-01 4.29314375e-01 -8.53445232e-02
-8.94957066e-01 4.14910883e-01 1.94101050e-01 -8.24822724e-01
-8.57465029e-01 -4.71861541e-01 -1.21183670e+00 -3.28738615e-02
-3.15781444e-01 1.85233116e-01 8.59721541e-01 1.31009662e+00
6.65471077e-01 8.00151646e-01 4.22904313e-01 -1.11319733e+00
8.62371027e-02 -1.17402363e+00 -2.30906233e-01 3.84580910e-01
4.06882375e-01 -5.76151669e-01 -1.50198340e-01 -1.54330403e-01]
|
[13.223611831665039, 5.282009124755859]
|
75bfd7f7-ccbc-4c1a-ba66-c599b92c6a0c
|
fine-grained-image-analysis-with-deep
|
2111.06119
| null |
https://arxiv.org/abs/2111.06119v2
|
https://arxiv.org/pdf/2111.06119v2.pdf
|
Fine-Grained Image Analysis with Deep Learning: A Survey
|
Fine-grained image analysis (FGIA) is a longstanding and fundamental problem in computer vision and pattern recognition, and underpins a diverse set of real-world applications. The task of FGIA targets analyzing visual objects from subordinate categories, e.g., species of birds or models of cars. The small inter-class and large intra-class variation inherent to fine-grained image analysis makes it a challenging problem. Capitalizing on advances in deep learning, in recent years we have witnessed remarkable progress in deep learning powered FGIA. In this paper we present a systematic survey of these advances, where we attempt to re-define and broaden the field of FGIA by consolidating two fundamental fine-grained research areas -- fine-grained image recognition and fine-grained image retrieval. In addition, we also review other key issues of FGIA, such as publicly available benchmark datasets and related domain-specific applications. We conclude by highlighting several research directions and open problems which need further exploration from the community.
|
['Serge Belongie', 'Jian Yang', 'Jinhui Tang', 'Yuxin Peng', 'Jianxin Wu', 'Oisin Mac Aodha', 'Yi-Zhe Song', 'Xiu-Shen Wei']
|
2021-11-11
| null | null | null | null |
['fine-grained-image-recognition']
|
['computer-vision']
|
[ 1.70016527e-01 -6.70588851e-01 -1.34845704e-01 -5.32739878e-01
-5.83442628e-01 -7.46502578e-01 6.82186246e-01 5.20892330e-02
-2.51738191e-01 5.85286140e-01 1.17445670e-01 1.23762697e-01
-4.32606608e-01 -8.86090636e-01 -5.94767451e-01 -7.01146245e-01
4.76303920e-02 2.82279074e-01 7.02645183e-02 9.80357639e-03
3.71462703e-01 8.40235710e-01 -2.09748387e+00 8.12988997e-01
5.08130848e-01 1.43745565e+00 6.89535290e-02 5.66603780e-01
-2.49867454e-01 6.49956763e-01 -7.07948625e-01 -2.43416369e-01
1.02486089e-01 6.01190142e-02 -9.94835734e-01 2.59310454e-01
9.42653656e-01 -1.26740351e-01 -2.83883154e-01 1.17463696e+00
3.31208467e-01 1.07988313e-01 8.36468577e-01 -1.38812971e+00
-1.01948655e+00 1.10123321e-01 -7.13201702e-01 5.28274596e-01
-2.00202569e-01 -1.24614552e-01 1.23122108e+00 -8.73149097e-01
1.57883197e-01 1.56363606e+00 5.56667805e-01 2.13123560e-01
-8.14928174e-01 -6.10499501e-01 4.74301904e-01 5.00695586e-01
-1.54401648e+00 -1.99678540e-01 5.13701558e-01 -7.44101524e-01
9.54532266e-01 4.27255809e-01 1.53381690e-01 5.92341423e-01
3.04725796e-01 7.43972719e-01 1.41941321e+00 -4.40442473e-01
4.52128574e-02 -3.19372356e-01 4.40611094e-01 7.38353372e-01
1.96982309e-01 2.68614233e-01 -2.70110577e-01 -1.96991917e-02
8.73569429e-01 3.23477685e-01 2.17301130e-01 -4.24140126e-01
-1.36882281e+00 1.01286173e+00 7.53482759e-01 4.49371517e-01
-2.62646526e-01 1.05602279e-01 4.53511298e-01 2.78677106e-01
4.50557649e-01 3.41431677e-01 -3.87721300e-01 1.92159399e-01
-9.44031775e-01 3.85823011e-01 4.32112813e-01 7.29824126e-01
1.00937402e+00 9.53360349e-02 -3.13301384e-01 1.08435404e+00
1.16288289e-01 4.65156198e-01 5.01966536e-01 -6.29930973e-01
-1.04297828e-02 6.55397832e-01 -2.46012751e-02 -1.10507631e+00
-3.18250448e-01 -5.42024732e-01 -1.22559273e+00 2.75031120e-01
3.45741838e-01 4.31710988e-01 -1.08097923e+00 1.36947262e+00
1.39317527e-01 6.61952123e-02 -5.79449236e-01 9.41919804e-01
1.27683997e+00 5.50879657e-01 2.02802554e-01 1.29565656e-01
1.51378405e+00 -1.23696530e+00 -2.49611020e-01 -2.52575397e-01
4.16015275e-02 -9.05294538e-01 1.17173862e+00 3.38310301e-01
-7.25890815e-01 -9.08909321e-01 -8.18322659e-01 -1.85913265e-01
-8.40953946e-01 1.97906718e-02 9.01379287e-01 3.61708939e-01
-1.00174356e+00 4.30801183e-01 -5.65376639e-01 -3.75069082e-01
6.85207427e-01 3.32553208e-01 -5.25550365e-01 -1.90773010e-01
-9.85157430e-01 6.08958721e-01 2.86205769e-01 -5.89920720e-03
-8.76282156e-01 -7.61570036e-01 -6.99225366e-01 1.00133337e-01
1.77102000e-01 -7.79241025e-01 1.26131630e+00 -6.46493018e-01
-8.72668266e-01 1.52203774e+00 -9.09262076e-02 -3.82744998e-01
6.58268332e-02 -2.47179836e-01 -5.25427997e-01 -7.73178190e-02
3.47040117e-01 6.01596951e-01 1.16309547e+00 -9.74702358e-01
-9.97252643e-01 -5.63502014e-01 2.83468038e-01 -1.06366605e-01
-2.49065638e-01 2.81110108e-01 -5.22724807e-01 -9.12136197e-01
-4.38437313e-01 -9.35482085e-01 -1.85665995e-01 -2.61623245e-02
-2.71276057e-01 -5.59905708e-01 7.57341862e-01 8.89971182e-02
1.32941043e+00 -2.16523504e+00 1.13229692e-01 -1.82408273e-01
6.88267469e-01 4.18631494e-01 -2.52519399e-01 4.08910632e-01
-2.61949837e-01 5.55278026e-02 1.00420222e-01 -2.22493894e-03
2.91628629e-01 1.64643094e-01 -6.30882621e-01 5.27331948e-01
2.35245898e-01 1.28127921e+00 -6.18440807e-01 -3.72930348e-01
5.83668172e-01 2.66069174e-01 -1.85636580e-01 2.28240252e-01
-1.36713162e-01 1.86592042e-01 -7.13199675e-01 7.71921396e-01
7.49481082e-01 -7.95029700e-01 -4.94895875e-01 -7.24088192e-01
-3.12088221e-01 -2.41669238e-01 -1.10667014e+00 1.17745602e+00
-3.06023747e-01 5.45936108e-01 1.36284858e-01 -1.26571488e+00
7.75489807e-01 -3.87023509e-01 3.39590400e-01 -8.47380579e-01
1.53271750e-01 4.72516082e-02 -9.89760980e-02 -1.66967407e-01
6.84982419e-01 -2.61541426e-01 -4.46398646e-01 5.00329435e-01
8.75219852e-02 -1.56680599e-01 2.53932148e-01 -9.15808082e-02
6.64997339e-01 -3.91693175e-01 7.29281127e-01 -4.74548399e-01
6.71510994e-01 -4.98921908e-02 2.26871759e-01 1.00997567e+00
-5.58641613e-01 5.03113389e-01 -1.39671341e-01 -1.05520189e+00
-8.44758689e-01 -1.08663726e+00 -2.63482302e-01 1.78838205e+00
2.71713048e-01 -3.56435120e-01 -6.92492127e-01 -5.75979114e-01
3.69454324e-01 -2.02830866e-01 -1.08661366e+00 5.87736908e-03
-5.63513756e-01 -9.39757466e-01 4.33227211e-01 7.10541487e-01
6.68153882e-01 -1.25917220e+00 -3.35508227e-01 -1.56230908e-02
-1.23503543e-02 -1.08360744e+00 -4.26543325e-01 8.96714553e-02
-4.91215318e-01 -1.03879976e+00 -8.49897146e-01 -9.71783519e-01
2.47894138e-01 7.77002454e-01 1.71711898e+00 2.25704044e-01
-8.34608614e-01 3.00645262e-01 -3.78071010e-01 -5.43853104e-01
1.01208299e-01 9.62361023e-02 1.43514976e-01 1.36805207e-01
6.56325281e-01 -1.92451909e-01 -5.63942611e-01 5.27060628e-01
-1.02346826e+00 -3.17463726e-01 7.19382405e-01 9.59915280e-01
1.12618780e+00 3.57700229e-01 5.29702306e-01 -8.88029277e-01
6.79535747e-01 -3.15345615e-01 -7.26990461e-01 4.31983232e-01
-2.56737292e-01 4.29103076e-02 6.09068751e-01 -8.26873258e-02
-7.63230801e-01 -3.32825184e-01 -2.96495229e-01 -4.31348950e-01
-6.62937164e-01 4.50739056e-01 -1.22183315e-01 -5.08533955e-01
6.20707989e-01 1.29618376e-01 -4.58765656e-01 -6.79009557e-01
5.55918813e-01 7.28720367e-01 5.59191287e-01 -6.03857815e-01
7.13148177e-01 6.76281989e-01 -1.54185548e-01 -9.73501623e-01
-1.44810629e+00 -7.37188399e-01 -6.26624703e-01 5.71879074e-02
8.98029625e-01 -8.19393396e-01 -8.42293739e-01 7.49712825e-01
-7.60240257e-01 -9.55496132e-02 -1.31971225e-01 5.51903695e-02
-4.96111631e-01 3.30561399e-01 -4.96494293e-01 -1.26191631e-01
-4.57898706e-01 -1.23640072e+00 1.65483534e+00 3.61781716e-01
-5.48476651e-02 -1.05872011e+00 3.59156467e-02 5.26415944e-01
5.09972572e-01 2.16431111e-01 8.58672500e-01 -2.15271190e-01
-6.15364194e-01 -7.17648715e-02 -8.29887867e-01 2.45554239e-01
4.35670882e-01 -1.96682215e-02 -1.08766615e+00 -5.16709685e-01
-2.99144089e-01 -5.69186866e-01 1.28513503e+00 6.43714309e-01
1.64712894e+00 7.17440248e-02 -4.26340342e-01 8.93786788e-01
1.41533983e+00 -5.37087722e-03 2.45887607e-01 4.87826973e-01
8.15401018e-01 4.60596979e-01 8.35456192e-01 3.63687903e-01
3.17320347e-01 6.17177308e-01 5.09124160e-01 -2.41252869e-01
-4.72446650e-01 3.79689522e-02 -4.40989554e-01 5.29499829e-01
-9.15841162e-02 -2.78029770e-01 -8.26614678e-01 5.54364860e-01
-1.72960401e+00 -9.57559764e-01 2.08623707e-01 1.69561994e+00
4.26584274e-01 -1.24282844e-01 1.81576312e-01 2.27449834e-02
9.19290304e-01 5.14809549e-01 -7.95920193e-01 -2.76807010e-01
-3.24459702e-01 4.87000674e-01 3.24133039e-01 2.51771629e-01
-1.76775610e+00 1.20487750e+00 6.72119474e+00 1.20399833e+00
-1.36197913e+00 -1.74761280e-01 7.62088895e-01 2.83250570e-01
2.86778454e-02 -7.03082681e-01 -9.55298781e-01 2.60738075e-01
4.25325871e-01 -1.10812418e-01 3.30831707e-01 8.50328207e-01
-2.32319400e-01 2.35725328e-01 -1.10161257e+00 1.40961993e+00
1.92436446e-02 -1.80232215e+00 3.14888299e-01 -8.92570689e-02
9.30537939e-01 4.79575753e-01 3.65783453e-01 2.63542354e-01
3.28282565e-01 -1.48611450e+00 6.02169812e-01 3.31131101e-01
1.17673779e+00 -5.95338404e-01 5.90022385e-01 1.44892409e-01
-1.60890043e+00 -1.72138765e-01 -5.75586140e-01 -2.59176403e-01
-1.99308544e-01 5.54480314e-01 -1.00168176e-01 4.30166423e-01
1.24716175e+00 8.82967412e-01 -6.34727597e-01 9.82066751e-01
1.55179724e-01 4.31390405e-01 3.40338983e-02 1.64547428e-01
6.11248851e-01 9.16241556e-02 1.49979284e-02 1.47462237e+00
-1.26263961e-01 -3.44558358e-02 5.10216594e-01 5.66720486e-01
-1.77930206e-01 -2.80822873e-01 -3.68164986e-01 -2.94445455e-01
3.61797899e-01 1.40901244e+00 -8.41951311e-01 -3.22159886e-01
-6.07124567e-01 7.36840725e-01 4.99737114e-01 2.33612835e-01
-4.93885666e-01 -4.12389636e-01 1.12153745e+00 -1.19252205e-01
6.36135340e-01 -1.43163949e-01 -1.50474846e-01 -1.28932595e+00
-2.75191456e-01 -1.13477671e+00 7.63941467e-01 -4.85051632e-01
-1.88344693e+00 8.71508181e-01 -1.40458554e-01 -1.08040571e+00
-2.27320939e-01 -9.55578923e-01 -3.30340147e-01 8.45795929e-01
-1.97172415e+00 -1.50228167e+00 -6.16496980e-01 9.48042393e-01
6.67941988e-01 -1.90349311e-01 9.36811328e-01 4.26894277e-01
-3.52496594e-01 6.74151659e-01 2.86880344e-01 2.26773590e-01
7.12141991e-01 -1.18578196e+00 6.88792706e-01 4.92289692e-01
2.71709591e-01 6.06263518e-01 4.54698592e-01 -1.75788522e-01
-1.31537926e+00 -1.54572570e+00 6.23973966e-01 -4.49909240e-01
8.35473359e-01 -3.35833579e-01 -8.19992363e-01 5.20534217e-01
5.02915569e-02 6.21433318e-01 6.95889771e-01 2.08744690e-01
-8.07278633e-01 -2.68683761e-01 -1.07946515e+00 1.99270144e-01
8.68784249e-01 -8.37007225e-01 -5.61396241e-01 3.21455985e-01
4.71308380e-01 -1.94714054e-01 -8.81058216e-01 4.64827746e-01
7.07523108e-01 -1.06252313e+00 1.37966144e+00 -7.86594152e-01
1.27741218e-01 -3.92487437e-01 -4.10081327e-01 -1.40305996e+00
-1.05054557e+00 -1.46330506e-01 4.51904349e-02 7.10467458e-01
-4.17362422e-01 -4.33867037e-01 8.58526349e-01 8.08696747e-02
-2.38637596e-01 -6.99881136e-01 -4.89473522e-01 -8.08543503e-01
4.70227212e-01 -3.02392781e-01 8.38481545e-01 6.86189890e-01
-5.22204041e-01 3.06103230e-01 -3.84367704e-01 1.07839674e-01
8.81965756e-01 1.08030796e+00 6.74429119e-01 -1.52427423e+00
-1.03729211e-01 -7.38989055e-01 -6.80597961e-01 -1.01386011e+00
2.73643792e-01 -6.54286861e-01 1.60830524e-02 -1.54647160e+00
4.38760638e-01 -4.54963833e-01 -6.35851562e-01 4.38494056e-01
-1.81915537e-01 1.10663199e+00 2.60340273e-01 3.22116524e-01
-1.00159895e+00 2.44568810e-01 1.37281966e+00 -4.69870329e-01
4.02844906e-01 -1.09744454e-02 -1.04198873e+00 9.37008977e-01
7.29867637e-01 2.95455102e-02 -3.03680718e-01 -6.39383912e-01
1.00950137e-01 -3.45374554e-01 4.08273816e-01 -6.21013582e-01
5.87868057e-02 -5.05643606e-01 4.37610209e-01 -7.83581853e-01
2.44525462e-01 -5.25864065e-01 -1.11968599e-01 1.63446397e-01
-1.42521933e-01 1.56633481e-02 3.85223091e-01 3.84242773e-01
-8.30969393e-01 2.37852842e-01 1.13859141e+00 -4.41982448e-01
-1.47183216e+00 8.50385964e-01 -2.46935412e-01 2.94990897e-01
9.90965545e-01 -2.09978849e-01 -5.65164328e-01 -1.69030565e-03
-6.42431498e-01 -4.33296524e-02 2.79550701e-01 7.21319377e-01
4.59618300e-01 -1.45757830e+00 -7.64389455e-01 2.97028422e-01
7.34195948e-01 -3.36139679e-01 6.39214873e-01 3.05786371e-01
-2.55421877e-01 1.04957259e+00 -6.34150743e-01 -6.49632752e-01
-1.38168657e+00 8.94804478e-01 2.04329401e-01 -3.19150090e-01
-3.38611066e-01 1.18058562e+00 1.15076363e+00 -3.15688938e-01
1.22073047e-01 -3.38928789e-01 -4.28151041e-01 4.64290865e-02
1.01205003e+00 2.64849275e-01 4.32255358e-01 -8.92680347e-01
-7.12226570e-01 9.68892932e-01 -4.60744977e-01 8.06782067e-01
1.49056196e+00 -2.42596596e-01 -2.71501422e-01 2.51635641e-01
1.39061987e+00 -2.39676416e-01 -1.01000440e+00 -4.74565506e-01
-1.06281318e-01 -5.29117763e-01 1.12194501e-01 -7.89186001e-01
-1.01621783e+00 1.07701647e+00 6.55942380e-01 4.42957044e-01
1.37657833e+00 4.79983985e-01 6.54319465e-01 4.30013537e-01
5.14203608e-01 -7.90866911e-01 9.21817347e-02 7.63350129e-01
1.00482452e+00 -1.56478417e+00 -9.13551543e-03 -2.70034581e-01
-1.12159610e-01 9.25706685e-01 3.88626307e-01 -4.74068761e-01
9.79016006e-01 2.75409371e-01 4.55153212e-02 -3.91344905e-01
-6.16887450e-01 -3.80328655e-01 8.24775934e-01 6.40556395e-01
6.24018788e-01 4.78912055e-01 1.89441070e-01 6.19556665e-01
-2.68117547e-01 3.45549434e-02 -7.97951818e-02 6.69455051e-01
-7.60533035e-01 -1.04328775e+00 -5.44485211e-01 7.76613176e-01
-4.37676638e-01 -2.54028827e-01 -4.24790621e-01 5.94443500e-01
3.38933557e-01 8.08123827e-01 3.03810269e-01 -2.88232177e-01
1.45170003e-01 -6.12482488e-01 6.71954274e-01 -6.02863133e-01
-4.25487459e-01 -1.97558343e-01 -2.47310966e-01 -8.04370105e-01
-5.69789648e-01 -3.67588967e-01 -7.23587453e-01 -6.54190183e-01
-4.61748056e-02 1.80457104e-02 2.91054100e-01 1.11087823e+00
4.61087257e-01 4.80307847e-01 4.23903555e-01 -9.64685202e-01
-2.22708866e-01 -6.90405786e-01 -7.92111218e-01 2.77782768e-01
7.22791731e-01 -1.03332293e+00 -1.26049742e-01 4.94294614e-02]
|
[9.63174819946289, 2.0216856002807617]
|
a4cf044a-0e82-4d46-be19-fe3f4c2f4ef6
|
analogy-as-nonparametric-bayesian-inference
|
2006.04156
| null |
https://arxiv.org/abs/2006.04156v1
|
https://arxiv.org/pdf/2006.04156v1.pdf
|
Analogy as Nonparametric Bayesian Inference over Relational Systems
|
Much of human learning and inference can be framed within the computational problem of relational generalization. In this project, we propose a Bayesian model that generalizes relational knowledge to novel environments by analogically weighting predictions from previously encountered relational structures. First, we show that this learner outperforms a naive, theory-based learner on relational data derived from random- and Wikipedia-based systems when experience with the environment is small. Next, we show how our formalization of analogical similarity translates to the selection and weighting of analogies. Finally, we combine the analogy- and theory-based learners in a single nonparametric Bayesian model, and show that optimal relational generalization transitions from relying on analogies to building a theory of the novel system with increasing experience in it. Beyond predicting unobserved interactions better than either baseline, this formalization gives a computational-level perspective on the formation and abstraction of analogies themselves.
|
['Ruairidh M. Battleday', 'Thomas L. Griffiths']
|
2020-06-07
| null | null | null | null |
['analogical-similarity']
|
['reasoning']
|
[ 1.95324644e-02 5.51568985e-01 7.44676515e-02 -5.90019822e-01
-2.02303022e-01 -5.69695652e-01 9.87940609e-01 6.78324044e-01
-8.44943821e-02 5.27670801e-01 7.05451488e-01 -4.37615395e-01
-8.01186025e-01 -1.15731740e+00 -1.15356028e+00 3.38232182e-02
-3.42491537e-01 9.03240085e-01 4.37776625e-01 -4.86027449e-01
4.31634068e-01 6.71650708e-01 -1.55722332e+00 3.90763879e-01
9.06231225e-01 3.58332515e-01 4.37612563e-01 5.12923419e-01
9.13853794e-02 1.03313649e+00 -1.72517389e-01 -5.74007809e-01
2.67025292e-01 -4.01058167e-01 -9.65955555e-01 -6.10276282e-01
8.08386326e-01 -4.38682079e-01 -7.45512128e-01 5.55278420e-01
2.61973590e-02 7.94575036e-01 1.00660479e+00 -9.03075576e-01
-1.26836360e+00 1.25633776e+00 1.09269924e-01 2.48849690e-01
8.75315130e-01 -1.49542719e-01 1.24929702e+00 -1.02713549e+00
8.37411344e-01 1.64563096e+00 1.00283229e+00 2.09539428e-01
-1.90981853e+00 -5.01604497e-01 3.32711309e-01 5.03218949e-01
-1.53955078e+00 -1.96970969e-01 4.02030438e-01 -7.87378788e-01
1.25923240e+00 -5.34846820e-02 8.62982273e-01 9.91908014e-01
6.24383567e-03 3.86293977e-01 1.08077526e+00 -5.05215585e-01
2.03533128e-01 1.97544530e-01 2.01816693e-01 4.07169372e-01
5.77821374e-01 6.49937510e-01 -9.57798719e-01 -1.33197695e-01
8.10881197e-01 6.81868047e-02 -2.97034886e-02 -5.61153889e-01
-9.24653471e-01 4.61715132e-01 8.09070468e-01 2.86673725e-01
-2.71615654e-01 1.68594897e-01 6.71067787e-03 4.18220103e-01
8.66074786e-02 1.11330116e+00 -5.66599071e-01 2.76259959e-01
-5.50595462e-01 5.69026113e-01 8.87660682e-01 1.28637004e+00
9.73536134e-01 -4.43566412e-01 1.91435933e-01 5.71068645e-01
4.12219346e-01 2.11652562e-01 2.01281220e-01 -1.45059240e+00
7.55110085e-02 2.76050568e-01 -2.63809711e-02 -1.07755327e+00
-2.84482539e-01 -2.93908328e-01 -1.32562844e-02 -9.29814130e-02
3.55994463e-01 3.68936092e-01 -3.92084777e-01 2.06527066e+00
1.74860820e-01 2.65539497e-01 2.41690010e-01 2.90922880e-01
7.09959507e-01 6.68582916e-01 1.54748634e-01 -2.80866385e-01
9.72082257e-01 -5.46159446e-01 -3.07547420e-01 -1.98377475e-01
6.81712329e-01 -4.54329073e-01 1.21592641e+00 6.22448683e-01
-1.19812989e+00 -8.83418322e-01 -1.18236351e+00 -3.49053591e-01
-5.57469726e-01 -8.15164268e-01 8.34333122e-01 4.02888536e-01
-1.04005122e+00 1.16312945e+00 -6.78076506e-01 -9.34987247e-01
1.56507678e-02 1.25695705e-01 -3.39532435e-01 -1.33268461e-01
-1.38865781e+00 1.65972352e+00 7.13073254e-01 -2.39898503e-01
-8.77115548e-01 -1.00864637e+00 -8.05545092e-01 7.09177479e-02
3.57232630e-01 -9.94463563e-01 1.42681754e+00 -2.51887888e-01
-1.28328741e+00 5.16165495e-01 7.45805446e-04 -2.00578243e-01
-1.88324317e-01 -2.79531449e-01 -1.50674194e-01 -2.18259364e-01
2.69934107e-02 3.40445757e-01 5.14766946e-02 -1.80347884e+00
-3.84101003e-01 -4.38044339e-01 4.68760878e-01 4.90010709e-01
-1.71937317e-01 -2.26534680e-01 1.10962898e-01 -5.14361978e-01
4.68412369e-01 -8.17416787e-01 -1.12343185e-01 -1.39865637e-01
5.40242530e-02 -4.75999475e-01 2.81112036e-03 -4.42501426e-01
1.04919112e+00 -2.08232236e+00 4.97928411e-01 4.61526513e-01
4.05445069e-01 -4.27046359e-01 -3.45120393e-03 9.28689003e-01
-3.04845780e-01 1.61051318e-01 3.41121584e-01 4.66181673e-02
3.60916317e-01 3.82034570e-01 -7.37223923e-01 -6.43438026e-02
-3.57485563e-01 8.02359939e-01 -1.24257171e+00 -5.13367653e-01
7.01115206e-02 -1.38582274e-01 -9.52025950e-01 2.42262706e-01
-2.58628577e-01 6.86196610e-02 -4.37195003e-02 1.69526264e-01
3.57170314e-01 2.75827255e-02 6.74392104e-01 -2.22745225e-01
1.43218949e-01 6.25022471e-01 -1.04722989e+00 1.83737838e+00
-6.85415089e-01 4.64416921e-01 -6.10409915e-01 -8.02488446e-01
9.47230160e-01 1.41376164e-02 -2.36426190e-01 -2.51727492e-01
-2.24519640e-01 -5.26006985e-03 4.35991049e-01 -5.21909058e-01
4.57662731e-01 -6.31668806e-01 -1.31311879e-01 6.31844163e-01
4.59864676e-01 -9.98598933e-01 -6.61511570e-02 7.15307057e-01
8.71880412e-01 5.11594176e-01 5.39769828e-01 -3.85274917e-01
-6.58226088e-02 -1.68518394e-01 2.90982157e-01 1.28408575e+00
3.27722669e-01 -1.09431922e-01 1.56502247e-01 -3.14283997e-01
-9.64012742e-01 -1.99858725e+00 -4.21742052e-01 1.58800554e+00
2.59375334e-01 -7.11071730e-01 -3.18736166e-01 -2.05187425e-01
3.09393883e-01 1.47744608e+00 -8.20802510e-01 -4.28241611e-01
-3.31294745e-01 -1.67139038e-01 3.10636252e-01 8.84099066e-01
-2.01653600e-01 -9.77012813e-01 -1.55739725e-01 8.24470073e-02
2.23459750e-02 -7.33216465e-01 9.30811241e-02 4.32825029e-01
-1.10089672e+00 -5.74472427e-01 4.08591360e-01 -4.59455043e-01
1.93210304e-01 2.81774644e-02 1.24151850e+00 -2.70329379e-02
-3.16893347e-02 6.71076179e-01 -3.61839682e-01 -5.37140071e-01
-5.19351363e-01 -1.28284380e-01 5.98220110e-01 -8.03009033e-01
4.77119088e-01 -1.31512189e+00 -4.07328941e-02 1.99290246e-01
-5.54758489e-01 -2.17138097e-01 5.98422348e-01 6.90358222e-01
2.13290468e-01 8.07064697e-02 5.15443444e-01 -8.47781241e-01
8.33109319e-01 -7.36963511e-01 -3.58924568e-01 5.33101141e-01
-7.20080853e-01 2.51796663e-01 4.75733548e-01 -5.73807120e-01
-1.57356274e+00 -9.93204862e-02 5.27453542e-01 -3.80153283e-02
-1.47817865e-01 9.13146615e-01 -2.57428020e-01 2.02556401e-01
1.42128611e+00 3.11432667e-02 -2.19999865e-01 -3.29005063e-01
1.02121174e+00 2.67371625e-01 7.28688419e-01 -1.60800254e+00
8.63850653e-01 8.57045352e-02 4.34289686e-02 -5.95781624e-01
-1.20599508e+00 3.17352973e-02 -9.86072063e-01 -2.23045330e-02
6.82576656e-01 -8.25846612e-01 -8.60296845e-01 -2.08958074e-01
-1.20143163e+00 -5.38001359e-01 -7.80599535e-01 7.12365031e-01
-9.05917764e-01 2.91619390e-01 -5.06613493e-01 -6.39228046e-01
6.88625872e-01 -7.02079713e-01 4.16821182e-01 2.62066349e-02
-1.00040758e+00 -1.19016337e+00 4.57453191e-01 1.27622262e-01
2.44466245e-01 -1.80998743e-01 1.33205891e+00 -1.07873642e+00
-8.58692646e-01 -3.56458277e-02 -6.50237277e-02 -6.25571981e-02
-8.31299499e-02 -2.14819536e-02 -8.12479615e-01 1.78500488e-01
5.07246144e-02 -6.16854787e-01 5.92029274e-01 -2.15978734e-02
9.27419782e-01 -4.88453135e-02 -5.85795581e-01 3.17184091e-01
1.18358552e+00 3.01407427e-01 5.01071692e-01 6.66413680e-02
4.75321472e-01 9.43899930e-01 5.38752377e-01 1.69899508e-01
7.65147507e-01 5.59741259e-01 -6.80473372e-02 6.07111454e-01
4.62974273e-02 -8.64984572e-01 1.35539770e-01 8.34989130e-01
-4.20874953e-01 2.77130127e-01 -1.13141501e+00 6.19484305e-01
-1.86831105e+00 -1.37315536e+00 3.73513214e-02 2.16135359e+00
1.44374514e+00 2.31730476e-01 -4.99230623e-02 -3.69942188e-01
4.53822613e-01 -4.89925355e-01 -4.73209858e-01 -4.23667312e-01
1.27965868e-01 4.63254780e-01 -1.50801778e-01 1.01829565e+00
-3.74179810e-01 1.18046832e+00 7.89425564e+00 6.09780192e-01
-1.36117721e-02 -7.95299336e-02 -8.17980096e-02 1.45464405e-01
-5.81443548e-01 4.93035465e-01 -7.42846191e-01 -2.94931233e-01
1.07333302e+00 -5.09633958e-01 9.24626887e-01 7.15831995e-01
-2.43798539e-01 -2.10097641e-01 -2.15154624e+00 3.75778764e-01
2.23347568e-03 -1.17015088e+00 4.94565904e-01 2.81519666e-02
6.59728169e-01 -2.65147120e-01 4.79928553e-02 6.04398787e-01
1.22032511e+00 -1.20039713e+00 7.36851275e-01 1.04093468e+00
1.93768069e-01 -4.02288318e-01 4.14290875e-01 4.33944106e-01
-8.66503835e-01 -3.13982934e-01 -5.79501629e-01 -7.17160463e-01
-6.01425506e-02 1.06677404e-02 -8.05416346e-01 4.16425288e-01
8.02929282e-01 6.55517817e-01 -6.39742196e-01 7.24273741e-01
-5.47011495e-01 2.25814372e-01 -3.19105119e-01 -2.33789608e-02
-5.33542395e-01 -2.15470672e-01 1.98663726e-01 8.91730666e-01
3.03140879e-01 8.20520103e-01 2.21969903e-01 1.20170367e+00
2.68455356e-01 -1.37164235e-01 -1.09839702e+00 3.73316050e-01
1.23113358e+00 6.63497567e-01 -3.14053446e-01 -5.77596068e-01
-5.01834810e-01 3.04174781e-01 7.38559842e-01 3.51866335e-01
-4.04877305e-01 -2.97737837e-01 2.33387575e-01 1.94075495e-01
2.38365218e-01 -2.35229641e-01 -4.39786643e-01 -9.18130577e-01
-3.49008590e-01 -9.13762987e-01 3.54336828e-01 -1.20423615e+00
-1.73754537e+00 1.81943655e-01 1.05201840e+00 -6.03289723e-01
-3.13232243e-01 -6.83216751e-01 -4.41990674e-01 8.73620093e-01
-4.67351377e-01 -9.83862877e-01 -3.45080465e-01 4.11576480e-01
4.20892760e-02 -2.15386674e-01 1.02112949e+00 -3.60368490e-01
6.77841082e-02 5.02440810e-01 -1.26052186e-01 -3.22739720e-01
8.21099758e-01 -1.48373413e+00 4.46737468e-01 4.22516137e-01
2.24635348e-01 1.69287038e+00 1.05718780e+00 -7.87703872e-01
-9.80230927e-01 -5.75790942e-01 8.21763515e-01 -1.22117531e+00
1.21397138e+00 -5.12881577e-01 -1.22041297e+00 1.34118390e+00
1.84677809e-01 -2.34680712e-01 1.05623412e+00 1.05936968e+00
-1.08401644e+00 -2.34210372e-01 -9.91802454e-01 8.19931746e-01
1.67101192e+00 -1.01975453e+00 -1.69936991e+00 4.38384295e-01
1.04328728e+00 -8.49491879e-02 -1.49394214e+00 3.00766557e-01
7.13222802e-01 -1.04366362e+00 1.24696863e+00 -8.73192608e-01
5.66577733e-01 -3.71061444e-01 -7.55016804e-01 -1.41549146e+00
-8.39451134e-01 -3.69180501e-01 -9.49702859e-02 8.93179297e-01
5.64801991e-01 -6.02165461e-01 4.24088240e-01 8.12814593e-01
-2.73982495e-01 -4.43903148e-01 -4.75561678e-01 -1.14593554e+00
6.46703243e-01 -5.02963483e-01 5.99162817e-01 1.11048341e+00
7.37744451e-01 7.56403565e-01 1.89217106e-01 2.83806026e-01
7.39273250e-01 6.19779974e-02 8.31308603e-01 -1.71078598e+00
-8.58308852e-01 -3.93570215e-01 -5.29871166e-01 -1.17248023e+00
2.61865884e-01 -1.19566739e+00 8.92373845e-02 -1.47748733e+00
4.82979238e-01 -3.05554330e-01 -1.23037674e-01 1.22801639e-01
-9.31452215e-02 -3.16371113e-01 2.47642905e-01 1.75216854e-01
-4.29049373e-01 3.69252741e-01 9.76833224e-01 2.41136894e-01
-2.64754415e-01 -2.73686230e-01 -9.59553480e-01 1.00345659e+00
4.79758531e-01 -5.26680529e-01 -8.90876055e-01 -2.11299166e-01
8.57077122e-01 1.82982162e-01 4.70409393e-01 -6.91489637e-01
6.63536668e-01 -4.38005120e-01 4.23106492e-01 -5.05383849e-01
4.90821809e-01 -8.00210655e-01 2.86181599e-01 3.52097809e-01
-6.80562496e-01 1.37966322e-02 2.38362908e-01 8.62360477e-01
2.25630596e-01 -3.49934459e-01 4.51109588e-01 -2.23692715e-01
-5.75721741e-01 -3.25484276e-01 -5.09122074e-01 1.87508568e-01
8.59051466e-01 -3.75856459e-01 -3.13733578e-01 -5.47775447e-01
-1.53918183e+00 3.45505797e-03 5.46466947e-01 7.00734407e-02
6.49085581e-01 -1.28910875e+00 -5.89620352e-01 -3.40342909e-01
3.84438753e-01 -8.00203606e-02 -6.03135899e-02 3.83375973e-01
-2.74305105e-01 1.56296864e-01 -1.86244413e-01 -3.81221563e-01
-6.58947825e-01 1.06482470e+00 3.47868234e-01 2.47716159e-01
-1.81440949e-01 1.06792521e+00 4.88633007e-01 -8.39657903e-01
1.90297157e-01 -5.16989648e-01 8.71791095e-02 -1.08643562e-01
9.99670699e-02 4.42342818e-01 -3.68979931e-01 -4.51170616e-02
-1.01016827e-01 5.84023416e-01 -3.65143001e-01 -4.49504375e-01
1.20918536e+00 -1.85775757e-01 -4.12016779e-01 1.11526668e+00
5.10545194e-01 5.88435493e-02 -9.63580728e-01 -6.43352091e-01
2.61812806e-01 -3.68625849e-01 -3.84428173e-01 -8.52919936e-01
1.78057715e-01 7.79746592e-01 -3.45849954e-02 2.11375296e-01
6.38329268e-01 6.62851572e-01 -1.88710075e-02 1.39299881e+00
4.92554218e-01 -9.21363473e-01 2.69454151e-01 6.58685207e-01
1.19407296e+00 -6.81765854e-01 6.64284766e-01 -5.72021246e-01
-3.29817832e-01 9.66515005e-01 9.05048847e-01 -2.86376655e-01
9.42477942e-01 1.07550673e-01 -5.88468850e-01 -2.72824615e-01
-1.07220733e+00 1.49198681e-01 2.24499568e-01 1.11288464e+00
5.56133270e-01 1.22254863e-02 6.80617318e-02 8.43153000e-01
-8.57352495e-01 -3.04089487e-01 6.67234123e-01 7.15873301e-01
-5.90150476e-01 -7.88817823e-01 -2.06685662e-01 2.96105504e-01
4.43807364e-01 -4.87562984e-01 -6.44210458e-01 1.04274094e+00
2.78071165e-01 6.88477516e-01 1.78376630e-01 -4.83909905e-01
6.01959646e-01 1.14168637e-01 1.25508320e+00 -1.26618147e+00
-3.06643546e-01 -5.12548506e-01 2.28943348e-01 -5.45553029e-01
-2.69454092e-01 -1.02638483e+00 -1.18609273e+00 -4.41610575e-01
-2.84804225e-01 2.01546103e-01 7.46439621e-02 1.10239363e+00
3.00399475e-02 1.96738869e-01 1.33755669e-01 -7.98481822e-01
-9.11268592e-01 -1.11512125e+00 -5.88690400e-01 3.88414562e-01
-9.85291973e-02 -9.81043756e-01 -3.83260965e-01 1.82885259e-01]
|
[9.533038139343262, 6.9905686378479]
|
33c045ff-6373-4930-b0f1-1dfc08a0eeb7
|
hierarchical-regression-network-for-spectral
|
2005.04703
| null |
https://arxiv.org/abs/2005.04703v1
|
https://arxiv.org/pdf/2005.04703v1.pdf
|
Hierarchical Regression Network for Spectral Reconstruction from RGB Images
|
Capturing visual image with a hyperspectral camera has been successfully applied to many areas due to its narrow-band imaging technology. Hyperspectral reconstruction from RGB images denotes a reverse process of hyperspectral imaging by discovering an inverse response function. Current works mainly map RGB images directly to corresponding spectrum but do not consider context information explicitly. Moreover, the use of encoder-decoder pair in current algorithms leads to loss of information. To address these problems, we propose a 4-level Hierarchical Regression Network (HRNet) with PixelShuffle layer as inter-level interaction. Furthermore, we adopt a residual dense block to remove artifacts of real world RGB images and a residual global block to build attention mechanism for enlarging perceptive field. We evaluate proposed HRNet with other architectures and techniques by participating in NTIRE 2020 Challenge on Spectral Reconstruction from RGB Images. The HRNet is the winning method of track 2 - real world images and ranks 3rd on track 1 - clean images. Please visit the project web page https://github.com/zhaoyuzhi/Hierarchical-Regression-Network-for-Spectral-Reconstruction-from-RGB-Images to try our codes and pre-trained models.
|
['Lai-Man Po', 'Tingyu Lin', 'Yuzhi Zhao', 'Wei Liu', 'Qiong Yan']
|
2020-05-10
| null | null | null | null |
['spectral-reconstruction']
|
['computer-vision']
|
[ 8.96802187e-01 -1.55021399e-01 1.56665370e-01 -4.02975768e-01
-9.34183478e-01 -4.61298674e-01 1.39521703e-01 -6.88647151e-01
-1.58424288e-01 4.99346703e-01 2.01092750e-01 -4.39323187e-01
-1.80861995e-01 -8.00974727e-01 -9.77905273e-01 -8.87831688e-01
3.76972705e-01 -3.05819631e-01 -3.81587535e-01 -1.91338509e-01
-3.34473550e-02 3.88947845e-01 -1.39116156e+00 6.47162497e-01
8.60691190e-01 9.85253334e-01 7.52087891e-01 7.66090810e-01
1.20320059e-01 9.19202745e-01 -1.80618763e-01 6.02475256e-02
8.90271425e-01 -5.37479937e-01 -8.28791797e-01 3.58606100e-01
5.11791945e-01 -4.27423775e-01 -4.52593058e-01 1.64180565e+00
6.45303488e-01 7.71989152e-02 3.27298343e-01 -1.02818632e+00
-1.19252396e+00 6.56533301e-01 -9.86788988e-01 -1.25624999e-01
6.24674559e-02 4.00852352e-01 9.42577720e-01 -8.29919457e-01
2.22758397e-01 8.46534252e-01 6.70736492e-01 3.41378480e-01
-1.12340391e+00 -7.41901517e-01 -3.04092020e-02 5.87435007e-01
-1.53382778e+00 -3.74792933e-01 6.41201675e-01 -2.04501361e-01
1.08072209e+00 4.76783186e-01 6.73412085e-01 9.78516936e-01
-3.85168672e-01 6.10790253e-01 1.60506392e+00 -3.87436718e-01
-2.84160197e-01 -1.54511079e-01 -5.85599318e-02 5.85694373e-01
4.76960838e-03 4.27981347e-01 -3.00333411e-01 4.90382969e-01
6.93564475e-01 1.70846805e-01 -9.55375314e-01 2.69787610e-01
-9.95521963e-01 5.90007067e-01 1.12335777e+00 5.98469190e-02
-5.98622084e-01 1.41001746e-01 -6.35609627e-02 2.40901023e-01
1.84307963e-01 3.64364773e-01 -4.96255010e-01 6.03186190e-01
-8.70552480e-01 -3.32144409e-01 1.58805370e-01 9.59384263e-01
8.79498243e-01 2.20641613e-01 1.27404436e-01 1.08365548e+00
3.33640069e-01 5.68698466e-01 3.78562063e-01 -7.69453764e-01
4.37710345e-01 3.27169925e-01 -4.37064581e-02 -3.69910479e-01
-5.90629458e-01 -6.86655760e-01 -1.17415786e+00 3.46995533e-01
-1.23339318e-01 -8.20718780e-02 -1.18432713e+00 1.35705125e+00
-1.51233539e-01 1.97833911e-01 8.07011127e-02 1.38990879e+00
9.74863291e-01 1.10911453e+00 -4.07148823e-02 -1.15161955e-01
1.06349862e+00 -1.36039305e+00 -4.11870986e-01 -4.74993914e-01
1.09044135e-01 -5.71561337e-01 1.24530327e+00 7.46419013e-01
-7.90055633e-01 -6.66387916e-01 -1.22350919e+00 -3.93982172e-01
-4.43526745e-01 4.40834016e-01 5.94923496e-01 5.30346751e-01
-1.15120983e+00 6.88664794e-01 -5.99618018e-01 -2.04345077e-01
5.21371067e-01 2.60544240e-01 -4.05304462e-01 -2.80691832e-01
-1.09165442e+00 8.25464547e-01 5.20245314e-01 5.10365844e-01
-9.29050624e-01 -6.05130970e-01 -5.72592854e-01 7.88994506e-02
2.27372438e-01 -3.78688484e-01 1.04841542e+00 -1.33565795e+00
-1.51943707e+00 7.45990992e-01 6.55620843e-02 -2.00407282e-01
1.50773764e-01 -1.96029037e-01 -6.30551934e-01 1.90847769e-01
-1.14747413e-01 7.59570420e-01 8.98755252e-01 -1.37501311e+00
-5.07476151e-01 -4.15450811e-01 -1.63382459e-02 5.24064004e-01
-2.28828639e-01 -2.85891861e-01 -3.03323328e-01 -3.31749171e-01
4.76655722e-01 -8.70795488e-01 -2.33879507e-01 -1.45902917e-01
-7.43083417e-01 3.89349043e-01 5.59509158e-01 -1.04927254e+00
7.80935585e-01 -2.08661842e+00 1.31221011e-01 1.13498485e-02
-5.10427020e-02 2.41057351e-01 -3.83874148e-01 1.12397961e-01
-6.93299115e-01 2.46933103e-01 -5.70829093e-01 -7.81506076e-02
-1.14594921e-01 -2.18892828e-01 -4.14187670e-01 6.71547174e-01
1.56711504e-01 8.62907767e-01 -6.96139336e-01 1.24936081e-01
3.79911900e-01 6.25827610e-01 -2.11071059e-01 2.04594284e-01
-1.94224760e-01 5.91958463e-01 -1.15914233e-01 9.65284407e-01
1.13202822e+00 -6.00265682e-01 5.59510663e-02 -8.73476446e-01
-3.61909062e-01 1.69759303e-01 -7.75358379e-01 2.00483060e+00
-4.25671071e-01 6.26706898e-01 3.56933594e-01 -9.31838274e-01
5.54786325e-01 1.47096083e-01 3.53881270e-01 -8.77735555e-01
-6.54777884e-02 7.18808249e-02 -1.93872258e-01 -7.06452429e-01
3.69838059e-01 -2.60630727e-01 5.94111204e-01 1.92469656e-01
-9.41799209e-02 -4.15983260e-01 -2.89646000e-01 -4.16590512e-01
8.10446918e-01 4.74220634e-01 3.37333530e-01 7.82127976e-02
4.97548640e-01 7.30632944e-03 3.31296414e-01 7.32432246e-01
-1.32364884e-01 1.05139208e+00 -1.88204050e-02 -3.21198463e-01
-1.16895604e+00 -8.59479070e-01 -2.83754259e-01 1.10590100e+00
1.56306386e-01 1.18044689e-02 -3.17169219e-01 -3.40749472e-01
-3.26766282e-01 6.64490759e-01 -5.45170009e-01 1.57057643e-01
1.00528896e-01 -1.19026661e+00 4.19876039e-01 1.88026488e-01
1.04551148e+00 -1.11300874e+00 -4.76317406e-01 -8.52925107e-02
-4.07745540e-01 -9.91726339e-01 -3.91198210e-02 5.58289528e-01
-7.18773663e-01 -1.18771255e+00 -4.26045984e-01 -4.76932615e-01
4.22320843e-01 8.24681699e-01 7.04707503e-01 -2.23984003e-01
-5.03776133e-01 -8.30070600e-02 -5.97622573e-01 -3.68500084e-01
-7.61210099e-02 -3.25454436e-02 -5.62812090e-01 -2.26706807e-02
3.66336584e-01 -4.68402475e-01 -8.68608117e-01 2.45511737e-02
-1.22036278e+00 5.81301987e-01 7.61921704e-01 8.48215342e-01
5.31316698e-01 3.42495859e-01 2.06569403e-01 -6.94720864e-01
2.04689294e-01 -4.45793808e-01 -8.48966599e-01 3.57603163e-01
-5.02501428e-01 -1.34722903e-01 6.80528045e-01 1.25831231e-01
-1.16348708e+00 5.39077640e-01 -2.17266902e-01 -4.94082868e-01
-2.71150261e-01 6.54762506e-01 -2.82802701e-01 -1.12609930e-01
8.64105225e-01 4.81421560e-01 -2.67072350e-01 -3.75879169e-01
4.54560548e-01 9.99565840e-01 6.17424250e-01 -5.73450932e-03
8.77552986e-01 5.93246996e-01 -8.75277296e-02 -9.76580441e-01
-1.18331265e+00 -6.09757006e-01 -3.43751878e-01 -2.16169104e-01
9.40747797e-01 -1.54381466e+00 -6.37205064e-01 5.58911085e-01
-1.02020943e+00 -6.42927229e-01 8.39904323e-02 4.76529300e-01
-2.59442270e-01 3.99043709e-01 -5.32229662e-01 -6.84465051e-01
-5.26161075e-01 -1.26273680e+00 1.12273669e+00 2.61655867e-01
6.71915174e-01 -5.26024401e-01 -3.57673854e-01 6.49953306e-01
5.97729445e-01 6.11842312e-02 7.03376472e-01 1.61713153e-01
-7.38263249e-01 2.15451300e-01 -8.81203830e-01 7.74236858e-01
2.27114633e-01 -2.18572259e-01 -1.65755320e+00 -2.37952322e-01
2.34749034e-01 -6.28123581e-01 1.37568212e+00 5.96342921e-01
1.86569321e+00 -1.30456284e-01 1.41836077e-01 1.33339846e+00
2.03072977e+00 1.05609834e-01 1.08987200e+00 4.83628124e-01
1.07085502e+00 5.50205708e-01 1.28886610e-01 3.45057040e-01
1.43867016e-01 2.69600689e-01 1.01037276e+00 -5.34438550e-01
-2.16493204e-01 1.06083840e-01 4.88720983e-01 4.43436623e-01
-2.83788979e-01 -3.78981978e-01 -6.89801335e-01 1.53881252e-01
-1.63549709e+00 -1.09741044e+00 -4.27975446e-01 2.12841797e+00
8.60998750e-01 -3.90531421e-01 -4.54245895e-01 -9.53068119e-03
8.30110013e-01 3.89921099e-01 -7.34170914e-01 2.61144757e-01
-5.77507138e-01 2.84349322e-01 1.11024201e+00 6.28032804e-01
-1.28264904e+00 9.72054064e-01 5.15296316e+00 6.52979732e-01
-1.51511860e+00 3.29558998e-01 6.73990130e-01 -1.17703073e-01
-1.60764813e-01 1.19992886e-02 -3.51092309e-01 9.58272815e-02
7.28039265e-01 4.16262448e-01 1.17912447e+00 4.25748050e-01
3.72356713e-01 -7.83140734e-02 -6.32178128e-01 1.35220599e+00
9.68576521e-02 -1.07079077e+00 -2.90401280e-03 3.76432277e-02
8.23275805e-01 6.95777833e-01 2.29196087e-01 5.11757098e-02
2.22427160e-01 -1.44047177e+00 6.95198238e-01 5.47559142e-01
1.14755213e+00 -4.29111123e-01 4.61269200e-01 6.26155585e-02
-1.13185096e+00 -3.67148787e-01 -8.08024585e-01 1.67766705e-01
-1.66021720e-01 5.38638115e-01 -7.06092715e-01 8.47509503e-01
8.26170266e-01 9.67556536e-01 -5.74105740e-01 1.07692969e+00
-4.87497598e-01 5.91464698e-01 -1.63167253e-01 5.17419577e-01
3.00965995e-01 -5.69255173e-01 2.13855475e-01 1.15730572e+00
6.21322572e-01 4.22760248e-01 -1.68147445e-01 1.06947637e+00
-2.53820062e-01 -1.89091563e-01 -6.21874034e-01 -4.21064794e-02
8.94789919e-02 1.53663695e+00 -3.98672611e-01 -1.40692554e-02
-6.17347002e-01 1.21441734e+00 3.36260535e-02 6.24272585e-01
-9.83033717e-01 -2.06800714e-01 3.84941190e-01 -1.28494993e-01
2.20158130e-01 1.11860763e-02 -2.49994442e-01 -1.21703398e+00
-2.65746891e-01 -1.07826769e+00 2.01768011e-01 -1.83709264e+00
-1.24827182e+00 7.25302517e-01 -3.98624808e-01 -1.36325407e+00
4.51334715e-01 -1.03695500e+00 -9.12717506e-02 1.19544160e+00
-2.13909125e+00 -1.35223997e+00 -1.07740092e+00 7.95030951e-01
5.87901592e-01 8.80865380e-02 8.14830363e-01 3.01738024e-01
-6.50553763e-01 -2.62389565e-03 2.39782661e-01 -3.49836685e-02
7.23948240e-01 -1.23091030e+00 1.40143052e-01 1.00313091e+00
-1.39532566e-01 3.04745615e-01 4.26909328e-01 -4.62715924e-01
-1.61416614e+00 -1.51042426e+00 -5.34653058e-03 2.20512804e-02
5.70119143e-01 -4.11684439e-02 -8.67115915e-01 7.68911958e-01
5.57124913e-01 7.50527903e-02 6.15524113e-01 -3.18743974e-01
-5.46112835e-01 -3.81051451e-01 -7.87665606e-01 3.51496547e-01
1.00486255e+00 -8.59157324e-01 -6.88297004e-02 8.38136673e-01
7.19119251e-01 -3.20389777e-01 -5.80710232e-01 5.65216243e-01
3.65352064e-01 -1.17474854e+00 1.06433439e+00 -1.39773339e-01
8.29241991e-01 -6.18966222e-01 -4.63935614e-01 -1.43256700e+00
-4.92768705e-01 -4.29967552e-01 4.63585109e-01 6.10357940e-01
6.68715298e-01 -4.33719993e-01 6.53722942e-01 1.81345299e-01
-5.54585338e-01 -8.39277208e-02 -3.54442120e-01 -5.47433496e-01
-3.08704257e-01 -5.05735457e-01 7.03329206e-01 1.11646533e+00
-2.79002696e-01 3.55204165e-01 -6.60082281e-01 8.11676085e-01
7.71829247e-01 3.28474432e-01 4.02827412e-01 -8.63441110e-01
-5.40130138e-01 -4.92734730e-01 1.23270057e-01 -9.04842377e-01
1.63995363e-02 -1.10056615e+00 2.71239907e-01 -1.73726571e+00
3.36989522e-01 -1.51086822e-01 -4.18280154e-01 7.44950771e-01
-5.23762554e-02 5.16455173e-01 1.94073960e-01 2.09695831e-01
-3.54657054e-01 5.26811838e-01 1.32207632e+00 -5.21756291e-01
-1.29616693e-01 -3.29736739e-01 -8.46105278e-01 4.71469760e-01
9.97429311e-01 -3.59686017e-01 -3.46670896e-01 -7.79792428e-01
4.65013623e-01 1.80805475e-01 6.94745958e-01 -9.55777645e-01
4.17840093e-01 -3.02642077e-01 6.17344558e-01 -4.13606167e-01
4.78271604e-01 -9.44730043e-01 5.34176052e-01 2.13363841e-01
-2.94981539e-01 -2.94809908e-01 1.79629713e-01 3.09696674e-01
-1.55399263e-01 -4.40639347e-01 8.36558282e-01 -4.75085527e-01
-9.70508218e-01 3.45223486e-01 -3.76763120e-02 -6.34712875e-01
5.48427403e-01 -2.27017671e-01 -6.92308187e-01 -3.26283067e-01
-7.40016341e-01 -1.04452029e-01 3.91669631e-01 6.61109462e-02
7.09711373e-01 -8.70562017e-01 -8.58844280e-01 2.18557373e-01
2.33180717e-01 3.52292880e-02 5.52614987e-01 7.24816322e-01
-9.84252274e-01 2.14641154e-01 -4.33788151e-01 -4.10184354e-01
-9.56527531e-01 3.40275705e-01 8.41032624e-01 1.07430071e-02
-5.69041073e-01 1.06583166e+00 3.51395965e-01 -5.63786328e-01
3.83280148e-03 -3.81039649e-01 -7.05166683e-02 -1.51621148e-01
4.49107796e-01 1.97503671e-01 3.70636195e-01 -4.26512688e-01
-3.47909331e-02 4.42290395e-01 3.33971918e-01 -6.13580383e-02
1.69754696e+00 -2.08768368e-01 -4.85878021e-01 2.55558789e-02
1.26564407e+00 -2.98133016e-01 -1.45566416e+00 -1.28802508e-01
-4.47188616e-01 -4.24988329e-01 6.51359677e-01 -1.19707644e+00
-1.34243309e+00 9.78740096e-01 8.86399031e-01 1.64679006e-01
1.72703850e+00 -6.15757227e-01 1.98155195e-01 4.92155701e-01
8.39988813e-02 -8.81651580e-01 -1.16354235e-01 5.22435606e-01
1.13732505e+00 -1.67729306e+00 2.08163500e-01 -3.46974641e-01
-6.46131456e-01 1.20876813e+00 5.81705570e-01 2.49248490e-01
4.28659767e-01 6.65269792e-02 7.73037001e-02 -2.91786849e-01
-3.83448184e-01 -5.95763922e-01 1.69355825e-01 6.90586686e-01
4.89476800e-01 2.15508789e-01 3.73199701e-01 3.00627470e-01
-1.08449019e-01 6.65364265e-02 7.55689204e-01 4.92103189e-01
-5.36535084e-01 -6.63388729e-01 -5.60318470e-01 4.00437355e-01
-2.36148477e-01 -7.98787177e-01 -2.52245843e-01 5.57038248e-01
1.08504012e-01 1.17261899e+00 -3.09563607e-01 -5.13247073e-01
2.00937673e-01 6.51770458e-02 4.85927880e-01 -5.63319206e-01
-5.08199155e-01 3.59851092e-01 -1.05781354e-01 -6.12380683e-01
-6.74420416e-01 -4.06731486e-01 -9.00186419e-01 -1.42082378e-01
-3.81821513e-01 -1.35960624e-01 1.04329848e+00 6.00808799e-01
1.59044042e-01 6.92739010e-01 6.72257662e-01 -1.08989859e+00
-4.97007906e-01 -1.26475322e+00 -9.18472588e-01 1.68010369e-01
5.88773191e-01 4.23165597e-02 -3.80916655e-01 3.07065725e-01]
|
[10.21755313873291, -2.046236276626587]
|
43540870-51f1-4311-bc18-7982b33380d8
|
pushing-the-frontiers-of-unconstrained-face
| null | null |
http://openaccess.thecvf.com/content_cvpr_2015/html/Klare_Pushing_the_Frontiers_2015_CVPR_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2015/papers/Klare_Pushing_the_Frontiers_2015_CVPR_paper.pdf
|
Pushing the Frontiers of Unconstrained Face Detection and Recognition: IARPA Janus Benchmark A
|
Rapid progress in unconstrained face recognition has resulted in a saturation in recognition accuracy for current benchmark datasets. While important for early progress, a chief limitation in most benchmark datasets is the use of a commodity face detector to select face imagery. The implication of this strategy is restricted variations in face pose and other confounding factors. This paper introduces the IARPA Janus Benchmark A (IJB-A), a publicly available media in the wild dataset containing 500 subjects with manually localized face images. Key features of the IJB-A dataset are: (i) full pose variation, (ii) joint use for face recognition and face detection benchmarking, (iii) a mix of images and videos, (iv) wider geographic variation of subjects, (v) protocols supporting both open-set identification (1:N search) and verification (1:1 comparison), (vi) an optional protocol that allows modeling of gallery subjects, and (vii) ground truth eye and nose locations. The dataset has been developed using 1,501,267 million crowd sourced annotations. Baseline accuracies for both face detection and face recognition from commercial and open source algorithms demonstrate the challenge offered by this new unconstrained benchmark.
|
['Jordan Cheney', 'Austin Blanton', 'Alan Mah', 'Kristen Allen', 'Anil K. Jain', 'Patrick Grother', 'Emma Taborsky', 'Brendan F. Klare', 'Ben Klein']
|
2015-06-01
| null | null | null |
cvpr-2015-6
|
['robust-face-recognition']
|
['computer-vision']
|
[ 3.91219109e-02 -2.97702789e-01 -1.56963784e-02 -7.47689903e-01
-9.17170584e-01 -7.09637105e-01 7.46347189e-01 -5.99175334e-01
-4.90472853e-01 6.53371274e-01 1.58135459e-01 4.39845212e-02
2.60051078e-04 -2.51737505e-01 -4.59207803e-01 -6.19169891e-01
-3.19450915e-01 6.91133380e-01 -2.35607237e-01 3.72030102e-02
-3.05933319e-02 9.77882385e-01 -1.93261290e+00 3.06056648e-01
-6.82599396e-02 1.20895052e+00 -1.91777751e-01 5.98601878e-01
3.87361705e-01 3.01081818e-02 -6.01112008e-01 -7.78908908e-01
8.19132507e-01 -1.73104629e-01 -6.93469346e-01 2.11331651e-01
1.28977096e+00 -7.53163576e-01 -2.58984953e-01 7.86013424e-01
1.20631528e+00 3.31265628e-02 4.82791156e-01 -1.51400828e+00
-4.01999295e-01 -2.54492071e-02 -8.36965859e-01 2.45854318e-01
6.74525261e-01 7.91211963e-01 4.79974687e-01 -1.20940411e+00
7.61880517e-01 1.58784997e+00 9.55143750e-01 9.03190076e-01
-1.31247401e+00 -1.05029249e+00 -3.94011140e-01 -2.08712742e-01
-1.94550526e+00 -1.45522523e+00 1.29612342e-01 -4.75688308e-01
1.04246867e+00 3.54822189e-01 2.71909714e-01 1.32434523e+00
-4.08439517e-01 3.55230495e-02 1.24272966e+00 -3.48477840e-01
-7.81669095e-02 3.93592298e-01 -2.07967907e-01 6.68666959e-01
2.62051612e-01 3.64685386e-01 -7.19562650e-01 -4.29002404e-01
4.39262390e-01 -2.32851118e-01 -2.30005205e-01 -1.61273051e-02
-7.19523847e-01 6.43380284e-01 1.61515072e-01 -2.64811218e-01
-1.03063181e-01 -2.84037858e-01 4.88981515e-01 2.19511509e-01
2.00401813e-01 9.08297375e-02 -3.40789169e-01 1.98752150e-01
-1.16615784e+00 3.33693594e-01 9.04709756e-01 7.96147764e-01
6.32114649e-01 8.42811093e-02 -4.91638184e-01 1.10416639e+00
5.87678015e-01 9.88952518e-01 3.92516732e-01 -8.80390108e-01
2.99597919e-01 2.90871173e-01 1.45784765e-01 -1.01595879e+00
-4.06583786e-01 -1.46743702e-02 -1.88544989e-01 4.10557777e-01
5.85320115e-01 -2.93722898e-01 -9.44630325e-01 1.85020828e+00
5.09421051e-01 -3.20029408e-02 -1.65358096e-01 6.90254748e-01
1.25569940e+00 1.71882078e-01 -1.07549414e-01 -8.84851366e-02
1.42809653e+00 -3.41139227e-01 -2.36325607e-01 -2.07622081e-01
1.00668527e-01 -9.40702438e-01 6.50340676e-01 1.35547861e-01
-8.13747227e-01 -4.86285537e-01 -6.20488107e-01 1.56363472e-01
-5.27914464e-01 3.12712044e-01 4.50472444e-01 1.50279307e+00
-1.62829292e+00 -2.79143099e-02 -1.23938605e-01 -8.40427458e-01
8.60896170e-01 9.19846416e-01 -9.96932209e-01 -2.41250962e-01
-6.55493200e-01 6.57239735e-01 -4.49675694e-02 1.07742481e-01
-1.22350693e+00 -7.82122910e-01 -8.31629753e-01 -3.80087167e-01
3.75846088e-01 -7.04173073e-02 9.24087524e-01 -9.50306177e-01
-1.16413236e+00 1.59662259e+00 -2.55981892e-01 -1.86206117e-01
5.61505139e-01 1.15630038e-01 -6.77440345e-01 1.88046291e-01
3.12745363e-01 1.13935304e+00 1.02792633e+00 -9.96118903e-01
-4.26271170e-01 -7.65457153e-01 -3.81967932e-01 -9.58917141e-02
-2.60711432e-01 7.34440982e-01 -6.90435171e-01 -4.22452748e-01
-5.70835829e-01 -9.81918991e-01 4.49564934e-01 2.28658110e-01
-4.47146237e-01 -2.01295763e-01 9.92037058e-01 -9.52089667e-01
5.74371338e-01 -2.19413710e+00 -4.77269709e-01 3.34679663e-01
6.72501773e-02 4.90378499e-01 -5.19466281e-01 1.19243763e-01
-3.15365523e-01 7.10914806e-02 3.04588616e-01 -6.68975830e-01
-1.92522593e-02 -1.18877128e-01 -6.46466389e-02 9.18598950e-01
2.59658366e-01 6.38535678e-01 -3.31682533e-01 -5.37854254e-01
1.36638388e-01 6.05206430e-01 -8.26662302e-01 1.21754117e-01
2.63111413e-01 3.42892975e-01 1.53066248e-01 1.43389487e+00
9.78168845e-01 2.29528844e-01 -8.65443870e-02 -4.51524556e-01
-3.87345888e-02 -2.79500842e-01 -1.44667745e+00 1.13823402e+00
-4.14529070e-02 7.25326896e-01 7.99499333e-01 -2.01280758e-01
7.84237027e-01 4.14802819e-01 4.81635273e-01 -2.43340507e-01
2.65932113e-01 1.40445769e-01 8.01774040e-02 -3.15096498e-01
8.41733590e-02 1.08496308e-01 4.68200147e-01 3.65845561e-01
5.66252470e-01 2.63676822e-01 2.88225561e-01 -7.51709938e-02
8.15528750e-01 2.72624884e-02 1.08965427e-01 -4.42310542e-01
4.46090251e-01 -3.49080473e-01 5.99810064e-01 6.38961375e-01
-7.74103284e-01 8.91276956e-01 3.12550157e-01 -4.00365829e-01
-6.97568238e-01 -7.48186350e-01 -6.79567337e-01 1.22884321e+00
-5.26552558e-01 -3.19189429e-01 -9.24166977e-01 -4.00933951e-01
4.27410573e-01 9.34493169e-03 -1.03487992e+00 2.43651211e-01
-1.49162099e-01 -9.29953516e-01 1.11000395e+00 2.95220256e-01
6.70520127e-01 -9.66722310e-01 -3.63638937e-01 -4.30910438e-01
1.27749592e-01 -1.13082194e+00 -8.53901148e-01 -2.51989514e-01
-1.67905942e-01 -1.46471167e+00 -7.31731176e-01 -7.42967486e-01
5.74510753e-01 1.63125351e-01 1.12323892e+00 2.15555504e-01
-8.54898095e-01 9.05978739e-01 3.29790488e-02 -4.17294711e-01
-1.03787147e-02 -4.23307419e-01 5.96369386e-01 4.90345061e-01
5.62997818e-01 4.67372984e-02 -6.85400248e-01 7.92368054e-01
-4.20415610e-01 -7.46937156e-01 3.66211504e-01 6.98624313e-01
2.95265585e-01 -3.19376647e-01 4.49883729e-01 -6.23711526e-01
2.38653198e-01 -3.45685989e-01 -8.40204537e-01 3.89540374e-01
-3.26822966e-01 -6.25045240e-01 -7.47686103e-02 -1.59696952e-01
-1.09487545e+00 4.63505238e-01 -1.40188605e-01 -4.15704370e-01
-3.74102950e-01 -2.94607431e-01 -5.77152431e-01 -6.88466251e-01
8.80683124e-01 -7.54325092e-02 5.61926067e-01 -4.19677794e-01
1.90203451e-02 9.03783023e-01 6.28634036e-01 -5.87673485e-01
6.63827240e-01 5.76746702e-01 -2.31604148e-02 -1.38594186e+00
-2.35171422e-01 -4.51896340e-01 -7.60779381e-01 -2.68958718e-01
6.93954825e-01 -1.23238492e+00 -1.05613005e+00 7.30938315e-01
-7.95125604e-01 -2.30298862e-01 6.04768991e-02 1.94000095e-01
3.39171737e-02 1.73548032e-02 -2.53845215e-01 -9.87162411e-01
-2.63545811e-01 -1.41691995e+00 1.36459625e+00 3.61474872e-01
-2.44938508e-01 -6.29396737e-01 -1.72163486e-01 5.69619596e-01
3.39112461e-01 3.58245909e-01 8.35544392e-02 -8.69903684e-01
-1.49797887e-01 -6.24493420e-01 -3.97456944e-01 4.12851840e-01
2.63762444e-01 2.75045127e-01 -1.57064128e+00 -7.21733212e-01
-3.79953384e-01 -8.06232989e-01 6.07338727e-01 3.02014381e-01
8.30745161e-01 -2.33165815e-01 -3.88427556e-01 9.55306232e-01
1.15152347e+00 5.46966381e-02 5.13938725e-01 -1.19545624e-01
5.47356665e-01 8.71234894e-01 3.33104730e-02 4.25822318e-01
1.88125260e-02 8.59309971e-01 1.75022170e-01 -7.79732876e-03
-4.06715006e-01 3.90858063e-03 3.91384482e-01 -3.22731346e-01
-8.37430134e-02 1.19437799e-01 -1.00733161e+00 3.81343544e-01
-9.33390200e-01 -1.11910605e+00 2.55296797e-01 2.37556124e+00
6.15446210e-01 -4.30027455e-01 6.80200815e-01 -1.70009390e-01
9.52085376e-01 1.97189618e-02 -5.50229430e-01 -2.87647191e-02
-2.95405954e-01 3.10953081e-01 5.67063093e-01 3.77367586e-01
-1.20115364e+00 6.51131868e-01 7.18962908e+00 6.60150707e-01
-1.11714590e+00 2.52754599e-01 9.50879335e-01 -7.25435138e-01
5.88055372e-01 -5.52919328e-01 -1.59948301e+00 5.03460824e-01
8.05891216e-01 1.97909653e-01 7.39040375e-01 7.96474814e-01
-2.05186754e-01 -1.79071408e-02 -1.20392025e+00 1.43664014e+00
7.27062464e-01 -1.08962882e+00 -4.66982014e-02 2.70319045e-01
6.32540286e-01 2.74095148e-01 3.23466718e-01 2.24576026e-01
-1.62953436e-01 -1.56111717e+00 5.23454905e-01 2.21434295e-01
1.66022074e+00 -5.43390632e-01 6.37764037e-01 -3.39495689e-01
-1.07051444e+00 -3.65859181e-01 -2.71336615e-01 4.07916427e-01
-3.43692154e-01 -7.58350566e-02 -9.12233591e-01 1.67486414e-01
1.06858277e+00 3.16679627e-01 -1.05950511e+00 9.45405543e-01
3.32363516e-01 5.22662342e-01 -7.20949113e-01 2.54333884e-01
-1.20778628e-01 1.68639317e-01 4.67061311e-01 1.18881428e+00
8.35248157e-02 1.58051196e-02 7.40201920e-02 5.70542157e-01
-4.38762158e-01 -9.21399891e-02 -6.69972479e-01 2.92518903e-02
6.08130813e-01 1.48239052e+00 -7.63824403e-01 1.90208867e-01
-7.65493751e-01 6.50408447e-01 -4.39705560e-03 3.10921192e-01
-5.25521517e-01 -1.02716908e-01 1.11700809e+00 3.31426620e-01
2.17805296e-01 4.28693801e-01 1.09009534e-01 -5.59612155e-01
-8.42382461e-02 -1.35440278e+00 5.63647151e-01 -5.27893305e-01
-1.23466420e+00 6.76616788e-01 5.88276647e-02 -6.48143649e-01
-1.50939211e-01 -1.07758355e+00 -4.88252372e-01 1.34509659e+00
-9.30997849e-01 -1.43918276e+00 -4.53100681e-01 8.89580786e-01
3.21594894e-01 -8.85992110e-01 8.80298853e-01 6.43226326e-01
-1.03190446e+00 1.21119010e+00 9.38509684e-03 4.00022566e-01
1.10331094e+00 -8.44924569e-01 1.79945156e-01 6.28949583e-01
1.00391641e-01 9.24875855e-01 4.33275223e-01 -6.54078424e-01
-1.58066976e+00 -1.22450721e+00 3.79112035e-01 -1.12607026e+00
2.84692589e-02 -5.65841556e-01 -4.24302787e-01 8.74421537e-01
-1.24644853e-01 4.37370390e-01 8.19093704e-01 1.67967364e-01
-3.81509334e-01 -5.54011285e-01 -1.78492200e+00 3.76154780e-01
9.40596700e-01 -6.93354666e-01 -3.95626388e-02 6.22541964e-01
-4.82937582e-02 -4.48078305e-01 -5.77346027e-01 2.98428714e-01
9.15388346e-01 -1.08737326e+00 1.15873861e+00 -3.82296860e-01
-8.67062435e-03 -1.27851769e-01 -4.10569191e-01 -7.13457346e-01
-2.00879970e-03 -8.23415399e-01 2.92741150e-01 1.80901098e+00
2.41591364e-01 -7.76979625e-01 7.06023276e-01 1.07934749e+00
3.63503665e-01 -5.23787856e-01 -1.16743720e+00 -7.20265687e-01
-2.61218458e-01 -7.87546486e-02 6.97505295e-01 6.14858508e-01
-8.03142905e-01 -1.16380416e-01 -3.36016268e-01 1.59425154e-01
6.86664701e-01 -2.77324080e-01 1.03787243e+00 -1.24369967e+00
-6.98764175e-02 -3.62635881e-01 -6.29474878e-01 -3.06670129e-01
1.92831188e-01 -8.19673419e-01 -1.22054771e-01 -7.87923694e-01
4.66400832e-01 -2.08395794e-01 2.28588015e-01 7.54169583e-01
1.89845830e-01 1.00825584e+00 -5.39489016e-02 2.28175506e-01
-3.90929312e-01 -1.33504355e-02 5.17278731e-01 1.90314248e-01
9.21029225e-02 -2.21568808e-01 -7.30381966e-01 4.95992333e-01
3.68527234e-01 -1.67210326e-01 -1.02188043e-01 -2.17913583e-01
-3.17117810e-01 -3.82928848e-01 8.55792999e-01 -1.01693964e+00
1.30196050e-01 2.27520525e-01 1.14611924e+00 -3.31411809e-01
6.21054947e-01 -6.72121286e-01 1.87211201e-01 4.11125630e-01
1.08965665e-01 8.48656669e-02 7.26243377e-01 3.07074815e-01
2.41818368e-01 -1.05870314e-01 1.18843889e+00 -1.41725227e-01
-7.35535443e-01 5.40826142e-01 2.13550419e-01 3.19188237e-01
1.01571047e+00 -5.12249291e-01 -2.54216313e-01 -8.81311595e-02
-3.69658917e-01 1.93621162e-02 6.57572806e-01 4.62230355e-01
1.59153193e-01 -1.18503845e+00 -9.86543894e-01 7.53578365e-01
2.09121689e-01 -6.39304876e-01 8.16338882e-03 5.68031669e-01
-4.23403233e-01 5.07446170e-01 -4.80604529e-01 -6.75261796e-01
-1.80914497e+00 1.94083408e-01 6.36766553e-01 5.36887288e-01
-2.28931203e-01 1.32915437e+00 3.51677015e-02 -4.58836883e-01
4.30050284e-01 5.61627388e-01 3.24681886e-02 4.02827471e-01
1.04329562e+00 5.52002013e-01 2.40373001e-01 -1.43401766e+00
-9.02227402e-01 3.50945652e-01 -2.51944736e-02 -1.43689364e-01
1.14385879e+00 6.80267438e-02 -2.19080850e-01 -1.79717764e-01
1.21633637e+00 1.83829620e-01 -1.27956915e+00 8.43951777e-02
-2.00038910e-01 -8.51450980e-01 -1.40449196e-01 -9.81760919e-01
-1.27862418e+00 4.08360034e-01 1.32455647e+00 -3.01282853e-01
8.96596789e-01 -6.72128201e-02 2.33857587e-01 2.49520168e-01
3.32733989e-01 -8.84542167e-01 -1.45053804e-01 2.21829981e-01
1.28013122e+00 -1.55264556e+00 1.03303783e-01 -2.18787327e-01
-3.68032843e-01 8.44021499e-01 7.50323117e-01 4.03124869e-01
6.03348315e-01 3.14429969e-01 2.03774512e-01 -2.69981056e-01
-2.66903430e-01 -1.17918335e-01 3.56706738e-01 1.00128758e+00
4.45030987e-01 -1.95070297e-01 2.49887481e-01 5.37453771e-01
-2.87171721e-01 -1.79891363e-01 -1.60090238e-01 6.27032280e-01
7.75039047e-02 -7.76774466e-01 -8.97892535e-01 7.99313724e-01
-6.50183260e-01 8.37166160e-02 -7.25446463e-01 9.36476290e-01
3.51419181e-01 1.20196974e+00 7.37731606e-02 -7.15250075e-02
2.61313826e-01 3.37329715e-01 6.45007372e-01 -7.61261344e-01
-7.75882840e-01 -3.87744993e-01 -2.65320367e-03 -7.62750864e-01
-1.19885162e-01 -9.51011419e-01 -4.34784055e-01 -4.56267178e-01
-1.70114040e-01 -1.98382378e-01 6.38427973e-01 5.75283706e-01
5.82650959e-01 -1.84557185e-01 5.02483189e-01 -1.23891783e+00
-6.66657209e-01 -1.15170419e+00 -7.92184293e-01 4.64054883e-01
3.36214870e-01 -9.09035742e-01 -5.84387243e-01 1.04963973e-01]
|
[13.390953063964844, 0.8259932398796082]
|
521373b5-a880-43dd-8058-929e7581166d
|
diverse-and-relevant-visual-storytelling-with
| null | null |
https://aclanthology.org/2020.conll-1.34
|
https://aclanthology.org/2020.conll-1.34.pdf
|
Diverse and Relevant Visual Storytelling with Scene Graph Embeddings
|
A problem in automatically generated stories for image sequences is that they use overly generic vocabulary and phrase structure and fail to match the distributional characteristics of human-generated text. We address this problem by introducing explicit representations for objects and their relations by extracting scene graphs from the images. Utilizing an embedding of this scene graph enables our model to more explicitly reason over objects and their relations during story generation, compared to the global features from an object classifier used in previous work. We apply metrics that account for the diversity of words and phrases of generated stories as well as for reference to narratively-salient image features and show that our approach outperforms previous systems. Our experiments also indicate that our models obtain competitive results on reference-based metrics.
|
['Bernt Schiele', 'Vera Demberg', 'Khushboo Mehra', 'Asad Sayeed', 'Rakshith Shetty', 'Xudong Hong']
|
2020-11-01
| null | null | null |
conll-2020
|
['visual-storytelling']
|
['natural-language-processing']
|
[ 3.53880316e-01 3.90571579e-02 -2.80931801e-01 -3.88325363e-01
-6.95582449e-01 -6.46318972e-01 1.40622365e+00 4.21386659e-01
-3.02693814e-01 5.56752980e-01 1.02769601e+00 2.65227258e-01
-2.15836223e-02 -8.69729757e-01 -4.49993521e-01 -2.07438201e-01
1.81770474e-01 3.79527271e-01 4.92339611e-01 -2.86011994e-01
6.88675582e-01 4.50770408e-01 -1.78178167e+00 6.14151716e-01
2.26953655e-01 5.00936091e-01 4.15785730e-01 7.40920484e-01
-3.33000451e-01 1.34979117e+00 -7.39894390e-01 -4.22767103e-01
-1.85523853e-01 -6.42372310e-01 -1.00701618e+00 5.92437088e-01
6.60858274e-01 -3.34382266e-01 -5.02299011e-01 6.82549953e-01
2.34130919e-01 2.45077193e-01 1.20254254e+00 -1.05674338e+00
-8.86623204e-01 7.19218910e-01 -3.96141142e-01 3.39542717e-01
7.98664272e-01 -8.68950188e-02 1.22737432e+00 -1.17483580e+00
1.41063154e+00 1.35172141e+00 4.35433447e-01 3.32770318e-01
-1.20797634e+00 -1.91439420e-01 2.21103296e-01 4.49512780e-01
-1.47014248e+00 -5.58623970e-01 7.28955686e-01 -8.44306707e-01
1.27083218e+00 3.97557944e-01 6.83212578e-01 1.21389771e+00
1.17565520e-01 8.51636112e-01 6.26786947e-01 -5.72297037e-01
-5.57030514e-02 3.06018829e-01 1.53285787e-02 5.71161747e-01
1.27666399e-01 -3.10390502e-01 -6.39040649e-01 4.99031646e-03
9.31056142e-01 -2.55225867e-01 -2.25110605e-01 -7.06352055e-01
-1.58823395e+00 1.01094556e+00 2.78117090e-01 4.76856112e-01
-4.34625536e-01 3.14871728e-01 3.88301730e-01 -1.61743134e-01
5.68260312e-01 8.22474360e-01 9.82355848e-02 -1.66028336e-01
-1.05164289e+00 7.28827059e-01 6.47850752e-01 1.25620985e+00
5.57064354e-01 -3.52199525e-02 -4.55784142e-01 8.98511052e-01
-8.53927061e-02 1.25031978e-01 3.78896296e-01 -1.06305850e+00
2.99119681e-01 5.90575635e-01 7.30598941e-02 -1.54025221e+00
-1.85776487e-01 -2.51574248e-01 -2.70495147e-01 -1.15146441e-02
2.07561776e-02 4.31742430e-01 -6.29638076e-01 1.60048580e+00
8.81075636e-02 -2.59570569e-01 -8.13380908e-03 6.55870616e-01
1.03355515e+00 6.17815554e-01 2.12540150e-01 -1.27032995e-01
1.44415903e+00 -9.16756332e-01 -5.75074196e-01 -3.20980191e-01
6.96364403e-01 -8.62494230e-01 1.05728269e+00 4.52602170e-02
-1.22451556e+00 -4.06001091e-01 -9.53143954e-01 -2.84183741e-01
-4.62329417e-01 -4.30277176e-02 5.23431480e-01 2.24570170e-01
-1.04970372e+00 3.77988487e-01 -2.46661425e-01 -8.78215492e-01
4.79637831e-01 -2.15725973e-01 -3.41667593e-01 -4.57571223e-02
-7.22905040e-01 1.16804409e+00 5.89065850e-01 -5.35201609e-01
-8.08507383e-01 -5.31185389e-01 -1.17609382e+00 -1.45087570e-01
2.70354956e-01 -7.46181428e-01 1.36771047e+00 -8.69144201e-01
-8.84547770e-01 1.09733582e+00 -2.14732036e-01 -3.70549917e-01
2.36160174e-01 -1.53763741e-01 -2.35142469e-01 5.82643390e-01
4.05638278e-01 9.81816590e-01 6.34647846e-01 -1.49801183e+00
-5.84280610e-01 1.66719720e-01 3.85561436e-01 5.07388592e-01
-2.99745709e-01 3.84161204e-01 -2.63512552e-01 -9.83375371e-01
-4.59193587e-02 -5.94645381e-01 -1.87949657e-01 1.01336557e-02
-3.18838835e-01 -1.39157236e-01 8.39514911e-01 -5.63918769e-01
1.13879633e+00 -1.93094015e+00 1.57922357e-01 -5.32435402e-02
1.62025005e-01 -4.21871752e-01 -2.33081311e-01 7.50451446e-01
-1.02914758e-02 2.95638829e-01 -1.08483665e-01 -1.35301173e-01
-1.05092585e-01 3.28899682e-01 -5.17293990e-01 1.45617202e-01
3.51200849e-01 8.76642883e-01 -1.13008785e+00 -1.04570055e+00
1.81858122e-01 3.39695692e-01 -5.13752580e-01 -2.99346335e-02
-4.26942080e-01 -1.64331123e-02 -4.72994685e-01 2.29830727e-01
5.78554813e-03 -3.48426819e-01 6.15971796e-02 -1.08319975e-01
-1.73384100e-02 6.25419676e-01 -8.41990888e-01 1.60524309e+00
-3.30193967e-01 1.08572066e+00 -6.12204075e-01 -7.06643045e-01
8.23158324e-01 4.16482836e-01 2.84783185e-01 -3.76632065e-01
-3.98147628e-02 -2.41009787e-01 -3.62041354e-01 -7.36858189e-01
9.94167984e-01 -2.18033016e-01 -3.16313684e-01 6.82311296e-01
1.72261372e-01 -6.65841579e-01 4.42033440e-01 7.75948167e-01
1.07396257e+00 4.05735821e-01 6.14690244e-01 -2.77555257e-01
3.08184236e-01 4.31615740e-01 -2.59176344e-02 8.61607909e-01
2.40331352e-01 1.01608455e+00 4.96759832e-01 -3.21499854e-01
-1.32307541e+00 -1.12144315e+00 5.04700467e-03 9.98273313e-01
1.64064065e-01 -8.08880210e-01 -5.78602612e-01 -6.04833007e-01
-3.10601234e-01 1.17776000e+00 -6.65761292e-01 7.66707584e-02
-5.45020401e-01 -3.32465798e-01 1.99162364e-01 7.14980245e-01
-1.55005539e-02 -1.28770161e+00 -9.23703432e-01 2.38760948e-01
-4.86833811e-01 -1.35148990e+00 -4.05751050e-01 -2.16716617e-01
-4.89290863e-01 -9.13287342e-01 -5.31649470e-01 -7.64627516e-01
7.01534271e-01 3.80672067e-01 1.54635060e+00 1.12417497e-01
-5.58943868e-01 8.47198963e-01 -6.82693541e-01 -3.71217728e-01
-4.80475903e-01 -1.10656925e-01 -2.36445040e-01 -1.83685020e-01
2.48132512e-01 -3.28120410e-01 -3.09322000e-01 1.45754263e-01
-9.97924089e-01 5.07894337e-01 1.92552775e-01 6.77746177e-01
3.66242349e-01 -2.62325983e-02 2.70453185e-01 -6.95501506e-01
6.43650889e-01 -4.25499022e-01 -2.87145749e-02 1.90521747e-01
-1.71795070e-01 7.11811008e-04 4.23037820e-02 -4.54224110e-01
-1.02286530e+00 -6.24141619e-02 2.89761394e-01 -1.58746928e-01
-1.56556442e-01 3.79423708e-01 -3.80767509e-02 3.86477232e-01
8.67776275e-01 1.60540193e-01 -3.72208089e-01 -2.65964549e-02
6.69731736e-01 3.38383704e-01 5.88353872e-01 -5.99400401e-01
8.37133408e-01 5.79494953e-01 -1.67934790e-01 -1.07296503e+00
-9.52851415e-01 -6.14267528e-01 -6.79961085e-01 -4.73809540e-01
1.00923193e+00 -9.10072684e-01 -3.10350489e-02 -6.05073273e-02
-1.49574542e+00 1.13083972e-02 -7.05033243e-01 5.50032139e-01
-1.05338073e+00 2.73723364e-01 -5.25654614e-01 -6.92877591e-01
1.93838507e-01 -7.83722639e-01 1.12131631e+00 -1.54540300e-01
-9.95370984e-01 -1.03512573e+00 2.88085714e-02 1.86606601e-01
2.35286474e-01 6.88555539e-01 9.89058793e-01 -5.45808971e-01
-7.43275881e-01 -1.15470596e-01 -3.34944904e-01 -1.54028609e-01
8.62819180e-02 9.92912799e-02 -7.95282900e-01 1.98241383e-01
1.52528530e-03 -4.63378221e-01 9.81896222e-01 2.42363811e-01
8.43667209e-01 -5.25154591e-01 -4.54399168e-01 5.35166897e-02
1.50663280e+00 4.17732634e-02 6.64907873e-01 4.98873055e-01
7.74504602e-01 9.87207472e-01 5.52743733e-01 5.86750507e-01
4.82112378e-01 7.14759707e-01 2.12446257e-01 1.24191888e-01
-3.34869057e-01 -5.73388577e-01 2.91160464e-01 4.83604580e-01
-1.60114709e-02 -5.74476361e-01 -9.80175793e-01 1.11639476e+00
-1.87961805e+00 -1.54341328e+00 -3.60995650e-01 1.62462962e+00
7.52590239e-01 1.78239390e-01 1.20171525e-01 2.55161375e-02
6.25998139e-01 4.95862126e-01 2.30224021e-02 -5.63067019e-01
-4.30245399e-01 2.07228716e-02 3.51301283e-01 3.51813912e-01
-1.08209395e+00 1.04614282e+00 7.96194935e+00 8.45474482e-01
-4.00471658e-01 1.05922610e-01 3.95129830e-01 -2.88691908e-01
-7.20183611e-01 2.19259933e-01 -4.51745391e-01 -1.36553109e-01
6.12788856e-01 -3.57787937e-01 1.39541745e-01 8.65333080e-01
4.32713404e-02 -1.94998413e-01 -1.47799635e+00 9.08525109e-01
7.44874656e-01 -1.74819994e+00 5.46346545e-01 -2.40951078e-03
7.91556120e-01 -4.16669577e-01 -6.66622669e-02 -3.10600009e-02
5.33448756e-01 -1.13834965e+00 1.31501615e+00 6.27103329e-01
5.80062449e-01 -5.77310085e-01 3.81424069e-01 5.46841025e-02
-1.04404175e+00 2.19307229e-01 -2.45749086e-01 -3.62787336e-01
2.99270064e-01 1.67339802e-01 -1.10140228e+00 3.35819572e-01
3.30212414e-01 9.08953011e-01 -8.09416175e-01 7.77728081e-01
-3.68621320e-01 1.89080641e-01 8.13421309e-02 -3.33599180e-01
4.01306301e-01 2.90876716e-01 6.77941024e-01 1.56429374e+00
4.28856820e-01 8.96839499e-02 1.11948848e-02 1.01654685e+00
1.07990719e-01 2.69331604e-01 -1.06760132e+00 -1.84044123e-01
2.46862799e-01 1.12615037e+00 -1.00116467e+00 -6.22455359e-01
-3.67228746e-01 8.78856659e-01 2.34658286e-01 2.55649894e-01
-6.38756573e-01 -1.89184979e-01 4.68177736e-01 4.90316093e-01
5.28210461e-01 -4.28664595e-01 -2.90295094e-01 -1.04793036e+00
-2.64254585e-02 -6.81515932e-01 2.39977077e-01 -1.45396268e+00
-1.27358925e+00 5.78774691e-01 5.02524853e-01 -1.18551278e+00
-5.88321686e-01 -3.85832101e-01 -5.21672964e-01 4.80728030e-01
-1.00485885e+00 -1.43393683e+00 -1.80121034e-01 3.95364106e-01
9.55582798e-01 -3.82307768e-02 8.64352107e-01 -4.11770791e-01
6.07387796e-02 1.05048634e-01 -2.69890785e-01 1.60043463e-01
4.67227072e-01 -1.17045188e+00 4.66558933e-01 7.97754049e-01
7.48013020e-01 4.87105161e-01 1.06469035e+00 -6.63873732e-01
-8.53945494e-01 -8.76040518e-01 1.20299435e+00 -9.54478800e-01
8.37882340e-01 -4.09630299e-01 -5.59614182e-01 7.11318552e-01
5.71468353e-01 -4.44071651e-01 6.74686849e-01 6.29970655e-02
-8.20075631e-01 5.01338065e-01 -8.43959868e-01 9.41962063e-01
1.42770135e+00 -7.01695740e-01 -9.43799853e-01 6.28962874e-01
6.03153110e-01 -2.01613992e-01 -5.45103908e-01 -3.35922725e-02
4.65532929e-01 -9.08611536e-01 1.06409526e+00 -5.76370537e-01
1.14239168e+00 -1.22757271e-01 -3.75649750e-01 -1.04574072e+00
-5.50180078e-01 -2.19496161e-01 6.17665313e-02 1.28936660e+00
3.48515928e-01 3.49592827e-02 4.43188310e-01 5.15880167e-01
-1.60047431e-02 -3.61570567e-01 -5.23764968e-01 -7.72538722e-01
-9.22737718e-02 -4.66945469e-01 4.17254597e-01 9.82868910e-01
2.30424240e-01 6.10396326e-01 -2.09634840e-01 -2.00160503e-01
4.28453505e-01 1.56941220e-01 7.38913894e-01 -1.00677717e+00
-1.05674155e-01 -7.38132954e-01 -8.24116290e-01 -7.78509676e-01
2.56318986e-01 -9.28994417e-01 9.00406539e-02 -2.04634380e+00
6.57524288e-01 -1.80589169e-01 8.22941586e-02 2.44525805e-01
7.35908672e-02 4.66140896e-01 3.41315567e-01 2.46569514e-01
-8.16462219e-01 3.15535277e-01 1.06553471e+00 -2.43683442e-01
5.33993468e-02 -8.08764338e-01 -8.73513401e-01 9.57654834e-01
5.34508049e-01 -5.92148781e-01 -5.20987570e-01 -6.29709482e-01
3.35227728e-01 -1.21523350e-01 6.59744620e-01 -9.75632012e-01
4.00503911e-02 -4.35862511e-01 5.60130775e-01 -6.98452353e-01
5.61264873e-01 -5.27244747e-01 3.45127493e-01 6.19875528e-02
-7.11005688e-01 1.54800996e-01 -4.98838574e-02 5.71643114e-01
-2.19414324e-01 -4.54401106e-01 4.59367096e-01 -4.07769769e-01
-7.72213876e-01 -6.44260570e-02 -6.43873632e-01 1.22087598e-01
1.09751511e+00 -6.20827317e-01 -3.26587558e-01 -7.71708429e-01
-5.84480643e-01 -7.52904266e-02 8.24916661e-01 7.56866217e-01
9.18740094e-01 -1.57871389e+00 -8.90869021e-01 -2.38277778e-01
5.95353365e-01 -2.95054108e-01 2.52298918e-02 3.25520158e-01
-6.13315821e-01 2.33231321e-01 -1.45697519e-01 -6.09029412e-01
-1.29861617e+00 6.78215742e-01 -1.17176976e-04 -1.44856676e-01
-7.47161984e-01 7.71245897e-01 5.22091031e-01 1.30598128e-01
-1.13195539e-01 -1.93427905e-01 -4.58573639e-01 3.34309787e-01
5.45262933e-01 -5.03227040e-02 -4.05253351e-01 -1.26419199e+00
-2.97887206e-01 5.80810368e-01 -1.44922301e-01 -5.49358666e-01
1.42733014e+00 -1.83365479e-01 -2.04881839e-02 8.00640523e-01
1.04469836e+00 3.09479032e-02 -9.60230768e-01 -2.92040557e-01
2.62983680e-01 -6.24785423e-01 -1.30331546e-01 -4.40803319e-01
-4.84490603e-01 6.23495340e-01 7.24277087e-03 2.65462160e-01
7.23359108e-01 4.85384703e-01 4.91990507e-01 2.03410581e-01
3.02985579e-01 -1.01796818e+00 5.07120788e-01 4.22156751e-01
1.30135596e+00 -8.50206017e-01 4.95663255e-01 -4.04983044e-01
-8.84752452e-01 1.17906785e+00 3.28325510e-01 -2.03385994e-01
2.95170695e-01 9.49292332e-02 -1.86760262e-01 -4.70116973e-01
-8.32300127e-01 -3.10031503e-01 3.14823329e-01 7.03969657e-01
4.66587156e-01 -1.21316597e-01 -3.61605734e-01 1.64078236e-01
-4.74756688e-01 -2.40829051e-01 8.41059089e-01 1.11558235e+00
-4.94486630e-01 -8.87402356e-01 -3.36684644e-01 4.45016742e-01
-3.26705873e-01 -6.56638741e-02 -6.60543323e-01 9.26322579e-01
9.53232199e-02 9.81934428e-01 3.81167233e-01 -2.34920055e-01
1.68496475e-01 -1.33471228e-02 7.63743699e-01 -9.85782266e-01
-5.17061710e-01 -1.15268808e-02 6.01935744e-01 -3.90192807e-01
-7.67738760e-01 -9.12564218e-01 -1.07006145e+00 -1.13432494e-03
-1.82641655e-01 -1.19842395e-01 5.53470612e-01 8.89962018e-01
8.61554965e-02 4.67553973e-01 3.20642889e-01 -9.11425591e-01
1.00562364e-01 -7.35076249e-01 -3.12066585e-01 8.24536324e-01
4.79929224e-02 -6.85145319e-01 -1.79552019e-01 5.92107654e-01]
|
[11.001441955566406, 1.0040435791015625]
|
212cf36a-f301-4328-b8c4-f82db2d1402e
|
dense-multi-path-u-net-for-ischemic-stroke
|
1810.07003
| null |
http://arxiv.org/abs/1810.07003v1
|
http://arxiv.org/pdf/1810.07003v1.pdf
|
Dense Multi-path U-Net for Ischemic Stroke Lesion Segmentation in Multiple Image Modalities
|
Delineating infarcted tissue in ischemic stroke lesions is crucial to
determine the extend of damage and optimal treatment for this life-threatening
condition. However, this problem remains challenging due to high variability of
ischemic strokes' location and shape. Recently, fully-convolutional neural
networks (CNN), in particular those based on U-Net, have led to improved
performances for this task. In this work, we propose a novel architecture that
improves standard U-Net based methods in three important ways. First, instead
of combining the available image modalities at the input, each of them is
processed in a different path to better exploit their unique information.
Moreover, the network is densely-connected (i.e., each layer is connected to
all following layers), both within each path and across different paths,
similar to HyperDenseNet. This gives our model the freedom to learn the scale
at which modalities should be processed and combined. Finally, inspired by the
Inception architecture, we improve standard U-Net modules by extending
inception modules with two convolutional blocks with dilated convolutions of
different scale. This helps handling the variability in lesion sizes. We split
the 93 stroke datasets into training and validation sets containing 83 and 9
examples respectively. Our network was trained on a NVidia TITAN XP GPU with 16
GBs RAM, using ADAM as optimizer and a learning rate of 1$\times$10$^{-5}$
during 200 epochs. Training took around 5 hours and segmentation of a whole
volume took between 0.2 and 2 seconds, as average. The performance on the test
set obtained by our method is compared to several baselines, to demonstrate the
effectiveness of our architecture, and to a state-of-art architecture that
employs factorized dilated convolutions, i.e., ERFNet.
|
['Christian Desrosiers', 'Jose Dolz', 'Ismail Ben Ayed']
|
2018-10-16
| null | null | null | null |
['ischemic-stroke-lesion-segmentation']
|
['medical']
|
[ 8.77413005e-02 -1.63278788e-01 2.40013702e-03 -2.09662870e-01
-4.43050265e-01 -6.21780276e-01 2.90580899e-01 3.84442881e-02
-8.69392633e-01 8.45086932e-01 8.39479342e-02 -5.07141948e-01
3.36192027e-02 -1.16429222e+00 -7.13619888e-01 -6.36931300e-01
-2.60317206e-01 1.59462407e-01 7.22363114e-01 3.10934726e-02
8.62928631e-04 8.31052244e-01 -1.08983970e+00 4.46744978e-01
9.82074201e-01 1.00331759e+00 2.14544058e-01 7.57556319e-01
-1.27567902e-01 6.83151066e-01 -4.27446455e-01 -2.73670524e-01
3.47025454e-01 -3.07916313e-01 -8.40002120e-01 -1.90651655e-01
3.15412641e-01 -6.29066586e-01 -7.28788853e-01 7.57429242e-01
7.00627983e-01 2.52696931e-01 6.45229876e-01 -7.82957971e-01
-2.18782276e-01 4.24365908e-01 -5.04629552e-01 7.06709206e-01
-3.91474485e-01 1.29172355e-01 5.25979161e-01 -4.76053447e-01
6.49178445e-01 8.80295455e-01 5.57244539e-01 5.60889721e-01
-9.42629457e-01 -5.50074875e-01 6.63563162e-02 4.05532867e-01
-9.00608838e-01 -2.19149049e-02 3.91367614e-01 -5.96235633e-01
9.42110300e-01 2.29043871e-01 7.58676529e-01 1.07768035e+00
2.72421360e-01 7.86418855e-01 1.09789693e+00 -1.40665948e-01
3.00339758e-01 -2.49261305e-01 4.27453935e-01 4.91652399e-01
3.07228684e-01 7.47848898e-02 3.08097769e-02 -5.08744754e-02
1.05465627e+00 3.60977203e-01 -6.08939409e-01 -2.49816984e-01
-1.06419981e+00 8.46501529e-01 9.26964819e-01 3.73921007e-01
-4.18027937e-01 6.97420686e-02 6.87839687e-01 1.17070945e-02
2.90633947e-01 1.15787640e-01 -4.72150236e-01 -1.62332401e-01
-8.61287832e-01 6.53639510e-02 5.18177927e-01 5.80661833e-01
4.04890925e-01 -1.85009032e-01 -3.41410875e-01 9.35234606e-01
-2.20354870e-01 1.35883361e-01 6.43284976e-01 -6.25297487e-01
5.24136901e-01 7.89251804e-01 -2.49094337e-01 -4.88371551e-01
-8.22903574e-01 -7.82788336e-01 -1.33463049e+00 7.13256001e-01
8.26652288e-01 -7.02839553e-01 -1.26373804e+00 1.54167044e+00
-1.54469842e-02 2.64360338e-01 -1.78579837e-01 9.83234584e-01
7.25915194e-01 2.35424444e-01 1.08620800e-01 2.19631776e-01
1.48675990e+00 -1.04269719e+00 -1.22646928e-01 -1.98606327e-01
7.56844461e-01 -4.86739576e-01 9.48183417e-01 2.03253612e-01
-1.04504371e+00 -3.13359767e-01 -9.59128797e-01 -5.19422032e-02
-5.18520594e-01 2.53178447e-01 5.45666873e-01 6.96863353e-01
-1.07695174e+00 8.86899352e-01 -9.63961184e-01 -3.47687781e-01
1.07962382e+00 2.63545036e-01 -3.37057650e-01 9.57066044e-02
-1.15332639e+00 1.03571475e+00 5.42494893e-01 1.30013496e-01
-5.71982205e-01 -7.86404371e-01 -4.40770179e-01 3.14678907e-01
9.04854462e-02 -7.90803611e-01 8.02679181e-01 -8.14063072e-01
-1.29259980e+00 6.15153432e-01 -9.67131555e-03 -7.70994902e-01
1.01467884e+00 -2.14225084e-01 1.08508598e-02 4.57307488e-01
-2.09114805e-01 6.49078727e-01 4.21056241e-01 -7.82072484e-01
-6.39690816e-01 -4.59576249e-01 2.84824371e-01 1.61068793e-02
-3.24180305e-01 1.21234946e-01 -4.41355258e-01 -7.73328125e-01
1.29070222e-01 -7.80769467e-01 -3.58858198e-01 1.73148230e-01
-1.71852231e-01 1.06354952e-01 8.42008471e-01 -7.31275916e-01
1.03229499e+00 -1.79844749e+00 1.03609830e-01 2.38616616e-01
4.48569566e-01 5.70829272e-01 1.05807118e-01 -3.15105058e-02
-1.52576014e-01 1.23359617e-02 -5.95855772e-01 -1.65582195e-01
-4.61264759e-01 2.78505534e-01 -7.27480352e-02 4.09390897e-01
-9.59139764e-02 1.09064901e+00 -6.79763436e-01 -3.84371191e-01
3.25778872e-01 6.71594381e-01 -4.61109817e-01 2.70309243e-02
2.05023676e-01 5.70103824e-01 -3.96177620e-01 2.76689917e-01
8.12123835e-01 -8.12958777e-02 -8.26582685e-02 -9.76307690e-02
-2.05316573e-01 -1.85580045e-01 -1.03592169e+00 1.56213629e+00
-4.07817006e-01 6.29732490e-01 -1.03686795e-01 -1.34173369e+00
7.49735475e-01 3.44083339e-01 5.79044521e-01 -5.67630649e-01
5.11503220e-01 2.80879766e-01 1.93759903e-01 -4.15722400e-01
-6.42057210e-02 6.53700009e-02 3.24552298e-01 5.66300511e-01
-5.43344580e-02 3.83727014e-01 4.26325262e-01 2.44204044e-01
1.27280128e+00 -7.50109032e-02 -2.86072306e-02 -2.30286643e-01
3.06330383e-01 -5.32392636e-02 4.46445584e-01 1.00649595e+00
-4.53956455e-01 8.92852962e-01 8.50020289e-01 -7.19247043e-01
-9.24150527e-01 -1.05154443e+00 -4.51003730e-01 6.37475431e-01
-6.70980290e-03 -1.95320658e-02 -1.10501826e+00 -7.84909248e-01
-2.47760639e-01 3.18708062e-01 -8.94171953e-01 4.28623222e-02
-9.69306827e-01 -1.13216150e+00 6.50741100e-01 1.13950634e+00
9.44035351e-01 -1.36524725e+00 -1.37329459e+00 2.73636669e-01
-5.92024364e-02 -9.56773877e-01 -2.04667643e-01 2.82824099e-01
-1.22711837e+00 -1.41908062e+00 -1.19912672e+00 -7.12795496e-01
7.31298804e-01 2.19376445e-01 9.73833799e-01 2.17660233e-01
-4.66897249e-01 -1.31406292e-01 -4.82770443e-01 -2.87408859e-01
8.69036242e-02 3.68922234e-01 -4.89404052e-01 -6.31680712e-02
2.45405231e-02 -8.02849770e-01 -1.14996159e+00 1.31067961e-01
-1.03671801e+00 1.33265287e-01 8.00624251e-01 8.32889855e-01
4.61190909e-01 -2.51372278e-01 5.69722950e-01 -9.51781392e-01
5.35092235e-01 -4.75253761e-01 -3.66641700e-01 3.41010660e-01
-1.70850158e-01 -1.94712937e-01 7.66714752e-01 -4.06319320e-01
-8.02080810e-01 8.06745216e-02 -5.72658889e-02 -3.01083624e-01
-2.99412251e-01 3.20722908e-01 6.97974414e-02 -1.77525461e-01
9.08250451e-01 1.17121823e-01 9.29150283e-02 -4.98860180e-01
3.33259255e-01 5.48617065e-01 6.34435594e-01 -4.57154661e-01
4.11917478e-01 6.11069262e-01 9.20702424e-03 -6.17481709e-01
-4.99163657e-01 -2.79427946e-01 -9.31694567e-01 -2.12283701e-01
1.03035021e+00 -4.12773579e-01 -4.59771633e-01 8.49422753e-01
-1.22454953e+00 -6.62952602e-01 -9.30100530e-02 6.22238398e-01
-2.96091169e-01 2.79421031e-01 -7.51051664e-01 -1.50622994e-01
-7.38265336e-01 -1.37940991e+00 4.10522223e-01 4.87949699e-01
1.04633868e-01 -9.10906911e-01 -1.46213815e-01 1.93806335e-01
7.56606817e-01 6.46763742e-01 9.73595679e-01 -4.41225708e-01
-2.84664303e-01 -2.98380584e-01 -9.31618452e-01 4.31052923e-01
7.30683878e-02 -3.22350383e-01 -7.41515636e-01 -3.31708878e-01
-2.07550481e-01 -8.55255052e-02 1.14662957e+00 7.02228427e-01
1.36661625e+00 1.42135113e-01 -4.02850538e-01 8.36518764e-01
1.34246004e+00 2.17759386e-01 9.24444914e-01 6.24279618e-01
5.47690213e-01 2.36851737e-01 -1.63173288e-01 2.37438917e-01
1.36631116e-01 4.33911413e-01 5.48964620e-01 -5.67056417e-01
-4.49253201e-01 5.00461042e-01 -1.69623494e-01 2.39577591e-02
-3.65433961e-01 7.19714817e-03 -8.58049214e-01 4.14507061e-01
-1.88839734e+00 -8.73441339e-01 -2.72035956e-01 2.20039916e+00
7.21879840e-01 2.74848551e-01 2.14557052e-01 -2.62752734e-02
6.64232552e-01 2.85222009e-03 -6.60618186e-01 -3.44444305e-01
-1.97004303e-01 6.45004451e-01 6.37714803e-01 3.50629658e-01
-1.29487407e+00 8.12761426e-01 5.35202074e+00 6.47969127e-01
-1.45331395e+00 2.25435361e-01 9.05697346e-01 -2.93413192e-01
2.47432262e-01 -2.65277885e-02 -3.17809075e-01 5.62809169e-01
4.59889501e-01 8.55958834e-02 4.19200838e-01 6.04072690e-01
1.88759997e-01 -2.73725152e-01 -7.13551283e-01 6.16737127e-01
-1.95802823e-01 -1.49015176e+00 -1.51101455e-01 -1.06213942e-01
6.42162681e-01 4.35690343e-01 -4.28827614e-01 2.28396311e-01
8.03024881e-03 -1.10019648e+00 5.24105906e-01 3.82369906e-01
6.79202557e-01 -7.67237902e-01 9.68920887e-01 3.23503315e-01
-1.15166104e+00 -3.41861211e-02 -4.24565047e-01 -5.79968328e-03
2.73822188e-01 5.10300875e-01 -4.08813298e-01 4.41328198e-01
1.03097546e+00 4.12433892e-01 -4.69211102e-01 1.40287614e+00
-2.78908372e-01 5.48727572e-01 -3.65428001e-01 2.42544845e-01
3.82107258e-01 -2.51818091e-01 3.95827204e-01 1.39734483e+00
2.00301021e-01 3.92771631e-01 3.09297014e-02 7.18214750e-01
-9.15372595e-02 1.34311616e-01 -2.66341209e-01 7.74805367e-01
1.52519345e-01 1.26835775e+00 -1.06873155e+00 -5.89271426e-01
-4.33475405e-01 8.77922654e-01 5.27648449e-01 4.80655581e-01
-8.34664345e-01 -7.97243237e-01 3.06035668e-01 2.18558580e-01
3.40153217e-01 -2.19627604e-01 -7.78675675e-01 -1.09392202e+00
1.78354070e-01 -4.45996940e-01 5.58826983e-01 -4.43213820e-01
-9.50286806e-01 9.68510032e-01 -1.20392241e-01 -1.02546501e+00
1.95531368e-01 -7.16467977e-01 -1.11568439e+00 1.20786560e+00
-1.68748844e+00 -9.15342093e-01 -5.97339392e-01 7.44680822e-01
3.11124235e-01 -6.11484936e-03 7.70750463e-01 4.60748971e-01
-7.48362243e-01 5.28000236e-01 2.87287366e-02 6.03152812e-01
5.58329821e-01 -1.16703963e+00 3.64873648e-01 1.05475569e+00
-2.64039516e-01 4.94624645e-01 1.28952250e-01 -5.14187217e-01
-7.43875444e-01 -1.23307002e+00 4.70192343e-01 -2.13030353e-02
6.60198808e-01 -8.00078213e-02 -1.08951616e+00 5.15661716e-01
1.12237506e-01 5.14307082e-01 3.81663203e-01 -2.05094367e-01
-1.85322285e-01 1.32412240e-01 -1.19468689e+00 6.01837397e-01
1.05418563e+00 -7.20591024e-02 -4.72306967e-01 2.42234483e-01
1.48987338e-01 -6.94330096e-01 -7.76901782e-01 4.45707798e-01
6.26773477e-01 -9.92772281e-01 9.58355486e-01 -7.36466944e-01
5.49879909e-01 -4.75202687e-02 2.65924960e-01 -1.36722291e+00
-2.86794752e-01 -1.55724213e-01 -1.50081543e-02 5.09051323e-01
3.84678751e-01 -1.01821399e+00 9.29616690e-01 5.71691871e-01
-4.63726968e-01 -1.34288526e+00 -9.74241972e-01 -5.47959983e-01
6.54154122e-01 -2.70113140e-01 4.17176455e-01 5.37096679e-01
-1.48822039e-01 -1.14078417e-01 -7.78851733e-02 -2.95664594e-02
5.71152508e-01 1.13221794e-01 3.62830997e-01 -1.18478608e+00
-2.14843243e-01 -8.84504616e-01 -2.98292816e-01 -8.12817991e-01
-1.05112135e-01 -1.04940963e+00 -3.61486346e-01 -1.70546961e+00
2.73100704e-01 -6.48367941e-01 -4.82789934e-01 8.27853799e-01
-3.51429045e-01 5.11948109e-01 2.47870460e-01 1.85693979e-01
8.91851783e-02 2.31320024e-01 1.41779351e+00 -8.86677131e-02
-3.92798632e-01 9.27666947e-02 -4.70416546e-01 7.53400326e-01
1.16017377e+00 -2.16412559e-01 -3.30076694e-01 -7.44419813e-01
-3.30671698e-01 -1.11225515e-03 6.72997236e-01 -1.12915468e+00
1.82115033e-01 1.94391161e-01 6.21864855e-01 -4.28855956e-01
3.76350917e-02 -5.78054070e-01 -2.82070994e-01 7.57655323e-01
-1.19276159e-01 -7.96918795e-02 3.75163347e-01 4.56715971e-02
2.25183275e-03 -2.98214942e-01 1.02649748e+00 -2.15937853e-01
-4.54635292e-01 6.19164586e-01 -3.85900170e-01 3.08235474e-02
1.11553121e+00 -3.00557524e-01 -5.57695091e-01 1.63318366e-02
-9.11637902e-01 1.94871515e-01 1.41043484e-01 2.53628284e-01
5.59614778e-01 -9.96063590e-01 -8.13734055e-01 5.35028838e-02
-3.56239557e-01 1.45215288e-01 6.20601177e-01 1.18456233e+00
-9.30815876e-01 2.72022277e-01 -6.10097945e-01 -4.71946359e-01
-1.17468131e+00 4.03643623e-02 6.44842327e-01 -4.88480628e-01
-1.31572926e+00 6.66554868e-01 2.16676831e-01 -3.20635810e-02
2.25608140e-01 -3.65136057e-01 -5.03598094e-01 6.26107380e-02
6.92305863e-01 4.87968713e-01 3.09148908e-01 -3.50118220e-01
-3.21163028e-01 4.86749977e-01 -2.24896014e-01 -4.10519913e-02
1.36747563e+00 4.35611844e-01 -1.56076595e-01 -3.26768339e-01
1.30742788e+00 -5.98237157e-01 -1.41424406e+00 -1.32692441e-01
-2.86896765e-01 -2.44280800e-01 3.86944741e-01 -1.01463485e+00
-1.70911527e+00 1.12177062e+00 9.38454628e-01 -9.60462317e-02
1.11741388e+00 -2.77066916e-01 1.12385976e+00 9.96681973e-02
2.03887478e-01 -8.34062278e-01 -2.47878283e-01 4.96712536e-01
8.42028975e-01 -1.05130196e+00 -2.09628776e-01 -1.69248268e-01
-3.94169301e-01 1.45299411e+00 6.10344350e-01 -3.91632110e-01
7.39003122e-01 2.71818519e-01 2.09213004e-01 -1.05211347e-01
-1.08843751e-01 -1.27785206e-01 1.34274811e-01 4.12101418e-01
2.98369527e-01 1.73552051e-01 -5.70789754e-01 5.09592116e-01
-3.55077498e-02 2.84106433e-01 5.05132496e-01 1.09409499e+00
-4.60455984e-01 -1.15971065e+00 -1.89475492e-01 7.83671618e-01
-4.76285905e-01 -6.73532113e-02 -6.00737371e-02 6.81206346e-01
3.12714219e-01 6.19244397e-01 1.96659360e-02 -3.33423652e-02
4.58944917e-01 -8.99171680e-02 4.92873192e-01 -2.53578514e-01
-8.56456339e-01 -1.87674031e-01 -3.37153703e-01 -5.37842453e-01
-6.98888525e-02 -6.14660084e-01 -1.51964581e+00 -1.51045233e-01
2.90411618e-02 -5.45838960e-02 6.00647509e-01 9.87888277e-01
3.45147610e-01 8.58847439e-01 3.70975971e-01 -9.10262108e-01
-3.73660058e-01 -1.07171559e+00 -3.05668801e-01 3.00721645e-01
-1.56275481e-02 -6.74874723e-01 -1.68381199e-01 -1.36116117e-01]
|
[14.360928535461426, -2.071216583251953]
|
888e010e-24ef-42d3-b55f-709aa1c1a6d4
|
three-dependency-and-boundary-models-for
| null | null |
https://aclanthology.org/D12-1063
|
https://aclanthology.org/D12-1063.pdf
|
Three Dependency-and-Boundary Models for Grammar Induction
| null |
['Valentin I. Spitkovsky', 'Hiyan Alshawi', 'Daniel Jurafsky']
|
2012-07-01
| null | null | null |
emnlp-2012-7
|
['dependency-grammar-induction']
|
['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]
|
[-7.310418605804443, 3.7660129070281982]
|
e330bd29-f5a4-4c06-83ed-bee428ee5136
|
bootstrap-your-flow
|
2111.1151
| null |
https://arxiv.org/abs/2111.11510v4
|
https://arxiv.org/pdf/2111.11510v4.pdf
|
Bootstrap Your Flow
|
Normalizing flows are flexible, parameterized distributions that can be used to approximate expectations from intractable distributions via importance sampling. However, current flow-based approaches are limited on challenging targets where they either suffer from mode seeking behaviour or high variance in the training loss, or rely on samples from the target distribution, which may not be available. To address these challenges, we combine flows with annealed importance sampling (AIS), while using the $\alpha$-divergence as our objective, in a novel training procedure, FAB (Flow AIS Bootstrap). Thereby, the flow and AIS improve each other in a bootstrapping manner. We demonstrate that FAB can be used to produce accurate approximations to complex target distributions, including Boltzmann distributions, in problems where previous flow-based methods fail.
|
['José Miguel Hernández-Lobato', 'Gregor N. C. Simm', 'Vincent Stimper', 'Laurence Illing Midgley']
|
2021-11-22
| null |
https://openreview.net/forum?id=Rzwf6LeM-6E
|
https://openreview.net/pdf?id=Rzwf6LeM-6E
|
pproximateinference-aabi-symposium-2022-2
|
['normalising-flows']
|
['methodology']
|
[ 1.33314684e-01 -3.20350647e-01 -1.46942884e-01 -3.86310309e-01
-1.10044229e+00 -6.63807690e-01 4.78269428e-01 5.42726144e-02
-5.23683071e-01 1.48607159e+00 -6.15741387e-02 -4.12026703e-01
-1.18600771e-01 -7.69875348e-01 -6.90145433e-01 -6.92075431e-01
2.01249734e-01 7.94062734e-01 2.87429214e-01 3.24532986e-02
3.92617583e-01 6.46565795e-01 -1.32538927e+00 -1.66155353e-01
1.37250590e+00 7.76733398e-01 -2.65507132e-01 7.08657265e-01
-4.25609916e-01 7.45796621e-01 -6.75758719e-01 -5.97273171e-01
6.17437474e-02 -5.74712515e-01 -7.44601071e-01 -4.71062452e-01
4.10996556e-01 -5.42350173e-01 -8.21123347e-02 1.14166558e+00
3.65330815e-01 7.37949252e-01 1.16579950e+00 -1.22491014e+00
-3.39337230e-01 4.93133247e-01 -5.12854338e-01 3.91868055e-01
-6.36969283e-02 2.55370080e-01 9.10020411e-01 -7.22841024e-01
2.85184771e-01 1.32744861e+00 7.50505745e-01 7.07431495e-01
-1.63423920e+00 -6.56090736e-01 2.24703565e-01 1.41662536e-02
-1.25632012e+00 -5.12398243e-01 7.20462203e-01 -3.37566912e-01
7.10027635e-01 9.82959121e-02 5.85857928e-01 1.31680262e+00
-3.89142446e-02 8.99707079e-01 8.58051360e-01 -1.60010621e-01
7.79714406e-01 3.34604084e-01 3.97860892e-02 4.73066241e-01
1.74038753e-01 1.17161043e-01 -4.27224964e-01 -6.48362100e-01
6.67648733e-01 -2.17248902e-01 -3.97545874e-01 -3.98192644e-01
-5.46849966e-01 1.16109240e+00 3.66325349e-01 -1.39753103e-01
-3.01889360e-01 3.51380974e-01 2.51196504e-01 1.14108538e-02
7.56275713e-01 4.04459625e-01 -2.37646461e-01 -6.00180268e-01
-1.20980132e+00 8.43227029e-01 9.80388761e-01 6.92021489e-01
8.85420442e-01 1.66988850e-01 -3.63671750e-01 7.91762829e-01
3.07193726e-01 5.08461475e-01 -9.40807816e-03 -1.16603529e+00
4.12569523e-01 3.23087424e-02 5.40865064e-01 -3.24217409e-01
1.87597554e-02 -3.67708951e-01 -5.01821399e-01 4.22030687e-01
9.62257862e-01 -2.68740386e-01 -1.02260828e+00 1.97391009e+00
5.54966807e-01 2.18578011e-01 -2.33798951e-01 7.41118491e-01
-4.97760102e-02 7.27552950e-01 2.39815786e-01 -9.16080102e-02
7.80766129e-01 -9.15180087e-01 -3.71417016e-01 -2.18567237e-01
2.71837682e-01 -5.88810146e-01 1.33871055e+00 5.14788747e-01
-1.37510777e+00 -2.28012189e-01 -8.63026321e-01 1.45681739e-01
7.59333521e-02 -5.21551013e-01 5.34743786e-01 8.68547440e-01
-1.00804794e+00 1.06612635e+00 -9.64653492e-01 6.95067793e-02
8.70065749e-01 2.31229246e-01 3.17510307e-01 -1.59428850e-01
-9.56454396e-01 7.35599935e-01 1.62102863e-01 1.29727572e-02
-8.96514714e-01 -1.15446746e+00 -6.70022488e-01 2.38047898e-01
1.96804762e-01 -8.91720533e-01 1.26898956e+00 -8.61361265e-01
-1.81521869e+00 -1.14216870e-02 -2.77003556e-01 -4.90442455e-01
9.37159121e-01 -2.92481691e-01 2.52116948e-01 1.52113229e-01
-1.42755523e-01 6.01096749e-01 1.11698043e+00 -1.05501139e+00
-3.22338015e-01 -9.78403538e-02 2.32767239e-02 1.02151968e-01
-2.62769401e-01 -3.97459924e-01 -1.17921017e-01 -5.11869133e-01
-5.50463080e-01 -6.99103355e-01 -4.00101364e-01 9.79622081e-02
-2.87806392e-01 -6.91365376e-02 5.43396533e-01 -2.66652465e-01
1.09684908e+00 -1.94843483e+00 1.06715061e-01 4.68664229e-01
1.63945660e-01 4.33494031e-01 3.88845429e-02 2.80899346e-01
4.91113901e-01 3.60338628e-01 -6.03001297e-01 -3.83566946e-01
1.40924931e-01 3.24857920e-01 -4.12732214e-01 3.20652157e-01
2.63945311e-01 7.53255665e-01 -1.14441395e+00 -6.24371946e-01
1.60298243e-01 5.94705164e-01 -1.14572752e+00 3.22954863e-01
-4.11175609e-01 5.77861309e-01 -5.16997397e-01 3.22971076e-01
8.51379514e-01 -2.22879395e-01 -1.34241842e-02 7.95671064e-03
2.37909079e-01 2.75196671e-01 -1.00635898e+00 1.29293990e+00
-5.47028601e-01 2.75656253e-01 2.04575315e-01 -1.02882636e+00
7.08742082e-01 -1.22852713e-01 2.45154560e-01 -6.59360439e-02
1.44411489e-01 2.84685552e-01 -4.99587059e-02 -2.34068520e-02
4.59668934e-01 -7.54453540e-01 2.96014637e-01 4.73115712e-01
1.21742748e-01 -5.27533293e-01 3.30212802e-01 7.29715675e-02
1.10805619e+00 3.67161483e-01 -1.08201385e-01 -2.64929622e-01
3.72698754e-01 -1.01683907e-01 5.19001722e-01 1.21926486e+00
-3.33099157e-01 7.22513855e-01 7.12119162e-01 -6.86266422e-02
-1.21371937e+00 -1.71047270e+00 -3.15711886e-01 7.82914817e-01
1.40575990e-02 -6.13634586e-02 -8.63033891e-01 -1.00206363e+00
2.07743853e-01 9.96212542e-01 -4.34693336e-01 -2.63007432e-01
-5.50895333e-01 -9.51929033e-01 4.61095124e-01 6.06376886e-01
1.63821489e-01 -8.88308764e-01 -3.57528389e-01 5.35350442e-01
-6.81063682e-02 -6.70588195e-01 -6.07311785e-01 2.19982624e-01
-9.13700283e-01 -8.60425949e-01 -9.96916711e-01 -1.22145697e-01
6.61330581e-01 -3.21398586e-01 1.30016816e+00 -1.25769272e-01
-9.75906923e-02 1.94056243e-01 2.58263340e-03 -2.51060963e-01
-4.35774207e-01 3.33783925e-01 -6.47066310e-02 -2.30722368e-01
1.56658366e-01 -9.46634114e-01 -7.66986012e-01 1.94857746e-01
-9.04695809e-01 -4.31487143e-01 2.16938049e-01 1.21868825e+00
2.71889001e-01 -1.65287763e-01 8.95485878e-01 -9.46161389e-01
9.18368578e-01 -6.12940907e-01 -7.28973508e-01 5.69887273e-02
-5.93328774e-01 4.88764554e-01 9.74770963e-01 -7.42671967e-01
-1.19457006e+00 -3.02118748e-01 -4.91576165e-01 -7.14209020e-01
8.00801441e-02 1.59084514e-01 2.27638986e-02 -9.66341197e-02
8.04718733e-01 -2.43051648e-02 1.44628793e-01 -3.70395541e-01
3.29312503e-01 5.49487770e-01 4.52526718e-01 -1.19552779e+00
6.84215248e-01 3.54538113e-01 2.44874656e-02 -4.92890894e-01
-1.18195033e+00 -1.87207267e-01 -2.14710981e-01 -1.66837275e-01
4.01595235e-01 -4.81286228e-01 -7.18110025e-01 3.91907632e-01
-9.46061075e-01 -5.56180179e-01 -6.52250886e-01 5.75619400e-01
-6.47914886e-01 3.57106239e-01 -6.70978129e-01 -1.33039510e+00
-2.35434487e-01 -1.11947608e+00 9.40691769e-01 6.00412488e-01
-3.90014559e-01 -1.16037953e+00 2.19125599e-02 1.51587218e-01
8.31403255e-01 4.76291366e-02 1.04641569e+00 -5.88951349e-01
-4.72537965e-01 -1.21326320e-01 -1.50955305e-01 5.59107184e-01
3.92303951e-02 2.04668850e-01 -9.78070736e-01 -2.30557740e-01
-1.50616571e-01 -6.77043676e-01 9.41001832e-01 5.80641925e-01
1.26770818e+00 -2.27240711e-01 -1.99746147e-01 5.73362648e-01
1.18866861e+00 3.92943695e-02 4.24886614e-01 -4.07143496e-02
3.86267006e-01 3.46842885e-01 2.94232786e-01 6.23966515e-01
7.10939318e-02 3.44629079e-01 1.45811528e-01 3.77930254e-01
2.83674687e-01 -5.98738372e-01 2.63618529e-01 3.37557256e-01
2.43332654e-01 -4.39826369e-01 -7.60020316e-01 5.62326550e-01
-1.82534945e+00 -1.00709867e+00 2.47917280e-01 2.37375021e+00
1.06295991e+00 5.24668992e-01 3.58597875e-01 -1.17850132e-01
6.13516212e-01 1.10452354e-01 -1.16668963e+00 -3.83981854e-01
3.67397755e-01 5.54690123e-01 3.92174900e-01 7.85091400e-01
-7.48913705e-01 7.64223576e-01 7.14241362e+00 1.11159313e+00
-9.43167984e-01 8.12983438e-02 8.49883258e-01 -3.12100649e-01
-8.97660196e-01 1.33711010e-01 -7.82493234e-01 7.38332689e-01
1.00844419e+00 -1.97858289e-01 7.39609838e-01 7.38101363e-01
-7.40894228e-02 -2.79631972e-01 -1.07696402e+00 7.55333722e-01
-2.45902270e-01 -1.22219884e+00 9.90198851e-02 -1.04354747e-01
5.38339972e-01 -1.51934057e-01 1.01422511e-01 5.63043475e-01
7.89223492e-01 -1.09409773e+00 6.21800303e-01 5.37267327e-01
5.34764111e-01 -9.84967172e-01 5.27695179e-01 5.03868699e-01
-8.90633285e-01 7.36976415e-02 -4.48425233e-01 7.88379833e-02
6.55939162e-01 1.02055085e+00 -6.71692073e-01 2.04924300e-01
4.74850625e-01 1.38429955e-01 1.76057339e-01 1.15166664e+00
-7.29041472e-02 9.45962965e-01 -8.18825901e-01 -4.68925416e-01
2.53539950e-01 -4.57291305e-01 7.13921964e-01 8.95372570e-01
3.11592996e-01 -3.47022623e-01 8.50450620e-02 1.51192582e+00
-1.50051624e-01 -1.08106636e-01 -3.36527616e-01 -7.23361224e-02
7.08991826e-01 1.13698804e+00 -6.72484219e-01 -3.90616030e-01
-1.09341070e-02 7.19646871e-01 7.32923150e-01 5.59852362e-01
-1.10004115e+00 -5.74969530e-01 9.89292502e-01 8.12078342e-02
4.38126564e-01 -1.69756144e-01 -1.61401615e-01 -1.25664079e+00
2.55329302e-03 -5.59560895e-01 2.41954029e-01 -4.37386006e-01
-1.70943081e+00 4.68627423e-01 3.56373101e-01 -7.97660410e-01
-4.12394941e-01 -5.01567841e-01 -7.51352966e-01 1.01407158e+00
-1.47881091e+00 -5.42469203e-01 1.29717141e-01 3.10768336e-01
3.74111146e-01 2.37886161e-01 4.26155627e-01 2.73907214e-01
-5.85097194e-01 7.31720448e-01 3.17501336e-01 -3.39019626e-01
4.89147902e-01 -1.44820797e+00 3.21515411e-01 5.23954332e-01
-6.66470751e-02 5.53537905e-01 7.75868416e-01 -5.40425301e-01
-9.15617824e-01 -9.38610971e-01 2.22569138e-01 -3.90043736e-01
6.24295712e-01 -4.50800270e-01 -1.05793023e+00 2.98162729e-01
-2.72116065e-01 2.88698822e-01 6.33311450e-01 2.29036063e-01
-2.74876446e-01 -2.11219378e-02 -1.59812403e+00 6.76199913e-01
9.83698428e-01 -2.87811190e-01 -2.44188875e-01 4.56241369e-02
3.93630654e-01 -2.64659345e-01 -7.27333128e-01 1.35809660e-01
5.26424825e-01 -9.62010503e-01 9.62729871e-01 -8.37913871e-01
2.48273388e-01 -1.83771625e-01 -1.79997494e-03 -1.67101836e+00
-3.92474495e-02 -7.59213328e-01 -3.05901378e-01 1.37920785e+00
4.77416545e-01 -8.20229530e-01 1.03012085e+00 8.11960638e-01
1.50254115e-01 -7.90090442e-01 -1.18841469e+00 -8.66710007e-01
5.96735716e-01 -6.19032085e-01 7.54114211e-01 3.07906151e-01
-1.34589732e-01 2.66324431e-01 -3.13266486e-01 -2.16137305e-01
9.24214065e-01 -7.87081271e-02 6.82936132e-01 -1.13994586e+00
-4.83158380e-01 -6.06413424e-01 4.47427221e-02 -1.32254982e+00
4.34819430e-01 -5.75808942e-01 3.39047194e-01 -1.14287996e+00
5.88608049e-02 -7.45323181e-01 -1.77972063e-01 1.59002468e-01
-5.53165734e-01 2.77504977e-02 1.00087523e-01 6.17086980e-03
-3.66059244e-01 1.18367577e+00 9.92739618e-01 -7.20600337e-02
-1.19854271e-01 1.71442911e-01 -6.44385219e-01 7.08052814e-01
7.20640123e-01 -6.31730199e-01 -6.74919367e-01 -2.04176530e-02
7.40537867e-02 5.57965934e-02 1.45742849e-01 -9.80268478e-01
-4.27439399e-02 -3.91437322e-01 4.42000210e-01 -3.02938163e-01
3.39020371e-01 -4.10861284e-01 -9.30473879e-02 2.49294505e-01
-3.14447999e-01 -2.80757666e-01 1.97684228e-01 6.67377770e-01
-1.12671100e-01 -6.19734526e-01 1.01145208e+00 -1.50934443e-01
1.21416800e-01 4.66424495e-01 -4.64955539e-01 7.14871049e-01
7.09160924e-01 8.40101019e-02 -2.36477420e-01 -5.59304655e-01
-3.00793082e-01 3.24736923e-01 4.86898184e-01 -2.11898252e-01
4.58263665e-01 -1.28953183e+00 -5.47670960e-01 3.94319259e-02
-3.79314989e-01 4.49325472e-01 1.68501556e-01 7.12472975e-01
-5.42044461e-01 -2.39026487e-01 8.43800381e-02 -5.77997208e-01
-4.78639305e-01 3.18993568e-01 5.97840488e-01 -5.12124658e-01
-3.83599937e-01 9.50717092e-01 1.56967536e-01 -6.74231648e-01
1.18671529e-01 -9.98305008e-02 2.63179839e-01 -1.76975429e-01
6.27304256e-01 4.60346490e-01 -1.80523291e-01 8.38491842e-02
-2.40396693e-01 2.46542245e-01 -1.77266568e-01 -5.21898150e-01
1.15446115e+00 -9.87748429e-02 3.34708989e-01 4.16476071e-01
1.09441638e+00 5.82332499e-02 -1.85658956e+00 -7.15339882e-03
-1.94315597e-01 -7.52908885e-01 -5.59042841e-02 -5.87996721e-01
-9.09336269e-01 9.86707330e-01 3.48048538e-01 1.14623964e-01
8.36821914e-01 -2.73505569e-01 9.71052468e-01 1.94956854e-01
2.56906450e-01 -9.62434053e-01 -9.61005390e-02 5.11507392e-01
3.99890810e-01 -9.96828973e-01 -7.53538311e-02 -1.09742343e-01
-4.93180931e-01 1.08820283e+00 7.66693711e-01 -1.23339623e-01
7.44712889e-01 5.51620781e-01 -5.58003373e-02 4.39747930e-01
-7.65688062e-01 -1.03526264e-01 2.06509456e-02 7.47876823e-01
2.33903378e-01 -3.83076042e-01 -8.54780301e-02 2.47204900e-01
-1.34600833e-01 2.11007655e-01 3.54578227e-01 9.32766795e-01
-3.30220252e-01 -1.29630911e+00 -1.87695056e-01 7.88870215e-01
-5.76867223e-01 -2.56206542e-01 9.35258940e-02 4.72617447e-01
-4.10124838e-01 6.52919650e-01 2.41882578e-01 -5.53994533e-03
1.37047231e-01 2.42249236e-01 5.95550299e-01 -2.00793594e-01
-3.23291302e-01 -1.64961040e-01 8.91179070e-02 -2.98497319e-01
-1.41005293e-01 -7.16851354e-01 -9.34541583e-01 -6.63043082e-01
-3.45700622e-01 6.20840013e-01 2.49137282e-01 1.01234663e+00
1.27746433e-01 2.33714178e-01 6.17689967e-01 -1.16376352e+00
-1.12870955e+00 -8.34458470e-01 -5.09238183e-01 4.00048941e-01
4.45507169e-01 -9.41801965e-01 -7.96937466e-01 -4.14168924e-01]
|
[7.022350311279297, 3.916487455368042]
|
72d34a86-f4a0-489d-ad83-72f12aafcb4c
|
middle-level-fusion-for-lightweight-rgb-d
|
2104.11543
| null |
https://arxiv.org/abs/2104.11543v3
|
https://arxiv.org/pdf/2104.11543v3.pdf
|
Middle-level Fusion for Lightweight RGB-D Salient Object Detection
|
Most existing lightweight RGB-D salient object detection (SOD) models are based on two-stream structure or single-stream structure. The former one first uses two sub-networks to extract unimodal features from RGB and depth images, respectively, and then fuses them for SOD. While, the latter one directly extracts multi-modal features from the input RGB-D images and then focuses on exploiting cross-level complementary information. However, two-stream structure based models inevitably require more parameters and single-stream structure based ones cannot well exploit the cross-modal complementary information since they ignore the modality difference. To address these issues, we propose to employ the middle-level fusion structure for designing lightweight RGB-D SOD model in this paper, which first employs two sub-networks to extract low- and middle-level unimodal features, respectively, and then fuses those extracted middle-level unimodal features for extracting corresponding high-level multi-modal features in the subsequent sub-network. Different from existing models, this structure can effectively exploit the cross-modal complementary information and significantly reduce the network's parameters, simultaneously. Therefore, a novel lightweight SOD model is designed, which contains a information-aware multi-modal feature fusion (IMFF) module for effectively capturing the cross-modal complementary information and a lightweight feature-level and decision-level feature fusion (LFDF) module for aggregating the feature-level and the decision-level saliency information in different stages with less parameters. Our proposed model has only 3.9M parameters and runs at 33 FPS. The experimental results on several benchmark datasets verify the effectiveness and superiority of the proposed method over some state-of-the-art methods.
|
['Jungong Han', 'Qiang Zhang', 'Nianchang Huang']
|
2021-04-23
| null | null | null | null |
['rgb-d-salient-object-detection']
|
['computer-vision']
|
[ 2.23644361e-01 -1.54094398e-01 -1.64828405e-01 -1.19378209e-01
-6.74125969e-01 -3.83710600e-02 3.11868966e-01 1.62860155e-01
-4.94203985e-01 1.94205463e-01 1.20905705e-01 1.84196942e-02
-2.42346585e-01 -8.30371320e-01 -4.10233736e-01 -9.69201684e-01
1.14022240e-01 -4.26216006e-01 1.00644529e+00 -2.67877758e-01
2.17998967e-01 3.84284407e-01 -2.20375156e+00 3.62437695e-01
8.76648128e-01 1.62110054e+00 3.83238494e-01 3.32907259e-01
-2.51827300e-01 6.52064204e-01 -2.83676088e-02 1.12267464e-01
3.35232526e-01 -2.68452793e-01 -3.52575481e-01 1.22312501e-01
9.21177268e-02 -5.37509024e-01 -4.11331445e-01 1.10462153e+00
7.34407544e-01 -1.91381667e-02 2.46260196e-01 -1.41410279e+00
-3.52234364e-01 8.18431228e-02 -9.29369807e-01 3.68209034e-01
3.87543619e-01 4.10883665e-01 6.71029627e-01 -1.03302550e+00
2.34805524e-01 1.39062107e+00 3.88412565e-01 2.94051915e-01
-6.15880609e-01 -5.30983925e-01 3.84005994e-01 2.20676601e-01
-1.21536851e+00 -3.87438953e-01 1.24409270e+00 -1.20583866e-02
5.89076519e-01 1.03118762e-01 8.04233551e-01 4.62539583e-01
-8.42223503e-03 1.24934304e+00 1.28165650e+00 -2.89677590e-01
1.06487744e-01 9.21886321e-03 1.93379804e-01 8.64371121e-01
1.23899639e-01 7.76337534e-02 -7.62267172e-01 1.20862603e-01
8.45514119e-01 3.86777103e-01 -3.94211113e-01 -4.10474360e-01
-1.36387992e+00 4.96452719e-01 7.96776772e-01 3.29847366e-01
-6.12437904e-01 -2.20212162e-01 2.74898499e-01 -1.58534199e-01
8.47686827e-02 -5.03715932e-01 -3.54267716e-01 1.08323442e-02
-7.68181622e-01 5.28867133e-02 2.55694866e-01 1.04009068e+00
1.22471976e+00 -1.87102675e-01 -2.75541037e-01 6.21903121e-01
5.76658487e-01 5.95171034e-01 7.59993732e-01 -5.74558616e-01
5.72819889e-01 1.13764119e+00 4.96428758e-02 -1.14526463e+00
-7.47157335e-01 -3.03299606e-01 -8.27023506e-01 1.32472157e-01
-1.39055336e-02 7.40369186e-02 -1.02876723e+00 1.56418359e+00
7.56584466e-01 -4.95035164e-02 1.76058143e-01 1.29741144e+00
1.25050759e+00 4.86267805e-01 -7.76159167e-02 -2.81191409e-01
1.56541324e+00 -8.44454229e-01 -6.25478387e-01 -1.90238684e-01
2.67210275e-01 -7.70424128e-01 1.27287126e+00 1.58803642e-01
-1.29052544e+00 -8.42045128e-01 -1.24945295e+00 -3.09694588e-01
-4.03616041e-01 3.86423111e-01 7.03688443e-01 4.04916584e-01
-7.58151054e-01 1.64651319e-01 -6.48446262e-01 -1.22224800e-01
4.64624017e-01 4.74799365e-01 -3.49730790e-01 -1.20260403e-01
-1.20326996e+00 5.61820209e-01 5.29537320e-01 4.04953986e-01
-5.26509881e-01 -4.83251780e-01 -9.16605294e-01 7.74558410e-02
5.15359402e-01 -5.64254224e-01 9.89583492e-01 -8.59244645e-01
-1.43109059e+00 4.24377650e-01 -2.02653021e-01 2.25073919e-01
1.15692787e-01 5.51390611e-02 -4.12528425e-01 5.94365716e-01
-1.17638614e-02 7.03059971e-01 8.68638039e-01 -1.31295621e+00
-1.29615486e+00 -5.99891722e-01 2.88363487e-01 6.28797889e-01
-7.33091772e-01 -1.28635347e-01 -6.06810391e-01 -3.81376117e-01
6.72838449e-01 -4.33812499e-01 1.29954830e-01 2.21585482e-01
-2.46099278e-01 -2.70092577e-01 1.14109778e+00 -3.33230138e-01
1.22994435e+00 -2.28581238e+00 9.52682178e-03 -5.66473156e-02
3.21289629e-01 3.65185469e-01 -6.91725016e-02 -1.34058846e-02
1.49759755e-01 -2.26955131e-01 -1.83828622e-01 -3.40320528e-01
-1.07498236e-01 -6.42603412e-02 6.49446398e-02 4.12395090e-01
4.73564118e-01 7.50694692e-01 -9.39741194e-01 -8.08299839e-01
6.49227023e-01 5.55393517e-01 -3.27530026e-01 1.06301695e-01
1.62863702e-01 4.96294908e-03 -7.18689501e-01 1.12630093e+00
1.16733539e+00 5.59033863e-02 -4.85244751e-01 -7.80436754e-01
-3.77817869e-01 7.34022213e-03 -1.46889031e+00 1.86420047e+00
-3.12194496e-01 2.25880910e-02 2.15463623e-01 -8.01079810e-01
8.86556804e-01 -3.64999175e-02 6.01356745e-01 -9.87401307e-01
4.28492397e-01 3.45303059e-01 -1.90640405e-01 -7.53456950e-01
4.14231241e-01 -2.00020015e-01 -1.11067183e-01 2.30402097e-01
4.19254415e-02 2.96273649e-01 5.55452928e-02 1.61783233e-01
8.52013052e-01 3.38823080e-01 4.46702950e-02 -4.87632155e-02
1.10568810e+00 -3.13913703e-01 7.35053182e-01 3.17666650e-01
-5.41058064e-01 5.48433721e-01 2.53004521e-01 -3.65628481e-01
-3.42811108e-01 -1.13323748e+00 7.30159357e-02 7.85289764e-01
1.08980978e+00 -2.98695505e-01 -4.12583023e-01 -7.03188896e-01
4.31284085e-02 9.71322507e-02 -5.32164812e-01 -4.91756141e-01
-3.14238518e-01 -7.20116019e-01 2.77925402e-01 5.60494542e-01
1.09696591e+00 -9.81269419e-01 -1.17129529e+00 1.43491760e-01
-3.65748405e-01 -1.05652082e+00 -1.75916180e-01 1.24268495e-01
-8.93382013e-01 -1.02659702e+00 -8.09897780e-01 -8.19287241e-01
6.20219886e-01 8.66468012e-01 4.90378261e-01 1.63781121e-01
-2.53438711e-01 2.79203326e-01 -5.53374827e-01 -4.27917212e-01
4.22262281e-01 2.63917968e-02 -3.12858596e-02 4.69710648e-01
4.31714684e-01 -5.64944446e-01 -1.00671577e+00 2.58443922e-01
-1.35821104e+00 3.25888604e-01 1.17921460e+00 6.54611766e-01
7.12324560e-01 2.23689321e-02 5.91548979e-01 -3.02627999e-02
2.02650234e-01 -2.56779194e-01 -2.80017287e-01 1.19378574e-01
-2.41259187e-01 -2.10235909e-01 5.14882743e-01 -4.03706342e-01
-1.14792991e+00 2.10215628e-01 -1.87215884e-03 -5.89317203e-01
-9.41029713e-02 3.48469824e-01 -5.38528085e-01 -1.78426370e-01
7.31702670e-02 6.14804804e-01 -1.19542487e-01 -4.42783237e-01
3.15275699e-01 9.16456163e-01 6.68453932e-01 -4.08495367e-01
8.48645926e-01 8.21330905e-01 6.33751601e-02 -5.19413888e-01
-7.50584543e-01 -5.40611863e-01 -6.32874548e-01 -4.83446360e-01
7.49800503e-01 -1.07199287e+00 -9.69886482e-01 8.28782201e-01
-9.19984043e-01 2.92136163e-01 -3.03389937e-01 5.78180373e-01
-3.96069258e-01 3.60891610e-01 -4.40380335e-01 -8.60985100e-01
-4.38232243e-01 -1.32459891e+00 1.61103940e+00 8.68110001e-01
6.85396969e-01 -3.11595708e-01 -4.34107900e-01 1.78909048e-01
3.04277062e-01 3.04958940e-01 6.47740245e-01 -3.43833566e-02
-6.59365535e-01 -1.15616366e-01 -7.64073968e-01 1.88066497e-01
3.05725664e-01 -2.54897594e-01 -1.06337571e+00 -7.76467249e-02
1.24114029e-01 -3.09017718e-01 8.77408028e-01 2.32557759e-01
1.08381534e+00 1.67002320e-01 -2.63665318e-01 6.06647313e-01
1.57267308e+00 -4.78776507e-02 4.85825598e-01 4.47250456e-01
8.33511412e-01 5.07176936e-01 1.03309321e+00 5.44358969e-01
8.54208708e-01 4.82998222e-01 7.64718950e-01 -4.66669440e-01
2.40653730e-03 -1.67832732e-01 3.62126768e-01 8.31024885e-01
-2.07810421e-02 1.39465734e-01 -4.83860761e-01 5.68192303e-01
-1.91981411e+00 -6.74703300e-01 -5.36115319e-02 1.93716061e+00
7.18351424e-01 2.48265073e-01 3.19468468e-01 4.89504367e-01
7.02569485e-01 2.19593465e-01 -8.19332361e-01 5.43659292e-02
-3.31637561e-01 -3.68531644e-02 4.56213802e-01 -1.22129917e-01
-1.13370275e+00 5.04592478e-01 3.81062818e+00 1.12780392e+00
-1.09575558e+00 4.60077710e-02 3.46771270e-01 -2.32329115e-01
-3.04367810e-01 1.32383630e-01 -6.53334320e-01 6.74022317e-01
2.39352360e-01 2.92228423e-02 5.71773946e-02 7.10864544e-01
1.27974525e-01 -6.25012159e-01 -7.86977112e-01 1.25150728e+00
1.07414238e-01 -9.32980180e-01 -5.12532890e-03 -1.47177786e-01
3.65311474e-01 -1.81033939e-01 -5.90652265e-02 1.73399329e-01
-3.91209394e-01 -3.27007711e-01 1.02607834e+00 6.88491523e-01
5.64164698e-01 -1.04750860e+00 9.17169750e-01 4.36731398e-01
-1.77314675e+00 -2.93213278e-01 -5.42804539e-01 -2.12140661e-02
1.21450767e-01 7.82923996e-01 2.26377785e-01 1.13129568e+00
1.08202767e+00 8.93234193e-01 -6.77699387e-01 1.00846958e+00
1.99708417e-02 -4.13320661e-02 -5.64214170e-01 -9.41181704e-02
3.81645232e-01 2.32405607e-02 3.91587913e-01 9.58166540e-01
3.05160046e-01 2.41098747e-01 2.81458318e-01 6.15873635e-01
2.36341238e-01 1.40857780e-02 -1.66672960e-01 3.87241125e-01
4.65083510e-01 1.71156621e+00 -9.30468082e-01 -4.26420599e-01
-6.08848870e-01 9.24368799e-01 2.36334465e-02 1.35935619e-01
-8.79689455e-01 -8.64876807e-01 3.51038605e-01 -5.41159995e-02
5.13221920e-01 2.17432063e-02 -1.92944974e-01 -1.21655118e+00
4.64603812e-01 -6.29623830e-01 4.85911310e-01 -1.00099182e+00
-1.00622225e+00 5.25198936e-01 -2.23354790e-02 -1.74572980e+00
3.21380377e-01 -5.38518786e-01 -5.04006028e-01 9.35169041e-01
-2.23435116e+00 -1.68769825e+00 -7.81500399e-01 1.07604873e+00
3.67962450e-01 2.54869729e-01 1.97642192e-01 2.39508063e-01
-7.63320148e-01 4.60933566e-01 -2.52034009e-01 -1.17450178e-01
4.57499772e-01 -8.46801400e-01 -3.01936239e-01 8.62524569e-01
-5.57563663e-01 5.32340169e-01 9.75773260e-02 -3.68300825e-01
-1.79796350e+00 -7.79116094e-01 4.04382497e-01 1.78854242e-01
3.38987857e-01 -1.57737628e-01 -7.28882432e-01 1.96582749e-02
-1.89545229e-01 3.52452427e-01 3.14149290e-01 -5.48349917e-01
-1.15947589e-01 -5.17894924e-01 -1.32303441e+00 3.91391963e-01
1.08945525e+00 -4.96458799e-01 -6.04084015e-01 2.41259672e-02
8.72069776e-01 -4.22868252e-01 -9.50401962e-01 8.39830339e-01
6.61510110e-01 -1.46337259e+00 1.01677561e+00 1.28907964e-01
4.31137443e-01 -9.37271535e-01 -2.34980732e-01 -7.35091090e-01
-1.16828203e-01 -3.84176999e-01 -2.53704935e-01 1.48598361e+00
-2.13432416e-01 -6.63403332e-01 4.84370679e-01 2.78130978e-01
-3.45986515e-01 -1.24781263e+00 -1.00825310e+00 -5.43936849e-01
-5.51048577e-01 -3.54341149e-01 8.52123618e-01 5.06417990e-01
-7.63759911e-02 6.50785044e-02 -8.16898122e-02 2.44541764e-01
6.51356459e-01 4.95978385e-01 5.79645514e-01 -1.09498227e+00
1.01922110e-01 -4.96974289e-01 -6.12195671e-01 -1.08732009e+00
-3.35663617e-01 -4.86140877e-01 9.64871421e-02 -1.42748022e+00
1.78357974e-01 -5.11959851e-01 -6.69257879e-01 5.75478196e-01
-4.31131899e-01 2.85577148e-01 2.56424397e-01 1.78833395e-01
-6.74232423e-01 9.28292453e-01 1.35311449e+00 6.06390685e-02
-2.58584440e-01 -2.58619100e-01 -8.31029356e-01 8.77846181e-01
4.53708529e-01 -2.15916499e-01 -5.35559654e-01 -2.13287860e-01
-3.44985053e-02 -1.21066466e-01 6.81217313e-01 -1.28144503e+00
4.50829387e-01 -1.41297340e-01 7.20312893e-01 -1.13124549e+00
3.23197603e-01 -9.45746720e-01 -5.64438820e-01 4.41904604e-01
1.92279801e-01 -3.78082357e-02 3.81362200e-01 5.27465880e-01
-4.10908252e-01 1.63886160e-01 7.26140499e-01 -1.07341908e-01
-1.17080224e+00 4.76788133e-01 3.08290441e-02 -1.72609612e-01
1.21204484e+00 -6.49502575e-01 -3.54807585e-01 4.47416790e-02
-3.43350112e-01 3.69266897e-01 5.23205221e-01 5.11401594e-01
9.70656395e-01 -1.59771466e+00 -3.71243149e-01 6.12668633e-01
3.23617369e-01 3.67589474e-01 7.89381981e-01 1.18406904e+00
-1.95736259e-01 1.02740183e-01 -5.20715475e-01 -9.15556967e-01
-9.71071005e-01 5.82057238e-01 1.97050080e-01 -6.58517852e-02
-4.10411686e-01 8.08498621e-01 2.32482821e-01 -1.19902559e-01
7.80381262e-02 -4.39206123e-01 -3.37558359e-01 3.04856956e-01
6.13173306e-01 3.72715503e-01 8.19840357e-02 -1.03863275e+00
-6.61533892e-01 1.04763937e+00 1.28464371e-01 4.54401551e-03
1.27450883e+00 -5.65611184e-01 -1.84962660e-01 5.56619465e-01
1.25765169e+00 -4.20387924e-01 -1.32526278e+00 -5.38948357e-01
-4.84670639e-01 -6.54243886e-01 3.94979656e-01 -4.81840849e-01
-1.31106627e+00 1.06806493e+00 8.38812232e-01 6.74802884e-02
2.04971719e+00 -1.44239739e-01 1.11911678e+00 -3.32622975e-02
4.97480959e-01 -1.10103035e+00 2.19283819e-01 -1.65962875e-02
5.16802549e-01 -1.11907566e+00 2.56951451e-01 -5.05468965e-01
-3.91579300e-01 1.10507107e+00 8.90353024e-01 -8.72547626e-02
6.91544712e-01 1.35322630e-01 -1.59769833e-01 -2.69512862e-01
-3.06523919e-01 -6.20025396e-01 3.91069382e-01 6.33741856e-01
-1.62273422e-01 -3.32166910e-01 -2.40059793e-01 9.48057413e-01
3.05356354e-01 1.91000059e-01 2.18687445e-01 1.34924710e+00
-7.13567972e-01 -6.52461648e-01 -3.93864155e-01 4.68962997e-01
-1.01509290e-02 -1.87348481e-03 5.51626645e-02 7.34464109e-01
7.15479195e-01 1.11676979e+00 -4.10483740e-02 -8.14106882e-01
4.83367562e-01 -2.81054795e-01 3.49174052e-01 -6.35948032e-02
-5.10058582e-01 2.13916913e-01 -3.64599764e-01 -8.40260804e-01
-1.02100253e+00 -6.05158687e-01 -1.48580682e+00 -7.76030868e-02
-6.16582036e-01 -2.16029659e-01 5.85763752e-01 9.42180336e-01
4.89452422e-01 6.67839885e-01 8.36986959e-01 -1.33486819e+00
-2.63669640e-01 -8.58777881e-01 -4.44723666e-01 1.96372136e-01
5.40144563e-01 -1.08036971e+00 -4.35778379e-01 -1.20982789e-01]
|
[9.676178932189941, -0.8601583242416382]
|
1780eb5d-b6ce-40b0-a518-8285f4dbeabf
|
arabsign-a-multi-modality-dataset-and
|
2210.03951
| null |
https://arxiv.org/abs/2210.03951v1
|
https://arxiv.org/pdf/2210.03951v1.pdf
|
ArabSign: A Multi-modality Dataset and Benchmark for Continuous Arabic Sign Language Recognition
|
Sign language recognition has attracted the interest of researchers in recent years. While numerous approaches have been proposed for European and Asian sign languages recognition, very limited attempts have been made to develop similar systems for the Arabic sign language (ArSL). This can be attributed partly to the lack of a dataset at the sentence level. In this paper, we aim to make a significant contribution by proposing ArabSign, a continuous ArSL dataset. The proposed dataset consists of 9,335 samples performed by 6 signers. The total time of the recorded sentences is around 10 hours and the average sentence's length is 3.1 signs. ArabSign dataset was recorded using a Kinect V2 camera that provides three types of information (color, depth, and skeleton joint points) recorded simultaneously for each sentence. In addition, we provide the annotation of the dataset according to ArSL and Arabic language structures that can help in studying the linguistic characteristics of ArSL. To benchmark this dataset, we propose an encoder-decoder model for Continuous ArSL recognition. The model has been evaluated on the proposed dataset, and the obtained results show that the encoder-decoder model outperformed the attention mechanism with an average word error rate (WER) of 0.50 compared with 0.62 with the attention mechanism. The data and code are available at github.com/Hamzah-Luqman/ArabSign
|
['Hamzah Luqman']
|
2022-10-08
| null | null | null | null |
['sign-language-recognition']
|
['computer-vision']
|
[-1.11022800e-01 -3.34304273e-01 -2.40406301e-02 -4.80118454e-01
-8.35026264e-01 -3.54159921e-01 6.23093009e-01 -4.68379378e-01
-6.89570487e-01 4.36163425e-01 4.74735051e-01 5.89260645e-02
7.49169216e-02 -2.47679785e-01 -2.57927597e-01 -8.65704894e-01
1.59994334e-01 2.63771862e-01 1.13365866e-01 -7.61856362e-02
5.97091019e-01 6.06943429e-01 -1.72549546e+00 5.51981516e-02
7.95427680e-01 7.46772110e-01 1.97985306e-01 8.50107729e-01
-5.76297492e-02 9.33309078e-01 -6.49072289e-01 -5.21106064e-01
1.47049218e-01 -8.63882363e-01 -4.94813383e-01 1.61515161e-01
6.68128729e-01 -6.26638114e-01 -4.96473789e-01 8.46141398e-01
1.00016093e+00 8.90156720e-03 6.02857769e-01 -9.24364567e-01
-6.51004016e-01 3.37178320e-01 -3.95177722e-01 -1.03736125e-01
4.50421751e-01 2.68342823e-01 1.02823591e+00 -9.33968782e-01
7.66974628e-01 9.77133512e-01 3.37272257e-01 7.60800183e-01
-2.17514560e-01 -5.03939033e-01 -1.03866816e-01 5.03531516e-01
-1.38222289e+00 -4.47760135e-01 7.09715128e-01 -3.39893401e-01
8.40875924e-01 1.61908314e-01 7.91419089e-01 8.51615012e-01
-7.52484426e-02 1.25447965e+00 1.27742791e+00 -7.25832343e-01
6.55144081e-02 -1.51860282e-01 3.60401273e-01 8.30159187e-01
2.08947390e-01 -2.65203416e-01 -6.50598466e-01 3.21236461e-01
5.21362841e-01 -3.75581145e-01 -3.48011076e-01 2.37076003e-02
-1.04433572e+00 5.53403437e-01 1.89245611e-01 5.78790843e-01
-3.24433863e-01 1.19111739e-01 4.83345956e-01 2.55609602e-01
-6.82843626e-02 -1.68021500e-01 -3.13793033e-01 -6.99895918e-01
-8.25193703e-01 6.51953788e-03 7.46755064e-01 9.03735995e-01
-1.32526398e-01 2.37858847e-01 2.23379612e-01 8.74119103e-01
8.38648081e-01 1.00059485e+00 8.05338264e-01 -4.80257750e-01
5.76550901e-01 4.65958953e-01 -1.84994310e-01 -7.68080592e-01
-2.01281101e-01 8.03598017e-02 -4.81695443e-01 3.06661874e-01
7.79180944e-01 -1.50118306e-01 -1.13505733e+00 1.11202812e+00
-1.48498923e-01 -2.28159070e-01 1.71148539e-01 1.41117871e+00
1.19680548e+00 4.74731982e-01 -6.30518869e-02 1.91283584e-01
1.38891280e+00 -9.51072037e-01 -8.72216225e-01 1.24590278e-01
6.41726136e-01 -9.23827052e-01 1.20853686e+00 5.81124663e-01
-9.50908780e-01 -2.60741740e-01 -8.59756052e-01 -7.71367773e-02
-3.20069134e-01 7.10727990e-01 3.05960804e-01 1.11361289e+00
-7.56457388e-01 -1.61357090e-01 -9.30118263e-01 -7.65108526e-01
2.75964111e-01 1.78429872e-01 -3.98061365e-01 -1.81746989e-01
-7.65058637e-01 9.71754909e-01 4.65802029e-02 5.84212303e-01
-3.70978504e-01 1.47299737e-01 -7.08720207e-01 -4.62753624e-01
-1.23076141e-01 5.62973209e-02 1.11953819e+00 -9.36562777e-01
-1.94072270e+00 1.04350269e+00 -2.47732490e-01 -1.21832140e-01
7.60352790e-01 -2.34029308e-01 -5.82381904e-01 2.52104491e-01
-3.22159797e-01 3.89614552e-01 5.01405597e-01 -9.72099066e-01
-4.56828952e-01 -3.68485987e-01 -2.23584384e-01 1.85639948e-01
-7.21879527e-02 6.71446502e-01 -6.58590913e-01 -5.86975396e-01
2.01491788e-01 -8.58459592e-01 2.35000446e-01 -2.79413909e-02
-1.50720894e-01 -5.35683986e-03 8.12313437e-01 -1.24997342e+00
1.11446881e+00 -2.06351328e+00 -8.37533921e-02 1.17300250e-01
-4.51017380e-01 6.29316032e-01 -2.04993173e-01 4.54839110e-01
2.65525043e-01 -1.59706801e-01 -4.83676791e-01 -2.60904878e-01
6.66871741e-02 4.16362733e-01 -6.21622019e-02 7.25679576e-01
3.18088643e-02 8.78799796e-01 -5.60794175e-01 -6.66417241e-01
3.02164882e-01 7.02383220e-01 -2.24719107e-01 5.93622066e-02
1.56090245e-01 5.02644479e-01 -3.18600178e-01 1.29385352e+00
6.01702213e-01 2.34819606e-01 3.65001447e-02 -3.26635651e-02
-3.31019878e-01 -3.13987173e-02 -1.19248378e+00 1.40317976e+00
-2.75772244e-01 9.91174996e-01 -1.09176859e-01 -6.79466009e-01
1.06238496e+00 4.45594013e-01 2.76105940e-01 -8.88165295e-01
5.01944184e-01 7.30055392e-01 2.00837567e-01 -9.67391312e-01
4.15287554e-01 6.98950216e-02 1.05847809e-02 3.83465648e-01
-1.54101759e-01 -7.62773007e-02 5.43413818e-01 -3.03640783e-01
7.59915352e-01 2.05830783e-01 1.45669833e-01 3.41477901e-01
9.51098084e-01 -5.73751107e-02 2.33449712e-01 4.60400492e-01
-4.61245626e-01 7.96929955e-01 2.64581561e-01 -2.03765184e-01
-8.09974074e-01 -8.86149645e-01 -1.70111060e-01 6.40498996e-01
-8.45904723e-02 -8.15288574e-02 -5.13817191e-01 -5.26882648e-01
-2.28976145e-01 4.70380396e-01 -2.78778285e-01 5.47044575e-01
-8.09277296e-01 -5.24295628e-01 9.41545665e-01 6.04465902e-01
1.02649319e+00 -1.52780056e+00 -9.99934614e-01 -1.37273818e-01
-1.37621194e-01 -1.08209670e+00 -4.51501697e-01 -3.57096583e-01
-8.27787817e-01 -1.30892694e+00 -1.22324181e+00 -9.50020730e-01
8.17654550e-01 -3.18604350e-01 5.43565154e-01 -8.59480575e-02
-3.34803402e-01 5.64585388e-01 -8.17442894e-01 -2.92805225e-01
-2.42620960e-01 -1.28704995e-01 -2.01966509e-01 1.83338299e-01
6.11486256e-01 8.13476667e-02 -2.79416978e-01 3.27565610e-01
-8.48504305e-01 -2.26582423e-01 8.52923334e-01 8.41428816e-01
2.03984737e-01 -6.27241373e-01 3.03762048e-01 -2.22606987e-01
5.74164808e-01 9.43819284e-02 -7.84576774e-01 2.55320996e-01
-2.37717301e-01 -3.96210421e-03 2.36486301e-01 -1.74764395e-01
-9.56997216e-01 1.64723799e-01 -4.83622193e-01 1.43787786e-01
-4.11020666e-01 6.22790337e-01 -2.46131510e-01 -1.68560922e-01
7.55329877e-02 5.88824272e-01 2.58200228e-01 -6.73637271e-01
1.67921141e-01 1.40376425e+00 4.96176481e-01 -1.49984568e-01
2.32743695e-01 3.77002418e-01 -2.23844916e-01 -1.35765696e+00
-3.38409156e-01 -5.25851905e-01 -7.57772923e-01 -6.71761096e-01
7.93148935e-01 -6.94300234e-01 -8.56366634e-01 1.15675986e+00
-1.01096952e+00 -1.58129781e-01 -1.57501459e-01 8.08044016e-01
-5.72103560e-01 7.02345848e-01 -5.65675437e-01 -1.12815416e+00
-4.05374289e-01 -1.13984776e+00 9.58326280e-01 2.47291401e-01
-1.50760740e-01 -6.98639989e-01 1.27583593e-02 6.98965788e-01
3.90011251e-01 1.39107913e-01 3.06675881e-01 -4.35796022e-01
-5.72240233e-01 -6.32535219e-01 -3.26439947e-01 6.01361692e-01
1.50049955e-01 7.22251832e-02 -8.54766846e-01 -7.23824650e-02
-2.85649270e-01 -5.12770414e-01 8.07030678e-01 3.89526814e-01
6.47985458e-01 -2.18627658e-02 3.81046295e-01 2.32170954e-01
1.29079235e+00 4.81675327e-01 9.02097166e-01 3.68255407e-01
5.30128241e-01 3.89049858e-01 6.68260932e-01 5.15834808e-01
4.15606052e-01 6.97893858e-01 1.50353238e-01 1.84707582e-01
-5.47339797e-01 -1.10063851e-01 6.93551183e-01 1.32429612e+00
-5.04899800e-01 -2.90726632e-01 -1.11959136e+00 6.42981589e-01
-1.59003663e+00 -9.52163696e-01 -5.03552735e-01 2.03644490e+00
5.86040735e-01 -3.30569685e-01 2.53546089e-01 5.45143247e-01
3.70988339e-01 1.82890743e-01 -1.54845759e-01 -6.05339348e-01
-5.03085017e-01 1.34609818e-01 5.97826838e-01 7.41193533e-01
-9.81352210e-01 1.00190163e+00 5.62754154e+00 2.91779041e-01
-1.54381120e+00 -1.34620145e-01 -1.03934914e-01 1.31122824e-02
2.02845231e-01 -3.17942679e-01 -8.65237713e-01 7.08488047e-01
8.75129402e-01 1.03130914e-01 1.47802740e-01 5.77620625e-01
4.67759609e-01 -3.46330315e-01 -5.61968863e-01 1.12530518e+00
7.67342687e-01 -7.51991868e-01 5.11131287e-02 6.29433542e-02
4.66582477e-01 3.45452964e-01 -7.36692324e-02 -3.14605720e-02
-1.48220778e-01 -9.31756496e-01 8.06350768e-01 9.32064176e-01
8.92300427e-01 -4.47350442e-01 1.21439540e+00 2.67996073e-01
-1.01414490e+00 -3.17819268e-02 1.37710366e-02 -8.47313926e-03
4.11805302e-01 -8.52091610e-03 -7.07455993e-01 3.41364115e-01
4.41693366e-01 8.20152521e-01 -6.06981277e-01 1.54199696e+00
-5.58724344e-01 8.75989616e-01 -4.81552273e-01 -5.48131764e-01
2.61839300e-01 -3.36752981e-01 5.52798629e-01 1.29077017e+00
5.60975015e-01 -7.72039071e-02 -4.00800586e-01 2.27871001e-01
2.72101194e-01 5.26919782e-01 -4.72486109e-01 -3.02275926e-01
-8.40773955e-02 5.92929959e-01 -5.51291525e-01 -2.25269124e-01
-6.16299868e-01 1.09744442e+00 -4.36486095e-01 2.97988743e-01
-6.65051222e-01 -7.01787472e-01 3.14320534e-01 -1.28755361e-01
4.38945562e-01 -5.04130661e-01 -3.14264417e-01 -1.23814142e+00
4.46387053e-01 -8.42395127e-01 3.50309104e-01 -8.83366585e-01
-1.07615292e+00 5.43399513e-01 -2.86698580e-01 -1.41504562e+00
-2.50456303e-01 -1.04119694e+00 -2.38183051e-01 8.96714568e-01
-1.53634226e+00 -1.33092701e+00 -5.78016579e-01 4.64757204e-01
6.18876636e-01 -5.19330680e-01 8.69091868e-01 6.05971634e-01
-3.84617329e-01 7.10735083e-01 1.64816499e-01 6.62146628e-01
6.30798221e-01 -9.80345726e-01 4.20137346e-02 9.49400663e-01
2.51699060e-01 2.96986967e-01 4.79484349e-01 -5.43464065e-01
-1.35506117e+00 -5.23506522e-01 1.60502529e+00 -3.53165537e-01
5.63430429e-01 1.43699721e-01 -4.24287021e-01 4.56329733e-01
1.64074168e-01 -4.67565469e-02 5.35175383e-01 -5.55667698e-01
-1.75329506e-01 -1.65674035e-02 -1.18333244e+00 5.35200715e-01
7.75614738e-01 -5.08706331e-01 -6.89944804e-01 5.89846000e-02
-3.56777340e-01 -4.66469139e-01 -5.51514089e-01 4.85806279e-02
1.04765260e+00 -8.54809701e-01 4.40056056e-01 -3.07009310e-01
4.03820425e-01 -4.84949768e-01 -3.53386492e-01 -8.40390861e-01
4.40917522e-01 1.43879820e-02 -4.96565290e-02 1.07847846e+00
2.44010523e-01 -7.81214893e-01 8.94084632e-01 5.37985325e-01
-2.85368841e-02 -5.90293646e-01 -1.09575963e+00 -8.68304729e-01
-8.05370882e-02 -6.45447433e-01 1.63089424e-01 4.28437024e-01
-9.89123210e-02 -1.81632206e-01 -4.97853637e-01 1.22875096e-02
5.25356412e-01 6.95770830e-02 8.66930604e-01 -9.72190619e-01
-4.39041480e-02 -5.32268226e-01 -8.50953043e-01 -1.23053002e+00
-3.68547104e-02 -8.44855845e-01 -1.51395900e-02 -1.78590631e+00
-1.55732274e-01 -4.20223214e-02 8.04029480e-02 6.23167634e-01
3.59784901e-01 4.62085068e-01 4.89759594e-01 2.90396273e-01
-3.65379602e-01 5.81124902e-01 1.19395375e+00 -9.20588151e-02
-7.13665634e-02 2.82259714e-02 -8.35979432e-02 6.79005742e-01
9.07655656e-01 5.90566583e-02 1.39075488e-01 -5.91238737e-01
-2.19093919e-01 -1.62086651e-01 3.66022021e-01 -1.01099575e+00
2.96253026e-01 2.34250799e-01 1.54677173e-02 -9.30889904e-01
3.43814999e-01 -8.40323985e-01 -2.56879598e-01 6.58039391e-01
-1.74689889e-01 -3.82348672e-02 2.88643539e-02 1.11129202e-01
-6.77761257e-01 -3.24873388e-01 7.35958874e-01 5.58444932e-02
-9.94826317e-01 -1.37911573e-01 -6.21712565e-01 -7.43484646e-02
8.69652390e-01 -6.57745481e-01 -5.90078272e-02 -4.65233892e-01
-5.76181114e-01 6.51631430e-02 2.70954221e-01 4.37106878e-01
8.81633401e-01 -1.20484257e+00 -9.01610494e-01 4.62692261e-01
2.64949918e-01 -4.41393852e-01 1.37408882e-01 1.02110744e+00
-1.21711254e+00 7.66255379e-01 -4.13028061e-01 -4.01917964e-01
-1.85837424e+00 -4.09764320e-01 2.75379807e-01 1.81841776e-01
-7.34767616e-01 7.76880383e-01 -7.83289492e-01 -4.14046764e-01
4.71651018e-01 -4.06478226e-01 -3.07366610e-01 1.83161676e-01
4.33388203e-01 3.60593766e-01 -7.41180927e-02 -1.20778239e+00
-5.07443130e-01 1.04501545e+00 1.57990217e-01 -4.48636800e-01
1.31807077e+00 9.28742662e-02 -1.72351804e-02 4.67174470e-01
9.67808068e-01 2.70803690e-01 -8.02084267e-01 6.10438036e-03
1.91734388e-01 -7.94277191e-01 -1.24297500e-01 -1.09574914e+00
-1.03639305e+00 9.46042717e-01 8.72819841e-01 -3.98268133e-01
1.18766654e+00 -1.03891954e-01 8.31533074e-01 3.88483852e-01
4.74639624e-01 -1.24380910e+00 -3.13156456e-01 8.78610253e-01
1.24653590e+00 -1.26846337e+00 -2.30112374e-01 -3.25072999e-03
-9.00052011e-01 1.17911911e+00 1.91731066e-01 -2.57921126e-02
5.46388328e-01 1.90605074e-01 7.08307683e-01 -5.02072200e-02
-7.85461366e-02 -4.86321807e-01 3.25439334e-01 4.45028901e-01
8.59644413e-01 6.30540252e-02 -9.27386045e-01 3.82530004e-01
-3.39220911e-01 4.37106282e-01 6.21064782e-01 1.13277674e+00
-2.21483871e-01 -1.23416424e+00 -5.51559448e-01 2.32880458e-01
-2.67553121e-01 6.99170083e-02 -6.16728306e-01 8.85618091e-01
2.29114778e-02 7.64665663e-01 -9.92893130e-02 1.72114857e-02
5.52935123e-01 2.67914385e-01 6.50079012e-01 -7.31339976e-02
-3.92396480e-01 -1.13375716e-01 2.64077276e-01 -3.11551273e-01
-6.52484655e-01 -1.04333997e+00 -1.33790636e+00 -1.13473879e-02
-1.53183594e-01 -1.14691846e-01 8.82657588e-01 8.66537213e-01
-1.09123804e-01 1.33146197e-01 1.62828773e-01 -5.66530287e-01
-4.61311787e-01 -1.11117840e+00 -7.21696794e-01 3.32925946e-01
2.05919206e-01 -4.37019676e-01 -3.89783412e-01 2.58683503e-01]
|
[9.116056442260742, -6.421206474304199]
|
bb07989b-1584-4e70-961e-432dd8cb18ff
|
mind-the-gap-alleviating-local-imbalance-for
|
2205.11888
| null |
https://arxiv.org/abs/2205.11888v2
|
https://arxiv.org/pdf/2205.11888v2.pdf
|
Mind The Gap: Alleviating Local Imbalance for Unsupervised Cross-Modality Medical Image Segmentation
|
Unsupervised cross-modality medical image adaptation aims to alleviate the severe domain gap between different imaging modalities without using the target domain label. A key in this campaign relies upon aligning the distributions of source and target domain. One common attempt is to enforce the global alignment between two domains, which, however, ignores the fatal local-imbalance domain gap problem, i.e., some local features with larger domain gap are harder to transfer. Recently, some methods conduct alignment focusing on local regions to improve the efficiency of model learning. While this operation may cause a deficiency of critical information from contexts. To tackle this limitation, we propose a novel strategy to alleviate the domain gap imbalance considering the characteristics of medical images, namely Global-Local Union Alignment. Specifically, a feature-disentanglement style-transfer module first synthesizes the target-like source-content images to reduce the global domain gap. Then, a local feature mask is integrated to reduce the 'inter-gap' for local features by prioritizing those discriminative features with larger domain gap. This combination of global and local alignment can precisely localize the crucial regions in segmentation target while preserving the overall semantic consistency. We conduct a series of experiments with two cross-modality adaptation tasks, i,e. cardiac substructure and abdominal multi-organ segmentation. Experimental results indicate that our method achieves state-of-the-art performance in both tasks.
|
['Kaizhu Huang', 'Jie Sun', 'Yuyao Yan', 'Qiufeng Wang', 'Xi Yang', 'Kai Yao', 'Zixian Su']
|
2022-05-24
| null | null | null | null |
['cardiac-segmentation']
|
['medical']
|
[ 4.41654503e-01 -1.10156111e-01 -4.41195279e-01 -5.07533014e-01
-1.08907902e+00 -5.26989102e-01 3.75970930e-01 2.22884431e-01
-4.04094011e-01 5.24052918e-01 2.74719834e-01 1.08419821e-01
-2.48948276e-01 -6.07238889e-01 -5.27459681e-01 -9.99179900e-01
4.66321737e-01 2.59700477e-01 3.64543885e-01 -1.76425979e-01
6.50482923e-02 1.97257370e-01 -9.87799466e-01 4.37158465e-01
1.21942747e+00 8.71683240e-01 4.23512876e-01 5.26321866e-03
-2.55498976e-01 2.01360986e-01 -4.09419060e-01 -7.65962154e-02
3.17651331e-01 -8.71062875e-01 -7.59193718e-01 2.14718297e-01
2.26661146e-01 -1.96398329e-02 6.75108209e-02 1.34263039e+00
7.73352981e-01 -1.48722082e-01 8.30125928e-01 -1.06509078e+00
-4.06215161e-01 2.91675717e-01 -9.75748181e-01 3.45175564e-01
1.25637829e-01 1.74851865e-01 7.68314123e-01 -5.40528774e-01
5.82062662e-01 8.94880176e-01 6.31624639e-01 4.97527868e-01
-1.24299502e+00 -8.35145116e-01 1.62590310e-01 -9.34526548e-02
-1.29038692e+00 -1.66869491e-01 1.11476707e+00 -4.92910683e-01
3.04191470e-01 2.90081859e-01 2.33222902e-01 1.01647329e+00
3.04144233e-01 5.18855631e-01 1.20649958e+00 -4.63823944e-01
-7.72214755e-02 2.29108498e-01 -1.65922403e-01 6.20115221e-01
-3.21463565e-03 -9.08681825e-02 -3.53543937e-01 -7.68780932e-02
7.31959879e-01 7.14130104e-02 -2.91439503e-01 -7.60260522e-01
-1.55535102e+00 6.66768849e-01 4.64906961e-01 7.03146636e-01
-1.91742077e-01 -5.36967635e-01 6.86720014e-01 2.62144417e-01
4.46585625e-01 3.45857382e-01 -6.49544775e-01 3.23970348e-01
-7.25602210e-01 -9.02139619e-02 3.84971827e-01 8.01824808e-01
7.58955956e-01 -4.71886814e-01 -4.29980367e-01 1.08268189e+00
1.11954652e-01 2.65018165e-01 7.32895911e-01 -5.30563831e-01
7.05289364e-01 7.84778535e-01 -3.56523305e-01 -1.01056397e+00
-4.37852025e-01 -5.26616096e-01 -1.17507172e+00 -1.19203188e-01
6.80203915e-01 -6.71988237e-04 -1.01044953e+00 1.92334545e+00
6.34355247e-01 9.17734113e-03 1.01344637e-03 1.08726883e+00
8.66896629e-01 3.47802758e-01 4.17796910e-01 -3.71422678e-01
1.65703940e+00 -8.84338081e-01 -6.03469551e-01 -2.17544451e-01
6.46106064e-01 -9.81996357e-01 1.18164515e+00 -1.41410932e-01
-6.27708137e-01 -6.49884224e-01 -8.56786549e-01 1.77433208e-01
-2.15255797e-01 6.44775629e-02 3.78614277e-01 4.06639814e-01
-5.02631962e-01 3.27663600e-01 -5.91675580e-01 -4.29054916e-01
3.97228032e-01 2.30729938e-01 -6.24032080e-01 -2.24148393e-01
-1.27748907e+00 8.29320967e-01 6.30349100e-01 -1.84208855e-01
-4.92978841e-01 -1.02503610e+00 -9.02394831e-01 -1.57510176e-01
3.25145036e-01 -8.00766170e-01 8.51691127e-01 -1.40724099e+00
-1.21952009e+00 1.20940340e+00 5.10515608e-02 3.91481407e-02
6.12816751e-01 2.45390892e-01 -5.73602676e-01 1.98670447e-01
4.96045470e-01 6.29432261e-01 8.64234984e-01 -1.26837099e+00
-7.13415146e-01 -4.80922848e-01 -3.24082613e-01 5.66669226e-01
-5.20961463e-01 -1.53999720e-02 -6.35055959e-01 -9.13459480e-01
3.81035149e-01 -9.35647726e-01 -3.44422460e-01 7.73955882e-02
-3.87518764e-01 -5.21510877e-02 5.34477651e-01 -6.98848128e-01
1.18123937e+00 -2.36269498e+00 1.35177642e-01 3.63977849e-01
5.98727427e-02 1.42695736e-02 -2.07202271e-01 1.54403625e-02
-3.25909346e-01 -5.28835431e-02 -5.37690401e-01 -8.30821842e-02
-3.35157126e-01 2.18268260e-01 1.23422006e-02 5.57836115e-01
2.26008102e-01 6.13460362e-01 -9.27681565e-01 -1.00180304e+00
1.72821894e-01 1.19270355e-01 -6.50381327e-01 2.74656743e-01
2.97145639e-02 1.19966567e+00 -6.95531964e-01 6.92178369e-01
9.47235107e-01 -2.42666200e-01 2.34857708e-01 -6.24980211e-01
5.45789413e-02 2.06528995e-02 -1.07342970e+00 2.25626636e+00
-3.22995603e-01 -1.14961289e-01 5.54611161e-02 -1.31963480e+00
8.50701034e-01 3.32589537e-01 1.05556536e+00 -1.03898656e+00
-1.03737414e-02 4.71146822e-01 1.04504332e-01 -6.10863090e-01
-9.18840170e-02 -4.96543407e-01 -3.50527227e-01 -1.11930305e-02
9.64017808e-02 1.37168327e-02 -6.51439056e-02 -6.13260455e-02
6.28653646e-01 1.73072442e-01 4.87129360e-01 -5.27914226e-01
7.35325754e-01 9.18310955e-02 9.42733884e-01 4.37730014e-01
-4.15985882e-01 1.05345774e+00 5.00454843e-01 -3.31458092e-01
-8.14830184e-01 -1.08766711e+00 -2.75392711e-01 1.03902566e+00
6.23106420e-01 -8.59112293e-02 -7.05627978e-01 -1.20317972e+00
-1.47143424e-01 2.17873409e-01 -7.61794031e-01 -4.19447690e-01
-6.87825978e-01 -1.08286917e+00 4.30778265e-01 4.02279794e-01
5.81143975e-01 -1.03773093e+00 -4.59534436e-01 9.43810195e-02
-4.85928595e-01 -9.63580608e-01 -9.21738446e-01 9.95861962e-02
-9.98022437e-01 -1.00336814e+00 -1.00099492e+00 -1.00235319e+00
9.43001390e-01 2.08336353e-01 1.17336237e+00 -3.30557376e-02
-1.17918715e-01 1.48441689e-03 -4.34274703e-01 -9.59552303e-02
-3.51347178e-01 3.08759689e-01 -2.90176660e-01 1.56934291e-01
1.55856714e-01 -4.66475308e-01 -9.08104300e-01 6.63062036e-01
-9.82918084e-01 2.81734824e-01 8.04680049e-01 1.20639193e+00
9.16917741e-01 -1.18253045e-01 5.81114650e-01 -8.97033393e-01
2.82587707e-01 -5.23353517e-01 -1.33610219e-01 4.80453700e-01
-4.80157852e-01 -1.48855457e-02 6.24788761e-01 -5.42779386e-01
-1.13815749e+00 2.28053674e-01 -8.68212283e-02 -2.44235799e-01
-3.93208623e-01 5.03637016e-01 -4.88730997e-01 1.16853015e-02
6.08701468e-01 3.84386480e-01 1.28177151e-01 -3.99809122e-01
7.87416566e-03 4.35524434e-01 4.78037179e-01 -7.58326292e-01
6.53540492e-01 4.71134096e-01 -5.44220470e-02 -3.31843942e-01
-8.71568561e-01 -6.41273320e-01 -7.31249630e-01 5.22992574e-02
1.00359952e+00 -9.23006296e-01 -7.64790410e-03 4.37271625e-01
-8.16521823e-01 -4.25343809e-04 -3.07553828e-01 5.92233837e-01
-4.01498377e-01 4.47587967e-01 -3.35842490e-01 3.21573168e-02
-3.26237828e-01 -1.36157572e+00 1.06822836e+00 5.25152385e-01
-7.29704574e-02 -9.87902462e-01 2.20316768e-01 2.43779629e-01
3.10149550e-01 4.03947085e-01 1.02532721e+00 -7.78303564e-01
-1.61068231e-01 1.74709320e-01 -5.15175164e-01 2.91827112e-01
6.14591658e-01 -4.06642616e-01 -7.54275918e-01 -3.83950412e-01
4.51289192e-02 -4.91456203e-02 6.88258827e-01 4.74687457e-01
1.23820734e+00 6.54651411e-03 -5.42217910e-01 8.57581854e-01
1.40122271e+00 2.13948950e-01 4.11344826e-01 2.87145138e-01
8.00535142e-01 6.93307281e-01 9.97381449e-01 3.03800881e-01
3.35404515e-01 7.80243039e-01 2.20816284e-01 -6.65520072e-01
-3.55803579e-01 -2.44141385e-01 -1.12240911e-02 8.55215430e-01
1.90832824e-01 5.07262014e-02 -9.72820520e-01 7.22089589e-01
-1.73138404e+00 -4.79954392e-01 3.39410186e-01 2.18196750e+00
1.28712022e+00 3.84182855e-02 6.69561476e-02 -2.80125141e-01
8.52016687e-01 5.62808402e-02 -4.45684582e-01 1.00775935e-01
-2.82537583e-02 -3.12299933e-03 3.54878902e-01 2.08020329e-01
-1.44141495e+00 6.30955160e-01 4.96786165e+00 1.16912425e+00
-1.24726188e+00 2.96403885e-01 7.35645175e-01 2.66601562e-01
-2.92788088e-01 -9.31056440e-02 -5.79825044e-01 7.67149508e-01
2.97910959e-01 5.51587567e-02 3.99342626e-02 6.23410940e-01
-9.24364850e-02 -6.24503829e-02 -1.02392793e+00 9.06939864e-01
8.16367790e-02 -9.66778457e-01 3.05311158e-02 -1.07868373e-01
7.30281293e-01 -2.67939240e-01 3.06040421e-02 2.53307372e-01
-2.70948261e-01 -7.79940426e-01 3.57499331e-01 4.21695352e-01
1.08859384e+00 -6.78877354e-01 8.53023767e-01 2.94079036e-01
-1.14640820e+00 1.83010802e-01 -1.43232629e-01 6.05857551e-01
-2.04391107e-02 6.80533946e-01 -5.79250574e-01 9.65690672e-01
7.90921628e-01 6.97950244e-01 -6.31276131e-01 9.45755780e-01
2.09385887e-01 1.46707684e-01 -2.12257490e-01 5.73807776e-01
2.54815053e-02 -2.56253541e-01 6.06883824e-01 1.34916174e+00
3.45922321e-01 1.20591267e-03 4.22702402e-01 6.31748199e-01
1.01214252e-01 3.54238093e-01 -4.94196087e-01 4.09043729e-01
2.84237474e-01 1.22963703e+00 -8.26751292e-01 -2.71232933e-01
-4.80338991e-01 1.01577401e+00 -2.97374725e-02 1.88187107e-01
-1.09283900e+00 -1.61099881e-01 4.56436425e-01 8.96853581e-02
1.08380631e-01 1.62895501e-01 -5.35636604e-01 -1.30814946e+00
4.32767868e-02 -1.01718593e+00 8.64428699e-01 -2.51725882e-01
-1.58250713e+00 5.01468122e-01 1.21732682e-01 -1.78141022e+00
2.31797211e-02 -1.35102198e-01 -4.67846483e-01 9.16095495e-01
-1.65304172e+00 -1.42240465e+00 -4.52390313e-01 9.25384939e-01
4.94239330e-01 -5.10428995e-02 8.00551534e-01 7.19067276e-01
-4.97937799e-01 9.27197635e-01 -3.68529633e-02 1.39939219e-01
1.33358788e+00 -1.12288785e+00 -3.52596879e-01 6.58625960e-01
-2.60639757e-01 5.97774148e-01 5.11864543e-01 -6.10755980e-01
-8.16207588e-01 -1.05253088e+00 6.41487360e-01 -2.33530954e-01
2.51750886e-01 -2.88166162e-02 -1.11582839e+00 3.94287914e-01
1.44416049e-01 4.13717777e-01 7.26270080e-01 -6.41309544e-02
-3.21978211e-01 -4.63208467e-01 -1.36085105e+00 5.22417188e-01
9.68588114e-01 -3.26190710e-01 -6.30792260e-01 3.61728817e-01
5.85927844e-01 -6.91987395e-01 -1.18637776e+00 8.58760536e-01
4.47434068e-01 -8.53743076e-01 9.89744604e-01 -4.68204260e-01
4.91512120e-01 -5.27032495e-01 -3.67505215e-02 -1.34337604e+00
-3.14052165e-01 -1.63779199e-01 3.73012632e-01 1.60147440e+00
2.86085933e-01 -6.33132756e-01 4.94502574e-01 3.97684187e-01
-2.33118713e-01 -6.37782693e-01 -1.02204633e+00 -4.72197622e-01
2.74330288e-01 1.44689322e-01 4.93553221e-01 1.38203895e+00
-7.34935552e-02 2.00614795e-01 -1.35049984e-01 1.60593137e-01
4.99466151e-01 5.12443244e-01 5.80925584e-01 -8.79813492e-01
-2.30056107e-01 -4.73420382e-01 -8.95669684e-02 -7.00333595e-01
-1.96077805e-02 -9.89388347e-01 1.28135100e-01 -1.21793580e+00
5.83520114e-01 -7.38515198e-01 -8.69906187e-01 5.36441147e-01
-4.46733534e-01 3.70500028e-01 7.24320933e-02 3.39666039e-01
-6.16021216e-01 4.01966274e-01 1.81247735e+00 -1.51112333e-01
-2.36637846e-01 -3.01601198e-02 -8.27337384e-01 6.35608435e-01
7.95595109e-01 -7.24409163e-01 -3.45756978e-01 -3.32291037e-01
-2.28998557e-01 -1.04603823e-02 3.26092869e-01 -9.11484182e-01
9.44814011e-02 -3.72168481e-01 5.74805915e-01 -3.24278295e-01
-2.30625585e-01 -9.65366900e-01 4.05644886e-02 4.60835934e-01
-2.60146409e-01 -7.47880191e-02 2.07301170e-01 4.18236315e-01
-4.90309477e-01 -1.80013720e-02 1.20579576e+00 -1.61063746e-01
-7.52979994e-01 3.42342824e-01 2.45014697e-01 2.70611972e-01
1.12931120e+00 -1.49079710e-01 -1.22717351e-01 1.20981745e-01
-7.32166946e-01 3.47902179e-01 5.28638959e-01 4.70904946e-01
3.55383992e-01 -1.39110088e+00 -7.82854974e-01 4.42430615e-01
3.88968706e-01 2.06658259e-01 7.10910916e-01 1.31814945e+00
-2.97629446e-01 1.18171364e-01 -4.80698973e-01 -9.83258188e-01
-1.19297373e+00 5.19914865e-01 5.56984603e-01 -7.61248767e-01
-5.61451793e-01 7.88941622e-01 8.43241632e-01 -6.73998117e-01
-1.49288073e-01 -1.92089364e-01 -2.82381147e-01 1.38672277e-01
2.14618713e-01 -2.16456205e-01 1.22781537e-01 -6.85466170e-01
-7.13766515e-01 9.38685179e-01 -2.11639360e-01 2.25011498e-01
1.06516302e+00 -2.76680708e-01 -1.09271027e-01 1.16499513e-01
1.31745493e+00 2.88418140e-02 -1.36428511e+00 -4.54210639e-01
-2.55803764e-01 -5.58390200e-01 -3.03076774e-01 -8.96361411e-01
-1.28842485e+00 7.60578156e-01 9.65461910e-01 -1.45463765e-01
1.62066329e+00 1.43330947e-01 7.96907663e-01 -4.78218079e-01
1.11949638e-01 -1.08777022e+00 1.14426851e-01 2.72907168e-01
7.87468970e-01 -1.46984923e+00 5.26275784e-02 -6.44041538e-01
-9.16710496e-01 8.82159710e-01 8.93501699e-01 8.62519592e-02
5.54197252e-01 1.83072705e-02 1.71142802e-01 -9.59723145e-02
-8.68484452e-02 -1.52027428e-01 5.83398938e-01 5.14569283e-01
4.38710451e-01 4.06699628e-02 -6.21176302e-01 6.61406577e-01
3.13304722e-01 -1.97440878e-01 -2.32547253e-01 7.72641420e-01
-4.68468443e-02 -1.31657839e+00 -4.23070014e-01 2.45863616e-01
-7.53106654e-01 1.89908799e-02 9.96096209e-02 8.67422521e-01
5.57899356e-01 4.94439632e-01 -2.00000722e-02 -2.29810894e-01
5.15354335e-01 -1.37849480e-01 4.57689881e-01 -5.64519525e-01
-6.59481525e-01 4.47789043e-01 -3.05494070e-01 -4.92759824e-01
-6.27353191e-01 -6.66561604e-01 -1.18959010e+00 2.09661216e-01
-2.89864242e-01 -1.05935618e-01 2.92716503e-01 9.52205241e-01
3.86653721e-01 7.03398824e-01 7.55544543e-01 -4.89620775e-01
-5.57431042e-01 -8.65623593e-01 -4.38898385e-01 9.94958758e-01
3.63666058e-01 -6.30957961e-01 -9.91762951e-02 2.06852555e-01]
|
[14.501206398010254, -1.9448487758636475]
|
a7cbdfd5-4617-4fa3-bf20-4bac703351af
|
intent-recognition-and-unsupervised-slot
|
2104.01287
| null |
https://arxiv.org/abs/2104.01287v3
|
https://arxiv.org/pdf/2104.01287v3.pdf
|
Intent Recognition and Unsupervised Slot Identification for Low Resourced Spoken Dialog Systems
|
Intent Recognition and Slot Identification are crucial components in spoken language understanding (SLU) systems. In this paper, we present a novel approach towards both these tasks in the context of low resourced and unwritten languages. We present an acoustic based SLU system that converts speech to its phonetic transcription using a universal phone recognition system. We build a word-free natural language understanding module that does intent recognition and slot identification from these phonetic transcription. Our proposed SLU system performs competitively for resource rich scenarios and significantly outperforms existing approaches as the amount of available data reduces. We observe more than 10% improvement for intent classification in Tamil and more than 5% improvement for intent classification in Sinhala. We also present a novel approach towards unsupervised slot identification using normalized attention scores. This approach can be used for unsupervised slot labelling, data augmentation and to generate data for a new slot in a one-shot way with only one speech recording
|
['Sai Krishna Rallabandi', 'William Zeng', 'Saloni Mittal', 'Akruti Kushwaha', 'Olivia Deng', 'Alan W Black', 'Akshat Gupta']
|
2021-04-03
| null | null | null | null |
['intent-recognition']
|
['natural-language-processing']
|
[ 6.37035072e-01 4.75335330e-01 -5.99802658e-02 -6.82583451e-01
-1.25664485e+00 -4.55269396e-01 7.20734119e-01 1.04486711e-01
-6.95691705e-01 6.22235894e-01 7.67333567e-01 -6.57704353e-01
3.96063179e-01 -5.27617335e-01 -2.95854419e-01 -4.11201119e-01
2.88339585e-01 9.62248445e-01 1.05063714e-01 -3.40910405e-01
3.45693588e-01 3.39998901e-02 -1.69278371e+00 2.35064402e-01
6.35283232e-01 6.41523361e-01 5.69878697e-01 1.05473137e+00
-7.59388566e-01 7.39119291e-01 -4.56079781e-01 5.40266871e-01
-1.05398506e-01 -5.56207716e-01 -1.24071038e+00 1.14956163e-01
5.02793975e-02 -2.01535627e-01 9.69982520e-02 3.32138091e-01
7.23529160e-01 4.44382697e-01 5.62447309e-01 -7.60178745e-01
-3.48440297e-02 8.35987568e-01 1.58072293e-01 1.31564721e-01
6.73720181e-01 -4.13605750e-01 1.01890445e+00 -1.01804078e+00
4.45830643e-01 1.19838595e+00 3.33348334e-01 7.50892162e-01
-9.70900178e-01 -2.52758771e-01 1.89940981e-03 2.00768672e-02
-1.24385297e+00 -1.06848848e+00 4.91891563e-01 1.21051716e-02
1.60415912e+00 3.52918476e-01 3.46979350e-02 7.79774308e-01
-4.04052883e-01 1.07656646e+00 9.15369809e-01 -1.17656946e+00
5.57630956e-01 2.30105564e-01 3.77897620e-01 3.99454713e-01
-5.57753146e-01 -3.38709474e-01 -7.61144578e-01 9.63834897e-02
3.78067762e-01 -2.93592423e-01 1.82399064e-01 3.13129164e-02
-1.06411433e+00 1.01526725e+00 -1.55486062e-01 3.25765878e-01
-3.08744431e-01 -6.86878413e-02 7.68761814e-01 2.65464693e-01
5.92551291e-01 3.05188686e-01 -5.86322963e-01 -6.64927542e-01
-8.95490766e-01 -1.70603514e-01 1.03780448e+00 8.18281293e-01
7.45319664e-01 3.92340302e-01 -2.77600773e-02 1.46451974e+00
5.25894880e-01 4.07840580e-01 1.16206479e+00 -5.83199143e-01
3.68128270e-01 2.63621420e-01 -3.12605202e-02 -5.98613061e-02
-4.98356432e-01 1.63848549e-01 -1.83728516e-01 -2.85134792e-01
8.04175586e-02 -1.70267567e-01 -1.42751980e+00 1.63313746e+00
3.68068218e-01 2.13936582e-01 6.97189987e-01 4.46794480e-01
7.68566370e-01 1.08795536e+00 3.41367036e-01 -2.48980358e-01
1.58087921e+00 -1.15831232e+00 -8.75293970e-01 -7.41201103e-01
1.05923474e+00 -1.01328456e+00 1.30497146e+00 1.45687953e-01
-7.35582888e-01 -5.11623383e-01 -7.71301031e-01 -3.35419029e-01
-6.00118518e-01 1.22503065e-01 6.06688142e-01 1.08268666e+00
-1.15009844e+00 -1.99946687e-01 -6.31354451e-01 -7.46184349e-01
6.31331801e-02 4.74491686e-01 -3.16356063e-01 -7.24685788e-02
-1.08947611e+00 5.75933278e-01 6.32822216e-01 -2.59961337e-01
-5.84923148e-01 -1.40005216e-01 -1.36375654e+00 -7.68169835e-02
3.56251448e-01 1.75212584e-02 1.66897595e+00 -5.93496323e-01
-1.83783507e+00 9.24389780e-01 -5.65744758e-01 -8.06589901e-01
-2.78645426e-01 -3.36721301e-01 -3.38380784e-01 -8.69878829e-02
2.32018247e-01 8.47601235e-01 5.61614931e-01 -8.53698730e-01
-8.20193291e-01 -2.53153712e-01 -4.63938624e-01 7.90335298e-01
-9.98882130e-02 5.15381340e-03 -4.07382965e-01 -5.50971091e-01
3.07331234e-01 -7.94860184e-01 -2.72242725e-02 -6.91416085e-01
-1.18098170e-01 -4.47807312e-01 1.00653791e+00 -7.42923141e-01
1.18838739e+00 -2.05988717e+00 -2.30749130e-01 8.40893015e-02
-4.49755490e-01 4.94706541e-01 -2.31184334e-01 6.40817702e-01
2.70457268e-01 -2.23684520e-01 -2.86341250e-01 -7.05348730e-01
1.99769735e-01 8.60376537e-01 -5.58657408e-01 -1.25972137e-01
2.37395037e-02 8.30451369e-01 -5.46990275e-01 -3.35029066e-01
5.71087122e-01 3.58212978e-01 -5.55396616e-01 3.88392359e-01
-2.38543004e-01 2.42577374e-01 -1.03289545e-01 5.25075972e-01
2.76190609e-01 4.89631504e-01 3.77087861e-01 2.87312567e-01
-3.33959430e-01 1.16730356e+00 -1.12566435e+00 1.85955060e+00
-1.16443086e+00 6.13380134e-01 3.97713743e-02 -1.43572736e+00
1.22766328e+00 7.02801645e-01 1.51883215e-02 -8.55747283e-01
1.26039430e-01 4.83811766e-01 -1.79591566e-01 -9.33632106e-02
8.22256923e-01 -3.56637657e-01 -5.84420085e-01 8.76189351e-01
5.28419673e-01 -2.12974712e-01 -6.94915950e-02 -3.48427892e-02
8.64234626e-01 -2.26762980e-01 6.95541322e-01 -2.77121574e-01
6.51971459e-01 -1.03835098e-01 1.44629627e-01 1.05167449e+00
-2.29017749e-01 5.24442077e-01 -9.07435343e-02 -3.56425762e-01
-1.10659170e+00 -8.27760279e-01 -9.76853073e-02 1.72656751e+00
-2.53079057e-01 -2.02283964e-01 -8.09170723e-01 -4.92633075e-01
-5.35841048e-01 1.01979589e+00 -2.43967220e-01 5.76635711e-02
-4.70802456e-01 -3.21853191e-01 7.86742330e-01 4.23296541e-01
5.00702739e-01 -1.63608372e+00 -3.12138140e-01 5.63236892e-01
-2.65047163e-01 -1.25931025e+00 -3.23295712e-01 7.13296354e-01
-6.45567358e-01 -3.04742694e-01 -5.22249758e-01 -1.23541605e+00
3.23660344e-01 7.65502229e-02 8.91966939e-01 -2.17969999e-01
-2.98552513e-01 4.64354247e-01 -7.22381294e-01 -6.15521550e-01
-6.52870238e-01 4.23595667e-01 3.88061166e-01 -1.44528434e-01
8.47232163e-01 -2.48933300e-01 -1.57318115e-01 1.71060637e-01
-8.50295067e-01 2.85927970e-02 3.91981572e-01 8.00356448e-01
4.00567800e-01 -1.52267724e-01 9.29554582e-01 -9.78619337e-01
5.97496629e-01 -3.23751003e-01 -2.09739029e-01 -2.00434998e-02
-4.19569701e-01 2.56315440e-01 4.78306800e-01 -1.94056928e-02
-1.26954830e+00 4.37528372e-01 -1.02451134e+00 3.57686222e-01
-8.09657633e-01 4.21120822e-01 -2.71813482e-01 2.86209792e-01
5.22364497e-01 6.58176422e-01 2.40174904e-02 -6.77152753e-01
5.18912256e-01 1.50325549e+00 5.15928864e-01 -2.19330236e-01
1.51295781e-01 1.78434432e-01 -8.22684288e-01 -1.57444584e+00
-6.75723076e-01 -1.12323439e+00 -7.73668349e-01 1.26224741e-01
7.16868043e-01 -1.01260781e+00 -2.97007054e-01 5.30807316e-01
-8.25688004e-01 -5.95433235e-01 -4.93353665e-01 3.59989583e-01
-7.68030345e-01 3.66087973e-01 -3.61740023e-01 -1.36194980e+00
-5.36151648e-01 -9.87325490e-01 1.35713613e+00 2.08835140e-01
-5.89095592e-01 -1.18519783e+00 3.63523036e-01 6.94600880e-01
4.56864864e-01 -8.76757681e-01 5.95707893e-01 -1.30970240e+00
-6.22676499e-02 -1.20224863e-01 2.52553280e-02 4.12488908e-01
2.95448780e-01 -8.18201482e-01 -1.34757495e+00 7.51919374e-02
-1.37494981e-01 -5.18231034e-01 5.95755935e-01 2.63296366e-01
3.97103339e-01 -2.82415032e-01 6.38922304e-02 -5.55246323e-03
1.09507024e+00 7.53193498e-01 7.29481399e-01 1.69383198e-01
5.79153657e-01 6.91904187e-01 7.83537865e-01 2.53686130e-01
4.46987063e-01 7.22619295e-01 -2.67566651e-01 9.39072743e-02
-1.22877397e-01 -2.38858491e-01 4.99004900e-01 1.18246782e+00
6.23693347e-01 -3.78499955e-01 -1.19791198e+00 9.97969747e-01
-1.81526804e+00 -5.38596869e-01 2.85605133e-01 2.10823178e+00
8.13215554e-01 9.90079716e-02 8.53704885e-02 3.31182688e-01
2.70357251e-01 7.61801526e-02 -1.29440367e-01 -1.10205948e+00
2.06505120e-01 6.23505116e-01 2.80805290e-01 1.08228397e+00
-1.19640911e+00 1.64806187e+00 6.69120502e+00 8.91348183e-01
-9.54585969e-01 2.35548645e-01 5.78658462e-01 3.36527526e-01
-1.43881395e-01 -1.31381363e-01 -8.06860924e-01 1.72253370e-01
1.56673074e+00 2.14848980e-01 2.16050208e-01 1.04033387e+00
1.71298549e-01 -6.15728557e-01 -6.56940758e-01 9.58700001e-01
1.70762122e-01 -1.25814807e+00 -4.98944409e-02 -1.45173177e-01
3.09102952e-01 3.02708328e-01 -2.55138308e-01 5.38211703e-01
2.88900107e-01 -8.47994208e-01 2.20071644e-01 3.61340307e-02
7.67200172e-01 -9.22401011e-01 9.14275646e-01 4.41378802e-01
-1.24414790e+00 2.45661244e-01 -2.99223810e-01 -3.96935493e-01
3.97486836e-01 4.45597135e-02 -1.56371176e+00 1.69325620e-01
2.10183531e-01 2.28308335e-01 -6.94306567e-02 5.09180486e-01
7.83644021e-02 1.00875628e+00 -7.50439286e-01 -1.74081594e-01
5.72697639e-01 5.82800666e-03 2.69464225e-01 1.36879265e+00
2.35494286e-01 7.73153976e-02 3.68038297e-01 1.70658119e-02
1.30478337e-01 5.98078489e-01 -8.33663762e-01 -2.31509492e-01
5.47790527e-01 8.57071519e-01 -1.08812773e+00 -5.94146907e-01
-3.27316076e-01 1.38619733e+00 5.55824228e-02 -5.34121133e-02
-1.36586770e-01 -4.89248246e-01 7.08358765e-01 -4.57373023e-01
3.03429455e-01 -3.08949530e-01 -2.94294119e-01 -9.74237561e-01
-4.28335726e-01 -7.10000455e-01 2.36397922e-01 -5.95554769e-01
-6.42485261e-01 7.69901216e-01 -1.75625786e-01 -6.12240016e-01
-9.29702461e-01 -5.04337251e-01 -5.58722556e-01 8.90471280e-01
-1.29680443e+00 -9.34914947e-01 2.00796574e-01 1.73683718e-01
1.48297071e+00 -5.51120758e-01 1.43897951e+00 3.09583813e-01
-3.14669997e-01 4.45655137e-01 1.92248866e-01 7.05181211e-02
4.24490303e-01 -1.26688766e+00 9.53489661e-01 7.73060501e-01
6.91781044e-01 4.30103391e-01 7.40809441e-01 -4.37801838e-01
-9.53634083e-01 -7.69305944e-01 1.39237070e+00 -4.59053278e-01
4.15119410e-01 -7.83209682e-01 -8.94966424e-01 6.84407473e-01
2.28802219e-01 -5.86783528e-01 1.21002722e+00 2.42223084e-01
-5.84740341e-02 3.18177879e-01 -9.02243495e-01 5.22547603e-01
7.80297160e-01 -9.69115257e-01 -9.00128603e-01 2.58562118e-01
7.54775524e-01 -2.98246145e-01 -3.31594199e-01 2.46721990e-02
4.81212705e-01 -6.10722959e-01 7.50007451e-01 -2.70452082e-01
-1.58186466e-01 -1.05379984e-01 -4.08783764e-01 -1.05262625e+00
2.12128788e-01 -5.60558736e-01 3.47236723e-01 1.19479287e+00
5.33221304e-01 -5.88184118e-01 8.50103080e-01 2.94475973e-01
-3.09027374e-01 -1.56669825e-01 -1.26962924e+00 -4.06236291e-01
-2.57275432e-01 -1.02977812e+00 2.03680843e-01 5.32554686e-01
3.15539747e-01 9.35617983e-01 -5.70562422e-01 -2.00304054e-02
1.52187496e-01 -1.58593282e-01 5.99673271e-01 -8.58302832e-01
-9.71447378e-02 7.87593797e-02 -4.18683559e-01 -1.28527308e+00
2.40928039e-01 -7.04506516e-01 7.12595522e-01 -1.57309711e+00
-3.36907655e-01 -6.24270499e-01 -7.56061971e-02 7.46123791e-01
1.26505643e-01 6.07583165e-01 -1.94648206e-02 1.11981787e-01
-6.04143560e-01 5.77398121e-01 1.60384193e-01 1.93949074e-01
-4.62825894e-01 4.91801612e-02 -5.32671094e-01 5.43713331e-01
9.23206627e-01 -2.84506291e-01 -6.63665831e-01 -9.03197527e-02
-3.41420263e-01 -1.55379742e-01 -3.55949610e-01 -1.06348562e+00
5.17167039e-02 2.06915975e-01 -3.58696193e-01 -6.64823353e-01
5.00485182e-01 -6.21369302e-01 -4.18969333e-01 2.82001793e-01
-4.15843636e-01 -4.96228933e-01 5.15629292e-01 2.35280260e-01
-3.08929920e-01 -5.75743318e-01 6.00922048e-01 -1.80813402e-01
-1.36490178e+00 -3.63480031e-01 -1.28819597e+00 -8.59168842e-02
6.84443533e-01 -3.39428693e-01 6.78600669e-02 -6.87549353e-01
-9.78026688e-01 3.63867469e-02 -1.11975996e-02 7.17849016e-01
5.02532601e-01 -9.85241175e-01 -3.30929458e-01 7.59516060e-01
3.34962547e-01 -1.22127935e-01 3.76811773e-01 2.66302824e-01
-6.06975555e-01 1.18165684e+00 -3.53902616e-02 -4.51839536e-01
-1.42332399e+00 -2.23849118e-02 5.23066334e-02 -1.95512444e-01
-4.58227009e-01 9.02135313e-01 1.85718581e-01 -1.14812422e+00
3.78697604e-01 -2.74199456e-01 -4.34182763e-01 1.85357869e-01
7.77283013e-01 -1.53392449e-01 3.10194492e-01 -9.47687864e-01
-4.17245895e-01 9.72589627e-02 -3.37864101e-01 -6.35666907e-01
1.11686063e+00 -5.70704877e-01 2.92427510e-01 8.63929451e-01
1.18997741e+00 -9.70423147e-02 -8.26992989e-01 -2.82357395e-01
2.32383326e-01 7.69531280e-02 1.86929867e-01 -7.28211105e-01
-1.00210801e-01 8.21546435e-01 7.77922869e-01 3.28414202e-01
9.03337657e-01 2.34799534e-01 9.45867360e-01 6.96199000e-01
2.21760079e-01 -1.54878139e+00 -1.87854052e-01 1.08307517e+00
4.37746853e-01 -1.42015350e+00 -6.29168332e-01 -9.67006460e-02
-8.54894519e-01 8.05891395e-01 2.52074957e-01 1.96553290e-01
4.85786587e-01 3.51288497e-01 6.64670765e-01 1.44762956e-02
-6.88763082e-01 -7.09312141e-01 1.79820210e-01 8.21795762e-01
6.40572608e-01 1.33640498e-01 -3.19895774e-01 2.18371853e-01
-4.52363819e-01 -3.54565799e-01 4.54923213e-01 1.19974279e+00
-1.03410482e+00 -1.57369375e+00 -3.35890263e-01 4.10908282e-01
-3.24429989e-01 -4.95867878e-01 -2.70516187e-01 4.25207436e-01
-3.74831498e-01 9.38650966e-01 4.50911343e-01 -1.99135542e-01
1.14808120e-01 9.88549590e-01 -2.00440124e-01 -1.36162376e+00
-1.20042376e-01 4.99031544e-01 5.38349688e-01 -3.71466279e-01
-3.95069480e-01 -7.05314815e-01 -1.55783796e+00 3.82393539e-01
-2.87091345e-01 4.46172357e-01 9.44757819e-01 1.37965405e+00
3.15231115e-01 4.94368345e-01 5.68713427e-01 -8.46559703e-01
1.21139184e-01 -1.06641889e+00 -4.97650832e-01 -2.22444028e-01
2.18262553e-01 -3.50250572e-01 -1.44032165e-01 1.76929370e-01]
|
[14.065542221069336, 6.912117958068848]
|
d87e2df1-1eef-442d-a1c4-22a13a19891f
|
mga-medical-generalist-agent-through-text
|
2303.08562
| null |
https://arxiv.org/abs/2303.08562v1
|
https://arxiv.org/pdf/2303.08562v1.pdf
|
MGA: Medical generalist agent through text-guided knowledge transformation
|
Multi-modal representation methods have achieved advanced performance in medical applications by extracting more robust features from multi-domain data. However, existing methods usually need to train additional branches for downstream tasks, which may increase the model complexities in clinical applications as well as introduce additional human inductive bias. Besides, very few studies exploit the rich clinical knowledge embedded in clinical daily reports. To this end, we propose a novel medical generalist agent, MGA, that can address three kinds of common clinical tasks via clinical reports knowledge transformation. Unlike the existing methods, MGA can easily adapt to different tasks without specific downstream branches when their corresponding annotations are missing. More importantly, we are the first attempt to use medical professional language guidance as a transmission medium to guide the agent's behavior. The proposed method is implemented on four well-known X-ray open-source datasets, MIMIC-CXR, CheXpert, MIMIC-CXR-JPG, and MIMIC-CXR-MS. Promising results are obtained, which validate the effectiveness of our proposed MGA. Code is available at: https://github.com/SZUHvern/MGA
|
['Shanshan Wang', 'Rui Yang', 'Mingtong Dai', 'Cheng Li', 'Hao Yang', 'Weijian Huang']
|
2023-03-15
| null | null | null | null |
['clinical-knowledge']
|
['miscellaneous']
|
[ 4.52650078e-02 1.77805498e-01 -5.00459611e-01 -4.30359900e-01
-1.26062548e+00 -3.86605531e-01 3.76535147e-01 4.49520856e-01
-2.91018516e-01 8.40260148e-01 4.78451043e-01 -4.88420427e-01
-5.25957346e-01 -7.84026206e-01 -5.12125075e-01 -7.03718483e-01
1.86594471e-01 6.41487598e-01 -4.58187796e-02 -3.04563679e-02
-1.57692358e-01 1.78965852e-01 -7.61425674e-01 5.33502579e-01
9.07115817e-01 5.06875992e-01 3.27992439e-01 4.90837008e-01
3.37661743e-01 1.06167781e+00 -2.65932441e-01 -3.62302303e-01
-8.38395953e-02 -3.49795610e-01 -9.44076240e-01 -1.49816275e-01
-3.49752873e-01 -2.22805962e-01 -5.09427845e-01 7.74144828e-01
8.33489001e-01 -1.33996713e-03 6.19463742e-01 -8.61895382e-01
-7.55206585e-01 7.52909243e-01 -4.82349783e-01 2.38430664e-01
5.26258111e-01 2.90265262e-01 7.29187906e-01 -3.65933776e-01
9.07043874e-01 8.16874623e-01 6.35322928e-01 8.14987361e-01
-7.94039249e-01 -7.03208089e-01 7.45547414e-02 2.19597697e-01
-1.20308697e+00 -4.10304256e-02 6.59381390e-01 -5.66742659e-01
5.92368066e-01 5.39423108e-01 3.24153662e-01 1.54693949e+00
2.64620274e-01 8.56142759e-01 1.22532475e+00 -1.32906750e-01
1.63357809e-01 1.61854327e-01 2.05338210e-01 7.13972867e-01
-1.66939832e-02 9.91139114e-02 -3.32123816e-01 -6.01893842e-01
6.50065422e-01 2.43419766e-01 -6.86273575e-01 1.62853062e-01
-1.54037130e+00 8.61067355e-01 5.01731992e-01 4.77890283e-01
-5.86941957e-01 -1.30283013e-01 5.20982921e-01 8.98207054e-02
4.81375962e-01 3.27842712e-01 -4.68197763e-01 -1.47858679e-01
-3.17041010e-01 1.73698232e-01 4.84880388e-01 9.51134324e-01
1.69712782e-01 -5.88726103e-01 -4.24147576e-01 8.92242134e-01
4.26062286e-01 3.27965826e-01 9.95069325e-01 -7.33618855e-01
4.02022570e-01 5.45656919e-01 -1.14202425e-01 -9.34606731e-01
-6.60670459e-01 -5.57479262e-01 -1.04915702e+00 -4.74188536e-01
3.69723067e-02 -1.75960556e-01 -8.84235382e-01 1.56507468e+00
5.03298163e-01 2.38011450e-01 2.39575088e-01 9.87085938e-01
1.14592695e+00 4.32384998e-01 3.39974642e-01 -2.18372107e-01
1.63007951e+00 -8.37468207e-01 -7.51153171e-01 -6.86014891e-02
1.08551669e+00 -7.78693736e-01 1.10676408e+00 3.16686213e-01
-8.37893128e-01 -1.89204305e-01 -7.12505817e-01 1.38239488e-01
-2.78121233e-01 2.44757175e-01 1.01973283e+00 1.95204362e-01
-6.32644594e-01 2.85191298e-01 -1.00863528e+00 -2.87517160e-01
4.74273980e-01 2.44974867e-01 -3.77196312e-01 -3.24770093e-01
-1.36565292e+00 7.75665343e-01 2.30159000e-01 1.18543319e-02
-8.37978482e-01 -1.07942319e+00 -8.06610763e-01 -3.56734514e-01
5.19091964e-01 -9.81480002e-01 1.52959955e+00 -3.05974513e-01
-1.43451500e+00 9.30939138e-01 1.00540377e-01 2.08248291e-03
6.38257086e-01 -1.20961018e-01 -4.87604022e-01 2.52318531e-01
2.58359730e-01 3.43421698e-01 1.74118266e-01 -1.18602359e+00
-4.45535868e-01 -3.86509657e-01 2.78585672e-01 6.01013340e-02
-2.30028361e-01 -4.39186692e-02 -5.27264178e-01 -9.97543871e-01
-2.48884410e-01 -9.05113280e-01 -7.56392956e-01 -1.31654605e-01
-6.51890039e-01 -2.46272370e-01 3.18782717e-01 -5.21600068e-01
1.35494828e+00 -2.08275986e+00 -7.74532743e-03 1.23372011e-01
3.64680082e-01 1.72043324e-01 -4.82623205e-02 2.32917190e-01
-2.55291164e-01 -2.35640556e-02 -3.17246675e-01 -1.60914250e-02
-3.31614614e-01 1.35216862e-01 1.06838234e-01 3.51341009e-01
-3.74700837e-02 9.16855693e-01 -1.10872936e+00 -6.43867433e-01
1.80544868e-01 4.36333001e-01 -6.23821855e-01 2.28425115e-01
-1.70870468e-01 7.99649239e-01 -1.10832989e+00 6.77067459e-01
3.40713441e-01 -8.88029754e-01 1.81596845e-01 -3.56917560e-01
2.10522994e-01 2.64996767e-01 -7.03348339e-01 1.94259214e+00
-7.24299431e-01 -3.24413478e-02 -9.45721716e-02 -1.12330282e+00
4.75309342e-01 7.12631047e-01 9.26177382e-01 -5.02333701e-01
2.62991458e-01 6.95300773e-02 9.90275759e-03 -9.48676884e-01
2.94523314e-02 -3.49601895e-01 -8.09654519e-02 5.79296231e-01
-2.83141226e-01 -1.48419272e-02 -8.75162985e-03 2.47035593e-01
1.37444103e+00 -9.02403668e-02 5.97805917e-01 7.05595315e-02
5.53038359e-01 1.94598719e-01 6.67255640e-01 6.92304254e-01
-1.65259749e-01 5.96570075e-01 1.37642428e-01 -3.14623863e-01
-2.83708632e-01 -1.01554954e+00 -4.15608495e-01 7.74715483e-01
2.15602573e-02 -5.33517122e-01 -3.52107793e-01 -8.59273911e-01
-1.38462842e-01 6.90950334e-01 -7.51105964e-01 -2.77441472e-01
-3.59140337e-01 -1.25454295e+00 4.67077047e-01 5.83576143e-01
8.83619562e-02 -1.08529234e+00 -5.16152918e-01 4.19524819e-01
-5.50949037e-01 -1.26447487e+00 -5.70563436e-01 -2.48391926e-02
-8.22378218e-01 -1.34402406e+00 -9.53492284e-01 -4.90135580e-01
7.19836950e-01 -1.22049928e-01 9.75115955e-01 4.94950823e-02
-6.04395628e-01 7.31507063e-01 -6.29334867e-01 -4.07119483e-01
-6.83595240e-01 1.77061886e-01 -1.35655329e-01 -8.38187411e-02
3.48169476e-01 -3.63975853e-01 -8.14821184e-01 2.36131012e-01
-9.53875244e-01 3.47409606e-01 7.74841189e-01 8.26342285e-01
8.04576933e-01 -3.10819149e-01 8.60209584e-01 -1.37116134e+00
8.57001841e-01 -7.35949576e-01 -2.63867937e-02 2.41714090e-01
-6.67504251e-01 -1.48218185e-01 5.90321779e-01 -4.55412090e-01
-1.19390166e+00 -7.40241036e-02 -3.07746559e-01 -2.62562305e-01
-3.36624473e-01 9.11100149e-01 1.70976549e-01 3.31218570e-01
7.50086188e-01 2.55399138e-01 -3.34914811e-02 -3.88748527e-01
1.60147443e-01 9.93137479e-01 3.90219361e-01 -5.98160625e-01
5.72953522e-01 5.93052745e-01 -1.49715871e-01 -4.34298933e-01
-1.05437326e+00 -5.62355459e-01 -2.14725420e-01 -4.98949215e-02
8.32816601e-01 -8.69092643e-01 -7.55281329e-01 1.32153332e-01
-1.09366453e+00 -2.44421184e-01 -2.10568696e-01 8.50526214e-01
-5.01480699e-01 3.19050133e-01 -9.25872564e-01 -2.35169247e-01
-6.35413051e-01 -1.39120054e+00 1.00337791e+00 3.12383380e-03
-3.17526221e-01 -1.24617910e+00 1.49432927e-01 5.72080910e-01
2.89038002e-01 3.38226765e-01 1.09467041e+00 -6.00029290e-01
-4.10771906e-01 -2.35191181e-01 -1.16188362e-01 2.48913988e-02
6.42935812e-01 -3.34889650e-01 -6.87347472e-01 -1.77766487e-01
6.65926039e-02 -3.41086060e-01 6.03238583e-01 5.73884487e-01
1.49871337e+00 -2.08718717e-01 -7.25006521e-01 5.79232514e-01
1.09025192e+00 4.95979667e-01 3.47323149e-01 4.20267403e-01
6.19997263e-01 1.98863044e-01 8.78695130e-01 6.22097909e-01
6.30579472e-01 5.23289502e-01 1.78061724e-01 -3.51408005e-01
6.32186309e-02 5.00186607e-02 -2.21835705e-03 1.22818398e+00
-2.85204321e-01 -1.54787093e-01 -1.15572941e+00 4.88051921e-01
-1.87254047e+00 -5.69787443e-01 3.86554701e-03 1.66026282e+00
1.30729771e+00 -1.46270946e-01 -2.92244047e-01 -2.18705237e-01
3.79469186e-01 -6.48612604e-02 -4.94817704e-01 1.41965881e-01
3.65115374e-01 2.28191450e-01 2.14819938e-01 2.75363982e-01
-1.01636624e+00 5.68660617e-01 5.79276180e+00 7.66094089e-01
-1.13803327e+00 4.96471614e-01 6.84899747e-01 -6.63959906e-02
-3.69607627e-01 -4.29884285e-01 -4.37730581e-01 5.66449344e-01
6.72338724e-01 -3.63570422e-01 -4.20406498e-02 8.30217898e-01
4.45344925e-01 5.75347506e-02 -1.33443987e+00 1.02882588e+00
-9.12093073e-02 -1.51909316e+00 7.68165663e-02 -7.51859229e-03
3.71256948e-01 2.74405051e-02 4.53582704e-02 2.93921798e-01
3.52393478e-01 -8.65070939e-01 8.70503783e-02 7.30597734e-01
9.29954171e-01 -3.68544489e-01 8.89470577e-01 1.61772013e-01
-8.56501222e-01 2.51560688e-01 6.26884308e-03 3.60506058e-01
3.11948448e-01 6.56002104e-01 -1.06186819e+00 1.17526388e+00
5.31664908e-01 9.68797326e-01 -4.24959213e-01 8.73055279e-01
-5.66693284e-02 7.42396593e-01 -1.96376234e-01 1.46359369e-01
1.54273972e-01 8.77404679e-03 3.57693672e-01 1.24376214e+00
3.22732657e-01 6.04669511e-01 3.16094458e-01 5.86225629e-01
-4.61880602e-02 3.12192649e-01 -6.04234397e-01 -5.81071451e-02
1.74906835e-01 1.20895576e+00 -4.35704648e-01 -4.33729678e-01
-6.33938849e-01 6.95423901e-01 -8.61592442e-02 3.01665872e-01
-1.08266342e+00 -9.78756845e-02 2.94192493e-01 2.59444505e-01
-3.36562186e-01 3.26716423e-01 -7.29836524e-02 -1.21247172e+00
-1.87347367e-01 -1.12216854e+00 8.00749421e-01 -7.32326508e-01
-1.57034159e+00 7.15909719e-01 4.06916738e-02 -1.49351060e+00
-2.64581740e-01 -5.73748410e-01 -2.52688020e-01 4.58827585e-01
-1.58657801e+00 -1.26326716e+00 -3.92965674e-01 8.32749963e-01
4.52252746e-01 -3.17518890e-01 1.17003071e+00 8.56977642e-01
-5.48257053e-01 5.91068149e-01 -9.37385857e-02 2.48375639e-01
9.80896533e-01 -9.53120530e-01 -2.44364813e-01 2.04177648e-01
-3.66564125e-01 7.50730455e-01 3.39594245e-01 -5.94750464e-01
-1.25501156e+00 -1.24270010e+00 3.58466327e-01 -5.27842164e-01
6.86534822e-01 2.08711967e-01 -9.65646446e-01 9.63076413e-01
3.22379060e-02 1.13216266e-01 1.17165864e+00 7.25583136e-02
-1.44285575e-01 6.67559654e-02 -1.04022515e+00 5.45917034e-01
1.00705361e+00 -4.09392327e-01 -6.63623273e-01 8.77390742e-01
7.09184229e-01 -7.41631806e-01 -1.33333111e+00 6.38034105e-01
1.32853314e-01 -3.45285624e-01 9.93427694e-01 -7.74480224e-01
7.30199158e-01 -2.26157323e-01 -1.60520582e-03 -1.34036636e+00
-2.41773322e-01 -4.07417178e-01 1.26662990e-02 8.85329485e-01
5.25642931e-01 -8.64969552e-01 5.15328586e-01 4.51978594e-01
-1.68184072e-01 -1.10198796e+00 -8.87245595e-01 -3.65474135e-01
-8.37845355e-02 -6.07617557e-01 3.69058967e-01 1.46613753e+00
1.93207085e-01 2.37033948e-01 -1.97074905e-01 3.59960288e-01
4.52457935e-01 1.83551416e-01 4.75270152e-01 -9.56145704e-01
-7.34043360e-01 -1.68923989e-01 -1.60873279e-01 -7.78796554e-01
-3.33665945e-02 -1.26111054e+00 -1.08027726e-01 -1.68263650e+00
6.71902835e-01 -7.58946002e-01 -5.31898081e-01 8.10149372e-01
-4.08792883e-01 -4.07591164e-02 -1.64878741e-01 1.97426602e-01
-4.98024642e-01 4.73357409e-01 1.66863358e+00 -2.87556976e-01
-9.71843898e-02 1.93884686e-01 -9.15555358e-01 8.87428164e-01
8.05033863e-01 -7.22905219e-01 -4.47041720e-01 -4.04507995e-01
1.62352070e-01 3.63259137e-01 4.03195739e-01 -6.38859212e-01
1.66066185e-01 -3.39892596e-01 -2.65226845e-04 -3.50303411e-01
1.65344045e-01 -7.36029565e-01 2.36723453e-01 6.15741253e-01
-3.81997824e-01 -4.69799936e-02 2.31444865e-01 5.31459391e-01
-2.71494508e-01 -2.68213212e-01 6.01331830e-01 -3.70206863e-01
-3.35522532e-01 4.68758106e-01 -3.37295204e-01 7.60603473e-02
1.27427983e+00 3.00913066e-01 -3.69871020e-01 -2.21243039e-01
-8.66896510e-01 3.42113435e-01 -6.32466972e-02 5.66153884e-01
6.72454596e-01 -1.19126773e+00 -8.20284247e-01 -2.11734876e-01
3.67725581e-01 3.02847087e-01 6.45531237e-01 1.17499459e+00
-4.05418396e-01 4.11984503e-01 2.40719944e-01 -6.16231561e-01
-1.19370592e+00 6.05694711e-01 3.74693066e-01 -6.63382292e-01
-9.04811859e-01 4.67226684e-01 4.43611532e-01 -6.06408775e-01
-1.94335319e-02 -4.24996823e-01 -1.57743096e-02 -5.15900142e-02
5.36636591e-01 -7.57826120e-02 1.06969871e-01 -2.56052971e-01
-5.02193391e-01 5.51761150e-01 -5.65701485e-01 1.95316672e-01
1.58067513e+00 3.25336605e-02 9.59344953e-02 2.94463158e-01
9.45452929e-01 -9.99464914e-02 -4.96581137e-01 -3.45586121e-01
-1.73562124e-01 -2.13832542e-01 3.19621637e-02 -1.03093517e+00
-1.22164392e+00 5.77831626e-01 5.92036009e-01 -2.67140836e-01
1.17051542e+00 4.14597094e-01 6.74896777e-01 3.75679970e-01
4.02741224e-01 -7.80861437e-01 1.49435416e-01 -4.77585085e-02
9.84027267e-01 -1.44548547e+00 1.59795448e-01 -5.63156426e-01
-8.24014664e-01 7.39546120e-01 6.39962375e-01 3.60307097e-01
8.20937157e-01 4.15909410e-01 4.73640800e-01 -4.66729462e-01
-7.45884240e-01 9.76672843e-02 3.28231491e-02 5.68322301e-01
6.60199761e-01 3.05537015e-01 -3.77649784e-01 9.67379153e-01
7.01186508e-02 4.80306774e-01 5.62722325e-01 1.06895459e+00
1.72175065e-01 -1.08088279e+00 -2.60231316e-01 5.97910583e-01
-7.32002616e-01 -1.33493468e-01 9.47258994e-02 9.26046610e-01
-1.10258654e-01 6.77353382e-01 -5.94338357e-01 -2.03599751e-01
5.33070982e-01 -1.96978584e-01 3.23282421e-01 -8.29304576e-01
-5.01169324e-01 -2.18056366e-02 -1.33863986e-01 -5.28961301e-01
-6.53065085e-01 -5.41206419e-01 -1.51160109e+00 1.28733948e-01
-6.03816099e-02 1.78316832e-01 2.39590034e-01 7.94392228e-01
4.99805897e-01 1.17079914e+00 3.58871520e-01 -1.62843466e-01
-4.91735905e-01 -9.60357547e-01 -2.57497460e-01 5.76010883e-01
1.71210200e-01 -6.77908123e-01 -4.93175723e-02 2.59336263e-01]
|
[14.917399406433105, -1.8517369031906128]
|
167b0cac-d140-40e3-8a85-7e11da9817e5
|
constrained-clustering-and-its-application-to
| null | null |
http://openaccess.thecvf.com/content_cvpr_2013/html/Wu_Constrained_Clustering_and_2013_CVPR_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2013/papers/Wu_Constrained_Clustering_and_2013_CVPR_paper.pdf
|
Constrained Clustering and Its Application to Face Clustering in Videos
|
In this paper, we focus on face clustering in videos. Given the detected faces from real-world videos, we partition all faces into K disjoint clusters. Different from clustering on a collection of facial images, the faces from videos are organized as face tracks and the frame index of each face is also provided. As a result, many pairwise constraints between faces can be easily obtained from the temporal and spatial knowledge of the face tracks. These constraints can be effectively incorporated into a generative clustering model based on the Hidden Markov Random Fields (HMRFs). Within the HMRF model, the pairwise constraints are augmented by label-level and constraint-level local smoothness to guide the clustering process. The parameters for both the unary and the pairwise potential functions are learned by the simulated field algorithm, and the weights of constraints can be easily adjusted. We further introduce an efficient clustering framework specially for face clustering in videos, considering that faces in adjacent frames of the same face track are very similar. The framework is applicable to other clustering algorithms to significantly reduce the computational cost. Experiments on two face data sets from real-world videos demonstrate the significantly improved performance of our algorithm over state-of-theart algorithms.
|
['Bao-Gang Hu', 'Baoyuan Wu', 'Yifan Zhang', 'Qiang Ji']
|
2013-06-01
| null | null | null |
cvpr-2013-6
|
['face-clustering']
|
['computer-vision']
|
[-1.03425540e-01 -1.61838755e-01 -1.09126136e-01 -6.60597742e-01
-4.15993363e-01 -3.32008988e-01 3.50021958e-01 -4.29062396e-01
-1.87251836e-01 3.84044200e-01 -8.60763267e-02 4.64512318e-01
-3.19983929e-01 -4.77830797e-01 -6.87482834e-01 -1.26263213e+00
-2.28125602e-01 3.38887364e-01 2.09945485e-01 3.09211344e-01
7.88293555e-02 4.86420900e-01 -1.75439107e+00 2.80892938e-01
6.21903598e-01 9.62740481e-01 5.92174649e-01 2.52659410e-01
-1.77278355e-01 7.34234989e-01 -2.34834284e-01 -1.48852319e-01
1.80100054e-01 -3.15981358e-01 -6.24946952e-01 8.33670676e-01
3.86134714e-01 -1.36592786e-03 -3.79390806e-01 1.16296363e+00
2.15615109e-01 4.32935387e-01 6.50946617e-01 -1.48578727e+00
-3.67954761e-01 2.41158083e-01 -8.31941843e-01 -5.23043759e-02
2.84092069e-01 -2.62984246e-01 6.11174881e-01 -9.13358927e-01
7.85072863e-01 1.42168641e+00 2.75968999e-01 7.02143073e-01
-1.05405593e+00 -8.03652763e-01 6.37027204e-01 4.76356179e-01
-1.77056634e+00 -7.13683903e-01 8.29063773e-01 -6.33096099e-01
3.05014342e-01 1.52142242e-01 5.34597635e-01 5.53186297e-01
-9.92672741e-02 6.01314545e-01 5.39389014e-01 -3.62421393e-01
3.09932888e-01 -2.63058953e-02 -1.77825391e-01 8.80507171e-01
-9.30991992e-02 -2.54875332e-01 -5.04815459e-01 -3.22486490e-01
8.58973086e-01 4.56626356e-01 -4.86130387e-01 -6.22921646e-01
-1.00466180e+00 8.24401319e-01 2.45161578e-01 3.13717455e-01
-2.74318099e-01 -3.96488979e-02 -1.04444064e-01 -9.80072990e-02
3.58717620e-01 -5.12962997e-01 -2.37036467e-01 4.68871117e-01
-9.82192576e-01 1.26981729e-04 6.02828264e-01 1.25233412e+00
1.19253111e+00 -2.81257808e-01 -1.34962037e-01 8.01264107e-01
7.93157995e-01 4.63623822e-01 -4.34341356e-02 -1.24437022e+00
1.81651995e-01 4.21204746e-01 1.22775361e-01 -1.29498756e+00
-1.84169449e-02 -1.45245437e-02 -9.23444569e-01 -1.74884424e-01
3.18417698e-01 -7.48096630e-02 -9.32785213e-01 1.82288992e+00
6.90944552e-01 8.94450545e-01 -2.67281979e-01 8.28879058e-01
6.34448469e-01 6.75347149e-01 -1.30794704e-01 -9.28199768e-01
1.17516959e+00 -7.01128960e-01 -1.06116569e+00 2.01521203e-01
2.04829350e-01 -8.43121767e-01 2.43753865e-01 2.34597966e-01
-1.04479718e+00 -6.40985548e-01 -5.60089350e-01 3.54243964e-01
1.37531146e-01 2.60384589e-01 4.19071198e-01 3.53580087e-01
-1.16222811e+00 2.97094643e-01 -1.03934598e+00 -1.87056556e-01
4.68029201e-01 7.26183474e-01 -4.23539937e-01 -3.99850965e-01
-7.09180057e-01 2.70284295e-01 2.10161999e-01 5.59740365e-01
-9.70940769e-01 -3.15028042e-01 -8.05636525e-01 -8.72989818e-02
5.60971498e-01 -3.17101359e-01 9.08824027e-01 -1.10203934e+00
-1.27575862e+00 7.49717653e-01 -8.09445679e-01 1.71118498e-01
8.57002512e-02 1.13851391e-01 -4.86039907e-01 5.06968975e-01
1.50509775e-01 6.81918204e-01 1.21152544e+00 -1.55504334e+00
-6.92693770e-01 -3.33350867e-01 -2.28614807e-01 1.37567356e-01
-3.36465031e-01 3.59601825e-01 -1.19745278e+00 -5.06746173e-01
2.18051255e-01 -1.07983768e+00 -3.49303663e-01 1.26552716e-01
-1.25802949e-01 -3.99043649e-01 1.16851175e+00 -5.33628166e-01
1.26725340e+00 -2.40770960e+00 4.53569740e-01 4.65089232e-01
1.71171173e-01 -1.01942485e-02 6.31558290e-03 1.72426715e-01
5.71284220e-02 -1.58996582e-01 -2.09512025e-01 -4.48889256e-01
-3.23835969e-01 4.66895163e-01 1.37897417e-01 7.67841697e-01
1.20663419e-01 4.16392714e-01 -8.40073168e-01 -8.65879238e-01
1.83103248e-01 7.74817646e-01 -7.16862082e-01 2.81929463e-01
-1.33312091e-01 7.86606133e-01 -6.05367541e-01 5.12616098e-01
1.09032798e+00 -3.35269928e-01 5.31820357e-01 -1.72207296e-01
-6.18725531e-02 -3.62352252e-01 -1.53123128e+00 1.68708372e+00
5.66140637e-02 8.36880207e-02 5.49968600e-01 -1.00506842e+00
6.75595164e-01 5.21000683e-01 8.08222711e-01 5.12145683e-02
-3.52779105e-02 -1.90447122e-01 -8.66781846e-02 -5.52079737e-01
1.54025201e-02 1.03214122e-01 3.46266299e-01 2.41299659e-01
2.62923747e-01 5.01534879e-01 3.82282645e-01 2.12769717e-01
5.71188271e-01 6.42280430e-02 -1.34403706e-01 -3.98057580e-01
8.94379973e-01 -4.35267001e-01 1.01414764e+00 2.66724080e-01
-6.22701012e-02 5.64950585e-01 1.07930981e-01 -2.60358930e-01
-4.59762454e-01 -8.94854486e-01 -2.21931398e-01 9.03612852e-01
2.86301106e-01 -6.13065541e-01 -1.12434816e+00 -5.59815168e-01
-2.24455163e-01 -2.15166152e-01 -6.10435486e-01 2.73201138e-01
-5.82580984e-01 -7.69645512e-01 -1.97828233e-01 3.61314714e-01
3.94608557e-01 -9.50212240e-01 -9.35702026e-02 2.06359699e-01
-3.20401609e-01 -1.29675722e+00 -7.89244890e-01 -2.27440804e-01
-6.67394996e-01 -1.21852267e+00 -5.99161923e-01 -1.29232597e+00
1.12570155e+00 6.33049667e-01 7.17963517e-01 5.21322846e-01
-2.06693247e-01 5.63811958e-01 -3.86296958e-01 6.79840669e-02
1.02140307e-02 -4.83322918e-01 3.43083978e-01 8.02807152e-01
3.70547801e-01 -4.13112193e-01 -4.99767035e-01 7.03032911e-01
-8.83527696e-01 -2.16565892e-01 1.81624398e-01 6.65058792e-01
8.73208463e-01 6.97377026e-01 1.39193460e-01 -6.47302449e-01
-1.91619903e-01 -5.71660936e-01 -6.31666541e-01 3.72447461e-01
1.73918530e-02 -3.35550874e-01 3.71714383e-01 -6.29717827e-01
-1.28761339e+00 6.93109453e-01 3.67743999e-01 -9.39562857e-01
-2.78552741e-01 1.93804920e-01 -6.51507378e-01 -2.48398528e-01
-1.94564730e-01 6.69551343e-02 7.36579597e-02 -4.22756881e-01
2.59193331e-01 5.11572480e-01 4.26831245e-01 -5.89212060e-01
9.32745397e-01 8.62317801e-01 -4.98203114e-02 -1.07540977e+00
-6.22530222e-01 -6.80416703e-01 -1.08477783e+00 -5.74769080e-01
1.09983313e+00 -1.06665695e+00 -8.87810111e-01 5.03136992e-01
-1.05720365e+00 3.83816399e-02 2.56109923e-01 6.27400994e-01
-4.79537576e-01 6.07658088e-01 -5.95829546e-01 -9.66455340e-01
3.98853183e-01 -1.31509626e+00 1.03772438e+00 3.60432684e-01
3.04421812e-01 -1.10143101e+00 -1.34182513e-01 1.77969038e-01
-1.99983761e-01 6.05610684e-02 5.88306785e-01 -1.25672773e-01
-8.99346292e-01 1.37136593e-01 1.26696333e-01 2.87862778e-01
3.70804816e-01 4.29266155e-01 -8.61256003e-01 -5.06183922e-01
2.38833487e-01 2.07421526e-01 7.62806773e-01 8.15611184e-01
1.16568398e+00 -3.27531606e-01 -8.44698191e-01 5.91563046e-01
1.22547889e+00 4.20238286e-01 5.26295662e-01 -2.91554868e-01
9.98859107e-01 8.46952736e-01 6.48283541e-01 5.34515142e-01
2.79443681e-01 8.67142379e-01 2.10477740e-01 1.12909436e-01
2.35428169e-01 -9.79487449e-02 5.47136128e-01 9.49848831e-01
-2.89710581e-01 -2.07657784e-01 -5.89870453e-01 2.69131839e-01
-2.11274481e+00 -1.31428456e+00 -2.38105521e-01 2.19710088e+00
4.25021231e-01 -5.36968768e-01 1.62079066e-01 3.71950082e-02
1.43362534e+00 3.98840895e-03 -2.74243772e-01 4.45963115e-01
1.74926221e-02 -1.41231701e-01 5.18126078e-02 5.69205105e-01
-1.30549514e+00 7.99223542e-01 6.17465401e+00 8.41609001e-01
-6.47469699e-01 6.90907016e-02 5.79086185e-01 -2.62667358e-01
8.23817030e-03 1.34838149e-01 -1.13311470e+00 7.60152578e-01
5.18431008e-01 6.17090389e-02 6.90664113e-01 5.83754361e-01
3.24077934e-01 -3.94997634e-02 -1.05551875e+00 1.16183829e+00
1.41557515e-01 -1.06123507e+00 -4.68203090e-02 4.41396266e-01
8.85791779e-01 -3.60671848e-01 4.98829111e-02 -1.95730910e-01
1.90778151e-01 -8.50403190e-01 5.37488282e-01 6.63255334e-01
6.21224105e-01 -1.03054035e+00 3.88438553e-01 3.18590194e-01
-1.70094037e+00 -1.68031633e-01 -6.92290604e-01 1.17192842e-01
2.98419863e-01 6.41160846e-01 -2.56548911e-01 5.67434072e-01
9.92718995e-01 1.10391474e+00 -2.77210981e-01 8.95836115e-01
7.52217090e-03 3.02570403e-01 -3.74677926e-01 4.64672118e-01
6.99123181e-03 -6.75739229e-01 2.60485262e-01 8.31804097e-01
4.20665473e-01 5.59325397e-01 6.66817546e-01 5.63115537e-01
1.19421251e-01 1.78503796e-01 -3.95047694e-01 2.06659883e-01
5.40012896e-01 1.56490982e+00 -9.66087699e-01 -2.38258064e-01
-6.96386218e-01 8.68050158e-01 2.37959549e-01 5.94335735e-01
-8.80609751e-01 1.60566524e-01 7.07337976e-01 1.12263918e-01
7.71651208e-01 -1.32447481e-01 6.10303879e-01 -1.30918574e+00
1.58769161e-01 -5.94682276e-01 4.22485232e-01 -4.67337042e-01
-1.40534055e+00 6.02163076e-01 1.10924967e-01 -1.32409048e+00
-1.40241673e-02 -3.93838882e-01 -6.11991346e-01 5.77551961e-01
-1.30367231e+00 -9.63888049e-01 -2.85483867e-01 1.19742060e+00
3.28309864e-01 -1.16630748e-01 5.55697143e-01 4.18643326e-01
-7.95514107e-01 2.49897972e-01 2.05379352e-01 2.53957182e-01
6.12808406e-01 -6.64483666e-01 -3.02863717e-01 7.93945551e-01
2.15876207e-01 7.72538245e-01 2.49387309e-01 -7.63029277e-01
-1.31058455e+00 -1.24677765e+00 6.19483650e-01 -2.41808012e-01
4.33758706e-01 -4.57569242e-01 -9.63354111e-01 7.78966188e-01
1.87972844e-01 3.18603843e-01 8.71570826e-01 -7.62968212e-02
1.84553638e-02 -4.30587120e-02 -1.05912244e+00 2.68268883e-01
1.29139709e+00 -3.16389859e-01 -1.44196108e-01 4.88448530e-01
2.58805543e-01 -1.04435571e-01 -9.83944774e-01 3.25716019e-01
3.76750618e-01 -8.22490096e-01 8.47485721e-01 -4.08811867e-01
-1.00368164e-01 -8.24531019e-01 -3.33330214e-01 -9.81142819e-01
-6.32547021e-01 -7.56008685e-01 -1.86915323e-01 1.64052153e+00
-1.61778316e-01 -2.73772359e-01 6.48900568e-01 6.21341169e-01
1.32422656e-01 -4.55488116e-01 -9.79451180e-01 -6.83549523e-01
-4.21857476e-01 -9.56945568e-02 5.59029639e-01 9.97423291e-01
-1.06471941e-01 4.46509644e-02 -4.61774886e-01 5.99367261e-01
1.02795649e+00 1.36136115e-01 4.99319553e-01 -1.50599861e+00
-2.89072216e-01 -1.26584604e-01 -5.09007692e-01 -1.03859544e+00
7.29633749e-01 -7.80561328e-01 1.90182820e-01 -1.13953447e+00
6.05822146e-01 -4.27963763e-01 -1.19741164e-01 3.46086621e-01
-1.89795494e-01 1.45827323e-01 2.96931565e-01 3.75014931e-01
-7.28873610e-01 5.88624060e-01 1.13876390e+00 -5.09570688e-02
-1.15348376e-01 3.07603795e-02 -2.74286598e-01 9.89380538e-01
4.57708269e-01 -5.14847696e-01 -5.07695556e-01 -3.30936968e-01
-3.98806006e-01 2.10268661e-01 1.61736771e-01 -7.97331154e-01
4.88986701e-01 -3.61874580e-01 5.63959897e-01 -6.09558523e-01
4.38300043e-01 -1.05658209e+00 5.48670709e-01 9.44811478e-02
2.05891188e-02 -1.64215341e-01 -1.70956090e-01 8.53658020e-01
-3.48210514e-01 -1.46756500e-01 8.64184499e-01 -2.10551292e-01
-6.09461963e-01 8.47841144e-01 -4.37917799e-01 -2.80800283e-01
1.43024218e+00 -2.14495569e-01 1.93825379e-01 -3.00909877e-01
-1.12002826e+00 4.41863745e-01 6.14919126e-01 3.49516004e-01
6.66593969e-01 -1.54017913e+00 -5.04974902e-01 5.21290660e-01
-7.65082836e-02 2.26396471e-01 6.01619482e-01 9.52758849e-01
-2.49076877e-02 2.85200924e-01 -3.33715677e-02 -9.82784152e-01
-1.55621874e+00 8.43499243e-01 2.36027464e-01 2.37094834e-01
-4.42603767e-01 7.59883523e-01 7.38303483e-01 -1.16007447e-01
3.84171814e-01 1.09325044e-01 -4.32469904e-01 1.39642984e-01
8.09644699e-01 2.75176883e-01 -2.43284807e-01 -1.32732952e+00
-6.15922809e-01 1.06156421e+00 -1.75267309e-01 9.71426889e-02
1.37822807e+00 -4.19827670e-01 -4.12965357e-01 1.96354136e-01
1.23015797e+00 -4.41201031e-04 -1.59169006e+00 -1.79588273e-01
-2.12103859e-01 -6.95453107e-01 -9.47397500e-02 -9.16717295e-03
-1.68133616e+00 6.23439848e-01 4.47760552e-01 -7.35542327e-02
1.30243278e+00 1.79728538e-01 3.83011967e-01 6.14863485e-02
6.62737072e-01 -9.49059546e-01 -4.89342660e-02 1.61527708e-01
4.74648029e-01 -8.56733441e-01 -1.24319315e-01 -9.13200438e-01
-5.56257486e-01 9.24738824e-01 5.10242045e-01 -9.97632518e-02
1.01001680e+00 3.12617034e-01 -6.43108189e-02 -1.28434137e-01
-6.92257345e-01 -2.37867072e-01 1.51049361e-01 6.34222329e-01
3.17657411e-01 -1.73619747e-01 -1.79944038e-02 4.06914532e-01
3.81382823e-01 -1.03818160e-02 2.87129581e-01 8.13711345e-01
-4.70598161e-01 -1.21906281e+00 -6.24760449e-01 2.05708474e-01
-3.32914799e-01 2.57718235e-01 2.26150546e-02 4.50762957e-01
3.83492529e-01 1.28446758e+00 2.24804685e-01 -2.42779344e-01
-1.84949949e-01 6.56179711e-02 7.01156616e-01 -6.35551512e-01
-6.35818467e-02 7.09229112e-01 -4.89687979e-01 -5.28824389e-01
-1.14535677e+00 -1.04744756e+00 -1.40989840e+00 -3.23802769e-01
-5.59324682e-01 5.15888631e-01 2.13190898e-01 9.46301460e-01
4.21573192e-01 1.25433356e-01 1.04850352e+00 -1.20700300e+00
1.54755428e-01 -6.88939393e-01 -9.55506504e-01 3.50885451e-01
1.42333642e-01 -1.06652915e+00 -4.36928272e-01 6.70434415e-01]
|
[13.46142578125, 1.0875228643417358]
|
297244f1-7b71-475c-bd46-738426807c2c
|
towards-incremental-learning-of-word
| null | null |
https://aclanthology.org/P19-2022
|
https://aclanthology.org/P19-2022.pdf
|
Towards Incremental Learning of Word Embeddings Using Context Informativeness
|
In this paper, we investigate the task of learning word embeddings from very sparse data in an incremental, cognitively-plausible way. We focus on the notion of {`}informativeness{'}, that is, the idea that some content is more valuable to the learning process than other. We further highlight the challenges of online learning and argue that previous systems fall short of implementing incrementality. Concretely, we incorporate informativeness in a previously proposed model of nonce learning, using it for context selection and learning rate modulation. We test our system on the task of learning new words from definitions, as well as on the task of learning new words from potentially uninformative contexts. We demonstrate that informativeness is crucial to obtaining state-of-the-art performance in a truly incremental setup.
|
["Aur{\\'e}lie Herbelot", 're', 'Alex Kabbach', 'Kristina Gulordava']
|
2019-07-01
| null | null | null |
acl-2019-7
|
['learning-word-embeddings']
|
['methodology']
|
[ 4.63715792e-01 1.95329189e-01 -1.61240682e-01 -3.59843314e-01
-8.39037299e-01 -5.88219702e-01 7.17340529e-01 4.67731804e-01
-1.12558317e+00 7.79139102e-01 6.94227219e-01 -4.80071217e-01
-5.29421568e-02 -6.02281392e-01 -6.89779401e-01 -4.15436119e-01
-2.23605022e-01 3.66611391e-01 5.29579110e-02 -2.20874593e-01
2.60721326e-01 2.56663024e-01 -1.66683710e+00 1.49086133e-01
5.28466523e-01 5.96805692e-01 2.30849460e-01 7.06656516e-01
-2.75582582e-01 6.67437434e-01 -4.41809893e-01 -5.63845694e-01
1.38721704e-01 -1.34286866e-01 -1.10868323e+00 -2.90867835e-01
6.11818075e-01 -3.23225617e-01 -1.39521256e-01 8.12281787e-01
4.88689899e-01 5.70268154e-01 6.70347214e-01 -7.53278852e-01
-1.03789532e+00 1.07070220e+00 1.56228945e-01 6.72546268e-01
1.42981663e-01 5.89663014e-02 1.54727495e+00 -1.48746419e+00
6.12724781e-01 1.08810306e+00 4.49442357e-01 8.74613643e-01
-1.23074508e+00 -4.17160779e-01 7.60330200e-01 3.24210137e-01
-1.14851356e+00 -7.74454594e-01 6.01460457e-01 -3.01129013e-01
1.06112361e+00 3.13969046e-01 4.58979756e-01 1.38089240e+00
-2.52470404e-01 9.37061667e-01 9.64743316e-01 -9.15968895e-01
5.49603224e-01 1.64551586e-01 5.08296907e-01 5.61489880e-01
5.26032448e-01 2.70924658e-01 -8.72108996e-01 -2.55931038e-02
2.81951815e-01 -9.29492563e-02 -2.18469992e-01 -2.08477154e-01
-9.53563094e-01 1.07623804e+00 1.85220927e-01 3.69791865e-01
-2.49823749e-01 4.47505534e-01 2.85631180e-01 3.48477870e-01
6.13303185e-01 8.06830883e-01 -6.24487698e-01 -3.58856142e-01
-6.16238654e-01 1.68838911e-02 7.41907001e-01 8.00027609e-01
5.96855521e-01 1.74834549e-01 -2.51240164e-01 7.77071714e-01
2.29406387e-01 1.81991622e-01 7.03121901e-01 -8.19637656e-01
1.24103159e-01 8.35304409e-02 -3.60024869e-02 -1.18749827e-01
-3.05855423e-01 -4.45613444e-01 -1.89635381e-01 -1.26169458e-01
2.45810702e-01 -3.59672278e-01 -8.47154796e-01 2.10456944e+00
5.03269471e-02 3.21701467e-01 -1.30781069e-01 4.86201286e-01
5.66813767e-01 4.26471531e-01 3.29512656e-01 -3.72038811e-01
1.22408485e+00 -8.96422207e-01 -7.51643121e-01 -6.20770693e-01
5.69366157e-01 -5.60331106e-01 1.40318930e+00 5.01032114e-01
-1.06474733e+00 -3.86406422e-01 -9.83000576e-01 -3.68472666e-01
-6.32968545e-01 -3.19748998e-01 9.99159813e-01 7.65511572e-01
-1.04907656e+00 5.17642379e-01 -5.66112757e-01 -2.17318311e-01
4.47911143e-01 1.23868473e-01 -1.35437101e-01 -2.99482167e-01
-1.39460146e+00 1.12298524e+00 4.59822774e-01 -2.62249440e-01
-6.86610103e-01 -6.70156121e-01 -1.02224660e+00 1.21457867e-01
5.79545677e-01 -7.31996417e-01 1.48646855e+00 -9.81760561e-01
-1.21341419e+00 6.19762778e-01 -3.49038303e-01 -6.16988122e-01
9.64095145e-02 -5.83685160e-01 -4.33281213e-01 -4.76499014e-02
-2.26587266e-01 5.80321431e-01 8.90208423e-01 -1.11175478e+00
-7.30667591e-01 -8.89308974e-02 4.39250141e-01 2.40985826e-01
-9.14325595e-01 -2.84114778e-01 -2.44999453e-01 -8.27929676e-01
-4.16324675e-01 -8.25192273e-01 -1.14882432e-01 1.82019547e-01
1.58302382e-01 -6.07171357e-01 3.70484114e-01 -4.68186468e-01
1.44284475e+00 -2.24805546e+00 -5.62578551e-02 1.51842877e-01
3.13607901e-01 4.63618517e-01 -4.40683395e-01 2.66248405e-01
-5.45799062e-02 2.92785227e-01 -1.06488191e-01 -4.94291425e-01
2.12038413e-01 3.69759470e-01 -4.42735076e-01 4.20561619e-02
3.56605560e-01 1.10348845e+00 -1.34011745e+00 -1.24664240e-01
9.48772356e-02 4.55249637e-01 -7.62461245e-01 9.97589752e-02
-2.43313000e-01 -2.81956822e-01 -1.79460183e-01 2.79199868e-01
1.62164822e-01 -5.26526012e-02 2.36694217e-01 2.82343645e-02
4.29057293e-02 7.38948882e-01 -1.20753026e+00 1.70027411e+00
-9.41245735e-01 7.60792792e-01 -1.91507787e-01 -8.81479919e-01
4.86259192e-01 7.28537217e-02 2.16183178e-02 -5.69981873e-01
-1.14087373e-01 1.35451511e-01 3.75724733e-02 -3.43504667e-01
8.36400747e-01 -3.85952950e-01 1.03042843e-02 7.23746061e-01
3.74959141e-01 7.98804462e-02 1.48312062e-01 2.28720248e-01
1.01184309e+00 7.22271279e-02 5.91195405e-01 -3.72909725e-01
8.19482133e-02 -4.72301245e-01 7.00322539e-02 9.53276813e-01
-1.49444744e-01 2.84702569e-01 1.24781385e-01 -2.77861238e-01
-8.15194488e-01 -1.35637915e+00 -1.50910199e-01 1.85398579e+00
-1.50923699e-01 -6.16887569e-01 -3.18402976e-01 -9.50883269e-01
-2.51113251e-02 1.34188187e+00 -7.94744432e-01 -5.01847982e-01
-5.62915087e-01 -5.49905956e-01 1.88604146e-01 6.62049353e-01
-2.27752551e-01 -1.01607883e+00 -6.82163775e-01 2.60511518e-01
2.22063791e-02 -8.62629712e-01 -6.54823959e-01 7.26360381e-01
-7.67480671e-01 -7.29908109e-01 -2.25272655e-01 -7.85931885e-01
5.67391872e-01 4.61214364e-01 1.49309313e+00 4.12290134e-02
-3.10064942e-01 7.72358060e-01 -5.28898895e-01 -6.38492942e-01
-1.26341820e-01 1.79381564e-01 2.67996311e-01 -2.23014385e-01
6.86240375e-01 -5.54909766e-01 -5.48934758e-01 -4.49260384e-01
-1.17347288e+00 -2.19777897e-01 5.43334663e-01 9.21524405e-01
3.64633292e-01 -2.86709666e-01 8.92222464e-01 -1.37012506e+00
9.70708013e-01 -5.92601359e-01 -1.26817614e-01 1.15747795e-01
-8.25556457e-01 5.46084821e-01 5.72086096e-01 -5.72634399e-01
-9.07009661e-01 -7.49414116e-02 -4.07529801e-01 3.98805290e-02
2.93696839e-02 4.29475814e-01 2.56972834e-02 1.12695754e-01
8.47985268e-01 1.28419504e-01 -3.75806510e-01 -3.43943983e-01
1.15810764e+00 5.68133175e-01 2.12145016e-01 -9.36221361e-01
7.37390459e-01 2.95596749e-01 -4.42845583e-01 -8.62403214e-01
-1.12158942e+00 -5.19222677e-01 -4.80447084e-01 7.97406137e-02
4.53321278e-01 -9.03143406e-01 -3.71888310e-01 -2.71879017e-01
-1.10682476e+00 -3.43817204e-01 -1.00308657e+00 3.95393640e-01
-3.26164722e-01 2.09389552e-01 -3.15188438e-01 -8.95592272e-01
-2.77837306e-01 -6.81375980e-01 7.62673438e-01 8.58970359e-02
-4.42438394e-01 -1.35915804e+00 1.51321203e-01 2.34504268e-02
6.89687133e-01 -2.77386159e-01 1.04901159e+00 -1.14444745e+00
-4.23086196e-01 -8.09871033e-02 6.31595328e-02 4.92267460e-01
1.14477932e-01 -4.17373568e-01 -1.46515620e+00 -2.80134916e-01
-3.37289297e-03 -6.24551654e-01 1.35389411e+00 6.03153855e-02
1.26830983e+00 -6.11880720e-01 -1.68417376e-02 3.31768304e-01
1.51413584e+00 -1.61767036e-01 2.98347563e-01 1.06612891e-01
5.54000735e-01 3.60675424e-01 4.07464683e-01 4.80654866e-01
5.67777574e-01 3.29469979e-01 2.41630509e-01 2.27943331e-01
-3.70308191e-01 -4.03847486e-01 5.35782516e-01 9.42085743e-01
1.18143462e-01 -1.56548515e-01 -6.59735620e-01 8.99093151e-01
-1.46471047e+00 -1.05285931e+00 4.98113930e-01 2.23227906e+00
1.33967280e+00 4.74877208e-01 -5.29819354e-02 -1.02642521e-01
4.14687783e-01 3.69872659e-01 -4.83856440e-01 -6.98695719e-01
4.27389331e-02 9.39529061e-01 2.57592082e-01 9.36060011e-01
-1.09786916e+00 9.43763256e-01 6.98363590e+00 7.79526174e-01
-9.04247820e-01 3.90271872e-01 4.25999910e-01 -4.90433484e-01
-8.51790071e-01 -6.79690987e-02 -8.68361652e-01 3.29369545e-01
1.16256034e+00 -3.25846672e-01 6.62423432e-01 9.00161624e-01
-2.79892117e-01 1.83450442e-03 -1.55974293e+00 8.19489777e-01
4.22527164e-01 -1.26106346e+00 2.97028959e-01 -1.97029531e-01
7.11789906e-01 4.87227254e-02 1.98138103e-01 6.86698675e-01
6.38554156e-01 -1.00817025e+00 6.40835762e-01 3.63394082e-01
7.76379406e-01 -7.49857187e-01 3.60620797e-01 1.80974603e-01
-9.94099915e-01 -2.04788208e-01 -2.78859913e-01 -2.62779266e-01
7.10970014e-02 7.69664526e-01 -7.11505115e-01 2.09606215e-01
2.49864310e-02 3.89273971e-01 -8.30965579e-01 8.14328969e-01
-5.09480417e-01 8.70937288e-01 -3.26136738e-01 -3.70997041e-01
1.73138306e-01 3.89373183e-01 2.34539524e-01 1.53997552e+00
9.12225544e-02 6.51287585e-02 1.77574977e-01 6.17600739e-01
-4.75499183e-01 -1.25589799e-02 -7.55120754e-01 2.68251151e-02
9.24270749e-01 1.04800129e+00 -2.88693458e-01 -3.67474794e-01
-5.55100560e-01 8.10554683e-01 7.54549623e-01 2.74390817e-01
-4.38634813e-01 -3.45334053e-01 6.63071454e-01 -2.67414659e-01
4.98743027e-01 -3.79182994e-01 -2.20438823e-01 -1.19316757e+00
7.41660967e-02 -7.08024323e-01 4.80434090e-01 -4.30282444e-01
-1.50326884e+00 3.30732793e-01 -6.85099959e-02 -7.11931407e-01
-2.96517372e-01 -7.85267830e-01 -5.43434978e-01 6.56648219e-01
-1.93223548e+00 -8.65890861e-01 1.03300117e-01 3.37991536e-01
9.03324723e-01 1.01431444e-01 1.08274770e+00 8.77627805e-02
-2.56569982e-01 8.14254224e-01 1.03994042e-01 -1.11241683e-01
7.78776109e-01 -1.56637681e+00 6.70021117e-01 1.01118064e+00
8.12065125e-01 9.57553983e-01 6.10070705e-01 -2.62813061e-01
-1.43660498e+00 -1.12126362e+00 1.21294606e+00 -8.51179361e-01
7.41686046e-01 -6.68528199e-01 -5.67518890e-01 7.26959705e-01
2.25321740e-01 1.04278803e-01 1.19652450e+00 6.27335429e-01
-6.96614027e-01 1.53844655e-01 -8.06754291e-01 8.27829897e-01
1.28464794e+00 -8.92562509e-01 -1.21237326e+00 2.89715320e-01
1.15066826e+00 5.13860099e-02 -4.07803237e-01 -1.19623065e-01
5.15280902e-01 -4.79745179e-01 1.13396657e+00 -8.84102523e-01
2.75236756e-01 1.65054590e-01 -3.60414386e-01 -1.52700460e+00
-5.94499946e-01 -6.73590600e-01 -6.97241783e-01 1.14362752e+00
4.86230731e-01 -4.51716304e-01 5.48613966e-01 6.29470825e-01
-9.56612304e-02 -7.51911521e-01 -1.06239951e+00 -7.49358177e-01
3.92680287e-01 -8.11829090e-01 2.47247577e-01 1.02913213e+00
1.42933100e-01 7.92307138e-01 -2.00721264e-01 -1.15050785e-01
3.65518123e-01 -2.35538438e-01 1.39356300e-01 -1.21342909e+00
-3.61556500e-01 -4.41355258e-01 2.84453183e-02 -1.15592706e+00
2.50046402e-01 -1.11660671e+00 1.63818866e-01 -1.43463790e+00
1.39253035e-01 -3.28930974e-01 -7.23183692e-01 3.98486137e-01
-6.39212549e-01 1.21242359e-01 2.81915873e-01 -2.93139577e-01
-8.40286255e-01 4.92191225e-01 8.12262952e-01 2.38864049e-02
5.09505272e-02 -2.40518540e-01 -1.24483144e+00 6.84083700e-01
6.48068547e-01 -4.29675817e-01 -7.06221700e-01 -7.77208269e-01
4.96377587e-01 -7.06288695e-01 2.08922029e-01 -6.08264923e-01
2.50867486e-01 -1.07551031e-01 2.13144898e-01 7.83837400e-03
3.92728299e-01 -7.84724534e-01 -8.13377380e-01 2.75251895e-01
-8.95821095e-01 4.57405150e-02 2.12897196e-01 6.90015435e-01
1.84970453e-01 -5.55607378e-01 5.03643572e-01 -2.03957379e-01
-1.09827626e+00 2.29995027e-01 -3.10833693e-01 7.41298139e-01
5.88992953e-01 1.63112327e-01 -2.93659925e-01 -1.46149293e-01
-7.29364634e-01 -8.45558643e-02 1.16584711e-01 6.25722706e-01
8.07418644e-01 -1.45852852e+00 -6.75867081e-01 2.47047797e-01
4.81297731e-01 -2.03282580e-01 -1.64819807e-01 2.44652659e-01
1.34661764e-01 2.90349901e-01 2.41457582e-01 2.40680221e-02
-9.12246227e-01 6.09475672e-01 -5.58518022e-02 -1.32576823e-01
-4.31530893e-01 1.13511503e+00 -6.31610304e-02 -2.53496438e-01
5.03737986e-01 -4.86620814e-01 -1.78523168e-01 4.84314919e-01
7.62567997e-01 1.15857109e-01 1.11854799e-01 1.13722742e-01
-3.81849647e-01 1.10719495e-01 -3.90747458e-01 -3.99260730e-01
1.33610153e+00 -1.14400379e-01 2.46495709e-01 7.68878996e-01
1.18407393e+00 2.84825534e-01 -1.08900356e+00 -7.07454681e-01
3.66649926e-01 -3.83763343e-01 2.62679886e-02 -9.69390571e-01
-4.46909308e-01 9.18511212e-01 7.33182251e-01 1.51769802e-01
7.72955358e-01 2.13415455e-02 7.30991721e-01 8.33267212e-01
3.22342753e-01 -1.38536990e+00 2.83363253e-01 8.52942169e-01
5.73421359e-01 -1.05323815e+00 3.33054736e-02 1.61617815e-01
-4.39316869e-01 1.12368441e+00 4.27099913e-01 -9.18352455e-02
7.01119900e-01 1.72447428e-01 -2.27187052e-01 4.77348156e-02
-1.23251998e+00 -5.73761344e-01 3.26674044e-01 7.52785742e-01
7.37614691e-01 1.09181991e-02 -2.37355933e-01 7.33430266e-01
-1.40647247e-01 9.50503536e-03 5.54509759e-01 1.03994370e+00
-8.43421936e-01 -1.09943318e+00 9.33792070e-02 5.03883362e-01
-2.85387695e-01 -8.01907003e-01 -2.83800244e-01 5.43862998e-01
2.84974933e-01 7.85042226e-01 9.97778550e-02 -1.16875000e-01
2.77310818e-01 6.45389974e-01 5.72173834e-01 -1.13847864e+00
-3.94393772e-01 -3.92374218e-01 4.20779198e-01 -3.35794628e-01
-2.15373501e-01 -5.50120771e-01 -9.82875109e-01 -1.51172876e-02
-2.04399556e-01 4.68727387e-02 6.09641850e-01 1.13700771e+00
2.27066293e-01 5.91824830e-01 6.14837766e-01 -5.44343770e-01
-9.25976396e-01 -7.33645439e-01 -2.50221193e-01 5.21642685e-01
5.72905838e-01 -4.91439670e-01 -5.03938556e-01 3.27246413e-02]
|
[10.636524200439453, 8.559222221374512]
|
3b7c0722-3640-4c1d-a8cf-dc4190efb4c4
|
learning-joint-latent-space-ebm-prior-model-1
|
2306.06323
| null |
https://arxiv.org/abs/2306.06323v1
|
https://arxiv.org/pdf/2306.06323v1.pdf
|
Learning Joint Latent Space EBM Prior Model for Multi-layer Generator
|
This paper studies the fundamental problem of learning multi-layer generator models. The multi-layer generator model builds multiple layers of latent variables as a prior model on top of the generator, which benefits learning complex data distribution and hierarchical representations. However, such a prior model usually focuses on modeling inter-layer relations between latent variables by assuming non-informative (conditional) Gaussian distributions, which can be limited in model expressivity. To tackle this issue and learn more expressive prior models, we propose an energy-based model (EBM) on the joint latent space over all layers of latent variables with the multi-layer generator as its backbone. Such joint latent space EBM prior model captures the intra-layer contextual relations at each layer through layer-wise energy terms, and latent variables across different layers are jointly corrected. We develop a joint training scheme via maximum likelihood estimation (MLE), which involves Markov Chain Monte Carlo (MCMC) sampling for both prior and posterior distributions of the latent variables from different layers. To ensure efficient inference and learning, we further propose a variational training scheme where an inference model is used to amortize the costly posterior MCMC sampling. Our experiments demonstrate that the learned model can be expressive in generating high-quality images and capturing hierarchical features for better outlier detection.
|
['Tian Han', 'Ying Nian Wu', 'Jiali Cui']
|
2023-06-10
|
learning-joint-latent-space-ebm-prior-model
|
http://openaccess.thecvf.com//content/CVPR2023/html/Cui_Learning_Joint_Latent_Space_EBM_Prior_Model_for_Multi-Layer_Generator_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Cui_Learning_Joint_Latent_Space_EBM_Prior_Model_for_Multi-Layer_Generator_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['outlier-detection']
|
['methodology']
|
[ 1.29259899e-01 6.77103698e-02 -1.86541930e-01 -2.17168376e-01
-1.01309073e+00 -1.26176067e-02 5.53824544e-01 -3.19998324e-01
-3.15232426e-02 6.74978316e-01 2.22156808e-01 1.65132552e-01
7.49332532e-02 -9.22608316e-01 -9.44623768e-01 -1.03482866e+00
3.10027841e-02 3.94717067e-01 1.31073639e-01 5.26531994e-01
-1.92031279e-01 2.42086172e-01 -1.37797546e+00 4.11623031e-01
1.04906392e+00 7.79747128e-01 4.73600656e-01 6.29565597e-01
-5.02671711e-02 7.75183380e-01 -5.67612112e-01 -3.39171350e-01
-1.74778417e-01 -6.62756622e-01 -2.64885098e-01 3.08752120e-01
9.40755829e-02 -2.78871626e-01 1.04252011e-01 1.15477812e+00
4.35906261e-01 -5.36227375e-02 1.24415612e+00 -1.47235382e+00
-6.14363492e-01 5.04458725e-01 -9.03843820e-01 -4.93580639e-01
-1.52543485e-01 1.05738573e-01 1.08429825e+00 -1.09332502e+00
3.88943732e-01 1.49769831e+00 7.62856960e-01 5.09214997e-01
-1.58238101e+00 -7.79174328e-01 3.85274857e-01 1.04831681e-01
-1.60485840e+00 -1.80046216e-01 9.36228395e-01 -6.29005730e-01
7.13090003e-01 -1.45261800e-02 6.56573892e-01 1.53650010e+00
5.00682473e-01 1.08623445e+00 8.87415648e-01 -3.50738496e-01
4.19147402e-01 6.66025504e-02 -4.21669215e-01 7.26913035e-01
3.30213755e-01 -1.61922589e-01 -6.98984563e-01 -4.92994696e-01
1.06228995e+00 4.72819693e-02 6.97475821e-02 -4.12930578e-01
-8.84232938e-01 1.09733665e+00 1.16243057e-01 -1.73472106e-01
-5.49557269e-01 6.33998930e-01 9.41082165e-02 -3.80421638e-01
6.12704575e-01 -1.12117693e-01 -1.84355929e-01 2.69717753e-01
-1.48093557e+00 2.18116209e-01 4.66353595e-01 1.10581112e+00
1.08661401e+00 2.02964291e-01 -4.57355410e-01 9.35050488e-01
1.07797587e+00 4.77965355e-01 2.90443987e-01 -1.09228206e+00
3.06674689e-01 4.04247314e-01 -3.11274659e-02 -7.19634295e-01
2.12536857e-01 -4.59046543e-01 -1.43730485e+00 3.93874168e-01
-1.19053490e-01 -1.61758900e-01 -9.91543472e-01 1.89873064e+00
2.76171535e-01 5.54385960e-01 -2.04165399e-01 3.26077104e-01
2.99703002e-01 8.32957864e-01 1.21920943e-01 -5.69885612e-01
1.14541769e+00 -1.09569395e+00 -8.17623377e-01 -8.74497816e-02
2.71922410e-01 -6.35401905e-01 1.01163411e+00 4.54736978e-01
-1.32782006e+00 -6.09996319e-01 -9.68896925e-01 -9.90558490e-02
2.80399993e-02 4.18135136e-01 4.16731030e-01 2.06386611e-01
-9.09803569e-01 5.44825733e-01 -1.25877655e+00 2.26181939e-01
6.27851307e-01 9.94079486e-02 1.61972776e-01 -1.13845356e-01
-9.80089784e-01 4.00785774e-01 2.54979312e-01 2.31256727e-02
-1.49015498e+00 -5.25781333e-01 -9.05252934e-01 3.92955318e-02
1.06520183e-01 -9.49233055e-01 8.37742448e-01 -6.22430265e-01
-1.62775111e+00 4.20924872e-01 -5.62126338e-01 -1.55204669e-01
6.39572680e-01 -2.33044833e-01 -9.52039734e-02 3.19403037e-02
5.50405346e-02 6.47755206e-01 1.51809895e+00 -1.52663934e+00
-1.57063082e-01 1.47107780e-01 -4.08005714e-01 5.43129519e-02
-1.81633741e-01 -1.75880179e-01 -7.49935865e-01 -9.53536689e-01
1.24289006e-01 -7.63059855e-01 -3.65855932e-01 5.79101592e-02
-4.29784089e-01 -2.82933086e-01 5.92499375e-01 -6.98199451e-01
1.38746226e+00 -2.10350204e+00 3.54287863e-01 9.78872404e-02
2.24411294e-01 -1.42021939e-01 -2.70510148e-02 2.71688700e-01
8.83147568e-02 9.83487219e-02 -5.81751049e-01 -1.16934824e+00
2.52909720e-01 6.03922546e-01 -4.88779783e-01 3.29871535e-01
1.51625171e-01 9.75241065e-01 -8.16263556e-01 -8.99738848e-01
2.25023225e-01 9.03105140e-01 -9.47666347e-01 4.71603274e-01
-4.78944063e-01 4.93451327e-01 -4.11826044e-01 4.65773344e-01
9.61780846e-01 -5.58911502e-01 7.60691538e-02 -3.35784644e-01
2.45625585e-01 -5.69730811e-02 -1.50164366e+00 1.82668173e+00
-3.89939219e-01 1.78534642e-01 1.69339418e-01 -6.60078347e-01
6.24055326e-01 4.10657853e-01 3.41258049e-01 -5.95312491e-02
-2.18365118e-01 3.08351405e-02 -4.75209862e-01 -3.60337138e-01
3.02612215e-01 -2.82930046e-01 -1.45942941e-01 4.83278960e-01
1.82705536e-01 -1.29419371e-01 -9.64768045e-03 4.03732419e-01
5.31767488e-01 5.62695920e-01 2.11545244e-01 -5.89306876e-02
3.89886141e-01 -7.20256448e-01 8.64508092e-01 8.27100933e-01
4.19334203e-01 8.23620379e-01 5.72624743e-01 3.36385146e-02
-1.22094154e+00 -1.41459882e+00 -1.01906493e-01 8.31493616e-01
-1.63294405e-01 -7.30442762e-01 -8.02811682e-01 -4.40212220e-01
-2.56654143e-01 7.63841093e-01 -5.06137013e-01 -1.80772483e-01
-3.40174258e-01 -1.23942769e+00 3.39538425e-01 5.90332508e-01
4.85885143e-01 -9.57270086e-01 -3.73849124e-01 2.18172163e-01
-2.07609370e-01 -7.82349646e-01 -2.84293145e-01 9.56202000e-02
-9.77371395e-01 -5.78563511e-01 -7.10166037e-01 -3.49423915e-01
8.84830773e-01 -3.71330649e-01 1.19763505e+00 -9.18779597e-02
-1.66711167e-01 1.75181732e-01 -1.77389123e-02 -6.55125082e-02
-5.28777599e-01 -2.39357315e-02 9.70709398e-02 1.79581121e-01
-1.27489805e-01 -8.70456994e-01 -5.25730789e-01 8.66296515e-02
-9.49861825e-01 5.92352927e-01 9.45167482e-01 7.94649422e-01
1.07305741e+00 2.04812422e-01 2.17882395e-01 -8.67597759e-01
3.50827336e-01 -6.33461952e-01 -6.67865157e-01 3.91873121e-01
-5.19041121e-01 2.90571064e-01 4.74502087e-01 -6.00154698e-01
-1.27788317e+00 -1.30733654e-01 -1.27805173e-01 -7.75223196e-01
-5.25342263e-02 4.26436543e-01 -6.95593297e-01 4.13925827e-01
2.24875376e-01 3.15350801e-01 -3.84287566e-01 -7.74110138e-01
5.27817190e-01 2.34437257e-01 4.90809798e-01 -8.41383278e-01
8.20504725e-01 4.99965549e-01 1.72439247e-01 -5.40822506e-01
-9.58789587e-01 -3.77228409e-02 -6.60078824e-01 -2.79072411e-02
1.23989356e+00 -1.23871422e+00 -2.91655719e-01 7.41185725e-01
-1.15589774e+00 -5.18256426e-01 -3.48693013e-01 4.73636597e-01
-5.86754680e-01 4.07798409e-01 -6.97421312e-01 -9.49296057e-01
2.91189235e-02 -1.26057553e+00 1.32338238e+00 2.94769052e-02
-1.42470896e-01 -1.32618785e+00 1.36066511e-01 2.40077898e-02
1.99769065e-01 3.62158000e-01 9.88827527e-01 1.10210739e-01
-9.98009622e-01 -8.22816491e-02 -2.94798184e-02 6.20719969e-01
-2.65314654e-02 1.22996517e-01 -1.08556366e+00 -4.78475094e-01
1.42522886e-01 -3.23608905e-01 1.13454247e+00 5.86605847e-01
1.66250038e+00 -5.23771524e-01 -5.29914141e-01 9.25658584e-01
1.52464783e+00 -5.43943048e-01 7.19984293e-01 -3.61710548e-01
1.09530604e+00 1.74219906e-01 1.63840950e-01 7.19319820e-01
4.37060684e-01 4.42295849e-01 3.73622477e-01 -9.00839120e-02
-1.31006926e-01 -6.50379777e-01 6.68919504e-01 1.18399572e+00
4.82187085e-02 -3.98274124e-01 -5.39433539e-01 4.42292899e-01
-1.94082069e+00 -9.60526288e-01 -1.15392685e-01 2.05773091e+00
1.12889099e+00 3.90495472e-02 -1.22293063e-01 -1.14409812e-01
5.88929117e-01 2.82050401e-01 -7.68271506e-01 1.55935049e-01
-3.13424580e-02 2.06174962e-02 2.30484545e-01 7.44348228e-01
-8.91836524e-01 6.64276540e-01 6.77050591e+00 1.29560578e+00
-6.24734521e-01 3.11357528e-01 5.04871964e-01 -2.75579810e-01
-8.58589411e-01 1.68197095e-01 -1.02767789e+00 8.80534470e-01
7.53349841e-01 1.30649880e-01 3.68402004e-01 7.93458998e-01
1.88705802e-01 -1.17025472e-01 -1.05733466e+00 9.64135885e-01
1.81712396e-02 -1.25047052e+00 4.72929716e-01 4.67171103e-01
1.13030791e+00 -6.03889450e-02 6.60323054e-02 2.99090654e-01
7.37640500e-01 -1.17700052e+00 7.57102907e-01 9.67381120e-01
9.37833369e-01 -7.55902767e-01 3.73077512e-01 6.75320566e-01
-1.30112910e+00 3.09584558e-01 -5.52736163e-01 9.58429202e-02
3.86106521e-01 9.90530491e-01 -3.23499382e-01 4.99847472e-01
6.29824519e-01 7.67406464e-01 -3.96237880e-01 6.87500238e-01
-6.64278686e-01 8.07225883e-01 -3.19436401e-01 5.88126302e-01
-2.24689245e-02 -4.13735658e-01 4.55447346e-01 1.16773987e+00
3.99396807e-01 -2.66736925e-01 2.63171941e-01 1.33870220e+00
-9.84392762e-02 -3.69189471e-01 -1.81139439e-01 1.81406379e-01
5.80740988e-01 1.29853809e+00 -6.66688442e-01 -5.62998712e-01
-3.80579740e-01 1.22923458e+00 4.69324589e-01 7.50952423e-01
-1.16800201e+00 1.10229202e-01 5.51830113e-01 -3.42650525e-02
4.90790844e-01 -4.15881902e-01 -2.41798282e-01 -1.50582063e+00
-6.71240091e-02 -5.64406991e-01 3.17522913e-01 -7.73834825e-01
-1.54578900e+00 1.35807708e-01 3.47796410e-01 -1.02327776e+00
-5.10911822e-01 -4.35876161e-01 -8.47224176e-01 1.25733352e+00
-1.52445388e+00 -1.35855603e+00 -2.40962178e-01 6.11132383e-01
3.53989422e-01 -4.75058286e-03 8.20120394e-01 1.47489548e-01
-7.68562794e-01 6.79974198e-01 1.78013608e-01 -2.18612254e-01
5.54210246e-01 -1.29109955e+00 9.94189084e-02 9.97588754e-01
1.25265881e-01 6.71675444e-01 4.95392561e-01 -8.49910200e-01
-8.71152222e-01 -1.32405257e+00 6.57749951e-01 -4.77935493e-01
3.61580729e-01 -6.90308809e-01 -1.07827318e+00 8.75151813e-01
1.49456784e-01 -2.73563489e-02 8.63749743e-01 -1.06810182e-01
-1.30780190e-01 2.13101953e-01 -8.81979823e-01 4.54375803e-01
7.66446471e-01 -7.20513403e-01 -2.31693983e-01 1.66362509e-01
7.86084533e-01 -2.42798001e-01 -8.59281003e-01 3.96568328e-01
2.79898226e-01 -9.26796913e-01 1.05106628e+00 -1.54860213e-01
4.48692858e-01 -5.46810925e-01 -2.83259362e-01 -1.08672035e+00
-4.44851637e-01 -7.20118940e-01 -9.18387413e-01 1.49814022e+00
2.32462585e-01 -1.16507463e-01 7.53240883e-01 4.33670700e-01
-1.49990413e-02 -9.79593992e-01 -7.93725908e-01 -5.31946957e-01
1.73636302e-01 -8.52418303e-01 6.01436675e-01 6.46629333e-01
-7.36001134e-01 3.03563297e-01 -7.57335603e-01 4.61395800e-01
1.15486658e+00 -1.08775288e-01 7.67664850e-01 -1.04752851e+00
-6.90260589e-01 -2.50863969e-01 1.67863667e-01 -1.22788227e+00
2.54038453e-01 -8.97613406e-01 8.51618871e-02 -1.76492822e+00
6.15927815e-01 -2.05384001e-01 -2.33612895e-01 2.11799309e-01
-4.42091137e-01 1.52886122e-01 -1.08804144e-01 5.56605637e-01
-6.95566177e-01 1.05200720e+00 1.17828381e+00 1.35093242e-01
1.84221715e-01 -1.06138848e-01 -3.58416855e-01 1.15872502e+00
3.17683220e-01 -7.07349062e-01 -6.43640935e-01 -5.41760743e-01
4.58788037e-01 -1.75435841e-01 6.96556091e-01 -9.15075958e-01
3.01183552e-01 -3.46048266e-01 7.40148067e-01 -9.21005547e-01
4.82913107e-01 -5.84005415e-01 4.12888080e-01 1.13893911e-01
-3.10667306e-01 -3.27382445e-01 -1.62588894e-01 6.70137644e-01
-8.23275894e-02 -2.95541555e-01 9.66439784e-01 -2.56587565e-01
-2.45699391e-01 7.08772540e-01 -3.32011640e-01 6.60540611e-02
7.66034305e-01 3.96157913e-02 3.15129936e-01 -3.01487684e-01
-8.48180175e-01 3.69229764e-01 6.87377751e-01 -6.67282566e-02
7.41921186e-01 -1.74730933e+00 -7.17187583e-01 5.06841481e-01
-1.62055790e-01 6.65989220e-01 3.99098039e-01 6.23440325e-01
-5.53750359e-02 -3.07830036e-01 2.51480252e-01 -8.04340363e-01
-7.65129983e-01 3.55168641e-01 2.02529311e-01 -7.04650342e-01
-5.78378558e-01 1.12903881e+00 6.09520614e-01 -1.51531115e-01
2.08288327e-01 -2.64248610e-01 1.05856113e-01 2.26525337e-01
3.70197296e-01 4.18420553e-01 -7.25848436e-01 -4.27912205e-01
-1.48680001e-01 5.50335765e-01 1.75521162e-03 -3.91692370e-01
1.28758848e+00 -3.40127259e-01 -2.05863729e-01 7.84617066e-01
1.18530083e+00 1.01411268e-01 -2.06910229e+00 -2.73058802e-01
-3.49526733e-01 -2.79099584e-01 1.20065153e-01 -4.85275567e-01
-7.97780216e-01 1.21924829e+00 2.58805573e-01 -1.95407793e-01
1.07993293e+00 1.96770966e-01 7.76528358e-01 -1.43908069e-01
1.07609324e-01 -1.30065358e+00 5.36806643e-01 4.10674334e-01
8.38674903e-01 -8.54320645e-01 2.71880418e-01 -1.50262013e-01
-6.13364458e-01 6.83981180e-01 5.05568326e-01 -2.59079903e-01
7.71868944e-01 3.77225548e-01 -4.06536072e-01 -6.83585107e-02
-8.69540155e-01 1.93912178e-01 4.39201921e-01 5.33991635e-01
2.90011078e-01 1.84803098e-01 2.73388445e-01 8.53311539e-01
4.64474820e-02 -2.36729207e-03 1.48165405e-01 5.31137168e-01
-1.51274949e-01 -1.29703331e+00 -1.93701938e-01 2.71797895e-01
-3.65924686e-01 -3.63235325e-01 2.99091786e-01 3.36278200e-01
6.70109808e-01 4.33182657e-01 1.43062457e-01 -6.50884258e-03
-2.21865907e-01 2.69861937e-01 5.29165149e-01 -6.78403854e-01
2.43803039e-01 6.08109891e-01 -5.20958185e-01 -5.29155374e-01
-4.14128244e-01 -8.36330831e-01 -1.08524811e+00 -2.34878482e-03
-3.15662771e-01 1.04805097e-01 4.67622101e-01 9.54983115e-01
3.13977599e-01 7.93872595e-01 5.03466368e-01 -1.11040032e+00
-5.05409241e-01 -1.06141508e+00 -7.09835708e-01 3.79850715e-01
2.97467679e-01 -7.11963058e-01 -5.27014911e-01 3.51769596e-01]
|
[7.142113208770752, 3.8205103874206543]
|
b7b6a8d9-437b-467b-bfcf-6dd787e9b73d
|
upop-unified-and-progressive-pruning-for
|
2301.13741
| null |
https://arxiv.org/abs/2301.13741v3
|
https://arxiv.org/pdf/2301.13741v3.pdf
|
UPop: Unified and Progressive Pruning for Compressing Vision-Language Transformers
|
Real-world data contains a vast amount of multimodal information, among which vision and language are the two most representative modalities. Moreover, increasingly heavier models, \textit{e}.\textit{g}., Transformers, have attracted the attention of researchers to model compression. However, how to compress multimodal models, especially vison-language Transformers, is still under-explored. This paper proposes the \textbf{U}nified and \textbf{P}r\textbf{o}gressive \textbf{P}runing (\textbf{\emph{UPop}}) as a universal vison-language Transformer compression framework, which incorporates 1) unifiedly searching multimodal subnets in a continuous optimization space from the original model, which enables automatic assignment of pruning ratios among compressible modalities and structures; 2) progressively searching and retraining the subnet, which maintains convergence between the search and retrain to attain higher compression ratios. Experiments on various tasks, datasets, and model architectures demonstrate the effectiveness and versatility of the proposed UPop framework. The code is available at https://github.com/sdc17/UPop.
|
['Jiaqi Wang', 'Chun Yuan', 'Zhendong Yang', 'Ying Jin', 'Chaofan Tao', 'Dachuan Shi']
|
2023-01-31
| null | null | null | null |
['visual-reasoning', 'visual-reasoning']
|
['computer-vision', 'reasoning']
|
[ 4.91831452e-01 -8.24009720e-03 -4.26424295e-01 -2.58625656e-01
-7.06944823e-01 -4.14991081e-01 4.03067052e-01 4.32411442e-03
-4.74458009e-01 4.23799127e-01 1.71515152e-01 -1.97869852e-01
-3.34135234e-01 -4.93055016e-01 -7.38116562e-01 -7.19146371e-01
3.32471222e-01 6.09020710e-01 2.36645788e-02 1.69824511e-02
1.93835869e-01 1.42398611e-01 -1.69298196e+00 3.24728757e-01
9.87640858e-01 1.43252790e+00 7.93553174e-01 4.50873762e-01
-3.16784710e-01 7.82083273e-01 -1.52534142e-01 -9.46056843e-01
4.13912162e-02 -2.66151458e-01 -8.83756816e-01 9.42257717e-02
4.36813533e-01 -3.36616635e-01 -8.00152123e-01 1.24333549e+00
3.71176571e-01 2.01359600e-01 3.92658740e-01 -1.25270367e+00
-6.25104427e-01 9.20358062e-01 -6.72164917e-01 2.83348262e-01
1.24680363e-01 3.98263186e-02 1.13184392e+00 -8.54272664e-01
4.95302826e-01 1.36593997e+00 2.65331864e-01 4.80634063e-01
-9.11420763e-01 -6.80577457e-01 2.11189240e-01 5.18091381e-01
-1.52784693e+00 -5.21992505e-01 6.89441204e-01 3.62142362e-02
1.01015604e+00 6.88493609e-01 5.93531907e-01 8.67641807e-01
4.45215106e-02 1.11568415e+00 8.08222950e-01 -2.06855938e-01
-9.44789797e-02 -3.01798061e-02 1.93799540e-01 9.29120243e-01
-1.54901370e-01 -1.74380362e-01 -8.25539231e-01 -8.64659622e-02
5.72990596e-01 1.81054220e-01 -4.09372866e-01 9.86334160e-02
-9.41439390e-01 5.34149110e-01 2.63072729e-01 1.79837704e-01
-2.53573835e-01 2.54856050e-01 3.85131508e-01 6.79718703e-02
9.33688805e-02 3.08085773e-02 -3.37679356e-01 -5.32267988e-01
-1.01707733e+00 1.10937990e-01 4.20762867e-01 1.42849493e+00
5.46945810e-01 -2.90469471e-02 -5.86419068e-02 1.31670725e+00
2.53519505e-01 5.92261255e-01 5.87127209e-01 -1.20719671e+00
8.61158490e-01 8.03316295e-01 -4.89951104e-01 -8.77875149e-01
-2.93588012e-01 -3.23800117e-01 -9.85980034e-01 -7.48606443e-01
-7.42025748e-02 9.62502882e-02 -8.64529967e-01 1.62586641e+00
1.39860600e-01 -6.36860430e-02 -2.40728468e-01 7.54217803e-01
9.03197646e-01 9.21880066e-01 1.55798599e-01 -2.55861431e-01
1.49546015e+00 -1.04873908e+00 -6.98099434e-01 -2.60862261e-01
3.87847036e-01 -8.71794164e-01 1.22838640e+00 4.37757432e-01
-1.52325177e+00 -4.45217907e-01 -6.75934076e-01 -4.16145831e-01
-1.96825683e-01 3.17310691e-01 4.54011381e-01 3.14184904e-01
-1.05940866e+00 3.93528759e-01 -8.15142214e-01 -1.91663980e-01
5.30081332e-01 5.24498701e-01 -3.22821707e-01 -3.16191018e-01
-9.75142062e-01 6.34278834e-01 7.90855289e-01 3.66841227e-01
-8.44224393e-01 -4.15422022e-01 -9.46077406e-01 2.36512691e-01
6.90018952e-01 -5.06786942e-01 1.08937109e+00 -7.39879072e-01
-1.17543960e+00 8.15260291e-01 -4.19500023e-01 -3.22102368e-01
1.52958855e-01 5.11361808e-02 -4.76320207e-01 3.61914963e-01
-3.15726757e-01 8.93162668e-01 9.29535508e-01 -1.17459178e+00
-7.72235632e-01 -5.23284733e-01 1.38054147e-01 4.44299340e-01
-7.44215250e-01 2.37114891e-01 -1.12995529e+00 -7.93151319e-01
3.49963397e-01 -9.13430095e-01 2.51715273e-01 -1.44688457e-01
-6.44480884e-01 -3.10640246e-01 7.30262578e-01 -9.30820227e-01
1.81928885e+00 -2.10228658e+00 6.55475676e-01 3.13084304e-01
5.91973305e-01 2.09608719e-01 -3.03209841e-01 4.04032320e-01
1.52859241e-01 1.09502010e-01 -2.99167216e-01 -5.01087904e-01
9.42408890e-02 4.49229479e-01 -1.76928878e-01 2.18898281e-01
-2.72512197e-01 9.33125496e-01 -3.64260912e-01 -8.99750769e-01
1.34701312e-01 2.44772971e-01 -5.30603230e-01 -2.55007241e-02
-2.75486261e-01 9.05577093e-03 -5.55801094e-01 1.04437912e+00
6.24204874e-01 -4.04329002e-01 1.64755255e-01 -4.68227506e-01
6.15750365e-02 9.26651359e-02 -8.88577044e-01 1.63479042e+00
-1.44689053e-01 4.69851792e-01 2.21241087e-01 -9.46857929e-01
7.08183229e-01 6.49780035e-02 5.85653901e-01 -1.02733600e+00
4.80726510e-01 7.51379728e-02 -1.17982283e-01 -6.45590365e-01
8.53692651e-01 -9.09578614e-03 -5.71278892e-02 2.64060616e-01
1.83084324e-01 -7.25376830e-02 5.75158834e-01 2.49456763e-01
6.97378993e-01 3.49852256e-02 -1.80955872e-01 4.42505740e-02
6.86814368e-01 -1.15647741e-01 4.27666813e-01 3.46899629e-01
-5.33808433e-02 2.69580156e-01 5.21285772e-01 -5.07333577e-02
-9.56838787e-01 -8.17481160e-01 -1.54097199e-01 1.46922994e+00
4.26024586e-01 -8.61760557e-01 -8.55307937e-01 -3.19219798e-01
-3.02124470e-01 6.42427325e-01 -2.82918990e-01 -2.24770889e-01
-6.45884097e-01 -7.02467263e-01 7.80931532e-01 5.10107040e-01
6.08221054e-01 -1.01421607e+00 -3.71529520e-01 -1.04033977e-01
-7.64945567e-01 -1.14047706e+00 -7.61821449e-01 4.94014397e-02
-9.88327444e-01 -8.84754777e-01 -5.92662096e-01 -9.23318386e-01
7.40609586e-01 9.50463042e-02 9.55009162e-01 2.44649827e-01
1.57173479e-03 3.85333687e-01 -2.63711751e-01 -2.82820374e-01
-1.66793481e-01 9.49561074e-02 -1.76448181e-01 -7.99918324e-02
2.84440726e-01 -4.61883038e-01 -3.21997911e-01 4.64390844e-01
-1.12768078e+00 5.21027207e-01 6.37092888e-01 5.83395362e-01
1.16475296e+00 2.08319902e-01 -7.45345801e-02 -5.10531247e-01
6.36667490e-01 -1.72652408e-01 -5.02601266e-01 4.61958975e-01
-4.73876506e-01 -2.35467032e-03 6.39301062e-01 -5.19212782e-01
-8.08330178e-01 -3.04992169e-01 -9.42440107e-02 -9.21221375e-01
2.03774780e-01 7.59445846e-01 -3.14959019e-01 1.09424233e-01
-1.96101181e-02 7.99177945e-01 -1.38599901e-02 -5.49588561e-01
2.09860921e-01 5.59569180e-01 6.12554312e-01 -6.40904605e-01
5.32347143e-01 2.81464189e-01 -1.75463349e-01 -8.64478111e-01
-5.38537383e-01 -1.95961758e-01 -1.71864688e-01 -2.75333613e-01
6.89411044e-01 -7.01834083e-01 -7.69846857e-01 3.56014282e-01
-9.07114089e-01 -1.29960284e-01 -1.97939962e-01 3.96719187e-01
-5.25626719e-01 6.13506794e-01 -8.04256201e-01 -6.74406111e-01
-5.18002629e-01 -1.34637594e+00 1.07574332e+00 4.71480817e-01
1.07247792e-01 -6.11697257e-01 -4.11475480e-01 9.86161292e-01
1.37514219e-01 -2.48099312e-01 1.28077769e+00 -4.53715742e-01
-6.66565955e-01 -1.53450817e-01 -3.87720644e-01 4.43994284e-01
-2.73227781e-01 -8.36399496e-02 -6.78512096e-01 -3.56558561e-01
-1.17671201e-02 -5.01446605e-01 8.23307633e-01 4.03370559e-01
1.67311084e+00 -6.20163083e-01 -3.08288395e-01 8.05269301e-01
1.02705073e+00 3.63900483e-01 7.06101179e-01 -2.88196607e-03
7.27071524e-01 4.20893103e-01 4.44466680e-01 4.64157045e-01
7.04911292e-01 4.35585439e-01 7.23659992e-01 2.10538104e-01
-8.11313242e-02 -4.05631423e-01 2.91790158e-01 1.37236392e+00
-4.26266879e-01 -4.06528145e-01 -7.14007497e-01 3.18148673e-01
-1.64253533e+00 -8.71123135e-01 2.87020087e-01 1.90480483e+00
7.62645841e-01 -3.57142054e-02 -1.25194833e-01 4.51225936e-02
6.89436615e-01 2.76920110e-01 -7.31736600e-01 -9.73225385e-02
-2.65749604e-01 6.72642291e-02 3.83086413e-01 1.87794223e-01
-8.48162472e-01 7.82751262e-01 4.70484114e+00 1.29440272e+00
-1.08877790e+00 1.48933917e-01 6.33896530e-01 -3.99294525e-01
-4.08901632e-01 -5.97033463e-02 -8.96318495e-01 6.26706243e-01
5.91473341e-01 -1.89886302e-01 9.54831362e-01 6.43056870e-01
7.28319213e-02 -1.39209377e-02 -9.04691696e-01 1.41719615e+00
2.95734227e-01 -1.17033768e+00 4.23442990e-01 2.35307276e-01
3.75277877e-01 -9.68105122e-02 2.79836327e-01 4.41827506e-01
-1.94134966e-01 -1.05796134e+00 1.08364296e+00 5.39629877e-01
9.41346109e-01 -7.08277643e-01 3.89868319e-01 5.03100455e-01
-1.58543074e+00 -2.51246125e-01 -5.33222735e-01 5.82822561e-01
1.84786081e-01 2.03010589e-01 -1.13124870e-01 6.70527518e-01
8.68244231e-01 6.89471006e-01 -6.22444034e-01 6.76486731e-01
1.08256280e-01 4.31971967e-01 -3.85869503e-01 7.34981745e-02
3.30028862e-01 -3.61512721e-01 5.79851747e-01 1.04047811e+00
4.16227639e-01 3.08565736e-01 1.95075080e-01 6.09492958e-01
-4.37179208e-01 1.84231952e-01 -5.10651469e-02 -3.82393301e-01
5.02017021e-01 1.07363939e+00 -6.47915781e-01 -1.48408711e-01
-2.08915368e-01 6.23912156e-01 3.13528180e-01 2.46375903e-01
-1.16537738e+00 -1.44677281e-01 2.11116880e-01 4.22175191e-02
3.58099222e-01 -6.70836344e-02 -1.16018625e-02 -1.18499410e+00
2.19402388e-01 -1.08132541e+00 7.54699528e-01 -9.10312951e-01
-8.72089207e-01 6.89299285e-01 2.98480570e-01 -1.20336568e+00
1.89632908e-01 -4.17503238e-01 -3.93384434e-02 6.61247551e-01
-1.32341814e+00 -1.25020528e+00 -3.91873628e-01 9.73017395e-01
8.05479646e-01 -2.49027297e-01 3.69426101e-01 5.16436219e-01
-8.29387188e-01 7.06095755e-01 -6.97983354e-02 -1.52911723e-01
6.05999790e-02 -6.81418061e-01 -2.48588145e-01 7.19249427e-01
6.57094941e-02 5.30088305e-01 4.15413529e-01 -5.33138394e-01
-1.88393188e+00 -9.01959360e-01 8.30833077e-01 7.00863302e-02
6.23842061e-01 -1.80728182e-01 -8.26744616e-01 7.68498361e-01
2.65052676e-01 -4.31114882e-01 5.42252541e-01 -3.31360728e-01
-2.58675724e-01 -2.58161962e-01 -9.77619946e-01 8.01195085e-01
1.19388163e+00 -5.29529870e-01 -3.66478473e-01 3.69305938e-01
6.16220117e-01 -5.56860805e-01 -9.82664585e-01 3.76639277e-01
4.99684960e-01 -7.56685197e-01 9.72120047e-01 -3.01388025e-01
6.93751752e-01 -1.93469524e-02 -4.75244105e-01 -6.18220806e-01
-6.05581030e-02 -6.47472799e-01 -6.16546988e-01 1.18730211e+00
4.10772800e-01 -4.23084557e-01 6.83236897e-01 6.40829682e-01
-1.89009994e-01 -1.07648969e+00 -1.20794916e+00 -4.35502887e-01
-2.48538002e-01 -5.71127176e-01 6.00452602e-01 6.95799053e-01
-1.11273326e-01 1.34143084e-01 -5.44955850e-01 -8.33222941e-02
5.52925408e-01 9.12212282e-02 3.97379369e-01 -7.99344718e-01
-3.75156701e-01 -6.83522046e-01 7.64146866e-03 -1.46135056e+00
3.21406648e-02 -1.26313925e+00 -2.85929799e-01 -1.43961072e+00
6.02645278e-01 -3.91035885e-01 -2.91256130e-01 6.91880465e-01
9.83750075e-02 1.64288893e-01 5.65434515e-01 5.82460165e-01
-7.54143834e-01 8.63819301e-01 1.49793947e+00 -3.09314638e-01
-8.39722008e-02 -2.81315595e-01 -6.63081229e-01 7.50807345e-01
6.19369626e-01 -3.00877869e-01 -6.88361943e-01 -8.61682773e-01
2.20896363e-01 2.92959899e-01 2.48605877e-01 -6.79110646e-01
4.47239757e-01 -1.80945158e-01 1.64464980e-01 -8.26989770e-01
8.39589179e-01 -9.30451691e-01 2.89390206e-01 3.57819319e-01
-4.04849589e-01 5.58224976e-01 3.05697322e-01 3.65461826e-01
-2.85301834e-01 -3.29352349e-01 7.43633449e-01 -1.15587838e-01
-7.23942816e-01 5.60002983e-01 -3.58389579e-02 7.85879418e-02
7.85825908e-01 -3.53018343e-01 -5.41411161e-01 -2.58965045e-01
-7.63102591e-01 5.00736117e-01 1.59718052e-01 4.74718839e-01
1.02943742e+00 -1.15965331e+00 -5.11628330e-01 1.33164108e-01
-3.68375555e-02 -1.08019775e-02 6.07401013e-01 1.16026497e+00
-4.69575316e-01 4.73202199e-01 7.14617223e-03 -7.23933697e-01
-1.63469613e+00 5.11729360e-01 2.81546623e-01 -3.58218372e-01
-4.72892821e-01 8.83475125e-01 1.29931882e-01 -2.52484590e-01
5.92128396e-01 -3.35850030e-01 -1.32930815e-01 9.21728238e-02
4.00472343e-01 6.86429501e-01 -1.56603530e-01 -8.94731998e-01
-2.08226755e-01 4.67377275e-01 -2.98038185e-01 4.54305522e-02
1.27376568e+00 -3.44980359e-01 -5.39238513e-01 1.27567708e-01
1.06739092e+00 -3.73877674e-01 -9.62918639e-01 -3.43350768e-01
-2.98489779e-01 -3.57059240e-01 6.44292384e-02 -7.68841743e-01
-1.38704383e+00 7.55800545e-01 4.63929892e-01 4.23584767e-02
1.64546013e+00 2.50442863e-01 1.09473860e+00 4.05352235e-01
4.15300131e-01 -1.30491078e+00 -4.12269011e-02 4.07314271e-01
9.29252326e-01 -5.79022586e-01 3.46710011e-02 -4.67083007e-01
-7.69843102e-01 9.00905848e-01 5.28579116e-01 3.93806458e-01
4.99444246e-01 2.24124432e-01 -5.12877941e-01 -2.05565780e-01
-7.24082768e-01 -5.99857382e-02 2.97662228e-01 1.47807077e-01
7.51925111e-02 8.85471553e-02 -3.32239419e-01 7.85288513e-01
-3.63311887e-01 2.78359391e-02 1.90070048e-02 1.04015660e+00
-3.66962463e-01 -9.36489165e-01 -3.24180037e-01 8.33740056e-01
-3.92325044e-01 -3.39828134e-01 -1.37193203e-01 6.13513589e-01
3.64056885e-01 8.10735226e-01 -8.66253600e-02 -4.73641545e-01
3.22291672e-01 -2.88937154e-04 6.08961105e-01 -7.44490400e-02
-4.58155811e-01 3.68897229e-01 -7.39311706e-03 -6.05133653e-01
-4.30680245e-01 -7.05954671e-01 -1.27719927e+00 -6.37379229e-01
-2.78688461e-01 -1.17212705e-01 5.15649974e-01 9.17203605e-01
3.01038057e-01 3.49116355e-01 3.14773351e-01 -6.76361799e-01
-5.53230822e-01 -8.33845377e-01 -5.58973908e-01 2.80078411e-01
-3.74009907e-02 -4.97807384e-01 -6.63671084e-03 2.31814399e-01]
|
[10.230467796325684, 0.9294155240058899]
|
d15cf739-6fc2-49bf-b792-523fad16b369
|
speech-enhancement-guided-by-contextual
|
2011.07442
| null |
https://arxiv.org/abs/2011.07442v5
|
https://arxiv.org/pdf/2011.07442v5.pdf
|
Improving Speech Enhancement Performance by Leveraging Contextual Broad Phonetic Class Information
|
Previous studies have confirmed that by augmenting acoustic features with the place/manner of articulatory features, the speech enhancement (SE) process can be guided to consider the broad phonetic properties of the input speech when performing enhancement to attain performance improvements. In this paper, we explore the contextual information of articulatory attributes as additional information to further benefit SE. More specifically, we propose to improve the SE performance by leveraging losses from an end-to-end automatic speech recognition (E2E-ASR) model that predicts the sequence of broad phonetic classes (BPCs). We also developed multi-objective training with ASR and perceptual losses to train the SE system based on a BPC-based E2E-ASR. Experimental results from speech denoising, speech dereverberation, and impaired speech enhancement tasks confirmed that contextual BPC information improves SE performance. Moreover, the SE model trained with the BPC-based E2E-ASR outperforms that with the phoneme-based E2E-ASR. The results suggest that objectives with misclassification of phonemes by the ASR system may lead to imperfect feedback, and BPC could be a potentially better choice. Finally, it is noted that combining the most-confusable phonetic targets into the same BPC when calculating the additional objective can effectively improve the SE performance.
|
['Yu Tsao', 'Shinji Watanabe', 'Jeih-weih Hung', 'Ching-Feng Liu', 'Cheng Yu', 'Chia-Yu Chang', 'Yen-Ju Lu']
|
2020-11-15
| null | null | null | null |
['speech-denoising', 'speech-dereverberation']
|
['speech', 'speech']
|
[ 3.95772278e-01 -1.83043443e-02 1.86891496e-01 -5.13558626e-01
-1.29550231e+00 -2.78306842e-01 2.68491209e-01 -2.93451667e-01
-4.81013536e-01 2.14869007e-01 7.27419019e-01 -5.05866647e-01
-1.03078440e-01 -3.44286859e-01 -6.33685648e-01 -7.08009243e-01
2.99993038e-01 -2.23701909e-01 -6.72390983e-02 -4.87959415e-01
1.29361078e-01 3.89026880e-01 -1.81464434e+00 4.79877740e-01
1.16051352e+00 1.00000894e+00 8.74666035e-01 9.85037446e-01
1.58130124e-01 3.98036182e-01 -7.91013300e-01 -3.87233913e-01
2.56833881e-01 -3.46266925e-01 -1.97774500e-01 -3.53258103e-02
1.99803784e-01 -3.69274855e-01 -2.32833534e-01 1.07401645e+00
9.74880576e-01 5.09042263e-01 3.72389138e-01 -8.33702564e-01
-7.28308976e-01 6.30016983e-01 -6.48406744e-02 8.05148333e-02
8.36458430e-02 1.97878033e-01 1.00197589e+00 -1.36748493e+00
6.83104992e-02 1.60756421e+00 7.05582261e-01 5.56112230e-01
-1.01222944e+00 -6.44952655e-01 2.08521456e-01 5.26114941e-01
-1.15303457e+00 -1.09473896e+00 8.59823465e-01 -9.65935811e-02
1.20093620e+00 5.16595781e-01 2.68261671e-01 9.71003473e-01
-2.03381941e-01 8.48508716e-01 1.11917198e+00 -6.01025701e-01
7.56004453e-02 9.68458503e-02 3.09159104e-02 2.07619503e-01
-3.87146235e-01 8.88758779e-01 -5.92907608e-01 1.30947635e-01
3.33289117e-01 -7.41890192e-01 -3.76058161e-01 5.20540833e-01
-6.67295516e-01 5.89543521e-01 2.19418317e-01 2.16600969e-01
-5.59068322e-01 7.98705220e-03 4.20151502e-01 3.11970979e-01
5.11035621e-01 5.40526390e-01 -6.33876681e-01 -4.73449200e-01
-8.38727057e-01 -1.64180920e-01 3.71132910e-01 4.65568691e-01
5.57380438e-01 6.60136640e-01 -4.01840329e-01 1.80131090e+00
4.98277783e-01 7.70575225e-01 6.04973435e-01 -1.16911387e+00
4.31147009e-01 -1.96319833e-01 -1.13158271e-01 -6.62247539e-01
-1.33396164e-02 -7.09165215e-01 -4.30141896e-01 3.44387740e-01
-7.74234310e-02 -1.53191030e-01 -1.16012669e+00 1.83414352e+00
2.52353791e-02 1.67902902e-01 2.12973043e-01 9.41710770e-01
4.62521106e-01 1.00832677e+00 2.65816420e-01 -2.70816654e-01
1.05649126e+00 -1.00534010e+00 -1.20979440e+00 -4.36122954e-01
5.26036263e-01 -1.06025839e+00 1.22912276e+00 4.10450220e-01
-1.24806082e+00 -1.09669602e+00 -9.09141362e-01 5.25737889e-02
-1.39278218e-01 4.71526980e-01 -1.11335106e-02 8.50750268e-01
-1.27743208e+00 5.80885530e-01 -6.83926642e-01 1.72159806e-01
-9.65927821e-03 2.64218688e-01 2.49644946e-02 7.78330043e-02
-1.47590768e+00 1.13728642e+00 3.30490410e-01 4.13773984e-01
-8.91226947e-01 -7.01151252e-01 -1.01919341e+00 2.08322942e-01
1.61802918e-01 -1.77234083e-01 1.35873318e+00 -9.40743148e-01
-1.96989691e+00 2.90219426e-01 -5.33335030e-01 -3.34588230e-01
3.96485329e-02 -3.75469565e-01 -8.10062885e-01 7.61683807e-02
-2.97391534e-01 7.08885670e-01 1.02453816e+00 -1.52836847e+00
-7.24879086e-01 -1.87962666e-01 -4.07383829e-01 6.05227649e-01
-3.82629573e-01 3.51082146e-01 -1.32828876e-01 -1.00681710e+00
2.90197320e-03 -7.56009698e-01 6.16990253e-02 -4.16861951e-01
-3.79921049e-01 -1.96999803e-01 7.91775107e-01 -1.54297519e+00
1.43110669e+00 -2.48834753e+00 -6.94003254e-02 1.14989035e-01
-4.30851460e-01 8.29947591e-01 -4.73509401e-01 1.99103311e-01
-1.51515380e-01 2.43361846e-01 -3.20294619e-01 -6.95669413e-01
1.37679487e-01 2.81902254e-01 -3.26435357e-01 -6.09210804e-02
4.89233881e-01 5.83858609e-01 -6.23167813e-01 -1.08741514e-01
5.01176238e-01 7.95935094e-01 -7.55068123e-01 3.15152407e-01
2.23538458e-01 3.02986205e-01 1.75025240e-01 4.25133854e-01
6.82261825e-01 5.99874139e-01 -2.22502202e-01 -3.34835738e-01
-1.04398988e-01 8.62663269e-01 -9.99740183e-01 1.03220141e+00
-8.63220572e-01 6.92139387e-01 5.55873871e-01 -9.23912704e-01
1.14004076e+00 5.70113122e-01 1.46130532e-01 -7.70594299e-01
-1.41977504e-01 5.27912617e-01 5.92990875e-01 -3.41474295e-01
6.75304115e-01 -3.03710848e-01 4.28123176e-01 -1.04313098e-01
6.46049008e-02 -4.14813638e-01 -2.94142216e-01 -2.95056641e-01
8.05050910e-01 -1.14340886e-01 2.80415872e-03 -6.18389761e-03
7.04638004e-01 -5.64464867e-01 7.34155834e-01 8.31919134e-01
-5.72083414e-01 6.82607710e-01 -2.54374623e-01 5.71217656e-01
-1.06537950e+00 -1.34184527e+00 -3.86948794e-01 1.32829154e+00
-3.24819386e-01 -1.51329100e-01 -8.28424275e-01 -2.83606321e-01
-2.14603081e-01 1.25027525e+00 -2.06064936e-02 -4.66177136e-01
-7.53550887e-01 -4.14285958e-01 7.18519747e-01 7.07898855e-01
3.07345450e-01 -1.21744478e+00 2.10700229e-01 5.33954561e-01
-3.98273408e-01 -1.03459156e+00 -7.46879339e-01 3.19791317e-01
-5.15897453e-01 -3.62305492e-01 -7.66051114e-01 -9.37603652e-01
3.18293810e-01 1.72384948e-01 4.67318982e-01 -8.05644020e-02
2.84043610e-01 4.14284438e-01 -6.30664468e-01 -4.59582984e-01
-1.13508582e+00 -2.39786759e-01 2.99305439e-01 -3.66507806e-02
2.03937963e-01 -4.22667086e-01 -4.05049622e-01 4.63999808e-01
-6.25384212e-01 -1.94398925e-01 5.19493580e-01 1.04738963e+00
5.60930908e-01 2.28553250e-01 1.26312375e+00 -2.21432224e-01
8.47370565e-01 -7.00059235e-02 -2.04464212e-01 1.65271033e-02
-6.89771473e-01 -1.30309731e-01 5.66200316e-01 -5.63416183e-01
-1.63632965e+00 -2.00843349e-01 -1.25522220e+00 -4.45974141e-01
-1.91310227e-01 5.64092755e-01 -6.01928890e-01 2.40105480e-01
4.99865085e-01 4.65538770e-01 1.60593867e-01 -5.63979626e-01
3.30113769e-01 1.42311072e+00 6.50815606e-01 -5.69314897e-01
5.54857433e-01 -1.70571432e-01 -6.24889374e-01 -1.18303180e+00
-5.39329350e-01 -5.78196228e-01 -1.03718936e-01 -2.95961201e-01
5.22123277e-01 -9.73244011e-01 -3.23932588e-01 7.83175766e-01
-1.20819426e+00 -4.47817743e-01 -2.67891407e-01 9.03505087e-01
-7.21321106e-01 4.40705478e-01 -7.37703919e-01 -1.25239897e+00
-3.69642496e-01 -1.52233660e+00 9.33561563e-01 -1.03203356e-01
-7.65851513e-02 -7.10440695e-01 -2.40149468e-01 6.49836838e-01
7.72439122e-01 -7.25991428e-01 9.32291806e-01 -7.35415518e-01
-1.20976821e-01 1.82488069e-01 1.25819087e-01 1.37284827e+00
4.21320587e-01 4.86338139e-02 -1.49109447e+00 -1.67362660e-01
3.50100964e-01 1.00257732e-01 8.15044701e-01 8.16248596e-01
1.20643520e+00 -4.23490375e-01 1.81477457e-01 4.66222763e-01
8.89976501e-01 7.60058284e-01 1.00067365e+00 9.11102891e-02
3.55231643e-01 7.99312592e-01 7.62164533e-01 8.22772756e-02
1.96472943e-01 7.94930756e-01 2.24765882e-01 -1.75077334e-01
-8.18563879e-01 -2.94642478e-01 9.12226856e-01 1.58495212e+00
1.85783774e-01 -2.62299240e-01 -5.15563905e-01 6.79263175e-01
-1.21982181e+00 -9.14148569e-01 1.08328968e-01 2.16833639e+00
1.16091621e+00 1.30088240e-01 -2.38484532e-01 4.94203120e-01
1.23950803e+00 1.78931132e-01 -4.00770187e-01 -8.14942360e-01
-3.71610224e-01 3.13310683e-01 2.13061303e-01 9.33989346e-01
-6.93879545e-01 9.45233226e-01 5.91543531e+00 1.17417252e+00
-1.12404692e+00 1.78363115e-01 5.48471510e-01 1.57798976e-01
-3.35486382e-01 -2.18948320e-01 -9.34079111e-01 5.54998577e-01
1.24025965e+00 1.04705185e-01 8.26101780e-01 5.89959919e-01
8.93264353e-01 1.64726377e-01 -7.50597298e-01 6.43570900e-01
-1.23500966e-01 -7.64136910e-01 -1.33351982e-01 -4.51767258e-02
5.24621308e-01 -3.26844044e-02 4.55441862e-01 5.74913204e-01
1.10623248e-01 -8.32342207e-01 8.77139568e-01 2.29447022e-01
7.29182422e-01 -5.83825588e-01 6.44979179e-01 2.56979287e-01
-1.20189309e+00 -3.38039577e-01 -2.19779864e-01 1.94805130e-01
4.81541902e-01 4.02683794e-01 -1.12002659e+00 3.79412293e-01
4.97846842e-01 3.56556326e-01 -1.92041919e-01 1.05722749e+00
-4.39584732e-01 1.21864426e+00 -2.58147627e-01 2.01486707e-01
-1.32949889e-01 -1.69125162e-02 9.84305024e-01 1.44359791e+00
6.44726336e-01 7.20273878e-04 -3.21499825e-01 5.43156683e-01
1.47035411e-02 1.63959652e-01 -2.41091937e-01 -3.69525924e-02
9.98585999e-01 7.75746882e-01 7.00667277e-02 -2.54544437e-01
-2.54287124e-01 7.68897414e-01 4.02733274e-02 5.23095667e-01
-6.13456666e-01 -3.57870907e-01 1.10336876e+00 -1.21832266e-01
4.44265902e-01 4.28773550e-04 -3.57588738e-01 -5.10289669e-01
-6.62147477e-02 -1.10683501e+00 -1.54395252e-01 -1.21603858e+00
-1.33392918e+00 7.05273986e-01 -3.07317883e-01 -9.24438000e-01
-2.91599244e-01 -7.94372022e-01 -5.87004423e-01 1.21045160e+00
-1.83987284e+00 -7.98954308e-01 2.89271235e-01 3.21059912e-01
8.80234361e-01 2.03773100e-02 6.71966553e-01 6.31141782e-01
-4.48293716e-01 8.46580386e-01 2.19227910e-01 -1.73092987e-02
8.74734163e-01 -1.10284185e+00 2.63914168e-01 1.03837407e+00
-7.95819908e-02 5.01796484e-01 7.83457637e-01 -6.80294871e-01
-7.83991754e-01 -1.25790346e+00 8.21869910e-01 8.78860131e-02
3.08324456e-01 9.50109214e-03 -1.18464875e+00 3.65453809e-01
6.31179363e-02 -3.29007059e-01 6.54559612e-01 1.71033219e-02
-8.18207934e-02 -2.73476213e-01 -1.03087389e+00 7.07192481e-01
1.05039728e+00 -1.07665884e+00 -8.69748771e-01 -2.09839195e-01
1.14726937e+00 -2.53783941e-01 -8.50730002e-01 6.98296487e-01
3.09043288e-01 -4.75653052e-01 1.16370142e+00 -2.60357976e-01
2.32279643e-01 -3.18719834e-01 -6.99648321e-01 -2.08226013e+00
-9.89626497e-02 -5.31490564e-01 2.19713245e-03 1.64382243e+00
5.79363465e-01 -6.23013437e-01 2.43640602e-01 3.65552038e-01
-1.03737521e+00 -3.05851072e-01 -1.09159315e+00 -1.02704787e+00
1.40933841e-01 -1.01146030e+00 4.98256683e-01 4.56687838e-01
-2.37795994e-01 -5.17296419e-02 -4.33845192e-01 5.75146019e-01
2.29847193e-01 -6.77190244e-01 2.92953044e-01 -8.44668984e-01
-4.35243547e-01 -4.91425574e-01 -1.80149917e-02 -1.24436915e+00
2.82433152e-01 -9.33596969e-01 7.50302613e-01 -1.41455042e+00
-4.94911283e-01 -5.03473878e-01 -5.39018214e-01 3.38427454e-01
-6.61792457e-01 -2.70609379e-01 4.68173444e-01 -1.21801302e-01
1.18994169e-01 9.58925784e-01 1.33810234e+00 -1.12522282e-01
-4.58789706e-01 3.70059520e-01 -6.47650540e-01 6.63337708e-01
7.50895977e-01 -3.92044276e-01 -2.49067307e-01 -3.88880789e-01
-3.64885479e-01 2.75384426e-01 2.19858944e-01 -8.20074081e-01
2.27558941e-01 5.25246710e-02 -8.67287219e-02 -6.61268055e-01
9.92177308e-01 -7.24284828e-01 -1.51770592e-01 2.58675992e-01
-5.43292820e-01 -3.33115876e-01 4.73217458e-01 2.76586562e-01
-5.41112125e-01 -4.80171859e-01 8.50090742e-01 3.03933144e-01
-6.11807823e-01 -2.40105316e-01 -7.72567332e-01 -2.37761304e-01
3.54933351e-01 -3.37225884e-01 -2.65925199e-01 -5.77838898e-01
-7.92258024e-01 -1.37398437e-01 -1.01068191e-01 4.99751598e-01
8.57671857e-01 -1.24211502e+00 -8.88260126e-01 3.67835462e-01
-2.18115896e-01 -4.34446812e-01 4.86905098e-01 6.04148984e-01
2.32384339e-01 3.58437270e-01 9.88807678e-02 -4.02272105e-01
-1.33909416e+00 1.68941185e-01 5.73057234e-01 5.82515299e-02
-1.46647170e-01 8.46481204e-01 2.88432270e-01 -7.80522108e-01
4.36149120e-01 -1.49935186e-01 -2.01167911e-01 -1.21501461e-01
3.52776080e-01 7.06874847e-01 3.88806194e-01 -8.87875974e-01
-1.65264904e-01 2.62131363e-01 1.12987712e-01 -5.86587429e-01
1.23856771e+00 -6.37277067e-01 3.84806931e-01 1.57297894e-01
1.22732043e+00 4.78992611e-01 -1.44491792e+00 -1.63899139e-01
-2.06008554e-01 -4.01155978e-01 5.74782133e-01 -1.34135664e+00
-9.12921846e-01 1.12835276e+00 8.10476422e-01 4.42414097e-02
1.55197179e+00 -3.04231316e-01 8.47849965e-01 8.97389352e-02
-2.52790842e-02 -1.48634219e+00 -6.54272269e-03 7.48042107e-01
1.17981851e+00 -1.15357685e+00 -6.80473328e-01 -5.24230957e-01
-9.04626966e-01 1.00444829e+00 5.24787962e-01 3.34171623e-01
7.58387506e-01 4.00485098e-01 2.75332838e-01 4.05046523e-01
-5.32942235e-01 -5.11339664e-01 4.53215927e-01 9.13864017e-01
2.30367228e-01 2.17496574e-01 -2.48097405e-01 7.28234708e-01
-5.19196928e-01 -4.01779562e-01 2.80328691e-01 3.85963410e-01
-6.24757528e-01 -1.22681355e+00 -6.55866385e-01 4.06491071e-01
-4.74811047e-01 -5.89704216e-01 8.47800821e-02 2.87273020e-01
-4.54074405e-02 1.56727755e+00 1.21071162e-02 -5.90626299e-01
5.82095087e-01 3.00900847e-01 -2.31070668e-02 -5.94125628e-01
-6.48099780e-01 3.31599534e-01 3.98280382e-01 -2.38639042e-01
-1.49437800e-01 -6.06005967e-01 -1.23381793e+00 9.72132012e-02
-7.69440174e-01 -2.47555636e-02 1.10495830e+00 8.87636721e-01
4.27154928e-01 8.54976237e-01 1.01437795e+00 -6.20330334e-01
-9.76297796e-01 -1.26787460e+00 -3.97690773e-01 3.61894459e-01
5.53987026e-01 -5.52896559e-01 -7.98552096e-01 9.00648311e-02]
|
[14.868931770324707, 6.1361613273620605]
|
4f2d2a06-cb35-4357-929e-9b2abfaa461a
|
contrastive-distillation-is-a-sample
|
2212.11353
| null |
https://arxiv.org/abs/2212.11353v1
|
https://arxiv.org/pdf/2212.11353v1.pdf
|
Contrastive Distillation Is a Sample-Efficient Self-Supervised Loss Policy for Transfer Learning
|
Traditional approaches to RL have focused on learning decision policies directly from episodic decisions, while slowly and implicitly learning the semantics of compositional representations needed for generalization. While some approaches have been adopted to refine representations via auxiliary self-supervised losses while simultaneously learning decision policies, learning compositional representations from hand-designed and context-independent self-supervised losses (multi-view) still adapts relatively slowly to the real world, which contains many non-IID subspaces requiring rapid distribution shift in both time and spatial attention patterns at varying levels of abstraction. In contrast, supervised language model cascades have shown the flexibility to adapt to many diverse manifolds, and hints of self-learning needed for autonomous task transfer. However, to date, transfer methods for language models like few-shot learning and fine-tuning still require human supervision and transfer learning using self-learning methods has been underexplored. We propose a self-supervised loss policy called contrastive distillation which manifests latent variables with high mutual information with both source and target tasks from weights to tokens. We show how this outperforms common methods of transfer learning and suggests a useful design axis of trading off compute for generalizability for online transfer. Contrastive distillation is improved through sampling from memory and suggests a simple algorithm for more efficiently sampling negative examples for contrastive losses than random sampling.
|
['Charysse Redwood', 'François Charton', 'Kurt Shuster', 'Hugh Leather', 'Amy Zhang', 'Gabriel Synnaeve', 'Chris Lengerich']
|
2022-12-21
| null | null | null | null |
['self-learning']
|
['natural-language-processing']
|
[ 1.11142725e-01 2.64948130e-01 -5.22224247e-01 -4.98960167e-01
-7.77786553e-01 -4.86810982e-01 9.90314543e-01 -1.34311423e-01
-5.52440703e-01 1.06289566e+00 3.48577529e-01 2.05139462e-02
-1.32852912e-01 -8.75014424e-01 -7.58876264e-01 -7.04990506e-01
-3.61170739e-01 8.16239476e-01 1.20171003e-01 -3.42329264e-01
3.47695462e-02 3.55417848e-01 -1.52012455e+00 2.86424041e-01
6.98070228e-01 7.46413767e-01 3.21033597e-01 6.23176455e-01
-3.99170667e-01 8.83404672e-01 -2.83300608e-01 -1.02539420e-01
2.93241382e-01 -8.09957325e-01 -7.67949283e-01 -2.17693210e-01
2.80409902e-01 -1.29479542e-01 -3.98486614e-01 9.07604694e-01
6.27518356e-01 6.28851533e-01 1.21237445e+00 -1.08605933e+00
-1.00908494e+00 6.28066838e-01 -1.93392321e-01 2.48097315e-01
-5.93314972e-03 4.65304643e-01 1.09490919e+00 -9.21784163e-01
6.46152675e-01 1.44876170e+00 6.92724943e-01 9.78918910e-01
-1.75113690e+00 -8.56678426e-01 2.49140158e-01 1.21601276e-01
-9.70044851e-01 -4.74631190e-01 7.31020391e-01 -3.70125413e-01
1.11712861e+00 -9.14984047e-02 6.48749590e-01 1.50770140e+00
1.18513271e-01 9.95608568e-01 1.24337780e+00 -3.18086565e-01
5.30693650e-01 5.78314006e-01 -2.88143337e-01 6.88189209e-01
-2.98730526e-02 3.76069307e-01 -7.28929818e-01 -1.45294055e-01
8.58302951e-01 2.39771962e-01 4.62839864e-02 -9.72448528e-01
-1.08176494e+00 1.06582928e+00 5.02398252e-01 3.00143331e-01
-1.47140786e-01 3.19804937e-01 6.80396080e-01 1.02145278e+00
6.87246740e-01 7.76553154e-01 -7.03497171e-01 4.76195402e-02
-9.80740964e-01 1.89219832e-01 8.74365687e-01 9.54357147e-01
1.19953179e+00 3.44617277e-01 -3.14157903e-01 6.90221429e-01
-5.74139878e-02 4.95189667e-01 1.01401651e+00 -9.40650105e-01
3.53215694e-01 3.09640586e-01 -1.42007502e-05 -4.58425313e-01
-3.30626249e-01 -4.80743885e-01 -7.97669411e-01 5.76822400e-01
3.16246957e-01 -3.13474655e-01 -8.64007115e-01 2.08276820e+00
5.61073311e-02 -3.31299715e-02 2.21290097e-01 6.90801978e-01
7.00709820e-02 7.47834682e-01 2.51773298e-01 -1.91606626e-01
7.89691687e-01 -8.50098372e-01 -4.97346699e-01 -1.58569366e-01
6.86521113e-01 -9.33805183e-02 1.61163747e+00 2.35714093e-01
-1.30951941e+00 -4.90683436e-01 -9.45128500e-01 -1.21041134e-01
-5.15121639e-01 -3.75372142e-01 6.13691330e-01 3.26042026e-01
-1.18994009e+00 9.70669508e-01 -8.28677118e-01 -5.40019989e-01
7.48876393e-01 5.07599711e-01 -2.78978229e-01 1.26735196e-01
-1.42373371e+00 1.17228687e+00 3.87173980e-01 -6.63316011e-01
-1.23357260e+00 -1.06947362e+00 -8.09896111e-01 9.68986526e-02
8.50216523e-02 -9.98099744e-01 1.21774125e+00 -1.62171352e+00
-2.10888076e+00 8.99334490e-01 1.32719889e-01 -8.76666009e-01
4.77208614e-01 -1.99305952e-01 -1.52829170e-01 2.23094046e-01
2.25969013e-02 9.41323102e-01 1.33486462e+00 -9.85918820e-01
-4.58926082e-01 -2.71144658e-01 -6.38441294e-02 6.27399206e-01
-6.64096832e-01 -2.91931212e-01 3.81435484e-01 -8.09165955e-01
-3.83281022e-01 -5.06271899e-01 -1.76394016e-01 3.44313473e-01
1.99088007e-01 -2.50442892e-01 7.36044228e-01 -2.51010895e-01
9.14309740e-01 -1.90541255e+00 5.22720635e-01 -5.22941016e-02
-7.38304332e-02 4.80801463e-02 -2.92695433e-01 5.22941589e-01
6.82186196e-03 -9.52410251e-02 -4.31949288e-01 -2.97332585e-01
1.64677203e-01 2.06503510e-01 -6.79020047e-01 5.03764868e-01
3.38092059e-01 1.03582263e+00 -1.34831989e+00 -6.20125756e-02
1.19123096e-03 2.41144925e-01 -7.99161255e-01 4.27302122e-01
-6.02858365e-01 4.50576186e-01 -2.64299184e-01 2.43788153e-01
4.69950996e-02 -2.02956930e-01 -1.00215487e-01 3.36669326e-01
1.75205618e-01 4.92466241e-01 -7.95944512e-01 2.04698372e+00
-7.26172090e-01 5.84742963e-01 1.04277618e-01 -1.40917921e+00
9.87319767e-01 2.77874708e-01 3.58478814e-01 -7.61114538e-01
-1.62035242e-01 1.43857449e-01 -3.19408655e-01 -3.41562867e-01
2.22810403e-01 -7.43845224e-01 -9.24178585e-02 8.19990337e-01
6.95246041e-01 -4.02003944e-01 -3.16073895e-01 1.47500873e-01
1.00975978e+00 6.14387929e-01 3.08992803e-01 -4.65746611e-01
1.56510100e-01 2.84787584e-02 2.80487776e-01 7.29215562e-01
-2.22686395e-01 4.36223716e-01 2.82484978e-01 -4.22597796e-01
-1.19228959e+00 -1.39078188e+00 1.32989109e-01 1.70379555e+00
-2.10862294e-01 -3.68527472e-02 -4.35388595e-01 -7.35149562e-01
1.95631981e-01 1.02092659e+00 -8.17928612e-01 -7.12457061e-01
-5.09490967e-01 -4.76028115e-01 5.13880908e-01 5.31505525e-01
2.68108398e-01 -1.43416655e+00 -6.50583982e-01 3.73850346e-01
4.13681328e-01 -2.64155626e-01 -7.35885203e-01 7.63706565e-01
-1.16755712e+00 -7.05434918e-01 -1.06037223e+00 -7.94677973e-01
4.77349669e-01 2.26156376e-02 1.15173733e+00 -5.47024250e-01
-3.01120996e-01 8.87050211e-01 6.95451498e-02 -2.92735398e-01
-3.05836290e-01 3.32726806e-01 4.09427941e-01 -8.27299282e-02
3.66777241e-01 -1.05209982e+00 -6.33634746e-01 4.36052820e-03
-9.06895101e-01 -1.78619489e-01 6.03422105e-01 1.22492290e+00
4.44513947e-01 -5.93464732e-01 1.05456364e+00 -9.90819156e-01
8.37222040e-01 -9.11953807e-01 -2.96131551e-01 1.11044228e-01
-8.50955904e-01 6.77713931e-01 9.25972998e-01 -9.31851208e-01
-1.25258470e+00 -1.14725411e-01 3.72908384e-01 -7.76255488e-01
-9.18962732e-02 -7.70967687e-03 6.13857619e-02 1.93249911e-01
1.09164369e+00 4.13100004e-01 4.20664132e-01 -2.13366821e-01
9.86342192e-01 2.33964831e-01 2.15040043e-01 -8.73330593e-01
7.66530633e-01 5.99090040e-01 -3.13397318e-01 -7.16308475e-01
-8.63999188e-01 -2.42656693e-01 -6.66987360e-01 1.29558131e-01
7.18579888e-01 -9.40358400e-01 -2.46327356e-01 1.19711518e-01
-6.38518691e-01 -1.01088011e+00 -1.25759995e+00 3.93859714e-01
-1.27948964e+00 -1.85661256e-01 -6.60641193e-01 -7.87772179e-01
-3.42778504e-01 -4.47285235e-01 7.34268129e-01 -8.39005634e-02
-5.01157463e-01 -1.43813455e+00 4.78838325e-01 -3.63345265e-01
8.32312286e-01 -1.29348233e-01 9.81113672e-01 -6.70374930e-01
-4.94269252e-01 2.07007527e-01 1.25603318e-01 3.38634908e-01
2.12613747e-01 -7.15060771e-01 -9.54316497e-01 -5.55732667e-01
9.58500952e-02 -1.10077691e+00 1.26721978e+00 2.41449237e-01
1.03627515e+00 -5.57913542e-01 -2.31431723e-01 7.84736395e-01
1.14501584e+00 -1.76216960e-01 2.46469229e-01 3.14796299e-01
4.55417871e-01 7.98727453e-01 3.17051768e-01 3.93513739e-01
2.03445941e-01 1.41782850e-01 1.27868682e-01 8.07252452e-02
-4.58007678e-02 -6.35913551e-01 7.82656252e-01 6.44571900e-01
2.49157712e-01 3.35990161e-01 -5.22247076e-01 6.02444649e-01
-1.72720528e+00 -1.19806457e+00 9.95004117e-01 2.23100972e+00
1.25812066e+00 1.61028311e-01 3.19690406e-01 -3.44045907e-01
4.47070241e-01 2.43025616e-01 -1.23133850e+00 -3.37025344e-01
-9.77384672e-02 4.11062032e-01 5.00934482e-01 5.16493082e-01
-8.96358430e-01 1.25387251e+00 6.84865427e+00 8.78691912e-01
-1.18433332e+00 3.01982075e-01 7.13266253e-01 -4.53512639e-01
-6.67643428e-01 -1.59313099e-03 -7.85762846e-01 8.15329701e-02
1.14385915e+00 -3.06333214e-01 7.40628541e-01 8.15203369e-01
-1.47120077e-02 4.42150176e-01 -1.35324597e+00 8.91709805e-01
2.12368444e-02 -1.37053823e+00 2.62447685e-01 -2.35790089e-01
9.97420490e-01 1.57849073e-01 2.78632522e-01 9.85767543e-01
8.05552721e-01 -1.08329189e+00 7.40197301e-01 7.31693208e-01
9.57725346e-01 -6.07627034e-01 -1.19169340e-01 5.06523371e-01
-7.83141375e-01 -3.95721406e-01 -5.22230625e-01 -2.07192793e-01
-2.35169232e-01 -1.12578878e-02 -6.84830785e-01 -8.08271021e-02
4.01241362e-01 9.15538609e-01 -1.44131303e-01 5.24067342e-01
-9.62136313e-02 5.93903542e-01 -1.08126633e-01 -2.50642687e-01
4.50193584e-01 -2.83463866e-01 5.91256320e-01 1.25634325e+00
2.98386723e-01 -1.48287177e-01 1.76660597e-01 1.13368869e+00
-2.13833302e-01 9.31907818e-02 -1.08385086e+00 -4.34032790e-02
2.95438915e-01 8.18553865e-01 -3.73593003e-01 -4.38098133e-01
-2.79833466e-01 1.15415919e+00 8.81779373e-01 8.22834611e-01
-3.98275942e-01 -2.53192872e-01 8.24611247e-01 3.87709886e-01
5.71507990e-01 -1.80567041e-01 -1.46975875e-01 -1.26484978e+00
-3.57433170e-01 -7.06613421e-01 5.10330379e-01 -4.03008759e-01
-1.70408297e+00 3.12976301e-01 1.00467257e-01 -1.22864711e+00
-5.42378068e-01 -5.15975893e-01 -7.58260727e-01 9.06601846e-01
-1.52530038e+00 -1.01492298e+00 4.43618953e-01 9.71699715e-01
8.52246404e-01 -7.35963404e-01 1.01538229e+00 -2.31674612e-01
-2.32756697e-02 6.59509301e-01 3.64692092e-01 -2.49643177e-01
1.00925732e+00 -1.46034122e+00 7.79307112e-02 1.62189156e-01
4.56008762e-02 6.21760488e-01 6.48812950e-01 -3.29479694e-01
-1.31842458e+00 -1.20213604e+00 4.13409144e-01 -3.71907473e-01
9.47166264e-01 -5.23192525e-01 -1.02634609e+00 7.06108332e-01
3.61992836e-01 8.41127932e-02 5.08842170e-01 1.13792099e-01
-5.92964232e-01 -3.49818438e-01 -1.13889885e+00 7.60242701e-01
1.06100118e+00 -8.95530641e-01 -7.74855852e-01 3.28793287e-01
8.60898256e-01 1.67125806e-01 -5.53990245e-01 -6.83697015e-02
3.28561276e-01 -7.27792084e-01 9.20895576e-01 -1.11949027e+00
9.77033377e-02 1.26158506e-01 -2.37551285e-03 -1.66931820e+00
-4.92701799e-01 -9.87886608e-01 -4.42090034e-01 8.75386596e-01
4.14554596e-01 -8.16915393e-01 7.51180053e-01 2.89268315e-01
-4.10097726e-02 -7.60186195e-01 -7.94365227e-01 -9.21327353e-01
7.48090804e-01 -1.23568930e-01 9.93430018e-02 1.04303920e+00
1.97372362e-01 6.07992470e-01 -3.13115865e-01 -4.55282539e-01
6.67947531e-01 1.63802937e-01 4.90976244e-01 -1.04165363e+00
-4.79026258e-01 -6.89885676e-01 -1.25788301e-01 -1.21712136e+00
6.00855887e-01 -1.35926282e+00 -5.16673960e-02 -1.13119423e+00
1.75383866e-01 -4.09497172e-01 -5.31957030e-01 3.06413233e-01
6.65437132e-02 -2.78443813e-01 -4.22746614e-02 3.50893229e-01
-6.81967080e-01 1.20054245e+00 1.27107549e+00 -3.72017592e-01
-6.19731605e-01 5.71194664e-02 -7.14137912e-01 5.95916331e-01
8.96776617e-01 -6.09549224e-01 -8.31939816e-01 -1.99284375e-01
1.09106459e-01 -9.03365389e-02 1.77238196e-01 -8.40353251e-01
2.95846552e-01 -2.79037178e-01 4.72201377e-01 -1.62547320e-01
4.24711019e-01 -5.83241582e-01 -3.92971098e-01 5.49385309e-01
-9.86833572e-01 -1.02945037e-01 1.66044161e-02 9.81891692e-01
-6.71114400e-02 -7.58029670e-02 9.16004360e-01 -5.94844759e-01
-6.08513236e-01 5.83820164e-01 -5.11280656e-01 5.82352877e-01
9.66086566e-01 -1.84599429e-01 1.18643835e-01 -5.58826149e-01
-1.10844684e+00 2.93900549e-01 4.01769966e-01 3.57003033e-01
6.60519481e-01 -1.40127933e+00 -6.47204041e-01 3.01966608e-01
-7.20695639e-03 -1.96540520e-01 1.68410931e-02 4.44749504e-01
-3.44305485e-02 1.70052931e-01 -5.42093873e-01 -4.30475444e-01
-4.20694381e-01 8.27930391e-01 4.00610924e-01 -2.85761476e-01
-8.20097983e-01 8.68766427e-01 4.43287790e-01 -8.18799198e-01
5.11284173e-01 -2.51920223e-01 3.29103395e-02 3.39103550e-01
5.04793704e-01 3.00843596e-01 -3.17054152e-01 -5.81823587e-02
2.07415044e-01 2.32004821e-01 -1.34182855e-01 -4.17456657e-01
1.42853498e+00 -2.44127557e-01 1.72884151e-01 1.16971850e+00
1.28220928e+00 -5.95272362e-01 -1.94149375e+00 -5.80697596e-01
2.07146794e-01 -1.29304649e-02 -2.87680566e-01 -5.69832206e-01
-6.34854972e-01 1.25005233e+00 6.06674910e-01 -2.40872148e-02
8.56662512e-01 6.29067123e-02 5.86021423e-01 8.85466814e-01
4.25166458e-01 -1.38938963e+00 7.91039944e-01 7.28531897e-01
1.06439579e+00 -1.13926923e+00 -3.57962251e-01 4.79557604e-01
-8.21449339e-01 1.09978318e+00 5.84275126e-01 -6.73507154e-01
8.78245473e-01 1.21719219e-01 -1.66352376e-01 3.91636044e-02
-1.08738124e+00 -1.90096870e-01 5.44010773e-02 8.60354841e-01
2.35198826e-01 3.89154628e-03 3.15577120e-01 3.78904939e-01
-8.40371300e-04 -1.56219229e-01 -6.77694231e-02 7.52786458e-01
-6.87040031e-01 -9.78140414e-01 -7.66098723e-02 5.23163736e-01
-1.05447218e-01 -1.44934863e-01 -8.55187997e-02 6.91320777e-01
-3.01636517e-01 1.53780371e-01 1.28767982e-01 -2.78329290e-02
5.56952506e-02 5.18230259e-01 6.69568717e-01 -9.73686695e-01
-5.57321787e-01 -2.00305477e-01 -3.27070445e-01 -5.13157308e-01
-2.18982667e-01 -7.46386051e-01 -1.26735747e+00 4.83759865e-02
1.29525587e-01 6.38543889e-02 2.56152719e-01 8.22263300e-01
3.77217680e-01 4.57152724e-01 6.20159805e-01 -1.19680870e+00
-1.34102392e+00 -1.01690900e+00 -7.17875123e-01 6.09854877e-01
5.41613340e-01 -7.02064216e-01 -5.13992131e-01 1.12284631e-01]
|
[4.254665374755859, 1.6076141595840454]
|
5c323702-2e1f-43ae-8a5c-f722730fc0e3
|
automatic-differentiation-to-simultaneously
|
2009.0881
| null |
https://arxiv.org/abs/2009.08810v2
|
https://arxiv.org/pdf/2009.08810v2.pdf
|
Automatic Differentiation to Simultaneously Identify Nonlinear Dynamics and Extract Noise Probability Distributions from Data
|
The sparse identification of nonlinear dynamics (SINDy) is a regression framework for the discovery of parsimonious dynamic models and governing equations from time-series data. As with all system identification methods, noisy measurements compromise the accuracy and robustness of the model discovery procedure. In this work, we develop a variant of the SINDy algorithm that integrates automatic differentiation and recent time-stepping constrained motivated by Rudy et al. for simultaneously (i) denoising the data, (ii) learning and parametrizing the noise probability distribution, and (iii) identifying the underlying parsimonious dynamical system responsible for generating the time-series data. Thus within an integrated optimization framework, noise can be separated from signal, resulting in an architecture that is approximately twice as robust to noise as state-of-the-art methods, handling as much as 40% noise on a given time-series signal and explicitly parametrizing the noise probability distribution. We demonstrate this approach on several numerical examples, from Lotka-Volterra models to the spatio-temporal Lorenz 96 model. Further, we show the method can identify a diversity of probability distributions including Gaussian, uniform, Gamma, and Rayleigh.
|
['J. Nathan Kutz', 'Steven L. Brunton', 'Kadierdan Kaheman']
|
2020-09-12
| null | null | null | null |
['model-discovery']
|
['miscellaneous']
|
[ 1.30636171e-02 -4.15045589e-01 4.12825495e-01 2.56393909e-01
-6.07350767e-01 -7.61613011e-01 7.23907650e-01 -2.45431542e-01
-2.27881327e-01 8.87106538e-01 -8.87285024e-02 -3.90606821e-01
-6.51503086e-01 -2.86925286e-01 -2.24976808e-01 -1.10467768e+00
-4.69583452e-01 6.15032256e-01 -1.06533259e-01 -8.86091813e-02
-8.07347149e-02 8.29764485e-01 -1.35454142e+00 -5.50964594e-01
7.26546824e-01 9.14240301e-01 -9.98947993e-02 8.40774953e-01
3.46448928e-01 6.44076347e-01 -4.94350225e-01 1.90771863e-01
3.15866977e-01 -4.58094001e-01 -1.32965088e-01 -4.98895980e-02
-1.18280120e-01 -5.51726855e-02 -5.91443479e-01 1.07717407e+00
4.84194547e-01 4.06454057e-01 9.00955260e-01 -1.08557773e+00
-1.56273142e-01 5.19002557e-01 -1.99314922e-01 4.49301690e-01
-2.40807589e-02 2.75823742e-01 5.10567307e-01 -7.78800189e-01
4.39268738e-01 1.12026346e+00 8.81892562e-01 2.33477995e-01
-1.73248744e+00 -4.73828018e-01 -1.77639276e-01 -2.76054237e-02
-1.69837153e+00 -6.76233947e-01 1.05584896e+00 -8.20944846e-01
6.86913490e-01 4.19550747e-01 6.23812973e-01 1.12677908e+00
2.99812317e-01 1.28654554e-01 1.14201224e+00 -3.00059050e-01
3.72659147e-01 -1.43651351e-01 1.47175655e-01 3.22936922e-01
1.62138864e-01 6.54927015e-01 -8.89336632e-04 -7.25767732e-01
8.12105715e-01 -2.61258632e-01 -1.93509698e-01 -2.58122027e-01
-9.92846727e-01 8.42536807e-01 -2.23357633e-01 4.21162635e-01
-5.90666056e-01 2.39187002e-01 3.05204421e-01 3.91960591e-01
6.69578075e-01 5.55854201e-01 -3.23654234e-01 -2.11464763e-01
-1.19284344e+00 3.94360065e-01 1.07663405e+00 4.50025320e-01
5.51369905e-01 8.51220787e-01 2.54458576e-01 5.03864348e-01
1.24450602e-01 8.96920085e-01 3.99216205e-01 -1.27176034e+00
-1.06254064e-01 4.66803387e-02 2.75411069e-01 -7.85077989e-01
-5.98985434e-01 -7.00741708e-01 -1.27367830e+00 2.60805726e-01
6.39204383e-01 -5.92349231e-01 -6.41675651e-01 1.71610427e+00
3.52009624e-01 5.27499378e-01 1.34485826e-01 7.22946763e-01
1.87497795e-01 8.43360364e-01 -3.71164799e-01 -8.32954943e-01
9.49728131e-01 -2.25781813e-01 -7.93550551e-01 2.07164586e-01
2.01330215e-01 -8.06264281e-01 4.13895518e-01 4.49303329e-01
-9.93179321e-01 -4.14862514e-01 -7.12292194e-01 5.28486192e-01
2.53799520e-02 6.56030923e-02 1.77548617e-01 4.46792364e-01
-9.57311749e-01 9.40068960e-01 -1.12618935e+00 1.58660337e-02
-2.93967962e-01 2.42771953e-01 -2.66617835e-01 5.08870482e-01
-1.07983148e+00 9.13429916e-01 1.18548982e-01 2.48240992e-01
-1.03853190e+00 -8.33236933e-01 -5.92477977e-01 1.38415322e-01
3.18609983e-01 -6.60754979e-01 1.02086723e+00 -7.70869195e-01
-1.68419123e+00 2.82544225e-01 -2.14719355e-01 -7.39362121e-01
4.94250238e-01 1.08652420e-01 -5.86243331e-01 2.35270262e-01
-1.31106421e-01 -2.52604842e-01 1.26654184e+00 -1.07557416e+00
1.52753843e-02 -5.20275789e-04 -5.80114663e-01 -1.72563359e-01
1.57623038e-01 3.05062514e-02 8.00497904e-02 -8.90140355e-01
2.42006883e-01 -1.10276949e+00 -5.33706725e-01 -1.91979542e-01
-3.04644048e-01 2.16088235e-01 7.61029005e-01 -8.59045684e-01
1.36703372e+00 -2.26016402e+00 5.68346262e-01 6.71997070e-01
2.93063343e-01 1.41334638e-01 7.11011142e-02 7.82790244e-01
-3.14274311e-01 1.03117861e-02 -4.60387200e-01 -4.23050135e-01
-5.70133403e-02 2.41253003e-01 -5.66440046e-01 8.02793443e-01
2.40742728e-01 3.57179761e-01 -6.27418637e-01 9.41063017e-02
4.29134727e-01 5.52860200e-01 -2.50120908e-01 1.77400485e-01
2.02659845e-01 8.35285187e-01 -2.53630936e-01 3.78560364e-01
3.63187313e-01 -8.58949497e-02 -5.36260754e-03 -1.96227953e-01
-4.00738984e-01 -6.31507263e-02 -1.56716442e+00 7.66338944e-01
-3.33718330e-01 7.47885108e-01 5.40608287e-01 -1.26566124e+00
1.01887023e+00 6.60155356e-01 8.78305316e-01 -1.36569723e-01
3.10574740e-01 5.24833500e-01 2.41353601e-01 -5.21589339e-01
5.37463278e-02 -3.34249377e-01 -9.55838338e-02 4.28167731e-01
3.06046396e-01 -3.37423354e-01 2.69393772e-01 -8.22959021e-02
1.01122439e+00 -1.76698729e-01 5.84142387e-01 -6.17889225e-01
5.93603075e-01 -1.40299842e-01 4.43482488e-01 8.10453594e-01
-1.17355511e-01 3.61745924e-01 4.33282733e-01 -2.67113835e-01
-1.38741314e+00 -9.12736773e-01 -3.56810480e-01 2.52615392e-01
-2.99890369e-01 -1.70786902e-01 -6.34268641e-01 3.36146593e-01
8.97705033e-02 7.21022725e-01 -5.34309208e-01 -3.08113813e-01
-7.32388079e-01 -8.72715533e-01 6.49684727e-01 -3.37992795e-02
-1.90934446e-02 -7.05881953e-01 -3.11930448e-01 5.98853171e-01
2.51392107e-02 -1.12788415e+00 -3.25774550e-01 4.55604434e-01
-7.75360107e-01 -8.41059148e-01 -4.87715781e-01 -1.34865746e-01
2.94699609e-01 -2.63654053e-01 7.61829317e-01 -3.51747513e-01
-4.12035823e-01 6.33111715e-01 1.92052834e-02 -6.47294149e-02
-9.61599946e-01 -3.13565344e-01 5.69870830e-01 2.24041566e-01
-2.92890936e-01 -1.05587375e+00 -1.79913372e-01 2.44143352e-01
-6.80160463e-01 -3.99952143e-01 8.91326144e-02 9.04706538e-01
4.91254449e-01 5.08233666e-01 4.25235450e-01 -3.26768726e-01
6.74985766e-01 -7.51870394e-01 -1.08901989e+00 -1.01763241e-01
-6.68440700e-01 8.20115283e-02 1.03675938e+00 -9.57956970e-01
-5.84628761e-01 2.11360276e-01 -9.25860032e-02 -7.98115194e-01
-1.34290576e-01 6.95195079e-01 3.71387094e-01 -4.68875259e-01
5.78350782e-01 6.01874292e-01 3.08506668e-01 -6.37588501e-01
2.25897044e-01 2.80143887e-01 7.60629594e-01 -5.95453441e-01
1.17091465e+00 4.83022690e-01 3.11090171e-01 -1.34244275e+00
-2.00680658e-01 -4.60190713e-01 -5.69040060e-01 -2.21199304e-01
4.15892333e-01 -7.81910181e-01 -7.11148560e-01 7.52532542e-01
-9.75827456e-01 -2.35110059e-01 -5.62598765e-01 7.04963684e-01
-5.96280694e-01 4.25016433e-01 -5.67121565e-01 -1.35320246e+00
-6.04948252e-02 -9.46044087e-01 8.29895496e-01 -5.56537993e-02
-3.47004294e-01 -1.25658715e+00 4.46759820e-01 -4.54602659e-01
6.69038951e-01 4.87657040e-01 7.74277389e-01 -5.93483746e-01
-1.42503127e-01 -4.10407037e-01 3.40869278e-01 3.34829301e-01
-6.44172877e-02 5.43672979e-01 -7.23179460e-01 -4.12458777e-01
7.22391188e-01 2.86835492e-01 6.30418479e-01 8.45918179e-01
5.55511594e-01 -5.11357605e-01 -1.51377052e-01 9.92539942e-01
1.33261704e+00 4.81459528e-01 1.90092310e-01 -8.62852335e-02
5.25420487e-01 5.44032335e-01 -9.76303816e-02 6.43351674e-01
1.93278473e-02 5.99829316e-01 2.58961737e-01 3.99642475e-02
2.78127819e-01 2.07393780e-01 4.56170648e-01 9.78415370e-01
-1.22769393e-01 2.47095409e-03 -8.06033432e-01 5.15115798e-01
-1.72319746e+00 -1.17528844e+00 -1.37233272e-01 2.08800125e+00
6.69438243e-01 -9.44866985e-02 3.26654673e-01 3.07478756e-01
6.17794275e-01 4.87672128e-02 -6.88987195e-01 -9.56707299e-02
-3.67228001e-01 9.14103761e-02 5.14499068e-01 7.73322403e-01
-1.15571392e+00 4.38585371e-01 7.02600288e+00 7.60142565e-01
-1.30904305e+00 -3.71942557e-02 2.81733036e-01 8.10305551e-02
-9.43489373e-03 3.91142210e-03 -6.92968428e-01 5.72237670e-01
1.31972587e+00 -7.44723260e-01 8.47394288e-01 5.02627790e-01
7.49274015e-01 1.74298450e-01 -6.79933846e-01 1.06840312e+00
-3.02847743e-01 -1.03244054e+00 -3.64793241e-01 2.58840155e-02
6.83909774e-01 5.02991863e-02 -6.45970255e-02 -2.05808878e-03
2.14772627e-01 -9.37470019e-01 8.90000403e-01 9.97617424e-01
5.21593451e-01 -3.97195041e-01 4.78577852e-01 6.35435462e-01
-1.15775490e+00 -2.93651909e-01 -1.84050158e-01 -1.96039915e-01
5.17112792e-01 9.99480724e-01 -1.99061498e-01 4.76805419e-01
4.19117570e-01 7.56370306e-01 -1.15491763e-01 1.13499808e+00
1.12295568e-01 1.11569214e+00 -9.27617729e-01 1.40919387e-01
1.77973345e-01 -6.77883744e-01 1.26187277e+00 9.84857857e-01
7.22801208e-01 3.78715903e-01 2.05609500e-01 9.49786484e-01
5.81108689e-01 -3.46543521e-01 -5.92608035e-01 -3.13047320e-01
5.85551381e-01 1.03915155e+00 -5.60564578e-01 -1.62932798e-01
-3.87569657e-03 4.18896884e-01 -2.25214839e-01 8.64629686e-01
-6.08263135e-01 -1.53417941e-02 6.60619199e-01 -2.43793055e-02
4.57307190e-01 -6.74728215e-01 -1.56393245e-01 -1.23291373e+00
-1.80011034e-01 -9.79077637e-01 1.93575099e-01 -4.42544997e-01
-1.38597322e+00 7.60924459e-01 2.02200487e-01 -1.35787499e+00
-7.47781157e-01 -5.35364985e-01 -6.55665815e-01 1.11520493e+00
-1.03914654e+00 -5.86679518e-01 1.00281216e-01 5.52532136e-01
8.89369771e-02 -2.67018527e-01 8.69801521e-01 1.86453864e-01
-7.19061971e-01 -5.63020073e-02 8.32690358e-01 -2.78587013e-01
2.43397698e-01 -1.15886986e+00 3.35509151e-01 9.96432483e-01
-2.21708566e-02 6.35136366e-01 1.29811525e+00 -4.50950563e-01
-1.37824726e+00 -8.73988509e-01 5.28162479e-01 -1.15386419e-01
1.31087613e+00 -3.41027588e-01 -1.13056505e+00 4.15671796e-01
-2.17447057e-01 5.85673712e-02 2.37848535e-01 -3.37933123e-01
9.76764830e-04 -9.58323851e-02 -9.21488225e-01 5.06954253e-01
4.52910393e-01 -5.74254334e-01 -4.60051060e-01 2.48241112e-01
2.49040782e-01 -3.40566337e-01 -9.25408959e-01 2.29849771e-01
4.47769552e-01 -7.38554358e-01 1.05473530e+00 -4.64904904e-01
-9.08243209e-02 -4.85619187e-01 -1.29656017e-01 -1.45293033e+00
-5.88310421e-01 -1.45570111e+00 -5.29575467e-01 9.58207369e-01
2.20163211e-01 -8.82753909e-01 1.20402262e-01 3.04423511e-01
8.06410797e-03 -4.20612037e-01 -1.38455546e+00 -1.09419811e+00
2.05224052e-01 -4.37827170e-01 2.44785473e-01 9.16972637e-01
-3.13391626e-01 1.09051950e-01 -6.77608788e-01 4.58120883e-01
9.19232309e-01 -1.13436446e-01 6.40868723e-01 -1.38095093e+00
-5.07969439e-01 -6.85335219e-01 -2.64986724e-01 -6.87133670e-01
1.91640750e-01 -5.71661413e-01 2.77391123e-03 -7.13467598e-01
-4.39841032e-01 -3.49122703e-01 -1.07200347e-01 -3.63343284e-02
-1.46661624e-01 -2.75795218e-02 5.63126020e-02 5.06547332e-01
1.75600320e-01 5.62074900e-01 6.70603633e-01 1.95298314e-01
-1.49970263e-01 2.96457976e-01 -2.08030447e-01 7.20070660e-01
5.00617027e-01 -6.17863417e-01 -2.57050276e-01 1.43500313e-01
1.26131310e-03 5.68797112e-01 7.60664880e-01 -1.02137983e+00
2.92725086e-01 -7.48779923e-02 2.66311541e-02 -3.57062936e-01
3.87097239e-01 -9.79491949e-01 8.42076302e-01 6.21954203e-01
-5.73211014e-02 2.13398620e-01 4.42766070e-01 6.06404543e-01
-2.88457543e-01 -2.44784430e-01 9.79263723e-01 1.37644395e-01
-3.79857570e-01 8.91608223e-02 -8.37893724e-01 8.17799717e-02
6.82994366e-01 -7.87528418e-03 -2.23752812e-01 -7.68927872e-01
-1.18097222e+00 5.92606179e-02 2.11776540e-01 -2.53130078e-01
1.45863861e-01 -1.13223815e+00 -9.87428725e-01 4.11224157e-01
-4.84718531e-01 -5.17706573e-01 2.63692170e-01 1.16488755e+00
-2.68878639e-01 1.91154018e-01 7.34247789e-02 -6.91381633e-01
-8.97429347e-01 4.46596801e-01 7.83556402e-01 -3.00332546e-01
-6.00983560e-01 4.20547724e-01 -1.51540145e-01 -3.45724910e-01
-1.43547356e-01 -3.35401624e-01 -6.28855005e-02 -3.95440264e-03
3.35123867e-01 6.68323457e-01 -2.39583045e-01 -8.50061297e-01
-2.71619439e-01 8.29507351e-01 5.73312938e-01 -2.65794098e-01
1.32820630e+00 -2.34699607e-01 -3.81358117e-01 8.36584330e-01
1.17345822e+00 -1.50324870e-02 -1.47322762e+00 -2.36619428e-01
1.33771691e-02 1.00644901e-01 5.17924502e-02 -5.10336876e-01
-7.37591743e-01 6.93134785e-01 4.23810929e-01 8.60706389e-01
1.12411392e+00 -4.46817309e-01 4.06564742e-01 2.93948680e-01
1.07672326e-01 -7.35253155e-01 -4.40535516e-01 6.49701118e-01
9.70899165e-01 -7.19169617e-01 -4.31868583e-02 -2.17413634e-01
-2.37126961e-01 1.29266858e+00 -2.45705411e-01 -6.48874879e-01
1.03809261e+00 7.12002635e-01 -6.14233222e-03 -2.73536216e-03
-8.83119881e-01 -8.67639557e-02 5.63306093e-01 5.06105661e-01
2.32728068e-02 4.73073162e-02 -2.31520608e-01 6.44057155e-01
-1.94295183e-01 -3.30232143e-01 6.21185184e-01 3.68735999e-01
-2.63049513e-01 -5.95310390e-01 -6.89728796e-01 3.06586891e-01
-4.19720083e-01 -9.66312960e-02 -1.09478839e-01 8.38401020e-01
-3.77398610e-01 9.01989043e-01 -1.19368628e-01 -5.28055355e-02
3.91528368e-01 2.79022932e-01 2.89803557e-02 -2.99925297e-01
-4.93537754e-01 7.00881720e-01 1.23108797e-01 -3.42457652e-01
-2.48169437e-01 -9.28933203e-01 -7.23832071e-01 -3.64104986e-01
-4.20936108e-01 3.45223129e-01 4.98484671e-01 1.14760292e+00
1.46454021e-01 5.33470929e-01 8.25619578e-01 -1.11481655e+00
-1.11726952e+00 -8.88409972e-01 -8.68757963e-01 8.11617449e-02
7.72462130e-01 -6.78443968e-01 -1.13512766e+00 3.52092274e-02]
|
[6.551994323730469, 3.517589569091797]
|
a93cb22e-db6f-4b2e-ab0c-916be30d1d8e
|
zero-shot-bird-species-recognition-by
|
2206.01466
| null |
https://arxiv.org/abs/2206.01466v1
|
https://arxiv.org/pdf/2206.01466v1.pdf
|
Zero-Shot Bird Species Recognition by Learning from Field Guides
|
We exploit field guides to learn bird species recognition, in particular zero-shot recognition of unseen species. The illustrations contained in field guides deliberately focus on discriminative properties of a species, and can serve as side information to transfer knowledge from seen to unseen classes. We study two approaches: (1) a contrastive encoding of illustrations that can be fed into zero-shot learning schemes; and (2) a novel method that leverages the fact that illustrations are also images and as such structurally more similar to photographs than other kinds of side information. Our results show that illustrations from field guides, which are readily available for a wide range of species, are indeed a competitive source of side information. On the iNaturalist2021 subset, we obtain a harmonic mean from 749 seen and 739 unseen classes greater than $45\%$ (@top-10) and $15\%$ (@top-1). Which shows that field guides are a valuable option for challenging real-world scenarios with many species.
|
['Konrad Schindler', 'Jan D. Wegner', 'Rodrigo Caye Daudt', "Stefano D'Aronco", 'Andrés C. Rodríguez']
|
2022-06-03
| null | null | null | null |
['generalized-zero-shot-learning', 'generalized-zero-shot-learning']
|
['computer-vision', 'methodology']
|
[ 2.52631366e-01 -2.32261315e-01 6.60836324e-02 -4.08373356e-01
-5.85045755e-01 -1.11844933e+00 7.23328531e-01 7.61964992e-02
-6.62238479e-01 7.28008091e-01 3.48610282e-01 8.73577446e-02
-1.26489744e-01 -8.38508964e-01 -9.27625000e-01 -6.14730418e-01
-4.95406061e-01 -1.82871241e-02 2.37239152e-01 -1.48167819e-01
2.25221410e-01 4.70507383e-01 -2.14250636e+00 2.05461234e-01
4.72746849e-01 9.19697940e-01 1.87828749e-01 8.42708647e-01
9.23642702e-03 7.97254145e-01 -8.52872074e-01 -6.14093184e-01
6.19723380e-01 -3.50790590e-01 -3.24784517e-01 2.62798648e-02
1.36961496e+00 -5.93549132e-01 -5.17859101e-01 1.04793799e+00
4.91038412e-01 4.91852164e-01 1.16783524e+00 -1.22135067e+00
-8.52196991e-01 4.17469800e-01 -6.34249508e-01 6.33340597e-01
2.58510470e-01 7.31254339e-01 1.15878963e+00 -7.75807977e-01
7.14313924e-01 9.95783031e-01 6.34338975e-01 5.77681720e-01
-1.20938993e+00 -1.02337575e+00 2.79045016e-01 2.47621924e-01
-1.37997389e+00 -5.73256552e-01 4.89145398e-01 -7.09850490e-01
1.06302416e+00 3.62303466e-01 8.24548781e-01 1.19652569e+00
-2.68789917e-01 6.42056882e-01 9.77333784e-01 -8.28359127e-02
3.09962779e-01 2.96783477e-01 -1.21517805e-03 7.19498754e-01
2.23204494e-01 6.58876181e-01 -6.86285555e-01 6.69795424e-02
6.23533249e-01 3.88817221e-01 -5.15310407e-01 -4.42547023e-01
-1.10842311e+00 7.76880741e-01 1.11397958e+00 -2.00406283e-01
-5.67441620e-02 7.34324530e-02 3.70820053e-02 1.90876856e-01
1.84912652e-01 8.95757675e-01 3.18778935e-03 -1.55007422e-01
-1.01179183e+00 6.66833669e-02 8.14610481e-01 1.17209196e+00
1.05785632e+00 3.54706824e-01 3.79023212e-03 9.77590740e-01
-1.08708717e-01 1.02088869e+00 2.24456370e-01 -7.51939058e-01
1.61370620e-01 4.56214130e-01 6.22465163e-02 -1.10452425e+00
-4.86730263e-02 -5.72290301e-01 -5.63707829e-01 2.99217165e-01
1.83939472e-01 -1.92205295e-01 -1.24893999e+00 1.94002223e+00
9.37897637e-02 2.46938869e-01 -5.64907268e-02 9.63005960e-01
1.56320071e+00 7.94280052e-01 7.47122392e-02 2.75411129e-01
1.05127931e+00 -1.01063716e+00 6.91871569e-02 -4.91358697e-01
3.07786018e-01 -5.09820223e-01 9.66262698e-01 5.87845668e-02
-7.14784145e-01 -4.83201295e-01 -1.17077875e+00 1.35157481e-01
-8.44824314e-01 -1.92447796e-01 5.83819509e-01 3.88539821e-01
-1.13546109e+00 5.30948043e-01 -2.58894682e-01 -3.25261950e-01
6.16749525e-01 9.64926183e-02 -5.70293844e-01 -2.16130614e-01
-7.80468464e-01 7.38799334e-01 8.09466019e-02 -3.21717471e-01
-1.55183291e+00 -1.04054737e+00 -1.16487527e+00 2.06058398e-01
4.07840937e-01 -2.11968288e-01 9.94691551e-01 -9.15772557e-01
-7.02783763e-01 1.18785548e+00 4.25944924e-01 -2.94867158e-01
3.69140774e-01 7.19884783e-02 -5.17007470e-01 4.45712000e-01
1.37477562e-01 1.36777246e+00 1.17751706e+00 -1.29763460e+00
-9.10051167e-01 -1.30090639e-01 4.20694888e-01 2.74457902e-01
-5.26262641e-01 -5.26884675e-01 -5.34069026e-03 -8.10226023e-01
-4.82905865e-01 -7.43414640e-01 1.04043655e-01 6.77342534e-01
-2.01875284e-01 1.75552011e-01 5.94615161e-01 -3.44030321e-01
8.50984335e-01 -2.30377698e+00 1.27707034e-01 -1.73340917e-01
4.94458765e-01 3.37409586e-01 -4.79269415e-01 6.83648229e-01
2.70997752e-02 5.49684465e-02 -3.71138304e-01 5.57146408e-02
8.70506316e-02 3.92831832e-01 -4.64041889e-01 5.41273117e-01
2.28063062e-01 8.04735363e-01 -1.08360291e+00 -3.66480678e-01
5.20785689e-01 3.85789841e-01 -4.46183145e-01 3.76133084e-01
1.23784142e-02 -1.63524285e-01 -9.77665465e-03 7.68119454e-01
6.50063813e-01 -6.91741407e-02 -3.88511598e-01 7.02397376e-02
-1.48628265e-01 -2.23905876e-01 -9.76061583e-01 1.52485609e+00
-3.19660217e-01 8.38115871e-01 1.60380706e-01 -5.53426266e-01
7.15144992e-01 -2.98454404e-01 -1.15849271e-01 -5.29143333e-01
9.67915822e-03 -1.12547666e-01 -1.03456959e-01 -1.82369202e-01
5.67927897e-01 -3.44324350e-01 -1.30319774e-01 4.54577774e-01
5.57290792e-01 -4.68655646e-01 3.58903021e-01 4.44387466e-01
1.08216524e+00 -2.71809936e-01 2.87227094e-01 -4.95154470e-01
-8.18052739e-02 1.26353130e-01 3.61671835e-01 1.05376160e+00
-4.31229085e-01 5.65884233e-01 8.56964961e-02 -5.99248409e-01
-9.50373411e-01 -1.52971554e+00 -2.61038601e-01 1.14710259e+00
3.18733513e-01 -2.45587274e-01 -5.32475054e-01 -6.36590123e-01
4.04467672e-01 6.67800903e-01 -1.00029993e+00 -4.50785935e-01
1.42681956e-01 -1.80761024e-01 4.84596580e-01 8.70634377e-01
3.59302849e-01 -1.16458344e+00 -1.07444763e+00 -4.11704004e-01
3.48759979e-01 -7.14154005e-01 -8.04023087e-01 4.82122004e-01
-2.93840498e-01 -1.18448913e+00 -1.08165026e+00 -9.22516286e-01
8.09258759e-01 9.31439817e-01 1.26621330e+00 3.46462876e-01
-8.56677234e-01 8.25612307e-01 -5.34960866e-01 -2.15290308e-01
-2.87759602e-02 -2.03743324e-01 2.04237178e-01 -7.87448213e-02
3.86445969e-01 -7.75736928e-01 -7.29512334e-01 2.87118822e-01
-7.53951490e-01 -1.36342317e-01 7.86567330e-01 8.72872174e-01
2.96792060e-01 -2.12662399e-01 3.01940471e-01 -7.86133826e-01
1.07379280e-01 -5.29153526e-01 -6.91608131e-01 3.47321421e-01
-2.85731331e-02 -6.64097897e-05 8.71865392e-01 -6.02142870e-01
-6.45789564e-01 5.92314713e-02 1.28811017e-01 -7.75245428e-01
-4.99345183e-01 3.21497060e-02 1.79924995e-01 -3.11575532e-01
1.03601217e+00 3.72044653e-01 -2.29836643e-01 -4.90348369e-01
5.92703402e-01 6.59835398e-01 9.28038538e-01 -3.27333033e-01
1.17322087e+00 4.53784108e-01 -2.30183423e-01 -1.29091024e+00
-7.18635619e-01 -6.63283408e-01 -6.16752684e-01 -3.22664231e-01
7.28820026e-01 -1.07621682e+00 -5.41837811e-01 2.35835597e-01
-4.44204211e-01 -4.49362338e-01 -6.16683900e-01 3.74983102e-01
-3.82171631e-01 2.13228524e-01 -3.21862549e-01 -8.07349145e-01
-2.81284124e-01 -8.17155600e-01 1.14089680e+00 6.99997962e-01
1.34602021e-02 -8.21689427e-01 -1.29749686e-01 -7.55772069e-02
2.10135475e-01 3.25626910e-01 6.92966700e-01 -6.51853025e-01
-6.33712113e-01 -2.47428536e-01 -5.96743524e-01 2.23654017e-01
6.87509477e-02 1.24491394e-01 -1.20575023e+00 -6.38720930e-01
-3.91190290e-01 -6.69079363e-01 1.12519395e+00 1.14366829e-01
8.11930776e-01 -4.29006368e-01 -2.95390040e-01 1.02965426e+00
1.60323775e+00 2.04696357e-01 2.93300360e-01 -2.10503340e-01
5.19376278e-01 4.98655885e-01 2.35954881e-01 5.82684159e-01
2.23729342e-01 3.07704777e-01 5.53971291e-01 -1.90789461e-01
-3.32402706e-01 -7.94495165e-01 1.75496906e-01 4.67839301e-01
2.95885175e-01 -2.50778377e-01 -8.71638834e-01 9.77470040e-01
-1.23866808e+00 -1.24314809e+00 3.78447115e-01 2.27416658e+00
5.59185624e-01 -1.46299396e-02 2.75784016e-01 -2.91051775e-01
7.22621143e-01 4.38763887e-01 -8.72210681e-01 -2.00394034e-01
-2.51916498e-01 8.04727003e-02 6.17743254e-01 5.66703938e-02
-1.18164504e+00 8.08170855e-01 7.35358429e+00 8.58645618e-01
-9.66519475e-01 -3.60174119e-01 2.02011138e-01 -4.34613317e-01
-2.98902631e-01 -2.62138665e-01 -6.94869101e-01 6.46712482e-01
5.59800208e-01 -4.30502325e-01 6.96676612e-01 1.12541795e+00
-6.09265149e-01 -2.62128413e-01 -1.18080783e+00 1.34933126e+00
5.07989168e-01 -1.35089350e+00 2.40313247e-01 9.16776359e-02
8.79313290e-01 1.38432220e-01 3.66213709e-01 4.43107456e-01
7.46028483e-01 -1.16944110e+00 7.04187036e-01 2.04939470e-01
1.20196283e+00 -7.25500822e-01 4.19474065e-01 3.49832296e-01
-1.31468046e+00 -2.45864540e-01 -7.50547886e-01 -1.85333550e-01
-2.14347705e-01 -2.00022291e-02 -5.66614866e-01 5.01577295e-02
1.07779050e+00 1.21606493e+00 -1.00728905e+00 1.29970145e+00
-3.48350167e-01 3.88190925e-01 -4.55357641e-01 -3.43873382e-01
5.07650852e-01 1.54501721e-01 6.58841670e-01 1.29203665e+00
1.46943659e-01 1.99091762e-01 1.94614843e-01 9.08025682e-01
-3.56424242e-01 -1.57001391e-01 -1.10540962e+00 -3.84819478e-01
4.06606972e-01 1.34952104e+00 -6.61472201e-01 -4.01868284e-01
-4.61235255e-01 8.01544189e-01 3.79894435e-01 2.36637771e-01
-6.87989712e-01 -5.26002407e-01 8.85918915e-01 -1.04410842e-01
7.85371065e-01 1.36687774e-02 3.80736798e-01 -1.34487176e+00
-2.20055431e-01 -6.69835269e-01 6.70125008e-01 -1.02711165e+00
-1.74647450e+00 7.85443246e-01 1.09345198e-01 -1.51555526e+00
-1.19564310e-01 -7.34439254e-01 -5.38494885e-01 5.42594135e-01
-1.45172060e+00 -1.12895548e+00 -6.64445162e-01 5.55742443e-01
6.41328514e-01 -2.46285647e-01 8.86983514e-01 7.21233338e-02
-1.67405665e-01 7.55497694e-01 3.64805460e-01 1.90690577e-01
5.51376939e-01 -1.39038873e+00 3.46573919e-01 9.34907198e-01
7.55156457e-01 3.32695812e-01 4.70983893e-01 -3.90180051e-01
-1.37199295e+00 -9.95453894e-01 -3.46646528e-03 -5.85249305e-01
6.55457854e-01 -4.44766611e-01 -6.63710356e-01 5.35266101e-01
2.28842050e-02 1.13212079e-01 9.02656972e-01 8.51930156e-02
-7.69492447e-01 -1.10575579e-01 -1.17454040e+00 5.54701805e-01
1.36451590e+00 -7.74791181e-01 -7.87350535e-01 5.86468801e-02
4.41902548e-01 -7.30021447e-02 -7.04840243e-01 7.06690736e-03
6.54527187e-01 -8.18652391e-01 1.09585762e+00 -7.32324302e-01
7.19534039e-01 -2.57113129e-01 -4.97800201e-01 -1.66945040e+00
-4.13827956e-01 -2.72237539e-01 1.74337402e-01 1.14120162e+00
3.12144846e-01 -3.48162413e-01 6.02145076e-01 5.58213532e-01
-1.90166652e-01 -2.63233155e-01 -7.68689394e-01 -1.12833571e+00
1.80106405e-02 5.67116663e-02 3.05968940e-01 7.50240386e-01
-2.17692405e-01 3.02262485e-01 -5.72348356e-01 -4.39419970e-02
8.75511885e-01 4.22768593e-01 9.16813314e-01 -1.10530365e+00
-3.20707887e-01 -4.81571734e-01 -8.41517985e-01 -1.12705028e+00
-3.13343480e-02 -9.47262883e-01 2.29182214e-01 -1.28137529e+00
4.56207335e-01 -1.58621192e-01 -3.35569739e-01 5.41926086e-01
-1.03941008e-01 6.32687032e-01 4.16481733e-01 1.77905068e-01
-5.93295455e-01 4.73091483e-01 1.04012001e+00 -4.24264312e-01
2.92876959e-01 -3.91522110e-01 -8.34841132e-01 5.33128917e-01
1.52410060e-01 -4.16196942e-01 -5.44172645e-01 -5.87592602e-01
-2.55749732e-01 -2.50768781e-01 6.12502277e-01 -1.21412778e+00
1.82018638e-01 -1.90486982e-01 5.11204898e-01 -5.46124578e-01
6.91093385e-01 -8.24308574e-01 -5.42901531e-02 4.09986526e-01
-3.31447482e-01 -7.52492100e-02 3.69414777e-01 8.98007333e-01
2.72866953e-02 -2.77772486e-01 9.74978864e-01 -4.93896931e-01
-1.43591893e+00 4.12794799e-01 -1.64150849e-01 5.07251322e-01
9.96257484e-01 -4.62695360e-01 -7.79284060e-01 -6.19479656e-01
-2.38920853e-01 3.28869253e-01 8.99368703e-01 5.17564774e-01
6.13025725e-01 -9.84199762e-01 -5.58609724e-01 4.32591498e-01
5.57405174e-01 -1.87770531e-01 5.17922223e-01 2.96529144e-01
-5.89505613e-01 1.38151348e-01 -7.40892112e-01 -6.46286309e-01
-1.10233390e+00 1.09068131e+00 1.27948090e-01 6.76171124e-01
-8.91626000e-01 1.42805481e+00 7.22434461e-01 -1.15906723e-01
4.30276811e-01 1.32941633e-01 8.25019479e-02 2.52922922e-01
8.78491998e-01 1.51277304e-01 -4.18519825e-01 -9.12241995e-01
-4.71304387e-01 6.15154445e-01 -4.28049304e-02 5.03888577e-02
1.74018288e+00 7.48301074e-02 3.89385164e-01 4.23428714e-01
1.18330181e+00 -1.42954826e-01 -1.81086719e+00 -1.61037952e-01
-4.53641802e-01 -1.03512561e+00 -8.59303772e-02 -1.07115161e+00
-1.14241159e+00 1.23615253e+00 6.49363279e-01 2.66682804e-01
9.29986060e-01 4.11700755e-02 5.77851951e-01 4.25356388e-01
6.49000406e-01 -8.67585003e-01 1.62540138e-01 -8.70817751e-02
9.25242960e-01 -1.37643373e+00 3.62798646e-02 5.63918753e-03
-7.21022844e-01 7.81820118e-01 7.27789640e-01 -3.46354663e-01
5.06787956e-01 2.72941798e-01 6.04725257e-02 -1.87061384e-01
-9.35567379e-01 -6.20419323e-01 3.60519916e-01 1.05866957e+00
-2.27871791e-01 1.84162870e-01 4.48240072e-01 3.97332996e-01
-4.24279422e-01 -5.05635917e-01 5.05808711e-01 8.75593364e-01
-5.52567899e-01 -2.84587592e-01 -6.71731159e-02 7.65882790e-01
1.58223107e-01 -3.15003783e-01 -8.62128794e-01 7.48338997e-01
7.06841499e-02 8.99208605e-01 1.71585485e-01 -5.98388910e-01
1.97242737e-01 -1.96663484e-01 3.83096457e-01 -8.04150224e-01
-6.13704264e-01 -4.21902984e-01 -1.09034991e-02 -1.29626915e-01
-1.43244326e-01 -3.43155205e-01 -5.10707438e-01 -5.11532962e-01
-2.67432541e-01 6.31975979e-02 3.20836484e-01 4.31598186e-01
1.53318897e-01 2.08220541e-01 6.30294204e-01 -1.03763127e+00
-3.36242169e-01 -6.46119952e-01 -8.59439075e-01 7.55984128e-01
4.16394919e-01 -1.12924838e+00 -9.78759646e-01 -1.49749322e-02]
|
[9.837918281555176, 2.2608370780944824]
|
c7acc28b-39f4-429f-b547-f810182aa6c0
|
texture-features-in-medical-image-analysis-a
|
2208.02046
| null |
https://arxiv.org/abs/2208.02046v1
|
https://arxiv.org/pdf/2208.02046v1.pdf
|
Texture features in medical image analysis: a survey
|
The texture is defined as spatial structure of the intensities of the pixels in an image that is repeated periodically in the whole image or regions, and makes the concept of the image. Texture, color and shape are three main components which are used by human visual system to recognize image contents. In this paper, first of all, efficient and updated texture analysis operators are survived with details. Next, some state-of-the-art methods are survived that use texture analysis in medical applications and disease diagnosis. Finally, different approaches are compared in terms of accuracy, dataset, application, etc. Results demonstrate that texture features separately or in joint of different feature sets such as deep, color or shape features provide high accuracy in medical image classification.
|
['Faeze Kiani']
|
2022-08-02
| null | null | null | null |
['texture-classification']
|
['computer-vision']
|
[ 1.75584868e-01 -5.72341979e-01 -9.52495113e-02 -2.70701468e-01
-1.72569692e-01 -8.77203569e-02 3.29150707e-01 1.22813970e-01
-2.26381212e-01 4.82527912e-01 -1.37997329e-01 6.71731532e-02
-3.24360609e-01 -9.99605894e-01 3.73711996e-02 -1.15628111e+00
-3.95655558e-02 1.55717418e-01 4.29905027e-01 -2.05009207e-01
4.93878126e-01 6.45439088e-01 -1.71031940e+00 7.58348405e-01
4.62456793e-01 1.48100913e+00 3.42070051e-02 6.59812748e-01
-6.31385267e-01 8.89969587e-01 -5.05311430e-01 -1.42547086e-01
-2.72919834e-01 -4.86596346e-01 -8.83762479e-01 4.39147115e-01
-1.37085795e-01 -3.67497839e-02 -1.99407578e-01 1.16225994e+00
4.58292216e-01 -1.05455540e-01 9.01704431e-01 -8.10781896e-01
-8.46904457e-01 -1.16925286e-02 -8.20049107e-01 4.78794605e-01
1.15611203e-01 -3.32960904e-01 1.43586084e-01 -8.26512814e-01
6.43650889e-01 1.18939292e+00 6.19594991e-01 2.30017632e-01
-7.19788432e-01 -2.09649801e-02 -2.73536414e-01 5.95703542e-01
-1.14280093e+00 -5.88230081e-02 8.91494930e-01 -3.26135486e-01
5.49699724e-01 7.59032905e-01 7.54711926e-01 5.02720177e-01
1.03777015e+00 8.82313848e-01 1.60929024e+00 -6.54031754e-01
-1.74565718e-01 1.36615783e-01 2.03574285e-01 1.05638087e+00
6.19717278e-02 -8.42506737e-02 -4.29371417e-01 6.80042952e-02
7.18092203e-01 3.03293914e-01 -9.24342275e-02 -3.05169430e-02
-1.11028910e+00 4.61416841e-01 1.97777927e-01 7.02044845e-01
-1.90805241e-01 -1.68630019e-01 5.62794805e-01 4.97746736e-01
6.35366619e-01 -1.53507695e-01 -2.72139668e-01 1.78366918e-02
-5.76598406e-01 -3.10366064e-01 3.66796494e-01 3.88928294e-01
6.39057398e-01 -2.10380301e-01 -3.31772268e-01 1.13990128e+00
1.08365357e-01 7.94934452e-01 7.66898513e-01 -5.02774596e-01
-2.87622690e-01 6.60503447e-01 -3.29137534e-01 -1.53975976e+00
-5.53918660e-01 -1.43991932e-01 -1.07990563e+00 4.98131275e-01
6.05540387e-02 4.84772563e-01 -1.20652747e+00 8.61083269e-01
5.44011652e-01 -1.06934771e-01 -9.24612135e-02 6.84766948e-01
1.35257411e+00 4.46136177e-01 -1.15644962e-01 -7.26681799e-02
1.91761684e+00 -9.47066724e-01 -1.05645239e+00 1.28852114e-01
2.25859761e-01 -1.23885179e+00 8.05616319e-01 8.23075294e-01
-9.33326960e-01 -6.56579614e-01 -8.36202264e-01 -1.09312944e-02
-7.43258178e-01 3.26318175e-01 6.99012280e-01 6.93191230e-01
-9.50136662e-01 7.16129482e-01 -6.97795451e-01 -5.03112912e-01
4.09927249e-01 1.86559394e-01 -5.62280953e-01 -1.33764431e-01
-6.12148345e-01 8.60745668e-01 1.57809362e-01 4.20522690e-01
-2.97573388e-01 -1.28351361e-01 -6.61159515e-01 -2.96800017e-01
-1.19914323e-01 -4.13968980e-01 7.22457111e-01 -1.30810905e+00
-1.40474510e+00 1.42774284e+00 -2.95803785e-01 2.32669592e-01
4.81947243e-01 1.91832110e-01 -7.46546328e-01 5.70189714e-01
-1.71446696e-01 1.66157991e-01 8.01378548e-01 -1.11083877e+00
-8.25223327e-01 -4.27069783e-01 -5.92692196e-01 9.78127345e-02
-2.02914685e-01 1.76938638e-01 -7.11038709e-01 -6.54932141e-01
5.36458969e-01 -6.46049976e-01 -7.44946953e-03 2.78607309e-01
-3.77988040e-01 -2.18829826e-01 1.26137912e+00 -8.69371116e-01
1.17346263e+00 -2.44888139e+00 -1.04958154e-01 6.67341053e-01
1.74591586e-01 -3.10951993e-02 -2.06508730e-02 2.02333719e-01
4.47750986e-02 2.13693187e-01 -6.30804151e-02 8.02392438e-02
-3.73118043e-01 3.04988563e-01 3.10720891e-01 7.40532100e-01
1.38225881e-02 7.90954769e-01 -4.13328826e-01 -1.09608126e+00
5.26372254e-01 5.29235780e-01 1.01638287e-01 -1.91940203e-01
4.34046954e-01 6.07309997e-01 -7.93493867e-01 1.06536019e+00
9.86861229e-01 -2.78933764e-01 2.85408609e-02 -4.77285415e-01
-3.00218109e-02 -4.50566709e-01 -9.65794146e-01 1.15046763e+00
-9.00248215e-02 5.68856776e-01 -1.02416888e-01 -1.29127288e+00
1.05436409e+00 3.26416522e-01 6.95970893e-01 -1.16484594e+00
3.41929406e-01 3.49376291e-01 -1.43096363e-02 -1.08786023e+00
2.14720592e-01 -1.39577631e-02 2.07041740e-01 2.84154296e-01
-2.80656427e-01 -7.34454617e-02 2.67128617e-01 -5.01740754e-01
6.21195495e-01 -7.88446963e-02 4.41490650e-01 -5.01820147e-01
9.09015238e-01 1.42895594e-01 2.96151519e-01 6.00223720e-01
-3.23350698e-01 6.73772156e-01 4.40676004e-01 -8.61458004e-01
-1.04007411e+00 -8.15032363e-01 -7.73123920e-01 7.78987765e-01
5.43300092e-01 3.21874231e-01 -6.81392372e-01 -4.25169021e-01
5.29325493e-02 -3.51701140e-01 -1.19818759e+00 7.72766098e-02
-4.51127440e-01 -1.08738768e+00 1.25362992e-01 2.45692968e-01
9.54333544e-01 -1.42179716e+00 -7.08444118e-01 3.92616875e-02
-2.32421234e-02 -7.94734299e-01 -1.92050099e-01 1.10476781e-02
-1.06184042e+00 -1.23058498e+00 -8.26913357e-01 -1.16661119e+00
9.23207104e-01 2.14140385e-01 9.16071892e-01 7.57400930e-01
-9.48786199e-01 3.28101844e-01 -6.25781476e-01 -2.43629143e-01
-1.50342882e-01 -4.31297630e-01 -4.85355616e-01 3.26399267e-01
2.38095239e-01 -1.67154431e-01 -8.40081632e-01 2.55209893e-01
-8.60104978e-01 4.77506556e-02 7.80698895e-01 1.05408919e+00
1.00928497e+00 4.92191851e-01 -1.44147221e-02 -1.00090599e+00
3.46411318e-01 -1.58177433e-03 -7.72704324e-03 7.30655253e-01
-2.70161420e-01 6.95862016e-03 1.68058753e-01 -2.36887038e-01
-1.09462452e+00 -2.16309831e-01 -6.29218891e-02 1.18797310e-02
-2.03771502e-01 7.05132842e-01 3.00997019e-01 -4.46727306e-01
4.74032342e-01 6.42569721e-01 4.17442262e-01 -6.19346976e-01
-2.55364954e-01 7.51860976e-01 4.25153315e-01 -3.86662275e-01
1.28110230e-01 8.42460990e-01 1.38365239e-01 -8.73492897e-01
-7.30857253e-01 -5.06625593e-01 -7.96037197e-01 -3.90309960e-01
9.47615147e-01 -3.19663793e-01 -5.80108225e-01 1.10373962e+00
-6.53940916e-01 -2.05849051e-01 1.72761738e-01 3.69207948e-01
-2.69895226e-01 5.76446533e-01 -8.03134561e-01 -7.15932190e-01
-5.99712789e-01 -1.10589314e+00 1.05885279e+00 3.93895209e-01
2.43834615e-01 -1.40959811e+00 -1.81919217e-01 1.61805943e-01
5.64721346e-01 5.50518751e-01 1.08450508e+00 6.79026619e-02
-3.74100469e-02 -1.59551561e-01 -4.26146477e-01 2.56065637e-01
6.92891955e-01 3.74820471e-01 -8.00644398e-01 -7.52387047e-02
1.20642096e-01 6.11697733e-02 9.49894488e-01 6.87860370e-01
1.65927482e+00 -1.35427728e-01 -5.79739213e-01 6.51175976e-01
1.64435828e+00 6.63955927e-01 1.08882558e+00 5.48385322e-01
2.43556201e-01 6.20534122e-01 6.64665699e-01 3.30366641e-01
-2.45775908e-01 3.84574801e-01 -6.43738508e-02 -7.24647999e-01
-3.00076753e-01 6.00503206e-01 -1.67144537e-01 8.88091028e-01
-3.55832458e-01 8.07429254e-02 -7.58505166e-01 1.85683683e-01
-1.54164422e+00 -9.96550143e-01 -3.64847571e-01 1.96845579e+00
7.83268094e-01 6.13760836e-02 -2.01599449e-01 4.46541309e-01
8.92346203e-01 -2.52855539e-01 -2.87603110e-01 -4.82394099e-01
-5.32247543e-01 4.69906718e-01 4.90130424e-01 1.65013954e-01
-1.43273079e+00 6.79730535e-01 7.39393091e+00 1.13673604e+00
-1.66564846e+00 3.13973315e-02 1.25525582e+00 4.38117206e-01
7.62173012e-02 -5.37079811e-01 6.43428862e-02 4.15267885e-01
5.48191592e-02 1.45326987e-01 2.66006202e-01 4.77456033e-01
1.58831626e-01 -7.21053541e-01 -4.15915668e-01 1.25707698e+00
2.41443604e-01 -1.13545454e+00 -3.73455808e-02 -6.71788529e-02
6.57796562e-01 -3.66537482e-01 4.96737629e-01 -3.47233117e-01
-1.96248040e-01 -1.17353702e+00 6.35634780e-01 9.30540442e-01
1.07583332e+00 -5.39400697e-01 1.16955793e+00 -3.38420272e-01
-1.26945508e+00 5.97970597e-02 -4.28056121e-01 2.31270254e-01
-2.04956338e-01 7.38538444e-01 -8.82292837e-02 6.93594515e-01
1.21196342e+00 8.41068327e-01 -8.18752706e-01 1.23428822e+00
1.25109449e-01 2.75676072e-01 -1.43220425e-02 -1.18813038e-01
2.49084547e-01 -2.09537163e-01 -5.38874082e-02 1.37560141e+00
4.47604120e-01 8.74438584e-02 1.33630678e-01 3.79478395e-01
6.70186579e-01 3.76203626e-01 -4.11543190e-01 2.22665593e-01
6.86614141e-02 1.45053577e+00 -1.33212781e+00 -6.37219787e-01
-2.92134196e-01 9.69383597e-01 -2.90757537e-01 3.84659648e-01
-5.52973449e-01 -4.40835088e-01 2.22740933e-01 -1.53950796e-01
3.51319350e-02 1.63188025e-01 -6.18576586e-01 -7.51342773e-01
-1.21097147e-01 -7.38056600e-01 4.53688383e-01 -8.11292768e-01
-1.41586971e+00 7.06278801e-01 -4.99985874e-01 -1.36306560e+00
4.94703889e-01 -1.17340696e+00 -5.68185091e-01 7.58203030e-01
-1.44113147e+00 -1.13983798e+00 -6.65697634e-01 8.67920280e-01
4.15673941e-01 -3.00043225e-01 7.51869082e-01 2.70953119e-01
-6.25084162e-01 2.93550193e-01 4.92473692e-01 3.64149243e-01
5.05252063e-01 -1.05802059e+00 -3.38181049e-01 2.83044726e-01
-2.22303540e-01 2.86722153e-01 4.46813524e-01 -5.91061532e-01
-1.11075544e+00 -4.99633372e-01 6.20158613e-01 2.49163769e-02
1.41139939e-01 1.25843793e-01 -7.59756565e-01 5.26367547e-03
1.60905793e-01 2.89497465e-01 5.02350450e-01 -1.73002139e-01
4.69993725e-02 -3.15927476e-01 -1.35823834e+00 3.88213485e-01
5.47113538e-01 -3.21622372e-01 -2.69661427e-01 2.93401033e-01
-1.22959994e-01 -6.11577690e-01 -9.12151814e-01 5.98195314e-01
8.65775883e-01 -1.11355650e+00 9.24206376e-01 -3.42702508e-01
3.40622008e-01 -2.60291398e-01 -2.91800369e-02 -6.95258081e-01
-6.67682827e-01 1.26507685e-01 5.41822553e-01 7.41062820e-01
2.50556022e-01 -5.59733570e-01 6.93032503e-01 -5.34017719e-02
-6.41302094e-02 -7.65335798e-01 -8.85643363e-01 -4.59938526e-01
-6.94342107e-02 -7.26420954e-02 4.38449740e-01 9.86699939e-01
1.06573179e-01 -3.69588315e-01 -2.84209102e-01 -3.70425105e-01
3.86036038e-01 1.78819284e-01 2.47224852e-01 -1.28616357e+00
2.18806982e-01 -8.31056535e-01 -7.35391200e-01 -2.43958503e-01
-4.69613731e-01 -7.01347828e-01 -2.36835092e-01 -1.86244631e+00
6.21897280e-01 -7.29805231e-01 -6.77668273e-01 6.58106983e-01
-1.51824698e-01 7.51291573e-01 -3.09226424e-01 3.81902426e-01
-3.94101918e-01 1.85191184e-01 1.76221704e+00 -4.28796887e-01
1.02317601e-01 -2.15178251e-01 -3.19589049e-01 5.14774859e-01
7.73856223e-01 -2.69015431e-01 -1.16530634e-01 -3.03084970e-01
-1.40936017e-01 -3.91769111e-02 3.19344193e-01 -9.47563112e-01
8.49004462e-02 -4.28901464e-01 8.67796779e-01 -5.08750319e-01
1.51172146e-01 -7.76455879e-01 2.64474303e-01 8.56561244e-01
1.02787025e-01 3.20579171e-01 1.92873135e-01 1.33459046e-01
-7.25972474e-01 -2.16153830e-01 1.05708098e+00 -4.04088557e-01
-9.98502970e-01 2.62648374e-01 -4.82780933e-01 -4.52459723e-01
9.66429412e-01 -7.53852606e-01 -3.05258811e-01 3.26575674e-02
-8.49057615e-01 -3.83790314e-01 2.28222847e-01 2.36525118e-01
7.84685850e-01 -1.39960670e+00 -6.20745599e-01 2.91625321e-01
1.30150050e-01 -4.72579181e-01 7.84847915e-01 1.37619162e+00
-1.37696648e+00 1.17558204e-01 -7.34991252e-01 -9.44217801e-01
-1.65251756e+00 4.99232173e-01 7.32115328e-01 -2.20025867e-01
-8.72145653e-01 5.53898633e-01 4.10135180e-01 -6.06547110e-02
-6.87451661e-02 -2.39875227e-01 -6.22041166e-01 -6.71540871e-02
7.13378072e-01 8.19678679e-02 3.27416390e-01 -7.38951743e-01
-3.31805736e-01 1.32957423e+00 1.14214420e-01 3.70005704e-02
1.12508321e+00 -2.97537237e-01 -7.53763974e-01 4.61242199e-01
1.21898580e+00 -6.79362714e-02 -5.97570062e-01 -2.81619370e-01
-1.18628457e-01 -5.90393722e-01 1.56918973e-01 -1.02620089e+00
-1.59332192e+00 9.12380636e-01 1.28265941e+00 2.40496367e-01
1.51932669e+00 -1.37403220e-01 5.50416052e-01 1.06787339e-01
2.90886432e-01 -1.37111294e+00 2.25130409e-01 3.14218819e-01
6.22400701e-01 -1.28973591e+00 8.51540864e-02 -5.27486920e-01
-7.32370198e-01 1.61210275e+00 3.52219284e-01 -9.34119709e-03
1.07390749e+00 3.88148487e-01 5.15435338e-01 -4.28033203e-01
-4.05042052e-01 -1.30644500e-01 5.21844923e-01 5.73627114e-01
6.33267462e-01 5.48576117e-02 -6.89223766e-01 1.61621436e-01
4.68983911e-02 -3.83157656e-02 1.14118315e-01 1.07043684e+00
-6.99735284e-01 -1.00439250e+00 -7.89995372e-01 6.17521524e-01
-9.05765235e-01 7.64693990e-02 -1.66538283e-01 7.11787641e-01
3.85504931e-01 8.97429049e-01 2.36681312e-01 -3.47206563e-01
8.18982050e-02 -2.18920782e-01 6.87963068e-01 -6.52416609e-03
-3.66074324e-01 2.97719449e-01 -2.03361794e-01 -4.06530619e-01
-6.43427312e-01 -4.98240709e-01 -1.13617957e+00 -5.06851614e-01
-1.01894386e-01 3.99185568e-02 5.93888164e-01 9.09906268e-01
-8.11821073e-02 5.29561341e-01 4.91783082e-01 -3.09630245e-01
2.32119083e-01 -1.02395737e+00 -1.16659057e+00 7.66982079e-01
2.91259259e-01 -7.92922795e-01 -9.40939039e-02 2.52016306e-01]
|
[10.384881973266602, -0.4168902635574341]
|
19fd68a3-8960-4f13-ba8e-a9c926932eb9
|
s3net-3d-lidar-sparse-semantic-segmentation
|
2103.08745
| null |
https://arxiv.org/abs/2103.08745v1
|
https://arxiv.org/pdf/2103.08745v1.pdf
|
S3Net: 3D LiDAR Sparse Semantic Segmentation Network
|
Semantic Segmentation is a crucial component in the perception systems of many applications, such as robotics and autonomous driving that rely on accurate environmental perception and understanding. In literature, several approaches are introduced to attempt LiDAR semantic segmentation task, such as projection-based (range-view or birds-eye-view), and voxel-based approaches. However, they either abandon the valuable 3D topology and geometric relations and suffer from information loss introduced in the projection process or are inefficient. Therefore, there is a need for accurate models capable of processing the 3D driving-scene point cloud in 3D space. In this paper, we propose S3Net, a novel convolutional neural network for LiDAR point cloud semantic segmentation. It adopts an encoder-decoder backbone that consists of Sparse Intra-channel Attention Module (SIntraAM), and Sparse Inter-channel Attention Module (SInterAM) to emphasize the fine details of both within each feature map and among nearby feature maps. To extract the global contexts in deeper layers, we introduce Sparse Residual Tower based upon sparse convolution that suits varying sparsity of LiDAR point cloud. In addition, geo-aware anisotrophic loss is leveraged to emphasize the semantic boundaries and penalize the noise within each predicted regions, leading to a robust prediction. Our experimental results show that the proposed method leads to a large improvement (12\%) compared to its baseline counterpart (MinkNet42 \cite{choy20194d}) on SemanticKITTI \cite{DBLP:conf/iccv/BehleyGMQBSG19} test set and achieves state-of-the-art mIoU accuracy of semantic segmentation approaches.
|
['Liu Bingbing', 'Yuan Ren', 'Ryan Razani', 'Ran Cheng']
|
2021-03-15
|
s3net-3d-lidar-sparse-semantic-segmentation-1
|
https://arxiv.org/abs/2103.08745
|
https://arxiv.org/pdf/2103.08745
| null |
['lidar-semantic-segmentation']
|
['computer-vision']
|
[ 2.25048929e-01 -9.66462716e-02 4.55622599e-02 -7.33814895e-01
-5.57897627e-01 -1.92652807e-01 4.17756587e-01 -1.87568232e-01
-3.84279132e-01 3.32217962e-01 -1.10044345e-01 -2.88218230e-01
-1.97211742e-01 -9.44068968e-01 -1.08118439e+00 -4.10069644e-01
3.25721920e-01 5.98637938e-01 5.80070078e-01 -2.26515666e-01
3.37596655e-01 5.27556360e-01 -1.75303912e+00 1.63195897e-02
1.22102571e+00 1.31861591e+00 6.70355678e-01 1.47212923e-01
-5.86089909e-01 2.35527694e-01 -1.83823630e-01 -2.46456563e-01
4.13907677e-01 1.47505039e-02 -5.11894643e-01 -2.98432764e-02
6.05105996e-01 -1.64756462e-01 -2.66105950e-01 1.23481154e+00
3.63172382e-01 1.28418505e-01 5.80238640e-01 -1.25609779e+00
-4.96015072e-01 3.51860225e-01 -9.04189050e-01 2.48842895e-01
-1.76593512e-01 1.76740736e-01 8.78556490e-01 -1.21663475e+00
3.26575845e-01 1.45955336e+00 6.14091337e-01 1.69099614e-01
-9.74565268e-01 -1.08085418e+00 5.59882581e-01 3.61047655e-01
-1.56122303e+00 -1.50571123e-01 9.39807892e-01 -4.23356026e-01
1.00060594e+00 5.66635057e-02 5.70612133e-01 8.64405990e-01
1.47645712e-01 8.91880870e-01 1.01291478e+00 1.95280820e-01
1.72144994e-01 -5.00716604e-02 2.06312448e-01 5.81176639e-01
2.20295757e-01 1.30028114e-01 -5.05325019e-01 4.02820259e-01
7.60675967e-01 1.92243978e-01 -8.35448783e-03 -5.23355544e-01
-1.00806391e+00 8.84403408e-01 1.10518920e+00 -2.99927630e-02
-2.62446791e-01 2.54581302e-01 1.45958886e-01 -2.27371767e-01
5.29891670e-01 5.10518998e-02 -3.87850910e-01 1.87702939e-01
-8.95638466e-01 3.01573247e-01 2.10802019e-01 1.25377846e+00
1.00317872e+00 2.69056797e-01 1.60473183e-01 7.70359993e-01
5.35980523e-01 8.06924045e-01 1.66252613e-01 -9.71212983e-01
7.40532041e-01 8.19541872e-01 -2.32750833e-01 -9.82454956e-01
-3.96964163e-01 -4.17750061e-01 -7.93867648e-01 1.66823357e-01
-2.66651094e-01 2.59085834e-01 -1.34395385e+00 1.45465171e+00
3.08069646e-01 4.63372558e-01 -1.02251090e-01 1.13801968e+00
1.09785330e+00 6.64499879e-01 1.95862547e-01 3.42802763e-01
1.08361602e+00 -9.43070531e-01 -2.76867956e-01 -5.88528514e-01
2.34656096e-01 -5.59438705e-01 1.08717287e+00 1.52731001e-01
-8.56072545e-01 -8.44009519e-01 -1.16177034e+00 -4.69224840e-01
-6.00890815e-01 -1.13408484e-01 6.36255383e-01 3.31861496e-01
-8.46998632e-01 3.05302739e-01 -9.13162351e-01 -4.10718679e-01
8.53952110e-01 4.29973930e-01 -9.76218283e-02 -2.89461523e-01
-9.21064198e-01 6.26865566e-01 5.07324576e-01 3.69814426e-01
-8.98907483e-01 -7.22196341e-01 -1.08502924e+00 1.69626828e-02
3.65598708e-01 -7.42926240e-01 8.30266237e-01 -5.68076432e-01
-1.13765633e+00 9.14897442e-01 -2.77125299e-01 -4.94137555e-01
4.06618357e-01 -5.30869603e-01 -2.06088901e-01 1.68476277e-03
4.78382498e-01 1.30174541e+00 6.36860013e-01 -1.47521532e+00
-7.96705067e-01 -5.73686361e-01 -3.27258445e-02 4.42817509e-01
1.04508221e-01 -3.00267279e-01 -6.22373521e-01 -3.84972572e-01
6.49481714e-01 -8.11828256e-01 -3.28369200e-01 -1.18876085e-01
-5.85279703e-01 -1.76379561e-01 1.14343226e+00 -5.07960081e-01
6.98284149e-01 -2.23287988e+00 1.05890697e-02 3.25391255e-02
1.15588412e-01 2.35081568e-01 -2.04432681e-02 6.70872815e-03
-3.39520052e-02 2.08834499e-01 -7.16893017e-01 -4.37396377e-01
6.88317567e-02 3.72288704e-01 -4.69337285e-01 3.29244792e-01
3.86096418e-01 1.08803606e+00 -6.19950831e-01 -5.41677594e-01
6.77342594e-01 5.66642225e-01 -6.62364960e-01 6.38603494e-02
-4.15899634e-01 7.04436600e-01 -7.84910858e-01 7.49968886e-01
1.07482708e+00 -1.05972417e-01 -5.04764438e-01 -2.45440871e-01
-3.30022663e-01 2.64996409e-01 -1.14463925e+00 2.16069341e+00
-4.38883513e-01 2.64951408e-01 1.49491310e-01 -1.06637442e+00
9.86987889e-01 -1.19555913e-01 4.59498972e-01 -8.42723250e-01
2.53896624e-01 2.67389923e-01 -1.75115883e-01 -3.00152481e-01
5.58871865e-01 6.88359980e-03 -1.80925086e-01 -1.14653915e-01
2.72674710e-02 -6.80294991e-01 -1.89335823e-01 4.92022969e-02
6.30628109e-01 3.84926617e-01 -1.83866784e-01 -2.61596024e-01
5.38760364e-01 1.34587154e-01 6.95311308e-01 3.61721873e-01
-7.83341080e-02 8.46199274e-01 6.32299259e-02 -3.44843000e-01
-9.19870079e-01 -1.21407747e+00 -4.13437426e-01 5.42213321e-01
7.80638993e-01 -1.46728791e-02 -5.91842175e-01 -4.74715173e-01
2.27106556e-01 8.65726292e-01 -3.71224254e-01 -2.10941494e-01
-6.85230255e-01 -5.92673004e-01 2.68105686e-01 7.86897779e-01
1.05191040e+00 -1.05622315e+00 -7.28247106e-01 5.25137521e-02
-1.22981593e-01 -1.50344229e+00 -1.66529119e-01 3.66033316e-01
-9.29591954e-01 -8.98374200e-01 -2.69747168e-01 -5.80754817e-01
5.07363319e-01 5.48991740e-01 9.06849742e-01 -2.43566871e-01
-1.83181494e-01 -7.11527979e-03 -1.95450693e-01 -5.40389836e-01
4.16747510e-01 3.65677506e-01 -7.12043867e-02 -4.61591817e-02
7.01886833e-01 -7.14691222e-01 -7.25610852e-01 3.11153412e-01
-6.62877917e-01 1.97905436e-01 7.46206284e-01 5.66265523e-01
9.50976551e-01 -9.86145362e-02 3.03642899e-01 -7.42130756e-01
3.67860049e-02 -5.38700938e-01 -7.62666821e-01 -3.10526907e-01
-5.53225815e-01 -1.87524244e-01 4.60236788e-01 2.54619718e-01
-9.32402492e-01 2.04964489e-01 -4.77548391e-01 -8.51435602e-01
-3.93908799e-01 2.94893712e-01 -5.21979213e-01 -1.30904973e-01
2.07748413e-01 2.95917302e-01 -3.25835168e-01 -5.45126677e-01
4.88379478e-01 5.19158125e-01 5.63632071e-01 -4.79959905e-01
9.29928184e-01 6.74889624e-01 -4.11664285e-02 -7.77705491e-01
-9.08919096e-01 -5.26375890e-01 -7.48235106e-01 1.80932116e-02
1.22920406e+00 -1.22606838e+00 -4.83565778e-01 3.46951514e-01
-1.11672997e+00 -1.28333926e-01 -2.89697558e-01 4.48692143e-01
-6.24984205e-01 1.51930243e-01 -1.96803063e-01 -6.68365002e-01
-2.49441803e-01 -1.47981346e+00 1.45958006e+00 4.15988177e-01
2.87497967e-01 -4.25365061e-01 -4.76268560e-01 6.50291324e-01
2.87251055e-01 2.34457374e-01 8.65269721e-01 -5.04292309e-01
-1.24249029e+00 1.43724859e-01 -7.83593297e-01 2.94291943e-01
-2.16297925e-01 -3.26212317e-01 -1.25174642e+00 4.15229388e-02
-8.94256830e-02 -2.68641949e-01 1.21237004e+00 5.35390258e-01
1.48524833e+00 1.88070953e-01 -4.87398803e-01 1.06123137e+00
1.50425935e+00 1.51596591e-01 6.01507425e-01 1.16789900e-01
1.16242254e+00 5.34265876e-01 6.79711640e-01 8.14998224e-02
7.54174352e-01 5.70458949e-01 9.35298502e-01 -1.41709268e-01
-1.35886237e-01 -4.55884278e-01 3.69645990e-02 7.65952170e-01
9.80701298e-02 -1.20763473e-01 -8.61730635e-01 6.84914708e-01
-1.77300096e+00 -5.48727989e-01 -2.22739369e-01 1.97914171e+00
2.54531473e-01 3.77695143e-01 -2.86337495e-01 -2.82740686e-02
5.06871581e-01 3.73766571e-01 -9.75355327e-01 -3.53211880e-01
-1.34573922e-01 4.16691363e-01 7.64748275e-01 4.75144476e-01
-1.12885284e+00 1.49059570e+00 4.88380575e+00 9.71564651e-01
-1.17457771e+00 2.79120862e-01 4.98611301e-01 4.17521372e-02
-4.35960472e-01 -1.96142003e-01 -1.00643384e+00 4.97556597e-01
6.79990649e-01 3.21537226e-01 2.27357417e-01 8.22269559e-01
2.30244294e-01 -2.27355734e-01 -7.15851843e-01 1.02311802e+00
-6.29695952e-02 -1.17804635e+00 5.66194504e-02 1.54462203e-01
6.26051724e-01 5.91091156e-01 1.34582087e-01 3.82297635e-01
3.02158833e-01 -1.18884838e+00 9.29107130e-01 3.62365395e-01
8.13831985e-01 -8.00578773e-01 6.85264170e-01 4.50247407e-01
-1.51495516e+00 -6.31440133e-02 -5.17337322e-01 1.01103708e-01
2.85790861e-01 6.30786240e-01 -5.24371505e-01 7.71766007e-01
1.04973972e+00 9.38552380e-01 -4.33656603e-01 9.52121854e-01
-2.56896734e-01 4.92436588e-01 -5.49068809e-01 2.40202874e-01
5.15567720e-01 -4.49709505e-01 6.04486763e-01 9.25322413e-01
4.16602135e-01 1.77556634e-01 3.50806564e-01 1.35173404e+00
-9.88044403e-03 -6.48074504e-03 -7.02737153e-01 2.22758040e-01
4.45021391e-01 1.14739919e+00 -8.69001389e-01 -1.36936933e-01
-4.19731885e-01 8.55461061e-01 1.56105608e-01 4.45426434e-01
-1.00159574e+00 -2.33500674e-01 8.94544482e-01 1.70398593e-01
6.96194232e-01 -4.78266358e-01 -7.89483607e-01 -9.52134907e-01
7.94852003e-02 -2.76730627e-01 -1.19106518e-03 -9.24592733e-01
-9.33216214e-01 5.13086677e-01 8.27499107e-02 -1.14146197e+00
2.20406830e-01 -5.25796890e-01 -4.59131449e-01 1.01636314e+00
-1.98278511e+00 -1.44818068e+00 -5.43244660e-01 5.39673686e-01
8.24879527e-01 1.26111925e-01 2.90637493e-01 4.89706993e-01
-5.05289316e-01 1.61970377e-01 -2.70344406e-01 -6.75835013e-02
2.53325552e-01 -1.12421227e+00 6.15575671e-01 7.83271015e-01
1.21402368e-02 3.33369315e-01 2.68332005e-01 -7.01586485e-01
-1.18777776e+00 -1.54027474e+00 7.37797976e-01 -4.66949701e-01
3.66306931e-01 -5.64030886e-01 -8.30550313e-01 6.17762566e-01
-8.20774511e-02 1.32179663e-01 1.65580705e-01 -2.53698647e-01
-2.41274238e-01 -2.31626198e-01 -1.04498041e+00 3.62397254e-01
1.51974189e+00 -4.71322119e-01 -5.14236331e-01 1.19079314e-01
1.16156149e+00 -6.40948057e-01 -5.83350003e-01 8.58425558e-01
1.54722899e-01 -1.10081542e+00 1.27099264e+00 -2.62581050e-01
5.02548039e-01 -6.93424582e-01 -5.49208879e-01 -8.51082981e-01
-1.80042818e-01 6.57274351e-02 1.97469935e-01 1.06976748e+00
2.63391703e-01 -5.23964167e-01 8.64785910e-01 2.54288584e-01
-8.69420946e-01 -8.12264919e-01 -1.12509918e+00 -5.75705469e-01
1.18777178e-01 -8.78054023e-01 7.01402009e-01 7.27303207e-01
-7.87359536e-01 4.81341153e-01 8.57497230e-02 5.14576137e-01
6.13379478e-01 4.02609855e-01 7.02931762e-01 -1.34047568e+00
3.36544901e-01 -5.49430966e-01 -6.20062351e-01 -1.37407529e+00
3.34723920e-01 -1.02952492e+00 1.87323317e-01 -1.70795739e+00
-7.65513554e-02 -6.67593479e-01 -2.84536958e-01 2.93686390e-01
-9.27152708e-02 3.62544626e-01 1.65519342e-01 1.27767295e-01
-5.80921173e-01 9.57007229e-01 1.15327477e+00 -2.93323994e-01
-9.53805074e-03 -1.06649995e-01 -6.66170239e-01 7.99416006e-01
5.65777540e-01 -3.45786035e-01 -4.85084116e-01 -8.22510958e-01
1.94569555e-04 -3.14421684e-01 7.61911154e-01 -1.33266294e+00
1.67702109e-01 -1.50732651e-01 3.33000034e-01 -1.28893614e+00
7.08485961e-01 -9.69417930e-01 -5.92819676e-02 2.75884360e-01
1.65149778e-01 3.48569192e-02 1.59964949e-01 6.90433502e-01
-3.33730608e-01 1.49236560e-01 8.40689123e-01 -2.43447036e-01
-1.19693291e+00 7.21741199e-01 1.34365708e-01 7.37030655e-02
8.91520679e-01 -5.57096899e-01 -1.74023472e-02 7.07781082e-03
-3.76639158e-01 7.13592529e-01 4.15822387e-01 5.69971025e-01
8.27670693e-01 -1.18205833e+00 -5.02245486e-01 4.36621428e-01
2.19783321e-01 9.13476169e-01 4.42121953e-01 8.54290724e-01
-4.57418025e-01 5.84345460e-01 -1.35582104e-01 -1.13180101e+00
-5.59706509e-01 3.66803437e-01 2.95418113e-01 1.49129599e-01
-8.76650155e-01 1.20720637e+00 6.47249341e-01 -7.84381092e-01
1.29337057e-01 -6.36321127e-01 -2.49197587e-01 -1.71866477e-01
1.81110995e-03 7.01927245e-02 1.57781392e-01 -1.00670254e+00
-4.80815440e-01 1.03988051e+00 2.29193360e-01 1.38830647e-01
1.28354180e+00 -2.96799242e-01 7.41098002e-02 4.67473924e-01
9.85276520e-01 -3.31662953e-01 -1.46658623e+00 -2.59634882e-01
-1.57164752e-01 -4.76200312e-01 2.53002405e-01 -6.10682666e-01
-1.34854221e+00 1.38488936e+00 8.73214841e-01 -2.71959782e-01
9.40269411e-01 1.06206015e-02 1.15223837e+00 6.55921251e-02
5.09597063e-01 -9.80491221e-01 -2.39323959e-01 7.55292535e-01
7.84669459e-01 -1.48818207e+00 5.38791704e-04 -6.59716964e-01
-6.03823841e-01 6.45677328e-01 9.30315018e-01 -2.82932192e-01
8.19947183e-01 -6.96996674e-02 -2.86842138e-02 -3.99264574e-01
-2.42219165e-01 -5.49936235e-01 2.43501082e-01 5.84900439e-01
7.71421567e-02 1.40694872e-01 2.44874749e-02 6.98481143e-01
-4.32084858e-01 -2.39854440e-01 -4.29763868e-02 6.21144295e-01
-6.01660430e-01 -5.68512261e-01 -1.48703203e-01 4.25131649e-01
8.78796577e-02 -2.51151770e-01 -2.28822201e-01 7.79869139e-01
6.78493202e-01 7.55355656e-01 3.30048114e-01 -5.35271943e-01
4.58860695e-01 3.97669412e-02 1.72057003e-01 -6.68692231e-01
-2.84360796e-01 1.08050980e-01 -1.84284240e-01 -7.54512370e-01
-4.01706070e-01 -6.42652690e-01 -1.64065385e+00 -1.06028073e-01
-2.75441140e-01 -1.35095760e-01 8.36223483e-01 1.07099926e+00
5.45379817e-01 6.91463172e-01 5.78931689e-01 -1.07533908e+00
5.35567349e-04 -8.72090220e-01 -4.90858167e-01 8.09204429e-02
1.25809446e-01 -1.00071573e+00 -1.29349664e-01 -2.16887265e-01]
|
[8.186901092529297, -2.733492612838745]
|
91a7db31-b1c6-48ac-ac4b-898a80bd68e7
|
maestro-open-ended-environment-design-for
|
2303.03376
| null |
https://arxiv.org/abs/2303.03376v1
|
https://arxiv.org/pdf/2303.03376v1.pdf
|
MAESTRO: Open-Ended Environment Design for Multi-Agent Reinforcement Learning
|
Open-ended learning methods that automatically generate a curriculum of increasingly challenging tasks serve as a promising avenue toward generally capable reinforcement learning agents. Existing methods adapt curricula independently over either environment parameters (in single-agent settings) or co-player policies (in multi-agent settings). However, the strengths and weaknesses of co-players can manifest themselves differently depending on environmental features. It is thus crucial to consider the dependency between the environment and co-player when shaping a curriculum in multi-agent domains. In this work, we use this insight and extend Unsupervised Environment Design (UED) to multi-agent environments. We then introduce Multi-Agent Environment Design Strategist for Open-Ended Learning (MAESTRO), the first multi-agent UED approach for two-player zero-sum settings. MAESTRO efficiently produces adversarial, joint curricula over both environments and co-players and attains minimax-regret guarantees at Nash equilibrium. Our experiments show that MAESTRO outperforms a number of strong baselines on competitive two-player games, spanning discrete and continuous control settings.
|
['Tim Rocktäschel', 'Roberta Raileanu', 'Jakob Foerster', 'Jack Parker-Holder', 'Minqi Jiang', 'Michael Dennis', 'Akbir Khan', 'Mikayel Samvelyan']
|
2023-03-06
| null | null | null | null |
['continuous-control']
|
['playing-games']
|
[ 4.50874902e-02 2.15538263e-01 2.25858856e-02 5.09163439e-02
-9.67230916e-01 -1.06715727e+00 6.99332952e-01 1.10886991e-01
-8.95060897e-01 1.21127725e+00 1.47537440e-01 -1.35561898e-01
-5.83820999e-01 -9.30445611e-01 -9.37722683e-01 -8.34773958e-01
-4.03940827e-01 9.15029883e-01 7.89443925e-02 -7.56877244e-01
1.50685310e-01 -4.21412811e-02 -1.48924458e+00 -1.66088060e-01
1.12365472e+00 2.09548116e-01 2.04245821e-01 1.22204173e+00
4.21569258e-01 1.01641202e+00 -7.49939084e-01 -5.20243347e-01
6.76692605e-01 -5.69444716e-01 -7.60491669e-01 1.61409695e-02
1.57422319e-01 -4.27327782e-01 -2.28535607e-01 1.16585350e+00
8.76628041e-01 7.16229260e-01 6.81973994e-01 -1.59551394e+00
-4.99353439e-01 1.09246504e+00 -3.94370794e-01 -4.28656228e-02
2.08346814e-01 7.23119378e-01 1.21985114e+00 -4.08271588e-02
7.32565880e-01 1.34525025e+00 3.89825255e-01 9.27549124e-01
-1.36651337e+00 -4.61674631e-01 4.88224417e-01 3.63188870e-02
-6.56673074e-01 -1.35072013e-02 4.96618956e-01 -3.14981878e-01
6.70184076e-01 1.34465590e-01 6.48407936e-01 1.48142493e+00
2.21700981e-01 9.36028242e-01 1.39115250e+00 -4.11127150e-01
7.00589240e-01 -2.36471966e-02 -6.16675735e-01 5.87045908e-01
1.49345830e-01 7.16123819e-01 -3.19305986e-01 -2.93500811e-01
7.80471504e-01 -5.18168926e-01 3.54587287e-02 -7.83372760e-01
-1.01730347e+00 1.04326284e+00 8.79767910e-02 -2.75666356e-01
-4.99614447e-01 4.84496057e-01 4.96893227e-01 8.64731610e-01
-4.98830713e-02 1.37913263e+00 -4.00692284e-01 -4.66184109e-01
-2.54700631e-01 1.04064381e+00 9.65414464e-01 9.48882103e-01
3.95372093e-01 2.77401447e-01 -1.72559157e-01 7.13710070e-01
9.71330330e-02 3.05070758e-01 4.50399339e-01 -1.52306020e+00
5.48088372e-01 -4.80804257e-02 5.39463401e-01 -3.97089094e-01
-3.80121738e-01 -6.25847340e-01 -2.55262643e-01 1.01295924e+00
5.19376338e-01 -1.00972831e+00 -5.80402374e-01 2.23880553e+00
6.53009295e-01 1.61284953e-01 5.04601538e-01 9.79373991e-01
4.97409552e-01 3.25366467e-01 9.20656621e-02 -1.02209188e-01
1.06196213e+00 -1.31097269e+00 -4.66380745e-01 -4.51499313e-01
3.83402169e-01 -3.69954854e-01 1.12504339e+00 4.43260670e-01
-1.40258765e+00 -2.02343822e-01 -9.05880988e-01 5.62699556e-01
-2.15096563e-01 -7.71512568e-01 4.09375310e-01 6.21113360e-01
-1.01614821e+00 6.47274435e-01 -5.79283416e-01 3.80138978e-02
2.63155729e-01 5.61154366e-01 -1.54340174e-02 3.23655546e-01
-1.35461259e+00 9.34512854e-01 4.13761824e-01 -5.27086079e-01
-1.56417203e+00 -6.97722554e-01 -8.31681490e-01 1.62906662e-01
1.02851915e+00 -7.52484620e-01 1.96283555e+00 -1.09794211e+00
-2.05559134e+00 3.55952173e-01 8.77932310e-01 -4.29988027e-01
1.03498662e+00 4.62191924e-02 1.75896361e-01 -3.90389860e-02
1.49902925e-01 7.89263427e-01 6.08087063e-01 -1.50390351e+00
-1.12389863e+00 5.67142293e-02 7.69817352e-01 1.00199187e+00
-1.79768831e-01 -1.51758492e-01 2.53065765e-01 -4.89132047e-01
-7.65348971e-01 -9.75081801e-01 -9.46236789e-01 -4.97570783e-01
-2.26306468e-01 -1.37743101e-01 2.09008023e-01 2.08139926e-01
7.23210692e-01 -1.73337650e+00 6.44001961e-01 5.47078950e-03
2.55121142e-01 -5.54535128e-02 -7.14491069e-01 5.19615352e-01
7.97183514e-02 -5.33368550e-02 7.36043528e-02 -3.39521378e-01
7.66260326e-01 3.93842787e-01 -9.04870182e-02 2.28579134e-01
6.57599419e-04 9.49230850e-01 -1.45103776e+00 -3.78507167e-01
-3.62822153e-02 -1.95645183e-01 -1.06628323e+00 4.04971480e-01
-8.70309591e-01 6.06204212e-01 -6.39373600e-01 1.91774607e-01
2.56505251e-01 2.03988254e-01 3.24611723e-01 8.37546468e-01
-1.12239502e-01 5.82478940e-02 -1.51128685e+00 1.54336476e+00
-3.81335169e-01 3.05892676e-01 3.14182401e-01 -6.64719462e-01
5.74386299e-01 3.47718179e-01 6.22940898e-01 -6.95387363e-01
1.90563843e-01 6.52452558e-02 4.24388140e-01 -4.87123311e-01
8.25694144e-01 -3.90353113e-01 -5.30398071e-01 6.93215549e-01
1.60204768e-01 -4.84801739e-01 4.32977527e-01 2.00499147e-01
1.31694543e+00 3.65740329e-01 2.49241397e-01 -3.32788795e-01
1.24229535e-01 2.27480546e-01 7.57346451e-01 1.37413061e+00
-6.57512486e-01 3.41231346e-01 8.49348962e-01 -2.39726216e-01
-1.26209509e+00 -1.02043378e+00 3.69075716e-01 1.70181119e+00
2.34224036e-01 -3.83685715e-02 -7.99125552e-01 -6.29836977e-01
1.79456234e-01 7.10789382e-01 -8.52742493e-01 -1.42232716e-01
-6.94196463e-01 -5.80802321e-01 6.96085095e-01 3.01911145e-01
2.20883161e-01 -1.44762826e+00 -1.01748466e+00 4.98430729e-01
-1.97504368e-02 -9.08994794e-01 -7.49641657e-01 5.99687934e-01
-1.41063035e-01 -9.69759881e-01 -5.99146247e-01 -7.71702409e-01
3.12785566e-01 -2.85176784e-01 1.30138433e+00 -3.39308769e-01
-4.87950817e-02 7.16372550e-01 -3.37377548e-01 -6.19279563e-01
-7.51343369e-01 2.52384603e-01 3.83044571e-01 -4.04330045e-01
-4.86464985e-02 -6.82828724e-01 -5.52551746e-01 2.66955853e-01
-8.00079048e-01 6.10069372e-02 3.65871251e-01 1.03883171e+00
2.43278876e-01 2.26879001e-01 8.31921995e-01 -9.36804771e-01
1.26200593e+00 -6.26261294e-01 -9.57537055e-01 1.16694555e-01
-3.86372000e-01 3.11371028e-01 9.56528604e-01 -8.67923081e-01
-1.11256027e+00 -9.02994424e-02 7.93218464e-02 -2.23166287e-01
-2.36576512e-01 2.84953088e-01 -2.31287509e-01 1.90308597e-02
1.05394828e+00 9.15220901e-02 -5.43067865e-02 1.31057993e-01
4.95552063e-01 2.68760920e-01 4.89871472e-01 -1.53547442e+00
1.05508852e+00 -1.32077143e-01 -3.76880616e-02 -1.93288118e-01
-5.99400938e-01 5.27352467e-02 -2.52908021e-01 -4.13782328e-01
7.20673919e-01 -9.83552158e-01 -1.09962618e+00 4.59713548e-01
-6.70455337e-01 -1.26948166e+00 -8.18886459e-01 3.15343589e-01
-1.23479092e+00 -4.15774249e-02 -4.93405163e-01 -7.74055421e-01
6.79917410e-02 -1.52589929e+00 5.62950253e-01 5.62324882e-01
4.66016531e-02 -1.13408530e+00 6.51639402e-01 3.23662460e-01
3.28671455e-01 4.07296568e-01 6.78251505e-01 -6.54046237e-01
-5.05265892e-01 4.78581756e-01 5.19674778e-01 -9.68832448e-02
-1.17858179e-01 -3.13872457e-01 -6.20969594e-01 -4.86896843e-01
-4.25629348e-01 -1.18446910e+00 2.30290622e-01 2.90391952e-01
8.03066194e-01 -4.99313354e-01 2.96924829e-01 5.16938031e-01
1.35285091e+00 2.71941930e-01 3.39145422e-01 1.11720824e+00
2.61962622e-01 4.99713004e-01 8.61020267e-01 8.47311020e-01
4.98391807e-01 5.04226565e-01 8.43796670e-01 2.59791762e-01
3.50817919e-01 -3.15007061e-01 5.30786455e-01 2.13675141e-01
-8.21620226e-02 -3.34456265e-01 -6.61733449e-01 5.15498996e-01
-2.16127944e+00 -1.28596199e+00 3.75862926e-01 1.96292961e+00
1.34231007e+00 2.33097479e-01 6.36883795e-01 -4.20434743e-01
5.14122188e-01 1.83942273e-01 -1.08907461e+00 -7.11575210e-01
-1.18086256e-01 3.09924453e-01 4.94477481e-01 6.90165460e-01
-1.11415780e+00 1.01279867e+00 6.08230400e+00 8.10788095e-01
-5.10711670e-01 1.05705798e-01 4.62698489e-01 -4.92714167e-01
-3.61849010e-01 -3.54651332e-01 -5.29570580e-01 2.76793540e-01
7.75613070e-01 -4.00663674e-01 1.03251195e+00 9.93882179e-01
3.31272513e-01 1.08777873e-01 -1.28593981e+00 4.26549643e-01
-4.33402926e-01 -1.07861304e+00 -5.39795518e-01 1.97464064e-01
1.38411880e+00 5.45588285e-02 3.84327680e-01 9.21408951e-01
1.58674896e+00 -1.14523149e+00 9.55810726e-01 -1.87814072e-01
6.49760544e-01 -1.10678875e+00 3.29681665e-01 5.33291161e-01
-8.82166028e-01 -4.04445738e-01 -1.85063541e-01 -3.33780736e-01
-1.97315991e-01 -5.48227668e-01 -7.76395500e-01 3.64361078e-01
3.33889574e-01 -6.98832050e-02 -4.81280237e-02 1.07579672e+00
-1.93528071e-01 3.81929904e-01 -1.57065347e-01 -2.57417887e-01
9.15599346e-01 -3.73537779e-01 8.29666734e-01 8.34436059e-01
-2.73413241e-01 3.71401608e-01 7.06930280e-01 8.61397445e-01
-9.40869376e-02 -9.99303758e-02 -4.78993475e-01 7.78008476e-02
6.40915632e-01 1.21281123e+00 -4.69912291e-01 2.40431931e-02
-5.47040552e-02 7.02965617e-01 6.60378695e-01 3.83679599e-01
-1.08267283e+00 -3.26324940e-01 1.27840865e+00 -3.57301950e-01
3.32848817e-01 -3.73204350e-02 -9.48823765e-02 -1.05327213e+00
-4.06835705e-01 -1.68395054e+00 4.92284089e-01 -4.37014252e-01
-1.34973562e+00 3.53647143e-01 -5.51232733e-02 -1.22856498e+00
-5.57175696e-01 -4.65013385e-01 -7.98148394e-01 4.80549544e-01
-1.52700996e+00 -7.96236932e-01 7.74369389e-02 5.23100913e-01
7.21316397e-01 -4.80598688e-01 7.73277462e-01 -5.58151677e-02
-6.06589079e-01 8.52680922e-01 5.71267605e-01 7.39919916e-02
7.65431285e-01 -1.97678304e+00 3.50538582e-01 5.96234977e-01
-1.74326420e-01 1.05996706e-01 1.14958572e+00 -3.77249449e-01
-1.41775072e+00 -9.51515853e-01 -2.58518547e-01 -4.39031929e-01
9.44077134e-01 -2.92317241e-01 -2.76997924e-01 5.90131462e-01
4.77506459e-01 -2.97583967e-01 5.25389135e-01 2.44580746e-01
-1.41336799e-01 2.51916528e-01 -1.21189284e+00 1.28697968e+00
1.10047376e+00 -7.67624825e-02 -5.09638667e-01 3.00533652e-01
9.96391475e-01 -1.04131413e+00 -8.48726034e-01 -6.60315007e-02
2.99322248e-01 -7.04264879e-01 9.31186259e-01 -1.22827005e+00
9.12882328e-01 1.27955433e-02 2.70746890e-02 -2.11865282e+00
-4.34711635e-01 -1.21209800e+00 2.71799415e-02 8.63787591e-01
3.52349609e-01 -5.03913462e-01 9.90724862e-01 6.31298900e-01
-2.38477558e-01 -7.52937853e-01 -7.57566690e-01 -8.68737936e-01
8.76161456e-01 -8.10451359e-02 6.26902044e-01 8.63959670e-01
1.77390903e-01 2.74680585e-01 -4.93499756e-01 2.65934885e-01
8.88610840e-01 2.89750868e-04 1.08692396e+00 -8.80221725e-01
-1.01820254e+00 -7.01998889e-01 1.11691333e-01 -8.85268271e-01
3.41250449e-01 -4.63098168e-01 4.64881808e-01 -1.04016650e+00
1.56975299e-01 -9.08681035e-01 -2.64467329e-01 3.94549280e-01
-4.31090474e-01 -2.61801779e-01 4.62380588e-01 -3.53792459e-01
-1.17705250e+00 8.21015179e-01 1.75464940e+00 -2.00401440e-01
-4.65352893e-01 1.09680191e-01 -1.01800549e+00 5.03957272e-01
1.03089309e+00 -4.07518357e-01 -7.23788321e-01 -5.29355049e-01
2.77977675e-01 2.54688382e-01 7.87926316e-02 -8.70401263e-01
2.98145890e-01 -1.09354019e+00 -9.01414081e-03 2.43965700e-01
2.22578570e-01 -5.88018239e-01 -1.73547909e-01 4.97166872e-01
-8.53171527e-01 1.07570104e-01 9.05855298e-02 6.64588749e-01
2.40541279e-01 -6.09477341e-01 7.67537653e-01 -5.93875647e-01
-4.90390241e-01 3.64125252e-01 -7.51672387e-01 7.95947015e-01
1.36405027e+00 -9.79267247e-03 -5.02808750e-01 -6.84325457e-01
-6.18230462e-01 9.71841812e-01 3.82708997e-01 3.06385100e-01
2.47736350e-01 -1.15297866e+00 -1.00570118e+00 -2.20773175e-01
-3.02611236e-02 3.07173729e-01 1.39340147e-01 2.79683080e-02
-3.15069795e-01 -1.37753010e-01 -5.71695089e-01 -7.21227378e-02
-1.26819539e+00 3.36031616e-01 7.78617382e-01 -7.43136287e-01
-4.05956119e-01 8.53712738e-01 1.57862008e-01 -1.08850014e+00
5.95095992e-01 1.35206833e-01 -1.74711481e-01 -6.09182902e-02
4.21280384e-01 2.16710806e-01 -4.06724274e-01 -3.38129550e-02
2.64589012e-01 -1.45552773e-02 -2.25198567e-01 -7.97220409e-01
1.61926973e+00 5.41367829e-02 4.50831085e-01 4.26130034e-02
4.37976688e-01 -1.06130093e-01 -1.93445063e+00 -2.47946635e-01
-2.10668921e-01 -3.83372575e-01 -2.45987520e-01 -9.70838845e-01
-6.96472585e-01 3.26133430e-01 2.62954473e-01 2.73356020e-01
8.14458489e-01 -3.54871720e-01 3.69771123e-01 6.74246252e-01
7.32006431e-01 -1.52373505e+00 7.43832648e-01 7.96235979e-01
6.86482728e-01 -1.27872682e+00 -2.17150211e-01 2.45375052e-01
-1.13954318e+00 8.84278178e-01 1.35035729e+00 -3.02329153e-01
2.84176599e-02 4.77256358e-01 1.95323750e-01 3.38141583e-02
-1.25259030e+00 -4.30048525e-01 -3.16997945e-01 1.09557521e+00
-1.44267127e-01 3.45259368e-01 5.60134766e-04 5.82531273e-01
-5.52746475e-01 -4.92456287e-01 1.02171135e+00 1.15752757e+00
-5.40463209e-01 -1.46329355e+00 -5.73947370e-01 1.63095385e-01
-3.08405697e-01 2.34067619e-01 -2.93432385e-01 7.64947414e-01
1.27069995e-01 9.70845282e-01 -1.37483284e-01 -2.13845566e-01
3.09403300e-01 -2.55654424e-01 7.64538407e-01 -5.88728070e-01
-1.16854179e+00 -1.72956631e-01 1.83917910e-01 -3.44555885e-01
1.82987552e-03 -7.77310312e-01 -1.20263314e+00 -4.70699638e-01
-1.82282776e-01 3.85913044e-01 2.64107794e-01 7.88674474e-01
2.19265223e-02 9.38143075e-01 8.26466739e-01 -8.27356219e-01
-1.44743383e+00 -5.91458678e-01 -4.22581524e-01 4.30703312e-01
4.50490445e-01 -7.55126476e-01 -1.62663728e-01 -2.63233572e-01]
|
[3.7963340282440186, 1.762798547744751]
|
288a8e8f-1781-4a45-b7ba-0c6788661057
|
tsfd-net-tissue-specific-feature-distillation
| null | null |
https://www.sciencedirect.com/science/article/pii/S0893608022000612?via%3Dihub
|
https://www.sciencedirect.com/science/article/pii/S0893608022000612?via%3Dihub
|
TSFD-Net: Tissue specific feature distillation network for nuclei segmentation and classification
|
Nuclei segmentation and classification of hematoxylin and eosin-stained histology images is a challenging task due to a variety of issues, such as color inconsistency that results from the non-uniform manual staining operations, clustering of nuclei, and blurry and overlapping nuclei boundaries. Existing approaches involve segmenting nuclei by drawing their polygon representations or by measuring the distances between nuclei centroids. In contrast, we leverage the fact that morphological features (appearance, shape, and texture) of nuclei in a tissue vary greatly depending upon the tissue type. We exploit this information by extracting tissue specific (TS) features from raw histopathology images using the proposed tissue specific feature distillation (TSFD) backbone. The bi-directional feature pyramid network (BiFPN) within TSFD-Net generates a robust hierarchical feature pyramid utilizing TS features where the interlinked decoders jointly optimize and fuse these features to generate final predictions. We also propose a novel combinational loss function for joint optimization and faster convergence of our proposed network. Extensive ablation studies are performed to validate the effectiveness of each component of TSFD-Net. The proposed network outperforms state-of-the-art networks such as StarDist, Micro-Net, Mask-RCNN, Hover-Net, and CPP-Net on the PanNuke dataset, which contains 19 different tissue types and 5 clinically important tumor classes, achieving 50.4% and 63.77% mean and binary panoptic quality, respectively.
|
['Friso De Boer', 'Hyongsuk Kim', 'Sami Azam', 'Abbas Khan', 'Zubaer Ibna Mannan', 'Talha Ilyas']
|
2022-03-09
| null | null | null |
elsevier-neural-networks-2022-3
|
['panoptic-segmentation']
|
['computer-vision']
|
[ 2.39389122e-01 -1.51868090e-01 -3.11046909e-03 -3.32541496e-01
-9.46976483e-01 -5.83359480e-01 1.32203281e-01 3.32848310e-01
-7.21018910e-01 8.07194948e-01 -7.26444498e-02 -5.43881170e-02
-5.84295020e-02 -5.70063114e-01 -1.96400791e-01 -1.32842588e+00
-8.01057369e-02 2.84534067e-01 1.89204752e-01 1.15390331e-01
1.56961650e-01 7.56175399e-01 -8.99990857e-01 2.59485751e-01
8.13357472e-01 1.09017539e+00 9.45764631e-02 7.79873312e-01
1.20360970e-01 4.49699491e-01 -3.63688171e-01 -3.15293223e-01
1.18263483e-01 -2.18189299e-01 -6.91003382e-01 8.19469318e-02
2.69956023e-01 -2.44912520e-01 -2.55900830e-01 1.25391006e+00
5.67800224e-01 -3.15478891e-01 9.24470603e-01 -1.08014154e+00
-3.52240354e-01 4.34178084e-01 -9.53494668e-01 3.23400646e-01
-5.13090432e-01 2.52814740e-01 9.16104317e-01 -7.10836112e-01
6.68866515e-01 7.00671852e-01 8.34562182e-01 3.66071761e-01
-1.25184298e+00 -7.62906611e-01 -3.47593784e-01 7.53652602e-02
-1.71134996e+00 -1.37095436e-01 4.42025155e-01 -4.84512210e-01
6.54087782e-01 3.83260608e-01 7.56785691e-01 4.75934923e-01
5.99948823e-01 6.69322193e-01 1.11419725e+00 -9.65519845e-02
-2.79474650e-02 9.02630109e-03 3.27501893e-02 9.82506990e-01
3.20452780e-01 -3.49934191e-01 -9.84914154e-02 -5.17734922e-02
8.32570016e-01 1.67183161e-01 -4.34255034e-01 -1.06667951e-01
-1.42037439e+00 6.27818584e-01 6.77090466e-01 3.41645926e-01
-2.33919024e-01 1.41713858e-01 4.76928830e-01 -1.82473093e-01
3.21277976e-01 1.82847589e-01 -1.76799476e-01 9.34818909e-02
-1.08616745e+00 3.43821757e-02 5.09092391e-01 5.28610170e-01
6.71735585e-01 -2.41099477e-01 -4.10161912e-01 7.59337246e-01
5.16537964e-01 1.89557970e-01 6.90238476e-01 -8.21873307e-01
-6.94378689e-02 7.64324844e-01 -2.85783917e-01 -1.07659984e+00
-8.24440300e-01 -7.70667851e-01 -1.36405694e+00 1.23199582e-01
4.83795166e-01 -1.70748219e-01 -1.20311713e+00 1.65214658e+00
5.23775816e-01 1.69827163e-01 -1.87396571e-01 8.40724349e-01
9.45937932e-01 2.84754604e-01 6.23478666e-02 -2.34600589e-01
1.24787962e+00 -1.09818399e+00 -6.83690906e-01 1.18700638e-01
8.78573477e-01 -6.11340761e-01 5.73723614e-01 1.29171282e-01
-8.00156534e-01 -6.05486706e-02 -1.07673311e+00 -8.66260901e-02
-4.42037821e-01 4.15766448e-01 6.36393487e-01 4.00704741e-01
-1.25941229e+00 5.27186334e-01 -9.78559792e-01 -2.56243438e-01
8.21492314e-01 8.36045086e-01 -5.44933259e-01 6.63146973e-02
-6.68060005e-01 5.40508926e-01 2.99034476e-01 5.24409711e-01
-6.85311317e-01 -9.12386477e-01 -6.77567184e-01 -1.53815839e-02
-1.35850444e-01 -8.12759876e-01 9.52197731e-01 -4.97101665e-01
-1.21884632e+00 9.72134411e-01 -1.02583930e-01 -1.62381887e-01
5.63829243e-01 5.12670279e-01 -6.41641468e-02 3.71638864e-01
4.71299849e-02 1.05494690e+00 2.94450432e-01 -8.95028830e-01
-7.32197046e-01 -4.67688918e-01 -4.24832672e-01 2.87648857e-01
-2.66225278e-01 -3.82104158e-01 -5.51319420e-01 -5.32781661e-01
4.81879078e-02 -8.84819388e-01 -4.70093101e-01 3.80730957e-01
-8.79570663e-01 4.90416288e-02 7.83909440e-01 -7.16408253e-01
1.14064574e+00 -2.27072501e+00 1.43635288e-01 5.22208214e-01
6.51870131e-01 -3.87766697e-02 -1.11245774e-01 -1.61968380e-01
1.17815897e-01 3.88748169e-01 -2.69845188e-01 -2.98509657e-01
-1.03181124e-01 -1.34803038e-02 4.30177689e-01 8.85854542e-01
-2.10273489e-02 8.31092715e-01 -8.38713169e-01 -1.01499557e+00
5.84210865e-02 6.28034472e-01 -5.64923882e-01 -8.07392970e-02
3.22159439e-01 2.07417324e-01 -2.63340414e-01 1.02996445e+00
7.07219183e-01 -5.65390646e-01 2.08373189e-01 -6.48669839e-01
8.24173540e-02 -3.94148499e-01 -8.42634618e-01 1.39768517e+00
3.30345593e-02 5.08971453e-01 3.42786193e-01 -6.48060024e-01
5.15974760e-01 2.55718440e-01 7.71940231e-01 -1.88179299e-01
4.54958647e-01 3.51290733e-01 2.81778634e-01 -3.08043629e-01
1.77656442e-01 -2.80795813e-01 7.80470967e-02 1.86841071e-01
2.82281011e-01 -1.05848312e-01 5.28224409e-01 2.28508785e-02
1.33067775e+00 -5.06703615e-01 3.29104692e-01 -2.87717521e-01
5.17553747e-01 -1.43115103e-01 8.58177185e-01 3.35485011e-01
-5.78523397e-01 8.39578152e-01 7.55015135e-01 -1.49898663e-01
-7.98635602e-01 -9.87487555e-01 -3.52469534e-01 4.95384604e-01
1.11687288e-01 -9.80513692e-02 -6.46208942e-01 -9.65707421e-01
-6.05367646e-02 7.99076185e-02 -1.01694393e+00 -3.42909545e-02
-2.27176785e-01 -1.37219918e+00 6.68704331e-01 4.34399873e-01
4.40628409e-01 -7.12930441e-01 -2.49827236e-01 9.85146761e-02
-1.80417478e-01 -9.74295318e-01 -6.32297158e-01 3.73695761e-01
-8.47244740e-01 -1.12550914e+00 -6.94256365e-01 -8.82318914e-01
1.29573333e+00 1.12529792e-01 7.26435602e-01 2.95851260e-01
-7.34529018e-01 -8.01669508e-02 -2.97494810e-02 -3.50976773e-02
-2.10161805e-01 1.62927970e-01 -4.56027329e-01 6.41964525e-02
9.28472653e-02 -5.40134668e-01 -9.94224489e-01 2.55807310e-01
-1.05080903e+00 2.14229465e-01 1.00134885e+00 1.13455176e+00
1.16591716e+00 2.56039619e-01 2.42050409e-01 -9.13383067e-01
4.19201761e-01 -4.36433613e-01 -3.20072472e-01 2.88481534e-01
-3.30273032e-01 -2.94699192e-01 4.86196548e-01 -8.11281428e-02
-6.64013505e-01 2.05501109e-01 -5.96220940e-02 -2.55514264e-01
6.25036657e-03 5.89059770e-01 1.27670631e-01 -5.11215389e-01
2.79036164e-01 1.46009535e-01 2.66979903e-01 1.43033281e-01
-1.57354046e-02 5.95998287e-01 5.07164359e-01 -3.40716657e-03
7.57704675e-01 7.24840939e-01 2.19853252e-01 -5.43907166e-01
-6.19682729e-01 -4.69243258e-01 -5.12729764e-01 -1.69390664e-01
9.52643037e-01 -6.76644206e-01 -7.75066733e-01 8.47300172e-01
-9.18713629e-01 -2.43774682e-01 -1.62518751e-02 4.31039333e-01
-2.32538059e-01 2.09734142e-01 -1.13896549e+00 -1.31507635e-01
-7.01467037e-01 -1.38475096e+00 9.88505661e-01 5.64858198e-01
-1.42386407e-01 -1.04948580e+00 3.38222943e-02 3.82089913e-01
3.44314843e-01 6.36238813e-01 1.22442472e+00 -5.84990203e-01
-2.83599168e-01 -3.11591983e-01 -5.29773891e-01 3.15687180e-01
4.77040261e-01 6.00251853e-01 -7.13024378e-01 -4.54519302e-01
-4.62812364e-01 -2.88122594e-01 8.22372913e-01 6.39969110e-01
1.35238683e+00 -1.24281481e-01 -6.81074321e-01 1.02387106e+00
1.72068703e+00 2.23030910e-01 4.34660703e-01 3.12836051e-01
5.98014534e-01 3.31218839e-01 3.27451348e-01 3.64770919e-01
5.09077191e-01 2.33844325e-01 5.70558906e-01 -6.89145565e-01
-1.99848518e-01 3.32382321e-01 -1.31613582e-01 7.99275458e-01
7.78248087e-02 -1.28841490e-01 -1.00148630e+00 7.88564503e-01
-1.51671517e+00 -6.41406298e-01 2.31639463e-02 1.65718484e+00
1.11643887e+00 -4.37951088e-02 -1.49843663e-01 3.76602361e-06
7.59368420e-01 -1.41610458e-01 -6.73890769e-01 -2.19028462e-02
7.14349234e-03 4.42776382e-02 6.73863292e-01 2.01184273e-01
-1.20924985e+00 4.94988054e-01 5.99282169e+00 1.05154979e+00
-1.41320896e+00 -5.87993488e-02 1.23963320e+00 -1.64129674e-01
2.13705171e-02 -4.70828265e-01 -6.83332384e-01 4.43162113e-01
3.57236773e-01 -1.83493569e-01 1.20271161e-01 3.39623570e-01
1.45209476e-01 -1.53474092e-01 -9.31541145e-01 8.11117053e-01
-1.09931037e-01 -1.54222703e+00 -7.83819258e-02 2.09889784e-01
7.30217397e-01 1.85568362e-01 2.47713342e-01 -1.26143470e-02
3.23674858e-01 -1.21187735e+00 2.95474946e-01 3.77867728e-01
9.59908485e-01 -7.74370670e-01 1.25111175e+00 -1.44906655e-01
-1.05369449e+00 9.38192084e-02 -1.92796469e-01 6.85012877e-01
-2.25143805e-01 7.35346973e-01 -1.28778052e+00 4.39083487e-01
5.46929896e-01 6.92339540e-01 -7.99374044e-01 1.47556174e+00
1.69872552e-01 5.76202631e-01 -4.74191517e-01 -6.72959015e-02
3.27373773e-01 -3.80801819e-02 2.58496404e-01 1.39668226e+00
3.93352717e-01 -6.12500980e-02 -9.84403566e-02 8.62129509e-01
-2.22502232e-01 5.07510379e-02 1.44660667e-01 -1.49165228e-01
4.59887385e-01 2.04488015e+00 -1.36656415e+00 9.28204414e-03
-2.49192521e-01 5.23621798e-01 3.31506908e-01 2.83755183e-01
-9.04190779e-01 -5.45305014e-01 5.13479710e-01 1.78517252e-02
3.06050926e-01 1.12937793e-01 -4.17530149e-01 -7.22418725e-01
-3.53151411e-01 -6.84658289e-01 3.70633781e-01 -3.27026635e-01
-1.34963119e+00 6.80634201e-01 -4.72253084e-01 -1.23428118e+00
3.04392755e-01 -4.53402311e-01 -6.27660215e-01 5.67750692e-01
-1.61573219e+00 -1.17801464e+00 -5.68117321e-01 4.38522875e-01
1.29836962e-01 6.79581612e-02 7.29780018e-01 3.10563654e-01
-9.24398363e-01 8.08364272e-01 1.82493865e-01 4.76429880e-01
5.78352511e-01 -1.46810496e+00 -2.99080133e-01 5.55249155e-01
-4.44831669e-01 4.48077828e-01 3.29389483e-01 -3.89338344e-01
-1.19102812e+00 -1.40307224e+00 5.20674229e-01 2.69103646e-01
6.56778932e-01 -1.50049791e-01 -6.05427563e-01 3.87009829e-01
1.13641232e-01 6.35676205e-01 1.25460684e+00 -4.32003289e-01
1.30242348e-01 -3.33462417e-01 -1.47169816e+00 6.55856431e-01
4.05746460e-01 -1.87306315e-01 1.41144544e-01 3.77845526e-01
3.33430439e-01 -5.67366660e-01 -1.13650072e+00 5.18960834e-01
5.21033823e-01 -8.87175798e-01 5.86265504e-01 -2.05648746e-02
3.97030056e-01 -4.91720229e-01 -1.06107835e-02 -1.28485596e+00
-6.24102056e-01 -2.91476548e-01 4.61631060e-01 1.06141281e+00
5.83679616e-01 -4.89485383e-01 1.09551895e+00 3.34460288e-01
-3.40985298e-01 -1.39894664e+00 -1.08837950e+00 -1.52851358e-01
1.01458393e-01 1.86305642e-01 3.77366364e-01 8.20302427e-01
9.25870705e-03 2.07850672e-02 2.45946810e-01 1.39682457e-01
6.89696729e-01 -8.07683021e-02 3.21787179e-01 -9.50011790e-01
-5.85311502e-02 -7.18108416e-01 -7.47355759e-01 -4.19239730e-01
-8.10430571e-02 -1.18203640e+00 9.59351510e-02 -1.60662520e+00
6.18030608e-01 -5.38291931e-01 -7.40789473e-01 7.61384189e-01
-2.01072589e-01 7.15053141e-01 8.06283765e-03 2.07094811e-02
-7.70329654e-01 3.60182345e-01 1.73314726e+00 -3.42969328e-01
5.03301099e-02 -4.10316557e-01 -8.48092079e-01 7.47807920e-01
6.52738869e-01 -4.58942652e-01 -1.56888172e-01 -2.26738006e-01
-2.24620868e-02 8.25927928e-02 3.81983012e-01 -1.14420676e+00
4.63078856e-01 -6.20784760e-02 8.73075604e-01 -6.53871834e-01
1.99289992e-01 -7.14745045e-01 8.06815699e-02 5.37352920e-01
-1.38232648e-01 -1.92339703e-01 2.08360851e-01 4.06819284e-01
-4.34733868e-01 7.17098266e-02 9.47007537e-01 -5.06332144e-02
-7.50605986e-02 6.02662504e-01 -4.08633590e-01 -2.10793674e-01
1.25086343e+00 -5.77550054e-01 -5.99559367e-01 5.98033406e-02
-5.93509972e-01 4.50760245e-01 3.44934821e-01 -2.42577523e-01
6.45906448e-01 -1.28668797e+00 -7.40756094e-01 5.36377318e-02
-1.56846583e-01 4.64707226e-01 5.97078383e-01 1.55163980e+00
-9.06813443e-01 3.02474171e-01 -3.30225706e-01 -8.33873928e-01
-1.37461042e+00 -1.95293143e-01 6.93543315e-01 -7.25994587e-01
-1.49480090e-01 1.26504350e+00 5.06697237e-01 -4.43245500e-01
6.01570569e-02 -5.23537517e-01 -1.77845150e-01 -5.54834306e-02
2.03463644e-01 2.81977713e-01 2.26435542e-01 -4.92104083e-01
-5.17161131e-01 4.82194722e-01 -5.49587071e-01 3.37646335e-01
1.36197209e+00 1.56615257e-01 -5.50571620e-01 1.61949061e-02
1.40018260e+00 -2.74028748e-01 -1.28924394e+00 -9.28610340e-02
-3.04513484e-01 -1.09832674e-01 3.79738301e-01 -1.00913048e+00
-1.55849004e+00 6.07199192e-01 7.41352558e-01 -9.41122249e-02
1.35852408e+00 -2.27801129e-01 8.50853562e-01 -5.77267446e-02
-2.78622061e-02 -8.14685881e-01 -1.31663352e-01 1.59537628e-01
5.18663883e-01 -9.85291779e-01 7.71420449e-02 -5.40684819e-01
-3.09336931e-01 1.25048196e+00 7.11607456e-01 3.91477607e-02
6.46214545e-01 6.86123729e-01 1.46308243e-01 -2.54510701e-01
-8.70954633e-01 1.36886716e-01 1.43793836e-01 3.39833915e-01
4.42353547e-01 5.05119115e-02 -2.19666243e-01 7.79040813e-01
-1.76996440e-01 7.74410889e-02 5.19330919e-01 8.43221188e-01
-2.74716884e-01 -7.70495713e-01 -1.04225248e-01 8.39358091e-01
-8.07620585e-01 -1.70786008e-02 -3.42994511e-01 9.43707824e-01
2.78100997e-01 4.62727576e-01 7.89902508e-02 -4.79112208e-01
-6.96185157e-02 -5.52049041e-01 4.12865847e-01 -4.46291387e-01
-6.86001480e-01 4.20029402e-01 -2.52760053e-01 -2.11178467e-01
-2.96757221e-01 -6.13943636e-01 -1.59820449e+00 -2.23869830e-01
-4.53744531e-01 6.94163516e-02 4.36776221e-01 7.95605063e-01
2.00698301e-01 7.74052083e-01 6.94104731e-01 -6.27305925e-01
-3.58585536e-01 -8.40747476e-01 -9.14725959e-01 4.33012210e-02
4.66962039e-01 -5.21861494e-01 -4.89702374e-01 1.02755472e-01]
|
[15.038466453552246, -3.059568166732788]
|
7bc60c7a-dbcc-45f2-8b81-daf71857f0ce
|
deep-neural-network-based-respiratory
|
2106.12174
| null |
https://arxiv.org/abs/2106.12174v1
|
https://arxiv.org/pdf/2106.12174v1.pdf
|
Deep Neural Network Based Respiratory Pathology Classification Using Cough Sounds
|
Intelligent systems are transforming the world, as well as our healthcare system. We propose a deep learning-based cough sound classification model that can distinguish between children with healthy versus pathological coughs such as asthma, upper respiratory tract infection (URTI), and lower respiratory tract infection (LRTI). In order to train a deep neural network model, we collected a new dataset of cough sounds, labelled with clinician's diagnosis. The chosen model is a bidirectional long-short term memory network (BiLSTM) based on Mel Frequency Cepstral Coefficients (MFCCs) features. The resulting trained model when trained for classifying two classes of coughs -- healthy or pathology (in general or belonging to a specific respiratory pathology), reaches accuracy exceeding 84\% when classifying cough to the label provided by the physicians' diagnosis. In order to classify subject's respiratory pathology condition, results of multiple cough epochs per subject were combined. The resulting prediction accuracy exceeds 91\% for all three respiratory pathologies. However, when the model is trained to classify and discriminate among the four classes of coughs, overall accuracy dropped: one class of pathological coughs are often misclassified as other. However, if one consider the healthy cough classified as healthy and pathological cough classified to have some kind of pathologies, then the overall accuracy of four class model is above 84\%. A longitudinal study of MFCC feature space when comparing pathological and recovered coughs collected from the same subjects revealed the fact that pathological cough irrespective of the underlying conditions occupy the same feature space making it harder to differentiate only using MFCC features.
|
['Jer Ming Chen', 'Dorien Herremans', 'Khai Pin Lee', 'Sung Shin Teng', 'Oon Hoe Teoh', 'Saumitra Kapoor', 'Hwan Ing Hee', 'Balamurali B T']
|
2021-06-23
| null | null | null | null |
['sound-classification']
|
['audio']
|
[ 3.12988281e-01 -1.14644229e-01 -2.53301173e-01 -3.30252163e-02
-3.40333909e-01 -5.27473032e-01 4.65240553e-02 2.67328292e-01
-5.87881804e-01 4.99201953e-01 1.75508428e-02 -5.19106388e-01
-1.44129813e-01 -7.89440453e-01 -4.33150619e-01 -9.36634004e-01
1.60317719e-01 6.92183137e-01 3.46724242e-02 3.43471259e-01
-6.68646097e-02 1.83286846e-01 -1.99172127e+00 4.59363788e-01
9.06506479e-01 7.64811993e-01 5.87502718e-01 1.29021633e+00
-9.57039297e-02 3.20173740e-01 -8.22407603e-01 1.79995686e-01
-3.09728116e-01 -4.00070995e-01 -1.09275830e+00 -5.21549702e-01
3.96133423e-01 -5.70969641e-01 3.41343731e-01 4.99787033e-01
7.02730179e-01 -2.66370326e-01 7.13009059e-01 -6.75879896e-01
-5.75487971e-01 2.68020391e-01 1.24185517e-01 3.37155849e-01
2.72452414e-01 5.09819537e-02 9.41561520e-01 -4.89055246e-01
-2.37945025e-03 8.64018738e-01 8.97873759e-01 9.25541103e-01
-1.07365739e+00 -5.59414983e-01 -2.73874819e-01 1.16217874e-01
-5.99414229e-01 1.69444486e-01 1.49473682e-01 -6.10857368e-01
1.08451819e+00 5.11231363e-01 1.08033514e+00 1.18540764e+00
1.97291285e-01 3.86329979e-01 1.12920332e+00 -2.52908617e-01
1.39801025e-01 2.31312260e-01 6.37683809e-01 2.98228920e-01
5.21866739e-01 2.96854198e-01 9.83807296e-02 -3.81421030e-01
1.38344780e-01 3.61814290e-01 -1.53598204e-01 4.16184366e-01
-1.36410201e+00 8.50268066e-01 3.98587823e-01 1.01209044e+00
-5.52289605e-01 1.30045950e-01 3.70729566e-01 3.81678104e-01
3.44594628e-01 4.41622227e-01 -7.17072189e-01 -7.55916312e-02
-8.74739707e-01 1.57048568e-01 9.28966582e-01 4.52761650e-02
4.50932562e-01 -8.77165608e-03 -3.10338110e-01 1.11966753e+00
1.54628441e-01 9.21705306e-01 9.06937480e-01 -6.22135639e-01
9.78431851e-02 4.24206197e-01 -8.52329209e-02 -3.77621293e-01
-6.33000314e-01 -4.48301584e-01 -8.41150165e-01 -4.67570037e-01
2.60926664e-01 -2.11199164e-01 -1.06574643e+00 1.49895179e+00
1.15808077e-01 1.39325812e-01 1.72804773e-01 6.18764758e-01
1.19786644e+00 3.91283214e-01 9.85407308e-02 -1.89145371e-01
1.46096051e+00 -5.24621665e-01 -8.36463451e-01 1.20046968e-03
7.65687644e-01 -7.16643929e-01 9.63331163e-01 3.62915188e-01
-9.27806914e-01 -8.18008959e-01 -7.94340789e-01 3.23061138e-01
-7.13384569e-01 1.57295555e-01 2.15148598e-01 7.41134226e-01
-1.09448564e+00 8.53918552e-01 -5.96586585e-01 -5.35142779e-01
-4.01109457e-02 5.17718017e-01 1.69996899e-02 2.59879053e-01
-1.44469631e+00 9.46914732e-01 3.33643079e-01 1.00129822e-04
-4.85442728e-01 -5.00521362e-01 -4.51424420e-01 -7.90555216e-03
-5.21538734e-01 -8.89965773e-01 1.21921909e+00 -4.25820768e-01
-9.15588200e-01 1.07411385e+00 -3.34574342e-01 -4.41560954e-01
-1.92414317e-02 -3.85006309e-01 -4.61013615e-01 1.53705701e-01
1.05452508e-01 4.00360107e-01 1.20153046e+00 -1.14476144e+00
-8.15456450e-01 -3.05198491e-01 -4.65702444e-01 -2.14225605e-01
-2.30386600e-01 -1.09730691e-01 5.24115086e-01 -2.77170449e-01
-1.17223039e-01 -1.05777168e+00 2.85084367e-01 -7.27136552e-01
-3.89007330e-01 -6.77571654e-01 1.02955914e+00 -6.59901679e-01
1.39810932e+00 -2.03369570e+00 -3.22721243e-01 -4.95587289e-02
2.53263324e-01 8.08073699e-01 3.11060641e-02 3.98631811e-01
-2.88178623e-01 3.59672964e-01 -4.45411086e-01 -2.25646660e-01
-1.20831743e-01 8.80327404e-01 -1.71246514e-01 2.03301117e-01
5.35097241e-01 8.28516006e-01 -9.47633266e-01 -4.41053897e-01
3.15233707e-01 6.56658232e-01 -3.44609708e-01 5.48787355e-01
5.29747494e-02 5.34622431e-01 -2.28872359e-01 4.58696038e-01
5.42597234e-01 -1.84793055e-01 -2.81410664e-01 -3.01774070e-02
-1.13281034e-01 5.30196905e-01 -4.83093441e-01 8.57472599e-01
-7.30972171e-01 4.28004503e-01 7.25779682e-02 -1.05374265e+00
7.21249104e-01 1.06788242e+00 3.09053510e-01 -2.59179085e-01
1.41282350e-01 8.12567353e-01 5.79975307e-01 -1.01920891e+00
-5.82812242e-02 -2.10084066e-01 3.74633282e-01 6.23701632e-01
6.98925788e-03 -3.21829528e-01 8.73593912e-02 -5.58428168e-01
1.06879330e+00 -2.81777292e-01 -1.65261105e-01 -1.60980746e-01
4.51255500e-01 -6.46237954e-02 4.85280417e-02 7.38835037e-01
-3.99929553e-01 7.11902797e-01 -4.82524782e-02 -4.16943789e-01
-7.28545010e-01 -1.06378448e+00 -6.55101240e-01 1.21259367e+00
-5.33282042e-01 4.06003654e-01 -8.35048378e-01 -5.53128302e-01
9.95308980e-02 7.03195870e-01 -5.79699218e-01 -1.69488221e-01
-8.87770534e-01 -9.62798417e-01 7.51952767e-01 4.53070998e-01
2.39565283e-01 -1.60439944e+00 -1.13548255e+00 1.44828945e-01
-4.07795310e-01 -6.87478006e-01 -1.20520376e-01 8.05389225e-01
-8.52988780e-01 -1.43215382e+00 -1.00292623e+00 -1.08202529e+00
2.30858058e-01 3.94774787e-02 1.04365087e+00 5.44095516e-01
-5.92765510e-01 5.27708590e-01 -3.45315099e-01 -7.89977551e-01
-9.34629798e-01 1.31259218e-01 1.24913074e-01 -4.36111510e-01
4.99429017e-01 -3.94756198e-01 -9.21910107e-01 -1.52540877e-01
-8.84289980e-01 -5.11241972e-01 5.32846928e-01 8.65397274e-01
3.73144537e-01 5.53312637e-02 9.74985063e-01 -5.65659583e-01
8.57304394e-01 -6.23792231e-01 3.74035209e-01 9.13196802e-03
-7.97968924e-01 -2.87149400e-01 5.91515243e-01 -6.36835933e-01
-4.18891221e-01 -4.33064818e-01 -6.46757901e-01 -3.44964057e-01
-6.85858607e-01 1.41542315e-01 8.60527396e-01 5.92731416e-01
3.99280429e-01 3.77655059e-01 -1.48748025e-01 -7.37391293e-01
-1.83218062e-01 1.08672237e+00 5.17465770e-01 -1.87649220e-01
4.31475550e-01 8.43474045e-02 -1.80130526e-01 -9.03232992e-01
-9.05830383e-01 -8.58346462e-01 -7.88798869e-01 5.25153540e-02
1.47852898e+00 -4.85892564e-01 -8.44754517e-01 5.56350768e-01
-1.20607841e+00 -1.71405464e-01 -3.77851784e-01 8.00942898e-01
-2.44080290e-01 2.00931937e-01 -7.00302839e-01 -1.12909532e+00
-1.14188969e+00 -9.80650544e-01 1.14869404e+00 8.43114704e-02
-7.20890462e-01 -1.25272989e+00 4.87722695e-01 5.62604666e-01
6.43516302e-01 2.49551281e-01 1.61621320e+00 -1.20739329e+00
3.90428215e-01 -4.06682789e-01 -3.22217308e-02 1.00356257e+00
5.16271830e-01 2.52876639e-01 -1.46947694e+00 -3.48202854e-01
4.89963740e-01 -3.96599561e-01 1.09927440e+00 4.62403059e-01
9.56292152e-01 -1.92585334e-01 -2.04475135e-01 -3.68321314e-02
1.13320947e+00 5.25870621e-01 7.23396018e-02 -1.96147889e-01
6.47594035e-01 7.36338079e-01 2.85628915e-01 -2.16360033e-01
1.70193300e-01 9.71207023e-02 5.71300209e-01 -2.82654881e-01
-1.91026688e-01 1.01742819e-01 1.71169981e-01 1.31905448e+00
-1.93916634e-01 -1.50739819e-01 -9.54077363e-01 9.40970182e-01
-1.22542560e+00 -8.52525115e-01 -4.79635298e-01 2.32432771e+00
8.58990014e-01 -5.93730360e-02 1.55338347e-01 6.92906022e-01
7.89872944e-01 -1.30798807e-02 -2.75952220e-01 -1.24119711e+00
2.66550809e-01 1.03360057e+00 1.76695555e-01 1.54712379e-01
-1.10486603e+00 1.24923252e-01 6.66040897e+00 5.29997826e-01
-1.48795235e+00 6.87168017e-02 2.57860690e-01 1.80514127e-01
-1.77824907e-02 -5.13595521e-01 -4.48869675e-01 6.57360077e-01
1.33278930e+00 7.10082531e-01 4.05317634e-01 3.93006027e-01
1.78574160e-01 -2.04865932e-01 -1.05168831e+00 5.55564880e-01
-3.39213699e-01 -6.81190610e-01 -2.78931737e-01 -2.63944473e-02
3.75896454e-01 4.89138603e-01 5.17714024e-02 4.24389869e-01
-2.41308764e-01 -1.21533644e+00 1.51635893e-02 4.26363379e-01
8.90801966e-01 -3.96456391e-01 1.06584132e+00 7.50272155e-01
-1.06412351e+00 -1.72393039e-01 -1.09071806e-01 -1.79772396e-02
-4.60883915e-01 5.20442009e-01 -1.24151230e+00 1.71075836e-01
1.07259369e+00 4.16421115e-01 -2.69017041e-01 5.70855439e-01
2.26780713e-01 1.12335169e+00 -3.11646938e-01 -2.06359908e-01
1.98532388e-01 1.84904560e-02 3.72611225e-01 1.42222738e+00
2.99204707e-01 -2.80867368e-01 -2.32064053e-01 1.00722384e+00
3.35782081e-01 -1.68838650e-01 -7.32407212e-01 -2.85674363e-01
2.32616946e-01 1.02122641e+00 -3.87233377e-01 -4.32738364e-01
-1.93577975e-01 7.52587676e-01 -2.30802149e-01 1.70987427e-01
-6.08980656e-01 -2.98006862e-01 4.95586962e-01 1.62967313e-02
2.55814940e-01 5.86854339e-01 -1.15953021e-01 -5.98177612e-01
-2.04583392e-01 -6.01563156e-01 1.02176525e-01 -4.79375184e-01
-1.24060202e+00 4.46509480e-01 -3.85059714e-02 -9.53058004e-01
-5.25123835e-01 -5.96384585e-01 -1.19183648e+00 1.04719186e+00
-1.53573823e+00 -8.55696559e-01 -9.43852514e-02 3.26626241e-01
3.58752370e-01 4.13135052e-01 1.41010499e+00 4.72041011e-01
-4.07980889e-01 2.46582121e-01 3.04632932e-02 1.68408453e-01
6.82863116e-01 -1.64974582e+00 8.18795189e-02 -4.11004573e-03
-2.97622114e-01 7.50464737e-01 4.93046790e-01 -5.58556676e-01
-8.15958083e-01 -1.29152012e+00 1.66761243e+00 -4.17619467e-01
3.08255225e-01 1.96508229e-01 -1.21243787e+00 -3.74023244e-02
2.11762026e-01 -2.15192124e-01 9.29904163e-01 -2.24697724e-01
-9.15252566e-02 2.28967845e-01 -1.33523738e+00 -1.24660358e-01
4.91127998e-01 -7.38024175e-01 -9.60568428e-01 4.86346483e-01
9.23890710e-01 -7.14280382e-02 -1.23252726e+00 8.60464752e-01
7.58427739e-01 -1.14365864e+00 1.02343404e+00 -4.44759607e-01
4.15002227e-01 1.40014723e-01 9.49133858e-02 -1.16403854e+00
-4.76536416e-02 3.78316529e-02 -1.93711206e-01 7.34695494e-01
4.84246239e-02 -1.07668579e+00 4.24651325e-01 1.69422310e-02
-1.77802160e-01 -1.19175303e+00 -9.31083858e-01 -5.64564884e-01
5.26077628e-01 -3.68107706e-01 5.24459004e-01 7.95300663e-01
-3.74277562e-01 2.99093693e-01 1.31982476e-01 -2.11112231e-01
7.84315616e-02 3.89517456e-01 1.89913243e-01 -1.65848207e+00
-2.38024816e-01 -4.98202145e-01 2.56175160e-01 -3.44094068e-01
-1.40178412e-01 -8.56803179e-01 1.26174726e-02 -1.95934105e+00
5.22606708e-02 -5.13070524e-01 -6.00571215e-01 4.36437577e-01
-3.06124538e-01 1.73674092e-01 -1.15684206e-02 3.24676871e-01
2.47172654e-01 -1.94299996e-01 1.52989399e+00 -1.86019130e-02
-1.28033817e-01 8.63254845e-01 -4.47406918e-02 7.57020116e-01
1.06896961e+00 -4.47745204e-01 -2.05201253e-01 -1.39838547e-01
4.02257480e-02 1.89389288e-01 6.61339939e-01 -9.95016575e-01
-4.96237069e-01 1.21171542e-01 3.23763072e-01 -1.08506095e+00
4.45077866e-01 -9.51940238e-01 -5.47994040e-02 1.47361255e+00
-3.01649898e-01 -1.54918879e-01 -1.03433384e-03 3.39631408e-01
-2.32091203e-01 -4.49404627e-01 7.37171113e-01 -2.05045030e-01
9.72885936e-02 -1.27901495e-01 -7.14060545e-01 4.82812524e-02
4.52725589e-01 -3.47588688e-01 -4.01262730e-01 -4.93367575e-03
-9.87496614e-01 -2.08872720e-03 -3.29685323e-02 3.72228622e-01
5.22004426e-01 -8.06196332e-01 -5.77109277e-01 5.23936987e-01
-2.51333490e-02 9.97502580e-02 6.82184696e-02 1.20865548e+00
-3.80720615e-01 9.86561954e-01 3.03338021e-01 -9.69320834e-01
-1.39846659e+00 3.64559531e-01 4.29462194e-01 -5.39465725e-01
-2.83652216e-01 8.02699745e-01 2.39437316e-02 -7.34018624e-01
3.38918358e-01 -1.10650063e+00 -5.93563378e-01 3.23950171e-01
2.60857996e-02 5.69392085e-01 2.80578613e-01 -5.33602297e-01
-3.23590368e-01 8.28230023e-01 3.74657989e-01 6.74868107e-01
9.20473278e-01 1.22587942e-01 -3.11215281e-01 8.36064041e-01
1.48361480e+00 -2.31093094e-01 -1.21573493e-01 1.06849514e-01
-1.79977626e-01 2.43448764e-01 -2.74984181e-01 -8.85313213e-01
-6.95416033e-01 1.33116388e+00 1.32902288e+00 9.49933171e-01
1.07105088e+00 -7.62510374e-02 1.38379598e+00 4.19465601e-01
-1.13421075e-01 -9.89424288e-01 -1.31972209e-01 6.10838592e-01
5.54637492e-01 -1.32623863e+00 -4.89114136e-01 8.72839689e-02
-2.75830835e-01 1.10308039e+00 1.83055192e-01 -4.47219849e-01
8.63379240e-01 -1.56350642e-01 2.30545998e-01 -3.56562644e-01
-7.91714549e-01 -4.01029289e-01 4.93701458e-01 7.35531807e-01
7.95327425e-01 4.73024905e-01 -4.77549493e-01 4.25376147e-01
-3.78287613e-01 -1.69042647e-01 4.13189530e-02 7.20496655e-01
-7.34434247e-01 -9.66930032e-01 -5.39638519e-01 1.14577734e+00
-8.85378540e-01 -1.76069543e-01 -5.36295831e-01 7.41291165e-01
8.66812289e-01 1.22149396e+00 3.87290299e-01 -4.53736871e-01
6.62738532e-02 5.58152437e-01 3.36918473e-01 -9.77604866e-01
-1.33223271e+00 7.41443783e-02 -5.26521988e-02 3.38082276e-02
-7.72561491e-01 -1.62483215e-01 -1.57784927e+00 -6.46684319e-02
-5.01206815e-01 1.99074298e-01 6.35532200e-01 1.03052068e+00
-3.05122465e-01 6.76788807e-01 6.84794247e-01 -5.89088440e-01
-6.07661128e-01 -1.19251072e+00 -4.67241913e-01 4.17381883e-01
1.21297002e+00 -3.18842977e-01 -7.34716773e-01 -6.78418204e-02]
|
[14.515698432922363, 3.841136932373047]
|
f95ce77b-0d88-4fc7-80f9-00c804718296
|
multi-modal-multi-kernel-graph-learning-for
|
2303.03388
| null |
https://arxiv.org/abs/2303.03388v2
|
https://arxiv.org/pdf/2303.03388v2.pdf
|
Multi-modal Multi-kernel Graph Learning for Autism Prediction and Biomarker Discovery
|
Due to its complexity, graph learning-based multi-modal integration and classification is one of the most challenging obstacles for disease prediction. To effectively offset the negative impact between modalities in the process of multi-modal integration and extract heterogeneous information from graphs, we propose a novel method called MMKGL (Multi-modal Multi-Kernel Graph Learning). For the problem of negative impact between modalities, we propose a multi-modal graph embedding module to construct a multi-modal graph. Different from conventional methods that manually construct static graphs for all modalities, each modality generates a separate graph by adaptive learning, where a function graph and a supervision graph are introduced for optimization during the multi-graph fusion embedding process. We then propose a multi-kernel graph learning module to extract heterogeneous information from the multi-modal graph. The information in the multi-modal graph at different levels is aggregated by convolutional kernels with different receptive field sizes, followed by generating a cross-kernel discovery tensor for disease prediction. Our method is evaluated on the benchmark Autism Brain Imaging Data Exchange (ABIDE) dataset and outperforms the state-of-the-art methods. In addition, discriminative brain regions associated with autism are identified by our model, providing guidance for the study of autism pathology.
|
['Shirui Pan', 'Yi Pan', 'Hulin Kuang', 'Hanhe Lin', 'Jin Liu', 'Junbin Mao']
|
2023-03-03
| null | null | null | null |
['disease-prediction']
|
['medical']
|
[ 1.01626813e-01 2.55623549e-01 6.18762597e-02 -1.24222428e-01
-4.76041108e-01 -1.13907486e-01 3.19486946e-01 6.62796378e-01
-8.25638846e-02 1.95732996e-01 4.06708449e-01 2.16080979e-01
-4.65311140e-01 -7.57278144e-01 -3.63604009e-01 -7.85721064e-01
-2.33069122e-01 5.37045181e-01 2.75220633e-01 -2.27878526e-01
-1.47455782e-01 3.83184075e-01 -1.07812285e+00 6.71949923e-01
1.05076003e+00 7.94090331e-01 1.82500705e-01 3.72154415e-01
-1.34410173e-01 8.41885746e-01 -9.42224935e-02 -4.66814220e-01
-1.12684689e-01 -3.31447929e-01 -7.94185042e-01 2.83611834e-01
3.06307137e-01 -2.95845360e-01 -3.74017209e-01 1.27712715e+00
6.58668697e-01 -2.04275683e-01 7.81467557e-01 -1.58002722e+00
-8.44167113e-01 9.70297754e-01 -7.83562660e-01 1.26995081e-02
4.00029957e-01 9.63565558e-02 1.09790003e+00 -6.76538408e-01
5.44414759e-01 1.39693689e+00 5.80818117e-01 4.48956609e-01
-1.39517188e+00 -4.37347770e-01 3.50467175e-01 4.56608653e-01
-1.13401151e+00 -3.49718779e-02 1.05480170e+00 -9.81906712e-01
9.60086048e-01 1.05530797e-02 1.16087604e+00 8.93427730e-01
4.53227818e-01 6.54140890e-01 8.80291343e-01 -1.44107327e-01
-2.24886045e-01 -5.08966267e-01 2.19236657e-01 1.39942324e+00
1.00852780e-01 -1.42177984e-01 -4.64030087e-01 -5.81174731e-01
3.99877131e-01 4.14678693e-01 -3.56969655e-01 -6.10916674e-01
-1.44714558e+00 7.67085791e-01 7.65511572e-01 4.74329114e-01
-4.71708000e-01 2.57283095e-02 5.01266718e-01 1.19808875e-01
5.71198344e-01 2.07914747e-02 -1.53171718e-01 6.33156300e-01
-5.49291432e-01 -1.31748104e-03 3.60979140e-01 4.47382063e-01
9.33472753e-01 -1.49798110e-01 -2.77554333e-01 1.09260619e+00
7.91445851e-01 1.80254951e-01 5.95840931e-01 -3.60269964e-01
6.56408310e-01 1.47476494e+00 -7.47677267e-01 -1.21556938e+00
-9.57551539e-01 -1.34512410e-01 -1.23293090e+00 6.23777248e-02
1.16689876e-01 -5.47483284e-03 -8.17632794e-01 1.68988860e+00
6.73023641e-01 4.01566356e-01 -1.16201580e-01 5.92777371e-01
1.07734478e+00 4.85007614e-02 2.36206234e-01 -1.38867572e-01
1.55863965e+00 -1.10175693e+00 -5.56643903e-01 -1.63369700e-01
1.05794477e+00 -2.68331796e-01 5.40824354e-01 3.52021717e-02
-7.12138355e-01 -4.94843982e-02 -7.78530896e-01 1.88984439e-01
-5.24099469e-01 1.40725458e-02 6.47173166e-01 1.02740349e-02
-1.12245917e+00 2.07797706e-01 -8.36927712e-01 -1.51290655e-01
4.76328641e-01 3.76593947e-01 -9.71721709e-01 -2.63695359e-01
-1.13702226e+00 8.33305717e-01 7.03162611e-01 -2.61695117e-01
-6.72324181e-01 -8.91800284e-01 -1.11463213e+00 -8.58524293e-02
1.67964935e-01 -9.09328759e-01 2.97457337e-01 -6.98135257e-01
-8.88949513e-01 7.09772944e-01 3.44998181e-01 -8.29211175e-02
3.38921733e-02 2.58563727e-01 -5.88307798e-01 6.00291789e-01
1.55189320e-01 5.14446318e-01 1.19007158e+00 -1.08753276e+00
-4.24211472e-01 -9.46046770e-01 1.16802625e-01 2.64317691e-01
-6.28930509e-01 -3.96609344e-02 -2.62064785e-01 -7.55244970e-01
2.59949178e-01 -9.49067831e-01 -1.62607357e-01 -3.00692409e-01
-5.54734468e-01 -4.21868354e-01 8.92959833e-01 -9.67921615e-01
1.33595860e+00 -2.10202384e+00 8.73681128e-01 5.05197532e-02
1.21989155e+00 -9.11215395e-02 -4.58048463e-01 2.52400219e-01
-4.47581291e-01 -9.29948762e-02 -3.15396637e-01 -4.30549353e-01
-4.54889908e-02 -1.61283299e-01 4.98196036e-01 5.66243649e-01
2.99638599e-01 1.03662670e+00 -1.15606010e+00 -7.57889628e-01
3.78948569e-01 6.35067344e-01 -6.33939326e-01 2.92223573e-01
1.42747322e-02 5.82254708e-01 -4.36321974e-01 8.90198767e-01
5.27546406e-01 -7.09788442e-01 1.77858189e-01 -8.47270072e-01
5.15910685e-01 -4.30195987e-01 -8.32800925e-01 1.72529995e+00
-1.27959147e-01 -9.11635309e-02 9.19502228e-02 -1.37161326e+00
4.42807019e-01 3.98738742e-01 1.08736682e+00 -2.77802795e-01
3.28035772e-01 1.43016949e-02 9.89818871e-02 -6.08315289e-01
-1.67302325e-01 5.69603667e-02 1.00307636e-01 3.51216495e-01
6.42710745e-01 1.36659473e-01 1.50049746e-01 3.96504641e-01
1.57241273e+00 -3.43311727e-01 2.78772324e-01 7.12351128e-02
7.07668960e-01 -4.80190128e-01 4.45357502e-01 -2.80528533e-04
-2.91907430e-01 2.26002932e-01 7.16012239e-01 -3.13996047e-01
-6.21852040e-01 -1.03219724e+00 1.02060497e-01 1.13086998e+00
8.84069223e-03 -5.63231587e-01 -7.40452111e-01 -1.13490188e+00
2.37432882e-01 -5.44603989e-02 -8.18725586e-01 -4.88563180e-01
-1.20175220e-01 -1.07138085e+00 4.19292659e-01 2.88348615e-01
3.18611652e-01 -8.27024281e-01 4.05409187e-02 7.72713348e-02
-1.62196919e-01 -1.14542270e+00 -5.48789918e-01 5.31261601e-02
-7.32314646e-01 -1.57775950e+00 -6.49549723e-01 -7.45135427e-01
9.64139640e-01 8.15645456e-02 8.25871825e-01 2.03553125e-01
-3.93584490e-01 8.41925561e-01 -6.04064405e-01 2.07410119e-02
-4.89409119e-01 -7.16365203e-02 -3.71600948e-02 5.43463707e-01
1.64612755e-02 -6.78472042e-01 -5.25148690e-01 -8.41533244e-02
-1.11826646e+00 3.47851247e-01 5.43684065e-01 1.13950729e+00
6.94874167e-01 1.45373112e-02 7.39538133e-01 -6.43223822e-01
7.43225217e-01 -7.94857681e-01 -4.96660411e-01 6.01361513e-01
-5.22650003e-01 8.49928111e-02 4.81937885e-01 -6.21401846e-01
-6.06714547e-01 2.26815000e-01 1.47915646e-01 -7.22429752e-01
6.16209283e-02 8.04659784e-01 -2.51586825e-01 -4.12009776e-01
3.78753603e-01 -3.27116624e-02 1.40442640e-01 -2.60044426e-01
5.79269826e-01 2.76145816e-01 5.92637900e-03 -9.43562016e-02
5.16722620e-01 3.14191341e-01 1.97632566e-01 -7.13665903e-01
-6.48187041e-01 -4.13890809e-01 -7.23396599e-01 -4.81572926e-01
1.39407420e+00 -8.67521703e-01 -5.11995435e-01 8.77607286e-01
-1.05881691e+00 -1.69421911e-01 1.03986599e-01 6.20999515e-01
-3.55015576e-01 6.27393067e-01 -6.47523105e-01 -3.33777338e-01
-4.89612281e-01 -1.44920290e+00 1.25870693e+00 -5.64196073e-02
2.19823912e-01 -1.34131491e+00 3.40353966e-01 7.87172854e-01
-1.06990106e-01 5.53553581e-01 1.29979253e+00 -7.13593483e-01
-2.38983244e-01 -9.23778564e-02 -3.79700214e-01 3.65507364e-01
4.12418664e-01 -3.56449097e-01 -5.81077576e-01 -5.18819213e-01
-4.24635679e-01 -3.41407984e-01 1.00189447e+00 4.87035066e-02
1.02377474e+00 -2.19845220e-01 -5.11130333e-01 6.22155964e-01
1.37739587e+00 -1.70605719e-01 9.27047953e-02 -5.25417812e-02
1.57248211e+00 6.24134183e-01 1.62907634e-02 4.53216851e-01
8.63359630e-01 5.27624309e-01 7.24907339e-01 -2.21809205e-02
-1.50214210e-01 1.17363207e-01 4.00306940e-01 1.19829488e+00
-5.42374477e-02 -9.61074084e-02 -1.12108994e+00 5.97096562e-01
-2.18622041e+00 -7.29570806e-01 -6.72858059e-02 1.76983678e+00
5.96819818e-01 -3.02333295e-01 2.48412147e-01 -1.40307397e-01
9.42303836e-01 2.69518554e-01 -5.91051161e-01 1.71478614e-01
-2.68224299e-01 -1.35271519e-01 6.26377687e-02 2.65456736e-01
-1.16027439e+00 7.80419052e-01 5.32852077e+00 6.80205822e-01
-9.89156425e-01 4.60588902e-01 2.02613533e-01 1.59797341e-01
-3.38988096e-01 -2.53444403e-01 -3.67308825e-01 4.51598704e-01
6.40559435e-01 1.48001000e-01 7.68446147e-01 4.70005304e-01
-3.26764196e-01 1.91393554e-01 -9.50681090e-01 1.21322405e+00
2.54909605e-01 -1.18437088e+00 3.25249970e-01 1.13568291e-01
5.95598936e-01 5.02949893e-01 -1.82652339e-01 5.14621809e-02
3.67381155e-01 -7.76244879e-01 3.42880309e-01 7.83906102e-01
5.62216818e-01 -6.44023418e-01 4.97529000e-01 9.58825052e-02
-1.68837845e+00 -2.12261349e-01 -9.57186148e-02 6.60565972e-01
-7.49989748e-02 8.88271153e-01 -7.64623106e-01 9.83923733e-01
5.48935533e-01 1.00813079e+00 -9.22233403e-01 8.94933164e-01
-5.51461093e-02 1.86007351e-01 7.30588511e-02 3.95058602e-01
-3.40419188e-02 -2.48971805e-01 6.71414971e-01 1.06207788e+00
1.76940545e-01 -1.42755121e-01 4.61832017e-01 8.87255251e-01
-2.22311631e-01 5.01364410e-01 -8.62060666e-01 -5.40567040e-01
3.28306742e-02 1.62024832e+00 -4.90178168e-01 -2.31722265e-01
-8.77242863e-01 9.43592966e-01 9.39822495e-01 2.13673621e-01
-7.33741701e-01 -1.35273576e-01 3.56833637e-01 -2.39800528e-01
-9.66549814e-02 -7.68942386e-02 7.08148703e-02 -1.47881508e+00
-1.48474976e-01 -7.19004154e-01 7.96048224e-01 -6.46568239e-01
-1.78227746e+00 6.26247883e-01 -3.23317870e-02 -1.11973274e+00
6.76534697e-02 -6.18281960e-01 -3.25443923e-01 7.17430294e-01
-1.24824655e+00 -1.93145001e+00 -3.47928971e-01 1.10436380e+00
-9.86255929e-02 -5.88958859e-01 8.78362000e-01 5.00986040e-01
-6.32295072e-01 4.71086830e-01 -2.48403937e-01 1.90922797e-01
4.80080247e-01 -1.26189625e+00 -3.03007156e-01 5.82486033e-01
-8.52297619e-02 2.19643757e-01 -1.35251582e-01 -9.98477876e-01
-1.43219674e+00 -1.43042684e+00 3.17709833e-01 -1.89221710e-01
1.19835103e+00 -1.15293644e-01 -1.10013270e+00 6.49753809e-01
6.22235872e-02 4.17602271e-01 9.62310076e-01 -3.20471474e-03
-5.47531009e-01 1.10894917e-02 -1.10932970e+00 4.69132274e-01
1.04231000e+00 -7.92973638e-01 -6.50186658e-01 5.81858933e-01
1.02392399e+00 -1.62735865e-01 -1.40451908e+00 7.45235920e-01
2.83025384e-01 -7.40221679e-01 9.19844031e-01 -7.36757815e-01
3.15610170e-01 -1.96271271e-01 -1.27413377e-01 -1.74442434e+00
-6.38767540e-01 -1.82542145e-01 -4.76072401e-01 1.01267934e+00
1.61067382e-01 -8.20449889e-01 1.89947292e-01 2.09585667e-01
-2.85992980e-01 -9.28646982e-01 -9.78224099e-01 -3.39983970e-01
-1.84616283e-01 -2.41042584e-01 7.36975908e-01 1.31600738e+00
2.96661675e-01 4.29994494e-01 -1.99286401e-01 4.45483029e-01
8.44395518e-01 -2.58670896e-02 2.68751234e-01 -1.33689439e+00
-2.45157227e-01 -6.38302863e-01 -9.64074254e-01 4.07361835e-02
5.24847031e-01 -1.51769269e+00 -5.30620813e-01 -1.83677363e+00
5.16708255e-01 -1.33937463e-01 -7.21816182e-01 8.82819355e-01
-1.98243588e-01 8.74009579e-02 4.45592813e-02 -8.54832083e-02
-7.34989882e-01 6.45365775e-01 1.59898007e+00 -7.08691597e-01
-8.07110146e-02 -6.03672147e-01 -5.00952601e-01 7.64678657e-01
5.67005873e-01 -1.26201168e-01 -5.94355643e-01 -2.81766981e-01
2.35843524e-01 1.01434022e-01 6.12282157e-01 -9.22337472e-01
1.87843159e-01 -1.61800116e-01 1.46649957e-01 -4.90581185e-01
3.14735800e-01 -9.08759773e-01 2.38438636e-01 4.25344735e-01
-7.61769861e-02 -1.56789401e-03 1.10638821e-02 7.01983154e-01
-5.74805401e-02 1.16840638e-01 8.94209921e-01 1.74752425e-03
-4.87922728e-01 9.63194430e-01 1.40863573e-02 1.92593765e-02
1.24305522e+00 2.17075154e-01 -6.78298771e-01 2.79881239e-01
-1.34290302e+00 3.95017385e-01 2.52091795e-01 5.21429539e-01
9.58086431e-01 -1.69321823e+00 -6.80938840e-01 2.58478224e-01
6.25422478e-01 -1.63099632e-01 6.68612659e-01 1.44731569e+00
-7.76906088e-02 -7.29480758e-02 -4.47289318e-01 -6.60354495e-01
-1.43213522e+00 6.65374458e-01 5.10480523e-01 -7.09957778e-01
-6.56080961e-01 6.97517395e-01 4.91034657e-01 -5.79938829e-01
-3.91496643e-02 -1.77872539e-01 -5.87690353e-01 3.84296775e-01
4.81755227e-01 4.00859565e-01 2.40193471e-01 -9.51909304e-01
-4.18696970e-01 4.85115737e-01 -3.60144228e-02 2.37943158e-01
1.57047594e+00 -1.43597890e-02 -9.55981910e-01 4.09600109e-01
1.25107169e+00 -2.63001651e-01 -6.44967973e-01 -2.58983642e-01
-2.29485482e-01 -5.16354814e-02 1.59623742e-01 -7.21770167e-01
-1.53230906e+00 6.33439183e-01 7.87171781e-01 4.55572307e-01
1.30099559e+00 3.54379803e-01 7.09602535e-01 1.46373406e-01
3.14194411e-01 -8.39965940e-01 4.86526310e-01 4.57546204e-01
1.24865139e+00 -1.25040174e+00 -1.27972350e-01 -3.84953797e-01
-4.69319820e-01 1.11830223e+00 8.39842677e-01 -3.78977396e-02
1.08028698e+00 5.32129221e-02 -3.16426575e-01 -8.04723680e-01
-4.53975677e-01 -2.88181692e-01 7.21071720e-01 8.19341063e-01
1.43218294e-01 3.84564698e-01 -1.43621579e-01 7.24233568e-01
2.72249281e-01 -3.06481093e-01 -5.88073097e-02 6.50567651e-01
-2.80602545e-01 -1.18623793e+00 -2.33739197e-01 9.99143124e-01
-1.40040010e-01 -1.64092451e-01 -5.59913278e-01 2.76237309e-01
4.04715627e-01 7.55797803e-01 -4.22554433e-01 -9.67057645e-01
8.41965452e-02 9.94592234e-02 5.99233985e-01 -7.35386252e-01
-6.09328747e-01 7.85384253e-02 -7.79647529e-02 -7.61719644e-01
-5.33190906e-01 -3.98648858e-01 -1.44251955e+00 1.02953628e-01
-4.78166312e-01 -3.42477918e-01 2.81618357e-01 1.09895122e+00
6.42980039e-01 9.73027825e-01 4.10941839e-01 -7.77394235e-01
-1.61273573e-02 -8.73465359e-01 -8.99267197e-01 5.98484516e-01
3.88385683e-01 -1.10993898e+00 -3.76736760e-01 -1.55330911e-01]
|
[12.350234985351562, 3.3848330974578857]
|
774892c9-04db-4b4c-a06c-8cf278c67cb4
|
variation-of-gender-biases-in-visual
|
2303.07615
| null |
https://arxiv.org/abs/2303.07615v1
|
https://arxiv.org/pdf/2303.07615v1.pdf
|
Variation of Gender Biases in Visual Recognition Models Before and After Finetuning
|
We introduce a framework to measure how biases change before and after fine-tuning a large scale visual recognition model for a downstream task. Deep learning models trained on increasing amounts of data are known to encode societal biases. Many computer vision systems today rely on models typically pretrained on large scale datasets. While bias mitigation techniques have been developed for tuning models for downstream tasks, it is currently unclear what are the effects of biases already encoded in a pretrained model. Our framework incorporates sets of canonical images representing individual and pairs of concepts to highlight changes in biases for an array of off-the-shelf pretrained models across model sizes, dataset sizes, and training objectives. Through our analyses, we find that (1) supervised models trained on datasets such as ImageNet-21k are more likely to retain their pretraining biases regardless of the target dataset compared to self-supervised models. We also find that (2) models finetuned on larger scale datasets are more likely to introduce new biased associations. Our results also suggest that (3) biases can transfer to finetuned models and the finetuning objective and dataset can impact the extent of transferred biases.
|
['Vicente Ordonez', 'Baishakhi Ray', 'Tianlu Wang', 'Jaspreet Ranjit']
|
2023-03-14
| null | null | null | null |
['object-recognition']
|
['computer-vision']
|
[ 2.95228601e-01 5.15710190e-02 -2.20437795e-01 -8.66955638e-01
-2.99914856e-03 -7.56825268e-01 9.01684046e-01 1.94516063e-01
-8.88870239e-01 5.54625809e-01 6.24355197e-01 -1.08967155e-01
7.16722235e-02 -8.47959995e-01 -1.21352565e+00 -2.55184919e-01
4.67835516e-02 3.27701509e-01 1.30626485e-01 -3.50685984e-01
4.74293619e-01 4.67451453e-01 -1.55259991e+00 4.48457778e-01
6.93980753e-01 6.84579730e-01 -7.06904233e-02 4.81293976e-01
1.35024255e-02 4.78711843e-01 -6.39142215e-01 -4.32963192e-01
6.79320276e-01 -1.74751699e-01 -6.65678263e-01 -3.75896066e-01
1.32977641e+00 -3.03738087e-01 -2.57479221e-01 1.08153057e+00
4.19830620e-01 -4.40379381e-02 7.33852923e-01 -1.29300511e+00
-1.25446773e+00 8.31719100e-01 -3.09863389e-01 6.56100571e-01
-4.54157740e-01 7.48233378e-01 8.61867130e-01 -7.24504411e-01
7.28161633e-01 1.60981929e+00 9.27880764e-01 6.23595417e-01
-1.75262785e+00 -1.24131441e+00 5.66662610e-01 -2.13599995e-01
-9.51248944e-01 -5.63176274e-01 4.82682437e-01 -7.56805480e-01
1.04303920e+00 5.93339168e-02 7.58272767e-01 1.38310111e+00
2.86678970e-01 1.09838478e-01 1.68521571e+00 -2.37375155e-01
9.38151330e-02 5.76779664e-01 5.20269632e-01 3.80143940e-01
7.34180212e-01 7.62527466e-01 -7.95623958e-01 2.90124677e-02
7.65271544e-01 -3.73565070e-02 1.07166963e-02 -5.16855896e-01
-1.43924677e+00 9.15125132e-01 1.08286786e+00 1.69171005e-01
-1.34037957e-01 3.72515917e-01 3.23365331e-01 4.82265264e-01
3.68486613e-01 1.26778579e+00 -7.60966063e-01 2.47395977e-01
-8.09731722e-01 2.42445588e-01 4.75823849e-01 8.59768450e-01
1.25938761e+00 -5.49318444e-04 -4.05044049e-01 8.03240240e-01
1.42828450e-01 5.92967212e-01 6.67004704e-01 -7.32950211e-01
3.47625196e-01 7.92917371e-01 -5.25737479e-02 -9.56637442e-01
-2.78618097e-01 -6.03669941e-01 -4.87264365e-01 2.83692479e-01
4.51755404e-01 -2.47579411e-01 -1.28783131e+00 2.12734652e+00
-1.00585097e-03 -5.15197158e-01 -2.29979798e-01 7.90268302e-01
7.07558453e-01 3.20409507e-01 5.88167250e-01 4.31051135e-01
9.73563969e-01 -8.35125864e-01 -2.49928698e-01 -9.53269720e-01
6.96234226e-01 -4.86731708e-01 1.45715809e+00 -3.09555046e-02
-7.40429163e-01 -8.36564481e-01 -1.20146489e+00 -1.86717361e-01
-8.59173596e-01 -2.42210910e-01 5.31366706e-01 8.35112453e-01
-1.30191255e+00 6.06934965e-01 -1.72614560e-01 -7.08253264e-01
9.09221411e-01 4.46828663e-01 -2.16749400e-01 -1.86093956e-01
-1.17742944e+00 1.35652757e+00 3.00772488e-01 -2.70325154e-01
-1.40108967e+00 -1.25040567e+00 -6.40372157e-01 2.04156656e-02
-2.79803693e-01 -6.97463334e-01 1.01749694e+00 -1.70610416e+00
-8.34689558e-01 1.20429444e+00 2.84070615e-02 -3.03402871e-01
3.13018054e-01 -1.00895174e-01 -3.55779380e-01 -5.51915467e-01
5.20254672e-02 1.43917596e+00 9.87956941e-01 -1.41521633e+00
-5.47838151e-01 -5.35619736e-01 2.70071119e-01 1.99251771e-01
-7.62657940e-01 -3.58046256e-02 2.47850150e-01 -5.83318293e-01
-3.46089721e-01 -1.05169940e+00 -1.68198586e-01 1.87734410e-01
-7.20680282e-02 1.72517464e-01 4.14910555e-01 -3.48644584e-01
9.26790714e-01 -1.97823048e+00 -3.65578309e-02 2.82875985e-01
1.92530885e-01 1.91777200e-01 -6.67753339e-01 -2.19881963e-02
-2.98693627e-01 5.59199512e-01 4.73391525e-02 2.52071738e-01
1.95195489e-02 5.97854964e-02 -3.12790871e-01 3.05183738e-01
3.61687094e-01 9.64662910e-01 -7.08386064e-01 -1.83839306e-01
4.50829752e-02 1.74531490e-01 -9.77468491e-01 7.00902268e-02
-1.21062333e-02 2.09226146e-01 1.24261826e-01 3.24348450e-01
6.33077860e-01 -1.41869068e-01 8.58946145e-02 -3.96460444e-01
-1.08923353e-01 3.67864490e-01 -7.35264301e-01 1.25312042e+00
-3.74024630e-01 8.34344208e-01 -1.38231978e-01 -6.28631055e-01
1.10414875e+00 -4.25418794e-01 -3.69977206e-01 -1.12040186e+00
-1.74281560e-02 1.09335534e-01 6.90136790e-01 -9.39488113e-02
5.76939046e-01 -4.26979512e-01 5.03800400e-02 3.98807853e-01
3.08322132e-01 -1.44866154e-01 8.76158997e-02 1.15902446e-01
9.86086369e-01 -2.78756380e-01 8.32278654e-02 -7.85965860e-01
-6.70850053e-02 3.97646934e-01 4.49597627e-01 9.01400208e-01
-4.79968846e-01 2.72145063e-01 3.70059937e-01 -5.63811839e-01
-1.23597693e+00 -8.35834324e-01 -3.24040771e-01 1.80737674e+00
-1.19412735e-01 -1.72712915e-02 -4.60422158e-01 -7.46961117e-01
6.69547141e-01 7.06563473e-01 -1.26569867e+00 -7.22881794e-01
-2.28391781e-01 -8.03500235e-01 6.17170751e-01 8.54490638e-01
4.55997348e-01 -1.02233362e+00 -4.45765316e-01 -3.22050542e-01
4.27955180e-01 -5.34046948e-01 -4.35385287e-01 4.69287872e-01
-1.34271872e+00 -9.60191846e-01 -4.83453095e-01 -6.25051975e-01
1.16043711e+00 1.81588486e-01 1.58101499e+00 2.03319356e-01
-2.85100974e-02 4.12183344e-01 8.47701207e-02 -8.61041903e-01
-4.67383623e-01 2.00598717e-01 7.73996487e-02 -3.59679639e-01
5.80254257e-01 -2.61029929e-01 -6.70691967e-01 6.69736028e-01
-7.65494168e-01 -2.04755530e-01 8.96905899e-01 7.80774474e-01
2.89742351e-01 -5.48189759e-01 6.81162238e-01 -1.32677150e+00
8.68563354e-01 -4.39632595e-01 -4.64272201e-01 2.21634269e-01
-1.22161055e+00 3.19424003e-01 2.23909765e-01 -6.59361601e-01
-1.14113796e+00 -4.02181238e-01 4.21962678e-01 -2.50725061e-01
-3.37439418e-01 2.34561950e-01 -2.47239638e-02 -1.87826440e-01
1.36993873e+00 -3.17351550e-01 -2.26612747e-01 -2.01245576e-01
6.18183851e-01 2.56065339e-01 8.29733387e-02 -7.15313673e-01
7.85693586e-01 4.06160027e-01 -3.96101415e-01 -3.39772642e-01
-8.55500042e-01 3.07706259e-02 -7.87342846e-01 -1.16673604e-01
7.52430499e-01 -1.17853308e+00 -2.08180562e-01 3.93123239e-01
-5.89911222e-01 -9.60843265e-01 -3.37396592e-01 2.20780641e-01
1.52230427e-01 -4.52668697e-01 -2.48183444e-01 -1.20848462e-01
-1.52029827e-01 -9.87922549e-01 5.94800949e-01 4.26233321e-01
-6.89203262e-01 -1.14831412e+00 3.41580302e-01 2.32894376e-01
8.80482614e-01 -3.07830870e-01 1.28222966e+00 -6.98426008e-01
-2.68481553e-01 1.84031293e-01 -5.80629408e-01 4.79151368e-01
2.30080858e-01 7.44510395e-03 -1.17909396e+00 -4.79351163e-01
-5.46238661e-01 -7.20299900e-01 1.27261233e+00 3.91251296e-01
1.12550783e+00 -3.88636351e-01 -4.31392789e-01 6.85894787e-01
1.27672076e+00 2.55678073e-02 5.21290362e-01 6.26263499e-01
6.92422926e-01 7.24854589e-01 4.43693519e-01 -1.63609594e-01
2.12497845e-01 1.49214134e-01 2.71835059e-01 -1.18217140e-01
-3.25017631e-01 -6.25644088e-01 4.58895177e-01 2.92741656e-01
2.43834987e-01 1.66197404e-01 -1.16046274e+00 9.26072180e-01
-1.34779656e+00 -7.32940555e-01 1.75000966e-01 2.18265152e+00
1.14636624e+00 3.63788158e-01 -1.20292522e-01 -4.24040765e-01
7.25044072e-01 3.44891578e-01 -1.09956610e+00 -6.69137239e-01
-2.43980274e-01 2.43108749e-01 7.61573851e-01 3.17388624e-01
-8.61494005e-01 1.09264863e+00 7.53218699e+00 3.58879007e-02
-1.41142750e+00 -2.52097715e-02 8.68665874e-01 -5.15831530e-01
-6.00664735e-01 3.00291609e-02 -9.30887043e-01 3.57209653e-01
9.61226821e-01 -1.75236210e-01 4.14447993e-01 9.18390691e-01
-1.97357610e-01 -8.42620656e-02 -1.62415314e+00 6.29220068e-01
1.31117195e-01 -1.31624711e+00 4.82812524e-01 1.09362006e-02
1.17900062e+00 3.16851437e-01 4.49953467e-01 7.97804594e-01
8.73793364e-01 -1.38662481e+00 8.13173890e-01 3.54713202e-01
8.22457433e-01 -5.51897764e-01 3.27335984e-01 -5.66058746e-03
-2.67331839e-01 -3.05681914e-01 -5.59803545e-01 -4.46923822e-01
-5.85025907e-01 5.45476913e-01 -9.29920435e-01 -6.16145194e-01
1.17938912e+00 7.38363087e-01 -1.42593014e+00 7.04755604e-01
-2.61533886e-01 7.39222288e-01 5.43431006e-02 -1.10783044e-03
-6.67377561e-02 1.52389839e-01 4.53778505e-02 1.37518334e+00
6.39246851e-02 -3.07288706e-01 -3.55670661e-01 9.68699098e-01
-4.80620623e-01 -2.91491270e-01 -7.30096459e-01 -3.11755002e-01
6.47402167e-01 1.16397989e+00 -4.27404940e-01 -4.81329381e-01
-2.71488160e-01 4.95882392e-01 5.82520187e-01 5.23341119e-01
-5.16765177e-01 -5.23334444e-02 1.23627520e+00 1.60280704e-01
1.38083488e-01 -9.16089956e-03 -8.87568891e-01 -8.89560282e-01
-4.54397947e-01 -1.25941730e+00 4.72548723e-01 -1.02114093e+00
-1.59010303e+00 2.25370899e-01 -4.94863540e-02 -4.86443430e-01
3.26289117e-01 -7.91659057e-01 -3.70061755e-01 9.39453661e-01
-1.59856236e+00 -8.87548923e-01 -6.11523211e-01 6.15244389e-01
2.39569694e-01 -1.88587397e-01 7.15223789e-01 3.31263058e-02
-5.59335113e-01 6.85948670e-01 -1.21898793e-01 3.74774724e-01
1.57905030e+00 -1.08340621e+00 5.27398884e-01 8.30258310e-01
1.26553895e-02 1.13698542e+00 6.75880313e-01 -7.75207937e-01
-9.47992563e-01 -1.33795464e+00 6.01390243e-01 -9.28352714e-01
4.65285957e-01 -3.64865005e-01 -9.20010209e-01 1.22842896e+00
3.87861937e-01 8.31891224e-02 7.29175389e-01 7.26147175e-01
-1.02321613e+00 -4.04265791e-01 -1.12954426e+00 5.92484832e-01
1.54881406e+00 -5.75704753e-01 -8.30264807e-01 -7.99375847e-02
6.00771368e-01 -1.95773914e-01 -7.60172486e-01 3.28697562e-01
6.25169098e-01 -9.60068524e-01 9.51847970e-01 -1.26052189e+00
8.83918226e-01 -9.18625109e-03 -2.19771832e-01 -2.01305151e+00
-8.22328508e-01 2.49957681e-01 3.95198077e-01 1.22290790e+00
8.17906618e-01 -9.49009120e-01 4.79310662e-01 8.54582310e-01
2.10509915e-02 -2.46115983e-01 -3.07473838e-01 -5.79049230e-01
5.43525636e-01 -5.93971312e-02 7.37334967e-01 1.26770937e+00
-4.64131117e-01 4.41480368e-01 9.96195823e-02 1.37162939e-01
5.59199929e-01 3.02397050e-02 8.79728973e-01 -1.29049611e+00
2.06002384e-01 -6.23242199e-01 -1.62078768e-01 -5.39583802e-01
1.62567437e-01 -1.03407025e+00 -7.48649763e-04 -1.31472278e+00
5.70448518e-01 -8.88115168e-01 -7.06286132e-01 6.38010144e-01
-4.08076793e-01 1.70074850e-01 2.71272063e-01 1.73557594e-01
-2.48680651e-01 3.46641004e-01 1.13080227e+00 -3.65893275e-01
-7.30665326e-02 -6.55121505e-01 -1.35681987e+00 6.53798580e-01
7.64996588e-01 -6.59174025e-01 -4.67430413e-01 -9.52119112e-01
5.66844523e-01 -1.07992828e+00 5.30326128e-01 -9.50320542e-01
-1.03973791e-01 -4.68439043e-01 1.07530451e+00 7.39540718e-03
-4.93996479e-02 -6.21821344e-01 -9.06644091e-02 5.50832689e-01
-9.68887091e-01 2.00205013e-01 7.98597276e-01 4.61350530e-01
7.49366209e-02 1.22679695e-01 1.08355999e+00 -2.57866442e-01
-1.06011665e+00 8.85900110e-02 -2.69420981e-01 4.38008279e-01
6.66209638e-01 -6.57666773e-02 -9.59123433e-01 -2.12667454e-02
-3.12016457e-01 1.87707409e-01 7.86795676e-01 9.13281918e-01
2.65446663e-01 -1.39829683e+00 -5.39879799e-01 2.68873453e-01
5.35133898e-01 -3.68917346e-01 -4.70224991e-02 4.02345091e-01
-4.44687903e-02 4.52773154e-01 -8.63717198e-01 -6.80144608e-01
-1.13822198e+00 4.93654311e-01 4.63358223e-01 1.15195904e-02
2.56773770e-01 1.30678713e+00 6.47577345e-01 -9.62304354e-01
-2.79145204e-02 -7.57844865e-01 -2.45383140e-02 3.91864508e-01
1.83051005e-01 2.37174854e-01 3.07725482e-02 -3.67077529e-01
-4.48295504e-01 4.12613958e-01 -3.27851266e-01 3.25169340e-02
1.38850558e+00 7.37831593e-02 7.87890404e-02 4.53274488e-01
9.38892603e-01 -2.55516231e-01 -1.43033683e+00 -2.73923546e-01
-2.45787323e-01 -5.57642996e-01 1.60350740e-01 -1.31672394e+00
-1.10406804e+00 1.03079081e+00 1.03555250e+00 -5.12884676e-01
8.38372290e-01 -2.16348752e-01 4.41742837e-02 5.72427034e-01
3.34586054e-01 -1.38399982e+00 3.54798287e-01 6.50834322e-01
1.17806828e+00 -1.49226415e+00 1.20046452e-01 2.60860562e-01
-7.45213985e-01 5.91597736e-01 1.25644076e+00 -2.49932155e-01
5.82993865e-01 -2.46275254e-02 3.84727448e-01 -1.47524342e-01
-8.80057096e-01 -2.75263283e-03 3.05091649e-01 8.83651733e-01
5.36460638e-01 5.36617301e-02 8.65851417e-02 4.10552919e-01
-4.24379647e-01 -4.75555658e-02 3.37037504e-01 7.08870173e-01
-3.75765860e-01 -6.92976177e-01 -5.19726694e-01 8.15660715e-01
1.06951073e-01 -3.50065947e-01 -9.77684021e-01 8.64254355e-01
4.64879036e-01 6.14920020e-01 4.21464205e-01 -3.69524300e-01
5.54221749e-01 3.70135307e-01 6.39005482e-01 -9.45047855e-01
-1.06534755e+00 -8.19981396e-01 1.68007955e-01 -6.74484372e-01
-2.88457334e-01 -6.85131907e-01 -7.81675696e-01 -3.13930959e-01
-6.02616630e-02 -4.74601090e-01 5.58377326e-01 5.56920409e-01
6.33677781e-01 6.01595402e-01 2.07335040e-01 -7.93292940e-01
-5.88961124e-01 -1.23134947e+00 -2.23824218e-01 8.27957928e-01
3.31135660e-01 -7.96186388e-01 -4.37248290e-01 1.06185883e-01]
|
[10.206670761108398, 2.2266547679901123]
|
dbebca04-1a8a-4b08-82d3-b82c2a4b7ecf
|
aifb-webscience-at-semeval-2022-task-12
|
2203.05325
| null |
https://arxiv.org/abs/2203.05325v2
|
https://arxiv.org/pdf/2203.05325v2.pdf
|
AIFB-WebScience at SemEval-2022 Task 12: Relation Extraction First -- Using Relation Extraction to Identify Entities
|
In this paper, we present an end-to-end joint entity and relation extraction approach based on transformer-based language models. We apply the model to the task of linking mathematical symbols to their descriptions in LaTeX documents. In contrast to existing approaches, which perform entity and relation extraction in sequence, our system incorporates information from relation extraction into entity extraction. This means that the system can be trained even on data sets where only a subset of all valid entity spans is annotated. We provide an extensive evaluation of the proposed system and its strengths and weaknesses. Our approach, which can be scaled dynamically in computational complexity at inference time, produces predictions with high precision and reaches 3rd place in the leaderboard of SemEval-2022 Task 12. For inputs in the domain of physics and math, it achieves high relation extraction macro F1 scores of 95.43% and 79.17%, respectively. The code used for training and evaluating our models is available at: https://github.com/nicpopovic/RE1st
|
['Michael Färber', 'Walter Laurito', 'Nicholas Popovic']
|
2022-03-10
| null | null | null | null |
['joint-entity-and-relation-extraction']
|
['natural-language-processing']
|
[-2.71623787e-02 4.44423020e-01 -1.75948814e-01 -5.02225757e-01
-8.18669379e-01 -7.46994734e-01 7.53463447e-01 5.75945854e-01
-3.38220984e-01 9.58560348e-01 -1.30911589e-01 -5.86627603e-01
-1.89930797e-01 -9.40949142e-01 -8.18654835e-01 3.15090045e-02
-1.44039933e-02 7.40200818e-01 3.30998152e-01 -1.48219556e-01
1.08661279e-01 2.47881114e-01 -1.18891883e+00 2.99816102e-01
1.06058335e+00 8.77500594e-01 5.18499129e-02 6.11775517e-01
-2.92158931e-01 1.09396517e+00 -5.32668591e-01 -1.00503874e+00
-1.61380805e-02 6.87504634e-02 -9.93526220e-01 -6.37861013e-01
5.35125017e-01 -5.67104407e-02 -4.13124114e-01 8.18745315e-01
4.00485128e-01 6.05992861e-02 5.48084915e-01 -1.20420504e+00
-6.06647432e-01 1.14147997e+00 -1.34866059e-01 1.41231060e-01
5.73252678e-01 -3.70265722e-01 1.27463806e+00 -1.21901190e+00
7.47306764e-01 1.02727735e+00 6.57288611e-01 2.47346699e-01
-1.18482792e+00 -8.53930950e-01 -4.40552905e-02 3.54291946e-01
-1.69820654e+00 -6.17120624e-01 4.65730667e-01 -5.32304645e-01
1.51672041e+00 2.17175499e-01 2.40188271e-01 8.12802613e-01
-3.50631191e-03 8.05444419e-01 9.50437784e-01 -5.86541474e-01
-2.07202300e-01 2.61713833e-01 4.01404858e-01 8.50611389e-01
3.60960752e-01 -9.91132855e-02 -7.33052313e-01 -9.46984440e-02
4.96325225e-01 -6.76411748e-01 -1.27653172e-02 -1.08454973e-01
-1.23162711e+00 3.48227441e-01 1.56882614e-01 1.83770299e-01
-2.65966386e-01 -5.09701259e-02 3.11607659e-01 1.12554260e-01
4.04389113e-01 6.45594954e-01 -8.54713023e-01 -3.21484596e-01
-1.07277179e+00 5.85791230e-01 1.21254015e+00 1.47811806e+00
3.59355867e-01 -3.96161824e-01 -2.92348862e-01 7.49883175e-01
1.47579208e-01 2.98682541e-01 1.21754274e-01 -8.16144884e-01
7.72388935e-01 6.04945421e-01 1.20692886e-01 -8.54364514e-01
-3.27840269e-01 -5.26821852e-01 -5.54647684e-01 -2.18285903e-01
5.42717397e-01 -1.66558638e-01 -7.10694849e-01 1.59259713e+00
3.18933010e-01 2.02476129e-01 1.97588027e-01 4.18523639e-01
1.43628752e+00 4.16395724e-01 3.69939774e-01 2.82285344e-02
1.48836648e+00 -9.25083101e-01 -9.05834496e-01 -2.57808622e-02
7.22812593e-01 -1.02738345e+00 6.57677114e-01 2.86371320e-01
-1.36435831e+00 -6.07808828e-01 -9.43069160e-01 -4.57273275e-01
-6.72359824e-01 3.82531375e-01 7.90401816e-01 2.93875545e-01
-7.31016278e-01 8.39229524e-01 -8.79684567e-01 -3.53264362e-01
2.94907272e-01 4.82521266e-01 -3.03096116e-01 2.02910930e-01
-1.43254161e+00 1.21182227e+00 7.29148030e-01 7.05066370e-03
-2.58752137e-01 -8.54853153e-01 -8.54425848e-01 8.20070282e-02
4.59781170e-01 -6.25939548e-01 1.47079384e+00 -4.88077700e-02
-1.35757041e+00 6.96633995e-01 -3.55262399e-01 -5.96620262e-01
5.32119632e-01 -6.22193754e-01 -4.44371700e-01 -1.31541744e-01
3.76618281e-02 5.25339007e-01 -1.58809051e-01 -7.62607515e-01
-7.38030136e-01 1.28256008e-01 2.25135133e-01 4.53306101e-02
-3.82573530e-02 3.31388086e-01 -7.34320998e-01 -5.06251812e-01
-4.23830859e-02 -7.10082591e-01 -2.89278105e-02 -4.10657853e-01
-6.71732247e-01 -6.56324267e-01 3.10593367e-01 -8.91571760e-01
1.45076799e+00 -1.66770422e+00 9.90733802e-02 1.42548874e-01
2.55461365e-01 3.04227889e-01 2.47860491e-01 6.05594397e-01
-1.37958422e-01 1.96066111e-01 -2.56222695e-01 -3.61504018e-01
2.78616518e-01 -3.24937031e-02 -2.90070117e-01 -2.84461901e-02
5.04811108e-01 1.11241078e+00 -9.32768345e-01 -7.66805172e-01
1.73230022e-01 4.34888810e-01 -3.00403029e-01 1.84685111e-01
-2.45679826e-01 2.64081031e-01 -4.75370944e-01 6.92789674e-01
5.23057580e-01 -3.84568423e-01 4.73358154e-01 -2.57083505e-01
-2.02901036e-01 1.03452635e+00 -1.24947011e+00 1.65614462e+00
-4.21884149e-01 6.06978774e-01 -4.45014596e-01 -8.68834019e-01
9.87667680e-01 5.12432694e-01 3.85981411e-01 -6.14230275e-01
-4.34936024e-02 3.87247056e-01 4.76062261e-02 -4.25670922e-01
6.12784028e-01 1.77041888e-01 -2.17295840e-01 -7.74275959e-02
3.92991453e-01 -1.22374572e-01 8.62646043e-01 4.53464448e-01
1.30550730e+00 6.87451780e-01 6.27082944e-01 -1.68957442e-01
7.57542670e-01 -1.02622733e-02 4.77268428e-01 6.70619905e-01
2.27570295e-01 1.06467932e-01 4.97156084e-01 -1.65674672e-01
-1.01104939e+00 -1.04863250e+00 -4.93273407e-01 7.97291458e-01
-1.79446131e-01 -1.08333540e+00 -3.65973055e-01 -6.95806384e-01
8.92434940e-02 1.00696123e+00 -1.25344992e-01 2.25435257e-01
-7.16962576e-01 -3.51364881e-01 8.23598981e-01 5.14650464e-01
4.71351057e-01 -9.32147503e-01 -2.73650140e-01 2.33604178e-01
-3.29966307e-01 -1.71721029e+00 2.33594924e-01 1.65527225e-01
-5.97016811e-01 -1.06158268e+00 -2.04100490e-01 -8.36287737e-01
3.96347910e-01 -6.25207305e-01 1.61307192e+00 -1.94105264e-02
-5.98853491e-02 -2.00074762e-02 -2.48441458e-01 -3.77228469e-01
-3.32832813e-01 4.35929149e-01 -1.75404638e-01 -6.98322058e-01
5.95439136e-01 -5.74504435e-01 -8.88048783e-02 -8.30010399e-02
-2.84265220e-01 4.91016239e-01 6.14460707e-01 5.58617592e-01
4.79544789e-01 -3.06071769e-02 3.94384265e-01 -1.33862305e+00
3.82567942e-01 -5.06497085e-01 -7.73492515e-01 4.75933731e-01
-7.66614497e-01 4.32334244e-02 6.16680086e-01 -3.13773192e-02
-9.70132053e-01 3.16523075e-01 -2.95850515e-01 4.38150913e-02
-2.21953601e-01 7.63478637e-01 -9.56811830e-02 2.29283482e-01
3.83018523e-01 -5.43204807e-02 -6.60662949e-01 -7.55534649e-01
5.47992289e-01 6.18658900e-01 6.94930434e-01 -9.56607461e-01
9.30011690e-01 -2.75478154e-01 1.66125149e-01 -4.58930641e-01
-9.53243673e-01 -3.04682612e-01 -8.68102789e-01 1.82361871e-01
4.43448693e-01 -1.05517876e+00 -8.54965210e-01 1.97383419e-01
-1.36666524e+00 -1.52861282e-01 -1.64837465e-01 5.12786210e-01
-3.42505217e-01 2.26713032e-01 -7.36934364e-01 -8.10262263e-01
-4.24269766e-01 -7.53091693e-01 9.87961888e-01 2.71460503e-01
-5.20907879e-01 -9.38146472e-01 -1.65534645e-01 2.75839388e-01
1.31513134e-01 3.40708613e-01 1.04988980e+00 -1.00735462e+00
-6.58897519e-01 -2.90418059e-01 -3.39607239e-01 7.12442994e-02
-3.46173532e-03 1.41481891e-01 -6.57824814e-01 8.87820050e-02
-8.29699516e-01 -3.90182137e-01 5.83611906e-01 -1.43711433e-01
1.04614091e+00 -2.19566494e-01 -5.57861984e-01 3.65497261e-01
1.30539310e+00 7.79028833e-02 5.41906893e-01 2.61634171e-01
6.50841594e-01 6.13576233e-01 7.59011984e-01 2.61775374e-01
7.75385678e-01 9.11968112e-01 -1.44225001e-01 7.26484358e-02
-2.95213133e-01 -4.69457299e-01 1.39161870e-01 8.58697474e-01
-1.90271273e-01 -3.85384619e-01 -1.24656272e+00 5.88628829e-01
-1.84556210e+00 -7.88389504e-01 -4.87798303e-01 1.93368697e+00
1.41822350e+00 4.22498941e-01 -7.28452280e-02 7.05085322e-02
4.68069643e-01 -3.15726668e-01 -9.43063125e-02 -3.06261122e-01
1.30915390e-02 8.78406167e-01 5.46849906e-01 7.18549788e-01
-1.28221250e+00 1.41610515e+00 5.81876135e+00 8.30260932e-01
-6.21156812e-01 -8.88923258e-02 3.13129932e-01 7.82253500e-03
9.16828681e-03 3.02337587e-01 -1.24007881e+00 3.03666234e-01
1.32245994e+00 -3.28897297e-01 2.44410172e-01 6.00059688e-01
-6.34955019e-02 6.11185506e-02 -1.33466458e+00 6.36069238e-01
-3.93605798e-01 -1.35100567e+00 -2.31597930e-01 -3.37111562e-01
3.76344502e-01 -8.45279172e-02 -2.35871106e-01 6.82452559e-01
5.51384926e-01 -1.10913563e+00 6.95401311e-01 6.73783660e-01
7.42507339e-01 -6.60149634e-01 7.48230934e-01 3.62794459e-01
-1.35160995e+00 3.47525269e-01 -4.25073951e-02 -1.97157711e-01
1.80662304e-01 7.60422409e-01 -1.11366069e+00 9.94154274e-01
5.42138755e-01 7.17023671e-01 -6.69600010e-01 8.52312624e-01
-7.59337604e-01 7.77619779e-01 -4.38381910e-01 -5.09992279e-02
-2.35372186e-01 4.68000546e-02 4.36452985e-01 1.56412983e+00
2.52359957e-01 3.02416325e-01 1.53361157e-01 9.05189574e-01
-2.88578182e-01 2.43329570e-01 -5.25640428e-01 -2.33055517e-01
9.54819381e-01 1.34338033e+00 -3.73524100e-01 -6.33708000e-01
-4.73836273e-01 6.84078455e-01 7.68714607e-01 1.91018194e-01
-1.09962380e+00 -8.34986687e-01 1.97739348e-01 5.69494031e-02
4.94959265e-01 -4.79048014e-01 -4.06901389e-01 -9.88313913e-01
2.41455913e-01 -7.22414613e-01 2.94826478e-01 -6.27260625e-01
-1.08211565e+00 6.06180429e-01 2.55315393e-01 -8.45155537e-01
-4.47853804e-01 -6.54263198e-01 -2.01755345e-01 1.03962159e+00
-1.49687970e+00 -1.34144092e+00 -1.16584457e-01 -2.50568707e-03
2.40594253e-01 -9.62487310e-02 1.07290387e+00 6.26853943e-01
-5.49137592e-01 9.98419285e-01 -1.52473241e-01 5.89265585e-01
6.90158606e-01 -1.42071366e+00 7.51961112e-01 9.00883973e-01
3.51416647e-01 9.72425342e-01 8.62523377e-01 -8.76404762e-01
-1.10455036e+00 -1.10468209e+00 1.89017391e+00 -7.20747173e-01
9.29402351e-01 -5.29210448e-01 -7.92118192e-01 8.91559660e-01
3.16603929e-01 -6.09159796e-03 6.67028010e-01 5.09519517e-01
-4.01648819e-01 2.11126357e-01 -9.05574501e-01 4.18485403e-01
1.29146492e+00 -3.22787434e-01 -6.31412446e-01 5.50044596e-01
6.82842314e-01 -1.04690063e+00 -1.52047181e+00 7.53432810e-01
5.53369462e-01 -4.10015047e-01 9.41075742e-01 -8.16608667e-01
8.33554804e-01 -3.79211366e-01 -1.82810262e-01 -8.45566571e-01
-2.57079363e-01 -6.68102562e-01 -6.61697328e-01 1.61330116e+00
9.65965927e-01 -3.84210467e-01 5.16957819e-01 7.04948902e-01
-1.69512406e-02 -1.05276370e+00 -7.47675121e-01 -7.98775434e-01
9.79543552e-02 -4.54089433e-01 5.61032116e-01 9.31406856e-01
1.14875011e-01 5.76441765e-01 -1.09345093e-01 3.49646688e-01
5.28097808e-01 3.64469111e-01 7.14857161e-01 -1.20086539e+00
-5.03624797e-01 -1.94738209e-01 -2.38267004e-01 -8.95615339e-01
3.72415066e-01 -1.28312325e+00 -1.82871655e-01 -1.64036393e+00
3.13452408e-02 -6.26993120e-01 -3.29198301e-01 8.24352860e-01
-1.53869167e-01 7.99725205e-02 1.80313393e-01 3.13316584e-02
-6.43481374e-01 2.57442355e-01 7.94539690e-01 4.91840169e-02
1.59813389e-02 -7.07080811e-02 -5.52584589e-01 6.97512329e-01
8.62672567e-01 -4.16141570e-01 -4.86362465e-02 -2.56310225e-01
3.36206257e-01 -3.22064534e-02 2.59647876e-01 -8.93740356e-01
3.70458573e-01 4.69704755e-02 3.54892850e-01 -7.37938344e-01
3.24142635e-01 -5.40482700e-01 1.36363536e-01 1.70389965e-01
-4.80585277e-01 7.57180601e-02 4.43849385e-01 1.02204807e-01
-1.09670632e-01 -2.62507498e-01 3.39656860e-01 -2.84665427e-03
-6.38884366e-01 1.29852341e-02 -7.54684024e-03 6.85131401e-02
8.56402397e-01 3.35555226e-01 -5.88319361e-01 -2.94208117e-02
-7.11118877e-01 3.34770590e-01 6.85038865e-02 4.31620926e-01
3.28497469e-01 -1.28845704e+00 -8.59249473e-01 -1.95898235e-01
8.66419226e-02 6.26548529e-02 -2.05428243e-01 8.55314136e-01
-3.76144171e-01 7.88263738e-01 4.57536317e-02 -2.12777987e-01
-1.41324806e+00 3.12735945e-01 1.26657695e-01 -7.04691052e-01
-4.73548472e-01 8.84985268e-01 -4.77951556e-01 -5.52578568e-01
1.98929772e-01 -3.55462909e-01 -3.91142726e-01 -2.00237647e-01
2.38915071e-01 1.99033916e-01 3.07757825e-01 -4.48607236e-01
-6.11755133e-01 2.23795399e-01 -2.10826248e-01 1.09360870e-02
1.34244597e+00 3.07444453e-01 -2.29472533e-01 4.40909356e-01
8.35387468e-01 3.32533419e-01 -5.05765676e-01 -4.03400719e-01
4.32934612e-01 -1.56230718e-01 -1.21561609e-01 -1.34073973e+00
-6.47388518e-01 6.45060003e-01 1.83009595e-01 1.72783881e-02
8.93308401e-01 1.94263637e-01 7.95903325e-01 5.24615526e-01
2.37906605e-01 -8.67309451e-01 -6.44662559e-01 7.94483125e-01
6.47286534e-01 -1.14477968e+00 3.32000583e-01 -9.35799479e-01
-3.90793353e-01 1.01709831e+00 7.50941992e-01 -4.24653701e-02
4.81075317e-01 6.19491696e-01 -3.90692234e-01 -1.36544019e-01
-1.05981612e+00 -2.60503352e-01 6.29752696e-01 2.83003241e-01
1.17418253e+00 2.17869490e-01 -6.26443624e-01 8.84204090e-01
-6.96830869e-01 1.30436391e-01 2.03802630e-01 9.14703786e-01
-1.72312886e-01 -1.53073323e+00 1.29882336e-01 3.97177428e-01
-5.96691489e-01 -5.62414050e-01 -4.67562914e-01 9.17382777e-01
2.91296631e-01 7.98143208e-01 -1.24816120e-01 -4.53382313e-01
5.72394252e-01 4.39172238e-01 7.16111958e-01 -7.02163577e-01
-5.79712689e-01 -3.38328302e-01 8.04169714e-01 -3.69396031e-01
-3.44386339e-01 -7.10530818e-01 -1.61185789e+00 -3.88658673e-01
-1.79895222e-01 3.05584013e-01 5.06400645e-01 1.00199473e+00
5.10218859e-01 7.83073723e-01 2.78920885e-02 -1.68718219e-01
-4.16689843e-01 -1.21222210e+00 -5.55659719e-02 2.04619244e-01
-1.54015005e-01 -7.70393789e-01 2.81199574e-01 1.45633638e-01]
|
[9.527030944824219, 8.824071884155273]
|
52e8fa24-743c-4141-bda2-aaa4d4a04179
|
data-and-knowledge-dual-driven-automatic
|
2206.15035
| null |
https://arxiv.org/abs/2206.15035v1
|
https://arxiv.org/pdf/2206.15035v1.pdf
|
Data-and-Knowledge Dual-Driven Automatic Modulation Recognition for Wireless Communication Networks
|
Automatic modulation classification is of crucial importance in wireless communication networks. Deep learning based automatic modulation classification schemes have attracted extensive attention due to the superior accuracy. However, the data-driven method relies on a large amount of training samples and the classification accuracy is poor in the low signal-to-noise radio (SNR). In order to tackle these problems, a novel data-and-knowledge dual-driven automatic modulation classification scheme based on radio frequency machine learning is proposed by exploiting the attribute features of different modulations. The visual model is utilized to extract visual features. The attribute learning model is used to learn the attribute semantic representations. The transformation model is proposed to convert the attribute representation into the visual space. Extensive simulation results demonstrate that our proposed automatic modulation classification scheme can achieve better performance than the benchmark schemes in terms of the classification accuracy, especially in the low SNR. Moreover, the confusion among high-order modulations is reduced by using our proposed scheme compared with other traditional schemes.
|
['Zhu Han', 'Qihui Wu', 'Fuhui Zhou', 'Hao Zhang', 'Rui Ding']
|
2022-06-30
| null | null | null | null |
['automatic-modulation-recognition']
|
['time-series']
|
[ 5.70619702e-01 -3.66882741e-01 -3.51211041e-01 -4.55022663e-01
-7.92618334e-01 1.46777593e-02 5.12222707e-01 -4.74289106e-03
-1.89091712e-01 7.51707196e-01 1.06867971e-02 -3.71978372e-01
-5.42723656e-01 -1.00454187e+00 -1.59855753e-01 -1.20699584e+00
5.73994853e-02 -1.97340429e-01 -9.07395855e-02 -1.58512846e-01
2.80335724e-01 2.79664725e-01 -1.41007900e+00 2.58169889e-01
9.23566222e-01 1.54105890e+00 4.06570435e-01 2.89940298e-01
-2.42960200e-01 6.98246121e-01 -7.32130945e-01 1.28461853e-01
-5.60771972e-02 -7.28868663e-01 -3.21652651e-01 -3.15317139e-02
-3.49517286e-01 8.12788829e-02 -5.09363115e-01 1.10939741e+00
7.60099232e-01 -1.89617291e-01 8.47923636e-01 -1.30252171e+00
-3.09043854e-01 3.73962313e-01 -7.31218696e-01 2.14823693e-01
8.59349892e-02 -2.10160896e-01 7.04699695e-01 -5.47138929e-01
-3.83781940e-02 1.10222495e+00 4.71634209e-01 2.06854731e-01
-1.05170763e+00 -1.02560592e+00 -3.49990577e-02 8.84326756e-01
-1.54269505e+00 -5.35361290e-01 1.13643885e+00 -2.22892180e-01
2.47722685e-01 1.70518830e-01 6.08280897e-01 7.83996582e-01
2.18568623e-01 5.38008094e-01 1.35004532e+00 -6.63841665e-01
1.66569740e-01 1.94872797e-01 -2.32584655e-01 6.04610860e-01
1.03086308e-01 1.06433637e-01 -4.17388529e-01 1.61865987e-02
5.33937037e-01 -8.09235051e-02 -4.62906033e-01 -3.28822553e-01
-1.02288675e+00 7.51958549e-01 5.78411758e-01 5.88987648e-01
-2.91123509e-01 1.14771113e-01 2.09183365e-01 4.64632690e-01
1.77084282e-01 2.20079213e-01 -1.40067995e-01 -3.29579785e-02
-8.46401930e-01 -4.39763427e-01 4.15499240e-01 8.28889430e-01
6.93156660e-01 4.68600869e-01 -1.94105700e-01 8.57815266e-01
5.39048970e-01 4.90384489e-01 5.12739956e-01 -5.86090803e-01
3.83048594e-01 3.70172530e-01 -2.65889913e-01 -1.24762928e+00
-6.18690848e-01 -1.09456146e+00 -1.29565489e+00 2.03334942e-01
5.84107712e-02 6.63914159e-02 -9.27611828e-01 1.62627292e+00
3.29915509e-02 7.19394758e-02 3.85601878e-01 7.75779784e-01
9.39403415e-01 7.87516356e-01 3.27415049e-01 -5.46520591e-01
1.31265426e+00 -3.56574059e-01 -1.10584641e+00 5.47187962e-02
6.71543121e-01 -8.25265646e-01 7.64817894e-01 4.57665443e-01
-4.96194601e-01 -7.47995317e-01 -1.79494762e+00 4.18161422e-01
-2.20045075e-01 3.62926781e-01 6.03134751e-01 1.13670540e+00
-4.03339714e-01 2.43639931e-01 -2.79370934e-01 -1.26131371e-01
1.01836383e+00 5.81120670e-01 -8.95123780e-02 9.81238186e-02
-1.43988132e+00 5.56578755e-01 6.16042316e-01 3.35116908e-02
-5.46264708e-01 -2.43736356e-01 -7.31457770e-01 2.13936776e-01
2.19654143e-01 -3.31137806e-01 9.72370207e-01 -1.22706068e+00
-1.52034295e+00 3.46317500e-01 3.16016818e-03 -3.09903562e-01
1.60337105e-01 3.66489649e-01 -1.04059708e+00 3.75948399e-01
-1.51904881e-01 6.99568614e-02 1.28280628e+00 -1.26294303e+00
-9.89982963e-01 -3.10331285e-01 4.83612902e-02 3.02643836e-01
-6.31452620e-01 -4.12562340e-01 -9.17668045e-02 -8.76377761e-01
4.13085252e-01 -5.17949820e-01 1.28806025e-01 -2.39158764e-01
-1.23808324e-01 1.55787334e-01 1.20179248e+00 -5.78698158e-01
1.29707563e+00 -2.15835023e+00 2.45115738e-02 5.69544435e-01
1.86746642e-01 3.80276531e-01 1.38102576e-01 1.37814339e-02
2.16751285e-02 -1.35560393e-01 -7.81109631e-02 3.28301489e-01
-2.36020625e-01 -1.34490952e-01 9.56706405e-02 5.11722028e-01
7.43159791e-04 4.01947707e-01 -7.02557266e-01 -6.62344694e-01
3.17236334e-01 4.90287751e-01 -9.61949080e-02 8.66935477e-02
8.82882848e-02 6.75112665e-01 -6.76226914e-01 5.94516218e-01
8.44057798e-01 -1.92395926e-01 3.29094350e-01 -8.08595300e-01
2.03972325e-01 -4.58623469e-02 -1.04125917e+00 1.41474223e+00
-9.35326755e-01 6.75138474e-01 5.40594710e-03 -1.45595837e+00
1.04604495e+00 4.25384998e-01 3.73900205e-01 -1.20569456e+00
3.86133850e-01 2.49843195e-01 3.85758996e-01 -4.40204442e-01
-1.28970206e-01 -3.73111010e-01 -1.29101664e-01 1.96594521e-01
5.64149301e-03 1.33050129e-01 -1.14873081e-01 -1.25917390e-01
5.80369532e-01 -1.32117802e-02 5.90222538e-01 -9.40708593e-02
1.18298733e+00 -3.67435604e-01 5.50541162e-01 5.09632289e-01
8.69760886e-02 2.41950061e-02 2.12533861e-01 -1.32117301e-01
-5.91973126e-01 -6.43106401e-01 -4.02808338e-01 8.49012792e-01
5.83467603e-01 -1.72751337e-01 -5.13266385e-01 -7.12864280e-01
-3.93523782e-01 4.97188419e-01 -3.00849795e-01 -6.57417774e-01
-2.07120717e-01 -1.06533408e+00 3.90139133e-01 3.50067943e-01
1.02241445e+00 -8.93911242e-01 -2.49751255e-01 2.01536655e-01
-1.78135544e-01 -1.13161194e+00 1.65349156e-01 1.50441065e-01
-5.25148809e-01 -7.96234190e-01 -6.57845438e-01 -1.01029587e+00
5.28052688e-01 2.73019016e-01 4.64494854e-01 1.00777231e-01
-1.66987225e-01 8.98267999e-02 -4.61017042e-01 -4.97228622e-01
-5.63300550e-01 2.93874115e-01 6.35800324e-03 6.18866980e-01
3.71367157e-01 -7.56081223e-01 -6.75321162e-01 4.06120807e-01
-6.21454477e-01 1.35708615e-01 1.25612199e+00 7.55914807e-01
5.49545705e-01 5.68187594e-01 1.06151974e+00 -6.49480045e-01
3.51065069e-01 -5.38295329e-01 -5.01204967e-01 2.42684558e-02
-4.47425425e-01 1.98414624e-01 9.59864438e-01 -3.64617348e-01
-1.02028787e+00 -1.83480546e-01 -1.81845918e-01 2.64237434e-01
-7.96338636e-03 4.20967758e-01 -7.96118021e-01 -5.54213107e-01
4.08194542e-01 5.42785585e-01 2.39039175e-02 -4.69655782e-01
-7.65213976e-03 1.34415412e+00 2.76925296e-01 -1.74418837e-01
9.96671975e-01 3.81752819e-01 4.85219747e-01 -1.09972894e+00
-8.52964282e-01 -1.78399384e-01 -3.62696290e-01 -2.58833766e-01
6.51836753e-01 -9.31769252e-01 -6.39041483e-01 4.60008025e-01
-7.36263990e-01 1.91119581e-01 2.18418255e-01 7.60434508e-01
-6.73530400e-01 5.51555395e-01 -1.02728218e-01 -8.00616741e-01
-3.08121234e-01 -1.11046278e+00 5.71273446e-01 4.89209563e-01
9.04270485e-02 -7.11632371e-01 -6.01761520e-01 2.44885609e-01
6.03877425e-01 6.54115751e-02 1.54206777e+00 -5.44125140e-01
-4.75779176e-01 -2.33556870e-02 -4.91149396e-01 1.96718201e-01
5.31527460e-01 -6.60200894e-01 -9.66080964e-01 -2.43101820e-01
2.13413030e-01 -6.84230253e-02 6.24596119e-01 2.86827445e-01
1.73599541e+00 -1.08575054e-01 -4.05197918e-01 7.45780110e-01
1.38254189e+00 6.90029502e-01 7.78545141e-01 4.98637021e-01
5.41764617e-01 1.58230886e-01 9.54091966e-01 4.47361052e-01
3.54521759e-02 9.33923721e-01 4.57423866e-01 -3.24262053e-01
-1.27721280e-01 2.95819342e-02 -1.50820957e-02 5.41604578e-01
8.65987763e-02 -6.56196848e-02 -5.24090469e-01 1.30331159e-01
-1.63651717e+00 -8.63403678e-01 -1.14404097e-01 2.26921463e+00
6.65573299e-01 3.92376810e-01 -1.60891060e-02 9.71560895e-01
8.14858377e-01 2.32805789e-01 -2.78040200e-01 -1.06064558e-01
-1.63722172e-01 3.06894809e-01 3.15675348e-01 3.20504159e-01
-1.41995609e+00 4.30626482e-01 5.05652857e+00 1.51170361e+00
-1.29156005e+00 5.27456217e-02 5.92674613e-01 2.56453276e-01
8.17627087e-02 -1.96962282e-01 -4.71398592e-01 6.37038112e-01
8.34095120e-01 2.39618886e-02 2.33924091e-01 5.53530633e-01
2.52823353e-01 2.72979140e-02 -6.63733304e-01 1.31895041e+00
-2.93534547e-02 -1.24317968e+00 3.70102346e-01 1.76821545e-01
1.64219350e-01 -9.59855735e-01 3.40970039e-01 2.93722630e-01
-5.67993760e-01 -1.23968363e+00 4.25361544e-01 6.21752203e-01
1.09485507e+00 -1.20822442e+00 9.95805740e-01 2.05771342e-01
-1.36239731e+00 -3.36881727e-01 -1.88925013e-01 9.49529111e-02
-1.81285530e-01 3.83044600e-01 -4.71231848e-01 9.18857336e-01
3.92326087e-01 5.88555157e-01 -4.77457881e-01 1.06114900e+00
-7.65147898e-03 7.28900731e-01 -5.10320663e-02 -1.84343025e-01
4.18320373e-02 -1.07350871e-01 4.08741862e-01 9.76157844e-01
5.16529977e-01 -7.07062408e-02 1.12880908e-01 1.98500961e-01
-6.19519800e-02 3.73008698e-01 -3.05314273e-01 3.55349854e-02
6.03510737e-01 1.35834670e+00 -5.56962132e-01 -1.58991471e-01
-4.27911669e-01 8.38752389e-01 -3.01179111e-01 2.67115682e-01
-9.04040396e-01 -1.00130367e+00 1.85278252e-01 2.97171205e-01
3.81278425e-01 -7.74404630e-02 -2.80211598e-01 -4.60715622e-01
-1.40349671e-01 -8.74579549e-01 3.31230670e-01 -4.64498997e-01
-8.74314666e-01 5.21935821e-01 -1.51622489e-01 -1.80899358e+00
-4.71761487e-02 -4.72500801e-01 -3.01061362e-01 7.76559293e-01
-1.78696144e+00 -1.13451672e+00 -7.28896499e-01 5.33466339e-01
3.98918062e-01 -7.08145857e-01 9.08323944e-01 6.34105325e-01
-2.99008101e-01 9.81464982e-01 3.17799360e-01 9.72302034e-02
3.94061327e-01 -1.02875078e+00 -4.51905727e-01 3.88656855e-01
4.68082204e-02 1.42166421e-01 4.86708015e-01 -1.48982033e-01
-1.16469550e+00 -1.06850922e+00 2.98033863e-01 4.43956941e-01
3.42391729e-01 -2.44825408e-01 -7.22232878e-01 -6.49042726e-02
-2.76929350e-03 -2.09680982e-02 1.11128640e+00 -3.28409582e-01
-1.03730615e-02 -6.55527115e-01 -1.14721608e+00 4.69484270e-01
8.18832040e-01 -4.15887117e-01 -1.63786605e-01 2.48371124e-01
2.04161346e-01 -7.20840916e-02 -9.31439579e-01 5.88979006e-01
7.07010269e-01 -6.64322734e-01 9.50924039e-01 -2.05800280e-01
2.07649712e-02 -7.08886445e-01 -4.40306842e-01 -1.23581898e+00
-4.83797222e-01 -4.44092095e-01 -3.16405177e-01 1.16505361e+00
3.13361853e-01 -6.22020185e-01 6.62294626e-01 -5.73173106e-01
3.27806681e-01 -7.89942145e-01 -1.07162988e+00 -8.05987179e-01
-4.55953062e-01 -2.52229422e-01 6.22795999e-01 7.76822627e-01
-3.38374168e-01 7.07507908e-01 -4.69677389e-01 2.15294003e-01
8.08024764e-01 1.75168529e-01 6.19757473e-01 -1.36676610e+00
-4.30216759e-01 -3.07897687e-01 -1.03965926e+00 -1.00393713e+00
7.40472227e-02 -9.85338211e-01 -3.05058628e-01 -1.52104425e+00
1.23785548e-01 -7.45051205e-01 -7.11614132e-01 1.46426186e-01
6.74647987e-02 6.15523696e-01 -1.08761244e-01 1.77054599e-01
-5.44207513e-01 8.71505558e-01 1.12673211e+00 -3.47353160e-01
-4.62635979e-02 3.07981551e-01 -9.40698981e-01 7.86110520e-01
9.97150183e-01 -3.40049118e-01 -4.67356116e-01 2.53025964e-02
-9.54307541e-02 9.70639959e-02 1.44227922e-01 -1.62017834e+00
-1.37354508e-01 8.92599672e-02 8.30896616e-01 -3.64289194e-01
4.66741115e-01 -1.06078053e+00 -5.65749332e-02 5.82973897e-01
-1.51215300e-01 -6.74874961e-01 4.55222093e-02 9.05934751e-01
-2.86443561e-01 -7.38354623e-02 1.10054827e+00 2.62460768e-01
-8.44555080e-01 1.59696743e-01 -5.50941825e-01 -1.77126989e-01
1.01407778e+00 -4.77009058e-01 -3.66132148e-02 -8.72629166e-01
-6.84640825e-01 -1.37900665e-01 -4.77028862e-02 4.08837527e-01
5.72993398e-01 -1.80936086e+00 -6.05886638e-01 2.88049757e-01
4.00181442e-01 -5.52124500e-01 2.69630581e-01 9.51769829e-01
-3.40337247e-01 3.02492976e-01 -3.62421036e-01 -6.89741433e-01
-1.44035375e+00 2.29066178e-01 3.47782433e-01 -1.54267728e-01
-4.06999320e-01 2.22134873e-01 8.83248448e-02 1.60268143e-01
2.50695497e-01 3.09148967e-01 -8.18401694e-01 3.24112698e-02
4.95744497e-01 2.51308173e-01 1.54075161e-01 -9.49511051e-01
-3.67601514e-01 9.11388755e-01 -1.34949043e-01 1.30218500e-02
1.06921065e+00 -8.14670622e-02 1.66429505e-01 1.15119293e-01
1.45936966e+00 5.36186881e-02 -7.39497006e-01 -4.48794603e-01
-1.33014666e-02 -5.97669482e-01 3.83018076e-01 -7.19159484e-01
-1.01001310e+00 1.11059582e+00 1.33241081e+00 3.59936237e-01
1.55508590e+00 -3.83625716e-01 6.79929078e-01 2.81515270e-01
4.16949362e-01 -1.20153534e+00 -7.40283728e-02 2.86060981e-02
5.79221487e-01 -1.09322989e+00 2.55515516e-01 -5.96319854e-01
-3.01240921e-01 1.17602885e+00 3.17385226e-01 3.58512223e-01
7.00206637e-01 -2.07825582e-02 1.78343490e-01 9.82808415e-03
6.36501387e-02 -2.08588570e-01 3.52787465e-01 9.16895628e-01
2.90877372e-01 -5.94791472e-02 -4.32248145e-01 8.63833725e-01
-1.84357509e-01 -3.03160101e-01 1.85981825e-01 1.02232218e+00
-7.07162201e-01 -1.19018519e+00 -4.56048757e-01 7.55963862e-01
-5.99884570e-01 1.09208070e-01 9.07902326e-03 6.11286998e-01
2.13405594e-01 1.29016113e+00 -2.12610900e-01 -6.40806258e-01
1.11881666e-01 -2.16281936e-01 6.64331913e-01 -2.18971238e-01
1.54222623e-01 4.49611358e-02 1.10954732e-01 -2.60192066e-01
-5.73336899e-01 -6.68743551e-02 -1.29556775e+00 1.25388533e-01
-4.12923664e-01 3.12690914e-01 6.77685261e-01 1.15519392e+00
3.53428483e-01 8.78405869e-01 1.24576473e+00 -6.78497314e-01
-3.19094867e-01 -8.73693466e-01 -5.83034396e-01 3.10741395e-01
3.87002379e-01 -9.70171869e-01 -3.69104534e-01 -2.64299154e-01]
|
[6.546184539794922, 1.4743773937225342]
|
d4bb0814-d23f-46e4-944d-ceb5620c5f42
|
towards-bengali-word-embedding-corpus
| null | null |
https://aclanthology.org/2020.icon-main.61
|
https://aclanthology.org/2020.icon-main.61.pdf
|
Towards Bengali Word Embedding: Corpus Creation, Intrinsic and Extrinsic Evaluations
|
Distributional word vector representation or word embedding has become an essential ingredient in many natural language processing (NLP) tasks such as machine translation, document classification, information retrieval and question answering. Investigation of embedding model helps to reduce the feature space and improves textual semantic as well as syntactic relations. This paper presents three embedding techniques (such as Word2Vec, GloVe, and FastText) with different hyperparameters implemented on a Bengali corpus consists of 180 million words. The performance of the embedding techniques is evaluated with extrinsic and intrinsic ways. Extrinsic performance evaluated by text classification, which achieved a maximum of 96.48% accuracy. Intrinsic performance evaluated by word similarity (e.g., semantic, syntactic and relatedness) and analogy tasks. The maximum Pearson (rˆ) correlation accuracy of 60.66% (Ssrˆ) achieved for semantic similarities and 71.64% (Syrˆ) for syntactic similarities whereas the relatedness obtained 79.80% (Rsrˆ). The semantic word analogy tasks achieved 44.00% of accuracy while syntactic word analogy tasks obtained 36.00%.
|
['Mohammed Moshiul Hoque', 'Md. Rajib Hossain']
| null |
towards-bengali-word-embedding-corpus-1
|
https://aclanthology.org/2020.icon-main.61/
|
https://aclanthology.org/2020.icon-main.61
|
icon-17th-international-conference-on-natural
|
['word-similarity']
|
['natural-language-processing']
|
[-2.25231364e-01 2.56502777e-01 -4.82241102e-02 -2.94825077e-01
-5.02797723e-01 -5.23826957e-01 9.83955801e-01 6.50432229e-01
-1.01517045e+00 5.40551007e-01 6.39190614e-01 -4.71989572e-01
-3.16075057e-01 -8.62514853e-01 7.42236301e-02 -4.05189931e-01
3.32156109e-04 4.01727885e-01 -8.75852406e-02 -5.51465034e-01
6.70031369e-01 3.08173090e-01 -1.31860042e+00 2.15429649e-01
7.80916512e-01 7.24447072e-01 2.36660868e-01 8.79116654e-01
-7.20977902e-01 4.56769854e-01 -8.69001150e-01 -5.29886365e-01
-7.25395307e-02 4.03292887e-02 -9.06796515e-01 -7.19767094e-01
-2.52211630e-01 2.90861994e-01 -2.72193968e-01 1.00713778e+00
5.52087486e-01 4.66476023e-01 9.60061371e-01 -1.02293241e+00
-1.51322949e+00 5.65750182e-01 -3.73174012e-01 4.52556044e-01
3.75872165e-01 -4.31314886e-01 1.28217053e+00 -1.28793609e+00
4.77344722e-01 1.27733338e+00 5.00782907e-01 3.66183817e-01
-9.57811654e-01 -5.74535429e-01 -4.08426315e-01 3.25374961e-01
-1.25796008e+00 1.11204088e-01 4.59374100e-01 -4.02217507e-01
1.75792766e+00 4.20331448e-01 3.01077187e-01 7.53582716e-01
4.46218401e-01 3.43355805e-01 1.07149363e+00 -8.32757592e-01
2.46998444e-02 7.26152718e-01 9.28538680e-01 1.70747250e-01
4.09630626e-01 -1.41244918e-01 -2.56957591e-01 -4.32382196e-01
2.94368744e-01 -1.22956097e-01 8.18641018e-03 3.90181661e-01
-9.85700071e-01 1.37900615e+00 1.86418399e-01 9.37052488e-01
-3.77401114e-01 7.58165196e-02 7.26949513e-01 5.27949333e-01
4.41217452e-01 9.14556861e-01 -6.53957903e-01 -2.61629492e-01
-2.15108767e-01 5.25278524e-02 7.69474626e-01 7.88225353e-01
6.94400251e-01 2.01013446e-01 -5.98743856e-02 1.35411990e+00
5.01728117e-01 6.29745603e-01 1.27135324e+00 -3.97909462e-01
2.95780808e-01 7.74120212e-01 -1.90139487e-01 -1.39912653e+00
-2.88265646e-01 -5.16385436e-02 -5.50041735e-01 -1.80800438e-01
-1.65291391e-02 -1.46239609e-01 -6.89128280e-01 1.40590465e+00
1.02445200e-01 -3.61690938e-01 6.07631624e-01 5.65669060e-01
1.27618098e+00 1.22633719e+00 3.70820761e-01 -3.84508483e-02
1.95450747e+00 -8.54207516e-01 -1.03518796e+00 -1.79944992e-01
1.11779714e+00 -1.34261405e+00 1.17227924e+00 -3.95911187e-03
-6.97904527e-01 -6.17586613e-01 -1.02315092e+00 -1.19012333e-01
-1.01428044e+00 -6.36833385e-02 6.16829038e-01 8.15107465e-01
-8.77249300e-01 3.40958238e-01 -3.73077869e-01 -6.17705584e-01
-9.22212526e-02 4.42252606e-01 -8.03559601e-01 1.81471258e-01
-1.60827637e+00 1.26737309e+00 5.57440281e-01 -5.13471127e-01
1.71962474e-02 -5.99163532e-01 -1.07375216e+00 1.88028812e-03
-2.28291944e-01 -1.74768969e-01 6.30922198e-01 -6.39209807e-01
-1.15665555e+00 8.70302320e-01 -1.52037054e-01 -3.99173111e-01
-3.76214176e-01 -4.65158939e-01 -6.68534338e-01 -2.49430221e-02
1.52271897e-01 4.42798674e-01 1.61382526e-01 -7.43714750e-01
-2.94446051e-01 -4.06237513e-01 -2.13606924e-01 2.89710104e-01
-8.99691939e-01 5.50944328e-01 2.53187537e-01 -8.10847223e-01
-6.38680439e-03 -7.01846302e-01 -1.70373023e-02 -5.44505775e-01
6.02801926e-02 -8.97496819e-01 7.77331710e-01 -9.63953674e-01
1.41921926e+00 -2.00787115e+00 -6.90513775e-02 2.21757263e-01
-2.76177879e-02 6.14743650e-01 -2.13709950e-01 1.03489506e+00
-3.85667026e-01 3.54732960e-01 3.76065560e-02 9.08114836e-02
9.83846858e-02 4.03356761e-01 -1.98422924e-01 2.51010209e-01
1.88843712e-01 1.05047250e+00 -7.74649441e-01 -4.49317575e-01
2.93065161e-01 7.63027370e-01 -3.67455423e-01 1.09932989e-01
3.57921630e-01 -3.80391628e-01 -4.91966814e-01 3.46064836e-01
1.94996089e-01 1.91206157e-01 -5.77485003e-02 -1.59103051e-01
-4.40947935e-02 2.63669431e-01 -9.73347723e-01 1.42456639e+00
-8.10607195e-01 9.48188424e-01 -6.46552622e-01 -1.07801747e+00
1.57308125e+00 4.72436398e-01 2.91691840e-01 -7.62136996e-01
4.23946023e-01 1.98211774e-01 2.75824159e-01 -8.80787492e-01
9.91492391e-01 -1.26878023e-01 -1.10921659e-01 6.83419228e-01
2.06198335e-01 -1.29268449e-02 -4.37453054e-02 4.08349693e-01
1.09136367e+00 -3.21436435e-01 6.60096586e-01 -5.95653534e-01
7.62845933e-01 9.65091307e-03 -7.49206468e-02 1.66455492e-01
-1.19073413e-01 2.52249956e-01 3.96751255e-01 -3.25234473e-01
-1.20008159e+00 -8.24352384e-01 -3.95386994e-01 9.59932208e-01
2.94284876e-02 -6.11798704e-01 -3.80181223e-01 -3.65931660e-01
6.33957461e-02 1.21899581e+00 -7.54759967e-01 -4.66692895e-01
-3.88416141e-01 -6.37889504e-01 6.65222585e-01 7.25984871e-01
9.64933708e-02 -1.24441922e+00 -2.54508138e-01 4.81830463e-02
1.89152434e-01 -7.53365755e-01 -1.86127439e-01 1.75079167e-01
-7.15839088e-01 -7.80137360e-01 -4.72461134e-01 -9.20874357e-01
4.56404537e-01 3.31157329e-03 1.00110364e+00 -3.55483368e-02
-4.55663651e-01 2.30659291e-01 -8.30988228e-01 -4.33778375e-01
-2.95067251e-01 -9.72148776e-02 2.90454268e-01 -4.62265849e-01
1.30250907e+00 -4.12109584e-01 -4.37751502e-01 7.21828863e-02
-1.06244302e+00 -6.44824922e-01 6.04071021e-01 1.02642655e+00
2.31858686e-01 -4.19018805e-01 8.72655213e-01 -6.99092865e-01
1.34971845e+00 -6.26542926e-01 1.14686349e-02 3.33011210e-01
-9.51096892e-01 2.63804793e-01 3.30974877e-01 -5.37120342e-01
-7.28082478e-01 -7.40780771e-01 -4.01309490e-01 1.00971155e-01
-6.78046793e-02 5.89984000e-01 3.03846449e-01 2.93626487e-01
9.15794253e-01 1.55752674e-01 5.42890280e-02 -3.09712291e-01
4.91167575e-01 1.30380642e+00 6.36562705e-02 -5.15018761e-01
5.69706678e-01 -1.62756145e-02 -3.98178458e-01 -1.18038154e+00
-3.99018377e-01 -8.94628704e-01 -4.28050637e-01 2.42754534e-01
1.21228826e+00 -4.35054511e-01 -5.63306272e-01 -2.59981513e-01
-1.37615442e+00 5.03118277e-01 -2.68605918e-01 1.04653513e+00
5.88364564e-02 4.25409079e-01 -3.40410411e-01 -8.71755004e-01
-8.75196338e-01 -8.68486106e-01 7.06146538e-01 5.40724322e-02
-9.40979302e-01 -1.42675722e+00 2.63528109e-01 1.96481407e-01
6.11550570e-01 6.52440339e-02 1.26758456e+00 -1.44808674e+00
6.45759106e-01 -6.13809347e-01 -4.25110847e-01 5.05123079e-01
2.93110520e-01 -6.87221065e-02 -8.61339629e-01 1.02002494e-01
-3.31067815e-02 -1.29207477e-01 4.61575985e-01 1.15143657e-01
5.77704847e-01 -3.18918794e-01 -2.17393916e-02 1.23377748e-01
1.54334080e+00 5.33650279e-01 8.11960816e-01 7.11596549e-01
4.43638951e-01 9.02970493e-01 5.93490899e-01 2.19245389e-01
4.43595983e-02 4.64747697e-01 4.64056917e-02 4.33998376e-01
-7.68707469e-02 -3.07572423e-03 2.87556350e-01 1.34822309e+00
9.20730084e-02 -2.14148268e-01 -1.20404744e+00 6.57344639e-01
-1.61019099e+00 -7.70170152e-01 -4.59797263e-01 1.92934430e+00
7.35710680e-01 -1.33056998e-01 -3.93877894e-01 2.80656040e-01
5.42116106e-01 1.61393628e-01 3.90357226e-02 -1.46146739e+00
-5.47274053e-02 6.78688586e-01 3.80867869e-01 7.74213493e-01
-6.66091621e-01 1.06742501e+00 5.17338657e+00 8.76290381e-01
-6.42249763e-01 1.69670388e-01 1.55258700e-01 4.83758867e-01
-6.11301780e-01 -4.41664308e-02 -7.04875886e-01 3.43856394e-01
1.13460529e+00 -5.43193340e-01 3.69447134e-02 8.81445706e-01
-2.12397546e-01 2.22885590e-02 -7.72425473e-01 1.11015725e+00
4.66057688e-01 -1.11374784e+00 3.00128073e-01 -1.15004808e-01
5.64366102e-01 -1.54750407e-01 2.06015725e-02 5.51034629e-01
2.87657399e-02 -1.36210608e+00 -6.19267300e-02 3.14032197e-01
6.08049989e-01 -1.00945222e+00 1.36616802e+00 2.25549728e-01
-1.03680921e+00 1.03301518e-01 -8.88573468e-01 -2.81649858e-01
5.34400940e-02 3.98302048e-01 -6.71842277e-01 4.62376565e-01
5.78896165e-01 4.44644451e-01 -3.99746150e-01 2.14896709e-01
-3.77209097e-01 5.05333722e-01 -2.68617898e-01 -8.34995806e-01
5.13972640e-01 -4.25371200e-01 5.52866280e-01 1.41505337e+00
1.45403311e-01 2.49174878e-01 -5.40218294e-01 3.71572942e-01
9.04603228e-02 9.56422687e-01 -7.06187010e-01 -3.81873935e-01
7.14495063e-01 1.12367094e+00 -4.21987295e-01 -4.34748828e-01
-3.48089397e-01 8.04429471e-01 1.11245573e-01 1.80901021e-01
-7.60338724e-01 -1.20317137e+00 8.22929740e-01 -3.18530589e-01
-1.67845681e-01 -2.87628233e-01 -4.10682023e-01 -8.58304620e-01
-1.25875935e-01 -5.47362387e-01 2.99440086e-01 -6.94479227e-01
-1.39499724e+00 8.45333517e-01 4.07641195e-02 -7.62006462e-01
-1.88335344e-01 -1.00035167e+00 -7.44158089e-01 1.24466670e+00
-1.10091996e+00 -8.48577261e-01 -1.62177965e-01 2.15313733e-01
6.33913636e-01 -7.42934227e-01 1.32644796e+00 1.64376915e-01
-1.88837990e-01 7.50493944e-01 3.33306134e-01 2.97198534e-01
6.85119987e-01 -1.28200841e+00 1.60031423e-01 2.44137779e-01
2.07775503e-01 1.00242805e+00 6.66110635e-01 -4.98366863e-01
-1.35995924e+00 -7.46638238e-01 1.86508286e+00 -4.61893141e-01
1.05603969e+00 -6.25851899e-02 -9.75924611e-01 3.46136034e-01
5.60695767e-01 -3.24341297e-01 1.31124449e+00 1.68526977e-01
-4.37684149e-01 1.38526306e-01 -1.13711739e+00 8.05823028e-01
4.02932525e-01 -6.93543673e-01 -1.30585074e+00 4.03255731e-01
1.02856255e+00 3.73632669e-01 -1.21180201e+00 -1.02366144e-02
6.49831891e-01 -3.94990414e-01 1.01914454e+00 -9.67221737e-01
6.68093383e-01 -2.53262408e-02 -5.32909513e-01 -1.18040776e+00
-3.68832678e-01 -8.35298225e-02 1.51209636e-02 1.37567914e+00
6.92633033e-01 -1.12951696e+00 4.01546001e-01 7.61936247e-01
-9.99228880e-02 -9.51806486e-01 -6.88890696e-01 -8.62436533e-01
3.74465227e-01 -5.02920508e-01 3.86720181e-01 1.32350302e+00
1.83284581e-01 7.59185374e-01 4.36345190e-02 -1.34757668e-01
1.52980164e-01 -3.65860790e-01 5.03402710e-01 -9.38200533e-01
-3.44880647e-03 -4.17083114e-01 -1.04015267e+00 -4.96938676e-01
2.11781248e-01 -1.10898614e+00 -7.27486014e-01 -1.67020905e+00
1.22451298e-01 -2.22056583e-01 -3.90426189e-01 3.77573729e-01
-2.79611081e-01 1.55653238e-01 -2.02895135e-01 7.44922385e-02
1.23248503e-01 6.56506956e-01 6.80119872e-01 -2.52647027e-02
-3.13092470e-02 -5.87461412e-01 -6.46258175e-01 5.53840339e-01
1.15488887e+00 -5.92288554e-01 -4.34653759e-01 -5.56638300e-01
5.40255308e-01 -4.65399176e-01 -1.61583036e-01 -2.60141969e-01
7.54081383e-02 7.95660540e-02 2.79376000e-01 -3.78545463e-01
3.37676436e-01 -7.60568917e-01 -2.78544337e-01 6.73735976e-01
-5.18085301e-01 6.58695281e-01 2.61105627e-01 3.82874846e-01
-4.71558720e-01 -7.46427357e-01 4.60886925e-01 2.54793435e-01
-7.71291018e-01 -1.79969504e-01 -1.72185868e-01 7.44719878e-02
1.26702905e+00 -5.20398796e-01 -2.53394872e-01 -1.98822722e-01
-4.97269779e-01 3.68430503e-02 -1.04290411e-01 8.77366364e-01
9.67083037e-01 -1.48891771e+00 -8.52971077e-01 5.17241508e-02
4.16926861e-01 -6.69528484e-01 -1.31694332e-01 6.30513132e-01
-7.79441833e-01 6.92159891e-01 -7.70060942e-02 -5.49537651e-02
-1.58078146e+00 3.71616989e-01 -2.98693538e-01 -5.51537126e-02
-4.35648978e-01 1.02290797e+00 -5.60311899e-02 -5.15891075e-01
-1.23465501e-01 4.95681539e-02 -9.71755862e-01 2.25364089e-01
5.58049440e-01 6.29088521e-01 -1.57065257e-01 -9.35082018e-01
-4.81454790e-01 8.87568295e-01 -2.22294554e-01 -2.89093912e-01
1.38397324e+00 1.13206938e-01 -4.61085171e-01 5.47699690e-01
1.74421775e+00 7.11450502e-02 2.19138265e-01 -1.99808195e-01
3.89042825e-01 -5.72335064e-01 -4.34694812e-02 -6.42786503e-01
-2.76655197e-01 9.76731300e-01 6.65040791e-01 4.01537955e-01
4.17320341e-01 4.45274711e-02 9.14879560e-01 6.57410383e-01
-1.37022868e-01 -1.32619190e+00 -2.46471781e-02 7.86726654e-01
8.51575375e-01 -1.26465535e+00 -4.95503098e-03 -8.25652182e-02
-7.69147813e-01 1.40339136e+00 2.47562870e-01 -3.83740872e-01
8.74986410e-01 -7.25321323e-02 1.30497336e-01 -3.81440222e-01
-7.88273036e-01 5.91723993e-02 5.88245451e-01 5.71217775e-01
1.07152784e+00 1.54982492e-01 -1.14695394e+00 7.19322503e-01
-5.95042765e-01 -8.87376964e-01 8.92987996e-02 7.95755982e-01
-6.68343306e-01 -1.15984011e+00 -2.01888725e-01 5.25457740e-01
-7.14399815e-01 -4.36196983e-01 -4.98239279e-01 8.69712114e-01
-1.64409876e-01 1.11967432e+00 1.10701509e-01 -3.35157156e-01
2.24783882e-01 2.03747943e-01 -3.52357849e-02 -8.20400238e-01
-8.53338361e-01 -4.04630631e-01 2.96147794e-01 -2.23707736e-01
-1.31165713e-01 -1.94952399e-01 -1.52383244e+00 -3.17758948e-01
-6.23054385e-01 7.87278771e-01 1.31647217e+00 8.33790123e-01
1.56154662e-01 4.47423220e-01 5.04123986e-01 1.46792799e-01
-6.13637209e-01 -1.40370595e+00 -3.94569039e-01 5.48403323e-01
-3.62736851e-01 -3.93896133e-01 -5.70904791e-01 -1.71308771e-01]
|
[10.477995872497559, 8.708149909973145]
|
9277c6d7-ae6d-445f-8009-ef339d591f67
|
learning-end-to-end-goal-oriented-dialog-with-2
|
1907.07638
| null |
https://arxiv.org/abs/1907.07638v1
|
https://arxiv.org/pdf/1907.07638v1.pdf
|
Learning End-to-End Goal-Oriented Dialog with Maximal User Task Success and Minimal Human Agent Use
|
Neural end-to-end goal-oriented dialog systems showed promise to reduce the workload of human agents for customer service, as well as reduce wait time for users. However, their inability to handle new user behavior at deployment has limited their usage in real world. In this work, we propose an end-to-end trainable method for neural goal-oriented dialog systems which handles new user behaviors at deployment by transferring the dialog to a human agent intelligently. The proposed method has three goals: 1) maximize user's task success by transferring to human agents, 2) minimize the load on the human agents by transferring to them only when it is essential and 3) learn online from the human agent's responses to reduce human agents load further. We evaluate our proposed method on a modified-bAbI dialog task that simulates the scenario of new user behaviors occurring at test time. Experimental results show that our proposed method is effective in achieving the desired goals.
|
['Jatin Ganhotra', 'Lazaros Polymenakos', 'Janarthanan Rajendran']
|
2019-07-17
|
learning-end-to-end-goal-oriented-dialog-with-3
|
https://aclanthology.org/Q19-1024
|
https://aclanthology.org/Q19-1024.pdf
|
tacl-2019-3
|
['goal-oriented-dialog']
|
['natural-language-processing']
|
[-2.45493278e-01 5.76859236e-01 5.74541569e-01 -7.88134575e-01
-6.74033821e-01 -7.56302655e-01 2.23824099e-01 -2.13283479e-01
-6.78396285e-01 8.23335767e-01 1.52953252e-01 -3.52471083e-01
-2.11141575e-02 -5.90024948e-01 -1.35364071e-01 -5.10163486e-01
2.61120014e-02 1.25207853e+00 3.11097413e-01 -8.01687062e-01
1.14765257e-01 3.36009771e-01 -1.06898415e+00 2.50791520e-01
7.07011342e-01 6.18711233e-01 5.27225375e-01 8.68847787e-01
-1.30071387e-01 1.12732863e+00 -8.26997519e-01 -1.26117215e-01
4.22118515e-01 -4.93726820e-01 -1.01077533e+00 2.59000301e-01
-2.70061225e-01 -9.46050227e-01 -2.62084454e-01 7.87418664e-01
7.87022650e-01 8.38176489e-01 3.91781479e-01 -1.39853358e+00
-3.74211252e-01 6.70631707e-01 -1.49646373e-02 -2.27548070e-02
2.80602843e-01 6.03309691e-01 6.48216248e-01 -5.10909617e-01
2.57492065e-01 1.51850235e+00 1.30388349e-01 1.20437574e+00
-8.22185457e-01 -3.01376134e-01 1.46636888e-01 -2.77092420e-02
-8.39042723e-01 -6.02245629e-01 6.22228265e-01 -3.49915504e-01
1.15658605e+00 1.11758761e-01 -6.78623617e-02 7.06712186e-01
-1.78165242e-01 7.45163023e-01 8.44845772e-01 -2.57645428e-01
4.61408854e-01 5.70464134e-01 5.32490849e-01 7.23608792e-01
-7.03397632e-01 -1.27489790e-01 -2.22819239e-01 -2.15222359e-01
3.59025121e-01 -1.20131634e-01 1.51613832e-01 2.15997681e-01
-8.70476782e-01 1.09703648e+00 2.38507763e-01 3.37167174e-01
-7.92650461e-01 -2.09896222e-01 4.00816768e-01 5.59761465e-01
3.45563650e-01 4.72915024e-01 -6.66141152e-01 -3.80130261e-01
-1.12431452e-01 3.98294747e-01 1.18143606e+00 1.30564415e+00
5.33545673e-01 3.01892817e-01 -3.21518272e-01 1.07196438e+00
1.21884175e-01 4.09403682e-01 5.58915317e-01 -9.96903360e-01
5.67608953e-01 8.11005116e-01 6.26326084e-01 -4.16287273e-01
-8.50668788e-01 1.74410626e-01 -4.76906091e-01 2.39885047e-01
4.82062429e-01 -8.94872785e-01 -7.60145843e-01 1.89536297e+00
3.34797114e-01 -4.02750611e-01 3.97463202e-01 1.00637734e+00
8.53755474e-01 8.84405732e-01 1.95684940e-01 -4.30398017e-01
1.38777590e+00 -9.51499522e-01 -8.02223504e-01 -4.28396553e-01
7.32757747e-01 -5.28796375e-01 1.36685610e+00 2.42612258e-01
-1.44958723e+00 -5.23919225e-01 -5.45701325e-01 2.98944235e-01
-1.87549874e-01 -8.49255174e-02 5.59156835e-01 7.50428677e-01
-1.17249537e+00 9.16923359e-02 -5.41915894e-01 -4.98284310e-01
-2.99397886e-01 8.98383677e-01 -8.60870071e-03 3.90659302e-01
-1.16241062e+00 1.02191246e+00 3.21066618e-01 9.54558700e-02
-1.09036684e+00 -9.85333845e-02 -7.42315829e-01 4.86248642e-01
6.51647627e-01 -4.01951343e-01 1.83521962e+00 -5.79930842e-01
-1.98081696e+00 4.30530876e-01 -1.77782774e-02 -4.89176065e-01
3.95310700e-01 1.01610452e-01 -5.84816523e-02 -2.67382711e-01
-1.80696771e-01 8.20232332e-01 2.18203247e-01 -1.09701347e+00
-1.03950119e+00 -4.06172037e-01 4.49908346e-01 6.44709826e-01
-4.86402601e-01 8.68341923e-02 -1.89511031e-01 4.45886515e-02
-2.43221700e-01 -1.19450831e+00 -2.73392797e-01 -7.84629643e-01
-9.68730897e-02 -7.70173728e-01 8.39167774e-01 -5.69000602e-01
7.36765027e-01 -2.05045605e+00 1.26307920e-01 -7.61746839e-02
2.13493630e-01 3.25340420e-01 -3.11710507e-01 5.58432043e-01
4.77050334e-01 -2.39983201e-01 1.09809816e-01 -4.21748161e-01
2.69341946e-01 1.96450263e-01 5.83749376e-02 -2.11506516e-01
-1.85393810e-01 6.60891593e-01 -6.18990779e-01 -2.97050357e-01
3.31465542e-01 -1.15151361e-01 -6.54318869e-01 1.15618443e+00
-4.29870367e-01 5.76213896e-01 -4.13121402e-01 2.33887702e-01
3.64798129e-01 1.36273205e-01 8.06013197e-02 2.58323729e-01
5.19824997e-02 2.48305053e-01 -7.19303429e-01 1.14924538e+00
-5.95169008e-01 1.89652026e-01 4.21980858e-01 -7.47226238e-01
1.16027856e+00 5.17717838e-01 2.41669416e-01 -8.35524321e-01
4.82323766e-01 -3.42803001e-01 5.23893714e-01 -6.12211764e-01
3.82679373e-01 -3.11223716e-01 -4.61548090e-01 6.83882773e-01
1.43559232e-01 6.41109049e-03 1.98406264e-01 2.32985765e-01
1.00497663e+00 -4.80089337e-01 1.17075056e-01 -7.23933652e-02
6.62203074e-01 9.84077156e-02 3.87522519e-01 9.18285608e-01
-5.16513586e-01 -7.53623024e-02 3.36593479e-01 -5.24809718e-01
-7.86417723e-01 -7.64921963e-01 9.38680053e-01 1.79735935e+00
-2.00629514e-02 3.78088087e-01 -1.13760173e+00 -8.31387460e-01
-4.22103494e-01 1.29013920e+00 -2.46131495e-01 -3.04485470e-01
-7.43601739e-01 -4.18471962e-01 2.59036034e-01 3.96509558e-01
7.55181193e-01 -1.65088558e+00 -7.42919743e-01 4.27803308e-01
-3.82653266e-01 -1.10886860e+00 -6.83411300e-01 1.49274185e-01
-5.86679101e-01 -5.35981476e-01 -6.43274844e-01 -1.07764602e+00
5.35979569e-01 3.12675714e-01 6.74302399e-01 8.52669254e-02
9.61125344e-02 3.62793058e-01 -3.61747295e-01 -4.17852879e-01
-7.41254508e-01 2.92451084e-01 3.07990015e-01 -2.05079034e-01
5.15745223e-01 -2.18502507e-01 -4.81850415e-01 8.17033052e-01
-5.02911806e-01 1.24177940e-01 2.28343681e-01 8.20563197e-01
-4.39741492e-01 1.30529061e-01 1.15828264e+00 -9.59768951e-01
1.56860065e+00 -3.37299556e-01 -6.24609053e-01 3.17684263e-01
-4.28652048e-01 2.99575571e-02 8.93711686e-01 -6.04676783e-01
-1.57831693e+00 2.71654222e-02 -2.44178906e-01 2.22817674e-01
-4.45047319e-01 3.23286414e-01 -7.84401745e-02 2.32417330e-01
5.69746912e-01 1.16628855e-01 1.32271230e-01 -2.58812636e-01
2.34388873e-01 1.21928978e+00 3.22143167e-01 -5.65002263e-01
4.85965788e-01 -3.24758589e-01 -5.85792899e-01 -7.88316071e-01
-4.35580939e-01 -5.69171190e-01 -3.01908255e-01 -4.12709832e-01
8.45780432e-01 -5.95306873e-01 -1.72101951e+00 4.85184669e-01
-1.19937921e+00 -7.61979997e-01 2.02139720e-01 2.06797540e-01
-7.06825852e-01 -4.35816422e-02 -7.67724216e-01 -1.35005343e+00
-7.65574694e-01 -1.35830677e+00 6.74489260e-01 5.77246070e-01
-3.70056629e-01 -1.03889084e+00 -2.45166600e-01 6.63688838e-01
6.97519898e-01 -3.90785575e-01 9.11503673e-01 -1.22004271e+00
-2.04469204e-01 -3.05645108e-01 -3.50566395e-02 2.49410287e-01
2.86924571e-01 -9.64590192e-01 -7.70321071e-01 -5.54776073e-01
3.64623785e-01 -7.58310676e-01 -3.00208926e-01 1.84935421e-01
1.92782864e-01 -6.31561816e-01 1.10222466e-01 -1.99124515e-01
8.16401958e-01 1.20117879e+00 2.48097092e-01 -8.35940149e-03
2.30106100e-01 1.01182199e+00 9.20235217e-01 5.62711895e-01
4.87680435e-01 8.05222988e-01 3.46476376e-01 -1.04852445e-01
3.53057235e-01 -7.60670006e-02 4.72720057e-01 6.42167389e-01
1.38473123e-01 -5.68845093e-01 -8.69850278e-01 4.06573623e-01
-2.23547077e+00 -5.56088448e-01 2.54428655e-01 2.01412868e+00
6.27576888e-01 2.56285042e-01 5.86515844e-01 -5.18142283e-01
6.73567712e-01 -3.43675196e-01 -6.64433122e-01 -5.95228195e-01
6.16596222e-01 -2.49755815e-01 1.77676544e-01 9.56895769e-01
-5.80844223e-01 1.40367007e+00 5.29479694e+00 1.57470837e-01
-7.62815654e-01 4.32875633e-01 6.24355435e-01 4.14312491e-03
4.45011586e-01 -1.72871411e-01 -8.95864069e-01 3.24219614e-01
1.00768256e+00 -2.49616668e-01 8.16996515e-01 1.03151727e+00
4.54059899e-01 -2.10492797e-02 -1.19530725e+00 6.14939332e-01
-1.05826959e-01 -6.82714105e-01 -3.20325553e-01 -1.14450336e-01
7.87480772e-02 -3.33470434e-01 -2.22962871e-01 9.82762456e-01
7.65448928e-01 -7.33781099e-01 1.85189277e-01 2.79408365e-01
2.17658222e-01 -8.66404653e-01 1.14945388e+00 1.03167152e+00
-6.66053236e-01 -2.55569190e-01 -2.89160967e-01 -2.65719295e-01
4.96236384e-01 -5.14352918e-01 -1.82515228e+00 -1.83172673e-01
3.05722713e-01 -4.97430950e-01 -1.18393160e-01 4.02692676e-01
6.30417764e-02 3.69377226e-01 -2.68050343e-01 -6.66395962e-01
5.16513646e-01 -1.86463177e-01 3.42355669e-01 9.50925708e-01
1.66704059e-02 8.02130878e-01 5.84863603e-01 6.20126307e-01
1.48278311e-01 6.82618767e-02 -7.29666054e-01 1.06861509e-01
3.50568354e-01 1.19894588e+00 -5.78113139e-01 -3.23388547e-01
-6.92731366e-02 1.07845545e+00 4.58763003e-01 3.02149683e-01
-8.08791399e-01 -4.35631394e-01 1.52739763e-01 -1.46695171e-02
-3.34296733e-01 -2.52708346e-01 7.20434710e-02 -5.20563185e-01
-2.48231992e-01 -1.02066374e+00 4.03244257e-01 -9.24697220e-01
-1.00730205e+00 1.04963720e+00 4.66191284e-02 -7.50816703e-01
-7.64540732e-01 -5.00928402e-01 -6.25596583e-01 9.85527575e-01
-5.14764309e-01 -1.02664101e+00 -3.47372264e-01 7.07900465e-01
1.28611648e+00 -6.53761804e-01 9.20564771e-01 2.78039485e-01
-3.92785102e-01 7.47585952e-01 -5.32854378e-01 -5.17886803e-02
6.24380589e-01 -1.11796415e+00 3.60818326e-01 4.05633211e-01
-5.70637286e-01 7.16147304e-01 8.82870793e-01 -6.77994490e-01
-1.35120511e+00 -6.75638139e-01 6.34999156e-01 -1.72740042e-01
2.75786012e-01 -5.04570305e-01 -7.64550805e-01 8.49220216e-01
5.70287585e-01 -8.78729701e-01 3.98770392e-01 1.96782961e-01
3.64883184e-01 -1.84053257e-02 -1.48696232e+00 9.05213356e-01
6.17559493e-01 -1.13179376e-02 -5.58189690e-01 7.20396996e-01
9.12448704e-01 -4.65667605e-01 -5.11189818e-01 9.90547836e-02
5.41157760e-02 -6.96947694e-01 6.59683645e-01 -8.69541287e-01
-6.26004636e-02 6.60209730e-02 2.55977772e-02 -1.52904022e+00
-3.36685985e-01 -9.21809614e-01 3.12433183e-01 1.20318651e+00
5.73695719e-01 -7.44684458e-01 8.49724650e-01 1.47910476e+00
-1.46600336e-01 -4.15231943e-01 -7.28067577e-01 -4.05784965e-01
-2.14750990e-01 1.47002339e-01 4.36394572e-01 5.13763487e-01
5.70622563e-01 9.39261496e-01 -8.50285172e-01 1.51353255e-01
1.74994439e-01 -3.65830690e-01 1.17344415e+00 -8.72772038e-01
-3.32125574e-01 -9.49159861e-02 5.33835851e-02 -1.21517193e+00
1.54370278e-01 -3.26329470e-01 6.39277458e-01 -1.27331555e+00
7.35764802e-02 -4.02486563e-01 1.32877812e-01 6.44563198e-01
-2.05205649e-01 -5.42012334e-01 4.03746367e-01 4.25057076e-02
-6.51544750e-01 3.88105810e-01 9.75496233e-01 -1.38269424e-01
-7.93738306e-01 5.32197535e-01 -5.57191432e-01 6.26181066e-01
1.11432886e+00 -3.45847011e-01 -9.19098377e-01 -2.67808825e-01
-2.82438397e-01 9.48238373e-01 -2.97062248e-01 -8.24181736e-01
5.73322713e-01 -4.01878327e-01 -2.89351821e-01 -3.71418566e-01
6.07125461e-01 -9.79145110e-01 -3.00617605e-01 5.49596786e-01
-5.51385880e-01 3.09289604e-01 3.23458135e-01 1.51971802e-01
1.01619184e-01 -6.51813626e-01 8.43805492e-01 -3.29418033e-01
-5.28002977e-01 -2.74596065e-02 -9.03578222e-01 -3.01034570e-01
1.27069867e+00 1.68647394e-01 -4.14448529e-01 -1.21429014e+00
-9.23951805e-01 8.28706026e-01 -7.34835863e-02 5.17780066e-01
5.30813634e-01 -6.66096747e-01 -5.63239753e-01 -5.94236329e-02
-5.08615747e-02 -2.41800740e-01 3.53547961e-01 3.16420704e-01
-4.01863843e-01 5.47776997e-01 -4.46011871e-01 -2.48476416e-01
-1.64157987e+00 5.06345928e-01 4.35300589e-01 -4.60418135e-01
-2.86367655e-01 7.35829115e-01 5.20985007e-01 -9.07954216e-01
7.58698165e-01 2.54138522e-02 -3.62269580e-01 -2.46452153e-01
4.69898909e-01 2.80486465e-01 -6.94567040e-02 -3.03496689e-01
-3.59358750e-02 -2.29197443e-01 -5.71337342e-01 -6.38062418e-01
1.24683380e+00 -2.07921237e-01 1.86130643e-01 3.05016935e-01
7.14954853e-01 -5.09032011e-01 -1.04755640e+00 -1.52450383e-01
6.57292735e-03 -1.52179554e-01 -3.80798250e-01 -1.14908671e+00
-6.74246013e-01 7.88967848e-01 8.19067419e-01 6.42977476e-01
9.99306142e-01 -1.93962336e-01 1.10139453e+00 1.01437223e+00
6.50351882e-01 -1.19029486e+00 2.39502862e-01 7.64752328e-01
8.27051163e-01 -1.54291630e+00 -5.57319522e-01 -2.14648396e-01
-1.26186037e+00 8.01635444e-01 1.30850494e+00 2.64485568e-01
2.40059182e-01 1.43360555e-01 3.91769886e-01 -3.47751886e-01
-9.67512608e-01 -6.57275468e-02 -2.67134547e-01 6.76316261e-01
2.32638463e-01 1.55069008e-01 -1.56981885e-01 6.99314654e-01
-2.07877144e-01 -1.55181721e-01 5.90538323e-01 9.73879814e-01
-7.64930546e-01 -8.68636191e-01 -2.42468834e-01 1.88670188e-01
-1.14059925e-01 7.37098455e-02 -6.53931618e-01 5.98402500e-01
-4.08352166e-01 1.49655831e+00 -2.39729255e-01 -4.26953256e-01
8.86124790e-01 4.99153614e-01 6.49644136e-02 -7.52171099e-01
-1.01634991e+00 6.16732351e-02 4.75350171e-01 -9.90497395e-02
1.85547739e-01 -8.37287828e-02 -1.37963152e+00 -3.41616720e-01
-2.95161694e-01 4.02302653e-01 6.97226942e-01 9.85170603e-01
2.44689435e-01 4.36990857e-01 9.18492436e-01 -6.97941601e-01
-8.94094467e-01 -1.59059274e+00 -2.81950086e-01 6.03056490e-01
-2.42528692e-02 -4.47369128e-01 -2.81628847e-01 1.21948645e-02]
|
[12.877631187438965, 8.011702537536621]
|
e272e176-d153-4423-8d02-531358b6c37e
|
deep-denoising-prior-based-spectral
|
2306.17096
| null |
https://arxiv.org/abs/2306.17096v1
|
https://arxiv.org/pdf/2306.17096v1.pdf
|
Deep Denoising Prior-Based Spectral Estimation for Phaseless Synthetic Aperture Radar
|
Incoherent processing for synthetic aperture radar (SAR) is a promising approach that enables low implementation costs, simplified hardware designs and operations in high frequency spectrum compared to the conventional imaging methods using coherent processing. Existing non-convex phaseless imaging algorithms offer recovery guarantees over limited range of forward models. In recent years, several deep learning (DL) based techniques have been introduced with the goal of extending applicability of phaseless imaging techniques to wave-based imaging modalities by addressing fundamental challenges, such as, lack of redundancy, non-uniqueness issues encountered commonly with inverse scattering models. In this paper, we introduce a DL-based phaseless SAR imaging approach that is designed under the premise that the spectral estimation technique, widely used for initializing non-convex phase retrieval algorithms, has significance far beyond generating good initial points. We extend the iterative power method for spectral estimation by using deep denoisers at each iteration, and subsequently design a deep imaging network within the plug-and-play framework. Finally, we verify the feasibility of our approach using synthetic SAR measurements.
|
['Birsen Yazıcı', 'Bariscan Yonel', 'Samia Kazemi']
|
2023-06-29
| null | null | null | null |
['retrieval']
|
['methodology']
|
[ 7.71857738e-01 -2.31109008e-01 6.17566466e-01 -3.37789536e-01
-1.15794647e+00 -2.14743093e-01 3.95411015e-01 -5.78969955e-01
-2.64439851e-01 6.81566119e-01 3.36589158e-01 -1.64120737e-02
-9.14424419e-01 -6.28652990e-01 -3.94056410e-01 -1.19401109e+00
-4.39887762e-01 4.81630474e-01 -3.48500967e-01 -1.18061803e-01
-6.63264990e-02 6.11764967e-01 -1.15937006e+00 -3.07862666e-02
7.42888987e-01 1.10376072e+00 2.20443144e-01 4.93661970e-01
6.21864498e-01 7.53359377e-01 -1.82478845e-01 2.25276023e-01
7.90380836e-01 -2.94061393e-01 -1.76292300e-01 6.14664108e-02
3.30945045e-01 -4.84926641e-01 -4.00352091e-01 1.11820412e+00
8.86124551e-01 5.67738488e-02 5.76434374e-01 -8.81113052e-01
-3.76091301e-01 4.63294268e-01 -8.42349231e-01 1.46092504e-01
9.69702080e-02 7.79926032e-02 7.09300995e-01 -1.10627389e+00
2.45328382e-01 8.41514528e-01 1.15944231e+00 -1.26092052e-02
-1.03616393e+00 -4.61166382e-01 -4.06085610e-01 2.83445001e-01
-1.42420220e+00 -8.53949964e-01 9.58261132e-01 -2.54627019e-01
8.24571073e-01 2.43753284e-01 6.69249058e-01 6.72774494e-01
1.31957814e-01 5.22445381e-01 1.43331981e+00 -6.16447866e-01
2.95658976e-01 -7.40984142e-01 1.33655876e-01 6.04506433e-01
4.68874156e-01 4.84057397e-01 -5.75830460e-01 -1.97659120e-01
7.10369527e-01 -3.54800701e-01 -7.81388044e-01 -5.89859903e-01
-1.33180010e+00 8.14672232e-01 3.74322981e-01 2.49452516e-01
-7.79198408e-01 3.38507921e-01 -1.34827226e-01 3.58547777e-01
6.22693658e-01 6.08103573e-01 -3.67910445e-01 4.85563189e-01
-1.58952224e+00 3.11493069e-01 7.83984303e-01 5.63489139e-01
7.17062175e-01 7.67041385e-01 -5.84512241e-02 4.57549721e-01
7.87550807e-01 9.34089839e-01 2.28227809e-01 -9.96873915e-01
-1.08887032e-02 -3.15903783e-01 2.53656656e-01 -7.20187008e-01
-6.33447230e-01 -1.25116372e+00 -1.29350269e+00 2.06373230e-01
-1.27968201e-02 -5.58523655e-01 -9.94399786e-01 1.61523926e+00
3.11884612e-01 7.23605454e-01 6.22524619e-01 1.14436924e+00
6.05670929e-01 7.23631144e-01 -2.45171368e-01 -7.33977795e-01
1.31210196e+00 -6.64406419e-01 -8.79634917e-01 -5.83695769e-01
2.24866837e-01 -1.08099842e+00 3.13137531e-01 6.52337015e-01
-1.07365835e+00 -2.74392962e-01 -1.35324967e+00 3.10479522e-01
3.20326030e-01 4.18285802e-02 7.61990368e-01 6.85309947e-01
-1.21913433e+00 4.42568392e-01 -8.50562215e-01 1.87511101e-01
4.37535226e-01 2.06012964e-01 -5.66903241e-02 -3.78799587e-01
-1.12474656e+00 7.52190948e-01 6.56570345e-02 6.15867555e-01
-8.94169688e-01 -1.14794064e+00 -7.89422214e-01 -9.52063873e-03
7.97670558e-02 -1.14594066e+00 9.85518754e-01 -1.02295589e+00
-1.42458367e+00 5.93405604e-01 1.01706840e-01 -8.47071350e-01
2.16175854e-01 -5.41134059e-01 -4.50095147e-01 4.15808111e-01
3.22120003e-02 2.13275492e-01 1.57201219e+00 -1.15306389e+00
-8.12883303e-02 -2.79442072e-01 -4.43219692e-01 3.91383082e-01
3.43175262e-01 -3.81881356e-01 4.29591328e-01 -6.52211964e-01
8.55585873e-01 -6.32652104e-01 -5.77551425e-01 -3.32629401e-03
-7.87974522e-02 5.98010361e-01 7.20493019e-01 -7.55116701e-01
6.37126923e-01 -2.11179829e+00 1.31706640e-01 2.30068982e-01
1.68297097e-01 1.20920226e-01 -2.43144706e-01 4.29234713e-01
-4.40756679e-01 -5.90748608e-01 -6.48952246e-01 -1.71431899e-01
-1.52997896e-01 -6.47080168e-02 -3.64774883e-01 9.87640321e-01
1.61789045e-01 7.49288142e-01 -4.92088705e-01 -5.71397915e-02
2.16771781e-01 5.51935911e-01 -6.21883571e-01 9.99116674e-02
-1.60710216e-01 6.47938430e-01 -4.24606383e-01 6.90387309e-01
1.31912649e+00 -3.29586148e-01 6.62525222e-02 -9.32204902e-01
-5.34041882e-01 -1.92368239e-01 -1.50998664e+00 1.71538770e+00
-6.07818186e-01 5.98616183e-01 8.42395127e-01 -1.49771416e+00
5.69849730e-01 4.05602962e-01 9.24928069e-01 -7.71502316e-01
8.51091295e-02 4.35714424e-01 2.43710637e-01 -4.58945245e-01
1.35042354e-01 -6.24806345e-01 2.17227742e-01 5.47982991e-01
-1.31196305e-01 -4.26636934e-01 -3.30444843e-01 -1.75092041e-01
1.16097534e+00 1.45086408e-01 5.84637344e-01 -4.10878867e-01
5.25537908e-01 2.94529766e-01 3.89975518e-01 6.71110690e-01
6.40969947e-02 8.19521308e-01 -4.85803336e-01 -3.96617353e-01
-9.42794561e-01 -1.18430567e+00 -3.63911837e-01 4.35181230e-01
-2.09198073e-02 3.48957419e-01 -3.05054456e-01 3.55142534e-01
-5.26555061e-01 4.12545592e-01 -1.59339998e-02 2.57736444e-01
-8.74670684e-01 -1.38608289e+00 2.29860350e-01 -1.11313730e-01
1.04321623e+00 -5.00400424e-01 -8.98260832e-01 4.82850373e-01
-3.33814442e-01 -1.24562192e+00 1.21182770e-01 3.13478440e-01
-9.55357969e-01 -9.25115883e-01 -9.21381950e-01 -7.28594899e-01
3.80025774e-01 6.22928858e-01 1.05788684e+00 -4.16900814e-01
-4.53247875e-01 5.48296034e-01 -3.84108692e-01 -2.32828155e-01
2.57787090e-02 -4.62101132e-01 7.41805509e-02 2.19944313e-01
-1.86805829e-01 -1.12510800e+00 -9.37054813e-01 -2.64583945e-01
-9.13982749e-01 1.03109010e-01 1.04344964e+00 1.08239472e+00
5.99066198e-01 1.56723201e-01 4.87685353e-01 -5.83808482e-01
4.74965096e-01 -3.67003024e-01 -7.32942343e-01 5.12857884e-02
-4.99578804e-01 1.68489501e-01 3.58450592e-01 -1.83590636e-01
-1.11856234e+00 2.24852681e-01 -4.24653560e-01 -2.81794846e-01
1.61916330e-01 9.67565000e-01 -2.44731363e-03 -6.78319693e-01
7.43970513e-01 5.54437518e-01 8.94690454e-02 -2.87860096e-01
2.34196201e-01 4.10535336e-01 4.81316954e-01 -1.10183224e-01
1.42273009e+00 8.50753844e-01 5.65078378e-01 -1.20434844e+00
-1.18375325e+00 -5.87618768e-01 -2.57858187e-01 -1.16506040e-01
5.79691589e-01 -1.42575729e+00 -2.78245598e-01 5.26969552e-01
-8.85699749e-01 -1.52439430e-01 -3.70516658e-01 9.04795408e-01
-7.27784455e-01 7.27265418e-01 -2.64110953e-01 -6.96920991e-01
-8.30047905e-01 -8.57514143e-01 1.21272910e+00 3.45844515e-02
5.93989827e-02 -9.41417158e-01 3.42675388e-01 4.18343633e-01
8.18908691e-01 2.43670419e-01 6.67651296e-01 -1.44210294e-01
-1.02223051e+00 1.30050123e-01 -2.59091675e-01 2.65550792e-01
-1.79379687e-01 -9.23486054e-01 -1.02116287e+00 -6.60065413e-01
7.17384934e-01 -2.16653690e-01 7.11833477e-01 1.10500872e+00
5.52480638e-01 -2.82862872e-01 -9.58093852e-02 1.29624319e+00
2.07808614e+00 -8.26838687e-02 8.22179139e-01 2.56633192e-01
1.98508725e-01 2.08603159e-01 5.92161655e-01 5.48864543e-01
-2.76651114e-01 3.14478010e-01 4.88583088e-01 -2.43696362e-01
-2.13324085e-01 5.38716018e-01 1.32346585e-01 6.99429572e-01
8.22958052e-02 -3.55469763e-01 -6.54045522e-01 4.63920563e-01
-1.63652527e+00 -1.03254819e+00 -3.61936450e-01 1.90710330e+00
4.30053681e-01 -3.82351488e-01 -8.22636187e-01 1.54786989e-01
8.08949396e-02 6.28265798e-01 -3.90831620e-01 3.39607477e-01
-3.03990513e-01 9.43677545e-01 7.68099546e-01 6.63290620e-01
-1.19974923e+00 5.15258968e-01 6.30859041e+00 6.54977977e-01
-1.13827622e+00 3.72558951e-01 9.91987586e-02 1.82896331e-01
-3.52077961e-01 4.43000495e-02 -2.88528770e-01 -6.50587380e-02
6.84286952e-01 -7.51023144e-02 3.69525999e-01 3.11224371e-01
5.16112924e-01 -1.00754760e-01 -7.20567584e-01 1.16408360e+00
1.57626629e-01 -1.57006788e+00 -2.43398294e-01 -1.68277964e-01
7.23785400e-01 2.63186276e-01 1.98730648e-01 -6.17128424e-02
1.52768418e-01 -9.60978806e-01 5.09082675e-01 6.55891597e-01
6.17094338e-01 -5.23906529e-01 7.01840341e-01 4.65618998e-01
-8.71256173e-01 2.79302001e-02 -1.03679568e-01 -2.80133903e-01
6.17015243e-01 1.13554728e+00 -6.84312642e-01 9.73168254e-01
4.54368800e-01 6.10084236e-01 1.10296629e-01 1.21695840e+00
-2.99292672e-02 6.12440944e-01 -4.93457943e-01 5.83763063e-01
3.68504465e-01 -4.34362262e-01 9.50648010e-01 8.37087631e-01
9.66430366e-01 5.19501925e-01 2.50920743e-01 7.25656927e-01
4.23419386e-01 -3.30632359e-01 -5.50478876e-01 2.65132517e-01
2.54532099e-01 1.40234590e+00 -3.14866066e-01 7.25991502e-02
-3.55176955e-01 5.08772254e-01 -6.22762501e-01 5.04209816e-01
-5.96457481e-01 1.05823934e-01 4.12829876e-01 3.87827218e-01
3.65932822e-01 -6.03614151e-01 -9.60496739e-02 -1.03719282e+00
-1.94056004e-01 -9.95146453e-01 1.32279962e-01 -1.20179260e+00
-1.23301947e+00 5.21161497e-01 7.91607723e-02 -1.66046488e+00
-4.19606715e-01 -4.70718235e-01 -3.83361518e-01 1.00780010e+00
-2.18990254e+00 -1.55127811e+00 -5.31677604e-01 5.07450402e-01
5.23639679e-01 -2.38287836e-01 7.91591108e-01 3.40106398e-01
7.72190616e-02 -2.19531059e-01 2.02740744e-01 -1.53074324e-01
5.22326291e-01 -6.80281937e-01 -8.94486830e-02 1.47547889e+00
-6.28920570e-02 3.97840232e-01 1.14925063e+00 -4.21046615e-01
-1.89548159e+00 -1.03701389e+00 2.96083987e-01 5.20250380e-01
7.39423335e-01 3.39485295e-02 -5.18125355e-01 5.90757310e-01
5.55640101e-01 1.08077288e-01 7.26405561e-01 -3.92529488e-01
1.28164306e-01 -1.73857316e-01 -1.13441288e+00 2.92218238e-01
8.05739343e-01 -3.68487462e-02 -5.52693129e-01 6.75233424e-01
4.04841304e-01 -3.64735007e-01 -6.03504181e-01 8.85649860e-01
3.04783970e-01 -9.05415475e-01 1.53150415e+00 1.61599323e-01
1.81484252e-01 -5.02763212e-01 -5.65241158e-01 -1.44883168e+00
-7.12840557e-01 -9.22847569e-01 -1.08743258e-01 4.88872588e-01
1.51584536e-01 -7.56840110e-01 8.84573400e-01 -4.27254885e-01
-7.10902810e-01 -1.70272246e-01 -1.13937759e+00 -5.44255495e-01
-2.77170241e-01 -4.31325465e-01 5.90815581e-02 9.03981864e-01
-8.61128092e-01 4.01406676e-01 -6.66249156e-01 1.14773822e+00
1.54015148e+00 5.02226532e-01 3.04744482e-01 -1.33703470e+00
-7.45316923e-01 1.91920146e-01 -1.67780407e-02 -1.25506330e+00
-2.26901826e-02 -7.23891795e-01 3.35979223e-01 -1.57705843e+00
-1.33548081e-01 -5.54833233e-01 2.30393440e-01 -1.80377886e-02
4.23871011e-01 5.35978734e-01 -2.01695517e-01 4.66243297e-01
-1.28164485e-01 4.70242321e-01 1.28185809e+00 -3.20956200e-01
2.04336315e-01 1.64537221e-01 -5.11583865e-01 8.50670576e-01
5.91334939e-01 -3.25613588e-01 -5.42704701e-01 -9.15075600e-01
5.82817614e-01 5.09444237e-01 6.54478371e-01 -1.47455120e+00
6.19884193e-01 1.25907719e-01 3.54767710e-01 -7.40477622e-01
6.64990604e-01 -1.06534779e+00 6.57599866e-01 6.83442831e-01
2.43917271e-01 -2.85683483e-01 -3.46402861e-02 6.05639398e-01
-5.08854568e-01 -4.51184392e-01 1.09667444e+00 -3.92035902e-01
-9.07805145e-01 3.50289583e-01 -4.74437565e-01 -5.62042668e-02
7.35935748e-01 -3.23305905e-01 -3.99634838e-02 -4.80971545e-01
-8.60754967e-01 -1.17160581e-01 -1.39371976e-01 -4.96924162e-01
8.03868890e-01 -9.30379152e-01 -1.09161758e+00 5.35326779e-01
-2.32573286e-01 -1.35661170e-01 7.39622295e-01 1.08462417e+00
-9.87674952e-01 2.76046008e-01 -1.88948810e-01 -6.36785984e-01
-1.11104190e+00 1.68646667e-02 6.34441197e-01 -4.77273434e-01
-7.19649792e-01 7.64955342e-01 8.70753452e-02 -2.24235758e-01
-2.63773054e-01 -1.46614268e-01 -7.57790208e-02 1.13527685e-01
4.69207764e-01 1.18134148e-01 3.69519055e-01 -4.16345894e-01
-2.91030139e-01 8.53263557e-01 3.84272873e-01 -5.02447784e-01
1.90821540e+00 -5.88122010e-02 -2.01254368e-01 -1.96868986e-01
9.40078259e-01 1.29190460e-01 -1.18448603e+00 -2.93395281e-01
-2.68273115e-01 -1.93409458e-01 6.97874188e-01 -6.54513121e-01
-9.89968956e-01 7.77427435e-01 1.09576714e+00 4.60413620e-02
1.59314692e+00 -4.23395097e-01 7.17513680e-01 4.44997698e-01
4.62479144e-01 -6.58362150e-01 1.14317037e-01 5.82015336e-01
7.62288094e-01 -1.09965193e+00 5.96661806e-01 -5.97786069e-01
1.39651462e-01 1.00382161e+00 -2.24467635e-01 -4.70517427e-01
1.03561223e+00 7.32217133e-01 6.04616851e-03 -5.36301255e-01
-5.11137962e-01 -2.41651937e-01 1.08058922e-01 8.94391835e-01
3.31892371e-01 -7.59856179e-02 -3.27812850e-01 1.07479125e-01
-1.88966617e-01 1.02345040e-02 5.64247847e-01 9.81409311e-01
-7.54942358e-01 -9.45196807e-01 -6.76451206e-01 2.94937581e-01
-3.26478034e-01 -5.64166546e-01 4.27553028e-01 7.81956553e-01
-7.28027746e-02 6.65589690e-01 -2.04890464e-02 2.47930959e-01
8.69620293e-02 -1.86225861e-01 7.60009766e-01 -4.87345994e-01
-2.45770589e-01 5.13930857e-01 2.50021845e-01 -2.63143122e-01
-8.90352786e-01 -7.47593641e-01 -7.10094988e-01 2.31673613e-01
-2.41824105e-01 9.80982035e-02 4.77572531e-01 1.00645387e+00
1.72435686e-01 5.98721683e-01 5.77048481e-01 -9.30973351e-01
-8.78410876e-01 -9.08713698e-01 -6.90183103e-01 -3.19928199e-01
6.02076232e-01 -3.97992253e-01 -5.39427936e-01 -4.30011004e-03]
|
[10.555087089538574, -2.1955349445343018]
|
e1d4b62f-d4cd-421c-90c1-42d31e90f312
|
deep-learning-for-ecg-classification
| null | null |
http://doi.org/10.1088/1742-6596/913/1/012004
|
http://iopscience.iop.org/article/10.1088/1742-6596/913/1/012004/pdf
|
Deep Learning for ECG Classification
|
The importance of ECG classification is very high now due to many current medical applications where this problem can be stated. Currently, there are many machine learning (ML) solutions which can be used for analyzing and classifying ECG data. However, the main disadvantages of these ML results is use of heuristic hand-crafted or engineered features with shallow feature learning architectures. The problem relies in the possibility not to find most appropriate features which will give high classification accuracy in this ECG problem. One of the proposing solution is to use deep learning architectures where first layers of convolutional neurons behave as feature extractors and in the end some fully-connected (FCN) layers are used for making final decision about ECG classes. In this work the deep learning architecture with 1D convolutional layers and FCN layers for ECG classification is presented and some classification results are showed.
|
['Natasha Kazachenko', 'Nick Mikhailovsky', 'Boris Pyakillya']
|
2017-01-01
| null | null | null |
journal-of-physics-conference-series-2017-1
|
['ecg-classification', 'electrocardiography-ecg']
|
['medical', 'methodology']
|
[ 2.49576956e-01 2.25329310e-01 1.86188668e-01 -6.37380421e-01
-1.40591100e-01 -2.43764855e-02 3.43441188e-01 5.16280711e-01
-6.11199200e-01 7.87580311e-01 -1.46129504e-01 -2.08514646e-01
-5.29897392e-01 -8.91910732e-01 -1.75184488e-01 -5.99345505e-01
-3.21750551e-01 3.48000556e-01 1.85548469e-01 -1.78699896e-01
2.03825250e-01 8.41808796e-01 -1.52730262e+00 5.82129896e-01
6.10106587e-01 1.21979296e+00 2.13789120e-02 9.12590981e-01
-2.60871798e-01 9.55258250e-01 -6.02442741e-01 1.37177194e-02
1.74552843e-01 -7.46003985e-01 -7.48876870e-01 -2.48353899e-01
-2.81570405e-01 1.63313339e-03 3.20313513e-01 6.77144945e-01
7.81203687e-01 -2.63226867e-01 8.34501743e-01 -9.11648929e-01
9.96958390e-02 5.38861632e-01 -1.95811272e-01 4.08904672e-01
1.90874174e-01 -1.46064326e-01 5.43167889e-01 -6.36932850e-01
2.54236251e-01 8.28070819e-01 8.46540451e-01 3.72162282e-01
-9.33920622e-01 -2.40113392e-01 -4.00897861e-01 4.12500143e-01
-1.23296130e+00 -9.55656394e-02 8.48821878e-01 -5.13735056e-01
8.33747089e-01 3.04235548e-01 7.87813246e-01 7.16766179e-01
4.54765379e-01 3.76974553e-01 1.24277747e+00 -4.95365292e-01
2.17769921e-01 4.28972363e-01 3.60170394e-01 6.08618021e-01
3.23392034e-01 6.23802748e-03 -8.65971074e-02 -1.73062027e-01
4.57239151e-01 2.08640605e-01 -1.89423099e-01 7.61251058e-03
-6.63390458e-01 7.43496835e-01 5.24266183e-01 9.69049573e-01
-7.15712368e-01 -1.10804467e-02 6.75301909e-01 4.20010805e-01
9.27295387e-02 5.55183053e-01 -6.47084177e-01 -1.03561513e-01
-9.56773996e-01 2.17894852e-01 6.90063000e-01 2.47989118e-01
5.60665548e-01 1.32863000e-01 3.03680003e-02 6.84861422e-01
2.11225629e-01 1.46196246e-01 7.52952516e-01 -3.79761904e-01
-4.25520614e-02 9.97102082e-01 -3.75468552e-01 -1.13362694e+00
-8.05029452e-01 -7.50440001e-01 -1.01822519e+00 6.65487766e-01
3.42119277e-01 -4.11503226e-01 -8.27340961e-01 1.08820879e+00
1.34159308e-02 -1.02834642e-01 2.27765143e-01 6.76984906e-01
8.87332976e-01 4.33935404e-01 2.48792954e-02 -9.53062177e-02
1.39983428e+00 -1.97713450e-01 -7.15248466e-01 1.93276271e-01
6.52927637e-01 -5.63982964e-01 6.80818975e-01 6.20939076e-01
-7.27425873e-01 -8.89088452e-01 -1.23117304e+00 1.43055782e-01
-6.76424444e-01 2.81144768e-01 4.29817140e-01 6.85885489e-01
-8.70572984e-01 9.75777328e-01 -6.27432883e-01 -3.67305070e-01
3.92170966e-01 7.69751191e-01 -3.47983718e-01 2.17481032e-01
-1.13807070e+00 1.09925830e+00 8.44342411e-01 6.58144236e-01
-5.44699788e-01 -1.66652247e-01 -5.43785870e-01 2.33358502e-01
-6.80646077e-02 -4.62112367e-01 8.07041943e-01 -1.35207379e+00
-1.23614681e+00 7.79372334e-01 3.07675391e-01 -8.76669347e-01
6.72665358e-01 2.37148046e-03 -4.24553901e-01 3.26640368e-01
-3.82670552e-01 4.11941797e-01 7.64771044e-01 -9.79476631e-01
-5.97619951e-01 -3.36615831e-01 -1.28566906e-01 -1.44318491e-01
-4.58696276e-01 -1.48010358e-01 3.67386818e-01 -3.07092935e-01
2.84823030e-01 -7.08281755e-01 -3.69979769e-01 -2.95008689e-01
-3.25822949e-01 -2.82237977e-01 9.36166525e-01 -6.30605817e-01
1.20193982e+00 -2.01065612e+00 -9.15742517e-02 3.57818276e-01
2.55848914e-01 6.13308907e-01 2.10854411e-01 3.46774459e-01
-2.58225650e-01 -9.37092826e-02 -1.28840446e-01 1.77229524e-01
-3.16890776e-01 3.12524319e-01 1.61352620e-01 2.44917691e-01
3.36729974e-01 6.31574750e-01 -4.88015264e-01 -6.45722091e-01
5.37670493e-01 6.46684587e-01 -3.16509545e-01 2.36016631e-01
5.46888933e-02 5.10909021e-01 -6.33470535e-01 2.17439100e-01
5.07294595e-01 1.00961789e-01 2.64991522e-01 -3.91666442e-01
-8.64107758e-02 -4.66534719e-02 -1.18941832e+00 1.27483368e+00
-2.81489551e-01 4.81299222e-01 -4.44992930e-01 -1.44113314e+00
1.19493175e+00 6.84298933e-01 5.85819304e-01 -4.54617381e-01
6.71528935e-01 5.34460366e-01 5.24464488e-01 -9.07041192e-01
-1.75610796e-01 -3.30559462e-01 2.48243436e-01 8.28944966e-02
1.67504132e-01 2.29462937e-01 6.45413697e-02 -3.78190339e-01
1.06225097e+00 3.21039632e-02 4.73117501e-01 -3.61676186e-01
9.37016666e-01 -2.04774007e-01 6.56674504e-01 4.83386070e-01
-8.98642614e-02 6.57208502e-01 8.64168167e-01 -1.15924466e+00
-7.50856280e-01 -4.29198712e-01 -3.52126807e-01 2.47419059e-01
-4.46800739e-01 -2.17746124e-01 -7.46680439e-01 -7.00437367e-01
-2.43261889e-01 2.34151915e-01 -7.68448353e-01 -2.11509168e-01
-6.56838059e-01 -7.96724200e-01 5.41336119e-01 4.25452739e-01
6.91945970e-01 -1.50855815e+00 -1.46188486e+00 4.50478196e-01
2.48175994e-01 -6.60673678e-01 6.38884246e-01 6.91495657e-01
-1.34884036e+00 -1.18211985e+00 -7.04920411e-01 -5.93653679e-01
4.76033866e-01 -5.12483120e-01 8.13112140e-01 3.91525865e-01
-6.83784246e-01 -2.21206412e-01 -4.86840695e-01 -7.74787486e-01
-6.03002250e-01 4.36297357e-01 -3.42666060e-01 1.99760884e-01
4.50742513e-01 -7.20063806e-01 -6.41956508e-01 -3.33190084e-01
-7.21013606e-01 -1.56964973e-01 9.60736334e-01 6.71624005e-01
2.87414193e-01 7.43547408e-03 8.19344580e-01 -1.12194765e+00
6.15364313e-01 -1.79762289e-01 -2.77076125e-01 6.44211024e-02
-7.12974489e-01 2.14114353e-01 8.06190610e-01 5.00831716e-02
-4.51261342e-01 1.77462026e-01 -7.90822208e-01 -2.71256864e-02
-5.15426278e-01 4.66403395e-01 1.64017472e-02 -8.81482661e-02
8.49201083e-01 1.13504373e-01 1.47956595e-01 -5.76882541e-01
-3.71393204e-01 7.60733545e-01 4.35724556e-02 -1.98256880e-01
2.75936902e-01 2.39748918e-02 4.81729448e-01 -8.95346582e-01
-6.57720506e-01 -1.11792974e-01 -8.93713057e-01 -3.56572747e-01
1.24904692e+00 -2.92641103e-01 -7.84003258e-01 3.64165306e-01
-9.68037069e-01 5.83056621e-02 -3.92629504e-01 4.82478023e-01
-4.27061349e-01 1.61497235e-01 -4.01146829e-01 -1.07855797e+00
-5.88812053e-01 -1.08393991e+00 4.54094797e-01 4.11122650e-01
-3.35418582e-01 -8.82802546e-01 -1.99717119e-01 -7.16849267e-02
5.78671992e-01 7.76739359e-01 1.27556312e+00 -8.47837567e-01
-8.53878818e-03 -5.40752769e-01 2.64708269e-02 7.32703328e-01
9.86897871e-02 -1.90664127e-01 -1.15499878e+00 -8.39829296e-02
2.62786210e-01 -2.57500112e-01 7.10707486e-01 6.71174347e-01
1.10650730e+00 -8.54335055e-02 -2.37291828e-01 4.17727321e-01
1.66369486e+00 4.34378445e-01 7.44501531e-01 2.41489440e-01
2.42617235e-01 5.56629241e-01 3.06634128e-01 3.96302670e-01
-1.16313502e-01 3.50956202e-01 4.76899356e-01 -4.28550452e-01
1.44165307e-02 2.68721551e-01 -1.65172964e-01 5.68421423e-01
-5.42275131e-01 1.10068269e-01 -9.31357861e-01 3.73374254e-01
-1.73541546e+00 -9.19439197e-01 -4.88724381e-01 2.12136364e+00
4.18176293e-01 5.37030399e-01 2.20174760e-01 1.02975786e+00
3.92700285e-01 -2.57986546e-01 -1.39484093e-01 -7.84137368e-01
1.74767524e-03 6.15664482e-01 9.39599574e-02 1.49153456e-01
-1.22004032e+00 1.71279013e-01 5.57535505e+00 1.78354204e-01
-1.45347571e+00 -4.65307981e-02 6.74754679e-01 3.09228301e-01
3.89470160e-01 -1.91953495e-01 -3.19772840e-01 2.62374818e-01
1.04344499e+00 2.88603127e-01 -1.95839271e-01 7.44080245e-01
3.10604155e-01 -1.27838001e-01 -8.71790051e-01 1.05159760e+00
-7.32706487e-02 -1.06224358e+00 -1.96603630e-02 -4.91538085e-02
3.08981597e-01 -2.75618911e-01 -2.45973617e-01 1.20202288e-01
-5.22460639e-01 -1.14055777e+00 2.75528520e-01 7.18023777e-01
4.47158545e-01 -1.02379191e+00 1.35085785e+00 3.06718141e-01
-8.68809640e-01 -5.42952657e-01 -4.99849409e-01 -3.08446705e-01
-1.38469428e-01 5.99730790e-01 -9.02821362e-01 7.03786969e-01
7.55700648e-01 6.29166186e-01 -5.66400707e-01 1.25878847e+00
-1.15627535e-02 5.51613510e-01 -2.10241467e-01 -2.15666980e-01
2.65273154e-01 -7.92160816e-03 1.66181356e-01 1.27089846e+00
3.31803173e-01 4.58449647e-02 -3.26627807e-04 6.32126391e-01
4.06090379e-01 3.95036429e-01 -6.82105601e-01 5.48600070e-02
-2.00278014e-01 1.45058167e+00 -1.08691263e+00 -2.89743632e-01
-1.98577881e-01 9.48083282e-01 5.23867644e-02 -3.40773538e-02
-5.16836405e-01 -8.04049432e-01 2.19615936e-01 5.04811049e-01
1.31199419e-01 1.45057082e-01 -3.71764123e-01 -7.34493017e-01
-1.97781801e-01 -7.49926507e-01 5.52042425e-01 -3.37168306e-01
-8.61782551e-01 9.28672671e-01 -2.44842485e-01 -1.14086819e+00
-3.11393112e-01 -8.88120115e-01 -5.70657432e-01 9.25404131e-01
-1.30368459e+00 -9.30614054e-01 -4.13885087e-01 4.66698766e-01
4.23707783e-01 -3.20999086e-01 1.12817144e+00 4.92266685e-01
-1.45213515e-01 1.66170448e-01 -2.19365537e-01 4.04001534e-01
2.62510985e-01 -1.28924036e+00 -4.36107129e-01 5.73176384e-01
1.07452936e-01 3.38436395e-01 6.64912820e-01 -3.21270138e-01
-8.66704464e-01 -6.78651989e-01 1.27108824e+00 3.29617374e-02
-4.48301136e-02 -2.10187972e-01 -8.17749560e-01 1.39400780e-01
1.90868676e-01 2.94179320e-01 6.76821232e-01 4.91950801e-03
2.56311417e-01 -3.51445079e-01 -1.19582295e+00 2.17466615e-02
4.25550610e-01 -1.50415182e-01 -6.87137425e-01 1.71266422e-02
-1.87564760e-01 -7.87883028e-02 -8.95564079e-01 5.36455333e-01
7.56767392e-01 -1.33194864e+00 6.75136387e-01 -6.68563306e-01
3.02988708e-01 -4.10060078e-01 2.45009854e-01 -1.08284223e+00
6.98307157e-03 -6.98318854e-02 2.53772169e-01 8.44896019e-01
6.19164526e-01 -5.25628269e-01 7.48381793e-01 2.34991327e-01
4.78833318e-02 -1.09656131e+00 -7.62809455e-01 -3.10211807e-01
-1.68448210e-01 -2.22608715e-01 2.25594059e-01 1.02719152e+00
-2.21308753e-01 5.29650152e-01 -2.59812623e-01 -2.09946334e-01
3.27561975e-01 1.98138848e-01 2.88831145e-01 -1.76738966e+00
-3.07939291e-01 -2.69232154e-01 -1.06567705e+00 4.37824950e-02
-3.76159638e-01 -8.91034901e-01 -3.88187796e-01 -1.72988546e+00
-8.46213102e-03 -5.53428769e-01 -7.43218720e-01 5.70985019e-01
3.78448330e-02 2.29192331e-01 1.79722965e-01 -2.43796080e-01
-1.71914995e-02 2.95748562e-02 9.27983761e-01 1.18965350e-01
-2.78510511e-01 3.54928881e-01 -2.81135947e-01 6.21484995e-01
1.00729620e+00 -6.09235704e-01 -3.79834056e-01 -1.31785482e-01
2.21534908e-01 1.37784500e-02 2.10113689e-01 -1.62657118e+00
-4.30281013e-02 4.73894387e-01 1.04600668e+00 -6.71808302e-01
3.60857472e-02 -1.12285542e+00 3.04162741e-01 1.09562349e+00
-4.67898399e-01 2.42670909e-01 -2.56271139e-02 1.53463632e-01
-4.38940614e-01 -5.88727951e-01 8.64242673e-01 -4.34824616e-01
-4.97776002e-01 5.34766987e-02 -4.81317282e-01 -3.59354585e-01
1.01521623e+00 -4.76281524e-01 5.02605200e-01 -2.08803162e-01
-1.11868954e+00 -2.29866698e-01 -1.93017453e-01 1.65469497e-01
6.19933665e-01 -1.00509179e+00 -6.99183702e-01 2.02338815e-01
-6.69697765e-03 -1.94536760e-01 2.58033186e-01 1.01284182e+00
-9.88742471e-01 3.77077609e-01 -7.43526995e-01 -5.67297280e-01
-1.30977261e+00 5.92008889e-01 7.02649474e-01 -2.48256713e-01
-7.46127188e-01 4.86837000e-01 -5.06320119e-01 -9.79014412e-02
9.15456265e-02 -4.65900987e-01 -8.87131393e-01 1.40212476e-01
5.40605009e-01 3.82745594e-01 3.90863329e-01 -3.36201787e-01
-3.72345060e-01 5.97546995e-01 1.31419614e-01 2.04239383e-01
1.60585856e+00 4.88378137e-01 -4.46648672e-02 5.30494869e-01
1.23014033e+00 -3.62397879e-01 -6.34251356e-01 3.49850118e-01
4.88479793e-01 6.81253374e-02 1.00135943e-03 -9.95416045e-01
-1.03774822e+00 1.41460514e+00 1.25144923e+00 4.13559675e-01
1.40268326e+00 -5.03669500e-01 4.14458007e-01 5.30928195e-01
1.97023839e-01 -1.19555640e+00 -3.77854943e-01 1.58889011e-01
6.71923101e-01 -1.07773399e+00 -5.14228307e-02 -9.90113467e-02
-4.93993223e-01 1.88690472e+00 3.25133473e-01 -4.57620561e-01
9.25015807e-01 5.46133161e-01 2.64394224e-01 -3.94812852e-01
-3.61490160e-01 -2.49471590e-01 1.29132763e-01 4.86136705e-01
8.72105360e-01 -1.83814969e-02 -1.05203187e+00 7.12459028e-01
2.13078372e-02 3.71917218e-01 2.57205546e-01 1.08080924e+00
-4.88147348e-01 -1.37396467e+00 -1.74243718e-01 8.53634834e-01
-8.52607131e-01 2.79088080e-01 -3.26857239e-01 7.89308786e-01
6.58414841e-01 7.47713506e-01 -3.17280054e-01 -3.96852970e-01
4.38865721e-01 4.50002193e-01 4.92599994e-01 -5.30188262e-01
-1.28940094e+00 -6.78268895e-02 -5.98434992e-02 -2.92050391e-01
-2.81260312e-01 -4.04071510e-01 -1.33795822e+00 1.94893599e-01
-2.21095383e-01 3.27173829e-01 8.37502301e-01 9.90522921e-01
2.06297651e-01 7.40826964e-01 4.83343393e-01 -4.59445655e-01
-4.36528593e-01 -1.18063605e+00 -6.14283144e-01 5.10711372e-01
2.94660032e-01 -4.12982464e-01 6.69822171e-02 9.98941213e-02]
|
[14.24092960357666, 3.2976226806640625]
|
fa5df241-5721-4d1a-970f-1ae07fb502db
|
condition-random-fields-based-grammatical
| null | null |
https://aclanthology.org/W15-4416
|
https://aclanthology.org/W15-4416.pdf
|
Condition Random Fields-based Grammatical Error Detection for Chinese as Second Language
| null |
['Wan-Ling Tsai', 'Ya-Ting Li', 'Kai-Hsiang Yu', 'Chan-Kun Yeh', 'Jui-Feng Yeh']
|
2015-07-01
| null | null | null |
ws-2015-7
|
['grammatical-error-detection']
|
['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]
|
[-7.3621439933776855, 3.7058639526367188]
|
87d17beb-1108-42ad-9f02-0cdd63a7729b
|
superbench-a-super-resolution-benchmark
|
2306.1407
| null |
https://arxiv.org/abs/2306.14070v1
|
https://arxiv.org/pdf/2306.14070v1.pdf
|
SuperBench: A Super-Resolution Benchmark Dataset for Scientific Machine Learning
|
Super-Resolution (SR) techniques aim to enhance data resolution, enabling the retrieval of finer details, and improving the overall quality and fidelity of the data representation. There is growing interest in applying SR methods to complex spatiotemporal systems within the Scientific Machine Learning (SciML) community, with the hope of accelerating numerical simulations and/or improving forecasts in weather, climate, and related areas. However, the lack of standardized benchmark datasets for comparing and validating SR methods hinders progress and adoption in SciML. To address this, we introduce SuperBench, the first benchmark dataset featuring high-resolution datasets (up to $2048\times2048$ dimensions), including data from fluid flows, cosmology, and weather. Here, we focus on validating spatial SR performance from data-centric and physics-preserved perspectives, as well as assessing robustness to data degradation tasks. While deep learning-based SR methods (developed in the computer vision community) excel on certain tasks, despite relatively limited prior physics information, we identify limitations of these methods in accurately capturing intricate fine-scale features and preserving fundamental physical properties and constraints in scientific data. These shortcomings highlight the importance and subtlety of incorporating domain knowledge into ML models. We anticipate that SuperBench will significantly advance SR methods for scientific tasks.
|
['Michael W. Mahoney', 'Zarija Lukic', 'Omer San', 'Shashank Subramanian', 'N. Benjamin Erichson', 'Pu Ren']
|
2023-06-24
| null | null | null | null |
['super-resolution', 'retrieval']
|
['computer-vision', 'methodology']
|
[-5.26237637e-02 -5.40550888e-01 9.08690915e-02 -7.19393566e-02
-5.55699885e-01 -6.47406518e-01 9.87058163e-01 3.01984161e-01
-1.50751188e-01 1.01636004e+00 3.54693562e-01 -2.03592837e-01
-3.28617096e-01 -9.73153949e-01 -6.68500602e-01 -6.91143334e-01
-3.66439939e-01 2.84161955e-01 5.01622148e-02 -1.18432760e-01
2.23612875e-01 1.11496472e+00 -1.49415720e+00 1.76069334e-01
9.21676457e-01 7.37805247e-01 2.54691660e-01 5.29688418e-01
-2.09290981e-01 6.81828797e-01 -2.68587232e-01 3.81990999e-01
1.41069457e-01 -1.79415286e-01 -7.22832382e-01 -4.14192528e-01
8.00739884e-01 1.51338577e-02 -5.31849742e-01 7.74681509e-01
4.48291272e-01 5.55250466e-01 4.31761682e-01 -6.68190956e-01
-8.95571649e-01 2.07084306e-02 -5.08035421e-01 7.45404422e-01
5.56717254e-03 3.66789520e-01 8.84004712e-01 -9.16196585e-01
8.29687357e-01 1.38268125e+00 8.83500516e-01 3.22950959e-01
-1.45149004e+00 -6.34607852e-01 1.23081602e-01 7.27692023e-02
-1.35237372e+00 -6.70167565e-01 5.53271532e-01 -8.19263339e-01
1.09075475e+00 3.83322358e-01 5.02498865e-01 9.87567961e-01
4.24919933e-01 2.40685374e-01 1.26961327e+00 -2.16882639e-02
3.92280251e-01 -1.02190614e-01 -1.02311797e-01 4.00051177e-01
3.13837528e-01 6.89566791e-01 -1.01491451e+00 -1.87652856e-01
9.99310255e-01 -3.47991019e-01 -2.27449954e-01 -3.13773036e-01
-1.41063964e+00 6.50154054e-01 5.85933864e-01 3.12403202e-01
-3.67896497e-01 2.95307152e-02 1.90760896e-01 -2.49005458e-03
9.60523367e-01 8.53737235e-01 -6.47783220e-01 -1.36860102e-01
-1.09946418e+00 7.03677893e-01 4.05900031e-01 5.46045244e-01
5.01273036e-01 4.73446816e-01 4.33995202e-02 7.09980488e-01
4.09834571e-02 7.91792154e-01 2.00437270e-02 -1.28937948e+00
1.64947271e-01 3.42841893e-01 2.81379730e-01 -1.22009885e+00
-6.55698299e-01 -7.48477519e-01 -1.18227875e+00 5.43421984e-01
1.44462138e-01 2.10447442e-02 -6.46394491e-01 1.61732304e+00
5.25607049e-01 6.57229125e-01 1.25163838e-01 1.08892691e+00
1.07858479e+00 8.95785809e-01 2.02522367e-01 -6.79761842e-02
1.08785009e+00 -4.70471799e-01 -5.10031223e-01 -8.09889808e-02
3.42593729e-01 -6.31140530e-01 9.55773413e-01 1.82928219e-01
-8.82711887e-01 -7.73953378e-01 -7.19152808e-01 -2.00936064e-01
-5.87925673e-01 -3.17555428e-01 5.88246942e-01 -2.52055228e-02
-9.22760904e-01 1.04714715e+00 -1.06976318e+00 -3.79626662e-01
4.00276750e-01 -1.38482526e-01 -3.22189659e-01 1.34515479e-01
-1.32559717e+00 1.01347148e+00 -1.38807207e-01 5.49659319e-02
-7.99977899e-01 -1.66764283e+00 -9.38439012e-01 3.92158814e-02
-2.74066538e-01 -7.16803968e-01 5.97821534e-01 -1.08456582e-01
-9.74852383e-01 5.17959535e-01 -2.79779315e-01 -4.45486546e-01
4.34465021e-01 -1.58082828e-01 -6.01549983e-01 -6.47422224e-02
8.12209919e-02 5.90647995e-01 4.40384597e-01 -1.27903962e+00
-5.79951763e-01 -3.79670769e-01 -2.00803369e-01 1.78997651e-01
-1.31545141e-01 -1.77042097e-01 -9.14394408e-02 -6.98702335e-01
-9.92798209e-02 -7.23815322e-01 -2.72341520e-01 3.79761428e-01
1.25692919e-01 1.02193542e-01 1.02542257e+00 -7.19790757e-01
8.51936579e-01 -2.11644483e+00 3.57514381e-01 -2.35686958e-01
4.31839466e-01 2.83736616e-01 -1.35915369e-01 3.94444108e-01
1.68399498e-01 1.44753575e-01 -6.12807393e-01 -3.91876400e-01
-2.88215578e-01 2.51337528e-01 -6.59098029e-01 6.96691692e-01
2.48352930e-01 7.74297178e-01 -8.41394126e-01 -3.42946708e-01
5.73933959e-01 8.90931070e-01 -2.88877159e-01 -3.81430537e-02
-3.30173194e-01 1.19776344e+00 -4.37807649e-01 4.30588722e-01
9.15352285e-01 -3.08258504e-01 -2.77475148e-01 -6.28043339e-02
-6.32305384e-01 2.90675759e-01 -1.34552133e+00 1.53743660e+00
-6.80777669e-01 8.12282741e-01 3.90390038e-01 -7.70502031e-01
8.78649831e-01 5.00413887e-02 7.56385326e-01 -1.13184798e+00
-5.79137683e-01 -9.82729942e-02 -1.91525161e-01 -3.93388480e-01
5.17458320e-01 -1.82505786e-01 2.87954509e-01 3.69155884e-01
-3.91933680e-01 -4.63773847e-01 6.98026121e-02 1.45467743e-01
7.37840116e-01 3.37858528e-01 -2.19664630e-02 -7.35432625e-01
3.78080606e-01 3.94772440e-01 5.67140102e-01 7.34134376e-01
-5.64413182e-02 6.72241688e-01 1.56147303e-02 -6.42747343e-01
-1.34528399e+00 -1.08035624e+00 -8.25289905e-01 8.66821766e-01
1.46430908e-02 -1.78726882e-01 2.89530344e-02 1.33448057e-02
6.23584509e-01 5.64591646e-01 -7.44032562e-01 3.76529247e-02
-6.44859552e-01 -1.08231032e+00 4.85585421e-01 5.00583708e-01
4.60278302e-01 -1.00426304e+00 -5.89608073e-01 2.57795006e-01
1.77660175e-02 -1.28947866e+00 1.34011880e-01 -4.17427987e-01
-9.65056658e-01 -9.00000989e-01 -4.20864105e-01 -1.95755973e-01
1.71661884e-01 2.24200457e-01 1.22155499e+00 -1.35182336e-01
-6.50977075e-01 2.05330402e-01 -7.65089318e-02 -1.91877306e-01
-3.51800948e-01 1.24020167e-01 2.86014467e-01 -4.09798980e-01
-1.71431571e-01 -7.59512782e-01 -6.28990471e-01 -4.37935069e-02
-8.44277978e-01 2.03864112e-01 1.55739710e-01 4.35452849e-01
6.26364827e-01 6.63897675e-03 5.28939843e-01 -5.90878785e-01
4.54245150e-01 -6.41657531e-01 -8.48936439e-01 -2.96399612e-02
-7.17690229e-01 4.83177006e-02 7.54832268e-01 -5.22798896e-02
-1.26459515e+00 -3.20534974e-01 -6.88325288e-03 -3.89456034e-01
-1.66602120e-01 7.17706621e-01 2.88443536e-01 -3.56970191e-01
9.50857282e-01 3.38083535e-01 8.93077850e-02 -8.88556302e-01
2.96542734e-01 2.25289449e-01 6.58967495e-01 -8.87155473e-01
7.76829422e-01 8.78830850e-01 5.69335461e-01 -1.19490504e+00
-7.74290264e-01 -3.15638185e-01 -5.87732852e-01 -1.42866969e-02
4.90920097e-01 -1.11720335e+00 -4.15700972e-01 3.87338877e-01
-6.82698727e-01 -4.00561124e-01 -3.27673644e-01 4.84602004e-01
-2.82487273e-01 3.66862833e-01 -6.69734299e-01 -5.05016565e-01
-3.46324265e-01 -9.44559813e-01 1.14978015e+00 1.53712267e-02
-1.75228015e-01 -1.28685462e+00 5.07651091e-01 1.46761522e-01
9.68399227e-01 6.54106617e-01 8.66661370e-01 1.88187540e-01
-4.97137070e-01 3.42140704e-01 -4.58130270e-01 -6.83078021e-02
2.99320281e-01 3.04259688e-01 -9.88009274e-01 -4.18751180e-01
-1.41992554e-01 -1.13293022e-01 1.07738435e+00 4.39123482e-01
1.24643707e+00 -4.31053489e-02 -3.36666763e-01 9.21809137e-01
1.43512678e+00 -2.20353112e-01 1.51554242e-01 3.20520461e-01
7.68475056e-01 5.67219615e-01 3.69233787e-01 5.22515059e-01
2.99791247e-01 6.65054202e-01 2.52717167e-01 -4.10893381e-01
-5.12102723e-01 1.15669332e-01 -1.18322745e-01 3.95498812e-01
6.41049817e-02 6.92021698e-02 -1.26234698e+00 5.66156089e-01
-1.69658005e+00 -9.63147581e-01 -3.42838526e-01 1.97226417e+00
9.39019561e-01 -1.50684401e-01 -2.69078612e-01 -2.95567930e-01
3.10513645e-01 5.75225592e-01 -9.24183607e-01 -1.19288564e-01
-4.66980338e-01 3.12334776e-01 2.60954857e-01 8.40904236e-01
-1.17279613e+00 9.81989026e-01 6.40033102e+00 5.19241452e-01
-1.58181000e+00 -3.09960805e-02 4.92232800e-01 -3.32275808e-01
-3.82374108e-01 -2.32120320e-01 -8.08457017e-01 2.81610817e-01
1.07443905e+00 -3.26084614e-01 6.52578831e-01 4.66557622e-01
7.75202334e-01 2.80239410e-03 -9.09048319e-01 5.99835157e-01
-4.16898429e-01 -2.10824013e+00 2.87759714e-02 -7.14782439e-03
9.19665098e-01 5.87632120e-01 2.05224350e-01 4.98937964e-02
3.83887649e-01 -1.26269603e+00 4.36356992e-01 7.92509377e-01
9.59625363e-01 -5.25163472e-01 3.03853959e-01 2.25928113e-01
-1.33558679e+00 3.83449882e-01 -4.19016629e-01 -2.12139562e-01
-5.71774468e-02 9.25380290e-01 -2.71508813e-01 7.01535046e-01
1.07821286e+00 1.24911916e+00 -4.65633571e-01 8.40350270e-01
4.73808557e-01 6.94320858e-01 -4.83506501e-01 6.23820543e-01
1.70266882e-01 -1.85976982e-01 7.49233425e-01 1.37749553e+00
1.96536541e-01 2.69012243e-01 4.74133454e-02 1.22837186e+00
8.10799450e-02 -2.51007944e-01 -7.06878066e-01 1.30444001e-02
5.54816067e-01 1.13295138e+00 -3.77048284e-01 -1.83462620e-01
-3.86860251e-01 3.04333866e-01 3.60287398e-01 5.58651507e-01
-5.15092194e-01 1.26823723e-01 1.36832321e+00 1.15688212e-01
7.34001473e-02 -7.51531363e-01 -8.81824672e-01 -1.19672823e+00
-1.59109607e-01 -8.03415656e-01 3.34591091e-01 -7.17260659e-01
-1.28826582e+00 3.62952113e-01 9.69153345e-02 -1.03136814e+00
1.17971249e-01 -4.13412720e-01 -4.82635736e-01 1.23426616e+00
-1.87352514e+00 -9.44121480e-01 -4.96415883e-01 2.18452185e-01
3.80650669e-01 -3.77152232e-03 7.26864398e-01 1.18834712e-01
-5.95219314e-01 -1.39021024e-01 8.45000625e-01 -3.83407414e-01
6.03193581e-01 -1.09867406e+00 9.06069338e-01 7.54847705e-01
-1.97950061e-02 6.93044543e-01 1.04885781e+00 -7.47428119e-01
-1.59793019e+00 -1.36073613e+00 6.61995173e-01 -6.82696342e-01
8.03979695e-01 -3.12936425e-01 -1.50391352e+00 2.38171443e-01
-2.56732672e-01 4.85044241e-01 3.21937233e-01 1.46416724e-01
-2.49139592e-01 -1.73386738e-01 -1.26448774e+00 3.87320668e-01
1.07992053e+00 -6.77723706e-01 -3.69104981e-01 3.29551309e-01
4.50624615e-01 -5.47488213e-01 -1.27850103e+00 8.72535646e-01
3.84839147e-01 -6.42993450e-01 1.34401691e+00 -7.20883369e-01
6.32379472e-01 -4.31333512e-01 -7.07465485e-02 -1.27724850e+00
-5.84198713e-01 -9.62265134e-02 -4.21345890e-01 9.86506402e-01
9.95964855e-02 -5.10372579e-01 5.92586100e-01 5.49533427e-01
5.02251908e-02 -5.06611109e-01 -1.01178944e+00 -7.41305590e-01
7.87716925e-01 -3.46280187e-01 7.22163141e-01 1.29351020e+00
-3.69627953e-01 -1.42440647e-01 -2.35228047e-01 6.24550641e-01
8.65904689e-01 6.57542348e-01 5.57424843e-01 -1.48178458e+00
8.27203020e-02 -5.79994142e-01 1.36713117e-01 -5.52814960e-01
1.55249536e-01 -6.71986997e-01 -2.68897504e-01 -1.63217080e+00
-4.25929911e-02 -7.83666193e-01 -3.84021997e-01 1.96589813e-01
-2.13916212e-01 2.89580405e-01 3.72108980e-03 5.87648869e-01
-2.35928252e-01 8.01652968e-01 1.43853652e+00 1.74904261e-02
-2.15471350e-02 -4.51832563e-01 -3.91648233e-01 3.92196178e-01
8.09249699e-01 -1.93756402e-01 -1.58610150e-01 -6.62485123e-01
1.58382375e-02 -2.23565195e-02 6.86427772e-01 -1.13068020e+00
1.27331063e-01 -6.68288350e-01 7.10476518e-01 -4.26025808e-01
2.86682606e-01 -4.97577399e-01 2.78347015e-01 3.44445825e-01
-3.80537570e-01 -6.81465119e-02 8.83745849e-01 4.50246602e-01
-9.18237120e-02 5.33670545e-01 1.00898743e+00 -3.22570711e-01
-8.98685217e-01 4.81503755e-01 -3.34733397e-01 1.93466663e-01
6.89897060e-01 1.33409739e-01 -5.40195405e-01 -2.41806507e-02
-6.42444491e-01 4.63429034e-01 6.56652153e-01 6.02219880e-01
3.47813547e-01 -9.27958131e-01 -1.04238951e+00 3.40713620e-01
3.65037359e-02 3.34877074e-01 5.27592480e-01 3.80889833e-01
-6.47789061e-01 6.67476416e-01 -3.05754364e-01 -6.02816880e-01
-9.49338377e-01 3.72452021e-01 6.09761059e-01 -6.03192933e-02
-1.10752571e+00 6.52531326e-01 2.71922886e-01 -6.94452703e-01
-4.93692234e-02 -3.34817976e-01 -2.42423147e-01 -2.07739636e-01
4.13313359e-01 5.69511175e-01 1.11463197e-01 -5.89211404e-01
-4.99677390e-01 6.64060593e-01 2.59944618e-01 2.90135723e-02
1.65602314e+00 -2.11668029e-01 -1.20465174e-01 3.44646335e-01
8.96014214e-01 7.42406249e-02 -1.71783006e+00 -5.85603237e-01
-1.35925710e-01 -5.37411034e-01 5.33329010e-01 -8.51106644e-01
-1.10480142e+00 8.99148226e-01 5.23547173e-01 4.64436188e-02
7.38157213e-01 -5.56519777e-02 8.25441957e-01 1.84463263e-01
2.78770596e-01 -7.65538096e-01 -3.78101081e-01 5.41502953e-01
1.17999339e+00 -1.29757941e+00 4.57725346e-01 -9.59355012e-02
-2.83299029e-01 9.41906214e-01 6.85113490e-01 -1.04930543e-03
7.98377931e-01 5.75881422e-01 1.99032165e-02 -3.55223656e-01
-8.90061140e-01 1.29576707e-02 4.03519988e-01 4.42375511e-01
3.80528688e-01 -5.60283773e-02 1.14779063e-01 -1.45342603e-01
-1.92036390e-01 9.26309675e-02 3.15408200e-01 9.04632986e-01
-4.44569826e-01 -7.26357996e-01 -4.55034882e-01 4.11867172e-01
-1.34156615e-01 -2.12173790e-01 -1.28459409e-01 5.85920036e-01
7.01091215e-02 6.55898929e-01 3.44116986e-01 1.45390868e-01
1.66521057e-01 -2.55007178e-01 1.55825555e-01 -4.61583793e-01
-1.99568406e-01 -3.31746578e-01 1.90655887e-03 -6.64023280e-01
-2.86361367e-01 -9.09922123e-01 -1.17585564e+00 -7.98257470e-01
4.08541888e-01 2.72972673e-01 7.24177063e-01 8.99875939e-01
8.09823871e-01 6.50960326e-01 2.33545780e-01 -9.43347335e-01
-1.28566071e-01 -8.22529733e-01 -3.71421486e-01 2.65059769e-01
7.97117889e-01 -7.92815089e-01 -3.82018328e-01 -3.15190181e-02]
|
[6.573493480682373, 3.139309883117676]
|
233b10d4-5840-454e-a658-915603d5d321
|
balanced-supervised-contrastive-learning-for
|
2305.16687
| null |
https://arxiv.org/abs/2305.16687v1
|
https://arxiv.org/pdf/2305.16687v1.pdf
|
Balanced Supervised Contrastive Learning for Few-Shot Class-Incremental Learning
|
Few-shot class-incremental learning (FSCIL) presents the primary challenge of balancing underfitting to a new session's task and forgetting the tasks from previous sessions. To address this challenge, we develop a simple yet powerful learning scheme that integrates effective methods for each core component of the FSCIL network, including the feature extractor, base session classifiers, and incremental session classifiers. In feature extractor training, our goal is to obtain balanced generic representations that benefit both current viewable and unseen or past classes. To achieve this, we propose a balanced supervised contrastive loss that effectively balances these two objectives. In terms of classifiers, we analyze and emphasize the importance of unifying initialization methods for both the base and incremental session classifiers. Our method demonstrates outstanding ability for new task learning and preventing forgetting on CUB200, CIFAR100, and miniImagenet datasets, with significant improvements over previous state-of-the-art methods across diverse metrics. We conduct experiments to analyze the significance and rationale behind our approach and visualize the effectiveness of our representations on new tasks. Furthermore, we conduct diverse ablation studies to analyze the effects of each module.
|
['Jong-Hwan Kim', 'Young-Min Kim', 'Tae-Min Choi', 'In-Ug Yoon']
|
2023-05-26
| null | null | null | null |
['class-incremental-learning', 'few-shot-class-incremental-learning', 'incremental-learning']
|
['computer-vision', 'methodology', 'methodology']
|
[ 2.21930549e-01 -1.71743780e-01 -1.14983842e-01 -4.26214069e-01
-7.84991324e-01 -4.59334582e-01 5.40349782e-01 2.40870863e-02
-5.38242817e-01 7.73645222e-01 1.80322498e-01 -6.37867972e-02
-2.98272759e-01 -3.52016032e-01 -6.64085984e-01 -6.99369013e-01
-1.61964502e-02 6.74271490e-03 4.49084848e-01 -1.28523842e-01
2.16909796e-01 4.02995318e-01 -1.66243041e+00 3.51954639e-01
9.78557944e-01 8.72929215e-01 1.88001141e-01 3.34495306e-01
5.62584996e-02 7.43660450e-01 -6.74890757e-01 -4.50476199e-01
5.50142303e-02 -2.59528726e-01 -6.84485137e-01 -7.83701316e-02
7.17641473e-01 -3.86216909e-01 -4.69938427e-01 7.00586855e-01
6.26376271e-01 4.25522119e-01 5.53888738e-01 -1.36855781e+00
-6.85209632e-01 7.67064393e-01 -5.21407127e-01 5.38740218e-01
-1.58194125e-01 3.19975287e-01 9.45290029e-01 -1.16665089e+00
4.92876619e-01 8.65711927e-01 8.27974558e-01 8.55180442e-01
-1.37198508e+00 -8.08731079e-01 6.90531254e-01 3.45353007e-01
-1.03537977e+00 -7.06098974e-01 5.95923603e-01 -4.38545793e-01
8.30531538e-01 -1.19831553e-02 4.07133192e-01 1.49512339e+00
8.58944654e-02 1.09094656e+00 8.76481116e-01 -2.10681677e-01
3.30223501e-01 2.78644383e-01 8.68208230e-01 4.88099366e-01
4.30631459e-01 3.44228372e-02 -9.40326631e-01 -7.79832229e-02
3.53855342e-01 4.33630645e-01 -2.84250081e-01 -7.50171781e-01
-9.22258198e-01 5.19232392e-01 3.72516662e-01 2.85277665e-01
-6.89500198e-02 1.30006909e-01 4.52661574e-01 2.81102926e-01
4.82259154e-01 5.34910023e-01 -5.66669941e-01 -2.30049957e-02
-9.54838932e-01 1.45120367e-01 4.99068111e-01 7.56445050e-01
6.23935461e-01 2.07739979e-01 -7.16314673e-01 8.39004397e-01
-2.77018338e-01 2.75847167e-01 6.03015661e-01 -7.09613144e-01
3.20120394e-01 4.79027659e-01 -1.31757900e-01 -5.28879523e-01
-4.46884543e-01 -9.65659142e-01 -5.95020950e-01 1.59663424e-01
2.18562245e-01 -3.78358662e-02 -1.14369607e+00 2.01441979e+00
-5.01208343e-02 3.02597106e-01 4.52835113e-03 3.47466499e-01
7.53016055e-01 3.21354359e-01 3.61893803e-01 -1.87570840e-01
1.04918456e+00 -1.14533746e+00 -6.12489879e-01 -5.93344986e-01
5.57539642e-01 -2.10817277e-01 1.34833407e+00 1.44456610e-01
-1.00424206e+00 -7.81832874e-01 -1.33540606e+00 -5.18592037e-02
-3.68592858e-01 1.90101370e-01 7.19955385e-01 4.39898789e-01
-8.67737114e-01 8.34918320e-01 -8.07082415e-01 -1.45389080e-01
7.60379910e-01 1.70207500e-01 -2.21624821e-01 -6.28228337e-02
-1.01667643e+00 9.02151346e-01 3.02713156e-01 -2.25583628e-01
-1.21085083e+00 -1.14145172e+00 -6.67907238e-01 5.98183811e-01
3.68895680e-01 -6.74790978e-01 1.35918486e+00 -7.81451702e-01
-1.00395501e+00 6.80106163e-01 -2.58099049e-01 -7.40532279e-01
3.80831391e-01 -4.27760214e-01 -2.00214013e-01 -8.47911090e-02
1.55128213e-03 4.61781681e-01 1.11969125e+00 -1.09189558e+00
-7.43640542e-01 -4.80006635e-01 3.23275886e-02 2.73768932e-01
-7.31439471e-01 -5.31183302e-01 -3.20408195e-01 -7.39987135e-01
-3.33460905e-02 -7.24323332e-01 9.48133096e-02 -4.14859690e-02
-8.51265248e-03 -3.69827062e-01 8.85059297e-01 -5.36135197e-01
1.29704046e+00 -2.58286762e+00 8.06858987e-02 -2.81449050e-01
4.43661034e-01 4.15645123e-01 -2.82800078e-01 4.18794565e-02
-1.74521074e-01 -8.31475258e-02 -1.60390362e-01 -7.06934094e-01
-1.46411091e-01 -8.54352713e-02 -6.48299694e-01 1.44167632e-01
1.53077245e-01 9.41730142e-01 -9.13112581e-01 4.94069681e-02
4.66003418e-02 3.94964695e-01 -5.13091147e-01 9.40702707e-02
5.90697527e-02 1.78766027e-01 -5.61556108e-02 5.33202529e-01
5.26975930e-01 -3.41242760e-01 9.96788815e-02 -2.30569631e-01
1.46875486e-01 2.48740137e-01 -9.22130942e-01 1.91631293e+00
-4.04949367e-01 5.85798740e-01 -3.68991286e-01 -8.98045599e-01
7.66502738e-01 1.04344591e-01 2.53856510e-01 -9.26968038e-01
-7.30012059e-02 7.18019754e-02 -2.01054230e-01 -2.03362390e-01
4.07392263e-01 5.02414554e-02 -3.45492666e-03 5.09593427e-01
7.01723278e-01 3.95051777e-01 2.61444867e-01 4.73954409e-01
1.19804752e+00 1.38635904e-01 1.28472954e-01 -1.64096057e-01
1.46191552e-01 -3.73154283e-01 5.67102134e-01 1.21133375e+00
-5.96706688e-01 6.19197190e-01 4.42141086e-01 -6.36253476e-01
-6.91495359e-01 -1.26436400e+00 3.21357921e-02 1.55071998e+00
2.14447472e-02 -3.68263304e-01 -4.02160853e-01 -1.09540761e+00
9.54829901e-02 9.44119096e-01 -9.18061972e-01 -8.94434333e-01
-4.47306573e-01 -1.00088489e+00 1.79903522e-01 7.94528186e-01
4.88845617e-01 -9.37959611e-01 -6.68230951e-01 1.74084216e-01
-1.59184784e-01 -9.39578712e-01 -4.46077317e-01 4.93604898e-01
-1.03821266e+00 -1.24542153e+00 -6.17638588e-01 -7.79533744e-01
5.98667026e-01 6.91488147e-01 1.19055569e+00 -1.07951581e-01
-3.69838983e-01 4.94918406e-01 -2.06043363e-01 -4.69621211e-01
2.03511566e-01 5.76295793e-01 6.29295781e-02 6.14665709e-02
2.86367267e-01 -7.44533062e-01 -6.64885700e-01 1.09499492e-01
-8.07648599e-01 -6.39761314e-02 6.36707246e-01 8.71665657e-01
3.12910408e-01 -1.90676138e-01 7.35374331e-01 -1.22359228e+00
6.34950459e-01 -3.68329048e-01 -2.76000202e-01 5.29497325e-01
-8.60012949e-01 1.74137920e-01 4.52450216e-01 -5.78727126e-01
-1.26697683e+00 -7.44669512e-02 1.85169503e-01 -4.30064648e-01
2.61487782e-01 2.53764331e-01 -4.96783741e-02 1.70313120e-02
9.32240069e-01 4.01298553e-01 -1.79746285e-01 -6.41461909e-01
4.40848172e-01 2.67419785e-01 6.31601214e-01 -4.73960727e-01
8.99182379e-01 5.82300425e-01 -4.95072573e-01 -5.51173747e-01
-1.47032809e+00 -3.81347358e-01 -6.84933484e-01 -7.87455589e-03
3.50002527e-01 -1.06156516e+00 -5.57398260e-01 7.30405271e-01
-8.92078400e-01 -4.06979829e-01 -7.64318287e-01 2.03569621e-01
-2.49014735e-01 -4.87813959e-03 -4.34611589e-01 -6.10829890e-01
-4.62926894e-01 -9.63611484e-01 7.34031081e-01 4.43145692e-01
-7.01209083e-02 -8.60104918e-01 1.73051134e-01 1.66160554e-01
7.02366889e-01 -9.47533846e-02 9.49758828e-01 -7.74359643e-01
-4.89963353e-01 -1.01080388e-01 -2.98239201e-01 5.72668970e-01
5.39955944e-02 -3.83017242e-01 -1.35727406e+00 -7.42280483e-01
4.87614907e-02 -5.32069802e-01 1.69060421e+00 2.07817808e-01
1.34783709e+00 2.97708828e-02 -3.50908190e-01 9.01335776e-01
1.29320800e+00 1.54333651e-01 6.71532273e-01 3.60100776e-01
4.58697259e-01 3.05665284e-01 2.92922646e-01 3.23519915e-01
2.85768211e-01 3.81378502e-01 1.08229049e-01 4.14261967e-02
-5.06857336e-01 -3.04260820e-01 4.28969413e-01 5.17043233e-01
1.87759101e-01 1.73103571e-01 -6.33303642e-01 6.86405122e-01
-1.90999627e+00 -1.06820178e+00 7.00364888e-01 2.28999639e+00
8.52543890e-01 4.40931529e-01 -9.14727375e-02 2.95019224e-02
4.03276235e-01 3.80194813e-01 -9.46602702e-01 5.51620349e-02
6.41300529e-03 3.23071152e-01 1.62238523e-01 1.39577389e-01
-1.29415047e+00 9.46363330e-01 6.46579170e+00 7.89778113e-01
-9.72215354e-01 2.60667831e-01 7.03291893e-01 -5.47668219e-01
-8.29499438e-02 9.18884389e-03 -1.31471157e+00 3.52050066e-01
7.66868174e-01 -3.27192068e-01 4.62463647e-01 1.07133985e+00
-2.88401783e-01 6.58244267e-02 -1.28515732e+00 1.06903446e+00
5.14554799e-01 -1.34285533e+00 1.50806397e-01 -4.40378934e-01
7.91461825e-01 1.03742555e-02 3.90139371e-01 9.68148053e-01
1.56063408e-01 -6.74854279e-01 6.33569121e-01 6.16732955e-01
7.51305997e-01 -5.17975867e-01 3.78579378e-01 1.64834812e-01
-9.77936685e-01 -4.86705542e-01 -2.57093847e-01 5.63705154e-02
-2.52619505e-01 5.47913730e-01 -5.50252557e-01 2.93304771e-01
8.49012136e-01 8.54363739e-01 -1.05511975e+00 1.27515411e+00
-2.89642066e-01 4.99235243e-01 8.34630877e-02 3.47587794e-01
-5.87798618e-02 3.94996047e-01 3.84883225e-01 9.57481742e-01
1.01666115e-01 -1.83123738e-01 1.18809998e-01 7.45309472e-01
-2.91710854e-01 -2.51137078e-01 -5.39677262e-01 1.23753339e-01
6.21219456e-01 1.11968243e+00 -3.84272963e-01 -3.39757442e-01
-3.26642931e-01 9.85133529e-01 8.58020902e-01 5.77583015e-01
-7.53428698e-01 -4.81667429e-01 7.62447655e-01 7.46236891e-02
3.14368218e-01 -1.99464470e-01 -3.37729394e-01 -1.49495757e+00
1.20868079e-01 -7.42606342e-01 5.71007133e-01 -5.10630727e-01
-1.23263407e+00 6.40866101e-01 -3.96622196e-02 -9.92217362e-01
8.52877498e-02 -4.25228953e-01 -7.60364056e-01 5.89198709e-01
-1.69704497e+00 -1.10370600e+00 -4.84894782e-01 6.64580107e-01
8.65012348e-01 -4.20110196e-01 6.87508941e-01 3.23340207e-01
-7.52109110e-01 9.57450390e-01 1.98311269e-01 -2.37594005e-02
9.32662368e-01 -9.46199179e-01 5.59025645e-01 9.36762691e-01
1.67761162e-01 7.23750412e-01 3.88765246e-01 -4.31012213e-01
-1.12147427e+00 -1.15920913e+00 6.72085464e-01 -6.74201906e-01
4.18372333e-01 -6.02637589e-01 -9.80617940e-01 9.06906188e-01
-1.51184604e-01 -4.42789234e-02 5.30611932e-01 5.66514313e-01
-6.93112612e-01 -5.33007681e-01 -6.92409158e-01 5.76716006e-01
1.23221660e+00 -5.19384146e-01 -7.71426857e-01 2.19838887e-01
8.27877760e-01 -1.89699605e-01 -1.47357792e-01 5.72449386e-01
6.10797524e-01 -9.18825984e-01 1.10887635e+00 -8.28868866e-01
1.35228649e-01 1.41135946e-01 8.75008851e-02 -1.50399470e+00
-6.00796878e-01 -3.69233638e-01 -5.26280403e-01 1.13153887e+00
4.70503658e-01 -5.78565955e-01 7.97689378e-01 4.55799550e-01
-1.59933671e-01 -6.73202932e-01 -6.93059087e-01 -8.84550452e-01
-6.87158778e-02 -1.95828319e-01 3.64295334e-01 8.42603385e-01
-2.93760389e-01 7.04275429e-01 -5.81696212e-01 -1.83596253e-01
8.35585356e-01 2.63857152e-02 6.69266462e-01 -1.38662195e+00
-1.48601711e-01 -3.37444603e-01 -3.67761496e-03 -6.65849268e-01
5.06560802e-02 -9.39867377e-01 -1.47258863e-01 -1.40934646e+00
6.19201660e-01 -4.11640584e-01 -9.34600055e-01 8.31552446e-01
-5.49932122e-01 4.60823961e-02 3.83848876e-01 2.42150456e-01
-1.04321563e+00 8.15574467e-01 9.76086259e-01 -2.67366171e-01
-3.80912364e-01 1.01379119e-01 -1.08736444e+00 5.89654684e-01
7.02703714e-01 -5.34764767e-01 -8.01782787e-01 -6.25422418e-01
1.80077944e-02 -5.56370437e-01 4.42943633e-01 -1.41565847e+00
3.98950011e-01 2.08297953e-01 5.78821063e-01 -5.07697046e-01
3.07325959e-01 -5.96967936e-01 -3.54795665e-01 4.01429474e-01
-5.55208147e-01 -1.93124786e-01 3.26587349e-01 7.64665842e-01
6.02548644e-02 -1.03246726e-01 9.51257050e-01 -1.38293102e-01
-1.01342046e+00 4.60582227e-01 5.14523722e-02 2.57029682e-01
1.04905939e+00 2.15239897e-02 -7.00387359e-01 -1.47348419e-01
-1.04127109e+00 3.38915169e-01 1.09959036e-01 7.34126687e-01
7.65497983e-01 -1.19101524e+00 -5.98370910e-01 5.49702108e-01
2.77938187e-01 -3.65155041e-01 5.86781621e-01 6.87430382e-01
9.11448896e-02 2.66063780e-01 -3.96969318e-01 -3.33585441e-01
-1.07443213e+00 5.89117169e-01 3.13073725e-01 -5.11584580e-01
-5.04708230e-01 1.01269603e+00 5.03749549e-01 -3.19772661e-01
8.79695833e-01 -1.76084321e-02 -2.28415281e-01 3.29585493e-01
8.83693814e-01 3.22720021e-01 3.39160860e-01 1.85818195e-01
-2.47064874e-01 3.89958397e-02 -8.18745613e-01 2.21964285e-01
1.48662961e+00 -1.65797576e-01 3.58821571e-01 7.17537761e-01
9.78713930e-01 -4.22143191e-01 -1.63369608e+00 -5.70342422e-01
-3.13049927e-02 -2.62714356e-01 -6.00008629e-02 -1.05948353e+00
-1.01089919e+00 9.84560430e-01 9.03703332e-01 -2.48674408e-01
1.07037938e+00 -2.36682937e-01 8.06851923e-01 5.25569320e-01
3.70847195e-01 -1.15960431e+00 5.19883037e-01 6.72816575e-01
6.63427889e-01 -1.20784664e+00 1.12445690e-02 -2.54504220e-03
-6.26780510e-01 8.06254745e-01 9.13411736e-01 -5.31245023e-02
7.22722590e-01 -7.69837350e-02 -1.94707930e-01 -1.85816884e-01
-1.06100965e+00 -2.28880599e-01 3.48554224e-01 4.38460380e-01
1.68401003e-01 -2.70088851e-01 4.56313277e-03 9.47443008e-01
2.39591375e-01 8.26774687e-02 2.49868691e-01 1.09368122e+00
-5.35360515e-01 -9.29414451e-01 1.46210492e-01 5.81129611e-01
-2.86218405e-01 -3.46178949e-01 -1.16475500e-01 7.52947688e-01
-2.97637731e-02 5.23090839e-01 -4.32571694e-02 -3.92531037e-01
5.26893377e-01 5.98765135e-01 3.44920963e-01 -7.56350994e-01
-5.91004193e-01 -2.35567257e-01 -3.24907511e-01 -5.68458736e-01
-3.59904729e-02 -4.65373665e-01 -6.16966724e-01 -6.36943728e-02
-2.73296386e-01 4.57312725e-02 4.34402585e-01 9.24267590e-01
6.21226609e-01 8.78374994e-01 6.47540629e-01 -7.06841886e-01
-9.46247160e-01 -9.04708147e-01 -6.18764460e-01 6.14152253e-01
3.92451257e-01 -9.33109105e-01 -5.37725329e-01 -2.25062221e-02]
|
[9.792716979980469, 3.406764268875122]
|
a08c03fa-582c-45b7-ad6a-188e66df65e0
|
ksconf-a-light-weight-test-if-a-convnet
|
1804.04171
| null |
http://arxiv.org/abs/1804.04171v1
|
http://arxiv.org/pdf/1804.04171v1.pdf
|
KS(conf ): A Light-Weight Test if a ConvNet Operates Outside of Its Specifications
|
Computer vision systems for automatic image categorization have become
accurate and reliable enough that they can run continuously for days or even
years as components of real-world commercial applications. A major open problem
in this context, however, is quality control. Good classification performance
can only be expected if systems run under the specific conditions, in
particular data distributions, that they were trained for. Surprisingly, none
of the currently used deep network architectures has a built-in functionality
that could detect if a network operates on data from a distribution that it was
not trained for and potentially trigger a warning to the human users. In this
work, we describe KS(conf), a procedure for detecting such outside of the
specifications operation. Building on statistical insights, its main step is
the applications of a classical Kolmogorov-Smirnov test to the distribution of
predicted confidence values. We show by extensive experiments using ImageNet,
AwA2 and DAVIS data on a variety of ConvNets architectures that KS(conf)
reliably detects out-of-specs situations. It furthermore has a number of
properties that make it an excellent candidate for practical deployment: it is
easy to implement, adds almost no overhead to the system, works with all
networks, including pretrained ones, and requires no a priori knowledge about
how the data distribution could change.
|
['Christoph H. Lampert', 'Rémy Sun']
|
2018-04-11
| null | null | null | null |
['image-categorization']
|
['computer-vision']
|
[-1.19106565e-02 -2.54208714e-01 1.43842444e-01 -7.35715389e-01
-3.91017854e-01 -6.60146594e-01 6.18071675e-01 3.17574441e-01
-6.71155095e-01 5.51050246e-01 -7.74552047e-01 -5.77359676e-01
-5.80850951e-02 -6.39517069e-01 -7.12427437e-01 -6.85188532e-01
-3.61842752e-01 5.13637722e-01 7.69854426e-01 -1.29837707e-01
2.10962832e-01 8.23254108e-01 -1.95788288e+00 2.62836397e-01
4.93084878e-01 1.45775378e+00 2.01984532e-02 8.42420459e-01
2.19526663e-02 6.36474848e-01 -9.30589318e-01 -3.16358417e-01
4.43720102e-01 -7.44169131e-02 -6.44224942e-01 4.07359190e-02
5.49370468e-01 -4.32449847e-01 -2.10169718e-01 1.28272998e+00
3.89797896e-01 -2.15824172e-01 5.56218803e-01 -1.52327478e+00
-4.26871389e-01 4.25906450e-01 -1.81864858e-01 4.18963999e-01
-7.41205812e-02 5.45667350e-01 7.20752120e-01 -6.49605215e-01
3.48341763e-01 9.93847728e-01 6.89095736e-01 2.54377753e-01
-1.31904280e+00 -5.22382796e-01 5.40797189e-02 3.43733132e-01
-1.19935572e+00 -4.39944148e-01 1.99893355e-01 -6.12516344e-01
8.73351157e-01 2.86491483e-01 4.20815200e-01 9.75620806e-01
2.10900500e-01 5.73164523e-01 1.14281213e+00 -4.00137067e-01
6.51771486e-01 5.35252810e-01 2.83411801e-01 4.54872876e-01
4.19683397e-01 1.02993570e-01 -4.72290888e-02 -6.14476316e-02
4.75934684e-01 -2.51555145e-01 -2.35970452e-01 -5.34056783e-01
-8.43742192e-01 7.70644367e-01 3.29558194e-01 5.15242279e-01
-2.71645963e-01 1.95073664e-01 6.91231430e-01 6.51925683e-01
1.89688161e-01 3.55452210e-01 -6.40659511e-01 -8.42790380e-02
-8.70889068e-01 1.42485514e-01 8.54208052e-01 7.46115029e-01
7.62199461e-01 6.76111551e-03 3.68053392e-02 6.14010155e-01
-1.57971941e-02 3.52980942e-01 6.84343159e-01 -7.45471478e-01
-1.17688186e-01 5.46539187e-01 3.43960486e-02 -9.91999805e-01
-3.99771720e-01 -4.62718993e-01 -6.56717181e-01 6.76432371e-01
6.88411713e-01 -1.52620330e-01 -9.31817234e-01 1.48929191e+00
1.31246999e-01 -9.14480761e-02 -3.78600173e-02 6.71479881e-01
5.36298096e-01 2.49899730e-01 -1.01872794e-01 9.15160775e-02
1.13524878e+00 -2.18106568e-01 -3.20739806e-01 -2.88914084e-01
4.60518658e-01 -7.16965199e-01 1.11228144e+00 8.59131634e-01
-6.52049541e-01 -6.10624492e-01 -1.20395994e+00 3.84153515e-01
-6.32896304e-01 1.29113153e-01 5.67611158e-01 7.87233174e-01
-1.26936412e+00 9.29573178e-01 -7.70512640e-01 -4.95633274e-01
4.44490701e-01 4.23519403e-01 -3.98012698e-01 1.53504506e-01
-1.01716769e+00 9.82372880e-01 5.49846411e-01 1.29780725e-01
-9.44364786e-01 -2.98085809e-01 -6.14592195e-01 1.48363501e-01
2.95565754e-01 -1.74976721e-01 1.49223018e+00 -1.49519122e+00
-1.26511145e+00 9.94792223e-01 2.67445415e-01 -6.66281343e-01
8.45276296e-01 -4.29508723e-02 -4.71318573e-01 5.68720810e-02
-5.88894039e-02 5.10972202e-01 1.02036095e+00 -1.06641042e+00
-8.10921669e-01 -3.72410268e-01 1.74206138e-01 -4.05643255e-01
-2.58947402e-01 7.67317936e-02 -4.76856977e-01 -2.62264729e-01
-1.85846105e-01 -8.63262355e-01 -1.13880493e-01 2.49351561e-01
-4.09483224e-01 -2.44773909e-01 7.19540179e-01 -2.76781559e-01
8.01709175e-01 -2.28386617e+00 -6.15229607e-01 4.86450493e-01
-7.50264078e-02 6.39536440e-01 2.25706510e-02 2.52820700e-01
-3.11453372e-01 1.24365814e-01 -1.06047094e-01 4.67611030e-02
2.00308450e-02 2.03022256e-01 -2.83140570e-01 7.39909410e-01
2.55162954e-01 3.74213099e-01 -5.80734074e-01 -1.84854325e-02
3.07942361e-01 2.14684069e-01 -3.14873844e-01 2.28371650e-01
-2.28289574e-01 -6.83571771e-02 -5.63450828e-02 1.73147947e-01
7.28615642e-01 -2.57309228e-01 1.23701237e-01 -8.87211859e-02
-1.26608387e-01 3.63022611e-02 -1.40323496e+00 9.83205557e-01
-2.37295941e-01 9.82151389e-01 3.69507750e-03 -1.19211709e+00
8.85358870e-01 -4.11965400e-02 2.19200164e-01 -6.08886361e-01
4.26273823e-01 1.99141189e-01 3.57053429e-01 -4.68088746e-01
2.60516644e-01 1.44746259e-01 1.84095860e-01 3.16321015e-01
2.11101875e-01 1.58444315e-01 4.23421621e-01 2.06530057e-02
1.33921337e+00 -4.77815390e-01 2.74616778e-01 -4.48405147e-01
4.56507295e-01 1.52929295e-02 3.42347503e-01 1.05875587e+00
-3.67103845e-01 5.00772595e-01 8.52173388e-01 -5.22963226e-01
-9.63847756e-01 -1.01048255e+00 -3.23480725e-01 9.37935710e-01
8.32165256e-02 -9.60732251e-02 -8.04004192e-01 -7.79025376e-01
-5.21971434e-02 6.37145758e-01 -6.09717250e-01 -1.73588768e-01
-9.65871215e-02 -5.69267452e-01 5.64376593e-01 3.72418135e-01
5.19619882e-01 -1.13123977e+00 -8.53108466e-01 1.42994061e-01
4.39212948e-01 -1.00643468e+00 1.64009914e-01 5.12161374e-01
-6.80536211e-01 -1.49310732e+00 -3.83825213e-01 -5.71761668e-01
5.74247241e-01 1.64251432e-01 1.12668657e+00 3.02679598e-01
-4.96841371e-01 2.99057782e-01 -3.49867821e-01 -4.53902304e-01
-6.26691282e-01 -1.65886179e-01 6.07655123e-02 -3.08654699e-02
5.83150744e-01 -3.93103749e-01 -5.40270507e-01 5.92974126e-01
-9.82122660e-01 -4.86436576e-01 5.91187000e-01 7.45074749e-01
2.25962445e-01 6.07606530e-01 4.72044885e-01 -9.31103349e-01
5.48230290e-01 -3.50766629e-01 -9.77124393e-01 1.64757341e-01
-7.17136919e-01 4.51816358e-02 7.42250621e-01 -4.40926850e-01
-5.24701774e-01 8.15771520e-02 -3.44835699e-01 -3.70433390e-01
-7.37751544e-01 3.06666195e-01 -1.65067106e-01 1.21971339e-01
9.65281725e-01 1.06262252e-01 1.59965679e-01 -3.77216935e-01
1.67194560e-01 7.71573961e-01 5.97341418e-01 -2.20429346e-01
7.74633229e-01 3.72843325e-01 -3.11143965e-01 -9.82578695e-01
-5.22231758e-01 -4.81606632e-01 -5.29747188e-01 -2.16457888e-01
5.92355728e-01 -6.41746700e-01 -8.70484531e-01 8.23661625e-01
-1.05179203e+00 -4.70552623e-01 -6.12718239e-02 1.74952611e-01
-2.16439798e-01 2.85270900e-01 -2.45739147e-01 -7.64674842e-01
-3.17404866e-02 -1.23658597e+00 3.93214077e-01 2.55275548e-01
-6.35745749e-02 -8.43815982e-01 -3.95023346e-01 -2.19775796e-01
5.42494297e-01 9.44583789e-02 8.09121668e-01 -1.04160571e+00
-3.14751834e-01 -5.64986050e-01 -2.75133342e-01 9.92523074e-01
1.17092924e-02 4.86295491e-01 -1.22074747e+00 -3.94956380e-01
-2.81074550e-02 -2.55397677e-01 7.30511904e-01 3.88306707e-01
1.58171690e+00 -2.79813230e-01 1.26982713e-02 3.49997431e-01
1.36355364e+00 3.04760844e-01 6.64204478e-01 5.67633808e-01
2.01310456e-01 5.20713568e-01 5.01058161e-01 3.00609291e-01
-4.58699651e-02 5.98461092e-01 6.25535667e-01 -4.26066779e-02
2.83013076e-01 2.62390167e-01 3.80069911e-01 5.01450188e-02
3.49474400e-01 -2.10376963e-01 -8.60981107e-01 4.55524236e-01
-1.78685939e+00 -8.61530423e-01 -2.54431039e-01 2.50313187e+00
6.69605732e-01 7.71406531e-01 1.64328709e-01 3.51955622e-01
8.23279202e-01 -3.12623799e-01 -7.24804580e-01 -4.83949423e-01
1.19420424e-01 2.47127160e-01 7.16339588e-01 1.36522830e-01
-1.27972388e+00 6.52866602e-01 6.43832731e+00 6.80069745e-01
-1.32941115e+00 -1.79459497e-01 8.25739980e-01 2.56726623e-01
2.21938148e-01 -1.64903268e-01 -5.71650624e-01 4.83109027e-01
9.30635214e-01 -2.93231346e-02 1.70563385e-01 1.41462278e+00
1.95600331e-01 -4.70516086e-01 -1.44080091e+00 1.01563084e+00
-4.30736784e-03 -1.18663752e+00 -3.13085556e-01 -1.89989675e-02
2.66910791e-01 3.68437558e-01 8.74606296e-02 3.60968113e-01
4.12110507e-01 -1.00355136e+00 8.92464161e-01 1.16302058e-01
7.25616813e-01 -7.38568187e-01 9.96996880e-01 4.93089586e-01
-4.96722192e-01 -4.36165333e-02 -5.67850828e-01 -4.47723735e-03
-3.65101993e-01 8.57518733e-01 -1.03501511e+00 1.17985457e-01
9.48472917e-01 2.83051223e-01 -9.37052250e-01 1.31289411e+00
-1.93726033e-01 6.01736128e-01 -4.26045358e-01 -3.81823741e-02
1.63852200e-01 3.18494231e-01 3.26487184e-01 1.32262671e+00
1.93171233e-01 -3.58054370e-01 7.03383163e-02 5.07128000e-01
1.57851994e-01 -2.02523664e-01 -7.09634006e-01 1.29977256e-01
3.17459643e-01 1.27956295e+00 -1.01859319e+00 -3.61701161e-01
-2.52278209e-01 8.29724610e-01 2.38842100e-01 1.42571092e-01
-6.61737263e-01 -5.52658379e-01 7.46046603e-01 8.67412016e-02
5.14030337e-01 -5.63214310e-02 -9.54342037e-02 -1.06165874e+00
5.55762127e-02 -1.05071509e+00 3.87094349e-01 -5.22035420e-01
-1.51286542e+00 8.38734508e-01 -8.70237052e-02 -1.14418936e+00
-3.36999476e-01 -1.18014467e+00 -6.03607297e-01 6.65113389e-01
-1.34317732e+00 -4.04029876e-01 -4.17012870e-01 7.11050510e-01
3.94911468e-01 -3.22841257e-01 7.41361737e-01 2.88434118e-01
-5.15134215e-01 6.41318321e-01 5.98127320e-02 3.22905838e-01
7.22979009e-01 -1.40106380e+00 3.47855270e-01 1.06070054e+00
2.50972331e-01 4.95397627e-01 1.14143145e+00 -2.00247839e-01
-1.06272995e+00 -1.02830946e+00 3.25028360e-01 -1.84911102e-01
7.37439275e-01 -4.19193387e-01 -1.08220208e+00 4.19110149e-01
3.84572074e-02 1.52045995e-01 4.44423497e-01 1.60825700e-01
-4.59751874e-01 -3.83292764e-01 -1.10867178e+00 2.56297886e-01
4.87047404e-01 -2.52291352e-01 -4.46294636e-01 3.81714046e-01
1.52050823e-01 -3.57694954e-01 -4.89538431e-01 2.40300924e-01
4.00781274e-01 -1.56601667e+00 6.57704294e-01 -5.67831814e-01
2.15816274e-01 -2.69644469e-01 -5.62030338e-02 -1.34609330e+00
-1.31390914e-01 -4.39171284e-01 2.44981393e-01 1.19009912e+00
4.10766512e-01 -7.58384466e-01 6.22705400e-01 8.40026259e-01
-2.52400432e-02 -4.51343447e-01 -7.77389526e-01 -1.02029383e+00
-2.10397422e-01 -9.27163720e-01 3.82644713e-01 7.73432910e-01
-5.06628573e-01 6.59753680e-02 -3.43313180e-02 4.79140162e-01
5.18311322e-01 -2.70683318e-01 9.41224337e-01 -1.31695580e+00
-4.92986262e-01 -5.83426654e-01 -9.24376488e-01 -7.30613351e-01
6.50781617e-02 -5.65947533e-01 2.53086388e-01 -1.17987514e+00
-2.08974108e-01 -5.98934710e-01 -4.23842132e-01 6.24654710e-01
1.23607397e-01 2.98317343e-01 1.03699163e-01 1.31590560e-01
-5.16131461e-01 1.40305534e-01 4.80468154e-01 -5.02687469e-02
-2.27693766e-02 4.01979208e-01 -5.86885214e-01 9.84772444e-01
8.32925856e-01 -4.95643765e-01 -2.91340530e-01 -1.68382093e-01
1.74084827e-01 -4.59710926e-01 5.37218034e-01 -1.31103516e+00
9.32346955e-02 1.12449378e-02 4.36370015e-01 -4.12805974e-01
-8.87291059e-02 -1.07784283e+00 4.88618538e-02 4.74411964e-01
-3.84072751e-01 4.35591415e-02 3.56661469e-01 4.83282506e-01
-1.38137117e-01 -5.37017882e-01 1.20693374e+00 3.20585165e-03
-1.03465319e+00 1.82604119e-01 -5.86088717e-01 5.53955557e-03
1.21658397e+00 -9.03674364e-02 -4.03830320e-01 -2.76663572e-01
-6.01367354e-01 1.85858011e-01 5.08095562e-01 3.88854146e-01
4.15020078e-01 -9.30649221e-01 -4.00542319e-01 3.53305787e-01
3.00949663e-01 -1.33173421e-01 3.84686887e-02 5.83245993e-01
-7.78610528e-01 4.65850858e-03 -2.49271229e-01 -1.02096438e+00
-1.35278928e+00 4.60892320e-01 5.34159899e-01 8.09151977e-02
-6.64218962e-01 6.80980086e-01 1.91422794e-02 -1.36864677e-01
5.08879900e-01 -4.47506934e-01 -1.06571972e-01 8.02646056e-02
9.40262377e-01 -3.00548505e-02 5.71660936e-01 -2.70650476e-01
-4.65347350e-01 -1.16437741e-01 -2.87014276e-01 7.99060892e-03
1.37267733e+00 1.24818988e-01 -3.15913721e-03 6.23653769e-01
1.09789169e+00 -4.66708750e-01 -1.31421745e+00 1.37589443e-02
1.18675411e-01 -6.04209185e-01 1.25275746e-01 -1.01869893e+00
-1.18524432e+00 9.29730415e-01 9.25770521e-01 8.12690973e-01
1.12442875e+00 -4.22491767e-02 2.55965531e-01 6.62566841e-01
2.98751473e-01 -1.19493806e+00 -6.29490316e-02 5.67097902e-01
7.22260118e-01 -1.32706785e+00 -2.28840813e-01 4.81281020e-02
-6.45801544e-01 1.43791497e+00 5.86676300e-01 -2.88023829e-01
7.72817075e-01 3.11091572e-01 3.35634559e-01 -2.25733578e-01
-7.86720693e-01 -1.64746374e-01 1.75305068e-01 7.50193596e-01
6.72031790e-02 -8.34787712e-02 3.33290249e-02 1.07792616e-01
-1.46793738e-01 1.12070134e-02 7.75150239e-01 6.75633252e-01
-6.62117064e-01 -8.17128778e-01 -2.06832886e-01 6.67524755e-01
-5.55436432e-01 1.44321889e-01 -2.92316794e-01 9.79748487e-01
2.11184651e-01 9.62477744e-01 2.40389869e-01 -4.16726232e-01
4.14470553e-01 -3.76012474e-02 1.87756449e-01 -5.91354311e-01
-4.62004721e-01 -2.17403933e-01 8.05530325e-02 -7.88236976e-01
-1.89948738e-01 -6.33898735e-01 -1.00993693e+00 -4.29202348e-01
-3.14642578e-01 -1.18532598e-01 1.02093172e+00 9.59205866e-01
1.64701894e-01 3.28911543e-01 6.28765941e-01 -7.14279354e-01
-7.60766447e-01 -8.50899816e-01 -7.09095836e-01 4.81281251e-01
3.37210983e-01 -6.01792276e-01 -6.89421177e-01 -8.55771527e-02]
|
[8.91755199432373, 2.667086124420166]
|
2a51ea3d-4ebf-4204-bfc3-ac905d942b05
|
symmetric-uncertainty-aware-feature
|
2306.00386
| null |
https://arxiv.org/abs/2306.00386v1
|
https://arxiv.org/pdf/2306.00386v1.pdf
|
Symmetric Uncertainty-Aware Feature Transmission for Depth Super-Resolution
|
Color-guided depth super-resolution (DSR) is an encouraging paradigm that enhances a low-resolution (LR) depth map guided by an extra high-resolution (HR) RGB image from the same scene. Existing methods usually use interpolation to upscale the depth maps before feeding them into the network and transfer the high-frequency information extracted from HR RGB images to guide the reconstruction of depth maps. However, the extracted high-frequency information usually contains textures that are not present in depth maps in the existence of the cross-modality gap, and the noises would be further aggravated by interpolation due to the resolution gap between the RGB and depth images. To tackle these challenges, we propose a novel Symmetric Uncertainty-aware Feature Transmission (SUFT) for color-guided DSR. (1) For the resolution gap, SUFT builds an iterative up-and-down sampling pipeline, which makes depth features and RGB features spatially consistent while suppressing noise amplification and blurring by replacing common interpolated pre-upsampling. (2) For the cross-modality gap, we propose a novel Symmetric Uncertainty scheme to remove parts of RGB information harmful to the recovery of HR depth maps. Extensive experiments on benchmark datasets and challenging real-world settings suggest that our method achieves superior performance compared to state-of-the-art methods. Our code and models are available at https://github.com/ShiWuxuan/SUFT.
|
['Bo Du', 'Mang Ye', 'Wuxuan Shi']
|
2023-06-01
| null | null | null | null |
['super-resolution']
|
['computer-vision']
|
[ 4.67408061e-01 -9.81328413e-02 2.71495610e-01 -2.34245926e-01
-1.00641322e+00 -5.61550073e-02 2.27473974e-01 -4.19123232e-01
-5.80384098e-02 8.50082517e-01 4.14306581e-01 2.34720513e-01
-1.36357665e-01 -1.08172417e+00 -7.11886942e-01 -8.62435997e-01
2.46047392e-01 -1.94926590e-01 5.16190827e-01 -2.31778771e-01
2.36310527e-01 3.39268118e-01 -1.92835402e+00 5.35028994e-01
1.18894720e+00 1.21410418e+00 6.06107056e-01 4.45936054e-01
-2.62675375e-01 8.78476083e-01 -1.91418946e-01 5.48325367e-02
4.46498275e-01 -2.48714924e-01 -7.17774689e-01 -3.99637558e-02
4.35688317e-01 -9.19378221e-01 -7.60537803e-01 1.31638002e+00
4.98115063e-01 2.38557592e-01 -3.67927700e-02 -8.54819894e-01
-8.68964493e-01 2.67971694e-01 -9.64941502e-01 1.22745328e-01
5.19826293e-01 1.08708635e-01 3.53719503e-01 -1.10653782e+00
6.77966774e-01 1.29011130e+00 4.78907526e-01 4.70145911e-01
-1.08136821e+00 -7.91190326e-01 1.50180370e-01 2.20343307e-01
-1.40622556e+00 -4.19933707e-01 9.60190773e-01 -2.14466751e-02
4.73175585e-01 2.01596498e-01 6.60848081e-01 1.00659585e+00
-7.11380541e-02 4.42579865e-01 1.43326497e+00 -1.44437179e-01
1.65725976e-01 -2.40731552e-01 -2.85555422e-01 4.56028432e-01
2.24133097e-02 4.96089965e-01 -1.04070735e+00 7.50640556e-02
1.34427512e+00 1.06907867e-01 -8.04330885e-01 -2.08696537e-02
-1.18950629e+00 3.37558150e-01 7.85937369e-01 1.27661020e-01
-3.12696934e-01 1.06828347e-01 -1.17215835e-01 -2.82075349e-02
6.35559320e-01 -3.67190912e-02 -3.72189015e-01 -3.76743190e-02
-7.75581360e-01 -6.69811368e-02 6.35886714e-02 9.79020238e-01
1.35536408e+00 -1.63880691e-01 -1.59968346e-01 8.76174808e-01
2.15972215e-01 5.24484515e-01 3.21561992e-01 -1.36534607e+00
5.42970300e-01 3.56595546e-01 2.49745622e-01 -7.72204697e-01
-1.74056932e-01 -1.50724426e-01 -1.12264073e+00 2.96425015e-01
4.12689835e-01 2.05626130e-01 -1.07445598e+00 1.39771593e+00
5.24781406e-01 5.39219260e-01 1.30680278e-01 1.19324195e+00
1.12674010e+00 7.11603642e-01 -5.44336140e-01 -1.98984131e-01
1.11363018e+00 -7.47684062e-01 -8.09518516e-01 -2.10461453e-01
-1.95687100e-01 -7.40887582e-01 1.04486310e+00 5.22175431e-01
-1.06759417e+00 -6.28140748e-01 -1.04253662e+00 -6.88361645e-01
-9.36013609e-02 -1.09600216e-01 5.59011400e-01 2.81048179e-01
-1.05826497e+00 8.19114864e-01 -8.04750025e-01 1.51485607e-01
4.32568282e-01 -1.50016949e-01 -3.52043480e-01 -8.88374150e-01
-1.25262916e+00 4.71930206e-01 1.20598301e-01 4.54155445e-01
-6.10791504e-01 -9.41297233e-01 -8.52235675e-01 -4.23597336e-01
4.13243115e-01 -5.39318979e-01 7.57636368e-01 -5.48550844e-01
-1.52936649e+00 3.67245048e-01 -4.55686808e-01 1.98814049e-01
6.22967839e-01 -2.63587743e-01 -3.25697064e-01 4.67643678e-01
1.65412560e-01 5.64240694e-01 9.34763730e-01 -1.54417300e+00
-8.81785095e-01 -5.88114142e-01 3.30949537e-02 4.13982689e-01
2.97661554e-02 -3.95456582e-01 -7.35222399e-01 -5.03871858e-01
8.74192297e-01 -3.88503373e-01 -2.08384559e-01 2.58652389e-01
-4.03224230e-01 4.17472273e-01 6.76536322e-01 -8.09858561e-01
8.89824927e-01 -2.34969521e+00 5.91903143e-02 1.81580633e-02
2.89270967e-01 -2.63870627e-01 -1.92480832e-01 3.25731747e-02
2.55566277e-02 -1.08727135e-01 -3.64210308e-01 -4.78442848e-01
-4.56135184e-01 2.40449965e-01 -3.96881223e-01 5.69762111e-01
4.22640741e-02 5.84286332e-01 -1.10835862e+00 -3.53792310e-01
6.02087319e-01 1.10763466e+00 -2.19527066e-01 1.95780441e-01
-2.02845838e-02 1.01251280e+00 -3.26304466e-01 9.25020516e-01
1.39499092e+00 -8.24583769e-02 -3.63491774e-01 -7.12137282e-01
-4.24649775e-01 3.02150756e-01 -1.45571315e+00 2.16029525e+00
-4.62878317e-01 4.27252084e-01 5.84588461e-02 -1.24599114e-01
8.17423880e-01 -1.45438299e-01 5.42564034e-01 -1.14794266e+00
-1.29597813e-01 2.82418013e-01 -5.85341454e-01 -3.23577136e-01
7.33017266e-01 -4.57820892e-02 3.95816505e-01 1.30301744e-01
-3.15088362e-01 -3.72309238e-01 -3.32873046e-01 7.64601976e-02
8.86689365e-01 3.62068504e-01 -2.34567761e-01 1.77477732e-01
5.78097761e-01 -4.09786850e-01 8.83748889e-01 5.54399908e-01
-1.78062007e-01 1.27409136e+00 4.62946482e-02 -2.30318919e-01
-9.03620005e-01 -1.21522152e+00 -3.63394231e-01 5.14930069e-01
6.99465096e-01 -2.14808092e-01 -3.80620629e-01 -1.58367306e-01
-1.40066206e-01 3.53341371e-01 -7.51206994e-01 5.23571260e-02
-3.90340090e-01 -6.47211134e-01 6.02486245e-02 4.34574425e-01
1.11958683e+00 -6.58896267e-01 -5.08263767e-01 5.70811294e-02
-7.15876281e-01 -1.36284161e+00 -3.03277284e-01 -2.88151447e-02
-9.47677195e-01 -9.62597966e-01 -8.42817187e-01 -2.19126403e-01
6.29114628e-01 7.08000422e-01 9.40603614e-01 -8.48901123e-02
-3.71875167e-01 2.16365412e-01 -4.61816192e-01 1.38097733e-01
1.48303255e-01 -5.50128698e-01 -2.27632061e-01 1.48359567e-01
-2.03798488e-02 -7.20930338e-01 -1.08994520e+00 2.79141337e-01
-1.06520939e+00 3.84199142e-01 4.29468334e-01 8.15709651e-01
1.03526080e+00 4.32200372e-01 8.99879187e-02 -4.71656740e-01
1.70974936e-02 -2.50912428e-01 -7.25406468e-01 -5.26694879e-02
-4.15059626e-01 3.75614315e-02 3.16668659e-01 -2.42496043e-01
-1.52214527e+00 1.01749726e-01 -3.87360752e-02 -6.84552610e-01
-2.37602573e-02 1.43990023e-02 -3.15756083e-01 -3.44171524e-01
4.19207871e-01 3.87931705e-01 -1.96285695e-01 -6.05569780e-01
4.69741195e-01 5.76470971e-01 8.10152173e-01 -5.45418024e-01
8.65103543e-01 1.17061043e+00 1.04089357e-01 -8.12736273e-01
-8.48550737e-01 -4.24962521e-01 -4.18873906e-01 -2.67675519e-01
7.16997743e-01 -1.31130016e+00 -2.66351253e-01 7.14537442e-01
-9.18100715e-01 -4.12068427e-01 -3.26210469e-01 4.44361925e-01
-4.34021324e-01 5.68173468e-01 -7.53824353e-01 -8.22451651e-01
-1.92611396e-01 -1.13461685e+00 1.32701755e+00 6.40473068e-01
5.85477829e-01 -5.10531247e-01 -1.79128647e-01 3.21443915e-01
5.42027771e-01 1.55069575e-01 4.05374974e-01 6.97824836e-01
-1.16401553e+00 3.79655570e-01 -7.36827374e-01 3.45612794e-01
4.24204856e-01 -8.37249309e-02 -1.38642120e+00 -9.39218402e-02
9.73493084e-02 -2.46867999e-01 9.29894388e-01 4.29247230e-01
1.37960875e+00 1.29401296e-01 4.76292260e-02 1.20409393e+00
1.87209284e+00 -1.11465558e-01 1.22434139e+00 4.47773576e-01
9.68480170e-01 6.17393076e-01 9.34497714e-01 5.29835582e-01
5.83933353e-01 6.36038840e-01 5.82940221e-01 -1.85247481e-01
-4.73238558e-01 -3.83752853e-01 2.49778539e-01 2.96348304e-01
-2.67159998e-01 2.78923094e-01 -3.94233823e-01 4.95025009e-01
-1.72374392e+00 -6.60068274e-01 -3.04099113e-01 2.39398241e+00
9.85092580e-01 -2.19303459e-01 -4.02971238e-01 1.22937433e-01
6.81494892e-01 3.61124456e-01 -7.20669508e-01 2.94268101e-01
-5.05613327e-01 3.25868070e-01 7.16743171e-01 6.59023583e-01
-6.94026947e-01 6.95082903e-01 4.89778519e+00 9.65018868e-01
-1.12922466e+00 1.67461246e-01 6.70335412e-01 -2.52538860e-01
-5.86017787e-01 -3.60910222e-02 -5.93181849e-01 5.68658650e-01
4.48670655e-01 1.99656725e-01 8.18868518e-01 4.34165150e-01
2.56666780e-01 -7.53125310e-01 -8.20093513e-01 1.47185695e+00
-1.03406496e-01 -1.18329239e+00 -3.09612334e-01 1.35617480e-02
9.59957659e-01 1.93293124e-01 2.05968022e-01 -2.53840297e-01
2.54387975e-01 -8.65983307e-01 5.79610407e-01 7.97822833e-01
1.22343493e+00 -8.88712943e-01 5.78988850e-01 -8.44090537e-04
-1.42130721e+00 -4.50011566e-02 -7.01732278e-01 2.47526601e-01
1.73727706e-01 1.16541982e+00 1.19732268e-01 1.07020485e+00
1.26497209e+00 9.32539821e-01 -5.09187043e-01 8.94861877e-01
-4.78594273e-01 -1.88195840e-01 -3.53079647e-01 8.87134731e-01
-1.77163571e-01 -3.70580107e-01 3.39415699e-01 5.84575593e-01
5.48477829e-01 4.35279340e-01 -2.54722118e-01 1.19090867e+00
1.32087469e-01 -4.73082751e-01 -3.49111468e-01 4.71725345e-01
6.36929870e-01 1.27888298e+00 -5.50175905e-01 -1.80401117e-01
-4.89677697e-01 1.36515045e+00 5.40723093e-02 5.89487433e-01
-7.42468596e-01 -3.18518013e-01 8.11966002e-01 1.72901377e-01
2.39752695e-01 1.11966692e-02 -5.55389285e-01 -1.35375524e+00
2.97048926e-01 -5.12281239e-01 1.30905479e-01 -1.28566349e+00
-1.28084862e+00 5.97073376e-01 -2.42796928e-01 -1.49927306e+00
1.90912709e-01 -1.47251472e-01 -8.32528174e-02 1.30247307e+00
-2.16738605e+00 -8.88666630e-01 -1.07103372e+00 9.23523188e-01
3.22406650e-01 5.89923918e-01 3.97380739e-01 4.01656538e-01
-4.11058456e-01 2.37638801e-01 1.17631555e-01 -1.95500135e-01
8.30896258e-01 -9.93781149e-01 1.58342645e-01 1.08207023e+00
-4.89482969e-01 3.84236515e-01 4.44796652e-01 -7.56552577e-01
-1.47165382e+00 -1.20056462e+00 2.32494041e-01 -1.43716037e-01
2.44246751e-01 -1.62606910e-01 -1.06459105e+00 2.75380105e-01
-2.67267466e-01 5.86657941e-01 1.25238672e-01 -4.38084841e-01
-5.54532290e-01 -3.93939763e-01 -1.59293258e+00 2.94043601e-01
1.23973572e+00 -8.58770549e-01 -1.70532614e-01 -6.46660849e-02
9.63541150e-01 -8.64543438e-01 -1.03322113e+00 5.87442458e-01
5.79962134e-01 -1.65249324e+00 1.29922903e+00 4.80027705e-01
9.21383977e-01 -8.44113052e-01 -5.28986335e-01 -1.17998266e+00
-1.21511839e-01 -4.62745875e-01 -2.50338107e-01 1.19903123e+00
-1.22532889e-01 -7.26484418e-01 5.19858539e-01 7.35557199e-01
-1.60049111e-01 -5.83307326e-01 -1.06629324e+00 -5.18637598e-01
-3.01807731e-01 -4.60591018e-01 8.35657358e-01 8.23496759e-01
-2.76324660e-01 -2.82186121e-01 -3.50350201e-01 6.85090959e-01
1.20798361e+00 1.89047158e-01 5.05573750e-01 -8.56441021e-01
-1.82994276e-01 -6.41917512e-02 -6.68356419e-02 -1.15727782e+00
-4.79293615e-01 -1.74346030e-01 3.29477996e-01 -1.81345952e+00
8.73687342e-02 -6.37597203e-01 -2.04770088e-01 2.21509531e-01
-2.84480006e-01 5.05980492e-01 -2.23480314e-02 2.46232659e-01
-2.70335138e-01 7.07277656e-01 1.58137095e+00 2.48584881e-01
-2.46914357e-01 -3.73263299e-01 -6.68541729e-01 6.12665951e-01
4.07292604e-01 -1.70427367e-01 -4.73242700e-01 -5.69641292e-01
2.53717721e-01 3.25817913e-01 5.14274776e-01 -1.05727959e+00
2.21234128e-01 -2.34143227e-01 8.34767818e-01 -8.95544827e-01
6.71313763e-01 -8.69886875e-01 3.04217756e-01 -2.54061148e-02
3.31940167e-02 -4.46410388e-01 9.26970020e-02 6.42558634e-01
-2.58905977e-01 2.55542785e-01 1.04040301e+00 -2.01905251e-01
-7.97211587e-01 4.84835804e-01 2.15731859e-01 -1.26025811e-01
6.23476446e-01 -3.19874704e-01 -7.24319875e-01 -3.26137245e-01
-5.05697429e-01 9.84882265e-02 9.40790415e-01 2.47476593e-01
1.07455230e+00 -1.29455483e+00 -5.09754479e-01 5.10348618e-01
1.67292416e-01 8.19765806e-01 9.37262356e-01 7.02052891e-01
-4.47226703e-01 -7.71150887e-02 -2.31401384e-01 -6.41516209e-01
-8.43914032e-01 3.22236568e-01 4.46544886e-01 2.93964129e-02
-1.10946178e+00 1.08385384e+00 3.28739613e-01 -1.29327133e-01
1.87121034e-01 -6.01531863e-01 5.91787137e-02 -1.40105188e-01
1.01408136e+00 5.85373461e-01 1.16796732e-01 -5.45553625e-01
-2.31090471e-01 8.25784624e-01 1.30814403e-01 -2.72628933e-01
1.46939170e+00 -7.84604013e-01 -3.07732880e-01 4.34729487e-01
1.06308806e+00 9.40295234e-02 -1.78352559e+00 -4.96758819e-01
-6.81358099e-01 -1.21293330e+00 5.10557771e-01 -6.78412735e-01
-1.49292374e+00 7.38031030e-01 8.19019794e-01 -1.99210197e-01
1.64359581e+00 -1.16773568e-01 9.64915514e-01 -2.86817223e-01
7.45059967e-01 -1.05002928e+00 -2.76397187e-02 2.89862633e-01
8.14719617e-01 -1.24801195e+00 2.59977341e-01 -8.53482664e-01
-3.72589767e-01 1.10686076e+00 6.51390195e-01 -4.28933427e-02
5.31768084e-01 2.84168571e-01 -6.46815971e-02 -2.57213078e-02
-4.17370230e-01 -3.55720937e-01 -4.44028676e-02 7.92129397e-01
1.50393983e-02 -2.78947353e-01 1.22140296e-01 3.98317903e-01
1.12981275e-02 3.28036517e-01 7.56561458e-01 8.73565435e-01
-2.82541484e-01 -7.49704480e-01 -6.47637069e-01 1.38497412e-01
-7.23524913e-02 -3.13574404e-01 1.87965170e-01 3.71649325e-01
3.50215614e-01 1.06126750e+00 1.47860363e-01 -5.59407651e-01
2.14478090e-01 -4.46097225e-01 7.19426274e-01 -2.74316669e-01
1.13069057e-01 2.68196374e-01 -1.55090153e-01 -1.36219776e+00
-5.90867698e-01 -4.99380708e-01 -1.38140285e+00 -2.78870314e-01
-3.30964662e-02 -3.69796753e-01 6.94930673e-01 5.57829142e-01
4.30346996e-01 7.20872104e-01 7.46253729e-01 -1.20304167e+00
1.26485556e-01 -8.90519440e-01 -8.54876280e-01 2.77053446e-01
7.83783913e-01 -7.57638395e-01 -8.10591102e-01 -2.37420648e-01]
|
[9.704601287841797, -2.4474592208862305]
|
0ca97a4d-41c3-4731-9fd2-ade4096e85cc
|
combinatorial-multi-armed-bandit-with
|
1707.07443
| null |
http://arxiv.org/abs/1707.07443v1
|
http://arxiv.org/pdf/1707.07443v1.pdf
|
Combinatorial Multi-armed Bandit with Probabilistically Triggered Arms: A Case with Bounded Regret
|
In this paper, we study the combinatorial multi-armed bandit problem (CMAB)
with probabilistically triggered arms (PTAs). Under the assumption that the arm
triggering probabilities (ATPs) are positive for all arms, we prove that a
class of upper confidence bound (UCB) policies, named Combinatorial UCB with
exploration rate $\kappa$ (CUCB-$\kappa$), and Combinatorial Thompson Sampling
(CTS), which estimates the expected states of the arms via Thompson sampling,
achieve bounded regret. In addition, we prove that CUCB-$0$ and CTS incur
$O(\sqrt{T})$ gap-independent regret. These results improve the results in
previous works, which show $O(\log T)$ gap-dependent and $O(\sqrt{T\log T})$
gap-independent regrets, respectively, under no assumptions on the ATPs. Then,
we numerically evaluate the performance of CUCB-$\kappa$ and CTS in a
real-world movie recommendation problem, where the actions correspond to
recommending a set of movies, the arms correspond to the edges between the
movies and the users, and the goal is to maximize the total number of users
that are attracted by at least one movie. Our numerical results complement our
theoretical findings on bounded regret. Apart from this problem, our results
also directly apply to the online influence maximization (OIM) problem studied
in numerous prior works.
|
['A. Ömer Sarıtaç', 'Cem Tekin']
|
2017-07-24
| null | null | null | null |
['movie-recommendation']
|
['miscellaneous']
|
[-6.59666955e-02 1.41319498e-01 -6.98466718e-01 -1.86914057e-01
-1.04089153e+00 -9.95425940e-01 -1.17725387e-01 7.13927448e-02
-3.35090369e-01 1.03397417e+00 -1.76394939e-01 -7.66495943e-01
-1.09780931e+00 -9.90447700e-01 -1.20958126e+00 -8.69547725e-01
-2.33635813e-01 7.94131219e-01 -5.69176488e-02 -1.35409623e-01
1.96954310e-01 3.03216249e-01 -9.46071386e-01 7.55493641e-02
6.19193852e-01 1.57295537e+00 -7.78324902e-02 5.15070200e-01
1.48048811e-02 5.93161643e-01 -2.55419552e-01 -8.58650923e-01
5.16584992e-01 -4.82693881e-01 -8.26593399e-01 4.06869268e-03
-3.28065634e-01 -4.02342975e-01 -2.92629451e-01 1.15396774e+00
1.09005384e-01 4.44301456e-01 5.09360373e-01 -1.16219664e+00
-3.79372865e-01 1.20254004e+00 -1.23607504e+00 1.99075475e-01
1.12318307e-01 -3.43210667e-01 1.31575334e+00 -1.22324705e-01
2.51942426e-01 1.04072082e+00 3.04511368e-01 4.59458917e-01
-1.05668533e+00 -5.07568419e-01 5.99352956e-01 -5.98985851e-02
-9.75637615e-01 2.14759000e-02 4.11723644e-01 -1.31242186e-01
4.47887629e-01 8.37112010e-01 5.90465605e-01 4.02156562e-01
-2.90584981e-01 1.08968568e+00 9.51197684e-01 -6.35493219e-01
4.89168614e-01 1.38036191e-01 3.55417937e-01 4.29059505e-01
5.88357151e-01 1.93957195e-01 -2.73338437e-01 -3.00067812e-01
7.20204055e-01 2.00621083e-01 -3.09012562e-01 -4.75354604e-02
-6.10319674e-01 1.10546267e+00 2.78068095e-01 -2.97353622e-02
-5.87541580e-01 4.38639641e-01 -6.94138110e-02 3.13053966e-01
4.87825215e-01 1.93805367e-01 -4.31720674e-01 -1.06083415e-01
-6.30320132e-01 9.80223492e-02 8.09519231e-01 1.35438156e+00
3.02979112e-01 -3.42261642e-01 -6.28396630e-01 5.64264297e-01
2.97775775e-01 7.62600601e-01 -3.23244363e-01 -1.07410884e+00
1.00886917e+00 5.46230562e-02 1.10440099e+00 -7.81576872e-01
-1.15886435e-01 -5.09934366e-01 -5.32787979e-01 -2.87061244e-01
7.39032567e-01 -5.43431044e-01 -4.65064079e-01 1.83585000e+00
4.31128949e-01 -4.85553801e-01 -5.04160762e-01 1.08378601e+00
2.07092166e-02 6.58873677e-01 -4.96675104e-01 -8.56461465e-01
1.18367863e+00 -8.61433327e-01 -6.88286841e-01 -2.04077765e-01
4.52194721e-01 -5.38165212e-01 7.59445786e-01 6.81701243e-01
-1.48176229e+00 1.67983741e-01 -8.18624198e-01 7.18744695e-01
1.24908552e-01 -1.79373860e-01 9.51573908e-01 1.12290359e+00
-6.58276796e-01 5.29117286e-01 -5.38433731e-01 1.21848039e-01
5.12706459e-01 4.02418077e-01 4.44687188e-01 -3.47680420e-01
-7.96014428e-01 2.75031179e-01 -1.62006676e-01 2.60813504e-01
-9.89050686e-01 -3.05819154e-01 -1.69192076e-01 1.62814334e-01
1.05094409e+00 -4.23843443e-01 1.26443624e+00 -7.74069011e-01
-1.47542775e+00 2.32406110e-01 -6.03190474e-02 -4.58747655e-01
6.46802187e-01 -5.53996712e-02 2.30268911e-01 -1.54947471e-02
-3.49962190e-02 -1.46543244e-02 3.52202922e-01 -1.10267973e+00
-1.02659452e+00 -6.33741081e-01 5.98088384e-01 -3.50021981e-02
-4.03430730e-01 2.16728002e-01 -5.82769990e-01 -4.76545662e-01
1.02096442e-02 -9.61193323e-01 -6.85686707e-01 -6.00593686e-01
-5.69097161e-01 -1.59456477e-01 -1.75950855e-01 -2.05914512e-01
1.39253342e+00 -1.79678810e+00 2.66329199e-01 6.45222247e-01
-2.32247576e-01 -3.62623602e-01 -1.26487032e-01 4.06212777e-01
4.26969975e-01 3.17537785e-01 3.52284759e-01 -1.92497686e-01
2.05092803e-01 2.18534902e-01 -4.32963759e-01 4.70292777e-01
-8.00546706e-01 7.67875969e-01 -6.83847785e-01 -1.09173454e-01
-1.47191659e-01 -4.97601777e-01 -6.19275749e-01 -8.30835849e-03
-5.40018499e-01 2.37051807e-02 -8.41564536e-01 4.33489174e-01
7.52414048e-01 -5.45253932e-01 5.49144924e-01 3.41836244e-01
3.07490993e-02 -3.58254127e-02 -1.29775584e+00 9.62195337e-01
-4.94462579e-01 -6.85315207e-02 3.57803196e-01 -1.02088892e+00
4.01637673e-01 6.73402920e-02 5.21402955e-01 -4.51254785e-01
5.37095129e-01 6.16467409e-02 -3.65532875e-01 -1.66192785e-01
1.94314048e-01 -3.89026195e-01 -2.34310076e-01 1.00920570e+00
-4.62286025e-01 1.76925048e-01 3.05336207e-01 3.79498482e-01
9.94368672e-01 -4.59344327e-01 7.77589828e-02 -2.01074436e-01
-1.58586428e-02 -5.86129203e-02 4.75580543e-01 1.47139406e+00
1.00809611e-01 5.56628220e-02 1.02477658e+00 -1.57057375e-01
-8.33785415e-01 -8.88682961e-01 6.29181415e-02 1.36331737e+00
5.14574170e-01 1.23426042e-01 -6.32828653e-01 -6.62782311e-01
1.56740457e-01 9.56603289e-01 -9.62528169e-01 2.97482252e-01
8.68499093e-03 -9.37638342e-01 -1.58773288e-01 5.44882953e-01
1.01669312e-01 -5.39845169e-01 -2.76356220e-01 3.81746411e-01
-3.16651046e-01 -9.16035354e-01 -7.96873033e-01 9.05536488e-02
-8.60715330e-01 -9.53425288e-01 -6.78680718e-01 -9.43984091e-02
8.46264064e-01 4.83725280e-01 6.31760538e-01 -3.69675845e-01
8.91237482e-02 6.00959122e-01 -7.54695654e-01 -6.07118726e-01
1.65188447e-01 -2.89874882e-01 -9.42750648e-02 1.81726232e-01
4.86697927e-02 -2.78410763e-01 -8.88507485e-01 7.34736860e-01
-7.16340840e-01 -9.54273045e-02 4.65671837e-01 6.41360223e-01
9.02655423e-01 1.51382342e-01 6.10760391e-01 -9.41396356e-01
5.39033890e-01 -6.08046353e-01 -1.20592022e+00 6.02671027e-01
-6.62351966e-01 -8.26635584e-02 5.39885461e-01 -5.82613409e-01
-9.17488635e-01 -2.30468392e-01 3.14962715e-01 -3.08485121e-01
4.00247455e-01 5.79006016e-01 1.28181845e-01 1.05297901e-01
4.74598885e-01 -5.25338985e-02 -3.65845948e-01 -5.13504863e-01
3.57100755e-01 7.08265483e-01 -9.13910568e-02 -8.58439326e-01
2.72082478e-01 7.59419382e-01 1.23296574e-01 -8.12988132e-02
-1.58440721e+00 -2.69655108e-01 2.62139022e-01 -2.88359076e-01
4.05386060e-01 -3.81929785e-01 -1.42597187e+00 -1.37589246e-01
-6.94172800e-01 -3.02081347e-01 -2.39661455e-01 6.46566331e-01
-8.76858711e-01 3.11659276e-01 -5.64313471e-01 -1.80890274e+00
-3.43750954e-01 -8.48905146e-01 3.74367982e-01 3.62479389e-01
3.10275316e-01 -4.56111133e-01 -3.41688901e-01 9.40717757e-01
1.76327899e-01 9.82004479e-02 7.74896562e-01 -3.51959527e-01
-7.55276620e-01 -4.45110053e-01 -2.03893706e-01 1.09139174e-01
-1.78347483e-01 -4.90684360e-01 -2.49036491e-01 -5.26934028e-01
-7.00108632e-02 -2.36343816e-01 5.80639839e-01 1.04971588e+00
1.37554312e+00 -9.93663549e-01 -5.84975719e-01 1.72777221e-01
1.44913507e+00 6.71480179e-01 4.25489873e-01 3.44768792e-01
-9.24818888e-02 3.69780868e-01 9.12168682e-01 1.14022255e+00
-5.79020120e-02 6.63209438e-01 7.71754563e-01 4.00988817e-01
6.56535208e-01 -9.50114951e-02 1.83607206e-01 1.39128298e-01
-2.96617746e-01 -7.77776003e-01 -2.00853512e-01 7.00140774e-01
-2.13556504e+00 -7.55087912e-01 -6.06067590e-02 2.78684354e+00
8.80744636e-01 3.12458962e-01 5.55350244e-01 -2.34778393e-02
8.93007815e-01 -2.96399206e-01 -7.67111897e-01 -5.85458398e-01
6.33048243e-04 2.82176286e-01 1.06169224e+00 5.35202861e-01
-7.70278096e-01 4.94652301e-01 5.73040199e+00 1.27619076e+00
-3.40985060e-01 3.87515247e-01 1.05617559e+00 -1.17851198e+00
-3.93566072e-01 1.17529575e-02 -7.87845492e-01 6.72302663e-01
7.75802076e-01 -1.62850246e-01 1.02839684e+00 9.58514631e-01
2.21095756e-01 -4.14101094e-01 -9.23687756e-01 6.26390874e-01
-3.21935773e-01 -1.39077020e+00 -3.20640832e-01 4.47302043e-01
1.05628538e+00 -1.80142388e-01 2.88358301e-01 -1.69904716e-02
1.09241116e+00 -7.88453281e-01 8.00585330e-01 2.44345114e-01
6.64407790e-01 -1.21463668e+00 7.65735984e-01 5.61146498e-01
-7.57256269e-01 -7.52669036e-01 -5.37805259e-01 -7.01588765e-02
5.54795921e-01 8.17007780e-01 -2.37857953e-01 7.10544050e-01
9.46100950e-01 -1.75318018e-01 4.83694106e-01 1.15528572e+00
-3.03135872e-01 7.13510156e-01 -7.10315526e-01 -7.78048933e-01
4.90508169e-01 -5.38165808e-01 4.09794211e-01 5.65588295e-01
2.91471362e-01 6.27808213e-01 -5.13556078e-02 5.14843524e-01
-1.41066492e-01 1.94422349e-01 1.84218243e-01 1.16001725e-01
4.91852939e-01 9.50350761e-01 -8.78070295e-01 -4.92870174e-02
-1.24986626e-01 5.59233665e-01 2.63202935e-01 3.30778360e-01
-1.21769953e+00 -3.19417328e-01 4.03883964e-01 8.62938240e-02
8.78799558e-01 1.87530994e-01 -6.05964422e-01 -6.11714780e-01
6.07788526e-02 -4.00060296e-01 8.29260767e-01 -4.66726929e-01
-1.30212629e+00 8.53105634e-02 9.23238620e-02 -8.51213574e-01
1.96166709e-01 -3.65444958e-01 -2.65808761e-01 5.57629526e-01
-1.10378253e+00 -6.86552644e-01 3.83040845e-01 4.06315207e-01
1.93103254e-01 2.70181030e-01 3.02917391e-01 1.30760729e-01
-7.08531678e-01 8.34817111e-01 8.81557643e-01 -1.52795553e-01
-1.12363905e-01 -8.77727449e-01 -4.30763453e-01 6.65070713e-01
5.45398071e-02 4.26374853e-01 8.26427341e-01 -3.66996348e-01
-1.45897257e+00 -7.62811363e-01 1.52588144e-01 -2.61978686e-01
6.72227800e-01 -9.77403373e-02 1.81349054e-01 7.81227946e-01
-1.65246025e-01 -1.60177186e-01 6.76850736e-01 5.42071104e-01
-1.95264518e-01 -3.39008003e-01 -1.31295359e+00 5.55979133e-01
1.20244563e+00 2.79629767e-01 8.77374932e-02 6.01133585e-01
7.05637217e-01 -3.19445372e-01 -8.31504345e-01 2.88598180e-01
7.65938163e-01 -8.74323308e-01 7.34070420e-01 -1.13119686e+00
2.21842349e-01 2.20314354e-01 -4.14315045e-01 -9.55917954e-01
-4.75429803e-01 -7.76630282e-01 -1.11349650e-01 9.35843229e-01
8.58014941e-01 -6.15228653e-01 9.15009141e-01 8.26286197e-01
3.09634060e-01 -1.12565923e+00 -1.49182618e+00 -1.00878596e+00
1.58357143e-01 -5.15576661e-01 5.96344173e-01 4.90225941e-01
3.23041886e-01 -1.84174068e-02 -7.51550555e-01 1.07240967e-01
7.16700435e-01 7.31004536e-01 3.81674945e-01 -6.80924296e-01
-1.04494369e+00 -5.71549654e-01 4.72283542e-01 -1.36189771e+00
-5.07044971e-01 -4.07012522e-01 -4.14092578e-02 -1.51200044e+00
7.08429694e-01 -8.33882511e-01 -6.08100057e-01 5.10389328e-01
1.69675395e-01 1.12970080e-02 2.03630149e-01 1.67378900e-03
-1.07709575e+00 3.29032958e-01 1.38680995e+00 -1.05475150e-01
-3.10717747e-02 7.30865359e-01 -1.07886088e+00 4.43770200e-01
4.88746494e-01 -7.34487355e-01 -3.78288180e-01 -4.08057898e-01
9.88163531e-01 8.52800786e-01 -2.58584559e-01 -5.78221530e-02
1.17246680e-01 -7.15957344e-01 -5.69769666e-02 -7.92466104e-01
2.83210784e-01 -7.27284849e-01 1.26706570e-01 4.85697567e-01
-4.53237623e-01 -4.54485625e-01 -9.52431560e-03 1.10563922e+00
5.53589523e-01 -7.14626253e-01 6.32115304e-01 -1.58468902e-01
4.57757682e-01 4.15623903e-01 -3.45632553e-01 -1.16753437e-01
1.36368740e+00 6.52488172e-02 -4.25625801e-01 -9.35060620e-01
-8.40910733e-01 5.59895813e-01 -2.61225373e-01 -1.58352718e-01
2.63337970e-01 -1.05906796e+00 -4.18622702e-01 -5.14764011e-01
-2.13724315e-01 -1.55855045e-01 6.31429672e-01 1.09088945e+00
8.33069161e-03 6.27854526e-01 4.64410514e-01 -8.99110138e-02
-8.60640287e-01 9.69376922e-01 2.26484627e-01 -5.68785489e-01
2.77587801e-01 1.21977961e+00 -1.18529864e-01 9.37221348e-02
3.95467877e-01 -7.74687305e-02 2.01058596e-01 -1.31992206e-01
4.72376317e-01 7.87669599e-01 -2.30560198e-01 2.51134127e-01
-1.54948905e-01 1.74382329e-01 -2.98486263e-01 -5.48109233e-01
1.51121759e+00 -3.51344198e-01 -1.62813693e-01 -1.72638074e-02
5.92702091e-01 2.24655256e-01 -9.75834966e-01 -2.33379155e-01
-1.92703351e-01 -8.07059050e-01 8.64813477e-02 -1.10321057e+00
-1.24252605e+00 2.90889144e-01 1.46117121e-01 8.01342189e-01
1.10090983e+00 4.57106739e-01 5.54682374e-01 2.55226225e-01
1.02423275e+00 -1.18692875e+00 -1.08066194e-01 1.22227989e-01
6.69860125e-01 -8.59317601e-01 3.85634303e-02 -4.68580633e-01
-5.47855675e-01 5.53382695e-01 4.26087558e-01 1.05520532e-01
7.59096742e-01 1.20968603e-01 -7.02188492e-01 8.53228346e-02
-8.12755644e-01 -2.46899217e-01 -2.45753583e-02 -5.19104563e-02
-8.62923414e-02 4.42691863e-01 -9.19801414e-01 1.39779770e+00
-9.18756332e-03 1.05589591e-01 5.04216015e-01 6.91917956e-01
-5.57829082e-01 -1.05620790e+00 -5.46608567e-01 9.54806924e-01
-8.49464834e-01 2.57886034e-02 -1.95326746e-01 4.00712371e-01
-4.05388594e-01 1.55599654e+00 -4.27463613e-02 -2.48150170e-01
2.65335023e-01 -4.69722658e-01 8.45052600e-01 -1.25538781e-01
-2.31157437e-01 4.08058584e-01 2.37923577e-01 -3.38071495e-01
-7.89163858e-02 -5.10058999e-01 -8.43089640e-01 -6.53113127e-01
-9.47729290e-01 6.93938494e-01 5.40499032e-01 9.40025210e-01
2.64109999e-01 1.61513731e-01 1.17714071e+00 -2.27929428e-01
-1.06005490e+00 -1.01929808e+00 -1.06054354e+00 1.91659406e-02
-2.67017454e-01 -6.54008210e-01 -5.60309708e-01 -7.10101843e-01]
|
[4.57266902923584, 3.3716907501220703]
|
a79d9c4f-93c9-42cc-a4fb-83491c7b78c4
|
an-energy-activity-dataset-for-smart-homes
|
2208.13416
| null |
https://arxiv.org/abs/2208.13416v2
|
https://arxiv.org/pdf/2208.13416v2.pdf
|
An Energy Activity Dataset for Smart Homes
|
A smart home energy dataset that records miscellaneous energy consumption data is publicly offered. The proposed energy activity dataset (EAD) has a high data type diversity in contrast to existing load monitoring datasets. In EAD, a simple data point is labeled with the appliance, brand, and event information, whereas a complex data point has an extra application label. Several discoveries have been made on the energy consumption patterns of many appliances. Load curves of the appliances are measured when different events and applications are triggered and utilized. A revised longest-common-subsequence (LCS) similarity measurement algorithm is proposed to calculate energy dataset similarities. Thus, the data quality prior information becomes available before training machine learning models. In addition, a subsample convolutional neural network (SCNN) is put forward. It serves as a non-intrusive optical character recognition (OCR) approach to obtain energy data directly from monitors of power meters. The link for the EAD dataset is: https://drive.google.com/drive/folders/1zn0V6Q8eXXSKxKgcs8ZRValL5VEn3anD
|
['Chen Li']
|
2022-08-29
| null | null | null | null |
['miscellaneous']
|
['miscellaneous']
|
[ 1.09392237e-02 -6.64439857e-01 -3.05392832e-01 -6.35994434e-01
-3.81589085e-01 -6.06341958e-01 3.77358735e-01 3.43805730e-01
-2.32367873e-01 5.20322263e-01 1.38565898e-01 -2.03071460e-01
-1.22109540e-01 -1.08338964e+00 -4.17448819e-01 -8.06169033e-01
2.96766132e-01 1.34179994e-01 -3.45233560e-01 2.91951030e-01
1.18126251e-01 6.57974124e-01 -1.74409711e+00 2.15526104e-01
7.11414218e-01 1.45510578e+00 3.99488837e-01 5.96053660e-01
-2.74839878e-01 7.23886490e-01 -6.59872293e-01 -2.06886724e-01
3.76831412e-01 -1.98391140e-01 -7.49784529e-01 -1.73904374e-01
1.87966935e-02 -6.20014608e-01 -2.67358512e-01 1.13330555e+00
6.42375767e-01 -1.85009420e-01 5.58500588e-01 -1.69649053e+00
-7.85585999e-01 7.59617805e-01 -4.43386585e-01 3.67745638e-01
6.01564288e-01 1.96303234e-01 7.03476131e-01 -2.19798148e-01
-1.18815890e-02 5.01465321e-01 3.60883623e-01 1.54039443e-01
-7.23033786e-01 -9.02156353e-01 -4.27854121e-01 8.23151290e-01
-1.36820626e+00 -2.70926803e-01 1.04498053e+00 -2.62657672e-01
1.35257924e+00 5.60644388e-01 7.35086322e-01 1.33916664e+00
-1.17089441e-02 8.19021463e-01 9.37481701e-01 -2.77867794e-01
6.56043410e-01 1.60267279e-01 2.58726656e-01 3.50276828e-01
3.69545966e-01 -1.58722058e-01 -2.60591179e-01 -1.80601761e-01
5.27485311e-02 6.98187411e-01 -2.07433254e-01 1.37527108e-01
-7.48317540e-01 4.34850216e-01 1.64201900e-01 4.81705517e-01
-4.45826441e-01 -4.23383303e-02 5.63482940e-01 2.36195996e-01
2.08015203e-01 4.17910777e-02 -6.47940457e-01 -5.82185268e-01
-8.28078687e-01 -2.47666150e-01 7.70542026e-01 1.19475257e+00
9.62303996e-01 -3.60572860e-02 1.01973154e-01 6.48161888e-01
1.12960689e-01 6.09528363e-01 7.83544660e-01 -6.45915568e-01
4.55641627e-01 1.17606556e+00 6.01539463e-02 -7.33346164e-01
-7.14414418e-01 2.24128231e-01 -8.61193955e-01 2.71755662e-02
8.66074413e-02 6.48124423e-03 -2.77026176e-01 1.14902079e+00
3.40299249e-01 2.12945029e-01 -2.43113056e-01 4.33196336e-01
7.92250395e-01 7.84825861e-01 2.00412050e-01 -3.86619568e-01
1.42668319e+00 -4.62345511e-01 -8.87635350e-01 3.67906183e-01
5.55090547e-01 -6.11015260e-01 1.16425753e+00 4.47705597e-01
-7.32455194e-01 -4.73396897e-01 -1.14227617e+00 -7.80952200e-02
-9.31764603e-01 1.89066127e-01 4.09751177e-01 7.18064427e-01
-6.21029019e-01 7.27923214e-01 -9.41648543e-01 -6.44561231e-01
5.81291914e-01 2.25014448e-01 -1.84659716e-02 2.31514141e-01
-6.98874354e-01 6.63435280e-01 6.39808536e-01 -6.52924702e-02
-5.72621644e-01 -7.55638361e-01 -5.46031237e-01 -2.74466258e-02
4.88473549e-02 -3.45065556e-02 1.17964768e+00 -6.34288967e-01
-1.75294864e+00 6.73371792e-01 2.04192489e-01 -3.14306766e-01
4.91028756e-01 -1.75158858e-01 -1.20935905e+00 -1.31678671e-01
-1.43890381e-01 -2.00596899e-02 6.40752435e-01 -6.35527730e-01
-8.20089519e-01 -5.08529186e-01 -3.60073388e-01 -3.46213043e-01
-6.75382435e-01 1.25897005e-01 -1.00024000e-01 -4.45806265e-01
-4.62062597e-01 -7.08671629e-01 5.87210000e-01 -2.37802193e-01
-7.08230257e-01 -6.53010666e-01 9.73128617e-01 -6.79831386e-01
1.54840755e+00 -2.17262101e+00 -5.40134192e-01 1.89089239e-01
-8.66712853e-02 3.01792979e-01 1.72483310e-01 5.89807272e-01
-4.22764093e-01 -4.70070317e-02 -8.68406594e-02 2.16661599e-02
4.96740580e-01 -6.34929612e-02 -6.34304248e-03 5.19340575e-01
-5.37852049e-02 1.15526772e+00 -7.68940270e-01 -3.27369750e-01
7.69459188e-01 3.75737578e-01 9.84119773e-02 2.87067503e-01
-1.73252255e-01 1.05150387e-01 -4.33512002e-01 8.20335209e-01
8.57512355e-01 -4.38263863e-01 1.00007266e-01 -7.69222379e-01
-3.28663707e-01 5.18335938e-01 -1.24065852e+00 1.79136026e+00
-4.95807976e-01 4.97625977e-01 -4.00840193e-01 -8.19151521e-01
8.18860114e-01 2.72414356e-01 9.95501459e-01 -1.06917942e+00
5.15988827e-01 3.33081000e-02 -6.14155173e-01 -8.21588814e-01
4.22751397e-01 5.48922420e-01 -5.65678366e-02 7.21468925e-01
-2.68141657e-01 2.26211414e-01 3.29866350e-01 -2.76253283e-01
1.29216170e+00 2.58513968e-02 5.62830389e-01 -1.82677582e-01
4.14130390e-01 -3.05993706e-01 4.87137228e-01 9.81314257e-02
-3.71849865e-01 8.05556551e-02 -1.01953559e-01 -5.38071990e-01
-1.30567539e+00 -9.22358751e-01 -3.54131848e-01 8.59105647e-01
-1.39465630e-01 -4.96716201e-01 -7.87029505e-01 -3.95907670e-01
3.16837057e-02 1.15625036e+00 -2.92399853e-01 -1.41543314e-01
-3.49535346e-01 -6.90721035e-01 4.51836556e-01 7.10483432e-01
9.24085021e-01 -1.12554395e+00 -1.09330893e+00 3.22306693e-01
-3.62794325e-02 -1.01555407e+00 -3.99734020e-01 4.69748527e-01
-5.82013071e-01 -1.27512801e+00 -5.97619601e-02 -4.91371363e-01
3.36327016e-01 6.16043657e-02 1.15578234e+00 -3.34247291e-01
-6.63239837e-01 4.13134366e-01 -5.57623208e-01 -5.06619930e-01
-3.15671712e-01 1.13139145e-01 3.21465999e-01 -2.09144186e-02
1.46276009e+00 -9.39166307e-01 -7.96873689e-01 3.28846931e-01
-8.72887194e-01 -2.53608584e-01 3.07711571e-01 -2.26518176e-02
4.62608576e-01 3.86539668e-01 2.75669962e-01 -2.46258780e-01
5.88345349e-01 -7.44974673e-01 -9.12688971e-01 2.56214023e-01
-9.80724156e-01 -1.10021971e-01 9.65938687e-01 -4.01541591e-01
-6.54774964e-01 1.06299043e-01 -8.06113183e-02 -3.62566143e-01
-7.94655025e-01 -2.38303944e-01 -5.63179731e-01 5.99172652e-01
2.87741780e-01 3.88062775e-01 -4.73733127e-01 -7.93136001e-01
2.31986880e-01 1.20012558e+00 5.91033399e-01 -2.95700461e-01
7.49658763e-01 3.99734318e-01 -5.39269388e-01 -8.56215298e-01
-5.93725666e-02 -3.79722416e-01 -5.40691078e-01 4.17853482e-02
9.90875900e-01 -7.40464270e-01 -1.30772388e+00 8.01272571e-01
-8.78794789e-01 -2.16118038e-01 -5.47175825e-01 4.31281388e-01
-3.01891357e-01 2.88079768e-01 -5.08979678e-01 -7.82083452e-01
-6.93962216e-01 -7.85979271e-01 8.16814661e-01 5.40309191e-01
-3.93714488e-01 -5.88858306e-01 -8.21609497e-02 1.94604859e-01
3.67922962e-01 3.58864754e-01 9.12863433e-01 -9.37461913e-01
-3.96391153e-01 -2.62565047e-01 -7.22775906e-02 4.65041518e-01
8.23208869e-01 3.24157208e-01 -9.33334768e-01 -3.05772811e-01
1.50635898e-01 -7.09068477e-02 2.10232094e-01 1.68490171e-01
1.55659926e+00 -4.66915160e-01 -1.86045602e-01 7.32858419e-01
1.60221469e+00 8.56351793e-01 6.89029455e-01 4.89031911e-01
5.15442371e-01 -1.18757099e-01 -1.26627326e-01 9.20759022e-01
5.75625062e-01 5.52141428e-01 4.16472673e-01 9.97523367e-02
1.87103137e-01 -1.10260829e-01 5.03569901e-01 7.80473888e-01
2.77968973e-01 -5.17868042e-01 -5.64397275e-01 3.86890113e-01
-1.33847308e+00 -1.00962257e+00 -5.65671511e-02 1.91970706e+00
6.53228164e-01 -3.69230323e-02 3.41417938e-01 5.29284656e-01
6.83431208e-01 2.23670423e-01 -1.06254506e+00 -3.93712461e-01
-5.42912446e-02 1.62718832e-01 5.51645577e-01 -1.00637600e-01
-7.93389559e-01 6.59483075e-02 5.06570864e+00 6.91319764e-01
-1.00284100e+00 9.44572315e-02 2.68366992e-01 -5.34773231e-01
-6.91388594e-03 -7.31298983e-01 -7.39743590e-01 1.38692868e+00
1.24800122e+00 -5.74121773e-02 8.53370249e-01 8.99741054e-01
4.61237907e-01 -2.67785966e-01 -1.39553618e+00 1.64991641e+00
-1.89082369e-01 -1.15425837e+00 -3.24797601e-01 1.17591701e-01
4.08843458e-01 3.56230974e-01 -4.91025597e-02 -2.27698117e-01
5.27746856e-01 -5.41359544e-01 5.38165271e-01 6.54567063e-01
8.99983108e-01 -7.82545686e-01 5.11551142e-01 1.45916715e-01
-1.57329082e+00 -5.49007714e-01 -2.86350042e-01 2.83672869e-01
4.93351556e-02 8.52410972e-01 -7.98061550e-01 6.78065419e-01
1.37532401e+00 6.84880555e-01 -5.31042457e-01 8.25038493e-01
-1.33084701e-02 5.71678340e-01 -6.12582386e-01 -2.61099517e-01
-3.81843179e-01 -5.50865114e-01 6.00813478e-02 1.06319737e+00
5.37776828e-01 1.45579576e-01 -2.30314359e-01 1.15674639e+00
-3.53935242e-01 2.25266770e-01 -3.76070231e-01 -3.16635340e-01
6.56377017e-01 1.49298251e+00 -6.02456689e-01 -3.31768215e-01
-6.35139406e-01 1.28747606e+00 -4.08528931e-02 7.07290471e-02
-8.42429101e-01 -5.88602722e-01 1.16254640e+00 -6.16444238e-02
4.73630339e-01 -1.86913565e-01 8.82054269e-02 -1.01552153e+00
1.60411805e-01 -5.14228642e-01 4.34307784e-01 -1.00207281e+00
-1.62485540e+00 2.70830572e-01 -1.35419875e-01 -1.31396687e+00
-1.54700607e-01 -5.81812382e-01 -8.05201352e-01 5.35751104e-01
-1.39020228e+00 -7.72667944e-01 -7.70186305e-01 1.06462562e+00
5.71886957e-01 -2.58232653e-01 8.81732702e-01 7.31819868e-01
-9.54563200e-01 5.93619049e-01 6.63967073e-01 4.96454448e-01
2.37688318e-01 -1.10523462e+00 4.49226379e-01 5.45970857e-01
1.01051247e-02 1.58870205e-01 4.48310286e-01 -7.26818323e-01
-1.69845319e+00 -1.20195079e+00 5.21789730e-01 -4.74316001e-01
7.29116023e-01 -3.64981532e-01 -7.73249626e-01 5.77813447e-01
3.75956655e-01 -1.70230135e-01 9.99985278e-01 -5.06982803e-01
-3.29452783e-01 -5.80543995e-01 -1.44602120e+00 5.31067476e-02
9.70835149e-01 -8.52937222e-01 -4.66482401e-01 2.32533783e-01
3.99216801e-01 1.09479420e-01 -1.12399554e+00 -5.22252619e-02
3.78486723e-01 -8.29543114e-01 5.82357764e-01 -4.84557420e-01
1.08485878e-01 -6.30464613e-01 -5.01668632e-01 -9.22978222e-01
-3.76950949e-01 -5.87116003e-01 -6.30963922e-01 1.71292043e+00
-7.51259997e-02 -5.50077021e-01 6.34521008e-01 7.01791286e-01
7.05134794e-02 -3.80826920e-01 -9.10327554e-01 -8.57159078e-01
-2.99423426e-01 -7.97217011e-01 1.56820917e+00 8.84616375e-01
1.75325975e-01 -4.84027602e-02 -1.60981014e-01 4.25251573e-01
4.32684481e-01 2.05739513e-01 3.45231086e-01 -1.12329662e+00
2.00156599e-01 -3.99516404e-01 -5.27335644e-01 -6.23663187e-01
-5.60219027e-02 -8.51777256e-01 -6.09008253e-01 -1.33838236e+00
5.49059585e-02 -1.15353301e-01 -6.62431240e-01 5.95159531e-01
4.77819830e-01 -2.09363252e-02 3.25468406e-02 7.52785876e-02
-3.98317426e-01 3.90273273e-01 1.90685019e-01 -4.63494807e-01
-1.96124434e-01 6.96050525e-02 -3.45187366e-01 6.16753757e-01
1.21663129e+00 -3.44172746e-01 -4.77876544e-01 -4.52465057e-01
3.13208669e-01 -6.39962792e-01 3.75976294e-01 -1.18229961e+00
3.72383237e-01 -1.24985531e-01 6.21645868e-01 -1.20499992e+00
-3.55220437e-02 -1.49788928e+00 8.30389142e-01 4.59878355e-01
-3.61125264e-03 4.49854702e-01 -2.89393123e-02 3.60353827e-01
5.21344662e-01 -1.63070001e-02 5.33387542e-01 -6.04499988e-02
-7.99008071e-01 4.58011061e-01 -3.27581644e-01 -3.77514064e-01
1.14367890e+00 -5.65290809e-01 -3.67132932e-01 5.58218993e-02
-2.82609582e-01 7.78022632e-02 6.00705087e-01 7.57642090e-01
1.35283843e-01 -1.67762756e+00 -3.40625346e-01 3.19729686e-01
4.27550495e-01 -2.34633625e-01 2.89313346e-01 3.09019536e-01
-2.70639569e-01 2.85068005e-01 -4.30060357e-01 -5.36456645e-01
-9.92501974e-01 9.27013934e-01 3.51192176e-01 2.76958495e-01
-7.64367819e-01 2.81181872e-01 -6.54759526e-01 -2.85558365e-02
2.05727249e-01 -8.69927704e-01 -2.68611372e-01 4.76056963e-01
8.56395304e-01 1.04086578e+00 4.64713067e-01 -4.31464940e-01
-5.15675962e-01 4.74809855e-01 1.58029407e-01 7.72186220e-01
1.72813463e+00 -4.28736776e-01 -1.12315468e-01 5.09039044e-01
1.66170692e+00 -3.74933302e-01 -9.98891354e-01 -1.81158736e-01
2.16719732e-01 -2.80414790e-01 -1.96779042e-01 -8.29859912e-01
-1.18482661e+00 5.08274138e-01 1.51442909e+00 6.10714138e-01
1.55253482e+00 1.01954147e-01 1.10667729e+00 3.74714196e-01
4.87100363e-01 -1.64676487e+00 -2.79648244e-01 -5.02338111e-02
3.59274417e-01 -1.11009264e+00 -6.82732984e-02 2.17999369e-01
-1.85608387e-01 1.16841698e+00 2.02273294e-01 3.37759167e-01
7.43985653e-01 5.71596622e-01 -2.45450675e-01 -2.42236361e-01
-2.75621295e-01 -3.86288203e-02 -1.56434074e-01 6.11383080e-01
4.96134795e-02 3.15490335e-01 -4.54856269e-02 5.91551363e-01
-5.11000335e-01 3.26298326e-01 2.79791951e-01 9.59121704e-01
-1.80740327e-01 -8.24148893e-01 -1.49200335e-01 5.41419566e-01
-2.68018156e-01 6.25044629e-02 -5.15616797e-02 2.21328124e-01
4.62769985e-01 1.14458215e+00 3.81966233e-01 -5.46822667e-01
4.76322412e-01 4.44108576e-01 6.69341013e-02 3.04687954e-02
-4.14809853e-01 -4.09214377e-01 -3.36086839e-01 -8.15734386e-01
-3.63943040e-01 -6.93168700e-01 -1.26649284e+00 -7.96991646e-01
-9.74535421e-02 5.65064885e-03 9.53141868e-01 7.95650899e-01
5.81195772e-01 5.77853918e-01 1.08757699e+00 -5.59839427e-01
-2.43996546e-01 -7.65736699e-01 -8.05528402e-01 5.96263051e-01
3.63514632e-01 -3.59950662e-01 -3.26381117e-01 1.81477532e-01]
|
[6.001276016235352, 2.581291913986206]
|
e5382308-b3f4-4f4a-a762-5ab58d751a96
|
high-fidelity-speech-regeneration-with
|
2102.00429
| null |
https://arxiv.org/abs/2102.00429v1
|
https://arxiv.org/pdf/2102.00429v1.pdf
|
High Fidelity Speech Regeneration with Application to Speech Enhancement
|
Speech enhancement has seen great improvement in recent years mainly through contributions in denoising, speaker separation, and dereverberation methods that mostly deal with environmental effects on vocal audio. To enhance speech beyond the limitations of the original signal, we take a regeneration approach, in which we recreate the speech from its essence, including the semi-recognized speech, prosody features, and identity. We propose a wav-to-wav generative model for speech that can generate 24khz speech in a real-time manner and which utilizes a compact speech representation, composed of ASR and identity features, to achieve a higher level of intelligibility. Inspired by voice conversion methods, we train to augment the speech characteristics while preserving the identity of the source using an auxiliary identity network. Perceptual acoustic metrics and subjective tests show that the method obtains valuable improvements over recent baselines.
|
['Yaniv Taigman', 'Ori Kabeli', 'Yossi Adi', 'Lior Wolf', 'Adam Polyak']
|
2021-01-31
| null | null | null | null |
['speaker-separation']
|
['speech']
|
[ 2.60003597e-01 1.43607646e-01 2.62816638e-01 -2.71437645e-01
-1.07234836e+00 -5.90374708e-01 3.02635401e-01 -4.12881583e-01
-5.09949513e-02 4.33410436e-01 9.44166780e-01 -1.63216516e-01
3.11069548e-01 -3.70041341e-01 -4.25102711e-01 -7.36347795e-01
2.97087550e-01 -4.74339336e-01 -2.30595842e-01 -5.71058571e-01
-3.35378647e-01 2.47931778e-01 -1.62504971e+00 2.63491333e-01
9.67155337e-01 8.89159441e-01 3.35641146e-01 8.69791031e-01
1.71672806e-01 5.51402092e-01 -9.39361751e-01 -3.67915988e-01
2.61702746e-01 -6.98401392e-01 -3.09061885e-01 1.66490301e-01
2.74495304e-01 -4.31165844e-01 -7.34434366e-01 1.18547225e+00
9.94487166e-01 4.43549663e-01 4.56731230e-01 -7.40506649e-01
-8.59240234e-01 8.50666285e-01 -2.40293846e-01 -3.41879800e-02
3.32259595e-01 -2.60686595e-02 9.95709658e-01 -1.18823230e+00
1.50421903e-01 1.38025975e+00 7.89329231e-01 7.68691123e-01
-1.30345833e+00 -8.53724301e-01 -1.64431900e-01 1.43958807e-01
-1.19089496e+00 -1.14181149e+00 1.03207648e+00 6.12276420e-03
7.46606171e-01 4.93971020e-01 3.29528242e-01 1.31841993e+00
-3.79911780e-01 7.43040323e-01 8.18634212e-01 -5.33812106e-01
7.69688189e-02 -2.02756934e-02 -1.42239571e-01 1.08342141e-01
-4.86982852e-01 5.72881997e-01 -6.73421800e-01 6.13429025e-03
4.59216684e-01 -5.74060321e-01 -6.34350300e-01 2.68551201e-01
-7.78900146e-01 5.62132299e-01 1.14217438e-01 2.54864901e-01
-5.43707192e-01 8.25771713e-04 2.50196993e-01 3.74144703e-01
6.04741454e-01 4.51449096e-01 -1.25756815e-01 -1.71251833e-01
-1.15875304e+00 3.54309119e-02 4.99122709e-01 8.04381013e-01
3.90590966e-01 9.54537988e-01 -4.62920964e-01 1.42593467e+00
1.16959438e-01 7.51018345e-01 8.17085922e-01 -1.02988541e+00
1.90874755e-01 -2.92802125e-01 -1.77740920e-02 -4.56808269e-01
8.74615014e-02 -8.03954720e-01 -9.78406250e-01 2.74827719e-01
-2.38279045e-01 -3.89712662e-01 -1.11296833e+00 1.86509657e+00
2.29027897e-01 4.83483285e-01 3.48787725e-01 7.61210978e-01
8.19329560e-01 1.03291440e+00 -2.51523077e-01 -2.01340735e-01
1.07361972e+00 -1.10046911e+00 -1.27536607e+00 -9.75922570e-02
-1.92896128e-01 -1.04113913e+00 1.08949268e+00 4.85144466e-01
-1.37992775e+00 -1.13681996e+00 -1.21721196e+00 -1.19721003e-01
-6.86244443e-02 3.07400048e-01 3.72899068e-03 1.12406015e+00
-1.40952492e+00 6.13466680e-01 -4.61069703e-01 2.64009207e-01
-5.76240681e-02 1.16834804e-01 -2.42501497e-01 1.79020196e-01
-1.41233528e+00 5.37225544e-01 1.94111958e-01 1.13120966e-01
-1.08420730e+00 -7.44230032e-01 -1.14443970e+00 2.15864152e-01
6.50244281e-02 -4.53316718e-01 1.39913630e+00 -7.69984365e-01
-2.10291910e+00 2.19064415e-01 -4.60537165e-01 -5.27788877e-01
1.69907346e-01 -3.04775834e-01 -9.54812169e-01 1.13655388e-01
-2.93115526e-01 4.00765568e-01 1.45226622e+00 -1.30753446e+00
-5.59221625e-01 -2.64280960e-02 -4.54432100e-01 3.17924470e-01
-6.45521462e-01 1.99605286e-01 -3.64458382e-01 -1.30174148e+00
1.25334725e-01 -7.54601955e-01 -5.01297135e-03 -4.71741170e-01
-4.65509236e-01 2.12978423e-01 1.03582895e+00 -1.44947505e+00
1.12443864e+00 -2.75259805e+00 6.42973483e-02 3.40784229e-02
-4.51854896e-03 7.12531388e-01 -5.26033759e-01 3.11074674e-01
-2.53688276e-01 -1.57927144e-02 -1.93045467e-01 -9.19140041e-01
1.49357855e-01 -1.47332801e-02 -7.81157374e-01 7.72293434e-02
3.69347900e-01 6.00185573e-01 -7.17116773e-01 -1.01184346e-01
2.55215287e-01 1.00715041e+00 -7.17572927e-01 4.66698885e-01
3.05471480e-01 3.54165018e-01 2.70829976e-01 3.65292042e-01
7.77300537e-01 6.99100494e-01 -7.90055692e-02 -3.15343559e-01
-6.58123344e-02 5.78450084e-01 -1.16983962e+00 1.43251467e+00
-6.77564979e-01 5.66694021e-01 9.44024384e-01 -5.42775333e-01
1.13179171e+00 8.94421637e-01 2.19112545e-01 -4.06568319e-01
1.16229048e-02 3.72941166e-01 1.73497126e-01 -2.11678207e-01
6.42495155e-01 -2.49658704e-01 4.16579217e-01 8.20825472e-02
3.23509693e-01 -5.89781761e-01 -1.83111593e-01 -8.07135087e-03
9.47336912e-01 -1.74044549e-01 3.42216007e-02 1.19058535e-01
6.35432601e-01 -7.17502236e-01 4.81084049e-01 5.21760583e-01
-2.66731054e-01 9.82225299e-01 -2.53385872e-01 5.98478436e-01
-1.05231988e+00 -1.47573340e+00 5.46240993e-02 1.29307687e+00
-2.49703646e-01 -2.01112002e-01 -1.06574416e+00 -8.20909366e-02
-1.80122316e-01 9.89389956e-01 -2.86728293e-01 -5.50167859e-01
-6.36226118e-01 -2.87284464e-01 9.61777687e-01 5.27134478e-01
4.10364747e-01 -9.64859068e-01 4.05548364e-01 4.59086776e-01
-4.71222460e-01 -1.08313036e+00 -1.10535300e+00 2.05276117e-01
-4.78272110e-01 -2.13425562e-01 -1.03891683e+00 -9.95030403e-01
2.68740892e-01 4.14642245e-01 5.03705919e-01 -3.29567939e-01
1.08917587e-01 1.90609857e-01 -4.81197745e-01 -4.77175504e-01
-1.05957520e+00 -1.75945178e-01 4.54863876e-01 3.55565637e-01
-2.18624055e-01 -9.86201286e-01 -2.81729221e-01 1.93738580e-01
-8.84789884e-01 -3.36599678e-01 6.14558339e-01 1.05723763e+00
3.77177596e-01 3.84692758e-01 1.22161841e+00 -1.63138479e-01
9.95215297e-01 -1.41280383e-01 -8.58178362e-02 -2.18762085e-01
-3.68788034e-01 -1.94203302e-01 7.52482772e-01 -7.76370823e-01
-1.62143564e+00 -1.31653070e-01 -8.28821421e-01 -6.48020029e-01
-1.34222835e-01 2.76477635e-01 -6.63873553e-01 9.41406190e-02
6.50941610e-01 4.96497393e-01 2.25157648e-01 -7.65155673e-01
6.66737497e-01 1.37283444e+00 1.18627274e+00 -3.05672228e-01
1.09479034e+00 2.24733666e-01 -5.79653919e-01 -1.23519123e+00
-5.35952985e-01 -6.35783315e-01 -2.94423491e-01 9.22932625e-02
3.72291446e-01 -1.07088315e+00 -3.33842188e-01 6.52933836e-01
-1.26413631e+00 -2.10734457e-01 -6.93946421e-01 7.42833376e-01
-4.05886918e-01 5.14472723e-01 -9.71218109e-01 -1.14438486e+00
-5.53719163e-01 -1.04382157e+00 9.11677003e-01 6.04724437e-02
-1.12169512e-01 -4.61344361e-01 -4.98157218e-02 2.94424653e-01
9.21932697e-01 -4.10367399e-01 7.61259735e-01 -5.09818554e-01
3.22065316e-02 -1.04003720e-01 2.46747538e-01 1.25980711e+00
6.49112523e-01 -3.03430885e-01 -1.54472959e+00 -3.71644557e-01
4.41968143e-01 1.11372136e-01 8.06227028e-01 4.82513338e-01
7.89850831e-01 -4.43519622e-01 7.54824504e-02 8.24973583e-01
5.85992992e-01 5.91528594e-01 8.62337172e-01 -3.56478572e-01
4.28510547e-01 6.73507273e-01 1.93093628e-01 2.71282226e-01
-1.48002848e-01 5.36737680e-01 2.37278715e-01 -4.79723275e-01
-1.01688147e+00 -4.48798239e-01 7.38867700e-01 1.37237799e+00
7.23435581e-02 -2.75112838e-01 -2.15628460e-01 7.07401097e-01
-1.11402428e+00 -1.05985820e+00 2.85651684e-01 2.22662377e+00
1.21285534e+00 -1.30427435e-01 7.94741437e-02 5.67260981e-01
1.03903413e+00 3.35858911e-01 -5.48137546e-01 -3.64823967e-01
-4.90557194e-01 5.21959305e-01 4.24582278e-03 7.11437404e-01
-8.53662670e-01 8.48801911e-01 6.90317345e+00 1.09548903e+00
-1.27686787e+00 1.32446423e-01 4.12021518e-01 -8.24050158e-02
-2.83025414e-01 -4.98919457e-01 -4.68258530e-01 2.19242200e-01
9.64944661e-01 -3.09464723e-01 8.82852077e-01 6.53115153e-01
5.43598175e-01 7.26910174e-01 -8.18760514e-01 8.95332575e-01
2.51883656e-01 -8.49188745e-01 -1.53487280e-01 -1.36135118e-02
6.10669732e-01 -1.42245546e-01 5.12953043e-01 5.44161916e-01
9.79962423e-02 -9.76349592e-01 1.03307486e+00 1.69417173e-01
9.45734978e-01 -7.34943151e-01 4.45756406e-01 2.30488420e-01
-1.26998854e+00 -1.69772610e-01 -1.17783263e-01 1.76802427e-01
3.30907136e-01 4.46919709e-01 -1.04509914e+00 5.82797289e-01
4.93894637e-01 3.19663078e-01 -6.07339777e-02 9.93772388e-01
-4.56696302e-01 1.11499870e+00 -5.39460815e-02 4.22794223e-01
-2.28699937e-01 -1.37559786e-01 1.05805981e+00 1.29709232e+00
6.58694923e-01 6.49183465e-04 -2.34873250e-01 8.16960335e-01
-4.10429358e-01 1.47012129e-01 -3.04102719e-01 -1.57270834e-01
7.03414798e-01 1.05923796e+00 2.08664089e-01 -1.69016153e-01
-8.17480385e-02 1.20929408e+00 -3.25068116e-01 6.54420376e-01
-7.50324249e-01 -8.08107138e-01 1.02511811e+00 -4.90439981e-02
4.31935579e-01 -1.02496915e-01 1.50291935e-01 -8.19224775e-01
1.44002676e-01 -1.19856417e+00 -2.54955888e-01 -8.62270117e-01
-1.22298062e+00 9.21724975e-01 -5.27645648e-01 -1.01634300e+00
-4.17255729e-01 -3.48092794e-01 -7.05007672e-01 1.27729380e+00
-1.58514857e+00 -1.16569710e+00 1.28553897e-01 5.95086396e-01
7.49028027e-01 -1.99590340e-01 7.74115324e-01 5.77976048e-01
-3.90052587e-01 8.25988412e-01 3.29881608e-01 1.11055396e-01
8.72802734e-01 -1.16861153e+00 7.55507112e-01 1.09787941e+00
1.48331344e-01 5.56979895e-01 8.17153513e-01 -5.16049266e-01
-1.12937737e+00 -1.27463055e+00 6.80570245e-01 1.55611858e-01
5.00323892e-01 -5.22834659e-01 -1.00603962e+00 4.70387727e-01
3.76303881e-01 -2.12951884e-01 6.34173572e-01 -1.04935445e-01
-3.97986263e-01 -3.00396174e-01 -1.02078927e+00 7.27977633e-01
9.96779978e-01 -9.68868732e-01 -8.14850986e-01 -2.15857640e-01
1.31286669e+00 -3.74589771e-01 -5.17131567e-01 2.81822681e-01
3.52542579e-01 -6.82872891e-01 1.24513674e+00 -3.10035884e-01
5.32415882e-03 -3.35929751e-01 -4.07526135e-01 -1.89731979e+00
-2.73540944e-01 -1.42109489e+00 -2.43959669e-03 1.92966020e+00
4.22259659e-01 -6.61654532e-01 4.05965686e-01 4.19411063e-03
-7.28301823e-01 4.26553153e-02 -8.84479523e-01 -1.17961323e+00
5.96737722e-03 -5.35216331e-01 8.31378222e-01 7.08011508e-01
-2.66380310e-02 3.88101846e-01 -7.65458465e-01 4.36581045e-01
5.17560244e-01 -2.94089466e-01 5.33915222e-01 -8.36524725e-01
-4.75669563e-01 -3.56296688e-01 5.17181866e-02 -1.20737863e+00
2.36483991e-01 -6.72567546e-01 6.13018811e-01 -1.29193091e+00
-4.48362082e-01 1.35247126e-01 -3.89715642e-01 2.04500079e-01
-4.13575709e-01 1.68281928e-01 2.93719292e-01 -1.80505186e-01
2.30913192e-01 1.07748151e+00 1.26559770e+00 -2.38924786e-01
-4.45184171e-01 2.94332504e-01 -8.70371938e-01 6.49995506e-01
6.30611598e-01 -2.82088578e-01 -4.33747917e-01 -2.56671846e-01
-5.94939291e-01 2.72713065e-01 6.71696663e-02 -1.12670231e+00
1.99078694e-01 2.37657398e-01 -1.17841735e-02 -3.73376012e-01
1.04906356e+00 -6.46510422e-01 7.33223706e-02 1.55237928e-01
-5.49521208e-01 -4.46687967e-01 2.70114094e-01 4.82626766e-01
-7.25132048e-01 -1.64959207e-01 8.64611328e-01 3.48921716e-01
-1.11853331e-01 3.66374515e-02 -5.79794586e-01 -1.85552791e-01
3.61038864e-01 8.12681988e-02 -2.50899881e-01 -9.42423284e-01
-7.77954996e-01 -3.44479173e-01 -2.59267632e-02 3.77457887e-01
6.41455412e-01 -1.35359383e+00 -1.16715574e+00 3.60533535e-01
-3.51628602e-01 -4.79623437e-01 4.51209128e-01 4.37644482e-01
2.24044621e-01 6.73440173e-02 1.31411940e-01 -7.54508451e-02
-1.25634253e+00 6.04337752e-01 4.98811454e-01 2.70667493e-01
-5.00268400e-01 8.66677523e-01 4.19068128e-01 -3.42157841e-01
3.85509074e-01 -2.59324312e-01 -2.31662691e-01 -7.29078054e-02
7.48248160e-01 5.53823769e-01 2.82305270e-01 -8.86510849e-01
3.12995315e-02 1.90451562e-01 2.46145412e-01 -6.50440335e-01
1.22549987e+00 -5.43141007e-01 1.78678989e-01 9.08731669e-02
1.14446032e+00 9.90169644e-01 -1.24142265e+00 -3.05315346e-01
-4.34939265e-01 -3.39011520e-01 4.52889413e-01 -9.20396566e-01
-9.54441488e-01 1.01746786e+00 5.50117195e-01 4.43615556e-01
1.44212770e+00 -2.65321016e-01 1.17918563e+00 -2.11724993e-02
-1.48617461e-01 -1.11536121e+00 2.89252520e-01 5.77860117e-01
1.27879286e+00 -8.43663752e-01 -6.28564715e-01 -6.04434848e-01
-5.78602612e-01 8.81978750e-01 1.65111706e-01 1.48898080e-01
5.64253032e-01 5.44771075e-01 4.58923370e-01 5.20773888e-01
-3.08992416e-01 -4.30223793e-01 3.29334021e-01 1.05669725e+00
2.81864733e-01 -2.15055625e-04 2.72965342e-01 9.59957659e-01
-7.62169659e-01 -5.26996791e-01 3.97833943e-01 2.60688484e-01
-4.74688441e-01 -1.17708755e+00 -7.79312551e-01 -8.11118931e-02
-5.35665989e-01 -6.38053298e-01 -3.22013676e-01 2.20672116e-01
-1.00034170e-01 1.70306683e+00 -1.61475286e-01 -6.12218499e-01
6.65159345e-01 2.30725676e-01 1.18956529e-02 -5.36575556e-01
-6.26373053e-01 7.64617503e-01 2.51642257e-01 -1.22992232e-01
-1.94116328e-02 -4.44951326e-01 -1.12907064e+00 4.69938666e-02
-6.13390088e-01 2.64089406e-01 8.75008464e-01 5.92842698e-01
3.81703228e-01 9.90210533e-01 1.12064886e+00 -1.07384765e+00
-1.02097344e+00 -1.31828582e+00 -8.58489633e-01 3.90792817e-01
8.47665787e-01 -1.19186029e-01 -7.18827069e-01 2.36203954e-01]
|
[15.041340827941895, 6.099411487579346]
|
20991ea1-a9fe-4d65-8e7e-b140261610fd
|
from-key-points-to-key-point-hierarchy
|
2306.03853
| null |
https://arxiv.org/abs/2306.03853v1
|
https://arxiv.org/pdf/2306.03853v1.pdf
|
From Key Points to Key Point Hierarchy: Structured and Expressive Opinion Summarization
|
Key Point Analysis (KPA) has been recently proposed for deriving fine-grained insights from collections of textual comments. KPA extracts the main points in the data as a list of concise sentences or phrases, termed key points, and quantifies their prevalence. While key points are more expressive than word clouds and key phrases, making sense of a long, flat list of key points, which often express related ideas in varying levels of granularity, may still be challenging. To address this limitation of KPA, we introduce the task of organizing a given set of key points into a hierarchy, according to their specificity. Such hierarchies may be viewed as a novel type of Textual Entailment Graph. We develop ThinkP, a high quality benchmark dataset of key point hierarchies for business and product reviews, obtained by consolidating multiple annotations. We compare different methods for predicting pairwise relations between key points, and for inferring a hierarchy from these pairwise predictions. In particular, for the task of computing pairwise key point relations, we achieve significant gains over existing strong baselines by applying directional distributional similarity methods to a novel distributional representation of key points, and further boost performance via weak supervision.
|
['Roy Bar-Haim', 'Yoav Kantor', 'Lilach Eden', 'Arie Cattan']
|
2023-06-06
| null | null | null | null |
['natural-language-inference', 'specificity']
|
['natural-language-processing', 'natural-language-processing']
|
[ 3.92558537e-02 1.72911718e-01 -6.57897115e-01 -4.85508054e-01
-1.04368126e+00 -1.00361276e+00 7.69082129e-01 1.15370870e+00
-4.76243645e-02 3.29853743e-01 9.72351015e-01 -4.03516859e-01
-2.23712176e-01 -6.99486315e-01 -7.93515801e-01 -3.90349299e-01
1.23386413e-01 4.89873379e-01 1.45019591e-01 -3.29255193e-01
7.44893909e-01 1.96805939e-01 -1.33160424e+00 8.76222789e-01
6.76418841e-01 1.13126242e+00 -8.84258747e-02 2.97333509e-01
-3.01541150e-01 6.23992264e-01 -4.33295697e-01 -8.80160272e-01
4.53723036e-02 -6.96703345e-02 -1.10875678e+00 -1.64416417e-01
6.97389960e-01 -6.07454963e-02 3.25669318e-01 9.84596550e-01
-3.11324082e-04 -1.80342704e-01 8.15669358e-01 -1.37855256e+00
-7.21108377e-01 1.03831041e+00 -6.70718372e-01 1.57091588e-01
6.66138053e-01 -4.80543077e-01 2.44658422e+00 -1.23159492e+00
6.63080633e-01 1.13305342e+00 9.18977797e-01 -3.96535322e-02
-1.45716214e+00 -4.85561550e-01 3.05445820e-01 1.55738201e-02
-1.32973564e+00 -7.95109496e-02 6.97903991e-01 -5.92432737e-01
1.12948859e+00 4.05546635e-01 4.95014429e-01 9.33267355e-01
1.57573655e-01 8.55440795e-01 9.04659152e-01 -2.43066072e-01
5.98963648e-02 1.10596694e-01 7.59992599e-01 4.18935865e-01
2.41269618e-01 -6.62986517e-01 -7.10082471e-01 -8.02353621e-01
-3.48693109e-04 2.98963815e-01 -2.21325368e-01 -2.64220387e-01
-1.29830313e+00 9.93339598e-01 4.96600211e-01 1.66046202e-01
-2.45307192e-01 -3.05646956e-01 6.67354703e-01 1.88579872e-01
8.12144995e-01 8.37425411e-01 -7.96811700e-01 3.94940376e-04
-7.99187481e-01 6.74879313e-01 1.08961010e+00 9.74278808e-01
8.92050683e-01 -1.11947918e+00 -2.63745487e-01 7.87447691e-01
2.85861760e-01 7.01484606e-02 4.58147049e-01 -7.36221969e-01
8.06449413e-01 1.16451514e+00 1.25742033e-01 -1.59480679e+00
-3.57692242e-01 -1.71620846e-01 -6.68369949e-01 -5.76647222e-01
-1.32000670e-01 2.62345970e-01 -3.01457644e-01 1.34131837e+00
2.70697951e-01 2.16083154e-02 -1.57884240e-01 4.55723286e-01
9.14210021e-01 7.45300770e-01 -4.18170482e-01 -3.51351708e-01
1.38908350e+00 -8.26322794e-01 -4.72707808e-01 6.88812882e-02
8.76989841e-01 -7.31800795e-01 1.32423091e+00 4.68684942e-01
-7.26485610e-01 -2.45721564e-01 -8.43488157e-01 -3.86662155e-01
-4.88669604e-01 -2.93277949e-01 3.92037988e-01 -6.49948716e-02
-8.67457092e-01 6.62139058e-01 -3.40739518e-01 -9.02914554e-02
2.81329930e-01 -4.13537100e-02 -2.86143571e-01 1.39246926e-01
-1.29853642e+00 6.13327384e-01 6.00551963e-02 -2.14788690e-01
-2.80482501e-01 -1.40762579e+00 -9.66373920e-01 1.65644959e-01
4.88809943e-01 -6.22394443e-01 1.19476593e+00 -2.82098472e-01
-7.94518411e-01 1.01441002e+00 -3.65805835e-01 -5.91665149e-01
-3.89647782e-02 -6.18193507e-01 -3.69897425e-01 -2.91666444e-02
6.66071415e-01 5.16788244e-01 6.58250988e-01 -1.28778636e+00
-8.69919837e-01 -2.71615565e-01 3.15296888e-01 7.54192024e-02
-5.15715539e-01 2.43500948e-01 -2.15476230e-01 -6.09455585e-01
-5.08706179e-03 -8.80529165e-01 -1.51284024e-01 -4.93586540e-01
-9.09905791e-01 -9.21150744e-01 4.45435286e-01 -3.05729032e-01
1.69690418e+00 -1.98139858e+00 2.20021501e-01 1.52281210e-01
8.58743131e-01 -2.64942378e-01 1.03917770e-01 8.95582616e-01
-7.10820481e-02 5.62475324e-01 -1.50134802e-01 -2.73791671e-01
4.19412196e-01 1.72168717e-01 -9.86514926e-01 -1.92830674e-02
1.30104721e-01 1.09870589e+00 -9.84561682e-01 -4.44799632e-01
-2.54117340e-01 -1.18510075e-01 -8.25586081e-01 8.46440718e-02
-3.81924391e-01 -3.09424937e-01 -3.61585379e-01 5.80576241e-01
4.07051921e-01 -5.76627791e-01 1.90582648e-02 -8.10922921e-01
-3.56450789e-02 7.92289436e-01 -7.47033179e-01 1.33171308e+00
-3.92525882e-01 5.27659833e-01 -3.28298330e-01 -9.26384628e-01
8.88912857e-01 9.65195373e-02 6.69049442e-01 -1.99491173e-01
-1.53138325e-01 5.53532317e-02 -3.03097755e-01 -3.36710155e-01
8.72958481e-01 -1.44804522e-01 -7.06292987e-01 5.18875539e-01
-9.63366926e-02 -4.14269328e-01 2.82032818e-01 7.35917807e-01
1.19023895e+00 -4.82834399e-01 6.44256055e-01 -3.73205990e-01
1.95045099e-01 2.96980366e-02 5.38681686e-01 4.93640721e-01
1.18184738e-01 6.48246229e-01 1.10234761e+00 -4.00733292e-01
-9.60679770e-01 -9.85707819e-01 -1.36479303e-01 1.19556141e+00
1.46939278e-01 -1.53849816e+00 -1.61144286e-01 -1.02331889e+00
4.88845289e-01 6.09966993e-01 -8.42107058e-01 1.17996626e-01
-3.84842128e-01 -3.95315260e-01 4.09809530e-01 4.73090172e-01
-1.37678921e-01 -4.55383986e-01 -9.85584036e-02 -6.57677650e-02
-4.90205377e-01 -1.34256124e+00 -7.23839045e-01 1.16702877e-01
-4.86914724e-01 -1.12358141e+00 -3.64907146e-01 -5.84936619e-01
5.82190037e-01 4.85334903e-01 1.82149827e+00 -8.57589841e-02
1.26164436e-01 1.21581241e-01 -7.32975423e-01 -3.95049334e-01
-1.66417047e-01 1.20715052e-01 2.03894109e-01 -4.04047184e-02
6.66722536e-01 -5.58408260e-01 -5.05620122e-01 4.30499643e-01
-8.09617877e-01 1.40282521e-02 3.72112364e-01 8.54610562e-01
7.37359166e-01 3.20245288e-02 3.59740496e-01 -1.27476215e+00
1.07832623e+00 -6.37039125e-01 -1.61408424e-01 2.41358340e-01
-6.40195847e-01 2.07080185e-01 5.85707545e-01 4.19661216e-02
-6.44388318e-01 -5.21321177e-01 -4.89071943e-02 -7.95528889e-02
8.74036551e-02 1.13428259e+00 5.45219816e-02 4.36408103e-01
5.58660209e-01 -1.78706825e-01 -3.59353632e-01 -4.60166186e-01
8.40682149e-01 8.25702071e-01 2.26964578e-01 -8.58393550e-01
6.93697453e-01 4.10717160e-01 -1.89316288e-01 -5.80121577e-01
-1.62929964e+00 -1.18268263e+00 -8.96939516e-01 3.14455092e-01
5.13670743e-01 -8.62809598e-01 -6.51411235e-01 -2.30666846e-01
-1.23984098e+00 2.18729645e-01 -3.15203190e-01 5.99873858e-03
-2.19067723e-01 4.78421599e-01 -6.22381866e-01 -1.32497743e-01
-3.73324275e-01 -7.47439027e-01 1.52164817e+00 -3.00881386e-01
-1.03326893e+00 -1.12108815e+00 3.78384471e-01 5.40868580e-01
-1.01685330e-01 1.62878454e-01 1.33361948e+00 -1.06603694e+00
-1.07920177e-01 -3.08406949e-01 -3.86608481e-01 4.74101752e-01
3.95044565e-01 1.41216069e-01 -4.59419012e-01 -5.47909923e-02
-1.26078427e-01 -5.66064715e-01 1.07325220e+00 2.54684836e-02
1.13454735e+00 -8.13425303e-01 -6.66701794e-01 1.53543442e-01
1.10061932e+00 -5.30207694e-01 3.96161936e-02 1.70279130e-01
1.05382764e+00 8.74777615e-01 7.37186849e-01 5.82668185e-01
8.13239872e-01 6.88935518e-01 3.27686548e-01 1.76044807e-01
3.87979984e-01 -5.35212815e-01 2.92543411e-01 1.35461688e+00
2.11524606e-01 2.25026552e-02 -1.04589474e+00 6.69064462e-01
-1.97929478e+00 -9.27327394e-01 -1.92917764e-01 1.71395504e+00
1.19951379e+00 3.04794818e-01 3.17962915e-01 6.08677305e-02
5.28391600e-01 2.59018004e-01 -2.18752787e-01 -4.01702821e-01
-8.96673556e-03 -1.55705407e-01 1.86623812e-01 5.30413270e-01
-7.89945483e-01 6.08529270e-01 5.97675085e+00 8.31034899e-01
-7.67206252e-01 -1.18238464e-01 5.68488300e-01 2.52365880e-02
-9.20278430e-01 4.17102464e-02 -1.18718505e+00 4.36961621e-01
7.38515198e-01 -3.99204195e-01 -3.62192392e-01 8.49055588e-01
3.03012021e-02 2.32607484e-01 -1.61907172e+00 9.02492881e-01
3.28103244e-01 -1.68381715e+00 5.04492640e-01 -1.75905842e-02
8.84857655e-01 1.48773715e-01 1.07237369e-01 6.75350055e-02
5.38513362e-01 -9.53722656e-01 5.70947886e-01 2.41294295e-01
3.99439216e-01 -6.63938820e-01 7.57314205e-01 3.46576959e-01
-1.20082176e+00 -1.48961255e-02 -4.10503149e-01 -2.00138822e-01
1.35015890e-01 9.34369504e-01 -7.89344013e-01 6.36139333e-01
8.23568761e-01 1.39692414e+00 -5.34175158e-01 5.65882564e-01
-1.88725621e-01 6.28186166e-01 -1.84563205e-01 -2.54099667e-01
4.37869161e-01 -1.60876513e-01 3.19057375e-01 1.50140536e+00
1.89455435e-01 1.23090474e-02 2.85909146e-01 7.65294731e-01
-4.08966869e-01 1.85175642e-01 -7.19798326e-01 -1.64108068e-01
6.39476180e-01 1.55674124e+00 -5.68904579e-01 -3.85327756e-01
-5.70079744e-01 7.09322631e-01 5.35906851e-01 -4.17284109e-03
-6.29959166e-01 -3.73542488e-01 8.11806619e-01 2.93158501e-01
2.74135023e-01 -1.28298700e-01 -2.79156238e-01 -1.35682106e+00
2.70735651e-01 -9.30533826e-01 6.09159410e-01 -7.14694738e-01
-1.88528061e+00 3.58295470e-01 1.36564836e-01 -1.31404042e+00
-1.37410313e-01 -6.98892593e-01 -4.75021333e-01 6.48102224e-01
-1.29732943e+00 -1.32542992e+00 -7.60142505e-02 3.57868403e-01
7.49222040e-01 1.71319187e-01 8.44493747e-01 -4.06507105e-02
-1.86936826e-01 4.62964088e-01 -7.72395507e-02 5.06945103e-02
8.87960255e-01 -1.68688035e+00 6.55590236e-01 4.25132334e-01
7.26067483e-01 1.12400341e+00 8.34100485e-01 -5.47811389e-01
-1.06532943e+00 -1.04870713e+00 1.43522871e+00 -1.22659266e+00
1.38277912e+00 -7.05891192e-01 -9.97552037e-01 6.75319254e-01
2.52090096e-01 2.02520043e-02 1.22859049e+00 9.29769754e-01
-9.37375426e-01 -5.00528440e-02 -5.20258248e-01 6.75529063e-01
1.05213296e+00 -8.64163637e-01 -1.17358112e+00 5.44121027e-01
1.29969597e+00 -9.77825224e-02 -1.07284105e+00 3.31006110e-01
3.32246661e-01 -8.37349772e-01 9.61998284e-01 -8.11537504e-01
9.98374403e-01 -1.62840277e-01 -3.42987120e-01 -1.61414015e+00
-4.03427839e-01 -6.42763078e-01 -1.74517334e-01 1.52398872e+00
6.87214911e-01 -3.30845147e-01 6.53235674e-01 3.43615443e-01
-3.54083106e-02 -1.03885508e+00 -3.89282942e-01 -4.45213795e-01
1.72470793e-01 -5.88363051e-01 5.90442479e-01 1.08450520e+00
7.92907715e-01 9.74368751e-01 -2.11695358e-01 9.07705724e-02
3.71257097e-01 6.29135072e-01 7.08978236e-01 -1.43741810e+00
-1.42587036e-01 -6.28436506e-01 -3.95672798e-01 -1.25120008e+00
4.83593136e-01 -1.11384749e+00 -3.08813415e-02 -1.51766944e+00
4.63481784e-01 -5.20001888e-01 -4.36823756e-01 2.45924160e-01
-2.98439115e-01 1.85420364e-01 9.98923462e-03 5.40997982e-01
-1.02266455e+00 2.64590681e-01 9.05202627e-01 -3.97599488e-01
3.76657657e-02 3.35928425e-02 -1.34972322e+00 9.59999979e-01
4.02896821e-01 -3.41131687e-01 -4.56278026e-01 -2.27094144e-01
1.35766411e+00 -1.61712050e-01 2.88005412e-01 -1.48475155e-01
3.67423356e-01 -1.43911257e-01 -1.69795677e-01 -8.86506319e-01
-8.96110758e-02 -6.54379189e-01 -3.41674000e-01 3.98678593e-02
-9.44639027e-01 3.18847656e-01 -1.18477844e-01 6.89981282e-01
-6.27439916e-01 -1.01194687e-01 3.23293172e-02 6.46290835e-03
-5.44723153e-01 2.15517849e-01 2.80077994e-01 3.86938453e-01
6.18874013e-01 1.71824515e-01 -4.44393069e-01 -1.43731281e-01
-4.38492745e-01 1.81843042e-01 3.20811540e-01 5.90027690e-01
7.25945294e-01 -1.52985716e+00 -7.16738343e-01 -2.48448431e-01
7.49842942e-01 2.04704016e-01 -2.34955192e-01 8.98061395e-01
-6.65575564e-02 5.75189650e-01 2.56900549e-01 -6.12900257e-01
-1.48494256e+00 6.46987259e-01 -2.56832927e-01 -5.33417165e-01
-6.44863904e-01 1.00915718e+00 4.73787010e-01 -4.98958498e-01
-1.45296991e-01 -9.96261537e-01 -4.06650573e-01 7.02816129e-01
5.35936534e-01 4.97373864e-02 2.14766830e-01 -6.49688125e-01
-4.87474352e-01 8.01404297e-01 -5.49032271e-01 2.49053031e-01
1.38221991e+00 -3.70437890e-01 -6.88158035e-01 1.02239001e+00
1.57500839e+00 5.50867617e-01 -8.15682769e-01 -6.73303127e-01
5.93757153e-01 -4.30841953e-01 -2.81186759e-01 -4.91957307e-01
-2.75330514e-01 7.15877116e-01 -5.10171890e-01 8.71270537e-01
6.09577417e-01 5.84897578e-01 8.74450624e-01 4.73159313e-01
1.39910549e-01 -8.45718026e-01 3.00319582e-01 7.32627690e-01
9.89184380e-01 -1.37751508e+00 7.89235458e-02 -6.54042542e-01
-7.22434938e-01 9.76453781e-01 1.62349746e-01 -6.52129501e-02
7.96103299e-01 2.45463967e-01 -1.30959690e-01 -6.33059204e-01
-1.13396263e+00 -5.05904406e-02 8.88668001e-01 2.39886507e-01
6.55652583e-01 1.98962420e-01 -2.10423246e-01 9.94711697e-01
-5.95258415e-01 -4.94120151e-01 4.01278198e-01 5.07009208e-01
-4.82925087e-01 -8.20913196e-01 1.86122641e-01 8.43779266e-01
-6.58010423e-01 -5.24898529e-01 -7.22904444e-01 3.82758915e-01
3.81336883e-02 7.97227442e-01 1.41161367e-01 -5.51371694e-01
2.94639885e-01 -2.73124933e-01 -3.45999561e-02 -1.04182637e+00
-4.73499954e-01 -3.22766930e-01 2.97591299e-01 -5.53390503e-01
-4.73989248e-01 -6.37315094e-01 -1.02275300e+00 -2.91867077e-01
-3.57431918e-01 6.05697274e-01 1.85336292e-01 1.26841962e+00
5.15296161e-01 3.28968287e-01 7.77103007e-01 -4.85760450e-01
-5.80916643e-01 -7.63027847e-01 -4.10767794e-01 9.63751435e-01
4.63159114e-01 -3.96751285e-01 -4.73078728e-01 1.32639199e-01]
|
[11.269048690795898, 8.786035537719727]
|
1ccd9ccd-8e24-46ba-a933-f541099f04f8
|
results-of-semtab-2022
| null | null |
https://www.semanticscholar.org/paper/Results-of-SemTab-2022-Abdelmageed-Chen/64dfbc1da6ad7402a6365c3e41667069a63599a6
|
https://ceur-ws.org/Vol-3320/paper0.pdf
|
Results of SemTab 2022
|
SemTab 2022 was the fourth edition of the Semantic Web Challenge on Tabular Data to Knowledge Graph Matching, successfully collocated with the 21st International Semantic Web Conference (ISWC) and the 17th Ontology Matching (OM) Workshop. SemTab provides a common framework to conduct a systematic evaluation of state-of-the-art systems. In this paper, we give an overview of the 2022’s edition of the challenge and summarize the results.
|
['Kavitha Srinivas', 'Juan Sequeda', 'Ernesto Jiménez-Ruiz', 'Madelon Hulsebos', 'Oktie Hassanzadeh', 'Vasilis Efthymiou', 'Vincenzo Cutrona', 'Jiaoyan Chen', 'Nora Abdelmageed']
|
2022-10-25
| null | null | null |
semtab-iswc-2022-10
|
['graph-matching', 'ontology-matching', 'column-type-annotation', 'cell-entity-annotation']
|
['graphs', 'knowledge-base', 'natural-language-processing', 'natural-language-processing']
|
[-3.18961106e-02 5.54221034e-01 -4.72445995e-01 -8.52105021e-02
-2.36348778e-01 -6.61282122e-01 8.64572763e-01 6.14779532e-01
-1.44431710e-01 4.67115551e-01 3.58165652e-01 -2.10117385e-01
-8.01894367e-01 -1.05928421e+00 -3.66127372e-01 6.44774258e-01
-1.54915107e-02 9.96121883e-01 7.83684194e-01 -7.19540298e-01
1.50213242e-01 5.24217598e-02 -2.28414488e+00 7.80128241e-01
7.11784303e-01 1.34060216e+00 -2.79544681e-01 -1.69002086e-01
-1.00874174e+00 1.11475575e+00 -2.23826505e-02 -6.73990726e-01
2.16229245e-01 5.72829731e-02 -1.57913494e+00 -5.72907150e-01
9.08877552e-01 7.16196656e-01 -3.05750161e-01 1.38367760e+00
1.36833191e-01 5.32648079e-02 1.01954319e-01 -1.83714557e+00
-6.08089745e-01 6.75006568e-01 6.22772053e-02 3.64656635e-02
1.14514351e+00 -8.73841524e-01 1.17192948e+00 -6.49431467e-01
1.42960560e+00 1.38615704e+00 8.72732997e-01 5.60939610e-01
-6.84645951e-01 -5.35783648e-01 -9.20223221e-02 1.20859444e+00
-1.51601982e+00 -6.37291074e-01 3.41641068e-01 -4.74223256e-01
1.36953628e+00 6.84477150e-01 5.08184612e-01 1.74111053e-01
-2.14155301e-01 2.00206026e-01 8.26372325e-01 -7.25062490e-01
1.19194530e-01 2.78998286e-01 2.14158148e-01 7.00764716e-01
7.09534585e-01 -2.07428098e-01 -6.15583897e-01 -4.83767897e-01
9.15016010e-02 -7.24569857e-01 7.65564069e-02 -9.48922515e-01
-8.09865475e-01 4.93411988e-01 2.19832048e-01 5.23906291e-01
-1.53302968e-01 -2.04851568e-01 7.01214671e-01 6.03426158e-01
1.85847804e-02 7.36573577e-01 -4.07756597e-01 -2.13809721e-02
-4.13029104e-01 5.00950456e-01 1.15209019e+00 1.56479585e+00
6.19788706e-01 -4.67701644e-01 3.36158305e-01 1.06099200e+00
5.21081924e-01 1.34727061e-01 4.11167264e-01 -1.06765211e+00
6.82403564e-01 1.15597963e+00 1.67887837e-01 -1.09938097e+00
-4.52909499e-01 1.18179128e-01 1.29003167e-01 -1.29734529e-02
2.13262603e-01 3.20954502e-01 -6.01643801e-01 1.17061341e+00
2.69513488e-01 -2.65453070e-01 1.80546165e-01 5.69166899e-01
1.41728437e+00 -1.63377255e-01 3.40651870e-01 -7.71375448e-02
1.72222769e+00 -6.93094194e-01 -1.07328844e+00 -3.45951170e-01
6.94220066e-01 -8.97009909e-01 4.80907530e-01 -1.38016775e-01
-9.89855230e-01 -2.64974624e-01 -1.15605152e+00 -1.14311881e-01
-1.48612094e+00 -1.02006543e+00 7.03398049e-01 7.53432512e-01
-1.00899792e+00 4.53197449e-01 -3.27630222e-01 -1.30763161e+00
1.88295543e-01 -5.00279889e-02 -7.23237872e-01 -3.33146036e-01
-1.98765481e+00 1.64861131e+00 9.79237258e-01 -6.90757275e-01
1.73876733e-01 -1.02609241e+00 -8.48307550e-01 -2.29108647e-01
7.87402332e-01 -7.89061368e-01 1.00168157e+00 -3.74415815e-01
-8.24393153e-01 1.34617174e+00 -1.25012428e-01 -6.62529826e-01
8.27351436e-02 2.83931762e-01 -1.58810186e+00 -2.99918074e-02
2.85559118e-01 4.78482872e-01 -1.55710325e-01 -9.17048633e-01
-9.15784359e-01 -5.82933664e-01 2.94714481e-01 -3.40763256e-02
-3.29535604e-01 5.72604120e-01 -7.39849031e-01 -1.44574493e-01
2.77429074e-01 -3.73083830e-01 1.97369188e-01 -3.80110592e-02
2.82963574e-01 -7.09223032e-01 9.90374267e-01 -5.20481825e-01
1.59503698e+00 -1.42131090e+00 3.11948098e-02 3.32244039e-01
1.16108440e-01 1.04472935e-02 2.31720731e-02 1.28700125e+00
-3.52434665e-01 3.32825720e-01 2.89390653e-01 9.79166627e-02
5.85763454e-01 1.96003959e-01 -2.15550870e-01 -1.06443383e-01
-6.29122853e-01 9.39159036e-01 -1.08146715e+00 -7.32771397e-01
2.37591397e-02 -6.79270700e-02 -2.16734838e-02 -2.95548350e-01
-4.07912731e-01 -5.63552618e-01 -2.40678400e-01 7.81648040e-01
3.80492598e-01 -3.64708066e-01 9.63267982e-01 -5.58738172e-01
-2.18362108e-01 7.85850108e-01 -1.16830444e+00 2.06550813e+00
-1.01394737e-02 3.47632080e-01 -2.62888849e-01 -6.53623223e-01
7.09012628e-01 6.42562330e-01 9.10097003e-01 -1.41394556e+00
-2.04176947e-01 7.81725585e-01 -4.91896540e-01 -6.99428380e-01
5.81137836e-01 -1.12971857e-01 -5.53107932e-02 8.12600330e-02
1.80710524e-01 -2.19900146e-01 7.98532426e-01 1.85861468e-01
1.13969243e+00 4.19451147e-01 7.33546555e-01 -6.70132637e-01
7.39622653e-01 5.84508061e-01 3.66841286e-01 5.01302838e-01
2.42883265e-01 2.13316292e-01 -5.81415892e-02 -6.96315229e-01
-8.04833174e-01 -6.64411545e-01 -2.74034411e-01 5.88214219e-01
4.46339458e-01 -1.34860170e+00 -6.35122657e-01 -9.27420855e-01
3.23979199e-01 8.74787271e-01 -5.11469722e-01 1.44016501e-02
-2.95680553e-01 -1.21333160e-01 7.08532870e-01 2.01304823e-01
4.94459003e-01 -1.13500953e+00 -4.10562724e-01 1.34642094e-01
-3.74565721e-01 -1.44289649e+00 2.67015100e-02 -3.25901181e-01
-5.59313416e-01 -1.85701501e+00 4.04830039e-01 -7.72035003e-01
8.70060474e-02 1.33818239e-02 1.60097253e+00 2.67976105e-01
-1.46289557e-01 5.78693032e-01 -6.25653505e-01 -5.36229551e-01
-2.88936347e-01 2.50161886e-01 -1.08821364e-02 -6.61024153e-01
1.23628283e+00 -3.96737188e-01 -2.12761775e-01 4.60282177e-01
-1.00351679e+00 -2.50279885e-02 -4.09127414e-01 8.23824853e-02
2.98946768e-01 3.17443371e-01 6.54231250e-01 -1.01458538e+00
5.02373874e-01 -5.76914668e-01 -1.01332593e+00 8.70423913e-01
-1.36054802e+00 -1.01547968e-02 2.61162043e-01 6.44740999e-01
-8.46524477e-01 -2.92423338e-01 -9.07987878e-02 -4.66921972e-03
1.20511346e-01 1.20229495e+00 -2.26522490e-01 -3.71215165e-01
4.92647260e-01 -1.89065516e-01 -2.48298451e-01 -9.05414939e-01
3.72576147e-01 8.25265169e-01 6.45724177e-01 -7.93263674e-01
1.02817523e+00 5.32487512e-01 9.11228582e-02 -1.75975069e-01
-8.01024437e-01 -7.70597935e-01 -5.36789119e-01 -3.81753057e-01
6.11316621e-01 -6.20856583e-01 -7.13699877e-01 2.03983590e-01
-9.80108082e-01 1.38217822e-01 -3.21929276e-01 -1.39021114e-01
-4.19005543e-01 3.03551525e-01 2.65189230e-01 -4.54624504e-01
-4.84936953e-01 -3.11518401e-01 5.88623703e-01 -5.60132079e-02
-4.36481833e-01 -1.46504033e+00 3.68528962e-01 8.82218719e-01
5.29307485e-01 1.01183005e-01 1.07583034e+00 -9.34161544e-01
3.97082865e-02 -5.14665842e-01 -4.51553911e-01 -8.88208076e-02
3.54688466e-01 -4.56565201e-01 -5.92200756e-01 -1.86600789e-01
-6.37038529e-01 -1.20379716e-01 8.51820186e-02 -5.72802663e-01
8.70773137e-01 -7.86928013e-02 -5.83778083e-01 2.66800016e-01
1.79189360e+00 5.53451657e-01 1.02968109e+00 1.24658954e+00
4.42944407e-01 9.36162472e-01 7.64151871e-01 -1.51921466e-01
1.02975154e+00 1.29163647e+00 4.79403883e-01 6.11813605e-01
-5.63412905e-01 -4.07652050e-01 -2.64937997e-01 9.74811673e-01
-1.44669697e-01 -5.78325912e-02 -1.40949547e+00 5.47006190e-01
-2.20567942e+00 -1.00859177e+00 -1.63706794e-01 2.28350306e+00
7.25362837e-01 -5.67675792e-02 4.27304655e-02 -1.32365199e-02
7.01018631e-01 -1.19165845e-01 -2.64325496e-02 -1.04782596e-01
-4.52300131e-01 5.03505528e-01 6.60489380e-01 4.71718073e-01
-7.58799493e-01 1.01093483e+00 7.20129251e+00 6.98714495e-01
-4.05840427e-01 2.90137798e-01 -7.31768370e-01 4.78881985e-01
-5.31740367e-01 5.64962327e-01 -8.93153489e-01 2.28213966e-01
1.18218291e+00 -1.07164419e+00 8.00736248e-01 5.21145761e-01
-5.71159065e-01 3.93645406e-01 -9.91425514e-01 7.03580201e-01
2.99449116e-01 -1.99991035e+00 3.80432367e-01 -1.45666972e-01
4.77895796e-01 3.35541278e-01 -6.16230786e-01 2.21609220e-01
3.45766664e-01 -9.60743129e-01 8.54489446e-01 5.59791028e-01
9.22419488e-01 -4.02675658e-01 7.39183664e-01 -4.15430367e-01
-1.53915954e+00 -1.90053657e-02 -2.33780712e-01 8.03917646e-02
4.03287411e-02 -1.06062479e-01 -1.37655765e-01 1.68369973e+00
1.01164436e+00 9.85648572e-01 -5.54820001e-01 1.27677739e+00
1.56568646e-01 -2.90565103e-01 -1.82588443e-01 2.11691603e-01
-1.03311419e-01 4.82629873e-02 6.02253735e-01 9.30305064e-01
2.51124978e-01 -7.15699866e-02 -1.07784495e-01 4.10140902e-01
-2.95988917e-01 2.41132334e-01 -3.86088192e-01 -2.02717394e-01
1.18435264e+00 8.88993442e-01 -6.84576854e-02 -5.70211172e-01
-7.17450440e-01 3.49402696e-01 3.04455340e-01 -2.26970240e-02
-3.45422894e-01 -8.09776068e-01 7.09903777e-01 3.99381191e-01
1.97673049e-02 3.60712767e-01 -1.51021674e-01 -1.13071740e+00
1.02148063e-01 -1.04382765e+00 1.25352395e+00 -1.23374462e+00
-1.53704178e+00 6.60822690e-01 4.57002938e-01 -1.14948142e+00
-2.61281610e-01 -5.54567814e-01 1.48550883e-01 8.30489814e-01
-1.63124907e+00 -1.05464005e+00 -4.90783811e-01 5.78503668e-01
-7.02130273e-02 -3.15395892e-01 1.47425795e+00 1.01258409e+00
2.04451699e-02 3.49496633e-01 -2.52648711e-01 -1.62456438e-01
6.27741754e-01 -1.22160244e+00 6.61202908e-01 5.23066044e-01
-3.96742709e-02 7.47227013e-01 7.96899796e-01 -8.38622093e-01
-1.59744215e+00 -9.58757102e-01 1.59914422e+00 -6.87100708e-01
1.35060620e+00 -1.29728079e-01 -6.74445391e-01 9.65000689e-01
5.47048271e-01 -1.17098495e-01 3.92777026e-01 -2.78861471e-03
-8.49423110e-01 -2.85502493e-01 -1.26681054e+00 4.21723366e-01
1.55087698e+00 -5.66974998e-01 -1.11065018e+00 3.96794617e-01
6.30850554e-01 -4.79392618e-01 -1.73394263e+00 7.81538308e-01
9.47742343e-01 -5.21509707e-01 1.00288534e+00 -9.17711914e-01
-1.33156583e-01 -6.50874853e-01 -6.25988901e-01 -9.08577502e-01
-1.78048670e-01 -6.27747476e-01 -2.39630684e-01 1.03165233e+00
3.49180639e-01 -9.55818892e-01 7.97781348e-01 7.14766920e-01
-3.71515483e-01 4.81492728e-02 -1.28194511e+00 -1.30185962e+00
-1.19246475e-01 -3.97514492e-01 1.12718606e+00 1.43648887e+00
1.01385272e+00 -1.26622573e-01 1.00889675e-01 -9.23956409e-02
9.98453081e-01 1.05721541e-01 5.46951294e-01 -1.87783360e+00
4.14460927e-01 -4.83772188e-01 -9.07982767e-01 -2.71477193e-01
-1.03414685e-01 -1.29369593e+00 -6.70001030e-01 -2.22792006e+00
2.10355744e-01 -2.54271299e-01 -6.27175689e-01 9.14936244e-01
3.14857244e-01 1.70502290e-01 2.61084009e-02 1.48576438e-01
-1.00337851e+00 -7.56239146e-02 6.96451783e-01 -2.20209911e-01
5.36345065e-01 -7.15565383e-01 -7.04949141e-01 1.30027041e-01
4.76261944e-01 -6.36795759e-01 -6.46695077e-01 -3.81003678e-01
8.35372388e-01 -2.01865628e-01 8.84976014e-02 -1.04229307e+00
7.36052811e-01 -4.67245817e-01 -5.27756929e-01 -4.46628362e-01
2.02697054e-01 -1.26705897e+00 1.04553652e+00 3.16368937e-01
-1.20832957e-02 1.22868851e-01 6.24566317e-01 -1.16923168e-01
-5.41476786e-01 -3.85133713e-01 3.91423434e-01 -1.14098147e-01
-1.32051241e+00 2.18140811e-01 -8.87887180e-02 4.14404869e-01
7.27832496e-01 -3.32026958e-01 -1.12552392e+00 1.39506444e-01
-5.67273498e-01 6.38994217e-01 5.47936022e-01 1.16041923e+00
3.03503215e-01 -1.62292194e+00 -3.65633994e-01 -3.21753204e-01
1.09157550e+00 -9.53034520e-01 -3.89356762e-02 6.18201435e-01
-4.26875621e-01 1.15561831e+00 -6.92226350e-01 3.66232783e-01
-1.18491888e+00 5.64154387e-01 5.96666574e-01 -2.54386693e-01
-4.83623862e-01 9.63773653e-02 -7.90067017e-01 -8.47178876e-01
2.21893284e-02 5.60200572e-01 -5.62425137e-01 1.51406363e-01
5.81880033e-01 6.26977742e-01 7.10096896e-01 -3.69274735e-01
-1.06445622e+00 4.78976548e-01 2.04235196e-01 4.02461514e-02
1.30534697e+00 -1.92990437e-01 -9.08746183e-01 4.28744346e-01
8.91550481e-01 -3.20136398e-01 1.80293232e-01 -3.77015769e-01
1.14998853e+00 -3.57714653e-01 1.00498892e-01 -1.24466789e+00
-8.70037854e-01 -1.03550874e-01 3.95759940e-01 7.86593795e-01
7.22348511e-01 3.53424884e-02 9.26287711e-01 3.66954476e-01
9.65780020e-01 -1.37749279e+00 -8.78961265e-01 5.82971513e-01
9.29945111e-01 -6.70265436e-01 5.26324920e-02 -8.92706156e-01
6.99289739e-02 1.18172932e+00 6.48537576e-01 5.55598199e-01
7.21170545e-01 2.22199168e-02 2.22275749e-01 -1.09540212e+00
-9.12528813e-01 -4.80684876e-01 8.14631999e-01 9.97590542e-01
4.61032987e-01 -4.13156778e-01 -5.20578563e-01 2.84261286e-01
-3.09632480e-01 4.17967558e-01 -1.36321053e-01 1.16454840e+00
-3.82586300e-01 -1.85706365e+00 -9.57630947e-02 3.19926530e-01
-1.92781344e-01 -3.18689138e-01 -7.80896723e-01 1.02324831e+00
-5.80002880e-03 1.11214995e+00 1.17958069e-01 -7.20003307e-01
1.08831775e+00 4.64177758e-01 6.28984392e-01 -6.74523354e-01
-6.99621558e-01 -7.47893155e-01 1.18350506e+00 -9.41565096e-01
-6.65872335e-01 -7.70967662e-01 -1.22826076e+00 -7.38962889e-01
-4.06853780e-02 8.21433306e-01 7.71413565e-01 7.05847025e-01
5.34339607e-01 3.58468801e-01 -2.44586587e-01 4.07402277e-01
1.21031597e-01 -7.60354877e-01 -5.65913737e-01 6.56181812e-01
-6.57220483e-01 -9.14257646e-01 -1.08617224e-01 2.51453370e-02]
|
[9.319365501403809, 8.074041366577148]
|
405e5419-f66a-4be5-ab0f-aa75df9cd977
|
examining-deep-learning-architectures-for
|
1812.00602
| null |
http://arxiv.org/abs/1812.00602v1
|
http://arxiv.org/pdf/1812.00602v1.pdf
|
Examining Deep Learning Architectures for Crime Classification and Prediction
|
In this paper, a detailed study on crime classification and prediction using
deep learning architectures is presented. We examine the effectiveness of deep
learning algorithms on this domain and provide recommendations for designing
and training deep learning systems for predicting crime areas, using open data
from police reports. Having as training data time-series of crime types per
location, a comparative study of 10 state-of-the-art methods against 3
different deep learning configurations is conducted. In our experiments with
five publicly available datasets, we demonstrate that the deep learning-based
methods consistently outperform the existing best-performing methods. Moreover,
we evaluate the effectiveness of different parameters in the deep learning
architectures and give insights for configuring them in order to achieve
improved performance in crime classification and finally crime prediction.
|
['Theodoros Semertzidis', 'Petros Daras', 'Panagiotis Stalidis']
|
2018-12-03
| null | null | null | null |
['crime-prediction']
|
['miscellaneous']
|
[-3.80224228e-01 -2.20835954e-01 -1.50617078e-01 -7.08592594e-01
-6.55911684e-01 -1.27697974e-01 5.83239257e-01 5.33787966e-01
-7.95015991e-01 6.23700023e-01 5.40387571e-01 -5.84631741e-01
-4.04741138e-01 -1.27657521e+00 -5.41009128e-01 -2.80837297e-01
-3.03321183e-01 5.95132053e-01 -2.10461125e-01 -2.02786177e-01
4.99429673e-01 5.59302807e-01 -1.10370278e+00 6.64465189e-01
7.92318702e-01 6.48708999e-01 -3.19215178e-01 5.44996858e-01
3.61860767e-02 1.00864112e+00 -6.94680274e-01 -9.99101043e-01
2.27819890e-01 2.86082804e-01 -1.09977245e+00 -5.70653319e-01
5.60776830e-01 -7.06770003e-01 -5.28430283e-01 6.87322021e-01
7.02232838e-01 2.89390624e-01 9.60247099e-01 -6.20281398e-01
-1.12636900e+00 6.32433057e-01 -3.84639561e-01 9.04158235e-01
2.87560612e-01 -6.71499819e-02 6.80487335e-01 -6.16456747e-01
3.16186875e-01 9.21207249e-01 1.14382768e+00 5.01605868e-01
-9.65744615e-01 -5.51218271e-01 -2.63887886e-02 6.92264855e-01
-1.27560413e+00 -4.47268963e-01 7.14608729e-01 -9.80494618e-01
1.57649469e+00 1.01702497e-03 3.99136722e-01 1.39458764e+00
1.68483675e-01 7.69408464e-01 6.72458708e-01 -3.68308663e-01
3.11948299e-01 -6.26426116e-02 8.50160122e-01 4.78196442e-01
4.84160870e-01 1.28675893e-01 -1.36249915e-01 -3.55346739e-01
4.07891631e-01 3.46595973e-01 3.88988316e-01 4.93086517e-01
-1.42152652e-01 1.19771147e+00 5.16476452e-01 5.21141768e-01
-6.13306642e-01 -9.96352658e-02 8.44548583e-01 7.12252632e-02
1.27423692e+00 2.49405056e-01 -4.29284900e-01 -3.05991113e-01
-9.75587547e-01 6.38311803e-01 3.72145504e-01 1.67676508e-01
5.86277843e-01 4.86676283e-02 -3.63240629e-01 1.17624664e+00
-2.77764976e-01 1.31549537e-01 2.41496563e-01 -5.19471407e-01
1.03876078e+00 7.81331778e-01 -1.59324065e-01 -1.34381104e+00
-8.86975646e-01 -1.80246413e-01 -1.00962555e+00 -2.08951250e-01
3.31227779e-01 -5.32858014e-01 -6.91983700e-01 1.06652534e+00
-1.13009930e-01 5.06419897e-01 -2.54294537e-02 5.87866604e-01
9.08473551e-01 4.73061979e-01 3.55270147e-01 2.30904967e-01
8.66935909e-01 -4.94124621e-01 -5.72637916e-01 -2.57363170e-01
1.17529798e+00 -1.43089518e-01 5.72483838e-01 3.80436540e-01
-8.67359400e-01 -6.72609925e-01 -4.84814972e-01 -2.59307861e-01
-9.11983371e-01 2.54534334e-01 7.85908282e-01 1.05856490e+00
-8.24653804e-01 6.72029793e-01 -9.15167809e-01 -2.22671047e-01
9.81530368e-01 1.35339364e-01 -3.03447485e-01 -1.19166419e-01
-1.23431492e+00 1.01823652e+00 3.98981661e-01 3.91304642e-02
-4.91088659e-01 -8.43022108e-01 -1.02179396e+00 1.67825684e-01
-4.88393933e-01 -4.45419513e-02 8.21362019e-01 -4.25776184e-01
-6.07348025e-01 1.07382405e+00 -5.91837578e-02 -9.06548679e-01
2.77996093e-01 -4.11805153e-01 -4.11704212e-01 -1.24202602e-01
3.42375427e-01 1.69125974e-01 2.41622478e-02 -6.59745395e-01
-8.34194541e-01 -7.04179287e-01 3.69927853e-01 -4.66495633e-01
-9.10402238e-01 5.40070176e-01 8.46996382e-02 -5.94934046e-01
-4.79364097e-01 -4.62617189e-01 -4.72525179e-01 -9.03297365e-01
-1.87969565e-01 -5.61583877e-01 4.43966687e-01 -9.90473986e-01
1.74899924e+00 -1.85300326e+00 -3.66844982e-01 1.15260221e-01
5.89089701e-03 6.73905134e-01 -1.77739173e-01 4.07028824e-01
-2.57807761e-01 2.05138788e-01 3.48912016e-03 -6.48346007e-01
-1.83270618e-01 1.47136465e-01 -4.43910718e-01 5.27805269e-01
2.58728653e-01 7.22465038e-01 -7.54359365e-01 -1.98450863e-01
4.54336315e-01 5.10066867e-01 -9.30616081e-01 -1.68220103e-02
4.70113277e-01 4.83555906e-02 -3.61939311e-01 5.36924064e-01
8.65836740e-01 1.89051896e-01 -1.12250715e-01 3.77450198e-01
-1.54575378e-01 5.15699208e-01 -6.49310648e-01 1.15157616e+00
-6.94648802e-01 1.04783237e+00 -5.01169860e-01 -1.49373138e+00
1.10577893e+00 2.60625839e-01 4.87105727e-01 -1.01924479e+00
6.41901493e-02 -9.35711265e-02 -1.53982952e-01 -8.09294939e-01
6.87940419e-01 3.83307189e-01 -1.35847226e-01 7.54786313e-01
1.54927194e-01 6.45513773e-01 3.45008045e-01 -2.23981872e-01
1.36054444e+00 -4.85593170e-01 -1.22915953e-01 -2.06736565e-01
2.98307210e-01 -2.76302963e-01 4.68198776e-01 1.03803158e+00
-2.71044493e-01 5.84580839e-01 5.98932683e-01 -1.77448726e+00
-9.84181702e-01 -5.03441513e-01 -5.82502127e-01 1.36235011e+00
-8.29847813e-01 -2.89787859e-01 -1.16025543e+00 -5.52352726e-01
-1.66956782e-01 6.17921770e-01 -1.15994632e+00 1.96148325e-02
-7.62011528e-01 -1.51214051e+00 1.01851630e+00 9.33953106e-01
5.66161692e-01 -1.12892413e+00 -9.36139345e-01 2.38825411e-01
-2.73832321e-01 -9.92575645e-01 4.67140347e-01 -1.01656001e-03
-6.05883896e-01 -1.29499078e+00 -4.71261322e-01 -5.24315357e-01
2.43559122e-01 -4.99124117e-02 1.08523071e+00 2.94795424e-01
-3.18895847e-01 -6.86966702e-02 -7.22607970e-01 -7.00670362e-01
1.49745243e-02 5.95346093e-01 7.84148052e-02 -1.31115243e-02
1.17350841e+00 -5.08913457e-01 -3.67583424e-01 -5.72803020e-01
-5.63413024e-01 -2.21765459e-01 -1.76713154e-01 6.62352145e-01
-1.59930021e-01 1.52358562e-01 5.74834466e-01 -9.46598291e-01
1.13922668e+00 -8.83836567e-01 -4.85504746e-01 2.10286424e-01
-2.92567015e-01 -3.40059191e-01 5.47105551e-01 6.95939735e-02
-9.50853348e-01 -3.03134352e-01 -5.50123394e-01 -9.14974138e-02
-7.15063870e-01 7.04626799e-01 5.64281344e-01 3.57152998e-01
1.02460909e+00 -8.11132118e-02 -9.65539396e-01 -7.71576345e-01
-1.53131738e-01 8.33132327e-01 2.70657837e-01 -6.11961544e-01
-3.63238826e-02 4.21152234e-01 -3.62208486e-01 -7.67537177e-01
-9.32956100e-01 -3.43054682e-01 -1.11527252e+00 -9.37033519e-02
1.13153970e+00 -6.70237839e-01 -5.62710583e-01 7.55810976e-01
-1.79816580e+00 -4.25268620e-01 2.81821877e-01 2.19059527e-01
-2.49477208e-01 -1.97585046e-01 -9.88317251e-01 -1.14447010e+00
-5.28029203e-01 -9.44057465e-01 1.00280929e+00 8.28211606e-02
-1.72161639e-01 -1.55003321e+00 5.89745045e-01 6.02541983e-01
1.72737002e-01 5.20840704e-01 8.70242476e-01 -9.59991336e-01
2.81168252e-01 -3.99035364e-01 -3.06392908e-01 1.68861538e-01
-6.39222264e-02 3.41208875e-02 -1.41406155e+00 3.57831009e-02
-6.45924568e-01 -2.47627050e-01 1.38215804e+00 8.72426808e-01
1.76006520e+00 -4.07934964e-01 -1.71866089e-01 8.14653814e-01
1.26870418e+00 1.67577118e-01 7.96505570e-01 9.23671186e-01
6.30392134e-01 8.09921920e-01 2.04040036e-01 7.66403437e-01
6.04826212e-01 3.45659494e-01 2.07322255e-01 -2.26608321e-01
5.43424547e-01 -1.54976305e-02 -8.29943568e-02 1.02733463e-01
-8.18855107e-01 1.63385347e-02 -1.72090578e+00 9.21830535e-01
-2.04692817e+00 -1.53649580e+00 -2.28451565e-01 1.59911656e+00
3.85984957e-01 5.52783497e-02 3.60774308e-01 4.72763926e-01
4.90707248e-01 2.10746229e-01 5.26312441e-02 -1.00417125e+00
-9.15257260e-02 8.46999645e-01 4.84481305e-01 3.94717753e-01
-1.68123889e+00 1.16697991e+00 7.39259577e+00 7.53652692e-01
-9.04541016e-01 4.78332520e-01 1.11245167e+00 -3.65303338e-01
2.37407357e-01 -4.78240460e-01 -7.57367671e-01 4.75417018e-01
1.20058465e+00 1.49539322e-01 1.61772802e-01 8.20374489e-01
4.69118237e-01 -1.50679275e-02 -8.43466043e-01 8.78419459e-01
-1.60243303e-01 -1.84380233e+00 -9.14730355e-02 1.40443906e-01
7.70454586e-01 3.49451244e-01 1.55924559e-01 5.10490477e-01
4.52409089e-01 -1.21081412e+00 4.42997098e-01 5.75321198e-01
3.39407325e-01 -1.23624039e+00 1.15938461e+00 4.25213933e-01
-5.97288609e-01 -7.76830375e-01 -7.27252722e-01 -8.45668793e-01
2.35749669e-02 4.99689519e-01 -5.55279374e-01 3.32028836e-01
1.19640613e+00 1.03525305e+00 -7.46229947e-01 8.32491159e-01
-2.88571175e-02 8.55720162e-01 1.31379515e-01 1.81626484e-01
6.73427224e-01 -3.18294913e-02 -3.33253741e-01 1.89051831e+00
1.71251610e-01 5.10921240e-01 -1.07644927e-02 8.07502627e-01
1.63296565e-01 -6.48571178e-02 -9.38668966e-01 4.38272804e-01
4.39741313e-01 9.02959943e-01 -3.68518263e-01 -3.08337867e-01
-4.99345392e-01 3.24267209e-01 8.70731354e-01 1.91622987e-01
-6.72407568e-01 -3.42820019e-01 1.00521290e+00 2.67857015e-01
8.91139209e-02 -2.68901372e-03 -7.27754116e-01 -1.04289353e+00
-1.34960100e-01 -4.43425357e-01 7.55298615e-01 -3.92553627e-01
-1.40425682e+00 6.72300756e-01 3.43343914e-01 -6.71694398e-01
-4.51223522e-01 -7.72693276e-01 -1.12365139e+00 7.23878980e-01
-1.33582664e+00 -1.12799573e+00 2.38277793e-01 7.11350977e-01
5.24231493e-01 -7.51976907e-01 9.63966548e-01 8.13625097e-01
-1.09150863e+00 7.27579176e-01 2.36512065e-01 1.04041624e+00
1.59079507e-01 -8.10778975e-01 8.86117280e-01 1.00605500e+00
1.36397406e-01 5.09124041e-01 2.48366520e-01 -4.40932631e-01
-3.48582268e-01 -1.34837782e+00 1.22173846e+00 -6.34266317e-01
4.99466866e-01 -3.18511963e-01 -8.93688500e-01 8.29999328e-01
-1.76047925e-02 -2.57711232e-01 1.06250000e+00 9.14674044e-01
-3.11190218e-01 -7.83183202e-02 -1.13343918e+00 1.31805375e-01
8.50835443e-01 -5.41583061e-01 -4.85227913e-01 4.27809954e-01
1.16532080e-01 -2.25658849e-01 -7.20598817e-01 2.14247748e-01
3.61295879e-01 -1.40625441e+00 9.21924293e-01 -1.31292391e+00
1.03065073e+00 8.96456897e-01 1.53441653e-02 -1.30467188e+00
-6.79337978e-01 2.76087791e-01 6.67860433e-02 9.71583664e-01
3.16898078e-01 -4.56587225e-01 9.29061890e-01 1.06065440e+00
-6.13192879e-02 -8.38134170e-01 -1.21449983e+00 -4.23195958e-01
7.58359909e-01 -1.08626246e+00 9.93654370e-01 1.38799620e+00
1.96568176e-01 -5.37600480e-02 -5.22384346e-01 2.42716506e-01
2.97205299e-01 -3.08547199e-01 6.66557074e-01 -1.25060713e+00
2.38022551e-01 -5.54135084e-01 -7.64748156e-01 -1.84317574e-01
6.65565312e-01 -5.09081066e-01 -8.99338067e-01 -1.41297626e+00
4.22853142e-01 -5.47067106e-01 -4.39146221e-01 9.30747628e-01
4.21634912e-02 1.95557922e-01 1.06371745e-01 -3.38740557e-01
-3.75813961e-01 2.73634270e-02 2.40515873e-01 -3.83360058e-01
-2.42760360e-01 1.34525687e-01 -5.87968409e-01 9.33864653e-01
9.69958067e-01 -6.14899158e-01 5.15851518e-03 -1.42593718e+00
2.69076049e-01 -1.18377581e-01 5.32871246e-01 -1.16458511e+00
9.62440819e-02 -2.48899251e-01 6.35391414e-01 -6.86550200e-01
1.72961727e-01 -3.98320019e-01 -5.23180604e-01 3.26002330e-01
-5.10289192e-01 1.55233040e-01 4.25434679e-01 -1.86326131e-02
-8.92190114e-02 -3.18925411e-01 4.12764788e-01 -1.17814038e-02
-5.59717238e-01 4.44910318e-01 -4.98761296e-01 -1.87368438e-01
7.51956224e-01 -2.38838047e-01 -4.19787884e-01 -1.25263005e-01
-7.55787611e-01 1.21283323e-01 -3.15613151e-01 7.52055407e-01
7.47518718e-01 -1.33374000e+00 -1.12148666e+00 4.23308969e-01
-1.17933005e-01 -3.37371856e-01 5.69343626e-01 2.87233174e-01
-6.18065000e-01 8.23890030e-01 -5.36540627e-01 -2.09309861e-01
-1.14299726e+00 3.30756634e-01 4.81691420e-01 -6.28669381e-01
-3.39097381e-01 6.51916265e-01 -2.63730019e-01 -7.50764370e-01
9.51770246e-02 -1.64117321e-01 -9.30422366e-01 1.25411770e-03
1.02245021e+00 9.25806105e-01 5.06401241e-01 -6.52387381e-01
-2.80476063e-01 5.47827519e-02 -1.77542210e-01 1.67474881e-01
2.18713570e+00 5.39039850e-01 5.66658862e-02 2.17452273e-02
1.16693974e+00 -6.81894660e-01 -7.23385334e-01 -4.47859466e-02
1.53493479e-01 -6.57441914e-01 3.89418304e-01 -4.88619328e-01
-1.33529842e+00 1.28422475e+00 6.81172729e-01 3.79009247e-01
1.06315064e+00 -2.62556791e-01 6.71176791e-01 6.10774159e-01
3.29023033e-01 -1.47097683e+00 -2.36272871e-01 1.07108700e+00
4.99742180e-01 -1.41140521e+00 -1.29087910e-01 3.49596143e-01
-4.08810556e-01 1.26386130e+00 7.03423798e-01 -5.58451712e-01
1.02181244e+00 1.73698217e-01 -1.18977025e-01 -4.29834574e-01
-7.31267750e-01 -7.24334121e-02 1.39504433e-01 8.44279706e-01
7.30566084e-01 3.19635063e-01 -2.41375372e-01 9.68771815e-01
-4.53322142e-01 -6.33140057e-02 4.79952574e-01 5.75958073e-01
-2.63549477e-01 -8.82804036e-01 -5.73428929e-01 6.84977114e-01
-1.02075934e+00 -1.68338731e-01 -4.06818926e-01 6.52296066e-01
6.10289872e-01 1.21229565e+00 6.72387540e-01 -5.06199539e-01
3.82612407e-01 -2.06819817e-01 2.32718110e-01 -6.50734544e-01
-9.29171145e-01 -6.53529167e-01 1.29905775e-01 -4.12253916e-01
-2.80324399e-01 -7.59587049e-01 -5.90872705e-01 -8.40048969e-01
2.83587840e-03 -9.28139165e-02 4.05219972e-01 1.18000889e+00
2.77128071e-01 2.29636535e-01 5.77027798e-01 -8.97835732e-01
-7.29281977e-02 -1.23601842e+00 -4.96917516e-01 4.57255691e-01
3.56147856e-01 -6.43903315e-01 2.25500152e-01 -1.38196588e-01]
|
[6.738507270812988, 1.9519312381744385]
|
d1f36613-efe9-48ca-a46f-9e34ebeddbbe
|
using-poisson-binomial-glms-to-reveal-voter
|
1802.01053
| null |
http://arxiv.org/abs/1802.01053v1
|
http://arxiv.org/pdf/1802.01053v1.pdf
|
Using Poisson Binomial GLMs to Reveal Voter Preferences
|
We present a new modeling technique for solving the problem of ecological
inference, in which individual-level associations are inferred from labeled
data available only at the aggregate level. We model aggregate count data as
arising from the Poisson binomial, the distribution of the sum of independent
but not identically distributed Bernoulli random variables. We relate
individual-level probabilities to individual covariates using both a logistic
regression and a neural network. A normal approximation is derived via the
Lyapunov Central Limit Theorem, allowing us to efficiently fit these models on
large datasets. We apply this technique to the problem of revealing voter
preferences in the 2016 presidential election, fitting a model to a sample of
over four million voters from the highly contested swing state of Pennsylvania.
We validate the model at the precinct level via a holdout set, and at the
individual level using weak labels, finding that the model is predictive and it
learns intuitively reasonable associations.
|
['Evan Rosenman', 'Nitin Viswanathan']
|
2018-02-04
| null | null | null | null |
['holdout-set']
|
['computer-vision']
|
[ 2.60149777e-01 8.46718997e-02 -4.89015281e-01 -5.05469620e-01
-6.83304489e-01 -7.41245210e-01 5.24638712e-01 4.11743999e-01
-9.21322823e-01 1.33084857e+00 3.32644045e-01 -6.24344945e-01
-2.04113945e-01 -8.08046401e-01 -1.08088195e+00 -4.66796607e-01
-5.55599511e-01 5.72348595e-01 -6.30508423e-01 2.29086310e-01
4.16993164e-03 -3.79291736e-02 -1.31361175e+00 -1.70158625e-01
8.54136884e-01 5.16874671e-01 2.45932024e-02 7.53514111e-01
4.37331647e-01 7.62553990e-01 -2.43366688e-01 -3.07314456e-01
1.81531191e-01 5.38640544e-02 -4.85551417e-01 -4.59847718e-01
7.77635276e-01 -4.21239823e-01 -3.08688320e-02 8.84404838e-01
2.77108818e-01 -3.23356926e-01 1.10608745e+00 -1.14939570e+00
-8.02912235e-01 9.70746100e-01 -8.20486307e-01 2.00056329e-01
3.15493345e-01 -5.26954457e-02 1.45456171e+00 -4.00002301e-01
3.67525488e-01 1.49373996e+00 1.07786751e+00 1.89734530e-02
-2.25260377e+00 -9.55313802e-01 3.98310244e-01 -3.13140750e-01
-1.42413604e+00 -1.33985952e-01 1.59152746e-01 -8.13714743e-01
5.07964134e-01 2.69813001e-01 7.80627787e-01 1.36622345e+00
3.34397852e-01 5.41002214e-01 1.70960760e+00 -6.80057779e-02
3.99582088e-01 -9.69425589e-02 3.50728869e-01 6.32502139e-01
6.44178569e-01 3.86351854e-01 -5.77921331e-01 -9.82240438e-01
7.89322913e-01 -9.70079452e-02 1.06839389e-01 1.02706529e-01
-9.57441807e-01 9.47801352e-01 5.08025110e-01 -3.90136451e-01
-5.68361640e-01 4.95973200e-01 -1.08001538e-01 3.57715297e-03
8.32052588e-01 2.25548506e-01 -6.28089368e-01 3.89673948e-01
-9.46597815e-01 6.25007868e-01 1.10303795e+00 4.11222667e-01
8.73552024e-01 -3.17336112e-01 8.91115740e-02 5.76555550e-01
4.39193636e-01 1.00984561e+00 -1.20340481e-01 -1.08742917e+00
9.81724188e-02 2.69378930e-01 5.32769799e-01 -9.49126363e-01
-5.42442679e-01 -4.83255178e-01 -7.90457666e-01 -1.63728464e-03
8.56056035e-01 -4.74339753e-01 -4.14204478e-01 2.50379157e+00
2.85735637e-01 2.69171685e-01 -5.21525025e-01 5.28169096e-01
3.44956011e-01 5.44342935e-01 6.87689185e-01 -2.94942498e-01
1.31784892e+00 1.99564442e-01 -2.78962344e-01 -3.65977645e-01
2.46825874e-01 -1.77496597e-02 9.53653455e-01 2.68031269e-01
-8.46644700e-01 -2.31856287e-01 -5.92544734e-01 1.21172003e-01
-3.51436645e-01 -4.25668173e-02 8.72263610e-01 4.61119682e-01
-1.07712305e+00 3.54023159e-01 -4.61263776e-01 -3.79298896e-01
3.76612186e-01 5.28500140e-01 -1.82712823e-01 2.74856269e-01
-1.10167706e+00 6.29188776e-01 -9.15635973e-02 -1.59285292e-02
-8.76309395e-01 -8.12971473e-01 -7.69274592e-01 2.06474856e-01
1.38215408e-01 -8.09114754e-01 1.07006168e+00 -6.55711293e-01
-7.18042314e-01 9.74379122e-01 -3.09755921e-01 -5.14699459e-01
4.21773434e-01 3.79492305e-02 5.06608225e-02 -4.82347190e-01
5.54767311e-01 7.94017732e-01 2.06205279e-01 -1.28563368e+00
-7.70616472e-01 -6.61983550e-01 -7.50769302e-02 -1.04491739e-02
-1.51422083e-01 -3.55668485e-01 6.65987849e-01 -3.59925926e-01
-1.28755122e-01 -9.64700639e-01 -4.58449543e-01 -1.44614637e-01
-2.41286933e-01 -3.22397411e-01 -6.98961318e-02 -8.46902013e-01
1.11896706e+00 -1.78596699e+00 -7.94272870e-02 3.82902056e-01
3.59562755e-01 -7.45401442e-01 -2.76401997e-01 1.92931384e-01
2.12507650e-01 3.87414545e-01 -4.46420044e-01 -1.64525107e-01
1.21969730e-01 2.15571940e-01 -4.70486462e-01 6.76418841e-01
-3.53450328e-02 9.45235252e-01 -1.07892084e+00 -5.14958858e-01
-2.77584493e-01 4.47430089e-02 -6.93942904e-01 -8.85970294e-02
-2.66247720e-01 3.76476705e-01 -1.74405336e-01 5.69388807e-01
5.59426546e-01 -2.96034068e-01 4.79047269e-01 2.73131520e-01
-2.55329102e-01 2.24510282e-01 -8.14641833e-01 1.01892793e+00
-3.66267085e-01 4.67079073e-01 3.48218799e-01 -7.06260979e-01
7.74691999e-01 -1.39777780e-01 3.62194449e-01 -3.56653005e-01
-8.81432891e-02 6.12629205e-02 -1.45088527e-02 -1.41490519e-01
4.83672678e-01 -4.83316958e-01 -6.77904010e-01 6.05438828e-01
-2.35931396e-01 7.91202784e-02 -2.21482188e-01 8.92514654e-04
9.37095702e-01 1.32728815e-01 5.56483269e-01 -7.17395365e-01
-8.58595818e-02 1.03674658e-01 5.95781803e-01 1.33920801e+00
-3.15166591e-03 2.84100585e-02 5.40668130e-01 -5.85003793e-01
-1.16425323e+00 -1.44165373e+00 -5.47986388e-01 1.39131451e+00
-7.67294243e-02 8.13637860e-04 -4.04269665e-01 -1.43975958e-01
4.93472368e-01 8.84172499e-01 -9.90982711e-01 8.38742554e-02
4.90404926e-02 -1.19628370e+00 5.48750460e-01 2.98076421e-01
8.70006159e-02 -7.20908940e-01 -3.80706906e-01 1.20036446e-01
-4.56416249e-01 -8.32072139e-01 -1.83201090e-01 4.30186033e-01
-6.66467667e-01 -8.66952419e-01 -1.66633502e-01 -4.38865989e-01
4.13737655e-01 -1.18240207e-01 1.28265750e+00 -1.02263093e-02
-3.14580262e-01 4.22092944e-01 5.28518021e-01 -5.03056228e-01
-2.36191258e-01 1.29791245e-01 5.30091763e-01 -1.89746156e-01
8.37092698e-01 -9.17216301e-01 -3.10059220e-01 -5.66485412e-02
-5.74136972e-01 -7.94288516e-02 3.51654172e-01 8.12189937e-01
4.30713534e-01 -4.52501506e-01 7.66729116e-01 -6.24403238e-01
7.39751101e-01 -9.85635519e-01 -9.27710295e-01 1.81176171e-01
-2.66425043e-01 9.46123060e-03 5.33640265e-01 -6.96327627e-01
-6.37432575e-01 6.02377988e-02 5.36003768e-01 4.52770412e-01
-3.84553403e-01 7.40657806e-01 1.90120012e-01 2.33325452e-01
7.69319177e-01 -9.95364413e-02 -1.42005980e-01 -3.01382452e-01
3.39791775e-01 8.14373851e-01 7.98977971e-01 -1.13498497e+00
4.27238971e-01 5.22890031e-01 4.07645077e-01 -9.78797853e-01
-1.10322475e+00 -5.88720702e-02 -6.63877130e-01 -2.12400138e-01
7.03376353e-01 -1.28440082e+00 -1.55164170e+00 2.84146778e-02
-8.70695472e-01 -8.32302451e-01 -2.73778409e-01 8.31498384e-01
-6.21151626e-01 7.39657134e-02 -6.18705213e-01 -1.33901644e+00
9.53456312e-02 -4.78159428e-01 1.05522370e+00 1.00170441e-01
-6.59305036e-01 -1.24532986e+00 6.27260745e-01 1.82855606e-01
-2.27700602e-02 5.40010512e-01 1.11419022e+00 -4.87542123e-01
-3.87998790e-01 -1.38371274e-01 -1.82526544e-01 -3.16443592e-01
-5.78417927e-02 2.33959496e-01 -9.75038648e-01 -1.32533714e-01
-2.20142916e-01 -6.62387490e-01 1.06567109e+00 9.95429933e-01
1.28033412e+00 -5.57974756e-01 -3.28736335e-01 5.36246240e-01
1.40302169e+00 -5.93000293e-01 4.33274209e-02 2.04369009e-01
3.96698087e-01 8.04659188e-01 -5.42089902e-02 7.00368464e-01
1.06891704e+00 1.29085526e-01 4.22529548e-01 1.60621136e-01
6.60885096e-01 -8.20307016e-01 3.62351060e-01 2.10319057e-01
3.53162661e-02 6.91614673e-02 -8.72656107e-01 5.44272363e-01
-1.94878578e+00 -1.32758772e+00 -3.90700459e-01 2.58975077e+00
1.11536741e+00 -1.44422442e-01 5.64217567e-01 -4.93325055e-01
7.51427352e-01 -9.32845324e-02 -6.86070144e-01 -2.85038918e-01
-3.83388788e-01 2.02607624e-02 1.08567524e+00 1.18410504e+00
-1.05501986e+00 7.72458196e-01 8.23315716e+00 6.06008887e-01
-6.76171362e-01 -2.26071000e-01 1.15790725e+00 -2.30108112e-01
-5.96745491e-01 1.63756013e-01 -6.84819341e-01 5.54435194e-01
1.03006053e+00 -1.83420643e-01 7.11795151e-01 5.09673953e-01
4.29307818e-01 -6.81869149e-01 -1.08580852e+00 5.18217325e-01
-3.48805577e-01 -7.60545194e-01 -3.88120472e-01 7.34732747e-01
7.72501349e-01 1.11723691e-01 2.71052986e-01 2.41360262e-01
1.52537060e+00 -1.52113295e+00 7.00133204e-01 7.02692568e-01
8.29034269e-01 -4.69524294e-01 3.28390658e-01 8.72850180e-01
-7.81193793e-01 -3.83275509e-01 -5.66697121e-01 -1.00420928e+00
-6.23029694e-02 6.12234175e-01 -7.69380033e-01 -1.22370854e-01
6.57679081e-01 4.83133435e-01 -4.76020724e-01 6.83805168e-01
-2.30211094e-01 1.02730799e+00 -9.88325179e-01 -1.78221941e-01
5.51323630e-02 -3.33936661e-01 1.73751369e-01 9.60397482e-01
2.46322423e-01 1.13423616e-01 1.19316205e-01 1.35528445e+00
-1.21048443e-01 -8.41537118e-02 -8.54011536e-01 3.38812292e-01
7.68430948e-01 1.35692811e+00 -3.73498529e-01 -1.64011970e-01
-2.42626533e-01 2.60251492e-01 6.97848916e-01 4.89636391e-01
-5.73951304e-01 5.92913926e-01 5.36687553e-01 -6.54365942e-02
1.13375652e-04 -3.74089569e-01 -7.34944165e-01 -1.06394112e+00
-3.88716042e-01 -5.68465650e-01 4.37151760e-01 -6.00005269e-01
-1.95520532e+00 -1.81752592e-01 4.75589424e-01 -4.45230514e-01
-2.23255649e-01 -6.82281137e-01 -4.85859603e-01 1.12601161e+00
-1.09423304e+00 -1.03625202e+00 -7.11912662e-02 1.55896768e-01
-3.30903530e-01 3.04256231e-01 7.53307402e-01 -1.60416439e-01
-4.22984570e-01 2.20938280e-01 2.34057233e-01 -2.51759380e-01
4.65835571e-01 -1.47287989e+00 3.97607267e-01 3.34615231e-01
6.10103123e-02 6.49441600e-01 7.84679353e-01 -8.80883098e-01
-9.85553980e-01 -6.91256940e-01 1.05457938e+00 -6.79024398e-01
8.99210453e-01 -7.16967583e-01 -5.08992255e-01 1.01405656e+00
-1.68507844e-01 -2.79838830e-01 1.21058142e+00 5.22528946e-01
-4.30423141e-01 5.29979216e-03 -1.27245378e+00 7.08508790e-01
9.70023572e-01 -6.88275874e-01 -4.03958887e-01 1.46233290e-01
3.01314831e-01 3.14978421e-01 -8.66317391e-01 8.84857923e-02
1.10562503e+00 -6.45349801e-01 9.53897953e-01 -9.03576970e-01
6.47763073e-01 -1.21228024e-01 -6.54833496e-01 -1.31049728e+00
-6.48840487e-01 -2.58828700e-01 2.08692759e-01 9.47030485e-01
4.51175272e-01 -5.27325690e-01 5.50934017e-01 7.87492812e-01
7.64056802e-01 -4.40954745e-01 -1.11213696e+00 -4.25137490e-01
5.62981009e-01 -2.94094026e-01 5.76300323e-01 9.66857314e-01
-1.42468154e-01 4.45994228e-01 -5.97859442e-01 3.91009390e-01
1.32924342e+00 9.56379697e-02 8.07877243e-01 -1.82187939e+00
-3.87365311e-01 -2.74466455e-01 -2.62729730e-03 -1.03291523e+00
5.82152307e-01 -7.51859426e-01 1.80746496e-01 -1.11433756e+00
7.87856400e-01 -5.50979257e-01 1.51075646e-01 3.21092606e-01
-4.48907375e-01 2.55269200e-01 -1.40266374e-01 1.64574206e-01
-1.98145851e-01 3.41023773e-01 6.92610025e-01 -1.34291962e-01
1.54011510e-02 -1.19310863e-01 -9.51782048e-01 8.41679573e-01
7.15041101e-01 -7.02910542e-01 2.00881943e-01 -3.61894876e-01
7.11866081e-01 4.97049317e-02 9.85006690e-01 -5.16105294e-01
2.12760776e-01 -7.18684077e-01 7.63676345e-01 -5.48293889e-01
2.62456983e-01 -7.94625401e-01 2.37804472e-01 5.37274957e-01
-8.00369442e-01 7.15336353e-02 7.38071650e-02 6.90878808e-01
5.44776917e-01 8.36439058e-02 5.31753898e-01 -2.25801378e-01
-6.96448423e-03 2.80661523e-01 -4.56613421e-01 1.23112872e-01
5.55233300e-01 2.05423966e-01 -3.75001639e-01 -7.22694576e-01
-6.01613343e-01 4.77795780e-01 5.47216833e-01 -2.11152181e-01
-1.61560196e-02 -1.07509184e+00 -1.20098162e+00 -3.52640031e-03
-1.63425207e-01 -1.92920431e-01 -1.67507067e-01 6.87157810e-01
-6.31571859e-02 8.87617618e-02 -8.70654657e-02 -6.50702000e-01
-8.58146131e-01 3.91051173e-01 3.09735924e-01 -2.55158067e-01
1.24945097e-01 5.50946951e-01 3.60516280e-01 -9.14954484e-01
-4.17869054e-02 -2.25800663e-01 5.05951680e-02 1.18945040e-01
3.38898182e-01 3.39349538e-01 -1.01211154e+00 -5.29991508e-01
-2.99128205e-01 3.68843853e-01 3.80024225e-01 -4.52435851e-01
1.53752613e+00 -4.53227073e-01 -3.88629735e-01 9.41460848e-01
7.78242946e-01 1.16389401e-01 -1.26554537e+00 -3.05308759e-01
1.49450913e-01 -2.93245465e-01 -1.84770420e-01 -8.13633978e-01
-8.08798447e-02 6.56819224e-01 2.74729401e-01 5.22781789e-01
4.32520628e-01 1.66530132e-01 1.25383168e-01 4.10432518e-01
4.26671177e-01 -7.66656816e-01 -5.35965204e-01 2.59669483e-01
4.79186088e-01 -1.31945527e+00 4.18307424e-01 1.25195846e-01
-2.24365532e-01 6.54318810e-01 3.77398759e-01 -4.09852952e-01
5.68366766e-01 3.35883647e-01 -3.15383643e-01 5.55299483e-02
-8.62459540e-01 5.18725067e-02 2.44781841e-02 6.08035922e-01
5.50569594e-01 6.33929372e-01 -3.59447092e-01 7.89272547e-01
-6.85461938e-01 -2.96328124e-02 5.67107439e-01 3.32545042e-01
-5.77862859e-01 -6.75379515e-01 -5.67553222e-01 8.63500655e-01
-5.40689111e-01 -4.27332729e-01 -6.78044915e-01 6.60796881e-01
1.29348665e-01 8.71840060e-01 4.32855487e-01 -1.28176853e-01
-2.16760769e-01 4.16565910e-02 3.83773893e-01 -3.95557433e-01
-3.95416558e-01 -1.80759624e-01 -6.92944229e-02 -7.68804997e-02
-4.99314487e-01 -8.78086686e-01 -5.26942968e-01 -8.56673837e-01
-6.88822418e-02 2.23571017e-01 4.72909957e-01 7.37602651e-01
-9.76843387e-02 -1.65864646e-01 8.10580194e-01 -8.55093241e-01
-9.15876448e-01 -1.10197628e+00 -8.14911067e-01 8.13630745e-02
5.07837474e-01 -2.73541570e-01 -7.25680053e-01 -1.85127065e-01]
|
[7.910074710845947, 4.710550785064697]
|
5ffbea3b-66a6-414d-b9a9-02e094cf79d9
|
adjust-a-dictionary-based-joint
|
2112.11406
| null |
https://arxiv.org/abs/2112.11406v2
|
https://arxiv.org/pdf/2112.11406v2.pdf
|
ADJUST: A Dictionary-Based Joint Reconstruction and Unmixing Method for Spectral Tomography
|
Advances in multi-spectral detectors are causing a paradigm shift in X-ray Computed Tomography (CT). Spectral information acquired from these detectors can be used to extract volumetric material composition maps of the object of interest. If the materials and their spectral responses are known a priori, the image reconstruction step is rather straightforward. If they are not known, however, the maps as well as the responses need to be estimated jointly. A conventional workflow in spectral CT involves performing volume reconstruction followed by material decomposition, or vice versa. However, these methods inherently suffer from the ill-posedness of the joint reconstruction problem. To resolve this issue, we propose 'A Dictionary-based Joint reconstruction and Unmixing method for Spectral Tomography' (ADJUST). Our formulation relies on forming a dictionary of spectral signatures of materials common in CT and prior knowledge of the number of materials present in an object. In particular, we decompose the spectral volume linearly in terms of spatial material maps, a spectral dictionary, and the indicator of materials for the dictionary elements. We propose a memory-efficient accelerated alternating proximal gradient method to find an approximate solution to the resulting bi-convex problem. From numerical demonstrations on several synthetic phantoms, we observe that ADJUST performs exceedingly well compared to other state-of-the-art methods. Additionally, we address the robustness of ADJUST against limited and noisy measurement patterns. The demonstration of the proposed approach on a spectral micro-CT dataset shows its potential for real-world applications. Code is available at https://github.com/mzeegers/ADJUST.
|
['Kees Joost Batenburg', 'Tristan van Leeuwen', 'Ajinkya Kadu', 'Mathé T. Zeegers']
|
2021-12-21
| null | null | null | null |
['inference-optimization', 'spectral-reconstruction', 'multispectral-object-detection', 'low-dose-x-ray-ct-reconstruction']
|
['audio', 'computer-vision', 'computer-vision', 'medical']
|
[ 4.60075378e-01 -3.29942137e-01 1.52394414e-01 -1.32168725e-01
-1.02679837e+00 -2.09249094e-01 1.89891025e-01 2.20157340e-01
-4.70805287e-01 5.80829740e-01 2.91603088e-01 3.14758122e-02
-1.71193257e-01 -5.93000174e-01 -5.30176640e-01 -1.14531207e+00
1.39457420e-01 7.00741529e-01 2.62087375e-01 9.73113701e-02
5.70179783e-02 5.62600017e-01 -1.13227999e+00 3.14633369e-01
6.45312786e-01 1.00234413e+00 6.19281888e-01 4.68860567e-01
-3.27846482e-02 5.49102187e-01 2.55272933e-03 2.59288311e-01
3.06880563e-01 -3.73718321e-01 -6.84855521e-01 4.01548892e-01
7.31418207e-02 -3.22448611e-01 -3.21283072e-01 1.07538557e+00
6.48725331e-01 2.22041681e-01 7.78096735e-01 -6.25832140e-01
6.19844086e-02 2.32643723e-01 -7.39356875e-01 2.05589026e-01
3.67966801e-01 1.19040854e-01 6.58959329e-01 -1.08277428e+00
5.97572029e-01 6.76712394e-01 6.79825664e-01 2.55647123e-01
-1.40909529e+00 -4.48069215e-01 -4.16907489e-01 9.87743437e-02
-1.40133238e+00 -4.37523842e-01 9.17418480e-01 -7.15225101e-01
6.19336009e-01 3.72680664e-01 7.41741896e-01 7.25574374e-01
9.43776891e-02 3.43370736e-01 1.33618093e+00 -5.13872445e-01
2.82742679e-01 2.25525841e-01 -1.93580031e-01 7.90570021e-01
1.16126627e-01 4.41069417e-02 -5.68045199e-01 -4.45753217e-01
8.37541878e-01 -2.06188075e-02 -6.63134634e-01 -6.28658533e-01
-1.33140445e+00 6.98641598e-01 3.47002625e-01 4.82645661e-01
-7.28829026e-01 6.78789392e-02 2.84079283e-01 -1.18966974e-01
6.18169904e-01 1.48456842e-01 1.33767039e-01 2.93135405e-01
-1.04146683e+00 8.27782601e-02 5.39213240e-01 4.63374674e-01
6.84793830e-01 -1.04088401e-02 1.27616793e-01 1.00203705e+00
6.73229277e-01 5.91439784e-01 3.88836086e-01 -6.64288819e-01
2.55152166e-01 3.29193354e-01 -5.46277948e-02 -7.59635687e-01
-4.34191614e-01 -4.71879005e-01 -7.63425648e-01 7.83019885e-02
5.95868587e-01 1.21551156e-01 -9.19194043e-01 1.46198869e+00
7.79910207e-01 5.13019562e-01 -3.48282427e-01 1.20554149e+00
6.96731627e-01 6.50120497e-01 -4.05345224e-02 -7.31089056e-01
1.22808766e+00 -3.71668071e-01 -5.50794661e-01 -1.82198465e-01
3.75617296e-01 -1.08528244e+00 6.45882726e-01 3.43489647e-01
-1.18069923e+00 -1.71362698e-01 -1.07020152e+00 2.27162346e-01
2.05263272e-01 2.03877762e-02 3.73014063e-01 6.31932855e-01
-7.66959369e-01 5.43797135e-01 -9.86938119e-01 -2.28775740e-02
3.29210103e-01 4.58247304e-01 -1.74757704e-01 -3.79028469e-01
-6.73787653e-01 7.99458742e-01 2.37368837e-01 5.55004738e-02
-8.24882269e-01 -9.69330609e-01 -5.41705251e-01 -3.78880113e-01
4.73188668e-01 -8.12873900e-01 1.00141358e+00 -6.53596818e-01
-1.38957775e+00 8.16225708e-01 -6.74933270e-02 -7.68717527e-02
6.43124878e-01 1.52050659e-01 -3.28025997e-01 5.57745755e-01
1.52818173e-01 6.89613372e-02 8.91562045e-01 -1.31467772e+00
-8.34059417e-02 -2.45108932e-01 -4.32359666e-01 2.70901859e-01
-6.11949190e-02 -3.61955948e-02 -4.46037620e-01 -5.80162406e-01
7.01754451e-01 -9.85883236e-01 -4.20909286e-01 1.87488735e-01
-3.52451712e-01 4.84849900e-01 3.81200552e-01 -8.78000319e-01
9.99909222e-01 -2.19797659e+00 1.88478380e-01 5.94393134e-01
2.69619375e-01 -1.64166525e-01 2.68373460e-01 4.58493680e-01
-2.23819733e-01 -5.36556661e-01 -8.43515813e-01 -1.75120011e-01
-3.02675009e-01 3.67642939e-02 6.67377412e-02 1.09872949e+00
-4.10408556e-01 3.87186408e-01 -1.04259717e+00 -5.90050697e-01
4.63858694e-01 6.96273506e-01 -4.87990916e-01 3.71799394e-02
-9.38056707e-02 9.05502200e-01 -6.09673440e-01 6.18574142e-01
8.71208131e-01 -3.71501744e-01 3.20381463e-01 -7.33356118e-01
-2.66719133e-01 8.33467543e-02 -1.39379656e+00 1.86562443e+00
-5.56920171e-01 1.33104771e-01 5.84694803e-01 -1.16457665e+00
3.38729680e-01 6.07922137e-01 1.33887100e+00 -7.00700462e-01
1.27564073e-01 8.12600374e-01 -2.23307218e-02 -5.22008002e-01
6.12352006e-02 -8.05468559e-01 3.86259496e-01 5.07433057e-01
-1.12816647e-01 -4.98296738e-01 -3.77870910e-02 1.57093391e-01
1.03474188e+00 -2.61486042e-02 2.22206041e-01 -5.29817581e-01
5.50669253e-01 6.11544065e-02 1.84787095e-01 4.60558861e-01
1.20223306e-01 7.62249470e-01 -1.94921896e-01 -2.12007821e-01
-1.16975379e+00 -1.24315834e+00 -5.88246405e-01 4.02109116e-01
-3.01075522e-02 -2.59144045e-02 -5.98799527e-01 -1.33654252e-01
-7.19064400e-02 3.34207505e-01 -3.43041867e-01 1.56590804e-01
-7.67218530e-01 -1.26110697e+00 -9.31202173e-02 9.19027776e-02
2.06333622e-01 -6.54474854e-01 -7.64442861e-01 5.80390573e-01
-3.58846158e-01 -1.21282792e+00 -3.46339017e-01 2.29330823e-01
-1.04108572e+00 -9.47608590e-01 -7.44892657e-01 -3.52715582e-01
7.44080961e-01 3.10581893e-01 9.59159017e-01 3.69860195e-02
-6.28200710e-01 6.27582788e-01 -1.19512774e-01 -3.15483334e-03
-5.32077551e-01 -5.90009868e-01 9.56132784e-02 2.62512267e-01
-4.21230912e-01 -8.76352072e-01 -9.16426659e-01 2.66940773e-01
-9.87498760e-01 1.67672098e-01 5.09650826e-01 7.87623763e-01
8.40849221e-01 4.66283001e-02 2.47168913e-01 -9.69893932e-01
3.17708462e-01 -6.41517699e-01 -5.12785256e-01 8.16536769e-02
-3.61687750e-01 1.24833547e-01 3.54337543e-01 -2.91095793e-01
-1.12915039e+00 4.10117179e-01 -2.41768479e-01 -2.76452839e-01
1.25711232e-01 6.52457952e-01 9.24466252e-02 -5.55001974e-01
7.32149780e-01 4.09639955e-01 2.50687413e-02 -4.09435332e-01
2.50580022e-03 4.02708769e-01 5.29597521e-01 -7.65706956e-01
6.80931032e-01 9.46430504e-01 3.43771338e-01 -1.21224737e+00
-4.49387550e-01 -1.10941577e+00 -4.99796212e-01 -6.17526293e-01
6.60657227e-01 -7.20278978e-01 -5.49970686e-01 3.23848456e-01
-7.63813853e-01 -2.21606463e-01 -2.94749916e-01 8.69080663e-01
-6.56482697e-01 6.58856213e-01 -5.80394626e-01 -6.41517043e-01
-2.11327106e-01 -1.41720414e+00 9.43638444e-01 -3.23861122e-01
-3.61005869e-03 -1.01264787e+00 2.39160448e-01 5.42048156e-01
4.41024244e-01 4.18254107e-01 8.81792188e-01 -4.03446779e-02
-7.28376508e-01 -1.00998729e-01 -1.43943578e-01 1.74475953e-01
1.46043032e-01 -5.07020772e-01 -7.92876244e-01 -3.71669322e-01
7.01573431e-01 -2.29176786e-02 7.23482132e-01 7.96409070e-01
9.30674434e-01 -1.13602439e-02 -3.45087230e-01 7.22432017e-01
1.88231444e+00 7.10865259e-02 3.36166561e-01 3.20255756e-02
8.26401711e-01 4.58731025e-01 4.86172467e-01 7.64514446e-01
-8.30822960e-02 9.20912623e-01 3.17968011e-01 1.58796944e-02
-3.05778891e-01 2.99769580e-01 7.67365657e-03 1.09180403e+00
-2.99932361e-01 1.48678169e-01 -1.12132907e+00 4.66481775e-01
-1.55171931e+00 -7.47783780e-01 -4.24278975e-01 2.39207482e+00
7.77330041e-01 -3.03476602e-01 1.20949689e-02 1.77239895e-01
4.05484915e-01 -1.41902030e-01 -4.93879318e-01 2.42776111e-01
1.51939631e-01 4.81386572e-01 7.14273870e-01 6.19354784e-01
-7.84958303e-01 1.79773614e-01 5.91423273e+00 8.90127778e-01
-1.23153853e+00 6.38194621e-01 2.75901228e-01 -1.96004361e-01
-4.67432171e-01 -2.23953500e-02 -1.66591510e-01 5.04998744e-01
8.27606976e-01 -2.82811094e-02 5.80802321e-01 1.72507256e-01
3.11463505e-01 -6.17447734e-01 -7.43794203e-01 1.14796817e+00
-1.20910905e-01 -1.19849050e+00 -3.23881507e-01 1.85361773e-01
5.30179203e-01 1.98768467e-01 -1.65847793e-01 -2.58656830e-01
-7.41852298e-02 -6.43857658e-01 7.53927052e-01 5.17884791e-01
7.34633565e-01 -4.32150126e-01 3.69613022e-01 3.02133560e-01
-1.17606008e+00 1.58135429e-01 -1.61151886e-01 3.40818286e-01
4.91475195e-01 1.17870438e+00 -1.05475581e+00 7.37797678e-01
3.81547898e-01 3.79376501e-01 -8.65085870e-02 1.16519237e+00
2.35922307e-01 5.21889925e-01 -6.77676082e-01 4.56629902e-01
8.52887407e-02 -5.33699811e-01 6.91257894e-01 1.18259227e+00
4.93267715e-01 5.52985907e-01 3.15052003e-01 8.11453402e-01
2.14687973e-01 2.92081535e-01 -2.53816277e-01 2.80717850e-01
8.71013254e-02 1.17241335e+00 -1.11295593e+00 -2.37977102e-01
-4.77540493e-01 7.20621765e-01 -2.97148656e-02 3.65399897e-01
-6.81695163e-01 6.50999546e-01 1.53718209e-02 5.98864913e-01
2.28571589e-03 -2.93335110e-01 -1.99935317e-01 -1.02217376e+00
2.29066432e-01 -7.90765524e-01 3.92058849e-01 -6.84351623e-01
-1.08309031e+00 3.07966471e-01 2.71345645e-01 -1.29693913e+00
1.29177766e-02 -5.36518514e-01 -8.54992196e-02 8.17359388e-01
-1.26021314e+00 -9.70992923e-01 -2.82815278e-01 7.24002779e-01
2.86059648e-01 3.21248174e-01 6.67278469e-01 6.41177237e-01
-7.19264299e-02 -1.84713036e-01 3.79888177e-01 -2.68325984e-01
3.87968779e-01 -1.05333686e+00 -2.91082352e-01 5.50117731e-01
5.81124891e-03 2.93058038e-01 1.00507855e+00 -7.07373977e-01
-1.76024139e+00 -5.88928401e-01 2.39079967e-02 8.83918703e-02
7.39625633e-01 -1.75525814e-01 -9.28679764e-01 4.13440585e-01
-5.79084828e-02 3.85657340e-01 6.74578190e-01 -3.96411061e-01
1.03942066e-01 -4.03137393e-02 -1.30668879e+00 9.95789543e-02
7.69227982e-01 -5.44822633e-01 -5.91108948e-02 7.42746711e-01
4.95885871e-02 -7.51272798e-01 -1.03961110e+00 3.43945235e-01
4.63733166e-01 -1.11492205e+00 1.34980655e+00 2.20725149e-01
2.32450277e-01 -2.29530126e-01 -3.33923727e-01 -1.23672211e+00
-2.69330144e-01 -1.65396854e-01 6.65671453e-02 6.81972504e-01
1.71319082e-01 -7.07448661e-01 7.53881872e-01 4.95405853e-01
-4.69031990e-01 -7.35573649e-01 -1.24899709e+00 -6.02543175e-01
-2.21841916e-01 -6.16185308e-01 -5.34397811e-02 1.07230628e+00
4.53945389e-03 -2.63566859e-02 -2.49206454e-01 4.31287646e-01
1.08787417e+00 2.11708859e-01 7.74718001e-02 -9.66439307e-01
-7.82361507e-01 -2.19905347e-01 -1.56620860e-01 -6.69463038e-01
-2.14314446e-01 -9.61194515e-01 1.06445268e-01 -1.43121755e+00
4.56313372e-01 -7.92231441e-01 -1.43960148e-01 9.04261395e-02
1.52488112e-01 3.35379064e-01 -5.02211265e-02 4.54200983e-01
-1.00111468e-02 2.56403595e-01 1.47950578e+00 -1.88950002e-01
-1.23613060e-01 -3.55363153e-02 -3.55925374e-02 6.16902113e-01
6.15786731e-01 -5.82062542e-01 -3.52064997e-01 -2.41437942e-01
2.34154269e-01 6.16489053e-01 6.20381892e-01 -1.13882899e+00
3.74844551e-01 -1.02975197e-01 1.18334487e-01 -4.89777297e-01
7.12364674e-01 -1.08498776e+00 8.14865768e-01 6.09251261e-01
1.19518735e-01 -3.78968835e-01 1.90571725e-01 4.78202581e-01
-1.28053322e-01 -5.82299352e-01 1.00658143e+00 -4.74458694e-01
-3.02309126e-01 2.82012492e-01 -2.53142029e-01 -1.44029468e-01
6.96568131e-01 -2.59692788e-01 2.37329766e-01 -6.87260926e-02
-9.20779228e-01 -2.85779893e-01 4.43197966e-01 -3.91606182e-01
7.91422188e-01 -1.18110609e+00 -8.37358892e-01 8.93383250e-02
-2.03484416e-01 6.90748021e-02 6.61188066e-01 1.48939586e+00
-7.28963017e-01 1.02034613e-01 -5.51688299e-02 -8.56520832e-01
-1.11475623e+00 5.00242054e-01 5.72610795e-01 -3.17405909e-01
-7.30554283e-01 7.00350165e-01 2.15638965e-01 -1.19238220e-01
-3.46430838e-01 -2.81737357e-01 2.58845180e-01 -1.00532286e-02
1.78713933e-01 3.62570167e-01 5.42449892e-01 -9.27664042e-01
-4.59034473e-01 8.03765535e-01 1.87360913e-01 -3.79113108e-01
1.61479712e+00 -1.28248379e-01 -2.24365279e-01 4.81869340e-01
1.29873025e+00 1.62123740e-01 -1.01516604e+00 -4.20964450e-01
-3.11115235e-01 -4.89125699e-01 4.87508148e-01 -5.56160569e-01
-1.19011092e+00 5.17868042e-01 7.27156520e-01 3.56253460e-02
1.25672066e+00 -8.96750763e-02 6.80580258e-01 -2.76252240e-01
5.05411148e-01 -8.82812679e-01 -9.90781281e-03 -1.55070484e-01
9.06176865e-01 -1.10470068e+00 7.46352792e-01 -8.53665411e-01
-8.25695768e-02 1.03281486e+00 -1.61743268e-01 4.14202511e-02
8.73888135e-01 2.62473732e-01 -9.33628827e-02 -5.00811696e-01
-2.62373090e-01 -5.82424216e-02 4.04036134e-01 1.48471862e-01
4.70733315e-01 2.53300488e-01 -2.55462170e-01 -1.65237024e-01
2.75281817e-02 -9.65709835e-02 4.35456753e-01 1.05845344e+00
-3.86409998e-01 -1.05143070e+00 -8.56358349e-01 4.04426038e-01
-4.75674987e-01 -1.14509307e-01 1.65581688e-01 4.34091300e-01
-5.66103905e-02 6.61799371e-01 -5.00797510e-01 1.19317517e-01
1.87917069e-01 -1.75843865e-01 9.31634843e-01 -7.04207540e-01
-3.94523650e-01 5.80816031e-01 -6.59464896e-02 -5.75869143e-01
-7.74509668e-01 -9.87883568e-01 -1.33980191e+00 -8.20796005e-03
-2.87825972e-01 -4.30321917e-02 9.34681177e-01 9.16014075e-01
-2.62215912e-01 6.03171170e-01 5.59448361e-01 -1.02396536e+00
-4.08699751e-01 -6.27261698e-01 -6.98996902e-01 5.67410469e-01
2.88338631e-01 -7.73375511e-01 -2.96155781e-01 7.69240037e-02]
|
[12.991758346557617, -2.65207839012146]
|
b1911603-fe9f-47b5-8a24-69fabe77f8bf
|
end-to-end-supervised-multilabel-contrastive
|
2307.03967
| null |
https://arxiv.org/abs/2307.03967v1
|
https://arxiv.org/pdf/2307.03967v1.pdf
|
End-to-End Supervised Multilabel Contrastive Learning
|
Multilabel representation learning is recognized as a challenging problem that can be associated with either label dependencies between object categories or data-related issues such as the inherent imbalance of positive/negative samples. Recent advances address these challenges from model- and data-centric viewpoints. In model-centric, the label correlation is obtained by an external model designs (e.g., graph CNN) to incorporate an inductive bias for training. However, they fail to design an end-to-end training framework, leading to high computational complexity. On the contrary, in data-centric, the realistic nature of the dataset is considered for improving the classification while ignoring the label dependencies. In this paper, we propose a new end-to-end training framework -- dubbed KMCL (Kernel-based Mutlilabel Contrastive Learning) -- to address the shortcomings of both model- and data-centric designs. The KMCL first transforms the embedded features into a mixture of exponential kernels in Gaussian RKHS. It is then followed by encoding an objective loss that is comprised of (a) reconstruction loss to reconstruct kernel representation, (b) asymmetric classification loss to address the inherent imbalance problem, and (c) contrastive loss to capture label correlation. The KMCL models the uncertainty of the feature encoder while maintaining a low computational footprint. Extensive experiments are conducted on image classification tasks to showcase the consistent improvements of KMCL over the SOTA methods. PyTorch implementation is provided in \url{https://github.com/mahdihosseini/KMCL}.
|
['Mahdi S. Hosseini', 'Konstantinos N. Plataniotis', 'Samir Khaki', 'Ahmad Sajedi']
|
2023-07-08
| null | null | null | null |
['contrastive-learning', 'image-classification', 'contrastive-learning', 'representation-learning']
|
['computer-vision', 'computer-vision', 'methodology', 'methodology']
|
[ 1.86383024e-01 1.01732194e-01 -3.89996439e-01 -6.19789064e-01
-9.88092482e-01 -2.17260748e-01 3.48286361e-01 7.80742392e-02
-3.09020758e-01 6.45083904e-01 -1.38620973e-01 -2.57757008e-01
-1.36418432e-01 -4.95140702e-01 -7.66038358e-01 -9.19312775e-01
2.73497105e-01 2.79479384e-01 -1.42640993e-01 1.74637780e-01
-1.96477417e-02 2.89192498e-01 -1.47241795e+00 2.41849288e-01
8.94920230e-01 1.29051459e+00 -1.32310914e-03 2.81363308e-01
-1.29690943e-02 1.12402427e+00 -1.01932339e-01 -4.25544739e-01
4.20481861e-01 -3.18312883e-01 -5.91191173e-01 1.38675436e-01
3.84705514e-01 -7.99260736e-02 -1.97541997e-01 1.19009173e+00
5.24112999e-01 -7.45850056e-02 7.63366222e-01 -1.57128751e+00
-6.84477508e-01 4.66591924e-01 -9.30603325e-01 -3.31318408e-01
-2.60206550e-01 -2.57325321e-02 1.02620435e+00 -1.13656390e+00
3.15232188e-01 1.12334144e+00 6.53395951e-01 4.97619241e-01
-1.24339736e+00 -7.00752139e-01 1.87502563e-01 2.34322265e-01
-1.50388730e+00 -2.08343700e-01 1.02134931e+00 -6.19133234e-01
3.17510486e-01 2.63914555e-01 2.89294004e-01 9.03099597e-01
-2.00996567e-02 9.27517533e-01 1.37612951e+00 -5.45617402e-01
2.02694371e-01 5.11805177e-01 5.75733244e-01 8.08670223e-01
2.30861783e-01 7.02745393e-02 -1.96931809e-01 -2.59776026e-01
5.40218830e-01 1.57531694e-01 -2.36280426e-01 -8.03738117e-01
-9.39581454e-01 8.69358063e-01 6.97246134e-01 -1.58038475e-02
-2.52532452e-01 1.49956360e-01 5.60772181e-01 2.80322611e-01
6.69275343e-01 1.32241622e-01 -5.60218275e-01 3.95514041e-01
-8.65939856e-01 1.07787825e-01 5.44668913e-01 8.92121613e-01
1.09915543e+00 -2.18222314e-03 -1.56746700e-01 9.30846512e-01
5.23674607e-01 1.97965026e-01 4.04130071e-01 -5.71095526e-01
4.15934741e-01 7.48039544e-01 -2.33684331e-01 -9.51425493e-01
-4.87624258e-01 -7.86859810e-01 -9.91872668e-01 2.86825567e-01
2.63418555e-01 -2.66541373e-02 -9.05580878e-01 2.01131845e+00
5.10380208e-01 3.27914834e-01 -4.46501300e-02 9.86741364e-01
7.49189198e-01 4.10865366e-01 2.19562769e-01 -9.52050611e-02
1.44753683e+00 -1.26373184e+00 -6.63661361e-01 -5.80276400e-02
1.11956394e+00 -5.40214241e-01 1.16713667e+00 2.27509066e-01
-8.50562215e-01 -4.52366054e-01 -1.07726145e+00 -1.81226298e-01
-4.59840566e-01 5.94893634e-01 5.22469163e-01 5.60102999e-01
-9.02487993e-01 2.75343031e-01 -5.76075494e-01 -5.08402362e-02
4.81001019e-01 3.57343584e-01 -4.73918557e-01 -3.07575047e-01
-1.16443729e+00 7.73714423e-01 5.77986717e-01 1.99238509e-01
-5.84431350e-01 -9.23161268e-01 -9.54238415e-01 -7.93722868e-02
3.49581808e-01 -5.49469590e-01 1.01019335e+00 -1.26215565e+00
-1.39108455e+00 8.65956545e-01 1.65664077e-01 -1.71049863e-01
8.10945332e-01 -2.57114023e-02 -1.62251741e-01 -1.00877985e-01
1.11014359e-02 5.19578159e-01 7.39515722e-01 -1.51133883e+00
-4.56090182e-01 -4.53582108e-01 -1.49976006e-02 2.99974024e-01
-4.93342519e-01 -1.62653208e-01 -3.68706375e-01 -6.94249868e-01
6.88041747e-02 -1.09572327e+00 -2.28872836e-01 2.28778988e-01
-4.71543968e-01 -2.65327007e-01 8.43675673e-01 -6.53482795e-01
1.12313342e+00 -2.25830030e+00 -3.57237346e-02 8.88761580e-02
1.76800787e-01 4.19298828e-01 -2.62728304e-01 2.58481294e-01
-4.63280439e-01 5.58024570e-02 -3.42505008e-01 -7.87635684e-01
5.95728029e-03 1.62811816e-01 -1.05697751e-01 7.67026186e-01
3.88422757e-01 7.86145091e-01 -8.55383992e-01 -5.22078454e-01
2.28220776e-01 6.74629271e-01 -3.81765425e-01 2.41271213e-01
-7.72804916e-02 3.18293422e-01 -3.48573089e-01 6.68172657e-01
1.03582764e+00 -4.50977296e-01 1.09401137e-01 -5.30721784e-01
1.99916631e-01 -4.79297303e-02 -1.20470476e+00 1.63586140e+00
-5.21266460e-01 1.47761449e-01 1.28717527e-01 -1.23733127e+00
7.50911832e-01 2.67448574e-01 2.97560662e-01 -4.81833041e-01
2.70522565e-01 3.25783670e-01 -1.90268904e-01 -2.33791515e-01
2.41575181e-01 -1.29423693e-01 1.52257821e-02 2.30084866e-01
1.26033098e-01 3.74783844e-01 -1.01382606e-01 8.44300464e-02
7.63803959e-01 1.70488819e-01 3.23866278e-01 -3.41342241e-01
5.27309060e-01 -1.74491331e-01 7.83023894e-01 4.30710137e-01
-1.59559950e-01 7.45359361e-01 4.80358481e-01 -2.80844629e-01
-7.93816030e-01 -7.89368510e-01 -3.05723310e-01 8.76675427e-01
9.84108597e-02 -2.61481881e-01 -6.30872905e-01 -1.09879947e+00
5.24502993e-02 6.80253386e-01 -7.71448195e-01 -3.96369129e-01
-2.45779067e-01 -1.01274765e+00 4.13208663e-01 4.46540207e-01
3.99400800e-01 -6.52936161e-01 -2.01560259e-01 -9.40397456e-02
-1.44197419e-01 -9.17286158e-01 -4.23227996e-01 5.15139341e-01
-6.30214036e-01 -1.11936355e+00 -5.38022041e-01 -7.42435992e-01
9.67172563e-01 1.62656531e-01 8.11879694e-01 -1.04887873e-01
-2.68664956e-01 2.11903781e-01 -4.18343484e-01 -3.80246878e-01
-9.43650082e-02 1.69297885e-02 -6.39312267e-02 4.21500176e-01
3.03752899e-01 -4.11450684e-01 -7.01779246e-01 2.32333779e-01
-1.12102497e+00 9.39457119e-02 6.99786007e-01 1.14286554e+00
6.45495415e-01 -5.40197380e-02 7.91204631e-01 -1.12889838e+00
3.40634555e-01 -6.87968493e-01 -5.17475367e-01 4.30118054e-01
-9.03846681e-01 -5.96315712e-02 6.73889279e-01 -5.57034731e-01
-8.88786912e-01 1.73508093e-01 4.46193777e-02 -7.83056438e-01
-5.16831838e-02 6.84663713e-01 -4.20113087e-01 -8.03446323e-02
5.05833328e-01 1.39568076e-01 9.37215164e-02 -4.59997237e-01
3.22998554e-01 7.17650950e-01 9.51990038e-02 -6.16253316e-01
6.74912095e-01 3.95559520e-01 1.05268560e-01 -5.12287319e-01
-1.00457823e+00 -5.50644517e-01 -4.44185436e-01 -9.69997421e-02
6.05685711e-01 -1.12966883e+00 -3.21671933e-01 5.62006891e-01
-9.42477226e-01 -2.21405864e-01 -3.65790457e-01 6.59712255e-01
-5.37121236e-01 2.59881824e-01 -6.95487022e-01 -7.94695318e-01
-4.15044188e-01 -1.25544524e+00 8.63082528e-01 7.46603757e-02
2.49453098e-01 -1.09694922e+00 -2.48067342e-02 6.04014635e-01
1.72461942e-01 2.84906417e-01 1.10870802e+00 -8.26680243e-01
-3.75277281e-01 -5.61727107e-01 -6.19546473e-01 8.59364390e-01
9.60010141e-02 -2.37736836e-01 -1.29189909e+00 -5.32837331e-01
-1.25870228e-01 -7.13222742e-01 7.67824531e-01 1.30288944e-01
1.26738381e+00 -2.52665043e-01 -2.08281219e-01 5.48227847e-01
1.68992567e+00 -1.70131221e-01 5.35725117e-01 8.65106285e-02
1.01806438e+00 7.63359129e-01 6.86760068e-01 3.89326960e-01
5.55289865e-01 6.74368382e-01 5.54090321e-01 -5.15098512e-01
-2.00501606e-01 -1.94113210e-01 2.08542109e-01 9.49330389e-01
3.31659675e-01 -2.88590461e-01 -7.81232595e-01 4.06137079e-01
-2.14056015e+00 -4.35901016e-01 -2.77065217e-01 2.26917195e+00
7.90919006e-01 -1.71513617e-01 -6.33328706e-02 1.18790977e-01
7.39955127e-01 3.14144716e-02 -5.32618463e-01 -8.26708898e-02
-7.58745298e-02 -2.06177279e-01 5.54700792e-01 3.14032942e-01
-1.13628221e+00 7.52260208e-01 4.31792021e+00 1.10999012e+00
-1.20070517e+00 2.34975561e-01 7.94770777e-01 1.78137615e-01
-1.77640855e-01 1.20812207e-01 -8.13354671e-01 4.82002765e-01
7.21217692e-01 1.87263891e-01 3.93278003e-02 9.56742585e-01
-6.49740770e-02 7.63189420e-02 -1.10537052e+00 1.05705237e+00
2.08910421e-01 -8.38791549e-01 1.90702733e-02 1.60823300e-01
4.98539954e-01 -4.82261181e-02 2.26361796e-01 4.51763362e-01
1.50947735e-01 -8.03225577e-01 8.57917905e-01 4.46505427e-01
8.71283472e-01 -7.47887790e-01 8.32092106e-01 4.52407092e-01
-1.03840446e+00 5.18588610e-02 -4.05273914e-01 1.55914783e-01
-1.33359879e-01 9.97552693e-01 -7.71326005e-01 8.87649834e-01
4.77049083e-01 6.42343104e-01 -5.27597189e-01 8.69799614e-01
-1.11348167e-01 7.62893796e-01 -1.54118836e-01 3.92306775e-01
2.06385031e-01 -1.85080722e-01 1.81040257e-01 1.30470753e+00
1.57382533e-01 -2.97116548e-01 4.68719661e-01 8.23606908e-01
-2.09007099e-01 2.80566335e-01 -5.35181403e-01 1.68663770e-01
2.58419007e-01 1.65915239e+00 -5.04038870e-01 -1.43895239e-01
-5.63737214e-01 8.63008320e-01 7.13513792e-01 3.33252877e-01
-8.71003330e-01 -2.61189878e-01 3.89926761e-01 1.30311981e-01
7.51004070e-02 1.69935375e-01 -1.85549513e-01 -1.26959431e+00
2.12335721e-01 -6.46288693e-01 3.82315546e-01 -4.16128218e-01
-1.64707351e+00 4.45081949e-01 -1.60277128e-01 -1.41263044e+00
4.59833927e-02 -5.71837902e-01 -3.26962590e-01 9.56533194e-01
-1.96842813e+00 -1.63396609e+00 -2.73542464e-01 5.04998505e-01
2.02379450e-01 2.80334111e-02 6.79323554e-01 7.26445615e-01
-8.12026143e-01 9.80325997e-01 2.26041093e-01 1.81290671e-01
9.18414950e-01 -1.24443626e+00 -3.67536604e-01 4.81191039e-01
-3.03123534e-01 3.35672110e-01 3.79725873e-01 -4.07654136e-01
-1.12613952e+00 -1.54052937e+00 8.52396131e-01 -2.88980812e-01
5.79451382e-01 -5.27188957e-01 -1.04660797e+00 6.57623768e-01
-8.04942027e-02 6.80267930e-01 1.07665491e+00 -8.37863460e-02
-7.28433192e-01 -2.35222384e-01 -1.18694353e+00 3.69078547e-01
7.89210021e-01 -5.03974974e-01 1.47360759e-02 4.32889670e-01
5.75570822e-01 -1.97051451e-01 -8.99616718e-01 6.55343592e-01
4.50310558e-01 -7.05602109e-01 7.31055856e-01 -5.57732999e-01
3.61542612e-01 -4.38718617e-01 -3.32711011e-01 -1.24108493e+00
-4.22580034e-01 1.35477334e-02 5.28333560e-02 1.47847521e+00
4.41013157e-01 -7.78959274e-01 6.87471807e-01 5.71395993e-01
-1.38463080e-01 -1.12611091e+00 -7.91026235e-01 -8.41415107e-01
2.11181045e-01 -1.73577979e-01 2.87051469e-01 1.29694211e+00
-1.45966843e-01 3.40420067e-01 -6.21959209e-01 2.21271351e-01
8.03013206e-01 2.62388005e-03 6.26238227e-01 -1.20547867e+00
-1.60276875e-01 -8.06131884e-02 -4.30384815e-01 -6.78453743e-01
3.82585168e-01 -1.27988839e+00 4.33687121e-03 -1.30170977e+00
3.78167957e-01 -8.44419301e-01 -7.15463519e-01 6.31392419e-01
-2.98855245e-01 1.83465064e-01 2.26528421e-01 2.39831865e-01
-7.45707989e-01 7.31848121e-01 1.09101450e+00 2.93226894e-02
1.19404405e-01 -1.38418645e-01 -6.69132471e-01 6.46835208e-01
9.17543948e-01 -6.34945333e-01 -6.75283492e-01 -2.76335955e-01
1.35677665e-01 3.12439259e-03 5.20479739e-01 -8.68846238e-01
1.77575111e-01 -7.76662901e-02 1.90407142e-01 -4.32612538e-01
3.21473300e-01 -9.92998958e-01 6.98022991e-02 2.97997296e-01
-4.90745485e-01 -1.57506630e-01 -3.62655036e-02 7.28319645e-01
-2.97663242e-01 -1.26201496e-01 1.03966939e+00 8.98699760e-02
-3.73119146e-01 4.80595261e-01 1.21740960e-01 -1.39055131e-02
1.14526784e+00 1.89745594e-02 -3.49979222e-01 -4.11950052e-02
-4.38917547e-01 3.56144011e-01 4.35849100e-01 3.40193152e-01
3.61349821e-01 -1.44825423e+00 -7.75764108e-01 1.72030166e-01
5.15895844e-01 7.43794516e-02 3.88268828e-01 1.09630954e+00
-2.11355761e-01 2.20953137e-01 1.11186557e-01 -4.74588156e-01
-1.14499545e+00 6.76435113e-01 3.41389298e-01 -6.70364261e-01
-2.89401740e-01 8.48468482e-01 5.77522039e-01 -9.49753642e-01
3.54773253e-01 -5.96629409e-03 9.32446215e-03 2.47588262e-01
3.67097974e-01 2.86261231e-01 2.65578687e-01 -5.99633276e-01
-2.97066152e-01 4.18780416e-01 -3.28998625e-01 2.78702945e-01
1.18818152e+00 -9.95116979e-02 -1.05456039e-01 5.40057182e-01
1.63001025e+00 -3.77702504e-01 -1.28223288e+00 -5.47211945e-01
-2.75600031e-02 -2.82236934e-01 2.19083473e-01 -9.15286422e-01
-1.23501325e+00 9.06662822e-01 8.95484328e-01 -1.58007890e-01
1.11509681e+00 -1.96176440e-01 5.95714450e-01 6.10281304e-02
2.99709827e-01 -1.00795841e+00 3.30814272e-02 2.17112243e-01
7.51630008e-01 -1.53479350e+00 -1.34651840e-01 -5.91359675e-01
-6.64174318e-01 9.34689343e-01 7.34640837e-01 -7.27489516e-02
9.29958582e-01 8.00622478e-02 2.28008032e-01 -3.32936309e-02
-7.10203588e-01 -1.16593778e-01 3.02693278e-01 3.43469888e-01
3.52384746e-01 1.66977808e-01 -3.72568786e-01 7.19971478e-01
3.52940828e-01 1.30271539e-01 1.59767359e-01 9.36037838e-01
1.11608580e-01 -1.17950642e+00 -1.46149352e-01 5.48954904e-01
-3.36994439e-01 -7.01156259e-02 -1.26256242e-01 6.56134307e-01
2.98601627e-01 8.77079070e-01 -2.10504428e-01 -4.93391842e-01
2.53004849e-01 1.21821173e-01 2.99605906e-01 -5.47553122e-01
-4.73807663e-01 1.32777495e-02 -1.41497448e-01 -2.80060083e-01
-3.53007287e-01 -5.07788539e-01 -1.18327498e+00 -3.60305533e-02
-7.02516794e-01 1.87140614e-01 7.15296865e-01 5.97686231e-01
4.18606102e-01 5.15592396e-01 7.82032728e-01 -5.70664048e-01
-8.90713871e-01 -9.25217927e-01 -9.26870048e-01 5.65330803e-01
3.24958056e-01 -7.10517704e-01 -5.78993559e-01 -1.65452227e-01]
|
[9.501811981201172, 3.7284326553344727]
|
840b6f65-0bfd-46e8-9d71-0cdb606ed10a
|
multi-task-neural-network-for-non-discrete
|
1708.04828
| null |
http://arxiv.org/abs/1708.04828v1
|
http://arxiv.org/pdf/1708.04828v1.pdf
|
Multi-task Neural Network for Non-discrete Attribute Prediction in Knowledge Graphs
|
Many popular knowledge graphs such as Freebase, YAGO or DBPedia maintain a
list of non-discrete attributes for each entity. Intuitively, these attributes
such as height, price or population count are able to richly characterize
entities in knowledge graphs. This additional source of information may help to
alleviate the inherent sparsity and incompleteness problem that are prevalent
in knowledge graphs. Unfortunately, many state-of-the-art relational learning
models ignore this information due to the challenging nature of dealing with
non-discrete data types in the inherently binary-natured knowledge graphs. In
this paper, we propose a novel multi-task neural network approach for both
encoding and prediction of non-discrete attribute information in a relational
setting. Specifically, we train a neural network for triplet prediction along
with a separate network for attribute value regression. Via multi-task
learning, we are able to learn representations of entities, relations and
attributes that encode information about both tasks. Moreover, such attributes
are not only central to many predictive tasks as an information source but also
as a prediction target. Therefore, models that are able to encode, incorporate
and predict such information in a relational learning context are highly
attractive as well. We show that our approach outperforms many state-of-the-art
methods for the tasks of relational triplet classification and attribute value
prediction.
|
['Luu Anh Tuan', 'Yi Tay', 'Siu Cheung Hui', 'Minh C. Phan']
|
2017-08-16
| null | null | null | null |
['value-prediction']
|
['computer-code']
|
[-7.43869841e-02 3.73794317e-01 -8.57799649e-01 -5.99227250e-01
-5.99028945e-01 -4.92523313e-01 6.33070052e-01 9.69849885e-01
-1.18340217e-01 1.04331791e+00 1.32548928e-01 -1.59653768e-01
-6.54861867e-01 -1.59547102e+00 -1.09551167e+00 -4.51307118e-01
-2.34221682e-01 9.92223799e-01 1.31351640e-02 -3.88619900e-01
-2.72891670e-01 3.87548059e-01 -1.64213800e+00 5.14286458e-01
9.31613564e-01 1.40335894e+00 -3.03923815e-01 -9.67317149e-02
-4.40257698e-01 1.18530369e+00 -2.90846378e-01 -8.71263862e-01
1.07693058e-02 2.64753371e-01 -9.91989553e-01 -3.37692261e-01
4.06983465e-01 4.98103872e-02 -3.50148648e-01 8.04853678e-01
1.59385741e-01 2.90038548e-02 8.90131950e-01 -1.48828101e+00
-8.46232831e-01 1.06729674e+00 -3.80968839e-01 -1.41002700e-01
4.34284717e-01 -6.23112142e-01 1.65484321e+00 -4.95323896e-01
6.73623025e-01 1.00464475e+00 6.68638587e-01 -7.29671344e-02
-1.41051638e+00 -4.50504631e-01 7.08087310e-02 4.98544663e-01
-1.43053019e+00 -2.71466970e-01 9.18979824e-01 -4.71033633e-01
8.05469513e-01 2.19165608e-01 6.03718281e-01 1.11390984e+00
-1.94448128e-01 6.82670474e-01 8.12097371e-01 -2.63560772e-01
-1.36815816e-01 3.37037057e-01 1.78100526e-01 6.96956635e-01
4.60859299e-01 -5.63250333e-02 -5.86570263e-01 -1.79700822e-01
5.90482056e-01 1.23210832e-01 5.00844717e-02 -7.09372580e-01
-1.33394349e+00 7.21606672e-01 8.50429475e-01 2.08605751e-01
-5.13843834e-01 3.10347408e-01 3.22835028e-01 4.80089366e-01
6.05845869e-01 4.54796612e-01 -7.91627109e-01 1.00852206e-01
-4.07771558e-01 1.97848588e-01 1.13665521e+00 1.09899139e+00
1.15215945e+00 -2.08654657e-01 -2.52761722e-01 8.57935250e-01
2.29509547e-01 2.16031522e-01 8.34371969e-02 -5.78227162e-01
8.43517900e-01 9.96998608e-01 -5.84563613e-02 -1.10109627e+00
-5.18497169e-01 -4.24637944e-01 -9.43057239e-01 -3.82440686e-01
4.40383285e-01 8.14063922e-02 -7.54173398e-01 1.84300303e+00
5.14076233e-01 -1.06078148e-01 2.14754462e-01 5.09400129e-01
1.20266485e+00 3.12834024e-01 2.91313440e-01 -1.91344861e-02
1.45862830e+00 -6.08341932e-01 -6.74743295e-01 -3.92678156e-02
9.82402742e-01 -1.67987719e-01 6.01470351e-01 -1.14600416e-02
-7.09510505e-01 -4.14329916e-01 -8.15162659e-01 -3.03989053e-01
-1.06657195e+00 6.43758327e-02 1.14016819e+00 4.81880158e-01
-5.90169549e-01 5.30710220e-01 -5.01034141e-01 -1.46207809e-01
6.27980947e-01 4.94407922e-01 -7.65158355e-01 -2.11786538e-01
-1.63994932e+00 1.16543984e+00 7.23998666e-01 -1.87583715e-01
-1.32867888e-01 -8.35966170e-01 -1.19794536e+00 4.07164991e-01
9.07600284e-01 -8.82013440e-01 5.92527211e-01 -4.75805938e-01
-7.57517040e-01 9.36162114e-01 1.64040729e-01 -5.94276726e-01
2.50429869e-01 6.02535345e-02 -5.03837109e-01 -1.65918633e-01
1.10275730e-01 2.45906353e-01 3.73943537e-01 -1.30003369e+00
-5.24879277e-01 -6.76430762e-01 4.20336127e-01 8.16816017e-02
-5.05032599e-01 -4.16594505e-01 -2.32138410e-01 -4.14300948e-01
-2.26666145e-02 -5.98373294e-01 8.58981162e-02 -1.91298768e-01
-8.24528575e-01 -5.47563791e-01 4.21309680e-01 -2.98649669e-01
1.09089041e+00 -1.76891041e+00 3.40882927e-01 4.91320640e-01
5.35545349e-01 -3.54823396e-02 2.08830740e-02 5.72963119e-01
-6.03225194e-02 1.45012975e-01 1.33745641e-01 -1.14647701e-01
1.69257492e-01 4.33009565e-01 -1.75852757e-02 1.55975252e-01
3.18811655e-01 1.26124513e+00 -8.55581522e-01 -6.29793942e-01
-2.81214658e-02 5.49228668e-01 -2.47968256e-01 1.78595614e-02
-6.00235283e-01 8.57168585e-02 -8.08818817e-01 9.47767615e-01
4.22804505e-01 -6.09829664e-01 5.11137068e-01 -6.20149434e-01
2.63828188e-01 3.95736307e-01 -1.11301565e+00 1.44692791e+00
-5.30157268e-01 1.10292368e-01 -3.02418053e-01 -1.55969727e+00
1.06267571e+00 2.60192841e-01 8.63838434e-01 -6.68815017e-01
-1.19648851e-03 1.88885435e-01 -9.04618874e-02 -4.18899983e-01
5.25574982e-01 -1.61904112e-01 -3.41562897e-01 1.77582309e-01
1.56288832e-01 9.79642868e-02 3.69853437e-01 2.07500905e-01
9.88877892e-01 6.15734793e-02 4.73127812e-01 6.03991263e-02
5.03797650e-01 -1.20133042e-01 6.85673475e-01 5.02587974e-01
4.06818360e-01 2.06494004e-01 8.42067122e-01 -6.28901362e-01
-9.30296898e-01 -9.25700963e-01 -3.38605374e-01 1.20630908e+00
1.38075752e-02 -5.00566423e-01 1.43534720e-01 -7.86460042e-01
8.07600856e-01 4.29062754e-01 -7.89445043e-01 -2.97827303e-01
-3.09276730e-01 -7.00044513e-01 4.46817130e-01 6.57103419e-01
9.64352712e-02 -8.38868141e-01 8.09543282e-02 3.00717801e-01
-2.55218685e-01 -1.55386508e+00 3.77714247e-01 6.00380123e-01
-5.69414258e-01 -1.28381026e+00 -1.05731733e-01 -6.24735773e-01
3.75916332e-01 -2.62028426e-01 1.59486139e+00 1.91133976e-01
4.58304547e-02 1.80447280e-01 -4.90028113e-01 -3.53012472e-01
-2.42430627e-01 4.26608503e-01 -1.76336065e-01 2.68962860e-01
5.89537263e-01 -8.04173410e-01 3.24647129e-02 4.45909537e-02
-8.28077674e-01 2.45011784e-02 7.50886381e-01 8.89626265e-01
8.32073212e-01 6.65252134e-02 8.87319803e-01 -1.52946353e+00
4.20674890e-01 -8.98776472e-01 -4.97462630e-01 7.66772687e-01
-7.17673063e-01 4.59634721e-01 7.07533598e-01 -1.88456699e-01
-8.58514369e-01 7.82311410e-02 1.48755595e-01 -1.44267261e-01
-5.60441166e-02 1.25232756e+00 -4.26084429e-01 4.11682064e-03
4.24864292e-01 2.66842674e-02 -2.14586079e-01 -4.49527204e-01
5.34439981e-01 3.82752538e-01 2.95943230e-01 -1.03124428e+00
8.48404109e-01 1.88204780e-01 6.69545829e-01 -4.28429991e-01
-1.28056312e+00 -3.47407192e-01 -9.88656521e-01 1.68538958e-01
6.38008833e-01 -1.16577971e+00 -8.80533993e-01 1.44703895e-01
-9.38245237e-01 2.07585413e-02 -3.41035903e-01 3.31901461e-01
-6.22591436e-01 -1.33021735e-02 -3.72007489e-01 -6.18881047e-01
-8.62416029e-02 -8.23295355e-01 8.90604019e-01 -5.87610416e-02
1.47602975e-01 -1.31206620e+00 -1.76585957e-01 5.92085004e-01
3.49614203e-01 6.24420106e-01 1.54502773e+00 -1.23925829e+00
-9.11528826e-01 -2.30622873e-01 -4.59103614e-01 -1.97244093e-01
2.77473062e-01 -2.56732553e-01 -5.97558618e-01 1.17390431e-01
-7.38618016e-01 -7.39890695e-01 1.06900799e+00 1.67450145e-01
1.23269439e+00 -3.99426192e-01 -6.28737330e-01 6.88966334e-01
1.60782313e+00 -3.08073789e-01 3.88110012e-01 3.56043935e-01
1.09773982e+00 7.94455767e-01 6.53202295e-01 4.46312845e-01
1.21946359e+00 9.90546644e-01 5.81607699e-01 -5.94826713e-02
1.20662540e-01 -3.79168659e-01 -4.15584058e-01 4.10017252e-01
-3.65515739e-01 -2.64135480e-01 -1.08864820e+00 6.01931155e-01
-2.08040047e+00 -9.96826172e-01 -3.02022040e-01 2.13395786e+00
1.35915697e+00 -4.89147156e-02 1.76343679e-01 1.30774140e-01
4.76873070e-01 2.19772130e-01 -4.32835340e-01 -5.29991686e-02
-3.81895751e-01 3.16211954e-02 5.50131321e-01 5.08490065e-03
-1.42305815e+00 8.67136478e-01 5.02641439e+00 6.40743613e-01
-5.53553700e-01 -1.63380668e-01 5.35504580e-01 1.15118295e-01
-6.49409831e-01 -9.79839042e-02 -8.57948363e-01 2.36187384e-01
9.05215561e-01 -3.19226682e-01 3.36004615e-01 7.19809890e-01
-5.74813843e-01 1.25992805e-01 -1.52681971e+00 9.21202779e-01
-1.44688711e-01 -1.45893717e+00 1.82328507e-01 2.59795964e-01
6.08716965e-01 -5.38174883e-02 -8.13078042e-03 5.17650306e-01
4.81889784e-01 -1.22641206e+00 4.09477770e-01 7.55268157e-01
8.20259154e-01 -7.38565922e-01 8.16957176e-01 1.96389630e-01
-1.39276147e+00 -2.18341291e-01 -3.87545556e-01 1.36240929e-01
-1.62738692e-02 9.33633745e-01 -4.80622202e-01 1.17011404e+00
5.89924514e-01 1.07684028e+00 -6.95130527e-01 8.62328410e-01
-2.43283674e-01 -2.55831480e-02 -3.24247718e-01 1.29173964e-01
-1.45048443e-02 -1.86888620e-01 4.12651561e-02 8.34851503e-01
2.74582058e-01 3.45057957e-02 2.59616196e-01 8.36195827e-01
-6.84165955e-01 2.16033474e-01 -9.35912371e-01 -3.26039493e-01
6.91031098e-01 1.14852738e+00 -2.87533909e-01 -3.58690411e-01
-7.81196594e-01 1.90119520e-01 9.72222149e-01 2.51820117e-01
-6.44320488e-01 -2.61099130e-01 6.43420339e-01 6.28768727e-02
3.27276021e-01 7.55157787e-04 -4.15551871e-01 -1.27254546e+00
1.48098484e-01 -6.14707112e-01 6.68667376e-01 -3.63296837e-01
-1.63408267e+00 3.96466792e-01 2.67650019e-02 -1.03235734e+00
-4.87332433e-01 -5.71387410e-01 -5.54431789e-02 6.92717075e-01
-2.03376102e+00 -1.75234091e+00 -3.54773730e-01 8.01786244e-01
-3.59059691e-01 -3.09798986e-01 9.59207594e-01 5.25430799e-01
-4.29007292e-01 6.72528267e-01 1.69288322e-01 5.71219921e-01
7.30892718e-01 -1.63503206e+00 6.71726326e-03 1.35645866e-01
4.55809206e-01 5.00107348e-01 4.31465298e-01 -5.50955653e-01
-1.68700039e+00 -9.18366075e-01 1.13685131e+00 -6.38481617e-01
8.79121721e-01 -3.79714340e-01 -1.11488521e+00 1.00044322e+00
-3.75048846e-01 5.17805696e-01 9.56242621e-01 9.60990906e-01
-7.66680241e-01 -4.70015556e-01 -9.34974492e-01 8.77457764e-03
1.03575134e+00 -6.23551667e-01 -5.78492582e-01 5.17330527e-01
7.51043260e-01 -2.92422801e-01 -1.70976341e+00 5.85477054e-01
5.71595848e-01 -7.48666704e-01 1.39493358e+00 -9.41931665e-01
7.38465309e-01 1.01624228e-01 -3.01090479e-01 -1.32756817e+00
-1.51565105e-01 -6.68954104e-03 -7.70503640e-01 1.43771648e+00
8.36762607e-01 -6.27917945e-01 8.77675235e-01 8.53484511e-01
1.31896690e-01 -8.89274657e-01 -8.55190396e-01 -7.64970899e-01
1.58480089e-02 -5.09163961e-02 8.89839768e-01 1.30488408e+00
1.39119491e-01 6.97066963e-01 -5.00064611e-01 1.17978528e-02
7.09881604e-01 6.83937430e-01 5.51824450e-01 -2.03176618e+00
-4.53643091e-02 -2.55218029e-01 -6.77498937e-01 -5.48437119e-01
5.01319170e-01 -1.15761626e+00 -6.23573363e-01 -1.88725388e+00
2.58269340e-01 -1.13081777e+00 -4.49607283e-01 9.34320629e-01
-7.50481784e-02 -4.08653654e-02 1.92515403e-02 6.17197938e-02
-6.27293348e-01 7.83268332e-01 1.08192265e+00 -3.34414512e-01
7.32044689e-03 1.02957964e-01 -7.96267927e-01 3.34973305e-01
5.08011162e-01 -6.22865975e-01 -2.85686612e-01 -4.07775879e-01
8.65816355e-01 4.35505927e-01 3.47821027e-01 -6.21376395e-01
4.41390187e-01 -2.74902165e-01 4.67378914e-01 -5.79277277e-01
7.49829412e-01 -1.19349909e+00 8.59204158e-02 -3.27113196e-02
-3.03046674e-01 -1.67182326e-01 -2.05791697e-01 7.59811103e-01
-6.58854902e-01 5.74718714e-02 2.00240195e-01 -1.04635835e-01
-6.97035968e-01 6.10121667e-01 4.55454826e-01 1.28469348e-01
1.05690730e+00 8.41304660e-03 -6.69995189e-01 -1.81346938e-01
-8.89636338e-01 4.95277375e-01 1.79847255e-01 4.31383014e-01
4.52070117e-01 -1.60357249e+00 -6.95669174e-01 1.52549103e-01
5.68496168e-01 1.96168438e-01 -6.48208443e-05 7.18673587e-01
5.27927559e-03 4.52583194e-01 -3.79343599e-01 -2.54493058e-01
-1.00336576e+00 8.54807496e-01 8.78984332e-02 -5.89766920e-01
-2.92552948e-01 5.72812498e-01 -2.50850201e-01 -6.36563957e-01
1.92951426e-01 -1.08675830e-01 -6.51588202e-01 7.30459511e-01
-1.12974837e-01 4.43453901e-02 1.60189226e-01 -6.55030251e-01
-2.91845322e-01 3.91721845e-01 -4.51636165e-02 5.69409072e-01
1.80739653e+00 6.07555956e-02 -3.70801330e-01 5.16631663e-01
1.08873332e+00 -1.06949799e-01 -6.52353704e-01 -7.79225647e-01
4.13491249e-01 -4.04256016e-01 -1.52152464e-01 -8.92295063e-01
-1.22646022e+00 7.06905961e-01 -2.82404214e-01 5.71919739e-01
9.08345222e-01 4.35706884e-01 5.09847343e-01 7.73102522e-01
6.17930710e-01 -8.82117093e-01 -1.45154074e-01 4.91870999e-01
6.67486727e-01 -1.54386115e+00 1.64959088e-01 -7.57057846e-01
-6.40562296e-01 1.03871417e+00 7.07625151e-01 3.47090006e-01
8.45957935e-01 7.88612291e-02 -2.28095666e-01 -4.13193554e-01
-1.03380239e+00 -6.25827909e-01 6.19742393e-01 8.19486082e-01
5.74145675e-01 1.62954405e-01 -2.00751168e-03 6.82746887e-01
-2.14640796e-01 -2.05599502e-01 3.83373171e-01 6.86958671e-01
-1.28795266e-01 -1.30389082e+00 -3.53105962e-02 9.74297464e-01
-4.17394429e-01 -1.60058379e-01 -4.09489840e-01 8.16251218e-01
1.30024701e-01 7.03346789e-01 -2.78113782e-02 -4.03422832e-01
3.57671231e-01 1.37421116e-01 4.74890232e-01 -7.50843704e-01
-4.63133305e-01 -6.03912830e-01 4.92364675e-01 -4.14085001e-01
-5.93277037e-01 -3.92395407e-01 -1.04856646e+00 -3.22711617e-01
-2.15641931e-01 8.67333859e-02 5.14676988e-01 1.07070518e+00
2.66669393e-01 6.57273531e-01 3.35732639e-01 -2.24321917e-01
-5.90465486e-01 -8.06604743e-01 -8.85407567e-01 7.01402128e-01
2.60304570e-01 -1.06016004e+00 6.52176812e-02 -2.42932320e-01]
|
[8.762755393981934, 7.872171401977539]
|
884473d3-6dfb-4205-9f7f-e572d6ed369b
|
stochastic-planner-actor-critic-for
|
2112.07415
| null |
https://arxiv.org/abs/2112.07415v2
|
https://arxiv.org/pdf/2112.07415v2.pdf
|
Stochastic Planner-Actor-Critic for Unsupervised Deformable Image Registration
|
Large deformations of organs, caused by diverse shapes and nonlinear shape changes, pose a significant challenge for medical image registration. Traditional registration methods need to iteratively optimize an objective function via a specific deformation model along with meticulous parameter tuning, but which have limited capabilities in registering images with large deformations. While deep learning-based methods can learn the complex mapping from input images to their respective deformation field, it is regression-based and is prone to be stuck at local minima, particularly when large deformations are involved. To this end, we present Stochastic Planner-Actor-Critic (SPAC), a novel reinforcement learning-based framework that performs step-wise registration. The key notion is warping a moving image successively by each time step to finally align to a fixed image. Considering that it is challenging to handle high dimensional continuous action and state spaces in the conventional reinforcement learning (RL) framework, we introduce a new concept `Plan' to the standard Actor-Critic model, which is of low dimension and can facilitate the actor to generate a tractable high dimensional action. The entire framework is based on unsupervised training and operates in an end-to-end manner. We evaluate our method on several 2D and 3D medical image datasets, some of which contain large deformations. Our empirical results highlight that our work achieves consistent, significant gains and outperforms state-of-the-art methods.
|
['Siwei Lyu', 'Xi Wu', 'Qi Song', 'Youbing Yin', 'Bin Kong', 'Shu Hu', 'Xin Wang', 'Jing Hu', 'Ziwei Luo']
|
2021-12-14
| null | null | null | null |
['deformable-medical-image-registration']
|
['medical']
|
[ 2.77476668e-01 3.13620836e-01 -2.90161341e-01 -1.65243015e-01
-1.20671380e+00 -3.11356366e-01 5.25064886e-01 7.24081025e-02
-4.99262631e-01 5.06249368e-01 2.98769444e-01 2.78411098e-02
-2.98964471e-01 -4.81843531e-01 -7.16496408e-01 -1.01983356e+00
-3.30876261e-01 8.10129642e-01 -6.25389069e-02 -4.38623101e-01
-1.14602111e-02 4.40589726e-01 -5.24910569e-01 -2.41638899e-01
8.28182757e-01 5.66807508e-01 1.07936636e-01 5.66372633e-01
2.62627184e-01 4.52563018e-01 -2.88555861e-01 -1.62761778e-01
4.38664734e-01 -4.79506940e-01 -9.74549532e-01 3.19348276e-01
2.99655855e-01 -3.43773037e-01 -2.80237406e-01 8.99474204e-01
9.99293387e-01 2.14987159e-01 4.66414422e-01 -8.76439273e-01
-5.57884157e-01 4.27478373e-01 -5.25346994e-01 5.04665412e-02
3.50467503e-01 3.90213549e-01 6.13516808e-01 -7.06199646e-01
6.87523603e-01 1.22695863e+00 7.49303877e-01 8.49281192e-01
-1.45817792e+00 -1.80146769e-01 1.02113195e-01 -4.08437073e-01
-9.93097067e-01 -2.62161970e-01 9.60710526e-01 -4.25685525e-01
6.65664613e-01 2.23929109e-03 6.85660124e-01 1.00105083e+00
5.82453310e-01 6.77707791e-01 1.02567029e+00 -1.86836466e-01
1.76093608e-01 -6.80864990e-01 -5.57703257e-01 8.22654545e-01
-1.82129875e-01 3.69095683e-01 -3.43683623e-02 -2.88116783e-01
1.07297707e+00 3.08710247e-01 -2.87235975e-01 -6.75804853e-01
-1.73481929e+00 8.90807033e-01 7.84139037e-01 3.07156682e-01
-6.90646529e-01 5.29677153e-01 2.22653985e-01 1.72058716e-01
3.57752889e-01 5.44765592e-01 -4.39530700e-01 7.31844902e-02
-6.45983815e-01 3.96133244e-01 5.84139109e-01 5.24488211e-01
6.30867779e-01 9.23750997e-02 -3.14551443e-01 6.56169355e-01
4.60418016e-01 3.82326722e-01 7.05645502e-01 -9.79144394e-01
4.88686562e-01 4.51808363e-01 -4.84190602e-03 -8.42636585e-01
-7.28634596e-01 -4.33353752e-01 -1.22576487e+00 2.87750572e-01
4.19043481e-01 -3.81603032e-01 -9.18853998e-01 1.80834544e+00
7.50321686e-01 3.53981972e-01 2.59563088e-01 1.04523158e+00
6.13396823e-01 4.12406832e-01 1.53255165e-01 -3.76287371e-01
1.11403036e+00 -1.04177761e+00 -7.65265822e-01 -2.87967324e-01
5.10486066e-01 -5.60765982e-01 9.43747401e-01 2.03044415e-02
-1.43022883e+00 -4.17514116e-01 -6.70216799e-01 2.69027501e-01
1.56518877e-01 -2.35993981e-01 4.71806824e-01 2.31293797e-01
-1.11336589e+00 9.00619447e-01 -1.40692425e+00 -2.41198693e-03
5.58131695e-01 6.98391020e-01 -3.61970931e-01 1.42166182e-01
-1.06733394e+00 9.90859687e-01 1.83562964e-01 4.09905255e-01
-1.06995463e+00 -8.41510475e-01 -1.00148487e+00 -3.37610006e-01
4.88303065e-01 -8.72560501e-01 1.22488785e+00 -7.68687367e-01
-2.00398731e+00 8.05361688e-01 2.94719934e-01 -3.13311636e-01
8.80540013e-01 -1.58174932e-01 -6.39102310e-02 -1.78995207e-02
-1.52987570e-01 6.92941785e-01 8.01855206e-01 -1.15959895e+00
9.08837318e-02 -3.40545058e-01 -5.78787364e-02 3.44531566e-01
-1.16588801e-01 2.78681573e-02 -4.05899376e-01 -9.58106399e-01
1.68403625e-01 -1.23605406e+00 -9.55460668e-01 1.54162720e-01
-3.11317593e-01 -9.93873402e-02 3.82137656e-01 -4.21796024e-01
9.36505258e-01 -1.81594193e+00 5.76275408e-01 1.38781667e-01
2.26901829e-01 2.93347985e-01 -2.84666866e-01 4.40834314e-01
-7.34111890e-02 -5.39911762e-02 -5.11617005e-01 -2.76481420e-01
-1.46749482e-01 3.93616259e-01 4.71221469e-02 8.58537078e-01
2.76598811e-01 1.29656255e+00 -1.33739221e+00 -6.51489913e-01
2.23836169e-01 6.32196069e-01 -7.48431385e-01 4.65129495e-01
-1.76471367e-01 1.11432242e+00 -8.47212911e-01 5.38177133e-01
4.78752881e-01 -3.23006809e-01 3.43206465e-01 -2.87936568e-01
2.29975432e-01 -2.41043806e-01 -1.20307589e+00 2.12603498e+00
-3.58867586e-01 9.41252150e-03 -8.26307107e-03 -1.28460741e+00
1.03501368e+00 5.70890009e-01 1.13126361e+00 -4.64585841e-01
1.74187154e-01 3.04490775e-01 -1.75911151e-02 -7.30175674e-01
7.46526867e-02 -4.37071919e-01 -1.82368413e-01 4.16925102e-01
3.32615426e-04 -2.97781110e-01 -2.04701200e-01 -3.30635607e-01
1.04927528e+00 4.15739924e-01 3.45765859e-01 -1.99108541e-01
6.83344007e-01 -1.34947449e-01 8.65326464e-01 3.95321310e-01
-3.60851824e-01 6.21583223e-01 3.88154775e-01 -8.23220909e-01
-8.54380369e-01 -1.04995191e+00 -3.22401077e-02 6.20060086e-01
1.96967989e-01 6.15868121e-02 -7.17926681e-01 -7.84888089e-01
-1.02621496e-01 -3.57269719e-02 -6.36252701e-01 -2.10430816e-01
-1.24630904e+00 -1.01546240e+00 4.23124462e-01 5.34583449e-01
3.54955792e-01 -1.25999081e+00 -7.60472476e-01 6.24409258e-01
-7.48784617e-02 -9.53504860e-01 -8.70807111e-01 -4.47938628e-02
-1.29214537e+00 -1.03895271e+00 -1.03701746e+00 -9.99609351e-01
1.16005206e+00 -2.54833907e-01 1.12551951e+00 1.17084242e-01
-2.91004717e-01 3.21838439e-01 -1.96674079e-01 -1.94696873e-01
-7.07131445e-01 9.83146057e-02 -3.43798078e-03 -3.57474983e-02
-4.16397482e-01 -4.33583349e-01 -9.50751245e-01 4.59207475e-01
-1.00291836e+00 -1.57300070e-01 6.91816151e-01 1.30003822e+00
9.31163907e-01 -3.99945021e-01 5.66454411e-01 -9.61396933e-01
6.76177263e-01 -2.18370736e-01 -5.57771981e-01 2.69073159e-01
-4.58131433e-01 3.75470102e-01 6.63211286e-01 -7.30932951e-01
-7.35255957e-01 5.51831007e-01 -2.40226552e-01 -6.01030171e-01
1.35357484e-01 5.03711581e-01 1.55416667e-01 -3.43467861e-01
8.22997153e-01 5.13871536e-02 4.93246466e-01 -2.47200042e-01
2.93602288e-01 9.20700431e-02 6.75203323e-01 -6.24534965e-01
9.98312414e-01 5.35391331e-01 2.25602746e-01 -1.81252062e-01
-6.29223764e-01 -2.37899840e-01 -9.14071560e-01 -3.26805264e-01
8.00553322e-01 -7.17851222e-01 -8.48514020e-01 4.86803502e-01
-9.51161623e-01 -7.82278538e-01 -4.94055033e-01 6.60359740e-01
-9.49699223e-01 4.00747716e-01 -5.62563419e-01 -2.75571615e-01
-6.18622482e-01 -1.53087711e+00 1.23550189e+00 1.31100148e-01
-2.93526947e-02 -1.34536493e+00 6.67197466e-01 1.69483259e-01
5.73400199e-01 1.07863224e+00 6.14285111e-01 -4.37906712e-01
-3.87845546e-01 -2.11777285e-01 4.85031009e-01 -1.55261997e-02
4.07476634e-01 -3.42076927e-01 -4.84382808e-01 -7.50132978e-01
9.25892293e-02 -5.56967854e-01 4.91969585e-01 5.58941543e-01
1.14475167e+00 -4.63972360e-01 -2.51430869e-01 8.16948235e-01
1.47868443e+00 1.08593432e-02 5.30835509e-01 2.87320197e-01
7.27859855e-01 4.44848061e-01 6.75575197e-01 2.60657012e-01
3.94060373e-01 9.01341856e-01 6.49109006e-01 -5.57252169e-01
-1.36884050e-02 -4.88266945e-02 3.13570112e-01 8.61359656e-01
-3.27931613e-01 2.20084235e-01 -1.01715684e+00 3.93046737e-01
-2.04131556e+00 -7.22407281e-01 3.88586104e-01 2.15931273e+00
1.25983262e+00 -6.92472234e-02 1.42030209e-01 -2.08223596e-01
5.87221980e-01 3.54516804e-01 -7.04254031e-01 -8.28933045e-02
3.73823255e-01 1.34991139e-01 3.77716690e-01 5.91861129e-01
-1.16133833e+00 9.53385234e-01 6.45500422e+00 4.54452515e-01
-1.38040006e+00 -4.14819419e-02 6.11937404e-01 1.70078650e-01
-1.68692932e-01 -2.28839815e-01 -3.23557287e-01 2.16209501e-01
7.15062737e-01 1.02442555e-01 5.04526854e-01 5.89556932e-01
4.41680640e-01 3.87283117e-01 -1.17358506e+00 8.01462114e-01
-1.30300999e-01 -1.32068622e+00 -2.10726157e-01 -1.02110663e-02
8.37211311e-01 8.65706429e-03 4.15844098e-03 1.03092954e-01
5.08770883e-01 -1.12070131e+00 2.96126932e-01 5.17800570e-01
7.94137478e-01 -6.55134022e-01 5.51290035e-01 4.53253508e-01
-1.03233969e+00 9.80410054e-02 -3.24051201e-01 4.76988107e-01
4.04579759e-01 2.10614726e-01 -7.90674448e-01 4.81757551e-01
3.00623983e-01 8.12946200e-01 -1.66700691e-01 9.36987817e-01
-1.47009239e-01 1.26926273e-01 -4.41016592e-02 1.28043026e-01
2.57724404e-01 -2.22549170e-01 5.38450718e-01 9.36309397e-01
1.74570367e-01 1.16966248e-01 6.77727461e-01 6.82006240e-01
-1.31797642e-01 -1.89555492e-02 -5.51146150e-01 2.46796265e-01
8.00535679e-02 1.33349216e+00 -8.08717966e-01 -2.10370235e-02
-2.35427797e-01 8.10947835e-01 3.69904697e-01 2.49719888e-01
-8.56645703e-01 -6.50658309e-02 3.99799109e-01 -9.18346941e-02
1.64578438e-01 -3.23910981e-01 1.66349486e-01 -1.09577763e+00
1.81535780e-02 -1.03918922e+00 4.21969980e-01 -2.37398282e-01
-1.28597617e+00 6.97961569e-01 -4.08343635e-02 -1.53764236e+00
-5.13634443e-01 -1.87761709e-01 -5.91983974e-01 6.52635694e-01
-1.66612995e+00 -1.29828131e+00 -3.36632937e-01 9.59733784e-01
4.74374503e-01 -1.88198835e-02 9.76080060e-01 2.14611590e-01
-3.99927735e-01 6.37799561e-01 8.13557953e-02 4.31153774e-01
7.81288147e-01 -1.38388574e+00 3.51048499e-01 4.56417322e-01
6.62050247e-02 2.65510947e-01 5.39933324e-01 -6.23282075e-01
-1.80210733e+00 -1.19358671e+00 6.70176268e-01 -3.19660544e-01
5.12835681e-01 -6.14621341e-02 -7.69745648e-01 6.70650899e-01
1.54459178e-01 7.89681137e-01 4.03638661e-01 -3.58755231e-01
2.68549025e-01 -1.97996393e-01 -1.22678542e+00 6.36531770e-01
8.96911323e-01 -1.71054244e-01 -5.49432874e-01 6.16423905e-01
5.58352888e-01 -1.17553186e+00 -1.41502321e+00 5.47360003e-01
2.94508725e-01 -4.35388654e-01 1.30936015e+00 -8.61490488e-01
3.42900723e-01 -3.03657860e-01 4.04899508e-01 -1.66453695e+00
-3.47394049e-01 -1.20371735e+00 -1.36554435e-01 7.47702599e-01
2.30406567e-01 -4.76704955e-01 8.06357265e-01 6.11873507e-01
-2.08952367e-01 -1.28693795e+00 -1.04348850e+00 -6.74445391e-01
3.76135439e-01 -1.05578667e-02 5.72484076e-01 1.05919552e+00
-6.20951056e-02 -4.13085110e-02 -4.87395465e-01 1.17013775e-01
6.70183420e-01 1.46515772e-01 7.80174732e-01 -8.86029541e-01
-3.87867838e-01 -3.47751260e-01 -3.82645547e-01 -8.20129454e-01
9.17932168e-02 -9.50404227e-01 4.10179645e-01 -1.44546342e+00
1.37407929e-02 -5.44327497e-01 -2.41342708e-01 6.17004752e-01
-3.92948955e-01 8.19499195e-02 7.26495907e-02 4.29951668e-01
-4.87291366e-01 6.98555410e-01 1.87740529e+00 -2.46937469e-01
-5.52507102e-01 2.93903232e-01 -4.71411198e-01 6.78775191e-01
6.09030962e-01 -6.50274813e-01 -2.34170526e-01 -5.04391372e-01
1.44982502e-01 5.80927968e-01 2.96043247e-01 -6.09222889e-01
3.11568230e-01 -3.86746943e-01 2.65348047e-01 -1.15754880e-01
5.65141737e-02 -8.28630149e-01 5.95907159e-02 9.88283873e-01
-5.87695718e-01 3.16765070e-01 -1.16333120e-01 6.39258444e-01
-2.70429522e-01 9.65454280e-02 1.09425080e+00 -2.66452163e-01
-2.49718353e-01 8.12643349e-01 -6.49698675e-02 3.65911186e-01
1.08369005e+00 5.54391928e-02 1.82204157e-01 -1.73693821e-01
-8.18216503e-01 3.69481206e-01 2.29130134e-01 2.37653658e-01
6.77441537e-01 -1.62808239e+00 -9.84212816e-01 1.33080885e-01
-1.75668567e-01 5.96002817e-01 1.31938383e-01 1.04569757e+00
-6.22494638e-01 -1.19742699e-01 -2.46677384e-01 -8.19738865e-01
-1.02528465e+00 5.47600448e-01 6.98764920e-01 -8.21494758e-01
-9.30361748e-01 4.96708632e-01 9.53767821e-02 -6.65203989e-01
2.05267176e-01 -4.96125579e-01 -2.93942779e-01 -2.96147645e-01
8.88776034e-02 1.27401441e-01 -1.67588461e-02 -5.93450248e-01
-3.07806462e-01 1.02545655e+00 1.88441258e-02 3.80522124e-02
1.66043830e+00 1.65433750e-01 7.78025240e-02 -1.00445569e-01
1.21193886e+00 -3.17810506e-01 -1.74930835e+00 -5.66040337e-01
-2.67030526e-04 -4.15113211e-01 -6.10493012e-02 -4.61169481e-01
-1.54386950e+00 6.22637928e-01 6.68618739e-01 -3.45935635e-02
1.08380413e+00 6.91742450e-02 9.30940807e-01 4.05625939e-01
1.56746745e-01 -1.12655473e+00 5.96252561e-01 3.59205842e-01
1.09006131e+00 -1.36651099e+00 1.47849783e-01 -1.90372262e-02
-7.56833494e-01 1.23242557e+00 4.44615752e-01 -5.16881168e-01
7.30804801e-01 3.76259416e-01 2.34064505e-01 -2.52695024e-01
-4.42088187e-01 1.11478388e-01 4.77658004e-01 4.70860451e-01
3.62071991e-01 -1.94531858e-01 -2.98200041e-01 1.54813007e-01
8.88546333e-02 3.56455036e-02 1.96428731e-01 1.14383233e+00
-1.83661673e-02 -1.53845644e+00 -3.83884072e-01 3.97532731e-02
-6.39787734e-01 3.49926382e-01 1.42841294e-01 8.11672568e-01
-2.27928013e-01 3.17197591e-01 -7.71087781e-02 3.03313471e-02
4.96189117e-01 -3.54243308e-01 6.09825373e-01 -5.14702082e-01
-9.38499928e-01 4.76792306e-01 -4.44160730e-01 -9.08262074e-01
-8.23428273e-01 -7.98130810e-01 -1.45005119e+00 7.74948299e-02
-1.17988281e-01 -1.83613971e-01 4.84004617e-01 9.27003920e-01
1.58090502e-01 5.61879873e-01 9.89424586e-01 -1.15681541e+00
-1.07242036e+00 -6.62488937e-01 -1.40981242e-01 7.00372934e-01
6.89417541e-01 -5.10289788e-01 1.04972020e-01 1.67481601e-01]
|
[14.17306900024414, -2.5496015548706055]
|
03d0f791-33ea-483b-83f0-b23381704a74
|
adversarial-learning-with-mask-reconstruction
| null | null |
https://github.com/GaranWu/ALMR
|
https://pan.baidu.com/s/1-FZHhxtGMxKexVY82_Ceyw?pwd=6jj5
|
Adversarial Learning with Mask Reconstruction for Text-Guided Image Inpainting
|
Text-guided image inpainting aims to complete the corrupted patches coherent with both visual and textual context. On one hand, existing works focus on surrounding pixels of the corrupted patches without considering the objects in the image, resulting in the characteristics of objects described in text being painted on non-object regions. On the other hand, the redundant information in text may distract the generation of objects of interest in the restored image. In this paper, we propose an adversarial learning framework with mask reconstruction (ALMR) for image inpainting with textual guidance, which consists of a two-stage generator and dual discriminators. The two-stage generator aims to restore coarse-grained and fine-grained images, respectively. In particular, we devise a dual-attention module (DAM) to incorporate the word-level and sentence-level textual features as guidance on generating the coarse-grained and fine-grained details in the two stages. Furthermore, we design a mask reconstruction module (MRM) to penalize the restoration of the objects of interest with the given textual descriptions about the objects. For adversarial training, we exploit global and local discriminators for the whole image and corrupted patches, respectively. Extensive experiments conducted on CUB-200-2011, Oxford-102 and CelebA-HQ show the outperformance of the proposed ALMR (e.g., FID value is reduced from 29.69 to 14.69 compared with the state-of-the-art approach on CUB-200-2011). Codes are available at \href{https://github.com/GaranWu/ALMR}
|
['and Wenyin Liu', 'Qing Li', 'Yi Yu', 'Zhenguo Yang', 'Jiaqi Zeng', 'Yucheng Xie', 'Xingcai Wu']
|
2021-07-28
| null | null | null |
conference-2021-7
|
['image-inpainting']
|
['computer-vision']
|
[ 3.05302203e-01 1.35979299e-02 1.65910125e-01 -1.37115017e-01
-1.08164728e+00 -2.12941572e-01 4.22206789e-01 -3.33488524e-01
-1.67249367e-01 8.54660749e-01 3.03460568e-01 8.46223682e-02
3.17493111e-01 -8.41320932e-01 -9.22447324e-01 -8.92965198e-01
5.55090547e-01 -8.04250017e-02 -4.36509959e-02 -2.58936822e-01
1.56145051e-01 3.33600819e-01 -1.33080149e+00 7.07786679e-01
1.20343995e+00 9.69954014e-01 6.70410037e-01 4.42461550e-01
-6.53034300e-02 8.97863865e-01 -7.97542214e-01 -5.12716651e-01
2.10080668e-01 -7.79964089e-01 -3.38961065e-01 5.27979851e-01
4.99710947e-01 -6.87306643e-01 -7.31581390e-01 1.26141167e+00
4.44692701e-01 1.82270408e-01 6.08623445e-01 -9.70583975e-01
-1.19187248e+00 3.24925482e-01 -1.01084614e+00 1.34089947e-01
2.55258501e-01 5.41194618e-01 6.54992759e-01 -1.07388675e+00
5.66738904e-01 1.41754174e+00 2.77893186e-01 6.61717713e-01
-1.13034022e+00 -6.19387686e-01 1.48688436e-01 1.96305051e-01
-1.43150127e+00 -5.24430573e-01 1.01426828e+00 -3.06652069e-01
4.41801667e-01 4.50484961e-01 2.27974117e-01 1.13467181e+00
4.09943253e-01 8.73090208e-01 1.21340632e+00 -3.63336295e-01
-8.95994306e-02 1.26168728e-01 -5.26399493e-01 5.39937615e-01
1.97285622e-01 1.20345421e-01 -2.97447890e-01 1.80480242e-01
9.54312563e-01 1.89615130e-01 -5.12163162e-01 3.57965410e-01
-1.05928838e+00 6.90435886e-01 4.96674120e-01 2.97905594e-01
-6.05382085e-01 -1.40080452e-02 2.68789530e-01 1.51919439e-01
8.53315055e-01 1.88433483e-01 -6.03744015e-02 3.98298115e-01
-1.05194724e+00 3.48362058e-01 1.75553665e-01 1.03023279e+00
8.36098015e-01 3.58162880e-01 -7.72888184e-01 9.51112270e-01
1.63341686e-01 5.67061484e-01 4.74789381e-01 -8.37656915e-01
7.49017596e-01 3.78172249e-01 1.86335176e-01 -1.13299775e+00
3.86942625e-01 -5.05339444e-01 -1.24500632e+00 1.78016752e-01
-2.40165088e-02 -1.32794335e-01 -1.15256274e+00 1.53823996e+00
1.70716122e-01 1.78890213e-01 4.93378974e-02 1.09964848e+00
1.02474451e+00 1.09212363e+00 7.72157982e-02 -2.64963865e-01
1.25999618e+00 -1.34564877e+00 -9.82581735e-01 -6.15174532e-01
2.05960032e-02 -1.00097549e+00 1.19041228e+00 1.50723487e-01
-1.30370212e+00 -9.16851521e-01 -9.41462457e-01 -2.56754518e-01
-1.13642560e-02 2.87020683e-01 3.70449796e-02 1.30983159e-01
-7.97058225e-01 4.17097181e-01 -4.61970657e-01 1.19582199e-01
7.15108514e-01 -2.04225808e-01 -2.60339648e-01 -5.84471405e-01
-1.21159899e+00 6.24740601e-01 3.10281932e-01 1.78875580e-01
-1.19061065e+00 -5.47704220e-01 -8.90117466e-01 -3.42766233e-02
3.64510238e-01 -5.72881997e-01 8.48491132e-01 -1.22452307e+00
-1.13001394e+00 8.56377125e-01 -5.23513779e-02 -1.72065422e-01
8.36408496e-01 -8.47985521e-02 -3.97794902e-01 2.81010449e-01
5.04448116e-01 8.89034808e-01 1.26086819e+00 -1.64156520e+00
-5.43164492e-01 -1.19153343e-01 -7.72837773e-02 5.09174645e-01
-1.44932836e-01 -1.84361190e-01 -7.98014402e-01 -1.47111893e+00
-1.81999251e-01 -4.18365717e-01 -1.58142447e-01 1.87179029e-01
-6.85735047e-01 1.41340867e-01 8.27988863e-01 -1.40078568e+00
1.17398560e+00 -2.49820995e+00 1.32405102e-01 -3.85692239e-01
3.86073291e-02 2.31904119e-01 -4.16113794e-01 3.73542517e-01
-4.58813757e-02 1.45650923e-01 -5.88652909e-01 -6.21553719e-01
-1.64763898e-01 2.54717499e-01 -4.94378179e-01 4.60934788e-01
4.52556342e-01 9.67786670e-01 -7.20978081e-01 -5.32513380e-01
3.22150648e-01 4.99688923e-01 -3.42223078e-01 5.69000602e-01
-3.38514179e-01 7.75065660e-01 -4.97330397e-01 6.39522672e-01
1.11669612e+00 4.54976559e-02 -3.60241234e-01 -3.10693860e-01
1.11114867e-01 -9.64085013e-02 -9.54477608e-01 1.75952077e+00
-4.02892709e-01 3.20804358e-01 3.56455803e-01 -7.87704587e-01
7.39767730e-01 2.19131082e-01 1.31354630e-01 -9.22686160e-01
4.48882021e-02 1.01209491e-01 -2.34345257e-01 -5.98119378e-01
5.32875359e-01 -2.11450711e-01 2.48280987e-02 3.30512583e-01
-1.84789047e-01 -1.27405390e-01 6.55453205e-02 1.92099363e-01
7.32861459e-01 1.25122949e-01 -3.63764353e-02 4.68724668e-02
6.88195527e-01 -1.15718208e-01 5.04235506e-01 5.85064709e-01
5.00029959e-02 1.19835567e+00 2.12537080e-01 -2.03744054e-01
-1.17319489e+00 -8.35502446e-01 4.20347191e-02 7.72840858e-01
4.47018296e-01 -1.80564485e-02 -8.65210414e-01 -6.05177641e-01
-1.46345645e-01 9.13286209e-01 -7.86889732e-01 -2.92805254e-01
-5.08388400e-01 -4.62183535e-01 3.52691412e-01 1.88502461e-01
9.01726961e-01 -1.54210043e+00 -1.53260872e-01 9.30893049e-02
-5.50543547e-01 -1.05303335e+00 -1.00909972e+00 -2.91851699e-01
-6.24204814e-01 -6.82669640e-01 -1.06860864e+00 -8.78203273e-01
9.93419290e-01 3.93978626e-01 9.08191860e-01 2.81324834e-01
-3.08324754e-01 -1.39385879e-01 -5.20720899e-01 7.66289681e-02
-5.64930081e-01 -3.32001835e-01 -4.20494944e-01 4.40673739e-01
-3.16278458e-01 -4.31111872e-01 -7.80176878e-01 1.52927846e-01
-1.44668746e+00 3.99916798e-01 1.03365612e+00 1.10451043e+00
7.29820371e-01 3.35254282e-01 4.76237476e-01 -8.32392991e-01
5.54295838e-01 -6.61666155e-01 -1.73828244e-01 1.18369877e-01
-3.29798847e-01 -2.51508296e-01 8.57845485e-01 -5.38592577e-01
-1.23342609e+00 -2.33269393e-01 -2.64510036e-01 -7.98386872e-01
-2.32330918e-01 3.34747136e-01 -6.36722922e-01 2.15527534e-01
4.01280999e-01 8.78130853e-01 -8.41705650e-02 -6.25753462e-01
4.22825873e-01 6.99391961e-01 6.27455473e-01 -5.13742268e-01
9.47091341e-01 4.59593266e-01 -3.80335778e-01 -4.53772962e-01
-8.46214294e-01 2.87037948e-03 -8.73174742e-02 1.07532088e-03
8.54243159e-01 -1.16640532e+00 -6.16540492e-04 6.42963588e-01
-1.28551531e+00 -3.62282515e-01 -4.54902470e-01 3.34740803e-02
-5.39552152e-01 5.67775667e-01 -7.53948748e-01 -5.88758111e-01
-5.59970379e-01 -1.22238231e+00 1.22831309e+00 2.88197786e-01
2.70014614e-01 -6.08212650e-01 -3.79504889e-01 5.73027790e-01
3.52095187e-01 4.29974139e-01 6.79164350e-01 -1.52027637e-01
-6.72897637e-01 -3.88794914e-02 -3.93529564e-01 7.09903359e-01
2.49249622e-01 -3.71522546e-01 -9.15540814e-01 -4.93420333e-01
2.87742764e-01 -3.06197166e-01 9.28358436e-01 3.68983686e-01
1.36134231e+00 -7.03035474e-01 -5.42976409e-02 6.27341092e-01
1.49246383e+00 1.74507126e-01 1.14614248e+00 2.64334381e-01
8.10891390e-01 5.84893823e-01 9.11858141e-01 4.21045303e-01
3.80067639e-02 6.03862882e-01 4.81219143e-01 -4.57481533e-01
-6.91406071e-01 -5.29890060e-01 3.95220429e-01 5.27900159e-01
2.63674021e-01 -5.44812977e-01 -2.98080951e-01 7.21051514e-01
-1.68132782e+00 -9.47968960e-01 -5.84540032e-02 1.91170442e+00
1.10943913e+00 -8.52166936e-02 -3.77928883e-01 -5.92497513e-02
1.04691672e+00 5.51834404e-01 -6.26182258e-01 -7.97339529e-02
-4.33765560e-01 1.02495544e-01 3.09466690e-01 5.61640799e-01
-9.20653403e-01 9.05742347e-01 4.30733109e+00 1.62283301e+00
-9.84254479e-01 4.07347918e-01 1.26805258e+00 1.58241298e-02
-3.64738703e-01 -1.46955132e-01 -4.04347360e-01 9.34822440e-01
3.68131697e-01 6.38299342e-03 6.55100882e-01 5.10718584e-01
3.99916500e-01 -6.84513450e-02 -5.78820229e-01 9.86310840e-01
2.50158906e-01 -1.23735833e+00 3.47787827e-01 -1.52910486e-01
1.09430039e+00 -3.84230673e-01 3.87116104e-01 1.53256461e-01
-1.57532334e-01 -1.10544586e+00 1.09608579e+00 6.36226535e-01
1.34075260e+00 -8.49434793e-01 5.96286833e-01 4.59667832e-01
-9.50791121e-01 2.29267441e-02 -5.04537582e-01 2.37376660e-01
7.96276331e-02 7.30110645e-01 -1.74813658e-01 8.82320762e-01
7.40845501e-01 7.43069768e-01 -5.14588594e-01 6.42168045e-01
-5.12550652e-01 3.90967697e-01 2.11530045e-01 5.70124865e-01
2.06342250e-01 -2.20197454e-01 6.06935382e-01 1.06449652e+00
4.24419671e-01 2.01407880e-01 1.07909761e-01 1.28283787e+00
-2.72630006e-01 8.70821476e-02 -4.49797034e-01 6.67523965e-02
3.62000436e-01 1.28597438e+00 -3.99368733e-01 -5.29553473e-01
-4.32484269e-01 1.48240983e+00 1.41078591e-01 5.04719138e-01
-1.00047541e+00 -4.68571603e-01 3.57878000e-01 3.02102238e-01
5.07847130e-01 1.95738718e-01 -4.28776324e-01 -1.24991369e+00
3.32504034e-01 -1.20957685e+00 7.05900639e-02 -1.06597686e+00
-1.41430461e+00 8.59301448e-01 -2.46157035e-01 -1.42443287e+00
1.13053739e-01 -4.82950397e-02 -7.25993514e-01 1.36151564e+00
-1.50654864e+00 -1.37192881e+00 -5.06797493e-01 6.03320956e-01
1.03071010e+00 -7.98641369e-02 4.11055803e-01 4.30029809e-01
-5.85091889e-01 6.29857242e-01 1.96999192e-01 1.45217821e-01
8.66987884e-01 -8.02825212e-01 4.10584956e-01 1.21011209e+00
-1.57948434e-01 2.35503152e-01 6.17403984e-01 -9.81151223e-01
-1.19346893e+00 -1.71738458e+00 5.47397792e-01 3.84186697e-03
2.07098484e-01 -2.94380963e-01 -9.80096698e-01 5.23762882e-01
5.72254539e-01 7.76482373e-02 -6.62461063e-03 -9.11190748e-01
-2.02119216e-01 -1.25629067e-01 -1.39920032e+00 7.29740441e-01
7.30849326e-01 -4.08635497e-01 -2.83633560e-01 3.09096247e-01
8.73612523e-01 -5.89106798e-01 -5.38055301e-01 2.93890506e-01
4.79881875e-02 -9.89324689e-01 9.28162336e-01 -5.42373657e-02
1.03282320e+00 -5.71207643e-01 -2.57124156e-01 -1.23494053e+00
-4.17442083e-01 -5.30040264e-01 -3.62777226e-02 1.48178387e+00
7.42918551e-02 -4.69676614e-01 4.65355247e-01 2.25847155e-01
-3.69178176e-01 -8.48738372e-01 -7.78599977e-01 -4.53204393e-01
4.62772772e-02 -5.62888682e-02 5.32446504e-01 8.73718441e-01
-7.40355730e-01 2.22435609e-01 -8.11195552e-01 7.45793954e-02
5.86420476e-01 2.19294727e-01 5.74545562e-01 -2.72402167e-01
-4.45993334e-01 -2.42123604e-01 1.82326995e-02 -1.15842724e+00
8.42759386e-02 -6.76192045e-01 2.16831252e-01 -1.59846187e+00
3.67191225e-01 -2.69375861e-01 -1.11210778e-01 4.90336329e-01
-6.08796597e-01 6.57404602e-01 1.94033846e-01 4.52674776e-01
-2.56576449e-01 9.03796792e-01 1.85402429e+00 -4.44329202e-01
1.96430340e-01 -2.48300686e-01 -9.44149315e-01 2.69106358e-01
7.20843077e-01 -5.12599826e-01 -3.01024437e-01 -7.15879798e-01
-3.62156004e-01 1.98629946e-01 6.40989780e-01 -7.27679014e-01
-8.16497952e-02 -2.22607359e-01 7.21931159e-01 -5.97704649e-01
3.84233475e-01 -6.39643192e-01 3.00506592e-01 4.54206228e-01
-2.71238148e-01 -9.18271169e-02 2.88213104e-01 7.08622098e-01
-4.36363846e-01 -1.48236707e-01 1.02730250e+00 -3.38046849e-01
-5.02049029e-01 3.88174057e-01 -8.27329382e-02 7.60475323e-02
1.08208048e+00 -9.78377759e-02 -4.28881705e-01 -4.74303305e-01
-4.86237586e-01 6.90085515e-02 6.69930398e-01 3.76903594e-01
9.48750436e-01 -1.49069738e+00 -1.26728749e+00 3.40036154e-01
-3.15631069e-02 2.62532920e-01 9.00267541e-01 6.78828895e-01
-6.70241594e-01 8.31191540e-02 -2.90325850e-01 -2.73777664e-01
-9.22790051e-01 7.83413410e-01 1.75519243e-01 -3.65605295e-01
-6.19490743e-01 7.68865049e-01 7.00153410e-01 -1.79201867e-02
5.17959930e-02 1.03322573e-01 3.31008323e-02 -1.98583424e-01
5.53532600e-01 1.06708810e-01 -1.55639529e-01 -8.13187540e-01
-6.92434385e-02 3.98326248e-01 -3.20789397e-01 -6.16420433e-02
1.24052751e+00 -2.73840487e-01 -3.10385555e-01 3.67003269e-02
1.05880940e+00 1.63529932e-01 -1.52771866e+00 -2.30683833e-01
-7.05657125e-01 -7.72220075e-01 2.62052976e-02 -9.08990383e-01
-1.38643706e+00 8.40685248e-01 5.25204003e-01 -8.42603073e-02
1.48994040e+00 -4.60964032e-02 9.77110028e-01 -4.32867050e-01
-4.73828241e-02 -5.81396461e-01 2.65079796e-01 5.28900214e-02
1.43818617e+00 -1.05996001e+00 5.90844713e-02 -4.15834785e-01
-8.41318607e-01 6.35427117e-01 8.48622262e-01 -4.22793895e-01
1.17330618e-01 -2.37079132e-02 2.63251401e-02 1.42220050e-01
-5.90426266e-01 1.00151084e-01 3.57722878e-01 4.81870204e-01
1.39941856e-01 -3.32984142e-02 -4.46030080e-01 7.62560785e-01
2.49727950e-01 -2.82854259e-01 4.44102973e-01 8.23059618e-01
-2.38481835e-01 -9.13016915e-01 -6.32077157e-01 3.38207930e-01
-5.58482945e-01 -5.76595664e-01 -1.82357758e-01 5.32455802e-01
4.78222400e-01 1.08933032e+00 -6.83104992e-02 -3.21917295e-01
2.09103197e-01 -3.67280781e-01 3.43795151e-01 -6.66894197e-01
-4.12987798e-01 3.87928277e-01 -1.51618972e-01 -3.90244633e-01
-1.31835744e-01 -4.31428820e-01 -8.96214366e-01 -2.84140527e-01
-1.18880473e-01 -1.16133075e-02 2.33140200e-01 6.43011093e-01
4.77395117e-01 7.48953521e-01 8.32483768e-01 -9.78927970e-01
-3.70745867e-01 -1.27151585e+00 -7.29278862e-01 5.42504966e-01
3.68864447e-01 -3.21997106e-01 -4.11493659e-01 3.36967379e-01]
|
[11.362399101257324, -1.1649285554885864]
|
ef202f17-8ac4-429f-80a7-3a864d2ff54b
|
graph-similarity-learning-for-change-point
|
2203.1547
| null |
https://arxiv.org/abs/2203.15470v1
|
https://arxiv.org/pdf/2203.15470v1.pdf
|
Graph similarity learning for change-point detection in dynamic networks
|
Dynamic networks are ubiquitous for modelling sequential graph-structured data, e.g., brain connectome, population flows and messages exchanges. In this work, we consider dynamic networks that are temporal sequences of graph snapshots, and aim at detecting abrupt changes in their structure. This task is often termed network change-point detection and has numerous applications, such as fraud detection or physical motion monitoring. Leveraging a graph neural network model, we design a method to perform online network change-point detection that can adapt to the specific network domain and localise changes with no delay. The main novelty of our method is to use a siamese graph neural network architecture for learning a data-driven graph similarity function, which allows to effectively compare the current graph and its recent history. Importantly, our method does not require prior knowledge on the network generative distribution and is agnostic to the type of change-points; moreover, it can be applied to a large variety of networks, that include for instance edge weights and node attributes. We show on synthetic and real data that our method enjoys a number of benefits: it is able to learn an adequate graph similarity function for performing online network change-point detection in diverse types of change-point settings, and requires a shorter data history to detect changes than most existing state-of-the-art baselines.
|
['Xiaowen Dong', 'Mihai Cucuringu', 'Henry Kenlay', 'Deborah Sulem']
|
2022-03-29
| null | null | null | null |
['graph-similarity', 'temporal-sequences']
|
['graphs', 'reasoning']
|
[ 1.96258307e-01 -1.24803595e-01 -3.00718069e-01 3.77453752e-02
2.51444757e-01 -6.38091326e-01 7.30156898e-01 6.22751355e-01
-2.63746083e-01 4.17208254e-01 -9.08031613e-02 -3.09605122e-01
-4.09779161e-01 -1.01583779e+00 -5.81047177e-01 -4.13995177e-01
-8.79368126e-01 5.37412465e-01 6.77227795e-01 -4.29424763e-01
-5.10890968e-03 7.71561146e-01 -7.90828109e-01 -3.73265952e-01
5.85341632e-01 7.46295094e-01 -7.81095177e-02 8.39031935e-01
1.09784372e-01 7.93659985e-01 -5.16902864e-01 -3.92127514e-01
1.28982812e-01 -4.62688029e-01 -6.74879551e-01 1.46871537e-01
2.82069385e-01 -8.50313753e-02 -9.80039895e-01 1.11687589e+00
4.58513379e-01 1.50957689e-01 5.99815309e-01 -1.46670532e+00
-3.50300819e-01 6.22027576e-01 -5.48730016e-01 9.55644011e-01
2.04092145e-01 2.91851640e-01 1.05753863e+00 -2.89777040e-01
9.09426570e-01 1.29133737e+00 1.05456853e+00 3.98636550e-01
-1.48552036e+00 -4.75816280e-01 6.26526177e-01 3.27630818e-01
-1.02222061e+00 -3.06564271e-01 1.10721242e+00 -5.48070610e-01
7.52463460e-01 5.75503297e-02 1.01686966e+00 1.31558275e+00
4.25098926e-01 5.81211627e-01 5.52403629e-01 7.36020207e-02
3.15827936e-01 -6.85187161e-01 1.41437322e-01 8.91434371e-01
2.67483681e-01 -1.38033643e-01 -3.51004809e-01 -3.28460634e-01
9.08885539e-01 4.48959291e-01 -2.53097206e-01 -5.22205472e-01
-1.32751071e+00 7.47728646e-01 7.80113995e-01 4.98594046e-01
-3.39909673e-01 6.29391432e-01 6.71511769e-01 7.26932466e-01
6.68126941e-01 1.75004929e-01 -3.09069455e-01 -1.22800976e-01
-6.38621032e-01 4.57996801e-02 1.02288461e+00 5.66956520e-01
5.75779676e-01 -6.78787306e-02 -6.44286796e-02 6.29445672e-01
9.88350660e-02 4.22107339e-01 4.13819313e-01 -5.94817460e-01
5.52832186e-01 7.11267948e-01 -3.45455021e-01 -1.74362898e+00
-7.29204595e-01 -5.93343258e-01 -1.33113301e+00 -2.03572750e-01
4.25129324e-01 -2.14536369e-01 -8.60042095e-01 1.95198572e+00
4.11410958e-01 7.23545134e-01 -4.82102811e-01 3.00745666e-01
5.49139500e-01 2.34243974e-01 -3.57782751e-01 -1.15159482e-01
1.06510079e+00 -5.46052039e-01 -5.35190403e-01 -3.63013446e-01
4.77162689e-01 1.07825384e-01 7.50818193e-01 9.32231545e-03
-8.13774407e-01 -5.83531298e-02 -8.64621520e-01 4.75249678e-01
-4.92464662e-01 -7.71110892e-01 7.22680867e-01 5.15915453e-01
-1.23432600e+00 9.36533809e-01 -1.27518058e+00 -8.63558471e-01
5.68051636e-01 3.31148237e-01 -2.27341771e-01 9.76967737e-02
-1.14780366e+00 4.31890696e-01 3.69834483e-01 1.06827300e-02
-8.54676604e-01 -6.74161315e-01 -8.52952838e-01 1.89523011e-01
6.82031810e-01 -8.00197423e-01 8.79867017e-01 -1.09243059e+00
-1.34416258e+00 6.45105839e-01 6.69282302e-02 -7.17461824e-01
7.38980711e-01 4.12282169e-01 -7.01187551e-01 3.13649118e-01
-3.09874602e-02 2.92627010e-02 9.84948814e-01 -5.71927547e-01
-2.48842537e-01 -3.12046200e-01 3.10048871e-02 -8.42389166e-02
-4.94349599e-01 -1.72791600e-01 -6.29126430e-01 -8.72333765e-01
6.21829703e-02 -9.50286806e-01 -3.68386507e-01 1.77208364e-01
-5.36754310e-01 -1.30832821e-01 8.20758462e-01 -3.02499264e-01
1.29840791e+00 -1.88162565e+00 2.76115060e-01 6.59467161e-01
9.99361873e-01 6.15926757e-02 -3.21037889e-01 7.84848928e-01
9.82919335e-02 1.99393660e-01 -4.05557543e-01 -2.00129464e-01
-1.87890455e-01 1.41525969e-01 3.99119854e-02 6.56973660e-01
1.46232113e-01 1.23708284e+00 -1.29694414e+00 -1.87108457e-01
-1.22195557e-01 2.96146035e-01 -3.93755674e-01 -8.30090493e-02
-1.64788127e-01 3.28197390e-01 -3.84751379e-01 4.52912390e-01
1.84225112e-01 -7.60364830e-01 5.52840471e-01 1.61914930e-01
5.92312753e-01 -9.33943200e-04 -1.13100660e+00 1.41757870e+00
-1.42957926e-01 8.16606760e-01 1.09131344e-01 -1.32278979e+00
7.39490271e-01 1.09851353e-01 7.35227585e-01 -5.79198539e-01
1.65638123e-02 -2.99555305e-02 2.57122844e-01 -3.17651629e-01
-2.49100942e-02 2.98826247e-01 -9.30145681e-02 6.32389903e-01
-1.70566496e-02 2.90823191e-01 4.11870867e-01 5.78503072e-01
2.03969312e+00 -7.09579229e-01 4.83851999e-01 -1.08195171e-01
3.52112293e-01 -4.93295819e-01 7.34758675e-01 9.04950559e-01
-3.59241635e-01 1.11224443e-01 1.03945827e+00 -5.63730419e-01
-7.08500564e-01 -1.18909037e+00 3.52679044e-01 9.44009960e-01
2.56375790e-01 -3.00755531e-01 -4.63418514e-01 -9.69487309e-01
3.82276893e-01 -1.55589610e-01 -7.06463516e-01 -4.52382088e-01
-7.12357163e-01 -8.07275772e-01 5.00656188e-01 2.32889086e-01
3.40477794e-01 -1.02690125e+00 -2.07847372e-01 5.23463666e-01
5.76696061e-02 -1.05968034e+00 -9.57559586e-01 -2.02705383e-01
-1.07598698e+00 -1.56013119e+00 -4.27010864e-01 -7.16591179e-01
7.60398507e-01 2.97130704e-01 1.37532711e+00 2.97718763e-01
-2.72209644e-01 8.05709243e-01 -7.32730255e-02 1.48327267e-02
-4.84998167e-01 3.82669508e-01 1.54256850e-01 4.43825781e-01
-1.02483369e-01 -1.19861090e+00 -7.20557570e-01 2.55459309e-01
-8.67604852e-01 -4.41575259e-01 3.72419924e-01 7.73824930e-01
3.02581728e-01 2.49170378e-01 8.35132003e-01 -1.35470736e+00
1.04867411e+00 -8.79811823e-01 -5.72206497e-01 1.85327366e-01
-8.18671823e-01 -4.23346870e-02 7.76558399e-01 -7.63210356e-01
-2.43731022e-01 -2.95231521e-01 3.04679990e-01 -4.19523269e-01
2.10704029e-01 8.41993034e-01 9.29047167e-02 -2.68929511e-01
6.08351529e-01 1.88031986e-01 2.49223992e-01 -2.26382032e-01
3.95968080e-01 -1.07696205e-02 6.02063060e-01 -1.13534302e-01
9.36216116e-01 5.36113858e-01 3.59160453e-01 -7.59552717e-01
-1.31383464e-01 -5.15927613e-01 -7.17649221e-01 -5.00283301e-01
1.57553747e-01 -5.98382235e-01 -1.02265763e+00 7.89190710e-01
-8.92505109e-01 -6.95245087e-01 -1.59454793e-01 1.09802216e-01
-2.99403936e-01 6.92272067e-01 -8.91033530e-01 -4.20645028e-01
-4.74256366e-01 -5.69172680e-01 5.95015168e-01 -5.83084486e-02
-9.51284096e-02 -1.80504036e+00 2.73440212e-01 -4.13416713e-01
5.64988971e-01 8.55335295e-01 8.78664613e-01 -7.24955440e-01
-5.58987260e-01 -3.43288332e-01 -1.27332464e-01 -1.47721753e-01
5.20259082e-01 -9.75173898e-03 -2.53768504e-01 -7.84349203e-01
-3.82937074e-01 1.49996787e-01 9.08203244e-01 4.39965189e-01
1.03273213e+00 -7.03739643e-01 -7.13545799e-01 6.41135037e-01
1.15211177e+00 -5.17780613e-03 2.34430075e-01 4.47113588e-02
9.29668128e-01 3.15372586e-01 -1.70052961e-01 2.15630293e-01
5.09910941e-01 7.64925420e-01 5.92845798e-01 -6.62167221e-02
-1.32302865e-01 -3.36620897e-01 3.86099964e-01 9.47447896e-01
2.31041521e-01 -5.13633966e-01 -1.06928694e+00 6.27944350e-01
-2.20036721e+00 -1.12045038e+00 1.31007740e-02 2.14693284e+00
5.08411944e-01 4.76081580e-01 7.32178211e-01 -3.41456826e-03
1.00645494e+00 5.47017694e-01 -1.10116315e+00 1.49395123e-01
-2.28969404e-03 -7.44472891e-02 5.42916059e-01 3.89566630e-01
-1.01203132e+00 6.62670255e-01 6.28943825e+00 4.96297091e-01
-1.16477776e+00 1.21111847e-01 5.45204461e-01 -5.11044823e-02
-2.19795272e-01 -2.48145893e-01 -6.56821132e-02 6.67952895e-01
9.87295747e-01 -4.17879850e-01 7.95784175e-01 3.75319153e-01
2.69080848e-01 3.91077161e-01 -1.19146442e+00 1.09428978e+00
2.32586730e-02 -1.48685479e+00 -1.05916575e-01 1.06421135e-01
5.27500212e-01 4.51197505e-01 -1.12696730e-01 3.60974297e-02
6.43473923e-01 -8.48834753e-01 2.52284676e-01 6.39202476e-01
5.97446382e-01 -5.80255508e-01 4.42492962e-01 2.78284818e-01
-1.53485084e+00 -1.23268269e-01 2.30210200e-02 9.35455970e-03
1.10809550e-01 9.88816500e-01 -9.20098424e-01 5.22767484e-01
4.44385052e-01 1.42666996e+00 -6.71604633e-01 1.18640518e+00
-2.60933656e-02 7.43472815e-01 -5.15792787e-01 -1.57781512e-01
4.95802313e-02 -1.57794669e-01 9.10191774e-01 1.07903099e+00
1.39624625e-01 -3.72535050e-01 3.71654242e-01 6.58969402e-01
-5.19907773e-01 -5.98367453e-02 -8.89293790e-01 -1.88097298e-01
6.74597621e-01 1.13894010e+00 -1.12461030e+00 -1.15625016e-01
-2.21711934e-01 1.13167739e+00 6.73549414e-01 4.34015751e-01
-5.35226464e-01 -5.05134106e-01 7.26138413e-01 3.16518962e-01
3.55455875e-01 -3.94044787e-01 2.99554288e-01 -1.32154739e+00
1.65148199e-01 -7.44741023e-01 6.62222147e-01 -9.25025269e-02
-1.72382832e+00 5.19639850e-01 -1.70388907e-01 -1.07132852e+00
-3.17264259e-01 -4.08608943e-01 -1.23292494e+00 2.63800114e-01
-1.35480118e+00 -8.87691975e-01 -3.89596194e-01 8.78322601e-01
1.37754932e-01 -1.78076416e-01 2.76622117e-01 2.88348794e-01
-7.27707505e-01 6.02188170e-01 1.73627615e-01 5.73125839e-01
2.57664412e-01 -1.36218417e+00 1.04212308e+00 1.14700472e+00
3.94609928e-01 3.92638028e-01 5.65473914e-01 -9.63687599e-01
-1.56380701e+00 -1.38549125e+00 4.55527484e-01 -3.59942228e-01
1.29585421e+00 -5.72706580e-01 -8.81845593e-01 7.84276724e-01
-2.62965202e-01 5.03269792e-01 1.05802238e-01 2.26503849e-01
-3.82959157e-01 -4.10755515e-01 -1.07025790e+00 6.90257668e-01
1.63203335e+00 -6.68560624e-01 2.21162033e-03 5.99902868e-01
6.48014903e-01 -3.16569328e-01 -8.97476614e-01 1.60476372e-01
2.81473100e-01 -5.57377338e-01 9.76790309e-01 -7.59040236e-01
-8.74636918e-02 -1.16178364e-01 5.90017974e-01 -1.59248340e+00
-5.54749131e-01 -1.23857057e+00 -7.89693475e-01 1.06018806e+00
3.07346106e-01 -1.12420082e+00 7.88869679e-01 1.55112624e-01
4.03058290e-01 -6.20493710e-01 -1.22759151e+00 -1.01050758e+00
-3.78480673e-01 -2.00998873e-01 6.64947033e-01 1.18216848e+00
-1.05557460e-02 4.52636629e-01 -3.57406080e-01 1.83730170e-01
6.63214445e-01 -1.41274676e-01 7.80027509e-01 -1.67434776e+00
-3.83027643e-01 -8.99591267e-01 -9.07652020e-01 -8.99603009e-01
1.77880570e-01 -1.22363234e+00 -4.31984514e-01 -1.45638549e+00
-9.34006274e-03 -3.25946003e-01 -4.22973007e-01 3.97247881e-01
-2.36097991e-01 -2.64713187e-02 8.31019729e-02 2.17826620e-01
-6.87606215e-01 5.16417563e-01 9.57845807e-01 -3.10435861e-01
-4.48728621e-01 3.71813267e-01 -4.45127100e-01 4.99745697e-01
6.86892748e-01 -6.46648645e-01 -5.56973159e-01 -1.10100441e-01
4.30832922e-01 1.63393274e-01 5.46061039e-01 -9.75122631e-01
5.23034394e-01 -1.77037641e-02 -4.62489873e-02 -5.48844524e-02
-7.50641972e-02 -7.83833921e-01 2.44656563e-01 8.32962513e-01
-2.63893336e-01 4.75002021e-01 -1.18142948e-01 1.50999212e+00
1.52367860e-01 2.99155414e-01 6.18936419e-01 -6.04588948e-02
-5.49731374e-01 1.08274162e+00 -3.33636373e-01 4.43217903e-01
1.06003404e+00 -1.03034250e-01 -4.25966918e-01 -6.78524911e-01
-6.40364349e-01 5.30801177e-01 4.95090932e-01 4.99561638e-01
4.91296828e-01 -1.41106927e+00 -6.76294744e-01 -3.32495309e-02
1.29721552e-01 -2.77734339e-01 -9.50498786e-03 9.72827196e-01
-4.17909116e-01 -1.73102126e-01 2.54463986e-04 -7.33865499e-01
-1.03221560e+00 6.03314698e-01 6.79276943e-01 -6.37340128e-01
-1.07477188e+00 4.41533208e-01 -2.01791018e-01 -4.69015479e-01
1.66233361e-01 -4.36373979e-01 -1.91017762e-01 1.45318165e-01
4.08059388e-01 4.62233514e-01 -6.75019473e-02 -2.73837596e-01
-4.58763540e-01 3.44885558e-01 -4.86977100e-02 2.29003638e-01
1.51867485e+00 -2.30244622e-01 -2.22409070e-01 6.82873905e-01
1.14315820e+00 -2.65195251e-01 -1.32043052e+00 -7.60277867e-01
2.75086045e-01 -3.27054948e-01 -1.88434333e-01 -3.96388054e-01
-1.60394871e+00 4.28052604e-01 3.74529868e-01 8.61148536e-01
9.84824240e-01 9.28962156e-02 8.24875712e-01 7.49221981e-01
3.05402100e-01 -7.86218405e-01 3.33483785e-01 5.86841404e-01
5.82295299e-01 -1.17897165e+00 -1.32407129e-01 -2.89288700e-01
-1.56562850e-01 1.17813015e+00 2.84855694e-01 -2.86453873e-01
1.03526497e+00 -9.18971077e-02 -4.54094708e-01 -7.05099106e-01
-7.93074608e-01 -1.38899341e-01 2.27762014e-01 7.07762420e-01
-1.38573661e-01 2.74400823e-02 7.02054054e-02 -6.67480603e-02
1.93377122e-01 -2.08743140e-01 7.22474396e-01 6.18013799e-01
-6.96065798e-02 -7.94983685e-01 3.06090891e-01 9.13274884e-01
-2.68558145e-01 2.84595471e-02 -5.52559555e-01 6.02611482e-01
-5.34508824e-01 7.79136896e-01 1.39068484e-01 -3.43800694e-01
3.53885710e-01 -2.23149836e-01 1.65204272e-01 -5.85078478e-01
-5.12140572e-01 -5.04275084e-01 -3.42778787e-02 -7.62207389e-01
-3.82206053e-01 -7.11580753e-01 -8.81962121e-01 -7.95778275e-01
-8.18934962e-02 -1.14209458e-01 1.22897319e-01 8.58150542e-01
7.27012575e-01 6.17219567e-01 8.21203947e-01 -7.17022538e-01
-2.19312623e-01 -7.45899916e-01 -7.28090107e-01 7.53966153e-01
6.83032990e-01 -6.12146318e-01 -4.64774966e-01 -2.64512181e-01]
|
[7.030505180358887, 5.9414215087890625]
|
d391207f-3541-407d-8aa1-1d4f12c6957b
|
the-clear-benchmark-continual-learning-on
|
2201.06289
| null |
https://arxiv.org/abs/2201.06289v3
|
https://arxiv.org/pdf/2201.06289v3.pdf
|
The CLEAR Benchmark: Continual LEArning on Real-World Imagery
|
Continual learning (CL) is widely regarded as crucial challenge for lifelong AI. However, existing CL benchmarks, e.g. Permuted-MNIST and Split-CIFAR, make use of artificial temporal variation and do not align with or generalize to the real-world. In this paper, we introduce CLEAR, the first continual image classification benchmark dataset with a natural temporal evolution of visual concepts in the real world that spans a decade (2004-2014). We build CLEAR from existing large-scale image collections (YFCC100M) through a novel and scalable low-cost approach to visio-linguistic dataset curation. Our pipeline makes use of pretrained vision-language models (e.g. CLIP) to interactively build labeled datasets, which are further validated with crowd-sourcing to remove errors and even inappropriate images (hidden in original YFCC100M). The major strength of CLEAR over prior CL benchmarks is the smooth temporal evolution of visual concepts with real-world imagery, including both high-quality labeled data along with abundant unlabeled samples per time period for continual semi-supervised learning. We find that a simple unsupervised pre-training step can already boost state-of-the-art CL algorithms that only utilize fully-supervised data. Our analysis also reveals that mainstream CL evaluation protocols that train and test on iid data artificially inflate performance of CL system. To address this, we propose novel "streaming" protocols for CL that always test on the (near) future. Interestingly, streaming protocols (a) can simplify dataset curation since today's testset can be repurposed for tomorrow's trainset and (b) can produce more generalizable models with more accurate estimates of performance since all labeled data from each time-period is used for both training and testing (unlike classic iid train-test splits).
|
['Deva Ramanan', 'Deepak Pathak', 'Jia Shi', 'Zhiqiu Lin']
|
2022-01-17
| null | null | null | null |
['unsupervised-pre-training']
|
['methodology']
|
[ 1.77255459e-02 -4.87719297e-01 -2.24874869e-01 -6.62105262e-01
-9.29915488e-01 -1.05896366e+00 9.28893030e-01 5.92730008e-02
-8.79833817e-01 8.61080468e-01 -1.01653516e-01 -4.97318119e-01
-3.36585678e-02 -3.72417271e-01 -1.05556345e+00 -4.42128211e-01
-1.32415980e-01 5.56916058e-01 2.25012809e-01 -2.88752973e-01
-1.20083816e-01 1.62575036e-01 -1.76416445e+00 6.41532838e-01
7.17572749e-01 8.24254155e-01 1.97277814e-01 6.64487600e-01
-6.28632680e-02 9.43320692e-01 -6.38304293e-01 -4.83151227e-01
5.24798274e-01 -4.70188767e-01 -8.26543510e-01 2.52538860e-01
8.40065777e-01 -1.80396214e-01 -8.58547539e-02 8.02010834e-01
3.07377994e-01 -7.07099289e-02 4.24226999e-01 -1.73344600e+00
-9.99734461e-01 8.33773375e-01 -6.29034162e-01 2.65772432e-01
5.39325699e-02 1.01185906e+00 9.37050402e-01 -9.34559286e-01
1.03372145e+00 9.99677062e-01 8.98827851e-01 6.60358310e-01
-1.48153889e+00 -7.86562800e-01 4.59859848e-01 2.91473418e-01
-1.27473223e+00 -4.17664319e-01 5.99213839e-01 -6.73180223e-01
7.19276965e-01 2.01977476e-01 8.14768434e-01 1.53000498e+00
-2.87320882e-01 1.14550769e+00 1.30757368e+00 -3.16753238e-01
3.75193119e-01 1.91696450e-01 1.92465752e-01 4.35835838e-01
-7.13382661e-02 2.43427470e-01 -5.69994152e-01 3.18352103e-01
1.86897963e-01 1.58998668e-02 -2.80499548e-01 -3.59144509e-01
-1.48638213e+00 6.05190396e-01 5.28954566e-01 3.27022523e-01
-1.81971222e-01 1.56318784e-01 5.55541992e-01 5.18171132e-01
4.18644935e-01 5.39275885e-01 -6.19327843e-01 -2.23331138e-01
-1.38763535e+00 2.46582672e-01 4.38498825e-01 9.91037905e-01
9.05619860e-01 1.55694872e-01 -2.60313928e-01 7.53182948e-01
-1.38434529e-01 5.45066893e-01 6.24355376e-01 -9.76842582e-01
4.26332563e-01 4.32284057e-01 -3.49318832e-02 -4.29684877e-01
-2.61423886e-01 -6.28890455e-01 -6.06796980e-01 3.38229030e-01
7.47218013e-01 -2.10784346e-01 -1.24766767e+00 1.85484147e+00
1.49096578e-01 2.79902041e-01 1.84736744e-01 7.82824934e-01
7.89586067e-01 3.83599073e-01 2.01870978e-01 1.79284215e-01
9.65516031e-01 -1.11333799e+00 -2.54322916e-01 -4.15577352e-01
7.25825429e-01 -3.30229163e-01 1.53742015e+00 5.13311982e-01
-7.20367610e-01 -7.38956213e-01 -9.95731533e-01 6.94459081e-02
-5.90478778e-01 -7.08610984e-03 8.77066970e-01 3.54468226e-01
-1.24202919e+00 4.22220796e-01 -7.12696671e-01 -3.01331490e-01
8.33256662e-01 1.09010443e-01 -6.24051392e-01 -4.81793225e-01
-9.44068611e-01 3.96101266e-01 3.89081299e-01 1.36983559e-01
-1.43740857e+00 -1.05459213e+00 -8.06505799e-01 -5.48887491e-01
5.37201583e-01 -2.32837811e-01 1.18982303e+00 -1.66457510e+00
-8.60368431e-01 1.15037727e+00 -2.61295401e-03 -8.33088815e-01
8.76969814e-01 -1.38924211e-01 -3.73324782e-01 5.12488186e-02
4.19100374e-01 1.29856479e+00 7.05110312e-01 -1.49582505e+00
-7.01679170e-01 -1.63727492e-01 1.60816029e-01 2.00219527e-02
-2.41091147e-01 -3.23520184e-01 -6.55104578e-01 -5.04615724e-01
-3.09493124e-01 -1.01051152e+00 1.49002615e-02 1.20624397e-02
-3.88794422e-01 -7.14848340e-02 7.47816622e-01 -3.04508060e-01
7.53569126e-01 -2.19450355e+00 9.86889564e-03 -2.59987503e-01
1.58859387e-01 2.74760932e-01 -4.76493716e-01 3.19525272e-01
-1.94544613e-01 2.63598830e-01 -3.64686251e-01 -7.50458062e-01
-2.03919411e-01 3.21553349e-01 -4.74807769e-01 4.48789477e-01
3.74087900e-01 1.20452845e+00 -1.08113086e+00 -3.17106396e-01
4.73365486e-02 1.51225403e-01 -5.17933309e-01 -2.22492870e-02
-5.79942167e-01 5.68160713e-01 1.53573036e-01 8.03698957e-01
4.99504298e-01 -3.77556920e-01 -1.69918403e-01 3.77149582e-02
-3.37122977e-01 -2.95342177e-01 -7.57286727e-01 1.97580183e+00
-2.50849485e-01 1.02033091e+00 -1.63631946e-01 -1.01403189e+00
5.16141832e-01 2.10737307e-02 4.07481879e-01 -1.06995177e+00
-1.71874374e-01 6.51544407e-02 -1.30880415e-01 -4.48449373e-01
3.47621024e-01 1.91517621e-02 -2.60903081e-03 3.80063891e-01
3.47774088e-01 -2.13011950e-01 5.29114485e-01 3.88771296e-01
1.13002908e+00 4.18459356e-01 -3.83842617e-01 -2.00434655e-01
2.96383113e-01 6.99904501e-01 7.92657912e-01 9.60467100e-01
-5.50171673e-01 7.89193511e-01 3.96048635e-01 -5.84663272e-01
-1.17743146e+00 -1.02717638e+00 -1.59237698e-01 1.30412161e+00
-6.46119704e-03 -9.21408758e-02 -5.37185490e-01 -1.02520382e+00
2.00164631e-01 8.53157222e-01 -9.73869741e-01 5.24285398e-02
-3.50102276e-01 -7.75131404e-01 7.12677658e-01 4.61464047e-01
6.18202507e-01 -1.13727045e+00 -5.10350704e-01 6.88280761e-02
-6.16675057e-02 -1.05374718e+00 -1.84005111e-01 2.96835691e-01
-3.69385242e-01 -1.06221557e+00 -7.64367700e-01 -6.51876211e-01
3.97516221e-01 3.70379537e-01 1.45526087e+00 -1.85166355e-02
-5.36709487e-01 7.07814813e-01 -4.48618948e-01 -5.61088920e-01
-2.87485987e-01 -1.32971704e-01 -2.59366375e-03 1.27669781e-01
4.42374438e-01 -4.50409025e-01 -6.29679024e-01 3.27267915e-01
-9.28650379e-01 2.75270343e-01 4.05514091e-01 8.86425138e-01
7.90481091e-01 -9.06479731e-02 7.57110417e-01 -1.17476416e+00
1.97605833e-01 -8.11046004e-01 -6.35749102e-01 4.31433231e-01
-6.40486777e-01 -1.58480972e-01 6.33601069e-01 -8.62467587e-01
-9.37367380e-01 -1.20119313e-02 2.21630201e-01 -7.84844697e-01
-2.86266714e-01 5.22832692e-01 2.78360695e-01 2.35965356e-01
1.05532634e+00 3.23059320e-01 -1.54637665e-01 -2.05725461e-01
6.73496962e-01 4.62108046e-01 9.78825569e-01 -7.11181223e-01
9.33967173e-01 8.01197410e-01 -5.10766685e-01 -5.42587817e-01
-1.04884458e+00 -4.54863876e-01 -7.45484591e-01 -4.76588339e-01
8.20821702e-01 -1.35756707e+00 -6.69998586e-01 6.29238844e-01
-6.25995159e-01 -1.02535415e+00 -5.82366407e-01 3.66409779e-01
-3.15767288e-01 -9.47903171e-02 -3.51017803e-01 -7.13750720e-01
-7.13519230e-02 -1.05025554e+00 1.02431297e+00 2.68965036e-01
2.66962778e-02 -1.06500614e+00 7.04222452e-03 5.46839952e-01
4.09082830e-01 3.32563519e-01 4.22413975e-01 -6.39065146e-01
-5.67023396e-01 -2.60022003e-02 -2.24832937e-01 6.17731452e-01
-1.37793332e-01 1.66533589e-01 -1.21160102e+00 -4.50843364e-01
-2.66478568e-01 -1.12804794e+00 1.03360868e+00 1.24103636e-01
1.12908876e+00 2.93606520e-02 -3.60918492e-02 7.28740036e-01
1.58135056e+00 1.00231551e-01 4.85419542e-01 5.35257161e-01
7.57606864e-01 6.63952172e-01 7.15250850e-01 1.40410841e-01
5.91581464e-01 1.76751554e-01 3.40854228e-01 -2.15611875e-01
-3.68732661e-01 -2.52469838e-01 4.30072844e-01 5.15854716e-01
1.63102120e-01 -3.25922251e-01 -1.36456573e+00 1.02981532e+00
-1.81696534e+00 -1.02216494e+00 -5.54605089e-02 2.18851233e+00
1.04570842e+00 3.44769537e-01 5.85536249e-02 -3.77484411e-02
4.24766660e-01 -7.05830695e-04 -8.33511531e-01 2.44359881e-01
-6.44687474e-01 6.88298345e-02 6.98068023e-01 2.71756440e-01
-1.17394412e+00 1.16589987e+00 5.94005775e+00 9.08055544e-01
-1.49713266e+00 2.07829937e-01 8.40794086e-01 -5.09535551e-01
-4.40853298e-01 7.20617995e-02 -7.26416051e-01 4.04280335e-01
9.01804447e-01 -1.37710050e-02 4.66462314e-01 7.22550929e-01
-7.56846592e-02 -3.19815934e-01 -1.33421516e+00 1.08092320e+00
1.58544481e-01 -1.58105934e+00 -1.13499165e-01 -2.37757042e-01
9.95355725e-01 7.32178986e-01 3.39853734e-01 7.23048031e-01
8.16953480e-01 -1.05935466e+00 1.14080048e+00 5.12803197e-01
1.08592486e+00 -4.40353066e-01 4.83344942e-01 4.83762771e-01
-8.95890951e-01 -3.82225394e-01 -2.54265755e-01 1.25794753e-01
-3.43571976e-02 5.14565766e-01 -7.09406674e-01 4.27687168e-01
9.32694197e-01 1.12028587e+00 -1.25469637e+00 8.84292781e-01
-1.29807979e-01 9.84684169e-01 -2.50963569e-01 2.34686941e-01
6.22445583e-01 -3.11391149e-02 3.68455410e-01 1.25432694e+00
-4.88065109e-02 -3.03351343e-01 3.60793501e-01 8.33242357e-01
-2.73054212e-01 -1.36686563e-01 -7.15748012e-01 -1.36868998e-01
3.61760646e-01 1.07705534e+00 -6.74626350e-01 -3.97977531e-01
-5.33821523e-01 8.59680474e-01 5.29227436e-01 6.65473938e-01
-7.85596907e-01 9.78129804e-02 3.90691400e-01 6.42730147e-02
2.66903251e-01 -1.41606674e-01 -1.46473929e-01 -1.31704485e+00
3.97149138e-02 -1.09777164e+00 4.93775457e-01 -9.55838978e-01
-1.53909874e+00 6.96420908e-01 -9.22331307e-03 -1.21774137e+00
-2.00567320e-01 -2.62490988e-01 -3.99610281e-01 5.37422240e-01
-1.68256831e+00 -1.60298049e+00 -4.52145606e-01 9.38381791e-01
7.39165306e-01 -3.49201173e-01 5.29791474e-01 2.78448850e-01
-5.75085044e-01 7.02419817e-01 1.42943591e-01 3.99804950e-01
9.95676875e-01 -1.35104847e+00 3.58199567e-01 9.51452136e-01
6.23904586e-01 4.01261747e-01 5.86613595e-01 -6.41921520e-01
-1.35147107e+00 -1.41120732e+00 4.19337839e-01 -7.70347238e-01
8.89342129e-01 -6.64669096e-01 -8.80906105e-01 9.68089640e-01
1.54719621e-01 4.91246462e-01 4.58459347e-01 1.57376572e-01
-7.74424911e-01 -3.43203247e-01 -9.14636731e-01 5.50996602e-01
1.24152505e+00 -7.09483027e-01 -4.23498631e-01 6.07309699e-01
9.22533929e-01 -9.85290930e-02 -5.69697976e-01 2.71498650e-01
3.94973427e-01 -1.07221365e+00 8.16185534e-01 -7.21780896e-01
4.92982715e-01 -3.11342806e-01 -2.24817306e-01 -1.15575194e+00
4.86296602e-02 -5.76787829e-01 4.59359884e-01 1.46787870e+00
6.90226614e-01 -4.48739022e-01 7.88015723e-01 5.12251675e-01
-7.89575279e-02 -5.03355801e-01 -7.63886690e-01 -1.00976372e+00
2.50088811e-01 -1.03017950e+00 3.28067958e-01 1.18080974e+00
-4.79166716e-01 1.87245250e-01 -2.22524866e-01 -3.01929638e-02
7.89021730e-01 9.66958478e-02 1.00852835e+00 -1.03677070e+00
-3.23768497e-01 -2.85046935e-01 -3.95353794e-01 -6.95484161e-01
2.36348063e-01 -1.05728233e+00 5.81944585e-02 -1.10876477e+00
3.03180039e-01 -8.69564354e-01 -2.91378021e-01 8.42371583e-01
-1.09329686e-01 5.97746015e-01 2.45279208e-01 4.36142832e-01
-9.43878472e-01 4.21700239e-01 1.01426995e+00 -3.35845262e-01
-2.99999744e-01 -4.00106847e-01 -5.32428682e-01 5.37379622e-01
5.05197108e-01 -4.66653645e-01 -7.15887129e-01 -7.64794469e-01
2.51588106e-01 -3.41092885e-01 6.51171565e-01 -1.33704114e+00
3.81541789e-01 -1.85934991e-01 3.93229216e-01 -5.24206340e-01
1.37899503e-01 -7.18700409e-01 1.32486209e-01 8.47503468e-02
-5.42031288e-01 3.12271472e-02 4.33815449e-01 7.37233102e-01
-3.29516411e-01 1.51105104e-02 6.54909372e-01 -1.83978349e-01
-9.99036133e-01 4.45003569e-01 -6.66171080e-03 5.30032277e-01
1.22095823e+00 -9.35870260e-02 -6.27255917e-01 -3.05400223e-01
-6.27002418e-01 6.88972056e-01 6.47430241e-01 4.92851853e-01
3.89721692e-01 -1.03550529e+00 -1.01019907e+00 7.13055581e-02
5.88170767e-01 2.21680492e-01 3.82540166e-01 6.60159111e-01
-4.16972965e-01 1.03512190e-01 -2.92197257e-01 -1.06056619e+00
-1.02241063e+00 7.68054903e-01 2.37714440e-01 -8.54302943e-02
-6.72376215e-01 1.16533482e+00 2.53992081e-01 -5.34122586e-01
3.04059714e-01 -1.23404056e-01 -8.67539942e-02 3.52297634e-01
7.01137125e-01 -2.65278965e-01 -1.14227228e-01 -4.94710714e-01
-2.87303299e-01 1.90344781e-01 -1.07021704e-01 -3.89575988e-01
1.66057360e+00 -8.72925967e-02 1.40915960e-01 9.43376780e-01
1.13039160e+00 -1.06936127e-01 -1.80589044e+00 -3.43051732e-01
-8.23914707e-02 -2.34199688e-01 -8.83212239e-02 -1.12440479e+00
-1.17880893e+00 8.47631216e-01 5.89060128e-01 -6.93613440e-02
9.83622193e-01 1.65586974e-02 4.69386250e-01 3.74135852e-01
4.70695406e-01 -1.15343237e+00 1.50827259e-01 3.06657493e-01
9.06549931e-01 -1.62043118e+00 -1.36305004e-01 1.84966445e-01
-1.16109908e+00 7.05504715e-01 6.93738222e-01 1.24756806e-01
4.93776560e-01 1.74647182e-01 3.46569568e-01 -1.93243220e-01
-1.11309516e+00 -3.32506001e-01 1.98071718e-01 7.29453027e-01
2.98589431e-02 -6.21993607e-03 3.26024711e-01 4.65624273e-01
-1.72166958e-01 1.34575799e-01 4.11313832e-01 9.29370165e-01
-2.33670883e-02 -7.89884388e-01 -9.55313370e-02 2.30582297e-01
-2.20563576e-01 -2.04639614e-01 -3.03269804e-01 1.00961852e+00
5.25545061e-01 9.33622837e-01 2.25224361e-01 -3.88513148e-01
2.13888317e-01 1.98063239e-01 2.43072227e-01 -6.18604541e-01
-6.67311013e-01 -2.69332141e-01 3.00164055e-03 -6.10270500e-01
-5.91391742e-01 -8.73360813e-01 -1.04312658e+00 -2.47763589e-01
2.35491171e-02 -6.91769496e-02 8.17193568e-01 9.75880980e-01
3.24158281e-01 3.79919231e-01 4.73448843e-01 -7.04207361e-01
-1.59563094e-01 -8.46800566e-01 -3.47084403e-01 9.26635742e-01
4.79790568e-01 -4.91666228e-01 -5.43412149e-01 5.02316236e-01]
|
[9.90649700164795, 1.8781208992004395]
|
2d517f6e-3cfa-4763-8450-a9e534b8c4fb
|
bridging-speech-and-textual-pre-trained
|
2211.03025
| null |
https://arxiv.org/abs/2211.03025v1
|
https://arxiv.org/pdf/2211.03025v1.pdf
|
Bridging Speech and Textual Pre-trained Models with Unsupervised ASR
|
Spoken language understanding (SLU) is a task aiming to extract high-level semantics from spoken utterances. Previous works have investigated the use of speech self-supervised models and textual pre-trained models, which have shown reasonable improvements to various SLU tasks. However, because of the mismatched modalities between speech signals and text tokens, previous methods usually need complex designs of the frameworks. This work proposes a simple yet efficient unsupervised paradigm that connects speech and textual pre-trained models, resulting in an unsupervised speech-to-semantic pre-trained model for various tasks in SLU. To be specific, we propose to use unsupervised automatic speech recognition (ASR) as a connector that bridges different modalities used in speech and textual pre-trained models. Our experiments show that unsupervised ASR itself can improve the representations from speech self-supervised models. More importantly, it is shown as an efficient connector between speech and textual pre-trained models, improving the performances of five different SLU tasks. Notably, on spoken question answering, we reach the state-of-the-art result over the challenging NMSQA benchmark.
|
['Hung-Yi Lee', 'Ann Lee', 'Shinji Watanabe', 'Paola Garcia', 'Dongji Gao', 'Holam Chung', 'Chan-Jan Hsu', 'Jiatong Shi']
|
2022-11-06
| null | null | null | null |
['spoken-language-understanding', 'spoken-language-understanding']
|
['natural-language-processing', 'speech']
|
[ 5.07433295e-01 6.12923324e-01 4.66130339e-02 -7.76953101e-01
-1.10747409e+00 -2.95844883e-01 9.55994010e-01 1.43225923e-01
-4.44225997e-01 3.96575511e-01 6.95045888e-01 -3.52579117e-01
1.80636913e-01 -5.75372696e-01 -8.70569229e-01 -2.42943496e-01
3.65187347e-01 4.78341907e-01 2.35349283e-01 -3.57649088e-01
-1.15179777e-01 -2.92004079e-01 -1.83778071e+00 6.93835676e-01
8.97914350e-01 1.07821047e+00 4.82536942e-01 5.72726667e-01
-8.85201037e-01 1.12136829e+00 -5.27255058e-01 1.13031335e-01
-4.38893169e-01 -7.13362277e-01 -1.19690740e+00 3.13997328e-01
4.02049124e-01 -1.00654550e-01 -3.40486825e-01 7.88122833e-01
3.03399354e-01 1.10754371e-01 5.85357845e-01 -9.12416160e-01
-5.84663272e-01 1.23222649e+00 2.15129063e-01 -2.31804162e-01
6.69672787e-01 -1.69656143e-01 1.17872703e+00 -1.02408338e+00
4.86271054e-01 1.45102096e+00 2.41642326e-01 9.45226133e-01
-1.08453286e+00 -1.88424855e-01 2.36645699e-01 3.67874831e-01
-1.17320085e+00 -1.02248394e+00 6.69941127e-01 -1.03638887e-01
1.25644898e+00 3.45401347e-01 -2.88334605e-03 1.39670205e+00
-5.77345371e-01 1.39800477e+00 1.10374892e+00 -8.25907588e-01
4.55312937e-01 4.87547934e-01 5.12439907e-01 3.50782424e-01
-5.32941520e-01 -2.90021360e-01 -9.63642776e-01 1.75644726e-01
1.35416791e-01 -2.38465309e-01 -3.58124822e-01 -6.58095777e-02
-1.07874966e+00 6.53451383e-01 1.93860486e-01 6.80123985e-01
-1.98492080e-01 -1.25052407e-01 5.66995025e-01 2.79757798e-01
6.62836909e-01 1.54059306e-01 -6.26231372e-01 -2.41926923e-01
-9.77958262e-01 -4.63144571e-01 9.90318954e-01 1.00026512e+00
7.21320987e-01 1.56044826e-01 -2.67800719e-01 1.25260675e+00
5.74949265e-01 6.99762762e-01 6.78693831e-01 -6.09097779e-01
4.81625468e-01 6.01998448e-01 -4.20702040e-01 -2.83020049e-01
-2.53008157e-01 -3.98452014e-01 -6.14114881e-01 -4.57369357e-01
2.78459698e-01 1.97803471e-02 -1.00747979e+00 1.70410943e+00
6.42896742e-02 1.55975625e-01 8.49503517e-01 7.22554743e-01
1.35926569e+00 9.39750373e-01 2.28678912e-01 -1.19614802e-01
1.48164380e+00 -1.23445094e+00 -1.11088204e+00 -5.56169927e-01
9.24965501e-01 -5.29164910e-01 1.43541682e+00 1.48169115e-01
-8.93365681e-01 -6.38447881e-01 -9.79957998e-01 -1.23010434e-01
-7.52251446e-01 3.52713645e-01 1.19482763e-01 6.46089911e-01
-1.06136334e+00 2.23789081e-01 -4.74440515e-01 -7.41668284e-01
1.43089026e-01 -1.68184321e-02 -2.89525479e-01 -3.02363597e-02
-1.61695457e+00 9.18919265e-01 4.67948973e-01 2.27563411e-01
-9.09910023e-01 -5.49329698e-01 -1.24474037e+00 2.54188299e-01
9.26225483e-01 -4.27841723e-01 1.51991498e+00 -1.00422609e+00
-1.98864007e+00 8.93031597e-01 -5.53884923e-01 -8.30022693e-01
1.08005762e-01 -2.58795351e-01 -3.95698756e-01 2.68423647e-01
-1.30585238e-01 5.31119347e-01 8.50766778e-01 -1.27991951e+00
-4.53432769e-01 -3.88776451e-01 -2.33613595e-01 1.40775055e-01
-5.73135793e-01 -2.13135276e-02 -1.75256804e-01 -3.90260667e-01
2.03288913e-01 -3.90314162e-01 2.02036634e-01 -7.23231733e-01
-4.06204969e-01 -7.76133120e-01 9.08683479e-01 -7.65718818e-01
1.16932738e+00 -2.15257192e+00 2.01204970e-01 -2.22075880e-02
-1.61447197e-01 4.09394085e-01 -4.27883208e-01 7.29415357e-01
-8.39018263e-03 -1.06917219e-02 -3.65424901e-01 -9.29856181e-01
4.34757143e-01 6.12003803e-01 -7.19493985e-01 -1.35272339e-01
3.67475599e-01 1.03239524e+00 -9.30565536e-01 -3.02118838e-01
4.91255909e-01 2.78646380e-01 -1.02951467e-01 5.51757455e-01
-5.84642947e-01 3.84700954e-01 -3.48206669e-01 3.57904166e-01
3.75290334e-01 -1.27790183e-01 3.25001568e-01 -1.04485471e-02
-2.16594171e-02 9.83975410e-01 -8.01436961e-01 2.06114411e+00
-8.44661593e-01 4.21124697e-01 1.49443492e-01 -1.35163140e+00
1.04166782e+00 6.77234232e-01 6.79532513e-02 -9.77293849e-01
2.12296009e-01 5.12873828e-01 -3.75782281e-01 -6.30672038e-01
3.06897789e-01 -2.46833965e-01 -4.39706184e-02 4.03353423e-01
6.51136398e-01 -4.13489759e-01 5.38150594e-02 3.84767532e-01
9.24042702e-01 6.58131763e-02 1.43541053e-01 -1.58435613e-01
1.00347519e+00 -2.78469861e-01 -2.61432737e-01 8.77691686e-01
2.24581771e-02 7.51470923e-01 2.78254062e-01 2.21584111e-01
-6.02157891e-01 -1.04854846e+00 -4.80243125e-06 1.38823891e+00
-2.78305382e-01 -7.10356891e-01 -1.08322370e+00 -8.83303523e-01
-3.03609341e-01 1.15085328e+00 -4.01522219e-01 -2.79531591e-02
-2.92492002e-01 -1.51376337e-01 6.80591643e-01 4.77731019e-01
4.06598032e-01 -1.27710640e+00 7.76144769e-03 2.40410015e-01
-3.58887434e-01 -1.74206281e+00 -3.63546103e-01 2.63313115e-01
-6.05952919e-01 -7.08569407e-01 -7.91293979e-01 -8.86252701e-01
4.51052189e-01 1.71565846e-01 9.96910751e-01 -1.05458580e-01
2.38054797e-01 8.74941349e-01 -9.43502128e-01 -2.88752079e-01
-9.32627201e-01 3.26416045e-01 1.02755642e-02 4.78172690e-01
3.99735272e-01 -3.54776859e-01 3.43830548e-02 2.24698573e-01
-1.18278253e+00 -6.15011975e-02 4.59221900e-01 7.23691940e-01
3.70709360e-01 -5.43350339e-01 1.01349318e+00 -7.66589522e-01
5.30222654e-01 -3.44363451e-01 -1.26672313e-01 6.33304656e-01
-2.17627868e-01 3.67409855e-01 7.12109745e-01 -1.98995724e-01
-1.17837358e+00 2.91604567e-02 -5.51036000e-01 -2.88721174e-01
-5.93118072e-01 7.51980424e-01 -4.31179702e-01 5.33199489e-01
5.27850568e-01 8.08612168e-01 3.17930311e-01 -6.92294240e-01
6.84182644e-01 1.28517258e+00 3.87273729e-01 -5.50351441e-01
4.19792801e-01 2.87586361e-01 -5.75760782e-01 -1.60620463e+00
-1.18754780e+00 -7.26348042e-01 -6.82544708e-01 -1.44167930e-01
9.05593038e-01 -8.94791186e-01 -4.40126628e-01 5.98734498e-01
-1.30506539e+00 -4.41050321e-01 -3.52993518e-01 3.31141561e-01
-5.87466657e-01 7.35048831e-01 -3.36652279e-01 -1.17298377e+00
-3.72702271e-01 -9.56789553e-01 1.25997698e+00 8.08403865e-02
-3.27908903e-01 -1.05442381e+00 -1.09351836e-01 7.73185372e-01
5.49556732e-01 -6.49918258e-01 6.52610481e-01 -1.29668999e+00
-3.84216338e-01 1.60760060e-01 -1.48881197e-01 9.20722187e-01
2.51853198e-01 -4.40669298e-01 -1.59287369e+00 -2.09585410e-02
2.37367362e-01 -7.51991868e-01 9.66201127e-01 8.77444893e-02
9.12657559e-01 -2.94066668e-01 6.22780398e-02 2.06777249e-02
7.88083911e-01 4.17254902e-02 6.26232684e-01 1.31834066e-02
3.93688947e-01 1.19310081e+00 3.97343904e-01 -3.46510783e-02
5.90992332e-01 5.57390034e-01 1.67955711e-01 8.02594982e-03
-3.48739177e-01 -5.62631547e-01 7.75615335e-01 1.46622014e+00
6.07551873e-01 -7.92033672e-02 -9.42298353e-01 6.44904256e-01
-2.03979468e+00 -5.93005717e-01 -1.01562381e-01 1.94553936e+00
9.68696475e-01 -2.12331451e-02 -1.80011675e-01 8.44171122e-02
3.52306187e-01 4.09689903e-01 -2.76564121e-01 -2.04703256e-01
-3.65016788e-01 3.38051021e-01 -8.90849382e-02 6.85966074e-01
-1.02588332e+00 1.23970437e+00 5.64827633e+00 1.15145123e+00
-9.23078418e-01 4.06122684e-01 3.16520721e-01 2.95628965e-01
-4.82764274e-01 -6.38765171e-02 -7.84118950e-01 2.16969565e-01
1.38250625e+00 -4.39321361e-02 2.07195923e-01 7.39631355e-01
2.73440748e-01 -2.28047878e-01 -1.37612450e+00 9.36731458e-01
4.80479300e-01 -1.19829094e+00 4.23644483e-01 -4.42620754e-01
3.29864770e-01 7.47974440e-02 -2.95058787e-01 5.78573048e-01
-1.18110888e-01 -9.90506470e-01 8.71496260e-01 4.02618080e-01
4.70717877e-01 -2.92733520e-01 8.94701123e-01 6.59309387e-01
-1.02344048e+00 2.33928517e-01 -8.46603438e-02 -1.09541170e-01
3.85147035e-01 2.98830420e-01 -1.05672264e+00 8.50770950e-01
5.30643165e-01 7.53399730e-01 -3.39259684e-01 3.11460048e-01
-4.71247822e-01 9.91792202e-01 -4.32888180e-01 -4.16759551e-01
4.75473106e-01 -1.09831199e-01 5.29094875e-01 1.32415318e+00
3.35983820e-02 -6.30458519e-02 5.16169183e-02 8.91845703e-01
-3.88495228e-03 5.15925825e-01 -6.62679255e-01 -4.47879165e-01
1.56095460e-01 7.64864802e-01 -4.03617382e-01 -5.89947581e-01
-5.75428665e-01 9.81854856e-01 1.90552369e-01 4.41304952e-01
-3.79715502e-01 -2.03565851e-01 3.99178833e-01 -3.40186269e-03
1.13413468e-01 -3.64409447e-01 -1.10750079e-01 -1.32894015e+00
1.64062157e-01 -8.56216311e-01 3.22620183e-01 -8.75497997e-01
-1.39700472e+00 8.84651303e-01 -1.70114204e-01 -9.63993490e-01
-5.72289824e-01 -5.77714264e-01 -2.95631111e-01 7.67714024e-01
-1.77095234e+00 -1.43559945e+00 1.27616376e-02 5.74146152e-01
1.17357123e+00 -2.46739492e-01 9.78946328e-01 9.03270617e-02
-3.55559200e-01 3.28202069e-01 7.97204860e-03 2.54636467e-01
5.98576546e-01 -1.07345366e+00 1.42366275e-01 7.22767830e-01
5.40829599e-01 5.32054007e-01 5.95921099e-01 -3.58010828e-01
-1.52928376e+00 -7.10584998e-01 1.27685964e+00 -6.00398183e-01
1.03834355e+00 -7.58210182e-01 -1.21809351e+00 6.28836811e-01
6.10310555e-01 -4.64764148e-01 6.16148353e-01 2.02327833e-01
-3.11944187e-01 -3.56963165e-02 -5.42907417e-01 4.23259199e-01
8.50136459e-01 -1.23510826e+00 -1.11499381e+00 2.01552317e-01
1.01226032e+00 -4.86578830e-02 -5.45536458e-01 3.52198124e-01
7.17337504e-02 -7.28552163e-01 8.18656266e-01 -7.08753288e-01
2.89236873e-01 -1.32725954e-01 -4.76143360e-01 -1.30081367e+00
7.28023827e-01 -4.96044129e-01 -9.18069575e-03 1.50376570e+00
7.63091028e-01 -7.80065835e-01 3.89301747e-01 2.61786729e-01
-3.32386434e-01 -5.21032155e-01 -1.09924424e+00 -8.07416260e-01
-9.54220220e-02 -9.00660634e-01 1.39185697e-01 7.23201454e-01
3.96442205e-01 9.30408120e-01 -2.14866713e-01 1.01969160e-01
4.98997957e-01 -2.17673197e-01 7.59762406e-01 -9.30653036e-01
-3.50310244e-02 -3.77309024e-01 5.24222925e-02 -1.65296197e+00
7.23568261e-01 -1.05002499e+00 4.34544027e-01 -1.74944711e+00
-8.44527856e-02 3.27509344e-02 -5.87167926e-02 6.30185664e-01
-8.34322651e-04 -3.20405245e-01 2.29708284e-01 -8.46926123e-02
-7.13108838e-01 1.20815277e+00 9.37645376e-01 -3.55161995e-01
-1.40360236e-01 -4.05868236e-03 -2.86328226e-01 5.57384789e-01
5.49936235e-01 -1.33901462e-01 -5.19203484e-01 -3.62777501e-01
-2.01045781e-01 3.30625661e-02 2.07584858e-01 -8.09018493e-01
4.21760708e-01 3.64685953e-01 -4.97745514e-01 -6.78131402e-01
5.44216871e-01 -7.68373430e-01 -6.74725413e-01 1.49303511e-01
-5.14964938e-01 -8.58491361e-01 2.20066786e-01 2.71094322e-01
-8.98488104e-01 -4.46212351e-01 4.73174483e-01 -3.00205946e-02
-9.52826440e-01 -2.90653199e-01 -6.71499252e-01 6.22921996e-02
4.80344921e-01 -2.41388492e-02 -2.84157306e-01 -8.37237298e-01
-1.03216827e+00 4.74414080e-01 -3.20320666e-01 7.97240257e-01
7.18602300e-01 -8.51860225e-01 -6.20288074e-01 2.92065144e-01
3.58363688e-01 -6.57155663e-02 4.63049889e-01 9.81516004e-01
9.84985605e-02 9.64601099e-01 3.52978557e-01 -7.67665565e-01
-1.16238904e+00 2.15862349e-01 2.18368366e-01 -6.24944642e-02
-3.19852680e-01 5.99312723e-01 2.05915481e-01 -1.00085402e+00
6.18979990e-01 -6.47889614e-01 -3.50377977e-01 2.86552846e-01
5.46504676e-01 9.51861814e-02 1.31289706e-01 -5.83009779e-01
-3.87940407e-01 3.65002155e-01 1.30111545e-01 -4.50744957e-01
1.22075629e+00 -5.08509457e-01 -3.22127603e-02 8.23281407e-01
1.24024177e+00 -1.72741815e-01 -8.75819325e-01 -6.96123004e-01
4.52740043e-01 2.72224069e-01 -7.04280511e-02 -7.84717679e-01
-5.04192770e-01 1.23897362e+00 1.14416055e-01 5.79056442e-01
8.32392216e-01 6.21177197e-01 9.14503992e-01 7.09610343e-01
3.59831721e-01 -1.17419755e+00 1.88952640e-01 1.13206780e+00
1.09904826e+00 -1.36580098e+00 -7.94913590e-01 -5.98493576e-01
-7.40865111e-01 1.08862758e+00 4.15501773e-01 3.76629621e-01
6.11200213e-01 1.57389976e-02 3.87007713e-01 5.25215920e-03
-6.41560018e-01 -8.15713704e-01 4.53539163e-01 3.50290298e-01
3.30680221e-01 -6.09844103e-02 -9.47351679e-02 9.45145726e-01
-1.11903548e-01 -5.35795391e-02 2.34965727e-01 7.80584216e-01
-6.41183019e-01 -1.28642249e+00 -2.52465695e-01 2.91474126e-02
3.17691490e-02 -3.57183963e-01 -6.91481590e-01 5.87694049e-01
-4.34177130e-01 1.35732865e+00 -1.48710888e-02 -2.29250982e-01
5.61711907e-01 7.86170542e-01 3.11089516e-01 -9.14498448e-01
-4.31851923e-01 1.46797583e-01 4.39909995e-01 -5.66754818e-01
-6.34971440e-01 -3.01346809e-01 -1.43846714e+00 4.73282039e-01
-2.01189831e-01 4.23563421e-01 7.58212030e-01 1.49434173e+00
2.55922258e-01 5.08349478e-01 6.55035198e-01 -4.73677069e-01
-7.07022846e-01 -1.45239341e+00 -4.40472811e-01 4.31029111e-01
2.96987742e-01 -3.16086441e-01 -6.62639201e-01 1.58153683e-01]
|
[14.010128021240234, 6.984279632568359]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.