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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
101c8f1c-7206-4f02-a7f3-89af9d535cb8
|
human-centered-trust-framework-an-hci
|
2305.03306
| null |
https://arxiv.org/abs/2305.03306v2
|
https://arxiv.org/pdf/2305.03306v2.pdf
|
Human-centered trust framework: An HCI perspective
|
The rationale of this work is based on the current user trust discourse of Artificial Intelligence (AI). We aim to produce novel HCI approaches that use trust as a facilitator for the uptake (or appropriation) of current technologies. We propose a framework (HCTFrame) to guide non-experts to unlock the full potential of user trust in AI design. Results derived from a data triangulation of findings from three literature reviews demystify some misconceptions of user trust in computer science and AI discourse, and three case studies are conducted to assess the effectiveness of a psychometric scale in mapping potential users' trust breakdowns and concerns. This work primarily contributes to the fight against the tendency to design technical-centered vulnerable interactions, which can eventually lead to additional real and perceived breaches of trust. The proposed framework can be used to guide system designers on how to map and define user trust and the socioethical and organisational needs and characteristics of AI system design. It can also guide AI system designers on how to develop a prototype and operationalise a solution that meets user trust requirements. The article ends by providing some user research tools that can be employed to measure users' trust intentions and behaviours towards a proposed solution.
|
['David Lamas', 'Paulo Martins', 'Jose Cravino', 'Sonia Sousa']
|
2023-05-05
| null | null | null | null |
['misconceptions']
|
['miscellaneous']
|
[-1.25434905e-01 6.90044701e-01 -1.53880352e-02 -5.63958704e-01
2.59717077e-01 -3.09442759e-01 5.03979504e-01 5.53751826e-01
-2.07896337e-01 1.60657942e-01 6.62865460e-01 -7.92092144e-01
-1.41755827e-02 -4.27519642e-02 -2.59299636e-01 -1.85508981e-01
5.24399996e-01 -1.04707554e-01 -2.33473077e-01 -4.65810180e-01
7.80095041e-01 2.14809731e-01 -1.37826943e+00 2.11833268e-01
7.99810827e-01 5.50356209e-01 -3.65002342e-02 3.95533830e-01
5.75084507e-01 1.25583935e+00 -2.75032610e-01 -7.16654658e-01
-2.82512844e-01 -3.55635732e-01 -1.09719944e+00 -1.60747826e-01
-1.12196289e-01 -7.22821832e-01 2.35863507e-01 6.77018404e-01
6.30113423e-01 -5.31618357e-01 2.80499369e-01 -1.62753916e+00
-1.01483512e+00 7.10268497e-01 4.34010416e-01 -1.97513565e-01
8.56065691e-01 1.65335000e-01 6.16339743e-01 -5.56965828e-01
3.62430334e-01 1.15410376e+00 1.01583493e+00 3.62094998e-01
-8.36671531e-01 -7.32296944e-01 -3.18871140e-01 2.18741298e-01
-1.02023733e+00 -9.82905805e-01 5.23555219e-01 -5.94465137e-01
1.13002765e+00 5.11902988e-01 1.27374089e+00 1.05707741e+00
2.12699145e-01 5.76804698e-01 1.34836316e+00 -9.92765069e-01
4.17271167e-01 1.26059055e+00 6.25885010e-01 2.42722958e-01
7.95606256e-01 -1.82186943e-02 -1.82398275e-01 -5.99699318e-01
4.44818527e-01 -2.34212905e-01 -2.03404695e-01 -1.80752367e-01
-8.24992657e-01 7.49366641e-01 3.11767876e-01 1.07899070e+00
-5.24619281e-01 -2.41808742e-01 6.80916131e-01 5.76713085e-01
-3.61980521e-03 7.64434695e-01 -2.46696159e-01 -9.04942989e-01
-8.12918022e-02 -4.71932068e-02 1.27004731e+00 7.30016828e-01
8.20946693e-02 -2.64309943e-01 1.12074472e-01 2.77971357e-01
1.03735316e+00 1.29240587e-01 7.35900626e-02 -8.75961363e-01
-8.58198583e-01 8.44119191e-01 7.52049744e-01 -1.09860301e+00
-2.11982086e-01 1.84546690e-02 1.10362336e-01 3.11389029e-01
-1.17878720e-01 -1.21829823e-01 -3.14532846e-01 9.99514937e-01
2.55776942e-01 -3.16606939e-01 2.97738403e-01 9.16478753e-01
7.49190092e-01 2.15052813e-01 3.42168242e-01 -3.57202619e-01
1.26888740e+00 -4.23456103e-01 -1.14223564e+00 -1.57869712e-01
9.05291915e-01 -7.58343935e-01 1.16616368e+00 3.47205400e-01
-1.21603239e+00 -2.49048546e-01 -1.49029112e+00 -2.38501630e-03
-1.58405021e-01 -2.34048754e-01 7.51932323e-01 1.60456252e+00
-1.18393254e+00 1.21127740e-01 -7.22170711e-01 -1.05996633e+00
9.97009724e-02 7.08450079e-02 7.57992044e-02 -8.09981450e-02
-1.08124530e+00 1.65599084e+00 1.67185038e-01 7.78917909e-01
-2.99988300e-01 -2.98829824e-01 -6.62440121e-01 -5.18655062e-01
-2.10184559e-01 -7.62601197e-01 1.27060008e+00 -1.59877813e+00
-1.20883024e+00 5.49955010e-01 5.01636028e-01 -2.48016104e-01
2.61657000e-01 -2.92578131e-01 -4.71823990e-01 -1.86444223e-01
1.11215688e-01 9.07460526e-02 -2.77069118e-03 -1.77472198e+00
-4.42599446e-01 -2.98175365e-01 2.95114815e-01 2.52285469e-02
-5.41569352e-01 7.02070355e-01 3.42607319e-01 6.99070469e-02
-1.81341931e-01 -7.76137352e-01 3.26457694e-02 -2.69157112e-01
4.31782365e-01 -1.02213636e-01 7.27281153e-01 -4.59610373e-01
1.74254572e+00 -1.86651790e+00 -5.29962838e-01 2.38194540e-01
2.63714194e-01 4.33934122e-01 2.86168188e-01 1.36375451e+00
3.07595789e-01 3.67710799e-01 2.22710103e-01 3.08226556e-01
2.61608303e-01 1.80277407e-01 2.25103334e-01 6.11095607e-01
-2.03038514e-01 4.72294122e-01 -1.10994554e+00 -1.57482490e-01
9.17918906e-02 8.55339289e-01 -6.26158938e-02 4.23798263e-01
6.56095803e-01 -7.74575025e-02 -4.30462390e-01 5.88160157e-01
5.15247762e-01 9.54503417e-02 3.66886824e-01 -3.14113319e-01
-7.02731133e-01 5.44638157e-01 -7.03673959e-01 9.67397213e-01
-3.43802720e-01 4.08013016e-01 1.79826960e-01 -4.71688628e-01
9.74029183e-01 1.04919314e+00 2.59106874e-01 -5.69759786e-01
7.03486502e-01 2.99751699e-01 4.37212706e-01 -9.50936615e-01
1.19714409e-01 -1.17256336e-01 -1.99660435e-02 9.32029307e-01
-3.72223645e-01 -1.98427483e-01 -7.28381813e-01 2.55924314e-01
9.49275494e-01 2.13256478e-01 4.90367085e-01 -6.34984374e-01
2.20874533e-01 -1.46349579e-01 2.78563857e-01 4.65640008e-01
-7.15816796e-01 -2.85369754e-01 4.46100026e-01 -7.68736005e-01
-1.03073394e+00 -7.02560395e-02 -2.32110381e-01 5.51715791e-01
1.71511412e-01 -4.79332536e-01 -9.66427386e-01 -4.58403677e-01
-2.70416021e-01 1.34389746e+00 -8.27584386e-01 -5.87546706e-01
5.45631409e-01 7.65446350e-02 5.26398420e-01 1.76399246e-01
3.73143524e-01 -1.15009701e+00 -1.88738680e+00 2.42204726e-01
8.72266665e-02 -7.38867223e-01 1.89756706e-01 -1.71979502e-01
-5.05579948e-01 -1.02970564e+00 1.60552010e-01 -6.95466697e-01
7.74803281e-01 4.81020153e-01 7.81203091e-01 1.06760240e+00
1.99328199e-01 8.15681458e-01 -8.58355701e-01 -9.04980659e-01
-6.82637453e-01 -5.89787722e-01 7.06312060e-02 -5.12896538e-01
8.80738616e-01 -2.44253501e-01 -4.16482478e-01 6.12281799e-01
-8.24799716e-01 2.20826432e-01 3.96439970e-01 6.27330244e-01
-7.72191226e-01 -1.02340423e-01 8.81938756e-01 -8.41335058e-01
1.33117592e+00 -6.84508801e-01 2.90373325e-01 4.23767895e-01
-1.37674558e+00 -6.29053354e-01 -2.53608167e-01 -1.49249434e-01
-1.15204120e+00 -5.07819772e-01 1.17401592e-01 4.35751438e-01
-1.90192714e-01 6.77460968e-01 -6.95729181e-02 -4.14574832e-01
9.89289820e-01 -6.78329945e-01 6.41014576e-01 3.62871230e-01
7.99332038e-02 1.38082910e+00 -1.85346812e-01 -6.55113280e-01
1.07112154e-01 -4.65359259e-03 -9.71660078e-01 -6.98753595e-01
-1.66744009e-01 -4.54801828e-01 -5.70903242e-01 -8.42776716e-01
5.45791328e-01 -7.70701289e-01 -9.72472131e-01 1.50955496e-02
-9.31961596e-01 -3.97638798e-01 2.46377945e-01 4.89815265e-01
-4.62185413e-01 7.92198852e-02 -3.63007843e-01 -1.74841082e+00
-6.51513517e-01 -9.94564831e-01 1.82660490e-01 2.21919984e-01
-1.08536482e+00 -1.02520835e+00 1.14873588e-01 6.94330812e-01
6.88864291e-01 4.01035994e-01 6.03647053e-01 -4.97449368e-01
2.46131495e-01 -4.82408881e-01 4.21941280e-02 4.71750408e-01
2.05009103e-01 3.69721025e-01 -9.06509221e-01 -2.63718933e-01
6.18509531e-01 -7.55699694e-01 -5.53149760e-01 -2.52312660e-01
-4.60805476e-01 -9.39719856e-01 -1.56079814e-01 -7.07410574e-01
1.43210208e+00 7.92524219e-01 9.75711644e-01 5.62501371e-01
3.38393658e-01 9.74696994e-01 1.11878467e+00 6.70721591e-01
7.02044249e-01 2.56950110e-01 9.87951383e-02 -6.62297681e-02
7.06682026e-01 2.77282757e-04 4.86337125e-01 7.25596309e-01
-1.50805369e-01 3.75186741e-01 -1.33378828e+00 6.53034210e-01
-2.06054640e+00 -6.34250462e-01 -5.76519549e-01 2.07142305e+00
6.18426740e-01 3.34310144e-01 2.75584668e-01 3.01689208e-01
2.14391708e-01 -5.57205796e-01 6.24639988e-02 -1.27397048e+00
7.45036304e-01 -1.82388529e-01 5.46483509e-02 5.36102772e-01
-2.81506777e-01 7.44413197e-01 6.23071575e+00 -6.45912170e-01
-6.09763622e-01 2.40810305e-01 6.18080676e-01 5.23466229e-01
-9.68747735e-01 4.67768162e-01 2.02479467e-01 -9.07659009e-02
1.24501061e+00 -3.18368077e-01 4.87046316e-02 6.21686399e-01
5.66380620e-01 -4.30513024e-01 -1.04446781e+00 2.87375659e-01
1.04277536e-01 -5.93145549e-01 -7.55436778e-01 -1.99141860e-01
3.67285818e-01 -5.36743760e-01 -9.88737941e-02 2.24346034e-02
2.86583841e-01 -8.06714535e-01 8.86531532e-01 5.16813099e-01
1.03907622e-01 -8.55918527e-01 1.41175091e+00 1.67056337e-01
-2.16865137e-01 -1.90276846e-01 -2.18279988e-01 -9.45417166e-01
3.84845175e-02 -2.96236545e-01 -1.28017163e+00 -7.83656538e-02
8.06094050e-01 2.35786557e-01 -4.10478950e-01 6.83486938e-01
1.47708505e-01 4.60313261e-01 -9.71485227e-02 -4.05359060e-01
2.12891340e-01 -3.36378008e-01 -9.10191238e-02 1.04002500e+00
-9.38041136e-02 5.64763069e-01 -4.94665176e-01 5.88725984e-01
8.90305459e-01 2.07662821e-01 -1.13978434e+00 -2.46734500e-01
9.51863885e-01 1.11321247e+00 -6.21888101e-01 -1.03810586e-01
-7.16482937e-01 8.06679368e-01 -2.72580713e-01 3.07265937e-01
-5.43324888e-01 -5.70677780e-02 5.57159603e-01 4.55747664e-01
-2.20193118e-01 -2.86561679e-02 -7.41399884e-01 -2.76232779e-01
-3.26839209e-01 -1.20292687e+00 -3.10553491e-01 -8.13536346e-01
-7.86527932e-01 4.44916397e-01 1.36718035e-01 -7.18395174e-01
1.56109771e-02 1.09428406e-01 -4.15793628e-01 5.49074233e-01
-5.19074321e-01 -1.87267041e+00 -2.15917319e-01 4.87632438e-04
-2.97002316e-01 3.66094500e-01 1.10889924e+00 3.75879966e-02
-3.76288056e-01 4.36374724e-01 -4.34226602e-01 -6.34269118e-01
8.32247376e-01 -7.81369865e-01 3.31659168e-01 5.49950361e-01
-5.55978000e-01 1.15925169e+00 1.01976550e+00 -9.28606093e-01
-1.70702696e+00 8.85300115e-02 1.29322386e+00 -7.88676023e-01
7.34522462e-01 -1.77075639e-01 -9.70145106e-01 6.52633429e-01
8.30903590e-01 -6.31853640e-01 1.31354868e+00 2.51185864e-01
-5.13063893e-02 -6.26776069e-02 -1.79446530e+00 7.52665401e-01
5.66338718e-01 -7.36224294e-01 -4.40562755e-01 -2.10670158e-01
4.50643510e-01 1.44605443e-01 -1.29978848e+00 1.91751361e-01
1.38807678e+00 -1.60495591e+00 1.90354988e-01 -3.13691258e-01
2.45571379e-02 3.86389829e-02 1.17979020e-01 -7.20813215e-01
-5.00975251e-01 -1.00678873e+00 4.97689009e-01 1.42992234e+00
3.54206473e-01 -8.74153912e-01 1.00430146e-01 2.00390291e+00
-2.74540007e-01 -6.69834554e-01 -6.50501311e-01 5.03680885e-01
-3.90414804e-01 -5.43706656e-01 4.60201442e-01 1.42621160e+00
1.19152391e+00 1.33736372e-01 -1.92270964e-01 3.38059925e-02
1.11725070e-01 -1.18121839e+00 8.33587766e-01 -1.24878669e+00
3.04364562e-01 -1.90548837e-01 -4.93392229e-01 1.15953699e-01
-4.13645566e-01 1.47628024e-01 4.00312133e-02 -1.71580386e+00
1.17948167e-02 -4.97235417e-01 -8.64546224e-02 4.84851539e-01
1.33334547e-01 -4.51589614e-01 2.20012873e-01 1.45972505e-01
-1.58564359e-01 1.68651059e-01 8.05962205e-01 6.74103916e-01
-3.60122502e-01 -2.07931891e-01 -1.69613028e+00 4.80367750e-01
7.89898098e-01 -2.45889693e-01 -7.53686070e-01 3.35812010e-02
8.67683053e-01 6.40351325e-02 2.95914143e-01 -9.35892224e-01
2.91791499e-01 -3.67814034e-01 -1.66837290e-01 1.24489792e-01
-2.91594267e-01 -1.78992987e+00 8.61682713e-01 6.00218654e-01
-2.49593243e-01 2.85229743e-01 2.72943854e-01 -3.22860330e-01
2.42011905e-01 -5.39457560e-01 2.79538751e-01 2.59908527e-01
4.54232357e-02 -6.68518245e-01 -6.77226245e-01 -9.85514402e-01
1.34967518e+00 -8.07005823e-01 -1.14004627e-01 -5.80249965e-01
3.95449027e-02 2.15400785e-01 1.10496533e+00 4.97884959e-01
8.14829051e-01 -1.26963007e+00 -2.11631536e-01 2.54747391e-01
2.84722924e-01 -6.62543178e-01 6.19548969e-02 9.27903414e-01
-6.76823556e-01 3.17392498e-01 -7.12285399e-01 9.00142044e-02
-1.20132172e+00 4.30284709e-01 1.49451926e-01 7.98109591e-01
-2.36797929e-01 6.33605003e-01 -3.74555945e-01 9.74958241e-02
2.66635507e-01 -1.79964423e-01 -2.60637790e-01 1.55855436e-02
3.61927778e-01 5.10129869e-01 -2.18996361e-01 -7.75317311e-01
-3.69490504e-01 -1.72933146e-01 -2.49968871e-01 -5.52038252e-01
1.11462843e+00 -4.03490275e-01 -4.14348841e-01 8.99870813e-01
6.68521345e-01 -2.91187286e-01 -9.91841376e-01 5.45108557e-01
1.97556078e-01 -7.70414770e-01 1.31110713e-01 -1.17494655e+00
-7.24147484e-02 4.39085662e-01 9.26132202e-01 7.08267093e-01
9.36735988e-01 -1.77706540e-01 3.01793069e-01 1.81957558e-01
3.87104005e-01 -1.44607246e+00 -2.61561751e-01 -2.36178741e-01
1.28422701e+00 -1.04751730e+00 4.64884415e-02 -1.10560007e-01
-1.38637578e+00 1.13842690e+00 6.71765566e-01 3.74848276e-01
9.20415342e-01 3.69594485e-01 4.62912828e-01 -4.77140009e-01
-7.91610122e-01 2.14687377e-01 -2.37574503e-01 9.40539777e-01
1.31759119e+00 2.63563991e-01 -1.07909024e+00 8.83379459e-01
1.51577950e-01 4.63081121e-01 8.24787319e-01 1.31398010e+00
-4.77837473e-01 -1.29996502e+00 -3.17831904e-01 1.44123649e-02
-3.29975545e-01 3.58508170e-01 -1.03404939e+00 8.06845427e-01
7.80543387e-02 1.48479128e+00 -4.14580494e-01 -8.79234731e-01
5.99840760e-01 -1.98508337e-01 1.57212585e-01 -1.94883779e-01
-1.44716239e+00 -1.10757155e-02 8.03912699e-01 -1.50921956e-01
-6.18210852e-01 -7.16550529e-01 -8.27620566e-01 -4.89253551e-01
-5.72344720e-01 4.27753448e-01 8.64727199e-01 7.32697010e-01
3.66806298e-01 -8.22425932e-02 5.37359893e-01 -1.22123450e-01
-2.14028582e-01 -1.27703834e+00 2.27200314e-02 1.39776975e-01
2.54542530e-01 -3.43515694e-01 -2.70687491e-02 -1.63010582e-01]
|
[9.085637092590332, 6.321040630340576]
|
35c78124-4d50-4447-b859-2b2096904034
|
improve-text-classification-accuracy-with
|
2212.07649
| null |
https://arxiv.org/abs/2212.07649v1
|
https://arxiv.org/pdf/2212.07649v1.pdf
|
Improve Text Classification Accuracy with Intent Information
|
Text classification, a core component of task-oriented dialogue systems, attracts continuous research from both the research and industry community, and has resulted in tremendous progress. However, existing method does not consider the use of label information, which may weaken the performance of text classification systems in some token-aware scenarios. To address the problem, in this paper, we introduce the use of label information as label embedding for the task of text classification and achieve remarkable performance on benchmark dataset.
|
['Yifeng Xie']
|
2022-12-15
| null | null | null | null |
['task-oriented-dialogue-systems']
|
['natural-language-processing']
|
[ 3.04074407e-01 -5.50889075e-02 -4.29745823e-01 -5.34034431e-01
-7.04981238e-02 -4.29512560e-01 8.64478767e-01 5.45988739e-01
-6.94454730e-01 6.02025628e-01 4.31818664e-01 -4.26851571e-01
2.65934944e-01 -6.77323639e-01 3.98337692e-01 -6.56876266e-01
5.07958651e-01 9.17044580e-02 3.05308729e-01 -2.88396627e-01
5.34838080e-01 -8.04815292e-02 -1.15986180e+00 3.12590778e-01
7.81571507e-01 1.02448595e+00 -1.29182473e-01 1.39901459e-01
-6.53762400e-01 8.79783213e-01 -5.96696615e-01 -4.68911827e-01
-1.97295725e-01 -4.18659121e-01 -1.17254531e+00 2.40723486e-03
-7.55534247e-02 1.64700840e-02 -4.41340894e-01 1.12931538e+00
4.46250945e-01 1.38808832e-01 6.05276406e-01 -1.21159458e+00
-5.35237789e-01 7.43084013e-01 -5.17308831e-01 7.31258318e-02
1.57785028e-01 -2.03974187e-01 1.31149888e+00 -6.61952853e-01
2.15806052e-01 1.10153282e+00 4.22147840e-01 5.33326745e-01
-9.07136261e-01 -5.66169918e-01 3.14431250e-01 2.46452168e-01
-1.14123189e+00 -2.05714986e-01 9.90676880e-01 -4.52729166e-01
7.09401250e-01 1.69958770e-01 4.70302582e-01 9.86864209e-01
8.72346759e-03 1.13769436e+00 1.33033323e+00 -5.82653463e-01
1.76798627e-01 2.99009234e-01 6.09597921e-01 6.17114663e-01
4.10000235e-03 -3.98912370e-01 -3.58105451e-01 -1.87978178e-01
2.22480908e-01 9.81597900e-02 -3.05838943e-01 -6.37053102e-02
-1.14121723e+00 1.09418106e+00 2.80093431e-01 5.46363652e-01
4.36351858e-02 -1.72676131e-01 1.11974263e+00 3.33132595e-01
9.89040256e-01 4.58722383e-01 -3.75098944e-01 -4.62532818e-01
-7.23650455e-01 6.18362129e-02 8.44490349e-01 5.94144404e-01
3.62019420e-01 -8.88036340e-02 -2.27027699e-01 1.24726510e+00
1.88641265e-01 4.54837941e-02 7.54004419e-01 -4.57483202e-01
5.88516712e-01 1.17482948e+00 -2.17568845e-01 -1.02954400e+00
-6.90047741e-01 -2.55427510e-01 -9.82083321e-01 -2.67832190e-01
3.44058722e-01 -4.09514308e-01 -3.61984283e-01 1.39160216e+00
4.34185684e-01 -3.41898650e-01 1.47434279e-01 8.51858556e-01
8.41831267e-01 6.88765168e-01 1.34439036e-01 -2.63849199e-01
1.37127972e+00 -1.23999810e+00 -9.85442340e-01 -6.86914325e-02
1.00412071e+00 -7.67380834e-01 9.78842258e-01 1.84032395e-01
-4.23835278e-01 -3.60146314e-01 -1.06582153e+00 3.18691768e-02
-5.84750116e-01 9.40902159e-02 8.52052093e-01 8.28864038e-01
-5.10190487e-01 2.48750001e-01 -3.69401127e-01 -5.78354537e-01
1.56840533e-01 2.29383603e-01 -3.18792403e-01 -9.74680707e-02
-1.43489432e+00 8.79929483e-01 4.64674771e-01 1.95953012e-01
2.20524520e-03 -6.39379770e-02 -7.14265943e-01 7.43294656e-02
5.64700484e-01 -1.51199445e-01 1.22936177e+00 -7.93313861e-01
-1.65572774e+00 7.26959705e-01 4.09394577e-02 -2.15744406e-01
5.25930643e-01 -7.16657117e-02 -4.58224237e-01 -2.37686574e-01
-1.02001399e-01 2.86366373e-01 8.07373524e-01 -8.87854218e-01
-1.02507412e+00 -2.85020113e-01 2.21877113e-01 3.47946078e-01
-1.04652488e+00 8.65367502e-02 -1.76208869e-01 -8.43282461e-01
4.12045568e-02 -9.29633081e-01 -5.54614514e-02 -1.96326554e-01
-5.35195708e-01 -9.67633128e-01 9.83606577e-01 -2.86302924e-01
1.20297396e+00 -2.28376508e+00 -8.91444013e-02 -2.81701863e-01
2.72461921e-01 5.52966058e-01 1.84903741e-01 7.72570133e-01
1.54670417e-01 1.81102246e-01 -2.65250087e-01 -3.03603411e-01
1.93809390e-01 1.39651105e-01 -4.71553445e-01 4.13042217e-01
1.13261707e-01 5.55886269e-01 -9.32672381e-01 -6.70913875e-01
3.45110893e-01 9.18880925e-02 -3.31758708e-01 3.11336607e-01
-2.21939832e-01 2.65498608e-01 -9.31197643e-01 2.35029846e-01
4.18792993e-01 -1.23386718e-01 2.49915451e-01 -1.60281807e-01
-3.61278892e-01 6.06712043e-01 -8.68287086e-01 1.37360346e+00
-5.10715365e-01 7.74578750e-01 -2.05736667e-01 -1.25785649e+00
9.02861297e-01 5.39337575e-01 7.07439125e-01 -5.03559649e-01
3.15799147e-01 1.06339917e-01 2.63327360e-01 -5.51166415e-01
7.23907948e-01 -2.80913621e-01 -1.35062620e-01 8.41473937e-01
-1.81915641e-01 1.10238597e-01 1.40322909e-01 -8.14254861e-03
9.39124346e-01 -2.83448517e-01 2.42863134e-01 -1.94270343e-01
7.59451568e-01 7.50732869e-02 2.55463779e-01 4.18296039e-01
-3.36761594e-01 1.31876662e-01 5.91230333e-01 -5.96327126e-01
-7.54301488e-01 -1.92054555e-01 -4.48903590e-01 1.46710265e+00
2.63381690e-01 -6.21225536e-01 -6.66047215e-01 -1.32285488e+00
1.02019072e-01 3.10485452e-01 -5.61449885e-01 -3.04015547e-01
-4.55853641e-01 -1.05362940e+00 7.15037525e-01 3.84139150e-01
9.43709075e-01 -1.07148409e+00 -3.99712712e-01 3.78880620e-01
-4.04972821e-01 -1.05066526e+00 -5.57980776e-01 3.01132709e-01
-7.21026003e-01 -1.04332709e+00 -7.08942235e-01 -9.53494966e-01
4.41329241e-01 3.53689462e-01 6.68326318e-01 2.78565884e-01
1.44961134e-01 2.02218562e-01 -9.11326230e-01 -3.37724507e-01
-3.87534142e-01 6.01922810e-01 -5.77765703e-02 3.08844745e-01
5.26399255e-01 -1.52004035e-02 -2.49800161e-01 4.56831545e-01
-9.08530772e-01 2.90680140e-01 1.64970398e-01 1.13424420e+00
-1.56184867e-01 4.04300719e-01 1.03661346e+00 -1.19062662e+00
1.00567269e+00 -2.32557192e-01 -2.74274856e-01 2.24965319e-01
-8.93741786e-01 1.74954548e-01 7.69683540e-01 -2.69989103e-01
-1.09576905e+00 -3.10678154e-01 -3.67712915e-01 5.17320752e-01
-3.21442261e-02 8.13643754e-01 -1.67430982e-01 -1.50220498e-01
3.01930308e-01 2.63583392e-01 -1.86046407e-01 -6.56265795e-01
7.83293247e-02 1.30408382e+00 -2.61673421e-01 -4.42590296e-01
3.39801341e-01 2.88318306e-01 -2.01099992e-01 -8.05823267e-01
-1.01312649e+00 -7.38861740e-01 -5.50275803e-01 -6.53214604e-02
5.65627277e-01 -4.47374523e-01 -5.71968019e-01 5.78634381e-01
-1.14805865e+00 -9.55075622e-02 4.02649641e-02 5.39415002e-01
-1.46376863e-01 5.51034451e-01 -5.99861741e-01 -8.42592537e-01
-3.36429000e-01 -1.09134221e+00 6.49722576e-01 1.66726440e-01
-1.20640017e-01 -1.06526923e+00 -3.12522501e-02 4.88695055e-01
4.94586110e-01 -2.93212503e-01 1.26232505e+00 -1.15940273e+00
4.15612534e-02 -5.35288811e-01 -5.08354127e-01 5.00956833e-01
4.06710684e-01 -2.17211589e-01 -1.10913360e+00 -2.17872262e-01
1.56567529e-01 -5.74036300e-01 9.32972908e-01 -1.94479734e-01
1.01405299e+00 -9.64707807e-02 -2.74215698e-01 6.09085010e-03
9.23021615e-01 1.67605668e-01 1.84360579e-01 2.62215614e-01
8.36438298e-01 9.76184964e-01 6.71631336e-01 6.22111857e-01
5.19756258e-01 9.09407556e-01 2.36851484e-01 -2.96770960e-01
1.13362424e-01 -4.94889282e-02 -1.58122852e-02 1.16120517e+00
1.88413352e-01 -4.64281231e-01 -8.40543747e-01 2.62307346e-01
-1.93820763e+00 -5.59618235e-01 -9.91749614e-02 1.85006618e+00
1.09289503e+00 2.18564659e-01 9.17829126e-02 5.03721714e-01
8.71111512e-01 5.98798156e-01 -3.64463329e-01 -3.71786326e-01
4.57921289e-02 -1.85029730e-01 2.65779078e-01 6.04784526e-02
-1.20163381e+00 1.12365377e+00 6.23345995e+00 8.55319858e-01
-1.37716830e+00 1.47381499e-01 5.32298267e-01 5.50545037e-01
9.86478329e-02 -7.97531679e-02 -6.71874523e-01 6.58005297e-01
5.07472575e-01 -2.20296144e-01 4.06147465e-02 6.56125426e-01
1.47963353e-02 -2.92572025e-02 -1.00415182e+00 8.39270294e-01
1.52169675e-01 -7.65272975e-01 -4.98365872e-02 7.34019503e-02
4.37165737e-01 -1.56003073e-01 8.42704996e-02 4.83804613e-01
1.76664844e-01 -8.08736563e-01 3.33106488e-01 -2.05933288e-01
4.60639954e-01 -6.00391090e-01 1.22508526e+00 5.16435862e-01
-9.42700982e-01 2.25852244e-02 -4.65659469e-01 -2.67883748e-01
-5.95106333e-02 6.49232268e-01 -8.44761252e-01 4.13640767e-01
3.73391807e-01 7.96211302e-01 -4.70083565e-01 9.39496160e-01
-1.07472196e-01 9.94128227e-01 -1.13151498e-01 -4.39345300e-01
3.93111646e-01 -1.23788714e-01 1.42319664e-01 1.28508568e+00
-3.28412205e-01 -1.07703749e-02 5.28864622e-01 1.90361276e-01
-3.73260885e-01 6.93117917e-01 -4.50666845e-01 -3.47705185e-01
1.84421897e-01 1.38483989e+00 -8.58521461e-01 -2.16890231e-01
-6.65286660e-01 8.79834771e-01 2.92039365e-01 2.90703215e-02
-4.58764583e-01 -5.15434921e-01 2.20761985e-01 -2.85736352e-01
-1.18070163e-01 -3.01236272e-01 -2.80638278e-01 -1.25866520e+00
5.55415712e-02 -8.01234603e-01 4.04350609e-01 -6.04327470e-02
-1.31547236e+00 4.60454255e-01 -2.39682883e-01 -1.16565883e+00
-2.29446478e-02 -5.33483624e-01 -5.99875927e-01 6.86321914e-01
-1.61272800e+00 -1.09904099e+00 -1.37096748e-01 2.89563477e-01
7.46304631e-01 -2.05418050e-01 8.80959511e-01 4.37130868e-01
-7.19547093e-01 7.93387115e-01 3.78720433e-01 4.61641908e-01
1.09985292e+00 -1.32705927e+00 2.43717507e-01 2.42425561e-01
1.20792724e-01 5.03996968e-01 5.30584991e-01 -3.62787455e-01
-1.15269601e+00 -1.00108004e+00 1.17081451e+00 -2.08854243e-01
7.24601209e-01 -2.77490824e-01 -9.49840903e-01 2.43395150e-01
3.26208830e-01 -1.41481727e-01 8.42918932e-01 3.89402300e-01
-3.64158541e-01 -1.08891353e-01 -9.74176288e-01 6.08731687e-01
5.65333307e-01 -5.64010561e-01 -5.46869516e-01 4.31144357e-01
6.54873490e-01 -1.47920474e-01 -8.89160454e-01 3.42974722e-01
4.00171667e-01 -5.94762921e-01 4.80140895e-01 -4.44785208e-01
1.89520016e-01 7.19936006e-03 4.91221063e-02 -1.44309890e+00
4.52372478e-03 -2.86112100e-01 2.64320821e-01 1.57730007e+00
2.57311910e-01 -8.43312204e-01 7.77392685e-01 6.29941165e-01
-1.29922017e-01 -8.06253731e-01 -9.56870973e-01 -4.72546369e-01
2.89109588e-01 -8.31192732e-02 3.96876782e-01 1.29576766e+00
7.10886359e-01 7.23563373e-01 -5.06615162e-01 -5.07338822e-01
9.87240896e-02 2.61896819e-01 5.27824521e-01 -1.34916389e+00
9.97839496e-02 -7.05159903e-01 -4.08268601e-01 -1.21704578e+00
4.42219615e-01 -9.37660158e-01 3.00400108e-02 -1.48985982e+00
5.99582009e-02 -7.47958541e-01 -3.05043221e-01 6.89987898e-01
-4.35563326e-01 2.18950957e-01 3.63678075e-02 2.99801588e-01
-7.80148685e-01 8.69877517e-01 1.06911922e+00 -5.14152229e-01
1.28126383e-01 1.55338362e-01 -7.23632395e-01 6.48365200e-01
8.96503091e-01 -4.98523057e-01 -3.18350047e-01 -3.29826444e-01
2.48412117e-01 -1.88769072e-01 -2.05331504e-01 -7.32189357e-01
2.02823013e-01 -2.50184774e-01 -2.27627754e-02 -1.51999861e-01
9.85401496e-02 -9.53157425e-01 -6.34470761e-01 3.57631475e-01
-7.65770733e-01 -2.95908581e-02 -1.24919094e-01 4.06710207e-01
-5.71583450e-01 -6.50403500e-01 6.24770463e-01 8.57196674e-02
-6.00111127e-01 1.08573288e-01 -4.24405783e-01 1.01256287e-02
6.06091142e-01 9.38679799e-02 -4.91477937e-01 -8.80463272e-02
-8.29912722e-03 4.06094849e-01 3.44093204e-01 6.76998019e-01
2.74472654e-01 -1.11826944e+00 -6.61495268e-01 -5.46590164e-02
4.73007292e-01 -1.12154007e-01 1.79235693e-02 7.57748961e-01
-2.57010132e-01 5.55820644e-01 1.55414149e-01 -2.90743798e-01
-1.03998971e+00 3.88305128e-01 1.74145564e-01 -3.56433988e-01
-6.91258669e-01 4.10163373e-01 1.05920665e-01 -5.27493954e-01
3.86864543e-01 -4.57966663e-02 -8.34081829e-01 3.59103084e-01
4.36305881e-01 1.98775694e-01 1.23275131e-01 -6.64019585e-01
-8.74306187e-02 3.94405276e-01 -3.94084185e-01 -4.43024468e-03
9.45712209e-01 -1.87826484e-01 -1.23073027e-01 7.70116568e-01
1.20463598e+00 -7.14633241e-02 -6.65100574e-01 -6.39748871e-01
4.15559322e-01 -2.92961597e-01 2.91086137e-01 -7.11754084e-01
-9.69828904e-01 1.13082874e+00 4.70588028e-01 8.41848791e-01
6.31554246e-01 -3.61909688e-01 1.12949944e+00 6.87912524e-01
3.21020961e-01 -1.28250575e+00 6.26280159e-02 9.74721193e-01
3.58438551e-01 -1.50402582e+00 1.03999805e-02 -3.46785247e-01
-6.98230684e-01 1.07317972e+00 5.85176587e-01 3.09996456e-01
7.04023242e-01 -1.43168092e-01 3.10966700e-01 -6.71208650e-02
-5.94038010e-01 -2.10530147e-01 1.30946159e-01 4.30232108e-01
9.67853189e-01 -1.19954824e-01 -8.15311074e-01 5.21856189e-01
1.65249422e-01 -2.46104687e-01 2.66284734e-01 1.01967061e+00
-5.82716465e-01 -1.52882731e+00 1.03277154e-01 7.68887103e-01
-6.94607854e-01 -1.96316630e-01 -4.92296875e-01 5.05602717e-01
-1.13290414e-01 1.19324577e+00 -1.77298188e-01 -4.22020823e-01
2.78621465e-01 2.65905827e-01 1.56895518e-02 -8.30518723e-01
-9.54461932e-01 -3.01339835e-01 2.79363662e-01 2.06486240e-01
-6.02746606e-01 -3.86379600e-01 -1.14978981e+00 -3.90014678e-01
-8.50552738e-01 6.92611039e-01 7.22316086e-01 1.18715036e+00
3.30945253e-02 3.75726849e-01 1.06917000e+00 -5.02250612e-01
-7.05964208e-01 -1.31295645e+00 -7.34961271e-01 4.90681052e-01
6.41087145e-02 -8.15665960e-01 -4.24310595e-01 -1.42021969e-01]
|
[10.559078216552734, 7.547170162200928]
|
21b8b19b-be37-4ad5-ab74-c0d1dd5375e7
|
automatic-discovery-and-optimization-of-parts
|
1412.6598
| null |
http://arxiv.org/abs/1412.6598v2
|
http://arxiv.org/pdf/1412.6598v2.pdf
|
Automatic Discovery and Optimization of Parts for Image Classification
|
Part-based representations have been shown to be very useful for image
classification. Learning part-based models is often viewed as a two-stage
problem. First, a collection of informative parts is discovered, using
heuristics that promote part distinctiveness and diversity, and then
classifiers are trained on the vector of part responses. In this paper we unify
the two stages and learn the image classifiers and a set of shared parts
jointly. We generate an initial pool of parts by randomly sampling part
candidates and selecting a good subset using L1/L2 regularization. All steps
are driven "directly" by the same objective namely the classification loss on a
training set. This lets us do away with engineered heuristics. We also
introduce the notion of "negative parts", intended as parts that are negatively
correlated with one or more classes. Negative parts are complementary to the
parts discovered by other methods, which look only for positive correlations.
|
['Andrea Vedaldi', 'Pedro Felzenszwalb', 'Sobhan Naderi Parizi', 'Andrew Zisserman']
|
2014-12-20
| null | null | null | null |
['l2-regularization']
|
['methodology']
|
[ 4.21993345e-01 5.32132626e-01 -3.68904620e-01 -4.73963350e-01
-1.04402030e+00 -5.37814200e-01 6.11412466e-01 9.38185528e-02
6.79349061e-03 7.44197369e-01 2.86405444e-01 4.58920509e-01
-5.00979982e-02 -6.44986153e-01 -1.03104758e+00 -9.35203731e-01
-1.48723766e-01 7.37805009e-01 2.39217535e-01 -1.07866175e-01
2.64578313e-01 6.67501330e-01 -1.92094421e+00 7.30516851e-01
4.73496944e-01 1.05354691e+00 -1.65074021e-02 4.06520665e-01
-4.27327782e-01 7.11418092e-01 -6.53032780e-01 -3.68725985e-01
5.14530778e-01 -7.69842923e-01 -1.11642647e+00 7.97267497e-01
2.04177439e-01 2.76306123e-01 1.07328832e-01 1.00996006e+00
1.69915289e-01 1.18502110e-01 7.90419877e-01 -1.35241497e+00
-3.70118648e-01 8.42183590e-01 -4.66760874e-01 -2.06746891e-01
1.89558074e-01 -7.18934610e-02 1.32877767e+00 -9.19594288e-01
6.73978984e-01 1.27492058e+00 5.84285736e-01 8.47973526e-01
-1.60989511e+00 -3.01713914e-01 3.48098636e-01 -1.16345465e-01
-1.11901331e+00 -3.75586927e-01 1.05921638e+00 -5.37228942e-01
6.75321579e-01 4.78116661e-01 7.17488825e-01 9.86648381e-01
4.25405465e-02 1.34328806e+00 1.00574124e+00 -4.57525998e-01
5.40608883e-01 6.81992352e-01 4.04352427e-01 5.96650839e-01
3.77045184e-01 -2.11908460e-01 -3.44002008e-01 -3.06275785e-01
4.06273961e-01 3.83992493e-01 -4.56624985e-01 -1.04141021e+00
-7.24418163e-01 9.46386099e-01 6.44738555e-01 4.75206763e-01
-6.96222365e-01 1.55040612e-02 2.62936890e-01 2.86139250e-01
4.60996062e-01 6.81264102e-01 -7.44005978e-01 5.43230236e-01
-8.98905158e-01 9.99815241e-02 8.89840484e-01 1.01124752e+00
1.29914701e+00 -4.59309012e-01 -3.10879529e-01 1.12831569e+00
1.90345973e-01 -3.04035563e-02 6.37502611e-01 -4.62049067e-01
-9.40877479e-03 1.05891156e+00 1.13375317e-02 -3.42669785e-01
-1.78284481e-01 -4.24307495e-01 -6.90787137e-01 2.40212709e-01
9.71946493e-02 6.16093725e-02 -1.25560176e+00 1.73015666e+00
2.08852932e-01 -5.88580191e-01 -2.54425079e-01 6.41554475e-01
5.12246668e-01 3.01061213e-01 4.88178879e-02 -4.36712652e-02
1.35025477e+00 -1.08280742e+00 -3.09141904e-01 -3.11698258e-01
3.25344861e-01 -5.98444462e-01 7.50341654e-01 2.90543467e-01
-1.10933220e+00 -6.20715082e-01 -9.30594921e-01 3.44324678e-01
-6.10020578e-01 2.55630761e-01 6.32740378e-01 6.15741909e-01
-1.07881081e+00 8.18757117e-01 -5.82641363e-01 -2.33799681e-01
9.92192447e-01 4.94415194e-01 -4.30588931e-01 -1.32275790e-01
-5.08126855e-01 5.98643482e-01 3.01329404e-01 -2.54709750e-01
-9.16135728e-01 -4.37859356e-01 -8.61364365e-01 4.06874642e-02
2.33203217e-01 -7.08508790e-01 1.13874638e+00 -1.84123623e+00
-1.06538081e+00 1.07980227e+00 -1.56703487e-01 -5.49496710e-01
3.66237134e-01 -1.40454724e-01 -1.37399761e-02 2.02750936e-01
2.39879280e-01 1.03162992e+00 1.20001233e+00 -1.75730598e+00
-4.91143852e-01 -3.15485984e-01 -1.57376885e-01 4.20994237e-02
7.36132339e-02 -1.88857451e-01 -4.75783050e-01 -5.51826537e-01
4.06193823e-01 -9.01028812e-01 -4.97777373e-01 1.74015328e-01
-5.88910937e-01 -2.19901621e-01 3.99993330e-01 -1.71081141e-01
6.76748633e-01 -1.93902171e+00 -2.28785593e-02 4.46943820e-01
2.23268837e-01 1.55974939e-01 -3.97898257e-01 4.88155544e-01
-4.36781347e-01 1.62869349e-01 -5.00269949e-01 -2.86141336e-01
-1.18115485e-01 2.48040244e-01 -1.99043766e-01 4.17843699e-01
7.72257924e-01 8.86108875e-01 -7.47028828e-01 -3.96164596e-01
-6.13266863e-02 2.20263675e-01 -5.02923369e-01 2.21757516e-01
-3.74811143e-01 1.92294046e-02 -6.45741940e-01 7.93043375e-01
7.53516078e-01 -4.09527570e-01 9.07026902e-02 1.37292221e-02
2.55831808e-01 3.57488066e-01 -1.08055413e+00 1.27002048e+00
-8.58436450e-02 3.18661869e-01 -1.71452034e-02 -1.16129041e+00
9.14712012e-01 2.20847011e-01 7.50985920e-01 -3.03902209e-01
1.71774238e-01 2.12668806e-01 -6.51571676e-02 -3.14475089e-01
2.72773832e-01 -4.08710808e-01 8.65214691e-02 4.41142857e-01
3.79267722e-01 -5.54648489e-02 2.10818246e-01 1.39565571e-04
1.24277973e+00 1.55457810e-01 5.98869205e-01 -2.20318034e-01
4.75405604e-01 3.44122648e-01 5.27298391e-01 7.50195026e-01
-1.65092558e-01 9.84279811e-01 6.51241779e-01 -4.99398112e-01
-1.00823092e+00 -7.00151801e-01 -1.93714812e-01 8.40222716e-01
6.14687148e-03 -2.06173137e-01 -6.26591384e-01 -1.22447217e+00
8.77456069e-02 2.82539040e-01 -1.20377254e+00 -2.54292876e-01
-3.69585276e-01 -6.52984679e-01 -4.10831347e-02 6.19997501e-01
1.29408345e-01 -1.38061666e+00 -5.12245595e-01 1.78164333e-01
1.70119375e-01 -3.93609375e-01 -2.76386380e-01 8.64620566e-01
-1.03131640e+00 -1.32734692e+00 -1.01085269e+00 -8.73926103e-01
1.18404353e+00 5.40714622e-01 1.56560004e+00 2.03753054e-01
-3.67996603e-01 3.23181599e-01 -7.28299856e-01 -5.53379834e-01
-3.32418412e-01 -2.36166790e-01 -3.46382707e-01 3.32909822e-01
3.17077547e-01 -3.17332000e-01 -4.45271909e-01 3.31681788e-01
-8.04028809e-01 -2.12119356e-01 1.24085069e+00 1.13221669e+00
1.01654422e+00 1.02707118e-01 2.80975044e-01 -1.25482881e+00
2.33528972e-01 -5.16662717e-01 -3.05028290e-01 4.04891759e-01
-3.63295794e-01 3.54244471e-01 6.37062609e-01 -6.03492022e-01
-6.39903009e-01 5.16342878e-01 1.46240443e-01 -6.87192142e-01
-2.61630982e-01 -6.62195832e-02 -5.35687506e-01 -9.51760411e-02
6.89970613e-01 3.58791381e-01 6.69815093e-02 -6.80239499e-01
3.01442653e-01 3.64847422e-01 -6.29536659e-02 -4.89933580e-01
6.27557158e-01 4.56989050e-01 -1.47383912e-02 -8.16580892e-01
-7.37425983e-01 -9.07875061e-01 -7.33468592e-01 7.34976754e-02
5.42831481e-01 -8.21988821e-01 -1.54983535e-01 4.98342216e-02
-8.46964478e-01 -1.88761249e-01 -1.08432591e+00 2.65009642e-01
-5.51700592e-01 7.66995698e-02 -1.79718271e-01 -8.79864752e-01
-2.94165194e-01 -9.83993948e-01 1.20349956e+00 2.76581377e-01
-2.54309744e-01 -3.88824850e-01 1.74659029e-01 2.18869094e-02
-1.66848022e-02 -7.51481354e-02 8.04462254e-01 -1.31573713e+00
-6.37820721e-01 -6.30186737e-01 -6.85856268e-02 5.37493765e-01
5.08651845e-02 -2.09969625e-01 -1.24763334e+00 -2.13266090e-01
3.29302400e-02 -4.63443249e-01 1.07831323e+00 4.39329565e-01
1.23890603e+00 -4.97961462e-01 -6.39515281e-01 2.79948980e-01
1.51668346e+00 2.19932824e-01 6.46092951e-01 4.86446992e-02
4.57801402e-01 8.16193044e-01 3.47086519e-01 3.23925644e-01
-2.18683779e-01 2.80184031e-01 3.29525560e-01 -4.24321694e-03
-2.16805711e-01 -3.23789209e-01 7.92452991e-02 1.33134618e-01
3.35578412e-01 -1.96665719e-01 -4.72045720e-01 7.12523580e-01
-1.76810467e+00 -1.06692982e+00 7.27810934e-02 2.26877928e+00
6.00990534e-01 1.54844239e-01 5.05493820e-01 2.04472750e-01
7.80094624e-01 -1.04722604e-01 -4.26272690e-01 -1.27431512e-01
-1.71218127e-01 4.83075917e-01 3.94833773e-01 1.17879629e-01
-1.25437844e+00 6.60831213e-01 6.76744890e+00 5.97020984e-01
-8.94125760e-01 -2.60875583e-01 1.00094533e+00 -2.37683170e-02
-5.36964238e-01 2.33532950e-01 -7.02071965e-01 2.74649650e-01
5.05523205e-01 4.50131506e-01 1.64915130e-01 1.21746361e+00
-3.16141576e-01 -1.53612569e-01 -1.30510008e+00 6.66105747e-01
1.49853617e-01 -1.33154202e+00 2.64475733e-01 1.17630422e-01
1.01824152e+00 -6.06919825e-02 -3.51947695e-01 3.01588297e-01
2.81878561e-01 -8.42810750e-01 8.76388192e-01 6.11507237e-01
1.39044449e-01 -6.08313024e-01 7.38425314e-01 2.69044250e-01
-1.13348067e+00 -1.66953310e-01 -6.52204275e-01 3.26775163e-01
-3.10494512e-01 7.01986372e-01 -6.83357954e-01 3.07516396e-01
6.05508626e-01 5.73800623e-01 -6.42285645e-01 1.44430971e+00
-8.75492319e-02 5.11093557e-01 -2.26750910e-01 -2.01666921e-01
1.59437388e-01 -2.36484572e-01 3.99023712e-01 1.06756747e+00
-2.33492907e-02 -6.25303164e-02 3.72690290e-01 9.97642517e-01
-2.52100229e-01 5.43267839e-02 -7.08818674e-01 1.27777914e-02
2.59731054e-01 1.37614810e+00 -1.04969478e+00 -5.59283376e-01
-3.83843422e-01 1.06114030e+00 1.92746252e-01 1.76061615e-02
-2.95857280e-01 -3.19939643e-01 5.19244909e-01 1.36655316e-01
7.42322505e-01 6.13023281e-01 -4.22720611e-01 -1.01368022e+00
7.83711970e-02 -9.73381042e-01 3.87317628e-01 -5.61504006e-01
-1.56277239e+00 7.61210740e-01 -1.27313361e-01 -1.56125605e+00
-4.56686541e-02 -6.83173954e-01 -7.21360147e-01 8.28962564e-01
-1.22698855e+00 -1.13788891e+00 3.89352255e-02 3.44221205e-01
7.99720764e-01 -2.93001272e-02 6.96490586e-01 -2.14337021e-01
-5.45892894e-01 3.81442696e-01 -8.06874707e-02 2.10964367e-01
2.14484960e-01 -1.43497336e+00 2.08294079e-01 6.90605223e-01
4.72364753e-01 6.51890218e-01 6.39286816e-01 -5.54627597e-01
-1.12715936e+00 -9.79260087e-01 9.27918911e-01 -3.73885989e-01
2.73164630e-01 -4.69685018e-01 -9.37918365e-01 4.97348964e-01
-7.48317838e-02 3.75374019e-01 5.88361502e-01 1.49938658e-01
-5.55223584e-01 8.54931101e-02 -1.39170611e+00 3.97995621e-01
8.51637363e-01 -4.08969223e-01 -4.96017843e-01 4.82520640e-01
3.00285250e-01 2.54666746e-01 -3.63098979e-01 2.31238961e-01
4.32113856e-01 -1.07830298e+00 8.99616599e-01 -8.70197952e-01
2.71143138e-01 -1.58957154e-01 -3.06235075e-01 -1.20065093e+00
-6.22942567e-01 -5.85123301e-01 1.81927398e-01 1.20052528e+00
6.29269302e-01 -4.17720884e-01 1.13899302e+00 4.02131855e-01
-1.92177385e-01 -1.09460461e+00 -5.00613570e-01 -9.51081634e-01
-9.26732570e-02 -1.17938463e-02 5.38938701e-01 4.41657841e-01
-3.61586325e-02 5.72643280e-01 -2.10270137e-02 -2.34640181e-01
4.50772673e-01 5.99313915e-01 6.58688366e-01 -1.41590381e+00
-3.82879168e-01 -7.93873727e-01 -5.08594990e-01 -8.96116495e-01
-1.53699189e-01 -7.64039576e-01 6.11902833e-01 -1.30747378e+00
6.61899090e-01 -5.15252471e-01 -3.54106367e-01 6.79940403e-01
-1.60679832e-01 1.84901938e-01 -9.99483094e-02 5.76956332e-01
-5.78192234e-01 4.68426496e-01 9.93874192e-01 -2.69678980e-01
-2.54526645e-01 4.76009190e-01 -9.47252691e-01 6.79961324e-01
7.20798671e-01 -6.04626656e-01 -3.99636179e-01 2.79914886e-01
-1.81696624e-01 -8.74423534e-02 2.00315148e-01 -1.09326017e+00
-3.79202306e-01 -9.23586951e-04 8.69281352e-01 -6.78598881e-01
4.47387874e-01 -8.07101130e-01 9.52243283e-02 5.62594771e-01
-6.02749944e-01 -1.72343746e-01 -1.20573014e-01 6.66265428e-01
-1.67862698e-01 -4.87511426e-01 1.05227387e+00 -5.76920152e-01
-4.64311033e-01 3.17392409e-01 -2.59160995e-01 -2.32905746e-02
9.88906920e-01 -5.06341159e-01 1.35633588e-01 -7.25395679e-02
-6.63774490e-01 -9.17576477e-02 5.47335267e-01 2.62737900e-01
4.56657916e-01 -1.21953082e+00 -4.48746234e-01 5.31112134e-01
4.45725977e-01 -1.53454110e-01 -1.28992852e-02 5.92822254e-01
-3.69483344e-02 3.30362946e-01 -2.77027637e-01 -4.09094036e-01
-1.19376194e+00 9.18121517e-01 3.31885517e-01 -2.39725530e-01
-7.04852402e-01 1.35443449e+00 3.90862435e-01 -2.13347390e-01
1.59047186e-01 -1.63258970e-01 -2.04068348e-01 3.56295556e-01
3.82618934e-01 4.50717472e-02 1.06306449e-01 -6.43951833e-01
-5.51085949e-01 4.14369285e-01 -2.90954918e-01 2.16356311e-02
1.60968220e+00 2.11059093e-01 -6.85123354e-02 2.99895495e-01
1.51533413e+00 3.73961031e-02 -1.28800249e+00 -2.23295882e-01
2.22452253e-01 -3.51467103e-01 -4.59601313e-01 -8.36216986e-01
-1.08311296e+00 6.24048054e-01 4.84268576e-01 4.44632828e-01
1.09382522e+00 6.05116904e-01 4.44071651e-01 2.08898872e-01
2.77132750e-01 -9.52477813e-01 3.11295599e-01 2.91708767e-01
9.02804434e-01 -1.25795972e+00 -4.27956805e-02 -3.79265964e-01
-7.68442571e-01 9.63843405e-01 4.76457685e-01 -4.43334341e-01
6.27742290e-01 -1.47580844e-03 -1.04942530e-01 -4.57541436e-01
-9.32283461e-01 -7.39847422e-01 7.38952875e-01 6.44632936e-01
5.29798985e-01 2.64624894e-01 -2.72085100e-01 5.34198701e-01
9.12264213e-02 -2.76341975e-01 5.22500537e-02 1.14484644e+00
-8.32853496e-01 -1.34585702e+00 -3.98966372e-01 7.77133942e-01
-3.74398261e-01 2.50615269e-01 -1.07771564e+00 7.29254842e-01
3.36681753e-01 6.44117773e-01 5.45365103e-02 -4.16918367e-01
2.36300007e-01 3.45663399e-01 4.94181514e-01 -8.38847458e-01
-7.24776745e-01 1.85601875e-01 -5.56795746e-02 -6.44629657e-01
-1.99257046e-01 -7.64070451e-01 -9.28370774e-01 3.65546256e-01
-5.12225270e-01 2.61925936e-01 5.84761739e-01 8.92059505e-01
2.32893109e-01 3.86562914e-01 1.01308322e+00 -1.05217612e+00
-6.53622031e-01 -7.58830011e-01 -5.93200207e-01 6.62326455e-01
3.25101942e-01 -4.35014278e-01 -3.35780770e-01 -5.29353507e-02]
|
[9.494291305541992, 1.0529749393463135]
|
6fb2eaa3-a7f4-4abf-9ce0-de013d2700f4
|
hiera-a-hierarchical-vision-transformer
|
2306.00989
| null |
https://arxiv.org/abs/2306.00989v1
|
https://arxiv.org/pdf/2306.00989v1.pdf
|
Hiera: A Hierarchical Vision Transformer without the Bells-and-Whistles
|
Modern hierarchical vision transformers have added several vision-specific components in the pursuit of supervised classification performance. While these components lead to effective accuracies and attractive FLOP counts, the added complexity actually makes these transformers slower than their vanilla ViT counterparts. In this paper, we argue that this additional bulk is unnecessary. By pretraining with a strong visual pretext task (MAE), we can strip out all the bells-and-whistles from a state-of-the-art multi-stage vision transformer without losing accuracy. In the process, we create Hiera, an extremely simple hierarchical vision transformer that is more accurate than previous models while being significantly faster both at inference and during training. We evaluate Hiera on a variety of tasks for image and video recognition. Our code and models are available at https://github.com/facebookresearch/hiera.
|
['Christoph Feichtenhofer', 'Yanghao Li', 'Jitendra Malik', 'Judy Hoffman', 'Omid Poursaeed', 'Arkabandhu Chowdhury', 'Vaibhav Aggarwal', 'Po-Yao Huang', 'Haoqi Fan', 'Chen Wei', 'Daniel Bolya', 'Yuan-Ting Hu', 'Chaitanya Ryali']
|
2023-06-01
| null | null | null | null |
['video-recognition', 'action-classification', 'action-recognition-in-videos', 'action-recognition-in-videos-2']
|
['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision']
|
[ 1.44689560e-01 7.24651515e-02 -3.36445756e-02 -4.39155161e-01
-8.81944120e-01 -5.55603981e-01 7.22120583e-01 -1.87555954e-01
-2.85428166e-01 1.72411457e-01 6.20268658e-02 -7.83322573e-01
2.47408360e-01 -5.08017242e-01 -8.51466894e-01 -5.23336172e-01
3.46610487e-01 4.92143631e-01 6.15846395e-01 4.07100394e-02
2.34291211e-01 1.09834440e-01 -1.57335377e+00 6.51990831e-01
5.62864840e-01 1.02510560e+00 2.12106317e-01 1.13091552e+00
1.15652390e-01 1.52993095e+00 -3.37767124e-01 -7.49019682e-01
2.96187401e-01 -2.51829952e-01 -9.39351201e-01 1.88276500e-01
1.05875182e+00 -6.11182630e-01 -3.57157052e-01 8.81825745e-01
1.72890350e-01 -3.37985367e-01 5.83266437e-01 -1.28876972e+00
-4.96025890e-01 5.52114785e-01 -7.51615584e-01 2.93250084e-01
2.75510419e-02 5.01765251e-01 1.26681674e+00 -1.08172977e+00
6.76740780e-02 1.14452231e+00 9.73072886e-01 5.02404809e-01
-1.13811707e+00 -7.39057779e-01 7.05370679e-02 3.80418062e-01
-1.22003746e+00 -8.78804505e-01 4.51456666e-01 -5.66781819e-01
1.22054911e+00 6.65607527e-02 8.19354951e-01 1.00575972e+00
1.45033240e-01 1.06097841e+00 1.18436587e+00 -3.38094383e-01
3.86958057e-03 5.55165559e-02 3.90869856e-01 1.18229699e+00
2.54638553e-01 1.05255730e-01 -5.61444342e-01 4.81030308e-02
7.43409038e-01 -5.16189300e-02 -2.50036031e-01 -3.92829895e-01
-8.09537172e-01 8.21360111e-01 4.88553584e-01 -2.03900784e-02
-8.92622173e-02 5.67227066e-01 3.15060258e-01 1.75940648e-01
1.66276947e-01 1.67413145e-01 -2.84665525e-01 -1.76160187e-01
-1.21482754e+00 -1.47491246e-01 9.00765479e-01 8.65461290e-01
8.23261857e-01 2.49006256e-01 7.16726705e-02 6.99419916e-01
4.80277091e-01 1.75616071e-01 2.70083457e-01 -1.34437501e+00
9.14942026e-02 2.23702282e-01 -2.21142963e-01 -5.26696265e-01
-1.08846590e-01 -4.70328450e-01 -7.69842684e-01 5.41327775e-01
7.43085921e-01 4.06307317e-02 -1.34646773e+00 1.34940886e+00
5.93843795e-02 2.29919568e-01 -8.64391997e-02 7.47509837e-01
6.95008695e-01 6.68309748e-01 -3.38255093e-02 1.29197106e-01
1.59013641e+00 -1.49288833e+00 -2.65933454e-01 -6.20569885e-01
3.14011097e-01 -8.49636853e-01 1.17487454e+00 7.27613986e-01
-1.40887761e+00 -5.80954075e-01 -1.21481431e+00 -5.92876613e-01
1.97986793e-02 1.77992508e-02 9.46815908e-01 6.51070476e-01
-1.62160301e+00 4.22181278e-01 -1.06194139e+00 -3.93221498e-01
7.07687020e-01 2.84996361e-01 -1.80030204e-02 -3.87162305e-02
-5.01061320e-01 1.03543580e+00 -8.17080289e-02 1.10186428e-01
-1.38456082e+00 -5.07486820e-01 -8.68082762e-01 1.48405820e-01
3.16948265e-01 -8.38143647e-01 2.04936743e+00 -1.18078387e+00
-1.40553653e+00 8.70150745e-01 -3.07303250e-01 -7.58587241e-01
5.72180450e-01 -2.70900369e-01 2.16994137e-01 1.40391588e-01
-6.94927275e-02 8.75795245e-01 1.24816763e+00 -1.18565369e+00
-7.60159492e-01 -1.62316933e-01 2.71576643e-01 1.11453764e-01
-1.59385070e-01 -1.54612750e-01 -7.27114558e-01 -2.57422894e-01
-2.74160653e-01 -7.31541455e-01 5.60454093e-02 2.41868883e-01
-1.38421565e-01 -2.23242402e-01 7.10311830e-01 -5.53334534e-01
6.40971184e-01 -2.11989880e+00 -9.82109606e-02 -1.34966001e-01
6.66295290e-01 3.11929584e-01 -2.97633074e-02 1.00637369e-01
1.91348061e-01 6.86870748e-03 -1.27900681e-02 -6.60394907e-01
-1.15345404e-01 3.14863712e-01 -3.22109312e-01 3.76750529e-01
1.21415071e-01 9.80119646e-01 -6.94771469e-01 -6.13849103e-01
3.26932430e-01 5.51147342e-01 -6.70915961e-01 2.18852181e-02
-1.44811064e-01 -1.83546338e-02 -1.25375306e-02 6.97992980e-01
3.60827595e-01 -6.59348190e-01 1.80393353e-01 -4.21461135e-01
-1.35595262e-01 6.31626666e-01 -6.16758585e-01 1.45502150e+00
-3.24097484e-01 1.02463651e+00 2.18550920e-01 -1.01737046e+00
3.79337013e-01 1.44273117e-01 6.69220462e-02 -6.78059340e-01
3.43767464e-01 -1.33806601e-01 5.01555279e-02 -2.27815762e-01
5.59997976e-01 -1.60139173e-01 3.75695676e-01 1.54579625e-01
3.22694510e-01 -2.23514467e-01 5.97833842e-02 2.84819752e-01
1.07463789e+00 1.47436932e-01 8.00211430e-02 -1.58713937e-01
1.88735817e-02 2.87539005e-01 4.06021446e-01 9.79892492e-01
-4.59976286e-01 6.10343158e-01 4.19636041e-01 -3.12658191e-01
-1.30270076e+00 -1.12293375e+00 4.08292338e-02 1.16521490e+00
-6.87507316e-02 -6.53599381e-01 -6.86207592e-01 -4.76421684e-01
-1.95289761e-01 4.52168822e-01 -5.06471157e-01 8.88474584e-02
-1.58607543e-01 -4.44731057e-01 9.37917233e-01 7.69321263e-01
5.78542590e-01 -6.10398710e-01 -7.85154998e-01 -1.26310453e-01
-6.63848966e-02 -1.23307788e+00 -3.66648704e-01 3.42035621e-01
-8.10035646e-01 -1.07466412e+00 -5.50343812e-01 -8.54518950e-01
3.74018818e-01 5.00497878e-01 1.26036227e+00 1.47976220e-01
-2.85845339e-01 3.78875107e-01 -1.68826714e-01 -4.85282630e-01
-3.59177828e-01 -8.62301365e-02 -2.54521191e-01 -2.55416036e-01
3.58132780e-01 -4.59451556e-01 -5.41007221e-01 5.57487980e-02
-4.70848083e-01 4.85808492e-01 7.39848077e-01 1.01089156e+00
3.66905510e-01 -1.45268857e-01 -2.36201227e-01 -6.70648515e-01
2.52867192e-01 -1.21260406e-02 -8.52927744e-01 1.77682757e-01
-6.63277507e-01 6.52826279e-02 4.53234196e-01 -3.01567465e-01
-7.21143186e-01 1.86537847e-01 -2.45973974e-01 -7.16450751e-01
3.36853671e-03 2.85466015e-01 3.27585936e-01 -2.69099802e-01
6.63353026e-01 3.96832198e-01 2.27994788e-02 -2.96315998e-01
4.26541209e-01 6.06459498e-01 6.89508796e-01 -3.23512137e-01
7.00803995e-01 3.77084941e-01 -2.70004600e-01 -8.96015406e-01
-1.12418628e+00 -3.74052733e-01 -3.08821589e-01 -2.41362512e-01
7.83522725e-01 -1.17911828e+00 -9.96910036e-01 7.10042238e-01
-1.01605666e+00 -8.02378654e-01 -2.00637966e-01 2.49023438e-01
-4.12230432e-01 4.65085477e-01 -1.00015330e+00 -7.36743987e-01
-4.06642824e-01 -1.28253543e+00 8.48261416e-01 2.23103866e-01
-9.30702537e-02 -7.85875022e-01 -1.54759273e-01 7.94510722e-01
4.10449207e-01 -4.36741561e-01 5.79279482e-01 -2.27893829e-01
-7.36600935e-01 1.72093555e-01 -5.46635211e-01 6.15336359e-01
-1.97250202e-01 2.65853673e-01 -1.44107294e+00 -4.64584887e-01
3.19229588e-02 -7.36833513e-01 1.26894057e+00 4.32799160e-01
9.55385864e-01 -4.30455953e-01 -8.06053281e-02 9.43253815e-01
1.41476548e+00 -4.07561362e-02 6.72806799e-01 2.22873330e-01
8.37382853e-01 1.91060677e-01 9.41840783e-02 1.69449821e-01
8.04447591e-01 3.73272747e-01 4.97199088e-01 -2.43717447e-01
-3.49427640e-01 -1.96807146e-01 6.74651384e-01 8.79789889e-01
-1.58996567e-01 -1.71911627e-01 -1.15634727e+00 7.02143848e-01
-1.68880856e+00 -9.66674864e-01 -9.94936377e-02 1.82516277e+00
8.67302597e-01 4.42701966e-01 2.26343304e-01 1.61550820e-01
3.35539639e-01 1.68422461e-01 -6.04547501e-01 -6.24377131e-01
1.18526541e-01 4.29216921e-01 8.31192195e-01 9.27139640e-01
-1.03477871e+00 1.04839706e+00 7.08994961e+00 5.87801635e-01
-1.12074077e+00 2.45128796e-01 5.90322614e-01 -8.59215334e-02
-3.61066870e-02 3.04383785e-01 -8.53006601e-01 6.45128936e-02
1.05872476e+00 2.00775236e-01 5.55939078e-01 8.87862504e-01
-2.82350361e-01 -2.48553470e-01 -1.15519536e+00 1.08167028e+00
1.82137802e-01 -1.38993919e+00 2.12984174e-01 -2.06579473e-02
3.15101445e-01 6.10690236e-01 1.17025256e-01 3.22840571e-01
7.46395111e-01 -1.09225786e+00 9.84781325e-01 1.45749152e-01
6.65283442e-01 -3.37837905e-01 4.66387779e-01 1.65607899e-01
-1.13033843e+00 -7.89516568e-02 -4.20415699e-01 -4.05962169e-01
-3.02988082e-01 4.47710872e-01 -8.54850590e-01 1.74407318e-01
9.59549546e-01 7.09613621e-01 -8.17144454e-01 1.04395103e+00
-4.25617814e-01 9.50459123e-01 -3.94336700e-01 1.19315170e-01
4.27555263e-01 8.08997825e-02 2.87882537e-02 1.31161392e+00
5.54143898e-02 1.63050424e-02 7.94169158e-02 5.96133530e-01
-1.19229190e-01 -5.27345717e-01 -4.99452263e-01 1.89265274e-02
4.08212543e-01 1.26748979e+00 -6.47562623e-01 -6.80269539e-01
-7.16500342e-01 1.11409438e+00 4.13174480e-01 2.53027290e-01
-9.49213386e-01 -1.76899865e-01 5.71635723e-01 8.70324485e-03
6.47275448e-01 -2.68466383e-01 -3.36688459e-01 -1.36338151e+00
-1.21704251e-01 -1.09521174e+00 2.86571831e-01 -1.04248178e+00
-1.03517032e+00 3.14531386e-01 -3.79714966e-01 -8.97245288e-01
-2.56319433e-01 -7.95070052e-01 -5.15243828e-01 4.24437642e-01
-1.55815363e+00 -1.41806138e+00 -5.42935431e-01 8.25904608e-01
6.91273570e-01 2.04882935e-01 6.09349728e-01 2.52256721e-01
-4.45450157e-01 8.43005180e-01 -2.08817035e-01 4.31279004e-01
5.81760406e-01 -1.28621721e+00 7.29791343e-01 1.07836437e+00
4.27273989e-01 3.55461448e-01 7.55148411e-01 -1.36362135e-01
-1.55169725e+00 -8.01622927e-01 6.37665331e-01 -6.45575464e-01
8.99904311e-01 -4.80194300e-01 -7.14025795e-01 1.14852393e+00
5.55225611e-01 -2.42146760e-01 3.57928395e-01 2.52983212e-01
-9.16831553e-01 -1.83818251e-01 -6.54947579e-01 6.04710460e-01
8.42677534e-01 -8.45309198e-01 -7.74807751e-01 2.08479241e-01
3.32676709e-01 -3.28940183e-01 -6.46788538e-01 1.50262520e-01
8.88460636e-01 -1.28453732e+00 1.05847180e+00 -1.50750607e-01
5.57725668e-01 -3.18728477e-01 -1.63437188e-01 -9.50071812e-01
-5.73539734e-01 -5.21560133e-01 -2.57678360e-01 9.53046143e-01
4.61955220e-01 -5.18975317e-01 8.07161093e-01 3.67923230e-01
-2.11127520e-01 -5.84810555e-01 -6.37103498e-01 -6.12086833e-01
2.42857989e-02 -6.18601203e-01 -1.34289265e-01 7.44430125e-01
-7.19135404e-02 9.14088130e-01 -5.39152682e-01 1.44546837e-01
1.01200795e+00 -1.19311377e-01 8.15469444e-01 -1.06042886e+00
-6.70061946e-01 -5.88760555e-01 -4.68325049e-01 -1.37067521e+00
-2.82271832e-01 -7.98609078e-01 1.81701452e-01 -1.72485626e+00
4.56896365e-01 -1.07924163e-01 -1.39895648e-01 1.13519418e+00
8.74363184e-02 6.47750676e-01 4.19174284e-01 4.31470662e-01
-6.24178529e-01 1.20899729e-01 1.08318162e+00 -3.90813529e-01
2.71514267e-01 -2.21793115e-01 -9.05761719e-01 9.32625830e-01
8.52784753e-01 -2.46095255e-01 -5.28762579e-01 -9.02632117e-01
5.34164757e-02 -8.15655515e-02 8.42985213e-01 -1.20841050e+00
5.54279268e-01 2.67775774e-01 5.00736773e-01 -4.18485105e-01
7.60391891e-01 -5.53350747e-01 -1.70230582e-01 6.02372468e-01
-2.08226204e-01 9.19275507e-02 3.25615525e-01 1.54265970e-01
-1.07120566e-01 -5.43269999e-02 1.07293510e+00 -2.59446293e-01
-7.92526484e-01 1.61210164e-01 -6.10952675e-01 -5.34712628e-04
7.03954756e-01 -3.94460768e-01 -6.80523872e-01 -5.90547323e-01
-6.18897855e-01 1.41840115e-01 6.48794413e-01 1.49569020e-01
6.38744771e-01 -7.26471126e-01 -4.78615820e-01 2.54155129e-01
-6.88454062e-02 2.94682365e-02 -5.69370203e-02 9.83737707e-01
-5.73543906e-01 1.90146178e-01 -3.25076222e-01 -7.77096868e-01
-1.37718773e+00 4.65555221e-01 4.90845144e-01 -1.54140279e-01
-7.48841166e-01 1.17058945e+00 3.85007769e-01 -1.90497115e-02
4.77797061e-01 -5.55925906e-01 1.69106632e-01 -5.20763732e-02
5.81255257e-01 2.82687973e-02 -6.87654465e-02 -2.52358973e-01
-3.61289293e-01 5.26560187e-01 -2.80593246e-01 -2.99823195e-01
1.11577213e+00 2.66876668e-02 4.02997509e-02 4.01390076e-01
8.02010179e-01 -2.03249425e-01 -1.54544508e+00 -4.91780695e-03
-1.42714366e-01 -2.48652950e-01 3.98109466e-01 -7.07465351e-01
-1.06993794e+00 1.27672648e+00 5.55083513e-01 1.96876228e-01
1.15162730e+00 6.44775778e-02 6.29609704e-01 4.46104348e-01
2.38764450e-01 -7.31984735e-01 1.70564968e-02 6.91727638e-01
3.94795269e-01 -1.18425500e+00 1.40018046e-01 -1.73961684e-01
-6.95009172e-01 7.87355542e-01 6.37963057e-01 -1.87943950e-01
4.85759735e-01 7.81876981e-01 2.24894181e-01 -2.08477736e-01
-1.13385761e+00 -3.03983063e-01 -9.51256827e-02 6.70282722e-01
3.99522990e-01 -1.72560781e-01 2.18419820e-01 7.54518956e-02
-3.10600758e-01 1.84050694e-01 5.29672027e-01 9.13417161e-01
-6.46730065e-01 -7.88739920e-01 -2.27258429e-01 4.22945142e-01
-5.43431342e-01 -4.55569804e-01 -3.81640315e-01 7.47759342e-01
-1.46107569e-01 1.04753947e+00 1.94236338e-01 -6.22559011e-01
-5.35179395e-03 -5.56890853e-03 8.27126205e-01 -3.19431096e-01
-5.71228802e-01 1.20188333e-01 2.33025342e-01 -6.82519853e-01
-3.92900795e-01 -3.71464312e-01 -9.91074264e-01 -6.50909483e-01
-2.67976075e-01 -5.56824580e-02 5.16289711e-01 9.25955355e-01
1.72441438e-01 4.27294880e-01 2.56530792e-01 -1.05999136e+00
-6.11167967e-01 -8.21702063e-01 -1.28617167e-01 -3.05346735e-02
6.51159227e-01 -3.12351435e-01 -2.81238586e-01 2.57957667e-01]
|
[9.534127235412598, 1.493985891342163]
|
55c853fc-dabe-4091-afa0-9dcd5df18afe
|
unsupervised-human-pose-estimation-through
|
2105.04154
| null |
https://arxiv.org/abs/2105.04154v1
|
https://arxiv.org/pdf/2105.04154v1.pdf
|
Unsupervised Human Pose Estimation through Transforming Shape Templates
|
Human pose estimation is a major computer vision problem with applications ranging from augmented reality and video capture to surveillance and movement tracking. In the medical context, the latter may be an important biomarker for neurological impairments in infants. Whilst many methods exist, their application has been limited by the need for well annotated large datasets and the inability to generalize to humans of different shapes and body compositions, e.g. children and infants. In this paper we present a novel method for learning pose estimators for human adults and infants in an unsupervised fashion. We approach this as a learnable template matching problem facilitated by deep feature extractors. Human-interpretable landmarks are estimated by transforming a template consisting of predefined body parts that are characterized by 2D Gaussian distributions. Enforcing a connectivity prior guides our model to meaningful human shape representations. We demonstrate the effectiveness of our approach on two different datasets including adults and infants.
|
['Bernhard Kainz', 'Tomoki Arichi', 'Anna Lukens', 'Simon Ellershaw', 'Athanasios Vlontzos', 'Luca Schmidtke']
|
2021-05-10
| null |
http://openaccess.thecvf.com//content/CVPR2021/html/Schmidtke_Unsupervised_Human_Pose_Estimation_Through_Transforming_Shape_Templates_CVPR_2021_paper.html
|
http://openaccess.thecvf.com//content/CVPR2021/papers/Schmidtke_Unsupervised_Human_Pose_Estimation_Through_Transforming_Shape_Templates_CVPR_2021_paper.pdf
|
cvpr-2021-1
|
['template-matching']
|
['computer-vision']
|
[ 4.22765166e-01 4.23198044e-01 -8.25799257e-02 -5.30714095e-01
-4.97378707e-01 -5.90281546e-01 4.60224450e-01 1.12512782e-01
-5.47572851e-01 5.36014199e-01 3.04951578e-01 3.00449193e-01
-1.92172959e-01 -2.44770065e-01 -7.48080969e-01 -3.47103149e-01
-1.46898374e-01 7.56102145e-01 1.50929049e-01 5.87675050e-02
-1.34236515e-01 6.84944928e-01 -1.49556053e+00 -1.91841036e-01
6.04823411e-01 8.99230003e-01 -1.55600756e-01 5.75544596e-01
4.38583553e-01 -1.50817074e-02 -2.77999163e-01 -6.05243385e-01
4.49790537e-01 -2.30031341e-01 -4.19060022e-01 2.43219614e-01
8.18565607e-01 -3.75703007e-01 -1.95505723e-01 8.11440647e-01
8.55123281e-01 2.02174276e-01 8.01084101e-01 -1.25887370e+00
-1.17290519e-01 1.14103749e-01 -7.98491299e-01 -8.11866447e-02
7.69483745e-01 -2.73101389e-01 5.94953775e-01 -7.19020367e-01
6.73979580e-01 1.14785409e+00 1.09891415e+00 8.99227858e-01
-1.15367401e+00 -4.70739812e-01 7.61030838e-02 -2.33647361e-01
-1.29600239e+00 -5.54173410e-01 6.97913587e-01 -7.47482061e-01
7.07891166e-01 1.87299490e-01 8.84235919e-01 1.03329861e+00
6.04590625e-02 6.58393085e-01 6.84764624e-01 -3.84545743e-01
2.11068287e-01 -2.08120972e-01 -2.15261981e-01 7.93959022e-01
6.51430428e-01 1.32734656e-01 -7.01817334e-01 -2.17563346e-01
9.72121656e-01 1.40849873e-01 -2.21706301e-01 -1.10668600e+00
-1.24038470e+00 6.15824580e-01 3.11119795e-01 -3.24120894e-02
-4.80295449e-01 1.77210033e-01 2.34659210e-01 -1.77208781e-01
3.86510402e-01 2.60333210e-01 -4.82499361e-01 -6.20733313e-02
-9.64395702e-01 3.52906048e-01 7.50285804e-01 1.17015457e+00
4.13068265e-01 -2.64124602e-01 7.23731890e-02 6.37105942e-01
4.68902230e-01 4.77304518e-01 3.78362298e-01 -7.39003062e-01
3.54194909e-01 5.88874757e-01 5.55290356e-02 -9.31369543e-01
-9.02857006e-01 -1.20129690e-01 -6.87043726e-01 1.10008724e-01
5.69603801e-01 -2.90943593e-01 -9.05277014e-01 1.80272055e+00
8.73327732e-01 7.69248083e-02 -3.73744935e-01 8.31986487e-01
8.29926074e-01 -1.64538592e-01 1.21213719e-02 3.94324511e-02
1.41836011e+00 -7.65577853e-01 -2.72832781e-01 -3.93269986e-01
1.83177650e-01 -5.10975599e-01 4.06103760e-01 5.45642793e-01
-1.15458930e+00 -1.00538872e-01 -7.58144081e-01 -1.12021223e-01
-7.18247592e-02 4.40041050e-02 6.27527833e-01 8.46303284e-01
-9.28289413e-01 5.34816206e-01 -1.24040592e+00 -7.50461936e-01
5.56131124e-01 8.22986722e-01 -8.65787268e-01 1.63235232e-01
-4.89122272e-01 1.08556223e+00 1.11796774e-01 1.06423728e-01
-4.81433153e-01 -5.50534308e-01 -1.14955258e+00 -4.01729584e-01
8.59586336e-03 -9.51623797e-01 1.11089528e+00 -7.18059063e-01
-1.26688302e+00 1.38635659e+00 -1.57521188e-01 -4.41710770e-01
8.98030221e-01 -5.98398864e-01 4.64996547e-02 3.97624940e-01
-2.20765155e-02 6.88529909e-01 9.85555768e-01 -8.05914521e-01
-3.54636252e-01 -7.89237201e-01 -2.03824669e-01 3.00374150e-01
-1.03129394e-01 3.82982701e-01 -5.05977392e-01 -7.77706861e-01
6.24770284e-01 -1.20752954e+00 -3.22972208e-01 6.47700548e-01
-2.47658074e-01 1.76156789e-01 4.34027672e-01 -8.73100102e-01
7.55581439e-01 -1.77663374e+00 4.15458500e-01 3.62823397e-01
4.78091806e-01 -3.70860435e-02 6.14775717e-02 1.29207283e-01
1.70404866e-01 -4.36288655e-01 -2.61191219e-01 -5.47421932e-01
7.72877932e-02 -7.57182576e-03 3.84305894e-01 1.03372669e+00
5.54896370e-02 1.02126765e+00 -1.01603949e+00 -4.59087759e-01
2.35276088e-01 9.75103974e-01 -5.80407262e-01 3.16620052e-01
1.13693334e-01 8.96188140e-01 -4.16323364e-01 6.63818002e-01
4.05076265e-01 1.06954277e-01 1.12135829e-02 -1.65006131e-01
1.61384434e-01 -2.16850340e-02 -1.05180061e+00 1.87710190e+00
-1.57507882e-01 4.37477767e-01 2.63041735e-01 -8.06536794e-01
7.87560642e-01 5.07087827e-01 8.86744797e-01 -1.22623593e-01
4.61655527e-01 5.90744503e-02 -1.21595748e-02 -5.06604671e-01
7.82089308e-03 -2.15328738e-01 1.56494573e-01 3.31105918e-01
6.54143021e-02 -3.11130911e-01 -1.34812415e-01 -2.91350961e-01
1.06569779e+00 3.98752779e-01 6.51883483e-01 -2.25307852e-01
8.11137334e-02 -3.35410178e-01 4.31316346e-01 4.41372335e-01
-2.22078770e-01 1.19509077e+00 -6.38421774e-02 -6.54943943e-01
-8.80971849e-01 -1.28467000e+00 -1.84228003e-01 9.51801777e-01
-1.68014482e-01 -3.61724123e-02 -8.19233775e-01 -6.10370100e-01
9.07234997e-02 9.99088287e-02 -7.53254056e-01 -9.36773494e-02
-6.58527434e-01 -4.28396791e-01 5.35826564e-01 8.24904799e-01
3.35567328e-03 -1.01866186e+00 -1.32248402e+00 7.66887143e-02
1.32487863e-01 -1.18410516e+00 -7.35641420e-01 -2.44622603e-02
-8.02528143e-01 -1.22182560e+00 -1.11324501e+00 -8.51879358e-01
1.13918281e+00 -1.87255085e-01 9.33640242e-01 -5.36946654e-02
-6.91533327e-01 1.02104723e+00 -2.29146853e-01 -6.76841378e-01
-2.17845873e-03 -9.41216573e-02 5.40379047e-01 3.22008133e-02
2.83936232e-01 -6.82189107e-01 -9.00466561e-01 2.45377019e-01
-5.82641244e-01 -1.43358242e-02 2.61030287e-01 6.02616489e-01
6.33405805e-01 -5.93835413e-01 9.21719819e-02 -7.47348666e-01
2.84859300e-01 -2.67426133e-01 -5.08871436e-01 1.91697240e-01
-2.92291790e-01 -1.03173450e-01 -8.54929257e-03 -8.15703809e-01
-6.15488827e-01 7.58534729e-01 -2.74926629e-02 -4.66796517e-01
-4.75371063e-01 1.31232366e-01 -2.75077015e-01 -3.57399344e-01
5.57221174e-01 -4.64243405e-02 2.05554128e-01 -4.85119164e-01
3.81781340e-01 2.47142002e-01 1.00232315e+00 -7.04792678e-01
8.57927024e-01 4.04476345e-01 2.82703429e-01 -9.05489266e-01
-5.37017643e-01 -5.17676055e-01 -1.28464770e+00 -2.14976817e-01
1.04082012e+00 -7.60948777e-01 -6.51489615e-01 3.68212581e-01
-1.21899128e+00 -1.57441109e-01 -1.97389603e-01 7.37875521e-01
-8.59320998e-01 2.41378576e-01 -3.24974656e-01 -7.05165744e-01
-6.52194619e-01 -9.15464580e-01 1.40450895e+00 2.67648965e-01
-7.09573984e-01 -1.02042758e+00 2.70557195e-01 2.40776390e-01
1.06213696e-01 9.55635846e-01 3.48220080e-01 -6.84866428e-01
-1.36863455e-01 -5.85191548e-01 1.29508644e-01 -6.33314997e-02
2.68298447e-01 -1.92667201e-01 -8.72229099e-01 -5.33452034e-01
-2.41643757e-01 -2.56901920e-01 2.65993834e-01 5.66680551e-01
7.12967038e-01 -1.69634745e-01 -5.43355286e-01 7.70963252e-01
8.71816814e-01 1.99110489e-02 2.06702828e-01 1.30313575e-01
8.49549174e-01 9.84011531e-01 3.46574605e-01 4.67994273e-01
5.36030710e-01 6.87204421e-01 2.75244504e-01 -2.48751277e-03
-9.98405144e-02 -3.53761226e-01 6.03473932e-03 4.95067418e-01
-3.86119068e-01 2.30689570e-01 -1.15768659e+00 6.21082902e-01
-1.69681847e+00 -5.92890620e-01 1.47872254e-01 2.52025795e+00
6.67249262e-01 -1.00793526e-01 4.60376859e-01 2.56474316e-02
6.15534246e-01 -3.97456646e-01 -5.41792333e-01 -1.49583705e-02
3.38679224e-01 4.33312565e-01 2.87765354e-01 1.60824835e-01
-1.21472192e+00 5.38361967e-01 6.55034256e+00 -1.37599245e-01
-1.04530108e+00 1.34446537e-02 1.42790481e-01 -9.65470299e-02
2.18745992e-02 -2.97275156e-01 -5.89149773e-01 1.80958435e-01
2.93778598e-01 2.27760710e-02 2.31351987e-01 6.86858177e-01
-2.00326219e-02 1.06419340e-01 -1.59116042e+00 1.30788743e+00
3.22182387e-01 -5.55236161e-01 -4.64303017e-01 -8.41716602e-02
6.32512152e-01 2.13216376e-02 -1.45774260e-02 -2.23441780e-01
-1.49989218e-01 -1.16397047e+00 8.82074773e-01 5.49129128e-01
9.10408735e-01 -5.13727665e-01 2.11297810e-01 4.35693383e-01
-1.23988032e+00 2.96691656e-01 -2.17620119e-01 -3.40428539e-02
5.97960725e-02 1.08909514e-02 -8.50257456e-01 8.83940011e-02
6.36890352e-01 4.57026571e-01 -3.96088600e-01 1.51414013e+00
-1.74186498e-01 2.18259022e-01 -7.23747611e-01 9.21074226e-02
-4.14928168e-01 -1.72616899e-01 4.99611169e-01 1.08789754e+00
3.50923419e-01 2.17455924e-01 -2.90186075e-03 5.31723261e-01
9.09735858e-02 1.56136855e-01 -7.94428527e-01 2.77528167e-01
3.02378416e-01 1.19819045e+00 -9.50976670e-01 1.84265763e-01
-5.67050517e-01 9.87320662e-01 2.59005636e-01 1.33770689e-01
-5.89949548e-01 -1.80798739e-01 4.75669265e-01 4.93921518e-01
1.99216321e-01 -2.50440180e-01 -1.09180942e-01 -1.16846812e+00
1.72610268e-01 -8.38408709e-01 3.07816803e-01 -4.47528630e-01
-1.09857845e+00 5.49312711e-01 3.70424092e-01 -1.27080905e+00
-5.33119917e-01 -5.42940140e-01 -6.47877872e-01 5.08412480e-01
-8.89780283e-01 -1.38146830e+00 -4.29992884e-01 4.68802989e-01
2.44841635e-01 6.80432282e-03 7.65867949e-01 2.92958677e-01
-3.38359892e-01 8.26325893e-01 -3.08055937e-01 2.22436354e-01
4.69471693e-01 -1.14608061e+00 7.13649631e-01 6.85937405e-01
3.29587132e-01 8.04749906e-01 8.60296369e-01 -6.88212216e-01
-1.45788932e+00 -9.17324424e-01 6.68515444e-01 -8.47235680e-01
3.15432489e-01 -4.93832320e-01 -5.84748328e-01 8.44126642e-01
-2.45326266e-01 4.05639827e-01 5.21482408e-01 -7.50883073e-02
-2.49563843e-01 2.59760708e-01 -1.47203839e+00 3.70709449e-01
1.44020641e+00 -3.13953489e-01 -6.62952781e-01 1.86124206e-01
2.24676073e-01 -9.17350411e-01 -8.53424668e-01 7.62276113e-01
1.13385510e+00 -7.32526481e-01 1.08865249e+00 -6.36339247e-01
-6.46365136e-02 -1.33497477e-01 3.06236856e-02 -1.10630965e+00
4.75401394e-02 -7.35360324e-01 -3.11812162e-01 9.17543590e-01
1.11091241e-01 -5.82587659e-01 1.10725236e+00 1.14459682e+00
2.21951544e-01 -8.56904507e-01 -1.02759528e+00 -6.39724374e-01
-5.60086295e-02 -3.98481309e-01 5.18686116e-01 5.90739906e-01
-3.20565812e-02 -2.01391894e-03 -3.66958976e-01 2.68360645e-01
8.62916231e-01 -1.22458242e-01 8.42634380e-01 -1.50847173e+00
-2.78617948e-01 -2.46177539e-01 -8.57369423e-01 -9.37931299e-01
1.61055639e-01 -7.73207843e-01 2.06907168e-01 -1.47423458e+00
1.86714366e-01 -2.31139660e-01 1.00581408e-01 3.93082172e-01
-1.08017921e-02 3.14190865e-01 -8.98944363e-02 -2.32438907e-01
-4.44196969e-01 2.16017634e-01 7.95107365e-01 -3.36664310e-03
-1.54000759e-01 5.28323829e-01 -3.38567108e-01 1.04651964e+00
5.87002277e-01 -4.95131403e-01 -4.43330914e-01 -4.15078998e-01
3.67497094e-02 2.65675671e-02 4.33840424e-01 -9.70827758e-01
2.82367349e-01 1.41714394e-01 6.76458001e-01 -4.52873737e-01
4.74083304e-01 -9.34742093e-01 3.35716218e-01 4.89850372e-01
-3.53606381e-02 2.37691209e-01 8.04036260e-02 3.72659653e-01
1.84574381e-01 -8.92214477e-02 6.53482497e-01 -5.04894704e-02
-2.70459920e-01 6.31436527e-01 7.62540549e-02 3.08954805e-01
9.63969350e-01 -4.38332111e-01 3.05093586e-01 -3.96777540e-01
-8.10687780e-01 1.56705156e-01 6.56854212e-01 5.86893857e-01
9.30272222e-01 -1.16053975e+00 -6.45225286e-01 3.99146080e-01
1.40632972e-01 2.81040519e-01 -2.12285325e-01 9.70636606e-01
-6.05313122e-01 2.11546972e-01 -2.29839668e-01 -6.47198081e-01
-1.59804738e+00 4.37415302e-01 4.21835005e-01 3.14631462e-01
-7.14838207e-01 9.69390631e-01 2.91345865e-01 -5.27097166e-01
6.17133915e-01 -4.52536672e-01 -1.99350551e-01 -1.63977310e-01
4.57646072e-01 5.06333709e-01 -1.21456375e-02 -1.15359890e+00
-5.98654807e-01 1.03704941e+00 2.59773195e-01 4.05660924e-03
1.36521482e+00 -1.09991565e-01 -3.38843442e-03 9.00508091e-03
1.01953125e+00 1.13697179e-01 -1.31294966e+00 -1.60065100e-01
3.05220455e-01 -3.82548571e-01 -5.38309395e-01 -3.60708654e-01
-1.01933992e+00 8.67520452e-01 7.85474539e-01 -3.42739582e-01
9.52241004e-01 3.08095664e-01 7.36781240e-01 2.46059239e-01
8.22228789e-01 -6.77101314e-01 -1.15507297e-01 1.43725738e-01
1.06713998e+00 -1.26836979e+00 2.51046091e-01 -4.22844589e-01
-4.48150367e-01 9.98309016e-01 5.28747976e-01 -4.91558835e-02
7.06177890e-01 3.80385846e-01 2.12792963e-01 -2.34080002e-01
-3.02836031e-01 -2.46972337e-01 1.03152895e+00 9.74593103e-01
6.22545719e-01 1.03727922e-01 -3.25268835e-01 7.21939921e-01
-4.21675533e-01 -1.05189204e-01 3.28918584e-02 1.04988635e+00
-2.66746134e-01 -9.83254850e-01 -4.70355272e-01 5.94677269e-01
-5.81842899e-01 2.59281665e-01 -4.07661796e-01 6.27225101e-01
2.59255022e-01 4.99366105e-01 -1.60532713e-01 -1.45383254e-01
5.35971224e-01 4.01559062e-02 1.06389081e+00 -9.81692970e-01
-3.70320439e-01 1.16494231e-01 -1.53047502e-01 -5.85914195e-01
-3.92419845e-01 -1.11887527e+00 -1.31906486e+00 3.46641809e-01
-1.57555267e-01 -3.58371735e-01 7.39980638e-01 8.73418927e-01
3.02536301e-02 -4.28437814e-02 9.05764699e-02 -1.14169359e+00
-4.28166807e-01 -6.76221251e-01 -5.43765187e-01 4.87219870e-01
5.07769406e-01 -9.53320980e-01 1.62909329e-02 1.98150486e-01]
|
[7.037136077880859, -1.0859593152999878]
|
809b36bc-0167-4ea4-82a8-74768058de6d
|
unsupervised-video-summarization-via
| null | null |
https://link.springer.com/chapter/10.1007/978-3-030-37731-1_40
|
https://link.springer.com/chapter/10.1007/978-3-030-37731-1_40
|
Unsupervised Video Summarization via Attention-Driven Adversarial Learning
|
This paper presents a new video summarization approach that integrates an attention mechanism to identify the significant parts of the video, and is trained unsupervisingly via generative adversarial learning. Starting from the SUM-GAN model, we first develop an improved version of it (called SUM-GAN-sl) that has a significantly reduced number of learned parameters, performs incremental training of the model’s components, and applies a stepwise label-based strategy for updating the adversarial part. Subsequently, we introduce an attention mechanism to SUM-GAN-sl in two ways: (i) by integrating an attention layer within the variational auto-encoder (VAE) of the architecture (SUM-GAN-VAAE), and (ii) by replacing the VAE with a deterministic attention auto-encoder (SUM-GAN-AAE). Experimental evaluation on two datasets (SumMe and TVSum) documents the contribution of the attention auto-encoder to faster and more stable training of the model, resulting in a significant performance improvement with respect to the original model and demonstrating the competitiveness of the proposed SUM-GAN-AAE against the state of the art.
|
['Ioannis Patras', 'Vasileios Mezaris', 'Alexandros I. Metsai', 'Eleni Adamantidou', 'Evlampios Apostolidis']
|
2019-12-24
| null | null | null |
multimedia-modeling-mmm-2019-12
|
['unsupervised-video-summarization']
|
['computer-vision']
|
[ 4.13168073e-01 7.14968860e-01 3.04711014e-01 6.90303836e-03
-1.08499885e+00 -4.32482928e-01 7.09688127e-01 -5.95047534e-01
-1.34194970e-01 7.46168852e-01 5.60031474e-01 -1.54879823e-01
4.63698864e-01 -6.72110498e-01 -1.19314432e+00 -7.74845183e-01
2.13462934e-01 4.31755543e-01 1.78549588e-01 -1.69270664e-01
-2.20254049e-01 9.63541269e-02 -1.13377285e+00 2.49681935e-01
1.07155943e+00 9.26038086e-01 2.11245002e-04 8.56656432e-01
1.77196980e-01 1.54520488e+00 -8.00511599e-01 -7.94206023e-01
1.37192652e-01 -1.23768675e+00 -7.92052567e-01 3.43738616e-01
4.62155253e-01 -6.50010943e-01 -5.49893320e-01 6.66400671e-01
3.25625539e-01 2.08760664e-01 8.61975849e-01 -9.88129854e-01
-8.88784111e-01 5.76120138e-01 -5.55310428e-01 -7.24717826e-02
1.29570410e-01 2.17738952e-02 9.61369634e-01 -6.86376870e-01
6.07408583e-01 1.03898704e+00 6.24494195e-01 9.24765587e-01
-1.15721369e+00 -4.17480737e-01 1.38392389e-01 3.34266633e-01
-1.27504539e+00 -4.38061893e-01 9.28390622e-01 -3.55408579e-01
9.55043495e-01 4.38168585e-01 5.94374001e-01 1.41236579e+00
2.51907080e-01 9.83615994e-01 6.99373424e-01 -3.82041067e-01
3.51518422e-01 -9.94221792e-02 -3.06310117e-01 6.70392752e-01
-1.66802838e-01 -1.20866418e-01 -9.77414250e-02 9.77344811e-02
6.75602913e-01 -2.07947120e-01 -4.55978513e-01 -4.06467736e-01
-8.58004928e-01 8.27367067e-01 4.77076471e-01 1.96266174e-01
-7.16977537e-01 3.83151919e-01 6.20733142e-01 2.10511610e-01
6.32747948e-01 3.25065464e-01 -1.13402121e-02 -2.97054779e-02
-1.26293778e+00 2.03417897e-01 5.57753026e-01 1.00206685e+00
4.91550535e-01 6.83483839e-01 -6.82021201e-01 7.72740602e-01
5.09410426e-02 2.63222218e-01 6.32606626e-01 -1.08134782e+00
4.50421065e-01 3.16964090e-01 -3.96127515e-02 -5.05546153e-01
1.59953699e-01 -7.24353075e-01 -1.00148261e+00 2.01060578e-01
-1.97699204e-01 -3.18539530e-01 -1.10706306e+00 2.01412463e+00
2.54135001e-02 5.66680789e-01 1.46404147e-01 5.78449249e-01
6.58813715e-01 9.73567069e-01 -1.95684731e-02 -2.69823074e-01
9.19000864e-01 -1.68548560e+00 -7.54004896e-01 -2.26955011e-01
2.14816213e-01 -2.37618193e-01 8.36718142e-01 1.81001574e-01
-1.41838408e+00 -7.96491385e-01 -1.17654157e+00 -1.18589103e-01
2.42961850e-02 1.11072868e-01 6.53335527e-02 4.01742578e-01
-1.39169300e+00 5.09455800e-01 -8.37691963e-01 -7.35036582e-02
6.09802008e-01 2.53271073e-01 -2.37925306e-01 -2.75000539e-02
-1.04290748e+00 6.87421024e-01 3.90829593e-01 -2.36859992e-02
-1.35113180e+00 -5.33437371e-01 -1.15568328e+00 3.90544385e-01
4.60460126e-01 -1.20280492e+00 1.22718620e+00 -1.62448359e+00
-2.05382156e+00 3.32055688e-01 -1.69126689e-01 -6.35857522e-01
7.45562911e-01 -3.22555542e-01 -5.28752841e-02 4.15924400e-01
-1.26005754e-01 6.39507592e-01 1.44228685e+00 -1.68341565e+00
-3.45426291e-01 1.15575688e-02 9.04909223e-02 1.15588248e-01
-1.20932490e-01 -1.69092193e-01 -6.94956899e-01 -1.04186666e+00
-5.74689984e-01 -9.55955386e-01 -7.46990740e-02 -5.99926591e-01
-4.83217597e-01 5.34152426e-02 8.91968429e-01 -1.31458199e+00
1.37529159e+00 -1.95805275e+00 7.77091742e-01 -1.20629266e-01
2.20470950e-01 6.82013452e-01 -3.58145207e-01 4.75654393e-01
-3.19555789e-01 2.26023756e-02 -6.50141478e-01 -9.68369007e-01
-2.28627384e-01 3.94140512e-01 -3.61927658e-01 1.65475965e-01
5.40596604e-01 1.31497085e+00 -8.81124973e-01 -2.77770936e-01
2.28753462e-01 6.69425189e-01 -8.36048961e-01 6.66719377e-01
-3.10530037e-01 6.65506005e-01 -2.04160616e-01 3.83366674e-01
4.31978017e-01 2.14656498e-02 2.20091984e-01 -1.50669768e-01
1.88084453e-01 -4.92757447e-02 -6.31178558e-01 1.64364481e+00
-4.32834744e-01 4.64295775e-01 5.11352718e-02 -1.07390964e+00
5.70960701e-01 4.68862057e-01 2.97527790e-01 -4.93728936e-01
2.59292692e-01 -7.82801881e-02 -1.62833720e-01 -3.98174465e-01
5.58863163e-01 -7.99146891e-02 3.75567339e-02 3.36072803e-01
5.61030984e-01 2.13609099e-01 1.48093298e-01 4.46342111e-01
1.15442336e+00 5.60686111e-01 3.08730930e-01 7.53305033e-02
8.83505881e-01 -4.01260644e-01 5.08945167e-01 5.83449066e-01
1.43334717e-01 8.52617264e-01 6.46390975e-01 -6.31101653e-02
-1.37563682e+00 -8.83787513e-01 5.01851916e-01 9.24069583e-01
-3.32565427e-01 -4.66920584e-01 -1.30878520e+00 -1.01001120e+00
-4.50267494e-01 1.11503029e+00 -1.00054121e+00 -4.29524183e-01
-7.58318841e-01 -4.41833645e-01 3.64021927e-01 8.79512250e-01
6.49963498e-01 -1.13468838e+00 -4.30013537e-01 1.77109748e-01
-3.25675875e-01 -8.73762131e-01 -7.21907377e-01 -4.49360050e-02
-6.51418865e-01 -7.04864919e-01 -1.03623116e+00 -6.27687633e-01
5.82892478e-01 -1.88710973e-01 1.20059657e+00 -1.56896502e-01
3.06484312e-01 4.11543578e-01 -5.77346802e-01 -2.98624873e-01
-9.72033978e-01 2.16932982e-01 -3.94275218e-01 4.24894065e-01
-3.14031869e-01 -6.73918247e-01 -5.04307032e-01 -3.03196851e-02
-1.10078323e+00 2.11583540e-01 7.75709808e-01 1.05904436e+00
6.39088392e-01 -2.15981498e-01 7.36327469e-01 -1.13593221e+00
3.63527894e-01 -5.14906645e-01 -2.35118911e-01 1.81532979e-01
-3.88981819e-01 4.52634059e-02 9.41953361e-01 -2.79797286e-01
-1.19125843e+00 -1.21094391e-01 -5.72725773e-01 -1.00328207e+00
2.03138039e-01 3.01029325e-01 -4.43284035e-01 1.85388952e-01
1.52997434e-01 4.64586854e-01 4.56118211e-02 -4.43370730e-01
6.12512171e-01 5.62838972e-01 9.88605559e-01 -2.17755258e-01
8.81355107e-01 3.34845215e-01 -1.53398901e-01 -6.22449338e-01
-7.88697779e-01 -2.55065151e-02 -5.22447288e-01 -2.26817310e-01
1.12308764e+00 -9.66473758e-01 -2.06546471e-01 7.01793969e-01
-1.06322539e+00 -5.64832330e-01 -8.07155252e-01 5.85643649e-02
-8.50488365e-01 4.93644029e-01 -7.11828709e-01 -6.75799906e-01
-6.61364853e-01 -9.45015371e-01 1.12466574e+00 1.56630144e-01
-3.34302746e-02 -1.08684766e+00 2.63511717e-01 4.18484002e-01
4.63705748e-01 6.40910149e-01 9.31052387e-01 -7.50353098e-01
-5.18816710e-01 -2.59509355e-01 1.07286803e-01 1.10241127e+00
-1.10799059e-01 4.20792736e-02 -1.10889852e+00 -4.56752598e-01
1.32363483e-01 -3.85589600e-01 1.12967563e+00 5.07516205e-01
1.10361075e+00 -6.48916900e-01 2.23435040e-04 7.37311244e-01
1.38542247e+00 2.83123285e-01 1.22938609e+00 2.37884328e-01
9.07250464e-01 7.80443698e-02 7.82137960e-02 1.38974875e-01
4.31417257e-01 7.57864773e-01 7.35504150e-01 -3.66406620e-01
-3.96216482e-01 -4.87126887e-01 6.65164530e-01 1.35514021e+00
-4.31258887e-01 -6.41759634e-01 -2.07467437e-01 4.72781003e-01
-1.97305727e+00 -1.11603975e+00 1.94103420e-01 2.09920907e+00
5.05648971e-01 -1.12064756e-01 3.03315848e-01 7.44220614e-02
5.24782717e-01 3.90962631e-01 -5.35144567e-01 -6.85627937e-01
-1.36018135e-02 4.24124211e-01 1.34530202e-01 4.43584979e-01
-1.06297755e+00 8.23494434e-01 6.34819269e+00 8.56588125e-01
-8.83037627e-01 4.47424173e-01 6.11880064e-01 5.12150452e-02
-2.79831260e-01 -2.33532488e-01 -2.10468426e-01 5.92927873e-01
1.09457755e+00 -4.91109118e-02 6.21994853e-01 8.32802653e-01
-1.67547241e-02 3.18448573e-01 -1.19784105e+00 5.41714311e-01
6.15731835e-01 -1.18458581e+00 4.06856298e-01 1.23899423e-01
1.00624406e+00 -1.18644945e-01 -3.36357988e-02 7.26802051e-01
1.85233116e-01 -7.31796980e-01 1.12453592e+00 6.74653530e-01
9.27247643e-01 -1.00082636e+00 9.91410911e-01 3.17869127e-01
-9.51541424e-01 -8.38479027e-02 -3.33653539e-01 2.81566828e-01
3.10560882e-01 9.77211669e-02 -4.15356964e-01 1.08268774e+00
2.45334223e-01 6.30967796e-01 -5.10580003e-01 5.77986240e-01
-6.09380960e-01 9.50753510e-01 2.63100594e-01 6.19723320e-01
2.60609388e-01 -3.99311423e-01 8.09360623e-01 1.13735187e+00
3.05226088e-01 -1.19957998e-01 -6.75833449e-02 9.14483488e-01
-5.17942429e-01 -1.70749813e-01 -3.45088750e-01 -1.43693546e-02
5.25333136e-02 1.15565085e+00 -2.00564608e-01 -6.00638747e-01
-4.33433682e-01 1.56670833e+00 3.93823773e-01 4.32250053e-01
-1.26846027e+00 -2.54127949e-01 2.83041596e-01 8.29656124e-02
9.54569936e-01 1.02007933e-01 2.42485344e-01 -1.12094998e+00
-1.03968360e-01 -1.04944897e+00 2.52120733e-01 -9.55514431e-01
-1.01287842e+00 1.06780100e+00 -1.03273995e-01 -1.03316784e+00
-8.63133371e-01 -3.87156457e-02 -8.76473844e-01 9.57558155e-01
-1.40337491e+00 -1.60115421e+00 -2.50150591e-01 6.08957708e-01
8.00532758e-01 -2.77471036e-01 8.19739223e-01 1.53768554e-01
-8.55665088e-01 8.89923871e-01 2.98290163e-01 9.41848159e-02
3.77161115e-01 -1.33679211e+00 5.71889699e-01 1.15959132e+00
-2.21318863e-02 1.39737830e-01 6.07951283e-01 -4.95020330e-01
-1.03878796e+00 -1.63074911e+00 5.41300237e-01 -4.55206305e-01
4.21899945e-01 -4.38309580e-01 -9.82818067e-01 1.23005331e+00
8.50673437e-01 -3.92961591e-01 5.59106529e-01 -5.68463743e-01
-1.11611374e-01 5.81911169e-02 -1.06159282e+00 4.84873176e-01
8.36333156e-01 -2.61449635e-01 -7.81768382e-01 -2.64574345e-02
9.63517606e-01 -4.30997580e-01 -7.19243526e-01 3.01566303e-01
3.68598521e-01 -1.03236043e+00 7.82587767e-01 -5.74419558e-01
1.00618315e+00 -1.00849986e-01 -3.82432938e-02 -1.60506570e+00
-6.03509188e-01 -8.29875231e-01 -7.81140745e-01 1.54631412e+00
4.74362522e-02 -4.67150003e-01 4.85638201e-01 1.97326168e-01
-6.02167606e-01 -9.32156980e-01 -9.47696984e-01 -7.21455038e-01
1.56376705e-01 -8.29412267e-02 4.68309402e-01 6.04858935e-01
-6.14785850e-01 6.84808016e-01 -9.32566106e-01 -3.55916917e-02
5.51276863e-01 -1.62811831e-01 9.21547234e-01 -6.59912348e-01
-7.02614009e-01 -2.36409351e-01 -2.41704255e-01 -1.03827584e+00
3.22878867e-01 -7.75414884e-01 -4.32228409e-02 -1.68021595e+00
2.68694788e-01 3.61863613e-01 -2.04680905e-01 3.98233503e-01
-5.17783761e-01 3.79404217e-01 3.13432038e-01 2.39099383e-01
-5.69694936e-01 1.01897883e+00 1.04877806e+00 -1.93194017e-01
2.20758822e-02 -3.50791961e-02 -7.82071173e-01 6.04217410e-01
3.61481816e-01 -3.74684244e-01 -5.11752486e-01 -3.81582767e-01
-2.53296942e-01 6.51577190e-02 2.68528730e-01 -1.04522300e+00
-1.33491635e-01 4.81414348e-01 3.12677532e-01 -4.03108686e-01
2.32311189e-01 -7.41976619e-01 4.33852226e-01 4.08064574e-01
-2.75552869e-01 -1.80589817e-02 1.02397576e-01 7.24920392e-01
-2.66803235e-01 -3.37159306e-01 8.93398345e-01 -1.21108936e-02
-3.93018752e-01 7.99623057e-02 -4.05726939e-01 8.86920467e-02
1.13248956e+00 -1.39722258e-01 -1.41753927e-01 -6.13364577e-01
-7.15246379e-01 1.06919268e-02 6.06718421e-01 2.19494373e-01
3.30468833e-01 -1.37970293e+00 -9.75623310e-01 2.55310059e-01
-2.42921174e-01 8.25270787e-02 5.05003929e-01 6.84850574e-01
-5.19496083e-01 2.38532841e-01 -1.08892508e-01 -2.88989007e-01
-9.77389693e-01 8.55287194e-01 2.86280364e-01 -6.87603235e-01
-5.87462962e-01 8.37042689e-01 5.33211172e-01 1.08325943e-01
1.08204350e-01 -3.40651460e-02 -1.83073238e-01 -1.28595442e-01
4.46105361e-01 4.94638652e-01 3.50885540e-02 -8.96754205e-01
-8.76612514e-02 2.88120031e-01 -1.50138259e-01 1.03287883e-02
1.51470709e+00 -1.36957973e-01 4.06546779e-02 2.24544182e-01
1.09072721e+00 1.60713792e-01 -1.66098738e+00 1.88929550e-02
-6.66906178e-01 -1.56271964e-01 1.09309062e-01 -7.48530328e-01
-1.32611215e+00 6.82349324e-01 2.27551162e-01 1.45585582e-01
1.45529568e+00 1.93670467e-02 8.56117487e-01 -1.91081852e-01
1.08833693e-01 -7.04415619e-01 1.81941777e-01 2.69511312e-01
1.21605802e+00 -7.10879803e-01 -1.51607066e-01 -1.49877787e-01
-9.79179978e-01 7.37141192e-01 3.65544617e-01 -4.02300984e-01
2.32975110e-01 1.12832636e-01 -2.01795459e-01 4.76020724e-02
-8.75682414e-01 5.32078147e-02 5.13506413e-01 6.13135219e-01
1.28148079e-01 -2.83651203e-01 -1.29242063e-01 8.41151416e-01
1.61714852e-02 8.12924579e-02 4.41935658e-01 7.21676528e-01
-7.19437897e-02 -9.53541875e-01 -8.46820399e-02 3.03687304e-01
-5.29635787e-01 -1.73660249e-01 -4.14766461e-01 7.53681242e-01
1.11118346e-01 6.85549080e-01 1.02747217e-01 -5.16909122e-01
4.93815273e-01 2.88473815e-01 4.41397816e-01 -4.38857585e-01
-7.85803080e-01 3.13685685e-01 1.50382118e-02 -5.75091541e-01
-5.59391260e-01 -6.22012794e-01 -6.25540674e-01 -1.06079303e-01
-2.57155955e-01 1.92840680e-01 4.40419137e-01 9.20051873e-01
5.01327813e-01 1.41208231e+00 8.71674061e-01 -1.11213124e+00
-6.61264181e-01 -1.17453325e+00 -3.85671437e-01 4.22926754e-01
1.80947185e-01 -4.43437964e-01 -2.91214943e-01 4.16083246e-01]
|
[10.590965270996094, 0.2715260982513428]
|
d9676bcf-f1bb-4544-85a1-d11bfcef41d2
|
data-augmentation-and-squeeze-and-excitation
|
2206.12059
| null |
https://arxiv.org/abs/2206.12059v1
|
https://arxiv.org/pdf/2206.12059v1.pdf
|
Data Augmentation and Squeeze-and-Excitation Network on Multiple Dimension for Sound Event Localization and Detection in Real Scenes
|
Performance of sound event localization and detection (SELD) in real scenes is limited by small size of SELD dataset, due to difficulty in obtaining sufficient amount of realistic multi-channel audio data recordings with accurate label. We used two main strategies to solve problems arising from the small real SELD dataset. First, we applied various data augmentation methods on all data dimensions: channel, frequency and time. We also propose original data augmentation method named Moderate Mixup in order to simulate situations where noise floor or interfering events exist. Second, we applied Squeeze-and-Excitation block on channel and frequency dimensions to efficiently extract feature characteristics. Result of our trained models on the STARSS22 test dataset achieved the best ER, F1, LE, and LR of 0.53, 49.8%, 16.0deg., and 56.2% respectively.
|
['Yong-Hwa Park', 'Seung-Deok Choi', 'Deokki Min', 'Seong-Hu Kim', 'Hyeonuk Nam', 'Byeong-Yun Ko']
|
2022-06-24
| null | null | null | null |
['sound-event-localization-and-detection']
|
['audio']
|
[ 3.31769735e-01 -2.37468213e-01 5.12745738e-01 -1.08265102e-01
-1.23416376e+00 -5.71361899e-01 3.18163544e-01 1.67592540e-01
-3.45916718e-01 8.64666522e-01 3.29892069e-01 3.54610793e-02
1.68409869e-02 -4.17145044e-01 -4.92341518e-01 -5.11942744e-01
-3.47129852e-01 -2.71558940e-01 4.98695165e-01 2.71827243e-02
1.78648695e-01 2.90020049e-01 -1.71125662e+00 3.66345644e-01
4.30767357e-01 1.16217947e+00 3.05747807e-01 1.05406320e+00
3.30196880e-02 5.34994185e-01 -8.37698162e-01 1.36864454e-01
2.76093870e-01 -6.27369225e-01 -2.42845193e-01 -5.11711650e-02
-1.52837196e-02 -2.70090342e-01 -2.16447860e-01 7.88173914e-01
1.04641557e+00 2.82608032e-01 5.39829433e-01 -1.34899867e+00
6.40931353e-02 6.74605906e-01 -9.02861118e-01 4.99517828e-01
4.60736364e-01 -1.76198363e-01 6.27455711e-01 -8.59650373e-01
1.72163665e-01 8.92858148e-01 7.15145469e-01 1.77541092e-01
-1.00231731e+00 -1.08745944e+00 -2.70140886e-01 1.01896919e-01
-1.50589204e+00 -6.57788575e-01 8.18769872e-01 -3.54380399e-01
8.32293749e-01 4.73230273e-01 3.42867106e-01 1.04039776e+00
-2.13105068e-01 5.62583208e-01 1.30806422e+00 -7.49897182e-01
-4.62395437e-02 2.69915938e-01 6.61173165e-02 1.63578555e-01
-2.81471252e-01 2.20103353e-01 -7.78742373e-01 -2.40531102e-01
5.66650033e-01 -4.68633682e-01 -2.44399980e-01 4.91809249e-01
-1.03308797e+00 4.91706818e-01 -2.72623211e-01 1.41739964e-01
-3.22807461e-01 1.98922046e-02 6.50132716e-01 4.25915450e-01
2.41910651e-01 5.10081947e-01 -7.17747808e-01 -5.65720916e-01
-7.15373695e-01 2.39854425e-01 4.06839013e-01 9.18030679e-01
3.28432173e-01 4.22667563e-01 -2.12823302e-01 1.10407460e+00
1.71642959e-01 5.65229714e-01 5.95687091e-01 -7.04474926e-01
5.04860878e-01 -7.37604052e-02 1.61650971e-01 -9.32325304e-01
-6.52539551e-01 -7.03769147e-01 -6.11132860e-01 -2.73791075e-01
2.47821391e-01 -3.17389607e-01 -8.83694828e-01 1.85848427e+00
2.20188990e-01 5.65519929e-01 1.95166111e-01 6.99276209e-01
8.24677408e-01 8.67461443e-01 8.62886533e-02 -4.70423073e-01
1.36072290e+00 -5.82933664e-01 -1.07098818e+00 8.52722004e-02
4.58840787e-01 -1.20866323e+00 1.17586052e+00 8.09277177e-01
-8.85671794e-01 -7.86514342e-01 -9.93976891e-01 4.07397002e-01
-8.94357860e-02 4.44822252e-01 5.82453787e-01 8.21637809e-01
-5.40991426e-01 2.61635393e-01 -5.58867276e-01 -1.08012907e-01
2.77517170e-01 3.80681545e-01 -3.03255796e-01 4.43841398e-01
-1.26643562e+00 1.10439487e-01 3.32429051e-01 3.42683215e-03
-1.04383969e+00 -5.27512372e-01 -3.96456987e-01 -3.68898804e-03
2.65765041e-01 1.82918623e-01 1.18497562e+00 -4.19306427e-01
-1.38502765e+00 4.37119603e-01 9.52607989e-02 -5.29103577e-01
1.27883911e-01 -4.62924808e-01 -1.02906311e+00 3.34766321e-02
-6.92799464e-02 3.53062332e-01 6.55884862e-01 -9.09948349e-01
-8.22147131e-01 -1.81575969e-01 -3.39998543e-01 9.95706469e-02
-4.35990900e-01 2.36622050e-01 -1.98617101e-01 -8.07110727e-01
2.78080404e-01 -6.69150054e-01 -1.94700003e-01 -6.23924196e-01
-6.24010742e-01 3.06394696e-01 8.52492392e-01 -8.15558434e-01
1.45415854e+00 -2.58489609e+00 -4.66893971e-01 1.11475997e-01
-1.61449030e-01 3.67626488e-01 -2.30663195e-01 5.88420928e-01
-3.43313545e-01 1.81231853e-02 1.61947146e-01 -1.70277849e-01
-2.72316396e-01 -1.17406331e-01 -4.31833982e-01 2.11588472e-01
1.53148368e-01 6.80510849e-02 -5.38560987e-01 -2.99248189e-01
1.18524298e-01 3.50381136e-01 -7.15661108e-01 3.04164857e-01
2.52633780e-01 5.66170514e-01 -2.82668769e-01 6.40010655e-01
7.61212647e-01 3.84273827e-01 -4.44602221e-01 -3.92990917e-01
-2.23901898e-01 4.15698677e-01 -1.86429918e+00 1.69859445e+00
-5.39230287e-01 7.53778100e-01 -1.90468617e-02 -8.10505629e-01
1.00610232e+00 7.46717870e-01 4.67279702e-01 -6.18344903e-01
3.28792781e-01 1.45616248e-01 8.64922553e-02 -8.77446592e-01
3.87806594e-01 -9.18591619e-02 -1.88334376e-01 3.48724723e-01
2.73527890e-01 1.54379278e-01 -4.27656434e-03 8.00293460e-02
1.24352300e+00 -1.78104416e-01 2.28392482e-01 2.52894126e-02
4.91847515e-01 -3.67060512e-01 5.32076657e-01 8.25913429e-01
-2.22884849e-01 7.81133652e-01 5.26978850e-01 3.65792327e-02
-8.62557232e-01 -1.05410004e+00 -2.87430167e-01 1.10677850e+00
-2.68291801e-01 -4.81979370e-01 -4.93157864e-01 -2.10413724e-01
-3.01005393e-01 4.93852407e-01 -2.77188897e-01 -1.91701144e-01
-3.01223934e-01 -8.03983271e-01 1.08972061e+00 4.97900814e-01
5.37171423e-01 -9.81825471e-01 -4.90922123e-01 5.49480081e-01
-1.85490921e-01 -1.38295853e+00 -1.50361687e-01 5.14047742e-01
-5.38718760e-01 -8.41436982e-01 -2.06917301e-01 -4.98053879e-01
1.77593246e-01 -6.42686635e-02 6.25917614e-01 -3.37393492e-01
-4.98278081e-01 -4.91578504e-02 -7.07721651e-01 -7.35922039e-01
-2.04854369e-01 -2.14858741e-01 1.52255937e-01 6.10631742e-02
7.69640505e-02 -9.30222511e-01 -5.34419894e-01 3.56401324e-01
-7.18976855e-01 -2.14580163e-01 4.81289327e-01 7.64460862e-01
4.02414948e-01 3.92287105e-01 1.14875066e+00 -6.65617466e-01
6.82339430e-01 -4.37304258e-01 -5.22425294e-01 -3.21942747e-01
-2.74377555e-01 -3.69807988e-01 5.38640141e-01 -7.23291934e-01
-9.40463305e-01 2.36085117e-01 -5.64099431e-01 -6.15790896e-02
-5.00768900e-01 1.91256925e-01 -1.88948095e-01 2.29707867e-01
7.61661649e-01 1.59960091e-01 -4.78303552e-01 -6.51486874e-01
-6.10923991e-02 1.35163116e+00 6.43966734e-01 -4.93323445e-01
5.06581426e-01 1.49724275e-01 -9.68429372e-02 -9.49508667e-01
-6.54515564e-01 -5.82207024e-01 -2.58665115e-01 5.13778115e-03
5.60232401e-01 -1.13539398e+00 -4.64051902e-01 6.42049134e-01
-1.00073242e+00 1.22124881e-01 -1.89311877e-01 9.00820434e-01
-1.89019904e-01 1.52788639e-01 -4.58787352e-01 -1.38742936e+00
-3.36954117e-01 -8.37571681e-01 8.63645077e-01 1.04648501e-01
-2.58923769e-01 -1.92346141e-01 -3.15382853e-02 1.24336198e-01
4.96788770e-01 4.80750054e-01 4.98112768e-01 -8.13605607e-01
-1.52129322e-01 -4.23018336e-01 -4.83760163e-02 6.25045002e-01
2.24367961e-01 -2.01132968e-01 -1.43676317e+00 -2.34150281e-03
2.39244759e-01 -4.00214672e-01 5.18810451e-01 2.56200343e-01
1.31114542e+00 -8.93188044e-02 -1.40682995e-01 3.74886453e-01
1.24705637e+00 7.53354907e-01 6.31794274e-01 -9.85257775e-02
3.42522055e-01 2.22381592e-01 7.30557501e-01 8.09343755e-01
-1.78214461e-01 6.05394125e-01 1.40807018e-01 3.71992365e-02
-2.55589157e-01 -2.77372658e-01 1.74139768e-01 9.03009593e-01
2.78058946e-01 -5.11121094e-01 -7.70374298e-01 6.79858267e-01
-1.33059108e+00 -7.20116079e-01 -4.77852613e-01 2.22253847e+00
7.76446402e-01 3.30493093e-01 1.59916595e-01 8.60964239e-01
6.13726258e-01 -2.37850413e-01 -9.08808261e-02 -2.75711387e-01
-2.40029499e-01 4.37821716e-01 4.05868262e-01 2.13157430e-01
-1.07712901e+00 5.76290846e-01 6.40927696e+00 9.70773876e-01
-1.12749481e+00 2.94226855e-01 4.28904027e-01 -3.40044081e-01
2.48527974e-01 -1.39360368e-01 -7.63466716e-01 6.00555182e-01
1.33815837e+00 2.78820515e-01 3.14106405e-01 4.05119509e-01
3.76337200e-01 -3.70535791e-01 -8.51254761e-01 1.19675648e+00
-1.79746747e-01 -7.64045060e-01 -4.59969044e-01 -6.91253021e-02
5.90253055e-01 -9.67755094e-02 -1.88466664e-02 3.58023584e-01
-2.73389488e-01 -8.90047669e-01 6.32025063e-01 2.24914610e-01
8.26759994e-01 -8.62888694e-01 7.08726048e-01 3.64155829e-01
-1.01266098e+00 -1.84254050e-01 -1.41913161e-01 -2.43857488e-01
2.34146163e-01 6.78947508e-01 -1.01264226e+00 2.98913687e-01
6.77685380e-01 6.11800328e-03 -3.33178014e-01 1.18473232e+00
1.64393976e-01 1.17437243e+00 -9.00060117e-01 5.73541969e-02
1.05332769e-03 4.22056943e-01 5.36098778e-01 1.40703321e+00
5.03223240e-01 1.78537905e-01 5.31042442e-02 2.57263213e-01
7.37803355e-02 1.79931283e-01 -4.16016966e-01 4.49832492e-02
8.46561074e-01 1.29134774e+00 -5.32513440e-01 -3.93139711e-03
-3.07678849e-01 6.41513050e-01 -3.33487153e-01 2.24797100e-01
-9.93958235e-01 -8.47778380e-01 2.46203154e-01 9.53368247e-02
1.10005990e-01 -1.30361579e-02 -4.48299292e-03 -6.61697686e-01
-1.31859317e-01 -8.98392260e-01 3.24943423e-01 -7.91102529e-01
-8.29360366e-01 7.41807282e-01 -1.06487349e-01 -1.28841472e+00
-1.74181402e-01 -3.09877843e-01 -5.38089514e-01 8.14399481e-01
-9.97156143e-01 -9.47804391e-01 -2.74634421e-01 6.54018760e-01
5.70533812e-01 -3.85193080e-01 1.01052165e+00 9.38737988e-01
-4.85693425e-01 8.40385735e-01 -5.88251557e-03 6.67639496e-03
6.94720328e-01 -1.02029717e+00 7.55594596e-02 7.86282897e-01
2.69810140e-01 2.71703720e-01 9.55264986e-01 -2.89668798e-01
-1.21555686e+00 -1.00217295e+00 5.15966654e-01 -1.21402137e-01
5.21023154e-01 -6.92700326e-01 -7.72433817e-01 3.95150304e-01
4.29051928e-02 1.95912972e-01 7.93379247e-01 1.43856287e-01
3.10218986e-02 -3.05036217e-01 -1.24833405e+00 2.72191167e-01
7.41744757e-01 -4.62909043e-01 -2.45274171e-01 2.47112557e-01
5.70916593e-01 -5.34974813e-01 -8.89611244e-01 4.14345086e-01
4.97020155e-01 -7.68515050e-01 8.26465368e-01 -4.55367386e-01
9.99640301e-02 -4.05665427e-01 -7.01786399e-01 -1.17408478e+00
6.63183024e-03 -9.05762792e-01 -6.60170149e-03 1.86196065e+00
5.55625379e-01 -4.39214379e-01 4.49437380e-01 -1.35703042e-01
-3.42869133e-01 -6.58937514e-01 -8.93319666e-01 -7.09405363e-01
-4.67107803e-01 -8.96019459e-01 5.19201696e-01 7.97755182e-01
-1.01357788e-01 4.34878081e-01 -7.24975169e-01 3.62836957e-01
3.23998600e-01 -4.07973081e-01 6.78276896e-01 -1.04660428e+00
-4.91336137e-01 1.90350756e-01 -5.54236233e-01 -7.03517914e-01
-4.22164083e-01 -3.23307872e-01 -1.17013603e-02 -9.31044042e-01
-3.80296372e-02 -5.53667128e-01 -6.91827178e-01 2.84797192e-01
2.48809699e-02 4.87646729e-01 -6.06896915e-02 -1.88309029e-01
-3.82662594e-01 4.18273032e-01 8.92196715e-01 4.00291979e-01
-4.47593123e-01 1.44275352e-01 -5.35284460e-01 7.66154349e-01
9.98055995e-01 -5.98463535e-01 -6.40366435e-01 -2.23233938e-01
4.19805720e-02 3.32955241e-01 2.24841699e-01 -1.38924491e+00
7.21517801e-02 1.56447202e-01 4.49656308e-01 -8.83730710e-01
5.84701240e-01 -7.43435502e-01 3.43348682e-01 7.29843453e-02
-3.52730572e-01 -1.11216381e-02 6.46046042e-01 5.06221414e-01
-3.25701833e-01 -1.40027344e-01 7.43567407e-01 1.11952886e-01
-5.38748503e-01 -2.84169942e-01 -4.15830284e-01 1.04254737e-01
1.01108742e+00 -5.95118776e-02 -2.96283010e-02 -2.96435833e-01
-8.32986832e-01 -6.64023831e-02 -4.89038020e-01 4.70365018e-01
4.40813214e-01 -1.47038317e+00 -7.04684794e-01 5.00780404e-01
3.86365727e-02 -2.82171845e-01 4.40757304e-01 7.75727391e-01
-2.44650990e-01 2.61963367e-01 -2.59278536e-01 -3.99701923e-01
-1.36909795e+00 1.36893764e-01 5.85904159e-02 4.30792905e-02
-3.86531502e-01 1.07498324e+00 -5.59534170e-02 -7.06202816e-03
4.94190335e-01 -4.04035866e-01 -2.10411966e-01 1.04301028e-01
6.00710392e-01 5.52753747e-01 2.75191844e-01 -4.20001388e-01
-4.44681078e-01 3.70692194e-01 4.53879163e-02 -4.94139701e-01
1.26882148e+00 -2.19728306e-01 3.43069881e-01 5.10551631e-01
1.27997959e+00 3.86753857e-01 -8.69155824e-01 1.48591874e-02
-1.60462052e-01 -3.99057657e-01 2.11727098e-01 -1.04572654e+00
-8.73212337e-01 9.19821024e-01 1.14039910e+00 5.03452182e-01
1.56943500e+00 -2.18770072e-01 7.97087491e-01 4.56610322e-02
9.54836234e-02 -1.11160862e+00 -2.06844285e-02 3.47594053e-01
8.48159075e-01 -9.90001678e-01 -1.62781939e-01 -3.07663769e-01
-5.17467022e-01 7.71820664e-01 6.43839717e-01 -1.18215254e-03
7.52744257e-01 7.00063586e-01 2.00067505e-01 -5.39313518e-02
-7.70770311e-01 1.67374313e-02 7.87442625e-02 5.85801482e-01
5.31435966e-01 -1.92936897e-01 -3.04513097e-01 1.02867568e+00
-4.65147555e-01 -6.68284893e-02 6.19681537e-01 8.70622158e-01
-6.16010904e-01 -9.21290874e-01 -6.81354344e-01 6.72885895e-01
-1.07126725e+00 -2.22683661e-02 -1.09636234e-02 7.05893815e-01
3.44396681e-01 1.27997863e+00 -8.29585865e-02 -7.48861849e-01
6.18945837e-01 1.33277833e-01 2.70907760e-01 -3.81455541e-01
-5.24544418e-01 7.49042690e-01 2.15124071e-01 -3.25310022e-01
-7.60300979e-02 -6.70095086e-01 -1.31883609e+00 2.72479624e-01
-6.96248353e-01 4.02126908e-01 7.74226129e-01 6.84905291e-01
2.95194536e-01 1.01694059e+00 7.57223368e-01 -4.16127264e-01
-4.88418579e-01 -1.43712807e+00 -9.48485255e-01 3.22262913e-01
5.55879712e-01 -5.46526730e-01 -4.36150342e-01 1.66494980e-01]
|
[15.198298454284668, 5.236056804656982]
|
c336c5ca-f5b8-4594-a98e-7b05a92eec38
|
fs-net-fast-shape-based-network-for-category
|
2103.07054
| null |
https://arxiv.org/abs/2103.07054v2
|
https://arxiv.org/pdf/2103.07054v2.pdf
|
FS-Net: Fast Shape-based Network for Category-Level 6D Object Pose Estimation with Decoupled Rotation Mechanism
|
In this paper, we focus on category-level 6D pose and size estimation from monocular RGB-D image. Previous methods suffer from inefficient category-level pose feature extraction which leads to low accuracy and inference speed. To tackle this problem, we propose a fast shape-based network (FS-Net) with efficient category-level feature extraction for 6D pose estimation. First, we design an orientation aware autoencoder with 3D graph convolution for latent feature extraction. The learned latent feature is insensitive to point shift and object size thanks to the shift and scale-invariance properties of the 3D graph convolution. Then, to efficiently decode category-level rotation information from the latent feature, we propose a novel decoupled rotation mechanism that employs two decoders to complementarily access the rotation information. Meanwhile, we estimate translation and size by two residuals, which are the difference between the mean of object points and ground truth translation, and the difference between the mean size of the category and ground truth size, respectively. Finally, to increase the generalization ability of FS-Net, we propose an online box-cage based 3D deformation mechanism to augment the training data. Extensive experiments on two benchmark datasets show that the proposed method achieves state-of-the-art performance in both category- and instance-level 6D object pose estimation. Especially in category-level pose estimation, without extra synthetic data, our method outperforms existing methods by 6.3% on the NOCS-REAL dataset.
|
['Ales Leonardis', 'Linlin Shen', 'Jinming Duan', 'Hyung Jin Chang', 'Xi Jia', 'Wei Chen']
|
2021-03-12
| null |
http://openaccess.thecvf.com//content/CVPR2021/html/Chen_FS-Net_Fast_Shape-Based_Network_for_Category-Level_6D_Object_Pose_Estimation_CVPR_2021_paper.html
|
http://openaccess.thecvf.com//content/CVPR2021/papers/Chen_FS-Net_Fast_Shape-Based_Network_for_Category-Level_6D_Object_Pose_Estimation_CVPR_2021_paper.pdf
|
cvpr-2021-1
|
['6d-pose-estimation-using-rgbd']
|
['computer-vision']
|
[-1.69607297e-01 -1.93345115e-01 -1.07176900e-01 -3.60517859e-01
-4.67926323e-01 -4.46091413e-01 2.62066454e-01 -2.01103404e-01
-3.86706144e-01 2.78416842e-01 -7.52211735e-02 2.49674208e-02
-2.01776368e-03 -8.43813658e-01 -1.22467852e+00 -8.69166315e-01
1.17742650e-01 5.23249149e-01 9.27264467e-02 1.69966444e-01
2.33529627e-01 7.97864676e-01 -1.42330587e+00 -1.50047988e-01
4.92348999e-01 1.54628313e+00 1.40056744e-01 1.22470893e-01
2.09876999e-01 3.17493439e-01 -4.75582302e-01 -2.53485553e-02
4.95862305e-01 7.62393102e-02 -3.43225837e-01 3.58647704e-01
6.49867713e-01 -7.37551391e-01 -7.20270634e-01 9.61052418e-01
6.87806308e-01 2.59513911e-02 6.62718654e-01 -1.17673647e+00
-5.79736769e-01 1.89890668e-01 -6.83959067e-01 -2.82626301e-01
2.07216114e-01 1.88238055e-01 5.87957680e-01 -1.15382719e+00
6.98504686e-01 1.28659308e+00 5.92853308e-01 4.21583742e-01
-9.38917816e-01 -8.24806213e-01 2.29913726e-01 -9.43871122e-03
-1.47791445e+00 1.17265293e-02 1.08390677e+00 -3.33050162e-01
1.09286892e+00 -1.41427979e-01 6.57864630e-01 9.14490461e-01
2.20134243e-01 6.90838993e-01 9.37892020e-01 3.04711913e-03
1.80995107e-01 -3.35248679e-01 -3.35997820e-01 7.70301580e-01
4.84757394e-01 1.67736430e-02 -4.59712237e-01 2.07729802e-01
1.41011608e+00 3.05253029e-01 -1.79495826e-01 -8.78986835e-01
-1.47226930e+00 5.03362477e-01 9.37616527e-01 -4.28855941e-02
-3.82693380e-01 4.69399124e-01 2.38670141e-01 -2.38651726e-02
4.12443221e-01 6.46449253e-02 -6.66958570e-01 6.69462420e-03
-4.26876605e-01 3.40530246e-01 4.94600236e-01 1.24329638e+00
8.15168858e-01 1.03268094e-01 -1.22052722e-01 5.57754219e-01
5.63644409e-01 9.90551233e-01 3.76721442e-01 -8.05946529e-01
5.80414891e-01 1.04900241e+00 -1.62457228e-02 -1.29816473e+00
-5.38842738e-01 -6.54233694e-01 -9.97669995e-01 -9.25284848e-02
2.89425611e-01 2.80805588e-01 -1.13322377e+00 1.69311082e+00
5.45089900e-01 1.47248179e-01 -2.62154639e-01 1.13680780e+00
9.26368952e-01 3.42847466e-01 -3.35875332e-01 1.46240657e-02
1.26731765e+00 -6.07860148e-01 -3.40034693e-01 -2.06433937e-01
5.35856187e-01 -6.18794441e-01 7.90572703e-01 1.17588826e-01
-8.77723932e-01 -6.85297668e-01 -1.15999305e+00 -2.59591788e-01
-2.31471226e-01 4.67427045e-01 6.95085764e-01 2.53448039e-01
-4.33756173e-01 4.46285486e-01 -1.01505864e+00 6.57670200e-02
3.86041731e-01 7.30917156e-01 -4.62586641e-01 -9.76050869e-02
-7.38626003e-01 6.36971176e-01 4.47381556e-01 2.12383956e-01
-6.65891707e-01 -6.16224170e-01 -9.92569268e-01 -8.16591233e-02
4.67218906e-01 -8.12721491e-01 1.01012933e+00 -2.02391505e-01
-1.64965248e+00 7.53056169e-01 1.20549336e-01 -1.31334901e-01
3.78466368e-01 -2.86257535e-01 1.31930843e-01 2.67033242e-02
-1.19147329e-02 7.98650026e-01 1.00584984e+00 -1.11683178e+00
-2.04757512e-01 -9.46094573e-01 -5.12122400e-02 3.77408326e-01
-2.01443836e-01 -6.92521811e-01 -7.15465248e-01 -6.51579201e-01
9.42792535e-01 -1.04680502e+00 7.58252442e-02 3.32217127e-01
-2.90973395e-01 -3.03553492e-01 9.52699006e-01 -4.67264831e-01
6.49449468e-01 -2.26752353e+00 2.33587325e-01 7.30562359e-02
3.11697960e-01 -3.97137590e-02 5.21089584e-02 -2.00347394e-01
9.49731767e-02 -2.11299092e-01 -5.12062758e-02 -3.26569319e-01
1.52690142e-01 2.62494326e-01 -1.51492432e-01 7.40106702e-01
2.42790371e-01 1.14958560e+00 -7.24390686e-01 -2.04387397e-01
3.95105958e-01 6.98326170e-01 -7.25402653e-01 1.89774513e-01
-1.49517521e-01 3.02361041e-01 -5.96480668e-01 8.72720838e-01
1.09327590e+00 -2.90031999e-01 -6.77016973e-02 -7.27572680e-01
7.34875053e-02 3.46253097e-01 -1.29384565e+00 2.18996835e+00
-4.62618947e-01 9.77537781e-02 -2.32684866e-01 -9.31809604e-01
1.14245570e+00 -2.69977674e-02 5.32944262e-01 -6.78371429e-01
5.41103005e-01 3.98528337e-01 -1.48358032e-01 -4.11248431e-02
2.35553712e-01 1.72283664e-01 -2.51494229e-01 2.61043280e-01
2.04709783e-01 -4.57493842e-01 -2.88788736e-01 -1.67935088e-01
7.81161368e-01 5.01362920e-01 6.53882623e-02 5.17516099e-02
4.98516113e-01 -3.90495569e-01 5.59373975e-01 2.98010349e-01
-8.53447020e-02 5.50847650e-01 2.72476166e-01 -5.89623213e-01
-9.76499081e-01 -1.21329355e+00 -1.40757129e-01 4.18535620e-01
5.10066926e-01 -6.29984960e-02 -5.37143707e-01 -7.97455788e-01
4.25184071e-01 1.86523080e-01 -5.06859481e-01 -4.24527586e-01
-9.00789380e-01 -5.84400058e-01 2.76482254e-01 9.36252952e-01
8.29107165e-01 -6.60096109e-01 -6.57794893e-01 6.23956062e-02
-2.55823843e-02 -1.45089531e+00 -4.70024407e-01 1.96909800e-01
-1.29425526e+00 -7.83602178e-01 -5.72274506e-01 -8.09051037e-01
9.11005020e-01 2.88676411e-01 7.32555330e-01 -2.77809590e-01
-1.50786147e-01 1.17732538e-03 -1.43057466e-01 -1.62321329e-01
2.34231830e-01 1.44045189e-01 3.01649868e-01 -1.58756211e-01
3.05402994e-01 -7.34037876e-01 -8.67577076e-01 6.33081198e-01
-6.49894893e-01 1.75718784e-01 9.57663536e-01 7.42071569e-01
9.71974194e-01 -1.96504891e-01 1.78301543e-01 -6.09701015e-02
-6.03692792e-02 4.54418696e-02 -7.57396460e-01 -2.74433583e-01
-3.47148210e-01 2.70691335e-01 3.33559841e-01 -6.50904894e-01
-6.81536019e-01 3.55257064e-01 3.01653016e-02 -7.92430162e-01
-1.11554936e-01 2.55908847e-01 -3.91683161e-01 -3.00251245e-01
3.59795839e-01 3.13463658e-01 -2.22066399e-02 -5.23550451e-01
3.52379411e-01 6.31649435e-01 6.27061725e-01 -4.84162390e-01
1.11079705e+00 5.60254931e-01 3.92971218e-01 -4.86997932e-01
-7.89715171e-01 -2.78830141e-01 -8.14022303e-01 -1.58505529e-01
7.09123969e-01 -1.24110234e+00 -1.09382391e+00 8.02208364e-01
-1.18087173e+00 -1.12385668e-01 -1.33910954e-01 7.82101274e-01
-7.40440607e-01 1.73201039e-01 -6.19147360e-01 -4.34699237e-01
-4.82886255e-01 -1.34685338e+00 1.64395535e+00 6.05546832e-02
3.22071314e-01 -4.67299640e-01 -4.07703757e-01 1.85373783e-01
1.20061554e-01 5.11512578e-01 7.57906437e-01 -1.68160766e-01
-1.05394936e+00 -4.32733685e-01 -5.16041517e-01 3.46324384e-01
1.53518647e-01 -5.28376877e-01 -6.02312565e-01 -3.55051190e-01
1.81738302e-01 -3.44144464e-01 5.63533068e-01 4.19578940e-01
1.45241261e+00 -1.19507477e-01 -2.90495992e-01 1.02752435e+00
1.42556214e+00 -1.66796044e-01 3.77909213e-01 2.04723611e-01
1.15818942e+00 2.83214618e-02 7.27845371e-01 4.92212743e-01
4.54268485e-01 7.83387125e-01 7.77699590e-01 3.07129174e-01
-2.51479417e-01 -4.47826207e-01 2.69520491e-01 1.28640115e+00
-2.77175248e-01 5.06109409e-02 -6.21431291e-01 2.43678913e-01
-1.59306836e+00 -3.54000717e-01 1.02728315e-01 2.22571206e+00
6.39469624e-01 3.98360282e-01 -1.24920405e-01 2.10748762e-01
5.55059314e-01 7.53551871e-02 -9.30288851e-01 1.69260576e-01
9.56629291e-02 1.01394698e-01 8.46064627e-01 4.74211909e-02
-9.92688656e-01 9.43903148e-01 4.63184071e+00 8.96912456e-01
-1.38835442e+00 -1.26073062e-01 1.77673101e-01 -1.12055570e-01
-4.07668725e-02 -2.66203374e-01 -8.99814904e-01 3.14127862e-01
3.40670288e-01 3.53996992e-01 2.94458061e-01 1.06335104e+00
-1.49853542e-01 1.45285591e-01 -1.17410684e+00 1.20381200e+00
2.20340475e-01 -9.95769322e-01 1.35692343e-01 1.95008174e-01
5.49962521e-01 7.42939785e-02 8.96946341e-02 2.74944216e-01
-3.06269646e-01 -7.31008232e-01 8.26191545e-01 2.60094702e-01
1.07254529e+00 -8.90527070e-01 7.63372302e-01 3.44407231e-01
-1.33339787e+00 7.05972780e-03 -5.40089667e-01 -1.22091405e-01
-7.51473755e-02 6.27676964e-01 -7.13188946e-01 6.63798571e-01
7.25483239e-01 8.21226418e-01 -3.99638116e-01 7.60894239e-01
-3.70732486e-01 7.30041638e-02 -5.94467938e-01 -7.75443688e-02
-4.21746299e-02 1.24264106e-01 5.54476738e-01 5.72074950e-01
4.09440249e-01 -1.33129545e-02 6.17760643e-02 8.50906432e-01
-3.73350114e-01 -2.16847643e-01 -3.37071151e-01 4.89502735e-02
5.28289557e-01 1.12997186e+00 -7.34542012e-01 -9.89224762e-02
-2.00837687e-01 1.07799244e+00 1.84853300e-01 1.44980431e-01
-9.12844896e-01 -4.58315849e-01 5.72339773e-01 5.83768487e-02
6.37854338e-01 -6.47395015e-01 -3.95704389e-01 -1.31867623e+00
4.03505564e-01 -5.87599456e-01 -7.64576197e-02 -8.41278255e-01
-9.25827563e-01 4.21325058e-01 -1.26348371e-02 -1.41528630e+00
-8.29941407e-02 -1.12726796e+00 5.94438463e-02 5.20178974e-01
-1.32012618e+00 -1.45182180e+00 -7.07722902e-01 4.43924248e-01
4.01892990e-01 8.34970698e-02 5.81241608e-01 3.55302453e-01
-2.90052652e-01 7.43387222e-01 -2.05348149e-01 3.86540264e-01
6.63644016e-01 -9.56216931e-01 5.91113806e-01 3.59458148e-01
5.84778488e-02 5.80705881e-01 2.12084591e-01 -6.26049697e-01
-2.09662080e+00 -1.12707949e+00 5.18814027e-01 -4.38561857e-01
3.29775214e-01 -6.60807133e-01 -5.73397934e-01 6.06036067e-01
-6.46958709e-01 3.69002670e-01 6.17128685e-02 -2.68631667e-01
-5.44055462e-01 -3.34689707e-01 -1.08598876e+00 3.17056626e-01
1.49730027e+00 -5.34934878e-01 -5.73448122e-01 1.92724124e-01
1.10070729e+00 -1.09199917e+00 -1.07383084e+00 9.36840177e-01
7.38552034e-01 -6.37238503e-01 1.22154498e+00 4.46459651e-03
4.35367972e-01 -5.42218864e-01 -3.92946512e-01 -9.34886456e-01
-3.13539058e-01 -2.33345740e-02 -6.10340118e-01 8.01013410e-01
-1.57572925e-01 -5.10014176e-01 9.60174739e-01 8.84329528e-03
-2.29042515e-01 -1.10948706e+00 -9.86928940e-01 -8.28387141e-01
-1.48221016e-01 -4.24406052e-01 8.24117124e-01 5.90028107e-01
-7.17352033e-01 2.54320711e-01 -1.20031089e-01 3.14693004e-01
7.50035822e-01 5.03256977e-01 1.03965342e+00 -1.16596091e+00
-1.73909515e-01 -2.26216227e-01 -8.79605055e-01 -1.63473916e+00
2.15736628e-02 -7.62369812e-01 7.12961331e-02 -1.22672367e+00
4.34817076e-02 -1.54629245e-01 -1.90490365e-01 5.02925038e-01
1.77257314e-01 4.62903947e-01 7.45493621e-02 5.02889678e-02
-3.78172487e-01 7.72396326e-01 1.70711160e+00 -3.72352488e-02
-1.95106506e-01 -2.29927167e-01 -3.01518112e-01 7.52856433e-01
5.23207366e-01 -4.39079940e-01 -1.00886092e-01 -6.85979366e-01
2.91130245e-01 -1.50761411e-01 6.72549844e-01 -1.12746227e+00
1.07207946e-01 -4.24135402e-02 7.93577373e-01 -1.24078989e+00
3.87560636e-01 -8.17978084e-01 -1.68875441e-01 5.66001415e-01
1.28669605e-01 -4.43385988e-02 1.80381671e-01 6.21140242e-01
3.39505053e-03 2.51318753e-01 6.24306679e-01 -2.86125168e-02
-5.46565831e-01 8.43948126e-01 2.04676211e-01 -2.51474559e-01
8.36510837e-01 -3.16983432e-01 -1.97227165e-01 -5.39197586e-02
-3.77812624e-01 5.96667156e-02 5.53253293e-01 5.10569990e-01
7.46148825e-01 -1.76253676e+00 -4.23159570e-01 3.97875309e-01
2.72394270e-01 8.40994596e-01 2.86666572e-01 8.62047017e-01
-6.01697862e-01 4.40820783e-01 -2.48130187e-01 -9.47991312e-01
-7.66274631e-01 4.74766463e-01 2.04041004e-01 3.96337546e-02
-3.82478535e-01 8.79725099e-01 8.70718509e-02 -7.96794176e-01
3.32530707e-01 -7.32267141e-01 2.71627098e-01 -2.08518654e-01
1.31850958e-01 2.34812379e-01 1.97465599e-01 -7.31125414e-01
-6.06723428e-01 1.05832767e+00 1.47248611e-01 2.96801329e-01
1.47495365e+00 -8.25613644e-03 -7.54102170e-02 2.93388993e-01
1.51665294e+00 -2.79302210e-01 -1.41692877e+00 -2.16146737e-01
-5.28252602e-01 -5.66658854e-01 2.51724899e-01 -5.24062097e-01
-1.22190392e+00 1.00273967e+00 7.59959936e-01 -4.40783262e-01
1.09167194e+00 -6.42621249e-04 9.51342702e-01 5.18001735e-01
4.49978054e-01 -9.50531602e-01 2.95430779e-01 5.53052366e-01
9.97451067e-01 -1.20022047e+00 3.29554588e-01 -5.45075476e-01
8.90406966e-03 1.05543816e+00 9.56677437e-01 -4.42023247e-01
6.06593013e-01 5.95050529e-02 -2.48247281e-01 -2.07821682e-01
-2.60279387e-01 -1.53839253e-02 6.14818573e-01 4.61860567e-01
5.63848540e-02 1.15915909e-01 -6.40908331e-02 5.29258370e-01
-2.49711186e-01 -1.28479257e-01 -3.59763228e-03 8.99150848e-01
-2.61395603e-01 -8.42926204e-01 -3.00556809e-01 2.64346153e-01
-1.33767247e-01 2.64124513e-01 -2.21223921e-01 7.92275012e-01
2.66941398e-01 1.86143011e-01 2.96795785e-01 -7.10558414e-01
3.88004124e-01 -2.25805730e-01 1.04575813e+00 -3.99499714e-01
2.64168475e-02 2.06848457e-01 -4.12901759e-01 -7.37446427e-01
-4.70142603e-01 -3.51926953e-01 -1.37297821e+00 -7.67038539e-02
-6.23093963e-01 -4.41447735e-01 1.01526737e+00 9.22933280e-01
4.42157924e-01 4.12799388e-01 6.07940614e-01 -1.22768545e+00
-7.37134993e-01 -9.63590682e-01 -4.24607188e-01 3.75934154e-01
1.66690782e-01 -1.03034163e+00 -3.48536402e-01 -2.46839017e-01]
|
[7.456226348876953, -2.747234582901001]
|
20360a71-ff6d-4a1f-881a-e19ffd0f9f31
|
continuous-emotional-intensity-controllable
|
2211.0616
| null |
https://arxiv.org/abs/2211.06160v2
|
https://arxiv.org/pdf/2211.06160v2.pdf
|
Semi-supervised learning for continuous emotional intensity controllable speech synthesis with disentangled representations
|
Recent text-to-speech models have reached the level of generating natural speech similar to what humans say. But there still have limitations in terms of expressiveness. The existing emotional speech synthesis models have shown controllability using interpolated features with scaling parameters in emotional latent space. However, the emotional latent space generated from the existing models is difficult to control the continuous emotional intensity because of the entanglement of features like emotions, speakers, etc. In this paper, we propose a novel method to control the continuous intensity of emotions using semi-supervised learning. The model learns emotions of intermediate intensity using pseudo-labels generated from phoneme-level sequences of speech information. An embedding space built from the proposed model satisfies the uniform grid geometry with an emotional basis. The experimental results showed that the proposed method was superior in controllability and naturalness.
|
['Kyogu Lee', 'Yoseob Han', 'Juheon Lee', 'Yoori Oh']
|
2022-11-11
| null | null | null | null |
['emotional-speech-synthesis']
|
['speech']
|
[-7.44139180e-02 4.59428042e-01 -1.83588654e-01 -3.92080754e-01
-2.33587012e-01 -4.00590092e-01 7.97088802e-01 -3.18218708e-01
4.79390882e-02 8.51313949e-01 6.34222806e-01 3.24077010e-01
-1.70551836e-02 -7.92898893e-01 -2.93894470e-01 -7.82614291e-01
7.41039962e-02 1.53387532e-01 -3.09715271e-01 -4.67380702e-01
3.09494603e-03 3.84896994e-01 -1.80047631e+00 3.05186361e-01
1.06169105e+00 7.50923693e-01 3.22542377e-02 6.48431420e-01
-4.88827705e-01 8.55948269e-01 -6.66275918e-01 5.84967434e-02
-5.19950986e-02 -7.26310909e-01 -4.12484765e-01 1.87307179e-01
-5.44476271e-01 1.15395054e-01 -2.31420472e-01 1.08945727e+00
5.02638459e-01 3.33957940e-01 7.69977033e-01 -1.69891620e+00
-9.09658968e-01 5.83181381e-01 7.64304921e-02 -5.20683587e-01
4.15547520e-01 -1.54577523e-01 9.11309302e-01 -7.77566314e-01
5.90747416e-01 1.43758368e+00 3.50467145e-01 8.81530941e-01
-1.17489588e+00 -7.44830787e-01 1.17362980e-02 3.22736427e-02
-1.30665791e+00 -4.25099015e-01 1.07874918e+00 -4.04054999e-01
1.03538179e+00 4.17966038e-01 9.65265214e-01 1.31591606e+00
8.68364945e-02 4.86280411e-01 1.44090343e+00 -6.68516159e-01
4.15130049e-01 7.77213037e-01 -1.69398114e-01 5.83959103e-01
-6.50170386e-01 2.73722917e-01 -5.83783150e-01 -1.28453687e-01
5.87110579e-01 -4.11774248e-01 -1.23001290e-02 -1.75686888e-02
-1.19190812e+00 1.14160597e+00 1.29139856e-01 6.67955518e-01
-2.53875256e-01 -1.00475729e-01 4.95998800e-01 4.64997530e-01
7.45080173e-01 5.58634102e-01 -3.10753196e-01 -2.87681669e-01
-7.87885368e-01 -1.95865631e-01 8.76953661e-01 1.03425348e+00
5.34517705e-01 6.01080596e-01 -2.80033541e-03 9.43782568e-01
2.89099067e-01 4.84131277e-01 7.64040828e-01 -9.62232411e-01
9.64707509e-02 4.58556116e-01 4.76562306e-02 -1.08218563e+00
-3.02863330e-01 -9.18186009e-02 -1.03757334e+00 4.09267277e-01
-3.35153639e-01 -4.72119421e-01 -5.90241790e-01 2.00833058e+00
1.92626819e-01 -2.57910741e-03 5.85800469e-01 7.00485885e-01
6.11615181e-01 1.54447329e+00 8.37065876e-02 -8.27328026e-01
9.04234529e-01 -1.02236891e+00 -1.74145508e+00 1.75961822e-01
3.73102635e-01 -6.89391434e-01 1.47962284e+00 4.41289514e-01
-9.74812329e-01 -7.85625696e-01 -1.11929417e+00 2.73526162e-01
-6.48707211e-01 1.91038311e-01 6.42608941e-01 8.62537265e-01
-1.16139317e+00 3.59954596e-01 -3.66579115e-01 -1.00718282e-01
-4.09227014e-01 2.96669126e-01 -3.19514662e-01 9.91312981e-01
-1.93353856e+00 8.09585750e-01 3.99698645e-01 6.21429123e-02
-4.53536838e-01 -2.56718427e-01 -9.86604452e-01 1.54871345e-01
-7.91852996e-02 -3.16394031e-01 9.43753421e-01 -1.46589279e+00
-2.58267069e+00 3.73598129e-01 -1.90599542e-02 3.12003978e-02
2.66732872e-01 7.03207180e-02 -7.45462596e-01 1.64982319e-01
-1.65601850e-01 8.27744424e-01 9.06648397e-01 -1.15201414e+00
-2.25174963e-01 1.54538706e-01 -2.10382894e-01 1.84332579e-01
-7.43968904e-01 1.57386348e-01 1.88766807e-01 -6.76002502e-01
-1.44767642e-01 -9.81061816e-01 -1.71777889e-01 -9.33785290e-02
-8.09329301e-02 -2.87523210e-01 8.98788452e-01 -1.65508628e-01
1.33108354e+00 -2.22973895e+00 4.79386926e-01 -1.85059775e-02
-1.65403143e-01 6.22636601e-02 9.27948672e-03 7.63385177e-01
-3.49101871e-01 4.52237934e-01 8.20818990e-02 -2.82815576e-01
3.22133362e-01 3.24847728e-01 -6.25314891e-01 4.02364545e-02
1.56515598e-01 5.89605749e-01 -8.21879566e-01 -7.82582104e-01
4.53766614e-01 7.42802083e-01 -5.73790848e-01 6.32178187e-01
-2.51260728e-01 4.53895748e-01 -4.22203988e-01 6.78030699e-02
3.32177490e-01 2.23378256e-01 -1.52715340e-01 -1.24292690e-02
-4.76794869e-01 -6.76511601e-02 -1.23751366e+00 1.36742818e+00
-9.12735701e-01 4.66095239e-01 5.71271479e-02 -7.22022891e-01
1.41889453e+00 9.70949471e-01 5.10471582e-01 -2.42579266e-01
2.24584743e-01 4.26819250e-02 -2.70410359e-01 -7.93469548e-01
3.55356246e-01 -7.31441200e-01 -4.43268776e-01 1.11018538e-01
1.10683441e-01 -8.43154371e-01 -2.26063654e-01 -4.14812267e-02
3.80707860e-01 -6.24658614e-02 3.46424609e-01 -1.82892725e-01
7.89952457e-01 -4.49137479e-01 5.27026355e-01 1.13873668e-01
-1.03388749e-01 2.84469336e-01 6.15898252e-01 -4.69912402e-02
-1.14239073e+00 -9.39740181e-01 -1.78158090e-01 7.61616170e-01
-1.48059711e-01 -5.27279854e-01 -8.43288660e-01 -2.59302020e-01
-6.25604093e-01 1.07070982e+00 -5.59174120e-01 -3.55776995e-01
-6.50724024e-02 -2.56375790e-01 6.11994982e-01 2.85517037e-01
3.83683562e-01 -1.41889608e+00 -3.07856560e-01 2.08987221e-01
-3.64234000e-01 -1.10298765e+00 -5.18575191e-01 1.74235180e-01
-4.14681345e-01 -2.84532760e-03 -3.77322108e-01 -8.45089078e-01
5.46322346e-01 -5.29238582e-01 3.91278088e-01 -3.91038746e-01
-9.28001925e-02 1.99362993e-01 -5.78154445e-01 -3.41826379e-01
-8.45476091e-01 -1.54705182e-01 4.41240549e-01 2.97993392e-01
-1.57113016e-01 -6.48923576e-01 -3.63473445e-02 1.29658923e-01
-1.04088938e+00 4.73101616e-01 1.78843603e-01 9.45210993e-01
2.67372221e-01 1.89180598e-01 1.15969622e+00 -3.97558570e-01
1.14750791e+00 -2.58238286e-01 -3.11640769e-01 1.33858055e-01
-5.36889255e-01 3.79183322e-01 1.01867902e+00 -7.38047421e-01
-1.50577807e+00 4.65129539e-02 -3.51163059e-01 -3.09112966e-01
-2.63285637e-01 2.32343391e-01 -4.45849091e-01 3.20418239e-01
3.57710451e-01 1.06130391e-01 1.87856197e-01 1.54283315e-01
7.21758723e-01 1.20998967e+00 5.82377873e-02 -6.35349572e-01
4.69465107e-01 7.90177286e-02 2.69495025e-02 -1.03461254e+00
-3.96334648e-01 6.56145960e-02 -5.24473369e-01 -5.12495935e-01
1.19779634e+00 -7.05499649e-01 -7.34639466e-01 2.93890566e-01
-1.23562050e+00 -2.52309144e-01 -6.03780806e-01 7.64446557e-01
-1.10000873e+00 3.01013172e-01 -9.03471708e-01 -1.17654550e+00
-3.05999696e-01 -1.14292037e+00 9.63745594e-01 1.22982070e-01
-3.86043310e-01 -1.11704695e+00 1.82056457e-01 -2.81238966e-02
6.10687852e-01 3.74289721e-01 1.01979244e+00 -2.24553332e-01
-1.86774414e-02 -3.51662971e-02 4.51395780e-01 5.50268948e-01
4.28365946e-01 4.31305587e-01 -9.50148344e-01 1.30892336e-01
4.08366799e-01 -6.77960157e-01 1.63530856e-01 6.50792494e-02
9.30516899e-01 -5.94272196e-01 1.60098642e-01 5.15989125e-01
1.10411501e+00 4.24854606e-01 4.66119230e-01 -1.14254780e-01
1.07893564e-01 9.21129227e-01 6.80085361e-01 5.69839299e-01
-1.88129425e-01 5.98220706e-01 3.91732007e-02 -1.98606625e-01
1.16156004e-01 -4.72313076e-01 6.69916928e-01 1.72317946e+00
2.34180957e-01 -2.42890373e-01 -4.76935238e-01 2.22191319e-01
-1.74725926e+00 -1.21275961e+00 1.51707530e-01 1.73711920e+00
1.16688335e+00 1.73025355e-01 -2.31904551e-01 2.88402975e-01
6.97983027e-01 2.69731462e-01 -1.35845125e-01 -1.18825603e+00
-3.00723195e-01 7.39294803e-03 -1.00818336e-01 1.02620709e+00
-7.25866377e-01 1.08525836e+00 6.69200039e+00 9.23358321e-01
-1.31108713e+00 7.09096268e-02 5.23340225e-01 3.10010510e-03
-6.25291467e-01 -1.46284357e-01 -5.83142221e-01 4.05689597e-01
1.16560638e+00 -4.34587002e-01 5.19746363e-01 7.68741131e-01
6.37069583e-01 3.53392631e-01 -7.83200443e-01 8.87258708e-01
2.04711333e-01 -8.89747739e-01 1.55726358e-01 -2.88740963e-01
8.61075342e-01 -7.26621032e-01 2.47023687e-01 5.07305026e-01
-1.11280255e-01 -1.09494507e+00 7.00552046e-01 8.79957616e-01
1.20948052e+00 -9.58884716e-01 3.70613456e-01 7.24206328e-01
-1.09457564e+00 -2.99210828e-02 -1.68398336e-01 -3.15805614e-01
3.56163651e-01 2.03264400e-01 -7.06232071e-01 1.84656277e-01
1.60446301e-01 2.73648977e-01 3.46775092e-02 -2.32303236e-02
-4.23094153e-01 6.06833518e-01 -2.55331248e-01 -6.78905964e-01
3.32143933e-01 -4.95677888e-01 5.04515886e-01 1.31059170e+00
4.96838391e-01 2.35232636e-01 3.00814640e-02 9.16964173e-01
3.03128898e-01 5.24327099e-01 -9.71754849e-01 -3.53834897e-01
4.11084384e-01 1.29968905e+00 -4.40331668e-01 -3.90202343e-01
-2.32039347e-01 1.05931425e+00 -3.23358737e-02 3.13047171e-01
-1.08109546e+00 -6.35167420e-01 3.51861387e-01 -9.12231505e-02
-2.64402688e-01 -3.62042665e-01 -1.88504249e-01 -1.05886698e+00
-3.26397985e-01 -8.81029427e-01 -2.97998935e-01 -1.06206989e+00
-1.06234300e+00 1.04753911e+00 1.33694664e-01 -1.18042195e+00
-5.95204055e-01 -4.34854984e-01 -4.60645318e-01 7.95263290e-01
-8.44881892e-01 -1.16705287e+00 -1.24522246e-01 6.29283071e-01
7.42879868e-01 -4.52523530e-01 1.37778509e+00 -3.02332342e-02
-3.19635063e-01 5.47988415e-01 -5.49304448e-02 -3.01105052e-01
6.63206100e-01 -1.22196198e+00 -2.24751949e-01 3.30355406e-01
-1.55496046e-01 5.37637651e-01 9.82714176e-01 -5.42208552e-01
-9.97843266e-01 -7.41693556e-01 1.12506616e+00 4.24834006e-02
7.15584338e-01 -7.10567355e-01 -5.90660632e-01 3.35992068e-01
7.85227418e-01 -2.51256257e-01 8.14654052e-01 -1.30553469e-01
1.08483555e-02 -5.53455539e-02 -1.19674718e+00 8.18773031e-01
7.67660260e-01 -6.51314497e-01 -7.11911201e-01 2.73953438e-01
1.22377241e+00 8.64221975e-02 -8.90265942e-01 4.35301900e-01
4.20869529e-01 -6.54738009e-01 6.13778114e-01 -5.56316137e-01
3.74439389e-01 -1.75108582e-01 -1.84074491e-01 -1.63904989e+00
-4.90345091e-01 -9.87046957e-01 2.84019798e-01 1.47373819e+00
3.11821222e-01 -7.13766873e-01 4.01608855e-01 4.95603442e-01
-3.37617472e-02 -7.26493120e-01 -9.89590287e-01 -6.43001139e-01
1.40563190e-01 -1.29564255e-01 6.53384387e-01 9.63341415e-01
6.96761608e-01 6.64933026e-01 -6.71829402e-01 6.24881648e-02
1.76148951e-01 -2.90280804e-02 4.54679877e-01 -9.26773190e-01
-2.25583270e-01 -3.36992502e-01 -2.44299576e-01 -5.91117918e-01
8.68144512e-01 -9.10848618e-01 1.21256195e-01 -1.31136549e+00
-1.11474566e-01 -3.10767680e-01 -9.03760120e-02 2.47145325e-01
7.85894617e-02 -3.58193755e-01 1.43361613e-01 -2.29225591e-01
-9.05084014e-02 1.20961297e+00 1.34987295e+00 -5.39064175e-03
-4.61603612e-01 -1.85325116e-01 -1.80730045e-01 7.69990504e-01
1.17490208e+00 -3.25475991e-01 -9.27825212e-01 2.54733086e-01
2.90143132e-01 5.98823309e-01 -1.76484391e-01 -9.82327104e-01
2.00938761e-01 -5.22441745e-01 -2.17548497e-02 -3.62728804e-01
7.64949679e-01 -8.48680258e-01 3.89502913e-01 2.87528753e-01
-8.62807691e-01 -2.50083557e-03 8.45088586e-02 2.24767759e-01
-5.80886006e-01 -2.70477355e-01 8.74655485e-01 1.56480402e-01
-2.42470264e-01 2.00219359e-02 -8.83534431e-01 -1.29721984e-01
1.17610037e+00 -6.69203401e-02 7.77629763e-02 -8.50720704e-01
-9.05649602e-01 -1.93700716e-01 1.50831372e-01 6.52872682e-01
6.46758616e-01 -1.69038010e+00 -4.76095051e-01 4.80472505e-01
-4.66043130e-02 -6.95564985e-01 1.96013138e-01 2.65678406e-01
-8.93006623e-02 4.24073696e-01 -4.51361418e-01 -2.56555289e-01
-1.20074880e+00 8.73426676e-01 2.18966201e-01 -7.47719109e-02
-2.66824305e-01 4.41229254e-01 7.83581957e-02 -8.55908692e-01
2.58376062e-01 -3.19850177e-01 -3.63892794e-01 2.63683856e-01
1.30304456e-01 1.41924024e-01 -4.53482121e-01 -7.59406984e-01
-5.91194443e-02 6.55534744e-01 6.14798367e-01 -9.08869386e-01
1.04660869e+00 -1.36168808e-01 -1.83852673e-01 1.01357603e+00
1.43901527e+00 3.98264587e-01 -9.93612826e-01 2.01764762e-01
-4.02435303e-01 -1.96795776e-01 -1.36821613e-01 -5.25335252e-01
-6.47852957e-01 1.03620577e+00 6.20676398e-01 3.19339663e-01
1.11696112e+00 -3.01717997e-01 5.28761923e-01 2.72842348e-01
2.60424167e-01 -1.62070692e+00 4.89252567e-01 5.02186120e-01
1.22777736e+00 -7.95816243e-01 -5.77781558e-01 -3.79028141e-01
-1.02033830e+00 1.26556730e+00 5.51721334e-01 1.13757074e-01
8.79210413e-01 8.20082486e-01 3.03195089e-01 2.42380217e-01
-1.17150617e+00 1.79010302e-01 -5.38744852e-02 4.68590379e-01
5.67712545e-01 3.14706951e-01 -6.69180155e-01 7.11532772e-01
-5.32215774e-01 -6.81369528e-02 5.26498199e-01 4.59029317e-01
-5.24150550e-01 -1.21029353e+00 -3.52652848e-01 -1.88513771e-01
-2.18883172e-01 9.87320840e-02 -2.74089545e-01 4.55447823e-01
3.13764542e-01 1.38951623e+00 -1.00895569e-01 -6.01866007e-01
2.97766000e-01 4.54265535e-01 1.57432303e-01 -4.79416251e-01
-3.86731893e-01 1.95543170e-01 1.59131914e-01 -1.92017883e-01
-3.76581252e-01 -3.71155292e-01 -1.60814083e+00 1.85054958e-01
-3.02597791e-01 6.99482203e-01 8.25850725e-01 5.51329851e-01
2.19922498e-01 6.60894752e-01 1.27111745e+00 -5.68643332e-01
-7.10010767e-01 -1.12293768e+00 -6.37005091e-01 4.27651733e-01
1.02455780e-01 -5.80826104e-01 -8.23765039e-01 1.70691505e-01]
|
[14.666726112365723, 6.424015998840332]
|
77e1deec-7a84-493f-8920-ea3441a94b92
|
fine-grained-human-feedback-gives-better
|
2306.01693
| null |
https://arxiv.org/abs/2306.01693v1
|
https://arxiv.org/pdf/2306.01693v1.pdf
|
Fine-Grained Human Feedback Gives Better Rewards for Language Model Training
|
Language models (LMs) often exhibit undesirable text generation behaviors, including generating false, toxic, or irrelevant outputs. Reinforcement learning from human feedback (RLHF) - where human preference judgments on LM outputs are transformed into a learning signal - has recently shown promise in addressing these issues. However, such holistic feedback conveys limited information on long text outputs; it does not indicate which aspects of the outputs influenced user preference; e.g., which parts contain what type(s) of errors. In this paper, we use fine-grained human feedback (e.g., which sentence is false, which sub-sentence is irrelevant) as an explicit training signal. We introduce Fine-Grained RLHF, a framework that enables training and learning from reward functions that are fine-grained in two respects: (1) density, providing a reward after every segment (e.g., a sentence) is generated; and (2) incorporating multiple reward models associated with different feedback types (e.g., factual incorrectness, irrelevance, and information incompleteness). We conduct experiments on detoxification and long-form question answering to illustrate how learning with such reward functions leads to improved performance, supported by both automatic and human evaluation. Additionally, we show that LM behaviors can be customized using different combinations of fine-grained reward models. We release all data, collected human feedback, and codes at https://FineGrainedRLHF.github.io.
|
['Hannaneh Hajishirzi', 'Mari Ostendorf', 'Noah A. Smith', 'Prithviraj Ammanabrolu', 'Alane Suhr', 'Nouha Dziri', 'Weijia Shi', 'Yushi Hu', 'Zeqiu Wu']
|
2023-06-02
| null | null | null | null |
['long-form-question-answering']
|
['natural-language-processing']
|
[ 1.48975983e-01 1.23849250e-01 -2.57278562e-01 -4.85831261e-01
-9.49906528e-01 -6.84044838e-01 3.74232769e-01 4.18467611e-01
-4.77662295e-01 9.98093545e-01 4.70630080e-01 -6.84423983e-01
-3.67473103e-02 -7.22949088e-01 -8.17826211e-01 -2.35325798e-01
4.52641457e-01 4.60309654e-01 1.00530498e-01 -4.31679785e-01
3.78487200e-01 -5.72743011e-04 -1.26159072e+00 7.03392625e-01
1.13573921e+00 8.98342431e-01 2.97116399e-01 8.74909639e-01
-2.20418513e-01 1.13370860e+00 -8.62503946e-01 -4.20485973e-01
-1.41995281e-01 -5.06980717e-01 -9.15177822e-01 -2.65875667e-01
3.12349647e-01 -5.98751247e-01 4.22021225e-02 7.58100927e-01
4.80791360e-01 1.53424650e-01 8.01559746e-01 -1.14684558e+00
-1.07474518e+00 8.41227651e-01 -1.98851779e-01 1.26117975e-01
6.91395700e-01 5.46930015e-01 1.22301257e+00 -8.58770013e-01
3.65110487e-01 1.38321972e+00 3.78606617e-01 9.58295405e-01
-1.29408658e+00 -5.02594590e-01 2.93413490e-01 3.08786742e-02
-8.30690205e-01 -2.95030266e-01 4.64916795e-01 -4.88738716e-01
8.94459486e-01 4.35543984e-01 1.47185773e-01 1.39439785e+00
3.19182873e-01 1.10296512e+00 1.21329761e+00 -4.31449592e-01
3.03712189e-01 4.74658698e-01 2.65568107e-01 3.34407687e-01
3.45427170e-02 3.03061664e-01 -6.23550236e-01 -2.56491035e-01
4.72026587e-01 -4.38945070e-02 -2.70543724e-01 1.04277499e-01
-8.45923305e-01 8.60743642e-01 3.74775290e-01 3.69095393e-02
-2.39210218e-01 1.54355913e-01 1.09383024e-01 6.33485198e-01
1.86715186e-01 7.35960782e-01 -6.38410628e-01 -3.98057073e-01
-6.84783936e-01 3.62605780e-01 8.46591175e-01 1.01409423e+00
8.28331292e-01 4.00755778e-02 -9.54679966e-01 1.03692579e+00
2.71421432e-01 7.52736568e-01 7.71420121e-01 -9.12107885e-01
6.33291483e-01 5.22649765e-01 5.71958303e-01 -6.22389436e-01
-5.01461804e-01 -2.34568864e-01 -5.00983953e-01 -2.02598963e-02
5.91161370e-01 -5.71313381e-01 -7.26839662e-01 2.05060983e+00
-2.40757778e-01 -4.86526340e-01 -1.77643716e-01 9.39656258e-01
7.44753122e-01 5.07029653e-01 2.23824829e-01 -4.48305383e-02
1.03461862e+00 -7.71905422e-01 -6.67137802e-01 -4.38283294e-01
8.59354496e-01 -5.98627090e-01 1.82331812e+00 3.99211138e-01
-1.14507282e+00 -4.94432449e-01 -6.96552873e-01 6.79123327e-02
-2.58651853e-01 -3.58266160e-02 1.73330098e-01 6.69437885e-01
-1.03320134e+00 5.74656427e-01 -2.65040368e-01 -3.02609615e-02
8.00902098e-02 2.65994936e-01 2.25866154e-01 8.15730989e-02
-1.60418177e+00 9.23050404e-01 7.82931596e-02 -2.18493149e-01
-6.08688474e-01 -4.55122381e-01 -7.21981406e-01 2.26747423e-01
3.62868667e-01 -7.06951141e-01 1.87361073e+00 -1.11166584e+00
-1.40820622e+00 3.12919438e-01 -2.07897201e-01 -3.94830763e-01
6.03354037e-01 -3.90226394e-01 -4.15457964e-01 -1.43208936e-01
1.35199860e-01 6.59529150e-01 8.57105255e-01 -1.16115141e+00
-4.57769185e-01 -2.48835430e-01 2.18096852e-01 3.09689641e-01
-5.02355456e-01 -1.11297049e-01 1.43420994e-01 -6.89851701e-01
-4.64986682e-01 -6.54028416e-01 -2.09072027e-02 -3.57743591e-01
-4.68026668e-01 -4.81802493e-01 3.79421622e-01 -4.95897919e-01
1.80442905e+00 -1.97040498e+00 -2.74418443e-01 2.26551786e-01
-2.73005087e-02 1.85654566e-01 -4.14273322e-01 6.47567213e-01
2.52382070e-01 6.41488612e-01 -1.03756428e-01 -7.07546994e-02
3.40576261e-01 4.82064113e-02 -3.09012234e-01 -2.72313565e-01
3.85518819e-01 1.10849571e+00 -1.19003522e+00 -3.41505647e-01
-1.13674112e-01 -3.87865007e-02 -8.11970890e-01 3.52760643e-01
-5.45862257e-01 2.33583689e-01 -4.84796494e-01 4.10763204e-01
2.75579244e-01 -4.40675706e-01 -8.75182897e-02 1.82468370e-01
1.21988282e-01 5.39496422e-01 -1.00551617e+00 1.11061895e+00
-5.36012769e-01 3.22290272e-01 -2.59293318e-01 -2.80267537e-01
8.64557445e-01 2.91465461e-01 -3.29440013e-02 -1.01721752e+00
-2.86995787e-02 2.81584442e-01 1.29558027e-01 -5.60652792e-01
8.49858344e-01 -7.09522218e-02 -1.78562090e-01 7.98692703e-01
-1.02634160e-02 -2.19452053e-01 4.57024872e-01 3.67694020e-01
1.16130948e+00 -3.08276084e-03 1.66974545e-01 -5.30597419e-02
3.12968194e-01 -2.52254397e-01 3.30021739e-01 1.31552827e+00
-2.63290972e-01 5.60649991e-01 5.52937746e-01 1.27706647e-01
-8.42560351e-01 -1.18957663e+00 5.83579838e-02 1.56618106e+00
-5.41056655e-02 -4.43598062e-01 -7.02182710e-01 -7.53542304e-01
1.79212332e-01 1.31036460e+00 -5.03493071e-01 -5.21977484e-01
-3.13631386e-01 -3.20706159e-01 5.51422417e-01 6.49561703e-01
3.13170999e-02 -1.67138302e+00 -5.57208419e-01 2.49844328e-01
-5.09938598e-01 -5.81110060e-01 -1.02432871e+00 3.59756798e-01
-7.56480038e-01 -8.26808989e-01 -5.54046631e-01 -4.37775970e-01
6.01302087e-01 7.97164962e-02 1.47857571e+00 2.29644313e-01
2.80761778e-01 4.86152351e-01 -5.85983276e-01 -3.41809571e-01
-5.44421136e-01 -1.59799427e-01 9.80104972e-03 -1.58894062e-01
3.31451923e-01 -6.28670529e-02 -6.76318288e-01 3.81714731e-01
-1.07002664e+00 -1.76576599e-01 7.39253044e-01 1.19578803e+00
2.18027651e-01 -5.01821935e-01 1.01516891e+00 -1.34581089e+00
1.53651357e+00 -6.31705165e-01 -9.51787829e-02 5.08255303e-01
-8.60119939e-01 3.52089345e-01 7.77929068e-01 -4.48091179e-01
-1.15782988e+00 -3.45645487e-01 -3.51101547e-01 -7.95708075e-02
-2.72144049e-01 5.88108718e-01 1.67242840e-01 5.17229915e-01
1.14007092e+00 2.46489525e-01 -2.93058395e-01 -1.92593411e-01
4.92811590e-01 9.74895537e-01 1.94707826e-01 -9.97121215e-01
3.78714651e-01 -3.67305756e-01 -6.65481865e-01 -4.24422979e-01
-8.20113242e-01 -3.94031614e-01 -1.07684359e-01 -1.88282549e-01
5.31272709e-01 -5.24455667e-01 -8.28563094e-01 1.33196682e-01
-9.80652988e-01 -8.41838658e-01 -3.79652739e-01 1.92303896e-01
-7.40710735e-01 1.74353123e-01 -8.17757010e-01 -1.18255925e+00
-3.70343745e-01 -1.00900304e+00 8.44118416e-01 3.16034287e-01
-8.58961284e-01 -8.02584589e-01 -1.84367374e-01 2.38078594e-01
5.12103021e-01 -3.93353879e-01 1.09891510e+00 -6.99139118e-01
-2.11988315e-01 -3.81042548e-02 -9.54364426e-03 4.43236083e-01
1.70584861e-02 -3.82760987e-02 -8.31967413e-01 -9.47653502e-02
-1.90575406e-01 -9.33348775e-01 7.37394035e-01 3.16194892e-01
1.29634070e+00 -7.66328335e-01 1.84936494e-01 1.77085167e-03
9.31199133e-01 1.15955628e-01 4.51336175e-01 1.54983893e-01
2.30670556e-01 6.09750330e-01 7.69959509e-01 6.89423084e-01
4.33836162e-01 4.89971340e-01 2.57043630e-01 1.90228134e-01
4.47581522e-02 -6.98397279e-01 6.69057131e-01 3.82022053e-01
3.06925684e-01 -5.33063531e-01 -6.03489816e-01 2.90318400e-01
-1.93008018e+00 -1.08796287e+00 -4.32043523e-02 2.33180618e+00
1.23474157e+00 4.21595842e-01 1.51424482e-01 -4.79927398e-02
4.09059644e-01 -1.12925716e-01 -9.06926036e-01 -7.45555520e-01
-1.94438294e-01 7.37347975e-02 1.62827894e-01 8.00518215e-01
-5.58215916e-01 9.41342592e-01 5.79707003e+00 6.97732925e-01
-8.87158811e-01 -4.64366004e-02 8.66754770e-01 -3.21465917e-02
-1.06210291e+00 -2.77775228e-01 -7.25225687e-01 6.13081753e-01
9.23470020e-01 -1.86582848e-01 5.36114216e-01 6.42850637e-01
4.70683515e-01 -1.43636376e-01 -1.17089331e+00 6.83305442e-01
-1.78023681e-01 -9.45859373e-01 2.52619207e-01 -3.11484516e-01
6.92953467e-01 -1.60530671e-01 2.06829458e-01 7.25069463e-01
7.84817219e-01 -1.03971159e+00 1.10920429e+00 5.47873676e-01
7.83851266e-01 -5.72496295e-01 6.60459220e-01 8.20223987e-01
-5.52060604e-01 -3.46503615e-01 -1.69074327e-01 -2.26097211e-01
-3.69877517e-02 4.44084913e-01 -8.85502279e-01 2.25552037e-01
5.30551910e-01 3.12001348e-01 -6.57804668e-01 6.88405514e-01
-5.57096720e-01 8.10831249e-01 3.30680166e-03 -5.70728660e-01
3.58358994e-02 9.45414677e-02 2.49723315e-01 1.27377415e+00
3.04067343e-01 -2.69261096e-03 8.77465028e-03 1.20620668e+00
-1.85447663e-01 8.22946131e-02 -3.98439318e-01 -5.95794134e-02
7.82230735e-01 9.87079978e-01 7.31971581e-03 -3.09015989e-01
-3.32743466e-01 8.74557793e-01 5.69764316e-01 7.63460636e-01
-4.70868349e-01 -2.55289406e-01 3.95160764e-01 1.98019847e-01
-4.36091237e-02 2.69126683e-01 -5.92835665e-01 -1.05104101e+00
-5.53777330e-02 -1.06155026e+00 4.66492712e-01 -9.69714522e-01
-1.56332278e+00 4.80748057e-01 -3.52701604e-01 -1.04799163e+00
-7.87910819e-01 -5.75910628e-01 -5.45860827e-01 1.09328854e+00
-1.24815786e+00 -5.34305871e-01 -1.06071055e-01 4.82283950e-01
6.26054704e-01 2.20706761e-01 8.65017951e-01 -7.57885724e-02
-2.68320441e-01 1.02782154e+00 -8.17670226e-02 7.08366260e-02
9.44055021e-01 -1.50441241e+00 6.05268218e-02 3.79507899e-01
-8.63841847e-02 7.86053061e-01 7.27018535e-01 -6.20031178e-01
-1.07785344e+00 -1.01299620e+00 1.15977120e+00 -7.38985896e-01
4.97478575e-01 -2.09569126e-01 -7.86744952e-01 5.82667649e-01
8.33209604e-02 -4.38561708e-01 7.09466517e-01 3.14544708e-01
-2.79480487e-01 1.03431351e-01 -1.16246879e+00 8.50913763e-01
7.40138888e-01 -5.97945988e-01 -5.05718350e-01 8.88553858e-02
8.21387529e-01 -3.29469949e-01 -5.40880620e-01 2.09996805e-01
4.82702911e-01 -1.07023549e+00 6.17867887e-01 -7.96313703e-01
7.46549368e-01 -1.06854670e-01 -1.62694976e-01 -1.72027469e+00
-5.51543713e-01 -4.08704251e-01 -5.03735900e-01 9.58576620e-01
9.58607078e-01 -5.04741132e-01 4.86017197e-01 8.92234743e-01
-3.10803711e-01 -9.79621351e-01 -3.67054909e-01 -6.76156580e-01
3.64089400e-01 -4.49553519e-01 7.05904901e-01 6.05866849e-01
3.98373187e-01 5.76863647e-01 -4.43965286e-01 -3.19813043e-01
3.26277278e-02 -4.11988497e-02 5.68088055e-01 -9.18587089e-01
-4.76082802e-01 -7.63869047e-01 4.64905232e-01 -1.43623579e+00
-1.05116829e-01 -8.41953695e-01 3.04778188e-01 -1.50919187e+00
9.12816226e-02 -3.32494229e-01 -4.52470392e-01 5.16268194e-01
-6.87894762e-01 -1.99767545e-01 3.73216450e-01 1.28762141e-01
-6.81605339e-01 4.76847857e-01 1.37057114e+00 -1.29569113e-01
-2.85040796e-01 2.59165525e-01 -1.05014050e+00 4.60045159e-01
1.08610559e+00 -2.47330084e-01 -4.46274012e-01 -4.60520804e-01
5.50434470e-01 4.07716244e-01 2.13071480e-01 -6.53091013e-01
1.69728965e-01 -3.47063065e-01 4.81817186e-01 -3.85788530e-01
1.38164401e-01 -4.93533701e-01 -4.73140150e-01 4.27212656e-01
-1.08875012e+00 2.39767447e-01 9.06680748e-02 5.79684675e-01
-9.58827510e-02 -3.91042531e-01 4.06637788e-01 -2.67788023e-01
-4.98279184e-01 1.29335359e-01 -4.45408940e-01 4.94612038e-01
3.78059596e-01 -2.24785954e-02 -5.19819498e-01 -8.13219488e-01
-7.12949038e-01 5.31280816e-01 2.33691543e-01 6.73291326e-01
7.05819130e-01 -1.24194109e+00 -7.25772560e-01 1.16563886e-01
4.06186074e-01 -3.37655425e-01 1.12172663e-01 6.30377591e-01
1.24851227e-01 5.70318997e-01 8.17554519e-02 -3.08559507e-01
-8.78964365e-01 2.62236297e-01 2.37269565e-01 -4.74566728e-01
4.14243750e-02 7.75369406e-01 1.69900730e-01 -7.24113941e-01
3.43874156e-01 -4.73796874e-01 -2.22015277e-01 -7.80248418e-02
4.63344663e-01 3.18102837e-01 6.79956377e-02 -1.58004969e-01
-4.18670811e-02 -4.27036285e-02 -1.60666481e-01 -2.96782881e-01
6.97359622e-01 -2.29297519e-01 2.92756945e-01 7.34894931e-01
7.43715405e-01 4.55237404e-02 -1.33518934e+00 -2.97081232e-01
1.70219034e-01 -3.18266898e-01 -3.93594414e-01 -1.48644066e+00
-3.65252048e-01 9.47705507e-01 2.62492985e-01 3.64420593e-01
9.38567698e-01 -2.42367201e-02 6.65811062e-01 4.37394917e-01
3.65272015e-01 -1.48184943e+00 4.57558930e-01 7.69682109e-01
1.05799997e+00 -1.34748816e+00 -6.38899565e-01 1.22755595e-01
-9.14617598e-01 1.01714659e+00 1.04394829e+00 2.89703071e-01
3.38134319e-01 -2.98739243e-02 2.92104304e-01 1.55774578e-01
-1.15144098e+00 -1.13262884e-01 2.89090127e-01 4.37535405e-01
9.72283363e-01 5.27732372e-01 -4.79413062e-01 9.82065797e-01
-4.79111224e-01 1.03215791e-01 5.35369575e-01 8.22998345e-01
-5.79976261e-01 -1.10558784e+00 -2.19265312e-01 1.07859826e+00
-1.98681056e-01 -4.39219177e-01 -6.26838684e-01 2.72347271e-01
-1.69558227e-01 1.41290653e+00 -2.18590766e-01 -5.74485123e-01
5.16123176e-01 2.00569436e-01 2.98283368e-01 -8.72882843e-01
-8.62389147e-01 -2.42685005e-01 1.27224192e-01 -4.65770602e-01
3.31594467e-01 -5.49365282e-01 -1.45005202e+00 -2.97417641e-01
-2.83810645e-01 2.78143764e-01 1.77727535e-01 8.26274514e-01
2.98614860e-01 4.89078134e-01 6.30320847e-01 -3.22036862e-01
-1.39656603e+00 -1.31069040e+00 -5.23782372e-01 8.16273868e-01
4.93041337e-01 -2.81185180e-01 -4.29384738e-01 -1.62300542e-01]
|
[11.648268699645996, 8.622105598449707]
|
3e7ec6e9-9ae8-4a20-aff6-3517a0ae2e38
|
evidence-aggregation-for-answer-re-ranking-in
|
1711.05116
| null |
http://arxiv.org/abs/1711.05116v2
|
http://arxiv.org/pdf/1711.05116v2.pdf
|
Evidence Aggregation for Answer Re-Ranking in Open-Domain Question Answering
|
A popular recent approach to answering open-domain questions is to first
search for question-related passages and then apply reading comprehension
models to extract answers. Existing methods usually extract answers from single
passages independently. But some questions require a combination of evidence
from across different sources to answer correctly. In this paper, we propose
two models which make use of multiple passages to generate their answers. Both
use an answer-reranking approach which reorders the answer candidates generated
by an existing state-of-the-art QA model. We propose two methods, namely,
strength-based re-ranking and coverage-based re-ranking, to make use of the
aggregated evidence from different passages to better determine the answer. Our
models have achieved state-of-the-art results on three public open-domain QA
datasets: Quasar-T, SearchQA and the open-domain version of TriviaQA, with
about 8 percentage points of improvement over the former two datasets.
|
['Wei zhang', 'Shiyu Chang', 'Jing Jiang', 'Xiaoxiao Guo', 'Tim Klinger', 'Shuohang Wang', 'Murray Campbell', 'Mo Yu', 'Gerald Tesauro', 'Zhiguo Wang']
|
2017-11-14
|
evidence-aggregation-for-answer-re-ranking-in-1
|
https://openreview.net/forum?id=rJl3yM-Ab
|
https://openreview.net/pdf?id=rJl3yM-Ab
|
iclr-2018-1
|
['triviaqa']
|
['miscellaneous']
|
[-2.35124547e-02 2.40920499e-01 1.36425063e-01 -4.24877614e-01
-1.90199900e+00 -1.05885148e+00 5.39613485e-01 5.24516284e-01
-4.78203356e-01 1.14087868e+00 6.40059829e-01 -5.33999562e-01
-4.11574602e-01 -9.88526702e-01 -6.56509221e-01 7.22540170e-02
6.39579296e-01 1.16867232e+00 1.11490107e+00 -8.38905275e-01
7.44568825e-01 -2.52746522e-01 -1.67560482e+00 9.73244071e-01
1.75660062e+00 6.84716642e-01 1.25561059e-01 9.82370257e-01
-9.79117215e-01 1.24716699e+00 -8.38286400e-01 -7.80952334e-01
-4.58023660e-02 -7.93019354e-01 -1.69760466e+00 -5.35420656e-01
6.57274008e-01 -1.20730676e-01 -1.89694092e-02 8.16108286e-01
4.93602157e-01 1.97597206e-01 6.37632191e-01 -6.40430808e-01
-9.18494105e-01 3.62894118e-01 5.26112206e-02 5.73256910e-01
1.31491041e+00 -9.30778310e-02 1.37306094e+00 -6.64645612e-01
7.53299475e-01 1.14817917e+00 3.55313897e-01 4.77427840e-01
-9.43347633e-01 -9.16524380e-02 -3.03821981e-01 7.85158932e-01
-7.61451900e-01 -1.89279690e-01 4.73351359e-01 -2.13024020e-01
1.06220365e+00 5.23528755e-01 1.65093109e-01 5.42620063e-01
4.09620814e-02 7.49031425e-01 1.53508258e+00 -7.69328356e-01
2.18547612e-01 -1.12365529e-01 6.19394958e-01 5.30939341e-01
-1.08176440e-01 -4.82193232e-01 -5.73691428e-01 -6.00596130e-01
-1.05510149e-02 -6.08297586e-01 -4.54635531e-01 1.14191644e-01
-1.06696081e+00 1.06240535e+00 3.03383201e-01 1.29728630e-01
-5.02603769e-01 -5.64912915e-01 9.76617783e-02 8.54809880e-01
3.95127743e-01 1.10052311e+00 -7.90092587e-01 -2.07816824e-01
-9.02569115e-01 8.31651449e-01 1.24320900e+00 5.19541025e-01
7.99262106e-01 -1.03737569e+00 -8.46535265e-01 1.05393350e+00
3.23535413e-01 4.29889143e-01 6.06291294e-01 -1.32440829e+00
9.54573393e-01 9.90145504e-01 4.97062892e-01 -7.63159156e-01
-7.90472403e-02 -2.14329690e-01 8.78104791e-02 -3.10346872e-01
6.35961592e-01 -1.46621913e-01 -9.35630262e-01 1.25819135e+00
5.62539816e-01 -3.31090331e-01 3.39972973e-01 8.65000248e-01
1.43392062e+00 8.65463197e-01 -1.83536902e-01 9.79216471e-02
1.57056665e+00 -1.29050839e+00 -7.40723670e-01 -7.00998083e-02
4.38522607e-01 -9.85243678e-01 1.33819580e+00 3.65595877e-01
-1.28054368e+00 -5.59624672e-01 -7.91305542e-01 -5.96306264e-01
-1.87108904e-01 -1.81835502e-01 -8.22208077e-02 3.81618649e-01
-9.14153099e-01 2.24691093e-01 1.36944607e-01 -1.06011003e-01
-1.52810320e-01 -6.15370758e-02 -1.63078997e-02 -4.25050467e-01
-1.65911925e+00 1.17145061e+00 9.07213762e-02 -7.29448378e-01
-5.82538903e-01 -7.82951713e-01 -5.96812189e-01 1.70779377e-01
4.61085230e-01 -1.12995517e+00 1.85780466e+00 -4.18074220e-01
-1.58166254e+00 8.42423975e-01 -6.64905965e-01 -3.84507000e-01
2.30093390e-01 -4.80546713e-01 -4.54578549e-01 6.21029139e-01
5.75879395e-01 5.50168991e-01 5.37967563e-01 -1.03388011e+00
-8.45488846e-01 -1.84795275e-01 7.47491777e-01 3.29266429e-01
1.12031870e-01 1.91055223e-01 -4.12124962e-01 -9.47853997e-02
1.30826369e-01 -5.37134945e-01 -4.85154092e-02 -5.28023720e-01
-1.59804374e-01 -9.42549169e-01 2.04264268e-01 -1.03743064e+00
1.53508735e+00 -1.28056002e+00 2.34963775e-01 -1.47255376e-01
2.66819030e-01 2.08703890e-01 -5.90332150e-01 8.08850229e-01
4.09829468e-01 -7.06439316e-02 -2.09121212e-01 2.11338490e-01
4.27957959e-02 1.12634137e-01 -5.99749744e-01 -4.88086104e-01
2.71701843e-01 1.01937973e+00 -1.25781059e+00 -7.13139951e-01
-3.98381829e-01 -2.31001288e-01 -6.13318741e-01 5.15805423e-01
-9.67800021e-01 4.22329724e-01 -6.52942240e-01 3.92146349e-01
3.25093091e-01 -4.08600062e-01 -2.54143119e-01 1.55017108e-01
9.14881378e-02 1.31115568e+00 -7.79690802e-01 1.67591584e+00
-4.85326052e-01 3.17853987e-01 -3.64663780e-01 -3.86846453e-01
8.63648415e-01 2.91657716e-01 -2.31440179e-02 -1.18439317e+00
-1.50215164e-01 6.32646084e-01 1.99046340e-02 -1.02085543e+00
6.92055106e-01 8.15723166e-02 -6.62567988e-02 5.94315767e-01
2.37908661e-01 -6.17764831e-01 8.23764503e-01 5.91906428e-01
1.38315225e+00 7.75664523e-02 1.77143455e-01 -1.14014305e-01
1.01144505e+00 6.26140058e-01 5.60820401e-02 9.97134328e-01
8.02016631e-02 6.82869017e-01 4.46642786e-01 -2.19563335e-01
-6.62679136e-01 -1.05159473e+00 1.37494177e-01 1.20524096e+00
-3.93935945e-03 -5.41586936e-01 -8.30673397e-01 -1.14777064e+00
-4.39681113e-02 1.10653496e+00 -4.27761286e-01 7.34353885e-02
-6.15294814e-01 -2.10699886e-01 4.18609470e-01 1.20422736e-01
5.58158636e-01 -1.05236459e+00 -4.55307007e-01 2.72031128e-01
-1.22067010e+00 -7.99506068e-01 -1.45877942e-01 -2.04121694e-01
-7.30582058e-01 -1.16990745e+00 -9.59272981e-01 -6.05788171e-01
2.18631953e-01 2.18050525e-01 1.97609639e+00 2.84352154e-01
4.28585619e-01 5.03860414e-01 -9.61328268e-01 -4.56673175e-01
-5.65620601e-01 4.16431218e-01 -5.59270024e-01 -3.67156208e-01
6.67374969e-01 -9.03214291e-02 -6.42661750e-01 2.77854413e-01
-7.98630297e-01 -3.71846974e-01 4.55294490e-01 7.89881766e-01
5.56817055e-01 -4.97783363e-01 1.03092098e+00 -8.29438508e-01
1.38945150e+00 -8.04232538e-01 -3.21362972e-01 8.84469211e-01
-4.62406874e-01 4.43039477e-01 4.90658164e-01 -1.23118296e-01
-1.23265886e+00 -5.88632584e-01 -5.57854235e-01 3.24256569e-01
-6.53613880e-02 8.91972601e-01 2.07043961e-01 -1.26912696e-02
9.82423842e-01 1.15400195e-01 -3.72118086e-01 -5.61569273e-01
6.73724055e-01 8.44266057e-01 4.21238422e-01 -7.02215195e-01
6.41211152e-01 -8.48375782e-02 -5.29264867e-01 -2.60710657e-01
-1.61635959e+00 -8.66784692e-01 -3.45595062e-01 -1.96598426e-01
8.19006681e-01 -5.62274218e-01 -2.92304456e-01 2.10164580e-02
-1.45870769e+00 -1.40542323e-02 -3.56290996e-01 1.71743885e-01
-3.86873811e-01 5.74734747e-01 -5.51321149e-01 -4.77151066e-01
-5.13803542e-01 -7.91416824e-01 9.97440279e-01 6.37031972e-01
-4.84174758e-01 -7.57105350e-01 8.07491302e-01 1.33053493e+00
4.82263803e-01 -3.52891415e-01 1.22455418e+00 -1.02076626e+00
-5.77317119e-01 -1.03287898e-01 -6.97338730e-02 1.28371239e-01
-1.27868280e-01 -4.28676516e-01 -7.40353286e-01 2.42068723e-01
2.38762110e-01 -8.53502870e-01 9.98019576e-01 -1.06578143e-02
8.26733649e-01 -3.44438523e-01 1.15810931e-01 -3.34459901e-01
1.13444233e+00 -1.30306408e-01 9.19130445e-01 6.14289403e-01
2.71974709e-02 9.19472396e-01 7.91523755e-01 -1.06307261e-01
9.17568982e-01 3.99088413e-01 2.26515144e-01 4.39497113e-01
-3.40694249e-01 -2.79215783e-01 1.37273312e-01 1.22688186e+00
1.08834781e-01 -3.90150577e-01 -8.43509197e-01 1.00935662e+00
-1.62853920e+00 -9.86767352e-01 -6.52781844e-01 2.01291680e+00
1.17727911e+00 5.96570522e-02 1.16916165e-01 3.50575037e-02
2.06700072e-01 -4.84022498e-02 -2.52131253e-01 -5.49413383e-01
-2.00944304e-01 8.19495559e-01 -2.51027852e-01 6.86625004e-01
-4.92918283e-01 7.01376200e-01 6.71079731e+00 8.92197669e-01
-2.62595147e-01 2.58702397e-01 3.56075883e-01 3.11488330e-01
-9.21633184e-01 3.85146052e-01 -7.93017447e-01 3.04822594e-01
1.15933776e+00 -2.27160707e-01 1.68234631e-01 4.27734464e-01
-3.82302314e-01 -6.35100543e-01 -7.70703018e-01 3.05396259e-01
3.75199735e-01 -1.48696291e+00 3.57821137e-01 -4.56625521e-01
9.72663224e-01 8.63507167e-02 -2.90438592e-01 7.20102191e-01
6.43486202e-01 -8.23752224e-01 3.61660779e-01 8.13554168e-01
3.09870958e-01 -5.57506800e-01 1.02063119e+00 6.42077625e-01
-7.55278528e-01 -1.49136916e-01 -3.69082600e-01 -2.90790141e-01
4.31110531e-01 4.52040851e-01 -6.37363791e-01 9.76227105e-01
8.47504318e-01 -1.67286783e-01 -8.71973455e-01 1.32520676e+00
-7.48067975e-01 1.02202654e+00 -8.68026465e-02 -5.51648557e-01
2.66044289e-01 2.45818105e-02 4.62661982e-01 6.87765121e-01
3.36756378e-01 4.64043915e-01 -2.72394083e-02 7.29979753e-01
-1.77652061e-01 2.74215668e-01 1.94089836e-03 2.49228805e-01
5.93574941e-01 1.04046392e+00 -7.71076679e-02 -5.17202616e-01
-6.54745042e-01 8.40759099e-01 5.48202038e-01 1.12024903e-01
-5.40478051e-01 -7.71204054e-01 -1.29262097e-02 4.28659208e-02
2.34843448e-01 -4.11529131e-02 -8.37764665e-02 -1.15163088e+00
2.82055527e-01 -1.47292745e+00 9.46764767e-01 -1.15514672e+00
-1.71420372e+00 6.93110406e-01 -2.20609102e-02 -1.02829778e+00
-5.44019103e-01 -4.59047616e-01 -6.39145434e-01 1.08771646e+00
-1.92564130e+00 -5.30507803e-01 -2.02116475e-01 4.54020888e-01
8.41656506e-01 6.92227855e-02 8.80359292e-01 2.45877251e-01
2.49484196e-01 2.15516552e-01 -3.15808356e-02 5.53830080e-02
9.84241664e-01 -1.47759068e+00 3.00418317e-01 6.36430085e-01
4.21733797e-01 6.50191545e-01 7.04206884e-01 -7.28903413e-01
-1.03158927e+00 -4.82920885e-01 1.69884515e+00 -1.19748116e+00
6.49813175e-01 2.88339615e-01 -1.33850229e+00 1.46457076e-01
8.74200881e-01 -6.33917689e-01 8.76759827e-01 2.46869579e-01
-3.90644431e-01 6.57216311e-02 -1.15102458e+00 3.82192999e-01
5.57129681e-01 -6.12316489e-01 -1.81745791e+00 4.46562856e-01
1.05604768e+00 -5.18396735e-01 -7.28475690e-01 5.62867939e-01
2.30836004e-01 -1.03545630e+00 8.93755615e-01 -7.91404188e-01
7.14594603e-01 -5.25502980e-01 -1.77421406e-01 -1.43610322e+00
-2.34795883e-01 -4.42388207e-01 -3.04763973e-01 1.15152311e+00
9.34929848e-01 -6.10588133e-01 4.19521809e-01 3.48667324e-01
-3.21870744e-02 -7.47272253e-01 -1.05658829e+00 -5.17740190e-01
4.23217386e-01 -1.08385822e-02 7.93774188e-01 5.71211696e-01
7.85368010e-02 9.25866365e-01 2.87133604e-01 1.25014479e-03
2.10062757e-01 3.77938151e-01 7.48720169e-01 -1.34538579e+00
-2.85686821e-01 -2.09509358e-01 1.89545065e-01 -1.49402893e+00
-6.93873018e-02 -7.67986655e-01 9.47471261e-02 -2.29381108e+00
2.22187087e-01 -2.00616300e-01 -1.58596933e-01 -5.62715009e-02
-7.42567778e-01 2.89847162e-02 5.37789948e-02 2.22144753e-01
-1.17699361e+00 5.86859345e-01 1.43764305e+00 -1.49829611e-01
3.38516347e-02 7.31011331e-02 -9.22689497e-01 5.04794478e-01
6.94737196e-01 -5.75773835e-01 -5.27764320e-01 -8.76545370e-01
7.84811616e-01 2.66739011e-01 1.95834070e-01 -1.04162252e+00
4.66331273e-01 -1.67171165e-01 7.06181452e-02 -9.14317310e-01
1.40687168e-01 -1.42382562e-01 -5.62321484e-01 2.59394497e-01
-6.21833265e-01 5.04575729e-01 2.98580974e-02 4.69775468e-01
-6.30041242e-01 -8.34880829e-01 2.42758065e-01 -3.79486829e-01
-5.74063420e-01 -3.56578380e-01 -4.29570436e-01 8.75184834e-01
3.62672299e-01 1.67137846e-01 -9.42906618e-01 -7.52748609e-01
-1.99333981e-01 6.38252497e-01 -1.16385855e-02 5.90278506e-01
7.37866223e-01 -1.16775048e+00 -1.29859340e+00 -4.63524610e-01
3.67960304e-01 -1.47675246e-01 3.31469327e-01 6.50210023e-01
-5.22096395e-01 8.97812843e-01 6.58518746e-02 -3.50073099e-01
-1.11892211e+00 2.49546811e-01 1.90056175e-01 -8.67027462e-01
-1.26586426e-02 9.90417778e-01 -4.72229958e-01 -1.04529142e+00
-2.44947016e-01 -3.02231014e-01 -8.35639119e-01 1.04771197e-01
7.35331535e-01 4.73014593e-01 3.96499783e-01 -2.88430423e-01
-9.85336378e-02 6.57251120e-01 -1.66733600e-02 -5.47726035e-01
7.70787776e-01 -2.77213097e-01 -4.59871739e-01 2.45633721e-01
6.83422446e-01 4.56890374e-01 -4.89744753e-01 -5.57448685e-01
3.76687855e-01 -4.39489007e-01 -4.29380566e-01 -1.51841116e+00
-1.95436180e-01 7.80016124e-01 1.11793637e-01 5.04435003e-01
1.14279783e+00 4.03707147e-01 1.22952402e+00 7.17580557e-01
5.17672837e-01 -1.13050902e+00 4.90895867e-01 1.10812616e+00
1.12096560e+00 -1.17075121e+00 -3.63885880e-01 -3.54461998e-01
-4.74463463e-01 1.04562390e+00 7.95909643e-01 1.50015399e-01
2.46066097e-02 -5.21562755e-01 4.31776166e-01 -4.26673383e-01
-9.99215841e-01 -6.14000738e-01 9.06978011e-01 3.21142435e-01
5.12695611e-01 -2.39286065e-01 -8.12771440e-01 6.91012025e-01
-4.81421888e-01 -5.67069463e-02 4.14981514e-01 9.38851058e-01
-1.01821578e+00 -1.31102848e+00 -5.96697390e-01 6.19096518e-01
-5.59313118e-01 -2.77471155e-01 -8.70862901e-01 2.72610158e-01
-1.26508310e-01 1.83213413e+00 -2.83090115e-01 -1.68453604e-01
4.22195345e-01 4.23884064e-01 6.85906827e-01 -8.05534303e-01
-8.07307839e-01 -6.40152097e-01 6.18465543e-01 -2.34729081e-01
-5.13348460e-01 -4.45344239e-01 -1.09681761e+00 8.55590329e-02
-5.20910442e-01 7.67732441e-01 2.72749186e-01 1.24771452e+00
6.94826901e-01 4.34483677e-01 3.17976058e-01 5.85065894e-02
-9.79854226e-01 -1.14821780e+00 2.86088973e-01 4.61220145e-01
3.42584938e-01 -4.49325889e-01 -1.76404476e-01 -2.40749612e-01]
|
[11.330344200134277, 8.001605033874512]
|
1809136a-629e-42eb-b0df-17f9566d2a18
|
toward-imitating-visual-attention-of-experts
|
1903.0632
| null |
http://arxiv.org/abs/1903.06320v1
|
http://arxiv.org/pdf/1903.06320v1.pdf
|
Toward Imitating Visual Attention of Experts in Software Development Tasks
|
Expert programmers' eye-movements during source code reading are valuable
sources that are considered to be associated with their domain expertise. We
advocate a vision of new intelligent systems incorporating expertise of experts
for software development tasks, such as issue localization, comment generation,
and code generation. We present a conceptual framework of neural autonomous
agents based on imitation learning (IL), which enables agents to mimic the
visual attention of an expert via his/her eye movement. In this framework, an
autonomous agent is constructed as a context-based attention model that
consists of encoder/decoder network and trained with state-action sequences
generated by an experts' demonstration. Challenges to implement an IL-based
autonomous agent specialized for software development task are discussed in
this paper.
|
['Hideaki Hata', 'Nishanth Koganti', 'Yoshiharu Ikutani', 'Kenichi Matsumoto', 'Takatomi Kubo']
|
2019-03-15
| null | null | null | null |
['comment-generation']
|
['natural-language-processing']
|
[ 5.94443493e-02 7.25623071e-01 4.51480635e-02 -2.55227029e-01
-2.26385161e-01 -3.90321910e-01 4.71918523e-01 -1.17553256e-01
-1.23132899e-01 3.12971145e-01 -6.64890036e-02 -4.21955526e-01
3.23731601e-01 -2.11177379e-01 -9.09751832e-01 -1.37886703e-01
3.43044609e-01 -1.28715739e-01 1.86348811e-01 -1.78157732e-01
9.13024366e-01 2.30416860e-02 -1.51391184e+00 2.89770842e-01
1.33111000e+00 2.46668443e-01 8.58747065e-01 1.10689807e+00
-2.14738399e-01 1.67139041e+00 -7.62217402e-01 -2.97559530e-01
-6.30837306e-02 -6.61927938e-01 -8.91573727e-01 -9.26981196e-02
7.06621706e-02 -4.79018748e-01 -8.13401267e-02 1.38753247e+00
2.13653058e-01 -2.19179735e-01 7.22555280e-01 -1.54300928e+00
-1.57540846e+00 5.26611090e-01 -3.13219339e-01 3.45176786e-01
9.13977802e-01 6.31865203e-01 8.19060564e-01 -7.80010104e-01
8.11174273e-01 9.45397317e-01 3.22644502e-01 1.05882311e+00
-6.27466857e-01 -6.74713477e-02 3.17867458e-01 5.74489772e-01
-9.61772144e-01 -3.29028010e-01 7.89787948e-01 -1.08346975e+00
1.39037967e+00 -2.42284298e-01 6.49162352e-01 1.26386046e+00
6.41430616e-01 1.19269133e+00 5.80801129e-01 -7.94274926e-01
3.86756063e-01 5.96316695e-01 1.55140907e-01 8.96831155e-01
-8.84491056e-02 1.37665138e-01 -5.97770095e-01 -3.15863080e-02
1.00903332e+00 -1.27598837e-01 -3.37493271e-01 -5.16622007e-01
-1.11824679e+00 8.48606348e-01 4.00436014e-01 2.36354873e-01
-5.67088187e-01 3.91621381e-01 1.43058077e-01 7.34618247e-01
9.59010199e-02 7.65444994e-01 -2.45021746e-01 -2.78483540e-01
-3.01283240e-01 8.44415277e-02 1.00432491e+00 1.55591547e+00
5.11075079e-01 4.12344128e-01 -3.88566971e-01 4.53577876e-01
1.10879827e+00 5.76435447e-01 7.61967659e-01 -1.10348177e+00
2.38718539e-01 1.19829333e+00 3.55251282e-01 -4.60878342e-01
-2.11900491e-02 -7.71650299e-02 -5.40521964e-02 9.76820111e-01
2.06570071e-03 -3.85149032e-01 -4.08340961e-01 1.43252373e+00
9.30111036e-02 4.25358862e-02 6.17446184e-01 7.65515149e-01
8.66984665e-01 5.51357031e-01 -8.77949446e-02 3.59425731e-02
1.17812216e+00 -1.45995176e+00 -6.40841007e-01 -3.53801489e-01
4.04223531e-01 -3.07075292e-01 1.00905383e+00 2.51404911e-01
-1.21998274e+00 -6.13967836e-01 -9.66713011e-01 -4.36865464e-02
-2.51706183e-01 4.25322086e-01 1.45119473e-01 2.43344724e-01
-1.25342965e+00 3.26889813e-01 -5.88413060e-01 -4.48361933e-01
2.66067505e-01 -9.61755123e-03 5.59406877e-02 6.64211154e-01
-7.72832215e-01 1.10729444e+00 1.79796033e-02 3.69943641e-02
-1.44359314e+00 -3.35685849e-01 -9.97600436e-01 1.75050303e-01
7.68524483e-02 -8.26804042e-01 2.03219867e+00 -1.81759787e+00
-2.01309133e+00 9.06868577e-01 -5.64305522e-02 -4.23167884e-01
1.46263644e-01 -4.11300778e-01 -4.88335699e-01 1.30672738e-01
3.57265592e-01 4.22621399e-01 1.37936771e+00 -1.01029837e+00
-6.34615839e-01 -4.47158739e-02 3.15041602e-01 1.87749758e-01
4.29003350e-02 5.14241993e-01 -1.19630933e-01 -4.55681413e-01
-8.06463242e-01 -8.51651430e-01 -1.43314436e-01 2.64884382e-01
-1.10760488e-01 -6.34895504e-01 4.76062894e-01 -7.15755939e-01
1.28864050e+00 -2.10024095e+00 5.49370527e-01 -2.72742271e-01
3.83079678e-01 3.82964522e-01 -3.53747815e-01 6.14777148e-01
-9.95992124e-02 -3.66278499e-01 -6.64456487e-02 1.69217914e-01
1.87347621e-01 -3.18860739e-01 -2.43859380e-01 2.18648329e-01
4.10377949e-01 1.03225017e+00 -1.13759112e+00 -2.73261040e-01
-1.48146674e-01 6.70304596e-02 -5.89848399e-01 8.46712828e-01
-6.01378560e-01 3.92350495e-01 -7.01601923e-01 5.57062387e-01
9.81004909e-03 -6.43499553e-01 -1.29342839e-01 6.01639807e-01
-4.56514329e-01 1.09886028e-01 -6.12414479e-01 1.94516516e+00
-5.51664412e-01 1.10937035e+00 1.63779765e-01 -6.32122219e-01
8.75483274e-01 6.12404525e-01 -5.19389153e-01 -5.80375552e-01
2.16241763e-03 -3.31913717e-02 4.19188231e-01 -1.41383600e+00
2.07188919e-01 6.35281622e-01 1.48785532e-01 1.03754461e+00
4.44337763e-02 -4.94221486e-02 1.99949324e-01 3.47894728e-01
1.30151224e+00 5.92256188e-01 5.19231856e-01 -6.47199228e-02
7.06098616e-01 9.00719687e-02 1.48219302e-01 8.03349614e-01
-2.78682172e-01 1.20791458e-01 6.33912325e-01 -4.95551705e-01
-1.21712649e+00 -6.94216490e-01 3.35742086e-01 1.39049792e+00
7.10152984e-02 -1.73386991e-01 -1.21360600e+00 -7.49628782e-01
-3.55206102e-01 9.93158221e-01 -5.93398213e-01 -1.93521947e-01
-4.69135195e-01 2.02714592e-01 3.63095611e-01 7.47588336e-01
1.06357649e-01 -2.00613642e+00 -1.55581522e+00 3.01195532e-01
4.19453263e-01 -5.06926596e-01 -4.69482213e-01 -1.86189175e-01
-4.35956985e-01 -1.25215685e+00 -9.14570689e-01 -1.04176593e+00
1.21220279e+00 5.63927321e-03 1.10355902e+00 4.40711617e-01
-2.18450740e-01 7.93582439e-01 -4.98328537e-01 -7.54850805e-01
-1.15650845e+00 -1.71336398e-01 -3.41796339e-01 -2.51741648e-01
4.53955710e-01 -2.78220594e-01 -5.83804071e-01 7.13574737e-02
-5.77664554e-01 2.04889566e-01 8.75692964e-01 7.36458004e-01
-5.02402067e-01 -9.85212862e-01 6.01435900e-01 -8.07317257e-01
1.21193326e+00 -6.58049285e-01 -9.06501532e-01 7.66305089e-01
-5.16526103e-01 3.23270053e-01 6.60067379e-01 -4.70004082e-01
-1.52371991e+00 3.41384672e-02 2.13552505e-01 -2.04751149e-01
-4.36284184e-01 2.33034775e-01 3.87542218e-01 -1.66889220e-01
9.46471155e-01 4.27988738e-01 1.86530292e-01 -2.75550932e-01
3.11903656e-01 1.32580876e+00 4.32793945e-01 -3.75356674e-01
5.78883052e-01 -2.31930897e-01 -8.04238081e-01 -4.24448073e-01
-3.01764756e-01 -1.19903743e-01 -5.04794002e-01 -4.34794039e-01
8.66222620e-01 -8.93727839e-01 -1.12824595e+00 5.76991260e-01
-1.82209313e+00 -5.68802774e-01 -5.85604496e-02 3.05329889e-01
-7.60119975e-01 2.52709985e-01 -6.57871783e-01 -7.07700491e-01
-3.73793364e-01 -1.47303760e+00 8.00843120e-01 7.52273977e-01
-5.17045796e-01 -1.06446886e+00 5.11404753e-01 4.52677459e-02
6.70732975e-01 -3.43249559e-01 8.29526365e-01 -7.62642503e-01
-9.46391642e-01 1.47441134e-03 -2.28379756e-01 2.55628645e-01
-1.13919772e-01 3.21013063e-01 -8.42585325e-01 6.14550337e-03
9.32520553e-02 -1.65026963e-01 7.87371099e-02 7.24006966e-02
6.69215083e-01 -3.60839993e-01 -4.52553779e-01 2.36886054e-01
1.11649966e+00 7.38789797e-01 3.92286867e-01 2.65987992e-01
6.56680167e-01 6.07645452e-01 4.85577941e-01 4.78756696e-01
6.03850186e-01 2.84533232e-01 5.99788725e-01 2.71303982e-01
1.23719677e-01 -8.67254436e-02 7.89472461e-01 6.75248265e-01
-2.86409587e-01 -1.97036743e-01 -1.08744454e+00 6.86615169e-01
-2.10267687e+00 -7.99978256e-01 -1.77769735e-01 1.59628177e+00
7.56649435e-01 -1.16806850e-01 -1.05591357e-01 -6.36794150e-01
9.40007746e-01 -2.82578140e-01 -8.91722620e-01 -8.58554840e-01
6.28822982e-01 -1.51067048e-01 -1.52236924e-01 3.22510958e-01
-6.95300996e-01 9.15665507e-01 6.60340118e+00 6.14444725e-03
-8.89147937e-01 1.33952871e-01 -2.79220492e-01 4.14726555e-01
-1.74627960e-01 2.03741807e-03 -5.53459823e-01 6.88939929e-01
9.96458948e-01 -4.99651849e-01 5.23373961e-01 1.49705684e+00
6.57037571e-02 -1.17318757e-01 -1.62797320e+00 6.39831722e-01
4.60579067e-01 -1.26952350e+00 -2.98647135e-01 -3.69531125e-01
8.73009503e-01 -1.12476610e-01 1.15611315e-01 4.56472903e-01
8.36523950e-01 -8.28876555e-01 6.45106077e-01 7.40575135e-01
4.57032144e-01 -1.12318441e-01 3.10088009e-01 6.83438957e-01
-6.45399332e-01 -3.98825556e-01 -4.33329970e-01 -2.79608279e-01
-1.91077814e-02 -3.63372803e-01 -1.05215943e+00 -1.09055042e-01
5.98282874e-01 8.93777847e-01 -5.45272231e-01 1.08086729e+00
-9.61586654e-01 2.24717453e-01 4.83793497e-01 -7.38404274e-01
6.18665330e-02 4.73931944e-03 6.25259399e-01 9.72328782e-01
4.56680804e-01 -1.33989621e-02 -3.71746778e-01 1.61962688e+00
9.42073986e-02 -2.03031879e-02 -1.03245890e+00 -1.96536347e-01
3.14011127e-01 1.11246109e+00 -2.04206586e-01 -4.44733918e-01
-9.16203797e-01 1.27930999e+00 4.74679559e-01 4.96513158e-01
-8.43579948e-01 -8.29450607e-01 4.25151497e-01 -1.89792871e-01
4.51136798e-01 2.00613707e-01 1.22916214e-01 -1.05140996e+00
1.00032076e-01 -1.14624536e+00 -2.22797215e-01 -1.51648712e+00
-8.70573759e-01 8.86036992e-01 -3.55038404e-01 -1.54859829e+00
-6.61186874e-01 -9.20523584e-01 -1.23051059e+00 7.33970582e-01
-1.37689912e+00 -8.30366015e-01 -4.50028032e-01 4.03152376e-01
1.05613911e+00 -8.70833397e-01 7.67572522e-01 -2.77300090e-01
-6.33656085e-01 3.71510446e-01 -1.78343058e-01 2.46093020e-01
4.04596329e-01 -1.49874485e+00 7.59408474e-01 9.58431959e-01
3.12887924e-03 8.33769500e-01 5.97273707e-01 -6.69668674e-01
-1.36515081e+00 -6.74721599e-01 6.33862853e-01 -7.83836842e-01
7.89228022e-01 1.53440803e-01 -8.91480744e-01 1.04465425e+00
1.09598017e+00 -3.51444632e-01 4.99458402e-01 -5.76563835e-01
-1.87649280e-01 5.11081994e-01 -8.04011047e-01 8.15436363e-01
8.37189674e-01 -7.45778799e-01 -1.20600545e+00 3.34241062e-01
5.82139552e-01 -5.06331265e-01 -4.90346581e-01 -2.14263827e-01
2.67586380e-01 -9.05720830e-01 4.75646883e-01 -7.37133503e-01
9.08018768e-01 -5.52818298e-01 6.99044168e-01 -1.55907679e+00
-3.44340265e-01 -9.19243038e-01 -1.97060868e-01 8.56725156e-01
3.89007270e-01 -4.64171946e-01 1.61473826e-01 7.15970516e-01
-4.40366358e-01 -1.16099820e-01 -5.16828120e-01 -2.13811263e-01
-2.96378523e-01 2.31006481e-02 3.38195294e-01 7.55867124e-01
8.39278221e-01 5.98781168e-01 -1.88638985e-01 2.56044388e-01
1.17303520e-01 1.80890523e-02 8.22188079e-01 -1.14070785e+00
-5.15444160e-01 -4.95070279e-01 -1.15776189e-01 -1.28360832e+00
6.10965729e-01 -7.33919322e-01 4.28496480e-01 -1.36742461e+00
1.62281558e-01 3.04760873e-01 2.74541844e-02 3.84596735e-01
-3.06950837e-01 -4.86471415e-01 3.12143639e-02 2.38809913e-01
-9.91200745e-01 3.07733327e-01 1.15174913e+00 1.74931604e-02
-1.25052840e-01 2.89789677e-01 -7.03966081e-01 1.06550097e+00
4.76467460e-01 -5.17692149e-01 -3.04851860e-01 -7.37057447e-01
7.95679629e-01 5.28657556e-01 5.11423409e-01 -7.24656880e-01
8.04800510e-01 -1.47615314e-01 -2.38411818e-02 2.73452755e-02
-2.47198090e-01 -6.69344962e-01 -2.60935426e-01 6.98298872e-01
-6.81921482e-01 3.43487084e-01 -8.93991068e-02 4.32332784e-01
-1.36571974e-01 -1.13157856e+00 3.64453375e-01 -3.74478787e-01
-1.22396564e+00 -4.37868237e-01 -1.08741236e+00 -1.06267408e-01
1.46112585e+00 -1.41470745e-01 -6.53593898e-01 -2.84177244e-01
-5.68666101e-01 3.18650186e-01 6.45912409e-01 6.94379032e-01
8.03533673e-01 -8.43105376e-01 -7.09976494e-01 3.38385582e-01
4.62707251e-01 -2.52385259e-01 4.60901409e-02 5.60505331e-01
-7.73029566e-01 2.66298115e-01 -5.10826528e-01 -4.29489821e-01
-1.20600760e+00 8.35770249e-01 4.13184673e-01 3.39822867e-03
-5.27173817e-01 1.04815233e+00 4.33846265e-01 -4.05201375e-01
2.22058356e-01 -3.30910355e-01 -6.14648342e-01 -4.34499681e-01
7.51772225e-01 2.80290455e-01 -4.97414142e-01 -1.76096857e-01
-2.80705065e-01 5.43046594e-01 -2.35609248e-01 1.57053784e-01
1.06043684e+00 4.01830226e-02 -5.62611148e-02 4.43168074e-01
4.80214030e-01 -2.50506133e-01 -1.49139738e+00 -1.56441227e-01
1.79754242e-01 -6.93139359e-02 -4.84962612e-01 -8.81440878e-01
-6.31668329e-01 1.13202333e+00 5.20453751e-01 4.82314765e-01
5.80674231e-01 1.56395406e-01 7.26628751e-02 7.45091677e-01
2.28645951e-01 -1.27107370e+00 7.31393516e-01 5.29160976e-01
1.28480697e+00 -1.39724767e+00 -5.32318830e-01 3.25041153e-02
-1.14374745e+00 1.34017730e+00 1.16503716e+00 -4.18289661e-01
1.74843684e-01 3.94602418e-01 3.76467288e-01 -4.98284519e-01
-1.30789316e+00 5.03926240e-02 1.65379629e-01 9.12513137e-01
5.99724650e-01 -4.13057029e-01 7.32738078e-02 4.74869907e-01
3.16866398e-01 3.42065454e-01 9.08912599e-01 1.25529134e+00
-5.83854854e-01 -8.86933804e-01 -3.01968277e-01 -2.84612235e-02
-2.80719995e-01 -2.17728212e-01 -3.02421004e-01 3.12168032e-01
-1.05474144e-01 6.89561844e-01 1.56600490e-01 6.66078702e-02
2.42072716e-01 1.51191920e-01 4.30380523e-01 -1.21212757e+00
-7.89628565e-01 -4.35387760e-01 -3.01419884e-01 -3.88829380e-01
-3.63578498e-01 -4.63845134e-01 -1.18819439e+00 3.17218870e-01
-1.79447457e-01 9.41612646e-02 6.96225524e-01 8.33929777e-01
6.08499944e-01 8.75740945e-01 5.49641013e-01 -6.09496891e-01
-6.96569800e-01 -1.15788710e+00 -4.02855314e-02 1.44001469e-01
4.72925365e-01 -2.92761922e-01 -2.48464152e-01 6.11984313e-01]
|
[7.747600078582764, 7.842690944671631]
|
4eca75ef-211d-4970-9a20-0b69f2f1bd72
|
perspective-plane-program-induction-from-a-1
|
2006.14708
| null |
https://arxiv.org/abs/2006.14708v1
|
https://arxiv.org/pdf/2006.14708v1.pdf
|
Perspective Plane Program Induction from a Single Image
|
We study the inverse graphics problem of inferring a holistic representation for natural images. Given an input image, our goal is to induce a neuro-symbolic, program-like representation that jointly models camera poses, object locations, and global scene structures. Such high-level, holistic scene representations further facilitate low-level image manipulation tasks such as inpainting. We formulate this problem as jointly finding the camera pose and scene structure that best describe the input image. The benefits of such joint inference are two-fold: scene regularity serves as a new cue for perspective correction, and in turn, correct perspective correction leads to a simplified scene structure, similar to how the correct shape leads to the most regular texture in shape from texture. Our proposed framework, Perspective Plane Program Induction (P3I), combines search-based and gradient-based algorithms to efficiently solve the problem. P3I outperforms a set of baselines on a collection of Internet images, across tasks including camera pose estimation, global structure inference, and down-stream image manipulation tasks.
|
['Joshua B. Tenenbaum', 'Xiuming Zhang', 'Yikai Li', 'Jiajun Wu', 'William T. Freeman', 'Jiayuan Mao']
|
2020-06-25
|
perspective-plane-program-induction-from-a
|
http://openaccess.thecvf.com/content_CVPR_2020/html/Li_Perspective_Plane_Program_Induction_From_a_Single_Image_CVPR_2020_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2020/papers/Li_Perspective_Plane_Program_Induction_From_a_Single_Image_CVPR_2020_paper.pdf
|
cvpr-2020-6
|
['program-induction', 'shape-from-texture']
|
['computer-code', 'computer-vision']
|
[ 5.76882541e-01 3.21601965e-02 -1.88462034e-01 -5.10801017e-01
-6.30360723e-01 -7.08007514e-01 5.75591445e-01 6.10465221e-02
-4.32826765e-02 5.83182387e-02 4.49213356e-01 -1.40088573e-01
-3.09602432e-02 -6.07090652e-01 -1.57153273e+00 -3.99315357e-01
5.76184630e-01 4.61679369e-01 -7.29777012e-03 5.25397249e-02
8.25405002e-01 6.00542784e-01 -1.52862406e+00 4.22073573e-01
6.58108413e-01 5.78544080e-01 5.69112599e-01 8.69928360e-01
1.63338512e-01 1.00180960e+00 -2.04470709e-01 -3.49767148e-01
1.13772161e-01 -3.52149278e-01 -7.14619040e-01 4.40213591e-01
9.24091876e-01 -7.41256177e-01 -2.75577039e-01 1.16249847e+00
1.82782814e-01 4.23057616e-01 6.94707036e-01 -8.93026829e-01
-7.46840656e-01 3.09905618e-01 -8.93628597e-01 -2.82357365e-01
6.28788292e-01 4.31790084e-01 9.86317098e-01 -1.04107690e+00
8.88142586e-01 1.47470522e+00 4.02145386e-01 2.49324411e-01
-1.60506010e+00 -1.96700126e-01 4.62200850e-01 -1.35006653e-02
-1.10138285e+00 -5.46318412e-01 9.56136465e-01 -4.86923575e-01
9.60479319e-01 1.94702327e-01 7.33859658e-01 8.88647079e-01
8.59451108e-03 8.69899929e-01 8.69561374e-01 -4.23406750e-01
2.05439523e-01 -8.34080577e-02 -4.70139384e-02 1.10787594e+00
8.62145275e-02 -1.03630781e-01 -8.03916574e-01 2.40201410e-02
1.25184119e+00 2.42498498e-02 -2.84671843e-01 -7.50275970e-01
-1.31892562e+00 6.07483149e-01 4.75678831e-01 -2.40145341e-01
-3.88962269e-01 4.81300086e-01 1.90415923e-02 -2.51130760e-01
2.04937130e-01 8.58167291e-01 -3.69155586e-01 -2.63710530e-03
-7.17926264e-01 7.80102074e-01 6.58115387e-01 1.35306561e+00
8.44103336e-01 6.98775947e-02 -1.98238716e-01 6.50525689e-01
1.45045742e-01 5.83103478e-01 -1.36070237e-01 -1.72152197e+00
5.30917525e-01 5.50771475e-01 2.98303757e-02 -1.40132809e+00
-2.36297369e-01 -1.78872585e-01 -5.54011524e-01 6.78176805e-02
4.84160990e-01 3.11662436e-01 -7.37813294e-01 1.98703432e+00
2.75432050e-01 1.05985776e-01 -4.01709676e-01 8.69289577e-01
6.05284274e-01 7.24146783e-01 -2.60794401e-01 2.71658778e-01
1.47066522e+00 -1.26113653e+00 -2.73247868e-01 -6.42668128e-01
4.21300977e-01 -5.72207093e-01 1.25409365e+00 4.55328375e-01
-1.24537647e+00 -4.02483135e-01 -6.21628821e-01 -6.65731609e-01
1.29669547e-01 2.34908298e-01 7.17083514e-01 2.23896697e-01
-1.16618776e+00 4.58302438e-01 -5.48690259e-01 -3.20596725e-01
5.94827950e-01 1.70053586e-01 -3.65488291e-01 -3.28645468e-01
-1.15907401e-01 6.66457593e-01 6.02109842e-02 -1.21955127e-02
-1.13167322e+00 -9.02024865e-01 -1.02378893e+00 1.77314728e-01
6.09365702e-01 -1.17091906e+00 1.26763272e+00 -9.29342926e-01
-1.45429337e+00 1.05866790e+00 -6.24010682e-01 -1.67715088e-01
1.21258058e-01 -3.54621232e-01 7.38454401e-01 3.24226350e-01
1.35638878e-01 9.63775039e-01 9.65439200e-01 -1.59505010e+00
-3.02154034e-01 -6.08659565e-01 2.48175338e-01 5.06179690e-01
2.34124184e-01 -1.11983843e-01 -9.31963444e-01 -6.57168806e-01
3.52711886e-01 -1.00894415e+00 -3.99055004e-01 4.09983754e-01
-7.17016339e-01 8.24046880e-02 5.71697235e-01 -9.43161964e-01
6.37303889e-01 -1.94596505e+00 7.88928628e-01 2.11437970e-01
2.86336780e-01 -3.69668782e-01 -1.54555440e-01 4.70000654e-02
-1.36556372e-01 -8.43181554e-03 -1.23852223e-01 -5.73085427e-01
-1.80826068e-01 4.38063554e-02 -4.94962782e-01 3.16780865e-01
1.20368354e-01 1.01169264e+00 -9.52774942e-01 -3.26853991e-01
3.83455783e-01 2.43687645e-01 -1.29147148e+00 2.72003204e-01
-7.22802579e-01 7.14104533e-01 -3.35735351e-01 6.23064816e-01
5.42549670e-01 -2.93420047e-01 6.11095689e-02 -4.05835897e-01
-9.65742208e-03 1.63177252e-01 -7.50412285e-01 2.34562373e+00
-5.50526202e-01 7.54795492e-01 1.09646574e-01 -8.04850399e-01
5.39118290e-01 -4.11222130e-01 2.19620049e-01 -6.38509810e-01
1.48487568e-01 -3.62993509e-01 -4.43467498e-01 -5.23337483e-01
6.03783488e-01 5.28526843e-01 -3.89426649e-02 5.58479190e-01
-6.07010685e-02 -6.18382931e-01 -2.19781116e-01 4.18509364e-01
8.10544550e-01 7.77745426e-01 3.69007796e-01 -1.54909268e-01
4.87154573e-02 2.95233051e-03 3.15695941e-01 8.97119999e-01
2.43857279e-01 7.79968679e-01 8.14856589e-01 -4.12995160e-01
-1.15281403e+00 -8.54300201e-01 4.55815643e-01 1.28920484e+00
3.34775060e-01 -3.11200202e-01 -1.09092462e+00 -1.07735850e-01
-1.31061703e-01 9.90490556e-01 -6.25119805e-01 -1.57413229e-01
-7.72463977e-01 -8.52752477e-02 1.24130197e-01 4.28066462e-01
3.65318120e-01 -9.51030672e-01 -6.45321846e-01 -2.42973581e-01
-4.50885475e-01 -1.24215651e+00 -1.00407386e+00 -1.36882603e-01
-1.11811709e+00 -1.02146447e+00 -3.79741699e-01 -7.92092383e-01
1.15016770e+00 7.07125425e-01 1.28849101e+00 8.79591107e-02
-2.64450312e-01 3.60138088e-01 9.49831158e-02 -1.85252324e-01
-1.44213319e-01 -2.99287558e-01 -3.36627185e-01 -2.47010943e-02
-2.30156615e-01 -7.18813896e-01 -6.42247319e-01 -9.08752531e-02
-6.26083016e-01 7.29059398e-01 6.46807969e-01 6.58046901e-01
8.59701931e-01 -4.26967651e-01 -2.69845068e-01 -1.28270614e+00
2.27475971e-01 -7.32998848e-02 -8.09006929e-01 2.03431204e-01
-1.64353803e-01 4.28395569e-01 4.97833639e-01 -3.19297075e-01
-1.28043520e+00 4.18873459e-01 2.14634836e-01 -7.33939707e-01
-2.30344012e-01 2.99896270e-01 -4.78402972e-01 -1.87181786e-01
6.41800582e-01 4.40386325e-01 -2.33097672e-01 -2.80728221e-01
7.38436043e-01 -9.20633003e-02 9.55087125e-01 -1.13474333e+00
6.89789414e-01 5.95805705e-01 -1.10279107e-02 -9.49243307e-01
-1.00575888e+00 -2.30530053e-01 -9.49231625e-01 -5.28021120e-02
9.63340402e-01 -1.04320323e+00 -8.35765719e-01 4.94748563e-01
-1.70015788e+00 -5.14634132e-01 -5.27745485e-02 -2.13610139e-02
-9.59520578e-01 2.89127290e-01 -3.23409230e-01 -5.60302198e-01
-6.57771900e-03 -1.41201520e+00 1.60066020e+00 1.03642941e-01
-2.45154426e-01 -7.42412925e-01 -1.94458500e-01 5.50783277e-01
-8.52817018e-03 4.72871929e-01 1.32833624e+00 2.23131239e-01
-1.40078759e+00 3.22194025e-02 -3.83428633e-01 -2.52692048e-02
-2.53391981e-01 1.03722058e-01 -1.05349827e+00 8.23914185e-02
5.61738461e-02 -1.42339036e-01 7.06340253e-01 6.54830337e-01
1.81013334e+00 -4.46584493e-01 -2.15792120e-01 1.28494632e+00
1.33180606e+00 1.31211877e-01 5.90603471e-01 1.65973738e-01
1.25530481e+00 6.76347375e-01 4.37812865e-01 3.93556863e-01
7.33728170e-01 8.74681532e-01 5.81473827e-01 8.59718118e-03
-1.52481750e-01 -7.64877915e-01 1.48535684e-01 4.86591130e-01
6.85492679e-02 6.49379659e-03 -7.32477069e-01 3.82984132e-01
-1.77556384e+00 -9.58252549e-01 -2.89755892e-02 2.23288655e+00
6.84617460e-01 -3.37020941e-02 1.46631580e-02 -5.78345120e-01
4.60293561e-01 2.34121427e-01 -8.46246302e-01 -3.45534772e-01
4.86377180e-02 -1.37112159e-02 5.61534703e-01 5.87028682e-01
-8.73243213e-01 1.16567600e+00 5.85893583e+00 3.80535871e-01
-8.35454762e-01 -1.31006330e-01 8.12413990e-01 -1.31470874e-01
-5.09425044e-01 3.54775190e-01 -6.70920730e-01 1.78222761e-01
5.18279135e-01 -8.34210068e-02 1.10438287e+00 9.38802898e-01
3.80333424e-01 -3.43735039e-01 -1.57425439e+00 1.28183270e+00
3.97914261e-01 -1.68525076e+00 5.07960498e-01 3.07362191e-02
8.85992289e-01 -1.79338768e-01 1.45168215e-01 8.22601691e-02
4.44541603e-01 -7.92076826e-01 1.23686874e+00 6.13208115e-01
9.20968115e-01 -6.54430926e-01 -1.85008571e-01 6.26746655e-01
-9.11090314e-01 -6.08808249e-02 -2.79691249e-01 2.56603118e-02
6.17249124e-02 3.34288061e-01 -4.49472934e-01 1.24112472e-01
5.77148020e-01 8.19317162e-01 -7.15090036e-01 8.17643821e-01
-3.34159732e-01 1.21014863e-01 -2.38560200e-01 2.61613220e-01
-1.15680255e-01 -4.03933883e-01 5.14779687e-01 9.01755571e-01
3.88806984e-02 2.27398738e-01 2.02918332e-03 1.43547654e+00
-3.61978561e-01 -3.21152031e-01 -9.79727030e-01 7.38270655e-02
6.32674515e-01 1.00767767e+00 -7.31953740e-01 -1.93738386e-01
-2.25411952e-01 1.21648812e+00 5.52704394e-01 4.89099234e-01
-8.45246673e-01 -8.40559602e-02 6.16901398e-01 -3.74057032e-02
2.66920626e-01 -3.39297295e-01 -7.35029697e-01 -1.30761337e+00
2.53903829e-02 -1.17144012e+00 -1.34294316e-01 -1.37398887e+00
-5.00128269e-01 2.06048042e-01 2.68615335e-01 -9.42758143e-01
-3.41320723e-01 -5.79916358e-01 -6.32528186e-01 7.77474999e-01
-1.11947703e+00 -1.31711507e+00 -5.30692458e-01 4.01568979e-01
8.78997087e-01 2.17543900e-01 6.60383403e-01 -7.32440501e-02
-5.25448561e-01 4.94803905e-01 -2.33699009e-01 2.57017575e-02
5.41167498e-01 -1.11880827e+00 7.06787229e-01 1.01133537e+00
1.90733582e-01 9.36598837e-01 6.38142943e-01 -5.05534708e-01
-2.09049916e+00 -1.20102441e+00 2.85454184e-01 -9.42964077e-01
1.93722263e-01 -5.52747071e-01 -5.74945450e-01 8.21158111e-01
-3.27138114e-03 -1.48919076e-01 3.29486728e-02 1.27993040e-02
-7.07045615e-01 2.04580389e-02 -8.52508128e-01 9.38127279e-01
1.25466013e+00 -8.27929616e-01 -1.91692173e-01 6.33874893e-01
9.13785517e-01 -8.55929434e-01 -3.25466454e-01 -1.98012546e-01
5.84392250e-01 -1.01982915e+00 1.38226318e+00 -5.00184119e-01
1.12211359e+00 -2.99357325e-01 -1.64817050e-01 -1.22548664e+00
-2.70505518e-01 -6.43534362e-01 -4.75998484e-02 9.26464379e-01
1.26873821e-01 -1.72667190e-01 7.96156824e-01 8.16536844e-01
-1.83961734e-01 -6.86568916e-01 -3.34460020e-01 -2.24622145e-01
-2.41415977e-01 -4.79401082e-01 3.84073287e-01 6.49700820e-01
-5.42205572e-01 4.38769192e-01 -3.52520347e-01 3.13219488e-01
9.59336579e-01 4.60762858e-01 1.27158427e+00 -9.06904936e-01
-6.99284613e-01 -5.61771512e-01 -1.59794297e-02 -1.60956454e+00
2.44797662e-01 -7.96336412e-01 3.18729669e-01 -1.34561241e+00
5.73511422e-01 -1.99444532e-01 5.74827194e-01 2.66902983e-01
-1.82486624e-01 -8.93403813e-02 3.60196680e-01 1.60774454e-01
-4.99812007e-01 3.15327108e-01 1.48247838e+00 -1.05165914e-01
-2.68587440e-01 -1.80326581e-01 -1.10842335e+00 9.78765726e-01
3.05298060e-01 -3.73566628e-01 -5.68418086e-01 -1.18842959e+00
4.63730961e-01 5.79329073e-01 6.46347404e-01 -6.22165143e-01
4.49501723e-01 -4.64118034e-01 3.36131781e-01 -7.07688451e-01
5.70325315e-01 -6.92521214e-01 3.07086799e-02 2.01782703e-01
-6.14033878e-01 1.76492140e-01 1.25634950e-03 8.32362652e-01
8.36965293e-02 -1.57609895e-01 7.18236148e-01 -5.46092570e-01
-7.76100099e-01 4.08956647e-01 5.99562302e-02 5.63725568e-02
7.83015668e-01 -3.22627097e-01 -3.57808739e-01 -6.72449231e-01
-3.74472141e-01 4.50320691e-02 8.54592860e-01 3.62504810e-01
8.01571727e-01 -9.92446125e-01 -3.43398392e-01 2.89970368e-01
9.55452695e-02 6.41647637e-01 1.12481132e-01 7.35778511e-01
-1.00117838e+00 2.87936449e-01 -1.24926725e-02 -9.37057137e-01
-1.16484225e+00 4.95289356e-01 2.23860323e-01 1.21505782e-01
-7.18106985e-01 1.16440547e+00 9.77480590e-01 -5.56289613e-01
2.72246927e-01 -6.73913836e-01 2.42502093e-01 -4.14271384e-01
4.80339229e-01 7.95894936e-02 -1.58761039e-01 -4.86641407e-01
-1.66125167e-02 8.65391612e-01 -5.11052571e-02 -9.10241306e-02
1.41855526e+00 -1.69040665e-01 -4.87250894e-01 1.82276800e-01
1.08412004e+00 -6.28556311e-02 -1.59280896e+00 -2.52002358e-01
-2.07200646e-01 -8.65435600e-01 1.62938967e-01 -4.50700372e-01
-8.62107873e-01 9.96166587e-01 2.42988374e-02 -6.66933358e-01
9.68852043e-01 9.00156982e-03 4.56523657e-01 7.80674398e-01
5.56588471e-01 -7.18343079e-01 4.52206910e-01 6.54069006e-01
1.09582317e+00 -9.88068223e-01 1.67887613e-01 -7.51150906e-01
-5.70313215e-01 9.91716146e-01 7.95812011e-01 -3.84895295e-01
1.95000395e-01 2.11595848e-01 -5.56731105e-01 -3.74248862e-01
-7.92756438e-01 1.29569396e-01 2.11181253e-01 6.68301940e-01
5.00055216e-02 -6.27628565e-02 7.09924459e-01 3.25910538e-01
-2.89979666e-01 -2.64307588e-01 5.62003374e-01 7.78033197e-01
-3.59513164e-01 -4.92061853e-01 -4.36451226e-01 3.80090863e-01
9.32159573e-02 -1.49610341e-01 -4.15781409e-01 4.89029586e-01
-1.93902344e-01 4.06574935e-01 5.64465486e-02 -2.88411707e-01
4.59392786e-01 -3.62474859e-01 8.51438105e-01 -8.68212700e-01
-5.60304113e-02 7.57078454e-02 -2.49268651e-01 -1.08649778e+00
-2.02089265e-01 -7.56574214e-01 -9.87628758e-01 -3.10653329e-01
-6.72014058e-02 -6.11704469e-01 7.99499869e-01 9.11190629e-01
4.63208288e-01 5.38534403e-01 4.79707927e-01 -1.53409028e+00
-2.84921020e-01 -4.02241707e-01 -1.12245150e-01 5.49118578e-01
3.70487452e-01 -5.71204245e-01 -6.36128187e-02 5.81664920e-01]
|
[9.065896034240723, -3.0427165031433105]
|
c8e19d8d-5d2e-474a-8119-ba4c41963d71
|
neuraldome-a-neural-modeling-pipeline-on
|
2212.07626
| null |
https://arxiv.org/abs/2212.07626v1
|
https://arxiv.org/pdf/2212.07626v1.pdf
|
NeuralDome: A Neural Modeling Pipeline on Multi-View Human-Object Interactions
|
Humans constantly interact with objects in daily life tasks. Capturing such processes and subsequently conducting visual inferences from a fixed viewpoint suffers from occlusions, shape and texture ambiguities, motions, etc. To mitigate the problem, it is essential to build a training dataset that captures free-viewpoint interactions. We construct a dense multi-view dome to acquire a complex human object interaction dataset, named HODome, that consists of $\sim$75M frames on 10 subjects interacting with 23 objects. To process the HODome dataset, we develop NeuralDome, a layer-wise neural processing pipeline tailored for multi-view video inputs to conduct accurate tracking, geometry reconstruction and free-view rendering, for both human subjects and objects. Extensive experiments on the HODome dataset demonstrate the effectiveness of NeuralDome on a variety of inference, modeling, and rendering tasks. Both the dataset and the NeuralDome tools will be disseminated to the community for further development.
|
['Jingya Wang', 'Lan Xu', 'Jingyi Yu', 'Ye Shi', 'Qianyang Wu', 'Xinru Xu', 'Hongdi Yang', 'Haimin Luo', 'Juze Zhang']
|
2022-12-15
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Zhang_NeuralDome_A_Neural_Modeling_Pipeline_on_Multi-View_Human-Object_Interactions_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Zhang_NeuralDome_A_Neural_Modeling_Pipeline_on_Multi-View_Human-Object_Interactions_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['human-object-interaction-detection']
|
['computer-vision']
|
[-5.20132668e-02 -3.06540072e-01 4.01881754e-01 -5.22343278e-01
-1.19910985e-01 -3.33400428e-01 4.44663733e-01 -5.50144434e-01
1.09108523e-01 3.54957819e-01 8.48801881e-02 5.11307754e-02
1.93086162e-01 -6.03380680e-01 -6.79248512e-01 -2.77513832e-01
5.66619337e-02 7.62334347e-01 5.10160685e-01 3.03425714e-02
4.02119756e-02 6.80950046e-01 -1.78522301e+00 6.76303983e-01
3.14227790e-01 1.06823492e+00 2.72755206e-01 9.89610493e-01
3.33677351e-01 9.20935512e-01 -4.14358169e-01 -4.60063726e-01
3.65109980e-01 1.07874937e-01 -5.63887596e-01 2.51591742e-01
9.04086590e-01 -8.94110262e-01 -4.04917210e-01 7.15499163e-01
7.61351883e-01 5.30284643e-01 4.98649150e-01 -1.23343742e+00
-5.17751157e-01 -8.39902386e-02 -5.07708311e-01 1.68953821e-01
6.87567770e-01 4.99684602e-01 4.02891785e-01 -1.24615502e+00
9.05476213e-01 1.64041436e+00 7.55638421e-01 4.34212357e-01
-9.19092536e-01 -5.73819816e-01 3.38120073e-01 3.05140279e-02
-1.10349512e+00 -6.12115920e-01 7.27176070e-01 -7.28269100e-01
1.05652690e+00 3.36434364e-01 9.27070320e-01 1.32207263e+00
1.86011344e-01 8.24395478e-01 7.57213175e-01 8.42136368e-02
8.23955834e-02 -2.74293125e-01 1.38520211e-01 8.50730836e-01
6.95576370e-02 2.19939590e-01 -6.66337788e-01 -3.81501839e-02
1.32822037e+00 1.59634396e-01 -2.77289331e-01 -6.91670358e-01
-1.22347593e+00 2.39481166e-01 2.46994063e-01 -4.63871509e-01
-4.40854132e-01 2.76472569e-01 4.37520802e-01 -3.55742499e-02
5.75277388e-01 5.39652742e-02 -5.69146335e-01 -2.85276592e-01
-4.42699462e-01 6.45571887e-01 7.20685542e-01 1.29179037e+00
1.74042284e-01 -4.28111618e-03 -2.33668879e-01 6.79382622e-01
2.56446868e-01 5.64155936e-01 -1.66393176e-01 -1.43845451e+00
4.62736368e-01 5.56374729e-01 3.09266746e-01 -1.25619519e+00
-5.73747993e-01 1.44031011e-02 -8.76331210e-01 4.70667660e-01
4.48860914e-01 -6.25479817e-02 -8.15669179e-01 1.35060835e+00
8.06650281e-01 1.45243198e-01 -3.75069112e-01 1.25862551e+00
1.21640944e+00 3.94340515e-01 -8.53372440e-02 2.18710780e-01
1.61269569e+00 -1.15913439e+00 -6.86096489e-01 -2.52674758e-01
1.60676390e-01 -6.40272856e-01 1.27370071e+00 6.60701871e-01
-1.47100043e+00 -1.05101705e+00 -7.65535831e-01 -5.60230017e-01
-1.52719781e-01 2.43779346e-01 7.80484498e-01 1.28141314e-01
-7.55006015e-01 4.57383752e-01 -1.13202870e+00 -1.36860326e-01
4.91243422e-01 2.49094322e-01 -3.37257773e-01 -2.58098692e-01
-7.44018257e-01 6.59389317e-01 8.96784291e-02 3.88470083e-01
-9.79822874e-01 -7.40792572e-01 -8.67168725e-01 -1.61754072e-01
5.75027406e-01 -1.27951884e+00 1.22383857e+00 -4.43713069e-01
-1.38521934e+00 9.89775360e-01 -1.36628553e-01 1.24575131e-01
8.22418034e-01 -6.59529805e-01 -4.63791192e-02 1.00473411e-01
-1.17672548e-01 6.85953498e-01 8.95857632e-01 -1.25895143e+00
-2.87132412e-01 -6.54566586e-01 3.51809204e-01 3.78547996e-01
1.72475845e-01 1.35274470e-01 -9.06175554e-01 -7.91280508e-01
1.62958521e-02 -1.09788013e+00 -2.18775887e-02 5.64945281e-01
-5.21062791e-01 -1.81015879e-01 1.11214936e+00 -7.52319992e-01
7.28823543e-01 -2.14908648e+00 3.95655125e-01 -1.92603350e-01
4.15998846e-01 2.59256661e-01 1.38236091e-01 9.63772684e-02
-9.69149694e-02 -5.09240568e-01 2.02696353e-01 -7.36781657e-01
2.01896336e-02 -1.01625929e-02 -1.06126532e-01 3.71931374e-01
-1.00886799e-01 1.10362542e+00 -6.62405014e-01 -5.12902319e-01
5.81726730e-01 7.95030057e-01 -8.00413609e-01 6.78488672e-01
-4.71113265e-01 6.78967595e-01 -4.73006636e-01 8.97775531e-01
6.95501685e-01 -6.39742732e-01 -1.99388601e-02 -4.94619668e-01
1.25885695e-01 1.85888167e-02 -1.22560418e+00 1.92655718e+00
-2.46369451e-01 6.04676902e-01 3.69428158e-01 -4.35526937e-01
4.21995938e-01 3.01760942e-01 5.33286512e-01 -2.73993611e-01
3.96679491e-01 -4.55263853e-01 -2.82462358e-01 -9.07709420e-01
5.17106593e-01 3.84454131e-01 1.79532275e-01 4.85312223e-01
-3.10205549e-01 1.07032442e-02 -6.57732859e-02 8.86225849e-02
8.87197256e-01 7.06849337e-01 1.42109975e-01 1.54523016e-03
1.77168757e-01 -1.43141076e-01 3.20499808e-01 4.23444152e-01
-2.16025248e-01 9.41174805e-01 1.24992751e-01 -1.11480212e+00
-9.24052954e-01 -1.19382238e+00 -4.98319007e-02 1.18146884e+00
1.15195826e-01 -4.12743419e-01 -7.10687697e-01 -3.83258909e-01
7.09629878e-02 2.61728436e-01 -5.20179570e-01 2.92494502e-02
-7.81017542e-01 -4.88847971e-01 2.08521098e-01 1.02396536e+00
7.54157424e-01 -1.33586931e+00 -1.25931144e+00 -1.12426318e-01
-3.31245154e-01 -1.50412774e+00 -5.74624598e-01 -4.77265090e-01
-6.78532779e-01 -1.35661018e+00 -4.34304684e-01 -3.58410895e-01
7.67785490e-01 5.10492682e-01 1.48420942e+00 2.09999382e-01
-6.14524722e-01 4.98603076e-01 -2.30396278e-02 -3.27709973e-01
3.78738903e-02 -2.75467038e-01 1.88768923e-01 -2.87783027e-01
2.23215431e-01 -6.79367483e-01 -6.80216610e-01 4.61412370e-01
-5.58868110e-01 5.81327856e-01 2.27187946e-03 4.05430168e-01
5.39586306e-01 -1.79230630e-01 -5.86836748e-02 -8.17730486e-01
3.31391722e-01 -1.41594633e-01 -7.62238324e-01 8.16579387e-02
1.51066691e-01 -4.62015629e-01 3.83845180e-01 -5.83520114e-01
-1.15441906e+00 2.79623687e-01 -2.73583215e-02 -9.70048368e-01
-4.43189770e-01 1.27146810e-01 -4.57817405e-01 1.50103465e-01
5.12918055e-01 -1.42521560e-01 -1.68129474e-01 -5.21337211e-01
4.15961772e-01 3.33337128e-01 7.80046046e-01 -6.81246161e-01
5.32860577e-01 6.43618643e-01 -9.45022553e-02 -8.20244730e-01
-8.05952251e-01 -1.51923329e-01 -9.14472401e-01 -4.76473182e-01
1.19198334e+00 -1.13439202e+00 -1.35759509e+00 7.96380460e-01
-1.44707358e+00 -5.39128006e-01 2.44497824e-02 4.34180379e-01
-6.39713466e-01 1.61793977e-01 -6.90418482e-01 -7.96965361e-01
-2.10104018e-01 -1.27330685e+00 1.46283531e+00 5.52066378e-02
-4.29063737e-01 -8.79951179e-01 -1.85092986e-01 7.46122181e-01
5.64666949e-02 3.99376482e-01 5.64895689e-01 -5.21405749e-02
-1.08087635e+00 -7.77701735e-02 -3.16613525e-01 -5.70084751e-02
9.95347351e-02 1.75929084e-01 -1.09341443e+00 -3.98425281e-01
6.57314509e-02 -5.23322642e-01 3.45797509e-01 6.92714095e-01
1.61701131e+00 1.50238335e-01 -3.54448408e-01 1.03305757e+00
8.38127553e-01 3.17621976e-01 5.65035105e-01 -1.28594443e-01
1.31484795e+00 5.95015526e-01 5.07809043e-01 6.27020657e-01
6.51045918e-01 9.92981255e-01 3.39098513e-01 -8.87495726e-02
-1.35676801e-01 -5.60667217e-02 9.81557965e-02 7.79307127e-01
-5.32231092e-01 -2.99606174e-01 -8.83308530e-01 1.16661720e-01
-1.70508265e+00 -9.44092572e-01 -1.74751535e-01 1.84824133e+00
4.11241859e-01 1.07132867e-01 2.57290393e-01 -3.23580444e-01
3.89879555e-01 8.47771093e-02 -8.94952714e-01 2.26141699e-02
3.24959457e-01 -1.78363740e-01 -9.08447355e-02 3.40193957e-01
-1.09143472e+00 9.18179154e-01 6.59654856e+00 1.45648733e-01
-9.04430151e-01 -8.55831727e-02 4.62461889e-01 -3.73176247e-01
1.20905124e-01 -3.74401599e-01 -8.16338480e-01 3.39575112e-01
4.69580770e-01 4.59610790e-01 5.62562048e-01 9.98457372e-01
2.80362010e-01 -1.30522773e-01 -1.25976431e+00 1.34394300e+00
2.52661794e-01 -1.06180608e+00 -2.15150878e-01 -1.33524984e-01
4.72700477e-01 -6.19111583e-02 -5.01468666e-02 1.81851119e-01
5.73972106e-01 -9.58233118e-01 6.29978359e-01 8.31148863e-01
8.49263072e-01 -3.61440063e-01 2.63744235e-01 5.63371122e-01
-1.37671375e+00 2.19922930e-01 -1.04528636e-01 -3.15581828e-01
5.52341402e-01 4.29327726e-01 -4.59140718e-01 3.82452250e-01
1.08594632e+00 8.29114556e-01 -5.11803448e-01 5.25340378e-01
1.46524385e-01 -8.55330005e-02 -2.27954164e-01 4.62191224e-01
-4.25942272e-01 -2.04802275e-01 4.33530211e-01 8.87437940e-01
-8.17605332e-02 5.53718865e-01 4.81056243e-01 9.98888910e-01
8.65482092e-02 -6.04010761e-01 -6.93100750e-01 8.01707134e-02
3.56648296e-01 1.15556407e+00 -8.51152182e-01 -6.78950429e-01
-3.94787520e-01 1.17860568e+00 4.09557730e-01 4.83657598e-01
-1.07029605e+00 -1.23601807e-02 9.57680285e-01 3.11093003e-01
1.90097168e-01 -6.21946514e-01 -2.15340063e-01 -1.43869257e+00
2.94047952e-01 -1.03028011e+00 1.11909389e-01 -1.37419474e+00
-1.28313744e+00 6.71980381e-01 5.92780828e-01 -1.13561058e+00
-2.76377738e-01 -8.24414313e-01 -2.98303246e-01 6.40997052e-01
-7.97860026e-01 -1.31222880e+00 -1.05905855e+00 7.11369872e-01
9.03734863e-01 -1.49736688e-01 5.64325571e-01 4.74134982e-01
-7.19142675e-01 1.68990687e-01 -6.74304485e-01 1.74542934e-01
5.57043493e-01 -9.51323330e-01 9.08472657e-01 5.45448542e-01
-7.23378435e-02 8.56614828e-01 4.03272331e-01 -9.16177332e-01
-1.68259132e+00 -1.04357588e+00 3.09260786e-01 -1.16244888e+00
1.45923775e-02 -6.91539645e-01 -8.96320105e-01 1.08641422e+00
-4.53706346e-02 3.62972707e-01 2.95675874e-01 -3.62945162e-02
-8.29005763e-02 3.02151263e-01 -9.11388278e-01 6.85891211e-01
1.86121976e+00 -6.67349756e-01 -4.91696328e-01 3.14189464e-01
7.11547792e-01 -1.26224279e+00 -9.61676002e-01 4.61826235e-01
1.01537275e+00 -1.33841765e+00 1.54108119e+00 -6.84698462e-01
5.70602894e-01 -2.89179057e-01 -8.35062414e-02 -9.65554118e-01
-3.37159991e-01 -4.71531361e-01 -5.93516886e-01 7.32689321e-01
-4.12370205e-01 -2.58402407e-01 8.55807006e-01 1.11069775e+00
-1.72034055e-01 -8.20061803e-01 -4.86753464e-01 -3.39471549e-01
-4.27803963e-01 -4.87440050e-01 6.39752209e-01 7.62943268e-01
-4.93855894e-01 2.75977999e-01 -4.90402460e-01 2.07678124e-01
6.52342856e-01 9.16929990e-02 1.48239172e+00 -1.35336256e+00
-3.98242652e-01 2.09580697e-02 -1.26941979e-01 -1.44199455e+00
1.60058111e-01 -4.22020465e-01 -5.28221726e-02 -1.37331676e+00
1.25212178e-01 -2.07890093e-01 4.36901122e-01 2.28051111e-01
-1.95117608e-01 2.42705926e-01 2.26403221e-01 2.30289981e-01
-5.82519293e-01 4.58032012e-01 1.74645662e+00 1.77203551e-01
-1.44908741e-01 3.09811067e-02 -1.97420970e-01 1.26657605e+00
2.89005578e-01 -6.17082044e-02 -7.02640235e-01 -8.04297388e-01
2.32265338e-01 3.34188342e-01 1.00551021e+00 -1.10903966e+00
1.41583346e-02 -2.75961101e-01 8.93734574e-01 -1.08674002e+00
9.68965471e-01 -8.61189663e-01 6.35118008e-01 2.87135869e-01
-1.47040680e-01 5.12449503e-01 1.48760617e-01 3.41763854e-01
2.50623256e-01 4.66723949e-01 5.36106944e-01 -4.62445080e-01
-6.99293017e-01 6.60865843e-01 -2.64780037e-02 1.48675591e-01
8.60925972e-01 -3.94359022e-01 -2.66597480e-01 -3.49598825e-01
-9.53501582e-01 2.13447407e-01 6.37651026e-01 5.87139964e-01
9.01008248e-01 -1.38334215e+00 -4.26312357e-01 5.25526106e-01
-1.43821016e-01 5.49321175e-01 4.54693705e-01 6.33166492e-01
-7.43516386e-01 1.44101948e-01 -4.28937256e-01 -6.42434776e-01
-1.44358611e+00 6.40706062e-01 2.96101570e-01 9.62174535e-02
-1.13374257e+00 8.48736048e-01 5.55962384e-01 -5.32539666e-01
3.33152086e-01 -7.02926219e-01 3.29626687e-02 -3.38975668e-01
6.36208057e-01 6.64620996e-01 -1.91890851e-01 -6.34101450e-01
-2.15349093e-01 7.34454274e-01 1.61928877e-01 -2.40891464e-02
1.04302108e+00 -5.96670471e-02 9.02464911e-02 5.29786587e-01
9.86670077e-01 -3.03872645e-01 -1.70773113e+00 3.02405432e-02
-6.61510527e-01 -8.47347736e-01 -2.39016831e-01 -4.93717253e-01
-1.18782401e+00 1.03649795e+00 5.18480241e-01 -7.21837208e-02
8.28439295e-01 -2.07613200e-01 1.01856089e+00 5.60905516e-01
5.11353254e-01 -8.54049206e-01 3.73700738e-01 5.44936538e-01
1.21095192e+00 -1.15414858e+00 1.93111643e-01 -7.08778679e-01
-6.08966589e-01 9.57435966e-01 1.23631907e+00 -1.57878980e-01
7.14868844e-01 5.36448658e-01 -2.35304814e-02 -6.80962622e-01
-7.79017448e-01 2.78864712e-01 3.66067469e-01 6.83223546e-01
4.09475356e-01 -1.25211373e-01 3.76445115e-01 4.40107822e-01
-4.48460639e-01 3.24338317e-01 4.79109362e-02 9.47702944e-01
9.46569517e-02 -6.78842485e-01 -3.63413453e-01 3.13088208e-01
-1.48210898e-01 2.30986983e-01 -3.42660666e-01 7.13118315e-01
2.17810601e-01 5.10408163e-01 3.16218078e-01 -3.59330088e-01
5.96178710e-01 -1.01177186e-01 6.64376616e-01 -4.64227766e-01
-6.35079563e-01 2.20947918e-02 7.11155608e-02 -1.00462961e+00
-5.45590460e-01 -6.33468151e-01 -1.18222177e+00 -5.28096139e-01
-4.89952229e-02 -6.68372333e-01 2.33474836e-01 8.07993472e-01
5.67514896e-01 7.87339032e-01 2.61497553e-02 -1.47216320e+00
-2.68765748e-01 -9.21811163e-01 -2.00362459e-01 7.86157191e-01
2.55830467e-01 -1.11754155e+00 1.70731887e-01 4.08333778e-01]
|
[6.959239482879639, -1.061302661895752]
|
48723ebb-3c16-480a-9179-054632bd7e1d
|
voar-a-visual-and-integrated-ontology
| null | null |
https://aclanthology.org/L14-1658
|
https://aclanthology.org/L14-1658.pdf
|
VOAR: A Visual and Integrated Ontology Alignment Environment
|
Ontology alignment is a key process for enabling interoperability between ontology-based systems in the Linked Open Data age. From two input ontologies, this process generates an alignment (set of correspondences) between them. In this paper we present VOAR, a new web-based environment for ontology alignment visualization and manipulation. Within this graphical environment, users can manually create/edit correspondences and apply a set of operations on alignments (filtering, merge, difference, etc.). VOAR allows invoking external ontology matching systems that implement a specific alignment interface, so that the generated alignments can be manipulated within the environment. Evaluating multiple alignments together against a reference one can also be carried out, using classical evaluation metrics (precision, recall and f-measure). The status of each correspondence with respect to its presence or absence in reference alignment is visually represented. Overall, the main new aspect of VOAR is the visualization and manipulation of alignments at schema level, in an integrated, visual and web-based environment.
|
['Renata Vieira', 'Cassia Trojahn', 'Bernardo Severo']
|
2014-05-01
| null | null | null |
lrec-2014-5
|
['ontology-matching']
|
['knowledge-base']
|
[ 9.01630521e-02 1.60487592e-01 1.09933138e-01 -3.38158876e-01
-2.01025680e-01 -8.38681281e-01 6.67652309e-01 1.09237635e+00
-3.61093223e-01 3.01668614e-01 7.63436481e-02 -1.93138048e-01
-7.27625132e-01 -1.27351236e+00 -1.37964353e-01 -1.09358355e-01
-6.60216361e-02 8.34429979e-01 5.76505899e-01 -6.05547130e-01
1.57031238e-01 7.03789771e-01 -2.31674385e+00 3.49481165e-01
8.60637069e-01 5.28004169e-01 2.39668727e-01 4.89723802e-01
-9.00427997e-01 1.05620593e-01 -7.23722756e-01 -4.96927947e-01
1.13901883e-01 -2.68876791e-01 -9.50175524e-01 -3.76949221e-01
4.29758310e-01 4.90640938e-01 6.19095206e-01 1.29934657e+00
3.05431098e-01 -1.18787356e-01 4.40628469e-01 -1.63272583e+00
-2.93500036e-01 4.50987548e-01 3.02917838e-01 -8.10125470e-02
1.11401224e+00 -1.44723505e-01 8.43526363e-01 -6.07100725e-01
1.04973567e+00 1.28005314e+00 3.52686018e-01 -3.71689126e-02
-1.31946957e+00 -4.59054440e-01 -3.89760882e-01 3.85154128e-01
-1.32933712e+00 -2.43939877e-01 1.76989943e-01 -8.23182523e-01
9.32997048e-01 1.02767491e+00 8.37606072e-01 2.12498367e-01
1.49236560e-01 -2.91209459e-01 9.00784075e-01 -8.67519379e-01
2.86627263e-01 3.98166507e-01 3.76343757e-01 4.41775143e-01
4.85897154e-01 -3.85911942e-01 -4.51915354e-01 -4.49793637e-01
3.46051633e-01 -2.28984267e-01 -8.62741172e-02 -7.57016897e-01
-9.73408222e-01 2.28016928e-01 2.63096273e-01 7.25595474e-01
-2.77082592e-01 -4.18471634e-01 3.29037637e-01 5.67831755e-01
-9.26012248e-02 5.88128567e-01 -1.04031906e-01 1.41248675e-02
-4.43777859e-01 4.80749577e-01 8.39448571e-01 1.17627382e+00
1.11135960e+00 -6.88335419e-01 -1.26002654e-02 6.36534214e-01
6.28800035e-01 2.02138141e-01 5.29548466e-01 -6.50409102e-01
1.29392698e-01 1.50780940e+00 2.81880856e-01 -1.09066427e+00
-3.89218211e-01 -2.95148715e-02 -9.31993499e-02 8.63684058e-01
1.64113805e-01 7.08201587e-01 -4.29827511e-01 1.36717868e+00
8.32427382e-01 -4.63506192e-01 4.09326941e-01 4.93496537e-01
1.11669719e+00 1.24312051e-01 2.70581275e-01 -6.12531863e-02
1.81973541e+00 -2.71968722e-01 -1.15147960e+00 1.21524528e-01
8.33687067e-01 -1.24404430e+00 1.00161624e+00 1.89465024e-02
-1.08632135e+00 -4.46268916e-01 -1.26816130e+00 -2.66726702e-01
-1.11943865e+00 -4.01747733e-01 7.49507993e-02 3.93827885e-01
-1.06226528e+00 5.54839194e-01 -4.89014924e-01 -9.56460238e-01
-5.03593832e-02 3.33710164e-01 -8.06176424e-01 3.01961213e-01
-1.22573173e+00 1.29061520e+00 8.53141785e-01 -3.51167828e-01
1.59408078e-01 -5.88093340e-01 -9.29615378e-01 2.44096369e-01
8.11130106e-02 -6.79118216e-01 9.25584853e-01 -7.63891757e-01
-1.06608367e+00 1.35120082e+00 -1.00543536e-01 -2.90244132e-01
6.94351375e-01 -7.78551772e-02 -9.53921080e-01 -3.52638334e-01
2.32546806e-01 2.19095886e-01 1.29642850e-03 -1.29083920e+00
-8.86852443e-01 -6.27502620e-01 2.15984955e-01 6.86559752e-02
-2.74527758e-01 5.42618036e-01 -4.72807586e-01 -2.75916815e-01
1.60803512e-01 -3.77168477e-01 4.19501550e-02 1.65400654e-01
1.69820651e-01 -2.82997727e-01 6.58647180e-01 -7.47777939e-01
1.73394597e+00 -1.90524197e+00 2.18814969e-01 7.37501502e-01
8.23220834e-02 2.97874987e-01 3.41561735e-02 9.76303279e-01
-3.78778070e-01 2.03571662e-01 -2.43983567e-01 2.79372036e-01
3.06241035e-01 3.39442492e-01 3.58010709e-01 -9.93083417e-02
-2.34647885e-01 2.58406103e-01 -8.47815573e-01 -6.89817369e-01
5.62408805e-01 3.68522733e-01 -2.35159039e-01 2.22476304e-01
-2.50768960e-01 2.90635198e-01 -9.00375545e-02 3.12778860e-01
7.02187479e-01 1.21318415e-01 8.31654072e-01 -3.03111851e-01
-6.09538078e-01 1.99692607e-01 -1.77184200e+00 1.78068936e+00
-6.31834626e-01 2.91043490e-01 -2.08076805e-01 -6.29092276e-01
1.51530850e+00 7.35743105e-01 3.25428098e-01 -8.18402767e-01
-5.31765334e-02 4.94509667e-01 -2.30218694e-01 -8.20621669e-01
4.43291247e-01 5.72747409e-01 9.69235301e-02 4.61638212e-01
3.57606225e-02 1.37189448e-01 8.57307196e-01 8.51109903e-03
7.13806689e-01 3.74765843e-01 8.90600741e-01 -3.81025165e-01
1.06252706e+00 5.03295995e-02 1.73967510e-01 4.32331920e-01
4.99235153e-01 -9.45381373e-02 4.15035367e-01 -7.07541645e-01
-1.36462331e+00 -1.11803663e+00 -4.36997294e-01 8.74058008e-01
7.99940526e-02 -1.07404339e+00 -7.34977722e-01 -2.23037168e-01
-5.72144613e-02 5.96734226e-01 -3.54463339e-01 5.00282571e-02
-4.38648075e-01 -1.61474526e-01 1.07162803e-01 3.88914123e-02
2.43811961e-02 -1.12795496e+00 -1.07412899e+00 2.92380124e-01
-1.39032155e-01 -6.94990456e-01 3.50319356e-01 -3.80431175e-01
-8.45898390e-01 -1.27297974e+00 1.54883578e-01 -6.53612971e-01
5.95850229e-01 -2.23040730e-01 1.42108417e+00 4.26065534e-01
-4.20730829e-01 1.59861818e-01 -4.51554835e-01 -6.72011733e-01
-9.91355300e-01 -5.52275553e-02 -3.23872179e-01 -1.08829305e-01
4.74977940e-01 -7.35729218e-01 -1.43167451e-01 5.96009910e-01
-1.35756505e+00 1.56537995e-01 -1.10347062e-01 1.32755578e-01
6.64966404e-01 -1.64400294e-01 1.34857535e-01 -6.39967978e-01
8.15726221e-01 -2.61292696e-01 -1.20045924e+00 7.51405716e-01
-5.42703092e-01 2.10609779e-01 8.19215998e-02 7.17422143e-02
-7.18895257e-01 2.52519324e-02 -1.24585003e-01 2.17131913e-01
-3.65377545e-01 8.96408379e-01 -6.62073195e-01 8.56035352e-02
6.53415203e-01 -4.05052960e-01 1.81123808e-01 -8.47637951e-01
4.18141633e-01 9.89891350e-01 5.73208511e-01 -4.83963549e-01
6.51589513e-01 4.32122141e-01 -1.30424723e-01 -3.30214590e-01
-1.31028473e-01 -4.95047837e-01 -1.13616204e+00 -4.24084783e-01
7.39345729e-01 -3.21714848e-01 -7.22941279e-01 -3.60197052e-02
-1.14566922e+00 1.45026594e-01 -3.96902949e-01 2.61771172e-01
-3.58126730e-01 1.44812286e-01 5.54176450e-01 -6.61730647e-01
-4.03352976e-01 -1.19047832e+00 5.96497715e-01 2.53849179e-01
-8.34802151e-01 -8.25774848e-01 3.70766670e-01 8.33452046e-02
2.96320558e-01 5.40126622e-01 1.07328129e+00 -9.16903913e-01
-3.21511358e-01 -5.54646373e-01 1.39298454e-01 -1.38015285e-01
3.70370001e-01 4.47373271e-01 -5.96564114e-01 -2.97121517e-03
-6.62527680e-01 6.97272241e-01 -3.92795920e-01 -4.72491235e-01
6.37783766e-01 -2.30231643e-01 -5.99518836e-01 1.92043215e-01
1.50500393e+00 4.12539989e-01 9.41120207e-01 9.74634409e-01
2.85040259e-01 9.15894687e-01 8.43677521e-01 1.69590801e-01
2.18569294e-01 1.48797762e+00 5.39173245e-01 -2.30348110e-02
-9.25156567e-03 1.41731232e-01 -1.21798538e-01 4.96974260e-01
-2.49313563e-01 -1.04731217e-01 -1.18384147e+00 1.84105456e-01
-2.08907437e+00 -1.00700033e+00 -7.72398174e-01 2.97174716e+00
4.26533014e-01 -8.01683962e-02 1.89495474e-01 3.32485795e-01
9.81749892e-01 -3.23461831e-01 3.48169208e-01 -6.58779085e-01
3.13022145e-04 4.83831167e-01 1.07642002e-01 7.97945917e-01
-7.20741868e-01 5.91745615e-01 5.79796267e+00 1.85239226e-01
-9.36342180e-01 2.15013430e-01 -5.16300797e-01 1.97405592e-01
-5.39974034e-01 2.15604484e-01 -6.00140512e-01 3.93847108e-01
9.48625624e-01 -7.84272611e-01 2.23410353e-01 5.58750987e-01
2.48082727e-01 -7.11142272e-02 -1.13836515e+00 6.90170884e-01
-1.30294949e-01 -1.51577985e+00 2.94662356e-01 2.00873449e-01
1.33882314e-01 -2.17876852e-01 -7.16171265e-01 -3.16842377e-01
3.30414325e-01 -7.43890762e-01 8.81453872e-01 8.99491429e-01
6.76982105e-01 -7.27283478e-01 7.65868664e-01 -2.78141946e-01
-1.33563471e+00 2.03110397e-01 -1.44055560e-01 1.66530311e-01
3.71753931e-01 3.62605751e-01 -4.54363555e-01 1.33637679e+00
9.33128059e-01 1.92616999e-01 -7.22776651e-01 1.48756742e+00
-8.65045562e-02 -4.20002937e-01 -3.92710418e-01 2.55737185e-01
-4.00708646e-01 -5.87668061e-01 6.65567636e-01 1.28940535e+00
6.56975269e-01 -4.11711723e-01 7.34546781e-02 6.78921223e-01
3.50791067e-01 9.02562261e-01 -5.49173415e-01 2.33755589e-01
8.38711977e-01 1.22589135e+00 -5.41965187e-01 -4.97163624e-01
-3.16512138e-01 5.70071042e-01 7.85237625e-02 -6.40287697e-02
-5.38628757e-01 -7.40362465e-01 1.00207281e+00 3.06375533e-01
-7.25100264e-02 -2.08482482e-02 7.64404656e-03 -7.22956836e-01
3.83423448e-01 -1.10627520e+00 7.71844804e-01 -1.02438259e+00
-6.25757992e-01 6.49778426e-01 2.43738994e-01 -1.29141819e+00
-6.78071007e-02 -3.29875529e-01 -4.46951210e-01 1.13268995e+00
-8.20351243e-01 -8.96112084e-01 -7.58025527e-01 3.14020187e-01
-1.51158601e-01 1.61827713e-01 1.57232022e+00 8.23595107e-01
-3.17654908e-01 8.99130404e-02 -7.24215284e-02 -2.19521672e-02
6.98613346e-01 -1.29847753e+00 4.31720376e-01 6.45696104e-01
2.20799029e-01 6.74329281e-01 1.00697184e+00 -6.56375587e-01
-7.17532814e-01 -5.61802208e-01 1.31691980e+00 -3.76371831e-01
6.15839422e-01 -2.87413836e-01 -1.28591812e+00 6.62670135e-01
3.96643221e-01 -1.02050208e-01 8.88583004e-01 -5.59939593e-02
-4.47591513e-01 -2.65740573e-01 -1.27752364e+00 7.18951702e-01
1.10407078e+00 -2.98624694e-01 -7.86840975e-01 2.86178321e-01
1.97383687e-01 -3.34918737e-01 -1.74850404e+00 3.50452811e-01
8.65781248e-01 -1.22066760e+00 1.03418744e+00 -8.32046449e-01
-1.63507894e-01 -1.03879404e+00 -9.42473337e-02 -9.85121906e-01
-2.40156483e-02 -5.09764314e-01 4.64993775e-01 1.62184632e+00
3.93045515e-01 -8.86293650e-01 3.16370763e-02 3.57697755e-01
-7.80243278e-02 1.54107455e-02 -8.00807476e-01 -7.98315108e-01
-5.42301655e-01 -2.19159752e-01 1.49364018e+00 1.16497743e+00
5.41809142e-01 -1.22787513e-01 3.16176265e-01 6.02828264e-01
2.04741150e-01 -1.37055919e-01 1.13859189e+00 -2.08436489e+00
1.17191941e-01 -5.82228482e-01 -1.18680465e+00 3.95827264e-01
-4.07944977e-01 -1.13861597e+00 -5.51565945e-01 -2.21736169e+00
-4.42597896e-01 -5.40663362e-01 -1.56670660e-01 4.19150740e-01
1.78983942e-01 1.15321539e-01 2.13012993e-01 4.78131831e-01
-1.63293839e-01 -2.17797294e-01 5.08743227e-01 1.79060176e-01
-1.35558456e-01 -4.45667177e-01 -5.84561713e-02 5.14683545e-01
7.22499609e-01 -8.33920896e-01 -6.40453845e-02 -1.30038738e-01
6.13451719e-01 -3.41458887e-01 1.73744768e-01 -1.39731634e+00
1.12785362e-01 -2.33960345e-01 -1.09398954e-01 -2.78137177e-01
-3.46232913e-02 -1.22350585e+00 1.41292536e+00 6.44696057e-01
-2.46108532e-01 6.50579393e-01 1.73235163e-01 -1.77149504e-01
-2.56309360e-01 -7.77855098e-01 4.92420435e-01 -7.65037090e-02
-7.11542845e-01 -1.95720017e-01 -3.40266198e-01 -3.26661110e-01
1.25484753e+00 -4.30550545e-01 -3.89665842e-01 2.42034525e-01
-1.17815924e+00 1.29159629e-01 9.09504652e-01 7.31225133e-01
-3.18342708e-02 -1.43368089e+00 -4.52964395e-01 1.98584020e-01
9.01489198e-01 -2.20766142e-01 -1.70891266e-02 5.00718117e-01
-9.47875500e-01 -3.09619196e-02 -8.02815616e-01 -5.07873356e-01
-1.96628237e+00 6.40190303e-01 5.55888414e-01 -1.82651460e-01
-5.30692458e-01 9.35734902e-03 -3.52505416e-01 -8.40653658e-01
3.10444236e-02 -4.13524620e-02 -7.13689864e-01 2.94129342e-01
7.56395519e-01 4.74363863e-01 7.07599938e-01 -8.73006463e-01
-5.82893908e-01 8.31503272e-01 4.68245924e-01 -3.31057727e-01
1.26750302e+00 1.76320076e-02 -6.01563394e-01 4.79888141e-01
7.28984356e-01 3.12832236e-01 -2.48051435e-01 -6.61783665e-02
6.64069593e-01 -7.06978321e-01 -5.91376901e-01 -7.43390024e-01
-5.36785245e-01 4.35716331e-01 9.94010627e-01 7.08771944e-01
6.66142404e-01 1.26264274e-01 -1.92785829e-01 1.36742264e-01
2.55854726e-01 -8.27442765e-01 -6.80493951e-01 6.80323392e-02
1.10931957e+00 -5.74312449e-01 1.32993862e-01 -7.10286081e-01
3.45046399e-03 1.45981050e+00 3.94116282e-01 4.81142342e-01
3.69069487e-01 3.84519428e-01 5.91091990e-01 -7.85974443e-01
-3.68783861e-01 -5.19304454e-01 3.09009433e-01 8.91551256e-01
8.33769679e-01 -1.17262363e-01 -1.17921746e+00 -1.84650883e-01
-5.29451251e-01 2.21556127e-01 3.29597175e-01 1.12352169e+00
-5.66248655e-01 -1.79817104e+00 -7.54434049e-01 6.27179444e-02
-9.54155847e-02 2.39795700e-01 -4.01044011e-01 8.92620683e-01
4.75418627e-01 7.50157952e-01 7.82755077e-01 8.18820298e-02
9.20304239e-01 3.94899130e-01 4.03796226e-01 -5.96079469e-01
-9.76791382e-01 -3.07084441e-01 4.91792828e-01 -5.29977679e-01
-5.32843411e-01 -5.90114653e-01 -1.28927612e+00 -5.51385507e-02
-3.71399373e-01 4.42904234e-01 1.07052851e+00 8.21267247e-01
5.92571318e-01 5.72856963e-01 -1.14457607e-01 -2.21743673e-01
1.49651036e-01 -6.80159926e-01 6.98041171e-02 9.76614892e-01
-3.46062005e-01 -6.89424217e-01 1.15978234e-01 2.79355496e-01]
|
[9.198094367980957, 8.027632713317871]
|
166b05ec-3905-43cd-adbb-548386f266b7
|
nuwa-visual-synthesis-pre-training-for-neural
|
2111.12417
| null |
https://arxiv.org/abs/2111.12417v1
|
https://arxiv.org/pdf/2111.12417v1.pdf
|
NÜWA: Visual Synthesis Pre-training for Neural visUal World creAtion
|
This paper presents a unified multimodal pre-trained model called N\"UWA that can generate new or manipulate existing visual data (i.e., images and videos) for various visual synthesis tasks. To cover language, image, and video at the same time for different scenarios, a 3D transformer encoder-decoder framework is designed, which can not only deal with videos as 3D data but also adapt to texts and images as 1D and 2D data, respectively. A 3D Nearby Attention (3DNA) mechanism is also proposed to consider the nature of the visual data and reduce the computational complexity. We evaluate N\"UWA on 8 downstream tasks. Compared to several strong baselines, N\"UWA achieves state-of-the-art results on text-to-image generation, text-to-video generation, video prediction, etc. Furthermore, it also shows surprisingly good zero-shot capabilities on text-guided image and video manipulation tasks. Project repo is https://github.com/microsoft/NUWA.
|
['Nan Duan', 'Daxin Jiang', 'Yuejian Fang', 'Fan Yang', 'Lei Ji', 'Jian Liang', 'Chenfei Wu']
|
2021-11-24
| null | null | null | null |
['video-prediction', 'text-to-video-generation']
|
['computer-vision', 'natural-language-processing']
|
[ 2.25329652e-01 5.43164611e-02 -2.73108035e-01 -9.96461138e-02
-8.27347755e-01 -4.79112208e-01 8.62951517e-01 -6.18830204e-01
2.68508047e-02 6.18339300e-01 5.32744288e-01 -1.69923052e-01
7.04270661e-01 -6.23076737e-01 -1.12162292e+00 -5.20782053e-01
4.86237019e-01 3.31141293e-01 7.13464692e-02 -1.28484428e-01
2.16378465e-01 2.22984985e-01 -1.68068278e+00 7.73407817e-01
5.64030647e-01 8.72016668e-01 8.78532648e-01 9.98606503e-01
-1.33956686e-01 7.35522151e-01 -4.65487272e-01 -4.38834369e-01
1.36230141e-01 -6.40026331e-01 -6.21397555e-01 5.00296652e-01
6.53929412e-01 -7.65404463e-01 -8.10418069e-01 7.91960478e-01
6.47423446e-01 2.74545789e-01 7.57480085e-01 -1.40889192e+00
-1.35297036e+00 6.13081813e-01 -7.12210000e-01 -1.92940384e-02
6.61685467e-01 6.63323998e-01 7.00630665e-01 -1.27284777e+00
1.02413201e+00 1.63772798e+00 2.67053656e-02 9.51318026e-01
-9.14583147e-01 -6.52536511e-01 2.84011751e-01 1.32389650e-01
-1.18084741e+00 -7.06842184e-01 5.44292271e-01 -4.76523548e-01
8.78068984e-01 1.91393033e-01 5.87271035e-01 1.69778669e+00
2.14072362e-01 1.34604311e+00 5.21609962e-01 -3.32379192e-01
-3.34434301e-01 -1.13122635e-01 -5.59535265e-01 6.40741110e-01
-2.00046062e-01 3.72241214e-02 -7.71777570e-01 4.12694633e-01
1.04367638e+00 -1.42389253e-01 -5.00864148e-01 -1.86429054e-01
-1.68882525e+00 5.72395802e-01 8.90626311e-02 -3.96820396e-04
-1.61441028e-01 3.87176037e-01 4.09334004e-01 1.81938738e-01
4.47379023e-01 2.86777407e-01 -2.12039456e-01 -9.06810760e-02
-7.17370033e-01 3.47364098e-01 2.60943055e-01 1.65034235e+00
5.05501449e-01 4.89499718e-01 -7.65072465e-01 6.22651994e-01
3.07758093e-01 9.43466902e-01 6.05490029e-01 -8.97928715e-01
1.08452988e+00 2.83500493e-01 2.60897189e-01 -5.69616377e-01
8.72106403e-02 5.00484742e-02 -7.93013990e-01 -7.76190162e-02
8.94384831e-02 -2.70882994e-01 -1.35683429e+00 1.58816862e+00
2.36550316e-01 1.77848190e-02 1.25880182e-01 9.51419771e-01
1.27876496e+00 1.30374312e+00 -1.76972762e-01 -1.47790045e-01
9.78118360e-01 -1.47023726e+00 -8.40371251e-01 -2.98435152e-01
4.34432566e-01 -1.01327920e+00 1.27057862e+00 7.94350505e-02
-1.65733325e+00 -9.04094040e-01 -6.33896351e-01 -6.49026453e-01
-2.43711531e-01 4.71423328e-01 1.79962680e-01 1.85756266e-01
-1.21020699e+00 5.19864447e-02 -5.93179941e-01 -4.33393627e-01
3.28360170e-01 -7.06430599e-02 -3.92257839e-01 -2.79354036e-01
-1.04570460e+00 7.57342815e-01 4.31311220e-01 1.58958174e-02
-1.38964021e+00 -4.08770233e-01 -1.05933523e+00 -2.02820953e-02
4.02476132e-01 -9.24474597e-01 1.28312516e+00 -9.58895624e-01
-1.49472356e+00 9.02423263e-01 -3.61108243e-01 -1.33019879e-01
6.10069692e-01 -2.20043153e-01 -1.73954487e-01 2.84916341e-01
1.67367995e-01 1.40252531e+00 1.18178868e+00 -1.40916264e+00
-5.72966039e-01 3.37119289e-02 6.32000118e-02 5.75725079e-01
-1.30208597e-01 -7.06124678e-02 -9.42594051e-01 -1.11192214e+00
-3.81333262e-01 -8.81533563e-01 1.06166422e-01 1.21667817e-01
-6.12071514e-01 -2.15937525e-01 1.03243887e+00 -8.03417087e-01
1.02082896e+00 -2.04390097e+00 6.88225925e-01 -4.07301277e-01
6.63210079e-02 2.46186078e-01 -5.65514803e-01 4.92281199e-01
-5.19993603e-02 2.49344185e-01 5.25809191e-02 -5.01661658e-01
4.59264964e-02 -1.91861615e-02 -4.51449007e-01 -5.38996942e-02
2.85980672e-01 1.31150842e+00 -7.88311601e-01 -4.72137243e-01
4.31121677e-01 3.88754994e-01 -6.46531701e-01 4.20430958e-01
-6.08214557e-01 5.09822965e-01 -3.85017812e-01 7.24557698e-01
4.54967350e-01 -4.17822570e-01 -8.62401351e-02 -3.98375213e-01
-1.87160801e-02 -9.08063650e-02 -6.94080949e-01 1.93116212e+00
-6.08971953e-01 9.21077490e-01 -8.96688849e-02 -7.18886375e-01
6.75419807e-01 5.60047626e-01 9.79120061e-02 -6.36912227e-01
2.59666979e-01 -6.03230409e-02 -3.39150816e-01 -8.41323972e-01
7.08180547e-01 3.36015671e-01 -5.31958304e-02 4.55449402e-01
2.62427211e-01 -3.08577836e-01 3.39928031e-01 3.18941087e-01
5.75879514e-01 7.08264530e-01 4.03374545e-02 2.31833532e-02
3.82558942e-01 -5.44504039e-02 9.21341404e-02 6.43126488e-01
1.25222281e-01 1.06191421e+00 3.66271198e-01 -2.25171924e-01
-1.40122664e+00 -1.05936396e+00 4.31229055e-01 1.10493279e+00
3.73445660e-01 -3.26197833e-01 -6.58486068e-01 -5.23903012e-01
-2.57353961e-01 8.83404791e-01 -5.38455665e-01 -2.26071745e-01
-4.71623480e-01 -2.59894758e-01 3.57020319e-01 5.12512624e-01
5.47166228e-01 -1.17681623e+00 -3.48860443e-01 5.92726506e-02
-7.40031958e-01 -1.32600713e+00 -1.18086004e+00 -4.72444296e-01
-5.57555199e-01 -7.56929457e-01 -1.12438929e+00 -9.99547005e-01
8.16337824e-01 6.93496227e-01 1.00733316e+00 -5.53227514e-02
-1.50838107e-01 4.80218589e-01 -6.82105243e-01 -1.65232077e-01
-5.66065431e-01 -1.78735435e-01 -1.27708003e-01 1.00520723e-01
-1.78267941e-01 -1.69301450e-01 -5.39406836e-01 2.97289371e-01
-1.15421712e+00 7.23744273e-01 5.97276330e-01 9.45945799e-01
4.69539821e-01 -3.98171216e-01 3.51821512e-01 -4.44583654e-01
5.36442757e-01 -3.57950121e-01 -3.64958733e-01 4.73960489e-01
7.62613770e-03 1.10940812e-02 5.80044687e-01 -6.95633888e-01
-1.28482997e+00 1.07381098e-01 1.30890533e-01 -9.80429888e-01
-6.00354560e-02 2.32693389e-01 -3.65813643e-01 3.72787207e-01
4.31230396e-01 5.41443527e-01 -2.31059548e-03 -2.72742569e-01
7.21006632e-01 8.01422775e-01 4.40453112e-01 -4.00811493e-01
8.29305530e-01 2.06709832e-01 -2.95983315e-01 -6.87695861e-01
-5.84177613e-01 4.06171903e-02 -5.38247049e-01 -3.51515383e-01
1.24001706e+00 -1.23049593e+00 -3.41113925e-01 7.06509948e-01
-1.44387877e+00 -6.99402809e-01 4.22418416e-02 3.63045663e-01
-7.34114766e-01 3.75051200e-01 -6.82412922e-01 -4.99448061e-01
-3.78719985e-01 -1.61749804e+00 1.56090975e+00 9.29693058e-02
8.17875285e-03 -8.24425519e-01 -5.13493896e-01 7.40727901e-01
1.48435190e-01 4.22894880e-02 7.54699171e-01 -1.89500786e-02
-1.06418669e+00 -1.08475331e-02 -4.14925903e-01 2.84562826e-01
5.65305427e-02 1.50296882e-01 -6.12673819e-01 -3.34697545e-01
-4.60998297e-01 -6.39314473e-01 9.01630938e-01 4.26636606e-01
1.20762384e+00 -4.17069495e-01 -3.06975454e-01 5.41255593e-01
9.77398813e-01 4.49992597e-01 8.02537978e-01 -1.71787933e-01
1.02547646e+00 4.94827211e-01 7.21348286e-01 4.21171606e-01
5.79744577e-01 8.99784327e-01 4.97674853e-01 -1.10302985e-01
-6.99390352e-01 -4.95001823e-01 7.94318020e-01 6.63634837e-01
-3.06651089e-02 -1.05948889e+00 -6.88891649e-01 5.46710908e-01
-1.90751302e+00 -1.10667682e+00 -1.13784090e-01 1.87354743e+00
6.85468614e-01 -1.49083212e-01 -1.28946066e-01 -4.86330360e-01
1.09615278e+00 3.21453124e-01 -5.69080830e-01 -1.21920131e-01
-1.73640534e-01 -1.21558852e-01 2.89105237e-01 4.61013854e-01
-9.46675599e-01 1.24945819e+00 5.79566669e+00 9.75332022e-01
-1.16794312e+00 9.66272950e-02 7.37158239e-01 -4.70508963e-01
-4.12419319e-01 -3.07826310e-01 -6.95562780e-01 6.60393834e-01
4.71946269e-01 -1.90240279e-01 5.08248568e-01 5.44178724e-01
4.17157084e-01 7.53537863e-02 -1.10437489e+00 1.24873817e+00
5.28314054e-01 -1.67849827e+00 7.54986346e-01 -4.18120585e-02
1.02166128e+00 -1.25633970e-01 3.72339517e-01 3.35169017e-01
2.25010350e-01 -8.85032415e-01 1.10312390e+00 4.40083981e-01
1.41316295e+00 -4.28430259e-01 4.40248847e-01 2.16725305e-01
-1.12813699e+00 7.72436410e-02 -1.69964999e-01 2.99175262e-01
5.32315910e-01 5.37384003e-02 -5.46394408e-01 5.36464214e-01
6.81260407e-01 9.57529724e-01 -4.83671546e-01 7.21540689e-01
-2.00525224e-01 2.03863367e-01 1.14913635e-01 3.28003913e-02
3.84940118e-01 4.47203293e-02 6.23200715e-01 1.01858604e+00
8.82928669e-01 1.28521100e-01 1.71254471e-01 7.86380887e-01
-3.44224989e-01 -7.12716877e-02 -8.67784441e-01 -2.05864534e-01
3.67821306e-01 8.25186491e-01 -4.17263389e-01 -6.09962225e-01
-5.12961030e-01 1.43373537e+00 -5.52881062e-02 6.10175133e-01
-1.14432597e+00 -3.75317156e-01 4.24768567e-01 1.05370857e-01
5.05280674e-01 -2.82207966e-01 2.91399539e-01 -1.52255869e+00
3.89248393e-02 -9.43747818e-01 -2.72052921e-02 -1.73125362e+00
-9.34386075e-01 6.27346516e-01 1.87786698e-01 -1.42123973e+00
-3.33422452e-01 -6.21957600e-01 -4.34966177e-01 6.40975058e-01
-1.17380869e+00 -1.44306958e+00 -4.30318087e-01 7.43384838e-01
1.24356079e+00 -4.74266410e-01 3.56301695e-01 1.76478535e-01
-4.76585120e-01 5.43005586e-01 6.78623393e-02 1.66510865e-01
9.33885336e-01 -7.51617730e-01 7.18549728e-01 9.94141042e-01
-1.92673784e-02 4.90164235e-02 4.60022777e-01 -7.50725269e-01
-1.67793119e+00 -1.42109978e+00 6.65970564e-01 -4.97685522e-01
4.28618103e-01 -5.50730467e-01 -5.25968373e-01 9.07840848e-01
7.22360611e-01 -2.77291328e-01 2.11505264e-01 -7.23326027e-01
-1.58587486e-01 1.84801966e-01 -8.25381339e-01 1.14865041e+00
1.41364455e+00 -4.27918077e-01 -2.69003779e-01 5.68024814e-01
1.21002865e+00 -9.25779223e-01 -4.98271704e-01 1.52455911e-01
3.63191724e-01 -8.15042436e-01 9.88778472e-01 -6.08716786e-01
1.17874455e+00 -2.16221899e-01 -2.63832033e-01 -1.25036287e+00
-2.47796804e-01 -8.92118216e-01 -2.19855800e-01 1.04092252e+00
4.30635571e-01 -1.73829645e-01 5.14276147e-01 2.55638272e-01
-5.11064291e-01 -5.08278251e-01 -6.96053028e-01 -5.51576734e-01
-1.40984267e-01 -2.57021517e-01 5.41335344e-01 7.63886571e-01
-1.93391711e-01 6.50996149e-01 -1.07415640e+00 -7.60864615e-02
3.53652120e-01 1.62117139e-01 1.01671863e+00 -3.68406117e-01
-1.11794576e-01 -4.46371526e-01 -1.71846017e-01 -1.60472941e+00
2.39774927e-01 -1.10058963e+00 -4.62612174e-02 -1.72486138e+00
2.18170896e-01 2.21287236e-01 3.54171753e-01 5.34079015e-01
-2.81857461e-01 4.40156579e-01 5.94030797e-01 2.01003820e-01
-4.89039779e-01 9.68787491e-01 2.06191349e+00 -3.13560486e-01
-1.63591698e-01 -2.56742746e-01 -5.48891842e-01 2.27269620e-01
7.26101935e-01 8.11460987e-02 -6.07977509e-01 -1.16118467e+00
-3.10937199e-03 5.84087908e-01 4.03656572e-01 -6.97198570e-01
-9.94129851e-02 -4.36999470e-01 4.67072099e-01 -7.92372942e-01
6.05665147e-01 -4.64833438e-01 1.35038450e-01 1.97239205e-01
-5.65809011e-01 2.94412732e-01 8.54060873e-02 5.83908200e-01
-1.98583201e-01 -7.17347413e-02 6.57131493e-01 -3.14475834e-01
-7.54845083e-01 6.09620750e-01 -5.18089294e-01 -6.86932653e-02
1.14656317e+00 -2.08328918e-01 -6.81795478e-01 -8.65518808e-01
-7.53321528e-01 5.00553727e-01 5.03524125e-01 9.11169708e-01
9.56373155e-01 -1.64958262e+00 -9.05002952e-01 2.85732094e-03
2.34606281e-01 2.90139820e-02 5.86016059e-01 4.48240519e-01
-6.17184222e-01 4.98965949e-01 -2.35130176e-01 -5.61094940e-01
-1.26036870e+00 7.53182292e-01 1.04140952e-01 3.77610847e-02
-5.06574929e-01 8.84990513e-01 5.91321170e-01 -2.19509915e-01
1.04268327e-01 -2.13308901e-01 7.32371360e-02 4.44519334e-03
4.90947276e-01 6.05129786e-02 -5.01888394e-01 -7.53978252e-01
1.69853345e-01 5.83174169e-01 -1.56833410e-01 -1.47734478e-01
8.99808407e-01 -4.96805161e-01 1.21675096e-01 2.31755942e-01
1.11137378e+00 -3.19357991e-01 -1.47081816e+00 -1.28102779e-01
-7.53467798e-01 -6.10161245e-01 -1.40824631e-01 -7.28530586e-01
-1.29637730e+00 1.11422706e+00 2.63783753e-01 -2.12489739e-01
1.05052459e+00 6.79421127e-02 1.02297544e+00 2.68900484e-01
7.85170421e-02 -9.93792832e-01 6.23649240e-01 4.74692166e-01
1.37604988e+00 -1.31619012e+00 -9.73087698e-02 -1.93520173e-01
-1.22810125e+00 1.07718480e+00 9.23300624e-01 2.46576637e-01
1.47742733e-01 -1.07549958e-01 -5.13461530e-02 2.70649582e-01
-1.14118648e+00 -2.60586828e-01 2.86045879e-01 7.71094799e-01
3.97437841e-01 -1.04249462e-01 4.89273332e-02 2.23401025e-01
-2.79368553e-02 3.67708388e-03 8.55907679e-01 6.22899890e-01
-1.78366750e-01 -9.89624798e-01 -2.92900145e-01 3.24563444e-01
-1.09375834e-01 -3.32818538e-01 -2.02021956e-01 8.51468563e-01
-4.12664609e-03 8.78403127e-01 1.49567753e-01 -4.26895738e-01
1.19721472e-01 4.25707437e-02 6.57407761e-01 -6.86799288e-01
-2.62342960e-01 3.87899280e-01 1.21456593e-01 -4.79696602e-01
-4.42165226e-01 -5.99709034e-01 -9.92117226e-01 -2.50479877e-01
-1.42538652e-01 -2.79582798e-01 4.24900800e-01 6.51062965e-01
5.77977836e-01 5.60714900e-01 5.96801639e-01 -1.35492361e+00
-1.63538411e-01 -9.44812238e-01 -1.49153247e-01 4.81730133e-01
3.14746708e-01 -4.72945184e-01 -9.19803679e-02 6.54494107e-01]
|
[10.851738929748535, -0.37665703892707825]
|
897b0ba0-9776-45e6-9bd5-d6444f91abc7
|
true-global-optimality-of-the-pressure-vessel
|
1403.7793
| null |
http://arxiv.org/abs/1403.7793v1
|
http://arxiv.org/pdf/1403.7793v1.pdf
|
True Global Optimality of the Pressure Vessel Design Problem: A Benchmark for Bio-Inspired Optimisation Algorithms
|
The pressure vessel design problem is a well-known design benchmark for
validating bio-inspired optimization algorithms. However, its global optimality
is not clear and there has been no mathematical proof put forward. In this
paper, a detailed mathematical analysis of this problem is provided that proves
that 6059.714335048436 is the global minimum. The Lagrange multiplier method is
also used as an alternative proof and this method is extended to find the
global optimum of a cantilever beam design problem.
|
['Xin-She Yang', 'Mehmet Karamanoglu', 'Nawaz Khan', 'Christian Huyck']
|
2014-03-30
| null | null | null | null |
['cantilever-beam']
|
['miscellaneous']
|
[ 3.38717327e-02 1.57980267e-02 -2.31109574e-01 -1.84411518e-02
-1.29480034e-01 -3.58011663e-01 -3.54181916e-01 -1.63452715e-01
-2.56732583e-01 1.19875026e+00 -2.60396868e-01 -3.67811561e-01
-6.41456425e-01 -5.93153059e-01 -5.26671588e-01 -1.08945966e+00
-1.89848080e-01 -8.48076791e-02 -1.27085134e-01 -4.41150278e-01
7.01334417e-01 4.83957499e-01 -1.10768342e+00 -5.32992303e-01
5.95328748e-01 1.08792818e+00 4.66874838e-01 4.73743051e-01
3.43435884e-01 -5.81988916e-02 -7.23604679e-01 -1.75626770e-01
2.78624386e-01 -6.06439531e-01 -7.87832916e-01 -2.86654681e-01
-1.30346924e-01 -1.20055526e-01 3.42066854e-01 7.76024580e-01
8.67613494e-01 4.11468208e-01 5.87004364e-01 -1.00204277e+00
-6.38966203e-01 3.15905392e-01 -6.04218304e-01 1.09845027e-01
2.31556490e-01 -1.69454455e-01 1.10020435e+00 -7.86843121e-01
3.91823024e-01 8.87251258e-01 5.13585627e-01 5.37731290e-01
-9.65717852e-01 -5.12659192e-01 -2.33793214e-01 8.69443044e-02
-1.18222284e+00 1.38151973e-01 1.07612848e+00 -1.80242583e-01
6.82741225e-01 5.30983925e-01 7.53466010e-01 3.13567460e-01
1.00184309e+00 1.98196933e-01 1.05946648e+00 -4.81668323e-01
5.70274651e-01 4.07802872e-03 6.93343356e-02 8.15058172e-01
6.07936144e-01 3.36684704e-01 -3.92188638e-01 -6.54438585e-02
8.41241896e-01 -2.91103899e-01 -4.67804253e-01 -1.47982329e-01
-5.24513960e-01 1.10778034e+00 6.11526012e-01 6.16582572e-01
-1.95343971e-01 2.70057529e-01 -1.19309146e-02 2.61051416e-01
7.26373047e-02 8.72267485e-01 -3.75923872e-01 1.47417650e-01
-7.28350103e-01 6.20436132e-01 1.03137445e+00 5.59937954e-01
3.03926200e-01 2.94961989e-01 3.41198981e-01 4.34703827e-01
6.56715572e-01 5.53130805e-01 7.95206800e-02 -1.33350313e+00
-1.76966414e-02 4.09958810e-01 2.32408121e-01 -1.54620552e+00
-6.27441883e-01 -6.70526624e-01 -7.18671739e-01 5.10525286e-01
4.27464843e-01 -5.32004595e-01 3.85736488e-02 1.30273855e+00
3.91009301e-01 -1.83793396e-01 -2.22979054e-01 1.27755904e+00
7.15185583e-01 6.83001041e-01 -3.22161436e-01 -6.16352856e-01
1.05568600e+00 -7.75932908e-01 -8.49823177e-01 3.48839611e-02
6.14291206e-02 -7.55458415e-01 6.06087804e-01 5.79014957e-01
-9.95308638e-01 -2.37845168e-01 -1.30139589e+00 3.54532331e-01
-2.42145538e-01 -2.94074327e-01 6.57057405e-01 9.45566356e-01
-5.97172379e-01 6.85648024e-01 -4.46365744e-01 -1.96439046e-02
2.69825719e-02 4.06869382e-01 -2.22056010e-03 1.42998815e-01
-8.63031983e-01 1.13937342e+00 2.51551270e-01 5.06548822e-01
-2.30831191e-01 -8.08539629e-01 -3.66911411e-01 -2.97100782e-01
2.63002634e-01 -6.79512978e-01 7.36007631e-01 -1.58312812e-01
-2.05257607e+00 5.49687922e-01 -2.33202115e-01 -2.18948051e-01
2.98739225e-01 9.27733406e-02 6.03847615e-02 1.45346537e-01
-2.84864396e-01 1.20115243e-01 7.11428761e-01 -1.13927829e+00
-2.26803526e-01 -3.07664067e-01 7.13753253e-02 -2.16855630e-01
-1.71653196e-01 2.01015428e-01 3.33232790e-01 -7.82242119e-01
2.57786334e-01 -6.82440281e-01 -6.13718510e-01 1.47139058e-01
-3.49214762e-01 -8.29970315e-02 5.74956059e-01 -7.17895985e-01
1.49023151e+00 -1.59295499e+00 3.34419340e-01 5.96386135e-01
-2.42593884e-01 -2.45473281e-01 3.04782957e-01 6.40421331e-01
2.06053555e-01 1.69336900e-01 -6.58657312e-01 1.55423149e-01
-9.71282646e-02 1.76960573e-01 -9.58281085e-02 6.64995313e-01
-2.39859018e-02 6.11321092e-01 -5.90517521e-01 -4.14865851e-01
1.37995988e-01 3.20192277e-01 -7.12331653e-01 -1.68037266e-01
3.21186274e-01 4.73708391e-01 -6.26222253e-01 8.99353027e-01
8.04864824e-01 2.36935228e-01 1.57426283e-01 -3.77984643e-01
-4.91767734e-01 -4.49972242e-01 -1.24558687e+00 1.45450449e+00
-4.60756451e-01 4.80972916e-01 3.93264145e-01 -1.32673848e+00
1.24029303e+00 2.78851420e-01 8.88911009e-01 -4.29421723e-01
7.59486616e-01 4.20506388e-01 4.65063602e-02 -7.36939549e-01
1.18276224e-01 -7.39388764e-01 -4.36468236e-02 2.61499047e-01
-4.26323861e-01 -5.28803527e-01 3.03541392e-01 -5.05495310e-01
9.42578375e-01 9.07778367e-02 2.38091454e-01 -8.56157124e-01
1.04694414e+00 1.66851893e-01 7.10870862e-01 2.80356497e-01
-1.77057922e-01 5.61080754e-01 1.48587093e-01 -2.33515143e-01
-8.51157069e-01 -7.61248827e-01 -6.41088605e-01 3.57056379e-01
5.63429892e-01 -1.21896736e-01 -5.07355213e-01 6.23185337e-02
4.32000458e-01 7.25362659e-01 -4.93953586e-01 -9.58024263e-02
-8.09535563e-01 -9.00172770e-01 9.52616427e-03 2.11140141e-01
1.44414976e-01 -8.05086255e-01 -1.18554544e+00 6.00335538e-01
-5.72141318e-04 -6.77481830e-01 2.65511107e-02 1.03324577e-02
-1.26929939e+00 -1.31652427e+00 -7.54421532e-01 -7.17397571e-01
6.41539812e-01 -1.66618645e-01 7.83390045e-01 4.97504950e-01
-6.84673607e-01 -3.83534171e-02 -1.74047679e-01 -5.19716680e-01
-3.62625420e-01 -2.93700963e-01 1.68600202e-01 -3.90660107e-01
-3.07680517e-01 -4.76180941e-01 -7.22660303e-01 7.44697571e-01
-2.95415461e-01 -5.52598119e-01 3.45776498e-01 8.37483943e-01
6.20642006e-01 5.53122699e-01 1.06802082e+00 -5.84614910e-02
7.18556285e-01 -1.32163316e-01 -8.43253911e-01 -1.02228031e-01
-7.13396192e-01 -1.09492503e-02 7.99636841e-01 -1.03346802e-01
-7.72005200e-01 -3.53123322e-02 -2.75320768e-01 7.96493217e-02
3.36937606e-01 7.26789057e-01 -1.05495073e-01 -5.67464888e-01
3.22927594e-01 -1.22428097e-01 2.55716681e-01 -7.95853138e-01
3.43264975e-02 3.10732931e-01 3.81169409e-01 -6.64539397e-01
7.44814813e-01 3.92399967e-01 7.41206408e-01 -1.04411268e+00
-6.26709819e-01 -3.01937073e-01 -2.48835757e-01 -5.19555151e-01
7.35378563e-01 1.83861837e-01 -1.32043803e+00 3.19985338e-02
-9.64980245e-01 7.16982633e-02 7.00409561e-02 3.77162218e-01
-7.40800500e-01 3.37250143e-01 -1.53872386e-01 -1.23477519e+00
-6.06613755e-01 -8.46821666e-01 3.72448087e-01 2.30474800e-01
-8.80361646e-02 -1.01948702e+00 5.32218926e-02 3.05790395e-01
5.53190529e-01 1.10630667e+00 6.54878259e-01 2.31934533e-01
-2.22923517e-01 -4.40616995e-01 3.08464468e-01 1.09793097e-01
6.51628897e-02 4.78581786e-01 -4.33010191e-01 -4.34014350e-01
6.09962523e-01 1.93051592e-01 2.90057093e-01 8.12499464e-01
8.32025647e-01 -3.17991555e-01 -3.15953344e-01 4.07650232e-01
2.02266455e+00 4.59367394e-01 3.94771039e-01 6.58597529e-01
-2.07401044e-03 6.65142596e-01 8.99745166e-01 5.55992424e-01
-1.87861040e-01 6.04157925e-01 7.50338554e-01 2.75064260e-01
4.43273067e-01 4.91343349e-01 1.56995624e-01 4.97800797e-01
-3.88998300e-01 -9.85256284e-02 -6.41170681e-01 9.18303803e-02
-1.45957446e+00 -8.84327650e-01 -4.94769186e-01 1.97057605e+00
6.69632137e-01 6.96928054e-03 7.60027617e-02 6.97826445e-01
6.13054574e-01 -2.52042264e-01 -2.36963198e-01 -8.30748916e-01
-2.32504129e-01 4.63282764e-01 5.66339374e-01 5.17502010e-01
-7.73694575e-01 1.23618096e-01 7.34673023e+00 4.40132171e-01
-1.13872480e+00 -2.91793402e-02 1.94237769e-01 1.01654969e-01
1.24114484e-01 4.85056080e-02 -5.98501325e-01 6.59318328e-01
5.51915824e-01 -5.31048954e-01 2.43045494e-01 6.92369163e-01
6.21988118e-01 -4.49710339e-01 -7.34428227e-01 8.25566053e-01
-2.48888329e-01 -1.25939751e+00 -5.07021606e-01 1.98320538e-01
6.47648931e-01 -8.46809626e-01 -9.79620814e-02 -4.53535199e-01
-5.45052350e-01 -8.88646960e-01 7.15236604e-01 6.38451040e-01
5.42342924e-02 -9.52471375e-01 9.12753224e-01 3.66817772e-01
-9.12131667e-01 -4.11005437e-01 -5.85405707e-01 -2.84823269e-01
7.43679106e-01 7.39672780e-01 -5.05311668e-01 8.16675603e-01
6.89385712e-01 2.74568915e-01 -3.16555388e-02 1.55985618e+00
-2.03239948e-01 5.37055969e-01 -6.24241054e-01 -5.09979069e-01
-3.49987075e-02 -6.53744817e-01 9.08833623e-01 7.21119940e-01
4.12999630e-01 2.25713134e-01 -1.69791862e-01 1.09292221e+00
4.15564179e-01 5.11624515e-01 -1.02802783e-01 1.85607597e-01
2.34876007e-01 1.01692319e+00 -6.71983242e-01 5.02144873e-01
-6.02256954e-02 2.94504076e-01 -4.25170302e-01 9.30664316e-02
-1.05908823e+00 -5.48797429e-01 6.07710481e-01 4.87868600e-02
1.13793783e-01 -4.99267876e-01 -9.69290376e-01 -5.55212736e-01
-3.12655866e-01 -1.24395661e-01 2.12410688e-01 -5.03805280e-01
-1.15711510e+00 -7.35158175e-02 7.25778639e-02 -1.06880498e+00
1.59879252e-01 -1.17614937e+00 -8.02685320e-01 9.10032570e-01
-1.26187921e+00 -4.23238218e-01 -2.24380881e-01 3.84995304e-02
1.83533654e-01 -1.57854185e-01 5.51229358e-01 2.46528938e-01
-8.96984637e-01 5.02838314e-01 2.56711423e-01 -4.85070646e-01
1.14724122e-01 -1.00400460e+00 -5.14554799e-01 6.97027147e-01
-8.50299418e-01 6.61183894e-01 1.41193938e+00 -4.00935501e-01
-2.03504753e+00 -3.20427835e-01 6.58579588e-01 8.35299417e-02
7.00124800e-01 3.48702967e-01 -7.36530066e-01 -1.32782444e-01
6.92156181e-02 -4.61781591e-01 4.97961789e-01 -5.18854201e-01
7.40763783e-01 -1.63312137e-01 -1.37522244e+00 4.13139194e-01
7.55984843e-01 3.94511402e-01 -4.95870084e-01 1.64570287e-01
2.96202481e-01 -4.64387894e-01 -1.36924386e+00 6.11677885e-01
6.41060054e-01 -7.13785589e-01 1.25450158e+00 -1.19930714e-01
4.92176443e-01 -3.44872892e-01 -2.30563298e-01 -5.88219762e-01
-1.89197868e-01 -6.99817717e-01 -2.34239563e-01 9.98090625e-01
4.02219504e-01 -9.71755445e-01 9.97213483e-01 4.19584066e-01
-1.59692779e-01 -1.53782129e+00 -1.26831329e+00 -1.21384633e+00
3.07474643e-01 -2.09436297e-01 3.77314329e-01 6.21323347e-01
4.16128375e-02 1.27509609e-01 -2.62328476e-01 -1.21792741e-02
8.65346014e-01 4.65997428e-01 4.35109466e-01 -1.06845212e+00
-5.99253774e-02 -7.31487572e-01 -2.36672103e-01 -7.75257528e-01
-8.99021178e-02 -4.40160394e-01 -3.77364494e-02 -1.52017379e+00
-4.64405209e-01 -5.73401809e-01 -3.45896393e-01 -6.98849335e-02
1.79459438e-01 1.90530017e-01 -4.18641195e-02 -7.59880617e-02
4.97362942e-01 6.16574109e-01 1.82090461e+00 -4.70965430e-02
-1.25921443e-01 2.56994605e-01 -6.79888368e-01 4.02269721e-01
1.11578953e+00 -4.71154630e-01 -1.61294907e-01 1.69334054e-01
2.21313044e-01 2.50800163e-01 2.90365249e-01 -8.43192518e-01
-6.41008615e-02 -6.26750767e-01 7.52748474e-02 -6.80592299e-01
3.22587132e-01 -1.22601235e+00 2.97162622e-01 8.99087846e-01
1.55710235e-01 8.93817283e-03 -7.14346813e-03 5.06813526e-01
6.35671914e-02 -9.95691121e-01 9.93216693e-01 1.73395336e-01
-3.29713613e-01 -1.48926839e-01 -3.05653661e-01 -1.03094630e-01
1.09979737e+00 -7.32685864e-01 6.55318275e-02 2.87198927e-02
-6.11239910e-01 3.08067560e-01 4.02249008e-01 3.16779613e-02
7.20543921e-01 -1.18241692e+00 -5.17262042e-01 -2.57885724e-01
-6.35650873e-01 -2.39976227e-01 2.04330077e-03 1.01643062e+00
-1.07333100e+00 4.48134571e-01 -2.48656720e-01 -5.57578206e-01
-1.16690934e+00 4.56725299e-01 5.83539784e-01 2.58121580e-01
-4.92675781e-01 7.94263899e-01 -5.75864077e-01 4.22769263e-02
-1.02925271e-01 -3.75633538e-01 -1.67826638e-01 -5.76573312e-02
1.48513019e-01 9.08083260e-01 -1.02624521e-01 -4.86697525e-01
-6.53479755e-01 1.04814637e+00 1.05619347e+00 3.18202078e-02
1.55321264e+00 -1.63553536e-01 -5.64771235e-01 7.26273432e-02
1.33459759e+00 -1.31692931e-01 -6.51486278e-01 2.62083083e-01
-5.49347773e-02 -5.81852198e-01 1.43354461e-01 -4.27304447e-01
-1.17645252e+00 3.89105201e-01 4.67191488e-01 2.38320678e-01
1.26930475e+00 -4.79119390e-01 8.29307079e-01 4.67269093e-01
4.55062628e-01 -1.34907722e+00 3.92850526e-02 4.53867674e-01
1.46735287e+00 -1.08637953e+00 5.34711361e-01 -7.74315834e-01
1.65054634e-01 1.45323777e+00 4.79274273e-01 -4.17166919e-01
9.57678914e-01 2.74463415e-01 -3.20763171e-01 -1.15755782e-01
-1.77425817e-01 3.12061638e-01 2.66933352e-01 1.72201872e-01
4.93704349e-01 -2.48220712e-01 -1.31122506e+00 7.72761762e-01
-6.58702254e-01 9.26056877e-02 3.23725998e-01 1.56152487e+00
-6.74927771e-01 -9.46353078e-01 -8.11335683e-01 2.13292185e-02
-6.31271303e-01 3.34699035e-01 -6.61041886e-02 7.51755595e-01
9.05534476e-02 1.19216490e+00 -4.26636666e-01 -1.83742419e-02
4.47247654e-01 -1.63565367e-01 6.48122370e-01 -1.17293455e-01
-5.01462460e-01 -9.89867449e-02 -2.46568725e-01 -3.22803468e-01
-6.56956196e-01 -2.10199669e-01 -1.57935882e+00 -6.00300252e-01
-8.99612188e-01 4.57803816e-01 1.16077161e+00 7.72820473e-01
-1.74577571e-02 4.62047875e-01 7.74602294e-01 -8.12897325e-01
-4.54260021e-01 -3.31237257e-01 -4.95220661e-01 -2.74046332e-01
6.40965104e-02 -1.06275725e+00 -5.53948879e-01 -3.80327180e-02]
|
[5.8980021476745605, 3.4167845249176025]
|
9bb19c0c-5a1c-4e3b-b8d4-15043094357a
|
learning-from-small-data-through-sampling-an
|
2003.14297
| null |
https://arxiv.org/abs/2003.14297v5
|
https://arxiv.org/pdf/2003.14297v5.pdf
|
Generative Latent Implicit Conditional Optimization when Learning from Small Sample
|
We revisit the long-standing problem of learning from a small sample, to which end we propose a novel method called GLICO (Generative Latent Implicit Conditional Optimization). GLICO learns a mapping from the training examples to a latent space and a generator that generates images from vectors in the latent space. Unlike most recent works, which rely on access to large amounts of unlabeled data, GLICO does not require access to any additional data other than the small set of labeled points. In fact, GLICO learns to synthesize completely new samples for every class using as little as 5 or 10 examples per class, with as few as 10 such classes without imposing any prior. GLICO is then used to augment the small training set while training a classifier on the small sample. To this end, our proposed method samples the learned latent space using spherical interpolation, and generates new examples using the trained generator. Empirical results show that the new sampled set is diverse enough, leading to improvement in image classification in comparison with the state of the art, when trained on small samples obtained from CIFAR-10, CIFAR-100, and CUB-200.
|
['Idan Azuri', 'Daphna Weinshall']
|
2020-03-31
| null | null | null | null |
['small-data']
|
['computer-vision']
|
[ 3.20251077e-01 4.65463847e-01 -3.33419859e-01 -2.86424130e-01
-9.00577962e-01 -5.60925305e-01 7.74203539e-01 -1.74966797e-01
-3.58737558e-01 1.00540352e+00 2.05105934e-02 5.50682545e-02
3.00962448e-01 -8.48578095e-01 -8.76812577e-01 -9.64817226e-01
2.54705220e-01 7.02689588e-01 -8.51129293e-02 3.40887904e-01
7.15387315e-02 2.96304673e-01 -1.48215389e+00 1.94070041e-01
1.13420546e+00 9.47444677e-01 1.86245188e-01 4.13923085e-01
5.82548603e-02 5.72450578e-01 -6.48923755e-01 -9.07794908e-02
2.67199486e-01 -6.10811591e-01 -4.80926841e-01 6.11778796e-01
4.35135692e-01 -3.08149308e-01 9.95310247e-02 8.17750633e-01
1.24106422e-01 1.99926943e-01 1.09125638e+00 -1.11946261e+00
-9.27821219e-01 5.75373590e-01 -1.58857152e-01 -3.63429695e-01
-7.37185255e-02 3.62224947e-03 7.76046932e-01 -1.40331483e+00
6.02568507e-01 9.34808493e-01 4.96086657e-01 7.33784735e-01
-1.62200177e+00 -6.78743601e-01 1.30574763e-01 -2.67464221e-01
-1.44378352e+00 -4.69969690e-01 8.58677089e-01 -7.44548321e-01
3.39999557e-01 -6.80946484e-02 5.68280876e-01 1.17322326e+00
-4.42597240e-01 9.17600513e-01 1.09914720e+00 -7.59488404e-01
6.82641268e-01 5.30607343e-01 1.36524394e-01 5.64173996e-01
1.47284284e-01 -5.08487187e-02 -2.36517519e-01 -3.37467223e-01
8.77354264e-01 1.69593766e-01 -3.13619971e-01 -7.05022335e-01
-1.22125471e+00 1.11197031e+00 5.38177073e-01 7.52722844e-02
-4.90252793e-01 1.58994608e-02 -2.69460492e-02 1.23943247e-01
7.37133682e-01 3.37327778e-01 -1.46328598e-01 3.31732392e-01
-1.22557259e+00 1.73061833e-01 6.84285343e-01 1.02788341e+00
1.13648403e+00 3.82102728e-01 -2.34520763e-01 8.03750157e-01
2.87884682e-01 5.36462069e-01 7.03346670e-01 -7.81847835e-01
6.25211477e-01 5.24277091e-01 2.17535943e-01 -5.04137754e-01
4.17518705e-01 -7.98363388e-01 -7.74972320e-01 3.67841423e-01
4.56025362e-01 -1.05051994e-01 -1.18796396e+00 1.85705519e+00
7.76668414e-02 4.34134632e-01 1.49778396e-01 4.78888273e-01
3.17939371e-01 8.98583949e-01 -2.12834626e-01 -3.13119054e-01
7.84071624e-01 -1.26197302e+00 -3.84417415e-01 -3.24141979e-01
3.60697657e-01 -6.05539083e-01 1.28566110e+00 5.10737956e-01
-9.41927016e-01 -8.06627512e-01 -1.08643961e+00 1.39342561e-01
-2.71566391e-01 6.65133476e-01 4.83855754e-01 6.40584350e-01
-8.08116853e-01 3.76511663e-01 -8.53313088e-01 -2.54647769e-02
5.84369183e-01 2.06538960e-01 -2.53568351e-01 -2.67676383e-01
-6.90314531e-01 4.28573728e-01 4.36437368e-01 -1.39778936e-02
-1.29682148e+00 -5.74505329e-01 -9.04542327e-01 1.99890025e-02
2.99445510e-01 -5.13915896e-01 1.08096826e+00 -1.31082904e+00
-1.66877961e+00 5.88448644e-01 -3.28009456e-01 -5.39758861e-01
5.83675623e-01 -2.53633767e-01 6.78329021e-02 3.70169692e-02
2.11495951e-01 9.98257756e-01 1.21272278e+00 -1.48460138e+00
-2.61997432e-01 -1.13008782e-01 6.25199173e-03 2.69006211e-02
-3.97136450e-01 -5.34633696e-01 -3.56445312e-01 -8.69365513e-01
2.25558564e-01 -1.04013729e+00 -3.28164697e-01 -1.08930185e-01
-4.79268372e-01 -3.17378759e-01 7.19812632e-01 -2.39853606e-01
7.21796393e-01 -2.11129570e+00 3.38284135e-01 1.82351634e-01
2.42105514e-01 3.23292315e-01 -2.65033185e-01 3.18693638e-01
-2.12053165e-01 1.88508406e-02 -5.73337495e-01 -8.07453871e-01
-1.11661226e-01 2.45304704e-01 -6.31008089e-01 3.08456689e-01
6.26269355e-02 8.35041583e-01 -1.01151216e+00 -1.35051042e-01
1.68381110e-01 3.86178881e-01 -7.25899220e-01 3.04353356e-01
-5.01940250e-01 5.65692663e-01 -3.47806633e-01 2.87037164e-01
6.10770166e-01 -5.70612252e-01 -9.98078585e-02 7.79571831e-02
1.98913619e-01 1.71749160e-01 -1.04176760e+00 1.59619439e+00
-5.86188734e-01 5.47063470e-01 -5.07297039e-01 -1.26012027e+00
1.07513607e+00 5.79249322e-01 1.03531785e-01 1.26673782e-03
-1.04852811e-01 3.32362324e-01 -3.14947039e-01 -1.89562216e-01
-3.01695317e-02 -1.50718138e-01 1.22353889e-01 6.03030384e-01
3.58619243e-01 -7.94104189e-02 3.42247754e-01 1.79731429e-01
4.85410303e-01 2.10104108e-01 2.70431131e-01 8.97372738e-02
5.03439426e-01 -1.43844977e-01 5.26090026e-01 6.49358332e-01
4.10283685e-01 7.88441956e-01 2.98235774e-01 -2.73056746e-01
-1.24485421e+00 -1.29745400e+00 -1.64175540e-01 6.48669422e-01
-3.88803124e-01 -2.86350906e-01 -9.70241666e-01 -9.54238713e-01
-2.59272635e-01 8.24840724e-01 -7.89313078e-01 1.39277712e-01
-3.44522566e-01 -5.63231230e-01 1.81527719e-01 5.03931403e-01
4.08195972e-01 -1.24159265e+00 -1.73747420e-01 1.23068929e-01
-1.69725984e-01 -1.01176500e+00 -5.85823715e-01 1.18874602e-01
-1.17630804e+00 -8.30332160e-01 -1.16109610e+00 -9.26482618e-01
1.40508819e+00 -1.18169725e-01 1.00202298e+00 -3.84771109e-01
-1.07295722e-01 1.17517993e-01 -3.01442802e-01 -2.33447626e-01
-5.03894925e-01 2.35167786e-01 -4.16605733e-02 4.00175810e-01
1.89376161e-01 -7.11079061e-01 -3.50820184e-01 1.62326455e-01
-8.33687603e-01 3.50717217e-01 7.63672233e-01 1.12483263e+00
7.70720541e-01 -1.52503937e-01 7.98908055e-01 -1.22896779e+00
3.01300853e-01 -5.75184047e-01 -9.04806077e-01 1.03980526e-01
-6.73570752e-01 3.01387459e-01 9.79366243e-01 -8.38730514e-01
-9.20295238e-01 3.82686496e-01 2.11428881e-01 -6.81091428e-01
-3.57109547e-01 6.41801417e-01 -9.04338900e-03 2.94719577e-01
9.52275574e-01 4.25727248e-01 -8.61156434e-02 -7.03807890e-01
6.03111148e-01 5.50784647e-01 5.76679409e-01 -5.00556052e-01
1.10171127e+00 4.32195485e-01 -2.64543176e-01 -6.45632565e-01
-1.02667713e+00 -2.30593562e-01 -8.91840875e-01 2.50085026e-01
7.98839808e-01 -9.00886178e-01 -1.21149477e-02 3.72601420e-01
-9.35047209e-01 -5.46107888e-01 -7.48144448e-01 7.22255290e-01
-7.15142250e-01 1.56803541e-02 -4.43948925e-01 -7.65668213e-01
-1.67867422e-01 -1.20332801e+00 9.10070896e-01 -3.76671180e-02
-1.26643375e-01 -1.03795612e+00 2.32632533e-01 2.77405202e-01
1.38668358e-01 3.14250171e-01 7.24824309e-01 -5.37733972e-01
-9.96959209e-01 -4.73506123e-01 1.83987021e-01 8.42596114e-01
4.47351813e-01 -2.89470643e-01 -1.03535008e+00 -3.92607987e-01
5.04425727e-02 -7.42548645e-01 8.88332427e-01 3.12945992e-01
1.33114934e+00 -4.81389493e-01 -3.16684395e-01 6.74194574e-01
1.34649706e+00 1.71991318e-01 5.01552105e-01 -1.33395374e-01
5.22706687e-01 5.07145941e-01 3.80569786e-01 2.33846843e-01
6.05678409e-02 6.35630548e-01 1.68965325e-01 -2.18711048e-01
-7.03189895e-02 -5.24231911e-01 3.99668038e-01 9.09430504e-01
-1.06338479e-01 -7.88700506e-02 -6.96633458e-01 6.59365296e-01
-1.67288005e+00 -8.99152815e-01 3.05076838e-01 2.43886495e+00
1.02155817e+00 4.31499183e-02 -1.39740139e-01 2.86775678e-01
5.82314432e-01 -3.73378098e-02 -6.11163616e-01 2.84662873e-01
1.68125391e-01 4.10331070e-01 2.96564192e-01 7.62533665e-01
-9.57093775e-01 9.43891943e-01 6.44457293e+00 8.92743707e-01
-1.39714634e+00 1.80295721e-01 8.48654985e-01 -2.13348549e-02
-3.82980436e-01 1.17806494e-01 -8.68338346e-01 4.78725255e-01
7.51465797e-01 -2.16909319e-01 5.01979411e-01 1.13694286e+00
8.52415189e-02 1.35687649e-01 -1.40825689e+00 9.60358202e-01
3.71957600e-01 -1.46779931e+00 1.55608460e-01 1.63853645e-01
1.22381687e+00 -1.06626146e-01 3.01521689e-01 3.08315188e-01
3.19521070e-01 -1.21237338e+00 7.77274549e-01 5.43487728e-01
9.59843457e-01 -6.29784524e-01 5.15399575e-01 7.82065988e-01
-8.30381453e-01 1.88624427e-01 -5.04649103e-01 2.09998637e-01
1.31761227e-02 7.49785185e-01 -1.06300116e+00 2.93392539e-01
2.17492327e-01 7.34062195e-01 -5.83904386e-01 9.43687439e-01
-5.89192629e-01 1.13476086e+00 -2.82249928e-01 2.93602318e-01
2.23700881e-01 -2.41566971e-01 2.67750025e-01 6.75951660e-01
5.34362793e-01 -1.19942620e-01 3.96319598e-01 1.17559505e+00
-2.29231730e-01 3.94441858e-02 -6.10180438e-01 -9.31795537e-02
3.59419256e-01 1.19556999e+00 -6.58743143e-01 -8.54646504e-01
-2.42190987e-01 1.06906378e+00 3.45372498e-01 8.52828622e-01
-6.51895881e-01 -2.52237886e-01 -1.10485777e-01 1.64581746e-01
4.62947190e-01 -1.81594938e-01 -8.00104141e-02 -1.50943732e+00
8.68037865e-02 -6.96456492e-01 2.02177688e-02 -8.14894497e-01
-1.35885060e+00 9.20181334e-01 1.28429562e-01 -1.46211326e+00
-7.78854489e-01 -4.46855307e-01 -5.89916587e-01 1.16415358e+00
-1.37854469e+00 -1.15748680e+00 -3.86220843e-01 4.93957818e-01
6.83374107e-01 -5.80658734e-01 1.00766289e+00 1.49820685e-01
-1.43304154e-01 6.32993758e-01 2.28055418e-01 1.78264558e-01
6.46268547e-01 -1.27918887e+00 3.14501643e-01 6.70934200e-01
4.35410649e-01 7.91047454e-01 4.45973605e-01 -4.28663343e-01
-5.60977221e-01 -1.16288733e+00 8.24679852e-01 -4.91036445e-01
1.49649680e-01 -8.71027112e-01 -8.23166549e-01 1.07390893e+00
9.81487483e-02 6.19483948e-01 6.71416521e-01 -7.97063857e-02
-3.89259130e-01 -7.72201568e-02 -1.03782451e+00 4.52029318e-01
6.30807638e-01 -4.51778054e-01 -5.40938318e-01 4.17287081e-01
6.10015273e-01 -2.10316509e-01 -5.35052836e-01 2.37416208e-01
2.65774697e-01 -7.02748358e-01 9.47656274e-01 -3.10707062e-01
4.30594474e-01 -4.15768266e-01 -1.14956439e-01 -1.50279987e+00
-1.53112695e-01 -3.27938318e-01 -1.64848655e-01 1.23244083e+00
5.40979743e-01 -7.52385318e-01 1.29521072e+00 2.27543712e-01
-2.04492018e-01 -8.30520689e-01 -5.50851226e-01 -9.18441355e-01
2.83304781e-01 -2.62815356e-01 5.62032223e-01 8.90480518e-01
-4.58011627e-01 4.59705859e-01 -4.23167437e-01 9.11304653e-02
9.44217861e-01 2.93102831e-01 9.76063371e-01 -1.24712336e+00
-6.47358060e-01 7.68354014e-02 -1.05410799e-01 -1.37081122e+00
2.03561112e-01 -1.11943007e+00 8.04679915e-02 -1.40124154e+00
1.83430523e-01 -8.92801404e-01 -1.88427299e-01 6.76950216e-01
-1.84033066e-01 5.92006862e-01 7.98256025e-02 5.31560540e-01
-2.94927627e-01 9.15595591e-01 1.33448565e+00 -2.08853215e-01
-1.92914858e-01 2.69107431e-01 -5.12499332e-01 7.67383337e-01
5.49700975e-01 -5.86096168e-01 -9.07280028e-01 -2.45223477e-01
-7.44286403e-02 -3.63946222e-02 3.35093856e-01 -1.14581537e+00
-4.09377292e-02 -2.08124757e-01 5.57921290e-01 -6.64100230e-01
5.15002847e-01 -6.29323065e-01 5.45040295e-02 2.44319424e-01
-5.95201612e-01 -5.17980635e-01 -3.21395665e-01 5.09968460e-01
-4.33806032e-01 -6.13123417e-01 8.39363873e-01 -1.52122065e-01
-2.46381894e-01 4.63927239e-01 -2.20253587e-01 -2.30558366e-02
1.15024960e+00 -1.39349967e-01 -4.83386265e-03 -3.77583593e-01
-1.10535216e+00 -8.78359824e-02 5.15413761e-01 2.08546773e-01
5.14832735e-01 -1.62503886e+00 -5.96769154e-01 5.67714691e-01
2.24315494e-01 3.92463863e-01 -1.74761742e-01 3.31626147e-01
-2.74080426e-01 5.13075769e-01 -2.17938665e-02 -7.57267356e-01
-7.72547662e-01 6.44695103e-01 1.26834497e-01 -2.98957765e-01
-4.87102896e-01 7.17230797e-01 5.75679719e-01 -5.93242168e-01
1.84680581e-01 -3.82929295e-01 -1.75134510e-01 -3.35669331e-02
3.72162700e-01 4.53517772e-02 -2.45611176e-01 -3.83883893e-01
1.50204137e-01 3.83945525e-01 -1.32738203e-01 -5.11352181e-01
1.37237763e+00 1.90853924e-01 8.79837647e-02 7.67178297e-01
1.27436149e+00 3.34146261e-01 -1.65969205e+00 -3.90968859e-01
-2.50598520e-01 -5.61977506e-01 -3.11251879e-01 -4.83506322e-01
-1.04187536e+00 1.03552449e+00 5.13852894e-01 2.42592581e-02
8.99072051e-01 3.31167206e-02 4.85264480e-01 4.64173079e-01
4.12038714e-01 -6.70852423e-01 5.76651812e-01 3.10971379e-01
9.81772602e-01 -1.10583556e+00 -6.24429472e-02 -3.51454020e-01
-5.21914959e-01 9.04995501e-01 4.87567693e-01 -5.38724065e-01
6.62517726e-01 -1.00175828e-01 8.58925804e-02 1.85850784e-01
-7.66325831e-01 6.86050579e-02 3.42448682e-01 5.76280713e-01
4.60166574e-01 2.00186484e-02 -1.86034590e-02 2.64094263e-01
-3.36674988e-01 2.72986710e-01 3.44862461e-01 6.98268950e-01
-2.79724032e-01 -1.50013340e+00 -3.61461759e-01 4.14153785e-01
-2.21638545e-01 -1.43266499e-01 9.80866551e-02 5.67052662e-01
3.56122196e-01 5.15671432e-01 1.08425267e-01 1.35832489e-01
-2.44403303e-01 2.54571646e-01 3.35450232e-01 -1.18009579e+00
5.58129102e-02 2.26024806e-01 -3.38833123e-01 -9.27983150e-02
-3.91148627e-01 -7.34644532e-01 -1.04681695e+00 4.17282164e-01
-3.76059145e-01 5.17183483e-01 6.44438267e-01 9.38314855e-01
6.26792386e-02 2.64530122e-01 7.75763392e-01 -1.08353388e+00
-5.26451766e-01 -1.15740097e+00 -3.74645323e-01 4.00285155e-01
4.72536743e-01 -6.05713546e-01 -5.03014147e-01 5.96211910e-01]
|
[11.475314140319824, -0.0660107210278511]
|
7ae8e101-f2d8-4b75-9196-45136b514f66
|
hfn-heterogeneous-feature-network-for
|
2211.00277
| null |
https://arxiv.org/abs/2211.00277v2
|
https://arxiv.org/pdf/2211.00277v2.pdf
|
HFN: Heterogeneous Feature Network for Multivariate Time Series Anomaly Detection
|
Network or physical attacks on industrial equipment or computer systems may cause massive losses. Therefore, a quick and accurate anomaly detection (AD) based on monitoring data, especially the multivariate time-series (MTS) data, is of great significance. As the key step of anomaly detection for MTS data, learning the relations among different variables has been explored by many approaches. However, most of the existing approaches do not consider the heterogeneity between variables, that is, different types of variables (continuous numerical variables, discrete categorical variables or hybrid variables) may have different and distinctive edge distributions. In this paper, we propose a novel semi-supervised anomaly detection framework based on a heterogeneous feature network (HFN) for MTS, learning heterogeneous structure information from a mass of unlabeled time-series data to improve the accuracy of anomaly detection, and using attention coefficient to provide an explanation for the detected anomalies. Specifically, we first combine the embedding similarity subgraph generated by sensor embedding and feature value similarity subgraph generated by sensor values to construct a time-series heterogeneous graph, which fully utilizes the rich heterogeneous mutual information among variables. Then, a prediction model containing nodes and channel attentions is jointly optimized to obtain better time-series representations. This approach fuses the state-of-the-art technologies of heterogeneous graph structure learning (HGSL) and representation learning. The experiments on four sensor datasets from real-world applications demonstrate that our approach detects the anomalies more accurately than those baseline approaches, thus providing a basis for the rapid positioning of anomalies.
|
['Xiandong Ma', 'Qiucheng Miao', 'Canqun Yang', 'Chengkun Wu', 'Jun Zhan']
|
2022-11-01
| null | null | null | null |
['supervised-anomaly-detection', 'semi-supervised-anomaly-detection', 'graph-structure-learning']
|
['computer-vision', 'computer-vision', 'graphs']
|
[ 1.56567529e-01 -1.46824941e-01 -4.22939323e-02 -1.31646439e-01
-2.46668532e-01 -1.64085105e-01 2.04497427e-01 7.30736911e-01
3.45727772e-01 3.78800690e-01 -2.94872373e-02 -3.74895155e-01
-4.61920947e-01 -9.96680439e-01 -5.46393692e-01 -8.69875073e-01
-5.42294562e-01 1.16636202e-01 1.33823186e-01 -1.94101378e-01
1.63997397e-01 4.96193856e-01 -1.27272832e+00 -2.22469404e-01
1.01851356e+00 1.49091768e+00 -2.53165841e-01 2.58060396e-01
-5.20267665e-01 7.61890054e-01 -7.31350303e-01 1.23056928e-02
6.50672168e-02 -4.16186959e-01 -2.73468822e-01 1.00312136e-01
-1.24949388e-01 6.57394975e-02 -5.89278638e-01 1.32185137e+00
3.78529519e-01 1.19546600e-01 4.98398751e-01 -1.75059235e+00
-5.79130054e-01 7.26235032e-01 -8.12522531e-01 6.17347836e-01
2.16273829e-01 1.32876948e-01 9.64279056e-01 -4.03899789e-01
-3.10755102e-03 1.16994894e+00 4.25261945e-01 3.65811773e-02
-9.66061532e-01 -7.86636353e-01 6.31395578e-01 6.79562509e-01
-1.25537753e+00 2.09530756e-01 1.40471780e+00 -2.88001597e-01
6.38064563e-01 1.88335717e-01 7.93961823e-01 1.06100893e+00
5.57710767e-01 7.05755949e-01 4.75337356e-01 -9.38702524e-02
2.44436458e-01 -3.42454791e-01 1.71715662e-01 5.96531808e-01
4.93390858e-01 -1.31706968e-01 -3.62911969e-01 -4.76020247e-01
4.04268503e-01 7.45172203e-01 -1.58332780e-01 -2.63325125e-01
-1.06608140e+00 6.66936398e-01 5.09768367e-01 4.41186666e-01
-7.30741918e-01 -1.41112864e-01 8.78149867e-01 6.28005385e-01
5.41538596e-01 2.49514416e-01 -4.62630570e-01 -2.67218146e-02
-1.24090359e-01 -2.33333841e-01 6.10064805e-01 6.77495539e-01
7.43987143e-01 7.00764418e-01 4.22902629e-02 4.84887749e-01
4.44972724e-01 4.33699697e-01 8.27143967e-01 -1.46672547e-01
6.60203338e-01 1.13706160e+00 -6.07245505e-01 -1.70903325e+00
-4.18916702e-01 -4.46392804e-01 -1.29932189e+00 -5.00741750e-02
5.52314371e-02 -3.33589524e-01 -8.09207857e-01 1.53445005e+00
3.76459748e-01 8.28541040e-01 3.29996087e-02 5.30710578e-01
3.63935202e-01 7.33080447e-01 4.92586829e-02 -3.39994818e-01
8.25466752e-01 -3.51889640e-01 -1.06470525e+00 7.99073093e-03
8.02318871e-01 -3.09133410e-01 5.67705393e-01 3.87860596e-01
-3.80814016e-01 -4.28770304e-01 -1.29477096e+00 7.91463971e-01
-5.90846837e-01 -4.61216003e-01 5.24667978e-01 2.25338086e-01
-2.96484023e-01 6.68132424e-01 -9.19118524e-01 -2.89041936e-01
4.77086902e-01 3.48134607e-01 -2.15399280e-01 -1.84910104e-01
-1.37058222e+00 3.74177277e-01 4.48868394e-01 2.49085739e-01
-5.94531953e-01 -3.59099776e-01 -1.20155787e+00 2.37631816e-02
6.10215127e-01 -2.84875721e-01 5.08674979e-01 -8.31155658e-01
-1.03644025e+00 6.57799318e-02 1.71791703e-01 -5.25478840e-01
1.66163556e-02 -6.49741739e-02 -1.33384800e+00 -1.17687788e-02
-2.23127622e-02 -5.37046373e-01 1.02275002e+00 -8.12914133e-01
-6.29514873e-01 -8.56308818e-01 -3.18354189e-01 -1.75635427e-01
-7.27131426e-01 -3.02676529e-01 1.07377730e-01 -8.10096800e-01
5.76936364e-01 -6.07011914e-01 -4.46658611e-01 -3.74956876e-01
-7.19043374e-01 -3.38387281e-01 1.52452517e+00 -7.63448536e-01
1.64545047e+00 -2.41680551e+00 1.96919858e-01 1.01395738e+00
4.44267929e-01 1.04737177e-01 -4.75375578e-02 4.05801356e-01
-4.14871633e-01 -9.98633206e-02 -4.57176208e-01 2.20435441e-01
-2.01620251e-01 2.88238436e-01 -2.51464486e-01 6.59843743e-01
3.04441124e-01 5.71252286e-01 -9.67496634e-01 -3.13145161e-01
4.20801580e-01 4.79210652e-02 -5.84309064e-02 3.18346500e-01
-3.70200388e-02 6.06430531e-01 -1.08417940e+00 9.25401509e-01
4.49538022e-01 -2.59717464e-01 -7.62615800e-02 -2.93250650e-01
2.58808613e-01 -2.71734685e-01 -1.33486176e+00 1.61480618e+00
-2.28468731e-01 2.90127456e-01 -2.12560520e-01 -1.57730663e+00
1.18308544e+00 3.69248956e-01 1.14350855e+00 -7.33206332e-01
2.22229868e-01 3.20651859e-01 1.28604889e-01 -7.14397192e-01
-1.23420298e-01 3.67075890e-01 -1.73444211e-01 2.73523390e-01
1.84023846e-02 2.51046509e-01 -6.62832037e-02 2.59138554e-01
1.57689703e+00 -4.69196588e-01 2.90040046e-01 1.08025305e-01
9.04426634e-01 -3.74838531e-01 8.01169872e-01 3.09894204e-01
-1.22708403e-01 1.92515209e-01 6.68186903e-01 -5.95918894e-01
-6.22483730e-01 -1.02645111e+00 9.58657712e-02 4.99217957e-01
3.75573933e-01 -5.84100306e-01 -1.84065640e-01 -1.11432910e+00
3.97851765e-01 5.11126280e-01 -4.51442808e-01 -8.85195494e-01
-4.37342554e-01 -7.13631809e-01 2.89227754e-01 5.35062790e-01
4.25964653e-01 -1.12482536e+00 7.95744658e-02 3.60947877e-01
1.37881443e-01 -8.75695646e-01 -1.42659202e-01 3.10899079e-01
-8.58141363e-01 -1.27537751e+00 -1.06686369e-01 -4.60113287e-01
6.97204769e-01 4.43487056e-02 6.88712537e-01 2.59412557e-01
-3.71176273e-01 5.01233816e-01 -5.57676494e-01 -5.53407371e-01
-1.25901163e-01 -2.24004865e-01 3.49853814e-01 5.85019827e-01
5.35971940e-01 -9.19212282e-01 -3.17129374e-01 4.42383289e-02
-1.08914912e+00 -8.24033856e-01 6.95785880e-01 7.56760657e-01
6.37719691e-01 5.75039804e-01 8.66444647e-01 -8.94144237e-01
5.32393754e-01 -1.18194294e+00 -4.62717533e-01 -4.07577828e-02
-7.22676158e-01 -2.10756119e-02 1.07649887e+00 -4.21006501e-01
-5.31520307e-01 -2.75390327e-01 1.31396083e-02 -9.38543856e-01
-2.59035468e-01 8.55446815e-01 -4.58169401e-01 -1.47660822e-01
3.74059886e-01 2.87087768e-01 1.12005711e-01 -3.38159651e-01
-7.87234008e-02 5.59385896e-01 2.22896293e-01 -3.23795408e-01
1.03704607e+00 1.83535740e-01 3.09833288e-01 -9.45700347e-01
-4.22683120e-01 -4.69726622e-01 -3.60923022e-01 -1.42891839e-01
6.00350916e-01 -7.01386690e-01 -5.31381190e-01 6.60737336e-01
-7.78330445e-01 3.09049189e-01 -4.69136536e-01 7.07755387e-01
-1.57999143e-01 6.53305948e-01 -5.37821174e-01 -6.90885723e-01
-1.34388015e-01 -7.83829808e-01 8.79974186e-01 2.17608780e-01
-4.41112109e-02 -1.15199780e+00 -6.29973263e-02 -3.03634197e-01
4.23395723e-01 7.16058731e-01 1.14567780e+00 -1.26297164e+00
-5.08964777e-01 -6.33630097e-01 4.52970304e-02 4.05173600e-01
6.21206760e-01 2.34749485e-02 -5.01328826e-01 -4.10934657e-01
3.98507342e-02 7.51873553e-02 5.58589578e-01 2.95086384e-01
1.76860762e+00 -3.33585411e-01 -5.62689424e-01 5.50659359e-01
1.23457885e+00 4.08884555e-01 4.49726015e-01 2.66127080e-01
1.20947647e+00 3.40684682e-01 6.84453726e-01 8.00395608e-01
2.10728884e-01 2.54699796e-01 9.00160611e-01 -3.67360301e-02
6.95749044e-01 -1.59810662e-01 4.62133586e-01 1.18624163e+00
1.17096074e-01 -2.67454386e-01 -7.82967567e-01 4.52464730e-01
-1.98546553e+00 -9.57088232e-01 -8.98910016e-02 2.10421300e+00
1.45725146e-01 3.06030422e-01 7.57229775e-02 5.98769844e-01
1.03384244e+00 4.99523938e-01 -8.97917867e-01 -1.18211836e-01
-6.50409907e-02 -6.82804063e-02 3.04962635e-01 -9.39527526e-02
-1.11768758e+00 2.83600450e-01 4.52152109e+00 6.90669656e-01
-1.09981048e+00 -2.75196463e-01 4.33215618e-01 1.91682562e-01
-3.71529430e-01 -2.85380900e-01 -1.72426686e-01 8.53186429e-01
1.03370714e+00 -4.95788366e-01 3.03124875e-01 9.05332625e-01
-4.22111452e-02 6.55303478e-01 -8.95343423e-01 1.08559811e+00
1.67591304e-01 -7.56451547e-01 1.79282933e-01 -1.55257266e-02
4.26919252e-01 1.22345928e-02 1.21948775e-02 3.72450411e-01
2.72218376e-01 -6.63211107e-01 1.03520788e-01 4.74014580e-01
1.66366786e-01 -8.55978847e-01 1.05654097e+00 5.12184426e-02
-1.76575172e+00 -5.92705607e-01 -2.23079056e-01 1.56015307e-01
1.84417441e-01 9.38490927e-01 -3.86028200e-01 1.21949351e+00
8.09958577e-01 1.47690451e+00 -6.01013005e-01 1.02028358e+00
1.17761297e-02 7.60736763e-01 -3.13186198e-01 1.27362296e-01
6.21740893e-02 -3.69548500e-01 8.60644817e-01 6.48193538e-01
6.00737154e-01 3.42687890e-02 5.48284292e-01 5.56246936e-01
1.59635499e-01 2.87782490e-01 -1.15226221e+00 -3.26451063e-01
4.71009523e-01 1.21672106e+00 -4.75936472e-01 -5.65632693e-02
-6.91855550e-01 7.59351611e-01 1.70444250e-02 4.89580810e-01
-6.95925355e-01 -6.48929894e-01 7.47316182e-01 -1.95598692e-01
1.07513301e-01 -1.11901253e-01 -7.63562545e-02 -1.21276295e+00
3.34665626e-01 -8.81227076e-01 9.50217128e-01 -2.13723540e-01
-1.84912276e+00 5.19238293e-01 -2.78703451e-01 -1.77403247e+00
-3.71714383e-01 -4.06078190e-01 -1.06565428e+00 6.49633169e-01
-1.32002330e+00 -9.27416861e-01 -5.84074795e-01 1.11999619e+00
4.02814537e-01 -4.33171153e-01 6.00969911e-01 4.80101526e-01
-1.00696564e+00 4.48676616e-01 -6.81096092e-02 3.36390972e-01
3.19157690e-01 -1.21855664e+00 2.92211711e-01 1.02715862e+00
9.00390223e-02 2.76272595e-01 4.78082120e-01 -8.69666219e-01
-1.52623856e+00 -1.36734843e+00 1.60398871e-01 6.76936954e-02
1.09402955e+00 6.90749474e-03 -1.33254886e+00 8.43939662e-01
-1.18551128e-01 5.49143791e-01 6.40701771e-01 1.99935749e-01
-9.95167643e-02 -4.10693288e-01 -1.00431871e+00 4.38113540e-01
9.93487239e-01 -5.33064663e-01 -3.92467588e-01 3.38360041e-01
8.81709397e-01 -2.03730673e-01 -9.09221530e-01 7.56988227e-01
-2.05962490e-02 -5.80308497e-01 8.05151343e-01 -7.93379605e-01
8.93189311e-02 -5.05929053e-01 -2.47977301e-01 -1.65177166e+00
-3.51074934e-01 -4.18941885e-01 -7.22789228e-01 1.25733638e+00
1.90373644e-01 -1.20329010e+00 4.85449493e-01 1.25015706e-01
-5.10295033e-01 -7.28689015e-01 -9.30316269e-01 -7.12802470e-01
-6.01787627e-01 -5.53795516e-01 9.70296919e-01 1.31344104e+00
1.48795575e-01 1.30074888e-01 -1.11412376e-01 7.01260686e-01
5.89922965e-01 -3.40794437e-02 5.51184535e-01 -1.59465253e+00
5.45370467e-02 -2.82875657e-01 -1.33121133e+00 -3.75877678e-01
3.60889733e-01 -8.25973094e-01 -3.66387367e-01 -1.13325286e+00
-5.08781612e-01 -4.47430432e-01 -1.14193618e+00 2.71802038e-01
-2.65667140e-01 -3.04011613e-01 -3.70737314e-01 -2.55750492e-02
-6.19766831e-01 9.49488163e-01 7.99290240e-01 -4.48762208e-01
-2.13177353e-01 1.33922294e-01 -4.32274938e-01 6.79096162e-01
7.95341849e-01 -3.52718234e-01 -5.42699814e-01 -7.34815374e-02
-4.04346399e-02 1.78165302e-01 2.45383590e-01 -1.30267370e+00
3.46315950e-01 -1.35812372e-01 4.11525220e-01 -5.36684453e-01
-4.39993292e-02 -1.48899305e+00 -3.16606537e-02 4.66350794e-01
1.03997095e-02 7.05509663e-01 1.63001776e-01 1.26247585e+00
-6.34757459e-01 2.71178961e-01 2.15357289e-01 1.87855184e-01
-1.00022233e+00 9.27235007e-01 -1.03431568e-01 -9.15007573e-03
1.39073539e+00 -3.86394151e-02 -1.80010125e-01 -3.31839293e-01
-7.64545858e-01 4.12404537e-01 3.61598507e-02 7.74799705e-01
8.46347332e-01 -1.76634169e+00 -4.74629194e-01 6.95188820e-01
4.78045076e-01 9.00602937e-02 4.83757973e-01 8.80260527e-01
-1.32864222e-01 -1.10790856e-01 -2.88697839e-01 -8.59688938e-01
-8.89949143e-01 7.45115101e-01 1.65689588e-01 -2.10302144e-01
-8.59196365e-01 3.54075044e-01 -2.37486124e-01 -9.81228277e-02
1.55684441e-01 -1.34015575e-01 -5.44111133e-01 4.95116003e-02
3.24973971e-01 5.32383740e-01 1.44971102e-01 -5.38103521e-01
-3.35416615e-01 5.51833272e-01 -2.13899128e-02 5.26415884e-01
1.19312882e+00 -1.66597754e-01 -1.67466417e-01 8.36484373e-01
1.21199071e+00 -5.68024293e-02 -7.30796635e-01 -5.84234715e-01
1.53402641e-01 -5.69241047e-01 -6.06249757e-02 -1.10376373e-01
-1.68266630e+00 6.96401596e-01 7.26413250e-01 8.69438827e-01
1.39249659e+00 -8.46030787e-02 9.06448483e-01 4.47897285e-01
3.98115844e-01 -8.88334692e-01 2.78042227e-01 2.19421744e-01
4.32645023e-01 -1.26134157e+00 -2.45657250e-01 -3.84090990e-01
-5.53800642e-01 1.12240314e+00 9.25764263e-01 -1.97111189e-01
1.00651979e+00 1.66049153e-02 8.24117661e-03 -5.05311966e-01
-4.18399006e-01 -3.52701060e-02 2.33173162e-01 6.24542952e-01
2.33023651e-02 -5.92164434e-02 8.68064240e-02 5.52576125e-01
2.46662512e-01 -5.99008143e-01 2.06110924e-01 1.00240827e+00
-1.85058117e-01 -9.05144513e-01 -2.08580211e-01 1.03496933e+00
-5.17903447e-01 2.56102264e-01 -9.96880829e-02 5.57586551e-01
-6.15514405e-02 1.05486310e+00 1.46254271e-01 -8.82995307e-01
5.97289860e-01 5.52059561e-02 -1.72755584e-01 -4.67800617e-01
-1.92985401e-01 -3.72625679e-01 -3.91340166e-01 -8.95339787e-01
-1.22751959e-01 -6.27251446e-01 -1.43444037e+00 -1.75192997e-01
-4.43203092e-01 3.05418342e-01 3.13184232e-01 1.12407911e+00
5.47222793e-01 1.17302775e+00 1.18247294e+00 -4.46026623e-01
-3.87956321e-01 -8.56014013e-01 -1.12003291e+00 8.20798457e-01
5.41938424e-01 -8.16526592e-01 -8.50688517e-01 -4.63254303e-01]
|
[7.278781890869141, 2.681631326675415]
|
ec76d10b-8a0d-4770-a4c2-02a5dd52bd75
|
neural-symbolic-computing-an-effective
|
1905.06088
| null |
https://arxiv.org/abs/1905.06088v1
|
https://arxiv.org/pdf/1905.06088v1.pdf
|
Neural-Symbolic Computing: An Effective Methodology for Principled Integration of Machine Learning and Reasoning
|
Current advances in Artificial Intelligence and machine learning in general, and deep learning in particular have reached unprecedented impact not only across research communities, but also over popular media channels. However, concerns about interpretability and accountability of AI have been raised by influential thinkers. In spite of the recent impact of AI, several works have identified the need for principled knowledge representation and reasoning mechanisms integrated with deep learning-based systems to provide sound and explainable models for such systems. Neural-symbolic computing aims at integrating, as foreseen by Valiant, two most fundamental cognitive abilities: the ability to learn from the environment, and the ability to reason from what has been learned. Neural-symbolic computing has been an active topic of research for many years, reconciling the advantages of robust learning in neural networks and reasoning and interpretability of symbolic representation. In this paper, we survey recent accomplishments of neural-symbolic computing as a principled methodology for integrated machine learning and reasoning. We illustrate the effectiveness of the approach by outlining the main characteristics of the methodology: principled integration of neural learning with symbolic knowledge representation and reasoning allowing for the construction of explainable AI systems. The insights provided by neural-symbolic computing shed new light on the increasingly prominent need for interpretable and accountable AI systems.
|
['Luis C. Lamb', "Artur d'Avila Garcez", 'Marco Gori', 'Son N. Tran', 'Luciano Serafini', 'Michael Spranger']
|
2019-05-15
| null | null | null | null |
['explainable-models']
|
['computer-vision']
|
[ 2.11277366e-01 8.22268307e-01 -1.99117079e-01 -4.47931975e-01
1.23327449e-01 -4.37271625e-01 8.87416005e-01 1.23657800e-01
-7.83191025e-02 5.84454179e-01 3.06156665e-01 -5.12162209e-01
-6.50768340e-01 -8.73465478e-01 -6.97353065e-01 -3.62807661e-01
6.83618411e-02 6.54247284e-01 -1.77424535e-01 -4.14892942e-01
2.37756863e-01 8.00279498e-01 -1.60061002e+00 6.85780942e-01
8.82231414e-01 8.42397332e-01 -3.37501645e-01 4.14345026e-01
-2.16253057e-01 1.75659740e+00 -4.45335060e-01 -4.04712707e-01
-1.93509102e-01 -5.32449186e-01 -1.15731609e+00 -3.46508950e-01
4.75922450e-02 -1.36734098e-01 -2.78024584e-01 9.81288373e-01
-1.71562299e-01 -3.01857702e-02 5.83233953e-01 -1.32401490e+00
-1.22019660e+00 1.09508681e+00 1.89382225e-01 3.07287294e-02
1.83030427e-01 1.34680241e-01 1.09329140e+00 -5.62449157e-01
4.54021394e-01 1.44412231e+00 5.70869148e-01 6.87848508e-01
-1.27689314e+00 -4.29599732e-01 9.00805146e-02 5.80792785e-01
-9.68921542e-01 -3.26474190e-01 7.65434206e-01 -5.37296593e-01
1.24056518e+00 4.52883810e-01 1.12414622e+00 8.03143680e-01
3.05406719e-01 7.40949333e-01 1.08807278e+00 -7.47852504e-01
3.29784065e-01 3.85502011e-01 2.60378391e-01 7.74747849e-01
4.44306016e-01 3.70470494e-01 -6.47964835e-01 6.98369816e-02
9.04128551e-01 7.39929229e-02 9.20067448e-03 -3.11316460e-01
-1.25127792e+00 7.78941572e-01 6.46292627e-01 6.78872883e-01
-5.38375437e-01 5.50689280e-01 5.01354158e-01 3.83224189e-01
2.83746421e-01 9.52126801e-01 -5.14587879e-01 -8.79444480e-02
-9.62911248e-01 4.00331020e-01 8.74443650e-01 4.57767695e-01
4.70257312e-01 5.33980012e-01 2.19404265e-01 1.55841529e-01
5.27354300e-01 4.65895057e-01 6.00693882e-01 -1.39892483e+00
-6.17420860e-02 1.06768143e+00 -2.18422398e-01 -1.34581995e+00
-4.02062058e-01 -3.92353684e-01 -7.85869718e-01 5.12959778e-01
2.58620530e-01 7.65886530e-02 -5.02601862e-01 1.64239120e+00
-7.83994794e-02 -1.68894440e-01 4.19425547e-01 7.73019373e-01
6.62285984e-01 5.50145745e-01 1.17150797e-02 1.36136055e-01
9.87550735e-01 -8.54542375e-01 -6.65597677e-01 -2.56873131e-01
6.87003076e-01 3.46086062e-02 8.35390806e-01 4.46062833e-01
-1.26035821e+00 -4.97939110e-01 -9.67739046e-01 -2.22049952e-01
-6.33848965e-01 -1.10828355e-01 1.25977492e+00 4.12681937e-01
-1.16648054e+00 6.62617803e-01 -9.61749732e-01 -1.12853028e-01
5.78068674e-01 5.58775961e-01 -2.67580450e-01 1.93929926e-01
-1.24240983e+00 1.40235722e+00 7.33503699e-01 2.95103371e-01
-5.77511489e-01 -3.91895771e-01 -8.13251972e-01 2.55310744e-01
2.36683100e-01 -7.52974749e-01 1.22443390e+00 -1.74654877e+00
-1.36222255e+00 9.37753737e-01 -1.10998042e-01 -8.97893369e-01
3.51659507e-01 -2.71579713e-01 -3.58739346e-01 5.34797870e-02
-2.62510449e-01 4.28133160e-01 3.40246171e-01 -1.05850601e+00
-1.21988371e-01 -4.29065377e-01 2.93317109e-01 -6.59374148e-02
-1.25452802e-01 -6.40897974e-02 4.21758711e-01 -3.25015575e-01
2.65365452e-01 -8.77574801e-01 6.38612593e-03 1.04753107e-01
-6.00441918e-02 -3.33594471e-01 5.11520147e-01 -4.91731316e-01
9.08751786e-01 -1.89599252e+00 3.16837192e-01 2.65326619e-01
5.66044033e-01 4.67837542e-01 2.28227600e-01 3.53286088e-01
-4.25399184e-01 2.23431990e-01 -1.73295643e-02 1.71474561e-01
2.47228205e-01 5.49694955e-01 -7.94664621e-01 1.46194696e-01
3.98934543e-01 1.27768922e+00 -9.62548375e-01 -2.39186749e-01
3.36625993e-01 5.35189748e-01 -5.69381714e-01 8.88412222e-02
-3.58594060e-01 3.66089880e-01 -5.16700387e-01 4.86401856e-01
6.93594944e-03 -4.61621225e-01 3.72176468e-01 1.83662087e-01
-1.41408160e-01 3.40608776e-01 -8.55172932e-01 1.22389793e+00
-2.85898894e-01 1.02564561e+00 -3.44509929e-01 -1.36743677e+00
9.87147272e-01 6.57362223e-01 7.38130361e-02 -6.65847063e-01
1.75833091e-01 5.24167538e-01 4.61724430e-01 -6.41662180e-01
2.25696266e-01 -5.54388821e-01 1.84408799e-01 6.26960993e-01
-1.11842547e-02 -4.14016128e-01 -1.72634423e-01 2.77693868e-02
6.70485616e-01 5.02188206e-01 6.13900363e-01 -1.07103035e-01
6.97069645e-01 3.21977705e-01 1.98071912e-01 5.86194634e-01
-6.72451556e-02 1.24741413e-01 4.63004082e-01 -1.18127871e+00
-1.00360274e+00 -7.10826099e-01 7.78794959e-02 1.10788906e+00
-1.75097808e-01 -3.26255970e-02 -8.19587827e-01 -1.69136822e-01
-6.46489188e-02 1.16655827e+00 -7.91558802e-01 -4.79450047e-01
-5.99636316e-01 -3.34903359e-01 7.35775948e-01 6.81399763e-01
4.08384383e-01 -1.69779575e+00 -1.03419936e+00 1.30830720e-01
1.15044944e-01 -8.51849198e-01 8.51784408e-01 4.08029705e-01
-9.58662629e-01 -9.51922894e-01 -2.25091167e-02 -4.86438096e-01
5.87534189e-01 -9.93924122e-03 1.19680429e+00 5.75121045e-01
5.61482310e-02 3.55325460e-01 -1.69465289e-01 -7.43182540e-01
-9.11138475e-01 2.42973268e-02 1.71663567e-01 -2.47818381e-01
4.46831226e-01 -7.32484758e-01 7.71428794e-02 -1.36048108e-01
-1.01084936e+00 5.51404476e-01 5.81127226e-01 6.41572118e-01
1.83255345e-01 -3.56379747e-02 7.46397972e-01 -8.98804545e-01
6.04440928e-01 -5.32348990e-01 -3.14900011e-01 3.76026124e-01
-7.44644344e-01 4.13978696e-01 7.84732997e-01 -1.00749716e-01
-1.12216544e+00 -3.96589249e-01 1.06094800e-01 -4.90511209e-02
-3.52305830e-01 7.98436344e-01 4.79467288e-02 -1.09497458e-01
8.87697935e-01 3.83687556e-01 1.24815561e-01 -3.38375270e-02
6.28895164e-01 4.90882576e-01 5.60252786e-01 -8.63279104e-01
6.68762326e-01 4.46177751e-01 8.30907524e-02 -6.05115175e-01
-8.96128297e-01 3.16385865e-01 -7.71485686e-01 -1.92453787e-01
7.34623969e-01 -6.02957308e-01 -8.72282743e-01 1.87384576e-01
-1.47214973e+00 -2.97831327e-01 -7.45003939e-01 3.68143111e-01
-8.70792806e-01 -8.12559947e-02 -2.72166640e-01 -8.91698480e-01
-2.52189040e-01 -9.80461001e-01 3.63158196e-01 1.73829600e-01
-9.12543595e-01 -1.36201465e+00 -1.39369443e-02 4.75697368e-01
5.16469479e-01 5.16555309e-01 1.16129625e+00 -1.17317498e+00
-5.97686768e-01 -3.87099981e-01 -2.11221471e-01 4.89926547e-01
-1.85423806e-01 -6.27089962e-02 -1.19548571e+00 4.10087883e-01
1.12481192e-01 -6.13856852e-01 5.26537538e-01 1.47681236e-01
9.61384714e-01 -4.95711714e-01 -7.90411979e-02 3.63894612e-01
1.05826843e+00 2.69707561e-01 4.29872781e-01 5.01556635e-01
5.51632345e-01 7.77712822e-01 5.25026321e-02 -1.16958162e-02
4.07618105e-01 3.60048741e-01 4.92709875e-01 6.47634789e-02
-6.77077845e-02 -1.53904617e-01 3.08404952e-01 7.72390485e-01
-7.53959477e-01 2.18326077e-01 -1.32778513e+00 3.47957999e-01
-2.18380499e+00 -1.22940969e+00 -6.26887903e-02 1.65104997e+00
8.23984683e-01 2.71593183e-01 -2.50075698e-01 4.10984844e-01
2.96647638e-01 -7.20547363e-02 -5.98773479e-01 -1.06796825e+00
-2.15697348e-01 1.12591438e-01 -2.31524304e-01 5.73300719e-01
-5.21196783e-01 1.03069377e+00 6.96886444e+00 3.42443109e-01
-1.15393639e+00 -1.61317393e-01 4.34739172e-01 9.46809053e-02
-5.00612974e-01 3.12849246e-02 -2.73924053e-01 -7.34326392e-02
1.28417516e+00 -2.82063633e-01 7.70328999e-01 1.01332033e+00
2.22294658e-01 1.11497335e-01 -1.50897193e+00 5.50606668e-01
1.19264990e-01 -1.75100553e+00 3.88856828e-01 -8.44645500e-02
7.66657293e-01 -1.11254945e-01 3.92895920e-04 2.68257797e-01
2.99127907e-01 -1.56893325e+00 1.19197845e+00 7.24624395e-01
2.64919519e-01 -5.28854489e-01 8.08142364e-01 5.99021256e-01
-6.11938715e-01 -3.53173167e-01 -1.81421712e-01 -1.02585900e+00
-3.48614067e-01 3.85192752e-01 -5.66779792e-01 2.86305338e-01
1.83042362e-01 7.17228711e-01 -3.72406423e-01 4.70311224e-01
-5.63122392e-01 5.28768718e-01 -1.33012598e-02 -2.57405132e-01
3.07264060e-01 5.67627326e-02 3.86522591e-01 1.14693427e+00
-1.30025014e-01 1.26802877e-01 -3.75844508e-01 1.55590022e+00
2.79192060e-01 -4.05426979e-01 -7.22585201e-01 -4.16201621e-01
3.09587866e-01 9.01117146e-01 -7.31549382e-01 -4.77764994e-01
-2.53248274e-01 4.77229506e-01 3.77740622e-01 1.91532642e-01
-7.27847099e-01 -3.20283510e-02 3.68420690e-01 4.38994765e-02
-3.94313000e-02 -3.07939976e-01 -7.20086992e-01 -1.10815239e+00
-2.25302696e-01 -1.10922909e+00 -1.05156168e-01 -9.02556419e-01
-7.51562119e-01 6.34144127e-01 1.46424159e-01 -3.62115383e-01
-6.43751085e-01 -9.69827235e-01 -4.99332458e-01 8.33296657e-01
-1.37536919e+00 -1.51264536e+00 -1.59943655e-01 3.33771467e-01
2.89955437e-01 -3.35230410e-01 1.27777445e+00 -3.47346663e-01
-2.39112347e-01 3.62579226e-02 -6.39937222e-02 1.19973093e-01
-1.20948188e-01 -1.20350444e+00 3.50097358e-01 3.70656878e-01
3.70232463e-01 8.99738848e-01 8.02182853e-01 -2.46719837e-01
-1.34322226e+00 -7.27067113e-01 1.15669501e+00 -7.49819875e-01
7.05774367e-01 -1.16151407e-01 -1.07076621e+00 1.10680306e+00
1.15175128e-01 -1.45659864e-01 6.13373458e-01 1.57070458e-01
-5.34827352e-01 1.32185929e-02 -9.13587213e-01 6.23724997e-01
6.77701712e-01 -7.99703956e-01 -1.19962263e+00 2.36893415e-01
7.17519760e-01 -1.67026564e-01 -5.33783793e-01 7.15543255e-02
7.90013075e-01 -1.33079779e+00 8.53346944e-01 -9.26805735e-01
7.68633902e-01 -2.14110970e-01 -1.56813040e-02 -1.03195119e+00
-4.14569110e-01 -4.27866966e-01 -3.46106887e-01 8.36641431e-01
3.75543743e-01 -8.48962426e-01 6.29578590e-01 1.18506896e+00
-1.78371444e-01 -8.17166328e-01 -7.00335860e-01 -4.63065565e-01
3.07869375e-01 -7.65980184e-01 6.34065449e-01 1.05004776e+00
4.22182798e-01 2.27670312e-01 -8.54334701e-03 -1.14773564e-01
2.32307360e-01 1.94742039e-01 5.08206785e-01 -1.68665218e+00
-2.92710364e-01 -6.86067224e-01 -6.50490642e-01 -3.53388041e-01
6.09931946e-01 -1.17064202e+00 -3.04346949e-01 -1.73837352e+00
2.00738475e-01 1.01599485e-01 -3.10587555e-01 8.83903682e-01
3.44833374e-01 7.35650286e-02 2.50906080e-01 3.93847704e-01
-4.49676037e-01 5.61407767e-02 1.10373747e+00 1.02720834e-01
-1.76797900e-02 -4.16752666e-01 -9.55053449e-01 1.35531950e+00
7.76146293e-01 -3.53675485e-01 -4.09425944e-01 -5.29919147e-01
1.02530384e+00 -1.93532929e-01 9.06025946e-01 -1.19134843e+00
2.93598711e-01 -4.67645884e-01 3.74933183e-01 8.87566060e-02
1.80152506e-01 -1.06151605e+00 3.10409307e-01 8.26848686e-01
-8.03699672e-01 -8.88875425e-02 2.84635603e-01 1.88660428e-01
-2.95805752e-01 -1.47924751e-01 5.62997282e-01 -3.65117162e-01
-7.57667124e-01 -4.44256097e-01 -3.91432643e-01 -6.01300448e-02
9.26800609e-01 -4.54004318e-01 -2.96401858e-01 -3.10290933e-01
-8.29106748e-01 -1.40204996e-01 2.09864810e-01 3.35928977e-01
5.77642977e-01 -1.04131258e+00 -5.28821588e-01 2.18382865e-01
-2.37685978e-01 -4.40558679e-02 -1.61922261e-01 6.60170913e-01
-7.26853609e-01 9.02329087e-01 -5.26650488e-01 -8.37121829e-02
-8.11143875e-01 4.01088268e-01 6.14836574e-01 -8.42078701e-02
-5.53297281e-01 6.22002125e-01 1.30513355e-01 -5.24180830e-01
7.34428316e-02 -7.02859461e-01 -2.49992847e-01 -2.75610566e-01
6.17580473e-01 4.07559276e-01 -2.08003521e-01 -4.53143448e-01
-3.56959909e-01 3.19777071e-01 1.37489527e-01 4.46512178e-02
1.60579705e+00 2.94859648e-01 -5.12586892e-01 9.43963587e-01
5.50812542e-01 -4.55341548e-01 -7.40989506e-01 -6.36717752e-02
2.26052850e-01 1.10705718e-01 1.89488474e-02 -1.10620260e+00
-6.52586579e-01 1.18969464e+00 -5.18641155e-03 6.47621512e-01
7.53366053e-01 3.20864655e-02 2.79559225e-01 7.85295367e-01
2.09249854e-01 -8.26025784e-01 4.07199040e-02 7.22178102e-01
1.08803844e+00 -9.91361141e-01 1.05369434e-01 5.76559827e-02
-5.41626573e-01 1.58418190e+00 4.61493224e-01 -2.20207781e-01
3.06479573e-01 2.15010028e-02 -1.13272056e-01 -5.44678211e-01
-6.89590633e-01 2.68980652e-01 4.07912314e-01 6.73102081e-01
7.63403475e-01 2.55755007e-01 1.34259850e-01 7.39956141e-01
-5.09953916e-01 5.16527951e-01 3.10833305e-01 6.92637682e-01
-6.54562533e-01 -7.26162910e-01 -4.61697608e-01 -4.12616730e-02
-2.58310139e-01 -1.42373070e-01 -8.17155063e-01 9.99165833e-01
5.38272202e-01 6.26375973e-01 1.18698208e-02 -1.83526516e-01
2.37726532e-02 3.77003461e-01 3.68368208e-01 -5.85952282e-01
-6.95741236e-01 -6.86439693e-01 5.76207228e-02 -4.73602951e-01
-6.59786463e-01 -4.13491249e-01 -1.66349280e+00 -4.97110516e-01
-9.20653790e-02 2.21130520e-01 5.51883280e-01 1.42205882e+00
2.32784897e-01 7.36394465e-01 -8.83116722e-02 -7.81869352e-01
-5.89838088e-01 -5.45087516e-01 -4.17620987e-01 9.42169577e-02
4.29664910e-01 -3.67666215e-01 -3.47150087e-01 2.62054026e-01]
|
[9.129398345947266, 6.655424118041992]
|
9de858cc-dc4c-41de-92ef-183b8cd4a612
|
neural-pipeline-for-zero-shot-data-to-text-1
|
2203.16279
| null |
https://arxiv.org/abs/2203.16279v1
|
https://arxiv.org/pdf/2203.16279v1.pdf
|
Neural Pipeline for Zero-Shot Data-to-Text Generation
|
In data-to-text (D2T) generation, training on in-domain data leads to overfitting to the data representation and repeating training data noise. We examine how to avoid finetuning pretrained language models (PLMs) on D2T generation datasets while still taking advantage of surface realization capabilities of PLMs. Inspired by pipeline approaches, we propose to generate text by transforming single-item descriptions with a sequence of modules trained on general-domain text-based operations: ordering, aggregation, and paragraph compression. We train PLMs for performing these operations on a synthetic corpus WikiFluent which we build from English Wikipedia. Our experiments on two major triple-to-text datasets -- WebNLG and E2E -- show that our approach enables D2T generation from RDF triples in zero-shot settings.
|
['Ondřej Dušek', 'Zdeněk Kasner']
|
2022-03-30
| null |
https://aclanthology.org/2022.acl-long.271
|
https://aclanthology.org/2022.acl-long.271.pdf
|
acl-2022-5
|
['data-to-text-generation']
|
['natural-language-processing']
|
[ 3.03465456e-01 1.09856021e+00 -1.09908640e-01 -3.07633400e-01
-1.06188726e+00 -3.81144673e-01 1.15473640e+00 3.31571072e-01
-1.23327605e-01 1.06104100e+00 6.93331838e-01 -2.67207742e-01
1.92174613e-01 -1.34233809e+00 -1.26699221e+00 1.79517999e-01
1.77031621e-01 1.10565972e+00 6.36092573e-02 -5.49436271e-01
1.66472152e-01 -1.80856183e-01 -1.74658060e+00 9.73748446e-01
1.28425860e+00 6.73136711e-01 7.38772750e-02 5.02369761e-01
-9.87881303e-01 9.04991031e-01 -5.96886933e-01 -8.36371720e-01
3.68556499e-01 -4.53319222e-01 -9.13787663e-01 1.05311833e-01
5.51454186e-01 -1.78948209e-01 -1.42976046e-01 6.27050400e-01
4.16770160e-01 1.78051572e-02 9.41539645e-01 -9.97861266e-01
-9.57249880e-01 1.34058106e+00 -1.26176134e-01 -5.42618155e-01
4.81935203e-01 2.04308718e-01 1.14774680e+00 -1.24004233e+00
1.27759635e+00 1.32871568e+00 6.86185658e-01 9.30519640e-01
-1.59356046e+00 -1.74874857e-01 -4.80794042e-01 -2.88049549e-01
-1.19987798e+00 -8.45301986e-01 4.59797569e-02 -2.70603538e-01
1.67296803e+00 8.86323750e-02 5.31864405e-01 1.32001889e+00
-2.39847362e-01 9.14398909e-01 5.64941704e-01 -6.44342661e-01
3.08415860e-01 2.69635856e-01 -3.48926216e-01 6.02723777e-01
7.81491280e-01 -1.90891981e-01 -8.57621312e-01 -1.45095393e-01
5.36458433e-01 -7.53063321e-01 2.17234582e-01 -3.31028402e-01
-1.30824912e+00 7.39167333e-01 -4.73904684e-02 -4.71113436e-02
-2.94460803e-01 3.67238939e-01 6.06438339e-01 4.71507758e-01
5.34195840e-01 7.04046726e-01 -5.35564184e-01 -1.57082230e-01
-9.19312060e-01 7.57116139e-01 1.15970922e+00 1.85422611e+00
8.98035765e-01 1.54188514e-01 -6.12773299e-01 8.05791438e-01
6.95653558e-02 6.83613062e-01 5.97207248e-01 -7.99597681e-01
1.16490483e+00 6.99235380e-01 3.76516908e-01 -1.23175293e-01
-1.76223427e-01 1.22736998e-01 -5.23291826e-01 -3.90441090e-01
2.97760099e-01 -3.95027488e-01 -1.16128993e+00 1.38754416e+00
9.97319818e-02 -4.64131743e-01 6.15482450e-01 3.28919530e-01
1.02906191e+00 6.34647429e-01 2.93488622e-01 7.85674676e-02
9.91694629e-01 -6.56944573e-01 -5.25392890e-01 -3.33498657e-01
1.41210079e+00 -4.56524968e-01 1.23900318e+00 1.40902773e-01
-1.58383214e+00 -4.81704950e-01 -8.27695072e-01 -6.51512206e-01
-6.05034947e-01 1.24554873e-01 5.33057868e-01 6.01379871e-01
-8.84930491e-01 7.05404758e-01 -5.40456116e-01 -6.83887005e-01
4.60737526e-01 -3.42697091e-02 -4.36474830e-01 -6.61994666e-02
-1.24506903e+00 8.14915180e-01 9.47874546e-01 -4.49349701e-01
-8.83996129e-01 -1.06405509e+00 -9.69474256e-01 -6.62576333e-02
4.68639433e-01 -1.21225154e+00 1.36643815e+00 -4.25810516e-01
-1.44378066e+00 8.16191792e-01 -1.72054380e-01 -7.94413328e-01
4.93845403e-01 -1.40412450e-01 -3.06530595e-01 -1.90169170e-01
3.15534472e-01 1.01202798e+00 6.29624128e-01 -1.13907719e+00
-6.21350706e-01 -2.28207875e-02 -9.78873074e-02 1.82899773e-01
-3.89368922e-01 -3.67667437e-01 -1.43440217e-01 -5.37995100e-01
-2.40760550e-01 -5.47297060e-01 -1.40798427e-02 -3.30146044e-01
-8.85325551e-01 -3.36697042e-01 3.81202698e-01 -5.37967741e-01
1.20949447e+00 -1.50210273e+00 1.44745409e-01 1.19998969e-01
-1.04632750e-01 1.17629074e-01 -4.09309298e-01 1.09682226e+00
4.01119515e-02 6.45495713e-01 -1.56202242e-01 -4.67095196e-01
5.88106334e-01 3.78180146e-01 -4.90769774e-01 -4.77657050e-01
4.96327490e-01 1.27458549e+00 -9.16759551e-01 -6.78449094e-01
-1.21784292e-01 -1.42623872e-01 -9.85220432e-01 3.32324058e-01
-1.20081353e+00 -3.13194767e-02 -3.65158647e-01 3.80027831e-01
3.92970711e-01 -2.08726257e-01 4.98209119e-01 -2.92035908e-01
-9.86624360e-02 8.01003575e-01 -9.49353397e-01 2.29454064e+00
-6.45565331e-01 1.96871325e-01 -6.35783374e-01 -7.56266713e-01
9.93799627e-01 2.54585654e-01 7.94797763e-02 -9.03263152e-01
-3.18553865e-01 4.66290385e-01 -4.50371772e-01 -9.77506042e-01
1.02639878e+00 -2.84191400e-01 -4.21489894e-01 6.65850163e-01
6.43350422e-01 -6.12010300e-01 9.75552917e-01 5.47971249e-01
1.07819510e+00 5.68481266e-01 8.91959369e-02 -2.70776629e-01
-4.35017347e-02 3.52869332e-01 1.56417817e-01 8.83995533e-01
1.06065166e+00 5.63216031e-01 6.16315603e-01 -2.44815424e-01
-1.75359356e+00 -1.03322554e+00 1.02067657e-01 8.56461287e-01
-5.15945911e-01 -9.76281106e-01 -7.65623569e-01 -8.48164201e-01
2.49568775e-01 1.69329655e+00 -4.78186697e-01 -1.70536757e-01
-7.15451419e-01 -7.22693205e-01 9.71961260e-01 4.05509144e-01
3.86142403e-01 -8.09178233e-01 -2.29338974e-01 4.50916708e-01
-3.83389026e-01 -1.10392129e+00 -3.75863314e-01 2.10988864e-01
-7.84884453e-01 -6.69743717e-01 -3.61033618e-01 -5.58500588e-01
7.55579710e-01 -2.03851625e-01 1.70190048e+00 -1.78856283e-01
-2.60351568e-01 2.16047585e-01 -3.86629075e-01 -5.43984294e-01
-1.12035263e+00 5.46428025e-01 -1.97301522e-01 -4.70877618e-01
4.91924912e-01 -4.56973135e-01 -6.49877116e-02 -2.77281284e-01
-1.16870475e+00 6.06113732e-01 4.71728146e-01 7.42446899e-01
4.67930883e-01 -3.03219527e-01 7.33305275e-01 -1.28937650e+00
7.36524761e-01 -6.52120352e-01 -4.28179562e-01 6.45721138e-01
-5.40537357e-01 7.20043898e-01 6.59063995e-01 3.67611013e-02
-1.47251105e+00 -3.49533409e-01 3.75612490e-02 -3.46870869e-02
1.64730083e-02 6.76093400e-01 -3.08445632e-01 6.57163084e-01
1.15266967e+00 3.14293414e-01 -3.05362016e-01 -5.03128946e-01
1.01733232e+00 4.88204688e-01 2.33292818e-01 -1.27957451e+00
8.66044164e-01 1.85269251e-01 -1.53658256e-01 -7.44633913e-01
-8.50386679e-01 2.44987801e-01 -4.99360919e-01 2.51059622e-01
7.17240453e-01 -9.66816723e-01 -2.20572412e-01 1.46377593e-01
-1.12941718e+00 -6.81002796e-01 -1.01980400e+00 -1.69911478e-02
-8.17130744e-01 1.82902336e-01 -5.66448092e-01 -5.45051098e-01
-6.09572351e-01 -5.37501574e-01 1.30343950e+00 -6.42570630e-02
-2.50077963e-01 -1.06479836e+00 1.22325428e-01 1.89652115e-01
3.93657267e-01 -2.80276127e-02 1.32246590e+00 -9.07569706e-01
-7.23432183e-01 8.67858529e-02 -2.39127710e-01 2.58143902e-01
-8.27263445e-02 -2.34273411e-02 -6.82036340e-01 1.31629288e-01
-6.50080800e-01 -8.66499841e-01 6.80990040e-01 -2.69681871e-01
9.63858902e-01 -7.30661094e-01 -2.01591760e-01 5.38794577e-01
1.49907219e+00 -3.45519960e-01 9.33098495e-01 3.14757019e-01
6.18298709e-01 6.80957556e-01 4.80043024e-01 6.49067938e-01
7.70138800e-01 5.36073923e-01 1.20124780e-01 3.31716239e-01
-5.00972271e-01 -1.12608159e+00 3.87304693e-01 6.42383277e-01
6.63441513e-03 -5.74825525e-01 -8.91059577e-01 8.77370477e-01
-1.79667211e+00 -1.07967401e+00 -1.92995876e-01 2.10834312e+00
1.38801539e+00 1.49491832e-01 -6.46294802e-02 -3.50792348e-01
2.76774049e-01 -2.39739046e-01 -3.74895513e-01 -4.31356013e-01
-4.49624449e-01 4.79357153e-01 6.04389369e-01 1.15594923e-01
-5.60042083e-01 1.05515218e+00 6.15160131e+00 1.00586700e+00
-6.15713179e-01 -1.54587897e-02 3.52919102e-01 -3.09280813e-01
-1.12282693e+00 2.09602386e-01 -1.13119447e+00 3.33916247e-01
1.28396797e+00 -6.99695766e-01 5.03893912e-01 5.42405903e-01
1.80957451e-01 1.53073072e-01 -1.47667885e+00 4.21768278e-01
3.54593014e-03 -1.92173970e+00 8.45848739e-01 -6.18739352e-02
1.00721776e+00 7.07679987e-03 -2.96966761e-01 7.22411275e-01
9.30815220e-01 -1.06119955e+00 1.03173721e+00 7.09038019e-01
1.10350478e+00 -4.18354720e-01 2.74524003e-01 3.85281742e-01
-8.35361421e-01 4.81145754e-02 -5.89928150e-01 3.04087818e-01
2.81043142e-01 7.90301144e-01 -1.40140557e+00 9.09051061e-01
3.09299707e-01 6.03373468e-01 -3.74047101e-01 4.43500280e-01
-2.38097832e-01 2.86561131e-01 -3.32567453e-01 4.13768888e-02
-4.25562710e-02 -2.91855540e-02 2.99586713e-01 1.43730140e+00
9.65707242e-01 -2.35239655e-01 -1.01722233e-01 1.27394462e+00
-5.91253400e-01 1.08366370e-01 -9.09440517e-01 -4.73526090e-01
6.02751315e-01 1.11465621e+00 3.38061270e-03 -7.42727995e-01
-5.44300497e-01 7.57426381e-01 5.59712291e-01 2.77609646e-01
-5.46269655e-01 -4.90400523e-01 3.74906182e-01 5.26584685e-01
4.82112557e-01 -1.82245180e-01 -3.25490236e-01 -1.55305314e+00
1.82129368e-01 -9.81416523e-01 2.53312588e-01 -9.65172648e-01
-1.39169168e+00 3.13446552e-01 2.45360613e-01 -9.73222256e-01
-7.54152298e-01 -2.73949534e-01 -3.47109824e-01 8.13738465e-01
-1.39981091e+00 -1.49035013e+00 -9.25850719e-02 3.67045939e-01
4.96063560e-01 -1.25788525e-01 8.88215899e-01 1.74988851e-01
-2.22147420e-01 5.48999786e-01 -9.14526451e-03 4.84454446e-02
6.99296951e-01 -1.50540066e+00 1.18086505e+00 5.97835064e-01
-1.06541567e-01 7.90760517e-01 7.08206177e-01 -1.03527331e+00
-1.64657021e+00 -1.48079276e+00 1.36368012e+00 -5.37866056e-01
9.16215241e-01 -6.86127722e-01 -6.76823676e-01 1.01760483e+00
3.65537196e-01 -2.95233428e-01 5.11127412e-01 -1.15306899e-01
-5.58346093e-01 1.43656343e-01 -1.09164786e+00 7.24571705e-01
1.59281218e+00 -4.37543362e-01 -7.66391158e-01 5.54570198e-01
8.58831942e-01 -4.99557674e-01 -1.30350983e+00 1.51581019e-01
2.85323650e-01 -6.64716542e-01 7.83463955e-01 -9.58647966e-01
9.34658885e-01 -1.80864513e-01 -3.14232558e-01 -1.68135703e+00
2.46732786e-01 -8.64480436e-01 -2.53303677e-01 1.43266249e+00
9.15503979e-01 -3.45318764e-01 8.01611662e-01 8.88565540e-01
-4.88391459e-01 -7.05042332e-02 -7.55924106e-01 -8.37126672e-01
4.07450050e-01 -3.26952726e-01 1.10519803e+00 6.20589793e-01
4.27786171e-01 5.83656013e-01 -3.47485989e-01 -5.69272101e-01
6.97527826e-01 -1.64729938e-01 1.26347876e+00 -9.95664179e-01
-2.49630108e-01 6.51648119e-02 3.71485680e-01 -1.14132762e+00
-4.07329760e-02 -1.41315222e+00 2.06594206e-02 -1.85368943e+00
1.31509349e-01 -6.15363181e-01 6.94985390e-01 5.56905508e-01
8.38078484e-02 -2.25663215e-01 3.80777866e-02 -4.50722761e-02
-4.80004698e-01 7.80003607e-01 1.31361985e+00 -1.17583096e-01
1.03836857e-01 -6.54879451e-01 -7.29670823e-01 2.23329410e-01
5.34646511e-01 -4.40457016e-01 -5.53491592e-01 -9.76873338e-01
8.43066394e-01 3.16034526e-01 1.51662722e-01 -8.56255412e-01
1.45533428e-01 -2.80784905e-01 2.46545613e-01 -3.27927023e-01
1.26073450e-01 -4.29660857e-01 3.15365285e-01 1.16711641e-02
-7.62511134e-01 -1.95751593e-01 1.75308898e-01 1.67106032e-01
9.19342190e-02 -5.99035382e-01 2.16840863e-01 -5.87897480e-01
-5.35915375e-01 1.40328348e-01 -1.99442968e-01 6.05469048e-01
4.72863823e-01 -4.82119620e-02 -1.00259614e+00 -9.81730074e-02
-6.11460328e-01 1.81636766e-01 6.61700428e-01 3.12486172e-01
3.72832000e-01 -1.44693363e+00 -1.08971536e+00 3.20196658e-01
5.98629653e-01 1.06266782e-01 5.64653166e-02 3.15492779e-01
-3.14001113e-01 6.90282941e-01 -7.60994032e-02 -9.84590426e-02
-4.93916780e-01 4.02303547e-01 1.72777921e-01 -5.11803746e-01
-5.78901291e-01 3.95383775e-01 -1.20791085e-01 -8.42409849e-01
-3.16836715e-01 -5.92590392e-01 2.12140501e-01 1.84800215e-02
2.73011744e-01 2.42371753e-01 3.47675085e-01 8.31983238e-02
3.08033437e-01 -7.24902973e-02 -4.31375094e-02 -3.05967003e-01
1.36192429e+00 -7.32179172e-03 -1.18650965e-01 2.47360855e-01
9.74655986e-01 5.86556550e-03 -8.68579268e-01 -2.47601002e-01
4.58218336e-01 -8.41708779e-02 -4.50557470e-01 -9.02806997e-01
-4.84365672e-01 6.35726869e-01 -2.89336890e-01 3.64508808e-01
5.45456529e-01 -5.40008992e-02 9.88975227e-01 9.84175563e-01
5.94220400e-01 -1.46596622e+00 8.64464492e-02 6.83743000e-01
9.12430704e-01 -9.63226676e-01 -1.84833601e-01 -2.11373672e-01
-7.62372613e-01 1.23022294e+00 7.41391957e-01 8.87908340e-02
1.55702978e-01 3.11260670e-01 -4.22116131e-01 -2.18274370e-01
-1.44895124e+00 -2.97009051e-01 8.29441771e-02 9.07526016e-01
5.80579937e-01 -1.42896876e-01 -1.24404892e-01 4.70580637e-01
-4.13110077e-01 5.38074732e-01 8.29244375e-01 9.10038769e-01
-5.33417940e-01 -1.58714056e+00 5.58537096e-02 9.06306446e-01
-1.48587659e-01 -4.98849213e-01 -1.77600339e-01 1.01004744e+00
2.87335273e-03 7.41239905e-01 1.14974827e-01 -1.70748726e-01
5.11470318e-01 4.81147915e-01 6.15765035e-01 -1.12944698e+00
-5.67829251e-01 -3.39962780e-01 1.17193711e+00 -2.91887045e-01
-2.64942497e-01 -7.45951951e-01 -1.38118005e+00 -4.11399484e-01
1.03369497e-01 2.82231659e-01 7.31481016e-01 7.89880395e-01
7.15538919e-01 4.07052517e-01 2.70374445e-03 -5.05224526e-01
-6.93077564e-01 -1.23046029e+00 -4.17763621e-01 7.47994661e-01
-3.14106494e-01 -1.82672396e-01 5.64054921e-02 2.93148041e-01]
|
[11.419304847717285, 8.777178764343262]
|
094b2805-b10e-4ea6-8980-0b08426e8b75
|
contrastive-learning-for-sleep-staging-based
|
2305.03178
| null |
https://arxiv.org/abs/2305.03178v1
|
https://arxiv.org/pdf/2305.03178v1.pdf
|
Contrastive Learning for Sleep Staging based on Inter Subject Correlation
|
In recent years, multitudes of researches have applied deep learning to automatic sleep stage classification. Whereas actually, these works have paid less attention to the issue of cross-subject in sleep staging. At the same time, emerging neuroscience theories on inter-subject correlations can provide new insights for cross-subject analysis. This paper presents the MViTime model that have been used in sleep staging study. And we implement the inter-subject correlation theory through contrastive learning, providing a feasible solution to address the cross-subject problem in sleep stage classification. Finally, experimental results and conclusions are presented, demonstrating that the developed method has achieved state-of-the-art performance on sleep staging. The results of the ablation experiment also demonstrate the effectiveness of the cross-subject approach based on contrastive learning.
|
['Bei Wang', 'Tongxu Zhang']
|
2023-05-05
| null | null | null | null |
['sleep-staging', 'automatic-sleep-stage-classification']
|
['medical', 'medical']
|
[-2.16600984e-01 -2.80759513e-01 -5.21851659e-01 -6.35402799e-01
-3.79454404e-01 4.22467962e-02 4.11222637e-01 -4.38232422e-02
-7.89362133e-01 7.56990790e-01 6.94509828e-03 -2.80440296e-03
-3.38478059e-01 -1.19404361e-01 3.73004074e-03 -7.32629180e-01
-3.80747944e-01 3.88510264e-02 4.03988222e-03 -8.13820362e-02
4.04516131e-01 3.92477103e-02 -1.77637935e+00 -2.63964720e-02
9.88162339e-01 1.03886700e+00 1.17138170e-01 3.93774927e-01
6.00621887e-02 2.32904911e-01 -9.42673028e-01 -1.89066857e-01
-8.09797794e-02 -7.51691580e-01 -6.68901145e-01 -2.77070999e-01
3.95788789e-01 3.39484066e-01 2.52318289e-02 8.92132461e-01
7.78431058e-01 2.46560574e-01 5.55578887e-01 -1.21555328e+00
-6.61600471e-01 3.25499952e-01 -8.90829444e-01 1.32613063e+00
2.50882894e-01 -1.51180714e-01 1.07564390e+00 -5.37886262e-01
-7.41424039e-02 7.29525566e-01 6.78247631e-01 8.23815703e-01
-1.12185919e+00 -1.07062125e+00 -1.49865776e-01 5.93937695e-01
-1.28666341e+00 -4.45365250e-01 7.76832819e-01 -2.94591725e-01
8.15829456e-01 2.01325387e-01 1.20454681e+00 1.02146375e+00
9.74301696e-01 7.41354346e-01 1.70659411e+00 -3.56971651e-01
2.93493092e-01 1.49715811e-01 7.08374441e-01 6.26607358e-01
2.41900712e-01 1.59670189e-01 -9.65154052e-01 4.91856873e-01
2.37522870e-01 -6.01317771e-02 4.22629528e-02 -3.30949612e-02
-7.44230688e-01 5.87876916e-01 4.46627796e-01 9.63658690e-01
5.03004268e-02 -1.09385513e-01 4.83216286e-01 2.42826506e-01
9.47190106e-01 3.87541175e-01 -3.52700710e-01 -1.06961414e-01
-1.75475085e+00 -2.20119581e-02 4.41642761e-01 3.55120629e-01
5.29897213e-01 8.36442783e-02 -4.08900261e-01 7.49174356e-01
4.75170702e-01 4.09634203e-01 9.62812126e-01 -4.58261102e-01
4.76684459e-02 6.16274714e-01 -4.40929949e-01 -7.32318401e-01
-1.18916249e+00 -9.15213585e-01 -9.87104595e-01 2.04567045e-01
2.15646699e-01 1.25522211e-01 -5.99897802e-01 1.65577257e+00
-1.45142481e-01 2.84521341e-01 -1.06862403e-01 8.30015540e-01
9.32897091e-01 1.79396659e-01 3.40877056e-01 -4.05454636e-01
1.73017275e+00 -9.01657462e-01 -1.13153040e+00 -5.16017497e-01
2.00349718e-01 -4.39648688e-01 1.12992167e+00 4.04845655e-01
-1.20406640e+00 -9.95930493e-01 -1.25896907e+00 -2.20734000e-01
-3.44635844e-01 3.67917418e-01 8.15527320e-01 8.89191985e-01
-1.36864161e+00 5.54726958e-01 -1.11718059e+00 -6.73361659e-01
6.64385796e-01 8.16551566e-01 -1.01020858e-01 6.12202048e-01
-1.11398435e+00 9.87010777e-01 -2.11128537e-02 2.10937649e-01
-7.54423618e-01 -7.72293687e-01 -3.81381899e-01 1.81634381e-01
-2.45820522e-01 -8.27262580e-01 1.07345736e+00 -8.23726475e-01
-1.19524407e+00 1.40853584e+00 -5.55136144e-01 -4.26600844e-01
-1.26845270e-01 -3.00085187e-01 -6.31810844e-01 -3.73223089e-02
2.97004521e-01 4.98373508e-01 6.11844897e-01 -7.15297580e-01
-5.47421157e-01 -6.31792665e-01 -1.27591625e-01 1.92986056e-01
-5.36777139e-01 1.26091048e-01 -1.49741352e-01 -2.58894831e-01
2.72738338e-02 -8.33817005e-01 1.89058006e-01 -9.03724059e-02
4.12519090e-02 -6.38666332e-01 4.48505104e-01 -3.03633630e-01
1.44482350e+00 -2.19031334e+00 2.98716217e-01 -2.82076597e-01
4.90800679e-01 8.01122040e-02 2.38494545e-01 7.55745545e-02
-1.52376175e-01 6.23938069e-02 -1.65060177e-01 -8.60164165e-01
-2.18489528e-01 1.81336582e-01 1.73155516e-01 8.38463664e-01
-1.52094394e-01 9.24057961e-01 -6.82798624e-01 -6.24649882e-01
1.35663031e-02 1.30328417e-01 -4.48863357e-01 1.26948118e-01
5.95780492e-01 5.84318399e-01 -1.49969816e-01 4.13030297e-01
5.63433051e-01 -1.49854660e-01 -1.00672305e-01 -2.07545996e-01
-4.06717956e-01 2.13902146e-01 -3.32150072e-01 1.94012523e+00
-5.40148795e-01 1.09141576e+00 -1.93764716e-01 -1.06123328e+00
6.75209522e-01 3.35018570e-03 3.95789266e-01 -9.87442374e-01
4.53297555e-01 -4.44336236e-02 5.18788695e-01 -3.77582371e-01
4.75257576e-01 -5.85581422e-01 -6.04211800e-02 3.85270506e-01
4.20153797e-01 1.88527673e-01 2.77840078e-01 -1.85224697e-01
8.32706094e-01 -1.50185287e-01 6.03236794e-01 -8.79203498e-01
7.66289830e-01 -5.20767689e-01 5.12534738e-01 4.81001288e-01
-7.11090028e-01 1.62047938e-01 4.97311682e-01 -1.68181583e-01
-4.55003351e-01 -1.15671551e+00 -2.59923995e-01 1.06308293e+00
1.80026174e-01 -4.04252350e-01 -7.95966804e-01 -4.14192349e-01
-4.94965553e-01 3.74497026e-01 -1.06579828e+00 -4.67993826e-01
-1.29760414e-01 -1.27358162e+00 6.08329892e-01 4.85322654e-01
8.00369143e-01 -1.14236856e+00 -7.91384995e-01 -4.72563177e-01
-6.52024597e-02 -8.65203083e-01 -1.07192233e-01 5.69921315e-01
-1.28669238e+00 -9.39678550e-01 -5.58297992e-01 -9.19563890e-01
5.17628372e-01 5.43867469e-01 1.20815468e+00 2.07183644e-01
-4.56847340e-01 3.12619805e-01 -2.50141203e-01 -3.50410461e-01
8.55642781e-02 5.20593941e-01 3.71832758e-01 -2.08315805e-01
7.31015623e-01 -9.07795727e-01 -9.80486214e-01 2.09525272e-01
-5.01258790e-01 -2.95328677e-01 7.67987430e-01 7.72953868e-01
2.69127309e-01 -2.58362442e-02 7.12942004e-01 -5.97638607e-01
8.07741940e-01 -5.13981998e-01 -3.98541987e-01 -8.69981200e-02
-1.08630276e+00 -9.30674560e-03 4.33285564e-01 -1.46802515e-01
-9.58197773e-01 -7.44659781e-01 -1.61627978e-01 -1.51207983e-01
-1.18641749e-01 3.90032530e-01 -8.70278664e-03 -2.69461684e-02
6.53446317e-01 4.58006382e-01 -2.54553743e-02 -3.08109373e-01
-8.84285271e-02 5.42322338e-01 3.19727957e-01 -8.48459601e-02
4.69742686e-01 5.13238549e-01 2.06890777e-01 -1.03637230e+00
-1.36500263e+00 -7.46227920e-01 -8.99420440e-01 -3.31631571e-01
1.39814758e+00 -1.03644013e+00 -8.69158387e-01 4.99728501e-01
-4.33590561e-01 -3.65885168e-01 -2.04096884e-01 4.70361143e-01
-4.74109650e-01 2.89727718e-01 -3.43889475e-01 -7.64355183e-01
-6.04314208e-01 -8.97702992e-01 9.40969288e-01 7.26336598e-01
-3.89733016e-01 -1.39855647e+00 4.55108553e-01 5.80342650e-01
2.56203324e-01 -5.24996996e-01 7.30428994e-01 -5.00586212e-01
-9.04773846e-02 8.01745206e-02 1.84650719e-03 1.88722998e-01
2.01104313e-01 -4.18496788e-01 -1.24963224e+00 -2.46781349e-01
3.36303592e-01 -1.26916215e-01 8.10042024e-01 6.83173060e-01
9.85719323e-01 3.33863497e-01 -3.31927508e-01 6.97685361e-01
1.07268262e+00 7.77782053e-02 6.96107388e-01 3.89975250e-01
2.81921446e-01 4.55243289e-01 4.28941250e-01 2.87093878e-01
4.77007300e-01 6.54985666e-01 1.04394928e-01 -2.58336365e-01
-2.93666273e-01 3.59446496e-01 3.38518828e-01 1.04269671e+00
-5.40453792e-02 -1.52500244e-02 -6.87961876e-01 3.56690109e-01
-1.54607463e+00 -1.17644250e+00 -3.50711346e-01 1.83011818e+00
5.26292026e-01 1.76721975e-01 2.79959619e-01 1.99107662e-01
2.25861475e-01 2.84593582e-01 -4.20396894e-01 -4.02380168e-01
-4.32512164e-02 6.06651545e-01 1.29813641e-01 5.44889085e-02
-1.15147269e+00 9.70845222e-01 7.30305481e+00 7.16195881e-01
-1.11690092e+00 5.44644952e-01 4.12293822e-01 -2.57644802e-01
3.84939522e-01 -3.19275945e-01 -8.51428866e-01 6.64956570e-01
1.18446660e+00 -1.39442340e-01 3.66228789e-01 5.45797527e-01
6.31039679e-01 -8.53596985e-01 -8.90166104e-01 1.14713013e+00
3.43598276e-01 -7.91479707e-01 -7.70382404e-01 1.73763618e-01
6.83633924e-01 -1.54778035e-02 3.39394152e-01 5.10171533e-01
-6.18781507e-01 -7.35184014e-01 3.39445055e-01 7.23553658e-01
4.60937679e-01 -5.49735904e-01 9.25131798e-01 2.83030450e-01
-1.07232511e+00 -1.42390862e-01 -4.88324821e-01 -4.96746451e-01
-2.06005290e-01 5.26453257e-01 -3.35752636e-01 3.56142521e-01
1.02373767e+00 1.06097317e+00 -1.27411616e+00 1.22990656e+00
-3.24703455e-01 8.20038021e-01 2.12348223e-01 -2.61980474e-01
-1.31262898e-01 -2.75322199e-01 4.92481925e-02 1.01246452e+00
2.28608176e-01 -2.09547684e-01 -2.56188452e-01 9.17578876e-01
2.10594013e-01 -9.60085839e-02 -2.78267294e-01 1.62474424e-01
-8.63655061e-02 1.67000234e+00 -1.20753014e+00 -7.56591931e-02
-5.12903452e-01 9.11040664e-01 3.24498743e-01 -3.56488600e-02
-7.34496772e-01 -2.96977796e-02 7.77835071e-01 -7.18849897e-02
5.75679317e-02 -2.53870308e-01 -7.56353199e-01 -1.12474787e+00
-2.93868065e-01 -4.22318399e-01 3.08634222e-01 -6.43462598e-01
-1.42550027e+00 4.00293410e-01 3.06316584e-01 -9.32685554e-01
2.09892228e-01 -4.06997263e-01 -9.74937081e-01 7.81904578e-01
-1.46056104e+00 -9.20283973e-01 -2.48319030e-01 4.24350083e-01
6.94833040e-01 -3.09423864e-01 7.36031532e-01 5.61027169e-01
-8.30351830e-01 7.44633317e-01 -1.62353702e-02 -3.06690037e-01
6.72082126e-01 -1.39619899e+00 7.22411647e-02 8.43183517e-01
1.42101973e-01 9.92955863e-01 6.61555469e-01 -3.63524616e-01
-9.41196322e-01 -4.65154082e-01 9.43748653e-01 -6.50956035e-01
7.43188262e-01 -5.88538170e-01 -7.61890233e-01 2.21430346e-01
6.40939593e-01 -1.23327434e-01 1.35792089e+00 8.08415711e-01
2.43842930e-01 -5.47350049e-01 -9.29797947e-01 5.59214413e-01
9.14646149e-01 -5.27187169e-01 -9.48299587e-01 -5.84582314e-02
-2.18288898e-02 2.32351385e-02 -7.02243328e-01 3.29504609e-02
6.94338620e-01 -1.39617634e+00 6.85254991e-01 -2.82280862e-01
3.77977341e-01 -1.15985468e-01 3.27724874e-01 -1.20525682e+00
-4.05691385e-01 -2.36790210e-01 -9.03970823e-02 1.22762644e+00
1.23525739e-01 -6.79211989e-02 9.29043174e-01 2.95600832e-01
-5.06936729e-01 -8.41353297e-01 -1.15410101e+00 -7.20794797e-01
1.01464413e-01 -2.42467552e-01 -1.54336959e-01 5.40621459e-01
2.14688092e-01 8.24695885e-01 -4.15760577e-01 -1.80753678e-01
4.29568917e-01 5.19413725e-02 3.67056847e-01 -1.46791816e+00
9.53432638e-03 -5.14434695e-01 -5.29708803e-01 -6.01484656e-01
5.92739999e-01 -9.84764695e-01 1.51510328e-01 -1.50373101e+00
6.73308730e-01 -8.30277335e-03 -8.12493622e-01 2.78172255e-01
-4.84668076e-01 7.02281296e-01 8.88261795e-02 1.84878573e-01
-8.86506259e-01 6.54725373e-01 1.11243522e+00 1.34590462e-01
-3.01371664e-01 2.88488805e-01 -9.78816390e-01 6.74242616e-01
1.17802691e+00 -6.18498445e-01 -5.80453157e-01 1.46080013e-02
2.91915953e-01 -3.55794042e-01 2.44490951e-01 -1.63132167e+00
2.84366548e-01 2.93739468e-01 5.90859294e-01 -5.42875111e-01
4.17114556e-01 -5.83408237e-01 -2.95892388e-01 5.48539162e-01
-2.98284292e-01 3.71914566e-01 3.47303927e-01 4.56043184e-01
1.05640916e-02 -3.07716280e-01 9.04172301e-01 -3.75050418e-02
-6.00463569e-01 2.01135017e-02 -8.02634537e-01 1.96949869e-01
6.37015939e-01 -3.23899508e-01 -1.94933683e-01 2.28325631e-02
-7.23711848e-01 1.18292116e-01 9.97015275e-03 3.34214985e-01
3.70083243e-01 -9.10420597e-01 -7.83908814e-02 2.92052954e-01
-4.35227826e-02 -7.38982320e-01 5.36861300e-01 1.40446603e+00
-2.07656734e-02 4.63664860e-01 -6.82891726e-01 -6.97228551e-01
-1.45669115e+00 4.70197380e-01 3.95308346e-01 -4.96166706e-01
-2.43337870e-01 9.46132183e-01 3.84229362e-01 1.30545080e-01
-5.88199608e-02 -3.55730057e-01 -6.28259778e-01 6.70417130e-01
3.64667177e-01 5.49433112e-01 1.31101727e-01 -4.71620977e-01
-5.87841213e-01 7.64437437e-01 -3.39563414e-02 3.03100757e-02
1.23799682e+00 -4.45659965e-01 -3.04701477e-01 1.08908522e+00
1.03103483e+00 -1.48443906e-02 -7.77103901e-01 3.76166165e-01
1.50293887e-01 4.15801220e-02 4.00236189e-01 -9.04488206e-01
-1.02913117e+00 1.16383576e+00 1.30350840e+00 1.76462531e-01
1.37787032e+00 -1.39590129e-02 4.91209328e-01 3.14094305e-01
1.25088617e-01 -1.02581561e+00 2.77626336e-01 2.77045071e-01
3.25776786e-01 -1.34737611e+00 5.81556141e-01 1.59702748e-01
-5.51463723e-01 9.00252342e-01 7.32353032e-01 -3.53654623e-01
9.15501893e-01 -6.86821938e-02 1.14603139e-01 -4.17724222e-01
-5.01456320e-01 -4.77198064e-01 7.15219796e-01 5.21539330e-01
9.10227180e-01 -1.06819853e-01 -9.50920582e-01 8.78531575e-01
-5.28919816e-01 4.45624217e-02 3.11434776e-01 5.59536517e-01
-3.06970835e-01 -1.17196369e+00 7.16518890e-03 5.59128165e-01
-9.89644051e-01 -1.50999591e-01 -2.43636921e-01 1.03744948e+00
3.94834340e-01 1.09072292e+00 2.00622797e-01 -5.45004368e-01
1.84823666e-02 3.50868590e-02 6.74980104e-01 -7.91465342e-01
-1.02762341e+00 9.72682238e-03 -3.74135345e-01 -3.94401252e-01
-1.06024611e+00 -7.04991937e-01 -9.70016599e-01 -2.26931311e-02
-2.91396677e-01 4.16002661e-01 4.37353045e-01 1.19057465e+00
2.25402683e-01 9.28731799e-01 3.43211472e-01 -8.49238396e-01
1.67544752e-01 -1.19317365e+00 -9.10210609e-01 1.27119268e-03
4.37837303e-01 -9.32106793e-01 -3.63299608e-01 -2.32833430e-01]
|
[13.537264823913574, 3.5275843143463135]
|
4fa76e9a-c3fd-42a0-90c8-d70d60772ec1
|
isic-2017-skin-lesion-segmentation-using-deep
|
1807.09083
| null |
http://arxiv.org/abs/1807.09083v1
|
http://arxiv.org/pdf/1807.09083v1.pdf
|
ISIC 2017 Skin Lesion Segmentation Using Deep Encoder-Decoder Network
|
This paper summarizes our method and validation results for part 1 of the
ISBI Challenge 2018. Our algorithm makes use of deep encoder-decoder network
and novel skin lesion data augmentation to segment the challenge objective.
Besides, we also propose an effective testing strategy by applying multi-model
comparison.
|
['Ngoc-Quang Nguyen']
|
2018-07-24
| null | null | null | null |
['skin-lesion-segmentation']
|
['medical']
|
[ 6.28981769e-01 1.43298581e-01 -8.77810359e-01 -3.49620581e-01
-1.51193237e+00 -1.10742785e-01 4.16711211e-01 -1.27126977e-01
-6.82946980e-01 8.01613510e-01 1.61862209e-01 -3.66021693e-02
2.92901218e-01 -2.98509330e-01 -7.29725420e-01 -3.56876105e-01
6.89364895e-02 4.34388131e-01 2.81990349e-01 -5.79875968e-02
-3.83401476e-02 -1.78680718e-01 -7.60582447e-01 9.20272946e-01
8.41422319e-01 1.14850616e+00 -5.88952377e-02 8.46795321e-01
3.03536147e-01 6.99225128e-01 -7.44879663e-01 -9.55977559e-01
5.30379117e-02 -6.05338871e-01 -1.12006879e+00 -1.33879147e-02
6.51047885e-01 -5.10522783e-01 -5.12172103e-01 8.90030086e-01
1.10375440e+00 -1.13971807e-01 6.26299679e-01 -1.08524990e+00
-3.93025041e-01 4.54844087e-01 -6.44972026e-01 5.41705668e-01
1.05256684e-01 2.94218604e-02 6.82320058e-01 -4.69059318e-01
8.56519639e-01 6.81725264e-01 9.26690340e-01 1.11282468e+00
-7.00112104e-01 -4.75542605e-01 -6.40798137e-02 8.42688620e-01
-1.36165369e+00 -1.09033734e-01 5.86849451e-01 -1.11764923e-01
1.02269185e+00 5.40128469e-01 6.64058685e-01 1.86147702e+00
1.32840857e-01 1.44650900e+00 1.32588828e+00 -2.72925675e-01
-2.22029850e-01 -7.02168718e-02 9.48780030e-02 6.87985659e-01
-5.09373732e-02 7.28988126e-02 -3.16868424e-01 7.11269677e-02
4.28091824e-01 -5.03816605e-01 -1.07227586e-01 5.00049472e-01
-7.77830124e-01 7.21768320e-01 5.19515276e-01 -4.61632945e-02
-2.02948049e-01 3.24253768e-01 9.59322393e-01 1.08191438e-01
7.04635203e-01 6.15859367e-02 -3.16131979e-01 -1.22950502e-01
-1.05954039e+00 -1.20663263e-01 4.55351919e-01 4.74177182e-01
-2.25904584e-01 -3.57164919e-01 -7.03374505e-01 1.26793790e+00
2.02075705e-01 1.03192352e-01 4.73929763e-01 -4.70650166e-01
5.75133502e-01 2.59957254e-01 -6.17651582e-01 -2.84647882e-01
-3.24649245e-01 -8.35872769e-01 -8.21990252e-01 -2.06104547e-01
-1.76916987e-01 -2.18431860e-01 -1.42540824e+00 1.45039248e+00
8.93412381e-02 5.11482060e-01 -3.10334027e-01 7.45715082e-01
1.06890488e+00 1.34484723e-01 2.31961384e-01 1.44734547e-01
1.26857972e+00 -1.63505268e+00 -8.48822474e-01 -1.64348990e-01
8.69429410e-01 -4.52335984e-01 7.14363337e-01 7.03540623e-01
-1.22589326e+00 -3.61477345e-01 -1.10765052e+00 -8.45850632e-02
-1.91327870e-01 3.26344937e-01 4.84773099e-01 7.02242911e-01
-1.05306458e+00 2.66362261e-02 -8.18337381e-01 -2.11387172e-01
6.56788588e-01 3.43645722e-01 -6.37757242e-01 -3.56049836e-01
-1.45818484e+00 1.42764199e+00 4.24609005e-01 1.49736390e-01
-9.72820461e-01 -6.24459505e-01 -8.85759711e-01 -6.04227245e-01
1.81314722e-01 -6.06090009e-01 1.41060054e+00 -8.26854587e-01
-1.45817566e+00 1.34618115e+00 -1.46854922e-01 -7.26675689e-01
6.06738865e-01 -2.01520458e-01 -7.31370330e-01 5.47624528e-01
-9.46958885e-02 7.56816745e-01 2.72624642e-01 -7.13317275e-01
-3.66764694e-01 1.72734621e-03 -1.32147789e-01 1.47613496e-01
-2.27771878e-01 3.97685945e-01 -1.04433572e+00 -6.77869558e-01
-5.16667068e-01 -7.58244812e-01 -4.18338388e-01 -7.74910301e-02
-9.40201223e-01 -1.77868024e-01 5.73301725e-02 -1.41993284e+00
1.39657116e+00 -1.90637994e+00 -2.81267092e-02 3.75817209e-01
3.74213457e-01 4.42796886e-01 -4.73829985e-01 1.33626357e-01
-3.10491472e-01 3.82795006e-01 -1.68862343e-01 -5.37313938e-01
-2.22232133e-01 -6.28705993e-02 3.03121239e-01 5.50507367e-01
4.51844990e-01 1.14014089e+00 -5.76153338e-01 -6.84851408e-01
1.60866618e-01 4.83369499e-01 -6.32903695e-01 5.57099655e-02
7.68494606e-02 3.07121664e-01 -1.11742482e-01 9.29507315e-01
6.74703538e-01 -2.28945717e-01 -3.58664840e-02 -5.62308788e-01
2.70382792e-01 4.67865080e-01 -4.82208312e-01 2.09256506e+00
-4.32380676e-01 5.25881410e-01 8.53653103e-02 -1.04922593e+00
3.09245557e-01 4.88259196e-01 5.44325948e-01 -9.85787332e-01
2.06652969e-01 1.69850543e-01 1.95420235e-01 -7.88826883e-01
-1.03466757e-01 -9.82729867e-02 3.00867796e-01 -8.51155370e-02
3.43699962e-01 9.79121327e-02 3.34219754e-01 -4.74559441e-02
1.25228524e+00 9.76938158e-02 2.75396973e-01 1.58159345e-01
4.86328125e-01 -1.32150382e-01 5.46245456e-01 4.77839231e-01
-6.25248671e-01 9.53654051e-01 6.02678061e-01 -1.54761011e-02
-8.02413166e-01 -9.99113917e-01 -5.36303341e-01 6.21457875e-01
-9.37187746e-02 -5.15434504e-01 -7.26684153e-01 -1.14603770e+00
-3.88176501e-01 5.23220658e-01 -1.34062099e+00 -2.53716618e-01
-2.89630890e-01 -1.16973722e+00 1.11467409e+00 6.41722202e-01
6.50474608e-01 -6.34996176e-01 8.41737092e-02 -5.34886084e-02
-5.33268034e-01 -1.07159376e+00 -4.73803967e-01 1.85249429e-02
-4.15835917e-01 -1.57550728e+00 -9.59480286e-01 -9.80434954e-01
4.42913502e-01 -1.33562297e-01 8.81062627e-01 4.93712258e-03
-8.26248586e-01 3.08954358e-01 -3.58732998e-01 -5.13972700e-01
-3.09576571e-01 2.82923043e-01 -4.93815005e-01 -2.28100181e-01
3.42233270e-01 4.65314761e-02 -8.76979351e-01 2.72981524e-01
-9.05555189e-01 4.15539205e-01 8.89976382e-01 1.00593960e+00
7.44247675e-01 -5.94214737e-01 6.66417599e-01 -8.68829489e-01
6.13158941e-01 -6.92499995e-01 1.45671040e-01 8.22600663e-01
-3.00057948e-01 -4.29954439e-01 -1.02126710e-01 -2.73862630e-01
-7.84567297e-01 -2.83292532e-01 -1.06835639e+00 -3.70878875e-02
6.40026107e-02 5.69904089e-01 3.85536738e-02 -3.51133257e-01
6.75940812e-01 1.47962451e-01 1.44499745e-02 -5.43543220e-01
2.10495517e-01 8.47185612e-01 3.82467508e-01 -1.55451640e-01
2.84309417e-01 2.14378864e-01 -2.91436553e-01 -4.85692263e-01
-9.67675328e-01 -4.80663687e-01 -5.61500609e-01 -3.82806987e-01
9.86459851e-01 -8.77852321e-01 -3.65963906e-01 7.32911348e-01
-1.12241495e+00 -5.21905720e-01 5.17630652e-02 5.28140664e-01
-3.81420374e-01 3.69110405e-01 -9.37406182e-01 -4.88625944e-01
-8.18616867e-01 -1.25700986e+00 9.69607472e-01 -1.62953623e-02
-1.28465563e-01 -1.07375956e+00 5.48462451e-01 8.34677994e-01
3.94995928e-01 4.40633118e-01 3.63274515e-01 -9.13418949e-01
2.67020166e-01 -7.44286478e-01 -3.41129839e-01 8.19431663e-01
-9.65954270e-03 5.31165935e-02 -1.00689781e+00 -3.43856990e-01
-2.71991670e-01 -9.35782731e-01 1.43187296e+00 4.33171511e-01
1.89732194e+00 -7.72778317e-02 -5.07769465e-01 9.23276305e-01
1.28705180e+00 -1.36794060e-01 9.60817277e-01 3.67200643e-01
4.76438820e-01 5.77238858e-01 3.67756009e-01 -2.04233900e-01
6.28900468e-01 7.11744308e-01 4.26721781e-01 -5.68143308e-01
-6.52783215e-01 -2.67485343e-02 2.20684499e-01 7.71501064e-01
-1.81963980e-01 -2.49397606e-01 -9.87556219e-01 7.25226283e-01
-1.29450369e+00 -7.37151563e-01 -1.38323694e-01 1.76151884e+00
1.22835553e+00 1.61608413e-01 2.66311109e-01 3.92570719e-02
6.31534159e-01 -5.13314595e-03 -4.68891114e-01 -4.84641492e-01
-3.08819771e-01 8.57256651e-01 4.87311393e-01 3.79411876e-01
-1.43215168e+00 6.95606649e-01 8.25717068e+00 1.28792465e+00
-1.07158804e+00 7.19477057e-01 9.48391199e-01 -4.32194799e-01
-1.49394125e-01 -7.85330057e-01 -6.30225360e-01 5.01643896e-01
1.10712612e+00 2.44314522e-01 -5.77318408e-02 4.60832477e-01
-3.35904241e-01 -1.06144287e-01 -7.47363389e-01 7.90572107e-01
6.82709217e-01 -1.42166162e+00 -3.56084287e-01 -2.39034608e-01
7.02529371e-01 6.97107732e-01 2.96057612e-01 3.48915875e-01
-1.38766423e-01 -1.37083542e+00 1.67106375e-01 3.93384218e-01
1.36303592e+00 -3.94812047e-01 9.93917465e-01 -1.92027107e-01
-7.28878856e-01 2.58303255e-01 1.69447869e-01 5.42498946e-01
-3.35324779e-02 3.05244237e-01 -1.00217199e+00 7.50120640e-01
4.09541309e-01 6.17337525e-01 -6.93412304e-01 1.70067549e+00
-5.91237605e-01 1.12681508e+00 -1.71040833e-01 1.07066877e-01
1.94299191e-01 3.43840420e-01 2.46365979e-01 1.72013247e+00
-1.31519675e-01 -3.58081400e-01 -2.35578567e-02 3.27704936e-01
-2.05188751e-01 9.95919630e-02 -3.00429434e-01 8.02829489e-02
-3.53191011e-02 1.24910474e+00 -1.45399794e-01 -1.69996202e-01
-5.70648372e-01 1.29555821e+00 3.91623765e-01 2.42263362e-01
-1.04792488e+00 -4.24013555e-01 3.69064838e-01 -1.66975379e-01
-2.27371469e-01 3.00899781e-02 -3.51951331e-01 -1.16173840e+00
-2.28208005e-01 -9.34040248e-01 6.68202877e-01 -4.26605105e-01
-1.35067987e+00 7.04035521e-01 -2.19516665e-01 -1.05923080e+00
-1.53179690e-01 -7.70769060e-01 -8.09244931e-01 8.67315173e-01
-1.95109189e+00 -1.70059335e+00 -1.59777790e-01 5.24481714e-01
4.97559637e-01 -1.05695643e-01 9.75831509e-01 7.15583205e-01
-1.14109588e+00 1.43126547e+00 -7.77551485e-03 4.08737004e-01
9.74802077e-01 -1.18095601e+00 6.48799360e-01 7.44020104e-01
-2.18371749e-01 2.04459995e-01 7.42506385e-02 -7.05605090e-01
-8.38823140e-01 -1.17583752e+00 6.58897936e-01 -4.33746010e-01
8.69960189e-01 -3.86191130e-01 -5.24076343e-01 5.28120100e-01
5.49431920e-01 -1.90067187e-01 1.42983115e+00 1.95415676e-01
-3.53264302e-01 5.11509888e-02 -1.39828801e+00 5.20927906e-01
8.20233643e-01 -5.69668710e-01 -3.10353190e-01 7.29785740e-01
6.46285892e-01 -7.58485854e-01 -9.44498122e-01 7.70079911e-01
4.98875707e-01 -3.18845570e-01 8.96677792e-01 -1.13937974e+00
8.75757396e-01 3.12265277e-01 -3.39980200e-02 -1.31950593e+00
-2.22600833e-01 -3.78309637e-01 -3.80340815e-01 8.46160650e-01
6.78519547e-01 -2.86112398e-01 8.11871588e-01 1.75378308e-01
-3.58412743e-01 -1.34871757e+00 -1.14599276e+00 -4.61065024e-01
2.01461047e-01 -7.37830102e-01 -2.07150131e-01 7.63751745e-01
-7.97914788e-02 -3.25984582e-02 -5.47308981e-01 -2.29959428e-01
6.80322170e-01 -6.16034448e-01 2.55367100e-01 -3.90682161e-01
-3.90081376e-01 -5.13152957e-01 -4.86671776e-01 -6.68807983e-01
2.16070749e-02 -1.29016161e+00 -1.39407203e-01 -1.76840580e+00
5.73929906e-01 8.96357372e-02 -9.27251399e-01 6.66031003e-01
-4.11802649e-01 1.05073714e+00 1.32820266e-03 -3.00948918e-01
-9.09891248e-01 2.80909091e-01 1.27157331e+00 -4.03101921e-01
4.61297780e-01 1.47795215e-01 -7.51726508e-01 3.47923160e-01
1.05647433e+00 -4.89275865e-02 -1.15088992e-01 -7.26976871e-01
1.63721725e-01 -4.56380069e-01 4.15571272e-01 -8.99683714e-01
6.16195612e-02 -1.81027856e-02 5.40077627e-01 -7.14837015e-01
4.76223618e-01 -3.03040862e-01 -4.46121931e-01 8.86616886e-01
-5.63933909e-01 -3.10573310e-01 5.78062713e-01 2.99995154e-01
-3.01085502e-01 -1.70637146e-01 8.48343432e-01 2.83187687e-01
-6.69596136e-01 4.93532091e-01 -2.06397787e-01 1.34145290e-01
1.16307592e+00 4.24344502e-02 -6.08683884e-01 1.79816056e-02
-7.56600797e-01 7.59823263e-01 -1.99410766e-01 4.15680707e-01
7.56707013e-01 -1.38760686e+00 -1.28886163e+00 8.25062394e-02
3.94311637e-01 -5.39492965e-01 5.52290559e-01 1.40987778e+00
-7.71466374e-01 4.75360870e-01 -4.14539158e-01 -3.16536665e-01
-1.38242817e+00 5.28881699e-02 6.95897222e-01 -6.55579209e-01
-2.17126027e-01 1.55663121e+00 6.73565641e-02 -3.65506351e-01
3.77960652e-01 -8.81039444e-03 -2.56957293e-01 -1.78465948e-01
9.24055576e-01 3.75563920e-01 1.98341802e-01 -7.85403922e-02
-4.49476451e-01 7.14459196e-02 -3.09955269e-01 -1.80418789e-01
1.13154578e+00 2.99202263e-01 -1.45814270e-01 1.08144730e-01
1.45765686e+00 -2.76898921e-01 -8.54970872e-01 -6.57668188e-02
-5.36472313e-02 -1.18024610e-01 2.08063155e-01 -1.73512042e+00
-1.07610667e+00 1.06413960e+00 8.98698449e-01 -3.30273181e-01
1.32151377e+00 -5.61005250e-02 1.05730891e+00 -1.74223259e-02
-2.30742823e-02 -1.26677251e+00 4.86527458e-02 3.77135217e-01
1.00820446e+00 -1.51127470e+00 -3.94893646e-01 -4.85499263e-01
-7.85164535e-01 8.44205320e-01 9.48564172e-01 -1.25795722e-01
5.30154049e-01 4.50796068e-01 6.40205592e-02 -6.16872981e-02
-9.62994695e-01 -3.05070758e-01 9.38727796e-01 7.86709487e-01
6.83882296e-01 4.40697558e-02 -9.49613512e-01 1.07232082e+00
8.58243778e-02 8.60279426e-02 -9.63544846e-02 4.70289826e-01
-6.90026069e-03 -1.33967364e+00 2.79952973e-01 8.59090984e-01
-9.59207594e-01 -3.40227902e-01 -7.34763980e-01 8.49355698e-01
2.05799416e-01 6.70059323e-01 -6.00005165e-02 -8.97512376e-01
8.85184556e-02 2.01024991e-02 8.52017641e-01 -7.15904295e-01
-7.27581203e-01 3.79210226e-02 5.84964335e-01 -5.95191717e-01
-2.02683553e-01 -1.64883584e-01 -9.02783334e-01 3.91647667e-02
-2.05951333e-01 -2.31410652e-01 9.88943577e-01 9.86432314e-01
8.40695202e-02 8.99914682e-01 5.81440091e-01 -5.10652699e-02
-4.71993685e-01 -1.42326474e+00 -3.21591049e-01 2.80656904e-01
3.34066063e-01 -4.58399534e-01 9.41342637e-02 -3.86316925e-01]
|
[15.663786888122559, -2.950857162475586]
|
5d3cde83-7192-416b-9b5a-5b6f6a2cfd68
|
bytesing-a-chinese-singing-voice-synthesis
|
2004.11012
| null |
https://arxiv.org/abs/2004.11012v1
|
https://arxiv.org/pdf/2004.11012v1.pdf
|
ByteSing: A Chinese Singing Voice Synthesis System Using Duration Allocated Encoder-Decoder Acoustic Models and WaveRNN Vocoders
|
This paper presents ByteSing, a Chinese singing voice synthesis (SVS) system based on duration allocated Tacotron-like acoustic models and WaveRNN neural vocoders. Different from the conventional SVS models, the proposed ByteSing employs Tacotron-like encoder-decoder structures as the acoustic models, in which the CBHG models and recurrent neural networks (RNNs) are explored as encoders and decoders respectively. Meanwhile an auxiliary phoneme duration prediction model is utilized to expand the input sequence, which can enhance the model controllable capacity, model stability and tempo prediction accuracy. WaveRNN neural vocoders are also adopted as neural vocoders to further improve the voice quality of synthesized songs. Both objective and subjective experimental results prove that the SVS method proposed in this paper can produce quite natural, expressive and high-fidelity songs by improving the pitch and spectrogram prediction accuracy and the models using attention mechanism can achieve best performance.
|
['Yu Gu', 'Yuan Wan', 'Yuxuan Wang', 'Yang Zhang', 'Benlai Tang', 'Zejun Ma', 'Yonghui Rao', 'Xiang Yin', 'Jitong Chen']
|
2020-04-23
| null | null | null | null |
['singing-voice-synthesis']
|
['speech']
|
[-2.39140049e-01 -7.19973668e-02 -1.79858521e-01 1.38654366e-01
-1.83470309e-01 -1.57149896e-01 8.48654285e-02 -7.93361664e-01
6.93050250e-02 6.44980907e-01 7.15839684e-01 -2.31545255e-01
7.79773518e-02 -3.76006693e-01 -2.63722092e-01 -5.69281757e-01
3.86233360e-01 -5.04566766e-02 2.74799932e-02 -2.77344376e-01
1.55070469e-01 1.04556181e-01 -1.73442018e+00 6.54971600e-03
8.02854121e-01 9.74874198e-01 7.78805971e-01 1.18223560e+00
1.29979877e-02 1.05733693e+00 -6.70354366e-01 2.30677649e-01
-9.69252810e-02 -8.55889022e-01 -4.53289598e-01 -2.90909857e-01
-3.48314017e-01 -2.54144788e-01 -8.75806034e-01 6.66085005e-01
1.09964371e+00 5.45036435e-01 4.79600877e-01 -6.25520170e-01
-1.11951387e+00 1.12110531e+00 1.78088740e-01 2.22050548e-01
6.22876287e-02 2.46948197e-01 1.01802909e+00 -8.32405627e-01
2.12062538e-01 1.07535660e+00 7.82281220e-01 9.56630647e-01
-5.33071220e-01 -7.10851669e-01 -4.55193609e-01 6.74966812e-01
-1.15392900e+00 -7.48541653e-01 1.04371321e+00 -1.58342049e-02
1.16736996e+00 7.33160317e-01 8.79687786e-01 7.71422625e-01
1.66849494e-01 8.81761074e-01 5.48601091e-01 -6.65615797e-01
-2.22046986e-01 1.33572534e-01 -1.91130564e-01 8.93182978e-02
-8.31277668e-01 5.51452458e-01 -6.59169853e-01 4.65615779e-01
9.37085330e-01 -3.16409916e-01 -4.75255609e-01 5.98075509e-01
-1.00168622e+00 6.30281866e-01 1.93775222e-01 5.62368929e-01
-5.65799236e-01 2.28563890e-01 7.15791881e-01 3.21084052e-01
1.76897988e-01 7.37843990e-01 -3.49734932e-01 -6.64026082e-01
-1.06916177e+00 3.61622870e-03 4.97631639e-01 9.52483952e-01
-1.01637222e-01 1.30775821e+00 -4.17029023e-01 1.30670381e+00
1.28663495e-01 5.01552165e-01 1.35677302e+00 -1.03264093e+00
2.30551645e-01 1.69450238e-01 -1.39355406e-01 -6.67214453e-01
-1.78546906e-01 -7.00917423e-01 -7.02665031e-01 -3.40436459e-01
-3.90540212e-01 -4.64256674e-01 -5.99701941e-01 1.56901169e+00
-2.44586244e-02 2.70127922e-01 1.44039288e-01 9.21717644e-01
1.20817769e+00 1.50178945e+00 -2.95348555e-01 -6.92234099e-01
8.80201817e-01 -1.53551531e+00 -1.60496712e+00 3.46821398e-01
2.04911590e-01 -9.73611593e-01 1.35559285e+00 3.38831425e-01
-1.47476113e+00 -1.26107407e+00 -1.00137269e+00 -2.37330005e-01
-6.36328831e-02 5.72142899e-01 -2.70181987e-02 6.20848119e-01
-8.97298157e-01 9.31496799e-01 -5.24007022e-01 3.14428568e-01
-3.36016178e-01 4.08762544e-01 3.48570496e-01 8.73575687e-01
-1.45855141e+00 7.04491556e-01 5.05396962e-01 4.61320966e-01
-1.01953840e+00 -6.52044833e-01 -4.71720070e-01 3.07631850e-01
9.58766267e-02 -2.43141934e-01 1.51991832e+00 -7.30452716e-01
-2.37876105e+00 -9.07601044e-02 -1.31476939e-01 -3.30764055e-01
-1.54705897e-01 -2.76046723e-01 -8.93205345e-01 1.98428798e-03
-5.34682095e-01 4.41433519e-01 7.55917728e-01 -6.87948227e-01
-5.62671423e-01 3.55105013e-01 -6.24779582e-01 6.00558579e-01
-6.45368636e-01 1.23850681e-01 -2.71353070e-02 -9.18917358e-01
-1.35749176e-01 -7.45443821e-01 6.83223233e-02 -8.37731183e-01
-5.76410472e-01 -3.46147537e-01 1.04275572e+00 -1.15877461e+00
2.22856307e+00 -2.04151273e+00 4.77319807e-01 -1.84721231e-01
-1.95626184e-01 8.90417397e-01 -5.33712693e-02 5.08270323e-01
1.00181676e-01 3.52644920e-03 1.41999125e-01 -1.66911647e-01
6.17346577e-02 2.09542572e-01 -4.26357746e-01 -4.78226431e-02
-3.48551542e-01 9.85546768e-01 -4.22844410e-01 -4.00808901e-01
4.58433241e-01 4.95799780e-01 -4.86301810e-01 7.89797843e-01
-1.86547086e-01 5.49435914e-01 1.07666694e-01 2.23306373e-01
1.02964401e-01 4.56795067e-01 -2.09825590e-01 -1.02071166e-02
-5.70241034e-01 6.94296181e-01 -9.87078965e-01 1.35164833e+00
-7.71205425e-01 7.10046291e-01 -3.35402191e-02 -6.04694605e-01
1.44021797e+00 9.73329306e-01 7.95104802e-02 -6.98766887e-01
3.17728966e-01 2.83748895e-01 3.05000782e-01 -8.85367692e-01
9.84382451e-01 -4.54025179e-01 4.46637899e-01 1.11250170e-01
2.26596475e-01 -3.02843779e-01 -3.88564587e-01 -4.86304969e-01
3.38907957e-01 3.18615586e-01 -1.40163317e-01 -8.04820284e-02
8.07888806e-01 -4.13398027e-01 8.72568190e-01 1.30061790e-01
2.07600936e-01 5.70749104e-01 4.48490493e-02 -1.50481597e-01
-1.41710436e+00 -5.35171688e-01 6.20498806e-02 1.28937483e+00
-1.26881316e-01 -5.23977816e-01 -7.96892941e-01 2.76229084e-01
-5.61033666e-01 1.04261661e+00 -2.28000656e-01 -3.64196420e-01
-7.76691675e-01 1.64188556e-02 9.94968653e-01 6.62313282e-01
3.03342670e-01 -1.71812904e+00 -2.24568650e-01 6.95723534e-01
-2.73369879e-01 -6.21487558e-01 -1.11945045e+00 -1.24290781e-02
-8.31894219e-01 -3.19008023e-01 -9.65819180e-01 -1.12173748e+00
-3.55356842e-01 -1.55141383e-01 4.70099062e-01 -1.56585529e-01
9.91658345e-02 -8.29868689e-02 -5.23365200e-01 -4.74076599e-01
-7.37153113e-01 2.86946595e-01 5.06467044e-01 -1.66064814e-01
-1.99321541e-03 -9.12120581e-01 -5.65444708e-01 7.53622949e-02
-6.34877443e-01 2.47946218e-01 1.87622741e-01 9.12346363e-01
5.08301914e-01 2.81626303e-02 1.25121987e+00 -3.94308567e-01
1.10431707e+00 -2.19476297e-01 -4.15810585e-01 4.87265848e-02
-8.16857517e-01 -2.55905479e-01 1.51374662e+00 -7.95627117e-01
-1.14241552e+00 -5.69121480e-01 -5.91537893e-01 -8.12615216e-01
2.62113690e-01 4.68125820e-01 -1.05296157e-01 2.98412293e-01
3.22149038e-01 8.30092132e-01 8.13522562e-02 -9.06596005e-01
1.89537928e-01 1.47956657e+00 7.63991416e-01 -1.51926696e-01
3.73198271e-01 -5.50709307e-01 -4.90136951e-01 -1.16576660e+00
-2.16818213e-01 -2.35926270e-01 -2.89851934e-01 -4.27268833e-01
7.87299335e-01 -7.70325243e-01 -1.04525602e+00 4.83337551e-01
-1.26302898e+00 -3.66984993e-01 -5.13603806e-01 1.04062331e+00
-8.34789276e-01 1.82890698e-01 -1.32782233e+00 -1.34123170e+00
-7.95822620e-01 -1.20695794e+00 3.46921057e-01 6.62011802e-01
-2.72082165e-03 -8.73534203e-01 9.33131948e-02 2.99233109e-01
8.28364074e-01 -1.69822097e-01 8.16581786e-01 -3.11994165e-01
-1.53858259e-01 1.69462129e-01 4.16939020e-01 9.41423714e-01
5.99996001e-02 -4.76205349e-02 -1.02360404e+00 1.37208283e-01
2.29078010e-01 -2.55189508e-01 4.26447242e-01 5.98607361e-01
1.31207430e+00 -6.97085321e-01 4.33186620e-01 7.38364518e-01
9.87753928e-01 1.05591524e+00 8.59357893e-01 -1.38220146e-01
8.14547956e-01 3.57410282e-01 5.21849573e-01 4.84372616e-01
1.91135228e-01 5.33567607e-01 1.51798548e-02 -3.33934166e-02
-5.61970294e-01 -5.60814857e-01 6.06419086e-01 2.40327191e+00
-4.57035065e-01 -3.01310152e-01 -3.13870877e-01 5.56322992e-01
-1.53731191e+00 -1.03065503e+00 -1.46757916e-01 2.04232335e+00
1.08458030e+00 -3.25864464e-01 1.46090016e-01 4.53254461e-01
9.28792417e-01 3.86173457e-01 -6.83850348e-01 -1.11968672e+00
-8.64412189e-02 3.77539873e-01 1.17535375e-01 4.29798752e-01
-4.16067034e-01 1.12585437e+00 6.22001886e+00 1.36096859e+00
-1.26447165e+00 1.61251262e-01 2.59841919e-01 -3.60964209e-01
-3.15100789e-01 -3.18297207e-01 -7.23580539e-01 6.78269923e-01
1.53904212e+00 -9.67520475e-02 1.10833013e+00 7.96866953e-01
7.62855947e-01 7.07242489e-01 -5.75791061e-01 9.93406475e-01
-9.51901302e-02 -1.38848913e+00 -1.04007320e-02 -1.55120805e-01
7.07715929e-01 -3.00980300e-01 2.47355342e-01 5.06318390e-01
-2.84958214e-01 -1.23275959e+00 8.82653892e-01 7.98848927e-01
1.36472797e+00 -1.03454483e+00 6.90592527e-01 3.65669847e-01
-1.39997494e+00 -1.66613087e-01 -5.96070588e-01 -2.71162361e-01
3.38471025e-01 -1.22040607e-01 -7.39592075e-01 5.17523527e-01
2.03420639e-01 6.89403236e-01 3.29191349e-02 9.82809365e-01
-9.41992626e-02 1.40717494e+00 -8.55703577e-02 -5.13134241e-01
3.22541714e-01 -2.26687238e-01 7.18815863e-01 1.19374692e+00
5.20377934e-01 2.08291411e-01 -3.14384371e-01 7.21688747e-01
-9.62105617e-02 4.47536558e-01 -1.90195367e-01 -4.86245364e-01
7.42847383e-01 8.51524830e-01 7.97391459e-02 -1.89056352e-01
1.08373143e-01 6.51002407e-01 -2.04854652e-01 4.10807699e-01
-9.88809824e-01 -9.01693761e-01 1.70275345e-01 -9.32028815e-02
3.85810405e-01 -2.11810052e-01 -1.11680485e-01 -7.57655799e-01
-3.59821320e-01 -9.03819978e-01 -2.94992298e-01 -1.09911144e+00
-5.64413726e-01 8.47408533e-01 -5.25883198e-01 -1.42390490e+00
-4.32302833e-01 -2.93287247e-01 -9.33728218e-01 1.11912251e+00
-1.36817586e+00 -9.30171430e-01 1.78837463e-01 3.29129189e-01
1.02643192e+00 -5.98661721e-01 7.93097973e-01 3.30939800e-01
-7.01139271e-01 5.53835452e-01 5.47634900e-01 -1.81420282e-01
2.98612684e-01 -1.13107061e+00 2.51853522e-02 5.28505147e-01
8.91740695e-02 3.29151422e-01 6.82108700e-01 -3.46917301e-01
-1.33564425e+00 -9.48270142e-01 9.35530663e-01 3.73807907e-01
4.35546279e-01 -6.73332736e-02 -8.87729645e-01 2.85845667e-01
6.37359798e-01 -6.39969885e-01 6.77540779e-01 -7.12973699e-02
2.60264963e-01 -2.25137770e-01 -4.22912151e-01 6.13966346e-01
7.78901398e-01 -7.05622971e-01 -8.28063488e-01 1.21158659e-01
1.32088614e+00 -6.46667540e-01 -1.08958781e+00 2.24122748e-01
7.04124749e-01 -8.52492630e-01 6.67932689e-01 -5.46050787e-01
6.03754401e-01 -2.23638967e-01 -1.22015715e-01 -1.36782074e+00
-4.32790667e-01 -1.07015097e+00 -3.91186744e-01 1.51067913e+00
3.97240132e-01 -2.98894525e-01 3.31629843e-01 2.66217738e-02
-7.93693364e-01 -8.32918942e-01 -9.40961182e-01 -9.91784334e-01
-3.41856740e-02 -4.22110081e-01 8.37300003e-01 6.09304130e-01
1.73078656e-01 6.61520779e-01 -1.20792150e+00 -2.61897624e-01
-2.28558257e-01 -1.14489473e-01 3.53222013e-01 -6.22805953e-01
-5.34866571e-01 -4.60552841e-01 1.75557494e-01 -1.26325166e+00
3.21686007e-02 -5.91283083e-01 3.34642023e-01 -1.28570747e+00
-4.39780623e-01 -3.77681106e-01 -4.87089306e-01 1.03854686e-01
-7.30114952e-02 4.61129583e-02 5.23810208e-01 1.41180322e-01
-1.86990187e-01 1.33071816e+00 1.71478832e+00 2.15733647e-01
-7.94970334e-01 3.61547559e-01 -2.03370497e-01 4.67170268e-01
9.31576252e-01 -7.40117952e-02 -5.47607899e-01 -2.02236101e-01
-2.42485225e-01 9.65600193e-01 -2.67686993e-01 -1.13538992e+00
3.78060907e-01 6.77376911e-02 7.95330778e-02 -7.70222962e-01
7.88562834e-01 -3.70291382e-01 2.38274977e-01 5.53941548e-01
-7.67010212e-01 -5.19087017e-02 2.03394607e-01 2.86117107e-01
-3.98380518e-01 -4.59474593e-01 6.28428698e-01 1.33944638e-02
-3.94206196e-01 2.35743687e-01 -5.46374261e-01 -1.94468558e-01
4.09665674e-01 -3.04505020e-01 2.84112673e-02 -7.36182034e-01
-8.32120240e-01 -3.72392349e-02 -3.59398991e-01 5.26075065e-01
8.04101050e-01 -1.60331881e+00 -7.13637352e-01 1.87472984e-01
-4.27200198e-01 -3.37617964e-01 6.51284218e-01 5.46593130e-01
-5.83985388e-01 7.61099041e-01 -1.76216051e-01 -1.33079916e-01
-1.24995613e+00 5.15627801e-01 3.52523088e-01 -4.54198159e-02
-4.74187046e-01 8.52639377e-01 -1.12446763e-01 -5.96111655e-01
4.49793309e-01 -1.73628286e-01 -7.93773174e-01 -1.67513549e-01
4.73797321e-01 8.16167533e-01 -4.69013065e-01 -6.61407173e-01
1.75118089e-01 4.11713511e-01 2.80866653e-01 -3.08314890e-01
1.26037920e+00 -4.54328537e-01 -3.84371281e-02 9.59112704e-01
1.05602157e+00 2.92880327e-01 -1.04053628e+00 2.40172688e-02
-3.88810158e-01 -5.13201207e-02 3.98923606e-01 -8.55955601e-01
-9.74705696e-01 1.16759789e+00 4.16969776e-01 2.18385741e-01
1.33103430e+00 -5.06855071e-01 1.62578738e+00 -6.40567765e-02
-3.94762665e-01 -1.56887233e+00 2.33061761e-02 6.34303391e-01
1.01274335e+00 -4.77389902e-01 -6.29723728e-01 2.39175484e-01
-9.39667165e-01 1.37421799e+00 6.89466476e-01 -2.40755543e-01
4.23266083e-01 7.20200762e-02 9.93623585e-03 4.43687230e-01
-9.72571194e-01 3.17523032e-02 2.79411644e-01 4.55549389e-01
6.36305332e-01 2.16156885e-01 -6.68693006e-01 1.05645144e+00
-9.36242402e-01 4.60999012e-02 6.26267552e-01 -4.46106233e-02
-6.90398872e-01 -9.68122721e-01 -4.35377240e-01 4.10583884e-01
-5.86858332e-01 -4.36253786e-01 1.01140104e-01 1.72259733e-01
2.58825570e-01 1.13599575e+00 1.39017673e-02 -1.14054966e+00
2.09293082e-01 2.38966867e-01 1.04250468e-01 -3.75007898e-01
-1.02968347e+00 5.40135920e-01 8.49559307e-02 -9.34358463e-02
-1.04640923e-01 -2.07748100e-01 -1.27800286e+00 -2.02834889e-01
-9.14078057e-01 6.92345619e-01 7.10027635e-01 8.63975286e-01
2.97954470e-01 1.28325891e+00 1.23688626e+00 -4.78075027e-01
-5.68144917e-01 -1.47176659e+00 -9.71499622e-01 -2.55163938e-01
3.64425659e-01 -8.97651538e-02 -2.30549619e-01 2.45414283e-02]
|
[15.529462814331055, 6.193066120147705]
|
00d735ba-8cb0-48ff-9074-c0bfdd58c8a9
|
structure-plp-slam-efficient-sparse-mapping
|
2207.06058
| null |
https://arxiv.org/abs/2207.06058v3
|
https://arxiv.org/pdf/2207.06058v3.pdf
|
Structure PLP-SLAM: Efficient Sparse Mapping and Localization using Point, Line and Plane for Monocular, RGB-D and Stereo Cameras
|
This paper presents a visual SLAM system that uses both points and lines for robust camera localization, and simultaneously performs a piece-wise planar reconstruction (PPR) of the environment to provide a structural map in real-time. One of the biggest challenges in parallel tracking and mapping with a monocular camera is to keep the scale consistent when reconstructing the geometric primitives. This further introduces difficulties in graph optimization of the bundle adjustment (BA) step. We solve these problems by proposing several run-time optimizations on the reconstructed lines and planes. Our system is able to run with depth and stereo sensors in addition to the monocular setting. Our proposed SLAM tightly incorporates the semantic and geometric features to boost both frontend pose tracking and backend map optimization. We evaluate our system exhaustively on various datasets, and show that we outperform state-of-the-art methods in terms of trajectory precision. The code of PLP-SLAM has been made available in open-source for the research community (https://github.com/PeterFWS/Structure-PLP-SLAM).
|
['Didier Stricker', 'Alain Pagani', 'Jiaxuan Wang', 'Fangwen Shu']
|
2022-07-13
| null | null | null | null |
['camera-localization']
|
['computer-vision']
|
[-3.09044600e-01 -3.63643587e-01 1.78159531e-02 -3.89316708e-01
-5.72127819e-01 -8.86405051e-01 4.66451824e-01 1.46065816e-01
-5.36133349e-01 3.58701974e-01 -1.65155306e-01 -2.15153009e-01
1.33462891e-01 -7.77665317e-01 -9.63433385e-01 -1.51873097e-01
4.79192324e-02 9.76583183e-01 5.26898623e-01 -1.39826670e-01
3.38276178e-01 8.33614230e-01 -1.39219427e+00 -4.68500763e-01
5.98388314e-01 7.62823939e-01 5.34321487e-01 7.26429999e-01
1.08902305e-01 3.48433614e-01 6.31432459e-02 -2.35167176e-01
5.99633813e-01 4.74427007e-02 -4.23358083e-01 1.30003601e-01
8.43470752e-01 -2.82309413e-01 -5.71490049e-01 1.16531575e+00
4.85332698e-01 2.39430275e-02 3.09965834e-02 -1.38432622e+00
1.52921587e-01 -1.66716844e-01 -7.64919817e-01 -3.74019712e-01
7.92981923e-01 5.28402850e-02 7.32539594e-01 -1.09143150e+00
8.52112412e-01 1.01718688e+00 9.78484809e-01 7.48962909e-02
-1.05117810e+00 -7.16160953e-01 -1.42632169e-03 2.92498115e-02
-1.82250309e+00 -6.36809886e-01 6.59650445e-01 -4.53159690e-01
9.30606008e-01 2.61275738e-01 8.41928482e-01 7.73241699e-01
2.86702991e-01 1.70480430e-01 6.18418634e-01 -3.82513225e-01
4.62794155e-02 -6.40317425e-02 -1.20559588e-01 1.02092421e+00
5.00089645e-01 2.12061226e-01 -8.08776200e-01 -1.18031360e-01
1.09390414e+00 2.37297744e-01 -4.14930254e-01 -1.21459043e+00
-1.61828792e+00 6.44758105e-01 6.34798765e-01 -2.29036570e-01
-1.56562284e-01 4.60781127e-01 -5.79724126e-02 7.18277693e-02
2.16121823e-01 2.91779608e-01 -2.08867103e-01 -1.95879444e-01
-8.48343372e-01 2.03792587e-01 6.64466262e-01 1.57836449e+00
1.15144086e+00 -3.50663453e-01 6.89462602e-01 4.54218000e-01
5.03227770e-01 9.42303717e-01 3.35994810e-02 -1.21812904e+00
6.32564545e-01 4.09656584e-01 3.47276598e-01 -1.19921517e+00
-6.43998861e-01 -2.84956247e-01 -5.19980788e-01 3.21572930e-01
2.88010716e-01 -2.11612992e-02 -5.05980492e-01 1.25495839e+00
6.80792093e-01 1.54314473e-01 -2.89325535e-01 1.03109288e+00
5.72277009e-01 3.46639186e-01 -6.76031947e-01 1.95431143e-01
1.08720803e+00 -1.15418005e+00 -4.71258193e-01 -6.77241623e-01
4.85400259e-01 -1.12966073e+00 6.26122296e-01 1.57245591e-01
-7.75214434e-01 -3.51784766e-01 -1.11795259e+00 -3.56482685e-01
3.03562209e-02 1.29529551e-01 5.47766745e-01 1.38018176e-01
-1.16133893e+00 3.89864922e-01 -1.07099485e+00 -7.80487061e-01
-4.41638641e-02 4.87813115e-01 -8.43046486e-01 -2.73842275e-01
-5.36541879e-01 9.53922272e-01 2.48347998e-01 6.67810291e-02
-5.80979526e-01 -5.00666857e-01 -1.17231417e+00 -3.70804638e-01
4.37574118e-01 -1.04199100e+00 1.13252378e+00 -2.68661231e-01
-1.43110430e+00 9.75950241e-01 -4.48820740e-01 -2.79685795e-01
8.86381030e-01 -3.92766446e-01 1.60070434e-01 -4.82914550e-03
1.82151213e-01 7.94333100e-01 3.85767877e-01 -1.31357777e+00
-6.44102633e-01 -5.80939412e-01 -6.47554100e-02 6.42108619e-01
3.70382279e-01 -3.17238271e-01 -1.04369235e+00 -9.21503007e-02
9.07709479e-01 -1.28518832e+00 -3.37365478e-01 3.76697510e-01
-3.30765963e-01 4.82181102e-01 6.50285125e-01 -6.09028459e-01
6.91785276e-01 -1.91938448e+00 2.85314322e-01 2.28754312e-01
1.51690647e-01 -3.55455458e-01 1.61731914e-01 5.96030951e-01
4.09254700e-01 -5.76367497e-01 3.14453132e-02 -9.55711484e-01
-8.33225176e-02 2.85666853e-01 -2.17995942e-01 1.15046954e+00
-6.45706177e-01 7.16569543e-01 -7.99115419e-01 -3.19685996e-01
6.78808749e-01 4.84214395e-01 -4.80230242e-01 1.38942033e-01
-2.93439683e-02 7.41455495e-01 -1.44062817e-01 7.10121274e-01
9.80250001e-01 -1.17676005e-01 1.18071787e-01 -1.95655823e-01
-5.48291326e-01 3.15911740e-01 -1.48339152e+00 2.49648213e+00
-3.45816106e-01 6.40374184e-01 3.25324357e-01 -1.74147502e-01
8.58398378e-01 -1.99471772e-01 4.60746527e-01 -5.43996871e-01
9.67762917e-02 3.37133050e-01 -5.49355865e-01 1.96945354e-01
8.17632914e-01 2.71081179e-01 8.34107399e-03 9.59523531e-05
-1.68171123e-01 -5.64500391e-01 -1.97481215e-01 1.01309337e-01
9.73478556e-01 5.79660654e-01 6.23531997e-01 -2.01861680e-01
4.45449769e-01 4.79474008e-01 6.81377470e-01 3.89046371e-01
8.41161907e-02 6.94513619e-01 -9.67574567e-02 -3.80542725e-01
-1.18625951e+00 -1.12085927e+00 -7.89060630e-03 3.96347791e-01
8.82354677e-01 -6.93298876e-01 -3.28377962e-01 -1.40435621e-01
3.64182323e-01 2.85802752e-01 -2.09669620e-02 3.56769085e-01
-5.18346429e-01 -1.25369608e-01 2.07375064e-01 2.83812106e-01
4.77671295e-01 -3.12888324e-01 -9.30869699e-01 3.88229340e-02
-1.82567179e-01 -1.35015368e+00 -4.60989356e-01 -4.07123566e-02
-7.75343239e-01 -1.20292342e+00 -3.74809206e-01 -6.16549551e-01
8.20646405e-01 8.48628581e-01 8.18088472e-01 7.69891217e-02
-6.72528669e-02 4.68219191e-01 -1.93029374e-01 -1.82892546e-01
2.76283771e-02 5.41476607e-02 2.51819402e-01 -3.25366616e-01
1.07815284e-02 -6.37285829e-01 -5.08723974e-01 6.27825499e-01
-1.36854112e-01 5.32469749e-01 1.67024747e-01 3.41765612e-01
1.09276378e+00 -1.94799259e-01 -6.43382728e-01 -4.74707037e-01
-1.35131881e-01 -5.47350496e-02 -1.49644017e+00 -1.51960656e-01
-5.00249922e-01 -1.88614100e-01 1.65754095e-01 -2.69303471e-02
-6.03985190e-01 7.16003299e-01 -7.00414330e-02 -6.79139197e-01
6.89039752e-02 2.22959474e-01 -1.57457203e-01 -7.77476013e-01
4.28197056e-01 1.47473156e-01 8.16217214e-02 -5.97482979e-01
3.68614256e-01 3.21916103e-01 7.58146048e-01 -3.24144155e-01
1.26161730e+00 9.51599538e-01 4.05562550e-01 -7.80884981e-01
-6.83091998e-01 -9.58428204e-01 -1.11537671e+00 -2.85068631e-01
6.54171705e-01 -1.32405531e+00 -7.57764459e-01 5.75359285e-01
-1.25292647e+00 -3.25849444e-01 2.33847424e-02 5.95275342e-01
-7.96832561e-01 6.15757048e-01 -3.62344086e-01 -5.21510124e-01
-1.36651933e-01 -1.35343969e+00 1.35565305e+00 3.94695103e-02
-7.12716132e-02 -8.77903700e-01 4.01959270e-01 1.22611627e-01
9.70002413e-02 4.73733872e-01 -5.31800687e-02 1.38966411e-01
-1.22206628e+00 -2.52294898e-01 -8.84749964e-02 -3.98027897e-01
-1.29220381e-01 -1.86927021e-01 -7.43641019e-01 -7.20268309e-01
-7.36120045e-02 1.87211648e-01 5.57391047e-01 2.15787709e-01
5.38550794e-01 -1.99533044e-03 -6.05907500e-01 1.38900173e+00
1.74840498e+00 -1.19463772e-01 4.43891466e-01 8.43605936e-01
1.10564911e+00 3.35166842e-01 9.59680557e-01 4.69196916e-01
9.03329730e-01 1.19859695e+00 8.43886912e-01 5.79932332e-03
-7.35686300e-03 -6.48636997e-01 2.78255373e-01 8.62623692e-01
1.56863436e-01 1.33341685e-01 -1.26055360e+00 2.49480247e-01
-2.14883971e+00 -5.64395487e-01 -5.75336516e-01 2.55355906e+00
2.95447916e-01 -1.70866534e-01 -1.91437542e-01 -2.73005188e-01
5.36189914e-01 2.07815856e-01 -2.90659994e-01 2.03086600e-01
-3.99930961e-03 -3.41432244e-01 1.20507836e+00 1.14156187e+00
-1.10698438e+00 1.23973012e+00 5.24335670e+00 1.59593269e-01
-1.14697206e+00 1.99614868e-01 -4.00716394e-01 -7.59181306e-02
9.38413665e-03 5.55304110e-01 -1.14465129e+00 1.82879984e-01
6.73900425e-01 -7.20868707e-02 6.47035301e-01 9.99199152e-01
-2.06225496e-02 -3.92157435e-01 -1.12904882e+00 1.39676511e+00
3.56814116e-02 -1.40114117e+00 -4.60256457e-01 4.36586857e-01
6.12349272e-01 7.24518895e-01 -3.53600532e-01 -1.43250868e-01
2.32939348e-01 -6.23585165e-01 1.13864863e+00 4.11991328e-01
8.94648433e-01 -6.20698452e-01 5.95002711e-01 7.07768440e-01
-1.49799669e+00 2.79011607e-01 -5.87110579e-01 -2.60595918e-01
3.75860244e-01 5.71392953e-01 -9.86768067e-01 9.30396914e-01
6.97413146e-01 8.88418913e-01 -5.56786478e-01 1.36163223e+00
-3.16918343e-01 -2.91687280e-01 -6.98917806e-01 4.38309550e-01
-4.88881171e-02 -4.31281596e-01 7.32764482e-01 9.01102483e-01
5.72405398e-01 -2.73995906e-01 4.85133737e-01 5.75369060e-01
5.57639748e-02 -7.16758147e-02 -7.53622532e-01 5.74519932e-01
7.99221337e-01 1.18226242e+00 -6.07338727e-01 -3.94641273e-02
-3.53992164e-01 1.20193577e+00 4.35153067e-01 -4.94859144e-02
-8.27898204e-01 -2.86033507e-02 8.40501964e-01 3.65834832e-01
8.51542577e-02 -9.48698878e-01 -3.65261346e-01 -1.48084521e+00
3.26516330e-01 -4.57996011e-01 7.15506151e-02 -9.67589915e-01
-6.29121482e-01 4.56608891e-01 -1.13279335e-01 -1.48438561e+00
-1.00919545e-01 -4.06020850e-01 -9.37645957e-02 9.88040507e-01
-1.63733089e+00 -1.25518394e+00 -1.04244947e+00 6.25631809e-01
4.26745474e-01 1.88508406e-01 6.94850802e-01 2.89241076e-01
-1.26319751e-01 1.35846958e-01 4.95113470e-02 -4.52239551e-02
8.81546378e-01 -1.05650663e+00 8.13511074e-01 1.21473575e+00
3.83921236e-01 5.93770504e-01 7.91224420e-01 -8.16750467e-01
-1.96486330e+00 -1.02281988e+00 8.25769365e-01 -7.69563675e-01
4.24348742e-01 -7.19779432e-01 -4.67846274e-01 1.18172443e+00
-1.64912671e-01 6.03611171e-02 2.83245333e-02 -1.29497781e-01
-2.43563667e-01 -1.11566551e-01 -9.59388554e-01 4.49278951e-01
1.30933774e+00 -3.23422790e-01 -2.46122405e-01 4.71083343e-01
7.63449013e-01 -1.35237920e+00 -5.88839948e-01 3.30477297e-01
5.62272012e-01 -1.12835348e+00 1.14956272e+00 3.40411931e-01
-3.80170047e-01 -9.18106139e-01 -4.46794301e-01 -1.12913072e+00
-3.00374120e-01 -7.56850421e-01 -1.30064497e-02 9.16048229e-01
-4.94291596e-02 -8.68062079e-01 7.84403145e-01 2.61258006e-01
-8.85215178e-02 -2.51782715e-01 -1.09205925e+00 -8.26580465e-01
-6.68186843e-01 -4.89858091e-01 5.90237856e-01 7.47399628e-01
-3.06358844e-01 1.70266569e-01 -4.53717798e-01 8.74257565e-01
1.00330698e+00 2.34417528e-01 1.54089487e+00 -1.14008617e+00
-2.37882674e-01 -1.06972806e-01 -7.90289879e-01 -1.37104678e+00
1.45741459e-02 -7.57556677e-01 1.37131855e-01 -1.66310680e+00
-1.13001570e-01 -6.08300447e-01 5.37401676e-01 2.09346741e-01
3.79362226e-01 1.57730833e-01 3.88984710e-01 6.77368641e-01
-6.46994472e-01 3.68950665e-01 8.82237017e-01 3.00522149e-01
-4.70219031e-02 -2.91947220e-02 -8.73000398e-02 9.13284302e-01
5.88987231e-01 -4.97511923e-01 -1.32870868e-01 -9.10099626e-01
2.47305885e-01 2.36638308e-01 5.44166923e-01 -1.32762992e+00
7.22036839e-01 -1.49320573e-01 2.98516512e-01 -1.17458200e+00
7.91891694e-01 -1.25643587e+00 8.17877054e-01 5.46948254e-01
3.52549314e-01 5.62719882e-01 8.67860243e-02 4.47425067e-01
-1.12521509e-02 -5.97864613e-02 6.56969190e-01 -1.85040891e-01
-1.00137222e+00 5.87351859e-01 3.76846462e-01 -3.21068943e-01
1.17501760e+00 -2.30567694e-01 -4.13616657e-01 -4.88378853e-01
-2.15709418e-01 4.24609303e-01 1.63777506e+00 3.80094141e-01
6.86141193e-01 -1.32801402e+00 -3.75741631e-01 3.61786038e-01
3.22248250e-01 6.00762367e-01 1.20265866e-02 9.84733641e-01
-1.51142442e+00 4.90598410e-01 -1.37751192e-01 -1.08524561e+00
-1.32116210e+00 4.57008094e-01 3.91481727e-01 -2.35287230e-02
-9.53278005e-01 5.90708137e-01 8.63073021e-02 -8.26417923e-01
3.35092098e-01 -1.09191343e-01 5.35741031e-01 -3.76290202e-01
4.33105558e-01 5.43627143e-01 1.40319347e-01 -9.91931140e-01
-7.84257352e-01 1.13343573e+00 3.81444693e-01 -2.65437901e-01
1.23438716e+00 -5.69186449e-01 -3.21569800e-01 2.33011484e-01
1.13782191e+00 4.54163998e-01 -1.36327267e+00 -2.46879876e-01
-3.17388438e-02 -8.26175272e-01 7.89919868e-02 -3.17842692e-01
-6.79611266e-01 7.47987628e-01 4.03667271e-01 -4.95013177e-01
6.40567243e-01 -1.51147798e-01 6.49062455e-01 3.65157068e-01
1.25443697e+00 -5.72863758e-01 -4.42937762e-01 8.20178330e-01
8.79428029e-01 -1.26213121e+00 3.45912397e-01 -8.38507056e-01
-3.48892689e-01 1.13344717e+00 4.59840566e-01 -2.93949187e-01
3.15075547e-01 4.55185086e-01 1.62077755e-01 -2.08699241e-01
-2.23995402e-01 -9.36518013e-02 2.10027605e-01 6.04600310e-01
-2.27128770e-02 1.32705092e-01 1.26281127e-01 -3.04245263e-01
-4.88857538e-01 -3.38643789e-01 4.04872745e-01 1.09709275e+00
-5.48625410e-01 -1.22002172e+00 -7.34980345e-01 -2.40714014e-01
2.26903990e-01 1.00476749e-01 -2.70679772e-01 9.12894070e-01
-1.95317194e-01 5.89094758e-01 -1.65407788e-02 -5.12513697e-01
4.26409096e-01 -2.98460841e-01 7.18756378e-01 -6.58216119e-01
-1.99534044e-01 2.02486396e-01 1.15520425e-01 -1.22893202e+00
-1.20779522e-01 -8.77598405e-01 -1.37602270e+00 -5.75947940e-01
-1.26342937e-01 -1.63887098e-01 1.17657638e+00 4.37712967e-01
5.44273436e-01 3.68012860e-03 3.71622145e-01 -1.31130624e+00
-3.86053771e-01 -4.68175232e-01 -3.73193264e-01 9.87453684e-02
4.76853698e-01 -7.45213151e-01 -1.19737230e-01 -2.66646147e-01]
|
[7.3963189125061035, -2.21466064453125]
|
bea4973c-2381-4115-8836-55e06974603f
|
sparse-insar-data-3d-inpainting-for-ground
|
2203.02407
| null |
https://arxiv.org/abs/2203.02407v1
|
https://arxiv.org/pdf/2203.02407v1.pdf
|
Sparse InSAR Data 3D Inpainting for Ground Deformation Detection Along the Rail Corridor
|
Monitoring of ground movement close to the rail corridor, such as that associated with landslips caused by ground subsidence and/or uplift, is of great interest for the detection and prevention of possible railway faults. Interferometric synthetic-aperture radar (InSAR) data can be used to measure ground deformation, but its use poses distinct challenges, as the data is highly sparse and can be particularly noisy. Here we present a scheme for processing and interpolating noisy, sparse InSAR data into a dense spatio-temporal stack, helping suppress noise and opening up the possibility for treatment with deep learning and other image processing methods.
|
['Nantheera Anantrasirichai', 'Alin Achim', 'David Bull', 'Juliet Biggs', 'Odysseas Pappas']
|
2022-03-04
| null | null | null | null |
['3d-inpainting']
|
['computer-vision']
|
[ 5.10104060e-01 -3.41908723e-01 1.85041860e-01 -2.42427826e-01
-8.30881596e-01 -2.36489609e-01 3.54294598e-01 1.44625753e-01
-5.40438652e-01 8.88424456e-01 8.20936784e-02 -1.79248694e-02
-5.57590127e-01 -1.31933689e+00 -7.29642868e-01 -9.39035356e-01
-4.92332160e-01 5.48905373e-01 2.51196742e-01 -6.86089396e-01
4.72369716e-02 9.65814888e-01 -1.57947016e+00 5.64884730e-02
9.40728366e-01 6.61216795e-01 4.29874569e-01 5.07948622e-02
4.18252051e-01 2.47508422e-01 -4.23538506e-01 1.70231670e-01
9.01114717e-02 -1.07125603e-01 -5.85933268e-01 2.17526138e-01
3.88796151e-01 -3.14972609e-01 -5.69531798e-01 1.04746008e+00
3.47490758e-01 1.62982911e-01 2.74952203e-01 -4.89051193e-01
-1.17551580e-01 3.20001870e-01 -6.32028997e-01 5.15936196e-01
-1.88252386e-02 -2.85497494e-02 5.55449665e-01 -6.42826319e-01
5.46636522e-01 9.18587744e-01 7.88591385e-01 -2.72891462e-01
-1.30130851e+00 -2.15217337e-01 -1.43622547e-01 2.20751032e-01
-1.36433375e+00 -2.60187417e-01 8.10471773e-01 -5.27895212e-01
6.65096760e-01 2.07993060e-01 1.00007427e+00 8.56946647e-01
2.10877791e-01 5.49661398e-01 1.14254415e+00 -2.67340153e-01
1.86275691e-01 -9.07150626e-01 1.10427834e-01 3.06237280e-01
7.15739071e-01 3.00183296e-01 -3.86795074e-01 1.69633582e-01
6.94913983e-01 3.03812981e-01 -4.47979063e-01 1.56357095e-01
-1.12430823e+00 6.59400344e-01 8.66712332e-01 4.51411366e-01
-8.55114877e-01 9.63011906e-02 5.43917716e-02 3.10185701e-01
6.90203190e-01 3.66609275e-01 -3.95016432e-01 6.88841045e-02
-1.41179550e+00 5.71898460e-01 1.90894101e-02 9.05118957e-02
1.05055940e+00 4.33889478e-01 3.78376961e-01 6.95626080e-01
2.62664139e-01 1.04900086e+00 1.66028515e-01 -5.54569542e-01
3.76239449e-01 3.04434836e-01 -5.32339253e-02 -8.99701238e-01
-8.09618056e-01 -6.42492771e-01 -1.15850234e+00 3.49211335e-01
1.88087165e-01 -3.43831986e-01 -1.02810717e+00 1.08134580e+00
9.15959775e-02 2.31668994e-01 3.69146802e-02 1.06097209e+00
4.71614152e-01 4.98293728e-01 -1.49932787e-01 -2.88618207e-01
1.40735412e+00 -1.16036879e-02 -8.07110429e-01 -8.67346942e-01
5.54977655e-01 -5.51486909e-01 5.16758919e-01 1.44166276e-01
-5.16590774e-01 -2.51600027e-01 -1.01014495e+00 3.91629696e-01
-2.62668222e-01 1.21696936e-02 6.90223217e-01 -2.93108318e-02
-5.73394656e-01 9.67684269e-01 -1.29383564e+00 -9.11087096e-02
5.77073574e-01 6.04210719e-02 -5.29406905e-01 -3.69196534e-01
-1.39883065e+00 9.78776634e-01 1.48220062e-01 1.16167641e+00
-3.29987586e-01 -3.91914815e-01 -1.27769148e+00 -1.34647340e-01
4.14788984e-02 -2.09756330e-01 2.93011248e-01 -3.24888080e-01
-7.24561930e-01 9.27073002e-01 2.41760835e-01 -8.07722926e-01
5.50944269e-01 -7.38285124e-01 -4.80440021e-01 -8.71285424e-02
3.62410456e-01 -8.80241245e-02 5.73061943e-01 -8.71639848e-01
-3.86637807e-01 -1.03208315e+00 -5.07823527e-01 1.13764502e-01
2.66000122e-01 1.91957457e-03 1.04485489e-01 -8.23339760e-01
9.32979286e-01 -9.66698766e-01 -5.96784592e-01 -4.36559290e-01
-2.95825303e-01 5.03596008e-01 1.03390563e+00 -9.13296223e-01
8.67060423e-01 -2.17104030e+00 -1.84635654e-01 2.22025678e-01
-3.39235663e-01 5.07979333e-01 2.41137445e-01 5.61834097e-01
2.99106142e-03 -3.42363745e-01 -8.01470637e-01 2.54372895e-01
-6.41621470e-01 4.14467603e-01 -2.00724378e-01 9.65365529e-01
3.92467469e-01 8.44970345e-01 -7.09138751e-01 -1.00966714e-01
2.89644599e-01 5.49832225e-01 -5.62735498e-02 -9.53851715e-02
1.24194540e-01 8.10933053e-01 -5.00571609e-01 7.55978703e-01
1.07492352e+00 2.31218770e-01 -7.60261854e-03 -2.18999192e-01
-6.06518567e-01 2.68348932e-01 -1.24241745e+00 1.33346236e+00
-9.49790999e-02 8.05062056e-01 2.76070088e-01 -1.28660190e+00
1.21490049e+00 3.77649255e-02 3.64858419e-01 -1.05278099e+00
-7.53148869e-02 2.39092633e-01 -1.11161083e-01 -1.00497842e+00
4.50974047e-01 -5.63642979e-01 -1.91223428e-01 -4.50365767e-02
-5.26326597e-01 -2.49378726e-01 -1.05451150e-02 -4.83263433e-01
9.95959938e-01 1.46728903e-01 -1.18251942e-01 -3.33430439e-01
2.01696068e-01 2.99621344e-01 7.15719521e-01 5.62444329e-01
2.37487093e-01 7.73948133e-01 1.16358191e-01 -8.60578299e-01
-6.92014694e-01 -1.09888661e+00 -5.17429054e-01 2.61634678e-01
1.84672579e-01 4.84711409e-01 -1.88827619e-01 4.91188988e-02
2.15825602e-01 1.69687998e-02 -3.69709224e-01 -2.30228603e-01
-9.93474305e-01 -1.45937932e+00 3.94807041e-01 6.13382518e-01
7.47963488e-01 -1.04873359e+00 -6.49154484e-01 6.49238110e-01
-4.81547117e-01 -1.14640176e+00 5.56462288e-01 4.56137121e-01
-1.36993158e+00 -9.36373115e-01 -5.33582330e-01 -6.44382000e-01
4.43406463e-01 2.34083131e-01 9.03853059e-01 1.91631336e-02
-4.91775066e-01 -3.35725099e-01 -2.47838587e-01 -9.34361666e-02
-1.15851492e-01 -1.88998267e-01 -1.77538529e-01 1.65413111e-01
4.72239330e-02 -6.83581710e-01 -4.34138268e-01 2.50844747e-01
-7.68087089e-01 -2.35309333e-01 5.38962603e-01 8.34341705e-01
6.62782550e-01 3.22954059e-01 5.24961233e-01 -8.86810839e-01
1.08485155e-01 -3.87367129e-01 -8.49753618e-01 -2.93388039e-01
-1.12854289e-02 -2.11507723e-01 -9.46029499e-02 3.02342325e-01
-9.51753974e-01 1.14229858e-01 -6.85713172e-01 7.02090934e-02
-4.65425760e-01 1.10893059e+00 -9.60925780e-03 -1.12411104e-01
7.22660720e-01 1.11857377e-01 1.85781583e-01 -6.69171870e-01
-1.16603792e-01 5.00635326e-01 8.46962333e-01 -1.92617122e-02
9.29521143e-01 8.43585491e-01 3.58141094e-01 -1.48903072e+00
-1.13827157e+00 -4.61159706e-01 -7.58410811e-01 -5.37188612e-02
8.41828704e-01 -1.21353006e+00 -2.13881638e-02 7.82511830e-01
-8.56729209e-01 -3.12203944e-01 -2.94843078e-01 8.67668331e-01
-2.83201337e-01 2.45382920e-01 -6.29023314e-01 -5.72129369e-01
-2.62234598e-01 -7.65333652e-01 1.04599285e+00 5.69533110e-02
-5.39847501e-02 -8.13127935e-01 3.80756184e-02 4.43599671e-01
4.41385925e-01 9.39282358e-01 4.20036763e-01 -7.96874091e-02
-5.17402828e-01 -6.15574181e-01 -2.31298804e-02 3.38865101e-01
2.57290810e-01 -1.10870250e-01 -7.57646680e-01 -2.09533170e-01
2.56294250e-01 8.11547562e-02 1.15317070e+00 8.97920728e-01
5.98798871e-01 -1.00202352e-01 -2.63631076e-01 7.23375916e-01
1.64400303e+00 -1.56832993e-01 9.67907548e-01 6.18802547e-01
5.23416221e-01 5.42431235e-01 9.71138179e-01 5.13376176e-01
-9.07000005e-02 6.61578655e-01 6.21616006e-01 -4.16385114e-01
1.01699755e-01 2.98132598e-01 2.39500850e-02 1.39023587e-01
-5.26456296e-01 2.36650303e-01 -1.03391111e+00 8.60135674e-01
-1.82543731e+00 -1.29487038e+00 -9.13016498e-01 2.19602633e+00
1.36967242e-01 1.04457006e-01 -4.01271552e-01 5.11159122e-01
6.90694571e-01 4.04436648e-01 -2.70246062e-02 1.95268854e-01
-6.63216650e-01 3.82942796e-01 9.06834185e-01 4.66910660e-01
-1.26988328e+00 6.31038487e-01 6.53569603e+00 2.01697037e-01
-1.43838799e+00 -1.82493925e-01 6.54367879e-02 4.47371721e-01
-1.53580636e-01 -1.81335770e-02 -5.42309880e-01 4.63067263e-01
7.02137113e-01 3.83251637e-01 -3.75202596e-02 3.75010103e-01
6.64998710e-01 -4.61640090e-01 -1.52920336e-01 6.63903892e-01
-3.70136946e-01 -1.62581730e+00 -3.76490325e-01 1.53349251e-01
5.87009668e-01 6.53186560e-01 -3.55741233e-01 -1.82601333e-01
-6.84586540e-02 -5.37924349e-01 5.21892011e-01 5.74056685e-01
4.08345073e-01 -7.98699260e-01 9.85767007e-01 2.04149991e-01
-9.42552328e-01 -7.43068457e-02 -1.73640072e-01 -5.35621643e-01
5.63637316e-01 1.09876788e+00 -3.62371981e-01 6.58581913e-01
9.05639648e-01 1.02976680e+00 -2.58569270e-01 1.16283166e+00
-6.33267462e-01 6.41163588e-01 -5.39129913e-01 6.17709517e-01
3.09316546e-01 -5.25478542e-01 6.65091455e-01 7.83002138e-01
6.01460874e-01 1.10656515e-01 2.61475205e-01 5.58376372e-01
2.78758556e-01 -4.11567867e-01 -9.52564061e-01 4.41765845e-01
2.18643948e-01 9.86705303e-01 -6.32521629e-01 -9.61320326e-02
-1.71098977e-01 5.79552710e-01 -2.49149144e-01 3.21283877e-01
-5.23286223e-01 -2.41074368e-01 9.06300008e-01 8.77940238e-01
3.40058565e-01 -6.27280414e-01 -3.08190405e-01 -1.04360676e+00
2.58767545e-01 -3.15836042e-01 3.65237743e-01 -6.21570289e-01
-9.10804868e-01 5.27837098e-01 -6.45349696e-02 -1.39326751e+00
6.35389090e-02 -2.92415142e-01 -7.38735795e-01 7.94314802e-01
-1.84225297e+00 -1.19745219e+00 -6.37746871e-01 4.38082725e-01
1.65421292e-01 1.46204174e-01 5.33908010e-01 5.14366627e-01
-3.14686060e-01 -4.04466450e-01 2.78985709e-01 2.20002696e-01
2.37251833e-01 -9.37933207e-01 5.83765864e-01 1.11493278e+00
-1.04804099e-01 -7.09880292e-02 1.09228325e+00 -8.06694925e-01
-1.41590035e+00 -1.36907434e+00 8.30055952e-01 1.47559226e-01
7.09304094e-01 9.75905266e-03 -1.38578427e+00 7.29054153e-01
-4.94558752e-01 5.74763596e-01 2.86847562e-01 5.01344502e-02
5.23908377e-01 -3.60884011e-01 -1.12158656e+00 1.37721300e-01
6.52959645e-01 -2.61814624e-01 -6.59700274e-01 4.91179079e-01
7.81238675e-02 -4.58320498e-01 -6.58920407e-01 8.80202413e-01
2.53132731e-01 -9.31010067e-01 1.02460778e+00 -3.87340263e-02
2.80825704e-01 -3.33224446e-01 -1.58052780e-02 -1.31293678e+00
-3.89843762e-01 -3.99535537e-01 4.60882932e-01 8.07763398e-01
-2.55355723e-02 -6.30605221e-01 1.00402915e+00 -1.37251586e-01
-4.62140441e-01 -2.21769810e-01 -1.06809902e+00 -8.58130276e-01
-2.30729267e-01 -4.95309651e-01 2.99354374e-01 8.84565830e-01
-5.19907534e-01 3.61753665e-02 -4.53685611e-01 7.93467045e-01
9.44873333e-01 2.64777750e-01 4.22292113e-01 -1.69228494e+00
2.78721303e-01 8.30519125e-02 -8.10391545e-01 -6.07580841e-01
-1.01758331e-01 -5.23872495e-01 4.43698853e-01 -1.70256627e+00
-6.96620882e-01 -4.02519047e-01 -6.19001985e-02 4.15511668e-01
-4.42927470e-03 7.03163564e-01 -3.65890622e-01 3.07340920e-01
2.74719864e-01 6.40243173e-01 1.07807899e+00 -1.08889826e-01
-1.61113054e-01 3.84625226e-01 -5.15342318e-02 9.52117920e-01
4.86307532e-01 -6.58483565e-01 2.94498354e-01 -5.54157794e-01
3.57671410e-01 3.75855625e-01 5.02457142e-01 -1.33318150e+00
-9.57593545e-02 -9.63502601e-02 2.27842942e-01 -9.28365648e-01
2.93226779e-01 -7.91115701e-01 3.84733468e-01 7.23484933e-01
3.77031595e-01 -1.06343105e-01 2.18815371e-01 5.03801823e-01
-9.18533921e-01 -9.84911770e-02 1.11199808e+00 -1.92451134e-01
-8.68793786e-01 3.07917923e-01 -7.54787922e-01 -1.16934031e-01
6.90131962e-01 -1.69350654e-01 -1.02557905e-01 1.82110876e-01
-9.93297815e-01 2.30978072e-01 3.75241667e-01 1.01935677e-01
8.25735986e-01 -1.03274560e+00 -1.08928144e+00 6.95945084e-01
1.17617562e-01 3.78602117e-01 7.79605031e-01 9.58828688e-01
-1.04611695e+00 -1.44103169e-01 -4.25321907e-01 -8.14162850e-01
-1.02009690e+00 3.51634552e-03 4.41170990e-01 -2.04480708e-01
-1.33012807e+00 5.65536857e-01 -3.90702724e-01 -1.36756212e-01
-6.06184602e-01 -2.41276383e-01 -2.58477509e-01 5.35800695e-01
4.26861435e-01 2.82067001e-01 6.22728527e-01 -7.28180766e-01
-4.71974641e-01 7.40172625e-01 3.61671656e-01 9.11483690e-02
2.02301264e+00 -1.87973693e-01 -1.90266445e-01 3.22592407e-01
7.86799669e-01 -3.32833201e-01 -1.20090032e+00 -3.69181722e-01
1.68523595e-01 -6.07966900e-01 4.99453276e-01 -1.46410093e-01
-1.20891774e+00 8.44267368e-01 5.19772649e-01 2.15930521e-01
9.96688366e-01 5.79756536e-02 7.15155482e-01 5.40719092e-01
3.64753962e-01 -9.32129920e-01 -4.45341349e-01 5.31146884e-01
8.96397531e-01 -1.24721205e+00 3.85464042e-01 -2.98480004e-01
-3.09833914e-01 1.22377598e+00 6.11559115e-02 -5.18160284e-01
8.39396477e-01 5.10203779e-01 5.89678213e-02 -6.75523937e-01
-1.63980275e-01 -5.59815645e-01 -8.23959261e-02 5.74490666e-01
2.58327514e-01 1.08641133e-01 -3.21518004e-01 2.88075348e-03
-7.40228221e-02 -6.76967576e-03 6.92633808e-01 1.47239375e+00
-7.81257868e-01 -6.95423782e-01 -7.15615213e-01 5.72465301e-01
-4.46577519e-01 8.26860145e-02 5.37274718e-01 6.87921584e-01
1.63714245e-01 4.10402566e-01 4.82406020e-01 2.08371565e-01
5.95104992e-01 -3.60831529e-01 4.04146999e-01 -6.15099192e-01
-3.91375780e-01 2.75910258e-01 2.45015964e-01 -3.71204764e-01
-6.79654479e-01 -1.03030622e+00 -1.20838320e+00 1.57364160e-01
-3.06291282e-01 9.26356763e-02 6.72811568e-01 1.38766539e+00
-6.17131367e-02 5.19693315e-01 6.25171900e-01 -1.15128088e+00
-1.00820750e-01 -8.52056026e-01 -1.21500444e+00 2.90928751e-01
7.01301217e-01 -7.95623362e-01 -3.61507148e-01 6.53292835e-02]
|
[10.015395164489746, -1.953804850578308]
|
d90db96d-ebc4-4deb-a1ac-8c8a213b31ee
|
local-implicit-grid-representations-for-3d
|
2003.08981
| null |
https://arxiv.org/abs/2003.08981v1
|
https://arxiv.org/pdf/2003.08981v1.pdf
|
Local Implicit Grid Representations for 3D Scenes
|
Shape priors learned from data are commonly used to reconstruct 3D objects from partial or noisy data. Yet no such shape priors are available for indoor scenes, since typical 3D autoencoders cannot handle their scale, complexity, or diversity. In this paper, we introduce Local Implicit Grid Representations, a new 3D shape representation designed for scalability and generality. The motivating idea is that most 3D surfaces share geometric details at some scale -- i.e., at a scale smaller than an entire object and larger than a small patch. We train an autoencoder to learn an embedding of local crops of 3D shapes at that size. Then, we use the decoder as a component in a shape optimization that solves for a set of latent codes on a regular grid of overlapping crops such that an interpolation of the decoded local shapes matches a partial or noisy observation. We demonstrate the value of this proposed approach for 3D surface reconstruction from sparse point observations, showing significantly better results than alternative approaches.
|
['Thomas Funkhouser', 'Matthias Nießner', 'Chiyu Max Jiang', 'Avneesh Sud', 'Jingwei Huang', 'Ameesh Makadia']
|
2020-03-19
| null | null | null | null |
['3d-shape-representation']
|
['computer-vision']
|
[ 1.26719266e-01 3.40461761e-01 1.80276394e-01 -2.77920216e-01
-6.73498690e-01 -4.09265369e-01 3.96273971e-01 1.61838140e-02
4.30832595e-01 3.29393566e-01 3.49892974e-01 5.54473773e-02
-7.62171894e-02 -1.22709692e+00 -1.15915668e+00 -7.45261967e-01
6.16027825e-02 7.39302218e-01 1.42008513e-01 7.89193902e-03
4.02835943e-02 1.10718989e+00 -1.71163762e+00 3.00675184e-01
3.47215354e-01 1.03357923e+00 7.17814505e-01 4.64163870e-01
-2.39929765e-01 1.30915016e-01 -3.01501244e-01 7.96769783e-02
3.86117637e-01 -1.60150174e-02 -3.42613399e-01 7.86817908e-01
3.94751042e-01 -5.62623322e-01 -2.30811715e-01 8.25674593e-01
3.99937779e-01 -1.10319957e-01 1.04986227e+00 -6.42592192e-01
-8.72775853e-01 -1.42650064e-02 -3.63193691e-01 -6.15478754e-01
4.25360680e-01 -2.34129921e-01 8.86489272e-01 -1.22576761e+00
7.60216355e-01 1.17336154e+00 9.66330647e-01 2.23494262e-01
-1.52105069e+00 -1.56784207e-02 6.40449971e-02 -4.14605379e-01
-1.51279974e+00 -3.73581439e-01 1.07830799e+00 -4.25565004e-01
1.02855802e+00 7.71335736e-02 7.08829939e-01 1.06260800e+00
1.28692567e-01 7.08657146e-01 8.27814341e-01 -4.42784458e-01
5.70063651e-01 1.00554734e-01 -3.78357589e-01 6.33033693e-01
2.91826218e-01 1.80296581e-02 -3.36762249e-01 -4.66233522e-01
1.46252537e+00 3.33633959e-01 -2.58971184e-01 -6.96799397e-01
-9.01192784e-01 8.86357188e-01 4.48081583e-01 3.95898223e-01
-7.56820917e-01 1.41996086e-01 -3.12776655e-01 1.79206774e-01
8.67820263e-01 9.58971605e-02 -5.68884075e-01 3.78036499e-01
-8.02437723e-01 1.86622217e-01 9.17005301e-01 1.15542650e+00
1.12865794e+00 1.90317452e-01 3.42496604e-01 8.18408251e-01
4.99724478e-01 8.69096577e-01 1.13887973e-02 -1.05416095e+00
-2.08037183e-01 6.86719179e-01 3.02977979e-01 -1.05871654e+00
-7.22126067e-02 -2.45130420e-01 -9.16437685e-01 4.60683435e-01
1.02275960e-01 2.90299922e-01 -8.02214444e-01 1.39561570e+00
6.14362240e-01 3.32781404e-01 1.86672620e-02 8.57009709e-01
8.26622725e-01 8.01542521e-01 -5.90572357e-01 2.27743343e-01
1.09665096e+00 -2.23961979e-01 -3.64566714e-01 -2.07793593e-01
9.38160717e-02 -8.21555257e-01 7.69453883e-01 2.59078503e-01
-1.24725723e+00 -4.84242618e-01 -8.41263294e-01 -2.13872522e-01
-1.74773946e-01 3.03466737e-01 5.00564814e-01 4.00174856e-01
-1.19880486e+00 6.36875391e-01 -8.73886168e-01 -3.20079803e-01
3.30848455e-01 7.09761828e-02 -4.27861899e-01 -1.68763652e-01
-3.06339681e-01 5.43909669e-01 -1.26954526e-01 -1.41815394e-01
-1.01760876e+00 -5.90041876e-01 -1.00018847e+00 6.69745505e-02
-4.58106538e-03 -6.62905633e-01 9.03680861e-01 -7.64692843e-01
-1.56193721e+00 1.08870161e+00 -4.36990201e-01 -2.28139415e-01
-9.81360376e-02 -1.54480323e-01 8.56097192e-02 -2.76551526e-02
2.11695492e-01 6.66816652e-01 1.23938918e+00 -1.84262753e+00
-7.08383024e-02 -5.83242357e-01 -6.95570782e-02 1.70435831e-01
1.85482517e-01 -6.59252346e-01 -3.98147345e-01 -6.43419802e-01
7.57088482e-01 -7.40205705e-01 -2.66670108e-01 4.87778008e-01
-2.14397848e-01 9.21831746e-03 8.81247401e-01 -6.87297344e-01
3.29803765e-01 -2.37539577e+00 2.96627134e-01 3.69428635e-01
-2.14686636e-02 -4.17274773e-01 -3.40414613e-01 5.19632936e-01
9.06554237e-02 -7.44962990e-02 -5.13552129e-01 -7.30064988e-01
3.49763222e-02 6.47726059e-01 -4.09131408e-01 5.14171004e-01
4.76575732e-01 7.43214488e-01 -5.68118036e-01 1.78770602e-01
2.71819025e-01 9.11557794e-01 -7.51841664e-01 4.03692782e-01
-5.56727052e-01 2.78679162e-01 -6.14610851e-01 8.30495477e-01
9.52874780e-01 -5.13370693e-01 -8.94970521e-02 -1.82794899e-01
-2.96810597e-01 1.31130785e-01 -1.44683194e+00 1.98499310e+00
-5.47981977e-01 3.57503027e-01 4.36334014e-01 -1.03287876e+00
1.44495547e+00 4.93346483e-01 6.27380371e-01 -8.44255462e-02
-2.99191400e-02 3.07943642e-01 -8.01383197e-01 -2.26653516e-01
3.57370228e-01 -1.05878413e-01 2.58428872e-01 3.34831089e-01
5.36131673e-02 -8.74673486e-01 -6.75721943e-01 -2.89906859e-01
9.81325507e-01 2.42186666e-01 3.10297251e-01 -2.11160243e-01
1.76936716e-01 -1.50558859e-01 2.78690785e-01 5.95243394e-01
4.29678410e-01 1.13903749e+00 1.66437432e-01 -6.20818436e-01
-1.60263109e+00 -1.09881830e+00 -2.99288839e-01 4.20392066e-01
1.25904202e-01 9.43094585e-03 -4.41326439e-01 -2.32276529e-01
3.86125743e-01 5.40177763e-01 -8.29111159e-01 2.31439382e-01
-3.48530591e-01 -2.32046261e-01 -4.03830335e-02 4.25853968e-01
2.35111654e-01 -1.04301894e+00 -7.19822228e-01 2.47740269e-01
1.87558472e-01 -9.82104778e-01 -2.09529266e-01 3.48103762e-01
-1.24903524e+00 -7.59805620e-01 -9.16117013e-01 -1.10044789e+00
1.01426816e+00 5.15274584e-01 1.21175039e+00 1.27013202e-03
-1.24029696e-01 6.82321608e-01 -3.99628133e-01 -3.12146455e-01
-3.83661866e-01 -4.15715128e-01 -1.37016013e-01 1.67844504e-01
1.41204596e-01 -1.00641549e+00 -2.44238868e-01 1.40071556e-01
-7.93208539e-01 3.45157348e-02 5.68998933e-01 9.08298016e-01
1.21624064e+00 -2.80449260e-03 1.63412377e-01 -3.81479412e-01
1.94613665e-01 -5.30942738e-01 -6.88744366e-01 -8.32035840e-02
1.15023308e-01 2.04604000e-01 4.28993076e-01 -3.55169713e-01
-9.02072847e-01 4.51216429e-01 -5.01949668e-01 -7.42352009e-01
-7.75608242e-01 2.89158046e-01 -2.50112504e-01 -1.68809354e-01
6.53631032e-01 5.88998616e-01 1.68237075e-01 -8.01533520e-01
3.26409310e-01 5.09218156e-01 1.60840794e-01 -6.08344734e-01
6.81060791e-01 8.86688054e-01 4.08115238e-02 -1.31777859e+00
-5.07025599e-01 -3.43350947e-01 -6.72816634e-01 8.43288898e-02
7.73694098e-01 -1.03421450e+00 -4.29603964e-01 1.87679023e-01
-1.41930246e+00 -5.07680595e-01 -8.57411563e-01 3.35481375e-01
-9.24018323e-01 2.16339543e-01 -3.74711275e-01 -8.39005888e-01
-2.72718910e-02 -8.84821296e-01 1.77588594e+00 -7.14941397e-02
1.39128834e-01 -8.66673410e-01 2.07216933e-01 -1.23663381e-01
3.95626307e-01 4.53282386e-01 8.55067611e-01 -1.20237306e-01
-8.88455451e-01 -2.24854663e-01 -2.40042865e-01 3.04836541e-01
3.08898300e-01 -3.58436942e-01 -1.06924891e+00 -2.41979510e-01
3.26882869e-01 -2.00920328e-01 5.70194721e-01 8.84414077e-01
1.11709368e+00 -2.57747680e-01 -3.00797731e-01 6.43567324e-01
1.70989811e+00 -3.78728881e-02 4.63544041e-01 -1.62913889e-01
3.31678689e-01 4.91336256e-01 4.86333184e-02 6.99362695e-01
2.10870206e-01 3.37105095e-01 7.47604012e-01 8.03498179e-02
-2.32976645e-01 -3.86619329e-01 4.98250388e-02 8.24868321e-01
-2.11328909e-01 -1.44531354e-01 -6.02086306e-01 8.63278449e-01
-1.46669996e+00 -5.54703236e-01 1.51083410e-01 2.16549134e+00
4.70170528e-01 -1.73826605e-01 -4.03928846e-01 5.99216372e-02
4.01472092e-01 1.69575766e-01 -5.58883965e-01 -1.11549616e-01
-4.42336440e-01 2.85472870e-01 3.18802238e-01 6.63612187e-01
-8.47188592e-01 7.71266043e-01 6.21413326e+00 6.60758674e-01
-9.17883694e-01 -1.88098121e-02 3.45302492e-01 4.75953519e-01
-8.46017122e-01 -1.33433156e-02 -6.38278246e-01 -4.01092246e-02
3.32671136e-01 4.71892059e-01 6.09276056e-01 1.00558782e+00
-2.46434398e-02 2.15930934e-03 -1.05463970e+00 1.02360094e+00
9.84887332e-02 -1.53535140e+00 3.00162341e-02 7.80210122e-02
1.00624335e+00 7.70578459e-02 -1.25777768e-02 -1.99171945e-01
4.10517842e-01 -8.86410534e-01 9.11296308e-01 5.94484687e-01
7.35655963e-01 -3.95452529e-01 4.15932149e-01 6.54509187e-01
-1.12404287e+00 2.52712816e-01 -9.33634162e-01 -8.36137980e-02
2.22005516e-01 9.10345435e-01 -7.14023948e-01 3.86744618e-01
7.33484328e-01 7.11683869e-01 1.02149047e-01 8.08471978e-01
-1.49180636e-01 3.20993572e-01 -7.84609020e-01 -1.01034358e-01
1.11675069e-01 -2.92143673e-01 7.06234813e-01 8.79121542e-01
1.01560009e+00 3.56672555e-01 1.39063492e-01 1.36688566e+00
-4.81373891e-02 -6.89860582e-02 -1.30515516e+00 2.03218266e-01
5.93657494e-01 7.51083672e-01 -5.68708062e-01 -1.60959497e-01
-6.43212914e-01 1.02323925e+00 2.11808473e-01 4.07434881e-01
-1.22822754e-01 1.01663478e-01 6.74559653e-01 2.60960072e-01
8.76323700e-01 -6.16044164e-01 -6.23236835e-01 -9.75819826e-01
-9.18702856e-02 -6.23549521e-01 -2.23595813e-01 -1.14321125e+00
-1.49643469e+00 3.13962728e-01 -2.68837214e-01 -1.24332261e+00
-3.60569241e-03 -6.51663601e-01 -3.53775740e-01 8.24067771e-01
-1.24073720e+00 -1.28726375e+00 -3.42538387e-01 5.34904659e-01
5.13101518e-01 -1.01205342e-01 1.38925481e+00 -2.26444677e-01
3.72142017e-01 -1.86181366e-01 3.20766062e-01 -1.58485457e-01
2.96449903e-02 -1.18700111e+00 4.87060130e-01 4.17179227e-01
4.74520415e-01 1.64837778e-01 4.77767855e-01 -7.81187892e-01
-1.71916509e+00 -8.56952488e-01 8.47292662e-01 -4.14943427e-01
1.63745999e-01 -4.07845736e-01 -8.99300814e-01 7.75313437e-01
2.97573209e-02 1.43203616e-01 4.93063748e-01 -6.99013397e-02
-1.98958054e-01 2.27210462e-01 -1.34305751e+00 3.58533591e-01
1.16220307e+00 -6.13000035e-01 -6.74048066e-01 1.80670336e-01
7.31174231e-01 -5.46310663e-01 -1.03935564e+00 3.91292393e-01
3.95236135e-01 -1.06754041e+00 1.33409560e+00 -1.14488676e-01
4.54390645e-01 -3.05326074e-01 -8.87558877e-01 -1.40454721e+00
-4.90589738e-01 -2.79082537e-01 -2.90419668e-01 6.69851840e-01
2.11776141e-02 -3.49229574e-01 1.12298954e+00 1.70007899e-01
-4.59731668e-01 -6.33614898e-01 -9.29849267e-01 -5.32665491e-01
-3.29047889e-02 -4.42077130e-01 9.09352899e-01 8.06852698e-01
-7.46388257e-01 -3.74952927e-02 -2.91050613e-01 7.12440789e-01
9.20283079e-01 7.52479911e-01 8.82508397e-01 -1.36971009e+00
-2.67375410e-01 -6.70487806e-02 -3.96083832e-01 -1.73768651e+00
1.09168604e-01 -8.07793438e-01 1.20104045e-01 -1.58026016e+00
-1.65239826e-01 -5.91839612e-01 3.19783747e-01 3.14587444e-01
3.78238291e-01 1.19789295e-01 -1.51765883e-01 5.52177094e-02
8.28420222e-02 8.87452364e-01 1.38129306e+00 -2.31867790e-01
-2.53701985e-01 -5.30547760e-02 -4.18798417e-01 7.60093868e-01
5.79066515e-01 -3.77324909e-01 -1.49389327e-01 -8.59791577e-01
1.55896053e-01 2.18374327e-01 5.80793023e-01 -8.47132742e-01
1.17956288e-01 -2.66877748e-02 6.81888938e-01 -1.01514971e+00
7.42194355e-01 -1.29070318e+00 5.08252859e-01 3.34158689e-02
9.14440602e-02 -4.05747682e-01 2.51296699e-01 7.76277184e-01
-8.20558816e-02 -2.76819646e-01 5.42571962e-01 -3.50399166e-01
-3.93610179e-01 4.51073378e-01 -2.39157349e-01 -4.01254475e-01
6.23525262e-01 -4.72791076e-01 4.67194200e-01 -3.78039390e-01
-9.21336889e-01 -4.41779852e-01 8.20407391e-01 1.97329819e-01
1.03602934e+00 -1.65393794e+00 -8.16150427e-01 8.78059447e-01
-2.01807674e-02 5.88741302e-01 1.92380175e-01 1.64866433e-01
-5.64420879e-01 2.61990994e-01 -2.63442189e-01 -9.55329955e-01
-7.68332839e-01 3.07374805e-01 2.15073392e-01 1.80732042e-01
-1.02321911e+00 9.13122535e-01 3.82804990e-01 -7.27490067e-01
1.88074932e-01 -6.32158399e-01 1.03189088e-01 -3.92399549e-01
1.17583551e-01 1.22297956e-02 3.95057201e-02 -6.39094412e-01
1.88275687e-02 1.11383533e+00 6.36505127e-01 1.44825265e-01
1.92354620e+00 3.15928943e-02 -1.79161251e-01 5.36503196e-01
1.12502027e+00 1.14372514e-01 -1.82057643e+00 -4.88131404e-01
-2.69899726e-01 -6.77012026e-01 1.23157002e-01 -2.13868365e-01
-9.15979326e-01 8.24432671e-01 5.06952107e-01 4.50102121e-01
9.72766280e-01 3.99272352e-01 5.30809224e-01 4.29846108e-01
6.57263279e-01 -6.70819104e-01 4.72173840e-02 7.03246236e-01
1.36381209e+00 -1.14042020e+00 -1.41130298e-01 -4.12471473e-01
-1.30284458e-01 1.15971017e+00 -1.06043488e-01 -7.42294014e-01
1.10925794e+00 5.15678883e-01 -3.61909777e-01 -4.10935134e-01
-5.71045637e-01 -1.89172342e-01 3.20812792e-01 8.66986752e-01
1.83530197e-01 1.88257515e-01 3.44709545e-01 4.93488699e-01
-2.26424560e-02 -1.13527298e-01 1.73734054e-01 8.15182507e-01
-6.50623143e-01 -9.40765858e-01 -6.52677000e-01 3.16297889e-01
7.29485154e-02 1.74345970e-01 -1.95884556e-01 4.10255581e-01
2.43525684e-01 5.10075450e-01 2.81803697e-01 -7.30561689e-02
4.06880796e-01 1.10715060e-02 7.02726901e-01 -6.11369967e-01
1.97587967e-01 4.28678662e-01 -2.75481105e-01 -5.52791297e-01
-4.03603524e-01 -8.01316619e-01 -9.58440721e-01 -2.92154029e-02
-2.38392875e-01 -1.91227704e-01 8.21927309e-01 5.70885897e-01
4.80198413e-01 -8.46221484e-03 7.11658001e-01 -1.50722504e+00
-3.51222605e-01 -7.25746930e-01 -9.62830663e-01 3.01312536e-01
5.22478044e-01 -6.23157322e-01 -3.93551320e-01 2.72764653e-01]
|
[8.70195484161377, -3.584632396697998]
|
f56f94ce-88fa-4dd9-8419-02b950a405dc
|
adversarial-pretraining-of-self-supervised
|
2210.13463
| null |
https://arxiv.org/abs/2210.13463v1
|
https://arxiv.org/pdf/2210.13463v1.pdf
|
Adversarial Pretraining of Self-Supervised Deep Networks: Past, Present and Future
|
In this paper, we review adversarial pretraining of self-supervised deep networks including both convolutional neural networks and vision transformers. Unlike the adversarial training with access to labeled examples, adversarial pretraining is complicated as it only has access to unlabeled examples. To incorporate adversaries into pretraining models on either input or feature level, we find that existing approaches are largely categorized into two groups: memory-free instance-wise attacks imposing worst-case perturbations on individual examples, and memory-based adversaries shared across examples over iterations. In particular, we review several representative adversarial pretraining models based on Contrastive Learning (CL) and Masked Image Modeling (MIM), respectively, two popular self-supervised pretraining methods in literature. We also review miscellaneous issues about computing overheads, input-/feature-level adversaries, as well as other adversarial pretraining approaches beyond the above two groups. Finally, we discuss emerging trends and future directions about the relations between adversarial and cooperative pretraining, unifying adversarial CL and MIM pretraining, and the trade-off between accuracy and robustness in adversarial pretraining.
|
['Mubarak Shah', 'Guo-Jun Qi']
|
2022-10-23
| null | null | null | null |
['miscellaneous']
|
['miscellaneous']
|
[ 4.78441983e-01 2.88647741e-01 1.94609433e-01 -4.07016397e-01
-4.87537920e-01 -1.07381761e+00 6.01519644e-01 -3.16655278e-01
-4.91340697e-01 7.90476918e-01 -3.72651778e-02 -4.75092500e-01
2.59841800e-01 -1.01296115e+00 -1.06657827e+00 -8.11118782e-01
-2.25067630e-01 2.15343207e-01 1.36735260e-01 -2.10302576e-01
-2.23991707e-01 8.22126389e-01 -1.13093948e+00 4.44059253e-01
9.55674171e-01 8.93073797e-01 -5.21521747e-01 9.23324525e-01
4.79354784e-02 1.25621665e+00 -1.07660079e+00 -1.00347877e+00
6.66746199e-01 -2.81095535e-01 -8.71457160e-01 2.47928817e-02
7.39973843e-01 -4.50585753e-01 -1.02799201e+00 1.49442995e+00
7.89235353e-01 6.85530677e-02 5.65967202e-01 -1.49538112e+00
-1.20115411e+00 8.47007453e-01 -1.92364603e-01 3.01721126e-01
-3.37559700e-01 6.18284702e-01 1.84159055e-01 -7.23041177e-01
1.68236509e-01 1.14413524e+00 7.42306471e-01 9.42365646e-01
-9.30978298e-01 -1.05297232e+00 4.04062778e-01 1.14997767e-01
-1.17453456e+00 -4.62626457e-01 7.00167954e-01 -4.51427579e-01
9.68833208e-01 2.79597044e-01 9.35464799e-02 1.22435391e+00
3.50775123e-01 6.71220243e-01 1.22464240e+00 -2.54046381e-01
2.31495172e-01 2.68197268e-01 -1.71278030e-01 6.61834598e-01
1.18235499e-01 9.55372274e-01 -1.95959672e-01 -1.91777349e-01
8.37603152e-01 1.23950407e-01 -8.71407539e-02 -1.18762948e-01
-8.16890836e-01 7.54962862e-01 8.05860579e-01 -1.64111197e-01
-4.19496670e-02 3.36458266e-01 6.53429747e-01 7.74841309e-01
3.57675731e-01 3.93143356e-01 -3.97178352e-01 6.26685619e-01
-7.71834195e-01 -1.13319241e-01 4.60989803e-01 1.08725822e+00
9.49796975e-01 9.56414044e-01 -2.21207216e-01 4.42221165e-01
-3.40516283e-03 6.76579833e-01 5.91213226e-01 -5.17442167e-01
4.25322682e-01 1.69399381e-01 -3.75765294e-01 -6.66758597e-01
-6.39183521e-02 -6.59173071e-01 -1.33365405e+00 7.21797645e-01
-4.73813079e-02 -8.25052917e-01 -1.48732269e+00 1.75545669e+00
7.21584260e-02 6.04735613e-01 6.52256310e-01 6.38774753e-01
1.22714972e+00 3.33191156e-01 2.52931446e-01 -1.54410034e-01
9.07277822e-01 -1.34432745e+00 -6.12069428e-01 -7.13106334e-01
1.12401590e-01 -8.25960636e-01 6.64974391e-01 6.73402250e-02
-1.30962026e+00 -9.35941577e-01 -1.13645911e+00 3.57363462e-01
-6.97275162e-01 -2.87132949e-01 5.47155082e-01 1.03174341e+00
-1.10056877e+00 8.74316752e-01 -6.48525894e-01 2.98305303e-01
8.12615395e-01 7.48826563e-01 -3.28188211e-01 1.06800444e-01
-1.48421633e+00 8.65527689e-01 4.42213774e-01 -4.39109392e-02
-1.61690593e+00 -7.52617598e-01 -1.04087317e+00 -1.84665114e-01
1.19325146e-01 -5.66532493e-01 1.11003673e+00 -1.59332609e+00
-1.35578215e+00 9.92821515e-01 2.89374799e-01 -7.69955695e-01
7.23270953e-01 -2.41879061e-01 -8.20895851e-01 -3.11488155e-02
-3.00206989e-01 5.24730980e-01 1.19516456e+00 -1.47142065e+00
-2.33735785e-01 -5.39124794e-02 2.23748907e-01 1.91439480e-01
-4.04990524e-01 6.73271492e-02 -1.57915995e-01 -1.08322859e+00
-1.94018766e-01 -7.30436325e-01 -6.67807460e-01 -1.92760862e-02
-6.48621857e-01 4.51040417e-01 1.05857325e+00 -2.29383901e-01
9.18498099e-01 -2.35945630e+00 -1.95165768e-01 2.23715857e-01
4.12897676e-01 8.61922503e-01 -3.08161914e-01 2.63542861e-01
-6.56124771e-01 2.26281896e-01 -3.09301853e-01 -3.32801789e-01
-9.33197811e-02 3.67864996e-01 -8.62730622e-01 5.42677999e-01
5.52060068e-01 1.26027083e+00 -9.44135368e-01 -2.99528897e-01
1.61197498e-01 1.56712562e-01 -4.10081148e-01 7.27608263e-01
3.15526091e-02 5.61010480e-01 1.30935153e-02 7.35379219e-01
9.92185116e-01 1.25362605e-01 -1.13887720e-01 -3.05480845e-02
4.81410295e-01 -1.57845110e-01 -1.06143904e+00 1.07202613e+00
-2.76585668e-01 4.74277914e-01 1.55692071e-01 -8.80043983e-01
7.71054089e-01 4.97660309e-01 -7.77369961e-02 -3.51192862e-01
1.93912670e-01 4.57819812e-02 2.37223413e-02 -2.56490648e-01
3.67772877e-01 -3.22903246e-01 -1.25461712e-01 4.45823461e-01
4.42909986e-01 -2.35302709e-02 -3.41018349e-01 2.35982835e-01
1.18486702e+00 -1.80424586e-01 2.58672357e-01 -1.49189855e-03
5.85651398e-01 -1.19241580e-01 4.11311746e-01 1.01857746e+00
-4.89132106e-01 6.93253279e-01 4.55508456e-02 -4.78395104e-01
-7.45209932e-01 -1.47644389e+00 1.20813534e-01 1.12856758e+00
1.43755674e-01 -1.87074840e-02 -7.66080916e-01 -1.29352736e+00
-1.66169312e-02 2.38998368e-01 -8.50077033e-01 -7.75297284e-01
-6.45268142e-01 -7.13653684e-01 1.18358183e+00 8.86603177e-01
1.04895341e+00 -1.36199760e+00 8.65259245e-02 -4.92528267e-02
3.01748037e-01 -1.06496978e+00 -4.61198092e-01 7.78208733e-01
-8.81129682e-01 -9.60896790e-01 -5.41030645e-01 -1.19890273e+00
1.24498725e+00 1.73711907e-02 1.44335330e+00 2.97981769e-01
-9.21019614e-02 3.19615960e-01 -2.13105008e-01 -6.57636225e-01
-8.65739882e-01 -3.17692190e-01 4.35932487e-01 -1.75636709e-01
8.52085426e-02 -8.13191175e-01 -4.77609813e-01 3.61363113e-01
-1.10387790e+00 -4.99092281e-01 7.98101544e-01 1.09712422e+00
5.50881803e-01 2.62102425e-01 6.34315968e-01 -1.32302082e+00
3.00398976e-01 -6.48954213e-01 -3.92663985e-01 1.76293701e-01
-4.77902114e-01 -2.97387749e-01 1.18146920e+00 -9.76776600e-01
-7.91035712e-01 1.12757899e-01 -4.16146636e-01 -1.04965329e+00
-2.68164009e-01 1.12362206e-01 -4.47718412e-01 -6.30189717e-01
1.27968717e+00 3.68241251e-01 -3.57205793e-02 1.92433354e-02
5.16783953e-01 3.66776764e-01 9.67099130e-01 -4.64422494e-01
1.59058893e+00 3.23901296e-01 -3.56602341e-01 -2.50432849e-01
-6.41545296e-01 1.47578254e-01 -6.48022950e-01 -1.08099140e-01
4.72903490e-01 -1.02614701e+00 -3.65345806e-01 8.70670855e-01
-9.47584450e-01 -6.65779412e-01 -9.15236831e-01 5.76127917e-02
-5.37968814e-01 4.04095590e-01 -8.72246265e-01 -6.64961338e-01
-5.14717221e-01 -1.29452288e+00 2.61455685e-01 2.32861504e-01
1.72515929e-01 -1.26672328e+00 -1.37501970e-01 1.76622003e-01
6.08919084e-01 5.86649358e-01 5.73178470e-01 -1.10526812e+00
-5.26145160e-01 -4.71926242e-01 2.53481343e-02 8.08107853e-01
6.88836202e-02 -2.35116974e-01 -1.40126455e+00 -6.50078058e-01
8.35659131e-02 -7.65530944e-01 8.52867782e-01 7.12430030e-02
1.20046115e+00 -7.43604124e-01 -1.47289857e-01 1.05677724e+00
1.47400880e+00 1.41481534e-01 9.42789674e-01 2.22659290e-01
8.74006093e-01 2.47989640e-01 3.98562998e-01 3.46783884e-02
-2.94969112e-01 -1.42706871e-01 1.02820361e+00 -6.11800313e-01
-2.15617284e-01 -2.14039460e-01 6.06933475e-01 4.26691413e-01
1.22870393e-01 -3.94046813e-01 -6.10689700e-01 3.63293648e-01
-1.44215477e+00 -1.01132405e+00 3.47349435e-01 2.02901554e+00
1.08609796e+00 3.73868287e-01 -9.37428325e-02 2.18358561e-01
1.00364625e+00 2.83702374e-01 -8.89114976e-01 -5.27278066e-01
-2.60819763e-01 5.96301436e-01 8.51411104e-01 5.44228077e-01
-1.62337625e+00 1.29974747e+00 7.66211557e+00 7.20884919e-01
-1.19981980e+00 2.73167133e-01 7.73925722e-01 4.77793850e-02
-1.39548585e-01 -2.52319604e-01 -5.86878479e-01 3.00527036e-01
8.64769042e-01 1.61772724e-02 4.89412546e-01 1.10443723e+00
-5.68678200e-01 4.81261849e-01 -1.07254577e+00 5.73843241e-01
3.12692689e-04 -1.40564930e+00 2.32363164e-01 -2.56270021e-01
1.21516454e+00 1.72096834e-01 5.85987031e-01 5.74570715e-01
1.06564760e+00 -1.46574867e+00 5.42351961e-01 1.05853178e-01
1.23183107e+00 -9.25022662e-01 9.51040566e-01 1.86527163e-01
-1.02256072e+00 -1.53365076e-01 -3.50534379e-01 3.31449918e-02
-2.74843365e-01 2.03896314e-01 -4.67954636e-01 5.22053957e-01
7.28810668e-01 4.07367975e-01 -6.46855235e-01 5.91547966e-01
-3.74350965e-01 7.86103368e-01 1.95507601e-01 4.06689942e-01
2.18258560e-01 4.05654907e-01 3.97743136e-01 1.20265543e+00
-1.84508681e-01 1.40478805e-01 1.26713589e-01 6.34043217e-01
-3.93906057e-01 -3.20329875e-01 -6.28778100e-01 7.31341308e-03
7.44217396e-01 9.82950211e-01 -3.45411748e-01 -4.36480850e-01
-3.83972555e-01 1.36299217e+00 3.89664292e-01 5.01443744e-01
-1.05429828e+00 -4.23093736e-01 9.59586084e-01 -9.79840308e-02
9.51851159e-02 -6.93330169e-02 -4.66004550e-01 -9.31657851e-01
-3.11109841e-01 -1.27245569e+00 4.01891738e-01 -3.25626135e-01
-1.62895727e+00 1.04473531e+00 -3.79377007e-01 -1.51385212e+00
-8.83265883e-02 -6.56242311e-01 -1.22998238e+00 1.00124145e+00
-1.61329842e+00 -1.22810090e+00 -3.52791339e-01 1.13851249e+00
3.30576658e-01 -7.76090443e-01 1.11469257e+00 3.38193148e-01
-6.87524557e-01 1.53942335e+00 1.86187446e-01 7.10028589e-01
9.13643837e-01 -1.26275289e+00 8.94976735e-01 1.38982916e+00
-8.67349058e-02 6.63919926e-01 5.91859341e-01 -6.56186461e-01
-9.56591129e-01 -1.89199924e+00 1.66693017e-01 -4.09131736e-01
6.04367316e-01 -1.16539761e-01 -9.45336342e-01 1.14219332e+00
4.04558718e-01 6.14652455e-01 9.55262721e-01 -4.09742534e-01
-5.17039895e-01 -4.19872329e-02 -1.62697101e+00 7.51128495e-01
1.13421094e+00 -6.08926117e-01 -3.33046049e-01 4.81193453e-01
8.55156124e-01 -8.92920792e-01 -8.51063311e-01 5.57835340e-01
-1.16554476e-01 -9.62166131e-01 1.35148537e+00 -8.97293448e-01
4.72222686e-01 -2.32631236e-01 3.27657610e-02 -1.43472672e+00
-4.77276534e-01 -9.35517550e-01 -3.73570681e-01 1.27006865e+00
2.68520862e-01 -8.63947093e-01 9.90536332e-01 4.11480993e-01
-4.87094790e-01 -5.94854474e-01 -6.36487305e-01 -9.12670434e-01
6.05105937e-01 -2.36516729e-01 6.57466173e-01 1.05318940e+00
-5.15349030e-01 -6.47281036e-02 -6.59567297e-01 7.32005596e-01
5.92803836e-01 -3.92706990e-01 7.99941182e-01 -5.64254403e-01
-4.46113855e-01 -1.33537158e-01 -6.04112148e-01 -8.25913787e-01
2.72083402e-01 -9.77384448e-01 2.23341212e-01 -9.58034515e-01
-1.36803716e-01 -6.04955792e-01 -4.30594236e-01 7.48739898e-01
-5.55598974e-01 8.16215634e-01 1.99459624e-02 3.17665458e-01
-3.61921579e-01 1.45499527e-01 1.27393270e+00 -4.50580150e-01
5.70674166e-02 2.66855955e-01 -8.47062647e-01 7.47756422e-01
1.02429068e+00 -6.31658196e-01 -6.20271385e-01 -6.02294683e-01
-3.01720470e-01 -3.38841617e-01 4.63171214e-01 -1.20336390e+00
5.08712649e-01 -2.85447568e-01 7.87351847e-01 -2.83324867e-01
1.43993288e-01 -9.54904377e-01 6.64915740e-02 6.66218936e-01
-3.50248456e-01 4.87790443e-02 5.12386918e-01 5.85293353e-01
-2.88093477e-01 -2.81831145e-01 1.30767190e+00 -3.39831531e-01
-8.51993978e-01 6.65235996e-01 -4.55767274e-01 3.55621219e-01
1.28386891e+00 -2.11177841e-01 -6.34195685e-01 -2.72936940e-01
-1.06253183e+00 1.90331757e-01 2.77639210e-01 2.10262626e-01
7.85053194e-01 -1.40282214e+00 -6.53007150e-01 5.76884866e-01
-2.00939268e-01 2.25673914e-01 4.89086181e-01 2.67500132e-01
-4.91725832e-01 -2.02071786e-01 -4.34666753e-01 -1.20342650e-01
-1.31452036e+00 1.20526564e+00 6.64430320e-01 -3.01072896e-01
-7.39480779e-02 1.39366913e+00 2.93888777e-01 -6.60320699e-01
6.23592079e-01 3.85420948e-01 8.21238500e-05 -5.11353493e-01
5.85237384e-01 1.32487327e-01 2.23865211e-02 -4.75271702e-01
-3.30362171e-01 2.11377919e-01 -3.18066090e-01 3.24157596e-01
8.33058298e-01 2.15086952e-01 -6.20430820e-02 -1.99186519e-01
1.11336136e+00 1.91335939e-02 -1.42592287e+00 -5.97417951e-01
-6.85150504e-01 -1.80990875e-01 -2.20897973e-01 -8.30452323e-01
-1.55159509e+00 8.82973135e-01 7.70642459e-01 9.36019719e-02
1.41758144e+00 -2.58698136e-01 7.11973011e-01 3.50955814e-01
3.11007798e-02 -7.60564506e-01 4.99722034e-01 6.65114105e-01
8.44564974e-01 -1.33488786e+00 6.82664961e-02 -6.05346084e-01
-7.76188850e-01 8.83450508e-01 1.18899369e+00 -6.04552209e-01
9.88780499e-01 8.04362297e-01 6.32475674e-01 2.14518681e-01
-3.67557406e-01 -2.01434150e-01 2.08291009e-01 1.12912357e+00
3.63554619e-02 -2.29885593e-01 4.22202319e-01 6.81542218e-01
-4.13200945e-01 -4.67791438e-01 1.21841840e-01 1.19637680e+00
-1.48729041e-01 -1.09460163e+00 -5.17407119e-01 3.05414766e-01
-8.13611686e-01 -4.78715301e-01 -4.66482848e-01 5.74574828e-01
5.23010075e-01 9.56076562e-01 1.19969815e-01 -1.05133414e+00
2.99318641e-01 -1.83397308e-01 3.04878205e-01 -7.97960937e-01
-1.25015259e+00 -5.99528790e-01 -1.39463902e-01 -2.03391537e-01
-2.06206769e-01 8.61138031e-02 -1.06871223e+00 -7.16965079e-01
-4.13061589e-01 3.11748534e-02 3.68507393e-02 8.63887370e-01
1.49884075e-01 8.44678998e-01 1.09638989e+00 -9.35773969e-01
-8.56082857e-01 -7.24661827e-01 -2.84918129e-01 2.13546112e-01
4.86634433e-01 -1.87122867e-01 -6.40289724e-01 2.63998836e-01]
|
[5.572829723358154, 7.92638635635376]
|
65410f14-a96b-4054-80a7-507d15b67822
|
learning-generative-models-for-active
|
2208.08713
| null |
https://arxiv.org/abs/2208.08713v2
|
https://arxiv.org/pdf/2208.08713v2.pdf
|
Learning Generative Models for Active Inference using Tensor Networks
|
Active inference provides a general framework for behavior and learning in autonomous agents. It states that an agent will attempt to minimize its variational free energy, defined in terms of beliefs over observations, internal states and policies. Traditionally, every aspect of a discrete active inference model must be specified by hand, i.e. by manually defining the hidden state space structure, as well as the required distributions such as likelihood and transition probabilities. Recently, efforts have been made to learn state space representations automatically from observations using deep neural networks. In this paper, we present a novel approach of learning state spaces using quantum physics-inspired tensor networks. The ability of tensor networks to represent the probabilistic nature of quantum states as well as to reduce large state spaces makes tensor networks a natural candidate for active inference. We show how tensor networks can be used as a generative model for sequential data. Furthermore, we show how one can obtain beliefs from such a generative model and how an active inference agent can use these to compute the expected free energy. Finally, we demonstrate our method on the classic T-maze environment.
|
['Bart Dhoedt', 'Tim Verbelen', 'Bram Vanhecke', 'Samuel T. Wauthier']
|
2022-08-18
| null | null | null | null |
['tensor-networks']
|
['methodology']
|
[ 1.25061888e-02 2.58949310e-01 3.33541930e-02 -4.93103862e-01
-2.70999581e-01 -6.79200351e-01 1.01855612e+00 6.77254573e-02
-5.01396775e-01 7.28173494e-01 1.76573396e-01 -2.77725190e-01
-1.84673429e-01 -1.15246832e+00 -6.04784727e-01 -9.40950036e-01
-3.30266744e-01 9.79213297e-01 6.90974807e-03 -3.55112702e-01
3.92922223e-01 5.54108918e-01 -1.32918942e+00 -1.48469821e-01
8.50975275e-01 5.99161565e-01 -5.08560939e-03 8.81894350e-01
-1.81418777e-01 1.25002766e+00 -3.80025148e-01 -1.96947634e-01
-9.93616972e-03 -5.77219546e-01 -1.04027998e+00 -2.89375875e-02
5.11403345e-02 -4.08676714e-01 -3.92887294e-01 1.20210063e+00
-1.18949890e-01 5.73688865e-01 9.56266761e-01 -1.17499304e+00
-5.55163980e-01 6.89660788e-01 4.94811773e-01 7.46086091e-02
8.05298686e-02 3.68938327e-01 1.27659297e+00 -4.60117847e-01
5.61971784e-01 1.13505280e+00 1.74758241e-01 5.31643450e-01
-2.00453281e+00 -4.01306599e-02 3.82118323e-03 3.91194791e-01
-1.23918688e+00 -5.28036177e-01 7.99655378e-01 -5.28795660e-01
1.12347019e+00 2.82875240e-01 1.00682473e+00 1.14875174e+00
4.56576109e-01 7.86608934e-01 1.03883684e+00 -5.23146808e-01
7.99998939e-01 -5.02831675e-02 4.54118460e-01 1.09016669e+00
8.16569701e-02 3.29424679e-01 -5.13084471e-01 -5.97974300e-01
7.96394348e-01 5.65048158e-02 1.48244962e-01 -6.98783219e-01
-1.27260196e+00 1.00845850e+00 3.80454391e-01 1.38458060e-02
-3.33456963e-01 6.71786010e-01 8.03248063e-02 1.40789688e-01
2.87461251e-01 6.08087182e-01 -1.81807831e-01 -6.69206083e-02
-5.71894467e-01 4.65705603e-01 1.17500401e+00 3.95665318e-01
1.23117805e+00 -4.03789543e-02 -3.74485552e-02 3.49330485e-01
8.31784666e-01 5.20015717e-01 9.50536057e-02 -1.45336354e+00
-1.63440421e-01 4.72754717e-01 3.45272899e-01 -7.40809560e-01
-2.24483132e-01 -1.07599773e-01 -4.65728343e-01 4.58863139e-01
5.01453757e-01 -5.01245745e-02 -9.09248769e-01 2.04803252e+00
7.17881843e-02 -9.05444380e-03 9.39812064e-02 5.57094097e-01
2.41019279e-01 7.80546606e-01 -1.60317004e-01 -3.47906530e-01
8.21678698e-01 -5.13803720e-01 -6.46593094e-01 -7.82560483e-02
6.14248335e-01 -2.51304746e-01 6.80358469e-01 3.35407823e-01
-1.20683765e+00 -1.85681447e-01 -8.80971849e-01 -2.27933172e-02
-3.93011600e-01 -2.70830750e-01 1.05673730e+00 5.54177999e-01
-1.07090235e+00 9.35564816e-01 -1.82316732e+00 -1.32563964e-01
8.15458372e-02 5.74026406e-01 -5.95525652e-02 4.95795578e-01
-1.04473114e+00 1.13511336e+00 5.82463503e-01 1.39000416e-01
-1.47196960e+00 1.30549846e-02 -7.88285077e-01 3.97159085e-02
3.55803668e-01 -8.56291890e-01 1.38454330e+00 -5.33510625e-01
-1.97063899e+00 2.56531447e-01 -1.46463186e-01 -5.84396362e-01
-6.94894642e-02 2.55091548e-01 -4.92249168e-02 -1.22897159e-02
-2.33373463e-01 5.38258135e-01 6.42177105e-01 -7.69515634e-01
-3.14640515e-02 -3.96601349e-01 5.89747906e-01 1.24932282e-01
-6.33104965e-02 -5.55817723e-01 6.12860732e-02 1.61494806e-01
5.41039586e-01 -1.36737037e+00 -5.84832788e-01 -1.58761665e-01
-6.26349866e-01 -4.99581724e-01 3.55958045e-01 -1.76267162e-01
6.98030412e-01 -1.85276484e+00 7.05236197e-01 5.50306618e-01
5.31274676e-01 -2.66606688e-01 9.34996158e-02 5.32411277e-01
1.60580531e-01 3.08711901e-02 -2.98743278e-01 -3.77474815e-01
5.63543081e-01 7.65572965e-01 -2.84977555e-01 4.91918504e-01
3.55699807e-01 1.02054703e+00 -9.43876922e-01 -3.94184947e-01
1.60051793e-01 3.82979035e-01 -8.88372004e-01 1.99532121e-01
-6.03618085e-01 6.55587494e-01 -5.40453613e-01 -7.28138387e-02
1.62054881e-01 -4.90136147e-01 4.37141091e-01 8.38700274e-04
-3.32618117e-01 9.07621443e-01 -1.04404819e+00 1.69257414e+00
-3.18965644e-01 4.99144435e-01 -2.40135476e-01 -1.24944770e+00
6.64557338e-01 2.74112582e-01 5.54733455e-01 -3.69364917e-01
1.93753988e-01 2.29180660e-02 3.10778379e-01 -3.62707675e-01
5.09587407e-01 -6.67258278e-02 -2.09062159e-01 8.75174284e-01
3.49653244e-01 -2.85934091e-01 4.93771583e-01 5.84530890e-01
1.07611489e+00 3.02742779e-01 3.76707762e-02 -3.11618298e-01
3.18457484e-01 2.74224672e-02 3.89815360e-01 8.92183661e-01
1.45029910e-02 -1.38602644e-01 6.11827374e-01 -4.44537133e-01
-1.22110009e+00 -1.47943282e+00 -1.06926017e-01 9.27381277e-01
-7.85149708e-02 -6.75621152e-01 -6.78031206e-01 -2.83010066e-01
-3.54882091e-01 8.59915137e-01 -7.93386936e-01 -3.49580705e-01
-3.72861445e-01 -7.98945189e-01 2.30833262e-01 1.46836102e-01
4.25965756e-01 -9.50063705e-01 -6.58459127e-01 2.71287739e-01
-2.16925159e-01 -6.07391834e-01 5.36084101e-02 5.35842717e-01
-9.97502923e-01 -7.12323487e-01 -8.12654123e-02 -2.34649360e-01
7.09170222e-01 -3.92563671e-01 1.12706673e+00 -9.13140699e-02
-1.40566513e-01 5.46289563e-01 3.18173692e-02 -8.01515281e-02
-7.88618147e-01 9.78669226e-02 3.00322652e-01 -5.71090318e-02
2.79501021e-01 -7.61848152e-01 -2.49203324e-01 -1.56283259e-01
-8.98954749e-01 1.86542884e-01 2.44168147e-01 8.50837827e-01
4.04885679e-01 -4.82693166e-02 -1.87551856e-01 -7.00364649e-01
6.57947123e-01 -1.91550851e-01 -9.81554806e-01 2.96658128e-01
-4.15413409e-01 9.52971399e-01 3.04009050e-01 -3.07455897e-01
-1.03410888e+00 1.86342314e-01 -9.83576402e-02 5.51014915e-02
-1.78838983e-01 7.00046539e-01 1.49925500e-01 -9.61884782e-02
8.06823432e-01 4.52708960e-01 2.24244326e-01 -1.67643964e-01
4.89489228e-01 2.85608500e-01 3.08349356e-02 -8.99813235e-01
7.66327918e-01 5.40873110e-01 3.14997703e-01 -7.80108988e-01
-9.41605330e-01 6.48058951e-02 -9.01551783e-01 -7.45999515e-02
9.58792090e-01 -3.00819278e-01 -1.34441674e+00 3.31053346e-01
-1.07133698e+00 -4.09752458e-01 -3.62118721e-01 7.16282964e-01
-9.46787536e-01 3.69715005e-01 -7.02191770e-01 -1.15322769e+00
4.16818187e-02 -1.47936845e+00 6.01248324e-01 1.27730235e-01
-2.07901537e-01 -1.38095832e+00 7.09916532e-01 1.44926816e-01
4.90941644e-01 -3.63784507e-02 8.48830462e-01 -4.11735266e-01
-1.07630289e+00 4.36135232e-02 4.00137514e-01 2.01047763e-01
1.30307302e-02 3.49706471e-01 -7.48199761e-01 -1.65849939e-01
4.80991341e-02 -4.60994095e-01 9.11349535e-01 2.95490056e-01
8.91009569e-01 -4.97393370e-01 -1.13122791e-01 3.88986379e-01
1.11673951e+00 1.76235303e-01 5.73891997e-01 -1.51621819e-01
5.74057639e-01 3.30480933e-01 -1.83708310e-01 2.43996501e-01
4.53452080e-01 6.57885432e-01 4.22884256e-01 4.81798798e-01
5.96247792e-01 -1.75150651e-02 5.98695517e-01 1.05025661e+00
-4.30217385e-01 1.46193713e-01 -1.02517307e+00 4.18837853e-02
-1.93485177e+00 -1.37774837e+00 4.88496162e-02 2.17877650e+00
9.15668786e-01 3.33962411e-01 4.42460701e-02 -1.25067487e-01
2.19022289e-01 -6.32633641e-02 -6.91524506e-01 -2.91399926e-01
1.49772227e-01 3.05402011e-01 2.81041712e-01 8.44224691e-01
-9.11263049e-01 7.82541037e-01 7.41072512e+00 3.01809162e-01
-9.76959527e-01 1.70516118e-01 1.89050928e-01 1.22527480e-02
-4.38114792e-01 5.91114700e-01 -6.03966594e-01 2.44281575e-01
1.17499626e+00 4.84310905e-04 9.09357727e-01 5.75474143e-01
-1.62772126e-02 -2.13744551e-01 -1.39186275e+00 7.67063141e-01
-3.61871660e-01 -1.54080200e+00 -5.70714734e-02 5.54928839e-01
6.10415399e-01 2.19880521e-01 -6.76409900e-02 3.06479245e-01
9.51860666e-01 -8.90741706e-01 8.24019551e-01 9.35035050e-01
4.76113707e-02 -5.62466800e-01 2.02865124e-01 7.48671234e-01
-7.23423421e-01 4.78955843e-02 -4.30779666e-01 -4.75879073e-01
1.56853423e-01 3.32555145e-01 -8.35122168e-01 1.47383595e-02
1.08555473e-01 5.17199039e-01 -3.13128740e-01 7.68134832e-01
-1.95902899e-01 7.80085444e-01 -7.17197120e-01 -5.73794544e-01
2.97359735e-01 -7.31610477e-01 5.66493571e-01 6.56507015e-01
-8.02659988e-02 1.51802078e-01 2.96119899e-01 1.64341497e+00
3.55182469e-01 -4.11292404e-01 -5.27824402e-01 -7.23333418e-01
1.90156356e-01 1.10857797e+00 -7.93175042e-01 -5.22398770e-01
-3.51982214e-03 7.89180696e-01 6.80028021e-01 5.05735457e-01
-7.24981546e-01 -1.20908402e-01 7.40919471e-01 -4.43490446e-01
9.86142755e-02 -1.06602001e+00 1.27618223e-01 -1.71323454e+00
-4.82726842e-01 -5.19981921e-01 -1.84103221e-01 -6.53669298e-01
-1.05523551e+00 3.38872373e-01 2.94841409e-01 -6.46363974e-01
-6.95498586e-01 -7.70067453e-01 -5.75736701e-01 8.75875533e-01
-7.81005561e-01 -4.94528055e-01 1.13784604e-01 6.46685541e-01
3.12343929e-02 -5.68244755e-02 1.20596170e+00 -3.20902944e-01
-7.68153787e-01 -7.89322704e-02 2.71628410e-01 3.36388320e-01
-1.90273598e-01 -1.50883698e+00 5.28970301e-01 7.97247231e-01
7.03817308e-01 1.30433095e+00 9.69894528e-01 -3.95249575e-01
-1.78015423e+00 -5.07930219e-01 4.84760493e-01 -7.24084795e-01
8.28404367e-01 -6.32280111e-01 -6.78152084e-01 9.65236366e-01
1.42137259e-01 -9.48620029e-03 7.13569522e-01 5.31051755e-01
-2.96320736e-01 1.32946476e-01 -8.19289863e-01 7.49063015e-01
8.76680017e-01 -9.69094038e-01 -5.92868745e-01 5.01471519e-01
3.78722429e-01 -2.62185037e-01 -7.92323351e-01 -1.68460026e-01
4.77423787e-01 -1.05464613e+00 8.00927222e-01 -9.05791461e-01
8.13752115e-02 -2.60031641e-01 -2.36917213e-01 -1.30986929e+00
-7.53880084e-01 -5.67487478e-01 -3.94617081e-01 4.94639158e-01
6.56441450e-01 -7.82130003e-01 8.17553759e-01 9.08447146e-01
-2.54336130e-02 -3.08652550e-01 -8.24351966e-01 -5.24540126e-01
4.25023101e-02 -5.87387800e-01 3.79587948e-01 6.97368562e-01
1.47275239e-01 5.51229894e-01 -8.00854638e-02 1.46378651e-01
1.02717376e+00 6.77763745e-02 5.62088549e-01 -1.37672937e+00
-6.36635065e-01 -2.46268168e-01 -5.24284422e-01 -1.18563735e+00
3.50177974e-01 -1.01098371e+00 9.51013044e-02 -1.48828745e+00
3.46853524e-01 -4.39296722e-01 -1.18157327e-01 4.18673456e-01
3.34382236e-01 1.62001267e-01 6.90309629e-02 4.00994420e-01
-6.98908746e-01 6.59735203e-01 1.13909853e+00 -2.33184397e-01
-5.15453480e-02 -1.06350191e-01 -1.78787515e-01 7.51700878e-01
8.69753659e-01 -5.07863939e-01 -4.62004423e-01 -2.02960983e-01
9.90274489e-01 1.23310260e-01 6.82708859e-01 -7.27781534e-01
3.59091759e-01 -5.78240275e-01 2.62173384e-01 -5.62853515e-01
8.29279304e-01 -3.76528770e-01 3.97161722e-01 5.35549104e-01
-6.27273619e-01 -3.09222728e-01 -3.86168510e-01 4.36184496e-01
5.04668392e-02 -4.37326223e-01 5.71717024e-01 -5.03921211e-01
-4.97245669e-01 2.36263573e-01 -8.36393118e-01 -1.59739777e-01
6.13737464e-01 -2.01065410e-02 -3.04318190e-01 -3.02730143e-01
-1.41656411e+00 -8.16730037e-02 4.66574430e-01 -1.60193518e-01
3.83099705e-01 -1.31995285e+00 -2.98025280e-01 2.26415843e-01
-6.21671136e-03 -2.40699694e-01 8.46975222e-02 9.78394747e-01
-6.33145094e-01 4.83825415e-01 -3.24130327e-01 -7.56186962e-01
-7.57214844e-01 2.47533441e-01 7.30656385e-01 -1.44219890e-01
-3.80666733e-01 6.54438317e-01 -2.18871105e-02 -6.16991460e-01
-1.40366405e-01 -6.28916204e-01 1.18018137e-02 -2.11759746e-01
3.15841615e-01 9.70121697e-02 -1.75952151e-01 -6.15715563e-01
-3.02640527e-01 8.96957144e-03 -2.82989908e-02 -6.59178019e-01
1.24295139e+00 -4.26989701e-03 -5.10018289e-01 9.56609488e-01
1.08739614e+00 -3.16416562e-01 -1.23855364e+00 -2.15362653e-01
-4.17189077e-02 -3.85747217e-02 3.35782766e-01 -4.52031314e-01
-6.03736460e-01 9.28430200e-01 4.28098470e-01 9.55373824e-01
3.33440065e-01 1.03252493e-01 2.82063484e-01 1.10590279e+00
6.66608453e-01 -1.06311059e+00 2.79108137e-01 9.86136973e-01
4.37640131e-01 -1.14389169e+00 -1.87176153e-01 1.31069213e-01
-2.50965089e-01 1.22757614e+00 2.99726985e-02 -3.72758359e-01
7.66852796e-01 1.87224850e-01 -3.41652721e-01 -5.26586473e-01
-1.07351434e+00 -2.21741393e-01 2.72611737e-01 3.68769526e-01
5.08307815e-01 4.59011883e-01 2.39683568e-01 -2.69990742e-01
-4.88966614e-01 -1.45677507e-01 7.81766057e-01 1.01757669e+00
-6.82633638e-01 -1.26310635e+00 -2.69936651e-01 4.24591064e-01
-1.11309879e-01 -4.80599329e-03 -3.18811148e-01 1.56483471e-01
-5.60644381e-02 7.56629527e-01 3.11820060e-01 -2.88179040e-01
-2.79498726e-01 3.25183600e-01 1.01660323e+00 -8.64660561e-01
2.69492250e-02 -3.02106917e-01 7.96828941e-02 -6.54147983e-01
-5.14366269e-01 -8.09471965e-01 -1.31739128e+00 -5.12116611e-01
-4.70447123e-01 5.08633435e-01 8.46374929e-01 1.27787352e+00
8.02388564e-02 3.86071742e-01 3.45928818e-01 -9.16341305e-01
-8.40894043e-01 -9.80208635e-01 -5.96052945e-01 1.82333305e-01
5.19927621e-01 -8.53420615e-01 -5.03675759e-01 7.67264366e-02]
|
[5.7667646408081055, 4.775277614593506]
|
5f6b987e-3b36-437d-992b-13a07ab942e2
|
dumlp-pin-a-dual-mlp-dot-product-permutation
|
2203.04007
| null |
https://arxiv.org/abs/2203.04007v2
|
https://arxiv.org/pdf/2203.04007v2.pdf
|
DuMLP-Pin: A Dual-MLP-dot-product Permutation-invariant Network for Set Feature Extraction
|
Existing permutation-invariant methods can be divided into two categories according to the aggregation scope, i.e. global aggregation and local one. Although the global aggregation methods, e. g., PointNet and Deep Sets, get involved in simpler structures, their performance is poorer than the local aggregation ones like PointNet++ and Point Transformer. It remains an open problem whether there exists a global aggregation method with a simple structure, competitive performance, and even much fewer parameters. In this paper, we propose a novel global aggregation permutation-invariant network based on dual MLP dot-product, called DuMLP-Pin, which is capable of being employed to extract features for set inputs, including unordered or unstructured pixel, attribute, and point cloud data sets. We strictly prove that any permutation-invariant function implemented by DuMLP-Pin can be decomposed into two or more permutation-equivariant ones in a dot-product way as the cardinality of the given input set is greater than a threshold. We also show that the DuMLP-Pin can be viewed as Deep Sets with strong constraints under certain conditions. The performance of DuMLP-Pin is evaluated on several different tasks with diverse data sets. The experimental results demonstrate that our DuMLP-Pin achieves the best results on the two classification problems for pixel sets and attribute sets. On both the point cloud classification and the part segmentation, the accuracy of DuMLP-Pin is very close to the so-far best-performing local aggregation method with only a 1-2% difference, while the number of required parameters is significantly reduced by more than 85% in classification and 69% in segmentation, respectively. The code is publicly available on https://github.com/JaronTHU/DuMLP-Pin.
|
['Shuo Zhang', 'Huanjun Deng', 'Mingyang Li', 'Zhidong Deng', 'Wenlei Liu', 'Ziyu Zhu', 'Jiajun Fei']
|
2022-03-08
| null | null | null | null |
['point-cloud-classification']
|
['computer-vision']
|
[ 1.00407518e-01 -1.23541981e-01 -3.94860059e-02 -3.55881929e-01
-7.84788489e-01 -6.35597944e-01 3.86069626e-01 4.02353853e-02
-1.37699574e-01 7.61445105e-01 -3.88857096e-01 -2.86066115e-01
-5.34170806e-01 -1.11284602e+00 -9.46060896e-01 -9.98151660e-01
-7.00969025e-02 5.80100894e-01 4.38045621e-01 -1.60520211e-01
1.15073860e-01 8.25504124e-01 -1.71205175e+00 1.81143612e-01
9.35366690e-01 1.35057473e+00 1.27332836e-01 2.56605893e-01
-1.46647379e-01 6.85722455e-02 -2.93242246e-01 -3.83908778e-01
7.10484266e-01 -3.78028080e-02 -6.70459569e-01 4.16700430e-02
5.37082613e-01 1.28987148e-01 1.23404646e-02 1.19920659e+00
4.75980520e-01 -8.64873827e-02 5.55792212e-01 -1.56720233e+00
-4.99942303e-01 7.15648770e-01 -6.86493933e-01 -1.38233334e-01
-1.02204025e-01 8.45113918e-02 1.16426146e+00 -9.53199208e-01
2.55537361e-01 1.13628638e+00 8.72629702e-01 -1.51703795e-02
-1.23487484e+00 -7.62147188e-01 6.95874766e-02 9.91866440e-02
-1.59056461e+00 4.70231883e-02 7.55266190e-01 -3.81748468e-01
5.75524390e-01 6.17598593e-01 4.63609606e-01 5.60356975e-01
6.78741708e-02 6.28435612e-01 1.25013423e+00 -1.83091626e-01
1.02306999e-01 -1.36152655e-01 4.50805008e-01 7.57733226e-01
3.69876385e-01 -3.39654654e-01 -2.02164903e-01 -2.58739322e-01
8.37175250e-01 8.37931037e-02 -2.49301195e-01 -4.29210991e-01
-1.29337382e+00 6.13759100e-01 6.44276261e-01 3.58723819e-01
-3.01049322e-01 1.51751880e-02 4.14778233e-01 2.68632859e-01
4.92780834e-01 3.53798598e-01 -6.85847819e-01 1.76731393e-01
-5.48047662e-01 1.51331693e-01 6.89297795e-01 1.04467237e+00
1.06982279e+00 -1.13783240e-01 -1.10916555e-01 9.02776659e-01
3.26886587e-03 6.29322410e-01 2.69821256e-01 -6.82311952e-01
6.15983844e-01 8.16314518e-01 -1.59355462e-01 -1.28179085e+00
-5.34103096e-01 -5.87984562e-01 -1.13268971e+00 2.76690781e-01
3.31767082e-01 -1.34131266e-02 -8.25882792e-01 1.73889482e+00
2.71687776e-01 1.88022152e-01 -1.03328973e-01 6.63930476e-01
8.13841581e-01 7.81490803e-01 -2.87032187e-01 -4.17206585e-02
1.40018237e+00 -8.62186253e-01 -2.72430688e-01 1.95694447e-01
3.94748539e-01 -8.11935544e-01 1.17767251e+00 4.72507060e-01
-1.01233447e+00 -6.80001378e-01 -1.06212497e+00 4.91988473e-02
-4.86604422e-01 4.00700659e-01 8.06123197e-01 4.67500418e-01
-1.01880848e+00 7.32952774e-01 -7.85938799e-01 -2.79520392e-01
5.67806721e-01 7.80745924e-01 -5.63212216e-01 1.15299195e-01
-8.17977369e-01 2.89365649e-01 5.78623831e-01 1.57140419e-01
-2.91982621e-01 -8.64501595e-01 -6.36993825e-01 2.23780721e-01
3.37135583e-01 -7.33636022e-01 7.89547324e-01 -7.61398852e-01
-1.26053143e+00 6.73376739e-01 1.17496371e-01 -3.85318041e-01
5.93438268e-01 -2.46992894e-02 -1.44656330e-01 2.27559637e-02
2.65408874e-01 7.49872088e-01 6.60435498e-01 -1.19462025e+00
-9.47440386e-01 -5.33945858e-01 2.15636566e-01 2.07134575e-01
-3.38941902e-01 -8.57555047e-02 -6.21840119e-01 -5.36539495e-01
2.74717391e-01 -1.07119691e+00 -1.08801745e-01 1.31026149e-01
-6.08044624e-01 -5.89784920e-01 1.03282499e+00 -2.72434384e-01
9.29255247e-01 -2.21270609e+00 -4.14672084e-02 3.56859356e-01
1.62971288e-01 3.56312931e-01 -8.71938244e-02 2.12041125e-01
-2.67872989e-01 2.99808592e-01 -6.48808122e-01 -2.88456589e-01
4.79089171e-02 3.45823914e-01 -2.21457943e-01 4.36472982e-01
1.84441820e-01 6.12789989e-01 -6.81951642e-01 -3.51969838e-01
3.48576903e-01 3.00426662e-01 -4.37397778e-01 -3.38447213e-01
-8.20708796e-02 1.02769963e-01 -4.29854155e-01 8.03399503e-01
1.19528651e+00 -1.16096765e-01 -2.02759326e-01 -4.18471098e-01
-2.48674333e-01 -3.14675644e-02 -1.46003723e+00 1.42828572e+00
-4.39970285e-01 4.15890038e-01 1.78181287e-03 -1.05088937e+00
1.09605920e+00 5.85758798e-02 7.19595313e-01 -3.39892864e-01
-1.91826709e-02 5.09437680e-01 1.04625389e-01 -9.09690857e-02
3.12457204e-01 1.14391878e-01 -1.84955806e-01 1.87772080e-01
-4.49824259e-02 6.19158372e-02 4.05148566e-01 -2.21770167e-01
1.00369406e+00 -7.52019659e-02 1.59127608e-01 -6.62518382e-01
7.63356745e-01 -1.84670374e-01 7.86585152e-01 6.34487808e-01
1.69958860e-01 8.08557987e-01 6.89536393e-01 -3.44563961e-01
-8.03492010e-01 -1.15292180e+00 -5.05765975e-01 7.16150582e-01
3.27381760e-01 -2.33503923e-01 -5.50328135e-01 -6.57740653e-01
2.08284616e-01 3.76117617e-01 -3.41654330e-01 5.70527352e-02
-6.23079181e-01 -9.44853783e-01 5.15810192e-01 6.34433091e-01
1.03445792e+00 -9.59944248e-01 -1.37481868e-01 -1.19110942e-02
4.48929518e-02 -1.17766297e+00 -3.20542365e-01 2.05097020e-01
-8.97881508e-01 -9.35959697e-01 -4.61451083e-01 -9.12546575e-01
7.72166252e-01 2.25718081e-01 8.34153056e-01 -1.11808002e-01
1.13771208e-01 -9.94749665e-02 -3.15625012e-01 -4.58777964e-01
1.19391322e-01 3.22901428e-01 6.93438621e-03 3.51296306e-01
1.70963138e-01 -8.35392058e-01 -4.87894654e-01 6.80933297e-01
-9.57383811e-01 -5.95424287e-02 8.02410364e-01 6.45615458e-01
9.29403365e-01 2.84088314e-01 5.04626632e-01 -7.44376659e-01
3.98127973e-01 -3.28113496e-01 -7.13836908e-01 -1.47410836e-02
-3.08306754e-01 2.91104782e-02 8.95007253e-01 -2.09346756e-01
-6.41120315e-01 2.61508942e-01 -3.01660329e-01 -5.28661489e-01
-1.39786884e-01 4.61434901e-01 -5.66990018e-01 -3.63783538e-01
3.76723588e-01 7.94782862e-02 -1.50279090e-01 -5.56571007e-01
1.57436073e-01 5.56511223e-01 6.07100844e-01 -7.29884863e-01
9.46536362e-01 7.11232126e-01 3.98610204e-01 -9.21849668e-01
-5.48434198e-01 -4.76560682e-01 -6.91004276e-01 1.15297109e-01
6.88580513e-01 -5.20995855e-01 -6.73316300e-01 7.00892389e-01
-1.07987177e+00 1.90088451e-02 -4.00200725e-01 2.07793713e-01
-4.97428179e-01 5.47887087e-01 -3.24984878e-01 -3.44750404e-01
-4.06334281e-01 -1.38907838e+00 1.12520492e+00 3.28095397e-03
2.12529272e-01 -8.36498857e-01 -3.16603363e-01 1.30444318e-01
8.73568654e-02 6.20336592e-01 1.09657371e+00 -8.50394726e-01
-8.92838180e-01 -2.09135637e-01 -3.64456445e-01 7.08281755e-01
1.61851466e-01 1.54596239e-01 -7.40909994e-01 -1.30391955e-01
-4.80848439e-02 1.22344695e-01 6.52584612e-01 3.33361357e-01
1.69867718e+00 -3.01430106e-01 -3.79486442e-01 8.74677122e-01
1.67798662e+00 2.09617212e-01 6.38941705e-01 2.50290811e-01
8.25712502e-01 2.54406989e-01 6.36532545e-01 1.61958426e-01
1.58534929e-01 6.96676552e-01 6.83565140e-01 -1.49580017e-01
9.63013694e-02 -9.63456370e-03 1.24396324e-01 9.60776210e-01
-3.48548383e-01 -2.39729390e-01 -9.00558114e-01 6.01224601e-01
-1.98533714e+00 -7.51867354e-01 -4.92500812e-01 2.35397482e+00
5.29032111e-01 2.80986786e-01 -2.81003118e-02 3.54097873e-01
8.45004141e-01 1.19134814e-01 -3.62757117e-01 -4.25872386e-01
-2.95202643e-01 4.40810472e-01 8.31204534e-01 3.31411302e-01
-1.37436950e+00 6.73740387e-01 4.94706821e+00 1.13936865e+00
-1.02092147e+00 1.81289520e-02 5.43187082e-01 2.56513596e-01
-6.19848184e-02 -4.27553691e-02 -8.98406267e-01 6.94936454e-01
3.24817568e-01 -3.86805534e-02 3.39841396e-02 8.57574880e-01
-1.73398927e-02 1.99866369e-01 -1.06256950e+00 1.07372200e+00
-1.30106509e-01 -1.27055025e+00 1.53767914e-01 2.12624088e-01
7.40917385e-01 1.44381851e-01 1.75445452e-01 1.75326288e-01
1.26356512e-01 -7.28824437e-01 5.66240668e-01 3.71452659e-01
6.78082883e-01 -9.20250237e-01 8.71393442e-01 3.46520215e-01
-1.35282123e+00 -9.62587830e-04 -5.91639161e-01 3.38862091e-02
1.02603557e-02 7.91899323e-01 -3.05379450e-01 1.03749979e+00
8.17717493e-01 7.28675544e-01 -4.34485197e-01 1.24803901e+00
-1.17871817e-02 5.03456473e-01 -7.99463749e-01 8.28587636e-02
3.74167740e-01 -6.08716011e-01 7.30710149e-01 9.36556458e-01
4.98206288e-01 -1.51267520e-03 2.38341898e-01 7.74875343e-01
-1.26525864e-01 1.28081158e-01 -6.43574655e-01 3.99541974e-01
3.60144287e-01 1.54726052e+00 -7.78663576e-01 -1.43293530e-01
-2.65920728e-01 5.58405578e-01 6.22000359e-02 3.46140265e-02
-9.88691330e-01 -6.98709846e-01 8.21693242e-01 5.32686301e-02
5.74834406e-01 -1.61125347e-01 -5.06946385e-01 -9.47338164e-01
3.68262827e-01 -7.09830284e-01 3.55233759e-01 -7.78337240e-01
-1.40085018e+00 8.19118798e-01 1.54830545e-01 -1.48676467e+00
2.11481676e-01 -9.96645451e-01 -6.58304214e-01 6.86198235e-01
-1.36180508e+00 -1.20302427e+00 -5.07220805e-01 6.33922517e-01
3.45667094e-01 -1.80903986e-01 5.53937078e-01 5.29056609e-01
-6.01963341e-01 6.91684246e-01 2.43879035e-01 1.43244699e-01
3.87783587e-01 -1.34489346e+00 8.13015699e-02 8.49884748e-01
6.79514855e-02 4.42921370e-01 3.86138648e-01 -3.29923183e-01
-1.17354274e+00 -1.29532158e+00 5.96652508e-01 -3.00446898e-01
6.04651392e-01 -4.64523405e-01 -9.31527674e-01 7.27819145e-01
1.81323737e-01 1.54396877e-01 3.57935101e-01 -1.17074572e-01
-1.08475164e-01 -6.40234113e-01 -1.28418946e+00 5.45487344e-01
1.27858019e+00 -6.31811097e-02 -3.69119734e-01 5.81100285e-01
9.50691402e-01 -4.40468341e-01 -1.05399394e+00 8.27208281e-01
2.39493832e-01 -1.06560469e+00 1.06319857e+00 -1.91173539e-01
4.44258898e-01 -5.65261424e-01 -3.04067403e-01 -1.14978623e+00
-4.07112181e-01 -4.79683965e-01 3.56612116e-01 1.47024763e+00
3.97182941e-01 -1.24670696e+00 6.72773540e-01 5.41059300e-02
-4.59239095e-01 -1.03195870e+00 -1.00864732e+00 -1.10415971e+00
4.01891507e-02 -4.26571101e-01 1.06527209e+00 8.55382800e-01
-5.94405472e-01 2.75232315e-01 2.11926877e-01 4.27360594e-01
5.20143747e-01 3.83340716e-01 8.09183478e-01 -1.42134237e+00
-1.28787279e-01 -5.66459715e-01 -7.60385275e-01 -9.10625935e-01
1.49684295e-01 -1.18234694e+00 -2.46567518e-01 -1.52224195e+00
-1.70621842e-01 -9.82763648e-01 -3.69336635e-01 6.54618919e-01
2.49579549e-01 3.55522841e-01 2.60170400e-01 3.05805475e-01
-1.87025428e-01 4.03580129e-01 1.14687467e+00 -1.34221032e-01
-1.97338179e-01 3.72995853e-01 -6.96508408e-01 1.03748906e+00
1.02513015e+00 -3.00341964e-01 -3.24534059e-01 -4.71750230e-01
1.83313891e-01 -3.07955861e-01 3.57794136e-01 -1.35215557e+00
1.27810925e-01 1.73208080e-02 2.83364509e-03 -9.71823394e-01
3.83147597e-01 -1.02519047e+00 2.23799601e-01 1.79611638e-01
1.26872733e-01 3.98160994e-01 3.11713934e-01 2.85458475e-01
-3.58008146e-01 -2.74102777e-01 7.56129920e-01 5.95978424e-02
-6.94521010e-01 5.94851732e-01 2.47888446e-01 -2.44340390e-01
1.09200990e+00 -3.06525677e-01 -4.73796159e-01 4.76734899e-02
-6.69934750e-01 1.43691629e-01 2.89402664e-01 2.01612204e-01
2.65410602e-01 -1.59632170e+00 -6.47975743e-01 2.59986490e-01
3.00538111e-02 5.22457421e-01 9.86443087e-02 1.00839663e+00
-7.99548626e-01 5.15204370e-01 -3.47977757e-01 -1.03867829e+00
-1.31267262e+00 4.57728505e-01 2.81912148e-01 -3.78839493e-01
-6.13268077e-01 6.34397566e-01 4.73070174e-01 -7.26352930e-01
4.46548350e-02 -7.82458603e-01 5.73901124e-02 -1.00820601e-01
3.90234925e-02 4.59209770e-01 2.22680777e-01 -6.50032759e-01
-3.47989351e-01 1.00609136e+00 1.27100840e-01 3.03159595e-01
1.41086733e+00 3.68571579e-01 -5.61025023e-01 2.14990258e-01
1.40974176e+00 4.17583287e-02 -9.81218636e-01 -1.89210966e-01
-3.46867830e-01 -4.76597369e-01 -2.54490167e-01 -5.34844458e-01
-1.31715286e+00 8.24366808e-01 5.25518417e-01 5.61365724e-01
1.35709107e+00 4.72662225e-02 7.19625711e-01 4.01466548e-01
5.10423601e-01 -7.22199082e-01 -2.89159089e-01 4.90052640e-01
1.16643131e+00 -9.09462154e-01 4.23788838e-02 -8.48366857e-01
-3.81178737e-01 9.49616611e-01 6.91244125e-01 -4.80558425e-01
8.01641703e-01 2.08792433e-01 -3.70538026e-01 -1.37465507e-01
-3.84252936e-01 -2.21188426e-01 3.20587635e-01 3.71765912e-01
-1.10827431e-01 2.44826689e-01 -4.73222941e-01 5.19202590e-01
-5.99121153e-01 -2.31523857e-01 2.51781464e-01 6.12080216e-01
-3.58414948e-01 -1.07732916e+00 -4.83352870e-01 6.34676933e-01
-3.14105511e-01 8.20790604e-02 -2.09039927e-01 1.19334245e+00
5.77541888e-01 4.03636098e-01 3.99337739e-01 -3.45117241e-01
4.84582961e-01 -1.51761666e-01 2.88698941e-01 -5.42785287e-01
-5.50214291e-01 -1.81974515e-01 -7.55187422e-02 -4.96658593e-01
-4.89252865e-01 -6.90175712e-01 -1.14683199e+00 -2.65548676e-01
-2.84755170e-01 -6.57640323e-02 5.31599343e-01 6.86980665e-01
4.79312599e-01 3.97188246e-01 6.54982507e-01 -6.39283359e-01
-4.62902188e-01 -7.23707497e-01 -5.54844141e-01 1.66097984e-01
1.04143083e-01 -6.80377007e-01 -4.16151702e-01 -2.67310858e-01]
|
[7.932216167449951, -3.4199299812316895]
|
7a113e86-794f-49b2-bde6-29befd0b4643
|
diffusionstr-diffusion-model-for-scene-text
|
2306.16707
| null |
https://arxiv.org/abs/2306.16707v1
|
https://arxiv.org/pdf/2306.16707v1.pdf
|
DiffusionSTR: Diffusion Model for Scene Text Recognition
|
This paper presents Diffusion Model for Scene Text Recognition (DiffusionSTR), an end-to-end text recognition framework using diffusion models for recognizing text in the wild. While existing studies have viewed the scene text recognition task as an image-to-text transformation, we rethought it as a text-text one under images in a diffusion model. We show for the first time that the diffusion model can be applied to text recognition. Furthermore, experimental results on publicly available datasets show that the proposed method achieves competitive accuracy compared to state-of-the-art methods.
|
['Masato Fujitake']
|
2023-06-29
| null | null | null | null |
['scene-text-recognition']
|
['computer-vision']
|
[ 5.15220702e-01 -6.24694049e-01 1.54115081e-01 -4.30140704e-01
-3.77573967e-01 -5.09005845e-01 1.30434263e+00 -2.60071725e-01
-3.72733116e-01 -3.49896044e-01 3.86153638e-01 -3.33030164e-01
2.47133896e-01 -4.51201826e-01 -3.54061544e-01 -7.72542775e-01
6.31677270e-01 5.61842561e-01 4.61417317e-01 1.35243580e-01
8.06734443e-01 3.91845107e-01 -1.05588436e+00 7.43410587e-01
5.01019478e-01 7.24801660e-01 3.18661928e-02 1.31276166e+00
-4.10369217e-01 1.30273461e+00 -6.35417700e-01 -4.52253193e-01
1.49012670e-01 -5.57095170e-01 -6.51785672e-01 7.24139750e-01
1.02567112e+00 -6.45972610e-01 -1.05624115e+00 9.61074114e-01
6.30152345e-01 -2.07919739e-02 1.22900832e+00 -8.92326534e-01
-1.29769909e+00 3.99072617e-01 -7.93692291e-01 2.08633021e-01
5.35135925e-01 -1.52625933e-01 5.13603866e-01 -1.59352410e+00
8.72688472e-01 1.24723947e+00 4.34430778e-01 3.19058597e-01
-7.61910260e-01 -1.03062488e-01 1.03372231e-01 1.07261218e-01
-1.42823851e+00 -7.20331430e-01 6.48300231e-01 -6.77949548e-01
1.41034782e+00 3.86684448e-01 3.18938404e-01 1.40300417e+00
5.77850580e-01 1.69023454e+00 7.64048755e-01 -7.48921931e-01
1.20058656e-01 -8.28533024e-02 2.39183605e-01 7.52657056e-01
1.01162326e-02 -2.76291877e-01 -9.57522452e-01 1.71103522e-01
5.13855398e-01 1.59828037e-01 5.52701913e-02 -3.47099900e-01
-1.50069511e+00 7.14152932e-01 -2.88336664e-01 5.41009188e-01
-3.42332184e-01 -7.53983259e-02 6.52251303e-01 4.43824977e-01
8.77491117e-01 -5.27688682e-01 2.84468174e-01 -4.55683738e-01
-1.52182186e+00 -1.29078209e-01 9.12665367e-01 1.00261235e+00
1.13519188e-02 6.55085504e-01 -3.36814106e-01 8.30757082e-01
5.52199721e-01 1.00453007e+00 1.07658148e+00 -6.18498400e-02
7.35953331e-01 6.26031339e-01 -3.02693307e-01 -9.21410859e-01
1.11244299e-01 3.98696512e-01 -1.06131017e+00 4.02265266e-02
9.32780094e-04 1.39419851e-03 -1.23168015e+00 1.75405800e-01
-9.94555559e-03 3.43066081e-02 3.25222820e-01 7.16403365e-01
7.69385993e-01 8.70016038e-01 -4.10996228e-01 1.41362667e-01
7.49435425e-01 -1.66185260e+00 -9.41228747e-01 -1.84148818e-01
9.50699747e-01 -1.27340221e+00 9.00098860e-01 7.70360589e-01
-7.68323421e-01 -3.04926246e-01 -9.31486368e-01 -2.36873284e-01
-6.70730472e-01 5.11228502e-01 1.63953587e-01 1.02712929e+00
-1.26568913e+00 1.01340562e-01 -9.19595063e-01 -1.21237707e+00
3.60767990e-01 8.85170475e-02 -2.59919912e-01 -2.92211562e-01
-2.10049301e-01 6.12467885e-01 6.19635507e-02 -3.81451808e-02
-8.57062459e-01 1.63420290e-01 -6.03728831e-01 -2.70411670e-01
1.07202515e-01 -2.96166509e-01 1.11365938e+00 -1.32867253e+00
-1.87289822e+00 9.54966009e-01 -5.54841280e-01 -5.37419140e-01
1.06931412e+00 -2.80116737e-01 -5.87623775e-01 3.08372974e-01
-2.57865638e-01 3.65046024e-01 1.65659630e+00 -8.39144468e-01
-5.57360113e-01 -3.90637010e-01 -7.33213723e-01 4.50946748e-01
-7.99491704e-01 3.60648185e-01 -8.64962697e-01 -1.06287634e+00
1.99031293e-01 -6.59594595e-01 3.21207911e-01 4.52504903e-01
-5.02920449e-01 -2.21030176e-01 1.87134707e+00 -6.86081052e-01
9.81463313e-01 -2.25080156e+00 3.84297743e-02 -8.23322311e-02
3.48201662e-01 3.17964524e-01 -2.94606268e-01 7.84791946e-01
2.36736238e-01 4.64754924e-02 1.52465543e-02 -9.19011652e-01
1.56340539e-01 2.94355899e-02 -7.04897523e-01 9.72565174e-01
-2.85885423e-01 1.14604199e+00 -3.41672778e-01 -6.98135972e-01
6.37459517e-01 5.34578264e-01 -7.93103576e-02 9.67921689e-02
-4.19351868e-02 -2.18404397e-01 -5.85095942e-01 7.51664400e-01
8.40747178e-01 3.69101986e-02 -3.72668028e-01 1.98796824e-01
-1.76963910e-01 -4.78954971e-01 -1.03472221e+00 1.67815137e+00
2.15669960e-01 1.49895346e+00 -9.46368054e-02 -9.01095390e-01
9.35438395e-01 1.03437327e-01 3.97940338e-01 -4.88779217e-01
2.31440455e-01 -3.65856625e-02 -3.91024172e-01 -6.14001572e-01
1.19456565e+00 4.05019134e-01 3.59496623e-01 7.44273782e-01
-9.78601351e-02 -3.82506341e-01 1.72809720e-01 5.34390748e-01
1.05031538e+00 9.07588005e-03 -2.14650512e-01 -1.15916215e-01
4.40346301e-01 1.84904858e-01 -6.31731272e-01 1.10352135e+00
-1.72473639e-01 8.82205606e-01 3.73240039e-02 -3.60551953e-01
-1.15839374e+00 -7.21488357e-01 4.33940813e-02 1.11601818e+00
-7.62372166e-02 -4.12130743e-01 -9.69900787e-01 -9.19628024e-01
1.06015112e-02 6.25433385e-01 -7.05832481e-01 2.59781867e-01
-9.46623012e-02 -5.91971517e-01 1.26323926e+00 4.59914088e-01
1.01150954e+00 -6.98705018e-01 -1.53607488e-01 -1.51543185e-01
3.16140175e-01 -1.44468868e+00 -1.07084394e+00 -1.98848560e-01
-8.30975831e-01 -5.50722241e-01 -1.31536412e+00 -1.07506108e+00
7.34333634e-01 7.24574685e-01 3.77094299e-01 -1.65911585e-01
-3.37088943e-01 1.11189413e+00 -6.79520607e-01 -2.20745847e-01
-6.03017807e-01 -2.04510197e-01 -6.34835213e-02 6.06850386e-01
6.50657356e-01 2.75907725e-01 -4.61376101e-01 4.60210621e-01
-1.36691380e+00 2.34889925e-01 2.86694348e-01 5.81827521e-01
3.09961081e-01 1.58514857e-01 -1.31821096e-01 -8.90667796e-01
9.69277263e-01 -8.32141284e-03 -4.39291149e-01 5.26961386e-01
-8.18804562e-01 -2.59158611e-01 3.14500928e-01 -7.59207845e-01
-1.18891203e+00 3.18378031e-01 6.37502074e-02 -5.80319405e-01
-3.29770923e-01 5.69043338e-01 3.71339202e-01 -4.25289929e-01
6.73714399e-01 1.19499397e+00 -2.18576908e-01 -2.35047847e-01
5.56804836e-01 1.41770101e+00 2.06957892e-01 -2.73686647e-01
5.48510432e-01 9.45665836e-01 -3.72931987e-01 -1.68569529e+00
-1.92410722e-01 -7.79852569e-01 -1.12064993e+00 -3.91473532e-01
9.70006466e-01 -7.44343281e-01 -2.03952298e-01 1.28245068e+00
-1.00386095e+00 -6.57314718e-01 -5.58464378e-02 2.40688846e-01
-5.15642822e-01 1.17564607e+00 -7.75117934e-01 -8.04734170e-01
-3.98934990e-01 -8.93710792e-01 1.66110837e+00 -1.28890514e-01
2.02157274e-01 -1.53329515e+00 1.54388726e-01 3.11995149e-01
5.00751972e-01 -3.65021020e-01 4.10179466e-01 -8.31414759e-01
-4.37924057e-01 -7.09917665e-01 -3.20938110e-01 3.05474639e-01
1.47676060e-03 4.81028110e-01 -8.76214981e-01 -2.39870191e-01
1.00658864e-01 -2.44578212e-01 9.91080523e-01 1.50322065e-01
8.42722833e-01 -1.38421087e-02 -3.62773567e-01 5.73083401e-01
1.11140633e+00 -4.77682054e-02 9.22670782e-01 2.45514721e-01
1.05776405e+00 3.84486094e-02 2.18455821e-01 3.86835277e-01
3.02973181e-01 4.40492600e-01 -3.88455749e-01 -2.41016880e-01
-5.07485688e-01 -3.74310493e-01 7.16731906e-01 1.18403780e+00
6.36867225e-01 -1.05111790e+00 -1.25800657e+00 2.13146597e-01
-1.96515024e+00 -8.83130729e-01 -5.66459477e-01 1.56726742e+00
1.52905017e-01 1.74828358e-02 -4.72346731e-02 2.07801715e-01
6.51514947e-01 4.76226747e-01 -5.20228446e-01 -4.90960658e-01
-7.46298611e-01 -4.49213475e-01 5.57367504e-01 3.81002754e-01
-1.02287102e+00 1.75013947e+00 7.46460629e+00 1.10083163e+00
-1.61746407e+00 -7.28241401e-04 4.03990716e-01 3.53697330e-01
3.38875532e-01 -2.19280869e-01 -7.09083080e-01 -1.46766260e-01
7.29131699e-01 -4.48219299e-01 4.22232091e-01 7.59893954e-01
3.55485529e-01 -2.46398658e-01 -1.16493285e+00 1.28763390e+00
1.12257504e+00 -1.06808031e+00 6.76432133e-01 -3.00772692e-04
9.86221611e-01 4.49055851e-01 2.96921343e-01 -7.60210603e-02
2.19038844e-01 -8.32238615e-01 9.84047771e-01 5.10508657e-01
8.52595270e-01 5.20713888e-02 2.96172678e-01 5.17113268e-01
-1.11905324e+00 4.02146369e-01 -5.88939488e-01 3.35748851e-01
-1.01804934e-01 3.05577934e-01 -1.01290083e+00 2.11362779e-01
1.96451902e-01 1.40969384e+00 -9.47892785e-01 8.34886551e-01
3.18389267e-01 9.29939568e-01 -1.74243063e-01 -4.67725843e-01
3.76080126e-01 -7.56253377e-02 4.58708376e-01 1.88106370e+00
4.56114709e-01 -2.94821709e-01 7.50204697e-02 5.53814113e-01
-1.44568712e-01 4.87356961e-01 -1.11012232e+00 -7.42433548e-01
-3.47850740e-01 9.68117476e-01 -1.22246814e+00 -7.36402869e-01
-4.88764644e-01 2.18306637e+00 -2.05342382e-01 6.70920491e-01
-4.49173361e-01 -3.88847291e-01 -3.26825351e-01 -2.34053060e-01
6.23920500e-01 -5.69127321e-01 -4.97799546e-01 -1.68081832e+00
-3.37326750e-02 -8.60871136e-01 1.61077783e-01 -1.31693316e+00
-1.33495224e+00 5.11081278e-01 -3.88253212e-01 -1.10038674e+00
-1.98498424e-02 -9.27861989e-01 -5.32367110e-01 6.14984214e-01
-1.11736369e+00 -1.55325198e+00 -2.58188099e-01 1.02482831e+00
1.55580437e+00 -5.48312485e-01 6.61934197e-01 -9.35538858e-02
-5.43011725e-01 6.15451217e-01 9.61349607e-01 3.69728655e-01
9.12802041e-01 -1.05492508e+00 9.02483046e-01 1.15854561e+00
6.25850201e-01 1.00121558e-01 4.35824692e-01 -1.01311123e+00
-2.10611320e+00 -1.14013433e+00 7.28096843e-01 -8.82745624e-01
8.99249256e-01 -6.02470994e-01 -8.36694181e-01 6.43564820e-01
6.94332898e-01 -2.70803332e-01 2.68563241e-01 -5.48738658e-01
-2.93546289e-01 3.70119005e-01 -9.20089066e-01 8.47730577e-01
7.18413234e-01 -8.36915851e-01 -4.69361544e-01 6.62919283e-01
5.13838232e-02 -5.49077749e-01 -5.02703309e-01 -4.56925839e-01
5.40061414e-01 -5.89381158e-01 4.07913744e-01 3.49027440e-02
4.05452579e-01 -3.51497419e-02 -3.34883332e-01 -7.23961711e-01
1.70576409e-01 -6.82021916e-01 -1.00338213e-01 1.13411045e+00
1.38339669e-01 -5.06684482e-01 8.18050921e-01 4.45203871e-01
2.21066356e-01 1.40205190e-01 -9.11338866e-01 -8.60949814e-01
3.64103734e-01 -6.80686951e-01 -1.10260904e-01 1.24697638e+00
-1.35083275e-03 3.61377269e-01 -5.11386752e-01 -1.10027865e-01
5.19672513e-01 -2.87579596e-01 9.49250400e-01 -8.63389909e-01
1.62341923e-01 -6.59086883e-01 -6.29270792e-01 -1.69654858e+00
1.52058437e-01 -8.59369278e-01 7.25019574e-02 -1.67466903e+00
2.70523280e-01 4.40761656e-01 3.74760360e-01 9.15512592e-02
2.68410176e-01 1.10036500e-01 2.22202659e-01 6.59595191e-01
-8.81974697e-01 6.32540703e-01 1.05764985e+00 -7.88229764e-01
3.33889946e-02 -3.78251165e-01 -1.09602958e-01 5.16829371e-01
4.37274933e-01 -4.53845680e-01 -3.02826732e-01 -9.68378186e-01
-5.99000715e-02 -1.12171508e-01 -1.10589430e-01 -8.56529951e-01
9.98146117e-01 1.96591462e-03 3.69527221e-01 -9.92867351e-01
1.71041653e-01 -8.06075931e-01 -4.12005633e-01 1.58408180e-01
-5.50342262e-01 1.97674617e-01 1.97683528e-01 7.55264044e-01
-2.49928936e-01 -2.03788072e-01 4.12472606e-01 4.48210716e-01
-5.90230882e-01 8.31755921e-02 -1.17095447e+00 -3.84804428e-01
9.41282094e-01 -6.76955462e-01 -5.57861745e-01 -6.46109760e-01
-2.46655360e-01 -2.69270539e-01 6.47076488e-01 8.19888115e-01
1.20079076e+00 -8.89709055e-01 -8.76414001e-01 4.78115052e-01
1.76781714e-01 -4.44118202e-01 1.22778432e-03 7.52593875e-01
-8.64741266e-01 6.26560688e-01 4.13783073e-01 -1.01993406e+00
-1.67218149e+00 5.47656775e-01 3.51084143e-01 3.08314916e-02
-9.36943769e-01 4.53498870e-01 1.98647037e-01 -6.71931058e-02
4.79493499e-01 -3.38028908e-01 1.53559119e-01 -3.13034773e-01
6.58123970e-01 1.91570774e-01 2.45157287e-01 -8.87708604e-01
-5.53341843e-02 8.84825766e-01 -5.72897732e-01 -4.42328751e-01
1.02894771e+00 -4.19873089e-01 9.78989899e-02 7.28636265e-01
9.54669178e-01 1.10603562e-02 -8.76678467e-01 -3.77507061e-01
-7.26830661e-02 -6.07838809e-01 5.90926528e-01 -8.45904827e-01
-7.13459134e-01 1.17302585e+00 1.06378889e+00 2.03322127e-01
8.78315210e-01 -4.56263304e-01 5.21610796e-01 1.15978050e+00
-2.47334391e-01 -1.37560248e+00 4.55640256e-01 9.68922555e-01
9.40922379e-01 -1.09160352e+00 6.08829316e-03 -1.98176429e-01
-1.08376896e+00 1.50680923e+00 1.69171259e-01 -1.46296650e-01
7.21858382e-01 5.52079141e-01 2.88543642e-01 -1.60995036e-01
-8.20481896e-01 5.15352897e-02 3.52960467e-01 6.77201271e-01
4.83543515e-01 -3.30143064e-01 2.87744790e-01 -1.90212801e-01
3.06440771e-01 1.57588497e-01 6.84744000e-01 1.33568263e+00
-3.86178046e-01 -9.10186052e-01 -5.99399865e-01 5.03201842e-01
-2.84272343e-01 -4.46577191e-01 -1.30435884e+00 4.55031216e-01
-9.47303295e-01 9.44678307e-01 1.01580173e-02 -5.13822973e-01
3.06643784e-01 2.73271501e-01 3.89755487e-01 -2.95058191e-01
-5.10384381e-01 5.41643441e-01 -2.87591130e-01 7.96354711e-02
-4.21227425e-01 -7.63152122e-01 -9.51688588e-01 -6.61492586e-01
-5.10079026e-01 -4.23760980e-01 9.98803139e-01 8.67494524e-01
3.63735884e-01 1.73389927e-01 5.05487859e-01 -7.07828581e-01
-2.21568078e-01 -1.17539668e+00 -9.23528314e-01 2.12542504e-01
2.25836158e-01 1.14593342e-01 -2.06132576e-01 9.57798779e-01]
|
[11.988736152648926, 2.3028056621551514]
|
c191d6ed-b9c5-4bce-b784-dd3df7603e1e
|
quantifying-gender-biases-towards-politicians
|
2112.12014
| null |
https://arxiv.org/abs/2112.12014v2
|
https://arxiv.org/pdf/2112.12014v2.pdf
|
Quantifying Gender Biases Towards Politicians on Reddit
|
Despite attempts to increase gender parity in politics, global efforts have struggled to ensure equal female representation. This is likely tied to implicit gender biases against women in authority. In this work, we present a comprehensive study of gender biases that appear in online political discussion. To this end, we collect 10 million comments on Reddit in conversations about male and female politicians, which enables an exhaustive study of automatic gender bias detection. We address not only misogynistic language, but also other manifestations of bias, like benevolent sexism in the form of seemingly positive sentiment and dominance attributed to female politicians, or differences in descriptor attribution. Finally, we conduct a multi-faceted study of gender bias towards politicians investigating both linguistic and extra-linguistic cues. We assess 5 different types of gender bias, evaluating coverage, combinatorial, nominal, sentimental, and lexical biases extant in social media language and discourse. Overall, we find that, contrary to previous research, coverage and sentiment biases suggest equal public interest in female politicians. Rather than overt hostile or benevolent sexism, the results of the nominal and lexical analyses suggest this interest is not as professional or respectful as that expressed about male politicians. Female politicians are often named by their first names and are described in relation to their body, clothing, or family; this is a treatment that is not similarly extended to men. On the now banned far-right subreddits, this disparity is greatest, though differences in gender biases still appear in the right and left-leaning subreddits. We release the curated dataset to the public for future studies.
|
['Isabelle Augenstein', 'Karolina Stańczak', 'Sara Marjanovic']
|
2021-12-22
| null | null | null | null |
['gender-bias-detection', 'gender-bias-detection']
|
['miscellaneous', 'natural-language-processing']
|
[-1.45225197e-01 5.05383730e-01 -8.57870996e-01 -6.09188974e-01
-5.00567257e-01 -1.14410257e+00 1.26534414e+00 4.86683577e-01
-7.04114676e-01 6.88061178e-01 1.25335503e+00 -6.54259622e-01
8.74237809e-03 -6.33530974e-01 -1.62252918e-01 -5.94717264e-01
6.46199882e-01 4.08600777e-01 -4.90644664e-01 -5.65505028e-01
5.80151439e-01 2.14489698e-01 -1.13234878e+00 -1.90441925e-02
5.88774681e-01 3.42337757e-01 -8.11132967e-01 5.20341210e-02
-1.80681065e-01 8.97686779e-01 -6.90041780e-01 -1.14453781e+00
1.66163631e-02 -2.11213201e-01 -9.04307246e-01 2.34638108e-03
1.04483867e+00 -1.35465115e-01 -2.82113373e-01 1.08849359e+00
6.24334097e-01 -3.54695916e-01 7.82987952e-01 -6.89007759e-01
-4.37399149e-01 1.08653474e+00 -9.89347339e-01 4.24932301e-01
2.28044271e-01 6.07357994e-02 1.19348383e+00 -5.82954168e-01
1.17252839e+00 1.86865819e+00 4.44459200e-01 5.25199592e-01
-1.42524195e+00 -1.05810142e+00 3.25921506e-01 -5.29645860e-01
-1.08120608e+00 -7.01804519e-01 4.57890213e-01 -1.00656843e+00
-2.62190867e-02 6.25170529e-01 6.79254770e-01 1.40015602e+00
2.35262245e-01 3.22625227e-02 1.76342952e+00 1.78231671e-02
-1.56691894e-01 4.24848676e-01 2.62910247e-01 1.39114052e-01
9.00904059e-01 -2.62010008e-01 -4.75866139e-01 -7.76418030e-01
2.37915516e-01 -4.31228638e-01 3.41499597e-02 2.31563881e-01
-1.25120997e+00 1.05138540e+00 1.57795697e-01 5.35946846e-01
-2.29962796e-01 1.02783270e-01 9.28578854e-01 -7.46425660e-03
9.21885908e-01 5.23190200e-01 -3.03594649e-01 -4.77649838e-01
-8.65771294e-01 5.63489914e-01 1.07593465e+00 3.19512337e-01
4.56599325e-01 -1.62850246e-01 -3.62200052e-01 7.86046982e-01
6.69120923e-02 7.77744412e-01 2.05922365e-01 -9.29622531e-01
5.99831998e-01 4.45959061e-01 2.23518670e-01 -1.74903870e+00
-2.49184564e-01 -4.42376614e-01 -5.70041239e-01 6.08879440e-02
9.89013195e-01 -8.04684907e-02 -3.85297537e-01 2.03605747e+00
4.88661140e-01 -1.10744178e+00 -2.52920985e-01 1.00778282e+00
8.34280968e-01 2.62420028e-01 6.03281260e-01 -1.58990458e-01
1.98518085e+00 7.13790432e-02 -6.96117163e-01 -4.90468144e-01
3.22751820e-01 -1.04292178e+00 8.36375833e-01 -8.40082914e-02
-7.65493095e-01 9.10531580e-02 -4.02146250e-01 -1.93711221e-01
-3.47691357e-01 -4.05525565e-02 6.79477036e-01 9.85966861e-01
-3.09631497e-01 5.22343338e-01 -3.64206672e-01 -7.48398304e-01
4.50730950e-01 -8.56626853e-02 -2.24240974e-01 4.37924653e-01
-1.01518750e+00 8.59496891e-01 -2.59469092e-01 -1.15561053e-01
-3.27433378e-01 -5.60193777e-01 -9.04677689e-01 -4.97633755e-01
5.66593349e-01 -1.81390584e-01 1.18471158e+00 -1.23910296e+00
-8.06773365e-01 1.80434132e+00 -3.14083427e-01 4.99784686e-02
7.05587208e-01 -1.94189444e-04 -5.62865496e-01 -2.36514971e-01
9.71032441e-01 5.19973695e-01 5.54824889e-01 -1.17418909e+00
-6.25179648e-01 -6.73242986e-01 2.61584789e-01 2.00499535e-01
-1.71752974e-01 7.53669679e-01 2.89129645e-01 -8.09743047e-01
2.44565591e-01 -8.92950535e-01 6.66035190e-02 -3.76105547e-01
-6.61554217e-01 -5.71774244e-01 5.36294639e-01 -8.18852782e-01
1.34505594e+00 -2.19291973e+00 -2.39026427e-01 2.37856492e-01
5.42786837e-01 -4.21871275e-01 6.22574687e-01 6.51523769e-01
-6.56772181e-02 7.22300947e-01 1.85341194e-01 1.80768266e-01
3.27300817e-01 2.70692617e-01 -5.91668069e-01 1.02016175e+00
-2.50664294e-01 5.73852599e-01 -8.85116518e-01 -6.60679936e-01
-4.32392627e-01 1.39466763e-01 -3.52563918e-01 -7.95296311e-01
1.75821394e-01 5.97720861e-01 -3.37652862e-01 1.05189574e+00
5.24711907e-01 3.21846753e-02 7.24329412e-01 -3.03840190e-01
-7.99221456e-01 7.82675862e-01 -6.65489018e-01 8.89869809e-01
-1.77501783e-01 1.05387759e+00 7.79241323e-01 -4.23006147e-01
6.68365777e-01 1.38803734e-04 1.69857875e-01 -4.47735876e-01
5.98534107e-01 6.73284292e-01 4.72731918e-01 -1.65811703e-01
1.15070295e+00 -4.49450105e-01 -8.37423742e-01 5.52011371e-01
-5.80721140e-01 -2.57384896e-01 4.03665751e-01 3.02807331e-01
3.55998337e-01 -8.30606222e-02 4.32721943e-01 -1.02395821e+00
6.14026785e-02 3.13882440e-01 8.30205202e-01 5.63510537e-01
-2.41244733e-01 1.73764437e-01 1.04782987e+00 -3.78341407e-01
-9.35520351e-01 -5.69426179e-01 -5.57499230e-01 1.56341791e+00
1.60193935e-01 -5.14484942e-01 -4.67650771e-01 -6.20078266e-01
2.16666654e-01 6.57081187e-01 -7.25747287e-01 3.57648402e-01
-5.98278821e-01 -7.44446456e-01 6.19865239e-01 -4.55509648e-02
2.00086683e-01 -2.38563970e-01 -5.88049173e-01 -2.14311644e-01
-3.74947429e-01 -9.30091798e-01 -4.46951240e-01 -2.72645980e-01
-4.43317235e-01 -1.11319709e+00 -5.99393189e-01 -2.38770276e-01
5.40535569e-01 -6.85687736e-02 1.20467198e+00 -4.60884757e-02
5.80990352e-02 2.69107431e-01 1.22688383e-01 -6.48692906e-01
-7.40994573e-01 2.30167076e-01 8.90511572e-02 -1.74636766e-01
5.19831061e-01 -3.11668426e-01 -4.85800534e-01 2.17330232e-01
-5.57852149e-01 -3.03612560e-01 2.55878031e-01 3.97792816e-01
-1.72655657e-01 -5.87117434e-01 1.87835276e-01 -1.36338139e+00
6.83805823e-01 -7.31724799e-01 -2.82612264e-01 -3.28866094e-01
-6.66339219e-01 -4.46032614e-01 1.90445647e-01 -2.83446252e-01
-8.37770641e-01 -1.01276767e+00 1.84973553e-01 5.88418365e-01
3.20073068e-02 4.31470215e-01 3.21000934e-01 2.44268119e-01
8.82427633e-01 -4.52696443e-01 1.63166836e-01 -3.51537675e-01
-2.89699044e-02 9.28041935e-01 3.70971054e-01 -1.00332463e+00
7.98271239e-01 7.77427554e-01 -3.45364183e-01 -7.83577144e-01
-1.17880988e+00 -1.25858530e-01 -1.84011031e-02 -3.64181966e-01
9.48803723e-01 -1.23263633e+00 -1.08661044e+00 1.11027092e-01
-8.54014993e-01 7.59952329e-03 -1.05979741e-01 5.39480805e-01
-6.62054867e-02 2.57519066e-01 -7.51740575e-01 -8.79949510e-01
-4.05541137e-02 -9.10074532e-01 9.28226113e-01 3.39794084e-02
-1.17968416e+00 -9.24795926e-01 -8.72198641e-02 5.85114777e-01
3.36309522e-01 6.34192646e-01 8.30538511e-01 -7.52404213e-01
3.05274665e-01 1.40603874e-02 -3.17134529e-01 -1.94138259e-01
2.66868055e-01 2.26001620e-01 -6.96717978e-01 -1.54173240e-01
-3.94028388e-02 -4.57877874e-01 5.30132771e-01 2.58176595e-01
6.40159667e-01 -7.93338418e-01 -4.32312369e-01 -9.61907879e-02
1.21953630e+00 -3.69873673e-01 1.07468188e-01 6.92615449e-01
4.31620598e-01 1.05169928e+00 6.46970809e-01 6.06915891e-01
7.54581809e-01 5.72576880e-01 1.49088815e-01 9.08783171e-03
2.52485983e-02 -3.63764495e-01 3.94619435e-01 2.21835300e-01
-3.46039981e-01 2.23802119e-01 -1.16222286e+00 7.05843925e-01
-1.30703020e+00 -1.26782084e+00 -4.68993306e-01 1.88100708e+00
1.04749787e+00 1.85575426e-01 3.07001352e-01 2.11080518e-02
9.34949994e-01 7.39666045e-01 -6.00065477e-02 -8.14491928e-01
-2.64259964e-01 -7.05884993e-02 8.95356238e-01 5.92232168e-01
-1.11989582e+00 7.82508075e-01 6.69993448e+00 8.70483220e-01
-1.35997367e+00 2.77081057e-02 1.10474038e+00 -7.30443895e-02
-6.51015162e-01 1.91511348e-01 -9.19455886e-01 4.46238548e-01
4.17192131e-01 -4.56291139e-01 -7.01473877e-02 8.06737483e-01
3.82857233e-01 -3.97977084e-01 -7.36737669e-01 6.86145484e-01
-4.30815779e-02 -1.07111275e+00 -4.73912835e-01 5.86645484e-01
6.07549429e-01 -2.36502707e-01 1.77377298e-01 1.47580266e-01
5.48456609e-01 -9.99381423e-01 1.50303996e+00 -1.34991929e-01
9.72480953e-01 -4.44004238e-01 5.45024633e-01 -1.53660532e-02
-3.36007655e-01 -2.06112444e-01 -1.62755981e-01 -6.39700353e-01
3.07789803e-01 6.39943957e-01 -4.20768887e-01 1.36350065e-01
5.99249184e-01 5.32491505e-01 -3.48142564e-01 -3.37125026e-02
-1.76092878e-01 8.73915732e-01 -2.20980123e-01 -3.22224088e-02
4.87209767e-01 -2.77800053e-01 9.22310054e-01 1.12781620e+00
-4.48902398e-02 -7.92160165e-03 3.52929272e-02 5.81951380e-01
2.25648228e-02 3.03347200e-01 -5.87823689e-01 -4.18830812e-01
6.03434801e-01 1.69174480e+00 -8.69094908e-01 -4.07847196e-01
-3.59792858e-01 1.83305010e-01 1.60626262e-01 2.58490860e-01
-6.57895684e-01 2.39787266e-01 8.39143217e-01 7.75886893e-01
-4.42273945e-01 -1.96920827e-01 -5.64204812e-01 -9.90334213e-01
-2.83962369e-01 -1.17777550e+00 5.27861118e-01 -6.49389476e-02
-1.33221304e+00 3.69446538e-02 2.88406789e-01 -6.83563769e-01
-3.89563292e-03 -3.57080013e-01 -1.96623370e-01 8.65058661e-01
-8.36731493e-01 -1.15166998e+00 -1.33518735e-02 -1.38646826e-01
5.67945242e-02 6.38806075e-02 5.08415580e-01 2.80033827e-01
-3.30766350e-01 2.68297851e-01 -3.52682441e-01 2.73839802e-01
1.22000968e+00 -9.78111148e-01 -2.37612482e-02 2.69474715e-01
-3.51985991e-01 7.14300156e-01 1.31311131e+00 -7.21250415e-01
-9.16667581e-01 -5.27503073e-01 1.42082894e+00 -6.96112692e-01
1.00075662e+00 -2.97096968e-01 -5.64853288e-02 7.78500319e-01
5.88104188e-01 -6.79950774e-01 8.89788747e-01 6.79828584e-01
-6.78707778e-01 1.94413900e-01 -1.16686916e+00 8.49520445e-01
1.07176518e+00 -3.57516855e-01 -5.68445623e-01 2.62302876e-01
2.43508369e-01 -4.31875974e-01 -8.07091177e-01 1.36038825e-01
8.61054718e-01 -9.22916174e-01 5.19194901e-01 -6.04002714e-01
8.47169518e-01 -8.83297413e-04 -2.79607713e-01 -9.27541316e-01
-4.63180810e-01 -4.93439734e-01 1.03056026e+00 1.64231980e+00
3.82555842e-01 -7.38603890e-01 4.52460527e-01 8.45182955e-01
-3.08471955e-02 -5.02735496e-01 -1.01184666e+00 -2.71225333e-01
4.76065934e-01 -9.42826550e-03 2.58337498e-01 1.63857055e+00
1.42062739e-01 4.68861908e-01 -1.59486711e-01 -2.93557465e-01
3.85308146e-01 3.40915591e-01 1.05889368e+00 -1.25294375e+00
1.65529642e-02 -7.71879673e-01 -2.92019963e-01 -4.22670633e-01
2.85586834e-01 -8.54313016e-01 -5.74712038e-01 -1.25112486e+00
4.70299244e-01 -4.44538802e-01 6.57400310e-01 1.68198720e-01
1.40742302e-01 3.93492550e-01 1.73699632e-01 3.85604262e-01
-1.53894741e-02 2.58436147e-02 1.29411614e+00 -1.67801961e-01
-4.79067788e-02 -2.71033376e-01 -1.63669145e+00 1.04411101e+00
6.49544716e-01 -4.81215745e-01 3.11389238e-01 -2.04225585e-01
9.61745143e-01 -3.38719964e-01 5.22168458e-01 -3.98170173e-01
-2.80855060e-01 -6.22112751e-01 2.55183488e-01 -3.12049598e-01
2.02838138e-01 -4.35120642e-01 1.55886203e-01 5.38761675e-01
-1.96752205e-01 3.11011188e-02 5.02981283e-02 1.54030621e-01
-3.10944289e-01 9.21394601e-02 5.82076788e-01 -4.01160359e-01
-2.11899858e-02 -1.54600143e-01 -7.61180937e-01 4.80159760e-01
7.25756943e-01 -1.65932044e-01 -7.25784719e-01 -7.12715268e-01
-5.87816000e-01 -7.89037570e-02 8.54850411e-01 2.65959829e-01
-4.33654249e-01 -1.37159216e+00 -1.09457242e+00 -6.82948351e-01
2.70626009e-01 -6.15023911e-01 -3.52747142e-02 1.36207843e+00
-2.17866421e-01 2.56724767e-02 2.10222527e-01 -1.78804681e-01
-1.44408154e+00 8.51440653e-02 3.74678709e-02 1.00397401e-01
-2.63178498e-01 4.68045235e-01 3.65863144e-01 -5.60547531e-01
-3.17137599e-01 6.45605847e-02 -3.49230111e-01 1.00377178e+00
2.36895438e-02 6.61993027e-01 -5.55322886e-01 -1.32493126e+00
-5.61496675e-01 2.99203187e-01 6.52762651e-02 -4.84708428e-01
1.04652870e+00 -2.36851111e-01 -7.71741629e-01 6.95861697e-01
6.60571039e-01 1.22142065e+00 -1.96274400e-01 1.93160921e-01
-1.35029918e-02 -8.46044540e-01 -5.44915617e-01 -5.74838102e-01
-5.02684653e-01 9.60303769e-02 -1.55844375e-01 6.24700129e-01
2.31809288e-01 4.48225707e-01 1.95633665e-01 -3.32729012e-01
2.05398053e-01 -1.46750021e+00 -3.74469668e-01 4.00035560e-01
1.03916657e+00 -1.00476706e+00 4.97992843e-01 -7.22234488e-01
-6.50393605e-01 6.51538074e-01 5.15475988e-01 4.59663011e-02
4.14343089e-01 -5.64395450e-02 4.07295346e-01 -7.25621819e-01
-5.96413791e-01 7.92061388e-02 -2.54240092e-02 -4.44661602e-02
1.28770745e+00 4.06098694e-01 -1.72490013e+00 6.34829819e-01
-9.49451208e-01 -6.24866545e-01 7.82903135e-01 5.94030857e-01
-1.85795814e-01 -9.49335277e-01 -8.30174208e-01 4.87506062e-01
-1.35104859e+00 5.98202497e-02 -9.71860170e-01 1.10530031e+00
3.51013988e-01 1.01682019e+00 2.74419814e-01 -2.35967152e-02
6.84573650e-02 -2.39528641e-01 2.16420650e-01 -3.79862398e-01
-1.05677938e+00 7.33613819e-02 1.13620281e+00 -2.24420667e-01
-6.27869546e-01 -9.57041025e-01 -5.92851698e-01 -9.62594271e-01
-9.26051959e-02 1.83982164e-01 6.60448968e-01 6.95507705e-01
1.43589124e-01 -1.25363007e-01 3.71482402e-01 -3.99165958e-01
-7.20405161e-01 -1.04407024e+00 -6.85622573e-01 5.82680106e-01
1.51176993e-02 -5.72427988e-01 -5.50753474e-01 -3.71668607e-01]
|
[9.033770561218262, 10.009323120117188]
|
2f6a8d42-7432-4ec2-bd97-c95aae8cfaec
|
towards-unsupervised-visual-reasoning-do-off
|
2212.10292
| null |
https://arxiv.org/abs/2212.10292v1
|
https://arxiv.org/pdf/2212.10292v1.pdf
|
Towards Unsupervised Visual Reasoning: Do Off-The-Shelf Features Know How to Reason?
|
Recent advances in visual representation learning allowed to build an abundance of powerful off-the-shelf features that are ready-to-use for numerous downstream tasks. This work aims to assess how well these features preserve information about the objects, such as their spatial location, their visual properties and their relative relationships. We propose to do so by evaluating them in the context of visual reasoning, where multiple objects with complex relationships and different attributes are at play. More specifically, we introduce a protocol to evaluate visual representations for the task of Visual Question Answering. In order to decouple visual feature extraction from reasoning, we design a specific attention-based reasoning module which is trained on the frozen visual representations to be evaluated, in a spirit similar to standard feature evaluations relying on shallow networks. We compare two types of visual representations, densely extracted local features and object-centric ones, against the performances of a perfect image representation using ground truth. Our main findings are two-fold. First, despite excellent performances on classical proxy tasks, such representations fall short for solving complex reasoning problem. Second, object-centric features better preserve the critical information necessary to perform visual reasoning. In our proposed framework we show how to methodologically approach this evaluation.
|
['David Picard', 'Tomasz Trzciński', 'Tom Monnier', 'Monika Wysoczańska']
|
2022-12-20
| null | null | null | null |
['visual-reasoning', 'visual-reasoning']
|
['computer-vision', 'reasoning']
|
[-3.57193276e-02 2.23581731e-01 1.16787516e-01 -4.52197105e-01
-4.79433537e-01 -7.44225800e-01 9.34095383e-01 6.22269571e-01
-6.02204680e-01 2.91672826e-01 4.90920156e-01 -3.03365022e-01
-3.30855936e-01 -7.87628591e-01 -7.64218867e-01 -5.65927625e-01
1.55429110e-01 4.11228597e-01 3.36968005e-01 -4.52332854e-01
5.59154510e-01 7.22697079e-01 -1.93873179e+00 8.61725986e-01
3.74522090e-01 8.77361178e-01 7.73724616e-02 6.34626389e-01
-2.54081964e-01 1.19360042e+00 -5.90306818e-01 -4.62650597e-01
-3.08657810e-02 -3.62862855e-01 -1.17373550e+00 -5.72245009e-02
7.65830278e-01 -2.05519408e-01 -3.43594521e-01 8.84663105e-01
2.41823375e-01 3.23911399e-01 9.38322604e-01 -1.13186824e+00
-9.33973551e-01 3.91670555e-01 -1.33334503e-01 6.06575072e-01
6.14060521e-01 3.37296367e-01 1.38113642e+00 -9.11144495e-01
8.13251555e-01 1.28744435e+00 4.45356458e-01 4.07272309e-01
-1.21590090e+00 5.88343441e-02 3.37499261e-01 5.15754342e-01
-1.10866296e+00 -5.19883215e-01 7.07076132e-01 -7.48876333e-01
1.06901836e+00 3.04650068e-01 6.31526470e-01 9.95263577e-01
-2.38591488e-02 7.51445234e-01 1.19282150e+00 -4.78683978e-01
2.77363181e-01 3.38556945e-01 3.66870373e-01 7.41235733e-01
2.27284625e-01 -7.12052211e-02 -5.14935374e-01 1.31505713e-01
6.21398151e-01 1.69655532e-01 -2.68352807e-01 -7.21008837e-01
-1.39013028e+00 8.08247447e-01 1.06267488e+00 7.57872164e-01
-4.18625563e-01 3.48402560e-01 3.19970280e-01 2.66402930e-01
7.87550434e-02 4.90544111e-01 -1.28052443e-01 2.52971321e-01
-7.79716313e-01 4.27482903e-01 4.16506886e-01 6.34148896e-01
7.70139456e-01 -2.53529668e-01 -8.21361661e-01 4.21241820e-01
3.69416595e-01 8.86477381e-02 4.72838283e-01 -7.94970691e-01
6.13684654e-01 8.75878513e-01 7.30484724e-02 -1.22344601e+00
-6.29425645e-01 -1.68060049e-01 -5.64864278e-01 4.88204390e-01
6.80656850e-01 4.30610806e-01 -8.27624142e-01 1.76027286e+00
3.03892553e-01 -4.94733185e-01 1.03859000e-01 1.13539028e+00
1.10264063e+00 3.47795099e-01 2.70193726e-01 3.06881160e-01
1.81081343e+00 -9.56811011e-01 -4.78304863e-01 -1.86718658e-01
6.27697885e-01 -4.38380957e-01 1.23896933e+00 5.96308447e-02
-1.22593391e+00 -7.00131655e-01 -9.49848652e-01 -6.13321841e-01
-1.00496197e+00 -5.79471700e-02 6.68138266e-01 3.95125359e-01
-1.34668183e+00 5.64311683e-01 -3.10562521e-01 -5.31429529e-01
6.93005025e-01 7.68159851e-02 -7.04744935e-01 -6.05761781e-02
-1.00391686e+00 1.30804241e+00 3.60488087e-01 1.81068584e-01
-8.87234032e-01 -5.06001770e-01 -9.88071740e-01 4.29379970e-01
1.81262419e-01 -8.64545524e-01 1.07784605e+00 -9.71186280e-01
-8.79881084e-01 1.31111526e+00 -1.75210938e-01 -2.54785269e-01
6.09305382e-01 5.70573099e-02 -3.06716692e-02 5.13253808e-01
1.61446154e-01 5.26424110e-01 7.50055909e-01 -1.46476531e+00
-2.46722102e-01 -4.50970590e-01 5.85609674e-01 8.14578459e-02
-7.86889344e-02 -1.82391509e-01 -3.87601882e-01 -4.54639703e-01
3.08600459e-02 -4.97490764e-01 -4.88224588e-02 4.76224691e-01
-2.51367182e-01 -2.21181810e-01 4.50238377e-01 -5.06362140e-01
6.87801838e-01 -2.14265966e+00 3.21527332e-01 7.72629529e-02
5.20086586e-01 2.95843571e-01 -2.29227811e-01 6.19316816e-01
-3.62954646e-01 -4.24916930e-02 -1.16176680e-01 -2.23961562e-01
2.32051864e-01 9.05935764e-02 -3.54573280e-01 5.77754080e-01
4.55854714e-01 1.41904044e+00 -8.49137306e-01 -5.60246050e-01
4.67953354e-01 6.02781475e-01 -5.64135373e-01 1.90106481e-01
-2.00387672e-01 3.21045339e-01 -4.10110414e-01 6.13802910e-01
5.03226876e-01 -3.20675969e-01 7.76000172e-02 -3.85043472e-01
2.58123800e-02 1.53726041e-01 -8.15683901e-01 1.63359952e+00
-4.58358288e-01 8.75986159e-01 -1.72262982e-01 -1.21189463e+00
7.07947791e-01 5.83832674e-02 8.94052237e-02 -1.24671733e+00
1.92045748e-01 -2.76933610e-01 -6.32719249e-02 -7.31373787e-01
3.09525251e-01 -2.52055645e-01 1.40460662e-03 4.83731449e-01
4.16572541e-01 -1.25396103e-02 2.84076750e-01 3.33492011e-01
1.04802024e+00 3.16536427e-01 6.15343571e-01 -2.43983224e-01
7.13185847e-01 -7.62353418e-03 -2.13264301e-01 6.97593749e-01
-2.61305094e-01 7.39547551e-01 8.51806045e-01 -6.02832973e-01
-8.91173840e-01 -1.14439964e+00 1.08167827e-01 1.22531462e+00
1.30134255e-01 -4.15266275e-01 -5.47623217e-01 -7.38301814e-01
1.08052514e-01 6.78157628e-01 -1.23638391e+00 -1.92768604e-01
-4.12851125e-01 -3.13633740e-01 2.68143207e-01 6.17566764e-01
2.66310304e-01 -1.47291899e+00 -1.12512982e+00 -2.53173739e-01
8.01485479e-02 -9.01991785e-01 9.37825665e-02 2.19122455e-01
-5.58509469e-01 -1.53162932e+00 -6.87515795e-01 -6.02472067e-01
7.41057813e-01 4.37396020e-01 1.44655657e+00 4.09654588e-01
-4.30375278e-01 8.03391933e-01 -3.42980534e-01 -1.71458885e-01
-5.39067239e-02 -1.75487563e-01 -5.58064461e-01 6.32740632e-02
2.02457979e-01 -4.97643858e-01 -7.58570969e-01 -4.42931578e-02
-1.10415113e+00 -1.81301624e-01 7.92038739e-01 6.07205749e-01
4.32230055e-01 -3.58384162e-01 4.46309522e-02 -8.76917601e-01
6.42806351e-01 -5.00959158e-01 -2.39181161e-01 5.79045415e-01
-2.64440805e-01 5.39870262e-01 5.41483998e-01 -1.19574942e-01
-9.58439052e-01 -1.72290225e-02 -1.33680671e-01 -3.04715246e-01
-2.52041459e-01 3.16741943e-01 -1.72662158e-02 -1.84074175e-02
8.12706113e-01 1.77326575e-01 -2.43830934e-01 -2.36529037e-01
8.48810077e-01 9.08501074e-02 4.61050987e-01 -5.78866065e-01
8.17231476e-01 6.76512539e-01 3.33734713e-02 -6.00369930e-01
-9.94341969e-01 -5.72615027e-01 -9.28927302e-01 -1.83516145e-01
9.79473114e-01 -6.79015875e-01 -1.12826192e+00 -7.38983601e-02
-1.28409171e+00 -2.85235405e-01 -5.05733848e-01 1.09070882e-01
-8.68465483e-01 3.36923718e-01 -2.51357168e-01 -7.81860530e-01
-6.78744540e-02 -9.22289312e-01 1.24551189e+00 7.88297281e-02
-2.07499653e-01 -1.10481930e+00 2.54326850e-01 3.92777622e-01
3.80998075e-01 3.49790931e-01 1.06923556e+00 -5.60012877e-01
-6.07777894e-01 7.42850499e-03 -7.15592265e-01 -1.95209365e-02
-2.65621006e-01 -9.99472290e-02 -1.35296500e+00 -1.36929363e-01
-1.10487908e-01 -4.07275856e-01 1.29109216e+00 7.12946579e-02
1.14791000e+00 -1.19709931e-01 -2.69839793e-01 5.62019944e-01
1.58211207e+00 -2.36986712e-01 7.02771366e-01 3.36842746e-01
6.24880433e-01 9.90388989e-01 3.53090316e-01 1.68592662e-01
4.54277039e-01 7.12090790e-01 7.86959350e-01 -1.34283230e-01
-3.36702824e-01 -3.18196982e-01 7.51145184e-02 1.01292744e-01
-2.35570431e-01 -1.80102408e-01 -9.61308837e-01 6.37313545e-01
-1.91575301e+00 -1.20952380e+00 -1.43572137e-01 2.00896406e+00
4.15311962e-01 1.59694906e-02 1.82355270e-01 1.36204556e-01
3.07326645e-01 3.94056112e-01 -1.57899886e-01 -6.09522104e-01
-5.17483652e-02 1.58558920e-01 3.69520225e-02 3.50554734e-01
-9.29784238e-01 7.43576586e-01 6.37666321e+00 2.33321905e-01
-8.91960740e-01 1.48366198e-01 3.30830187e-01 3.27990018e-02
-5.32656431e-01 1.02452755e-01 -4.10986215e-01 -2.21283291e-03
8.51140738e-01 2.60936886e-01 3.65663886e-01 7.15412199e-01
-1.41370013e-01 -1.77589878e-01 -1.39701331e+00 9.80602026e-01
4.08653140e-01 -1.47196865e+00 4.00018841e-01 -1.87878743e-01
1.15653507e-01 -2.57700086e-01 1.78431094e-01 3.69399399e-01
1.51796579e-01 -1.34477341e+00 1.06800413e+00 9.34380710e-01
4.57945585e-01 -4.45014536e-01 6.26485169e-01 -2.56587714e-02
-1.08606529e+00 -1.69407487e-01 -3.88661087e-01 -1.93313554e-01
-4.17977870e-02 2.63201207e-01 -5.96791863e-01 5.94402075e-01
7.42345035e-01 6.59489095e-01 -1.17797756e+00 9.32519674e-01
-3.88797164e-01 -2.17582509e-02 1.82038546e-01 -2.13432908e-02
4.42324966e-01 1.86730862e-01 1.71821445e-01 1.29925632e+00
3.88684943e-02 -1.50885075e-01 -3.82320732e-01 1.20862484e+00
1.40591815e-01 1.13674887e-01 -8.02698255e-01 -5.33891432e-02
-1.35619655e-01 1.32927394e+00 -8.69589031e-01 -3.29930663e-01
-4.69286680e-01 9.94508564e-01 9.85403895e-01 4.30922717e-01
-7.93135166e-01 -1.85998291e-01 6.09552503e-01 1.58362567e-01
5.02390325e-01 -8.81179199e-02 3.55532058e-02 -1.29549289e+00
1.81692302e-01 -6.15760386e-01 5.35147846e-01 -1.20302188e+00
-1.38252473e+00 7.60489345e-01 2.52648424e-02 -9.23181415e-01
-1.68731406e-01 -1.11446738e+00 -6.64228559e-01 6.91923678e-01
-1.73338568e+00 -1.34227514e+00 -5.69969416e-01 8.75762641e-01
2.33686715e-01 2.97535211e-02 8.12741101e-01 -7.58686885e-02
-1.85842916e-01 2.72582173e-01 -2.49409452e-01 2.92072415e-01
3.87012988e-01 -1.36813962e+00 7.75288194e-02 5.68410456e-01
6.83776200e-01 7.25351751e-01 6.46042407e-01 -4.92547490e-02
-1.26264143e+00 -6.40191913e-01 9.12112236e-01 -8.79317760e-01
5.73420525e-01 -6.26314044e-01 -9.39484239e-01 7.49532342e-01
2.23770708e-01 3.82328421e-01 4.37761903e-01 2.81622142e-01
-9.20399189e-01 4.71968837e-02 -1.03799570e+00 6.98644340e-01
9.93527651e-01 -7.99612224e-01 -1.09126019e+00 3.65687370e-01
5.00007749e-01 -7.12323710e-02 -5.51724434e-01 -1.25910426e-02
5.63294411e-01 -1.53636110e+00 1.29918730e+00 -9.25960600e-01
4.22863036e-01 -2.77767897e-01 -3.61305714e-01 -1.02760768e+00
-4.37585771e-01 -1.70905322e-01 4.55835760e-02 1.15018725e+00
2.32619837e-01 -4.76767153e-01 4.57390547e-01 3.95412236e-01
1.61682114e-01 -4.09902543e-01 -8.20720792e-01 -3.09618771e-01
7.85935223e-02 -2.35680968e-01 3.49708647e-01 7.84063101e-01
7.73073956e-02 5.21058738e-01 1.06311284e-01 -5.81364259e-02
2.16228902e-01 4.17608082e-01 6.10323906e-01 -1.29351962e+00
-1.57478020e-01 -6.94772303e-01 -8.40597451e-01 -5.27880609e-01
2.61816949e-01 -1.04779112e+00 -1.43280432e-01 -1.83208740e+00
3.76651376e-01 -3.09315436e-02 -4.76812392e-01 5.77651680e-01
-8.91699642e-02 2.88608760e-01 5.29505610e-01 1.95462746e-03
-9.03957486e-01 4.41032529e-01 1.27813387e+00 -3.27459842e-01
1.65433645e-01 -2.84006178e-01 -7.98351526e-01 6.31655991e-01
6.45229995e-01 -2.06852794e-01 -4.52576667e-01 -3.13478380e-01
5.12949824e-01 -2.50805140e-01 1.03999496e+00 -9.55867052e-01
5.91234975e-02 -3.41863371e-02 6.83317065e-01 -3.88212591e-01
4.34444994e-01 -8.93640578e-01 -2.39246950e-01 4.12057936e-01
-5.85210085e-01 2.06621498e-01 2.11449385e-01 5.51155329e-01
-4.35438365e-01 -3.77139539e-01 6.15850687e-01 -4.46435034e-01
-9.24287260e-01 -6.85699806e-02 -2.38556936e-01 1.59630015e-01
1.11528993e+00 -1.78352803e-01 -5.86192131e-01 -3.69740188e-01
-8.28553021e-01 -1.07046030e-01 4.13511008e-01 3.37172955e-01
7.14801311e-01 -1.24344206e+00 -4.98144537e-01 -3.01289819e-02
6.10688388e-01 -3.78109068e-01 4.22895521e-01 7.54070044e-01
-6.17650867e-01 6.56347156e-01 -5.49742162e-01 -4.25492913e-01
-1.11603260e+00 1.27396679e+00 4.14241552e-01 -2.10545570e-01
-7.53701270e-01 7.05711246e-01 5.64282000e-01 -3.52779180e-01
7.51971751e-02 -7.17860103e-01 -5.61075687e-01 4.48003471e-01
6.62939966e-01 -6.89051971e-02 1.50690868e-01 -8.27773392e-01
-5.73275089e-01 6.78723276e-01 2.17323899e-01 -2.07361057e-02
1.39725959e+00 -1.05689671e-02 -1.35372862e-01 5.48420787e-01
1.18613696e+00 -6.50234800e-03 -1.17781830e+00 -8.32840428e-02
4.86213453e-02 -4.01356190e-01 -1.47869080e-01 -7.10054636e-01
-1.05455112e+00 1.27936029e+00 4.34926867e-01 4.03949648e-01
9.21245813e-01 3.21696967e-01 -1.26485884e-01 4.63298887e-01
6.08387291e-02 -6.75976753e-01 4.07295465e-01 1.64946884e-01
1.23586273e+00 -1.31132209e+00 2.06569657e-01 -9.03051347e-02
-7.11841047e-01 1.26583922e+00 3.61513674e-01 -1.91412047e-01
4.16737825e-01 -2.27245599e-01 -2.45222764e-04 -7.47697830e-01
-8.03199887e-01 -7.96493351e-01 7.03714609e-01 7.87067652e-01
4.94697660e-01 -1.94451615e-01 7.50265345e-02 2.74791300e-01
-5.61540239e-02 -3.16104800e-01 2.65058994e-01 8.29125226e-01
-4.06805485e-01 -6.26556933e-01 -3.31145257e-01 1.27512693e-01
-1.50295153e-01 -1.36384696e-01 -5.76420069e-01 1.06878221e+00
1.11717492e-01 7.74629831e-01 1.45061299e-01 2.54513212e-02
4.99521136e-01 1.72562554e-01 7.24989891e-01 -5.14188230e-01
-7.95366108e-01 -6.90828681e-01 -1.09133534e-01 -9.15580153e-01
-6.82019114e-01 -5.73743820e-01 -1.01685786e+00 -5.24719581e-02
1.58931151e-01 -1.95109934e-01 5.63232183e-01 1.01900339e+00
1.85506269e-01 7.27336764e-01 7.18020042e-03 -9.39736128e-01
-2.89352119e-01 -6.54828608e-01 -3.83136392e-01 8.28778923e-01
5.91912329e-01 -9.32777524e-01 -2.65393078e-01 -1.26094148e-01]
|
[10.672818183898926, 1.9343249797821045]
|
9779c813-0f7b-45bf-8a2c-0dae88278a46
|
doubly-reparameterized-importance-weighted
|
2206.11352
| null |
https://arxiv.org/abs/2206.11352v1
|
https://arxiv.org/pdf/2206.11352v1.pdf
|
Doubly Reparameterized Importance Weighted Structure Learning for Scene Graph Generation
|
As a structured prediction task, scene graph generation, given an input image, aims to explicitly model objects and their relationships by constructing a visually-grounded scene graph. In the current literature, such task is universally solved via a message passing neural network based mean field variational Bayesian methodology. The classical loose evidence lower bound is generally chosen as the variational inference objective, which could induce oversimplified variational approximation and thus underestimate the underlying complex posterior. In this paper, we propose a novel doubly reparameterized importance weighted structure learning method, which employs a tighter importance weighted lower bound as the variational inference objective. It is computed from multiple samples drawn from a reparameterizable Gumbel-Softmax sampler and the resulting constrained variational inference task is solved by a generic entropic mirror descent algorithm. The resulting doubly reparameterized gradient estimator reduces the variance of the corresponding derivatives with a beneficial impact on learning. The proposed method achieves the state-of-the-art performance on various popular scene graph generation benchmarks.
|
['Josef Kittler', 'Miroslaw Bober', 'Daqi Liu']
|
2022-06-22
| null | null | null | null |
['scene-graph-generation']
|
['computer-vision']
|
[ 5.44115305e-01 4.16920990e-01 2.08611730e-02 -3.02929968e-01
-9.83658016e-01 -8.78381953e-02 9.01180446e-01 6.05472177e-03
-4.76794809e-01 8.93055797e-01 1.48314029e-01 1.28911352e-02
-1.57301471e-01 -7.04775989e-01 -1.06126511e+00 -9.40068126e-01
4.29624975e-01 6.24801159e-01 4.66210842e-02 2.71505892e-01
3.91220570e-01 2.47331619e-01 -1.29829466e+00 -2.50874549e-01
9.51195955e-01 8.43168437e-01 5.67083001e-01 6.08589649e-01
-4.07335795e-02 9.21604216e-01 -1.90435529e-01 -6.54475629e-01
-1.39818219e-02 -3.28840077e-01 -6.87568188e-01 3.01638752e-01
6.43037379e-01 -3.22239071e-01 -2.73981720e-01 1.39340758e+00
9.50161070e-02 6.36068285e-01 1.20067477e+00 -1.09297955e+00
-6.62292302e-01 5.64869165e-01 -8.72935057e-01 -1.00219324e-01
-4.24636453e-02 -9.35635157e-03 1.36724555e+00 -8.80253494e-01
6.62887633e-01 1.36823523e+00 2.45899513e-01 3.96021724e-01
-1.85282302e+00 -1.93995193e-01 3.58643651e-01 3.59303415e-01
-1.52576780e+00 -1.14460297e-01 1.23452902e+00 -6.93822801e-01
5.45086265e-01 2.12788489e-02 6.37538552e-01 1.22669506e+00
-3.65615599e-02 9.51022923e-01 7.31688678e-01 -3.18805009e-01
5.35649598e-01 1.34308502e-01 -8.54741335e-02 1.00165045e+00
4.47300106e-01 -2.63879895e-01 -4.01731372e-01 -2.87862599e-01
7.17260122e-01 -1.92223433e-02 -2.90438145e-01 -6.71403527e-01
-1.02426720e+00 1.03319299e+00 4.65290785e-01 -3.96777272e-01
-4.25805420e-01 4.92406845e-01 1.46962509e-01 -5.02365172e-01
5.93511403e-01 -7.98929781e-02 -2.44135484e-02 2.92270184e-01
-1.08752811e+00 5.43425143e-01 6.55480564e-01 8.04002523e-01
9.71944690e-01 2.80184060e-01 -4.99004632e-01 7.86041737e-01
1.07727754e+00 5.41771710e-01 -8.68750587e-02 -1.06142318e+00
2.57087260e-01 3.08825314e-01 1.80095792e-01 -9.59114730e-01
1.33608624e-01 -2.94989467e-01 -9.98078525e-01 3.01415861e-01
4.29042667e-01 -1.55442625e-01 -8.54599655e-01 1.85941720e+00
6.45200312e-01 5.41342556e-01 -2.28456914e-01 8.97923946e-01
6.22119486e-01 8.93603742e-01 2.46833682e-01 -2.21138075e-01
1.13328505e+00 -8.70407999e-01 -5.32309353e-01 -1.61311164e-01
-1.03049278e-01 -7.07675159e-01 8.86343837e-01 3.38227361e-01
-1.01305342e+00 -3.93281400e-01 -1.06342566e+00 -1.92519277e-01
8.30306038e-02 8.20801109e-02 5.86151481e-01 4.06783015e-01
-7.11278796e-01 7.43947327e-01 -9.09533799e-01 5.87677844e-02
4.83430684e-01 -1.05480216e-01 -1.31522998e-01 5.29437996e-02
-8.24118197e-01 6.58141315e-01 6.75834358e-01 1.96732253e-01
-1.21696782e+00 -6.48165703e-01 -1.05901301e+00 3.24441977e-02
5.77886999e-01 -1.19741368e+00 1.06275082e+00 -4.50181782e-01
-1.79835093e+00 8.34212482e-01 -2.95271754e-01 -4.81308818e-01
7.24506915e-01 -9.33498293e-02 4.20803368e-01 -3.45030730e-03
-2.92959269e-02 6.27012134e-01 1.50915229e+00 -1.48404586e+00
-2.28877127e-01 -3.09527874e-01 -2.37282626e-02 3.12806666e-01
1.76217675e-01 -6.09346569e-01 -4.83580023e-01 -5.97024143e-01
4.02523205e-02 -8.82404864e-01 -3.40511322e-01 1.13547176e-01
-6.28842890e-01 -2.61814564e-01 3.88143569e-01 -5.76846182e-01
8.14763665e-01 -1.79293787e+00 5.37900567e-01 8.13365057e-02
3.22190523e-01 -2.16653228e-01 1.82263464e-01 2.92151928e-01
2.19490618e-01 -1.49988085e-01 -5.72588623e-01 -6.53455675e-01
2.34492093e-01 8.38833898e-02 -3.62195522e-01 7.18110859e-01
1.18855886e-01 7.87717164e-01 -1.03313625e+00 -7.12367833e-01
5.42971134e-01 7.42070615e-01 -7.73913801e-01 3.81796598e-01
-5.12175977e-01 2.96004385e-01 -4.46479410e-01 -1.47135630e-02
9.52044785e-01 -4.97130483e-01 1.64257169e-01 -3.20526719e-01
5.38211577e-02 -1.52570307e-01 -1.36528051e+00 1.89264929e+00
-2.34180883e-01 7.20356286e-01 2.63520032e-02 -1.18931508e+00
8.07673573e-01 -8.17894116e-02 2.86181718e-01 1.21447921e-01
5.60120232e-02 -2.36530676e-01 -5.52828014e-01 -4.07295227e-01
6.25767708e-01 -3.47418278e-01 2.95102358e-01 8.11745524e-02
8.02087560e-02 -5.24060905e-01 1.31043345e-02 4.56868291e-01
6.03822768e-01 7.14182019e-01 3.37623060e-01 -2.67111868e-01
5.73939145e-01 -3.03628474e-01 3.87882829e-01 1.02202654e+00
1.69860423e-01 7.14737594e-01 5.46237469e-01 1.07375078e-03
-1.14284670e+00 -1.53352368e+00 -2.15293467e-01 9.23338354e-01
2.48125941e-01 -1.48425236e-01 -8.33643198e-01 -4.13269877e-01
3.60239670e-02 1.09627450e+00 -4.58587497e-01 -1.32385835e-01
-9.15417597e-02 -8.42520595e-01 1.38920918e-01 5.12562096e-02
3.82459670e-01 -8.91179860e-01 -2.98619151e-01 1.92821696e-01
-1.75782785e-01 -1.00787234e+00 -5.56896150e-01 -2.95258194e-01
-6.62677288e-01 -8.42966199e-01 -1.09650218e+00 -4.19731855e-01
6.77047968e-01 8.03453773e-02 9.08285856e-01 -5.09231567e-01
-2.91298836e-01 3.07720929e-01 7.76385292e-02 -1.50219679e-01
-2.62698114e-01 -2.44990647e-01 -1.21144161e-01 5.91368377e-01
-1.43529698e-02 -4.66053009e-01 -6.12474680e-01 -3.09137911e-01
-8.37363601e-01 4.70757067e-01 4.27183837e-01 1.01398563e+00
8.89312267e-01 -2.60477304e-01 9.66049582e-02 -8.46623421e-01
5.87034643e-01 -5.60022712e-01 -1.10210788e+00 2.64648855e-01
-5.32282114e-01 7.53299713e-01 3.69779855e-01 -3.18669349e-01
-1.63066638e+00 8.12921524e-02 -3.60757695e-03 -6.34638011e-01
-5.13949506e-02 4.87331420e-01 1.33696618e-02 -3.48981731e-02
3.64797145e-01 4.84538645e-01 -2.21929133e-01 -3.90456885e-01
6.92362905e-01 1.93406537e-01 5.16105175e-01 -9.13135886e-01
7.55460680e-01 7.08702385e-01 5.13595581e-01 -1.09366357e+00
-1.02859271e+00 -3.37648511e-01 -3.04730952e-01 -4.48225141e-01
1.06522918e+00 -8.15474570e-01 -1.04765022e+00 3.81304562e-01
-1.40029967e+00 -3.23925912e-01 -2.25374788e-01 6.09295964e-01
-7.66974628e-01 8.91327202e-01 -3.90615910e-01 -1.24619329e+00
-2.15472028e-01 -1.06258643e+00 1.36247826e+00 3.05300444e-01
4.79394719e-02 -1.03138697e+00 3.75725865e-01 4.38523114e-01
-8.69235024e-02 6.39854729e-01 8.56752455e-01 -3.46340984e-02
-9.04082060e-01 6.78585991e-02 -4.91005957e-01 2.92148948e-01
-2.87770301e-01 3.60757500e-01 -1.08323824e+00 -1.17736951e-01
-5.09474315e-02 -2.93238997e-01 1.18372512e+00 8.77465785e-01
1.09865057e+00 -2.39869118e-01 -8.12465698e-02 6.76740468e-01
1.54516482e+00 -2.47021690e-01 4.02104318e-01 -2.55826622e-01
9.85428929e-01 5.96997797e-01 2.15077788e-01 7.23217845e-01
4.22539055e-01 5.00974596e-01 7.04919279e-01 4.78159815e-01
6.58633262e-02 -6.96769774e-01 1.85404837e-01 6.03217602e-01
-1.50415227e-01 -3.25354546e-01 -6.36469126e-01 5.62168479e-01
-2.11463475e+00 -1.01566732e+00 -2.09847808e-01 2.35159111e+00
7.97868252e-01 2.13364512e-01 -2.69295335e-01 -3.57904166e-01
8.80083203e-01 4.72814173e-01 -7.44665980e-01 4.59011197e-02
1.77399993e-01 4.89164479e-02 4.14660722e-01 9.59483206e-01
-8.43776703e-01 1.02292919e+00 5.34626818e+00 1.22182417e+00
-6.55204475e-01 1.19782738e-01 4.37014490e-01 -2.08621547e-02
-4.20574933e-01 2.55339235e-01 -7.36484766e-01 3.94139677e-01
4.59586322e-01 -3.37778658e-01 5.54061949e-01 1.08746970e+00
3.92672896e-01 -3.10069472e-01 -1.04830039e+00 1.27915466e+00
5.77127896e-02 -1.62129450e+00 2.50705004e-01 2.04949498e-01
9.81883347e-01 2.79428605e-02 7.14194553e-04 1.24068581e-01
4.69755143e-01 -7.19548225e-01 9.68612850e-01 9.62898076e-01
4.97144431e-01 -6.35983944e-01 2.69440532e-01 3.56193006e-01
-1.07657266e+00 3.49618286e-01 -5.47931910e-01 -9.46388859e-03
5.96897602e-01 8.03745747e-01 -6.18481815e-01 3.20713133e-01
2.99150169e-01 5.60031354e-01 -1.13014475e-01 1.00691855e+00
-4.53549355e-01 5.96705377e-01 -3.93414617e-01 -3.36105019e-01
2.46370628e-01 -8.00770938e-01 9.97982383e-01 1.04228640e+00
7.96864703e-02 -2.20644459e-01 5.80484308e-02 1.44561982e+00
-3.39653760e-01 3.89201231e-02 -5.44379294e-01 3.87903191e-02
1.97495610e-01 1.37142706e+00 -6.97711647e-01 -4.77517456e-01
-2.03023717e-01 1.02335274e+00 6.50916994e-01 5.77646971e-01
-9.25420284e-01 -8.75854492e-02 5.18812478e-01 -9.95384455e-02
4.08187479e-01 -1.44477487e-01 -4.74070646e-02 -1.49561739e+00
-1.45587876e-01 -3.26893151e-01 1.41621828e-01 -1.00596929e+00
-1.35948288e+00 1.41457424e-01 3.66157413e-01 -7.40709960e-01
-3.43271583e-01 -6.72645390e-01 -7.00563610e-01 9.76907015e-01
-1.44560194e+00 -9.94329870e-01 -3.66823733e-01 5.61020255e-01
6.16719067e-01 1.36959285e-01 4.09734458e-01 -6.37666602e-03
-7.19714224e-01 1.82623789e-01 3.05093795e-01 -1.91247150e-01
4.00486410e-01 -1.53719926e+00 1.88103795e-01 8.98779333e-01
2.04951271e-01 4.64174718e-01 1.03789473e+00 -6.83626890e-01
-1.27101433e+00 -1.12456036e+00 4.15179521e-01 -2.08976686e-01
7.74426699e-01 -4.03194577e-01 -8.45694005e-01 5.87985158e-01
2.70920813e-01 -1.50006801e-01 1.91838861e-01 -1.35170624e-01
-2.32245907e-01 3.78656052e-02 -1.02502263e+00 9.25297916e-01
8.13125074e-01 -6.47706032e-01 -4.66057718e-01 5.67971945e-01
7.12833166e-01 -1.42820165e-01 -6.93700075e-01 7.02454522e-02
5.21915913e-01 -7.71660924e-01 9.29667890e-01 -5.75378180e-01
6.11610711e-01 -2.65765518e-01 -2.46462986e-01 -1.18880010e+00
-3.55018288e-01 -9.47712898e-01 -5.14116406e-01 9.50351179e-01
2.48300195e-01 -3.76414865e-01 9.58198547e-01 5.95824957e-01
2.25357845e-01 -6.57341301e-01 -7.96564758e-01 -4.75871056e-01
-2.45972842e-01 -5.80122411e-01 7.00563267e-02 5.65169752e-01
-3.43960494e-01 6.84912860e-01 -7.21542060e-01 1.21978454e-01
1.72052634e+00 -6.45235851e-02 7.73490429e-01 -1.20127964e+00
-6.91151798e-01 -4.49767590e-01 -2.14744270e-01 -1.42880309e+00
4.80756193e-01 -9.50947464e-01 2.69317091e-01 -1.48936617e+00
4.61222112e-01 3.40175301e-01 -2.84923576e-02 -2.53598332e-01
-4.91564363e-01 -5.26954606e-02 1.79463804e-01 3.84741020e-03
-4.90298778e-01 1.06533015e+00 1.25379610e+00 -3.42939258e-01
1.06283380e-02 9.15796459e-02 -4.39310104e-01 1.00032771e+00
3.74469370e-01 -6.84384048e-01 -7.35212564e-01 -2.53470123e-01
4.19614494e-01 1.43942624e-01 7.52340138e-01 -5.33966541e-01
2.26618826e-01 -4.16545749e-01 1.13685161e-01 -6.83376968e-01
6.23709440e-01 -4.10863250e-01 1.99252442e-01 2.49084592e-01
-4.57457364e-01 -6.53190255e-01 -1.99806631e-01 1.19255030e+00
7.67153129e-02 -7.08965361e-01 1.01373601e+00 -1.91491887e-01
-5.03993154e-01 4.85317618e-01 -2.45353714e-01 2.66079724e-01
7.66816378e-01 -8.75312239e-02 1.91554606e-01 -4.42168564e-01
-7.61525631e-01 -6.04419224e-03 1.89336270e-01 -9.59144011e-02
7.81582057e-01 -1.22777843e+00 -9.26355124e-01 -1.53022274e-01
2.45782025e-02 2.00270519e-01 4.98421192e-01 6.28779531e-01
-4.87365335e-01 5.21102734e-02 2.30208933e-01 -7.70631552e-01
-9.53017056e-01 3.89987767e-01 2.74609178e-01 -3.91087472e-01
-5.27573943e-01 1.04300416e+00 4.58586127e-01 -2.39026085e-01
1.46745116e-01 -1.51584670e-01 -9.87494588e-02 5.59104756e-02
2.73031116e-01 6.07068539e-01 -4.84786719e-01 -6.09960079e-01
-3.98989059e-02 4.46092188e-01 -5.22558130e-02 -4.16969031e-01
1.12152219e+00 -2.23321304e-01 -1.12686064e-02 5.46443999e-01
1.28801024e+00 -2.36231267e-01 -1.89283156e+00 -3.80671084e-01
-1.93465039e-01 -3.71485919e-01 4.74592090e-01 3.90320569e-02
-7.74663448e-01 9.07796502e-01 3.49565536e-01 2.66621977e-01
4.70502943e-01 -5.29375486e-03 4.71872747e-01 5.26700437e-01
1.81601822e-01 -1.09371519e+00 7.82232452e-03 1.73049316e-01
9.01416719e-01 -1.44204867e+00 3.19136441e-01 -2.75631070e-01
-5.62184155e-01 9.39394474e-01 2.87277043e-01 -5.35722673e-01
7.48631775e-01 -2.64893711e-01 -7.36285806e-01 -1.88068241e-01
-5.77370584e-01 -1.46903172e-01 5.32268167e-01 4.19679016e-01
7.38912448e-02 1.59727827e-01 -2.91452229e-01 1.51151448e-01
7.34087080e-02 -1.06112629e-01 3.48916531e-01 3.80594075e-01
-4.32798266e-01 -5.26701629e-01 -1.76090121e-01 3.81739438e-01
-3.83436233e-01 -3.60358000e-01 -1.56171126e-02 4.67288375e-01
-2.82554299e-01 7.53070831e-01 -1.70860216e-01 2.81865537e-01
2.02006493e-02 -1.48556530e-01 7.43984520e-01 -4.73726660e-01
2.47254014e-01 1.83063045e-01 -2.75793552e-01 -3.81440371e-01
-3.82596642e-01 -7.27942586e-01 -9.73128378e-01 -2.29911521e-01
-2.95057535e-01 -3.39949839e-02 6.78711593e-01 1.02790105e+00
8.43589082e-02 4.87158597e-01 3.05028886e-01 -1.29569733e+00
-7.80865371e-01 -1.02654302e+00 -6.89946771e-01 5.23266077e-01
2.60664880e-01 -7.77597010e-01 -5.42867899e-01 3.54377866e-01]
|
[7.173470973968506, 3.690725803375244]
|
28130a68-cc7d-46de-ba53-e5a6d8ffc6c1
|
quantum-circuit-components-for-cognitive
|
2302.03012
| null |
https://arxiv.org/abs/2302.03012v3
|
https://arxiv.org/pdf/2302.03012v3.pdf
|
Quantum Circuit Components for Cognitive Decision-Making
|
This paper demonstrates that some non-classical models of human decision-making can be run successfully as circuits on quantum computers. Since the 1960s, many observed cognitive behaviors have been shown to violate rules based on classical probability and set theory. For example, the order in which questions are posed in a survey affects whether participants answer 'yes' or 'no', so the population that answers 'yes' to both questions cannot be modeled as the intersection of two fixed sets. It can, however, be modeled as a sequence of projections carried out in different orders. This and other examples have been described successfully using quantum probability, which relies on comparing angles between subspaces rather than volumes between subsets. Now in the early 2020s, quantum computers have reached the point where some of these quantum cognitive models can be implemented and investigated on quantum hardware, by representing the mental states in qubit registers, and the cognitive operations and decisions using different gates and measurements. This paper develops such quantum circuit representations for quantum cognitive models, focusing particularly on modeling order effects and decision-making under uncertainty. The claim is not that the human brain uses qubits and quantum circuits explicitly (just like the use of Boolean set theory does not require the brain to be using classical bits), but that the mathematics shared between quantum cognition and quantum computing motivates the exploration of quantum computers for cognition modeling. Key quantum properties include superposition, entanglement, and collapse, as these mathematical elements provide a common language between cognitive models, quantum hardware, and circuit implementations.
|
['Emmanuel Pothos', 'Jyoti Rani', 'Dominic Widdows']
|
2023-02-06
| null | null | null | null |
['decision-making-under-uncertainty', 'decision-making-under-uncertainty']
|
['medical', 'reasoning']
|
[ 3.11496884e-01 1.83359385e-01 4.37779188e-01 -2.90517777e-01
-7.32749999e-02 -8.56503010e-01 6.85116529e-01 1.90170974e-01
-5.62094688e-01 5.35315275e-01 8.29910859e-02 -7.75191426e-01
-4.26623762e-01 -1.47116828e+00 -2.87705541e-01 -4.83643591e-01
1.21637680e-01 6.08995199e-01 -2.16995589e-02 -6.80164933e-01
5.92726946e-01 1.02016911e-01 -1.66139996e+00 2.21891597e-01
5.94657719e-01 3.86058658e-01 -2.44973063e-01 7.51670599e-01
1.69398934e-02 7.90980637e-01 -3.40592742e-01 -4.95467633e-01
2.14197263e-01 -7.56057322e-01 -9.17891920e-01 -4.52684522e-01
-3.05490773e-02 -2.99712986e-01 -7.98216283e-01 1.52038026e+00
1.25131741e-01 2.34720618e-01 6.47829056e-01 -8.54033530e-01
-1.08903813e+00 6.55598581e-01 3.23231727e-01 4.88898382e-02
7.39867866e-01 2.18308851e-01 1.14584160e+00 -1.18190259e-01
3.45938981e-01 1.41536665e+00 3.98970753e-01 4.08637583e-01
-1.70713663e+00 -6.08244658e-01 -6.34435117e-01 5.10368705e-01
-1.79338109e+00 -2.57249296e-01 9.02932435e-02 -4.89972711e-01
1.10344696e+00 2.45556921e-01 1.03152776e+00 4.55634028e-01
1.02242434e+00 -7.95855746e-02 1.69876754e+00 -6.18827820e-01
7.56134927e-01 4.32269610e-02 7.40117550e-01 6.22119486e-01
6.35265291e-01 8.81528974e-01 -6.54169083e-01 -1.47494555e-01
6.31381929e-01 -1.25310317e-01 -1.41390488e-01 1.22120850e-01
-1.12806296e+00 7.03156352e-01 2.05547169e-01 4.47120488e-01
-3.16234678e-01 3.85962784e-01 -5.90603016e-02 6.16663575e-01
-5.88905871e-01 6.14875734e-01 2.85748485e-02 -8.54164660e-02
-4.55673277e-01 5.20498991e-01 1.00793600e+00 6.48166537e-01
1.00170934e+00 -7.01852441e-01 -4.98517118e-02 -3.92153740e-01
4.51900780e-01 8.89215648e-01 2.43467048e-01 -1.13186753e+00
-2.49510869e-01 4.03756171e-01 1.20693900e-01 -8.69577348e-01
-5.68447530e-01 -3.86997089e-02 -4.84745741e-01 2.43727028e-01
5.11204362e-01 4.59951907e-02 -7.53882468e-01 1.85254264e+00
-4.24794823e-01 -4.09526601e-02 2.48540953e-01 7.87177742e-01
3.09293091e-01 5.76107204e-01 -7.15987850e-03 -5.41159697e-02
1.75827384e+00 3.28782171e-01 -9.15223777e-01 2.51063220e-02
1.03742003e+00 -6.16458416e-01 6.30550563e-01 5.61384976e-01
-1.00286376e+00 -3.77929091e-01 -1.25434172e+00 -1.45496175e-01
-4.32749987e-01 -8.33852410e-01 1.23734522e+00 1.61548185e+00
-1.35710537e+00 6.91290677e-01 -8.25773716e-01 -2.04092056e-01
8.43334049e-02 4.80916232e-01 -1.21222191e-01 -3.11364472e-01
-1.70918322e+00 1.29441345e+00 5.19552410e-01 2.01684330e-02
-5.21616518e-01 2.61239652e-02 -4.78244692e-01 3.33991528e-01
1.48014188e-01 -9.93836641e-01 9.87084925e-01 -4.20494705e-01
-1.56833243e+00 9.18589771e-01 -1.37865007e-01 -3.26680839e-01
-2.70555586e-01 4.48376238e-01 -5.33418119e-01 -1.88019499e-01
-1.67889982e-01 4.48129892e-01 5.47468029e-02 -4.88499820e-01
-2.10196003e-01 -6.51549160e-01 6.82597339e-01 1.73497692e-01
3.04802060e-01 -4.29083556e-02 3.19874972e-01 5.18533111e-01
8.95154715e-01 -1.32866991e+00 -1.03636175e-01 -6.04975462e-01
-2.12236315e-01 -1.79824114e-01 -3.11415315e-01 7.35561848e-02
1.14281058e+00 -1.99696958e+00 -6.24756142e-02 6.29139304e-01
2.87294745e-01 -3.69135380e-01 4.91383933e-02 4.93537128e-01
3.36221829e-02 2.98110783e-01 -1.74180195e-01 3.45145226e-01
7.32300580e-01 3.49243164e-01 -2.53654718e-01 6.68678939e-01
-2.73165554e-02 9.65150714e-01 -8.11919987e-01 -1.40777454e-01
-2.60976842e-03 2.53128141e-01 -1.00870955e+00 -4.70874846e-01
3.83469075e-01 1.54161349e-01 -3.25560570e-01 2.33739167e-01
6.05450630e-01 -1.52297646e-01 5.49032152e-01 2.82663792e-01
-3.80868137e-01 9.08333540e-01 -1.48255503e+00 1.47046924e+00
5.06934002e-02 8.21764290e-01 -4.36439753e-01 -6.01283133e-01
4.53184634e-01 4.69757140e-01 -3.02985460e-01 -1.10617840e+00
4.91864443e-01 2.14159667e-01 1.10578334e+00 -1.65406287e-01
6.05876148e-01 -7.61990190e-01 -4.28890705e-01 1.00868583e+00
7.10947514e-02 -9.20017540e-01 4.25751030e-01 3.62194419e-01
1.24106956e+00 -1.98795557e-01 1.94994628e-01 -5.96265256e-01
3.49406153e-01 -4.72295880e-02 3.71791333e-01 1.00425434e+00
-4.48164523e-01 1.44587502e-01 4.70592350e-01 -2.65143782e-01
-8.66540968e-01 -1.57408464e+00 -6.66655004e-01 6.63420141e-01
4.79848176e-01 -5.26659906e-01 -8.24448287e-01 7.25239575e-01
-2.48289913e-01 1.34921002e+00 -6.34600878e-01 -6.32531703e-01
6.56557679e-02 -9.37542200e-01 6.63604438e-01 -4.00526933e-02
3.23709577e-01 -7.27772951e-01 -9.36734259e-01 1.61294311e-01
-6.37061670e-02 -6.89558983e-01 1.21634282e-01 5.40532351e-01
-7.92021990e-01 -9.94019747e-01 4.47594583e-01 -1.33219853e-01
4.97756213e-01 2.39436179e-01 8.98128986e-01 1.79251641e-01
-2.27530152e-01 5.78539252e-01 -2.33301342e-01 -5.16632974e-01
-6.07082844e-01 -5.08281469e-01 4.46400404e-01 -3.34813088e-01
1.07383287e+00 -6.10010028e-01 -4.50214446e-01 -1.47429824e-01
-1.12746060e+00 -7.16132894e-02 5.23109734e-01 5.87973177e-01
2.18877792e-01 3.31778049e-01 -6.37976751e-02 -5.63356280e-01
8.21660399e-01 -2.12550148e-01 -4.52637523e-01 2.81491220e-01
-3.47542644e-01 4.39486504e-01 1.30311877e-01 1.61177471e-01
-8.59682202e-01 -5.49777329e-01 3.74896377e-01 6.10671937e-01
-1.87628746e-01 6.95116699e-01 6.34863675e-02 -1.27465129e-01
8.85869920e-01 3.18423569e-01 -9.53470767e-02 4.66539651e-01
5.19375265e-01 4.89543855e-01 1.56260669e-01 -7.37110078e-01
7.96030760e-01 6.29610062e-01 4.66701001e-01 -6.69520915e-01
-4.22093779e-01 4.93121520e-02 -8.60882938e-01 -1.44298002e-02
1.19699132e+00 -6.85675621e-01 -1.47958016e+00 4.57450598e-02
-1.06173742e+00 1.71226025e-01 -3.13012153e-01 9.59419489e-01
-7.22536504e-01 2.30123639e-01 -6.87839925e-01 -1.25486183e+00
3.18273574e-01 -1.15665352e+00 3.96545589e-01 3.75882000e-01
-4.53669697e-01 -8.81543756e-01 1.83130592e-01 2.43411928e-01
5.04350305e-01 -3.25238526e-01 1.18295681e+00 -4.00945604e-01
-1.12463272e+00 -4.54457402e-01 3.66013162e-02 -1.47903413e-02
-1.59265026e-01 -1.58249855e-01 -9.59113896e-01 -2.09951028e-01
3.93133134e-01 -3.58331323e-01 5.50554752e-01 8.87928829e-02
2.89615601e-01 4.15900052e-01 -6.61711991e-02 9.22536850e-02
1.39075208e+00 5.87245286e-01 1.24581683e+00 9.30018425e-02
7.47107714e-02 5.80454886e-01 -1.63089544e-01 6.73515648e-02
7.02730179e-01 1.20477416e-01 -6.05697650e-03 8.31599295e-01
3.51415217e-01 9.78514254e-02 4.67834771e-01 1.20732367e+00
-3.42088759e-01 1.95920661e-01 -1.01080310e+00 -3.96954641e-02
-1.23550320e+00 -1.61384785e+00 -3.02223027e-01 2.58462644e+00
6.62428081e-01 4.55771714e-01 -5.56421638e-01 2.49604762e-01
5.10580003e-01 -4.64225024e-01 -1.16186894e-01 -7.89558232e-01
-3.16713542e-01 9.03205693e-01 4.50794339e-01 6.34443700e-01
-5.61297953e-01 7.53195584e-01 7.17152214e+00 2.16040537e-01
-5.93241572e-01 1.58881292e-01 2.22708374e-01 2.88259555e-02
-6.84878945e-01 8.58505487e-01 -3.90383929e-01 2.92598903e-01
1.45643878e+00 -5.51621079e-01 9.80937004e-01 -1.19598813e-01
-1.58796832e-01 -7.03609347e-01 -1.48431575e+00 9.62297082e-01
-2.49659926e-01 -1.03005028e+00 2.32687965e-02 3.92965019e-01
7.32273757e-01 -2.67765969e-01 1.46741077e-01 5.04032969e-01
6.35017157e-01 -1.40283620e+00 9.53592002e-01 8.53589475e-01
4.06302482e-01 -5.41761696e-01 7.74512410e-01 4.16056305e-01
-7.55631626e-01 -2.13964403e-01 -6.14437699e-01 -1.07887030e+00
5.40503338e-02 1.75359651e-01 -3.20374936e-01 3.60699803e-01
1.53387263e-01 -2.54056633e-01 -3.12112987e-01 6.56823158e-01
-2.07209691e-01 4.61719334e-01 -4.44333881e-01 -4.19020087e-01
1.50983050e-01 -8.59969616e-01 1.84283018e-01 5.96230388e-01
4.84395236e-01 1.05384529e+00 -3.53806436e-01 1.46618211e+00
5.02151787e-01 -2.43273422e-01 -4.64486599e-01 -4.28147167e-01
6.73323929e-01 6.37545228e-01 -9.33265626e-01 -3.65315139e-01
-4.22963500e-01 6.16546750e-01 -2.82588571e-01 7.59987980e-02
-7.03719616e-01 -4.86087412e-01 6.63007140e-01 -1.22104429e-01
-2.84109324e-01 -5.60248256e-01 -5.78979015e-01 -1.24335289e+00
-5.87982953e-01 -6.67995870e-01 -2.13359147e-01 -7.56807625e-01
-8.84168744e-01 8.21317062e-02 -2.08855078e-01 -6.46209240e-01
2.40516663e-02 -8.85209799e-01 -4.74955738e-01 1.25050557e+00
-7.28816092e-01 -2.96101332e-01 2.68900186e-01 4.65990841e-01
-6.80149317e-01 2.36388981e-01 1.44338512e+00 3.41644324e-02
-2.99606770e-01 1.53751805e-01 1.80069342e-01 8.68034363e-03
2.54123151e-01 -1.21936476e+00 3.53789032e-01 7.17561960e-01
3.45235139e-01 1.48901772e+00 8.85589004e-01 -3.68060440e-01
-1.82530975e+00 -1.23177692e-01 1.01595998e+00 -9.23390627e-01
6.42150819e-01 -4.07056272e-01 -7.35334337e-01 6.42461240e-01
6.37615398e-02 -6.53350413e-01 1.13733280e+00 4.81113136e-01
-4.19157833e-01 2.76645482e-01 -8.45118284e-01 7.81280935e-01
1.05674732e+00 -1.46293771e+00 -1.04983711e+00 2.84920603e-01
2.80326039e-01 -1.00784458e-01 -5.87319076e-01 -1.67280063e-01
7.76465058e-01 -1.37067807e+00 5.85971236e-01 -6.15346611e-01
-7.09661181e-05 -5.94484627e-01 -6.33428872e-01 -1.06985307e+00
-7.14639783e-01 -4.86997545e-01 6.08593285e-01 2.93106884e-01
2.21563786e-01 -1.04050040e+00 4.39119041e-01 1.18430734e+00
7.61158094e-02 1.10162266e-01 -1.25778222e+00 -5.74249148e-01
4.82081145e-01 -7.53822923e-01 7.61015058e-01 9.21163857e-01
7.11079597e-01 3.98606718e-01 4.82399672e-01 1.77287966e-01
7.95579910e-01 -1.11031914e-02 2.11911976e-01 -1.56435537e+00
-3.86152476e-01 -6.79472089e-01 -1.05288363e+00 -7.50479341e-01
-1.17033690e-01 -1.30786526e+00 1.84155069e-02 -1.29921544e+00
4.33568567e-01 -2.55820006e-01 -4.40852612e-01 -9.13969874e-02
-4.31334786e-03 9.76308659e-02 3.86948347e-01 2.69924819e-01
-3.34218770e-01 2.11653575e-01 1.07866240e+00 -5.88710904e-02
5.13561442e-02 -4.59518731e-01 -8.81136775e-01 5.99801660e-01
5.33865929e-01 -4.97250795e-01 -2.56900251e-01 -1.68043792e-01
1.21978855e+00 3.01760465e-01 4.74309891e-01 -1.37464499e+00
6.11892223e-01 -6.15038984e-02 1.71812877e-01 -1.67730987e-01
4.27378148e-01 -4.33140606e-01 4.69792128e-01 1.03474820e+00
-2.13130757e-01 4.80063409e-02 2.79551726e-02 4.25328314e-01
2.11420447e-01 -2.62627125e-01 5.19968688e-01 -3.40531737e-01
-5.77646792e-01 -3.13438892e-01 -9.99696255e-01 -2.96403885e-01
8.59101832e-01 -4.83520091e-01 -6.10762060e-01 -2.26710007e-01
-9.10126030e-01 -9.97840613e-02 6.06121719e-01 -2.25876510e-01
3.41207623e-01 -1.02572381e+00 -5.30179799e-01 1.04421757e-01
-1.14196889e-01 -6.13614559e-01 5.66098392e-01 7.41958082e-01
-8.05374265e-01 1.13387311e+00 -6.21665835e-01 -3.14192832e-01
-4.38937753e-01 3.86828482e-01 4.87417281e-01 1.78457156e-01
2.21308097e-02 8.60401034e-01 4.00875568e-01 -4.67446804e-01
-5.47278702e-01 -4.62634385e-01 1.60129070e-01 -1.44868895e-01
5.76633632e-01 2.20092252e-01 -2.29042828e-01 -5.67950130e-01
-5.30766964e-01 3.50280941e-01 1.87307507e-01 -6.05286241e-01
7.04056740e-01 -1.74981579e-01 -7.35108435e-01 8.44601989e-01
6.96729124e-01 -1.96513668e-01 -3.82289261e-01 -1.22507803e-01
-1.83887526e-01 -1.47919580e-01 7.89067000e-02 -7.43107975e-01
5.44535816e-02 1.07853782e+00 7.71989465e-01 5.38775146e-01
7.16458976e-01 -1.84048072e-01 3.30104947e-01 1.02008736e+00
1.09150386e+00 -1.17424059e+00 -3.35573196e-01 9.04954374e-01
1.96007952e-01 -7.63762295e-01 9.87150297e-02 -1.53192848e-01
1.60224307e-02 1.26475847e+00 4.09974530e-02 -2.82859027e-01
6.80406153e-01 -4.16636206e-02 -2.86052078e-01 -4.85474050e-01
-8.49485517e-01 -4.57563937e-01 -1.35870308e-01 2.95100033e-01
7.71066904e-01 9.25730109e-01 -5.54266453e-01 4.69411105e-01
-8.56673837e-01 1.00339368e-01 1.24399662e+00 9.88829374e-01
-8.51010144e-01 -1.14139616e+00 -9.41867948e-01 2.93574154e-01
-9.16524157e-02 -3.58782917e-01 -2.19849586e-01 4.05717403e-01
4.85704482e-01 1.25763702e+00 5.31277537e-01 -6.65419996e-01
7.73831978e-02 4.72895861e-01 1.13525796e+00 -8.41150641e-01
-3.16205680e-01 -6.70934916e-01 -2.82058716e-01 -4.67709690e-01
-2.25182936e-01 -8.67258608e-01 -1.55864811e+00 -1.10332894e+00
-4.83954251e-01 3.82355332e-01 5.88073909e-01 1.16094327e+00
1.25820950e-01 6.81245327e-01 3.54150729e-03 -4.42442328e-01
-9.67783630e-01 -6.71213865e-01 -1.08982086e+00 1.75725728e-01
-1.12930901e-01 -6.51374578e-01 -3.64505857e-01 -2.01052353e-01]
|
[5.653796195983887, 4.940537929534912]
|
a636a643-9e64-4a82-b48d-2f5aa049c65b
|
talking-face-generation-by-adversarially
|
1807.0786
| null |
http://arxiv.org/abs/1807.07860v2
|
http://arxiv.org/pdf/1807.07860v2.pdf
|
Talking Face Generation by Adversarially Disentangled Audio-Visual Representation
|
Talking face generation aims to synthesize a sequence of face images that
correspond to a clip of speech. This is a challenging task because face
appearance variation and semantics of speech are coupled together in the subtle
movements of the talking face regions. Existing works either construct specific
face appearance model on specific subjects or model the transformation between
lip motion and speech. In this work, we integrate both aspects and enable
arbitrary-subject talking face generation by learning disentangled audio-visual
representation. We find that the talking face sequence is actually a
composition of both subject-related information and speech-related information.
These two spaces are then explicitly disentangled through a novel
associative-and-adversarial training process. This disentangled representation
has an advantage where both audio and video can serve as inputs for generation.
Extensive experiments show that the proposed approach generates realistic
talking face sequences on arbitrary subjects with much clearer lip motion
patterns than previous work. We also demonstrate the learned audio-visual
representation is extremely useful for the tasks of automatic lip reading and
audio-video retrieval.
|
['Yu Liu', 'Ping Luo', 'Hang Zhou', 'Ziwei Liu', 'Xiaogang Wang']
|
2018-07-20
| null | null | null | null |
['talking-face-generation']
|
['computer-vision']
|
[ 3.75034779e-01 2.92833060e-01 -1.39355510e-01 -4.12203342e-01
-8.61510217e-01 -6.22171521e-01 8.72114897e-01 -1.04787004e+00
2.95719981e-01 7.39931822e-01 6.61417127e-01 3.19702655e-01
2.66347468e-01 -3.15435559e-01 -7.31038690e-01 -1.08605874e+00
1.03251569e-01 3.28358203e-01 -2.44469002e-01 -2.04655170e-01
-6.38821051e-02 4.67219472e-01 -2.02099252e+00 6.84624135e-01
3.36607337e-01 8.02775025e-01 1.32363349e-01 9.54428554e-01
-4.27602296e-04 6.12275481e-01 -6.43998802e-01 -4.39256608e-01
4.02917504e-01 -7.24830747e-01 -4.64121372e-01 2.06428826e-01
7.75864065e-01 -3.45676452e-01 -5.14152765e-01 9.71640825e-01
7.76389480e-01 -2.12509502e-02 7.93775976e-01 -1.60248697e+00
-8.30026925e-01 4.93321121e-01 -6.96166992e-01 -1.21579207e-01
6.57549024e-01 3.44906807e-01 6.52166069e-01 -9.68314409e-01
6.19782507e-01 1.67572331e+00 2.06738070e-01 1.19399667e+00
-1.28228080e+00 -1.06781387e+00 1.98241565e-02 3.41733605e-01
-1.39605248e+00 -1.14269400e+00 1.17355740e+00 -5.49832582e-01
1.97284251e-01 5.52174985e-01 6.24685884e-01 1.66993761e+00
-6.77007735e-02 8.15995574e-01 1.00309741e+00 -3.09205502e-01
-8.69061947e-02 3.91102850e-01 -5.44246137e-01 5.52135766e-01
-1.53294310e-01 3.85423601e-01 -8.91305923e-01 -1.20396249e-01
6.28231406e-01 -1.27455205e-01 -6.42561793e-01 -5.74520409e-01
-1.27271771e+00 6.55148804e-01 6.58243746e-02 8.64474773e-02
-1.01834111e-01 1.71990648e-01 7.16827810e-02 2.18692958e-01
2.74594098e-01 7.32989460e-02 8.15868676e-02 2.83883750e-01
-1.06276321e+00 2.91811109e-01 7.91813016e-01 8.14694166e-01
4.11700308e-01 5.04874170e-01 -2.58013129e-01 7.80679822e-01
4.53716546e-01 9.68003869e-01 6.40527606e-01 -1.03162920e+00
3.12363833e-01 -8.14041272e-02 -6.08170852e-02 -8.52027655e-01
2.74059474e-01 1.86254948e-01 -7.94643998e-01 4.88932878e-01
4.30234134e-01 -1.32725924e-01 -8.54916573e-01 2.38269281e+00
3.37428540e-01 6.13364697e-01 2.22954810e-01 9.23982203e-01
9.59421575e-01 6.35040820e-01 -4.92648743e-02 -3.59732985e-01
1.45849526e+00 -7.91937113e-01 -1.27130866e+00 -7.79944956e-02
-4.41120625e-01 -9.24300492e-01 8.36285830e-01 6.32890984e-02
-1.39703989e+00 -5.53486466e-01 -1.05228996e+00 -2.83534024e-02
-1.85374081e-01 2.17501789e-01 2.56192952e-01 5.81226468e-01
-1.19899607e+00 2.07051352e-01 -2.98337787e-01 -8.20315704e-02
3.73526514e-01 4.40732449e-01 -6.23794138e-01 1.13623694e-01
-1.31248033e+00 6.19909227e-01 -2.21252203e-01 9.38192531e-02
-1.10683596e+00 -6.73091888e-01 -1.07215166e+00 -1.05969161e-02
2.20234469e-01 -1.03527212e+00 1.33875799e+00 -1.29267585e+00
-1.84775984e+00 1.08445036e+00 -3.55727851e-01 -3.38532664e-02
4.76135880e-01 6.77810460e-02 -5.58260500e-01 4.56260413e-01
-1.86682045e-01 8.54165137e-01 1.76505995e+00 -1.58516622e+00
9.79925506e-03 -6.14025712e-01 -3.34608078e-01 4.33279842e-01
-1.17907047e-01 -2.77816341e-03 -2.39534318e-01 -9.42158937e-01
-2.14769900e-01 -9.58478272e-01 3.44650865e-01 4.29177731e-01
-2.93340087e-01 2.07418546e-01 1.27496719e+00 -8.02126229e-01
6.54758513e-01 -2.18656492e+00 3.85531545e-01 -1.90825775e-01
2.90861130e-01 2.21995190e-01 -4.73075926e-01 2.80196846e-01
-5.53174257e-01 1.00019746e-01 1.33328646e-01 -6.34757102e-01
-9.30536538e-02 -9.01680738e-02 -6.30201340e-01 5.19006729e-01
2.42238417e-01 9.30290222e-01 -8.13627839e-01 -5.91755509e-01
2.77859941e-02 9.22872961e-01 -5.36240101e-01 3.79191756e-01
-1.37897938e-01 7.17328191e-01 -2.69306809e-01 5.50774992e-01
6.31576300e-01 2.45584592e-01 1.41701877e-01 -3.98768723e-01
5.03971815e-01 -1.39235072e-02 -7.92588890e-01 1.88216507e+00
-4.97685134e-01 9.45055604e-01 5.91467798e-01 -7.33066976e-01
7.59312510e-01 8.13514113e-01 3.04584414e-01 -3.46528411e-01
1.33696213e-01 -1.98300585e-01 4.57613915e-03 -8.61130655e-01
1.39410287e-01 -5.26116312e-01 2.00107753e-01 5.98831177e-01
3.49650681e-01 -3.96300733e-01 -3.98252457e-01 9.69411209e-02
5.82596004e-01 2.27042928e-01 1.48004144e-01 -1.33778706e-01
6.56168580e-01 -8.78963053e-01 2.97497660e-01 2.06681281e-01
-1.81679726e-01 8.50883722e-01 3.70277852e-01 3.16079035e-02
-8.81870627e-01 -1.49209332e+00 1.57222375e-01 7.80889869e-01
1.01099640e-01 9.22730267e-02 -9.61869597e-01 -3.94347340e-01
-1.40007913e-01 6.75834656e-01 -8.18818092e-01 -4.00157392e-01
-5.77357173e-01 -1.02234133e-01 6.99899733e-01 2.99808115e-01
1.70251280e-01 -1.25534081e+00 -6.93300441e-02 -3.94694984e-01
-3.62943441e-01 -1.11712563e+00 -1.03775835e+00 -6.86947763e-01
-3.62377703e-01 -9.52237487e-01 -1.05809546e+00 -8.75445068e-01
5.83715796e-01 3.28229129e-01 9.43690062e-01 -3.82544160e-01
-5.92337728e-01 5.89421272e-01 7.61379302e-02 -3.08855236e-01
-7.29779005e-01 -5.55064023e-01 3.07437003e-01 7.55536139e-01
3.64536494e-02 -9.63008344e-01 -6.52316689e-01 2.52720147e-01
-8.77262771e-01 1.96544945e-01 3.61690164e-01 9.41347659e-01
1.39440939e-01 -5.11022687e-01 7.27316737e-01 -4.59230900e-01
6.47732079e-01 -4.58741546e-01 -2.18595102e-01 3.89999270e-01
6.22511096e-02 1.46513760e-01 3.66236389e-01 -9.08669531e-01
-1.43864584e+00 3.41974139e-01 1.13384791e-01 -1.00042605e+00
-2.93426096e-01 -4.10237819e-01 -8.33454072e-01 1.54781491e-01
5.48715472e-01 3.48776042e-01 5.85384011e-01 -2.65287340e-01
7.84428835e-01 7.95998514e-01 8.38077426e-01 -4.62386340e-01
9.25638616e-01 6.09746993e-01 -1.41743898e-01 -1.04359627e+00
-2.51866043e-01 -2.00419351e-02 -4.62855488e-01 -3.50068599e-01
9.56867635e-01 -1.01032209e+00 -1.05856884e+00 3.85755092e-01
-1.35171330e+00 -3.10341995e-02 -4.44897711e-01 3.07373792e-01
-8.88716638e-01 2.64427990e-01 -3.21843207e-01 -9.50638413e-01
-1.24733143e-01 -1.36894143e+00 1.46676958e+00 7.55813122e-02
-3.02802384e-01 -7.60265052e-01 4.41230610e-02 5.85651815e-01
2.12507963e-01 2.26603255e-01 7.07991362e-01 -3.96290988e-01
-7.27692068e-01 -5.32677434e-02 1.61050797e-01 2.96680093e-01
4.91080403e-01 3.02441400e-02 -1.45387912e+00 -1.66667193e-01
2.72102535e-01 -3.73050809e-01 7.67340899e-01 3.42358381e-01
9.57095861e-01 -7.73874700e-01 -2.36255467e-01 5.58506668e-01
8.89371932e-01 2.37501472e-01 6.29667878e-01 -7.75088191e-01
6.73796594e-01 1.17114556e+00 6.99583218e-02 1.18168205e-01
6.58915844e-03 1.06750762e+00 3.31245124e-01 5.96097345e-03
-6.52705312e-01 -5.55839121e-01 7.54521132e-01 7.64671564e-01
1.12713866e-01 -3.45999211e-01 -5.21231949e-01 4.52319652e-01
-1.55337131e+00 -1.61521387e+00 4.77807045e-01 2.04920053e+00
8.65523696e-01 -6.56614304e-01 4.34939787e-02 5.51813692e-02
9.85468626e-01 2.36225381e-01 -6.26194417e-01 -1.51332080e-01
-1.36126131e-01 2.15121046e-01 -2.15737954e-01 6.74980581e-01
-7.24295497e-01 7.08249390e-01 6.45946407e+00 7.49462724e-01
-1.37281621e+00 1.70749739e-01 4.37241912e-01 -5.57055354e-01
-5.32901347e-01 -3.62123489e-01 -4.40961331e-01 4.54466283e-01
7.60345459e-01 -3.96334291e-01 4.72208917e-01 4.73071992e-01
3.09056818e-01 3.20587009e-01 -1.42374051e+00 1.39748359e+00
6.76314950e-01 -1.24555790e+00 4.93493140e-01 7.66853988e-02
5.07442534e-01 -6.43527269e-01 7.93221235e-01 -1.58068627e-01
5.52004315e-02 -1.37625754e+00 8.11966240e-01 7.55845308e-01
1.22507679e+00 -4.32904124e-01 1.39517009e-01 2.50318050e-01
-1.04479074e+00 6.02284893e-02 3.16407621e-01 3.49678099e-01
2.88952410e-01 -1.39348716e-01 -8.97099257e-01 2.86810696e-01
4.07922924e-01 3.99043411e-01 -1.45804077e-01 4.05857384e-01
-3.16093192e-02 2.55551606e-01 6.95680380e-02 2.21596465e-01
-4.45800334e-01 -1.32961646e-01 9.85566139e-01 7.91663408e-01
3.92372608e-01 1.62395984e-01 -3.28748494e-01 1.23724091e+00
-1.81360871e-01 -7.31839091e-02 -1.23685336e+00 -1.06569655e-01
2.33809501e-01 1.13358963e+00 -2.12708130e-01 -1.79361209e-01
-2.18389660e-01 1.09171999e+00 -9.48264524e-02 6.14253521e-01
-7.55272448e-01 -2.51914784e-02 1.16978025e+00 2.09131211e-01
1.52454272e-01 1.77781805e-01 1.80431992e-01 -1.39206231e+00
-5.34972884e-02 -1.06774414e+00 -1.64739609e-01 -1.20396698e+00
-1.14634335e+00 8.20467889e-01 3.58339883e-02 -1.27696693e+00
-8.08838785e-01 -3.46837789e-01 -5.67829669e-01 9.51143503e-01
-1.04137146e+00 -1.39783502e+00 -2.78340042e-01 9.74816918e-01
8.79942834e-01 -5.12928307e-01 1.04049039e+00 1.14115909e-01
-3.32048744e-01 8.26053202e-01 -2.03397125e-01 -1.32403839e-02
9.23337996e-01 -7.57271826e-01 6.33117110e-02 5.05690813e-01
3.50115269e-01 5.92966676e-01 7.64544725e-01 -3.81264478e-01
-1.24277234e+00 -8.27500939e-01 5.76922297e-01 -4.78217870e-01
3.31315398e-01 -7.57421553e-01 -6.41799212e-01 5.50525904e-01
7.49365449e-01 -2.46988758e-02 9.51465070e-01 -5.58013022e-01
-4.69088614e-01 -2.86420554e-01 -1.26920938e+00 9.71886039e-01
1.00294590e+00 -9.26382959e-01 -6.95763290e-01 2.03159600e-01
7.39692807e-01 -1.12309687e-01 -3.71074080e-01 2.47439533e-01
7.87323892e-01 -9.55185175e-01 1.14902902e+00 -6.65323555e-01
4.29497570e-01 -2.19955519e-01 -2.10391968e-01 -1.26350141e+00
1.63396984e-01 -1.11981034e+00 -1.62037835e-01 1.43489218e+00
2.06065178e-01 -3.96830887e-01 5.71316600e-01 3.40656281e-01
3.12038571e-01 -4.38849896e-01 -8.23566854e-01 -4.10415530e-01
-1.42572289e-02 -1.50260581e-02 6.87043905e-01 9.30602193e-01
1.00133866e-01 6.10788703e-01 -7.61694252e-01 1.32215962e-01
6.93774283e-01 8.47483948e-02 7.54789233e-01 -9.69359517e-01
-3.20245951e-01 -2.67309874e-01 -4.56617028e-01 -8.27409029e-01
7.62114108e-01 -8.48802805e-01 6.18008636e-02 -7.64969587e-01
2.26066411e-01 1.57599822e-01 1.52305782e-01 2.06503615e-01
7.65799880e-02 3.05701882e-01 3.34538102e-01 1.33221835e-01
8.18141624e-02 8.64859462e-01 1.58144295e+00 -2.79358685e-01
3.03418674e-02 4.95969243e-02 -5.58184564e-01 6.87926948e-01
5.17973483e-01 -2.05046415e-01 -9.31341887e-01 -2.44521290e-01
-4.64694619e-01 7.67804384e-01 6.49191916e-01 -6.42622530e-01
1.35997713e-01 -2.74449550e-02 4.22302604e-01 -2.64958769e-01
1.10163903e+00 -8.68331492e-01 4.07641411e-01 1.82952508e-01
-5.84956229e-01 -1.51816666e-01 1.92340076e-01 7.39808381e-01
-3.79861176e-01 1.03174888e-01 8.39456856e-01 -1.33666828e-01
-1.62604332e-01 4.26885873e-01 -2.83902675e-01 -1.07293457e-01
1.23142004e+00 -3.15268040e-01 -1.97658669e-02 -1.14178574e+00
-1.22923577e+00 -1.95015177e-01 2.01599747e-01 7.93658853e-01
8.39230657e-01 -1.49099815e+00 -6.89771295e-01 6.21923327e-01
-2.05244452e-01 -4.65339303e-01 3.94832253e-01 5.03537714e-01
-4.14718650e-02 2.93435693e-01 -4.90527779e-01 -5.78277886e-01
-1.69862306e+00 7.50843287e-01 3.47462714e-01 3.01888108e-01
-3.42471063e-01 7.97648072e-01 7.93966413e-01 -1.91726863e-01
2.75307894e-01 2.44138792e-01 -1.87437713e-01 1.88897923e-01
6.95232272e-01 -2.66946778e-02 -4.63949740e-01 -1.13861418e+00
-1.00377433e-01 7.26774514e-01 1.00274496e-01 -5.90213060e-01
1.02570903e+00 -1.70282304e-01 1.31765664e-01 4.78411824e-01
1.40483773e+00 3.49405497e-01 -1.23401904e+00 -2.86291335e-02
-7.40405023e-01 -4.53242958e-01 -4.01442230e-01 -5.69781899e-01
-1.34721947e+00 1.25041819e+00 7.54760802e-01 -1.25175506e-01
1.15918648e+00 1.28841877e-01 4.79004979e-01 -1.74162209e-01
1.41802609e-01 -4.48909283e-01 4.77023333e-01 -2.77158320e-01
1.47478247e+00 -9.89049196e-01 -3.55768502e-01 -6.55622900e-01
-7.26468205e-01 8.70713532e-01 6.41989768e-01 6.41821623e-02
7.17129648e-01 4.84785825e-01 2.34330714e-01 1.21326372e-01
-9.15998161e-01 -3.07300370e-02 5.96128583e-01 9.58555460e-01
1.79553002e-01 -1.52281761e-01 1.76706240e-01 5.35634160e-01
-5.02844214e-01 -2.12715149e-01 3.05179954e-01 3.05459231e-01
1.74095064e-01 -1.07460785e+00 -5.32135844e-01 -1.30354658e-01
-3.98442119e-01 1.86813492e-02 -6.46630526e-01 7.57652283e-01
4.56962511e-02 7.78836846e-01 9.13779736e-02 -3.92056316e-01
-1.99073534e-02 5.38941741e-01 7.93296933e-01 -4.80330259e-01
-6.61886111e-02 1.17362700e-01 -2.68211633e-01 -6.93344831e-01
-4.90974128e-01 -6.83777213e-01 -8.80105078e-01 -2.06716076e-01
-2.01309934e-01 -3.71366041e-03 6.47589684e-01 6.32052302e-01
4.19404626e-01 3.63128185e-01 6.94051743e-01 -1.31146479e+00
-6.24003470e-01 -8.37453842e-01 -5.96636057e-01 6.39748931e-01
8.53855193e-01 -6.96421683e-01 -5.93962610e-01 5.76956213e-01]
|
[13.216524124145508, -0.3931722939014435]
|
caaff2d0-c34e-45d3-8c79-ae872caa5476
|
a-keypoint-based-global-association-network
|
2204.07335
| null |
https://arxiv.org/abs/2204.07335v1
|
https://arxiv.org/pdf/2204.07335v1.pdf
|
A Keypoint-based Global Association Network for Lane Detection
|
Lane detection is a challenging task that requires predicting complex topology shapes of lane lines and distinguishing different types of lanes simultaneously. Earlier works follow a top-down roadmap to regress predefined anchors into various shapes of lane lines, which lacks enough flexibility to fit complex shapes of lanes due to the fixed anchor shapes. Lately, some works propose to formulate lane detection as a keypoint estimation problem to describe the shapes of lane lines more flexibly and gradually group adjacent keypoints belonging to the same lane line in a point-by-point manner, which is inefficient and time-consuming during postprocessing. In this paper, we propose a Global Association Network (GANet) to formulate the lane detection problem from a new perspective, where each keypoint is directly regressed to the starting point of the lane line instead of point-by-point extension. Concretely, the association of keypoints to their belonged lane line is conducted by predicting their offsets to the corresponding starting points of lanes globally without dependence on each other, which could be done in parallel to greatly improve efficiency. In addition, we further propose a Lane-aware Feature Aggregator (LFA), which adaptively captures the local correlations between adjacent keypoints to supplement local information to the global association. Extensive experiments on two popular lane detection benchmarks show that our method outperforms previous methods with F1 score of 79.63% on CULane and 97.71% on Tusimple dataset with high FPS. The code will be released at https://github.com/Wolfwjs/GANet.
|
['Tianzhu Zhang', 'Chen Qian', 'Fei Wang', 'Tianrui Hui', 'Shaofei Huang', 'Yinchao Ma', 'Jinsheng Wang']
|
2022-04-15
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Wang_A_Keypoint-Based_Global_Association_Network_for_Lane_Detection_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Wang_A_Keypoint-Based_Global_Association_Network_for_Lane_Detection_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['lane-detection']
|
['computer-vision']
|
[-3.73073667e-01 -3.43114197e-01 -4.23373163e-01 -4.33557034e-01
-4.39682722e-01 -6.67372346e-01 4.38589424e-01 -6.93697948e-03
-1.31031945e-01 4.92840677e-01 9.99534577e-02 -5.33027887e-01
-1.65281892e-01 -9.03792679e-01 -6.68137074e-01 -6.18514776e-01
-2.08053246e-01 1.19307794e-01 6.74303710e-01 -4.18257892e-01
4.66941088e-01 9.22423720e-01 -1.25120914e+00 -9.31258649e-02
1.10055470e+00 7.14266479e-01 4.47588898e-02 4.34460342e-01
-4.26979095e-01 2.88785130e-01 -1.92283630e-01 -1.85369160e-02
4.76913899e-01 3.36338915e-02 -1.25277564e-01 2.55954534e-01
6.64316714e-01 -3.72644722e-01 -4.13660109e-01 8.63714337e-01
-8.57910421e-03 7.75692686e-02 5.83248079e-01 -1.59417713e+00
-9.56568122e-02 1.70443535e-01 -1.21200264e+00 7.31326714e-02
-8.41755718e-02 1.38099134e-01 1.04406118e+00 -1.03487980e+00
4.00650322e-01 1.05443692e+00 7.55041718e-01 -2.26682991e-01
-1.12047470e+00 -8.56963038e-01 6.00672185e-01 1.58791631e-01
-1.60588491e+00 -2.21252680e-01 1.11852086e+00 -2.87713259e-01
3.22672516e-01 3.41587484e-01 4.24386740e-01 3.72646540e-01
1.23066008e-01 7.51560748e-01 8.24896216e-01 4.29050438e-02
-2.78867126e-01 -2.29255259e-01 4.15179551e-01 7.92984068e-01
4.12025779e-01 -1.24137878e-01 -3.23629946e-01 2.09141187e-02
9.84797478e-01 1.58180207e-01 -1.99857637e-01 -7.42752194e-01
-1.38223779e+00 7.33788610e-01 7.01929748e-01 -1.60677329e-01
-1.06191486e-01 -4.77798395e-02 3.59644383e-01 4.21856456e-02
7.45831355e-02 1.44224569e-01 -3.12278181e-01 1.83028713e-01
-6.29014492e-01 3.26739401e-01 5.53593695e-01 1.10939169e+00
1.30106258e+00 -2.02587500e-01 7.88025334e-02 7.77344048e-01
2.54545093e-01 5.66544294e-01 -2.63087839e-01 -8.55046451e-01
8.80906641e-01 8.08127761e-01 1.51700944e-01 -1.41425717e+00
-7.65354395e-01 -4.79257971e-01 -1.03644705e+00 1.78359747e-01
7.17754424e-01 -2.55879194e-01 -6.28959894e-01 1.34561706e+00
3.72653782e-01 5.34172535e-01 -2.67695159e-01 1.03852713e+00
3.63836616e-01 8.98464620e-01 -9.26567167e-02 1.65633392e-02
1.24229538e+00 -1.09555447e+00 -4.40984428e-01 -2.98228532e-01
7.71027565e-01 -7.89822996e-01 1.00324094e+00 3.04888666e-01
-4.88645226e-01 -6.51508987e-01 -1.02163386e+00 3.65633778e-02
-3.97178769e-01 4.86058831e-01 6.41902626e-01 1.92670301e-01
-7.81743407e-01 -4.52042930e-02 -4.08542633e-01 -3.21033150e-02
1.43646508e-01 1.16561867e-01 -3.21906716e-01 2.21633092e-02
-1.01970732e+00 3.51323336e-01 3.12417686e-01 3.91527742e-01
-7.74471015e-02 -8.54181588e-01 -9.18728113e-01 -1.16342977e-02
8.28717530e-01 -2.83971250e-01 6.66906893e-01 -4.56742167e-01
-1.03509426e+00 2.51402736e-01 -3.76013726e-01 -2.42386296e-01
6.52528405e-01 -2.51353860e-01 -7.43778646e-01 -2.17065036e-01
3.59458715e-01 5.95326662e-01 5.40254891e-01 -1.44883251e+00
-1.29115176e+00 -1.56505313e-03 -4.86844517e-02 8.67497921e-02
9.91280675e-02 -4.72936928e-01 -7.45507002e-01 -5.39974630e-01
4.74551260e-01 -9.53336954e-01 -3.23494941e-01 -5.81133040e-03
-7.34623611e-01 -3.70364636e-01 1.12857652e+00 -4.49468613e-01
1.72395563e+00 -2.31914401e+00 -4.46282864e-01 6.70334041e-01
4.26197022e-01 1.00390173e-01 -5.65730520e-02 4.97656256e-01
4.72852811e-02 2.58169957e-02 -6.62874505e-02 7.31407627e-02
1.06791742e-02 8.40737596e-02 -6.50526106e-01 5.03049612e-01
1.54414639e-01 6.87123060e-01 -8.88742030e-01 -4.04940128e-01
4.90946174e-01 1.32425949e-01 -2.64909863e-01 -1.07230969e-01
3.17887753e-01 1.15979478e-01 -5.00601172e-01 6.26846552e-01
1.07493746e+00 -6.74024085e-03 1.44999381e-02 -6.63784266e-01
-6.56972170e-01 -7.46421982e-04 -1.41829395e+00 1.19233322e+00
-3.00570786e-01 7.75131226e-01 -1.77280843e-01 -7.53633797e-01
1.42773890e+00 -3.41681272e-01 4.16667640e-01 -6.03914320e-01
-2.78278619e-01 2.27928132e-01 5.11108199e-03 -1.19178824e-01
6.68672144e-01 4.16583240e-01 -6.63300097e-01 -7.53794834e-02
-7.37550914e-01 2.53797442e-01 2.89048553e-01 5.49921989e-02
8.80494058e-01 -5.84955849e-02 2.52480447e-01 -1.59013852e-01
7.87923753e-01 9.87125114e-02 8.54371309e-01 7.56959736e-01
-2.43984550e-01 3.96072000e-01 6.21162236e-01 -7.64576912e-01
-9.98039782e-01 -1.26967168e+00 -3.82601768e-02 7.85659432e-01
6.87891304e-01 -4.11155283e-01 -2.68492430e-01 -8.81830215e-01
3.50023985e-01 5.04533172e-01 -2.75198996e-01 1.69149444e-01
-9.67520654e-01 -3.39605480e-01 3.19772720e-01 5.36891818e-01
7.11642385e-01 -4.38383073e-01 -3.28314364e-01 3.05210620e-01
-3.36178727e-02 -1.04246676e+00 -8.46440971e-01 -4.95124497e-02
-4.80442822e-01 -1.07173359e+00 -3.38634521e-01 -6.53378129e-01
7.90236294e-01 7.96247900e-01 6.12532556e-01 8.28822777e-02
2.62347579e-01 -2.79258937e-01 -1.73135057e-01 -6.11594245e-02
1.06387548e-01 2.88930714e-01 -1.11793168e-01 1.93101764e-01
2.28508025e-01 -2.87818104e-01 -6.57755911e-01 9.20532048e-01
-3.70563686e-01 4.46398139e-01 8.76895308e-01 6.96398079e-01
6.54495656e-01 1.55447736e-01 4.95981038e-01 -7.55691946e-01
3.41880858e-01 -6.82650864e-01 -8.64039898e-01 5.64263836e-02
-3.52364480e-01 -9.86217409e-02 9.18699741e-01 -1.77614003e-01
-8.12105238e-01 3.50714207e-01 8.93964991e-02 -5.45042634e-01
-4.72476155e-01 5.16122699e-01 -3.71187776e-01 -2.14050245e-02
1.11418299e-01 3.23570967e-01 1.27868988e-02 -3.11919093e-01
3.54204744e-01 3.83839160e-01 7.18603551e-01 -3.53209883e-01
1.27726209e+00 5.45578837e-01 3.88723671e-01 -7.00395823e-01
-5.68371594e-01 -8.24393213e-01 -8.18760276e-01 -4.40551221e-01
3.73739213e-01 -7.47512162e-01 -6.77052736e-01 4.76397306e-01
-9.28293586e-01 -2.45751560e-01 3.61020178e-01 1.50814548e-01
-2.36927867e-01 4.76682037e-01 -2.40823299e-01 -4.12061036e-01
9.28495824e-02 -9.74982560e-01 9.84967232e-01 3.63050282e-01
-2.21469309e-02 -8.78963768e-01 -1.70310482e-01 4.68321480e-02
3.85831669e-02 4.18832600e-01 6.95643723e-01 -3.97168249e-01
-9.18821394e-01 -5.06771028e-01 -5.73502421e-01 -1.39136851e-01
4.39121783e-01 2.34857276e-01 -3.59970808e-01 -4.58170213e-02
-7.40991592e-01 5.13280869e-01 8.62604320e-01 3.26805562e-01
1.16604745e+00 -1.52255252e-01 -6.64012671e-01 7.06938028e-01
1.34125876e+00 2.03821048e-01 6.10607743e-01 5.40240645e-01
1.06116056e+00 5.96863210e-01 1.05456185e+00 2.37990260e-01
7.21401930e-01 7.87463486e-01 2.46272326e-01 -3.79882574e-01
8.40667710e-02 -4.63366687e-01 2.39694476e-01 5.70624053e-01
1.42645657e-01 -2.91191608e-01 -1.01476085e+00 5.33302605e-01
-2.12863803e+00 -9.42449450e-01 -7.42372572e-01 2.18452501e+00
2.80858111e-02 5.64589024e-01 4.00979906e-01 5.85297495e-02
8.83284509e-01 3.04385424e-01 -4.80460167e-01 -2.14493260e-01
-2.63179988e-02 -8.35395277e-01 1.11219323e+00 5.21246731e-01
-1.37488186e+00 9.17041421e-01 4.88469982e+00 9.30840373e-01
-1.27851176e+00 -4.75810498e-01 6.96639359e-01 4.75484699e-01
-1.28609255e-01 1.97993353e-01 -1.22135413e+00 7.10731149e-01
3.84725869e-01 2.26460099e-02 1.90586254e-01 6.47503257e-01
7.17097759e-01 -1.68854922e-01 -5.48114181e-01 7.98117220e-01
-2.98466533e-01 -1.42957914e+00 1.47448286e-01 2.55040556e-01
5.28459609e-01 -2.50203367e-02 -9.18058604e-02 2.38598466e-01
8.87402073e-02 -5.27415991e-01 5.72323143e-01 6.58158481e-01
6.42444134e-01 -9.85032022e-01 6.15183353e-01 4.59682941e-01
-1.85804343e+00 -1.25096053e-01 -1.51607767e-01 4.51770052e-02
3.61454606e-01 5.25829077e-01 -1.00937521e+00 7.35258579e-01
5.13586700e-01 8.90338063e-01 -5.45882583e-01 1.21096754e+00
-1.86804742e-01 6.57014370e-01 -4.96597618e-01 1.34307876e-01
5.78810573e-01 -4.47640687e-01 6.48856282e-01 1.27879190e+00
4.35965449e-01 -1.08827211e-01 6.77904189e-01 5.65989375e-01
2.65445322e-01 1.35630608e-01 -5.73312938e-01 7.32259393e-01
8.31558943e-01 1.41755736e+00 -8.43321562e-01 -1.24193802e-01
-7.98869669e-01 3.19346070e-01 2.14489266e-01 4.01472002e-01
-1.10549331e+00 -8.65792215e-01 7.80137658e-01 5.39483130e-01
3.24630946e-01 -5.58984339e-01 -3.32592696e-01 -8.14788938e-01
1.04171075e-01 -2.95474350e-01 2.67738044e-01 -6.79197609e-01
-1.28554547e+00 1.79119900e-01 -1.58428457e-02 -1.81101882e+00
1.19334169e-01 -4.15591180e-01 -9.22004998e-01 8.45810473e-01
-1.74120188e+00 -1.27961469e+00 -6.24700904e-01 4.19553518e-01
5.76044261e-01 1.81005210e-01 6.29442111e-02 1.95996031e-01
-6.27438247e-01 6.10811949e-01 1.35863870e-01 4.24448460e-01
8.41170609e-01 -1.06425202e+00 5.33439815e-01 1.01141953e+00
-2.73111820e-01 5.30665338e-01 5.65868676e-01 -7.63482451e-01
-1.31734860e+00 -1.41708612e+00 7.17978895e-01 -1.00567944e-01
9.66890275e-01 -3.63635629e-01 -1.13928592e+00 5.58351040e-01
-2.16307774e-01 1.97859913e-01 2.70861268e-01 1.34741785e-02
-2.50171363e-01 -6.61580861e-01 -6.48082674e-01 8.46889734e-01
1.13647509e+00 -5.82805313e-02 -2.01105759e-01 -1.47971168e-01
3.55239272e-01 -2.25579038e-01 -7.26809800e-01 5.61720610e-01
4.98691380e-01 -9.70482647e-01 1.04800332e+00 1.11041501e-01
1.03295609e-01 -1.17276287e+00 1.51917353e-01 -1.27103436e+00
-4.96022046e-01 -4.45632100e-01 9.13989991e-02 1.26273263e+00
2.61972398e-01 -9.88218069e-01 7.70389616e-01 2.06216961e-01
-6.13627076e-01 -8.43960702e-01 -7.67700672e-01 -7.24073052e-01
-2.85990477e-01 -4.00576770e-01 9.36831355e-01 8.58677089e-01
-1.36482865e-01 1.73583388e-01 -4.27053541e-01 7.20471025e-01
5.84058702e-01 3.43743950e-01 1.30116320e+00 -1.26071906e+00
2.20793024e-01 -7.04575241e-01 -4.40146357e-01 -1.54613149e+00
1.38964996e-01 -7.94987440e-01 1.47516683e-01 -1.30683339e+00
-2.46086627e-01 -9.00650680e-01 -2.54249781e-01 4.96768385e-01
-2.61477172e-01 -3.85617604e-03 8.29951316e-02 3.85809898e-01
-5.93699932e-01 1.61072165e-01 1.25057125e+00 -1.49789542e-01
-4.20803457e-01 8.79328847e-02 -4.50329542e-01 7.63086677e-01
1.08098197e+00 7.83512220e-02 -4.62147415e-01 -1.72907650e-01
6.22590184e-02 1.03418671e-01 3.78668249e-01 -1.08310950e+00
4.78750706e-01 -4.20970708e-01 3.40472400e-01 -1.28350735e+00
9.83773768e-02 -9.25276816e-01 1.90079421e-01 2.60241956e-01
1.09278023e-01 2.66754478e-01 3.24782878e-01 7.02718556e-01
-2.23098755e-01 2.41795078e-01 3.74954313e-01 4.42953914e-01
-1.44360447e+00 3.74623477e-01 -3.42628032e-01 -4.45131153e-01
1.34361506e+00 -5.68347275e-01 -7.19160080e-01 -2.90070742e-01
-2.76044995e-01 7.15475917e-01 5.49330771e-01 4.60961521e-01
5.96494138e-01 -1.41890216e+00 -5.99095106e-01 3.52693796e-01
3.39917868e-01 1.24475196e-01 3.47599536e-01 9.83174145e-01
-5.60435295e-01 4.82847959e-01 -1.60196647e-01 -7.81951487e-01
-1.14397192e+00 4.63399589e-01 2.47875974e-01 -1.19431317e-01
-7.29668677e-01 1.53853267e-01 3.93184215e-01 -2.79041439e-01
-1.71398878e-01 -4.39513385e-01 -3.22667301e-01 2.13305769e-03
3.77001107e-01 5.30298769e-01 -2.57564932e-01 -8.80612135e-01
-3.14931899e-01 1.02152550e+00 -1.27258852e-01 5.35290599e-01
1.00914896e+00 -4.73076284e-01 6.87588006e-02 4.08839792e-01
9.98118758e-01 4.27082270e-01 -1.62925351e+00 -5.31413741e-02
1.60453647e-01 -6.47128463e-01 -8.37487727e-02 -4.81570810e-01
-1.10730934e+00 4.52167869e-01 3.71212512e-01 1.71191573e-01
1.01656294e+00 -1.35924801e-01 8.97063255e-01 2.53343076e-01
4.69185680e-01 -9.01437163e-01 -3.26020300e-01 5.09153128e-01
6.49223506e-01 -1.17113137e+00 -8.32131281e-02 -1.00209498e+00
-5.33505678e-01 1.44879556e+00 8.13619554e-01 -3.87281656e-01
5.85839450e-01 2.38563895e-01 2.50597596e-02 2.23214403e-02
-3.65450442e-01 -1.20667689e-01 4.41308260e-01 2.65069425e-01
1.31525621e-01 3.98677915e-01 -2.02618316e-01 3.32612783e-01
7.04856291e-02 -3.09493572e-01 5.72576106e-01 6.77315474e-01
-6.23179078e-01 -9.67893362e-01 -5.21517038e-01 6.87738955e-01
4.27163750e-01 1.70017660e-01 1.13199435e-01 1.13574684e+00
2.87217498e-01 6.24836624e-01 4.26162332e-01 -5.90384722e-01
5.40856421e-01 -2.50043541e-01 -4.04740304e-01 -1.69570237e-01
9.76054370e-02 2.79621989e-01 1.55258015e-01 -5.46902120e-01
7.92325586e-02 -9.61119831e-01 -1.52032542e+00 -5.57834804e-01
-1.70213252e-01 5.96872829e-02 3.00421566e-01 6.03875160e-01
6.37144089e-01 3.12553167e-01 1.02366233e+00 -8.02263856e-01
1.92201361e-02 -4.79397684e-01 -4.53191549e-01 1.71658352e-01
4.49986339e-01 -8.29821169e-01 -3.58527184e-01 -2.22545415e-01]
|
[8.019418716430664, -1.565313696861267]
|
e3296d46-65d6-4db6-8c64-ca20f8ffba41
|
cut-and-paste-neural-rendering-1
|
2010.05907
| null |
https://arxiv.org/abs/2010.05907v2
|
https://arxiv.org/pdf/2010.05907v2.pdf
|
Cut-and-Paste Object Insertion by Enabling Deep Image Prior for Reshading
|
We show how to insert an object from one image to another and get realistic results in the hard case, where the shading of the inserted object clashes with the shading of the scene. Rendering objects using an illumination model of the scene doesn't work, because doing so requires a geometric and material model of the object, which is hard to recover from a single image. In this paper, we introduce a method that corrects shading inconsistencies of the inserted object without requiring a geometric and physical model or an environment map. Our method uses a deep image prior (DIP), trained to produce reshaded renderings of inserted objects via consistent image decomposition inferential losses. The resulting image from DIP aims to have (a) an albedo similar to the cut-and-paste albedo, (b) a similar shading field to that of the target scene, and (c) a shading that is consistent with the cut-and-paste surface normals. The result is a simple procedure that produces convincing shading of the inserted object. We show the efficacy of our method both qualitatively and quantitatively for several objects with complex surface properties and also on a dataset of spherical lampshades for quantitative evaluation. Our method significantly outperforms an Image Harmonization (IH) baseline for all these objects. They also outperform the cut-and-paste and IH baselines in a user study with over 100 users.
|
['David A. Forsyth', 'Anand Bhattad']
|
2020-10-12
|
cut-and-paste-neural-rendering
|
https://openreview.net/forum?id=IfEkus1dpU
|
https://openreview.net/pdf?id=IfEkus1dpU
| null |
['image-harmonization']
|
['computer-vision']
|
[ 5.84774613e-01 1.85499161e-01 9.32195783e-01 -4.13157642e-01
-4.60270435e-01 -4.82488096e-01 5.55307329e-01 -1.11231823e-02
9.38460007e-02 3.97369176e-01 7.53424168e-02 -7.01670274e-02
3.09830934e-01 -7.79615343e-01 -9.86744821e-01 -9.06502008e-01
4.62762207e-01 3.61085415e-01 4.96420532e-01 -3.44310433e-01
2.28263989e-01 4.90377367e-01 -1.84036779e+00 5.21064162e-01
1.03583407e+00 7.67344475e-01 5.61498284e-01 8.89495969e-01
-5.69270290e-02 5.54117262e-01 -6.89862430e-01 -3.41393620e-01
6.29629731e-01 -6.26061440e-01 -4.13386196e-01 4.89341825e-01
1.06585503e+00 -5.25818169e-01 1.10699147e-01 9.76590097e-01
3.91691059e-01 2.42965575e-02 5.90130687e-01 -9.12194967e-01
-5.33406615e-01 -4.26144928e-01 -1.07901359e+00 -6.91603720e-01
5.18614233e-01 2.17971504e-01 7.48865545e-01 -7.51275182e-01
6.20344818e-01 1.06824350e+00 6.20735168e-01 7.95444846e-02
-1.53302503e+00 -3.01541001e-01 -1.17911480e-01 -4.40636784e-01
-1.17554796e+00 -3.50000113e-01 7.32876778e-01 -4.33991253e-01
7.19495118e-01 5.59262335e-01 6.44636333e-01 3.61059517e-01
1.92156136e-01 1.91341951e-01 1.24185407e+00 -7.64229298e-01
2.56556362e-01 5.19155741e-01 -2.54356295e-01 5.69692671e-01
3.20568457e-02 -1.40697574e-02 -3.92229617e-01 -2.28036344e-01
7.54343331e-01 -2.73382157e-01 -6.67402387e-01 -5.74502468e-01
-8.09326589e-01 3.80867869e-01 4.51627910e-01 -8.22795928e-02
-3.37494642e-01 1.81044504e-01 -2.47111827e-01 -9.38631073e-02
4.21823025e-01 3.74316216e-01 -1.40917957e-01 4.53547299e-01
-9.31455135e-01 4.05137956e-01 7.71785319e-01 7.65264869e-01
1.05829418e+00 1.53503995e-02 6.54874668e-02 8.42735887e-01
2.73026317e-01 7.84229517e-01 8.58424813e-04 -1.42222071e+00
1.10440388e-01 4.13845778e-01 5.86601257e-01 -1.07771456e+00
-7.75426403e-02 -1.57415837e-01 -5.75473249e-01 1.08973610e+00
4.29681510e-01 1.92247033e-02 -9.29935515e-01 1.66639042e+00
5.76607227e-01 7.80543983e-02 -1.10624939e-01 9.43533838e-01
6.77101493e-01 8.18998992e-01 -5.79348266e-01 -4.85467128e-02
1.34347236e+00 -8.73513043e-01 -4.63605255e-01 -4.11263049e-01
1.23545647e-01 -1.23623431e+00 1.43398798e+00 5.68657279e-01
-1.48056281e+00 -5.19883931e-01 -1.17479026e+00 -3.11150283e-01
2.15343803e-01 4.18151468e-02 4.85333234e-01 7.21563578e-01
-1.49780405e+00 5.04430652e-01 -5.58087766e-01 -2.29522333e-01
4.43201512e-02 6.07548915e-02 -2.50430495e-01 -8.92000645e-02
-4.03803855e-01 6.77222788e-01 -1.09581627e-01 -1.82456687e-01
-4.96722013e-01 -9.61229384e-01 -7.34116316e-01 1.72832817e-01
9.77247506e-02 -8.49508345e-01 1.13877535e+00 -1.41995287e+00
-1.63747227e+00 1.00252938e+00 -3.61068666e-01 -3.15174535e-02
5.02313554e-01 -2.14321032e-01 1.37078241e-01 2.21399441e-01
-2.96633691e-02 4.30719882e-01 9.45824265e-01 -2.11251163e+00
-3.63551050e-01 -1.74421281e-01 5.73066957e-02 5.51056206e-01
2.14669645e-01 -9.97873023e-02 -5.76120377e-01 -4.09979463e-01
3.23021710e-01 -6.73622310e-01 -3.29865552e-02 3.56945962e-01
-6.23047173e-01 5.71119249e-01 8.17431152e-01 -8.93079281e-01
6.43167138e-01 -2.29618216e+00 -2.86428541e-01 3.63219559e-01
1.07620530e-01 4.26080935e-02 -1.04880989e-01 4.59855676e-01
-2.84678370e-01 -1.41283363e-01 -4.05168742e-01 -7.66020298e-01
-2.04326138e-01 3.45117934e-02 -5.94047785e-01 4.71849650e-01
-1.85785353e-01 5.73296666e-01 -5.58206975e-01 -1.73107520e-01
2.92795986e-01 7.84889340e-01 -6.83869600e-01 5.54455996e-01
-2.70600766e-01 4.36397314e-01 2.36078426e-01 1.95514187e-01
1.07931244e+00 -1.15473367e-01 1.20625339e-01 -3.94640028e-01
-2.90189177e-01 1.90451398e-01 -1.36626661e+00 1.37515807e+00
-5.24188340e-01 5.88398993e-01 4.42400455e-01 -3.56974244e-01
8.73546600e-01 1.26567800e-02 4.29127395e-01 -9.32010174e-01
-2.07597092e-01 1.77435115e-01 -3.24604362e-01 -2.46493042e-01
4.27724510e-01 -3.84192675e-01 4.91458625e-01 6.59475505e-01
-5.72167575e-01 -9.25396085e-01 -1.91719264e-01 2.96698302e-01
8.24358642e-01 2.61713147e-01 -4.14918251e-02 -3.07938308e-01
2.18880430e-01 -3.75076562e-01 2.65715003e-01 7.15100944e-01
4.88712102e-01 1.45269704e+00 3.98053139e-01 -2.15701833e-01
-1.25286734e+00 -1.19219887e+00 -5.81795583e-03 5.67214608e-01
4.09158230e-01 -1.89416185e-01 -1.10119975e+00 1.19137891e-01
-7.60628060e-02 1.02431369e+00 -5.95877409e-01 2.21227631e-01
-3.78698826e-01 -4.01582062e-01 -1.92152902e-01 2.96730064e-02
5.84846199e-01 -8.10207188e-01 -1.00099730e+00 -9.85570475e-02
-4.23404664e-01 -9.24537897e-01 -6.06733322e-01 -1.84679657e-01
-5.76737225e-01 -1.04785359e+00 -4.82365906e-01 -6.30697191e-01
9.14875925e-01 6.85502052e-01 1.38500571e+00 4.78744894e-01
-4.57070082e-01 4.30632204e-01 -3.48339812e-03 -4.71921682e-01
-5.02340913e-01 -8.71607184e-01 -3.37801844e-01 3.69214267e-01
-5.61778665e-01 -6.76950812e-01 -8.55052531e-01 5.11181831e-01
-1.09943557e+00 8.04281592e-01 -5.84797934e-02 4.55925703e-01
4.54831392e-01 2.36687943e-01 -2.49784186e-01 -6.23624861e-01
1.93686798e-01 1.07817240e-01 -9.93337333e-01 2.11392090e-01
-3.14723700e-01 -2.58005083e-01 3.92428309e-01 -1.30699649e-01
-1.65164638e+00 1.33277521e-01 1.00821428e-01 -2.20441028e-01
-5.40359654e-02 -2.16119438e-01 -3.77614647e-01 -6.30427822e-02
7.56429911e-01 1.26181263e-03 -2.81879045e-02 -5.42340636e-01
3.16747576e-01 3.93077403e-01 8.01345229e-01 -6.21249974e-01
9.73832548e-01 1.18666387e+00 4.59780246e-02 -1.14758372e+00
-7.06496894e-01 -1.82437986e-01 -4.06602800e-01 -1.47221625e-01
7.09881365e-01 -6.45309627e-01 -7.37294972e-01 7.69843102e-01
-1.47466111e+00 -8.26525033e-01 -5.53986788e-01 9.65702608e-02
-5.63853264e-01 4.70531464e-01 -2.49637082e-01 -9.21921670e-01
-1.90633200e-02 -1.05760002e+00 1.38653505e+00 3.12384337e-01
3.02545503e-02 -7.67916739e-01 7.87469074e-02 3.80256742e-01
5.11302531e-01 4.07073915e-01 9.69670594e-01 4.77782786e-01
-7.88233995e-01 7.12495595e-02 -5.06607533e-01 4.70591396e-01
2.64246762e-01 4.10350442e-01 -1.52439225e+00 -3.22292089e-01
3.57165962e-01 -4.15380485e-02 5.65451026e-01 6.56680822e-01
9.75609660e-01 -2.85067230e-01 -6.89889416e-02 7.66874254e-01
1.71181285e+00 1.63295522e-01 1.29237485e+00 1.21120915e-01
8.02682519e-01 1.09096098e+00 2.89353132e-01 4.50930089e-01
3.39151442e-01 9.14612889e-01 6.08702660e-01 -7.53315389e-01
-5.85831285e-01 -5.98384403e-02 2.12409809e-01 2.00255975e-01
-4.33581062e-02 -4.03773665e-01 -6.52853072e-01 2.69407451e-01
-1.50531936e+00 -6.98956907e-01 -5.82436740e-01 2.71178937e+00
8.08935285e-01 -8.61263350e-02 -2.55218834e-01 -9.25661027e-02
4.48568732e-01 -9.65635255e-02 -3.32013547e-01 -4.97644007e-01
-3.17415893e-01 2.17339769e-01 2.61224866e-01 1.14364541e+00
-5.51804721e-01 6.06645644e-01 6.19384480e+00 4.10259575e-01
-1.08534133e+00 6.37987722e-03 7.51875818e-01 8.41070414e-02
-7.87821233e-01 2.70928919e-01 -4.63929415e-01 3.44155639e-01
5.05766869e-01 8.37097615e-02 6.30318880e-01 3.30542952e-01
4.71639246e-01 -7.66833723e-01 -1.03942704e+00 9.86630917e-01
2.49408513e-01 -1.05993009e+00 -1.99812323e-01 2.91949827e-02
8.79140198e-01 -1.48194268e-01 1.60185192e-02 -4.08569664e-01
2.75672197e-01 -9.93849039e-01 9.35800374e-01 5.39931893e-01
8.23204339e-01 -2.28663191e-01 3.71990085e-01 2.96530962e-01
-9.85667408e-01 3.97288680e-01 -2.06957370e-01 -3.67939062e-02
3.17577511e-01 8.37621748e-01 -8.26049149e-01 3.43519837e-01
8.64935517e-01 1.08663172e-01 -4.70182925e-01 1.04805529e+00
-3.22393358e-01 3.31239909e-01 -6.43105507e-01 6.37017429e-01
-2.58802980e-01 -5.95794380e-01 4.57713842e-01 9.70542550e-01
2.75635302e-01 2.45235518e-01 5.97689748e-02 1.19047809e+00
1.11585520e-01 1.55322984e-01 -4.44921762e-01 4.81439501e-01
-7.37124905e-02 1.19932127e+00 -7.01003253e-01 -2.89439648e-01
-3.72356415e-01 1.32529044e+00 3.66657227e-02 8.70685279e-01
-6.99017525e-01 -3.13296795e-01 5.05943716e-01 6.75630927e-01
9.95316729e-02 -3.29941362e-02 -6.06852472e-01 -8.48112166e-01
3.86280447e-01 -6.31730258e-01 -2.56066352e-01 -1.67414975e+00
-9.00629878e-01 4.55108762e-01 -2.78853495e-02 -1.04683530e+00
1.51899934e-01 -3.23953360e-01 -7.97290027e-01 1.40447378e+00
-1.49170971e+00 -1.10193920e+00 -8.20647657e-01 3.38058591e-01
2.34781355e-01 5.63995540e-01 6.51845992e-01 1.08497545e-01
2.48134863e-02 -6.87613487e-02 1.13878556e-01 -4.37572479e-01
7.52233386e-01 -1.34219730e+00 3.93247634e-01 8.53145719e-01
-4.97335121e-02 5.25794387e-01 1.12015557e+00 -6.26853526e-01
-1.12304568e+00 -6.82884276e-01 5.70166826e-01 -3.21792930e-01
-4.79148813e-02 -5.24052620e-01 -1.18846464e+00 5.24218500e-01
6.23466134e-01 -2.49004111e-01 2.16250911e-01 -3.85859549e-01
-3.54660869e-01 -1.40566796e-01 -1.23172736e+00 6.98884487e-01
6.34220541e-01 -2.05473661e-01 -1.79289103e-01 4.88328159e-01
6.40008688e-01 -6.72723651e-01 -2.31894076e-01 2.83425748e-01
7.33139038e-01 -1.82322323e+00 1.14911652e+00 -3.16539072e-02
4.16136354e-01 -6.49596870e-01 -2.98176676e-01 -1.27711201e+00
-1.20534673e-01 -7.03258157e-01 4.37502116e-01 1.19596887e+00
1.37483180e-01 -6.76479399e-01 5.66930652e-01 1.07322323e+00
-2.14130253e-01 -4.39393699e-01 -3.21262479e-01 -4.02065724e-01
-1.96990073e-01 -2.81686783e-01 3.23101401e-01 6.93064749e-01
-6.42842293e-01 9.28392112e-02 -1.44466653e-01 4.36480194e-01
8.18217874e-01 3.89903396e-01 1.20496428e+00 -1.08540928e+00
-6.99317396e-01 -2.73034513e-01 8.68133828e-02 -1.06031239e+00
-3.18449736e-01 -3.74783128e-01 3.20714027e-01 -1.92087603e+00
2.49851406e-01 -5.37335694e-01 4.66004431e-01 4.14471924e-01
-6.40885979e-02 4.41554099e-01 1.89669028e-01 1.90321386e-01
-3.04038320e-02 4.39026982e-01 1.17084539e+00 9.21045542e-02
-4.33953971e-01 -1.51309133e-01 -7.26284504e-01 1.06983304e+00
5.37329078e-01 -2.53142416e-01 -5.07002413e-01 -5.70653796e-01
3.41141284e-01 -1.49637088e-01 6.35057926e-01 -6.91255987e-01
-1.20645180e-01 -8.11958686e-02 3.49436462e-01 -5.04967153e-01
7.78902769e-01 -9.49717402e-01 6.30387127e-01 3.32878023e-01
5.10610752e-02 -3.19796175e-01 3.29121739e-01 3.00332427e-01
2.88032353e-01 -3.46745104e-01 1.20224977e+00 -3.50289904e-02
-1.08526528e-01 -2.72123188e-01 2.30423119e-02 -2.00140886e-02
7.38841772e-01 -4.66280133e-01 -4.04086083e-01 -7.85509288e-01
-2.39085272e-01 -2.02103406e-01 1.18593943e+00 5.67517020e-02
6.77236199e-01 -1.07241058e+00 -7.77321875e-01 4.40907180e-01
-1.40214354e-01 2.56490529e-01 1.70045570e-01 5.95007598e-01
-1.02460539e+00 -2.35598341e-01 1.98032975e-01 -7.27830410e-01
-1.57514989e+00 2.02529535e-01 6.97725832e-01 1.38244033e-01
-8.74645770e-01 6.42863810e-01 1.15562844e+00 -4.17329282e-01
1.60098016e-01 -3.23177457e-01 4.14534122e-01 -4.51379538e-01
6.98664486e-01 2.20637992e-01 2.74527341e-01 -5.97036779e-01
-8.80030319e-02 8.85149658e-01 3.01952571e-01 -4.52482730e-01
1.28674126e+00 -3.45456004e-01 -2.92351007e-01 3.27295095e-01
1.10790622e+00 5.49974799e-01 -1.56778038e+00 -8.64267051e-02
-6.78733349e-01 -1.11580205e+00 6.96937963e-02 -9.35738385e-01
-1.02437186e+00 7.78240323e-01 4.09068018e-01 1.71845928e-01
1.24656022e+00 -1.72293887e-01 6.35549843e-01 -5.00102341e-02
3.00729424e-01 -8.74072433e-01 2.74873346e-01 3.01181003e-02
1.27337480e+00 -9.79985118e-01 2.23992661e-01 -8.53697896e-01
-6.15332544e-01 8.77099097e-01 4.19592857e-01 1.52366422e-02
5.56630313e-01 4.84439045e-01 2.13856399e-01 -3.56722385e-01
-3.43402833e-01 -1.15420828e-02 4.34857547e-01 5.56843996e-01
2.92285085e-01 -1.92495972e-01 2.75295317e-01 -8.97159353e-02
-1.93080679e-01 -3.48646879e-01 6.91531122e-01 6.94940031e-01
-5.86601973e-01 -8.00765157e-01 -6.57731354e-01 1.19000152e-01
3.59872803e-02 -1.35371149e-01 -3.98849428e-01 5.12256742e-01
8.74161348e-02 1.01199114e+00 1.11223169e-01 1.86407909e-01
4.85950589e-01 -2.76228786e-01 7.18744636e-01 -6.95657134e-01
-3.46065760e-01 5.65933943e-01 -6.51788637e-02 -6.77234054e-01
-2.89539993e-01 -5.20681441e-01 -1.36420584e+00 -1.75958678e-01
-3.35206002e-01 -2.63223380e-01 9.96313632e-01 4.64418054e-01
2.50743866e-01 3.57373923e-01 7.88025081e-01 -1.28696549e+00
5.06270006e-02 -5.03934860e-01 -8.71200442e-01 8.20542216e-01
2.84842223e-01 -4.64721829e-01 -5.94399631e-01 3.68503034e-01]
|
[9.850878715515137, -3.0560662746429443]
|
5c84e56b-39d1-49f6-8b5e-8dbd65357a75
|
mask-editor-an-image-annotation-tool-for
|
1809.06461
| null |
http://arxiv.org/abs/1809.06461v1
|
http://arxiv.org/pdf/1809.06461v1.pdf
|
Mask Editor : an Image Annotation Tool for Image Segmentation Tasks
|
Deep convolutional neural network (DCNN) is the state-of-the-art method for
image segmentation, which is one of key challenging computer vision tasks.
However, DCNN requires a lot of training images with corresponding image masks
to get a good segmentation result. Image annotation software which is easy to
use and allows fast image mask generation is in great demand. To the best of
our knowledge, all existing image annotation software support only drawing
bounding polygons, bounding boxes, or bounding ellipses to mark target objects.
These existing software are inefficient when targeting objects that have
irregular shapes (e.g., defects in fabric images or tire images). In this paper
we design an easy-to-use image annotation software called Mask Editor for image
mask generation. Mask Editor allows drawing any bounding curve to mark objects
and improves efficiency to mark objects with irregular shapes. Mask Editor also
supports drawing bounding polygons, drawing bounding boxes, drawing bounding
ellipses, painting, erasing, super-pixel-marking, image cropping, multi-class
masks, mask loading, and mask modifying.
|
['Zhiyu Chen', 'Chuanhai Zhang', 'Kurt Loken', 'Zhiyong Xiao', 'Gary Kunkel']
|
2018-09-17
| null | null | null | null |
['image-cropping']
|
['computer-vision']
|
[ 5.18120170e-01 -1.09293222e-01 2.02780679e-01 -3.43184739e-01
1.26683384e-01 -6.43195570e-01 2.17316430e-02 -7.70624354e-02
-3.87740225e-01 3.46065253e-01 -6.37647927e-01 -7.31121540e-01
1.52868152e-01 -9.72776949e-01 -6.44300222e-01 -3.35945874e-01
4.10937726e-01 3.88968438e-01 7.50933170e-01 -1.43201351e-01
4.35415536e-01 1.05952919e+00 -1.52726483e+00 2.48873785e-01
8.35085928e-01 7.52124071e-01 5.06743073e-01 6.55231118e-01
-8.33290994e-01 1.05496801e-01 -9.81937885e-01 -1.57040030e-01
4.71094966e-01 -3.29986885e-02 -5.84907651e-01 4.60792184e-01
1.94664732e-01 -5.36752045e-01 1.23200960e-01 1.07936442e+00
2.73976505e-01 -2.48065069e-01 5.82042038e-01 -1.25254607e+00
-8.11698675e-01 4.64035034e-01 -7.73371994e-01 -8.29001293e-02
-2.87434608e-01 1.36246443e-01 4.59922738e-02 -5.68862498e-01
3.79708469e-01 9.38400030e-01 6.90122068e-01 5.29568017e-01
-8.85812938e-01 -8.20295274e-01 -2.24051625e-01 -3.90360355e-01
-1.27813816e+00 6.73021376e-02 8.02819788e-01 -5.85024238e-01
6.47749722e-01 6.06122136e-01 5.12653649e-01 7.94703215e-02
4.84120324e-02 5.91781974e-01 1.00126159e+00 -4.97727484e-01
-5.09529971e-02 2.33205050e-01 3.23524743e-01 7.51782358e-01
2.23511204e-01 -2.56717205e-01 2.52424687e-01 2.56936044e-01
1.46473801e+00 1.85361773e-01 -1.57253444e-01 -2.01673526e-02
-9.89634037e-01 5.69141865e-01 2.09165588e-01 2.75531352e-01
-1.54147685e-01 2.90624052e-01 4.13712233e-01 1.35934679e-03
1.87975153e-01 3.56961221e-01 -6.25439703e-01 8.82727504e-02
-9.71061230e-01 -1.21596316e-02 3.08832020e-01 1.12906313e+00
1.01336002e+00 2.09003329e-01 -1.75529957e-01 1.18411279e+00
3.05892155e-02 4.03317779e-01 2.34544009e-01 -9.90023017e-01
2.38178939e-01 1.03113246e+00 2.38136634e-01 -9.74103332e-01
-6.12313271e-01 -2.38205977e-02 -8.42134237e-01 6.79847002e-01
4.81396705e-01 -3.97576362e-01 -1.42636275e+00 9.43129957e-01
3.70055586e-01 -3.10037196e-01 -1.98681563e-01 7.31108367e-01
1.20300722e+00 9.02632296e-01 -7.37418458e-02 9.15077776e-02
1.62832940e+00 -1.02198839e+00 -8.50547016e-01 -3.13811094e-01
4.69779998e-01 -1.12335098e+00 1.38804877e+00 4.51953024e-01
-9.26488042e-01 -7.86432862e-01 -9.28176224e-01 -2.53974348e-01
-8.76762748e-01 7.91086018e-01 7.59531140e-01 8.34599614e-01
-5.95332861e-01 4.36914295e-01 -6.44263983e-01 -1.12167642e-01
5.10513842e-01 7.01457798e-01 -3.20903152e-01 2.41233096e-01
-6.38231874e-01 4.86655116e-01 6.57390058e-01 1.56879738e-01
-3.99556011e-01 -5.28206408e-01 -8.07852626e-01 8.92950818e-02
5.05730152e-01 8.24200436e-02 1.06155586e+00 -1.20762098e+00
-1.51510489e+00 9.63507235e-01 2.44206816e-01 -6.46089911e-02
4.25070733e-01 -1.33880377e-01 -3.41955870e-01 1.01089321e-01
-1.80397466e-01 1.05643153e+00 9.34526443e-01 -1.33872986e+00
-4.86283273e-01 -4.32018796e-03 1.07436851e-01 -2.13432740e-02
-1.25041336e-01 2.85813302e-01 -7.32292116e-01 -7.31283426e-01
-1.53266340e-02 -5.90203881e-01 -3.27864021e-01 3.00714642e-01
-7.51327813e-01 -2.67037034e-01 1.78892338e+00 -7.57290781e-01
1.42137921e+00 -2.18515062e+00 -6.15110338e-01 3.13739538e-01
-1.22910179e-01 8.02875340e-01 1.40181914e-01 1.80980712e-01
-1.95494175e-01 3.97571683e-01 -5.82145572e-01 -7.99310505e-02
5.28823351e-04 3.83118600e-01 2.20117986e-01 5.02077229e-02
2.50972420e-01 6.58009529e-01 -2.16807142e-01 -5.89239001e-01
6.52329683e-01 3.78952652e-01 -3.01684976e-01 6.36953786e-02
-4.37400967e-01 2.03760356e-01 -2.23736793e-01 7.71816850e-01
1.13454664e+00 -1.40704378e-01 -2.31591389e-01 -4.74343926e-01
-4.25297737e-01 -5.51798403e-01 -1.36276913e+00 1.13926864e+00
-1.76754519e-01 8.09851050e-01 2.04607785e-01 -8.29596639e-01
1.35604322e+00 2.03128885e-02 1.67095080e-01 -1.98166072e-01
4.23481315e-01 1.87891230e-01 -2.63150986e-02 -7.67624319e-01
5.79246938e-01 4.74219501e-01 1.64837480e-01 4.51817423e-01
-5.20051181e-01 -6.26972497e-01 5.02386153e-01 -1.00829884e-01
5.62247157e-01 5.86529337e-02 -5.59326150e-02 -2.35839352e-01
4.58968133e-01 1.59638643e-01 4.15271878e-01 3.87410998e-01
2.73837566e-01 1.00458431e+00 6.55446351e-01 -8.26744318e-01
-1.40784192e+00 -5.02371252e-01 -1.50334969e-01 7.00863898e-01
7.55864903e-02 1.16815113e-01 -1.32860184e+00 -3.91475111e-01
-2.57085562e-01 4.49204743e-01 -3.14536393e-01 3.98678482e-01
-7.34077871e-01 -5.62562585e-01 2.77729034e-01 5.62105238e-01
1.02469134e+00 -1.48976433e+00 -9.59480524e-01 -4.77747247e-02
4.47049409e-01 -9.75117981e-01 -7.35774934e-01 1.89214647e-01
-7.98011541e-01 -1.04787576e+00 -8.98541749e-01 -1.27920449e+00
1.49898648e+00 1.52048424e-01 8.40807736e-01 6.75009489e-01
-6.13452196e-01 -2.49665052e-01 -3.17195237e-01 -6.27669990e-01
-2.79319793e-01 -8.70489255e-02 -6.87985539e-01 -1.82705343e-01
6.35817200e-02 -8.51605237e-02 -5.51694632e-01 6.17067933e-01
-1.48821712e+00 5.57836533e-01 3.15337479e-01 4.13234770e-01
7.03067183e-01 4.14344430e-01 1.79929361e-01 -1.21708751e+00
5.96439600e-01 1.51062995e-01 -9.16609883e-01 2.62080967e-01
-1.64108992e-01 -2.49750912e-01 7.28680551e-01 -4.74581510e-01
-8.25175226e-01 2.48808607e-01 -2.40227252e-01 -2.61649340e-01
-5.78188300e-01 2.79960364e-01 -2.67693639e-01 -4.71355207e-02
2.76079863e-01 2.89885104e-02 1.73704624e-01 -5.83353162e-01
2.87106812e-01 9.46316123e-01 5.88286400e-01 -2.34771550e-01
6.26276493e-01 2.56019503e-01 -1.95845231e-01 -7.42524564e-01
-3.35956186e-01 -9.44822729e-02 -7.23031163e-01 -3.00426096e-01
1.14676619e+00 -2.20443934e-01 -5.74355960e-01 8.55545342e-01
-1.29911458e+00 -6.71016157e-01 -8.14920366e-02 -1.34028450e-01
2.06937660e-02 2.62806088e-01 -5.55824757e-01 -6.50446415e-01
-6.44267321e-01 -1.41351962e+00 8.67155850e-01 8.35629642e-01
3.62716895e-03 -5.71302772e-01 -7.41454363e-01 1.70901477e-01
3.99367303e-01 3.92088681e-01 1.08434641e+00 -3.19381028e-01
-6.05820417e-01 -2.51746476e-01 -5.98350048e-01 5.28115869e-01
2.90475696e-01 6.41846657e-01 -5.15865803e-01 2.52218187e-01
-6.69743538e-01 2.12794438e-01 5.00469685e-01 5.01810014e-01
2.06455374e+00 -9.82511714e-02 -5.18802702e-01 5.65995753e-01
1.36854231e+00 9.42430615e-01 1.06023383e+00 3.54457319e-01
8.28101754e-01 4.96940851e-01 4.40337300e-01 2.00265020e-01
-3.02949306e-02 4.48984474e-01 4.24497753e-01 -9.21462595e-01
-2.74778865e-02 3.03980052e-01 -2.91948467e-01 1.88625842e-01
4.53125201e-02 -2.78246373e-01 -9.77988303e-01 3.94964755e-01
-1.36231697e+00 -4.85065877e-01 -7.07639217e-01 1.90319228e+00
7.49182284e-01 2.46789426e-01 -4.03178595e-02 2.32448339e-01
1.05607867e+00 -3.30105275e-01 -3.48230273e-01 -7.44526982e-01
8.93110931e-02 4.92306352e-01 8.96475255e-01 2.74482936e-01
-1.14663994e+00 1.20335686e+00 5.97263050e+00 1.03008437e+00
-1.28449214e+00 -8.36600587e-02 8.93895566e-01 4.09840286e-01
-1.32120222e-01 -1.09637462e-01 -8.00939322e-01 5.56734204e-01
-7.33872503e-02 7.23362148e-01 4.07096326e-01 7.85942316e-01
2.57767022e-01 -4.83084798e-01 -5.38058162e-01 9.19359565e-01
-1.88191235e-01 -1.48270559e+00 -9.79449600e-02 -1.61818832e-01
7.53716350e-01 -4.74440575e-01 -1.16219074e-01 -8.25649723e-02
1.80010244e-01 -1.00634038e+00 6.27541244e-01 2.92150408e-01
9.73048806e-01 -8.58986676e-01 7.65678763e-01 2.96671297e-02
-9.38905180e-01 1.18290178e-01 -4.98555630e-01 2.38038853e-01
2.35193789e-01 7.43686914e-01 -6.18131578e-01 7.68930539e-02
6.38566792e-01 -9.38832667e-03 -6.28736019e-01 1.17817783e+00
-6.31585717e-02 3.44650894e-01 -5.18395245e-01 -4.25180383e-02
3.09358895e-01 -4.45997685e-01 -7.23020658e-02 1.30311823e+00
2.91025519e-01 1.15627674e-02 2.42600530e-01 9.73502636e-01
-3.82655254e-03 1.69063285e-01 -2.07523867e-01 -1.89894333e-01
5.12617886e-01 1.38758278e+00 -1.60564947e+00 -3.93850029e-01
-2.86018014e-01 9.86059487e-01 -2.29283288e-01 3.42167109e-01
-1.00371516e+00 -1.11485612e+00 7.78956860e-02 3.11005026e-01
4.26937044e-01 -5.36964893e-01 -5.93989372e-01 -2.41054535e-01
-1.82863027e-01 -7.37024486e-01 2.74291467e-02 -1.10996723e+00
-6.68745458e-01 3.60641867e-01 1.63970277e-01 -9.39975679e-01
6.06554389e-01 -1.04464018e+00 -1.00006843e+00 7.49829292e-01
-1.11105394e+00 -1.16718435e+00 -6.77612424e-01 4.19514358e-01
7.63124645e-01 -8.91981199e-02 4.44648951e-01 5.07893324e-01
-8.56102049e-01 1.94549054e-01 1.46403253e-01 6.26890779e-01
2.77827740e-01 -1.10152543e+00 5.23780704e-01 5.97666204e-01
-2.61255383e-01 3.97754490e-01 3.82318556e-01 -7.50820637e-01
-1.00594306e+00 -9.31145191e-01 2.07023486e-01 1.41231522e-01
2.76616346e-02 -4.67260838e-01 -9.54550207e-01 5.02491295e-01
3.18998218e-01 1.73111796e-01 4.19676483e-01 -6.21599138e-01
2.14471653e-01 -1.86806455e-01 -1.32024002e+00 7.02330291e-01
4.84739959e-01 5.76274432e-02 9.02033448e-02 5.69868326e-01
7.01849401e-01 -8.18421543e-01 -5.42337835e-01 4.74212915e-01
5.68965673e-01 -8.82866025e-01 9.05557632e-01 -1.63281187e-01
3.18519175e-01 -8.80800843e-01 4.21011806e-01 -7.37185419e-01
1.00385696e-01 -5.32175124e-01 5.59685230e-01 1.37345970e+00
5.80035269e-01 -4.92016762e-01 8.33315134e-01 7.18007624e-01
-5.33482134e-01 -7.15046883e-01 -2.76616544e-01 -5.14745355e-01
-2.81379163e-01 -4.00283217e-01 9.09361541e-01 8.50641906e-01
-6.39991283e-01 -4.59223896e-01 -1.35837406e-01 1.98088914e-01
1.51092201e-01 -1.01283580e-01 9.38788295e-01 -1.06014907e+00
-3.79896648e-02 -6.15415454e-01 2.87371948e-02 -8.82796943e-01
-3.00615370e-01 -2.98238486e-01 -1.51800886e-01 -1.97995615e+00
-3.09265137e-01 -8.95297289e-01 5.47032654e-01 8.12321544e-01
4.13517095e-02 6.19524896e-01 2.53885090e-01 -2.60984488e-02
-2.51021795e-02 -1.67173922e-01 1.60604000e+00 -1.54082432e-01
-3.86610478e-01 -1.42784104e-01 -3.36113662e-01 9.03651834e-01
1.18281066e+00 -3.94797921e-01 -3.29458117e-01 -6.84586227e-01
-2.03023218e-02 -4.70580012e-01 3.02626967e-01 -8.76933694e-01
1.78776570e-02 -3.06322724e-01 7.76890814e-01 -1.06583583e+00
7.20514506e-02 -9.88126278e-01 4.47740197e-01 4.63013381e-01
3.14723910e-03 1.53923720e-01 5.58901668e-01 -2.47189090e-01
1.84302688e-01 -8.57863545e-01 9.70043898e-01 -6.06376886e-01
-7.26925075e-01 2.16387957e-01 -5.18395007e-01 -4.62438256e-01
1.39000678e+00 -9.44009602e-01 -3.62471849e-01 2.96869446e-02
-6.37169540e-01 1.99679285e-01 5.06718993e-01 1.79976821e-01
5.61632991e-01 -1.03934264e+00 -2.49637008e-01 4.94345486e-01
-2.41898775e-01 5.65180898e-01 3.49515438e-01 4.49308902e-01
-1.63145697e+00 9.15024951e-02 -2.83494860e-01 -4.00406718e-01
-1.56205320e+00 5.19330740e-01 2.71567881e-01 1.96880862e-01
-6.39599025e-01 8.72015655e-01 1.78337187e-01 -2.80286640e-01
1.12885766e-01 -7.02694237e-01 -2.79183656e-01 -2.88190007e-01
5.15995383e-01 4.53585356e-01 -2.65708603e-02 -1.10563241e-01
4.02399041e-02 9.37106073e-01 -1.14126794e-01 2.95439065e-01
9.57215726e-01 1.30368963e-01 -3.76235187e-01 -8.63216445e-02
7.27261305e-01 3.01493891e-02 -1.37476087e+00 3.36656839e-01
-3.43662620e-01 -4.02183414e-01 3.56336832e-02 -9.48718846e-01
-1.34107912e+00 9.35574412e-01 7.55876601e-01 6.70712769e-01
1.13469481e+00 -2.71833867e-01 8.51323664e-01 8.77950415e-02
1.44529507e-01 -1.35677397e+00 -3.29583511e-02 3.26262295e-01
8.87683272e-01 -8.81771922e-01 -6.68381326e-05 -7.94602811e-01
-5.26049972e-01 1.53948414e+00 9.85496223e-01 -1.44428238e-01
5.88730156e-01 7.09862649e-01 2.37340018e-01 -2.84136951e-01
1.54712170e-01 -1.12549372e-01 2.64559567e-01 7.58664072e-01
3.37555796e-01 1.00605696e-01 -5.19343436e-01 2.94762135e-01
-1.52128026e-01 -1.23979568e-01 6.09611034e-01 8.85743737e-01
-7.87103891e-01 -1.20657682e+00 -8.02836716e-01 7.02172101e-01
-6.64908230e-01 -1.08324297e-01 -3.20304722e-01 9.86725688e-01
6.04916215e-01 7.47754574e-01 4.90770012e-01 2.44970080e-02
-1.73307881e-02 -2.42644534e-01 1.85263678e-01 -7.22108305e-01
-6.38662636e-01 9.07455385e-02 -4.64890525e-02 2.52679698e-02
-1.88272148e-01 -1.49662912e-01 -1.67949736e+00 -4.90113869e-02
-5.25571704e-01 -1.27981007e-01 1.01154399e+00 7.23530054e-01
3.15119743e-01 5.99800825e-01 1.04769751e-01 -7.92860389e-01
3.58631462e-01 -9.13005590e-01 -4.91694570e-01 1.14173107e-01
-7.09677637e-02 -4.22753423e-01 2.06644967e-01 5.20735741e-01]
|
[9.624958038330078, -0.041415661573410034]
|
abb1bb15-2009-47eb-b434-25546b2970a8
|
gans-n-roses-stable-controllable-diverse
|
2106.06561
| null |
https://arxiv.org/abs/2106.06561v1
|
https://arxiv.org/pdf/2106.06561v1.pdf
|
GANs N' Roses: Stable, Controllable, Diverse Image to Image Translation (works for videos too!)
|
We show how to learn a map that takes a content code, derived from a face image, and a randomly chosen style code to an anime image. We derive an adversarial loss from our simple and effective definitions of style and content. This adversarial loss guarantees the map is diverse -- a very wide range of anime can be produced from a single content code. Under plausible assumptions, the map is not just diverse, but also correctly represents the probability of an anime, conditioned on an input face. In contrast, current multimodal generation procedures cannot capture the complex styles that appear in anime. Extensive quantitative experiments support the idea the map is correct. Extensive qualitative results show that the method can generate a much more diverse range of styles than SOTA comparisons. Finally, we show that our formalization of content and style allows us to perform video to video translation without ever training on videos.
|
['David Forsyth', 'Min Jin Chong']
|
2021-06-11
| null | null | null | null |
['multimodal-generation']
|
['natural-language-processing']
|
[ 5.69156289e-01 2.70107746e-01 1.34866595e-01 -5.20929158e-01
-7.74185956e-01 -9.78288770e-01 7.78347611e-01 -8.72743547e-01
-1.46672633e-02 9.27123189e-01 1.75968304e-01 1.39835432e-01
3.74381781e-01 -7.32663155e-01 -1.13194203e+00 -5.79566300e-01
-1.56701375e-02 5.95289052e-01 -2.17147484e-01 -3.35296333e-01
-5.92315197e-03 4.52388883e-01 -1.49946165e+00 4.03078139e-01
5.09820998e-01 8.08097064e-01 -3.98750082e-02 1.10864246e+00
-4.10652021e-03 7.12266386e-01 -7.94736087e-01 -1.01570606e+00
6.46963239e-01 -1.16069651e+00 -6.77368164e-01 4.38835263e-01
1.18889761e+00 -5.50659537e-01 -3.77154112e-01 1.17041826e+00
2.39570022e-01 -2.11954311e-01 8.82408679e-01 -1.63693857e+00
-8.88950169e-01 4.66263503e-01 -2.66938359e-01 -4.45147842e-01
7.58507669e-01 4.15389717e-01 9.07668293e-01 -7.56879628e-01
1.09148872e+00 1.48794019e+00 5.89897633e-01 1.09269810e+00
-1.37576449e+00 -7.43859708e-01 -1.21011831e-01 -4.74457026e-01
-1.18088651e+00 -5.07076204e-01 7.24382699e-01 -3.47124159e-01
8.89182091e-02 3.15994471e-01 9.22711253e-01 1.60662675e+00
2.56504774e-01 6.43465340e-01 1.11832201e+00 -2.47315273e-01
1.06921166e-01 2.86881715e-01 -7.45011628e-01 9.34046805e-01
1.78539194e-02 1.68040574e-01 -6.29626036e-01 -1.74687386e-01
1.02729011e+00 -3.71631473e-01 -3.31649661e-01 -4.97106433e-01
-1.20436120e+00 8.78929675e-01 2.12133884e-01 -1.98022142e-01
1.34408567e-02 4.66280788e-01 1.50404066e-01 6.65739715e-01
3.91435483e-03 5.41642308e-01 4.86422330e-03 -1.83923379e-01
-1.01119924e+00 5.96282363e-01 1.05788863e+00 1.41205823e+00
6.90867722e-01 3.93906057e-01 -6.42946959e-02 5.25838435e-01
4.93839756e-02 1.22383952e+00 -4.95679304e-02 -1.60101247e+00
1.44085899e-01 -4.29549925e-02 2.40269050e-01 -9.32749271e-01
2.50532359e-01 -2.05036346e-02 -6.28571332e-01 6.00099027e-01
5.99841893e-01 -4.28769857e-01 -9.38620448e-01 2.26748538e+00
-5.35565801e-02 -4.39044051e-02 -1.54287711e-01 6.88654482e-01
3.55197847e-01 6.19007945e-01 -2.52238005e-01 -8.36877525e-02
9.00906026e-01 -5.20308018e-01 -6.37553275e-01 -1.35696605e-01
-1.87719725e-02 -1.02149701e+00 1.08609140e+00 3.38432580e-01
-1.62930620e+00 -4.45113152e-01 -1.11914361e+00 2.12613463e-01
-4.02033776e-02 -2.02583686e-01 5.73824108e-01 9.32518780e-01
-1.51449418e+00 5.61136425e-01 -2.92216927e-01 -4.49066281e-01
3.47502053e-01 4.06457305e-01 -6.72159195e-01 -4.26022746e-02
-1.06549549e+00 7.87826777e-01 1.24867216e-01 -3.58289570e-01
-1.14464927e+00 -5.11507332e-01 -9.37888861e-01 -2.22452492e-01
-1.18131757e-01 -1.10066783e+00 1.25293303e+00 -1.88032973e+00
-1.58054483e+00 1.13912952e+00 -2.14248359e-01 -2.87235945e-01
9.40950751e-01 -6.40535280e-02 -3.61691862e-01 5.35667479e-01
5.58644608e-02 1.38230884e+00 1.22751641e+00 -1.67350352e+00
-4.89086092e-01 1.56322926e-01 3.21817666e-01 3.10112000e-01
-2.30116010e-01 -1.82035193e-02 -4.56157327e-01 -9.29480731e-01
-3.38109016e-01 -1.21210241e+00 4.02642012e-01 7.15692997e-01
-4.59664702e-01 4.98727530e-01 7.77190804e-01 -5.06451666e-01
7.47003853e-01 -2.23321271e+00 4.54544067e-01 2.65093654e-01
3.95872146e-02 -1.71101928e-01 -4.38275158e-01 2.73814470e-01
-6.87262788e-02 3.92658681e-01 -4.57825392e-01 -2.60532439e-01
1.75820678e-01 2.52824396e-01 -5.43259144e-01 4.27965194e-01
4.22821224e-01 8.41795504e-01 -8.68392885e-01 -6.48693085e-01
-5.19255102e-02 5.02839863e-01 -9.99332309e-01 3.81431878e-01
-3.79429251e-01 3.92573655e-01 -5.69873005e-02 6.03506744e-01
7.22514451e-01 1.50779858e-02 1.00089066e-01 3.94160077e-02
4.87615675e-01 -4.04560626e-01 -9.47116911e-01 1.68277597e+00
-2.54102975e-01 8.59535158e-01 3.25135499e-01 -2.82001019e-01
7.56896615e-01 2.70912200e-01 3.36728245e-01 -2.14700520e-01
2.83415206e-02 2.89130181e-01 -1.70197278e-01 -2.13037968e-01
5.33406436e-01 -7.18197882e-01 -2.63240963e-01 4.78579134e-01
2.89779842e-01 -7.12994218e-01 9.76497978e-02 4.48294610e-01
9.65498626e-01 4.89996016e-01 -2.11254865e-01 -2.61319101e-01
2.74587840e-01 -1.39894888e-01 5.06232560e-01 7.62407064e-01
-1.80979192e-01 1.00364017e+00 7.29561329e-01 -1.83032826e-01
-1.47141290e+00 -1.55325794e+00 -2.66430546e-02 6.38837457e-01
2.55041063e-01 -1.21885709e-01 -1.07221723e+00 -7.33794808e-01
1.08446414e-02 5.45822978e-01 -8.94483209e-01 -1.11572012e-01
-6.12703800e-01 -1.23934701e-01 8.51946592e-01 2.05927283e-01
4.78660703e-01 -1.05808401e+00 -1.16667882e-01 -2.31069252e-01
-2.16510281e-01 -1.03958416e+00 -9.53983307e-01 -3.46301705e-01
-4.88202482e-01 -8.72968614e-01 -9.27811921e-01 -9.83856380e-01
7.41455078e-01 -6.19809441e-02 1.43075514e+00 -2.12630462e-02
-3.30287069e-02 6.61149383e-01 -1.38321981e-01 -2.24843487e-01
-1.07170439e+00 -3.32567245e-01 6.93421289e-02 6.66900352e-02
-5.34212962e-02 -6.16418004e-01 -5.01140833e-01 3.41573119e-01
-1.03089571e+00 1.82670593e-01 2.98452169e-01 9.43386257e-01
2.69000888e-01 -2.31150344e-01 4.63626832e-01 -9.71975088e-01
3.12942833e-01 -3.16725463e-01 -3.80156785e-01 3.21521401e-01
-7.16887414e-02 9.94671136e-02 7.72626281e-01 -5.48961043e-01
-1.06320310e+00 1.43381894e-01 -3.55892219e-02 -6.05100334e-01
-1.33722737e-01 -2.66599149e-01 -3.87598693e-01 -2.16744334e-01
6.79563940e-01 1.54229611e-01 3.16755027e-01 1.59581065e-01
6.56836748e-01 2.98704565e-01 7.70673394e-01 -9.34004307e-01
1.24357069e+00 6.44491792e-01 8.98778290e-02 -6.18204474e-01
-3.71487647e-01 4.69634324e-01 -2.82558262e-01 -3.12209189e-01
8.90069723e-01 -9.84865069e-01 -5.11810064e-01 5.09309292e-01
-1.17210364e+00 -4.28664833e-01 -4.20699447e-01 9.74876583e-02
-1.22682202e+00 1.75537989e-01 -6.61772728e-01 -5.54111958e-01
1.18894577e-01 -1.10165477e+00 1.13065708e+00 1.42051145e-01
-4.56575036e-01 -1.05133498e+00 -9.07422081e-02 -1.31916385e-02
3.62596333e-01 6.20927155e-01 7.45049417e-01 -1.15967818e-01
-7.06382275e-01 -8.90130848e-02 -9.10066068e-03 4.02633995e-01
1.77659795e-01 4.11390305e-01 -8.35839093e-01 -3.45828056e-01
-1.60280421e-01 -6.41656816e-01 5.27421534e-01 1.68587700e-01
8.28058302e-01 -5.16468585e-01 7.25250468e-02 9.39194500e-01
1.51328707e+00 2.72115856e-01 7.79469371e-01 -1.36479586e-01
6.84884071e-01 5.53782940e-01 3.28453064e-01 2.68472075e-01
-4.04669791e-02 6.38864994e-01 3.16129863e-01 -4.85000610e-02
-3.08897227e-01 -4.93386418e-01 8.20790350e-01 3.65068018e-01
5.11543117e-02 -3.97022069e-01 -3.66059184e-01 3.85818571e-01
-1.39271545e+00 -1.44717646e+00 4.26120311e-01 2.06305742e+00
1.17267227e+00 1.42235056e-01 3.35226446e-01 -2.63565749e-01
8.75360966e-01 8.85773152e-02 -4.41907555e-01 -7.48048544e-01
-5.07692814e-01 2.59838641e-01 3.92452091e-01 6.62454247e-01
-1.00974417e+00 8.94577801e-01 8.10024071e+00 6.86813533e-01
-9.72665668e-01 -2.17716634e-01 7.23433673e-01 -3.63097519e-01
-9.71810102e-01 -2.17214212e-01 -4.73986715e-01 6.59902453e-01
7.16953993e-01 -3.47700983e-01 8.93816769e-01 7.75841653e-01
-1.36632800e-01 3.01517725e-01 -1.54209340e+00 9.74629164e-01
4.11859632e-01 -1.36468792e+00 4.83853877e-01 1.27149716e-01
1.15102506e+00 -6.32056534e-01 4.18622762e-01 1.63267106e-01
6.59920037e-01 -1.16870236e+00 1.23327124e+00 5.39604664e-01
1.26588225e+00 -8.94161522e-01 2.83438802e-01 -1.87466338e-01
-8.44688296e-01 1.16915338e-01 -1.42844096e-01 3.59543085e-01
1.59829959e-01 1.96851507e-01 -3.28297496e-01 2.61816561e-01
2.31739953e-01 5.45064867e-01 -4.71876651e-01 7.01501727e-01
-2.24756181e-01 2.79050767e-01 -1.02048129e-01 2.82446027e-01
3.86351682e-02 -1.87385619e-01 7.86611736e-01 1.25370538e+00
5.90143025e-01 -1.38443410e-01 7.68981725e-02 1.15094161e+00
-4.20435011e-01 -3.14226359e-01 -1.15856206e+00 -1.47708878e-01
4.05664146e-01 9.60993886e-01 -3.76083702e-01 -4.28048998e-01
-2.12001488e-01 1.52850342e+00 -8.70996267e-02 5.30938983e-01
-1.12681031e+00 -4.13250804e-01 9.36681151e-01 -6.66082092e-03
2.16017857e-01 -3.38071957e-02 -3.76046479e-01 -1.28522551e+00
-5.20662032e-02 -1.35309196e+00 -7.57450536e-02 -1.14848137e+00
-1.34791458e+00 7.12528825e-01 8.17720369e-02 -1.36730528e+00
-6.02814317e-01 -6.27941668e-01 -6.80028737e-01 8.21462870e-01
-8.17574203e-01 -1.23918664e+00 -1.66017950e-01 7.48585641e-01
4.67325121e-01 -4.37187582e-01 8.50565195e-01 6.63969740e-02
-1.24792069e-01 1.00379324e+00 -2.23336250e-01 2.43673712e-01
9.11150277e-01 -1.41806734e+00 3.94672841e-01 9.13200080e-01
4.83855978e-02 5.73266029e-01 1.00664783e+00 -5.39435208e-01
-1.48306763e+00 -9.27206814e-01 5.48992336e-01 -7.24251747e-01
5.68935156e-01 -4.13011074e-01 -3.99246514e-01 8.09000194e-01
6.21467710e-01 -1.90820143e-01 4.29430872e-01 -4.13739026e-01
-6.55042171e-01 7.68566132e-02 -1.51948798e+00 9.32558537e-01
1.29536617e+00 -7.35747933e-01 -4.07854199e-01 1.11891501e-01
6.78736925e-01 -4.03285235e-01 -7.26424396e-01 2.52751261e-01
8.06531727e-01 -1.16481304e+00 9.23478425e-01 -5.55469751e-01
9.69122112e-01 -2.65043944e-01 -4.58027989e-01 -1.27555513e+00
-1.00340247e-01 -9.48585451e-01 -2.99859662e-02 1.24960041e+00
3.48968118e-01 -2.62433976e-01 7.12129951e-01 7.74873674e-01
7.27063939e-02 -3.76180917e-01 -6.07004762e-01 -9.09398377e-01
4.33476776e-01 -6.59160018e-02 7.08256841e-01 8.30059767e-01
-2.00938612e-01 9.24768597e-02 -6.80352926e-01 -1.50885716e-01
8.09336603e-01 5.41363880e-02 9.01002645e-01 -8.16474378e-01
-5.53876698e-01 -5.99710763e-01 -5.08265853e-01 -7.21462548e-01
3.91431481e-01 -6.91870868e-01 1.13539234e-01 -1.14325202e+00
3.93997014e-01 -3.29217970e-01 1.55333146e-01 1.88492239e-01
-5.18071502e-02 7.44782567e-01 5.77393532e-01 5.49556836e-02
-2.82586038e-01 3.89442474e-01 1.65653598e+00 -2.05017045e-01
2.59013504e-01 -3.09441447e-01 -8.12071383e-01 7.66125917e-01
7.09534705e-01 -2.45164812e-01 -6.23135328e-01 -3.91938895e-01
3.54247630e-01 2.03303844e-01 5.37782133e-01 -9.31602716e-01
-3.60557228e-01 -3.72404248e-01 6.61119938e-01 4.38036025e-02
4.80334759e-01 -6.60491943e-01 5.80483735e-01 2.42140338e-01
-4.98593956e-01 7.72276074e-02 1.58085957e-01 4.10516083e-01
-9.99488235e-02 -8.08161572e-02 1.08249986e+00 -2.03815475e-01
-4.47797328e-01 3.29679310e-01 -2.77357191e-01 3.15804720e-01
8.63347948e-01 -4.30356145e-01 -2.22457275e-01 -9.97297585e-01
-7.73120522e-01 -1.47871360e-01 1.24883175e+00 3.53667706e-01
6.21141732e-01 -1.90106857e+00 -8.23873699e-01 3.97363305e-01
-1.94490254e-02 -4.91979778e-01 -1.06721982e-01 1.75782084e-01
-7.92663753e-01 -2.85752445e-01 -6.82597399e-01 -4.34083670e-01
-1.15504146e+00 4.03327137e-01 4.46610600e-01 3.23249638e-01
-4.07947302e-01 8.96671295e-01 5.70663095e-01 -1.62065178e-01
1.69313952e-01 2.15771705e-01 4.78436053e-01 -2.18506739e-01
4.85612571e-01 -2.75535379e-02 -6.63411796e-01 -8.83067489e-01
-2.02240869e-01 7.50982106e-01 2.64839798e-01 -7.93494642e-01
8.17648053e-01 -1.44446552e-01 2.65867244e-02 2.42029026e-01
1.31172252e+00 3.06549579e-01 -1.70914054e+00 3.24926794e-01
-6.90005243e-01 -7.69320726e-01 -6.45086825e-01 -7.80791223e-01
-1.04924083e+00 6.88774228e-01 4.55085874e-01 7.19076991e-02
1.19335115e+00 -2.39523828e-01 8.24683011e-01 2.48353630e-01
5.51184535e-01 -8.73231471e-01 3.00319284e-01 2.54400730e-01
1.17871296e+00 -1.00125206e+00 -3.04230124e-01 -3.03933233e-01
-9.53178287e-01 1.00811851e+00 5.44891596e-01 -2.83743024e-01
3.18612874e-01 4.06548351e-01 1.86555773e-01 1.87740415e-01
-6.00685239e-01 9.39985216e-02 2.19681293e-01 8.29289734e-01
3.02974582e-01 6.53175041e-02 1.57486886e-01 -1.28883794e-01
-6.02912605e-01 -4.56625111e-02 7.75891721e-01 7.13948548e-01
-3.07669491e-01 -1.24121547e+00 -5.36544979e-01 1.79270670e-01
-4.66002792e-01 7.22704083e-02 -7.62043834e-01 7.84384787e-01
2.78489798e-01 8.24259102e-01 1.67848662e-01 -6.11016691e-01
-8.35717991e-02 1.21324360e-01 1.05940402e+00 -3.48796219e-01
-4.24047261e-01 -7.47194663e-02 6.09829314e-02 -6.18120372e-01
-4.04189885e-01 -6.51179194e-01 -9.65301156e-01 -8.45505714e-01
1.05641529e-01 -1.07581273e-01 4.39482749e-01 4.63395387e-01
6.85073286e-02 8.46226588e-02 9.96999860e-01 -9.29567873e-01
-3.20101649e-01 -4.34169799e-01 -6.01737976e-01 9.44392264e-01
3.68615508e-01 -4.68602479e-01 -5.66897511e-01 6.55878842e-01]
|
[11.7262601852417, -0.3665771484375]
|
1e748cdd-3ed1-43af-8f9e-3ccbdec037f4
|
private-multi-winner-voting-for-machine-1
|
2211.1541
| null |
https://arxiv.org/abs/2211.15410v1
|
https://arxiv.org/pdf/2211.15410v1.pdf
|
Private Multi-Winner Voting for Machine Learning
|
Private multi-winner voting is the task of revealing $k$-hot binary vectors satisfying a bounded differential privacy (DP) guarantee. This task has been understudied in machine learning literature despite its prevalence in many domains such as healthcare. We propose three new DP multi-winner mechanisms: Binary, $\tau$, and Powerset voting. Binary voting operates independently per label through composition. $\tau$ voting bounds votes optimally in their $\ell_2$ norm for tight data-independent guarantees. Powerset voting operates over the entire binary vector by viewing the possible outcomes as a power set. Our theoretical and empirical analysis shows that Binary voting can be a competitive mechanism on many tasks unless there are strong correlations between labels, in which case Powerset voting outperforms it. We use our mechanisms to enable privacy-preserving multi-label learning in the central setting by extending the canonical single-label technique: PATE. We find that our techniques outperform current state-of-the-art approaches on large, real-world healthcare data and standard multi-label benchmarks. We further enable multi-label confidential and private collaborative (CaPC) learning and show that model performance can be significantly improved in the multi-site setting.
|
['Xiao Wang', 'Nicolas Papernot', 'Somesh Jha', 'Muhammad Ahmad Kaleem', 'Ali Shahin Shamsabadi', 'Vinith Menon Suriyakumar', 'Natalie Dullerud', 'Christopher A Choquette-Choo', 'Adam Dziedzic']
|
2022-11-23
|
private-multi-winner-voting-for-machine
|
https://openreview.net/forum?id=JedTK_aOaRa
|
https://openreview.net/pdf?id=JedTK_aOaRa
| null |
['multi-label-learning']
|
['methodology']
|
[ 5.10194004e-01 3.36050659e-01 -7.75363803e-01 -8.39341700e-01
-1.48249245e+00 -9.76734221e-01 3.00491899e-01 3.24865401e-01
-7.87821770e-01 9.01753008e-01 1.20337784e-01 -5.39572656e-01
-1.35454506e-01 -4.93337959e-01 -8.00587296e-01 -1.15161264e+00
-1.42278299e-01 4.13080245e-01 -3.55130613e-01 3.97997558e-01
-1.53075531e-01 1.73822194e-02 -8.57825518e-01 6.26516104e-01
2.11507380e-01 1.10038030e+00 -8.53323817e-01 5.53525269e-01
2.97311395e-01 7.75220454e-01 -2.34443754e-01 -1.03019440e+00
9.24096584e-01 -1.51132077e-01 -8.38743210e-01 -4.99683291e-01
5.90482414e-01 -2.89453089e-01 -3.96124363e-01 1.19422197e+00
7.07324266e-01 -3.57577980e-01 5.79644144e-01 -1.37027574e+00
-5.28392613e-01 9.12409365e-01 -6.96495116e-01 -4.04566258e-01
5.18022776e-02 -7.65759125e-02 1.49108422e+00 -3.82001579e-01
7.73061574e-01 9.83262599e-01 1.02074969e+00 7.40585148e-01
-1.65484738e+00 -1.29031289e+00 -1.44780010e-01 -1.44265726e-01
-1.31644499e+00 -5.12401700e-01 4.11805451e-01 -2.31498703e-01
5.63699126e-01 7.29988277e-01 6.41883835e-02 9.97960389e-01
2.00466901e-01 8.85531187e-01 1.79425108e+00 -3.58212441e-01
3.07590276e-01 2.64910251e-01 3.26722354e-01 7.93543220e-01
5.01307011e-01 1.55723512e-01 -6.31958723e-01 -1.14937770e+00
-4.02706340e-02 2.90850013e-01 -3.94284338e-01 -5.94633281e-01
-1.04723871e+00 1.09300971e+00 8.08690116e-02 -8.96811709e-02
2.85001043e-02 6.08277380e-01 6.54777586e-01 6.58599854e-01
5.54654837e-01 2.44611487e-01 -8.39889884e-01 2.70128965e-01
-8.78754139e-01 4.69109833e-01 1.06727004e+00 1.03162575e+00
6.42095685e-01 -9.25101042e-01 -4.19532299e-01 6.03704870e-01
1.20074116e-01 6.67539060e-01 9.16946754e-02 -1.34831858e+00
5.19270301e-01 3.05150092e-01 1.99690595e-01 -6.92023993e-01
-3.89837265e-01 -2.15201154e-01 -1.08669078e+00 4.61165346e-02
5.05249619e-01 -4.04994756e-01 -5.08717060e-01 2.09518671e+00
5.16498744e-01 -2.75586545e-01 -5.69468886e-02 4.36385006e-01
3.62641901e-01 2.76258409e-01 3.52453023e-01 -4.68515515e-01
1.73095906e+00 -6.04463160e-01 -6.31454766e-01 1.70152064e-03
1.14217222e+00 -3.33885550e-01 3.40034634e-01 3.98112893e-01
-6.77281260e-01 5.09177685e-01 -6.27541900e-01 -1.73762500e-01
-2.81131953e-01 -3.75529170e-01 1.10999048e+00 1.32420015e+00
-9.87766743e-01 3.89756590e-01 -5.43670595e-01 1.57748923e-01
1.06026220e+00 6.03604615e-01 -8.40082526e-01 -4.92065549e-01
-1.12999594e+00 3.77207637e-01 -3.24808247e-03 -6.34219825e-01
-5.54432094e-01 -9.34709907e-01 -7.52568960e-01 -3.22631836e-01
3.15391213e-01 -7.28862226e-01 1.32969058e+00 -5.14343619e-01
-8.56384039e-01 1.43514001e+00 -4.91381511e-02 -5.89683294e-01
8.94241154e-01 3.81075770e-01 -2.26614736e-02 -3.67520005e-02
1.56950429e-01 4.35642689e-01 4.63705838e-01 -1.19396484e+00
-1.02293777e+00 -6.97154403e-01 -5.02975732e-02 3.52185423e-04
-1.69705376e-01 -4.73177880e-02 -6.61058873e-02 -5.77369392e-01
3.59881409e-02 -1.30028784e+00 -4.68230277e-01 3.32230300e-01
-6.34057045e-01 -2.84433633e-01 6.88003063e-01 -3.58125657e-01
9.60412443e-01 -2.16387463e+00 -4.26816285e-01 4.60973471e-01
5.59337616e-01 -1.98471501e-01 2.01448560e-01 1.05575651e-01
1.17456369e-01 2.72461593e-01 -4.65019852e-01 -9.19367194e-01
3.15893054e-01 2.48011276e-01 -2.25875512e-01 1.19284952e+00
-7.39668071e-01 1.07977068e+00 -7.92724192e-01 -6.18903399e-01
-4.19903010e-01 9.72448662e-02 -7.10113406e-01 -2.92868853e-01
-1.59528434e-01 2.51851350e-01 -3.83822620e-01 8.66477668e-01
9.26750481e-01 -5.75099051e-01 6.66427076e-01 9.25058573e-02
4.57126588e-01 -1.12049498e-01 -1.11892641e+00 1.58966947e+00
-1.49772123e-01 1.19791515e-01 5.08973658e-01 -7.75386810e-01
2.91121036e-01 5.44422328e-01 7.25290537e-01 -5.27705193e-01
1.62381098e-01 3.29235911e-01 -5.61160862e-01 -7.70335346e-02
2.50304699e-01 -6.19801998e-01 -7.48384058e-01 1.02952349e+00
-1.69910222e-01 3.10946852e-01 -6.61674678e-01 2.39572778e-01
1.47608709e+00 -7.07150578e-01 4.08444107e-01 -3.74945998e-01
-5.71617708e-02 -2.86275893e-01 8.72025311e-01 1.22809553e+00
-4.33555633e-01 4.45074260e-01 5.52337229e-01 -3.63675058e-01
-8.84165823e-01 -5.92082083e-01 -3.60474557e-01 1.46565211e+00
-5.51986173e-02 -2.20905140e-01 -4.79499608e-01 -1.42997408e+00
6.43757522e-01 4.15814966e-01 -8.88381660e-01 2.63575464e-01
-2.05723643e-01 -1.04423344e+00 8.28391612e-01 1.12262450e-01
2.85341769e-01 -5.78930557e-01 -3.54638904e-01 9.53625217e-02
-1.61444023e-01 -8.09504747e-01 -9.58160341e-01 6.40325725e-01
-4.67649519e-01 -1.32008719e+00 -6.87541068e-01 -4.70681012e-01
6.22722208e-01 1.70613751e-01 8.90004337e-01 -2.00132459e-01
-2.65942365e-01 5.11016846e-01 -1.71891842e-02 -3.87841642e-01
-3.85303944e-01 8.13575163e-02 -4.94131073e-02 2.55868971e-01
4.41731751e-01 -3.27940494e-01 -1.05585325e+00 2.66753733e-01
-8.32744896e-01 -3.37184936e-01 4.32959199e-01 9.39448714e-01
9.41047549e-01 -1.35112196e-01 3.29398602e-01 -1.90071678e+00
4.52105045e-01 -6.45576715e-01 -5.50472498e-01 4.26150888e-01
-1.19578683e+00 2.40279347e-01 5.07739365e-01 -2.77701497e-01
-5.85644484e-01 3.33998829e-01 6.32133782e-02 -2.14055493e-01
7.70058855e-02 3.13057099e-03 -1.06912971e-01 -4.18503821e-01
5.88647664e-01 1.33756027e-01 -3.55031877e-03 -3.49211663e-01
6.42397344e-01 8.66031885e-01 4.22396928e-01 -6.76489890e-01
4.23775971e-01 8.75033975e-01 3.23502213e-01 2.38792915e-02
-8.42870116e-01 -6.83906138e-01 8.57451633e-02 4.30730551e-01
7.08584607e-01 -1.05954337e+00 -1.35299635e+00 5.27357340e-01
-8.49981546e-01 -3.07940692e-01 -4.79054153e-01 1.76262081e-01
-5.19360721e-01 5.07292211e-01 -7.59451151e-01 -8.64311874e-01
-6.83456123e-01 -9.86116827e-01 1.19639695e+00 -2.41667077e-01
-1.53538346e-01 -8.61435115e-01 3.16565305e-01 5.91127813e-01
2.91711479e-01 4.27767336e-01 9.44775879e-01 -1.06202638e+00
-5.07472694e-01 -5.32024384e-01 -1.70209527e-01 1.47525311e-01
-1.87470153e-01 -8.32201719e-01 -1.20815337e+00 -7.24033773e-01
4.11487278e-03 -6.49770081e-01 1.17648351e+00 2.36729532e-01
1.49794805e+00 -9.26244915e-01 -6.72195494e-01 8.20643544e-01
1.44112277e+00 -2.40428731e-01 1.75525367e-01 -6.25742078e-02
3.92321825e-01 3.11192721e-01 2.61153430e-01 7.77157605e-01
7.01542854e-01 4.89553273e-01 3.30873609e-01 -8.82265717e-02
3.76814812e-01 -2.90906042e-01 -3.25517319e-02 2.03578934e-01
4.44050282e-01 -6.13484494e-02 -5.50698817e-01 4.56169575e-01
-1.83722997e+00 -8.80078673e-01 4.14896086e-02 2.22158122e+00
1.36393785e+00 -2.53975570e-01 -4.14065234e-02 -5.34568215e-03
4.18856949e-01 3.23921055e-01 -7.14283526e-01 -4.01832074e-01
-4.37135696e-01 4.99632329e-01 1.62247682e+00 3.71153057e-01
-1.43384743e+00 4.19721395e-01 6.37748528e+00 9.80660141e-01
-8.37767065e-01 8.78180444e-01 1.23016346e+00 -4.59203869e-01
-7.57524610e-01 -2.62592256e-01 -8.58421981e-01 5.20880282e-01
7.66407371e-01 -2.84105390e-02 3.97958040e-01 1.03992629e+00
-6.09368026e-01 1.31942600e-01 -1.44160891e+00 1.32536030e+00
-1.82731420e-01 -1.49851751e+00 -6.41354501e-01 6.93035007e-01
1.18102586e+00 1.78481355e-01 4.77233261e-01 2.21402913e-01
1.02998734e+00 -9.81693149e-01 6.08046830e-01 2.89271891e-01
1.57287991e+00 -7.28668571e-01 5.78058362e-01 5.29901445e-01
-6.95213616e-01 -4.15408313e-01 -1.75176039e-01 5.69222867e-01
-1.55618265e-01 6.92425489e-01 -4.65083510e-01 4.62302715e-01
5.61307549e-01 1.19200401e-01 -2.58376628e-01 5.91537297e-01
3.69762667e-02 6.07513070e-01 -6.43666804e-01 1.23557314e-01
1.07906945e-01 2.94295579e-01 3.15397345e-02 1.32039607e+00
9.78201628e-02 5.18905401e-01 5.48852049e-02 2.61793941e-01
-8.98500264e-01 2.79033542e-01 -6.95277810e-01 3.42737824e-01
7.79681861e-01 9.61137593e-01 -1.75703928e-01 -1.82243034e-01
-3.68556619e-01 1.12268627e+00 1.76120728e-01 -3.56456861e-02
-5.07804751e-01 -1.18974783e-01 9.09530759e-01 -1.71170861e-01
2.99394250e-01 3.04115087e-01 -6.18703842e-01 -1.17036867e+00
-7.93846473e-02 -1.08286369e+00 1.22195363e+00 1.40729874e-01
-1.66950917e+00 -8.64327773e-02 -3.10824484e-01 -7.64230430e-01
3.80881061e-03 -4.02323097e-01 -1.88109688e-02 7.58005619e-01
-1.53766108e+00 -1.18497634e+00 4.15231287e-01 9.98332977e-01
-3.28642935e-01 -1.44189522e-02 1.37333393e+00 3.18383127e-01
-9.80189592e-02 1.63911974e+00 6.52770579e-01 4.59429361e-02
9.93983865e-01 -1.31178403e+00 9.02484506e-02 4.46275473e-01
2.14186430e-01 4.19832647e-01 3.57491553e-01 -4.44373995e-01
-1.62240422e+00 -1.26831830e+00 1.24172699e+00 -6.75254285e-01
1.74272656e-01 -6.90256476e-01 -4.22249675e-01 9.35647964e-01
-2.71574974e-01 7.99492300e-01 1.45524001e+00 1.93901405e-01
-1.18290746e+00 -3.68014842e-01 -2.00609326e+00 1.96610391e-01
9.63621557e-01 -1.02865732e+00 1.26137927e-01 5.96320450e-01
8.08965862e-01 -2.87184834e-01 -9.60921764e-01 2.73696780e-01
9.36602354e-01 -7.80317843e-01 8.30593228e-01 -6.06789827e-01
-5.89986816e-02 1.11931518e-01 -6.54069543e-01 -7.71507978e-01
-2.20546618e-01 -9.54341352e-01 -1.53887525e-01 8.98042083e-01
6.39339447e-01 -9.82093215e-01 1.28068221e+00 1.26576221e+00
6.98378086e-01 -7.72621870e-01 -1.58278978e+00 -3.32013965e-01
4.26067233e-01 -5.05921602e-01 7.84152091e-01 1.37021148e+00
5.77697530e-02 -3.35482568e-01 -7.45884895e-01 1.53591931e-01
1.23240018e+00 1.53620929e-01 5.78542829e-01 -1.08710015e+00
-7.33367980e-01 -1.78407028e-01 -1.61413357e-01 -7.24517286e-01
3.75208527e-01 -1.21432996e+00 -1.48569569e-01 -8.64249706e-01
7.56447911e-01 -1.07467806e+00 -6.16692364e-01 1.07864702e+00
-1.06380850e-01 5.52488029e-01 1.20801225e-01 2.81934172e-01
-1.01004887e+00 -3.39737311e-02 7.18547642e-01 -4.57707286e-01
2.69333839e-01 2.18176723e-01 -1.21304119e+00 1.68193772e-01
5.29618621e-01 -1.02055287e+00 1.03784144e-01 2.77890842e-02
3.13157827e-01 2.42633894e-01 1.32126376e-01 -3.38301867e-01
4.33635294e-01 -7.18595609e-02 5.27870953e-02 -1.64856270e-01
1.24766812e-01 -1.09701777e+00 4.95706141e-01 6.78557217e-01
-8.06067109e-01 -4.16771948e-01 -3.19005340e-01 9.29994643e-01
4.28940862e-01 2.30564535e-01 9.98181701e-01 -1.68638453e-01
-9.87525806e-02 7.78491616e-01 2.05695435e-01 2.80338079e-01
1.18884504e+00 1.50703862e-01 -5.69938719e-01 -3.35597485e-01
-5.58459938e-01 4.34588015e-01 5.83015382e-01 -2.31854692e-01
1.16273493e-01 -1.20292449e+00 -6.75798893e-01 1.43330306e-01
4.63369966e-01 -1.00804269e-01 3.99526447e-01 7.40994990e-01
8.08775201e-02 4.30282772e-01 4.51543272e-01 -2.76592076e-01
-1.61588335e+00 6.03035092e-01 3.29797596e-01 -6.28604114e-01
-4.64007765e-01 1.16971099e+00 1.79746673e-01 -7.02825010e-01
5.30649722e-01 7.25149587e-02 7.54913509e-01 -4.74549830e-02
5.35318255e-01 2.31705084e-01 4.68561761e-02 -4.32921797e-01
-3.64191443e-01 2.86185712e-01 -2.09669486e-01 -1.26769081e-01
1.22517240e+00 -1.32682500e-02 -2.36136869e-01 6.82293102e-02
1.73550904e+00 1.54973775e-01 -1.07312655e+00 -7.17800200e-01
-1.64400741e-01 -5.06275952e-01 -1.36234954e-01 -1.03062451e+00
-1.14419436e+00 4.11052227e-01 7.37345397e-01 1.57886781e-02
8.90382230e-01 2.68123239e-01 8.88947010e-01 3.28036547e-01
9.66819406e-01 -8.67079258e-01 -7.49032736e-01 -1.35099635e-01
1.59197912e-01 -1.37563217e+00 1.44209936e-02 -1.65634736e-01
-6.03052497e-01 4.44407672e-01 -2.60127097e-01 4.01450723e-01
1.10405421e+00 3.39387566e-01 8.37368965e-02 -1.40262932e-01
-9.08017159e-01 4.18748766e-01 -4.90940139e-02 3.02975178e-01
-5.01095057e-02 7.14293659e-01 -4.63352412e-01 9.84577537e-01
3.41248838e-03 -8.28748196e-02 2.03145551e-03 1.02046204e+00
-9.35420990e-02 -1.54814374e+00 -2.13120028e-01 8.24674666e-01
-1.22962177e+00 -1.82357505e-02 -1.44670710e-01 3.65742326e-01
3.85155499e-01 7.87878692e-01 -3.86693388e-01 -3.10503364e-01
1.20654084e-01 3.65265965e-01 3.41584355e-01 -4.00215417e-01
-9.83104467e-01 -3.33905548e-01 1.87609140e-02 -7.80181348e-01
-1.74945906e-01 -8.73230577e-01 -9.23473179e-01 -7.35372663e-01
-2.12350458e-01 1.44369781e-01 6.84835434e-01 6.17823899e-01
5.14805734e-01 -4.47027832e-01 9.40902531e-01 -2.80956905e-02
-1.26760542e+00 -4.15052801e-01 -9.92397070e-01 6.56380117e-01
4.80591476e-01 -4.99360077e-03 -4.57187474e-01 -2.30826974e-01]
|
[5.932338237762451, 6.678913593292236]
|
185232b1-1f99-4263-9692-7b474ada2ae8
|
weakly-supervised-video-moment-retrieval-from
|
1904.03282
| null |
https://arxiv.org/abs/1904.03282v2
|
https://arxiv.org/pdf/1904.03282v2.pdf
|
Weakly Supervised Video Moment Retrieval From Text Queries
|
There have been a few recent methods proposed in text to video moment retrieval using natural language queries, but requiring full supervision during training. However, acquiring a large number of training videos with temporal boundary annotations for each text description is extremely time-consuming and often not scalable. In order to cope with this issue, in this work, we introduce the problem of learning from weak labels for the task of text to video moment retrieval. The weak nature of the supervision is because, during training, we only have access to the video-text pairs rather than the temporal extent of the video to which different text descriptions relate. We propose a joint visual-semantic embedding based framework that learns the notion of relevant segments from video using only video-level sentence descriptions. Specifically, our main idea is to utilize latent alignment between video frames and sentence descriptions using Text-Guided Attention (TGA). TGA is then used during the test phase to retrieve relevant moments. Experiments on two benchmark datasets demonstrate that our method achieves comparable performance to state-of-the-art fully supervised approaches.
|
['Amit K. Roy-Chowdhury', 'Niluthpol Chowdhury Mithun', 'Sujoy Paul']
|
2019-04-05
|
weakly-supervised-video-moment-retrieval-from-1
|
http://openaccess.thecvf.com/content_CVPR_2019/html/Mithun_Weakly_Supervised_Video_Moment_Retrieval_From_Text_Queries_CVPR_2019_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2019/papers/Mithun_Weakly_Supervised_Video_Moment_Retrieval_From_Text_Queries_CVPR_2019_paper.pdf
|
cvpr-2019-6
|
['moment-retrieval']
|
['computer-vision']
|
[ 2.04583764e-01 -4.04808402e-01 -5.93730092e-01 -4.56392676e-01
-1.06535244e+00 -5.27013600e-01 5.85654438e-01 9.25558731e-02
-5.73926687e-01 5.29713809e-01 4.15703386e-01 1.86399445e-01
-3.59316655e-02 -1.96449861e-01 -7.61510730e-01 -6.15296423e-01
4.76221330e-02 2.76506543e-01 2.58967519e-01 1.31372482e-01
3.34635973e-01 1.32595047e-01 -1.42356586e+00 5.00881970e-01
2.21051842e-01 1.02371061e+00 4.56771284e-01 5.61361372e-01
-1.73678339e-01 1.05992103e+00 -3.45843703e-01 -4.21943292e-02
6.34105727e-02 -4.99517828e-01 -9.54833984e-01 4.71214086e-01
7.63669670e-01 -7.69141734e-01 -6.94788814e-01 8.17757785e-01
3.82644594e-01 6.06378138e-01 5.81142366e-01 -1.30194187e+00
-6.16067648e-01 2.35165596e-01 -6.33602500e-01 4.99203801e-01
7.67734468e-01 -2.48567060e-01 1.25378680e+00 -1.13886607e+00
1.01213539e+00 1.07792020e+00 2.18279436e-01 5.88940084e-01
-8.55305314e-01 -2.87182093e-01 3.94576907e-01 5.65521240e-01
-1.42585385e+00 -6.22861564e-01 9.76294994e-01 -5.60833395e-01
9.17084694e-01 -1.82379745e-02 5.18313229e-01 1.17469621e+00
-2.39910278e-02 1.09291708e+00 4.72834468e-01 -4.10470724e-01
-6.44199476e-02 5.54803535e-02 1.79646127e-02 8.20054531e-01
-3.25580180e-01 -4.68422800e-01 -8.57769668e-01 -5.02512492e-02
6.47762954e-01 3.85837674e-01 -4.60980207e-01 -6.12696648e-01
-1.40234816e+00 7.48779774e-01 6.33335486e-02 3.82800341e-01
-2.80446798e-01 2.21647814e-01 7.84027874e-01 3.40405226e-01
6.73868835e-01 8.03418458e-02 -2.46921644e-01 -2.47336835e-01
-1.22984028e+00 5.63006550e-02 6.99693739e-01 1.05787170e+00
8.21713150e-01 -3.05196345e-01 -2.57567704e-01 6.92225933e-01
2.62752533e-01 3.36753488e-01 5.22399962e-01 -9.39012766e-01
7.14446723e-01 2.96864241e-01 1.56973734e-01 -1.08272970e+00
1.57540858e-01 2.19426528e-01 -3.82879734e-01 -5.72712302e-01
2.23850444e-01 2.79577166e-01 -6.35233402e-01 1.49465132e+00
2.38526538e-01 3.05788219e-01 -6.09893799e-02 1.05995274e+00
6.06213570e-01 9.39178050e-01 1.60157797e-03 -4.92293596e-01
1.17218709e+00 -1.40758455e+00 -9.62350368e-01 -1.20017663e-01
6.86676860e-01 -8.51123691e-01 1.03043091e+00 -5.28074205e-02
-1.18606067e+00 -5.51185608e-01 -7.99092948e-01 -4.19098407e-01
-2.70977408e-01 1.81543529e-01 1.97445422e-01 -1.57058820e-01
-1.13012886e+00 3.94688189e-01 -8.69868934e-01 -6.93550408e-01
9.28483829e-02 1.92518055e-01 -6.34528100e-01 -3.89116973e-01
-1.17997098e+00 4.82141078e-01 3.72032136e-01 -4.10396270e-02
-1.14672005e+00 -2.82772213e-01 -1.12313366e+00 9.83118266e-02
6.92180395e-01 -5.35507202e-01 1.19036007e+00 -1.22106874e+00
-1.27017438e+00 8.75344932e-01 -4.72916305e-01 -3.27480018e-01
2.28578299e-01 -5.18996239e-01 -2.12604254e-02 1.04013562e+00
4.28914279e-01 8.92977595e-01 1.21943617e+00 -9.37980294e-01
-6.04402423e-01 -2.14631066e-01 3.29985857e-01 4.04385448e-01
-6.79774404e-01 3.57806176e-01 -1.10794628e+00 -6.28572404e-01
-2.22672541e-02 -9.18223977e-01 1.80826187e-01 2.51419038e-01
1.85540114e-02 -5.51221550e-01 1.21897054e+00 -7.95038044e-01
1.26028502e+00 -2.34533954e+00 3.95299286e-01 -2.86140829e-01
-3.57999327e-03 4.84288260e-02 -2.82999426e-01 6.93536699e-01
-2.21671835e-02 4.78222128e-03 6.59798235e-02 -6.11950755e-01
-2.04935372e-02 1.99640691e-01 -5.64976990e-01 6.10271394e-01
1.78400308e-01 8.56292009e-01 -1.12595379e+00 -1.06955659e+00
3.07501316e-01 5.37528336e-01 -5.03640056e-01 5.36952913e-01
-2.72825658e-01 3.91165614e-01 -8.20945024e-01 7.68109024e-01
1.05233766e-01 -4.47583318e-01 -3.18526328e-02 -2.85002649e-01
3.00468132e-03 1.73126653e-01 -7.11329103e-01 2.24436092e+00
-2.87539572e-01 1.01197124e+00 -2.53546871e-02 -1.31837273e+00
3.91874433e-01 6.82586074e-01 7.81911373e-01 -5.44190526e-01
-9.31253806e-02 3.63129787e-02 -7.44210422e-01 -1.04274666e+00
4.79613364e-01 1.35950372e-01 -6.66860566e-02 6.23171806e-01
2.72967428e-01 5.38327023e-02 5.80268621e-01 4.75664169e-01
9.15903032e-01 5.29697120e-01 3.14811528e-01 2.16658443e-01
6.34026587e-01 -9.57672521e-02 4.17068124e-01 5.50075650e-01
-4.47743565e-01 7.34309793e-01 5.42747915e-01 -4.44428504e-01
-1.14413798e+00 -7.30033517e-01 2.25048929e-01 1.30845475e+00
1.97446927e-01 -6.40951455e-01 -5.86375058e-01 -8.78129423e-01
-3.55255127e-01 1.98035121e-01 -5.84217727e-01 3.15788090e-02
-6.54827952e-01 1.35458767e-01 1.31903112e-01 4.90873307e-01
3.57639819e-01 -1.10221386e+00 -4.92498904e-01 1.08094379e-01
-6.76144421e-01 -1.61907184e+00 -1.14505911e+00 -2.75933534e-01
-9.40870047e-01 -9.77407217e-01 -1.00778759e+00 -1.12713075e+00
7.81649053e-01 7.39512682e-01 9.64426458e-01 8.09990838e-02
-1.91167682e-01 9.39678133e-01 -7.29380071e-01 3.27706486e-01
1.61685482e-01 -6.67436793e-02 -4.57912199e-02 1.66945621e-01
3.26928854e-01 -2.84241021e-01 -6.28044963e-01 2.42328286e-01
-1.18047118e+00 -2.59907208e-02 4.06051010e-01 9.47793067e-01
8.12410474e-01 -2.76983827e-01 1.95936561e-01 -4.99634862e-01
2.38600880e-01 -4.97556269e-01 -2.81010866e-01 4.90342766e-01
-1.11984558e-01 1.86323021e-02 5.97193837e-01 -4.85975087e-01
-8.14269602e-01 2.27585375e-01 3.15411031e-01 -1.12320721e+00
-8.00792426e-02 6.74716234e-01 2.44931765e-02 2.37920657e-01
-1.40709430e-02 4.29280907e-01 -2.49776661e-01 -3.75491887e-01
1.32047281e-01 6.15312099e-01 3.90301228e-01 -4.78461146e-01
6.54934883e-01 7.05934465e-01 -1.74344510e-01 -8.58135939e-01
-1.15961671e+00 -1.09044385e+00 -9.30301011e-01 -3.87073308e-01
1.08287871e+00 -1.01098204e+00 -2.60167062e-01 -8.19375888e-02
-1.25933075e+00 -2.00861812e-01 -1.18470930e-01 7.11539626e-01
-8.46102178e-01 8.22049797e-01 -5.72604060e-01 -4.74836856e-01
-2.87540644e-01 -1.10191023e+00 1.47506595e+00 -1.28562257e-01
3.25868502e-02 -1.09053946e+00 5.53166866e-02 6.54529631e-01
1.48872500e-02 1.49478108e-01 6.13636494e-01 -5.98167777e-01
-7.95040905e-01 -4.35789675e-01 -2.45488852e-01 1.66924998e-01
1.99797049e-01 -4.70050564e-03 -6.95697546e-01 -6.24164462e-01
1.17416777e-01 -6.29963994e-01 8.69071722e-01 3.56528729e-01
1.28093278e+00 -2.46020943e-01 -4.86145377e-01 5.12910187e-01
1.33490503e+00 4.34567519e-02 3.79349113e-01 4.19207156e-01
7.59938359e-01 7.39930511e-01 1.12758362e+00 3.88605654e-01
3.01795214e-01 8.92528474e-01 2.56566018e-01 2.03563511e-01
4.19102646e-02 -2.80910522e-01 6.45106912e-01 9.92983997e-01
6.22461848e-02 -4.44213063e-01 -6.79015279e-01 8.12206030e-01
-2.06971121e+00 -1.21232235e+00 4.86045778e-01 2.06669140e+00
7.23672092e-01 -1.59054503e-01 -5.87243885e-02 -1.32152379e-01
7.07138538e-01 5.40872872e-01 -3.80113423e-01 3.49416733e-02
2.40036890e-01 -3.46991986e-01 9.00404230e-02 3.51201892e-01
-1.33702433e+00 8.94797444e-01 5.37136555e+00 7.34134316e-01
-1.11417770e+00 2.76281536e-01 3.47372800e-01 -4.03522700e-01
4.83470857e-02 1.80690616e-01 -5.77389479e-01 3.31151277e-01
8.05144966e-01 -1.17095046e-01 2.28494316e-01 7.73578167e-01
4.13221985e-01 4.09411592e-03 -1.41634643e+00 1.10513282e+00
7.54724324e-01 -1.17350256e+00 2.71481633e-01 -2.50485480e-01
8.53302658e-01 -1.28882244e-01 -2.98920460e-02 2.12743863e-01
-6.42275333e-01 -5.70595682e-01 6.50772452e-01 5.29266477e-01
9.05803978e-01 -5.44049919e-01 6.34763241e-01 2.00757787e-01
-1.46359944e+00 5.24542294e-02 -4.11022246e-01 2.21093878e-01
3.28320652e-01 2.11378112e-01 -6.08246565e-01 5.06241798e-01
8.54934394e-01 1.31407249e+00 -4.47715908e-01 8.00434411e-01
-2.46211931e-01 3.30712110e-01 -5.70237637e-03 1.56656101e-01
4.69579577e-01 -5.25856344e-03 6.08009100e-01 1.16097176e+00
3.54370803e-01 3.39406244e-02 5.85288048e-01 3.77948225e-01
-2.11395502e-01 2.38537207e-01 -8.51406455e-01 -4.94416445e-01
2.11902186e-01 1.04600942e+00 -7.46021390e-01 -4.89148885e-01
-9.17142689e-01 1.38114524e+00 3.15170616e-01 5.82218349e-01
-8.42447937e-01 -4.63314116e-01 1.13622069e-01 1.15940057e-01
4.08484727e-01 -3.40347290e-01 7.66085088e-01 -1.50058520e+00
3.48732412e-01 -6.63076997e-01 5.06124914e-01 -1.00685501e+00
-1.12906051e+00 3.68413806e-01 2.37593159e-01 -1.52706099e+00
-3.56765717e-01 -3.59689951e-01 -3.04961056e-01 4.50236291e-01
-1.80241203e+00 -1.23037517e+00 -3.23538184e-01 8.75949264e-01
1.20079708e+00 -4.41630259e-02 5.52827060e-01 4.57998037e-01
-3.43022674e-01 3.39228094e-01 2.68603653e-01 3.30363005e-01
1.22345436e+00 -9.47461665e-01 -1.22134507e-01 7.56621838e-01
4.13227648e-01 6.13557994e-01 3.96009028e-01 -4.91081864e-01
-1.46952820e+00 -1.01614010e+00 1.20414782e+00 -3.36726427e-01
8.28171849e-01 -2.88079798e-01 -7.71195889e-01 9.10386741e-01
4.41451341e-01 2.38363460e-01 6.90348208e-01 -4.05486256e-01
-1.87819064e-01 -1.12025611e-01 -4.81959999e-01 5.56804419e-01
7.63908684e-01 -1.07868028e+00 -7.86440730e-01 6.54464245e-01
8.64627779e-01 -2.73698807e-01 -6.39425755e-01 7.16909692e-02
4.46981817e-01 -6.32041216e-01 1.02569091e+00 -4.31133330e-01
5.92009306e-01 -2.73874253e-01 -2.70748764e-01 -7.47478843e-01
1.18218631e-01 -6.65505826e-01 -2.10483342e-01 1.23649311e+00
-9.48235765e-02 3.23842391e-02 5.87389231e-01 3.57503027e-01
-9.17221829e-02 -6.13167942e-01 -9.60552692e-01 -6.05154932e-01
-4.27982032e-01 -1.35472789e-01 -8.00814927e-02 1.02351248e+00
1.46391392e-01 4.21716034e-01 -6.30069971e-01 1.49346981e-02
4.62733418e-01 3.53478551e-01 6.61701560e-01 -8.70921612e-01
-1.41406164e-01 -5.69725335e-02 -4.92353678e-01 -1.37670219e+00
5.48038125e-01 -7.71565437e-01 2.49685362e-01 -1.46026003e+00
7.41092682e-01 3.60180110e-01 -3.89371753e-01 3.88356775e-01
-1.14840411e-01 2.77129501e-01 2.37516806e-01 6.24479055e-01
-1.39027524e+00 5.92800379e-01 1.28028607e+00 -3.72410029e-01
-1.65639445e-02 -3.71002704e-01 1.21271387e-02 5.58173418e-01
3.77094775e-01 -6.46813273e-01 -6.11972809e-01 -6.50543630e-01
1.69788450e-01 3.84857982e-01 3.76913249e-01 -8.86223257e-01
4.40900773e-01 -1.66696757e-01 1.52038246e-01 -7.87601888e-01
5.51306367e-01 -9.17445779e-01 -3.37246299e-01 8.19429085e-02
-6.07238173e-01 6.93902224e-02 -7.43365288e-02 9.37545478e-01
-7.58657634e-01 -4.63126987e-01 2.98565507e-01 -1.23159058e-01
-9.14984345e-01 6.19406521e-01 -3.05989683e-01 7.17967525e-02
1.03595424e+00 -1.26608223e-01 6.16418719e-02 -6.70781970e-01
-7.59366095e-01 4.00843859e-01 5.27132690e-01 4.99323875e-01
8.83260906e-01 -1.30699456e+00 -3.67544651e-01 -7.88619220e-02
3.67005199e-01 -1.84778422e-01 3.11083555e-01 9.62412357e-01
-3.61233413e-01 8.31855893e-01 -1.63491890e-02 -7.79267848e-01
-1.43911278e+00 9.16759372e-01 -9.16057825e-02 -2.58557469e-01
-6.91730976e-01 6.44936025e-01 4.48001713e-01 1.68127269e-01
5.64848661e-01 -6.49026558e-02 -4.49541271e-01 3.43255877e-01
5.70196748e-01 2.17306167e-02 -3.58016163e-01 -8.98463130e-01
-2.48651043e-01 9.41636622e-01 -3.37502241e-01 -1.30682379e-01
1.23884296e+00 -6.26908958e-01 -1.11249305e-01 7.10616946e-01
1.74810994e+00 -3.37225080e-01 -1.45074403e+00 -5.13359308e-01
6.02485426e-02 -6.93887234e-01 2.63043791e-02 -6.16724938e-02
-9.98600364e-01 1.16499472e+00 4.81516510e-01 -1.65870748e-02
1.12145078e+00 1.52006075e-01 1.05060494e+00 7.91057110e-01
1.54711396e-01 -1.28893685e+00 6.49875164e-01 4.77943182e-01
7.48855293e-01 -1.51990700e+00 1.07638344e-01 -1.99542493e-02
-6.33234262e-01 1.28315747e+00 4.29938793e-01 -8.72792825e-02
4.22335953e-01 -3.53290796e-01 -7.00451583e-02 -1.46150604e-01
-9.77433681e-01 -1.78238422e-01 4.54120636e-01 1.92660525e-01
4.66728330e-01 -5.88878393e-01 -2.47675568e-01 1.04252294e-01
6.55066192e-01 6.20546080e-02 3.21530491e-01 1.32208526e+00
-3.66736978e-01 -1.01793611e+00 -1.43971294e-01 1.02805771e-01
-7.73163438e-01 -1.23424254e-01 -3.57478172e-01 7.92655468e-01
-3.08483988e-01 8.55636954e-01 1.90203607e-01 7.87795410e-02
3.11551280e-02 2.92210400e-01 3.72344047e-01 -8.33598733e-01
-6.40405416e-02 4.84741688e-01 -2.69539714e-01 -7.80624509e-01
-1.03225267e+00 -6.47986352e-01 -1.20809913e+00 1.26380682e-01
-3.01689595e-01 4.46260512e-01 4.34359938e-01 1.04637969e+00
1.90460846e-01 2.98823923e-01 8.27963054e-01 -1.12789714e+00
-2.46458381e-01 -7.98266411e-01 -4.44251090e-01 6.48906946e-01
6.80322587e-01 -6.08853102e-01 -5.33817828e-01 7.36297309e-01]
|
[10.14433479309082, 0.8042119145393372]
|
3675a577-fba1-4d52-9f34-3ae17e8f57d8
|
mae-gebd-winning-the-cvpr-2023-loveu-gebd
|
2306.15704
| null |
https://arxiv.org/abs/2306.15704v1
|
https://arxiv.org/pdf/2306.15704v1.pdf
|
MAE-GEBD:Winning the CVPR'2023 LOVEU-GEBD Challenge
|
The Generic Event Boundary Detection (GEBD) task aims to build a model for segmenting videos into segments by detecting general event boundaries applicable to various classes. In this paper, based on last year's MAE-GEBD method, we have improved our model performance on the GEBD task by adjusting the data processing strategy and loss function. Based on last year's approach, we extended the application of pseudo-label to a larger dataset and made many experimental attempts. In addition, we applied focal loss to concentrate more on difficult samples and improved our model performance. Finally, we improved the segmentation alignment strategy used last year, and dynamically adjusted the segmentation alignment method according to the boundary density and duration of the video, so that our model can be more flexible and fully applicable in different situations. With our method, we achieve an F1 score of 86.03% on the Kinetics-GEBD test set, which is a 0.09% improvement in the F1 score compared to our 2022 Kinetics-GEBD method.
|
['Jie Tang', 'Xu Cheng', 'Feng Hu', 'Zuwei Huang', 'Youzeng Li', 'Rui He', 'Yuanxi Sun']
|
2023-06-27
| null | null | null | null |
['boundary-detection', 'pseudo-label']
|
['computer-vision', 'miscellaneous']
|
[-7.40646720e-02 -6.14107996e-02 -1.65169001e-01 -4.01584715e-01
-7.45838881e-01 -4.45659131e-01 4.21347857e-01 1.46888927e-01
-7.42453396e-01 5.10546148e-01 3.42169702e-02 3.01951729e-02
1.33864209e-01 -6.64036930e-01 -7.15072274e-01 -5.49781263e-01
-3.09095562e-01 4.51247573e-01 9.62050438e-01 1.40804812e-01
1.00738280e-01 4.61509198e-01 -1.49858725e+00 4.13486272e-01
7.36646652e-01 9.61798847e-01 1.72776699e-01 5.86441338e-01
2.63072699e-01 3.12215179e-01 -7.66647100e-01 -3.67921203e-01
1.89039931e-01 -5.75700164e-01 -8.81182492e-01 1.03640772e-01
3.19578469e-01 -1.24891683e-01 -3.63886625e-01 6.98162377e-01
5.76466322e-01 2.77787209e-01 8.74730885e-01 -1.27218497e+00
8.05161744e-02 5.68941414e-01 -5.79003215e-01 5.66429257e-01
2.61077881e-01 -4.12771069e-02 7.35484958e-01 -6.72668040e-01
7.55126238e-01 9.99429643e-01 7.48137414e-01 6.53524160e-01
-1.03690898e+00 -6.55572653e-01 5.62502265e-01 6.60030246e-01
-1.47058487e+00 -1.22373976e-01 5.47490835e-01 -5.91991782e-01
8.09824824e-01 2.67865747e-01 7.64483750e-01 1.21779597e+00
1.44545928e-01 9.42231476e-01 7.06105113e-01 -4.68582302e-01
1.15256451e-01 -1.15780076e-02 1.54037893e-01 5.80628395e-01
-2.38504130e-02 -1.65713966e-01 -1.94207162e-01 1.47867054e-01
5.35774767e-01 -2.97350913e-01 -1.95335731e-01 -3.01520139e-01
-1.23502314e+00 7.15495527e-01 1.20485373e-01 1.86993241e-01
-1.86882168e-01 -2.16423661e-01 7.48208165e-01 3.13260779e-02
6.16387188e-01 2.75084436e-01 -6.28426731e-01 -3.07966799e-01
-1.15227211e+00 3.91671658e-01 8.01302373e-01 8.49548340e-01
3.41033131e-01 -4.00155693e-01 -5.99189758e-01 9.20128942e-01
7.24511743e-02 1.03568248e-01 4.01461363e-01 -9.99987066e-01
4.11539316e-01 3.71473640e-01 3.20619345e-02 -7.98714101e-01
-7.39149153e-01 -3.89305294e-01 -5.44569552e-01 -3.39825638e-02
5.96397638e-01 -2.78204590e-01 -1.08941722e+00 1.78723752e+00
2.95297146e-01 2.81624377e-01 -3.38374585e-01 9.19914842e-01
6.90932274e-01 7.21343696e-01 2.77290165e-01 -3.32137257e-01
1.38036501e+00 -1.06820023e+00 -6.67233288e-01 6.86111003e-02
7.18300581e-01 -7.45753109e-01 1.00102568e+00 7.79554427e-01
-8.58691871e-01 -8.24243009e-01 -9.94181871e-01 3.74038190e-01
-1.35534912e-01 1.63403526e-01 4.09554213e-01 5.34298897e-01
-8.48061323e-01 5.36975265e-01 -1.02703750e+00 -5.43687403e-01
2.91973114e-01 2.53666312e-01 -1.19882554e-01 9.68880206e-02
-1.35937846e+00 7.19892561e-01 7.28368878e-01 3.96982394e-02
-1.00936747e+00 -4.62301761e-01 -7.47645497e-01 -9.93619114e-02
6.27799273e-01 -5.69047213e-01 1.21389782e+00 -4.41527009e-01
-1.27597678e+00 9.29779172e-01 -1.45454213e-01 -6.25816226e-01
9.30846155e-01 -1.75792143e-01 -5.75353026e-01 3.03299367e-01
1.54484082e-02 9.66419399e-01 4.06780511e-01 -7.96739876e-01
-9.51392651e-01 -2.35211495e-02 -9.93238539e-02 9.24151391e-02
-9.82295498e-02 1.98781669e-01 -9.65248048e-01 -8.58002722e-01
-6.78505152e-02 -1.03866398e+00 -7.26111978e-02 -3.26082379e-01
-3.42030488e-02 -4.61957783e-01 8.08724165e-01 -9.42553043e-01
1.58518577e+00 -2.39658904e+00 2.13439599e-01 1.05933063e-01
3.38111147e-02 3.61390352e-01 -8.08223039e-02 2.10645959e-01
-1.58712327e-01 1.97066754e-01 -2.36984000e-01 -3.78361106e-01
-1.39178783e-01 -1.28329135e-02 1.34219065e-01 2.77568042e-01
1.69643939e-01 5.42438328e-01 -7.37046540e-01 -8.06588054e-01
1.09686382e-01 2.07858875e-01 -7.47517884e-01 2.26445436e-01
-2.59715170e-01 3.68611008e-01 -1.68312952e-01 5.10261655e-01
5.46783328e-01 4.64566238e-02 -1.30903021e-01 -1.83970913e-01
-8.47591273e-03 3.98047641e-02 -1.31526518e+00 1.74224031e+00
-1.89069033e-01 8.31403971e-01 -2.70626843e-01 -1.05188835e+00
9.07451272e-01 2.64508039e-01 8.36575747e-01 -5.25211632e-01
-2.52006156e-03 -7.43357763e-02 -2.84970440e-02 -8.05834711e-01
2.14580372e-01 1.78406790e-01 -8.87792408e-02 2.27342583e-02
4.62323129e-02 1.11022964e-01 8.40951025e-01 1.20455965e-01
9.76355076e-01 4.59026724e-01 8.97524133e-02 -2.23785102e-01
4.89191413e-01 -4.82404232e-02 8.15864980e-01 7.45352030e-01
-5.34417629e-01 8.57536137e-01 6.04082167e-01 -3.56665134e-01
-7.99606740e-01 -1.12891352e+00 -4.25074846e-01 9.28212464e-01
2.33331531e-01 -6.49334610e-01 -1.01390088e+00 -1.06903875e+00
-3.06789428e-01 6.58982873e-01 -4.91985381e-01 -2.57879496e-01
-8.14864814e-01 -1.10224175e+00 6.63871527e-01 4.85579431e-01
6.97155356e-01 -1.08785021e+00 -3.08421493e-01 5.28407395e-01
-5.86386621e-01 -1.18485153e+00 -6.71792984e-01 -3.22666462e-03
-6.66294634e-01 -1.13919604e+00 -8.65526497e-01 -8.34517896e-01
2.11755469e-01 -1.39284387e-01 1.06426525e+00 -2.36922577e-01
-3.20315927e-01 1.51178658e-01 -6.71805263e-01 -3.59072238e-01
-3.77728999e-01 3.44998121e-01 -1.80719361e-01 -9.40272063e-02
2.81464309e-01 -8.60608444e-02 -6.70307219e-01 7.83642471e-01
-8.65337968e-01 -7.63992295e-02 2.05457434e-01 5.59035301e-01
5.87887764e-01 1.80844247e-01 7.68657327e-01 -8.30359042e-01
2.48540759e-01 -4.85622436e-01 -5.50006330e-01 2.13096067e-01
-7.41354048e-01 -2.13880435e-01 2.82008588e-01 -6.35228693e-01
-1.18713450e+00 -8.48971456e-02 -4.59874749e-01 -2.51754612e-01
-2.38393903e-01 3.22740376e-01 -4.08529341e-01 2.64275938e-01
5.06656826e-01 -9.36334878e-02 -2.60734499e-01 -6.05725110e-01
-2.33191755e-02 6.86476707e-01 4.40308064e-01 -5.06280482e-01
-7.38516301e-02 2.45492846e-01 -1.65657669e-01 -5.63045204e-01
-7.63235569e-01 -6.34156704e-01 -5.67566097e-01 -4.51753318e-01
1.00471640e+00 -9.17235136e-01 -5.02064049e-01 8.65577221e-01
-9.20704067e-01 -6.41514182e-01 -8.49569961e-02 7.33027697e-01
-5.42247236e-01 7.59164929e-01 -8.26264024e-01 -4.36749339e-01
5.42442389e-02 -1.15718102e+00 9.54601645e-01 1.65369496e-01
-3.22243214e-01 -9.66495931e-01 1.15141712e-01 2.45491281e-01
2.80880537e-02 4.35795456e-01 5.28696299e-01 -7.28406966e-01
-1.60282478e-01 -4.68180664e-02 -1.02844484e-01 5.00224113e-01
-8.71341303e-02 2.62603462e-01 -7.72475839e-01 -3.02642703e-01
-5.09149730e-02 8.56610015e-02 1.17627144e+00 6.16412222e-01
1.40829182e+00 8.79697651e-02 -5.56800127e-01 5.35476804e-01
1.00159907e+00 5.89956999e-01 8.28706861e-01 5.50040662e-01
3.91453236e-01 5.95342636e-01 1.13554299e+00 4.27544117e-01
3.64719599e-01 1.10815179e+00 2.63694733e-01 -1.32851183e-01
-3.45327348e-01 -4.97509614e-02 5.84933937e-01 4.32726890e-01
-4.48070765e-02 -6.71608508e-01 -8.12028587e-01 6.10616207e-01
-2.03827763e+00 -9.62244630e-01 -3.51301372e-01 2.06906748e+00
8.47844660e-01 5.81256688e-01 6.08843565e-01 3.07945069e-02
1.04645336e+00 4.18509059e-02 -2.53580153e-01 -4.28915083e-01
-6.07181399e-04 -9.49878767e-02 3.69470537e-01 2.27088630e-01
-1.48435032e+00 8.85470867e-01 6.65557146e+00 1.21035564e+00
-1.14595270e+00 1.26009703e-01 6.12117231e-01 -6.46074489e-02
3.16526473e-01 -3.01530510e-01 -1.15257478e+00 9.65262949e-01
1.13305867e+00 2.51643937e-02 9.38836634e-02 4.61858511e-01
3.84093463e-01 -3.22188377e-01 -1.21927726e+00 7.97150850e-01
6.51544482e-02 -8.41360271e-01 -2.27193579e-01 3.81182134e-02
6.97713435e-01 4.23963182e-02 -4.26927149e-01 5.73056936e-01
-1.47635132e-01 -6.78904235e-01 8.50417554e-01 4.29452270e-01
5.29704094e-01 -6.30283415e-01 9.46630955e-01 3.92159611e-01
-1.13326573e+00 7.11860731e-02 -7.50945136e-02 7.34860972e-02
5.07059336e-01 7.51542151e-01 -1.00285184e+00 6.46410048e-01
9.07286823e-01 7.91853309e-01 -6.97006285e-01 1.66081536e+00
-1.00180387e-01 1.01599431e+00 -4.93414342e-01 1.43337265e-01
7.79210031e-02 -9.47504193e-02 5.30943632e-01 1.70880473e+00
3.00558805e-01 -2.11611867e-01 3.50425899e-01 4.74545568e-01
2.05441087e-01 7.67155364e-02 -2.35853329e-01 4.46887434e-01
3.22925001e-01 1.15757346e+00 -1.06178069e+00 -4.15605932e-01
-2.14120924e-01 8.96012962e-01 -2.36047618e-03 3.92012477e-01
-1.45720851e+00 -4.53851640e-01 3.28874499e-01 1.87185198e-01
5.25266290e-01 -1.35996848e-01 7.82204568e-02 -1.08514559e+00
-1.77522600e-02 -6.72332048e-01 8.18873227e-01 -6.04737520e-01
-1.07224131e+00 5.77040434e-01 6.26293778e-01 -1.20874393e+00
-2.02571407e-01 -4.07672793e-01 -4.41073507e-01 4.02893871e-01
-1.02316105e+00 -8.38341117e-01 -2.32661664e-01 3.56182694e-01
7.80639231e-01 1.41072467e-01 3.61024588e-01 7.66842484e-01
-7.80507088e-01 6.89834654e-01 -1.46615788e-01 1.12272456e-01
1.00721431e+00 -1.05559206e+00 3.06534320e-01 1.05154824e+00
1.53193757e-01 1.27283290e-01 7.40598440e-01 -5.70690393e-01
-2.01998040e-01 -1.18768919e+00 7.35809207e-01 -2.63209939e-01
4.99170333e-01 -4.68391806e-01 -1.00908113e+00 7.20879316e-01
3.83467204e-03 -3.19276839e-01 4.42635506e-01 3.25624570e-02
6.08640630e-03 -1.70208067e-01 -1.18563282e+00 4.63471174e-01
1.33097625e+00 6.63633943e-02 -5.16433716e-01 2.23525017e-01
7.88363516e-01 -5.20981312e-01 -1.03532100e+00 7.36269712e-01
3.17168534e-01 -7.39259005e-01 8.06010365e-01 -4.54801142e-01
-2.51047276e-02 -3.94320577e-01 9.68260691e-02 -1.12875223e+00
-3.74099016e-01 -2.66361594e-01 -1.66247994e-01 1.53552186e+00
2.88087577e-01 -4.45783347e-01 6.28878713e-01 3.85886163e-01
-3.93214136e-01 -6.66553557e-01 -1.03524160e+00 -9.63040054e-01
-1.79724887e-01 -5.31931520e-01 3.45750332e-01 5.75875223e-01
-2.85321891e-01 -1.02327270e-02 -2.90821016e-01 1.85414348e-02
3.63429010e-01 -1.26535445e-01 5.20818830e-01 -1.09264708e+00
-2.30318949e-01 -4.90006328e-01 -4.34628904e-01 -1.16835272e+00
1.55682266e-01 -6.72975004e-01 1.33766428e-01 -1.31986916e+00
3.75968337e-01 -3.68387222e-01 -5.28121471e-01 5.09862721e-01
-2.40306944e-01 5.15974879e-01 2.27105618e-01 -1.94770359e-02
-8.87924492e-01 3.17429453e-01 1.19786549e+00 -5.62575571e-02
-2.33507693e-01 2.77781576e-01 -3.67776394e-01 6.84404075e-01
8.25462520e-01 -5.56901097e-01 -1.85606018e-01 -4.87686060e-02
-5.97202145e-02 -1.45353630e-01 3.34834009e-01 -1.10785544e+00
-4.28913534e-02 -1.05033271e-01 2.79568613e-01 -6.99971497e-01
1.02773659e-01 -6.67151809e-01 3.24210852e-01 5.63757658e-01
-1.71627358e-01 -1.59663692e-01 4.00815070e-01 4.52397913e-01
-2.33134553e-01 -4.01716143e-01 5.80419898e-01 1.16356060e-01
-1.03418481e+00 1.23618186e-01 -5.79510808e-01 1.84440643e-01
1.38866258e+00 -2.90421873e-01 -6.30362108e-02 -1.27231672e-01
-1.15418959e+00 4.48030293e-01 4.91186559e-01 4.52937305e-01
3.40685606e-01 -1.09168446e+00 -7.27486670e-01 1.52342856e-01
1.33120403e-01 -1.20914653e-02 2.96215266e-01 1.01237965e+00
-6.37867272e-01 2.62842238e-01 -1.47408932e-01 -8.86997938e-01
-1.35497701e+00 6.06970251e-01 3.44688207e-01 -4.24321741e-01
-6.26874983e-01 6.78771675e-01 2.26235822e-01 -1.13886565e-01
4.62687194e-01 -5.21254122e-01 -4.44471002e-01 3.10124218e-01
3.26165169e-01 6.97643518e-01 2.49899611e-01 -3.81783366e-01
-5.34445643e-01 5.76415241e-01 -1.47556797e-01 -7.04469606e-02
1.14487135e+00 -2.39274427e-01 2.65337735e-01 3.93798470e-01
1.03212261e+00 3.53442021e-02 -1.61034107e+00 2.27606997e-01
9.55230892e-02 -1.73748851e-01 -1.04211077e-01 -9.57882106e-01
-1.09950948e+00 6.20180130e-01 7.05508292e-01 3.26773942e-01
1.32737303e+00 1.39684483e-01 7.37046540e-01 -7.08001554e-02
3.78497034e-01 -1.06556904e+00 -7.13966414e-02 4.72820014e-01
6.42703414e-01 -1.10394633e+00 -1.35153815e-01 -5.15461087e-01
-6.77821994e-01 9.88776445e-01 6.64205313e-01 1.07836582e-01
5.99271357e-01 -5.40004857e-03 -1.33780599e-01 5.46482578e-02
-5.97544372e-01 -6.06137775e-02 5.32990098e-01 5.39538443e-01
3.28287661e-01 1.24969501e-02 -9.14519310e-01 6.69097662e-01
1.43487051e-01 2.27950320e-01 5.15932858e-01 5.19593298e-01
-3.74432981e-01 -1.13846338e+00 -2.06841677e-01 3.19323808e-01
-5.75212181e-01 2.53120184e-01 -1.48768812e-01 1.14215243e+00
4.73541766e-01 9.35256183e-01 1.54799208e-01 -3.64266366e-01
5.98743737e-01 1.38080284e-01 5.42561293e-01 -5.77032328e-01
-4.78461087e-01 3.05092841e-01 2.22050503e-01 -5.67702830e-01
-5.91490328e-01 -9.67913568e-01 -1.44581747e+00 -9.03368592e-02
-3.78484845e-01 3.28339607e-01 4.39291924e-01 7.71642387e-01
3.56512338e-01 7.33956397e-01 4.32704538e-01 -8.09458017e-01
-2.41844341e-01 -9.88266289e-01 -5.74997663e-01 4.70847934e-01
-1.36920199e-01 -8.78284514e-01 -3.08153838e-01 3.07844549e-01]
|
[8.668274879455566, 0.3287557363510132]
|
6d8d2088-f472-4b02-9c26-522435a065dd
|
part-aware-measurement-for-robust-multi-view-1
|
2106.11589
| null |
https://arxiv.org/abs/2106.11589v1
|
https://arxiv.org/pdf/2106.11589v1.pdf
|
Part-Aware Measurement for Robust Multi-View Multi-Human 3D Pose Estimation and Tracking
|
This paper introduces an approach for multi-human 3D pose estimation and tracking based on calibrated multi-view. The main challenge lies in finding the cross-view and temporal correspondences correctly even when several human pose estimations are noisy. Compare to previous solutions that construct 3D poses from multiple views, our approach takes advantage of temporal consistency to match the 2D poses estimated with previously constructed 3D skeletons in every view. Therefore cross-view and temporal associations are accomplished simultaneously. Since the performance suffers from mistaken association and noisy predictions, we design two strategies for aiming better correspondences and 3D reconstruction. Specifically, we propose a part-aware measurement for 2D-3D association and a filter that can cope with 2D outliers during reconstruction. Our approach is efficient and effective comparing to state-of-the-art methods; it achieves competitive results on two benchmarks: 96.8% on Campus and 97.4% on Shelf. Moreover, we extends the length of Campus evaluation frames to be more challenging and our proposal also reach well-performed result.
|
['Chu-Song Chen', 'Jia-Da Li', 'Ching-Hsien Hsu', 'Yao-Chih Lee', 'Jia-Hong Lee', 'Hau Chu']
|
2021-06-22
|
part-aware-measurement-for-robust-multi-view
|
https://openaccess.thecvf.com/content/CVPR2021W/AMFG/html/Chu_Part-Aware_Measurement_for_Robust_Multi-View_Multi-Human_3D_Pose_Estimation_and_CVPRW_2021_paper.html
|
https://openaccess.thecvf.com/content/CVPR2021W/AMFG/papers/Chu_Part-Aware_Measurement_for_Robust_Multi-View_Multi-Human_3D_Pose_Estimation_and_CVPRW_2021_paper.pdf
| null |
['3d-pose-estimation', '3d-human-pose-tracking']
|
['computer-vision', 'computer-vision']
|
[-2.53243923e-01 -1.65941268e-01 -5.78338280e-02 -2.86484420e-01
-9.74693775e-01 -4.59442288e-01 2.77164370e-01 -1.54798463e-01
-3.19001436e-01 4.68680412e-01 1.84086084e-01 4.67990279e-01
6.10270649e-02 -3.79218370e-01 -8.88612866e-01 -2.55285770e-01
7.40920156e-02 7.54638374e-01 6.06309295e-01 -3.37804645e-01
1.12151220e-01 5.80604732e-01 -1.65740824e+00 -8.28425735e-02
4.42596078e-01 7.62188137e-01 4.93233427e-02 6.53895438e-01
3.48511517e-01 8.03701207e-02 -4.50476289e-01 -5.25690734e-01
6.06067717e-01 -1.31390303e-01 -5.96494198e-01 6.23594284e-01
1.08047390e+00 -3.86982381e-01 2.26534326e-02 9.95700121e-01
9.43980336e-01 3.60580981e-02 1.29817858e-01 -1.24015212e+00
1.93201408e-01 -7.12400079e-02 -9.82493281e-01 -4.56470363e-02
1.18841028e+00 -7.27180913e-02 5.46162963e-01 -1.24728549e+00
8.87256026e-01 1.31524014e+00 1.09251010e+00 4.10636663e-01
-9.83872235e-01 -4.89165843e-01 3.79841477e-01 6.73522875e-02
-1.62110305e+00 -4.16864157e-01 6.22490942e-01 -3.56101632e-01
5.74843228e-01 4.26945776e-01 1.08864236e+00 1.23111451e+00
2.16666669e-01 6.23109639e-01 1.01967824e+00 -3.69682550e-01
-2.80057311e-01 -8.90715718e-02 -4.53114100e-02 8.45037639e-01
3.69353533e-01 -3.30117121e-02 -7.47532964e-01 -2.31270656e-01
9.65679646e-01 3.30859363e-01 -3.15425813e-01 -8.72658610e-01
-1.70198739e+00 2.37134904e-01 8.78653303e-02 -4.66845669e-02
-3.44012052e-01 1.60243466e-01 4.18351501e-01 -3.23604606e-02
3.38281333e-01 2.21266001e-02 -5.96234441e-01 -1.22748084e-01
-1.00085819e+00 4.05328661e-01 5.34943759e-01 1.39161813e+00
4.63531911e-01 -3.26555669e-01 2.01738596e-01 6.17889524e-01
2.00102717e-01 5.57500541e-01 3.42908531e-01 -1.06378937e+00
5.24896741e-01 4.59354967e-01 3.81439269e-01 -1.15105653e+00
-7.30350375e-01 -4.64383692e-01 -6.00728214e-01 6.98878691e-02
6.68409467e-01 2.11905688e-01 -6.39427125e-01 1.38503957e+00
8.03510487e-01 2.42748648e-01 -3.13677877e-01 1.08011055e+00
7.13985622e-01 -1.23848142e-02 -4.54071522e-01 -2.49590382e-01
1.45053005e+00 -1.13156760e+00 -6.87014103e-01 -3.67827654e-01
2.99849689e-01 -1.09307063e+00 6.81446254e-01 6.08919859e-01
-1.37256026e+00 -9.94031370e-01 -1.01782298e+00 7.64185116e-02
-4.68109660e-02 2.68263876e-01 2.13702098e-01 6.61644161e-01
-7.24739850e-01 6.31719053e-01 -9.02022660e-01 -6.50926530e-01
-1.17021613e-01 4.24377143e-01 -8.15429986e-01 -4.44598719e-02
-7.60630548e-01 1.04960835e+00 2.56578922e-01 1.92676425e-01
-5.48973501e-01 -4.69773352e-01 -7.87496805e-01 -4.03334349e-01
6.93320692e-01 -1.05593383e+00 1.09704673e+00 -4.16516662e-01
-1.31047380e+00 1.11245108e+00 -1.92482561e-01 -1.08484328e-01
1.12982011e+00 -8.77802789e-01 -2.79887259e-01 6.13859296e-02
2.04601273e-01 4.52289432e-01 6.53119087e-01 -1.51526141e+00
-5.37547231e-01 -7.28851914e-01 -2.32143208e-01 4.26298916e-01
2.23820861e-02 -2.62411356e-01 -1.19870591e+00 -5.69458604e-01
9.42371905e-01 -1.31003964e+00 -4.13334399e-01 2.19987303e-01
-3.59890014e-01 2.02524498e-01 5.01834869e-01 -8.22107732e-01
1.05162489e+00 -1.73025000e+00 3.53703886e-01 1.86504707e-01
1.82690024e-01 -6.57596663e-02 3.10944676e-01 2.49474391e-01
-1.31459370e-01 -3.37643534e-01 3.86092782e-01 -7.42458045e-01
-2.55886521e-02 1.25977471e-01 1.06142975e-01 1.01134527e+00
-3.96061927e-01 4.89113033e-01 -7.81151474e-01 -7.03775585e-01
3.39870244e-01 4.37040508e-01 -5.71151733e-01 1.61340669e-01
2.90088773e-01 6.14947259e-01 -2.26166055e-01 7.97104299e-01
9.13936555e-01 -1.79675907e-01 3.03789526e-01 -4.93608057e-01
-4.91153076e-02 -7.12094903e-02 -1.90234494e+00 2.26840162e+00
-2.10414723e-01 -2.68024188e-02 -1.57499194e-01 -5.17494678e-01
9.35899377e-01 4.85298008e-01 8.40835214e-01 -1.41589835e-01
4.61632647e-02 3.45345289e-01 -5.35931051e-01 -2.66904235e-01
5.62592626e-01 1.31853029e-01 -9.88931432e-02 2.14316268e-02
4.24818136e-02 -1.73010245e-01 -8.86224210e-02 -3.33525687e-02
7.67153323e-01 9.25281167e-01 7.59977102e-01 -3.21452171e-02
6.76404297e-01 -2.73498595e-01 7.25597680e-01 3.41636777e-01
-4.49560225e-01 9.79597807e-01 1.40001997e-01 -6.78166330e-01
-9.58257258e-01 -1.15831864e+00 1.66684493e-01 7.20433354e-01
4.31915939e-01 -7.61885107e-01 -5.46495557e-01 -8.61631930e-01
6.04551435e-02 6.87669367e-02 -4.93668556e-01 2.10237443e-01
-9.30986524e-01 -3.97873461e-01 1.38072208e-01 7.00963795e-01
4.62269306e-01 -6.79054409e-02 -7.43041933e-01 1.06068462e-01
-5.67448676e-01 -1.50280917e+00 -7.57952571e-01 -1.02173269e-01
-1.08984172e+00 -1.30779803e+00 -8.64365041e-01 -4.97697204e-01
7.18477011e-01 5.72131634e-01 1.35570490e+00 1.10527843e-01
-1.62301674e-01 7.04171717e-01 -2.61048734e-01 -1.73998341e-01
1.78773664e-02 -1.59906328e-01 5.39764524e-01 -2.21474662e-01
1.13572352e-01 -5.99712551e-01 -7.26471424e-01 9.00447965e-01
-1.49195448e-01 4.18455638e-02 2.81748235e-01 7.42725194e-01
1.00929987e+00 -3.21485460e-01 4.10955362e-02 -6.92478657e-01
-2.01503053e-01 1.27966180e-01 -6.51992679e-01 3.48047823e-01
-4.43421900e-01 -1.88843295e-01 2.21245199e-01 -3.00143003e-01
-8.37270319e-01 7.85584509e-01 -1.81381837e-01 -7.41263509e-01
-2.30716676e-01 -2.15780094e-01 -2.41425216e-01 -2.20566615e-01
6.31825626e-01 7.63117298e-02 -1.88185368e-02 -6.37053490e-01
1.46159336e-01 1.46763474e-01 7.56765187e-01 -5.67173660e-01
9.41229761e-01 6.52575135e-01 2.28270009e-01 -4.99957383e-01
-1.04705667e+00 -1.03531504e+00 -1.38450539e+00 -6.42578006e-01
7.84622431e-01 -1.38161492e+00 -8.93525600e-01 2.67903566e-01
-1.28219664e+00 4.55064446e-01 4.17541303e-02 7.25821674e-01
-8.89041245e-01 8.17544162e-01 -2.86211610e-01 -7.92919874e-01
-2.51818866e-01 -1.28972352e+00 1.55680859e+00 -1.52119815e-01
-3.83800417e-01 -6.81253850e-01 1.59488380e-01 5.17145634e-01
-2.67184228e-01 6.60742521e-01 -1.34261176e-01 -3.79214376e-01
-5.92049539e-01 -5.21309733e-01 2.63536513e-01 -8.63023847e-02
-2.12945610e-01 -1.02636985e-01 -8.18135738e-01 -5.08322477e-01
-5.52039854e-02 -2.15142574e-02 3.28050762e-01 3.11664522e-01
8.21318746e-01 1.46226034e-01 -6.05769575e-01 5.41679204e-01
1.30386853e+00 -3.76569033e-01 5.51558614e-01 5.84634244e-01
7.47163296e-01 5.98808646e-01 1.07243729e+00 7.57205844e-01
5.41784286e-01 1.35789323e+00 6.93135440e-01 1.96664155e-01
9.34752636e-03 -2.40917563e-01 3.71406525e-01 1.05392754e+00
-5.97749472e-01 2.22213462e-01 -8.04837584e-01 2.85578698e-01
-2.08935285e+00 -1.03758395e+00 -4.86181796e-01 2.56809354e+00
3.96927238e-01 4.08506662e-01 6.56702876e-01 2.10005626e-01
6.89301729e-01 1.57974847e-02 -2.47556046e-01 3.05884570e-01
1.35571912e-01 -1.10675544e-01 7.33505726e-01 3.58429968e-01
-1.19433033e+00 5.86869955e-01 6.33104801e+00 4.16095793e-01
-5.04729033e-01 1.77822173e-01 1.13939703e-01 -2.52104163e-01
1.97085157e-01 -1.87378109e-01 -1.27722740e+00 1.81348458e-01
4.52638865e-01 3.35971087e-01 -9.97813866e-02 1.02679431e+00
-3.57585102e-02 -2.90802211e-01 -1.15701854e+00 1.37791765e+00
3.96703482e-01 -9.66005325e-01 -3.91479701e-01 1.53457168e-02
6.38453364e-01 -3.91923666e-01 -2.57063240e-01 2.05525793e-02
-8.35579708e-02 -4.76929188e-01 1.09942508e+00 6.48953557e-01
5.76147974e-01 -8.63415241e-01 7.16243327e-01 6.96266830e-01
-1.61437500e+00 2.46380404e-01 -3.00661087e-01 1.23912863e-01
3.00008297e-01 5.13802290e-01 -6.84658885e-01 1.06791985e+00
9.20322299e-01 8.53837430e-01 -7.81585276e-01 1.15909481e+00
-8.16787928e-02 -2.39215329e-01 -3.26294065e-01 3.62032175e-01
-2.78506964e-01 -1.95124187e-02 5.60745537e-01 9.18897092e-01
6.30163789e-01 -1.03797302e-01 8.37668121e-01 9.21571255e-02
3.27891678e-01 2.09936090e-02 -5.58600962e-01 8.66840184e-01
3.53186607e-01 1.22913253e+00 -8.07802558e-01 -3.74698788e-01
-4.24201220e-01 1.39557385e+00 2.31427476e-01 -1.74366474e-01
-1.32948697e+00 2.90789098e-01 4.15071726e-01 4.06055421e-01
3.01371396e-01 -5.36401331e-01 -2.16352716e-01 -1.46161377e+00
3.25985193e-01 -8.98423612e-01 6.10032320e-01 -8.22624803e-01
-9.64854777e-01 5.47766447e-01 2.10504204e-01 -1.87772799e+00
-3.00485760e-01 -4.98163581e-01 -2.38554999e-02 5.75667262e-01
-1.04269350e+00 -1.37326050e+00 -6.93725467e-01 6.19399548e-01
7.32277095e-01 2.38177836e-01 9.05769646e-01 5.46668947e-01
-4.18106288e-01 6.51767194e-01 -2.90544093e-01 -1.48183435e-01
1.21299148e+00 -1.20520651e+00 4.49878454e-01 9.96461689e-01
1.72583088e-01 5.07221341e-01 8.91458571e-01 -7.36668348e-01
-1.53098643e+00 -7.44215131e-01 7.56749570e-01 -1.01466167e+00
1.32080019e-01 -3.57978165e-01 -4.25624192e-01 9.47539568e-01
-8.89661610e-02 3.26127917e-01 5.15595913e-01 9.05974805e-02
-2.99114317e-01 -7.27761388e-02 -1.07328379e+00 5.26621521e-01
1.50015044e+00 -5.40420860e-02 -4.11881149e-01 4.62112188e-01
5.05395651e-01 -1.31934381e+00 -1.11523271e+00 4.11942482e-01
9.02942300e-01 -1.32914937e+00 1.56631112e+00 -2.67033905e-01
-1.04363427e-01 -5.36121726e-01 -4.13297057e-01 -9.46104228e-01
-2.02933937e-01 -7.58146286e-01 -2.24267811e-01 8.25025797e-01
-1.53524831e-01 -1.74304351e-01 9.80252624e-01 3.56472880e-01
-1.28016278e-01 -6.32495522e-01 -9.67680395e-01 -1.03593242e+00
-6.83039427e-01 -3.53327394e-01 3.79688531e-01 7.90164292e-01
-3.60536486e-01 1.81448087e-01 -8.92872274e-01 5.07458985e-01
1.04785812e+00 2.52475500e-01 1.45768797e+00 -1.31395972e+00
-3.15503001e-01 7.85915107e-02 -6.37130678e-01 -1.18703115e+00
-1.36728540e-01 -3.56821626e-01 6.97375555e-03 -1.13400257e+00
2.05043778e-01 5.18054329e-02 2.10065112e-01 8.74475539e-02
-2.66570091e-01 3.87412310e-01 4.39028323e-01 2.02154130e-01
-9.51056302e-01 2.72738904e-01 1.34618509e+00 3.82626086e-01
1.74827054e-02 3.52720529e-01 -1.08557619e-01 1.16215503e+00
3.57927382e-01 -4.88018274e-01 -1.08581875e-02 -1.90955952e-01
1.59347191e-01 3.79787475e-01 6.36498988e-01 -1.35824227e+00
3.29871893e-01 9.40172747e-03 8.66874278e-01 -1.40598261e+00
7.57095933e-01 -1.17449427e+00 6.18730187e-01 6.68279290e-01
2.18102008e-01 6.26546204e-01 1.32171335e-02 7.72005916e-01
-2.85914354e-02 5.19972574e-03 8.64528298e-01 -6.23539984e-01
-7.39644706e-01 3.77018422e-01 2.06508785e-01 -9.69090238e-02
1.19321024e+00 -6.30999506e-01 1.94755405e-01 -4.65708911e-01
-1.12904871e+00 1.44215509e-01 7.54107833e-01 3.93683702e-01
7.36103535e-01 -1.69992566e+00 -5.20558953e-01 1.41323879e-01
1.42324433e-01 2.44150832e-02 3.43962193e-01 1.12996173e+00
-5.75615287e-01 1.96636409e-01 -3.51509064e-01 -1.06007147e+00
-1.77559865e+00 5.08188009e-01 2.56350070e-01 -4.33424979e-01
-6.66523993e-01 6.12001240e-01 -2.02795401e-01 -5.57162762e-01
3.64758909e-01 -1.28412813e-01 -6.07467480e-02 1.26552224e-01
4.02654111e-01 7.76859105e-01 2.17059568e-01 -9.84602749e-01
-6.19487464e-01 1.18353581e+00 2.18096420e-01 1.44714806e-02
1.23136640e+00 -4.87282395e-01 1.82313412e-01 4.51220870e-01
1.11764312e+00 4.00145262e-01 -1.16663778e+00 -2.25921720e-01
-1.32796124e-01 -8.64100397e-01 -4.29075211e-01 -4.15231109e-01
-8.76361310e-01 6.28377140e-01 7.71796703e-01 -2.07478270e-01
9.40772772e-01 -1.87482417e-01 8.57037544e-01 1.96877688e-01
6.92166924e-01 -1.18216312e+00 4.72835422e-01 4.32252437e-01
9.94063199e-01 -1.31368637e+00 6.89673126e-01 -9.00643945e-01
-5.03894567e-01 1.32473135e+00 9.35756505e-01 -1.82695046e-01
5.05208194e-01 1.06522717e-01 -1.53589085e-01 -2.82073230e-01
-3.43424022e-01 -1.73582658e-01 5.27291894e-01 4.95480627e-01
5.10460734e-01 -1.50117218e-01 -1.16979785e-01 3.29845101e-01
-1.57861128e-01 -1.73974380e-01 3.99182558e-01 1.00201285e+00
-3.67309868e-01 -1.18698788e+00 -1.07093990e+00 -1.30994052e-01
-4.55356091e-01 4.63726521e-01 -3.68397385e-02 1.14764595e+00
1.37889117e-01 5.03664672e-01 -2.56988257e-01 -4.67663258e-01
9.61480677e-01 1.65175825e-01 9.18345571e-01 -5.33086360e-01
-6.27879083e-01 6.68103337e-01 9.60830152e-02 -1.17808557e+00
-7.92269289e-01 -1.01450634e+00 -9.86454904e-01 -3.48379582e-01
-5.69292068e-01 -3.14709902e-01 3.99541795e-01 6.30805850e-01
3.04818958e-01 4.12787020e-01 5.10739326e-01 -1.22966301e+00
-7.67727971e-01 -5.98309100e-01 -4.46670592e-01 6.33943498e-01
5.39024696e-02 -1.09189332e+00 -2.51226332e-02 1.72186375e-01]
|
[7.064300537109375, -1.0143779516220093]
|
77b0117d-703a-4b5f-984d-a5b2ffb2f965
|
information-retrieval-for-label-noise
|
2203.06408
| null |
https://arxiv.org/abs/2203.06408v1
|
https://arxiv.org/pdf/2203.06408v1.pdf
|
Information retrieval for label noise document ranking by bag sampling and group-wise loss
|
Long Document retrieval (DR) has always been a tremendous challenge for reading comprehension and information retrieval. The pre-training model has achieved good results in the retrieval stage and Ranking for long documents in recent years. However, there is still some crucial problem in long document ranking, such as data label noises, long document representations, negative data Unbalanced sampling, etc. To eliminate the noise of labeled data and to be able to sample the long documents in the search reasonably negatively, we propose the bag sampling method and the group-wise Localized Contrastive Estimation(LCE) method. We use the head middle tail passage for the long document to encode the long document, and in the retrieval, stage Use dense retrieval to generate the candidate's data. The retrieval data is divided into multiple bags at the ranking stage, and negative samples are selected in each bag. After sampling, two losses are combined. The first loss is LCE. To fit bag sampling well, after query and document are encoded, the global features of each group are extracted by convolutional layer and max-pooling to improve the model's resistance to the impact of labeling noise, finally, calculate the LCE group-wise loss. Notably, our model shows excellent performance on the MS MARCO Long document ranking leaderboard.
|
['Fan Wang', 'Xing Hu', 'Jiajia Ding', 'Chunyu Li']
|
2022-03-12
| null | null | null | null |
['document-ranking']
|
['natural-language-processing']
|
[ 5.26084080e-02 -4.91310954e-01 -2.62395829e-01 -5.15047729e-01
-1.43281710e+00 -3.60171765e-01 5.16358376e-01 4.03668016e-01
-5.34845650e-01 5.41572571e-01 4.34444815e-01 2.63695300e-01
-3.78007889e-01 -6.27059221e-01 -4.53218579e-01 -9.16110039e-01
2.23658279e-01 6.12729907e-01 2.79993117e-01 -2.59936661e-01
6.09701514e-01 2.49549091e-01 -1.59554052e+00 7.34445989e-01
9.49685276e-01 1.19736660e+00 3.45146149e-01 6.13254189e-01
-4.29140508e-01 6.88615620e-01 -8.31432700e-01 -2.76172340e-01
1.27679959e-01 -1.10255316e-01 -7.95852542e-01 -2.21326485e-01
4.88643706e-01 -6.92235529e-01 -2.08718181e-01 9.01991129e-01
1.23580945e+00 4.02470380e-01 7.12416589e-01 -8.43724251e-01
-4.76814538e-01 4.87782210e-01 -7.63987184e-01 1.13154881e-01
1.94685385e-01 -2.60413527e-01 1.18480504e+00 -1.23415089e+00
4.66185182e-01 1.77016950e+00 1.48660794e-01 6.40138209e-01
-8.52681816e-01 -7.48674810e-01 2.63987631e-01 1.61775306e-01
-1.30708873e+00 -4.10336792e-01 7.53533363e-01 -3.20273042e-01
5.10118365e-01 2.95657426e-01 3.92174840e-01 9.19603884e-01
1.68506190e-01 1.39195597e+00 6.46298468e-01 -5.23195267e-01
-1.12202555e-01 -2.71308385e-02 8.89335096e-01 3.12615424e-01
-4.64864820e-02 7.16774957e-03 -3.73096704e-01 -6.07619584e-02
2.10510641e-01 2.38896459e-01 -3.13000470e-01 1.91596635e-02
-8.73954713e-01 8.31206799e-01 7.05782592e-01 6.49115667e-02
-1.89458147e-01 2.13988364e-01 5.84548831e-01 2.98392266e-01
6.97338462e-01 2.03123167e-01 -2.48080283e-01 1.32837489e-01
-1.06157458e+00 6.69236362e-01 4.22908396e-01 7.02370048e-01
4.54991162e-01 -6.65338337e-01 -1.05145597e+00 1.36169589e+00
4.57053244e-01 3.65900338e-01 8.49641919e-01 -3.17521334e-01
8.96836936e-01 4.35350269e-01 1.43386602e-01 -1.16946030e+00
-1.87171131e-01 -8.97378445e-01 -1.03355718e+00 -4.00195211e-01
-2.22209878e-02 3.38573366e-01 -1.10117018e+00 1.45058751e+00
-2.29335204e-02 -4.71308500e-01 1.82559136e-02 1.01338756e+00
1.07869720e+00 1.15025556e+00 1.82580017e-02 -4.16931629e-01
1.22439778e+00 -1.31590712e+00 -7.64652550e-01 -3.42853725e-01
6.85106277e-01 -8.87946665e-01 1.29431283e+00 4.27154124e-01
-1.18452084e+00 -7.69015610e-01 -1.01554859e+00 -5.65037310e-01
-5.84167540e-01 4.15536582e-01 4.05322164e-01 2.64956355e-01
-7.53160119e-01 2.45802388e-01 -2.39374116e-01 1.40022695e-01
4.54056323e-01 2.34170750e-01 8.97450298e-02 -8.67024720e-01
-1.53923547e+00 6.63048208e-01 4.10305172e-01 4.64241385e-01
-9.32509303e-01 -3.70261014e-01 -5.71564078e-01 4.02552158e-01
2.78346222e-02 -4.46192384e-01 1.04087794e+00 -7.06390023e-01
-1.03583467e+00 7.62668729e-01 -2.84574926e-01 -1.36969343e-01
5.43266356e-01 -4.49792385e-01 -2.89846033e-01 -8.57202411e-02
-1.00570954e-02 8.46576750e-01 8.24359238e-01 -1.20616519e+00
-4.53304738e-01 -5.48142076e-01 -6.36511222e-02 6.86998665e-01
-4.38305587e-01 1.05127193e-01 -8.35702956e-01 -6.39424622e-01
4.77491230e-01 -6.52838528e-01 7.47582763e-02 -4.46733624e-01
-3.08329672e-01 -6.25002027e-01 5.30116558e-01 -6.93226695e-01
1.69742775e+00 -2.47875071e+00 -4.86833975e-02 2.82644451e-01
4.40499522e-02 1.63440987e-01 -5.49187481e-01 3.08630884e-01
6.34547099e-02 8.17416161e-02 2.53466129e-01 -4.81948346e-01
2.15940978e-02 -2.00671539e-01 -7.44845510e-01 -2.34534424e-02
-2.37982050e-02 9.04021204e-01 -9.16654229e-01 -5.93841374e-01
-2.62326568e-01 3.35518241e-01 -3.95108402e-01 2.96874940e-01
-3.27228665e-01 -8.74401033e-02 -6.17193401e-01 5.23167908e-01
9.22639847e-01 -1.33169815e-01 -5.25281310e-01 -1.66847065e-01
3.76500994e-01 5.31043589e-01 -9.88170981e-01 1.51135302e+00
-3.70214134e-01 3.91333282e-01 -1.59482613e-01 -7.02510297e-01
9.77311969e-01 -4.00108807e-02 -1.66839100e-02 -1.01132751e+00
-2.13510860e-02 4.39966977e-01 -3.02694321e-01 -5.31298399e-01
9.82556343e-01 1.55812085e-01 -6.29161373e-02 4.04411912e-01
-2.74777859e-01 -4.46606949e-02 4.25691307e-01 4.66038674e-01
5.47547817e-01 -1.65052727e-01 -4.56956238e-01 4.74123247e-02
6.08854413e-01 -2.27424771e-01 2.65866965e-01 8.95426691e-01
2.20719755e-01 1.00844514e+00 3.79787773e-01 -1.07217804e-01
-6.20160758e-01 -6.23626232e-01 -1.59249276e-01 1.45498991e+00
2.77316183e-01 -3.98568988e-01 -4.25836205e-01 -7.39955246e-01
-5.83186783e-02 4.73712713e-01 -3.37317586e-01 -7.63500929e-01
-5.54093063e-01 -1.01822269e+00 2.59843498e-01 3.32434267e-01
4.29571003e-01 -1.14730155e+00 4.33713049e-02 2.64808536e-01
-3.72107446e-01 -2.83041775e-01 -5.80769718e-01 2.73496628e-01
-7.49485373e-01 -7.19834805e-01 -1.19086993e+00 -9.82958913e-01
8.93811703e-01 4.62607354e-01 1.20979941e+00 2.99437344e-01
-2.81215638e-01 -1.10728361e-01 -5.41789711e-01 -2.80644536e-01
2.80022994e-02 3.04283738e-01 -3.53035927e-01 -2.37901926e-01
6.35303915e-01 1.45546466e-01 -8.29914272e-01 2.98642367e-01
-9.84843969e-01 -2.23996505e-01 7.15672314e-01 1.29633260e+00
7.57762611e-01 3.60742211e-01 7.31510580e-01 -6.00196183e-01
1.22252178e+00 -2.42526054e-01 -4.28154320e-01 5.33301413e-01
-7.44930089e-01 -1.32653445e-01 3.98486704e-01 -3.77881080e-01
-8.09778512e-01 -6.76637292e-01 -1.48842946e-01 -6.90219253e-02
4.78750318e-01 6.96791828e-01 -2.62843192e-01 3.07921350e-01
4.10237610e-01 3.11665684e-01 -2.59168416e-01 -5.95229149e-01
1.85192525e-01 1.10325825e+00 -3.98260728e-02 -4.55901444e-01
3.42502207e-01 -1.36773691e-01 -3.21262509e-01 -3.59189361e-01
-1.42045987e+00 -8.13119948e-01 -1.90176442e-01 -3.28065753e-02
5.89232802e-01 -9.79004860e-01 -3.17971587e-01 7.17303395e-01
-1.21654510e+00 1.46770865e-01 -1.42861530e-01 4.56912547e-01
6.88484088e-02 1.65352896e-01 -6.28668666e-01 -9.67431068e-01
-7.97809601e-01 -1.24942791e+00 1.32955837e+00 3.56990099e-01
1.84390664e-01 -4.30200279e-01 -8.45197141e-02 4.09525305e-01
4.00086910e-01 -5.55344343e-01 1.20920062e+00 -7.66984940e-01
-6.63724542e-01 -8.74220729e-01 -5.93327105e-01 7.58431017e-01
-3.81849289e-01 -2.15497300e-01 -1.06227565e+00 -7.00055599e-01
-2.25795090e-01 -8.52274239e-01 1.56169176e+00 4.94720995e-01
1.60117900e+00 -1.76132753e-01 -3.28109145e-01 2.81268626e-01
1.18990779e+00 2.06153482e-01 7.69621611e-01 1.22293361e-01
6.49504602e-01 7.75576472e-01 1.10429049e+00 6.28524423e-02
9.59323272e-02 4.21891510e-01 1.89700857e-01 1.52087927e-01
-1.66297138e-01 -3.42165291e-01 3.86583716e-01 1.03981221e+00
5.12353301e-01 -7.59426594e-01 -5.63010156e-01 3.99010420e-01
-1.62411284e+00 -8.84282053e-01 4.29670066e-02 2.39453864e+00
9.46614087e-01 4.95055556e-01 -4.30804163e-01 1.12554662e-01
5.94327569e-01 4.84351754e-01 -2.97021449e-01 -5.00879101e-02
-2.84490079e-01 -4.33412492e-02 7.34714717e-02 4.83873367e-01
-1.02275479e+00 8.34864557e-01 5.80427217e+00 1.57123566e+00
-8.17541420e-01 -1.57580674e-01 9.99812543e-01 -3.52272749e-01
-4.81669992e-01 -1.32287830e-01 -1.49398017e+00 5.33710718e-01
6.09227955e-01 1.84383273e-01 2.92127103e-01 6.41532123e-01
3.67038473e-02 -2.84992248e-01 -8.78071845e-01 1.06266701e+00
3.45501065e-01 -8.46150517e-01 4.82682884e-01 -1.99034199e-01
7.21372545e-01 -1.92084610e-01 2.67953604e-01 8.68991792e-01
-1.84889495e-01 -1.07917130e+00 4.29602832e-01 7.21654356e-01
4.79249746e-01 -9.50827062e-01 1.08002722e+00 5.61214149e-01
-7.66562581e-01 -3.46807867e-01 -9.21793282e-01 3.74635398e-01
-1.83926657e-01 1.07485473e+00 -4.26002890e-01 3.45110565e-01
4.32017356e-01 6.04698837e-01 -8.41650367e-01 1.27579606e+00
7.59967417e-02 3.71422738e-01 -2.13385835e-01 -3.25890481e-01
4.82328713e-01 -2.53022045e-01 3.71139973e-01 1.18160772e+00
2.44132653e-01 -1.69573233e-01 2.92903155e-01 4.54300821e-01
-4.26629901e-01 4.13663298e-01 -8.25969875e-02 -1.66522413e-01
3.52460861e-01 1.06261206e+00 -2.66075581e-01 -4.60669816e-01
1.82155311e-01 9.49528158e-01 2.83065170e-01 5.70417643e-01
-3.55733395e-01 -7.93627858e-01 -2.56874830e-01 -1.34920850e-01
-1.25343144e-01 2.33269706e-01 -6.40805289e-02 -9.68466222e-01
4.70778525e-01 -7.91140199e-01 5.93408763e-01 -9.58674014e-01
-1.36171401e+00 4.33708429e-01 -1.75624028e-01 -1.25896978e+00
-1.14297248e-01 -2.75422037e-01 -2.74905235e-01 1.31761038e+00
-1.63020301e+00 -1.00801766e+00 -4.44459736e-01 2.50153124e-01
7.70896733e-01 -2.54627436e-01 6.15758836e-01 7.20961273e-01
-4.64122266e-01 7.55853832e-01 7.96323061e-01 -2.22386718e-02
9.76877511e-01 -1.06469846e+00 -1.98354498e-01 4.45226699e-01
-1.70536101e-01 8.76339734e-01 1.82329848e-01 -6.39200032e-01
-9.62650061e-01 -9.83148277e-01 1.08127511e+00 -1.57815665e-01
-6.78712130e-02 -4.57145393e-01 -9.67988253e-01 1.84695527e-01
-3.24191809e-01 -6.91973139e-03 4.62890387e-01 5.36012501e-02
-3.30267042e-01 -5.95349669e-01 -9.36084867e-01 4.15967256e-01
4.08455312e-01 -6.37755036e-01 -5.32000065e-01 6.89518929e-01
8.32424521e-01 -3.26348007e-01 -3.32733393e-01 3.64991128e-01
7.23142684e-01 -5.92844784e-01 1.06785667e+00 -5.45725405e-01
6.50423646e-01 -1.99488908e-01 -9.46267992e-02 -1.22980773e+00
-3.06653470e-01 -5.00669330e-02 -5.28847501e-02 1.57251358e+00
3.71739060e-01 -1.12495787e-01 6.23226404e-01 3.00279260e-01
-1.27343133e-01 -1.05238235e+00 -5.45781195e-01 -4.71908689e-01
2.44570181e-01 -1.03883788e-01 5.50037444e-01 4.93136287e-01
-3.65645409e-01 5.87405920e-01 -3.61747324e-01 -3.56927663e-01
3.55523854e-01 3.14966440e-01 4.92788583e-01 -1.31468892e+00
-3.79482619e-02 -5.75061321e-01 2.40410283e-01 -1.80513120e+00
7.80486874e-03 -9.75234509e-01 3.05303991e-01 -1.80999172e+00
6.61388516e-01 -6.80357456e-01 -6.61428034e-01 -2.17240527e-02
-5.58670700e-01 1.01213288e-02 -1.29115283e-01 3.73072833e-01
-9.62270439e-01 7.62272537e-01 1.54226327e+00 -4.45730925e-01
-1.72609940e-01 2.17129990e-01 -5.98370314e-01 3.08341622e-01
3.46133322e-01 -5.40480614e-01 -4.00582254e-01 -5.13599575e-01
4.83160436e-01 -8.75148475e-02 1.18609257e-01 -6.68624520e-01
2.16019988e-01 1.43703282e-01 8.20197642e-01 -1.36291313e+00
4.06015933e-01 -6.52604103e-01 -5.74829757e-01 2.61845261e-01
-1.07274437e+00 -1.86942205e-01 -2.34835610e-01 5.52386582e-01
-6.43259823e-01 -7.35699058e-01 7.16927350e-01 -1.30104110e-01
-3.66984844e-01 5.52098572e-01 1.47343978e-01 1.47479206e-01
4.27829921e-01 3.45282368e-02 -5.38989842e-01 -2.31339142e-01
-3.71750265e-01 8.52817893e-01 2.13672984e-02 5.17218053e-01
7.34003484e-01 -1.44800174e+00 -7.98473597e-01 2.65706956e-01
3.43861550e-01 4.75141704e-01 4.27160889e-01 5.08804917e-01
-3.32431763e-01 5.74117720e-01 3.41350704e-01 -3.37609351e-01
-1.21815789e+00 3.92400235e-01 1.27664506e-01 -8.86833787e-01
-2.84917146e-01 1.28720617e+00 1.99767455e-01 -3.00532490e-01
7.64516413e-01 7.68902376e-02 -6.65931940e-01 6.06175423e-01
9.15748775e-01 2.71762997e-01 1.72850817e-01 -3.42842609e-01
-6.40242845e-02 5.57053506e-01 -8.05996835e-01 -1.10440552e-01
1.09491181e+00 -3.34774286e-01 -5.14527917e-01 5.06678998e-01
1.73016524e+00 7.84904808e-02 -6.17661059e-01 -1.89699233e-01
2.65963487e-02 -5.68649411e-01 5.16113460e-01 -1.07550192e+00
-8.81726384e-01 1.35373116e+00 8.98715496e-01 2.24960625e-01
1.02850175e+00 -1.89682499e-01 1.04348981e+00 6.16841376e-01
2.64146384e-02 -1.23604143e+00 2.19387203e-01 8.39372396e-01
1.20928407e+00 -1.16013670e+00 7.55518004e-02 -7.61933858e-03
-3.66428763e-01 1.05403316e+00 6.12966835e-01 -1.28774447e-02
4.43335235e-01 -4.33374614e-01 -7.84287825e-02 -1.47847176e-01
-6.06885016e-01 1.10221587e-01 6.54440224e-01 1.27177328e-01
5.88180244e-01 -2.88418144e-01 -5.51089346e-01 6.75202608e-01
-4.17192318e-02 -1.61102563e-01 -2.54519861e-02 6.56255364e-01
-6.30239904e-01 -9.85474467e-01 -2.73214698e-01 9.53780293e-01
-6.53637528e-01 -5.49073398e-01 -2.36030534e-01 1.24561712e-01
-1.54371664e-01 7.95948207e-01 7.32076019e-02 -1.30292252e-01
2.52702355e-01 2.22134382e-01 2.46822119e-01 -5.57667851e-01
-5.81285477e-01 3.36118698e-01 1.78572442e-02 -3.54485154e-01
5.49575407e-03 -1.84747711e-01 -1.05625951e+00 1.48611948e-01
-9.12103474e-01 5.46005726e-01 6.79438174e-01 6.66146994e-01
2.46505916e-01 5.74444413e-01 8.64998281e-01 -5.57150304e-01
-1.04799926e+00 -1.31424665e+00 -7.54373491e-01 5.11662424e-01
5.79179525e-01 -4.71123397e-01 -6.71690166e-01 -4.49240297e-01]
|
[11.485816955566406, 7.59336519241333]
|
868742a2-5aef-40d1-a656-d40b6c8ac37f
|
mask-and-infill-applying-masked-language
|
1908.08039
| null |
https://arxiv.org/abs/1908.08039v1
|
https://arxiv.org/pdf/1908.08039v1.pdf
|
"Mask and Infill" : Applying Masked Language Model to Sentiment Transfer
|
This paper focuses on the task of sentiment transfer on non-parallel text, which modifies sentiment attributes (e.g., positive or negative) of sentences while preserving their attribute-independent content. Due to the limited capability of RNNbased encoder-decoder structure to capture deep and long-range dependencies among words, previous works can hardly generate satisfactory sentences from scratch. When humans convert the sentiment attribute of a sentence, a simple but effective approach is to only replace the original sentimental tokens in the sentence with target sentimental expressions, instead of building a new sentence from scratch. Such a process is very similar to the task of Text Infilling or Cloze, which could be handled by a deep bidirectional Masked Language Model (e.g. BERT). So we propose a two step approach "Mask and Infill". In the mask step, we separate style from content by masking the positions of sentimental tokens. In the infill step, we retrofit MLM to Attribute Conditional MLM, to infill the masked positions by predicting words or phrases conditioned on the context1 and target sentiment. We evaluate our model on two review datasets with quantitative, qualitative, and human evaluations. Experimental results demonstrate that our models improve state-of-the-art performance.
|
['Xing Wu', 'Tao Zhang', 'Liangjun Zang', 'Songlin Hu', 'Jizhong Han']
|
2019-08-21
| null | null | null | null |
['text-infilling']
|
['natural-language-processing']
|
[ 4.37405229e-01 2.15856835e-01 -1.10534832e-01 -9.23448026e-01
-7.03012109e-01 -6.26471043e-01 5.69489360e-01 4.81861942e-02
-5.39821804e-01 9.96870935e-01 7.10566878e-01 -3.91017467e-01
7.52237856e-01 -8.55497718e-01 -8.29189539e-01 -5.66405356e-01
7.42505550e-01 2.17676222e-01 -1.68093190e-01 -6.86830163e-01
2.88709283e-01 -7.69371390e-02 -1.15820360e+00 9.59427357e-01
6.77201450e-01 6.39363885e-01 2.70903140e-01 5.18775403e-01
-8.08381557e-01 9.90208030e-01 -8.32186103e-01 -9.41918015e-01
5.47128059e-02 -7.55544662e-01 -7.26995349e-01 1.87362090e-01
1.61416885e-02 -1.70683920e-01 1.19158905e-03 1.24079204e+00
4.72478032e-01 -1.46777004e-01 7.67060399e-01 -8.31688166e-01
-1.30565321e+00 1.18079472e+00 -6.61235094e-01 -3.16510081e-01
3.38397086e-01 -1.89168841e-01 1.20031691e+00 -1.18978167e+00
4.31081086e-01 1.36703825e+00 4.84683931e-01 9.79572833e-01
-1.18516409e+00 -6.47312343e-01 5.41189313e-01 -2.14561865e-01
-1.06254363e+00 -4.82316494e-01 7.80866027e-01 -2.94476718e-01
8.41266036e-01 3.37679982e-01 4.00941402e-01 1.26402962e+00
4.36245650e-01 8.10790241e-01 1.21531987e+00 -5.93236387e-01
4.78625223e-02 6.27738595e-01 8.15759897e-02 1.84032619e-01
-5.26222698e-02 -4.75169808e-01 -4.69553620e-01 2.86573112e-01
1.22670978e-01 -7.93447047e-02 -6.15432858e-02 1.98390856e-01
-1.26044405e+00 8.85152340e-01 3.35163832e-01 1.29427433e-01
-3.20025802e-01 -5.40462648e-03 5.93620420e-01 4.62451160e-01
8.91711354e-01 3.56058478e-01 -7.10237622e-01 2.24924788e-01
-7.94077039e-01 1.03060290e-01 7.23903477e-01 1.16520476e+00
8.09289694e-01 -4.82014753e-02 -5.36341429e-01 8.98361206e-01
3.17588270e-01 7.34923780e-01 6.42480671e-01 -2.00644717e-01
6.83747768e-01 5.36626875e-01 3.83097455e-02 -7.41662204e-01
-2.24673182e-01 -4.48289633e-01 -1.08469462e+00 -8.86009708e-02
-1.31893754e-01 -4.92861360e-01 -1.05541050e+00 1.71599770e+00
8.54934305e-02 -5.96193075e-01 4.26833987e-01 7.70252287e-01
1.06690443e+00 9.59630728e-01 2.08383754e-01 -1.17794052e-01
1.42762041e+00 -1.09587502e+00 -1.01522827e+00 -6.94747746e-01
7.61308074e-01 -1.05845511e+00 1.51026428e+00 1.34418994e-01
-1.11490393e+00 -5.35836279e-01 -1.03813744e+00 -3.42740476e-01
-6.65692389e-01 2.31306225e-01 3.51257712e-01 6.94249570e-01
-9.58545685e-01 3.14549118e-01 -2.43322089e-01 1.54962689e-01
2.53658593e-01 1.97378129e-01 -2.97802389e-01 1.21039398e-01
-1.65521073e+00 9.77794349e-01 1.30643368e-01 3.68942559e-01
-3.87134045e-01 -4.97074485e-01 -1.07809496e+00 -6.66005723e-03
1.13053754e-01 -6.81997299e-01 1.23378253e+00 -1.55474293e+00
-1.61616290e+00 1.02593398e+00 -6.33659244e-01 -1.96649283e-01
3.39072466e-01 -3.38298321e-01 -3.61584127e-01 -5.53143620e-01
3.39194715e-01 7.70912051e-01 8.22831154e-01 -1.37132263e+00
-4.61299181e-01 -1.43482298e-01 1.20115668e-01 4.03361470e-01
-6.76548719e-01 2.98831224e-01 -3.45454842e-01 -9.40035582e-01
-1.58836812e-01 -8.07349801e-01 -3.23960274e-01 -5.17526388e-01
-7.37345457e-01 -5.54521568e-04 5.00834107e-01 -6.71915472e-01
1.38517976e+00 -2.12824011e+00 1.62244409e-01 9.80750471e-02
-1.52927399e-01 5.55119663e-02 -3.03703189e-01 5.03435254e-01
-2.69359976e-01 3.80600750e-01 -4.95134771e-01 -8.17362130e-01
1.20587707e-01 2.85655856e-01 -8.16184163e-01 6.63310438e-02
4.66123223e-01 1.06316185e+00 -8.95419657e-01 -3.47985536e-01
-1.71466455e-01 4.32396829e-01 -6.18423164e-01 9.75261331e-02
-3.39675903e-01 3.96932930e-01 -2.73860276e-01 3.30134034e-01
7.41370320e-01 -1.44850872e-02 5.05015925e-02 -1.84773147e-01
-1.86874345e-03 8.39262724e-01 -8.93621802e-01 1.39027524e+00
-8.84337068e-01 4.84665126e-01 -1.63576573e-01 -6.64594054e-01
1.22968698e+00 3.10462475e-01 1.37419142e-02 -5.78147233e-01
1.48365662e-01 5.54580502e-02 -1.29319116e-01 -2.94095039e-01
9.44374859e-01 -7.14973330e-01 -5.49187064e-01 5.38461685e-01
-2.13844463e-01 -4.20787871e-01 1.88479483e-01 2.66082525e-01
7.46835887e-01 1.55954182e-01 1.84454650e-01 -2.18705043e-01
8.66151154e-01 -1.61315426e-01 3.78529817e-01 4.47099328e-01
3.55613947e-01 8.01892519e-01 8.37654591e-01 -1.36707902e-01
-1.06875169e+00 -8.28409612e-01 7.90468454e-02 1.39077425e+00
-6.62438124e-02 -5.63657403e-01 -8.19845974e-01 -9.59172010e-01
-2.92369366e-01 1.14481449e+00 -9.00045514e-01 -3.57770979e-01
-5.55728555e-01 -9.82688248e-01 3.30602646e-01 5.46131492e-01
3.96904558e-01 -1.50893509e+00 1.70232236e-01 2.63070166e-01
-4.19664085e-01 -9.61467147e-01 -6.74925745e-01 2.72952616e-01
-3.65628600e-01 -2.40512490e-01 -5.08979082e-01 -1.03269815e+00
9.30574298e-01 4.72578667e-02 1.18408751e+00 -1.92228444e-02
4.52579081e-01 -3.19131821e-01 -6.30499601e-01 -6.33607328e-01
-7.78456450e-01 3.54392231e-01 -1.86970934e-01 2.62815356e-01
6.02861762e-01 -2.65205950e-01 -3.26586783e-01 -4.25004214e-03
-1.18351650e+00 2.91250408e-01 8.36439312e-01 9.05163288e-01
6.08660698e-01 -1.94244146e-01 9.08574224e-01 -1.40555525e+00
9.89550948e-01 -4.55180973e-01 -8.09841454e-02 1.92543387e-01
-4.32867885e-01 8.55636746e-02 1.03801298e+00 -4.45400357e-01
-1.28919244e+00 9.85892266e-02 -5.84192514e-01 1.14781417e-01
1.21908903e-01 7.18927622e-01 -5.60466528e-01 6.86282754e-01
4.00908768e-01 3.21401477e-01 -1.35603338e-01 -2.23861247e-01
7.07340181e-01 1.03133440e+00 4.11224961e-01 -3.26236635e-01
6.68743730e-01 4.07411426e-01 -4.13364828e-01 -4.35217798e-01
-1.43983126e+00 -1.17275357e-01 -7.15576768e-01 -5.86322397e-02
9.47158515e-01 -1.04962146e+00 -1.27872765e-01 4.04961973e-01
-1.50822759e+00 -1.76213384e-01 -5.20119786e-01 1.54519826e-01
-2.01059192e-01 1.44884497e-01 -6.73399568e-01 -5.33918738e-01
-5.29570401e-01 -9.43375587e-01 1.14708221e+00 -3.09393518e-02
-6.55376017e-01 -1.05875790e+00 -1.39807742e-02 2.83575624e-01
4.26218510e-01 -1.72887921e-01 9.92303491e-01 -6.20391130e-01
8.33133683e-02 -4.05594796e-01 -1.53178453e-01 9.01436865e-01
4.15783256e-01 -1.19723670e-01 -1.08140731e+00 -9.65383500e-02
1.88285634e-01 -3.64371449e-01 9.72982407e-01 4.37840540e-03
1.06503868e+00 -5.15407741e-01 4.09009419e-02 3.26774746e-01
1.04609871e+00 -3.54051292e-02 8.60433102e-01 3.03724051e-01
8.24999690e-01 8.90554309e-01 5.64296722e-01 3.25352252e-01
6.30034089e-01 9.88022238e-02 1.46453306e-01 -4.01767850e-01
-2.36923844e-01 -4.77007449e-01 8.69645715e-01 1.40445828e+00
5.19255757e-01 -4.96687382e-01 -4.29262161e-01 6.19355083e-01
-1.64392865e+00 -7.39029884e-01 -3.20011079e-01 1.78030574e+00
1.35075402e+00 4.24443752e-01 -4.24110025e-01 3.94801982e-02
7.28666425e-01 2.77051300e-01 -2.73017526e-01 -8.82513702e-01
-5.43804705e-01 1.33274809e-01 3.83729577e-01 8.08030963e-01
-9.02074456e-01 1.29299772e+00 5.65711546e+00 7.45849907e-01
-1.16059172e+00 1.21215284e-01 8.55120540e-01 -1.72292516e-01
-1.00432086e+00 7.31162429e-02 -1.04349685e+00 5.32266855e-01
7.47002959e-01 -4.03274074e-02 2.00610250e-01 5.51427126e-01
2.55027831e-01 1.90646768e-01 -1.12752366e+00 5.11935532e-01
4.31550622e-01 -1.08637309e+00 4.52806473e-01 -4.09472048e-01
9.47011590e-01 -3.02333593e-01 2.02559680e-01 6.24546528e-01
2.48188093e-01 -1.01872933e+00 9.72488642e-01 4.78666782e-01
8.58488441e-01 -6.98425949e-01 1.00901926e+00 2.03824252e-01
-8.68928611e-01 2.59476662e-01 -3.87778580e-01 -3.01022410e-01
2.47597992e-01 7.78512716e-01 -5.25763512e-01 3.05862248e-01
5.92053413e-01 8.73120546e-01 -5.89164972e-01 1.12609185e-01
-7.25911379e-01 5.77693462e-01 2.35249847e-01 -5.10536015e-01
2.23721117e-01 -2.34999865e-01 2.41195157e-01 1.57621145e+00
7.59803727e-02 -1.15731746e-01 -2.30918571e-01 8.54272664e-01
-4.72539902e-01 4.78660941e-01 -5.25879920e-01 3.96853350e-02
2.19082057e-01 1.41249800e+00 -5.83215594e-01 -5.72564125e-01
-6.07798934e-01 1.24998808e+00 3.24995607e-01 3.37837577e-01
-6.86243892e-01 -5.86281240e-01 4.00834441e-01 -1.75594538e-02
3.73675704e-01 1.47037745e-01 -8.64299655e-01 -1.15663695e+00
4.77864146e-01 -1.10160303e+00 -8.92968755e-03 -9.19623017e-01
-1.60973835e+00 8.62290621e-01 -4.22058225e-01 -1.25980592e+00
-7.86847249e-02 -4.68854666e-01 -6.30236745e-01 1.22028840e+00
-1.60571790e+00 -1.23932242e+00 1.85859978e-01 3.47091734e-01
6.82475507e-01 -5.48451096e-02 8.19225430e-01 2.13954270e-01
-4.60952580e-01 6.81738615e-01 4.71154042e-03 3.89745235e-01
1.02537489e+00 -1.29262078e+00 6.95513248e-01 8.78902972e-01
-1.62545055e-01 7.49441981e-01 8.16678166e-01 -8.06588531e-01
-1.02723789e+00 -1.40969205e+00 1.54808474e+00 -5.83469450e-01
7.86639035e-01 -9.34854209e-01 -1.00229764e+00 7.68061340e-01
6.38043523e-01 -4.06700969e-01 8.59958529e-01 7.00280368e-02
-2.84618884e-01 -1.00922495e-01 -7.50957847e-01 1.01012182e+00
6.96696103e-01 -5.60092747e-01 -8.47483814e-01 3.09193373e-01
1.18956721e+00 -2.77520716e-01 -3.69602293e-01 3.65196764e-01
3.39406759e-01 -5.90834796e-01 4.72885907e-01 -7.70494759e-01
1.08108139e+00 -4.10192996e-01 -2.65949517e-01 -1.74653351e+00
-2.39127010e-01 -5.41972399e-01 3.93843800e-01 1.67335665e+00
1.09692633e+00 -5.36036968e-01 5.46853185e-01 5.69220662e-01
-2.51916885e-01 -7.73696721e-01 -5.79088330e-01 -2.54190356e-01
4.19617712e-01 -4.23686028e-01 8.26749206e-01 8.47342610e-01
7.52982199e-02 9.62074876e-01 -4.50822383e-01 -1.22121155e-01
3.61058079e-02 1.94342926e-01 7.05187678e-01 -4.91973877e-01
-2.61404723e-01 -4.53120232e-01 1.90171063e-01 -1.22648478e+00
5.23727119e-01 -1.04211771e+00 3.34410042e-01 -1.67461538e+00
2.76368022e-01 -3.28490049e-01 -1.73982292e-01 5.42190850e-01
-6.11836791e-01 3.32963467e-01 1.46963403e-01 -1.41757578e-01
-3.86705935e-01 9.20437515e-01 1.40484858e+00 -2.92495131e-01
-1.17610589e-01 -4.22493694e-03 -1.21651196e+00 7.04128385e-01
9.26949143e-01 -6.88252926e-01 -3.45177531e-01 -5.85595787e-01
8.24801564e-01 -2.97334373e-01 -8.91839713e-02 -2.35380813e-01
-8.59659631e-03 -9.96529311e-02 4.40793097e-01 -6.87183440e-01
2.76020110e-01 -5.91656029e-01 -6.19697213e-01 1.71480745e-01
-8.15791368e-01 3.83611083e-01 2.58882165e-01 1.53100580e-01
-4.03164089e-01 -4.12491858e-01 4.74542201e-01 -1.03907190e-01
-2.09539369e-01 -9.10725892e-02 -6.37240171e-01 1.04017667e-01
6.51217699e-01 9.88837332e-02 -3.48231912e-01 -6.16218925e-01
-4.89444941e-01 2.47373223e-01 2.55447745e-01 7.87861168e-01
7.05577970e-01 -1.24928999e+00 -1.06261826e+00 3.85737777e-01
8.09059814e-02 7.01859295e-02 9.03487504e-02 4.25510675e-01
-1.18004896e-01 1.15401089e-01 8.24756920e-02 -1.61917478e-01
-1.10479462e+00 6.99391603e-01 1.76719636e-01 -2.10499048e-01
-1.69061795e-01 9.30825472e-01 4.86474454e-01 -7.77444959e-01
-2.82742344e-02 -3.23691905e-01 -4.22099948e-01 1.83416680e-01
5.30979037e-01 -3.24899256e-01 1.83802575e-01 -6.88960552e-01
-1.65967107e-01 4.10053015e-01 -4.31013614e-01 -3.65141690e-01
1.08958173e+00 -4.50569659e-01 -7.44114578e-01 6.97028160e-01
1.21817756e+00 5.16441941e-01 -8.69741917e-01 -1.83383793e-01
-1.71297461e-01 -5.81680387e-02 -1.33259296e-01 -8.19565892e-01
-9.43476439e-01 9.59005237e-01 -1.98533878e-01 1.02335095e-01
1.16167927e+00 1.64139941e-02 8.52998316e-01 3.50124657e-01
-3.13602209e-01 -1.18871045e+00 -9.94546805e-03 8.69018555e-01
1.11758780e+00 -1.22774816e+00 -9.60579589e-02 -4.04429913e-01
-1.14128006e+00 9.23274279e-01 7.92292416e-01 2.61923466e-02
6.39781654e-01 4.78842735e-01 5.51320791e-01 9.80641097e-02
-9.06322360e-01 3.02614328e-02 1.92560151e-01 2.03608036e-01
8.66126418e-01 1.71776146e-01 -4.57355678e-01 9.63906765e-01
-7.91950405e-01 -4.62725103e-01 7.76214242e-01 7.68550038e-01
-1.75825179e-01 -1.23513842e+00 -2.84051985e-01 4.73153710e-01
-5.83910227e-01 -6.88622713e-01 -7.14703977e-01 3.59242320e-01
8.55351985e-02 1.12252545e+00 1.44973576e-01 -4.53455597e-01
5.48126638e-01 9.52925906e-02 3.76031660e-02 -9.63746846e-01
-8.73068869e-01 7.30374902e-02 2.76164472e-01 2.68768780e-02
-2.68838078e-01 -5.95754981e-01 -1.39061618e+00 -1.72474444e-01
-1.95152819e-01 1.26581669e-01 7.59216368e-01 1.10176945e+00
1.20033942e-01 7.76285946e-01 9.06259537e-01 -4.11463499e-01
-5.96100628e-01 -1.21086323e+00 -3.55661899e-01 6.11774385e-01
3.95765841e-01 7.52621591e-02 -3.29591781e-01 2.70832747e-01]
|
[11.515189170837402, 8.987691879272461]
|
763046a5-f70b-440e-9a58-f2c6f8d005c0
|
document-level-event-extraction-via-parallel
| null | null |
https://aclanthology.org/2021.acl-long.492
|
https://aclanthology.org/2021.acl-long.492.pdf
|
Document-level Event Extraction via Parallel Prediction Networks
|
Document-level event extraction (DEE) is indispensable when events are described throughout a document. We argue that sentence-level extractors are ill-suited to the DEE task where event arguments always scatter across sentences and multiple events may co-exist in a document. It is a challenging task because it requires a holistic understanding of the document and an aggregated ability to assemble arguments across multiple sentences. In this paper, we propose an end-to-end model, which can extract structured events from a document in a parallel manner. Specifically, we first introduce a document-level encoder to obtain the document-aware representations. Then, a multi-granularity non-autoregressive decoder is used to generate events in parallel. Finally, to train the entire model, a matching loss function is proposed, which can bootstrap a global optimization. The empirical results on the widely used DEE dataset show that our approach significantly outperforms current state-of-the-art methods in the challenging DEE task. Code will be available at https://github.com/HangYang-NLP/DE-PPN.
|
['Taifeng Wang', 'Jun Zhao', 'Kang Liu', 'Yubo Chen', 'Dianbo Sui', 'Hang Yang']
|
2021-08-01
| null | null | null |
acl-2021-5
|
['document-level-event-extraction']
|
['natural-language-processing']
|
[ 2.90865272e-01 3.79173197e-02 -9.44802314e-02 -5.82748532e-01
-1.54760110e+00 -6.51086450e-01 7.32884407e-01 3.62671852e-01
-4.54662025e-01 8.03409994e-01 6.03019059e-01 -2.60179043e-01
5.52043580e-02 -8.34429681e-01 -9.74050999e-01 -3.46468508e-01
1.04939282e-01 3.66224051e-01 5.76954894e-02 2.93057337e-02
4.02210914e-02 1.66549787e-01 -1.14222372e+00 8.14694107e-01
7.74696410e-01 8.11700225e-01 3.08582753e-01 7.35830903e-01
-3.19458276e-01 1.13541830e+00 -8.18549037e-01 -5.96305490e-01
-2.39493743e-01 -6.34870410e-01 -8.39143813e-01 1.38439655e-01
-6.31246343e-02 -2.44396046e-01 -2.79219061e-01 7.29112685e-01
4.78057832e-01 5.76354899e-02 6.50878251e-01 -9.38235521e-01
-3.49411964e-01 1.26036513e+00 -4.36593354e-01 3.66452426e-01
4.00900543e-01 -1.85878158e-01 1.38182354e+00 -1.19083166e+00
6.81740403e-01 1.17707384e+00 2.38948897e-01 2.20152751e-01
-8.62674654e-01 -5.60757637e-01 5.20669103e-01 1.91239312e-01
-1.00818050e+00 -4.06604558e-01 9.82216656e-01 -1.62621424e-01
1.15621173e+00 1.39726564e-01 5.29847562e-01 1.53386855e+00
4.27940398e-01 1.36153674e+00 6.17980778e-01 -3.38209093e-01
1.52377456e-01 -2.79736340e-01 2.18564287e-01 4.04571205e-01
8.01988840e-02 -2.24271744e-01 -8.04604650e-01 4.21410576e-02
1.90041810e-01 1.46654934e-01 -1.33725405e-01 3.96071553e-01
-1.30377078e+00 8.22591424e-01 1.14770748e-01 3.48449916e-01
-8.36574376e-01 1.63453534e-01 6.49575353e-01 -3.50567978e-03
5.56752801e-01 1.41586468e-01 -5.60391366e-01 -3.51619810e-01
-9.59250689e-01 5.08700907e-01 8.98898900e-01 8.21370959e-01
1.74619958e-01 -2.16101259e-01 -5.22227168e-01 6.93681121e-01
3.00962895e-01 1.94167688e-01 3.12215656e-01 -4.13233846e-01
1.17199707e+00 4.97507870e-01 1.55535027e-01 -6.59036815e-01
-2.86802828e-01 -4.66323793e-01 -7.62767613e-01 -3.71437371e-01
1.80641785e-01 -5.07524431e-01 -5.58053613e-01 1.81701875e+00
4.90597695e-01 1.83961406e-01 3.27128947e-01 5.38063765e-01
8.86141241e-01 1.27569497e+00 1.72870278e-01 -3.38743210e-01
1.62312388e+00 -1.09954631e+00 -1.06932843e+00 -6.15131617e-01
3.74586880e-01 -6.60124362e-01 1.00894690e+00 3.13908458e-01
-1.35484755e+00 -2.99945951e-01 -1.03832519e+00 -4.29376960e-01
-3.55115384e-01 4.17420357e-01 4.91449445e-01 -2.21990049e-03
-1.88606530e-01 2.52357483e-01 -1.07062769e+00 8.49682614e-02
5.44726670e-01 3.45647894e-02 -2.91703660e-02 1.87681258e-01
-1.45980072e+00 7.42047846e-01 9.14062917e-01 2.04517350e-01
-7.98040152e-01 -5.52691698e-01 -1.03401411e+00 3.34763944e-01
6.23739004e-01 -6.04993403e-01 1.51071286e+00 -4.30585057e-01
-1.44472086e+00 6.35304213e-01 -5.20520449e-01 -5.93110204e-01
3.37154776e-01 -5.82522392e-01 -4.52122092e-01 2.41085663e-01
2.64578134e-01 1.98914573e-01 5.78572750e-01 -9.04421031e-01
-8.24554980e-01 -1.39421180e-01 1.03843652e-01 1.59815222e-01
-1.85043573e-01 4.94767457e-01 -5.28995574e-01 -9.66331661e-01
-1.18410811e-01 -6.22976482e-01 -2.01749206e-02 -6.64790571e-01
-8.27227354e-01 -5.62269092e-01 5.51236629e-01 -8.68391812e-01
1.58266306e+00 -2.15183473e+00 2.98912883e-01 -2.44972080e-01
1.14695542e-02 -8.29296336e-02 8.58153850e-02 8.75114739e-01
-1.38433367e-01 -5.29519245e-02 -3.71148080e-01 -6.51227713e-01
2.73894101e-01 -1.03110252e-02 -7.06746399e-01 1.37012735e-01
6.88338161e-01 1.03350794e+00 -9.56626952e-01 -6.34908557e-01
-4.29293737e-02 4.09805775e-01 -3.74405116e-01 3.89827281e-01
-5.73096812e-01 3.21391761e-01 -6.32386744e-01 4.80153650e-01
3.90654027e-01 -6.61930680e-01 1.30004779e-01 -8.48073140e-02
-1.15165725e-01 1.00974369e+00 -1.23220551e+00 1.74780643e+00
-6.15775526e-01 4.53724235e-01 -2.94367045e-01 -1.14785755e+00
5.51795900e-01 6.55023038e-01 2.93081999e-01 -3.66164148e-01
2.91828275e-01 2.84439266e-01 -2.31521562e-01 -3.21988016e-01
4.80725139e-01 -6.44659922e-02 -6.78803563e-01 5.77600121e-01
2.29333863e-01 -3.27793770e-02 7.63819516e-01 3.39493066e-01
1.07489729e+00 2.24950574e-02 6.72156692e-01 2.23899364e-01
5.44844627e-01 -1.42436713e-01 7.11642385e-01 7.75532663e-01
3.71662050e-01 6.38557673e-01 7.18855679e-01 -2.52950132e-01
-8.48698735e-01 -1.02416408e+00 -2.34624613e-02 8.18417311e-01
-9.10212845e-02 -8.36002648e-01 -6.23315811e-01 -1.03655732e+00
-4.29714501e-01 1.07704103e+00 -4.34875190e-01 1.23506472e-01
-7.83699155e-01 -9.42166865e-01 3.41373593e-01 7.08135188e-01
3.70890021e-01 -1.27508700e+00 -5.56634486e-01 7.51664102e-01
-6.97347939e-01 -1.24954784e+00 -5.50654352e-01 4.20071959e-01
-5.78864813e-01 -8.73797655e-01 -5.31234324e-01 -6.73278153e-01
4.54014748e-01 -2.61603594e-01 1.19532096e+00 -3.75207037e-01
-7.31579214e-02 -1.04081564e-01 -5.67983806e-01 -8.76226544e-01
-4.23080117e-01 2.92432457e-01 -5.45931816e-01 9.53444764e-02
5.09102285e-01 -4.56884712e-01 -4.75436747e-01 -1.27314523e-01
-1.04327309e+00 3.07576567e-01 6.71343684e-01 7.80399919e-01
8.07080626e-01 1.58891737e-01 8.16964447e-01 -1.06080520e+00
7.42480993e-01 -6.87135398e-01 -5.25299251e-01 3.89749408e-01
-6.35407120e-02 1.42496541e-01 8.90273869e-01 -5.09821296e-01
-1.33257484e+00 -1.13797141e-02 -4.26005363e-01 1.31627619e-01
-6.23350665e-02 7.45255530e-01 -4.26003069e-01 1.09793603e+00
3.04795980e-01 3.82114381e-01 -7.01033413e-01 -3.41512650e-01
3.71911854e-01 6.93572581e-01 5.34689963e-01 -6.90866411e-01
5.72031379e-01 3.80843878e-01 -3.41826588e-01 -4.94474500e-01
-1.53076923e+00 -2.02477366e-01 -3.54895800e-01 2.33627036e-02
9.23459113e-01 -1.12726724e+00 -4.08530414e-01 3.39436948e-01
-1.65096426e+00 -2.96054929e-01 -3.80453527e-01 5.72921574e-01
-4.73671615e-01 4.36785370e-02 -9.02968585e-01 -6.74947381e-01
-5.13846695e-01 -8.85994256e-01 1.38536692e+00 3.40349197e-01
-3.73437285e-01 -7.98279464e-01 9.67603251e-02 1.81762844e-01
-1.16650805e-01 3.04599941e-01 6.75060630e-01 -9.00349140e-01
-5.54557681e-01 -1.89972222e-01 -8.46720189e-02 1.96564361e-01
2.35638320e-01 -9.39751863e-02 -6.74254000e-01 5.51322429e-03
1.92738786e-01 -3.64713609e-01 9.25244927e-01 2.38006324e-01
1.29754484e+00 -5.29486001e-01 -2.85979629e-01 3.53742808e-01
1.15286815e+00 3.48790348e-01 4.06487346e-01 3.15181464e-01
4.97765481e-01 5.39054930e-01 6.88289940e-01 7.62246549e-01
7.02655971e-01 3.82734239e-01 9.88074392e-02 7.49729276e-02
-5.55241816e-02 -5.21793425e-01 6.07471883e-01 9.76411819e-01
3.88132632e-01 -8.19055080e-01 -6.86876833e-01 6.84498787e-01
-2.06322050e+00 -1.07458460e+00 -1.58439726e-01 1.66233456e+00
1.15797651e+00 3.29489529e-01 -1.27682880e-01 2.15020344e-01
5.71521938e-01 5.62422812e-01 -3.87353539e-01 -1.63030654e-01
-1.41378060e-01 2.15794131e-01 -1.47470804e-02 3.83791238e-01
-1.35623646e+00 9.07974899e-01 5.25308609e+00 8.93782437e-01
-8.41698885e-01 2.08288983e-01 4.91135180e-01 -3.42622131e-01
-3.22007149e-01 9.82171819e-02 -1.28331757e+00 9.11946237e-01
1.24673343e+00 -3.85556579e-01 4.39496711e-02 5.58165848e-01
3.10749859e-01 7.64650404e-02 -1.24323058e+00 7.50021338e-01
1.00058153e-01 -1.43279815e+00 -3.44912522e-02 -1.34185314e-01
5.64964473e-01 -1.41263634e-01 -3.14180434e-01 3.60332519e-01
2.97247529e-01 -5.72822690e-01 9.97942209e-01 3.71884495e-01
3.08899969e-01 -8.18643928e-01 5.91180980e-01 5.58435440e-01
-1.40025961e+00 -4.35044914e-02 -1.55849829e-02 1.51070923e-01
9.95207906e-01 1.00711930e+00 -6.43158436e-01 7.81672657e-01
3.87127101e-01 8.70294571e-01 -2.99757600e-01 6.00726008e-01
-9.26046431e-01 8.24408889e-01 -4.55331832e-01 -2.32255831e-01
2.85121381e-01 4.57155108e-02 6.11993432e-01 1.58394206e+00
4.22783136e-01 2.23215550e-01 9.77156758e-02 9.66039360e-01
-4.41339284e-01 6.22074045e-02 -3.43987077e-01 -2.63694286e-01
3.94858181e-01 1.14067411e+00 -7.14055359e-01 -6.01996481e-01
-6.38190627e-01 1.16018605e+00 5.04163623e-01 2.28604347e-01
-1.21768486e+00 -5.62546313e-01 1.39022410e-01 -3.21106553e-01
5.54537177e-01 -1.85511336e-01 -2.58791968e-02 -1.56827343e+00
4.10329163e-01 -8.97517085e-01 6.93874776e-01 -7.49762774e-01
-1.33492947e+00 6.20395243e-01 7.72456974e-02 -1.16432106e+00
-5.88919163e-01 -3.86942476e-01 -7.74525583e-01 6.40643179e-01
-1.43366790e+00 -1.06649327e+00 1.31282508e-01 3.98772031e-01
1.08161259e+00 1.05510786e-01 5.85721433e-01 4.86165226e-01
-7.53601074e-01 3.25425088e-01 -1.31935209e-01 5.11710584e-01
4.81555790e-01 -1.32564151e+00 5.23087502e-01 1.32219589e+00
5.07579029e-01 5.37860990e-01 5.13402760e-01 -8.17438960e-01
-1.31732929e+00 -1.20993125e+00 1.55215216e+00 -3.97115469e-01
7.14616060e-01 -6.31151557e-01 -8.53855789e-01 1.01672685e+00
3.62327069e-01 -1.02650836e-01 7.02054143e-01 6.88902438e-02
-1.06384447e-02 1.05799697e-02 -5.52174568e-01 7.35614479e-01
7.87499607e-01 -6.41185582e-01 -1.09106779e+00 5.65128624e-01
8.55428874e-01 -6.75612628e-01 -6.97294056e-01 2.66362697e-01
1.86670825e-01 -5.77229261e-01 9.14433599e-01 -6.21354818e-01
9.63890731e-01 -2.59721398e-01 1.54206809e-03 -1.07390654e+00
6.50755912e-02 -7.36012578e-01 -7.09878385e-01 1.61817038e+00
8.29599321e-01 -4.38541323e-01 4.38706666e-01 2.72635341e-01
-1.01807244e-01 -8.60524476e-01 -7.00856149e-01 -6.67516828e-01
-1.59004584e-01 -7.36615419e-01 5.57178497e-01 4.89587754e-01
-2.11020429e-02 7.54465997e-01 -3.69491816e-01 4.63740826e-01
5.43027282e-01 4.23939198e-01 4.54840571e-01 -8.48674595e-01
-5.18256605e-01 -2.84302175e-01 3.93237501e-01 -1.20355201e+00
4.40685451e-01 -9.40948546e-01 2.09528327e-01 -1.82379544e+00
3.25235993e-01 1.60711437e-01 -3.04523379e-01 3.48200619e-01
-6.05200529e-01 -4.33405995e-01 2.06309661e-01 -3.48824523e-02
-8.18167329e-01 8.23549151e-01 1.00388765e+00 -6.99009821e-02
-1.89058468e-01 1.18749455e-01 -7.44577885e-01 7.68188596e-01
8.71493578e-01 -9.68210995e-01 -3.06920767e-01 -3.98334622e-01
4.09805030e-01 3.62520069e-01 2.60270119e-01 -6.88065588e-01
3.79604965e-01 -1.85291842e-02 3.69963139e-01 -1.14907014e+00
3.07527989e-01 -5.01352966e-01 -3.49965021e-02 1.87674657e-01
-5.83580315e-01 2.07500324e-01 6.42868876e-02 4.86566514e-01
-6.63982213e-01 -4.37080622e-01 2.65933663e-01 -1.44390941e-01
-2.80193508e-01 3.04353863e-01 -3.29541296e-01 4.29049969e-01
8.97741437e-01 3.91550571e-01 -2.48205021e-01 -1.63394883e-01
-5.40566683e-01 2.95421988e-01 -1.65668234e-01 3.96489084e-01
6.02052927e-01 -1.26977372e+00 -1.04108143e+00 -1.68694764e-01
-3.40288156e-03 4.20103759e-01 3.08232516e-01 5.85681438e-01
-9.95617136e-02 3.61999065e-01 4.61222261e-01 -2.99418002e-01
-1.09131145e+00 4.62976485e-01 -1.33695722e-01 -9.28853631e-01
-7.95919418e-01 8.13617289e-01 1.81210041e-01 -1.64544791e-01
1.60068661e-01 -5.64230025e-01 -1.15171857e-01 3.05921346e-01
7.50072658e-01 -4.48685624e-02 1.98818296e-02 -2.14724109e-01
-3.41813624e-01 2.17192963e-01 -2.89149076e-01 -4.90669817e-01
1.48066449e+00 2.68273614e-02 -6.83541177e-03 5.81614733e-01
1.19428146e+00 1.51774600e-01 -1.41581118e+00 -2.55260408e-01
2.07186878e-01 9.13408305e-03 -1.64449498e-01 -6.97350144e-01
-5.78250468e-01 8.78983855e-01 -2.62825370e-01 2.57633477e-01
1.22254860e+00 3.57807130e-01 1.10114980e+00 3.84470224e-01
-4.22396325e-02 -1.01889753e+00 7.31400996e-02 7.00160205e-01
1.10546029e+00 -1.18517220e+00 -4.96408716e-02 -4.27968889e-01
-7.20419109e-01 1.01708746e+00 2.21608952e-01 -1.20473273e-01
4.83592540e-01 5.52231491e-01 -4.15668815e-01 -2.43266702e-01
-1.10000896e+00 -1.07476905e-01 3.14852118e-01 -2.24256188e-01
5.44384181e-01 -8.64006877e-02 -5.65604508e-01 1.20055556e+00
-3.27710956e-01 -1.10612819e-02 3.57656509e-01 1.10412824e+00
-1.00172617e-01 -1.33457494e+00 -1.04559436e-01 2.81206787e-01
-1.02308524e+00 -3.10056895e-01 -1.89157575e-01 6.31180406e-01
-6.77287057e-02 9.88347590e-01 -7.77136758e-02 2.28516102e-01
4.61872071e-01 3.91676843e-01 3.23367596e-01 -7.86609828e-01
-6.90434158e-01 2.67764926e-01 4.43454593e-01 -5.14114678e-01
-3.51441711e-01 -8.65638852e-01 -1.51614177e+00 2.97101915e-01
4.98406636e-03 2.76986867e-01 6.22099221e-01 1.04357409e+00
4.80558395e-01 1.08095145e+00 5.61634183e-01 -6.30294323e-01
-4.66211885e-01 -8.77012372e-01 -2.60702968e-01 3.38145018e-01
4.07526493e-01 -3.75359714e-01 -3.97559404e-01 3.48151177e-01]
|
[9.070380210876465, 9.15941047668457]
|
f64c398a-93ac-4afe-a1a1-cbab746d7778
|
carvenet-carving-point-block-for-complex-3d
|
2107.13452
| null |
https://arxiv.org/abs/2107.13452v1
|
https://arxiv.org/pdf/2107.13452v1.pdf
|
CarveNet: Carving Point-Block for Complex 3D Shape Completion
|
3D point cloud completion is very challenging because it heavily relies on the accurate understanding of the complex 3D shapes (e.g., high-curvature, concave/convex, and hollowed-out 3D shapes) and the unknown & diverse patterns of the partially available point clouds. In this paper, we propose a novel solution,i.e., Point-block Carving (PC), for completing the complex 3D point cloud completion. Given the partial point cloud as the guidance, we carve a3D block that contains the uniformly distributed 3D points, yielding the entire point cloud. To achieve PC, we propose a new network architecture, i.e., CarveNet. This network conducts the exclusive convolution on each point of the block, where the convolutional kernels are trained on the 3D shape data. CarveNet determines which point should be carved, for effectively recovering the details of the complete shapes. Furthermore, we propose a sensor-aware method for data augmentation,i.e., SensorAug, for training CarveNet on richer patterns of partial point clouds, thus enhancing the completion power of the network. The extensive evaluations on the ShapeNet and KITTI datasets demonstrate the generality of our approach on the partial point clouds with diverse patterns. On these datasets, CarveNet successfully outperforms the state-of-the-art methods.
|
['Yang Liu', 'Wei Feng', 'Lei Ma', 'Di Lin', 'Felix Juefei-Xu', 'Zhijie Wang', 'Qing Guo']
|
2021-07-28
| null | null | null | null |
['point-cloud-completion']
|
['computer-vision']
|
[-2.76602566e-01 -1.74315497e-01 2.97304094e-01 -2.50741065e-01
-3.22804779e-01 -6.80023432e-01 3.34259182e-01 -2.10399270e-01
3.37998271e-02 -5.43506667e-02 -3.59420590e-02 -3.26257795e-01
2.55745519e-02 -8.66424143e-01 -1.07064033e+00 -5.23025334e-01
7.22103715e-02 6.44297421e-01 1.24221094e-01 -2.12681338e-01
1.70496002e-01 1.11789191e+00 -1.32599747e+00 -5.29566780e-02
8.73189807e-01 1.18174803e+00 5.35892129e-01 3.63353193e-01
-3.75567138e-01 -1.03231883e-02 -1.09932013e-01 -1.54365376e-01
4.95404065e-01 6.42603099e-01 -1.26333237e-01 3.95684630e-01
3.22202682e-01 -6.56823516e-01 -3.26235652e-01 8.05624723e-01
1.64726406e-01 -1.22732356e-01 5.37674904e-01 -1.16992009e+00
-8.03803265e-01 -5.96022699e-03 -7.71284223e-01 -4.24517006e-01
1.32823184e-01 1.96274862e-01 5.37120521e-01 -1.66602039e+00
2.72238761e-01 1.37013757e+00 8.24242115e-01 3.06718558e-01
-8.52637351e-01 -9.49671626e-01 1.99543387e-01 -2.58658916e-01
-1.64443588e+00 -3.19913596e-01 1.21804643e+00 -5.01870394e-01
6.74334109e-01 7.41896704e-02 7.19886243e-01 4.44838703e-01
-2.27359325e-01 7.93436348e-01 5.88519990e-01 1.95617061e-02
1.60479367e-01 -3.24699163e-01 -1.81649700e-01 4.99504656e-01
2.80028850e-01 2.01163843e-01 7.82761499e-02 -3.00182849e-01
1.25158536e+00 7.76175857e-01 -3.02626014e-01 -7.12953627e-01
-1.27065170e+00 3.64143401e-01 7.47084737e-01 -9.03684571e-02
-7.03403115e-01 1.94428369e-01 5.63226733e-03 -1.89590156e-01
4.80564862e-01 -7.82892331e-02 -7.11480439e-01 5.33967055e-02
-5.06945848e-01 5.56339145e-01 4.46912080e-01 1.61429954e+00
1.04070461e+00 1.16163671e-01 1.90460771e-01 7.71079957e-01
7.58634329e-01 1.09465790e+00 -2.82659054e-01 -8.20542216e-01
8.42757046e-01 1.16199994e+00 3.43100727e-01 -1.06814480e+00
-3.95955682e-01 -3.52173299e-01 -1.15142763e+00 3.44454676e-01
-8.17665160e-02 -1.37766570e-01 -1.16528010e+00 1.29008079e+00
6.85696363e-01 6.10992193e-01 -8.25983435e-02 1.15764725e+00
1.02358234e+00 7.61360347e-01 -3.54061723e-01 3.28945726e-01
1.06574690e+00 -5.91318667e-01 -3.08425426e-01 -7.76847005e-02
1.70045033e-01 -6.14866972e-01 9.51762676e-01 1.78307116e-01
-9.92175758e-01 -6.49640918e-01 -1.03233969e+00 -1.13475405e-01
-1.15648426e-01 4.39065248e-01 6.79272175e-01 9.02105309e-03
-7.90251017e-01 4.05972809e-01 -1.00949121e+00 5.30785173e-02
8.50092411e-01 3.45926136e-01 -4.53318447e-01 -5.61074555e-01
-2.92898983e-01 2.96512097e-01 2.03940257e-01 3.62178117e-01
-9.99310374e-01 -1.13095689e+00 -8.28095257e-01 2.06947520e-01
2.90669620e-01 -5.95593750e-01 9.97481287e-01 -2.09079072e-01
-1.09706199e+00 6.43735766e-01 -5.85288741e-02 5.00221662e-02
3.11554879e-01 -2.38364935e-01 -4.38226908e-02 -9.28371996e-02
-8.55386704e-02 7.28018761e-01 9.18178022e-01 -1.74622929e+00
-4.75095600e-01 -6.99892521e-01 -6.81215897e-02 2.59232402e-01
1.17854431e-01 -4.70276207e-01 -9.52411175e-01 -4.87146765e-01
7.26648748e-01 -9.45251048e-01 -2.86309958e-01 3.84341598e-01
-4.44556504e-01 -3.10766548e-01 1.18870270e+00 -5.42518675e-01
6.40508890e-01 -2.51347780e+00 -4.63809706e-02 3.99773091e-01
4.27476674e-01 2.09481433e-01 -3.24189156e-01 3.41031730e-01
-6.20734654e-02 1.17919512e-01 -3.65000367e-01 -6.65731132e-01
1.46239609e-01 4.76886421e-01 -5.84436178e-01 3.91078085e-01
3.93223941e-01 1.11695397e+00 -7.13291347e-01 -2.39499751e-02
5.31805634e-01 7.36086726e-01 -4.61272329e-01 3.16917270e-01
-3.05661350e-01 4.09374893e-01 -7.67240107e-01 1.02471066e+00
1.57479262e+00 -2.34229207e-01 -6.13749921e-01 -3.34080666e-01
-2.74975091e-01 -2.10571334e-01 -1.24047339e+00 2.11071396e+00
-3.54624927e-01 3.62531915e-02 4.05156642e-01 -4.82731819e-01
1.36561120e+00 1.93744585e-01 6.47675216e-01 -2.97146440e-01
6.78423792e-02 3.90437722e-01 -2.24679425e-01 -2.31268778e-01
3.82241577e-01 6.09983206e-02 1.55179694e-01 1.62968516e-01
-3.18989724e-01 -5.64229429e-01 -5.19655406e-01 1.79794282e-02
8.87761533e-01 3.26152056e-01 -2.77508289e-01 -3.34430509e-03
4.26254123e-01 1.88712589e-02 6.48637772e-01 2.56286353e-01
1.06933184e-01 9.79943335e-01 4.89547551e-02 -5.99102795e-01
-1.38600469e+00 -1.09337091e+00 -1.49760485e-01 1.52304098e-01
4.30254966e-01 -1.58135518e-01 -4.25620943e-01 -3.95951718e-01
4.70138937e-01 2.65471071e-01 -3.27659816e-01 6.39416799e-02
-6.73336446e-01 -2.41204754e-01 3.95452082e-02 6.24027848e-01
6.45976424e-01 -9.90700603e-01 -3.81685019e-01 1.18663058e-01
1.58839732e-01 -1.31106889e+00 -4.69696760e-01 -2.88112700e-01
-1.19628787e+00 -1.11117756e+00 -5.09946942e-01 -6.90423608e-01
1.00577855e+00 8.15314889e-01 9.19538260e-01 3.17645252e-01
2.05289274e-01 3.55126143e-01 -5.18639684e-01 -7.25339413e-01
1.48197278e-01 -1.79271311e-01 -3.36336456e-02 1.02391161e-01
3.86867911e-01 -1.00470066e+00 -7.76622593e-01 3.78627062e-01
-9.53621507e-01 1.67553023e-01 9.05038297e-01 5.16373217e-01
1.10965788e+00 2.56708153e-02 2.06666827e-01 -5.26425421e-01
3.72759044e-01 -5.87191463e-01 -7.17571437e-01 -1.65740609e-01
-9.98232812e-02 -2.69667178e-01 6.28980339e-01 -4.46702152e-01
-6.15825593e-01 5.11859000e-01 -3.28843892e-01 -1.50121045e+00
-3.13162655e-01 3.88263524e-01 -4.58081871e-01 -2.95462400e-01
3.24191004e-01 4.22251016e-01 4.44336422e-02 -8.99064958e-01
3.45228523e-01 5.10432363e-01 6.03858054e-01 -5.64107537e-01
1.39480925e+00 8.87355208e-01 7.14156851e-02 -6.65957451e-01
-3.75696093e-01 -6.04031920e-01 -6.76419199e-01 -3.66983041e-02
7.22638786e-01 -1.17258966e+00 -8.01037133e-01 6.57637000e-01
-1.54877508e+00 -1.55855492e-01 -2.46525228e-01 2.94694394e-01
-4.36840028e-01 2.49814942e-01 -2.00471923e-01 -7.56843388e-01
-5.81476092e-01 -1.06778753e+00 1.51496053e+00 1.11773439e-01
5.36519229e-01 -5.56859553e-01 -1.27386779e-01 7.04042763e-02
7.91530162e-02 4.16180491e-01 6.77872598e-01 -2.70359039e-01
-1.19217825e+00 -4.29121345e-01 -5.15265226e-01 2.94699252e-01
1.88578591e-01 -4.54221144e-02 -8.90530109e-01 -3.88922751e-01
-2.98943371e-02 1.72299296e-02 4.66529012e-01 2.90681809e-01
1.52704656e+00 -1.25465199e-01 -4.31028813e-01 1.05801976e+00
1.47925389e+00 1.33396253e-01 5.54006457e-01 -2.57102340e-01
1.08183146e+00 1.99166253e-01 8.02061260e-01 7.13761508e-01
6.33798718e-01 4.15400505e-01 1.12573183e+00 -2.67873853e-01
6.11683987e-02 -6.74296081e-01 -2.35732645e-01 9.93433297e-01
-1.50482938e-01 1.62231356e-01 -1.07540405e+00 6.10881090e-01
-1.86486602e+00 -3.22682768e-01 -3.35294187e-01 2.02604485e+00
1.64660498e-01 -6.78233206e-02 -4.23201472e-01 -2.47966051e-02
5.59547842e-01 5.82317188e-02 -1.02366519e+00 2.06259355e-01
9.20105651e-02 1.25974849e-01 3.79199356e-01 2.73657590e-01
-8.95307302e-01 8.94688845e-01 4.85582352e+00 6.81652904e-01
-1.04952121e+00 -6.15617260e-02 1.62130117e-01 2.37166107e-01
-5.27513325e-01 7.02093244e-02 -7.29007721e-01 4.98877794e-01
6.96695922e-03 2.91759610e-01 5.64558625e-01 1.00617802e+00
2.63001531e-01 4.15908098e-01 -9.14883196e-01 1.41059244e+00
2.21851598e-02 -1.41087866e+00 1.43991649e-01 1.97194532e-01
6.92714810e-01 4.67810571e-01 -1.47282153e-01 2.61636645e-01
3.75879049e-01 -7.59851456e-01 7.04964042e-01 6.44869089e-01
1.13776958e+00 -7.48556197e-01 5.87063909e-01 8.21696937e-01
-1.37894237e+00 4.21719924e-02 -7.76561677e-01 3.90800796e-02
1.03856571e-01 7.45989919e-01 -7.34474897e-01 6.08611405e-01
8.28214705e-01 1.04289079e+00 -2.33835787e-01 1.15613210e+00
-1.34665921e-01 2.40945101e-01 -7.01423645e-01 1.85870469e-01
2.14178026e-01 -4.83816624e-01 6.87546074e-01 7.06924081e-01
5.85651934e-01 6.25981748e-01 3.57027978e-01 1.31852412e+00
-1.10622779e-01 -1.36295065e-01 -7.55032957e-01 2.24304497e-01
8.79177094e-01 1.52893984e+00 -2.04333827e-01 -1.23214141e-01
-4.97306585e-01 4.89633799e-01 5.24515390e-01 4.27787155e-01
-4.30537969e-01 -2.76077271e-01 8.52926314e-01 1.74374312e-01
6.68758869e-01 -6.95129812e-01 -5.81968129e-01 -1.09874678e+00
4.33771163e-01 -4.42727894e-01 -1.55840069e-01 -1.26055825e+00
-1.29351282e+00 3.87341380e-01 -8.93021300e-02 -1.63132751e+00
3.89022559e-01 -6.05736256e-01 -8.81609261e-01 1.16961408e+00
-1.66879439e+00 -1.52824783e+00 -9.08848107e-01 8.26362610e-01
3.35821629e-01 1.26735176e-04 3.80683392e-01 3.05042893e-01
-1.66608423e-01 8.16963520e-03 -1.65962160e-01 2.11385369e-01
2.79953554e-02 -8.47544134e-01 8.05145860e-01 6.09771073e-01
-1.55469075e-01 5.53055286e-01 5.44759333e-02 -8.19359839e-01
-2.06560135e+00 -1.43103886e+00 2.07298592e-01 -4.77172643e-01
2.51361132e-01 -6.74286842e-01 -1.07973456e+00 7.07515121e-01
-4.31947708e-01 3.43899757e-01 6.57882690e-02 -1.84250653e-01
-2.77415484e-01 -2.28231117e-01 -1.12853742e+00 5.09813666e-01
1.32075059e+00 -1.86356843e-01 -4.87240493e-01 2.64954329e-01
1.17927063e+00 -7.82469153e-01 -8.93941224e-01 8.11691701e-01
2.87181646e-01 -5.53002238e-01 1.23860335e+00 -2.67430842e-01
5.18273592e-01 -7.61742294e-01 -2.42402211e-01 -1.38032043e+00
-3.78534436e-01 -2.48014838e-01 -3.46553385e-01 9.80248094e-01
1.17598034e-01 -4.72433329e-01 1.07462144e+00 5.66617906e-01
-8.10197830e-01 -1.06876171e+00 -9.12297368e-01 -3.54065895e-01
8.67139772e-02 -7.23000288e-01 1.48565757e+00 8.79367709e-01
-6.22555852e-01 -1.17261056e-02 -2.24701986e-02 6.61246419e-01
6.05485678e-01 3.78918469e-01 1.32473826e+00 -1.30882418e+00
1.61128566e-01 -1.37784287e-01 -4.29452151e-01 -1.67482638e+00
-5.61748706e-02 -7.83385396e-01 1.85928531e-02 -1.53415930e+00
-1.30545929e-01 -1.18575120e+00 1.13939099e-01 5.67988992e-01
-1.16331138e-01 -7.67111555e-02 2.81519175e-01 5.10932326e-01
-1.31865770e-01 8.90241563e-01 1.82856119e+00 -1.90970898e-01
-3.42505068e-01 5.92281930e-02 -6.07834160e-01 6.46748304e-01
5.92470884e-01 -9.53786597e-02 -3.97434026e-01 -9.49251235e-01
1.91161111e-01 1.15034029e-01 5.69467187e-01 -9.49144900e-01
3.64736319e-01 -2.36581385e-01 4.84189332e-01 -1.68673229e+00
7.24351346e-01 -1.53003860e+00 2.34934255e-01 5.95503710e-02
4.38555032e-01 1.67034030e-01 3.71766597e-01 5.74258089e-01
3.66135081e-03 9.48226601e-02 3.60264510e-01 -2.88885117e-01
-4.45147216e-01 1.30399239e+00 7.26257205e-01 -3.17125380e-01
9.54698682e-01 -3.25837433e-01 -1.48375928e-01 -5.13001159e-02
-4.12532806e-01 6.63864911e-01 7.15178251e-01 5.52156866e-01
1.34683239e+00 -1.78567171e+00 -9.16242599e-01 7.33395040e-01
3.19654912e-01 1.31919956e+00 4.60674465e-01 5.49813271e-01
-6.50734305e-01 1.29511386e-01 8.06100070e-02 -1.07755053e+00
-8.09238434e-01 7.49741852e-01 2.76517481e-01 2.23488718e-01
-9.95867014e-01 4.29523110e-01 4.33067560e-01 -1.11677158e+00
8.36156905e-02 -7.35543251e-01 -1.54597461e-02 -5.97698033e-01
3.32694650e-01 1.18293680e-01 1.47580698e-01 -5.61518967e-01
-1.97915480e-01 1.05215931e+00 9.27649364e-02 2.71065772e-01
1.48913550e+00 1.44348249e-01 -2.33945847e-01 1.60773039e-01
1.01998913e+00 -1.17073841e-02 -1.72684097e+00 -3.77300054e-01
-5.17407537e-01 -7.77706742e-01 2.22426374e-03 -5.04465461e-01
-1.36978090e+00 1.04715264e+00 2.96057284e-01 -1.58162981e-01
9.20061529e-01 1.68119464e-02 9.17799532e-01 3.74271482e-01
5.61042011e-01 -5.04854560e-01 -2.58509189e-01 6.03359818e-01
1.29174876e+00 -1.14237654e+00 5.46053387e-02 -7.17652798e-01
-2.99473614e-01 9.90318954e-01 8.03786397e-01 -5.30586243e-01
9.14655447e-01 1.63133666e-01 -1.40339345e-01 -5.35297275e-01
-5.71429908e-01 -2.77098292e-03 2.93240637e-01 7.27367878e-01
-4.74611819e-01 3.53886575e-01 3.36241603e-01 5.70083678e-01
-1.91628918e-01 7.66970515e-02 2.19600871e-01 7.45750248e-01
-3.68043572e-01 -6.23947978e-01 -7.34647512e-01 4.76545125e-01
2.97647238e-01 8.05387869e-02 -2.83873975e-01 7.09860325e-01
2.95354396e-01 6.48842275e-01 3.38964492e-01 -6.42858744e-01
8.39082122e-01 -4.40622032e-01 5.88354468e-02 -6.67277038e-01
-1.92837611e-01 9.59702581e-02 -5.19824564e-01 -5.18667459e-01
-9.29411948e-02 -6.40956223e-01 -1.50491107e+00 -4.16690767e-01
-1.40894309e-01 -1.16072118e-01 9.90148604e-01 7.18544185e-01
8.04618955e-01 1.93420932e-01 9.30990934e-01 -1.33368504e+00
-5.14051497e-01 -8.86531830e-01 -4.66759235e-01 5.76578617e-01
4.34763253e-01 -7.24874377e-01 -1.68672353e-01 -3.59415233e-01]
|
[8.305142402648926, -3.532452344894409]
|
e573f314-bcdb-446f-92b9-66fa7fd17081
|
context-aware-language-modeling-for-goal-1
|
2204.10198
| null |
https://arxiv.org/abs/2204.10198v2
|
https://arxiv.org/pdf/2204.10198v2.pdf
|
Context-Aware Language Modeling for Goal-Oriented Dialogue Systems
|
Goal-oriented dialogue systems face a trade-off between fluent language generation and task-specific control. While supervised learning with large language models is capable of producing realistic text, how to steer such responses towards completing a specific task without sacrificing language quality remains an open question. In this work, we formulate goal-oriented dialogue as a partially observed Markov decision process, interpreting the language model as a representation of both the dynamics and the policy. This view allows us to extend techniques from learning-based control, such as task relabeling, to derive a simple and effective method to finetune language models in a goal-aware way, leading to significantly improved task performance. We additionally introduce a number of training strategies that serve to better focus the model on the task at hand. We evaluate our method, Context-Aware Language Models (CALM), on a practical flight-booking task using AirDialogue. Empirically, CALM outperforms the state-of-the-art method by 7% in terms of task success, matching human-level task performance.
|
['Mengjiao Yang', 'Sergey Levine', 'Yi Su', 'Justin Fu', 'Charlie Snell']
|
2022-04-18
| null |
https://aclanthology.org/2022.findings-naacl.181
|
https://aclanthology.org/2022.findings-naacl.181.pdf
|
findings-naacl-2022-7
|
['goal-oriented-dialogue-systems']
|
['natural-language-processing']
|
[ 2.58850515e-01 4.46850181e-01 -3.33142281e-01 -5.72280169e-01
-8.81684422e-01 -6.73538446e-01 8.60734701e-01 1.10901967e-01
-5.47922313e-01 9.22924101e-01 4.62465525e-01 -4.72622246e-01
-2.75193732e-02 -4.90080446e-01 -2.07723692e-01 -3.54522288e-01
1.06445670e-01 8.75193596e-01 1.01262145e-01 -7.78466046e-01
3.62093866e-01 2.93436289e-01 -1.05080354e+00 4.15308744e-01
9.40261960e-01 5.49814343e-01 3.34003508e-01 1.11019230e+00
-7.10156187e-02 1.28066683e+00 -5.85333884e-01 -2.21652514e-03
-2.15306506e-02 -7.09295332e-01 -1.20338035e+00 3.68444324e-01
6.91439360e-02 -1.88210011e-01 3.92203555e-02 6.29498601e-01
4.27118152e-01 5.70404291e-01 7.13937461e-01 -9.65317190e-01
-4.38848212e-02 4.83388573e-01 -2.33480446e-02 -2.53649801e-02
3.95379484e-01 5.19429505e-01 8.38777721e-01 -2.46370479e-01
6.15352511e-01 1.57789528e+00 1.12420142e-01 1.02266955e+00
-1.52485752e+00 -1.82611480e-01 5.47931194e-01 -3.00834388e-01
-9.25184965e-01 -5.98998964e-01 5.11319458e-01 -7.47593641e-01
1.07611322e+00 2.84661409e-02 3.70662004e-01 9.95289087e-01
5.42425334e-01 6.64439499e-01 1.37903976e+00 -5.47889769e-01
4.09050763e-01 2.60976255e-01 -4.87597287e-03 7.96049058e-01
-3.03635329e-01 2.49688521e-01 -7.59907424e-01 -2.23620191e-01
5.56503713e-01 -5.13559043e-01 -3.11505705e-01 -5.87386608e-01
-1.28477573e+00 1.13801122e+00 5.13135754e-02 1.79511949e-01
-6.17802501e-01 1.12660997e-01 4.60648209e-01 5.24283409e-01
5.20050168e-01 1.02484238e+00 -6.29103243e-01 -3.80411208e-01
-6.80117309e-01 6.82859778e-01 1.30491126e+00 8.19570959e-01
5.58989942e-01 2.18260601e-01 -7.09154665e-01 9.02347386e-01
2.71449685e-01 2.29856521e-01 3.84900331e-01 -1.19796121e+00
5.15341043e-01 4.38927174e-01 4.93049830e-01 -4.79408383e-01
-6.04690135e-01 -1.87954664e-01 -3.79912764e-01 3.24126750e-01
5.95116973e-01 -7.41482258e-01 -7.26435959e-01 1.93561888e+00
3.57509851e-01 -5.15082657e-01 3.41272384e-01 9.82664645e-01
4.41312455e-02 7.39456713e-01 4.25068319e-01 -4.01012808e-01
1.26734710e+00 -1.13570106e+00 -8.66658926e-01 -5.77945590e-01
8.92410696e-01 -5.86196303e-01 1.15859330e+00 5.63591123e-01
-1.10150850e+00 -4.46427077e-01 -7.45742381e-01 3.05686265e-01
-2.77599990e-02 -2.36261547e-01 6.02050781e-01 3.41645241e-01
-1.06436825e+00 5.15397489e-01 -7.24922657e-01 -4.73950267e-01
-2.94891745e-01 2.82439798e-01 -8.00113678e-02 1.90948173e-01
-1.38064659e+00 1.25871372e+00 5.16821623e-01 -1.03606425e-01
-1.27158463e+00 -3.48423243e-01 -7.84744501e-01 -1.48546144e-01
8.87656510e-01 -8.78172457e-01 2.08353162e+00 -8.43810737e-01
-2.09292650e+00 6.88617527e-01 -5.40756211e-02 -6.69707358e-01
5.43162823e-01 -2.48974532e-01 -5.22937477e-02 -3.76590947e-03
1.20075233e-02 8.47646534e-01 8.34094524e-01 -1.21367705e+00
-8.90164435e-01 -1.18780985e-01 4.66043979e-01 6.83129668e-01
-1.27176508e-01 -2.74441708e-02 -8.31020176e-02 -2.86397099e-01
-3.77460331e-01 -1.16913700e+00 -6.82808757e-01 -4.33648974e-01
-2.55741090e-01 -4.45925683e-01 3.36169660e-01 -4.81646121e-01
1.32348239e+00 -1.52391112e+00 4.87968087e-01 -2.85250694e-01
1.67200208e-01 2.89449602e-01 -1.39124632e-01 7.02739358e-01
2.41523147e-01 -7.75930192e-03 -2.96079010e-01 -4.27788943e-01
7.97191262e-02 2.66021490e-01 -3.18658561e-01 1.23748317e-01
2.76926070e-01 6.88247144e-01 -1.20106375e+00 -5.05495131e-01
2.84819186e-01 -7.94968009e-02 -7.73755968e-01 7.03211904e-01
-9.85545456e-01 8.42579126e-01 -7.65503347e-01 1.21430628e-01
-1.06394000e-01 8.16335380e-02 5.68515301e-01 4.16588396e-01
-2.02415749e-01 5.49278140e-01 -8.08349609e-01 1.82859635e+00
-9.96741116e-01 2.42599756e-01 4.97978359e-01 -7.14782596e-01
8.94145608e-01 3.35056335e-01 1.23107001e-01 -5.28126717e-01
-2.04688907e-02 -1.30640835e-01 2.77400196e-01 -6.60314023e-01
6.95584774e-01 -4.90637720e-01 -4.10136938e-01 6.38016284e-01
1.46060199e-01 -6.00665689e-01 3.17035019e-01 1.83597356e-01
8.02470505e-01 2.31744185e-01 4.30085480e-01 -5.11627674e-01
5.65947413e-01 3.84016484e-01 3.00734997e-01 9.96698499e-01
-4.08342332e-01 1.04602270e-01 6.62656307e-01 -4.47854072e-01
-6.71509087e-01 -4.83366251e-01 3.51487249e-01 1.64739048e+00
-2.41559446e-01 -3.99862796e-01 -1.04888177e+00 -7.96457767e-01
-2.75159717e-01 1.33127558e+00 -2.40306780e-01 -2.40285084e-01
-7.00280964e-01 -5.55916131e-01 2.32609823e-01 -1.69891934e-03
3.86736423e-01 -1.30167496e+00 -7.45293200e-01 5.22049844e-01
-4.69192505e-01 -1.15927994e+00 -7.19360888e-01 3.36949140e-01
-8.63041282e-01 -8.54336917e-01 -4.55751479e-01 -4.44436669e-01
2.92246372e-01 -5.98148555e-02 1.41953468e+00 -1.10811479e-01
2.89616048e-01 6.12818658e-01 -2.68742859e-01 -4.12731707e-01
-1.09823525e+00 4.08275217e-01 1.37582406e-01 1.16070714e-02
1.38117269e-01 -1.42399415e-01 -3.26920778e-01 1.82474196e-01
-7.50308871e-01 3.71354401e-01 4.56372470e-01 1.04241812e+00
2.59204924e-01 -2.21668765e-01 5.82981944e-01 -9.94637489e-01
1.40027905e+00 -2.20045403e-01 -6.38359427e-01 2.05509245e-01
-8.65723193e-01 5.16278863e-01 7.23013997e-01 -4.59669411e-01
-1.27829051e+00 2.31979057e-01 -9.37421620e-02 5.07759377e-02
-3.49085689e-01 5.05582213e-01 6.74756542e-02 1.68467864e-01
8.60942721e-01 3.74943793e-01 1.73714027e-01 -1.66972414e-01
5.60147464e-01 7.12953925e-01 1.76277354e-01 -9.55699682e-01
3.98801178e-01 -9.18517783e-02 -1.41958162e-01 -7.98772395e-01
-1.11185002e+00 -3.86846423e-01 -6.55675471e-01 -3.44689250e-01
9.17011499e-01 -8.58625472e-01 -7.41539061e-01 2.26876631e-01
-1.20613027e+00 -1.11173081e+00 -1.40401185e-01 2.41765678e-01
-1.16506481e+00 -5.42004742e-02 -4.97633308e-01 -1.25725901e+00
-2.33222872e-01 -1.08429253e+00 1.09895718e+00 -2.73228996e-02
-6.15665674e-01 -1.24989796e+00 2.58581698e-01 5.39855897e-01
5.02797425e-01 1.42858386e-01 9.46667552e-01 -8.39200377e-01
-3.87458563e-01 -1.20795956e-02 3.64875793e-01 2.72884965e-01
2.26013083e-02 -4.44189131e-01 -7.30345488e-01 -4.24316138e-01
3.63576829e-01 -7.71250725e-01 5.28856516e-01 1.49625033e-01
5.72363079e-01 -5.23331642e-01 -6.60965145e-02 1.34575767e-02
9.15743351e-01 2.39268661e-01 2.28648365e-01 2.90639341e-01
3.95811558e-01 1.10812008e+00 1.11852014e+00 5.38767099e-01
5.79566419e-01 9.47415650e-01 3.45727503e-01 2.03812778e-01
8.88023302e-02 -4.94362295e-01 5.43080866e-01 5.88439465e-01
2.33207524e-01 -4.01816636e-01 -1.06415021e+00 4.45869476e-01
-2.05647826e+00 -7.86952794e-01 1.35343522e-01 2.26483917e+00
1.05390227e+00 3.33941638e-01 2.91962504e-01 -3.75588000e-01
4.93424147e-01 4.05165941e-01 -4.32066739e-01 -7.75786817e-01
5.54103315e-01 -2.01020658e-01 2.35954523e-01 1.05840170e+00
-9.90493774e-01 1.43425727e+00 6.38838339e+00 6.09720767e-01
-1.08628631e+00 1.62655994e-01 5.32943070e-01 -6.04473799e-03
5.31415790e-02 3.31641287e-02 -9.81819630e-01 1.23183489e-01
1.35545599e+00 -2.31220573e-01 6.73105836e-01 8.93649578e-01
7.31348574e-01 -2.74562806e-01 -1.26874435e+00 4.59332317e-01
-1.09721228e-01 -9.30673420e-01 6.72263876e-02 1.67785902e-02
5.33804357e-01 -2.80217677e-01 -3.09837937e-01 8.60112906e-01
7.89157808e-01 -9.23794627e-01 7.76515007e-01 4.55563247e-01
5.18904805e-01 -6.29657388e-01 3.65437776e-01 1.07282519e+00
-6.81646228e-01 -6.40874058e-02 -1.06387027e-01 -3.80833626e-01
2.47760549e-01 5.84520288e-02 -1.30313206e+00 3.25527847e-01
7.76678510e-03 1.40657529e-01 -1.76477805e-01 3.76914144e-01
-4.18126345e-01 6.62315011e-01 1.13874339e-01 -3.31470162e-01
5.18385530e-01 -9.50662643e-02 6.73043370e-01 1.21663201e+00
-2.49472529e-01 2.54115343e-01 8.46602976e-01 7.82854140e-01
2.98598677e-01 1.20369427e-01 -7.78585017e-01 -3.79440039e-01
1.45993054e-01 1.22127712e+00 -4.19202745e-01 -3.43136758e-01
1.60233062e-02 9.00015891e-01 4.85158741e-01 3.69474918e-01
-4.92509425e-01 -8.28510225e-02 7.80033469e-01 1.76677570e-01
-1.61805540e-01 -5.65830469e-01 -7.55500793e-02 -1.01970780e+00
-4.26180512e-01 -1.32339394e+00 2.29771450e-01 -5.40487647e-01
-7.72103429e-01 8.14839005e-01 1.25168920e-01 -9.59001601e-01
-1.00119293e+00 -5.10196030e-01 -3.14322293e-01 1.07710898e+00
-1.56455898e+00 -9.20286536e-01 -6.46423623e-02 4.22542363e-01
1.11698461e+00 -1.75168023e-01 1.09697676e+00 -2.63992429e-01
-4.30954963e-01 1.10419720e-01 -4.31409359e-01 -1.80519387e-01
9.26576674e-01 -1.34094870e+00 3.21274638e-01 6.36779368e-01
-2.47778237e-01 5.20731807e-01 1.06768811e+00 -6.05324864e-01
-1.46625292e+00 -1.01200485e+00 1.12519705e+00 -5.77166617e-01
6.91489875e-01 -5.82889497e-01 -7.03612030e-01 5.11721432e-01
3.50417167e-01 -5.65252721e-01 3.60311449e-01 2.25307703e-01
1.62173122e-01 1.84280962e-01 -8.65717292e-01 8.65772188e-01
7.63730407e-01 -4.18232501e-01 -7.37164855e-01 7.30469167e-01
9.35490906e-01 -6.76281452e-01 -6.69194758e-01 9.31460857e-02
1.02776915e-01 -8.24662149e-01 5.54126859e-01 -1.01366889e+00
3.44450861e-01 -1.08639542e-02 9.69879236e-03 -1.73167098e+00
-2.97617137e-01 -1.09274673e+00 -5.97165748e-02 1.00824034e+00
4.50159222e-01 -5.60282707e-01 5.38880944e-01 8.86866093e-01
-1.97654024e-01 -6.64562404e-01 -5.00472724e-01 -6.53439999e-01
2.75430113e-01 -2.73430705e-01 1.61025092e-01 6.07672155e-01
2.28797942e-01 9.41168487e-01 -6.70047879e-01 -1.14415653e-01
2.46387079e-01 4.02045473e-02 1.01308942e+00 -1.00061226e+00
-3.27546716e-01 -2.84967154e-01 4.04038310e-01 -1.45394194e+00
5.70163846e-01 -6.81781292e-01 6.84084356e-01 -1.44689620e+00
-1.94628835e-01 -4.04486626e-01 3.60978283e-02 4.82355028e-01
-1.60771921e-01 -5.80553055e-01 4.56824005e-01 7.56157637e-02
-9.06391025e-01 5.86423218e-01 1.49088597e+00 8.03190097e-03
-6.29307926e-01 2.47364685e-01 -8.93087029e-01 6.26050889e-01
8.52820933e-01 -4.78411198e-01 -6.21392608e-01 -2.56269306e-01
-2.08777368e-01 8.51252496e-01 -2.94006467e-02 -8.13619375e-01
3.26447159e-01 -6.94142759e-01 -2.81117857e-01 1.15448907e-02
3.98436487e-01 -4.42202240e-01 -3.51647764e-01 6.67414188e-01
-1.05151641e+00 6.69656880e-03 1.34969771e-01 6.01407051e-01
-1.90184325e-01 -3.16746086e-01 8.19553494e-01 -4.03573632e-01
-7.06296563e-01 1.22430809e-01 -1.02607036e+00 3.40724975e-01
9.44659173e-01 2.08532423e-01 -9.55597162e-02 -8.54219198e-01
-7.63308406e-01 7.42886484e-01 2.16952115e-01 5.58650196e-01
3.28628749e-01 -8.51876497e-01 -9.09312665e-01 -8.86805877e-02
2.02906132e-01 -8.09157565e-02 3.12913992e-02 5.44897139e-01
-2.26690724e-01 1.02193487e+00 -1.67735349e-02 -5.35249650e-01
-1.04946280e+00 4.08482075e-01 5.98462284e-01 -1.01081491e+00
-2.34620720e-01 5.39566755e-01 4.18279231e-01 -8.37360680e-01
2.49063373e-01 -1.45467475e-01 -3.76668066e-01 4.91735302e-02
4.10045594e-01 -3.43216993e-02 -5.22202738e-02 -2.95042753e-01
-4.86854166e-02 8.54853541e-02 -1.15266494e-01 -4.95311320e-01
8.17377508e-01 -2.95315742e-01 1.57274917e-01 6.92531347e-01
5.42840958e-01 -3.63684177e-01 -1.52287436e+00 -2.07769677e-01
2.79031754e-01 -1.50439873e-01 1.71981066e-01 -1.26715887e+00
-3.43404859e-01 8.91052663e-01 2.91765273e-01 4.01813745e-01
7.96711802e-01 -3.37237656e-01 4.85298634e-01 7.03842521e-01
6.58002377e-01 -1.18954635e+00 4.22385812e-01 9.55321014e-01
1.04365122e+00 -1.38642347e+00 -3.63979161e-01 -1.57262966e-01
-1.21836102e+00 1.01901865e+00 8.62665594e-01 5.70275821e-02
2.28818759e-01 1.33247048e-01 3.60662341e-01 -1.10106610e-01
-1.50559568e+00 -3.43879312e-01 1.26495793e-01 6.22263908e-01
5.65925062e-01 1.68700993e-01 -3.19124311e-01 3.15581203e-01
-1.11060955e-01 1.26938805e-01 5.19699931e-01 1.01588666e+00
-7.07741022e-01 -1.35342598e+00 -3.52623820e-01 2.87909269e-01
-2.91112155e-01 -6.55985205e-03 -6.73944652e-01 7.12641597e-01
-4.70492899e-01 1.26077998e+00 -2.16826066e-01 -1.43070936e-01
6.19013667e-01 5.26525438e-01 2.89294302e-01 -1.14142013e+00
-8.64373744e-01 6.05728067e-02 6.52185798e-01 -5.22879362e-01
-2.52146870e-01 -5.24399996e-01 -1.07309186e+00 -1.19875878e-01
-1.81668818e-01 4.17216033e-01 5.02244115e-01 1.02189457e+00
2.36244231e-01 7.48846233e-01 6.92215800e-01 -8.72272313e-01
-1.24006009e+00 -1.00356317e+00 -3.22704136e-01 1.89166725e-01
3.82595271e-01 -5.02020061e-01 -1.80598930e-01 4.98121604e-02]
|
[13.048102378845215, 8.048705101013184]
|
7e240495-2e20-4a91-a406-2c6601ae1b7e
|
real-time-slam-pipeline-in-dynamics
|
2303.02272
| null |
https://arxiv.org/abs/2303.02272v1
|
https://arxiv.org/pdf/2303.02272v1.pdf
|
Real-time SLAM Pipeline in Dynamics Environment
|
Inspired by the recent success of application of dense data approach by using ORB-SLAM and RGB-D SLAM, we propose a better pipeline of real-time SLAM in dynamics environment. Different from previous SLAM which can only handle static scenes, we are presenting a solution which use RGB-D SLAM as well as YOLO real-time object detection to segment and remove dynamic scene and then construct static scene 3D. We gathered a dataset which allows us to jointly consider semantics, geometry, and physics and thus enables us to reconstruct the static scene while filtering out all dynamic objects.
|
['Lingjie Kong', 'Alex Fu']
|
2023-03-04
| null | null | null | null |
['real-time-object-detection']
|
['computer-vision']
|
[-5.05315103e-02 -3.65748823e-01 4.40244555e-01 -5.45035899e-01
-2.83502400e-01 -6.91373348e-01 6.63964391e-01 1.00777142e-01
-5.92139781e-01 6.75549746e-01 -1.22906737e-01 -1.11929797e-01
-2.72443205e-01 -8.66432250e-01 -5.67866683e-01 -2.24300861e-01
-2.12323502e-01 8.68682444e-01 7.86209345e-01 -5.77300787e-01
2.58960664e-01 9.07188356e-01 -1.72295892e+00 -2.06599578e-01
7.28676915e-01 6.71296120e-01 6.07113481e-01 8.11746955e-01
-1.85452521e-01 6.14749610e-01 -2.56460696e-01 3.85310084e-01
6.50809944e-01 -3.85780334e-01 -6.56395376e-01 1.69026434e-01
8.63221526e-01 -9.80956629e-02 -6.99991345e-01 1.09811223e+00
3.74305844e-01 4.81052607e-01 -2.43814606e-02 -1.09724486e+00
3.33435126e-02 -1.98014930e-01 -3.63674164e-01 9.98568311e-02
1.11692071e+00 1.36630908e-01 5.11206269e-01 -8.58697534e-01
1.24011111e+00 1.31970143e+00 7.63865054e-01 1.01086117e-01
-1.09580529e+00 -2.97373205e-01 7.32882097e-02 2.97785759e-01
-1.52835894e+00 -3.89891952e-01 5.80756962e-01 -1.98170215e-01
1.12530315e+00 5.41473150e-01 1.02337396e+00 6.78604901e-01
4.08068180e-01 4.93339866e-01 1.24738705e+00 -5.37647843e-01
2.18025237e-01 -6.51294515e-02 1.33730680e-01 7.29233146e-01
3.55609477e-01 3.85849506e-01 -7.33682454e-01 1.35588899e-01
7.02055216e-01 4.34945047e-01 -1.95656925e-01 -1.17580426e+00
-1.56152225e+00 6.04917526e-01 6.48095727e-01 -1.13010220e-01
-3.44803542e-01 3.85532051e-01 1.68881565e-01 2.27198109e-01
5.60500808e-02 1.60326153e-01 -4.45972294e-01 -6.45208955e-02
-9.86594677e-01 5.31147897e-01 6.05161488e-01 1.19645977e+00
1.35372746e+00 -4.25895303e-03 4.10445124e-01 -2.88332049e-02
4.76849377e-01 9.35574174e-01 2.64109164e-01 -1.14412141e+00
6.49363101e-02 7.72772372e-01 2.61673719e-01 -9.77929890e-01
-6.02031827e-01 -2.48342410e-01 -3.49841386e-01 4.60464031e-01
3.32524953e-03 3.81978184e-01 -1.15362322e+00 8.57162297e-01
7.87749588e-01 3.15766543e-01 4.31051329e-02 1.10608137e+00
7.47045755e-01 1.58395618e-01 -3.36536437e-01 8.91338289e-02
9.03795063e-01 -8.63707066e-01 -8.79307389e-01 -3.66425455e-01
6.66442454e-01 -9.19055223e-01 6.77909076e-01 6.08490109e-01
-4.15752351e-01 -6.74598336e-01 -1.03075254e+00 -3.80883396e-01
-5.62995315e-01 -2.49320984e-01 1.14085257e+00 3.49990785e-01
-1.24875104e+00 4.64729995e-01 -1.02546549e+00 -9.81491446e-01
-1.63813978e-01 2.80885488e-01 -7.96246827e-01 -2.46663690e-01
-9.02312577e-01 1.32044637e+00 7.66301990e-01 3.32824528e-01
-1.14695454e+00 -3.60042602e-01 -1.12639296e+00 -5.99078953e-01
5.07081270e-01 -9.17786241e-01 7.37217188e-01 -4.51035708e-01
-1.42232406e+00 1.01121664e+00 -3.48451227e-01 -5.75725973e-01
6.39029980e-01 -5.19778728e-01 -1.79349601e-01 5.40100709e-02
1.07891917e-01 4.48839366e-01 2.23293304e-01 -1.36869061e+00
-8.55376184e-01 -6.07872963e-01 5.99277345e-03 4.45889831e-01
7.91813314e-01 -3.90736639e-01 -4.43314850e-01 2.24146888e-01
1.29174829e+00 -9.88948703e-01 -5.10957062e-01 -1.55392230e-01
-2.52279431e-01 4.61186737e-01 1.12573099e+00 -4.45526987e-01
5.77816069e-01 -1.85306311e+00 4.08793092e-01 1.66367859e-01
1.81472138e-01 -3.19729984e-01 1.78965330e-01 4.14778650e-01
3.35187733e-01 -5.48735023e-01 -8.79531726e-03 -7.14540660e-01
1.23979986e-01 7.95362353e-01 -3.16438168e-01 1.05531049e+00
-4.36742187e-01 8.07529151e-01 -9.00269449e-01 -5.95093906e-01
1.13395882e+00 4.75204647e-01 -3.99058431e-01 -5.80169857e-02
-2.84204155e-01 7.70399749e-01 -4.28855151e-01 8.91358674e-01
1.15074289e+00 4.44440484e-01 -1.18644476e-01 2.37460732e-01
-6.74847543e-01 2.42439225e-01 -1.78320396e+00 2.69433236e+00
-3.11577022e-01 7.25659311e-01 3.91300738e-01 -5.64132333e-01
1.00625360e+00 -2.91566581e-01 5.51090360e-01 -8.88583839e-01
8.24690461e-02 1.68462530e-01 -5.37445068e-01 -2.85741836e-01
1.07283163e+00 -1.37652069e-01 -1.82448611e-01 -3.42182636e-01
1.78626508e-01 -9.38223839e-01 -1.86143070e-01 5.77257812e-01
1.20067334e+00 1.12624085e+00 2.68817544e-01 -3.95630836e-01
5.48626125e-01 8.61912131e-01 4.87881869e-01 8.10106754e-01
-3.99242878e-01 6.14216447e-01 -2.96074390e-01 -7.14696825e-01
-7.59246588e-01 -1.38220692e+00 -7.38039836e-02 7.21179068e-01
9.77355838e-01 -3.00659359e-01 5.33973351e-02 -2.36143947e-01
2.76135027e-01 3.97415400e-01 -4.13765281e-01 1.53939873e-01
-5.05805135e-01 -3.58597517e-01 -1.09296381e-01 -3.91277611e-01
4.08731133e-01 -7.76318014e-01 -1.01723170e+00 2.54025996e-01
-7.65582696e-02 -1.14137661e+00 2.47689575e-01 4.89328176e-01
-9.33616042e-01 -1.17988622e+00 2.04273313e-01 -2.13983238e-01
2.18815312e-01 5.90930104e-01 1.10513294e+00 -1.70902669e-01
-7.91853011e-01 6.31651521e-01 -5.49998820e-01 -4.16719675e-01
-1.26954153e-01 -2.08980113e-01 2.16574430e-01 -4.04183894e-01
2.63912588e-01 -4.91832852e-01 -3.53695512e-01 8.54106694e-02
-5.49624860e-01 3.01479369e-01 2.98857570e-01 -7.11267292e-02
9.35965180e-01 -1.02767041e-02 -4.91765082e-01 -7.16640651e-01
-4.64588284e-01 -2.32761025e-01 -1.06122386e+00 -1.22179300e-01
-2.71965384e-01 -3.08942586e-01 -8.68232772e-02 1.02535263e-01
-9.38997328e-01 8.43733788e-01 1.33527012e-03 -3.86509478e-01
-4.24787641e-01 5.90778794e-03 -7.23962709e-02 -6.36419713e-01
5.60204268e-01 3.92405033e-01 -1.74162745e-01 -7.61260748e-01
4.90106821e-01 2.58590907e-01 8.33392441e-01 -2.73163438e-01
1.07284617e+00 1.22402287e+00 3.99540633e-01 -8.97240520e-01
-7.25857675e-01 -1.09262991e+00 -1.30901515e+00 -3.53657305e-01
9.13533390e-01 -1.09110808e+00 -7.16117144e-01 1.88099220e-01
-1.17206061e+00 -1.78333566e-01 -5.50067604e-01 7.38638520e-01
-7.31318414e-01 5.64985693e-01 -1.50467351e-01 -9.44267035e-01
2.08859935e-01 -9.42185342e-01 1.34397471e+00 -4.05766368e-02
2.02771530e-01 -7.21955180e-01 5.00414610e-01 2.16742977e-03
3.04151177e-01 8.41902673e-01 -1.44309998e-01 -2.29318012e-02
-1.27483475e+00 -3.04059803e-01 -1.16817661e-01 -3.62239480e-01
5.37930652e-02 -2.20621735e-01 -8.11151266e-01 -1.81607649e-01
2.23010898e-01 7.06991032e-02 9.07905281e-01 1.23026341e-01
5.66832460e-02 4.35086340e-01 -5.64260840e-01 1.18905485e+00
2.04520702e+00 -2.44030192e-01 7.36950219e-01 9.19313729e-01
9.91233885e-01 3.59288305e-01 1.11312008e+00 5.00515461e-01
6.51068449e-01 8.30679774e-01 9.18385088e-01 -5.72223440e-02
-3.42333317e-01 -2.49686971e-01 1.64524063e-01 6.59243107e-01
-1.45161839e-03 1.74662799e-01 -1.05213296e+00 3.32588732e-01
-2.06952000e+00 -5.82159102e-01 -8.67333055e-01 2.13660622e+00
2.87085533e-01 2.24966869e-01 -3.14130813e-01 -6.38722181e-02
1.39489889e-01 1.33928627e-01 -1.98022142e-01 -9.34044421e-02
-3.08271050e-01 1.28665134e-01 1.29987395e+00 9.90602076e-01
-9.81451392e-01 1.42797244e+00 6.80662441e+00 3.14659685e-01
-9.64100361e-01 3.28813732e-01 -8.81508827e-01 -1.38458312e-01
-2.77199119e-01 7.50647128e-01 -1.06890738e+00 2.57521216e-02
8.31788421e-01 2.29506731e-01 3.74224156e-01 9.52767909e-01
2.90720284e-01 -9.27565753e-01 -8.23625326e-01 9.72245932e-01
3.21451295e-03 -1.25182915e+00 -2.02917710e-01 9.55916792e-02
7.51469791e-01 5.77138901e-01 -7.59506047e-01 2.86011875e-01
6.43350005e-01 -5.84727883e-01 1.00287759e+00 8.38230491e-01
2.39892378e-01 -2.39838183e-01 5.84581852e-01 6.85435474e-01
-1.27147400e+00 2.37397566e-01 -5.60970783e-01 -6.53250098e-01
3.08078319e-01 6.52016461e-01 -8.52834284e-01 1.19264388e+00
8.03268909e-01 7.31719911e-01 -7.68479288e-01 1.32199228e+00
-1.68819144e-01 -2.86841780e-01 -7.40245879e-01 2.84481704e-01
1.96211398e-01 -4.36435610e-01 7.41579235e-01 1.03250837e+00
2.76160866e-01 -1.60829380e-01 7.45826423e-01 4.51286972e-01
7.49939978e-01 -2.08874896e-01 -8.65228772e-01 7.91182697e-01
9.91259888e-02 1.07659483e+00 -8.05923402e-01 -3.93247545e-01
-1.82372794e-01 1.27119863e+00 1.06862858e-02 1.52672574e-01
-6.77555799e-01 -1.27831548e-01 6.98206007e-01 2.81769037e-01
1.97320938e-01 -1.23257744e+00 -1.70630127e-01 -1.33621347e+00
-3.82347465e-01 -1.54217795e-01 1.68324620e-01 -1.15962076e+00
-4.24950391e-01 3.83001536e-01 5.39992824e-02 -1.11971462e+00
1.00276925e-01 -4.44487602e-01 4.11358625e-02 9.10108328e-01
-1.84901667e+00 -1.43761289e+00 -8.21801901e-01 8.72182488e-01
4.57475275e-01 2.85645604e-01 7.42246270e-01 2.04190537e-01
1.03557184e-01 -7.29706049e-01 1.33510813e-01 -4.36041564e-01
7.14925706e-01 -1.30104053e+00 2.09566534e-01 1.25829017e+00
4.30172652e-01 4.49533522e-01 1.33315027e+00 -1.09339845e+00
-2.20597601e+00 -8.00325871e-01 7.01320052e-01 -9.99066591e-01
6.28111005e-01 -7.00267553e-01 -5.14896691e-01 1.06885493e+00
-3.06861401e-01 2.79515177e-01 -1.94936231e-01 -2.04771891e-01
3.62383723e-02 -1.83070540e-01 -1.18022168e+00 9.70535055e-02
1.46771431e+00 -3.38273019e-01 -7.15158761e-01 5.02022386e-01
9.37902451e-01 -1.14790356e+00 -3.25918525e-01 5.68518162e-01
2.05494553e-01 -1.34223318e+00 1.01657176e+00 -7.15558529e-02
-6.83374345e-01 -7.61021376e-01 -5.89280665e-01 -6.94685817e-01
-6.94126710e-02 -6.74046814e-01 -6.14377931e-02 7.54489601e-01
-4.71535742e-01 -6.23760998e-01 9.12690699e-01 2.85383850e-01
-6.12340808e-01 4.38875884e-01 -1.32202637e+00 -7.13439643e-01
-8.51430655e-01 -7.43667781e-01 4.13392991e-01 7.53948092e-01
-7.75453925e-01 -2.07901478e-01 -3.91810656e-01 5.77057660e-01
9.51499522e-01 6.06971681e-01 1.32398903e+00 -1.52448034e+00
3.74811329e-02 1.49335802e-01 -1.03549612e+00 -9.37816560e-01
9.24729854e-02 -7.31391311e-01 6.19562209e-01 -1.84945607e+00
-1.37598827e-01 -6.25035882e-01 -8.36315006e-03 1.85786545e-01
4.94005889e-01 4.70674127e-01 -3.00804209e-02 4.01871979e-01
-1.08850801e+00 4.45825994e-01 1.12147963e+00 4.68785614e-01
-2.20297709e-01 -3.38639438e-01 1.74953312e-01 7.72303522e-01
1.17105439e-01 -6.36696458e-01 4.99462597e-02 -4.71494794e-01
3.12682480e-01 -8.98119528e-03 6.53884768e-01 -1.30230188e+00
3.07815939e-01 -4.32365268e-01 3.36614460e-01 -1.30275404e+00
6.64875627e-01 -1.26655066e+00 4.76180077e-01 7.49047160e-01
6.43144906e-01 -5.09253666e-02 2.66900271e-01 6.63031578e-01
-1.60558864e-01 -7.72246867e-02 5.07721543e-01 -6.86052859e-01
-1.68746650e+00 2.61522442e-01 -7.09257051e-02 -6.40506625e-01
1.12430394e+00 -6.63107872e-01 1.27117224e-02 -2.88335830e-02
-9.51297283e-01 3.47584724e-01 1.04489934e+00 4.13906127e-01
6.00051045e-01 -1.15285623e+00 -3.75681609e-01 5.43500900e-01
-1.20417383e-02 4.61892962e-01 3.95127654e-01 9.30837452e-01
-1.37092710e+00 6.11937642e-01 -5.00585973e-01 -1.02261114e+00
-1.17117512e+00 6.60615206e-01 2.98620701e-01 2.45457932e-01
-9.86078858e-01 5.80031574e-01 -7.75835514e-02 -8.54159892e-01
1.49903521e-01 -3.35357040e-01 2.57274330e-01 -2.47554243e-01
2.20365673e-01 4.17603552e-01 2.86666214e-01 -9.78606164e-01
-9.47846949e-01 8.27997327e-01 8.05635452e-01 -2.37476096e-01
1.41596770e+00 -9.18075681e-01 -4.04312611e-01 7.53053784e-01
8.96970451e-01 3.55278760e-01 -1.12508297e+00 -1.76691636e-01
2.19306588e-01 -9.06349599e-01 1.41549706e-01 -3.79030436e-01
-2.96055138e-01 6.64037287e-01 8.75984848e-01 -2.02299636e-02
8.54123890e-01 -4.66087051e-02 3.86215270e-01 4.21041012e-01
1.13275635e+00 -7.97473252e-01 -1.88982770e-01 1.04969275e+00
6.89846218e-01 -1.37823105e+00 6.92095518e-01 -5.16205549e-01
-2.58879036e-01 1.11091983e+00 4.41029429e-01 -4.98103589e-01
3.02857876e-01 3.49664360e-01 9.66922194e-02 -4.87425506e-01
-2.20998332e-01 -7.68563390e-01 -4.84347343e-02 6.87688291e-01
-1.43064469e-01 -5.32635450e-02 1.61298946e-01 -3.73882383e-01
-4.07581359e-01 5.79876676e-02 6.35639429e-01 1.33228946e+00
-9.82875586e-01 -9.93848801e-01 -7.07970560e-01 -2.73997039e-01
2.42592305e-01 2.01795653e-01 -2.18773201e-01 1.25431132e+00
4.98585075e-01 4.38516229e-01 2.61838101e-02 -2.25278676e-01
5.00855863e-01 7.64982998e-02 8.14143896e-01 -8.86865675e-01
-3.86178881e-01 1.55154467e-01 -3.56261916e-02 -1.20117915e+00
-4.90588665e-01 -6.59796834e-01 -1.53702617e+00 -3.24027777e-01
-2.87120104e-01 -1.40411466e-01 1.34944177e+00 8.90490294e-01
1.34121537e-01 4.05686677e-01 2.28353888e-01 -1.36054802e+00
-7.14450106e-02 -8.56291056e-01 -8.62069726e-01 3.36310446e-01
7.68141627e-01 -6.95808411e-01 -2.06086949e-01 -1.69409931e-01]
|
[7.2951226234436035, -2.254598379135132]
|
db778a28-da02-41ca-b94c-976a3a4f7fc9
|
towards-unified-keyframe-propagation-models
|
2205.09731
| null |
https://arxiv.org/abs/2205.09731v1
|
https://arxiv.org/pdf/2205.09731v1.pdf
|
Towards Unified Keyframe Propagation Models
|
Many video editing tasks such as rotoscoping or object removal require the propagation of context across frames. While transformers and other attention-based approaches that aggregate features globally have demonstrated great success at propagating object masks from keyframes to the whole video, they struggle to propagate high-frequency details such as textures faithfully. We hypothesize that this is due to an inherent bias of global attention towards low-frequency features. To overcome this limitation, we present a two-stream approach, where high-frequency features interact locally and low-frequency features interact globally. The global interaction stream remains robust in difficult situations such as large camera motions, where explicit alignment fails. The local interaction stream propagates high-frequency details through deformable feature aggregation and, informed by the global interaction stream, learns to detect and correct errors of the deformation field. We evaluate our two-stream approach for inpainting tasks, where experiments show that it improves both the propagation of features within a single frame as required for image inpainting, as well as their propagation from keyframes to target frames. Applied to video inpainting, our approach leads to 44% and 26% improvements in FID and LPIPS scores. Code at https://github.com/runwayml/guided-inpainting
|
['Soumyadip Sengupta', 'Peter Michael', 'Patrick Esser']
|
2022-05-19
| null | null | null | null |
['video-inpainting']
|
['computer-vision']
|
[ 2.91460723e-01 -1.69726819e-01 1.05566248e-01 -1.34007305e-01
-9.84527647e-01 -5.27501643e-01 6.68999791e-01 1.55611917e-01
-3.09612364e-01 5.78721404e-01 5.08354604e-01 2.80941248e-01
-5.68450429e-02 -5.87505579e-01 -1.11525989e+00 -5.14979362e-01
-2.31846124e-01 1.77432045e-01 4.66772735e-01 -1.84993073e-01
4.28705156e-01 5.03479600e-01 -1.49106729e+00 5.51373243e-01
6.73072219e-01 6.72863305e-01 2.49227598e-01 1.17303348e+00
2.18518168e-01 9.98776674e-01 -4.27051544e-01 -1.02739677e-01
3.38950038e-01 -5.09176791e-01 -8.36112320e-01 1.39488727e-01
1.14065218e+00 -7.47101903e-01 -3.26740921e-01 9.32165921e-01
3.97847503e-01 4.40829128e-01 3.46823037e-01 -8.65715086e-01
-5.31282783e-01 1.51325107e-01 -6.73665583e-01 4.14554626e-01
7.22903430e-01 3.93071979e-01 9.12874937e-01 -1.19922602e+00
1.13527071e+00 1.25997984e+00 7.47696638e-01 4.88827169e-01
-1.25163257e+00 -2.64679730e-01 3.01510602e-01 1.13935478e-01
-1.01311946e+00 -6.08426809e-01 6.71338856e-01 -7.41313756e-01
1.10969162e+00 3.60199928e-01 6.73580766e-01 7.71587253e-01
5.27043760e-01 5.93272090e-01 5.81109405e-01 -2.77802020e-01
-1.09613672e-01 -3.93307269e-01 -8.17954838e-02 7.85626650e-01
-8.79126191e-02 3.68966877e-01 -1.02127099e+00 -1.07296139e-01
1.15762973e+00 -6.93887845e-02 -5.30020297e-01 -3.49322483e-02
-1.41223860e+00 6.88533127e-01 3.43301773e-01 1.90421924e-01
-6.26870036e-01 5.67755461e-01 1.54049054e-01 2.79871672e-01
7.64529347e-01 5.40715873e-01 -4.13107425e-01 -3.58159870e-01
-1.10094273e+00 5.74967146e-01 4.94355649e-01 8.05638134e-01
8.60178113e-01 -1.73216507e-01 -2.88999915e-01 6.93246901e-01
3.46995071e-02 2.50026166e-01 1.63347468e-01 -1.44730663e+00
4.08036172e-01 4.66770492e-02 3.03084075e-01 -1.38654327e+00
-1.91195890e-01 -5.73867983e-05 -3.69683594e-01 5.50903141e-01
3.71022463e-01 -1.71700522e-01 -1.03898287e+00 1.79804242e+00
5.88713288e-01 4.30299342e-01 -4.51897442e-01 1.12026036e+00
6.32602453e-01 7.40218759e-01 -7.65272155e-02 -8.71431977e-02
1.12860584e+00 -1.15276599e+00 -6.66525900e-01 -2.58259147e-01
2.70383865e-01 -1.24107277e+00 7.75147676e-01 3.00012469e-01
-1.53244829e+00 -6.32241011e-01 -8.02233219e-01 -4.82053906e-01
2.02129930e-01 -3.90577495e-01 3.35177451e-01 -1.01692006e-01
-1.12252378e+00 1.00536728e+00 -1.05074275e+00 -3.76893222e-01
4.30216163e-01 3.84434372e-01 -5.90630412e-01 -1.62407160e-01
-7.64542997e-01 8.81215394e-01 -1.17748633e-01 -9.70072299e-02
-9.47246850e-01 -1.03948975e+00 -8.23519766e-01 -3.14477496e-02
3.33049119e-01 -9.01244879e-01 1.23660755e+00 -1.22768176e+00
-1.36799967e+00 4.36921209e-01 -3.89179200e-01 -1.52736604e-01
7.40936100e-01 -8.77706230e-01 3.29756021e-01 4.47623253e-01
1.87802449e-01 9.27183986e-01 1.11782527e+00 -1.16676199e+00
-7.89367199e-01 6.32035285e-02 1.73934296e-01 5.06875217e-01
-6.92800954e-02 1.55145809e-01 -6.53074026e-01 -1.08020818e+00
-5.63277826e-02 -8.60423148e-01 -6.61964640e-02 3.81039023e-01
-7.37838745e-02 1.26101896e-01 1.04047668e+00 -1.02219701e+00
9.38548863e-01 -2.05810404e+00 6.93094552e-01 -7.31370598e-02
2.96398103e-01 -3.95884439e-02 -1.94588035e-01 3.89969409e-01
-7.36097172e-02 -3.06361951e-02 -2.09221363e-01 -5.99390626e-01
-4.41325814e-01 1.08899921e-01 -2.90618539e-01 4.64371473e-01
5.92516780e-01 7.22831845e-01 -1.15003943e+00 -3.93598109e-01
3.93700570e-01 8.64806294e-01 -1.08469975e+00 2.64325798e-01
-2.94917136e-01 7.46368527e-01 -1.87706515e-01 5.21031916e-01
4.72330898e-01 -1.37737110e-01 -2.24072143e-01 -1.94749981e-01
-2.20795199e-01 2.80263901e-01 -1.08692229e+00 1.97606409e+00
-2.63003230e-01 9.20879841e-01 1.83133915e-01 -5.63434601e-01
3.48903507e-01 3.22978765e-01 7.17363715e-01 -2.82785565e-01
-1.06470816e-01 -3.13268825e-02 -1.67449564e-01 -5.52215934e-01
6.63661957e-01 -4.11661752e-02 3.58674109e-01 3.95836174e-01
2.39060000e-01 -3.04353416e-01 1.55099630e-01 3.69157225e-01
1.23871422e+00 5.73145270e-01 2.41598990e-02 -1.21281184e-01
1.81707427e-01 -4.57994118e-02 4.56563324e-01 7.21408248e-01
4.56559882e-02 1.27348888e+00 3.53363633e-01 -5.91217101e-01
-1.11367869e+00 -8.55706871e-01 2.17298225e-01 1.11633587e+00
1.74965471e-01 -7.43560970e-01 -7.41001248e-01 -5.11537194e-01
3.23930047e-02 2.86896706e-01 -7.58048594e-01 -7.38819316e-02
-8.52463782e-01 -2.72188783e-01 1.53487977e-02 5.34401417e-01
1.93402842e-01 -1.04008007e+00 -7.35659301e-01 4.05116171e-01
-3.78140271e-01 -1.07600427e+00 -8.53693187e-01 -8.19619969e-02
-9.33719337e-01 -1.04020631e+00 -8.06706727e-01 -5.53907394e-01
8.33960474e-01 2.25962222e-01 1.13607264e+00 5.37336469e-01
-5.22273004e-01 5.23522615e-01 -3.86429161e-01 5.25616184e-02
-3.83375287e-01 -1.87674120e-01 -2.01638073e-01 1.38381934e-02
-2.88073450e-01 -5.20970166e-01 -7.72018790e-01 1.79636374e-01
-8.68207693e-01 2.14355230e-01 2.59158105e-01 9.39157546e-01
5.71500182e-01 -3.75114113e-01 5.28891608e-02 -7.61219919e-01
3.67189914e-01 -3.48272651e-01 -4.23009783e-01 -1.09635465e-01
-5.77963598e-04 -7.42060244e-02 4.03434575e-01 -5.51894665e-01
-8.80843580e-01 1.44372150e-01 -1.04485445e-01 -7.92105019e-01
-5.17073460e-02 3.40643167e-01 3.11646193e-01 -2.96004802e-01
7.33531892e-01 -2.88320143e-06 2.13015117e-02 -3.48986745e-01
2.84493029e-01 -2.32971441e-02 5.95772207e-01 -7.78826356e-01
7.99676001e-01 5.29003561e-01 -1.78901777e-01 -9.02804971e-01
-6.69966400e-01 -2.07633018e-01 -5.04012406e-01 -4.72128153e-01
1.01772809e+00 -8.14399183e-01 -2.70400226e-01 5.67623734e-01
-1.35238171e+00 -6.25795364e-01 -4.26411062e-01 3.97038877e-01
-8.08302224e-01 4.27931249e-01 -8.27209294e-01 -3.02623063e-01
-1.15407169e-01 -1.29634452e+00 1.34556675e+00 9.87460241e-02
-5.46368003e-01 -9.57642078e-01 6.06897213e-02 6.57871217e-02
3.65227431e-01 4.78854924e-01 5.97531319e-01 4.54567820e-02
-8.93504918e-01 -4.35232036e-02 8.09079781e-02 2.75913715e-01
3.44079614e-01 3.17781180e-01 -8.18903148e-01 -3.66123557e-01
1.34305954e-02 -1.78959101e-01 8.68287444e-01 5.46293676e-01
8.53897333e-01 -2.67811805e-01 -1.04267761e-01 7.10843861e-01
1.16163516e+00 -5.31914793e-02 6.38582587e-01 1.40162721e-01
9.29813921e-01 4.89486337e-01 6.79579854e-01 3.67977053e-01
1.63435519e-01 8.83135796e-01 3.42579782e-01 -2.01678365e-01
-5.69728673e-01 -9.74628255e-02 4.84966576e-01 6.30248308e-01
-4.89162147e-01 -1.18286826e-01 -7.18635261e-01 5.00462055e-01
-1.97519231e+00 -1.13843274e+00 -3.15034948e-02 2.22423697e+00
1.00194550e+00 6.48846701e-02 1.32415686e-02 -1.87996328e-01
6.78408980e-01 1.37555435e-01 -2.95317441e-01 -2.78214514e-01
8.45806971e-02 2.29042247e-01 1.83327198e-01 1.28124154e+00
-1.07247972e+00 9.83305037e-01 6.19740868e+00 5.74643314e-01
-1.27918875e+00 2.78547078e-01 5.90960979e-01 -4.14789438e-01
-2.98497915e-01 4.90405373e-02 -5.11324644e-01 5.39517581e-01
4.89039987e-01 1.83671549e-01 5.44396937e-01 4.36114401e-01
2.43351683e-01 -2.89149106e-01 -1.23618639e+00 7.32101023e-01
9.97057855e-02 -1.59440053e+00 8.59943181e-02 -7.37783983e-02
9.28635061e-01 9.19744670e-02 -2.39693467e-02 -1.64267719e-01
5.77614494e-02 -8.24226797e-01 1.02679503e+00 7.33299196e-01
7.41737664e-01 -6.67859554e-01 3.39524448e-01 2.97547132e-03
-1.09082139e+00 2.36171722e-01 -1.57402292e-01 -1.45646498e-01
2.81928599e-01 5.38979471e-01 -4.64078873e-01 3.79209787e-01
7.31314063e-01 9.68253613e-01 -3.02310616e-01 1.02818227e+00
-2.02129945e-01 3.77689242e-01 -4.79492396e-01 5.48682272e-01
1.50812224e-01 2.39670575e-02 8.98359418e-01 1.22651720e+00
3.16661745e-01 2.64235556e-01 1.51644886e-01 6.83564365e-01
1.02213673e-01 -1.83402479e-01 -4.38624293e-01 7.64959604e-02
1.62961081e-01 1.08321846e+00 -6.61500216e-01 -3.94473404e-01
-4.63503420e-01 1.30372560e+00 3.85570318e-01 5.01254737e-01
-1.00811815e+00 -3.83383006e-01 9.26463664e-01 4.52810079e-01
4.36825216e-01 -3.75921041e-01 -1.21786185e-01 -1.24922943e+00
1.45789146e-01 -7.65998781e-01 1.89793244e-01 -8.87707829e-01
-9.10991669e-01 4.96236175e-01 -6.69595078e-02 -1.12012947e+00
-3.11272770e-01 -2.55631387e-01 -6.27170920e-01 7.37026989e-01
-1.32887983e+00 -1.03548300e+00 -5.63701391e-01 7.31270909e-01
8.71167541e-01 3.05847049e-01 6.02487743e-01 3.89379233e-01
1.95789430e-03 3.88743848e-01 -2.18986765e-01 -7.75567591e-02
1.00397193e+00 -1.05103576e+00 5.13828278e-01 1.06279421e+00
1.99910343e-01 6.04794502e-01 9.70588624e-01 -7.91749537e-01
-1.43462253e+00 -9.47436929e-01 8.14508379e-01 -5.44897854e-01
3.45389754e-01 -1.84802055e-01 -9.71435308e-01 8.38363528e-01
1.62680745e-01 4.02032971e-01 7.27033988e-02 -8.14890936e-02
-2.03219384e-01 1.47802442e-01 -9.59906042e-01 5.88266492e-01
1.01916480e+00 -5.24468720e-01 -4.54040170e-01 5.08021832e-01
5.67809224e-01 -9.05716419e-01 -9.46359813e-01 3.03506821e-01
4.78980213e-01 -1.11014366e+00 9.35560405e-01 -4.39689398e-01
1.01764631e+00 -3.77154082e-01 -1.06660565e-02 -1.20637119e+00
-5.19602954e-01 -1.16201544e+00 -2.93468058e-01 1.04398787e+00
7.43990391e-02 -3.51282924e-01 5.59881628e-01 3.96242231e-01
-3.39683771e-01 -7.71178603e-01 -8.10088456e-01 -2.88599432e-01
-1.22311205e-01 -3.65201324e-01 9.88314822e-02 1.00393832e+00
-1.97990254e-01 -6.39982522e-02 -4.83518422e-01 1.14424087e-01
3.85937154e-01 -5.56750894e-02 8.31574440e-01 -6.79861724e-01
-5.31603575e-01 -2.81790614e-01 -5.73646009e-01 -1.14069641e+00
-9.51367095e-02 -5.94792902e-01 2.80957311e-01 -1.48870051e+00
2.41974115e-01 -2.49353930e-01 -1.40680289e-02 4.92615581e-01
-4.49856311e-01 4.86097932e-01 4.40075785e-01 2.24973366e-01
-4.36993539e-01 2.62407511e-01 1.35768664e+00 9.61830765e-02
-1.20080434e-01 -4.08472151e-01 -3.90072852e-01 7.67348647e-01
3.68511260e-01 -5.82375944e-01 -2.60912329e-01 -9.02237177e-01
1.18213452e-01 1.73367471e-01 5.96456707e-01 -1.07234955e+00
1.86469674e-01 -2.37441093e-01 6.51686430e-01 -2.48367429e-01
4.94983464e-01 -5.75300872e-01 3.28174919e-01 3.05924386e-01
-1.92717761e-01 1.97894335e-01 2.47045115e-01 4.39259052e-01
-2.93458194e-01 2.05291416e-02 8.35352838e-01 -1.53427005e-01
-6.00728214e-01 3.00784349e-01 -3.13718468e-01 2.46166557e-01
9.78599370e-01 -3.54472786e-01 -2.09742054e-01 -6.52249694e-01
-1.10185456e+00 -2.26588212e-02 7.80113459e-01 5.63322842e-01
6.92237675e-01 -1.15402842e+00 -8.91997576e-01 3.59541595e-01
-3.91997993e-01 2.21855402e-01 3.50878716e-01 9.67673838e-01
-8.68147016e-01 -4.00131866e-02 -2.26369783e-01 -8.99342477e-01
-1.43012118e+00 1.14285529e-01 4.03767198e-01 1.25177845e-01
-9.36108351e-01 1.36098301e+00 4.58291590e-01 2.04065606e-01
1.27838179e-01 -4.63879526e-01 3.05974096e-01 5.42127080e-02
5.96206546e-01 2.12950096e-01 2.64337417e-02 -7.42142677e-01
-2.91834116e-01 9.87189412e-01 -2.61394560e-01 -2.22211316e-01
1.43209887e+00 -6.25938326e-02 -1.66256383e-01 2.07379147e-01
1.11632359e+00 2.27378815e-01 -2.03572679e+00 1.75054595e-01
-3.34651321e-01 -9.02843654e-01 -1.24849733e-02 -5.88082492e-01
-1.21338403e+00 6.65892363e-01 2.91519165e-01 1.09602036e-02
1.05645764e+00 2.17413381e-02 8.25251043e-01 -1.62373051e-01
2.36407369e-01 -7.92729318e-01 3.43066573e-01 5.77877402e-01
1.22425115e+00 -1.00346267e+00 2.97023267e-01 -5.38234711e-01
-4.40532118e-01 1.20026457e+00 5.80603957e-01 -5.93287289e-01
4.85204160e-01 5.22590637e-01 -2.74175629e-02 -1.21278085e-01
-9.56465960e-01 1.72802031e-01 3.99623960e-01 4.37864363e-01
5.57506084e-01 -4.10029173e-01 -9.00154263e-02 -6.65831892e-03
-7.24545941e-02 2.55160891e-02 5.89709342e-01 1.19032395e+00
-3.22651386e-01 -9.36601698e-01 -4.09082353e-01 2.70282447e-01
-6.52124465e-01 -2.25818485e-01 -2.41171286e-01 6.97389126e-01
2.25540191e-01 6.37568533e-01 2.53785521e-01 -2.66600639e-01
2.36488834e-01 -1.34029061e-01 7.90948451e-01 -6.98961616e-01
-8.06050181e-01 5.30778408e-01 1.21615253e-01 -1.04657555e+00
-4.27582264e-01 -7.76241243e-01 -1.30375421e+00 -2.97436208e-01
-3.04560840e-01 -1.53300941e-01 3.64273906e-01 7.45571494e-01
4.91758764e-01 6.51116192e-01 3.10368448e-01 -1.64373207e+00
-1.14924029e-01 -7.65085757e-01 -2.18763538e-02 7.15997159e-01
8.11518371e-01 -6.16210699e-01 -4.31550950e-01 5.70459664e-01]
|
[10.746176719665527, -1.2579433917999268]
|
6b57c578-c637-4098-8441-d847878ff4cc
|
translation-based-supervision-for-policy
| null | null |
https://aclanthology.org/2021.emnlp-main.130
|
https://aclanthology.org/2021.emnlp-main.130.pdf
|
Translation-based Supervision for Policy Generation in Simultaneous Neural Machine Translation
|
In simultaneous machine translation, finding an agent with the optimal action sequence of reads and writes that maintain a high level of translation quality while minimizing the average lag in producing target tokens remains an extremely challenging problem. We propose a novel supervised learning approach for training an agent that can detect the minimum number of reads required for generating each target token by comparing simultaneous translations against full-sentence translations during training to generate oracle action sequences. These oracle sequences can then be used to train a supervised model for action generation at inference time. Our approach provides an alternative to current heuristic methods in simultaneous translation by introducing a new training objective, which is easier to train than previous attempts at training the agent using reinforcement learning techniques for this task. Our experimental results show that our novel training method for action generation produces much higher quality translations while minimizing the average lag in simultaneous translation.
|
['Anoop Sarkar', 'Hassan S. Shavarani', 'Ashkan Alinejad']
| null | null | null | null |
emnlp-2021-11
|
['action-generation']
|
['computer-vision']
|
[ 5.83526492e-01 3.32037181e-01 -6.22521460e-01 -2.39754900e-01
-1.57371867e+00 -7.69079864e-01 9.72868621e-01 8.67379084e-02
-4.09503758e-01 1.18754101e+00 -5.21758050e-02 -5.74405551e-01
3.92380476e-01 -8.11498523e-01 -1.01978362e+00 -7.25057721e-01
2.88473845e-01 1.14900565e+00 6.80024996e-02 -7.09078684e-02
4.44066226e-01 2.43840039e-01 -1.12995028e+00 2.13526294e-01
9.32018459e-01 2.42480680e-01 5.18069744e-01 9.44052041e-01
-1.19306520e-01 9.71014500e-01 -9.96200204e-01 -2.12847441e-01
4.24208313e-01 -1.29564393e+00 -1.02288532e+00 1.90625593e-01
1.15831224e-02 -3.91472012e-01 2.66399622e-01 9.06007886e-01
5.02029419e-01 -1.50492504e-01 5.72387874e-01 -1.10883653e+00
-4.57396179e-01 9.80143487e-01 -1.21414840e-01 1.84065863e-01
5.05588055e-01 4.05616045e-01 1.12953627e+00 -4.13536519e-01
7.89063752e-01 1.14099741e+00 9.92925167e-02 7.69653678e-01
-1.46786678e+00 -3.17246497e-01 -1.76237017e-01 9.60650574e-03
-1.07337630e+00 -4.81360853e-01 1.82361811e-01 -9.42482278e-02
1.55781424e+00 4.01700795e-01 5.19596934e-01 1.01729059e+00
4.96235341e-01 7.61593103e-01 1.36759400e+00 -8.30773056e-01
4.64721501e-01 1.95658177e-01 -6.02441192e-01 8.04934025e-01
3.51593941e-02 1.38152301e-01 -6.26168668e-01 -3.36728603e-01
9.07671630e-01 -4.50714916e-01 4.05470133e-02 -6.20384887e-02
-1.58311820e+00 8.22914124e-01 -1.21805988e-01 2.26143762e-01
-4.82947350e-01 5.58064461e-01 1.62655383e-01 8.04587245e-01
2.80849814e-01 9.76561487e-01 -6.12030149e-01 -5.57863057e-01
-8.18563581e-01 3.49907011e-01 1.00204825e+00 1.06523490e+00
9.96848881e-01 -2.72663832e-02 -2.94606149e-01 4.79157597e-01
2.22314056e-02 9.76540685e-01 4.09387410e-01 -1.12541366e+00
6.42831445e-01 3.63999993e-01 5.53655803e-01 -1.15610450e-03
1.85419783e-01 -1.48712382e-01 -2.58719534e-01 8.11475664e-02
2.31105193e-01 -4.08644527e-01 -6.97428584e-01 1.66676188e+00
2.21709028e-01 3.94380502e-02 3.18604022e-01 5.37378967e-01
-4.40389097e-01 1.01164615e+00 -4.33537304e-01 -7.14502633e-01
9.35536146e-01 -1.20279348e+00 -5.45110524e-01 -4.26666170e-01
9.63389874e-01 -9.82490897e-01 1.00689030e+00 4.75973450e-02
-1.39093363e+00 -1.11971393e-01 -9.69623566e-01 3.36524069e-01
1.85237482e-01 5.91205135e-02 5.18114924e-01 5.20866275e-01
-1.15361524e+00 7.24227428e-01 -1.15174842e+00 -2.74536192e-01
-1.36003926e-01 7.07499325e-01 -7.90947862e-03 2.51063794e-01
-8.23144734e-01 1.35281551e+00 3.54688793e-01 -3.20886463e-01
-1.20374012e+00 -1.16553716e-01 -5.75345874e-01 -9.97965485e-02
6.38083816e-01 -7.33862996e-01 1.93466818e+00 -1.39730835e+00
-2.18955803e+00 4.79722172e-01 -3.46017152e-01 -5.03325343e-01
5.28275251e-01 -2.05880124e-02 3.12619090e-01 -2.43076477e-02
4.60397273e-01 5.72532356e-01 7.43270159e-01 -1.00397730e+00
-7.48641312e-01 -2.62587843e-03 2.60403100e-02 6.26833200e-01
-3.61249335e-02 1.65460154e-01 -1.01030827e-01 -2.03337580e-01
-3.32288384e-01 -1.05091536e+00 -3.79930377e-01 -6.01255059e-01
-3.57072532e-01 -4.45425123e-01 1.18356705e-01 -2.97298878e-01
8.46042454e-01 -1.36467314e+00 7.10353553e-01 1.63208827e-01
-3.80273372e-01 1.70137230e-02 -4.92371053e-01 8.42869878e-01
5.21769404e-01 1.52778625e-01 -1.88199863e-01 -2.39257887e-01
-9.07816440e-02 4.76962239e-01 -3.12155783e-01 5.97749390e-02
3.18046689e-01 1.01617193e+00 -1.19299459e+00 -4.47029293e-01
-2.91696101e-01 -2.04034060e-01 -3.41379374e-01 6.93593860e-01
-8.00755203e-01 4.57096249e-01 -4.49115485e-01 3.16681951e-01
-1.79827493e-02 -2.90238947e-01 7.31674194e-01 7.97612429e-01
-7.98035935e-02 6.05488479e-01 -8.99380624e-01 1.49443150e+00
-6.98662102e-01 3.52641106e-01 -2.15123087e-01 -7.75564730e-01
9.33067322e-01 3.76959473e-01 2.25643337e-01 -9.53047872e-01
5.79042640e-03 5.83963573e-01 4.57669236e-03 -4.36123848e-01
2.91346222e-01 -3.92915130e-01 -8.05270448e-02 1.20629716e+00
-3.08665454e-01 -4.97748226e-01 3.46121460e-01 9.69980359e-02
1.22243249e+00 2.05672354e-01 2.33722463e-01 -9.11864787e-02
3.23879480e-01 2.98841238e-01 5.62266827e-01 1.05138636e+00
2.80259073e-01 2.34089330e-01 6.63864136e-01 -4.13217276e-01
-1.62350178e+00 -8.15008819e-01 5.93001008e-01 9.61721301e-01
-9.33325514e-02 -3.03903818e-01 -1.05620706e+00 -7.52137661e-01
-3.08900267e-01 8.60650897e-01 -2.38618210e-01 -3.19834687e-02
-9.61064398e-01 -6.05756283e-01 5.91499388e-01 4.24492598e-01
1.33898959e-01 -1.31786633e+00 -8.61318111e-01 3.87181789e-01
-5.03535151e-01 -8.96573007e-01 -8.72777283e-01 5.27013063e-01
-1.01801610e+00 -8.34574580e-01 -2.97896296e-01 -9.89463985e-01
9.20016527e-01 1.38117999e-01 1.13277352e+00 2.67691851e-01
2.01102093e-01 -1.92219257e-01 -1.29242420e-01 -3.90976280e-01
-1.30297875e+00 5.43392181e-01 8.13312680e-02 -4.64240849e-01
2.57017501e-02 -2.08314940e-01 -1.76247418e-01 2.81244218e-01
-7.76175022e-01 4.30961519e-01 9.23219562e-01 1.10203481e+00
5.87704301e-01 -2.20535114e-01 6.15643501e-01 -7.93206811e-01
9.46693599e-01 -1.91981867e-01 -9.68700588e-01 4.66177255e-01
-8.72483730e-01 6.69451833e-01 1.01707363e+00 -5.13938785e-01
-8.30159783e-01 1.61793739e-01 1.35321215e-01 1.28357440e-01
1.83938012e-01 2.23031312e-01 6.87579811e-02 3.80227536e-01
7.77499259e-01 7.72517622e-01 3.86427820e-01 -1.35301426e-01
2.22064570e-01 7.42849112e-01 7.61393607e-02 -7.18450487e-01
7.89959490e-01 -6.35072440e-02 -1.96659207e-01 -2.41107076e-01
-4.77836460e-01 -1.17576487e-01 -4.49144870e-01 2.16973856e-01
5.75293660e-01 -7.28031456e-01 -5.33642948e-01 3.09537292e-01
-1.23986292e+00 -1.02168512e+00 -2.29068711e-01 2.20210418e-01
-1.28371286e+00 1.96976755e-02 -5.28280616e-01 -7.15716183e-01
-4.41504031e-01 -1.44583344e+00 1.34488773e+00 1.30092073e-02
-4.24838424e-01 -8.34017098e-01 4.81686950e-01 4.06945825e-01
4.97919321e-01 -2.28675067e-01 1.09527111e+00 -7.50401795e-01
-1.17136312e+00 3.71897891e-02 3.27260256e-01 2.16314152e-01
3.66995871e-01 -1.87445045e-01 -6.66179359e-02 -3.78278375e-01
-1.23350471e-01 -6.57732368e-01 2.03596771e-01 5.20108305e-02
4.52514797e-01 -7.96526611e-01 -2.90157646e-01 8.26629400e-02
1.44183350e+00 5.13567626e-01 5.32437861e-01 6.04446888e-01
3.43045801e-01 1.00627184e-01 8.05710196e-01 2.84814358e-01
7.74298087e-02 9.70158517e-01 6.69337735e-02 2.47869670e-01
3.41592044e-01 -3.49347144e-01 8.71909082e-01 6.85778558e-01
-1.46658523e-02 -3.92123282e-01 -8.19905400e-01 5.91903627e-01
-2.12119532e+00 -1.19286692e+00 5.12163281e-01 2.35383821e+00
1.41917753e+00 1.82218716e-01 2.80683249e-01 -1.15640141e-01
5.48993230e-01 -2.28038132e-01 -3.23317051e-01 -8.72972488e-01
3.29314500e-01 5.23428321e-01 6.59925938e-01 9.67132568e-01
-4.25662100e-01 1.30464387e+00 6.98900509e+00 5.81271946e-01
-1.09058475e+00 6.75613061e-02 5.71857274e-01 -2.66273439e-01
-5.88848114e-01 3.01340342e-01 -8.36117804e-01 4.54306841e-01
1.42526126e+00 -2.46547699e-01 1.00723290e+00 5.67999542e-01
7.47788548e-01 -1.59015089e-01 -1.48688436e+00 3.62677515e-01
-5.47287241e-02 -1.70028543e+00 4.90430705e-02 7.60241151e-02
1.03161180e+00 -3.41935456e-02 -3.37355763e-01 9.51846987e-02
8.44771504e-01 -9.15961683e-01 7.06983030e-01 8.57421905e-02
5.88638484e-01 -8.97726536e-01 7.09886253e-01 9.22919393e-01
-5.31594038e-01 7.45976567e-02 -3.72898728e-01 -2.49457657e-01
1.46736354e-01 2.22346753e-01 -1.62343955e+00 2.23711178e-01
-2.41805717e-01 1.15344435e-01 -1.60618857e-01 5.75271070e-01
-5.07421196e-01 5.67749321e-01 -2.73948431e-01 -6.63826108e-01
3.30257475e-01 -2.92349875e-01 3.25351059e-01 9.93376434e-01
4.03669804e-01 -2.08413810e-01 4.91550118e-01 7.05956757e-01
-1.94004536e-01 1.44636124e-01 -6.71367168e-01 -2.99363524e-01
6.16241753e-01 7.94840157e-01 -8.16595912e-01 -6.00105822e-01
6.09157532e-02 1.44324136e+00 5.77994406e-01 1.44507408e-01
-8.80701721e-01 -1.06377468e-01 3.37880909e-01 -1.70797823e-04
2.69381374e-01 -4.09281462e-01 -2.91192293e-01 -1.07034612e+00
2.36372441e-01 -1.49067950e+00 -1.73396304e-01 -4.32915419e-01
-7.22507775e-01 5.39585590e-01 -4.91119549e-02 -9.65889513e-01
-1.01464510e+00 -6.33531213e-02 -6.20338142e-01 8.54162097e-01
-1.25132966e+00 -7.71576047e-01 3.50960225e-01 2.30651036e-01
9.16726589e-01 -3.30481380e-01 9.85209703e-01 -3.42346549e-01
-4.89849180e-01 5.24004698e-01 3.42434943e-01 -2.05232963e-01
6.30256355e-01 -1.46579802e+00 6.21434331e-01 8.69498491e-01
3.43211681e-01 6.48520589e-01 8.02091897e-01 -7.17540205e-01
-1.82128549e+00 -8.91679943e-01 1.39070559e+00 -4.86771852e-01
5.04162312e-01 -2.88242787e-01 -1.85544699e-01 9.86609697e-01
7.37656176e-01 -7.01997221e-01 3.80468816e-01 -2.71134645e-01
-1.02838434e-01 -8.22314844e-02 -8.82196724e-01 7.65415370e-01
8.67761493e-01 -4.66347188e-01 -3.66915047e-01 9.33369935e-01
6.61921620e-01 -5.31813145e-01 -5.74625432e-01 -2.77523279e-01
2.05054417e-01 -6.24196351e-01 4.37625349e-01 -5.86735785e-01
6.90514028e-01 -3.21021765e-01 1.50515944e-01 -1.63319600e+00
-1.72245651e-02 -1.20909822e+00 1.20456800e-01 8.14849854e-01
8.44262958e-01 -5.23907185e-01 8.68745685e-01 8.66234228e-02
1.16280921e-01 -9.13509548e-01 -8.16207170e-01 -1.04573655e+00
9.53970999e-02 2.92532295e-01 4.77098227e-01 5.40862620e-01
1.40892550e-01 6.68998718e-01 -5.82945585e-01 -5.09586036e-02
4.97515172e-01 3.27977419e-01 1.08153474e+00 -4.58927959e-01
-8.22552621e-01 -1.62126616e-01 1.35050444e-02 -1.23544276e+00
3.49640250e-01 -1.03472281e+00 6.61779881e-01 -1.61713433e+00
5.60787439e-01 -3.22671950e-01 2.16123551e-01 6.52296722e-01
-3.10760558e-01 -7.37366453e-02 1.04481652e-01 3.74890059e-01
-4.83784080e-01 2.60890007e-01 1.27958047e+00 -1.33180425e-01
-3.39060724e-01 2.41631821e-01 -6.36437535e-01 -4.06217761e-02
8.64253879e-01 -8.50911260e-01 -5.47010005e-01 -6.52726352e-01
4.32457238e-01 5.09270847e-01 -1.53551623e-01 -4.78999496e-01
1.38946891e-01 -8.07024479e-01 -1.13524869e-01 -1.31064788e-01
-1.27729535e-01 -4.02048200e-01 1.80781543e-01 8.29784513e-01
-6.74687326e-01 7.27572739e-01 -2.18673572e-01 3.91866326e-01
1.04203336e-01 -4.18454409e-01 6.90762103e-01 -4.72378969e-01
-1.43654868e-01 -6.78990260e-02 -8.99883270e-01 -8.66243988e-02
1.14396560e+00 5.34856990e-02 -1.26565486e-01 -5.45349479e-01
-4.54147607e-01 3.02670330e-01 7.61796892e-01 3.81929219e-01
4.20051664e-01 -1.18546772e+00 -8.43211830e-01 3.27362046e-02
-2.08342671e-01 -2.12656066e-01 -7.46676862e-01 6.31539762e-01
-7.21646786e-01 3.73514801e-01 -2.41702214e-01 -5.58168530e-01
-1.29268754e+00 3.79332751e-01 3.62781614e-01 -8.18723977e-01
-2.89681911e-01 6.25597477e-01 -5.72247148e-01 -3.65856349e-01
-1.27536252e-01 -2.87657063e-02 6.36086404e-01 -3.97682279e-01
3.92031431e-01 1.82812423e-01 1.38685703e-01 -1.10487871e-01
-2.05841511e-01 1.13521852e-01 -2.77834445e-01 -7.34651923e-01
1.32142425e+00 9.37367789e-03 -2.85207331e-01 2.04838246e-01
1.01262319e+00 1.53424531e-01 -1.15526342e+00 -4.17908467e-02
1.97211266e-01 -5.41495442e-01 -5.11824310e-01 -8.93563867e-01
-5.13304591e-01 4.18940902e-01 1.44891575e-01 4.67050821e-02
7.32172728e-01 -9.94883180e-02 7.93129861e-01 7.88493276e-01
7.77727604e-01 -1.27999353e+00 5.88017702e-01 4.14481491e-01
4.41447407e-01 -1.09418929e+00 -1.25185788e-01 -2.39043146e-01
-7.17711151e-01 1.08607090e+00 6.73797190e-01 -6.51585460e-02
-4.33253556e-01 6.75864398e-01 3.67092073e-01 1.38314456e-01
-1.19990087e+00 1.20579496e-01 -4.93464053e-01 4.42588180e-01
3.80586028e-01 3.08469146e-01 -6.88039303e-01 -5.52629590e-01
-2.75421590e-01 -9.94073674e-02 9.24396873e-01 1.25566459e+00
-6.46265686e-01 -1.99759769e+00 -2.28803426e-01 2.46250287e-01
-3.99794400e-01 -1.53012410e-01 -7.64738083e-01 3.13310415e-01
-1.65361091e-01 8.17255139e-01 5.41662164e-02 -1.12516269e-01
7.56706949e-03 3.95474195e-01 8.66207659e-01 -9.82415497e-01
-7.16039360e-01 7.68025592e-02 3.73461783e-01 -4.56805736e-01
-2.81597614e-01 -7.47383595e-01 -1.42436182e+00 -3.06177944e-01
-5.49366713e-01 4.70384955e-01 5.38151801e-01 1.27454579e+00
3.24461758e-01 1.74228832e-01 1.07611549e+00 -6.12514198e-01
-1.16123545e+00 -7.06607699e-01 1.20401323e-01 3.02577436e-01
1.44905195e-01 -8.14889893e-02 -4.24008965e-02 1.08725674e-01]
|
[11.807792663574219, 9.198775291442871]
|
fb84bc10-d6d3-4641-b675-53fc62878816
|
efficient-hybrid-transformer-learning-global
|
2109.08937
| null |
https://arxiv.org/abs/2109.08937v4
|
https://arxiv.org/pdf/2109.08937v4.pdf
|
UNetFormer: A UNet-like Transformer for Efficient Semantic Segmentation of Remote Sensing Urban Scene Imagery
|
Semantic segmentation of remotely sensed urban scene images is required in a wide range of practical applications, such as land cover mapping, urban change detection, environmental protection, and economic assessment.Driven by rapid developments in deep learning technologies, the convolutional neural network (CNN) has dominated semantic segmentation for many years. CNN adopts hierarchical feature representation, demonstrating strong capabilities for local information extraction. However, the local property of the convolution layer limits the network from capturing the global context. Recently, as a hot topic in the domain of computer vision, Transformer has demonstrated its great potential in global information modelling, boosting many vision-related tasks such as image classification, object detection, and particularly semantic segmentation. In this paper, we propose a Transformer-based decoder and construct a UNet-like Transformer (UNetFormer) for real-time urban scene segmentation. For efficient segmentation, the UNetFormer selects the lightweight ResNet18 as the encoder and develops an efficient global-local attention mechanism to model both global and local information in the decoder. Extensive experiments reveal that our method not only runs faster but also produces higher accuracy compared with state-of-the-art lightweight models. Specifically, the proposed UNetFormer achieved 67.8% and 52.4% mIoU on the UAVid and LoveDA datasets, respectively, while the inference speed can achieve up to 322.4 FPS with a 512x512 input on a single NVIDIA GTX 3090 GPU. In further exploration, the proposed Transformer-based decoder combined with a Swin Transformer encoder also achieves the state-of-the-art result (91.3% F1 and 84.1% mIoU) on the Vaihingen dataset. The source code will be freely available at https://github.com/WangLibo1995/GeoSeg.
|
['Peter M. Atkinson', 'Xiaoliang Meng', 'Shenghui Fang', 'Rui Li', 'Chenxi Duan', 'Ce Zhang', 'Libo Wang']
|
2021-09-18
| null | null | null | null |
['scene-segmentation']
|
['computer-vision']
|
[ 1.19125254e-01 -3.24642479e-01 -1.98164582e-01 -4.03061897e-01
-4.42791998e-01 -1.90430120e-01 2.53830522e-01 -1.90384895e-01
-5.88418722e-01 4.34842557e-01 -1.71792373e-01 -5.85148036e-01
-5.29966026e-04 -1.27911973e+00 -6.43605888e-01 -7.10755885e-01
1.70477301e-01 1.55790865e-01 4.13308859e-01 -1.26024321e-01
-7.16299713e-02 3.76520872e-01 -1.69374728e+00 -6.04619421e-02
1.23196745e+00 1.30153275e+00 6.95295095e-01 5.42364120e-01
-3.56132053e-02 6.58245504e-01 -3.70599896e-01 -1.01671875e-01
3.22677642e-01 -5.76899983e-02 -7.48817921e-01 -1.04814060e-01
2.20001996e-01 -5.98362684e-01 -3.18771005e-01 1.08743536e+00
4.82735008e-01 -5.10716401e-02 1.66233703e-01 -1.07278347e+00
-3.42006654e-01 4.17484641e-01 -7.27397799e-01 3.39450568e-01
-4.00436401e-01 2.60404527e-01 8.12509537e-01 -5.68483353e-01
3.12279344e-01 1.13958406e+00 5.88024437e-01 5.56589849e-02
-6.99688017e-01 -9.93633866e-01 2.34406263e-01 2.91656315e-01
-1.60560691e+00 -1.14942223e-01 3.44411880e-01 -2.21035048e-01
9.21934485e-01 3.50576907e-01 8.56792629e-01 4.92790759e-01
2.37288609e-01 1.01132715e+00 9.31563318e-01 1.30207986e-01
5.79134710e-02 -3.53726268e-01 4.31919210e-02 7.44969010e-01
2.01702267e-01 -3.12355580e-04 6.79548755e-02 3.06843013e-01
9.83703017e-01 2.69862831e-01 -1.37259886e-01 3.90258372e-01
-1.06068981e+00 8.34979296e-01 1.10766745e+00 2.19138026e-01
-4.41457838e-01 4.77440387e-01 4.51026350e-01 -2.95197740e-02
6.77261412e-01 -2.54797805e-02 -5.64162970e-01 -6.22997172e-02
-8.61360431e-01 1.58002764e-01 2.82193244e-01 8.59636188e-01
8.84392023e-01 1.95192724e-01 7.14324089e-03 8.55156541e-01
3.96922976e-01 8.90655518e-01 3.93260628e-01 -7.60254562e-01
4.46897566e-01 7.68274307e-01 -2.34421045e-01 -9.59761262e-01
-4.82773811e-01 -7.76062012e-01 -1.16050708e+00 -5.72010837e-02
-6.10841177e-02 -4.07243371e-02 -1.17911255e+00 1.49846029e+00
4.15332675e-01 2.88316756e-01 -4.76626083e-02 1.01202536e+00
9.49110806e-01 1.03670907e+00 3.94694269e-01 3.35944921e-01
1.46450448e+00 -1.01940656e+00 -2.95836538e-01 -5.45254350e-01
5.14214575e-01 -6.22172475e-01 8.62864435e-01 3.29998732e-02
-5.59747517e-01 -8.29562783e-01 -1.00603294e+00 -2.72934049e-01
-3.76334161e-01 3.84836376e-01 7.80373693e-01 3.96352142e-01
-9.38459158e-01 2.92750627e-01 -1.11640370e+00 -6.10452950e-01
7.96055198e-01 3.70442867e-01 1.33140963e-02 -1.67256802e-01
-1.16620493e+00 5.73432982e-01 5.59186220e-01 3.94782513e-01
-8.39022398e-01 -5.04189610e-01 -7.95111060e-01 1.14085853e-01
2.07051307e-01 -6.06701910e-01 1.14497650e+00 -9.23704565e-01
-1.27676260e+00 6.79777086e-01 -9.70573351e-02 -5.39824545e-01
3.97768050e-01 -3.50956798e-01 -2.74176449e-01 4.80240323e-02
4.01963115e-01 1.09638011e+00 5.06480396e-01 -7.44979799e-01
-9.97989357e-01 -3.72692406e-01 1.20560482e-01 2.78837860e-01
-2.19326556e-01 -7.93970972e-02 -5.36321521e-01 -5.93075812e-01
2.08831951e-01 -8.24029326e-01 -4.39911515e-01 1.02439202e-01
-2.80003816e-01 -1.80719972e-01 1.21178758e+00 -6.86942160e-01
1.04989672e+00 -2.18570495e+00 -3.31145883e-01 5.86707965e-02
1.30553991e-01 5.62951267e-01 -3.02727092e-02 4.58892100e-02
1.49108872e-01 2.35959828e-01 -5.60103655e-01 -1.00960836e-01
-1.90856159e-01 2.82144159e-01 -1.68344751e-01 4.22719657e-01
2.22899579e-02 1.21530402e+00 -7.76442111e-01 -4.34809208e-01
5.40765584e-01 5.36795020e-01 -3.54788095e-01 2.84032263e-02
-8.56183246e-02 4.42336142e-01 -7.22872317e-01 8.25582385e-01
9.81042862e-01 -2.95676827e-01 -1.60431817e-01 -2.03724399e-01
-4.39109445e-01 1.22338338e-02 -8.43126535e-01 1.54989028e+00
-4.98268783e-01 8.04826498e-01 6.41661286e-02 -1.06778085e+00
9.12613750e-01 2.78116148e-02 3.38287711e-01 -1.07567108e+00
4.82053816e-01 4.00198132e-01 -1.28448069e-01 -4.60167825e-01
5.76805413e-01 1.00735262e-01 -1.30080402e-01 2.20113480e-03
-2.00787291e-01 4.11593542e-02 4.84196953e-02 -1.16304293e-01
8.44715476e-01 1.53253213e-01 2.59249896e-01 -4.56619918e-01
4.08655971e-01 2.51644045e-01 5.35576999e-01 4.34787124e-01
-2.08175018e-01 3.81482184e-01 -1.12425007e-01 -5.49823463e-01
-8.13031435e-01 -7.61703134e-01 -4.44936842e-01 9.06288624e-01
5.42164862e-01 -1.65552810e-01 -8.91181588e-01 -3.50870818e-01
-1.69083297e-01 3.79821599e-01 -2.42817670e-01 -3.22175659e-02
-5.47109842e-01 -9.87855792e-01 7.81529427e-01 7.22540915e-01
1.66924751e+00 -9.55215931e-01 -1.06836402e+00 2.44858220e-01
-3.34610671e-01 -1.32065046e+00 -1.55771971e-01 -6.26188889e-02
-1.03078926e+00 -8.90075505e-01 -5.24528027e-01 -8.95605266e-01
3.99132401e-01 5.89046419e-01 7.99524784e-01 1.47340909e-01
-3.22681546e-01 -1.98873773e-01 -2.91128963e-01 -1.96286649e-01
1.39990211e-01 5.05560815e-01 -5.06756961e-01 -1.12517416e-01
2.51521468e-01 -4.34300900e-01 -9.06609058e-01 3.84163857e-01
-9.53141749e-01 5.18798947e-01 7.15938270e-01 6.21419609e-01
6.50120437e-01 2.57298797e-01 4.24421906e-01 -6.45934045e-01
2.10594703e-02 -5.32809496e-01 -7.60831118e-01 -7.94436336e-02
-3.40284765e-01 -3.47710401e-01 6.84727788e-01 1.03542104e-01
-1.07234049e+00 -9.87531058e-03 -6.52030587e-01 -1.61975086e-01
-2.39069134e-01 7.25469768e-01 -2.48713583e-01 1.68539099e-02
2.93956310e-01 2.37987682e-01 -1.94255501e-01 -4.05027896e-01
-1.36247454e-02 9.43133175e-01 5.12149215e-01 -2.51845986e-01
6.58299029e-01 5.36940932e-01 -2.15549916e-01 -1.09381783e+00
-7.31325865e-01 -4.65549290e-01 -2.36244202e-01 -1.52121544e-01
1.11638916e+00 -1.41573370e+00 -7.33839273e-01 9.68098164e-01
-9.47857440e-01 -5.96614718e-01 -1.26744192e-02 3.28079402e-01
-1.60372436e-01 8.95329937e-02 -4.95965570e-01 -4.98541296e-01
-7.28803813e-01 -1.38486969e+00 1.22072709e+00 6.08401477e-01
1.80004239e-01 -7.93185174e-01 -5.35629392e-01 5.02230823e-01
7.01765299e-01 2.94283837e-01 6.24316871e-01 -1.43320471e-01
-9.74120498e-01 -7.79455751e-02 -7.59306729e-01 4.48623657e-01
-1.40736653e-02 -9.38507989e-02 -1.05933762e+00 -1.79427773e-01
-1.77177101e-01 3.04018427e-03 1.04091704e+00 5.05037069e-01
1.44252133e+00 -1.55042470e-01 -4.50354576e-01 9.54067767e-01
1.71773827e+00 3.08855295e-01 8.90765607e-01 4.37928647e-01
9.75194275e-01 1.03680819e-01 6.98043764e-01 3.34861785e-01
7.18880117e-01 4.34549809e-01 8.66977572e-01 -3.96069795e-01
-1.32519424e-01 -1.75553411e-01 1.63315147e-01 6.38111949e-01
-1.17523931e-01 -4.17975277e-01 -1.11683583e+00 6.45769656e-01
-1.88268089e+00 -7.63705194e-01 -2.85034627e-01 1.87920046e+00
4.76649791e-01 6.96067140e-03 -3.32791984e-01 2.43164338e-02
6.94027007e-01 4.22448397e-01 -7.11583793e-01 -8.68140981e-02
3.92315760e-02 3.96373540e-01 1.01349688e+00 4.21379387e-01
-1.45884657e+00 1.33949292e+00 4.40086699e+00 1.21514356e+00
-1.42999697e+00 2.28724360e-01 8.38791907e-01 3.53990704e-01
5.37935682e-02 -1.50208712e-01 -7.48460233e-01 5.09436309e-01
8.78625631e-01 1.47665352e-01 1.19556122e-01 9.02617276e-01
4.44765121e-01 -3.46460313e-01 -2.31700152e-01 8.50822806e-01
-4.28076595e-01 -1.21306384e+00 5.30867185e-03 1.53628603e-01
6.05174243e-01 6.34749055e-01 -5.46809053e-03 3.04993033e-01
1.37675539e-01 -1.06859922e+00 7.15251982e-01 1.18880838e-01
1.00857782e+00 -9.28760588e-01 9.81870592e-01 5.06041110e-01
-1.77173948e+00 -9.79267880e-02 -4.03704256e-01 -1.52695045e-01
-2.20889356e-02 6.21629000e-01 -5.17049253e-01 6.13614798e-01
9.89882946e-01 1.10039425e+00 -5.45153081e-01 9.84542429e-01
-2.27378190e-01 8.75849068e-01 -5.18633664e-01 1.54575035e-01
6.31316245e-01 -2.80131072e-01 3.16856414e-01 1.12979639e+00
4.08316731e-01 3.55832160e-01 4.06139731e-01 6.38721168e-01
-9.43498909e-02 -2.35292632e-02 -2.84782320e-01 1.42165154e-01
3.88064146e-01 1.35217130e+00 -1.00224400e+00 -3.54843140e-01
-1.35944307e-01 8.54377329e-01 -7.82600939e-02 3.18937600e-01
-1.21054614e+00 -5.35125017e-01 8.96353602e-01 -7.19469637e-02
3.56995851e-01 -2.73579270e-01 -4.08952951e-01 -1.08596468e+00
-1.63758367e-01 -6.77488089e-01 6.17691465e-02 -7.16214478e-01
-6.22726858e-01 7.09109306e-01 -5.60795888e-02 -1.16450620e+00
2.66443074e-01 -4.73154098e-01 -5.87980509e-01 7.71823823e-01
-1.85730183e+00 -1.34248281e+00 -9.50087667e-01 4.16424483e-01
7.50193417e-01 2.65181214e-01 5.80886245e-01 6.05161846e-01
-8.16854000e-01 2.33863086e-01 1.06009454e-01 3.84109408e-01
1.42898172e-01 -8.09046865e-01 7.58962512e-01 1.09988487e+00
-3.35048914e-01 2.63579965e-01 2.55057544e-01 -5.22769690e-01
-1.15961516e+00 -1.75998831e+00 6.59663975e-01 3.43661040e-01
3.22105169e-01 -1.26272872e-01 -6.71312213e-01 6.54406250e-01
1.01317890e-01 1.82479203e-01 1.32577121e-01 -4.77018416e-01
2.05040816e-02 -4.74758714e-01 -1.16444194e+00 4.47851211e-01
1.08579290e+00 -3.10504735e-01 1.11379810e-01 2.05633387e-01
8.31413507e-01 -6.29980624e-01 -6.75856411e-01 6.41185343e-01
4.53948140e-01 -9.78831410e-01 8.75816107e-01 1.91320807e-01
4.63275373e-01 -5.08340895e-01 -3.36929709e-01 -8.99717927e-01
-3.36687893e-01 -4.93674316e-02 3.21503580e-01 1.00841033e+00
2.73698959e-02 -1.02271628e+00 6.38507903e-01 6.93619922e-02
-4.81907338e-01 -9.45951343e-01 -9.59253550e-01 -5.84629357e-01
-1.66010514e-01 -6.40876353e-01 7.48579443e-01 4.86399919e-01
-7.43712246e-01 2.09440932e-01 3.63996215e-02 3.86745334e-01
5.01456320e-01 2.62735546e-01 6.60661280e-01 -1.17734313e+00
2.44399965e-01 -4.61782187e-01 -6.19385481e-01 -1.48694801e+00
1.73131935e-02 -8.58529031e-01 5.98144755e-02 -1.80464959e+00
1.27869099e-01 -6.26670361e-01 -6.23055175e-02 7.49644041e-01
-1.32501423e-01 5.12351274e-01 2.09532231e-01 1.25942558e-01
-4.60749298e-01 7.35883594e-01 1.25906479e+00 -3.33827347e-01
9.60986540e-02 2.31185593e-02 -6.04435742e-01 7.06865609e-01
1.16814625e+00 -3.17734271e-01 -3.74485523e-01 -8.26618671e-01
-3.64677794e-02 -2.51813620e-01 7.47116566e-01 -1.41247630e+00
1.27506703e-01 -1.70309171e-02 2.83851057e-01 -7.35542297e-01
1.86964303e-01 -8.39446366e-01 2.59021014e-01 7.94807076e-01
2.12884530e-01 -5.30265272e-02 4.09958124e-01 3.04674983e-01
-4.54079568e-01 1.41696170e-01 8.23372781e-01 -1.33820564e-01
-1.22597265e+00 6.73083425e-01 -4.25080538e-01 -2.52348296e-02
1.01947367e+00 -2.66445428e-01 -5.43156207e-01 -8.40407461e-02
-1.62017807e-01 5.12204468e-01 2.81199127e-01 3.75822902e-01
6.25929654e-01 -1.03400111e+00 -7.55397439e-01 2.92741746e-01
-9.59120914e-02 6.09909356e-01 5.10713935e-01 8.13420594e-01
-9.79638517e-01 4.54652846e-01 -1.02879986e-01 -9.74771798e-01
-1.04167581e+00 -2.57323444e-01 4.46677178e-01 -1.99579343e-01
-6.58175826e-01 7.84504533e-01 4.18935478e-01 -3.87790918e-01
-2.06859097e-01 -5.96683979e-01 -1.07613377e-01 -7.60377245e-03
2.18748316e-01 4.63149935e-01 1.41922325e-01 -7.10129797e-01
-3.93775284e-01 7.28779137e-01 1.42277971e-01 2.64464647e-01
1.29324675e+00 -1.39136881e-01 -1.93254426e-01 -4.67883684e-02
1.27351177e+00 -5.10168493e-01 -1.33576763e+00 -1.48659483e-01
-2.59318739e-01 -4.69583780e-01 4.91444260e-01 -7.28528738e-01
-1.52819729e+00 9.57355440e-01 8.17629755e-01 -1.74921229e-02
1.43913615e+00 -2.05001935e-01 1.33057249e+00 1.39448687e-01
5.66302061e-01 -7.82893360e-01 -4.26618814e-01 8.28943729e-01
5.20084679e-01 -1.38030732e+00 -4.59738355e-03 -4.71294284e-01
-4.52208817e-01 9.11799431e-01 6.84111059e-01 -1.61810756e-01
6.58175409e-01 2.53210962e-01 9.12361741e-02 -2.18946561e-01
-2.78878421e-01 -5.45638144e-01 3.62057462e-02 2.51127630e-01
2.48863056e-01 4.82389063e-01 1.79031044e-02 3.61571848e-01
-2.55264223e-01 5.06898351e-02 1.03777342e-01 8.96537423e-01
-6.01524413e-01 -6.20023906e-01 -1.41122282e-01 5.83786786e-01
-2.55442888e-01 -2.93926597e-01 3.14860314e-01 8.19196939e-01
4.85832572e-01 9.17232394e-01 2.64466703e-01 -4.65041816e-01
1.44415945e-01 -4.13961709e-01 -1.03690103e-02 -3.87947321e-01
-6.50882661e-01 6.10126182e-02 -1.20842338e-01 -6.04992628e-01
-5.69703400e-01 -4.61312652e-01 -1.49743772e+00 -5.17465055e-01
-3.24295610e-01 -1.16728298e-01 7.16643393e-01 1.05326641e+00
4.38753426e-01 7.22672164e-01 5.02286434e-01 -8.66167307e-01
2.30404101e-02 -8.99959445e-01 -3.07049334e-01 -2.47425824e-01
2.70478308e-01 -5.25702059e-01 4.11273651e-02 -1.53814062e-01]
|
[9.244217872619629, -0.7299501895904541]
|
adb3625e-5cb6-4106-ab74-675707f81b9a
|
hand-gesture-recognition-using-802-11ad
|
2211.0709
| null |
https://arxiv.org/abs/2211.07090v1
|
https://arxiv.org/pdf/2211.07090v1.pdf
|
Hand gesture recognition using 802.11ad mmWave sensor in the mobile device
|
We explore the feasibility of AI assisted hand-gesture recognition using 802.11ad 60GHz (mmWave) technology in smartphones. Range-Doppler information (RDI) is obtained by using pulse Doppler radar for gesture recognition. We built a prototype system, where radar sensing and WLAN communication waveform can coexist by time-division duplex (TDD), to demonstrate the real-time hand-gesture inference. It can gather sensing data and predict gestures within 100 milliseconds. First, we build the pipeline for the real-time feature processing, which is robust to occasional frame drops in the data stream. RDI sequence restoration is implemented to handle the frame dropping in the continuous data stream, and also applied to data augmentation. Second, different gestures RDI are analyzed, where finger and hand motions can clearly show distinctive features. Third, five typical gestures (swipe, palm-holding, pull-push, finger-sliding and noise) are experimented with, and a classification framework is explored to segment the different gestures in the continuous gesture sequence with arbitrary inputs. We evaluate our architecture on a large multi-person dataset and report > 95% accuracy with one CNN + LSTM model. Further, a pure CNN model is developed to fit to on-device implementation, which minimizes the inference latency, power consumption and computation cost. And the accuracy of this CNN model is more than 93% with only 2.29K parameters.
|
['Hao Xu', 'Chirag Patel', 'Daniel Fontijne', 'Ilia Karmanov', 'Yin Huang', 'Andrian Beletchi', 'Jiuyuan Lu', 'Yuwei Ren']
|
2022-11-14
| null | null | null | null |
['hand-gesture-recognition', 'hand-gesture-recognition-1', 'gesture-recognition']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[ 4.65193689e-01 -4.30576950e-01 -1.03475727e-01 -5.08186758e-01
-3.87740880e-01 -2.60511935e-01 4.77385044e-01 -7.55558550e-01
-6.44439459e-01 4.61580694e-01 6.90197051e-02 -3.96588296e-01
-9.87560451e-02 -7.34171987e-01 -1.06606930e-01 -6.57324135e-01
-3.67447227e-01 1.39603436e-01 1.81205738e-02 7.82348365e-02
-1.87671497e-01 7.74591982e-01 -1.52943230e+00 2.58354247e-01
2.96843320e-01 1.31512499e+00 -1.17226362e-01 1.27905834e+00
-8.46729800e-02 3.15033704e-01 -1.20791864e+00 -2.91382875e-02
3.11007112e-01 -4.29275446e-02 5.27552105e-02 -6.38406396e-01
3.70446563e-01 -1.28729796e+00 -7.92649508e-01 4.71697062e-01
1.21493983e+00 -1.22955829e-01 3.52451384e-01 -1.50337362e+00
-1.34306133e-01 3.96517605e-01 -4.37013477e-01 1.27334535e-01
6.23019338e-01 2.98470169e-01 1.35711908e-01 -5.47681987e-01
2.67481446e-01 1.32762504e+00 6.70009732e-01 7.67621994e-01
-4.22820628e-01 -1.28641152e+00 -2.10065678e-01 2.31295899e-01
-1.43410516e+00 -3.77029687e-01 5.63771546e-01 -1.00167386e-01
1.12136102e+00 5.76087356e-01 7.73724794e-01 1.63666785e+00
1.01900570e-01 7.24794626e-01 4.31297570e-01 -3.19148958e-01
6.97927475e-02 -7.06204832e-01 3.90691459e-01 5.66835523e-01
2.50893414e-01 5.95371485e-01 -8.57897222e-01 -1.23603372e-02
9.70497370e-01 6.61305547e-01 -1.21260814e-01 7.22998142e-01
-1.42427492e+00 1.23453461e-01 1.43397734e-01 2.62180924e-01
-4.55177575e-01 5.88869691e-01 1.84190243e-01 3.92685115e-01
-5.61987311e-02 -3.52693260e-01 -3.49363089e-01 -8.60325396e-01
-1.01574576e+00 2.30717123e-01 8.39263082e-01 1.26579177e+00
1.26636207e-01 3.32945466e-01 -3.60593200e-01 2.55441993e-01
6.20275080e-01 1.49683058e+00 3.22413087e-01 -6.49255991e-01
6.52867496e-01 1.02763616e-01 1.54108539e-01 -8.45057786e-01
-7.94659913e-01 -2.54249461e-02 -1.37573195e+00 1.41948536e-01
5.27866364e-01 -7.33827353e-01 -1.29454315e+00 1.38218856e+00
1.23334989e-01 5.47519445e-01 -1.96964979e-01 1.04872525e+00
1.10180783e+00 1.99894965e-01 2.22823583e-02 -2.89975226e-01
1.47347128e+00 -3.81530911e-01 -9.50647354e-01 1.28683195e-01
5.43112122e-02 -5.55985332e-01 7.78860927e-01 5.60061514e-01
-4.83274728e-01 -8.52289498e-01 -1.15674865e+00 2.68864512e-01
-9.86627042e-02 2.51397073e-01 7.73772299e-01 1.32158291e+00
-5.55380940e-01 2.74431825e-01 -1.11565554e+00 -3.30240130e-01
4.06516761e-01 6.13988221e-01 1.66268170e-01 -4.01909143e-04
-1.37186110e+00 1.43351346e-01 6.02302141e-03 6.36708915e-01
-3.20247442e-01 -6.16601944e-01 -5.11865139e-01 -7.74465650e-02
-2.40721166e-01 -2.97508568e-01 1.22367048e+00 -2.77347207e-01
-1.96656775e+00 1.56712830e-01 -3.88883471e-01 -2.58493662e-01
5.66983461e-01 -3.79855752e-01 -1.08391798e+00 1.12576999e-01
-5.97410738e-01 4.15139675e-01 1.22601533e+00 -5.64885855e-01
-8.01474690e-01 -4.41109061e-01 -2.21301496e-01 -1.65867031e-01
-2.93047547e-01 1.57218859e-01 -5.01710236e-01 -7.21023500e-01
5.48793674e-01 -9.13294673e-01 1.01817526e-01 -1.67329967e-01
-5.05489111e-01 1.52917467e-02 9.92027581e-01 -6.85363889e-01
1.34483445e+00 -2.00558496e+00 -7.33233333e-01 5.66079378e-01
8.66752714e-02 4.55220163e-01 -1.01938313e-02 9.73087400e-02
4.06340271e-01 -1.59435511e-01 9.39181745e-02 -2.19348907e-01
2.22714052e-01 2.41812542e-01 -5.81839681e-01 1.51066005e-01
-1.28270447e-01 1.09566092e+00 -4.73763734e-01 -2.10432574e-01
3.13092500e-01 6.34322405e-01 -2.07994670e-01 2.12424070e-01
2.98057824e-01 5.51586568e-01 -6.00797117e-01 1.18199790e+00
9.45208013e-01 1.05342560e-01 1.87205613e-01 -4.74449992e-01
-6.99221417e-02 2.61453800e-02 -1.44873333e+00 1.50291169e+00
-2.56606728e-01 9.25707102e-01 4.22185920e-02 -7.19845116e-01
1.06037056e+00 3.32437366e-01 5.29750109e-01 -1.06448317e+00
4.22269404e-01 4.73000705e-02 7.14942142e-02 -1.06074667e+00
2.14986399e-01 2.63789773e-01 -1.66283339e-01 8.88394117e-01
-3.14304084e-01 4.88416821e-01 -3.24114621e-01 -2.78349578e-01
1.36938095e+00 5.31217605e-02 -1.37129799e-01 4.57447171e-01
1.22389711e-01 -3.60276282e-01 3.08784842e-01 1.28113222e+00
-2.85430640e-01 3.33046168e-01 -2.84336030e-01 -4.40122008e-01
-3.10330749e-01 -1.13222563e+00 5.95264845e-02 1.21773338e+00
2.47387171e-01 8.60864222e-02 -4.41650987e-01 -3.06565583e-01
3.84432115e-02 1.75022781e-01 1.36595024e-02 2.32714489e-01
-1.14076340e+00 -6.72168493e-01 1.44556642e+00 7.64776647e-01
1.02810490e+00 -1.22867537e+00 -1.37667167e+00 4.36454535e-01
-9.10282955e-02 -1.14105821e+00 -1.87012017e-01 -1.88331194e-02
-8.18099439e-01 -9.32634592e-01 -8.00127685e-01 -5.96124291e-01
9.16516110e-02 2.97752053e-01 6.12683058e-01 1.56840801e-01
-6.84827209e-01 6.32136941e-01 -3.95631075e-01 -7.11445272e-01
3.64189953e-01 -2.70101558e-02 3.90726924e-01 -1.41650751e-01
9.22227204e-01 -6.82999492e-01 -6.09671891e-01 1.79310650e-01
-3.98144364e-01 -3.00897777e-01 7.57386446e-01 4.14838612e-01
-8.48767639e-04 -2.49073669e-01 3.58948797e-01 -3.81759629e-02
6.51765823e-01 -5.46441562e-02 -7.46624321e-02 1.17339164e-01
-2.81081408e-01 -2.81199068e-01 2.29306534e-01 -9.58893061e-01
-1.08545959e+00 1.34946570e-01 -3.63795072e-01 -3.49788129e-01
-6.23002827e-01 2.63780445e-01 -2.91326880e-01 3.98293324e-02
4.69808906e-01 3.96683723e-01 1.28134549e-01 -4.39158529e-01
3.11596870e-01 1.43731225e+00 8.74241173e-01 -2.84316272e-01
8.09492886e-01 6.49921715e-01 -1.73306525e-01 -1.41697061e+00
2.73760930e-02 -2.03212261e-01 -6.59727752e-01 -3.70126128e-01
5.18429637e-01 -7.95361638e-01 -1.58678007e+00 9.71019208e-01
-1.31702089e+00 -4.56949085e-01 2.79794395e-01 1.00078249e+00
-6.34665117e-02 3.63905281e-01 -6.40567899e-01 -1.37657440e+00
-7.70618558e-01 -6.05663717e-01 1.10598767e+00 5.70213854e-01
-3.37793022e-01 -2.78743446e-01 -3.73557359e-01 -2.83036195e-02
6.51455164e-01 2.51018405e-01 3.80705476e-01 -5.35743177e-01
-7.40324736e-01 -3.98361623e-01 -2.71536350e-01 -3.03095341e-01
1.61850333e-01 -9.24461484e-02 -1.11437964e+00 -2.37523869e-01
-1.23057999e-01 9.54405889e-02 7.08201468e-01 5.53911030e-01
1.03595817e+00 -2.20239192e-01 -7.40791798e-01 9.31136250e-01
7.32679427e-01 7.26346314e-01 8.60222578e-01 -1.23826386e-02
7.55313218e-01 7.92205334e-02 7.42845058e-01 6.66987538e-01
2.12352604e-01 6.88559115e-01 1.85264692e-01 -5.98049164e-02
-1.53112039e-01 1.32180259e-01 4.22491968e-01 4.33435202e-01
-4.53700691e-01 -4.96496826e-01 -7.64639199e-01 -1.03850044e-01
-1.60911095e+00 -1.21249986e+00 -3.92217904e-01 2.01584053e+00
3.88776660e-01 7.19510317e-02 2.01567069e-01 4.70546126e-01
7.30012774e-01 1.17500968e-01 -8.25939298e-01 -6.29257634e-02
1.52423665e-01 8.15293729e-01 5.60621977e-01 3.89256388e-01
-1.17509961e+00 6.86540067e-01 6.23440647e+00 4.95391369e-01
-1.64038956e+00 -2.20693931e-01 -1.03119612e-01 -4.54955727e-01
2.78024346e-01 -8.86578977e-01 -8.60363781e-01 6.38661742e-01
8.82120609e-01 5.03179431e-01 4.79734004e-01 4.74640101e-01
2.70103931e-01 1.34060442e-01 -1.00792360e+00 1.66881132e+00
-5.98829202e-02 -1.01846373e+00 -3.36762518e-01 1.64495528e-01
-2.32450962e-01 -4.96724993e-02 -1.13509096e-01 4.43357587e-01
-1.15470804e-01 -1.13335097e+00 4.12898928e-01 8.03373098e-01
1.52850580e+00 -6.46958470e-01 6.63196862e-01 3.30040246e-01
-1.51903307e+00 -2.68262595e-01 -6.31818548e-02 -4.27186698e-01
3.32841069e-01 5.43936610e-01 -7.63811707e-01 3.31872255e-01
8.70307624e-01 3.40138704e-01 -2.07621790e-02 5.13029397e-01
-2.42648404e-02 7.42690742e-01 -8.00263643e-01 -5.33487022e-01
-3.60619485e-01 6.56027794e-02 4.13413137e-01 1.44993496e+00
6.90526426e-01 8.04984093e-01 -8.72716010e-02 3.51540357e-01
4.42390978e-01 -6.59605026e-01 -2.03233629e-01 6.99510723e-02
1.03727078e+00 9.32997167e-01 -5.04632831e-01 -2.88128942e-01
-1.13391079e-01 1.00185299e+00 -7.45132923e-01 6.40115559e-01
-1.05786026e+00 -1.04999709e+00 1.05644548e+00 -3.48126113e-01
2.18395680e-01 -7.39344299e-01 -3.50449622e-01 -8.85355115e-01
2.00821400e-01 -5.40583789e-01 1.89662457e-01 -3.69317204e-01
-9.53655779e-01 3.85320544e-01 -3.72059464e-01 -1.35284090e+00
-6.90464437e-01 -6.05525255e-01 -7.07512140e-01 7.78931379e-01
-1.22639894e+00 -1.05769706e+00 -1.12172675e+00 8.41700673e-01
2.71828324e-01 -3.40030551e-01 1.01732969e+00 7.76443899e-01
-5.12846351e-01 1.15004265e+00 -1.38338119e-01 6.58416867e-01
5.62481999e-01 -7.19321966e-01 8.47510457e-01 7.69598961e-01
2.68509723e-02 8.38866949e-01 3.53880733e-01 -8.54759812e-01
-1.89665604e+00 -9.40518141e-01 6.79280579e-01 -2.32875615e-01
1.24342203e-01 -4.30434376e-01 -3.97538185e-01 4.61868137e-01
-4.82261449e-01 -7.72439539e-02 7.22489655e-01 -1.01297408e-01
-8.88716429e-02 -2.29928941e-01 -1.19103611e+00 4.81962085e-01
1.62176406e+00 -4.45343137e-01 -3.68782252e-01 -1.52218476e-01
5.66572070e-01 -6.42535627e-01 -6.43249631e-01 2.25781620e-01
1.51214385e+00 -8.69551659e-01 1.31873739e+00 -4.42490488e-01
-2.90798843e-01 -1.33800238e-01 -2.75308609e-01 -4.78348792e-01
-9.27943885e-02 -9.31084931e-01 -8.48850369e-01 1.12755406e+00
-1.11098446e-01 -7.14844465e-01 1.10585725e+00 6.21812046e-01
1.90093026e-01 -3.73378873e-01 -1.18412960e+00 -7.58344412e-01
-8.12086821e-01 -1.12497473e+00 1.13180244e+00 4.42637980e-01
1.55728066e-03 -4.06663828e-02 -5.62980652e-01 3.40612173e-01
7.36745119e-01 -7.25026652e-02 1.06195676e+00 -1.45130301e+00
-9.47910920e-02 -2.16318443e-01 -3.90673012e-01 -1.65837336e+00
-4.68689591e-01 -3.52984548e-01 8.98149088e-02 -1.25133383e+00
-6.75225496e-01 -5.53484023e-01 -7.54425377e-02 5.24457693e-01
2.83175349e-01 3.94457132e-01 2.53807634e-01 2.42886245e-01
-3.53198908e-02 1.81486577e-01 7.92545140e-01 -3.33474785e-01
-7.24730492e-01 3.76560628e-01 -1.61885396e-01 6.13641083e-01
7.63640523e-01 -7.09600374e-02 -2.04692006e-01 -3.97131056e-01
-2.06801534e-01 3.54000628e-01 6.72954738e-01 -1.35484660e+00
5.01435935e-01 -1.87785536e-01 8.67106795e-01 -9.06305850e-01
5.67640424e-01 -8.88163984e-01 8.28504264e-02 9.38070297e-01
5.84697574e-02 -3.01434040e-01 -7.18093244e-03 4.27243561e-01
4.41140771e-01 3.83761346e-01 2.03537390e-01 4.82111603e-01
-7.15886354e-01 4.84986663e-01 -4.86791492e-01 -3.78911614e-01
4.95023996e-01 -5.91991484e-01 -4.65702415e-01 -5.71818769e-01
-5.67947388e-01 -1.62102774e-01 -5.49997628e-01 5.84545553e-01
8.76776397e-01 -1.47146440e+00 -5.04409134e-01 7.63124466e-01
-3.26326698e-01 -1.96376458e-01 3.13749492e-01 6.44693673e-01
-4.70512062e-01 6.05224788e-01 -2.61789352e-01 -8.45068812e-01
-1.48070884e+00 -1.29148990e-01 2.18502909e-01 2.49626696e-01
-9.52362895e-01 8.50117922e-01 -8.24604332e-01 -6.96040019e-02
8.26078117e-01 -6.77629888e-01 -1.81682169e-01 1.47845209e-01
1.13485277e+00 7.23902702e-01 -7.69510865e-03 -1.40005827e-01
-7.76888669e-01 7.53591061e-01 4.09660786e-01 -4.37451333e-01
1.04670906e+00 -2.90062092e-02 4.53634053e-01 6.42001927e-02
9.07212019e-01 -1.97787732e-01 -1.24107444e+00 -1.05705798e-01
-2.81545997e-01 -3.73679280e-01 -1.99254811e-01 -9.03241456e-01
-9.98720407e-01 9.77465332e-01 1.33545792e+00 1.35094807e-01
1.11076307e+00 -6.07385278e-01 1.33980775e+00 9.31094885e-01
5.55032969e-01 -8.20882559e-01 -1.38346210e-01 7.59795427e-01
5.21542966e-01 -8.02918673e-01 -6.43861666e-02 -2.95373201e-02
-1.73796967e-01 1.41897070e+00 3.74781251e-01 8.19500089e-02
8.71846437e-01 8.67668033e-01 5.08509636e-01 -9.66031402e-02
-2.91279759e-02 -3.04627508e-01 1.43839702e-01 1.20128334e+00
1.38898551e-01 1.35767162e-01 -7.69920722e-02 1.04411066e+00
-6.79996014e-01 3.82338643e-01 1.20425463e-01 9.65507746e-01
-4.23998058e-01 -6.00233614e-01 -5.14373779e-01 6.78913355e-01
-2.86337212e-02 1.73787698e-01 -1.25951588e-01 8.00244689e-01
2.75739312e-01 1.25905943e+00 4.17990953e-01 -1.04743612e+00
2.58493751e-01 1.66869476e-01 5.03879786e-01 9.55376998e-02
-3.64605069e-01 1.40760288e-01 -1.62438840e-01 -7.50514448e-01
-3.47392559e-01 -4.05617386e-01 -1.63463807e+00 -5.02370119e-01
9.88113210e-02 -4.93199497e-01 8.97150457e-01 1.21228170e+00
4.45113719e-01 7.26959884e-01 4.19748425e-01 -1.15178394e+00
-5.15147865e-01 -1.07757545e+00 -5.98556578e-01 -1.12643987e-01
5.41832387e-01 -4.78100240e-01 -9.14569274e-02 2.76400819e-02]
|
[6.677103519439697, 0.20469580590724945]
|
7f4af9ca-4eb6-443c-b6ee-81f7fb3acc2b
|
multi-head-linear-attention-generative
|
2012.10898
| null |
https://arxiv.org/abs/2012.10898v1
|
https://arxiv.org/pdf/2012.10898v1.pdf
|
Multi-Head Linear Attention Generative Adversarial Network for Thin Cloud Removal
|
In remote sensing images, the existence of the thin cloud is an inevitable and ubiquitous phenomenon that crucially reduces the quality of imageries and limits the scenarios of application. Therefore, thin cloud removal is an indispensable procedure to enhance the utilization of remote sensing images. Generally, even though contaminated by thin clouds, the pixels still retain more or less surface information. Hence, different from thick cloud removal, thin cloud removal algorithms normally concentrate on inhibiting the cloud influence rather than substituting the cloud-contaminated pixels. Meanwhile, considering the surface features obscured by the cloud are usually similar to adjacent areas, the dependency between each pixel of the input is useful to reconstruct contaminated areas. In this paper, to make full use of the dependencies between pixels of the image, we propose a Multi-Head Linear Attention Generative Adversarial Network (MLAGAN) for Thin Cloud Removal. The MLA-GAN is based on the encoding-decoding framework consisting of multiple attention-based layers and deconvolutional layers. Compared with six deep learning-based thin cloud removal benchmarks, the experimental results on the RICE1 and RICE2 datasets demonstrate that the proposed framework MLA-GAN has dominant advantages in thin cloud removal.
|
['Rui Li', 'Chenxi Duan']
|
2020-12-20
| null | null | null | null |
['cloud-removal']
|
['computer-vision']
|
[ 3.72424006e-01 -2.46004105e-01 4.50080723e-01 -1.91663578e-01
-3.97261828e-01 -3.46265733e-01 2.01545596e-01 -4.49449599e-01
-3.78835574e-02 6.60025358e-01 -1.09750554e-01 -3.60042155e-01
1.50823817e-01 -1.08036542e+00 -6.70337737e-01 -1.43586624e+00
6.01605713e-01 -3.89829278e-02 -1.23290889e-01 -2.31509835e-01
-1.76390246e-01 5.29850304e-01 -1.28010714e+00 2.50876456e-01
1.42198265e+00 7.89439380e-01 6.71446145e-01 1.74971014e-01
-4.16864306e-01 5.62615752e-01 -6.77393079e-01 3.12976874e-02
4.05871361e-01 -4.63457853e-01 -5.96748553e-02 3.16105604e-01
2.36501127e-01 -5.90284288e-01 -3.98115516e-01 1.50906181e+00
3.88585299e-01 -1.90495118e-01 4.55596030e-01 -8.12415779e-01
-1.00328851e+00 3.73794764e-01 -1.00890231e+00 1.58558220e-01
-5.29843569e-01 2.59069204e-01 5.70858121e-01 -9.86296833e-01
8.53408277e-02 1.03994703e+00 4.90708679e-01 3.02374423e-01
-7.20018446e-01 -8.72973680e-01 6.43692434e-01 -2.55512614e-02
-1.52423620e+00 -2.37443179e-01 7.20420837e-01 -3.23875308e-01
3.70042741e-01 4.51423585e-01 7.44813025e-01 3.60221058e-01
2.01759696e-01 7.96109438e-01 1.14452660e+00 -1.48245633e-01
-2.75392503e-01 7.64004365e-02 2.89122341e-03 7.68152401e-02
5.48859358e-01 -2.65009925e-02 2.82551795e-01 2.77366698e-01
6.46505773e-01 5.95570207e-01 -6.53431177e-01 1.78904042e-01
-5.84512472e-01 5.86383820e-01 9.00494993e-01 1.79314733e-01
-5.07557988e-01 7.89295807e-02 1.81592777e-02 6.69037476e-02
9.95703757e-01 -7.21290261e-02 -7.60234073e-02 5.79752684e-01
-1.05711043e+00 2.13517144e-01 3.76892686e-02 7.38900483e-01
8.19401622e-01 6.54324651e-01 2.79107951e-02 5.86598217e-01
2.79332578e-01 1.09900308e+00 1.57511383e-01 -4.93742228e-01
4.69724625e-01 6.43299818e-01 3.42524618e-01 -9.17151034e-01
3.22721675e-02 -8.20084155e-01 -1.34918630e+00 4.26551551e-01
-1.42828673e-01 -1.45351797e-01 -1.09803295e+00 1.32788396e+00
2.15537563e-01 2.02028409e-01 1.18800089e-01 1.26889229e+00
8.44908774e-01 1.12030280e+00 6.27643690e-02 -3.50230724e-01
1.05110228e+00 -9.44023550e-01 -9.69184339e-01 -4.02762175e-01
6.86023459e-02 -6.14660084e-01 9.49488223e-01 2.33520344e-01
-8.45653832e-01 -5.35833776e-01 -1.02458549e+00 -1.79337524e-02
-4.22406226e-01 1.42933503e-01 7.84770072e-01 4.29956406e-01
-8.07722330e-01 2.98256725e-01 -6.62170947e-01 1.83220029e-01
5.70648432e-01 -6.76216185e-02 3.37760039e-02 -3.78566474e-01
-1.13840389e+00 6.41062856e-01 2.10326657e-01 1.01448858e+00
-9.48481441e-01 -5.70067167e-01 -4.21738714e-01 4.18876052e-01
1.99329749e-01 -4.95263755e-01 7.30920374e-01 -1.37728775e+00
-1.01903117e+00 5.54677546e-01 -1.96164563e-01 -1.13892458e-01
5.40316105e-01 -4.28382874e-01 -3.65629822e-01 -5.82208037e-02
4.99206148e-02 2.27003589e-01 1.07385576e+00 -1.58075118e+00
-4.94307429e-01 -4.87374783e-01 -8.68594423e-02 4.54704374e-01
-2.30057195e-01 -1.17596507e-01 -3.77254218e-01 -6.75776064e-01
2.12400541e-01 -7.41573393e-01 -1.81107298e-01 -8.63875300e-02
-5.17536640e-01 4.12134796e-01 1.28601384e+00 -1.02763009e+00
9.42876518e-01 -2.34469771e+00 1.23335823e-01 3.20204273e-02
3.95254850e-01 5.60702682e-01 -2.47842774e-01 3.35834771e-02
-2.27458820e-01 4.15835112e-01 -6.99448526e-01 -7.76172727e-02
-3.46887618e-01 2.82942474e-01 -5.53430259e-01 6.10526979e-01
3.63846719e-01 7.45766997e-01 -6.37174070e-01 -1.91855416e-01
2.83293366e-01 7.50157416e-01 -1.43472776e-01 2.58214355e-01
-1.84754908e-01 7.95357287e-01 -6.81657851e-01 6.79727614e-01
1.54594779e+00 6.70640543e-02 -2.57809669e-01 -1.04261609e-03
-3.94098312e-01 -2.28217840e-01 -7.70478606e-01 1.13835573e+00
-4.53174859e-01 5.22722781e-01 5.13445854e-01 -6.76175237e-01
8.84460449e-01 1.90364152e-01 2.58358538e-01 -7.02010989e-01
8.49345978e-03 3.28736126e-01 9.91208572e-03 -5.20520926e-01
3.42148453e-01 -3.44444931e-01 4.00329381e-01 1.11043260e-01
-6.64219081e-01 -3.60829204e-01 -3.78839731e-01 6.37216400e-03
5.44102192e-01 1.42436922e-01 -1.62257880e-01 7.50982612e-02
4.67609167e-01 -1.28098503e-01 8.34922373e-01 3.74653846e-01
-3.18826176e-02 8.15958977e-01 4.44751047e-02 -4.56736028e-01
-1.00376141e+00 -9.43455756e-01 -1.44199833e-01 5.58329761e-01
3.89002979e-01 2.78387874e-01 -7.51766324e-01 -3.11990976e-01
-4.24423255e-03 6.35652423e-01 -6.43314362e-01 -1.93063706e-01
-4.82691824e-01 -1.20105314e+00 2.92588115e-01 3.46311897e-01
9.77640450e-01 -1.20832646e+00 -2.35551700e-01 2.72406358e-02
-4.09338087e-01 -9.16254163e-01 -2.62851447e-01 3.70580591e-02
-8.27330291e-01 -8.33885670e-01 -7.74197280e-01 -3.46112281e-01
7.58416414e-01 1.14713120e+00 9.67100501e-01 5.15275300e-01
-1.25611335e-01 -3.53345513e-01 -5.77294350e-01 -8.90152931e-01
-1.21072635e-01 -4.77240160e-02 -4.27251011e-01 1.97580442e-01
1.92332640e-01 -6.44849241e-01 -6.89829171e-01 -1.64328367e-01
-1.26021898e+00 2.50988483e-01 8.97785246e-01 7.86747098e-01
7.26130247e-01 6.92713737e-01 2.34292403e-01 -1.06692863e+00
1.51720062e-01 -6.07281506e-01 -7.47349262e-01 1.52635738e-01
-4.52838928e-01 -5.05322456e-01 5.40588856e-01 -2.92102043e-02
-1.42905021e+00 -3.85855208e-04 -2.62929220e-02 -7.00941801e-01
5.09033352e-02 5.41876674e-01 -8.08619440e-01 -3.84945273e-02
-2.85295229e-02 6.77167237e-01 -3.89004558e-01 -5.31884909e-01
2.18435392e-01 7.31929064e-01 4.03286815e-01 1.29258305e-01
1.18360496e+00 6.93125069e-01 -2.54940599e-01 -1.08136117e+00
-8.23958874e-01 -3.72465491e-01 -3.87869120e-01 -2.72989362e-01
1.18204379e+00 -1.14957178e+00 -3.19009542e-01 9.75845754e-01
-1.22264051e+00 -1.84001103e-01 -3.14395726e-02 1.94391027e-01
3.01816970e-01 4.21821713e-01 -4.08886701e-01 -1.18354154e+00
-7.75301039e-01 -1.02294755e+00 1.06442726e+00 6.04611814e-01
8.31190228e-01 -6.17460191e-01 -3.47183019e-01 4.10739422e-01
5.40012896e-01 2.98411638e-01 9.76787388e-01 1.01148769e-01
-9.60689843e-01 -1.26665220e-01 -5.73185027e-01 7.08409965e-01
5.08692801e-01 2.52593517e-01 -1.35009623e+00 -2.51012802e-01
2.63400555e-01 2.01361448e-01 1.21449566e+00 5.63720167e-01
1.34416866e+00 -2.08779901e-01 -1.67226493e-01 1.03572702e+00
1.76358008e+00 2.21691847e-01 1.04975629e+00 3.98944199e-01
1.19279158e+00 4.25005734e-01 5.24643898e-01 3.10829550e-01
7.02144625e-03 1.38199434e-01 1.19535530e+00 -7.36220121e-01
-6.77822903e-02 1.24584407e-01 2.36535877e-01 7.70022750e-01
-3.98098588e-01 -4.57349330e-01 -5.84168136e-01 5.02099931e-01
-1.53897965e+00 -1.07572341e+00 -6.04755104e-01 2.01777244e+00
5.47345102e-01 -1.87381387e-01 -5.64623535e-01 6.43051043e-02
9.05547082e-01 4.88676161e-01 -8.48633051e-01 -5.43556996e-02
-5.81597805e-01 2.44161278e-01 7.64228404e-01 2.76538968e-01
-1.05619788e+00 9.46895480e-01 4.86056566e+00 7.07973421e-01
-1.48553073e+00 3.01172107e-01 5.48269749e-01 1.84845016e-03
-6.02880061e-01 -4.13091965e-02 -5.08061945e-01 7.47726560e-01
2.78547257e-01 9.93076265e-02 5.39359927e-01 5.61343312e-01
6.82228684e-01 1.24060893e-02 -2.99080819e-01 7.07093716e-01
-2.74501026e-01 -8.12321365e-01 2.36262307e-01 5.41910082e-02
8.43453348e-01 1.60684213e-01 3.21882576e-01 1.84329599e-01
1.71775788e-01 -1.03894985e+00 7.36949742e-01 6.41621232e-01
7.80480325e-01 -7.55547583e-01 1.02954495e+00 4.93614852e-01
-1.22098243e+00 -4.10653278e-02 -8.49934697e-01 -5.78695796e-02
-8.05332735e-02 1.21972048e+00 -2.37521395e-01 9.24210370e-01
8.20421278e-01 7.53699720e-01 -4.55176026e-01 8.90145242e-01
-6.29872024e-01 6.29479885e-01 -5.91091290e-02 6.95250809e-01
3.33954126e-01 -8.34536552e-01 4.80112612e-01 9.42753136e-01
3.94243985e-01 4.49829727e-01 4.17047292e-02 1.19717813e+00
-1.62385717e-01 -5.03718220e-02 -6.02932751e-01 -1.06853992e-01
3.33167315e-01 1.30242932e+00 -4.55397159e-01 -1.93290532e-01
-5.05878031e-01 1.12219715e+00 -1.28625244e-01 5.45807421e-01
-9.57161486e-01 -2.40597367e-01 7.10727870e-01 6.99120834e-02
3.57834339e-01 -1.08565926e-03 -3.54187727e-01 -1.25309348e+00
2.97201008e-01 -9.44805443e-01 -1.48469284e-01 -1.19970119e+00
-1.08433938e+00 6.24549329e-01 -5.66405058e-01 -1.39550364e+00
5.57636738e-01 -2.82149047e-01 -1.06477594e+00 1.56285584e+00
-2.16289759e+00 -1.46147573e+00 -1.02950382e+00 3.59367043e-01
5.51922441e-01 2.30781525e-01 7.33982027e-01 3.85724694e-01
-8.99416387e-01 1.21490946e-02 4.79920685e-01 1.17802650e-01
3.71545106e-01 -9.99926805e-01 2.86529988e-01 1.35855281e+00
-3.90896678e-01 3.59064966e-01 5.44326961e-01 -7.36542225e-01
-1.22697198e+00 -1.75152838e+00 2.84470767e-01 1.59088150e-01
1.50842845e-01 -2.07639754e-01 -1.45436382e+00 6.09510064e-01
3.30627084e-01 6.47730082e-02 3.67322415e-01 -4.41672236e-01
-2.18621388e-01 -4.04209912e-01 -1.03493083e+00 3.81367385e-01
6.35586500e-01 -3.97288859e-01 -1.94765702e-01 5.50678253e-01
8.20189416e-01 -2.90935248e-01 -3.59073520e-01 5.06750584e-01
1.81657463e-01 -1.03629255e+00 8.39764237e-01 -2.93878704e-01
6.23505354e-01 -5.84910572e-01 -1.46783635e-01 -1.49131370e+00
-5.26529133e-01 -1.17603596e-03 3.35919917e-01 1.42209792e+00
6.53208271e-02 -6.43085778e-01 6.41314507e-01 2.39475608e-01
-5.46917260e-01 -2.18887880e-01 -5.22701025e-01 -4.22883570e-01
3.56620580e-01 6.14595674e-02 9.09345388e-01 1.06146455e+00
-1.09321404e+00 2.99879368e-02 -4.28067416e-01 8.43855262e-01
6.80497885e-01 5.27501822e-01 6.80557489e-01 -1.18353760e+00
-1.23930044e-01 -4.08220559e-01 1.48264483e-01 -6.76236808e-01
2.12871417e-01 -6.74052477e-01 2.45487824e-01 -1.67945445e+00
3.89903903e-01 -7.29304314e-01 -3.55717927e-01 4.57837969e-01
-7.99512804e-01 3.88502359e-01 2.11063579e-01 5.28498709e-01
-7.76356757e-02 8.61533701e-01 1.63559449e+00 -6.32246494e-01
1.25669800e-02 7.13998601e-02 -7.12665021e-01 7.29945958e-01
8.53157222e-01 -4.49764371e-01 -2.63968885e-01 -1.19160485e+00
2.54265726e-01 -1.25970855e-01 4.62485045e-01 -8.00934732e-01
-1.71766445e-01 -3.32896173e-01 6.19071841e-01 -8.59659374e-01
2.61479914e-01 -1.02893591e+00 2.44281173e-01 4.79470789e-01
2.43358031e-01 -2.84887850e-01 2.97965854e-01 5.04957795e-01
-3.93892825e-01 -1.37838617e-01 8.79010022e-01 -4.63388741e-01
-5.86363792e-01 5.28409362e-01 -4.00743842e-01 -3.08610499e-01
7.57539153e-01 -9.08665136e-02 -3.82829815e-01 -3.57478857e-01
-4.39733803e-01 3.12314361e-01 5.24314225e-01 1.45932987e-01
7.71281123e-01 -9.89946067e-01 -1.16414273e+00 3.40033591e-01
-3.18734013e-02 5.95876753e-01 7.33783484e-01 7.50513136e-01
-6.72978520e-01 9.26323757e-02 -7.77766034e-02 -4.29193467e-01
-1.08019686e+00 6.14537776e-01 6.04347229e-01 -1.14240989e-01
-7.20923722e-01 8.19642961e-01 9.21239913e-01 -2.32683316e-01
-2.80952990e-01 -2.97708035e-01 -2.36463219e-01 -2.68314421e-01
5.25055230e-01 1.09884858e-01 2.33762980e-01 -6.01451695e-01
-1.19123841e-02 4.36075181e-01 5.84104098e-02 3.84883106e-01
1.51131868e+00 -3.05921167e-01 -5.62320054e-01 3.14564914e-01
6.43531740e-01 2.92853862e-01 -1.22879827e+00 -2.19937429e-01
-7.08673358e-01 -7.03923106e-01 5.00629127e-01 -6.49310887e-01
-1.89427888e+00 1.18921161e+00 6.97708011e-01 2.11698860e-01
1.51193750e+00 -5.19043088e-01 7.74195969e-01 5.89332581e-02
6.96384311e-02 -5.17315030e-01 -4.23088044e-01 3.43996793e-01
8.66815448e-01 -1.23951721e+00 1.25607222e-01 -7.57189274e-01
-4.55029577e-01 7.57220149e-01 6.76286101e-01 -1.16738372e-01
3.88200521e-01 3.44640046e-01 3.48520070e-01 -1.42203689e-01
-2.52320200e-01 -3.05004835e-01 -1.97274819e-01 6.14705205e-01
2.54167169e-01 2.21500456e-01 8.56556073e-02 6.01036370e-01
2.33101606e-01 -2.19214752e-01 5.87942839e-01 5.98736703e-01
-4.28370893e-01 -7.35484481e-01 -7.26317704e-01 5.20196438e-01
-4.74678844e-01 -5.83529651e-01 -3.37525487e-01 7.33275712e-01
4.57838595e-01 9.31618392e-01 1.19532496e-01 -1.94934055e-01
4.89331670e-02 -1.89254716e-01 1.95332319e-01 -7.84886599e-01
-6.15279555e-01 3.28676105e-01 -4.61069316e-01 -8.92734230e-02
-4.83668059e-01 -6.27835810e-01 -1.13561833e+00 -5.21152377e-01
-7.61349022e-01 1.01530492e-01 7.10814774e-01 9.59999979e-01
1.29968554e-01 9.82531905e-01 8.95833135e-01 -8.86526704e-01
-3.28084916e-01 -1.20279300e+00 -1.13540983e+00 1.88268110e-01
4.44229513e-01 -5.45829356e-01 -4.78332013e-01 5.36628708e-04]
|
[10.026091575622559, -1.9400556087493896]
|
37546ace-f49a-47a6-8594-a3b1f9d05b3c
|
rcps-rectified-contrastive-pseudo-supervision
|
2301.055
| null |
https://arxiv.org/abs/2301.05500v1
|
https://arxiv.org/pdf/2301.05500v1.pdf
|
RCPS: Rectified Contrastive Pseudo Supervision for Semi-Supervised Medical Image Segmentation
|
Medical image segmentation methods are generally designed as fully-supervised to guarantee model performance, which require a significant amount of expert annotated samples that are high-cost and laborious. Semi-supervised image segmentation can alleviate the problem by utilizing a large number of unlabeled images along with limited labeled images. However, learning a robust representation from numerous unlabeled images remains challenging due to potential noise in pseudo labels and insufficient class separability in feature space, which undermines the performance of current semi-supervised segmentation approaches. To address the issues above, we propose a novel semi-supervised segmentation method named as Rectified Contrastive Pseudo Supervision (RCPS), which combines a rectified pseudo supervision and voxel-level contrastive learning to improve the effectiveness of semi-supervised segmentation. Particularly, we design a novel rectification strategy for the pseudo supervision method based on uncertainty estimation and consistency regularization to reduce the noise influence in pseudo labels. Furthermore, we introduce a bidirectional voxel contrastive loss to the network to ensure intra-class consistency and inter-class contrast in feature space, which increases class separability in the segmentation. The proposed RCPS segmentation method has been validated on two public datasets and an in-house clinical dataset. Experimental results reveal that the proposed method yields better segmentation performance compared with the state-of-the-art methods in semi-supervised medical image segmentation. The source code is available at https://github.com/hsiangyuzhao/RCPS.
|
['Lichi Zhang', 'Ying Mao', 'Xuehai Wu', 'Qian Wang', 'Sheng Wang', 'Zengxin Qi', 'Xiangyu Zhao']
|
2023-01-13
| null | null | null | null |
['semi-supervised-medical-image-segmentation']
|
['computer-vision']
|
[ 5.69311500e-01 2.79929370e-01 -4.96155709e-01 -6.72020674e-01
-9.30716455e-01 -2.17764527e-01 2.13648397e-02 2.07560193e-02
-4.87151533e-01 7.74949253e-01 -6.32371530e-02 -4.52118292e-02
-2.06703931e-01 -5.22499382e-01 -5.39915025e-01 -9.86393273e-01
4.53835517e-01 5.14611959e-01 3.57381195e-01 1.78519174e-01
2.45511848e-02 2.30670303e-01 -1.12842679e+00 7.39536062e-02
1.48557103e+00 9.49608147e-01 3.69972765e-01 -3.37680131e-02
-2.04695299e-01 3.36978257e-01 -2.91743726e-01 9.08315554e-02
2.48912379e-01 -5.14485419e-01 -9.04030442e-01 5.85010648e-01
4.47671190e-02 -2.05178395e-01 8.08141902e-02 1.41896069e+00
4.23739135e-01 3.47078852e-02 7.14441061e-01 -1.11121762e+00
-4.73235101e-01 5.84748268e-01 -9.33392584e-01 1.34230658e-01
-1.78912878e-01 1.08031429e-01 7.15894878e-01 -6.57185197e-01
5.11177123e-01 8.06263506e-01 4.47829217e-01 4.79564458e-01
-1.18894839e+00 -8.00153673e-01 4.13990021e-02 -2.39618167e-01
-1.48737991e+00 -1.04089953e-01 8.95203650e-01 -4.22607899e-01
2.91641504e-01 1.67350739e-01 4.70949739e-01 4.91836339e-01
4.65260819e-02 9.29551601e-01 1.32941329e+00 -2.14306027e-01
1.60151735e-01 2.32710600e-01 3.31036806e-01 8.96776974e-01
2.72111595e-01 5.63066527e-02 5.49939573e-02 6.06500283e-02
9.17467058e-01 2.18330592e-01 -4.26984787e-01 -5.59395075e-01
-1.11186147e+00 6.68608904e-01 8.20743024e-01 3.29506785e-01
-2.43498772e-01 -3.50902766e-01 4.53551203e-01 -1.96894154e-01
6.23706758e-01 1.12645045e-01 -3.06041449e-01 3.46045107e-01
-1.19903064e+00 -3.32248867e-01 2.89261103e-01 9.24081862e-01
7.99162269e-01 -7.75960684e-02 3.61146312e-03 8.92163038e-01
4.80145276e-01 3.45598340e-01 5.90014219e-01 -7.30526567e-01
4.01419282e-01 8.65096033e-01 -2.19301224e-01 -7.72506535e-01
-5.34406424e-01 -5.40865004e-01 -1.08775520e+00 -9.77840498e-02
3.93919408e-01 -2.65731756e-02 -1.34718359e+00 1.40672874e+00
6.05951130e-01 2.14439690e-01 -1.43633604e-01 1.20407498e+00
9.17765141e-01 5.36684632e-01 1.71042114e-01 -4.85557258e-01
1.13359904e+00 -1.14516962e+00 -7.48415709e-01 -1.26594365e-01
7.21755385e-01 -5.55288374e-01 1.12343407e+00 2.58506596e-01
-8.71901155e-01 -5.45486629e-01 -1.11822164e+00 2.31609389e-01
-7.38238767e-02 2.83067763e-01 5.75921118e-01 6.72883153e-01
-5.38434982e-01 2.79801905e-01 -1.19429898e+00 2.54408028e-02
9.01216269e-01 5.86279452e-01 -3.07920963e-01 -2.32295468e-01
-1.13681078e+00 4.52965498e-01 5.42490780e-01 4.34113622e-01
-6.54838741e-01 -6.11708522e-01 -1.05836868e+00 -2.65708089e-01
6.30531192e-01 -1.39746234e-01 9.08873081e-01 -1.04088175e+00
-1.29607153e+00 9.31676090e-01 8.18475522e-03 -1.11633651e-01
6.44033968e-01 1.58499151e-01 -2.71248937e-01 4.77827847e-01
3.98598194e-01 8.34748209e-01 6.39580369e-01 -1.44749665e+00
-5.18227041e-01 -6.02789700e-01 -4.47906077e-01 3.88596088e-01
-1.77608952e-01 -2.24031806e-01 -5.99860311e-01 -6.69004083e-01
6.87386870e-01 -9.55908775e-01 -6.49690568e-01 5.25354631e-02
-6.96880817e-01 3.17040533e-02 7.73148537e-01 -6.53558910e-01
9.55572605e-01 -2.13504529e+00 -3.44715193e-02 4.38091934e-01
1.25369936e-01 3.92866403e-01 1.71576470e-01 -4.26394910e-01
-5.02135381e-02 1.63889602e-01 -9.30819869e-01 -2.30051816e-01
-4.53001738e-01 2.95950174e-01 2.57970393e-01 7.27757275e-01
1.19690493e-01 8.07239532e-01 -9.83836949e-01 -1.08338976e+00
4.16319519e-01 5.04113019e-01 -2.95823902e-01 1.06629394e-01
-9.47074145e-02 9.97062147e-01 -8.47552240e-01 8.88503611e-01
8.93244267e-01 -4.81467009e-01 -3.49574462e-02 -2.02725321e-01
1.90772548e-01 -1.39001697e-01 -1.20045030e+00 1.88906074e+00
-1.70857236e-01 -8.80589485e-02 2.65661299e-01 -1.24604225e+00
8.46598148e-01 3.42467189e-01 7.54933119e-01 -4.85101223e-01
4.60150689e-01 4.67550218e-01 -1.76247671e-01 -5.01663804e-01
-2.00869268e-04 -3.57966185e-01 1.06514543e-01 3.06136787e-01
-1.19962476e-01 -2.63783962e-01 1.83852002e-01 1.41423553e-01
4.04980510e-01 3.09576958e-01 1.09648265e-01 -3.73265833e-01
7.23090649e-01 2.21664503e-01 1.02239490e+00 3.11916471e-01
-5.33381581e-01 9.06229496e-01 2.62086928e-01 2.93556359e-02
-5.71275353e-01 -1.00221992e+00 -5.68940878e-01 4.44314390e-01
5.09497941e-01 2.42852956e-01 -9.56995189e-01 -1.04822850e+00
-1.78677693e-01 2.30859086e-01 -5.14627814e-01 -1.01210296e-01
-4.90085334e-01 -1.06160343e+00 3.37998807e-01 6.18773162e-01
8.34812105e-01 -9.33112085e-01 -4.01564747e-01 1.18076481e-01
-3.21964771e-01 -1.03691137e+00 -7.12395132e-01 2.31258526e-01
-1.11902165e+00 -1.18340647e+00 -1.11319816e+00 -1.10131156e+00
1.35875392e+00 1.66598380e-01 6.94131970e-01 2.48548925e-01
-4.11133975e-01 -9.65858176e-02 -3.00743967e-01 -1.49650112e-01
-3.24488878e-01 1.19368024e-01 -1.94316342e-01 -9.59283859e-03
1.20521985e-01 -2.54446685e-01 -7.94931889e-01 5.78218699e-01
-1.07464886e+00 1.38818577e-01 7.29761481e-01 1.16415083e+00
1.19687712e+00 3.32955182e-01 7.56675303e-01 -1.36676800e+00
1.66303173e-01 -2.98817396e-01 -7.09623575e-01 2.66085505e-01
-9.27416623e-01 -2.18826011e-01 5.20960808e-01 -2.63335735e-01
-1.42238820e+00 4.77751195e-01 -1.11017175e-01 -2.26940081e-01
-1.85151294e-01 5.65790951e-01 -2.24555641e-01 -1.56173497e-01
4.12041575e-01 2.15589061e-01 3.72967839e-01 -2.34907299e-01
2.61876911e-01 8.09939981e-01 4.75293398e-01 -3.66900474e-01
4.86045122e-01 7.19105899e-01 -4.35457826e-02 -4.92808431e-01
-1.00963330e+00 -7.22204089e-01 -8.32552195e-01 -5.48807234e-02
8.98923695e-01 -8.34972739e-01 -1.83789060e-01 5.08404672e-01
-5.32818556e-01 -3.48833382e-01 -2.43008584e-01 7.37928331e-01
-3.93190295e-01 6.52017415e-01 -6.75911784e-01 -3.77124310e-01
-5.28299212e-01 -1.65298903e+00 9.33511078e-01 5.52697957e-01
1.39644578e-01 -1.02854669e+00 -3.00017446e-01 8.26255202e-01
1.29613727e-01 3.08276772e-01 6.33801043e-01 -6.27000630e-01
-5.30770123e-01 -1.48258999e-01 -4.58288968e-01 6.37222648e-01
5.06375074e-01 -2.79143721e-01 -5.96448779e-01 -2.87257433e-01
1.47246704e-01 -4.42026705e-01 9.18370962e-01 8.40466559e-01
1.29406738e+00 5.62945232e-02 -4.69096214e-01 6.46794915e-01
1.33893490e+00 1.38573810e-01 3.24296921e-01 1.15075834e-01
9.85463738e-01 6.98387742e-01 1.06005287e+00 2.47831091e-01
1.99410185e-01 2.20971465e-01 2.30063871e-01 -5.83900988e-01
-6.00142665e-02 -7.74003640e-02 -3.32202554e-01 7.89259195e-01
2.35171169e-01 2.54132934e-02 -9.21163619e-01 6.28540874e-01
-1.79559779e+00 -2.79067159e-01 -1.83006421e-01 2.03046441e+00
1.12538409e+00 3.24785680e-01 -1.39355510e-01 2.23253146e-01
9.17131603e-01 -1.04210470e-02 -7.40554154e-01 1.39301449e-01
1.03573978e-01 4.38347459e-02 5.75873911e-01 4.42101240e-01
-1.20228541e+00 8.23853374e-01 5.05502892e+00 1.02945733e+00
-1.18844855e+00 1.95434883e-01 1.10182762e+00 1.80968642e-01
-1.84283778e-01 -1.67066678e-01 -6.36150479e-01 4.78458703e-01
4.02174503e-01 2.63348997e-01 1.23231672e-02 7.66378880e-01
3.20653915e-01 -4.38226759e-01 -7.59034514e-01 7.85888791e-01
7.23308623e-02 -1.13676202e+00 -9.45299417e-02 -1.32306814e-01
9.89236355e-01 -5.03936410e-02 -8.63180123e-03 -4.96741347e-02
-8.36069807e-02 -9.02973771e-01 1.65156990e-01 1.45238325e-01
8.88661146e-01 -7.13036299e-01 8.98007870e-01 3.65155876e-01
-1.08053005e+00 2.12896645e-01 -1.66851595e-01 4.78240252e-01
2.92188376e-01 7.57320702e-01 -6.29883409e-01 6.37970388e-01
4.66064751e-01 7.49644518e-01 -4.10594285e-01 9.73242879e-01
-3.63370270e-01 6.64499521e-01 -2.43991539e-01 3.04706514e-01
3.58193696e-01 -4.03213084e-01 2.47809127e-01 8.56773078e-01
-1.54732898e-01 2.63806731e-01 5.20522475e-01 8.80009234e-01
3.25748846e-02 2.78528243e-01 7.66876712e-02 1.46561131e-01
2.78098434e-01 1.35397589e+00 -1.37822330e+00 -3.60114157e-01
-2.98333704e-01 7.59125769e-01 -6.05156496e-02 2.96342194e-01
-8.24148238e-01 -2.87733793e-01 -4.32561159e-01 1.42194286e-01
-1.22526929e-01 1.77462056e-01 -6.52385950e-01 -1.05281079e+00
-2.29397137e-02 -5.71786702e-01 6.17970943e-01 -5.09319007e-01
-1.11591041e+00 5.92553914e-01 1.81275353e-01 -1.25578928e+00
2.34004408e-01 -2.66156286e-01 -4.01256889e-01 7.37017155e-01
-1.71009791e+00 -1.20800841e+00 -4.49662089e-01 5.00129223e-01
5.12859941e-01 1.28710985e-01 4.04197186e-01 4.84952867e-01
-8.02010715e-01 5.02716482e-01 9.25222188e-02 1.93941832e-01
7.00445712e-01 -1.20850348e+00 -3.11697096e-01 7.33437300e-01
-2.30889499e-01 4.55941290e-01 2.80893385e-01 -7.74762034e-01
-7.74817407e-01 -1.20586812e+00 2.66711980e-01 9.65861529e-02
2.22238675e-01 2.94349771e-02 -1.06489909e+00 5.49935460e-01
-2.17390686e-01 5.77844262e-01 7.13218510e-01 -4.27159905e-01
2.54667103e-01 -6.79277554e-02 -1.55542219e+00 3.73863101e-01
6.41341984e-01 -1.60107359e-01 -4.17753309e-01 5.37085235e-01
6.58842027e-01 -6.77752495e-01 -9.36401546e-01 7.98197985e-01
2.40225047e-01 -6.23949528e-01 7.33347833e-01 3.59239057e-02
2.32747108e-01 -4.78349626e-01 3.08158100e-01 -1.02202952e+00
1.57087818e-02 -2.26992607e-01 4.52586830e-01 1.23551857e+00
4.79630500e-01 -7.36616254e-01 1.14689219e+00 8.00462246e-01
-3.69598866e-01 -1.08712733e+00 -8.67202938e-01 -6.58812642e-01
1.00701332e-01 4.61496450e-02 2.67503738e-01 9.97702301e-01
-1.96365416e-02 3.71868648e-02 -3.99650820e-02 2.06312284e-01
9.00594950e-01 1.83241516e-01 1.86415613e-01 -1.06716967e+00
-4.94928248e-02 -1.79657400e-01 -1.81590796e-01 -7.76156008e-01
1.87598854e-01 -1.05267525e+00 3.42489094e-01 -1.67988765e+00
4.52105492e-01 -8.64606380e-01 -4.52427328e-01 5.97068965e-01
-3.80225092e-01 4.90573108e-01 -3.70307386e-01 4.99412835e-01
-5.90052605e-01 5.93866944e-01 1.85090184e+00 -1.99927598e-01
-4.03795153e-01 1.80264965e-01 -5.49718082e-01 7.58595049e-01
9.08924103e-01 -6.61226511e-01 -7.36437738e-01 -1.75331518e-01
-3.93361926e-01 2.14657068e-01 1.48165822e-01 -7.45213389e-01
1.98458195e-01 -1.66558400e-01 3.21147233e-01 -6.34517848e-01
-3.87433829e-04 -8.03896606e-01 -1.24578476e-01 6.16319478e-01
-3.64427119e-01 -5.84419847e-01 -5.59525490e-02 4.98669297e-01
-4.55793858e-01 -3.10062468e-01 1.10643625e+00 -2.66459942e-01
-5.47679067e-01 5.14816880e-01 3.58769968e-02 1.82418361e-01
1.05993617e+00 -3.01220953e-01 -3.95019427e-02 8.77500698e-02
-8.39160740e-01 7.22028971e-01 2.82734454e-01 2.80481614e-02
6.56163871e-01 -1.01006281e+00 -4.58038628e-01 1.75167963e-01
5.61760627e-02 6.79807961e-01 5.83458483e-01 1.18467367e+00
-5.86449504e-01 2.80621499e-01 -1.90097108e-01 -8.90474677e-01
-1.14013481e+00 4.70583260e-01 3.20179731e-01 -2.92701989e-01
-5.44907570e-01 8.19571614e-01 2.42144629e-01 -7.11614013e-01
1.83924213e-01 -3.64641517e-01 -1.89274877e-01 -1.91759318e-01
1.22581743e-01 2.15188801e-01 -2.57722009e-02 -8.87334347e-01
-4.35396761e-01 6.09996080e-01 -3.27871799e-01 1.53493062e-01
9.98130798e-01 -2.52886742e-01 -1.22418582e-01 1.72654346e-01
1.22973645e+00 -3.86912644e-01 -1.45349765e+00 -3.40085834e-01
-1.85945153e-01 -3.10629874e-01 4.29833114e-01 -7.63984084e-01
-1.46494067e+00 8.26992273e-01 8.39873493e-01 -2.47237235e-01
1.19688439e+00 -6.19261079e-02 9.59729075e-01 -9.45933908e-03
2.18669564e-01 -1.21954763e+00 3.16306949e-02 -1.23355247e-01
3.92442495e-01 -1.75803685e+00 3.63160223e-01 -9.66212094e-01
-8.48766208e-01 6.82252109e-01 7.37328887e-01 -7.66157135e-02
9.11211610e-01 5.39817363e-02 2.80548960e-01 -2.43844837e-01
5.92638850e-02 -5.01485355e-02 4.04993415e-01 3.25416297e-01
3.82466406e-01 6.45250902e-02 -6.22191727e-01 6.76304996e-01
2.96930999e-01 1.78140789e-01 2.82516181e-01 1.11471927e+00
-3.43364984e-01 -9.04920340e-01 -3.19937438e-01 5.77487111e-01
-5.61889112e-01 8.28887820e-02 9.63442102e-02 7.08593130e-01
1.71649784e-01 9.18951929e-01 -2.17016935e-01 5.68386018e-02
4.85913828e-02 -2.90783852e-01 2.46151611e-01 -7.92148054e-01
-3.73370707e-01 5.88589311e-01 -3.35819960e-01 -2.90500700e-01
-7.85365462e-01 -4.86135632e-01 -2.02479625e+00 5.03169835e-01
-7.85647035e-01 2.84191608e-01 5.38861692e-01 9.51331258e-01
6.61153942e-02 5.69842696e-01 7.17239797e-01 -5.99761844e-01
-5.34481585e-01 -8.03279459e-01 -8.09943974e-01 4.17572856e-01
1.33514971e-01 -7.13936865e-01 -3.56485635e-01 1.25980571e-01]
|
[14.589546203613281, -2.020796775817871]
|
31779f06-c97c-4972-95ab-61bdb7ce23be
|
on-mitigating-hard-clusters-for-face
|
2207.11895
| null |
https://arxiv.org/abs/2207.11895v1
|
https://arxiv.org/pdf/2207.11895v1.pdf
|
On Mitigating Hard Clusters for Face Clustering
|
Face clustering is a promising way to scale up face recognition systems using large-scale unlabeled face images. It remains challenging to identify small or sparse face image clusters that we call hard clusters, which is caused by the heterogeneity, \ie, high variations in size and sparsity, of the clusters. Consequently, the conventional way of using a uniform threshold (to identify clusters) often leads to a terrible misclassification for the samples that should belong to hard clusters. We tackle this problem by leveraging the neighborhood information of samples and inferring the cluster memberships (of samples) in a probabilistic way. We introduce two novel modules, Neighborhood-Diffusion-based Density (NDDe) and Transition-Probability-based Distance (TPDi), based on which we can simply apply the standard Density Peak Clustering algorithm with a uniform threshold. Our experiments on multiple benchmarks show that each module contributes to the final performance of our method, and by incorporating them into other advanced face clustering methods, these two modules can boost the performance of these methods to a new state-of-the-art. Code is available at: https://github.com/echoanran/On-Mitigating-Hard-Clusters.
|
['Qianru Sun', 'Yun Liang', 'Tao Wang', 'Jianqiang Huang', 'Chen Shen', 'Chong Chen', 'Huasong Zhong', 'Yingjie Chen']
|
2022-07-25
| null | null | null | null |
['face-clustering']
|
['computer-vision']
|
[-2.19156042e-01 -1.52645335e-01 -2.85377890e-01 -5.54531634e-01
-6.08841360e-01 -3.82751673e-01 5.77678740e-01 -2.56013423e-01
1.50286973e-01 2.72255391e-01 -4.54322211e-02 -7.36897960e-02
-1.69240981e-01 -6.23504519e-01 -4.93461341e-01 -1.09546924e+00
-2.63872355e-01 7.41987944e-01 2.29026884e-01 3.25358361e-01
8.22835267e-02 5.14598787e-01 -1.90030551e+00 4.72448051e-01
6.49235547e-01 9.41550910e-01 -3.86508442e-02 2.08265677e-01
-1.76632375e-01 5.63802600e-01 -5.27434826e-01 -2.42096812e-01
2.35108748e-01 -6.12873077e-01 -6.10909462e-01 2.03124717e-01
3.42037827e-01 -8.32435116e-02 -1.02951348e-01 1.34195065e+00
3.18889797e-01 1.32224709e-02 8.83641243e-01 -1.47287524e+00
-6.04049623e-01 7.06276238e-01 -1.13978422e+00 7.79920965e-02
1.19512826e-01 -1.03917509e-01 4.77608740e-01 -1.07578063e+00
4.71352249e-01 1.68321443e+00 7.38295138e-01 6.75619364e-01
-1.18963289e+00 -8.93156290e-01 2.71657050e-01 4.59063470e-01
-2.00862145e+00 -8.22122514e-01 6.99486732e-01 -6.01159751e-01
2.91229248e-01 2.79295862e-01 1.48320779e-01 8.28213453e-01
-4.13806051e-01 5.89690387e-01 1.24878585e+00 -2.96112359e-01
4.76879627e-01 7.49074444e-02 1.18261412e-01 6.29970372e-01
1.96516737e-01 -1.29921302e-01 -3.48739207e-01 -4.70995367e-01
4.17258590e-01 2.73754030e-01 -2.94619650e-01 -3.02813441e-01
-6.80364311e-01 9.25543845e-01 2.59211689e-01 3.43371809e-01
-7.84475729e-02 -4.76465002e-02 9.40823033e-02 2.70282347e-02
4.80354339e-01 -4.61507529e-01 -2.07328141e-01 -2.46079231e-04
-1.33793640e+00 -1.50184274e-01 9.06104028e-01 6.13440752e-01
1.03217888e+00 -3.03328156e-01 -9.32715759e-02 9.19444621e-01
4.68358964e-01 5.46197116e-01 3.92999500e-01 -1.09371889e+00
-1.55568928e-01 5.46840191e-01 -2.58704722e-01 -1.11152077e+00
-1.32813230e-01 -2.72826683e-02 -9.52746034e-01 1.93154037e-01
5.28601587e-01 -1.91378489e-01 -1.18106401e+00 1.76234031e+00
6.22088194e-01 8.32010090e-01 -4.06272173e-01 7.77393937e-01
4.85563993e-01 5.77204645e-01 -7.66341239e-02 -4.67286348e-01
1.17480886e+00 -7.03415155e-01 -6.18589044e-01 7.02659339e-02
3.71438950e-01 -6.85426652e-01 7.96076417e-01 4.63977695e-01
-7.58701622e-01 -4.55461532e-01 -7.38876462e-01 5.51893950e-01
-2.55871236e-01 1.47607878e-01 4.75236624e-01 9.89286840e-01
-1.36611688e+00 7.12237597e-01 -1.08301556e+00 -3.85069728e-01
8.56998384e-01 6.00118458e-01 -2.85629272e-01 -4.41872895e-01
-5.30962706e-01 9.84734073e-02 8.20591599e-02 7.42507204e-02
-7.52430797e-01 -7.83958137e-01 -6.09489262e-01 -5.89032024e-02
4.39574212e-01 -1.78643674e-01 6.63427532e-01 -8.55724216e-01
-1.30155849e+00 6.76661611e-01 -6.44525468e-01 2.27048025e-02
3.04978758e-01 1.63449213e-01 -4.33173507e-01 3.21395725e-01
8.44400898e-02 6.77221656e-01 1.02252102e+00 -1.60133934e+00
-5.57980776e-01 -7.69096136e-01 -6.68784142e-01 -2.78636307e-01
-4.92590517e-01 1.51167884e-01 -9.27842438e-01 -4.36163515e-01
5.14791608e-02 -1.20826292e+00 -1.04866691e-01 -1.08986966e-01
-5.52471340e-01 -5.64724922e-01 1.22410071e+00 -3.37408036e-01
1.32049751e+00 -2.43592334e+00 1.64686143e-02 6.71046793e-01
4.38566178e-01 1.43629640e-01 -4.14398536e-02 5.09733818e-02
6.87520579e-02 2.10814670e-01 -4.43419039e-01 -5.26381433e-01
-1.51483286e-02 1.25716031e-01 8.70935433e-03 7.61303246e-01
2.14002416e-01 5.02638102e-01 -8.22195470e-01 -5.61986685e-01
1.52878806e-01 7.56353974e-01 -6.12540841e-01 6.46839142e-02
-2.88748555e-03 4.72167581e-01 -1.12363175e-01 8.27043891e-01
1.21100652e+00 -4.47713703e-01 6.08123302e-01 -1.64344251e-01
2.10961372e-01 -2.59379804e-01 -1.47217858e+00 1.22967088e+00
1.94892827e-02 1.73612192e-01 4.07180876e-01 -1.04798460e+00
6.93022668e-01 6.97899014e-02 7.53289700e-01 -7.96574131e-02
7.24295154e-02 1.80782154e-02 -7.20285326e-02 -9.55966953e-03
-7.69456923e-02 7.13012740e-02 2.55531013e-01 5.49757838e-01
7.74175376e-02 5.80017269e-01 2.75094241e-01 2.21064121e-01
1.16509008e+00 -3.07170272e-01 -1.62435904e-01 -4.75930691e-01
3.74517649e-01 -3.10864061e-01 6.46624207e-01 6.00678980e-01
-3.48208666e-01 7.19738841e-01 5.70051253e-01 -5.18201403e-02
-6.59853995e-01 -1.24236119e+00 -3.68565232e-01 1.00397015e+00
-1.05172925e-01 -5.29798687e-01 -1.01297021e+00 -8.87258708e-01
1.73199013e-01 5.39829694e-02 -7.71780908e-01 -1.90666631e-01
-2.77729690e-01 -1.13922906e+00 3.85835826e-01 3.29718798e-01
2.73196101e-01 -7.05809653e-01 1.87069550e-01 -1.90466017e-01
-2.71466617e-02 -1.09591889e+00 -5.23133755e-01 5.78146540e-02
-6.20840967e-01 -1.12903130e+00 -5.98704338e-01 -7.65256524e-01
9.91830647e-01 3.46184343e-01 1.02752352e+00 4.39741343e-01
-3.46785069e-01 3.03448170e-01 -3.94641012e-01 -2.35888120e-02
-3.49025220e-01 -1.59530774e-01 4.41289037e-01 4.47035313e-01
7.91735053e-01 -6.26969159e-01 -7.65880644e-01 5.49751103e-01
-7.13563025e-01 -6.71236515e-01 4.14365292e-01 6.67393446e-01
6.08612061e-01 4.42830741e-01 5.80463946e-01 -1.01910377e+00
3.67119133e-01 -9.10960376e-01 -4.15449798e-01 2.00705916e-01
-6.64166808e-01 -1.39736772e-01 5.58404684e-01 -5.98244846e-01
-8.24638486e-01 3.51391345e-01 -3.18231136e-02 -7.78629363e-01
-3.94277811e-01 2.01859489e-01 -3.70844036e-01 -1.93689689e-01
3.84719580e-01 6.08662404e-02 3.02605420e-01 -4.65242118e-01
4.78878558e-01 7.95450389e-01 4.48048204e-01 -5.68694174e-01
8.25581193e-01 8.84390831e-01 -1.48944497e-01 -9.02084231e-01
-5.23654461e-01 -7.64003038e-01 -7.33006358e-01 -1.68651551e-01
5.93705356e-01 -1.02677393e+00 -8.90297771e-01 4.33163583e-01
-5.65638185e-01 -4.07680809e-01 -2.88660955e-02 1.62468016e-01
-1.01687141e-01 7.93979764e-01 -7.16989994e-01 -7.77888179e-01
7.06501976e-02 -1.17383838e+00 1.04314983e+00 3.03659141e-01
-6.68253377e-02 -7.92382658e-01 -1.18250899e-01 2.44168669e-01
2.38094643e-01 1.33217603e-01 6.85241520e-01 -5.85948646e-01
-3.05026323e-01 7.24183396e-02 -1.91438422e-01 3.27763051e-01
3.61292869e-01 3.36051702e-01 -1.20014954e+00 -4.68498439e-01
2.61538122e-02 -1.96810305e-01 1.05111492e+00 5.84678531e-01
1.44561863e+00 -4.55801010e-01 -7.92236567e-01 4.84222174e-01
1.38906312e+00 4.64451090e-02 7.38309681e-01 -3.93013239e-01
7.85086036e-01 5.94519794e-01 2.84845948e-01 7.38765419e-01
4.59751934e-01 7.18000054e-01 2.03840837e-01 -1.74722001e-02
-3.17024320e-01 -1.74154174e-02 5.42432845e-01 8.23961556e-01
3.61364447e-02 1.75642818e-02 -8.92766893e-01 5.46627223e-01
-1.81823730e+00 -1.14821386e+00 -5.47439717e-02 2.28107476e+00
1.02342236e+00 -2.48960271e-01 4.63937402e-01 1.89398631e-01
1.21023452e+00 -2.13645011e-01 -4.96979833e-01 9.75288674e-02
1.91505745e-01 2.04444468e-01 2.11166099e-01 4.14777905e-01
-1.10900891e+00 9.28892314e-01 6.35184002e+00 1.31142354e+00
-1.15849447e+00 2.74925321e-01 1.08803797e+00 -6.76593781e-02
5.84430210e-02 -2.38807425e-01 -1.17122436e+00 9.34245884e-01
1.25335121e+00 1.43952474e-01 6.27788603e-01 9.24930096e-01
1.40824646e-01 -4.89279591e-02 -1.09051335e+00 1.16262650e+00
2.10126847e-01 -1.13007843e+00 -1.20899621e-02 2.58755654e-01
8.16974401e-01 5.13167456e-02 2.96884328e-01 2.04838872e-01
5.99523842e-01 -1.15493155e+00 4.09182698e-01 2.78233021e-01
7.81501889e-01 -8.23769450e-01 5.41636288e-01 1.54133916e-01
-1.42610013e+00 -1.79469809e-01 -5.18206358e-01 3.16140294e-01
-2.27393419e-01 1.12214351e+00 -7.03338146e-01 2.90989488e-01
1.03487980e+00 6.06857300e-01 -6.26510620e-01 7.86257625e-01
1.60374597e-01 8.99747670e-01 -6.31593287e-01 3.13919634e-01
-2.08518043e-01 -3.54467243e-01 6.16118461e-02 1.18390596e+00
4.37833428e-01 5.34449518e-02 4.29479063e-01 9.20845449e-01
-3.05545837e-01 -1.13756880e-01 -3.69833708e-01 1.72701493e-01
9.21129167e-01 1.54309201e+00 -1.13708866e+00 -3.50938588e-01
-4.49213445e-01 9.53352749e-01 3.88449490e-01 2.22936198e-01
-7.93256044e-01 9.81895719e-03 7.69763649e-01 3.11530888e-01
6.26186669e-01 -1.33743957e-01 -1.85268000e-02 -9.40528989e-01
-3.31582017e-02 -8.60982955e-01 5.17219245e-01 -8.50160569e-02
-1.73372757e+00 6.91460133e-01 -1.31886095e-01 -9.27669168e-01
-1.16029866e-02 -4.25552398e-01 -6.78796351e-01 3.83893192e-01
-1.34784198e+00 -8.72845292e-01 -3.99478287e-01 9.69863772e-01
2.39687577e-01 -3.72973792e-02 6.05121076e-01 6.57305717e-01
-7.89291143e-01 9.10597861e-01 3.10301542e-01 2.22230628e-01
8.90751719e-01 -1.12758052e+00 -9.94905755e-02 6.78964078e-01
1.80693507e-01 6.10376000e-01 2.79802293e-01 -6.13155007e-01
-1.27608776e+00 -1.45664024e+00 3.18002909e-01 -5.66411138e-01
4.92991835e-01 -4.55782652e-01 -9.94838893e-01 4.50323254e-01
-4.65363450e-02 4.89189088e-01 1.01102197e+00 1.01682879e-01
-4.83604699e-01 -1.62938610e-01 -1.38220990e+00 2.90191650e-01
1.10390079e+00 -3.14582050e-01 -8.60632025e-03 3.42616588e-01
3.74398530e-01 2.26483971e-01 -1.00195265e+00 3.43308359e-01
2.50695288e-01 -1.15395296e+00 9.57706273e-01 -1.48443073e-01
6.01069890e-02 -6.02792680e-01 -8.32008421e-02 -1.12347817e+00
-5.45214415e-01 -5.18839896e-01 -3.47141117e-01 1.56211507e+00
2.54960597e-01 -5.59239745e-01 1.06244969e+00 4.61171538e-01
1.00756459e-01 -6.20796084e-01 -1.02112293e+00 -9.09889579e-01
6.29968494e-02 -7.91359544e-02 8.39498043e-01 1.23205602e+00
7.05845132e-02 -2.02719681e-02 -7.88418651e-02 3.53353411e-01
1.00803828e+00 -1.77340563e-02 6.56072438e-01 -1.49547517e+00
1.95308588e-02 -5.15566289e-01 -5.05141139e-01 -8.15080762e-01
3.35920215e-01 -9.58839297e-01 1.20041661e-01 -1.02264297e+00
4.40597832e-01 -7.29529977e-01 -2.36207709e-01 6.02692246e-01
-3.04612339e-01 6.78734064e-01 5.01558334e-02 4.06133920e-01
-9.61801350e-01 3.75613153e-01 6.31673455e-01 -1.06621668e-01
-2.31624424e-01 -1.42964378e-01 -7.13958681e-01 7.44799614e-01
7.80721426e-01 -5.82194209e-01 -2.33373642e-01 5.95976859e-02
-3.78407300e-01 -3.27344894e-01 1.75437793e-01 -1.30041254e+00
4.34836477e-01 -2.89225895e-02 5.42866945e-01 -4.64917928e-01
2.94975787e-01 -7.68932700e-01 2.24114001e-01 3.51234585e-01
1.41400248e-01 -2.02163100e-01 1.08848833e-01 7.02005863e-01
-2.40372151e-01 1.80215448e-01 1.03504956e+00 -7.67297670e-02
-6.00745201e-01 6.10596299e-01 -2.79633611e-01 6.54074103e-02
1.22446859e+00 -1.76527992e-01 -3.39487284e-01 -2.47548506e-01
-8.03387403e-01 9.94330198e-02 6.37130797e-01 3.03265303e-01
5.63627779e-01 -1.39669776e+00 -7.36102104e-01 3.29283029e-01
-1.48537550e-02 -1.97288439e-01 2.75659800e-01 9.99622762e-01
-2.07124636e-01 6.06863536e-02 1.78945288e-01 -1.07926250e+00
-1.42132509e+00 6.62581801e-01 1.79675624e-01 1.17385276e-01
-2.76607871e-01 8.36389244e-01 1.23794511e-01 -3.64365697e-01
2.97470212e-01 1.41715169e-01 -1.61099315e-01 7.70809725e-02
7.60136783e-01 3.59445393e-01 3.53297889e-02 -7.28801548e-01
-6.57324672e-01 6.36010110e-01 -1.60061434e-01 2.68789411e-01
1.24484849e+00 -1.73766181e-01 -5.08073509e-01 1.86354309e-01
1.18582344e+00 8.95776004e-02 -1.26362789e+00 -6.97634742e-02
1.29630148e-01 -5.83107710e-01 -5.49338534e-02 -4.45710480e-01
-1.58745277e+00 5.86604118e-01 9.65612948e-01 2.51547247e-01
1.08172786e+00 4.30656314e-01 5.48656464e-01 1.09298162e-01
3.03986758e-01 -9.26412225e-01 5.03924526e-02 8.29991847e-02
3.11555266e-01 -1.29506946e+00 -9.18717608e-02 -7.31545627e-01
-4.64609355e-01 9.28355753e-01 5.96447945e-01 5.74850999e-02
1.37915409e+00 5.40952563e-01 8.65968317e-02 -2.67245680e-01
-6.36304080e-01 -2.60238051e-01 7.62918517e-02 7.96136439e-01
3.17912310e-01 2.53466904e-01 -1.09536685e-02 5.91296554e-01
8.42490420e-02 3.15012932e-02 3.19195181e-01 7.25764275e-01
-4.18822825e-01 -1.28475773e+00 -6.41272366e-01 6.18579924e-01
-5.42675376e-01 8.49324279e-03 -4.28838730e-01 3.80964994e-01
3.46432835e-01 1.26255023e+00 2.66804665e-01 -6.47427857e-01
-3.09463114e-01 9.40428674e-02 3.58494669e-01 -5.96615374e-01
-1.47495046e-01 1.41163200e-01 -4.04168487e-01 -6.14698052e-01
-4.96559650e-01 -7.03343749e-01 -1.34046519e+00 -7.36202955e-01
-2.68707037e-01 3.32511455e-01 4.05403644e-01 6.75148606e-01
7.23006964e-01 3.67642231e-02 9.85377848e-01 -9.24004674e-01
-4.47835833e-01 -8.91640067e-01 -7.90809155e-01 6.03352427e-01
2.81948764e-02 -7.89044738e-01 -7.32791185e-01 2.80987799e-01]
|
[13.463974952697754, 1.0476027727127075]
|
09a95933-ddc6-4633-99f2-32ae5aa58aec
|
needle-tip-force-estimation-by-deep-learning
|
2006.16675
| null |
https://arxiv.org/abs/2006.16675v1
|
https://arxiv.org/pdf/2006.16675v1.pdf
|
Needle tip force estimation by deep learning from raw spectral OCT data
|
Purpose. Needle placement is a challenging problem for applications such as biopsy or brachytherapy. Tip force sensing can provide valuable feedback for needle navigation inside the tissue. For this purpose, fiber-optical sensors can be directly integrated into the needle tip. Optical coherence tomography (OCT) can be used to image tissue. Here, we study how to calibrate OCT to sense forces, e.g. during robotic needle placement. Methods. We investigate whether using raw spectral OCT data without a typical image reconstruction can improve a deep learning-based calibration between optical signal and forces. For this purpose, we consider three different needles with a new, more robust design which are calibrated using convolutional neural networks (CNNs). We compare training the CNNs with the raw OCT signal and the reconstructed depth profiles. Results. We find that using raw data as an input for the largest CNN model outperforms the use of reconstructed data with a mean absolute error of 5.81 mN compared to 8.04 mN. Conclusions. We find that deep learning with raw spectral OCT data can improve learning for the task of force estimation. Our needle design and calibration approach constitute a very accurate fiber-optical sensor for measuring forces at the needle tip.
|
['A. Schlaefer', 'T. Saathoff', 'N. Gessert', 'M. Gromniak']
|
2020-06-30
| null | null | null | null |
['robust-design']
|
['miscellaneous']
|
[ 1.51216358e-01 1.02482907e-01 -7.31439888e-02 -3.10210615e-01
-6.49937987e-01 -6.60096884e-01 -2.38444030e-01 -4.08792309e-02
-6.65769339e-01 7.69412577e-01 -1.83048695e-01 -3.00450593e-01
7.92911798e-02 -5.04119217e-01 -1.11604023e+00 -6.75178230e-01
3.19045037e-01 4.29928780e-01 4.46168967e-02 6.36426881e-02
3.37716490e-01 8.34560990e-01 -5.46936512e-01 1.99042201e-01
7.98603415e-01 1.37157023e+00 5.70289612e-01 6.39030278e-01
3.40562046e-01 2.28553385e-01 -1.21207587e-01 -6.87218010e-02
4.04095560e-01 1.19205266e-01 -7.24023700e-01 -4.25026953e-01
8.36553037e-01 -9.45758343e-01 -3.00121427e-01 1.04758286e+00
8.33979070e-01 -6.04013860e-01 5.60983479e-01 -2.09075704e-01
-3.78575802e-01 7.05850422e-01 -3.61945599e-01 1.89401060e-01
2.69768894e-01 2.83048272e-01 2.12896451e-01 -6.53346717e-01
8.47978890e-01 6.52867734e-01 1.09143996e+00 9.67119694e-01
-1.18949020e+00 -6.39450550e-01 -6.59248769e-01 -1.44459754e-01
-7.92855203e-01 -3.10273439e-01 6.46255374e-01 -8.58496428e-01
7.40997195e-01 1.31680757e-01 1.00690579e+00 1.10618603e+00
1.20276010e+00 3.99357677e-01 8.99463892e-01 -7.75069147e-02
-1.29087493e-01 1.10849068e-01 -9.10655856e-02 5.95226943e-01
4.40008372e-01 3.71718287e-01 9.64424163e-02 1.26287937e-01
1.52656817e+00 9.85433981e-02 -6.68524623e-01 -2.52584785e-01
-1.39935768e+00 5.38907766e-01 1.21652603e+00 7.16620743e-01
-6.39705598e-01 6.90383434e-01 1.77386731e-01 7.98073262e-02
5.28280213e-02 1.30764520e+00 -3.47213328e-01 -3.43704671e-01
-7.31296241e-01 -2.36121323e-02 3.42891127e-01 3.58104914e-01
3.99325907e-01 -1.93725303e-01 -2.47043982e-01 4.64578241e-01
1.20853677e-01 6.63575411e-01 5.19001603e-01 -1.05279124e+00
2.11116955e-01 2.48429537e-01 3.03913653e-01 -8.36059749e-01
-1.15953100e+00 -7.54497051e-01 -7.94906735e-01 1.29193738e-01
6.59186363e-01 -5.03991723e-01 -9.01550174e-01 1.30296993e+00
7.45676979e-02 2.42149442e-01 -3.34116131e-01 1.28140616e+00
6.96699083e-01 -8.77562985e-02 -6.29299045e-01 -1.36685401e-01
9.43286061e-01 -3.37369621e-01 -7.14012444e-01 -2.88678229e-01
6.14427567e-01 -9.07692194e-01 9.25056398e-01 5.34117460e-01
-1.34428549e+00 -4.72576350e-01 -1.09194911e+00 -3.26567539e-03
1.46996938e-02 4.01367009e-01 9.23887014e-01 4.93431330e-01
-1.03806150e+00 1.16066515e+00 -1.36945093e+00 -9.68241468e-02
6.55134439e-01 7.38585651e-01 -2.52220631e-01 9.39336866e-02
-8.64364743e-01 9.20756817e-01 4.47502546e-02 4.92722809e-01
-4.96898532e-01 -1.11105144e+00 -3.09316963e-01 -3.27686697e-01
-2.46166661e-01 -9.27366972e-01 1.30355680e+00 -3.60544324e-01
-1.72108245e+00 9.27193761e-01 2.48523891e-01 -5.94752431e-01
5.66279888e-01 -1.82782918e-01 2.72967935e-01 3.67466807e-01
-2.01338023e-01 7.74693370e-01 6.94297731e-01 -1.02947140e+00
2.39546254e-01 -6.21602535e-01 3.55245098e-02 -4.61784840e-01
-4.02685881e-01 -3.82141888e-01 -1.74110394e-03 -2.13399976e-01
6.55122399e-01 -1.31405473e+00 -2.47807696e-01 4.82574403e-01
-6.30191684e-01 4.25574005e-01 4.10650849e-01 -4.68756318e-01
7.95300722e-01 -1.68932807e+00 4.92858440e-02 2.23149896e-01
5.86174905e-01 1.54363170e-01 -3.43759879e-02 -1.00510322e-01
-2.01179221e-01 -3.00472621e-02 1.23096012e-01 1.64696440e-01
-8.39936197e-01 -1.23227410e-01 2.72717685e-01 5.95626235e-01
-9.75728706e-02 1.38335872e+00 -8.99367511e-01 -1.82597071e-01
2.35059559e-01 5.67620337e-01 -6.74507558e-01 1.54767275e-01
1.83778685e-02 9.97637272e-01 -3.49944711e-01 7.31049776e-01
8.56311142e-01 -5.04142880e-01 7.91146010e-02 -1.06809807e+00
-2.21788753e-02 2.02504680e-01 -5.57943404e-01 2.21479392e+00
-7.15521812e-01 8.03062499e-01 2.33793288e-01 -5.54452181e-01
8.75311971e-01 1.74940452e-01 8.47689211e-01 -6.09956980e-01
6.01723373e-01 7.12203383e-01 5.09644926e-01 -7.82716870e-01
-7.60665610e-02 -5.77094257e-01 4.54508722e-01 2.13740245e-01
-9.68365967e-02 -6.64666712e-01 -4.04997081e-01 -2.40218773e-01
1.09828722e+00 -9.05315652e-02 -3.04308623e-01 -2.34419420e-01
-4.94817458e-03 -1.64984182e-01 2.08371226e-02 6.83693886e-01
-1.44749448e-01 6.29100502e-01 3.62338245e-01 -7.32647836e-01
-1.00591993e+00 -9.47722495e-01 -7.69160390e-01 6.20112233e-02
1.30978763e-01 2.78837442e-01 -6.74952328e-01 -1.60855666e-01
5.69974661e-01 -1.55070171e-01 -8.93423557e-01 -8.79931003e-02
-6.61450982e-01 -5.38211524e-01 1.83427513e-01 8.26421559e-01
2.08240882e-01 -7.36927629e-01 -7.69214332e-01 5.05912900e-01
-1.85890734e-01 -1.34831274e+00 -2.60637224e-01 1.36132449e-01
-1.37122393e+00 -1.08860207e+00 -7.77724206e-01 -3.46544951e-01
5.99943399e-01 -1.34437367e-01 6.89165115e-01 -2.96882242e-01
-4.72662747e-01 1.72508746e-01 5.99179491e-02 -5.39666414e-01
-2.88299173e-01 2.41535276e-01 1.34790629e-01 -2.81209320e-01
2.27182209e-01 -8.43177378e-01 -1.06313896e+00 4.73654494e-02
-5.03117263e-01 1.80397063e-01 9.22659934e-01 5.82920372e-01
4.70837623e-01 -1.01134086e+00 1.61483869e-01 -9.27526951e-01
8.41505527e-01 1.54233441e-01 -6.43300235e-01 -2.26691291e-01
-2.56463319e-01 -4.52119745e-02 4.16513950e-01 -3.98267180e-01
-3.22663248e-01 1.47592276e-01 -3.28039616e-01 -7.81272054e-01
3.72650057e-01 6.42180741e-01 7.38243401e-01 -8.70721638e-01
1.34703112e+00 -2.74263948e-01 5.08445799e-01 -1.16245724e-01
-3.02886724e-01 7.96424568e-01 4.94985431e-01 -3.12790424e-01
3.80068511e-01 7.37904966e-01 3.35512787e-01 -4.05523688e-01
-8.38919163e-01 -3.55695367e-01 -7.61527956e-01 -4.92644072e-01
7.40493655e-01 -5.53188503e-01 -1.27495420e+00 4.30314958e-01
-1.45428085e+00 -4.16620195e-01 4.73527312e-02 1.12990606e+00
-6.35233700e-01 1.35402363e-02 -1.15872872e+00 -2.06136182e-01
-7.40793765e-01 -1.35249186e+00 1.17202556e+00 4.27764319e-02
-7.97004253e-02 -1.04098320e+00 -2.57726461e-02 3.69973689e-01
7.78731167e-01 6.49192214e-01 3.00889403e-01 1.70235321e-01
-6.71600640e-01 -5.63363075e-01 -4.58754092e-01 3.64882529e-01
5.15340090e-01 -1.06260315e-01 -1.15166664e+00 -6.16286159e-01
2.25006759e-01 -3.66954714e-01 7.80098736e-01 1.28352523e+00
1.46579969e+00 1.32866919e-01 -6.47457659e-01 1.01212513e+00
1.63904178e+00 3.18095088e-01 9.56845284e-01 1.26335472e-01
8.51836264e-01 8.37091822e-04 2.43280739e-01 1.78895384e-01
-3.98666441e-01 6.41909897e-01 6.13967776e-01 -2.06441302e-02
-1.08524203e-01 2.70655721e-01 -2.02819616e-01 8.86676908e-01
-5.90145409e-01 3.65478128e-01 -1.09118223e+00 1.50502324e-01
-1.35480523e+00 -4.47044432e-01 -6.77988529e-02 2.12827730e+00
9.91525292e-01 3.26894939e-01 -4.19183284e-01 -1.23635970e-01
2.50569224e-01 -4.78671581e-01 -9.44366634e-01 -1.38965160e-01
3.41038227e-01 5.50426126e-01 8.38732183e-01 6.53480232e-01
-9.46201801e-01 3.49290252e-01 6.40692711e+00 1.27542198e-01
-2.08979940e+00 3.44549306e-02 3.28873813e-01 -1.13861129e-01
-1.78970903e-01 -6.79432154e-01 -6.12868965e-01 5.35216808e-01
5.96579134e-01 3.62295449e-01 3.38375717e-01 3.40225965e-01
4.49833632e-01 -1.33659989e-01 -1.42664003e+00 1.29353118e+00
-1.60369352e-01 -1.92441320e+00 -2.02168554e-01 1.87129378e-01
6.29081786e-01 4.68318760e-01 1.56637520e-01 -5.30778050e-01
-5.60373366e-01 -1.23210311e+00 2.10959552e-04 9.59887922e-01
1.30506361e+00 -7.19968602e-02 1.02919924e+00 1.66381553e-01
-6.02289319e-01 -2.70632654e-02 -5.83900154e-01 6.84681013e-02
-7.77058676e-02 1.17532802e+00 -1.37047005e+00 2.14378804e-01
5.65558195e-01 1.05748963e+00 -1.65581122e-01 1.15844274e+00
3.14574242e-01 1.32233575e-01 -4.03151780e-01 -1.56908676e-01
1.57329500e-01 -7.65875056e-02 3.72458160e-01 9.35614228e-01
4.72499251e-01 -3.18583339e-01 -2.98282921e-01 1.20555592e+00
-2.02857852e-01 -1.86811730e-01 -4.79789674e-01 -6.82881102e-02
2.26042062e-01 1.40587115e+00 -4.78969574e-01 1.16272412e-01
4.05358858e-02 5.71076453e-01 1.14033550e-01 1.23355508e-01
-6.17928028e-01 -3.86954546e-01 5.03197551e-01 7.69196153e-01
-1.20393991e-01 -3.90252888e-01 -5.05529404e-01 -1.11007595e+00
2.26387814e-01 -1.31601810e-01 -5.42097986e-01 -1.29287875e+00
-1.19438684e+00 4.21703935e-01 -7.19815135e-01 -1.28913438e+00
-1.28454804e-01 -1.24451041e+00 -2.81361848e-01 9.62143183e-01
-1.69326532e+00 -8.79991531e-01 -8.72212648e-01 3.86442333e-01
-1.38500720e-01 4.38034177e-01 7.59908020e-01 3.17410767e-01
-1.68969378e-01 3.76425177e-01 1.64016888e-01 4.51776862e-01
8.20405185e-01 -1.16690063e+00 1.62169822e-02 2.27890521e-01
-5.25029480e-01 7.64192224e-01 4.92796302e-01 -5.80061138e-01
-1.74422562e+00 -7.86289394e-01 3.71200681e-01 -4.68458235e-01
5.86519063e-01 -1.11319616e-01 -6.61890090e-01 7.35742867e-01
-6.58321083e-02 3.26390147e-01 3.98827285e-01 -2.30820745e-01
3.05653848e-02 -3.39077413e-01 -1.37761652e+00 3.01321983e-01
7.13797927e-01 -5.43081939e-01 -1.78855643e-01 6.77289546e-01
4.34917420e-01 -1.28463531e+00 -1.20396793e+00 7.12210953e-01
1.17430234e+00 -9.83617246e-01 9.36336696e-01 -3.06161314e-01
7.59200275e-01 1.75468490e-01 3.66251498e-01 -1.52749765e+00
-3.83739829e-01 -5.12028337e-01 2.48005435e-01 -6.39260635e-02
3.93639624e-01 -8.76383126e-01 1.28697932e+00 3.66664767e-01
-3.51683140e-01 -1.23085713e+00 -9.14060295e-01 -5.37983239e-01
3.73356849e-01 -2.75339305e-01 2.11952299e-01 7.71780252e-01
1.38014570e-01 3.93760279e-02 2.11316004e-01 -7.96415880e-02
4.37520355e-01 4.40798737e-02 2.77285188e-01 -1.37869096e+00
-2.04986155e-01 -2.18362764e-01 -6.58952534e-01 -1.34718037e+00
-3.24662507e-01 -9.58663940e-01 -1.15954667e-01 -1.67597294e+00
7.26632029e-02 -5.96081197e-01 -4.91297580e-02 1.84313267e-01
2.26031005e-01 3.17661643e-01 1.34693533e-02 1.47624314e-01
2.76938260e-01 -1.99795872e-01 2.37421441e+00 -1.56461358e-01
-3.25957648e-02 1.65308297e-01 -3.22200119e-01 4.43780810e-01
8.84736121e-01 -1.22839965e-01 -2.31835060e-02 -6.04953825e-01
6.44116759e-01 4.83531415e-01 4.97012317e-01 -1.36083531e+00
4.10141140e-01 1.69543698e-01 7.00245023e-01 -2.82971025e-01
4.54594374e-01 -9.31649625e-01 -1.65567212e-02 1.06777036e+00
-6.57792613e-02 -3.89331877e-01 4.45417106e-01 1.12567283e-01
-5.12032546e-02 -1.74718916e-01 1.04069436e+00 -3.68603528e-01
2.06495211e-01 3.43027681e-01 -1.16285264e-01 -2.91803062e-01
5.77344179e-01 -5.34032106e-01 -3.22431237e-01 -2.19084784e-01
-1.04417467e+00 -1.26158312e-01 1.43488780e-01 -1.51295677e-01
8.92583370e-01 -1.21469021e+00 -5.49478292e-01 3.66884947e-01
-1.82212994e-01 3.05966586e-02 2.40950301e-01 1.41493714e+00
-1.02328110e+00 6.72834754e-01 -5.98575175e-01 -1.16022789e+00
-8.33498240e-01 9.06022936e-02 1.26719820e+00 -7.53582865e-02
-1.53625682e-01 1.17327678e+00 -2.49954104e-01 -4.60572481e-01
-5.84319569e-02 -1.19516551e+00 -3.43324952e-02 -2.52536148e-01
8.60726610e-02 -1.57082841e-01 5.31956494e-01 2.38193020e-01
-1.86024532e-01 1.29894412e+00 -1.06228083e-01 4.66046244e-01
1.44139266e+00 1.84863612e-01 -1.86811373e-01 2.97374040e-01
1.37954664e+00 -8.39132816e-02 -1.18163431e+00 9.43469349e-03
-5.43871939e-01 -6.06129467e-01 4.30575162e-01 -9.77001011e-01
-1.43741739e+00 1.12264884e+00 9.18692350e-01 6.03431202e-02
8.34085286e-01 -9.87015739e-02 1.02133286e+00 4.52635199e-01
4.11143333e-01 -7.82128096e-01 2.31047958e-01 2.95149714e-01
1.14189208e+00 -1.27249706e+00 -1.59508377e-01 -8.41364384e-01
1.20099917e-01 1.56689644e+00 6.34363472e-01 -2.48593777e-01
9.21497285e-01 6.84733629e-01 2.58448720e-01 -3.50105107e-01
-2.18271866e-01 2.96607494e-01 2.19489321e-01 3.65406662e-01
6.93875015e-01 1.97596267e-01 -1.19245067e-01 1.23043500e-01
-2.59100437e-01 8.31435680e-01 7.64647484e-01 8.16507876e-01
-3.96943599e-01 -7.52177417e-01 -6.71674162e-02 1.07860291e+00
-4.49607313e-01 -1.52879199e-02 3.58154997e-02 3.53741288e-01
1.44366473e-01 2.19736084e-01 2.58497536e-01 -4.54445958e-01
5.68861067e-01 -4.44629848e-01 1.28393292e+00 -6.66866481e-01
-6.54695570e-01 1.68834120e-01 -3.30544144e-01 -8.92932236e-01
-5.57400405e-01 -2.58245498e-01 -1.13995779e+00 -1.86658338e-01
-4.58688349e-01 -2.79779911e-01 1.03800261e+00 5.40921867e-01
2.36937329e-01 7.22274601e-01 4.59879071e-01 -1.23870361e+00
-4.98621583e-01 -1.13338649e+00 -6.16889894e-01 1.35638878e-01
6.49433732e-01 -6.53777122e-01 -3.60395640e-01 -2.31331572e-01]
|
[13.861222267150879, -3.043362617492676]
|
d033a500-02c7-45d7-b787-726f0d36e957
|
cuffless-blood-pressure-estimation-from
| null | null |
https://arxiv.org/abs/1811.02214v1
|
https://arxiv.org/abs/1811.02214v1
|
Cuffless Blood Pressure Estimation from Electrocardiogram and Photoplethysmogram Using Waveform Based ANN-LSTM Network
|
Goal: Although photoplethysmogram (PPG) and electrocardiogram (ECG) signals can be used to estimate blood
pressure (BP) by extracting various features, the changes in morphological contours of both PPG and ECG signals due
to various diseases of circulatory system and interaction of other physiological systems make the extraction of such features
very difficult. Methods: In this work, we propose a waveformbased hierarchical Artificial Neural Network–Long Short Term
Memory (ANN-LSTM) model for BP estimation. The model consists of two hierarchy levels, where the lower hierarchy level
uses ANNs to extract necessary morphological features from ECG and PPG waveforms and the upper hierarchy level uses
LSTM layers to account for the time domain variation of the features extracted by lower hierarchy level. Results: The proposed model is evaluated on 39 subjects using the Association for the Advancement of Medical Instrumentations (AAMI) standard and the British Hypertension Society (BHS) standard. The method satisfies both the standards in the estimation of systolic blood pressure (SBP) and diastolic blood pressure (DBP). For the proposed network, the mean absolute error (MAE) and the root mean square error (RMSE) for SBP estimation are 1.10 and 1.56 mmHg, respectively, and for DBP estimation are 0.58 and 0.85 mmHg, respectively. Conclusion: The performance of the proposed hierarchical ANN-LSTM model is found to be better than the other feature engineering-based networks. It is shown that the proposed model is able to automatically extract the necessary features and their time domain variations to estimate BP reliably in a noninvasive continuous manner. Significance: The method is expected to greatly facilitate the presently available mobile health-care gadgets in continuous BP estimation.
|
['Md. Sayed Tanveer and Md. Kamrul Hasan∗']
|
2018-11-06
| null | null | null |
journal-2018-11
|
['blood-pressure-estimation']
|
['medical']
|
[ 1.95999011e-01 -6.54579978e-03 2.21819356e-01 -4.92504627e-01
-4.56052236e-02 -7.79505761e-04 -3.58828813e-01 1.85287386e-01
-3.51202905e-01 1.13496971e+00 -1.89715058e-01 -4.14380252e-01
-1.79688036e-01 -7.57464647e-01 -1.41764209e-01 -6.38635099e-01
-5.16076744e-01 -4.97472696e-02 -1.40957788e-01 2.05803700e-02
1.83088109e-01 6.85118139e-01 -1.13436437e+00 4.20316234e-02
1.19067705e+00 1.38294387e+00 -4.09939945e-01 9.21746910e-01
9.70723480e-03 4.73331243e-01 -6.77048326e-01 5.37711233e-02
2.38245353e-01 -5.76538146e-01 -3.62135172e-01 -3.26301277e-01
4.04938728e-01 -3.36087495e-01 -1.43088758e-01 8.28833878e-01
1.09061074e+00 -5.59359975e-02 4.62730795e-01 -7.50777423e-01
-3.55869591e-01 3.90545785e-01 -4.68909651e-01 3.66297245e-01
-1.63400158e-01 -1.50978481e-02 7.37389699e-02 -6.70264125e-01
-9.73302126e-02 6.44778371e-01 1.10096323e+00 3.63417387e-01
-1.22787154e+00 -4.81025219e-01 -3.76733720e-01 2.17851937e-01
-1.44970405e+00 -4.18115109e-01 7.02963650e-01 -5.87361574e-01
6.73952460e-01 4.86117899e-01 9.32476580e-01 4.13746417e-01
8.05350780e-01 -4.16681580e-02 1.49457777e+00 -5.18813193e-01
1.13744691e-01 6.57756925e-01 6.11776173e-01 6.45687759e-01
4.67450589e-01 1.27292812e-01 2.75427159e-02 -3.28384042e-01
1.03510416e+00 -9.99350026e-02 -3.30592304e-01 2.34030038e-01
-5.89394629e-01 5.38559020e-01 1.61350563e-01 6.65394664e-01
-8.21606338e-01 -1.28112733e-01 5.02536774e-01 2.99373835e-01
1.76977381e-01 4.02522355e-01 -5.08658171e-01 -2.77730078e-01
-1.12327826e+00 4.13869135e-02 1.07745719e+00 6.17389143e-01
1.77215353e-01 4.38378572e-01 -2.97043234e-01 8.98448586e-01
3.93460393e-01 5.55077076e-01 5.06408095e-01 -9.28688884e-01
6.70324787e-02 4.63650227e-01 3.87180269e-01 -1.38698912e+00
-8.90164495e-01 -4.82044280e-01 -1.17518306e+00 6.75758645e-02
5.72250903e-01 -6.65740073e-01 -8.83626521e-01 1.42175961e+00
3.56371664e-02 1.11091495e-01 -1.45313963e-01 6.23478591e-01
1.18456590e+00 3.89361084e-01 2.79358238e-01 -5.00410497e-01
1.40633821e+00 -2.53482908e-01 -9.88310874e-01 5.41692041e-02
9.10263062e-02 -2.41196513e-01 5.88051617e-01 2.24560902e-01
-1.16492796e+00 -9.41466570e-01 -1.10766244e+00 2.96576828e-01
-3.03116620e-01 2.49202490e-01 1.20002083e-01 1.02070057e+00
-8.66773307e-01 9.89860773e-01 -7.43050098e-01 -2.26202637e-01
2.16525167e-01 4.08142358e-01 1.38341174e-01 6.87989354e-01
-1.74071872e+00 1.20160878e+00 9.71012339e-02 1.00052810e+00
1.97521448e-01 -7.59906948e-01 -6.99039698e-01 1.24441855e-01
-5.18395603e-01 -6.49598956e-01 6.28786027e-01 -3.80677164e-01
-1.73265266e+00 5.19558132e-01 -1.68703616e-01 -5.49513102e-01
3.69443238e-01 -1.44320950e-01 -6.36372983e-01 2.83487976e-01
-4.94693339e-01 1.10804223e-01 7.05841601e-01 -7.09617555e-01
-2.48338684e-01 -6.79275453e-01 -6.32266462e-01 -1.28935307e-01
-2.50994861e-01 -8.74700472e-02 2.45774880e-01 -1.42413005e-01
2.68549651e-01 -5.91660976e-01 -4.21974748e-01 -3.07755079e-02
-2.08327502e-01 2.29034781e-01 4.40151066e-01 -1.29398215e+00
1.45770669e+00 -1.74458265e+00 -3.66899431e-01 6.27823830e-01
4.86700654e-01 6.84425056e-01 5.68872511e-01 1.64036989e-01
2.16229986e-02 1.31657809e-01 -2.14781210e-01 2.90291965e-01
-2.35544637e-01 8.61302838e-02 1.13747800e-02 4.73313600e-01
-8.46986547e-02 8.68830144e-01 -2.60921001e-01 -4.70548362e-01
6.15630686e-01 6.73421919e-01 2.02741921e-01 -2.93340292e-02
5.77873647e-01 6.15469694e-01 -2.97163278e-01 3.40759158e-01
7.67903209e-01 1.16905950e-01 1.54257715e-01 -5.78185380e-01
-3.97851050e-01 -1.05497040e-01 -1.22930181e+00 8.64523828e-01
-2.62947559e-01 7.31892526e-01 -2.02404812e-01 -9.47875559e-01
1.56799304e+00 8.04289162e-01 5.05073488e-01 -7.56243944e-01
3.92345846e-01 3.69887322e-01 5.84066749e-01 -1.01134670e+00
-2.02698320e-01 -4.39085901e-01 4.84083235e-01 9.19590592e-02
-2.10124239e-01 2.10155994e-01 6.48211911e-02 -4.17591691e-01
4.22826171e-01 -3.63596268e-02 6.06394529e-01 -5.74837506e-01
9.92319107e-01 -5.91197431e-01 5.52036703e-01 8.33688438e-01
-6.86510146e-01 1.17338918e-01 3.50368857e-01 -8.74662519e-01
-9.14579153e-01 -6.86249733e-01 -7.75918245e-01 1.95696458e-01
-2.35249609e-01 1.87407747e-01 -4.57436174e-01 1.99955896e-01
2.86638528e-01 4.41947192e-01 -5.26562393e-01 -1.04894944e-01
-7.17233896e-01 -8.61372948e-01 7.47403085e-01 6.93440795e-01
6.79762900e-01 -1.13073671e+00 -1.04392600e+00 6.36624455e-01
7.18215480e-02 -1.07754600e+00 2.76524127e-01 -1.28693655e-01
-1.29975569e+00 -7.04189122e-01 -8.09053659e-01 -4.68463272e-01
3.16630334e-01 -8.95468593e-01 8.09789062e-01 -1.72058165e-01
-6.12119794e-01 9.80397016e-02 1.50530711e-01 -6.86306655e-01
-2.97783375e-01 -2.97397912e-01 1.38622686e-01 1.11251675e-01
6.39891267e-01 -1.12599325e+00 -8.76746655e-01 1.39567971e-01
-1.05400123e-02 -2.96740144e-01 4.17147160e-01 3.19757581e-01
2.89795458e-01 -2.10998401e-01 1.07026160e+00 -7.06329703e-01
1.00294948e+00 1.17188245e-02 -5.38845897e-01 1.11301860e-03
-9.32109296e-01 -4.69277978e-01 7.72204280e-01 -2.08981723e-01
-7.22986639e-01 -3.21297884e-01 -2.36545578e-01 5.93200587e-02
-3.44620198e-01 5.35339415e-01 2.93356031e-01 -2.66889483e-01
8.65295231e-01 4.91618991e-01 1.46999896e-01 -4.08971786e-01
-2.42653131e-01 9.68359292e-01 6.48200214e-01 -5.49492478e-01
2.82714486e-01 -9.41455141e-02 4.03543651e-01 -1.30422020e+00
-4.47455257e-01 -5.23951426e-02 -8.35057437e-01 -4.14488018e-01
8.23650002e-01 -4.83476311e-01 -1.15090024e+00 6.29383326e-01
-8.88128757e-01 4.89130616e-02 -1.08399309e-01 6.54826164e-01
-2.58712888e-01 4.68823999e-01 -9.30549562e-01 -1.33215761e+00
-1.18426073e+00 -6.10580385e-01 2.57278848e-02 4.78864253e-01
-3.45196217e-01 -1.18968093e+00 -1.88123673e-01 -3.02822739e-02
1.03589904e+00 1.04026604e+00 9.92236912e-01 -4.11587268e-01
3.43095958e-01 -5.96386790e-01 -1.87277913e-01 7.17884064e-01
3.33912373e-01 1.99055895e-02 -9.43156958e-01 1.38125243e-02
7.24153519e-01 2.10547626e-01 4.04230714e-01 1.08526325e+00
8.95886540e-01 -3.00057918e-01 -2.95643006e-02 3.23360413e-01
1.53845608e+00 8.50397885e-01 1.07075453e+00 9.42010581e-02
4.59637195e-01 3.51730436e-01 -9.90682393e-02 5.09893894e-01
1.47841081e-01 2.01395467e-01 -1.88363135e-01 -4.53645200e-01
3.19865048e-01 3.21397990e-01 -1.16122000e-01 7.57765293e-01
-6.69057071e-01 5.34272850e-01 -8.88275564e-01 9.37828645e-02
-1.50276971e+00 -8.34929347e-01 -6.39939070e-01 2.41532421e+00
1.00808203e+00 1.87795386e-01 2.06526607e-01 4.02565747e-01
7.90228188e-01 -3.09208542e-01 -4.82454956e-01 -9.57066774e-01
3.14468205e-01 6.05742991e-01 4.42578346e-01 5.25805414e-01
-9.93128598e-01 4.63495627e-02 5.94022036e+00 -3.15492392e-01
-1.42736924e+00 -3.19760978e-01 6.51978970e-01 4.34886456e-01
6.08050823e-01 -4.33515251e-01 -6.86195433e-01 7.35815883e-01
1.44043887e+00 -3.97599578e-01 -4.26069722e-02 5.06559908e-01
6.33515298e-01 -9.58992839e-02 -8.31898689e-01 1.04774630e+00
-2.24711597e-01 -1.01305127e+00 -3.97415906e-01 -1.95808902e-01
1.86023816e-01 -3.97698730e-01 -1.56021044e-01 2.43190020e-01
-8.12339842e-01 -1.07848394e+00 -1.05401583e-01 1.12380219e+00
7.32245445e-01 -4.07390535e-01 1.19145656e+00 1.77799538e-01
-1.02492034e+00 -1.07634701e-01 -4.88681555e-01 -4.41843987e-01
2.23692432e-01 9.27455068e-01 -5.43477237e-01 3.70578915e-01
4.42604274e-01 3.57529074e-01 -3.98185909e-01 1.45807767e+00
8.17892477e-02 6.83615863e-01 -3.56889099e-01 -1.10670634e-01
-5.02196625e-02 -3.94560963e-01 4.82020468e-01 1.13955915e+00
1.57830492e-01 3.29699725e-01 -2.30214987e-02 1.18270230e+00
4.66050297e-01 2.85983711e-01 -2.16585681e-01 1.88534528e-01
4.78216171e-01 1.04357028e+00 -2.64277101e-01 -6.73698604e-01
-2.64050514e-01 3.15695316e-01 -3.69928837e-01 5.20861506e-01
-8.47169757e-01 -1.02260149e+00 5.88818975e-02 3.25036407e-01
-1.70285329e-01 1.41421899e-01 -9.62631583e-01 -7.16290176e-01
2.07650468e-01 -5.43857336e-01 3.58278960e-01 -2.94291228e-01
-1.08564007e+00 7.23615229e-01 -8.22730884e-02 -1.00540388e+00
-1.44613609e-01 -3.96855831e-01 -6.92210555e-01 1.64710724e+00
-1.38371110e+00 -6.12223744e-01 -5.16863704e-01 2.72989839e-01
-8.18734930e-04 1.11875124e-01 1.12163281e+00 4.90853727e-01
-5.91273606e-01 3.83317232e-01 -2.15443313e-01 2.91267514e-01
3.06011826e-01 -1.33039653e+00 7.38852844e-02 6.17373705e-01
-8.41237307e-01 9.47893798e-01 7.63335466e-01 -4.94403452e-01
-8.27534854e-01 -9.52755630e-01 1.19401538e+00 5.79477809e-02
1.97225213e-01 4.11701389e-02 -8.83960187e-01 3.98438036e-01
-5.40729649e-02 1.48488790e-01 8.42505395e-01 -1.81584954e-01
2.26674467e-01 -4.37081754e-01 -1.40588009e+00 2.28547573e-01
-9.94112622e-03 -1.46654278e-01 -7.41565049e-01 -9.13558975e-02
-1.72226764e-02 -5.58221757e-01 -1.75955355e+00 7.91729331e-01
1.25016773e+00 -9.17372882e-01 8.15858006e-01 -5.00554979e-01
2.37932861e-01 -1.83197364e-01 2.62721121e-01 -9.13035393e-01
-3.72938156e-01 -5.12470305e-01 -3.37155998e-01 8.65803957e-01
5.20225406e-01 -1.21949422e+00 5.35606682e-01 1.41770875e+00
1.12284357e-02 -1.03344929e+00 -7.40813315e-01 -5.45564234e-01
-9.87006798e-02 5.16296104e-02 1.23650476e-01 7.69719899e-01
3.20889711e-01 3.46909344e-01 -6.39852107e-01 1.31952083e-02
7.94763207e-01 4.93449494e-02 2.85607994e-01 -1.64744365e+00
-6.19619936e-02 -1.58482566e-01 -6.46976471e-01 -4.78246778e-01
-5.95200717e-01 -3.07681143e-01 -2.57459998e-01 -1.77279079e+00
-2.03896180e-01 -2.34121427e-01 -8.62094402e-01 4.44811374e-01
-1.50260776e-01 1.31577671e-01 -1.13327734e-01 -2.35894307e-01
5.05370319e-01 1.21736914e-01 1.10254192e+00 3.03362161e-01
-1.13580477e+00 4.79763001e-01 -4.93429273e-01 6.77672207e-01
1.14359725e+00 -3.66257608e-01 -1.04954824e-01 2.25373864e-01
-1.02105886e-01 4.91221070e-01 2.10871965e-01 -1.21919239e+00
4.12129350e-02 2.18803450e-01 1.07956421e+00 -4.74639803e-01
1.60931036e-01 -8.26605797e-01 1.45294413e-01 9.80690300e-01
-2.45821759e-01 -2.11824581e-01 4.57386225e-01 -4.30575944e-02
-1.26127824e-01 -1.66606739e-01 1.09527421e+00 -3.66659999e-01
-7.21012149e-03 1.14733063e-01 -5.53922415e-01 -2.65602767e-01
6.23042285e-01 -6.81518912e-01 -1.30175427e-01 -3.93446594e-01
-1.26162708e+00 -4.55981493e-02 -4.51319933e-01 -6.38804585e-02
6.92202389e-01 -1.15759456e+00 -8.30297589e-01 4.33312565e-01
-3.87149215e-01 -4.51212615e-01 5.41291654e-01 1.53726447e+00
-7.23480999e-01 5.54351509e-01 -5.32677650e-01 -5.21639943e-01
-1.30507803e+00 -2.89899064e-04 9.71723974e-01 1.83110368e-02
-8.54301929e-01 4.99313533e-01 -4.76886541e-01 -2.82780151e-03
2.96936572e-01 -5.40720701e-01 -6.09242320e-01 -1.83031812e-01
7.49936700e-01 8.20817888e-01 1.42271481e-02 -2.14764893e-01
-2.30897143e-01 6.07489288e-01 2.18683347e-01 2.86987633e-01
1.29804802e+00 -1.64658204e-01 -4.16213632e-01 6.73039436e-01
8.27428579e-01 -2.97491580e-01 -4.55820918e-01 -5.77186234e-02
-2.85483673e-02 -2.32598186e-01 -5.05993888e-02 -1.06802499e+00
-8.26705039e-01 1.09569240e+00 1.19129443e+00 3.95382643e-01
1.15395117e+00 -8.77928436e-01 9.80153799e-01 2.46989325e-01
-8.47216398e-02 -1.14193714e+00 -7.39859760e-01 1.52180672e-01
8.45802963e-01 -4.17554468e-01 1.60933733e-01 -2.06161588e-01
-3.91807348e-01 1.59708333e+00 6.02186322e-01 -1.73774913e-01
9.27586794e-01 1.64050996e-01 3.60968053e-01 -8.57136920e-02
-2.09152132e-01 2.45938823e-01 4.51371551e-01 5.18256724e-01
7.12556779e-01 1.08600982e-01 -9.76599693e-01 6.74181521e-01
-8.57749805e-02 6.01818502e-01 5.54962516e-01 7.01768577e-01
-8.37085366e-01 -5.59272349e-01 -3.51415217e-01 8.56697559e-01
-6.83701515e-01 -1.25980243e-01 1.59016927e-03 6.54082835e-01
8.84277821e-02 9.73974705e-01 -1.27359778e-01 -1.47546500e-01
5.93394041e-01 6.16668522e-01 5.66887975e-01 -1.62555456e-01
-5.30216515e-01 1.32475704e-01 1.30287990e-01 -2.21274182e-01
-3.46743405e-01 -1.62475601e-01 -1.20617366e+00 1.08740535e-02
-1.98758736e-01 1.93171486e-01 7.68362939e-01 9.11785483e-01
3.50700498e-01 5.66701591e-01 5.07973671e-01 -5.01558423e-01
-6.12061441e-01 -1.38731527e+00 -8.98520052e-01 9.74406302e-02
3.61175448e-01 -3.93691272e-01 -2.35925108e-01 1.19091697e-01]
|
[14.084672927856445, 2.9885177612304688]
|
9eddf2c7-7e06-4f8a-a030-84bc5b3722e2
|
a-robust-stereo-camera-localization-method
|
1912.05023
| null |
https://arxiv.org/abs/1912.05023v1
|
https://arxiv.org/pdf/1912.05023v1.pdf
|
A Robust Stereo Camera Localization Method with Prior LiDAR Map Constrains
|
In complex environments, low-cost and robust localization is a challenging problem. For example, in a GPSdenied environment, LiDAR can provide accurate position information, but the cost is high. In general, visual SLAM based localization methods become unreliable when the sunlight changes greatly. Therefore, inexpensive and reliable methods are required. In this paper, we propose a stereo visual localization method based on the prior LiDAR map. Different from the conventional visual localization system, we design a novel visual optimization model by matching planar information between the LiDAR map and visual image. Bundle adjustment is built by using coplanarity constraints. To solve the optimization problem, we use a graph-based optimization algorithm and a local window optimization method. Finally, we estimate a full six degrees of freedom (DOF) pose without scale drift. To validate the efficiency, the proposed method has been tested on the KITTI dataset. The results show that our method is more robust and accurate than the state-of-art ORB-SLAM2.
|
['Cheng-Zhong Xu', 'Lujia Wang', 'Dong Han', 'Zuhao Zou']
|
2019-12-02
| null | null | null | null |
['camera-localization']
|
['computer-vision']
|
[-1.55056417e-01 -6.49459839e-01 -1.25283793e-01 -4.89961594e-01
-4.72996801e-01 -5.16576111e-01 3.95341694e-01 9.86606553e-02
-5.79174280e-01 8.07476997e-01 -4.23512459e-01 -1.64785177e-01
-9.20795426e-02 -7.56416082e-01 -7.49314129e-01 -5.71943462e-01
2.39269182e-01 5.98282337e-01 3.32291037e-01 -3.44661586e-02
4.33134049e-01 8.09810400e-01 -1.38208759e+00 -9.04040337e-01
1.11652732e+00 8.52789104e-01 6.08785629e-01 1.16191663e-01
-4.27358598e-02 -9.31057036e-02 -3.68383348e-01 6.34225979e-02
3.04974526e-01 -4.10325974e-02 -1.52650252e-01 2.38898441e-01
6.40995443e-01 -2.28596076e-01 -1.25675753e-01 1.06022465e+00
5.50765812e-01 1.71542138e-01 3.55767429e-01 -1.29820943e+00
-8.20664763e-02 -1.27224952e-01 -8.82209122e-01 -4.96284097e-01
5.14169335e-01 -1.53431147e-01 5.97439408e-01 -1.06237471e+00
7.71658719e-01 1.01608062e+00 8.17534447e-01 -1.68783024e-01
-1.40036535e+00 -7.02585816e-01 7.32111186e-02 2.05545753e-01
-1.96723473e+00 -3.67414415e-01 7.96896398e-01 -3.61513942e-01
5.23924589e-01 8.55650008e-02 7.66183913e-01 6.81409299e-01
4.80333686e-01 2.31848061e-02 1.12129247e+00 -3.65412056e-01
1.38936579e-01 1.09664872e-01 -3.07947040e-01 7.62747884e-01
8.14039111e-01 7.20918039e-03 -6.71902120e-01 -1.30990565e-01
7.50627875e-01 4.54059035e-01 -3.50167900e-01 -1.23293400e+00
-1.39670074e+00 8.02328587e-01 7.99484432e-01 -9.38757882e-02
-2.10028619e-01 2.32655287e-01 -2.03933194e-01 -9.31916535e-02
2.15941533e-01 1.21805377e-01 1.35060355e-01 4.85750102e-02
-9.73737538e-01 9.77742299e-02 6.19343460e-01 1.19260895e+00
1.25082958e+00 -3.76155935e-02 5.40803194e-01 5.34224629e-01
8.15793574e-01 1.07508445e+00 1.09492444e-01 -7.94159830e-01
4.34001952e-01 5.94483972e-01 3.27726871e-01 -1.41183865e+00
-4.18984562e-01 -3.21924329e-01 -8.07177126e-01 2.89048105e-01
1.71857048e-02 8.60918462e-02 -7.60213554e-01 1.17572653e+00
5.47015846e-01 2.15485021e-02 -1.11002661e-01 1.14343548e+00
5.26797593e-01 4.81177270e-01 -5.13216317e-01 -2.45206758e-01
9.14696038e-01 -6.37899518e-01 -7.88215339e-01 -6.26338959e-01
2.07451701e-01 -1.11952078e+00 7.01343656e-01 1.84490368e-01
-3.97582918e-01 -4.54916388e-01 -1.45715213e+00 -9.47205871e-02
-9.65320840e-02 2.47953176e-01 5.58699131e-01 2.81714678e-01
-8.90625894e-01 2.84619927e-01 -8.96561027e-01 -8.35153580e-01
-4.12814528e-01 3.23884636e-01 -6.34975016e-01 -3.30652148e-01
-7.98026502e-01 1.06644857e+00 3.29326063e-01 3.91616970e-01
-5.19412756e-01 -6.51694536e-02 -1.13141906e+00 -3.04407328e-01
4.04882967e-01 -6.39147162e-01 8.37053299e-01 -1.24674410e-01
-1.49570954e+00 6.94846034e-01 -5.62003314e-01 -2.39405409e-01
5.86117148e-01 -2.03755289e-01 8.18785504e-02 -6.82158619e-02
3.80233020e-01 4.09428805e-01 5.76103210e-01 -1.49013042e+00
-4.00783509e-01 -4.94708985e-01 -2.68447071e-01 3.73357415e-01
1.91695258e-01 -5.24541855e-01 -6.72897816e-01 -1.22404590e-01
1.09186924e+00 -1.10857666e+00 -2.68117249e-01 2.18822554e-01
-2.79899329e-01 2.51371205e-01 8.31145465e-01 -4.36515361e-01
8.44913065e-01 -2.20807433e+00 2.39956900e-02 4.76753086e-01
1.79888848e-02 -3.04150105e-01 2.49355793e-01 5.45433164e-01
5.67835391e-01 -1.80203423e-01 -1.68706611e-01 -5.71697295e-01
-9.58801806e-02 5.07899821e-01 -9.01966169e-02 1.03213620e+00
-3.22853357e-01 4.67023879e-01 -6.98549628e-01 -6.18005157e-01
5.24072230e-01 5.26331782e-01 -2.27592602e-01 1.45465776e-01
1.94658518e-01 6.61997437e-01 -4.03139472e-01 1.00764000e+00
1.11421847e+00 2.85998620e-02 1.43075690e-01 -1.56418234e-01
-6.90167844e-01 7.87844062e-02 -1.53488123e+00 2.16731501e+00
-5.27603149e-01 6.16022944e-01 3.75487059e-01 -5.52592218e-01
1.39833593e+00 -2.20617130e-01 2.74691850e-01 -6.14334643e-01
1.38766235e-02 5.13452709e-01 -3.44035000e-01 -1.74072787e-01
7.27239311e-01 -9.52937733e-03 -1.08582620e-02 -5.29160462e-02
-2.56778181e-01 -6.10530794e-01 -6.52658939e-02 -5.37507832e-02
6.87523067e-01 5.24285257e-01 5.78571916e-01 -1.75267562e-01
5.11160493e-01 2.25394338e-01 9.38324451e-01 4.88044202e-01
3.32063511e-02 5.95132947e-01 -1.41112566e-01 -2.55493760e-01
-8.11734200e-01 -1.17447615e+00 -3.14560086e-01 -4.46833186e-02
9.86387372e-01 -4.83617425e-01 -1.18257202e-01 -1.71447709e-01
5.43169379e-01 1.31234616e-01 1.29137903e-01 1.91004425e-01
-4.66871738e-01 -3.59876126e-01 -9.76509452e-02 1.01099111e-01
7.30379164e-01 -3.11230272e-01 -6.14335537e-01 2.47131959e-01
-2.89017409e-01 -1.22402608e+00 -1.89612761e-01 -1.39292732e-01
-9.66463268e-01 -9.90444124e-01 -4.26592827e-01 -6.49911761e-01
7.88210213e-01 1.01810133e+00 7.18783677e-01 1.24430090e-01
-1.56719178e-01 1.63977787e-01 -1.83882609e-01 -2.80129820e-01
2.79854685e-01 8.72772485e-02 4.02145118e-01 -1.03759514e-02
1.08797051e-01 -5.86213052e-01 -5.09140313e-01 7.36705422e-01
-2.47052938e-01 1.17996428e-02 6.68519735e-01 6.67331517e-01
1.12712395e+00 -2.38297150e-01 -9.20701772e-02 -2.06679657e-01
6.67962208e-02 -6.42153621e-02 -1.36573684e+00 -4.52365503e-02
-8.11507523e-01 -3.10822334e-02 1.64961934e-01 -1.96583018e-01
-7.48473167e-01 7.70183623e-01 1.65568024e-01 -3.45986843e-01
8.05551708e-02 6.58015788e-01 -2.84010112e-01 -8.73261988e-01
3.83484274e-01 2.38587633e-01 1.35148093e-01 -6.36532009e-01
2.33960003e-01 8.22234213e-01 5.52910149e-01 -3.44860196e-01
1.22146928e+00 7.60738254e-01 5.80481470e-01 -1.10934412e+00
-2.80706406e-01 -8.19799542e-01 -9.13929999e-01 -2.96013594e-01
6.18824780e-01 -1.06671453e+00 -7.59429634e-01 2.77526587e-01
-1.25247145e+00 1.58669204e-01 2.80459195e-01 9.91458714e-01
-4.62944269e-01 7.65182674e-01 1.30361347e-02 -8.60143363e-01
1.50792198e-02 -1.35809171e+00 1.19557631e+00 3.01414073e-01
1.61078334e-01 -6.24027908e-01 2.74757743e-01 1.11769952e-01
2.23252624e-01 6.63500190e-01 7.43821412e-02 2.33584568e-01
-1.22932994e+00 -3.76855820e-01 -2.72788018e-01 -2.62190044e-01
1.65746450e-01 5.57693429e-02 -5.30585408e-01 -4.88106042e-01
5.36928028e-02 1.43373027e-01 5.27269304e-01 1.50655955e-01
4.50198025e-01 1.34206831e-01 -6.90063953e-01 1.06402004e+00
1.82925057e+00 3.61020789e-02 3.79663795e-01 7.36410379e-01
7.33185589e-01 2.96272039e-01 1.18148112e+00 2.87017792e-01
6.86985970e-01 1.04086781e+00 7.85585284e-01 -3.19756344e-02
3.20175916e-01 -4.71312612e-01 1.48821160e-01 7.66197145e-01
-5.83762452e-02 1.08206533e-01 -1.01201963e+00 3.43780279e-01
-2.13862801e+00 -4.86202270e-01 -5.23344994e-01 2.61615825e+00
3.40303928e-01 -9.32679772e-02 -4.58495557e-01 -1.32141903e-01
6.60344958e-01 2.08723918e-01 -2.96828479e-01 6.09696284e-03
-5.59017202e-03 -3.27737212e-01 1.11796951e+00 9.09139037e-01
-9.00759101e-01 1.01737094e+00 5.83601284e+00 2.86731541e-01
-1.27886331e+00 8.82235989e-02 -6.02333426e-01 1.80873454e-01
-1.71592906e-01 5.78970730e-01 -1.04286313e+00 3.37473273e-01
4.56143230e-01 -1.15109138e-01 3.13934356e-01 9.98643637e-01
3.25123072e-01 -6.16467357e-01 -7.15902328e-01 1.35711217e+00
3.04171711e-01 -1.08713484e+00 -3.91770005e-01 4.11012173e-01
4.49652672e-01 1.77946672e-01 -4.32837695e-01 -1.65366024e-01
-8.71390626e-02 -5.31445026e-01 7.38523424e-01 6.01233482e-01
6.90265059e-01 -7.07742274e-01 8.09485257e-01 5.84970772e-01
-1.52334642e+00 3.47505778e-01 -6.70820653e-01 -2.14271680e-01
5.40174186e-01 8.58388186e-01 -8.74144912e-01 8.98586333e-01
6.19766235e-01 7.71906257e-01 -6.40852511e-01 1.62872219e+00
-6.86638236e-01 -1.20998643e-01 -6.47172153e-01 -1.37202423e-02
-5.60591072e-02 -7.33351231e-01 7.19281495e-01 7.04043567e-01
7.98912883e-01 -3.14891487e-01 5.71678221e-01 7.55453587e-01
1.99606135e-01 1.55957326e-01 -1.10909617e+00 3.66883039e-01
5.75457931e-01 1.29402506e+00 -6.03114486e-01 6.02270998e-02
-2.79998660e-01 1.02615786e+00 1.26155779e-01 2.42942169e-01
-7.27349281e-01 -4.72848505e-01 5.99095345e-01 6.51780292e-02
1.34991184e-02 -1.13748658e+00 -2.76950091e-01 -1.38285375e+00
2.95432746e-01 -3.13689202e-01 -2.34282240e-01 -9.43272352e-01
-6.65126979e-01 3.24462980e-01 -9.47087333e-02 -1.65741420e+00
-2.57588744e-01 -4.65394586e-01 -3.17765087e-01 1.05884063e+00
-1.72648501e+00 -1.21216857e+00 -8.43628824e-01 3.38095605e-01
2.73182780e-01 1.75849959e-01 5.34663558e-01 2.75824517e-01
-1.66115493e-01 3.33556384e-02 2.53829420e-01 -2.61981696e-01
1.06081438e+00 -1.01010633e+00 2.53364503e-01 1.16123676e+00
2.29206368e-01 7.57554591e-01 7.84681559e-01 -9.28436577e-01
-1.81619382e+00 -7.85141826e-01 1.16161001e+00 -1.63559303e-01
4.97464538e-01 -6.94310546e-01 -6.34726524e-01 7.36606121e-01
-1.62800342e-01 1.80279724e-02 8.71044695e-02 -6.71063513e-02
1.05888039e-01 -5.12917519e-01 -1.09943831e+00 3.26683372e-01
9.97310698e-01 -3.52654338e-01 -4.12949294e-01 3.46489429e-01
6.33973956e-01 -8.57708633e-01 -5.88246882e-01 5.39144158e-01
7.10233092e-01 -8.92150044e-01 1.00487411e+00 4.84649360e-01
-6.62293017e-01 -1.07724404e+00 -4.15840328e-01 -1.18433177e+00
-1.51334405e-01 -5.36933064e-01 2.11418524e-01 1.28398061e+00
2.72927843e-02 -9.95919883e-01 6.39210761e-01 9.56298560e-02
1.76244900e-02 -1.17696308e-01 -1.16674399e+00 -1.09278309e+00
-8.30133736e-01 -7.93891400e-02 4.93009746e-01 6.81882799e-01
-3.46167624e-01 3.75709683e-01 -5.41022420e-01 7.23101318e-01
1.05281699e+00 4.13567185e-01 1.32076752e+00 -1.55920768e+00
8.93012993e-03 1.91799611e-01 -8.11674714e-01 -1.09955347e+00
1.28536448e-01 -5.77149630e-01 3.49305242e-01 -1.75018668e+00
-1.86988458e-01 -5.52868962e-01 2.78076470e-01 1.46369711e-01
1.84608623e-01 2.76204705e-01 1.83599189e-01 4.75032568e-01
-2.39212766e-01 6.31406069e-01 8.15054953e-01 -4.71770689e-02
-2.04530135e-01 -1.47621054e-02 -1.03390664e-02 6.83687806e-01
5.84879458e-01 -5.23169279e-01 -1.30088434e-01 -6.26847267e-01
3.49900872e-01 -4.56322432e-02 3.54121417e-01 -1.08212090e+00
4.24180061e-01 -4.02204990e-01 2.45416135e-01 -1.12284243e+00
6.51714921e-01 -1.20518708e+00 3.98628324e-01 5.44579685e-01
7.12048650e-01 3.13898116e-01 -1.04204684e-01 6.07726276e-01
-3.49996120e-01 -2.87648976e-01 6.17268085e-01 9.75915715e-02
-9.42896605e-01 3.71711493e-01 6.71022758e-02 -6.03929400e-01
9.58626330e-01 -2.71166682e-01 -1.63644508e-01 -4.17807162e-01
-3.44181776e-01 4.94698435e-01 1.24097168e+00 2.04860806e-01
9.06121790e-01 -1.63734937e+00 -3.25583041e-01 4.17670876e-01
4.75659579e-01 3.70616436e-01 -1.35796070e-01 1.13109183e+00
-1.02538478e+00 3.55680108e-01 -7.58097544e-02 -1.24270141e+00
-1.29800284e+00 3.73191595e-01 8.30392390e-02 3.73393685e-01
-4.56173331e-01 2.13263631e-01 -2.60813415e-01 -6.91753566e-01
1.18946888e-01 -3.42661709e-01 1.57499626e-01 -8.62768665e-02
1.71555474e-01 3.69165719e-01 -3.53380712e-03 -9.32973325e-01
-7.36836672e-01 1.38580394e+00 6.09124720e-01 -3.30273211e-01
1.07896709e+00 -5.95105112e-01 -3.21025372e-01 4.77517456e-01
9.61037219e-01 4.49257135e-01 -9.39944446e-01 -6.92593008e-02
9.17810597e-04 -9.69915628e-01 -2.01217085e-02 -4.20705155e-02
-5.70397615e-01 9.75566685e-01 7.21535683e-01 -1.20411932e-01
7.26990044e-01 -3.51674289e-01 3.57013553e-01 5.30598462e-01
1.04196429e+00 -8.97769332e-01 -4.06116456e-01 5.52343309e-01
9.11339998e-01 -1.45509434e+00 5.00350356e-01 -6.47898316e-01
-1.49953842e-01 1.04108131e+00 5.32081068e-01 -1.45231813e-01
3.03096056e-01 -4.87667434e-02 2.72927225e-01 1.29500255e-01
6.33802786e-02 -3.20036054e-01 1.21208377e-01 5.75820684e-01
1.81873962e-01 1.62703395e-02 -5.63835442e-01 -1.49461135e-01
-2.30654955e-01 -1.38583571e-01 4.32780653e-01 1.13205373e+00
-7.17323661e-01 -1.28472066e+00 -8.04732084e-01 -2.20152602e-01
2.10651383e-01 2.02336267e-01 -3.46321583e-01 1.06859803e+00
3.03874966e-02 9.08719182e-01 -1.22885536e-02 -3.95882398e-01
3.60601664e-01 -1.81119785e-01 4.98464078e-01 -4.31990534e-01
2.45817885e-01 2.72424608e-01 -5.10455295e-02 -8.50057185e-01
-4.41294968e-01 -6.77659273e-01 -1.27120078e+00 -2.28447720e-01
-5.88537097e-01 1.66403159e-01 1.44647002e+00 6.85491145e-01
3.49335998e-01 -7.66200796e-02 6.93174481e-01 -1.10110235e+00
-4.52791005e-01 -7.20035076e-01 -7.22598374e-01 -1.17484026e-01
4.27274108e-01 -1.05973065e+00 -3.68575066e-01 -4.10703331e-01]
|
[7.440664768218994, -2.1982641220092773]
|
7b2779e4-b527-4711-93ec-cb6f0720fbad
|
joint-self-attention-and-scale-aggregation
|
2008.02763
| null |
https://arxiv.org/abs/2008.02763v1
|
https://arxiv.org/pdf/2008.02763v1.pdf
|
Joint Self-Attention and Scale-Aggregation for Self-Calibrated Deraining Network
|
In the field of multimedia, single image deraining is a basic pre-processing work, which can greatly improve the visual effect of subsequent high-level tasks in rainy conditions. In this paper, we propose an effective algorithm, called JDNet, to solve the single image deraining problem and conduct the segmentation and detection task for applications. Specifically, considering the important information on multi-scale features, we propose a Scale-Aggregation module to learn the features with different scales. Simultaneously, Self-Attention module is introduced to match or outperform their convolutional counterparts, which allows the feature aggregation to adapt to each channel. Furthermore, to improve the basic convolutional feature transformation process of Convolutional Neural Networks (CNNs), Self-Calibrated convolution is applied to build long-range spatial and inter-channel dependencies around each spatial location that explicitly expand fields-of-view of each convolutional layer through internal communications and hence enriches the output features. By designing the Scale-Aggregation and Self-Attention modules with Self-Calibrated convolution skillfully, the proposed model has better deraining results both on real-world and synthetic datasets. Extensive experiments are conducted to demonstrate the superiority of our method compared with state-of-the-art methods. The source code will be available at \url{https://supercong94.wixsite.com/supercong94}.
|
['Zhixun Su', 'Yutong Wu', 'Cong Wang', 'Junyang Chen']
|
2020-08-06
| null | null | null | null |
['single-image-deraining']
|
['computer-vision']
|
[-1.48778975e-01 -3.20020616e-01 3.45806926e-01 -6.45764530e-01
-2.36236215e-01 -2.19076782e-01 1.30160749e-01 -3.32420141e-01
-5.18185377e-01 5.80134690e-01 -9.43965930e-03 -1.44119099e-01
1.66355595e-01 -9.18844938e-01 -8.99031699e-01 -7.72898495e-01
-5.66851422e-02 -4.09126788e-01 5.15453696e-01 -3.02264571e-01
-1.09129369e-01 6.13908112e-01 -1.66448855e+00 2.20726192e-01
1.36635244e+00 1.04727936e+00 6.09673500e-01 6.46266520e-01
-2.12827250e-01 6.29226685e-01 -3.75652701e-01 -1.08473569e-01
1.32301971e-01 -2.42082283e-01 -4.04191971e-01 2.27443010e-01
5.05739927e-01 -5.67765296e-01 -6.52047455e-01 1.25739801e+00
5.83492398e-01 1.92886069e-01 1.80524334e-01 -9.16210890e-01
-7.69647598e-01 2.39658445e-01 -6.92788661e-01 7.31553495e-01
-2.11627647e-01 3.45519066e-01 6.41992569e-01 -9.39814210e-01
2.12419465e-01 1.38658202e+00 5.52687705e-01 2.53595561e-01
-7.68824637e-01 -1.02643359e+00 5.01507521e-01 3.64969701e-01
-1.40699577e+00 -2.18637168e-01 7.21114099e-01 -6.42760172e-02
5.10914028e-01 2.55988836e-01 7.41331995e-01 5.99792004e-01
1.28122489e-03 9.89085734e-01 1.07368636e+00 -2.33761892e-02
-1.40490577e-01 7.44999126e-02 1.43497046e-02 6.42852426e-01
3.16124231e-01 1.48215026e-01 -6.76343441e-02 5.31428099e-01
1.10001385e+00 3.88946503e-01 -5.82539439e-01 1.00838348e-01
-1.15222275e+00 6.30112231e-01 1.13685822e+00 3.82418007e-01
-3.26441169e-01 1.02712475e-01 1.00656234e-01 2.09909841e-01
5.97527802e-01 1.38918728e-01 -5.31466186e-01 3.99730384e-01
-8.34299147e-01 1.27375141e-01 1.92503005e-01 1.04961491e+00
1.27273250e+00 1.58448890e-01 -3.52485538e-01 6.35917485e-01
2.51435339e-01 7.29144454e-01 3.37703675e-01 -6.76449537e-01
4.70791280e-01 7.43403792e-01 7.65242577e-02 -9.53147590e-01
-6.43648267e-01 -7.82611609e-01 -1.22759664e+00 9.63166356e-02
8.56677815e-02 -3.52797180e-01 -1.24537623e+00 1.41564965e+00
4.96823341e-01 5.72589159e-01 1.06378391e-01 1.14593923e+00
9.96274292e-01 9.98453856e-01 1.55075401e-01 -2.71856040e-01
1.33676183e+00 -1.14488816e+00 -6.43735886e-01 -4.49112713e-01
2.22814322e-01 -8.53764594e-01 1.18156219e+00 -3.67202121e-03
-8.88833165e-01 -1.09404957e+00 -1.10072958e+00 -1.15319423e-01
-3.64072472e-01 3.00494939e-01 7.04371929e-01 1.74426571e-01
-6.90871775e-01 4.09039587e-01 -6.36305213e-01 -2.51141369e-01
6.47282481e-01 1.43309295e-01 -4.43194062e-02 -2.56315947e-01
-1.48835289e+00 4.34565872e-01 5.30856729e-01 7.21781909e-01
-7.87870467e-01 -6.40114069e-01 -7.13662505e-01 5.26364557e-02
1.74718320e-01 -5.78379929e-01 1.00833213e+00 -1.20291519e+00
-1.01328719e+00 4.69137937e-01 3.48207504e-02 -9.72911939e-02
2.40758404e-01 -3.93038899e-01 -6.26214862e-01 2.62147337e-01
9.34193060e-02 6.65842533e-01 1.08700824e+00 -1.27700830e+00
-9.32036936e-01 -2.95579851e-01 2.09874973e-01 4.92443949e-01
-3.79520327e-01 -2.43098706e-01 -9.35468018e-01 -8.66698742e-01
1.03748143e-01 -5.89487314e-01 -3.79992902e-01 4.10349704e-02
-3.80667746e-01 1.84406161e-01 1.08521605e+00 -6.24641001e-01
1.34401393e+00 -2.39843035e+00 1.90502610e-02 9.19014856e-04
3.24969947e-01 6.24564290e-01 -1.52201191e-01 -1.60221636e-01
-2.64810286e-02 -3.26130763e-02 -3.74207675e-01 -9.98779759e-02
-4.11878943e-01 4.24208522e-01 -1.26564754e-02 4.90503252e-01
3.55392516e-01 1.01358390e+00 -6.26991689e-01 -6.23783410e-01
4.22949642e-01 6.34379268e-01 -3.26844424e-01 5.73660433e-01
-1.52947783e-01 6.30948484e-01 -7.60972738e-01 6.59966171e-01
1.26553774e+00 -2.77163535e-01 -4.48979348e-01 -5.45654416e-01
-4.23851788e-01 -4.87326622e-01 -1.29949021e+00 1.51558650e+00
-6.43080592e-01 3.86215150e-01 1.85602173e-01 -7.51360178e-01
9.32085335e-01 -1.63482577e-01 2.25154255e-02 -8.56394291e-01
2.34172285e-01 2.04792824e-02 -1.97361082e-01 -9.13387477e-01
2.76054412e-01 2.10258350e-01 2.01593444e-01 -2.64195260e-02
-7.03088790e-02 7.42814690e-02 -4.30453867e-02 7.06092268e-02
5.28303564e-01 5.06531522e-02 5.10759279e-03 -2.42191315e-01
8.65840137e-01 -2.96794415e-01 6.44355774e-01 5.84768653e-01
-5.33944108e-02 7.00584173e-01 5.17526828e-02 -6.13473952e-01
-1.06423283e+00 -8.67174029e-01 -3.53075951e-01 1.07219148e+00
6.54256463e-01 2.01311737e-01 -7.97679424e-01 -4.66522336e-01
3.88085730e-02 3.49799037e-01 -7.76486576e-01 -9.15330797e-02
-5.53476393e-01 -9.80699420e-01 3.10009986e-01 7.10162401e-01
1.21577752e+00 -1.21779418e+00 -5.39170563e-01 2.60664105e-01
-1.17985688e-01 -1.07412946e+00 -6.21765256e-01 8.12837631e-02
-7.66255736e-01 -9.20922577e-01 -9.22920644e-01 -1.01168382e+00
7.54110038e-01 7.09254384e-01 7.81014204e-01 4.40991551e-01
-3.20500463e-01 -1.83366492e-01 -6.25150204e-01 -3.95002812e-01
3.69772017e-01 9.19943079e-02 -4.47188258e-01 3.07316959e-01
1.43551037e-01 -7.31686890e-01 -1.14660394e+00 2.64766693e-01
-1.29427338e+00 1.46535754e-01 1.10251367e+00 8.13607514e-01
5.98460734e-01 5.42956516e-02 4.25622314e-01 -9.45993900e-01
2.98127562e-01 -4.06345397e-01 -5.56737304e-01 1.97345972e-01
-3.11004579e-01 -3.83961834e-02 7.88572967e-01 -2.49944150e-01
-1.27340186e+00 1.08299106e-01 -2.14854494e-01 -5.28775871e-01
-2.99291998e-01 2.76949048e-01 -5.29117525e-01 -2.33276024e-01
3.00826281e-01 5.43912947e-01 -1.52001709e-01 -5.66348493e-01
4.89396155e-01 7.79457510e-01 7.27836251e-01 -2.95974612e-01
1.04885554e+00 5.70941389e-01 -3.47065598e-01 -7.76440799e-01
-1.02368391e+00 -4.95217860e-01 -6.97084844e-01 -1.78576499e-01
1.07405078e+00 -1.28596723e+00 -5.92293739e-01 9.35469568e-01
-1.08468688e+00 -3.09060097e-01 6.51022494e-02 4.30147678e-01
-1.08848952e-01 3.21782976e-01 -5.49048960e-01 -5.12000442e-01
-5.49472809e-01 -1.16411066e+00 1.03537416e+00 1.02892637e+00
7.85389245e-01 -8.12326312e-01 -3.14036280e-01 2.81950235e-02
6.82895303e-01 1.02607772e-01 5.14925301e-01 -2.34532967e-01
-8.67521167e-01 9.49015915e-02 -8.25439334e-01 5.67785203e-01
-4.66702096e-02 1.69156790e-02 -1.03417194e+00 -3.75832170e-01
-1.28259286e-01 -9.61834118e-02 1.18031156e+00 3.77467066e-01
1.53321755e+00 -1.68422863e-01 -1.79774463e-01 1.11796474e+00
1.55623674e+00 6.21222816e-02 8.08923900e-01 4.90526736e-01
9.53311801e-01 4.45644587e-01 7.73405313e-01 4.54860330e-01
5.57229996e-01 3.23865294e-01 6.77308559e-01 -8.23566437e-01
-1.07748613e-01 -2.42812261e-02 5.00788679e-04 5.94584525e-01
-2.11804599e-01 -1.04338400e-01 -3.91890287e-01 4.12441313e-01
-1.80697083e+00 -8.35069418e-01 -2.19049633e-01 1.82580602e+00
6.43604696e-01 -6.53847605e-02 -2.45198801e-01 -2.29486227e-01
8.24339688e-01 4.28466886e-01 -6.78842783e-01 3.67474817e-02
-3.40333968e-01 2.55454183e-01 6.79741621e-01 4.09365982e-01
-1.41747952e+00 1.02641404e+00 5.02878475e+00 9.78638113e-01
-1.26319098e+00 1.57760590e-01 8.67863953e-01 1.90861106e-01
-2.66508907e-01 -1.90994352e-01 -7.86465764e-01 7.25053549e-01
3.49932820e-01 4.15916666e-02 3.16593528e-01 6.76371872e-01
3.52966726e-01 1.75041705e-03 -3.89980644e-01 7.45196879e-01
-1.16136983e-01 -1.13197827e+00 -5.36432937e-02 -2.53877729e-01
7.77103543e-01 9.52152014e-02 7.16378316e-02 3.13405663e-01
5.45145385e-02 -9.36911702e-01 4.79877234e-01 7.78848231e-01
9.29199755e-01 -9.52274740e-01 9.81774747e-01 5.87090552e-02
-1.61785710e+00 -1.56670287e-01 -7.03683317e-01 1.38762191e-01
-4.29292396e-02 7.90180445e-01 -9.95773599e-02 7.10484385e-01
1.19706917e+00 8.82483602e-01 -8.18555951e-01 1.12860680e+00
-3.93207639e-01 4.41954225e-01 -2.47278556e-01 1.97959989e-01
5.21647215e-01 -1.49457395e-01 2.47765422e-01 1.42052078e+00
2.72062272e-01 3.63257587e-01 1.94280729e-01 5.88545501e-01
2.07477938e-02 1.31473631e-01 -2.34762594e-01 5.09725153e-01
5.17108858e-01 1.54062545e+00 -4.53516006e-01 -3.77579212e-01
-6.06857598e-01 1.11411548e+00 2.76911944e-01 7.37678349e-01
-1.06673658e+00 -9.33360219e-01 7.60726213e-01 7.19905794e-02
7.58821130e-01 -1.58191577e-01 -4.46651131e-02 -1.09703112e+00
9.30276588e-02 -6.39774740e-01 2.25836396e-01 -9.16023850e-01
-1.14988339e+00 8.57915163e-01 -2.34580666e-01 -1.33685386e+00
5.39002717e-01 -4.02945787e-01 -1.08098316e+00 1.00960100e+00
-2.18875957e+00 -1.46736526e+00 -1.01453257e+00 8.75670254e-01
4.79683310e-01 1.70066729e-01 2.22824171e-01 5.87677479e-01
-8.42083693e-01 6.02083862e-01 2.20186606e-01 1.50937572e-01
7.85136402e-01 -9.64391887e-01 3.43066812e-01 1.15653908e+00
-2.87047595e-01 3.89111698e-01 4.14544374e-01 -4.49264973e-01
-1.10001278e+00 -1.72615802e+00 2.98376173e-01 1.82795018e-01
4.58427876e-01 -1.38724670e-01 -1.26063478e+00 5.51530898e-01
1.87237740e-01 5.70213139e-01 1.51483357e-01 -2.84789711e-01
-2.19768539e-01 -5.70696115e-01 -8.32879901e-01 2.80328065e-01
1.02331710e+00 -3.03847015e-01 -3.16521853e-01 2.97677219e-01
1.18905580e+00 -6.25080466e-01 -6.88455641e-01 5.50464988e-01
2.47299582e-01 -1.15878892e+00 1.05332649e+00 -2.53067791e-01
3.72732520e-01 -6.99912846e-01 -5.65728694e-02 -1.18439305e+00
-5.78789175e-01 -3.05132389e-01 4.95764762e-02 1.26024461e+00
2.48753458e-01 -5.70176721e-01 5.78857958e-01 -1.52512854e-02
-4.28887784e-01 -9.36532021e-01 -6.80582285e-01 -4.25179422e-01
-6.82985112e-02 -1.44019738e-01 7.81069219e-01 8.20655167e-01
-7.23892808e-01 1.91248506e-01 -5.68971694e-01 6.55923426e-01
6.22629344e-01 4.71550167e-01 6.92200422e-01 -9.61822271e-01
-8.23882297e-02 -1.29400790e-01 -4.10252750e-01 -1.44748700e+00
-1.27752349e-01 -5.84836304e-01 1.09899200e-01 -1.53489268e+00
5.42245917e-02 -4.63443935e-01 -4.12517309e-01 5.01170278e-01
-6.94313407e-01 3.64321589e-01 1.89709857e-01 2.06764355e-01
-7.79929161e-01 8.45938683e-01 1.85051644e+00 -1.32099345e-01
-2.13029936e-01 1.62140336e-02 -7.57398605e-01 7.07281053e-01
9.22424376e-01 -2.13028654e-01 -2.71335959e-01 -6.98875189e-01
-2.38939688e-01 -1.50810018e-01 5.15476286e-01 -1.25849044e+00
3.30576181e-01 -1.99641377e-01 8.58739138e-01 -6.64501071e-01
8.11282471e-02 -8.27361763e-01 -1.20843403e-01 4.70274091e-01
-6.94859624e-02 1.46175250e-02 3.99354100e-01 6.67826533e-01
-4.77430254e-01 5.33806048e-02 1.04812145e+00 -2.04313800e-01
-1.18969214e+00 7.31377840e-01 -1.10023342e-01 -1.75431697e-03
1.04085827e+00 -1.11616664e-01 -3.83934587e-01 -1.73281252e-01
-7.61796176e-01 7.49035895e-01 2.86442697e-01 3.00918370e-01
8.15994859e-01 -1.20882750e+00 -7.70997882e-01 4.48444664e-01
5.98143414e-02 4.94585574e-01 9.94542003e-01 8.66921246e-01
-7.15360641e-01 -3.11933886e-02 -4.21205491e-01 -5.31464279e-01
-9.93708909e-01 5.21843374e-01 4.33984905e-01 1.47072673e-02
-7.39975572e-01 1.02016413e+00 6.05573237e-01 -1.80263773e-01
1.12524144e-01 -3.05551440e-01 -4.12066251e-01 -9.67720821e-02
7.48221099e-01 1.17516175e-01 -1.73903555e-01 -5.42759418e-01
-2.87995875e-01 7.21058547e-01 -3.09320241e-01 3.67581666e-01
1.38848770e+00 -5.84875822e-01 -1.27098590e-01 5.51886931e-02
1.16540968e+00 -1.44541606e-01 -1.65563798e+00 -4.77654338e-01
-7.67671704e-01 -6.47315383e-01 2.49322712e-01 -4.88110721e-01
-1.63241267e+00 1.02021956e+00 8.39440346e-01 -8.45076591e-02
1.52412856e+00 -1.50933310e-01 9.37408864e-01 2.93684959e-01
-1.04580536e-01 -8.46825123e-01 1.44020528e-01 4.92437989e-01
6.92755461e-01 -1.33547819e+00 5.19658858e-03 -3.76429975e-01
-6.78717315e-01 1.06707680e+00 1.18531644e+00 -3.06175619e-01
7.53284991e-01 2.55924821e-01 2.42866576e-01 -1.23126246e-01
-2.82646149e-01 -5.45611322e-01 1.34750947e-01 5.08995175e-01
1.87294617e-01 -6.66654631e-02 -2.50415742e-01 6.98986888e-01
2.62316853e-01 -1.02114849e-01 3.75308633e-01 7.09780812e-01
-9.23837721e-01 -6.55212700e-01 -3.91490132e-01 4.20336187e-01
-1.55288160e-01 -3.61222416e-01 1.43535048e-01 7.79857457e-01
6.35628462e-01 8.21485996e-01 2.42722228e-01 -5.48900366e-01
4.88333941e-01 -5.00237942e-01 1.17903300e-01 -3.07756841e-01
-3.95123869e-01 2.19992891e-01 -3.94299477e-01 -6.09546661e-01
-6.12303793e-01 -3.07667047e-01 -1.34334433e+00 -2.95652688e-01
-2.84466505e-01 1.81045085e-01 2.51051396e-01 7.72700489e-01
4.40965533e-01 9.35884714e-01 9.42940474e-01 -1.04872835e+00
-2.01989591e-01 -9.93574798e-01 -6.95641518e-01 2.90933281e-01
5.08278191e-01 -4.96786535e-01 -3.03341955e-01 1.04330979e-01]
|
[10.843006134033203, -3.007000207901001]
|
5814b772-ec6e-48aa-b6c7-6df91ccf7d55
|
nighthazeformer-single-nighttime-haze-removal
|
2305.09533
| null |
https://arxiv.org/abs/2305.09533v1
|
https://arxiv.org/pdf/2305.09533v1.pdf
|
NightHazeFormer: Single Nighttime Haze Removal Using Prior Query Transformer
|
Nighttime image dehazing is a challenging task due to the presence of multiple types of adverse degrading effects including glow, haze, blurry, noise, color distortion, and so on. However, most previous studies mainly focus on daytime image dehazing or partial degradations presented in nighttime hazy scenes, which may lead to unsatisfactory restoration results. In this paper, we propose an end-to-end transformer-based framework for nighttime haze removal, called NightHazeFormer. Our proposed approach consists of two stages: supervised pre-training and semi-supervised fine-tuning. During the pre-training stage, we introduce two powerful priors into the transformer decoder to generate the non-learnable prior queries, which guide the model to extract specific degradations. For the fine-tuning, we combine the generated pseudo ground truths with input real-world nighttime hazy images as paired images and feed into the synthetic domain to fine-tune the pre-trained model. This semi-supervised fine-tuning paradigm helps improve the generalization to real domain. In addition, we also propose a large-scale synthetic dataset called UNREAL-NH, to simulate the real-world nighttime haze scenarios comprehensively. Extensive experiments on several synthetic and real-world datasets demonstrate the superiority of our NightHazeFormer over state-of-the-art nighttime haze removal methods in terms of both visually and quantitatively.
|
['ErKang Chen', 'Wenqi Ren', 'Tian Ye', 'Sixiang Chen', 'Zhongsheng Yan', 'Yun Liu']
|
2023-05-16
| null | null | null | null |
['image-dehazing']
|
['computer-vision']
|
[ 1.09634221e-01 -5.71525097e-01 8.03245366e-01 -5.19808114e-01
-7.17371941e-01 -3.93208802e-01 5.00764370e-01 -5.28022289e-01
-7.49507546e-02 8.16946030e-01 4.82518941e-01 -5.53589240e-02
5.18117808e-02 -6.89985275e-01 -6.27873719e-01 -1.23761797e+00
4.65978622e-01 -2.13648289e-01 3.81983429e-01 -4.91357923e-01
4.23137136e-02 -9.18460824e-03 -1.63153124e+00 3.39128017e-01
1.54553449e+00 6.87012434e-01 5.79012513e-01 7.17661202e-01
5.63414633e-01 6.76139235e-01 -7.72127926e-01 -2.43206576e-01
4.82331783e-01 -4.40153360e-01 2.92621851e-02 5.32074094e-01
4.93413955e-01 -5.43324232e-01 -4.89895254e-01 1.43893611e+00
6.71951830e-01 4.06021208e-01 4.32107240e-01 -1.06301761e+00
-8.20208013e-01 -2.46212035e-01 -4.99236703e-01 3.67443174e-01
-2.52147436e-01 6.58233762e-01 2.23658606e-01 -9.38165367e-01
-1.14616295e-02 1.07707143e+00 4.17696863e-01 3.84201974e-01
-7.54987359e-01 -9.24669147e-01 3.57043482e-02 4.23039198e-01
-1.40237176e+00 -7.33081222e-01 8.02542210e-01 -3.45089942e-01
4.34116215e-01 2.59994000e-01 5.24209142e-01 7.82393157e-01
3.69009316e-01 4.77954149e-01 1.44885957e+00 -1.12770990e-01
1.67365223e-01 1.22569390e-01 -4.12887424e-01 3.89896095e-01
1.72628745e-01 3.77926737e-01 -1.72473669e-01 2.09940970e-01
5.09445786e-01 1.17248662e-01 -6.58818722e-01 3.70132625e-02
-8.83557677e-01 4.85682577e-01 5.74955761e-01 -2.21897826e-01
-3.96101832e-01 -1.80404350e-01 -1.74186565e-02 3.91331106e-01
7.99611747e-01 1.69849843e-01 -2.51968801e-01 3.06395799e-01
-9.98146951e-01 2.65385360e-01 2.15612948e-01 9.32836831e-01
9.64784980e-01 2.16957942e-01 -3.43256801e-01 1.03865409e+00
5.80208302e-01 9.18223917e-01 3.98362696e-01 -7.33074248e-01
7.02084363e-01 -3.85819227e-02 4.66257989e-01 -7.64715791e-01
4.94050905e-02 -6.58811629e-01 -9.69566524e-01 3.48775387e-01
-1.69666395e-01 -1.82631329e-01 -1.47924924e+00 1.39689600e+00
2.90334225e-01 3.54340702e-01 1.63000822e-01 1.45257127e+00
7.75194049e-01 1.26151931e+00 -1.79455906e-01 -3.74658793e-01
1.18248022e+00 -1.38970613e+00 -1.10405433e+00 -5.65692186e-01
-5.75416721e-02 -1.02644897e+00 1.29276860e+00 5.49976349e-01
-8.78805578e-01 -6.85338259e-01 -1.11270201e+00 -8.34549293e-02
-3.19799453e-01 1.41249508e-01 1.23500619e-02 4.89939064e-01
-1.05535924e+00 1.04566820e-01 -6.82251632e-01 -3.00882496e-02
2.15706959e-01 -1.21269487e-01 -8.55851248e-02 -6.22831404e-01
-1.42479205e+00 9.03795123e-01 2.92780191e-01 6.77107394e-01
-1.65860188e+00 -5.99665225e-01 -8.81399333e-01 -1.03771366e-01
4.23892677e-01 -8.70712876e-01 1.13171852e+00 -8.86953473e-01
-1.43222237e+00 6.07481122e-01 -1.83734745e-01 -1.03811949e-01
4.49821323e-01 -5.11519432e-01 -1.04610062e+00 -1.59748048e-01
1.52242333e-01 3.00336421e-01 1.39784503e+00 -1.68284619e+00
-6.38537347e-01 -1.20501824e-01 1.17625415e-01 5.63432395e-01
-1.79136381e-01 -2.27567136e-01 -7.42551386e-01 -1.00791097e+00
-1.71219066e-01 -6.14160895e-01 -1.51711181e-01 -9.91041809e-02
-3.24670255e-01 3.81923497e-01 1.05758309e+00 -9.88977015e-01
1.15257347e+00 -2.50601912e+00 -1.31242320e-01 -2.02117369e-01
1.39681593e-01 5.15146911e-01 -1.46057710e-01 3.71193320e-01
-5.94456270e-02 -3.31457287e-01 -5.93227863e-01 -4.48799193e-01
-1.18412279e-01 3.24908495e-01 -5.64412534e-01 7.28071392e-01
2.24251673e-01 5.55473387e-01 -9.83651280e-01 -2.84071952e-01
5.74753404e-01 6.88302517e-01 -2.18225732e-01 7.68636942e-01
-1.70018867e-01 6.44736826e-01 -1.54481858e-01 5.86078525e-01
1.18224537e+00 2.03233063e-01 -5.21918297e-01 -4.06925380e-01
-3.29551011e-01 9.70615596e-02 -8.33667874e-01 1.48878121e+00
-5.69697976e-01 6.70980752e-01 1.86101660e-01 -4.61214691e-01
7.17580080e-01 1.34164646e-01 -4.40408915e-01 -1.06705797e+00
1.98513791e-02 1.68757349e-01 -3.12813163e-01 -1.02208376e+00
4.31597471e-01 -5.72298229e-01 3.54115039e-01 7.17765912e-02
-2.64304072e-01 -4.92647022e-01 -9.30253267e-02 -1.21414028e-01
7.38276184e-01 1.10373609e-01 -1.44758955e-01 -2.18866572e-01
8.45572233e-01 -1.56132445e-01 6.93587840e-01 3.76938850e-01
-3.19266349e-01 1.21560860e+00 -2.21654460e-01 -1.89136788e-01
-1.11448717e+00 -1.25576866e+00 5.11962129e-03 8.23582590e-01
5.75467765e-01 -1.70731217e-01 -7.31699288e-01 -2.41698056e-01
-5.08668244e-01 8.32292616e-01 -5.72623372e-01 -5.88822126e-01
-3.48213017e-01 -1.12030065e+00 2.86861002e-01 -9.02933553e-02
1.15583301e+00 -8.92103732e-01 -1.75457984e-01 -5.10668457e-02
-4.62683648e-01 -1.13199353e+00 -6.53975129e-01 -2.10891381e-01
-5.84963799e-01 -7.32317269e-01 -8.84343147e-01 -7.29714453e-01
7.28013933e-01 8.73319566e-01 8.50799382e-01 5.70096541e-03
-1.49065271e-01 -1.65628180e-01 -3.68288785e-01 -4.42591250e-01
-3.33030790e-01 -5.40659189e-01 -9.90262106e-02 5.24540305e-01
-9.86014754e-02 -6.51552379e-01 -1.18108547e+00 4.83420223e-01
-1.43743753e+00 1.37545794e-01 6.38363302e-01 7.58826494e-01
4.52694833e-01 8.95839691e-01 5.37510775e-02 -6.34822786e-01
5.87567985e-01 -5.64662158e-01 -7.89295495e-01 1.04644798e-01
-5.84851503e-01 -1.75873563e-01 8.20276141e-01 -2.56269991e-01
-1.65562284e+00 -2.40395993e-01 -8.13391358e-02 -7.35934079e-01
-2.63435990e-01 4.30389553e-01 -5.78757226e-01 -1.36813864e-01
6.56895459e-01 7.20374286e-01 -2.86969930e-01 -5.89903891e-01
3.78100961e-01 9.55787897e-01 9.99136567e-01 -1.59393206e-01
1.57158124e+00 7.34908462e-01 -4.22614604e-01 -7.83275962e-01
-1.15292466e+00 -4.25795406e-01 -1.49984092e-01 -2.14399517e-01
8.98544192e-01 -1.60962999e+00 -2.85729747e-02 9.47422802e-01
-8.77404332e-01 -6.43039942e-01 1.69460177e-01 2.31070846e-01
-2.66973585e-01 4.28828955e-01 -4.01088655e-01 -8.25739682e-01
-4.49289650e-01 -1.06679070e+00 1.18004155e+00 5.46345532e-01
8.22675109e-01 -8.56108010e-01 -1.18288619e-03 5.55244386e-01
5.95537543e-01 8.58338550e-03 6.61630034e-01 3.71505231e-01
-7.57560551e-01 1.24952368e-01 -4.74154413e-01 8.83604527e-01
4.67593908e-01 -2.45884955e-01 -1.23322821e+00 -4.79000181e-01
3.35561156e-01 -6.24679849e-02 1.00645483e+00 1.83542207e-01
1.04342651e+00 -3.99644256e-01 7.97211304e-02 1.11210871e+00
1.46530974e+00 8.19131061e-02 1.18965673e+00 6.24819100e-01
7.76654959e-01 4.53860790e-01 1.06954753e+00 3.84200960e-01
5.10669947e-01 4.47437525e-01 7.95925736e-01 -3.99434954e-01
-3.24010819e-01 -1.05435915e-01 4.99287546e-01 6.89577579e-01
-1.56538014e-03 -5.97822189e-01 -5.63755691e-01 7.43799448e-01
-1.64115977e+00 -6.91518366e-01 -9.37023535e-02 2.26015043e+00
9.74197984e-01 -9.39887424e-04 -3.83053690e-01 -1.08021140e-01
6.74830973e-01 4.57647204e-01 -3.74043405e-01 1.04657307e-01
-3.95873964e-01 -1.03293791e-01 5.96237361e-01 5.75278640e-01
-1.19457650e+00 1.05763972e+00 5.02939844e+00 7.83848643e-01
-1.23363984e+00 3.46914828e-01 4.05015588e-01 -4.98024151e-02
-4.24448222e-01 -1.75298452e-02 -4.42022532e-01 8.79535496e-01
8.64247620e-01 -2.05409471e-02 9.63415742e-01 1.65576607e-01
8.70063722e-01 -9.54054967e-02 -3.30585361e-01 1.04202378e+00
3.61828387e-01 -8.39101076e-01 -1.50908353e-02 -2.70491868e-01
9.78069723e-01 1.85262650e-01 1.74989581e-01 2.54043251e-01
1.65222093e-01 -8.15764666e-01 9.03475165e-01 6.54347956e-01
7.81774640e-01 -6.49733305e-01 7.32241094e-01 3.93532425e-01
-8.28954518e-01 -1.40655115e-01 -5.46883762e-01 -1.76021345e-02
2.61422992e-01 9.65438128e-01 -6.10004723e-01 6.73830390e-01
9.96228039e-01 8.64072561e-01 -6.42787933e-01 1.37749922e+00
-8.21680307e-01 7.97085941e-01 -4.48242314e-02 7.08303332e-01
2.55343616e-01 -3.73368919e-01 5.32322347e-01 1.09852302e+00
3.52993101e-01 4.80607212e-01 -1.89220205e-01 7.53137410e-01
2.33707190e-01 -5.99702954e-01 -2.68448383e-01 2.42129788e-01
1.52475238e-01 1.20136523e+00 -1.30511299e-01 -3.59254748e-01
-3.20101082e-01 1.34698153e+00 -3.78980845e-01 7.95087874e-01
-1.18624544e+00 -5.76094925e-01 8.89561653e-01 5.60378246e-02
3.17028761e-01 -2.43404344e-01 4.87136729e-02 -1.37294662e+00
1.29158705e-01 -1.00908923e+00 7.93388709e-02 -1.58960402e+00
-1.33592093e+00 6.16660476e-01 -1.09821513e-01 -1.54316652e+00
3.16014975e-01 -3.86873931e-01 -7.08723247e-01 1.24083722e+00
-2.16465425e+00 -1.26074421e+00 -1.19292712e+00 8.14959109e-01
8.84267390e-01 2.20968097e-01 2.62494236e-01 7.51922369e-01
-6.53809130e-01 2.46281818e-01 3.94482940e-01 -1.89978406e-01
1.08087242e+00 -1.04859602e+00 4.24305856e-01 1.46088016e+00
-4.30763781e-01 4.17693377e-01 1.20252633e+00 -5.99426985e-01
-1.20874310e+00 -1.82383347e+00 3.73877347e-01 -1.79362133e-01
4.01051432e-01 -4.62069571e-01 -1.02649379e+00 4.73832011e-01
3.14922154e-01 3.09691757e-01 9.78700444e-02 -6.48649991e-01
-2.32357278e-01 -5.25827765e-01 -9.90098059e-01 6.73535109e-01
7.57435918e-01 -6.28438830e-01 -6.42734587e-01 4.42444891e-01
1.00144672e+00 -5.91092050e-01 -3.80365103e-01 5.41236281e-01
2.25142881e-01 -1.17186797e+00 9.01842177e-01 -6.17251806e-02
5.20792902e-01 -1.05993569e+00 -2.13184714e-01 -1.85494936e+00
-4.79181498e-01 -7.48757064e-01 9.42919850e-02 1.29729211e+00
1.02438308e-01 -6.94809496e-01 3.02220255e-01 8.95602629e-02
-7.09330261e-01 -2.13005200e-01 -3.91317487e-01 -6.34322286e-01
-2.92117476e-01 -9.12927762e-02 7.85179973e-01 9.43211794e-01
-8.81230175e-01 1.28446266e-01 -7.80217409e-01 9.85955715e-01
8.33304346e-01 8.49929079e-02 8.13279033e-01 -6.55980825e-01
-1.90838352e-01 2.46775627e-01 -1.20012514e-01 -9.39692259e-01
-2.60296673e-01 -1.45134285e-01 8.07801664e-01 -1.62262237e+00
5.40419556e-02 -1.80419147e-01 -4.90711391e-01 2.92875677e-01
-5.71740687e-01 5.35120666e-01 -1.38675421e-01 2.96507269e-01
-5.14954805e-01 1.06627691e+00 1.60425889e+00 -3.66876006e-01
-9.47634354e-02 -4.86714132e-02 -6.56533360e-01 4.16829228e-01
6.27242088e-01 -4.18878257e-01 -7.16668248e-01 -9.53701437e-01
7.15555325e-02 -2.21459046e-02 5.70469201e-01 -1.12265015e+00
3.28137353e-02 -3.29095840e-01 3.49843532e-01 -4.27913070e-01
5.83559036e-01 -6.75084054e-01 1.97787389e-01 1.11428954e-01
1.46682739e-01 -1.98035821e-01 2.71764040e-01 6.63624585e-01
-5.44903815e-01 1.26107797e-01 9.31720316e-01 -4.13309745e-02
-8.46569657e-01 3.84750009e-01 -2.00367525e-01 7.63911614e-03
7.31782973e-01 -1.74846753e-01 -8.42353463e-01 -6.12485409e-01
-4.60115075e-01 3.24655533e-01 6.45658910e-01 4.95701700e-01
7.53697276e-01 -9.26157415e-01 -8.93031120e-01 5.35345793e-01
3.45643103e-01 2.21546203e-01 7.10506260e-01 8.59097302e-01
-6.75245047e-01 -3.47692743e-02 -1.66501135e-01 -3.20611566e-01
-1.03041518e+00 6.32691681e-01 4.99706358e-01 3.30071628e-01
-6.60068274e-01 7.82342076e-01 8.61419022e-01 -1.88076615e-01
-1.06535636e-01 -1.54302254e-01 3.09639852e-02 -4.35910761e-01
6.95761204e-01 1.94631249e-01 2.96573520e-01 -5.43258250e-01
-1.00719467e-01 4.23457980e-01 1.02951884e-01 -8.04870501e-02
1.34192038e+00 -7.61744380e-01 -3.15851122e-01 2.33668685e-01
8.76108408e-01 8.16989839e-02 -1.65045393e+00 -1.73649460e-01
-6.53502047e-01 -8.81312370e-01 4.02361721e-01 -1.09034586e+00
-1.19418526e+00 9.40657973e-01 9.18834448e-01 -1.63803846e-01
1.86938393e+00 -4.08260673e-01 9.31831598e-01 1.54987723e-01
1.32192731e-01 -6.87763155e-01 4.89449203e-02 3.66888404e-01
9.35429394e-01 -1.14373434e+00 1.37887551e-02 -5.19951165e-01
-6.16404295e-01 7.06259608e-01 7.64172196e-01 -1.20446227e-01
5.12618065e-01 -8.89221728e-02 6.19799495e-01 -8.78498033e-02
-6.64947212e-01 -1.99856013e-01 2.42613867e-01 6.14095390e-01
-1.21820979e-01 -1.40101507e-01 4.40464020e-02 4.60368872e-01
-2.42034242e-01 8.18219706e-02 7.18761861e-01 7.27344275e-01
-6.34575307e-01 -5.35193741e-01 -7.92186797e-01 8.46495703e-02
-1.55289412e-01 -3.89661133e-01 4.66677025e-02 3.52098465e-01
4.81146574e-01 1.29120803e+00 -1.09476566e-01 -4.19895828e-01
3.72869045e-01 -4.42228884e-01 2.68609852e-01 -5.01556218e-01
-2.12677449e-01 3.20480794e-01 -5.38205132e-02 -3.29269707e-01
-2.08756685e-01 -2.38986403e-01 -7.85852313e-01 -1.45942792e-01
-3.80444586e-01 2.05383599e-01 4.68609452e-01 9.73869085e-01
3.41075808e-01 6.79838836e-01 8.63904357e-01 -9.09441710e-01
-2.48771623e-01 -1.10849202e+00 -8.07400525e-01 4.31829244e-01
1.00016606e+00 -6.66161358e-01 -6.14755392e-01 3.94551992e-01]
|
[10.940521240234375, -3.1801352500915527]
|
bfacb815-50e1-4866-b9ab-f47010530f04
|
language-aware-multilingual-machine
|
2302.05008
| null |
https://arxiv.org/abs/2302.05008v1
|
https://arxiv.org/pdf/2302.05008v1.pdf
|
Language-Aware Multilingual Machine Translation with Self-Supervised Learning
|
Multilingual machine translation (MMT) benefits from cross-lingual transfer but is a challenging multitask optimization problem. This is partly because there is no clear framework to systematically learn language-specific parameters. Self-supervised learning (SSL) approaches that leverage large quantities of monolingual data (where parallel data is unavailable) have shown promise by improving translation performance as complementary tasks to the MMT task. However, jointly optimizing SSL and MMT tasks is even more challenging. In this work, we first investigate how to utilize intra-distillation to learn more *language-specific* parameters and then show the importance of these language-specific parameters. Next, we propose a novel but simple SSL task, concurrent denoising, that co-trains with the MMT task by concurrently denoising monolingual data on both the encoder and decoder. Finally, we apply intra-distillation to this co-training approach. Combining these two approaches significantly improves MMT performance, outperforming three state-of-the-art SSL methods by a large margin, e.g., 11.3\% and 3.7\% improvement on an 8-language and a 15-language benchmark compared with MASS, respectively
|
['Vedanuj Goswami', 'Jean Maillard', 'Haoran Xu']
|
2023-02-10
| null | null | null | null |
['cross-lingual-transfer']
|
['natural-language-processing']
|
[ 3.90016399e-02 -2.60433942e-01 -4.67394978e-01 -4.10255373e-01
-1.85314727e+00 -6.15337312e-01 7.49920547e-01 -1.28492385e-01
-6.16494536e-01 1.01703978e+00 2.34210044e-01 -6.95072174e-01
3.74449223e-01 -1.18571930e-01 -1.20187676e+00 -6.53063238e-01
1.99015766e-01 6.29534304e-01 -2.86073178e-01 -3.98076892e-01
-2.45927140e-01 -6.54154122e-02 -7.19282866e-01 4.31105405e-01
1.32252598e+00 5.69998622e-01 4.44503576e-01 3.30524832e-01
-1.73936352e-01 3.96576911e-01 -2.79992014e-01 -5.22506297e-01
4.00440544e-01 -4.60528463e-01 -6.35716856e-01 -5.24088033e-02
6.31987691e-01 -1.14149530e-03 1.13050371e-01 8.83120000e-01
7.08714306e-01 -3.13862771e-01 4.35503364e-01 -8.48920166e-01
-6.65719688e-01 8.85167420e-01 -7.76433408e-01 4.35262695e-02
-8.54321755e-03 1.91999719e-01 1.22246039e+00 -1.18472588e+00
6.38927221e-01 1.37499511e+00 7.27637410e-01 3.57454151e-01
-1.54921937e+00 -8.82216036e-01 8.87547135e-02 4.87209372e-02
-1.25645566e+00 -7.77468324e-01 5.58400571e-01 -3.19518536e-01
1.28213596e+00 -8.91072601e-02 9.58827063e-02 1.31573164e+00
5.62682986e-01 9.66102958e-01 1.64595306e+00 -6.15490437e-01
-1.80950537e-01 2.99562782e-01 -2.56754607e-01 7.10657775e-01
-4.02168594e-02 7.96627477e-02 -7.80692220e-01 1.13987014e-01
2.96673983e-01 -4.89352316e-01 -2.54370328e-02 -1.33506715e-01
-1.64226460e+00 7.54841268e-01 8.70342553e-02 3.29490274e-01
-1.90083474e-01 1.33187696e-01 5.65591872e-01 8.35902452e-01
1.04977679e+00 3.96961927e-01 -1.02240825e+00 -2.75404930e-01
-1.12618744e+00 -1.83034673e-01 6.94214761e-01 9.49268103e-01
9.47408378e-01 1.22673668e-01 -1.13926746e-01 1.07682729e+00
2.85152942e-01 9.50517476e-01 4.51645762e-01 -5.10724306e-01
1.09485936e+00 2.46557385e-01 -3.23228151e-01 -2.53483087e-01
-2.50642508e-01 -8.05852532e-01 -8.08472455e-01 -1.28343567e-01
3.35960418e-01 -4.53683585e-01 -7.28934944e-01 2.08230042e+00
6.45495206e-02 -2.14034994e-03 2.87439764e-01 7.35657811e-01
3.40138108e-01 8.67599487e-01 -7.43257329e-02 -3.59311670e-01
1.27147818e+00 -1.37929630e+00 -7.53311098e-01 -6.07633233e-01
1.08893239e+00 -1.42251539e+00 1.24305403e+00 2.45424062e-01
-1.07049489e+00 -4.41442132e-01 -1.01312947e+00 -2.78310657e-01
-2.04596832e-01 4.99191701e-01 5.00690281e-01 3.64719629e-01
-1.09487641e+00 3.78729135e-01 -9.21139121e-01 -3.38045120e-01
1.09356754e-01 4.30255711e-01 -4.67229903e-01 -2.53791660e-01
-1.35423112e+00 1.38449538e+00 4.06134613e-02 2.61695459e-02
-7.76113510e-01 -7.76474595e-01 -8.65263760e-01 -2.80368000e-01
3.27401608e-01 -6.32474422e-01 1.30109704e+00 -8.65774572e-01
-1.70832217e+00 8.84060800e-01 -5.04592896e-01 -4.32663947e-01
6.58548117e-01 -4.93282944e-01 -3.21568102e-01 -4.73843843e-01
4.31623250e-01 6.24825299e-01 7.15447605e-01 -1.11927283e+00
-5.32815993e-01 -2.76706785e-01 -3.40743512e-01 4.49763030e-01
-2.79020578e-01 1.49487004e-01 -6.39322579e-01 -8.31781685e-01
-4.92353179e-02 -1.19750714e+00 -4.09542508e-02 -4.42689598e-01
-3.80954236e-01 -1.74317330e-01 6.15184426e-01 -9.94529366e-01
1.12990022e+00 -1.95523143e+00 4.51839149e-01 -1.92204386e-01
-1.58445299e-01 3.30112159e-01 -5.24318039e-01 6.37128294e-01
1.16770729e-01 -7.94515461e-02 -3.17962468e-01 -9.31862891e-01
8.64978433e-02 3.22427571e-01 -7.89785162e-02 4.63528305e-01
3.60584259e-01 1.19544125e+00 -7.11092055e-01 -4.60380465e-01
-7.77780637e-02 5.31185985e-01 -5.49496293e-01 -6.40485575e-03
-3.03415179e-01 8.46700609e-01 -4.76152487e-02 5.59961498e-01
4.89922673e-01 -1.56141534e-01 4.07654643e-01 -1.55384704e-01
-3.24853659e-01 7.68599629e-01 -7.06926227e-01 2.23044086e+00
-9.63972628e-01 7.23941803e-01 2.38897413e-01 -9.49980974e-01
8.86495054e-01 4.97868747e-01 3.19688618e-01 -9.80322540e-01
-4.49386984e-02 8.46048474e-01 3.59270126e-02 -3.32561284e-01
1.79320186e-01 -4.40450430e-01 -2.98317015e-01 7.78413117e-01
3.62097412e-01 -1.54608950e-01 1.50554880e-01 1.15614515e-02
8.22692275e-01 3.73588473e-01 1.29754707e-01 -5.83110213e-01
4.74470347e-01 -9.31542292e-02 8.13371122e-01 5.33539653e-01
1.34971105e-02 2.81661749e-01 3.02110583e-01 -1.17046140e-01
-1.07764184e+00 -9.77805972e-01 -4.30089496e-02 1.28148866e+00
-2.69520372e-01 -5.14605999e-01 -7.60498524e-01 -7.24089742e-01
7.94715285e-02 5.88696003e-01 -2.83840984e-01 -9.75518487e-03
-9.23350632e-01 -1.15566647e+00 4.93017912e-01 2.19681427e-01
5.06057262e-01 -5.62011003e-01 3.15639824e-01 2.37982571e-01
-5.31211257e-01 -1.38511932e+00 -8.98502231e-01 5.70132613e-01
-8.99729848e-01 -4.16135728e-01 -7.08508134e-01 -9.35470879e-01
3.89956206e-01 1.74332976e-01 1.33578897e+00 -3.57207954e-01
3.09399873e-01 -1.74658239e-01 -1.08211063e-01 -1.82696015e-01
-6.71968222e-01 8.25662673e-01 2.06978440e-01 2.94869940e-04
2.43107915e-01 -6.27834737e-01 -1.83361396e-01 2.47521088e-01
-4.86880213e-01 3.73605877e-01 1.04653263e+00 1.03349388e+00
6.03120387e-01 -4.91893321e-01 6.86526597e-01 -1.00870049e+00
6.86153650e-01 -4.58844632e-01 -4.75005835e-01 4.80458915e-01
-8.52132022e-01 5.44271171e-01 6.48794830e-01 -3.71453881e-01
-8.80637109e-01 -1.16605863e-01 -1.26323730e-01 -1.40114695e-01
3.41409475e-01 7.22459614e-01 -2.10751563e-01 2.65276898e-02
4.82736170e-01 2.81286955e-01 1.13691963e-01 -6.80567563e-01
4.17262882e-01 7.02310979e-01 2.37252936e-01 -8.43226075e-01
8.78583133e-01 -4.80410084e-02 -1.80027023e-01 -3.17589968e-01
-1.05480647e+00 -3.20358783e-01 -6.98713064e-01 1.69634506e-01
7.87095606e-01 -1.40293121e+00 -2.56798208e-01 4.00974691e-01
-1.20565176e+00 -6.75960183e-01 1.27618119e-01 7.21671581e-01
-5.30897319e-01 1.55737057e-01 -8.86101127e-01 -2.35444427e-01
-5.63221693e-01 -1.60549116e+00 1.28588951e+00 -5.12818873e-01
-4.83530052e-02 -1.15433800e+00 1.14180997e-01 6.86319888e-01
6.87770545e-01 -2.64454484e-01 9.84620154e-01 -4.66850787e-01
-5.82282186e-01 1.22999258e-01 -2.50266075e-01 6.72307134e-01
2.54253268e-01 -4.81436521e-01 -7.67939091e-01 -6.35433912e-01
3.48612368e-02 -4.55694497e-01 7.76564300e-01 1.81024686e-01
4.04770225e-01 -2.37750441e-01 -5.99460974e-02 8.21237981e-01
1.26917827e+00 -2.21818149e-01 1.60548463e-01 4.04197276e-01
7.48443782e-01 4.34915900e-01 4.37832385e-01 -7.61683518e-03
8.08783889e-01 1.04525077e+00 -1.53044447e-01 -4.54196632e-01
-4.00646061e-01 -1.61668226e-01 8.94424200e-01 1.83330429e+00
1.43409610e-01 4.04196456e-02 -1.07123113e+00 3.61833304e-01
-1.91818857e+00 -4.03604418e-01 -1.39042959e-01 2.07903457e+00
1.46803451e+00 7.33392760e-02 -1.12559244e-01 -2.58503377e-01
3.75229925e-01 1.22269243e-01 -4.90278602e-01 -4.52464491e-01
-4.92248774e-01 3.26748759e-01 7.81244099e-01 9.29141462e-01
-9.89147365e-01 1.42188513e+00 5.80567122e+00 1.10044122e+00
-1.36910522e+00 6.97936594e-01 5.64520001e-01 -1.53116599e-01
-2.92371243e-01 8.01684633e-02 -1.00344300e+00 4.07077730e-01
1.20488596e+00 -1.13453880e-01 7.01540530e-01 3.25958133e-01
4.38973546e-01 9.29854661e-02 -1.19608557e+00 7.92181551e-01
1.70032695e-01 -1.02732122e+00 -8.65600854e-02 7.77683556e-02
1.05967593e+00 6.03275478e-01 1.85536757e-01 6.57385468e-01
3.78510565e-01 -8.15212965e-01 7.36567318e-01 8.30253661e-02
8.94181371e-01 -6.57508492e-01 6.40976191e-01 5.46961606e-01
-1.01490009e+00 2.99902499e-01 -2.27652676e-02 7.96179771e-02
2.48198107e-01 7.80314982e-01 -7.27155328e-01 6.98022664e-01
4.66705412e-01 8.90307248e-01 -5.10075867e-01 5.25105000e-01
-4.56752986e-01 8.99014175e-01 -2.36430749e-01 3.99409264e-01
6.00096345e-01 -3.60928684e-01 4.17336076e-01 1.52255619e+00
4.80811775e-01 -5.76410830e-01 2.45963395e-01 5.37666380e-01
-4.35906559e-01 3.63480836e-01 -4.53831226e-01 1.00507736e-01
2.16560572e-01 9.98325706e-01 -8.06010440e-02 -3.23105544e-01
-7.70893812e-01 1.23316729e+00 5.82000732e-01 4.17851478e-01
-8.01282585e-01 9.01365951e-02 8.08142126e-01 -1.05535001e-01
1.65184766e-01 -5.93222737e-01 -3.62235963e-01 -1.53232026e+00
1.74473599e-01 -1.32219946e+00 -1.98305808e-02 -3.65601003e-01
-1.31911421e+00 6.83475554e-01 -3.56071442e-01 -1.22551203e+00
-4.67737257e-01 -5.36865592e-01 -1.19759053e-01 1.27085531e+00
-2.00284410e+00 -1.53153253e+00 4.03375298e-01 4.92661506e-01
6.94581747e-01 -3.90604883e-01 8.07529986e-01 7.17582345e-01
-7.15995789e-01 9.24594164e-01 5.71200132e-01 -4.29548174e-02
1.35462594e+00 -1.22400010e+00 6.36989832e-01 8.90817821e-01
2.68515259e-01 5.72450638e-01 3.67441148e-01 -5.17788708e-01
-1.82900143e+00 -1.22774756e+00 1.70318639e+00 -5.57702482e-01
9.08171594e-01 -9.10835683e-01 -8.01391065e-01 8.38221669e-01
6.35800183e-01 -5.43865710e-02 5.89376748e-01 4.62671399e-01
-5.13068795e-01 -3.44386131e-01 -6.50743186e-01 6.77448809e-01
9.19781089e-01 -8.41206014e-01 -3.11983228e-01 5.69292486e-01
8.36029172e-01 -3.68282437e-01 -9.66896772e-01 3.93740922e-01
3.48543525e-01 -5.54619193e-01 9.30758357e-01 -3.72553945e-01
4.67313230e-01 -1.53218657e-01 -3.74255389e-01 -1.75503957e+00
-7.73532242e-02 -9.82124746e-01 1.88491359e-01 1.17026639e+00
9.10517395e-01 -8.33731353e-01 3.87154877e-01 -6.56770170e-02
-3.75662267e-01 -8.31944168e-01 -1.14533508e+00 -1.07872629e+00
6.46417439e-01 -5.17300963e-01 3.28728616e-01 1.09557366e+00
-6.19741641e-02 8.59104872e-01 -8.11989665e-01 -5.09140939e-02
7.58540630e-01 1.09788172e-01 7.93089569e-01 -6.55407250e-01
-4.63032573e-01 -4.41766053e-01 2.79440552e-01 -1.23296630e+00
5.05663216e-01 -1.50506318e+00 2.02325955e-01 -1.27250004e+00
2.48824760e-01 -5.37697077e-01 -2.87569404e-01 5.85600555e-01
-3.92965019e-01 2.80823201e-01 2.76686013e-01 4.11307722e-01
-4.97488081e-01 6.18838012e-01 1.31560516e+00 -1.57053694e-01
-1.09620333e-01 -2.30659589e-01 -6.23180568e-01 1.90891236e-01
7.87082434e-01 -4.55621868e-01 -1.51092663e-01 -1.23854876e+00
6.31209314e-02 9.55173001e-02 -1.80488676e-01 -6.44879222e-01
1.56912491e-01 7.00462610e-02 -7.82741755e-02 -2.96092182e-01
2.85832018e-01 -5.04859149e-01 -3.16390321e-02 2.42770195e-01
-3.34749788e-01 4.35753942e-01 3.83663893e-01 1.75970197e-01
-4.40638155e-01 2.27205679e-01 6.14626467e-01 -2.96366271e-02
-3.40470374e-01 1.57606423e-01 -2.69273907e-01 2.18033388e-01
5.43158650e-01 3.64669651e-01 -2.63489336e-01 -2.13471591e-01
-3.96774620e-01 3.03199708e-01 1.77186191e-01 6.01001740e-01
4.36776243e-02 -1.50033796e+00 -1.23929083e+00 3.38983834e-01
1.27718508e-01 -1.75362408e-01 -2.04913452e-01 1.41856134e+00
-1.31845534e-01 6.70985162e-01 1.64269939e-01 -7.29003727e-01
-1.07081378e+00 1.28097922e-01 1.91272721e-01 -7.26749659e-01
-3.03994983e-01 7.97012687e-01 8.93525779e-02 -1.00626910e+00
1.22563541e-01 -2.75811642e-01 5.48062503e-01 1.65477768e-02
-4.55764569e-02 -2.83468165e-04 4.48916703e-01 -7.04955280e-01
-4.00079221e-01 6.88779652e-01 -2.88437635e-01 -4.43370014e-01
1.32049191e+00 -3.14391226e-01 -2.80149341e-01 5.69548845e-01
1.50730181e+00 1.67047441e-01 -1.16670489e+00 -8.79251122e-01
2.49057308e-01 -1.19982682e-01 1.45633519e-01 -1.15543163e+00
-1.01391256e+00 1.07809973e+00 3.03571075e-01 -5.10175049e-01
1.03937304e+00 1.88757107e-02 9.93529916e-01 3.65595430e-01
4.67273474e-01 -1.09743893e+00 -5.57576418e-02 8.63248944e-01
8.36484551e-01 -1.69117415e+00 -2.18338639e-01 -3.44201177e-01
-5.47385931e-01 7.50696719e-01 3.14702570e-01 2.06189856e-01
4.38443720e-01 5.24555624e-01 4.86950725e-01 2.52826184e-01
-1.11037683e+00 -8.02953318e-02 4.46092546e-01 7.87477344e-02
9.71536934e-01 2.04664811e-01 -4.81673509e-01 3.75854731e-01
-2.30040804e-01 -1.51926652e-01 -9.51787308e-02 8.02372515e-01
-8.33985060e-02 -1.73343623e+00 -2.89446682e-01 1.98912308e-01
-5.66902936e-01 -6.97059095e-01 -1.35946706e-01 6.69479966e-01
8.67878944e-02 9.37129974e-01 -3.98108453e-01 -4.01418388e-01
1.59061506e-01 2.49067456e-01 5.52832782e-01 -6.07022345e-01
-8.07013094e-01 4.48909134e-01 2.14376792e-01 -4.03419048e-01
-2.89070040e-01 -8.40982139e-01 -7.64161944e-01 -4.38482493e-01
-2.43026480e-01 1.14140123e-01 9.55384135e-01 1.18355632e+00
4.70295310e-01 3.28061312e-01 7.01322973e-01 -7.26239383e-01
-7.97053039e-01 -1.09595227e+00 -1.10345269e-02 1.84134960e-01
3.59729707e-01 -2.97756970e-01 -2.92708486e-01 6.91370666e-02]
|
[11.45691967010498, 10.200719833374023]
|
29e1297a-9679-469d-b627-901e1c13adf4
|
tuning-multilingual-transformers-for-language
| null | null |
https://aclanthology.org/W19-3712
|
https://aclanthology.org/W19-3712.pdf
|
Tuning Multilingual Transformers for Language-Specific Named Entity Recognition
|
Our paper addresses the problem of multilingual named entity recognition on the material of 4 languages: Russian, Bulgarian, Czech and Polish. We solve this task using the BERT model. We use a hundred languages multilingual model as base for transfer to the mentioned Slavic languages. Unsupervised pre-training of the BERT model on these 4 languages allows to significantly outperform baseline neural approaches and multilingual BERT. Additional improvement is achieved by extending BERT with a word-level CRF layer. Our system was submitted to BSNLP 2019 Shared Task on Multilingual Named Entity Recognition and demonstrated top performance in multilingual setting for two competition metrics. We open-sourced NER models and BERT model pre-trained on the four Slavic languages.
|
['Alexey Sorokin', 'Mikhail Arkhipov', 'Yuri Kuratov', 'Maria Trofimova']
|
2019-08-01
| null | null | null |
ws-2019-8
|
['multilingual-named-entity-recognition']
|
['natural-language-processing']
|
[-6.82434261e-01 6.50657166e-05 -3.46605986e-01 -5.39492965e-01
-1.20337796e+00 -9.22860503e-01 7.10848212e-01 6.23214692e-02
-1.24885118e+00 1.36821783e+00 4.06195641e-01 -7.50013828e-01
6.46881402e-01 -4.76350427e-01 -7.55665600e-01 -8.58418718e-02
-1.12375543e-01 9.75116611e-01 1.11077344e-02 -3.68782669e-01
-8.48196596e-02 7.37508595e-01 -2.39535823e-01 3.40039521e-01
8.29652309e-01 1.22090027e-01 2.60505021e-01 5.22676289e-01
-3.91111851e-01 7.22394705e-01 -6.62835419e-01 -6.39294505e-01
3.21426302e-01 1.17294647e-01 -1.24429202e+00 -6.30354524e-01
4.47103232e-01 2.10171953e-01 -1.30407482e-01 8.74394596e-01
6.19012833e-01 3.07869036e-02 5.94325066e-01 -6.22086585e-01
-6.74303412e-01 1.44622850e+00 -3.81821543e-01 3.13301623e-01
1.79724693e-01 -2.03070566e-01 1.15271330e+00 -1.11413813e+00
1.26441002e+00 1.18472099e+00 8.19516182e-01 4.55242991e-01
-8.82720530e-01 -9.64725792e-01 7.09753409e-02 1.07699856e-01
-1.63572860e+00 -6.43107951e-01 1.58846185e-01 -3.64175707e-01
1.70459175e+00 -3.09256017e-01 8.80647898e-02 9.83460724e-01
1.15812324e-01 8.92516375e-01 1.41070771e+00 -6.47449613e-01
-1.78924412e-01 3.69845092e-01 4.71842289e-01 3.83753240e-01
2.67563552e-01 2.13465020e-01 -2.83791065e-01 7.63124153e-02
7.09926605e-01 -8.42350185e-01 1.33696407e-01 2.24395007e-01
-1.42570269e+00 6.42167091e-01 3.04810822e-01 8.94378543e-01
-3.82898986e-01 -5.90714067e-02 7.32576966e-01 5.79906642e-01
4.08543617e-01 7.64383614e-01 -1.32515860e+00 6.18145578e-02
-9.82567847e-01 -1.24084204e-01 9.91076827e-01 1.45264935e+00
7.49948859e-01 3.35200608e-01 -3.68874781e-02 1.16684103e+00
3.15866113e-01 5.70253015e-01 5.06093442e-01 -2.10544035e-01
8.42479110e-01 1.17262684e-01 7.88190290e-02 -1.10612176e-01
-4.50782299e-01 -4.25570339e-01 -6.20337486e-01 -1.34434670e-01
3.78476620e-01 -6.48856997e-01 -1.15329111e+00 1.69862449e+00
1.44146606e-01 1.28793433e-01 6.54052019e-01 3.23146731e-01
1.03611267e+00 7.57237077e-01 6.08018279e-01 1.31589577e-01
1.54013908e+00 -1.20361865e+00 -6.64526045e-01 -1.34901568e-01
8.44352245e-01 -9.69701946e-01 4.23032850e-01 7.84455389e-02
-8.87099802e-01 -5.77158213e-01 -8.85175467e-01 -1.93249166e-01
-1.03522146e+00 5.63186824e-01 5.42130232e-01 7.77876377e-01
-1.22876561e+00 2.91971087e-01 -8.47636998e-01 -5.28643072e-01
-7.47661814e-02 4.73249584e-01 -9.47352946e-01 6.40839478e-03
-1.54309762e+00 1.30613458e+00 1.19438875e+00 1.06440231e-01
-9.88660574e-01 -5.33942461e-01 -1.24036908e+00 -2.87809640e-01
-3.12118948e-01 -6.44616187e-02 1.18894613e+00 -5.93120575e-01
-1.48698521e+00 1.50622714e+00 -1.03967354e-01 -6.92766905e-01
2.68854052e-01 -4.24712151e-01 -8.32331538e-01 -3.62419575e-01
5.09652495e-01 7.98914313e-01 -1.67829409e-01 -9.71056581e-01
-6.87735081e-01 -5.65505214e-02 -1.76191285e-01 2.02700317e-01
1.37934402e-01 7.39124715e-01 -4.02143002e-01 -5.85900545e-01
-5.25516152e-01 -9.87975538e-01 -5.90805233e-01 -1.34952235e+00
-4.26042259e-01 -3.43315274e-01 1.66055858e-01 -1.23393261e+00
9.02156532e-01 -1.79005063e+00 -3.10347021e-01 -1.14131980e-01
-4.17404413e-01 5.50918400e-01 -4.26117718e-01 7.69307315e-01
-2.63196290e-01 3.03786039e-01 1.27124339e-01 -3.23881298e-01
-1.48282588e-01 1.59842446e-01 -2.32083187e-01 3.69552404e-01
5.20053566e-01 1.06470907e+00 -8.11565995e-01 -5.64871311e-01
7.70134851e-02 5.23262978e-01 -1.85196027e-01 1.68451831e-01
1.12543225e-01 9.09569383e-01 -1.51717201e-01 4.08068359e-01
8.81936133e-01 3.45090449e-01 3.06611627e-01 7.38114417e-02
-5.57374895e-01 7.75531232e-01 -1.07646477e+00 2.01319146e+00
-9.67953861e-01 3.92362088e-01 4.14112322e-02 -6.31018817e-01
1.18743718e+00 8.61155510e-01 -1.47389686e-02 -4.27818865e-01
3.43944132e-02 8.24094534e-01 -1.43833056e-01 -3.45046334e-02
7.77242720e-01 -3.70264232e-01 -7.51699030e-01 -5.20132557e-02
9.95572865e-01 7.04389811e-02 2.11152315e-01 9.12150294e-02
8.32617402e-01 3.88488978e-01 8.96056116e-01 -6.12411737e-01
6.99324250e-01 3.23737830e-01 8.88020515e-01 7.07301974e-01
-5.30700445e-01 2.78132528e-01 -1.26345828e-01 -3.57894391e-01
-1.05770361e+00 -1.21210086e+00 -4.31513876e-01 1.32015061e+00
-3.89299005e-01 -1.02369718e-01 -3.56852829e-01 -1.06281877e+00
-3.87465775e-01 1.00631022e+00 -3.59692395e-01 7.24101067e-01
-1.20474780e+00 -5.88900566e-01 1.49929380e+00 4.03408587e-01
5.15833914e-01 -1.62844396e+00 2.79598922e-01 6.60121381e-01
-2.48746965e-02 -1.60465503e+00 -3.75726551e-01 6.99222386e-01
-6.26903534e-01 -6.58945322e-01 -1.00938392e+00 -1.50304234e+00
7.98071176e-02 -5.24980903e-01 1.57374907e+00 -7.07436860e-01
5.04987128e-02 -1.05991215e-02 -2.77026564e-01 -5.62009573e-01
-5.43540418e-01 7.08760560e-01 -3.65464799e-02 -5.13265133e-01
5.84337413e-01 -3.86611640e-01 1.08514950e-01 1.55298663e-02
-4.22207624e-01 -1.56171605e-01 7.00292766e-01 7.78676927e-01
4.55919415e-01 -5.97560167e-01 1.06003606e+00 -1.30410969e+00
1.72102973e-01 -5.22235453e-01 -6.09312594e-01 4.75068539e-01
-1.41601890e-01 3.36952716e-01 6.15017474e-01 1.42302383e-02
-1.52049482e+00 3.83397460e-01 -7.40091085e-01 1.56925619e-01
-7.80450284e-01 5.83703816e-01 -5.13999403e-01 2.28288546e-01
5.59837699e-01 -4.62872311e-02 -1.25247037e+00 -7.85318911e-01
8.79130602e-01 9.25617218e-01 8.98419738e-01 -7.49149799e-01
5.24140298e-01 -1.82452984e-02 -5.76549709e-01 -8.41673017e-01
-8.56219947e-01 -1.03416753e+00 -1.35839438e+00 2.65707940e-01
1.21962166e+00 -1.51169777e+00 1.29889511e-02 4.04000729e-01
-1.77590656e+00 -2.66243279e-01 -5.56274801e-02 7.27889180e-01
-2.93429852e-01 -3.07548232e-02 -1.35146046e+00 -5.19158542e-01
-6.63088679e-01 -7.65921414e-01 9.47912574e-01 1.01955205e-01
2.15096418e-02 -1.47477686e+00 9.29549456e-01 6.08452186e-02
2.54363775e-01 1.59109324e-01 6.16879702e-01 -1.41279888e+00
7.28848130e-02 -4.80766259e-02 -5.23076542e-02 3.06836039e-01
-1.97186992e-01 -4.89029765e-01 -1.08186150e+00 -3.23937505e-01
-4.51171041e-01 -5.55965841e-01 8.43693197e-01 1.72248241e-02
-2.25486428e-01 7.32463738e-03 -3.01705331e-01 5.73347270e-01
1.75018883e+00 3.79992366e-01 3.96185011e-01 6.39910042e-01
6.79327071e-01 4.61617023e-01 3.15876931e-01 -2.80150235e-01
5.27190506e-01 3.29942912e-01 -3.53891104e-01 -2.46739641e-01
-2.27775142e-01 -3.00715178e-01 5.38489640e-01 1.57916284e+00
-3.57046202e-02 3.25150862e-02 -1.51656950e+00 1.06421435e+00
-1.54453433e+00 -7.15120137e-01 -2.69142419e-01 1.94345605e+00
1.22738671e+00 -5.72986677e-02 -2.42858291e-01 -8.91883552e-01
9.06232595e-01 -2.24606302e-02 7.50705451e-02 -6.92777991e-01
-6.02365494e-01 8.13966751e-01 8.94411266e-01 8.18241715e-01
-1.61427593e+00 2.04279423e+00 6.25544930e+00 9.41946685e-01
-1.03474927e+00 7.08732486e-01 3.11237633e-01 6.49144709e-01
8.09544474e-02 2.08084434e-01 -1.64058030e+00 -2.05867931e-01
1.65572858e+00 -1.72241732e-01 4.77437265e-02 8.37985814e-01
-1.11574493e-01 4.05856997e-01 -1.00167608e+00 7.06991792e-01
-1.13676690e-01 -1.03341556e+00 -6.07820004e-02 -1.08096473e-01
1.17643261e+00 1.21541870e+00 -6.32120788e-01 9.44067657e-01
1.29383790e+00 -1.12720966e+00 5.10768056e-01 1.95184737e-01
1.04049468e+00 -8.73026073e-01 1.15019763e+00 2.82667935e-01
-1.40750563e+00 6.95510447e-01 -6.37781322e-01 3.49381745e-01
5.11401892e-01 2.93386400e-01 -1.25738180e+00 9.73962545e-01
3.09610277e-01 6.35623634e-01 -4.98759359e-01 1.19727647e+00
-6.19958699e-01 1.02488458e+00 -2.95126736e-01 3.41861427e-01
5.81004918e-01 1.50297627e-01 5.08164585e-01 2.25904608e+00
-2.95221452e-02 -4.45494115e-01 4.10001189e-01 1.53435513e-01
-4.28628772e-01 8.70830238e-01 -8.62928569e-01 -8.98750201e-02
3.52975756e-01 1.25723636e+00 -5.95746100e-01 -6.15471721e-01
-4.59304154e-01 1.20575011e+00 1.01712453e+00 3.13944638e-01
-3.08810145e-01 -7.66340077e-01 1.69339433e-01 -4.73892123e-01
2.04786092e-01 -5.34232914e-01 -2.34441325e-01 -1.53694582e+00
-6.01382077e-01 -8.31306100e-01 5.02307355e-01 -2.70619482e-01
-1.52356243e+00 1.25868297e+00 -2.03632668e-01 -7.07603753e-01
-2.82076985e-01 -9.28600490e-01 -4.38678533e-01 1.20850575e+00
-1.72724915e+00 -1.65076292e+00 5.85795045e-01 5.82426071e-01
4.67671573e-01 -5.62586129e-01 1.22611904e+00 6.36217594e-01
-3.15761298e-01 6.81750238e-01 2.47171357e-01 9.16604638e-01
1.21360409e+00 -1.44921756e+00 1.04181170e+00 7.85634816e-01
6.72820330e-01 6.97644234e-01 2.04074994e-01 -7.67509997e-01
-4.82100189e-01 -1.30182958e+00 1.94521213e+00 -6.39996707e-01
1.04758394e+00 -5.51963866e-01 -6.71861172e-01 1.29735780e+00
9.81611073e-01 -1.93295598e-01 8.64209652e-01 5.74558914e-01
-5.95032990e-01 4.53610808e-01 -9.30289268e-01 2.41281867e-01
7.07140982e-01 -6.84226036e-01 -9.88470972e-01 4.47640240e-01
8.17787766e-01 -1.55178785e-01 -1.20876050e+00 2.71861374e-01
2.93901145e-01 -2.94427186e-01 6.83598161e-01 -1.07832599e+00
-1.71508357e-01 -2.01148782e-02 -2.58060068e-01 -1.50770795e+00
-2.12742865e-01 -4.90539253e-01 8.17972779e-01 1.60994458e+00
9.68706727e-01 -6.18187249e-01 2.80798078e-01 -4.24065255e-02
-1.85683280e-01 7.30353966e-02 -1.12405431e+00 -1.00712156e+00
9.00334001e-01 -6.55226350e-01 2.97379434e-01 1.50840223e+00
-5.72476014e-02 9.10632908e-01 -4.32477802e-01 3.65960240e-01
3.07965100e-01 -4.28794920e-01 5.73107004e-01 -1.17234576e+00
-2.95591891e-01 -1.39289975e-01 -3.96947205e-01 -7.92049825e-01
8.03092837e-01 -1.56753290e+00 3.28237474e-01 -1.54385090e+00
1.76338717e-01 -7.68807769e-01 -6.65705502e-01 7.72794068e-01
-9.08212811e-02 2.75385261e-01 3.77647996e-01 2.80263036e-01
-6.09411836e-01 2.53109366e-01 5.08112729e-01 1.20507171e-02
-3.12198848e-01 -2.74075955e-01 -2.04113871e-01 6.30356848e-01
8.11116457e-01 -8.44624758e-01 5.36369443e-01 -7.43210971e-01
1.83098540e-01 1.07484378e-01 -4.35091525e-01 -9.63045478e-01
7.63914138e-02 -1.96350254e-02 2.23137543e-01 -6.64289057e-01
-1.67277381e-01 -4.77769554e-01 -9.27137136e-02 1.95764646e-01
-2.77679712e-01 2.26947367e-01 6.00462437e-01 5.65717667e-02
-5.28628469e-01 -3.76719803e-01 9.60815549e-01 -3.42144728e-01
-9.82140124e-01 4.72905263e-02 -4.56002533e-01 3.88009310e-01
7.16979623e-01 5.59430420e-01 -1.21803209e-01 1.32737234e-01
-1.19715488e+00 3.06455940e-01 5.99911027e-02 4.87510949e-01
-1.81410462e-01 -1.25239325e+00 -1.20683706e+00 -7.33768567e-02
1.14909388e-01 -5.50110877e-01 -1.82184279e-01 4.93761778e-01
-8.13923001e-01 1.07774019e+00 -4.15599167e-01 -1.28335029e-01
-8.60025048e-01 3.01032722e-01 3.06049854e-01 -1.24234486e+00
-2.18362138e-01 8.07672262e-01 1.79599687e-01 -1.78026187e+00
-3.99435386e-02 3.96763049e-02 -4.84357119e-01 -7.20653161e-02
3.17626595e-01 1.53642789e-01 2.96994537e-01 -1.17969263e+00
-7.07163095e-01 3.48479718e-01 -3.19403291e-01 -7.12130129e-01
1.35515094e+00 7.78429070e-03 6.97653368e-02 7.69556761e-01
1.22521234e+00 6.81711376e-01 -3.99100691e-01 -3.49449098e-01
6.97608829e-01 4.52224672e-01 -3.10614258e-01 -1.20319736e+00
-6.80898190e-01 8.87433112e-01 3.77710402e-01 -7.03509629e-01
5.90012908e-01 1.91793703e-02 7.56377161e-01 7.96433151e-01
8.97328377e-01 -9.99847591e-01 -8.27685714e-01 1.47779715e+00
5.81968486e-01 -1.25043714e+00 -5.01980543e-01 -1.20275654e-01
-7.99945295e-01 8.86859059e-01 4.40795749e-01 -6.17477119e-01
8.66160631e-01 5.81605732e-01 6.85811222e-01 9.53705087e-02
-5.25890887e-01 -3.14357102e-01 3.33616614e-01 5.26581943e-01
1.06638789e+00 4.72008049e-01 -7.75915325e-01 5.68978310e-01
-3.63677889e-01 -2.51965612e-01 3.48200500e-01 9.46609676e-01
-2.12776423e-01 -1.46975470e+00 -3.07072163e-01 -2.06887841e-01
-1.29417431e+00 -8.66679013e-01 -3.91559631e-01 1.14584231e+00
2.19487786e-01 8.13829422e-01 -2.60457367e-01 3.82172689e-02
3.45726043e-01 5.98710835e-01 1.18754372e-01 -9.35202539e-01
-1.22368491e+00 1.53420016e-01 6.02359533e-01 -1.08420111e-01
-3.29789281e-01 -6.18751347e-01 -1.37679887e+00 3.43250275e-01
-3.04687768e-01 6.77690744e-01 7.70207405e-01 9.62801039e-01
2.19212398e-01 4.57132980e-02 1.78231359e-01 -6.15113258e-01
-6.78384826e-02 -1.39116609e+00 -6.14027262e-01 1.09777287e-01
-6.44033849e-02 -1.60239249e-01 7.45772794e-02 1.90939933e-01]
|
[9.986825942993164, 9.856061935424805]
|
929327f2-3aaf-419d-abe7-9a2dcebf41bb
|
opinion-spam-recognition-method-for-online
|
1807.11024
| null |
http://arxiv.org/abs/1807.11024v1
|
http://arxiv.org/pdf/1807.11024v1.pdf
|
Opinion Spam Recognition Method for Online Reviews using Ontological Features
|
Nowadays, there are a lot of people using social media opinions to make their
decision on buying products or services. Opinion spam detection is a hard
problem because fake reviews can be made by organizations as well as
individuals for different purposes. They write fake reviews to mislead readers
or automated detection system by promoting or demoting target products to
promote them or to damage their reputations. In this paper, we pro-pose a new
approach using knowledge-based Ontology to detect opinion spam with high
accuracy (higher than 75%). Keywords: Opinion spam, Fake review, E-commercial,
Ontology.
|
['V. M. Ngo', 'L. H. Nguyen', 'N. T. H. Pham']
|
2018-07-29
| null | null | null | null |
['spam-detection']
|
['natural-language-processing']
|
[-0.21535438 0.3774083 -0.11628152 -0.35599458 0.14038733 -0.6330953
0.5009818 0.74402165 -0.13888776 0.8801478 -0.21014097 -0.38066313
0.34021792 -1.2017398 -0.2517093 -0.21911736 0.6391769 0.7099188
0.9065988 -0.8353332 0.9675474 0.22005716 -1.3261628 0.43045154
1.1405032 0.78649545 -0.516579 0.06340156 -0.4489359 0.77119124
-1.1192675 -1.3060657 0.08475192 -0.4629628 -0.3472517 0.2937694
0.199398 -0.06272673 0.30382937 1.8340739 0.06570557 -0.38095063
0.5949814 -1.4644561 -1.3870227 0.4484153 -0.47008523 -0.01141255
0.26304346 -0.20217155 0.5830444 -0.75840163 0.63858557 1.5968286
0.51803946 0.3564287 -0.644044 -0.7274729 0.02061374 0.34933764
-0.828336 -0.134918 0.3484269 -0.33228198 0.52111995 0.11405408
1.0195109 0.5699233 0.8901872 0.47808465 1.3371538 -0.23341863
0.4465542 1.1650432 0.44012538 0.5840311 1.1277009 -0.2047477
-0.7275099 -0.6275263 -0.13197085 0.10079748 0.09937762 -0.1284922
-0.42311504 1.1895338 0.30669633 0.2963753 -0.23473082 -0.17022452
0.52020717 0.8377908 0.930669 0.6961172 -0.38577393 0.15735658
0.01919028 0.26827005 1.1154402 0.58482957 0.6327225 -0.1737079
0.31488368 0.6349073 0.7548565 0.8746522 0.9650668 -0.35003787
-0.0171141 1.2087328 0.69297004 -1.8297327 0.07011238 -0.21940157
-0.34269616 0.14288336 0.13740171 0.13617279 -0.6567442 0.6073991
0.3329969 -0.3802753 0.01415211 0.97584015 0.6836941 0.4828405
-0.06665603 -0.17933577 1.6277674 -1.2034909 -1.1911896 -0.3988545
0.70187795 -1.3032256 0.49933255 0.8721387 -0.6841548 -0.03200163
-1.02848 0.27894372 -1.2181838 -0.2768814 0.8747305 1.0520787
-0.40138325 0.6464199 -0.20495498 -0.5156458 0.5792987 0.38460794
-0.1988689 0.02790389 -1.5280477 1.4522706 0.00678485 0.03321703
-0.1734649 0.12899376 -0.29076156 -0.2911049 0.20318787 -0.23582187
1.0398898 -1.6941768 -1.304368 0.9889498 -0.09696904 -0.60923904
0.38496232 -0.18027872 -1.0828452 0.18196383 0.34797093 -0.00961022
1.2963463 -0.8457554 -0.67175055 -0.89183545 0.05993979 -0.01840456
-0.3438672 0.26793018 0.5499872 -0.42071763 0.5506964 -0.821633
-0.03560542 -0.35003334 -0.25173435 -0.47561783 1.2987641 -0.46356338
1.0219873 -1.5812708 -0.47869873 0.40713736 0.22595885 0.90389556
0.3068705 0.34764507 0.4044179 0.61561817 -0.007892 0.38286468
-0.16217233 -0.11946108 -0.25092515 0.4519959 0.09590442 0.9006998
-1.2684762 -0.20963904 0.05371865 0.10045112 0.14955863 -0.34245965
0.07731704 -0.14396593 -0.8267023 1.0207487 0.71472865 -0.24414325
0.1629573 0.00753866 0.2127509 0.45659685 -0.81695646 0.30830038
-0.3845325 0.65939075 0.01665232 -0.93353075 1.1537209 0.17070502
-0.2122862 -0.32141507 0.6488012 0.948793 0.09158443 -0.6268084
0.71369356 -0.48465464 0.23076162 0.7133287 -0.35222596 -0.34245688
0.27759197 0.39847633 0.98694503 -0.60687923 0.46593004 -0.32714155
1.0658194 0.43657747 0.12118568 0.48550323 -0.64437145 -0.03187428
0.57217956 -0.64939946 -0.77576417 -0.53479075 -0.04792484 0.5196682
0.5037564 -0.1469645 -0.52923566 -1.2362887 0.50521755 0.53706914
-0.32350877 -0.6437751 0.05158225 -0.78612936 -0.08156848 -0.3309076
0.8027064 -1.0241345 0.04661694 0.46561435 0.02673987 -0.78393924
-0.09058265 -0.24857408 -0.9567487 -1.2857848 -0.41248384 -0.6726145
1.2352929 0.70542425 0.99186033 0.6209959 0.20311622 -0.36026788
-0.92353326 -0.76497334 -0.9557057 0.03434283 0.33376646 0.17664006
1.3463223 -0.15767081 -0.38826275 0.8415811 -0.9094548 -0.7707312
0.53911185 0.6512946 -0.31234002 0.40965208 1.286518 -1.4837257
1.3246566 -0.47309566 -0.80081236 0.16356228 -1.1108727 -0.35869557
0.6899217 -0.3897612 -0.71797085 -0.45518515 0.23865901 0.47051102
0.30263388 0.21946208 0.17719524 -0.4213534 0.8419856 -0.07937014
0.24134669 -0.2668095 0.09149064 1.3334754 -0.31002185 0.35267216
0.9503137 0.6775493 -0.26109606 -0.52292824 -1.1872687 -0.56559515
-0.15974458 -0.19332573 0.42167136 -0.44720182 -0.68685544 0.75601333
-1.3913475 0.70677876 0.33630353 0.27418888 0.41824296 0.46890306
-0.55072093 -1.1140871 -0.42016104 -0.8431723 0.5281813 0.17544997
-0.11529025 -0.7890573 -0.37837133 0.8292179 0.59404176 -0.36366457
0.37469065 -0.9692092 -0.5266054 -0.8817982 -0.31792232 0.79527575
0.38588563 -0.11935963 -0.41370493 0.03561415 0.10726878 -0.03985459
0.5686376 -0.17460676 0.27611175 -0.48280796 -0.35315454 -0.6534526
1.2448583 0.3377734 0.89283544 0.57393914 0.23594117 0.9415438
1.0286771 0.05704288 -0.04509916 0.35110766 0.42378837 0.70772904
0.514876 -0.19493662 0.47882825 0.88955146 0.20774826 -0.3501466
-0.44898894 0.5524394 -1.9749252 -0.6774393 -0.89117855 2.1661382
0.41743723 0.67458946 -0.17164472 0.22361457 1.225713 -0.20582776
-0.43169162 -0.91179764 -0.03954729 -0.00632313 0.9462377 0.775394
-0.81808287 1.5097218 5.8433294 0.70644957 -1.0260514 0.5975789
0.5712779 0.6273879 -0.27129102 0.31381917 -0.93141806 0.7359674
0.63431865 -0.34381878 0.17618443 0.9623059 0.46881542 -0.6343198
0.08747903 0.5661427 0.49299228 -1.1125082 0.09482326 -0.02989425
0.9111957 -0.53949213 -0.33777565 0.0217513 0.37852073 -0.6561041
0.36414418 0.41018993 -0.4763559 -0.61121684 1.4732889 0.18216328
-0.43767512 -0.06624405 -0.71997344 -0.65616584 0.6519891 1.4034696
-0.6436325 0.01911782 0.6023432 0.71767384 -0.8392583 0.77522075
-0.9051904 0.5040951 -0.2565643 -0.9958676 0.04832058 -0.7091827
0.53512686 0.2832284 0.0523551 -0.10918415 -0.27549234 0.5799789
-0.12711833 0.51494664 -0.88995904 -0.421935 0.22913535 1.2371585
-1.0516658 -0.8600009 -0.20618026 1.0841479 -0.24088809 -0.05882017
-0.5990022 -0.714925 0.2814888 0.7013183 -0.0595293 0.1910518
-0.23241766 -1.2278762 0.0217845 -1.0662628 -0.3619961 -0.78422976
-1.5330904 0.5386903 -0.7567786 -1.1791759 0.4138865 -0.86951566
-0.46476194 0.46040666 -1.2058028 -0.61767113 0.00728667 -0.20408633
0.28595304 -0.5865099 0.39962193 0.32994458 -0.09973738 -0.1442596
0.02470023 -0.21610393 0.9211623 -0.99472255 0.16620725 0.35631323
-0.33399045 0.8776094 1.0873859 -1.1200415 -0.7782023 -0.5343973
1.5104865 -0.6652285 1.094144 -0.05585321 -0.6311837 0.05868631
0.11520464 -0.24791218 0.62766063 -0.03202778 -0.11861169 -0.24740084
-1.6026661 0.47800928 0.5291975 -0.4081112 -0.83807635 1.0000645
0.43181056 0.21535751 -0.47203642 -0.04018537 0.67722875 -1.2941562
0.4349237 -0.6999107 0.4646381 -0.6969845 0.4327272 -1.308116
0.22659172 -0.2949234 0.01649001 0.81476396 0.6731843 -1.3049296
1.0151839 0.30352882 0.2370015 -0.38948974 -0.40602964 -0.7710866
-0.26751062 0.01189919 0.4587474 1.0173289 0.36027136 0.69648755
-0.3459332 -0.02466871 0.48091143 -0.23543064 0.67444134 -1.4399941
0.11244062 -0.16764025 -0.8042568 -0.90750325 -0.08613951 -0.35626718
-0.31724307 -1.3974805 -0.29902953 -0.01326748 0.05543516 0.06396617
-0.04962273 0.6301122 -0.13823918 0.45131943 -0.8585229 0.28101704
1.4189516 -0.26846096 -0.08552245 0.3645646 -1.0027438 1.1624404
0.9894156 -0.92460364 0.01152859 0.13116583 1.0889268 -0.5447133
0.12536795 -0.4519534 0.09870907 -0.02503962 -0.25939023 -0.4745681
-0.0990274 -0.8218395 -0.21946739 0.7963135 0.15931977 0.18849415
-0.4761986 0.7805234 -0.49419948 -1.0552442 0.85213387 -0.68076056
-0.22665222 -0.3441159 -1.0301467 -0.5175368 0.93385977 -0.13463163
-0.889494 -0.8230582 -0.601503 0.23132025 0.5913271 0.70241225
0.367556 -1.1254752 -0.39434746 -0.1482558 0.39009497 -0.75119746
-0.31066912 0.7999078 -1.025649 0.32123157 0.05697625 -0.05872807
-1.173491 0.41945246 0.28167644 -0.12924981 0.24644297 0.6042927
-0.63986343 -0.5710627 -0.277896 0.22124155 -0.5085101 0.23347065
0.6742854 0.62078834 0.33966056 -0.95364654 -0.23266871 0.2617502
-0.33176726 0.18804444 0.8694997 -0.14949712 -0.9460713 0.17442793
0.9021945 0.39320385 0.06302475 0.26728502 0.203118 -0.92949086
-0.00881975 -1.161417 -0.8758459 0.45562 0.5777026 1.2518259
0.22087926 -0.12732579 0.8656313 0.773108 0.545443 -1.7253095
0.20008561 0.3028245 0.6588035 -1.8016044 0.31477246 -0.9964769
-0.5400361 1.0690236 0.5243647 -0.4791063 1.055241 -0.61351424
0.46180668 -0.5751628 -0.41608536 -0.09585138 0.01309428 0.676742
0.53789204 0.06623916 -1.6820197 0.34296566 0.05558383 0.10766637
1.0211856 0.89814293 -1.3171973 -1.6608981 -0.6429196 0.9814043
-0.66930777 -0.12360161 -1.0195293 0.3128732 0.29658872 1.6819277
-0.50611985 -0.34861824 0.20985295 -0.01303201 -0.13975258 -0.7108493
-0.7484877 -0.4628061 0.46553132 -0.17997214 -0.8340266 -0.28170696
-0.801107 -0.60381615 -1.3205867 0.6147853 0.95562464 1.0761056
0.30795836 -0.11049601 0.6044829 0.08693126 -0.617404 -1.2237058
-0.8118039 0.61039054 -0.44699556 -0.828075 -1.0273441 -0.30240768]
|
[7.861884117126465, 10.054539680480957]
|
36c150b0-afa3-43d5-bde5-bd44b2907f69
|
fastwordbug-a-fast-method-to-generate
|
2002.0076
| null |
https://arxiv.org/abs/2002.00760v1
|
https://arxiv.org/pdf/2002.00760v1.pdf
|
FastWordBug: A Fast Method To Generate Adversarial Text Against NLP Applications
|
In this paper, we present a novel algorithm, FastWordBug, to efficiently generate small text perturbations in a black-box setting that forces a sentiment analysis or text classification mode to make an incorrect prediction. By combining the part of speech attributes of words, we propose a scoring method that can quickly identify important words that affect text classification. We evaluate FastWordBug on three real-world text datasets and two state-of-the-art machine learning models under black-box setting. The results show that our method can significantly reduce the accuracy of the model, and at the same time, we can call the model as little as possible, with the highest attack efficiency. We also attack two popular real-world cloud services of NLP, and the results show that our method works as well.
|
['Wang minghua', 'Dou Goodman', 'Lv Zhonghou']
|
2020-01-31
| null | null | null | null |
['adversarial-text']
|
['adversarial']
|
[ 8.55717883e-02 8.88068229e-02 -2.02395201e-01 -1.91635534e-01
-7.99602509e-01 -7.35478997e-01 4.97439981e-01 4.72674042e-01
-3.20825934e-01 6.04007363e-01 2.63606995e-01 -5.87173641e-01
5.05519748e-01 -7.12260723e-01 -7.26628840e-01 -5.05180597e-01
3.08618218e-01 4.27019835e-01 2.89078474e-01 -4.57992166e-01
5.90335310e-01 1.96346164e-01 -1.13997400e+00 6.81231618e-01
6.90396965e-01 6.26478732e-01 -3.01384628e-01 8.59609008e-01
-3.05826217e-01 9.89990771e-01 -1.01413071e+00 -7.65583813e-01
3.26624691e-01 -7.68763870e-02 -8.36218894e-01 -3.20643425e-01
9.37139094e-02 -3.18703294e-01 -3.00675422e-01 1.29682124e+00
8.07271004e-01 -7.10837245e-02 3.60595763e-01 -1.37145925e+00
-5.78525484e-01 1.21544850e+00 -6.88440204e-01 1.90858960e-01
5.40346026e-01 2.54981846e-01 1.19448328e+00 -8.72640789e-01
1.86841086e-01 1.38344097e+00 5.10455251e-01 7.51815915e-01
-1.04163456e+00 -9.72938418e-01 5.67323089e-01 1.29477188e-01
-1.06184864e+00 -2.22763345e-01 5.73960662e-01 -7.40097985e-02
1.19523418e+00 4.51885939e-01 1.25747278e-01 1.48378932e+00
4.29973006e-01 9.61387396e-01 8.77032161e-01 -4.48627681e-01
4.63285714e-01 1.76366508e-01 3.74376595e-01 3.95131111e-01
4.81855750e-01 -4.62289676e-02 -7.70310581e-01 -8.98242593e-01
-4.32777584e-01 8.79775211e-02 -3.92045289e-01 2.76414841e-01
-9.16247904e-01 8.93924475e-01 -2.57875770e-01 -2.64153164e-03
-1.99320853e-01 2.99056441e-01 6.13624275e-01 2.74181634e-01
7.67631769e-01 3.01977754e-01 -9.31228518e-01 -3.76278073e-01
-6.98356926e-01 3.69908512e-01 1.19066560e+00 8.97624195e-01
3.70852381e-01 -3.45034264e-02 -3.39763999e-01 4.28188026e-01
-1.18108308e-02 7.80652642e-01 8.74944985e-01 -2.57161885e-01
7.23039627e-01 3.56881976e-01 2.41248950e-01 -1.13265395e+00
-2.39502728e-01 8.82828459e-02 -6.90965235e-01 -4.21430394e-02
1.25608385e-01 -4.02485698e-01 -6.56263411e-01 1.51594865e+00
2.73803502e-01 3.43952507e-01 -9.24777389e-02 4.54315782e-01
2.05508158e-01 8.06997180e-01 -1.95976317e-01 -2.33380601e-01
1.33845365e+00 -1.09492648e+00 -7.20287204e-01 -5.91960311e-01
1.01849294e+00 -7.28944719e-01 1.63979816e+00 7.52204955e-01
-8.05221438e-01 6.51792511e-02 -9.32358265e-01 3.96472663e-01
-4.57773656e-01 -3.00592870e-01 3.73708487e-01 8.45066547e-01
-6.23714447e-01 5.33073843e-01 -6.72735572e-01 3.63707915e-02
2.22824454e-01 1.39472678e-01 -6.78859130e-02 2.45147020e-01
-1.42755592e+00 8.04932415e-01 3.14951837e-01 -4.54135835e-01
-6.51980221e-01 -8.20589960e-01 -4.01210666e-01 3.20446491e-01
5.49594343e-01 -3.65696579e-01 1.58766747e+00 -8.09425294e-01
-1.64522851e+00 5.62557280e-01 -2.71228492e-01 -5.28143644e-01
4.41859156e-01 -5.28156400e-01 -4.05651540e-01 4.86550443e-02
-2.27205381e-01 -1.05586387e-01 1.03560650e+00 -1.01366687e+00
-6.43038332e-01 -3.37104529e-01 -1.30677819e-01 -2.27483228e-01
-1.00096834e+00 4.23097759e-01 -8.02131221e-02 -8.81699622e-01
-5.15463412e-01 -8.31369400e-01 -4.14393783e-01 -4.27517951e-01
-8.62389207e-01 -2.66987532e-01 7.58031905e-01 -4.64973122e-01
1.66539121e+00 -1.86496174e+00 -2.04717368e-01 4.07509446e-01
2.68794864e-01 4.73539799e-01 -1.96160719e-01 6.59141660e-01
-2.34759986e-01 7.68609762e-01 -1.52474076e-01 -3.56027752e-01
2.29298621e-02 7.14307325e-03 -1.23017359e+00 3.06145102e-01
-7.44358897e-02 9.42014158e-01 -8.71393025e-01 -1.88598439e-01
-1.82217553e-01 -1.22243412e-01 -7.82619059e-01 1.24301486e-01
-6.11098468e-01 -2.16463521e-01 -5.33071220e-01 4.38983798e-01
6.28300667e-01 -2.83731997e-01 1.07597575e-01 1.97054669e-01
4.61389184e-01 5.52919686e-01 -1.01136696e+00 1.01349175e+00
-4.95897770e-01 4.34319615e-01 -1.88525602e-01 -8.03732932e-01
6.58898473e-01 8.64871591e-02 6.72897249e-02 -2.91695386e-01
3.33895624e-01 -4.04166169e-02 -2.60322839e-01 -5.50850451e-01
4.21316326e-01 1.94776565e-01 -2.05674246e-01 1.10310888e+00
-4.35393065e-01 -1.54617801e-01 -1.22725762e-01 5.68336308e-01
1.38584185e+00 -5.52197397e-01 5.77207386e-01 2.99940817e-03
4.78588074e-01 -2.22040281e-01 2.88021415e-01 1.21187782e+00
-2.55065352e-01 3.14067721e-01 8.74214053e-01 -4.47026491e-01
-8.85364652e-01 -3.85929406e-01 3.31338674e-01 1.25699115e+00
6.89606415e-03 -9.76190567e-01 -8.90131772e-01 -1.22932613e+00
2.53213435e-01 1.12837553e+00 -6.16121471e-01 -4.67455477e-01
-4.61167365e-01 -9.92335320e-01 7.96683729e-01 2.76751876e-01
-4.15052585e-02 -1.14442122e+00 -9.42396969e-02 -2.08161364e-04
-1.88990459e-01 -1.35259748e+00 -8.66102993e-01 -3.63134593e-02
-5.06152272e-01 -9.37662601e-01 -8.56529921e-02 -4.34119105e-01
7.15159774e-01 3.35544854e-01 1.00813198e+00 3.22010189e-01
-5.52230775e-02 -2.05939308e-01 -8.11789691e-01 -5.11543810e-01
-8.44950497e-01 3.07307303e-01 2.33327359e-01 1.56060886e-02
8.02719593e-01 -5.10230899e-01 -2.69692451e-01 1.69983774e-01
-1.11377573e+00 -2.01458633e-01 2.11753324e-01 8.93947840e-01
2.63850152e-01 2.41428897e-01 6.44356966e-01 -1.31249571e+00
1.13595319e+00 -4.94946569e-01 -6.39738858e-01 1.68723002e-01
-1.04009175e+00 8.56823623e-02 1.33214962e+00 -7.81652510e-01
-5.26462257e-01 -3.06351304e-01 -3.50908220e-01 -2.28223279e-01
7.58778602e-02 5.44709384e-01 -2.30387077e-01 6.11675344e-02
1.00851953e+00 4.71109182e-01 -2.71099448e-01 -3.46842289e-01
5.04991055e-01 1.26967144e+00 2.27103204e-01 -5.39610445e-01
1.18088257e+00 4.96144265e-01 -3.84526730e-01 -3.21369082e-01
-1.05570757e+00 -3.50297272e-01 -7.74154887e-02 3.42882633e-01
1.33283690e-01 -6.23993039e-01 -9.44943666e-01 6.19787276e-01
-1.29654944e+00 -3.60707998e-01 1.63479168e-02 -5.32168038e-02
-3.29626620e-01 8.01316321e-01 -5.77625930e-01 -9.45877731e-01
-1.10618448e+00 -8.37175071e-01 9.91386175e-01 -2.18539283e-01
-3.35691482e-01 -6.41832888e-01 7.90317059e-02 1.44957215e-01
1.64481387e-01 -2.07609698e-01 1.17133045e+00 -1.44414425e+00
-6.48531988e-02 -6.29489005e-01 8.05537924e-02 4.65015799e-01
-3.44920554e-03 1.64491266e-01 -9.47371662e-01 -3.91920775e-01
3.22558619e-02 -2.15307310e-01 6.55914187e-01 -3.13168257e-01
1.64688003e+00 -9.15606737e-01 -3.20000350e-01 4.47852939e-01
9.68811393e-01 9.99551341e-02 5.77656746e-01 3.10639381e-01
5.20943344e-01 3.31389487e-01 6.72844887e-01 8.51310968e-01
4.89697326e-03 4.82540935e-01 4.54588532e-01 3.32657784e-01
4.42269772e-01 -5.70646405e-01 8.77060294e-01 7.28149295e-01
8.52308571e-01 -7.40109801e-01 -1.03430283e+00 4.73179340e-01
-1.96893859e+00 -9.47524905e-01 -8.10799897e-02 2.05662966e+00
1.24545026e+00 5.05505800e-01 -9.18011963e-02 3.20771128e-01
6.80518866e-01 2.07220390e-01 -5.08473754e-01 -6.60816789e-01
-2.50449963e-02 4.14606661e-01 6.33156061e-01 6.40043736e-01
-8.74701202e-01 1.29568422e+00 6.78643847e+00 1.01487625e+00
-1.14117730e+00 2.05379203e-01 6.28121495e-01 -5.74201345e-01
-5.09684563e-01 -1.00817069e-01 -9.97195542e-01 9.39337432e-01
1.06080472e+00 -8.11472893e-01 6.56819046e-01 1.17289448e+00
3.81699324e-01 4.02527183e-01 -9.46266890e-01 8.32175910e-01
3.24855834e-01 -1.06143868e+00 4.06124055e-01 -3.55071932e-01
5.89164197e-01 4.94850017e-02 -7.08782449e-02 2.77519822e-01
6.40002429e-01 -8.53464901e-01 5.59138298e-01 -1.13333717e-01
5.22229075e-01 -9.51647341e-01 6.50840223e-01 7.57302761e-01
-5.52619815e-01 -2.82792687e-01 -4.03756559e-01 -8.06585178e-02
5.85081056e-02 9.79692638e-01 -1.02721882e+00 1.09619983e-01
6.95819974e-01 3.34057868e-01 -5.41113198e-01 5.82892954e-01
-6.93023562e-01 1.12119555e+00 -3.43555480e-01 -6.72889650e-01
-3.50433588e-02 3.23399395e-01 6.54326916e-01 1.28381610e+00
1.62466839e-01 3.94226275e-02 1.60889402e-01 5.50425589e-01
-4.29738224e-01 4.37483817e-01 -4.85250086e-01 -1.91415280e-01
8.05927277e-01 1.20140767e+00 -3.05183887e-01 -4.52914923e-01
-2.87659258e-01 1.09267771e+00 3.38964343e-01 2.91409731e-01
-9.01940286e-01 -5.85078239e-01 8.41841459e-01 6.35479912e-02
1.39429584e-01 1.88326731e-01 -4.84564662e-01 -1.55483246e+00
2.37562001e-01 -1.46368515e+00 2.65459716e-01 -8.15690219e-01
-1.45987523e+00 5.95260859e-01 -2.96185344e-01 -7.70897865e-01
-2.87283808e-01 -6.75127208e-01 -7.55082846e-01 8.07828844e-01
-1.41368997e+00 -6.58061326e-01 -8.65016878e-02 4.77293998e-01
4.60571557e-01 -2.87663430e-01 1.01320982e+00 2.07798593e-02
-8.62274170e-01 1.09690726e+00 2.35315472e-01 3.29177529e-01
9.71904814e-01 -1.05034709e+00 1.29661500e+00 1.06909776e+00
1.37690052e-01 6.46215439e-01 9.22852218e-01 -8.32866371e-01
-1.29085064e+00 -1.21922088e+00 1.10084772e+00 -5.65808892e-01
9.82380331e-01 -8.30939949e-01 -7.68768072e-01 7.48714328e-01
5.44819757e-02 -6.07017912e-02 6.88973367e-01 5.01338691e-02
-8.02278101e-01 4.91489610e-03 -1.10606921e+00 1.05077183e+00
7.57938623e-01 -6.33420587e-01 -5.68694592e-01 8.67384076e-01
1.29302132e+00 -3.76961589e-01 -4.61721532e-02 7.12537840e-02
3.39617223e-01 -5.32438040e-01 7.82261968e-01 -1.48183835e+00
5.45339882e-01 3.34552675e-02 2.76169199e-02 -1.61279404e+00
-9.24415216e-02 -1.27581584e+00 -6.38705552e-01 1.00224984e+00
7.15158522e-01 -1.02562129e+00 8.38606954e-01 5.60628057e-01
3.52936536e-01 -8.36432159e-01 -8.50310922e-01 -7.48026311e-01
3.32012385e-01 -6.74013495e-01 8.71832073e-01 8.68463695e-01
5.06683350e-01 3.84065390e-01 -6.15259349e-01 2.91777939e-01
2.74372756e-01 -7.63829723e-02 9.70507562e-01 -8.76182854e-01
-5.68757474e-01 -3.90457809e-01 -8.14152285e-02 -1.01012635e+00
5.96875310e-01 -6.64902866e-01 -1.07368752e-02 -7.88411438e-01
2.95756340e-01 2.97521725e-02 -2.42896512e-01 9.27556038e-01
-8.34839046e-01 -1.48866192e-01 8.76322240e-02 1.04223803e-01
-2.76811302e-01 4.29685682e-01 6.35291636e-01 -2.43225873e-01
1.31276634e-03 2.23305494e-01 -1.26371479e+00 7.35892653e-01
1.07841468e+00 -1.02252030e+00 -2.63739824e-01 -2.04277903e-01
8.15705776e-01 -4.11189318e-01 -6.77391365e-02 -3.19300354e-01
2.47473359e-01 -3.36555332e-01 -2.77337641e-01 -4.46580023e-01
-1.49359539e-01 -7.55997181e-01 -7.89073706e-01 4.34719354e-01
-7.31001377e-01 1.86799809e-01 1.93726778e-01 6.43113017e-01
1.72546059e-01 -3.59527856e-01 8.07575166e-01 2.50834614e-01
3.27272601e-02 4.17325526e-01 -5.33230007e-01 3.42981577e-01
1.02781391e+00 5.22554278e-01 -7.36985743e-01 -7.50759661e-01
-1.19229034e-01 2.30086997e-01 2.78704673e-01 5.83234429e-01
4.94243681e-01 -9.09097314e-01 -8.84844065e-01 2.21945032e-01
2.94505626e-01 -3.62481534e-01 -2.82759726e-01 4.62628692e-01
-3.11145633e-01 2.70898759e-01 5.98931313e-01 -8.54971260e-02
-1.41303217e+00 1.12634373e+00 3.62103373e-01 -5.74019849e-01
-4.88567948e-01 7.52663255e-01 -1.47527065e-02 -4.33012217e-01
2.81875342e-01 -2.43755698e-01 7.87379369e-02 -2.12656707e-01
1.08733845e+00 7.73675814e-02 4.00200784e-01 -1.84985511e-02
-3.24724644e-01 -4.11120057e-02 -5.84020972e-01 -8.09235498e-03
8.97124112e-01 -2.53006723e-02 -1.53095871e-01 5.98190576e-02
1.01321232e+00 5.94802976e-01 -5.08352578e-01 -3.83155018e-01
-3.24834101e-02 -6.26922309e-01 4.57173511e-02 -1.05704582e+00
-8.63685846e-01 8.53854120e-01 -8.27578381e-02 5.28034210e-01
1.11124694e+00 -4.47705597e-01 1.24980938e+00 8.38547409e-01
2.70379543e-01 -1.19944346e+00 1.78882882e-01 7.84035683e-01
7.22658932e-01 -9.73936319e-01 -1.13361172e-01 -4.89718288e-01
-7.78398275e-01 9.08529639e-01 6.06197238e-01 5.79486750e-02
6.72413766e-01 7.72466719e-01 1.86158702e-01 2.23536104e-01
-1.41167617e+00 2.53628641e-01 -4.11799908e-01 2.54761517e-01
3.32052410e-02 -1.27535954e-01 -2.71829098e-01 1.07174444e+00
-4.91407812e-01 -1.21065669e-01 8.15953553e-01 8.58446360e-01
-5.05339801e-01 -1.25970984e+00 -4.46428537e-01 5.85143387e-01
-9.99371469e-01 -6.78134859e-01 -8.09236288e-01 2.96190470e-01
-6.12953722e-01 1.28725314e+00 -2.71939456e-01 -8.63687515e-01
3.70444685e-01 3.34503323e-01 -2.15265632e-01 -5.80279231e-01
-8.68389666e-01 -3.08058679e-01 1.53240666e-01 -6.59739137e-01
4.87408161e-01 -2.32680708e-01 -1.23515821e+00 -9.80058849e-01
-5.23759723e-01 2.57223845e-01 4.75556463e-01 9.50491905e-01
6.77645624e-01 2.18194276e-01 1.25533664e+00 -2.49057978e-01
-1.53745902e+00 -1.06128514e+00 -4.09616709e-01 5.89585781e-01
1.59138262e-01 7.25241238e-03 -9.42908704e-01 -6.71343282e-02]
|
[6.02384090423584, 8.062111854553223]
|
ba844be9-1d1c-4a49-a18a-5cfa4d14cc18
|
task-decoupled-framework-for-reference-based
| null | null |
http://openaccess.thecvf.com//content/CVPR2022/html/Huang_Task_Decoupled_Framework_for_Reference-Based_Super-Resolution_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Huang_Task_Decoupled_Framework_for_Reference-Based_Super-Resolution_CVPR_2022_paper.pdf
|
Task Decoupled Framework for Reference-Based Super-Resolution
|
Reference-based super-resolution(RefSR) has achieved impressive progress on the recovery of high-frequency details thanks to an additional reference high-resolution(HR) image input. Although the superiority compared with Single-Image Super-Resolution(SISR), existing RefSR methods easily result in the reference-underuse issue and the reference-misuse as shown in Fig.1. In this work, we deeply investigate the cause of the two issues and further propose a novel framework to mitigate them. Our studies find that the issues are mostly due to the improper coupled framework design of current methods. Those methods conduct the super-resolution task of the input low-resolution(LR) image and the texture transfer task from the reference image together in one module, easily introducing the interference between LR and reference features. Inspired by this finding, we propose a novel framework, which decouples the two tasks of RefSR, eliminating the interference between the LR image and the reference image. The super-resolution task upsamples the LR image leveraging only the LR image itself. The texture transfer task extracts and transfers abundant textures from the reference image to the coarsely upsampled result of the super-resolution task. Extensive experiments demonstrate clear improvements in both quantitative and qualitative evaluations over state-of-the-art methods.
|
['Dazhi He', 'Yan-Feng Wang', 'Ya zhang', 'Siheng Chen', 'Yu Fu', 'Xiaoyun Zhang', 'Yixuan Huang']
|
2022-01-01
| null | null | null |
cvpr-2022-1
|
['reference-based-super-resolution']
|
['computer-vision']
|
[ 5.52749276e-01 -1.28319278e-01 2.93329190e-02 -1.58497974e-01
-1.35789955e+00 -2.05822941e-02 4.83145475e-01 -7.67236531e-01
-7.73662776e-02 8.58325303e-01 3.41689020e-01 3.89861584e-01
-4.62902747e-02 -8.09586525e-01 -6.09878361e-01 -9.42333341e-01
4.08232659e-01 -1.03152230e-01 4.82799619e-01 -4.96560425e-01
2.26846620e-01 2.57791758e-01 -1.82626271e+00 6.80036962e-01
9.15975749e-01 1.02316976e+00 6.20019972e-01 4.63132620e-01
6.58247694e-02 6.71769083e-01 -3.87170583e-01 1.23358585e-01
3.11405689e-01 -4.90284890e-01 -6.44305468e-01 -5.42137660e-02
6.80923641e-01 -6.53798461e-01 -2.70329207e-01 1.26147091e+00
5.60399652e-01 1.92780867e-01 1.63860038e-01 -5.22545218e-01
-1.07881236e+00 2.52880633e-01 -1.18138862e+00 5.57478011e-01
5.01180708e-01 -1.72054023e-01 5.80616832e-01 -1.30340016e+00
8.12220693e-01 1.42559373e+00 5.94541907e-01 5.26407778e-01
-1.40475214e+00 -7.43710577e-01 -2.18101162e-02 8.95290375e-02
-1.53641880e+00 -6.56250000e-01 8.32949638e-01 1.47826318e-02
5.47292590e-01 2.20523208e-01 2.06900075e-01 1.12772012e+00
1.81554332e-01 1.47179753e-01 1.66544485e+00 -4.33060914e-01
-6.01560734e-02 2.07116872e-01 -3.59663069e-02 3.23908031e-01
1.19142070e-01 4.63138700e-01 -7.32630551e-01 1.35595664e-01
1.48638749e+00 1.32500129e-02 -5.43847084e-01 1.95500061e-01
-1.07040453e+00 3.81208718e-01 4.23284173e-01 5.61750293e-01
-3.09780240e-01 -4.07290936e-01 -9.53118503e-03 2.78373480e-01
7.50754058e-01 2.19967961e-01 -1.51608840e-01 2.58966267e-01
-1.06014442e+00 -2.49260175e-03 1.21742219e-01 8.44927311e-01
9.68462110e-01 -1.86889414e-02 -2.11998075e-01 1.16881621e+00
-1.56669930e-01 4.33212698e-01 4.79111820e-01 -9.99155581e-01
3.49299878e-01 2.27887303e-01 3.99359047e-01 -1.01137614e+00
-7.47329071e-02 -4.69707847e-01 -1.13743544e+00 2.57516116e-01
2.47666448e-01 3.70444506e-01 -7.27642655e-01 1.54907691e+00
4.59839702e-01 1.88702032e-01 8.53357986e-02 1.41104329e+00
1.04686785e+00 5.52803457e-01 -1.66143507e-01 -4.37254041e-01
1.40754914e+00 -9.05611813e-01 -9.66701567e-01 2.22482029e-02
-2.68911362e-01 -9.36089993e-01 1.18531168e+00 2.68069208e-01
-1.27712023e+00 -1.14384604e+00 -1.08955753e+00 -6.43640816e-01
-1.39950320e-01 2.81872272e-01 1.93242788e-01 1.09329104e-01
-1.11636674e+00 6.85967863e-01 -3.78783196e-01 -1.01631656e-01
2.93652833e-01 3.00887991e-02 -4.95533794e-01 -3.72372597e-01
-1.20941389e+00 9.10426795e-01 3.11444644e-02 2.86638439e-01
-4.98248816e-01 -9.06075895e-01 -6.25076592e-01 -9.94156599e-02
4.42826897e-01 -7.02084839e-01 7.71855235e-01 -1.09404874e+00
-1.54003203e+00 8.14286530e-01 -3.84141713e-01 -8.71508867e-02
4.20058221e-01 -3.61546934e-01 -6.47773385e-01 2.82153785e-01
1.80406481e-01 2.28429049e-01 1.24042606e+00 -1.58737373e+00
-8.26442122e-01 -3.80448431e-01 -1.45266935e-01 2.00040117e-01
1.93228468e-01 6.00640029e-02 -3.62791985e-01 -8.01008821e-01
2.57025898e-01 -4.11141783e-01 -5.95537908e-02 -2.41231248e-01
-1.02070235e-01 2.70886451e-01 9.48884845e-01 -9.03093159e-01
1.24815595e+00 -2.40283108e+00 4.88422215e-02 -3.54021966e-01
4.35802907e-01 1.84618011e-01 -3.34009230e-01 -1.03757121e-01
-3.06362569e-01 -3.64976861e-02 1.13267601e-01 -2.21061751e-01
-5.66663682e-01 -1.43515006e-01 -7.36873269e-01 4.97487962e-01
3.72138500e-01 7.70235837e-01 -8.30882967e-01 -5.82519770e-01
2.24860832e-01 1.11185610e+00 -1.25671044e-01 3.08111638e-01
1.94386914e-01 9.63182747e-01 -3.62318218e-01 5.59880316e-01
1.19686913e+00 -2.90696502e-01 -9.50762711e-04 -9.72361267e-01
-4.23897177e-01 1.61856279e-01 -1.19855130e+00 1.51406372e+00
-4.46469456e-01 2.56361634e-01 3.10837626e-01 -4.93698865e-01
1.12162554e+00 1.85583487e-01 4.41441953e-01 -1.23340595e+00
-3.44040602e-01 3.15054864e-01 -3.45706463e-01 -2.86458611e-01
7.70381033e-01 -5.01335382e-01 2.48394936e-01 1.44255921e-01
-5.44674993e-02 1.36484593e-01 -2.80153871e-01 -3.13478820e-02
6.48763895e-01 4.35945719e-01 3.12965125e-01 -2.04942003e-01
8.12676609e-01 -2.21786246e-01 6.03898525e-01 6.91600382e-01
-8.84322822e-03 9.87467229e-01 -8.57676715e-02 -5.59522212e-01
-1.32587481e+00 -1.14839196e+00 -2.57296979e-01 9.50861871e-01
4.97847229e-01 -2.72073060e-01 -6.98801458e-01 -2.46539369e-01
-3.54792804e-01 1.17218517e-01 -7.80574799e-01 1.26132861e-01
-7.23113239e-01 -7.43111789e-01 1.13552988e-01 3.83566260e-01
9.53610897e-01 -7.11621284e-01 -4.43338096e-01 7.76945129e-02
-6.91842139e-01 -1.45754242e+00 -4.98550892e-01 -3.45927715e-01
-7.88391232e-01 -7.81202734e-01 -8.19426596e-01 -4.28223133e-01
5.80811977e-01 7.90228844e-01 1.08105826e+00 4.15292867e-02
-4.22479510e-01 5.11128195e-02 -2.36131907e-01 3.39278102e-01
-1.91021308e-01 -2.49623150e-01 -6.90551847e-02 3.70007634e-01
1.39814969e-02 -6.93015993e-01 -6.90657914e-01 5.14973521e-01
-8.39726388e-01 4.94099140e-01 8.58313382e-01 1.02418184e+00
1.24287558e+00 4.18972820e-01 6.29853427e-01 -8.08805645e-01
2.92848855e-01 -2.79283494e-01 -5.98481476e-01 1.64014846e-01
-5.94334424e-01 2.65812557e-02 7.77238190e-01 -4.92338806e-01
-1.65911472e+00 -2.08951458e-01 1.59578264e-01 -6.32978857e-01
-5.76071143e-02 -1.40486464e-01 -3.08023781e-01 -2.83814639e-01
4.36844349e-01 4.62210864e-01 -9.77836251e-02 -8.44056964e-01
2.70107448e-01 5.61123133e-01 9.11971450e-01 -5.58202386e-01
1.10237920e+00 9.76897180e-01 2.68831067e-02 -9.03841317e-01
-1.37016392e+00 -4.99383688e-01 -6.34903789e-01 -9.97539982e-02
7.55771399e-01 -1.31518018e+00 -3.04691881e-01 2.84045815e-01
-9.05320168e-01 -1.06249623e-01 -3.34361732e-01 3.33815187e-01
-5.55421114e-01 3.30905646e-01 -6.96053267e-01 -6.57729805e-01
-4.33185041e-01 -1.12189460e+00 1.42774642e+00 4.23745364e-01
1.62412852e-01 -5.34265935e-01 -1.40794843e-01 4.60186988e-01
1.03014052e+00 1.18091986e-01 6.59943461e-01 2.02760726e-01
-9.30284441e-01 3.94615293e-01 -9.21057642e-01 2.75402069e-01
3.07326406e-01 -3.00147414e-01 -1.10166121e+00 -4.53262478e-01
4.73911524e-01 -4.36818302e-02 8.90321910e-01 3.28296483e-01
9.80210841e-01 -1.86618954e-01 -3.19010057e-02 9.27631915e-01
1.95725417e+00 -2.55838156e-01 1.03371310e+00 4.55788940e-01
7.30815291e-01 6.12200975e-01 9.61183250e-01 5.78272492e-02
2.82547265e-01 1.13609850e+00 -6.10014163e-02 -5.55066347e-01
-8.09528828e-01 -2.44402483e-01 4.07304496e-01 6.11928880e-01
-5.82673371e-01 4.94843572e-01 -2.47268826e-01 2.08437040e-01
-1.63661849e+00 -1.16326356e+00 -1.32778063e-01 2.38545966e+00
1.00604761e+00 -3.01110238e-01 -1.01694554e-01 -1.09457418e-01
7.78804421e-01 2.74832368e-01 -4.83823866e-01 6.73601106e-02
-4.73624736e-01 2.98049688e-01 3.41364950e-01 6.73853934e-01
-9.55548048e-01 9.35761511e-01 6.09211588e+00 1.15720069e+00
-1.27523160e+00 3.13453645e-01 7.64383852e-01 -1.65015772e-01
-1.74633175e-01 -1.05720609e-01 -1.03544986e+00 4.93729681e-01
8.16746533e-01 -1.23686761e-01 7.61263907e-01 3.76029342e-01
2.94330508e-01 -2.99618870e-01 -8.47222328e-01 1.20435143e+00
1.01318117e-02 -1.06549633e+00 2.81082630e-01 -1.08031882e-02
6.51976287e-01 -1.91696346e-01 3.11875999e-01 8.47458839e-02
-2.20101207e-01 -1.17718637e+00 5.05594790e-01 8.82277727e-01
1.55633652e+00 -7.02504694e-01 5.81742525e-01 -2.21498832e-02
-1.52125680e+00 2.25949828e-02 -5.51901460e-01 1.49170637e-01
1.10401222e-02 9.02469993e-01 -1.37387784e-02 9.89065230e-01
1.08562124e+00 6.41251504e-01 -5.51230907e-01 2.64584243e-01
-6.98589012e-02 9.11047384e-02 -8.30478407e-03 1.08218932e+00
-3.32312614e-01 -3.55990857e-01 6.68364882e-01 9.72950935e-01
2.33157903e-01 4.52955812e-01 1.41789624e-02 1.41418386e+00
2.11425647e-01 -1.65274933e-01 -4.11299706e-01 4.17016506e-01
4.60716188e-01 1.52253950e+00 -5.02255738e-01 -3.19577694e-01
-4.86359507e-01 1.10689819e+00 1.81715414e-01 6.67617142e-01
-6.55243635e-01 -2.25715429e-01 4.08189267e-01 4.34113562e-01
4.05089527e-01 6.92931935e-02 -3.88875902e-01 -1.43544436e+00
2.93912590e-01 -1.00207651e+00 1.07270442e-01 -9.90706146e-01
-1.33903182e+00 6.99883997e-01 -1.73388630e-01 -1.42955005e+00
8.64738151e-02 -4.01454605e-02 -2.03027546e-01 1.51530516e+00
-2.10027003e+00 -1.21312392e+00 -5.86491764e-01 6.75994396e-01
5.77566445e-01 2.57924438e-01 5.60639322e-01 4.15129691e-01
-4.84928936e-01 4.23645645e-01 -6.60717040e-02 -7.99187049e-02
9.62786973e-01 -8.93130779e-01 9.56201479e-02 9.42049444e-01
-4.53921229e-01 7.91773796e-01 6.08690083e-01 -6.98784411e-01
-1.39048266e+00 -9.17712808e-01 5.30867517e-01 -3.15371305e-01
3.59994143e-01 -2.65789419e-01 -1.29636860e+00 2.38226548e-01
-2.61421800e-01 3.51740241e-01 2.29882866e-01 -1.71241120e-01
-6.38215423e-01 -3.62047553e-01 -1.34787846e+00 3.60164940e-01
9.67299044e-01 -8.39537621e-01 -5.42787135e-01 -3.42711121e-01
8.72793674e-01 -4.54782009e-01 -1.23057163e+00 6.31429434e-01
6.94576740e-01 -1.40230989e+00 1.42944539e+00 1.69904977e-01
8.48744273e-01 -7.02062786e-01 -3.84524643e-01 -9.71622109e-01
-6.20063066e-01 -5.14295995e-01 -1.31658733e-01 1.36805236e+00
-9.36233178e-02 -6.30774856e-01 2.10061982e-01 3.70756269e-01
2.29176760e-01 -5.27352333e-01 -8.80842388e-01 -6.52081013e-01
-2.62520909e-01 3.28029335e-01 6.72371864e-01 9.45717156e-01
-4.19688374e-01 5.08768439e-01 -5.45197010e-01 4.85915333e-01
1.14317417e+00 4.87605929e-01 5.83922803e-01 -1.04166639e+00
-2.72997975e-01 -8.11581090e-02 1.07335441e-01 -8.81183147e-01
-2.32969061e-01 -3.53233755e-01 7.67854303e-02 -1.20782590e+00
6.04946792e-01 -3.05668235e-01 -3.70287061e-01 1.79739427e-02
-4.58194107e-01 4.44813818e-01 7.32919574e-02 6.30766451e-01
-6.03675485e-01 3.92686903e-01 1.71196699e+00 3.85368884e-01
-1.44527644e-01 -3.83697122e-01 -9.47095931e-01 5.39882481e-01
3.87414634e-01 -8.14721063e-02 -1.89049393e-01 -2.56876022e-01
-7.04375058e-02 3.01877648e-01 5.97973943e-01 -7.49126196e-01
5.38433231e-02 -6.04750216e-02 6.71830237e-01 -4.86696750e-01
3.77688199e-01 -7.03367949e-01 3.77880007e-01 -2.06891179e-01
-1.82290271e-01 -3.22915882e-01 5.26919067e-02 5.84756315e-01
-4.46485281e-01 4.01645660e-01 1.34982741e+00 -2.24771127e-01
-6.30872071e-01 1.30994007e-01 2.58028686e-01 -1.21237859e-01
4.74969685e-01 -3.52344096e-01 -7.30996668e-01 -7.94350877e-02
-5.97826302e-01 -2.52113253e-01 7.75298059e-01 2.80896246e-01
7.41636097e-01 -1.16814220e+00 -8.52547526e-01 4.67826515e-01
-1.76293433e-01 1.61958411e-01 8.21784496e-01 1.01743591e+00
1.57183781e-01 2.23810747e-01 -3.35897923e-01 -3.90083671e-01
-1.23804319e+00 6.92686737e-01 3.99468750e-01 -5.79698503e-01
-1.16604578e+00 4.29524690e-01 7.27189898e-01 6.46640956e-02
-1.31218746e-01 -9.08330679e-02 -3.38285625e-01 -1.43254593e-01
1.22956657e+00 5.08412182e-01 -5.08984998e-02 -8.51141810e-01
-8.07948411e-02 1.07757616e+00 -3.65291238e-01 -6.55360371e-02
1.50038958e+00 -7.84267604e-01 -3.13515872e-01 3.69485617e-01
1.05908513e+00 1.44361168e-01 -1.43289542e+00 -5.00336170e-01
-3.04911762e-01 -8.34750652e-01 2.85214514e-01 -8.26663375e-01
-1.07310784e+00 7.06330180e-01 7.38094568e-01 -6.05229624e-02
1.63853705e+00 -7.57180378e-02 8.30311835e-01 -3.39802176e-01
6.86835110e-01 -1.00300920e+00 2.10886099e-03 2.66864359e-01
9.74475503e-01 -1.19310594e+00 2.92254657e-01 -7.16450810e-01
-3.53680164e-01 1.06803870e+00 4.90712196e-01 -1.94391221e-01
3.78224194e-01 5.13634980e-01 -3.96912806e-02 -1.01094693e-01
-8.04533601e-01 -3.00207138e-01 3.19171339e-01 6.88308835e-01
3.13894093e-01 -1.81554556e-01 -2.07094342e-01 8.97413194e-01
7.70879984e-02 4.06851947e-01 4.90653485e-01 4.02107090e-01
-3.59248221e-01 -8.36400449e-01 -7.05788195e-01 4.95319031e-02
-6.81537509e-01 -2.66615599e-01 1.11106731e-01 5.37192583e-01
3.01228315e-02 8.23564231e-01 -5.67005165e-02 -3.82112503e-01
4.03608054e-01 -3.89465421e-01 5.89748859e-01 -3.42635751e-01
-2.29864985e-01 3.71335298e-01 -2.08636791e-01 -1.13610125e+00
-6.76482499e-01 -2.56593734e-01 -9.30899918e-01 -9.95727777e-02
-3.34326893e-01 -6.91794008e-02 3.09620321e-01 6.33046687e-01
5.31327248e-01 6.03898346e-01 7.05518365e-01 -1.25954628e+00
-4.36020553e-01 -9.48916197e-01 -8.59317780e-01 5.16953647e-01
7.33993411e-01 -7.16154754e-01 -6.36025488e-01 9.93412137e-02]
|
[10.956046104431152, -2.0663340091705322]
|
de0b4e5d-ae25-4ae1-9d3e-18be00a92b12
|
how-do-multilingual-encoders-learn-cross
|
2207.05737
| null |
https://arxiv.org/abs/2207.05737v1
|
https://arxiv.org/pdf/2207.05737v1.pdf
|
How Do Multilingual Encoders Learn Cross-lingual Representation?
|
NLP systems typically require support for more than one language. As different languages have different amounts of supervision, cross-lingual transfer benefits languages with little to no training data by transferring from other languages. From an engineering perspective, multilingual NLP benefits development and maintenance by serving multiple languages with a single system. Both cross-lingual transfer and multilingual NLP rely on cross-lingual representations serving as the foundation. As BERT revolutionized representation learning and NLP, it also revolutionized cross-lingual representations and cross-lingual transfer. Multilingual BERT was released as a replacement for single-language BERT, trained with Wikipedia data in 104 languages. Surprisingly, without any explicit cross-lingual signal, multilingual BERT learns cross-lingual representations in addition to representations for individual languages. This thesis first shows such surprising cross-lingual effectiveness compared against prior art on various tasks. Naturally, it raises a set of questions, most notably how do these multilingual encoders learn cross-lingual representations. In exploring these questions, this thesis will analyze the behavior of multilingual models in a variety of settings on high and low resource languages. We also look at how to inject different cross-lingual signals into multilingual encoders, and the optimization behavior of cross-lingual transfer with these models. Together, they provide a better understanding of multilingual encoders on cross-lingual transfer. Our findings will lead us to suggested improvements to multilingual encoders and cross-lingual transfer.
|
['Shijie Wu']
|
2022-07-12
| null | null | null | null |
['multilingual-nlp']
|
['natural-language-processing']
|
[-4.63316470e-01 4.58563268e-02 -6.79107368e-01 -5.07559121e-01
-1.38610506e+00 -1.06367004e+00 7.02470839e-01 -1.11187309e-01
-4.59487915e-01 1.01088774e+00 4.51291054e-01 -6.67567909e-01
2.95040816e-01 -6.18112385e-01 -1.26526892e+00 -2.08073556e-01
5.61265799e-04 6.10463262e-01 -4.13428754e-01 -6.91132903e-01
-4.04730856e-01 2.16840237e-01 -1.01223028e+00 5.27844906e-01
8.55948329e-01 3.73183608e-01 4.43092406e-01 3.30139071e-01
-3.43231767e-01 7.42506146e-01 -4.58013594e-01 -6.15619957e-01
3.87395084e-01 -1.29237071e-01 -7.61020422e-01 -4.33717519e-01
7.38496661e-01 5.66877834e-02 -1.08290330e-01 7.86910892e-01
5.19629300e-01 -3.13779205e-01 6.08014286e-01 -9.71058071e-01
-1.21599352e+00 1.18448818e+00 -6.34885192e-01 3.07615437e-02
2.72947162e-01 1.45298898e-01 1.33844328e+00 -9.07173932e-01
7.55254090e-01 1.49536490e+00 8.64371181e-01 3.44169408e-01
-1.47983730e+00 -9.61200416e-01 8.16200003e-02 -2.50949293e-01
-1.40530539e+00 -6.81976378e-01 2.66863018e-01 -5.26843429e-01
1.29257596e+00 -2.53251910e-01 2.29290530e-01 1.21149778e+00
3.81362945e-01 9.69557226e-01 1.25049305e+00 -6.61616027e-01
-6.64063752e-01 7.47425139e-01 4.22582217e-02 6.32521808e-01
2.64808297e-01 2.18994707e-01 -6.56288147e-01 3.48341256e-01
5.08794069e-01 -5.84090710e-01 -3.03977162e-01 -2.03873873e-01
-1.14507043e+00 9.14324820e-01 3.38446617e-01 7.57807493e-01
-1.04375631e-01 3.30465317e-01 7.45704472e-01 7.12933719e-01
4.98778611e-01 5.95154583e-01 -1.04294896e+00 -1.34830087e-01
-8.12256694e-01 -3.14277560e-01 8.16128671e-01 1.24065685e+00
1.28648138e+00 4.58351970e-01 3.04025501e-01 1.14129674e+00
1.25699535e-01 7.79133916e-01 6.89146578e-01 -5.91344595e-01
1.00460637e+00 1.58957362e-01 -4.15328562e-01 -5.82692325e-01
-4.87103350e-02 -5.88968933e-01 -5.44198155e-01 1.43564329e-03
3.37722510e-01 -4.58242327e-01 -5.31356156e-01 2.13299489e+00
-3.15505743e-01 -3.20059091e-01 5.39732516e-01 3.28210026e-01
4.18592453e-01 9.30125535e-01 9.73506421e-02 1.12787038e-01
1.20220149e+00 -9.57218170e-01 -5.09630680e-01 -3.60482395e-01
1.19640458e+00 -1.01789963e+00 1.32075465e+00 1.50238335e-01
-1.16270018e+00 -6.90011144e-01 -1.04763639e+00 -5.02611935e-01
-8.09704065e-01 3.47461611e-01 8.06708932e-01 5.57779670e-01
-1.19290662e+00 3.56420636e-01 -7.44106531e-01 -3.63201231e-01
6.49193898e-02 3.04771334e-01 -7.12177575e-01 -4.26059246e-01
-1.61973238e+00 1.37388968e+00 4.12774712e-01 -2.41697669e-01
-8.52376699e-01 -1.02003276e+00 -1.20002139e+00 -1.06508516e-01
-9.98812914e-03 -4.15529341e-01 1.11255288e+00 -1.39584255e+00
-1.21015310e+00 9.38423514e-01 -1.73558503e-01 -5.26799321e-01
2.39461944e-01 -3.94040018e-01 -5.73675990e-01 -4.29949552e-01
6.51485026e-01 8.42043519e-01 3.46319675e-01 -1.19733191e+00
-6.82492495e-01 -8.93098041e-02 6.71969652e-02 4.76853669e-01
-3.16332251e-01 5.98113313e-02 -4.38661963e-01 -4.96227801e-01
-4.56664383e-01 -8.40621233e-01 3.02420616e-01 -6.59101427e-01
-4.60897684e-02 -2.58241504e-01 4.40386444e-01 -9.36296225e-01
8.82542014e-01 -2.18208337e+00 2.18630031e-01 -1.69979885e-01
-3.27147931e-01 8.59868620e-03 -4.37419415e-01 6.30926073e-01
-3.15524161e-01 4.18348134e-01 9.57991034e-02 -4.56140012e-01
1.72399983e-01 5.36925375e-01 -3.28680307e-01 3.13494146e-01
4.49237257e-01 1.04040062e+00 -9.02302086e-01 -3.90424639e-01
1.80287082e-02 7.72349834e-01 -6.03661120e-01 -1.12482332e-01
1.03028789e-01 3.33030313e-01 8.15367624e-02 5.74356794e-01
3.31766218e-01 3.85116786e-02 2.44269133e-01 -7.56428912e-02
-2.65017807e-01 7.52703428e-01 -6.14857674e-01 2.10626936e+00
-1.20418143e+00 9.03544009e-01 1.20128609e-01 -9.50580895e-01
8.11448514e-01 5.58074296e-01 3.28066200e-01 -8.69904935e-01
-1.85921982e-01 8.02751601e-01 6.35072365e-02 -1.41593665e-01
7.00682163e-01 -5.81484139e-01 -4.24447507e-01 5.54261923e-01
6.19925797e-01 -1.24165654e-01 3.85431737e-01 1.78939968e-01
6.35150850e-01 3.46670866e-01 2.16111183e-01 -5.07324219e-01
3.30106705e-01 -2.76624337e-02 4.28709149e-01 3.74847025e-01
1.72461197e-01 2.00841814e-01 1.85353741e-01 -2.07549497e-01
-8.90127361e-01 -1.21950173e+00 -4.48870450e-01 1.56520486e+00
-4.45979595e-01 -3.97061408e-01 -4.33659285e-01 -6.17487848e-01
3.93804699e-01 8.86275649e-01 -4.41402048e-01 -4.97890189e-02
-7.19162166e-01 -6.48665249e-01 1.00021148e+00 4.36448991e-01
1.16473243e-01 -9.48946893e-01 3.04342330e-01 2.93463051e-01
-2.22314149e-01 -1.17452276e+00 -6.33179486e-01 8.07412148e-01
-6.63731396e-01 -7.55321920e-01 -7.04807937e-01 -1.05567753e+00
3.43676358e-01 1.47737116e-01 1.69992721e+00 -4.11718786e-01
-1.32699059e-02 5.10249138e-01 -1.92210630e-01 -4.45269555e-01
-7.32953966e-01 7.92521477e-01 3.07673186e-01 -4.23157275e-01
5.60806334e-01 -5.75956941e-01 2.52528846e-01 7.30140135e-02
-5.76693773e-01 -1.68792844e-01 8.51705134e-01 8.07531416e-01
4.91526037e-01 -1.56312212e-01 8.77497435e-01 -1.03643203e+00
6.91053927e-01 -7.62069106e-01 -5.16878486e-01 3.64008725e-01
-5.20551920e-01 3.67578655e-01 8.22016120e-01 -2.63825864e-01
-1.03622580e+00 -2.93515325e-01 7.95895159e-02 -2.53117591e-01
2.31763184e-01 9.25788105e-01 -1.89043313e-01 2.87484258e-01
8.65514398e-01 2.54364610e-01 -2.62134790e-01 -5.49520254e-01
8.62766922e-01 6.08230412e-01 3.66246611e-01 -1.22831285e+00
5.56372702e-01 -1.05501324e-01 -7.80968845e-01 -8.67386162e-01
-7.17166483e-01 -3.21691811e-01 -8.78052771e-01 2.81086832e-01
7.66104519e-01 -1.64747798e+00 -1.07178681e-01 2.93653905e-02
-1.24735415e+00 -5.68707347e-01 -4.99335051e-01 5.93895435e-01
-4.70185310e-01 -7.76206031e-02 -8.48379314e-01 -2.73507267e-01
-1.53071862e-02 -1.37075269e+00 1.04812276e+00 -2.18572453e-01
-3.06504458e-01 -1.53557456e+00 2.20299914e-01 3.74832898e-01
3.73070657e-01 -2.31264547e-01 1.01749301e+00 -4.49937046e-01
-4.55536693e-01 -8.05897042e-02 -1.29324615e-01 8.60512555e-01
4.34855342e-01 -1.60096928e-01 -9.56069052e-01 -6.51179373e-01
-4.88639235e-01 -8.53011608e-01 4.83248532e-01 1.84627429e-01
3.29677284e-01 -1.27962455e-01 -1.81724995e-01 8.86038005e-01
1.51691353e+00 -9.49423388e-02 2.55015820e-01 4.84679490e-01
8.38685155e-01 7.11598635e-01 2.84034073e-01 -2.14540198e-01
6.68154180e-01 4.86803353e-01 -2.40002528e-01 -1.30075261e-01
-3.97066891e-01 -4.49866414e-01 1.12205791e+00 1.74166715e+00
1.28564402e-01 1.12303765e-02 -1.11627460e+00 7.80500829e-01
-1.29482484e+00 -5.60251176e-01 6.43708631e-02 2.20223856e+00
1.53409898e+00 -7.50918314e-02 -2.50752419e-01 -4.99977201e-01
4.26369071e-01 -1.68118477e-02 -3.71503085e-01 -6.53952777e-01
-4.88164276e-01 4.00104403e-01 7.92043149e-01 8.52799833e-01
-8.52183044e-01 1.57749355e+00 6.40938616e+00 8.08460295e-01
-1.47736287e+00 4.02791232e-01 3.14637691e-01 3.00827366e-03
-4.57225174e-01 6.19039834e-02 -1.31752288e+00 2.13730380e-01
1.33477700e+00 -5.52215457e-01 5.97953141e-01 8.74307454e-01
-1.45684958e-01 2.99824029e-01 -1.55549419e+00 7.80378580e-01
8.49393383e-02 -1.10541284e+00 1.56324655e-01 2.77620584e-01
1.06372309e+00 8.85232568e-01 1.10248253e-01 9.38252747e-01
9.67411637e-01 -1.26322138e+00 7.41655529e-01 7.85757452e-02
1.22195530e+00 -9.09514785e-01 5.78691661e-01 2.69016594e-01
-1.34946632e+00 3.78120512e-01 -5.20379543e-01 1.10947102e-01
1.29590943e-01 2.64475018e-01 -8.32675636e-01 8.16153109e-01
4.75099653e-01 1.17069399e+00 -4.26655680e-01 7.74643272e-02
-3.74070108e-01 6.26243770e-01 -2.87782788e-01 5.74109197e-01
4.29871112e-01 -1.45055324e-01 1.29393190e-01 1.74789369e+00
3.69610727e-01 -7.94251561e-01 2.97495425e-01 7.53745854e-01
-3.89668196e-01 3.93631279e-01 -1.21734083e+00 -3.15239608e-01
4.52270120e-01 9.11431789e-01 1.51724115e-01 -4.33289707e-01
-7.28666306e-01 8.68168473e-01 6.76066160e-01 5.72833717e-01
-6.95724249e-01 -1.66111097e-01 8.09274197e-01 2.30749115e-01
-1.04909062e-01 -4.69233394e-01 -1.55503631e-01 -1.41407764e+00
-1.59756482e-01 -1.32741606e+00 2.11154565e-01 -6.51536524e-01
-1.48568106e+00 6.52973175e-01 3.16367857e-03 -1.04447174e+00
-6.00286543e-01 -7.81994581e-01 1.42961040e-01 1.42429435e+00
-2.04360366e+00 -1.51930761e+00 4.53167826e-01 7.06740081e-01
5.53899467e-01 -6.09776974e-01 1.05753541e+00 7.09611893e-01
-3.17140758e-01 1.02673483e+00 3.72138977e-01 5.62438309e-01
1.31234872e+00 -1.26237416e+00 2.94883668e-01 5.97711325e-01
5.03755629e-01 1.09943199e+00 2.01239586e-01 -3.92340839e-01
-1.38655245e+00 -1.19582367e+00 1.14159119e+00 -6.52443171e-01
1.01516163e+00 -3.26490730e-01 -8.64950657e-01 1.52416158e+00
7.69562662e-01 -1.41504586e-01 9.39047992e-01 7.62925982e-01
-6.85600758e-01 -1.50620699e-01 -5.01387179e-01 4.61116672e-01
5.72297156e-01 -1.11368799e+00 -4.95140433e-01 3.32595915e-01
9.05963004e-01 -1.24401405e-01 -1.16624486e+00 1.27756139e-02
5.06619334e-01 -5.39293349e-01 9.64829981e-01 -6.79566145e-01
5.90042114e-01 -5.36344796e-02 -5.95582724e-01 -1.81620824e+00
-2.87403017e-01 -3.55513453e-01 5.30426443e-01 1.46769726e+00
9.29748356e-01 -9.83802557e-01 2.62185544e-01 5.77317476e-02
-4.78532076e-01 -4.56189036e-01 -8.08703244e-01 -1.26501358e+00
1.05074763e+00 -6.87853575e-01 2.69774497e-01 1.56205380e+00
-4.71194834e-02 8.67566466e-01 -3.47864985e-01 1.16283007e-01
3.82181823e-01 -2.48292133e-01 8.72678280e-01 -8.66472304e-01
-5.50785542e-01 -3.95422906e-01 -1.61491841e-01 -1.07207012e+00
6.38258457e-01 -1.89267981e+00 5.76174930e-02 -1.17295468e+00
-5.15917763e-02 -7.55534828e-01 -3.63233358e-01 8.55702400e-01
8.04689229e-02 5.51291741e-02 3.92033398e-01 2.62215048e-01
-7.27987438e-02 4.13997322e-01 1.03384924e+00 -3.33605468e-01
-1.09763205e-01 -6.74239576e-01 -1.00505638e+00 6.52071297e-01
7.26093352e-01 -5.32008767e-01 -4.06356275e-01 -1.10262918e+00
3.46568942e-01 -1.38963714e-01 -3.94130647e-01 -7.67293990e-01
-1.79317594e-01 1.36386424e-01 1.40949354e-01 -9.56806093e-02
3.37042719e-01 -7.89518893e-01 -1.61067769e-01 4.01351415e-02
-3.26632738e-01 3.03176790e-01 5.67562401e-01 3.84952389e-02
-6.47353053e-01 -1.28451020e-01 7.58041203e-01 -3.38113725e-01
-6.42303348e-01 -8.14914517e-03 -3.43025148e-01 4.77062255e-01
7.87895381e-01 1.60605423e-02 -3.16287637e-01 -2.63706654e-01
-7.61518002e-01 3.15986186e-01 4.46468472e-01 6.88377380e-01
-1.48687497e-01 -1.57678258e+00 -1.05154800e+00 3.38629574e-01
1.25424594e-01 -9.94862467e-02 -2.37863034e-01 7.07711995e-01
-3.26141596e-01 7.51365483e-01 -2.21329033e-01 -7.26722658e-01
-8.50703239e-01 2.44743794e-01 4.39463496e-01 -6.19495094e-01
-7.22374097e-02 9.27978098e-01 5.85232079e-01 -1.19426155e+00
-6.04635887e-02 -3.23759526e-01 2.24394783e-01 3.43398660e-01
1.18666343e-01 -1.64707720e-01 1.75335053e-02 -9.41098273e-01
-3.17444354e-01 6.79558635e-01 -3.12871903e-01 -2.69812644e-01
1.32465792e+00 -3.67959775e-02 -1.82670787e-01 1.12372184e+00
1.61068583e+00 4.18217540e-01 -9.12403584e-01 -3.32255304e-01
-2.13963725e-02 -3.64963301e-02 -1.35603040e-01 -9.58937407e-01
-1.04498231e+00 1.22467434e+00 2.00567499e-01 -2.22759470e-01
6.83172107e-01 9.65804327e-03 5.40029764e-01 4.08579677e-01
8.01256180e-01 -1.08943892e+00 -1.61895946e-01 9.61212933e-01
1.07032430e+00 -1.35543203e+00 -1.00917503e-01 -1.32987246e-01
-7.58002579e-01 8.53103757e-01 5.77157438e-01 -1.05388142e-01
5.96387982e-01 5.29302299e-01 3.16979200e-01 6.79942518e-02
-8.90599966e-01 -2.98372339e-02 1.05164587e-01 5.17121375e-01
1.33065712e+00 4.44946349e-01 7.14731812e-02 5.66365480e-01
-7.17923939e-01 -7.58428797e-02 2.63359725e-01 8.04443181e-01
-1.07935585e-01 -1.62907505e+00 -2.89145887e-01 -5.24784625e-03
-4.90123183e-01 -6.93967819e-01 -1.22021697e-01 1.32201815e+00
4.29458529e-01 6.35137439e-01 -3.52552086e-02 -1.71648070e-01
1.61923513e-01 4.97743189e-01 4.90056843e-01 -1.01327336e+00
-7.27927685e-01 1.52313272e-02 3.16963762e-01 -3.16506833e-01
-1.64650425e-01 -7.93040276e-01 -1.08104932e+00 -4.01322335e-01
-2.37067223e-01 2.94827074e-01 8.21355700e-01 6.92803204e-01
4.25958216e-01 7.15916455e-01 3.05425048e-01 -6.97570622e-01
-4.73784596e-01 -1.14985371e+00 -5.54471970e-01 1.75346345e-01
1.71445265e-01 -3.44947726e-01 -2.57065356e-01 2.44046777e-01]
|
[11.037808418273926, 9.974750518798828]
|
8dd1f8ad-96fc-4229-b216-56049f33cb3d
|
a-nonconvex-low-rank-tensor-completion-model
|
2003.10271
| null |
https://arxiv.org/abs/2003.10271v2
|
https://arxiv.org/pdf/2003.10271v2.pdf
|
A Nonconvex Low-Rank Tensor Completion Model for Spatiotemporal Traffic Data Imputation
|
Sparsity and missing data problems are very common in spatiotemporal traffic data collected from various sensing systems. Making accurate imputation is critical to many applications in intelligent transportation systems. In this paper, we formulate the missing data imputation problem in spatiotemporal traffic data in a low-rank tensor completion (LRTC) framework and define a novel truncated nuclear norm (TNN) on traffic tensors of location$\times$day$\times$time of day. In particular, we introduce an universal rate parameter to control the degree of truncation on all tensor modes in the proposed LRTC-TNN model, and this allows us to better characterize the hidden patterns in spatiotemporal traffic data. Based on the framework of the Alternating Direction Method of Multipliers (ADMM), we present an efficient algorithm to obtain the optimal solution for each variable. We conduct numerical experiments on four spatiotemporal traffic data sets, and our results show that the proposed LRTC-TNN model outperforms many state-of-the-art imputation models with missing rates/patterns. Moreover, the proposed model also outperforms other baseline models in extreme missing scenarios.
|
['Xinyu Chen', 'Lijun Sun', 'Jinming Yang']
|
2020-03-23
| null | null | null | null |
['traffic-data-imputation']
|
['time-series']
|
[ 1.25509903e-01 -6.62525892e-01 -4.28243548e-01 -5.27025342e-01
-8.49017024e-01 1.08504377e-01 1.96505293e-01 -6.49489880e-01
-1.87127218e-01 8.54296029e-01 6.34009659e-01 -2.96837419e-01
-5.32237947e-01 -5.62685490e-01 -8.99177492e-01 -9.09472585e-01
6.68781623e-02 3.43631804e-01 -3.13085616e-01 -3.37889016e-01
3.75825614e-02 2.61755675e-01 -1.37309909e+00 3.47437441e-01
1.15391159e+00 9.48448479e-01 1.20555088e-01 -1.47184618e-02
-4.28576358e-02 1.00869024e+00 1.00677550e-01 -5.46986043e-01
5.41818619e-01 5.26742004e-02 -5.26396453e-01 3.87707710e-01
4.73055184e-01 -2.93767631e-01 -7.74849534e-01 8.94390523e-01
1.23088524e-01 3.91995579e-01 4.40868348e-01 -1.65268350e+00
-5.27804673e-01 4.63676780e-01 -9.35557246e-01 2.18324468e-01
-7.82494321e-02 2.87624393e-02 7.00533330e-01 -1.34524786e+00
3.84321511e-01 1.13170397e+00 7.50818670e-01 1.00280702e-01
-1.33404613e+00 -8.40148211e-01 2.20080033e-01 3.86538565e-01
-1.65090585e+00 -7.05145597e-01 9.42170680e-01 -8.36651921e-01
3.65431756e-01 3.21266562e-01 1.94227204e-01 1.05775583e+00
-9.93748754e-03 7.58599281e-01 1.04388261e+00 2.30404720e-01
9.31622982e-02 -4.49073404e-01 1.69323906e-01 4.48190629e-01
4.82749790e-01 -2.25629844e-02 -7.41261780e-01 -3.78827512e-01
4.54516709e-01 5.57549596e-01 5.84543049e-02 -1.97443232e-01
-1.53708148e+00 7.32410669e-01 1.46109462e-01 -2.45938957e-01
-8.31282377e-01 3.53969961e-01 2.01614186e-01 9.48601589e-02
5.86367667e-01 -6.16262138e-01 -1.44579858e-01 -2.29729429e-01
-1.00081348e+00 3.61732334e-01 2.22218290e-01 1.06855905e+00
8.85440350e-01 3.76156032e-01 -3.30121607e-01 1.05621970e+00
2.41604805e-01 9.16732013e-01 -1.92778826e-01 -1.35825729e+00
1.31647491e+00 3.57099116e-01 3.32308561e-01 -1.44835842e+00
-1.39703780e-01 -4.74942595e-01 -1.49811268e+00 -5.60074985e-01
4.34050411e-01 -2.93516964e-01 -5.38335264e-01 1.72870255e+00
4.42611754e-01 7.53379345e-01 -1.07455038e-01 1.19051778e+00
3.17433894e-01 7.24745095e-01 2.56724637e-02 -4.17260706e-01
1.05526757e+00 -3.92009646e-01 -1.05811012e+00 -2.95332447e-02
6.71303511e-01 -7.79110551e-01 5.75395942e-01 3.22829127e-01
-7.68146574e-01 -5.21196842e-01 -3.60963732e-01 -1.58515543e-01
9.82922390e-02 5.14865816e-01 6.40429437e-01 3.71312946e-01
-5.07482886e-01 2.11668760e-02 -8.21941376e-01 -2.19144728e-02
3.25368732e-01 1.91849545e-01 -3.53574932e-01 -6.97260201e-01
-1.08889449e+00 5.50012767e-01 -2.91918784e-01 9.74630952e-01
-9.73292291e-01 -8.88189673e-01 -7.69334197e-01 -3.22231442e-01
5.80048263e-01 -7.45535791e-01 5.91765523e-01 -1.96850419e-01
-7.97215879e-01 3.92363876e-01 -9.15145636e-01 -3.19612890e-01
4.97079760e-01 -3.89165641e-03 -7.54716277e-01 -3.87124717e-01
5.64681351e-01 2.10967436e-01 8.49198580e-01 -1.19980872e+00
-4.94989336e-01 -4.50715750e-01 -2.76143849e-01 -2.47499853e-01
-1.52300745e-01 -1.53558120e-01 -3.24405909e-01 -9.08001781e-01
4.75796133e-01 -1.00044000e+00 -4.63360220e-01 -1.43583938e-01
-5.47108054e-01 -7.42790252e-02 8.05777490e-01 -9.18297470e-01
1.35111153e+00 -2.13191390e+00 2.94364393e-01 3.57920796e-01
2.82216609e-01 -1.89054221e-01 -1.28928751e-01 4.64738756e-01
-2.44290312e-03 -1.93828702e-01 -6.10634625e-01 -5.48496187e-01
1.05660312e-01 6.05494678e-01 -4.43710059e-01 6.05504870e-01
-1.79041959e-02 5.21732092e-01 -6.80156708e-01 -4.58383083e-01
1.45566881e-01 5.95471859e-01 -4.41394687e-01 -3.23880129e-02
1.05065159e-01 6.84294939e-01 -6.78623676e-01 6.95076108e-01
1.16611207e+00 -5.38603328e-02 2.56856494e-02 -4.88058627e-01
-4.50525433e-01 -8.69158730e-02 -1.53377450e+00 1.70703566e+00
-3.51734966e-01 3.85636330e-01 3.35427225e-01 -1.20628631e+00
8.66447568e-01 2.46742740e-01 9.06567454e-01 -8.26791286e-01
-1.76312938e-01 2.00553089e-01 -3.71981412e-01 -8.36493671e-01
6.29266679e-01 5.92921302e-02 -1.25701487e-01 2.89342791e-01
-6.37223482e-01 6.66968763e-01 4.60254490e-01 2.70910293e-01
9.79712784e-01 -2.42653280e-01 -7.43075311e-01 -1.12371869e-01
6.08188987e-01 2.94925086e-02 1.26116383e+00 6.46630824e-01
-9.26498696e-02 5.07135391e-01 4.73854929e-01 -6.71765804e-01
-1.05608165e+00 -9.16584194e-01 -2.59544700e-01 1.00371408e+00
-2.10780039e-01 -5.55139855e-02 -2.93469548e-01 -2.43491396e-01
2.43982285e-01 4.21667337e-01 -5.85134566e-01 3.17268312e-01
-7.20028937e-01 -1.12490821e+00 3.50439280e-01 3.01655382e-01
6.13952935e-01 -2.94183433e-01 2.84083486e-01 1.95319191e-01
-1.09822679e+00 -1.60539532e+00 -6.51286840e-01 -6.04707479e-01
-9.98917520e-01 -7.56712198e-01 -5.28274834e-01 -4.02524203e-01
8.21914017e-01 8.53524268e-01 7.35322297e-01 -5.28669804e-02
1.23794228e-01 1.02973156e-01 -2.60711014e-01 1.36149704e-01
3.16118926e-01 -1.34417089e-02 3.13595831e-01 1.08635867e+00
3.75428855e-01 -6.95640802e-01 -5.33699512e-01 6.15895331e-01
-9.88948584e-01 1.95438772e-01 5.38520753e-01 6.70699298e-01
7.73965895e-01 9.70997587e-02 4.16006297e-01 -6.41431510e-01
4.69702870e-01 -1.15772367e+00 -6.86477363e-01 1.10036217e-01
-1.95040300e-01 1.35578038e-02 3.42783481e-01 -2.30413005e-01
-9.85003889e-01 2.03646347e-01 7.51590803e-02 -7.30212092e-01
2.83696532e-01 8.75418603e-01 -3.73878747e-01 5.13262302e-02
9.36367512e-02 3.11684042e-01 1.02491044e-02 -7.54169941e-01
2.15046436e-01 5.21712780e-01 3.71378928e-01 -8.96358967e-01
9.39021289e-01 9.29233015e-01 3.04107189e-01 -1.05962825e+00
-9.46938336e-01 -4.91735458e-01 -6.17727995e-01 -2.85526186e-01
5.85345149e-01 -1.18603516e+00 -8.94568622e-01 3.22994292e-01
-9.70898151e-01 5.53584993e-02 -1.04089119e-01 8.23224962e-01
-2.95212626e-01 4.83097106e-01 -4.09413040e-01 -8.84331405e-01
2.27546558e-01 -1.25560629e+00 8.91187787e-01 -5.02849519e-01
4.11974847e-01 -6.40061975e-01 5.97898886e-02 1.08645046e+00
5.38677216e-01 3.87260407e-01 7.26992905e-01 1.77039027e-01
-8.13109100e-01 4.93463501e-02 -3.56169283e-01 3.21018815e-01
-1.18240677e-02 -2.47584119e-01 -4.83431309e-01 -6.07431531e-02
-3.76260281e-02 1.78317323e-01 1.12537289e+00 6.21747494e-01
1.41929579e+00 -6.89300478e-01 -8.30712840e-02 7.19924629e-01
1.31063974e+00 -3.70934188e-01 8.34063113e-01 -1.16417959e-01
1.14694941e+00 6.99230194e-01 5.78963995e-01 7.95152366e-01
1.02577722e+00 6.62245631e-01 3.45882326e-01 -8.19977373e-03
1.37583628e-01 -1.19098440e-01 3.66585523e-01 1.25705564e+00
-2.76307106e-01 5.93468845e-02 -8.71718884e-01 9.24107075e-01
-2.26935625e+00 -1.29937053e+00 -8.84179533e-01 2.32935214e+00
3.46386075e-01 -3.67758662e-01 3.94984692e-01 1.46800578e-01
8.41637015e-01 2.96964705e-01 -4.55805749e-01 -9.43738893e-02
-3.34433317e-01 -2.40216643e-01 7.73505509e-01 4.37743306e-01
-8.93347800e-01 5.18703818e-01 5.87413597e+00 7.61708736e-01
-5.74272215e-01 4.73719805e-01 4.81027782e-01 -1.68351233e-01
-5.02783298e-01 3.99637744e-02 -5.94257295e-01 8.14805806e-01
9.38568473e-01 1.41834542e-01 7.00961888e-01 3.75495881e-01
1.15045226e+00 1.29935250e-01 -6.29492939e-01 1.11943030e+00
-1.49918333e-01 -1.42288804e+00 2.18307808e-01 5.59865415e-01
9.57402408e-01 2.15283677e-01 2.01141357e-01 1.08182974e-01
2.27638826e-01 -8.51016998e-01 3.05847168e-01 9.22441483e-01
5.71285367e-01 -7.43449450e-01 5.13993621e-01 2.46239290e-01
-1.29540074e+00 -2.33363792e-01 -5.95916092e-01 -1.59347221e-01
5.92825532e-01 1.21842766e+00 -1.65587023e-01 7.54082799e-01
6.26402557e-01 1.01645505e+00 -2.98475504e-01 9.85636830e-01
2.60327488e-01 9.29335475e-01 -3.54379177e-01 5.67002058e-01
2.18845800e-01 -8.30357373e-01 6.09224796e-01 8.74553680e-01
5.28411984e-01 3.35222781e-01 3.36320996e-01 7.84229040e-01
-1.39905497e-01 -1.33277625e-01 -6.04390025e-01 2.95596898e-01
6.39881372e-01 9.65990305e-01 2.13701025e-01 -1.19241513e-01
-5.40555835e-01 4.42278117e-01 1.75879881e-01 7.52415836e-01
-9.83282208e-01 1.63097724e-01 1.11097229e+00 2.55119205e-01
2.94014812e-01 -7.37684131e-01 -3.56699705e-01 -1.41925216e+00
5.76012313e-01 -7.45270610e-01 3.38759184e-01 -6.69656396e-01
-1.55394471e+00 9.30300429e-02 4.28829491e-02 -1.67177331e+00
1.64702892e-01 -1.50769934e-01 -2.73528099e-01 6.98078811e-01
-1.54770565e+00 -1.33632684e+00 -2.94229418e-01 1.15793335e+00
2.36570135e-01 1.64542831e-02 1.85933739e-01 1.07381690e+00
-1.22608483e+00 3.37399453e-01 5.42653978e-01 3.37800533e-01
2.32899830e-01 -4.64847296e-01 1.33251324e-01 1.26429772e+00
-2.48016983e-01 7.43571103e-01 6.77278757e-01 -7.48595417e-01
-1.98522139e+00 -1.50803375e+00 1.08844185e+00 -2.27415517e-01
6.93002760e-01 -1.95166618e-01 -6.97976589e-01 8.84311676e-01
-2.45240942e-01 2.65773177e-01 7.34615386e-01 2.59885877e-01
-5.24536490e-01 -7.55272031e-01 -9.98920441e-01 3.63093525e-01
1.19833553e+00 -3.98702919e-01 -2.03825280e-01 6.16202295e-01
6.17702246e-01 -1.77952975e-01 -9.63251829e-01 5.04986167e-01
5.75930238e-01 -5.76690435e-01 1.08372486e+00 -7.59239733e-01
2.58528978e-01 -8.22096288e-01 -7.43186653e-01 -8.18099320e-01
-4.76799875e-01 -5.97566903e-01 -2.25703865e-01 1.28441775e+00
4.17228729e-01 -5.67091167e-01 7.17507541e-01 9.93526578e-01
-2.12556884e-01 -3.80181551e-01 -1.49041760e+00 -6.89328671e-01
-1.70147076e-01 -1.04782701e+00 7.83083141e-01 1.06544971e+00
-6.37585640e-01 -2.07172502e-02 -1.30419755e+00 4.84285235e-01
1.34545410e+00 -7.66865583e-03 9.45047319e-01 -1.16658199e+00
2.24238142e-01 2.68594146e-01 -2.67860383e-01 -1.11174273e+00
2.59580582e-01 -6.99366152e-01 -3.01206201e-01 -1.38369799e+00
4.00561064e-01 -7.60744810e-01 -3.81497800e-01 4.03993934e-01
1.99653193e-01 1.81485534e-01 4.75799367e-02 5.35160840e-01
-7.11244464e-01 9.32345092e-01 1.03131878e+00 -4.45554972e-01
8.10541511e-02 1.01563679e-02 -5.95878243e-01 3.09789211e-01
4.66109425e-01 -7.08531499e-01 -3.17807198e-01 -1.10744190e+00
3.91243368e-01 3.81473899e-01 5.68801880e-01 -8.06866229e-01
4.30508465e-01 -7.71970630e-01 7.31611997e-02 -1.06145310e+00
6.02361619e-01 -1.09760034e+00 6.19412243e-01 4.43466455e-02
-1.37868807e-01 3.16835225e-01 -2.90046453e-01 8.32479119e-01
-1.59822002e-01 4.01231378e-01 1.23459771e-01 2.63035357e-01
-5.56638718e-01 7.13075519e-01 -5.49374402e-01 1.18234433e-01
6.57683492e-01 -1.74277216e-01 -3.21917683e-01 -3.16599488e-01
-7.11406052e-01 6.87798679e-01 -1.62962973e-02 4.49761927e-01
8.23632061e-01 -1.87334478e+00 -1.16937006e+00 2.52254725e-01
1.26388997e-01 -1.75047860e-01 9.25634921e-01 1.56882620e+00
-6.66066706e-02 4.24297243e-01 7.45696947e-02 -7.88239658e-01
-6.29687726e-01 4.18188542e-01 5.47130108e-02 5.71651012e-02
-3.48444134e-01 2.13437706e-01 -2.53493160e-01 -6.57024384e-01
9.41901356e-02 -2.06055865e-01 1.14773266e-01 -8.47529918e-02
3.66177768e-01 1.06797111e+00 2.06813514e-01 -1.21950376e+00
-3.75403494e-01 6.76347017e-01 2.46629104e-01 -7.17482865e-02
1.44328916e+00 -4.86666322e-01 -3.57113272e-01 3.47562253e-01
1.25047779e+00 -9.91319567e-02 -1.15857339e+00 -5.02290905e-01
-2.51695752e-01 -8.94802153e-01 8.17700401e-02 -2.20732227e-01
-1.54284012e+00 6.51484966e-01 4.58183616e-01 -1.54334009e-01
9.73580599e-01 -5.26615202e-01 1.20117319e+00 3.21631610e-01
5.64920723e-01 -1.04123747e+00 -5.68290830e-01 5.37386119e-01
8.04170072e-01 -1.26104093e+00 -5.55680059e-02 -6.07085049e-01
-7.06998944e-01 5.70302904e-01 2.17321724e-01 1.24230627e-02
8.31938326e-01 -1.66644454e-01 -2.16350332e-01 -2.01366082e-01
-7.17939675e-01 -3.68017614e-01 2.51199365e-01 4.18322533e-01
2.13892639e-01 3.31935644e-01 -4.35230136e-01 2.10326090e-01
4.90852725e-03 1.84440538e-01 5.57873964e-01 7.35176980e-01
3.56705822e-02 -1.12616313e+00 -7.21320510e-01 5.90014100e-01
-2.65315264e-01 4.73272055e-02 1.61242694e-01 3.43949556e-01
2.71599591e-01 1.55073583e+00 -4.74224798e-02 -5.41453421e-01
3.55682939e-01 -1.73393011e-01 -2.18777377e-02 -4.70234826e-03
-7.11372942e-02 -9.95285660e-02 9.93639529e-02 -6.52032971e-01
-6.60934567e-01 -1.04690611e+00 -7.26836622e-01 -9.66938615e-01
4.42586392e-02 3.71692806e-01 5.79374313e-01 1.05424714e+00
7.35492647e-01 3.11580580e-02 9.92760360e-01 -5.89250445e-01
-2.15630099e-01 -7.49269962e-01 -6.10576630e-01 6.74255073e-01
5.22836149e-01 -8.87513578e-01 -3.41945559e-01 9.98011231e-02]
|
[6.57404899597168, 2.126007556915283]
|
936ddfa8-d0ea-4e89-b836-5380a1b0abec
|
evaluating-the-logical-reasoning-ability-of
|
2304.03439
| null |
https://arxiv.org/abs/2304.03439v3
|
https://arxiv.org/pdf/2304.03439v3.pdf
|
Evaluating the Logical Reasoning Ability of ChatGPT and GPT-4
|
Harnessing logical reasoning ability is a comprehensive natural language understanding endeavor. With the release of Generative Pretrained Transformer 4 (GPT-4), highlighted as "advanced" at reasoning tasks, we are eager to learn the GPT-4 performance on various logical reasoning tasks. This report analyses multiple logical reasoning datasets, with popular benchmarks like LogiQA and ReClor, and newly-released datasets like AR-LSAT. We test the multi-choice reading comprehension and natural language inference tasks with benchmarks requiring logical reasoning. We further construct a logical reasoning out-of-distribution dataset to investigate the robustness of ChatGPT and GPT-4. We also make a performance comparison between ChatGPT and GPT-4. Experiment results show that ChatGPT performs significantly better than the RoBERTa fine-tuning method on most logical reasoning benchmarks. With early access to the GPT-4 API we are able to conduct intense experiments on the GPT-4 model. The results show GPT-4 yields even higher performance on most logical reasoning datasets. Among benchmarks, ChatGPT and GPT-4 do relatively well on well-known datasets like LogiQA and ReClor. However, the performance drops significantly when handling newly released and out-of-distribution datasets. Logical reasoning remains challenging for ChatGPT and GPT-4, especially on out-of-distribution and natural language inference datasets. We release the prompt-style logical reasoning datasets as a benchmark suite and name it LogiEval.
|
['Yue Zhang', 'Qiji Zhou', 'Jian Liu', 'Zhiyang Teng', 'Ruoxi Ning', 'Hanmeng Liu']
|
2023-04-07
| null | null | null | null |
['reading-comprehension', 'logical-reasoning']
|
['natural-language-processing', 'reasoning']
|
[-3.68748069e-01 5.11859179e-01 -1.44170508e-01 -5.21541536e-01
-8.40750277e-01 -7.32322156e-01 7.02564657e-01 5.47414012e-02
-1.05304027e-03 7.98219025e-01 3.63360554e-01 -1.09674048e+00
-5.60623586e-01 -1.34636128e+00 -9.88118529e-01 4.03410345e-02
1.92575499e-01 1.07658422e+00 2.63474405e-01 -4.89785939e-01
1.72503725e-01 3.85642312e-02 -1.27983093e+00 7.99085557e-01
8.64988863e-01 6.60076439e-01 -3.53241801e-01 8.33026230e-01
-2.93617457e-01 1.73037720e+00 -3.46636832e-01 -8.33422959e-01
1.06705099e-01 -2.13545263e-01 -1.50849509e+00 -8.75637591e-01
8.11351061e-01 -1.20211340e-01 -4.08580936e-02 6.96024239e-01
3.76291633e-01 3.36941294e-02 2.95187086e-01 -1.61414886e+00
-7.17286408e-01 1.54056346e+00 2.62828935e-02 1.35321289e-01
8.57246399e-01 5.42883813e-01 1.49230719e+00 -5.45937061e-01
7.88524926e-01 1.81117034e+00 1.03507888e+00 3.81726414e-01
-1.21875393e+00 -5.01015186e-01 7.63856694e-02 6.00895882e-01
-1.13791108e+00 -2.53066331e-01 2.20194280e-01 -9.11469385e-02
1.57840538e+00 3.90736014e-01 1.97121873e-01 1.17221332e+00
3.70943815e-01 9.09819841e-01 1.38019192e+00 -4.02130485e-01
2.31723383e-01 -1.49320900e-01 6.05820656e-01 1.14865303e+00
-1.14486717e-01 -3.40474129e-01 -6.34966850e-01 -1.67484432e-01
4.26613510e-01 -5.21175325e-01 7.50066116e-02 1.17397755e-01
-1.30336499e+00 6.29950047e-01 1.22879475e-01 1.54990926e-01
-1.31112978e-01 5.08501709e-01 6.25837564e-01 8.57260585e-01
1.58271112e-03 8.74062181e-01 -8.09569240e-01 -6.56486511e-01
-5.24379730e-01 1.00696945e+00 1.42513192e+00 1.10038137e+00
2.80365020e-01 -3.76317024e-01 -4.25072014e-01 5.90700865e-01
5.16247861e-02 4.69243944e-01 8.53480548e-02 -1.48802698e+00
6.21621251e-01 7.86423326e-01 -3.52555215e-01 -6.70917928e-01
-4.84947205e-01 5.84154949e-02 -4.39034104e-01 2.33931676e-01
9.16863501e-01 -3.52244377e-02 -4.22767133e-01 1.80726302e+00
-1.57359093e-01 -7.26224259e-02 5.48547208e-01 4.12033767e-01
1.18228793e+00 5.33040404e-01 3.87194604e-02 2.51133442e-01
1.58631384e+00 -9.83845115e-01 -4.51749146e-01 -8.37737992e-02
9.36740339e-01 -5.11221647e-01 1.71705985e+00 8.68916869e-01
-1.56506324e+00 -3.29540431e-01 -6.99867785e-01 -6.16153955e-01
-5.42528808e-01 -4.23630148e-01 1.07325625e+00 2.11141095e-01
-1.14029431e+00 3.40823293e-01 -7.40742505e-01 -5.31150222e-01
3.37954432e-01 -3.61758508e-02 -2.62429446e-01 -5.73877573e-01
-1.39216232e+00 1.22501147e+00 5.57025194e-01 -1.91015467e-01
-7.26260185e-01 -1.28920197e+00 -8.83126974e-01 2.27191731e-01
8.83705258e-01 -9.83016372e-01 1.56153286e+00 9.69721675e-02
-1.39593840e+00 8.51125240e-01 -1.69508800e-01 -7.61768699e-01
1.01531887e+00 -4.16908711e-01 -2.39459738e-01 -1.02451600e-01
3.81170392e-01 5.21201074e-01 2.93071233e-02 -6.62975490e-01
-2.73845881e-01 1.55566812e-01 6.88556135e-01 -1.45435154e-01
4.17954296e-01 -4.20816578e-02 -1.74656689e-01 -2.23509014e-01
-1.23769924e-01 -6.34110570e-01 5.13383031e-01 -1.59355417e-01
-7.14014113e-01 -9.32090878e-01 7.97835290e-01 -3.34560961e-01
8.37776661e-01 -1.63403535e+00 2.35186219e-02 -7.44468346e-03
2.66888201e-01 -4.91357036e-02 -1.36074305e-01 5.87906241e-01
-2.05457494e-01 3.36230248e-01 4.23637815e-02 2.52518542e-02
7.15266883e-01 7.18207061e-01 -6.34038627e-01 -2.11385101e-01
3.02017987e-01 1.44247937e+00 -9.40181196e-01 -8.49740148e-01
1.89823642e-01 -2.59218037e-01 -1.01954043e+00 7.45349750e-02
-1.33326054e+00 -5.14483312e-03 -4.74342555e-01 8.60035777e-01
4.75517482e-01 -4.64429736e-01 2.20375836e-01 -4.83956828e-04
1.85976066e-02 8.99953485e-01 -8.69000196e-01 1.84040534e+00
-6.67926729e-01 5.61798215e-01 -5.46962380e-01 -5.89534998e-01
5.09509742e-01 1.68459579e-01 -1.32393822e-01 -9.95172262e-01
-1.92250565e-01 2.24303771e-02 9.93236676e-02 -7.49800026e-01
3.46990466e-01 -3.05998713e-01 -4.06823993e-01 7.76188374e-01
4.26396355e-03 -5.79176307e-01 5.86232960e-01 6.19631588e-01
1.69745147e+00 4.18676645e-01 8.91564637e-02 -4.11872745e-01
4.88064319e-01 3.39137524e-01 2.12882012e-01 1.12270343e+00
3.27475786e-01 1.60363525e-01 1.27416253e+00 -4.84064966e-01
-6.00503385e-01 -1.22409356e+00 -1.84358448e-01 1.25196981e+00
-1.33291081e-01 -1.06861699e+00 -4.74899381e-01 -8.35631132e-01
1.65421322e-01 1.57831013e+00 -4.26470876e-01 1.97209865e-02
-6.27170324e-01 -4.24604774e-01 1.35067725e+00 7.11571455e-01
1.00137615e+00 -1.24129522e+00 -3.21700156e-01 9.36553814e-03
-4.99787480e-01 -1.43020868e+00 4.78372633e-01 9.02606770e-02
-3.98872286e-01 -1.39587343e+00 3.69638234e-01 -3.37223619e-01
8.26672241e-02 -4.77159470e-01 1.78224659e+00 4.30364817e-01
-2.81328168e-02 4.60141271e-01 -3.52872819e-01 -2.72315055e-01
-7.39837766e-01 2.37623781e-01 -4.38500136e-01 -9.29880023e-01
4.08157974e-01 -7.59991884e-01 2.93660283e-01 1.86594889e-01
-4.76533175e-01 4.27944988e-01 1.55505314e-01 7.07507908e-01
1.15680903e-01 7.28910446e-01 1.39283147e-02 -1.14261413e+00
7.59427726e-01 -4.83790755e-01 -5.06913185e-01 4.98654693e-01
-5.88912010e-01 5.52726030e-01 9.08832133e-01 1.20332323e-01
-1.28209078e+00 -1.21799600e+00 -2.22247288e-01 -6.22875057e-02
2.70382334e-02 7.11498201e-01 -1.82393849e-01 5.36654592e-01
7.87829638e-01 -8.03486034e-02 -2.62221336e-01 -2.31777295e-01
5.38665652e-01 -3.94179896e-02 9.43865180e-01 -1.67763138e+00
9.07876372e-01 1.55549701e-02 1.05143711e-01 -3.47066194e-01
-1.11186361e+00 4.05731678e-01 -1.52211010e-01 2.78973788e-01
6.84151113e-01 -8.99799466e-01 -1.44753647e+00 3.83861065e-01
-1.11918092e+00 -1.24761868e+00 -3.28319162e-01 1.20102167e-01
-7.10437894e-01 3.28241363e-02 -8.73529196e-01 -5.32216668e-01
-3.15171391e-01 -1.18433702e+00 9.47166562e-01 -1.09509528e-01
-8.67122233e-01 -1.50688112e+00 -1.23953655e-01 7.66413391e-01
3.23661923e-01 2.38438398e-01 1.73656869e+00 -7.14732409e-01
-8.38568330e-01 1.04535438e-01 -4.72127944e-01 2.66065925e-01
-4.02977049e-01 2.00380325e-01 -8.28135908e-01 3.24943691e-01
-3.31649035e-01 -9.23293769e-01 6.20624423e-01 -7.83639029e-03
1.30205274e+00 -3.35952163e-01 6.96800649e-02 5.85906565e-01
1.11637139e+00 -3.45674396e-01 8.93381476e-01 5.82873166e-01
5.99972486e-01 1.41147643e-01 5.30023634e-01 5.68315722e-02
1.12084365e+00 2.67821610e-01 2.42246956e-01 4.88445818e-01
-1.49911836e-01 -5.79780757e-01 4.49123532e-01 2.69631535e-01
-1.58108443e-01 -4.57866013e-01 -1.38039839e+00 1.38268799e-01
-2.04546952e+00 -1.31457138e+00 -4.23262805e-01 1.41970205e+00
1.35218418e+00 6.26269877e-01 -1.86592504e-01 1.98709041e-01
-1.58167660e-01 4.62130345e-02 -3.78221691e-01 -8.12856913e-01
-2.51698047e-01 7.78575122e-01 -3.35055217e-02 8.16338658e-01
-5.21888137e-01 1.25476158e+00 6.22537804e+00 7.73419619e-01
-6.87539279e-01 -8.62989202e-02 1.58897176e-01 -3.94392125e-02
-6.36561334e-01 2.72351891e-01 -7.55490482e-01 1.55095279e-01
9.87958908e-01 -1.48927405e-01 6.85444295e-01 6.94032729e-01
-8.90504792e-02 -5.33409715e-01 -1.71052015e+00 4.96411324e-01
-2.26111308e-01 -1.47386587e+00 6.11689687e-02 -3.73182833e-01
3.17597777e-01 -8.89009386e-02 -1.79147243e-01 1.16500330e+00
1.00837076e+00 -1.48720920e+00 9.05426502e-01 5.88774621e-01
4.61964786e-01 -4.57737416e-01 7.70831943e-01 4.27198589e-01
-7.76737571e-01 6.67947456e-02 -1.80178240e-01 -6.19619966e-01
1.21964356e-02 6.40395880e-01 -9.14189696e-01 6.86078668e-01
7.76423931e-01 9.61481869e-01 -9.43396807e-01 2.28265494e-01
-1.02459931e+00 7.23097682e-01 -4.34230894e-01 1.32935531e-02
1.76505774e-01 2.04765365e-01 2.77511686e-01 1.16210866e+00
-1.70034006e-01 2.86188185e-01 5.54024354e-02 1.54756212e+00
-1.84198603e-01 -5.83943725e-01 -1.84167892e-01 -1.73520327e-01
4.48232979e-01 1.10045874e+00 -2.36853287e-01 -8.04843485e-01
-1.42444715e-01 4.19362545e-01 5.36502600e-01 5.86625561e-02
-1.21784544e+00 -2.01652274e-02 6.08130395e-01 1.17493652e-01
-1.22272737e-01 3.45597491e-02 -4.88991976e-01 -1.22269487e+00
1.40402079e-01 -1.31920958e+00 7.65175939e-01 -1.53915608e+00
-1.51052022e+00 2.63630182e-01 8.04841280e-01 -1.11813754e-01
-4.24434155e-01 -7.34368622e-01 -7.26095498e-01 8.62962782e-01
-1.45496511e+00 -1.22481883e+00 -5.78053892e-01 6.98427200e-01
4.03414041e-01 -5.81987426e-02 9.31215167e-01 -5.85929900e-02
-5.23802638e-01 6.29454374e-01 -7.02247381e-01 2.64034957e-01
6.47534609e-01 -1.78698134e+00 8.25668216e-01 6.67491794e-01
-2.16601059e-01 9.89578068e-01 8.15712571e-01 -4.93399084e-01
-1.74961436e+00 -7.45018542e-01 1.03536808e+00 -1.04872763e+00
1.11496294e+00 -3.58828694e-01 -1.00076354e+00 1.48591983e+00
3.96759868e-01 -1.92559302e-01 1.56941667e-01 6.01790965e-01
-9.24802601e-01 1.10137589e-01 -1.15731120e+00 8.26681852e-01
1.40302229e+00 -7.85635471e-01 -1.35297501e+00 7.37052679e-01
7.43733883e-01 -9.43335176e-01 -1.14974904e+00 2.94451714e-01
6.27132356e-01 -1.29376781e+00 8.88613701e-01 -6.60697758e-01
1.13641179e+00 -1.46649376e-01 -2.18400031e-01 -1.08509076e+00
1.15319928e-02 -6.88478649e-01 2.94063836e-02 1.29493117e+00
4.71706539e-01 -1.12591565e+00 5.45274675e-01 8.85603487e-01
-3.25937904e-02 -5.94424307e-01 -4.43739355e-01 -5.95728219e-01
5.65161943e-01 -1.00319993e+00 6.11471355e-01 9.14225280e-01
3.80758762e-01 5.26424646e-01 3.70185167e-01 1.19173825e-02
5.40439188e-01 3.82129401e-01 1.23146915e+00 -9.63866770e-01
-7.16674387e-01 -5.25960445e-01 1.97113724e-03 -7.09889591e-01
7.06329167e-01 -1.34273791e+00 -1.50098875e-01 -1.76823425e+00
9.84460190e-02 -5.30544281e-01 2.49122724e-01 1.39019358e+00
1.98079273e-02 -2.50423431e-01 7.64158741e-02 -2.62151241e-01
-5.91472745e-01 2.44484097e-02 1.27297521e+00 -2.49907061e-01
2.60058969e-01 -3.00408810e-01 -8.19926798e-01 7.70567715e-01
6.78089797e-01 -2.18112752e-01 -8.03958178e-01 -5.68346500e-01
8.40316772e-01 1.66206583e-01 9.70592856e-01 -1.12707615e+00
1.34804174e-01 -3.44492257e-01 6.11851970e-03 -5.86059928e-01
9.88492966e-02 -1.99360877e-01 1.02867723e-01 1.87660009e-01
-6.15218699e-01 3.74365419e-01 4.45867747e-01 -1.64933562e-01
-8.22619498e-02 -2.18774397e-02 4.79925513e-01 -4.50253129e-01
-8.04733336e-01 -4.79742855e-01 -2.54396647e-01 8.68072093e-01
6.12698615e-01 8.97657871e-02 -1.07952487e+00 -2.17413872e-01
-4.37903970e-01 7.12628663e-01 3.49466264e-01 3.67859393e-01
3.59702379e-01 -8.37415993e-01 -8.27196002e-01 1.12168312e-01
1.44721195e-01 4.73952860e-01 6.24840483e-02 1.15850687e+00
-9.79265392e-01 5.03827393e-01 -1.44054055e-01 -4.70162183e-01
-1.13190091e+00 3.63381088e-01 5.28333008e-01 -7.64873028e-01
-8.55886996e-01 9.31832075e-01 -1.76386312e-01 -8.84803593e-01
-3.59452292e-02 -1.17117512e+00 4.54785734e-01 -4.44012582e-01
4.47643220e-01 2.92136341e-01 1.20070398e-01 3.57954681e-01
-4.66034532e-01 1.98272660e-01 -4.18445803e-02 -9.03844610e-02
1.41331160e+00 4.76687431e-01 -7.07645297e-01 4.44721967e-01
8.42372179e-01 8.05504471e-02 -4.77890998e-01 -1.49745047e-01
1.24068521e-02 -2.39022803e-02 -2.82326072e-01 -1.43769562e+00
-5.86507797e-01 7.00674891e-01 -4.99410808e-01 1.68055549e-01
7.07737625e-01 9.19554234e-02 6.24278307e-01 9.22775745e-01
7.27712572e-01 -5.75425506e-01 5.16711595e-03 1.12780738e+00
9.57229495e-01 -9.71166432e-01 2.45216668e-01 -2.23274872e-01
-5.01124680e-01 1.03606904e+00 7.81789422e-01 -1.14060389e-02
1.87763393e-01 6.45065427e-01 -9.37057585e-02 -6.17632866e-01
-1.49413955e+00 1.16447940e-01 -2.19638184e-01 2.73245811e-01
4.68666583e-01 4.43242639e-02 1.46961078e-01 4.31053549e-01
-1.12762237e+00 3.67156893e-01 4.54507858e-01 1.13562799e+00
1.79173410e-01 -1.26261961e+00 -4.22112167e-01 2.32062891e-01
-8.51614922e-02 -4.63519543e-01 -4.26161736e-01 1.41686237e+00
-1.56481881e-02 8.99376988e-01 -8.09269175e-02 -8.83051232e-02
5.17647684e-01 5.22365689e-01 9.79285181e-01 -5.98056436e-01
-8.69321942e-01 -8.71468365e-01 7.90483475e-01 -8.71462345e-01
-1.01763807e-01 -5.23730516e-01 -1.66003704e+00 -1.21051991e+00
2.43079588e-01 1.48574695e-01 -3.44370194e-02 1.24583733e+00
7.63831139e-02 7.94795275e-01 -5.59294641e-01 -1.29209995e-01
-5.81639647e-01 -8.35096896e-01 -1.86773270e-01 3.10776949e-01
-1.14600107e-01 -4.18820620e-01 -1.78134829e-01 -1.51165396e-01]
|
[9.625263214111328, 7.407289981842041]
|
876a917a-f9b6-4900-b723-f8b4a4e7b178
|
supervised-visualization-for-data-exploration
|
2006.08701
| null |
https://arxiv.org/abs/2006.08701v1
|
https://arxiv.org/pdf/2006.08701v1.pdf
|
Supervised Visualization for Data Exploration
|
Dimensionality reduction is often used as an initial step in data exploration, either as preprocessing for classification or regression or for visualization. Most dimensionality reduction techniques to date are unsupervised; they do not take class labels into account (e.g., PCA, MDS, t-SNE, Isomap). Such methods require large amounts of data and are often sensitive to noise that may obfuscate important patterns in the data. Various attempts at supervised dimensionality reduction methods that take into account auxiliary annotations (e.g., class labels) have been successfully implemented with goals of increased classification accuracy or improved data visualization. Many of these supervised techniques incorporate labels in the loss function in the form of similarity or dissimilarity matrices, thereby creating over-emphasized separation between class clusters, which does not realistically represent the local and global relationships in the data. In addition, these approaches are often sensitive to parameter tuning, which may be difficult to configure without an explicit quantitative notion of visual superiority. In this paper, we describe a novel supervised visualization technique based on random forest proximities and diffusion-based dimensionality reduction. We show, both qualitatively and quantitatively, the advantages of our approach in retaining local and global structures in data, while emphasizing important variables in the low-dimensional embedding. Importantly, our approach is robust to noise and parameter tuning, thus making it simple to use while producing reliable visualizations for data exploration.
|
['Kevin R. Moon', 'Jake S. Rhodes', 'Guy Wolf', 'Adele Cutler']
|
2020-06-15
| null | null | null | null |
['supervised-dimensionality-reduction']
|
['computer-vision']
|
[-6.77743321e-03 -2.02486888e-01 -1.44497547e-02 -3.09032112e-01
-2.67685615e-02 -7.75904179e-01 8.05095792e-01 6.72932923e-01
-2.40834922e-01 4.45187300e-01 3.27531755e-01 -3.26457769e-01
-6.03050768e-01 -8.18347633e-01 8.95589963e-02 -9.71396744e-01
-3.25754166e-01 3.98346454e-01 -4.56447825e-02 -8.21292028e-02
5.05087912e-01 8.75446498e-01 -1.77887475e+00 -1.68730631e-01
9.11103070e-01 6.19064391e-01 -9.72716510e-03 3.72706920e-01
-5.19243062e-01 2.87286043e-01 -6.82667494e-01 -6.01344779e-02
1.63020730e-01 -3.99133831e-01 -4.68154460e-01 1.51500376e-02
8.96224082e-02 1.59196854e-01 3.67453583e-02 1.15923274e+00
3.58760685e-01 1.10379137e-01 9.04906392e-01 -1.34990370e+00
-5.70264220e-01 1.81652874e-01 -7.54158795e-01 1.04705423e-01
2.53758073e-01 1.76851660e-01 8.46966386e-01 -9.67865944e-01
6.77072227e-01 1.43899357e+00 3.74204248e-01 1.17977686e-01
-1.87331235e+00 -4.40845370e-01 1.24012968e-02 1.39344499e-01
-1.41197908e+00 -1.81988403e-01 1.31022966e+00 -8.53522182e-01
5.04660547e-01 7.41299808e-01 7.31920540e-01 5.69190741e-01
-5.22930622e-02 2.52348244e-01 1.40976012e+00 -5.49233079e-01
5.52528501e-01 3.81305486e-01 3.53859484e-01 3.47801417e-01
5.46369255e-01 5.77599742e-02 -3.26072633e-01 -2.03298718e-01
5.11183143e-01 2.28388444e-01 -9.18651000e-02 -1.21122265e+00
-1.21796978e+00 9.74257529e-01 5.78320324e-01 3.51928979e-01
-2.41555244e-01 -3.03842127e-01 4.46081966e-01 2.69722760e-01
5.93419909e-01 7.71337450e-01 6.91655725e-02 -2.57276952e-01
-8.08500588e-01 1.54335231e-01 3.76888543e-01 3.30149561e-01
9.65718091e-01 -1.78919598e-01 1.14072293e-01 7.80878365e-01
2.63666153e-01 1.84390143e-01 6.75095677e-01 -7.42026091e-01
2.03578293e-01 1.08506596e+00 1.34643028e-02 -1.64171433e+00
-6.90967858e-01 -7.13465065e-02 -8.82334232e-01 9.39436316e-01
4.08818722e-01 1.11509219e-01 -9.04417038e-01 1.56567264e+00
5.47344863e-01 -5.92138350e-01 -5.21565303e-02 8.64379346e-01
5.66144884e-01 2.92139828e-01 2.49485344e-01 -2.51784652e-01
1.27228069e+00 -5.42354345e-01 -1.00735128e+00 1.45831093e-01
9.14443552e-01 -5.36357522e-01 1.59579730e+00 3.57309550e-01
-7.24031687e-01 -3.32072109e-01 -1.06910181e+00 -3.35508548e-02
-8.76272917e-01 -2.96994150e-02 6.40448809e-01 8.02214146e-01
-7.67789483e-01 7.26745605e-01 -1.02195001e+00 -4.36597645e-01
4.05949414e-01 2.85105735e-01 -5.45810401e-01 1.68216839e-01
-7.53921449e-01 8.83896172e-01 4.33197290e-01 -7.64243379e-02
-1.41664833e-01 -3.53857666e-01 -7.24387169e-01 1.04637668e-01
1.08256906e-01 -1.78115830e-01 3.86888176e-01 -6.09407187e-01
-1.16073036e+00 6.66159630e-01 -4.15855348e-02 -4.23602723e-02
4.68030542e-01 -2.77687237e-02 -4.78627235e-01 1.61713883e-02
-1.90409958e-01 6.60614789e-01 7.01211870e-01 -1.42582881e+00
-1.53147802e-01 -6.64640248e-01 -1.12208292e-01 3.37783754e-01
-7.51783669e-01 -1.02093764e-01 7.08268285e-02 -8.19711208e-01
4.84280825e-01 -8.05132031e-01 -2.65215069e-01 4.66619343e-01
-4.48068470e-01 -6.29009232e-02 1.38676429e+00 -4.83389705e-01
1.64246821e+00 -2.36374640e+00 3.16418409e-01 5.55117369e-01
7.14710057e-01 2.09282428e-01 2.30009481e-01 6.12128675e-01
-4.39685732e-01 4.42330599e-01 -4.39748853e-01 -3.15083504e-01
3.13063636e-02 1.45080924e-01 6.38786554e-02 5.32878578e-01
1.34438351e-01 5.81028998e-01 -8.39634180e-01 -5.20437300e-01
5.05035639e-01 6.51945293e-01 -3.34415764e-01 -2.66712811e-02
1.88612774e-01 3.54989856e-01 -8.86481181e-02 3.36379230e-01
5.75306118e-01 -1.05346655e-02 2.16846198e-01 -1.54383853e-01
-3.82022530e-01 2.97998428e-01 -1.37673950e+00 1.33827055e+00
-6.46816939e-02 1.01666021e+00 -1.71101972e-01 -9.68206048e-01
1.23602509e+00 -1.54329196e-01 5.14082253e-01 -5.17003179e-01
-1.67910472e-01 -2.31160466e-02 3.07541341e-01 -3.70231777e-01
4.82947111e-01 -9.96277556e-02 2.63682663e-01 7.30976820e-01
-3.43340039e-01 5.92861734e-02 3.76952648e-01 1.19800925e-01
7.61975229e-01 -9.63717401e-02 3.36991102e-01 -4.14526194e-01
3.41880977e-01 1.72377616e-01 2.49698669e-01 1.91383615e-01
1.40182033e-01 6.28454864e-01 7.77548015e-01 -4.97743160e-01
-1.04382825e+00 -8.77695799e-01 -2.32610539e-01 6.22624815e-01
1.57830685e-01 -6.99185908e-01 -6.02224231e-01 -6.99056566e-01
4.72762212e-02 7.06110597e-01 -8.75867546e-01 -2.91237801e-01
-3.65525991e-01 -9.14411902e-01 7.30588213e-02 2.87189424e-01
-1.15844691e-02 -8.50197136e-01 -8.50992620e-01 -6.26048520e-02
2.53469169e-01 -2.14471027e-01 9.18175653e-03 3.03346992e-01
-1.35503495e+00 -1.09287655e+00 -5.52945077e-01 -3.01105231e-01
1.14154410e+00 4.53111261e-01 6.98912024e-01 -1.59432605e-01
-6.46930575e-01 -7.78419599e-02 -2.15580344e-01 -2.47600332e-01
-2.00658917e-01 -6.05765134e-02 1.27727121e-01 -3.77078578e-02
6.09640658e-01 -7.82396793e-01 -6.50050282e-01 5.26805758e-01
-9.44300413e-01 2.23321281e-02 3.30629259e-01 7.38813877e-01
5.97960234e-01 2.79291511e-01 2.10536003e-01 -8.21726859e-01
1.12911904e+00 -3.22007418e-01 -3.40419620e-01 4.76130545e-02
-1.05575597e+00 3.26623738e-01 5.35623550e-01 -5.69939435e-01
-6.43427134e-01 -7.35293925e-02 4.36570317e-01 -5.19914031e-01
-2.37360716e-01 6.08726382e-01 -1.99212000e-01 -1.34792194e-01
1.01436901e+00 1.38984236e-03 5.02237737e-01 -9.13001239e-01
4.57562685e-01 6.54773831e-01 1.63003132e-02 -1.16973430e-01
8.52405012e-01 5.50360084e-01 1.30741194e-01 -8.59256446e-01
-3.19033973e-02 -2.64867008e-01 -1.10854757e+00 -6.83275610e-02
5.06783366e-01 -2.01341614e-01 -4.18423146e-01 4.90944320e-03
-5.59197128e-01 6.62287325e-03 -6.98640764e-01 4.60211933e-01
-2.74700612e-01 5.36181629e-01 1.01959899e-01 -8.12255621e-01
4.69899923e-02 -1.10082352e+00 4.93459493e-01 4.48648110e-02
-7.10327744e-01 -1.23090184e+00 1.53029963e-01 -2.10498676e-01
6.03308976e-01 6.25258446e-01 1.41097558e+00 -5.45079231e-01
-5.22688627e-02 -2.90881634e-01 -1.84064403e-01 8.55557770e-02
8.02113712e-01 4.07199293e-01 -8.06141436e-01 -3.90220910e-01
-2.24573657e-01 9.88326445e-02 6.14265680e-01 9.93248299e-02
1.14970911e+00 -4.14063364e-01 -4.99317825e-01 4.14014935e-01
1.04310918e+00 3.32503140e-01 4.93813664e-01 3.98341566e-01
8.24796975e-01 1.22874260e+00 6.37093246e-01 3.00909817e-01
5.58227114e-02 8.15222144e-01 3.00439298e-01 -4.24993396e-01
-1.05741188e-01 -2.30392888e-01 -1.56817764e-01 5.31805217e-01
-6.82311431e-02 1.35806769e-01 -1.12737334e+00 3.04182678e-01
-1.72384906e+00 -6.60815418e-01 -2.65923858e-01 2.49572015e+00
5.44969440e-01 1.56145453e-01 3.96934360e-01 7.82340765e-01
6.09544814e-01 2.03135133e-01 -5.00614047e-01 -5.11192739e-01
-5.19525856e-02 -2.99625933e-01 9.97247174e-02 4.16199982e-01
-1.01512969e+00 4.67477322e-01 6.08963966e+00 4.84712809e-01
-1.27586377e+00 -2.56535321e-01 4.05634761e-01 -1.99548423e-01
-5.28943360e-01 1.50625125e-01 -1.38761729e-01 5.69346845e-01
5.66927314e-01 -2.28341863e-01 2.74164945e-01 7.61268735e-01
4.88032967e-01 -2.19831526e-01 -9.86360371e-01 1.08204913e+00
-2.70897269e-01 -1.07836592e+00 6.33539706e-02 4.56934392e-01
2.73700327e-01 -5.82610905e-01 1.60321623e-01 -3.11131805e-01
1.57763422e-01 -1.10536766e+00 3.99677604e-01 4.19222236e-01
9.26587701e-01 -8.86989176e-01 3.14612120e-01 7.06532001e-02
-8.58400464e-01 -1.21030048e-01 -1.98463634e-01 -1.69676483e-01
5.92252612e-02 6.21239603e-01 -5.92406929e-01 1.77909564e-02
6.83431983e-01 5.76072574e-01 -8.36728036e-01 1.02188480e+00
-5.95430732e-02 3.29191864e-01 -4.71310973e-01 -1.06150940e-01
-8.62637593e-04 -5.29974461e-01 6.99388325e-01 1.02751172e+00
2.49048889e-01 -2.32467130e-01 -1.30098805e-01 9.09148335e-01
3.97147089e-01 4.35576499e-01 -7.77871788e-01 -3.35278600e-01
5.52966595e-01 1.19737053e+00 -1.09795403e+00 -1.19820543e-01
-2.11653158e-01 5.95733404e-01 1.54309437e-01 3.46911222e-01
-1.73138425e-01 -6.93891644e-01 9.91212666e-01 5.34700751e-01
-9.76498201e-02 -6.48580611e-01 -7.16755509e-01 -8.40757549e-01
-5.30388532e-03 -7.60686278e-01 4.18332905e-01 -4.44329947e-01
-1.04157996e+00 4.56986755e-01 2.21100882e-01 -1.45355284e+00
-2.69371063e-01 -4.75856632e-01 -5.07319987e-01 9.32327926e-01
-1.01910210e+00 -6.03273988e-01 -4.81037915e-01 4.70489949e-01
1.08884513e-01 -9.62691605e-02 9.26181555e-01 2.09437698e-01
-6.40639067e-01 3.09338748e-01 3.71561825e-01 -2.37036645e-01
6.56709850e-01 -1.49576831e+00 3.44044983e-01 5.89509428e-01
2.06545338e-01 9.09787655e-01 8.78781974e-01 -6.32889450e-01
-1.09843171e+00 -5.57722449e-01 6.55147076e-01 -2.35228509e-01
5.83556056e-01 -6.71515942e-01 -1.12541497e+00 1.68996409e-01
-2.30172217e-01 -7.85132051e-02 1.00664079e+00 3.86830479e-01
-2.80340075e-01 -6.48125261e-02 -1.32842314e+00 8.00451756e-01
7.64906883e-01 -3.20799977e-01 -3.99404168e-01 1.33472010e-01
2.92208046e-01 9.99093130e-02 -8.11416924e-01 1.20521188e-01
5.04006505e-01 -1.03152835e+00 9.87959743e-01 -6.03441834e-01
1.89031109e-01 -5.32951474e-01 1.69727847e-01 -1.41630924e+00
-5.42513371e-01 -4.05298889e-01 -1.85672954e-01 1.28319848e+00
3.81261051e-01 -6.66816294e-01 8.13760817e-01 8.53106022e-01
3.76417309e-01 -5.76944828e-01 -7.83935308e-01 -6.93448603e-01
-8.06734934e-02 -1.02847628e-01 6.38055086e-01 1.25465178e+00
2.74888575e-01 8.94551128e-02 -1.74859077e-01 -1.91172138e-01
6.51188135e-01 1.08086632e-03 8.56948018e-01 -1.68941319e+00
3.18655133e-01 -8.27852070e-01 -7.68928230e-01 -3.62186044e-01
-4.88113552e-01 -6.63442016e-01 -4.78925973e-01 -1.53481781e+00
-1.19596303e-01 -8.56081069e-01 -1.70628682e-01 5.11446118e-01
-7.16293156e-02 2.98741281e-01 1.44540980e-01 7.64696956e-01
-1.89453363e-02 5.62884271e-01 1.07859230e+00 -3.24346847e-03
-7.24847019e-01 -3.69709790e-01 -7.96158135e-01 5.97415626e-01
8.62412691e-01 -6.03587329e-01 -6.52917266e-01 -6.35526925e-02
9.31795090e-02 -5.01253545e-01 2.40084022e-01 -7.66007543e-01
3.43799703e-02 -2.42556766e-01 4.65370178e-01 -3.00757706e-01
3.66596192e-01 -9.31815684e-01 4.47587550e-01 4.55445558e-01
-2.84881711e-01 2.96183228e-01 6.14820011e-02 4.76474106e-01
-4.05854732e-01 -9.68116000e-02 7.23666787e-01 9.76863429e-02
-5.56821704e-01 -2.81558335e-01 -1.78789079e-01 -4.25709099e-01
1.07315695e+00 -6.64640367e-01 -2.99875736e-01 -3.41282368e-01
-8.74440432e-01 1.25278741e-01 8.07230055e-01 4.60055977e-01
4.95259255e-01 -1.42226708e+00 -3.04511905e-01 3.39244574e-01
2.25518793e-01 -7.96496794e-02 1.62943184e-01 8.37460220e-01
-6.24891996e-01 2.01415196e-01 -4.21468318e-01 -6.55965090e-01
-1.49703002e+00 9.30921495e-01 1.27982683e-02 -5.72081767e-02
-8.19551587e-01 3.28054905e-01 2.36441448e-01 -3.35397393e-01
2.94535458e-01 -1.44707382e-01 -6.21173978e-01 6.99412346e-01
5.42229712e-01 7.37795174e-01 -1.98954754e-02 -5.91873586e-01
-4.40472364e-01 6.56962931e-01 -3.32018957e-02 -9.81713161e-02
1.30749619e+00 -2.93524593e-01 -1.40855134e-01 6.60045981e-01
1.16673303e+00 -7.90001079e-02 -1.18807900e+00 -1.15234211e-01
3.21988076e-01 -8.85209680e-01 1.67221338e-01 -6.96754038e-01
-9.62630272e-01 1.33565795e+00 8.96561682e-01 7.17695713e-01
1.13731194e+00 -2.20092446e-01 -7.77268112e-02 1.71437979e-01
3.53224464e-02 -1.04492879e+00 4.23008986e-02 -1.38551116e-01
9.74141121e-01 -1.21178985e+00 2.40836665e-01 -4.05335635e-01
-5.90375543e-01 1.25341594e+00 3.81107539e-01 1.26464158e-01
5.95272362e-01 1.88045233e-01 2.99316853e-01 -5.54997802e-01
-3.11379343e-01 -3.52566652e-02 3.47704142e-01 7.23727107e-01
4.44396377e-01 1.44127473e-01 -6.68636441e-01 -1.18478894e-01
-3.11464161e-01 -5.93848228e-01 2.49795318e-01 1.12676215e+00
-3.68411630e-01 -1.18303478e+00 -5.57861328e-01 5.04328132e-01
-1.04351409e-01 1.99801385e-01 -7.17760384e-01 9.62250650e-01
-1.38784930e-01 5.82398355e-01 2.05933303e-01 -3.86437207e-01
3.90828609e-01 1.39402121e-01 3.74511667e-02 -3.86631995e-01
-2.86097795e-01 1.45826578e-01 -1.55994311e-01 -3.70008409e-01
-2.39006266e-01 -7.91495860e-01 -9.60971177e-01 -3.27720970e-01
-2.93105602e-01 2.14111060e-01 9.87109482e-01 4.33677405e-01
5.01690567e-01 3.39048862e-01 5.86099565e-01 -9.11506593e-01
-1.50042340e-01 -8.96614194e-01 -6.40946507e-01 6.25234306e-01
3.52881312e-01 -1.02845693e+00 -4.81495440e-01 -9.06851217e-02]
|
[8.018515586853027, 4.5713934898376465]
|
e372be3b-22cb-427a-bf5f-5f5d54bcc8f0
|
sheetcopilot-bringing-software-productivity
|
2305.19308
| null |
https://arxiv.org/abs/2305.19308v1
|
https://arxiv.org/pdf/2305.19308v1.pdf
|
SheetCopilot: Bringing Software Productivity to the Next Level through Large Language Models
|
Computer end users have spent billions of hours completing daily tasks like tabular data processing and project timeline scheduling. Most of these tasks are repetitive and error-prone, yet most end users lack the skill of automating away these burdensome works. With the advent of large language models (LLMs), directing software with natural language user requests become a reachable goal. In this work, we propose a SheetCopilot agent which takes natural language task and control spreadsheet to fulfill the requirements. We propose a set of atomic actions as an abstraction of spreadsheet software functionalities. We further design a state machine-based task planning framework for LLMs to robustly interact with spreadsheets. We curate a representative dataset containing 221 spreadsheet control tasks and establish a fully automated evaluation pipeline for rigorously benchmarking the ability of LLMs in software control tasks. Our SheetCopilot correctly completes 44.3\% of tasks for a single generation, outperforming the strong code generation baseline by a wide margin. Our project page:https://sheetcopilot-demo.github.io/.
|
['Zhaoxiang Zhang', 'Qing Li', 'Yuntao Chen', 'Jingran Su', 'Hongxin Li']
|
2023-05-30
| null | null | null | null |
['code-generation']
|
['computer-code']
|
[-4.38157581e-02 1.45016178e-01 -1.99539632e-01 -5.29034197e-01
-8.11629653e-01 -9.79896665e-01 7.20412791e-01 7.97520727e-02
4.98344796e-03 3.22297484e-01 2.16310918e-01 -4.27628040e-01
9.62662250e-02 -3.76991481e-01 -6.32824421e-01 3.05765718e-01
1.70939192e-01 6.38284862e-01 5.55328839e-02 -3.66550803e-01
6.03023767e-01 1.22321539e-01 -1.32346570e+00 5.69928885e-01
1.15351081e+00 3.29140931e-01 6.57457590e-01 8.19063723e-01
-1.95548803e-01 1.05133641e+00 -8.43068004e-01 -2.82337546e-01
1.63848922e-01 7.85855651e-02 -9.97049630e-01 2.20282838e-01
2.84834743e-01 -1.28202856e-01 7.43504241e-02 6.84106231e-01
3.76434445e-01 -1.11286081e-01 3.80624324e-01 -1.71931732e+00
-4.38828319e-01 8.19877028e-01 -4.97377276e-01 -1.43970549e-01
8.23115587e-01 5.51397383e-01 1.02753830e+00 -6.32699192e-01
8.10651422e-01 8.91566753e-01 4.31314886e-01 5.45658231e-01
-1.22452557e+00 -4.91863281e-01 7.74942115e-02 -5.25226831e-01
-1.23759604e+00 -6.61354244e-01 1.56960577e-01 -1.15456927e+00
1.58818722e+00 3.13052624e-01 2.72778720e-01 7.87412286e-01
6.37202501e-01 9.35896754e-01 5.60411513e-01 -3.71701837e-01
8.40681195e-02 2.51020100e-02 3.50390106e-01 9.33493435e-01
1.03636578e-01 -4.38850611e-01 -6.78465605e-01 -5.19249201e-01
5.65053523e-01 -1.76699515e-02 -2.37212166e-01 -2.90571421e-01
-1.56260300e+00 4.11537647e-01 -2.86393017e-01 3.73318017e-01
-2.12007806e-01 6.23674750e-01 4.51727271e-01 3.82641613e-01
1.82660148e-01 1.14203775e+00 -7.35298216e-01 -7.27188051e-01
-9.18380380e-01 7.01781273e-01 1.39469838e+00 1.77712405e+00
6.13875687e-01 -1.92122087e-01 -4.96493459e-01 5.99590480e-01
4.14359309e-02 3.70367497e-01 2.52476245e-01 -1.12515104e+00
9.05124962e-01 8.37881446e-01 5.54219961e-01 -5.71807325e-01
-4.50767457e-01 -3.49806815e-01 -3.51560920e-01 1.95686579e-01
4.26626772e-01 -3.48582923e-01 -5.59312344e-01 1.16904664e+00
-1.17106192e-01 -6.92135096e-01 -4.74703878e-01 6.16872966e-01
1.98868707e-01 6.67356253e-01 -4.61128987e-02 -3.28560323e-02
1.38704848e+00 -1.10349071e+00 -6.12110198e-01 -5.41571677e-01
9.16566074e-01 -1.07156050e+00 1.63573825e+00 8.93403649e-01
-1.33274031e+00 -3.87482285e-01 -8.33133221e-01 -1.46745861e-01
-2.27862950e-02 4.30996984e-01 9.45381761e-01 3.15362662e-01
-1.05076110e+00 5.17314196e-01 -1.09786916e+00 -1.76021904e-01
1.72218308e-01 8.67586285e-02 -1.54370517e-01 2.25091763e-02
-4.80415970e-01 8.33276212e-01 1.41434804e-01 -3.19221258e-01
-9.40933645e-01 -1.17102253e+00 -7.86713064e-01 3.29355866e-01
7.90891171e-01 -6.49540126e-01 2.16636920e+00 -1.56307787e-01
-1.19104016e+00 6.47464573e-01 9.27844569e-02 1.27489060e-01
5.82640767e-01 -5.24682164e-01 1.74488202e-01 -4.95342821e-01
4.71770108e-01 4.50660408e-01 6.22117460e-01 -7.82574177e-01
-7.02385604e-01 -7.13618547e-02 1.76259935e-01 -1.73700079e-01
-1.35151967e-01 4.39855754e-01 -7.06896305e-01 -5.34444511e-01
-4.78061557e-01 -9.79658425e-01 -1.84513286e-01 -2.31548339e-01
-6.96104228e-01 -6.21062577e-01 -6.14922233e-02 -6.64730728e-01
1.74545956e+00 -1.84928060e+00 3.46651524e-01 -1.78158477e-01
6.30525649e-01 -2.66472578e-01 -8.08579698e-02 8.73074830e-01
-1.93411913e-02 3.35242152e-01 -1.32498592e-01 -2.33820900e-01
7.44534016e-01 -5.00641823e-01 -1.98893666e-01 9.16958898e-02
5.06503060e-02 9.70743179e-01 -8.69548500e-01 -3.72814059e-01
-2.10122243e-01 -1.62683040e-01 -8.63848746e-01 4.29698080e-01
-1.08150005e+00 -1.98826045e-02 -5.81980705e-01 7.51370609e-01
2.41104916e-01 -7.13186502e-01 2.11050771e-02 4.01147991e-01
-4.58312452e-01 4.78360385e-01 -8.13575268e-01 2.34084558e+00
-9.79820490e-01 5.49392700e-01 1.00096799e-01 -1.04314089e-02
7.59488106e-01 2.17949539e-01 1.84223935e-01 -3.86583209e-01
-3.43781233e-01 3.96200269e-02 -2.05726624e-01 -6.59978867e-01
5.16948521e-01 3.93956810e-01 -7.63639033e-01 1.05026698e+00
-2.58130372e-01 -7.98834622e-01 5.68853140e-01 5.13394237e-01
1.66096187e+00 3.92887801e-01 4.93695766e-01 -2.52370447e-01
7.31088743e-02 4.16306615e-01 2.90680170e-01 7.00117648e-01
1.98814362e-01 4.40484613e-01 1.04972553e+00 -4.95350569e-01
-1.07180655e+00 -5.96276045e-01 3.56566191e-01 1.80500901e+00
-5.66026926e-01 -1.06799424e+00 -9.26382959e-01 -7.26570725e-01
-1.34521611e-02 1.34867156e+00 -9.65044498e-02 1.32601455e-01
-4.40450102e-01 -2.48689160e-01 5.47510922e-01 3.78168851e-01
2.83260047e-01 -1.20736349e+00 -7.68172026e-01 2.42984682e-01
-1.97065264e-01 -9.60394084e-01 -1.32325888e+00 1.69122331e-02
-2.08448693e-01 -1.01694775e+00 -4.65877235e-01 -3.75904113e-01
6.15621567e-01 2.50787199e-01 1.71075726e+00 4.94329304e-01
-5.41358769e-01 3.67139250e-01 -2.68537760e-01 -5.39446235e-01
-7.41140664e-01 5.28980851e-01 -3.26548010e-01 -7.30995238e-01
4.10295516e-01 -4.03336704e-01 -2.70069689e-01 3.11110675e-01
-7.12345421e-01 6.41540527e-01 5.73260427e-01 3.91452968e-01
-6.29461557e-02 -1.47087410e-01 4.57629800e-01 -1.19598818e+00
1.10692310e+00 -5.22283792e-01 -1.09419703e+00 4.72045928e-01
-6.89375103e-01 1.77926347e-01 5.80793679e-01 -4.58757058e-02
-1.11625886e+00 2.53912807e-01 2.97282755e-01 7.87572712e-02
-1.95571750e-01 6.09773040e-01 1.23006359e-01 5.29734910e-01
1.04502475e+00 2.83934832e-01 -1.01515047e-01 -4.92070943e-01
1.59673244e-01 7.75611699e-01 1.83136225e-01 -9.91919637e-01
7.85936594e-01 -1.10957928e-01 -7.22501934e-01 -3.23652774e-01
-5.72510421e-01 -4.04962212e-01 -2.69620091e-01 -2.02468615e-02
6.91407979e-01 -5.77549279e-01 -1.18061817e+00 2.65877575e-01
-1.66175878e+00 -1.01969540e+00 3.28869253e-01 -1.06907777e-01
-8.34681690e-01 1.10936671e-01 -5.43074906e-01 -6.14878833e-01
-4.54778939e-01 -1.46406007e+00 1.43455553e+00 -2.01396585e-01
-9.14213479e-01 -5.69941223e-01 1.52496681e-01 5.90176165e-01
6.12317681e-01 3.56941693e-03 1.10154498e+00 -5.64179659e-01
-8.98155153e-01 -3.56857300e-01 -1.22014284e-01 -2.14149542e-02
-2.89333407e-02 2.81388581e-01 -5.19425929e-01 -2.78702557e-01
-3.78202498e-01 -4.51494366e-01 1.99327439e-01 6.61367401e-02
1.13566780e+00 -3.17806751e-01 -4.73549306e-01 3.87372553e-01
9.88979578e-01 1.11670367e-01 3.04789305e-01 3.53660703e-01
5.35182059e-01 6.07818067e-01 8.82383585e-01 8.41596007e-01
5.23254812e-01 6.64786935e-01 1.50885612e-01 4.21929270e-01
2.32266337e-01 -1.47775888e-01 4.93993640e-01 6.63723052e-01
1.71744116e-02 -3.98962528e-01 -1.64050579e+00 5.62363625e-01
-1.86048687e+00 -5.78236163e-01 -3.06739844e-02 2.00494313e+00
1.14970970e+00 4.21623349e-01 -7.80631006e-02 -4.91745055e-01
4.13474768e-01 -1.95020840e-01 -4.82312799e-01 -2.17855304e-01
9.26563978e-01 -1.93486005e-01 2.05855876e-01 3.63756925e-01
-7.55302787e-01 9.48516607e-01 5.46043444e+00 7.27571309e-01
-6.76899433e-01 -5.40904589e-02 2.89282888e-01 -2.67773509e-01
-3.22414041e-01 -3.41091380e-02 -8.60047162e-01 6.47565365e-01
8.52746964e-01 -6.78280056e-01 8.41123521e-01 1.27841401e+00
5.15869021e-01 -1.75458580e-01 -1.77654457e+00 8.51592958e-01
-2.49594197e-01 -1.47701418e+00 -3.16014558e-01 -6.83220401e-02
6.80820763e-01 5.05345268e-03 -1.68483347e-01 6.29902184e-01
8.01749349e-01 -1.20750511e+00 9.02506053e-01 7.59196818e-01
9.47308838e-01 -4.11289692e-01 2.97000110e-01 8.82360816e-01
-7.77644813e-01 2.19648048e-01 -1.37672322e-02 -2.27119699e-01
1.06681742e-01 5.09237230e-01 -1.19059634e+00 1.03065357e-01
4.83667612e-01 4.44892347e-01 -6.58852816e-01 8.67601454e-01
-1.96282133e-01 2.97724634e-01 -3.48868314e-03 -1.52597278e-01
-1.36820599e-01 2.32807681e-01 3.65980476e-01 1.43436408e+00
2.37693742e-01 1.27244461e-03 4.22927290e-01 1.24083364e+00
-1.22556843e-01 -1.94678694e-01 -5.95994890e-01 -6.04348004e-01
7.13527381e-01 1.39245331e+00 -4.74481821e-01 -3.55315536e-01
-4.68681604e-01 8.70115876e-01 2.48736888e-01 3.13529164e-01
-7.52314806e-01 -9.22124207e-01 6.67285740e-01 3.62095833e-01
-2.52325982e-01 -6.75022006e-01 -4.61136848e-01 -1.27041125e+00
2.75126368e-01 -1.42866194e+00 -1.13314018e-02 -1.16708243e+00
-1.00515676e+00 5.65830886e-01 8.49091187e-02 -1.06395626e+00
-5.48644960e-01 -2.60109484e-01 -6.56734169e-01 8.20218444e-01
-1.01554942e+00 -8.71609509e-01 -6.42764091e-01 1.67021230e-01
1.07863200e+00 -3.49705368e-01 6.03541911e-01 9.23533067e-02
-6.06372714e-01 3.82032067e-01 -5.92633247e-01 -1.68496996e-01
9.17613268e-01 -1.54703534e+00 1.11993861e+00 6.51393712e-01
-3.57206315e-01 1.30094099e+00 8.74738693e-01 -9.14547205e-01
-1.79827964e+00 -8.13066959e-01 1.02500129e+00 -9.27701771e-01
9.31466103e-01 -9.58177269e-01 -6.36808157e-01 1.02794480e+00
4.47908282e-01 -5.72708845e-01 4.05353367e-01 1.15892403e-01
-2.85833985e-01 2.83230126e-01 -5.93418479e-01 6.72855258e-01
1.09956586e+00 -3.61552924e-01 -5.16132116e-01 1.08833492e+00
1.17009091e+00 -5.89063287e-01 -8.93121064e-01 -4.01317805e-01
2.62847573e-01 -7.38952398e-01 3.85413587e-01 -6.02017283e-01
9.81874347e-01 -2.38242745e-01 2.50952601e-01 -1.46791697e+00
-2.55146831e-01 -1.48051894e+00 1.97743431e-01 1.08404446e+00
7.00099766e-01 -3.22325557e-01 6.52306139e-01 1.23971713e+00
-5.34837186e-01 -4.54570562e-01 -1.01237282e-01 -6.41199946e-01
-2.91638941e-01 -7.74084449e-01 7.20581830e-01 8.60117793e-01
7.94878364e-01 5.18500328e-01 -8.74685571e-02 -1.74169019e-01
3.92581820e-01 2.35018833e-03 1.16354251e+00 -1.10521936e+00
-6.67667091e-01 -4.42112386e-01 3.20949376e-01 -1.21105123e+00
1.33794427e-01 -1.20906532e+00 2.28773892e-01 -1.74156117e+00
2.68146664e-01 -4.16742355e-01 6.14180326e-01 9.02210534e-01
1.18537389e-01 -5.67315280e-01 4.48571980e-01 3.06696743e-01
-1.07294011e+00 1.87328085e-01 1.09059834e+00 -9.45887193e-02
-4.35580164e-01 3.40494871e-01 -9.03497219e-01 7.47812450e-01
7.49832690e-01 -4.10044700e-01 -3.29382569e-01 -7.84869492e-01
1.10200572e+00 7.11643934e-01 2.70739403e-02 -8.18355978e-01
6.52011395e-01 -5.00741601e-01 -1.94782600e-01 -2.51198322e-01
-1.85396165e-01 -6.80668950e-01 5.51078394e-02 3.18421543e-01
-7.04980612e-01 3.59593123e-01 1.57061026e-01 1.52907237e-01
1.51859686e-01 -3.80698442e-01 1.56012997e-01 -1.78922102e-01
-4.36868608e-01 7.55522251e-02 -7.83314586e-01 2.86335021e-01
1.16521430e+00 8.28488469e-02 -9.05425787e-01 -2.16855720e-01
-5.10307133e-01 6.63729608e-01 6.09315038e-01 5.92983365e-01
3.40273649e-01 -7.34211147e-01 -6.46912694e-01 5.71876811e-03
3.56776565e-01 2.16461360e-01 -1.84914574e-01 7.76513815e-01
-9.22346473e-01 8.39710951e-01 -1.26723111e-01 -2.30667219e-01
-9.22307670e-01 4.25541222e-01 1.04712561e-01 -4.50185567e-01
-3.92145902e-01 8.05286884e-01 2.66411394e-01 -6.29576087e-01
2.44767532e-01 -7.05777109e-01 2.26659730e-01 -1.52337879e-01
5.66908360e-01 4.95248139e-01 3.23253930e-01 5.36134660e-01
-2.85147220e-01 -3.42108367e-04 -4.85503465e-01 -3.73596996e-02
1.55180526e+00 2.52845198e-01 -4.83891368e-01 2.65670896e-01
6.41855419e-01 6.34828433e-02 -1.20336413e+00 7.93910995e-02
5.60838580e-01 -4.59495455e-01 -1.76073551e-01 -1.35087311e+00
-5.81648350e-01 5.29225826e-01 -4.96127188e-01 4.46046442e-01
7.40771592e-01 8.70651752e-02 4.87017632e-01 8.07568967e-01
8.76850963e-01 -9.95709896e-01 3.93144786e-01 5.81914306e-01
1.40138602e+00 -1.11011112e+00 -3.15168761e-02 -5.29186010e-01
-8.71112287e-01 1.14364779e+00 1.03260171e+00 1.58003375e-01
1.68796733e-01 7.69179344e-01 -3.17008376e-01 -6.27156854e-01
-1.57259738e+00 4.49364811e-01 -4.76396270e-02 3.43926877e-01
9.65965092e-01 2.40634382e-02 -1.86383173e-01 9.85168457e-01
-4.38890815e-01 4.15512741e-01 1.14487004e+00 1.29229927e+00
-4.37285960e-01 -8.94087136e-01 -2.65645355e-01 6.92157626e-01
-3.84134442e-01 -3.51108521e-01 -2.99248427e-01 7.17088699e-01
-4.88514662e-01 7.41916776e-01 -7.13139400e-02 -2.76551872e-01
5.48362315e-01 1.81132540e-01 1.09703772e-01 -1.43407917e+00
-9.50482070e-01 -1.90134391e-01 4.50632453e-01 -8.75922322e-01
4.96981680e-01 -5.76653659e-01 -1.28534126e+00 -4.97859776e-01
-2.40886547e-02 -1.07156588e-02 7.07469881e-01 6.00630999e-01
7.09389567e-01 7.37504005e-01 1.88616067e-01 -8.84070992e-01
-9.68583167e-01 -1.02850366e+00 -1.85286209e-01 7.78427571e-02
9.27020162e-02 -3.31527203e-01 -2.36233789e-02 4.52618867e-01]
|
[7.975882053375244, 7.706299304962158]
|
4d0daca6-aa8b-43d8-86ed-bdd7edbeb9a0
|
learning-to-detect-semantic-boundaries-with
|
2212.07579
| null |
https://arxiv.org/abs/2212.07579v1
|
https://arxiv.org/pdf/2212.07579v1.pdf
|
Learning to Detect Semantic Boundaries with Image-level Class Labels
|
This paper presents the first attempt to learn semantic boundary detection using image-level class labels as supervision. Our method starts by estimating coarse areas of object classes through attentions drawn by an image classification network. Since boundaries will locate somewhere between such areas of different classes, our task is formulated as a multiple instance learning (MIL) problem, where pixels on a line segment connecting areas of two different classes are regarded as a bag of boundary candidates. Moreover, we design a new neural network architecture that can learn to estimate semantic boundaries reliably even with uncertain supervision given by the MIL strategy. Our network is used to generate pseudo semantic boundary labels of training images, which are in turn used to train fully supervised models. The final model trained with our pseudo labels achieves an outstanding performance on the SBD dataset, where it is as competitive as some of previous arts trained with stronger supervision.
|
['Suha Kwak', 'Sehyun Hwang', 'Namyup Kim']
|
2022-12-15
| null | null | null | null |
['boundary-detection', 'multiple-instance-learning']
|
['computer-vision', 'methodology']
|
[ 6.90861046e-01 7.56835759e-01 -6.86904132e-01 -5.27166426e-01
-1.03388262e+00 -1.90464944e-01 5.67645252e-01 -4.34828140e-02
-1.78716376e-01 7.34853923e-01 -1.93056464e-01 -4.73118797e-02
2.45833978e-01 -8.24774027e-01 -1.16615522e+00 -3.59266400e-01
2.56795973e-01 9.26571369e-01 4.68188792e-01 2.32115775e-01
2.66508192e-01 1.90468565e-01 -1.36839247e+00 6.00966275e-01
8.70382726e-01 1.50660622e+00 1.92822024e-01 2.76856244e-01
-4.11199987e-01 7.90856361e-01 -5.00491440e-01 -7.30486363e-02
2.99294561e-01 -3.32493663e-01 -1.31509686e+00 4.60498035e-01
9.80952322e-01 -1.79042339e-01 6.01869896e-02 1.11827135e+00
-2.18405828e-01 -2.55485982e-01 1.01723826e+00 -1.31336308e+00
-7.25313544e-01 5.55557668e-01 -6.68853939e-01 9.77564156e-02
5.68962693e-02 -4.58520919e-01 1.26846385e+00 -1.08413291e+00
6.15661442e-01 1.00443625e+00 7.47981369e-01 6.89393997e-01
-1.34712303e+00 -4.04450983e-01 4.96108741e-01 2.31211171e-01
-1.51483631e+00 -1.76926121e-01 9.39335108e-01 -6.66597426e-01
4.72600490e-01 -1.73163056e-01 2.67458349e-01 7.47108042e-01
-2.48430952e-01 1.13517797e+00 9.22603548e-01 -5.12496293e-01
1.66589126e-01 2.52084970e-01 2.30795875e-01 7.75581300e-01
2.70525843e-01 -2.81770259e-01 -3.42912555e-01 2.77528554e-01
9.93955493e-01 -1.45155087e-01 -3.46543938e-01 -6.06271088e-01
-1.11760843e+00 7.35180140e-01 9.07102823e-01 2.88915604e-01
2.82838922e-02 3.06659728e-01 -5.55345938e-02 -7.61452410e-03
7.46295273e-01 2.95381248e-01 -4.24753398e-01 6.17276490e-01
-1.05648386e+00 -1.86973900e-01 5.78208983e-01 1.07663000e+00
1.08378124e+00 -3.94105762e-01 4.59922031e-02 8.67734969e-01
2.90299982e-01 2.25049689e-01 4.95586507e-02 -1.07054770e+00
6.31636679e-01 8.49675298e-01 1.90458193e-01 -7.35795557e-01
-3.41944784e-01 -3.67541224e-01 -6.25875890e-01 5.14901161e-01
7.98968673e-01 8.88349712e-02 -1.20989251e+00 1.54270566e+00
3.44420940e-01 5.49807549e-01 2.53776852e-02 8.53785217e-01
9.19913828e-01 5.20300090e-01 7.64202476e-02 2.20227167e-01
1.26612818e+00 -1.40292871e+00 -5.14548182e-01 -6.67340279e-01
5.79819381e-01 -5.13377786e-01 8.20072949e-01 1.94785178e-01
-1.05165517e+00 -7.87973881e-01 -1.12714219e+00 -1.03832133e-01
-3.88890684e-01 3.55619550e-01 5.23962915e-01 6.87467754e-02
-1.16926730e+00 5.10731816e-01 -5.34761250e-01 3.35671939e-02
1.11831594e+00 3.15217853e-01 -2.42810220e-01 7.87379369e-02
-9.82900083e-01 7.71380782e-01 5.08754015e-01 4.76018973e-02
-1.13198531e+00 -5.30078590e-01 -1.19338512e+00 -1.26919851e-01
4.72011566e-01 -5.19757569e-01 1.05314124e+00 -1.44919491e+00
-8.56513739e-01 1.60124397e+00 -6.07795715e-02 -5.36496282e-01
6.09368622e-01 -1.51145041e-01 -1.00554720e-01 3.50077152e-01
5.49276412e-01 1.40423238e+00 1.01532757e+00 -1.76316583e+00
-9.45849776e-01 -1.99311838e-01 -3.21935713e-02 2.12058797e-01
4.57582586e-02 -4.66964006e-01 -4.24062103e-01 -5.08831978e-01
6.00659192e-01 -7.11541772e-01 -8.49300623e-02 2.08599135e-01
-8.12382638e-01 -6.69536173e-01 7.99580455e-01 -6.53385162e-01
8.10764730e-01 -1.89428484e+00 4.08481993e-02 1.41369700e-01
1.99232697e-01 6.18816018e-02 1.00446986e-02 -2.51477391e-01
-7.83476681e-02 2.24566951e-01 -5.77004254e-01 -2.99223781e-01
-4.00032289e-02 -7.24128820e-03 -3.25758606e-01 3.30968142e-01
6.19417071e-01 8.61160100e-01 -7.33085990e-01 -7.28942275e-01
-2.75805276e-02 1.15861498e-01 -2.18048707e-01 2.74317592e-01
-5.06384730e-01 3.76385391e-01 -4.61581618e-01 7.00405478e-01
6.61642849e-01 -6.50751114e-01 -1.77636772e-01 -2.54083186e-01
2.59935826e-01 1.93032801e-01 -9.69979882e-01 1.73265088e+00
-3.74565959e-01 7.17761874e-01 -5.09511307e-02 -1.43662512e+00
1.14078772e+00 2.10348219e-01 3.51230800e-01 -5.61641276e-01
-3.61866574e-03 2.21605361e-01 -4.77674603e-01 -4.38038349e-01
-1.35461703e-01 -2.39055231e-02 -2.49543171e-02 3.20609182e-01
7.26779103e-02 -4.40686136e-01 -6.44913688e-02 -5.80469891e-02
7.25281000e-01 5.63936412e-01 -1.43000752e-01 -3.48694891e-01
7.74599254e-01 2.76762456e-01 6.14992321e-01 7.42003798e-01
-3.97460938e-01 9.28792298e-01 6.10571742e-01 -6.94826484e-01
-1.27272630e+00 -1.23269022e+00 -6.94718778e-01 6.77484810e-01
8.26351643e-01 2.16338411e-01 -1.00805104e+00 -1.19822013e+00
5.31886378e-03 3.67675245e-01 -8.22961569e-01 3.01854424e-02
-4.17146564e-01 -2.18466371e-01 2.00635761e-01 8.82126629e-01
7.97312319e-01 -1.04498708e+00 -1.59620181e-01 2.79362444e-02
-2.68655330e-01 -1.32667851e+00 -4.84911263e-01 2.70083755e-01
-9.32481289e-01 -1.31397235e+00 -8.54847252e-01 -1.67419255e+00
1.12607348e+00 1.19165517e-01 1.20262265e+00 1.29860193e-01
-2.40968198e-01 1.72173858e-01 -6.59399554e-02 -1.69781923e-01
-2.79267281e-01 4.81547005e-02 -2.14719832e-01 1.69540584e-01
3.46006513e-01 -2.63003528e-01 -6.64785206e-01 5.56315541e-01
-6.21201277e-01 4.40257549e-01 5.12818038e-01 9.89791512e-01
9.03364360e-01 -6.99522644e-02 8.26062441e-01 -9.80104089e-01
-7.34950751e-02 -3.98339033e-01 -6.10756516e-01 4.20588136e-01
-4.29767966e-01 4.36476469e-02 3.97647381e-01 -2.01523840e-01
-1.24875844e+00 2.62863994e-01 6.80548847e-02 -2.54048973e-01
-5.51008582e-01 -2.74192933e-02 -1.64278299e-01 -3.74128558e-02
4.99869347e-01 -2.40691125e-01 -7.01422393e-02 -4.62869674e-01
3.01110119e-01 7.96744347e-01 6.16350114e-01 -5.01849234e-01
6.32758796e-01 8.22324634e-01 -1.75516278e-01 -5.12959182e-01
-1.78726542e+00 -5.51898718e-01 -1.02944791e+00 -2.33325079e-01
1.21231127e+00 -1.02741528e+00 -2.56515980e-01 6.05286062e-01
-1.11658895e+00 -8.45881820e-01 -1.89159989e-01 2.44652897e-01
-8.68935347e-01 -2.50333641e-02 -7.52058506e-01 -3.99675876e-01
-3.51540297e-02 -9.72829998e-01 1.44955850e+00 4.72008288e-01
-1.51755512e-02 -1.29362261e+00 -2.93440491e-01 1.05739105e+00
-5.68964630e-02 2.50314355e-01 8.70993078e-01 -6.94498181e-01
-8.46600413e-01 -2.28204839e-02 -6.46880150e-01 5.35323918e-01
1.40622541e-01 -3.37499171e-01 -1.12223005e+00 -6.77298084e-02
-3.06606412e-01 -6.16159439e-01 1.12173545e+00 5.77949643e-01
1.41653693e+00 -4.10562009e-02 -7.71250069e-01 5.90131223e-01
1.42793906e+00 5.65523934e-03 5.23755133e-01 4.45224017e-01
7.12896287e-01 6.51638150e-01 6.90581560e-01 -5.27292816e-03
2.33665466e-01 5.82911611e-01 5.95259488e-01 -5.64119339e-01
-5.53076267e-01 -3.02766502e-01 -3.62455659e-02 1.23891011e-01
3.35783303e-01 -6.86667413e-02 -1.01524603e+00 7.09170282e-01
-1.79874766e+00 -6.20354235e-01 -1.75711378e-01 2.11228848e+00
7.97980845e-01 6.18593216e-01 -6.53218925e-02 -2.90606730e-02
1.22721601e+00 -3.74863967e-02 -7.55151093e-01 6.48845080e-03
-5.67421950e-02 6.53960928e-02 3.20518762e-01 6.09759152e-01
-1.65991259e+00 1.19502079e+00 6.49573946e+00 7.59675980e-01
-6.47532344e-01 -1.13097867e-02 1.41209829e+00 5.18825591e-01
-1.72527000e-01 4.18415666e-02 -1.01217973e+00 2.84624815e-01
2.54767090e-01 3.50292265e-01 -1.51369736e-01 8.68116558e-01
-1.99644893e-01 -3.40561658e-01 -1.32266021e+00 8.16121519e-01
3.24232727e-01 -1.45206285e+00 -2.45358869e-02 -1.57544360e-01
1.21894300e+00 -2.53661745e-03 5.50467819e-02 -9.86477640e-03
3.29525203e-01 -1.24162066e+00 7.06118762e-01 4.78354096e-01
9.20366108e-01 -3.33394945e-01 8.18727016e-01 3.18219632e-01
-1.02138591e+00 8.57778341e-02 -4.71818298e-01 -1.63717091e-01
7.14477152e-02 5.92175007e-01 -1.02856660e+00 1.26856536e-01
5.52898526e-01 8.67009401e-01 -3.54408532e-01 1.11439824e+00
-6.84578657e-01 5.88502705e-01 -1.87800199e-01 3.84776682e-01
3.97027463e-01 -3.03645343e-01 4.95656766e-02 6.51649237e-01
-3.68714817e-02 -1.23710588e-01 3.94907683e-01 1.26864302e+00
-3.71431738e-01 -1.20887473e-01 -7.34193563e-01 4.23569113e-01
2.83942044e-01 1.12012792e+00 -1.25618279e+00 -5.09523213e-01
-4.47131366e-01 9.95156109e-01 6.77147686e-01 3.80832762e-01
-8.72933030e-01 -3.10539275e-01 3.64137977e-01 1.59477219e-01
2.93402791e-01 9.46102813e-02 -6.46151245e-01 -9.09520268e-01
-4.61047068e-02 -1.02517009e-01 4.44057435e-01 -1.01094842e+00
-1.37221885e+00 4.32518721e-01 -2.89719582e-01 -1.18664861e+00
3.35075885e-01 -6.97110116e-01 -6.87901080e-01 6.26938224e-01
-1.84406412e+00 -1.15554559e+00 -4.06515449e-01 2.65416682e-01
7.89864719e-01 1.92735150e-01 4.91674870e-01 1.77283272e-01
-3.44054312e-01 4.17290211e-01 -1.08796759e-02 6.26214743e-01
4.39386219e-01 -1.36517644e+00 2.47186840e-01 5.93041480e-01
4.00729567e-01 -1.53303146e-01 2.14964837e-01 -7.04874337e-01
-2.39336804e-01 -1.30975723e+00 9.08154905e-01 -5.71013689e-01
5.12880802e-01 -3.54135156e-01 -9.52411234e-01 7.74100661e-01
8.32422264e-03 4.17371362e-01 3.38588357e-01 -9.65012889e-03
-2.37524912e-01 -1.07174993e-01 -9.49620903e-01 2.38577262e-01
1.22850156e+00 -4.61328268e-01 -8.94244134e-01 5.84704638e-01
7.03258634e-01 -4.30566192e-01 -8.12038541e-01 5.81957579e-01
-4.38289270e-02 -7.47425318e-01 1.09738231e+00 -6.00880444e-01
6.44955933e-01 -3.99933815e-01 1.91128582e-01 -1.06688035e+00
6.88427612e-02 2.30452255e-03 2.53693163e-01 1.35946178e+00
8.36461186e-01 -3.14484268e-01 1.20036900e+00 4.39634919e-01
-2.84960181e-01 -8.11945319e-01 -8.72956038e-01 -6.69002771e-01
1.63358167e-01 -2.18155384e-01 3.54510963e-01 9.80276167e-01
-6.26377761e-02 4.72301185e-01 -4.76741455e-02 2.10162386e-01
8.93190861e-01 5.32549560e-01 3.23240817e-01 -1.38120008e+00
2.18828306e-01 -4.19646084e-01 -5.53649306e-01 -1.44334042e+00
5.59321940e-01 -1.14337003e+00 4.68361616e-01 -1.83997428e+00
3.10763717e-01 -1.00541294e+00 -2.94077605e-01 7.81232834e-01
-1.34048074e-01 7.49761820e-01 -1.57540008e-01 2.91391969e-01
-9.12756026e-01 4.49489415e-01 1.52265930e+00 -5.53745508e-01
-5.67237362e-02 3.06364536e-01 -4.40526932e-01 1.12932730e+00
6.36759639e-01 -4.24869508e-01 -2.75195062e-01 -3.80731523e-01
3.12648970e-03 4.81126504e-03 4.14495081e-01 -1.11469507e+00
1.22819088e-01 -3.31648737e-02 5.53375602e-01 -6.43518984e-01
1.84229746e-01 -8.91921103e-01 -2.64961034e-01 2.80796260e-01
-6.54056609e-01 -7.37691164e-01 -8.77114981e-02 5.50459087e-01
-4.26079333e-01 -4.84154791e-01 8.31267416e-01 -1.38514534e-01
-1.07516444e+00 3.78445894e-01 -1.52672958e-02 4.56685483e-01
1.39366901e+00 -4.00461584e-01 -3.40191215e-01 -1.29431650e-01
-1.14874911e+00 5.74090600e-01 4.82436746e-01 3.69461596e-01
5.91083288e-01 -1.35472369e+00 -5.16879976e-01 1.76489249e-01
2.24401489e-01 6.63438082e-01 -1.73379220e-02 4.75133389e-01
-5.29109716e-01 4.22557622e-01 -6.75197393e-02 -1.04214120e+00
-8.44827771e-01 5.41330695e-01 5.98015428e-01 1.51986908e-03
-7.42709339e-01 1.25981450e+00 5.79424202e-01 -2.34375745e-01
4.44393724e-01 -1.99713081e-01 -3.18409443e-01 4.09690067e-02
2.51932293e-01 -1.18190676e-01 -5.95166646e-02 -7.69803464e-01
-3.23023826e-01 9.08506870e-01 -4.09672521e-02 2.63307422e-01
1.18508196e+00 -1.29633933e-01 9.28486362e-02 5.11770070e-01
1.18432033e+00 -5.08587003e-01 -2.09308696e+00 -4.80955005e-01
3.79379898e-01 -2.44183242e-01 -4.88534942e-02 -7.76054740e-01
-1.17547381e+00 8.89152288e-01 3.85849863e-01 9.29599851e-02
8.01508605e-01 7.23966241e-01 7.79087543e-01 6.16007335e-02
4.14753079e-01 -1.45740330e+00 5.21300673e-01 3.27781379e-01
5.61322570e-01 -1.73898280e+00 -3.69139612e-01 -7.14241445e-01
-5.73182404e-01 1.10248148e+00 1.04723513e+00 -3.47078502e-01
6.05600178e-01 2.05137193e-01 1.89007282e-01 -2.07164258e-01
-4.01629746e-01 -5.39728701e-01 5.70138812e-01 6.37727380e-01
3.28078382e-02 -6.95457980e-02 -7.26898313e-02 3.79014015e-01
3.39418918e-01 -8.74361023e-02 2.80694366e-01 5.44333816e-01
-8.00278485e-01 -9.47020054e-01 -3.43619108e-01 5.42908251e-01
-2.01449275e-01 -3.49505059e-02 -3.54976535e-01 7.24701107e-01
3.62960726e-01 7.36114562e-01 5.26889682e-01 -1.77121349e-02
6.61108410e-03 1.12492092e-01 3.32313657e-01 -7.85484493e-01
4.77861203e-02 -1.76211163e-01 5.40381074e-02 -3.87900651e-01
-4.95920628e-01 -5.22978425e-01 -1.65176392e+00 5.62889874e-01
-5.10634482e-01 2.54634410e-01 4.57397461e-01 1.14958036e+00
1.38035091e-02 4.33977574e-01 6.37947738e-01 -8.68600070e-01
-1.43277526e-01 -7.15507090e-01 -5.90046942e-01 8.57487977e-01
3.88769478e-01 -9.12752628e-01 -5.90306163e-01 4.44441259e-01]
|
[9.54534912109375, 0.5411403179168701]
|
818b551f-8c54-4a35-8663-57b765093afd
|
efficient-real-time-camera-based-estimation
|
1909.01206
| null |
https://arxiv.org/abs/1909.01206v1
|
https://arxiv.org/pdf/1909.01206v1.pdf
|
Efficient Real-Time Camera Based Estimation of Heart Rate and Its Variability
|
Remote photo-plethysmography (rPPG) uses a remotely placed camera to estimating a person's heart rate (HR). Similar to how heart rate can provide useful information about a person's vital signs, insights about the underlying physio/psychological conditions can be obtained from heart rate variability (HRV). HRV is a measure of the fine fluctuations in the intervals between heart beats. However, this measure requires temporally locating heart beats with a high degree of precision. We introduce a refined and efficient real-time rPPG pipeline with novel filtering and motion suppression that not only estimates heart rate more accurately, but also extracts the pulse waveform to time heart beats and measure heart rate variability. This method requires no rPPG specific training and is able to operate in real-time. We validate our method on a self-recorded dataset under an idealized lab setting, and show state-of-the-art results on two public dataset with realistic conditions (VicarPPG and PURE).
|
['Amogh Gudi', 'Roelof Lochmans', 'Jan van Gemert', 'Marian Bittner']
|
2019-09-03
| null | null | null | null |
['photoplethysmography-ppg', 'heart-rate-variability']
|
['medical', 'medical']
|
[ 1.45740509e-01 -3.93521905e-01 1.56448305e-01 -3.51829916e-01
-3.65740478e-01 -5.50355613e-01 -1.23679757e-01 4.98970412e-02
-1.46454915e-01 7.02131808e-01 2.35046580e-01 3.86658400e-01
2.61083513e-01 -9.83162761e-01 1.54520199e-01 -4.53023940e-01
-2.06276238e-01 1.55002236e-01 -1.48296058e-01 1.04617052e-01
1.48233790e-02 4.49750423e-01 -1.22154963e+00 -2.55207956e-01
3.93472463e-01 1.05024326e+00 -5.82167983e-01 1.25866544e+00
4.78899956e-01 4.96369451e-01 -7.50421345e-01 5.11005782e-02
2.48527408e-01 -6.34949803e-01 -2.80828148e-01 -1.96891174e-01
6.36616945e-01 -5.66051245e-01 -3.06659251e-01 4.37358737e-01
1.12184405e+00 -2.38083303e-01 -2.08683953e-01 -8.59179795e-01
1.67899788e-03 -5.76921925e-02 -1.89305142e-01 7.27315784e-01
8.59281778e-01 5.02534568e-01 2.62199581e-01 -1.17387466e-01
5.23842990e-01 8.84532213e-01 1.26373041e+00 4.02511299e-01
-1.61324680e+00 -3.77024204e-01 -8.45719457e-01 -4.25765254e-02
-1.48477471e+00 -5.48295081e-01 8.75445604e-01 -4.03463542e-01
5.89109242e-01 8.39500189e-01 1.34855926e+00 1.03579223e+00
3.08037251e-01 -1.70993611e-01 1.57914114e+00 -6.80241138e-02
2.16557309e-01 2.12538049e-01 7.92283937e-02 4.55919176e-01
3.83000150e-02 1.00927845e-01 -7.55865633e-01 -3.20237517e-01
1.12419629e+00 3.28467488e-01 -8.05993795e-01 2.69906800e-02
-1.42958879e+00 2.84771711e-01 -5.36371656e-02 2.08058342e-01
-5.19559562e-01 2.59716481e-01 3.76855761e-01 2.04009175e-01
2.72761911e-01 2.70188034e-01 -3.87928486e-01 -9.27660048e-01
-1.25272906e+00 1.28592610e-01 1.15100038e+00 3.45948786e-01
3.22303504e-01 -1.18384421e-01 -5.26310861e-01 5.07992148e-01
2.03639641e-01 8.19858313e-01 3.65157068e-01 -1.47558820e+00
2.44052615e-02 4.60283577e-01 2.80196458e-01 -1.13400495e+00
-6.74954116e-01 -2.21379966e-01 -8.17272961e-01 9.63682607e-02
5.96308649e-01 -1.46968514e-01 -2.41426200e-01 1.44313109e+00
5.80793321e-01 5.28921068e-01 -5.41629672e-01 1.28449750e+00
8.73900175e-01 1.63233593e-01 -1.25406638e-01 -7.01166987e-01
1.67883015e+00 -1.14385135e-01 -8.70524645e-01 -2.21065298e-01
-2.10063625e-02 -4.00307000e-01 9.37394917e-01 4.09339428e-01
-1.06783962e+00 -7.77048767e-01 -8.30619633e-01 3.70640331e-03
3.95025909e-02 -1.95261508e-01 4.71074373e-01 1.10439336e+00
-8.92679811e-01 1.32784057e+00 -1.00377464e+00 -6.53611183e-01
1.84582770e-01 -1.64117202e-01 -1.74976423e-01 2.94893175e-01
-1.28130293e+00 6.49621487e-01 -3.44543457e-01 3.27669591e-01
-3.46651882e-01 -1.32385612e+00 -6.05774879e-01 -1.53049920e-03
-1.63311914e-01 -8.96857679e-01 7.53241658e-01 -1.16697669e-01
-1.97739732e+00 1.06004059e+00 -1.79407477e-01 -3.74986023e-01
8.60844135e-01 -4.27721411e-01 -6.02953374e-01 8.44206572e-01
-4.12403494e-01 2.86461413e-01 9.30241823e-01 -3.05844218e-01
4.04047519e-01 -6.04946256e-01 -7.31742918e-01 -1.09332344e-02
-4.17177603e-02 7.26339370e-02 -2.08316684e-01 -2.17588097e-01
2.30884120e-01 -9.96454537e-01 1.50174767e-01 3.83608818e-01
2.33617257e-02 4.80106920e-01 4.83028889e-01 -9.63573515e-01
1.34552574e+00 -2.01191306e+00 -4.69462633e-01 8.29080716e-02
4.79100078e-01 4.08400506e-01 6.28338635e-01 4.20567513e-01
6.53685629e-02 -7.39826995e-04 -5.48179820e-02 -3.01863879e-01
-1.96922824e-01 1.36839254e-02 -2.08601013e-01 9.08435464e-01
-4.98820692e-01 1.01357353e+00 -7.18998551e-01 -5.47389388e-01
8.64960194e-01 9.84258533e-01 -1.73802719e-01 2.89666086e-01
5.62066317e-01 9.95793283e-01 9.23464522e-02 6.62386775e-01
5.16563892e-01 -8.16143751e-02 2.38934964e-01 -4.18243796e-01
-7.19526410e-02 1.99769735e-01 -1.05258322e+00 1.69717968e+00
-2.38909453e-01 9.39843535e-01 -3.75046767e-02 -4.46288198e-01
1.37176967e+00 5.94030261e-01 7.52516568e-01 -7.87739515e-01
6.06326684e-02 -3.51778567e-01 -3.72735918e-01 -1.00622690e+00
-7.13454634e-02 -3.88752788e-01 4.89408858e-02 3.88016284e-01
-3.90596986e-01 -2.19559789e-01 -8.17868039e-02 -6.68345541e-02
1.26246512e+00 2.30177417e-01 6.36911988e-01 -3.00877690e-01
3.73516053e-01 -4.74669129e-01 6.41520560e-01 6.73392713e-01
-1.01607633e+00 1.08745086e+00 2.21508250e-01 -8.76952648e-01
-6.64280593e-01 -1.28887248e+00 -3.58539879e-01 2.37988666e-01
-4.99689616e-02 -5.64883173e-01 -4.54801649e-01 3.46784964e-02
2.50308335e-01 1.94172263e-01 -5.84075272e-01 2.74564251e-02
-4.93962079e-01 -5.76531410e-01 7.58331597e-01 5.70926011e-01
7.09678590e-01 -1.14103889e+00 -1.42978692e+00 3.53386670e-01
-5.90353012e-01 -1.27881074e+00 -1.63487092e-01 -4.84055042e-01
-1.26520383e+00 -1.11705315e+00 -5.07248521e-01 2.76716977e-01
5.81860207e-02 3.20766009e-02 1.50584483e+00 -7.93493167e-02
-1.25696552e+00 8.19322407e-01 2.13498604e-02 -2.60435909e-01
1.97664320e-01 -5.00805914e-01 1.46538168e-01 -1.61892861e-01
4.57174480e-01 -1.23652744e+00 -1.37837565e+00 2.03198060e-01
-6.94055185e-02 -2.95062363e-01 -1.22225903e-01 -8.27721134e-02
7.32611001e-01 -3.15144897e-01 4.31688398e-01 -5.72633445e-01
5.37106514e-01 6.83540553e-02 -4.19607460e-01 -2.74365842e-01
-5.91425776e-01 -6.15170240e-01 3.76263380e-01 -2.11147740e-01
-5.52023351e-01 -1.09681778e-01 2.09550574e-01 -3.73143822e-01
-5.32731891e-01 -1.20463230e-01 5.08264303e-01 1.28433153e-01
9.33625042e-01 9.78661180e-02 3.09057623e-01 -3.00210208e-01
1.29578248e-01 5.54633498e-01 1.21884573e+00 -2.92734087e-01
3.63936901e-01 9.61370051e-01 3.22881669e-01 -1.16129005e+00
-6.37115180e-01 -8.12639117e-01 -7.63523698e-01 -6.55767739e-01
8.56581450e-01 -1.10914731e+00 -1.56306684e+00 4.28635865e-01
-5.29501379e-01 -2.41276845e-01 -5.49164891e-01 6.03941679e-01
-7.30289042e-01 4.89681512e-01 -8.10653508e-01 -1.02427411e+00
-9.94495332e-01 -7.41213337e-02 9.76386428e-01 5.95406294e-01
-8.95196140e-01 -1.04918492e+00 7.11102903e-01 7.59393692e-01
8.44015718e-01 1.21008682e+00 -1.94808841e-01 2.28183106e-01
-7.97422975e-02 -4.47309136e-01 1.72553852e-01 1.67773724e-01
8.20651650e-02 -9.13267583e-02 -1.38561368e+00 -2.84490913e-01
4.44804728e-01 -6.20222948e-02 4.16432798e-01 6.35840535e-01
9.16783273e-01 -6.33074790e-02 -6.02822527e-02 7.84671545e-01
1.50296831e+00 -3.95868748e-01 1.31132519e+00 -6.68654963e-02
4.06849086e-01 3.63775164e-01 2.27406412e-01 8.79234850e-01
1.96427256e-01 6.71615243e-01 1.01953007e-01 -9.20234695e-02
-1.24280810e-01 -3.84066254e-02 3.92383188e-01 4.00569022e-01
-6.21949077e-01 4.58585531e-01 -5.62369347e-01 1.84883073e-01
-1.41512251e+00 -1.30891705e+00 -5.53221762e-01 2.64171481e+00
8.76257658e-01 -1.59647971e-01 4.80732918e-01 4.33127046e-01
7.67437935e-01 2.92680353e-01 -4.67325717e-01 -5.41916788e-01
3.52963507e-01 5.61299264e-01 3.87109011e-01 1.20198615e-01
-9.32965815e-01 1.13661699e-01 7.02431488e+00 -5.42257190e-01
-1.36197782e+00 -9.06695649e-02 3.28350455e-01 -5.69252491e-01
3.32014382e-01 -6.54394478e-02 -4.19448107e-01 7.23166287e-01
1.44634891e+00 -1.15798004e-01 5.48349977e-01 6.09192789e-01
7.54964769e-01 -4.02181000e-01 -9.95482504e-01 1.54985607e+00
3.45172510e-02 -1.10037041e+00 -1.07098198e+00 -6.51549222e-03
-6.72236606e-02 -1.43363819e-01 -4.67737675e-01 -5.00135235e-02
-6.60013020e-01 -8.94606411e-01 -3.36142965e-02 1.16555643e+00
9.77172613e-01 -4.20889199e-01 3.29970717e-01 8.92088488e-02
-1.21076691e+00 3.15995723e-01 -2.37323821e-01 -4.60179180e-01
2.77948707e-01 1.41388452e+00 -7.19909906e-01 1.50576293e-01
8.29689980e-01 8.52909029e-01 -5.54942131e-01 1.09591162e+00
-4.55019742e-01 7.19374657e-01 -5.49279571e-01 3.97083998e-01
-9.99156296e-01 -2.36772060e-01 6.96652472e-01 1.01349962e+00
3.45974267e-01 4.34985459e-01 -9.03755724e-02 1.13939393e+00
4.09296244e-01 -2.03630090e-01 -4.92083639e-01 9.14525315e-02
4.31336433e-01 1.81893849e+00 -5.31118572e-01 -2.69053966e-01
-6.19924515e-02 8.63485217e-01 -2.77985156e-01 -3.43916863e-02
-8.78439903e-01 -4.26602721e-01 7.64181912e-01 6.21464491e-01
-2.45631337e-01 -2.79851377e-01 -2.82749295e-01 -1.22129953e+00
1.85240686e-01 -4.39413458e-01 4.56898987e-01 -1.02648056e+00
-9.18967545e-01 1.33458689e-01 -4.36463058e-01 -1.32383943e+00
-1.98325619e-01 1.11064687e-01 -9.09067690e-01 1.00248623e+00
-1.36103714e+00 -4.43258524e-01 -1.19188666e+00 6.84846163e-01
-1.18949391e-01 8.86702418e-01 1.20992291e+00 3.00345361e-01
-5.44276536e-01 2.09476560e-01 -7.66297221e-01 1.01688765e-01
1.20721948e+00 -1.36415899e+00 1.11120082e-01 5.65222204e-01
-3.37711483e-01 6.49319947e-01 8.18241835e-01 -4.67964709e-01
-1.70507967e+00 -8.69538844e-01 8.11839044e-01 -8.15699041e-01
2.27659598e-01 -1.81888491e-01 -7.90010452e-01 3.46262455e-01
-1.50373608e-01 8.23336065e-01 9.31510687e-01 1.32078156e-02
4.64362428e-02 -5.81232488e-01 -1.53372550e+00 -6.08084761e-02
7.51068115e-01 -6.65148675e-01 -6.72277868e-01 9.70981717e-02
7.39495968e-04 -6.90378726e-01 -1.66116953e+00 2.24026278e-01
1.21316350e+00 -1.38470554e+00 1.07290399e+00 1.41920313e-01
9.68189165e-02 -3.58016551e-01 4.33260918e-01 -9.31638002e-01
-1.70610905e-01 -1.19561410e+00 -4.04075772e-01 1.31458402e+00
-5.38674355e-01 -8.25961471e-01 6.06174171e-01 1.12671506e+00
4.24379885e-01 -2.21906260e-01 -6.97495699e-01 -5.84332347e-01
-7.50326693e-01 -2.27796316e-01 4.63195182e-02 9.87118423e-01
4.75821912e-01 4.64748517e-02 -4.55821663e-01 8.92118067e-02
9.43785131e-01 4.16845381e-01 9.17109728e-01 -1.60604572e+00
-2.62062579e-01 4.04026568e-01 -7.48587430e-01 -1.22586809e-01
-6.36549592e-01 -3.15013558e-01 -1.98241279e-01 -1.27275407e+00
1.14208542e-01 5.84497936e-02 -3.03676724e-01 2.98113465e-01
-1.89746290e-01 8.68567467e-01 1.61324248e-01 2.18637303e-01
-3.49354446e-01 -3.71158011e-02 1.04806769e+00 5.22783816e-01
-7.66644180e-01 -5.53046577e-02 -3.76964450e-01 6.16751730e-01
7.66795874e-01 -3.62336576e-01 -2.15393111e-01 4.65348721e-01
3.73809367e-01 7.05396652e-01 8.10984850e-01 -1.60721266e+00
-4.27603535e-02 2.51189590e-01 8.82610440e-01 -4.09862280e-01
4.60726976e-01 -5.11734366e-01 6.82431579e-01 7.66463041e-01
-4.98166829e-02 -1.04998067e-01 -1.19982911e-02 4.29073066e-01
1.31700531e-01 4.04655308e-01 1.07274258e+00 -5.07413566e-01
-1.77652881e-01 1.26785785e-01 -1.18453525e-01 1.67822927e-01
7.48527586e-01 -4.48073477e-01 -7.19620228e-01 -4.96551067e-01
-1.02809274e+00 3.80214080e-02 5.62776685e-01 1.61786109e-01
5.08908391e-01 -1.21081793e+00 -6.64451838e-01 4.47518229e-01
-5.60460910e-02 -5.47320843e-01 6.65408611e-01 1.45371962e+00
-8.64022493e-01 2.03591496e-01 -4.25549567e-01 -9.93417323e-01
-1.38935292e+00 2.71432489e-01 6.15177989e-01 -9.22234207e-02
-1.29205370e+00 2.02677011e-01 -7.04040527e-01 1.36120483e-01
-1.46818012e-01 -2.34155819e-01 -3.09078336e-01 1.47686020e-01
1.01232183e+00 9.12096858e-01 9.96732637e-02 -7.96030462e-02
-5.54868579e-01 7.03458905e-01 6.84283853e-01 -6.70706034e-02
9.13104355e-01 -7.36879110e-01 -1.80095047e-01 6.40935481e-01
9.84686673e-01 1.00403361e-01 -1.18158436e+00 2.19709039e-01
-6.60699368e-01 -7.57946372e-01 7.34842122e-02 -7.41515219e-01
-1.13156819e+00 9.34048474e-01 1.11046588e+00 4.48907316e-01
1.25502968e+00 -6.28563523e-01 9.53682005e-01 -1.46851927e-01
3.38999391e-01 -1.07956862e+00 -3.08677237e-02 -2.40677729e-01
5.34456611e-01 -7.01988757e-01 4.61095363e-01 -4.56751883e-01
-4.14922416e-01 1.24271154e+00 1.26312479e-01 -1.92927405e-01
5.83296776e-01 3.52655619e-01 4.41108823e-01 -9.23840180e-02
-5.54814160e-01 -2.62401607e-02 -1.24714144e-01 7.27841616e-01
5.90886354e-01 7.50851855e-02 -4.83442217e-01 -3.33744474e-02
-4.95389581e-01 7.23236561e-01 6.32566571e-01 9.24781978e-01
-3.31547260e-01 -4.43163782e-01 -4.95678067e-01 2.84622520e-01
-7.63202310e-01 2.46775478e-01 -1.16187841e-01 3.67554277e-01
-1.11235611e-01 1.17967343e+00 1.45401716e-01 -1.61887873e-02
3.00697714e-01 5.66293716e-01 6.26998186e-01 -3.80037904e-01
-7.63684630e-01 -4.10212995e-03 5.63717559e-02 -1.28637195e+00
-6.34448886e-01 -5.68589568e-01 -9.76779580e-01 -4.99117762e-01
2.73696721e-01 -2.79709637e-01 7.37418950e-01 6.04322672e-01
7.18480468e-01 5.26229680e-01 6.73769951e-01 -5.96711516e-01
-1.13622673e-01 -7.87680924e-01 -9.16754723e-01 7.29096532e-01
2.53670186e-01 -2.73056440e-02 -5.77482820e-01 3.38154167e-01]
|
[13.894272804260254, 2.8623275756835938]
|
c61a275d-a12a-4a4c-b8e9-fd7cf4a42e79
|
scalable-transformer-for-pde-surrogate
|
2305.1756
| null |
https://arxiv.org/abs/2305.17560v1
|
https://arxiv.org/pdf/2305.17560v1.pdf
|
Scalable Transformer for PDE Surrogate Modeling
|
Transformer has shown state-of-the-art performance on various applications and has recently emerged as a promising tool for surrogate modeling of partial differential equations (PDEs). Despite the introduction of linear-complexity variant, applying attention to a large number of grid points can result in instability and is still expensive to compute. In this work, we propose Factorized Transformer(FactFormer), which is based on an axial factorized kernel integral. Concretely, we introduce a learnable projection operator that decomposes the input function into multiple sub-functions with one-dimensional domain. These sub-functions are then evaluated and used to compute the instance-based kernel with an axial factorized scheme. We showcase that the proposed model is able to simulate 2D Kolmogorov flow on a 256 by 256 grid and 3D smoke buoyancy on a 64 by 64 by 64 grid with good accuracy and efficiency. In addition, we find out that with the factorization scheme, the attention matrices enjoy a more compact spectrum than full softmax-free attention matrices.
|
['Amir Barati Farimani', 'Dule Shu', 'Zijie Li']
|
2023-05-27
| null | null | null | null |
['pde-surrogate-modeling']
|
['miscellaneous']
|
[-2.32222483e-01 -2.16684178e-01 4.06628489e-01 1.33725345e-01
-7.46270478e-01 -4.98824060e-01 5.28121054e-01 -1.61242828e-01
-4.85883653e-01 7.78691888e-01 1.04255661e-01 -4.25230175e-01
-2.66080320e-01 -6.16450548e-01 -9.02770698e-01 -9.45282280e-01
-2.20334440e-01 5.16721606e-01 -2.85457790e-01 3.14994678e-02
3.28858360e-03 6.64870203e-01 -1.44800007e+00 4.95290495e-02
8.35514009e-01 1.19927847e+00 8.82951468e-02 7.82268584e-01
4.21311483e-02 8.66822004e-01 -2.67510951e-01 -1.53913736e-01
5.58439255e-01 -1.96984813e-01 -7.77523160e-01 -7.38344118e-02
4.96578693e-01 -2.55994022e-01 -2.64506429e-01 9.92750108e-01
4.78428423e-01 6.86801195e-01 9.34907317e-01 -9.45745170e-01
-7.25162983e-01 2.37898365e-01 -3.35919142e-01 2.97221184e-01
-1.61026686e-01 1.01185575e-01 8.61318529e-01 -1.34493244e+00
3.07930648e-01 1.14901316e+00 8.32530558e-01 2.84439296e-01
-1.39979911e+00 -4.67891812e-01 -7.42551014e-02 2.57361662e-02
-1.39648354e+00 -7.29850829e-02 5.66261828e-01 -6.31740630e-01
1.14764690e+00 3.11140448e-01 8.10086668e-01 6.76708639e-01
4.49684322e-01 4.35866952e-01 1.15813828e+00 2.60168258e-02
2.97084153e-01 1.38577119e-01 -5.49404435e-02 6.73732638e-01
6.79536983e-02 1.34801477e-01 -3.25846195e-01 -3.42272490e-01
9.83646810e-01 -5.42623736e-02 -7.08225548e-01 -2.49243513e-01
-9.75947797e-01 1.03887606e+00 5.41260660e-01 1.12280384e-01
-6.17973864e-01 2.30013028e-01 3.67786020e-01 2.05665261e-01
7.89962471e-01 5.87090492e-01 -6.12778306e-01 -2.28475079e-01
-9.80996370e-01 5.09713650e-01 8.77971113e-01 3.25910658e-01
7.66299903e-01 4.22773302e-01 -9.99939889e-02 6.36886597e-01
1.05417088e-01 8.52831721e-01 5.82820356e-01 -1.02510512e+00
4.07261886e-02 3.47432464e-01 2.98208833e-01 -9.99416828e-01
-4.03599083e-01 -5.92601359e-01 -1.35128641e+00 6.99243069e-01
3.03533673e-01 -3.71063828e-01 -7.28795886e-01 1.64778543e+00
5.30826330e-01 8.26494932e-01 1.71027318e-01 1.29342842e+00
4.69012678e-01 1.04811645e+00 -2.02589020e-01 -2.41821289e-01
1.32718277e+00 -9.32815969e-01 -5.22281587e-01 2.15816170e-01
3.54900092e-01 -5.10321796e-01 1.10994101e+00 3.76927555e-01
-1.23700535e+00 -6.83899283e-01 -8.38491321e-01 -1.95526347e-01
-2.07199171e-01 3.59837897e-02 5.78080893e-01 4.88175243e-01
-1.09801269e+00 1.00615788e+00 -1.04928517e+00 2.55337834e-01
1.83927387e-01 2.09766358e-01 -5.07063627e-01 3.34015399e-01
-1.13897657e+00 8.47623527e-01 -8.97414461e-02 2.34920323e-01
-9.79683995e-01 -1.56682265e+00 -9.12304699e-01 4.78673875e-01
-5.82959503e-02 -9.68855083e-01 1.04391253e+00 -6.68113768e-01
-1.85337758e+00 6.03699744e-01 1.14870006e-02 -6.40419543e-01
5.61467826e-01 -5.39413512e-01 1.25829190e-01 8.47733021e-02
-2.34826095e-02 1.69158638e-01 1.14003789e+00 -7.21092165e-01
-2.36116886e-01 -1.57929450e-01 9.52211842e-02 4.45958525e-01
-4.66884226e-01 -3.63921523e-01 4.30561081e-02 -8.82526636e-01
-3.09596479e-01 -9.61697817e-01 -2.83873856e-01 2.24585131e-01
2.88634486e-02 2.74496265e-02 7.19995201e-01 -8.56673598e-01
9.75531220e-01 -2.11137104e+00 7.11246431e-01 -1.09934598e-01
2.32214555e-01 4.43937242e-01 1.93054602e-01 2.87148327e-01
-1.45176291e-01 -6.89140260e-02 -5.05678356e-01 -5.90289950e-01
2.72050127e-02 1.71032339e-01 -5.24984002e-01 8.52506340e-01
2.34895945e-01 6.84221148e-01 -7.68086076e-01 -6.80077374e-02
3.78365219e-01 9.65056658e-01 -8.21960747e-01 3.68012458e-01
-2.09606085e-02 5.62450886e-01 -3.36932808e-01 1.04284197e-01
9.00676608e-01 -3.99973661e-01 -3.08417976e-01 -2.52756387e-01
-2.01302707e-01 1.26899583e-02 -1.28168631e+00 1.67833817e+00
-8.97848606e-01 6.00904286e-01 5.53589821e-01 -1.11055148e+00
6.63416564e-01 3.74056429e-01 5.64304531e-01 -2.93539256e-01
1.67800426e-01 1.66201398e-01 -1.35284185e-01 -3.08742344e-01
3.86304438e-01 -5.68536580e-01 2.06420422e-01 5.93444824e-01
6.12310767e-02 -3.17680568e-01 2.96949167e-02 -1.06489457e-01
1.02131128e+00 9.85932536e-03 1.28564835e-01 -8.26590061e-01
8.21787953e-01 -3.25459301e-01 2.38370255e-01 3.90864342e-01
2.68065453e-01 5.34990311e-01 3.80134583e-01 -7.18033910e-01
-9.86503065e-01 -7.76501298e-01 -3.72680068e-01 7.69861877e-01
-1.66474774e-01 -2.09973693e-01 -7.97744751e-01 -2.65405357e-01
3.82055193e-01 5.35762668e-01 -1.05841982e+00 -5.15252240e-02
-5.51088214e-01 -8.90316129e-01 6.11592829e-01 5.21169662e-01
4.78013963e-01 -9.83787000e-01 -9.15356636e-01 1.46177858e-01
1.50730059e-01 -8.66518497e-01 -5.61304748e-01 3.46366405e-01
-7.29671240e-01 -1.11109340e+00 -1.15841424e+00 -5.50810516e-01
5.13845742e-01 -3.94586660e-02 1.01361692e+00 -3.19526613e-01
-2.08431602e-01 3.69068921e-01 -9.66037959e-02 -7.75802061e-02
-1.40872240e-01 -5.23350500e-02 2.42785066e-01 4.03218269e-01
-4.48879272e-01 -7.45017946e-01 -6.11306310e-01 3.65027017e-03
-1.02000666e+00 6.43059909e-02 1.51188672e-01 1.10013044e+00
5.35930395e-01 -4.24831584e-02 3.00307482e-01 -5.61917722e-01
8.12635303e-01 -4.56793457e-01 -8.84951055e-01 -1.17297582e-01
-4.16837156e-01 4.71641392e-01 1.05965507e+00 -5.12497723e-01
-1.01438320e+00 -4.81962189e-02 -1.62203133e-01 -1.20617032e+00
3.19546014e-01 6.31928205e-01 3.01813066e-01 -3.68356824e-01
6.94142103e-01 2.76574135e-01 1.23912066e-01 -7.28818178e-01
1.22113764e-01 2.89998770e-01 2.77408063e-01 -7.02645659e-01
8.34371448e-01 4.22199130e-01 3.43002528e-01 -9.73109305e-01
-5.92880607e-01 -2.65162855e-01 -2.83502191e-01 -2.91052051e-02
7.18853056e-01 -9.04221535e-01 -9.65159118e-01 6.67591214e-01
-1.04715765e+00 -7.80827284e-01 -5.48166633e-01 5.24296105e-01
-4.69423711e-01 1.66633695e-01 -9.49556649e-01 -8.53526354e-01
-6.65082633e-01 -1.05610585e+00 1.15294981e+00 2.94137951e-02
1.22291699e-01 -1.18038094e+00 3.95699203e-01 -8.96724313e-02
7.69124448e-01 1.58582285e-01 7.09807336e-01 -1.49825022e-01
-2.54010439e-01 1.67756960e-01 -2.30788365e-01 6.93569779e-01
5.93094267e-02 -1.61505163e-01 -1.05165684e+00 -4.80453283e-01
4.97619838e-01 -2.05569461e-01 9.37949121e-01 4.53831434e-01
9.95960772e-01 -3.25389087e-01 -5.20762578e-02 1.09963584e+00
1.36185968e+00 -2.49683276e-01 1.04218788e-01 -2.95204461e-01
7.42047250e-01 7.86742792e-02 1.31454095e-01 6.34106874e-01
-4.56770323e-02 4.94795382e-01 4.36189532e-01 -1.21321633e-01
6.94161430e-02 1.44806698e-01 3.32810909e-01 8.07681501e-01
-3.06453437e-01 -1.07953772e-01 -8.36165786e-01 4.84514296e-01
-1.60140383e+00 -7.51883507e-01 6.92762658e-02 2.13992620e+00
5.91487885e-01 -1.76611081e-01 -1.41489089e-01 9.51695070e-02
2.23492339e-01 1.17720157e-01 -5.90602100e-01 -3.76828074e-01
-2.39668265e-02 5.67669094e-01 5.07807851e-01 9.82854486e-01
-1.06402612e+00 5.90264559e-01 5.93935776e+00 7.86417544e-01
-1.57589340e+00 2.12579966e-01 5.81387639e-01 -2.58623026e-02
-2.29890972e-01 -4.66629833e-01 -5.14280558e-01 4.71426606e-01
8.79088759e-01 -3.20794135e-01 7.91028559e-01 9.32158709e-01
1.28900960e-01 2.85222888e-01 -8.54223132e-01 1.01510835e+00
4.22537699e-02 -1.35213947e+00 4.98255417e-02 8.59559607e-03
7.15607584e-01 1.92119777e-01 2.55805701e-01 4.57588881e-01
-2.21514888e-02 -1.24612427e+00 5.50105095e-01 3.97130936e-01
1.01020801e+00 -6.72821045e-01 7.15190053e-01 4.50452596e-01
-1.23638713e+00 6.69984967e-02 -5.48355937e-01 -4.28199917e-01
2.49591023e-01 6.02167308e-01 -5.00686765e-01 4.14914846e-01
7.61181653e-01 6.54717147e-01 3.60444449e-02 8.70091736e-01
2.22660169e-01 6.04037642e-01 -6.91428185e-01 1.04515299e-01
4.68457967e-01 -5.75680256e-01 6.78440213e-01 1.07434857e+00
7.14063168e-01 1.84524775e-01 1.50463536e-01 8.70780349e-01
-5.81096932e-02 2.37781197e-01 -5.33850372e-01 3.75506394e-02
-2.87569076e-01 1.25003326e+00 -5.20082474e-01 -4.27832603e-01
-2.93555558e-01 1.03483593e+00 4.74006534e-01 5.12328923e-01
-1.07157624e+00 -6.38523847e-02 1.25582337e+00 1.82507008e-01
6.73350513e-01 -1.53800353e-01 6.02714233e-02 -1.39782858e+00
-1.21992148e-01 -6.30997002e-01 1.52805805e-01 -8.05588722e-01
-1.42057347e+00 9.75221634e-01 -4.01584394e-02 -9.43560958e-01
-7.70336315e-02 -9.98684347e-01 -4.29462135e-01 1.16294968e+00
-1.53941238e+00 -8.49303901e-01 -3.69286537e-01 6.83045805e-01
3.74231458e-01 -1.17024602e-02 1.17196238e+00 5.13219714e-01
-2.97001839e-01 3.23716968e-01 3.29722166e-01 5.35618141e-03
2.18193576e-01 -1.39019775e+00 3.52893949e-01 5.69631934e-01
1.11128218e-01 5.35040021e-01 7.40705669e-01 -4.86142695e-01
-1.55820107e+00 -1.14092660e+00 5.65188050e-01 -2.17651263e-01
6.21424496e-01 -1.29029423e-01 -1.21623409e+00 5.84298432e-01
2.26609692e-01 6.91703141e-01 4.30802494e-01 -2.57113546e-01
-2.31008470e-01 -1.28827885e-01 -1.17542732e+00 2.08158329e-01
7.02633083e-01 -5.62327027e-01 -2.36052379e-01 3.77622038e-01
5.29016852e-01 -8.66561592e-01 -8.91286492e-01 4.84747142e-01
4.99326617e-01 -9.25886750e-01 9.38956559e-01 -6.43037677e-01
3.46176654e-01 -3.35453421e-01 -3.44430096e-02 -1.56514227e+00
-5.82253337e-01 -8.30554187e-01 -3.23647916e-01 6.52871132e-01
1.78412631e-01 -8.78736198e-01 6.04935586e-01 3.58788371e-01
-2.14828491e-01 -1.13023543e+00 -1.26292777e+00 -5.40976584e-01
4.82799858e-01 -5.24525642e-01 3.81363273e-01 8.47720504e-01
-6.53388724e-02 3.31432730e-01 -6.33819997e-01 3.27900976e-01
5.04008472e-01 3.87577385e-01 5.23335218e-01 -1.14880884e+00
-6.83383167e-01 -4.34134960e-01 -2.04722553e-01 -1.18593371e+00
3.94178331e-01 -9.91047323e-01 -1.04681745e-01 -9.59300280e-01
-1.59574345e-01 -4.44550663e-01 -4.06423628e-01 2.46293783e-01
-1.08884320e-01 4.50206071e-01 1.01569884e-01 -1.89689711e-01
-3.32379900e-02 7.56062448e-01 1.29133737e+00 -6.32426515e-02
-5.61499745e-02 -1.89134048e-03 -2.93583184e-01 5.01539052e-01
5.96264064e-01 -3.48759085e-01 -4.00088787e-01 -3.31594706e-01
2.43120536e-01 2.49637589e-01 3.92871380e-01 -1.08722603e+00
1.12306431e-01 3.91647518e-02 2.13640526e-01 -1.24576859e-01
6.82587385e-01 -8.21500361e-01 3.48908901e-01 4.70364600e-01
-1.39350832e-01 2.92101443e-01 6.13907874e-01 3.90806735e-01
-2.82849908e-01 -4.16132137e-02 8.41599405e-01 -1.70067072e-01
-3.35768163e-02 5.12499154e-01 -4.79724407e-01 1.31358430e-01
8.17244053e-01 3.65644604e-01 2.71023780e-01 -2.84762144e-01
-5.74277997e-01 -8.94820318e-03 3.88949484e-01 8.68747160e-02
3.80213201e-01 -1.25565040e+00 -9.08619165e-01 6.41464829e-01
-6.33004427e-01 1.74640059e-01 5.36206841e-01 8.89206648e-01
-8.57863963e-01 2.84847438e-01 -2.06825316e-01 -4.85704273e-01
-9.59412932e-01 6.98666334e-01 5.49578547e-01 -5.64625204e-01
-9.43637788e-01 9.69348013e-01 4.82665658e-01 -4.06345248e-01
3.72661054e-02 -7.77925909e-01 -9.13029909e-03 8.34451318e-02
6.11361980e-01 5.81197500e-01 1.47477284e-01 -6.99818850e-01
-2.11604938e-01 8.59282553e-01 4.90337700e-01 -5.62762618e-02
1.43511939e+00 4.24323738e-01 -8.57853517e-02 3.42862368e-01
1.19257808e+00 1.60879627e-01 -1.57402027e+00 -1.76826958e-02
-6.87425613e-01 -4.19322193e-01 3.02967697e-01 -3.93458158e-01
-1.40875041e+00 1.05119240e+00 3.07970107e-01 4.35821325e-01
1.00248599e+00 -6.29213035e-01 8.84097636e-01 2.22245321e-01
1.26350746e-01 -3.09669137e-01 -3.32261622e-01 7.00414538e-01
1.14721811e+00 -9.60997164e-01 -1.43889666e-01 -1.04802065e-01
-3.49806309e-01 1.02713633e+00 2.47740984e-01 -5.95481932e-01
1.05944312e+00 7.11519539e-01 1.05263181e-01 7.23707899e-02
-7.31823146e-01 1.75679013e-01 4.52949882e-01 2.39018679e-01
1.88534126e-01 -4.32124399e-02 7.20518455e-02 4.69449282e-01
-9.78567973e-02 4.20506895e-02 2.33466342e-01 3.81882936e-01
-3.85893136e-02 -7.60075152e-01 -3.94522816e-01 3.77799630e-01
-6.11379623e-01 -2.86507189e-01 3.32665533e-01 5.92369497e-01
-7.74310529e-02 2.70441473e-01 1.29589766e-01 -3.77538055e-02
1.17038742e-01 3.23886484e-01 3.72445315e-01 -3.72256130e-01
-8.22616458e-01 -3.69540378e-02 -3.22952569e-01 -4.44735855e-01
-2.54603744e-01 -6.59646511e-01 -1.00806379e+00 -3.24098557e-01
-8.19994882e-02 6.03829801e-01 2.88829058e-01 7.36667454e-01
6.65817380e-01 6.94135666e-01 3.06196600e-01 -1.47607076e+00
-7.17016160e-01 -1.13750422e+00 -6.68376088e-01 4.04729128e-01
7.54288793e-01 -8.30135584e-01 -7.17437744e-01 3.73121686e-02]
|
[6.55389404296875, 3.4138762950897217]
|
ced08641-c2b5-4ec9-87de-c5d9860ef599
|
av-nerf-learning-neural-fields-for-real-world
|
2302.02088
| null |
https://arxiv.org/abs/2302.02088v2
|
https://arxiv.org/pdf/2302.02088v2.pdf
|
AV-NeRF: Learning Neural Fields for Real-World Audio-Visual Scene Synthesis
|
Human perception of the complex world relies on a comprehensive analysis of multi-modal signals, and the co-occurrences of audio and video signals provide humans with rich cues. This paper focuses on novel audio-visual scene synthesis in the real world. Given a video recording of an audio-visual scene, the task is to synthesize new videos with spatial audios along arbitrary novel camera trajectories in that audio-visual scene. Directly using a NeRF-based model for audio synthesis is insufficient due to its lack of prior knowledge and acoustic supervision. To tackle the challenges, we first propose an acoustic-aware audio generation module that integrates our prior knowledge of audio propagation into NeRF, in which we associate audio generation with the 3D geometry of the visual environment. In addition, we propose a coordinate transformation module that expresses a viewing direction relative to the sound source. Such a direction transformation helps the model learn sound source-centric acoustic fields. Moreover, we utilize a head-related impulse response function to synthesize pseudo binaural audio for data augmentation that strengthens training. We qualitatively and quantitatively demonstrate the advantage of our model on real-world audio-visual scenes. We refer interested readers to view our video results for convincing comparisons.
|
['Chenliang Xu', 'Anurag Kumar', 'Yapeng Tian', 'Chao Huang', 'Susan Liang']
|
2023-02-04
| null | null | null | null |
['audio-generation']
|
['audio']
|
[ 2.59682089e-01 -2.35767812e-01 5.10340512e-01 -2.23097235e-01
-1.04350793e+00 -5.12421846e-01 3.86916131e-01 -1.51151642e-01
-6.65352046e-02 3.60244244e-01 5.55134535e-01 2.75698364e-01
2.31718704e-01 -5.00297368e-01 -1.12608862e+00 -6.47007406e-01
1.08958684e-01 -2.85025716e-01 3.13936949e-01 -1.27151757e-01
-4.47021239e-02 3.13905060e-01 -1.94912398e+00 5.24395585e-01
5.10558009e-01 9.45314109e-01 5.68598032e-01 1.10796404e+00
2.58761615e-01 6.99995875e-01 -5.36191642e-01 -6.26603961e-02
2.06272721e-01 -4.04580027e-01 -3.26533705e-01 2.72364646e-01
6.50873601e-01 -4.30552691e-01 -6.37659907e-01 8.73426557e-01
6.10433817e-01 4.43349153e-01 4.86792147e-01 -1.32218599e+00
-4.93526310e-01 4.27478641e-01 -3.56319994e-01 -1.01527041e-02
7.90616453e-01 3.72615308e-01 9.87092376e-01 -1.02816725e+00
5.27897239e-01 1.11798477e+00 4.74894881e-01 4.15167332e-01
-1.09134245e+00 -5.71959317e-01 1.62831455e-01 4.33130324e-01
-1.48440039e+00 -7.08890319e-01 1.27371800e+00 -4.98259395e-01
5.34928799e-01 2.63956875e-01 9.17692721e-01 1.18182468e+00
-1.20129675e-01 7.21084177e-01 6.74921870e-01 -5.41146398e-01
2.11584985e-01 7.37560615e-02 -6.27427101e-01 2.68351883e-01
-5.79133093e-01 3.35817635e-01 -9.93258476e-01 1.32737666e-01
7.84997940e-01 -2.58063525e-01 -5.66893399e-01 -3.87637466e-01
-1.35851598e+00 4.56410229e-01 5.18505216e-01 3.30408514e-01
-2.17845336e-01 3.53842854e-01 8.68952181e-03 -8.39388464e-03
3.18399221e-01 3.20528179e-01 -6.48296550e-02 -3.25284675e-02
-8.95446599e-01 3.00246805e-01 1.30502209e-01 9.46015954e-01
4.77942020e-01 5.15895367e-01 5.81703670e-02 9.33639824e-01
5.06311655e-01 7.83753514e-01 1.78656936e-01 -1.31739366e+00
3.47234249e-01 -1.32500827e-01 1.03219151e-01 -1.16169012e+00
-3.30307752e-01 -5.00596941e-01 -7.30367601e-01 2.61172280e-02
1.82968959e-01 -8.30404311e-02 -5.43953359e-01 1.93819034e+00
5.34916401e-01 7.87704229e-01 -2.30741858e-01 1.26336753e+00
6.71756685e-01 1.07505667e+00 -3.86453420e-01 -3.31491053e-01
1.04004240e+00 -8.33397448e-01 -8.55189800e-01 -2.10048426e-02
4.45702150e-02 -8.11318636e-01 1.16325665e+00 4.30415928e-01
-1.17109179e+00 -9.95962858e-01 -8.31044734e-01 -6.12171516e-02
1.67550355e-01 -3.46996486e-02 1.77349687e-01 2.20098749e-01
-1.06593359e+00 -2.72898152e-02 -7.31332004e-01 -1.19526796e-01
-1.73189759e-01 -2.22684965e-01 -3.28349143e-01 -9.67932791e-02
-1.15821421e+00 3.25088948e-01 3.88988815e-02 1.07574932e-01
-1.39302242e+00 -8.91657591e-01 -1.14823866e+00 -3.50754075e-02
4.72481728e-01 -8.85715604e-01 1.42929459e+00 -8.28805447e-01
-1.69492030e+00 9.82926190e-02 -3.75440359e-01 -1.37166649e-01
1.94625616e-01 -1.34149447e-01 -5.78962564e-01 5.46431482e-01
5.15576899e-02 7.32445538e-01 1.15763092e+00 -1.80433822e+00
-7.50692964e-01 -8.19697455e-02 1.78675979e-01 5.29687703e-01
-2.27949530e-01 -3.16339694e-02 -5.12105703e-01 -9.32452738e-01
-2.24796799e-03 -6.17428124e-01 -1.54061049e-01 -3.39782089e-02
-2.80785501e-01 3.21322441e-01 6.33593440e-01 -6.08869731e-01
1.09630275e+00 -2.31204414e+00 3.19079161e-01 1.55498579e-01
1.49505585e-01 -2.55626112e-01 -3.47546756e-01 1.68442443e-01
1.01094088e-02 -3.11662078e-01 -3.22598852e-02 -4.59069312e-01
-1.19298436e-01 -2.21118122e-01 -6.00182354e-01 2.07556009e-01
6.96832836e-02 4.83768821e-01 -1.04133201e+00 -4.14303631e-01
3.27521741e-01 9.19927716e-01 -1.08931112e+00 5.08497298e-01
-2.08694860e-01 9.63928580e-01 -2.15845034e-01 5.19806564e-01
5.21141827e-01 8.32253024e-02 -2.50198692e-01 -4.25537616e-01
-2.55575448e-01 -1.10508695e-01 -1.33325040e+00 2.17943645e+00
-7.39834130e-01 6.15995407e-01 3.11115354e-01 -6.38802886e-01
7.58262336e-01 5.37375510e-01 4.56671089e-01 -5.82595885e-01
-3.53581980e-02 1.19956471e-01 -2.65572876e-01 -6.81245506e-01
5.11153817e-01 -1.80654436e-01 9.72576141e-02 1.87740415e-01
3.63167405e-01 -4.31692868e-01 -1.85340345e-01 2.55226940e-01
9.32184815e-01 1.20556824e-01 -9.51617360e-02 4.13595945e-01
5.59806943e-01 -6.63681328e-01 4.16052312e-01 5.15600443e-01
9.25744176e-02 1.03991377e+00 -3.20675671e-02 1.19677454e-01
-9.79185104e-01 -1.48524380e+00 4.46450897e-02 1.25895941e+00
6.49660230e-02 -6.99983716e-01 -7.74220228e-01 -1.92264706e-01
-4.30055946e-01 6.39626026e-01 -2.75342911e-01 -1.47086129e-01
-5.29346168e-01 -1.86300933e-01 3.50743830e-01 3.70747089e-01
1.86721742e-01 -8.96591842e-01 -4.04681325e-01 1.97828516e-01
-6.93177938e-01 -1.34801817e+00 -6.76245868e-01 -2.51536161e-01
-3.37608069e-01 -7.43900716e-01 -9.73230302e-01 -6.99183345e-01
3.76375407e-01 3.92121285e-01 5.94515800e-01 -5.10290265e-01
-4.25154924e-01 9.31346059e-01 -3.53029788e-01 -3.05117726e-01
-4.30394769e-01 -5.67559123e-01 4.17083055e-01 5.84088385e-01
-3.34464282e-01 -9.35611546e-01 -5.19890308e-01 2.62364686e-01
-9.97099638e-01 3.12354922e-01 1.51092038e-01 7.18238592e-01
6.83276713e-01 4.53938618e-02 5.62704921e-01 -1.65432331e-03
3.05878371e-01 -4.15150136e-01 -4.81536686e-01 -2.33219955e-02
2.61429220e-01 -3.38339657e-01 7.23261237e-01 -5.81164002e-01
-1.26105487e+00 4.32399094e-01 -3.11933875e-01 -8.02527130e-01
-3.59405220e-01 3.80166441e-01 -5.00048280e-01 1.34143695e-01
7.02179313e-01 2.52136052e-01 -3.51903558e-01 -5.46562731e-01
5.82857251e-01 5.98993003e-01 9.87207174e-01 -5.21783829e-01
9.20902669e-01 6.45263076e-01 6.34422945e-03 -1.23341405e+00
-7.62352407e-01 -4.20714587e-01 -6.06090009e-01 -7.20099151e-01
9.35810208e-01 -1.10946524e+00 -8.45501721e-01 5.48270524e-01
-1.42546797e+00 -3.30476254e-01 -2.04163477e-01 8.99293542e-01
-9.41087723e-01 3.74096990e-01 -4.00621891e-01 -8.84213030e-01
3.38308960e-01 -1.19470954e+00 1.19964325e+00 7.72305056e-02
1.26651779e-01 -8.29908431e-01 2.56096810e-01 4.12221223e-01
2.23462969e-01 9.47939530e-02 6.02941155e-01 1.15941212e-01
-8.99200797e-01 8.80643725e-02 9.57551003e-02 4.58332956e-01
9.02077109e-02 -7.48594552e-02 -1.48790014e+00 -1.61427274e-01
1.31733924e-01 -1.59242779e-01 6.94525003e-01 7.55181611e-01
1.39998460e+00 -2.39965454e-01 -2.59377658e-02 8.09286952e-01
1.11170697e+00 3.10659349e-01 2.80452847e-01 -7.86229819e-02
9.43391025e-01 7.99739301e-01 5.97533822e-01 7.10776091e-01
5.40372908e-01 1.14276874e+00 5.81066906e-01 -1.05453208e-01
-4.49212521e-01 -6.57637894e-01 5.18313825e-01 1.20545638e+00
-3.72650921e-02 -3.54285210e-01 -7.78888404e-01 5.40675700e-01
-1.55658114e+00 -9.58498597e-01 1.34073645e-01 2.07457995e+00
7.67847836e-01 -1.83821827e-01 9.52948183e-02 2.34432518e-01
8.06154847e-01 6.65153340e-02 -4.19391125e-01 1.06667846e-01
-1.33200347e-01 1.80487707e-01 -2.20981985e-01 8.30615461e-01
-8.89092982e-01 6.79072261e-01 5.93056154e+00 8.30983818e-01
-1.25381780e+00 1.05513752e-01 5.37536629e-02 -4.78020102e-01
-6.06755674e-01 -1.78910956e-01 -4.52938676e-01 2.06872404e-01
8.24030161e-01 -1.37021378e-01 6.96605086e-01 4.03041363e-01
5.03251731e-01 7.89334998e-02 -1.30608428e+00 1.25481081e+00
4.40040767e-01 -1.27948284e+00 2.54840940e-01 -1.88698277e-01
5.33040583e-01 -2.81789660e-01 3.45919251e-01 -2.48016175e-02
-1.17113777e-02 -8.14003706e-01 1.17002678e+00 7.28171766e-01
8.71439695e-01 -6.62090719e-01 7.70250112e-02 2.76306152e-01
-1.42698431e+00 -9.14184898e-02 -3.43003422e-02 1.27683416e-01
6.99561298e-01 3.88479263e-01 -8.36768925e-01 5.58182061e-01
9.23618734e-01 8.32707763e-01 -3.29034269e-01 1.22590792e+00
-2.61201829e-01 5.02876937e-01 -3.81706476e-01 5.48322380e-01
-1.46773264e-01 1.63374200e-01 1.02746511e+00 1.10331118e+00
7.59202600e-01 1.02819905e-01 2.97456682e-02 8.50131214e-01
7.67603815e-02 1.14572793e-01 -8.55934024e-01 2.37224355e-01
4.10104334e-01 1.13682914e+00 -2.49630213e-01 1.50204729e-02
-4.55843180e-01 6.78241909e-01 -3.24743897e-01 7.02515900e-01
-9.96306419e-01 -3.83837342e-01 6.19383097e-01 2.68178165e-01
1.00191578e-01 -2.35784799e-01 1.47647381e-01 -1.17483580e+00
1.97982192e-01 -8.01268160e-01 -4.84927669e-02 -1.47676992e+00
-1.01054978e+00 7.64125228e-01 5.09274006e-02 -1.54770541e+00
-4.00975078e-01 -3.04771453e-01 -4.35476929e-01 5.56565762e-01
-1.37552333e+00 -1.22291708e+00 -4.22611773e-01 9.53147292e-01
5.89527726e-01 -1.98460311e-01 5.90277851e-01 5.77505827e-01
-6.72615319e-02 6.29718781e-01 -1.02360383e-01 -6.00295328e-02
9.92703676e-01 -8.34978044e-01 2.24082842e-01 8.55390608e-01
3.89016837e-01 2.64325202e-01 9.61363316e-01 -3.64363760e-01
-1.17327857e+00 -1.24889362e+00 4.90972936e-01 -3.98514867e-01
6.53174579e-01 -6.22088253e-01 -9.42793131e-01 6.58353865e-01
1.13086328e-01 1.32985950e-01 6.61705792e-01 -2.57272363e-01
-3.66495490e-01 -3.11007828e-01 -5.62665224e-01 6.30528808e-01
1.03223479e+00 -8.20881367e-01 -4.69592035e-01 9.40969810e-02
1.10684073e+00 -4.41751838e-01 -5.31899452e-01 3.96348029e-01
5.50707638e-01 -8.88851345e-01 1.20748103e+00 -2.97170162e-01
4.68036383e-01 -6.59147203e-01 -5.40679038e-01 -1.70578492e+00
-1.29367083e-01 -6.38900399e-01 -2.17083395e-02 1.28216255e+00
1.61646813e-01 -2.03302994e-01 2.67920285e-01 -8.69070217e-02
-4.94601995e-01 -2.51318425e-01 -9.05646503e-01 -5.99196970e-01
-3.04373533e-01 -1.08162272e+00 5.90641558e-01 8.40028524e-01
1.02284007e-01 3.25860083e-01 -6.29431427e-01 5.50423384e-01
7.36156702e-01 -1.80453863e-02 9.66833353e-01 -1.03498554e+00
-6.65453970e-01 -3.10798548e-02 -4.22618002e-01 -1.31546223e+00
3.01083744e-01 -7.03655779e-01 3.77427578e-01 -1.42700589e+00
-5.25155738e-02 -1.64898291e-01 -1.18905284e-01 3.78098302e-02
2.63751894e-01 3.48254323e-01 4.74384397e-01 -7.38246515e-02
-5.53270400e-01 1.01424456e+00 1.49561572e+00 -8.06852803e-02
-3.44911128e-01 2.37838868e-02 -3.67286533e-01 9.39354062e-01
3.29104245e-01 -3.82118523e-01 -6.21412992e-01 -7.39235282e-01
4.10669774e-01 6.94104850e-01 6.26168549e-01 -1.20601153e+00
4.32248920e-01 -2.07809508e-01 1.84677601e-01 -4.79467869e-01
9.61301446e-01 -8.70477498e-01 2.27037951e-01 -9.12972912e-02
-5.66697121e-01 -2.15735704e-01 2.14571744e-01 7.78576434e-01
-5.61282098e-01 1.03679150e-01 6.31681383e-01 3.81850414e-02
-3.87774080e-01 3.91058028e-01 -5.75551212e-01 4.48894762e-02
8.07210624e-01 -3.54710519e-02 -1.62402272e-01 -8.48986745e-01
-1.06306875e+00 -1.94346439e-02 2.34787121e-01 5.67728281e-01
9.14275944e-01 -1.72300279e+00 -6.97445631e-01 4.06564981e-01
2.38643631e-01 7.15579763e-02 8.25122654e-01 7.01267183e-01
-2.97716409e-01 2.12558076e-01 -2.87387460e-01 -9.52315986e-01
-1.11294615e+00 5.57595253e-01 3.59012872e-01 4.49676752e-01
-5.79878747e-01 9.11828756e-01 9.29750383e-01 -1.56052098e-01
4.73671705e-01 -3.83242369e-01 -1.50577277e-01 -9.92865041e-02
6.77139819e-01 3.26410532e-01 -1.73461765e-01 -8.78544509e-01
-1.26252025e-01 6.89975560e-01 4.95001733e-01 -7.36830354e-01
1.21763289e+00 -5.92685401e-01 1.81804672e-01 9.72785056e-01
1.26801848e+00 4.40545410e-01 -1.45004427e+00 -3.70093971e-01
-6.92356706e-01 -6.19028687e-01 1.83121026e-01 -6.46198034e-01
-1.04292524e+00 1.44312024e+00 4.11693394e-01 -1.22351959e-01
1.35668492e+00 8.88203532e-02 7.11579561e-01 1.56129450e-01
3.67555201e-01 -9.45902944e-01 6.40116811e-01 4.82837796e-01
1.23853159e+00 -7.88629889e-01 -5.15842497e-01 -4.65872347e-01
-7.26072073e-01 1.08772957e+00 6.11655533e-01 2.89152443e-01
7.69994497e-01 2.60876328e-01 1.86661735e-01 1.37348399e-01
-7.45807648e-01 -1.42108411e-01 4.72679704e-01 9.16132629e-01
1.41452238e-01 -1.86018124e-01 7.07239747e-01 6.99082732e-01
-5.50565243e-01 -1.93138316e-01 7.09375322e-01 3.94728005e-01
-4.05482054e-01 -5.93760729e-01 -6.32722318e-01 -3.23132545e-01
-1.58123657e-01 -5.27343825e-02 -1.39249722e-02 1.70216024e-01
1.05260946e-01 1.10827112e+00 2.39267975e-01 -5.96212208e-01
4.83394712e-01 3.14744160e-04 6.85851812e-01 -5.65033555e-01
-8.82672668e-02 7.77856171e-01 -3.81501257e-01 -5.93360364e-01
-4.11664873e-01 -6.64631486e-01 -1.31240034e+00 1.64311863e-02
-7.33175734e-03 9.32844207e-02 7.18480110e-01 5.75616837e-01
2.21914917e-01 8.96209240e-01 8.98442805e-01 -1.31989205e+00
-1.36117280e-01 -7.10503161e-01 -8.09445739e-01 2.56076396e-01
8.79186988e-01 -7.32289732e-01 -6.24519289e-01 5.72356164e-01]
|
[14.98573112487793, 5.089306831359863]
|
5e12ffc6-b23b-42cf-b5eb-1ad804a44bbe
|
simple-yet-powerful-native-language
| null | null |
https://aclanthology.org/W13-1720
|
https://aclanthology.org/W13-1720.pdf
|
Simple Yet Powerful Native Language Identification on TOEFL11
| null |
['Po-Hsiang Lai', 'Ching-Yi Wu', 'Vincent Ng', 'Yang Liu']
|
2013-06-01
| null | null | null |
ws-2013-6
|
['native-language-identification']
|
['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]
|
[-7.2176055908203125, 3.658822536468506]
|
b29b98c9-a973-4dae-a633-779905c7fcd0
|
a-few-brief-notes-on-deepimpact-coil-and-a
|
2106.14807
| null |
https://arxiv.org/abs/2106.14807v1
|
https://arxiv.org/pdf/2106.14807v1.pdf
|
A Few Brief Notes on DeepImpact, COIL, and a Conceptual Framework for Information Retrieval Techniques
|
Recent developments in representational learning for information retrieval can be organized in a conceptual framework that establishes two pairs of contrasts: sparse vs. dense representations and unsupervised vs. learned representations. Sparse learned representations can further be decomposed into expansion and term weighting components. This framework allows us to understand the relationship between recently proposed techniques such as DPR, ANCE, DeepCT, DeepImpact, and COIL, and furthermore, gaps revealed by our analysis point to "low hanging fruit" in terms of techniques that have yet to be explored. We present a novel technique dubbed "uniCOIL", a simple extension of COIL that achieves to our knowledge the current state-of-the-art in sparse retrieval on the popular MS MARCO passage ranking dataset. Our implementation using the Anserini IR toolkit is built on the Lucene search library and thus fully compatible with standard inverted indexes.
|
['Xueguang Ma', 'Jimmy Lin']
|
2021-06-28
| null | null | null | null |
['passage-ranking']
|
['natural-language-processing']
|
[ 2.44367778e-01 -1.51443586e-01 -4.93795246e-01 -1.61995426e-01
-1.16928864e+00 -6.32032216e-01 1.05387723e+00 4.75371510e-01
-2.82082856e-01 4.62783724e-01 9.31089818e-01 -1.32029220e-01
-8.03992748e-01 -5.40558577e-01 -3.33524615e-01 -5.75255990e-01
-3.08659017e-01 8.12253952e-01 1.58109348e-02 -7.43509412e-01
5.94791889e-01 4.36981201e-01 -1.84515965e+00 8.42653334e-01
4.31644171e-01 9.12963033e-01 2.74063319e-01 4.04397488e-01
-3.11499953e-01 1.19245267e+00 -3.54372144e-01 4.67345081e-02
2.15391815e-01 -3.88472863e-02 -1.20124221e+00 -6.99115455e-01
4.66940641e-01 -1.91102862e-01 -7.43431628e-01 7.00448871e-01
6.76810205e-01 3.88828307e-01 8.16401422e-01 -4.51072514e-01
-7.83545196e-01 8.75415802e-01 -5.14576256e-01 6.07840598e-01
8.47228289e-01 -6.47121727e-01 1.31209183e+00 -9.78400767e-01
1.03983271e+00 1.23731279e+00 6.88050330e-01 3.22964728e-01
-1.04991567e+00 -3.02420437e-01 -7.89160356e-02 5.37926435e-01
-1.53125274e+00 -6.15064561e-01 3.48052174e-01 -3.18732411e-01
1.33166671e+00 6.22121155e-01 2.88917542e-01 1.22477829e+00
-1.84466437e-01 1.09509981e+00 9.68446195e-01 -8.42033267e-01
-1.86475608e-02 1.35050654e-01 5.42548716e-01 4.79873180e-01
2.36055404e-01 2.26162359e-01 -7.35356212e-01 -3.71961594e-01
4.24464256e-01 1.03291981e-01 -3.25446457e-01 -4.76662159e-01
-1.10268581e+00 9.96907711e-01 5.16748607e-01 8.49075377e-01
-3.33404154e-01 -6.10506453e-04 6.18060470e-01 5.64232171e-01
4.95036364e-01 5.45802295e-01 -3.54195803e-01 -1.12564109e-01
-1.39265680e+00 4.23858225e-01 9.78673458e-01 8.18460763e-01
6.04603410e-01 -1.28271624e-01 -4.68665957e-01 1.20055676e+00
4.32300359e-01 1.31601050e-01 9.23532605e-01 -8.26853633e-01
3.74656886e-01 2.90482700e-01 -1.66765794e-01 -1.10210073e+00
-3.12382072e-01 -6.58497155e-01 -7.36265182e-01 -3.01019609e-01
-1.54637337e-01 3.36241633e-01 -8.20928693e-01 1.13493645e+00
-2.51052678e-01 -5.68465330e-02 5.91232702e-02 9.57491398e-01
1.28645849e+00 7.04656482e-01 -9.91428420e-02 -4.00736555e-02
1.22213805e+00 -1.00250411e+00 -6.46785438e-01 -1.90876648e-01
8.12553883e-01 -8.79031062e-01 5.03718078e-01 2.87634373e-01
-1.17907298e+00 -2.72118360e-01 -1.02367651e+00 -4.43152964e-01
-7.81205356e-01 -3.49407978e-02 1.12289941e+00 3.71478587e-01
-1.45356774e+00 6.20056510e-01 -5.63310325e-01 -6.97377443e-01
1.78665459e-01 2.07471982e-01 -3.73888701e-01 -4.54386443e-01
-1.29948103e+00 1.15169132e+00 4.77912128e-01 -1.20720990e-01
-1.07003868e+00 -6.02610409e-01 -8.29285502e-01 4.12690163e-01
3.47335011e-01 -6.70919716e-01 1.30565464e+00 -4.57831681e-01
-1.17057002e+00 1.11834955e+00 -1.51083082e-01 -7.49247074e-01
-1.73196107e-01 -5.13249516e-01 -4.47065055e-01 4.14989442e-01
7.41071105e-02 4.56973910e-01 5.20395219e-01 -1.23470879e+00
-1.48447961e-01 -3.69749784e-01 1.35444626e-01 3.68386388e-01
-3.78635526e-01 3.15419257e-01 -5.62074542e-01 -8.36988568e-01
7.14775771e-02 -8.50410402e-01 -2.48347804e-01 -3.89953732e-01
-8.19061846e-02 -2.69175500e-01 5.06075561e-01 -6.09535038e-01
1.61092246e+00 -1.97814023e+00 4.92598236e-01 5.24803698e-01
2.00552598e-01 2.82462090e-01 -5.39340973e-01 1.08572137e+00
-3.80424023e-01 -6.15834258e-02 4.72729467e-02 -2.88090110e-01
8.77939612e-02 2.03018323e-01 -5.38149655e-01 3.85055184e-01
-3.86434168e-01 8.50180566e-01 -1.17621934e+00 -5.78556061e-01
-1.09999925e-02 8.01941633e-01 -4.93160874e-01 1.08765168e-02
8.45382661e-02 -8.63065496e-02 -5.82378328e-01 9.10971940e-01
2.61620343e-01 -3.09761614e-01 4.24115717e-01 -1.08599864e-01
-3.25890750e-01 6.73397362e-01 -9.87606168e-01 2.39627290e+00
-2.84846038e-01 6.40833557e-01 -2.82547306e-02 -1.27466905e+00
6.39964283e-01 3.24521214e-01 4.56445634e-01 -8.91915798e-01
-1.95145831e-02 3.20881277e-01 -4.74408895e-01 -2.10292011e-01
1.16447234e+00 3.13044578e-01 1.47203818e-01 4.62524325e-01
5.79374492e-01 1.11855954e-01 4.79647785e-01 7.93708742e-01
1.19587612e+00 1.68248013e-01 3.98131967e-01 -5.26254654e-01
2.86576152e-01 -2.16732379e-02 -1.77085862e-01 1.10737443e+00
4.79492068e-01 9.69988286e-01 1.65579215e-01 -2.79617995e-01
-7.86013126e-01 -8.85602176e-01 -3.60804141e-01 1.60176730e+00
-2.39074618e-01 -1.05703187e+00 -2.59024829e-01 -2.10755512e-01
-1.77539006e-01 3.61377299e-01 -7.27395773e-01 2.39207894e-02
-4.83251184e-01 -5.58620930e-01 4.03545022e-01 3.83473247e-01
-1.02254763e-01 -1.23450065e+00 -2.87403524e-01 3.30982991e-02
-1.38959721e-01 -7.89292872e-01 -8.92136693e-02 4.32500601e-01
-9.35437381e-01 -9.21087861e-01 -1.11006796e+00 -7.98708320e-01
2.29864821e-01 6.82148933e-01 1.72280109e+00 3.60276073e-01
-4.83467102e-01 8.06802511e-01 -9.28825080e-01 -1.66900739e-01
5.60528375e-02 5.63078225e-01 -1.51995897e-01 -5.57062447e-01
4.09832895e-01 -5.31008542e-01 -8.74294639e-01 -2.62822866e-01
-1.24434447e+00 -5.70557058e-01 6.62038445e-01 8.19057822e-01
5.75469792e-01 -3.80059808e-01 1.61319971e-01 -1.01426458e+00
8.93640637e-01 -1.02076399e+00 -2.80339330e-01 2.85389185e-01
-7.76621521e-01 2.37467960e-01 -1.27352372e-01 4.85375077e-02
-8.99772346e-01 -2.05137476e-01 -2.36731961e-01 -4.02442127e-01
1.95564270e-01 1.08906543e+00 7.55381942e-01 -2.38653630e-01
8.70015144e-01 3.97512317e-01 -1.32463247e-01 -9.14165258e-01
7.35660017e-01 7.40923345e-01 3.74540091e-01 -7.23228872e-01
6.05519414e-01 3.99197608e-01 -1.71578422e-01 -9.17853832e-01
-9.26917970e-01 -1.05724442e+00 -3.92367780e-01 3.08478653e-01
2.11441085e-01 -1.14913118e+00 -2.90803581e-01 -1.77341953e-01
-8.00063729e-01 1.91843733e-01 -6.50788546e-01 2.05141917e-01
-4.48569655e-01 7.40584016e-01 -7.83318460e-01 -5.18980801e-01
-7.66160846e-01 -7.65361190e-01 1.34898508e+00 3.63722555e-02
-3.54797959e-01 -9.03134286e-01 7.04516292e-01 3.43216807e-01
9.06912446e-01 -1.90123692e-01 9.07259464e-01 -8.01764548e-01
-3.87131006e-01 -2.95757413e-01 -3.26709986e-01 1.63202152e-01
-4.39315438e-01 -4.12489951e-01 -9.37217414e-01 -6.60935760e-01
-2.02889353e-01 -6.57519042e-01 1.31623757e+00 2.61814535e-01
1.19114077e+00 -2.88755417e-01 -4.69380796e-01 5.57227254e-01
1.68235874e+00 -2.41192669e-01 1.11153531e+00 6.21644258e-01
3.69416445e-01 4.30490106e-01 2.45521307e-01 5.50780237e-01
1.64162785e-01 7.42302954e-01 2.44729862e-01 1.18481100e-01
-2.88863271e-01 -1.90742046e-01 1.86600938e-01 1.01952374e+00
-3.02359253e-01 -4.03809883e-02 -7.28703320e-01 7.50472605e-01
-1.89309108e+00 -1.26798916e+00 2.30914041e-01 2.00229168e+00
8.05294514e-01 -3.50716770e-01 -8.47025216e-02 3.45481709e-02
1.94575861e-01 6.55420661e-01 8.04343447e-02 -4.72166508e-01
-1.68284178e-01 9.55042541e-01 2.90549070e-01 4.00656313e-01
-9.31885898e-01 8.74689937e-01 7.65960741e+00 1.10766804e+00
-7.21589088e-01 1.11433834e-01 1.96316004e-01 -2.89249629e-01
-4.91427779e-01 5.65839373e-02 -7.75156558e-01 5.50728180e-02
1.10425842e+00 -1.05027832e-01 6.76218212e-01 8.82988811e-01
-4.41731513e-01 2.13026062e-01 -1.13836360e+00 1.00020039e+00
4.80363488e-01 -1.73019803e+00 2.95437008e-01 -1.85620070e-01
7.81370461e-01 6.74060881e-01 2.46124521e-01 7.45263219e-01
2.76624382e-01 -1.25016177e+00 2.95678526e-01 5.72502851e-01
6.83861673e-01 -3.23036104e-01 6.88293993e-01 4.29655500e-02
-1.09262657e+00 -1.30341411e-01 -5.27646363e-01 1.63266256e-01
-2.29108348e-01 3.09646070e-01 -3.92667621e-01 8.60095024e-01
7.86869049e-01 9.80512202e-01 -6.25643909e-01 1.22495615e+00
1.28414109e-01 3.60045344e-01 -2.78171331e-01 2.43181750e-01
4.98610258e-01 -2.53075529e-02 5.94478130e-01 1.63688672e+00
4.22333717e-01 -1.96089283e-01 1.10831954e-01 4.28311974e-01
-1.88935809e-02 2.96039551e-01 -7.51221895e-01 -1.56069055e-01
2.73692936e-01 1.13878632e+00 -3.35615695e-01 -4.23837602e-01
-4.28572327e-01 7.63243258e-01 4.38533545e-01 4.46927249e-01
-1.27305061e-01 -2.86918372e-01 3.92191261e-01 1.23191103e-01
4.79036659e-01 -1.44649679e-02 2.13183045e-01 -1.52845252e+00
-1.52282417e-01 -1.30119181e+00 8.15373659e-01 -8.39057207e-01
-1.40272474e+00 6.67075217e-01 3.63707542e-01 -1.08749342e+00
-5.36174893e-01 -3.83841813e-01 -1.14239752e-01 7.03651965e-01
-1.95004594e+00 -8.84160340e-01 -1.57114476e-01 8.45463812e-01
5.84098518e-01 -5.02105415e-01 1.52093983e+00 5.38716078e-01
-3.74770910e-02 4.18504089e-01 7.44087756e-01 -7.51155615e-02
5.03178954e-01 -1.09496260e+00 -4.74758595e-02 2.14973241e-01
7.81992316e-01 1.00729394e+00 5.01800299e-01 -2.67908555e-02
-1.51409864e+00 -5.12853742e-01 1.12188268e+00 -4.20285761e-01
8.17671716e-01 1.45313861e-02 -8.99047911e-01 7.70151436e-01
5.48093140e-01 -1.22202942e-02 7.98512459e-01 7.65908301e-01
-7.44289577e-01 1.46841789e-02 -8.55987191e-01 3.13325047e-01
9.69400108e-01 -8.64214182e-01 -1.05411780e+00 8.23386490e-01
3.90483201e-01 -4.00375545e-01 -8.11230242e-01 4.61336046e-01
6.91931069e-01 -8.60192597e-01 1.58129263e+00 -6.32419646e-01
4.37484920e-01 1.16463110e-01 -5.08198619e-01 -1.18482780e+00
-6.12401426e-01 -6.67043507e-01 -5.09709597e-01 9.05397236e-01
3.25442135e-01 -2.88366854e-01 6.44656777e-01 -7.86895156e-02
-7.75202811e-02 -7.22281575e-01 -8.18789959e-01 -4.95766073e-01
1.86687574e-01 -1.83457017e-01 3.28786373e-01 1.05066371e+00
3.80473077e-01 4.19276476e-01 -2.84331232e-01 -2.74499983e-01
4.30865973e-01 2.49764994e-01 2.21010417e-01 -1.34532094e+00
-2.98875242e-01 -3.71246845e-01 -3.92640233e-01 -1.19744015e+00
1.60126597e-01 -1.40394843e+00 -3.43696028e-01 -1.64219844e+00
6.04975343e-01 -3.55694324e-01 -8.96397114e-01 4.78523523e-01
3.58376443e-01 3.74798149e-01 1.05392113e-01 6.85986459e-01
-1.06065214e+00 4.57841814e-01 6.41754329e-01 -4.26217735e-01
5.46439476e-02 -4.98374552e-01 -1.13291597e+00 4.91940528e-01
4.79469270e-01 -4.90818679e-01 -3.86998713e-01 -7.11645246e-01
6.03164315e-01 4.16116305e-02 1.24361636e-02 -9.37629998e-01
3.49199384e-01 4.64996696e-01 1.87753320e-01 -6.69895947e-01
3.65678251e-01 -6.86351597e-01 -1.05961710e-01 1.80364281e-01
-8.00204337e-01 6.20538630e-02 1.63865075e-01 5.13102114e-01
-5.86425602e-01 -5.65251350e-01 3.11645031e-01 -5.28133214e-01
-9.86754000e-01 1.48648262e-01 -4.87965912e-01 7.68426880e-02
2.80784696e-01 1.85248315e-01 -4.37904477e-01 -4.68817681e-01
-8.09978426e-01 1.38071761e-01 -2.25663241e-02 6.13958180e-01
6.52025938e-01 -1.24235034e+00 -9.08945024e-01 -7.22380504e-02
4.52396899e-01 -4.97894108e-01 2.93660074e-01 6.01726472e-01
-4.77672249e-01 1.16126311e+00 7.34281018e-02 -3.28531951e-01
-1.26815522e+00 6.94209278e-01 -9.28631797e-02 -8.88094187e-01
-7.86471069e-01 6.40789330e-01 -1.28024518e-01 -3.23473692e-01
5.02955973e-01 1.71793569e-02 -8.11761141e-01 3.91797930e-01
7.83892035e-01 2.66849458e-01 3.14072788e-01 -5.68143785e-01
-3.41638774e-01 6.28397584e-01 -6.15138531e-01 1.51757792e-01
1.67006409e+00 1.07782250e-02 -3.74335676e-01 2.12291569e-01
1.45975506e+00 -2.50268161e-01 -1.07138984e-01 -4.88597751e-01
1.57220036e-01 -4.41004127e-01 4.85335529e-01 -1.00511396e+00
-9.17511225e-01 6.46427035e-01 8.66250336e-01 1.76707223e-01
9.19820428e-01 1.87063277e-01 6.45963728e-01 9.47414696e-01
4.31515545e-01 -8.39500487e-01 -3.56581472e-02 8.16707194e-01
1.14466262e+00 -1.10245109e+00 5.17945349e-01 -4.17409465e-02
-3.12503040e-01 9.00212884e-01 -1.63323388e-01 -4.93195921e-01
7.74475694e-01 9.06566456e-02 -2.70153254e-01 -6.67109609e-01
-8.11159790e-01 -5.52200198e-01 7.33004391e-01 5.73771596e-01
8.74102771e-01 -2.65274554e-01 -5.97875178e-01 1.78088307e-01
-7.10264891e-02 1.90351471e-01 1.26677041e-04 1.16331124e+00
-4.28385556e-01 -1.27867985e+00 -2.11842805e-01 4.49414521e-01
-7.66774297e-01 -7.56408274e-01 -2.71788120e-01 6.82014108e-01
-4.64001060e-01 6.40896916e-01 -1.54898912e-01 -1.53966561e-01
9.82950106e-02 4.56092842e-02 6.05572820e-01 -9.45995510e-01
-8.61733019e-01 -7.16296732e-02 2.36426190e-01 -8.16786110e-01
-7.24976301e-01 -4.77395147e-01 -5.56792557e-01 -6.02523573e-02
-1.98331609e-01 5.37130833e-01 7.16129482e-01 8.15759003e-01
5.36443174e-01 2.71545082e-01 3.08587849e-01 -1.20082498e+00
-6.08050764e-01 -1.19229138e+00 -6.48172677e-01 2.16837689e-01
1.73543438e-01 -4.66239274e-01 -4.36861515e-01 -2.91961759e-01]
|
[11.465315818786621, 7.599825859069824]
|
44fb00a9-d6a8-4bf1-aab5-7c2fae2ca9da
|
data-might-be-enough-bridge-real-world
|
2303.10828
| null |
https://arxiv.org/abs/2303.10828v1
|
https://arxiv.org/pdf/2303.10828v1.pdf
|
Data Might be Enough: Bridge Real-World Traffic Signal Control Using Offline Reinforcement Learning
|
Applying reinforcement learning (RL) to traffic signal control (TSC) has become a promising solution. However, most RL-based methods focus solely on optimization within simulators and give little thought to deployment issues in the real world. Online RL-based methods, which require interaction with the environment, are limited in their interactions with the real-world environment. Additionally, acquiring an offline dataset for offline RL is challenging in the real world. Moreover, most real-world intersections prefer a cyclical phase structure. To address these challenges, we propose: (1) a cyclical offline dataset (COD), designed based on common real-world scenarios to facilitate easy collection; (2) an offline RL model called DataLight, capable of learning satisfactory control strategies from the COD; and (3) a method called Arbitrary To Cyclical (ATC), which can transform most RL-based methods into cyclical signal control. Extensive experiments using real-world datasets on simulators demonstrate that: (1) DataLight outperforms most existing methods and achieves comparable results with the best-performing method; (2) introducing ATC into some recent RL-based methods achieves satisfactory performance; and (3) COD is reliable, with DataLight remaining robust even with a small amount of data. These results suggest that the cyclical offline dataset might be enough for offline RL for TSC. Our proposed methods make significant contributions to the TSC field and successfully bridge the gap between simulation experiments and real-world applications. Our code is released on Github.
|
['Jianming Deng', 'Liang Zhang']
|
2023-03-20
| null | null | null | null |
['offline-rl']
|
['playing-games']
|
[-3.49026531e-01 -3.68947744e-01 -3.84259313e-01 -1.60682797e-01
-7.54894376e-01 -4.61143404e-01 3.67246866e-01 -1.29212335e-01
-3.34246516e-01 9.13525641e-01 -3.65944952e-01 -7.87781179e-01
-1.41027421e-01 -9.28345501e-01 -7.12349534e-01 -5.89123905e-01
-4.95448411e-01 4.36282516e-01 8.56717646e-01 -6.48773134e-01
8.46382305e-02 8.49165082e-01 -1.83726096e+00 -2.95863777e-01
8.74210835e-01 8.93754065e-01 2.74302036e-01 5.82611144e-01
-1.69936810e-02 7.52943814e-01 -6.30033493e-01 2.30978414e-01
5.56361973e-01 -2.49675721e-01 -2.57957518e-01 -1.53863311e-01
-7.55764619e-02 -2.68487602e-01 -5.16716778e-01 4.92762506e-01
7.22637713e-01 3.05501848e-01 1.08647108e-01 -1.85863972e+00
2.62114078e-01 3.04056793e-01 -5.73604882e-01 1.67884335e-01
2.34260485e-01 8.31272304e-01 7.05239832e-01 -2.73998648e-01
3.15986365e-01 1.04790556e+00 4.34243858e-01 3.64702225e-01
-1.22318995e+00 -1.11929095e+00 4.27550673e-01 3.90029699e-01
-1.47230554e+00 -4.35037613e-01 9.21973467e-01 -1.22307010e-01
7.82680333e-01 2.30986238e-01 9.56053078e-01 7.72081137e-01
1.10286258e-01 8.46562624e-01 1.17044175e+00 -1.40820786e-01
5.96891701e-01 -4.24132720e-02 -1.38935894e-01 3.70023489e-01
2.02984765e-01 6.96998000e-01 -2.37011209e-01 2.00870018e-02
5.71461856e-01 -3.35433304e-01 -5.31508438e-02 -5.42794287e-01
-9.34465706e-01 8.52582932e-01 3.00535917e-01 -1.64031107e-02
-2.85612196e-01 3.59532803e-01 4.18767363e-01 5.93620241e-01
-5.64631931e-02 3.78541231e-01 -2.65286148e-01 -5.83787501e-01
-6.84714973e-01 5.14982045e-01 7.49121487e-01 1.20022547e+00
9.73676562e-01 4.38197285e-01 5.39234318e-02 6.53338969e-01
1.64492577e-01 7.37711668e-01 6.82043061e-02 -1.10450399e+00
6.50488794e-01 3.01201880e-01 4.60358053e-01 -9.05448437e-01
-6.77733898e-01 -3.31247389e-01 -3.11361492e-01 4.60194945e-01
4.78421211e-01 -4.45146322e-01 -2.55644798e-01 1.79269886e+00
3.25452209e-01 6.41472936e-01 4.25224192e-02 7.96586692e-01
3.03693712e-01 6.28930390e-01 -1.06198862e-01 -2.89179534e-01
7.71061838e-01 -7.87368357e-01 -6.60277903e-01 -3.05402249e-01
5.61502576e-01 -5.38143814e-01 1.15351641e+00 5.39720356e-01
-8.16031098e-01 -6.30370200e-01 -1.25532603e+00 7.94396460e-01
-5.17670691e-01 -1.42051175e-01 7.71299124e-01 9.73049045e-01
-1.01938987e+00 2.16330633e-01 -7.40290642e-01 -3.16884428e-01
8.14893246e-02 5.91938257e-01 1.19636849e-01 -4.34263311e-02
-1.41184759e+00 8.53769183e-01 6.37738258e-02 1.44348294e-01
-1.01718163e+00 -6.19474888e-01 -7.88312137e-01 -2.68263966e-01
8.99273038e-01 -1.35701463e-01 1.35224140e+00 -5.40972888e-01
-2.05689645e+00 -3.75496782e-02 -2.84499153e-02 -5.52070498e-01
5.72856665e-01 -6.73472881e-04 -7.68321693e-01 7.20019266e-02
9.28144380e-02 4.94867772e-01 5.72253704e-01 -1.59931827e+00
-9.32695866e-01 2.37040490e-01 2.39460662e-01 -5.70948049e-03
1.48820177e-01 -3.75824690e-01 -5.17965019e-01 -3.27488124e-01
-3.25191110e-01 -1.04985845e+00 -5.84731102e-01 -2.59616911e-01
6.83920756e-02 -8.13768134e-02 1.17245710e+00 -9.14433599e-02
1.45963514e+00 -1.92100549e+00 -5.88121712e-01 4.67599362e-01
-5.90873137e-02 5.46020150e-01 -2.87038803e-01 8.23738337e-01
1.73668981e-01 -1.45619571e-01 3.14376920e-01 1.50829896e-01
7.73410574e-02 5.50810575e-01 -4.05099720e-01 4.21860933e-01
-2.35262737e-02 7.15925694e-01 -1.16526532e+00 -4.96420473e-01
7.62842238e-01 -1.33684218e-01 -6.69848323e-01 1.63614452e-01
-1.59141839e-01 5.88483095e-01 -7.33326137e-01 4.26717371e-01
7.56118298e-01 3.23997259e-01 1.65961444e-01 -7.60801807e-02
-6.24123216e-01 6.98151886e-02 -1.53164852e+00 1.05375922e+00
-8.42099786e-01 7.04621136e-01 2.11549535e-01 -1.11114216e+00
9.42282736e-01 1.31909102e-01 8.33939612e-01 -1.23424733e+00
1.90902561e-01 1.93826929e-01 6.16721287e-02 -5.69141805e-01
5.69183648e-01 1.62054360e-01 -2.51467228e-01 4.91987079e-01
-3.21396112e-01 -5.05321085e-01 5.65141797e-01 7.09225163e-02
1.27577853e+00 -1.85492188e-02 -6.90346397e-03 -8.53337627e-03
5.17332256e-01 -2.40442716e-03 8.45068574e-01 8.69223118e-01
-6.55874014e-01 -3.39252874e-02 4.51240033e-01 -2.89376974e-01
-4.63927656e-01 -9.86553073e-01 2.25791603e-01 7.09276021e-01
5.91209292e-01 -2.24193946e-01 -2.85609037e-01 -5.31505167e-01
1.23897992e-01 8.35966408e-01 -1.22463606e-01 -1.64751455e-01
-6.37018085e-01 -4.23822343e-01 6.33248270e-01 4.55821335e-01
7.07943022e-01 -6.94656730e-01 -7.16573298e-01 5.69862127e-01
-2.05310330e-01 -1.34105599e+00 -3.59763473e-01 1.42270133e-01
-5.66154003e-01 -1.15191209e+00 -7.50576556e-02 -4.00751710e-01
3.05911154e-01 6.97981477e-01 7.34847009e-01 1.80503666e-01
-1.66384242e-02 4.84620154e-01 -3.30152363e-01 -5.62991917e-01
-4.02367175e-01 -1.45645499e-01 2.06553206e-01 4.11022007e-02
1.86678484e-01 -5.95584095e-01 -4.57621932e-01 9.14764106e-01
-5.22482574e-01 -2.94054925e-01 5.63212633e-01 6.30596876e-01
4.32441205e-01 5.50611377e-01 9.77214515e-01 -5.49919367e-01
5.66041410e-01 -3.55660200e-01 -1.04576457e+00 -1.51549965e-01
-7.80080795e-01 -1.14463709e-01 9.36224997e-01 -4.89526510e-01
-8.08964968e-01 -1.21659815e-01 -1.57137141e-01 -3.95893902e-01
-2.25113362e-01 2.74366289e-01 -2.09944293e-01 -2.38172054e-01
5.07497668e-01 3.50074768e-02 2.96131372e-01 3.60948092e-04
1.55087143e-01 7.42590725e-01 1.35889724e-01 -8.46146226e-01
1.04841268e+00 4.61736888e-01 6.80941120e-02 -9.38679874e-01
-5.11293709e-01 -4.97428834e-01 -2.97077864e-01 -7.54410267e-01
3.92648190e-01 -7.85833478e-01 -1.08303773e+00 2.96262681e-01
-4.81172621e-01 -9.12321508e-01 -3.01473171e-01 7.04066634e-01
-8.22682619e-01 1.41552269e-01 -3.71425241e-01 -1.00967705e+00
3.69789720e-01 -1.35209703e+00 7.41506219e-01 3.14279258e-01
7.83158317e-02 -9.16971803e-01 1.11156210e-01 2.44961455e-01
6.94993079e-01 2.09966078e-01 6.12258315e-01 -1.19609833e-01
-7.00141191e-01 -1.22223325e-01 2.78206915e-02 1.04751624e-01
2.57448182e-02 1.02357119e-01 -7.10773289e-01 -4.84865189e-01
-4.61488873e-01 -3.69327635e-01 3.14027786e-01 4.07835215e-01
1.11563587e+00 1.16320223e-01 -4.53012764e-01 1.22168519e-01
1.40604830e+00 7.16087282e-01 7.09604740e-01 6.06015265e-01
1.20479837e-01 3.73030692e-01 1.29345942e+00 5.79625666e-01
6.95607960e-01 9.97904539e-01 4.67556596e-01 -4.13754702e-01
-5.64812124e-02 -5.87079704e-01 6.03653848e-01 8.65638614e-01
9.40514728e-02 -1.44639298e-01 -7.57463396e-01 4.07685369e-01
-2.02993321e+00 -1.12231100e+00 -1.85091674e-01 2.37123227e+00
5.42146206e-01 5.40600181e-01 4.31011856e-01 2.78053045e-01
5.99820554e-01 2.96535492e-02 -6.64144218e-01 -5.09667754e-01
1.09481283e-01 2.17599615e-01 8.05889845e-01 4.02476877e-01
-8.08922231e-01 9.76508975e-01 6.66174555e+00 9.78964984e-01
-1.44171762e+00 -8.26074108e-02 8.24158415e-02 1.92911103e-01
6.77858219e-02 2.54557073e-01 -7.44022965e-01 5.15858293e-01
1.15149784e+00 -3.27743828e-01 5.21103024e-01 7.79842019e-01
1.04016590e+00 -4.51677024e-01 -9.81892109e-01 8.07131946e-01
-4.37559366e-01 -1.06384623e+00 -2.37984702e-01 9.13921669e-02
4.92284417e-01 5.97474836e-02 -1.33588731e-01 9.46910441e-01
6.60167694e-01 -7.97445953e-01 6.29821599e-01 3.90848756e-01
5.59328675e-01 -1.15329313e+00 6.63697958e-01 4.43710089e-01
-1.49574423e+00 -3.44064116e-01 -1.14504538e-01 -1.31706476e-01
2.81904399e-01 3.48141670e-01 -6.47537112e-01 7.52876401e-01
5.20199597e-01 7.67026603e-01 -6.17781639e-01 1.20953846e+00
-2.19008997e-01 1.06050289e+00 -3.72771829e-01 -3.38458508e-01
3.62914890e-01 -2.19009206e-01 5.76395512e-01 9.09988999e-01
3.26214209e-02 -6.72777668e-02 5.94137013e-01 6.31225169e-01
5.83220959e-01 -1.60461232e-01 -7.68547654e-01 3.32075417e-01
8.12976658e-01 1.00486672e+00 -5.18819928e-01 8.66515562e-02
-5.28918564e-01 6.14459030e-02 -1.06768586e-01 6.42176449e-01
-1.25000989e+00 -6.67619109e-01 8.08713734e-01 1.91801697e-01
2.55999267e-01 -4.73813653e-01 -3.19651105e-02 -5.56287289e-01
-2.67111391e-01 -9.76348698e-01 1.34075090e-01 -6.35340869e-01
-8.92738938e-01 3.23778480e-01 3.37887883e-01 -1.91990697e+00
-2.11596996e-01 -3.28486323e-01 -6.39276683e-01 3.16596776e-01
-1.81564367e+00 -7.36032844e-01 -1.99150756e-01 7.46075749e-01
6.29475355e-01 -3.13812822e-01 3.92997205e-01 6.26397908e-01
-8.28368485e-01 5.18086851e-01 -1.61488466e-02 -2.12039024e-01
5.66945136e-01 -9.08083856e-01 -2.66514048e-02 7.48567760e-01
-2.81633794e-01 2.82921851e-01 9.06847715e-01 -2.00267404e-01
-1.83247137e+00 -1.17578924e+00 2.02848479e-01 2.21857559e-02
8.83882999e-01 -3.75516385e-01 -3.76649350e-01 4.19139057e-01
-6.37572408e-02 -7.93277100e-02 3.62773210e-01 -7.43882805e-02
2.87173092e-01 -6.19496226e-01 -1.01256359e+00 9.60218310e-01
7.73422956e-01 -8.18687454e-02 -6.97358251e-02 2.98919771e-02
5.73011041e-01 -4.43207592e-01 -5.74761033e-01 1.72997952e-01
4.22942877e-01 -7.68655062e-01 6.85448885e-01 -8.48153681e-02
-4.10576701e-01 -7.93284535e-01 3.90975587e-02 -1.46792102e+00
6.38501421e-02 -9.49343204e-01 6.02719486e-02 1.02689755e+00
3.93011242e-01 -1.14423835e+00 6.37254417e-01 3.66183877e-01
-2.63240874e-01 -7.51616716e-01 -1.00635993e+00 -1.25995350e+00
-4.70215566e-02 -1.13545108e+00 8.31081808e-01 5.51279068e-01
-1.40690401e-01 1.04371056e-01 -3.32825363e-01 3.21461111e-01
5.30559242e-01 1.55478746e-01 1.48499417e+00 -9.80167091e-01
-2.20461339e-01 -3.01551431e-01 -4.87219572e-01 -1.33390081e+00
2.87811041e-01 -4.72077817e-01 2.18939558e-01 -1.33429945e+00
-4.19937998e-01 -1.06128955e+00 -9.19565931e-02 4.94711578e-01
3.12738329e-01 -2.09124714e-01 3.25402133e-02 -1.05500400e-01
-7.28217483e-01 8.07860017e-01 1.16359973e+00 -1.25728279e-01
-5.21648288e-01 4.71707702e-01 -4.08669174e-01 4.57683414e-01
1.00970316e+00 -3.69361877e-01 -7.34273732e-01 -1.39190136e-02
-1.26941964e-01 3.49731058e-01 2.20936596e-01 -1.35968971e+00
3.80446732e-01 -6.95689440e-01 -3.70157301e-01 -7.40602493e-01
2.20140412e-01 -1.16906679e+00 -4.58819158e-02 6.00319088e-01
6.74130246e-02 -4.69792113e-02 5.29204011e-01 6.83339357e-01
-3.39864850e-01 2.17764676e-01 7.22838879e-01 3.29885423e-01
-9.66458142e-01 1.85497925e-01 -9.12322223e-01 2.94312418e-01
1.32961345e+00 -4.55181032e-01 -2.74969101e-01 -7.79166281e-01
-4.32352781e-01 1.03165293e+00 1.46477997e-01 5.49052775e-01
5.08017719e-01 -1.28370380e+00 -2.79538065e-01 3.12030137e-01
-2.16630008e-02 -2.71578878e-01 6.23451658e-02 1.03029084e+00
-4.66671735e-01 5.28138340e-01 -7.55479932e-02 -7.19882667e-01
-8.91022921e-01 4.18368906e-01 4.08076674e-01 -3.75638604e-01
-5.28052151e-01 6.90013990e-02 -2.67224550e-01 -6.90075636e-01
2.49400645e-01 -3.85809511e-01 -6.31477162e-02 -1.60061136e-01
2.87171811e-01 5.15783191e-01 1.05263159e-01 -5.48166156e-01
-2.44062722e-01 5.56296647e-01 3.42386544e-01 -2.66646773e-01
1.22235608e+00 -2.97300875e-01 4.92960274e-01 4.38136250e-01
8.20907116e-01 1.43111274e-01 -1.42789960e+00 3.27941477e-02
-4.44677770e-02 -5.31283021e-01 2.00861722e-01 -4.71467167e-01
-1.36511242e+00 5.25735676e-01 6.90727532e-01 1.55452145e-02
1.23445547e+00 -4.31535661e-01 9.10182655e-01 4.46712881e-01
1.13393497e+00 -1.54410267e+00 1.03165373e-01 5.86669862e-01
6.20813549e-01 -1.05178022e+00 -1.09486155e-01 -3.80987167e-01
-7.57964432e-01 9.97505605e-01 8.60737622e-01 -6.62542656e-02
8.69082630e-01 5.44743538e-01 1.22466862e-01 -6.18209317e-03
-1.08075309e+00 -6.38749182e-01 -3.97725016e-01 8.00874889e-01
-1.78093284e-01 1.12983607e-01 -3.12233061e-01 4.19270217e-01
-1.74116641e-01 2.61814058e-01 7.18975842e-01 1.19330335e+00
-3.93439054e-01 -1.27913666e+00 -3.95592123e-01 1.81787059e-01
1.58585355e-01 6.17055595e-01 7.02234358e-02 1.25560486e+00
4.75692376e-02 1.59479964e+00 -1.19247459e-01 -5.58393598e-01
8.64095986e-01 -5.08010983e-01 1.52998373e-01 -2.40572304e-01
-3.97722393e-01 -2.22088676e-02 3.53488415e-01 -1.03885615e+00
-2.51324683e-01 -6.54766500e-01 -1.45831347e+00 -4.61393148e-01
-4.12795961e-01 3.41733217e-01 4.78244215e-01 8.80455792e-01
3.80265474e-01 6.35291696e-01 1.21395755e+00 -9.57517147e-01
-4.23834741e-01 -4.61032152e-01 -5.96497476e-01 -1.06535271e-01
2.24099815e-01 -1.14751744e+00 -4.14246202e-01 -4.09927607e-01]
|
[5.256855010986328, 1.376437783241272]
|
e1e08e6d-c33f-471b-8f18-814aedd5ae10
|
morph-kgc-scalable-knowledge-graph
| null | null |
https://content.iospress.com/articles/semantic-web/sw223135
|
https://content.iospress.com/download/semantic-web/sw223135?id=semantic-web%2Fsw223135
|
Morph-KGC: Scalable knowledge graph materialization with mapping partitions
|
Knowledge graphs are often constructed from heterogeneous data sources, using declarative rules that map them to a target ontology and materializing them into RDF. When these data sources are large, the materialization of the entire knowledge graph may be computationally expensive and not suitable for those cases where a rapid materialization is required. In this work, we propose an approach to overcome this limitation, based on the novel concept of mapping partitions. Mapping partitions are defined as groups of mapping rules that generate disjoint subsets of the knowledge graph. Each of these groups can be processed separately, reducing the total amount of memory and execution time required by the materialization process. We have included this optimization in our materialization engine Morph-KGC, and we have evaluated it over three different benchmarks. Our experimental results show that, compared with state-of-the-art techniques, the use of mapping partitions in Morph-KGC presents the following advantages: (i) it decreases significantly the time required for materialization, (ii) it reduces the maximum peak of memory used, and (iii) it scales to data sizes that other engines are not capable of processing currently.
|
['Oscar Corcho', 'María S. Pérez', 'Jhon Toledo', 'David Chaves-Fraga', 'Julián Arenas-Guerrero']
|
2022-08-25
| null | null | null |
semantic-web-2022-8
|
['knowledge-graphs-data-curation', 'data-integration']
|
['knowledge-base', 'knowledge-base']
|
[-4.03594878e-03 5.22973597e-01 -9.97201800e-02 -2.30942607e-01
-2.18917191e-01 -5.54306269e-01 5.55401206e-01 9.30582225e-01
-4.47126210e-01 8.43401313e-01 -9.59060267e-02 -3.12883914e-01
-4.61524904e-01 -1.64322269e+00 -8.34935546e-01 1.55599033e-02
-1.25178784e-01 9.62419987e-01 1.03953731e+00 -4.71039750e-02
7.49521777e-02 6.24746084e-01 -2.12569261e+00 2.94340372e-01
1.20012128e+00 1.06983590e+00 1.29107282e-01 -1.63067132e-02
-8.07356894e-01 8.02550197e-01 -5.45107126e-01 -4.45546061e-01
3.02060574e-01 -2.83170879e-01 -1.25621033e+00 -9.15053114e-02
2.38073856e-01 1.96112931e-01 3.38241249e-01 1.12729359e+00
2.03217063e-02 1.74374640e-01 3.26173961e-01 -1.27672887e+00
-2.05935493e-01 8.54748428e-01 -3.04052323e-01 -2.38403976e-01
4.87395197e-01 -6.04881525e-01 6.65509820e-01 -6.02185071e-01
1.24412906e+00 1.04062390e+00 4.04986441e-01 3.94487143e-01
-1.27243173e+00 -3.05255294e-01 1.20105058e-01 4.54732448e-01
-1.67823780e+00 -4.03859198e-01 4.46278274e-01 -3.26080561e-01
1.21185875e+00 5.28692484e-01 9.39893484e-01 -5.01538329e-02
-5.71423993e-02 2.56575972e-01 9.12095964e-01 -7.02447593e-01
5.47405899e-01 1.40544996e-01 1.84359401e-01 7.10361660e-01
8.84542048e-01 -4.78044748e-01 -4.45489615e-01 -3.58990192e-01
2.70120263e-01 -3.21274430e-01 -9.74033922e-02 -6.99112535e-01
-8.11239004e-01 5.35127461e-01 2.61177301e-01 5.12881637e-01
-3.53466690e-01 5.64198531e-02 3.92100275e-01 1.51797175e-01
2.42311075e-01 3.14920992e-01 -3.19014758e-01 -2.17037741e-02
-9.14741635e-01 3.38601768e-01 1.35790765e+00 1.14902580e+00
1.24034357e+00 -4.04900193e-01 5.08673072e-01 5.55061460e-01
2.39761591e-01 3.12916070e-01 1.26441777e-01 -6.56562626e-01
5.70501626e-01 1.48550308e+00 1.53088436e-01 -1.00364256e+00
-5.60725689e-01 9.01324600e-02 -3.52816880e-01 1.86097726e-01
3.47136050e-01 2.45957226e-01 -7.95541525e-01 1.34498298e+00
5.44685602e-01 -3.20567071e-01 2.16140434e-01 4.54493552e-01
8.48515451e-01 6.12644613e-01 1.00597918e-01 -3.59171927e-01
1.46429551e+00 -5.94107032e-01 -7.02019811e-01 7.24133104e-02
5.92285752e-01 -7.29576886e-01 7.79231787e-01 3.06141764e-01
-1.30993152e+00 -2.61182755e-01 -1.23027790e+00 -3.26977782e-02
-9.79117155e-01 -3.56598675e-01 8.00907314e-01 7.91158020e-01
-1.10120058e+00 4.63225007e-01 -8.91845644e-01 -5.39457560e-01
8.38640481e-02 4.69438732e-01 -4.34298933e-01 1.21060957e-03
-1.03547907e+00 9.84631300e-01 1.12218058e+00 -1.13162361e-01
-1.63390949e-01 -7.94895947e-01 -7.11652935e-01 4.38733488e-01
8.12602937e-01 -6.53846502e-01 7.71732390e-01 -6.75888062e-01
-1.23608935e+00 7.36165762e-01 -4.24798578e-02 -3.98675561e-01
4.37381953e-01 6.19983748e-02 -6.79716408e-01 4.89468575e-02
1.64007351e-01 4.66183096e-01 2.41741210e-01 -1.39063275e+00
-8.81121695e-01 -3.14267337e-01 3.61716151e-01 -8.14606175e-02
-3.87036711e-01 -5.59736975e-02 -9.54212546e-01 -2.00141847e-01
1.48754105e-01 -9.30968702e-01 -1.71603531e-01 -6.20476782e-01
-2.74079770e-01 -3.47433656e-01 7.09399283e-01 -2.29510486e-01
1.56827295e+00 -1.78843129e+00 2.22328186e-01 8.26362252e-01
1.32473474e-02 4.26354438e-01 3.73928457e-01 8.12998593e-01
2.09364220e-01 3.95250827e-01 -1.66382745e-01 1.59342766e-01
1.23392157e-01 3.95704597e-01 -2.95008067e-02 -1.96063127e-02
-1.36967808e-01 5.88170350e-01 -6.41243398e-01 -6.95525467e-01
1.89757302e-01 2.26891130e-01 -6.24358177e-01 -1.60031095e-01
-7.61635065e-01 -2.91467160e-01 -3.45150799e-01 4.45631862e-01
9.49386954e-01 -3.03408653e-01 8.64906847e-01 -2.86772758e-01
-2.26362377e-01 2.72573292e-01 -1.79390466e+00 1.58640254e+00
-3.58441412e-01 -6.98270276e-02 -1.35157779e-01 -7.73020148e-01
9.85344172e-01 1.53036624e-01 5.53100646e-01 -7.73455679e-01
-1.75548926e-01 5.12182176e-01 -3.54883105e-01 -3.59180659e-01
6.59360051e-01 5.18331937e-02 -2.49436259e-01 5.97700000e-01
-5.41531108e-02 -3.02140024e-02 1.16446686e+00 1.42340615e-01
1.09796941e+00 2.03944653e-01 4.75699633e-01 -6.30615115e-01
7.85939217e-01 3.31345201e-01 6.82914913e-01 3.01162064e-01
6.14293873e-01 -1.62455231e-01 7.51971304e-01 -6.66071951e-01
-6.87184632e-01 -1.07422590e+00 2.22755019e-02 6.67254150e-01
3.54325444e-01 -1.14265597e+00 -8.36684644e-01 -7.33758569e-01
1.23676635e-01 6.94346130e-01 -3.28955889e-01 1.58858880e-01
-6.07698023e-01 -6.34289443e-01 2.80752808e-01 3.74851286e-01
3.18641663e-01 -9.98446941e-01 -9.81425107e-01 3.89492124e-01
-1.78371653e-01 -1.23888254e+00 2.18501523e-01 -2.83737220e-02
-9.50724542e-01 -1.48096609e+00 1.26578584e-01 -5.16130328e-01
8.38402867e-01 1.16582468e-01 1.26021612e+00 4.02581811e-01
-1.40947014e-01 3.28270286e-01 -6.79933190e-01 -6.10983312e-01
-4.53325778e-01 3.91812652e-01 -3.11743528e-01 -1.10755742e-01
2.60630906e-01 -6.10352099e-01 -9.74306762e-02 3.64144504e-01
-1.39351368e+00 2.39615709e-01 2.69472599e-01 1.72746554e-01
8.57880712e-01 6.57709599e-01 6.04424834e-01 -1.40674841e+00
4.72549111e-01 -3.74256372e-01 -1.19012165e+00 6.45590127e-01
-9.29331124e-01 2.61845231e-01 7.07474947e-01 1.10484876e-01
-1.16675639e+00 1.51046321e-01 2.28180110e-01 1.34185016e-01
2.05328792e-01 9.66539860e-01 -4.40155834e-01 -7.56076118e-03
4.80757535e-01 -2.49991298e-01 -8.63617510e-02 -6.53719008e-01
3.25950176e-01 2.25279182e-01 3.02287519e-01 -7.31767058e-01
7.29314685e-01 5.51642060e-01 3.74654263e-01 -5.63814402e-01
-2.21369088e-01 -2.97529191e-01 -5.92730582e-01 -2.36935705e-01
7.19247222e-01 -4.50906247e-01 -6.40775859e-01 -4.39543463e-02
-1.10146987e+00 -1.29875168e-01 -5.70566297e-01 2.92149514e-01
-5.28590441e-01 3.40489715e-01 -8.83893445e-02 -3.65894794e-01
-1.79116309e-01 -1.04494512e+00 5.71884811e-01 2.61348486e-01
-1.63706690e-01 -8.44543457e-01 5.19457199e-02 2.13527322e-01
4.67961103e-01 4.45312560e-01 1.33445513e+00 -4.47091341e-01
-9.07573760e-01 -2.92065054e-01 -2.17166662e-01 -1.31146327e-01
1.45825356e-01 1.60089359e-01 -4.26502347e-01 7.11169466e-02
-5.39082825e-01 1.52682945e-01 3.02534461e-01 -2.76440382e-01
8.88261020e-01 -3.29225175e-02 -5.49921334e-01 2.09469110e-01
1.84812701e+00 3.22184443e-01 9.95855391e-01 6.27862751e-01
5.03983498e-01 6.70079708e-01 8.03268671e-01 3.29729259e-01
6.47907972e-01 9.15459096e-01 5.15890002e-01 -1.63016170e-02
-1.93722233e-01 -9.17644575e-02 2.07812106e-03 8.41726899e-01
-5.58019757e-01 -4.29533601e-01 -1.13443613e+00 5.59134305e-01
-2.26423144e+00 -7.54976869e-01 -3.06372881e-01 2.34666276e+00
7.72881150e-01 7.04175532e-02 1.51088119e-01 2.51903594e-01
6.35003388e-01 -1.96094751e-01 -1.25519231e-01 -5.21068096e-01
1.21077135e-01 4.63672161e-01 5.43308496e-01 5.30435443e-01
-7.08991408e-01 9.11974013e-01 5.79402781e+00 5.50586641e-01
-8.62973273e-01 1.42174736e-01 -2.39990860e-01 -1.91983934e-02
-5.05003989e-01 4.17347968e-01 -8.25856984e-01 2.99741745e-01
1.07181513e+00 -5.04417002e-01 4.20572728e-01 7.46251345e-01
-2.49172062e-01 -5.04265428e-01 -1.09156656e+00 6.18960023e-01
-8.96392390e-02 -1.54190361e+00 2.49616861e-01 2.12615058e-01
4.92238194e-01 -2.20944181e-01 -8.11320186e-01 4.42606956e-02
3.45219195e-01 -6.57041430e-01 8.97541821e-01 7.04190254e-01
3.34772676e-01 -1.03464735e+00 7.34358668e-01 8.76124278e-02
-1.50459254e+00 2.34053582e-01 -4.77981597e-01 2.10642532e-01
2.79821992e-01 8.60192299e-01 -7.08077013e-01 1.33381939e+00
8.82976472e-01 1.25133336e-01 -6.88490212e-01 9.15876687e-01
-5.71942069e-02 4.64918017e-02 -6.04755044e-01 4.87721637e-02
-2.10869670e-01 -3.28938454e-01 2.58902162e-01 1.05446970e+00
4.57069606e-01 -2.08195746e-01 1.35445684e-01 6.36436641e-01
-1.75354600e-01 5.97013652e-01 -2.89451331e-01 -4.07288037e-03
5.16673148e-01 9.86681461e-01 -1.31999350e+00 -6.02092683e-01
-5.50360739e-01 2.97213435e-01 4.06265587e-01 4.72757146e-02
-7.63613164e-01 -6.38463199e-01 2.17373028e-01 5.56995749e-01
5.12309968e-01 -1.14425115e-01 2.65321434e-02 -9.37594056e-01
4.80656922e-01 -5.22477627e-01 6.16617858e-01 -5.31206369e-01
-7.27947295e-01 7.54879951e-01 4.74050879e-01 -1.05053759e+00
-2.29651883e-01 -2.82051295e-01 -4.10390832e-02 6.01409495e-01
-1.33890951e+00 -9.89247918e-01 -4.12221998e-01 6.91189051e-01
-1.61144450e-01 2.00770602e-01 1.13366044e+00 6.74358249e-01
-2.13160649e-01 2.71065235e-01 -2.70340890e-01 -3.90003741e-01
5.18804252e-01 -1.13207746e+00 -1.23687545e-02 9.73715961e-01
-1.96621716e-02 6.39964998e-01 4.09070998e-01 -8.40559304e-01
-1.63498366e+00 -1.06760490e+00 1.00806761e+00 3.31647359e-02
5.14048755e-01 -3.49863172e-01 -9.15471911e-01 7.29567528e-01
1.09714717e-01 1.80167556e-01 7.67163336e-01 2.91029721e-01
-4.09788311e-01 -6.47599220e-01 -1.22938955e+00 3.17965984e-01
1.03536999e+00 -1.08318552e-01 -4.12964523e-01 2.50509173e-01
4.47474718e-01 -3.22540820e-01 -1.54102278e+00 4.86550033e-01
4.61913168e-01 -1.03342402e+00 6.86667860e-01 -3.14364851e-01
2.80168392e-02 -8.44650924e-01 -1.64842069e-01 -1.06200886e+00
-9.98297930e-02 -4.12801981e-01 -3.25043201e-01 1.52873123e+00
7.28818834e-01 -8.65626156e-01 6.55246079e-01 7.12656796e-01
-1.03966780e-01 -5.74978948e-01 -8.51092041e-01 -1.06354392e+00
-4.93544728e-01 -3.13427180e-01 1.30775344e+00 8.69308054e-01
3.88929725e-01 1.83193758e-01 6.84234574e-02 2.94249784e-02
3.99789035e-01 5.20323157e-01 9.06857431e-01 -1.73309553e+00
-1.18196003e-01 -2.05651194e-01 -6.45225167e-01 -1.50674149e-01
-1.00785993e-01 -1.09626734e+00 -3.55252087e-01 -2.15361285e+00
-1.04291327e-01 -9.26858842e-01 -6.85596764e-02 7.76304662e-01
1.98864177e-01 -3.47531140e-02 3.95394772e-01 1.89676926e-01
-6.88716173e-01 -5.59787415e-02 8.34830642e-01 3.15308012e-02
-4.32311445e-01 -3.58149678e-01 -5.95667243e-01 6.98315799e-01
6.83474064e-01 -6.84269726e-01 -7.16147959e-01 -2.77586311e-01
7.70298362e-01 -4.33760025e-02 -1.10801846e-01 -1.22860897e+00
3.03784549e-01 -4.40278292e-01 -6.32882863e-02 -6.20293617e-01
1.40628457e-01 -1.24965203e+00 1.12117434e+00 5.86951971e-01
3.50897104e-01 -4.60130200e-02 3.28672051e-01 2.81899899e-01
-3.27014536e-01 -5.98672569e-01 6.03482425e-01 -1.24169670e-01
-8.87693107e-01 3.46027166e-02 -2.63590425e-01 -9.48297232e-02
1.50086641e+00 -1.67866185e-01 -4.89279956e-01 2.74173409e-01
-7.37722695e-01 1.26374245e-01 8.89659703e-01 3.50186050e-01
2.02135354e-01 -1.32369781e+00 -2.18009442e-01 1.64163057e-02
2.11755499e-01 2.02306911e-01 1.71494614e-02 5.77203512e-01
-1.04706335e+00 3.88852447e-01 -3.98720384e-01 -2.69023061e-01
-1.26734769e+00 9.54361618e-01 -3.03857140e-02 -6.45074606e-01
-7.30175078e-01 3.37698720e-02 -2.28025272e-01 -1.80519357e-01
-1.89127520e-01 -3.96597475e-01 -2.68827140e-01 3.72902513e-01
3.08160603e-01 7.39692032e-01 6.15988374e-01 -5.22221148e-01
-6.64653301e-01 6.86907172e-01 6.04478968e-03 5.24101555e-02
1.48626363e+00 1.51699498e-01 -7.48822510e-01 6.74555078e-02
6.25310600e-01 5.50530434e-01 -3.27963412e-01 -7.43313655e-02
3.78737003e-01 -4.14901733e-01 -2.74183214e-01 -6.35343075e-01
-1.12847531e+00 1.61692843e-01 2.36405954e-01 9.06485081e-01
1.37760127e+00 1.42871797e-01 6.46289587e-01 3.38836789e-01
7.97423124e-01 -1.27689075e+00 -6.96920455e-01 2.85150558e-01
7.60214031e-01 -3.76226664e-01 4.16613996e-01 -1.20150840e+00
-1.72967732e-01 1.21214056e+00 4.44732577e-01 5.88158555e-02
8.31737995e-01 4.20304894e-01 -2.89816260e-01 -5.24384320e-01
-5.65494597e-01 -3.83473068e-01 7.65697062e-02 5.59784114e-01
1.31299749e-01 2.43282646e-01 -9.24945354e-01 5.22246063e-01
-1.44180968e-01 4.29256111e-01 6.12214625e-01 1.38672924e+00
-5.52885532e-01 -1.76244962e+00 -4.15919065e-01 3.82701606e-01
-3.71158063e-01 3.66254181e-01 -4.12455797e-01 1.24611759e+00
5.14336407e-01 9.03113902e-01 1.23375237e-01 -9.83971059e-02
8.55735004e-01 8.21446925e-02 7.38741577e-01 -6.86230183e-01
-4.70866501e-01 -1.73578098e-01 6.26127660e-01 -6.61222517e-01
-6.55539513e-01 -5.55133641e-01 -1.63404441e+00 -4.02651578e-01
-2.21187964e-01 5.68385541e-01 7.86034226e-01 6.42231882e-01
7.19320655e-01 5.70329070e-01 -2.97572929e-02 -4.88953739e-01
1.85338870e-01 -5.88810146e-01 -4.77175534e-01 4.76666629e-01
-5.61240494e-01 -9.74262059e-01 2.31261611e-01 3.08082253e-01]
|
[9.019852638244629, 7.7522292137146]
|
4e2aaa2a-6c14-453f-babf-bf074a008b53
|
deeprls-a-recurrent-network-architecture-with
|
2112.05505
| null |
https://arxiv.org/abs/2112.05505v1
|
https://arxiv.org/pdf/2112.05505v1.pdf
|
DeepRLS: A Recurrent Network Architecture with Least Squares Implicit Layers for Non-blind Image Deconvolution
|
In this work, we study the problem of non-blind image deconvolution and propose a novel recurrent network architecture that leads to very competitive restoration results of high image quality. Motivated by the computational efficiency and robustness of existing large scale linear solvers, we manage to express the solution to this problem as the solution of a series of adaptive non-negative least-squares problems. This gives rise to our proposed Recurrent Least Squares Deconvolution Network (RLSDN) architecture, which consists of an implicit layer that imposes a linear constraint between its input and output. By design, our network manages to serve two important purposes simultaneously. The first is that it implicitly models an effective image prior that can adequately characterize the set of natural images, while the second is that it recovers the corresponding maximum a posteriori (MAP) estimate. Experiments on publicly available datasets, comparing recent state-of-the-art methods, show that our proposed RLSDN approach achieves the best reported performance both for grayscale and color images for all tested scenarios. Furthermore, we introduce a novel training strategy that can be adopted by any network architecture that involves the solution of linear systems as part of its pipeline. Our strategy eliminates completely the need to unroll the iterations required by the linear solver and, thus, it reduces significantly the memory footprint during training. Consequently, this enables the training of deeper network architectures which can further improve the reconstruction results.
|
['Stamatios Lefkimmiatis', 'Daniil Selikhanovych', 'Iaroslav Koshelev']
|
2021-12-10
| null | null | null | null |
['image-deconvolution']
|
['computer-vision']
|
[ 2.57298410e-01 1.67280864e-02 3.20469290e-01 -6.96549490e-02
-7.09633708e-01 -1.93823978e-01 4.68572736e-01 -3.44336331e-01
-5.83617151e-01 6.27978444e-01 6.36047050e-02 -3.56594056e-01
-2.43061736e-01 -4.33689624e-01 -7.21382856e-01 -9.25987303e-01
2.71497071e-01 2.62941897e-01 1.25222147e-01 -1.43902004e-01
1.98818728e-01 6.57281160e-01 -1.44988549e+00 4.81868312e-02
9.25755203e-01 1.09245396e+00 5.58722258e-01 3.99846077e-01
1.66344643e-01 1.05647743e+00 -2.02695996e-01 -6.04484193e-02
3.68989915e-01 -4.80641782e-01 -6.22073650e-01 3.61004084e-01
4.58551198e-01 -5.43241739e-01 -1.84563011e-01 9.72490430e-01
6.30588233e-01 2.70479262e-01 5.19728601e-01 -7.13835299e-01
-4.02302235e-01 2.08109021e-01 -4.86709535e-01 -4.64573614e-02
-1.14869904e-02 5.62107414e-02 7.67380238e-01 -9.83616531e-01
4.90110546e-01 7.85069287e-01 7.25159228e-01 3.64592582e-01
-1.64078355e+00 -1.62262231e-01 -2.40343809e-02 2.07006961e-01
-1.37414479e+00 -7.96586752e-01 6.61324203e-01 -4.06586617e-01
8.58066022e-01 2.64666438e-01 5.18891454e-01 7.67327011e-01
-3.33924234e-01 5.58883846e-01 1.18561435e+00 -4.92389858e-01
3.40848684e-01 1.48735464e-01 -6.80959299e-02 6.43475831e-01
8.05963203e-02 -1.59405097e-02 -4.23357576e-01 -2.24601813e-02
9.82047558e-01 -1.03099115e-01 -7.66746223e-01 -4.92335171e-01
-1.02817237e+00 6.74491882e-01 5.60837507e-01 4.86306041e-01
-6.38067842e-01 1.82258829e-01 3.13280709e-02 5.04421145e-02
3.96256268e-01 3.72865289e-01 -2.98455119e-01 3.07794005e-01
-1.38199770e+00 4.28677350e-02 9.00438428e-01 3.35412681e-01
9.28791106e-01 2.66207635e-01 1.49205789e-01 9.65368569e-01
4.66117769e-01 4.04474497e-01 4.99317884e-01 -1.14351070e+00
1.19763337e-01 3.71451735e-01 2.90119410e-01 -1.17062700e+00
-4.84199762e-01 -8.15467000e-01 -1.11968517e+00 4.31806535e-01
4.03470397e-01 -9.14195180e-02 -7.34784544e-01 1.70909476e+00
1.12649389e-01 3.68914753e-01 1.33936763e-01 1.37507415e+00
6.22385561e-01 5.82152963e-01 -4.03390259e-01 -4.68269974e-01
1.19265485e+00 -9.44425166e-01 -7.01780915e-01 -5.30718267e-01
1.96256861e-01 -7.91080952e-01 6.65842950e-01 4.71891463e-01
-1.09866989e+00 -3.36420149e-01 -1.04433084e+00 -1.20348178e-01
6.43819198e-03 6.55261338e-01 4.69000638e-01 2.71567374e-01
-1.46469307e+00 8.42328250e-01 -8.43194246e-01 -8.86241794e-02
1.32312864e-01 4.95688647e-01 -4.11514282e-01 2.14746520e-02
-7.64209092e-01 9.47258294e-01 1.99308038e-01 8.08574080e-01
-8.10497701e-01 -5.48776507e-01 -6.83337092e-01 1.92855164e-01
5.25688589e-01 -6.79547668e-01 9.45182562e-01 -1.21161270e+00
-1.72070730e+00 6.82284653e-01 -4.25993353e-01 -5.24585783e-01
6.87639236e-01 -2.10283175e-01 9.16995332e-02 2.45888829e-01
-1.44746423e-01 3.73221219e-01 1.22179496e+00 -1.48710763e+00
-1.00691713e-01 -3.72912288e-02 -7.74762034e-02 6.00231066e-02
-3.92914325e-01 -1.05913050e-01 -7.52981365e-01 -6.46199644e-01
5.01486301e-01 -8.73701870e-01 -5.30669272e-01 -1.34066271e-03
-3.45892698e-01 3.56396317e-01 3.39951783e-01 -1.01591372e+00
9.56792772e-01 -2.20823002e+00 6.26109481e-01 2.29102790e-01
2.31891260e-01 2.53426552e-01 -1.34948581e-01 2.52404213e-01
-3.02201390e-01 -3.52636814e-01 -6.13965452e-01 -8.29879165e-01
-2.16894701e-01 1.62863538e-01 -3.82727444e-01 7.36216903e-01
1.79887608e-01 7.15403557e-01 -3.47291827e-01 -1.09781645e-01
3.37205231e-01 7.80496955e-01 -5.53947508e-01 3.16615403e-01
-1.34332582e-01 5.69031954e-01 -1.78637896e-02 9.72862020e-02
8.70663404e-01 -4.76311207e-01 4.54461128e-01 -4.15544748e-01
-4.63610142e-01 1.46062851e-01 -1.52128720e+00 1.81226969e+00
-6.05780244e-01 5.81344783e-01 4.70653236e-01 -1.26396441e+00
7.67885149e-01 3.66396993e-01 4.31351572e-01 -7.73185492e-01
9.39925238e-02 5.41526139e-01 -2.35948890e-01 -3.86206269e-01
2.76811182e-01 -2.89693803e-01 5.19395828e-01 5.14590025e-01
9.29267332e-02 1.20182134e-01 1.01976171e-01 -3.29577597e-03
8.28049779e-01 2.32348502e-01 1.31286532e-01 -2.38405064e-01
1.05260336e+00 -1.81631029e-01 5.16625941e-01 5.89499295e-01
2.76032031e-01 7.89658606e-01 4.84576970e-01 -3.47100824e-01
-9.43765640e-01 -6.82443678e-01 1.92358643e-02 6.51169598e-01
-9.14382264e-02 2.10744161e-02 -7.51994073e-01 -1.04079202e-01
-3.53562921e-01 3.73416901e-01 -4.28023458e-01 1.66319564e-01
-8.77184153e-01 -8.56576025e-01 1.47498161e-01 2.07019046e-01
4.67088044e-01 -9.27284718e-01 -5.68790436e-01 7.15217516e-02
-3.86132896e-01 -1.33027625e+00 -2.30943963e-01 2.60520250e-01
-1.04248989e+00 -9.66927826e-01 -1.08672857e+00 -7.70803690e-01
8.51952016e-01 3.45014393e-01 8.71206939e-01 1.68278679e-01
-1.24303937e-01 1.01748899e-01 -6.42959848e-02 2.81281024e-01
-2.23674178e-01 -1.51146621e-01 -6.40377775e-02 4.02513772e-01
-3.11902076e-01 -7.88999319e-01 -5.87735116e-01 1.71428710e-01
-9.19893682e-01 2.42501050e-01 7.34512031e-01 8.72570097e-01
6.95415020e-01 8.43028277e-02 3.08729231e-01 -5.60011446e-01
3.89118999e-01 -1.93892300e-01 -9.98400390e-01 1.90253124e-01
-4.98406738e-01 3.56105745e-01 8.91519547e-01 -2.95946687e-01
-1.03922331e+00 3.76479596e-01 -1.76136434e-01 -4.11657691e-01
1.22143790e-01 5.93091011e-01 7.07483068e-02 -3.88490915e-01
4.77873594e-01 6.04815960e-01 2.21333414e-01 -8.53587866e-01
3.15519542e-01 4.80978787e-01 7.81947374e-01 -1.68241739e-01
8.33518624e-01 6.28937125e-01 1.32896289e-01 -9.55183029e-01
-6.46298766e-01 -4.91558671e-01 -4.47798818e-01 -2.55254328e-01
7.60906756e-01 -1.02598763e+00 -1.05978668e+00 6.84174120e-01
-1.24041331e+00 -4.70487922e-01 -3.91917396e-03 3.68747175e-01
-6.13954365e-01 5.87573171e-01 -5.79957426e-01 -8.96933079e-01
-4.16315317e-01 -1.14765024e+00 7.56950259e-01 8.88944194e-02
1.50495604e-01 -9.58255470e-01 -5.98435067e-02 3.45882505e-01
6.83958292e-01 4.97880913e-02 5.68295956e-01 -1.10289410e-01
-9.49862778e-01 1.21350728e-01 -4.50021684e-01 6.54697537e-01
-1.71391442e-01 -4.79506075e-01 -1.00251710e+00 -4.54435557e-01
5.93258381e-01 -9.35884565e-02 1.21238995e+00 4.77882624e-01
8.82306576e-01 -4.04144198e-01 1.16243303e-01 8.78854156e-01
1.77413452e+00 -4.13798451e-01 6.62784576e-01 3.89294863e-01
5.96963942e-01 5.49254119e-01 1.06123060e-01 4.47380036e-01
3.02750647e-01 8.40054154e-01 6.43210709e-01 -4.37273592e-01
-2.14789584e-01 1.52297392e-01 4.50603515e-01 7.91098177e-01
-2.59920597e-01 3.00702602e-02 -7.27088273e-01 5.38403571e-01
-2.02566338e+00 -7.15122938e-01 -3.30311686e-01 2.37289453e+00
7.55441785e-01 -2.87694573e-01 -9.54768881e-02 2.44246945e-01
4.11183059e-01 1.32550552e-01 -4.98164564e-01 -5.67236468e-02
-2.13601649e-01 2.57276535e-01 5.26125014e-01 6.38679802e-01
-9.76600885e-01 6.90833628e-01 6.02727175e+00 5.01478076e-01
-1.34556401e+00 1.23703048e-01 3.16340387e-01 -7.17282146e-02
-4.32689078e-02 -2.61453670e-02 -3.67702514e-01 2.13371396e-01
8.70024979e-01 2.68082142e-01 1.01083064e+00 4.72943813e-01
5.50090194e-01 -2.04195604e-01 -7.96628475e-01 9.99189377e-01
2.36444965e-01 -1.28929090e+00 -2.09168166e-01 6.09167758e-03
5.74853837e-01 1.67290002e-01 2.04765741e-02 -3.55493695e-01
-1.24662280e-01 -9.36188638e-01 8.02573502e-01 7.47160733e-01
6.09852552e-01 -5.63858628e-01 7.13521481e-01 5.57933867e-01
-6.73438072e-01 -1.66188747e-01 -3.24338406e-01 -1.44303128e-01
3.58839065e-01 7.67473519e-01 -4.53275919e-01 5.48525333e-01
5.09972990e-01 8.12592983e-01 -2.49694273e-01 1.08458424e+00
-3.76963288e-01 3.52759957e-01 -2.34491512e-01 4.47820961e-01
1.22332901e-01 -5.10799408e-01 5.06709337e-01 1.00353789e+00
3.42134804e-01 7.07558319e-02 -1.00224487e-01 1.11485791e+00
7.08960444e-02 8.67877249e-03 -6.12340942e-02 1.57581195e-01
-1.47367716e-01 1.30408585e+00 -5.97416401e-01 -6.94135875e-02
-3.62048537e-01 1.13706160e+00 4.79472607e-01 6.11277997e-01
-5.82407415e-01 -4.41279709e-02 6.13583505e-01 3.00140008e-02
6.88396394e-01 -4.21216071e-01 -2.90325910e-01 -1.37564313e+00
1.90790847e-01 -9.71498132e-01 -8.15460831e-02 -7.87979305e-01
-9.25818741e-01 7.55393624e-01 -6.22835577e-01 -1.04911470e+00
-2.86609650e-01 -7.24176407e-01 -2.90536821e-01 1.20301425e+00
-2.00735140e+00 -8.56304109e-01 -3.89930546e-01 5.84546268e-01
1.27061114e-01 1.17716432e-01 8.63299072e-01 5.37745953e-01
-7.24346340e-01 1.86366752e-01 2.94983178e-01 -3.18120383e-02
5.46933055e-01 -1.05833292e+00 7.67037570e-02 1.22803330e+00
-2.80122440e-02 8.24615121e-01 9.02969956e-01 -2.88159728e-01
-1.50700486e+00 -6.21634483e-01 8.18088233e-01 1.81162804e-01
6.56144559e-01 -1.41811132e-01 -9.92617249e-01 4.42833751e-01
8.15770403e-03 7.27447569e-02 3.13318372e-01 -2.66861796e-01
-2.36538187e-01 -1.72103494e-01 -8.13218772e-01 2.42043003e-01
5.56415319e-01 -6.05407000e-01 -2.56080985e-01 2.30870530e-01
3.02310675e-01 -4.82755810e-01 -5.21458149e-01 1.20528765e-01
3.76888752e-01 -1.26529467e+00 1.14683115e+00 -2.26716921e-02
6.24771595e-01 -5.07537961e-01 4.02942859e-02 -1.12094975e+00
-2.63995796e-01 -5.41377306e-01 -3.13804060e-01 9.25737739e-01
2.97413528e-01 -7.75842249e-01 5.18387973e-01 5.86896002e-01
-5.25679290e-02 -5.95193148e-01 -8.48310471e-01 -5.49424350e-01
-3.67651165e-01 -2.96592504e-01 1.33948699e-01 6.85492635e-01
-4.25705194e-01 1.38204992e-01 -7.92924523e-01 4.11381155e-01
8.36972058e-01 2.45401248e-01 5.11043072e-01 -1.03415167e+00
-7.33377993e-01 -3.49980175e-01 -1.32729486e-01 -1.47283173e+00
2.15742350e-01 -7.73684263e-01 3.96588176e-01 -1.64502013e+00
1.78698123e-01 -3.78651232e-01 -1.42811313e-01 5.87482095e-01
2.67210584e-02 5.25851369e-01 3.11837584e-01 4.09773827e-01
-3.19755048e-01 4.27665085e-01 1.01427901e+00 9.31921452e-02
-2.22255617e-01 -5.09837568e-02 -7.13073194e-01 6.39392555e-01
4.41637218e-01 -3.97371411e-01 -1.69493943e-01 -7.31025219e-01
3.04190844e-01 1.57622546e-01 6.78342044e-01 -8.65276456e-01
4.01338309e-01 2.20115989e-01 2.24215329e-01 -1.87723771e-01
5.16296506e-01 -9.32174385e-01 2.32335165e-01 5.52354515e-01
-2.77775005e-02 -5.03591657e-01 1.48077101e-01 2.62301803e-01
-3.36087853e-01 -4.96918827e-01 9.25202847e-01 -7.11467490e-02
-5.11744976e-01 -1.43026248e-01 -4.13120002e-01 -3.59078735e-01
5.10447204e-01 -5.92596456e-03 -7.80027211e-02 -4.42214370e-01
-6.93233371e-01 4.63520689e-03 4.74350691e-01 -2.08403580e-02
5.40199220e-01 -9.86655176e-01 -7.73365855e-01 2.76649117e-01
-4.43964362e-01 -1.30337298e-01 3.68937135e-01 1.23132694e+00
-5.71120381e-01 3.75448078e-01 3.67081054e-02 -4.96773243e-01
-1.07448483e+00 3.11995029e-01 6.59982443e-01 -2.68526554e-01
-7.46594429e-01 6.82289064e-01 -2.13921480e-02 -2.87426531e-01
1.05127059e-01 -6.13969639e-02 -3.47914994e-01 7.79081509e-02
6.73726678e-01 5.42508662e-01 1.82772398e-01 -8.53153110e-01
-2.71114558e-01 5.53235292e-01 2.01011777e-01 -2.10887492e-01
1.82998776e+00 -2.17903316e-01 -6.83747411e-01 1.17964812e-01
1.17038500e+00 -8.11726004e-02 -1.32476735e+00 -2.75984228e-01
-8.21895823e-02 -2.11540863e-01 5.50446987e-01 -7.80587256e-01
-1.29847813e+00 7.06220388e-01 6.49757326e-01 9.20251384e-03
1.51294386e+00 -3.42281312e-01 6.25562251e-01 3.65559787e-01
1.07974283e-01 -6.88836575e-01 -9.11912844e-02 5.48519373e-01
1.02901447e+00 -1.13142991e+00 1.91023037e-01 -3.63512009e-01
-2.31411234e-01 1.19817781e+00 6.29554167e-02 -2.27940768e-01
3.67071897e-01 1.38947457e-01 1.46678284e-01 -2.44475100e-02
-3.93508136e-01 -2.89566576e-01 4.54275489e-01 1.84807777e-01
4.13312614e-01 -2.87093282e-01 -4.48196739e-01 4.47851121e-01
3.86296362e-02 2.30770126e-01 5.24775326e-01 4.27665532e-01
-3.19903344e-01 -1.06815946e+00 -5.12370348e-01 -1.09518543e-02
-2.74462193e-01 -2.19564199e-01 -1.08575545e-01 4.04891163e-01
-1.78246461e-02 8.99933398e-01 -1.82351574e-01 1.33174330e-01
2.06794634e-01 -1.96403056e-01 3.80108744e-01 -3.12488049e-01
-4.82856899e-01 4.15302277e-01 -2.25551084e-01 -7.71306157e-01
-5.64824104e-01 -6.46095276e-01 -1.25429440e+00 -7.58751109e-02
-4.24265325e-01 2.81059067e-03 8.84367824e-01 1.14052749e+00
2.35654831e-01 6.33392096e-01 4.74771142e-01 -1.25248671e+00
-5.67964613e-01 -8.52243185e-01 -4.62195665e-01 1.28608182e-01
6.41896427e-01 -3.61582369e-01 -5.13727784e-01 4.79652844e-02]
|
[11.681173324584961, -2.5352649688720703]
|
0cff8bb7-8a08-4d9d-8e64-acbd9a58e702
|
towards-characterizing-domain-counterfactuals
|
2306.11281
| null |
https://arxiv.org/abs/2306.11281v1
|
https://arxiv.org/pdf/2306.11281v1.pdf
|
Towards Characterizing Domain Counterfactuals For Invertible Latent Causal Models
|
Learning latent causal models from data has many important applications such as robustness, model extrapolation, and counterfactuals. Most prior theoretic work has focused on full causal discovery (i.e., recovering the true latent variables) but requires strong assumptions such as linearity or fails to have any analysis of the equivalence class of solutions (e.g., IRM). Instead of full causal discovery, we focus on a specific type of causal query called the domain counterfactual, which hypothesizes what a sample would have looked like if it had been generated in a different domain (or environment). Concretely, we assume domain-specific invertible latent structural causal models and a shared invertible observation function, both of which are less restrictive assumptions than prior theoretic works. Under these assumptions, we define domain counterfactually equivalent models and prove that any model can be transformed into an equivalent model via two invertible functions. This constructive property provides a tight characterization of the domain counterfactual equivalence classes. Building upon this result, we prove that every equivalence class contains a model where all intervened variables are at the end when topologically sorted by the causal DAG, i.e., all non-intervened variables have non-intervened ancestors. This surprising result suggests that an algorithm that only allows intervention in the last $k$ latent variables may improve model estimation for counterfactuals. In experiments, we enforce the sparse intervention hypothesis via this theoretic result by constraining that the latent SCMs can only differ in the last few causal mechanisms and demonstrate the feasibility of this algorithm in simulated and image-based experiments.
|
['David I. Inouye', 'Murat Kocaoglu', 'Ruqi Bai', 'Zeyu Zhou', 'Sean Kulinski']
|
2023-06-20
| null | null | null | null |
['causal-discovery']
|
['knowledge-base']
|
[ 4.25682694e-01 6.27199590e-01 -8.08234930e-01 -1.53258830e-01
-1.77714556e-01 -7.68018782e-01 9.17614281e-01 -2.04868123e-01
1.43192306e-01 1.03022826e+00 6.57622695e-01 -8.08291614e-01
-6.35696471e-01 -9.46333468e-01 -1.19074285e+00 -6.78644300e-01
-6.14928961e-01 6.02505982e-01 -1.40976354e-01 2.46202633e-01
2.11532190e-01 2.04323694e-01 -1.17765260e+00 2.66613245e-01
9.15266395e-01 1.40406489e-01 -9.48063806e-02 3.59652132e-01
1.49208322e-01 8.13418388e-01 -2.87008613e-01 -1.17672838e-01
2.67017066e-01 -7.08930075e-01 -9.98040378e-01 -2.15305865e-01
3.85883510e-01 -4.37490255e-01 -2.61836767e-01 1.04200637e+00
-1.11692715e-02 -1.86784461e-01 8.52358401e-01 -1.71046472e+00
-5.22423685e-01 1.25579488e+00 -6.71327949e-01 -2.04043314e-02
4.04099226e-01 7.47787282e-02 1.09419298e+00 -3.05213243e-01
8.35631728e-01 1.77381921e+00 3.12022805e-01 4.18483615e-01
-1.90384567e+00 -9.45820928e-01 3.57356578e-01 1.85276717e-02
-8.38480175e-01 -3.63480657e-01 8.22555602e-01 -5.50360262e-01
4.64389205e-01 3.91264111e-01 4.28676546e-01 1.34481514e+00
2.47555897e-01 5.51009297e-01 1.47929251e+00 -6.08612955e-01
5.01479805e-01 -2.29522705e-01 1.34541273e-01 6.12798810e-01
7.09242284e-01 6.64152384e-01 -6.11722648e-01 -5.65802515e-01
9.25519168e-01 -9.89844874e-02 -4.27985936e-01 -7.66321719e-01
-1.36563635e+00 1.04637778e+00 3.27389330e-01 1.10647596e-01
-2.73801774e-01 5.17139375e-01 7.76631534e-02 3.69408995e-01
1.06966533e-01 5.71084499e-01 -4.27335709e-01 4.81525898e-01
-7.88341045e-01 4.96148169e-01 6.61931634e-01 7.38570154e-01
4.84750003e-01 -6.47894442e-02 4.66679372e-02 7.11035207e-02
3.64528537e-01 5.71714163e-01 1.82442397e-01 -1.13315868e+00
3.46799403e-01 3.84664893e-01 3.24895054e-01 -8.90122354e-01
-2.48887852e-01 -3.09797645e-01 -8.22643340e-01 2.22253054e-01
6.05736017e-01 -3.36527377e-02 -8.52894247e-01 2.34529734e+00
2.25541607e-01 7.21603513e-01 -1.74948748e-03 7.39486933e-01
1.52662158e-01 3.73425663e-01 1.80044904e-01 -8.22928965e-01
9.71052527e-01 -1.66820213e-01 -6.33201480e-01 -8.17663148e-02
5.73017120e-01 -4.00437474e-01 1.09386051e+00 1.42119318e-01
-9.11647916e-01 -3.64440605e-02 -9.07599568e-01 4.03903604e-01
1.12948440e-01 -6.87458754e-01 1.00580156e+00 5.24600863e-01
-7.16607332e-01 5.03869474e-01 -7.98104405e-01 -1.79039210e-01
2.29119197e-01 2.64585406e-01 -3.43223244e-01 -6.06032871e-02
-1.35485494e+00 4.84747738e-01 3.41516644e-01 -4.55131918e-01
-1.35346949e+00 -9.72511888e-01 -6.26295745e-01 1.22235090e-01
7.60115266e-01 -1.07321811e+00 1.04960692e+00 -1.09921157e+00
-9.27160025e-01 6.19854271e-01 -3.72604072e-01 -7.58234560e-01
7.13539183e-01 -9.28764567e-02 -3.17443281e-01 -3.53209563e-02
4.22090560e-01 2.64432788e-01 8.29493642e-01 -1.42936158e+00
-4.66389179e-01 -5.42148292e-01 5.22404432e-01 2.60728691e-02
1.10349104e-01 -3.97421494e-02 1.15537204e-01 -4.62964803e-01
3.20245564e-01 -9.81125772e-01 -2.83588886e-01 -2.97069430e-01
-7.06819832e-01 -1.56880450e-02 7.16607690e-01 -1.37456581e-01
1.19449735e+00 -1.96549308e+00 4.68404479e-02 4.57048863e-01
3.70171487e-01 -4.76600111e-01 2.19960231e-02 3.70566547e-01
-7.81613588e-01 7.45264471e-01 -4.03580338e-01 2.58087516e-01
1.07811481e-01 1.32510856e-01 -1.03976119e+00 8.15703094e-01
-2.22972482e-01 8.38371038e-01 -9.98174548e-01 -4.73646134e-01
9.84312743e-02 -1.75259337e-01 -8.39751899e-01 -1.31841190e-02
-4.77224678e-01 4.03374344e-01 -3.52766305e-01 1.03124715e-01
6.15653455e-01 -2.24239767e-01 8.49310458e-01 8.82426202e-02
-1.85404450e-01 5.07990241e-01 -1.46990263e+00 1.19981468e+00
-2.41619483e-01 3.23409885e-01 -1.87291026e-01 -1.11341143e+00
2.00488135e-01 5.07112741e-01 4.00416464e-01 -5.02303958e-01
-2.29990005e-01 -1.78135689e-02 1.94663808e-01 -2.63379961e-01
-1.11742049e-01 -6.49593472e-01 -2.32142463e-01 7.38812327e-01
-2.89544910e-01 3.27750385e-01 -1.38185963e-01 4.74591881e-01
1.10247457e+00 7.10796937e-02 5.51170051e-01 -6.05816722e-01
-1.54735222e-01 1.69622526e-01 8.87251616e-01 1.06273293e+00
4.46628816e-02 1.15114905e-01 1.00496149e+00 -9.12431553e-02
-9.80645418e-01 -1.58994222e+00 -5.75462393e-02 7.91744292e-01
2.65364975e-01 -1.58272907e-01 -4.98134702e-01 -7.78447390e-01
1.40139148e-01 1.20634854e+00 -9.32415724e-01 -2.62521803e-01
-6.29633665e-01 -7.42855847e-01 4.05706167e-01 3.26777816e-01
3.42023641e-01 -6.55606389e-01 -6.59902215e-01 -1.72373876e-02
-3.17079872e-01 -3.06556195e-01 -2.73878306e-01 1.09242409e-01
-1.05497229e+00 -1.45994580e+00 -1.09817140e-01 -1.81952998e-01
5.83478212e-01 1.73236847e-01 1.11083460e+00 -2.52660394e-01
2.24356264e-01 1.12157933e-01 1.87876493e-01 -2.50116259e-01
-2.98709750e-01 -4.85900015e-01 3.94427925e-01 -1.73335150e-01
-6.37137368e-02 -8.40607524e-01 -4.38977748e-01 3.12817544e-01
-8.41003418e-01 3.24612200e-01 3.70757490e-01 9.08709347e-01
4.62819666e-01 4.69362199e-01 5.73296964e-01 -1.10678196e+00
4.96842802e-01 -6.52347267e-01 -7.77898669e-01 3.36686224e-01
-6.48607314e-01 5.85369647e-01 5.88222325e-01 -6.20782733e-01
-1.31109774e+00 -3.38102467e-02 7.07067609e-01 -3.52056682e-01
-1.92822233e-01 7.22485542e-01 -6.82576060e-01 7.69544303e-01
7.99732506e-01 -2.78672576e-01 -4.15468246e-01 -3.35695446e-01
7.20332265e-01 -6.35293638e-03 5.81811845e-01 -8.24824214e-01
9.25333798e-01 9.62621748e-01 3.70385259e-01 -5.25981069e-01
-5.82832217e-01 4.02191579e-02 -4.74354416e-01 5.59334718e-02
5.80081463e-01 -5.77094793e-01 -1.00681794e+00 -1.78063557e-01
-1.01085377e+00 -5.30094326e-01 -3.67020607e-01 6.72353446e-01
-7.61852324e-01 1.26319647e-01 -8.71251523e-02 -1.06847858e+00
4.50225174e-01 -7.67818689e-01 5.66607893e-01 -3.56422037e-01
-6.38863981e-01 -1.11379623e+00 1.81737110e-01 -1.21285953e-01
-1.83554322e-01 5.43548942e-01 1.47591865e+00 -3.95658702e-01
-7.45966256e-01 2.42095515e-01 1.16377370e-02 -4.88111764e-01
2.65532155e-02 -1.12339575e-02 -6.74687564e-01 -6.39633089e-02
7.64298365e-02 1.22180410e-01 9.62889016e-01 9.40567791e-01
7.94696927e-01 -8.73056710e-01 -9.21220422e-01 3.45832437e-01
1.26189709e+00 3.21102172e-01 4.40200478e-01 1.14689656e-01
4.08131599e-01 8.04384232e-01 4.48796809e-01 1.19416170e-01
1.25633270e-01 6.54342115e-01 5.92354476e-01 -6.72006905e-02
2.16524675e-01 -7.08499730e-01 4.51639980e-01 -8.75944495e-02
-2.03671753e-02 -1.27254114e-01 -9.00992274e-01 7.56099343e-01
-1.94729400e+00 -1.53366363e+00 -4.37245548e-01 2.59599113e+00
9.70967889e-01 2.25425079e-01 1.10306032e-01 6.81131706e-03
8.50974083e-01 -1.59512591e-02 -6.22880757e-01 -2.08834335e-01
-1.62874237e-01 1.29708797e-01 7.40655601e-01 8.65613818e-01
-9.71136808e-01 6.42134905e-01 6.69602156e+00 5.66677034e-01
-8.94759417e-01 2.59588957e-01 4.61014688e-01 -2.16413438e-01
-1.02554667e+00 6.26937270e-01 -3.89749527e-01 4.84364659e-01
8.81545484e-01 -6.74356639e-01 3.44976425e-01 6.91026032e-01
7.09219754e-01 -2.37496555e-01 -1.54676497e+00 2.69751400e-01
-4.95632678e-01 -1.40961826e+00 1.46761015e-01 4.58357394e-01
9.06964183e-01 -5.46742857e-01 1.69446066e-01 -6.46568015e-02
1.17352951e+00 -1.20151651e+00 9.44779456e-01 4.73993152e-01
9.57992911e-01 -6.66208923e-01 2.63853550e-01 5.01366913e-01
-8.00682962e-01 -2.74928361e-01 -1.11487927e-02 -4.45733517e-01
1.31270692e-01 7.51662850e-01 -9.38823283e-01 6.42735600e-01
3.31159502e-01 2.81424642e-01 1.73329890e-01 6.40214562e-01
-5.43355048e-01 1.07108974e+00 -3.92083913e-01 6.21395409e-01
-1.22000866e-01 3.10464278e-02 6.30276501e-01 8.95975709e-01
1.05836578e-01 2.81853914e-01 -3.62779684e-02 1.16907549e+00
-7.75843300e-03 -4.11169767e-01 -1.01929688e+00 1.71219558e-01
5.59267998e-01 3.95286679e-01 -5.74459434e-01 -4.46613044e-01
-1.59139231e-01 4.97681797e-01 -8.48689750e-02 6.25703931e-01
-9.92735386e-01 3.31273526e-01 7.70860612e-01 3.53091270e-01
-1.75099254e-01 -6.88745128e-03 -6.65787280e-01 -1.31785357e+00
-3.49184304e-01 -7.61298954e-01 7.40691185e-01 -6.05751872e-01
-1.19381189e+00 -2.71397024e-01 5.93750358e-01 -8.06609333e-01
-5.98938704e-01 -1.14080727e-01 -7.02314615e-01 9.60246384e-01
-9.65794504e-01 -9.72451687e-01 3.75986516e-01 7.43216932e-01
1.40487462e-01 4.15868282e-01 6.49628639e-01 -2.29296044e-01
-4.42983568e-01 2.32563600e-01 -6.63322508e-02 -2.13139117e-01
6.32587790e-01 -1.38271797e+00 -3.75763029e-02 1.19287503e+00
7.01214448e-02 1.25417876e+00 1.07752264e+00 -1.13382626e+00
-1.18457174e+00 -1.06016386e+00 9.02561724e-01 -4.41226363e-01
8.01119685e-01 -2.87796825e-01 -6.14256918e-01 1.23863673e+00
9.70796421e-02 -4.53676432e-01 4.03476268e-01 6.26506090e-01
-5.27302623e-01 8.34077373e-02 -8.87865007e-01 1.02590370e+00
1.40331578e+00 -4.27731812e-01 -9.38115656e-01 2.18672663e-01
9.24194515e-01 1.06256358e-01 -3.64651620e-01 6.52759850e-01
6.50094509e-01 -8.96209061e-01 1.08660734e+00 -1.22258484e+00
6.57035410e-01 -4.50729221e-01 -2.28941873e-01 -1.10482097e+00
-4.30893838e-01 -4.77965236e-01 -1.71133131e-01 1.12165833e+00
2.78766125e-01 -7.20308542e-01 4.84198570e-01 7.16845274e-01
1.54752821e-01 -3.07238638e-01 -9.85207021e-01 -9.50880408e-01
1.95258573e-01 -5.99794209e-01 6.36093259e-01 1.57330203e+00
1.20212175e-01 4.83237684e-01 -4.83532369e-01 5.53512275e-01
9.33319926e-01 6.45854473e-01 7.24461436e-01 -1.20517564e+00
-5.16109824e-01 -4.28201675e-01 1.15416087e-01 -7.12529600e-01
4.82611030e-01 -8.28503847e-01 -2.04281226e-01 -1.38042867e+00
5.49173713e-01 -5.77567935e-01 -1.27017155e-01 5.88937700e-01
-7.18149990e-02 -2.78211981e-01 -1.23349726e-02 3.24374110e-01
-1.20646015e-01 2.74973720e-01 1.03703022e+00 -5.52121326e-02
-3.68228376e-01 -9.31806788e-02 -9.20353651e-01 1.09052873e+00
6.31293356e-01 -6.42257929e-01 -7.03651845e-01 2.32990924e-02
2.96259880e-01 5.85466743e-01 9.42676663e-01 -2.58738607e-01
-4.26069871e-02 -9.58359540e-01 7.32017756e-02 -2.21827060e-01
-1.83926299e-01 -7.88565397e-01 8.25183988e-01 8.14560950e-01
-7.07053542e-01 -3.15196395e-01 -9.69093516e-02 7.20688999e-01
1.72346905e-01 -7.08253905e-02 5.74827492e-01 -6.63665161e-02
-5.35243630e-01 -7.10539967e-02 -2.31506109e-01 -5.88855520e-02
1.00753224e+00 -5.58789261e-02 -4.89478439e-01 -4.91128594e-01
-7.59431422e-01 1.02408350e-01 4.96121347e-01 7.13330880e-02
3.57042253e-01 -1.21403241e+00 -7.11192250e-01 -5.74892871e-02
-1.25928029e-01 -2.35368609e-01 1.64708957e-01 8.36974919e-01
2.35110298e-01 5.97530425e-01 3.24068181e-02 -3.99852782e-01
-1.16066849e+00 9.91667926e-01 2.87301570e-01 -5.02349138e-01
-3.27655673e-01 3.62314433e-01 1.15951610e+00 -1.51632726e-01
-2.27187306e-01 -2.41013572e-01 1.63767487e-01 -7.47843087e-02
2.21387923e-01 3.54501307e-01 -6.10468686e-01 -2.25795448e-01
-3.86025667e-01 6.82917014e-02 1.76973522e-01 -5.65227687e-01
1.21953130e+00 -1.03518002e-01 -2.62016833e-01 5.13128638e-01
7.27023065e-01 4.34655637e-01 -1.16970038e+00 -1.15278224e-02
1.69291779e-01 -7.31759429e-01 -2.21168354e-01 -9.08281803e-01
-5.22320926e-01 6.01940215e-01 2.78850913e-01 2.90681213e-01
1.14824843e+00 4.28590864e-01 -1.98582441e-01 -2.76704144e-04
5.09870112e-01 -5.08485496e-01 -1.79528758e-01 1.21885344e-01
1.02950382e+00 -6.34030342e-01 2.65780520e-02 -5.48700094e-01
-2.18802974e-01 5.36132038e-01 1.74612731e-01 -1.11539930e-01
3.53445113e-01 1.60807356e-01 -5.20553946e-01 -2.53599286e-01
-9.29515719e-01 -2.51720678e-02 3.94227095e-02 4.93978739e-01
4.80938405e-01 6.40371025e-01 -5.30327976e-01 5.71369946e-01
-4.66266930e-01 3.62026952e-02 5.56855619e-01 5.36311865e-01
-1.57278210e-01 -8.77588987e-01 -7.71826386e-01 4.72348988e-01
-2.84006417e-01 -3.02168727e-01 -3.22276443e-01 1.08661056e+00
1.68252900e-01 1.00712538e+00 9.36480388e-02 -1.41885690e-02
2.12229729e-01 2.74587050e-02 5.56527793e-01 -6.31268442e-01
7.72289336e-02 2.15272456e-01 1.16858319e-04 -6.30630136e-01
-5.14261067e-01 -9.26679969e-01 -1.23829043e+00 -7.07738161e-01
-3.74673635e-01 9.44633484e-02 1.19225256e-01 9.39963877e-01
1.01411663e-01 5.08275390e-01 4.60245520e-01 -1.28465757e-01
-5.26598871e-01 -7.03403056e-01 -5.77470541e-01 4.40107316e-01
3.75984251e-01 -9.32858884e-01 -4.20719832e-01 5.09374499e-01]
|
[8.053357124328613, 5.436249732971191]
|
ea5adbe2-62a4-4162-b2ea-49fe74ad6af2
|
v2v4real-a-real-world-large-scale-dataset-for
|
2303.07601
| null |
https://arxiv.org/abs/2303.07601v2
|
https://arxiv.org/pdf/2303.07601v2.pdf
|
V2V4Real: A Real-world Large-scale Dataset for Vehicle-to-Vehicle Cooperative Perception
|
Modern perception systems of autonomous vehicles are known to be sensitive to occlusions and lack the capability of long perceiving range. It has been one of the key bottlenecks that prevents Level 5 autonomy. Recent research has demonstrated that the Vehicle-to-Vehicle (V2V) cooperative perception system has great potential to revolutionize the autonomous driving industry. However, the lack of a real-world dataset hinders the progress of this field. To facilitate the development of cooperative perception, we present V2V4Real, the first large-scale real-world multi-modal dataset for V2V perception. The data is collected by two vehicles equipped with multi-modal sensors driving together through diverse scenarios. Our V2V4Real dataset covers a driving area of 410 km, comprising 20K LiDAR frames, 40K RGB frames, 240K annotated 3D bounding boxes for 5 classes, and HDMaps that cover all the driving routes. V2V4Real introduces three perception tasks, including cooperative 3D object detection, cooperative 3D object tracking, and Sim2Real domain adaptation for cooperative perception. We provide comprehensive benchmarks of recent cooperative perception algorithms on three tasks. The V2V4Real dataset can be found at https://research.seas.ucla.edu/mobility-lab/v2v4real/.
|
['Jiaqi Ma', 'Bolei Zhou', 'Hongkai Yu', 'Rui Song', 'Xiaoyu Dong', 'Hao Xiang', 'Zonglin Meng', 'Zhengzhong Tu', 'Shuo Zhang', 'Hanzhao Li', 'Jinlong Li', 'Xin Xia', 'Runsheng Xu']
|
2023-03-14
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Xu_V2V4Real_A_Real-World_Large-Scale_Dataset_for_Vehicle-to-Vehicle_Cooperative_Perception_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Xu_V2V4Real_A_Real-World_Large-Scale_Dataset_for_Vehicle-to-Vehicle_Cooperative_Perception_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['3d-object-tracking']
|
['computer-vision']
|
[-3.17330211e-01 5.84053509e-02 -1.87056392e-01 -6.16477728e-01
-8.28047633e-01 -8.79092932e-01 6.79333925e-01 5.08942753e-02
-4.89370942e-01 3.81292641e-01 -4.45339233e-01 -3.92559648e-01
9.05185416e-02 -7.89141357e-01 -9.36258972e-01 -5.96313477e-01
-1.67338759e-01 5.48322499e-01 1.06913197e+00 -7.51610160e-01
1.18845850e-01 5.48711002e-01 -2.24257040e+00 6.84079751e-02
7.73532987e-01 1.30923700e+00 8.53170693e-01 9.33654010e-01
1.77059934e-01 2.61413753e-01 -2.62961745e-01 -1.58247381e-01
5.38210809e-01 3.45386177e-01 -9.84577090e-03 1.18412236e-02
6.60859704e-01 -2.04006657e-01 -3.93111080e-01 9.22928452e-01
5.75624943e-01 1.80141151e-01 5.21898389e-01 -2.15262985e+00
-3.11898142e-01 -2.19484106e-01 -3.73030961e-01 4.02533174e-01
1.89366024e-02 5.48551321e-01 6.02042198e-01 -1.11756110e+00
4.41802293e-01 1.58534896e+00 4.78750080e-01 4.16749954e-01
-9.93422747e-01 -8.05082679e-01 6.60869479e-02 6.77662969e-01
-1.53536904e+00 -6.39418781e-01 5.36619306e-01 -6.82575047e-01
9.20090675e-01 2.05455581e-03 6.54198945e-01 9.01815474e-01
4.50679034e-01 7.44776547e-01 1.11923718e+00 2.28690460e-01
3.39855999e-01 3.50201219e-01 1.94193780e-01 6.26195490e-01
3.38395596e-01 5.31983614e-01 -7.70680010e-01 1.57117620e-01
1.07784763e-01 -1.95320711e-01 1.42287806e-01 -8.52850437e-01
-9.08068955e-01 1.02402627e+00 6.14470303e-01 -4.75086272e-01
-6.13658391e-02 2.87450880e-01 4.16374058e-01 4.17088509e-01
7.95328394e-02 -2.55784094e-01 -4.56305921e-01 3.66235003e-02
-4.48855162e-02 5.94839513e-01 1.19902045e-01 1.45379817e+00
1.14736223e+00 5.35689816e-02 3.12661588e-01 6.56440377e-01
6.18834317e-01 1.45879114e+00 4.02648039e-02 -1.56551600e+00
4.26594645e-01 5.74603915e-01 2.07230285e-01 -8.40045094e-01
-3.77011865e-01 3.69882770e-02 -4.14657414e-01 9.64847565e-01
4.96897548e-02 9.71402526e-02 -8.06474090e-01 1.40143442e+00
4.84917849e-01 -2.46634111e-01 6.09153986e-01 1.00746834e+00
1.24397933e+00 4.52795774e-01 -2.90673021e-02 3.33098292e-01
1.33905208e+00 -9.94320929e-01 -2.35198379e-01 -6.90296590e-01
4.93689328e-01 -6.13777459e-01 7.65869737e-01 1.46498904e-01
-6.73496962e-01 -1.04979825e+00 -1.20856631e+00 -1.58231303e-01
-7.64449716e-01 -1.50116608e-01 4.79474783e-01 5.84768593e-01
-1.05248713e+00 -5.24853826e-01 -5.73509395e-01 -5.34380317e-01
6.21835828e-01 8.33455920e-02 -5.49287856e-01 -4.83713299e-01
-1.18176091e+00 1.11656761e+00 3.66554707e-01 -7.54207149e-02
-1.52942669e+00 -5.84894478e-01 -9.98913050e-01 -4.51808602e-01
5.87320805e-01 -3.73974085e-01 1.14838409e+00 -2.92317748e-01
-7.58660197e-01 1.03345180e+00 -2.85869032e-01 -6.00078344e-01
5.48044145e-01 -9.39860493e-02 -6.50171280e-01 -1.22579016e-01
4.97393966e-01 1.25165915e+00 5.87471008e-01 -1.88459146e+00
-1.29433942e+00 -5.98971903e-01 -9.72998217e-02 3.97925943e-01
3.67387265e-01 -5.67832470e-01 -6.61913216e-01 3.37463766e-01
1.54614836e-01 -1.27838409e+00 -1.95054188e-01 2.42359802e-01
-2.50321310e-02 -6.50330007e-01 1.29458284e+00 1.39507681e-01
1.64570939e-02 -2.50696445e+00 -2.60833979e-01 -2.93937445e-01
3.25706810e-01 1.00834928e-01 -2.61811435e-01 2.40981430e-01
6.75120533e-01 -3.15731883e-01 1.04355207e-02 -5.40275574e-01
2.76617229e-01 7.43927658e-01 -4.45940167e-01 5.05515218e-01
-1.17300436e-01 1.09861887e+00 -8.36472809e-01 -5.88257849e-01
5.11770070e-01 2.74848342e-01 -2.75834471e-01 1.80680200e-01
-2.06363827e-01 3.04921895e-01 -4.55040365e-01 9.76009488e-01
1.23001218e+00 4.69765007e-01 -3.17775309e-01 6.09999783e-02
-5.23992121e-01 -1.30342990e-01 -8.84444356e-01 1.55744445e+00
-1.52105078e-01 1.16609085e+00 4.07985419e-01 -8.42484474e-01
1.23343360e+00 -1.96228847e-01 2.54005849e-01 -1.11287832e+00
-5.92977293e-02 2.80036151e-01 -2.43557915e-01 -5.34129322e-01
8.12089205e-01 2.81060427e-01 -4.11635071e-01 -3.24831188e-01
-3.63427013e-01 -7.76916504e-01 1.06831327e-01 3.85568023e-01
1.01608324e+00 -1.68595761e-01 -1.22619346e-01 -1.71812609e-01
2.76880383e-01 7.18039751e-01 7.67670095e-01 9.56417859e-01
-9.94682610e-01 2.89691776e-01 -3.66046727e-02 -2.15489760e-01
-9.03382123e-01 -1.60167670e+00 -2.81137258e-01 1.10600054e+00
1.07906735e+00 3.58362533e-02 -5.91366775e-02 -3.93115461e-01
8.27071071e-01 7.06774056e-01 -5.22532165e-01 -6.53782561e-02
-2.76441455e-01 -2.47704655e-01 6.10741019e-01 6.21040344e-01
8.70506406e-01 -7.79728711e-01 -1.04700959e+00 3.98966074e-02
-3.47997755e-01 -1.55717540e+00 7.75260255e-02 2.31683657e-01
-4.17352617e-01 -1.09403718e+00 1.29072115e-01 -6.59859121e-01
-4.95883636e-02 1.29866886e+00 1.07922173e+00 -1.72946289e-01
-2.48668239e-01 6.44406021e-01 -3.26727569e-01 -1.16729689e+00
-4.48006421e-01 -2.96390474e-01 4.94236976e-01 -3.87226194e-01
6.90327048e-01 -2.07414791e-01 -5.41240096e-01 8.49982977e-01
-2.15099901e-01 -1.78581268e-01 4.96280074e-01 2.34281510e-01
7.20723271e-01 4.17527631e-02 4.45599943e-01 -6.48577064e-02
-2.55540330e-02 -5.29035926e-01 -1.00004840e+00 -3.12686265e-01
-5.00797033e-01 -7.83780217e-01 1.55959919e-01 -5.86524718e-02
-8.69857907e-01 4.32674050e-01 2.08115011e-01 -4.85218912e-01
-5.26896060e-01 -4.24401835e-02 -3.77523512e-01 -1.10596023e-01
7.97901928e-01 2.94921815e-01 1.49423972e-01 -2.25990191e-02
5.86519599e-01 8.59293938e-01 1.01286983e+00 -1.99079305e-01
1.01627076e+00 8.23849142e-01 1.39743403e-01 -1.16170251e+00
-2.97934443e-01 -7.74735630e-01 -7.03165948e-01 -5.53296089e-01
1.04645860e+00 -1.55288756e+00 -9.42090273e-01 4.23851192e-01
-9.89689350e-01 -6.57196701e-01 -5.75280078e-02 5.11830211e-01
-7.51135468e-01 1.95958465e-01 1.08395867e-01 -9.63739038e-01
1.85191140e-01 -1.30801654e+00 1.14218974e+00 1.37258038e-01
3.70722294e-01 -3.93141180e-01 1.06135644e-02 9.31419492e-01
2.96759546e-01 3.19565088e-03 4.08887982e-01 -5.50171696e-02
-1.10272694e+00 -1.52868102e-03 -4.42607760e-01 1.07784070e-01
-3.16679507e-01 -1.64473698e-01 -1.17680430e+00 -3.73998255e-01
-5.95642984e-01 -6.51327193e-01 1.18054140e+00 2.20695421e-01
5.45977175e-01 6.19932413e-01 -6.55037940e-01 5.25390565e-01
1.33777547e+00 5.56984246e-01 5.24338484e-01 4.03191715e-01
7.08709717e-01 7.28340745e-01 1.25043285e+00 1.07770085e-01
1.21993911e+00 8.27122331e-01 1.23548913e+00 -9.55587253e-02
-2.02224165e-01 -7.96789378e-02 4.04057205e-01 2.56257296e-01
4.51688655e-02 -4.40657377e-01 -1.05130088e+00 8.56635094e-01
-1.91696966e+00 -8.16647470e-01 -5.51609337e-01 1.72825420e+00
-8.16280693e-02 4.30605561e-01 6.07762858e-02 -1.13192186e-01
4.04357076e-01 1.03949361e-01 -9.35839117e-01 -2.39489004e-01
-3.61442864e-01 -8.68640065e-01 1.03162217e+00 6.19097590e-01
-1.17216098e+00 1.11622989e+00 5.36920118e+00 7.14250743e-01
-7.23073840e-01 3.83550256e-01 -3.02507356e-02 6.47035912e-02
-2.07063287e-01 -1.02901809e-01 -1.21895230e+00 3.13717425e-01
8.27103496e-01 -3.14852968e-02 7.93722644e-02 1.26527858e+00
3.41937751e-01 -5.26199222e-01 -7.61306226e-01 1.18842471e+00
-1.43548712e-01 -1.23682320e+00 -4.75121647e-01 2.82491297e-01
4.26625490e-01 1.04100096e+00 3.10589224e-01 5.61856329e-01
8.02441776e-01 -8.47155452e-01 1.09928989e+00 1.59767330e-01
7.84131944e-01 -6.29318178e-01 6.85757935e-01 5.78348756e-01
-1.58452225e+00 -4.34106141e-01 -7.86935925e-01 5.27385958e-02
2.74562418e-01 2.22624123e-01 -9.40575838e-01 3.48916829e-01
1.26328588e+00 7.47142017e-01 -6.23681486e-01 7.43119895e-01
1.04426451e-01 2.39841476e-01 -3.06793064e-01 -9.17143896e-02
3.13851655e-01 -5.26404493e-02 8.67042899e-01 8.79908144e-01
2.88071066e-01 1.40890718e-01 4.81229097e-01 4.62742925e-01
2.77097315e-01 -3.11820358e-01 -1.17621970e+00 4.83956367e-01
8.47010314e-01 1.16754782e+00 -4.36694145e-01 -4.22171772e-01
-5.89969099e-01 4.50798154e-01 9.82651785e-02 2.43797719e-01
-8.98541331e-01 -1.16709203e-01 1.36626911e+00 1.37963116e-01
4.49505210e-01 -8.17758977e-01 -2.26095319e-01 -5.51863968e-01
-6.71733171e-02 -4.00799513e-01 1.95618477e-02 -1.04851639e+00
-8.33645642e-01 6.42774463e-01 1.72408894e-01 -1.27119839e+00
2.34874144e-01 -5.74001074e-01 -1.75768241e-01 5.83945215e-01
-1.85241270e+00 -1.23352206e+00 -1.09847093e+00 6.90349519e-01
8.48142743e-01 -5.49284637e-01 3.58455926e-01 2.47336641e-01
-2.24598676e-01 9.43879262e-02 2.23705750e-02 -4.14310902e-01
8.21711421e-01 -1.09631157e+00 5.97354114e-01 6.81381762e-01
-4.52921629e-01 -2.60202795e-01 8.41747701e-01 -4.87824738e-01
-1.75991631e+00 -1.45023608e+00 4.76362437e-01 -1.03783488e+00
3.95160586e-01 -6.80008054e-01 -6.83041930e-01 6.14391088e-01
6.43334761e-02 3.21976364e-01 3.73416841e-02 -4.73270088e-01
-2.87997693e-01 -5.87119222e-01 -1.07297671e+00 3.36148530e-01
1.37004733e+00 -3.53467941e-01 -3.23282510e-01 2.65437573e-01
9.29990828e-01 -4.04103220e-01 -5.41415393e-01 5.43607831e-01
6.24568343e-01 -1.18445849e+00 1.26609123e+00 8.97249430e-02
-2.18340829e-01 -7.55454004e-01 -1.08758867e+00 -1.12535894e+00
-3.14096779e-01 -1.20571321e-02 1.87374517e-01 7.65588224e-01
2.06418559e-01 -1.02244782e+00 7.61657715e-01 1.29176229e-01
-8.68618250e-01 -1.45403787e-01 -1.31021786e+00 -1.08537424e+00
1.17141930e-02 -9.40676332e-01 5.25616407e-01 2.81321555e-01
-7.09852278e-01 3.24404567e-01 -1.03258260e-01 7.12958992e-01
1.10144866e+00 2.67739654e-01 1.60160315e+00 -1.40603840e+00
2.13084027e-01 -2.78581619e-01 -9.36761856e-01 -1.42139435e+00
8.00539106e-02 -6.58455193e-01 6.36587560e-01 -1.51244259e+00
-9.98452306e-02 -8.26249719e-01 2.00609580e-01 3.81980747e-01
1.65181711e-01 5.16377807e-01 1.86839610e-01 2.62844801e-01
-7.30001986e-01 7.11201906e-01 1.18593752e+00 -5.18956006e-01
-4.22723070e-02 -8.39367658e-02 -4.33296919e-01 4.93519694e-01
9.14301395e-01 -2.38762602e-01 -7.85125852e-01 -4.32134211e-01
-2.14408249e-01 3.50625925e-02 7.70337939e-01 -1.10312378e+00
3.89119387e-01 -4.08680439e-01 1.14257149e-01 -1.43086922e+00
1.03656518e+00 -8.42716455e-01 -7.11299628e-02 6.61122024e-01
2.09347293e-01 5.37135005e-02 4.26284909e-01 8.22665870e-01
-1.21734828e-01 3.72401714e-01 8.95506501e-01 -2.37373170e-02
-1.98920822e+00 4.28255320e-01 -8.06260765e-01 -6.32302836e-02
1.55954933e+00 -3.66586417e-01 -7.47520983e-01 -2.65834570e-01
-4.05051559e-01 9.51961040e-01 5.78420758e-01 8.36808085e-01
8.96954715e-01 -1.27679443e+00 -8.15232873e-01 3.59487325e-01
8.53502274e-01 2.51455933e-01 4.48919833e-01 6.88490987e-01
-2.62022793e-01 6.80311561e-01 -4.12668228e-01 -1.36398458e+00
-1.62822354e+00 7.83554256e-01 2.10776672e-01 8.32905769e-01
-6.60509050e-01 8.05923641e-01 4.51593161e-01 -7.20494449e-01
-7.00834244e-02 -1.37163149e-02 -1.55930459e-01 -1.57844856e-01
3.39140743e-01 5.36756992e-01 7.85588566e-03 -1.30163240e+00
-8.00157607e-01 7.00934350e-01 3.69791687e-01 -2.83018611e-02
7.66912162e-01 -8.83568943e-01 4.26472127e-01 4.19929713e-01
1.03355515e+00 -2.11049542e-01 -1.59363270e+00 -1.61412060e-01
-4.70044643e-01 -6.04809165e-01 1.07210733e-01 -2.72409946e-01
-7.57990003e-01 9.04792190e-01 1.20900357e+00 2.73898631e-01
6.59292638e-01 4.53820676e-01 6.08682096e-01 6.51273608e-01
8.48007321e-01 -1.03642714e+00 1.22895159e-01 7.07381785e-01
8.32458854e-01 -2.09068155e+00 -2.79075980e-01 -6.32063091e-01
-9.60974872e-01 5.75983226e-01 9.83956754e-01 1.68512121e-01
5.50325751e-01 1.84364408e-01 7.70784616e-01 -3.84363115e-01
-9.59218502e-01 -6.85097873e-01 -3.31252106e-02 1.21672928e+00
-6.16917133e-01 5.02330601e-01 3.97830904e-01 1.51660711e-01
-3.03198457e-01 -5.17658770e-01 4.89576489e-01 8.89577746e-01
-1.00161302e+00 -5.12433350e-01 -5.18856525e-01 -1.12402095e-02
6.28817976e-01 5.01465619e-01 -1.08303078e-01 9.44685161e-01
6.86996400e-01 1.57550347e+00 2.30482116e-01 -6.76340461e-01
5.47831059e-01 -4.27956164e-01 3.36957723e-01 -2.41693482e-01
4.14454818e-01 -3.38110149e-01 9.90109146e-02 -7.88130701e-01
-2.22013474e-01 -8.48207831e-01 -1.38755190e+00 -5.91824174e-01
1.67955145e-01 -8.17792863e-02 1.09674919e+00 6.20456338e-01
5.14092803e-01 2.89803803e-01 7.03287661e-01 -1.16587734e+00
-1.01476692e-01 -6.38343871e-01 -5.88913620e-01 8.91365707e-02
7.47178674e-01 -1.28896868e+00 -2.75283903e-01 -6.05031811e-02]
|
[7.801602840423584, -1.9075980186462402]
|
9bf3cf58-fcca-42af-beed-f52e463d3a1d
|
context-aware-pretraining-for-efficient-blind
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Wang_Context-Aware_Pretraining_for_Efficient_Blind_Image_Decomposition_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Wang_Context-Aware_Pretraining_for_Efficient_Blind_Image_Decomposition_CVPR_2023_paper.pdf
|
Context-Aware Pretraining for Efficient Blind Image Decomposition
|
In this paper, we study Blind Image Decomposition (BID), which is to uniformly remove multiple types of degradation at once without foreknowing the noise type. There remain two practical challenges: (1) Existing methods typically require massive data supervision, making them infeasible to real-world scenarios. (2) The conventional paradigm usually focuses on mining the abnormal pattern of a superimposed image to separate the noise, which de facto conflicts with the primary image restoration task. Therefore, such a pipeline compromises repairing efficiency and authenticity. In an attempt to solve the two challenges in one go, we propose an efficient and simplified paradigm, called Context-aware Pretraining (CP), with two pretext tasks: mixed image separation and masked image reconstruction. Such a paradigm reduces the annotation demands and explicitly facilitates context-aware feature learning. Assuming the restoration process follows a structure-to-texture manner, we also introduce a Context-aware Pretrained network (CPNet). In particular, CPNet contains two transformer-based parallel encoders, one information fusion module, and one multi-head prediction module. The information fusion module explicitly utilizes the mutual correlation in the spatial-channel dimension, while the multi-head prediction module facilitates texture-guided appearance flow. Moreover, a new sampling loss along with an attribute label constraint is also deployed to make use of the spatial context, leading to high-fidelity image restoration. Extensive experiments on both real and synthetic benchmarks show that our method achieves competitive performance for various BID tasks.
|
['Yi Yang', 'Yifan Sun', 'Ruijie Quan', 'Zhedong Zheng', 'Chao Wang']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['image-reconstruction']
|
['computer-vision']
|
[ 5.60413003e-01 -2.79288799e-01 -1.31449506e-01 -2.49178484e-01
-9.08485591e-01 -2.62472749e-01 3.65899473e-01 -2.78656453e-01
-1.72151700e-01 4.51071501e-01 2.68210351e-01 -2.94505179e-01
-1.78932473e-01 -5.22491634e-01 -7.35792100e-01 -1.19564080e+00
2.64365762e-01 -2.49005869e-01 1.49449125e-01 -6.27019629e-02
1.53060809e-01 3.11038196e-01 -1.66991282e+00 4.79757220e-01
1.09818196e+00 1.45235860e+00 5.35076439e-01 5.25313437e-01
-6.38233349e-02 9.35573101e-01 -3.17301333e-01 -4.23393250e-01
5.42792916e-01 -3.55948776e-01 -5.82873821e-01 6.65779173e-01
3.59092265e-01 -4.01884824e-01 -3.67419809e-01 1.32623839e+00
5.51487803e-01 4.06012796e-02 2.49062717e-01 -1.06637573e+00
-4.63441044e-01 2.60011375e-01 -8.27506721e-01 9.42980498e-02
8.69072229e-02 3.03304285e-01 9.24130201e-01 -9.46300387e-01
2.73257285e-01 9.39848244e-01 5.43976247e-01 3.10621947e-01
-1.39491773e+00 -5.43621004e-01 3.95770282e-01 3.58004361e-01
-1.29044461e+00 -7.26715088e-01 9.90975738e-01 -3.50405008e-01
4.62496996e-01 2.87370890e-01 4.25084174e-01 9.55585659e-01
-2.21587643e-02 8.03472102e-01 1.37589765e+00 -3.67569685e-01
2.39595309e-01 -4.49147224e-02 -2.03089812e-03 5.68189144e-01
1.67668849e-01 1.90380856e-01 -4.67207611e-01 9.43486989e-02
6.95669651e-01 1.43669233e-01 -8.07796478e-01 -3.80263090e-01
-1.06040287e+00 4.21582729e-01 3.59947056e-01 2.54922897e-01
-4.04682100e-01 -1.70280352e-01 2.67637163e-01 3.54622126e-01
4.00880218e-01 -1.61009878e-02 -4.00059640e-01 4.06998247e-01
-9.80882764e-01 -3.52956690e-02 4.71919715e-01 7.88870990e-01
1.03842962e+00 2.16503683e-02 -2.25863501e-01 1.04042780e+00
1.91140428e-01 4.03737634e-01 5.44953585e-01 -8.83234739e-01
5.62460423e-01 4.32421446e-01 -1.33302212e-02 -1.09353745e+00
-2.16062590e-01 -7.13240743e-01 -1.23263013e+00 2.97314793e-01
1.99687034e-01 4.86978814e-02 -1.13905430e+00 1.76059163e+00
3.24563116e-01 4.68779176e-01 -6.69426750e-03 1.21540940e+00
5.44744074e-01 4.80268180e-01 -2.19314858e-01 -4.99506712e-01
1.52183497e+00 -1.01584196e+00 -7.43647516e-01 -2.69861311e-01
1.72140375e-01 -9.03657913e-01 8.98546517e-01 5.68711460e-01
-1.01027799e+00 -5.86678028e-01 -1.16169393e+00 -1.31944329e-01
6.78786263e-02 3.09192181e-01 5.13001621e-01 5.60112774e-01
-9.73316789e-01 3.64169508e-01 -6.30323529e-01 -1.09798469e-01
3.63151431e-01 3.17350596e-01 -5.08747876e-01 -4.32675600e-01
-7.93539703e-01 3.32769960e-01 1.46658167e-01 2.53444970e-01
-8.97628069e-01 -3.80013049e-01 -8.28523934e-01 5.59413545e-02
6.80016339e-01 -8.42294872e-01 9.67682064e-01 -1.17704713e+00
-1.45528769e+00 6.94592118e-01 -3.66973728e-01 -2.52158254e-01
4.47377443e-01 -6.97638616e-02 -5.94145119e-01 2.66253471e-01
1.84345439e-01 2.32432887e-01 1.41518426e+00 -1.72091758e+00
-8.18465412e-01 -3.79879653e-01 1.54553443e-01 3.74863863e-01
-4.49417830e-01 -1.76273793e-01 -1.02730262e+00 -1.03413129e+00
5.71275115e-01 -5.95211506e-01 -1.88820213e-01 6.61724880e-02
-6.48122609e-01 3.93986136e-01 6.39790356e-01 -8.66120577e-01
1.19023895e+00 -2.45272136e+00 2.74836749e-01 2.49155089e-01
4.09114748e-01 1.39711887e-01 -1.82079941e-01 3.10920719e-02
-2.60890871e-01 -3.02124739e-01 -5.54773033e-01 -6.29063129e-01
-1.74844965e-01 3.11291128e-01 -4.28914130e-01 4.69450712e-01
1.92592964e-02 5.36290348e-01 -6.50754452e-01 -5.50245404e-01
1.58499733e-01 3.94395500e-01 -5.48698068e-01 3.31295013e-01
5.02278395e-02 5.49134433e-01 -1.95798799e-01 7.61086643e-01
1.00939107e+00 -4.31635052e-01 4.09494579e-01 -6.31579340e-01
-3.67100835e-02 9.99462008e-02 -1.38741338e+00 1.83920169e+00
-5.63046038e-01 1.88269839e-01 5.45360386e-01 -1.12663794e+00
5.95541835e-01 3.98086399e-01 4.80460316e-01 -9.07801867e-01
3.40491869e-02 2.85446674e-01 -7.74170011e-02 -5.26804447e-01
3.06904376e-01 -1.06539652e-01 2.85278380e-01 2.93224335e-01
-2.36580684e-03 3.67348760e-01 -1.91255286e-02 -1.24102680e-03
1.07162344e+00 3.01512983e-02 8.33237767e-02 -8.13840553e-02
8.00011575e-01 -3.80048990e-01 1.11653721e+00 5.06099880e-01
-1.35095850e-01 8.98079097e-01 2.87172049e-01 -3.00665200e-01
-7.20755696e-01 -1.02349997e+00 2.79078353e-02 8.59868228e-01
5.81252992e-01 -3.91792059e-01 -6.83734596e-01 -6.38505220e-01
-2.67725170e-01 1.85686335e-01 -4.46457565e-01 -7.45374411e-02
-5.16711414e-01 -1.06481075e+00 5.45118488e-02 2.97801375e-01
7.57968724e-01 -5.59520245e-01 -2.67522573e-01 4.31923941e-02
-5.19043863e-01 -1.24532330e+00 -6.73241615e-01 2.12105080e-01
-6.73782945e-01 -1.08254361e+00 -5.66759825e-01 -8.26862872e-01
8.22878182e-01 7.78492033e-01 7.60354519e-01 2.85242260e-01
-1.05530664e-01 2.04515055e-01 -3.47031981e-01 1.91729218e-01
-2.91402470e-02 -2.98076749e-01 -3.13889310e-02 7.32674778e-01
-6.11351281e-02 -9.28306997e-01 -9.43610132e-01 3.94034296e-01
-1.13542116e+00 2.94612586e-01 1.09602880e+00 1.18701315e+00
8.65823448e-01 5.51233590e-01 3.35372180e-01 -5.90373158e-01
2.79279113e-01 -2.44649380e-01 -4.59086508e-01 3.59478384e-01
-6.78525448e-01 -5.40786348e-02 6.78100288e-01 -3.48018557e-01
-1.24437118e+00 2.61342108e-01 4.35869247e-02 -5.97285032e-01
-8.03240314e-02 2.96157718e-01 -8.25531006e-01 -1.43710837e-01
2.86385626e-01 6.17218256e-01 -8.36530328e-02 -7.92623043e-01
2.26725504e-01 6.46888852e-01 8.48476827e-01 -5.59929013e-01
1.00956190e+00 6.84411466e-01 -9.69440192e-02 -6.36949420e-01
-6.97177827e-01 -4.45625663e-01 -4.16567713e-01 -5.58798425e-02
7.50958025e-01 -1.10358119e+00 -6.74242914e-01 6.82899594e-01
-9.18702602e-01 -2.26667956e-01 -2.25890666e-01 3.16909552e-01
-4.17004824e-01 6.36156201e-01 -6.20039761e-01 -6.69389725e-01
-2.06448033e-01 -1.40691316e+00 9.34118271e-01 1.02172978e-01
5.47409832e-01 -5.66941440e-01 -4.19903189e-01 5.77636421e-01
3.49132210e-01 -1.94631442e-01 8.90127599e-01 -1.93188146e-01
-9.03492093e-01 9.70291793e-02 -6.51900351e-01 5.83583117e-01
3.11809182e-01 -6.57399535e-01 -1.27356601e+00 -3.92958581e-01
3.60476673e-01 3.96262202e-03 1.07447445e+00 2.25789770e-01
1.42390811e+00 -3.74545187e-01 -1.77930757e-01 9.44309652e-01
1.42315912e+00 1.15112931e-01 7.09892154e-01 2.94914484e-01
8.43531251e-01 5.02583444e-01 4.03437734e-01 3.73265982e-01
5.58645189e-01 8.86995196e-01 4.35499579e-01 -3.79111886e-01
-6.02357388e-01 -1.05760083e-01 3.89636040e-01 8.29214692e-01
5.26658595e-02 -1.90817356e-01 -5.15947163e-01 4.61948007e-01
-1.80436707e+00 -7.58244991e-01 1.15748607e-01 2.37505150e+00
9.35023844e-01 -8.18759128e-02 -1.70394555e-01 4.91218865e-01
7.32262373e-01 2.37012818e-01 -5.52789807e-01 2.54524440e-01
-3.90713781e-01 -5.70746735e-02 4.55908626e-01 4.37386662e-01
-1.22196496e+00 4.65821832e-01 4.87481260e+00 1.07947659e+00
-1.22640920e+00 2.73869634e-01 8.03745866e-01 9.46509242e-02
-2.24720150e-01 1.31453395e-01 -4.01150465e-01 6.05540812e-01
3.27870816e-01 2.39840075e-01 6.08612955e-01 3.99647206e-01
1.26717135e-01 -1.05575502e-01 -8.29639256e-01 1.12757587e+00
9.33559909e-02 -1.02676058e+00 -2.38032993e-02 1.21507585e-01
4.97248083e-01 -3.28152061e-01 1.46254316e-01 -1.68564618e-01
1.08392894e-01 -6.86674118e-01 7.75270045e-01 6.29505932e-01
9.26495314e-01 -6.59473896e-01 6.03853524e-01 2.43166089e-01
-1.30513644e+00 -3.49424571e-01 -9.60633159e-02 2.83418089e-01
1.51744857e-01 9.49632943e-01 -3.02197456e-01 1.12455773e+00
9.09175873e-01 7.54522443e-01 -4.28797960e-01 1.12114596e+00
-3.87080580e-01 4.42127138e-01 -1.06879376e-01 8.88317287e-01
-1.80364832e-01 -1.76278546e-01 6.89282775e-01 9.95818675e-01
2.61407703e-01 4.98386733e-02 3.93031508e-01 3.94116908e-01
-2.63194311e-02 -1.64234657e-02 -1.21816210e-01 4.02430147e-01
3.76795858e-01 1.32597852e+00 -6.20034933e-01 -1.31623074e-01
-6.73860669e-01 1.39435279e+00 -1.75035298e-02 5.68612635e-01
-5.66145658e-01 -2.61859596e-01 8.13286841e-01 8.69364440e-02
4.92849946e-01 -2.10173074e-02 -4.11985695e-01 -1.48260379e+00
4.71884787e-01 -1.16390193e+00 3.31264019e-01 -5.89881003e-01
-1.46745837e+00 7.33800709e-01 -5.17507255e-01 -1.50523341e+00
2.00848386e-01 -4.53877002e-01 -3.90874654e-01 9.47839856e-01
-2.02863836e+00 -1.40581369e+00 -4.13099051e-01 8.82474422e-01
4.66189027e-01 5.80772664e-03 4.84506667e-01 7.60680020e-01
-1.01163900e+00 7.67386258e-01 -5.92370667e-02 1.40049830e-01
8.50559115e-01 -1.09302473e+00 -5.73153608e-02 1.39726937e+00
-1.36573492e-02 5.24899721e-01 5.18728971e-01 -5.32169521e-01
-1.54856551e+00 -1.15998721e+00 4.95255232e-01 1.61263168e-01
4.60041046e-01 -3.22992772e-01 -1.00734890e+00 3.52332830e-01
-5.30221835e-02 3.55243623e-01 4.75114316e-01 -1.00141011e-01
-6.39598906e-01 -5.31048000e-01 -9.90650773e-01 5.00984013e-01
1.07568955e+00 -6.92295730e-01 -1.91548273e-01 2.83059627e-01
7.35248625e-01 -3.93191040e-01 -5.94421864e-01 4.72369045e-01
3.48537385e-01 -1.19969153e+00 1.12973464e+00 -7.30328560e-02
3.59043270e-01 -7.87083268e-01 -5.70205212e-01 -9.61153746e-01
-3.15506577e-01 -6.68154895e-01 -2.35662282e-01 1.39407527e+00
9.10086408e-02 -6.06362522e-01 5.29507458e-01 4.27507997e-01
-2.99669892e-01 -7.44825840e-01 -8.40368450e-01 -8.18241894e-01
-4.75444198e-01 -4.61745501e-01 6.97029591e-01 9.38645601e-01
-4.39913839e-01 1.75088331e-01 -7.62414455e-01 6.27031446e-01
7.81231999e-01 4.02267307e-01 6.58820748e-01 -1.04773939e+00
-7.04973340e-01 -3.43691111e-01 -3.65410566e-01 -1.30469489e+00
-1.19400479e-01 -6.56142056e-01 2.17908561e-01 -1.34254110e+00
3.28790545e-01 -5.57759106e-01 -4.37791973e-01 5.99782526e-01
-3.47955197e-01 3.34560752e-01 5.05946539e-02 5.06786764e-01
-5.80779135e-01 5.99888742e-01 1.13229871e+00 -1.83483765e-01
-2.91378498e-02 -3.52428444e-02 -9.56480324e-01 6.00106061e-01
4.92652893e-01 -2.71688014e-01 -6.26120806e-01 -7.01990306e-01
-1.05292685e-01 1.28951788e-01 5.94403803e-01 -1.00708055e+00
4.54544067e-01 -6.42472804e-02 3.29198569e-01 -2.57324517e-01
4.46361452e-01 -1.03954720e+00 -2.69827712e-02 1.14797689e-01
9.36516598e-02 -2.04418898e-01 -4.33249585e-02 8.63995790e-01
-5.16131222e-01 9.53036174e-02 8.40316474e-01 9.37389731e-02
-7.27961719e-01 5.13462842e-01 -4.23453301e-02 -3.45631063e-01
7.91187108e-01 -2.13927463e-01 -3.64134878e-01 -3.25255692e-01
-5.59054315e-01 1.34406000e-01 6.85629606e-01 1.85823396e-01
6.51305974e-01 -1.17384791e+00 -5.60549796e-01 5.17190576e-01
1.30766243e-01 -2.23565996e-02 7.51330972e-01 1.09907556e+00
-3.22953351e-02 -4.94550401e-03 1.19404644e-01 -5.24419546e-01
-1.11995149e+00 7.64304817e-01 4.17084455e-01 -2.34052658e-01
-9.51377153e-01 8.21649075e-01 7.32611358e-01 -2.10286155e-02
2.77935565e-01 2.25266833e-02 -1.18524425e-01 -1.04114011e-01
7.83053279e-01 1.14394628e-01 3.21281165e-01 -7.17448771e-01
-1.73690692e-01 4.85256702e-01 -2.29383662e-01 -9.58772078e-02
1.30695343e+00 -6.45279408e-01 -3.12703907e-01 4.91683632e-02
1.00783789e+00 1.24399699e-01 -1.44444394e+00 -6.55334413e-01
-3.51312645e-02 -7.86460578e-01 4.52546775e-01 -9.04334307e-01
-1.57786679e+00 7.71882236e-01 8.17497492e-01 6.20687492e-02
2.06217694e+00 -3.73703241e-01 9.36843693e-01 -1.43677406e-02
2.69731432e-01 -9.38733935e-01 1.07419550e-01 6.55499399e-02
7.73774385e-01 -1.25433552e+00 -1.13842618e-02 -6.86005831e-01
-5.08611798e-01 8.51219177e-01 5.25999844e-01 3.29849720e-01
6.27995312e-01 2.94434577e-01 8.44244212e-02 -5.44930324e-02
-6.00359440e-01 -3.61249119e-01 3.71568590e-01 5.35893321e-01
3.41177359e-02 -6.82660043e-02 5.27974823e-03 8.94577920e-01
2.14051947e-01 -2.16547295e-01 2.71743059e-01 8.74688983e-01
-2.61951089e-01 -1.30999935e+00 -5.25814712e-01 2.80737966e-01
-4.88814145e-01 -2.73004264e-01 8.17524455e-03 2.59305567e-01
4.36190814e-01 1.20224273e+00 -1.83499053e-01 -6.66397810e-01
1.71827435e-01 -2.07260534e-01 2.32987911e-01 -3.92028540e-01
-2.76716053e-01 4.76962596e-01 -1.89958155e-01 -7.21000791e-01
-4.26115483e-01 -4.94379550e-01 -9.26372111e-01 -1.51566073e-01
-3.69454741e-01 7.20252329e-03 5.78413904e-01 9.66380954e-01
5.46976686e-01 6.37926757e-01 9.23420131e-01 -7.93703198e-01
-2.68956989e-01 -6.22823238e-01 -7.07007289e-01 3.60588729e-01
7.39368856e-01 -6.91550553e-01 -3.76342714e-01 2.60417849e-01]
|
[11.067758560180664, -2.343198299407959]
|
e215ac16-b2a3-43ff-bb23-f0256fb796a4
|
online-learning-of-order-flow-and-market
|
2307.02375
| null |
https://arxiv.org/abs/2307.02375v1
|
https://arxiv.org/pdf/2307.02375v1.pdf
|
Online Learning of Order Flow and Market Impact with Bayesian Change-Point Detection Methods
|
Financial order flow exhibits a remarkable level of persistence, wherein buy (sell) trades are often followed by subsequent buy (sell) trades over extended periods. This persistence can be attributed to the division and gradual execution of large orders. Consequently, distinct order flow regimes might emerge, which can be identified through suitable time series models applied to market data. In this paper, we propose the use of Bayesian online change-point detection (BOCPD) methods to identify regime shifts in real-time and enable online predictions of order flow and market impact. To enhance the effectiveness of our approach, we have developed a novel BOCPD method using a score-driven approach. This method accommodates temporal correlations and time-varying parameters within each regime. Through empirical application to NASDAQ data, we have found that: (i) Our newly proposed model demonstrates superior out-of-sample predictive performance compared to existing models that assume i.i.d. behavior within each regime; (ii) When examining the residuals, our model demonstrates good specification in terms of both distributional assumptions and temporal correlations; (iii) Within a given regime, the price dynamics exhibit a concave relationship with respect to time and volume, mirroring the characteristics of actual large orders; (iv) By incorporating regime information, our model produces more accurate online predictions of order flow and market impact compared to models that do not consider regimes.
|
['Piero Mazzarisi', 'Fabrizio Lillo', 'Ioanna-Yvonni Tsaknaki']
|
2023-07-05
| null | null | null | null |
['change-point-detection']
|
['time-series']
|
[-3.74196380e-01 -6.88338995e-01 -3.11290711e-01 -1.56070605e-01
-4.07125980e-01 -1.14418304e+00 9.12692904e-01 4.23839808e-01
7.39970505e-02 6.01749599e-01 1.56533554e-01 -7.04505026e-01
-6.20907962e-01 -7.39756346e-01 -5.66920817e-01 -2.78390884e-01
-4.59055215e-01 5.26055396e-01 3.31906796e-01 3.70714664e-02
6.42067671e-01 6.00138009e-01 -1.09040070e+00 -1.56534046e-01
7.81205535e-01 1.24877906e+00 -2.28528291e-01 3.91448230e-01
-1.51488073e-02 7.10479915e-01 -4.72234577e-01 -2.54354179e-01
7.81765342e-01 -2.63410926e-01 -1.14067741e-01 2.69568682e-01
-8.69726390e-02 -5.92032492e-01 -3.28906059e-01 7.34960616e-01
-1.15914799e-01 7.03957304e-02 8.22624505e-01 -1.14169502e+00
-5.75356781e-01 3.82792264e-01 -5.24594784e-01 7.22207010e-01
2.79649943e-01 2.31986299e-01 1.37434852e+00 -6.14770114e-01
3.97700518e-01 8.83078039e-01 5.00918806e-01 -3.67748588e-01
-1.61601245e+00 -5.83910882e-01 4.22290623e-01 -1.55988470e-01
-1.09051466e+00 1.05364226e-01 9.42837417e-01 -6.90163732e-01
8.22923481e-01 2.55338281e-01 9.93463635e-01 7.29187906e-01
8.95135343e-01 6.84344411e-01 1.33103371e+00 -9.86529142e-02
4.48039174e-01 4.29258682e-02 1.51941389e-01 -2.15133429e-01
3.70669454e-01 5.34626603e-01 -3.04573238e-01 -5.79419971e-01
9.77932751e-01 4.17602062e-01 2.06034660e-01 -7.31835440e-02
-8.31919432e-01 7.89187670e-01 3.93116251e-02 3.31160218e-01
-7.29741931e-01 -2.48786733e-01 2.61302024e-01 5.67813873e-01
7.57174492e-01 3.49328548e-01 -7.21313179e-01 -3.80413949e-01
-1.20118320e+00 7.54650712e-01 1.07185006e+00 8.47267747e-01
2.25076079e-01 3.69153544e-02 -1.81589141e-01 3.32033962e-01
3.86411250e-01 5.60002029e-01 5.98904371e-01 -7.20990598e-01
4.94394392e-01 5.28984785e-01 4.84278798e-01 -8.51651549e-01
-3.18080008e-01 -5.92652023e-01 -5.35986722e-01 -4.73152474e-02
6.69609129e-01 1.89577520e-01 -5.09059548e-01 1.26415837e+00
-7.47723551e-03 2.30439305e-02 -1.08060040e-01 4.10028309e-01
-3.11994463e-01 7.02567160e-01 -1.55010685e-01 -8.17267299e-01
1.33806372e+00 -2.72004068e-01 -7.22029209e-01 1.96387038e-01
2.41462111e-01 -7.56109416e-01 8.20505977e-01 5.48707485e-01
-1.03555560e+00 -2.20347375e-01 -6.95885003e-01 7.47536182e-01
-5.22964224e-02 -4.24633265e-01 5.85695028e-01 3.70392501e-01
-6.90563202e-01 7.77904749e-01 -1.05181623e+00 7.23896772e-02
-1.03454124e-02 2.10463285e-01 4.22463208e-01 5.12318909e-01
-9.68396366e-01 3.50275934e-01 1.76744550e-01 1.52330950e-01
-6.34332061e-01 -9.75370705e-01 -2.59424090e-01 1.83064789e-01
3.62275720e-01 -2.41590247e-01 1.38710809e+00 -6.81677282e-01
-1.31781042e+00 3.84475030e-02 -2.48813685e-02 -6.94087386e-01
7.43260086e-01 -7.42207393e-02 -7.82176495e-01 6.99858274e-03
1.84217133e-02 -3.77251059e-01 6.15004897e-01 -1.12916112e+00
-8.72999132e-01 -3.57992768e-01 -2.31251165e-01 -1.67077273e-01
1.25399502e-02 4.64042984e-02 1.33629411e-01 -1.05780447e+00
2.95030624e-01 -9.43628430e-01 -7.81029090e-03 -7.94916451e-01
-2.76561320e-01 -2.50814021e-01 4.57189888e-01 -6.26697183e-01
1.82132828e+00 -2.02907133e+00 -6.19071364e-01 6.61089242e-01
5.01668826e-02 -2.81640887e-01 4.97390211e-01 1.03182602e+00
4.72380556e-02 1.69712692e-01 -1.11387216e-01 9.71113592e-02
3.26405823e-01 1.91371515e-01 -9.27153230e-01 4.64087635e-01
-5.49085848e-02 6.70210183e-01 -6.02761030e-01 1.84378028e-01
9.80920345e-02 -1.87556192e-01 -5.76994538e-01 1.47835001e-01
-1.29173130e-01 4.02611762e-01 -3.29681337e-01 8.66855323e-01
6.22465730e-01 -4.28332061e-01 3.53049636e-01 2.73649365e-01
-5.64856350e-01 4.84744668e-01 -1.13503420e+00 5.08147955e-01
-1.99535176e-01 4.67855781e-01 -3.47410738e-01 -7.10079312e-01
9.99774158e-01 2.64414579e-01 7.41429090e-01 -9.23919976e-01
-2.71752477e-02 5.34074366e-01 3.06685239e-01 -4.61280569e-02
4.01359230e-01 -3.65009636e-01 -7.30913579e-02 6.20172262e-01
-3.01973373e-01 1.36322901e-01 4.17679876e-01 -6.36353046e-02
1.08496118e+00 -2.56908238e-01 3.44697654e-01 -4.64685589e-01
-4.99796160e-02 -1.06190704e-02 7.96034038e-01 7.88002908e-01
-1.60617560e-01 -4.47869785e-02 8.38230014e-01 -3.76331031e-01
-1.09888911e+00 -1.31461203e+00 -4.93724972e-01 6.69352233e-01
1.87073424e-02 -4.57740985e-02 -9.20482799e-02 -3.05892408e-01
7.33119428e-01 7.63430536e-01 -3.82133037e-01 2.37371132e-01
-5.35638034e-01 -8.29660714e-01 -5.88227808e-02 6.16399109e-01
1.80499956e-01 -8.91030252e-01 -6.24954462e-01 6.29198432e-01
3.95795405e-01 -7.92315423e-01 -5.82798839e-01 2.04007760e-01
-1.33410907e+00 -1.09844577e+00 -5.20260334e-01 -2.26066619e-01
5.14556646e-01 -9.72808376e-02 1.08096647e+00 -5.02847433e-01
2.44606987e-01 5.05761325e-01 -1.35615811e-01 -4.37180638e-01
-2.52292752e-01 -1.45562902e-01 3.34024519e-01 1.85257092e-01
4.29558545e-01 -6.64057314e-01 -1.01382852e+00 4.09891665e-01
-1.03791070e+00 -6.37802184e-01 4.76982862e-01 7.35045850e-01
5.87923348e-01 3.25135976e-01 8.19220662e-01 -6.82215810e-01
1.01260424e+00 -6.30698502e-01 -1.31485391e+00 2.28235692e-01
-1.50765145e+00 -1.86516866e-01 5.93057811e-01 -7.80895829e-01
-1.00714827e+00 -4.04474020e-01 4.80473250e-01 -4.06315625e-01
1.51060671e-01 7.78549433e-01 4.77434635e-01 4.21670586e-01
-1.42105266e-01 3.94464672e-01 1.86310187e-01 -6.75084710e-01
-1.63461551e-01 5.64687550e-01 2.89436460e-01 -3.86242509e-01
9.10854995e-01 3.58275294e-01 -1.03915170e-01 -5.22472382e-01
-2.57070661e-01 -4.92936909e-01 -5.01205087e-01 -2.20508143e-01
1.50433391e-01 -8.08996618e-01 -1.08204007e+00 4.41823691e-01
-5.06170034e-01 -3.11827540e-01 -3.79114628e-01 8.67412627e-01
-5.35621345e-01 1.68888390e-01 -1.06910515e+00 -1.30744183e+00
3.80363571e-03 -8.40207636e-01 5.73040605e-01 2.18583673e-01
-4.86714780e-01 -1.42969000e+00 4.41982567e-01 8.24605487e-03
3.73134404e-01 1.62996382e-01 1.08911026e+00 -1.31944311e+00
-8.40829492e-01 -4.65775132e-01 6.89392909e-02 2.64764667e-01
4.08553213e-01 9.69451666e-02 -1.15500338e-01 -5.38563788e-01
2.80013591e-01 5.59417963e-01 3.41146886e-01 6.64510608e-01
5.68086684e-01 -7.89758265e-01 -8.78405422e-02 3.81828368e-01
1.44976830e+00 9.65114176e-01 2.01310992e-01 6.20656669e-01
3.17444988e-02 5.63775122e-01 6.77893698e-01 9.15647030e-01
3.47335815e-01 4.14595187e-01 9.06970128e-02 3.33527565e-01
6.70691729e-01 -5.83063006e-01 4.48216110e-01 9.75494921e-01
1.05502039e-01 -5.47331087e-02 -8.28779697e-01 7.34884083e-01
-1.85633624e+00 -9.43115532e-01 -1.49607554e-01 2.47412324e+00
5.96756637e-01 6.01713300e-01 6.92180455e-01 -4.97593246e-02
3.39355916e-01 8.96318555e-02 -7.99515009e-01 -3.78880858e-01
3.94442640e-02 -1.09596543e-01 1.00673938e+00 2.14064702e-01
-6.72052681e-01 3.48395050e-01 7.36292791e+00 4.75935906e-01
-1.00271869e+00 -3.21758658e-01 5.99644840e-01 -1.27630010e-01
-6.14629030e-01 3.08013260e-01 -8.90977740e-01 9.70713735e-01
1.08540368e+00 -6.19724691e-01 4.16321427e-01 4.77727264e-01
5.94933271e-01 -5.54416068e-02 -1.09957600e+00 5.77408969e-01
-5.00977814e-01 -1.06368828e+00 -1.14456244e-01 7.93658555e-01
7.39686251e-01 -3.44777912e-01 2.09067166e-01 2.03216985e-01
3.56797278e-01 -5.23178875e-01 8.02669585e-01 7.36885369e-01
1.40953615e-01 -8.12473595e-01 6.35636032e-01 3.84016514e-01
-1.16225421e+00 -5.13957679e-01 5.86467832e-02 -3.16369563e-01
3.97566944e-01 6.84959352e-01 -5.78691423e-01 5.31352460e-01
4.97875065e-01 6.05049610e-01 -2.54156768e-01 1.06947732e+00
1.72432572e-01 1.07574487e+00 -4.98550743e-01 3.43654715e-02
1.81578651e-01 -7.71643698e-01 5.96883535e-01 6.35664403e-01
5.45384169e-01 -2.98018958e-02 1.58828437e-01 1.00481200e+00
3.09165388e-01 5.64817823e-02 -5.14758289e-01 -3.30142796e-01
6.62477970e-01 5.74908733e-01 -9.49061751e-01 -3.11277896e-01
-6.65411055e-01 3.97824317e-01 -4.67566729e-01 5.28527379e-01
-5.54038882e-01 -1.81847408e-01 5.31325281e-01 6.15653098e-01
6.60034716e-01 -4.85981822e-01 -4.24674869e-01 -1.15597665e+00
3.41096222e-01 -7.87931502e-01 4.70437407e-01 -5.08027934e-02
-1.61331236e+00 1.69981182e-01 2.28472441e-01 -1.51969779e+00
-6.11819208e-01 -4.35240090e-01 -5.89871883e-01 5.64464152e-01
-1.28360498e+00 -5.16382933e-01 5.87052047e-01 4.35790479e-01
4.78808910e-01 -1.36978313e-01 1.42477691e-01 9.48469937e-02
-4.36669111e-01 2.27487773e-01 7.88166642e-01 3.04346178e-02
3.68044823e-01 -1.42419720e+00 2.96757430e-01 6.91710055e-01
1.88873217e-01 8.25058222e-01 8.73172641e-01 -1.23165751e+00
-1.25694203e+00 -7.06893623e-01 8.16912711e-01 -3.83086622e-01
1.45920086e+00 -2.07019985e-01 -1.06111121e+00 8.51028800e-01
9.88787487e-02 -4.22734350e-01 8.21660876e-01 1.69599876e-01
-1.36344045e-01 -4.23113108e-01 -9.61514711e-01 3.26754630e-01
4.05112505e-01 -4.09187794e-01 -8.33372831e-01 1.99070171e-01
3.88951153e-01 3.90110281e-03 -1.32666707e+00 3.33797127e-01
8.82317722e-01 -9.65748668e-01 5.45589805e-01 -4.22501743e-01
4.07002941e-02 -1.22853979e-01 -1.95539594e-01 -1.09022450e+00
-5.10489285e-01 -8.35553765e-01 -3.75249058e-01 1.40076959e+00
3.75178218e-01 -1.32222772e+00 2.61090726e-01 8.70962560e-01
3.64109933e-01 -6.66605830e-01 -1.03962278e+00 -1.39079809e+00
1.75309330e-01 -2.05278471e-01 7.44455934e-01 6.81641161e-01
5.26791960e-02 -2.37084299e-01 -2.58468658e-01 5.92222959e-02
6.87806845e-01 7.27230966e-01 5.07345796e-01 -1.34119081e+00
-4.60358262e-01 -6.42151952e-01 -1.32670328e-01 -1.19123983e+00
-1.98739856e-01 -4.29431081e-01 -2.10551053e-01 -9.23258007e-01
2.28690773e-01 -2.75515348e-01 -6.70766056e-01 -1.36015579e-01
1.83044270e-01 -1.73922539e-01 2.45115563e-01 8.65444779e-01
-1.85879841e-01 4.61980879e-01 8.43887568e-01 2.30920419e-01
-6.02259278e-01 5.15146792e-01 -6.03402674e-01 4.95520025e-01
6.59754872e-01 -3.27774853e-01 -2.01958448e-01 4.05099243e-01
3.27435344e-01 5.87745965e-01 3.37909192e-01 -5.10691464e-01
7.92490393e-02 -4.55046624e-01 2.91536868e-01 -1.00651133e+00
-2.43209511e-01 -8.91825676e-01 4.91459042e-01 6.85985982e-01
-2.41018042e-01 6.21007442e-01 1.28921270e-01 9.13455963e-01
-3.81796867e-01 1.45295501e-01 1.71596900e-01 1.04516670e-01
-1.75144330e-01 2.22285300e-01 -7.17160106e-01 -1.51152030e-01
1.05241263e+00 -2.21546978e-01 -5.95817231e-02 -5.38039207e-01
-6.12753034e-01 3.83705139e-01 4.49878544e-01 4.17148709e-01
1.77561417e-01 -1.34242344e+00 -4.23697233e-01 1.58983722e-01
-1.90732375e-01 -4.78322566e-01 1.02193549e-01 1.01979768e+00
-3.30490500e-01 6.97544932e-01 2.79767394e-01 -6.29708886e-01
-7.44124413e-01 5.95196247e-01 9.97232944e-02 -6.02600217e-01
-7.21789837e-01 -1.60088185e-02 1.58717886e-01 7.63290469e-03
-3.73092736e-03 -7.08399594e-01 1.40228644e-01 5.55532813e-01
4.20638204e-01 5.49209297e-01 -6.67544156e-02 -3.85422766e-01
-2.80608922e-01 3.68863463e-01 -2.38720477e-01 -1.90916404e-01
1.42685068e+00 -2.59493887e-01 -2.36348212e-02 1.09186637e+00
8.09278607e-01 1.53418913e-01 -1.70159566e+00 -1.40353307e-01
6.52815938e-01 -6.25858843e-01 -3.48088264e-01 -7.34126449e-01
-8.15879703e-01 1.20739333e-01 4.07754093e-01 1.06308222e+00
1.05281746e+00 -4.53576557e-02 8.35985601e-01 -2.93243080e-01
4.16291505e-01 -1.19899547e+00 -2.53648520e-01 3.91037017e-01
6.72110081e-01 -8.40346217e-01 -1.17710412e-01 -4.67527658e-03
-4.54518527e-01 8.37715268e-01 -5.43770492e-02 -5.28805733e-01
1.02580976e+00 1.10627234e-01 -1.11981884e-01 -1.59899861e-01
-1.09392834e+00 1.62985831e-01 3.02375883e-01 -1.00162655e-01
1.91570148e-01 4.02493834e-01 -7.07977831e-01 7.99155056e-01
-3.71980965e-01 -9.35355201e-02 5.15056491e-01 9.96536553e-01
-2.17816800e-01 -1.01224196e+00 -4.36256409e-01 6.98571801e-01
-7.88623095e-01 2.24070214e-02 -2.30211735e-01 1.10427201e+00
-5.40405393e-01 9.37230766e-01 6.35483444e-01 -1.64116040e-01
5.41783035e-01 1.93939969e-01 7.53040165e-02 -2.29372874e-01
-7.55151510e-01 7.19294012e-01 -2.76440114e-01 -4.57488954e-01
5.39425900e-03 -1.11452174e+00 -8.98778200e-01 -5.26762843e-01
-3.17081243e-01 1.30606338e-01 3.34083855e-01 1.00083518e+00
4.48371679e-01 4.84357685e-01 1.17238569e+00 -4.89095688e-01
-1.20521593e+00 -9.18993890e-01 -1.36443186e+00 4.88023579e-01
4.27863449e-01 -7.18644440e-01 -9.09759283e-01 -1.70859829e-01]
|
[4.841994285583496, 4.073649883270264]
|
6a8ae512-dbe1-4077-b75e-7adcf71ecc8b
|
learning-from-synthetic-data-using-a-stacked
|
1509.05463
| null |
http://arxiv.org/abs/1509.05463v2
|
http://arxiv.org/pdf/1509.05463v2.pdf
|
Learning from Synthetic Data Using a Stacked Multichannel Autoencoder
|
Learning from synthetic data has many important and practical applications.
An example of application is photo-sketch recognition. Using synthetic data is
challenging due to the differences in feature distributions between synthetic
and real data, a phenomenon we term synthetic gap. In this paper, we
investigate and formalize a general framework-Stacked Multichannel Autoencoder
(SMCAE) that enables bridging the synthetic gap and learning from synthetic
data more efficiently. In particular, we show that our SMCAE can not only
transform and use synthetic data on the challenging face-sketch recognition
task, but that it can also help simulate real images, which can be used for
training classifiers for recognition. Preliminary experiments validate the
effectiveness of the framework.
|
['Yanwei Fu', 'Shanshan Jiang', 'Gady Agam', 'Xi Zhang', 'Leonid Sigal']
|
2015-09-17
| null | null | null | null |
['sketch-recognition']
|
['computer-vision']
|
[ 3.19178104e-01 -1.88349456e-01 1.76785469e-01 -3.97604764e-01
-4.40462112e-01 -1.62108451e-01 5.93821883e-01 -7.74032474e-01
2.95366254e-02 7.49117136e-01 1.32202879e-02 1.11011356e-01
-9.63994041e-02 -7.23983109e-01 -8.32218885e-01 -7.08857477e-01
3.01463872e-01 2.83047557e-01 -2.47946065e-02 -1.35217577e-01
2.11906340e-02 8.32165956e-01 -1.94643426e+00 8.50402176e-01
6.26095474e-01 9.60749745e-01 2.28970218e-02 4.24637765e-01
-1.44253522e-01 6.00163996e-01 -7.25136817e-01 -3.23715091e-01
3.85824770e-01 -3.32041442e-01 -2.36356854e-01 3.79142493e-01
6.81985795e-01 -3.90179813e-01 -4.62166429e-01 7.71763742e-01
4.79037285e-01 -1.18603669e-01 8.32708776e-01 -1.57426012e+00
-6.13483369e-01 2.95866847e-01 -2.46920004e-01 -4.37545210e-01
2.35730410e-01 -9.13624167e-02 4.11502004e-01 -1.34261346e+00
5.88506937e-01 1.48678374e+00 7.88106501e-01 8.67733657e-01
-1.20305228e+00 -9.51313198e-01 -4.44501638e-01 1.85951725e-01
-1.38543522e+00 -6.11079633e-01 1.01700127e+00 -4.09638762e-01
4.00436074e-01 1.36970490e-01 6.44777536e-01 1.65083945e+00
-3.52271199e-01 1.23507214e+00 1.32016242e+00 -6.25239372e-01
1.88786447e-01 2.77729869e-01 -2.87858963e-01 6.35892749e-01
1.56415135e-01 2.31082350e-01 -5.18291175e-01 -8.08587819e-02
9.48680580e-01 6.55638427e-02 -3.33835036e-01 -3.98188531e-01
-1.27360332e+00 7.78129518e-01 4.59205098e-02 4.24672693e-01
-3.48262042e-01 -2.91456226e-02 1.70492813e-01 5.82166135e-01
1.80853799e-01 2.99773633e-01 2.31597908e-02 1.12685256e-01
-1.15420771e+00 1.79769844e-01 8.22120070e-01 7.36252129e-01
6.13833129e-01 6.28214002e-01 1.89179316e-01 1.11117160e+00
2.58579887e-02 6.71766698e-01 7.20679224e-01 -9.10788894e-01
2.32997134e-01 4.94229406e-01 -1.26811549e-01 -9.31005716e-01
4.28725220e-02 1.68126598e-01 -1.08707011e+00 3.88384044e-01
5.28406441e-01 1.73860267e-01 -7.39843249e-01 1.51659751e+00
1.68158382e-01 5.15264392e-01 2.43652612e-01 7.45081782e-01
8.65909696e-01 6.27249777e-01 -2.81693816e-01 7.11270049e-02
7.74772346e-01 -6.49676681e-01 -6.08306646e-01 5.32094017e-02
7.51697198e-02 -8.07450652e-01 9.96581614e-01 5.54894388e-01
-7.91850984e-01 -8.85038435e-01 -1.19333470e+00 2.56623983e-01
-5.34311950e-01 5.64712465e-01 3.89369816e-01 7.61757731e-01
-8.87177944e-01 6.96353137e-01 -6.32791400e-01 -3.55878562e-01
5.55994689e-01 2.32883677e-01 -6.96902871e-01 -1.55092984e-01
-1.14609313e+00 4.58787590e-01 3.78346831e-01 1.00524232e-01
-5.06850123e-01 -6.95864499e-01 -8.55962455e-01 5.52432938e-03
2.03137845e-01 -1.42194256e-01 8.41555238e-01 -1.18057406e+00
-1.63834727e+00 5.34613311e-01 1.84638590e-01 -2.76658386e-01
4.85197037e-01 1.53097570e-01 -7.59479642e-01 1.66313961e-01
-2.89013416e-01 7.30838001e-01 1.58115494e+00 -1.38264501e+00
-2.77981628e-02 -5.35152674e-01 -3.67646992e-01 -3.19869190e-01
-5.85187435e-01 -2.01865390e-01 -2.49473646e-01 -1.08126009e+00
-1.33114666e-01 -9.41321492e-01 2.42515787e-01 3.44658107e-01
-2.08230615e-01 -6.37004897e-02 1.46437728e+00 -5.90746045e-01
6.84732735e-01 -2.28517032e+00 -5.39345369e-02 3.01582843e-01
-8.26150849e-02 7.21172810e-01 -4.73395914e-01 6.77772462e-01
-2.92913616e-01 -2.51633555e-01 -3.61768961e-01 2.29481310e-02
-1.43063843e-01 3.82720053e-01 -6.93903387e-01 1.34776518e-01
4.49466616e-01 9.77783203e-01 -4.75721955e-01 -3.30337554e-01
3.73194635e-01 8.30223083e-01 -2.58903325e-01 3.21101636e-01
-1.59069330e-01 3.72831523e-01 -2.98131049e-01 7.14739442e-01
8.89793992e-01 -1.21825181e-01 2.39689201e-02 -4.09618944e-01
1.78873956e-01 -5.91504931e-01 -1.51896369e+00 1.17443705e+00
-5.20802915e-01 9.78316069e-01 2.53726672e-02 -1.25328588e+00
1.12346590e+00 2.54125476e-01 4.07142311e-01 -6.05383396e-01
-1.28523961e-01 1.68347105e-01 -2.22376391e-01 -5.17451704e-01
1.42260775e-01 -2.14387551e-01 2.86231458e-01 5.99511683e-01
2.74969399e-01 -2.90457308e-01 2.81562470e-02 -4.38832790e-02
4.65574563e-01 1.01868644e-01 1.65857539e-01 -5.76261207e-02
8.47783685e-01 -3.47843766e-01 1.79304898e-01 5.35328925e-01
2.34110355e-01 7.30872631e-01 3.78722310e-01 -5.71993947e-01
-1.29116488e+00 -1.08703852e+00 1.80565175e-02 5.04123211e-01
-2.30529726e-01 -4.70189601e-02 -8.30165327e-01 -7.28719056e-01
3.71818870e-01 3.74334246e-01 -6.98657632e-01 -1.15548261e-01
-5.65540731e-01 -2.63227850e-01 8.59817266e-01 7.79831648e-01
5.38124979e-01 -1.25410187e+00 -4.16237801e-01 4.87385541e-02
1.37778208e-01 -1.41974926e+00 -1.79514721e-01 -4.37418342e-01
-6.03342175e-01 -1.24102354e+00 -1.00501812e+00 -6.02975965e-01
5.38152397e-01 3.18259239e-01 8.45391333e-01 6.52771816e-02
-5.74719489e-01 8.19852471e-01 -2.38503963e-01 -6.15907013e-01
-8.72958541e-01 -4.21618253e-01 3.06377679e-01 6.53366566e-01
1.17042743e-01 -6.23144269e-01 -2.68751591e-01 4.31432247e-01
-1.22002852e+00 1.01223737e-01 7.42463470e-01 1.32422316e+00
2.88566709e-01 -2.38746509e-01 6.80220366e-01 -6.76904857e-01
6.99277163e-01 -1.13260411e-02 -6.58655882e-01 4.85120028e-01
-2.68653214e-01 2.60502875e-01 9.38123345e-01 -5.76996922e-01
-1.12469482e+00 1.81454137e-01 -8.57107788e-02 -8.86036396e-01
-3.29801947e-01 1.42843291e-01 -1.51815429e-01 -4.95143116e-01
5.93006730e-01 5.65795362e-01 5.91252983e-01 -4.32458967e-01
1.57490939e-01 1.00654304e+00 4.14242893e-01 -6.51442826e-01
8.53073597e-01 6.82072639e-01 1.77716419e-01 -1.59067059e+00
-2.61473149e-01 1.54070584e-02 -6.18818760e-01 -3.65719348e-01
3.01343888e-01 -6.88410997e-01 -7.22145379e-01 6.55911148e-01
-1.03073621e+00 -3.62365514e-01 -2.93904901e-01 4.62692916e-01
-7.20202923e-01 5.63103259e-01 -4.13879246e-01 -8.45087051e-01
-1.72266185e-01 -1.02749634e+00 1.21747160e+00 1.52600914e-01
3.29793185e-01 -8.39320123e-01 -7.32021704e-02 1.81799918e-01
4.84031886e-01 3.50860149e-01 7.34303892e-01 -4.62618351e-01
-5.84786773e-01 -4.27650779e-01 -3.72110128e-01 8.99687767e-01
2.96597439e-03 1.53959438e-01 -1.16364300e+00 -3.61219198e-01
-1.94079697e-01 -8.26057136e-01 9.01261151e-01 -5.18060252e-02
1.62504184e+00 -2.27748081e-01 -9.10678729e-02 3.63882065e-01
1.34333158e+00 7.04111978e-02 7.50614583e-01 -2.74785131e-01
4.02117491e-01 7.27546811e-01 4.24380511e-01 3.18541884e-01
-1.16641238e-01 8.74962151e-01 -6.50281832e-02 -1.86225940e-02
-4.26625311e-01 -4.33331877e-01 2.85146385e-01 8.52980316e-01
-2.24642036e-03 -5.81673703e-06 -7.98296988e-01 3.05561692e-01
-1.61207724e+00 -1.07538545e+00 3.22814733e-01 2.02154422e+00
5.00384212e-01 -4.61965472e-01 1.97135471e-02 5.46724856e-01
6.91270888e-01 1.34654492e-01 -2.89626926e-01 -1.87421039e-01
-2.24588871e-01 5.76733530e-01 -5.57360239e-02 1.97879553e-01
-8.76444697e-01 8.50686967e-01 6.33042145e+00 1.14006746e+00
-1.48313677e+00 -1.78574979e-01 3.81567746e-01 4.62236434e-01
2.15778667e-02 -3.79145265e-01 -4.23071980e-01 4.73126203e-01
6.51437700e-01 6.60878420e-02 6.47725701e-01 8.58611822e-01
-3.31176430e-01 2.42827564e-01 -1.26469195e+00 1.35539544e+00
4.61925656e-01 -1.44691682e+00 6.45798206e-01 -8.86449441e-02
6.12030149e-01 -3.63918871e-01 1.82534188e-01 2.36729503e-01
-1.48608625e-01 -1.10178077e+00 4.58962560e-01 6.42822504e-01
1.16310024e+00 -6.45971060e-01 4.57121611e-01 3.82627159e-01
-1.08630371e+00 -1.92575231e-01 -4.87401485e-01 2.40088031e-01
-3.03068727e-01 4.17916596e-01 -9.25375879e-01 4.87189382e-01
3.62964123e-01 5.68699360e-01 -5.48494101e-01 7.99893439e-01
7.98223615e-02 3.79154354e-01 -2.87260115e-01 -1.26943186e-01
-7.10271895e-02 -3.10778171e-01 2.67625153e-01 1.15241480e+00
5.91448843e-01 -6.08054474e-02 -1.77666932e-01 8.93541694e-01
2.68746894e-02 3.44613846e-03 -1.09297645e+00 -3.82446378e-01
3.51150721e-01 9.01191115e-01 -3.03720534e-01 -3.44218761e-01
-4.81290340e-01 1.12960100e+00 1.90817133e-01 6.11657500e-01
-5.22826314e-01 -4.26760972e-01 5.33524454e-01 -4.72301468e-02
4.83951747e-01 -1.65517300e-01 1.26449078e-01 -1.34848845e+00
1.56192914e-01 -1.11613870e+00 9.69645455e-02 -7.77073443e-01
-1.31344450e+00 6.02972448e-01 -7.40231946e-02 -1.45245767e+00
-4.18457866e-01 -1.06844592e+00 -5.63031793e-01 3.71379614e-01
-1.30007517e+00 -1.37935150e+00 -6.43532515e-01 7.70251572e-01
5.91770768e-01 -8.48276496e-01 9.83922243e-01 3.50947499e-01
-2.95025587e-01 8.57147217e-01 2.96488404e-01 4.05255854e-01
6.60501897e-01 -9.44449902e-01 4.12304074e-01 3.53331894e-01
7.49499261e-01 2.96858728e-01 3.53235334e-01 -2.67427921e-01
-1.61218095e+00 -1.15790951e+00 2.90979385e-01 -1.82132006e-01
3.69064212e-01 -5.83625495e-01 -9.64803398e-01 3.60725045e-01
-1.49737239e-01 2.76979864e-01 6.71041250e-01 -5.76273501e-01
-7.37290561e-01 -3.39474678e-01 -1.20715201e+00 6.12387121e-01
7.71930456e-01 -7.21643448e-01 -3.94939154e-01 1.49854228e-01
9.14141908e-02 -6.46338388e-02 -8.65873814e-01 6.82484388e-01
1.19547224e+00 -1.17362750e+00 1.25721490e+00 -6.01984620e-01
4.61142689e-01 5.26874699e-03 -2.69881725e-01 -1.51475751e+00
1.98545620e-01 -2.68066615e-01 -3.40040356e-01 1.01587093e+00
1.47455931e-01 -6.15368783e-01 9.84935760e-01 2.21562490e-01
4.51407492e-01 -6.72786176e-01 -9.16455865e-01 -1.07807279e+00
4.65685502e-02 -4.72169697e-01 7.58764148e-01 9.74478185e-01
-4.39989686e-01 -1.28700346e-01 -5.99276721e-01 -1.82578221e-01
8.82987618e-01 1.80878177e-01 1.12735629e+00 -1.32479262e+00
-1.33409217e-01 -2.12375671e-01 -6.71455204e-01 -7.79646158e-01
5.58945119e-01 -6.56133235e-01 -3.51602763e-01 -9.45933580e-01
-3.00130378e-02 -3.19303364e-01 -3.52565050e-02 1.37730092e-01
1.67751834e-01 5.65931320e-01 4.76270139e-01 -3.56191657e-02
-1.28381014e-01 9.00392592e-01 1.12768519e+00 -2.68678248e-01
1.46465451e-01 -2.49735489e-02 -1.04787953e-01 5.25880456e-01
6.02780163e-01 -1.53861567e-01 -3.08637619e-01 -5.37756942e-02
-3.90918970e-01 2.46779263e-01 5.13504386e-01 -1.33427882e+00
1.70654282e-01 -4.02729698e-02 7.09564865e-01 -4.74477351e-01
6.86476767e-01 -9.78700161e-01 1.75115868e-01 3.83007616e-01
-1.61286399e-01 -4.28428054e-01 2.70952046e-01 6.50595486e-01
-4.98762637e-01 -1.14394493e-01 8.69952798e-01 -6.71783239e-02
-6.24638617e-01 2.80607820e-01 -1.60956442e-01 -3.41376901e-01
1.04976833e+00 -5.13602614e-01 -8.55014548e-02 -7.10768402e-01
-5.78171849e-01 -2.32959390e-01 3.11088443e-01 6.31155133e-01
1.18316007e+00 -1.74046266e+00 -8.32566202e-01 9.60348785e-01
2.16843531e-01 -5.65789402e-01 2.67455071e-01 4.17160481e-01
-3.59260023e-01 5.13566911e-01 -6.51429713e-01 -5.91624260e-01
-1.26787293e+00 5.79686701e-01 3.54795277e-01 2.55502701e-01
-4.75371838e-01 4.89718825e-01 2.07761258e-01 -6.35357738e-01
2.55686611e-01 3.96137498e-02 -1.54611073e-03 -4.72158082e-02
7.99143791e-01 3.98478121e-01 -1.89475063e-02 -6.37322605e-01
-9.47546121e-03 6.75370872e-01 1.72998577e-01 -1.34163007e-01
1.43805575e+00 5.38220108e-01 2.50428542e-02 3.82301956e-01
1.22231627e+00 -1.90065369e-01 -1.22627270e+00 -2.35584825e-01
-3.30197401e-02 -7.80820727e-01 -3.35663855e-01 -4.44815576e-01
-1.23626792e+00 1.06149912e+00 7.62835205e-01 1.86752021e-01
1.03146827e+00 -2.62149066e-01 6.61893129e-01 7.71126270e-01
4.47661996e-01 -1.13829446e+00 4.99115258e-01 1.38683677e-01
1.45553076e+00 -1.22213686e+00 -2.50726789e-01 -4.97306436e-01
-6.65202260e-01 1.57582903e+00 6.70030534e-01 -2.60924160e-01
7.37273753e-01 4.52197522e-01 -4.81004901e-02 4.39902842e-02
-5.56292057e-01 6.09274469e-02 3.04386348e-01 8.03323388e-01
1.43971980e-01 -6.46022484e-02 1.91293042e-02 4.67735678e-01
1.08432651e-01 3.42687517e-01 4.08057421e-01 6.63070858e-01
-3.81890424e-02 -1.37591159e+00 -6.47781193e-01 4.47406858e-01
2.21657511e-02 3.51041645e-01 -6.72314823e-01 9.88205910e-01
8.64568353e-03 4.18612540e-01 3.59074399e-02 -5.48914135e-01
3.32604378e-01 4.33896929e-01 8.11972857e-01 -1.87678695e-01
1.40994843e-02 -2.93883413e-01 -1.34352177e-01 -5.07179618e-01
-4.88178551e-01 -4.50724304e-01 -6.16313875e-01 -6.15967773e-02
-3.14331591e-01 1.19694881e-01 8.04402709e-01 7.75710404e-01
2.95055956e-01 3.68848778e-02 8.99843872e-01 -9.73050177e-01
-8.03037405e-01 -1.06201291e+00 -6.08272076e-01 5.93615055e-01
4.57869172e-01 -8.76016080e-01 -2.25481078e-01 4.80624288e-02]
|
[11.99551010131836, 0.3630218207836151]
|
d72b23c6-fd40-470d-9f2b-6a19f7d362ab
|
word-class-representations-spontaneously
|
2302.07588
| null |
https://arxiv.org/abs/2302.07588v1
|
https://arxiv.org/pdf/2302.07588v1.pdf
|
Word class representations spontaneously emerge in a deep neural network trained on next word prediction
|
How do humans learn language, and can the first language be learned at all? These fundamental questions are still hotly debated. In contemporary linguistics, there are two major schools of thought that give completely opposite answers. According to Chomsky's theory of universal grammar, language cannot be learned because children are not exposed to sufficient data in their linguistic environment. In contrast, usage-based models of language assume a profound relationship between language structure and language use. In particular, contextual mental processing and mental representations are assumed to have the cognitive capacity to capture the complexity of actual language use at all levels. The prime example is syntax, i.e., the rules by which words are assembled into larger units such as sentences. Typically, syntactic rules are expressed as sequences of word classes. However, it remains unclear whether word classes are innate, as implied by universal grammar, or whether they emerge during language acquisition, as suggested by usage-based approaches. Here, we address this issue from a machine learning and natural language processing perspective. In particular, we trained an artificial deep neural network on predicting the next word, provided sequences of consecutive words as input. Subsequently, we analyzed the emerging activation patterns in the hidden layers of the neural network. Strikingly, we find that the internal representations of nine-word input sequences cluster according to the word class of the tenth word to be predicted as output, even though the neural network did not receive any explicit information about syntactic rules or word classes during training. This surprising result suggests, that also in the human brain, abstract representational categories such as word classes may naturally emerge as a consequence of predictive coding and processing during language acquisition.
|
['Patrick Krauss', 'Andreas Maier', 'Paul Stoewer', 'Achim Schilling', 'Kishore Surendra']
|
2023-02-15
| null | null | null | null |
['language-acquisition']
|
['natural-language-processing']
|
[ 5.10884941e-01 3.58223319e-01 -8.39870498e-02 -5.00538766e-01
3.69139671e-01 -7.21126854e-01 7.34338999e-01 5.49431443e-01
-5.20016909e-01 2.37743646e-01 2.63537914e-01 -7.83811271e-01
2.25136001e-02 -1.12336969e+00 -6.66434646e-01 -4.91920322e-01
9.33710765e-03 3.42121750e-01 1.99059378e-02 -5.49374223e-01
4.11319822e-01 3.31552029e-01 -1.80350089e+00 4.14159238e-01
9.55396235e-01 4.44127977e-01 6.66478097e-01 3.38195443e-01
-5.30657589e-01 6.42632663e-01 -2.74965674e-01 -2.54604131e-01
-6.49179816e-02 -7.24060059e-01 -9.60064411e-01 1.62399679e-01
1.09972253e-01 -1.29040420e-01 -2.72553861e-02 1.11179817e+00
-2.01547563e-01 5.53950407e-02 5.93678594e-01 -2.33026057e-01
-9.35873270e-01 1.07516849e+00 1.31742582e-01 1.12710893e-01
4.08588737e-01 2.60025948e-01 1.19016528e+00 -1.05846786e+00
5.89828730e-01 1.36472332e+00 1.93558633e-01 6.09884620e-01
-1.52140439e+00 -1.70323849e-01 4.55418408e-01 3.33490781e-02
-1.11582422e+00 -3.09342802e-01 6.03404880e-01 -8.87520611e-01
1.21587789e+00 -3.55808623e-02 1.23595715e+00 7.96207964e-01
4.00544077e-01 5.25259435e-01 1.15918398e+00 -9.69162285e-01
1.07989386e-01 1.91035345e-01 5.79100311e-01 7.71447599e-01
4.96561587e-01 3.79231930e-01 -5.76058924e-01 2.70930320e-01
6.72047138e-01 1.72616635e-02 -2.41251409e-01 9.12545398e-02
-1.12967932e+00 1.01774800e+00 3.14615428e-01 1.02091980e+00
-4.38864857e-01 -6.56662649e-03 2.64410526e-01 4.94827509e-01
1.50027692e-01 5.51205873e-01 -6.87203884e-01 1.84009835e-01
-5.69151878e-01 -8.13332871e-02 5.19636631e-01 3.83479416e-01
9.44168568e-01 2.28527606e-01 3.95943999e-01 8.45914125e-01
5.25521338e-01 3.66640896e-01 8.17775071e-01 -5.07981360e-01
-4.53006476e-02 7.03349113e-01 -6.11308455e-01 -1.01453841e+00
-2.62461543e-01 -3.03429216e-01 -4.77474838e-01 6.82034418e-02
5.58867872e-01 4.14520018e-02 -7.76185989e-01 2.32151604e+00
-3.93938608e-02 -4.34699863e-01 2.50995368e-01 5.52087963e-01
4.96341228e-01 7.79438436e-01 4.56048161e-01 -4.68830884e-01
1.43838966e+00 -2.55092412e-01 -4.59455997e-01 -7.06807077e-01
8.67176294e-01 -3.17762107e-01 1.22610152e+00 4.03303325e-01
-1.13722348e+00 -7.21925378e-01 -1.03600991e+00 -9.78873894e-02
-6.44776165e-01 -3.46750200e-01 1.02646911e+00 6.31504714e-01
-1.10256994e+00 7.31435895e-01 -6.41853333e-01 -6.16130114e-01
9.84589979e-02 1.58908010e-01 -3.05351794e-01 1.50472268e-01
-1.41728222e+00 1.06658721e+00 8.89072537e-01 1.59938782e-01
-7.20499933e-01 -2.54406869e-01 -9.68917429e-01 1.88126773e-01
7.41443932e-02 -4.48063195e-01 1.27973378e+00 -1.63412941e+00
-1.28207588e+00 1.38641536e+00 -3.14677835e-01 -2.66187012e-01
-5.29743373e-01 1.14673935e-01 -2.88228005e-01 1.98933226e-03
-6.78792894e-02 4.94731128e-01 6.06986463e-01 -1.16928089e+00
-4.46420968e-01 -5.44814825e-01 9.15851742e-02 2.59388257e-02
-1.44985929e-01 3.69941331e-02 2.30587944e-01 -5.84618330e-01
5.21780312e-01 -7.62179732e-01 -2.09483042e-01 -2.79755563e-01
4.80569191e-02 -6.32915556e-01 -2.40920931e-01 -4.19680089e-01
1.28160036e+00 -2.34548831e+00 1.97701931e-01 2.89723366e-01
1.50562108e-01 1.49600729e-01 -7.93937817e-02 5.01094103e-01
-4.07476395e-01 4.37873602e-01 -3.08877051e-01 2.94482708e-01
7.80591518e-02 6.10484779e-01 -5.52051246e-01 2.03043610e-01
2.24635854e-01 9.69046891e-01 -8.90586078e-01 -1.14321262e-01
2.54892632e-02 2.66216069e-01 -6.86262369e-01 2.80416831e-02
-4.38545436e-01 1.64173409e-01 -3.06956142e-01 1.97613120e-01
1.28049940e-01 -2.29542837e-01 7.99257994e-01 3.50000501e-01
-2.67933965e-01 9.24702048e-01 -7.21503437e-01 1.40502203e+00
-4.04897124e-01 7.42935002e-01 -2.75071919e-01 -1.39808929e+00
6.85525715e-01 5.23970783e-01 -3.81710052e-01 -8.13834846e-01
2.50858009e-01 4.75688070e-01 9.85103250e-01 -5.33128798e-01
1.57590628e-01 -6.39424741e-01 -2.18215913e-01 6.26186013e-01
2.25496337e-01 -1.30570427e-01 3.35057259e-01 -4.46076617e-02
6.76365077e-01 -4.41575572e-02 7.84334183e-01 -5.83944023e-01
4.96895373e-01 -1.28684267e-01 5.84184408e-01 6.38703585e-01
2.34101340e-01 1.11742625e-02 5.20090520e-01 -7.62235522e-01
-9.02172923e-01 -1.15707958e+00 -4.20835584e-01 1.58682847e+00
-3.22808117e-01 -2.68058717e-01 -8.57766926e-01 -1.02532484e-01
-4.89456356e-01 1.20759618e+00 -6.68985307e-01 -3.19981813e-01
-8.33189666e-01 -4.45798367e-01 1.27414048e-01 2.00367779e-01
-1.10604107e-01 -1.77617919e+00 -1.05524552e+00 5.13277352e-01
3.28001320e-01 -8.40982199e-01 -2.00561378e-02 3.92971396e-01
-1.07484126e+00 -8.04405630e-01 -3.45038548e-02 -9.63152230e-01
8.76420379e-01 -4.18635309e-02 1.22795534e+00 8.32983792e-01
-1.08568370e-01 3.19637179e-01 -3.26938540e-01 -4.59351301e-01
-7.47593462e-01 -2.02838220e-02 1.72296569e-01 -2.12274969e-01
6.67481244e-01 -9.36823428e-01 -3.65129918e-01 -3.80827785e-01
-1.20153284e+00 1.47659168e-01 6.03160322e-01 7.18274951e-01
1.46648377e-01 -1.98074028e-01 4.70245540e-01 -9.09487069e-01
6.37145162e-01 -4.82940435e-01 -3.90038282e-01 2.13616922e-01
-4.43218797e-01 3.72636616e-01 6.48493230e-01 -4.18640316e-01
-8.89097095e-01 -1.01105057e-01 -3.04966211e-01 4.92088795e-01
-5.06220639e-01 9.13893223e-01 -5.95704429e-02 5.36619961e-01
7.63367653e-01 8.44072282e-01 1.04732839e-02 -1.81011826e-01
3.46083015e-01 2.56548017e-01 9.81033295e-02 -9.84551370e-01
5.80584824e-01 1.10196667e-02 -3.05700332e-01 -1.16928411e+00
-8.76829684e-01 1.99994311e-01 -8.60642135e-01 -3.45390961e-02
1.18441594e+00 -6.33441269e-01 -5.76648951e-01 2.47578591e-01
-1.47183013e+00 -3.28816593e-01 -3.86515796e-01 5.61844170e-01
-5.28297067e-01 1.58087298e-01 -5.63015878e-01 -8.43520820e-01
-3.10511123e-02 -1.04809237e+00 4.40397888e-01 1.16688363e-01
-6.31370068e-01 -1.13130260e+00 -1.34238899e-02 -3.34280938e-01
2.31469288e-01 -3.60014774e-02 1.72999418e+00 -8.68309259e-01
-4.37046111e-01 2.60642201e-01 4.59490687e-01 2.13106245e-01
2.60482371e-01 7.99868535e-03 -8.02824259e-01 6.18981905e-02
2.80042648e-01 -3.20726812e-01 7.55463719e-01 1.31628916e-01
9.40548301e-01 -4.23408538e-01 -8.28168690e-02 3.34818572e-01
1.47238421e+00 6.33040309e-01 4.68980432e-01 -6.74581528e-02
2.43107215e-01 1.19035149e+00 -5.15885763e-02 -6.77673221e-02
2.79731274e-01 2.19653711e-01 -8.54013860e-02 3.87057006e-01
2.19380513e-01 -3.74494761e-01 5.52037179e-01 1.41455257e+00
-5.71522266e-02 2.29859389e-02 -1.31276941e+00 6.96514428e-01
-1.32971251e+00 -1.02907681e+00 -2.05143690e-02 2.26066852e+00
1.21583974e+00 4.10136491e-01 -4.06790674e-01 2.68353760e-01
5.42852223e-01 1.05407715e-01 -2.85365105e-01 -9.49506462e-01
-1.46576390e-01 6.87585056e-01 -2.71729738e-01 7.44874358e-01
-3.71005535e-01 1.39684772e+00 6.72229767e+00 4.03323442e-01
-1.29121137e+00 -1.45603716e-02 5.13046205e-01 4.25027162e-01
-7.38837302e-01 7.62577131e-02 -5.39056957e-01 2.37861112e-01
1.10864794e+00 -2.40714937e-01 5.45986414e-01 4.19466466e-01
4.40137610e-02 -1.70975342e-01 -1.46891654e+00 5.36173642e-01
-1.18584439e-01 -1.15039575e+00 4.34992403e-01 1.01751313e-01
1.88732594e-01 -2.80862868e-01 -1.45446688e-01 3.74850601e-01
1.92868710e-01 -1.32733774e+00 8.83162677e-01 4.54110444e-01
5.14917552e-01 -3.18457931e-01 1.27495155e-01 7.08582342e-01
-9.56305385e-01 8.06294158e-02 -4.89208043e-01 -8.31251979e-01
3.59814502e-02 4.11638409e-01 -4.77032483e-01 -2.25444257e-01
1.08944930e-01 3.46262723e-01 -5.98698139e-01 1.41945958e-01
-7.10819423e-01 8.63688886e-01 -1.44669801e-01 -3.68045866e-01
2.14491174e-01 -1.50555208e-01 2.60283202e-01 1.06905806e+00
3.02559793e-01 6.59250617e-01 -7.75580779e-02 9.85710859e-01
3.13765377e-01 3.48285645e-01 -9.69907224e-01 -3.37893546e-01
4.57941949e-01 8.56713593e-01 -9.08625066e-01 -3.85595530e-01
-5.37189066e-01 6.44591272e-01 6.11548662e-01 3.32594723e-01
-1.84424043e-01 1.06677316e-01 7.02887833e-01 2.85539359e-01
2.01099277e-01 -5.25977671e-01 -2.68335402e-01 -1.24034655e+00
-2.89222270e-01 -7.69034207e-01 -8.33611116e-02 -4.83995408e-01
-1.09507275e+00 5.94398856e-01 -1.13192290e-01 -4.38094050e-01
-6.90573454e-01 -1.15320575e+00 -6.26190066e-01 8.21196795e-01
-8.35357487e-01 -6.47939503e-01 4.02956009e-01 3.71530861e-01
6.96098149e-01 -1.39386043e-01 1.16434121e+00 -1.80165038e-01
-2.71013826e-01 3.15849543e-01 -2.79399097e-01 3.25831622e-01
-1.43506184e-01 -1.11695993e+00 3.47641051e-01 7.73441136e-01
6.34810507e-01 1.33240378e+00 6.57576442e-01 -4.74523753e-01
-1.20423543e+00 -4.83898282e-01 1.41520000e+00 -3.25740427e-01
6.43027604e-01 -6.84995353e-01 -1.38151431e+00 8.60421360e-01
4.89038527e-01 -1.65266320e-01 1.04147315e+00 1.16607964e-01
-3.67501885e-01 2.31042206e-01 -6.02096796e-01 7.73308933e-01
1.12889457e+00 -7.00298011e-01 -1.38930905e+00 1.19945221e-01
1.01149333e+00 2.91138083e-01 -4.97947127e-01 1.37579694e-01
5.32750189e-01 -1.03661633e+00 5.85195780e-01 -8.57131302e-01
6.17049694e-01 -9.83406603e-02 -3.20497334e-01 -1.32832336e+00
-5.68581223e-01 -1.86967418e-01 2.44154036e-01 9.26964223e-01
8.51871312e-01 -8.93164814e-01 1.92779228e-01 5.30170858e-01
-1.84766904e-01 -7.54806042e-01 -8.10824275e-01 -3.29275250e-01
7.38146722e-01 -7.10766137e-01 4.77660596e-01 1.12009442e+00
2.81380266e-01 7.74131835e-01 3.16773862e-01 -2.02142373e-01
2.19989017e-01 1.25803635e-01 1.54773980e-01 -1.43867660e+00
-4.10032541e-01 -6.79985225e-01 -1.94572523e-01 -9.62955475e-01
4.96348411e-01 -1.27893317e+00 1.81990564e-01 -1.34509909e+00
-3.06465495e-02 -1.10636219e-01 -2.01538265e-01 4.18674260e-01
-1.70886233e-01 -3.30397606e-01 3.44263077e-01 7.14184865e-02
7.19593316e-02 3.21669579e-01 9.51092422e-01 2.52231002e-01
-1.48832247e-01 -3.76657128e-01 -9.07200336e-01 1.26800954e+00
8.47199857e-01 -3.57712835e-01 -3.92490119e-01 -7.07451701e-01
8.68641019e-01 -3.47085968e-02 9.19823498e-02 -6.51628435e-01
1.64628308e-02 -5.38946748e-01 3.26729923e-01 -1.42544523e-01
-1.54778257e-01 -8.09838951e-01 -6.62247837e-02 8.36798012e-01
-3.85253280e-01 2.51915604e-01 1.15087688e-01 -4.17918079e-02
-7.14490563e-02 -5.12724757e-01 8.00090730e-01 -5.01808167e-01
-8.71857762e-01 -1.16161242e-01 -1.17230272e+00 -2.54273061e-02
7.92333961e-01 -4.09358829e-01 -2.84438627e-03 -2.44608857e-02
-9.18666184e-01 -2.00856134e-01 2.95630008e-01 4.37890172e-01
8.17635179e-01 -1.20233500e+00 -5.97825527e-01 4.33239281e-01
1.99383926e-02 -1.56509817e-01 -5.76112308e-02 4.06280339e-01
-4.77539331e-01 4.66886729e-01 -8.09924453e-02 -4.43863869e-01
-6.54426992e-01 7.99759030e-01 3.82912785e-01 1.98054407e-02
-3.72052610e-01 6.93234682e-01 8.36300075e-01 -5.44304490e-01
-3.00206482e-01 -6.24207973e-01 -5.06486535e-01 2.42394611e-01
6.26566410e-01 -4.39965814e-01 -4.53898609e-01 -8.36880147e-01
-2.26279512e-01 7.01884985e-01 -8.71847942e-02 -1.09708652e-01
1.27716899e+00 -1.06489263e-01 -6.53878093e-01 1.18391919e+00
8.98263872e-01 5.89186698e-02 -5.05619228e-01 -3.42807114e-01
4.11332756e-01 -1.24640819e-02 -5.28184116e-01 -5.57187140e-01
-6.93162262e-01 1.16457283e+00 2.63834924e-01 5.23836851e-01
9.89089489e-01 3.81945163e-01 3.34000349e-01 4.56128746e-01
3.69150877e-01 -9.79382753e-01 -1.59938522e-02 1.11153269e+00
1.02543020e+00 -9.37183142e-01 -3.39782536e-01 -3.44877273e-01
-1.67951271e-01 1.32421839e+00 5.58295667e-01 -3.45928967e-01
7.01766372e-01 -1.30521521e-01 -1.43429160e-01 -2.58475721e-01
-1.11468983e+00 -3.89816195e-01 2.56931603e-01 2.93131083e-01
1.09861517e+00 1.76750228e-01 -8.59956026e-01 6.33237541e-01
-6.81406975e-01 -4.35060561e-01 4.60839212e-01 7.89580464e-01
-1.02892721e+00 -1.14175510e+00 -2.83240587e-01 3.58438015e-01
-4.40173417e-01 -5.76706350e-01 -2.98111558e-01 6.55362427e-01
6.52586401e-01 6.55249655e-01 3.86281997e-01 -1.75127178e-01
6.65894225e-02 6.06138885e-01 5.80525219e-01 -1.34129333e+00
-6.65475667e-01 -2.13653743e-01 -2.36034498e-01 -1.80093184e-01
-4.04615253e-01 -7.76562393e-01 -1.65157342e+00 -3.74983363e-02
3.06734648e-02 2.20806673e-01 2.80609637e-01 1.28045177e+00
-2.24397570e-01 3.73877943e-01 2.73173273e-01 -4.92388368e-01
-2.63549179e-01 -9.47656989e-01 -4.76315260e-01 3.59771788e-01
2.76415437e-01 -2.87247747e-01 -3.29279840e-01 2.18323112e-01]
|
[10.340147972106934, 8.873058319091797]
|
4a4aca33-c4d1-449c-8e8d-dbbeaa9b602c
|
diable-efficient-dialogue-state-tracking-as
|
2305.1702
| null |
https://arxiv.org/abs/2305.17020v1
|
https://arxiv.org/pdf/2305.17020v1.pdf
|
Diable: Efficient Dialogue State Tracking as Operations on Tables
|
Sequence-to-sequence state-of-the-art systems for dialogue state tracking (DST) use the full dialogue history as input, represent the current state as a list with all the slots, and generate the entire state from scratch at each dialogue turn. This approach is inefficient, especially when the number of slots is large and the conversation is long. In this paper, we propose Diable, a new task formalisation that simplifies the design and implementation of efficient DST systems and allows one to easily plug and play large language models. We represent the dialogue state as a table and formalise DST as a table manipulation task. At each turn, the system updates the previous state by generating table operations based on the dialogue context. Extensive experimentation on the MultiWoz datasets demonstrates that Diable (i) outperforms strong efficient DST baselines, (ii) is 2.4x more time efficient than current state-of-the-art methods while retaining competitive Joint Goal Accuracy, and (iii) is robust to noisy data annotations due to the table operations approach.
|
['Lluis Marquez', 'Chao Shang', 'Momchil Hardalov', 'Yoshinari Fujinuma', 'Pietro Lesci']
|
2023-05-26
| null | null | null | null |
['dialogue-state-tracking']
|
['natural-language-processing']
|
[ 1.47686228e-01 5.65187633e-01 -2.29664251e-01 -3.13748270e-01
-1.24261928e+00 -9.74797666e-01 1.03802598e+00 2.95884997e-01
-4.07038093e-01 9.37453985e-01 7.13572085e-01 -5.11950374e-01
4.19170588e-01 -4.26222295e-01 -2.68880904e-01 -8.21792558e-02
-5.44822365e-02 1.32163858e+00 5.11338711e-01 -9.73470330e-01
2.31872559e-01 -8.26061442e-02 -1.03904283e+00 5.57589173e-01
6.72095895e-01 6.44430339e-01 3.00382525e-02 1.06914830e+00
-5.57555199e-01 1.07031214e+00 -1.04678774e+00 -4.47113007e-01
1.53793514e-01 -8.66065741e-01 -1.48286772e+00 2.26328686e-01
2.38573194e-01 -4.83811557e-01 -4.60243404e-01 6.15827978e-01
5.53502321e-01 4.33273077e-01 9.06694904e-02 -1.10450423e+00
2.58950770e-01 8.74031723e-01 -7.42930695e-02 -5.04389964e-02
9.73045826e-01 2.59758890e-01 1.12330532e+00 -5.29875755e-01
9.68224943e-01 1.68295932e+00 5.06883323e-01 8.97741616e-01
-1.35001278e+00 -1.79290786e-01 4.18959200e-01 -1.87312767e-01
-8.09922516e-01 -9.59436238e-01 2.39228308e-01 -6.88594058e-02
1.59756005e+00 4.49516445e-01 6.48683608e-01 8.51386189e-01
-8.39486793e-02 1.04137242e+00 9.51406062e-01 -7.13687718e-01
1.52801275e-01 -1.18169539e-01 1.90077692e-01 8.91470373e-01
-6.16285741e-01 -5.40401936e-01 -8.76150370e-01 -4.90231544e-01
3.94387960e-01 -6.39617264e-01 -1.76310744e-02 -2.76469648e-01
-1.36330771e+00 7.99586594e-01 -2.32330188e-01 3.47242435e-03
-4.71825302e-02 -1.34579048e-01 8.89879704e-01 5.48206568e-01
2.65874296e-01 5.94399273e-01 -4.84877735e-01 -9.94037032e-01
-5.77429891e-01 7.67521262e-01 1.63538563e+00 1.02451265e+00
3.40811074e-01 -1.58137783e-01 -4.93675083e-01 1.06721771e+00
1.73429977e-02 1.36360303e-01 3.46739888e-01 -1.34406650e+00
8.95840168e-01 6.17064118e-01 5.24696708e-01 -1.97643772e-01
-4.46692377e-01 2.20183000e-01 -3.85624319e-01 -1.00662269e-01
9.32431698e-01 -4.63514388e-01 -5.35260022e-01 1.84720278e+00
4.78203624e-01 -4.31323349e-01 3.84505093e-01 5.57608426e-01
7.99116373e-01 8.88359427e-01 -2.39557382e-02 -4.54170763e-01
1.55883789e+00 -1.11391020e+00 -1.15253282e+00 -7.38296747e-01
1.02890778e+00 -7.74472296e-01 1.01449609e+00 3.87882024e-01
-1.52786398e+00 -2.60967731e-01 -8.96345913e-01 -3.37819099e-01
-1.53506920e-01 -9.79608446e-02 7.40148425e-01 4.75158274e-01
-1.23390436e+00 3.99800301e-01 -8.86562288e-01 -5.36818206e-01
-3.59653920e-01 2.51521379e-01 -2.35091701e-01 1.60686716e-01
-1.54418552e+00 1.29203212e+00 6.40768886e-01 7.99024031e-02
-7.00022757e-01 -1.57271817e-01 -1.15112972e+00 -1.03351347e-01
8.53823483e-01 -3.64892900e-01 2.36856580e+00 -2.54419386e-01
-2.12695169e+00 6.85841262e-01 -4.38304037e-01 -4.89755690e-01
6.49962425e-01 -3.35740179e-01 -2.85532605e-02 -9.58388075e-02
1.57523438e-01 5.88079333e-01 7.23529533e-02 -8.36568713e-01
-7.68761814e-01 -2.41043702e-01 3.86211336e-01 7.93465376e-01
1.92192316e-01 3.02930921e-01 -9.72245514e-01 -1.31974384e-01
6.23770989e-02 -1.16831589e+00 -3.54313016e-01 -4.22000289e-01
-6.84190392e-01 -5.46992421e-01 5.24036467e-01 -8.42732847e-01
1.70944190e+00 -1.78504717e+00 2.94131488e-01 -2.26564288e-01
9.67251658e-02 3.74456644e-01 1.06289171e-01 1.09816098e+00
2.48557925e-01 -1.86269775e-01 -1.83402151e-01 -6.47921503e-01
3.32982898e-01 3.63931984e-01 -6.46113157e-02 8.79214108e-02
-1.05046354e-01 8.61060977e-01 -9.34515178e-01 -5.86738169e-01
2.67691314e-01 -1.89306095e-01 -6.50882602e-01 5.51047862e-01
-7.94124067e-01 2.65525609e-01 -3.17604005e-01 5.33218607e-02
2.71527395e-02 -2.07728624e-01 8.45842719e-01 1.61488861e-01
-4.17827845e-01 1.35566902e+00 -1.18320799e+00 2.16718745e+00
-3.87564570e-01 5.57859302e-01 2.20916837e-01 -4.93851691e-01
7.43169785e-01 5.58137000e-01 1.70240492e-01 -4.61363494e-01
3.02576125e-02 1.83271337e-02 1.03533883e-02 -2.73944110e-01
8.42867255e-01 7.24510550e-02 -6.89475954e-01 8.79068494e-01
1.84423238e-01 -3.17455739e-01 6.22742116e-01 7.12401092e-01
9.94064331e-01 1.36631474e-01 6.38721704e-01 3.84493209e-02
4.48228568e-01 3.39457661e-01 3.97266269e-01 8.95262539e-01
-1.62143916e-01 6.75396174e-02 1.10170364e+00 -3.78939211e-01
-1.04462922e+00 -6.42000854e-01 3.21146667e-01 1.58662832e+00
-2.88100410e-02 -9.45915461e-01 -9.79137182e-01 -6.89127386e-01
-2.31540158e-01 8.81194115e-01 -3.58021587e-01 7.46163204e-02
-9.29375172e-01 -3.79713804e-01 7.95988262e-01 2.93290973e-01
8.48172307e-01 -9.98459995e-01 -3.75697702e-01 6.18257582e-01
-8.16120386e-01 -1.19429898e+00 -7.48198569e-01 2.31717944e-01
-5.17899454e-01 -8.30018222e-01 -3.39422405e-01 -7.30314493e-01
4.35971245e-02 -1.76995128e-01 1.27433372e+00 -1.01170853e-01
2.00421989e-01 9.77740972e-04 -2.21801832e-01 -1.54377088e-01
-1.19258213e+00 2.99921006e-01 -1.18072212e-01 -3.85538548e-01
2.33531579e-01 1.46470636e-01 -1.14002384e-01 2.46155798e-01
-4.13820744e-01 4.72490460e-01 -3.95354873e-04 1.00390482e+00
-5.53732701e-02 -2.03358680e-01 4.36169118e-01 -1.35018790e+00
1.11165190e+00 -2.28263084e-02 -4.78772998e-01 4.13185745e-01
-4.33716089e-01 3.92913759e-01 2.60660768e-01 -5.12336865e-02
-1.41739678e+00 5.80706783e-02 -4.64003503e-01 3.74539167e-01
-1.09456651e-01 4.41937566e-01 -2.16045976e-01 5.59644639e-01
7.08483160e-01 2.70476103e-01 3.97101909e-01 -5.41744173e-01
5.28248489e-01 8.16504300e-01 6.76929832e-01 -7.91471541e-01
3.54923218e-01 -9.60659608e-02 -4.38511759e-01 -6.39382303e-01
-8.63817394e-01 -6.50110066e-01 -8.37786615e-01 -1.76257253e-01
7.23530829e-01 -8.66253793e-01 -1.03343356e+00 6.56100333e-01
-1.33385193e+00 -9.72781181e-01 -3.93204279e-02 3.82749140e-02
-7.23105609e-01 5.36966741e-01 -1.17258203e+00 -1.13385606e+00
-4.52761322e-01 -1.05563056e+00 1.01223087e+00 9.59848017e-02
-8.22752059e-01 -1.06958973e+00 9.26867649e-02 3.92863095e-01
1.68921009e-01 -6.07966669e-02 8.99540782e-01 -9.97453928e-01
-2.52770096e-01 -1.22116685e-01 1.67838886e-01 -1.03809148e-01
1.30473664e-02 -2.35794142e-01 -6.81289673e-01 -2.51523525e-01
-3.76240790e-01 -7.54671574e-01 3.75632226e-01 -1.12682290e-01
2.85501510e-01 -6.27473950e-01 -2.17206836e-01 -1.22336343e-01
6.91681683e-01 3.72978091e-01 4.33647156e-01 4.68381613e-01
2.06432074e-01 8.59127164e-01 8.96474183e-01 6.36395454e-01
8.12279940e-01 1.01203430e+00 -7.90631697e-02 3.31850015e-02
3.32378899e-03 -4.41279769e-01 5.72403669e-01 8.74570608e-01
3.84056956e-01 -4.31420773e-01 -9.69140291e-01 4.45150018e-01
-2.24914813e+00 -9.35415864e-01 6.70973435e-02 2.06370449e+00
1.47005355e+00 5.36606729e-01 5.84641695e-01 -4.08146605e-02
5.51766634e-01 4.26245034e-01 -4.14584041e-01 -8.66001785e-01
1.14015885e-01 -6.10852391e-02 1.42263219e-01 9.21836376e-01
-9.12291765e-01 1.46694839e+00 6.79833412e+00 4.62008059e-01
-5.80713928e-01 -1.94578022e-02 3.97123218e-01 2.66539883e-02
1.08957656e-01 9.81564969e-02 -1.05097556e+00 5.11134118e-02
1.12671673e+00 -3.10667813e-01 5.07956028e-01 6.15252852e-01
4.38943654e-02 -3.62762034e-01 -1.15768039e+00 5.67231357e-01
-9.63343009e-02 -1.21209288e+00 -1.89026520e-01 -1.79100066e-01
2.49289438e-01 -1.67466730e-01 -3.92826229e-01 7.52313435e-01
9.16017532e-01 -6.26427174e-01 7.17263341e-01 -7.01620709e-03
8.35412741e-01 -6.07999265e-01 5.24104118e-01 7.39055276e-01
-1.00379133e+00 9.31557938e-02 1.17630642e-02 -3.57466578e-01
4.29570824e-01 -3.37949991e-01 -1.43004894e+00 3.30867141e-01
3.49368662e-01 2.78788626e-01 -3.47155690e-01 5.43993175e-01
-2.61574626e-01 5.18851578e-01 -3.90189290e-01 -2.87653387e-01
4.67405200e-01 -3.85251269e-02 5.74602008e-01 1.51698446e+00
-3.58474374e-01 2.58117229e-01 7.18427300e-01 2.99744576e-01
-2.40241587e-02 -5.21208458e-02 -5.02532303e-01 -1.12272680e-01
7.34624445e-01 9.50570464e-01 -4.61849093e-01 -7.81840324e-01
-3.20977122e-01 1.06202781e+00 3.58587116e-01 1.11490905e-01
-3.39675456e-01 -4.43890154e-01 7.30387688e-01 -2.50890881e-01
3.78998043e-03 -4.54935491e-01 7.37703592e-02 -1.19976807e+00
-2.27111891e-01 -1.35319769e+00 6.52254879e-01 -6.57115996e-01
-7.10999846e-01 7.47233808e-01 2.02666432e-01 -7.39602864e-01
-1.04737234e+00 -2.13147745e-01 -4.61212426e-01 9.25975621e-01
-9.92359102e-01 -7.36666501e-01 2.54586209e-02 4.72047865e-01
1.16185880e+00 9.52379429e-04 1.31126726e+00 -1.34479791e-01
-6.06393516e-01 6.07508481e-01 -1.18751869e-01 5.62672377e-01
8.94129217e-01 -1.63345730e+00 1.13604009e+00 7.60340393e-01
-1.37786150e-01 7.53649235e-01 1.01952422e+00 -8.26426566e-01
-1.39924955e+00 -4.75065261e-01 1.30970919e+00 -6.85226142e-01
8.31507385e-01 -8.24087441e-01 -1.01627612e+00 8.63994896e-01
5.13915479e-01 -6.23992682e-01 5.57149172e-01 5.31973600e-01
-2.00296819e-01 3.38138908e-01 -9.06560481e-01 7.87311256e-01
8.63276064e-01 -6.94872320e-01 -8.94427955e-01 4.57289845e-01
7.70972788e-01 -1.25684977e+00 -9.91379917e-01 -8.43737870e-02
5.87844908e-01 -7.99490333e-01 5.61711073e-01 -7.92586327e-01
-1.11539541e-02 -3.02073061e-02 1.18029252e-01 -1.41231823e+00
-6.89398944e-02 -1.48385060e+00 -3.44018072e-01 1.16885245e+00
7.21182585e-01 -4.15966481e-01 7.74387479e-01 1.00117505e+00
-2.64835775e-01 -5.02831399e-01 -9.74150777e-01 -4.55130190e-01
-1.31025538e-01 -1.72369748e-01 5.39199412e-01 6.97967112e-01
8.47221613e-01 1.04805791e+00 -5.76448858e-01 -2.44620949e-01
1.75040528e-01 7.89466798e-02 1.11843538e+00 -9.61895347e-01
-4.21673805e-01 -3.35942686e-01 3.38158876e-01 -1.58071327e+00
7.62842894e-02 -4.52416092e-01 4.43072140e-01 -1.63321614e+00
-2.28199601e-01 -3.84815454e-01 3.44288260e-01 6.84649408e-01
-3.05722147e-01 -3.60104799e-01 5.21627009e-01 1.12650916e-01
-1.10213685e+00 3.45697701e-01 1.19687641e+00 3.33378185e-03
-4.58197057e-01 1.40810579e-01 -5.98322749e-01 4.41671729e-01
7.74711251e-01 -6.58595264e-02 -3.37994009e-01 -1.24300443e-01
-6.08895272e-02 9.75327075e-01 -3.14461142e-01 -7.05660820e-01
4.61014003e-01 -1.82098418e-01 -7.10702613e-02 -4.89977360e-01
8.01645398e-01 -1.40311658e-01 -2.69822538e-01 4.94151056e-01
-8.53779316e-01 1.98266998e-01 3.46876174e-01 3.22659463e-01
-7.86381289e-02 -1.57364622e-01 5.65113068e-01 -4.29217547e-01
-6.86847091e-01 -1.08819298e-01 -8.28937054e-01 4.61528242e-01
5.79805970e-01 -4.91459332e-02 -4.16112423e-01 -7.49850929e-01
-7.21880615e-01 6.07973456e-01 3.30898523e-01 4.82995480e-01
2.25096513e-02 -9.79158342e-01 -5.53972185e-01 -7.31940987e-03
2.26330996e-01 1.81804091e-01 -2.44061090e-02 4.15810466e-01
-4.27542865e-01 7.71148562e-01 -9.04801954e-03 -4.06607658e-01
-1.54863930e+00 1.29702196e-01 3.86875749e-01 -7.58094549e-01
-5.76717913e-01 8.19471657e-01 -2.68313646e-01 -6.87900484e-01
6.18677258e-01 -1.70436755e-01 -3.43031913e-01 1.59428060e-01
6.22939587e-01 1.40358955e-01 1.65098667e-01 -5.26629806e-01
-9.11022648e-02 -1.60663471e-01 -5.34257412e-01 -6.89802945e-01
8.88751268e-01 -3.29289049e-01 -5.42752333e-02 7.25796878e-01
7.59633183e-01 -4.82621342e-02 -1.25310695e+00 -6.47029936e-01
3.04401249e-01 -1.92732468e-01 -3.89042228e-01 -1.17380416e+00
-4.02366668e-02 5.32297075e-01 -8.08663219e-02 5.85529268e-01
5.43972850e-01 -6.27804846e-02 1.05270660e+00 8.63029122e-01
4.75285053e-01 -1.29254162e+00 1.55746430e-01 1.21264982e+00
7.10251272e-01 -1.11476469e+00 -1.30367890e-01 -3.52964759e-01
-1.20645320e+00 1.07306170e+00 8.52506459e-01 4.93003905e-01
2.04203334e-02 3.44998986e-01 3.91892165e-01 -1.51781961e-01
-1.41403675e+00 -1.77934483e-01 -5.86818345e-02 3.11840475e-01
7.03799486e-01 -1.28253445e-01 -1.22425936e-01 2.67916143e-01
-5.04529119e-01 -2.27572232e-01 6.23428047e-01 1.23877180e+00
-6.29153967e-01 -1.51703894e+00 -1.77212030e-01 3.17129105e-01
-4.78680342e-01 -4.16380167e-02 -8.13407123e-01 9.33755159e-01
-6.68587804e-01 1.24164319e+00 3.68308797e-02 -2.09632203e-01
6.29219830e-01 6.24944866e-01 3.67755443e-01 -1.07138801e+00
-1.02503908e+00 2.74060249e-01 1.22373497e+00 -5.97155929e-01
7.05093443e-02 -7.73991585e-01 -1.44135869e+00 -5.22947848e-01
-3.02807212e-01 6.92088842e-01 4.18560237e-01 9.54782605e-01
2.81331539e-01 2.92342693e-01 3.70422542e-01 -6.33983791e-01
-8.78489971e-01 -1.30023670e+00 -2.06770763e-01 2.39043057e-01
3.26483041e-01 -4.88484651e-01 6.93515912e-02 -1.45132333e-01]
|
[12.859674453735352, 7.9304118156433105]
|
09686b02-95d1-4e99-8342-bc6d713f7a22
|
optical-coherence-tomography-image
|
2306.1175
| null |
https://arxiv.org/abs/2306.11750v1
|
https://arxiv.org/pdf/2306.11750v1.pdf
|
Optical Coherence Tomography Image Enhancement via Block Hankelization and Low Rank Tensor Network Approximation
|
In this paper, the problem of image super-resolution for Optical Coherence Tomography (OCT) has been addressed. Due to the motion artifacts, OCT imaging is usually done with a low sampling rate and the resulting images are often noisy and have low resolution. Therefore, reconstruction of high resolution OCT images from the low resolution versions is an essential step for better OCT based diagnosis. In this paper, we propose a novel OCT super-resolution technique using Tensor Ring decomposition in the embedded space. A new tensorization method based on a block Hankelization approach with overlapped patches, called overlapped patch Hankelization, has been proposed which allows us to employ Tensor Ring decomposition. The Hankelization method enables us to better exploit the inter connection of pixels and consequently achieve better super-resolution of images. The low resolution image was first patch Hankelized and then its Tensor Ring decomposition with rank incremental has been computed. Simulation results confirm that the proposed approach is effective in OCT super-resolution.
|
['Hossein Rabbani', 'Andrzej Cichocki', 'Farnaz Sedighin']
|
2023-06-19
| null | null | null | null |
['image-super-resolution', 'image-enhancement', 'super-resolution']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[ 3.11581135e-01 -3.08875680e-01 5.91630749e-02 8.49542618e-02
-6.42620146e-01 3.17468792e-02 -7.07516745e-02 -3.74517143e-01
-1.29037723e-01 1.02623391e+00 1.99777618e-01 3.20266962e-01
-4.35880542e-01 -6.22709334e-01 -7.76405856e-02 -9.19644713e-01
-1.02883216e-03 7.06948861e-02 4.36376929e-01 1.87590092e-01
2.67715722e-01 3.23429048e-01 -1.50721121e+00 5.19202948e-01
1.20887947e+00 8.28997850e-01 3.76988083e-01 4.61623132e-01
1.08934790e-01 5.67335188e-01 1.88722033e-02 1.50969267e-01
3.11249226e-01 -5.17300248e-01 -7.81291485e-01 3.01085174e-01
5.37670374e-01 -5.98538160e-01 2.27821022e-02 1.27639651e+00
5.18056154e-01 -4.63066772e-02 3.67257267e-01 -1.29196167e-01
-3.43935668e-01 3.91981781e-01 -1.01292384e+00 5.19694030e-01
1.69002324e-01 -4.09376234e-01 7.67082989e-01 -1.05733681e+00
5.80965459e-01 1.11364782e+00 4.00889814e-01 2.82800436e-01
-1.73444879e+00 -6.08850777e-01 -4.20107514e-01 4.39186811e-01
-1.49445295e+00 -1.47779033e-01 7.94804931e-01 -5.38146257e-01
2.58169174e-01 2.09280193e-01 9.97773647e-01 4.80132043e-01
1.82597652e-01 3.26700807e-01 1.80609751e+00 -2.55423456e-01
-5.68062924e-02 -1.20716598e-02 1.50661901e-01 5.85022211e-01
6.83321118e-01 1.46298468e-01 -3.22615683e-01 -1.57963544e-01
1.52878118e+00 2.31656954e-02 -6.42929196e-01 -6.88204840e-02
-1.47793925e+00 4.05058563e-01 4.11800802e-01 9.10454273e-01
-5.99724770e-01 -1.56162128e-01 1.92015409e-01 -7.28351995e-02
4.43913609e-01 5.94728410e-01 3.26329321e-01 1.90923184e-01
-1.09846199e+00 4.04791124e-02 1.38737351e-01 3.29457611e-01
7.38027930e-01 1.42128214e-01 -1.35608509e-01 8.04299474e-01
-5.23850024e-02 3.06278944e-01 5.28125405e-01 -1.04586446e+00
2.06120625e-01 3.86136085e-01 2.49268606e-01 -9.09396887e-01
-2.60381937e-01 -6.09877110e-01 -1.33552134e+00 1.67245030e-01
2.09151149e-01 1.09491959e-01 -5.73515236e-01 9.22970653e-01
3.36611927e-01 6.00034773e-01 -6.12621754e-02 1.23403513e+00
5.22326350e-01 4.58341628e-01 -4.64533478e-01 -6.04231894e-01
1.44860494e+00 -4.30960834e-01 -9.44358349e-01 4.14943218e-01
5.69084994e-02 -1.11565053e+00 5.75234711e-01 7.09307611e-01
-1.12514734e+00 -6.85772955e-01 -1.09117341e+00 2.94547658e-02
4.32582617e-01 3.77867937e-01 1.85418248e-01 2.89419025e-01
-8.68988037e-01 8.88049841e-01 -9.07825291e-01 -1.33148298e-01
2.87971467e-01 1.37089103e-01 -3.65683109e-01 -3.55270028e-01
-6.90553367e-01 6.54345870e-01 4.23355579e-01 1.94280043e-01
-2.18250319e-01 -4.76113647e-01 -2.21803337e-01 -2.02971563e-01
8.35763365e-02 -8.22011352e-01 6.31568968e-01 -4.99571502e-01
-1.40661573e+00 4.58908856e-01 -4.14282054e-01 -2.50303298e-01
2.49918923e-01 6.32618368e-02 -4.23655301e-01 7.94550061e-01
2.85708606e-01 -1.25769168e-01 1.50495052e+00 -1.16934264e+00
-5.09904504e-01 -7.53252089e-01 -4.22610760e-01 6.48275912e-02
-5.79854138e-02 -3.22425336e-01 1.16488606e-01 -5.74433863e-01
8.55338275e-01 -7.16162682e-01 -3.40662003e-01 -4.63721812e-01
-2.89012760e-01 4.57761228e-01 6.46394074e-01 -7.55558729e-01
1.42623758e+00 -2.12753415e+00 1.29033327e-01 2.66549021e-01
8.39711547e-01 2.54287988e-01 8.53610709e-02 1.51437476e-01
-3.04151237e-01 -3.30643356e-02 -8.91471580e-02 1.21505074e-01
-9.52359319e-01 -4.35994416e-02 8.95130560e-02 5.14255226e-01
8.06988776e-02 3.07735234e-01 -7.76436329e-01 -7.49490440e-01
3.00629437e-01 8.23758781e-01 -6.47772908e-01 -2.63053402e-02
3.15938830e-01 1.29856920e+00 -8.60236943e-01 5.96970379e-01
1.29243350e+00 -4.80080456e-01 2.81953998e-02 -9.28934216e-01
-7.03141451e-01 -2.13329807e-01 -1.31020868e+00 1.20535588e+00
-3.99571121e-01 4.04649496e-01 2.14902371e-01 -7.21143246e-01
8.46095622e-01 5.24926066e-01 8.24940503e-01 -5.30038893e-01
-1.29315583e-02 4.65317935e-01 -1.35172289e-02 -7.34689653e-01
3.51002038e-01 -5.00791848e-01 9.02322114e-01 2.12612107e-01
-6.43376172e-01 3.76357101e-02 3.58177945e-02 -3.19368243e-01
7.70908773e-01 -1.96036994e-01 2.13923246e-01 -1.84221447e-01
8.82998586e-01 -2.46492680e-02 3.76708984e-01 2.01886311e-01
1.58932194e-01 6.81455135e-01 3.74974310e-01 -4.85189945e-01
-1.46390998e+00 -7.62248039e-01 -7.78528512e-01 -1.60140127e-01
1.87475353e-01 -3.02388608e-01 -6.96893513e-01 1.58288553e-01
-2.50069052e-01 -2.00648438e-02 -4.76623267e-01 3.32084537e-01
-6.69068933e-01 -7.74923027e-01 2.55678147e-02 1.50964916e-01
1.12146056e+00 -1.91233650e-01 -6.10930800e-01 3.40826184e-01
-5.61709106e-01 -1.40877783e+00 -2.09257200e-01 -7.85709560e-01
-1.60708463e+00 -1.03792834e+00 -1.15614021e+00 -5.97890198e-01
8.49776268e-01 6.89042211e-01 4.39088851e-01 -2.06081569e-01
-4.66573060e-01 2.33594198e-02 -4.33739245e-01 3.30975294e-01
3.18572216e-04 -4.88837659e-01 -6.33457722e-03 8.30831230e-01
-7.25430846e-02 -9.48369145e-01 -1.05114591e+00 3.58726203e-01
-8.31393480e-01 3.07124943e-01 1.11859143e+00 1.16648257e+00
1.13718402e+00 5.17235518e-01 1.13943063e-01 -7.48786092e-01
5.32786608e-01 4.31620181e-02 -8.44329059e-01 -1.53737560e-01
-5.31173468e-01 7.07468390e-02 5.05496264e-01 -2.41959020e-01
-1.08419597e+00 -2.44136259e-01 3.70166004e-01 -6.75606370e-01
2.26413965e-01 6.25451922e-01 2.86908716e-01 -5.86392283e-01
5.87024152e-01 4.45225924e-01 1.78050980e-01 -8.48628223e-01
-2.02011257e-01 5.03515959e-01 2.12772578e-01 -4.08596814e-01
6.82403862e-01 9.50440049e-01 6.02259278e-01 -1.40186226e+00
-6.09272838e-01 -5.85879982e-01 -6.82471633e-01 -8.24459791e-02
1.09316421e+00 -7.85326183e-01 -7.97683418e-01 2.76111126e-01
-1.08225501e+00 3.49911541e-01 1.11705936e-01 1.20817268e+00
-3.91393960e-01 8.30623209e-01 -6.60893023e-01 -7.06342816e-01
-2.26804212e-01 -1.13247848e+00 8.41208398e-01 4.83643487e-02
3.21441740e-01 -5.55616319e-01 6.28732741e-02 7.22471178e-01
4.15245295e-01 4.05808300e-01 1.01030052e+00 3.43810558e-01
-1.15909433e+00 -4.30690087e-02 -8.63342702e-01 5.87125242e-01
1.73164889e-01 -2.43845433e-01 -6.38649642e-01 -3.36007923e-01
5.06517112e-01 3.48564297e-01 6.36174858e-01 6.83403730e-01
1.04141080e+00 -2.75164902e-01 -1.21271707e-01 7.76519835e-01
1.95274818e+00 -2.08228856e-01 9.17169392e-01 2.98954219e-01
7.04597652e-01 4.27242815e-01 8.61624599e-01 6.64663136e-01
-1.96254149e-01 7.65629530e-01 2.17558414e-01 -4.46677580e-02
-1.87594727e-01 8.27435777e-02 -2.18432099e-01 9.79399323e-01
-1.03931665e+00 7.15754032e-01 -4.86600876e-01 4.85493124e-01
-1.73521924e+00 -9.87358570e-01 -5.06708682e-01 2.44691491e+00
6.53167963e-01 -3.76173377e-01 -7.41595179e-02 1.69971421e-01
8.27220380e-01 -1.02243414e-02 -3.91812503e-01 3.82107757e-02
-3.20352837e-02 4.62993741e-01 4.78267372e-01 5.47683895e-01
-6.50831699e-01 4.74827707e-01 5.76725435e+00 8.27351630e-01
-1.31685627e+00 3.24391693e-01 -5.53962030e-03 6.43957481e-02
-9.90217552e-02 6.28212020e-02 -6.85471654e-01 3.84534031e-01
3.66375685e-01 -2.46206015e-01 4.33240265e-01 1.54862136e-01
6.82705164e-01 -4.44465995e-01 -6.92214787e-01 1.46888685e+00
-2.51353770e-01 -1.13857949e+00 3.42745245e-01 4.41487670e-01
7.61542201e-01 -4.15108114e-01 2.58643627e-01 -6.09874368e-01
-5.03802717e-01 -9.61354792e-01 -2.86410093e-01 7.24197149e-01
1.03794134e+00 -6.38337255e-01 7.07443714e-01 5.15766926e-02
-1.27586377e+00 3.31892073e-02 -6.12647891e-01 1.27577886e-01
1.30492859e-02 1.16856730e+00 -8.11839223e-01 8.28037977e-01
6.42184079e-01 9.24021482e-01 -1.48385614e-01 1.20605624e+00
1.80076078e-01 2.91280776e-01 3.65927024e-03 4.21403885e-01
4.91224881e-03 -7.55146742e-01 1.12281275e+00 5.04850686e-01
7.73401380e-01 5.81503630e-01 -6.22548945e-02 9.79304552e-01
6.25785530e-01 3.63557816e-01 -4.83791977e-01 -3.41206826e-02
8.83989856e-02 1.25432551e+00 -3.85147899e-01 -1.70059338e-01
-4.43680406e-01 7.58500993e-01 -3.30267310e-01 5.57549596e-01
-3.68215144e-01 -6.89582378e-02 4.50767875e-01 9.34021235e-01
4.45526510e-01 -3.02481204e-01 -2.44483754e-01 -1.51641333e+00
1.55582532e-01 -7.83190191e-01 6.88635409e-02 -8.21215510e-01
-8.58079553e-01 7.54839361e-01 -7.32856020e-02 -1.84350574e+00
-1.98396016e-03 -4.66501087e-01 -1.29700661e-01 1.11355674e+00
-1.63095498e+00 -1.20031857e+00 -4.24534619e-01 9.06889617e-01
1.25682831e-01 9.88410637e-02 4.51935440e-01 4.11453605e-01
-3.57348621e-01 5.12935268e-03 3.11179817e-01 -1.40136093e-01
5.81132829e-01 -9.09749866e-01 -7.77051806e-01 7.65650988e-01
-6.18727326e-01 7.08246052e-01 5.65278590e-01 -8.57776403e-01
-1.03549790e+00 -9.27498579e-01 5.00607371e-01 1.89024538e-01
5.64808249e-01 4.32315916e-01 -1.03852069e+00 5.60467899e-01
-1.20006921e-02 -4.64244261e-02 6.16063178e-01 -1.94835842e-01
-6.17766418e-02 -4.56394851e-01 -1.28043246e+00 4.08088684e-01
4.68673170e-01 -3.98110986e-01 -4.49253350e-01 1.75317109e-01
2.24313721e-01 -2.41358206e-01 -1.38233542e+00 6.51269495e-01
7.10876703e-01 -1.28667438e+00 8.97193670e-01 -1.22527592e-01
3.09363067e-01 -6.38325810e-01 1.90674122e-02 -1.08078039e+00
-7.78726995e-01 -5.07589698e-01 2.51407087e-01 7.40901232e-01
2.95818429e-02 -9.72156107e-01 7.06221461e-01 1.43280134e-01
1.67448550e-01 -6.07104123e-01 -9.70636308e-01 -7.47992039e-01
-4.90308732e-01 3.24408233e-01 2.53845274e-01 9.44242954e-01
-4.47710231e-02 3.99744153e-01 -4.69077140e-01 4.16042030e-01
1.35492194e+00 3.72148484e-01 2.96453893e-01 -1.37670135e+00
-3.75997514e-01 -2.04484701e-01 -6.27557158e-01 -7.67185748e-01
-5.06944776e-01 -6.21445537e-01 -5.70523858e-01 -1.37723386e+00
4.28977221e-01 -3.52822900e-01 -1.92179188e-01 -3.39580059e-01
1.57952845e-01 5.05665541e-01 -1.70366243e-01 7.21378863e-01
2.19496340e-02 2.27549478e-01 2.19687700e+00 1.33721814e-01
-2.73247063e-01 1.68829471e-01 -3.27531308e-01 5.52777767e-01
5.42239904e-01 1.02239214e-02 -3.84277433e-01 -3.26137304e-01
1.57466054e-01 4.04952019e-01 5.16761005e-01 -1.16372740e+00
1.02994286e-01 1.11165322e-01 2.55142748e-01 -7.47549117e-01
6.32906497e-01 -7.86061168e-01 4.41865623e-01 3.65898252e-01
5.62822372e-02 -3.88252646e-01 -7.87895396e-02 4.98836219e-01
-6.96518242e-01 -1.05314091e-01 1.35246384e+00 -4.10310775e-01
-3.88222128e-01 3.80713940e-01 -1.54128253e-01 -3.75813782e-01
8.11058521e-01 -6.31307542e-01 -6.30610064e-02 -1.00290105e-01
-1.24582636e+00 -2.06281364e-01 4.09395188e-01 -4.52547342e-01
1.17134798e+00 -1.38180578e+00 -1.05234635e+00 3.05114716e-01
-7.20393611e-03 -1.25067040e-01 8.40163887e-01 1.59910631e+00
-7.61014223e-01 5.10886014e-01 -5.06941140e-01 -8.53087664e-01
-1.51517701e+00 4.12201792e-01 3.35198998e-01 -4.52692538e-01
-8.93042684e-01 3.52287918e-01 4.42839175e-01 4.86335605e-01
-4.71351385e-01 -3.90339673e-01 -5.87671340e-01 -1.07676759e-01
8.64991724e-01 8.23265672e-01 -5.03202751e-02 -6.92444861e-01
1.28373191e-01 1.55225515e+00 -2.53680587e-01 -2.51675308e-01
1.24485314e+00 -4.71696347e-01 -8.12351286e-01 1.91313982e-01
1.13376498e+00 3.00832868e-01 -6.24974906e-01 -4.66566294e-01
-3.14446807e-01 -1.10237968e+00 5.52514374e-01 -2.36129761e-01
-9.61570561e-01 9.58801746e-01 6.46339715e-01 1.87632382e-01
1.42505908e+00 -2.48155400e-01 9.60472167e-01 -2.78628115e-02
4.63214993e-01 -7.73277581e-01 1.43111855e-01 -1.52720120e-02
9.61236358e-01 -9.32362616e-01 1.95112556e-01 -9.53099132e-01
-1.41970560e-01 1.20098662e+00 2.96974182e-01 -2.12411165e-01
5.40330708e-01 -2.00669661e-01 -3.41613531e-01 -1.35369763e-01
-4.37038451e-01 -4.47206527e-01 2.64097482e-01 4.10893530e-01
4.24067974e-01 7.96226487e-02 -8.07410717e-01 2.29656234e-01
1.80783167e-01 4.11959738e-01 7.35820591e-01 3.66552889e-01
-5.49003601e-01 -1.07780421e+00 -8.24285805e-01 4.61406231e-01
-7.04441786e-01 -1.65969223e-01 1.71357751e-01 3.95014524e-01
1.49024218e-01 7.41782725e-01 -1.63674727e-01 -3.93025488e-01
2.21320704e-01 -3.63705724e-01 1.00071526e+00 -6.28539562e-01
3.04796338e-01 7.19927907e-01 -4.68136333e-02 -7.08774149e-01
-5.26416540e-01 -6.10054314e-01 -9.82982337e-01 -1.28505498e-01
-1.86133489e-01 3.27343076e-01 5.67623734e-01 3.91171217e-01
3.17380607e-01 3.10011446e-01 8.64150524e-01 -6.42072320e-01
-2.76080728e-01 -8.36873293e-01 -1.36154640e+00 4.13874313e-02
7.61982679e-01 -6.31951213e-01 -3.49668145e-01 -7.01051429e-02]
|
[11.103544235229492, -2.409626007080078]
|
c771d91f-7bc3-446a-8bff-d01f0b42368b
|
extending-context-window-of-large-language
|
2306.15595
| null |
https://arxiv.org/abs/2306.15595v2
|
https://arxiv.org/pdf/2306.15595v2.pdf
|
Extending Context Window of Large Language Models via Positional Interpolation
|
We present Position Interpolation (PI) that extends the context window sizes of RoPE-based pretrained LLMs such as LLaMA models to up to 32768 with minimal fine-tuning (within 1000 steps), while demonstrating strong empirical results on various tasks that require long context, including passkey retrieval, language modeling, and long document summarization from LLaMA 7B to 65B. Meanwhile, the extended model by Position Interpolation preserve quality relatively well on tasks within its original context window. To achieve this goal, Position Interpolation linearly down-scales the input position indices to match the original context window size, rather than extrapolating beyond the trained context length which may lead to catastrophically high attention scores that completely ruin the self-attention mechanism. Our theoretical study shows that the upper bound of interpolation is at least $\sim 600 \times$ smaller than that of extrapolation, further demonstrating its stability. Models extended via Position Interpolation retain its original architecture and can reuse most pre-existing optimization and infrastructure.
|
['Yuandong Tian', 'Liangjian Chen', 'Sherman Wong', 'Shouyuan Chen']
|
2023-06-27
| null | null | null | null |
['retrieval', 'document-summarization']
|
['methodology', 'natural-language-processing']
|
[ 1.94988191e-01 2.10633188e-01 -6.22693956e-01 -3.79777580e-01
-1.08286750e+00 -5.49081683e-01 4.92154330e-01 2.95715272e-01
-8.02306354e-01 8.67421269e-01 4.60488498e-01 -7.78046310e-01
1.50294993e-02 -4.56558764e-01 -9.39830840e-01 -2.43346363e-01
-3.28591615e-01 -6.81411996e-02 2.90997654e-01 -1.07831836e-01
6.29356503e-01 2.96190470e-01 -1.14015806e+00 4.99680042e-01
9.60734606e-01 7.27560043e-01 2.76561439e-01 9.35224473e-01
-2.94803739e-01 6.52739048e-01 -8.68996739e-01 -2.44301423e-01
9.65857208e-02 -1.14725702e-01 -9.51530218e-01 -4.60446775e-01
8.16552639e-01 -6.51034534e-01 -4.58661914e-01 4.88522738e-01
2.62721419e-01 4.57696736e-01 2.64021933e-01 -7.89320529e-01
-9.45040584e-01 9.97397423e-01 -7.22713590e-01 5.57547033e-01
-7.05154538e-02 1.87439188e-01 1.32300317e+00 -8.70716333e-01
3.40877682e-01 1.07056987e+00 7.63146639e-01 4.60115075e-01
-1.04583716e+00 -5.83697677e-01 5.64555645e-01 4.10835333e-02
-1.09066868e+00 -3.47517192e-01 1.91980302e-01 9.37706009e-02
1.68195188e+00 4.46408331e-01 3.91044617e-01 8.93979192e-01
4.25556213e-01 5.77120364e-01 7.60313451e-01 -4.32334244e-01
8.31755102e-02 -2.09837437e-01 3.71195942e-01 4.78556931e-01
4.58403260e-01 -3.18810374e-01 -7.23003805e-01 -4.92780246e-02
5.65609872e-01 -3.38369042e-01 -2.25424498e-01 3.08856070e-01
-1.31627381e+00 6.60403430e-01 6.15007579e-01 3.24214816e-01
-1.76386878e-01 5.99037886e-01 4.18293655e-01 3.29856724e-01
4.17345226e-01 9.50281262e-01 -5.38950384e-01 -2.95556098e-01
-1.41738534e+00 2.64911860e-01 6.17533684e-01 9.91784453e-01
5.89403450e-01 6.01019897e-02 -5.34165740e-01 6.36142373e-01
-1.39558166e-01 4.03823406e-01 3.68879288e-01 -1.33948112e+00
8.09868753e-01 3.63486528e-01 2.72957683e-01 -6.68405831e-01
-4.30283427e-01 -7.21262634e-01 -6.06619358e-01 -3.51745598e-02
6.04684114e-01 -3.39567840e-01 -6.87807083e-01 1.90287948e+00
-1.54910848e-01 1.12867281e-01 -1.39899418e-01 6.55163884e-01
4.86584514e-01 1.09685659e+00 1.86946452e-01 -1.69275522e-01
1.29955363e+00 -1.36534011e+00 -4.01759714e-01 -7.04270184e-01
7.60018885e-01 -7.91225195e-01 1.87510967e+00 3.50795299e-01
-1.43572664e+00 -4.47081625e-01 -1.48759556e+00 -6.32122755e-01
-1.14221312e-01 2.76824720e-02 7.94553101e-01 4.29366231e-01
-1.21259916e+00 9.48974073e-01 -7.05000401e-01 -8.37102830e-02
3.16261917e-01 4.44002271e-01 -1.47034377e-01 6.51503280e-02
-1.05521703e+00 8.38862956e-01 2.89145201e-01 1.28783733e-01
-3.98737490e-01 -7.96125650e-01 -6.64753675e-01 5.45987785e-01
3.64312261e-01 -7.31705010e-01 1.11342132e+00 -5.46797633e-01
-1.36012983e+00 4.83597547e-01 -4.40908730e-01 -8.61352324e-01
2.36040279e-01 -6.36939466e-01 -9.56514180e-02 1.55892313e-01
-1.97804540e-01 9.08994496e-01 6.70055568e-01 -9.39773917e-01
-4.28085327e-01 9.47650149e-03 4.73198175e-01 3.50210369e-01
-7.71742344e-01 1.12540638e-02 -6.74675226e-01 -6.17083132e-01
-2.07886308e-01 -7.74396777e-01 -2.48836070e-01 -3.14714342e-01
-2.45629966e-01 -2.05925480e-01 5.79821885e-01 -6.25585556e-01
1.82344186e+00 -1.75678241e+00 -2.00530395e-01 -6.87976629e-02
2.15517819e-01 3.94804180e-01 -6.12554669e-01 3.86473000e-01
2.35713437e-01 4.45040792e-01 -3.29239696e-01 -6.43392742e-01
-8.17371607e-02 3.13241839e-01 -5.25422633e-01 1.77048415e-01
1.26736164e-01 1.09191251e+00 -8.36673677e-01 -3.93172324e-01
-2.52351373e-01 2.35231683e-01 -9.59734917e-01 2.27957424e-02
-4.58988160e-01 1.35977361e-02 -1.49221927e-01 2.15498701e-01
4.56197560e-01 -5.30618131e-01 9.38838944e-02 -9.99834985e-02
-1.70499235e-01 7.83153594e-01 -7.59551823e-01 2.04929399e+00
-9.38898146e-01 1.05105126e+00 1.28437549e-01 -5.97664535e-01
6.12006843e-01 -9.44667682e-02 2.10464478e-01 -5.25483847e-01
-1.38302103e-01 1.09879673e-01 -6.26619086e-02 -3.82719994e-01
1.18522453e+00 1.68134972e-01 -8.87346733e-03 6.64803147e-01
-1.79709598e-01 -2.76282877e-02 3.75270814e-01 3.05126876e-01
1.05862713e+00 6.71665743e-02 -9.87234339e-02 -3.27597588e-01
3.95952910e-01 -9.22742784e-02 2.56096423e-01 1.06176662e+00
1.44175030e-02 4.54706728e-01 7.62700915e-01 -1.42571688e-01
-1.38206100e+00 -8.05380583e-01 -1.86492965e-01 1.58922923e+00
5.06784953e-02 -7.28757143e-01 -8.93068373e-01 -6.36217654e-01
-6.49327114e-02 1.00019109e+00 -6.19573295e-01 -1.26416996e-01
-9.51609969e-01 -8.88219237e-01 8.56537461e-01 7.62880623e-01
6.00779295e-01 -9.55599010e-01 -6.63765848e-01 2.51095504e-01
-2.69178718e-01 -9.59740222e-01 -9.56038237e-01 2.39459842e-01
-9.81710434e-01 -6.04927599e-01 -7.47426152e-01 -6.41067326e-01
4.35037911e-01 2.16883495e-01 1.22508180e+00 2.97686964e-01
3.80735621e-02 -1.70833632e-01 -8.01529288e-02 -2.16332197e-01
-2.09552079e-01 6.33209527e-01 -1.61015272e-01 -7.52200067e-01
2.40645304e-01 -5.63763022e-01 -9.27956760e-01 6.40667230e-02
-8.67834091e-01 -1.17944553e-01 7.05831409e-01 9.71959710e-01
3.29741895e-01 -4.49693233e-01 9.70969498e-01 -6.05433643e-01
8.43598247e-01 -3.88695002e-01 -4.58217770e-01 2.96322256e-01
-7.96121120e-01 3.28279048e-01 5.50109506e-01 -5.30897915e-01
-9.00973201e-01 -7.70393014e-01 -1.15593709e-01 -1.39686078e-01
3.26879680e-01 4.35035735e-01 1.87383726e-01 1.21665783e-01
7.65354037e-01 -2.06625715e-01 -1.46325439e-01 -4.07203764e-01
6.48439467e-01 4.94226843e-01 6.14717782e-01 -6.57258451e-01
3.84658217e-01 2.44430378e-01 -1.39753908e-01 -6.09795392e-01
-9.50536311e-01 -9.96507853e-02 -1.13203354e-01 3.09438288e-01
6.09234095e-01 -7.54998386e-01 -9.15958166e-01 -8.23114440e-03
-1.10277212e+00 -8.31100464e-01 -3.01761806e-01 2.98141062e-01
-3.07870775e-01 4.36717868e-01 -1.02055776e+00 -5.85507631e-01
-8.86344671e-01 -9.20740187e-01 8.89376044e-01 3.01092058e-01
-3.82415324e-01 -8.26572657e-01 -3.46674412e-01 1.78315833e-01
7.16903210e-01 -1.86123505e-01 1.16737151e+00 -4.58786607e-01
-5.30159235e-01 -6.35891482e-02 -4.10441041e-01 5.23096502e-01
-2.23961934e-01 -1.45765468e-01 -7.25080788e-01 -4.24503148e-01
-1.22956060e-01 -2.95985132e-01 1.27660835e+00 4.23931688e-01
1.27076459e+00 -5.23020208e-01 -1.43368408e-01 6.22166038e-01
1.21699560e+00 -6.68334514e-02 5.23323596e-01 5.57668865e-01
4.96310920e-01 2.44918525e-01 3.92815113e-01 2.53009081e-01
2.83413380e-01 4.50752646e-01 2.84150839e-01 -3.28547731e-02
-2.07525298e-01 -3.90134603e-01 6.05264962e-01 7.44794548e-01
1.37451276e-01 -5.08425117e-01 -7.75173962e-01 6.07880890e-01
-1.49632704e+00 -8.21149468e-01 1.75125077e-01 2.38638258e+00
1.15757716e+00 7.70453751e-01 -1.79983944e-01 -1.72722384e-01
3.90944004e-01 5.43681741e-01 -7.00266421e-01 -1.22510266e+00
-1.50733396e-01 2.08486676e-01 8.62327576e-01 7.73201048e-01
-7.87421286e-01 1.13135803e+00 7.47489691e+00 1.02750754e+00
-1.08172846e+00 -8.60289298e-03 8.28194201e-01 -7.37103879e-01
-5.78275800e-01 8.75364989e-02 -9.21335042e-01 3.38667214e-01
1.28274262e+00 -3.93391043e-01 6.29776955e-01 6.53357625e-01
3.83741707e-01 -3.17351699e-01 -1.22631991e+00 4.42588925e-01
-1.36529412e-02 -1.64417410e+00 1.86059892e-01 1.07648134e-01
9.16436911e-01 2.44910792e-01 2.81492949e-01 5.57906926e-01
1.68697879e-01 -1.20747185e+00 5.02749801e-01 1.58822000e-01
1.07906532e+00 -8.20508420e-01 5.01535356e-01 4.06233698e-01
-8.41300607e-01 -3.75342429e-01 -5.38534999e-01 -1.55639112e-01
2.74462372e-01 4.77384329e-01 -9.00705338e-01 1.99750960e-01
3.85382384e-01 1.31914482e-01 -7.53607094e-01 7.73879945e-01
-7.53002539e-02 9.95548844e-01 -5.76219857e-01 -8.32445472e-02
6.82663620e-01 5.31249046e-02 6.66053414e-01 1.53203392e+00
3.05645674e-01 -4.78983298e-02 -2.12596536e-01 8.08583200e-01
-4.97659981e-01 9.11782905e-02 -2.81557292e-01 2.05457062e-01
6.87904000e-01 9.44394410e-01 -2.70036012e-01 -5.18770635e-01
-3.33949178e-01 7.46602297e-01 3.68106008e-01 4.72152472e-01
-1.06278467e+00 -7.87448645e-01 4.56033021e-01 2.07623735e-01
3.49176705e-01 -6.04906082e-01 -8.18706095e-01 -9.43414688e-01
1.42367795e-01 -6.73278213e-01 5.38883023e-02 -6.92682803e-01
-7.14719713e-01 7.17423677e-01 1.51717231e-01 -7.29126871e-01
-1.51505321e-01 -3.80447835e-01 -8.36388767e-01 1.04310799e+00
-1.67549539e+00 -1.02495921e+00 3.87748219e-02 1.14486806e-01
7.36646771e-01 2.51258880e-01 5.47301412e-01 2.84894347e-01
-5.97502768e-01 1.18991435e+00 2.13770270e-01 -2.56063551e-01
8.51769447e-01 -1.18713212e+00 8.00728202e-01 8.44727159e-01
-6.30379841e-02 1.14064765e+00 7.25886583e-01 -4.63839620e-01
-1.13284636e+00 -8.83172750e-01 1.21482956e+00 -4.25155461e-01
8.12295318e-01 -3.60576868e-01 -1.12211251e+00 6.90429747e-01
4.91267681e-01 -2.29769737e-01 4.46927577e-01 2.34969899e-01
-3.83882880e-01 -1.76465347e-01 -7.97905564e-01 9.50975478e-01
1.12387633e+00 -4.13827807e-01 -7.43880928e-01 1.36539698e-01
1.10228944e+00 -5.11733055e-01 -8.23400199e-01 4.61614162e-01
6.73186123e-01 -8.98499370e-01 9.85630751e-01 -6.45959675e-01
7.60171294e-01 1.14360072e-01 2.37774849e-03 -9.46828544e-01
-2.06621155e-01 -8.63530695e-01 -4.41619128e-01 1.15652812e+00
6.76698208e-01 -4.64118063e-01 8.36891472e-01 7.00651646e-01
-3.58760417e-01 -1.19435585e+00 -8.46279740e-01 -7.95771599e-01
7.08020329e-01 -4.47784841e-01 6.62087917e-01 4.42898035e-01
1.36677638e-01 3.18918586e-01 -2.18350425e-01 2.27042977e-02
2.93695062e-01 1.08407542e-01 5.19973636e-01 -4.65971678e-01
-5.27577460e-01 -7.86233425e-01 5.60388267e-01 -1.82800353e+00
1.26507431e-01 -8.19684446e-01 -4.86476161e-02 -1.49313402e+00
7.31135756e-02 -7.04320788e-01 -5.06302536e-01 5.99445283e-01
-5.47653556e-01 2.85151273e-01 3.41839641e-01 1.68329000e-01
-5.89936137e-01 4.13654149e-01 1.26292717e+00 8.19783360e-02
-3.01328331e-01 -2.76821733e-01 -9.69963551e-01 7.01930761e-01
7.61844397e-01 -2.21138999e-01 -4.35377359e-01 -9.67972457e-01
4.43810970e-01 -1.03909433e-01 1.03249826e-01 -8.39482546e-01
2.91496478e-02 -2.30740309e-01 2.54208833e-01 -5.89669228e-01
2.15704471e-01 -2.53199846e-01 -5.30590832e-01 1.11017793e-01
-8.96919489e-01 2.93523967e-01 6.60185397e-01 4.07921582e-01
1.21468894e-01 -3.03469390e-01 4.46568012e-01 1.31434694e-01
-4.42289561e-01 -6.55477792e-02 -3.44598591e-01 2.08695531e-01
5.00893712e-01 -1.86447762e-02 -6.88137710e-01 -3.72356683e-01
-5.18026352e-01 5.37187994e-01 3.92715424e-01 3.65031093e-01
2.75897503e-01 -9.34507608e-01 -6.10364497e-01 -6.98588789e-02
-2.75306672e-01 1.83971807e-01 2.93337345e-01 6.16869986e-01
-4.65173900e-01 7.74805605e-01 1.91211998e-01 -2.25574046e-01
-9.95292485e-01 5.27609646e-01 1.24202482e-01 -7.45307386e-01
-7.34156072e-01 1.04728711e+00 9.01408792e-02 1.24706618e-01
3.73066962e-01 -7.73348510e-01 2.36170441e-02 -1.46552444e-01
7.37602711e-01 3.62178624e-01 2.01630652e-01 -5.34190647e-02
-2.52942175e-01 4.43495601e-01 -2.00868174e-01 -1.74242944e-01
1.05813694e+00 -3.25440139e-01 -8.60428531e-03 1.06302604e-01
1.24723911e+00 3.34597498e-01 -1.55731726e+00 -1.34987563e-01
7.24672154e-02 -2.17676207e-01 1.57223985e-01 -9.61253524e-01
-5.55258512e-01 9.47870553e-01 -3.53317298e-02 -2.98944041e-02
1.11943460e+00 -2.12578431e-01 1.24746752e+00 7.34315336e-01
4.92429584e-02 -1.08541298e+00 1.29444137e-01 6.63254142e-01
9.98167753e-01 -8.42585623e-01 2.46800378e-01 2.47582179e-02
-5.41392446e-01 9.34341133e-01 6.25645220e-01 -2.35454127e-01
9.27362368e-02 4.03683662e-01 -3.71177942e-01 2.95235634e-01
-1.28900695e+00 1.78754300e-01 2.99163759e-01 1.14999563e-01
5.61578333e-01 -2.39317223e-01 -6.29671812e-01 7.97768950e-01
-3.60595107e-01 3.52286398e-02 6.75958157e-01 7.56494343e-01
-6.54316962e-01 -1.08210492e+00 -2.88152784e-01 4.39762264e-01
-7.97999084e-01 -7.15587437e-01 1.95106283e-01 6.29713297e-01
-1.13863520e-01 9.88682866e-01 3.92915368e-01 -1.14860646e-01
2.47895300e-01 1.21112049e-01 3.99734795e-01 -4.69412655e-01
-8.27770889e-01 -1.81726757e-02 3.70505989e-01 -5.37872732e-01
1.50404036e-01 -2.02877268e-01 -1.39822578e+00 -6.20379090e-01
-3.90270263e-01 1.43971771e-01 6.53389633e-01 8.36034060e-01
5.90308845e-01 5.04593849e-01 4.59470868e-01 -9.13554430e-01
-8.60238850e-01 -1.04564631e+00 -8.62643719e-02 2.11463831e-02
5.12849748e-01 -1.98383138e-01 -4.06908780e-01 -1.22931488e-01]
|
[11.138648986816406, 8.189362525939941]
|
c3625ea1-fdbd-41d4-ab09-54498c72c3e9
|
generalized-representations-learning-for-time
|
2209.07027
| null |
https://arxiv.org/abs/2209.07027v4
|
https://arxiv.org/pdf/2209.07027v4.pdf
|
Out-of-Distribution Representation Learning for Time Series Classification
|
Time series classification is an important problem in real world. Due to its non-stationary property that the distribution changes over time, it remains challenging to build models for generalization to unseen distributions. In this paper, we propose to view the time series classification problem from the distribution perspective. We argue that the temporal complexity attributes to the unknown latent distributions within. To this end, we propose DIVERSIFY to learn generalized representations for time series classification. DIVERSIFY takes an iterative process: it first obtains the worst-case distribution scenario via adversarial training, then matches the distributions of the obtained sub-domains. We also present some theoretical insights. We conduct experiments on gesture recognition, speech commands recognition, wearable stress and affect detection, and sensor-based human activity recognition with a total of seven datasets in different settings. Results demonstrate that DIVERSIFY significantly outperforms other baselines and effectively characterizes the latent distributions by qualitative and quantitative analysis. Code is available at: https://github.com/microsoft/robustlearn.
|
['Xing Xie', 'Yiqiang Chen', 'Xinwei Sun', 'Jindong Wang', 'Wang Lu']
|
2022-09-15
| null | null | null | null |
['gesture-recognition']
|
['computer-vision']
|
[ 1.41562223e-01 -5.00432491e-01 -3.51004630e-01 -4.48606700e-01
-8.44133615e-01 -1.02026296e+00 5.15485823e-01 -5.81019707e-02
-1.20906726e-01 6.11850381e-01 3.18140864e-01 -1.11632615e-01
-2.30513453e-01 -4.10209805e-01 -6.09537125e-01 -8.38235199e-01
-2.98154086e-01 2.66182572e-01 -4.21751767e-01 2.12952182e-01
-6.94550350e-02 4.91789073e-01 -1.43493176e+00 1.24794126e-01
6.01748347e-01 1.01262105e+00 -3.76517981e-01 6.61834955e-01
2.45883241e-01 6.81982994e-01 -7.09188640e-01 -1.02440737e-01
5.34613542e-02 -3.77153188e-01 -3.96739751e-01 4.38977107e-02
1.14360325e-01 -3.87371451e-01 -5.40055156e-01 9.79316533e-01
6.50749326e-01 3.91652554e-01 8.08084488e-01 -1.88193476e+00
-7.13000357e-01 4.18199629e-01 -3.76752555e-01 5.75278103e-01
3.95102531e-01 -3.28357369e-02 8.18414450e-01 -5.67245781e-01
3.09853703e-01 1.14221299e+00 8.17822099e-01 7.22660780e-01
-1.27959108e+00 -8.13271582e-01 3.51377755e-01 3.58921975e-01
-1.26261675e+00 -4.40721393e-01 8.58483613e-01 -3.78512412e-01
7.75872409e-01 2.93267757e-01 3.47002357e-01 2.09193611e+00
4.14427593e-02 1.06908274e+00 1.28538430e+00 -2.18563862e-02
6.78142130e-01 -3.97977322e-01 2.52612799e-01 -2.14917157e-02
-1.18715517e-01 1.05422437e-01 -5.32182992e-01 -3.85312259e-01
4.31468159e-01 4.75321114e-01 -3.21404904e-01 -2.78975427e-01
-9.75951016e-01 5.11728764e-01 -8.50028917e-02 2.38430783e-01
-4.51874197e-01 2.39632979e-01 5.82701981e-01 5.80991924e-01
6.74954593e-01 1.16598327e-02 -6.79637194e-01 -8.34713221e-01
-7.09961355e-01 2.42478013e-01 8.67410064e-01 7.76595116e-01
1.56349242e-01 2.89819866e-01 -8.19276050e-02 8.77679825e-01
-2.21691892e-01 7.36998796e-01 1.00419903e+00 -8.95114779e-01
2.80890703e-01 1.62312433e-01 1.33845419e-01 -7.72611499e-01
-3.06704819e-01 -4.15963739e-01 -7.99498975e-01 -2.06053808e-01
6.40500546e-01 -3.53464454e-01 -8.53517413e-01 2.07894993e+00
3.26253146e-01 7.44530082e-01 -2.49774568e-02 6.45256162e-01
2.58032143e-01 6.62428200e-01 2.05203101e-01 -5.45917153e-01
8.35811496e-01 -5.48622966e-01 -8.93515825e-01 -3.62976175e-03
1.63859591e-01 -3.44395816e-01 1.27461135e+00 5.65015972e-01
-7.51886010e-01 -3.11737627e-01 -8.40314329e-01 2.48839185e-01
-1.63395688e-01 -4.97742631e-02 5.54548621e-01 6.53735638e-01
-6.45691276e-01 8.18474591e-01 -1.55455291e+00 -5.39755821e-01
4.29444164e-01 1.63077906e-01 -8.74828920e-02 1.02330856e-01
-9.03478861e-01 3.78656626e-01 1.04339376e-01 -6.99885935e-02
-1.07644439e+00 -7.41043746e-01 -6.59301758e-01 -3.89188994e-03
2.05794960e-01 -3.18309575e-01 1.53523445e+00 -1.08205581e+00
-1.43882644e+00 4.72709805e-01 -2.49730110e-01 -4.09250170e-01
4.67170686e-01 -4.71393198e-01 -6.00261748e-01 1.08148694e-01
-8.31630230e-02 -7.47256726e-02 9.98285055e-01 -8.22547257e-01
-1.52805731e-01 -5.57971060e-01 -2.13596746e-01 -9.27997157e-02
-6.27658546e-01 -3.16346660e-02 -1.12023965e-01 -9.57954466e-01
3.83801870e-02 -1.08039415e+00 9.89993513e-02 -2.02141553e-01
-2.00498298e-01 -3.31653327e-01 7.90687501e-01 -7.03039110e-01
1.28113163e+00 -2.43164802e+00 1.70234084e-01 3.05383913e-02
-4.20925245e-02 -1.85835049e-01 -1.83622852e-01 6.64109230e-01
-4.23652828e-01 -2.60752570e-02 -2.00421929e-01 -3.99194926e-01
3.97489786e-01 3.92126292e-01 -9.35359895e-01 7.35490322e-01
-1.05717055e-01 6.98379874e-01 -1.05358684e+00 3.43339592e-02
1.00332715e-01 4.83529001e-01 -3.04783791e-01 2.44578332e-01
-2.26620600e-01 5.46862423e-01 -4.61775690e-01 9.72051382e-01
4.79606569e-01 -1.83328599e-01 2.93855071e-01 -6.65753633e-02
2.71480054e-01 1.55193299e-01 -9.01781201e-01 1.69727993e+00
-2.58065313e-01 5.81394374e-01 -2.16820508e-01 -1.36820889e+00
6.39203489e-01 4.18335110e-01 8.94701242e-01 -6.05721176e-01
1.87145740e-01 8.25807899e-02 -1.88496009e-01 -5.96779406e-01
8.15215483e-02 -2.41384611e-01 -3.56231630e-01 6.62260056e-01
1.27145901e-01 2.32792661e-01 -1.70169741e-01 -1.33297771e-01
1.37452257e+00 3.04420799e-01 3.18885982e-01 1.10505357e-01
-1.12655602e-01 -3.85903716e-01 6.90754652e-01 6.03653073e-01
-6.68244958e-01 4.25177068e-01 6.17363393e-01 -3.09660137e-01
-8.68231773e-01 -1.56951785e+00 -3.07808514e-03 1.28137088e+00
-2.23536268e-01 -2.89638698e-01 -5.71783066e-01 -6.21546268e-01
2.17941869e-02 8.50008607e-01 -7.60199547e-01 -2.65763044e-01
-4.18218106e-01 -7.73727477e-01 6.97351277e-01 8.94197881e-01
6.25677733e-03 -9.55632806e-01 -4.51827824e-01 8.68841335e-02
-3.73877853e-01 -9.41026568e-01 -4.65388030e-01 2.66109049e-01
-1.05833364e+00 -8.95062327e-01 -5.18459737e-01 -2.78656602e-01
2.18628585e-01 -2.33647786e-02 8.35592330e-01 -5.62218547e-01
-2.30889857e-01 1.05062938e+00 -4.88820285e-01 -6.11330628e-01
-1.57143191e-01 3.37552987e-02 4.10024107e-01 1.26780614e-01
7.11222112e-01 -1.28958619e+00 -6.71177030e-01 2.34407902e-01
-1.05074668e+00 -6.66396856e-01 8.33032578e-02 6.86800480e-01
6.76527739e-01 7.93496743e-02 9.05055940e-01 -3.44766468e-01
8.05109262e-01 -1.08224916e+00 -2.01958403e-01 2.03063384e-01
-4.97809350e-01 -1.96403414e-01 6.32552981e-01 -1.13221908e+00
-7.58292437e-01 -6.50291815e-02 4.37027588e-02 -9.09588337e-01
-4.32145953e-01 5.05401552e-01 -1.87088162e-01 7.03451693e-01
7.08883166e-01 4.40401018e-01 6.22447543e-02 -5.42187631e-01
3.56156796e-01 7.03001440e-01 5.96874177e-01 -9.69611645e-01
7.36655056e-01 6.73763692e-01 -3.52365494e-01 -8.00005853e-01
-8.91371727e-01 -4.15467113e-01 -4.61253166e-01 -2.50132293e-01
4.23657715e-01 -9.41255867e-01 -5.70741594e-01 6.47994280e-01
-8.21568727e-01 -6.34714186e-01 -4.77863044e-01 5.43982327e-01
-8.33096385e-01 3.42766136e-01 -5.56406558e-01 -1.07289445e+00
-2.53045231e-01 -5.11537671e-01 1.14525902e+00 9.13574696e-02
-6.05610073e-01 -1.20989943e+00 3.52795333e-01 4.19681221e-02
2.61050463e-01 6.93582833e-01 6.85068727e-01 -1.05953693e+00
-1.11641467e-01 -1.88985780e-01 3.88408929e-01 5.04423440e-01
4.12777781e-01 -1.47217527e-01 -1.20068419e+00 -5.09155035e-01
3.60579550e-01 -5.23763657e-01 5.67643106e-01 4.27976578e-01
1.70128798e+00 -2.69641459e-01 -2.04388618e-01 7.63168752e-01
1.02577472e+00 2.86437988e-01 5.74068606e-01 1.50056258e-01
3.97022158e-01 4.50896412e-01 6.06459975e-01 8.41033280e-01
1.68100685e-01 3.56537968e-01 1.83033869e-01 4.99688715e-01
2.57896066e-01 -2.98025876e-01 7.29990423e-01 8.96863997e-01
1.17363393e-01 -4.08997536e-01 -8.96798432e-01 7.88150668e-01
-1.93299186e+00 -1.45196915e+00 3.92995536e-01 2.10907006e+00
8.68779480e-01 -1.94516480e-01 4.48135883e-01 4.22145754e-01
5.72321653e-01 1.44390106e-01 -1.14673638e+00 -1.51800066e-01
1.92775428e-02 4.48940814e-01 7.35635236e-02 -5.16727148e-03
-1.14949083e+00 4.32937056e-01 6.37534666e+00 7.86612630e-01
-1.40038109e+00 2.70716667e-01 4.79249388e-01 -6.64967000e-01
-1.77236155e-01 -4.35883611e-01 -2.76344180e-01 6.31108344e-01
1.20484412e+00 -6.56700790e-01 6.64014578e-01 8.88853133e-01
1.90704525e-01 4.23617244e-01 -1.46477306e+00 1.28182626e+00
7.31931925e-02 -6.19698405e-01 -3.08400661e-01 -1.02420948e-01
5.71367800e-01 2.45352656e-01 4.05934304e-01 6.24368727e-01
1.32912189e-01 -1.03137577e+00 7.52115667e-01 6.23071909e-01
6.46852195e-01 -5.92225373e-01 2.10929319e-01 4.91415769e-01
-9.51251149e-01 -3.95153165e-01 -8.77717976e-03 -2.46226266e-01
-4.14316030e-03 4.75910246e-01 -6.26275480e-01 2.84543097e-01
6.99162245e-01 1.00567448e+00 -2.28090912e-01 7.94532597e-01
-4.46980409e-02 1.17752981e+00 -3.84990066e-01 4.38756533e-02
-3.02642792e-01 -9.99985635e-02 6.18159354e-01 1.04436433e+00
6.73099279e-01 1.08754061e-01 2.01801091e-01 6.68877423e-01
-9.15587842e-02 -1.82889551e-01 -6.37180984e-01 -5.05618930e-01
7.41456509e-01 9.57016230e-01 -5.83591461e-01 -1.55466810e-01
-2.84904152e-01 1.09065688e+00 2.03909561e-01 7.28571773e-01
-1.09183228e+00 -2.02382237e-01 9.21295822e-01 -1.85611323e-01
1.70123145e-01 -4.36777771e-01 -2.09597778e-02 -1.48895705e+00
3.58034223e-01 -1.10458875e+00 6.97327375e-01 -6.85401380e-01
-1.90015721e+00 4.23938096e-01 3.00983161e-01 -1.43382692e+00
-4.92581248e-01 -5.20568311e-01 -5.23376048e-01 5.30986905e-01
-1.19552994e+00 -9.53998387e-01 -2.84261346e-01 8.14116120e-01
6.52512550e-01 6.36486858e-02 9.31002557e-01 3.80752981e-01
-6.85173988e-01 6.65151477e-01 4.52584088e-01 1.53368324e-01
7.14472473e-01 -1.28668809e+00 2.73817062e-01 6.76759839e-01
1.63260952e-01 7.13480234e-01 6.93722606e-01 -4.82732832e-01
-1.38239503e+00 -1.23618972e+00 4.08553332e-01 -7.42406487e-01
1.08900535e+00 -3.85648936e-01 -1.05541146e+00 9.15143728e-01
3.30769978e-02 3.97077054e-02 1.13681233e+00 2.21824870e-01
-6.42780662e-01 -2.01549634e-01 -1.00369966e+00 4.59905177e-01
1.20581067e+00 -8.50443900e-01 -6.41446531e-01 3.97576749e-01
4.86839175e-01 -3.52380604e-01 -9.37188923e-01 2.14958802e-01
7.66210616e-01 -6.20296478e-01 9.08301651e-01 -8.80118072e-01
3.07927907e-01 2.47797985e-02 -5.73498011e-01 -1.48011291e+00
-8.12527724e-03 -8.93820763e-01 -6.31863356e-01 1.29306495e+00
-5.02302758e-02 -7.49407947e-01 5.36371469e-01 5.29442966e-01
1.25310317e-01 -5.71166873e-01 -1.06671846e+00 -1.18567741e+00
3.26777995e-01 -8.06420624e-01 7.57042289e-01 1.18773496e+00
1.36750892e-01 3.12626399e-02 -3.88195217e-01 4.13943172e-01
6.87171876e-01 2.13829190e-01 6.37856305e-01 -1.07329869e+00
-4.87592906e-01 -2.10743248e-01 -3.47135007e-01 -9.62091088e-01
6.04301989e-01 -7.66206980e-01 -9.26661640e-02 -8.84668708e-01
1.34744465e-01 -1.15347877e-01 -7.38235474e-01 5.08023500e-01
2.82363431e-03 -1.70692369e-01 -9.78095904e-02 2.93979108e-01
-6.25958860e-01 8.61360848e-01 7.51263440e-01 -1.44486606e-01
-2.24341199e-01 1.85756296e-01 -6.07800066e-01 6.37918890e-01
1.26833498e+00 -4.32261854e-01 -8.00808132e-01 -2.46061370e-01
2.16682535e-02 1.91190660e-01 6.05567396e-01 -9.41826820e-01
-4.41102087e-02 -4.82683271e-01 3.15738708e-01 -4.42633092e-01
4.56669480e-01 -7.10733175e-01 5.11173345e-02 2.25302830e-01
-4.53641206e-01 1.13900125e-01 3.82085592e-01 8.34591746e-01
-1.03150271e-01 1.82462126e-01 5.25613248e-01 2.62220442e-01
-4.56950486e-01 5.42233109e-01 -3.46045226e-01 3.16711932e-01
9.40713644e-01 -6.40957663e-03 -2.65618324e-01 -7.33575583e-01
-1.01116121e+00 4.73250896e-02 2.41125301e-01 6.34856284e-01
4.85116512e-01 -1.49357283e+00 -3.11362356e-01 6.97759017e-02
1.71659783e-01 -4.56700832e-01 3.80550116e-01 8.33640277e-01
1.83159083e-01 -1.15579821e-01 -1.53772831e-01 -7.56211579e-01
-1.11775947e+00 6.15960300e-01 3.48997772e-01 1.24219984e-01
-3.71901512e-01 4.41531450e-01 -4.63531576e-02 -4.86755788e-01
4.81571883e-01 -6.37868464e-01 1.11427397e-01 8.82634968e-02
6.27949357e-01 6.16436124e-01 -1.09536976e-01 -3.07008028e-01
-3.57997268e-01 2.44203627e-01 1.96455419e-01 -1.99953556e-01
1.48406696e+00 -1.41056165e-01 8.62361416e-02 1.06185997e+00
1.36831188e+00 -1.05681725e-01 -1.34159338e+00 -2.08209619e-01
7.18352869e-02 -3.88149440e-01 -3.02678138e-01 -8.00611794e-01
-7.41270959e-01 8.58448625e-01 8.51903379e-01 3.35529834e-01
1.43829668e+00 -1.00317277e-01 7.72442758e-01 3.57573688e-01
2.40364850e-01 -1.01085413e+00 3.00787687e-01 4.07212526e-01
9.88160908e-01 -8.97235394e-01 -3.27024370e-01 1.57790557e-02
-5.47151446e-01 9.54538167e-01 4.14399236e-01 -1.47744998e-01
7.92067528e-01 3.14113200e-01 4.77182567e-02 4.15539145e-02
-1.00876927e+00 4.78283390e-02 4.62767854e-02 8.32470775e-01
4.16377753e-01 2.55133957e-01 4.38383296e-02 1.03569734e+00
-2.13857442e-01 2.25290433e-01 4.01004106e-01 9.73885953e-01
5.94142713e-02 -1.14219177e+00 -3.33111942e-01 3.26144487e-01
-5.93442559e-01 2.91804999e-01 -1.94040731e-01 5.82724154e-01
-1.57312050e-01 1.02793145e+00 1.89310201e-02 -4.33093101e-01
3.78535211e-01 5.21977484e-01 5.12641251e-01 -3.69691730e-01
-1.27929658e-01 5.86190186e-02 -1.01086117e-01 -7.13154554e-01
-4.55811501e-01 -1.19333827e+00 -1.11378849e+00 -2.03717321e-01
2.38505140e-01 -2.37226598e-02 5.46956778e-01 7.55611062e-01
5.10740161e-01 4.97666627e-01 7.52895176e-01 -6.46910489e-01
-1.22608316e+00 -1.07807827e+00 -6.65953696e-01 8.42035890e-01
6.25019193e-01 -7.11019993e-01 -6.76366270e-01 3.57806236e-01]
|
[7.432645320892334, 3.011188507080078]
|
3a74e86f-17e1-4335-bad7-a801b3507ddd
|
graph-based-time-series-anomaly-detection-a
|
2302.00058
| null |
https://arxiv.org/abs/2302.00058v2
|
https://arxiv.org/pdf/2302.00058v2.pdf
|
Graph-based Time-Series Anomaly Detection: A Survey
|
With the recent advances in technology, a wide range of systems continue to collect a large amount of data over time and thus generate time series. Time-Series Anomaly Detection (TSAD) is an important task in various time-series applications such as e-commerce, cybersecurity, vehicle maintenance, and healthcare monitoring. However, this task is very challenging as it requires considering both the intra-variable dependency and the inter-variable dependency, where a variable can be defined as an observation in time series data. Recent graph-based approaches have made impressive progress in tackling the challenges of this field. In this survey, we conduct a comprehensive and up-to-date review of Graph-based TSAD (G-TSAD). First, we explore the significant potential of graph representation learning for time-series data. Then, we review state-of-the-art graph anomaly detection techniques in the context of time series and discuss their strengths and drawbacks. Finally, we discuss the technical challenges and potential future directions for possible improvements in this research field.
|
['Narges Armanfard', 'Ali Karami', 'Thi Kieu Khanh Ho']
|
2023-01-31
| null | null | null | null |
['graph-anomaly-detection']
|
['graphs']
|
[ 1.72462389e-01 -3.79843235e-01 -2.15346530e-01 -1.23815276e-01
-5.23441983e-03 -3.92755806e-01 3.95253986e-01 8.20666015e-01
2.05449820e-01 1.91042066e-01 -2.60416001e-01 -8.02362442e-01
-4.68433797e-01 -9.69554126e-01 -1.97579503e-01 -5.31429052e-01
-1.14975870e+00 1.46505848e-01 2.79411823e-01 -4.13127571e-01
1.21567562e-01 7.85715342e-01 -1.30887401e+00 -2.35295981e-01
6.48829818e-01 1.29187107e+00 -7.52789080e-01 5.12080252e-01
-2.90407777e-01 7.28173494e-01 -8.48313868e-01 -5.58973104e-02
2.24817827e-01 -5.39253533e-01 -3.81792516e-01 6.73874989e-02
2.60712683e-01 1.67715177e-01 -7.15496182e-01 9.84332204e-01
1.28147095e-01 3.81831437e-01 5.22343159e-01 -2.14851475e+00
-3.72926593e-01 4.63749543e-02 -8.55747819e-01 9.36155915e-01
2.84451276e-01 -8.46102908e-02 9.27137434e-01 -3.74447405e-01
3.01285625e-01 9.45622087e-01 6.58434153e-01 1.95931762e-01
-8.83203566e-01 -5.64509034e-01 5.87523103e-01 5.66949129e-01
-1.00662100e+00 1.39164880e-01 1.49440134e+00 -4.49911684e-01
1.27949309e+00 4.01596934e-01 9.75229681e-01 7.37976432e-01
5.79939306e-01 6.41399503e-01 6.07452154e-01 -2.18639031e-01
2.82712460e-01 -9.27895546e-01 4.01671976e-01 5.70982635e-01
4.20920759e-01 -8.20213184e-02 -3.87716383e-01 -5.77948928e-01
6.08347833e-01 4.35067177e-01 1.18593343e-01 -5.21080256e-01
-1.02113307e+00 8.92751515e-01 2.20471039e-01 5.85215211e-01
-4.84079242e-01 -8.15416046e-04 9.75919545e-01 9.36933279e-01
7.20801651e-01 2.69276530e-01 -3.32086593e-01 -1.47846684e-01
-4.73066658e-01 1.42720252e-01 7.82439113e-01 8.17018688e-01
3.36518615e-01 7.34336495e-01 -2.08333880e-02 5.20192564e-01
1.24577925e-01 2.80658066e-01 3.61736894e-01 -1.56034738e-01
5.86484611e-01 9.28378642e-01 -5.36366820e-01 -1.54216349e+00
-7.42472589e-01 -3.20530832e-01 -1.25906575e+00 1.24378100e-01
3.60417098e-01 4.75485250e-02 -9.31071281e-01 1.41018558e+00
5.52616239e-01 8.64962161e-01 -3.59532088e-01 4.32810217e-01
4.01881278e-01 5.56136608e-01 -9.37826037e-02 -7.65034318e-01
9.68324482e-01 -5.03280163e-01 -1.06046021e+00 -1.13490433e-01
6.30670428e-01 -4.74921048e-01 5.72945833e-01 1.73019901e-01
-4.15553451e-01 -1.71065569e-01 -1.20647299e+00 3.92586112e-01
-5.52967906e-01 -7.25882769e-01 7.49170601e-01 3.70897591e-01
-7.05658317e-01 4.70962822e-01 -1.18821931e+00 -4.79567498e-01
3.11760902e-01 3.19232404e-01 -2.46954232e-01 8.57467130e-02
-1.20726252e+00 5.82192481e-01 4.08290625e-02 -5.50907701e-02
-4.83052194e-01 -4.50221628e-01 -1.12968647e+00 -3.10119569e-01
6.06695890e-01 -2.31594905e-01 9.60329592e-01 -3.56020063e-01
-8.11717391e-01 4.98802394e-01 -1.16489917e-01 -7.57877648e-01
2.27903426e-01 -8.06393474e-02 -1.45684469e+00 -3.36909443e-02
1.10194925e-02 -6.41217828e-01 9.84368861e-01 -4.61845636e-01
-5.31656504e-01 -6.38456166e-01 -2.93104589e-01 -2.33007520e-01
-2.76559204e-01 -6.60957471e-02 -4.49175648e-02 -9.82532263e-01
3.18752766e-01 -7.78146327e-01 -4.12803710e-01 -1.53940588e-01
-3.45044464e-01 -6.52135491e-01 1.56007302e+00 -3.64345163e-01
1.89764273e+00 -2.43173909e+00 -6.22139163e-02 7.21789837e-01
5.48484147e-01 2.70720929e-01 -2.65762955e-02 8.85433495e-01
-5.00908077e-01 -6.03723377e-02 -4.61240470e-01 -2.95805093e-02
-3.44812602e-01 3.32542479e-01 -5.33827603e-01 8.24772060e-01
2.65119404e-01 7.13863850e-01 -1.20739532e+00 -5.02927117e-02
4.97711152e-01 9.52725783e-02 6.60287887e-02 1.11926168e-01
-2.30155841e-01 4.85512525e-01 -8.44166756e-01 7.58032084e-01
3.40712160e-01 -2.23587364e-01 7.91903958e-02 1.78034231e-02
-5.19909598e-02 5.17377928e-02 -9.27713990e-01 1.31612146e+00
8.55757445e-02 7.35475779e-01 -4.29088116e-01 -1.61904645e+00
1.06782484e+00 2.11389378e-01 1.24369907e+00 -1.03997087e+00
1.09618716e-02 3.56167555e-01 2.20054552e-01 -5.55230498e-01
3.37853074e-01 1.63530931e-01 -1.37815505e-01 6.33182108e-01
-1.84784025e-01 7.08277449e-02 3.78518403e-01 2.52958447e-01
1.55470836e+00 -5.87406516e-01 7.27904439e-01 8.79985094e-02
6.77237391e-01 -9.35765356e-02 4.38887626e-01 3.19498301e-01
-4.23598468e-01 1.21447191e-01 9.29710329e-01 -8.32874179e-01
-8.53914857e-01 -7.48615682e-01 2.05868304e-01 7.13973939e-01
-1.74212918e-01 -7.01641023e-01 -1.42317533e-01 -1.01170552e+00
3.96347106e-01 6.47207081e-01 -6.70567572e-01 -5.07077813e-01
-8.26545596e-01 -6.57006443e-01 4.43468392e-01 4.02566582e-01
1.69439077e-01 -8.72547269e-01 -2.87938118e-01 4.00595039e-01
9.13962498e-02 -9.22979176e-01 -5.01405656e-01 -9.80480239e-02
-1.16782451e+00 -1.36719835e+00 -1.55871734e-01 -5.18762767e-01
5.45347512e-01 6.08886480e-01 1.12562788e+00 2.00781956e-01
-5.35565197e-01 6.65296912e-01 -4.81789052e-01 -6.92121267e-01
-2.55953789e-01 -1.89055860e-01 2.97113359e-01 3.34842831e-01
6.98646426e-01 -7.63028562e-01 -4.35271710e-01 2.86443442e-01
-1.02614057e+00 -8.33627343e-01 8.04429650e-02 5.57712853e-01
8.78033996e-01 5.56746721e-01 8.35149586e-01 -7.93611109e-01
8.52534473e-01 -1.06046629e+00 -7.37061024e-01 1.22187659e-01
-8.04261565e-01 -2.59170115e-01 7.33865380e-01 -4.05950069e-01
-8.33238587e-02 -3.24741900e-01 1.05508320e-01 -7.55544841e-01
1.87765643e-01 8.81430089e-01 2.60302842e-01 -1.46274909e-01
5.62608421e-01 1.10221222e-01 3.58208925e-01 -3.48619699e-01
-4.16925214e-02 2.99218714e-01 3.03367406e-01 -2.39056289e-01
8.74016583e-01 5.16292751e-01 6.09055519e-01 -1.19886911e+00
-5.48914135e-01 -8.22162211e-01 -4.54456896e-01 -3.60981166e-01
5.37816763e-01 -4.45178092e-01 -4.48942155e-01 7.60610640e-01
-8.32794070e-01 -9.81907267e-03 -4.36223060e-01 3.20428222e-01
-2.23175764e-01 6.34713769e-01 -3.16259712e-01 -8.27754736e-01
-3.13305438e-01 -5.41877687e-01 7.52870440e-01 -1.75576895e-01
-3.72084439e-01 -1.44433641e+00 3.02402437e-01 -3.03646952e-01
4.53648657e-01 1.00230992e+00 1.21061480e+00 -1.13212204e+00
-2.37743586e-01 -7.43613541e-01 2.14612514e-01 1.44478321e-01
6.71648264e-01 -1.42386071e-02 -3.60918134e-01 -5.85158288e-01
2.34881446e-01 4.12933826e-01 3.77581120e-01 2.89931357e-01
1.34595549e+00 -2.73396760e-01 -3.52048010e-01 3.20485383e-01
1.09040856e+00 3.57857317e-01 3.35963160e-01 1.26369998e-01
9.50691223e-01 5.00510395e-01 7.95128644e-01 6.05778575e-01
2.21547991e-01 4.24062580e-01 6.31709933e-01 1.13609843e-01
1.09077044e-01 -1.72604024e-01 1.17126286e-01 1.13560104e+00
-1.14511155e-01 -3.99522394e-01 -1.10402596e+00 5.67972958e-01
-2.04098201e+00 -1.12031686e+00 -6.42185688e-01 2.33050871e+00
6.92849904e-02 2.41787717e-01 5.29038906e-01 6.83012962e-01
7.58315086e-01 5.94015479e-01 -9.58382726e-01 -3.21032405e-01
5.35893738e-02 5.82970940e-02 2.85452843e-01 1.10377088e-01
-1.27750587e+00 4.42603201e-01 6.76872921e+00 5.27749658e-01
-1.36952543e+00 -3.11108887e-01 2.69313395e-01 1.64257988e-01
-8.61953422e-02 -3.13997507e-01 -8.50058645e-02 3.48423332e-01
1.03704929e+00 -6.58357620e-01 3.42724681e-01 7.71078646e-01
9.74337235e-02 2.45451778e-01 -1.06545389e+00 1.14485657e+00
1.58361793e-01 -1.01975322e+00 7.73922205e-02 1.41238704e-01
5.06760955e-01 2.86550909e-01 -2.81035039e-03 5.12363613e-02
-2.94080377e-01 -8.57060492e-01 1.90327771e-03 1.40136153e-01
6.70887768e-01 -7.12096751e-01 5.34384191e-01 -3.18174735e-02
-1.79211068e+00 -1.60222113e-01 5.59723116e-02 -3.45752686e-01
2.72088796e-01 1.01032031e+00 -4.85567182e-01 9.70722139e-01
5.31248987e-01 1.39336240e+00 -3.59342337e-01 1.14860725e+00
7.37827346e-02 6.68541908e-01 -3.69093567e-01 1.26303643e-01
1.26706049e-01 -4.57184553e-01 1.01484728e+00 7.09717453e-01
4.28572327e-01 5.08575216e-02 4.52606916e-01 1.90269560e-01
2.63122976e-01 -2.12375298e-02 -1.03417754e+00 -5.05079746e-01
3.47533613e-01 9.17653084e-01 -6.83098197e-01 -1.67134121e-01
-9.10060823e-01 5.89212596e-01 1.71867181e-02 5.48411369e-01
-6.26794875e-01 -5.29869318e-01 9.14176643e-01 1.56963959e-01
1.32135227e-02 -7.61130512e-01 -1.29878208e-01 -1.11343539e+00
3.27617258e-01 -9.88804996e-01 1.15080655e+00 -3.57916616e-02
-1.69721639e+00 5.25454223e-01 8.04730728e-02 -1.89369941e+00
-5.22067666e-01 -5.95385253e-01 -8.37890863e-01 4.19700474e-01
-1.25456738e+00 -7.62212098e-01 -2.83085436e-01 1.00549674e+00
4.62307006e-01 -3.29335004e-01 7.48308003e-01 4.63624835e-01
-5.87640226e-01 3.19373906e-01 -7.59251565e-02 3.59437883e-01
3.30789655e-01 -1.07057810e+00 9.76290822e-01 1.08950305e+00
1.78436577e-01 4.22300994e-01 6.98841572e-01 -9.39234376e-01
-1.73942900e+00 -1.18412113e+00 7.33717799e-01 -2.22655267e-01
1.29572260e+00 -2.59122252e-01 -1.36259747e+00 7.61700451e-01
-2.33814731e-01 5.35350323e-01 7.51305580e-01 3.37671310e-01
-3.66194785e-01 -2.55067259e-01 -9.85557199e-01 5.97515225e-01
1.13983095e+00 -6.37869656e-01 -5.88260651e-01 3.86648148e-01
5.93913198e-01 -2.72214383e-01 -7.96228290e-01 4.76915449e-01
7.36934617e-02 -6.55878007e-01 7.70532072e-01 -9.42630827e-01
-3.13391298e-01 -5.69755197e-01 1.08560055e-01 -1.31750691e+00
-3.04406643e-01 -1.00707722e+00 -9.70180094e-01 8.89777362e-01
-7.10642189e-02 -1.07596600e+00 5.14735222e-01 3.16990227e-01
-3.86007011e-01 -8.24338377e-01 -1.28520417e+00 -1.13114738e+00
-3.94493937e-01 -7.17402101e-01 7.89700925e-01 1.22628069e+00
2.92650700e-01 2.50072002e-01 -3.47772390e-01 2.21812487e-01
7.18868911e-01 6.75846338e-02 6.58510864e-01 -1.46524620e+00
1.94878906e-01 -4.52470452e-01 -1.18998444e+00 -6.14999533e-01
-7.12903170e-03 -7.28831232e-01 -4.86464113e-01 -1.38996983e+00
-6.03255749e-01 -2.07903713e-01 -7.23325074e-01 3.26467723e-01
-2.33443305e-02 -2.22592041e-01 -3.10172737e-01 1.92032829e-01
-6.09163940e-01 4.86376762e-01 9.92019355e-01 -1.96466029e-01
-3.20134789e-01 3.64094943e-01 -1.34940937e-01 4.60426539e-01
8.75465691e-01 -3.30550611e-01 -7.59193778e-01 -1.06306665e-01
2.18511119e-01 1.82965681e-01 1.95078596e-01 -7.90812671e-01
3.64582747e-01 -2.76355505e-01 -1.80787370e-02 -6.92625582e-01
3.13450731e-02 -1.10888457e+00 2.52004359e-02 5.44170678e-01
2.97014505e-01 9.64590907e-01 3.32120687e-01 1.08484077e+00
-6.02974057e-01 5.55174887e-01 3.30123007e-01 3.23828906e-01
-1.04157722e+00 9.71007347e-01 -4.24437046e-01 1.40571892e-01
1.33952713e+00 -9.28567871e-02 -3.84002566e-01 -4.75883633e-01
-4.19401735e-01 4.97146606e-01 1.33403866e-02 9.68135059e-01
7.22931147e-01 -1.54824769e+00 -4.39413548e-01 5.81708729e-01
5.10472119e-01 -1.35889485e-01 3.73400956e-01 1.01397419e+00
-2.20803499e-01 1.95877105e-01 -1.84739977e-01 -6.89373136e-01
-1.17830229e+00 9.76098299e-01 1.99905753e-01 -3.76366913e-01
-9.32558000e-01 1.31009430e-01 -2.07420543e-01 9.19432566e-02
7.72138238e-02 -2.92506665e-01 -2.80321062e-01 6.06959350e-02
5.46405315e-01 7.04880774e-01 2.87513107e-01 -5.15612006e-01
-5.86697221e-01 5.83757401e-01 1.32269878e-02 4.07158732e-01
1.28954566e+00 2.84264740e-02 -3.34731311e-01 9.31120515e-01
1.18433273e+00 -9.42147896e-02 -7.43600607e-01 -2.05836698e-01
3.60345185e-01 -4.22832817e-01 -2.75798976e-01 -6.08065762e-02
-1.15160382e+00 6.59784555e-01 3.67423296e-01 1.08738399e+00
1.32984924e+00 -1.07968226e-01 8.69861066e-01 1.96008474e-01
5.33136487e-01 -9.73645747e-01 3.14651608e-01 7.23653197e-01
9.07127082e-01 -9.92491543e-01 5.59030175e-02 -5.56345403e-01
-3.34642708e-01 1.13850117e+00 4.64298040e-01 -4.16774869e-01
1.09758210e+00 -5.36440723e-02 -8.57375935e-02 -5.54161668e-01
-4.99779344e-01 -1.28275841e-01 6.13548577e-01 6.46388531e-01
3.48302096e-01 5.30832335e-02 -2.64606178e-01 4.07873131e-02
6.97122440e-02 -4.15981978e-01 2.36341998e-01 1.14297855e+00
-5.87846488e-02 -1.22378767e+00 -2.31660396e-01 9.69450355e-01
-4.54462856e-01 2.65700758e-01 -3.59587491e-01 8.37095082e-01
-5.09391367e-01 1.09446657e+00 2.64092088e-01 -5.23743093e-01
6.39051199e-01 8.44399109e-02 1.14959702e-01 -3.63664001e-01
-1.31512806e-01 -1.69435233e-01 -8.20182177e-05 -7.30985761e-01
-3.26118946e-01 -7.00938165e-01 -1.36940789e+00 -5.03497839e-01
-1.71939284e-02 1.41707793e-01 4.71261472e-01 8.66595685e-01
6.12438023e-01 8.76889110e-01 7.70920157e-01 -2.26739183e-01
-2.37822339e-01 -6.52650654e-01 -7.57049561e-01 6.11564159e-01
7.31420338e-01 -7.89882600e-01 -5.98256648e-01 -3.59483510e-01]
|
[7.260163307189941, 2.8157758712768555]
|
4a6fc7af-fab9-4435-8e78-6130fb77b289
|
rl-grit-reinforcement-learning-for-grammar
|
2105.13114
| null |
https://arxiv.org/abs/2105.13114v1
|
https://arxiv.org/pdf/2105.13114v1.pdf
|
RL-GRIT: Reinforcement Learning for Grammar Inference
|
When working to understand usage of a data format, examples of the data format are often more representative than the format's specification. For example, two different applications might use very different JSON representations, or two PDF-writing applications might make use of very different areas of the PDF specification to realize the same rendered content. The complexity arising from these distinct origins can lead to large, difficult-to-understand attack surfaces, presenting a security concern when considering both exfiltration and data schizophrenia. Grammar inference can aid in describing the practical language generator behind examples of a data format. However, most grammar inference research focuses on natural language, not data formats, and fails to support crucial features such as type recursion. We propose a novel set of mechanisms for grammar inference, RL-GRIT, and apply them to understanding de facto data formats. After reviewing existing grammar inference solutions, it was determined that a new, more flexible scaffold could be found in Reinforcement Learning (RL). Within this work, we lay out the many algorithmic changes required to adapt RL from its traditional, sequential-time environment to the highly interdependent environment of parsing. The result is an algorithm which can demonstrably learn recursive control structures in simple data formats, and can extract meaningful structure from fragments of the PDF format. Whereas prior work in grammar inference focused on either regular languages or constituency parsing, we show that RL can be used to surpass the expressiveness of both classes, and offers a clear path to learning context-sensitive languages. The proposed algorithm can serve as a building block for understanding the ecosystems of de facto data formats.
|
['Walt Woods']
|
2021-05-17
| null | null | null | null |
['constituency-parsing']
|
['natural-language-processing']
|
[ 3.05851489e-01 3.03649157e-01 -3.08852673e-01 -5.50102234e-01
-5.04976928e-01 -9.91971254e-01 4.59834337e-01 3.01553398e-01
-6.60995990e-02 7.50946641e-01 -3.26090790e-02 -1.23846138e+00
-1.78246230e-01 -1.16645694e+00 -8.35376799e-01 -1.54626086e-01
-8.35239589e-02 4.03220952e-01 4.45571721e-01 -3.93579423e-01
5.00021815e-01 5.15679002e-01 -2.04321170e+00 5.60098648e-01
8.96640718e-01 7.02933133e-01 4.39166695e-01 7.05757797e-01
-9.65011775e-01 9.51169074e-01 -9.06971753e-01 -4.00758356e-01
1.29411544e-03 -1.95599169e-01 -9.87284720e-01 -3.14358741e-01
5.35849392e-01 -2.84753293e-01 1.80956349e-01 8.65239441e-01
-1.28889233e-02 -2.95320243e-01 2.79787064e-01 -1.34629965e+00
-4.96362090e-01 9.30696189e-01 1.98618248e-01 -1.06924221e-01
7.86570132e-01 9.94603187e-02 1.03350270e+00 -2.24032421e-02
6.55867457e-01 1.38427424e+00 5.10076940e-01 7.49202371e-01
-1.14758158e+00 -5.27790785e-01 2.28171408e-01 -2.78336138e-01
-6.81442916e-01 -2.87084728e-01 4.54870403e-01 -3.60583425e-01
1.06145501e+00 7.17591643e-01 7.01216459e-01 1.01601756e+00
5.24864435e-01 7.85361886e-01 1.41584980e+00 -6.54080391e-01
2.01431081e-01 1.76085666e-01 5.89336038e-01 7.53603935e-01
5.67234576e-01 3.44494879e-01 -3.53515297e-01 -3.82508844e-01
5.86217403e-01 -3.70580047e-01 -4.57048900e-02 -2.69496113e-01
-6.64832711e-01 7.90597022e-01 -2.22829416e-01 3.77668977e-01
4.27979469e-01 2.74299681e-01 4.54673111e-01 7.11503506e-01
-2.90736884e-01 5.67620516e-01 -5.28330863e-01 -5.72227478e-01
-6.24540567e-01 4.54737544e-01 1.45982611e+00 1.26103461e+00
8.61415744e-01 7.86971599e-02 2.56585360e-01 2.66504169e-01
5.84106803e-01 4.61450845e-01 1.15268357e-01 -8.53606999e-01
3.86067390e-01 5.53782165e-01 -1.88345730e-01 -6.19734168e-01
-2.09753528e-01 4.68549617e-02 -1.28050014e-01 4.09065247e-01
8.58043909e-01 -3.75394560e-02 -6.24654114e-01 1.76610994e+00
1.76973715e-01 -2.78171748e-01 3.01534116e-01 3.46602350e-01
6.58190906e-01 4.79407191e-01 2.98412710e-01 -9.95007753e-02
1.27715456e+00 -4.10185456e-02 -5.88953555e-01 -1.04021452e-01
1.07114732e+00 -6.69160187e-01 1.42291844e+00 6.61620617e-01
-1.08277678e+00 -4.02980953e-01 -1.11287606e+00 -2.33025968e-01
-8.12644482e-01 -4.12375987e-01 1.17812395e+00 1.13499165e+00
-8.49193633e-01 5.54924786e-01 -5.43300688e-01 -3.42642128e-01
3.78182121e-02 3.64159763e-01 -5.66805266e-02 9.04732421e-02
-1.24069691e+00 6.51060462e-01 8.23368192e-01 -2.34110162e-01
-4.99015898e-01 -6.67934954e-01 -8.14081252e-01 1.11400802e-02
7.75415778e-01 -6.91464186e-01 1.42217171e+00 -1.03243577e+00
-1.53271258e+00 5.97774446e-01 1.58744276e-01 -4.09784347e-01
2.92593569e-01 -2.90057570e-01 -5.93430102e-01 -2.44765252e-01
-2.01948985e-01 2.51529664e-01 7.80439615e-01 -1.14261806e+00
-8.60687613e-01 -2.33929574e-01 5.92546821e-01 -1.51791573e-01
9.51830000e-02 1.86729431e-01 -6.88441843e-02 -5.26461661e-01
-3.44918370e-01 -6.31399453e-01 1.33178219e-01 -2.42916018e-01
-2.92167574e-01 -2.64173001e-01 8.90193224e-01 -3.03368390e-01
1.67326558e+00 -2.03950715e+00 -1.78226858e-01 3.02808464e-01
1.03903219e-01 1.93715796e-01 6.44277260e-02 6.40568197e-01
3.94540764e-02 6.60150409e-01 -2.72423893e-01 3.96504998e-01
3.23502690e-01 8.11142504e-01 -8.59806180e-01 -2.15155706e-01
2.25475177e-01 5.63504159e-01 -8.77582610e-01 -3.58209163e-01
-5.66068925e-02 1.89917795e-02 -8.24027896e-01 3.11002374e-01
-8.60661805e-01 1.25863865e-01 -6.67642176e-01 4.84855980e-01
5.51104128e-01 -1.80985332e-02 6.54442608e-01 -2.65249442e-02
-3.10242891e-01 6.05263054e-01 -1.43658936e+00 1.51667833e+00
-5.56120694e-01 1.54906899e-01 6.34253025e-03 -8.22804749e-01
1.03778791e+00 5.33048138e-02 -5.08056916e-02 -6.41093671e-01
-1.48247406e-01 4.24932212e-01 2.48487256e-02 -7.17799127e-01
6.87693536e-01 -1.05041176e-01 -5.65596819e-01 7.28439450e-01
-1.23605020e-01 -5.06385267e-01 5.83066583e-01 2.17044696e-01
1.01695502e+00 6.05781913e-01 4.04787928e-01 -3.20673704e-01
4.62613225e-01 -1.43418729e-01 4.29114461e-01 1.16964543e+00
3.43041688e-01 1.16919115e-01 9.73207712e-01 -6.37562990e-01
-8.03074002e-01 -1.24446273e+00 -1.30275369e-01 1.31501341e+00
-2.48729259e-01 -9.27775800e-01 -7.26131141e-01 -8.63278210e-01
4.51098196e-02 1.06703568e+00 -1.13827340e-01 7.56812375e-03
-9.16879654e-01 -3.83349240e-01 9.36154664e-01 3.88019085e-01
2.91158140e-01 -1.18735588e+00 -9.55735564e-01 3.12815249e-01
2.38891348e-01 -1.02077734e+00 -1.01101018e-01 4.42409098e-01
-9.94786024e-01 -1.31525147e+00 4.28188950e-01 -6.19081199e-01
3.85339051e-01 -3.04402441e-01 1.39998245e+00 5.90726674e-01
7.55352825e-02 6.68471932e-01 -3.71098965e-01 -5.63552201e-01
-1.18737137e+00 2.13925228e-01 -3.45931768e-01 -4.50333774e-01
3.77971709e-01 -6.27620339e-01 2.62955338e-01 -6.78776354e-02
-1.49026811e+00 5.28906472e-02 3.63448262e-01 6.63058221e-01
1.76067919e-01 -1.42740503e-01 4.79740471e-01 -1.48177969e+00
9.78587389e-01 -3.41560930e-01 -8.46340597e-01 5.76549649e-01
-7.49990463e-01 7.33381927e-01 1.01985800e+00 -2.43922576e-01
-1.06748176e+00 -1.63986340e-01 -2.65463650e-01 1.88407227e-01
-5.19704521e-01 6.49941802e-01 -4.51454550e-01 1.55004784e-01
5.13633132e-01 3.16340208e-01 2.85181284e-01 -3.26045096e-01
5.62775433e-01 6.78811073e-01 2.53182650e-01 -1.62584305e+00
6.69526756e-01 -6.99225143e-02 -6.57258332e-02 -6.91978753e-01
-2.97319323e-01 2.07715780e-01 -3.51729125e-01 1.71653539e-01
5.40348649e-01 -2.04673290e-01 -7.87182391e-01 2.68213212e-01
-1.07871127e+00 -5.06181300e-01 -4.27799612e-01 -8.58773217e-02
-9.10563529e-01 6.32005274e-01 -4.81022507e-01 -7.15210617e-01
-2.99088806e-02 -1.28581297e+00 9.21512842e-01 1.36997595e-01
-3.79504412e-01 -1.08978343e+00 -4.67659235e-02 3.44246924e-02
3.92982751e-01 -4.14477587e-02 1.66678178e+00 -7.20019937e-01
-7.78485656e-01 1.61228999e-01 9.87104625e-02 1.89251021e-01
3.35939646e-01 4.85013008e-01 -7.40700424e-01 -1.21036202e-01
-1.41081989e-01 -4.23665643e-01 2.42806852e-01 -2.71819562e-01
1.28421807e+00 -5.84058225e-01 -2.87532397e-02 5.83538175e-01
1.43793023e+00 5.08924305e-01 9.09844756e-01 4.61736292e-01
3.75063539e-01 5.29674649e-01 4.40304428e-01 2.42496207e-01
3.91692877e-01 5.73251903e-01 2.74491251e-01 3.25407743e-01
-4.81462479e-02 -5.46489954e-01 5.65077424e-01 7.57760167e-01
4.55702618e-02 -1.18786916e-01 -8.86131406e-01 -1.34045668e-02
-1.50155210e+00 -1.07469916e+00 -3.25803980e-02 2.36938000e+00
1.13040233e+00 3.82177413e-01 5.60619570e-02 1.62642896e-01
3.75856847e-01 6.80790618e-02 -1.94956928e-01 -1.22363448e+00
-9.02668759e-02 5.97521305e-01 1.93659604e-01 6.34865403e-01
-8.02722991e-01 9.23753917e-01 7.26830959e+00 7.71494269e-01
-1.28072417e+00 -4.57594037e-01 1.18583977e-01 4.99288768e-01
-8.81170094e-01 4.42246050e-01 -1.10549951e+00 4.45884198e-01
1.28979337e+00 -2.58799583e-01 6.11395538e-01 7.42202520e-01
-5.01243174e-02 -8.66710320e-02 -1.50307465e+00 5.53785563e-01
-2.52016187e-01 -1.49460030e+00 5.88491380e-01 1.23357564e-01
-6.93004131e-02 -5.40702283e-01 -9.98173431e-02 5.95535517e-01
6.18423104e-01 -1.05083585e+00 8.80615413e-01 5.61023653e-01
6.95001304e-01 -5.71890533e-01 2.39607453e-01 3.83720160e-01
-1.03931105e+00 -3.02285969e-01 -2.33019561e-01 -3.00552189e-01
-2.06825703e-01 3.17747533e-01 -8.02749276e-01 6.89164996e-01
4.20242786e-01 3.55446368e-01 -5.65227866e-01 6.29158020e-01
-1.75612405e-01 5.43419659e-01 -4.51229244e-01 -3.72064590e-01
3.20812799e-02 -1.11991815e-01 5.08716226e-01 1.33685327e+00
3.67961168e-01 -1.09772809e-01 2.47404918e-01 1.02954459e+00
2.59569675e-01 6.35028305e-03 -9.28445816e-01 -2.73947626e-01
4.74096596e-01 8.68281603e-01 -4.16338593e-01 -3.43110412e-01
-6.63372874e-01 2.67196715e-01 3.82677257e-01 3.61406386e-01
-7.62444615e-01 -3.74888271e-01 5.08391798e-01 2.73644775e-01
2.87857383e-01 -3.76897097e-01 -2.14743063e-01 -1.12647581e+00
1.18489135e-02 -1.51198888e+00 6.30788445e-01 -6.83569908e-01
-1.13214886e+00 4.75947440e-01 5.32912433e-01 -1.02690351e+00
-5.01978755e-01 -8.44291806e-01 -5.25538206e-01 5.99825323e-01
-1.28821349e+00 -1.10489285e+00 3.23589817e-02 4.21145320e-01
2.26424396e-01 -2.36456364e-01 1.19795251e+00 2.94721033e-02
-2.40610197e-01 6.47070944e-01 -2.73385733e-01 3.82768773e-02
3.70132565e-01 -1.55097306e+00 2.60269672e-01 8.11497986e-01
3.55806798e-01 1.14339316e+00 6.59638464e-01 -5.97147226e-01
-1.90345597e+00 -7.12038934e-01 7.08798349e-01 -4.56450939e-01
8.10370386e-01 -4.70612884e-01 -1.14481354e+00 8.07715893e-01
5.33437096e-02 -3.28108460e-01 7.19798923e-01 1.17669404e-01
-7.29448974e-01 -2.18913421e-01 -1.01532602e+00 7.07321823e-01
1.01415920e+00 -6.31992638e-01 -9.93749857e-01 -3.92136872e-02
6.66215658e-01 -5.63166678e-01 -1.06941473e+00 1.92933217e-01
7.02139556e-01 -9.95550394e-01 6.10896051e-01 -1.06643510e+00
2.27108568e-01 -3.75956029e-01 -3.36250454e-01 -9.66453075e-01
1.44497558e-01 -1.07069647e+00 -3.21779668e-01 1.30475831e+00
4.15024608e-01 -8.18403244e-01 6.04621649e-01 8.22348058e-01
-3.50833803e-01 -7.49401748e-01 -6.91830099e-01 -8.29935312e-01
5.24266481e-01 -9.34401631e-01 9.90861118e-01 6.90397561e-01
1.79793894e-01 2.47262314e-01 3.62557732e-02 2.12362525e-03
2.66339451e-01 4.00003880e-01 1.01679659e+00 -1.30540264e+00
-6.93734884e-01 -4.72410589e-01 -2.12509915e-01 -1.18577969e+00
1.67530373e-01 -1.14680612e+00 -2.66815424e-01 -1.12445068e+00
-2.96136945e-01 -1.01590657e+00 1.23537272e-01 6.59647942e-01
1.47769839e-01 -6.06965899e-01 4.62368429e-01 -4.66905273e-02
-1.48192227e-01 -5.59841096e-02 1.16373551e+00 -1.42764434e-01
-8.44529420e-02 -2.38733813e-02 -9.34756160e-01 6.49866760e-01
8.87606382e-01 -4.26980942e-01 -6.91725016e-01 -3.67749572e-01
6.41849875e-01 8.17133486e-02 2.16551870e-01 -9.82226312e-01
9.39746350e-02 -6.41490340e-01 -9.32788625e-02 -1.60669982e-01
-1.78191319e-01 -9.11607444e-01 3.17296773e-01 3.43942136e-01
-3.91742945e-01 1.17157385e-01 4.75716054e-01 2.52009273e-01
5.58213657e-03 -6.73007548e-01 2.30241850e-01 -3.56170177e-01
-9.00317132e-01 -1.15852334e-01 -6.20784283e-01 3.85581166e-01
7.27612972e-01 -5.82068741e-01 -6.19022131e-01 -9.96599793e-02
-7.06043303e-01 -2.08052978e-01 6.02665544e-01 4.84751374e-01
4.78432655e-01 -8.96221042e-01 -1.70741662e-01 6.29600763e-01
-9.25074071e-02 -1.23578899e-01 -2.63458014e-01 2.76519775e-01
-6.59096301e-01 3.15195769e-01 -4.16344643e-01 -4.36016977e-01
-1.09814394e+00 5.55521548e-01 3.02649379e-01 -4.21214968e-01
-5.43744147e-01 5.16213067e-02 7.86237046e-02 -5.94810486e-01
2.16642693e-01 -9.64700043e-01 2.36319788e-02 -1.90478906e-01
5.34312665e-01 -1.58503935e-01 1.28010496e-01 2.99892984e-02
-8.75626653e-02 6.34052634e-01 -2.36623317e-01 1.51254833e-01
1.10924888e+00 1.85500503e-01 -1.90882906e-01 5.95301807e-01
7.31989384e-01 4.05680090e-01 -9.56182003e-01 2.96884388e-01
2.29930550e-01 -2.57364720e-01 -5.28314650e-01 -8.94746304e-01
-5.60215533e-01 8.74298275e-01 3.75696599e-01 7.42377043e-01
1.06844676e+00 -2.60258671e-02 6.33182943e-01 6.65450752e-01
8.71743679e-01 -1.03940177e+00 -2.55676091e-01 7.03736603e-01
8.09473574e-01 -6.04503393e-01 6.88036308e-02 -6.82021201e-01
-1.87196925e-01 1.79775476e+00 5.68010807e-01 1.70938656e-01
3.98378193e-01 9.26833272e-01 1.71484072e-02 -6.19748421e-02
-9.10173416e-01 -2.42907070e-02 -8.12454075e-02 1.02724302e+00
6.40081882e-01 -2.86453795e-02 -5.40961087e-01 5.83194315e-01
-5.11797190e-01 3.19328815e-01 8.99064064e-01 1.46586394e+00
-5.13570905e-01 -1.98782015e+00 -4.54609007e-01 5.17275095e-01
-6.04398251e-01 -1.33921146e-01 -2.23608673e-01 1.27426302e+00
1.77291751e-01 6.91555440e-01 9.14290026e-02 -2.06003472e-01
2.15475395e-01 3.45629156e-01 7.96658933e-01 -7.42976844e-01
-8.33523810e-01 -3.06794733e-01 5.10254383e-01 -5.63555181e-01
-3.09361905e-01 -6.47184193e-01 -1.45164943e+00 -4.01821405e-01
1.06637321e-01 2.70539016e-01 3.39989692e-01 8.61964583e-01
2.92134464e-01 2.06545532e-01 1.60522178e-01 -1.41567692e-01
-6.57608569e-01 -3.38689595e-01 -3.53185475e-01 4.35579896e-01
8.81538391e-02 -5.06291091e-01 -2.78357178e-01 1.81592584e-01]
|
[8.178014755249023, 7.3883891105651855]
|
db811cdd-093e-4a84-b75f-728010593a85
|
calculating-and-visualizing-counterfactual
|
2306.06506
| null |
https://arxiv.org/abs/2306.06506v1
|
https://arxiv.org/pdf/2306.06506v1.pdf
|
Calculating and Visualizing Counterfactual Feature Importance Values
|
Despite the success of complex machine learning algorithms, mostly justified by an outstanding performance in prediction tasks, their inherent opaque nature still represents a challenge to their responsible application. Counterfactual explanations surged as one potential solution to explain individual decision results. However, two major drawbacks directly impact their usability: (1) the isonomic view of feature changes, in which it is not possible to observe \textit{how much} each modified feature influences the prediction, and (2) the lack of graphical resources to visualize the counterfactual explanation. We introduce Counterfactual Feature (change) Importance (CFI) values as a solution: a way of assigning an importance value to each feature change in a given counterfactual explanation. To calculate these values, we propose two potential CFI methods. One is simple, fast, and has a greedy nature. The other, coined CounterShapley, provides a way to calculate Shapley values between the factual-counterfactual pair. Using these importance values, we additionally introduce three chart types to visualize the counterfactual explanations: (a) the Greedy chart, which shows a greedy sequential path for prediction score increase up to predicted class change, (b) the CounterShapley chart, depicting its respective score in a simple and one-dimensional chart, and finally (c) the Constellation chart, which shows all possible combinations of feature changes, and their impact on the model's prediction score. For each of our proposed CFI methods and visualization schemes, we show how they can provide more information on counterfactual explanations. Finally, an open-source implementation is offered, compatible with any counterfactual explanation generator algorithm. Code repository at: https://github.com/ADMAntwerp/CounterPlots
|
['David Martens', 'Raphael Mazzine Barbosa de Oliveira', 'Bjorge Meulemeester']
|
2023-06-10
| null | null | null | null |
['counterfactual-explanation']
|
['miscellaneous']
|
[ 2.76278704e-01 6.04444206e-01 -4.71109390e-01 -2.93310702e-01
-1.52798578e-01 -5.63942134e-01 9.23045516e-01 2.49617234e-01
-1.94519348e-02 1.32130492e+00 3.88163060e-01 -9.13090765e-01
-5.27243435e-01 -7.01839983e-01 -6.12417579e-01 -6.66267812e-01
-3.79355103e-01 2.41874844e-01 -2.44496882e-01 1.26008410e-02
6.70674920e-01 4.24558610e-01 -1.93235266e+00 3.80937815e-01
1.20544541e+00 7.63225257e-01 7.33313188e-02 3.71801674e-01
-2.59093851e-01 6.27030313e-01 -6.95559740e-01 -6.94693387e-01
2.39865571e-01 -6.81403816e-01 -6.36886656e-01 -3.63663107e-01
3.38090435e-02 -1.45305410e-01 3.20590049e-01 8.69968772e-01
1.52846143e-01 -1.61055535e-01 9.06466722e-01 -1.81243527e+00
-6.05640471e-01 9.73326981e-01 -5.70802391e-01 -3.19936946e-02
6.73472047e-01 3.05283457e-01 1.07126129e+00 -5.72079360e-01
6.63740456e-01 1.27467203e+00 3.21846128e-01 3.00960600e-01
-1.36255813e+00 -6.97764575e-01 2.67895788e-01 4.15348172e-01
-7.38522947e-01 1.88715100e-01 9.07644510e-01 -5.62775970e-01
7.33627021e-01 1.03089070e+00 1.08477080e+00 8.63715768e-01
3.73633325e-01 6.08393610e-01 1.56573272e+00 -6.11675560e-01
4.64560807e-01 3.19544852e-01 -3.01496554e-02 3.11447203e-01
7.55111039e-01 5.02417922e-01 -3.67487788e-01 -4.54735756e-01
5.12682796e-01 2.37373620e-01 -4.30489391e-01 -7.71856248e-01
-1.21642542e+00 9.83215690e-01 5.94144285e-01 2.95402646e-01
-3.44991654e-01 7.55835474e-02 1.18775502e-01 2.94103593e-01
3.65754187e-01 7.65554011e-01 -7.53074646e-01 -7.15762097e-03
-6.47951484e-01 7.06615984e-01 7.71284401e-01 2.35016197e-01
6.92005217e-01 5.08412495e-02 -2.32152507e-01 3.89624327e-01
2.33092085e-01 3.27957273e-01 7.45453238e-01 -7.78292537e-01
3.50248516e-01 8.98091495e-01 5.31471252e-01 -9.54178393e-01
-5.65176249e-01 -5.02431512e-01 -7.40617990e-01 7.74060726e-01
6.63172781e-01 -2.13460132e-01 -4.98715460e-01 1.73204970e+00
3.80314618e-01 -1.69675544e-01 -3.79792824e-02 8.68323565e-01
3.73998910e-01 3.79423618e-01 1.82602465e-01 -7.68073261e-01
1.20629454e+00 -5.75794280e-01 -7.01566279e-01 5.07760309e-02
8.44671965e-01 -6.23786092e-01 1.30350947e+00 3.62858385e-01
-8.06023061e-01 -1.46618769e-01 -1.01424217e+00 3.92038584e-01
-5.96012294e-01 -7.44501874e-02 1.01307905e+00 8.24046850e-01
-7.32134163e-01 1.02901971e+00 -4.02460635e-01 -3.21580954e-02
2.45622605e-01 2.34561026e-01 -1.76341891e-01 3.47909272e-01
-1.27439976e+00 1.04202557e+00 5.17968237e-01 -2.05474436e-01
-2.99795687e-01 -9.81736302e-01 -7.29552388e-01 5.44274092e-01
3.28084677e-01 -8.21979523e-01 1.04028702e+00 -1.16276062e+00
-1.20030117e+00 5.40840566e-01 6.52063452e-03 -5.84051192e-01
1.02831519e+00 2.16839407e-02 -4.25900400e-01 -3.72441113e-01
7.41069987e-02 5.19989431e-01 5.28822124e-01 -1.46969092e+00
-6.40881956e-01 -4.11082983e-01 1.41916990e-01 3.01599383e-01
3.64201665e-02 -2.26598412e-01 5.47511637e-01 -7.31991947e-01
-7.18439519e-02 -6.49211228e-01 -3.98289591e-01 -9.56003442e-02
-6.98352456e-01 -1.66645840e-01 5.82477510e-01 -2.40334868e-01
1.59324193e+00 -1.75126600e+00 -3.77005219e-01 1.25177085e-01
7.64923394e-02 5.24422042e-02 2.91244119e-01 4.34112042e-01
-7.70028234e-01 5.41202068e-01 -3.57455224e-01 3.34156662e-01
3.21976602e-01 -1.66365221e-01 -4.51064616e-01 1.24292463e-01
1.02553293e-02 7.67113328e-01 -9.61406708e-01 -2.02149853e-01
4.10148293e-01 1.47670418e-01 -5.53613603e-01 -7.32674673e-02
-1.60463795e-01 5.92375211e-02 -1.64115846e-01 1.77893803e-01
7.99052715e-01 -9.47345570e-02 6.61800742e-01 1.12405144e-01
-4.82414007e-01 5.24319232e-01 -1.44105911e+00 8.39482129e-01
-2.22923219e-01 4.31523293e-01 -6.37827396e-01 -7.14202464e-01
9.68691528e-01 3.69926900e-01 2.36130506e-01 -4.17866170e-01
-7.97004811e-03 3.53782594e-01 1.43681124e-01 -1.95452467e-01
3.77626300e-01 -5.22648036e-01 3.25619616e-02 8.13749969e-01
-4.55874771e-01 2.95055588e-03 2.99446225e-01 1.12709388e-01
6.84333026e-01 1.98087782e-01 1.26608121e+00 -3.67437661e-01
3.44166666e-01 -2.12300178e-02 5.60659349e-01 6.95282102e-01
5.44926040e-02 5.19081831e-01 1.16319561e+00 -9.98423457e-01
-7.61608899e-01 -9.82379556e-01 -1.56332299e-01 4.91994441e-01
-1.05689332e-01 -2.11439624e-01 -4.80775386e-01 -8.94988596e-01
3.73776734e-01 1.55923522e+00 -1.02976573e+00 -1.36475086e-01
-2.04581290e-01 -9.67644870e-01 -1.65066138e-01 2.03130543e-01
3.57230753e-01 -1.04487455e+00 -1.14309049e+00 -2.33993798e-01
-2.49756649e-01 8.09838250e-03 -4.33787927e-02 9.26403701e-02
-1.12489164e+00 -1.27930176e+00 -5.24094701e-01 1.47588000e-01
5.29767096e-01 2.17989489e-01 1.10286546e+00 9.49788019e-02
-4.12272289e-03 -1.43383473e-01 -1.63564816e-01 -8.55568230e-01
-3.68029535e-01 -5.19018233e-01 1.06734104e-01 -2.08953232e-01
2.03195527e-01 -7.74796009e-01 -9.44082081e-01 1.66099817e-01
-5.61003089e-01 3.88820499e-01 4.86627787e-01 7.62734175e-01
3.51025969e-01 -2.34323621e-01 6.80797219e-01 -1.10612988e+00
7.54097939e-01 -6.67742312e-01 -5.69008768e-01 2.71697015e-01
-1.21147895e+00 2.01007545e-01 7.73639917e-01 -2.32155710e-01
-1.21541810e+00 -6.64798915e-02 1.70927569e-01 -8.54869559e-02
-2.32474074e-01 4.50229019e-01 -1.10940278e-01 5.71681619e-01
9.81592536e-01 -4.68611419e-02 -8.36234614e-02 -5.05507052e-01
6.13208652e-01 2.37795398e-01 2.02766612e-01 -1.81117579e-01
7.20758557e-01 2.69426852e-01 -5.94106456e-03 -1.46373510e-01
-4.22990561e-01 2.54031837e-01 -5.68377912e-01 -3.21109682e-01
5.38189888e-01 -2.85453320e-01 -1.06572700e+00 -1.65451199e-01
-1.07116365e+00 -6.97423741e-02 -5.71223676e-01 6.26603782e-01
-6.89833403e-01 9.29706097e-02 1.81026474e-01 -1.16277540e+00
-1.06466770e-01 -8.65716815e-01 2.60638416e-01 3.17587584e-01
-5.90937674e-01 -9.96693790e-01 6.03673607e-02 1.41518548e-01
9.63395834e-02 7.92395949e-01 1.23407781e+00 -7.10984647e-01
-4.68677521e-01 -1.90760359e-01 -1.28426880e-01 -2.70273566e-01
1.35817900e-01 1.99761212e-01 -9.57129598e-01 -4.11627926e-02
-1.03016526e-01 1.55762464e-01 5.08221924e-01 6.80449128e-01
1.05932367e+00 -9.96815562e-01 -5.05759597e-01 1.62457481e-01
1.34118915e+00 4.80650097e-01 6.28375649e-01 4.89919871e-01
1.19545810e-01 7.04141855e-01 8.02179217e-01 6.77298546e-01
2.28083074e-01 7.53528833e-01 7.25102186e-01 -3.17863494e-01
5.83975613e-02 -4.82654393e-01 1.37593180e-01 -2.03911420e-02
-4.20949608e-01 7.43640289e-02 -7.15244889e-01 3.19563657e-01
-1.90663850e+00 -1.20339525e+00 -5.99807799e-01 2.56859779e+00
4.18686032e-01 2.85762727e-01 2.97631174e-01 4.34715629e-01
5.67816973e-01 1.32050663e-01 -4.97196347e-01 -9.87455964e-01
8.26826319e-02 -2.83011943e-01 1.46215081e-01 5.41986346e-01
-6.87738359e-01 2.36866161e-01 6.52606869e+00 5.68073571e-01
-9.34242785e-01 -1.70050040e-01 8.76225293e-01 -2.39388078e-01
-9.12152946e-01 3.66533995e-01 -7.35802501e-02 6.93394184e-01
7.46044338e-01 -9.87009048e-01 8.18700939e-02 1.05245566e+00
6.40818238e-01 -3.36149663e-01 -1.11513281e+00 6.18833423e-01
-4.68178540e-01 -1.63637567e+00 4.68928367e-01 1.03564680e-01
6.90486491e-01 -6.54945016e-01 4.66830619e-02 2.69993134e-02
4.79259968e-01 -8.19674551e-01 8.89015317e-01 4.12988544e-01
7.77811646e-01 -7.50473797e-01 6.44818723e-01 3.27548921e-01
-7.94611752e-01 -3.92447442e-01 -1.79286420e-01 -7.88013041e-01
-1.07136734e-01 7.66144574e-01 -9.64851797e-01 7.94718206e-01
4.84669864e-01 2.00074211e-01 -4.74865556e-01 9.88926113e-01
-5.87718248e-01 5.26188850e-01 8.34659189e-02 -2.65383512e-01
-1.31534025e-01 -1.56869426e-01 5.63251913e-01 9.57697451e-01
6.08852327e-01 4.55674902e-02 -6.15859032e-01 1.24281979e+00
2.78529197e-01 2.58337975e-01 -8.77224028e-01 4.93565440e-01
5.78347504e-01 9.59224641e-01 -8.59165311e-01 -3.21921736e-01
-2.65075088e-01 5.89712858e-01 6.81419820e-02 3.50650072e-01
-7.82329679e-01 -2.09464043e-01 7.49641418e-01 2.20356017e-01
-1.05077937e-01 4.85610992e-01 -8.93612564e-01 -1.10115755e+00
-3.31339687e-02 -9.25747454e-01 8.08788538e-01 -9.58039165e-01
-8.35178137e-01 3.25233281e-01 3.05407345e-01 -1.44504917e+00
-5.27610064e-01 -4.18427378e-01 -9.95419443e-01 9.06827390e-01
-1.10232508e+00 -7.38422453e-01 4.69000041e-02 2.02792421e-01
1.81685761e-01 -3.87969473e-03 9.08031702e-01 -1.96462706e-01
-4.04897064e-01 4.43590641e-01 1.70539301e-02 -5.97835183e-01
2.83346385e-01 -1.60088313e+00 2.80490607e-01 6.77362621e-01
1.61638692e-01 6.80262089e-01 1.24795413e+00 -5.81432283e-01
-6.36142790e-01 -4.65009809e-01 1.26076531e+00 -4.48661864e-01
4.64364052e-01 -3.43222991e-02 -7.42601633e-01 6.01241112e-01
5.40076680e-02 -5.08676767e-01 7.20836759e-01 3.50922704e-01
-2.14657903e-01 -1.10251710e-01 -1.25904179e+00 1.08058846e+00
7.12409794e-01 2.32987642e-01 -8.97499621e-01 2.06819534e-01
6.62034631e-01 1.34510789e-02 -4.92394716e-01 2.20940143e-01
8.77194703e-01 -1.81382751e+00 8.26981068e-01 -6.23148143e-01
8.03615093e-01 -2.81486928e-01 2.15839684e-01 -1.59338582e+00
-3.77594233e-01 -6.79437280e-01 6.85435236e-02 1.03473783e+00
8.06578577e-01 -1.08650684e+00 5.86659610e-01 9.67478573e-01
1.77759409e-01 -1.07287169e+00 -9.43310142e-01 -6.34918034e-01
5.07889204e-02 -4.13483381e-01 1.21471226e+00 1.22976053e+00
6.33550704e-01 7.40024596e-02 -3.20002168e-01 -2.70490050e-01
5.70272386e-01 6.28582537e-01 7.87956595e-01 -1.19416809e+00
-3.44116330e-01 -7.53577769e-01 -2.70534039e-01 -3.06523204e-01
-3.89581501e-01 -9.58020031e-01 -7.56764531e-01 -1.46152222e+00
4.09089625e-01 -2.27441609e-01 -3.04302245e-01 5.22521198e-01
-3.89532983e-01 -1.64853930e-01 5.72637737e-01 2.18849093e-01
1.71491757e-01 4.17196184e-01 1.00725675e+00 2.16215119e-01
-4.19229388e-01 1.41589940e-01 -1.01873744e+00 9.81305361e-01
8.46483290e-01 -6.11956120e-01 -5.43529630e-01 2.50355810e-01
2.68025517e-01 3.16450685e-01 4.80618954e-01 -6.68595493e-01
-3.63452613e-01 -5.70424855e-01 5.76578081e-01 -4.58567917e-01
-4.44521457e-02 -6.62262142e-01 6.52483582e-01 1.08739448e+00
-4.39254463e-01 1.88609228e-01 1.40669152e-01 4.25056517e-01
-3.75380702e-02 -2.43918747e-01 5.72924614e-01 -2.04151437e-01
-3.01792175e-01 -3.28760386e-01 -2.84636557e-01 -4.27612543e-01
1.16250741e+00 -4.46578681e-01 -5.37727416e-01 -5.14140785e-01
-6.10622883e-01 1.72583953e-01 4.55574363e-01 2.96982259e-01
3.77625585e-01 -1.49304998e+00 -5.92917264e-01 -4.64428850e-02
2.02593170e-02 -7.52971649e-01 4.67643142e-01 7.72806227e-01
-2.81767488e-01 5.62937856e-01 -5.52824795e-01 -2.38576923e-02
-1.20228946e+00 8.24769914e-01 2.42226169e-01 -6.42158806e-01
-4.50737119e-01 3.44807863e-01 5.60006440e-01 -3.65805507e-01
-2.54202574e-01 -3.74896169e-01 -5.04439414e-01 2.26632133e-01
6.30220413e-01 7.38634229e-01 -1.95920765e-01 -6.27658069e-02
-3.51393849e-01 1.81852907e-01 1.95115730e-01 -9.74526927e-02
1.25415778e+00 -8.67268816e-02 -4.75303130e-03 5.15062392e-01
6.60721302e-01 -1.17606863e-01 -1.22554100e+00 4.89632457e-01
2.42620438e-01 -8.18229556e-01 -5.48273563e-01 -1.35558510e+00
-4.75011706e-01 5.94349205e-01 4.19009328e-01 7.98876345e-01
1.21404231e+00 -2.88058102e-01 -1.70897678e-01 4.68498133e-02
2.27788910e-01 -6.62599266e-01 -4.18226004e-01 -1.53683931e-01
1.41302812e+00 -1.13144660e+00 3.61823350e-01 -3.68166238e-01
-8.69029284e-01 1.07714200e+00 2.66968638e-01 2.57180899e-01
4.31696475e-01 -7.45726153e-02 1.78200543e-01 -1.23052381e-01
-1.09897709e+00 -3.59572656e-03 2.74804950e-01 5.45497179e-01
7.81514764e-01 6.41695380e-01 -1.21200097e+00 8.47657979e-01
-7.98232377e-01 1.07359186e-01 7.05251336e-01 4.36975896e-01
-2.05178171e-01 -9.45515752e-01 -6.32291198e-01 8.11109424e-01
-2.93110371e-01 6.43637702e-02 -5.27686238e-01 1.19474912e+00
1.50401488e-01 8.49558234e-01 -4.83068079e-02 -2.52655834e-01
4.54403073e-01 2.11910263e-01 4.09965292e-02 -1.76734388e-01
-4.50171679e-01 -2.78194934e-01 1.77482754e-01 -7.08759367e-01
-1.43816933e-01 -8.04792702e-01 -1.18181169e+00 -4.55788404e-01
-1.97893783e-01 5.05531192e-01 6.13349557e-01 7.87497818e-01
2.65221715e-01 2.97339112e-01 6.58571005e-01 -8.08158159e-01
-5.60530484e-01 -8.81940961e-01 -7.17947960e-01 4.85213786e-01
1.29722789e-01 -9.23027337e-01 -6.76508963e-01 -2.77157515e-01]
|
[8.735918998718262, 5.636493682861328]
|
2b1264e8-3495-427a-bc26-7913ed21abe6
|
upcycling-models-under-domain-and-category
|
2303.0711
| null |
https://arxiv.org/abs/2303.07110v1
|
https://arxiv.org/pdf/2303.07110v1.pdf
|
Upcycling Models under Domain and Category Shift
|
Deep neural networks (DNNs) often perform poorly in the presence of domain shift and category shift. How to upcycle DNNs and adapt them to the target task remains an important open problem. Unsupervised Domain Adaptation (UDA), especially recently proposed Source-free Domain Adaptation (SFDA), has become a promising technology to address this issue. Nevertheless, existing SFDA methods require that the source domain and target domain share the same label space, consequently being only applicable to the vanilla closed-set setting. In this paper, we take one step further and explore the Source-free Universal Domain Adaptation (SF-UniDA). The goal is to identify "known" data samples under both domain and category shift, and reject those "unknown" data samples (not present in source classes), with only the knowledge from standard pre-trained source model. To this end, we introduce an innovative global and local clustering learning technique (GLC). Specifically, we design a novel, adaptive one-vs-all global clustering algorithm to achieve the distinction across different target classes and introduce a local k-NN clustering strategy to alleviate negative transfer. We examine the superiority of our GLC on multiple benchmarks with different category shift scenarios, including partial-set, open-set, and open-partial-set DA. Remarkably, in the most challenging open-partial-set DA scenario, GLC outperforms UMAD by 14.8\% on the VisDA benchmark. The code is available at https://github.com/ispc-lab/GLC.
|
['Changjun Jiang', 'DaCheng Tao', 'Guang Chen', 'Cewu Lu', 'Florian Roehrbein', 'Tianpei Zou', 'Sanqing Qu']
|
2023-03-13
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Qu_Upcycling_Models_Under_Domain_and_Category_Shift_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Qu_Upcycling_Models_Under_Domain_and_Category_Shift_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['source-free-domain-adaptation', 'universal-domain-adaptation']
|
['computer-vision', 'computer-vision']
|
[ 1.72733605e-01 -3.79231423e-01 -1.96548671e-01 -4.94776577e-01
-7.44116068e-01 -8.43136191e-01 5.66624939e-01 4.02408168e-02
-5.93170881e-01 8.46698761e-01 -8.71480256e-02 -1.17358543e-01
-1.11519620e-01 -6.06114686e-01 -6.05733693e-01 -1.04015696e+00
4.07162279e-01 7.05027580e-01 3.72639805e-01 -1.04575157e-01
-8.19654316e-02 4.61050510e-01 -1.42671525e+00 8.63673985e-02
1.10988212e+00 8.87395859e-01 3.95722866e-01 8.49627182e-02
-2.64045656e-01 1.83209091e-01 -6.19309127e-01 -2.28109047e-01
2.88215578e-01 -5.02799451e-01 -7.18071699e-01 1.25264615e-01
2.94929743e-01 -4.22607400e-02 2.96174223e-03 1.25881314e+00
6.74709082e-01 3.18650454e-01 9.95744348e-01 -1.31864536e+00
-5.58719754e-01 3.54752004e-01 -5.12778759e-01 2.85106599e-01
-3.01041842e-01 -9.19633172e-03 6.43980265e-01 -1.09066260e+00
6.68279350e-01 1.02038038e+00 3.88363361e-01 8.83460224e-01
-1.38814473e+00 -9.57336605e-01 4.02437150e-01 2.46406883e-01
-1.54199529e+00 -4.01593775e-01 9.92695153e-01 -5.57638288e-01
4.53522623e-01 -6.81311265e-02 1.26852319e-01 1.47625065e+00
-3.25491965e-01 7.47792780e-01 1.11167645e+00 -3.60401362e-01
6.08039379e-01 3.17579806e-01 3.82844746e-01 -3.64628136e-02
1.62257507e-01 -1.98556632e-01 -1.42225057e-01 -2.18685810e-02
5.43486714e-01 4.28347290e-02 -2.61359215e-01 -7.25071549e-01
-1.06066823e+00 7.40088105e-01 2.97162533e-01 4.09632295e-01
-2.31027514e-01 -5.35870075e-01 5.33359051e-01 3.47765923e-01
5.39496005e-01 7.71022215e-02 -7.87513435e-01 1.74166173e-01
-6.17667794e-01 1.72827527e-01 6.05687201e-01 1.12122607e+00
8.41085255e-01 1.05174361e-02 -1.10981062e-01 1.31542003e+00
1.42361701e-01 4.81677771e-01 8.77921879e-01 -6.64041758e-01
5.34371138e-01 6.34396911e-01 -1.09104656e-01 -6.14249229e-01
-3.82285535e-01 -8.44078541e-01 -1.13904846e+00 1.51566833e-01
5.43706417e-01 -2.68391848e-01 -1.11817932e+00 2.00014067e+00
5.69892883e-01 2.48364687e-01 4.28359926e-01 8.18561912e-01
6.76240623e-01 5.50739408e-01 1.89116120e-01 -2.18232438e-01
1.09512079e+00 -7.75886238e-01 -4.75303352e-01 -3.26903015e-01
7.18451679e-01 -4.18026209e-01 1.11464059e+00 3.15525085e-01
-5.18394649e-01 -6.31434619e-01 -1.03957200e+00 1.19629838e-01
-6.50069892e-01 1.95845932e-01 -2.98970640e-02 4.40969944e-01
-7.49860168e-01 2.33039051e-01 -5.65060735e-01 -5.69859684e-01
6.35685742e-01 3.23998392e-01 -3.79043818e-01 -4.27703053e-01
-1.25891423e+00 6.20819211e-01 7.81872571e-01 -3.39629591e-01
-8.30611825e-01 -7.19898820e-01 -5.76586008e-01 -4.43643332e-02
4.53700364e-01 -4.85541254e-01 1.19095767e+00 -1.28421855e+00
-1.41364145e+00 1.05272472e+00 -7.79343620e-02 -4.71866429e-01
4.07924652e-01 -5.35055399e-02 -6.20263875e-01 -1.07298590e-01
1.88462734e-01 6.95820630e-01 7.65390396e-01 -1.44392300e+00
-6.06230140e-01 -5.82808614e-01 -4.37867254e-01 3.16474199e-01
-5.63039780e-01 -5.90757839e-02 -3.02874237e-01 -8.52848887e-01
8.43378454e-02 -8.03071141e-01 7.83305466e-02 -1.49361491e-01
-2.39794686e-01 -4.18641418e-01 8.56548667e-01 -3.95781100e-01
1.23389816e+00 -2.37031341e+00 2.41743296e-01 2.40969256e-01
2.98252068e-02 6.88654065e-01 -1.31062225e-01 7.84482050e-04
-2.94590414e-01 -1.43468425e-01 -6.78147316e-01 -3.92486900e-01
-2.55477764e-02 3.46769750e-01 -1.92761824e-01 2.38428473e-01
1.57440737e-01 4.00791079e-01 -7.69714355e-01 -3.58286202e-01
9.29630734e-03 2.71760046e-01 -4.79629755e-01 4.22523022e-02
-1.78544968e-01 5.59955895e-01 -2.07243294e-01 5.61557651e-01
9.37216163e-01 -1.03924200e-01 1.45530745e-01 4.46934626e-02
7.36633763e-02 -5.10566160e-02 -1.38069308e+00 1.70652652e+00
-1.00192174e-01 3.82439464e-01 9.94922295e-02 -1.19334102e+00
1.17780304e+00 2.07690284e-01 3.46857369e-01 -7.26383507e-01
1.92088559e-01 5.35108745e-01 8.45168531e-02 -9.87292379e-02
6.25952631e-02 -3.35717499e-01 -1.24067262e-01 8.83130953e-02
3.25619996e-01 4.94714826e-01 7.24352449e-02 6.20970204e-02
9.55746651e-01 -3.79661545e-02 3.75348508e-01 -4.19725895e-01
6.18193805e-01 6.34511784e-02 9.75465596e-01 4.68095124e-01
-6.15726173e-01 8.05526555e-01 2.79243767e-01 -6.50694594e-02
-8.52148354e-01 -1.29493976e+00 -2.57701486e-01 1.28972781e+00
2.68369287e-01 2.22531363e-01 -8.27088535e-01 -8.57656181e-01
-6.63455799e-02 8.48390579e-01 -3.17594469e-01 -4.00091469e-01
-3.65437329e-01 -7.90507436e-01 4.60907847e-01 4.86090362e-01
6.73859894e-01 -9.62232769e-01 -7.10863844e-02 2.38214254e-01
-2.24151328e-01 -1.02360332e+00 -4.73835081e-01 4.10704195e-01
-8.06646943e-01 -7.59371042e-01 -1.11782050e+00 -1.03972161e+00
6.80739522e-01 2.12805033e-01 8.16977859e-01 -5.35561919e-01
3.33446175e-01 7.49627948e-02 -4.50062841e-01 -3.05963308e-01
-4.71496046e-01 3.31569493e-01 4.04805660e-01 3.07415813e-01
7.56003261e-01 -6.52259707e-01 -5.09910226e-01 6.22668564e-01
-9.67576325e-01 -9.86749679e-02 5.55720508e-01 7.15919793e-01
7.88090587e-01 1.83722705e-01 1.08043098e+00 -1.04627037e+00
4.87918943e-01 -9.46354270e-01 -3.92198414e-01 1.64923996e-01
-5.35142064e-01 -2.19106242e-01 9.11761463e-01 -6.88747764e-01
-1.18830657e+00 2.64083028e-01 -1.24474764e-01 -8.22484612e-01
-7.03837872e-01 3.12790006e-01 -8.81101787e-01 3.50536734e-01
1.00594866e+00 3.58891577e-01 -1.75636262e-01 -8.46086323e-01
1.86034277e-01 9.24080491e-01 7.03924000e-01 -5.65272570e-01
8.25059116e-01 4.78572845e-01 -4.85678285e-01 -6.28183842e-01
-6.09961510e-01 -7.62470841e-01 -1.05751657e+00 6.40825629e-02
6.92296863e-01 -1.17171252e+00 2.31967881e-01 8.41547132e-01
-9.87229526e-01 -6.69027567e-01 -2.12439492e-01 3.55493903e-01
-2.40338042e-01 2.86316723e-01 -3.80005985e-02 -4.53309923e-01
-2.86168337e-01 -9.63146210e-01 6.77872956e-01 2.96836674e-01
-6.94262832e-02 -9.31839585e-01 -3.26791033e-02 1.70246392e-01
1.96502879e-01 2.03753188e-01 8.50284576e-01 -1.35920835e+00
-5.39579876e-02 8.43144432e-02 -2.66398430e-01 7.77297854e-01
3.48036379e-01 -3.49839777e-01 -1.02775443e+00 -4.09050465e-01
-1.08585976e-01 -8.29093903e-02 8.76537144e-01 2.96157837e-01
1.12921691e+00 -8.45529884e-02 -3.99825662e-01 5.78452110e-01
1.35894465e+00 3.95555645e-01 3.97887170e-01 4.64089304e-01
7.61435509e-01 4.69822019e-01 6.89495564e-01 4.11467463e-01
3.44902277e-01 6.60092592e-01 1.20260119e-01 5.90662658e-02
-2.14577720e-01 -8.87299553e-02 4.16547000e-01 7.31794953e-01
3.66494268e-01 -4.85215098e-01 -1.11437726e+00 9.01092350e-01
-1.73331952e+00 -5.68802476e-01 -8.80666971e-02 2.28096557e+00
9.75328803e-01 2.89677233e-01 3.40081573e-01 1.58938795e-01
1.10136163e+00 -3.22738171e-01 -8.66543949e-01 -1.04195356e-01
-3.63675684e-01 9.00487378e-02 4.20996040e-01 1.51181504e-01
-1.36416233e+00 9.56869304e-01 4.06440401e+00 1.26300645e+00
-1.08203113e+00 2.89466977e-01 6.04054093e-01 8.11823383e-02
-3.88148315e-02 -2.88147926e-01 -8.49708915e-01 7.62940049e-01
8.99172425e-01 -4.93938401e-02 1.86540678e-01 1.03498936e+00
6.18524253e-02 1.96000338e-01 -9.70838726e-01 9.62758243e-01
-5.93891218e-02 -8.41796875e-01 3.93694676e-02 -5.20824417e-02
8.55077922e-01 2.42187530e-01 -1.66709591e-02 6.15502894e-01
2.36919254e-01 -4.00918305e-01 4.67335850e-01 9.37179849e-02
9.42779720e-01 -8.24542224e-01 6.56347990e-01 5.34450233e-01
-9.36828613e-01 -1.78292871e-01 -4.51019138e-01 2.52598584e-01
-1.58482999e-01 6.07290268e-01 -6.00645363e-01 5.58841586e-01
8.54115963e-01 7.53488064e-01 -4.68290240e-01 1.08352733e+00
2.70065516e-02 7.43157089e-01 -3.85653317e-01 4.13531363e-01
9.51327160e-02 -1.06378317e-01 6.81705415e-01 1.15387225e+00
3.35510880e-01 7.81317502e-02 3.76613326e-02 6.66841686e-01
-2.06897587e-01 1.49443775e-01 -4.41792011e-01 2.79808223e-01
7.45825171e-01 9.93577182e-01 -7.83737659e-01 -3.46834362e-01
-3.29563856e-01 1.12675130e+00 3.27925414e-01 5.86309493e-01
-7.56589949e-01 -4.37536716e-01 8.90543580e-01 1.19334504e-01
3.86032730e-01 -9.92660373e-02 -3.05527300e-01 -1.22412395e+00
3.00016254e-04 -8.10733020e-01 7.93348968e-01 -4.54457581e-01
-1.66035998e+00 4.02985185e-01 1.91306978e-01 -1.53107333e+00
9.31089520e-02 -5.67701161e-01 -5.63449502e-01 7.38229156e-01
-1.66441250e+00 -8.83921802e-01 -3.50157797e-01 9.81110215e-01
6.79368973e-01 -4.26011473e-01 5.52484632e-01 7.78174222e-01
-7.73835182e-01 9.51983273e-01 8.18771362e-01 3.14101785e-01
1.19343948e+00 -1.09594858e+00 3.66630554e-02 8.69612098e-01
-1.81305602e-01 4.86141264e-01 4.79841292e-01 -4.41621780e-01
-6.71474278e-01 -1.61349833e+00 6.29355788e-01 -2.68177927e-01
4.59721982e-01 -5.38369656e-01 -1.48623347e+00 5.91068864e-01
-1.42575160e-01 6.70817643e-02 6.96190000e-01 -1.80414200e-01
-5.02217472e-01 -4.24325168e-01 -1.33386481e+00 4.24021393e-01
1.07680106e+00 -2.40511388e-01 -5.48743010e-01 1.70674473e-01
7.95605004e-01 -2.14881986e-01 -8.29395473e-01 4.91676420e-01
6.97373152e-02 -7.36644387e-01 9.25028622e-01 -3.91302586e-01
2.27738902e-01 -5.39710879e-01 -3.01103652e-01 -1.53330445e+00
-3.96490723e-01 -1.34020790e-01 7.37527385e-02 1.71406245e+00
2.84613311e-01 -9.40739334e-01 6.09563887e-01 3.43387783e-01
-4.29986089e-01 -3.05925399e-01 -1.07841837e+00 -1.12032592e+00
6.04360878e-01 -2.44830042e-01 5.57270646e-01 1.27093995e+00
-3.57881248e-01 4.02666390e-01 -4.23487946e-02 2.76502311e-01
5.35537839e-01 -6.63637966e-02 5.73859990e-01 -1.52191472e+00
-1.18475534e-01 -4.52563852e-01 -2.00425148e-01 -9.37173545e-01
2.73085386e-01 -1.07058644e+00 1.76493935e-02 -1.21102953e+00
1.56520866e-02 -5.60768723e-01 -6.39416337e-01 5.38625360e-01
-1.39415011e-01 3.28346975e-02 1.05914399e-01 4.91830349e-01
-6.22439742e-01 7.16294587e-01 9.28365886e-01 5.32807875e-03
-4.29776669e-01 1.29169583e-01 -7.69090533e-01 4.63493407e-01
1.18439555e+00 -6.13391042e-01 -5.97423673e-01 -4.26128000e-01
-4.16054070e-01 -5.73510289e-01 3.77665132e-01 -1.33608258e+00
1.69312626e-01 -2.01391026e-01 2.75538325e-01 -4.72535223e-01
4.51448634e-02 -9.55213428e-01 -2.30982546e-02 1.84818506e-01
-1.80791691e-01 -3.42015088e-01 3.74285132e-01 7.71674156e-01
-2.57182211e-01 -7.95556828e-02 1.26295877e+00 1.26695901e-01
-1.16512740e+00 3.63847256e-01 -2.28777647e-01 2.89646417e-01
1.21578717e+00 -2.45452285e-01 -4.03066337e-01 5.08276038e-02
-7.43427455e-01 3.61746490e-01 4.32018578e-01 4.47468102e-01
3.11950177e-01 -1.40603435e+00 -6.54329956e-01 2.75055259e-01
4.90981966e-01 5.17223895e-01 4.87190336e-01 6.14970982e-01
-1.04102865e-01 1.79480910e-01 -2.38562390e-01 -6.94908559e-01
-1.16899765e+00 6.30082369e-01 3.84669244e-01 -1.01912469e-02
-3.41379255e-01 8.15572560e-01 6.28843665e-01 -9.21795011e-01
2.30569810e-01 -1.64803579e-01 -3.20849776e-01 1.19366184e-01
3.09975922e-01 2.89318770e-01 1.70173571e-01 -7.25178719e-01
-4.38995689e-01 3.75519454e-01 -2.01150969e-01 2.61400014e-01
1.12450111e+00 -3.07219625e-01 2.96960533e-01 4.02598768e-01
1.41216755e+00 -4.48615611e-01 -1.34966719e+00 -7.59633958e-01
1.49832323e-01 -4.97944765e-02 -2.11051852e-01 -9.39731002e-01
-1.00600874e+00 9.94509995e-01 8.86801600e-01 -1.27498299e-01
1.34505463e+00 1.06658176e-01 7.83904552e-01 1.91572040e-01
1.14376768e-01 -1.36920965e+00 -2.54454669e-02 6.19506896e-01
7.17041790e-01 -1.35936439e+00 -3.80296141e-01 -1.62915185e-01
-8.54002774e-01 6.78541362e-01 9.22562897e-01 1.88333429e-02
6.77604377e-01 -2.39914104e-01 1.21711925e-01 1.67053461e-01
-4.87469584e-01 -2.90476203e-01 9.08943862e-02 8.46892953e-01
-5.94052151e-02 1.92133307e-01 -2.95792380e-03 9.36727405e-01
1.83738798e-01 -1.06986210e-01 2.03204423e-01 7.78748512e-01
-4.27415550e-01 -1.20588470e+00 -4.04991448e-01 3.50712806e-01
-2.28977710e-01 1.36243790e-01 -4.65618521e-01 7.40264535e-01
4.27003533e-01 7.26319432e-01 5.89131936e-02 -3.42071950e-01
3.86941850e-01 4.66573656e-01 -1.74586587e-02 -7.01927304e-01
-2.48850152e-01 1.26814499e-01 -2.93681473e-01 3.54004349e-03
-3.67989123e-01 -8.43028188e-01 -1.40371740e+00 -1.52811660e-02
-1.42186359e-01 -1.12835929e-01 5.00654340e-01 8.37737858e-01
6.31935954e-01 3.24227422e-01 5.86987197e-01 -5.43134093e-01
-4.87095207e-01 -9.28732872e-01 -6.18635952e-01 6.92191958e-01
2.84256637e-01 -9.38289762e-01 -4.80353922e-01 1.21139050e-01]
|
[10.37541675567627, 2.9812381267547607]
|
56e06a71-9e77-49e2-a5b2-2c4a107fc497
|
ltg-at-semeval-2016-task-11-complex-word
| null | null |
https://aclanthology.org/S16-1154
|
https://aclanthology.org/S16-1154.pdf
|
LTG at SemEval-2016 Task 11: Complex Word Identification with Classifier Ensembles
| null |
['Marcos Zampieri', 'Mark Dras', 'Shervin Malmasi']
|
2016-06-01
| null | null | null |
semeval-2016-6
|
['complex-word-identification']
|
['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]
|
[-7.422660827636719, 3.613111972808838]
|
ec3b8a3d-0e85-4685-a6b7-4d38f80c8158
|
topological-sort-for-sentence-ordering
|
2005.00432
| null |
https://arxiv.org/abs/2005.00432v1
|
https://arxiv.org/pdf/2005.00432v1.pdf
|
Topological Sort for Sentence Ordering
|
Sentence ordering is the task of arranging the sentences of a given text in the correct order. Recent work using deep neural networks for this task has framed it as a sequence prediction problem. In this paper, we propose a new framing of this task as a constraint solving problem and introduce a new technique to solve it. Additionally, we propose a human evaluation for this task. The results on both automatic and human metrics across four different datasets show that this new technique is better at capturing coherence in documents.
|
['Alan W. black', 'Shrimai Prabhumoye', 'Ruslan Salakhutdinov']
|
2020-05-01
|
topological-sort-for-sentence-ordering-1
|
https://aclanthology.org/2020.acl-main.248
|
https://aclanthology.org/2020.acl-main.248.pdf
|
acl-2020-6
|
['sentence-ordering']
|
['natural-language-processing']
|
[ 2.65251160e-01 1.35909513e-01 -7.36607909e-02 -9.24709082e-01
-1.56133369e-01 -5.12894452e-01 7.16572940e-01 3.37655306e-01
-5.63631117e-01 7.49695122e-01 8.46009374e-01 -2.87456512e-01
-3.24586593e-02 -4.96838361e-01 -4.15411830e-01 -9.38833058e-02
-6.67507052e-02 6.68298841e-01 1.42482251e-01 -3.84543061e-01
7.49480128e-01 1.95544988e-01 -1.17198157e+00 7.37307549e-01
5.95338106e-01 5.66291571e-01 2.58516967e-01 8.34227979e-01
-2.19568238e-01 1.11828327e+00 -8.91384959e-01 -2.67910004e-01
-3.93900694e-03 -5.86175203e-01 -1.64053500e+00 7.07643926e-02
6.83918238e-01 -2.74329215e-01 -3.94346230e-02 9.03813303e-01
1.93805352e-01 3.12565476e-01 4.14134592e-01 -8.64009619e-01
-6.07118428e-01 8.64043355e-01 -1.13879815e-02 3.83665532e-01
9.71671462e-01 -3.95401388e-01 1.70150876e+00 -6.82776511e-01
8.82693768e-01 1.09374774e+00 6.21074200e-01 5.78606606e-01
-1.09518981e+00 -5.17719649e-02 2.52628326e-01 6.67205274e-01
-1.06134140e+00 -3.92820090e-01 6.60467684e-01 -5.26027203e-01
1.82877421e+00 6.18740678e-01 6.12841606e-01 8.67963314e-01
2.94372380e-01 9.23346877e-01 5.96485376e-01 -7.72284985e-01
2.53451914e-01 -2.34688073e-01 9.24267709e-01 4.17160004e-01
1.53389037e-01 -3.34533840e-01 -7.78790951e-01 2.11822078e-01
7.60631859e-02 -4.45439726e-01 -2.68915355e-01 2.21541926e-01
-1.23976505e+00 9.09844935e-01 2.41645813e-01 7.24448264e-01
-1.76045924e-01 -7.10677402e-03 7.10990727e-01 2.76718765e-01
6.47603333e-01 1.07565701e+00 -5.73762834e-01 -1.44970953e-01
-1.05298603e+00 5.44634819e-01 1.38593650e+00 8.31306756e-01
2.40612298e-01 -4.01606798e-01 -6.10485375e-01 5.83682477e-01
1.97943479e-01 -3.61215770e-01 5.07633984e-01 -9.68095839e-01
5.07861137e-01 5.29565573e-01 2.84214526e-01 -1.34802902e+00
-7.97306418e-01 -1.42565817e-01 -7.31778741e-01 -3.74900371e-01
1.18626073e-01 6.55581728e-02 -6.82083189e-01 1.65027666e+00
-1.52783841e-01 1.43981621e-01 -1.60890535e-01 1.08062375e+00
8.95352840e-01 8.58207405e-01 -1.67827904e-01 -5.57663202e-01
1.17637467e+00 -1.38425100e+00 -1.08354783e+00 -2.77572393e-01
6.78868592e-01 -6.35375500e-01 9.86188471e-01 6.50363266e-01
-1.15356219e+00 -5.02101779e-01 -1.30800760e+00 -3.96297157e-01
-1.90625727e-01 -2.40704678e-02 6.54418647e-01 4.52508450e-01
-1.48481214e+00 8.30897450e-01 -5.20135701e-01 -4.72615421e-01
6.81167021e-02 3.25537443e-01 -1.90216705e-01 1.50638267e-01
-1.42285395e+00 1.24054325e+00 6.63261116e-01 2.63630748e-01
-3.31079304e-01 -3.33277822e-01 -7.01947808e-01 4.99058902e-01
2.51811147e-01 -8.23846698e-01 1.61292493e+00 -8.12478781e-01
-1.51476336e+00 1.03708172e+00 -5.80087543e-01 -8.06475222e-01
2.51243204e-01 -6.18770301e-01 -3.65710378e-01 -1.62894279e-02
1.73720792e-01 4.51626509e-01 4.75950062e-01 -8.09232891e-01
-6.18847966e-01 5.83366454e-02 1.46491364e-01 1.98597461e-01
-4.10194010e-01 3.60641479e-01 -2.60679960e-01 -5.52526414e-01
3.33922287e-03 -8.54695022e-01 -2.48194501e-01 -7.04970717e-01
-7.99521327e-01 -8.46765876e-01 3.36940527e-01 -8.19608331e-01
1.69940305e+00 -1.52671945e+00 5.22927165e-01 -4.68913354e-02
4.09658760e-01 3.47554445e-01 -2.17409283e-01 7.76369274e-01
-3.33038568e-01 5.99885762e-01 -3.44596535e-01 -6.12262905e-01
1.68651879e-01 2.70030290e-01 -3.57686967e-01 1.22329794e-01
3.31547230e-01 7.98274934e-01 -8.13586533e-01 -5.40627122e-01
-1.45522431e-01 4.72788624e-02 -6.36169791e-01 3.43036652e-01
-6.19476795e-01 3.34757179e-01 8.07070360e-02 -9.37459525e-04
3.78775805e-01 -3.20843518e-01 7.90357411e-01 9.88363475e-02
-3.95135209e-02 7.04092145e-01 -8.31548631e-01 1.74770606e+00
-2.50304312e-01 1.22669971e+00 -5.23158431e-01 -1.10740566e+00
7.71783888e-01 4.02960300e-01 2.90611506e-01 -6.90810680e-01
1.36219338e-01 -1.24149226e-01 2.32868567e-01 -9.38921928e-01
9.55188692e-01 7.49425143e-02 -8.36781487e-02 6.25300825e-01
-1.13662608e-01 -1.05584107e-01 8.78122449e-01 4.62403327e-01
1.27018774e+00 -2.51442343e-01 4.87786829e-01 -3.76023412e-01
7.99217701e-01 6.02535866e-02 5.54682732e-01 8.81128550e-01
-1.67974591e-01 6.57216787e-01 7.44362652e-01 -9.63056803e-01
-1.26299834e+00 -4.16584283e-01 1.16739899e-01 9.34482396e-01
-1.65980563e-01 -9.64488387e-01 -8.68759096e-01 -5.54644346e-01
-2.21511990e-01 8.94161999e-01 -6.28912807e-01 1.06140874e-01
-9.85777736e-01 -5.28779209e-01 2.27135524e-01 5.48605442e-01
2.64669269e-01 -1.26434541e+00 -7.84103572e-01 3.19393754e-01
-7.57629752e-01 -1.44578147e+00 -6.23645306e-01 4.02886346e-02
-6.60031199e-01 -9.39864218e-01 -2.16805220e-01 -1.17156756e+00
5.01488805e-01 1.42327994e-01 1.63876677e+00 6.89393282e-01
-1.35734618e-01 -1.77181289e-01 -7.30381668e-01 -1.73193276e-01
-3.29467148e-01 2.48507619e-01 -1.62101593e-02 -2.46764839e-01
7.42918074e-01 -2.14195669e-01 -2.73277313e-01 -2.33544469e-01
-9.34576690e-01 2.84905851e-01 6.17812462e-02 9.31583345e-01
5.99959046e-02 1.05660856e-01 2.07375914e-01 -1.02717578e+00
1.54620028e+00 -2.09916696e-01 -2.96515971e-01 3.84529233e-01
-6.43478930e-01 1.02945775e-01 5.94833553e-01 6.87607378e-02
-6.87273383e-01 9.64104533e-02 -1.77772790e-01 3.88331681e-01
-1.85892642e-01 1.09624076e+00 2.14725956e-01 5.55423677e-01
4.97697800e-01 5.73751144e-02 -3.71266037e-01 -3.79539102e-01
8.56046900e-02 4.46740329e-01 3.34836066e-01 -3.04427564e-01
2.10367590e-01 1.13740690e-01 6.68412307e-03 -7.38672018e-01
-1.39630544e+00 -5.21227896e-01 -1.07479727e+00 -5.94999641e-02
9.89545345e-01 -4.16642874e-01 -8.70921135e-01 1.28610596e-01
-1.98560822e+00 -2.69112498e-01 1.32680312e-01 1.52201861e-01
-4.48576719e-01 6.65366948e-01 -5.99767089e-01 -4.95600969e-01
-4.92162526e-01 -9.51860845e-01 9.13735211e-01 7.37740099e-02
-9.61310506e-01 -1.00145423e+00 4.49302524e-01 2.44323581e-01
2.93600202e-01 1.04719259e-01 7.97147274e-01 -9.91623402e-01
-3.19838852e-01 -2.05645755e-01 -1.43919513e-01 2.51533329e-01
-1.51440352e-01 2.17333287e-01 -5.71366251e-01 -6.89726248e-02
2.26052940e-01 -1.03232943e-01 9.95121896e-01 4.50871408e-01
1.27088475e+00 -4.78500605e-01 -8.86163339e-02 3.06773305e-01
1.35579908e+00 2.47820288e-01 6.69154167e-01 5.60155094e-01
4.03920591e-01 8.07730258e-01 5.53815484e-01 4.94986892e-01
4.23717767e-01 8.58351290e-01 2.08418384e-01 1.64633468e-01
1.05740227e-01 1.93430737e-01 5.44922948e-02 1.09151244e+00
2.16426358e-01 -8.04510057e-01 -1.30174792e+00 4.74824488e-01
-2.33774567e+00 -1.27067280e+00 -5.01624525e-01 1.44939089e+00
1.00670338e+00 3.86174530e-01 -7.47928843e-02 3.20210993e-01
6.08901620e-01 1.72219008e-01 3.96973863e-02 -1.06310105e+00
-1.22427061e-01 2.28423566e-01 -1.47965863e-01 1.04610276e+00
-1.28494084e+00 1.13084018e+00 7.48943949e+00 2.12665841e-01
-9.47954357e-01 -2.90822566e-01 6.08272314e-01 -3.69482189e-02
-2.10584134e-01 3.36776078e-02 -8.87971103e-01 4.26879346e-01
1.04535246e+00 -2.40436360e-01 3.08605641e-01 4.34015900e-01
5.98321438e-01 -4.62683737e-01 -1.69724679e+00 6.39672697e-01
4.13476169e-01 -1.76842487e+00 6.15994260e-02 -4.87073213e-01
6.30976379e-01 -2.55325228e-01 -4.23966348e-01 6.29221573e-02
1.35374963e-01 -1.38238800e+00 7.35383332e-01 6.63278162e-01
1.41942143e-01 -4.95748401e-01 1.03438962e+00 3.21282178e-01
-7.60689080e-01 1.15671232e-01 -1.56071410e-01 -7.95279920e-01
4.15355384e-01 7.09137261e-01 -1.36644912e+00 4.94364470e-01
6.31494880e-01 9.78522182e-01 -7.55458534e-01 1.06252646e+00
-4.80974823e-01 4.85036373e-01 1.40303865e-01 -4.95712519e-01
2.79536903e-01 -7.67565379e-03 5.52889705e-01 1.67384613e+00
-1.79147288e-01 3.21171656e-02 2.67354280e-01 6.99680090e-01
5.76219149e-02 -6.42665252e-02 -4.68672395e-01 -8.08333531e-02
4.09714341e-01 1.03418362e+00 -7.90885031e-01 -4.15661305e-01
-1.91148847e-01 1.17664683e+00 6.99097574e-01 2.27223396e-01
-6.48174226e-01 -4.50390160e-01 4.23376501e-01 -4.36360866e-01
2.38820508e-01 -5.98149776e-01 -8.41066480e-01 -1.13149214e+00
2.75104344e-01 -7.66271234e-01 2.31042072e-01 -8.26208889e-01
-1.07755423e+00 9.10443187e-01 -1.25712827e-01 -1.02267909e+00
-3.93702120e-01 -7.12022245e-01 -7.33103752e-01 8.17165554e-01
-1.29769480e+00 -6.56716526e-01 -2.66662329e-01 3.03711910e-02
8.20458949e-01 -1.55694082e-01 9.80717957e-01 2.00663105e-01
-7.23801136e-01 2.95216233e-01 -6.23490028e-02 8.37826803e-02
4.68952447e-01 -1.53608406e+00 7.42801070e-01 1.05595243e+00
3.50913733e-01 9.42866385e-01 1.24970996e+00 -4.47837591e-01
-8.56153011e-01 -5.53803563e-01 2.06812429e+00 -5.39450765e-01
5.19462943e-01 -3.59776020e-01 -7.98055470e-01 6.55975163e-01
9.44272757e-01 -8.09328556e-01 7.43926227e-01 4.55446690e-01
-1.74259484e-01 2.79386431e-01 -6.51643097e-01 6.03560925e-01
1.01602781e+00 -4.89107311e-01 -9.32655156e-01 8.62358689e-01
1.29820716e+00 -5.15542626e-01 -5.28092563e-01 1.10295653e-01
3.47621828e-01 -1.13637531e+00 5.94582260e-01 -1.08447230e+00
1.16721892e+00 -2.29032516e-01 -7.27311373e-02 -1.44755864e+00
-5.82573235e-01 -5.43867171e-01 -1.80511206e-01 8.41747224e-01
4.87259150e-01 -2.29965206e-02 8.96698594e-01 6.42283440e-01
-4.56323534e-01 -7.68937707e-01 -8.56057167e-01 -7.20189273e-01
2.81663369e-02 -2.35471800e-01 4.68394995e-01 1.20869923e+00
4.95789915e-01 9.02821422e-01 -4.67169136e-01 -1.76069707e-01
-7.32699037e-02 8.65696892e-02 3.53925079e-01 -1.29460299e+00
-7.16609433e-02 -6.23469770e-01 -1.16380557e-01 -9.56429303e-01
6.06823027e-01 -1.01451933e+00 3.27100158e-01 -1.94736254e+00
2.89857447e-01 3.31419557e-01 1.41053393e-01 2.08093688e-01
-1.31282300e-01 -3.31765354e-01 2.89065003e-01 1.06383994e-01
-8.08694303e-01 3.40346158e-01 9.69135523e-01 -2.97092915e-01
-7.54457340e-02 -2.28622392e-01 -5.72331071e-01 3.61623496e-01
1.03454232e+00 -4.62916285e-01 -1.79206207e-01 -1.08702493e+00
7.54569054e-01 -2.14329213e-02 3.20949294e-02 -1.01617038e+00
6.30056620e-01 -1.63472936e-01 7.22480193e-02 -8.85904849e-01
5.28479517e-02 -7.24445581e-01 -1.34060442e-01 5.51677644e-01
-8.61239254e-01 4.31347191e-01 1.65097475e-01 3.38322431e-01
-5.28106868e-01 -6.97377145e-01 3.73701125e-01 -1.67628527e-01
-9.50917184e-01 -2.28925839e-01 -5.69102347e-01 -3.51486611e-03
8.49642158e-01 4.66647521e-02 -3.68168414e-01 -5.81673324e-01
-9.07132864e-01 4.44774449e-01 -5.92762837e-03 6.73334301e-01
6.79177225e-01 -9.74285364e-01 -8.42755020e-01 -1.57253832e-01
-1.60797998e-01 -1.16239689e-01 -1.85429066e-01 5.92250943e-01
-9.34571266e-01 1.16555572e+00 -1.01883180e-01 -5.26045561e-01
-1.58296549e+00 3.79201472e-01 3.58449042e-01 -7.04194069e-01
-3.45245451e-01 1.10328674e+00 -3.74122888e-01 -3.77139002e-01
3.90923679e-01 -3.99201065e-01 -7.23214567e-01 1.25672996e-01
6.24083281e-01 5.15714474e-02 3.38976443e-01 -4.10203665e-01
-4.76238370e-01 2.51149863e-01 -4.44118202e-01 -6.18405528e-02
1.43778634e+00 -2.40619048e-01 -7.65642285e-01 3.69463980e-01
1.21175838e+00 -3.99328262e-01 -4.12645966e-01 -1.36834413e-01
9.29431319e-01 -2.94078261e-01 -2.44448677e-01 -7.65429854e-01
-4.15468633e-01 7.55188167e-01 -8.86549335e-03 9.14786696e-01
8.74401271e-01 -2.52633810e-01 6.30006731e-01 7.30414152e-01
-1.25245139e-01 -1.49849510e+00 4.12457794e-01 1.53469491e+00
1.17399263e+00 -1.38981783e+00 2.33883202e-01 -4.63294148e-01
-6.09967768e-01 1.52740383e+00 7.58010924e-01 -1.56685188e-01
4.85728323e-01 -1.19254760e-01 -1.10708334e-01 -3.69458169e-01
-1.11817920e+00 -1.16900951e-02 3.51717710e-01 2.01365009e-01
1.12436044e+00 1.96074005e-02 -1.02169609e+00 4.69486535e-01
-5.37990630e-01 1.11941300e-01 9.26929414e-01 1.02971685e+00
-5.55306137e-01 -1.30369532e+00 -1.19165191e-02 4.16438341e-01
-2.47372046e-01 -3.97264481e-01 -9.89561498e-01 3.37560326e-01
-2.24042937e-01 1.24920225e+00 3.54174048e-01 -5.88992119e-01
1.97915763e-01 1.61684766e-01 2.99587339e-01 -1.08644152e+00
-8.54350090e-01 -3.56197834e-01 7.07111299e-01 -5.55253208e-01
-5.99573970e-01 -5.98427594e-01 -1.21360910e+00 -3.26299787e-01
-4.37700897e-02 1.65104374e-01 3.15173358e-01 1.21009994e+00
2.42334098e-01 7.99491405e-01 6.09164834e-01 -6.83622003e-01
-5.40121317e-01 -1.06843448e+00 -1.16036378e-01 6.03288114e-01
3.61556709e-01 -7.96184316e-02 -1.58216462e-01 2.14857295e-01]
|
[11.885749816894531, 9.28035831451416]
|
8270397b-3f5c-49a6-bc51-50839d1ee172
|
deep-learning-assessment-of-tumor
|
1610.03467
| null |
http://arxiv.org/abs/1610.03467v1
|
http://arxiv.org/pdf/1610.03467v1.pdf
|
Deep Learning Assessment of Tumor Proliferation in Breast Cancer Histological Images
|
Current analysis of tumor proliferation, the most salient prognostic
biomarker for invasive breast cancer, is limited to subjective mitosis counting
by pathologists in localized regions of tissue images. This study presents the
first data-driven integrative approach to characterize the severity of tumor
growth and spread on a categorical and molecular level, utilizing multiple
biologically salient deep learning classifiers to develop a comprehensive
prognostic model. Our approach achieves pathologist-level performance on
three-class categorical tumor severity prediction. It additionally pioneers
prediction of molecular expression data from a tissue image, obtaining a
Spearman's rank correlation coefficient of 0.60 with ex vivo mean calculated
RNA expression. Furthermore, our framework is applied to identify over two
hundred unprecedented biomarkers critical to the accurate assessment of tumor
proliferation, validating our proposed integrative pipeline as the first to
holistically and objectively analyze histopathological images.
|
['Manan Shah', 'Dayong Wang', 'Christopher Rubadue', 'David Suster']
|
2016-10-11
| null | null | null | null |
['severity-prediction']
|
['computer-vision']
|
[ 4.31461781e-01 5.82666732e-02 -9.82459247e-01 -2.05076654e-02
-1.39590383e+00 -4.62189496e-01 2.96984315e-01 1.06829035e+00
-5.16189754e-01 8.15599978e-01 2.09033534e-01 -5.21692872e-01
-2.92785406e-01 -6.40955687e-01 -1.34083688e-01 -1.28500819e+00
-2.30139732e-01 6.10998809e-01 -2.96531349e-01 6.13153130e-02
1.58276677e-01 7.12198555e-01 -8.89066935e-01 2.79571593e-01
6.70437157e-01 1.14054966e+00 -2.38579914e-01 9.50353742e-01
1.44275114e-01 5.56033254e-01 -1.43710032e-01 -2.35175714e-02
-2.72374213e-01 -4.32168916e-02 -6.07217729e-01 -2.66134441e-01
3.98390353e-01 -2.85591274e-01 -8.22362974e-02 7.75518715e-01
4.96121526e-01 -8.91376257e-01 1.10285914e+00 -9.60968494e-01
-3.32127005e-01 5.04220724e-01 -7.10457802e-01 4.18520868e-01
2.88819261e-02 3.93253118e-01 1.21844387e+00 -6.39139354e-01
8.62906635e-01 4.50291842e-01 7.51527250e-01 2.85776228e-01
-1.65080357e+00 -4.95383620e-01 -6.10981822e-01 5.82300052e-02
-1.36183691e+00 -3.27180296e-01 3.93664062e-01 -6.55437469e-01
5.46483278e-01 4.84607041e-01 1.03696060e+00 9.83452082e-01
8.87850463e-01 5.39150119e-01 1.56950915e+00 -8.25350210e-02
2.65979052e-01 -2.86460608e-01 3.21381569e-01 9.32763040e-01
6.20716274e-01 -1.17723577e-01 -5.27684093e-01 -1.88903406e-01
3.99823725e-01 1.07361466e-01 -2.11082593e-01 9.60380957e-02
-1.34877920e+00 4.47104722e-01 3.40753287e-01 5.34666419e-01
-9.34136361e-02 4.34333563e-01 6.24674618e-01 -1.27913773e-01
3.52371186e-01 2.56310344e-01 -5.84656000e-01 3.82951535e-02
-1.21700478e+00 -9.45675671e-02 4.79216993e-01 1.35970309e-01
5.00255525e-01 -5.57036102e-01 -2.72061020e-01 3.41805905e-01
4.14651424e-01 6.04546666e-01 7.69703388e-01 -6.55203283e-01
-5.58109105e-01 9.84611392e-01 -1.17086068e-01 -8.70042741e-01
-1.14449239e+00 -8.06907773e-01 -1.06011450e+00 1.53389871e-01
6.81594372e-01 4.34114605e-01 -6.22410238e-01 1.36374867e+00
9.52773318e-02 -1.72906548e-01 -1.44739270e-01 5.72466850e-01
8.75694335e-01 -6.58156201e-02 6.06833160e-01 -5.54034829e-01
1.74966264e+00 -5.31903744e-01 -5.22014797e-01 1.11054704e-01
1.21662509e+00 -1.09831132e-01 8.20866942e-01 3.41867536e-01
-7.16302216e-01 1.18591838e-01 -9.53636050e-01 -3.18501085e-01
-4.65654969e-01 5.11790395e-01 8.70189369e-01 3.67297262e-01
-1.09005737e+00 4.65032071e-01 -9.64954972e-01 -6.75740957e-01
1.07590604e+00 4.08482224e-01 -7.77725339e-01 9.10185650e-02
-6.23388410e-01 8.10531139e-01 9.93717015e-02 -9.69572589e-02
-1.15410733e+00 -1.32678854e+00 -3.93481821e-01 1.40316859e-01
-5.31934619e-01 -1.08475637e+00 9.83100474e-01 -4.48713899e-01
-1.18992674e+00 1.53038669e+00 -6.07966959e-01 -3.48743379e-01
3.33319664e-01 5.98834753e-01 -5.11854142e-02 3.96077096e-01
1.05229363e-01 6.12926304e-01 9.08760726e-02 -9.78679478e-01
-7.23226428e-01 -7.88411856e-01 -6.23537719e-01 -1.29533067e-01
-6.33442700e-01 -4.76458520e-01 -7.76636153e-02 -3.83018821e-01
1.88504264e-01 -6.81497991e-01 -4.26638484e-01 4.41686898e-01
-2.41525799e-01 1.33512944e-01 4.81725544e-01 -6.07192814e-01
1.16563940e+00 -2.05198741e+00 1.92306310e-01 1.73654348e-01
7.43590295e-01 -2.29689360e-01 1.11443751e-01 1.09444290e-01
-2.15800181e-02 7.12858737e-01 5.22598531e-03 -2.49807432e-01
-1.39509484e-01 -3.53600651e-01 2.49973729e-01 1.03379059e+00
1.77983150e-01 1.47842348e+00 -9.84680176e-01 -8.31608653e-01
-1.98496990e-02 2.35645175e-01 -1.86434284e-01 -2.11171493e-01
-5.96130751e-02 4.40971613e-01 -2.45375812e-01 1.35483205e+00
4.43481266e-01 -6.99657261e-01 5.96432686e-01 -3.28076243e-01
3.37061435e-01 -1.27024129e-01 8.11894834e-02 1.32454562e+00
-1.68297470e-01 7.47128904e-01 1.68143079e-01 -6.13192141e-01
6.55686319e-01 8.86130407e-02 8.81211877e-01 -5.92042267e-01
5.91981947e-01 3.23533297e-01 -4.35079969e-02 -3.25645238e-01
-1.55803204e-01 -4.14776683e-01 -1.61969706e-01 6.11768179e-02
-5.41537330e-02 1.19367041e-01 2.52304643e-01 -4.34749275e-02
1.66024029e+00 -5.81538022e-01 7.95551062e-01 -7.13403523e-01
6.32402897e-01 4.15451974e-01 4.98641223e-01 3.50369126e-01
-8.01443875e-01 3.31304461e-01 1.06016791e+00 -4.01765615e-01
-8.88098180e-01 -1.12923825e+00 -5.23808897e-01 9.57622766e-01
1.54117085e-02 -1.25142664e-01 -1.90342963e-01 -4.89675283e-01
3.97085488e-01 -3.76742110e-02 -1.26574612e+00 -1.80753335e-01
-4.45088446e-02 -1.30196595e+00 1.02054989e+00 4.48204219e-01
5.52520677e-02 -2.10086986e-01 -6.46520182e-02 2.74412129e-02
3.74632818e-03 -7.53805101e-01 8.45179334e-02 6.04278684e-01
-1.01998723e+00 -1.46844399e+00 -4.66477066e-01 -8.10715735e-01
1.06366026e+00 -1.22002892e-01 8.01386476e-01 2.42264405e-01
-9.17898476e-01 -5.14559485e-02 9.70397890e-03 -3.72730464e-01
-6.79960668e-01 2.54063368e-01 -2.07253918e-01 -2.76476085e-01
4.48673815e-01 -6.22119427e-01 -9.94405806e-01 4.10853740e-04
-5.99173009e-01 1.02305785e-01 1.11345959e+00 9.03593600e-01
1.10403872e+00 -1.08757153e-01 5.81925154e-01 -6.87123537e-01
2.52991974e-01 -6.07094169e-01 -3.08345169e-01 6.36505410e-02
-7.43061781e-01 -3.06695640e-01 5.03402829e-01 -6.56455010e-02
-6.59208655e-01 9.45417285e-02 -4.94579151e-02 1.99036807e-01
-2.34206378e-01 9.19383228e-01 2.71768361e-01 -2.98125893e-01
7.41781890e-01 2.39871323e-01 3.30648690e-01 2.11939186e-01
-1.39289975e-01 3.34375888e-01 6.79321945e-01 -1.37942746e-01
4.10184115e-01 8.75202119e-01 9.64006424e-01 -5.82245350e-01
-8.09921145e-01 -8.09836924e-01 -6.29805446e-01 -3.34248811e-01
7.39802897e-01 -1.01306117e+00 -1.40062523e+00 5.60873091e-01
-5.33191562e-01 -4.80877906e-01 7.65699521e-02 2.84772217e-01
-6.13181949e-01 1.51112422e-01 -1.10419071e+00 -2.71403283e-01
-5.48332274e-01 -7.68522918e-01 1.53090203e+00 7.44727477e-02
-4.47210610e-01 -1.11781991e+00 4.65855658e-01 6.23147726e-01
2.01047018e-01 7.65651047e-01 1.38548398e+00 -5.15520573e-01
-4.23953086e-01 -5.48590839e-01 -3.50864440e-01 -4.64630127e-01
1.89425126e-01 6.67588949e-01 -1.02324235e+00 -2.01261237e-01
-7.77876914e-01 -2.49089137e-01 1.07650197e+00 5.73844433e-01
1.18896067e+00 -1.08105287e-01 -1.10188389e+00 7.99811542e-01
1.70140064e+00 -3.30001086e-01 4.63132560e-01 5.03931522e-01
1.57758623e-01 2.18809113e-01 5.59776425e-01 3.12735677e-01
2.49180824e-01 4.95338328e-02 4.92204368e-01 -4.15918767e-01
-1.80675238e-01 -4.09187339e-02 -1.99312661e-02 3.98703754e-01
1.11562774e-01 -3.16186219e-01 -1.38782644e+00 6.16097093e-01
-1.39233220e+00 -7.47085094e-01 -2.91280627e-01 1.78792214e+00
9.07055318e-01 1.13796256e-01 -1.60838947e-01 1.93045467e-01
3.79999429e-01 -1.69077829e-01 -5.56134045e-01 -3.90044414e-02
-3.04869175e-01 -1.51958719e-01 8.08407187e-01 2.80591547e-01
-1.08500719e+00 5.65061867e-01 7.74121904e+00 8.58345091e-01
-1.34834933e+00 2.17536669e-02 1.33852255e+00 -3.74675393e-02
-1.59532696e-01 -1.96091086e-01 -6.20979965e-01 2.33930096e-01
9.54334378e-01 -6.44691169e-01 -2.55380511e-01 6.42946362e-01
5.84665835e-01 -3.80559146e-01 -1.27478600e+00 5.72826743e-01
-2.64297575e-01 -1.77099574e+00 -7.71827623e-02 5.78257501e-01
7.13689864e-01 9.63972658e-02 3.09555203e-01 -1.35599390e-01
3.17318887e-01 -1.38208222e+00 -1.06811590e-01 8.10363889e-01
1.18837273e+00 -3.68539810e-01 1.03171694e+00 3.33423793e-01
-8.28864992e-01 -2.38396555e-01 3.51154767e-02 1.96518078e-01
-3.24785620e-01 1.02578366e+00 -1.38087523e+00 1.14295982e-01
2.69653648e-01 8.75734508e-01 -8.87463391e-01 8.16553175e-01
2.42018551e-01 6.58337891e-01 -1.69660941e-01 -7.12886527e-02
-1.99131176e-01 4.64862257e-01 2.13785499e-01 1.47508562e+00
3.60822588e-01 1.64198503e-02 -1.06423244e-01 4.93181735e-01
-1.18508965e-01 2.68677056e-01 -1.33824036e-01 -4.86564904e-01
3.16948771e-01 1.90335906e+00 -1.08098102e+00 -1.09431691e-01
9.10082087e-02 3.92635405e-01 3.93530279e-01 -4.53672111e-02
-7.42660165e-01 2.73402542e-01 6.08138382e-01 3.35152805e-01
-4.37228262e-01 -9.20032188e-02 -9.92054582e-01 -6.22116983e-01
-6.21260464e-01 -3.70547712e-01 4.35336620e-01 -4.63312626e-01
-1.32326043e+00 -3.61474082e-02 -6.43286705e-01 -1.13052833e+00
4.28848624e-01 -9.10493016e-01 -6.36272371e-01 4.98125464e-01
-1.74649501e+00 -1.42925251e+00 -7.42837548e-01 -6.69042245e-02
-1.99892372e-02 1.28545880e-01 1.04781497e+00 1.05032668e-05
-8.11162174e-01 5.59594989e-01 3.70709777e-01 -1.29789278e-01
7.00324118e-01 -1.43245316e+00 -4.57682401e-01 5.46231754e-02
-7.94572473e-01 5.00759840e-01 7.29794145e-01 -5.37792504e-01
-1.52262354e+00 -1.21571994e+00 7.23541677e-01 -4.84537363e-01
1.06032264e+00 -3.73032689e-02 -3.48808646e-01 4.09171462e-01
-3.93943250e-01 3.27372134e-01 1.48025990e+00 1.17224023e-01
-2.87433535e-01 -4.30944383e-01 -1.19853401e+00 4.91347939e-01
5.88531971e-01 -4.35759485e-01 1.58554718e-01 4.35389161e-01
1.24052890e-01 -2.30546832e-01 -1.64481342e+00 7.57993698e-01
9.04491127e-01 -7.62329042e-01 7.49155939e-01 -3.41231346e-01
7.57930279e-01 -3.18300515e-01 1.43553372e-02 -9.09404457e-01
-7.77422965e-01 1.08430818e-01 7.09018186e-02 8.28681290e-01
6.95613801e-01 -4.78988022e-01 1.22045124e+00 -7.88442791e-02
3.88900116e-02 -1.35903692e+00 -1.12671268e+00 -2.77406156e-01
3.09162289e-01 5.35350256e-02 5.36945537e-02 6.88733697e-01
6.48570359e-01 -1.16992913e-01 4.35224831e-01 2.11948022e-01
5.58878183e-01 -1.69719681e-01 5.81862688e-01 -1.31155837e+00
5.42434193e-02 -1.05976379e+00 -1.08773828e+00 -1.50438145e-01
1.77414581e-01 -1.22999346e+00 -2.24580452e-01 -1.42401493e+00
9.34060991e-01 -2.75872827e-01 -7.20121741e-01 4.50572073e-01
-9.49137956e-02 7.44279206e-01 -3.81787002e-01 3.43519002e-01
-6.34568930e-01 9.59248245e-02 1.25949264e+00 -6.70393348e-01
3.55814695e-01 -5.52002072e-01 -1.08587170e+00 6.52716994e-01
7.89111316e-01 -3.02256107e-01 8.37976858e-02 2.23954007e-01
5.12676775e-01 1.05332389e-01 5.48916280e-01 -1.14838850e+00
2.40658358e-01 -5.14550865e-01 9.87261653e-01 -5.61929286e-01
8.31149891e-02 -7.19168544e-01 1.82952374e-01 9.22278285e-01
-4.59943205e-01 -5.51585376e-01 3.75479341e-01 6.76828146e-01
-8.14852938e-02 4.20540184e-01 8.97650242e-01 1.03455164e-01
-2.76863337e-01 4.82509851e-01 -6.51565075e-01 -5.43239594e-01
1.38955009e+00 -5.57743609e-01 -9.98338521e-01 1.91819400e-01
-9.10637796e-01 2.45577425e-01 7.61440039e-01 -2.88657367e-01
1.48727670e-01 -1.06291044e+00 -7.57501841e-01 -2.51203090e-01
5.74100018e-01 -2.44909272e-01 5.46707153e-01 1.50007808e+00
-8.81286740e-01 6.48612022e-01 -2.26110622e-01 -9.79654014e-01
-1.41581118e+00 4.74147826e-01 5.86137414e-01 -9.02703583e-01
-8.21801051e-02 8.14835072e-01 2.71076679e-01 4.16732952e-02
-1.07795902e-01 -2.42089227e-01 -2.47501165e-01 1.65318012e-01
4.06054884e-01 3.88246179e-01 1.63761646e-01 -4.81407613e-01
-4.54411775e-01 4.11753207e-01 -1.86938480e-01 3.90026629e-01
1.21146941e+00 -5.32975346e-02 -6.73771501e-01 6.13508761e-01
1.44127417e+00 -4.50051911e-02 -7.93563664e-01 2.97335654e-01
-2.19432358e-02 -1.37452528e-01 1.72901854e-01 -1.24223363e+00
-6.97832763e-01 4.48943496e-01 7.08973587e-01 -2.23926023e-01
1.23559093e+00 1.28422365e-01 5.34127057e-01 2.06007227e-01
1.85705110e-01 -7.75311470e-01 2.17273855e-03 -2.69571096e-02
5.21354914e-01 -1.28171015e+00 3.25561762e-01 -5.92498064e-01
9.04187262e-02 1.19416130e+00 4.63408291e-01 1.97998419e-01
6.15951061e-01 6.89813197e-01 1.85926631e-01 -3.32451373e-01
-1.22647166e+00 1.12168342e-01 5.72504418e-04 5.19997120e-01
7.92377591e-01 4.13251996e-01 -7.29216874e-01 9.64439690e-01
-1.19907953e-01 3.78348589e-01 5.04850030e-01 5.81109822e-01
-6.95214629e-01 -7.03420520e-01 5.41528724e-02 7.10622430e-01
-6.41469955e-01 5.60639948e-02 -5.96448839e-01 7.31281281e-01
-1.25703543e-01 3.16794813e-01 1.25308424e-01 -7.66770691e-02
-3.32520038e-01 -5.33680506e-02 2.59892553e-01 -3.37329298e-01
-3.87808710e-01 6.00825883e-02 -1.21380448e-01 -2.56274343e-01
-2.47069240e-01 -6.61117375e-01 -1.43621790e+00 -3.35078746e-01
2.78538000e-02 -3.65603626e-01 7.74597526e-01 7.76948690e-01
3.45321894e-01 7.01839745e-01 5.05487561e-01 -4.62853789e-01
-6.37813658e-02 -6.57910347e-01 -8.56079459e-01 -2.84916591e-02
5.26988029e-01 -3.58470172e-01 -4.81769681e-01 2.77885884e-01]
|
[15.138033866882324, -3.0931997299194336]
|
b5dbf310-ed68-480b-81e6-91943b1326a2
|
reformulation-of-matching-equation-in
|
2112.08742
| null |
https://arxiv.org/abs/2112.08742v1
|
https://arxiv.org/pdf/2112.08742v1.pdf
|
Reformulation of Matching Equation in Potential Energy Shaping
|
Stabilization of an underactuated mechanical system may be accomplished by energy shaping. Interconnection and damping assignment passivity-based control is an approach based on total energy shaping by assigning desired kinetic and potential energy to the system. This method requires solving a partial differential equation (PDE) related to he potential energy shaping of the system. In this short paper, we focus on the reformulation of this PDE to be solved easier. For this purpose, under a certain condition that depends on the physical parameters and the controller gains, it is possible to merely solve the homogeneous part of potential energy PDE. Furthermore, it is shown that the condition may be reduced into a linear matrix inequality form. The results are applied to a number of benchmark systems.
|
['Hamid D. Taghirad', 'M. Reza J. Harandi']
|
2021-12-16
| null | null | null | null |
['total-energy']
|
['miscellaneous']
|
[ 7.61292577e-02 8.48541737e-01 -1.57685652e-01 5.92936039e-01
-1.55320108e-01 -6.62128150e-01 3.09927016e-01 1.13743916e-01
-3.19102913e-01 1.06007826e+00 -5.08248687e-01 -1.05673626e-01
-2.96546251e-01 -4.23133194e-01 -5.17061532e-01 -9.34551716e-01
2.50407368e-01 1.37585536e-01 -1.11669816e-01 -6.56100810e-01
-1.40876383e-01 7.39876926e-01 -1.00394773e+00 -6.71806574e-01
8.44034910e-01 9.03698444e-01 1.01629734e-01 4.67482269e-01
6.34029865e-01 4.39864069e-01 -3.60527515e-01 4.47971255e-01
2.89755791e-01 -5.21720231e-01 -8.39133739e-01 5.07877052e-01
-4.08510298e-01 -1.38811767e-01 -5.73455803e-02 1.24028087e+00
3.08966875e-01 7.04785109e-01 8.52967262e-01 -1.18803346e+00
-3.00702397e-02 1.90022573e-01 -1.53005555e-01 -1.40212029e-01
1.21569574e-01 -5.95614165e-02 6.21501446e-01 -8.20440590e-01
5.51419497e-01 9.43396926e-01 3.75125498e-01 5.46885908e-01
-1.57798326e+00 -5.50453812e-02 -7.65338242e-02 -1.41860113e-01
-1.50619066e+00 -6.73468485e-02 7.74341702e-01 -5.52047074e-01
6.27913952e-01 5.16859710e-01 6.71680689e-01 3.11565071e-01
4.83300447e-01 1.96292996e-01 8.39783728e-01 -4.04661685e-01
3.99354428e-01 3.47187132e-01 2.53941298e-01 6.07597530e-01
6.33292377e-01 1.52748317e-01 3.00603062e-01 -4.30345416e-01
7.06887066e-01 -5.33748567e-01 -6.53226972e-01 -5.78631759e-01
-7.17952669e-01 9.70457554e-01 4.82514679e-01 3.52433562e-01
-4.25645232e-01 -1.76708940e-02 1.85674891e-01 1.33049160e-01
3.42695802e-01 9.92244601e-01 -6.89317882e-02 3.35595876e-01
-1.97497666e-01 3.56109470e-01 1.06951320e+00 5.16756654e-01
4.67949897e-01 5.07674456e-01 3.54094207e-01 3.29909980e-01
1.42432913e-01 6.20231509e-01 3.89132947e-02 -8.89748812e-01
-3.38836350e-02 6.17146134e-01 6.41108274e-01 -8.17186058e-01
-4.15232956e-01 -2.23096997e-01 -8.79329681e-01 4.75247234e-01
4.05377328e-01 -8.61402214e-01 -4.49827224e-01 1.76647723e+00
5.53500772e-01 -3.56264502e-01 -1.58377830e-02 1.22055387e+00
1.58256710e-01 1.07458270e+00 -3.31743389e-01 -7.49536693e-01
9.80256975e-01 -2.01000914e-01 -1.15119803e+00 9.44969282e-02
2.98801512e-01 -4.39842701e-01 5.59674442e-01 3.27609837e-01
-1.31890512e+00 -2.89765298e-01 -1.31247711e+00 3.58702511e-01
-2.39132687e-01 3.81471395e-01 -3.05913448e-01 1.10494688e-01
-9.68577564e-01 8.11894238e-01 -9.02741730e-01 -1.61007598e-01
-7.87992179e-01 7.07693517e-01 -8.23069438e-02 1.17948222e+00
-1.33986008e+00 1.30426478e+00 5.64346910e-01 5.27212262e-01
-2.73749322e-01 -6.98507845e-01 -5.90968132e-01 6.34762943e-02
6.02221370e-01 -6.95331037e-01 8.82784069e-01 -6.61496818e-01
-2.03064370e+00 2.64055848e-01 1.78067595e-01 -1.90308988e-01
6.29404545e-01 -4.26357388e-02 7.23833963e-02 3.36792260e-01
-2.86774755e-01 -3.35175455e-01 8.27683926e-01 -1.22805762e+00
9.07961130e-02 9.77161676e-02 1.59902871e-01 1.64636478e-01
-2.72340775e-01 -2.19866440e-01 2.34371260e-01 -4.39470053e-01
2.00074427e-02 -1.30741739e+00 -4.20845509e-01 -7.94036910e-02
-5.34668505e-01 -9.79638621e-02 8.08690310e-01 -5.22549391e-01
1.16833699e+00 -2.04384780e+00 9.42860842e-01 5.20804763e-01
2.02330705e-02 3.44412744e-01 5.17177224e-01 7.27917671e-01
-2.10359678e-01 -1.66244376e-02 -2.61618435e-01 3.16274971e-01
1.54626416e-02 -1.00791864e-01 -3.45728189e-01 7.42279351e-01
3.64688963e-01 2.22091004e-01 -4.67756093e-01 -2.73816645e-01
3.33304673e-01 5.09244621e-01 -4.74616319e-01 2.48165905e-01
-1.16264872e-01 4.22815919e-01 -9.09801245e-01 -1.82745144e-01
4.09559280e-01 1.08975507e-01 3.07663828e-01 -3.78096610e-01
-6.80488110e-01 -1.94826826e-01 -1.59400201e+00 7.78878331e-01
-5.22048891e-01 2.50091761e-01 1.11935246e+00 -1.45800364e+00
8.59649420e-01 6.41677380e-01 7.37717271e-01 2.54294515e-01
8.72824192e-01 2.62797982e-01 -6.92870989e-02 -3.41646433e-01
2.91038007e-01 -3.63284409e-01 9.81906205e-02 5.86527400e-02
-2.78347522e-01 -6.84224844e-01 2.51404107e-01 -1.29298732e-01
7.33806312e-01 -2.46892974e-01 3.94298851e-01 -1.00352013e+00
1.14827812e+00 4.34858382e-01 4.29639667e-01 -6.38819337e-02
2.47832779e-02 -2.76948541e-01 6.49201035e-01 2.49404773e-01
-1.17089856e+00 -5.64678609e-01 -4.24301922e-01 1.77753299e-01
1.76778495e-01 1.69033289e-01 -8.92043531e-01 4.25741076e-01
8.87699574e-02 4.49074149e-01 -5.55740237e-01 -5.37287056e-01
-8.28921854e-01 -7.27300823e-01 -1.58614889e-01 1.18935080e-02
3.58537138e-01 -1.96463808e-01 -7.82280445e-01 3.59210074e-01
-1.20273851e-01 -8.16411674e-01 -2.90877819e-01 2.03163028e-01
-7.65825450e-01 -9.94949043e-01 -6.89547777e-01 -7.72438526e-01
7.33489573e-01 -3.04455936e-01 3.63321424e-01 -1.67449027e-01
-2.34663516e-01 4.27997291e-01 1.77815780e-02 -2.11085722e-01
-5.86666524e-01 5.67284673e-02 4.40802187e-01 8.72723982e-02
-7.67001629e-01 -1.03722177e-01 -2.86612898e-01 5.79188824e-01
-7.71918774e-01 -3.03287208e-01 1.39006674e-01 8.66814256e-01
7.79841304e-01 3.21476847e-01 7.30566502e-01 -2.68785387e-01
8.06900620e-01 -2.65421961e-02 -1.18816268e+00 8.97518098e-02
-4.08965498e-01 1.31606236e-01 1.05097568e+00 -5.83495438e-01
-1.14771593e+00 5.47490954e-01 1.24881901e-01 -3.32704037e-01
5.70197523e-01 5.11968732e-01 -3.24589789e-01 -4.61820751e-01
3.90927851e-01 1.90152869e-01 5.84177196e-01 -3.29812407e-01
2.01499686e-01 3.27413559e-01 4.06890452e-01 -5.00645518e-01
9.15186584e-01 2.21552685e-01 8.76037419e-01 -1.01729965e+00
-2.10980967e-01 -1.79498315e-01 -3.84147644e-01 -4.59448457e-01
7.15516388e-01 -2.80550122e-01 -1.42325270e+00 1.05153313e-02
-8.61185431e-01 -4.89513695e-01 -4.79750454e-01 5.34466267e-01
-8.16950440e-01 2.18827799e-01 -6.94567084e-01 -1.36428893e+00
-1.02394447e-01 -8.41251075e-01 3.51316750e-01 1.25640854e-01
-2.49570996e-01 -1.21751773e+00 4.16713625e-01 -4.30138290e-01
3.03516686e-01 6.63762331e-01 5.76836646e-01 8.52883980e-02
-2.31699675e-01 -3.45654905e-01 3.16105604e-01 2.58568108e-01
-1.44719794e-01 1.67379543e-01 -3.33849072e-01 -6.99794412e-01
7.96093762e-01 -7.13436976e-02 3.18159103e-01 5.97726524e-01
3.46487075e-01 -5.48181355e-01 -4.99954909e-01 2.19741926e-01
1.82939029e+00 3.30638051e-01 -6.54326752e-02 9.56367105e-02
3.80130500e-01 7.92273581e-01 6.89741671e-01 4.45807070e-01
-1.41705409e-01 9.64675307e-01 4.34674114e-01 -1.59666464e-01
5.41847289e-01 4.00294602e-01 2.71944731e-01 4.85876799e-01
-2.68069774e-01 8.35922584e-02 -6.83238447e-01 4.06868637e-01
-1.90086484e+00 -6.87862873e-01 -4.16961581e-01 2.25027275e+00
8.13369870e-01 -2.92857975e-01 3.33918363e-01 5.20059884e-01
9.81613457e-01 -3.34209561e-01 -5.19240201e-01 -7.10868180e-01
-7.68955722e-02 -2.11714342e-01 7.66250432e-01 1.05939245e+00
-8.90255272e-01 6.40516207e-02 6.16133547e+00 5.13047695e-01
-1.36998153e+00 -2.66042531e-01 9.00941193e-02 1.15154043e-01
2.66694427e-01 -8.28701258e-02 -6.55485630e-01 3.72786850e-01
7.79378355e-01 -9.71461713e-01 5.00095665e-01 8.13342750e-01
7.58188128e-01 -3.13912481e-01 -6.25688910e-01 3.82446021e-01
-2.88441390e-01 -7.21143425e-01 -5.58238864e-01 9.31324288e-02
6.96377277e-01 -7.79492319e-01 1.88045248e-01 -1.96913466e-01
-7.62419701e-02 -4.13614482e-01 5.65227985e-01 4.10759330e-01
3.84527862e-01 -7.58583128e-01 5.63559294e-01 4.61605251e-01
-1.34575319e+00 -1.22121736e-01 -2.57009387e-01 -3.33912104e-01
6.35813892e-01 5.48583150e-01 -4.52936798e-01 7.70094991e-01
-1.63648352e-01 4.22856688e-01 1.47207141e-01 8.69883537e-01
5.67932129e-02 1.28162935e-01 -6.64025784e-01 -5.91031253e-01
1.12256318e-01 -6.64601028e-01 9.77544248e-01 6.08082950e-01
1.48737565e-01 6.26171589e-01 8.62844288e-02 1.17354369e+00
5.65618932e-01 2.21728683e-01 -4.41113740e-01 3.03171158e-01
-2.99642161e-02 1.19468808e+00 -5.46090245e-01 -2.58186758e-01
1.24659933e-01 5.61813056e-01 -4.04422954e-02 3.53638202e-01
-9.21520531e-01 -6.96942270e-01 5.96167386e-01 1.06675327e-01
-1.81795388e-01 -3.60578328e-01 -2.78781503e-01 -1.03901327e+00
-1.36774614e-01 -1.04329258e-01 2.38678828e-01 -3.11305791e-01
-7.37053990e-01 5.54266155e-01 2.61710435e-01 -1.38904643e+00
-5.66987932e-01 -5.28265715e-01 -5.04384696e-01 1.01584029e+00
-1.02760184e+00 -1.26505300e-01 2.51339763e-01 5.55146396e-01
1.86371151e-03 5.25461316e-01 7.06071019e-01 2.58538395e-01
-1.13126969e+00 -9.64807048e-02 4.92781818e-01 -3.14668268e-01
2.38826886e-01 -1.34270155e+00 -7.24798322e-01 9.66530740e-01
-1.29333818e+00 5.55480063e-01 1.51077461e+00 -5.56568146e-01
-1.63622129e+00 -1.06477737e+00 8.74988377e-01 2.76637673e-01
9.99383450e-01 -3.12157231e-03 -1.04782283e+00 5.13174176e-01
1.28177539e-01 -1.19472958e-01 -2.49520749e-01 -7.50443280e-01
7.85399139e-01 7.17731714e-02 -1.13731313e+00 2.84736872e-01
2.58632839e-01 -2.43314892e-01 -4.92115766e-01 2.31657982e-01
3.10945123e-01 -5.96185803e-01 -1.10355854e+00 4.94923711e-01
1.12867005e-01 6.37492239e-02 7.57192731e-01 -3.24233443e-01
1.20583810e-01 -5.63478827e-01 3.03385973e-01 -1.47371423e+00
-9.32306275e-02 -1.00234783e+00 -1.42638713e-01 8.68182838e-01
3.53808910e-01 -8.13580036e-01 5.10086119e-01 8.56340647e-01
-6.07657544e-02 -7.35162437e-01 -9.75424409e-01 -1.13962173e+00
1.86604068e-01 3.95486265e-01 -1.71925917e-01 8.30667019e-01
1.04939556e+00 3.60267907e-01 -3.20794404e-01 5.10833919e-01
5.19015253e-01 -9.03312042e-02 3.62123281e-01 -1.11945987e+00
-2.92329639e-01 -3.78411353e-01 8.89950544e-02 -7.68900514e-01
2.09479824e-01 -4.02969211e-01 3.44611257e-01 -1.41536391e+00
-4.74509150e-01 -1.02881119e-01 1.07702650e-01 5.14753126e-02
-9.57794711e-02 -3.54501680e-02 8.61638039e-02 1.90634772e-01
1.55803859e-01 8.05591166e-01 1.30850649e+00 3.27281617e-02
-7.49275029e-01 4.00382638e-01 2.25722492e-01 5.44434130e-01
1.00342858e+00 -5.86410575e-02 -5.55210531e-01 1.35213196e-01
1.22533411e-01 6.90204322e-01 4.18752402e-01 -8.66699755e-01
2.11128473e-01 -3.19638997e-01 -5.29931784e-01 -4.44924235e-01
4.92232561e-01 -9.43611503e-01 6.33565605e-01 1.25744009e+00
-2.40580291e-01 -1.46646187e-01 1.14159457e-01 3.08531731e-01
-3.42815310e-01 -3.95816803e-01 1.14949048e+00 3.67170125e-01
-1.62566155e-01 -3.62400323e-01 -9.74138558e-01 -3.20108831e-01
1.09708071e+00 -3.78611572e-02 -2.82090250e-02 -3.17365587e-01
-1.29817116e+00 2.73770690e-01 3.08913082e-01 -1.60285145e-01
3.38201709e-02 -1.13412535e+00 -2.52348095e-01 -1.23692557e-01
-5.48712969e-01 -3.66383344e-01 1.99735433e-01 1.18726587e+00
-5.47299564e-01 8.25406432e-01 -3.45741987e-01 -3.43049616e-01
-1.07672095e+00 5.43677211e-01 8.28697801e-01 5.27016781e-02
-3.13796073e-01 1.69342890e-01 -1.30429253e-01 1.82606801e-01
-1.81941077e-01 -4.83426869e-01 -3.87973756e-01 -6.62603006e-02
7.59380609e-02 8.03343177e-01 -1.39049828e-01 -6.67271852e-01
-3.68669301e-01 8.37266564e-01 7.81822681e-01 -3.41813952e-01
1.06472099e+00 -4.22268927e-01 -2.50354946e-01 2.70352244e-01
9.78258431e-01 -1.97321132e-01 -1.25982559e+00 1.05857290e-01
-2.42874354e-01 2.46986955e-01 1.71790197e-01 -2.07088217e-01
-1.03497219e+00 1.93858221e-01 2.74843723e-01 9.58758712e-01
1.07331288e+00 -3.65617365e-01 2.56186336e-01 4.78769779e-01
9.59577709e-02 -1.41996443e+00 -3.04127425e-01 4.65608001e-01
1.04702020e+00 -7.05426157e-01 1.85662150e-01 -1.22739112e+00
-1.08345255e-01 1.35598469e+00 5.67862272e-01 -6.88697338e-01
9.70421851e-01 4.13761407e-01 -3.91008019e-01 3.16629618e-01
-3.07843149e-01 -8.75366330e-02 3.46815825e-01 3.51236947e-02
4.72811937e-01 6.01350553e-02 -1.28613985e+00 4.61953193e-01
1.26656353e-01 1.23820174e-03 8.54207873e-01 6.92483306e-01
-6.95834517e-01 -7.70178318e-01 -6.14083052e-01 -3.08666766e-01
-3.72893810e-01 7.49249637e-01 -4.56170589e-01 1.18051112e+00
-2.23485857e-01 1.12032270e+00 -2.79474169e-01 1.88543335e-01
8.15592945e-01 -1.72396690e-01 2.32428461e-01 -3.84445399e-01
-3.97285044e-01 3.72479767e-01 1.40388697e-01 -3.43524694e-01
-4.09170926e-01 -4.57489431e-01 -1.45861614e+00 -2.57430702e-01
-8.29295874e-01 4.98153687e-01 5.74470639e-01 7.15455472e-01
-1.40622437e-01 7.06709325e-01 7.50992119e-01 -9.16087031e-01
-1.02388179e+00 -6.91367805e-01 -9.31743205e-01 -1.62430093e-01
5.31988502e-01 -9.16246176e-01 -7.89625943e-01 -1.00758299e-01]
|
[5.477224826812744, 2.653379440307617]
|
60bd6ef8-9eab-487d-b091-02afe8e66a3d
|
nonnegative-tensor-factorization-for
|
1411.501
| null |
http://arxiv.org/abs/1411.5010v2
|
http://arxiv.org/pdf/1411.5010v2.pdf
|
Nonnegative Tensor Factorization for Directional Blind Audio Source Separation
|
We augment the nonnegative matrix factorization method for audio source
separation with cues about directionality of sound propagation. This improves
separation quality greatly and removes the need for training data, with only a
twofold increase in run time. This is the first method which can exploit
directional information from microphone arrays much smaller than the wavelength
of sound, working both in simulation and in practice on millimeter-scale
microphone arrays.
|
['Noah D. Stein']
|
2014-11-18
| null | null | null | null |
['audio-source-separation']
|
['audio']
|
[ 1.16564007e-02 -8.14400613e-02 4.51698601e-01 6.08778000e-02
-9.44777668e-01 -9.03107285e-01 8.56081396e-02 -1.82373554e-01
-3.00359786e-01 5.92040360e-01 4.02328819e-01 -7.91324973e-01
-2.02968821e-01 -5.47834635e-01 -4.13578413e-02 -9.55520809e-01
-5.59695303e-01 -2.48204712e-02 1.45006567e-01 -6.58477619e-02
1.92979013e-03 5.43847024e-01 -1.40723920e+00 1.38744548e-01
3.33637565e-01 9.01747227e-01 -4.47820574e-02 1.50100911e+00
-5.59394211e-02 5.16490102e-01 -9.19057727e-01 1.04908809e-01
2.38290429e-01 -3.42745781e-01 -1.15789458e-01 -2.92247772e-01
2.90883273e-01 -8.20932537e-03 -3.04965645e-01 7.92454779e-01
9.27802384e-01 1.26341224e-01 4.46241349e-01 -9.72055733e-01
-9.44080576e-02 7.39072680e-01 -5.63488722e-01 5.30808449e-01
3.45443189e-01 -3.96644145e-01 8.05637896e-01 -8.93436193e-01
-1.59230996e-02 1.13821816e+00 8.50277662e-01 3.09558004e-01
-8.86786759e-01 -8.79274130e-01 -2.26728171e-01 -3.19011033e-01
-1.06765556e+00 -1.07629955e+00 9.14765775e-01 -3.77405792e-01
1.12967885e+00 7.34990537e-01 3.90802681e-01 3.73829514e-01
-2.46444941e-01 3.25144887e-01 7.74134040e-01 -7.67483890e-01
2.78266519e-01 4.00221348e-03 8.08565915e-02 4.29336071e-01
6.92815632e-02 3.78616601e-01 -6.80195570e-01 -7.05292106e-01
7.19591975e-01 -5.75490415e-01 -3.98540348e-01 -1.52544454e-01
-1.30228412e+00 5.11329353e-01 -6.35849079e-03 3.57109249e-01
-2.01069295e-01 4.81558353e-01 1.17895149e-01 6.64485276e-01
4.35135275e-01 7.89912522e-01 -4.13231432e-01 -4.64442283e-01
-8.68039846e-01 -1.99334845e-01 8.88709664e-01 6.05558276e-01
3.31419319e-01 7.43307292e-01 6.48083627e-01 1.02328122e+00
6.02068961e-01 1.18436885e+00 2.07777917e-01 -1.08325720e+00
4.86914426e-01 -1.73345193e-01 2.20535278e-01 -1.00803936e+00
-6.22261941e-01 -6.69065356e-01 -5.91236770e-01 3.37522537e-01
4.42015827e-01 -1.00614083e+00 -6.59953237e-01 1.56018198e+00
3.77440721e-01 4.93731976e-01 2.60025859e-01 7.42598116e-01
7.57843077e-01 5.78835130e-01 -7.90677488e-01 -4.81028259e-01
1.04889309e+00 -1.05702031e+00 -4.05422926e-01 -2.73535281e-01
2.02762142e-01 -1.36657405e+00 3.20664644e-01 8.85427356e-01
-1.05999982e+00 -2.90464848e-01 -1.23778605e+00 5.32349825e-01
1.61125273e-01 -2.75359780e-01 1.21960545e+00 1.42362869e+00
-1.18573189e+00 -9.93406475e-02 -9.17195976e-01 2.71098137e-01
-1.84239313e-01 5.83389103e-01 -2.77133465e-01 2.77463198e-01
-8.58767152e-01 -2.88590807e-02 -6.70296848e-01 1.86358079e-01
-3.42052400e-01 -9.42754805e-01 -5.18279135e-01 8.25661942e-02
-1.95830166e-01 -5.01486123e-01 1.36354363e+00 -6.61754072e-01
-1.62470376e+00 -9.42844823e-02 -5.79853117e-01 -2.39575803e-01
-9.18595120e-02 -4.46209051e-02 -9.74555194e-01 3.45203817e-01
5.22629917e-03 2.26208091e-01 1.16806722e+00 -1.37089145e+00
-6.20522201e-01 -2.19125167e-01 -8.16622376e-02 -1.65072232e-01
-5.23625076e-01 3.24315459e-01 -2.61126608e-01 -7.22378016e-01
6.35539770e-01 -1.00370991e+00 -5.00275791e-01 -3.78239959e-01
-1.55561566e-01 3.58233064e-01 5.64993083e-01 -5.56750894e-01
9.95980561e-01 -2.36175990e+00 1.00163333e-01 5.37077725e-01
2.77186900e-01 8.19668919e-02 -2.44893059e-01 6.47794127e-01
-2.94767827e-01 -1.74278930e-01 5.60133867e-02 -3.63919705e-01
-4.38673466e-01 -3.22381943e-01 -5.72050750e-01 3.79251301e-01
-4.58888322e-01 1.33124694e-01 -8.10870111e-01 8.28089789e-02
-4.51144017e-02 4.94412243e-01 -9.91996408e-01 1.78141698e-01
5.94537735e-01 5.85483670e-01 -9.98871177e-02 4.18702900e-01
6.84552610e-01 1.77823737e-01 -8.14449117e-02 4.59316336e-02
-4.35403645e-01 4.89578158e-01 -1.70231295e+00 1.46753991e+00
-1.06672704e+00 1.06491411e+00 1.12073600e+00 -6.94398165e-01
6.74501359e-01 8.75356376e-01 8.34463596e-01 -3.07934642e-01
-3.23571078e-02 3.50989908e-01 3.11099589e-01 -3.47868294e-01
1.39457375e-01 -5.72912037e-01 1.44284576e-01 8.79233837e-01
4.96480130e-02 -5.13158262e-01 -1.27238138e-02 2.58558333e-01
1.27536285e+00 -1.08202779e+00 -9.50240269e-02 -3.25322330e-01
3.71614218e-01 -4.44100827e-01 3.34324986e-01 4.18689489e-01
2.54958063e-01 4.65845764e-01 -8.49165767e-02 2.92275250e-02
-3.41803789e-01 -1.06870353e+00 -7.04806298e-02 1.22939622e+00
-1.43645024e-02 -7.06863999e-01 -4.06547934e-01 -9.72964764e-02
1.50652174e-02 3.36112201e-01 -1.10652946e-01 3.05924475e-01
-5.54277778e-01 -8.81593168e-01 7.30137885e-01 5.96367240e-01
-2.05818802e-01 -3.05313110e-01 -3.19798738e-01 2.39052400e-01
2.02846527e-02 -8.38417828e-01 -1.31267488e-01 6.73073173e-01
-8.30833614e-01 -7.49954522e-01 -4.81255263e-01 -6.73194289e-01
5.60598135e-01 7.95810938e-01 9.01768267e-01 -8.51109102e-02
-2.84097821e-01 6.71881378e-01 -1.83160126e-01 -8.20067585e-01
-7.63873607e-02 -4.65492696e-01 5.35350144e-01 -2.90269136e-01
-2.47294456e-01 -1.32321787e+00 -5.16176164e-01 3.93377274e-01
-3.53886992e-01 -4.62717384e-01 2.53398061e-01 5.10930479e-01
-1.08583026e-01 7.63208389e-01 7.80956924e-01 -3.61093342e-01
7.17862308e-01 -3.43912065e-01 -5.20774364e-01 -3.49211007e-01
-2.59030253e-01 -8.25423840e-03 6.66482747e-01 -2.41760790e-01
-9.54628646e-01 7.40396678e-02 -4.50883687e-01 8.92035961e-02
-1.30162463e-01 3.21242332e-01 -3.66362222e-02 -5.54511070e-01
7.60465980e-01 -2.87978262e-01 -3.08082879e-01 -6.97872996e-01
4.42093968e-01 7.98741698e-01 5.12563407e-01 -2.48788178e-01
9.60647762e-01 6.98280096e-01 1.98946595e-02 -1.46078956e+00
-1.21321909e-01 -1.06961441e+00 -3.94533187e-01 -4.82497411e-03
2.49436051e-01 -8.79117250e-01 -5.19303441e-01 3.31607461e-01
-1.05980134e+00 -1.52423635e-01 -6.67706951e-02 1.14418912e+00
-7.15981871e-02 1.67043865e-01 -3.77691925e-01 -1.07863665e+00
-1.00211509e-01 -5.70117712e-01 9.04874563e-01 -1.83967218e-01
-1.49365753e-01 -9.75882590e-01 4.78567094e-01 1.66580856e-01
6.91107571e-01 -5.02926648e-01 7.54018188e-01 -6.06847405e-02
-2.47758359e-01 -2.87979752e-01 -7.23647252e-02 2.78054744e-01
4.21701878e-01 -5.79724200e-02 -1.32541049e+00 -2.56350547e-01
5.46144247e-01 1.91761181e-01 9.34898674e-01 6.25125945e-01
3.59187126e-01 -2.32883748e-02 -4.41208601e-01 5.51242471e-01
1.15239286e+00 4.47693497e-01 1.59591824e-01 -1.55103788e-01
6.98369682e-01 3.06160659e-01 1.49409771e-01 3.32572877e-01
-3.64167422e-01 5.47315657e-01 1.86776921e-01 -2.85458118e-01
-4.28755641e-01 2.75825322e-01 2.88219333e-01 1.28822935e+00
2.68011857e-02 -3.65849048e-01 -8.82027149e-01 6.19616747e-01
-1.33063018e+00 -8.30932319e-01 -2.94209212e-01 2.15863585e+00
3.46212059e-01 -4.73385565e-02 2.25242853e-01 8.10465634e-01
3.33530664e-01 1.05080940e-01 -6.72105327e-02 -5.00002623e-01
2.83097416e-01 2.79788733e-01 5.87429881e-01 1.24500370e+00
-1.02952111e+00 2.78179109e-01 8.06702995e+00 4.64814335e-01
-1.50852156e+00 1.17635265e-01 -2.05890045e-01 -4.09442067e-01
-5.94740868e-01 -3.13283652e-01 -2.82542348e-01 3.03216800e-02
1.06967068e+00 1.78123545e-02 5.14587343e-01 6.01126969e-01
-7.62939453e-06 4.52328138e-02 -8.69250536e-01 1.45452166e+00
1.57298684e-01 -1.09621966e+00 -5.35287023e-01 -1.14668131e-01
5.29121518e-01 1.52352571e-01 2.42408872e-01 -3.28418255e-01
5.62484041e-02 -7.94042051e-01 5.00995696e-01 2.78566615e-03
3.27703387e-01 -8.04886341e-01 4.31345731e-01 3.53315324e-01
-1.39705896e+00 -2.16169044e-01 -2.69211978e-01 -5.20278275e-01
6.25889838e-01 1.05166674e+00 -1.02847624e+00 2.17006832e-01
4.92742419e-01 2.63453126e-01 -2.89252669e-01 1.29382443e+00
-1.62241593e-01 1.32123494e+00 -7.43858278e-01 1.66731268e-01
3.92453745e-02 -1.07448243e-01 1.09512019e+00 1.46793425e+00
6.86630428e-01 1.72171220e-01 -2.27379188e-01 -1.40244141e-01
2.30935201e-01 -1.14431299e-01 -6.47983909e-01 1.16243944e-01
5.10715425e-01 1.20329916e+00 -9.42007303e-01 1.88428327e-01
-4.48545039e-01 5.96121132e-01 -4.61534053e-01 5.57901382e-01
-3.97022486e-01 -7.92950392e-01 7.68799961e-01 9.21101496e-02
4.33907092e-01 -7.66065001e-01 -2.77592301e-01 -7.76433170e-01
-1.19512945e-01 -8.08526218e-01 -1.37666583e-01 -4.00051206e-01
-9.58457828e-01 6.59206450e-01 -3.82385790e-01 -1.26556110e+00
-5.79402983e-01 -5.91848135e-01 -7.11110294e-01 7.32863307e-01
-1.08288789e+00 -5.45000076e-01 2.08558530e-01 4.64714527e-01
1.28572583e-01 -3.53741527e-01 1.08758187e+00 6.85546517e-01
-1.51251540e-01 4.10948783e-01 5.34904122e-01 3.84734245e-03
7.18386471e-01 -1.35389483e+00 5.69567680e-01 7.50429273e-01
7.18284488e-01 1.01795745e+00 1.21626842e+00 8.94593373e-02
-1.59998274e+00 -4.34064418e-01 6.06332600e-01 -4.05797988e-01
8.65829825e-01 -5.39957702e-01 -3.04727912e-01 1.09917015e-01
3.59742135e-01 -7.03387186e-02 1.55922437e+00 7.47996926e-01
-6.07398808e-01 -3.05788487e-01 -7.82462299e-01 1.44997492e-01
8.81965816e-01 -5.44259191e-01 -3.40207040e-01 4.50709999e-01
7.63378024e-01 -1.21518403e-01 -5.67563415e-01 6.53719082e-02
7.06178665e-01 -1.00645828e+00 1.36637652e+00 -2.28844985e-01
-2.42335066e-01 -3.52239817e-01 -4.17003572e-01 -1.54669416e+00
-4.61422414e-01 -1.16063821e+00 -5.40818274e-02 1.19875002e+00
8.07273805e-01 -7.50499725e-01 7.74467826e-01 2.51286417e-01
-2.06538618e-01 -2.53926545e-01 -1.09900987e+00 -9.79960740e-01
-5.91923185e-02 -1.08472264e+00 6.53044522e-01 5.89591205e-01
3.84660542e-01 7.84298062e-01 -1.98620453e-01 7.29303062e-01
6.97147012e-01 2.70250678e-01 5.33327520e-01 -1.19584608e+00
-8.51272106e-01 -2.96800584e-01 -5.09562314e-01 -1.43868148e+00
-7.22763762e-02 -4.36774433e-01 1.49185881e-01 -1.24129784e+00
-5.65732956e-01 -9.31332827e-01 -4.50482994e-01 4.46199328e-02
1.14858240e-01 6.64929509e-01 -3.54772657e-02 -3.22488219e-01
-3.57746571e-01 1.43928334e-01 6.06601596e-01 -1.50150210e-01
-4.15103376e-01 4.90277737e-01 -8.99014235e-01 1.07876706e+00
6.19765520e-01 -5.60898960e-01 -8.32189262e-01 -9.48208511e-01
7.01907337e-01 3.81897748e-01 -4.97795902e-02 -1.33177245e+00
4.12961125e-01 1.54653400e-01 1.66822806e-01 -1.86770409e-01
5.36928117e-01 -9.10248101e-01 -9.23551545e-02 2.24539004e-02
-1.21882617e-01 -4.65375185e-02 5.08235872e-01 8.49117398e-01
-4.59427416e-01 -3.96397291e-03 4.39872265e-01 3.51903856e-01
-1.35219857e-01 -2.41693690e-01 -8.65612745e-01 -2.07640111e-01
4.60285336e-01 1.73479840e-02 -8.87985826e-02 -8.25187385e-01
-6.78439438e-01 -3.07966918e-01 -5.72650284e-02 2.26460159e-01
3.91307592e-01 -1.25395334e+00 -4.45182562e-01 5.71270227e-01
-4.34762627e-01 -3.94513875e-01 1.69005051e-01 7.76187837e-01
-2.93544978e-01 7.04386830e-01 1.59455776e-01 -4.04681981e-01
-1.69632995e+00 5.45264184e-01 1.51732355e-01 3.65541428e-01
-2.81160474e-01 1.81422639e+00 1.48037001e-01 -4.14535046e-01
1.55121535e-01 -2.44266748e-01 -3.44948024e-01 3.63911428e-02
9.61876810e-01 6.51456833e-01 3.18284959e-01 -5.84397137e-01
-6.40415251e-01 7.12700546e-01 5.01823425e-01 -9.12234426e-01
1.23979330e+00 -2.29535356e-01 -5.63466012e-01 5.24244785e-01
1.20907176e+00 1.48830926e+00 -7.28407443e-01 3.23651761e-01
-5.38655221e-01 -6.86850607e-01 5.19958973e-01 -6.07441068e-01
-1.18489707e+00 1.21178079e+00 5.99245369e-01 6.38689280e-01
1.10038424e+00 -2.79154181e-02 8.26782227e-01 6.66744947e-01
6.41144454e-01 -5.46691775e-01 2.72710770e-02 2.28638753e-01
6.79323673e-01 -7.36104190e-01 -8.66609290e-02 -7.12563992e-01
-4.04066816e-02 9.57940817e-01 -2.04688814e-02 -5.73428608e-02
1.38991749e+00 1.11202836e+00 6.59394383e-01 -6.75755227e-03
-5.95704794e-01 2.19030574e-01 3.51788461e-01 9.59812045e-01
7.35107422e-01 2.49494895e-01 3.20716590e-01 7.73512006e-01
-6.18331552e-01 -5.94256759e-01 2.86058962e-01 7.81576395e-01
-7.76614726e-01 -1.55143607e+00 -7.03035235e-01 3.44731599e-01
-5.12551129e-01 -3.60115647e-01 -4.78545278e-01 -2.69474871e-02
-1.31023869e-01 1.88206196e+00 -3.07554025e-02 -4.54744518e-01
1.89388797e-01 -1.56613216e-01 3.82137150e-01 -7.71210074e-01
-3.52822930e-01 5.94811738e-01 3.21199059e-01 -3.00724864e-01
-3.40689838e-01 -5.35852253e-01 -1.39086258e+00 -1.48031667e-01
-6.22121692e-01 9.58781362e-01 7.15643108e-01 6.33232236e-01
4.66452986e-01 7.45204806e-01 8.57956886e-01 -9.56056833e-01
8.47671367e-03 -5.56509018e-01 -7.41612613e-01 -4.68254387e-01
9.86377954e-01 -5.13096988e-01 -8.55828702e-01 -1.67792425e-01]
|
[15.20190715789795, 5.633871555328369]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.