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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
7a883491-24fa-40ec-858d-737cd18e3f75
|
a-review-of-3d-human-pose-estimation
|
2010.06449
| null |
https://arxiv.org/abs/2010.06449v3
|
https://arxiv.org/pdf/2010.06449v3.pdf
|
A review of 3D human pose estimation algorithms for markerless motion capture
|
Human pose estimation is a very active research field, stimulated by its important applications in robotics, entertainment or health and sports sciences, among others. Advances in convolutional networks triggered noticeable improvements in 2D pose estimation, leading modern 3D markerless motion capture techniques to an average error per joint of 20 mm. However, with the proliferation of methods, it is becoming increasingly difficult to make an informed choice. Here, we review the leading human pose estimation methods of the past five years, focusing on metrics, benchmarks and method structures. We propose a taxonomy based on accuracy, speed and robustness that we use to classify de methods and derive directions for future research.
|
['Philippe Montesinos', 'Pierre Slangen', 'Denis Mottet', 'Yann Desmarais']
|
2020-10-13
| null | null | null | null |
['markerless-motion-capture']
|
['computer-vision']
|
[-1.04670912e-01 3.74897160e-02 -5.47296405e-01 6.45923242e-03
-2.78916448e-01 -2.64432937e-01 4.82008070e-01 1.96621772e-02
-7.24727631e-01 7.82030404e-01 3.39371413e-01 1.67714238e-01
-1.59610584e-01 -3.53601456e-01 -4.31852221e-01 -2.83811659e-01
-3.65020752e-01 4.47083026e-01 3.04839432e-01 -1.67753294e-01
1.10875212e-01 7.05540895e-01 -1.29331553e+00 -5.19370735e-01
3.63790959e-01 9.65103269e-01 -1.87417820e-01 4.80500340e-01
5.57265460e-01 3.23394448e-01 -5.69567859e-01 -4.46353495e-01
2.58466333e-01 -2.23781005e-01 -5.88717699e-01 -1.14897251e-01
3.05769950e-01 -2.73738772e-01 -5.66746593e-01 7.01293707e-01
1.03657281e+00 1.93550721e-01 5.08982658e-01 -1.36747706e+00
-7.08889216e-02 2.21366078e-01 -4.93408889e-01 6.85808733e-02
6.89777911e-01 2.05940977e-02 5.63871264e-01 -8.83610010e-01
1.00875211e+00 9.95144188e-01 1.09063280e+00 7.65419841e-01
-5.31010509e-01 -4.44629699e-01 -2.44350545e-02 2.08757877e-01
-1.31341016e+00 -3.26848954e-01 8.84293854e-01 -3.27032089e-01
7.58475780e-01 1.78824663e-01 1.15386975e+00 1.25542307e+00
6.76899791e-01 1.06498289e+00 6.60419285e-01 -2.13245839e-01
2.16409400e-01 -4.56507325e-01 -2.13619694e-01 5.77773511e-01
7.85229862e-01 7.87982717e-02 -6.48909688e-01 3.40615325e-02
1.13023603e+00 -2.17729852e-01 -2.04974443e-01 -1.05696237e+00
-1.55946994e+00 5.93140900e-01 4.12614703e-01 1.04614474e-01
-4.22237009e-01 5.14651358e-01 5.94935417e-01 -2.09705740e-01
2.72188365e-01 6.04931891e-01 -3.76174271e-01 -6.58972144e-01
-8.07721853e-01 8.53169084e-01 7.72665560e-01 8.71927798e-01
-9.18591619e-02 -4.41721119e-02 2.09802911e-01 5.28543293e-01
3.71692359e-01 3.81852895e-01 4.29420173e-01 -1.19286394e+00
2.74824649e-01 3.98849219e-01 2.40244865e-01 -1.32609296e+00
-1.08301830e+00 -5.88224113e-01 -9.43772972e-01 1.01606965e-01
4.61029559e-01 -2.28054479e-01 -6.47417843e-01 1.47730899e+00
4.53000605e-01 -1.23425402e-01 -5.59965491e-01 1.35138845e+00
7.76357651e-01 -1.81352302e-01 -9.32151452e-03 2.03384385e-02
1.17295671e+00 -7.99550772e-01 -8.17257822e-01 -5.50194442e-01
5.84693730e-01 -6.19018734e-01 3.65772367e-01 6.63828671e-01
-1.11125886e+00 -3.94042045e-01 -1.25008571e+00 -3.91026400e-02
-8.38876218e-02 9.93985459e-02 9.33662355e-01 6.14938617e-01
-6.84748411e-01 8.16924214e-01 -1.14781499e+00 -6.22780442e-01
3.11963320e-01 6.16997957e-01 -4.78946239e-01 2.88876951e-01
-1.30043638e+00 1.31636035e+00 1.19545355e-01 3.93806100e-01
-4.20838535e-01 -3.71687889e-01 -8.75978649e-01 -6.12608790e-01
4.50021237e-01 -1.08923781e+00 1.23607659e+00 -8.48695561e-02
-1.74278343e+00 8.76756132e-01 2.98280686e-01 -6.33542478e-01
9.55734015e-01 -8.71486902e-01 -2.91143179e-01 -1.67993512e-02
-6.04772642e-02 7.01295018e-01 4.91539747e-01 -7.19830990e-01
-5.17657578e-01 -5.16025782e-01 4.51381318e-02 4.70720291e-01
-4.86374460e-03 -1.60895288e-01 -7.39446759e-01 -7.06025004e-01
4.35710132e-01 -1.35716772e+00 -5.76520741e-01 5.07344663e-01
-4.32656854e-01 -1.08634427e-01 4.08656210e-01 -4.51018006e-01
1.12748611e+00 -1.72988701e+00 5.52655756e-01 -1.41479252e-02
5.75393617e-01 2.16484368e-01 2.97878355e-01 1.52736917e-01
2.70335257e-01 -2.65485913e-01 2.72719725e-03 -2.54049301e-01
-1.39826089e-02 -5.53071592e-03 2.56301343e-01 8.97189677e-01
-1.29608691e-01 1.02929270e+00 -1.00045455e+00 -5.02209008e-01
5.51259995e-01 6.45641088e-01 -2.93403000e-01 -5.78142069e-02
2.81155944e-01 5.38765728e-01 -4.63630080e-01 7.38452733e-01
2.75781810e-01 -1.76795185e-01 6.62928522e-02 -3.12802255e-01
8.79460275e-02 2.49508053e-01 -1.41133130e+00 2.03647375e+00
-1.41152218e-01 8.35929692e-01 4.20658253e-02 -7.64895141e-01
8.00053954e-01 2.90863097e-01 1.02548611e+00 -3.76119703e-01
6.42812371e-01 4.09732044e-01 3.38452458e-02 -3.49375010e-01
7.99763441e-01 1.32685393e-01 -4.53510553e-01 1.38799340e-01
-1.23595625e-01 -9.52410921e-02 5.83113693e-02 -2.49374017e-01
1.01213694e+00 5.64743757e-01 6.24200821e-01 6.31683320e-02
3.19868177e-01 1.25432447e-01 5.17630994e-01 3.87857497e-01
-8.85191202e-01 8.35974336e-01 6.65266290e-02 -7.59397626e-01
-9.64809179e-01 -1.02469063e+00 -1.32891819e-01 5.60334384e-01
3.73004019e-01 -3.15155447e-01 -6.72161758e-01 -3.79425019e-01
2.19016969e-01 -3.43998939e-01 -6.05349720e-01 -1.96962327e-01
-1.09287918e+00 -4.76612478e-01 7.36358404e-01 9.76509333e-01
7.01225579e-01 -9.49321866e-01 -1.18833053e+00 2.48832941e-01
-3.02001894e-01 -1.28697586e+00 -1.24508515e-01 -1.09039389e-01
-1.11773860e+00 -1.05924416e+00 -1.19969749e+00 -6.18009865e-01
2.28025749e-01 5.69850095e-02 1.20891356e+00 -1.62297279e-01
-3.98738950e-01 3.37058246e-01 -2.66091436e-01 -4.39948052e-01
1.28262654e-01 3.83182794e-01 6.12603128e-01 -5.95638275e-01
2.38157138e-01 -2.79918045e-01 -9.20713484e-01 3.55194330e-01
-2.86595047e-01 -3.01099401e-02 6.64879024e-01 4.48706239e-01
4.20381546e-01 -6.28449857e-01 2.39289314e-01 -3.48149776e-01
5.76425910e-01 -2.12593421e-01 -1.79295868e-01 -2.16022387e-01
-3.30131173e-01 -1.77802458e-01 1.54406354e-02 -4.95340347e-01
-5.09650350e-01 2.49383703e-01 -2.89069504e-01 -3.22489679e-01
-2.21356139e-01 5.73203504e-01 9.05085802e-02 -3.05553496e-01
7.83991098e-01 -2.28066325e-01 2.05362365e-01 -3.64002079e-01
3.09668690e-01 4.52713370e-01 8.16837549e-01 -2.77965784e-01
5.21351337e-01 4.62944835e-01 3.15441400e-01 -8.57845604e-01
-6.67704821e-01 -4.80891913e-01 -8.94962907e-01 -6.17678523e-01
7.87371218e-01 -7.01405048e-01 -1.00663173e+00 5.88747323e-01
-1.21350169e+00 -6.03930280e-03 -1.13059722e-01 8.12528074e-01
-8.08022738e-01 3.73068184e-01 -5.87432325e-01 -6.21222854e-01
-3.25565666e-01 -1.13961601e+00 1.10864675e+00 1.19893953e-01
-9.17179823e-01 -9.05677736e-01 2.82747835e-01 1.41581535e-01
2.66214222e-01 7.42720068e-01 1.26882941e-01 -2.60155529e-01
-2.26554006e-01 -7.06834495e-01 1.30692586e-01 -1.94690481e-01
-3.46071869e-02 -3.43209118e-01 -4.79384750e-01 -3.23071986e-01
-1.95576474e-01 -1.88468039e-01 3.13121945e-01 6.44761801e-01
7.81373262e-01 2.44782954e-01 -7.32436895e-01 5.10763407e-01
9.56122994e-01 -5.05234785e-02 5.91196656e-01 7.67124653e-01
7.63562620e-01 5.29016018e-01 7.24146724e-01 3.16256583e-01
2.97925532e-01 1.05807245e+00 4.65993971e-01 2.75016546e-01
-2.00408027e-01 -1.20641254e-01 -7.87560567e-02 9.42139268e-01
-7.51973510e-01 -2.51656445e-03 -1.02448237e+00 5.01006663e-01
-1.87764168e+00 -5.77828109e-01 -2.03762390e-02 2.31298304e+00
4.80179816e-01 3.41546983e-01 4.26100731e-01 3.86993468e-01
6.32204294e-01 1.69157997e-01 -6.09342396e-01 2.23363768e-02
1.69900969e-01 7.83010274e-02 6.49075627e-01 -1.62724648e-02
-1.27248812e+00 6.61286592e-01 7.47681856e+00 3.43714297e-01
-1.10743105e+00 -3.01860124e-01 1.02502227e-01 -1.14694946e-01
4.05334294e-01 -4.31286633e-01 -7.65897989e-01 3.32308233e-01
4.86256838e-01 -8.65076706e-02 -1.19729646e-01 9.37278509e-01
6.11656867e-02 -2.91458964e-01 -1.11457932e+00 1.28076684e+00
1.45914719e-01 -1.20124340e+00 -2.92852402e-01 3.20044830e-02
5.74750721e-01 -2.36616489e-02 -5.86554892e-02 1.06296435e-01
-1.33803844e-01 -9.69911158e-01 6.62005723e-01 5.28756261e-01
5.74725926e-01 -8.07430625e-01 8.23122323e-01 3.33587646e-01
-1.27221930e+00 1.63300604e-01 -1.65505618e-01 -4.96392220e-01
4.98801261e-01 4.15520638e-01 -4.32235748e-01 4.99609321e-01
7.33509302e-01 7.54928410e-01 -2.61025965e-01 1.55540454e+00
-2.97274232e-01 1.53580919e-01 -4.29802746e-01 -4.33525473e-01
-7.67050907e-02 1.34882122e-01 7.61293590e-01 8.59477937e-01
1.45357892e-01 -1.01698652e-01 1.26330897e-01 3.71753752e-01
6.09877519e-02 -8.80577415e-02 -5.87711453e-01 2.16563225e-01
4.86583441e-01 1.16135252e+00 -9.21501100e-01 1.43190697e-01
-1.56853676e-01 9.84215915e-01 2.93055419e-02 -1.98021028e-02
-9.49392796e-01 -5.80744028e-01 8.65125000e-01 2.11329564e-01
-1.55893013e-01 -9.62621331e-01 -4.54304069e-01 -1.00546014e+00
2.05634654e-01 -6.73486054e-01 5.46357855e-02 -5.54711163e-01
-8.19808722e-01 2.09870517e-01 2.25670740e-01 -1.73455739e+00
-6.38621569e-01 -8.21794927e-01 -5.72278500e-02 2.68031567e-01
-9.22198057e-01 -8.28064620e-01 -3.21605653e-01 -3.61677259e-02
4.10484165e-01 1.86762661e-01 6.60749018e-01 3.92708808e-01
-3.89532685e-01 6.52519405e-01 -2.56853253e-01 3.90631884e-01
7.04259396e-01 -7.76024222e-01 7.76217639e-01 4.65735227e-01
-8.35566148e-02 6.77133322e-01 9.24707234e-01 -6.69062257e-01
-1.64594018e+00 -5.14751077e-01 9.81978118e-01 -7.40245342e-01
4.60264295e-01 -1.96961850e-01 -3.87651086e-01 5.13388336e-01
-2.45847568e-01 1.35058627e-01 2.24421844e-01 1.13276452e-01
1.79796293e-01 1.84922948e-01 -9.95097160e-01 9.36061502e-01
1.41831410e+00 -2.60190330e-02 -6.21008039e-01 1.18150041e-01
3.53863090e-01 -1.18895626e+00 -9.46168363e-01 8.00489008e-01
1.14826202e+00 -7.41317809e-01 1.09616184e+00 -3.36304218e-01
2.12073848e-01 -7.55786300e-02 2.57368505e-01 -1.09004951e+00
-4.11935985e-01 -6.95311308e-01 -5.41450500e-01 3.62763941e-01
3.48798223e-02 -2.62903959e-01 1.36776173e+00 5.23605168e-01
-4.61978540e-02 -1.14450336e+00 -1.09924388e+00 -7.70175397e-01
2.40289536e-03 -6.28912807e-01 3.46874952e-01 6.49083257e-01
2.11853370e-01 2.25200862e-01 -6.75908029e-01 -2.28366628e-01
5.87541521e-01 -2.39135399e-01 1.01164043e+00 -1.42277884e+00
1.42712772e-01 -6.84322715e-01 -1.12870526e+00 -1.36721337e+00
-2.65442282e-01 -2.97247678e-01 2.19650075e-01 -1.64095485e+00
-1.73440859e-01 -1.01927062e-02 -7.87552521e-02 5.11698090e-02
7.60429306e-03 7.93269396e-01 1.88933536e-01 8.14618990e-02
-7.86417723e-01 4.04215306e-01 1.28142524e+00 1.48553669e-01
-9.69067216e-02 2.47131929e-01 -1.64194062e-01 1.07347572e+00
5.92277110e-01 -1.22466080e-01 4.30349261e-02 -4.05737847e-01
6.81429625e-01 3.37511040e-02 4.16347057e-01 -1.63009250e+00
2.58547634e-01 3.62720736e-03 7.45790005e-01 -7.01416373e-01
5.05463004e-01 -7.03520358e-01 1.82412684e-01 8.12138557e-01
-1.73065990e-01 2.22596347e-01 1.12514719e-01 3.80095601e-01
-1.18403152e-01 1.42392144e-01 4.88277525e-01 -1.37668014e-01
-9.38824832e-01 4.51142997e-01 -4.37250674e-01 1.19018413e-01
9.84667838e-01 -6.89561725e-01 2.19053701e-01 -4.82759833e-01
-7.96502829e-01 6.95174113e-02 4.05577570e-01 8.28914940e-01
6.75972998e-01 -1.73990011e+00 -3.69266629e-01 -1.61705717e-01
8.30868185e-02 9.41228401e-03 4.74673882e-02 1.06763566e+00
-7.74776995e-01 6.69201136e-01 -4.99391526e-01 -7.02599227e-01
-1.02409720e+00 3.78348827e-02 2.09637523e-01 -8.91786814e-02
-5.72397590e-01 6.81905508e-01 -5.15898824e-01 -4.83427912e-01
4.12698001e-01 -2.16476768e-01 -3.08057159e-01 -2.12754488e-01
2.11853638e-01 1.00732398e+00 1.56925231e-01 -8.07595432e-01
-6.64859116e-01 8.33921492e-01 4.17908192e-01 -1.30626455e-01
1.17244577e+00 -8.38696882e-02 3.04438174e-01 5.29168665e-01
1.02700400e+00 -2.47189239e-01 -1.17426026e+00 1.79763198e-01
1.78197697e-01 -4.74600822e-01 -2.12048352e-01 -3.74993831e-01
-9.37625408e-01 7.39943862e-01 6.85275435e-01 -1.20708361e-01
7.34637201e-01 -8.85833651e-02 1.03896880e+00 3.72480333e-01
8.49296331e-01 -1.30245101e+00 8.79975706e-02 6.88197970e-01
8.85823488e-01 -1.14892697e+00 5.83789051e-01 -5.47332406e-01
-3.28945190e-01 9.97932374e-01 6.45154417e-01 -5.48809230e-01
8.09546173e-01 2.45112240e-01 7.41397589e-02 -1.36927530e-01
-8.99611786e-02 -1.02483854e-01 7.04682648e-01 7.62592435e-01
9.49072897e-01 -5.68001717e-02 -7.79446542e-01 4.66841429e-01
-3.89510095e-01 2.32665807e-01 1.76689461e-01 1.28590834e+00
-4.34140146e-01 -9.99055326e-01 -4.15706486e-01 4.11804020e-01
-4.51860160e-01 6.13949835e-01 -4.47937787e-01 1.26979578e+00
-7.51138031e-02 5.27752876e-01 5.14838509e-02 -5.43817401e-01
6.11374319e-01 -2.06274986e-01 8.34041357e-01 -3.47048461e-01
-3.28064531e-01 -8.30646753e-02 3.21322083e-01 -8.75719488e-01
-6.02306545e-01 -6.56779766e-01 -1.17603815e+00 -2.70637065e-01
-3.82319033e-01 -2.79094309e-01 8.03290546e-01 9.97470558e-01
3.35692883e-01 3.29174548e-01 2.68559828e-02 -1.49197423e+00
-3.50314587e-01 -9.68866587e-01 -3.00094932e-01 1.77539423e-01
1.15529135e-01 -1.28037012e+00 -1.34415295e-05 -3.47056389e-01]
|
[7.006270885467529, -0.686739444732666]
|
96e0e583-63df-4881-baba-1d3b369eb6f9
|
leveraging-textures-in-zero-shot
|
2203.11449
| null |
https://arxiv.org/abs/2203.11449v2
|
https://arxiv.org/pdf/2203.11449v2.pdf
|
How well does CLIP understand texture?
|
We investigate how well CLIP understands texture in natural images described by natural language. To this end, we analyze CLIP's ability to: (1) perform zero-shot learning on various texture and material classification datasets; (2) represent compositional properties of texture such as red dots or yellow stripes on the Describable Texture in Detail(DTDD) dataset; and (3) aid fine-grained categorization of birds in photographs described by color and texture of their body parts.
|
['Subhransu Maji', 'Chenyun Wu']
|
2022-03-22
| null | null | null | null |
['material-classification']
|
['computer-vision']
|
[ 2.22381040e-01 -1.44942343e-01 -2.20649764e-01 -4.50465053e-01
-2.43041456e-01 -7.88775206e-01 8.57522845e-01 -5.35692237e-02
2.06555769e-01 5.14820158e-01 3.98706079e-01 1.52839169e-01
2.68826094e-02 -9.19797182e-01 -6.27596319e-01 -8.23139787e-01
-2.52981991e-01 5.29836357e-01 1.97370619e-01 -2.25649163e-01
1.15623541e-01 5.35984874e-01 -1.92665398e+00 8.32360148e-01
2.10827872e-01 1.53474379e+00 -2.10716158e-01 7.22743154e-01
-1.66639075e-01 1.38340986e+00 -2.96326816e-01 -4.12942678e-01
8.63433629e-02 -3.40839952e-01 -9.82128561e-01 6.18611872e-01
1.09002566e+00 -4.89919037e-01 -3.90132904e-01 1.00793660e+00
-1.69818029e-01 1.73064142e-01 1.20123041e+00 -7.97477543e-01
-1.18435967e+00 3.00101280e-01 -6.23448431e-01 2.82406539e-01
3.59525710e-01 2.68892080e-01 8.14039528e-01 -7.08589017e-01
1.04158425e+00 1.54221427e+00 3.89844537e-01 7.51304686e-01
-1.60191822e+00 -1.43949911e-01 -2.05106944e-01 -8.81084299e-04
-1.33857250e+00 -5.91950417e-01 6.85556352e-01 -6.35080397e-01
6.54745877e-01 2.83298731e-01 8.57172787e-01 1.26330829e+00
7.16150761e-01 6.46732390e-01 1.51841104e+00 -8.05685818e-02
4.94776934e-01 -3.14539701e-01 9.47127119e-02 1.16112924e+00
3.82932350e-02 4.33498621e-01 -6.10113442e-01 3.34561244e-03
8.27518880e-01 -1.95449173e-01 1.18062258e-01 -2.00511396e-01
-1.19663560e+00 5.55767536e-01 5.01365960e-01 -2.27722570e-01
-1.11549072e-01 4.32350636e-01 4.63914663e-01 4.09646153e-01
1.09634089e+00 5.94644964e-01 -5.48624210e-02 -1.42518491e-01
-6.99016392e-01 2.31575862e-01 7.32028306e-01 1.17721760e+00
9.52050030e-01 4.58059371e-01 -1.24101199e-01 9.70474243e-01
-3.28924686e-01 7.75848389e-01 2.00503513e-01 -1.05248749e+00
-2.49599218e-01 2.98362255e-01 1.15162954e-01 -1.09833324e+00
-3.55148077e-01 3.72800410e-01 -1.01651800e+00 3.93527150e-01
4.98209178e-01 2.48049259e-01 -1.20501268e+00 1.26291299e+00
2.43030097e-02 -2.77758926e-01 -6.13505132e-02 1.07332456e+00
1.34751415e+00 5.53174734e-01 2.22582728e-01 4.24903691e-01
1.70096660e+00 -8.08136761e-01 -5.61300099e-01 -2.79624581e-01
1.90510571e-01 -6.63378537e-01 1.14235401e+00 2.60727435e-01
-8.38142097e-01 -9.08132076e-01 -9.30844724e-01 -3.19950968e-01
-5.45911312e-01 -1.21786799e-02 8.22045863e-01 3.47284347e-01
-8.80784214e-01 7.86901534e-01 -6.82874143e-01 -2.40102649e-01
9.98800278e-01 -1.94025978e-01 -6.68114841e-01 -2.81578265e-02
-6.32257760e-01 6.62189186e-01 1.04313664e-01 -2.65772820e-01
-1.26559889e+00 -7.70322680e-01 -1.25193310e+00 -4.60954346e-02
2.52338767e-01 -6.04729474e-01 1.01112521e+00 -1.50823176e+00
-1.55002260e+00 1.50739408e+00 -1.36630991e-02 -2.58259177e-01
2.63137162e-01 1.11340784e-01 -3.94852340e-01 5.98905683e-01
-1.28385514e-01 1.07005131e+00 1.22989786e+00 -1.28439116e+00
-3.64381433e-01 -5.84917590e-02 2.79723942e-01 -3.16892304e-02
1.69690251e-01 -2.81178147e-01 -2.16110740e-02 -1.06946158e+00
2.14943960e-02 -8.49978387e-01 8.64672065e-02 8.54809999e-01
-6.06474578e-01 3.61198410e-02 7.91535437e-01 -5.22267997e-01
5.25825143e-01 -2.27515340e+00 1.10747173e-01 -1.02458850e-01
5.59192359e-01 -2.56749958e-01 -2.92328119e-01 3.92739624e-01
2.35865965e-01 2.51303881e-01 -3.78501713e-02 -8.45524594e-02
8.09760839e-02 4.84180093e-01 -4.28279042e-01 7.81687737e-01
5.65544724e-01 1.16907799e+00 -8.93248439e-01 -7.32256413e-01
5.17889977e-01 2.83167362e-01 -3.35961878e-01 6.09415434e-02
-4.01053101e-01 2.81340003e-01 -2.94521749e-01 1.26192307e+00
5.77718437e-01 -1.56886116e-01 -1.13747127e-01 -5.50208688e-01
7.98369572e-03 -2.16768011e-01 -4.36209559e-01 1.39278471e+00
-3.29947650e-01 1.05509961e+00 -2.22971052e-01 -5.18175662e-01
1.02119160e+00 8.70549306e-03 2.60627598e-01 -9.66027796e-01
1.16419367e-01 -3.61503989e-01 -1.84034944e-01 -6.19068086e-01
4.66585398e-01 -4.67178345e-01 -4.31140363e-01 6.33665323e-01
2.19468296e-01 -6.17063582e-01 5.29353395e-02 -1.91615582e-01
7.76157081e-01 1.95407465e-01 2.84176767e-01 -1.02130020e+00
1.03081055e-01 1.50118083e-01 -1.16065629e-02 7.68716574e-01
-1.80638865e-01 7.97832787e-01 6.39312565e-01 -1.40486455e+00
-1.47303975e+00 -1.37945509e+00 -3.38084817e-01 1.32794499e+00
3.94691557e-01 -1.73800103e-02 -5.78900993e-01 -2.72999376e-01
3.03733200e-01 3.88587654e-01 -1.51019430e+00 -1.11868948e-01
1.87746719e-01 -2.50337839e-01 6.10097766e-01 6.06068254e-01
6.04642630e-01 -1.06810641e+00 -9.66850340e-01 -3.70507360e-01
3.93258557e-02 -9.43425655e-01 -4.43219811e-01 4.70229954e-01
-4.03634995e-01 -1.04334140e+00 -7.13316619e-01 -6.86114669e-01
8.11603427e-01 3.80531073e-01 1.54666400e+00 5.06906174e-02
-7.81890154e-01 5.29857337e-01 -5.45636892e-01 -5.81243113e-02
-2.35079840e-01 -4.00720924e-01 9.28631425e-02 2.10583508e-01
1.59080476e-01 -8.76268521e-02 -2.32593343e-01 5.01864016e-01
-1.07924724e+00 3.69722933e-01 9.51832309e-02 9.61091638e-01
7.52490699e-01 7.83132985e-02 -3.33555192e-02 -1.09035766e+00
4.08993691e-01 -1.94118187e-01 -5.61324656e-01 5.66350222e-01
1.02964632e-01 1.85324267e-01 5.33762753e-01 -6.71506345e-01
-1.04947698e+00 -5.77778220e-02 2.20521212e-01 -5.31516552e-01
-4.48429078e-01 1.43831715e-01 4.27952856e-01 -3.56713891e-01
9.49780226e-01 2.17739612e-01 -1.09809292e-02 -2.57914573e-01
4.43877816e-01 3.36801082e-01 6.40902519e-01 -1.06222320e+00
6.99668944e-01 9.53125775e-01 1.23912551e-01 -1.20016015e+00
-1.08801329e+00 7.32746965e-04 -9.73083138e-01 -3.40091735e-01
1.11629641e+00 -7.63171732e-01 -8.40569615e-01 5.92553318e-01
-6.22574329e-01 -7.63690293e-01 -4.74320769e-01 -1.69574201e-01
-1.01986027e+00 1.65465802e-01 -8.57246041e-01 -6.57468379e-01
-1.87352434e-01 -8.01033258e-01 1.52835917e+00 2.03110024e-01
-4.38499153e-01 -1.15985739e+00 -2.11021140e-01 -1.33199543e-01
4.58010584e-01 6.67044759e-01 1.18359792e+00 2.09155306e-01
-2.44200036e-01 4.88375351e-02 -6.53154135e-01 1.66886121e-01
3.01119328e-01 4.37082499e-01 -1.25754368e+00 -2.75313687e-02
-3.82617265e-01 -8.83513629e-01 8.16617191e-01 4.56778854e-01
1.54563236e+00 -3.13566387e-01 2.10210413e-01 9.40612793e-01
1.45584750e+00 -1.07069895e-01 5.52224457e-01 2.32594609e-02
4.00589406e-01 9.80460048e-01 3.28930050e-01 5.52180707e-01
-3.03198732e-02 5.32624006e-01 1.31665781e-01 -1.87848777e-01
-6.29104137e-01 -4.47027236e-01 -1.00820184e-01 3.44982445e-01
-1.75977468e-01 -6.48136809e-02 -9.55845177e-01 2.72375822e-01
-1.30609775e+00 -1.13943183e+00 1.43996269e-01 1.49269807e+00
7.45713949e-01 9.26463008e-02 1.41093001e-01 -4.64765072e-01
4.47611809e-01 5.56989789e-01 -6.17229044e-01 -6.74076200e-01
-6.04200661e-01 1.65149078e-01 4.56613779e-01 7.97096118e-02
-1.35970211e+00 1.14403069e+00 7.86487579e+00 8.20469081e-01
-1.06522775e+00 -2.36543611e-01 1.02482140e+00 1.15265198e-01
-1.69819877e-01 -2.09409535e-01 -2.01975808e-01 2.91640580e-01
3.78558636e-01 -3.84225816e-01 7.90588319e-01 9.49835718e-01
-2.94961214e-01 -3.03328961e-01 -1.25999248e+00 1.07345116e+00
2.66383011e-02 -1.49947321e+00 4.55048680e-01 -2.44898021e-01
8.96116912e-01 -4.02490705e-01 3.60309482e-01 7.00309128e-02
4.09734428e-01 -1.37400615e+00 1.09222007e+00 8.86978984e-01
1.51824582e+00 -2.89339572e-01 4.31260079e-01 -1.03052214e-01
-1.35333407e+00 5.87048568e-02 -1.07584441e+00 -2.06266344e-01
-3.55965495e-01 1.62534252e-01 -2.47746348e-01 1.87151015e-01
9.20826077e-01 1.08506036e+00 -9.50173736e-01 6.57260954e-01
-2.18058545e-02 4.52769667e-01 -7.49096349e-02 -1.91267729e-01
2.76192844e-01 9.04591084e-02 8.04196671e-02 1.21135569e+00
-1.72712579e-01 1.46942556e-01 1.19672582e-01 1.22647059e+00
6.57330677e-02 -2.58431107e-01 -7.28736937e-01 -4.66035783e-01
1.62361994e-01 1.07076526e+00 -9.39546347e-01 -6.15493715e-01
-1.96293414e-01 8.38938236e-01 2.81626314e-01 4.88523543e-01
-6.96436226e-01 -1.94333419e-01 8.63657892e-01 1.04323491e-01
2.14779317e-01 -1.39282107e-01 -2.05658570e-01 -1.16817725e+00
-5.06461263e-01 -5.85996330e-01 -3.99660990e-02 -1.48612869e+00
-1.79839504e+00 9.38828111e-01 1.01091802e-01 -1.50805604e+00
7.37644732e-02 -7.69464672e-01 -3.82909060e-01 6.65608704e-01
-8.98151457e-01 -1.41640246e+00 -5.47060549e-01 4.84205186e-01
7.70826757e-01 -1.62423223e-01 1.22203994e+00 -1.18714929e-01
-8.92290175e-02 2.02744380e-01 2.56467443e-02 2.11591914e-01
5.33046186e-01 -1.29421782e+00 6.55862570e-01 5.80136739e-02
1.68625742e-01 3.75796348e-01 7.79357910e-01 -5.57320535e-01
-1.59540224e+00 -1.01826668e+00 2.10026130e-01 -4.38605636e-01
7.35197425e-01 -5.63973784e-01 -7.71915317e-01 3.32386434e-01
-9.12924006e-04 5.41784108e-01 4.76492792e-01 1.67009071e-01
-9.75269437e-01 -1.28923595e-01 -1.07301807e+00 6.53619885e-01
8.10471952e-01 -9.93803144e-01 -6.49016559e-01 3.05556297e-01
2.45476231e-01 -3.88071775e-01 -9.69811320e-01 1.79092482e-01
1.05006349e+00 -7.72317529e-01 8.38364363e-01 -1.03469920e+00
1.10617495e+00 1.42255783e-01 -3.05318981e-01 -1.32189715e+00
-6.23603642e-01 -1.15226053e-01 2.33465329e-01 6.95887089e-01
-2.96757556e-02 -2.60796160e-01 5.80477536e-01 4.45801735e-01
2.65093856e-02 -6.28635347e-01 -8.05338979e-01 -6.80388451e-01
9.17320251e-02 -1.01517782e-01 5.87840140e-01 7.66299009e-01
-1.62835464e-01 -1.75585136e-01 -7.16814280e-01 -4.57976937e-01
6.94313765e-01 4.81078804e-01 6.76850915e-01 -1.34178638e+00
-6.61145225e-02 -6.11704409e-01 -5.94624996e-01 -8.48123252e-01
1.71582073e-01 -6.69433355e-01 1.84358805e-01 -1.20898104e+00
3.92460972e-01 -3.68819356e-01 3.85056846e-02 5.98447084e-01
3.93613726e-01 6.79562509e-01 -2.45384779e-02 9.72943082e-02
-9.25790131e-01 5.13231814e-01 1.57888019e+00 -6.82514071e-01
5.24420619e-01 -5.38470447e-01 -6.22206628e-01 7.92437792e-01
2.71481842e-01 -2.58063376e-01 -1.66198745e-01 -5.82356453e-01
-1.67922787e-02 3.00302148e-01 7.12094009e-01 -8.39625657e-01
-2.05531016e-01 -4.89554703e-01 8.58803332e-01 -7.89972395e-02
5.18851221e-01 -7.43345499e-01 1.23427480e-01 2.13006511e-01
-6.53174996e-01 -1.09038979e-01 4.45693284e-01 5.65523684e-01
-3.18212330e-01 9.10418555e-02 9.94132161e-01 -2.84624934e-01
-1.19311523e+00 4.01875287e-01 -6.91815138e-01 7.67514557e-02
7.53555477e-01 -2.88402528e-01 -1.13936234e+00 -3.52033585e-01
-7.21691847e-01 -2.32009903e-01 1.08460748e+00 6.48334146e-01
7.43229628e-01 -1.47993672e+00 -4.55638736e-01 5.24455845e-01
6.69117153e-01 -4.67629820e-01 6.74043775e-01 3.28906327e-01
-1.22482157e+00 9.48351398e-02 -9.73604321e-01 -6.81499124e-01
-9.64272261e-01 7.62976766e-01 4.22507614e-01 5.85194409e-01
-9.84359264e-01 9.72585678e-01 7.76581585e-01 1.10861063e-01
1.13284580e-01 -3.82963389e-01 5.83961606e-02 -1.08642153e-01
7.23426759e-01 -1.34681456e-03 -4.78628755e-01 -7.81614065e-01
-2.40488186e-01 9.21904922e-01 1.51643410e-01 1.53344199e-01
1.27863240e+00 -9.03750807e-02 -2.36026257e-01 8.88767302e-01
1.23387313e+00 -2.87270397e-01 -1.75129712e+00 -2.04743341e-01
-1.12208001e-01 -7.00899541e-01 3.31941098e-02 -7.60675609e-01
-9.21098471e-01 9.85440195e-01 4.99904811e-01 4.03414726e-01
9.85152125e-01 2.26492703e-01 4.15015876e-01 5.16751945e-01
6.42909706e-01 -9.92811501e-01 2.87663519e-01 3.61289203e-01
9.21004295e-01 -1.33705497e+00 2.87597567e-01 -3.34293514e-01
-1.16821015e+00 1.57391250e+00 3.40289652e-01 -5.22280037e-01
9.44379866e-01 3.86094540e-01 1.09623574e-01 -6.90699399e-01
-1.26164734e+00 -3.81796300e-01 9.11288500e-01 7.89220810e-01
3.07499141e-01 4.41043347e-01 4.50962573e-01 3.20980579e-01
-2.70879298e-01 -1.78101853e-01 3.96009356e-01 6.24611616e-01
-4.64617491e-01 -2.08989382e-01 -2.96268612e-02 6.75588429e-01
-1.33310780e-01 -3.46222147e-02 -7.00327873e-01 5.70880592e-01
1.73799887e-01 7.77878106e-01 3.78311008e-01 -6.94069624e-01
1.84456748e-03 -1.63635910e-01 7.83619285e-01 -6.77511811e-01
-2.15202123e-01 -7.03672618e-02 2.21692801e-01 -6.03548229e-01
-3.95585179e-01 -4.19104695e-01 -4.33394462e-01 -6.06671512e-01
2.83276707e-01 -3.59484971e-01 2.92167008e-01 8.35093498e-01
-2.69449204e-01 4.80488479e-01 3.59069794e-01 -1.00828910e+00
-1.81200102e-01 -8.52422178e-01 -1.13593507e+00 7.89095044e-01
4.11988378e-01 -1.00822198e+00 -4.03399378e-01 5.97229362e-01]
|
[10.288725852966309, -0.13394474983215332]
|
ac4f20b6-0443-44fc-8a74-96aa383314ec
|
multi-resolution-factor-graph-based-stereo
|
2202.01309
| null |
https://arxiv.org/abs/2202.01309v1
|
https://arxiv.org/pdf/2202.01309v1.pdf
|
Multi-Resolution Factor Graph Based Stereo Correspondence Algorithm
|
A dense depth-map of a scene at an arbitrary view orientation can be estimated from dense view correspondences among multiple lower-dimensional views of the scene. These low-dimensional view correspondences are dependent on the geometrical relationship among the views and the scene. Determining dense view correspondences is difficult in part due to presence of homogeneous regions in the scene and due to presence of occluded regions and illumination differences among the views. We present a new multi-resolution factor graph-based stereo matching algorithm (MR-FGS) that utilizes both intra- and inter-resolution dependencies among the views as well as among the disparity estimates. The proposed framework allows exchange of information among multiple resolutions of the correspondence problem and is useful for handling larger homogeneous regions in a scene. The MR-FGS algorithm was evaluated qualitatively and quantitatively using stereo pairs in the Middlebury stereo benchmark dataset based on commonly used performance measures. When compared to a recently developed factor graph model (FGS), the MR-FGS algorithm provided more accurate disparity estimates without requiring the commonly used post-processing procedure known as the left-right consistency check. The multi-resolution dependency constraint within the factor-graph model significantly improved contrast along depth boundaries in the MR-FGS generated disparity maps.
|
['Madhusudhanan Balasubramanian', 'Hanieh Shabanian']
|
2022-02-02
| null | null | null | null |
['stereo-matching-1']
|
['computer-vision']
|
[ 2.60775954e-01 -3.68064076e-01 2.65436292e-01 -3.95217031e-01
-4.82487589e-01 -5.74621618e-01 4.65733796e-01 1.18053705e-01
-2.44909421e-01 5.10037899e-01 3.37499201e-01 2.56813914e-01
-2.50452787e-01 -9.28035498e-01 -5.04286826e-01 -5.34070551e-01
4.14016664e-01 4.19877082e-01 7.67215431e-01 -1.38088912e-01
6.33591533e-01 6.86699450e-01 -1.86711955e+00 4.24633950e-01
7.42934704e-01 6.40452802e-01 6.03555143e-01 6.93367004e-01
2.95779295e-02 4.03885186e-01 7.03062536e-03 -3.25770169e-01
6.45194888e-01 -4.02643144e-01 -5.46957552e-01 4.04120892e-01
1.13443959e+00 -4.78245527e-01 -1.99549675e-01 1.08740795e+00
3.42828244e-01 7.01861233e-02 3.36387873e-01 -9.24821675e-01
1.86571300e-01 -3.70886564e-01 -1.04901147e+00 3.05110633e-01
7.99586415e-01 -2.88637012e-01 7.74861753e-01 -1.04963779e+00
1.03822374e+00 1.29281175e+00 3.45945001e-01 8.82809535e-02
-1.51415205e+00 -4.24012572e-01 4.25526984e-02 2.52869964e-01
-1.34416497e+00 -4.47343677e-01 9.01591241e-01 -8.18885446e-01
8.96006703e-01 2.14613512e-01 6.95004821e-01 3.37195307e-01
5.77691019e-01 5.76992594e-02 1.54655147e+00 -5.70955396e-01
1.73514277e-01 -2.13163495e-02 8.65031555e-02 7.23143458e-01
3.73999327e-01 4.49304491e-01 -8.27497840e-01 -2.53210306e-01
1.36359262e+00 -1.24423288e-01 -5.84691942e-01 -1.03326082e+00
-1.27738249e+00 6.28921092e-01 2.57334381e-01 2.86550403e-01
-3.28209966e-01 -3.74718904e-01 1.78959191e-01 1.31975144e-01
5.42229295e-01 1.63561821e-01 -1.17309920e-01 2.03713238e-01
-9.09762084e-01 1.00500375e-01 4.40915227e-01 9.25229549e-01
1.17859423e+00 -1.01362139e-01 3.54856074e-01 7.23038197e-01
1.79348573e-01 3.07660609e-01 7.51168206e-02 -1.25435424e+00
7.32509971e-01 6.44875705e-01 2.86214165e-02 -1.32207370e+00
-2.85418272e-01 -1.54749528e-01 -8.79404545e-01 6.23025358e-01
5.01384318e-01 3.46765280e-01 -6.84448123e-01 1.42322469e+00
5.23898959e-01 -7.88246319e-02 -1.65184021e-01 9.39885676e-01
6.20850980e-01 3.67824674e-01 -5.82289398e-01 -4.12607402e-01
1.02224720e+00 -6.44824624e-01 -5.89731634e-01 -2.94952959e-01
1.67563513e-01 -1.12168562e+00 5.61785400e-01 2.49564350e-01
-1.30054986e+00 -7.54742801e-01 -1.05365896e+00 -1.68106392e-01
4.01722677e-02 -2.69412696e-01 2.22672030e-01 3.74413639e-01
-1.30446398e+00 2.49704525e-01 -6.33769512e-01 -3.59691262e-01
-1.56079173e-01 3.68574560e-01 -9.10163105e-01 -6.15437388e-01
-5.55443466e-01 7.96737909e-01 2.21113816e-01 5.27454680e-03
-1.73359931e-01 -6.27024949e-01 -1.19555998e+00 -1.75189570e-01
1.31406844e-01 -1.06495249e+00 5.91878116e-01 -8.08530331e-01
-1.15386474e+00 1.17080653e+00 -7.15598345e-01 1.81818813e-01
7.93615401e-01 -7.85172656e-02 -1.26460686e-01 4.14848655e-01
2.41892368e-01 4.66533095e-01 6.31265879e-01 -1.55370498e+00
-7.33413100e-01 -7.85282552e-01 1.45554215e-01 7.16182232e-01
6.33215725e-01 -2.59063810e-01 -6.45695627e-01 -3.77497911e-01
8.58267605e-01 -6.95526600e-01 -2.08069205e-01 2.20802985e-02
-2.20160797e-01 5.68697810e-01 8.48593652e-01 -6.13715112e-01
9.27063048e-01 -2.04517126e+00 4.47886199e-01 3.48206669e-01
3.34690452e-01 -2.91576892e-01 4.41550426e-02 4.59374160e-01
-1.92994148e-01 -6.43823922e-01 -2.02388596e-02 -1.39044851e-01
-6.99239254e-01 1.73890397e-01 1.76737323e-01 8.35534990e-01
-4.46888417e-01 2.13359252e-01 -8.14436018e-01 -4.84305769e-01
8.54954123e-01 5.58011115e-01 -7.40626812e-01 3.09922814e-01
2.83773661e-01 8.56969476e-01 -1.76474780e-01 3.40331614e-01
1.11077154e+00 -3.29671726e-02 -7.06688222e-03 -6.54206157e-01
-4.72307533e-01 -4.49854992e-02 -1.58618891e+00 1.86686444e+00
-5.82005680e-01 6.55040503e-01 2.24022478e-01 -5.19644201e-01
8.24409187e-01 1.63807020e-01 5.69846570e-01 -8.73247921e-01
-1.21169157e-01 3.44130367e-01 -1.35073841e-01 -1.45874754e-01
6.91498280e-01 -2.82731146e-01 3.64918858e-01 1.22911833e-01
-3.60711813e-02 -2.85320163e-01 3.81472945e-01 9.84425098e-02
6.35692596e-01 6.09205803e-03 6.71964884e-01 -5.27330458e-01
9.35595870e-01 -2.90698647e-01 5.61092913e-01 3.67819518e-01
1.14881456e-01 1.09328580e+00 1.26149848e-01 -4.93861794e-01
-1.20996928e+00 -1.27329695e+00 -3.33628029e-01 3.34394693e-01
6.36813283e-01 -3.26532692e-01 -4.42365110e-01 -1.34238854e-01
-1.17874667e-01 2.53077567e-01 -3.53380024e-01 3.91861379e-01
-6.36034429e-01 -2.80577928e-01 -4.40759003e-01 2.14516610e-01
7.16668308e-01 -3.56147110e-01 -7.47592449e-01 1.23217419e-01
-4.66464013e-01 -1.50020063e+00 -4.78963047e-01 -1.22931369e-01
-1.15858579e+00 -1.31616461e+00 -7.29981303e-01 -5.71713865e-01
9.50679064e-01 7.67763436e-01 1.25241578e+00 -2.39682198e-01
-2.62530476e-01 5.14010489e-01 -3.13370787e-02 4.01125640e-01
-3.44684958e-01 -5.40099502e-01 -1.34705827e-01 1.77355334e-01
6.52094930e-02 -7.57972300e-01 -7.74117529e-01 8.18379164e-01
-7.85565913e-01 6.90725863e-01 1.70537114e-01 8.32058668e-01
7.98001647e-01 -5.03746085e-02 -1.56744644e-01 -8.97364080e-01
-4.02995721e-02 5.42703234e-02 -1.13422763e+00 7.99333602e-02
-3.45132113e-01 -2.67848419e-03 2.29694426e-01 2.56814837e-01
-1.19168222e+00 2.97558874e-01 5.99225648e-02 -4.87767100e-01
-5.65640703e-02 -9.17395502e-02 -3.22018445e-01 -3.10638249e-01
3.03788960e-01 6.76966012e-02 -2.00272039e-01 -3.36321592e-01
1.09907463e-01 2.18214259e-01 5.70819259e-01 -3.42076212e-01
7.26567090e-01 8.86499465e-01 4.89125371e-01 -1.00839829e+00
-6.48564994e-01 -9.01731193e-01 -1.26666379e+00 -4.36312973e-01
1.05382669e+00 -1.07562613e+00 -2.93635875e-01 5.75002968e-01
-1.24292839e+00 1.21160053e-01 1.89923700e-02 7.29589760e-01
-7.17898250e-01 6.93579853e-01 -4.40296471e-01 -4.59587216e-01
1.04686156e-01 -1.32991278e+00 1.25416565e+00 1.52739018e-01
-7.74985850e-02 -1.18491662e+00 1.66480839e-01 6.34203255e-01
1.71992227e-01 4.38748419e-01 9.97576833e-01 4.30760473e-01
-9.35702443e-01 1.16452426e-01 -3.58867288e-01 2.77663350e-01
3.47044557e-01 -6.03116713e-02 -9.42551076e-01 -3.62269759e-01
2.69005865e-01 2.76831031e-01 4.78250533e-01 7.77596176e-01
5.22559762e-01 2.57668197e-01 -9.28751901e-02 7.05118060e-01
2.06196022e+00 1.73468083e-01 7.36905575e-01 4.35687363e-01
9.26611245e-01 9.19696152e-01 6.78737581e-01 3.54882985e-01
2.80949295e-01 1.11066902e+00 3.22156727e-01 -3.76732290e-01
-1.51123255e-01 -2.36955613e-01 -2.68425215e-02 8.19407165e-01
-3.38446021e-01 -4.26198430e-02 -7.69707739e-01 2.36484155e-01
-1.61588264e+00 -9.65023458e-01 -6.23382092e-01 2.64012599e+00
2.53555208e-01 -2.77505256e-02 -1.76295951e-01 3.40589106e-01
1.01390743e+00 2.70179391e-01 -3.41541559e-01 -4.92472410e-01
-2.51253426e-01 -2.54283808e-02 6.20359480e-01 1.14530337e+00
-7.25369394e-01 4.67122257e-01 5.87781334e+00 4.16809827e-01
-9.15755332e-01 -1.42068028e-01 5.39565563e-01 1.06085218e-01
-3.12908739e-01 5.42488992e-02 -8.13608110e-01 1.59499750e-01
2.22432390e-01 -3.42131764e-01 1.70312524e-01 4.26039845e-01
2.87915647e-01 -8.83726537e-01 -1.39053082e+00 1.40867972e+00
2.63297409e-01 -1.23561025e+00 7.80803189e-02 3.82807940e-01
1.08051789e+00 -1.96372960e-02 -1.74487859e-01 -6.40625000e-01
-1.21476635e-01 -6.03055239e-01 4.93024141e-01 2.66314238e-01
9.17417824e-01 -7.61734903e-01 6.12096846e-01 1.61335662e-01
-1.60208774e+00 2.16318533e-01 -3.58277410e-01 5.93453757e-02
4.22784746e-01 7.52658844e-01 -1.95351049e-01 1.00692427e+00
5.76830089e-01 8.22854817e-01 -5.02327263e-01 1.01175892e+00
2.20509559e-01 -4.48326230e-01 -1.78752288e-01 1.02922320e+00
4.64451760e-02 -7.10921109e-01 6.55293405e-01 8.90541196e-01
3.59346122e-01 1.64553791e-01 -1.64288171e-02 5.78283966e-01
4.67486173e-01 1.12214843e-02 -8.74492288e-01 7.95970201e-01
8.84935856e-02 1.04029500e+00 -8.67346466e-01 -2.32309490e-01
-7.85984516e-01 1.11997461e+00 1.03161626e-01 4.36077565e-01
-3.33742201e-01 1.96508348e-01 5.79475999e-01 7.90308416e-01
2.25063875e-01 -3.02160859e-01 -2.54319251e-01 -1.33059621e+00
1.77481398e-01 -7.18814135e-01 3.52983057e-01 -1.00866437e+00
-1.09803414e+00 6.99467719e-01 3.02025527e-01 -1.67949557e+00
-4.55093652e-01 -3.15445542e-01 -2.21146241e-01 1.25940585e+00
-1.57521307e+00 -7.10353255e-01 -6.08423889e-01 9.75682914e-01
7.89898694e-01 1.08772650e-01 6.93500519e-01 3.85730684e-01
2.57471874e-02 -2.97983103e-02 5.71915396e-02 -2.50233710e-01
6.36049271e-01 -1.02009344e+00 8.27138498e-02 1.00929475e+00
-1.04907669e-01 4.21728879e-01 6.28583014e-01 -5.05044401e-01
-1.09155273e+00 -6.76238298e-01 8.54127228e-01 -1.98443264e-01
1.12734891e-01 -2.93883532e-01 -8.90259445e-01 4.59022284e-01
2.43574917e-01 1.04297705e-01 4.26527590e-01 -1.90944955e-01
-2.97497392e-01 -1.51475132e-01 -1.20850217e+00 1.70411929e-01
9.99199390e-01 -6.74371839e-01 -4.14925486e-01 -1.59685090e-01
-1.42226564e-02 -7.95674980e-01 -7.36095548e-01 4.69973594e-01
6.24973953e-01 -1.99707854e+00 1.08190835e+00 2.87113130e-01
5.83816051e-01 -5.58237493e-01 -4.77906585e-01 -1.23561931e+00
-3.54660451e-01 -1.96315467e-01 4.07134145e-01 1.02273655e+00
-9.86549631e-02 -6.42030895e-01 6.97435141e-01 6.30509615e-01
5.13255522e-02 -3.73325139e-01 -1.25479710e+00 -5.49190521e-01
-5.16785264e-01 2.48178300e-02 -2.18353122e-02 8.60104322e-01
-3.23932469e-01 3.18980217e-01 -1.21986620e-01 4.99480486e-01
1.13866150e+00 5.00276208e-01 8.69841814e-01 -1.31103766e+00
-4.88892466e-01 -1.78844467e-01 -9.01275158e-01 -1.10529888e+00
-1.42203361e-01 -4.94265229e-01 -2.46901944e-01 -1.70436871e+00
2.59040922e-01 -2.18168959e-01 1.49741977e-01 -4.51731950e-01
-2.25768127e-02 3.29053193e-01 1.70101926e-01 2.67324865e-01
-4.57842872e-02 2.60232419e-01 1.50775993e+00 4.20964092e-01
-1.54824078e-01 -6.18249085e-03 -8.32629949e-02 8.76924694e-01
3.25652570e-01 -1.50952414e-01 -6.09286606e-01 -5.46283126e-01
1.79418817e-01 6.07299328e-01 2.56966174e-01 -9.60956216e-01
2.49324739e-01 -3.05667836e-02 5.37141442e-01 -1.04454434e+00
7.50218987e-01 -1.10447419e+00 7.28823841e-01 3.22998166e-01
9.21070352e-02 5.63448846e-01 7.68182129e-02 6.52862251e-01
-5.80470681e-01 4.61839847e-02 1.22604442e+00 -3.08566689e-01
-9.16539669e-01 4.94358093e-02 -1.51813533e-02 -3.70251313e-02
1.00306451e+00 -8.72892082e-01 -2.37926260e-01 -4.15081203e-01
-5.53601563e-01 -1.27102122e-01 1.06556726e+00 2.54771829e-01
8.83446455e-01 -1.13082111e+00 -5.23239851e-01 7.07698762e-01
2.10464656e-01 1.87758327e-01 5.29788136e-01 8.52468550e-01
-8.99944901e-01 3.62423182e-01 -6.27896190e-01 -1.20468080e+00
-1.85606623e+00 3.09951723e-01 5.75144112e-01 -3.08280826e-01
-8.23590159e-01 4.60391492e-01 1.04294252e+00 7.01918546e-03
2.24006027e-02 -3.45104456e-01 -1.44895196e-01 -1.70708016e-01
4.57976609e-01 5.98126769e-01 2.36736670e-01 -1.20261157e+00
-2.62614638e-01 1.40723717e+00 7.26686604e-03 -3.50671887e-01
1.20537257e+00 -7.29076505e-01 -1.81334585e-01 5.09704411e-01
1.33048546e+00 2.65834987e-01 -1.16216099e+00 -2.46746480e-01
-4.73663092e-01 -1.27899313e+00 2.86662519e-01 -2.95374185e-01
-1.22800434e+00 7.54594803e-01 7.55242944e-01 -1.15414776e-01
1.27690685e+00 -2.29908481e-01 4.21328932e-01 -3.52289230e-01
8.25305402e-01 -7.98436582e-01 -1.50507569e-01 2.22648650e-01
8.50125492e-01 -1.19557190e+00 4.25992966e-01 -1.07772350e+00
-4.08614248e-01 1.29391325e+00 6.54422283e-01 -8.36864263e-02
6.32776439e-01 2.03562126e-01 4.21115980e-02 -3.27819020e-01
-4.56593663e-01 -7.37052336e-02 4.96931881e-01 6.60528004e-01
4.93958145e-01 -4.30324852e-01 -2.46081695e-01 -6.20477915e-01
1.19816291e-03 -2.08069056e-01 6.35256588e-01 7.44055092e-01
-3.01763117e-01 -1.09348452e+00 -5.66350520e-01 1.43345788e-01
-4.28557098e-02 5.58235012e-02 -9.73582119e-02 8.09737265e-01
1.26710117e-01 9.98060763e-01 3.10193330e-01 -1.74742550e-01
5.87193012e-01 -4.84061897e-01 8.82592797e-01 -6.44197166e-01
-3.72949898e-01 4.23178673e-01 8.91862586e-02 -1.04839718e+00
-1.03452837e+00 -6.93906486e-01 -8.41251373e-01 -3.90753359e-01
-1.57589391e-01 -2.18782827e-01 5.40053666e-01 6.57791138e-01
8.83266404e-02 8.99385512e-02 8.79426897e-01 -1.15708363e+00
2.97590613e-01 -3.30282956e-01 -8.28361392e-01 4.48950201e-01
3.63979906e-01 -7.70991802e-01 -5.25326014e-01 1.37502164e-01]
|
[9.247384071350098, -2.525109052658081]
|
aa708e29-eb3d-4404-98f7-023492adfb58
|
slgtformer-an-attention-based-approach-to
|
2212.10746
| null |
https://arxiv.org/abs/2212.10746v2
|
https://arxiv.org/pdf/2212.10746v2.pdf
|
SLGTformer: An Attention-Based Approach to Sign Language Recognition
|
Sign language is the preferred method of communication of deaf or mute people, but similar to any language, it is difficult to learn and represents a significant barrier for those who are hard of hearing or unable to speak. A person's entire frontal appearance dictates and conveys specific meaning. However, this frontal appearance can be quantified as a temporal sequence of human body pose, leading to Sign Language Recognition through the learning of spatiotemporal dynamics of skeleton keypoints. We propose a novel, attention-based approach to Sign Language Recognition exclusively built upon decoupled graph and temporal self-attention: the Sign Language Graph Time Transformer (SLGTformer). SLGTformer first deconstructs spatiotemporal pose sequences separately into spatial graphs and temporal windows. SLGTformer then leverages novel Learnable Graph Relative Positional Encodings (LGRPE) to guide spatial self-attention with the graph neighborhood context of the human skeleton. By modeling the temporal dimension as intra- and inter-window dynamics, we introduce Temporal Twin Self-Attention (TTSA) as the combination of locally-grouped temporal attention (LTA) and global sub-sampled temporal attention (GSTA). We demonstrate the effectiveness of SLGTformer on the World-Level American Sign Language (WLASL) dataset, achieving state-of-the-art performance with an ensemble-free approach on the keypoint modality. The code is available at https://github.com/neilsong/slt
|
['Yu Xiang', 'Neil Song']
|
2022-12-21
| null | null | null | null |
['sign-language-recognition']
|
['computer-vision']
|
[ 5.36873937e-02 -1.72168329e-01 -1.68249905e-01 -2.61415064e-01
-9.02545333e-01 -4.14987326e-01 6.19279623e-01 -4.82993484e-01
-4.28347141e-01 1.84518218e-01 9.72477257e-01 -1.62976846e-01
-3.85378748e-01 -4.34589177e-01 -5.76525390e-01 -6.67487025e-01
-2.03907207e-01 3.65016639e-01 4.51520421e-02 -2.26994723e-01
5.35119697e-02 5.33596575e-01 -1.61391389e+00 3.11952263e-01
7.32065916e-01 7.53240407e-01 9.60801244e-02 9.86825347e-01
-1.57821268e-01 8.41821253e-01 -1.77727506e-01 -2.64264792e-01
2.34580398e-01 -5.15834689e-01 -7.12502837e-01 -2.70355102e-02
1.20538330e+00 -4.95309234e-01 -9.15509999e-01 7.26753294e-01
9.10710812e-01 1.27707556e-01 5.26787341e-01 -9.95240569e-01
-7.14994490e-01 3.51698041e-01 -4.69683260e-01 3.25301796e-01
4.50706393e-01 5.55265248e-01 1.37642884e+00 -9.04255748e-01
8.55140269e-01 1.36359417e+00 6.31097555e-01 6.18724346e-01
-1.10644042e+00 -6.76557183e-01 6.79364383e-01 4.79268700e-01
-1.14813876e+00 -4.55172032e-01 8.61765802e-01 -6.12017810e-01
1.15478814e+00 1.50969461e-01 1.43912983e+00 1.14347935e+00
3.57406624e-02 1.09840417e+00 1.10205901e+00 -3.98365974e-01
-1.73055679e-01 -1.19486904e+00 2.54850775e-01 9.38651443e-01
-1.78356677e-01 2.77805060e-01 -1.08135700e+00 1.13642231e-01
7.83287287e-01 -1.28676027e-01 -5.59082866e-01 -2.78735071e-01
-1.20653665e+00 4.26027358e-01 5.02196550e-01 2.53052324e-01
-3.52658361e-01 6.47593200e-01 8.56364053e-03 2.10729659e-01
7.30354488e-02 -1.48554429e-01 -3.90895814e-01 -4.59633827e-01
-6.72842085e-01 2.00384945e-01 4.26517934e-01 7.19098330e-01
1.80399626e-01 -4.89537930e-03 -3.78745526e-01 7.67481923e-01
4.71643716e-01 7.41833687e-01 4.81656820e-01 -7.61441708e-01
4.51655120e-01 5.35907865e-01 -4.92637098e-01 -4.54353452e-01
-5.03440261e-01 -4.07553941e-01 -4.91350949e-01 5.47390878e-01
8.71634245e-01 1.74352258e-01 -1.72790980e+00 2.09130454e+00
1.49675414e-01 2.07958624e-01 -5.80627739e-01 1.11183465e+00
8.86264145e-01 5.04426397e-02 2.23712906e-01 2.17619568e-01
1.40453911e+00 -7.80143321e-01 -4.61143672e-01 -1.40090793e-01
3.25982302e-01 -3.90076399e-01 1.32793033e+00 4.22272056e-01
-1.01917124e+00 -1.00511625e-01 -4.63781089e-01 -4.88884717e-01
-2.38688722e-01 -1.55310100e-02 5.70396841e-01 3.04478616e-01
-1.13283587e+00 2.03603849e-01 -1.18284035e+00 -6.42887890e-01
6.28988147e-01 4.34272736e-01 -5.83476305e-01 -1.31686792e-01
-7.27756619e-01 7.42720008e-01 -3.20631295e-01 2.41118968e-01
-4.47177976e-01 -9.06555593e-01 -9.82014537e-01 -3.69863808e-01
2.42788568e-02 -1.00002873e+00 1.16740751e+00 -5.34275532e-01
-1.51970482e+00 1.09847760e+00 -5.86177111e-01 -2.11146966e-01
8.32493722e-01 -4.63757545e-01 -2.69170403e-01 2.20247820e-01
-9.38650221e-02 5.07833958e-01 1.02931142e+00 -9.12179649e-01
-5.86470485e-01 -5.44612885e-01 -3.06504786e-01 2.49881044e-01
4.49851081e-02 1.01277307e-01 -6.09597504e-01 -1.06877553e+00
6.20207429e-01 -7.77869046e-01 1.08359315e-01 5.33525169e-01
-3.27591151e-01 -2.06599280e-01 8.07677090e-01 -1.19130981e+00
1.20995891e+00 -2.05997849e+00 5.81042945e-01 3.83460671e-01
4.58889931e-01 1.03045568e-01 -3.89524013e-01 2.66686916e-01
-7.61044621e-02 -1.42462909e-01 -4.45331007e-01 -4.61543024e-01
5.69402333e-03 3.78298789e-01 -2.17550933e-01 6.22936368e-01
-2.16070786e-02 1.30936480e+00 -1.13840175e+00 -3.79958481e-01
4.01091635e-01 8.59108686e-01 -7.37096250e-01 -9.90038738e-02
-1.43245205e-01 7.45189190e-01 -2.94559270e-01 1.11102867e+00
1.34921625e-01 -3.04766893e-01 2.21790243e-02 -3.20193529e-01
-2.79843453e-02 2.62428969e-01 -7.74591744e-01 1.84089518e+00
-2.92648673e-01 8.39628100e-01 5.18694706e-02 -4.84000772e-01
3.52784514e-01 8.75380784e-02 6.74774289e-01 -8.67030501e-01
2.10749596e-01 2.26369843e-01 1.57419011e-01 -6.89669967e-01
-7.11407810e-02 -4.11291458e-02 2.48218626e-01 4.47514296e-01
1.76999703e-01 7.45772794e-02 8.62816349e-02 1.08134545e-01
1.44084680e+00 2.75900543e-01 2.94463515e-01 1.32304639e-01
2.46454716e-01 -5.16031504e-01 2.42333919e-01 5.81524909e-01
-3.77576828e-01 8.46295834e-01 1.88402116e-01 -2.07593620e-01
-7.30484962e-01 -1.59466887e+00 1.47278368e-01 1.16765058e+00
-2.55118072e-01 -3.73190671e-01 -2.81208187e-01 -5.90940714e-01
3.87408912e-01 3.27008963e-01 -7.83702910e-01 1.45025879e-01
-1.00977802e+00 -7.96064138e-02 5.72779775e-01 7.79328346e-01
2.72934377e-01 -1.19388783e+00 -6.99978888e-01 -4.66274507e-02
-1.33376211e-01 -9.88189161e-01 -1.11882460e+00 -3.35217714e-01
-4.70887363e-01 -1.17314672e+00 -1.11424303e+00 -7.55873859e-01
7.38626659e-01 -1.19053386e-01 6.83013320e-01 1.41064763e-01
-5.47187626e-01 1.05614185e+00 -4.20442641e-01 -2.33969186e-02
1.23987600e-01 -2.58229971e-01 -4.61062752e-02 2.69883931e-01
2.43129507e-02 -1.22866333e+00 -8.04344177e-01 -5.96244521e-02
-5.02764463e-01 1.42595753e-01 6.20527148e-01 1.10844612e+00
4.79038745e-01 -7.78702259e-01 -1.42256707e-01 -1.51946306e-01
4.15805012e-01 -2.36061905e-02 -4.28771496e-01 2.51811981e-01
-2.50199705e-01 2.46589556e-01 1.08957350e-01 -6.38740540e-01
-7.10000217e-01 2.11800430e-02 -1.56430542e-01 -5.38540840e-01
1.08059007e-03 4.03829545e-01 -1.38165489e-01 -4.77835506e-01
2.19212830e-01 5.46970129e-01 2.27628455e-01 -6.25607908e-01
6.03866756e-01 3.70907366e-01 9.66900349e-01 -6.57549262e-01
7.59936988e-01 7.76885569e-01 6.54157847e-02 -1.08808160e+00
-5.20603776e-01 -5.05847275e-01 -8.11184585e-01 -6.55212343e-01
7.81494975e-01 -5.00249267e-01 -9.80254829e-01 9.18755293e-01
-1.02699316e+00 -8.72449875e-01 -4.06105787e-01 5.16502619e-01
-6.96185589e-01 5.56678891e-01 -3.45345676e-01 -8.30624640e-01
-2.60377586e-01 -7.80664206e-01 1.31795061e+00 -1.65202945e-01
-4.67384726e-01 -7.14698434e-01 6.07555546e-02 4.20809001e-01
8.97667706e-02 4.64951754e-01 8.70556056e-01 -1.28128961e-01
-9.22955334e-01 -1.94049180e-02 -2.41901919e-01 1.48733318e-01
1.01643533e-01 -2.62949973e-01 -8.53060126e-01 -2.84427762e-01
-8.10682952e-01 -1.62096024e-01 1.16681337e+00 7.57048428e-01
9.15183961e-01 -3.22278559e-01 -6.35652756e-03 1.11966980e+00
9.16278064e-01 -4.21953574e-02 4.85980004e-01 5.29973842e-02
9.79326844e-01 5.79408050e-01 -7.97401071e-02 4.06058878e-01
7.50847876e-01 8.29390824e-01 3.97243761e-02 6.59160316e-02
-1.05521488e+00 -6.70404613e-01 4.35320467e-01 7.84230828e-01
-7.64432549e-01 2.39442606e-02 -1.24389684e+00 7.84168363e-01
-1.82406259e+00 -1.07193577e+00 7.16202185e-02 2.06645846e+00
6.87492609e-01 -2.67596185e-01 1.71745792e-01 4.28160846e-01
2.94276565e-01 2.57177740e-01 -6.85174584e-01 1.54938295e-01
-4.71175224e-01 5.66986442e-01 4.51283157e-01 9.35901999e-01
-9.28322315e-01 1.04589128e+00 5.60133314e+00 4.64525223e-01
-1.31456113e+00 1.11147217e-01 -1.62972540e-01 -4.09465194e-01
-3.77413243e-01 -1.70979306e-01 -3.82177740e-01 3.25900078e-01
2.24977717e-01 -6.55879527e-02 6.49031699e-01 1.87527955e-01
3.15546036e-01 1.94524080e-01 -1.02733135e+00 1.11023021e+00
1.24159612e-01 -1.06550002e+00 2.50608772e-01 3.17737073e-01
2.81862110e-01 4.49233651e-01 1.82005391e-01 -1.58623308e-01
3.35413069e-01 -1.06154835e+00 1.04653692e+00 8.52898777e-01
1.21347749e+00 -9.02534351e-02 5.75805306e-02 -1.25215724e-02
-1.63234270e+00 -3.94455940e-01 7.09247351e-01 -1.08770996e-01
4.38198447e-01 -4.37389128e-02 -4.22576189e-01 1.67772472e-01
8.37506056e-01 1.06114125e+00 -3.27132672e-01 1.22327793e+00
-6.51210964e-01 6.70616388e-01 -6.54107034e-01 1.28098667e-01
3.28591466e-02 -7.73537764e-03 1.25098717e+00 1.14505148e+00
3.11331034e-01 4.44355398e-01 -1.34262398e-01 5.90011358e-01
2.59356618e-01 3.92358042e-02 -3.22903961e-01 -7.67738149e-02
1.21110961e-01 4.69306111e-01 -3.45878690e-01 2.14774944e-02
-4.60740715e-01 1.10596585e+00 2.90897846e-01 8.64710927e-01
-2.58543640e-01 -1.26537615e-02 1.00665545e+00 3.45736206e-01
4.02029097e-01 -5.92880249e-01 -3.81283730e-01 -1.20054102e+00
4.09677327e-01 -7.19514906e-01 8.28909338e-01 -7.55183041e-01
-1.30532312e+00 2.73548573e-01 -2.28591129e-01 -1.26697826e+00
-2.33617634e-01 -8.50066543e-01 -5.55742800e-01 9.39414203e-01
-1.47083068e+00 -1.95956409e+00 -5.10629952e-01 1.04715383e+00
4.21434194e-01 -3.00524831e-02 7.01172948e-01 1.84971467e-01
-7.21563250e-02 1.00478184e+00 -4.36205506e-01 4.88645703e-01
4.89842564e-01 -1.16714573e+00 4.89294022e-01 9.70909834e-01
4.06106144e-01 5.94076574e-01 4.40714121e-01 -7.95481920e-01
-1.47354281e+00 -7.16321111e-01 1.17166162e+00 -6.18266523e-01
8.81767929e-01 -3.14931452e-01 -6.05689287e-01 7.25870907e-01
-3.65063280e-01 3.11468929e-01 3.02023470e-01 8.87628421e-02
-7.96971440e-01 3.17927934e-02 -7.35316217e-01 8.15400481e-01
1.94568610e+00 -9.77204084e-01 -6.54034555e-01 2.05538914e-01
4.56556946e-01 -6.14418209e-01 -6.74835622e-01 3.78291965e-01
1.36677992e+00 -7.57945895e-01 1.02869415e+00 -6.50952816e-01
1.58916056e-01 -2.42809311e-01 -2.78301477e-01 -9.51750994e-01
-3.06769848e-01 -9.16687012e-01 -4.15857285e-01 6.38059020e-01
1.42636284e-01 -8.79713655e-01 7.58642256e-01 5.01909733e-01
-1.78342178e-01 -7.50906050e-01 -1.55224872e+00 -9.02736783e-01
-3.36171612e-02 -9.35549319e-01 3.76753330e-01 7.00979054e-01
6.60063475e-02 -2.51165003e-01 -3.23040068e-01 2.07360864e-01
8.99529219e-01 2.68862154e-02 7.05199540e-01 -1.01181889e+00
-4.06905860e-01 -1.00467467e+00 -9.27746713e-01 -1.14907408e+00
8.73134509e-02 -1.11710620e+00 -8.58267695e-02 -1.75681865e+00
-1.54497236e-01 1.79010183e-01 -3.20321888e-01 7.76353300e-01
2.02661425e-01 7.26705119e-02 4.34696376e-01 8.05647895e-02
3.00838724e-02 5.09792864e-01 1.55875087e+00 -2.93605089e-01
-2.52142191e-01 -7.31276795e-02 -1.57273084e-01 6.79072678e-01
3.82530123e-01 9.07355472e-02 -1.35199398e-01 -5.25497258e-01
-1.11880526e-01 1.13289361e-03 9.52395380e-01 -8.28116059e-01
5.62165201e-01 -8.82697478e-02 1.60870012e-02 -6.69995368e-01
4.79912966e-01 -5.16741335e-01 -2.18107492e-01 5.10261178e-01
-2.13287637e-01 -2.28827327e-01 5.37962504e-02 4.63651448e-01
-8.40158090e-02 8.77782643e-01 6.63935959e-01 7.40648136e-02
-9.27038133e-01 7.91930556e-01 -1.41913831e-01 2.83160001e-01
3.67630154e-01 -5.84592283e-01 -1.17778108e-01 -6.48796618e-01
-1.10483909e+00 2.60570705e-01 1.79779142e-01 5.41795492e-01
9.24731135e-01 -1.31515920e+00 -7.45583713e-01 5.44009149e-01
3.92958730e-01 -2.62840271e-01 5.04010379e-01 1.11410201e+00
-3.05103928e-01 3.58436763e-01 -9.92964283e-02 -6.83904290e-01
-1.49481082e+00 -1.90465346e-01 5.50192773e-01 1.33147910e-01
-1.45875657e+00 1.30388582e+00 2.21985370e-01 -3.17185909e-01
5.29891074e-01 -8.09803724e-01 4.22575660e-02 -9.62014496e-02
4.00412232e-01 3.21039259e-01 -3.23385417e-01 -1.02818894e+00
-4.33977544e-01 1.30919099e+00 3.63605589e-01 -4.80229646e-01
1.25397944e+00 6.90989494e-02 -8.53172317e-02 3.04492593e-01
1.01981318e+00 1.99524358e-01 -1.41509974e+00 -5.15944242e-01
-2.07525641e-01 -6.21895134e-01 1.03138223e-01 -1.09340084e+00
-9.88426149e-01 9.87614870e-01 5.86445570e-01 -5.41325867e-01
1.22893345e+00 3.14977169e-01 9.12563622e-01 -8.32131226e-03
3.83938015e-01 -8.70235264e-01 7.37974271e-02 6.25351250e-01
1.60083258e+00 -9.48224723e-01 -2.89258271e-01 -2.38668963e-01
-1.81639224e-01 9.45841908e-01 2.26069331e-01 6.39973506e-02
9.68093812e-01 2.92410284e-01 2.88089275e-01 -2.28595465e-01
-3.92481476e-01 -7.91741133e-01 1.05059528e+00 7.22667217e-01
4.19082552e-01 2.42163256e-01 -2.03688473e-01 5.74580371e-01
-6.53798819e-01 3.49192619e-02 -2.40766838e-01 9.73413944e-01
-1.28485247e-01 -1.02634227e+00 -1.93878695e-01 5.60990691e-01
1.08968616e-01 -2.41821244e-01 -4.15815473e-01 8.09894919e-01
2.77037650e-01 4.20625061e-01 1.96451060e-02 -4.54585373e-01
5.49959302e-01 2.39696622e-01 9.65591073e-01 -3.35300267e-01
-2.02320680e-01 1.28122464e-01 9.55701321e-02 -9.50565040e-01
-2.39550099e-01 -1.11489379e+00 -1.42583799e+00 4.11076657e-02
4.14605290e-01 -6.53719604e-01 2.66147286e-01 1.05106676e+00
1.17277898e-01 4.53855276e-01 -2.26151228e-01 -8.72535169e-01
-2.10368261e-01 -9.21077549e-01 -5.35989344e-01 5.28460562e-01
9.59354103e-01 -8.50071430e-01 -3.13348621e-01 7.67694563e-02]
|
[9.189470291137695, -6.486324787139893]
|
d87df972-a513-42ba-8c3c-b7a411122a20
|
probing-sentiment-oriented-pre-training
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Feng_Probing_Sentiment-Oriented_Pre-Training_Inspired_by_Human_Sentiment_Perception_Mechanism_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Feng_Probing_Sentiment-Oriented_Pre-Training_Inspired_by_Human_Sentiment_Perception_Mechanism_CVPR_2023_paper.pdf
|
Probing Sentiment-Oriented Pre-Training Inspired by Human Sentiment Perception Mechanism
|
Pre-training of deep convolutional neural networks (DCNNs) plays a crucial role in the field of visual sentiment analysis (VSA). Most proposed methods employ the off-the-shelf backbones pre-trained on large-scale object classification datasets (i.e., ImageNet). While it boosts performance for a big margin against initializing model states from random, we argue that DCNNs simply pre-trained on ImageNet may excessively focus on recognizing objects, but failed to provide high-level concepts in terms of sentiment. To address this long-term overlooked problem, we propose a sentiment-oriented pre-training method that is built upon human visual sentiment perception (VSP) mechanism. Specifically, we factorize the process of VSP into three steps, namely stimuli taking, holistic organizing, and high-level perceiving. From imitating each VSP step, a total of three models are separately pre-trained via our devised sentiment-aware tasks that contribute to excavating sentiment-discriminated representations. Moreover, along with our elaborated multi-model amalgamation strategy, the prior knowledge learned from each perception step can be effectively transferred into a single target model, yielding substantial performance gains. Finally, we verify the superiorities of our proposed method over extensive experiments, covering mainstream VSA tasks from single-label learning (SLL), multi-label learning (MLL), to label distribution learning (LDL). Experiment results demonstrate that our proposed method leads to unanimous improvements in these downstream tasks. Our code is released on https://github.com/tinglyfeng/sentiment_pretraining
|
['Jufeng Yang', 'Jiaxuan Liu', 'Tinglei Feng']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['multi-label-learning']
|
['methodology']
|
[ 4.81596649e-01 2.31169220e-02 -6.87893480e-02 -5.83314598e-01
-5.23123682e-01 -6.80110455e-01 7.19995916e-01 8.39070603e-02
-4.62346435e-01 3.26615065e-01 5.70961833e-02 -4.63469148e-01
2.05218017e-01 -6.99890137e-01 -8.57748389e-01 -8.79653871e-01
4.44083422e-01 7.84281865e-02 -7.75863230e-02 -3.51584196e-01
3.05988580e-01 3.42007697e-01 -1.68899119e+00 6.16442084e-01
6.32498860e-01 1.35026217e+00 1.57841995e-01 4.51891512e-01
-3.07491869e-01 8.50299418e-01 -4.11166847e-01 -6.08923435e-01
1.01714768e-02 -2.79122114e-01 -7.02629328e-01 2.49585226e-01
3.83207351e-01 1.11003794e-01 3.68346006e-01 1.19109201e+00
4.42969620e-01 -9.98619795e-02 7.96765804e-01 -1.38027763e+00
-8.33348632e-01 2.14321941e-01 -6.71294928e-01 -6.25867397e-02
-8.24804157e-02 3.44987124e-01 1.00446153e+00 -1.12098050e+00
3.75762314e-01 1.26028836e+00 5.65933526e-01 6.04804099e-01
-1.04408514e+00 -7.01224804e-01 5.66866875e-01 3.19723822e-02
-9.58193958e-01 -4.03696597e-01 1.09108496e+00 -5.70426285e-01
5.29996216e-01 2.22241029e-01 7.20023394e-01 1.33203161e+00
-7.50332512e-03 9.48238134e-01 1.69442618e+00 -3.33566099e-01
2.91897714e-01 5.40975451e-01 1.77919731e-01 6.34748280e-01
2.48927116e-01 -6.95977062e-02 -5.72211623e-01 3.22922736e-01
4.37033772e-01 6.63829967e-02 -1.02539375e-01 -3.98205072e-01
-1.04624426e+00 6.31272078e-01 6.72558308e-01 1.54051393e-01
-3.41066986e-01 7.69310445e-02 5.37401199e-01 9.76276919e-02
4.65229481e-01 2.18755677e-01 -5.64740479e-01 3.22981596e-01
-6.63470089e-01 -2.26499513e-01 3.56535614e-01 8.13704491e-01
9.83187616e-01 8.43736529e-02 -2.74671465e-01 7.28915572e-01
3.49877179e-01 6.03548467e-01 5.77878356e-01 -4.43717033e-01
1.56511128e-01 8.32712889e-01 -7.78078958e-02 -1.16630673e+00
-4.31537807e-01 -6.41209602e-01 -9.55046237e-01 4.67483282e-01
2.47365773e-01 -7.68429711e-02 -1.11016309e+00 1.76265860e+00
3.36347193e-01 -5.25331087e-02 1.84011042e-01 9.45148766e-01
8.48761439e-01 5.83174050e-01 5.82159877e-01 -1.81244478e-01
1.59016585e+00 -1.21284354e+00 -4.06823993e-01 -4.25942481e-01
6.05134964e-01 -7.71511734e-01 1.35193110e+00 5.86449623e-01
-8.17323685e-01 -9.55028594e-01 -1.01109052e+00 -5.98433390e-02
-7.79032230e-01 3.28077823e-01 7.52619445e-01 8.11189711e-01
-1.15519357e+00 1.00829467e-01 -4.03707862e-01 -2.99865365e-01
7.60238051e-01 2.35009879e-01 -3.61499876e-01 3.87483984e-02
-8.42114985e-01 7.90388703e-01 3.34039092e-01 4.71226692e-01
-1.12672019e+00 -4.22011554e-01 -6.82079315e-01 -4.06320915e-02
3.47200930e-01 -6.75856888e-01 1.01691580e+00 -1.71470475e+00
-1.31935632e+00 1.15382826e+00 -2.79790431e-01 -1.10383138e-01
3.19264114e-01 -1.43672839e-01 -2.79769838e-01 1.62983984e-01
7.63195753e-03 7.98276126e-01 1.05997884e+00 -1.96360862e+00
-5.29065430e-01 -3.25836420e-01 3.21440578e-01 3.64525527e-01
-6.88192546e-01 -8.17816854e-02 -2.62912542e-01 -5.05289257e-01
-2.23140661e-02 -7.68689513e-01 -1.60123438e-01 6.71084151e-02
-4.35799807e-01 -2.44534343e-01 6.51578665e-01 -3.20904136e-01
8.93819332e-01 -2.15130353e+00 -7.04179928e-02 1.27209857e-01
1.80771127e-01 4.58500743e-01 -3.01333994e-01 1.72968909e-01
-3.07529151e-01 1.91785529e-01 -2.11099997e-01 -5.62024057e-01
5.50366528e-02 -4.86659445e-02 -3.56054842e-01 3.94890666e-01
4.00466770e-01 1.07205582e+00 -7.59683609e-01 -5.37079871e-01
2.89190233e-01 4.09248114e-01 -4.46488500e-01 2.49307841e-01
-3.35796386e-01 5.59266329e-01 -3.54847491e-01 9.64922130e-01
8.13341439e-01 -4.64514911e-01 1.77624896e-01 -5.69038510e-01
-6.19344786e-02 -8.41012001e-02 -9.39776838e-01 1.68336153e+00
-7.38241971e-01 2.25711748e-01 -4.57563326e-02 -1.36149037e+00
9.62651014e-01 1.06616400e-01 2.01083988e-01 -8.76601577e-01
5.64823270e-01 1.03440680e-01 -1.79303572e-01 -6.25080645e-01
3.02168757e-01 -3.91012579e-01 -1.43968552e-01 4.50815827e-01
3.31548542e-01 2.58589059e-01 9.06526204e-03 1.52219862e-01
4.29881394e-01 2.47311190e-01 2.83384919e-01 -3.47174615e-01
8.02031815e-01 1.15472458e-01 2.66621292e-01 4.81557757e-01
-3.44118059e-01 5.16328156e-01 6.54653311e-01 -3.92760009e-01
-9.17658627e-01 -7.75443733e-01 -6.42775595e-02 1.43019617e+00
4.07576650e-01 -4.15338911e-02 -6.85638905e-01 -8.51292610e-01
-2.12680012e-01 4.76879656e-01 -1.00757098e+00 -2.83320725e-01
-1.68579489e-01 -9.67710972e-01 2.80657113e-01 5.29037178e-01
5.63116312e-01 -1.53209662e+00 -5.12085915e-01 -1.49169043e-01
7.67090693e-02 -9.71322417e-01 -5.96254170e-02 3.97936106e-01
-5.40217757e-01 -9.47455585e-01 -6.55067086e-01 -1.01735795e+00
8.85982573e-01 4.27560359e-01 1.04315400e+00 2.44587839e-01
6.27592877e-02 2.79132903e-01 -4.70705479e-01 -6.02897108e-01
-1.43367633e-01 -1.05022125e-01 -1.91240132e-01 5.90406358e-01
4.82675910e-01 -4.80438173e-01 -9.34184849e-01 1.09040476e-01
-9.86144125e-01 2.99921006e-01 1.03753507e+00 7.23392248e-01
5.74459255e-01 -1.76832914e-01 8.52597058e-01 -1.11919856e+00
5.07500827e-01 -5.34212530e-01 -2.73736626e-01 3.39529067e-01
-6.40799224e-01 -2.12635234e-01 7.82689989e-01 -4.87956911e-01
-1.26202083e+00 1.35734364e-01 -2.79620469e-01 -3.72958422e-01
-5.84664941e-01 4.09300685e-01 -2.65424132e-01 -1.49038896e-01
5.48292935e-01 4.46605593e-01 4.14618440e-02 -1.15862437e-01
5.94548404e-01 6.71073914e-01 3.06719631e-01 -6.27698302e-01
6.42984688e-01 7.66632199e-01 -1.80063158e-01 -3.32493573e-01
-1.30415201e+00 -4.19969231e-01 -6.06876433e-01 -5.37927985e-01
1.08861804e+00 -1.11557162e+00 -1.01417589e+00 6.50421023e-01
-8.79717350e-01 -2.90578336e-01 -5.71798417e-04 -3.06290537e-02
-3.99360061e-01 1.74014941e-01 -2.86953539e-01 -9.21477795e-01
-4.13053274e-01 -1.02189434e+00 1.05910075e+00 5.13447344e-01
1.56653747e-01 -1.02318966e+00 -1.00187641e-02 5.53601444e-01
4.00984228e-01 1.53913289e-01 9.14264083e-01 -7.70844221e-01
-3.77607197e-01 1.94759201e-02 -6.36636257e-01 7.87580729e-01
-1.63461983e-01 -2.14962125e-01 -1.48786247e+00 -3.02323103e-01
2.26973332e-02 -8.51376176e-01 9.43757296e-01 2.28433937e-01
1.22379708e+00 -4.94896099e-02 -1.37340382e-01 5.87785184e-01
1.74458826e+00 2.06359506e-01 4.90577459e-01 5.37875593e-01
8.10200810e-01 7.54549086e-01 6.81894243e-01 2.72076190e-01
5.88933647e-01 1.54837817e-01 8.59631717e-01 -6.40631020e-01
-1.93880409e-01 -1.31336048e-01 3.90390486e-01 7.56774783e-01
-6.31968603e-02 -3.44272852e-01 -8.03096831e-01 5.92944026e-01
-1.76647496e+00 -6.26165867e-01 3.08098588e-02 1.76308095e+00
6.36885941e-01 3.31070095e-01 -6.31143674e-02 7.22339600e-02
5.36334276e-01 2.79487818e-01 -6.03874743e-01 -4.97116864e-01
-2.40766227e-01 1.93803549e-01 1.97494328e-01 1.77748621e-01
-1.25278473e+00 1.10299778e+00 5.10072851e+00 7.62512743e-01
-1.48665762e+00 2.24094823e-01 1.01187646e+00 1.74952194e-01
-4.95315701e-01 -3.59669589e-02 -6.59208059e-01 3.65695596e-01
6.18218899e-01 1.77053958e-01 1.01673521e-01 1.08147693e+00
7.21805990e-02 -3.73653881e-03 -6.97490275e-01 8.67823422e-01
3.11297476e-01 -1.14310348e+00 3.75787020e-01 -1.59885660e-01
8.49771261e-01 -1.92938790e-01 4.13392246e-01 5.29113889e-01
4.87702191e-02 -1.02712107e+00 9.51352179e-01 5.03677249e-01
8.77472222e-01 -6.90577865e-01 9.25728381e-01 1.79663047e-01
-1.05008817e+00 -2.45483756e-01 -4.24262762e-01 -8.78276005e-02
1.62067357e-02 5.84404171e-01 -4.82598126e-01 5.70827544e-01
6.48294389e-01 7.89354622e-01 -7.39948630e-01 5.74416101e-01
-3.29515249e-01 6.25808775e-01 2.13158444e-01 -1.36545226e-01
4.02908176e-01 -1.23132557e-01 -5.94867095e-02 1.19349563e+00
-4.45861071e-02 4.47623758e-03 7.57856816e-02 6.76155925e-01
-5.02059888e-03 2.00677231e-01 -4.49478745e-01 4.57891226e-02
6.15465939e-02 1.78977692e+00 -9.01467085e-01 -5.42592287e-01
-5.98111570e-01 7.64660478e-01 4.42566603e-01 5.59214294e-01
-8.67477179e-01 -1.30747065e-01 4.79292870e-01 -1.18415430e-01
4.04774725e-01 8.79034176e-02 -5.54917693e-01 -1.17018604e+00
-3.55944075e-02 -1.01197267e+00 1.74194828e-01 -9.67648745e-01
-1.31984317e+00 9.13139760e-01 -2.86984682e-01 -1.19406474e+00
1.95245788e-01 -1.05562413e+00 -6.04060650e-01 8.09591651e-01
-1.95298207e+00 -1.67716098e+00 -4.84553933e-01 7.63420522e-01
5.81459224e-01 1.28464475e-02 6.69161081e-01 2.93063194e-01
-5.77944994e-01 5.19970298e-01 -3.65421116e-01 6.60589314e-04
6.95688546e-01 -1.14372158e+00 -4.19142693e-02 9.30707455e-01
7.90613592e-02 6.10581458e-01 3.99584621e-01 -2.82319278e-01
-1.20233762e+00 -1.05314291e+00 6.33829832e-01 -4.17502433e-01
7.08162546e-01 -6.09890044e-01 -7.27047741e-01 4.92795199e-01
2.76576310e-01 1.97087433e-02 7.01918542e-01 2.22634152e-01
-4.83059853e-01 -2.45225310e-01 -7.91678011e-01 4.75112468e-01
8.98277700e-01 -5.74488699e-01 -4.13544953e-01 2.00319722e-01
4.74156082e-01 2.79586874e-02 -4.95933145e-01 5.92117250e-01
5.38157344e-01 -1.22099519e+00 9.76241589e-01 -5.96423745e-01
8.36857140e-01 -4.09756541e-01 -1.89910680e-01 -1.09275556e+00
-2.08599821e-01 -1.08532030e-02 2.45051041e-01 1.27538896e+00
3.40068460e-01 -5.64552844e-01 6.62373483e-01 1.29813910e-01
-3.15839320e-01 -9.94323373e-01 -3.98170412e-01 -2.73834467e-01
5.63823581e-02 -5.24274707e-01 3.95053029e-01 1.09264743e+00
-3.03633481e-01 5.19742250e-01 -3.23254645e-01 2.99547255e-01
5.45988977e-01 3.45338702e-01 7.94070959e-01 -1.00774419e+00
-1.53943181e-01 -5.57284117e-01 -1.38528377e-01 -9.46557581e-01
6.03650287e-02 -9.73964930e-01 1.62193879e-01 -1.52102458e+00
4.39617246e-01 -5.75588822e-01 -8.24706197e-01 6.05501771e-01
-3.29617947e-01 6.49054587e-01 2.61543304e-01 9.11006108e-02
-8.33278239e-01 5.86720765e-01 1.61026895e+00 -1.64746687e-01
2.05924898e-01 -1.88698828e-01 -1.23089921e+00 8.40961218e-01
8.43291581e-01 -4.12291437e-01 -6.13720357e-01 -3.49447548e-01
4.79252964e-01 -3.16472918e-01 6.30806684e-01 -8.30074072e-01
1.57553554e-01 -8.43786895e-02 3.78248721e-01 -4.40152049e-01
2.63314396e-01 -7.38742232e-01 -3.08211923e-01 2.64479727e-01
-2.90954083e-01 -1.65156409e-01 2.59896606e-01 4.47220087e-01
-3.91562939e-01 -1.01902291e-01 8.33601594e-01 -1.98077068e-01
-1.03265250e+00 1.10429861e-01 -1.37191832e-01 -1.58470079e-01
9.27428484e-01 -1.73025414e-01 -4.55868334e-01 -1.04060777e-01
-7.56520689e-01 6.42173961e-02 3.71150166e-01 4.21982050e-01
3.76801252e-01 -1.05022836e+00 -3.50878745e-01 2.80383140e-01
3.12469661e-01 -7.77067675e-04 5.62182784e-01 8.99500668e-01
-2.36634940e-01 2.83145458e-01 -4.41019863e-01 -7.04919577e-01
-9.82399285e-01 8.43693912e-01 6.59155697e-02 -1.52322575e-01
-1.85734957e-01 1.09151983e+00 7.38368571e-01 -4.96876627e-01
1.37051001e-01 -1.28077179e-01 -5.72298229e-01 3.12196851e-01
3.39216828e-01 -2.36845851e-01 3.12751578e-03 -7.74596870e-01
-3.74243349e-01 8.41674268e-01 -1.01848230e-01 1.38665855e-01
1.13754654e+00 -2.06245720e-01 -8.46071765e-02 3.97190779e-01
1.17817676e+00 -1.31987885e-01 -1.29960537e+00 1.11781051e-02
-3.17550421e-01 -9.86722186e-02 -6.93369433e-02 -1.04539311e+00
-1.14170575e+00 1.14332187e+00 6.00169182e-01 2.14754805e-01
1.52270401e+00 -6.97293598e-03 5.23305058e-01 2.26806238e-01
2.55423158e-01 -8.82265329e-01 5.18204391e-01 1.32883877e-01
7.08019912e-01 -1.58382523e+00 -1.80101395e-01 -2.55955964e-01
-1.14457214e+00 8.24488103e-01 7.69990087e-01 -2.71296173e-01
6.27264977e-01 2.31540035e-02 5.87399662e-01 -3.74378294e-01
-7.57054687e-01 -4.59805965e-01 3.56769234e-01 4.86942798e-01
2.16208607e-01 -2.14281566e-02 -1.08967446e-01 8.36241126e-01
-7.58983195e-02 -1.13255203e-01 2.73161858e-01 8.04352701e-01
-5.18577039e-01 -8.05426836e-01 -2.08016634e-01 1.00737579e-01
-2.23476395e-01 -2.61755288e-01 -7.63557926e-02 7.64396369e-01
5.87966800e-01 7.86444187e-01 -3.11499480e-02 -4.58193034e-01
2.48810828e-01 8.04028735e-02 1.64994344e-01 -3.98558497e-01
-6.91692054e-01 1.08852938e-01 -5.98005727e-02 -5.17386913e-01
-8.60563040e-01 -4.53143269e-01 -8.92777979e-01 4.35915068e-02
-1.76888019e-01 -1.32780030e-01 8.38586211e-01 9.54652309e-01
2.13589743e-01 6.97699070e-01 6.84917331e-01 -8.34190726e-01
-1.80196017e-01 -9.03926134e-01 -2.95411944e-01 6.38032198e-01
3.14304769e-01 -6.99733853e-01 -5.29215813e-01 3.17736506e-01]
|
[10.327132225036621, 2.895373582839966]
|
59e3e40e-17ae-42cc-9c97-6ec541979584
|
text-style-transfer-back-translation
|
2306.01318
| null |
https://arxiv.org/abs/2306.01318v1
|
https://arxiv.org/pdf/2306.01318v1.pdf
|
Text Style Transfer Back-Translation
|
Back Translation (BT) is widely used in the field of machine translation, as it has been proved effective for enhancing translation quality. However, BT mainly improves the translation of inputs that share a similar style (to be more specific, translation-like inputs), since the source side of BT data is machine-translated. For natural inputs, BT brings only slight improvements and sometimes even adverse effects. To address this issue, we propose Text Style Transfer Back Translation (TST BT), which uses a style transfer model to modify the source side of BT data. By making the style of source-side text more natural, we aim to improve the translation of natural inputs. Our experiments on various language pairs, including both high-resource and low-resource ones, demonstrate that TST BT significantly improves translation performance against popular BT benchmarks. In addition, TST BT is proved to be effective in domain adaptation so this strategy can be regarded as a general data augmentation method. Our training code and text style transfer model are open-sourced.
|
['Hao Yang', 'Zhengzhe Yu', 'Xiaoyu Chen', 'Jiaxin Guo', 'Minghan Wang', 'Zongyao Li', 'Hengchao Shang', 'Zhanglin Wu', 'Daimeng Wei']
|
2023-06-02
| null | null | null | null |
['style-transfer', 'text-style-transfoer']
|
['computer-vision', 'natural-language-processing']
|
[ 3.62726390e-01 -2.61630386e-01 -3.98986131e-01 -4.47756618e-01
-7.49736726e-01 -6.79517746e-01 7.29353130e-01 -3.83687347e-01
-3.44693244e-01 9.41320300e-01 2.46628568e-01 -6.09766424e-01
7.04578161e-01 -8.44252288e-01 -9.64502931e-01 -4.14889604e-01
8.42119992e-01 5.17310977e-01 1.21322996e-03 -8.41016769e-01
1.10992454e-02 8.43343213e-02 -7.46203184e-01 5.24718761e-01
1.15988052e+00 5.25656283e-01 2.39393711e-01 1.87292397e-01
-4.81699616e-01 3.95338863e-01 -7.31546819e-01 -9.48524117e-01
3.29522043e-01 -8.30769837e-01 -1.04715025e+00 -2.30255231e-01
2.28661865e-01 -7.66494796e-02 -1.53037190e-01 9.56944644e-01
5.58048368e-01 -1.34222060e-01 6.46458089e-01 -9.87020075e-01
-1.13445818e+00 6.90651536e-01 -4.97197390e-01 1.27207950e-01
3.31427753e-01 1.48686469e-01 8.59836400e-01 -1.20020866e+00
6.91649795e-01 1.24329865e+00 4.59617466e-01 9.17494118e-01
-1.31202328e+00 -7.49394178e-01 -1.14930123e-02 -4.99673374e-02
-1.02465749e+00 -2.62472242e-01 6.53408587e-01 -1.80300593e-01
7.89449871e-01 3.12518567e-01 3.06468606e-01 1.56801975e+00
4.38054293e-01 8.87945652e-01 1.30299938e+00 -6.50609553e-01
-6.55498654e-02 3.62386495e-01 -1.78196624e-01 1.97835058e-01
-8.09627026e-02 -7.68206129e-03 -4.72291768e-01 8.01414475e-02
6.67999566e-01 -9.59071442e-02 -3.94228786e-01 -8.83635413e-03
-1.55469549e+00 5.63901901e-01 4.76005733e-01 4.15617883e-01
-1.61985457e-01 -1.88401550e-01 7.12555528e-01 9.35302138e-01
8.18673074e-01 5.36614001e-01 -6.54929698e-01 -3.11725587e-01
-3.90141696e-01 2.40217030e-01 5.29977620e-01 1.52913070e+00
8.57042968e-01 -6.65794984e-02 -6.70380652e-01 1.19607067e+00
-2.31435239e-01 8.88277054e-01 7.80800700e-01 -2.46694177e-01
1.17018211e+00 6.57241940e-01 4.01424281e-02 -5.76465428e-01
4.22560833e-02 -5.32419026e-01 -1.07379127e+00 -2.12670028e-01
2.03339428e-01 -3.15025449e-01 -8.66969049e-01 1.84445345e+00
6.97974488e-02 -3.21957409e-01 1.38560906e-01 9.12401855e-01
7.26808965e-01 7.96108127e-01 -1.15050867e-01 -3.05976458e-02
1.06410909e+00 -1.33373177e+00 -6.98414505e-01 -4.39156353e-01
8.91400814e-01 -1.17528629e+00 1.73662293e+00 5.68679720e-02
-8.93703043e-01 -7.10091650e-01 -6.98932409e-01 -1.76135927e-01
-3.69451612e-01 2.76735365e-01 2.56959438e-01 4.23348606e-01
-7.74627745e-01 5.56178868e-01 -5.82815528e-01 -5.48574269e-01
2.93844730e-01 2.04997882e-01 -4.43000525e-01 -2.56847590e-01
-1.34599102e+00 1.09273386e+00 2.33836070e-01 -1.16918609e-01
-4.94994730e-01 -8.10415983e-01 -6.48650408e-01 -2.60791600e-01
1.57742620e-01 -9.35681999e-01 1.54929388e+00 -1.44513178e+00
-1.99909616e+00 8.92175913e-01 -3.01764339e-01 -2.02581123e-01
8.82574618e-01 -4.75464523e-01 -4.30467665e-01 -4.45245296e-01
2.69292921e-01 5.55347264e-01 9.76910293e-01 -8.56584370e-01
-3.27256024e-01 -1.62252143e-01 -1.80051163e-01 2.94121504e-01
-8.25980425e-01 2.33086258e-01 -5.38451791e-01 -1.16578388e+00
-3.82662296e-01 -1.10752237e+00 3.80263887e-02 -2.74013251e-01
-3.97368997e-01 -1.26326293e-01 6.02286279e-01 -6.77395880e-01
1.12962902e+00 -2.16535830e+00 5.16117454e-01 -2.32025713e-01
-1.66159272e-01 5.46564996e-01 -4.66185212e-01 6.48903072e-01
1.80667546e-02 1.16184749e-01 -4.64609206e-01 -3.49600762e-01
-1.44512773e-01 3.49498779e-01 -6.01502359e-01 -1.09160811e-01
5.89217365e-01 1.22685719e+00 -8.77179861e-01 -1.99359909e-01
-1.69062003e-01 9.17482302e-02 -4.54210103e-01 4.94875103e-01
-3.77186537e-01 7.46864259e-01 -5.53310931e-01 2.91467160e-01
6.82813704e-01 -7.76794851e-02 -2.41235465e-01 4.01070751e-02
4.44581248e-02 5.64029336e-01 -3.15543145e-01 1.93987596e+00
-7.95107961e-01 5.75398982e-01 -4.86203343e-01 -6.62933290e-01
1.32599807e+00 2.90085852e-01 1.30533037e-04 -9.97280598e-01
1.52455956e-01 4.25368160e-01 2.56265290e-02 -2.80843616e-01
5.34181297e-01 -2.58515984e-01 -1.11435615e-02 5.64232945e-01
-7.99310878e-02 -3.02425027e-01 1.05772667e-01 -8.65135342e-02
9.26512301e-01 5.17870843e-01 1.45876244e-01 -2.07349241e-01
6.03734672e-01 2.71415710e-01 6.32240117e-01 2.95357406e-01
1.55289128e-01 7.01866090e-01 2.21156985e-01 -3.78093958e-01
-1.16329432e+00 -8.18462610e-01 8.78386497e-02 1.21622360e+00
1.35603711e-01 -4.23798949e-01 -8.81980062e-01 -1.05462551e+00
-1.43588945e-01 7.03195810e-01 -6.47919655e-01 -5.49737096e-01
-8.67951870e-01 -6.99430168e-01 6.45305276e-01 6.23010635e-01
8.03334534e-01 -1.02574146e+00 7.83199146e-02 3.22487265e-01
-6.04499340e-01 -1.13884282e+00 -9.89181578e-01 3.69936489e-02
-1.09002864e+00 -4.68661666e-01 -1.29173708e+00 -8.80219221e-01
6.68085992e-01 4.72808212e-01 1.34208536e+00 -9.25950333e-02
3.37096155e-01 -3.09452236e-01 -8.01675260e-01 -4.57185447e-01
-9.54869449e-01 5.67958891e-01 -6.18142262e-02 -5.61651494e-03
4.74118382e-01 -4.66648936e-01 -2.85577357e-01 7.86349475e-01
-9.78110135e-01 5.18935859e-01 7.82407165e-01 1.08237088e+00
3.22429687e-01 -5.90876877e-01 5.56979477e-01 -1.10033917e+00
8.56282651e-01 -2.94337302e-01 -1.79656506e-01 4.28774774e-01
-5.97280979e-01 2.53679246e-01 1.23786247e+00 -6.43629014e-01
-1.12696767e+00 -3.85776281e-01 -1.82166651e-01 -3.91494215e-01
-1.87787618e-02 4.34588641e-01 -1.62980855e-01 -1.90627873e-02
9.72482204e-01 4.62015390e-01 -1.13864996e-01 -8.49546731e-01
2.03169823e-01 9.22063231e-01 2.49615133e-01 -8.97710919e-01
1.00682235e+00 1.98374409e-02 -2.41775185e-01 -3.05775762e-01
-5.96517742e-01 -1.49177566e-01 -6.34205341e-01 3.09980631e-01
5.18265784e-01 -8.64843905e-01 8.03174451e-02 4.71704334e-01
-1.29112804e+00 -5.14450252e-01 3.36488187e-02 3.03742975e-01
-5.09383082e-01 3.78047451e-02 -6.75412416e-01 -5.42034805e-02
-6.73970938e-01 -1.28620315e+00 1.13277996e+00 -8.14868808e-02
-2.75161475e-01 -9.16275024e-01 1.97477892e-01 2.26966396e-01
6.27881229e-01 -1.01221897e-01 1.15617836e+00 -5.22026658e-01
-1.31813109e-01 9.53568071e-02 -2.42521033e-01 6.14015400e-01
5.68924725e-01 -6.64290488e-02 -7.22065449e-01 -3.61289650e-01
-1.49089023e-01 -1.59470230e-01 5.29603481e-01 -3.21147472e-01
8.92795682e-01 -2.92753845e-01 -1.30253106e-01 6.96748018e-01
1.16239762e+00 1.72775369e-02 5.83023906e-01 5.01445889e-01
8.22082818e-01 4.87345368e-01 8.42449844e-01 -6.92102453e-03
1.84089258e-01 1.18736899e+00 -4.38023657e-02 -5.44583559e-01
-3.35798025e-01 -4.09657151e-01 7.79437065e-01 1.14865041e+00
-2.09509507e-01 -5.98009350e-03 -8.65417063e-01 4.04094428e-01
-1.68366265e+00 -5.27906299e-01 -2.29795322e-01 2.07181525e+00
1.41009605e+00 1.94840923e-01 2.63610452e-01 -1.88991189e-01
6.51678741e-01 -2.97731429e-01 -5.72943270e-01 -8.16117346e-01
-1.60044506e-01 4.38851595e-01 3.15974325e-01 7.71467835e-02
-7.25303888e-01 1.29967296e+00 5.83286190e+00 1.04229403e+00
-1.53225279e+00 2.77101755e-01 5.84294856e-01 1.41592324e-01
-4.76524174e-01 -1.58389181e-01 -7.01230288e-01 6.79306388e-01
6.28356576e-01 -5.07850409e-01 5.79272568e-01 6.84157968e-01
2.30216548e-01 5.18046379e-01 -1.40773952e+00 7.64425635e-01
-1.29060775e-01 -1.06932461e+00 4.91056949e-01 -7.41995275e-02
1.02475309e+00 6.88744262e-02 9.96266529e-02 6.61908627e-01
1.67342916e-01 -7.84697294e-01 6.83503091e-01 9.54712462e-03
1.13997257e+00 -7.40168393e-01 9.00619268e-01 4.22650129e-01
-8.06329012e-01 2.65543222e-01 -5.47819972e-01 -1.86350778e-01
-1.17906153e-01 5.18772423e-01 -7.93643296e-01 9.47130144e-01
6.23650372e-01 9.74952936e-01 -7.56241500e-01 5.26959777e-01
-4.35584992e-01 7.07645774e-01 9.35428739e-02 -6.26729429e-02
3.16911131e-01 -4.39513326e-01 3.30974817e-01 1.33867192e+00
4.23827261e-01 -3.42616022e-01 3.48478742e-02 9.81577873e-01
-4.74210501e-01 5.53998590e-01 -8.60696018e-01 -7.14233071e-02
2.00577706e-01 1.00672591e+00 -1.94889888e-01 -4.31082100e-01
-5.57805240e-01 1.44154239e+00 3.68701577e-01 4.53304470e-01
-7.79282391e-01 -4.08855617e-01 8.18662047e-01 -1.57752573e-01
6.92201331e-02 4.20265272e-02 -5.15720904e-01 -1.51441276e+00
3.35496962e-01 -1.39739525e+00 -9.72307026e-02 -6.26655102e-01
-1.36383379e+00 1.00874329e+00 -2.35306129e-01 -1.86281395e+00
-1.64624631e-01 -4.93218303e-01 -5.11822999e-01 1.25584853e+00
-1.56451464e+00 -1.26821637e+00 -1.71261460e-01 6.46233857e-01
8.45082939e-01 -3.05896163e-01 8.87722492e-01 5.10663867e-01
-6.04028165e-01 1.14343584e+00 4.09501523e-01 2.60071725e-01
1.30743694e+00 -1.15757895e+00 1.02600825e+00 8.20268631e-01
-2.85808593e-02 8.80950809e-01 6.25945628e-01 -5.90685606e-01
-1.46094251e+00 -1.54570901e+00 9.26216841e-01 -5.85556984e-01
5.25323331e-01 -4.72490668e-01 -1.03381515e+00 6.57949626e-01
4.32918131e-01 -3.56954038e-01 5.35070300e-01 -4.94700707e-02
-4.89765435e-01 -1.97191879e-01 -9.00847673e-01 9.29363906e-01
1.18137169e+00 -4.72311348e-01 -7.33242631e-01 3.68701637e-01
9.34265494e-01 -5.91779351e-01 -8.92228723e-01 5.68540931e-01
3.10540378e-01 -5.29165328e-01 6.11920655e-01 -7.59643495e-01
9.73717511e-01 -1.64720446e-01 2.29940861e-02 -1.90550971e+00
-2.02590793e-01 -8.05748701e-01 2.86290735e-01 1.46597147e+00
6.18329942e-01 -7.86701083e-01 4.08633649e-01 2.08371252e-01
-2.36557961e-01 -6.36504531e-01 -6.20080829e-01 -1.21712649e+00
7.47373521e-01 -1.68901440e-02 9.58672762e-01 1.06108081e+00
-4.59549725e-02 6.96156025e-01 -4.75414693e-01 -4.31856662e-01
3.97486724e-02 3.24142963e-01 1.17957771e+00 -7.53913403e-01
-3.82053316e-01 -5.50093591e-01 8.74509141e-02 -1.41570807e+00
1.79131985e-01 -1.13954020e+00 4.96790037e-02 -1.10731554e+00
1.09850869e-01 -4.91991043e-01 -1.13501862e-01 5.71482360e-01
-5.50814450e-01 4.49003458e-01 1.69178173e-01 4.36112851e-01
5.60264252e-02 8.13254893e-01 1.85141110e+00 -2.69210607e-01
-1.29232228e-01 2.76374042e-01 -6.83700204e-01 1.86815903e-01
9.15002644e-01 -4.85007137e-01 -3.22938472e-01 -1.02773750e+00
2.76586354e-01 -2.01172173e-01 -2.07643688e-01 -6.02601647e-01
-3.09199661e-01 -3.75752360e-01 1.80226211e-02 -2.66883045e-01
-7.20084086e-02 -8.02271664e-01 -7.59329647e-02 3.49296302e-01
-5.57553470e-01 3.82343858e-01 3.45858693e-01 1.17865130e-01
-4.05588210e-01 -2.97487024e-02 6.01895392e-01 4.46567796e-02
-2.87520111e-01 2.83196151e-01 -1.13739759e-01 9.79513675e-02
6.66920364e-01 5.44726066e-02 -3.15018564e-01 -2.72452235e-01
-1.26430646e-01 1.13216616e-01 7.54461944e-01 9.04791057e-01
2.75083125e-01 -1.66977823e+00 -1.00600946e+00 3.38900119e-01
6.79939687e-01 -1.70697889e-03 -3.46401960e-01 8.30107391e-01
-3.90090138e-01 4.14106458e-01 -3.67393106e-01 -5.52640438e-01
-1.11049807e+00 5.70200503e-01 1.30504295e-01 -1.98570833e-01
-5.78275979e-01 7.05299556e-01 4.07308698e-01 -8.93330932e-01
-1.18446387e-01 -5.64914048e-01 1.66252375e-01 -4.76684213e-01
5.20486116e-01 1.99589729e-01 4.24789548e-01 -4.25996304e-01
-1.86314851e-01 6.72881782e-01 -3.52786660e-01 -6.15673959e-02
1.18236518e+00 -7.42891505e-02 -3.20918441e-01 4.65463459e-01
1.22810531e+00 -3.71440290e-03 -9.01025057e-01 -7.50466824e-01
-4.26170230e-02 -5.37885249e-01 -4.31036830e-01 -1.00502527e+00
-9.09794867e-01 1.17874360e+00 2.14071304e-01 -1.50543869e-01
1.25519454e+00 -3.34608734e-01 1.15844584e+00 5.92613935e-01
5.69117010e-01 -8.44535410e-01 -8.43874365e-02 8.49826038e-01
1.13645101e+00 -1.29942441e+00 -4.40838903e-01 -5.01010060e-01
-6.98799372e-01 1.17952311e+00 8.89823675e-01 1.74831420e-01
4.34277356e-02 1.19085247e-02 3.86769056e-01 4.80565399e-01
-7.03881085e-01 6.60069957e-02 4.48513210e-01 4.98932064e-01
7.69553602e-01 1.38050914e-01 -6.14992440e-01 2.76103616e-01
-3.96088988e-01 1.69833750e-01 1.73993230e-01 7.03481674e-01
1.11027092e-01 -1.78144169e+00 -3.95443350e-01 1.69227242e-01
-3.66144031e-01 -3.82634938e-01 -7.61186540e-01 5.46333671e-01
-1.75136730e-01 9.05958652e-01 -2.37464324e-01 -6.45470798e-01
5.68959236e-01 3.98813002e-02 4.64830667e-01 -8.00750315e-01
-1.00772274e+00 -1.57726303e-01 4.31053676e-02 -3.31149966e-01
-1.26572043e-01 -2.30268195e-01 -9.12797034e-01 -5.18572986e-01
-2.15182096e-01 2.18657240e-01 5.95511019e-01 1.09004283e+00
5.26444852e-01 5.25541544e-01 1.05052173e+00 -4.00476038e-01
-7.59360254e-01 -1.42732751e+00 1.62990354e-02 8.20881844e-01
1.88073426e-01 -3.99435014e-01 1.49772212e-01 2.43245795e-01]
|
[11.647781372070312, 10.099991798400879]
|
7e408641-b8a5-47fb-803f-70ca9d7c491d
|
class-similarity-weighted-knowledge
| null | null |
http://openaccess.thecvf.com//content/CVPR2022/html/Phan_Class_Similarity_Weighted_Knowledge_Distillation_for_Continual_Semantic_Segmentation_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Phan_Class_Similarity_Weighted_Knowledge_Distillation_for_Continual_Semantic_Segmentation_CVPR_2022_paper.pdf
|
Class Similarity Weighted Knowledge Distillation for Continual Semantic Segmentation
|
Deep learning models are known to suffer from the problem of catastrophic forgetting when they incrementally learn new classes. Continual learning for semantic segmentation (CSS) is an emerging field in computer vision. We identify a problem in CSS: A model tends to be confused between old and new classes that are visually similar, which makes it forget the old ones. To address this gap, we propose REMINDER - a new CSS framework and a novel class similarity knowledge distillation (CSW-KD) method. Our CSW-KD method distills the knowledge of a previous model on old classes that are similar to the new one. This provides two main benefits: (i) selectively revising old classes that are more likely to be forgotten, and (ii) better learning new classes by relating them with the previously seen classes. Extensive experiments on Pascal-VOC 2012 and ADE20k datasets show that our approach outperforms state-of-the-art methods on standard CSS settings by up to 7.07% and 8.49%, respectively.
|
['Abdesselam Bouzerdoum', 'Long Tran-Thanh', 'Son Lam Phung', 'The-Anh Ta', 'Minh Hieu Phan']
|
2022-01-01
| null | null | null |
cvpr-2022-1
|
['continual-semantic-segmentation']
|
['computer-vision']
|
[ 3.73659343e-01 1.20021179e-01 3.09114177e-02 -4.85521674e-01
-2.30591819e-01 -5.88572323e-01 4.59869206e-01 5.13484001e-01
-7.51881897e-01 9.74120378e-01 -2.06331462e-01 4.03092392e-02
7.39771947e-02 -7.84711182e-01 -9.51104045e-01 -7.19162405e-01
1.26608640e-01 6.39483869e-01 1.23456991e+00 2.79897987e-03
1.58279538e-01 3.36280763e-01 -1.82474208e+00 4.27017242e-01
1.07309425e+00 8.55453908e-01 4.62622970e-01 3.87015551e-01
-4.19563323e-01 5.69909692e-01 -5.63642323e-01 -6.56648993e-01
3.02975215e-02 -3.00260276e-01 -1.28208578e+00 6.13002367e-02
6.99024558e-01 -9.09784213e-02 -1.95204720e-01 1.06997681e+00
7.60462806e-02 3.33129525e-01 3.75218600e-01 -1.18869340e+00
-7.25220621e-01 6.75124347e-01 -6.48799241e-01 2.54718810e-01
-2.77617246e-01 3.41954008e-02 5.20191729e-01 -9.65994179e-01
7.94799805e-01 1.31162369e+00 8.80630195e-01 9.60091412e-01
-1.01888263e+00 -6.58474505e-01 7.38427281e-01 6.76614046e-01
-1.29414034e+00 -7.54042789e-02 7.17449963e-01 -2.41248310e-01
6.31823897e-01 1.76673874e-01 8.43484163e-01 9.56319213e-01
-2.22199157e-01 1.29658771e+00 1.23072278e+00 -3.23171228e-01
3.80159199e-01 3.37524295e-01 5.43178022e-01 5.59910595e-01
4.76717919e-01 -3.42122540e-02 -5.79986811e-01 2.44921237e-01
2.44155064e-01 2.19115898e-01 -2.08663363e-02 -7.55606174e-01
-8.06349874e-01 5.43914557e-01 6.00550830e-01 3.31868589e-01
-3.99986580e-02 4.74314131e-02 2.58850813e-01 6.36665449e-02
4.03632671e-01 4.21412706e-01 -8.46813023e-01 1.44596159e-01
-8.54407012e-01 1.41363904e-01 4.78898525e-01 9.39626575e-01
1.13194489e+00 -1.90983593e-01 -6.66561723e-02 9.69305158e-01
3.58946472e-02 2.99770206e-01 7.09346414e-01 -6.83732867e-01
-4.91543487e-02 5.64783931e-01 -5.35442270e-02 -6.93812728e-01
-1.63252756e-01 -8.34430933e-01 -4.51522976e-01 5.63013777e-02
2.92222109e-02 7.30513483e-02 -1.58344793e+00 1.64700222e+00
4.88239199e-01 7.41699994e-01 2.21847340e-01 4.36373651e-01
8.51025343e-01 4.40415740e-01 2.94626921e-01 -1.80296287e-01
6.56073451e-01 -1.37083042e+00 -4.08758610e-01 -7.63287783e-01
1.59963116e-01 -5.60766220e-01 1.02389991e+00 3.96707088e-01
-6.26990020e-01 -9.89886403e-01 -1.08888423e+00 2.22543016e-01
-7.42027700e-01 -2.06970647e-01 7.04669356e-01 3.86736572e-01
-9.56143975e-01 9.62979436e-01 -8.04540873e-01 -5.50807416e-01
9.25844252e-01 3.52103205e-04 6.71826825e-02 -2.42135048e-01
-8.70582163e-01 7.13218212e-01 9.85544860e-01 -2.04681829e-01
-1.15540421e+00 -9.09723103e-01 -5.38639724e-01 -7.97501206e-02
7.20389128e-01 -4.96915877e-01 1.42265451e+00 -1.29601693e+00
-1.10768402e+00 8.70277882e-01 -2.03040257e-01 -5.38038850e-01
6.41266465e-01 -5.99137843e-01 -5.52746773e-01 4.22100686e-02
2.39682585e-01 9.70151722e-01 1.08465302e+00 -1.87246132e+00
-1.33999360e+00 -4.14508492e-01 -1.31057471e-01 3.00997138e-01
-3.19159567e-01 -7.67620146e-01 -6.99970305e-01 -7.36597896e-01
4.89023954e-01 -9.31378603e-01 -9.55053046e-02 4.22134846e-02
-3.14246625e-01 -3.20864737e-01 1.20333707e+00 -2.83222318e-01
1.06599367e+00 -2.06968856e+00 8.31572339e-02 -2.29499102e-01
3.00880194e-01 9.57892418e-01 -1.39919668e-01 -1.22865565e-01
5.05590672e-03 5.37891462e-02 -4.16578203e-01 -5.31388164e-01
-3.85660201e-01 6.50451958e-01 -4.16727602e-01 -2.01303989e-01
2.26943895e-01 7.63957918e-01 -1.36353528e+00 -3.26292157e-01
8.38866830e-02 1.37020484e-01 -2.70696521e-01 3.72546492e-03
-3.75363290e-01 3.01602602e-01 -7.33780861e-02 5.97545445e-01
9.88293171e-01 -1.07165597e-01 -8.73883367e-02 -1.39348507e-01
8.70999545e-02 -2.01990277e-01 -1.16240096e+00 1.73591399e+00
6.41025081e-02 4.77059096e-01 -5.89653790e-01 -9.04058099e-01
7.68889606e-01 -2.37419382e-01 -1.40147090e-01 -4.69917923e-01
-6.70261309e-02 1.94585770e-01 -2.86020279e-01 -4.44320619e-01
5.35928607e-01 -1.58831820e-01 3.42126548e-01 7.60630667e-02
4.18462425e-01 -1.95699051e-01 3.30486685e-01 3.37686568e-01
7.70661950e-01 1.72689646e-01 -1.16565101e-01 -2.17613995e-01
2.46965423e-01 -3.46093513e-02 1.00132763e+00 1.16350663e+00
-5.26854038e-01 5.44073105e-01 1.37093470e-01 -6.48958445e-01
-6.21694088e-01 -1.49986744e+00 -1.01059806e-02 1.13619971e+00
5.79180241e-01 -1.52235523e-01 -6.39159143e-01 -1.19576871e+00
1.31386802e-01 1.03800392e+00 -8.22581947e-01 -5.58890164e-01
-3.91252398e-01 -6.74931884e-01 7.28376359e-02 7.14417219e-01
7.99758017e-01 -1.04327452e+00 -6.84233785e-01 3.44742239e-01
-2.30271574e-02 -9.71243799e-01 -1.52958304e-01 2.67680705e-01
-1.21008456e+00 -1.28268731e+00 -9.80603874e-01 -1.08257639e+00
8.35568070e-01 5.91695786e-01 1.01327586e+00 2.13763252e-01
-3.33193690e-01 5.17271042e-01 -6.16264164e-01 -7.13821173e-01
-3.95959496e-01 1.63006946e-01 5.82359508e-02 2.08433867e-01
4.90252942e-01 -3.01831901e-01 -5.36915958e-01 1.25333264e-01
-9.40114439e-01 3.37453097e-01 4.60152477e-01 6.90461397e-01
8.16029131e-01 2.82805592e-01 7.31750786e-01 -1.46933448e+00
2.19741002e-01 -3.02866608e-01 -2.58356214e-01 6.03507876e-01
-1.04876435e+00 8.76830071e-02 2.55156845e-01 -6.71851575e-01
-1.39489758e+00 2.06278577e-01 7.04107583e-02 -3.15784514e-01
-3.13129097e-01 1.94252506e-01 8.04274380e-02 3.02701369e-02
6.36108100e-01 2.97508568e-01 -5.34805059e-01 -7.04377234e-01
5.54871619e-01 3.19782674e-01 1.01949453e+00 -6.08032703e-01
6.99775219e-01 5.97286820e-01 -5.22554815e-01 -6.58800781e-01
-1.45792317e+00 -5.27730703e-01 -9.51754034e-01 -1.99761406e-01
6.54658258e-01 -7.16185451e-01 7.07869977e-02 1.17137730e+00
-9.38319445e-01 -3.76503885e-01 -6.22414827e-01 8.51230323e-02
-1.97843298e-01 5.64914644e-01 -4.52170372e-01 -5.21809697e-01
-1.38011202e-01 -8.35605860e-01 6.66895807e-01 9.00117099e-01
3.84692140e-02 -7.57727206e-01 -4.63049337e-02 1.89452633e-01
4.43207353e-01 1.83917448e-01 9.73570347e-01 -6.38533950e-01
-3.56927484e-01 4.07345071e-02 -2.91258633e-01 5.14914036e-01
2.77866393e-01 -1.26353323e-01 -1.25843966e+00 -5.39178729e-01
-2.30869442e-01 -2.19159067e-01 1.62320328e+00 -1.62317734e-02
1.41084123e+00 -9.73940119e-02 -7.95877457e-01 4.69830900e-01
1.39656544e+00 5.38105667e-01 5.33479929e-01 3.40330958e-01
7.70960391e-01 3.25379491e-01 6.77182138e-01 2.52812147e-01
3.68591279e-01 2.20717147e-01 4.23732251e-01 -2.02876464e-01
-5.90745270e-01 -4.23202455e-01 -2.44852498e-01 7.54959881e-01
2.57367522e-01 -2.47482248e-02 -9.57706451e-01 1.06005585e+00
-2.07648206e+00 -5.94743669e-01 7.58536234e-02 2.23926401e+00
1.21653795e+00 6.44438863e-01 -2.19545558e-01 1.48754105e-01
8.77242565e-01 -8.69798511e-02 -1.04933059e+00 1.08192265e-02
-2.98669845e-01 5.48375607e-01 2.34328136e-01 4.78435278e-01
-1.24572337e+00 1.57434022e+00 5.48347998e+00 8.74591053e-01
-9.86425281e-01 2.62887985e-01 6.85438156e-01 2.76498735e-01
-6.07162490e-02 2.61617959e-01 -9.68137383e-01 3.35972369e-01
2.51132429e-01 -1.44702435e-01 1.57557115e-01 9.57609951e-01
-6.41663969e-01 -5.69998384e-01 -1.06950760e+00 1.05399907e+00
3.10647100e-01 -1.29853475e+00 4.37823921e-01 -7.76087284e-01
1.10424209e+00 -2.85230596e-02 9.20503661e-02 5.10471702e-01
3.62659335e-01 -5.06170452e-01 8.47101092e-01 6.38069034e-01
4.58493352e-01 -8.04911017e-01 7.61840105e-01 1.49178490e-01
-1.09531379e+00 -1.73220754e-01 -4.75271195e-01 3.11608523e-01
-1.91165268e-01 6.09964788e-01 -8.63216519e-01 3.93594205e-01
1.32753932e+00 9.74637151e-01 -1.21305740e+00 1.44420588e+00
-5.83742738e-01 6.75378919e-01 4.50548064e-03 1.59759060e-01
2.63126343e-01 2.87084103e-01 3.30700338e-01 1.07318640e+00
5.25591746e-02 1.84329245e-02 4.27724794e-02 7.73975313e-01
3.33557054e-02 -5.01999497e-01 -4.73371893e-02 9.55365747e-02
6.16244555e-01 9.12113905e-01 -1.23429811e+00 -8.06445479e-01
-3.25281359e-02 1.58954513e+00 3.86183560e-01 4.52367634e-01
-6.12756729e-01 -5.85759759e-01 4.32164401e-01 -1.56253904e-01
7.08768189e-01 -5.52311875e-02 -1.15855560e-01 -1.04009998e+00
-1.73078641e-01 -4.11929548e-01 6.41204059e-01 -8.62964272e-01
-1.44779181e+00 7.09727705e-01 -3.40457857e-02 -8.05416107e-01
3.65043402e-01 -3.76979679e-01 -3.64982396e-01 3.48840892e-01
-1.84033287e+00 -1.01864028e+00 -5.64402461e-01 4.37198192e-01
9.30621743e-01 -1.45622447e-01 5.78900993e-01 2.81668931e-01
-2.55370647e-01 5.45831323e-01 2.20015734e-01 -9.80942547e-02
6.71948254e-01 -1.33935118e+00 5.43792069e-01 9.49829817e-01
2.45863065e-01 2.59539694e-01 6.93124056e-01 -8.40786517e-01
-5.43896258e-01 -1.39506018e+00 8.14586341e-01 -4.86114889e-01
2.33837962e-01 -9.32884589e-02 -1.32992280e+00 5.30966878e-01
-1.22574642e-01 -7.35955164e-02 3.58530223e-01 -5.68890609e-02
-5.58265388e-01 -3.67811173e-01 -1.08128524e+00 6.06488466e-01
1.25970554e+00 -1.81142092e-01 -9.35486734e-01 1.98104128e-01
1.02796733e+00 -3.26495916e-01 -2.14133114e-02 6.12072825e-01
2.50592142e-01 -1.04227257e+00 9.50103700e-01 -6.64284825e-01
-5.70300333e-02 -4.97172534e-01 1.31543025e-01 -1.58542562e+00
-3.71179968e-01 -4.56405878e-01 -2.77025998e-01 1.29534519e+00
3.31982285e-01 -5.48865199e-01 8.34173560e-01 3.59631002e-01
-2.85982370e-01 -6.33437097e-01 -8.29788685e-01 -1.19961989e+00
1.80914462e-01 -2.20523313e-01 6.67518020e-01 1.17943323e+00
-6.34773135e-01 3.11255693e-01 -8.36557802e-03 1.16049923e-01
7.04457104e-01 8.87732506e-02 5.14258087e-01 -1.55705678e+00
-1.30757883e-01 -3.80185187e-01 -3.62363100e-01 -9.82164741e-01
-1.94202691e-01 -7.62420416e-01 1.60351321e-01 -1.60881603e+00
5.07959485e-01 -7.25199342e-01 -7.07811058e-01 8.81688237e-01
-5.54841518e-01 5.44207171e-03 3.55717510e-01 3.13456282e-02
-1.04603553e+00 5.91174841e-01 1.15491796e+00 -4.00566548e-01
-4.20766145e-01 1.13663055e-01 -7.58929968e-01 9.04384613e-01
7.12055445e-01 -8.47182095e-01 -5.75204492e-01 -5.88260591e-01
-3.87681201e-02 -7.06662118e-01 3.52447927e-01 -1.50353014e+00
2.88590252e-01 -1.68628674e-02 5.19212425e-01 -7.97186434e-01
1.69839367e-01 -7.19794810e-01 -9.26508661e-03 8.67323697e-01
-1.87234893e-01 -3.36992383e-01 5.79700887e-01 9.36437547e-01
3.47900800e-02 -4.61806178e-01 1.16075969e+00 -3.56015712e-01
-1.59061623e+00 4.07024354e-01 1.62749831e-02 3.34712118e-01
1.14114964e+00 -4.08688039e-01 -4.82432246e-01 5.86720929e-02
-1.08553326e+00 3.56507272e-01 2.96063453e-01 9.56277549e-01
7.95404315e-01 -1.08596933e+00 -4.31287467e-01 1.56117916e-01
3.89844865e-01 3.55105519e-01 3.47837895e-01 2.67366171e-01
-3.20651203e-01 3.73272821e-02 -7.78993815e-02 -7.20625043e-01
-1.33867788e+00 7.37971127e-01 3.02623451e-01 1.19823404e-01
-6.22287989e-01 1.45067382e+00 2.31300354e-01 -3.47433001e-01
5.18930316e-01 -3.21474850e-01 -1.60013705e-01 1.33733436e-01
5.22761583e-01 4.21440005e-01 1.27343699e-01 -2.55352616e-01
-3.30906421e-01 5.71967661e-01 -7.34886587e-01 2.57562757e-01
1.38905227e+00 -2.94918150e-01 1.00978859e-01 7.37596691e-01
8.18330109e-01 -5.38093328e-01 -1.75462472e+00 -8.51215303e-01
2.96750367e-01 -4.87583548e-01 -2.67403036e-01 -1.25478590e+00
-1.12771165e+00 9.35883164e-01 1.23463500e+00 -1.36435002e-01
1.09118223e+00 2.93122977e-01 1.03717935e+00 4.89867747e-01
4.80816215e-01 -1.58655548e+00 4.71188664e-01 7.36066818e-01
6.50165319e-01 -1.21313977e+00 3.20978202e-02 -4.35436457e-01
-6.94126546e-01 1.04619730e+00 1.02634263e+00 1.39688686e-01
7.05921590e-01 -2.96046823e-01 7.04434067e-02 -1.24185935e-01
-3.62616122e-01 -4.96931493e-01 1.27580464e-01 8.64157021e-01
-2.48822406e-01 1.08573206e-01 -1.02660619e-01 7.20541477e-01
9.27508101e-02 1.09282702e-01 4.93948251e-01 1.19749618e+00
-7.91463196e-01 -9.50094759e-01 8.95842463e-02 4.14328247e-01
-3.43083777e-03 -4.19162922e-02 -5.25469363e-01 6.37315929e-01
6.18585229e-01 8.27741563e-01 7.72367045e-02 -4.57374245e-01
2.66517520e-01 1.94342002e-01 5.51821172e-01 -8.18710268e-01
-1.10689223e-01 -2.10550800e-01 -4.06813204e-01 -3.06726158e-01
-4.81509477e-01 -6.21748686e-01 -1.46061420e+00 1.04148291e-01
-3.89290541e-01 6.56592622e-02 3.89433831e-01 1.04026592e+00
4.06248271e-01 7.66812623e-01 3.11339825e-01 -4.16929513e-01
-2.47516915e-01 -5.18390954e-01 -5.30069053e-01 6.10137880e-01
3.52567792e-01 -8.70340765e-01 -2.23846644e-01 3.40889931e-01]
|
[9.424365043640137, 2.1000163555145264]
|
8a119044-6f8f-42c4-8196-df9553213b2a
|
local-facial-makeup-transfer-via-disentangled
|
2003.12065
| null |
https://arxiv.org/abs/2003.12065v2
|
https://arxiv.org/pdf/2003.12065v2.pdf
|
Local Facial Makeup Transfer via Disentangled Representation
|
Facial makeup transfer aims to render a non-makeup face image in an arbitrary given makeup one while preserving face identity. The most advanced method separates makeup style information from face images to realize makeup transfer. However, makeup style includes several semantic clear local styles which are still entangled together. In this paper, we propose a novel unified adversarial disentangling network to further decompose face images into four independent components, i.e., personal identity, lips makeup style, eyes makeup style and face makeup style. Owing to the further disentangling of makeup style, our method can not only control the degree of global makeup style, but also flexibly regulate the degree of local makeup styles which any other approaches can't do. For makeup removal, different from other methods which regard makeup removal as the reverse process of makeup, we integrate the makeup transfer with the makeup removal into one uniform framework and obtain multiple makeup removal results. Extensive experiments have demonstrated that our approach can produce more realistic and accurate makeup transfer results compared to the state-of-the-art methods.
|
['Zhaoyang Sun', 'Shengwu Xiong', 'Ryan Wen Liu', 'Wenxuan Liu', 'Feng Liu']
|
2020-03-27
| null | null | null | null |
['facial-makeup-transfer']
|
['computer-vision']
|
[ 3.08778957e-02 2.24075355e-02 -6.41568229e-02 -5.02250075e-01
-2.67921507e-01 -8.95045519e-01 4.78657454e-01 -8.99641097e-01
2.25594372e-01 5.48559129e-01 3.11758846e-01 8.37121755e-02
2.76627362e-01 -1.06487679e+00 -7.84792006e-01 -8.42419744e-01
7.80172229e-01 7.29958490e-02 -2.71019518e-01 -4.44728881e-01
-5.79499826e-02 5.76848924e-01 -1.17305899e+00 2.45091975e-01
8.47165406e-01 7.78530300e-01 -3.06211889e-01 4.63749141e-01
-3.71427953e-01 5.55754244e-01 -5.82302690e-01 -7.55061507e-01
4.77046132e-01 -6.45955324e-01 -3.81369501e-01 -1.04870955e-02
6.86963201e-01 -7.01972425e-01 -4.20621455e-01 1.47543788e+00
5.98071218e-01 -2.77658373e-01 7.09326625e-01 -1.56077158e+00
-1.55009258e+00 2.98816413e-01 -7.83964396e-01 -3.19563597e-01
2.90573567e-01 1.65134773e-01 4.09239739e-01 -6.82264626e-01
5.13795197e-01 1.71257603e+00 6.18347704e-01 1.22027326e+00
-1.21178246e+00 -1.76653504e+00 2.98803955e-01 -1.84408933e-01
-1.52381897e+00 -6.38763845e-01 1.07902002e+00 -2.73680568e-01
-6.38363436e-02 5.43854952e-01 4.75687891e-01 1.08591032e+00
-5.50906770e-02 5.98593652e-01 1.52817500e+00 -9.96628478e-02
-1.08876012e-01 5.98045290e-02 -1.50180757e-01 8.61874878e-01
5.65117970e-02 9.41038504e-02 -2.07832113e-01 2.58719847e-02
1.29309487e+00 1.92016497e-01 -3.65271807e-01 -2.71620721e-01
-1.00289881e+00 5.19455612e-01 5.42372704e-01 5.93310501e-03
2.69507647e-01 2.62077659e-01 8.76096170e-03 5.13564825e-01
3.50718826e-01 1.67206973e-01 -2.61110336e-01 3.62200409e-01
-5.95824778e-01 1.05677359e-01 6.47443354e-01 9.66504633e-01
1.12453091e+00 1.04083233e-01 -2.50666440e-01 8.01978648e-01
3.70804369e-01 7.99187422e-01 3.27415019e-01 -8.16401064e-01
3.97681803e-01 4.92354304e-01 7.43987337e-02 -1.18154705e+00
1.40823334e-01 2.34548703e-01 -1.10852075e+00 5.74991524e-01
1.58547431e-01 -5.13482988e-01 -1.13796055e+00 2.18485594e+00
3.90397012e-01 4.09828126e-01 -1.16944797e-01 7.41312146e-01
1.11065340e+00 8.31818402e-01 1.11543261e-01 -2.82556098e-02
1.34582233e+00 -8.40946257e-01 -9.56716895e-01 3.90446279e-03
-8.95757526e-02 -9.66027319e-01 8.16217303e-01 2.11133987e-01
-7.05789328e-01 -6.95380211e-01 -1.32662249e+00 -3.27512711e-01
-2.00913742e-01 1.80233583e-01 6.55514002e-01 9.38365340e-01
-1.28710318e+00 3.50111753e-01 -4.49400395e-01 -1.40919179e-01
6.61644638e-01 5.57782948e-01 -9.46014941e-01 6.46381155e-02
-1.18514371e+00 6.89953566e-01 1.63568184e-01 1.60444856e-01
-1.01880145e+00 -8.05552542e-01 -6.47657752e-01 -1.76314086e-01
3.75544727e-01 -1.04970789e+00 8.52034509e-01 -1.47083580e+00
-1.94790518e+00 1.06967521e+00 -1.03518650e-01 4.30573404e-01
6.41328931e-01 -4.92233075e-02 -5.48417568e-01 -1.58470422e-01
-1.97570801e-01 7.09867954e-01 1.31242990e+00 -1.58807516e+00
-3.05009056e-02 -3.85474265e-01 3.27394396e-01 3.51282150e-01
-3.93002093e-01 2.89763480e-01 -5.15946329e-01 -9.31343496e-01
-7.05704018e-02 -8.66019547e-01 3.11431766e-01 5.60707211e-01
-5.59586525e-01 8.64572227e-02 9.98223782e-01 -6.49564683e-01
1.07590735e+00 -2.15202093e+00 4.98002678e-01 -4.25815023e-02
5.71858883e-01 3.54882330e-01 -4.79084998e-01 3.35259318e-01
-4.09534514e-01 6.01038098e-01 -9.08134207e-02 -2.02345639e-01
9.06499773e-02 2.75299281e-01 -3.38688493e-01 3.81729573e-01
5.60700335e-02 9.98848855e-01 -7.20881641e-01 -3.52771640e-01
2.39468649e-01 7.58191466e-01 -6.19578362e-01 2.11028129e-01
9.42777470e-02 6.41984463e-01 -6.84859455e-01 5.56506157e-01
1.36593318e+00 4.08509284e-01 -2.16594916e-02 -3.92416745e-01
1.67184725e-01 -3.64789039e-01 -9.63902831e-01 1.68503380e+00
-5.99679112e-01 2.17839211e-01 5.10776281e-01 -3.22956979e-01
9.93524194e-01 2.79350549e-01 1.21814094e-01 -1.82937786e-01
3.62292469e-01 -1.02892324e-01 -6.92533776e-02 -3.51939052e-01
1.98311612e-01 -7.56656587e-01 4.25296165e-02 3.16122800e-01
4.18021865e-02 -4.22408968e-01 -5.49909472e-01 -4.79239039e-02
4.92850810e-01 2.53883630e-01 1.07308075e-01 -5.35875201e-01
7.28204250e-01 -1.00151002e+00 7.77319431e-01 -1.20163530e-01
-4.98519540e-01 8.42962503e-01 7.67828405e-01 -3.49441022e-01
-8.73132110e-01 -1.24859977e+00 4.46304791e-02 9.06180143e-01
6.87107980e-01 -3.28967988e-01 -1.22213292e+00 -8.53818595e-01
7.80625045e-02 2.82919228e-01 -1.07146513e+00 -3.88374925e-01
-6.25766158e-01 -5.85372686e-01 8.06404352e-01 3.27696562e-01
9.94362295e-01 -9.35543239e-01 6.01711869e-01 -4.54006314e-01
-5.46685979e-02 -7.68286705e-01 -1.09281456e+00 -9.06902373e-01
-2.44016424e-01 -9.65113342e-01 -8.49475801e-01 -7.21072078e-01
8.60956609e-01 1.38684973e-01 7.41199970e-01 2.41616890e-01
4.50801523e-03 4.39341590e-02 -2.67672408e-02 -5.18673599e-01
-4.51157928e-01 -3.22340906e-01 2.03895077e-01 6.96235240e-01
2.09661782e-01 -9.50587809e-01 -7.96072245e-01 5.04046619e-01
-9.92229044e-01 3.85307312e-01 2.47868389e-01 6.43336058e-01
3.49844396e-01 -2.40701754e-02 7.04399228e-01 -9.07703102e-01
5.84923983e-01 -2.67320156e-01 -2.26884350e-01 5.12212932e-01
-3.94444615e-01 -9.27805901e-02 6.67232335e-01 -6.43329561e-01
-1.28528762e+00 -4.22125399e-01 -2.94688970e-01 -6.76436722e-01
-1.38465688e-01 -5.67161620e-01 -9.56124127e-01 -5.12432694e-01
3.38667572e-01 2.24802077e-01 3.72051686e-01 -2.95449406e-01
8.01117480e-01 6.16602004e-01 4.75570619e-01 -7.32957363e-01
1.30488741e+00 6.69561207e-01 -4.35418710e-02 -4.29211438e-01
-5.86666286e-01 3.13976586e-01 -5.20381689e-01 -1.84948727e-01
1.05715549e+00 -8.90603125e-01 -9.22628820e-01 1.04060721e+00
-1.00323999e+00 -1.01317473e-01 -8.15307647e-02 -2.19656184e-01
-2.75824070e-01 5.35652459e-01 -7.77768910e-01 -2.59292185e-01
-3.34982991e-01 -1.11182868e+00 1.06468642e+00 5.20048261e-01
1.42226785e-01 -8.54767501e-01 -1.22067697e-01 3.26245248e-01
3.98762196e-01 6.54720485e-01 7.09641159e-01 1.88676082e-03
-5.78381836e-01 2.45320424e-02 -4.72194880e-01 6.86627686e-01
7.56701171e-01 2.60898769e-01 -1.11637628e+00 -2.85032362e-01
9.49186683e-02 -1.37624875e-01 6.81683421e-01 -1.69677600e-01
1.06861293e+00 -6.51290357e-01 -1.72875479e-01 1.21857548e+00
1.40567195e+00 2.91792542e-01 8.66716683e-01 -3.21083888e-02
1.33316445e+00 5.83741009e-01 2.63346314e-01 7.41470009e-02
1.63859412e-01 5.55526853e-01 5.16478360e-01 -2.46470109e-01
-4.57428992e-01 -7.64816403e-01 4.96064305e-01 6.43868208e-01
-1.41880527e-01 -4.69716042e-02 -2.47692913e-01 7.69997165e-02
-1.43200016e+00 -9.49847281e-01 2.77096510e-01 1.91344047e+00
1.00249934e+00 -5.07690191e-01 -3.09327006e-01 -2.67976969e-01
1.03847253e+00 4.63446558e-01 -5.62548220e-01 -3.93189311e-01
-8.60087350e-02 5.02627753e-02 3.11775804e-01 5.71170211e-01
-9.97633040e-01 1.09671223e+00 6.05601788e+00 1.14030027e+00
-1.36913514e+00 2.83236742e-01 6.96682274e-01 -1.43278107e-01
-8.14392626e-01 -1.08930945e-01 -7.03535795e-01 9.10629034e-01
-1.95581205e-02 -2.49331653e-01 8.52222681e-01 4.36009288e-01
-1.63632289e-01 5.10045648e-01 -1.09760034e+00 1.27729225e+00
3.25533241e-01 -9.97238100e-01 4.74960536e-01 -7.92465210e-02
9.30763900e-01 -6.35243535e-01 3.85600984e-01 1.59744516e-01
3.89961630e-01 -1.24258530e+00 8.12704563e-01 4.82287854e-01
1.48137629e+00 -8.49547684e-01 2.23196939e-01 1.87598094e-01
-1.14920974e+00 1.07719257e-01 -2.94380307e-01 -3.03116404e-02
-3.06865033e-02 3.53762597e-01 -4.88070212e-02 9.02562916e-01
3.44022006e-01 6.67621434e-01 -3.40021104e-01 1.59831092e-01
-8.49975407e-01 7.45446086e-02 -1.81229457e-01 3.23892653e-01
-1.67175144e-01 -4.58017379e-01 4.88135576e-01 6.19434536e-01
4.44007993e-01 4.02544260e-01 -2.78310627e-02 9.67973471e-01
-7.18556523e-01 8.43135193e-02 -6.83148682e-01 2.93519527e-01
6.45389497e-01 1.27432919e+00 -2.05571234e-01 -3.19525689e-01
-3.29768136e-02 1.41757250e+00 1.34697303e-01 3.80129188e-01
-9.28193033e-01 -3.24195027e-01 1.32147038e+00 6.51741773e-02
-1.99137345e-01 6.07935302e-02 -2.09914580e-01 -1.54894876e+00
-4.99514788e-02 -8.41535032e-01 -1.23399980e-01 -1.11015320e+00
-1.48838079e+00 6.26400054e-01 -1.24182217e-01 -1.05860221e+00
3.34790587e-01 -6.09766603e-01 -9.95553792e-01 1.26867080e+00
-1.55091214e+00 -1.87824762e+00 -4.48749453e-01 8.19354951e-01
3.54296416e-01 -1.62398890e-01 8.13422918e-01 2.57220805e-01
-8.45218062e-01 9.89969671e-01 -8.37551206e-02 4.32096452e-01
1.13167787e+00 -9.23684597e-01 2.54859477e-01 9.31393266e-01
-2.27750942e-01 9.34115708e-01 3.08030725e-01 -5.63495576e-01
-1.33643377e+00 -1.09938455e+00 3.38230968e-01 -6.32711112e-01
4.71125335e-01 -5.49671948e-01 -7.08258688e-01 9.18231010e-01
4.19462442e-01 7.86653161e-02 6.78283453e-01 -7.80861452e-02
-8.82947087e-01 -5.58218956e-01 -1.46850908e+00 6.98660612e-01
1.23889220e+00 -7.21653700e-01 -6.92076921e-01 -7.72703141e-02
7.34787285e-01 -3.75613302e-01 -6.39670789e-01 3.78328949e-01
7.37197697e-01 -1.11168075e+00 1.07909000e+00 -3.05521548e-01
5.90831041e-01 -6.06284678e-01 1.58362076e-01 -1.35604370e+00
-4.78431344e-01 -1.09214032e+00 3.07946235e-01 1.96972322e+00
-1.69209212e-01 -9.26178932e-01 6.72830880e-01 6.36225700e-01
1.38311535e-01 -4.17530149e-01 -7.55508661e-01 -4.91633683e-01
4.36524749e-01 1.25387847e-01 1.38968158e+00 1.20097542e+00
-4.41576034e-01 3.09349358e-01 -8.14347565e-01 1.40093148e-01
5.84362030e-01 1.34371698e-01 9.19898868e-01 -8.75371039e-01
-8.93644169e-02 -4.74780768e-01 -3.96337181e-01 -8.75312805e-01
4.35012430e-01 -9.54898536e-01 -2.97982126e-01 -1.22613037e+00
3.86162519e-01 -6.56311691e-01 -4.05633003e-01 7.11561441e-01
-3.12363535e-01 5.89657903e-01 4.04744714e-01 1.60686359e-01
2.11534366e-01 5.65234065e-01 1.99955392e+00 -2.64626324e-01
1.59675255e-01 -3.30220819e-01 -1.33963931e+00 9.36631441e-01
7.13897884e-01 -1.46306083e-01 -8.08239639e-01 -7.22479045e-01
-1.82944734e-03 -1.28395930e-01 3.86564493e-01 -7.24994183e-01
-3.99524607e-02 -4.53234375e-01 4.54979926e-01 9.84230712e-02
3.51630658e-01 -7.01276720e-01 3.85870665e-01 1.71424091e-01
8.16403830e-04 -1.65249616e-01 2.54407257e-01 3.42097968e-01
-2.18205199e-01 2.12694585e-01 1.09509516e+00 -5.85090816e-02
-7.25687325e-01 7.84650147e-01 2.10742980e-01 -2.90934384e-01
1.34316254e+00 -2.49478921e-01 -5.43087304e-01 -3.14116061e-01
-7.62135446e-01 4.96123992e-02 8.41919839e-01 7.82125652e-01
4.78002489e-01 -1.67109597e+00 -8.67645860e-01 6.75104022e-01
-1.60122603e-01 -2.32729569e-01 8.71174753e-01 2.24501595e-01
-7.32149720e-01 -1.25299469e-01 -7.34412491e-01 -6.68196976e-02
-1.41573238e+00 9.31582808e-01 5.58297098e-01 1.07336864e-01
-3.15522015e-01 1.19781053e+00 1.26616323e+00 -6.96940482e-01
-4.43804234e-01 -3.52294161e-03 -1.98633328e-01 -9.56630707e-02
7.71168292e-01 1.26335770e-01 -5.64484954e-01 -8.86213064e-01
-1.95194438e-01 1.33982766e+00 6.74367473e-02 8.13214853e-02
8.09621513e-01 -3.39940429e-01 -6.48315668e-01 -1.63472956e-03
1.40347707e+00 5.03125250e-01 -1.38845313e+00 2.62608051e-01
-1.17653537e+00 -8.42562139e-01 -2.73380518e-01 -8.65290344e-01
-1.43623412e+00 9.77236986e-01 3.94940466e-01 -9.77545306e-02
1.48733461e+00 -2.61727154e-01 8.21494341e-01 -3.00211310e-01
5.44250250e-01 -4.54740435e-01 -2.70292431e-01 -4.18683840e-03
1.19163597e+00 -1.03923512e+00 -1.86326474e-01 -1.05095112e+00
-3.70127797e-01 7.58238077e-01 9.58894193e-01 -3.69783163e-01
8.43933761e-01 4.15071905e-01 2.78464407e-01 2.66127229e-01
-4.20629412e-01 3.45205903e-01 2.61244535e-01 7.63819456e-01
-5.83907031e-02 3.40859890e-01 -7.59790391e-02 7.37623572e-01
-3.73880923e-01 2.42148384e-01 3.21714610e-01 3.20802599e-01
-3.60123627e-02 -1.59182978e+00 -5.90628326e-01 -2.07836255e-02
-5.62335372e-01 -1.24835901e-01 -5.13927221e-01 7.23789454e-01
7.57565022e-01 7.57823944e-01 -7.68918693e-02 -7.15007544e-01
2.82078266e-01 -1.14026427e-01 7.37502933e-01 -5.31378508e-01
-4.85267371e-01 -2.60290951e-01 -1.95172653e-01 -6.20443523e-01
-2.15884551e-01 -1.16478026e-01 -9.40446734e-01 -1.00806093e+00
5.78948669e-02 -1.40272751e-01 2.66557008e-01 6.92425489e-01
4.07805294e-01 5.69022179e-01 8.99561107e-01 -7.67705083e-01
-1.99073896e-01 -7.99163282e-01 -7.65429616e-01 6.45467997e-01
3.92037034e-01 -5.70450425e-01 -1.31732360e-01 2.95964569e-01]
|
[12.71579360961914, -0.048966795206069946]
|
298cd552-83d4-46fc-9afb-15903d329037
|
one-peace-exploring-one-general
|
2305.11172
| null |
https://arxiv.org/abs/2305.11172v1
|
https://arxiv.org/pdf/2305.11172v1.pdf
|
ONE-PEACE: Exploring One General Representation Model Toward Unlimited Modalities
|
In this work, we explore a scalable way for building a general representation model toward unlimited modalities. We release ONE-PEACE, a highly extensible model with 4B parameters that can seamlessly align and integrate representations across vision, audio, and language modalities. The architecture of ONE-PEACE comprises modality adapters, shared self-attention layers, and modality FFNs. This design allows for the easy extension of new modalities by adding adapters and FFNs, while also enabling multi-modal fusion through self-attention layers. To pretrain ONE-PEACE, we develop two modality-agnostic pretraining tasks, cross-modal aligning contrast and intra-modal denoising contrast, which align the semantic space of different modalities and capture fine-grained details within modalities concurrently. With the scaling-friendly architecture and pretraining tasks, ONE-PEACE has the potential to expand to unlimited modalities. Without using any vision or language pretrained model for initialization, ONE-PEACE achieves leading results on a wide range of uni-modal and multi-modal tasks, including image classification (ImageNet), semantic segmentation (ADE20K), audio-text retrieval (AudioCaps, Clotho), audio classification (ESC-50, FSD50K, VGGSound), audio question answering (AVQA), image-text retrieval (MSCOCO, Flickr30K), and visual grounding (RefCOCO/+/g). Code is available at https://github.com/OFA-Sys/ONE-PEACE.
|
['Chang Zhou', 'Xinggang Wang', 'Jingren Zhou', 'Xiaohuan Zhou', 'Shuai Bai', 'Junyang Lin', 'Shijie Wang', 'Peng Wang']
|
2023-05-18
| null | null | null | null |
['audio-classification', 'self-supervised-image-classification', 'visual-grounding', 'action-classification']
|
['audio', 'computer-vision', 'computer-vision', 'computer-vision']
|
[ 2.07399622e-01 -2.18913227e-01 7.40691945e-02 -3.32172096e-01
-1.27977467e+00 -8.55396926e-01 5.56145310e-01 -1.02193199e-01
-4.70849067e-01 1.10400781e-01 2.84583151e-01 -1.23086229e-01
5.15938876e-03 -5.20148754e-01 -7.96263278e-01 -4.00424361e-01
2.91271836e-01 3.32394928e-01 6.34319335e-02 -3.59237283e-01
-8.79116282e-02 1.97577074e-01 -1.66777909e+00 9.13427711e-01
3.50237429e-01 1.36883104e+00 1.40926734e-01 1.17947876e+00
-8.69960561e-02 6.10834479e-01 -2.44236082e-01 -5.09900987e-01
3.36262211e-02 -1.15108922e-01 -1.29630327e+00 -1.06913067e-01
6.93915308e-01 -3.77709359e-01 -5.37276864e-01 7.34388769e-01
8.83931160e-01 1.72553092e-01 4.68077123e-01 -1.58218503e+00
-8.63465250e-01 5.21836460e-01 -5.64526677e-01 2.11217210e-01
5.77511668e-01 5.72158098e-01 1.07741284e+00 -7.89375424e-01
4.97881770e-01 1.48419690e+00 7.50152886e-01 6.16164029e-01
-1.09488571e+00 -6.74810648e-01 1.18179180e-01 5.36809981e-01
-1.38424850e+00 -6.22074544e-01 4.57658023e-01 -2.39784509e-01
1.15902805e+00 3.89329195e-01 5.25425792e-01 1.55132854e+00
-1.94889888e-01 9.89928603e-01 1.13692880e+00 -2.53670424e-01
-9.65722948e-02 -8.81760642e-02 1.33733556e-01 4.73146170e-01
-7.58902013e-01 -1.40047386e-01 -8.09086680e-01 7.14722574e-02
7.46281981e-01 -6.19684383e-02 -3.07895422e-01 1.35792971e-01
-1.39390409e+00 4.68110919e-01 6.26105666e-01 2.52818674e-01
-1.86286137e-01 4.03941840e-01 7.72409678e-01 4.53258425e-01
1.48391664e-01 2.10320413e-01 -5.61163604e-01 -1.31541088e-01
-7.99300849e-01 1.20808678e-02 3.47995996e-01 9.63156700e-01
6.17144585e-01 7.86428228e-02 -2.18133539e-01 1.10455167e+00
3.86968583e-01 8.21331978e-01 4.53187227e-01 -1.32828379e+00
4.10263628e-01 2.05341667e-01 -3.71783823e-01 -5.21741688e-01
-6.56473756e-01 -2.54556596e-01 -8.28115940e-01 -4.56697010e-02
1.05027147e-01 -8.23256075e-02 -1.38111329e+00 1.95072937e+00
2.46256635e-01 2.03379780e-01 3.36447731e-02 1.16680932e+00
1.55937696e+00 8.81478846e-01 4.12568390e-01 4.40013468e-01
1.73123240e+00 -1.13842773e+00 -4.02265817e-01 -1.98847532e-01
5.42885512e-02 -9.07811522e-01 1.47920716e+00 4.01728988e-01
-1.29303908e+00 -9.28598881e-01 -6.69165194e-01 -6.50936246e-01
-6.21789753e-01 -2.29750246e-01 6.11065447e-01 1.57128572e-01
-1.44252026e+00 6.47396147e-02 -6.27824187e-01 -7.37449765e-01
4.20390517e-01 2.01615229e-01 -6.93909585e-01 -9.18423831e-02
-1.37091625e+00 5.77529192e-01 4.64717060e-01 8.62307101e-03
-1.12437141e+00 -7.75391698e-01 -1.01315010e+00 -1.38024557e-02
2.59926736e-01 -1.14567792e+00 1.33750665e+00 -1.28671086e+00
-1.50654364e+00 1.23946095e+00 4.77258936e-02 -3.50069493e-01
6.37754351e-02 -4.89656888e-02 -6.98010206e-01 7.23060310e-01
6.64402917e-02 1.56970572e+00 1.22444820e+00 -1.09660745e+00
-4.98790264e-01 -2.07480982e-01 3.44525367e-01 5.35742462e-01
-3.15585583e-01 1.34664610e-01 -9.00285482e-01 -6.43044472e-01
-1.71500057e-01 -7.84422994e-01 2.52151221e-01 -5.28444126e-02
-4.90296513e-01 3.30201350e-02 7.90300250e-01 -8.65336597e-01
7.57689655e-01 -2.45543385e+00 2.45903403e-01 -1.12500954e-02
1.11596607e-01 6.47218227e-02 -9.31459248e-01 4.32029933e-01
-3.32947850e-01 3.41710746e-02 -1.36312082e-01 -5.58627427e-01
2.89368927e-01 1.73085198e-01 -4.04862523e-01 1.05028123e-01
2.90536791e-01 1.26966047e+00 -5.08244634e-01 -5.35589278e-01
2.66122133e-01 6.92207575e-01 -7.05746114e-01 1.19705826e-01
-2.60174632e-01 4.16360497e-01 6.37284443e-02 1.06752551e+00
5.97507060e-01 -4.16558057e-01 -2.99453139e-01 -6.69022679e-01
2.05063403e-01 1.33567646e-01 -1.16829550e+00 2.28785729e+00
-7.43797064e-01 7.71948576e-01 6.10124469e-01 -7.81021893e-01
3.08224618e-01 5.10261774e-01 4.32910711e-01 -9.79014874e-01
2.12728932e-01 -4.49586548e-02 -4.03012455e-01 -8.06016564e-01
6.52426720e-01 -7.74078369e-02 -2.90986180e-01 3.99117023e-01
8.99503291e-01 -2.41587743e-01 1.00890569e-01 4.18963253e-01
9.19405878e-01 -4.76097427e-02 -3.18222731e-01 3.35203290e-01
3.24345440e-01 -1.60694808e-01 -9.87625681e-03 6.62577987e-01
-9.53260139e-02 1.05030274e+00 7.59746358e-02 7.33967349e-02
-8.81001472e-01 -1.44985247e+00 -1.14415094e-01 1.81772268e+00
4.47251089e-02 -4.05745924e-01 -5.41386902e-01 -2.97278315e-01
9.79605466e-02 3.24242175e-01 -5.12791574e-01 -1.68601513e-01
-5.27108014e-02 -3.09785336e-01 1.09396875e+00 7.01005757e-01
7.58160651e-01 -1.14096439e+00 -2.66792178e-01 -1.10694222e-01
-6.79989994e-01 -1.25691676e+00 -6.01556420e-01 1.85455203e-01
-5.14403820e-01 -7.98359096e-01 -9.00016010e-01 -7.77463555e-01
8.70373920e-02 1.57580167e-01 1.17933106e+00 -1.50972813e-01
-2.99227834e-01 1.21335173e+00 -2.58797944e-01 -1.38275579e-01
-1.01706281e-01 2.56970525e-01 -2.49929070e-01 1.99705325e-02
-1.27035137e-02 -4.81703520e-01 -6.37112379e-01 1.12469055e-01
-1.29166365e+00 1.15785018e-01 3.55275065e-01 8.50872457e-01
7.41373837e-01 -4.26472098e-01 5.79794705e-01 -3.55787009e-01
4.78532463e-01 -5.44041574e-01 -8.40246975e-02 1.68749988e-01
-2.09047385e-02 -4.42941993e-01 2.68360376e-01 -5.25481761e-01
-9.51955974e-01 -1.96670339e-01 -6.79551244e-01 -8.01113307e-01
-6.78993106e-01 5.93253314e-01 -3.35625589e-01 8.57927352e-02
6.08272612e-01 2.21968397e-01 -3.00092809e-02 -5.69498241e-01
9.12650108e-01 8.76395166e-01 1.13381815e+00 -5.12167692e-01
4.43643898e-01 6.23468161e-01 -3.83161783e-01 -7.66441166e-01
-7.52993464e-01 -6.22801960e-01 -3.67975622e-01 -1.29122213e-01
1.13611281e+00 -1.31887329e+00 -7.86481440e-01 7.26094544e-01
-9.32419658e-01 -6.19557500e-01 -5.44179559e-01 2.28297770e-01
-5.93129039e-01 2.99153805e-01 -9.90198731e-01 -2.62860119e-01
-5.21917999e-01 -1.04786825e+00 1.42778766e+00 4.17636961e-01
-2.54659116e-01 -9.15250480e-01 -1.21689275e-01 9.76256967e-01
5.41075528e-01 -7.45152980e-02 6.50582969e-01 -5.95878601e-01
-3.96683365e-01 1.94550499e-01 -5.14186442e-01 4.97073770e-01
-2.90849328e-01 7.22236373e-03 -1.53456163e+00 -1.98603868e-01
-5.27381599e-01 -8.97466123e-01 1.20129585e+00 4.51327920e-01
1.29413617e+00 -8.39338601e-02 5.27480803e-02 1.05145776e+00
9.90705013e-01 -1.47935092e-01 6.79536045e-01 3.99830073e-01
7.45929539e-01 2.99591094e-01 2.66887367e-01 1.42858505e-01
8.51258874e-01 4.60606664e-01 6.49542093e-01 -4.73537683e-01
-4.20631856e-01 -6.32307753e-02 4.07681733e-01 7.50873923e-01
1.75244346e-01 -3.18583757e-01 -8.92496705e-01 7.51457155e-01
-1.57992995e+00 -1.02369738e+00 4.68251146e-02 1.65288615e+00
9.42160785e-01 -2.59806812e-01 3.71599168e-01 1.03813872e-01
4.49658275e-01 1.27893224e-01 -6.82539940e-01 -6.05025351e-01
-3.79516602e-01 2.95944870e-01 -1.32503221e-02 4.70710188e-01
-1.41117561e+00 8.54812026e-01 6.11569405e+00 1.09131539e+00
-1.35226274e+00 4.82533842e-01 5.79253793e-01 -2.78073817e-01
-4.17630792e-01 -3.19599152e-01 -3.84061903e-01 3.21146458e-01
9.55118537e-01 3.80745620e-01 6.64063931e-01 3.79010618e-01
-2.49400690e-01 -4.07072604e-02 -9.31156576e-01 1.25209618e+00
8.29393491e-02 -1.35540450e+00 1.52581081e-01 -3.81424189e-01
4.80579644e-01 4.93595451e-01 4.09477055e-01 4.94573891e-01
-3.61265354e-02 -1.06146026e+00 1.00738204e+00 5.97101450e-01
1.25239384e+00 -5.11762321e-01 5.12146175e-01 -8.81146267e-02
-1.19005096e+00 -1.40346453e-01 1.47222072e-01 2.92879224e-01
2.39306644e-01 1.03085801e-01 -4.15370762e-01 6.42661870e-01
1.22604990e+00 6.51274741e-01 -6.11636996e-01 9.14921522e-01
-6.16945811e-02 5.86663544e-01 -5.97144365e-01 7.17381299e-01
1.48360237e-01 3.54454219e-01 5.66199064e-01 1.67206872e+00
1.05702713e-01 -1.95215762e-01 3.81819382e-02 6.63905084e-01
-2.53491193e-01 -1.93480417e-01 -1.55156702e-01 -1.83792859e-01
3.57838988e-01 1.46147954e+00 -5.24841309e-01 -4.48546737e-01
-4.17245865e-01 1.19406104e+00 -9.75483730e-02 6.68068051e-01
-1.10718143e+00 -3.60895306e-01 6.99519396e-01 -6.14978373e-02
2.40011364e-01 1.96269490e-02 -2.55962521e-01 -1.27903795e+00
-3.82772446e-01 -1.22508514e+00 8.78308237e-01 -1.45522380e+00
-1.52949274e+00 6.80399716e-01 2.94044148e-02 -8.91084611e-01
-1.24760844e-01 -5.85102618e-01 -3.57644737e-01 9.33356345e-01
-1.54237163e+00 -1.77201653e+00 -4.39774990e-01 1.36595392e+00
4.43434596e-01 -4.90391068e-02 7.92787552e-01 6.26367509e-01
-3.89756650e-01 7.44990230e-01 -3.58242542e-01 1.63391590e-01
1.03387415e+00 -1.13512170e+00 2.01921865e-01 4.02568877e-01
2.17520416e-01 3.57620835e-01 2.82961071e-01 -2.13000253e-01
-1.32360327e+00 -9.47672963e-01 3.00596267e-01 -3.15286189e-01
9.99190748e-01 -2.79274136e-01 -8.58902335e-01 6.68085456e-01
6.13250077e-01 3.02502960e-02 8.66394341e-01 1.80080011e-01
-9.45412636e-01 -1.61150619e-01 -1.02972591e+00 3.74300361e-01
8.72846067e-01 -1.25158679e+00 -4.54608679e-01 1.79167524e-01
9.04727757e-01 -6.00166202e-01 -1.10873747e+00 3.94655645e-01
6.33365571e-01 -7.21784949e-01 1.26722479e+00 -5.34039617e-01
4.20251012e-01 -2.52277732e-01 -3.64297569e-01 -1.03094029e+00
-1.73564225e-01 -4.40864354e-01 6.60930648e-02 1.54147315e+00
2.90637732e-01 -5.54805994e-01 2.09063321e-01 2.26911545e-01
-3.93062353e-01 -3.00031811e-01 -1.09316862e+00 -4.84852791e-01
-2.48836838e-02 -9.33547914e-01 5.62122941e-01 1.13033056e+00
-1.26458749e-01 5.04889369e-01 -8.67205337e-02 3.00431103e-01
3.20578337e-01 9.38250348e-02 6.07661188e-01 -7.81189561e-01
-6.50660336e-01 -5.42544365e-01 -2.52462596e-01 -9.40811872e-01
4.68657687e-02 -1.06218660e+00 -2.18584940e-01 -1.72774506e+00
2.25246400e-01 -1.74329311e-01 -5.21628201e-01 1.08503354e+00
1.41158894e-01 8.59771371e-01 5.04249334e-01 1.26539856e-01
-9.72481251e-01 3.56963754e-01 1.21320653e+00 -3.65498066e-01
1.96831990e-02 -3.37411284e-01 -7.37971425e-01 7.58951306e-01
7.53871560e-01 1.42158195e-02 -3.36320639e-01 -8.11332047e-01
2.98577338e-01 -6.25387654e-02 1.06210053e+00 -9.58272755e-01
1.25890851e-01 1.65201306e-01 4.12018895e-01 -5.35323203e-01
8.30511510e-01 -7.57703483e-01 2.26223484e-01 -1.93783641e-01
-4.29923654e-01 1.23379948e-02 7.09623337e-01 2.92228669e-01
-4.89986956e-01 1.89067218e-02 6.86545968e-01 -1.11306876e-01
-1.02480221e+00 2.23120898e-01 -3.74683678e-01 1.78365812e-01
5.02337158e-01 -5.77701256e-02 -7.51850784e-01 -6.57977283e-01
-1.18410838e+00 4.40759271e-01 2.84645736e-01 6.49562359e-01
5.61745524e-01 -1.31978202e+00 -4.54323262e-01 6.13601170e-02
2.36251086e-01 -2.39577331e-02 9.15750086e-01 9.38673437e-01
-2.89708227e-01 1.90059796e-01 -3.16596836e-01 -8.59814465e-01
-1.22678685e+00 4.50106651e-01 3.98946851e-01 -4.87962104e-02
-2.57866889e-01 1.11938703e+00 8.82348567e-02 -6.41654789e-01
3.77071321e-01 -2.93873072e-01 3.94462198e-02 3.67908180e-01
5.01227260e-01 2.44676277e-01 9.28516835e-02 -7.85578132e-01
-4.40330297e-01 5.31680882e-01 9.34519917e-02 -4.33642477e-01
1.21666956e+00 -4.00213599e-01 -1.28228053e-01 5.14894068e-01
1.24435270e+00 -3.30361456e-01 -1.20999026e+00 -1.81609571e-01
-5.75425267e-01 9.24602374e-02 1.40797511e-01 -1.27489614e+00
-1.25581884e+00 1.25556707e+00 9.21985328e-01 2.49025315e-01
1.68269074e+00 3.52082998e-01 9.38309252e-01 1.10649772e-01
-4.76286300e-02 -1.12820411e+00 3.46516281e-01 7.00864375e-01
1.06952286e+00 -1.24202168e+00 -5.02425015e-01 -2.40636617e-02
-8.53206336e-01 9.25305665e-01 6.29404664e-01 3.08298707e-01
5.99402785e-01 3.70031744e-01 3.87039751e-01 -1.80226222e-01
-8.13113630e-01 -7.06432343e-01 5.95205188e-01 8.27205539e-01
3.06847721e-01 -8.64893347e-02 6.38910115e-01 7.45250046e-01
-6.88590184e-02 -6.53248876e-02 2.06774939e-02 7.39215970e-01
-1.18933670e-01 -8.10080528e-01 -5.86596847e-01 8.52468684e-02
-4.64230120e-01 -3.61692995e-01 -3.25816840e-01 6.48194373e-01
3.42079341e-01 1.05717623e+00 2.27312565e-01 -4.25875127e-01
3.50143284e-01 2.67009944e-01 4.68631506e-01 -3.14251125e-01
-9.16617513e-01 5.05487442e-01 9.53585431e-02 -6.92260563e-01
-5.48123240e-01 -3.63314837e-01 -1.34180224e+00 -2.24227861e-01
-8.45691189e-02 -3.14402193e-01 7.45257795e-01 7.13242173e-01
7.36765027e-01 6.60762131e-01 1.63416624e-01 -1.07138646e+00
-6.31368384e-02 -9.47162688e-01 -4.15108323e-01 3.69401455e-01
3.75417531e-01 -3.79997015e-01 -2.28354752e-01 2.88441002e-01]
|
[10.8817720413208, 1.5410795211791992]
|
7f2739a8-a3ea-4932-a9d1-7059eeb3e5bd
|
perspective-purposeful-failure-in-artificial
|
2102.12076
| null |
https://arxiv.org/abs/2102.12076v1
|
https://arxiv.org/pdf/2102.12076v1.pdf
|
Perspective: Purposeful Failure in Artificial Life and Artificial Intelligence
|
Complex systems fail. I argue that failures can be a blueprint characterizing living organisms and biological intelligence, a control mechanism to increase complexity in evolutionary simulations, and an alternative to classical fitness optimization. Imitating biological successes in Artificial Life and Artificial Intelligence can be misleading; imitating failures offers a path towards understanding and emulating life it in artificial systems.
|
['Lana Sinapayen']
|
2021-02-24
| null | null | null | null |
['artificial-life']
|
['miscellaneous']
|
[-2.96200975e-03 1.02770358e-01 2.53906816e-01 2.72683471e-01
7.53309488e-01 -5.93007505e-01 9.14555788e-01 -1.93886891e-01
-3.91856521e-01 9.66323733e-01 -1.93281636e-01 -5.84316194e-01
-3.61845642e-01 -8.60612631e-01 -3.12577963e-01 -8.08659315e-01
-1.23189658e-01 3.00672889e-01 -7.86778629e-02 -8.12260807e-01
5.88092387e-01 5.90614617e-01 -1.52989841e+00 -5.98135948e-01
9.44453359e-01 1.53522149e-01 1.72535524e-01 1.22216845e+00
2.38109931e-01 7.52007723e-01 -1.09546709e+00 -2.80407518e-01
4.83489633e-02 -1.04150224e+00 -6.61454797e-01 -2.80565381e-01
-5.79325914e-01 5.49099505e-01 -3.38463992e-01 6.25178456e-01
4.95440066e-01 -2.58123875e-01 6.76304817e-01 -1.24766064e+00
-8.22088957e-01 3.09287041e-01 7.32813999e-02 2.43170530e-01
3.06917846e-01 8.80241811e-01 6.50054216e-01 -2.72395551e-01
4.75518823e-01 1.52855396e+00 1.02514589e+00 7.89444089e-01
-1.33876634e+00 1.36648551e-01 -4.37547803e-01 -2.71484047e-01
-1.40756977e+00 -6.63830519e-01 -3.45232263e-02 -4.99248773e-01
1.29629755e+00 6.96384668e-01 1.40677643e+00 7.43885756e-01
1.19317877e+00 2.15187490e-01 8.17460775e-01 -3.49335730e-01
4.69447672e-01 -3.34519863e-01 -1.61764681e-01 6.85359895e-01
8.35847437e-01 8.94632757e-01 -5.74398160e-01 -4.73181665e-01
1.00753474e+00 3.45554352e-02 -2.31412143e-01 4.47562754e-01
-1.33458781e+00 2.81758636e-01 3.90424997e-01 4.85817075e-01
-3.48536432e-01 8.96218598e-01 4.19539474e-02 5.50003171e-01
-1.45186797e-01 1.59929466e+00 -8.23787153e-01 -5.44345140e-01
-1.68591291e-01 1.35569602e-01 1.11012948e+00 3.60691428e-01
3.69326353e-01 3.94166350e-01 4.98904020e-01 3.78270000e-01
2.81334758e-01 6.20583773e-01 7.65489459e-01 -1.07866812e+00
-9.30383921e-01 9.05074239e-01 7.47916400e-02 -8.56436968e-01
-6.72065556e-01 -7.25462615e-01 -5.78333735e-01 6.16187334e-01
2.96285003e-01 -1.16472162e-01 -6.98913872e-01 1.72883737e+00
2.37715468e-02 3.22920084e-03 -1.30186398e-02 4.31834131e-01
2.25444481e-01 5.30217290e-01 -7.08424821e-02 -3.99951696e-01
7.32407331e-01 -7.44824171e-01 -5.08925140e-01 -3.28960568e-01
7.66169667e-01 -5.33251226e-01 8.92637789e-01 5.27549744e-01
-1.26787126e+00 -3.30348730e-01 -1.30838287e+00 5.11083364e-01
-4.50694978e-01 -7.66009212e-01 1.09752488e+00 1.28324723e+00
-1.37919486e+00 9.11496699e-01 -1.08554971e+00 -7.05475092e-01
7.57794678e-02 4.02583063e-01 1.59103692e-01 6.05602384e-01
-7.76811600e-01 1.40753651e+00 3.35383087e-01 -1.73119858e-01
-9.33990717e-01 -4.72271174e-01 -3.68078440e-01 -1.17800154e-01
2.03284398e-01 -1.26725054e+00 8.31354499e-01 -1.00365543e+00
-1.34721971e+00 9.03869927e-01 -1.05392821e-01 -3.00682217e-01
3.90343256e-02 1.36544064e-01 -3.87969285e-01 -5.79128504e-01
-3.59653324e-01 3.21828246e-01 5.63828409e-01 -1.16678131e+00
-2.53515840e-01 -4.39146876e-01 -1.87191278e-01 1.06889144e-01
2.05044951e-02 -8.87104571e-02 4.12645161e-01 -7.07607031e-01
9.38013941e-02 -9.72154915e-01 -6.05989635e-01 -1.88172892e-01
1.19810971e-03 -2.79343188e-01 2.18881741e-01 -1.97348297e-01
1.32513487e+00 -1.72311580e+00 7.62814105e-01 -5.95433563e-02
6.10774755e-01 -6.89666485e-03 -4.94722873e-02 6.94094777e-01
3.86119723e-01 9.16821599e-01 -1.00966841e-01 3.77223730e-01
1.43313989e-01 5.15289843e-01 2.97331184e-01 4.76867557e-01
5.06917417e-01 1.31044304e+00 -9.20757055e-01 -1.36204004e-01
2.21232027e-02 3.91642570e-01 -3.22645634e-01 -1.88933015e-02
-8.37136731e-02 4.68556762e-01 -1.39084518e-01 9.21508133e-01
-1.05364680e-01 -3.95691246e-01 8.25906545e-02 7.34489620e-01
-2.76540220e-01 2.15130895e-01 -4.77913976e-01 1.01301682e+00
-2.14400634e-01 5.31138659e-01 -3.49708706e-01 -7.87285388e-01
7.46055901e-01 2.42312163e-01 2.28438437e-01 -8.11438918e-01
2.77707636e-01 4.60125923e-01 9.72573161e-01 -2.90924758e-01
2.05849886e-01 -2.02945054e-01 -1.54606685e-01 6.77169859e-01
1.17516965e-02 -1.07381010e+00 1.10463530e-01 -1.17754832e-01
1.51778889e+00 1.46930233e-01 4.04350489e-01 -7.18038797e-01
4.13564295e-01 1.75983116e-01 7.20716238e-01 1.05565965e+00
-4.11924243e-01 1.00301094e-01 1.12917081e-01 -8.39646459e-01
-1.28578722e+00 -8.82522047e-01 -1.63539767e-01 9.38396752e-01
1.88945070e-01 -2.88768142e-01 -7.33503938e-01 1.17487259e-01
-7.49550015e-02 6.50826871e-01 -8.45767856e-01 -7.84230351e-01
-5.89214742e-01 -1.33411157e+00 7.10281968e-01 -1.17791824e-01
1.07916944e-01 -1.30188990e+00 -1.01819360e+00 2.79782683e-01
4.01327968e-01 -2.45052978e-01 1.38071060e-01 5.64635277e-01
-9.21234787e-01 -9.69667852e-01 -1.86228886e-01 -5.72131574e-01
5.24004817e-01 1.53127730e-01 1.70421088e+00 1.55017769e+00
-7.77391493e-01 2.27755919e-01 -2.09013432e-01 -5.10685265e-01
-1.16950226e+00 -1.10938154e-01 6.44862294e-01 -1.13427055e+00
9.90957171e-02 -7.51593053e-01 -6.08089328e-01 6.27910614e-01
-8.91272664e-01 -1.11150727e-01 1.33071885e-01 1.16946959e+00
7.52089694e-02 2.72843421e-01 7.25200534e-01 -3.24016392e-01
7.51066267e-01 -4.46329534e-01 -2.65979320e-01 5.22496223e-01
-1.05451882e+00 -6.50841370e-02 7.20113575e-01 -2.35248119e-01
-3.96515816e-01 -3.66908938e-01 -4.14289311e-02 7.43658066e-01
-1.90169085e-02 3.24403822e-01 -4.43964750e-02 -5.08601010e-01
8.17723393e-01 4.04831678e-01 1.48671791e-01 -1.04226656e-01
1.65619135e-01 2.50638008e-01 3.30928862e-01 -6.63755000e-01
5.90115547e-01 1.60063818e-01 2.42484644e-01 -1.06503427e+00
1.01275273e-01 5.00442624e-01 -4.79615003e-01 -4.11526322e-01
4.11454082e-01 -1.98587239e-01 -1.08680701e+00 4.71247464e-01
-7.53645480e-01 -3.14755112e-01 -6.53735161e-01 1.85446162e-02
-7.94892728e-01 -1.24194063e-01 -3.02591264e-01 -1.06647348e+00
-5.85886277e-03 -8.58326674e-01 7.08123922e-01 8.73041451e-01
-6.64880574e-01 -1.26823735e+00 4.35526729e-01 -6.14040941e-02
5.93162417e-01 4.49600160e-01 8.01134706e-01 -4.88660604e-01
-3.08862269e-01 9.05292202e-03 5.82826734e-01 -1.94436550e-01
2.09977239e-01 7.18126595e-01 -6.12336814e-01 -2.08467096e-01
3.08475196e-01 -1.96034059e-01 4.18991864e-01 2.14680851e-01
3.27977031e-01 -2.84055650e-01 -5.21803141e-01 6.30770266e-01
1.44135678e+00 8.36785853e-01 7.12389350e-01 6.86053932e-01
7.45358691e-02 7.53082514e-01 2.09855422e-01 5.82263112e-01
1.00398287e-01 2.45138943e-01 2.40776807e-01 1.85220838e-02
1.00056432e-01 3.64538282e-02 4.19450939e-01 1.11890495e+00
-1.60547405e-01 -2.92704105e-01 -1.49018753e+00 3.43333840e-01
-1.55332077e+00 -1.03154528e+00 -1.63537279e-01 2.16753936e+00
8.02945137e-01 3.22727889e-01 4.46211278e-01 1.88377365e-01
4.89088178e-01 -6.67216539e-01 -8.33804011e-01 -5.39564788e-01
-8.28978121e-01 2.83073515e-01 1.09146900e-01 5.78254938e-01
-2.78279483e-01 1.04907084e+00 9.46373653e+00 4.14143443e-01
-8.19541931e-01 -2.11651437e-02 6.76594079e-01 7.63920397e-02
-4.69776005e-01 1.56032726e-01 -2.53556699e-01 5.79836965e-01
1.25082326e+00 -7.06908941e-01 9.57594156e-01 9.82081965e-02
3.80727530e-01 -4.04785037e-01 -1.19206297e+00 4.90157872e-01
-3.07314754e-01 -1.26658249e+00 -2.34262764e-01 2.43988410e-01
5.54042280e-01 -1.57702148e-01 4.41737436e-02 8.35887715e-03
9.21490550e-01 -1.76041925e+00 6.29507303e-01 7.98653901e-01
4.01703566e-02 -6.23809397e-01 5.07794023e-01 6.11215532e-01
-5.20175099e-01 -1.75748795e-01 -4.24562812e-01 -1.00195646e+00
1.32009774e-01 1.47952005e-01 -3.54356080e-01 -9.13959891e-02
6.19676232e-01 3.40410173e-01 -1.02293754e+00 1.25046003e+00
3.53205949e-03 5.74898481e-01 -5.11076629e-01 -6.08173609e-01
-1.12779349e-01 -2.43872538e-01 9.00704920e-01 7.75943756e-01
1.21525221e-01 2.34778240e-01 -3.99318546e-01 1.20937216e+00
7.05171585e-01 -4.54544902e-01 -7.65676260e-01 -9.42856550e-01
7.48591125e-01 7.49421537e-01 -1.17856860e+00 -3.30476761e-01
2.14121237e-01 8.79076123e-01 -3.07310522e-01 1.34576589e-01
-6.89741731e-01 -2.08122537e-01 9.81024861e-01 -3.25725347e-01
-5.33173501e-01 -5.86608768e-01 -8.85067761e-01 -9.26998794e-01
-7.40598381e-01 -1.11977792e+00 -2.14698926e-01 -6.85436487e-01
-1.16607988e+00 3.38015854e-01 -7.02850461e-01 -4.76734281e-01
-2.50383466e-01 -6.84079111e-01 -7.32709467e-01 5.82277298e-01
-4.71285820e-01 -5.92095673e-01 -5.53344786e-02 -1.24295168e-01
2.53350854e-01 -2.49671757e-01 8.34365726e-01 -5.50429881e-01
-9.31421936e-01 4.45782661e-01 4.23412740e-01 -7.75889695e-01
-1.08699590e-01 -1.22886062e+00 1.01291275e+00 7.03242838e-01
5.10514947e-03 1.07309163e+00 1.11436152e+00 -7.63540983e-01
-1.84175324e+00 -1.54769689e-01 3.55202734e-01 -8.36929739e-01
6.81127071e-01 -1.25461789e-02 -6.55608475e-01 1.47475109e-01
2.95298845e-01 -7.50317335e-01 6.37902319e-01 1.78941991e-02
1.53812855e-01 5.15803397e-01 -1.07148683e+00 1.07893658e+00
1.37936127e+00 -1.79775208e-01 -5.32324553e-01 2.68273205e-01
1.06732738e+00 3.01652193e-01 -1.00662446e+00 2.68082798e-01
9.52833176e-01 -1.01562035e+00 1.17867160e+00 -7.11922944e-01
1.50191560e-01 -5.42954683e-01 2.62915604e-02 -1.20715761e+00
-6.92795038e-01 -1.18073297e+00 7.51232058e-02 6.82971060e-01
4.69307214e-01 -1.11911774e+00 5.46574533e-01 6.57833338e-01
3.52762528e-02 -5.33989191e-01 -7.46248841e-01 -1.14234114e+00
4.23155665e-01 2.08203569e-01 7.63631344e-01 1.03420782e+00
1.94521159e-01 6.23192638e-02 2.31725052e-01 -2.09558293e-01
5.18553317e-01 -6.61261678e-01 6.82692647e-01 -1.41673875e+00
-5.45894504e-01 -1.29019904e+00 -7.40728438e-01 -3.54561716e-01
-3.31335008e-01 -4.58193660e-01 -1.23621132e-02 -1.00849581e+00
-7.10111707e-02 -5.05480051e-01 -2.42232502e-01 3.84717137e-02
-4.19189274e-01 2.97589183e-01 2.44940966e-01 4.02497172e-01
-3.10861707e-01 3.07285011e-01 1.06741512e+00 1.67745367e-01
-1.82077363e-01 -5.35240829e-01 -9.71564412e-01 8.45265329e-01
9.46728587e-01 -6.19296372e-01 -1.14916630e-01 -4.90827300e-02
9.21914756e-01 -7.22409710e-02 2.73160338e-01 -1.07918990e+00
-2.93703675e-02 -7.99513161e-01 3.37062776e-01 2.23461568e-01
8.25545117e-02 -6.01257741e-01 8.62335622e-01 1.45862460e+00
1.02667369e-01 5.01051128e-01 1.26426518e-01 1.28996029e-01
3.72108102e-01 -3.85339916e-01 7.86120594e-01 -5.04456460e-01
-4.07949924e-01 -3.40110093e-01 -8.92268836e-01 5.52563667e-02
1.09374666e+00 -8.33291352e-01 -8.04055333e-01 1.20562064e-02
-5.53396463e-01 3.84423733e-01 1.58711123e+00 3.66312563e-01
5.71784437e-01 -9.27823305e-01 -6.65497720e-01 8.54782388e-02
-2.56943911e-01 -7.39255488e-01 -3.88565123e-01 6.81695819e-01
-1.35683143e+00 2.01806471e-01 -5.90273201e-01 -4.41176295e-01
-9.08450246e-01 4.77638453e-01 1.06420851e+00 3.18231642e-01
-4.83079031e-02 1.31928146e+00 1.84034243e-01 -3.68869931e-01
-1.12571210e-01 2.58808404e-01 1.20729603e-01 -5.37750125e-01
4.12876278e-01 4.28767025e-01 -5.86059034e-01 -4.11861449e-01
-6.41245544e-01 5.98561943e-01 4.38469678e-01 -3.09642941e-01
1.48165905e+00 -3.33614439e-01 -8.69423866e-01 9.30185318e-01
2.18031168e-01 -3.74811858e-01 -5.68106890e-01 5.77676058e-01
4.53704506e-01 -3.89426082e-01 -1.80674359e-01 -1.19867945e+00
-5.67612126e-02 6.15482807e-01 5.31336308e-01 8.46250534e-01
1.07954729e+00 -1.71578333e-01 3.65384221e-01 7.59454846e-01
7.38192856e-01 -1.01842880e+00 1.18365951e-01 4.58133966e-01
9.64255035e-01 -1.03237391e+00 1.51007578e-01 1.60261497e-01
-4.45618667e-02 8.94902945e-01 7.83685565e-01 -3.35068822e-01
6.63244486e-01 6.50729835e-01 -2.96855360e-01 -2.34010190e-01
-1.28507042e+00 -2.36819744e-01 -2.27668703e-01 7.09012032e-01
9.22853351e-01 5.35467267e-02 -9.61354017e-01 2.19298705e-01
-4.64841008e-01 -2.12875918e-01 5.77916324e-01 1.05201924e+00
-9.25515413e-01 -1.35141730e+00 -7.12619901e-01 9.16778222e-02
-5.30536234e-01 -6.03173524e-02 -1.30314016e+00 6.14586532e-01
3.95505935e-01 1.03057671e+00 -1.44310564e-01 -7.55454600e-01
-1.13081589e-01 1.15770645e-01 6.46265984e-01 -4.07096177e-01
-1.26932883e+00 -2.57808149e-01 -6.32670969e-02 -3.50954652e-01
-1.10007010e-01 -4.65857625e-01 -1.16905725e+00 -1.07294464e+00
-6.12003922e-01 2.98385203e-01 6.87835813e-01 6.97010159e-01
4.14716721e-01 6.66427493e-01 5.38572192e-01 -6.18450224e-01
-1.69637084e-01 -6.70846224e-01 -3.70648265e-01 1.68189585e-01
2.13215947e-01 -4.76586878e-01 -6.11064434e-01 1.08335853e-01]
|
[5.5686516761779785, 4.150617599487305]
|
67ea0134-54ce-4305-8d86-decd4902c10a
|
speaking-multiple-languages-affects-the-moral
|
2211.07733
| null |
https://arxiv.org/abs/2211.07733v2
|
https://arxiv.org/pdf/2211.07733v2.pdf
|
Speaking Multiple Languages Affects the Moral Bias of Language Models
|
Pre-trained multilingual language models (PMLMs) are commonly used when dealing with data from multiple languages and cross-lingual transfer. However, PMLMs are trained on varying amounts of data for each language. In practice this means their performance is often much better on English than many other languages. We explore to what extent this also applies to moral norms. Do the models capture moral norms from English and impose them on other languages? Do the models exhibit random and thus potentially harmful beliefs in certain languages? Both these issues could negatively impact cross-lingual transfer and potentially lead to harmful outcomes. In this paper, we (1) apply the MoralDirection framework to multilingual models, comparing results in German, Czech, Arabic, Chinese, and English, (2) analyse model behaviour on filtered parallel subtitles corpora, and (3) apply the models to a Moral Foundations Questionnaire, comparing with human responses from different countries. Our experiments demonstrate that, indeed, PMLMs encode differing moral biases, but these do not necessarily correspond to cultural differences or commonalities in human opinions. We release our code and models.
|
['Kristian Kersting', 'Alexander Fraser', 'Constantin A. Rothkopf', 'Jindřich Libovický', 'Patrick Schramowski', 'Björn Deiseroth', 'Katharina Hämmerl']
|
2022-11-14
| null | null | null | null |
['cross-lingual-transfer']
|
['natural-language-processing']
|
[-2.93473035e-01 2.53928751e-01 -2.33737603e-01 -3.76771778e-01
-5.62148035e-01 -7.29461551e-01 7.84781635e-01 3.02678794e-01
-1.01160657e+00 1.09874547e+00 5.63849747e-01 -4.52652931e-01
1.12455651e-01 -5.43135107e-01 -5.78056633e-01 -5.72659910e-01
2.60232598e-01 6.57619715e-01 -1.57704636e-01 -5.79405665e-01
4.75542665e-01 6.70247227e-02 -9.66044247e-01 3.91124457e-01
1.18041313e+00 -4.49717641e-02 -2.06405744e-01 2.17148095e-01
1.35084316e-01 1.13309562e+00 -4.95394886e-01 -1.09268558e+00
-1.16843786e-02 -3.30922037e-01 -9.67848301e-01 -2.61200309e-01
3.53039950e-01 -6.24900050e-02 4.90219921e-01 1.20655322e+00
4.21862632e-01 -2.01047257e-01 1.15566683e+00 -1.00728202e+00
-9.98845637e-01 8.98968756e-01 -4.27036732e-01 -1.89601555e-01
4.71020907e-01 5.66862561e-02 1.04964221e+00 -5.94447672e-01
8.77390027e-01 1.77950060e+00 8.51042867e-01 7.64572620e-01
-1.31360757e+00 -5.35310209e-01 1.61664948e-01 -2.67397482e-02
-9.36201990e-01 -4.19457257e-01 5.09606719e-01 -8.03814828e-01
7.48087406e-01 -1.44012943e-01 3.99731040e-01 1.68147230e+00
5.80348969e-01 6.09474421e-01 1.84537005e+00 -4.36723113e-01
4.08527022e-03 8.41970801e-01 -7.98090175e-02 1.80307686e-01
2.28975654e-01 2.02718571e-01 -4.86737400e-01 -3.52750123e-01
9.19591635e-02 -7.95453608e-01 -1.62987053e-01 -2.38689706e-01
-1.27768934e+00 1.22418416e+00 1.24127857e-01 5.17496705e-01
-2.85078943e-01 -1.37647986e-01 6.80468619e-01 7.68597424e-01
4.41848606e-01 5.38194895e-01 -6.57325685e-01 -7.04989806e-02
-4.86150384e-01 3.21906805e-01 9.42314684e-01 6.01922274e-01
6.51326776e-01 -2.45334536e-01 3.08627754e-01 1.13699770e+00
3.60485166e-01 6.39879346e-01 6.07690811e-01 -1.06853521e+00
2.99890250e-01 1.91992715e-01 2.25775927e-01 -1.11929691e+00
-5.14353335e-01 -1.05282694e-01 -4.73235339e-01 2.95158654e-01
7.04250276e-01 -3.15824687e-01 -1.58809364e-01 2.18943954e+00
-6.72118291e-02 -8.07009280e-01 4.49863076e-01 7.83205390e-01
2.52539545e-01 4.41049814e-01 5.92783630e-01 -1.26452625e-01
1.18338430e+00 -3.62299204e-01 -3.51038307e-01 -7.20182836e-01
1.35445285e+00 -1.03058517e+00 1.36374259e+00 4.80679601e-01
-1.07772541e+00 -1.79184839e-01 -6.06466949e-01 -1.76138937e-01
-5.33768356e-01 -7.02083036e-02 5.03652692e-01 8.61315131e-01
-1.14439023e+00 5.21963835e-01 -3.46009225e-01 -8.68136406e-01
4.05040830e-02 6.21446855e-02 -3.75454903e-01 -7.02164173e-02
-1.70668375e+00 1.61219561e+00 2.12103337e-01 -2.44819000e-01
-4.03222531e-01 -2.34294504e-01 -7.17019856e-01 -5.73590875e-01
-1.46236464e-01 -3.77347589e-01 9.74001050e-01 -1.67030370e+00
-1.19226074e+00 1.58351672e+00 7.68155605e-02 -4.36613634e-02
6.71671927e-01 -3.58841002e-01 -4.20408845e-01 -4.66885895e-01
6.28890216e-01 6.51600182e-01 2.71848112e-01 -1.28210533e+00
-6.40038252e-01 -2.12087959e-01 2.19565988e-01 4.23296601e-01
-3.16988051e-01 6.81752264e-01 3.65593880e-01 -6.82577968e-01
-5.53016603e-01 -1.05055737e+00 3.88676450e-02 -3.56728315e-01
-6.71191094e-03 -4.90312725e-01 -6.23065196e-02 -7.56016672e-01
8.40213060e-01 -2.07803106e+00 2.76846170e-01 1.01025350e-01
-5.27584970e-01 -8.33179057e-02 -7.01288283e-02 5.89092433e-01
1.03731826e-01 4.28080499e-01 -2.74705023e-01 5.37677854e-02
2.36729980e-01 5.05482018e-01 -2.36255407e-01 7.27118611e-01
2.03784257e-01 5.57583690e-01 -9.37494218e-01 -6.95101500e-01
-1.67172909e-01 3.62562448e-01 -1.12463975e+00 -2.75280803e-01
1.30841702e-01 5.05392373e-01 -7.14057088e-02 1.94362834e-01
3.41927230e-01 4.14712876e-01 4.20977712e-01 2.32971564e-01
-1.96388319e-01 8.03801835e-01 -6.37306929e-01 1.41460896e+00
-5.95652640e-01 6.36329591e-01 2.20549852e-01 -8.06130528e-01
6.97033167e-01 3.50499868e-01 -8.23024511e-02 -7.29836166e-01
1.89277142e-01 7.62421608e-01 5.05282521e-01 -5.00667632e-01
3.72026443e-01 -9.64201689e-01 -4.98319179e-01 7.91902006e-01
-1.24031618e-01 -3.06247979e-01 1.94391280e-01 -2.04178272e-03
3.71508271e-01 2.76512176e-01 3.09060395e-01 -1.09704208e+00
8.90438378e-01 3.99550050e-02 9.18807685e-01 4.86300111e-01
-5.67889750e-01 -5.23485318e-02 8.86206686e-01 -3.31890047e-01
-8.83486032e-01 -8.14467251e-01 -4.85642433e-01 1.45644319e+00
-6.75261542e-02 -6.52530640e-02 -7.62672186e-01 -7.74335682e-01
-1.08660705e-01 1.46105564e+00 -6.33154035e-01 -2.07771629e-01
-5.09236991e-01 -8.56122494e-01 7.05785930e-01 1.69484049e-01
-3.23716663e-02 -1.22382510e+00 -5.83044767e-01 8.61903727e-02
-3.74329835e-01 -8.63818944e-01 -1.45884484e-01 1.32753417e-01
-5.72577477e-01 -7.85748541e-01 -4.20216888e-01 -8.15715492e-01
4.48636144e-01 -3.49042237e-01 1.49653161e+00 9.08403322e-02
5.51523566e-01 2.98909903e-01 -3.50060195e-01 -7.24028945e-01
-1.13416040e+00 1.34237632e-01 4.46674168e-01 -4.29862499e-01
9.40019369e-01 -3.06275576e-01 -7.39517016e-03 3.23284090e-01
-7.89689481e-01 -1.89307049e-01 2.99468189e-01 6.79278970e-01
-2.04993501e-01 -3.01129997e-01 6.67625546e-01 -1.19679475e+00
1.00570595e+00 -6.33125246e-01 -2.60156870e-01 2.54190713e-01
-5.70408762e-01 -2.78926380e-02 7.31821835e-01 -2.17171937e-01
-1.13504457e+00 -6.88831747e-01 -2.04027042e-01 5.12855589e-01
-3.50246519e-01 5.93782842e-01 8.68777744e-03 3.09160829e-01
9.07513678e-01 -3.31514210e-01 -4.72490191e-02 -2.85082340e-01
2.58618265e-01 9.29439485e-01 5.38736731e-02 -1.31783795e+00
7.35343397e-01 3.70411515e-01 -5.56427181e-01 -8.55312228e-01
-8.75437796e-01 1.90968245e-01 -7.26520240e-01 -1.51453421e-01
1.03539813e+00 -1.09230912e+00 -2.91630745e-01 3.59704882e-01
-1.13296628e+00 -6.88733459e-01 1.38141274e-01 7.87732899e-01
-5.65711737e-01 2.05306366e-01 -9.28986967e-01 -7.44420409e-01
-7.08996644e-03 -1.16424727e+00 5.92243910e-01 -2.43966773e-01
-9.97537792e-01 -1.68129122e+00 3.78331065e-01 3.16796869e-01
2.85484344e-01 5.51702306e-02 1.47902858e+00 -7.01729238e-01
5.07366657e-01 1.10495165e-01 1.31696969e-01 6.12043262e-01
3.86286303e-02 1.12810232e-01 -7.68821776e-01 -4.32550669e-01
2.17562899e-01 -1.11881924e+00 4.28862244e-01 3.46423499e-02
9.11132991e-02 -3.44089687e-01 1.30399659e-01 2.18731552e-01
1.26623845e+00 -6.49647340e-02 4.73008841e-01 7.08668232e-01
4.10708845e-01 1.53808045e+00 6.13811970e-01 -7.56895617e-02
8.48351717e-01 4.27142143e-01 -8.41893777e-02 5.72655462e-02
6.27162516e-01 -8.26359391e-02 1.28536820e+00 1.24467492e+00
-2.61591803e-02 1.22742683e-01 -1.13894784e+00 7.79705107e-01
-1.67728341e+00 -9.39808667e-01 -3.19256753e-01 2.06366086e+00
1.15457523e+00 7.67951608e-02 1.62995368e-01 -3.30625623e-01
4.83202875e-01 3.98145206e-02 -8.40489268e-02 -1.34645236e+00
-4.35719311e-01 -1.91633955e-01 3.96094352e-01 8.90295029e-01
-8.77313137e-01 1.12603354e+00 6.41831827e+00 4.12452132e-01
-1.06891131e+00 1.86492160e-01 7.05919445e-01 -1.58638861e-02
-6.66019261e-01 1.32669210e-01 -4.04905528e-01 2.61251181e-01
9.40098345e-01 -2.95367807e-01 3.93428504e-01 7.28202939e-01
1.23828530e-01 -1.20149247e-01 -1.32607663e+00 4.79459167e-01
1.71231449e-01 -2.06944615e-01 -2.49569997e-01 9.93816778e-02
7.43701696e-01 3.25505018e-01 4.57474552e-02 6.18584991e-01
7.77202427e-01 -1.09461129e+00 1.14506781e+00 1.09562881e-01
4.57304090e-01 -8.91645908e-01 9.75104153e-01 6.43482387e-01
-7.72415772e-02 1.94921479e-01 -6.84649408e-01 -4.61060345e-01
4.05003456e-03 2.13781036e-02 -2.75587857e-01 2.33417794e-01
7.35753715e-01 7.25594878e-01 -6.03543162e-01 2.14828807e-03
-6.51385605e-01 4.84556407e-01 -8.37380141e-02 -2.96707414e-02
4.97984290e-01 -5.12951851e-01 3.17156971e-01 1.46913731e+00
1.62028819e-01 -4.20617402e-01 -1.84607312e-01 7.50487626e-01
1.15124978e-01 6.05322003e-01 -9.23456907e-01 5.39127551e-02
6.96161389e-02 9.78883445e-01 -3.29400569e-01 -1.37630939e-01
-9.78262722e-01 9.40336764e-01 4.75551516e-01 2.96393543e-01
-7.25814641e-01 -5.55707291e-02 7.76572645e-01 2.50621028e-02
-4.67024803e-01 8.28415230e-02 -1.79889590e-01 -1.35018468e+00
-2.86262661e-01 -1.36930406e+00 3.46314162e-01 -6.68553829e-01
-1.89472783e+00 2.09079519e-01 5.08459508e-02 -7.07358003e-01
-4.77358609e-01 -1.04699969e+00 -2.29806200e-01 1.10824800e+00
-1.42495799e+00 -9.98504996e-01 6.06894016e-01 5.52231193e-01
-3.42161919e-04 -6.66001067e-02 9.03101802e-01 3.50640535e-01
-2.92021066e-01 5.42027175e-01 4.40841839e-02 2.71158606e-01
1.41619277e+00 -1.16942704e+00 1.44986838e-01 5.36315441e-01
-1.79011062e-01 8.42750371e-01 8.66479218e-01 -4.89881545e-01
-5.96452296e-01 -7.71621048e-01 1.51806355e+00 -9.03104067e-01
1.00157869e+00 -2.82045275e-01 -1.02974951e+00 1.00216496e+00
7.64527977e-01 -7.59202361e-01 1.06407189e+00 5.66255748e-01
-5.35601676e-01 3.06074828e-01 -1.10565937e+00 9.82981384e-01
8.63233864e-01 -6.40585542e-01 -1.02255094e+00 3.22934657e-01
2.71709055e-01 1.34674653e-01 -8.88553202e-01 1.11460179e-01
5.96840739e-01 -1.42030680e+00 5.16278684e-01 -9.68675196e-01
8.33027840e-01 1.52909264e-01 -3.62895817e-01 -1.79048204e+00
-4.44993943e-01 -2.72863060e-01 9.93769467e-01 1.41844428e+00
7.06706107e-01 -1.03390777e+00 1.97693128e-02 6.71299219e-01
2.08149582e-01 -4.75209683e-01 -9.64174509e-01 -6.14110947e-01
1.52225363e+00 -6.29011512e-01 2.69768924e-01 1.56868517e+00
2.65863717e-01 5.20288765e-01 -3.29800487e-01 -1.95497379e-01
6.62258863e-01 -2.76393920e-01 6.47620559e-01 -1.30214393e+00
-1.69905975e-01 -6.72226429e-01 -6.44276068e-02 -2.31458306e-01
9.17262912e-01 -1.17355812e+00 -1.19693978e-02 -1.31950498e+00
1.84440479e-01 -6.23960972e-01 -8.17370415e-02 2.97461092e-01
-2.47817203e-01 7.88146779e-02 2.12797076e-01 2.45755956e-01
-1.74352750e-01 3.69916111e-01 7.51740694e-01 3.45895857e-01
7.71967769e-02 -5.55088997e-01 -1.08993244e+00 1.32387292e+00
9.08114135e-01 -7.18593776e-01 -1.80770710e-01 -8.21039021e-01
9.04246330e-01 -4.45943922e-01 2.92381883e-01 -5.44540644e-01
-2.83080399e-01 -6.08062088e-01 8.39621797e-02 3.82481009e-01
-1.25530288e-01 -7.46809423e-01 -2.26062492e-01 6.90565050e-01
-4.55186158e-01 1.82997748e-01 2.15598464e-01 1.14611471e-02
-1.60312206e-01 -3.99560153e-01 8.99536967e-01 -3.78116816e-01
-2.66572595e-01 -4.62557435e-01 -8.57589304e-01 6.32545054e-01
9.01098847e-01 1.33773118e-01 -3.32326442e-01 -4.71135020e-01
-2.83554524e-01 2.21790880e-01 1.33940399e+00 6.62982583e-01
-1.38194934e-01 -1.48884082e+00 -1.16804314e+00 -1.56760097e-01
3.49477440e-01 -7.65596032e-01 -1.46770850e-01 1.12397408e+00
-4.82704997e-01 5.53107321e-01 -3.66439134e-01 -2.40514845e-01
-8.28876972e-01 4.75386500e-01 3.82434279e-01 -1.21873356e-01
2.45016124e-02 7.96484828e-01 5.73127031e-01 -1.19909000e+00
-3.36786419e-01 1.96935624e-01 -3.05704951e-01 5.05914450e-01
-8.16009715e-02 6.00326024e-02 -3.86401922e-01 -1.31388569e+00
-6.22524559e-01 7.28499413e-01 -9.95211080e-02 -6.03397191e-01
1.20971251e+00 -1.42769918e-01 -6.20010257e-01 1.00109375e+00
1.02727401e+00 6.37676179e-01 -6.42790556e-01 -6.41329736e-02
3.41793776e-01 -1.17248110e-01 -4.38827187e-01 -7.47424424e-01
-5.32619894e-01 9.70645905e-01 1.34255677e-01 -5.61068654e-02
7.39768803e-01 -8.40923563e-02 3.02034497e-01 1.86388075e-01
6.70623183e-01 -1.72655296e+00 -3.96168768e-01 6.57248497e-01
9.36350226e-01 -1.17829537e+00 -1.16328701e-01 1.75148278e-01
-1.28844392e+00 8.33651245e-01 5.66758573e-01 -2.21137181e-01
3.05760086e-01 4.63512428e-02 8.65628362e-01 3.75223048e-02
-1.07598722e+00 1.50647208e-01 -1.62209511e-01 6.64526105e-01
1.37380195e+00 2.26234093e-01 -9.68266666e-01 3.65425944e-01
-5.32957613e-01 -2.09224612e-01 8.46030533e-01 7.78049588e-01
2.95041818e-02 -1.41650271e+00 -6.66955173e-01 7.50202825e-03
-8.67389739e-01 -3.88142228e-01 -9.88803327e-01 1.10304880e+00
3.05610895e-01 1.07818580e+00 -8.11011493e-02 -2.18249455e-01
2.87519217e-01 4.10566092e-01 3.58967215e-01 -5.62553108e-01
-9.84174371e-01 1.59580912e-02 4.73230809e-01 -2.69872159e-01
-6.63200617e-01 -1.04599059e+00 -9.41480160e-01 -6.12347603e-01
1.04714714e-01 1.89356148e-01 2.88771421e-01 9.53516245e-01
7.67737906e-03 -1.62799787e-02 1.48616448e-01 -4.26328570e-01
-7.10589886e-01 -1.05118799e+00 -4.38913643e-01 7.58864760e-01
5.21418974e-02 -3.53324056e-01 -7.75000632e-01 -7.99221098e-02]
|
[9.851529121398926, 10.22383975982666]
|
107f64a6-0b73-4366-b53f-3cc3932413ec
|
a-convolutional-neural-network-approach-to
| null | null |
https://doi.org/10.1016/j.bspc.2019.101597
|
https://www.sciencedirect.com/science/article/pii/S1746809419301776/pdfft?md5=ca17956e278efdd4a39ec925adfa2b16&pid=1-s2.0-S1746809419301776-main.pdf
|
A convolutional neural network approach to detect congestive heart failure
|
Congestive Heart Failure (CHF) is a severe pathophysiological condition associated with high prevalence, high mortality rates, and sustained healthcare costs, therefore demanding efficient methods for its detection. Despite recent research has provided methods focused on advanced signal processing and machine learning, the potential of applying Convolutional Neural Network (CNN) approaches to the automatic detection of CHF has been largely overlooked thus far. This study addresses this important gap by presenting a CNN model that accurately identifies CHF on the basis of one raw electrocardiogram (ECG) heartbeat only, also juxtaposing existing methods typically grounded on Heart Rate Variability. We trained and tested the model on publicly available ECG datasets, comprising a total of 490,505 heartbeats, to achieve 100% CHF detection accuracy. Importantly, the model also identifies those heartbeat sequences and ECG’s morphological characteristics which are class-discriminative and thus prominent for CHF detection. Overall, our contribution substantially advances the current methodology for detecting CHF and caters to clinical practitioners’ needs by providing an accurate and fully transparent tool to support decisions concerning CHF detection.
|
['Mihaela Porumb', 'Leandro Pecchia', 'Sebastiano Massaro', 'Ernesto Iadanza']
|
2019-09-03
| null | null | null |
biomedical-signal-processing-and-control
|
['heart-rate-variability', 'congestive-heart-failure-detection', 'heartbeat-classification', 'electrocardiography-ecg']
|
['medical', 'medical', 'medical', 'methodology']
|
[ 4.77910995e-01 -2.48371288e-01 -5.02898805e-02 -2.63349444e-01
-7.20307052e-01 -2.81199664e-01 -2.06154227e-01 7.40297258e-01
-4.45534796e-01 7.37124443e-01 4.73448783e-02 -5.17539740e-01
-2.11478129e-01 -5.44273436e-01 1.06722154e-01 -5.05044878e-01
-6.84476852e-01 4.44016367e-01 -7.24205613e-01 1.69337496e-01
1.40754193e-01 6.63855612e-01 -1.03859818e+00 3.23172271e-01
8.60110402e-01 1.22100878e+00 -1.46167293e-01 1.03111148e+00
6.38638616e-01 5.25480330e-01 -7.33152449e-01 4.23064269e-02
1.14194606e-03 -9.34040964e-01 -4.92325932e-01 -1.93693191e-01
1.52189016e-01 -3.92148286e-01 3.71617777e-03 3.64334077e-01
1.27778733e+00 -4.37319249e-01 4.69681710e-01 -5.34399450e-01
-1.25326484e-01 4.55726057e-01 -2.35680073e-01 9.56010580e-01
1.29647583e-01 3.21355134e-01 8.84525776e-01 -6.59905255e-01
3.02930593e-01 3.88528734e-01 1.31526315e+00 4.20747489e-01
-1.26393080e+00 -5.41088343e-01 -5.63760936e-01 6.74031302e-02
-1.34028399e+00 -3.84014726e-01 8.45497549e-01 -5.16052008e-01
8.91812325e-01 3.53203058e-01 1.24051690e+00 9.98760343e-01
1.97775543e-01 4.12576824e-01 9.17353094e-01 -3.54509532e-01
5.58950193e-03 -4.17373091e-01 3.45049612e-02 3.75637203e-01
3.43840957e-01 1.03299402e-01 -3.52492988e-01 -4.82164055e-01
8.20208907e-01 1.17934734e-01 -5.66848695e-01 -1.01316730e-02
-1.70947003e+00 6.94870234e-01 6.42238855e-02 5.00723839e-01
-6.41031086e-01 -2.95839943e-02 9.62521434e-01 3.30734491e-01
2.76402086e-01 6.52682006e-01 -7.44102776e-01 -4.10012126e-01
-1.11936140e+00 2.85680860e-01 5.58585167e-01 2.02300817e-01
1.60191208e-01 8.68801922e-02 -4.21688318e-01 8.03294301e-01
4.64777127e-02 4.19862837e-01 4.95591760e-01 -8.96847665e-01
2.79307365e-01 7.12539315e-01 -2.08823413e-01 -1.24069166e+00
-8.55708718e-01 -9.93965268e-01 -1.36457765e+00 -1.09958239e-01
2.77085572e-01 -3.28507960e-01 -3.71125460e-01 1.40232360e+00
3.13789248e-02 1.91793442e-02 -1.05378069e-01 9.83602583e-01
7.46423900e-01 -1.72143027e-01 3.66963983e-01 -4.24731880e-01
1.48675454e+00 -1.09364472e-01 -4.59236473e-01 1.32238686e-01
8.06083143e-01 -4.22545612e-01 7.53802776e-01 4.76282686e-01
-9.01774943e-01 -4.97459799e-01 -1.09703505e+00 2.31812552e-01
-6.09202683e-02 5.76788127e-01 5.17740369e-01 7.20464706e-01
-9.34331894e-01 9.51645315e-01 -7.99906552e-01 -3.48862201e-01
8.65112364e-01 3.23480755e-01 -2.62875795e-01 1.80093899e-01
-1.40856278e+00 8.86573374e-01 3.57636213e-01 4.96583611e-01
-6.85176969e-01 -1.02064717e+00 -8.00008655e-01 8.41967091e-02
-2.50286728e-01 -1.16094625e+00 9.11853135e-01 -5.90975583e-01
-1.01520073e+00 1.11563325e+00 1.59780577e-01 -8.30781579e-01
7.59977698e-01 -2.01352194e-01 -3.66674036e-01 5.72082162e-01
-1.09441550e-02 1.37930229e-01 8.18995416e-01 -5.33244908e-01
-4.37453806e-01 -4.98635978e-01 -6.04676843e-01 3.99977975e-02
-2.26044998e-01 2.02398345e-01 3.84699494e-01 -8.60155284e-01
2.24844769e-01 -5.84306717e-01 -2.74126649e-01 -8.40804949e-02
-3.07909042e-01 8.32545161e-02 7.47440517e-01 -9.05116856e-01
1.37274313e+00 -2.15501189e+00 -1.97794795e-01 1.78738475e-01
1.02350485e+00 4.74455208e-01 4.34800088e-01 4.44441974e-01
-2.84041494e-01 4.43753660e-01 -4.11981940e-01 -1.39918894e-01
-2.06400052e-01 -1.21720396e-01 -8.32414429e-04 9.40976977e-01
3.56399387e-01 1.23562551e+00 -7.94776618e-01 -4.63898420e-01
5.73613465e-01 7.51154065e-01 -2.57495850e-01 1.96424797e-01
4.95977253e-01 9.20262039e-01 -1.23460047e-01 7.69738257e-01
4.43309397e-01 -4.01606560e-01 4.59866554e-01 -1.68258816e-01
5.96010461e-02 9.38796625e-02 -6.73364639e-01 1.45911348e+00
5.97594008e-02 6.68544710e-01 -6.84554055e-02 -1.24360132e+00
1.13309216e+00 7.50858009e-01 9.66417551e-01 -3.69493484e-01
3.33667606e-01 4.65932071e-01 2.77983189e-01 -6.86157227e-01
-2.81060219e-01 -3.64394009e-01 1.26125008e-01 1.80871978e-01
-2.42761642e-01 2.50378966e-01 3.03972475e-02 -1.66725010e-01
1.24911511e+00 -1.56959027e-01 7.54257500e-01 -2.70069987e-01
5.31153500e-01 -2.09378853e-01 6.88168287e-01 8.88853490e-01
-7.88167834e-01 1.02915406e+00 5.61662674e-01 -1.00543022e+00
-9.18612361e-01 -7.68849254e-01 -5.61227858e-01 2.78678447e-01
-3.93119514e-01 -3.59800696e-01 -3.73448610e-01 -4.81492013e-01
6.92569166e-02 -1.69661656e-01 -7.23755240e-01 -1.28474429e-01
-8.30084443e-01 -1.14873636e+00 1.08944225e+00 7.72176087e-01
5.10754645e-01 -1.34175730e+00 -1.45184922e+00 5.90437889e-01
-3.94752294e-01 -9.02050495e-01 -2.33087633e-02 1.75142631e-01
-1.27873671e+00 -1.48128402e+00 -1.18399906e+00 -6.71068788e-01
2.50956833e-01 -3.14411074e-01 1.39969599e+00 4.89827752e-01
-1.13540733e+00 1.89001765e-02 -2.70792007e-01 -5.59309542e-01
-4.45925713e-01 1.88347489e-01 -7.97027275e-02 4.21001092e-02
2.87322581e-01 -7.08452523e-01 -1.27441025e+00 -2.08632544e-01
-3.33543181e-01 -5.68868816e-02 7.68300176e-01 9.51056480e-01
4.78124112e-01 -2.71002829e-01 1.17779553e+00 -7.43668973e-01
8.07593703e-01 -3.23882550e-01 1.14659213e-01 -2.43317977e-01
-1.01664257e+00 -8.10496390e-01 4.21838403e-01 -1.64774116e-02
-2.81142622e-01 9.65907797e-03 -2.66390175e-01 -2.74662584e-01
-7.50805080e-01 6.15377605e-01 3.65689993e-01 1.45984858e-01
7.67182827e-01 2.89647490e-01 3.59364778e-01 -2.93357372e-01
-3.62739384e-01 5.07677197e-01 6.90714955e-01 -2.29588464e-01
2.60351539e-01 3.71184766e-01 1.60596341e-01 -7.69942462e-01
-5.57672143e-01 -7.71875858e-01 -8.13511789e-01 -3.20691496e-01
1.02325892e+00 -8.06723177e-01 -8.80835533e-01 5.09193778e-01
-8.75180542e-01 -6.11581393e-02 -3.28715116e-01 5.83621621e-01
-4.19568330e-01 6.51793838e-01 -7.92438209e-01 -8.59492183e-01
-1.03577709e+00 -5.76376736e-01 9.14925992e-01 -3.02464932e-01
-8.05789292e-01 -9.49682772e-01 3.93778414e-01 1.37009829e-01
6.05111480e-01 1.20389628e+00 1.15390205e+00 -6.36180758e-01
2.00394720e-01 -4.38886583e-01 -2.74684966e-01 4.72278476e-01
2.02440023e-01 3.17949504e-02 -1.02475715e+00 -3.32360536e-01
-2.28026398e-02 -5.98010197e-02 8.31000030e-01 7.51208365e-01
9.65959191e-01 1.37991056e-01 -1.17457107e-01 6.59398675e-01
1.29292321e+00 1.96654767e-01 7.03873038e-01 3.15499812e-01
5.56233346e-01 1.64440632e-01 6.29877718e-03 7.80934453e-01
1.67060345e-01 2.31157318e-01 5.23518443e-01 -7.77789056e-01
2.91142091e-02 3.28188300e-01 -2.26605460e-01 6.72825336e-01
-5.55699289e-01 3.23586762e-01 -1.29379749e+00 6.07807636e-01
-1.53037238e+00 -9.70444083e-01 -4.95514214e-01 1.96740961e+00
6.86396778e-01 6.61653792e-03 2.52084285e-01 8.02495122e-01
8.88977528e-01 -7.84890950e-02 -5.60740471e-01 -1.62436351e-01
-2.86247730e-01 5.17221451e-01 -9.31032747e-03 -3.86956513e-01
-1.43116176e+00 -5.55395298e-02 7.33517981e+00 -1.72363669e-01
-1.31837928e+00 -2.14029163e-01 9.09689546e-01 1.53981850e-01
4.13567871e-01 -3.37888569e-01 -1.59824744e-01 4.01052773e-01
8.38314056e-01 1.24548398e-01 8.68690461e-02 6.25134051e-01
6.11531556e-01 2.91039854e-01 -9.84411657e-01 1.34727359e+00
-9.59719252e-03 -1.50074792e+00 -3.68165910e-01 1.49862040e-02
2.65791297e-01 -2.14451645e-02 -6.03225566e-02 1.27675012e-01
-7.96515524e-01 -1.15324342e+00 3.33686709e-01 5.03191292e-01
1.13141596e+00 -6.42826498e-01 1.16685951e+00 -5.29640727e-02
-1.11112416e+00 -3.09606612e-01 8.71374458e-02 -3.42291147e-01
3.88476513e-02 9.83845174e-01 -9.73354042e-01 4.81213212e-01
7.92966843e-01 1.23829675e+00 -5.21258056e-01 1.22548819e+00
1.54982209e-01 9.20287490e-01 4.31480631e-02 5.11931717e-01
-3.00204217e-01 1.16874970e-01 5.85871577e-01 1.33483875e+00
3.48101109e-01 3.66821103e-02 3.06592673e-01 8.87101948e-01
2.14992967e-02 4.91925299e-01 -5.72014034e-01 -5.01043461e-02
1.32403567e-01 1.31297505e+00 -9.03373837e-01 -4.96552914e-01
-2.32421368e-01 4.76520926e-01 -9.66989025e-02 2.57808894e-01
-5.87712586e-01 -5.82852960e-01 4.11073148e-01 2.76280820e-01
-8.76567662e-02 8.88628066e-02 -9.26446557e-01 -1.07347381e+00
2.12134905e-02 -1.05892694e+00 7.21667349e-01 -1.82932824e-01
-1.12664783e+00 5.99349618e-01 -5.75176418e-01 -1.38838851e+00
-2.50164419e-01 -3.36533815e-01 -6.49279833e-01 1.16030216e+00
-1.62966919e+00 -8.28764319e-01 -5.46899796e-01 4.54235911e-01
1.98540017e-01 5.73201738e-02 1.40544939e+00 6.66640759e-01
-6.16166949e-01 3.27707916e-01 -5.10334730e-01 4.76151884e-01
7.03871667e-01 -1.44413340e+00 1.35280252e-01 4.86022055e-01
-4.35701221e-01 7.34918535e-01 4.66184288e-01 -4.56185818e-01
-7.96138585e-01 -1.00043619e+00 1.27379823e+00 -4.54513341e-01
2.12779731e-01 1.45237133e-01 -8.01951110e-01 1.39908224e-01
-1.17785305e-01 1.72447860e-01 1.01273358e+00 6.41563237e-02
1.56249963e-02 -1.08941220e-01 -1.10314763e+00 5.16421050e-02
5.58575332e-01 -5.24223685e-01 -5.98093927e-01 4.78886925e-02
-1.76630184e-01 -3.62913430e-01 -1.33106303e+00 1.09497595e+00
8.62728238e-01 -1.12533796e+00 9.61471081e-01 -5.56740820e-01
4.26001757e-01 2.53640376e-02 4.74091232e-01 -9.97567356e-01
-3.90612602e-01 -6.75457597e-01 -3.22525680e-01 6.79577887e-01
1.49236083e-01 -6.98384047e-01 7.04906762e-01 1.99259371e-01
-2.77476668e-01 -1.15798235e+00 -7.45801330e-01 -1.91365272e-01
-1.04392037e-01 -3.27938735e-01 3.30654800e-01 1.06862330e+00
4.84983176e-02 7.44952187e-02 -3.16417783e-01 -2.09353417e-01
4.48130250e-01 2.33678237e-01 3.59452516e-01 -1.69428754e+00
-7.47881457e-02 -6.53186262e-01 -6.68969393e-01 -1.14605449e-01
-3.34034085e-01 -8.99041533e-01 -1.87348574e-01 -1.50969863e+00
3.35793495e-01 -3.48109514e-01 -7.88043439e-01 4.41633165e-01
-2.69084752e-01 7.66613483e-01 -1.91354170e-01 3.77332002e-01
-9.62570608e-02 -6.17406107e-02 1.21187139e+00 3.71740647e-02
-4.17551279e-01 6.03009537e-02 -7.49137998e-01 5.82954168e-01
1.11073029e+00 -3.99190336e-01 -1.81874558e-01 2.85609365e-01
1.20915122e-01 2.20030129e-01 8.03593814e-01 -1.25728786e+00
-3.83848995e-01 3.75469089e-01 9.61602569e-01 -5.73420703e-01
-1.40791774e-01 -3.58267069e-01 1.32767186e-01 1.03870118e+00
-5.29490411e-01 2.06562653e-01 9.42155570e-02 3.85030746e-01
-3.18434745e-01 2.99324065e-01 7.93466508e-01 -2.95508862e-01
-2.27277011e-01 4.86368805e-01 -6.84790373e-01 3.82438987e-01
8.56821537e-01 -3.49982858e-01 1.16762765e-01 -1.50471792e-01
-1.24368560e+00 -2.30815098e-01 -1.44419577e-02 2.96056904e-02
8.33627105e-01 -1.08434772e+00 -1.18073928e+00 3.57878596e-01
2.82332122e-01 -2.11857393e-01 4.38267171e-01 1.53968358e+00
-1.01323485e+00 3.99431258e-01 -3.18829507e-01 -8.67768288e-01
-1.27015448e+00 2.93086380e-01 6.87445641e-01 -2.95822620e-01
-1.45120394e+00 4.21051800e-01 -2.84062266e-01 7.51231760e-02
1.84214190e-01 -4.62435305e-01 -5.31718075e-01 1.86519861e-01
5.39426923e-01 4.38502938e-01 3.81918252e-01 -4.15095240e-01
-4.69453186e-01 2.46462345e-01 2.12843552e-01 4.00010884e-01
1.15035808e+00 -4.12448570e-02 -2.46504620e-01 3.84099960e-01
9.84991670e-01 -3.66174638e-01 -7.87525833e-01 2.92987019e-01
-5.43626957e-02 -1.59709096e-01 -6.18823431e-02 -8.55986536e-01
-1.10440791e+00 1.11505723e+00 1.06259394e+00 3.83332223e-01
1.16540194e+00 -4.04750913e-01 9.33345914e-01 1.64830431e-01
8.33481625e-02 -7.64149904e-01 -1.62777036e-01 9.30903777e-02
5.50081253e-01 -1.08764875e+00 -1.18416332e-01 -1.38652042e-01
-4.05134559e-01 1.44250548e+00 1.88475773e-02 -2.91498929e-01
6.15790844e-01 -1.05367256e-02 6.31430268e-01 -4.56753820e-01
-4.06650305e-01 3.45172621e-02 6.06017038e-02 7.75365829e-01
7.72959113e-01 1.80850640e-01 -6.07115865e-01 5.90928316e-01
1.43849567e-01 3.71028066e-01 8.47532153e-02 1.03326631e+00
-2.81289190e-01 -8.61672103e-01 -7.19166175e-02 9.09211516e-01
-1.18263113e+00 -3.09134930e-01 -2.96101928e-01 5.73506236e-01
2.98376709e-01 7.35723436e-01 -3.63579273e-01 -1.02505714e-01
1.36846840e-01 4.01777983e-01 2.00253218e-01 -6.22862577e-01
-9.65271235e-01 8.44160840e-02 4.43073325e-02 -2.68641144e-01
-4.99467850e-01 -7.34468341e-01 -1.02735460e+00 1.56170532e-01
-9.77477431e-03 -3.16276252e-02 3.61390233e-01 8.49890769e-01
6.61319256e-01 8.54856312e-01 4.63552982e-01 -6.67538047e-01
-6.18123353e-01 -1.05735660e+00 -9.35213685e-01 5.90325892e-01
6.90762818e-01 -3.34743559e-01 -4.30785924e-01 3.27796370e-01]
|
[14.319178581237793, 3.279939889907837]
|
b6768ff0-2eed-466d-a8a7-14ef380e703f
|
an-iterative-bp-cnn-architecture-for-channel
|
1707.05697
| null |
http://arxiv.org/abs/1707.05697v1
|
http://arxiv.org/pdf/1707.05697v1.pdf
|
An Iterative BP-CNN Architecture for Channel Decoding
|
Inspired by recent advances in deep learning, we propose a novel iterative
BP-CNN architecture for channel decoding under correlated noise. This
architecture concatenates a trained convolutional neural network (CNN) with a
standard belief-propagation (BP) decoder. The standard BP decoder is used to
estimate the coded bits, followed by a CNN to remove the estimation errors of
the BP decoder and obtain a more accurate estimation of the channel noise.
Iterating between BP and CNN will gradually improve the decoding SNR and hence
result in better decoding performance. To train a well-behaved CNN model, we
define a new loss function which involves not only the accuracy of the noise
estimation but also the normality test for the estimation errors, i.e., to
measure how likely the estimation errors follow a Gaussian distribution. The
introduction of the normality test to the CNN training shapes the residual
noise distribution and further reduces the BER of the iterative decoding,
compared to using the standard quadratic loss function. We carry out extensive
experiments to analyze and verify the proposed framework. The iterative BP-CNN
decoder has better BER performance with lower complexity, is suitable for
parallel implementation, does not rely on any specific channel model or
encoding method, and is robust against training mismatches. All of these
features make it a good candidate for decoding modern channel codes.
|
['Cong Shen', 'Feng Wu', 'Fei Liang']
|
2017-07-18
| null | null | null | null |
['noise-estimation']
|
['medical']
|
[ 2.29168355e-01 -2.87626475e-01 1.77031755e-02 -1.03635557e-01
-6.06247842e-01 -9.12299380e-03 1.45590588e-01 1.45172656e-01
-5.80562294e-01 6.74627125e-01 -8.78776163e-02 -6.40555143e-01
2.07103223e-01 -6.89251423e-01 -9.77104485e-01 -1.02128446e+00
-7.03300759e-02 -1.66031197e-01 2.30389148e-01 9.16335508e-02
2.66568549e-02 3.84287357e-01 -9.76693034e-01 -7.83056915e-02
7.35179484e-01 1.42252827e+00 3.88667345e-01 7.92547822e-01
2.64832556e-01 8.01914871e-01 -6.35864556e-01 -5.87607622e-01
1.51104614e-01 -5.75206459e-01 -1.31454021e-01 -1.33753151e-01
-4.12159830e-01 -3.97858113e-01 -8.36213231e-01 1.39310181e+00
7.84738898e-01 -2.30077237e-01 4.39121902e-01 -7.17426658e-01
-3.24499816e-01 6.91785157e-01 -2.33295605e-01 1.19770482e-01
-1.55215021e-02 3.07299569e-02 7.87752390e-01 -7.23032594e-01
1.33001767e-02 9.72020686e-01 9.78328526e-01 3.39874983e-01
-1.09067488e+00 -9.05938327e-01 -2.79506058e-01 3.42635363e-01
-1.60632658e+00 -5.28531730e-01 5.67651451e-01 -5.17295534e-03
4.57259923e-01 5.24203293e-03 6.28841698e-01 1.00706339e+00
2.37447754e-01 6.33784711e-01 7.52625227e-01 -5.36964118e-01
3.79884988e-01 -2.99174450e-02 -1.82991594e-01 5.80653846e-01
2.99158752e-01 2.78379023e-01 -7.88944140e-02 1.61274448e-02
7.17161775e-01 -6.18809275e-02 -7.16276169e-01 -1.48474708e-01
-8.49529028e-01 6.16897762e-01 5.85293233e-01 1.66315109e-01
-3.52704138e-01 6.12816274e-01 3.56223673e-01 2.03017011e-01
2.74337959e-02 -1.10823698e-01 -2.49112874e-01 -1.85725302e-01
-9.35897112e-01 -2.48764511e-02 9.08231735e-01 7.46888578e-01
7.39253819e-01 2.18621150e-01 -2.74349838e-01 8.51205707e-01
6.78654015e-01 7.47227490e-01 2.73608983e-01 -5.22270203e-01
5.67696214e-01 1.06012665e-01 -2.00945258e-01 -1.09099066e+00
-2.28042156e-01 -1.07732749e+00 -1.32483995e+00 6.17389120e-02
3.14762831e-01 -4.15121078e-01 -9.63835716e-01 1.50466180e+00
-3.69881481e-01 4.56592560e-01 2.59858459e-01 8.40704679e-01
4.45520997e-01 8.88536811e-01 -2.72083461e-01 -1.69921756e-01
9.16635990e-01 -8.09925437e-01 -7.70093262e-01 -2.02427655e-01
5.34424007e-01 -6.90730035e-01 3.40848237e-01 4.44094092e-01
-1.08277643e+00 -4.36389625e-01 -1.49445713e+00 2.57345825e-01
1.77001417e-01 3.87092769e-01 8.15189630e-02 8.43923926e-01
-8.54034066e-01 5.75105429e-01 -8.00149143e-01 2.58357674e-01
5.15051723e-01 4.21859294e-01 -7.45329335e-02 -2.03980237e-01
-1.35803199e+00 8.05480838e-01 6.73658073e-01 5.54427326e-01
-9.93861735e-01 -1.01083107e-01 -8.12797844e-01 5.60516357e-01
9.36268941e-02 -3.63938630e-01 1.20330417e+00 -1.15294635e+00
-1.76897085e+00 6.87196031e-02 -2.26730481e-01 -7.72676110e-01
4.31162357e-01 -3.05254944e-04 -5.07003725e-01 -4.80952263e-02
-4.42011535e-01 1.21069565e-01 1.02067912e+00 -9.85017896e-01
-5.20748556e-01 1.49276733e-01 -2.25356132e-01 -1.50866257e-02
-2.42144138e-01 -2.44975895e-01 -9.71312463e-01 -6.55744255e-01
2.40826055e-01 -8.59709978e-01 -3.21867198e-01 -1.71172798e-01
-4.09416050e-01 5.00895143e-01 5.34990489e-01 -8.71266961e-01
1.46587455e+00 -2.50777721e+00 -1.07086964e-01 7.61613667e-01
1.20498456e-01 5.60892344e-01 3.63722257e-02 2.31074080e-01
9.12341699e-02 -1.12896964e-01 -4.06084955e-01 -1.89305067e-01
-2.77214110e-01 1.40538529e-01 1.72550976e-01 5.79274118e-01
5.68777807e-02 6.59712851e-01 -7.11938322e-01 2.15343274e-02
-8.27491209e-02 5.71442246e-01 -6.38671994e-01 2.82530785e-01
-8.31404328e-03 4.67885911e-01 -6.89872652e-02 2.42734104e-01
9.13289011e-01 -2.68363208e-01 1.07409693e-01 -1.16439112e-01
9.93375480e-02 3.20065081e-01 -1.24927938e+00 1.07287109e+00
-6.02359235e-01 9.69878376e-01 2.84046352e-01 -1.25063014e+00
1.11893702e+00 3.94554526e-01 -1.23404101e-01 -9.49150383e-01
5.52293241e-01 5.31705856e-01 3.35738301e-01 -2.72562742e-01
5.07658087e-02 -4.95189391e-02 1.14700839e-01 -6.73545301e-02
8.29197466e-02 1.94116026e-01 -1.57803074e-02 -9.58516374e-02
1.01206589e+00 -5.62911570e-01 2.15810329e-01 3.29427309e-02
7.28621185e-01 -6.72703445e-01 7.40660071e-01 8.48957300e-01
-4.36264585e-04 4.39061880e-01 5.06549597e-01 -1.30739212e-01
-1.18089664e+00 -9.49801028e-01 -1.88639328e-01 4.78969157e-01
3.64919364e-01 -1.82614893e-01 -5.78472495e-01 -3.16194475e-01
-2.06825733e-01 4.18171763e-01 -7.50572681e-02 -3.67290914e-01
-3.89182657e-01 -8.95848036e-01 8.14518631e-01 4.45209533e-01
9.25535679e-01 -7.07312882e-01 -3.57428156e-02 4.13111091e-01
-2.04591215e-01 -1.16387558e+00 -1.20510139e-01 6.92518234e-01
-6.31796598e-01 -8.37545276e-01 -9.80696619e-01 -8.89955401e-01
6.47443831e-01 -1.27464771e-01 6.26307607e-01 2.83309102e-01
2.04105720e-01 -1.89014345e-01 -5.34681857e-01 -3.54483426e-01
-7.80238032e-01 -1.57076046e-01 -2.42061928e-01 1.79925844e-01
4.34288643e-02 -3.54212463e-01 -6.44978166e-01 2.26829723e-01
-6.29034817e-01 -9.11531150e-02 9.85884011e-01 1.10830295e+00
3.91512483e-01 5.85246801e-01 3.40340167e-01 -4.89408910e-01
6.30494595e-01 -4.38652843e-01 -8.25512171e-01 -3.88373435e-02
-3.76514554e-01 1.72483668e-01 8.88386726e-01 -2.86774635e-01
-7.40182340e-01 8.75628516e-02 -8.49656880e-01 -9.91848707e-02
2.62451112e-01 7.81291723e-01 -2.00197637e-01 -3.75003785e-01
5.06444812e-01 4.58033502e-01 3.02430931e-02 -2.46364072e-01
-1.35637820e-02 9.43613410e-01 6.21975660e-01 -2.86438819e-02
8.76108289e-01 1.57947868e-01 1.20000122e-02 -8.25302184e-01
-4.02110994e-01 -2.65942246e-01 -1.23502210e-01 -1.30455285e-01
6.37226582e-01 -1.03445983e+00 -8.00554514e-01 7.70357370e-01
-1.29504442e+00 -3.54690194e-01 3.89930367e-01 7.43753970e-01
-2.14690223e-01 5.56984067e-01 -7.46202052e-01 -8.26647162e-01
-3.27588588e-01 -1.36672461e+00 3.78132403e-01 3.11402023e-01
3.36013854e-01 -9.75184202e-01 -2.95269042e-01 -3.24922293e-01
6.26984239e-01 -1.99312687e-01 8.26407433e-01 -5.32608092e-01
-5.91991603e-01 -5.73527634e-01 -5.14861822e-01 1.07187843e+00
-2.68019468e-01 -2.58081764e-01 -8.78716290e-01 -4.52460974e-01
8.94498304e-02 7.41945878e-02 1.11515427e+00 3.85871142e-01
1.42053735e+00 -2.68464416e-01 -1.44148260e-01 8.65416110e-01
1.63193488e+00 4.70070601e-01 1.12380779e+00 3.38877767e-01
4.50105697e-01 -2.94475704e-01 3.92174162e-02 6.30656838e-01
6.97137713e-02 5.34663379e-01 6.33430839e-01 -2.21369952e-01
3.67663614e-02 -7.15373755e-02 3.77448350e-01 1.03504145e+00
1.22600734e-01 -5.32797098e-01 -8.57392728e-01 1.56405732e-01
-1.61379302e+00 -7.56224513e-01 -2.73500293e-01 2.24160504e+00
8.94651413e-01 4.91909593e-01 -3.62582535e-01 5.16039610e-01
7.11078644e-01 -8.50888863e-02 -3.15782458e-01 -5.19089282e-01
-3.82859856e-02 4.40987825e-01 1.00309753e+00 3.54347974e-01
-1.13483679e+00 5.56158245e-01 6.13788128e+00 9.68751073e-01
-1.44842291e+00 -9.84543934e-02 7.31928587e-01 3.80090654e-01
7.60807618e-02 -1.26653135e-01 -6.95077062e-01 7.73492098e-01
1.01968837e+00 1.71017587e-01 5.95389843e-01 6.84559822e-01
2.84838736e-01 -3.06893438e-02 -8.65559101e-01 1.23447466e+00
-5.78183345e-02 -1.23646927e+00 -2.19772369e-01 -2.67393030e-02
5.69496095e-01 4.75040115e-02 3.59454416e-02 2.52761275e-01
3.74255441e-02 -1.07831764e+00 8.82483542e-01 6.51879489e-01
5.41576862e-01 -8.62475753e-01 1.39755654e+00 4.88181531e-01
-1.00995123e+00 -4.48012829e-01 -5.60116589e-01 -1.64887071e-01
4.17461619e-02 9.94614065e-01 -7.71316886e-01 3.82046640e-01
6.31993234e-01 4.74397361e-01 -1.12941705e-01 1.69033873e+00
-4.31822360e-01 8.16401362e-01 -2.03292787e-01 -2.38300368e-01
3.16605985e-01 6.12548850e-02 4.44029301e-01 1.48835993e+00
6.25147521e-01 -1.22668415e-01 5.51780276e-02 5.20527065e-01
-4.16558176e-01 1.41040739e-02 2.64768153e-02 9.06699076e-02
6.16313338e-01 7.43074000e-01 -4.60603207e-01 -2.32362211e-01
-5.69941580e-01 9.16100562e-01 2.13202715e-01 6.06195271e-01
-8.00085068e-01 -6.63732350e-01 5.11742473e-01 -2.01350704e-01
8.44435990e-01 -3.05437297e-01 -4.05800700e-01 -8.36856663e-01
1.09289654e-01 -8.24076474e-01 -2.09909081e-01 -4.05016184e-01
-9.35451210e-01 5.34651220e-01 -6.13879502e-01 -1.22247684e+00
3.32306586e-02 -5.10415077e-01 -5.81804812e-01 1.10615861e+00
-1.95720732e+00 -2.51379997e-01 -2.78218657e-01 5.04111528e-01
4.10608128e-02 -5.00114821e-02 6.75792933e-01 6.50879800e-01
-7.90840924e-01 7.70938635e-01 6.28367603e-01 5.06591856e-01
3.73873860e-01 -8.81586075e-01 2.98767090e-01 1.06213939e+00
-2.90258557e-01 4.81753320e-01 6.89610958e-01 -5.11433363e-01
-1.16476405e+00 -1.11291671e+00 5.39676428e-01 5.12043655e-01
5.53049207e-01 -3.46493632e-01 -8.53434086e-01 2.30102226e-01
-1.47604980e-02 1.10800616e-01 4.18723196e-01 -2.20458716e-01
-2.35973164e-01 -4.16962773e-01 -7.77514279e-01 4.28666234e-01
4.07505214e-01 -3.70611876e-01 -3.15124579e-02 -1.57941524e-02
3.16469759e-01 -6.67007744e-01 -5.03900647e-01 4.73774433e-01
5.48402369e-01 -9.70494390e-01 7.53430665e-01 2.90981114e-01
4.08015132e-01 -2.87399054e-01 -1.24361344e-01 -1.34566092e+00
-4.38519806e-01 -4.73573238e-01 -1.31037921e-01 6.52914405e-01
8.10018897e-01 -5.05868495e-01 6.35881305e-01 -4.00947127e-03
-2.74594009e-01 -8.81952524e-01 -9.85107183e-01 -8.36089492e-01
-2.59588182e-01 -7.60120511e-01 3.44555050e-01 3.36233050e-01
-1.72500744e-01 -4.01426777e-02 -6.63442433e-01 6.26792550e-01
5.60713530e-01 -5.65937281e-01 6.52338207e-01 -1.00599515e+00
-6.45534873e-01 -4.94430959e-01 -7.30336964e-01 -1.68808794e+00
-1.40362725e-01 -8.03849101e-01 5.87556720e-01 -1.40579832e+00
5.67531735e-02 -4.80974466e-01 -4.08251911e-01 2.60857552e-01
-1.70191810e-01 5.04163682e-01 3.62131037e-02 1.01952836e-01
-4.58082974e-01 6.63130581e-01 9.34191048e-01 -6.70568123e-02
-8.49138349e-02 4.64911610e-01 -5.46203017e-01 6.66300833e-01
7.85312593e-01 -5.71097314e-01 6.99061379e-02 -3.26087266e-01
4.32598144e-01 1.11990102e-01 3.59463304e-01 -1.48630059e+00
3.83535475e-01 5.95448554e-01 5.49707234e-01 -3.23583812e-01
1.19854681e-01 -1.18524694e+00 -5.40728867e-02 7.53243446e-01
-1.55899554e-01 -4.53118742e-01 1.44520551e-01 7.97422290e-01
-3.74390781e-01 -5.15136003e-01 1.07624698e+00 9.21997353e-02
-5.52354693e-01 1.83982253e-01 -6.83773100e-01 -4.05292749e-01
7.41596997e-01 -2.70014971e-01 9.79108661e-02 -7.37792909e-01
-7.51867115e-01 1.50219604e-01 1.06536895e-01 -7.81712309e-02
7.36137271e-01 -1.15016699e+00 -7.20154643e-01 5.33180475e-01
-8.31607729e-02 -2.20189497e-01 -3.01851258e-02 9.92397964e-01
-8.68760645e-01 3.24792266e-01 1.95006132e-01 -6.89944983e-01
-1.10569465e+00 -1.39791341e-02 4.64198530e-01 -2.21764594e-01
-5.61969161e-01 9.07618999e-01 -2.24150315e-01 5.44469804e-02
7.32771814e-01 -3.30875129e-01 -4.82490547e-02 -6.04527771e-01
6.41166091e-01 2.06251845e-01 2.88763076e-01 -4.31612462e-01
-1.42988130e-01 2.18096763e-01 7.23636225e-02 5.50162718e-02
1.08989131e+00 -1.35313392e-01 -6.02703765e-02 -1.16179907e-03
1.41092777e+00 -1.11184962e-01 -1.27498055e+00 -3.82498026e-01
-1.45969927e-01 -2.77686089e-01 4.50037032e-01 -7.70799577e-01
-1.23460448e+00 8.99767101e-01 6.47689641e-01 2.40546122e-01
1.34786487e+00 -4.56525683e-01 1.04031837e+00 3.64511907e-01
1.35790199e-01 -9.33066726e-01 3.47066447e-02 9.87973809e-01
5.13954103e-01 -9.35367465e-01 -1.48817584e-01 -1.87273234e-01
-3.08982819e-01 1.56205273e+00 3.81904878e-02 -6.14904091e-02
8.66396189e-01 4.64669615e-01 -1.18768716e-03 3.34761560e-01
-3.39388132e-01 -2.33894393e-01 1.59971729e-01 4.54831809e-01
1.73554853e-01 -6.85410053e-02 -3.16475481e-01 5.42758405e-01
-1.11568861e-01 7.71231996e-03 4.90194440e-01 5.50270915e-01
-6.94994688e-01 -8.86730850e-01 -3.75676900e-01 4.01392490e-01
-5.05643964e-01 -3.18648607e-01 1.56183824e-01 3.26925457e-01
1.01219900e-01 9.56964433e-01 -4.04019142e-03 -5.61750054e-01
1.65504679e-01 -1.86448932e-01 1.09511755e-01 -3.16765815e-01
-2.71634996e-01 2.78428584e-01 -1.25697488e-02 -2.78851360e-01
-1.07707083e-01 -1.91077948e-01 -1.26837659e+00 -4.69773620e-01
-8.25775445e-01 1.82764933e-01 9.99403775e-01 1.11459243e+00
6.46418706e-02 7.07754433e-01 8.89950037e-01 -6.85506761e-01
-5.30078053e-01 -7.18035698e-01 -4.77054060e-01 -4.72459085e-02
6.14578247e-01 -2.96857893e-01 -4.12775487e-01 -2.55777359e-01]
|
[6.4093499183654785, 1.488447904586792]
|
85753ce3-55bc-489a-be5f-843d5b0cc95b
|
semi-centralised-multi-agent-reinforcement
|
2209.01054
| null |
https://arxiv.org/abs/2209.01054v2
|
https://arxiv.org/pdf/2209.01054v2.pdf
|
Taming Multi-Agent Reinforcement Learning with Estimator Variance Reduction
|
Centralised training with decentralised execution (CT-DE) serves as the foundation of many leading multi-agent reinforcement learning (MARL) algorithms. Despite its popularity, it suffers from a critical drawback due to its reliance on learning from a single sample of the joint-action at a given state. As agents explore and update their policies during training, these single samples may poorly represent the actual joint-policy of the system of agents leading to high variance gradient estimates that hinder learning. To address this problem, we propose an enhancement tool that accommodates any actor-critic MARL method. Our framework, Performance Enhancing Reinforcement Learning Apparatus (PERLA), introduces a sampling technique of the agents' joint-policy into the critics while the agents train. This leads to TD updates that closely approximate the true expected value under the current joint-policy rather than estimates from a single sample of the joint-action at a given state. This produces low variance and precise estimates of expected returns, minimising the variance in the critic estimators which typically hinders learning. Moreover, as we demonstrate, by eliminating much of the critic variance from the single sampling of the joint policy, PERLA enables CT-DE methods to scale more efficiently with the number of agents. Theoretically, we prove that PERLA reduces variance in value estimates similar to that of decentralised training while maintaining the benefits of centralised training. Empirically, we demonstrate PERLA's superior performance and ability to reduce estimator variance in a range of benchmarks including Multi-agent Mujoco, and StarCraft II Multi-agent Challenge.
|
['David Mguni', 'Jun Wang', 'Kun Shao', 'Matthew Taylor', 'Jianhong Wang', 'Zipeng Dai', 'Tianpei Yang', 'Juliusz Ziomek', 'Taher Jafferjee']
|
2022-09-02
| null | null | null | null |
['starcraft-ii', 'starcraft']
|
['playing-games', 'playing-games']
|
[-4.61684614e-01 -3.49919647e-02 -4.70126271e-01 1.94782674e-01
-9.81564522e-01 -5.47585189e-01 7.74039447e-01 2.31885314e-01
-9.20668960e-01 1.25496352e+00 -1.47131845e-01 -4.74745542e-01
-1.28342137e-01 -5.75402379e-01 -7.88509548e-01 -8.47564697e-01
-5.40534198e-01 8.10386598e-01 2.57494152e-01 -2.09496394e-01
9.05839205e-02 3.64199847e-01 -1.28172112e+00 -3.03913355e-01
7.47815132e-01 7.98967600e-01 2.59927157e-02 1.08204186e+00
2.82622367e-01 1.05574989e+00 -1.07557130e+00 1.08830087e-01
3.76649588e-01 -4.79187012e-01 -4.00187463e-01 1.64572015e-01
7.94557631e-02 -9.24419820e-01 -9.85582024e-02 9.74655867e-01
4.07602698e-01 3.38968754e-01 4.25621539e-01 -1.48683798e+00
2.06211120e-01 5.89602768e-01 -7.38332629e-01 1.32996529e-01
3.54347080e-02 7.31785715e-01 7.34632134e-01 -7.88047463e-02
2.62394965e-01 1.37711465e+00 6.88168406e-01 5.85621238e-01
-1.34450710e+00 -4.43578303e-01 5.05993903e-01 -2.40929201e-01
-8.23957086e-01 -4.27126706e-01 2.23862410e-01 -1.27734512e-01
1.15217400e+00 -1.76244631e-01 7.42047727e-01 7.62513757e-01
5.59213519e-01 8.09302509e-01 1.22683692e+00 -2.40886167e-01
8.13771129e-01 1.18366517e-01 -6.29297137e-01 7.31040895e-01
3.32817644e-01 4.47362274e-01 -1.37617633e-01 -5.59053838e-01
1.02753818e+00 -6.91368878e-02 1.08738199e-01 -5.75818837e-01
-1.02079451e+00 1.01393878e+00 1.41765937e-01 -1.61343127e-01
-8.04757893e-01 8.33305776e-01 7.97013462e-01 6.69390261e-01
2.58134812e-01 4.11607891e-01 -7.44776130e-01 -6.00834310e-01
-6.54676020e-01 6.49975002e-01 1.07196164e+00 6.79624319e-01
6.73530281e-01 6.61510885e-01 1.78136323e-02 4.64322060e-01
3.15757692e-01 6.39023721e-01 5.53179562e-01 -1.54659855e+00
4.29038823e-01 2.95564950e-01 5.20489514e-01 -2.55709976e-01
-1.79961070e-01 -5.07691562e-01 -3.48825872e-01 9.38718796e-01
5.82544625e-01 -8.85765314e-01 -5.25147498e-01 1.76947546e+00
6.09416008e-01 6.23903722e-02 4.30127174e-01 6.28911734e-01
-4.70471025e-01 6.19662702e-01 2.74041351e-02 -6.03538454e-01
1.03529549e+00 -9.94300425e-01 -3.96284074e-01 -3.83174509e-01
6.68869555e-01 -3.57681632e-01 7.14877605e-01 4.61133122e-01
-1.27150905e+00 -9.59465951e-02 -1.04372430e+00 8.99863958e-01
-8.13896360e-04 -3.30027729e-01 6.53477132e-01 4.76199746e-01
-1.06321120e+00 6.99958980e-01 -1.25189114e+00 6.72068894e-02
3.00928205e-01 5.43667018e-01 1.21255487e-01 3.49150747e-01
-7.94744074e-01 1.20298064e+00 3.62696141e-01 -4.01741385e-01
-1.60762298e+00 -6.89691186e-01 -5.60522079e-01 9.03059244e-02
8.38902056e-01 -5.19153416e-01 2.13312507e+00 -1.14746499e+00
-2.15690589e+00 -2.09376007e-01 3.28141332e-01 -8.70820522e-01
8.08008611e-01 -1.57265797e-01 1.09846637e-01 1.57053158e-01
1.26697943e-01 3.65735412e-01 1.10182571e+00 -1.19712567e+00
-9.85256732e-01 -9.48953927e-02 2.90437907e-01 5.44301987e-01
-2.10588425e-02 -4.49328452e-01 1.69754192e-01 -2.04396963e-01
-7.69445300e-01 -9.75793302e-01 -6.97030008e-01 -3.13161194e-01
3.53480011e-01 -3.34249407e-01 7.13558018e-01 -1.86870277e-01
8.79521906e-01 -1.97813618e+00 -8.72817561e-02 2.52389401e-01
2.12133795e-01 3.13333154e-01 -2.18886465e-01 7.97368467e-01
3.59977752e-01 -2.52160162e-01 8.98985863e-02 -2.54048645e-01
2.71100014e-01 4.90353078e-01 -2.36942813e-01 6.79808140e-01
6.26727790e-02 6.24289453e-01 -1.27253997e+00 -3.78031194e-01
2.47543260e-01 1.23406470e-01 -7.49796748e-01 2.46494785e-01
-7.95678437e-01 1.92189172e-01 -6.66322052e-01 -1.60393957e-02
2.28201732e-01 -5.36321774e-02 5.77038705e-01 5.26323080e-01
-2.76780337e-01 2.29612544e-01 -1.30005169e+00 1.26000202e+00
-6.13387883e-01 2.70770431e-01 3.91333282e-01 -9.29755390e-01
6.00379765e-01 5.24392664e-01 7.59323955e-01 -7.24178672e-01
-1.34603726e-02 2.46914595e-01 1.45444036e-01 -1.53704002e-01
2.42662489e-01 -1.80251747e-01 6.85541844e-03 8.77342820e-01
-3.87151651e-02 -3.80343765e-01 4.95272279e-01 2.65803397e-01
1.23290706e+00 2.84973323e-01 4.51550603e-01 -1.74257785e-01
3.15553576e-01 1.26255840e-01 5.99110484e-01 1.01564693e+00
-3.76825988e-01 -5.26649296e-01 9.21445549e-01 -4.01294202e-01
-1.25451052e+00 -9.67113197e-01 3.59372884e-01 1.21467841e+00
-2.37351790e-01 -4.00579929e-01 -6.95575595e-01 -9.12297368e-01
4.12043661e-01 7.08575785e-01 -4.80474830e-01 -8.07212852e-03
-6.13736391e-01 -5.97656250e-01 5.04770875e-01 2.87167549e-01
5.02011240e-01 -9.68073010e-01 -1.33735645e+00 6.85709298e-01
4.31297332e-01 -6.81153119e-01 -6.20393038e-01 3.50241572e-01
-8.80170882e-01 -1.13438261e+00 -4.77574825e-01 -2.49054149e-01
5.59607148e-01 -6.75303116e-02 1.06639564e+00 -1.21363766e-01
5.02817072e-02 7.94899225e-01 1.10935338e-03 -4.90883052e-01
-7.93546975e-01 -5.07201441e-02 2.57940084e-01 -4.38920498e-01
-9.70864147e-02 -5.02597630e-01 -5.97385228e-01 7.08125830e-02
-7.91341662e-01 -2.87035882e-01 6.73282087e-01 1.08874917e+00
1.59047216e-01 1.91183075e-01 9.08007145e-01 -6.06851220e-01
1.14964461e+00 -4.20604318e-01 -1.21655881e+00 1.01842560e-01
-9.25832629e-01 4.29029018e-01 1.01735485e+00 -7.86475539e-01
-9.85140800e-01 -1.69624910e-01 2.46985063e-01 -3.51938516e-01
-6.26599742e-03 2.64906466e-01 6.52450860e-01 1.47199988e-01
6.83677554e-01 2.69880891e-01 8.06938827e-01 -9.18504819e-02
2.70762146e-01 2.99252808e-01 1.36402071e-01 -1.00463927e+00
5.23783863e-01 1.02878585e-01 8.26627687e-02 -5.20892680e-01
-3.73212665e-01 -8.13374147e-02 1.08885445e-01 -2.44187772e-01
9.60495546e-02 -9.97647345e-01 -1.45231378e+00 3.00715059e-01
-7.54311264e-01 -9.11079586e-01 -7.10635006e-01 5.47468483e-01
-9.85660791e-01 1.88984513e-01 -7.02135444e-01 -1.19389248e+00
-3.15762311e-01 -1.21403134e+00 5.79307795e-01 3.60395879e-01
1.38732865e-01 -1.24245226e+00 6.46271527e-01 -4.09488618e-01
6.49193048e-01 1.98619112e-01 7.25094378e-01 -5.37607372e-01
-3.17194611e-01 -5.86517230e-02 3.68582040e-01 3.94310415e-01
-3.18091065e-02 -1.01520345e-01 -4.27026510e-01 -1.03111768e+00
1.63101573e-02 -6.74247801e-01 2.62221843e-01 4.19656277e-01
2.85656333e-01 -7.89412081e-01 -4.41003181e-02 1.24736745e-02
1.65665042e+00 3.60733509e-01 1.63683463e-02 7.87073493e-01
9.42708030e-02 5.91034815e-02 7.06584632e-01 1.16309631e+00
2.83652157e-01 3.78304511e-01 5.95972478e-01 2.81698585e-01
3.67440820e-01 -1.38414547e-01 1.03298676e+00 4.64242458e-01
8.34455267e-02 2.41449609e-01 -6.60611868e-01 4.64957923e-01
-2.16823745e+00 -9.49163496e-01 5.04068971e-01 2.42603493e+00
1.14421284e+00 1.70488015e-01 7.89454699e-01 -3.33828896e-01
3.27844560e-01 1.21425427e-02 -1.22505200e+00 -7.17151821e-01
5.49346626e-01 1.16637737e-01 9.13523197e-01 7.23134696e-01
-6.57050371e-01 8.54809523e-01 6.53520107e+00 6.60724103e-01
-8.78177106e-01 1.32003665e-01 1.98120266e-01 -2.66751647e-01
1.70424253e-01 6.40976205e-02 -7.73866594e-01 5.02347469e-01
1.32401991e+00 -4.62095857e-01 1.04117644e+00 1.16047323e+00
4.99741673e-01 -5.82211673e-01 -1.05129850e+00 5.34837306e-01
-4.84518588e-01 -9.87291992e-01 -4.56511676e-01 3.91463518e-01
9.48963225e-01 4.04875010e-01 1.15174148e-02 7.40862548e-01
1.23351490e+00 -6.36437833e-01 5.78475595e-01 5.04682474e-02
4.13386017e-01 -1.28443909e+00 6.36640310e-01 7.63097525e-01
-9.55763400e-01 -4.47350144e-01 -3.52642298e-01 -3.77032489e-01
-1.54834688e-01 1.30841970e-01 -1.24636412e+00 1.95738837e-01
1.94977924e-01 2.48912200e-01 -1.10509865e-01 7.55934536e-01
-1.51800036e-01 5.79651773e-01 -6.05029464e-01 -3.11155409e-01
8.23049247e-01 -2.14664176e-01 4.77940738e-01 8.44759405e-01
-2.62054745e-02 -3.91690224e-01 6.75385654e-01 4.83370900e-01
2.42397904e-01 -2.03753114e-01 -4.70360309e-01 -6.07094206e-02
5.61612844e-01 1.05541730e+00 -4.72164124e-01 -5.62246323e-01
-2.54191607e-01 4.64166582e-01 5.68016648e-01 4.37362373e-01
-8.08472335e-01 -1.56415582e-01 9.07425463e-01 -3.26169312e-01
4.13607419e-01 -5.32170892e-01 3.57712716e-01 -8.80169511e-01
-2.38601401e-01 -1.46784067e+00 2.95315176e-01 -1.46176284e-02
-1.10845399e+00 2.60171443e-01 1.22450419e-01 -1.05071115e+00
-1.17603803e+00 -2.49914587e-01 -5.83183527e-01 6.06036842e-01
-1.43507636e+00 -4.55152631e-01 4.02228951e-01 5.40149271e-01
6.81564450e-01 -4.02482718e-01 7.49033093e-01 -1.48943096e-01
-6.27971590e-01 3.79750252e-01 6.93286121e-01 -1.95133865e-01
5.16437471e-01 -1.57462871e+00 2.27838367e-01 4.59162474e-01
-4.64372516e-01 2.88186491e-01 9.17879045e-01 -6.01478100e-01
-1.56912315e+00 -7.30745494e-01 -2.18897805e-01 -1.06881678e-01
1.03691936e+00 2.38402888e-01 -4.63783294e-01 7.52333701e-01
4.34160024e-01 -1.90452203e-01 5.03669716e-02 1.53277935e-02
9.84075945e-03 -2.18189731e-01 -1.07925522e+00 7.29972482e-01
9.84979123e-02 -2.26047501e-01 -2.93690056e-01 2.17065036e-01
3.46655995e-01 -4.08868730e-01 -9.88011301e-01 -1.44179851e-01
5.41190743e-01 -9.46801960e-01 5.49675107e-01 -6.22826099e-01
1.26063135e-02 -2.16417253e-01 1.73166916e-01 -1.93164599e+00
-2.64268778e-02 -1.01402581e+00 -5.67552149e-01 8.59252751e-01
3.69283706e-02 -1.08735156e+00 8.23672235e-01 4.38007444e-01
1.30669251e-01 -8.40767086e-01 -1.04834008e+00 -1.04391849e+00
2.62537897e-01 3.71879078e-02 3.28084767e-01 5.73857307e-01
1.36049986e-01 2.78459221e-01 -3.55327278e-01 1.37417480e-01
1.07586634e+00 -2.50457704e-01 1.04206884e+00 -5.19808650e-01
-9.35319662e-01 -5.39125264e-01 4.99063618e-02 -1.17777061e+00
1.98866904e-01 -2.70731688e-01 2.94320166e-01 -1.25843418e+00
5.46557307e-02 -4.73166764e-01 -1.01311311e-01 4.43207085e-01
-3.57314423e-02 -5.11476457e-01 4.32583213e-01 1.50595903e-01
-8.93098772e-01 6.00753188e-01 1.26773345e+00 8.67038518e-02
-4.71888930e-01 1.60448313e-01 -2.02025399e-01 6.25058413e-01
1.07498562e+00 -6.48438871e-01 -6.14933252e-01 -2.07508296e-01
1.47225589e-01 4.77063984e-01 2.13946685e-01 -9.67045844e-01
2.39585608e-01 -5.21457314e-01 1.63013637e-01 -1.57000095e-01
1.49198294e-01 -7.15867400e-01 -6.23405054e-02 1.01274025e+00
-4.17655766e-01 5.74948907e-01 2.42588222e-01 6.55412078e-01
4.31596227e-02 -2.92410672e-01 9.33268189e-01 -3.48117113e-01
-3.25898528e-01 1.46735474e-01 -9.20979619e-01 2.73537487e-01
1.22066534e+00 1.42531201e-01 -4.30750906e-01 -5.52801013e-01
-3.33673179e-01 8.96836936e-01 5.20596683e-01 -1.22709282e-01
4.09340262e-01 -1.05258381e+00 -6.09411895e-01 1.44454390e-02
-5.08214056e-01 -1.97091568e-02 -5.72305024e-02 8.47577870e-01
-3.58333737e-01 2.01450035e-01 -3.38640004e-01 -3.44258100e-01
-8.76314461e-01 4.73140299e-01 7.90396214e-01 -6.90723240e-01
-5.82037091e-01 1.50798351e-01 -2.25233018e-01 -2.80709624e-01
3.15925360e-01 -8.14784840e-02 2.70810306e-01 -1.79783508e-01
4.48184431e-01 5.99806905e-01 -3.80466431e-01 9.70593095e-02
-3.75168473e-02 1.87296439e-02 -3.54786426e-01 -6.57910466e-01
1.45005894e+00 9.07621533e-02 1.23604253e-01 4.88012940e-01
7.87910461e-01 -2.89281696e-01 -2.19940329e+00 -2.19279647e-01
5.59938364e-02 -2.18652576e-01 1.49580881e-01 -7.39248991e-01
-7.36192167e-01 3.99751574e-01 4.44879085e-01 5.05789816e-01
6.11519694e-01 -3.99817884e-01 4.95227426e-01 6.45998597e-01
5.17567813e-01 -1.44735444e+00 1.95953652e-01 6.52039409e-01
4.39023852e-01 -9.68582749e-01 3.42740417e-01 6.30513072e-01
-9.56042349e-01 1.04676235e+00 7.15862155e-01 -5.75955868e-01
1.60474420e-01 4.95227456e-01 3.65507565e-02 1.21310420e-01
-1.43087864e+00 -1.28762990e-01 -5.82570910e-01 6.11755848e-01
-1.02917790e-01 1.63074791e-01 2.09659841e-02 -2.54423946e-01
1.84100240e-01 -1.07734099e-01 6.76629484e-01 1.27591300e+00
-7.33258247e-01 -1.14999771e+00 -4.41382319e-01 3.95171970e-01
-4.81600702e-01 2.84994185e-01 2.26986974e-01 1.02352524e+00
-3.44709814e-01 7.98200130e-01 1.12643212e-01 3.87065768e-01
1.34345472e-01 -6.04064763e-02 6.13298059e-01 -3.61309916e-01
-7.37404704e-01 3.68368119e-01 7.58939907e-02 -6.16182923e-01
-1.32354960e-01 -6.72130704e-01 -1.33217096e+00 -6.41140997e-01
-1.39021456e-01 6.10751152e-01 7.97380149e-01 9.05068219e-01
2.06164628e-01 5.76473594e-01 9.42598581e-01 -1.13353372e+00
-1.72480476e+00 -7.57477105e-01 -5.49872339e-01 -3.32024992e-02
7.10874915e-01 -5.40774107e-01 -4.31041479e-01 -4.29338664e-01]
|
[3.9349992275238037, 2.1275386810302734]
|
e9ac8f29-6c44-45f0-9884-dd7a1908bb3d
|
vpair-aerial-visual-place-recognition-and
|
2205.11567
| null |
https://arxiv.org/abs/2205.11567v1
|
https://arxiv.org/pdf/2205.11567v1.pdf
|
VPAIR -- Aerial Visual Place Recognition and Localization in Large-scale Outdoor Environments
|
Visual Place Recognition and Visual Localization are essential components in navigation and mapping for autonomous vehicles especially in GNSS-denied navigation scenarios. Recent work has focused on ground or close to ground applications such as self-driving cars or indoor-scenarios and low-altitude drone flights. However, applications such as Urban Air Mobility require operations in large-scale outdoor environments at medium to high altitudes. We present a new dataset named VPAIR. The dataset was recorded on board a light aircraft flying at an altitude of more than 300 meters above ground capturing images with a downwardfacing camera. Each image is paired with a high resolution reference render including dense depth information and 6-DoF reference poses. The dataset covers a more than one hundred kilometers long trajectory over various types of challenging landscapes, e.g. urban, farmland and forests. Experiments on this dataset illustrate the challenges introduced by the change in perspective to a bird's eye view such as in-plane rotations.
|
['Daniel Cremers', 'Fahmi Rouatbi', 'Michael Schleiss']
|
2022-05-23
| null | null | null | null |
['visual-place-recognition']
|
['computer-vision']
|
[ 1.43411338e-01 -2.33239040e-01 3.20391804e-02 -6.42786682e-01
-9.71229225e-02 -9.99566615e-01 5.92574239e-01 -3.13351840e-01
-6.04780853e-01 7.62037158e-01 -4.68732238e-01 -4.66321826e-01
-4.37392034e-02 -1.06670702e+00 -7.02115297e-01 -4.45906669e-01
-4.15844202e-01 4.97704148e-01 3.40781391e-01 -8.28975439e-01
3.00387502e-01 1.04263568e+00 -1.82181180e+00 -5.72500288e-01
6.86959267e-01 5.39267957e-01 4.12724704e-01 8.58217359e-01
3.71602654e-01 1.03990473e-02 -9.36265513e-02 -1.70145288e-01
8.64832640e-01 6.79142103e-02 -1.19764142e-01 2.48695165e-01
8.31101239e-01 -3.57869387e-01 -4.76505488e-01 1.13334823e+00
2.25103930e-01 4.58016723e-01 2.44617417e-01 -1.49391413e+00
-5.00492156e-02 -5.53250551e-01 -4.81079221e-01 2.23917410e-01
4.60197061e-01 2.62129158e-01 5.27511537e-01 -7.63241053e-01
7.76369989e-01 7.40097046e-01 6.53272986e-01 4.65540774e-02
-8.46757054e-01 -4.45538044e-01 2.59392083e-01 2.28181839e-01
-2.02219057e+00 -4.96241897e-01 2.53330112e-01 -4.78717148e-01
8.72177124e-01 2.45585307e-01 6.80324495e-01 7.33148932e-01
4.49123383e-01 -6.76889345e-02 8.35327625e-01 2.04857569e-02
2.25553080e-01 -1.05570229e-02 -4.31724966e-01 7.51866579e-01
8.06542695e-01 4.01544958e-01 -5.58998168e-01 1.07781559e-01
7.29881585e-01 3.64475012e-01 -5.31037688e-01 -8.66560280e-01
-1.41372776e+00 7.09745109e-01 1.03819084e+00 -3.21825504e-01
-3.43260854e-01 -1.01858705e-01 -3.49149197e-01 4.10967767e-01
5.95475025e-02 4.38313454e-01 -2.58653164e-01 -1.03979558e-01
-8.61789942e-01 4.82272685e-01 6.05689585e-01 1.72901571e+00
1.12326849e+00 1.60018563e-01 8.12574089e-01 1.51735455e-01
2.55428702e-01 1.09476185e+00 1.10914208e-01 -9.93280292e-01
7.06381261e-01 3.14134449e-01 6.05466008e-01 -1.32696497e+00
-5.97877681e-01 -6.48725510e-01 -7.38327801e-01 5.77689350e-01
1.77688047e-01 -2.10547149e-01 -8.58823061e-01 1.38410330e+00
5.22058904e-01 1.06387317e-01 8.97333845e-02 1.25436640e+00
5.11226356e-01 4.40755874e-01 -7.43359864e-01 1.31421223e-01
1.12941289e+00 -8.86291206e-01 -5.16877651e-01 -8.97957027e-01
3.78413677e-01 -5.15912950e-01 6.88776970e-01 1.13296770e-01
-3.52201223e-01 -4.33869064e-01 -1.27592611e+00 -1.35935456e-01
-7.52689600e-01 3.81158553e-02 2.26414353e-01 4.50406313e-01
-1.24215174e+00 1.60040393e-01 -8.43931556e-01 -8.70840132e-01
-5.07851653e-02 2.33217895e-01 -8.84995282e-01 -3.08441639e-01
-1.18723810e+00 1.03361261e+00 -4.47747186e-02 4.47950095e-01
-8.69999170e-01 -4.17920023e-01 -1.28087997e+00 -3.05856109e-01
3.47698599e-01 -6.24545872e-01 8.49028289e-01 -3.87410700e-01
-1.18632364e+00 9.14126277e-01 -3.12694907e-01 -5.98187506e-01
7.12687612e-01 -3.41727197e-01 -7.80194938e-01 -2.08768398e-01
5.20483553e-01 6.56559408e-01 3.69998336e-01 -1.08021808e+00
-9.97446775e-01 -7.87251294e-01 3.56189817e-01 8.66218865e-01
6.39062285e-01 -6.82761133e-01 -2.21257508e-01 7.52212945e-03
7.56085515e-01 -1.47619486e+00 -4.50820804e-01 1.87718287e-01
-2.31778637e-01 8.64954829e-01 1.00277495e+00 -1.86618179e-01
5.56324482e-01 -2.01105547e+00 -6.82180822e-02 2.37120643e-01
-7.98054412e-02 -1.88600555e-01 2.34883681e-01 6.69900775e-01
3.10477048e-01 -1.71926066e-01 -1.91334516e-01 -1.57434598e-03
-2.80287027e-01 4.40976620e-01 -2.49900967e-01 9.23243165e-01
-4.28217530e-01 6.66361570e-01 -1.04893398e+00 1.23357773e-01
4.40902174e-01 2.90949136e-01 -3.79536569e-01 -9.56328399e-03
2.92051822e-01 5.72631776e-01 -2.05463618e-01 8.56701314e-01
1.14480901e+00 3.04753095e-01 -1.86521605e-01 2.67055601e-01
-9.70090389e-01 7.45090991e-02 -1.10430789e+00 1.83749545e+00
-3.63972485e-01 1.11836803e+00 2.36986592e-01 -1.42508954e-01
9.57419932e-01 -3.17659020e-01 -1.41467616e-01 -7.09973991e-01
-2.06838608e-01 2.53325939e-01 -1.86327994e-01 -2.04521954e-01
1.11085701e+00 -1.94585398e-01 -6.32226095e-02 -3.58525723e-01
-4.23272997e-01 -4.93821830e-01 5.94670773e-02 2.12829616e-02
9.74102974e-01 -1.29892547e-02 6.76701188e-01 -4.00949061e-01
4.46154952e-01 5.38652599e-01 4.09887850e-01 6.13279045e-01
-3.41228724e-01 7.96830773e-01 -1.57617144e-02 -5.50573766e-01
-7.88958311e-01 -1.19436765e+00 -1.66125551e-01 4.95977312e-01
8.64389002e-01 -3.29190731e-01 -8.58260095e-02 -1.87911049e-01
1.80830389e-01 4.56940114e-01 -4.48195636e-01 1.02033820e-02
-2.86537290e-01 -3.99284601e-01 2.97036916e-01 3.19159955e-01
1.05060160e+00 -3.27709496e-01 -1.08005917e+00 -1.04788631e-01
-3.12671423e-01 -1.28259587e+00 -8.64123926e-02 4.20683175e-02
-6.16793633e-01 -9.30223286e-01 -5.73001146e-01 -4.39523071e-01
7.89881051e-01 1.25460339e+00 8.33829284e-01 -4.05619264e-01
-1.01874806e-01 2.29114145e-01 -1.30015999e-01 -3.41903299e-01
4.83904511e-01 3.06526963e-02 4.28400576e-01 7.44960755e-02
2.20062807e-01 -5.01264870e-01 -6.58281922e-01 7.82966316e-01
-1.04337908e-01 -7.56488219e-02 3.96528751e-01 4.74542499e-01
7.41867542e-01 1.77506015e-01 -3.03829670e-01 -5.82774937e-01
-1.58615917e-01 -7.53906488e-01 -1.23773289e+00 -2.78224975e-01
-2.56500214e-01 -5.26127458e-01 5.23280382e-01 2.62332380e-01
-7.08774090e-01 2.58452743e-01 1.81623518e-01 -1.83103636e-01
-3.87956262e-01 3.97530615e-01 -2.47943118e-01 -5.66466272e-01
8.25690448e-01 5.12605496e-02 -1.57903180e-01 5.86213395e-02
3.27470481e-01 5.34264565e-01 7.88672924e-01 1.95736229e-01
1.35732925e+00 9.58263457e-01 5.38945079e-01 -1.46135354e+00
-3.45944762e-01 -5.94623148e-01 -1.11601996e+00 -2.17610806e-01
8.02923262e-01 -1.31262171e+00 -5.00199080e-01 2.25664720e-01
-5.25332391e-01 -3.28735083e-01 1.05368286e-01 6.29652977e-01
-4.63318080e-01 2.33083777e-02 2.76524782e-01 -5.03381789e-01
2.97177374e-01 -8.49428654e-01 1.04368091e+00 6.03203535e-01
1.11474231e-01 -9.43491876e-01 2.52881557e-01 1.46810815e-01
4.00759250e-01 6.02818549e-01 9.35171023e-02 1.71057060e-01
-1.21188927e+00 -3.33052188e-01 8.03123191e-02 -3.16271186e-01
2.29580864e-01 -1.91435799e-01 -7.42759407e-01 -6.35839760e-01
-1.54070720e-01 1.26684949e-01 6.58747911e-01 2.83924967e-01
2.68710315e-01 2.01825313e-02 -5.83919942e-01 1.23902845e+00
1.62589717e+00 4.32903916e-01 5.68816066e-01 6.21417582e-01
7.84502447e-01 5.58561504e-01 1.31715071e+00 3.33130985e-01
6.59763753e-01 8.34671021e-01 1.09116364e+00 -1.07624285e-01
4.35324609e-01 -3.50838900e-01 1.51994303e-01 5.76805770e-02
-1.65314734e-01 -3.83031338e-01 -1.15591300e+00 5.91228545e-01
-1.53331053e+00 -9.11886096e-01 -4.26794857e-01 2.71791720e+00
-3.06132883e-01 -2.02865135e-02 -4.74705607e-01 -3.19669008e-01
4.78365362e-01 3.96890730e-01 -7.66218364e-01 -1.57439001e-02
-4.81545813e-02 -6.86679900e-01 1.37440956e+00 9.50065374e-01
-1.18048179e+00 1.01623499e+00 5.70975542e+00 -1.46535501e-01
-1.30700183e+00 -1.91300452e-01 -3.06298405e-01 -6.88839406e-02
-2.49758735e-01 3.11932594e-01 -1.05995953e+00 1.63374096e-01
8.61430705e-01 -1.22765832e-01 4.95575547e-01 1.08903551e+00
8.87375847e-02 -6.43860817e-01 -5.18190742e-01 1.04621983e+00
1.74766928e-01 -1.10825706e+00 -3.21720630e-01 5.20236969e-01
8.55411589e-01 7.33153820e-01 2.60188013e-01 4.68546562e-02
9.71137583e-02 -7.55832434e-01 8.08494568e-01 2.51141131e-01
1.00206053e+00 -8.71594667e-01 5.60284972e-01 7.80008674e-01
-1.46694195e+00 1.53367743e-01 -6.24229610e-01 -3.32516253e-01
2.88712293e-01 1.32678226e-01 -1.14289093e+00 7.08517253e-01
7.13654816e-01 8.54866803e-01 -7.22808540e-01 1.12081754e+00
-2.74702579e-01 -2.08593696e-01 -5.52986205e-01 2.98582673e-01
6.35871053e-01 -6.42809331e-01 7.19621956e-01 5.34491122e-01
8.21323812e-01 2.11378679e-01 1.18133299e-01 4.99160618e-01
3.48433733e-01 -4.46118653e-01 -1.71064079e+00 3.94388795e-01
3.67035210e-01 1.35024559e+00 -7.42521405e-01 -4.92024906e-02
-2.22253025e-01 9.95428443e-01 -1.28201932e-01 5.18197834e-01
-9.21388686e-01 -6.76733971e-01 1.26644504e+00 5.34101963e-01
1.35561109e-01 -1.09939468e+00 2.11274400e-01 -1.19312322e+00
1.10132424e-02 -1.55561209e-01 -3.45669806e-01 -1.06936979e+00
-2.69891739e-01 6.28340065e-01 -3.11843734e-02 -1.78014624e+00
-3.81754160e-01 -5.64804375e-01 -3.43138397e-01 1.05641472e+00
-1.76687026e+00 -8.31748664e-01 -1.16000569e+00 6.78517222e-01
4.74538684e-01 -1.45324707e-01 7.00863481e-01 1.72138318e-01
-8.04201141e-02 -1.78940415e-01 4.36507881e-01 -3.67503822e-01
6.87762976e-01 -1.18212593e+00 9.11053419e-01 1.27536154e+00
8.45311433e-02 6.94324255e-01 8.40338469e-01 -7.56913662e-01
-1.57375896e+00 -1.32573807e+00 1.05786192e+00 -5.23565769e-01
2.55710036e-01 -6.25659227e-01 -4.84832853e-01 1.17647183e+00
-1.51290610e-01 3.68480980e-01 2.84790695e-01 -1.32427096e-01
2.45509401e-01 -3.40084881e-01 -1.33141184e+00 9.31017280e-01
1.41236234e+00 -4.41024274e-01 -1.43294707e-01 4.39926386e-01
2.39168212e-01 -1.05071211e+00 -1.05990261e-01 4.22711551e-01
6.61863029e-01 -1.24135172e+00 9.97701287e-01 -1.48658574e-01
-2.42822826e-01 -8.74365985e-01 -5.86161673e-01 -1.44026744e+00
-5.32666802e-01 -5.25554180e-01 5.43460190e-01 6.64167106e-01
9.57341865e-02 -9.03551996e-01 8.11718941e-01 1.67806268e-01
-3.07123870e-01 -1.54549941e-01 -1.05851078e+00 -9.12928104e-01
-6.77082777e-01 -2.47709423e-01 5.54972649e-01 5.68576336e-01
-6.34767950e-01 1.47422627e-01 -4.10431027e-01 1.09570861e+00
6.78735673e-01 8.73863623e-02 1.30819857e+00 -1.37915552e+00
5.88419616e-01 2.15791345e-01 -1.00822008e+00 -1.25303185e+00
4.83405553e-02 -7.46649444e-01 1.75825000e-01 -1.61080539e+00
-7.34484673e-01 -2.40130320e-01 4.08332735e-01 -6.31943420e-02
4.33424413e-01 2.77319670e-01 -2.34327018e-01 1.61744669e-01
-3.46462756e-01 6.28353059e-01 9.08691823e-01 8.57776552e-02
-2.67650355e-02 4.57558990e-01 -3.40290993e-01 8.37929845e-01
5.92375040e-01 -2.99405009e-01 -8.02991271e-01 -5.34946442e-01
4.81321484e-01 2.10407689e-01 4.87159103e-01 -1.52424014e+00
5.70999980e-01 -5.03629625e-01 3.85670960e-01 -1.11863708e+00
7.96965301e-01 -1.15154123e+00 4.92512941e-01 3.87069792e-01
3.79632175e-01 6.21338308e-01 7.94034004e-02 7.71319032e-01
-3.54245931e-01 9.03348550e-02 7.99819410e-01 -4.01952535e-01
-1.48549259e+00 4.37414914e-01 -5.57350993e-01 -6.20764121e-02
1.38868475e+00 -6.83941603e-01 -4.46533978e-01 -6.08338714e-01
-4.48666930e-01 5.54865062e-01 1.12095320e+00 6.63232625e-01
7.24539757e-01 -1.33161891e+00 -3.44967514e-01 1.12573755e+00
5.83567023e-01 1.40143543e-01 3.87568414e-01 9.15359795e-01
-1.31986642e+00 8.34848285e-01 -6.61059618e-01 -9.23877299e-01
-1.27572012e+00 2.81776786e-01 4.78022754e-01 5.27113438e-01
-6.71244800e-01 6.88077331e-01 3.26523930e-01 -7.66624153e-01
-3.95744443e-01 -5.25071263e-01 6.62932992e-02 -1.36704758e-01
4.74542886e-01 5.38769066e-01 2.13161841e-01 -1.26745391e+00
-7.92676508e-01 1.06674111e+00 3.41779590e-01 -4.38735813e-01
7.36686110e-01 -9.07242298e-01 3.06288868e-01 5.28743982e-01
8.91772926e-01 2.92402744e-01 -1.35710108e+00 1.02056429e-01
-3.01131278e-01 -1.08862412e+00 6.58184960e-02 -2.82327861e-01
-6.88875794e-01 9.60090041e-01 7.47950017e-01 -1.78821132e-01
7.27777064e-01 -4.61207956e-01 2.49881193e-01 1.01624990e+00
1.29118061e+00 -6.48231566e-01 -7.30965436e-01 1.00361061e+00
7.86924422e-01 -1.58926451e+00 8.30492228e-02 -1.92455173e-01
-7.61670649e-01 1.00126076e+00 6.50046349e-01 -7.32721910e-02
5.99006116e-01 -7.96093047e-02 3.85064274e-01 -1.22362874e-01
-3.99051249e-01 -5.53256989e-01 9.49200094e-02 9.44857240e-01
-1.69362277e-01 1.77234963e-01 4.16563332e-01 -2.97477812e-01
-7.20309615e-01 -3.50250870e-01 7.99513996e-01 1.24124992e+00
-6.56065881e-01 -4.05415595e-01 -4.04769659e-01 3.24286241e-03
4.69153821e-01 9.77664217e-02 -2.41856173e-01 1.22263396e+00
3.69738370e-01 8.91731441e-01 2.56207496e-01 -5.14796853e-01
5.98872304e-01 -2.95432299e-01 3.07747424e-01 -6.76222503e-01
2.52048597e-02 -3.75365347e-01 8.65685046e-02 -8.36693585e-01
-4.92763286e-03 -7.84203649e-01 -1.12821543e+00 -5.03260076e-01
9.23523605e-02 -2.28809211e-02 9.87476289e-01 5.70104957e-01
4.45044935e-01 1.06634296e-01 6.37655675e-01 -1.28262770e+00
9.18736160e-02 -4.58672851e-01 -1.12018347e+00 -3.81713390e-01
9.20775950e-01 -9.22476947e-01 -4.56126779e-01 -2.65984267e-01]
|
[7.266829967498779, -1.9391705989837646]
|
5292bc9b-657b-4a96-a813-76f4dd6cbd65
|
mts2graph-interpretable-multivariate-time
|
2306.03834
| null |
https://arxiv.org/abs/2306.03834v1
|
https://arxiv.org/pdf/2306.03834v1.pdf
|
MTS2Graph: Interpretable Multivariate Time Series Classification with Temporal Evolving Graphs
|
Conventional time series classification approaches based on bags of patterns or shapelets face significant challenges in dealing with a vast amount of feature candidates from high-dimensional multivariate data. In contrast, deep neural networks can learn low-dimensional features efficiently, and in particular, Convolutional Neural Networks (CNN) have shown promising results in classifying Multivariate Time Series (MTS) data. A key factor in the success of deep neural networks is this astonishing expressive power. However, this power comes at the cost of complex, black-boxed models, conflicting with the goals of building reliable and human-understandable models. An essential criterion in understanding such predictive deep models involves quantifying the contribution of time-varying input variables to the classification. Hence, in this work, we introduce a new framework for interpreting multivariate time series data by extracting and clustering the input representative patterns that highly activate CNN neurons. This way, we identify each signal's role and dependencies, considering all possible combinations of signals in the MTS input. Then, we construct a graph that captures the temporal relationship between the extracted patterns for each layer. An effective graph merging strategy finds the connection of each node to the previous layer's nodes. Finally, a graph embedding algorithm generates new representations of the created interpretable time-series features. To evaluate the performance of our proposed framework, we run extensive experiments on eight datasets of the UCR/UEA archive, along with HAR and PAM datasets. The experiments indicate the benefit of our time-aware graph-based representation in MTS classification while enriching them with more interpretability.
|
['Zahra Ahmadi', 'Abdul Hakmeh', 'Raneen Younis']
|
2023-06-06
| null | null | null | null |
['graph-embedding', 'time-series-classification']
|
['graphs', 'time-series']
|
[ 8.80263820e-02 -1.93273678e-01 -6.75691897e-03 -2.88261592e-01
-5.01955897e-02 -4.84823644e-01 5.72244167e-01 6.29364908e-01
-1.57520190e-01 4.49774086e-01 -2.05060933e-03 -2.15242058e-01
-8.71860862e-01 -7.57270157e-01 -4.37706590e-01 -7.40011215e-01
-8.15036952e-01 1.86107576e-01 -8.42550844e-02 -2.89563268e-01
-2.56947353e-02 7.85322189e-01 -1.60229933e+00 1.86776400e-01
6.18843675e-01 1.34641445e+00 -1.66520894e-01 3.57346565e-01
-1.06358014e-01 6.27688825e-01 -5.39754331e-01 -1.25858217e-01
1.30258664e-01 -2.88626015e-01 -3.10832620e-01 8.47755522e-02
-9.40258056e-03 3.71815920e-01 -5.63869357e-01 6.60079002e-01
1.14365049e-01 1.19220540e-01 5.96197009e-01 -1.55546570e+00
-3.43360245e-01 5.99182308e-01 -3.58875781e-01 6.53585196e-01
-6.83783367e-02 1.19841538e-01 1.15153241e+00 -4.81774211e-01
3.82143229e-01 9.29489851e-01 4.34512287e-01 5.56992181e-02
-1.36091852e+00 -4.22228724e-01 2.17201114e-01 4.77296084e-01
-1.11909688e+00 -1.41542763e-01 1.34472108e+00 -5.18299401e-01
9.50551748e-01 4.96725708e-01 9.32736516e-01 1.09120262e+00
3.06902826e-01 5.50294161e-01 8.63374352e-01 -1.48498908e-01
2.51403660e-01 -2.24514693e-01 5.34400880e-01 5.41769087e-01
7.13057593e-02 -1.50751024e-01 -5.21377981e-01 -1.00879498e-01
4.26810026e-01 3.87124926e-01 -2.66700834e-01 -2.29068860e-01
-1.17091012e+00 5.42449832e-01 6.75083697e-01 8.19603503e-01
-5.43122530e-01 1.51649401e-01 5.77379227e-01 3.91585201e-01
5.59636652e-01 4.67091531e-01 -4.89957184e-01 -5.41243777e-02
-5.83871603e-01 -1.03463769e-01 4.28034574e-01 3.48007113e-01
6.12237930e-01 2.08171785e-01 -2.15290617e-02 5.78597307e-01
-8.31236616e-02 4.37443554e-02 6.36822581e-01 -2.37566218e-01
4.30738866e-01 1.08351851e+00 -5.23819327e-01 -1.72836685e+00
-7.16502845e-01 -8.86304855e-01 -1.29614949e+00 -1.94446012e-01
1.78480640e-01 2.04440668e-01 -6.68842614e-01 1.73201871e+00
9.59292203e-02 3.39858174e-01 -1.05771452e-01 6.65739119e-01
4.13308471e-01 7.11957157e-01 4.63090427e-02 -3.02328825e-01
1.21950781e+00 -4.16712642e-01 -6.22392416e-01 -3.10516208e-02
3.71526510e-01 -9.91433412e-02 8.02888691e-01 2.58653432e-01
-5.29434085e-01 -6.43004596e-01 -1.05876374e+00 2.43875310e-01
-6.84795439e-01 1.61925241e-01 9.04472411e-01 6.48107976e-02
-8.42434525e-01 8.99625480e-01 -8.51383865e-01 -1.91245228e-01
4.07361031e-01 4.86636221e-01 -5.46325803e-01 2.85789430e-01
-1.20816910e+00 4.87182915e-01 4.78237391e-01 5.07102966e-01
-4.58913565e-01 -5.48297286e-01 -7.89814472e-01 4.35498685e-01
7.80014619e-02 -3.76936167e-01 4.07622039e-01 -1.11951303e+00
-7.95269907e-01 4.34235483e-01 -8.47705752e-02 -6.74167991e-01
2.04922706e-01 2.64623076e-01 -6.75798416e-01 1.39804631e-01
-1.09183297e-01 2.41774976e-01 1.04549205e+00 -8.96890521e-01
-4.17529881e-01 -6.67933702e-01 -6.36428073e-02 -1.97841629e-01
-8.50799859e-01 -3.42517853e-01 -2.05495104e-01 -7.35165656e-01
3.50385696e-01 -7.67128408e-01 -1.55994833e-01 -2.23845273e-01
-2.63341218e-01 -4.01064396e-01 1.08690453e+00 -5.45242071e-01
1.45081925e+00 -2.35753632e+00 4.26827848e-01 5.20962656e-01
6.65202379e-01 -4.24965657e-02 -1.11488946e-01 4.98791009e-01
-4.87154901e-01 1.71769589e-01 -3.44265252e-01 -3.01698416e-01
-1.17407858e-01 2.59055197e-01 -4.40819532e-01 4.18155938e-01
5.19597471e-01 8.68634522e-01 -7.82637835e-01 -1.78872734e-01
2.64613390e-01 4.81513232e-01 -1.62825853e-01 -5.90940826e-02
-2.10559174e-01 5.52379072e-01 -4.99889702e-01 3.76008958e-01
1.23401254e-01 -4.40303832e-01 3.88884872e-01 -5.09952068e-01
-4.73346449e-02 -4.16057482e-02 -7.92788565e-01 1.30102694e+00
-2.88091660e-01 9.11737740e-01 -5.52397251e-01 -1.53944039e+00
1.18990290e+00 2.09785968e-01 8.85456979e-01 -8.44809771e-01
2.43731588e-01 1.49988666e-01 1.80121735e-01 -5.42112529e-01
1.51645303e-01 1.15604773e-01 -8.50200504e-02 2.38541976e-01
1.11088537e-01 4.37251449e-01 1.92356184e-01 7.93355703e-02
1.20790303e+00 -3.30939621e-01 2.26667300e-01 -6.18992746e-02
5.05049348e-01 -2.18350172e-01 4.26760674e-01 9.98164713e-02
-5.85985258e-02 1.81841031e-01 8.64036918e-01 -9.83203590e-01
-8.77641201e-01 -7.01743960e-01 4.70095985e-02 7.00146794e-01
-1.98140278e-01 -4.74103659e-01 -1.64579526e-01 -5.00084400e-01
4.88560088e-03 3.92400831e-01 -8.89614701e-01 -3.63161951e-01
-6.03019476e-01 -7.56515324e-01 5.56445897e-01 5.16023397e-01
2.86189497e-01 -1.05992138e+00 -6.06011331e-01 2.81746686e-01
-2.75475889e-01 -1.04549801e+00 7.17598051e-02 4.58150119e-01
-1.15579641e+00 -1.01321280e+00 -1.70568794e-01 -5.50543725e-01
6.81051016e-01 1.58264205e-01 8.58413339e-01 1.26467705e-01
-4.52007413e-01 1.05982497e-01 -3.37461501e-01 -2.69295543e-01
-1.04459841e-02 4.26955260e-02 1.51995309e-02 5.82959831e-01
2.10677564e-01 -9.57990110e-01 -4.91960555e-01 2.97185015e-02
-1.05153871e+00 -1.15481010e-02 3.41890186e-01 6.43681586e-01
5.36503673e-01 4.22468066e-01 7.24249601e-01 -4.88647789e-01
8.06979001e-01 -6.75164044e-01 -4.25111711e-01 1.18560091e-01
-3.50787014e-01 2.09835529e-01 1.13101304e+00 -4.16271210e-01
-3.39985937e-01 3.73456674e-03 3.08446616e-01 -7.98514545e-01
-9.34825242e-02 9.68427658e-01 7.09515885e-02 2.02159181e-01
5.45693636e-01 3.62594038e-01 -1.20613538e-02 -3.06908101e-01
1.87051058e-01 2.51024663e-01 2.89300889e-01 -4.64943588e-01
7.27890730e-01 4.91726786e-01 5.25128484e-01 -9.16634142e-01
-3.86300653e-01 -2.60049164e-01 -7.67886579e-01 -4.17255521e-01
6.90581083e-01 -4.06868696e-01 -8.03296268e-01 3.50696474e-01
-9.44341063e-01 1.51201859e-01 -1.85759142e-01 3.85788321e-01
-2.67037839e-01 3.86234790e-01 -3.21524084e-01 -8.09792578e-01
-2.32898235e-01 -8.46731246e-01 8.35185528e-01 9.27983001e-02
-3.62997830e-01 -1.11106098e+00 5.34231681e-03 9.34661776e-02
2.33401805e-01 7.95891166e-01 1.45778751e+00 -8.24485004e-01
-4.48607802e-01 -5.01495540e-01 -4.54815999e-02 2.10155860e-01
1.89556614e-01 1.05228789e-01 -9.86954749e-01 -1.41978458e-01
6.74260184e-02 6.42106459e-02 8.08478296e-01 3.02360684e-01
1.54741502e+00 -3.38909745e-01 -3.25723618e-01 5.33884466e-01
1.34733534e+00 3.57759953e-01 3.68849903e-01 2.71740288e-01
7.50755250e-01 7.04589844e-01 2.68432558e-01 5.00261903e-01
2.63744742e-01 5.15278220e-01 6.75441206e-01 -4.40719873e-02
4.28876132e-01 -6.14469424e-02 1.57564938e-01 1.13822126e+00
-3.15121472e-01 -2.67170370e-01 -1.02849746e+00 5.14171958e-01
-1.93047345e+00 -9.73027706e-01 -2.30095074e-01 2.03089309e+00
3.18079054e-01 2.94265926e-01 3.19203213e-02 8.08062375e-01
5.26466012e-01 4.36619073e-01 -4.53920990e-01 -2.84371167e-01
-3.13840002e-01 2.57215619e-01 7.20623285e-02 -8.86642337e-02
-1.16690791e+00 4.84190643e-01 4.75983810e+00 5.80222964e-01
-1.49744403e+00 -3.03345710e-01 7.40460217e-01 3.17424424e-02
-3.15842420e-01 -2.71651298e-01 -7.22653493e-02 3.74182791e-01
1.04675007e+00 -3.44029903e-01 4.72094715e-01 4.79352236e-01
4.04688865e-01 4.25735027e-01 -1.31664240e+00 9.99367356e-01
-7.68530741e-02 -1.32310629e+00 1.51173353e-01 7.39386529e-02
2.46914133e-01 -1.95303127e-01 1.86728299e-01 2.50871658e-01
-4.06881660e-01 -1.02535129e+00 6.64102793e-01 6.01823688e-01
3.73791516e-01 -7.02802896e-01 5.40144265e-01 3.17025393e-01
-1.55507088e+00 -3.52796853e-01 -1.70994014e-01 -1.87067583e-01
-2.02992275e-01 6.38100445e-01 -8.32260847e-01 8.77122641e-01
5.60893059e-01 1.18411839e+00 -7.84335792e-01 8.39041948e-01
8.94897953e-02 5.58223546e-01 -2.08642319e-01 -9.72287282e-02
3.26258242e-01 -2.44597271e-01 5.28437853e-01 1.00361884e+00
3.43606025e-01 1.17679343e-01 -7.91927129e-02 8.45863283e-01
1.47266723e-02 4.65022847e-02 -9.26939845e-01 -5.81675470e-01
3.05627435e-01 1.41629052e+00 -1.06855059e+00 -1.08587213e-01
-2.34131545e-01 6.13526285e-01 4.86377925e-01 4.56290692e-01
-8.69793773e-01 -2.02036813e-01 6.20651603e-01 -1.58548616e-02
1.40961614e-02 -6.42875075e-01 -3.08645844e-01 -1.06212223e+00
2.43235290e-01 -7.95156479e-01 6.07588530e-01 -5.57185352e-01
-1.34505367e+00 8.51391435e-01 -1.52955592e-01 -1.55941665e+00
-1.71292394e-01 -5.34129083e-01 -6.49633229e-01 7.12645590e-01
-1.14810991e+00 -1.12322199e+00 -5.18351555e-01 7.90961385e-01
5.04012346e-01 -8.58389139e-02 7.86363721e-01 3.84964734e-01
-6.57997787e-01 1.99929729e-01 -7.54607916e-02 2.83357859e-01
-7.52530023e-02 -9.77518618e-01 3.31237793e-01 7.70933330e-01
4.36866790e-01 5.79787016e-01 4.46409345e-01 -3.60516667e-01
-1.72311318e+00 -1.08841395e+00 7.53724158e-01 -2.32987061e-01
9.77297366e-01 -4.13514405e-01 -1.10660696e+00 5.17036259e-01
-1.24795012e-01 2.36036815e-02 5.87674379e-01 1.65641993e-01
-3.79888624e-01 -4.25148845e-01 -6.22649252e-01 6.53109074e-01
9.34774935e-01 -6.11523211e-01 -4.25252557e-01 1.80376023e-01
7.03820825e-01 1.54226929e-01 -8.55602622e-01 4.71060514e-01
4.95947063e-01 -8.11275959e-01 9.33594823e-01 -8.83171141e-01
5.08676589e-01 -3.00674766e-01 -4.89472784e-03 -1.42048168e+00
-3.62223417e-01 -3.20317000e-01 -1.50483504e-01 9.82141316e-01
3.09247106e-01 -6.35174572e-01 4.52912480e-01 3.39377522e-01
-2.21765459e-01 -9.51687694e-01 -9.16825175e-01 -7.38307297e-01
-4.19856280e-01 -6.63746297e-01 6.65810525e-01 1.30110800e+00
1.59423515e-01 3.38503361e-01 -2.49583393e-01 2.30065197e-01
5.45234621e-01 2.38460824e-01 4.88452643e-01 -1.42908120e+00
-1.84038877e-01 -6.10246480e-01 -9.67445791e-01 -1.06263660e-01
2.40936324e-01 -1.07088900e+00 -4.53189522e-01 -1.25628912e+00
-1.83391735e-01 -4.39395577e-01 -6.94044411e-01 5.58426440e-01
-7.91227072e-03 -1.67315647e-01 1.92481697e-01 2.24969238e-01
-2.83167869e-01 6.14149153e-01 9.93007123e-01 -4.16118801e-01
-1.77897304e-01 -1.12328991e-01 -1.93584859e-01 4.80133593e-01
8.39100003e-01 -2.69917160e-01 -6.41480982e-01 -4.35542941e-01
3.06649238e-01 1.97296038e-01 6.04290009e-01 -1.00482309e+00
2.79081017e-01 -1.25859277e-02 4.72271532e-01 -4.56982166e-01
3.48275632e-01 -1.10385823e+00 5.57325184e-01 3.92915368e-01
-2.62041211e-01 3.54923368e-01 3.52437884e-01 6.87280357e-01
-5.84352255e-01 3.30979019e-01 3.55536282e-01 1.43583581e-01
-7.93602943e-01 5.15247285e-01 -2.55592942e-01 -2.96284795e-01
9.23322201e-01 -1.89153120e-01 -3.32443923e-01 -2.87100792e-01
-8.50034118e-01 7.83137158e-02 -1.35000169e-01 7.95013428e-01
7.38900781e-01 -1.32297909e+00 -3.15571845e-01 5.46187699e-01
2.95757592e-01 -3.36626023e-01 2.42072701e-01 1.00889432e+00
-2.48042852e-01 3.52501720e-01 -4.84804332e-01 -7.19633102e-01
-1.03757119e+00 6.16642952e-01 2.95184821e-01 -2.07527518e-01
-5.96715033e-01 3.17119420e-01 4.77060042e-02 1.18826538e-01
1.95390806e-01 -6.99779451e-01 -6.53836966e-01 4.23792869e-01
1.21674508e-01 2.31405467e-01 1.94364727e-01 -5.91912091e-01
-5.06112337e-01 4.80711907e-01 3.46587926e-01 1.51132271e-01
1.65439630e+00 2.21793383e-01 -4.05281335e-01 7.67074347e-01
1.34603429e+00 -3.60231429e-01 -6.76892400e-01 -1.38222754e-01
3.12684745e-01 -2.49131903e-01 -1.69085473e-01 -2.64497906e-01
-1.25849760e+00 9.94759917e-01 4.35315430e-01 6.65601015e-01
1.46476018e+00 -1.46739736e-01 4.19273823e-01 3.47986192e-01
2.56290644e-01 -5.99108338e-01 1.66996509e-01 4.14374083e-01
9.47355986e-01 -9.48635161e-01 -3.28645945e-01 -1.90073445e-01
-3.41443837e-01 1.50929654e+00 4.07761991e-01 -8.93200189e-02
7.37067819e-01 -4.95239682e-02 -2.79074281e-01 -6.09312654e-01
-8.70484650e-01 -4.73854169e-02 5.89350939e-01 2.96245694e-01
2.99417496e-01 2.75962055e-01 -2.83435702e-01 7.96118677e-01
-1.11150466e-01 -2.28657410e-01 1.50357693e-01 6.48683548e-01
-1.84531286e-01 -7.79345989e-01 -3.67325693e-02 5.13220131e-01
-1.52688578e-01 1.49692565e-01 -4.45438266e-01 7.46914625e-01
4.96569388e-02 7.69747198e-01 2.08509386e-01 -7.88418412e-01
3.81246567e-01 7.22720772e-02 9.54428911e-02 -2.39731371e-01
-6.24540269e-01 2.88651306e-02 -4.54496183e-02 -5.31520665e-01
-4.19403553e-01 -4.75882500e-01 -1.23790073e+00 -1.24676831e-01
1.36041909e-01 2.73417123e-02 6.16489291e-01 8.96222293e-01
4.67359662e-01 8.65775347e-01 9.04242635e-01 -6.99918509e-01
-1.73337281e-01 -6.96562767e-01 -5.45874178e-01 7.07865238e-01
3.15181583e-01 -6.43369257e-01 -4.31344450e-01 7.63240233e-02]
|
[7.15551233291626, 2.982861280441284]
|
4ce8fd74-44ba-401b-b9d1-ec62028c7e35
|
rethinking-few-shot-class-incremental
|
2207.09963
| null |
https://arxiv.org/abs/2207.09963v1
|
https://arxiv.org/pdf/2207.09963v1.pdf
|
Rethinking Few-Shot Class-Incremental Learning with Open-Set Hypothesis in Hyperbolic Geometry
|
Few-Shot Class-Incremental Learning (FSCIL) aims at incrementally learning novel classes from a few labeled samples by avoiding the overfitting and catastrophic forgetting simultaneously. The current protocol of FSCIL is built by mimicking the general class-incremental learning setting, while it is not totally appropriate due to the different data configuration, i.e., novel classes are all in the limited data regime. In this paper, we rethink the configuration of FSCIL with the open-set hypothesis by reserving the possibility in the first session for incoming categories. To assign better performances on both close-set and open-set recognition to the model, Hyperbolic Reciprocal Point Learning module (Hyper-RPL) is built on Reciprocal Point Learning (RPL) with hyperbolic neural networks. Besides, for learning novel categories from limited labeled data, we incorporate a hyperbolic metric learning (Hyper-Metric) module into the distillation-based framework to alleviate the overfitting issue and better handle the trade-off issue between the preservation of old knowledge and the acquisition of new knowledge. The comprehensive assessments of the proposed configuration and modules on three benchmark datasets are executed to validate the effectiveness concerning three evaluation indicators.
|
['Li Liu', 'Wei Peng', 'Zitong Yu', 'Yawen Cui']
|
2022-07-20
| null | null | null | null |
['few-shot-class-incremental-learning', 'open-set-learning']
|
['methodology', 'miscellaneous']
|
[ 2.15174884e-01 3.27362508e-01 -1.77187428e-01 -3.36692721e-01
-3.62854362e-01 -4.16464508e-01 4.72022742e-01 4.73676294e-01
-5.42052746e-01 7.79299259e-01 -2.65892237e-01 -7.02904090e-02
-5.53708673e-01 -1.00017798e+00 -6.48098528e-01 -9.56728816e-01
-3.93045172e-02 4.90581065e-01 5.55454373e-01 1.13451632e-03
2.66004831e-01 3.26074541e-01 -1.93663704e+00 2.73301512e-01
9.98951733e-01 1.06835926e+00 -4.57717590e-02 2.80325174e-01
-2.64648765e-01 7.46303618e-01 -2.90958107e-01 -5.43368340e-01
3.43354970e-01 -3.13976556e-01 -6.75631940e-01 -7.29119480e-02
2.90022314e-01 -8.36874694e-02 -1.28006011e-01 8.22838187e-01
5.28283119e-01 4.26128536e-01 4.27315593e-01 -1.37544513e+00
-5.49209535e-01 7.06370175e-01 -3.38808566e-01 1.90693647e-01
2.83085350e-02 2.58196056e-01 7.23272622e-01 -1.26729918e+00
5.53640127e-01 9.10583854e-01 7.80810297e-01 5.27860641e-01
-9.97868598e-01 -5.06193519e-01 3.92777205e-01 7.71865666e-01
-1.46145916e+00 -4.18109477e-01 7.55877078e-01 -2.31932089e-01
3.83097053e-01 1.00233607e-01 6.80248618e-01 1.05795085e+00
1.30964993e-02 7.79513896e-01 1.00358987e+00 -4.50654030e-01
7.84898400e-01 6.03473485e-01 6.91561222e-01 3.80288869e-01
2.68191338e-01 2.48891234e-01 -5.08446991e-01 1.03528857e-01
4.06096764e-02 5.05565464e-01 -4.62973677e-02 -7.74076998e-01
-8.77154291e-01 6.12370491e-01 5.53673923e-01 3.66286218e-01
-1.17454700e-01 -5.44296980e-01 5.62295496e-01 4.72500205e-01
3.30264866e-01 2.24352702e-01 -5.32774210e-01 1.04812905e-02
-9.28250551e-01 -3.20964716e-02 6.73687220e-01 9.96126831e-01
8.84493172e-01 -5.27930558e-02 -5.40181875e-01 7.34105051e-01
-1.51795864e-01 2.68953741e-02 9.64577019e-01 -4.74924713e-01
9.67823789e-02 9.90785599e-01 -2.07703456e-01 -6.05972111e-01
-5.50651789e-01 -1.07006586e+00 -8.30605328e-01 1.81472301e-01
3.01716268e-01 7.12618902e-02 -9.40112948e-01 1.65642643e+00
5.84194124e-01 5.22733867e-01 2.14165345e-01 4.39770609e-01
7.49136746e-01 5.78953207e-01 5.87196015e-02 -6.51081979e-01
9.85466480e-01 -9.16282594e-01 -3.73480618e-01 -5.04624099e-02
7.30380058e-01 -4.98868451e-02 1.26513004e+00 4.57411230e-01
-5.87161303e-01 -7.78032243e-01 -1.18925118e+00 2.34273434e-01
-7.79835641e-01 -2.16432452e-01 1.97138444e-01 5.36566079e-01
-5.02539694e-01 7.54547358e-01 -4.77203012e-01 -3.37018400e-01
4.54675406e-01 8.22743848e-02 -1.76231295e-01 -3.44713777e-01
-1.23072946e+00 6.18209779e-01 9.84987795e-01 1.20489709e-01
-7.47672319e-01 -9.22570765e-01 -3.42768818e-01 2.51563072e-01
5.36324620e-01 -3.82653981e-01 9.09390926e-01 -8.90312672e-01
-1.40320051e+00 5.10924041e-01 3.33397180e-01 -5.46640635e-01
6.96666121e-01 2.60826293e-02 -4.21987653e-01 -1.84710160e-01
-1.73930436e-01 5.89998364e-01 7.96329498e-01 -1.21252763e+00
-6.86422646e-01 -4.78621423e-01 3.62581760e-02 3.79246652e-01
-7.27096736e-01 -9.37521875e-01 -6.45851567e-02 -4.72992390e-01
2.06996903e-01 -9.43851650e-01 1.98820144e-01 -1.05455630e-01
-1.18993342e-01 -3.17593098e-01 9.57758129e-01 -1.73958734e-01
1.44472492e+00 -2.24930024e+00 -2.93772165e-02 -2.96799652e-02
9.65472683e-02 5.60128033e-01 -9.51970890e-02 1.08030736e-01
-1.26560181e-01 -3.73271137e-01 -3.98390204e-01 -2.49562114e-01
-1.84859425e-01 2.85922259e-01 -2.93304563e-01 1.76230386e-01
-1.19417503e-01 7.40639865e-01 -1.07382131e+00 -3.74212265e-01
9.70303118e-02 3.03289983e-02 -5.76937437e-01 7.70384595e-02
-2.13422045e-01 2.85284758e-01 7.63048232e-02 7.37512767e-01
8.45669627e-01 -1.19232699e-01 2.48344280e-02 -5.69890626e-02
-8.55224282e-02 -4.19557303e-01 -1.34704208e+00 1.63824308e+00
-4.29781556e-01 -7.81765208e-03 -6.37190521e-01 -9.57333922e-01
9.78554904e-01 6.59714118e-02 3.83835286e-01 -6.86359525e-01
5.83668984e-02 1.57518357e-01 2.50000623e-03 -4.42636400e-01
3.90405148e-01 -3.19838792e-01 9.41137746e-02 3.49152237e-01
4.30781811e-01 4.05482680e-01 1.82844177e-01 7.17307702e-02
1.04530704e+00 5.58831692e-02 4.29067671e-01 -7.54134059e-02
6.62426651e-01 -4.44721393e-02 8.49563897e-01 8.55136156e-01
-4.75702167e-01 4.91916537e-01 2.38391206e-01 -6.37991548e-01
-7.46854424e-01 -1.22956717e+00 -3.37501168e-01 1.38130522e+00
2.27371261e-01 -1.43006653e-01 -4.18229967e-01 -1.16003430e+00
6.28255829e-02 1.22610188e+00 -8.15671444e-01 -9.25857902e-01
-2.04061821e-01 -9.64038014e-01 2.73042530e-01 3.77153337e-01
4.99194235e-01 -1.10107517e+00 -6.71035826e-01 2.39928886e-01
3.28351222e-02 -5.62002182e-01 -2.02004761e-01 5.31012475e-01
-9.99065995e-01 -1.34336650e+00 -5.34068227e-01 -7.03685820e-01
6.24509215e-01 2.73510396e-01 6.56470895e-01 -2.63637841e-01
-2.17088670e-01 3.40842366e-01 -6.71148896e-01 -3.84310991e-01
-3.35504591e-01 4.99263674e-01 2.31066316e-01 4.09104019e-01
5.11330962e-01 -7.06487358e-01 -5.46261132e-01 1.96468413e-01
-1.06763506e+00 4.06418368e-03 6.19541943e-01 1.01584721e+00
5.72265387e-01 2.31125236e-01 1.06693339e+00 -1.11018825e+00
3.04419547e-01 -8.18007648e-01 -3.10296506e-01 5.10067642e-01
-1.19540095e+00 7.58155063e-02 8.80345702e-01 -8.00063610e-01
-1.17427945e+00 -1.85151488e-01 1.12036496e-01 -6.10567808e-01
-3.44453231e-02 2.73948044e-01 -2.30345100e-01 1.26318038e-01
7.86670566e-01 4.07948941e-01 -8.43225196e-02 -5.29920936e-01
5.67560494e-01 5.62672317e-01 4.63418126e-01 -2.07319930e-01
7.88063943e-01 3.69808584e-01 -2.02400371e-01 -4.68844354e-01
-1.18463159e+00 -3.47684771e-01 -9.87434864e-01 -3.34973305e-01
2.73120314e-01 -6.78929090e-01 -4.51515377e-01 6.50890410e-01
-6.17933512e-01 -1.52051106e-01 -1.00754392e+00 2.67330408e-01
-4.55689728e-01 2.10202560e-01 -2.85184145e-01 -6.21689141e-01
-3.68669301e-01 -6.88813746e-01 4.10903603e-01 4.85688835e-01
2.12832704e-01 -7.74631321e-01 2.68796742e-01 6.42557070e-02
4.93108928e-01 1.51517719e-01 1.13788581e+00 -1.16506183e+00
-2.78262824e-01 -2.86800057e-01 -3.85003425e-02 4.49822158e-01
4.35991026e-02 -3.72112066e-01 -1.20984840e+00 -6.46888912e-01
8.37952271e-02 -5.14202237e-01 8.62833500e-01 -1.79578766e-01
1.03836548e+00 -3.37020606e-01 -1.57670259e-01 5.40293634e-01
1.44426596e+00 3.46666008e-01 4.77660060e-01 4.72747922e-01
4.80015934e-01 5.28410256e-01 6.99189126e-01 6.26727283e-01
3.10523033e-01 4.79122847e-01 3.99996072e-01 3.85384679e-01
-2.59830475e-01 -4.38635558e-01 1.68336272e-01 9.56767619e-01
3.32009465e-01 -1.02211910e-04 -8.14044237e-01 3.26907277e-01
-1.85527670e+00 -9.45445597e-01 4.16918516e-01 2.59766889e+00
7.94168413e-01 4.91381168e-01 -3.64340208e-02 4.23084974e-01
8.30624938e-01 -1.47356391e-01 -9.98602211e-01 -6.08463138e-02
-1.14930168e-01 -1.83149517e-01 9.76571292e-02 2.90079951e-01
-9.86357450e-01 6.11763418e-01 4.92265368e+00 9.68523622e-01
-1.06844842e+00 3.32850337e-01 6.52282834e-01 -3.86401713e-01
-5.82666695e-02 1.17357925e-01 -9.47775304e-01 6.04741037e-01
8.52018237e-01 -3.62412900e-01 3.66344810e-01 9.84274149e-01
-2.11256236e-01 -1.72027387e-02 -1.20072508e+00 8.28019321e-01
1.28645197e-01 -1.03007245e+00 1.61000162e-01 -3.71455342e-01
7.03617036e-01 -1.62950680e-01 1.68261528e-01 9.75528121e-01
-1.33041576e-01 -2.24226177e-01 7.24645913e-01 8.84417653e-01
7.43376970e-01 -7.75018692e-01 7.85538912e-01 6.62018716e-01
-9.73255396e-01 -8.11870933e-01 -4.16220278e-01 2.14704707e-01
-2.16032892e-01 7.31701255e-01 -9.06688154e-01 6.07275426e-01
7.16789067e-01 6.53773427e-01 -1.11554623e+00 1.31001604e+00
1.43359125e-01 5.62449157e-01 -1.59819871e-01 2.67738074e-01
4.72609214e-02 -8.77948552e-02 5.63009560e-01 8.44421864e-01
2.85179108e-01 -7.50850067e-02 6.81446716e-02 5.30163109e-01
-4.38981596e-03 7.07463399e-02 -5.27064443e-01 4.79547441e-01
7.83285141e-01 1.09839594e+00 -7.62867033e-01 -4.40968782e-01
-2.15813369e-01 7.28424489e-01 5.77049911e-01 2.25910008e-01
-7.20925510e-01 -4.75143284e-01 -1.40453078e-04 2.53686786e-01
2.26434112e-01 2.62864441e-01 -1.32402852e-01 -1.07509124e+00
8.72647464e-02 -5.56550801e-01 8.45574737e-01 -5.45699716e-01
-1.37267089e+00 4.69075263e-01 1.32143840e-01 -1.48993802e+00
-1.72038246e-02 -1.02457210e-01 -6.68761849e-01 9.87228006e-02
-1.54968083e+00 -9.51943517e-01 -6.19199336e-01 6.24949157e-01
7.01125920e-01 -2.50355691e-01 5.58412671e-01 3.95295054e-01
-8.14430773e-01 9.73375738e-01 4.99066770e-01 -4.27507222e-01
7.22930789e-01 -1.11740315e+00 -1.01525389e-01 7.21097827e-01
-1.61400914e-01 3.86800557e-01 3.40455085e-01 -4.69332665e-01
-1.13924778e+00 -1.54420924e+00 5.56035697e-01 -2.75208771e-01
4.11778003e-01 -3.68878037e-01 -1.29596746e+00 4.85299349e-01
-3.56438816e-01 1.46003664e-01 6.89747453e-01 2.22857948e-02
-4.64780986e-01 -7.19932139e-01 -1.26222444e+00 4.08552200e-01
1.10880363e+00 -1.59744039e-01 -7.55077899e-01 2.10186407e-01
1.01834607e+00 1.03789434e-01 -6.89727902e-01 6.68403268e-01
4.87476557e-01 -9.67097640e-01 7.23665416e-01 -5.06920815e-01
-7.71090314e-02 -3.44441682e-01 -1.18551739e-01 -1.25754392e+00
-5.35446525e-01 -1.90495193e-01 -5.14352739e-01 1.41967523e+00
2.18766794e-01 -7.29612112e-01 7.90747046e-01 2.18052149e-01
-9.05575380e-02 -8.12817335e-01 -1.25598979e+00 -1.07096052e+00
7.17081875e-02 -4.42797877e-02 6.40951931e-01 9.55780327e-01
-4.94908616e-02 4.39781159e-01 -2.21324354e-01 -5.43652251e-02
6.72106087e-01 2.16671228e-01 4.86034364e-01 -1.51857507e+00
-2.61790812e-01 -2.31931746e-01 -3.90362412e-01 -3.78018618e-01
-1.00906797e-01 -1.12818289e+00 -5.38655929e-02 -8.85505319e-01
2.20103756e-01 -7.04585969e-01 -9.02010024e-01 5.58802605e-01
-2.90480882e-01 6.04955554e-02 4.30661052e-01 4.11610007e-01
-9.73717749e-01 9.69573319e-01 8.53004873e-01 -1.49007514e-01
-5.86401105e-01 2.24574283e-01 -6.32827282e-01 5.21330714e-01
6.37346029e-01 -6.33006036e-01 -8.40549290e-01 -6.22294843e-02
1.34553760e-01 -1.61280721e-01 1.65549278e-01 -1.51849389e+00
6.01312160e-01 5.82511351e-02 2.70578206e-01 -6.02590799e-01
2.69227251e-02 -9.37524498e-01 7.16968551e-02 5.72501361e-01
-4.56940472e-01 -2.39522770e-01 3.93326990e-02 9.71115232e-01
8.31667483e-02 -4.62982059e-01 1.11851823e+00 -5.53849451e-02
-7.66985893e-01 4.25865650e-01 1.12630902e-02 6.75427914e-02
1.44115949e+00 -4.11534280e-01 -3.77976447e-01 2.47822598e-01
-1.08273351e+00 2.24296287e-01 3.45929503e-01 5.89884520e-01
6.30870938e-01 -1.28807366e+00 -4.37086135e-01 3.99526656e-01
5.94879925e-01 1.29867345e-01 6.12729609e-01 6.96338594e-01
-8.32573026e-02 1.00414671e-01 -2.05336913e-01 -5.22694170e-01
-8.12672198e-01 1.13614666e+00 2.63661653e-01 -3.97339821e-01
-6.05348229e-01 4.63434458e-01 -1.07872533e-03 -7.89281130e-01
5.76608419e-01 1.60760935e-02 -3.42698753e-01 4.69193012e-01
5.45988619e-01 8.63778710e-01 3.99751544e-01 1.38446763e-02
-2.11088642e-01 2.04384074e-01 -4.21282828e-01 3.79941255e-01
1.25915730e+00 -2.65926868e-01 1.03473410e-01 1.02538025e+00
9.93030369e-01 -5.36242485e-01 -1.39449418e+00 -6.11133695e-01
8.76487941e-02 -2.41457857e-02 -2.47797608e-01 -9.30838227e-01
-7.06249952e-01 7.31770217e-01 1.20112383e+00 -3.46577838e-02
1.14847195e+00 -3.01889211e-01 5.82473218e-01 6.56579018e-01
5.05541265e-01 -1.27458704e+00 2.93260783e-01 5.43019354e-01
5.85299671e-01 -1.07214773e+00 -1.18256966e-02 5.52114062e-02
-4.00450647e-01 1.13582945e+00 8.59020293e-01 2.28424788e-01
9.04387772e-01 -2.89421558e-01 -2.99482435e-01 6.54672906e-02
-9.69264925e-01 8.63867626e-03 -1.14907706e-02 4.32495803e-01
-2.12235779e-01 6.51515787e-03 -2.58877456e-01 7.87292421e-01
1.17299289e-01 1.67803779e-01 5.62061429e-01 9.84709561e-01
-7.03285635e-01 -7.04479277e-01 -1.15571536e-01 5.40911019e-01
3.66560757e-01 2.51597185e-02 3.59209999e-03 6.50615931e-01
6.84284747e-01 7.27628529e-01 8.94427523e-02 -5.67475140e-01
4.44981694e-01 5.17664731e-01 1.93235829e-01 -6.36703551e-01
-4.24405992e-01 -5.93946278e-01 -5.72806954e-01 -1.99419782e-01
-8.86550173e-02 -5.37417054e-01 -9.24461842e-01 -1.41250238e-01
-4.11971241e-01 3.30389321e-01 2.73004234e-01 8.92920971e-01
5.74036658e-01 4.88489538e-01 9.58097160e-01 -4.28334236e-01
-9.66521442e-01 -9.47590530e-01 -6.71431184e-01 3.24077457e-01
1.08765028e-01 -7.72175610e-01 -5.87602556e-01 -1.05321288e-01]
|
[9.82714557647705, 3.324594497680664]
|
c8b9603c-6030-40fd-b47d-10a9ee4e3e3a
|
3d-semantic-segmentation-of-modular-furniture
| null | null |
https://ieeexplore.ieee.org/document/7926598
|
http://web-info8.informatik.rwth-aachen.de/media/papers/egpaper_final.pdf
|
3D Semantic Segmentation of Modular Furniture using rjMCMC
|
In this paper we propose a novel approach to identify and label the structural elements of furniture e.g. wardrobes, cabinets etc. Given a furniture item, the subdivision into its structural components like doors, drawers and shelves is difficult as the number of components and their spatial arrangements varies severely. Furthermore, structural elements are primarily distinguished by their function rather than by unique color or texture based appearance features. It is therefore difficult to classify them, even if their correct spatial extent were known. In our approach we jointly estimate the number of functional units, their spatial structure, and their corresponding labels by using reversible jump MCMC (rjMCMC), a method well suited for optimization on spaces of varying dimensions (the number of structural elements). Optionally, our system permits to invoke depth information e.g. from RGB-D cameras, which are already frequently mounted on mobile robot platforms. We show a considerable improvement over a baseline method even without using depth data, and an additional performance gain when depth input is enabled.
|
['Bastian (*equal contribution)', 'Markus; Leibe', 'Manu; Mathias', 'Ishrat; Tom*', 'Badami*']
|
2017-05-15
| null | null | null |
wacv-2017-2017-5
|
['furniture-segmentation']
|
['computer-vision']
|
[ 1.98048726e-01 -6.56801835e-02 -1.03685874e-02 -1.15450267e-02
-2.89803833e-01 -1.01462615e+00 5.83023489e-01 2.96840191e-01
-4.01598632e-01 7.28059709e-01 -1.35030180e-01 -1.85784519e-01
-1.35319382e-01 -8.29128087e-01 -6.51166081e-01 -7.39216149e-01
3.72234499e-03 9.22127426e-01 3.70542794e-01 5.92901446e-02
4.37518120e-01 9.87629652e-01 -1.93091798e+00 -2.74289455e-02
5.09524405e-01 9.29818630e-01 5.59738457e-01 7.52949834e-01
-7.88181424e-02 4.64374155e-01 -4.94105667e-01 2.35296845e-01
8.00454244e-02 -1.71637923e-01 -8.27415347e-01 6.72569513e-01
1.56291291e-01 -2.99156159e-01 1.37967616e-01 7.48979270e-01
7.90025070e-02 3.85278463e-01 1.15967596e+00 -1.01648688e+00
-1.79539695e-01 -3.51696461e-03 -5.75554788e-01 -4.36384469e-01
4.43915129e-01 -2.70148307e-01 9.45492685e-01 -7.84966350e-01
6.29350126e-01 1.06470013e+00 4.40209061e-01 2.67065674e-01
-1.52524674e+00 1.97676220e-03 1.94674984e-01 -6.40553162e-02
-1.61518729e+00 -4.42772806e-01 6.17981732e-01 -6.40619755e-01
8.40073526e-01 3.68914366e-01 4.74535495e-01 5.08768737e-01
-1.45768926e-01 6.63294613e-01 9.13711309e-01 -5.61927497e-01
7.13493288e-01 2.81884044e-01 4.42533903e-02 8.54397237e-01
1.71750903e-01 -5.55880308e-01 -5.15647940e-02 3.59633146e-03
1.20443141e+00 3.35255831e-01 -2.29779221e-02 -9.91324782e-01
-1.24663091e+00 5.44435978e-01 4.08230871e-01 3.41247767e-01
-3.33367646e-01 1.99918523e-01 8.76733568e-03 -4.98864800e-02
8.29885826e-02 3.81199390e-01 -5.26816726e-01 -1.53267505e-02
-7.72643447e-01 -1.40355021e-01 7.68508673e-01 1.20667624e+00
1.17838418e+00 -4.64976043e-01 6.44509435e-01 7.95662820e-01
1.86030120e-01 7.47709870e-02 1.85119058e-03 -1.15561461e+00
4.18772072e-01 7.96731770e-01 5.23986578e-01 -7.50076592e-01
-7.06218481e-01 1.49602473e-01 -8.82874966e-01 4.01129723e-01
5.61538756e-01 -2.62127374e-03 -7.89627552e-01 1.32021594e+00
4.93339121e-01 -5.57248853e-02 -2.50351578e-01 8.06608856e-01
2.83686191e-01 3.79624039e-01 -2.97632694e-01 -2.73861978e-02
1.53684819e+00 -7.26327002e-01 -4.54542398e-01 -4.84644249e-02
6.23210549e-01 -7.06910014e-01 9.13524389e-01 6.16779745e-01
-1.00616503e+00 -5.17895699e-01 -1.00467598e+00 -9.16700289e-02
-5.07884800e-01 4.73566085e-01 6.49404645e-01 7.72350669e-01
-1.01507425e+00 5.31918466e-01 -1.09552145e+00 -4.47027653e-01
-1.44218013e-01 7.33071327e-01 -5.60028076e-01 7.73408562e-02
-3.99236292e-01 7.92074919e-01 1.76385567e-01 3.25768977e-01
-5.82349479e-01 1.72761992e-01 -1.00187135e+00 -3.14421505e-02
5.30743241e-01 -3.87279809e-01 9.73624349e-01 -5.55208147e-01
-1.71484625e+00 9.07413542e-01 -1.19380899e-01 1.14680193e-01
3.93561810e-01 1.77569389e-01 1.77781910e-01 1.21322401e-01
-1.02622278e-01 6.43773019e-01 7.87161529e-01 -1.46204448e+00
-6.12657249e-01 -5.49917936e-01 4.67866689e-01 2.86028415e-01
-1.70650724e-02 -3.13952416e-01 -6.18721306e-01 -2.46086866e-01
6.93667829e-01 -1.19436061e+00 -3.01647961e-01 1.33748993e-01
-6.29823923e-01 -1.08186148e-01 5.31652987e-01 -3.50430667e-01
8.94588590e-01 -2.10089016e+00 6.29853070e-01 3.54266286e-01
7.08736479e-02 -2.87452817e-01 2.94322670e-01 4.75331366e-01
6.46842495e-02 -6.80595711e-02 -3.70945126e-01 -4.61965531e-01
2.70133525e-01 5.39262474e-01 2.41145968e-01 7.52987862e-01
-1.05781145e-01 4.19623911e-01 -7.31127143e-01 -4.48200464e-01
6.49094284e-01 4.19397384e-01 -3.39743376e-01 -1.37251288e-01
-8.82092044e-02 4.94045317e-01 -5.39999366e-01 7.82950342e-01
6.41112149e-01 -2.61368990e-01 4.26016152e-01 2.65988596e-02
-3.26252788e-01 2.23381117e-01 -1.74383676e+00 1.79543400e+00
-7.00069666e-01 4.34652448e-01 4.56846803e-01 -9.21839297e-01
9.07719731e-01 2.66871691e-01 4.90564644e-01 -2.81139135e-01
-9.85610187e-02 1.74288616e-01 -4.13930535e-01 -2.80489177e-01
8.42773318e-01 -1.13337934e-02 -3.63678277e-01 4.91446525e-01
-4.10855621e-01 -2.80842125e-01 2.61874571e-02 -2.94356108e-01
1.16256917e+00 3.24652940e-01 4.69681591e-01 -3.00258428e-01
4.88607079e-01 -7.75474533e-02 2.25156769e-01 4.24300849e-01
1.86699033e-01 8.32887173e-01 4.96814698e-01 -2.98888415e-01
-1.10496533e+00 -1.15403306e+00 -3.80192339e-01 9.52121198e-01
3.72402519e-01 -1.83930203e-01 -7.13787258e-01 -4.70122933e-01
-4.93043847e-03 2.61034161e-01 -6.52282834e-01 3.31271708e-01
-7.25241303e-01 -4.99346286e-01 -8.92544389e-02 4.27057594e-01
1.48457140e-01 -8.00596714e-01 -1.05761611e+00 2.76807338e-01
-6.66902214e-03 -8.09359014e-01 -8.16875398e-02 5.48413098e-01
-1.12326515e+00 -1.05359530e+00 -6.97051585e-01 -8.68495047e-01
1.11699748e+00 4.00476396e-01 9.44548726e-01 1.56726837e-02
-3.45418394e-01 5.22213221e-01 -3.34657460e-01 1.69269159e-01
1.58728268e-02 1.92469001e-01 4.75052372e-02 3.25714387e-02
-1.52708262e-01 -5.33215404e-01 -6.56584978e-01 7.52836525e-01
-8.91172349e-01 -6.04000129e-02 3.85854602e-01 4.51443374e-01
8.07824016e-01 3.00524205e-01 -2.14269400e-01 -8.41388106e-01
1.07889883e-01 -3.52005512e-01 -7.25054204e-01 2.46243626e-01
-1.42930314e-01 2.29249254e-01 3.77999872e-01 -1.47628769e-01
-7.33567834e-01 5.79288363e-01 9.14767757e-02 2.79406253e-02
-5.84139943e-01 2.42502689e-01 -4.73331749e-01 1.60940483e-01
2.33269453e-01 -1.53072923e-01 -2.48833477e-01 -8.12383175e-01
3.06357086e-01 7.40956008e-01 3.60789418e-01 -6.60596371e-01
3.08151126e-01 6.38879836e-01 2.76305199e-01 -1.09370363e+00
-3.76049355e-02 -6.80491149e-01 -1.21353757e+00 -1.10897243e-01
8.01118791e-01 -6.32922709e-01 -1.03465509e+00 2.62425274e-01
-1.18727827e+00 -3.58417541e-01 -4.49638069e-02 4.71642584e-01
-7.74539113e-01 2.89575696e-01 -5.61624348e-01 -9.77086544e-01
3.65130693e-01 -1.20653415e+00 1.44513559e+00 -2.11340964e-01
-3.41574192e-01 -9.19829488e-01 -6.62478283e-02 1.44156933e-01
-1.33909360e-01 4.00572389e-01 9.45781887e-01 3.38731334e-02
-8.14605772e-01 -2.55484104e-01 1.25228064e-02 -9.52065215e-02
4.25332904e-01 1.04531668e-01 -7.09854007e-01 -2.29519159e-01
-1.20866597e-01 1.14201657e-01 5.23624599e-01 5.37671506e-01
9.40032601e-01 -1.15330681e-01 -4.76829827e-01 2.78235078e-01
1.51421976e+00 2.52033621e-01 6.60585880e-01 4.12919074e-01
7.29978919e-01 1.03656590e+00 6.94025993e-01 5.51888168e-01
3.75090986e-01 9.83222306e-01 6.05194092e-01 1.94822401e-02
2.08129510e-01 1.73995867e-01 1.61966816e-01 5.71674168e-01
-2.89503723e-01 -3.48158002e-01 -9.00752664e-01 5.10190189e-01
-1.79481077e+00 -5.84131002e-01 -4.08299655e-01 2.45423293e+00
3.32367122e-01 7.50959367e-02 2.74477214e-01 3.60789657e-01
8.27565253e-01 -2.69443661e-01 -4.69158143e-01 -2.14301035e-01
1.57804161e-01 -1.31927982e-01 7.46569812e-01 5.59426785e-01
-1.05172050e+00 3.80458325e-01 6.30167150e+00 6.25396848e-01
-5.59486628e-01 -1.26888409e-01 5.57363153e-01 -5.78561686e-02
-1.10359415e-01 -3.38167287e-02 -7.96950638e-01 4.51742351e-01
6.40804231e-01 8.00911009e-01 5.67098439e-01 6.67120099e-01
3.02601699e-02 -7.86454320e-01 -1.29779470e+00 1.07479310e+00
-1.65696502e-01 -9.05223906e-01 -5.69031596e-01 3.33674312e-01
6.02215767e-01 -2.15796992e-01 -2.41512209e-02 -2.19449744e-01
2.18255863e-01 -6.52526796e-01 9.15748894e-01 4.28933889e-01
8.03314328e-01 -7.32370198e-01 4.05362248e-01 4.43231910e-01
-1.36371434e+00 9.46014971e-02 -5.00155509e-01 -2.44385988e-01
1.54008538e-01 2.85700113e-01 -6.64935172e-01 2.37443313e-01
4.30188060e-01 3.85371566e-01 -2.48021975e-01 8.77229273e-01
-2.93624490e-01 7.71990642e-02 -6.44246817e-01 -1.43114418e-01
1.88208625e-01 -5.90944648e-01 1.28982842e-01 8.06367576e-01
5.12877107e-01 7.57993832e-02 3.71515900e-02 5.42759001e-01
1.97927460e-01 -1.29642740e-01 -5.26115298e-01 3.54591548e-01
3.63390416e-01 1.36562073e+00 -1.50494635e+00 -1.48313433e-01
-3.66449744e-01 1.17308080e+00 2.60628182e-02 3.51914823e-01
-6.89587057e-01 -3.05325925e-01 5.80009043e-01 2.85850644e-01
5.16082883e-01 -6.76061630e-01 -1.67597532e-01 -9.81143773e-01
1.82767853e-01 -3.80491972e-01 5.13290353e-02 -7.34789252e-01
-8.39423954e-01 4.40054148e-01 -8.18112493e-02 -1.42779684e+00
-3.30863833e-01 -8.91542912e-01 7.98453912e-02 6.18289948e-01
-1.09707475e+00 -9.03912663e-01 -3.10071651e-02 4.58886862e-01
3.97851229e-01 2.70747989e-01 9.82701778e-01 1.78825468e-01
-5.87753475e-01 -1.96057297e-02 6.60668731e-01 -3.37245129e-02
2.82804459e-01 -1.56887889e+00 2.89795995e-01 3.83348435e-01
3.05906117e-01 5.40268064e-01 5.87620676e-01 -4.15654182e-01
-1.52329075e+00 -5.98881185e-01 5.83983004e-01 -7.05322385e-01
4.28025037e-01 -9.16739047e-01 -5.80555499e-01 6.28549695e-01
-2.11753443e-01 -2.25738138e-01 4.35614228e-01 3.32663745e-01
-5.38941752e-03 9.20649711e-03 -1.04560041e+00 4.13298011e-01
9.85049725e-01 -5.30571163e-01 -3.31529200e-01 3.88444483e-01
2.11181894e-01 -3.44453424e-01 -7.01785326e-01 5.97589416e-03
6.65113747e-01 -1.11779857e+00 1.06729281e+00 6.65448159e-02
9.53850672e-02 -6.10071957e-01 -3.11022907e-01 -9.04477715e-01
-2.94979066e-01 -4.03019756e-01 1.45602599e-01 1.04299033e+00
2.71296769e-01 -4.99976218e-01 1.17279184e+00 8.78228784e-01
-1.60823792e-01 -2.64556170e-01 -1.03047144e+00 -7.44481385e-01
-4.44708914e-01 -2.70173460e-01 6.85218275e-01 8.15682232e-01
-3.10441311e-02 3.59593570e-01 -2.14910492e-01 2.48551175e-01
3.60174358e-01 4.34876055e-01 6.65823102e-01 -1.43051791e+00
-5.26524842e-01 -3.32119972e-01 -5.66566229e-01 -1.35923564e+00
-1.02722339e-01 -5.12758851e-01 1.80661395e-01 -1.63373005e+00
-2.16968041e-02 -8.76758218e-01 -1.64445236e-01 3.70148212e-01
4.55048174e-01 3.14903706e-01 -1.07450537e-01 2.14091703e-01
-6.78030014e-01 1.67861372e-01 1.01492882e+00 -1.07701659e-01
-3.38713825e-01 1.35694489e-01 -9.17916652e-03 9.08167422e-01
6.87711060e-01 -3.88395727e-01 -2.14532718e-01 -5.56033731e-01
3.70459139e-01 4.06535834e-01 2.74046659e-01 -1.01734316e+00
1.23106658e-01 -9.64312255e-02 3.07611436e-01 -7.94400930e-01
7.62840509e-01 -1.39070308e+00 5.92705846e-01 3.46551031e-01
2.89816596e-02 1.74335480e-01 -1.84041802e-02 5.88259757e-01
2.68927068e-02 -6.87505186e-01 3.59220237e-01 -4.30714875e-01
-6.22875988e-01 -1.28089890e-01 -8.12061369e-01 -7.58485615e-01
1.07349539e+00 -6.47679508e-01 1.53506637e-01 -7.06998110e-02
-9.85883117e-01 -9.52154994e-02 1.14013958e+00 -2.28093384e-04
5.19503593e-01 -1.26989353e+00 -9.16131660e-02 2.80780196e-01
7.02076182e-02 3.10409576e-01 1.54064193e-01 5.32105684e-01
-9.09720600e-01 4.38371241e-01 -2.71693319e-01 -6.58654928e-01
-1.26472986e+00 6.41874790e-01 1.66867934e-02 1.05644893e-02
-3.99548352e-01 6.85210168e-01 3.87442499e-01 -3.81775647e-01
2.81951219e-01 -7.96183467e-01 -2.11200058e-01 4.12416667e-01
2.05909505e-01 6.07419312e-01 1.13764353e-01 -6.95780337e-01
-3.35928828e-01 8.81591737e-01 3.23309392e-01 -1.60941720e-01
1.36616647e+00 -5.13370454e-01 -4.42890644e-01 7.57007837e-01
1.04955375e+00 -5.88524751e-02 -1.43197310e+00 -7.75154680e-02
1.34900361e-01 -4.12544012e-01 -2.09404811e-01 -2.17964172e-01
-6.98602319e-01 9.93218422e-01 3.46943140e-01 4.17997599e-01
1.12918270e+00 5.75631149e-02 2.55224049e-01 5.86645722e-01
9.77269650e-01 -1.08295822e+00 -2.87492070e-02 2.60702223e-01
5.89028776e-01 -9.38108802e-01 1.13643244e-01 -4.88895506e-01
-2.17443541e-01 1.11681712e+00 -2.48410534e-02 -1.28886020e-02
3.13057691e-01 2.48739138e-01 -3.58254641e-01 -2.12131873e-01
-3.37510794e-01 -4.12718803e-01 1.06464118e-01 4.24805939e-01
2.86674798e-01 3.32140982e-01 1.08510174e-01 7.16874748e-03
5.97377121e-02 -5.36306381e-01 5.34570456e-01 1.26216459e+00
-6.11642241e-01 -1.26079905e+00 -7.34356403e-01 3.12231123e-01
-1.80363566e-01 1.28422037e-01 -2.33887866e-01 9.15026546e-01
3.02465975e-01 9.69931602e-01 4.28453594e-01 -2.41943374e-01
4.15947258e-01 -2.03967497e-01 7.59491563e-01 -8.00397992e-01
-1.86106771e-01 3.91827762e-01 -1.85632464e-02 -2.10507348e-01
-5.94245136e-01 -1.05661500e+00 -1.26462913e+00 -4.72754724e-02
-4.47017133e-01 4.26342189e-02 1.06213403e+00 7.06671834e-01
-2.27503162e-02 2.64645666e-01 6.37160361e-01 -1.15680683e+00
1.17481358e-01 -6.52476907e-01 -1.08282173e+00 5.28434105e-02
4.41209555e-01 -8.80685329e-01 -5.18859513e-02 2.82911569e-01]
|
[8.105355262756348, -2.499748468399048]
|
301d8185-01bf-4aa4-a1af-0f8a0926aaa5
|
generative-meta-learning-for-zero-shot
|
2305.01920
| null |
https://arxiv.org/abs/2305.01920v1
|
https://arxiv.org/pdf/2305.01920v1.pdf
|
Generative Meta-Learning for Zero-Shot Relation Triplet Extraction
|
The zero-shot relation triplet extraction (ZeroRTE) task aims to extract relation triplets from a piece of text with unseen relation types. The seminal work adopts the pre-trained generative model to generate synthetic samples for new relations. However, current generative models lack the optimization process of model generalization on different tasks during training, and thus have limited generalization capability. For this reason, we propose a novel generative meta-learning framework which exploits the `learning-to-learn' ability of meta-learning to boost the generalization capability of generative models. Specifically, we first design a task-aware generative model which can learn the general knowledge by forcing the optimization process to be conducted across multiple tasks. Based on it, we then present three generative meta-learning approaches designated for three typical meta-learning categories. Extensive experimental results demonstrate that our framework achieves a new state-of-the-art performance for the ZeroRTE task.
|
['Tieyun Qian', 'Wanli Li']
|
2023-05-03
| null | null | null | null |
['general-knowledge', 'zero-shot-relation-triplet-extraction']
|
['miscellaneous', 'natural-language-processing']
|
[ 2.70449281e-01 6.86207056e-01 -4.03965533e-01 -4.78910416e-01
-9.64234114e-01 -1.28821954e-01 9.98365223e-01 -3.06883872e-01
1.71765625e-01 7.70850956e-01 8.14740062e-02 -2.28558749e-01
-6.97929412e-02 -9.54240620e-01 -7.47230291e-01 -5.78826725e-01
2.70252913e-01 9.37416673e-01 -2.58479547e-02 -4.96980250e-01
-3.03669274e-01 3.91101581e-04 -1.30611324e+00 7.06692040e-01
1.02333713e+00 7.51329958e-01 -1.06503051e-02 4.64569569e-01
-1.49229258e-01 9.27229762e-01 -6.92937732e-01 -8.56999159e-01
-6.54757991e-02 -7.17967391e-01 -1.04762232e+00 2.09815770e-01
-1.27709195e-01 -6.48634955e-02 -5.75698674e-01 7.66602159e-01
4.18275952e-01 2.66150415e-01 9.87390578e-01 -1.70822740e+00
-1.23653316e+00 1.08348870e+00 -2.33807102e-01 1.38035163e-01
3.37067693e-01 -5.23810089e-02 1.26585603e+00 -1.10827291e+00
8.61015260e-01 1.29747581e+00 4.86286253e-01 8.61040652e-01
-1.36679792e+00 -5.99232435e-01 1.99878424e-01 4.00336653e-01
-1.37357116e+00 -5.36180019e-01 8.49313140e-01 -1.51627809e-01
1.20120430e+00 1.72378540e-01 4.95376617e-01 1.69876540e+00
9.58767459e-02 1.08967817e+00 8.73710334e-01 -4.45764244e-01
2.78281093e-01 -1.47631587e-02 2.01623917e-01 4.22424793e-01
3.64917099e-01 -5.94999418e-02 -8.40719819e-01 -7.14460462e-02
5.33108830e-01 -1.59884945e-01 -2.30028957e-01 -4.14027601e-01
-1.18505824e+00 6.36127949e-01 3.55573356e-01 2.46757910e-01
-2.21944660e-01 -8.31462257e-03 2.77135402e-01 3.85446906e-01
6.77932084e-01 4.25301462e-01 -5.87873518e-01 7.16114566e-02
-4.76197511e-01 2.13524774e-01 9.47803259e-01 1.53478777e+00
9.42229927e-01 -1.14762440e-01 -4.85743016e-01 9.93273973e-01
2.34924257e-01 8.24425966e-02 5.58986008e-01 -4.92142528e-01
8.49468291e-01 8.75938714e-01 -2.24792317e-01 -4.57211852e-01
-9.55272019e-02 -5.90045691e-01 -1.12461841e+00 -3.56373608e-01
-9.98112485e-02 -3.45957786e-01 -1.12399256e+00 1.56106341e+00
3.72110128e-01 1.91003501e-01 6.36155009e-01 3.09648216e-01
1.23824608e+00 5.71142375e-01 1.86392609e-02 -3.02999020e-01
1.22295678e+00 -1.27072132e+00 -9.21921253e-01 -5.30799091e-01
7.26529896e-01 -4.12056208e-01 9.52806652e-01 6.27166554e-02
-7.60759413e-01 -6.44622326e-01 -1.15019011e+00 -1.14872120e-01
-6.19817138e-01 -7.11722067e-03 1.01075077e+00 5.56766331e-01
-4.92075503e-01 4.54633713e-01 -8.26090813e-01 -1.62620634e-01
5.65766573e-01 8.80769938e-02 -1.54928103e-01 -2.00261384e-01
-1.33758307e+00 8.53185236e-01 9.54863787e-01 1.51231647e-01
-8.68176341e-01 -7.60805905e-01 -9.91052330e-01 5.13127968e-02
9.08364475e-01 -1.24851358e+00 1.32457304e+00 -3.69114459e-01
-1.48214316e+00 6.93005204e-01 -3.63080770e-01 -3.61091465e-01
2.99170583e-01 -3.83852303e-01 -4.27609921e-01 -3.68873745e-01
8.20847377e-02 4.05460268e-01 7.81531155e-01 -1.70797634e+00
-5.62442362e-01 3.18876542e-02 8.01681168e-03 1.89599544e-01
-2.13786870e-01 -2.93280125e-01 -4.14674461e-01 -8.79869044e-01
1.98089838e-01 -8.80663276e-01 -1.22381970e-01 -9.92839634e-01
-7.98268676e-01 -7.98208594e-01 8.26224148e-01 -8.03512633e-02
1.34679043e+00 -1.89792407e+00 3.31552476e-01 -6.33201674e-02
3.75369608e-01 3.06527585e-01 -1.76663250e-01 6.02808952e-01
-2.47239724e-01 1.30387440e-01 -1.12821266e-01 -5.41696966e-01
7.85411596e-02 5.60132563e-01 -2.93773830e-01 -3.17756623e-01
5.03351927e-01 1.50671780e+00 -1.24778271e+00 -5.44016719e-01
-5.91762317e-03 2.47189566e-01 -2.21303999e-01 5.76697886e-01
-4.81299371e-01 2.35715181e-01 -5.34969807e-01 7.19198525e-01
3.76278818e-01 -5.62792182e-01 5.04615605e-01 -3.28590065e-01
6.43505156e-01 4.01085883e-01 -8.50943208e-01 1.76758409e+00
-4.14947808e-01 2.64918387e-01 -8.79485190e-01 -1.05812716e+00
1.05213749e+00 4.73484457e-01 3.64658922e-01 -1.71412125e-01
1.39320090e-01 1.66634753e-01 2.57071108e-01 -5.67169845e-01
3.53564799e-01 -2.24989906e-01 -1.48738608e-01 5.53186476e-01
7.07509279e-01 -1.68506786e-01 3.76769781e-01 2.28596181e-01
9.73667979e-01 4.86178607e-01 5.11177361e-01 2.03491226e-01
1.16512530e-01 -1.05527967e-01 6.66690290e-01 7.38995016e-01
3.35600883e-01 3.83857310e-01 4.70385373e-01 -5.36985815e-01
-7.57283330e-01 -1.14668870e+00 1.29669115e-01 1.25231910e+00
-3.66985425e-02 -8.85608077e-01 -4.66941267e-01 -1.07384884e+00
-2.90335923e-01 1.05013800e+00 -8.28640938e-01 -6.49494112e-01
-5.62149346e-01 -1.30937433e+00 1.79333389e-01 6.52290046e-01
6.10823393e-01 -1.15798807e+00 -1.85185835e-01 3.93804014e-01
-4.14955378e-01 -1.28075290e+00 -1.43619835e-01 3.90725225e-01
-8.39783549e-01 -9.53162014e-01 -4.40749824e-01 -8.80387962e-01
5.68284571e-01 1.60492718e-01 1.46731114e+00 -2.22399041e-01
1.02031589e-01 1.16265573e-01 -7.46188521e-01 -4.46844429e-01
-7.67672598e-01 7.37733006e-01 -1.96935073e-01 5.23558930e-02
6.34207606e-01 -7.92402387e-01 3.13805640e-02 1.60359517e-02
-6.60542190e-01 2.90470272e-01 8.34743261e-01 1.13725567e+00
5.39980233e-01 7.15648159e-02 8.83745372e-01 -1.29452050e+00
9.24475968e-01 -5.18305361e-01 6.04850128e-02 8.21796179e-01
-7.89033353e-01 2.79735804e-01 3.86865139e-01 -6.51170611e-01
-1.29029703e+00 -1.69250786e-01 2.56828964e-01 -4.09158349e-01
6.45696279e-03 5.56014359e-01 -5.93914092e-01 3.77614528e-01
6.65340900e-01 3.03992838e-01 -4.11259264e-01 -2.54285932e-01
8.01431537e-01 4.79509145e-01 6.84343040e-01 -7.62047648e-01
9.42166090e-01 2.28367075e-02 6.82130978e-02 -4.75580662e-01
-1.35792804e+00 -1.14747606e-01 -8.30852807e-01 1.33266980e-02
6.77331746e-01 -8.64693344e-01 -3.55377287e-01 3.62902552e-01
-1.19975162e+00 -4.95590776e-01 -5.15676916e-01 2.45818123e-01
-7.53706038e-01 -1.94924185e-03 -5.13332129e-01 -7.55485117e-01
-5.54845333e-01 -8.13196599e-01 1.04726219e+00 2.00049028e-01
-3.40000659e-01 -1.16328704e+00 2.16228500e-01 1.72022551e-01
-1.58915464e-02 2.87729025e-01 1.13698184e+00 -9.03267562e-01
-7.50851989e-01 2.14152914e-02 1.45659139e-02 2.39033014e-01
5.73260427e-01 -1.94592178e-01 -1.06973612e+00 -5.34112491e-02
-5.56766130e-02 -5.56699455e-01 9.18825686e-01 -1.02794943e-02
1.03211987e+00 -3.93871665e-01 -7.99250722e-01 7.18198478e-01
1.01168203e+00 2.51501054e-01 7.28319764e-01 2.85607159e-01
1.01505923e+00 4.25854862e-01 6.90228879e-01 1.43381447e-01
7.28095353e-01 6.45383298e-01 9.72408149e-03 6.04163343e-03
-2.53432572e-01 -5.23794770e-01 6.84133619e-02 7.93295681e-01
-4.15617794e-01 -3.97038162e-01 -9.13392961e-01 5.28992116e-01
-2.22742033e+00 -1.01416552e+00 1.77411020e-01 1.84813225e+00
1.20038378e+00 3.14886332e-01 -4.97767664e-02 1.27188131e-01
5.31761467e-01 1.45308107e-01 -4.57438290e-01 -1.51649356e-01
-1.95651222e-02 3.79360646e-01 -1.94704399e-01 2.22724676e-01
-1.08677638e+00 1.39219761e+00 6.53915977e+00 9.98947322e-01
-6.62095666e-01 1.76918328e-01 5.34983218e-01 1.95945129e-01
-4.64234710e-01 6.50371909e-02 -9.62324560e-01 8.72107372e-02
8.51159096e-01 -3.95633101e-01 2.28425324e-01 8.46168816e-01
-4.97001231e-01 3.63994271e-01 -1.63836455e+00 9.73098278e-01
1.37770146e-01 -1.29799557e+00 4.41807896e-01 6.31962046e-02
9.18202996e-01 -3.26343954e-01 2.22085062e-02 8.70102227e-01
7.17959523e-01 -9.97039795e-01 2.85255075e-01 4.75790799e-01
6.81859791e-01 -7.29386687e-01 7.20236242e-01 4.63563681e-01
-1.20165348e+00 1.04832575e-01 -3.77289861e-01 9.90836993e-02
2.62069702e-01 6.00456178e-01 -1.17185879e+00 1.09263766e+00
3.26508254e-01 8.19391429e-01 -7.43682683e-01 4.55389887e-01
-6.41441226e-01 4.65390593e-01 3.60379666e-02 -6.07418902e-02
-1.20669708e-01 -6.32602870e-02 4.06078339e-01 1.07911408e+00
1.92936629e-01 1.74899653e-01 1.35908291e-01 9.14027750e-01
-3.00501019e-01 -2.45621473e-01 -6.53740525e-01 -4.98334393e-02
5.87855458e-01 1.23080754e+00 -5.21086156e-01 -5.72926164e-01
-3.07533950e-01 8.55922401e-01 6.21596456e-01 5.93227088e-01
-6.65220976e-01 -2.39240631e-01 3.30661416e-01 -3.90723825e-01
3.63582045e-01 -3.48826461e-02 -4.27831799e-01 -1.53145587e+00
-5.55971041e-02 -9.29682732e-01 4.12448347e-01 -6.58214808e-01
-1.59061432e+00 8.28071773e-01 4.93049175e-01 -9.78217423e-01
-8.60121608e-01 -3.62474799e-01 -5.88935792e-01 6.93659842e-01
-1.18924129e+00 -1.68791556e+00 -3.30449432e-01 3.74492973e-01
8.55709314e-01 -5.61629951e-01 9.94180739e-01 -1.33715659e-01
-7.09488213e-01 8.39307487e-01 -2.99541622e-01 3.23635221e-01
5.48420370e-01 -1.41377449e+00 8.21453929e-01 7.89447188e-01
5.68643630e-01 7.82389402e-01 5.20475209e-01 -8.24584901e-01
-1.27503824e+00 -1.22117805e+00 1.12484145e+00 -8.00547421e-01
5.61081886e-01 -6.10724449e-01 -9.62334514e-01 1.26813245e+00
3.04196984e-01 7.59535143e-03 9.72023845e-01 7.96886206e-01
-6.13504827e-01 -1.71722025e-01 -6.78968549e-01 7.51435518e-01
1.42920005e+00 -5.20256341e-01 -1.09999216e+00 4.45934385e-01
8.64158332e-01 -5.26035249e-01 -1.13993704e+00 7.32145131e-01
4.08438861e-01 -6.60491228e-01 1.06944215e+00 -1.05935049e+00
5.86729586e-01 1.00268118e-01 -1.21682085e-01 -1.63117170e+00
-4.70662504e-01 -8.88163626e-01 -9.36937213e-01 1.64106297e+00
7.43161142e-01 -6.08890235e-01 6.56669676e-01 5.21640480e-01
-3.01934808e-01 -1.16867697e+00 -6.93072557e-01 -9.47546780e-01
8.36246535e-02 -3.47724885e-01 1.01381874e+00 1.00485384e+00
9.41115171e-02 1.23067391e+00 -3.52929533e-01 2.01730318e-02
6.69143379e-01 3.33040982e-01 1.08928382e+00 -1.13856876e+00
-5.25382698e-01 -1.53139770e-01 -1.09276183e-01 -9.69223022e-01
3.65418136e-01 -1.02337766e+00 -2.82615907e-02 -1.71055782e+00
4.51005638e-01 -4.23992157e-01 -1.67687103e-01 7.19143212e-01
-8.02302241e-01 -1.37421116e-01 2.61918865e-02 3.27432901e-01
-5.96589506e-01 8.38182330e-01 1.43964601e+00 -2.20313817e-01
-2.94025511e-01 2.71367729e-01 -9.21640277e-01 4.71114576e-01
7.71628380e-01 -5.21956682e-01 -9.68694270e-01 -2.55039573e-01
2.65640944e-01 -3.01909596e-01 2.78782070e-01 -6.29326642e-01
8.38164538e-02 -2.13752165e-01 3.16332161e-01 -6.45830572e-01
3.46524537e-01 -3.71087909e-01 1.66410476e-01 9.31141526e-02
-2.45471686e-01 -2.13643655e-01 -1.66476697e-01 6.14150763e-01
-1.17805339e-01 -8.22980478e-02 2.33318895e-01 -1.91834211e-01
-6.79490030e-01 4.23184544e-01 5.76584488e-02 3.77476603e-01
8.40188682e-01 1.01091787e-01 -5.22731960e-01 -2.26336464e-01
-8.40282202e-01 1.88262194e-01 6.31405115e-02 6.87685370e-01
6.05722129e-01 -1.60455596e+00 -8.62518311e-01 8.59287754e-02
5.53770840e-01 4.66330916e-01 7.98522960e-03 4.49198306e-01
2.67221481e-01 1.27669230e-01 7.47474730e-02 -4.80932802e-01
-8.59610677e-01 9.60554242e-01 3.05229843e-01 -6.90898001e-01
-7.24305332e-01 9.44521785e-01 1.39227614e-01 -3.16185981e-01
-9.59445089e-02 -3.72984797e-01 -1.06223226e-01 3.22088227e-02
1.66934058e-01 1.39458388e-01 1.21038474e-01 -1.88035667e-01
-1.29187077e-01 9.34451893e-02 -3.96367997e-01 -1.41866252e-01
1.32050729e+00 1.38056234e-01 3.86048593e-02 7.63520837e-01
9.47966874e-01 -4.55899030e-01 -1.04374254e+00 -5.75434148e-01
2.88732916e-01 -2.24892274e-01 -4.23013657e-01 -6.49179101e-01
-7.45285630e-01 5.87951243e-01 -2.19149649e-01 2.67749310e-01
9.66274679e-01 4.66865689e-01 8.05496037e-01 6.95079207e-01
4.95312750e-01 -9.51777935e-01 3.40263337e-01 6.09822273e-01
7.94604182e-01 -1.23524415e+00 -1.63655933e-02 -8.90845597e-01
-7.15801001e-01 8.96877050e-01 8.06700528e-01 1.83532983e-01
5.88212132e-01 2.54144371e-01 -6.48412183e-02 -2.38120958e-01
-1.23046708e+00 -4.02176321e-01 5.94630063e-01 1.04967451e+00
5.09983838e-01 1.11384533e-01 -5.49016371e-02 8.50732505e-01
-4.88907605e-01 2.60020439e-02 2.29421154e-01 9.35771108e-01
-1.45154402e-01 -1.50749958e+00 2.27210447e-01 5.17372310e-01
9.58046541e-02 -1.67046621e-01 -5.81314981e-01 8.41968238e-01
2.12702483e-01 1.02577019e+00 -3.14228892e-01 -6.62356377e-01
4.43789274e-01 4.42213088e-01 7.52116263e-01 -1.11689603e+00
-3.51398021e-01 -6.21675290e-02 3.32029134e-01 -1.50245085e-01
-5.95396996e-01 -4.94056284e-01 -8.19126964e-01 -3.92523557e-02
-4.00442243e-01 2.32253298e-01 -7.24300891e-02 1.36914098e+00
3.18066627e-01 9.52535689e-01 5.15724599e-01 -5.41936338e-01
-4.94488209e-01 -1.34210801e+00 -3.29848200e-01 4.24033433e-01
3.38460971e-03 -8.97218883e-01 -5.80732301e-02 2.47409135e-01]
|
[9.476325035095215, 8.504120826721191]
|
8fdedd28-18a1-4bbb-b970-c2387ba71434
|
semantic-based-neural-network-repair
|
2306.07995
| null |
https://arxiv.org/abs/2306.07995v1
|
https://arxiv.org/pdf/2306.07995v1.pdf
|
Semantic-Based Neural Network Repair
|
Recently, neural networks have spread into numerous fields including many safety-critical systems. Neural networks are built (and trained) by programming in frameworks such as TensorFlow and PyTorch. Developers apply a rich set of pre-defined layers to manually program neural networks or to automatically generate them (e.g., through AutoML). Composing neural networks with different layers is error-prone due to the non-trivial constraints that must be satisfied in order to use those layers. In this work, we propose an approach to automatically repair erroneous neural networks. The challenge is in identifying a minimal modification to the network so that it becomes valid. Modifying a layer might have cascading effects on subsequent layers and thus our approach must search recursively to identify a "globally" minimal modification. Our approach is based on an executable semantics of deep learning layers and focuses on four kinds of errors which are common in practice. We evaluate our approach for two usage scenarios, i.e., repairing automatically generated neural networks and manually written ones suffering from common model bugs. The results show that we are able to repair 100% of a set of randomly generated neural networks (which are produced with an existing AI framework testing approach) effectively and efficiently (with an average repair time of 21.08s) and 93.75% of a collection of real neural network bugs (with an average time of 3min 40s).
|
['Jun Sun', 'Richard Schumi']
|
2023-06-12
| null | null | null | null |
['automl']
|
['methodology']
|
[ 2.80604154e-01 3.15378934e-01 2.02073336e-01 -2.42541224e-01
-1.40746564e-01 -7.12283373e-01 6.16409965e-02 4.43540663e-02
-1.10157229e-01 5.82177937e-01 -5.71796298e-01 -8.36190820e-01
6.65627494e-02 -1.09983563e+00 -1.35692763e+00 -1.42595395e-01
-1.36385471e-01 1.50332853e-01 5.70940435e-01 -1.72543123e-01
2.76255012e-01 2.75866240e-01 -1.92727399e+00 4.83359337e-01
1.06872809e+00 6.92838907e-01 -3.63370180e-02 9.49633956e-01
-1.44001022e-01 1.01635146e+00 -1.10162151e+00 -5.44916153e-01
1.41731590e-01 -1.44267336e-01 -1.02262533e+00 -2.95212060e-01
4.65096802e-01 -3.71909320e-01 3.17181170e-01 1.44686091e+00
6.29319996e-02 -2.38679454e-01 2.51468867e-01 -1.53426421e+00
-3.52062762e-01 1.08026469e+00 -2.92316768e-02 -2.46719681e-02
3.74782383e-02 4.19830471e-01 6.97155714e-01 -4.00510430e-01
3.70823085e-01 1.03316462e+00 8.81402016e-01 9.12342370e-01
-1.08994889e+00 -6.17056966e-01 -1.21917203e-01 -3.32988478e-04
-1.25164640e+00 -3.98858309e-01 4.74266648e-01 -7.71417856e-01
1.66985285e+00 3.87568414e-01 2.90775627e-01 1.18737960e+00
4.68779713e-01 1.85341984e-01 4.95685786e-01 -4.92247254e-01
5.51217318e-01 6.02454059e-02 3.35270077e-01 8.94957960e-01
5.47340155e-01 1.84096083e-01 9.89372060e-02 -2.64867008e-01
5.44226050e-01 -2.88455427e-01 -2.75022406e-02 1.88991323e-01
-8.37026715e-01 6.45962894e-01 1.56140119e-01 3.93767416e-01
-2.01287657e-01 6.78165913e-01 6.47021174e-01 3.29475075e-01
2.30907761e-02 8.39537501e-01 -7.08254218e-01 -2.75693715e-01
-8.25367153e-01 2.85421997e-01 1.07027805e+00 9.21087682e-01
7.22031713e-01 4.78299916e-01 6.00421391e-02 6.01670444e-01
3.95294242e-02 3.50415073e-02 5.82643688e-01 -8.07046711e-01
3.45305204e-01 9.14030254e-01 -6.04471676e-02 -9.28316891e-01
-3.05654049e-01 -3.20866615e-01 -6.97319448e-01 5.33591926e-01
3.93719256e-01 -5.64196408e-01 -8.67167711e-01 1.72416019e+00
1.47743955e-01 4.83272493e-01 1.12253718e-01 4.73351181e-01
5.17338455e-01 5.78530729e-01 -2.13229060e-01 1.94565520e-01
9.34878707e-01 -1.00010443e+00 -2.47257128e-01 -5.15789390e-01
9.56061244e-01 -2.94165641e-01 1.10632443e+00 6.84627712e-01
-1.27918053e+00 -4.95659202e-01 -1.37057114e+00 3.95019382e-01
-3.48043829e-01 8.43860209e-02 4.95881945e-01 6.74977779e-01
-1.37600732e+00 9.48894680e-01 -8.74768138e-01 -2.26900317e-02
1.50891647e-01 4.79093403e-01 -1.27585515e-01 1.70979738e-01
-1.04949212e+00 7.48011529e-01 8.74057949e-01 1.37330800e-01
-1.13108170e+00 -7.82991946e-01 -9.03009236e-01 2.72421241e-01
4.06318635e-01 -5.57563365e-01 1.61434555e+00 -1.36329710e+00
-1.22002912e+00 4.27820176e-01 3.43892366e-01 -7.73525655e-01
3.40488285e-01 -1.16461404e-01 -4.13899422e-01 -1.91219941e-01
-1.66218534e-01 5.44513404e-01 7.21548617e-01 -9.99865353e-01
-6.40731275e-01 1.52089208e-01 4.91835624e-01 -6.74070358e-01
-5.04606903e-01 3.18879813e-01 -1.88755125e-01 -5.00363708e-01
-4.44693148e-01 -1.04209793e+00 -1.31419182e-01 -2.28117153e-01
-6.47879004e-01 -1.03214934e-01 3.23203266e-01 -7.15501487e-01
1.36746669e+00 -1.82629561e+00 1.59789786e-01 3.02537560e-01
2.14073181e-01 6.48656607e-01 -1.95226058e-01 1.01570360e-01
-4.27968830e-01 5.72542787e-01 -5.25559962e-01 5.76573573e-02
3.70824970e-02 3.19577456e-01 -3.46203715e-01 -1.56890862e-02
5.39801657e-01 6.30326152e-01 -8.51027131e-01 -1.05728865e-01
-1.13149710e-01 3.27875286e-01 -8.65164876e-01 2.93218821e-01
-8.21946383e-01 -1.99334145e-01 -6.17769659e-02 6.46997094e-01
4.54390794e-01 -1.38165345e-02 1.40632555e-01 1.61745951e-01
-6.17096536e-02 3.98548722e-01 -1.05771136e+00 1.22774196e+00
-4.40783679e-01 5.57691574e-01 -2.43027940e-01 -8.36530805e-01
8.76812100e-01 4.20081139e-01 -2.84686446e-01 -4.04483676e-01
2.56279111e-01 3.39476258e-01 2.88544357e-01 -6.61898315e-01
2.77284473e-01 2.67828822e-01 -1.71350986e-01 5.91990829e-01
1.10476881e-01 2.51115680e-01 4.30848330e-01 -1.32638156e-01
1.79218423e+00 -6.87511498e-03 1.90487146e-01 -9.32480991e-02
4.73295718e-01 5.82433231e-02 6.72242761e-01 9.32129741e-01
2.25849524e-01 3.77014935e-01 1.12031889e+00 -7.18286872e-01
-1.16062713e+00 -5.79217255e-01 2.92789340e-01 7.45376706e-01
-5.75577259e-01 -5.17671406e-01 -1.40435231e+00 -8.52425575e-01
-3.53304803e-01 1.05744910e+00 -6.23032391e-01 -7.38276362e-01
-6.24604940e-01 -2.75634557e-01 1.19381607e+00 6.74391210e-01
3.37941617e-01 -1.56184375e+00 -1.02571988e+00 3.16303611e-01
2.78899968e-01 -7.86853790e-01 -5.33617884e-02 2.85947531e-01
-7.95412421e-01 -1.25904739e+00 -1.79922637e-02 -8.21766913e-01
9.28008795e-01 -3.72453988e-01 1.33615005e+00 8.24049592e-01
-2.24048406e-01 -1.27149731e-01 -2.72486150e-01 -1.79667562e-01
-1.10073149e+00 2.03063279e-01 -6.94184750e-02 -3.13568085e-01
1.44981593e-01 -6.17931545e-01 8.62478316e-02 5.95831871e-02
-1.30610120e+00 2.02153906e-01 4.80354249e-01 7.47407377e-01
1.22140132e-01 5.68945885e-01 3.75705779e-01 -1.05065799e+00
9.08033013e-01 -6.72232866e-01 -1.19479525e+00 4.28632051e-01
-5.99958301e-01 2.68816918e-01 1.00367224e+00 -6.09123111e-01
-7.37619877e-01 1.07293278e-01 -3.63468170e-01 -6.51809275e-01
-4.34456795e-01 8.56922805e-01 5.71436342e-03 -7.47550279e-02
1.11280477e+00 -9.08334255e-02 -1.28966853e-01 -2.09321260e-01
-1.46150053e-01 4.32728618e-01 6.10164940e-01 -8.58850420e-01
7.68948853e-01 -4.58607793e-01 -2.97269315e-01 -2.94574559e-01
-3.47344398e-01 3.75471234e-01 -2.51082718e-01 -1.34600893e-01
4.67579484e-01 -3.95194441e-01 -8.63437533e-01 8.20650220e-01
-1.58258355e+00 -8.61746013e-01 1.21550765e-02 -2.70841748e-01
-8.84538591e-02 2.50699632e-02 -6.55676603e-01 -6.45576477e-01
-5.92764974e-01 -1.46310103e+00 6.63465321e-01 9.50552821e-02
-4.37980115e-01 -8.50586772e-01 6.28763288e-02 -2.77916968e-01
5.49798369e-01 4.50400800e-01 1.15256953e+00 -6.74600840e-01
-5.08116126e-01 -1.83693543e-01 -4.24475735e-03 8.94722283e-01
-1.67467311e-01 5.79756737e-01 -9.07292783e-01 -9.40943807e-02
-1.29724368e-01 -1.52446136e-01 3.59955430e-01 -7.12271258e-02
1.36075974e+00 -8.66618872e-01 -7.02205598e-02 3.84985119e-01
1.39014053e+00 4.58419949e-01 8.54489744e-01 2.87340075e-01
7.61704266e-01 5.13803959e-01 1.78805530e-01 2.32537612e-01
1.38653204e-01 4.09162670e-01 9.28800225e-01 3.77075255e-01
3.54256898e-01 3.25827822e-02 6.63126528e-01 3.07338983e-01
1.22339249e-01 -1.61551490e-01 -1.51338136e+00 5.87697327e-01
-1.83083892e+00 -7.40504324e-01 -1.93531945e-01 2.10135150e+00
1.27205062e+00 5.22029757e-01 -1.82630777e-01 4.12271440e-01
6.03019893e-01 -4.77200091e-01 -5.22154927e-01 -9.30352211e-01
5.69600701e-01 6.32474422e-01 2.09283471e-01 5.08095145e-01
-9.21525776e-01 9.44713295e-01 5.93302822e+00 3.96404326e-01
-1.34019983e+00 -1.11650325e-01 3.33621591e-01 7.69557757e-03
-3.12741399e-01 3.60736959e-02 -6.74567342e-01 5.76320350e-01
1.49501419e+00 -9.61393416e-02 6.64118350e-01 1.29245532e+00
-2.82374650e-01 2.84261070e-02 -1.36544788e+00 3.85845780e-01
1.66051313e-02 -1.46894467e+00 -1.67859048e-01 -2.67467648e-01
6.34988070e-01 -6.28540516e-02 -3.18136394e-01 5.71412921e-01
4.87556577e-01 -1.23215652e+00 1.07344830e+00 4.61283833e-01
7.33791828e-01 -1.03180790e+00 8.60486209e-01 5.90076387e-01
-6.15764797e-01 -1.39690027e-01 -2.62217224e-01 -3.60568047e-01
-4.29975986e-01 8.45344126e-01 -1.23746002e+00 -1.98941529e-02
1.01228821e+00 4.37867314e-01 -8.95868480e-01 1.07015908e+00
-5.35014987e-01 7.28895187e-01 -1.40049785e-01 -1.01124696e-01
5.74261844e-02 3.58391702e-01 2.83971786e-01 1.15820622e+00
4.26336467e-01 -4.34729040e-01 -2.83335030e-01 1.37012339e+00
-4.89149578e-02 -5.46037614e-01 -6.33653224e-01 -1.72475368e-01
5.31457067e-01 1.13303411e+00 -5.15913010e-01 -3.61045539e-01
-1.63148075e-01 7.85913467e-01 4.62352604e-01 2.15563774e-01
-1.33891833e+00 -6.66380107e-01 7.54229903e-01 7.91940168e-02
2.83258706e-01 3.58638633e-03 -3.82886261e-01 -8.97722900e-01
3.81738335e-01 -1.13165510e+00 -1.15518436e-01 -8.70352089e-01
-9.37268138e-01 1.00822055e+00 -6.11695163e-02 -7.75211930e-01
-6.10884428e-01 -5.85279465e-01 -9.94452775e-01 7.54008055e-01
-1.06768930e+00 -6.37564301e-01 -3.23317379e-01 2.48629749e-01
1.67680457e-01 -1.80785090e-01 9.47402298e-01 4.47132856e-01
-9.35921550e-01 9.14577127e-01 -6.38248146e-01 1.15625881e-01
2.30035976e-01 -1.19146132e+00 9.10102010e-01 1.37922180e+00
-3.06847036e-01 9.46767867e-01 7.09566474e-01 -7.63916612e-01
-1.16808903e+00 -1.59014034e+00 9.91510928e-01 -1.52613193e-01
7.62378633e-01 -4.46128994e-01 -1.21688259e+00 9.95961130e-01
1.53145105e-01 -6.69350699e-02 1.80865169e-01 -1.86642990e-01
-6.23652279e-01 6.88903686e-03 -1.21057987e+00 7.34246492e-01
8.91372442e-01 -3.62898082e-01 -4.09733355e-01 2.66366869e-01
1.01938272e+00 -7.11171627e-01 -6.81181252e-01 4.30218339e-01
2.42136240e-01 -1.21749210e+00 6.73243284e-01 -6.18523955e-01
8.98767591e-01 -4.50639069e-01 2.17216194e-01 -1.33261085e+00
6.82895631e-02 -6.10702217e-01 -3.26867074e-01 1.29482293e+00
6.38955355e-01 -8.16789746e-01 4.89046335e-01 8.84403765e-01
-5.57876885e-01 -7.87625611e-01 -5.89482903e-01 -6.46380246e-01
-8.91139507e-02 -7.69280195e-01 8.89005959e-01 9.55798805e-01
-1.94291294e-01 9.00371447e-02 -1.08981557e-01 4.27169263e-01
8.74066949e-02 -4.32821542e-01 5.69242120e-01 -1.26340151e+00
-6.66176736e-01 -5.00098050e-01 -3.47092152e-01 -1.72306895e-01
5.13696373e-01 -8.05426598e-01 3.31986755e-01 -1.01799250e+00
-1.83818817e-01 -5.40772498e-01 -1.30205393e-01 1.33419824e+00
1.24734074e-01 -9.74511728e-02 -2.61815190e-01 -2.01902390e-01
-2.54674971e-01 -2.00518683e-01 3.76150012e-01 -2.14143246e-01
-7.20393509e-02 -8.33136290e-02 -6.93602860e-01 8.93591464e-01
9.91779625e-01 -6.67148292e-01 -4.23835695e-01 -6.63499832e-01
9.63014960e-01 2.84009743e-02 7.04311669e-01 -1.42931437e+00
2.83572555e-01 -1.74042851e-01 -1.36043385e-01 -1.59876779e-01
-3.24294835e-01 -8.27327847e-01 5.79708397e-01 6.65515304e-01
-3.50909323e-01 3.19224179e-01 5.81062317e-01 -5.10411076e-02
-1.49481043e-01 -9.18298960e-01 5.82469404e-01 -1.59290954e-01
-6.68921113e-01 -1.14036128e-01 -5.03363609e-01 -1.17729351e-01
1.17672968e+00 4.49398421e-02 -6.07389629e-01 1.67730406e-01
-4.70337242e-01 5.30792288e-02 5.50802112e-01 6.49626315e-01
6.41079545e-01 -9.53391731e-01 -2.57762998e-01 4.64944750e-01
8.64576176e-02 2.89652258e-01 2.91594653e-03 4.56977844e-01
-8.99186552e-01 1.78421274e-01 -3.28012973e-01 -4.40685123e-01
-1.16709840e+00 5.60494006e-01 8.01469684e-01 -1.83093965e-01
-3.78637731e-01 8.74836504e-01 -1.94030359e-01 -7.14973629e-01
3.76562268e-01 -9.62214053e-01 3.92099656e-03 -4.36190069e-01
6.81190789e-01 1.22931711e-01 5.74250877e-01 6.83491156e-02
-3.81648272e-01 5.76384366e-02 -8.22977629e-03 2.67194778e-01
1.33925641e+00 1.01993942e+00 -6.05473936e-01 1.87958777e-01
9.08248365e-01 -3.81092072e-01 -1.02225220e+00 1.98967949e-01
1.73293784e-01 1.27785638e-01 -7.25168586e-02 -1.02946532e+00
-1.31391144e+00 7.97166526e-01 2.39606053e-01 4.08410698e-01
1.16043663e+00 -3.08589995e-01 6.70990467e-01 7.12154746e-01
4.76079851e-01 -6.92239761e-01 -5.77405542e-02 8.57005775e-01
8.87671411e-01 -7.94638515e-01 -6.50443554e-01 -1.79759607e-01
-1.57897606e-01 1.22905207e+00 1.05678248e+00 -3.25602084e-01
2.47272566e-01 8.29639971e-01 -4.33452249e-01 -6.76684603e-02
-1.08952165e+00 3.68477821e-01 -2.36813594e-02 4.55362171e-01
2.63476193e-01 -4.62603197e-02 5.97256422e-02 9.24251616e-01
-3.19673866e-01 2.60500342e-01 9.14604723e-01 1.10559797e+00
-3.34766805e-01 -1.14616406e+00 -3.93818915e-01 4.33520526e-01
-4.03222352e-01 -3.25821042e-01 -3.91306013e-01 6.06894493e-01
4.74639028e-01 7.26661921e-01 5.32414168e-02 -8.24718177e-01
3.97486031e-01 2.40904629e-01 3.65120083e-01 -9.11139369e-01
-1.16354501e+00 -5.39707899e-01 4.13530916e-01 -7.08609104e-01
1.89266726e-01 -3.54438752e-01 -1.44228899e+00 -3.69733572e-01
-3.92820388e-02 -1.93832174e-01 6.88201249e-01 9.05381978e-01
4.57414657e-01 1.03090715e+00 1.71518028e-01 -6.27003074e-01
-5.20892978e-01 -6.94624722e-01 1.90474726e-02 3.17677334e-02
2.44370043e-01 -3.49237263e-01 -4.18013990e-01 2.81278908e-01]
|
[7.488786220550537, 7.674733638763428]
|
a44e4f60-574b-4f65-b750-ff47039915c4
|
vipr-visual-odometry-aided-pose-regression
|
1912.08263
| null |
https://arxiv.org/abs/1912.08263v3
|
https://arxiv.org/pdf/1912.08263v3.pdf
|
ViPR: Visual-Odometry-aided Pose Regression for 6DoF Camera Localization
|
Visual Odometry (VO) accumulates a positional drift in long-term robot navigation tasks. Although Convolutional Neural Networks (CNNs) improve VO in various aspects, VO still suffers from moving obstacles, discontinuous observation of features, and poor textures or visual information. While recent approaches estimate a 6DoF pose either directly from (a series of) images or by merging depth maps with optical flow (OF), research that combines absolute pose regression with OF is limited. We propose ViPR, a novel modular architecture for long-term 6DoF VO that leverages temporal information and synergies between absolute pose estimates (from PoseNet-like modules) and relative pose estimates (from FlowNet-based modules) by combining both through recurrent layers. Experiments on known datasets and on our own Industry dataset show that our modular design outperforms state of the art in long-term navigation tasks.
|
['Christoffer Löffler', 'Felix Ott', 'Christopher Mutschler', 'Tobias Feigl']
|
2019-12-17
| null | null | null | null |
['camera-localization']
|
['computer-vision']
|
[-3.97202045e-01 -1.65977776e-01 -3.56061459e-01 -3.03919375e-01
-1.81071028e-01 -5.10105133e-01 6.55669332e-01 -2.01798484e-01
-5.52072763e-01 7.31321454e-01 3.21334630e-01 -6.68828860e-02
3.74648571e-02 -7.86838591e-01 -9.12711799e-01 -2.34999686e-01
-2.44341746e-01 6.66032195e-01 3.89905810e-01 -7.76560009e-01
5.28608441e-01 5.03174841e-01 -1.47196209e+00 -2.96969771e-01
7.85218060e-01 1.06868613e+00 2.93556094e-01 8.82252216e-01
-1.34142682e-01 1.13692546e+00 -3.20105463e-01 2.26125166e-01
5.00012219e-01 -8.63086805e-02 -8.47399354e-01 -1.65369928e-01
1.24083722e+00 -5.69892347e-01 -8.57292235e-01 8.71158481e-01
5.12127101e-01 2.09535077e-01 3.05918723e-01 -1.20870566e+00
-4.90020156e-01 -1.31454572e-01 -2.98154324e-01 1.49169162e-01
6.49694681e-01 5.78324080e-01 7.76930392e-01 -7.45134413e-01
1.27872109e+00 1.38513684e+00 1.14643991e+00 4.68782932e-01
-1.21761179e+00 -4.48308736e-01 2.96028018e-01 2.35116817e-02
-1.07373238e+00 -4.54009742e-01 5.41021287e-01 -5.64021647e-01
1.50533283e+00 -5.40061891e-01 9.30947423e-01 1.26163793e+00
8.20131540e-01 5.42277277e-01 5.12994111e-01 2.68709660e-01
8.28979090e-02 -5.52362680e-01 -2.49121547e-01 1.19203544e+00
3.20867509e-01 4.21995431e-01 -1.08420169e+00 5.32785535e-01
1.15590227e+00 3.58298570e-02 -1.50333270e-01 -1.26836801e+00
-1.53376222e+00 7.15780556e-01 1.12431324e+00 -2.23391682e-01
-2.87300497e-01 9.84270334e-01 2.06139639e-01 5.93681395e-01
1.37194395e-01 6.42097056e-01 -5.28569221e-01 -4.23772782e-01
-7.41728246e-01 4.77708161e-01 8.43258619e-01 1.17312324e+00
1.29420793e+00 3.21530402e-01 2.89429605e-01 3.73994648e-01
5.01572967e-01 7.05198467e-01 5.57313025e-01 -1.38284647e+00
6.97365046e-01 4.73996401e-01 3.43959033e-01 -1.22610676e+00
-8.27184796e-01 -5.39596319e-01 -5.05432069e-01 6.71938300e-01
6.69239819e-01 -1.81627139e-01 -1.29545641e+00 1.78788471e+00
3.45405340e-02 5.36892563e-02 -1.31239057e-01 1.16689825e+00
7.89349198e-01 2.20403895e-01 -3.29851359e-01 5.77437520e-01
8.45919847e-01 -1.40197766e+00 -6.46266162e-01 -9.86026943e-01
7.19209552e-01 -3.25735450e-01 5.15053809e-01 3.07372570e-01
-6.93686247e-01 -6.88922405e-01 -1.61720979e+00 -8.23486865e-01
-6.16714418e-01 -2.14929014e-01 7.97466040e-01 2.91220158e-01
-1.49436390e+00 6.72801018e-01 -1.09432209e+00 -6.27758622e-01
3.10752183e-01 4.53909218e-01 -7.70761967e-01 -1.65289238e-01
-1.01033163e+00 1.14523900e+00 -7.52131119e-02 4.37561065e-01
-1.27460504e+00 -6.28263354e-01 -1.60651052e+00 -5.66420734e-01
2.05733299e-01 -1.30576968e+00 1.11192870e+00 -4.13024247e-01
-1.69536495e+00 5.17781675e-01 -3.70101422e-01 -7.63875008e-01
9.20611024e-01 -7.95409441e-01 1.76928252e-01 -6.46365434e-03
2.34308958e-01 1.35215664e+00 7.82193124e-01 -1.10965562e+00
-7.32799888e-01 -5.11964858e-01 3.70437622e-01 4.23924506e-01
3.45473975e-01 -1.22708833e+00 -4.75435883e-01 8.94026682e-02
7.91668117e-01 -1.01389611e+00 -7.65721738e-01 5.28381407e-01
-1.85840905e-01 3.28366160e-01 8.64750385e-01 -4.19865578e-01
6.40074253e-01 -1.65332341e+00 3.44002128e-01 -2.24286467e-01
4.16926980e-01 -2.82842219e-01 -1.52046144e-01 6.65845275e-02
4.60924149e-01 -3.51566046e-01 1.36175558e-01 -6.89220667e-01
-9.19355750e-02 4.94424343e-01 -2.67073750e-01 7.58362353e-01
2.65333891e-01 1.23245597e+00 -1.47349381e+00 -1.17842294e-01
7.86236942e-01 6.58248186e-01 -8.80259514e-01 -1.51387364e-01
-2.97419518e-01 7.76605368e-01 -1.19177923e-01 7.75154948e-01
5.80401957e-01 -4.70759831e-02 -1.63279846e-01 -1.88489690e-01
-3.39074671e-01 5.04404604e-01 -9.88556087e-01 2.86451840e+00
-7.34579206e-01 1.04582608e+00 -7.64634535e-02 -6.38069093e-01
9.73314881e-01 -2.06085458e-01 5.51770926e-01 -1.09078622e+00
1.95176080e-01 3.66022021e-01 -4.36068475e-01 -3.09858501e-01
1.12281382e+00 2.76830971e-01 -2.18432516e-01 -1.48569480e-01
5.32363653e-01 -4.85050261e-01 7.94211105e-02 -2.51574349e-02
1.28730261e+00 9.53139424e-01 2.21645743e-01 -2.75711007e-02
3.17887396e-01 4.69145834e-01 4.98740792e-01 1.02214313e+00
-7.69774377e-01 8.47576737e-01 2.15653390e-01 -7.64250159e-01
-1.25996387e+00 -1.30496466e+00 3.21150944e-02 5.85109890e-01
8.74920070e-01 -3.31407696e-01 -5.91380782e-02 -4.22477752e-01
5.34257174e-01 -1.15073532e-01 -7.84207702e-01 -7.78507218e-02
-8.97945106e-01 -1.83307350e-01 5.28843403e-01 6.32972240e-01
7.28992939e-01 -6.70894921e-01 -9.12032366e-01 5.96253157e-01
-1.46240026e-01 -1.08401752e+00 -2.70825595e-01 4.56803650e-01
-1.11169636e+00 -9.63687003e-01 -5.76803565e-01 -7.40123689e-01
3.21215749e-01 6.10853910e-01 1.23155653e+00 -4.50638592e-01
-1.46107003e-01 1.71441004e-01 -1.81831513e-02 5.87491728e-02
1.25370041e-01 3.74634206e-01 1.38036326e-01 -6.68247938e-01
1.25964403e-01 -7.01805651e-01 -8.35751891e-01 3.89047474e-01
-3.16210419e-01 -1.63749546e-01 5.67424119e-01 9.82027888e-01
3.85023206e-01 -9.10789967e-01 9.33383331e-02 -3.76722604e-01
1.35157719e-01 -4.98711407e-01 -7.98106551e-01 -4.56506252e-01
-6.43240690e-01 4.97223854e-01 1.52500495e-01 -1.33332938e-01
-8.92797172e-01 3.40358406e-01 -1.51807591e-01 -5.90378046e-01
-6.04845248e-02 2.49593243e-01 2.98782617e-01 -3.94573003e-01
8.82432759e-01 -7.74242058e-02 4.07055885e-01 -2.33826697e-01
6.28485322e-01 4.13242839e-02 1.04205739e+00 -1.66038617e-01
6.69646442e-01 1.09152770e+00 2.59004712e-01 -5.15000880e-01
-6.77113295e-01 -5.63397527e-01 -1.00331807e+00 -1.66673094e-01
8.28955352e-01 -1.41204631e+00 -9.14464474e-01 5.07622421e-01
-1.01894450e+00 -5.12727797e-01 -9.96084735e-02 5.32572150e-01
-8.18212569e-01 1.08952366e-01 -7.28591084e-01 -4.02463883e-01
-1.07272556e-02 -1.19275260e+00 1.33994734e+00 2.98742831e-01
-2.70628601e-01 -1.19575357e+00 4.33809608e-01 9.91154686e-02
5.00506997e-01 4.14721459e-01 -1.23917222e-01 2.84225076e-01
-1.16508389e+00 -1.97116420e-01 -1.33916318e-01 -2.29736716e-01
9.09298845e-03 -3.34409803e-01 -8.62871230e-01 -3.53287339e-01
-4.61235940e-01 -4.91783261e-01 1.37760127e+00 6.34063125e-01
2.06353843e-01 2.53793627e-01 -4.75338787e-01 1.23817194e+00
1.60767722e+00 -1.44003471e-03 7.30207741e-01 9.51741993e-01
1.15248716e+00 4.16862369e-01 6.07168436e-01 2.58177161e-01
8.99666846e-01 6.20145082e-01 1.08697319e+00 1.19439065e-01
-4.05925691e-01 -6.37964904e-01 7.31692910e-01 5.99137664e-01
4.01998200e-02 -3.36337760e-02 -8.77767801e-01 7.58904755e-01
-2.07585979e+00 -6.39014721e-01 1.70088466e-02 1.85190630e+00
3.54959190e-01 4.17359352e-01 -3.81680250e-01 -2.82946497e-01
2.16245130e-01 6.53485417e-01 -7.96164989e-01 -3.64366949e-01
-2.79666483e-01 -1.65720597e-01 9.33857381e-01 8.92084837e-01
-1.18342650e+00 1.24917507e+00 6.48396683e+00 6.30173311e-02
-1.38161814e+00 -1.09836474e-01 -3.39804180e-02 -9.80612636e-02
-6.17272221e-02 1.77150205e-01 -1.00142169e+00 -1.12442426e-01
6.45002186e-01 3.69584501e-01 2.55330682e-01 9.47132289e-01
-2.71622520e-02 -4.46716100e-01 -1.10699666e+00 9.82895911e-01
1.17711797e-01 -1.57689321e+00 -1.37943015e-01 2.47781739e-01
9.85127449e-01 8.39664936e-01 1.46601021e-01 3.51098508e-01
7.74654746e-01 -1.02410722e+00 1.06895113e+00 6.08608425e-01
5.99052191e-01 -4.45177227e-01 8.18736434e-01 1.26716167e-01
-1.53428137e+00 -2.99011588e-01 -2.98353285e-01 -6.14833117e-01
4.71175075e-01 3.57073039e-01 -8.28059316e-01 7.23507285e-01
7.02029228e-01 1.59262276e+00 -6.54564738e-01 1.11123776e+00
-2.60807455e-01 -4.01628882e-01 -3.17159325e-01 2.16538042e-01
7.83146918e-01 -2.22349782e-02 6.33871078e-01 7.75257885e-01
2.28704944e-01 -9.47384775e-01 1.83947876e-01 7.64728785e-01
2.43813232e-01 -5.35834193e-01 -1.17871118e+00 4.01789248e-01
2.58709162e-01 9.80118394e-01 -4.38015133e-01 -1.13505930e-01
-4.20825750e-01 1.15822136e+00 5.12367129e-01 3.90659988e-01
-4.89072561e-01 -4.54409152e-01 1.42501187e+00 -2.51802444e-01
4.31262642e-01 -1.08471751e+00 -6.38668656e-01 -1.21650279e+00
5.90886106e-04 1.58198085e-02 2.35696249e-02 -9.19203699e-01
-8.11353505e-01 4.23296392e-01 -5.91414452e-01 -1.67593062e+00
-5.78478456e-01 -7.82003641e-01 -3.52932550e-02 7.18366504e-01
-1.95763052e+00 -8.83064508e-01 -7.50604510e-01 3.36405873e-01
5.74991882e-01 3.37454230e-01 5.91293752e-01 2.81659216e-01
-1.18188009e-01 1.73102468e-01 -9.07439142e-02 2.22392589e-01
8.42180252e-01 -1.32651150e+00 1.03311932e+00 7.70026267e-01
1.18938744e-01 5.89448988e-01 8.04563344e-01 -6.71663821e-01
-1.68848789e+00 -9.06212926e-01 8.40190887e-01 -9.74355221e-01
5.02762258e-01 -3.78087968e-01 -3.15922290e-01 8.58966708e-01
-1.18397415e-01 4.26624894e-01 -4.42735285e-01 1.86962057e-02
-5.29109836e-01 -8.19773450e-02 -8.64196241e-01 7.57278860e-01
1.76911128e+00 -6.48612261e-01 -3.88785958e-01 -2.11008832e-01
7.71569967e-01 -1.06208789e+00 -6.13461256e-01 4.02424574e-01
9.93885338e-01 -1.18428206e+00 1.29303133e+00 -3.75527292e-01
3.93469423e-01 -6.68620884e-01 -6.87641576e-02 -1.40685236e+00
-1.41320035e-01 -5.01381755e-01 -2.96953201e-01 3.38796824e-01
1.84580922e-01 -5.68578959e-01 1.13554335e+00 -1.86988842e-02
-3.71060938e-01 -3.93276483e-01 -1.07453024e+00 -9.14412856e-01
-1.76418170e-01 -5.38538039e-01 3.26652169e-01 6.21962309e-01
-1.03941709e-01 3.62698913e-01 -6.62880063e-01 2.37407297e-01
4.77354676e-01 -1.15425929e-01 1.24112713e+00 -1.17252231e+00
3.26610833e-01 -3.30766678e-01 -1.12993681e+00 -1.93154585e+00
8.92213508e-02 -5.99785566e-01 6.94084048e-01 -1.77793837e+00
-6.02367222e-01 -2.01936156e-01 -8.55868980e-02 2.81483591e-01
2.27734059e-01 4.47637439e-01 4.54211095e-03 9.67453420e-02
-8.48421574e-01 7.47659206e-01 1.54912221e+00 -1.30724370e-01
-4.63895380e-01 -6.24361277e-01 -1.10082068e-01 5.73414922e-01
3.36133301e-01 -2.16583058e-01 -3.16021800e-01 -9.12028432e-01
5.11444628e-01 1.78691268e-01 5.75079679e-01 -1.64284039e+00
6.10298276e-01 1.26296222e-01 6.12270534e-01 -8.20001006e-01
5.97671628e-01 -5.19697189e-01 -1.38010168e-02 8.46552849e-01
7.05012977e-02 3.74441057e-01 5.87838590e-02 1.01414728e+00
-3.79901260e-01 4.44071889e-01 3.88107896e-01 -3.14357787e-01
-1.58555996e+00 5.40101111e-01 -3.23359609e-01 1.39256045e-02
3.37558866e-01 -6.65159047e-01 -6.35337949e-01 -5.84208488e-01
-5.45069814e-01 5.97655773e-01 8.21723700e-01 8.32893491e-01
6.45773053e-01 -1.33876288e+00 -6.53802454e-02 3.79652172e-01
3.64132136e-01 5.89981854e-01 1.38500273e-01 1.13931000e+00
-1.21597254e+00 6.93973422e-01 -6.94624126e-01 -1.18965328e+00
-3.52497667e-01 1.60661727e-01 5.91700912e-01 -2.30614051e-01
-9.03923154e-01 1.01581264e+00 6.19868152e-02 -9.38821197e-01
2.09676102e-01 -5.58386087e-01 -1.67216286e-01 8.57948512e-02
1.72880620e-01 3.10427785e-01 -1.27011696e-02 -7.98511684e-01
-4.18841958e-01 8.54715824e-01 2.59256065e-01 -4.24008846e-01
1.10631859e+00 -8.25960040e-01 3.06037605e-01 6.26600504e-01
1.34829867e+00 -3.73848647e-01 -2.25392890e+00 -2.94334114e-01
8.33776668e-02 -5.35452902e-01 -2.91900523e-02 -2.09202528e-01
-1.01006269e+00 9.61510956e-01 6.04790270e-01 -5.45358002e-01
4.04891282e-01 -4.56500202e-01 9.49885726e-01 8.85063231e-01
8.61927450e-01 -9.71773803e-01 4.50183898e-01 1.39615345e+00
7.35389233e-01 -1.50782073e+00 -4.52961773e-02 -1.19283073e-01
-3.32338959e-01 1.30695581e+00 8.53089929e-01 -5.41501999e-01
7.56242454e-01 1.81695580e-01 5.24267673e-01 -1.81644768e-01
-7.50982344e-01 -4.79083657e-01 -2.52161529e-02 9.12007451e-01
-3.27099934e-02 -4.62706566e-01 3.62003982e-01 -1.58413649e-01
-4.78419930e-01 2.97077578e-02 4.81077075e-01 1.38893020e+00
-5.57915151e-01 -6.96576715e-01 -2.94670165e-02 -3.04001607e-02
2.29051381e-01 4.42502722e-02 -5.87352291e-02 9.05700505e-01
2.42825866e-01 6.79357231e-01 4.15966302e-01 -6.49130702e-01
1.97584465e-01 -2.04367489e-01 7.15361774e-01 -3.12174618e-01
-2.73176342e-01 -4.10663724e-01 3.02394360e-01 -1.49640930e+00
-4.78644162e-01 -5.27663589e-01 -1.18576825e+00 -3.85849088e-01
8.50361362e-02 -5.88493526e-01 9.04414654e-01 9.84726489e-01
3.81049991e-01 6.58694386e-01 6.13944978e-02 -1.61213851e+00
-4.80255298e-03 -7.94500172e-01 -3.90616387e-01 1.70239717e-01
1.11791182e+00 -1.19502115e+00 -3.59077722e-01 -1.39870927e-01]
|
[8.038272857666016, -2.1512744426727295]
|
a4a4e986-2b52-4bb1-acce-2b99b11b6b0c
|
lite-light-field-transparency-estimation-for
|
1910.00721
| null |
https://arxiv.org/abs/1910.00721v4
|
https://arxiv.org/pdf/1910.00721v4.pdf
|
LIT: Light-field Inference of Transparency for Refractive Object Localization
|
Translucency is prevalent in everyday scenes. As such, perception of transparent objects is essential for robots to perform manipulation. Compared with texture-rich or texture-less Lambertian objects, transparency induces significant uncertainty on object appearances. Ambiguity can be due to changes in lighting, viewpoint, and backgrounds, each of which brings challenges to existing object pose estimation algorithms. In this work, we propose LIT, a two-stage method for transparent object pose estimation using light-field sensing and photorealistic rendering. LIT employs multiple filters specific to light-field imagery in deep networks to capture transparent material properties, with robust depth and pose estimators based on generative sampling. Along with the LIT algorithm, we introduce the light-field transparent object dataset ProLIT for the tasks of recognition, localization and pose estimation. With respect to this ProLIT dataset, we demonstrate that LIT can outperform both state-of-the-art end-to-end pose estimation methods and a generative pose estimator on transparent objects.
|
['Zheming Zhou', 'Odest Chadwicke Jenkins', 'Xiaotong Chen']
|
2019-10-02
| null | null | null | null |
['transparent-objects']
|
['computer-vision']
|
[ 4.50816154e-01 -5.03819957e-02 4.25613731e-01 -6.22836649e-01
-4.41100031e-01 -6.27292454e-01 3.78986388e-01 -4.96296912e-01
-2.09953710e-01 3.60202968e-01 -2.50017382e-02 2.89409190e-01
9.02316496e-02 -5.66887736e-01 -1.12513375e+00 -5.25722086e-01
3.24851900e-01 7.65284479e-01 2.42178306e-01 1.06829971e-01
2.02796772e-01 6.69108689e-01 -1.63863850e+00 3.98910224e-01
4.53497857e-01 1.37423611e+00 5.40666699e-01 6.19449914e-01
3.88408080e-02 6.46913707e-01 -3.72841179e-01 -4.11093622e-01
5.60862958e-01 3.86272609e-01 -4.03474331e-01 3.31156015e-01
1.41877401e+00 -8.79117310e-01 -3.98321211e-01 8.09116542e-01
4.15204525e-01 -2.92122997e-02 8.61787081e-01 -8.40664268e-01
-1.02709055e+00 2.55042493e-01 -6.27646446e-01 -1.93525374e-01
4.91098464e-01 3.41925532e-01 6.35765433e-01 -1.26586688e+00
6.88866973e-01 1.66358662e+00 5.15287757e-01 6.61836088e-01
-1.16939652e+00 -5.04190326e-01 3.58712375e-01 -2.26631299e-01
-9.97514486e-01 -6.61055326e-01 6.39046073e-01 -6.50220335e-01
7.71214426e-01 2.15875462e-01 7.68753529e-01 1.19171202e+00
5.07115006e-01 8.40267599e-01 1.23305023e+00 -1.68564066e-01
1.89428836e-01 5.64175546e-02 -3.55755597e-01 8.47235501e-01
3.49414855e-01 2.58243620e-01 -1.01484191e+00 -1.26335341e-02
1.13986707e+00 -2.11220160e-02 -3.81209373e-01 -9.85150635e-01
-1.23764932e+00 1.94573388e-01 6.61431968e-01 -5.55365860e-01
-1.91358119e-01 5.14162481e-01 -2.83657700e-01 -9.09487158e-02
6.83102727e-01 4.82527256e-01 -3.76668066e-01 -9.09317937e-03
-3.20887119e-01 1.37647260e-02 7.37093508e-01 1.20699573e+00
7.10238039e-01 8.31767172e-02 -2.14548767e-01 6.43337607e-01
8.35853279e-01 9.18772042e-01 -1.30457148e-01 -1.09625030e+00
3.45837623e-01 4.65480506e-01 3.94817680e-01 -6.93209648e-01
-3.84833038e-01 -4.47834462e-01 -2.23266929e-01 4.97169942e-01
5.12359381e-01 3.16117764e-01 -1.07465839e+00 1.41871381e+00
6.39388978e-01 -2.13887542e-01 -1.80102959e-01 1.25358593e+00
1.23586857e+00 1.61016330e-01 -5.00538766e-01 3.46235335e-01
1.38827670e+00 -9.00181234e-01 -4.19002503e-01 -6.93794906e-01
-1.37672931e-01 -9.96452570e-01 1.32293081e+00 7.27183878e-01
-1.11235976e+00 -2.19745696e-01 -7.18396962e-01 -4.82903898e-01
8.85861441e-02 5.10609984e-01 1.01323128e+00 6.92024589e-01
-6.91913188e-01 1.12993203e-01 -9.41704273e-01 -2.16776133e-01
6.53789103e-01 3.35911453e-01 -2.13686407e-01 -2.11436182e-01
-3.21024805e-01 8.56635094e-01 -2.43693665e-02 4.98347968e-01
-1.19293714e+00 -9.06162977e-01 -7.18038738e-01 -3.62759322e-01
4.19323534e-01 -9.69980836e-01 1.26624560e+00 -5.08783996e-01
-1.96961117e+00 1.05882049e+00 -7.25480765e-02 -1.00904927e-02
8.29873800e-01 -7.17739165e-01 1.89898804e-01 2.49428630e-01
-1.52013019e-01 6.93978786e-01 1.06173265e+00 -1.57310069e+00
-1.38778165e-01 -7.73425341e-01 1.02088638e-01 3.88840288e-01
-1.47229642e-01 -2.71036327e-02 -5.64903617e-01 -1.89343750e-01
5.45154512e-01 -1.14845240e+00 1.63819343e-01 9.06254411e-01
-5.84028840e-01 1.98502064e-01 9.05876100e-01 -3.36705714e-01
8.12224597e-02 -1.93658841e+00 -1.33071083e-03 -9.60548595e-02
4.75993305e-01 -1.77910820e-01 5.11483615e-03 7.57216942e-03
3.87241244e-01 -6.51802719e-01 1.74690053e-01 -6.57942832e-01
1.50147170e-01 -1.02288179e-01 -3.69402677e-01 9.25515950e-01
2.83914264e-02 8.22726250e-01 -6.50486946e-01 -1.20793179e-01
4.08952743e-01 8.32110941e-01 -4.95569438e-01 1.37824103e-01
-6.48736537e-01 8.02022338e-01 -3.95663351e-01 1.01624084e+00
9.67217922e-01 -1.00293338e-01 -2.48079255e-01 -4.65438306e-01
-1.45655170e-01 3.23596478e-01 -8.18668962e-01 1.94689035e+00
-6.72001123e-01 8.52016509e-01 3.00914735e-01 4.18018363e-03
9.39825773e-01 -1.46096662e-01 3.24416101e-01 -5.04648328e-01
4.04470921e-01 2.44269133e-01 -3.44465792e-01 -5.67993104e-01
4.24692065e-01 2.54894018e-01 3.85457754e-01 2.60654777e-01
-2.42896050e-01 -8.50054204e-01 -1.63516551e-01 -1.57670915e-01
8.87923121e-01 7.23873019e-01 -3.90477926e-01 2.56019062e-04
-2.99917966e-01 -4.37806308e-01 2.54118621e-01 5.86745560e-01
2.19163463e-01 9.65069890e-01 -2.43173167e-01 -7.27223516e-01
-7.08537638e-01 -1.58014238e+00 -4.19458359e-01 9.48867619e-01
4.36882764e-01 1.12732314e-01 -5.23501873e-01 -1.32223248e-01
4.11842078e-01 3.51553738e-01 -5.75179517e-01 -8.59156847e-02
-4.43601549e-01 -4.95334059e-01 -1.02088317e-01 3.05377871e-01
6.50411606e-01 -8.29976618e-01 -1.21881390e+00 -5.07981405e-02
-6.09609522e-02 -1.50430489e+00 -3.57397407e-01 9.94098410e-02
-6.59050643e-01 -8.83267522e-01 -5.66826642e-01 -3.23082805e-01
8.08504164e-01 6.10500932e-01 1.42047179e+00 -5.89328706e-01
-7.41737425e-01 7.12919116e-01 -1.61602572e-02 -7.88023233e-01
-6.72624633e-02 -2.04781488e-01 9.61748809e-02 1.75485626e-01
4.67807949e-02 -5.14101446e-01 -1.00565088e+00 4.95711803e-01
-6.66634381e-01 1.83071092e-01 7.49370158e-01 2.08876088e-01
4.66576487e-01 -6.51151299e-01 -4.62920129e-01 -5.88213801e-01
1.65887922e-01 8.84605795e-02 -9.78191853e-01 2.12702230e-01
-7.06163421e-02 -2.62487661e-02 -7.95801580e-02 -7.67899334e-01
-1.19851363e+00 4.39039469e-01 5.24399757e-01 -7.69867241e-01
9.43262652e-02 -5.06336764e-02 -9.59091559e-02 -7.18676925e-01
9.48562145e-01 -5.14596067e-02 -1.55271068e-01 -4.18989718e-01
1.85910180e-01 5.92435181e-01 4.97880071e-01 -6.28271878e-01
8.87681246e-01 1.12741542e+00 1.12996995e-01 -9.30717170e-01
-1.08470905e+00 -2.38294095e-01 -5.06784976e-01 -5.06000757e-01
7.27746010e-01 -1.07879674e+00 -1.30315042e+00 6.85845554e-01
-1.26605666e+00 -4.88063246e-01 -1.24151155e-01 7.22032905e-01
-7.18498230e-01 4.94609773e-02 -5.28022587e-01 -1.04068661e+00
-1.12521484e-01 -1.21017349e+00 1.84713912e+00 1.52575240e-01
4.89305444e-02 -5.02065837e-01 -2.01723471e-01 7.13596523e-01
3.09437126e-01 1.24696687e-01 1.99670568e-01 5.72853982e-01
-1.42570090e+00 4.79821078e-02 -3.66886437e-01 -1.26591874e-02
1.50377095e-01 1.16957881e-01 -1.45084810e+00 -3.17070305e-01
-6.60697892e-02 -7.32146144e-01 9.25259054e-01 5.84108531e-01
1.17403901e+00 -1.86666653e-01 -2.25135386e-01 1.08953679e+00
1.18253374e+00 -1.37597010e-01 4.61052746e-01 2.48428345e-01
1.01760983e+00 6.44039333e-01 6.41235530e-01 5.64863384e-01
4.38969970e-01 9.27004516e-01 9.52919006e-01 -4.92770076e-02
-3.19519728e-01 -7.47276917e-02 3.65375549e-01 3.59110422e-02
-1.02173761e-01 -1.71719849e-01 -7.74811566e-01 -3.88976708e-02
-1.33997297e+00 -4.95209575e-01 -1.80563867e-01 2.31257629e+00
7.98208714e-01 -1.83989909e-02 -2.08619744e-01 -6.36642396e-01
1.69506922e-01 -9.66323614e-02 -8.97399724e-01 3.47502151e-04
-1.62525043e-01 -2.61864036e-01 7.45106995e-01 3.56060565e-01
-8.58481288e-01 9.51581895e-01 5.87299109e+00 3.31198335e-01
-1.35946357e+00 -1.00689217e-01 1.96398303e-01 -4.27482605e-01
-5.12968957e-01 -2.88101017e-01 -1.01374173e+00 1.33228689e-01
7.04643968e-03 6.56160712e-01 6.67624474e-01 7.77889729e-01
2.71784281e-03 -4.51236993e-01 -1.41015685e+00 1.21815789e+00
2.50359088e-01 -7.98158407e-01 -1.87827796e-01 9.65480134e-02
8.25001240e-01 4.34803814e-01 7.22616673e-01 -4.31421340e-01
4.47545797e-01 -8.52044106e-01 1.41482127e+00 5.90937793e-01
7.73973346e-01 -9.17806178e-02 1.39046106e-02 2.15721935e-01
-9.51106906e-01 -1.31951571e-01 -7.03027248e-01 1.48689792e-01
4.09077890e-02 8.82833481e-01 -1.03156924e+00 6.90999255e-02
1.02586865e+00 4.77260917e-01 -4.03201103e-01 1.18844211e+00
-2.84103423e-01 1.26463160e-01 -6.29574120e-01 -9.20229778e-02
-2.39900723e-01 -2.07970113e-01 6.36861086e-01 6.72678888e-01
6.46510199e-02 -3.94331306e-01 2.50844717e-01 1.43749034e+00
-1.44549698e-01 -4.74472433e-01 -4.67804790e-01 1.20568126e-01
2.52977133e-01 1.30277789e+00 -6.63794518e-01 1.57122731e-01
-2.04345420e-01 1.01355672e+00 3.35903436e-01 5.23250341e-01
-7.08574355e-01 2.04833075e-01 5.79582393e-01 2.58664697e-01
1.36398584e-01 -5.96636057e-01 -1.65885478e-01 -1.30559182e+00
5.45169652e-01 -5.29380023e-01 -4.60188836e-01 -1.31887329e+00
-1.16241634e+00 5.08315325e-01 -8.02542176e-03 -1.40106082e+00
2.49658421e-01 -1.08438849e+00 1.39123008e-01 8.13242674e-01
-1.69902921e+00 -1.51241672e+00 -7.82265663e-01 3.03746939e-01
5.23788810e-01 1.81270301e-01 5.85463643e-01 -4.95652109e-02
-2.09230013e-04 2.38213524e-01 4.65040915e-02 -1.82663471e-01
8.23219359e-01 -9.26672697e-01 4.53867167e-01 5.29078364e-01
2.55710572e-01 8.02138746e-01 8.19040716e-01 -5.49338996e-01
-2.02710986e+00 -9.17808354e-01 -1.58886448e-01 -9.57026541e-01
2.28241578e-01 -1.01980507e+00 -3.53468955e-01 7.02212512e-01
-2.66900808e-01 4.72278893e-01 3.65200453e-02 -7.81274438e-02
-6.97724760e-01 -2.57381141e-01 -1.13929617e+00 7.75868475e-01
1.34361303e+00 -4.97922838e-01 -1.11289091e-01 7.36944079e-01
5.61033189e-01 -1.19970906e+00 -5.12979031e-01 4.79415357e-01
1.25210202e+00 -1.23589253e+00 1.35679293e+00 -2.13979539e-02
3.05472404e-01 -2.47824520e-01 -1.03091635e-01 -1.15547788e+00
-1.72336951e-01 -6.63818836e-01 -1.71387434e-01 7.56847501e-01
2.43296668e-01 -7.92100072e-01 1.04297543e+00 8.06332886e-01
-3.52480888e-01 -5.25645852e-01 -8.68476510e-01 -8.44160795e-01
-3.88371170e-01 -2.29957774e-01 2.72113144e-01 3.73363882e-01
-7.62571216e-01 1.17683470e-01 -2.24040911e-01 3.70363533e-01
1.05751204e+00 5.40184379e-01 1.05438483e+00 -1.51603115e+00
-3.94086212e-01 -1.40477672e-01 -3.07358325e-01 -1.40333331e+00
1.12589262e-01 -3.65377456e-01 5.10720253e-01 -1.51358438e+00
3.30311656e-01 -5.51061451e-01 5.20442843e-01 2.21519411e-01
1.95616961e-01 6.27128243e-01 8.08733106e-02 2.69595027e-01
-5.18527448e-01 7.89507508e-01 1.77493072e+00 -2.64489025e-01
-1.37379229e-01 1.44341171e-01 -2.97920316e-01 8.77234757e-01
5.83858311e-01 -2.16657713e-01 -4.44037825e-01 -1.20919752e+00
6.47571504e-01 -2.62623966e-01 6.31659627e-01 -1.09751952e+00
5.95347434e-02 -2.85245687e-01 5.81841528e-01 -5.88132322e-01
1.11734045e+00 -1.02742684e+00 9.80789065e-02 2.32775703e-01
-3.07857841e-01 -3.77292305e-01 2.79764961e-02 6.56710505e-01
5.51990032e-01 9.22890529e-02 8.25296223e-01 -3.27881306e-01
-4.20543939e-01 6.54402554e-01 -5.65576181e-02 -7.26106251e-03
7.11143136e-01 -6.70701027e-01 -6.89332128e-01 -2.38465965e-01
-2.75247157e-01 -4.26604375e-02 8.23317885e-01 6.13982081e-01
9.86162543e-01 -9.62599039e-01 -7.00384617e-01 2.24399149e-01
4.77582067e-01 5.80964088e-01 2.76405700e-02 8.63760173e-01
-8.01869333e-01 2.03588679e-01 -9.27884355e-02 -1.01112294e+00
-1.38142943e+00 1.45054072e-01 4.32951599e-01 5.22173584e-01
-6.07181549e-01 1.43674171e+00 8.72483671e-01 -6.49580836e-01
3.97033006e-01 -8.78470421e-01 4.33793098e-01 -6.74145222e-01
2.53487766e-01 1.89646900e-01 1.76243246e-01 -4.06680405e-01
-1.86257243e-01 9.56041694e-01 5.10972477e-02 -2.59011686e-01
1.27135444e+00 -2.99934089e-01 -1.64824650e-01 6.50410950e-01
8.09469640e-01 2.75899231e-01 -1.80824780e+00 -3.38885158e-01
-7.44695842e-01 -9.63266671e-01 1.71737894e-01 -8.62894177e-01
-1.17727172e+00 9.81099963e-01 5.95362961e-01 -3.82302552e-01
6.13158643e-01 2.10075527e-01 5.05873501e-01 7.63293564e-01
8.52386177e-01 -8.78338814e-01 4.78265285e-01 5.37439406e-01
1.28221881e+00 -1.30807054e+00 1.06666625e-01 -1.04777086e+00
-2.16652632e-01 1.06255460e+00 8.82721722e-01 1.15225591e-01
2.85533905e-01 6.37595177e-01 2.13877156e-01 -3.35858494e-01
-4.97658223e-01 2.33064257e-02 5.73974669e-01 7.90361524e-01
2.26063386e-01 -1.67344779e-01 6.25606596e-01 -2.84470886e-01
-3.74429524e-01 -1.78382739e-01 2.30627462e-01 8.39961231e-01
-4.59906667e-01 -4.63119656e-01 -5.97356081e-01 3.08436662e-01
-1.45174459e-01 -2.11545825e-02 -5.86362302e-01 2.96951622e-01
-4.92703132e-02 7.83944845e-01 1.54041201e-01 3.79774310e-02
1.61425143e-01 -7.09907293e-01 1.29426050e+00 -8.74121428e-01
-1.67543828e-01 5.31653464e-02 -3.72124240e-02 -9.22595799e-01
-4.27723229e-01 -4.94967610e-01 -9.46911216e-01 1.18503571e-01
-5.38735390e-01 -6.38079345e-01 1.03787231e+00 6.24113500e-01
2.19727203e-01 4.09629077e-01 4.27057564e-01 -1.56054008e+00
-5.67020297e-01 -9.75402832e-01 -5.94421864e-01 1.04071513e-01
4.14156854e-01 -1.02630377e+00 -3.35130721e-01 -9.38163996e-02]
|
[7.066502571105957, -2.108020782470703]
|
a1445c46-e678-4231-894d-066030b2e1b2
|
batchformer-learning-to-explore-sample
|
2203.01522
| null |
https://arxiv.org/abs/2203.01522v2
|
https://arxiv.org/pdf/2203.01522v2.pdf
|
BatchFormer: Learning to Explore Sample Relationships for Robust Representation Learning
|
Despite the success of deep neural networks, there are still many challenges in deep representation learning due to the data scarcity issues such as data imbalance, unseen distribution, and domain shift. To address the above-mentioned issues, a variety of methods have been devised to explore the sample relationships in a vanilla way (i.e., from the perspectives of either the input or the loss function), failing to explore the internal structure of deep neural networks for learning with sample relationships. Inspired by this, we propose to enable deep neural networks themselves with the ability to learn the sample relationships from each mini-batch. Specifically, we introduce a batch transformer module or BatchFormer, which is then applied into the batch dimension of each mini-batch to implicitly explore sample relationships during training. By doing this, the proposed method enables the collaboration of different samples, e.g., the head-class samples can also contribute to the learning of the tail classes for long-tailed recognition. Furthermore, to mitigate the gap between training and testing, we share the classifier between with or without the BatchFormer during training, which can thus be removed during testing. We perform extensive experiments on over ten datasets and the proposed method achieves significant improvements on different data scarcity applications without any bells and whistles, including the tasks of long-tailed recognition, compositional zero-shot learning, domain generalization, and contrastive learning. Code will be made publicly available at https://github.com/zhihou7/BatchFormer.
|
['DaCheng Tao', 'Baosheng Yu', 'Zhi Hou']
|
2022-03-03
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Hou_BatchFormer_Learning_To_Explore_Sample_Relationships_for_Robust_Representation_Learning_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Hou_BatchFormer_Learning_To_Explore_Sample_Relationships_for_Robust_Representation_Learning_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['compositional-zero-shot-learning']
|
['computer-vision']
|
[ 9.78057384e-02 -1.06713280e-01 -2.93351531e-01 -6.70619309e-01
-4.03368324e-01 -3.05672854e-01 3.88981283e-01 -2.14065779e-02
-2.90818006e-01 7.59740710e-01 9.15322006e-02 -2.15762436e-01
-3.34325999e-01 -7.79190183e-01 -6.03613675e-01 -9.14834440e-01
2.65465707e-01 3.71541172e-01 5.73657639e-02 -1.32450387e-01
-2.65130904e-02 2.60226041e-01 -1.60746109e+00 2.71575242e-01
8.48586023e-01 1.46348405e+00 4.84135933e-02 9.98487473e-02
-3.19346011e-01 6.25217855e-01 -6.52598619e-01 -2.56015629e-01
2.07691327e-01 -3.58922303e-01 -2.96272665e-01 1.31674856e-01
2.26564825e-01 -4.54120487e-01 -4.00865287e-01 9.29829895e-01
7.62546182e-01 3.25768083e-01 6.20579481e-01 -1.38172746e+00
-6.14579618e-01 6.03878021e-01 -6.91661656e-01 2.45928720e-01
-1.10560521e-01 2.05721244e-01 8.61848891e-01 -1.15654039e+00
2.12109506e-01 1.16253209e+00 5.16123414e-01 5.86877286e-01
-1.07411170e+00 -1.25098646e+00 2.29452431e-01 4.03478146e-01
-1.35525143e+00 -6.54723048e-01 9.62700486e-01 -3.90094340e-01
4.37581480e-01 3.69865634e-02 2.08816335e-01 1.30517995e+00
-1.26059353e-01 1.09407198e+00 8.50098848e-01 -1.70152456e-01
3.33434314e-01 2.92361706e-01 3.80130351e-01 1.75483987e-01
1.46586046e-01 -7.12815858e-03 -4.56396610e-01 -3.34284119e-02
4.12731618e-01 3.77456486e-01 -3.29801083e-01 -4.52085018e-01
-7.12378561e-01 8.17553163e-01 4.45727646e-01 1.34121761e-01
-1.72900334e-01 -2.68600106e-01 5.60355008e-01 3.57589215e-01
5.26910484e-01 9.42294076e-02 -6.88610852e-01 1.68280900e-01
-8.70952547e-01 1.38934106e-01 7.54221737e-01 9.22670424e-01
8.06130946e-01 7.45452419e-02 -3.16615492e-01 1.08537865e+00
1.18584715e-01 8.59274417e-02 7.42709577e-01 -2.98744291e-01
6.90253079e-01 4.78213400e-01 -1.27541602e-01 -8.28041136e-01
-4.15981501e-01 -8.93770099e-01 -1.18390548e+00 5.89797972e-03
3.91888738e-01 -2.91556299e-01 -9.28165674e-01 1.76384902e+00
5.51486194e-01 3.04454029e-01 8.76689553e-02 7.21187174e-01
8.83161902e-01 5.20135403e-01 -2.89289355e-02 -1.31723106e-01
1.28788471e+00 -9.03906643e-01 -5.55222273e-01 -2.85242409e-01
6.13804936e-01 -4.14715827e-01 1.08354902e+00 4.28867966e-01
-7.90301800e-01 -6.92673564e-01 -1.01878476e+00 6.81070611e-02
-4.82519835e-01 3.47118191e-02 4.37831730e-01 4.12232280e-01
-3.40707034e-01 5.60296834e-01 -5.41132629e-01 -7.15783015e-02
9.32667851e-01 2.27352232e-01 -1.85977831e-01 -4.54202950e-01
-1.37147057e+00 4.33244914e-01 5.47913313e-01 1.80073932e-01
-6.38573289e-01 -8.48445773e-01 -7.87848830e-01 3.82415384e-01
4.46106851e-01 -3.90729576e-01 1.00660491e+00 -1.19801962e+00
-1.32791066e+00 5.02608299e-01 1.10129073e-01 -3.67220432e-01
5.41427255e-01 -2.29700282e-01 -3.59226435e-01 -3.05252165e-01
-1.70305431e-01 4.37957078e-01 7.78542697e-01 -9.24515009e-01
-6.12818718e-01 -5.62501848e-01 -1.72955096e-01 4.39303517e-02
-7.08151340e-01 -2.28836253e-01 -2.24704847e-01 -7.55214095e-01
6.84155943e-03 -6.73497319e-01 1.27643839e-01 -3.96589935e-03
-3.58379453e-01 -4.88104373e-01 9.11954343e-01 -5.26905835e-01
1.16989410e+00 -2.61754060e+00 -6.92229420e-02 3.85867991e-02
1.48611888e-01 3.64313364e-01 -3.23638022e-01 2.42264837e-01
-2.83390731e-01 -1.82276405e-02 -2.91870624e-01 -4.17477846e-01
3.83401737e-02 3.44042480e-01 -5.03622055e-01 4.26654994e-01
3.95312816e-01 7.34856665e-01 -7.59085059e-01 -8.81164744e-02
-3.60891372e-02 2.96063423e-01 -5.40538430e-01 3.36362422e-01
-1.38616204e-01 5.13906121e-01 -3.78464550e-01 6.14736438e-01
1.09883869e+00 -1.90709069e-01 1.63108096e-01 -9.64505151e-02
1.73861533e-01 2.86895543e-01 -1.35291409e+00 1.45152795e+00
-3.90928507e-01 3.81393820e-01 3.00004520e-02 -1.47368395e+00
1.15346491e+00 1.98529616e-01 2.14799285e-01 -9.38220978e-01
1.25710890e-01 2.84080535e-01 2.48485669e-01 -3.63626510e-01
2.27432057e-01 -3.47097933e-01 2.52262447e-02 3.26356500e-01
1.45580083e-01 4.95313019e-01 3.01928259e-02 -2.11340055e-01
7.58740067e-01 -4.09351513e-02 1.93866834e-01 5.14355116e-03
3.78350854e-01 -4.91472423e-01 9.59559977e-01 6.26180351e-01
-1.99052632e-01 6.23520613e-01 5.79751492e-01 -5.47129512e-01
-8.37127686e-01 -9.76722479e-01 -3.41693163e-01 1.48134542e+00
1.36389405e-01 1.26181515e-02 -2.50736922e-01 -7.73023069e-01
2.06775084e-01 8.40345979e-01 -6.74125195e-01 -5.34424305e-01
-3.15580904e-01 -7.39765823e-01 4.72849369e-01 6.82551384e-01
4.38538641e-01 -9.60722208e-01 -4.19840604e-01 2.54376054e-01
3.26085575e-02 -8.66731346e-01 -3.82806033e-01 6.17503643e-01
-8.15362215e-01 -9.41504419e-01 -7.73435235e-01 -6.00177407e-01
5.51695168e-01 2.19995290e-01 8.91593099e-01 -1.91618297e-02
-1.04752280e-01 -1.27193317e-01 -4.35093343e-01 -4.60198998e-01
2.75125168e-02 2.62997657e-01 8.22532922e-02 2.54737198e-01
5.42580009e-01 -7.62634397e-01 -6.55926645e-01 4.75686193e-01
-1.14671183e+00 -1.31422862e-01 5.54283381e-01 1.31226683e+00
4.11884397e-01 1.75198674e-01 1.01167989e+00 -1.00708497e+00
5.47823429e-01 -1.06226230e+00 -3.49436224e-01 9.59101543e-02
-4.59107816e-01 2.75869220e-02 8.65052640e-01 -7.62525380e-01
-1.01288569e+00 -2.71245539e-01 -1.78132832e-01 -7.02561378e-01
-2.80402690e-01 5.41270852e-01 -5.22961557e-01 4.09941643e-01
5.59113145e-01 3.35926235e-01 2.59387866e-02 -7.48885393e-01
2.05331475e-01 9.00771081e-01 2.57967800e-01 -5.36543310e-01
6.14004314e-01 2.29699373e-01 -3.81693631e-01 -5.73386967e-01
-1.02744627e+00 -3.89550775e-01 -4.36129332e-01 1.57239258e-01
3.55215818e-01 -8.83270979e-01 -3.21337223e-01 6.48858428e-01
-9.09630656e-01 -2.75473297e-01 -4.29328889e-01 3.70988816e-01
-7.39466995e-02 1.36679411e-01 -3.17263842e-01 -7.14057088e-01
-3.11098158e-01 -1.01399457e+00 6.44118726e-01 6.03208721e-01
4.65262216e-03 -9.10715044e-01 -1.99726507e-01 2.15048015e-01
4.60949540e-01 3.28132994e-02 1.05082214e+00 -1.26949692e+00
-1.75539598e-01 -3.09276134e-01 -3.86656910e-01 6.46729231e-01
2.99451351e-01 -3.63335371e-01 -1.24270821e+00 -4.65099782e-01
1.19206697e-01 -6.13375604e-01 1.01000369e+00 2.20411152e-01
1.60772419e+00 -3.46229911e-01 -1.37535691e-01 8.18849444e-01
1.05581713e+00 2.77857393e-01 7.10398674e-01 2.91398883e-01
5.60400188e-01 7.45527923e-01 5.10619998e-01 8.28823447e-01
3.48263472e-01 4.01639730e-01 4.98886049e-01 4.17607054e-02
-5.49823930e-03 -2.19813570e-01 1.13443568e-01 5.63802302e-01
3.96478206e-01 -3.99355263e-01 -8.25500667e-01 4.09204215e-01
-1.69835854e+00 -6.81587815e-01 3.78000706e-01 2.17828417e+00
8.59151185e-01 1.67524770e-01 -3.08100190e-02 1.87761158e-01
6.80570722e-01 3.52649242e-01 -1.14319921e+00 -1.32732734e-01
1.50537612e-02 2.18577564e-01 2.14163333e-01 -9.52919647e-02
-1.00593972e+00 6.57936931e-01 4.55021477e+00 1.06462204e+00
-1.36596596e+00 -1.11365237e-03 8.94589305e-01 -2.46490300e-01
-2.20686287e-01 -1.98048741e-01 -9.53933060e-01 6.47288084e-01
5.73184609e-01 3.84226181e-02 2.98614442e-01 9.26291168e-01
-7.80098885e-02 1.88615099e-01 -1.19250810e+00 9.36928153e-01
-4.55302149e-02 -1.06087577e+00 -3.68685722e-02 -4.85718064e-02
5.97880542e-01 8.33683833e-03 2.09097385e-01 7.74008870e-01
7.67050916e-03 -9.93508339e-01 6.27752483e-01 4.82620299e-01
7.62271047e-01 -8.43386173e-01 8.59568596e-01 7.31830359e-01
-8.86857927e-01 -5.11456728e-01 -5.68569779e-01 -1.60106093e-01
-3.76696765e-01 9.12751734e-01 -7.03872740e-01 6.02609098e-01
5.69200575e-01 7.24032879e-01 -3.76718372e-01 1.07970619e+00
-4.74996082e-02 6.63825691e-01 -1.70956641e-01 8.37758854e-02
3.44626121e-02 -1.20375313e-01 2.59730577e-01 9.67067957e-01
4.23359066e-01 -1.04739875e-01 2.32979536e-01 7.99357176e-01
-3.07527959e-01 -9.35940221e-02 -3.98954332e-01 -9.54424813e-02
5.59663713e-01 1.08592975e+00 -3.15154791e-01 -3.33757967e-01
-4.51513499e-01 6.70016110e-01 3.82014126e-01 5.89301467e-01
-7.59674966e-01 -6.45544052e-01 7.20507801e-01 1.47083163e-01
4.84121561e-01 5.02586626e-02 -3.76243651e-01 -1.15934527e+00
1.90306976e-01 -9.65608180e-01 6.14035428e-01 -2.74015576e-01
-1.71538949e+00 3.93947244e-01 -1.06009446e-01 -1.29352593e+00
-1.27497807e-01 -5.02918363e-01 -8.34106147e-01 9.67562258e-01
-1.69579363e+00 -8.53226244e-01 -3.36013079e-01 5.54791093e-01
5.91472924e-01 -2.66968101e-01 4.64723200e-01 5.79249442e-01
-7.95806885e-01 9.81218338e-01 4.38107789e-01 3.06698948e-01
8.84951293e-01 -9.01781380e-01 9.36265588e-02 5.21961570e-01
-8.14427510e-02 5.62469304e-01 4.84903932e-01 -2.90339351e-01
-1.17016244e+00 -1.11302507e+00 4.78196412e-01 5.56824449e-03
5.79168379e-01 -5.79767227e-01 -1.35929263e+00 4.22235191e-01
-1.67686701e-01 3.62434983e-01 9.17316139e-01 2.75245637e-01
-6.20324671e-01 -6.50515854e-01 -9.86240804e-01 2.38351867e-01
8.93801212e-01 -3.44650865e-01 -6.00817561e-01 1.45259321e-01
6.22018158e-01 -2.69846559e-01 -5.46956897e-01 5.52393377e-01
6.17191494e-01 -7.99341440e-01 8.11237693e-01 -7.38844275e-01
5.26748955e-01 -2.28285100e-02 -2.16866434e-01 -1.39358294e+00
-2.45783657e-01 -1.45686269e-02 -2.16824338e-01 1.53398037e+00
1.44660771e-01 -9.01015401e-01 8.02221417e-01 5.33602834e-01
-1.83062851e-01 -1.09835446e+00 -9.88159001e-01 -7.66693234e-01
2.76075751e-01 -3.88622344e-01 7.71591067e-01 1.02713156e+00
-3.68536383e-01 2.73552954e-01 -5.54574192e-01 2.26845052e-02
4.60610986e-01 2.34062016e-01 8.11217248e-01 -1.30353343e+00
-4.58805919e-01 -3.42429459e-01 -1.35081872e-01 -1.28177035e+00
1.36345223e-01 -9.26616609e-01 5.92222363e-02 -1.21386623e+00
1.70189723e-01 -5.77102304e-01 -7.29700983e-01 5.57922244e-01
-2.30327919e-01 -3.17410454e-02 2.05092698e-01 1.73995838e-01
-4.79343414e-01 9.48156476e-01 1.25619638e+00 -1.89808533e-01
-2.21623987e-01 2.55156130e-01 -9.35175121e-01 6.58608317e-01
8.34285915e-01 -6.27725422e-01 -5.54464281e-01 -4.18018520e-01
5.56092560e-02 -7.91570023e-02 3.49982560e-01 -9.16799724e-01
2.67997533e-01 -1.05464317e-01 6.74539268e-01 -5.26051342e-01
2.34898955e-01 -7.66564012e-01 -2.00572282e-01 3.64634931e-01
-3.96139771e-01 -4.31658745e-01 2.28484318e-01 6.59982264e-01
-3.17850113e-01 -2.48977691e-01 8.62747848e-01 4.25847881e-02
-6.08490884e-01 5.50510526e-01 8.84276181e-02 3.50280225e-01
8.22794676e-01 -1.09853111e-01 -4.06535327e-01 -1.55614734e-01
-5.27992845e-01 6.10319853e-01 9.18143839e-02 5.11868179e-01
5.45547485e-01 -1.34998512e+00 -6.05321586e-01 5.07561147e-01
2.74870753e-01 4.05659378e-01 6.11485839e-01 8.63781214e-01
1.92342967e-01 5.17200679e-02 -2.72587150e-01 -4.99548763e-01
-8.37066054e-01 5.20458221e-01 2.53267050e-01 -4.05779660e-01
-4.14272517e-01 9.88726556e-01 7.05336392e-01 -7.31536269e-01
4.74394917e-01 -2.36115366e-01 -1.66199327e-01 4.91880745e-01
7.36732483e-01 1.80984035e-01 2.07550824e-02 -9.93337184e-02
-4.21894163e-01 1.29563853e-01 -3.91329706e-01 4.31975335e-01
1.49672651e+00 -6.80348575e-02 1.66828051e-01 7.00835109e-01
1.31196427e+00 -3.27430218e-01 -1.52531660e+00 -5.60680628e-01
-1.56092411e-02 -3.70793015e-01 -6.64568320e-02 -6.86907351e-01
-1.18363833e+00 1.20488930e+00 6.61419213e-01 1.42134428e-01
1.26748407e+00 -9.18090940e-02 8.82273436e-01 2.41747692e-01
-1.82073917e-02 -1.13468146e+00 1.49940491e-01 6.73479795e-01
7.39802659e-01 -1.41312599e+00 -2.47615635e-01 1.11834124e-01
-5.03494978e-01 1.18538332e+00 7.00489819e-01 -7.18997344e-02
7.52384961e-01 1.92545399e-01 -1.45071283e-01 1.84556901e-01
-8.92693996e-01 -2.45911721e-02 2.54938990e-01 4.31521505e-01
3.21972996e-01 6.97614551e-02 -1.10155717e-01 1.11355662e+00
2.15644613e-01 1.59568384e-01 1.37903631e-01 7.40150273e-01
-3.39392811e-01 -9.56912816e-01 -3.00316036e-01 6.62760437e-01
-2.02550784e-01 -1.47079527e-01 -5.81995398e-02 6.26334965e-01
3.87555182e-01 6.43599808e-01 1.68110594e-01 -4.94181216e-01
4.19127226e-01 2.56135821e-01 -5.68969995e-02 -6.49376094e-01
-3.06018323e-01 -8.53595659e-02 -1.28086120e-01 -2.87493378e-01
-3.48184071e-02 -4.31249321e-01 -9.91514504e-01 -6.62844777e-02
-3.25377375e-01 8.89562219e-02 4.73265111e-01 9.92589653e-01
4.35593307e-01 7.34019399e-01 8.78779233e-01 -6.96995437e-01
-1.11073327e+00 -1.09465420e+00 -8.27584147e-01 4.27437901e-01
4.62568671e-01 -7.64436781e-01 -5.95377803e-01 -3.37711960e-01]
|
[9.581974983215332, 3.523075819015503]
|
2c903133-e990-4363-95b8-c1c1f648aca0
|
distributional-lesk-effective-knowledge-based
| null | null |
https://aclanthology.org/W17-6931
|
https://aclanthology.org/W17-6931.pdf
|
Distributional Lesk: Effective Knowledge-Based Word Sense Disambiguation
| null |
['Gertjan van Noord', 'Dieke Oele']
|
2017-01-01
| null | null | null |
ws-2017-1
|
['learning-word-embeddings']
|
['methodology']
|
[-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]
|
[-7.394874095916748, 3.791229486465454]
|
4cca679d-acdb-4cc1-80c4-1a4ef93c298a
|
deep-multi-frame-filtering-for-hearing-aids
|
2305.08225
| null |
https://arxiv.org/abs/2305.08225v1
|
https://arxiv.org/pdf/2305.08225v1.pdf
|
Deep Multi-Frame Filtering for Hearing Aids
|
Multi-frame algorithms for single-channel speech enhancement are able to take advantage from short-time correlations within the speech signal. Deep filtering (DF) recently demonstrated its capabilities for low-latency scenarios like hearing aids with its complex multi-frame (MF) filter. Alternatively, the complex filter can be estimated via an MF minimum variance distortionless response (MVDR), or MF Wiener filter (WF). Previous studies have shown that incorporating algorithm domain knowledge using an MVDR filter might be beneficial compared to the direct filter estimation via DF. In this work, we compare the usage of various multi-frame filters such as DF, MF-MVDR, or MF-WF for HAs. We assess different covariance estimation methods for both MF-MVDR and MF-WF and objectively demonstrate an improved performance compared to direct DF estimation, significantly outperforming related work while improving the runtime performance.
|
['Andreas Maier', 'Alberto N. Escalante-B.', 'Tobias Rosenkranz', 'Hendrik Schröter']
|
2023-05-14
| null | null | null | null |
['speech-enhancement']
|
['speech']
|
[ 2.49533176e-01 -2.92991042e-01 3.33499283e-01 -1.19100578e-01
-1.10266674e+00 -3.93292010e-01 5.84938109e-01 -1.09505549e-01
-6.79405808e-01 6.03520274e-01 8.20759714e-01 -4.91156518e-01
-2.64000505e-01 -2.35684097e-01 -4.48404610e-01 -6.57255888e-01
-1.56308904e-01 -3.52253437e-01 3.45601857e-01 -8.40056017e-02
4.76071090e-02 3.81358564e-01 -1.74360073e+00 4.60606962e-01
5.35497665e-01 9.39063191e-01 7.28669345e-01 1.17404747e+00
2.25774541e-01 6.10016823e-01 -1.00549161e+00 -1.42322734e-01
2.02613741e-01 -1.38441145e-01 -3.08799565e-01 -3.36464942e-01
3.60406637e-01 -8.29716027e-01 -4.72027302e-01 9.42460060e-01
1.18823290e+00 4.57121909e-01 2.65260786e-01 -7.71825850e-01
1.53014526e-01 4.44887936e-01 -3.24897289e-01 5.30533016e-01
8.04334223e-01 8.44307989e-02 5.71372926e-01 -1.11521375e+00
4.82627869e-01 1.62655723e+00 8.72349501e-01 3.12582433e-01
-1.19920242e+00 -5.15257657e-01 -9.78418514e-02 4.72753078e-01
-1.31227100e+00 -9.55408931e-01 3.03977847e-01 -2.77935266e-01
1.62902284e+00 1.66131198e-01 3.18960190e-01 9.92328346e-01
2.91429281e-01 9.56456602e-01 1.28764176e+00 -5.04440486e-01
2.27493733e-01 -3.42188984e-01 8.03188384e-02 2.41744950e-01
-1.60944045e-01 6.85325682e-01 -8.68065715e-01 -2.45994881e-01
5.17992318e-01 -6.07737243e-01 -8.68145883e-01 3.48551989e-01
-1.16034484e+00 4.48170543e-01 -2.60636657e-01 2.05319196e-01
-4.17944700e-01 2.40461588e-01 6.12688005e-01 6.78728580e-01
5.74293494e-01 -8.05841833e-02 -5.91356814e-01 -6.93174958e-01
-1.39736271e+00 2.31548116e-01 9.27394092e-01 7.90488839e-01
4.15033042e-01 2.50438303e-01 -4.52074438e-01 8.59779000e-01
5.85142612e-01 6.20438159e-01 3.97314370e-01 -1.10112059e+00
2.92020559e-01 -6.02566123e-01 1.07066631e-01 -5.22796690e-01
-4.83883560e-01 -6.55463934e-01 -8.00686300e-01 4.71589744e-01
5.68991661e-01 -3.83994937e-01 -7.22187638e-01 1.65441787e+00
2.89916098e-01 5.44788122e-01 1.19333193e-01 8.12959611e-01
5.86996317e-01 5.41291535e-01 -1.93804950e-01 -7.22115278e-01
1.13638973e+00 -8.01747501e-01 -1.34666276e+00 -1.00006506e-01
2.96315402e-01 -1.37448955e+00 6.77742541e-01 9.10133004e-01
-1.42355645e+00 -8.80207598e-01 -1.09286165e+00 6.11357158e-04
-4.15364280e-02 1.01559274e-01 2.94053286e-01 1.18667305e+00
-1.42378139e+00 7.57296145e-01 -8.46152544e-01 -2.33827867e-02
1.29186451e-01 5.13976097e-01 -1.02898628e-01 2.36579850e-01
-1.06873047e+00 8.24335217e-01 -3.60613354e-02 -1.55442879e-01
-8.25122833e-01 -9.98808384e-01 -7.05185711e-01 1.36924922e-01
2.25694068e-02 -6.97920620e-01 1.63819659e+00 -7.30348408e-01
-2.02408981e+00 7.81892538e-02 -6.09292805e-01 -8.13299060e-01
5.62111497e-01 -7.81046867e-01 -7.18673766e-01 2.24910200e-01
-5.73095679e-01 1.93098441e-01 1.40034509e+00 -6.20261610e-01
-6.09726906e-01 7.82220811e-02 7.00616539e-02 -3.49086523e-02
-2.42612064e-01 2.51507968e-01 -9.55551639e-02 -1.08969760e+00
-3.77208292e-02 -4.57813710e-01 -4.75534424e-02 -1.89119175e-01
-9.00260657e-02 1.79022372e-01 7.43139446e-01 -1.22662354e+00
1.86010456e+00 -2.41567898e+00 -1.24300122e-01 -1.25674834e-03
-3.52114104e-02 5.95041871e-01 -1.85874179e-01 5.03728628e-01
-2.51436029e-02 -3.09817791e-01 8.26954469e-02 -3.63827616e-01
9.58380997e-02 -3.82018313e-02 -1.92672387e-01 5.75444937e-01
-6.20560236e-02 2.91650951e-01 -7.58453369e-01 -1.13959268e-01
4.98195499e-01 9.64702845e-01 -7.73245871e-01 2.14785621e-01
2.13020310e-01 1.77674085e-01 2.48068571e-01 1.38756230e-01
1.09212279e+00 2.06638977e-01 1.86628371e-01 -6.99072361e-01
-2.49408409e-01 4.48000938e-01 -1.65749967e+00 1.71499825e+00
-1.07089901e+00 1.09498966e+00 5.66638768e-01 -3.29547852e-01
6.03180468e-01 6.74579620e-01 1.51970223e-01 -5.01417518e-01
-8.58472884e-02 3.58839750e-01 2.09623218e-01 -2.32618108e-01
5.84501147e-01 -6.67722821e-02 7.37777948e-01 3.40312958e-01
3.26150894e-01 1.01168148e-01 8.10229108e-02 9.22528729e-02
1.30189288e+00 1.39713502e-02 3.75125408e-01 -4.23856139e-01
8.52470279e-01 -9.64742303e-01 1.13879643e-01 7.85966933e-01
-2.27989331e-01 6.43749654e-01 -1.54622704e-01 2.77664214e-01
-7.43026495e-01 -1.41809070e+00 -3.34516376e-01 8.09201896e-01
-3.29889536e-01 -6.76044285e-01 -8.58544707e-01 -1.36089042e-01
-2.44741127e-01 6.04706645e-01 3.29692550e-02 1.05726197e-01
-5.50395370e-01 -4.51078266e-01 5.63733041e-01 5.64241827e-01
4.62227374e-01 -3.63859296e-01 -5.37696481e-01 7.68926263e-01
-1.36386111e-01 -1.16284966e+00 -6.90563679e-01 9.31164324e-02
-7.59411037e-01 -5.27369559e-01 -9.23243999e-01 -3.23393703e-01
-7.09589720e-02 1.82751358e-01 9.49693918e-01 -2.03840390e-01
-8.24223012e-02 5.15525162e-01 -4.01445836e-01 -2.24088103e-01
-4.85834211e-01 -5.15566945e-01 2.40972653e-01 2.84816623e-02
2.12872773e-01 -6.62134290e-01 -8.54954243e-01 2.66510874e-01
-7.87223935e-01 -1.40481487e-01 4.62882012e-01 9.38560545e-01
3.41584116e-01 -3.11056455e-03 7.12735236e-01 -3.13041121e-01
9.85596716e-01 -1.24583051e-01 -5.40714443e-01 1.73820946e-02
-5.12784481e-01 1.95508339e-02 2.64814198e-01 -6.93510771e-01
-1.47121799e+00 -1.49151608e-01 -8.27063441e-01 -3.12928617e-01
5.12065031e-02 3.85025233e-01 -1.60669923e-01 -4.04499024e-02
5.74845076e-01 4.41474169e-02 -4.83161490e-03 -7.44120657e-01
3.75789285e-01 1.13825071e+00 4.13228601e-01 -3.48047823e-01
5.45649350e-01 3.86245608e-01 -1.20771162e-01 -1.03421342e+00
-2.88280219e-01 -6.73047304e-01 -2.60448188e-01 -2.89625973e-01
6.17013574e-01 -1.28056443e+00 -7.04474747e-01 5.61188519e-01
-1.36801851e+00 -3.26417685e-01 -1.91968605e-01 9.66395199e-01
-6.17570162e-01 6.73905015e-01 -7.98168659e-01 -1.17669213e+00
-4.64729100e-01 -1.16710508e+00 1.08544970e+00 -1.15939388e-02
-3.38782549e-01 -1.04518616e+00 -1.33235872e-01 -8.08370858e-02
8.39433193e-01 -1.01690374e-01 3.15956146e-01 -2.04504520e-01
-1.90526754e-01 1.48965240e-01 -1.41724553e-02 7.28173256e-01
8.83618817e-02 -7.67881125e-02 -1.42751014e+00 -6.58184707e-01
3.44926089e-01 3.73527706e-01 6.46174014e-01 9.24935043e-01
7.78171480e-01 -1.35052189e-01 1.43080233e-02 5.90474963e-01
1.38431478e+00 2.52317071e-01 8.13118339e-01 2.43845671e-01
-7.01528490e-02 1.17729656e-01 7.68088579e-01 8.35625529e-01
-1.20786250e-01 8.79723549e-01 -3.00855339e-01 3.29419188e-02
-7.46655881e-01 1.39185920e-01 8.17501724e-01 9.89526689e-01
-9.61275548e-02 -4.42820281e-01 -5.83616316e-01 3.66087079e-01
-1.57680607e+00 -1.16748738e+00 -2.59076238e-01 2.34137583e+00
6.27158225e-01 1.59943014e-01 1.01006746e-01 6.54766560e-01
6.34307683e-01 1.10855974e-01 -2.57845312e-01 -6.27004385e-01
-2.81954378e-01 8.22898626e-01 4.45046574e-01 8.88475657e-01
-9.05992389e-01 4.01532233e-01 6.92189693e+00 1.19463599e+00
-7.81450152e-01 5.18352091e-01 7.77584836e-02 -2.81560361e-01
-6.99592009e-02 -2.78439820e-01 -6.89502478e-01 2.86979049e-01
1.52821529e+00 -2.42746487e-01 7.21642673e-01 4.68093604e-01
8.52132022e-01 -2.51928717e-01 -8.71869624e-01 1.12278390e+00
-3.79822969e-01 -1.13216293e+00 -4.03466672e-01 2.11793363e-01
4.00999367e-01 7.55050406e-02 4.34581675e-02 1.27714783e-01
-5.63214011e-02 -5.95335186e-01 8.71518195e-01 6.65498257e-01
9.19177771e-01 -7.92147815e-01 6.36587024e-01 8.21968243e-02
-1.51960659e+00 -2.10557152e-02 -1.92839965e-01 -1.53743431e-01
4.19149548e-01 1.07684243e+00 -7.03537285e-01 6.62149072e-01
6.63153946e-01 5.29649198e-01 7.87169952e-03 1.21923196e+00
-1.82986632e-02 6.84027076e-01 -3.59835416e-01 1.43685520e-01
-2.16580033e-01 2.25578144e-01 9.05857563e-01 1.67310452e+00
5.50638378e-01 2.15269420e-02 -2.78479308e-01 2.34907821e-01
4.98210490e-01 -5.97121492e-02 1.01267286e-02 1.79974496e-01
4.79419053e-01 9.82598901e-01 -1.96586221e-01 -3.57569784e-01
-7.16366291e-01 9.32555318e-01 -4.23752844e-01 5.19400239e-01
-7.38112807e-01 -5.10790408e-01 1.22635555e+00 -1.54210599e-02
5.23762047e-01 -5.09649515e-01 -1.21372985e-02 -8.40800107e-01
-1.91220090e-01 -1.14159310e+00 1.74730912e-01 -9.08030987e-01
-8.23200941e-01 6.53804302e-01 -1.28388986e-01 -1.39084423e+00
-5.26967108e-01 -6.32734597e-01 -2.00753585e-01 1.38564026e+00
-1.63849819e+00 -4.58736360e-01 4.27298322e-02 7.22939610e-01
8.97793889e-01 -1.32406473e-01 8.13972592e-01 6.66770935e-01
-7.96507895e-02 7.74760902e-01 4.58145618e-01 -4.95763242e-01
6.87960207e-01 -1.14522648e+00 3.39778602e-01 1.09542799e+00
-1.00839838e-01 7.41745770e-01 1.00326777e+00 -4.81224865e-01
-1.54114485e+00 -6.44381642e-01 7.36565351e-01 2.77778395e-02
6.06730819e-01 -3.05861205e-01 -7.66512215e-01 6.97316751e-02
4.96493727e-01 -2.90800571e-01 7.28477001e-01 -1.60160977e-02
-8.88239443e-02 -7.66615495e-02 -1.24556243e+00 3.75717640e-01
1.02591872e+00 -8.76893878e-01 -3.85004997e-01 2.29667369e-02
6.75948262e-01 -5.96701741e-01 -1.11560333e+00 4.67022419e-01
7.66277969e-01 -1.26699877e+00 1.18813026e+00 1.21007547e-01
-8.30999538e-02 -5.34726560e-01 -5.08486629e-01 -1.37510931e+00
-1.96561158e-01 -1.19247806e+00 -7.63738453e-01 1.26892614e+00
2.46839210e-01 -4.85560030e-01 3.60988677e-01 2.90534105e-02
-2.93220907e-01 -2.85598218e-01 -1.08005798e+00 -9.96098340e-01
-4.14363742e-01 -9.64064777e-01 4.43338752e-01 1.46451890e-01
-2.55875260e-01 7.47345164e-02 -3.87874275e-01 5.32370090e-01
5.43421328e-01 -4.33444411e-01 5.81557572e-01 -8.18076372e-01
-9.53471780e-01 -2.95325905e-01 -3.97701532e-01 -1.49829710e+00
-6.01212047e-02 -3.64070356e-01 -3.01082898e-03 -1.38948655e+00
-4.85724509e-01 3.75383526e-01 -2.35593140e-01 -2.89517909e-01
-1.71107903e-01 -1.34646207e-01 1.92428380e-01 -3.03398818e-01
-7.80524760e-02 2.93919981e-01 1.08835351e+00 2.93648578e-02
-1.50788188e-01 2.36569896e-01 1.79198328e-02 6.34701669e-01
4.64763463e-01 -2.58432478e-01 -4.34159398e-01 -3.85947585e-01
-1.14154622e-01 3.97451997e-01 1.38828322e-01 -1.35678136e+00
4.11444396e-01 5.50462842e-01 2.38425419e-01 -6.53206229e-01
6.43354177e-01 -6.65592670e-01 3.55776012e-01 5.84014952e-01
-3.85854729e-02 4.71527465e-02 5.95053732e-01 6.34827018e-01
-3.92844528e-01 -1.09568879e-01 7.56442010e-01 2.31099918e-01
-7.39041805e-01 -3.13397974e-01 -1.08586884e+00 -2.40330353e-01
3.65778029e-01 -3.44972640e-01 -2.45574191e-01 -7.10874617e-01
-1.03943706e+00 -4.30675358e-01 -1.33385286e-01 2.29275882e-01
9.03972924e-01 -9.71133590e-01 -6.38906062e-01 2.19356135e-01
-5.96133292e-01 -6.66729391e-01 5.19957304e-01 1.06058896e+00
-3.52820396e-01 3.21356088e-01 1.99765846e-01 -5.77793896e-01
-1.57959664e+00 4.33711559e-01 4.36659306e-01 -2.43505761e-01
-3.92589420e-01 9.83432353e-01 -8.49696621e-02 2.48123214e-01
4.67652351e-01 -4.66791600e-01 8.51785913e-02 9.20692310e-02
1.16553974e+00 8.39323401e-01 5.17773449e-01 -4.34398830e-01
-3.89386177e-01 4.10445958e-01 2.16144964e-01 -7.54076064e-01
1.19523573e+00 -5.17902374e-01 2.71271944e-01 1.73986465e-01
1.27258861e+00 3.34172279e-01 -1.29503906e+00 -2.29830146e-01
9.95748118e-03 -6.55344248e-01 7.93124735e-01 -9.68102157e-01
-8.12372804e-01 1.04660559e+00 1.41061890e+00 3.26512158e-02
1.80587578e+00 -6.34695709e-01 8.63149285e-01 2.71972269e-02
5.33772409e-01 -1.10164917e+00 -5.27782068e-02 4.89119887e-01
8.67732108e-01 -6.84599221e-01 -2.71859169e-02 -5.10417461e-01
-2.41613928e-02 1.55068302e+00 -9.59338397e-02 2.49832511e-01
1.19061935e+00 9.98853147e-01 1.49378210e-01 4.02933985e-01
-7.68316150e-01 -6.61887288e-01 4.07353520e-01 8.09091568e-01
7.03032076e-01 -2.08856374e-01 -4.73437130e-01 4.29167420e-01
-1.60422832e-01 5.50085939e-02 4.22406912e-01 8.20250273e-01
-4.98607516e-01 -1.29757690e+00 -5.24211586e-01 3.44246238e-01
-5.88794708e-01 -5.49292088e-01 3.04336488e-01 3.99834841e-01
-6.09080456e-02 1.67755759e+00 -2.49018185e-02 -5.90094388e-01
4.87780869e-01 -1.13943614e-01 5.64313948e-01 -1.83408082e-01
-1.05321348e+00 7.61590719e-01 4.05233890e-01 -7.03853786e-01
-6.17811978e-01 -1.01181519e+00 -9.33213055e-01 -3.67444843e-01
-6.65008664e-01 -2.33136132e-01 8.59636784e-01 9.12043035e-01
2.76169926e-01 8.71478200e-01 4.93441880e-01 -9.08301055e-01
-5.00161946e-01 -1.01245427e+00 -4.33309376e-01 2.72309990e-03
7.28549719e-01 -4.79061365e-01 -6.50453985e-01 -1.00142248e-01]
|
[15.125001907348633, 5.8861083984375]
|
2e4ebafc-e30d-4377-9d44-9e3d9522fd0e
|
dense-network-expansion-for-class-incremental
|
2303.12696
| null |
https://arxiv.org/abs/2303.12696v1
|
https://arxiv.org/pdf/2303.12696v1.pdf
|
Dense Network Expansion for Class Incremental Learning
|
The problem of class incremental learning (CIL) is considered. State-of-the-art approaches use a dynamic architecture based on network expansion (NE), in which a task expert is added per task. While effective from a computational standpoint, these methods lead to models that grow quickly with the number of tasks. A new NE method, dense network expansion (DNE), is proposed to achieve a better trade-off between accuracy and model complexity. This is accomplished by the introduction of dense connections between the intermediate layers of the task expert networks, that enable the transfer of knowledge from old to new tasks via feature sharing and reusing. This sharing is implemented with a cross-task attention mechanism, based on a new task attention block (TAB), that fuses information across tasks. Unlike traditional attention mechanisms, TAB operates at the level of the feature mixing and is decoupled with spatial attentions. This is shown more effective than a joint spatial-and-task attention for CIL. The proposed DNE approach can strictly maintain the feature space of old classes while growing the network and feature scale at a much slower rate than previous methods. In result, it outperforms the previous SOTA methods by a margin of 4\% in terms of accuracy, with similar or even smaller model scale.
|
['Nuno Vasconcelos', 'Dashan Gao', 'Jiancheng Lyu', 'Yunsheng Li', 'Zhiyuan Hu']
|
2023-03-22
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Hu_Dense_Network_Expansion_for_Class_Incremental_Learning_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Hu_Dense_Network_Expansion_for_Class_Incremental_Learning_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['class-incremental-learning']
|
['computer-vision']
|
[ 1.42816737e-01 2.57662028e-01 1.12546295e-01 -1.15621567e-01
-1.86978236e-01 -3.25407177e-01 6.23313904e-01 1.86533213e-01
-7.19898164e-01 6.86487079e-01 -1.47711754e-01 4.18632962e-02
-3.21235567e-01 -6.09414101e-01 -6.73739254e-01 -6.04960084e-01
1.52616724e-01 4.64790553e-01 8.68417084e-01 -8.83095562e-02
2.37524644e-01 6.01339638e-01 -1.72233975e+00 3.63173157e-01
1.12316895e+00 1.02566659e+00 5.53730369e-01 2.21149072e-01
-3.73084366e-01 7.85093069e-01 -5.30283272e-01 -4.13068503e-01
3.16721141e-01 -6.99003637e-02 -8.66885900e-01 -5.23025729e-02
4.40399915e-01 -2.17975844e-02 -1.66249618e-01 8.46590519e-01
5.34636259e-01 4.27693993e-01 4.40691501e-01 -1.27306497e+00
-6.57863915e-01 5.71347713e-01 -5.77652454e-01 1.63168386e-01
-1.25904948e-01 -4.60848659e-02 7.39287674e-01 -1.08050787e+00
4.94824499e-01 1.10044050e+00 8.20658565e-01 5.02618790e-01
-1.31583595e+00 -7.99288988e-01 5.85494339e-01 3.14063489e-01
-1.39369261e+00 -3.15883845e-01 9.29153621e-01 -4.57667410e-01
1.07122231e+00 8.79662484e-02 7.28969276e-01 6.79568231e-01
2.67325163e-01 5.98260105e-01 9.05672669e-01 -5.47659516e-01
2.17752010e-01 5.44009387e-01 2.90041834e-01 4.84614432e-01
3.49220514e-01 -1.53391778e-01 -7.04727888e-01 1.73288196e-01
9.27411318e-01 2.20563501e-01 -2.22553581e-01 -7.30887890e-01
-9.79131103e-01 6.27348959e-01 8.11256111e-01 6.35267138e-01
-5.74186385e-01 7.69570544e-02 4.64085460e-01 5.48655331e-01
8.06949556e-01 7.55372524e-01 -4.74175781e-01 2.73326159e-01
-9.72782671e-01 8.57182369e-02 5.49471736e-01 7.04400063e-01
8.20407927e-01 1.00073077e-01 -3.96602184e-01 8.28678012e-01
3.03071514e-02 -8.58565234e-03 6.55498862e-01 -8.87957156e-01
3.82226318e-01 1.08196390e+00 -1.09843165e-01 -9.47214127e-01
-5.43630183e-01 -1.07001770e+00 -7.29682744e-01 4.84624922e-01
2.54034311e-01 -1.61930665e-01 -9.55419123e-01 1.90263152e+00
3.88527215e-01 8.85999948e-02 -1.15501612e-01 4.11078423e-01
4.70421880e-01 4.28379864e-01 2.27563962e-01 -1.27550170e-01
1.13061953e+00 -1.23149014e+00 -7.40942180e-01 -3.80270749e-01
5.15124857e-01 -4.85390633e-01 9.29961562e-01 3.72358084e-01
-1.21136093e+00 -1.05729306e+00 -9.01366949e-01 -2.25159824e-02
-8.07048678e-01 1.28319621e-01 4.93958801e-01 4.97410119e-01
-1.38147092e+00 5.61834455e-01 -6.05523527e-01 -3.95689338e-01
6.43707633e-01 6.89018428e-01 -4.02822614e-01 6.36387989e-02
-1.01511765e+00 1.10992742e+00 5.78214645e-01 2.17989758e-01
-7.25776613e-01 -8.92023265e-01 -8.40303898e-01 5.07537842e-01
4.98247355e-01 -8.70925009e-01 9.26635206e-01 -1.43911457e+00
-1.49057865e+00 3.29266816e-01 -1.87001705e-01 -4.87525731e-01
4.03191537e-01 -4.38904822e-01 -1.37023985e-01 9.60513484e-03
-1.49000660e-01 9.40144420e-01 1.13336682e+00 -1.35505307e+00
-6.28675044e-01 -3.55372787e-01 2.27410242e-01 3.39344561e-01
-8.49284887e-01 -1.93786174e-01 -4.54854131e-01 -6.29532635e-01
-7.11429343e-02 -8.77944827e-01 -1.84951380e-01 1.68627411e-01
4.53665070e-02 -4.28019106e-01 9.73870695e-01 -5.42883694e-01
1.56433272e+00 -2.25480938e+00 4.07848090e-01 1.62385404e-01
5.44358671e-01 6.62670970e-01 -1.17584236e-01 2.14153722e-01
-3.01468462e-01 -2.04724148e-01 -1.88744202e-01 -7.00674057e-01
-2.69443244e-01 1.71621684e-02 2.22547278e-01 7.22187757e-02
1.51867509e-01 9.67779577e-01 -7.72433937e-01 -2.39880741e-01
6.57681003e-02 4.13621962e-01 -6.58700287e-01 -8.86619613e-02
-4.90474254e-02 2.21026585e-01 -8.00650120e-02 1.29698947e-01
7.07115173e-01 -4.05121922e-01 1.53354108e-01 -1.47060782e-01
-2.20606431e-01 -2.88008843e-02 -1.22484624e+00 1.81133151e+00
-6.85033500e-01 5.22346854e-01 8.34270790e-02 -1.04921103e+00
8.97552848e-01 3.65984678e-01 3.06921810e-01 -5.16007364e-01
2.30914857e-02 1.27894029e-01 3.01344484e-01 -1.29508466e-01
3.93685549e-01 1.43566057e-01 2.17057064e-01 3.06016237e-01
1.99016646e-01 2.38597229e-01 2.53936023e-01 2.05252782e-01
9.49774981e-01 7.65749738e-02 2.49743730e-01 -5.39521575e-01
6.07523739e-01 -1.26643941e-01 5.03174305e-01 7.41628051e-01
-3.41683701e-02 3.19095403e-02 3.20124984e-01 -5.85553586e-01
-9.65733230e-01 -7.84862995e-01 1.05114788e-01 1.38134634e+00
-1.98306106e-02 -2.49850333e-01 -7.42058933e-01 -8.48142385e-01
2.07927570e-01 4.81789917e-01 -8.89336348e-01 -5.04372418e-01
-6.44862354e-01 -4.25857842e-01 1.02017067e-01 7.43101180e-01
8.42842340e-01 -1.22164905e+00 -7.98738718e-01 3.96351397e-01
1.69567451e-01 -8.29455376e-01 -4.27617252e-01 4.99111801e-01
-1.06452096e+00 -9.16064024e-01 -1.01036847e+00 -7.99444854e-01
8.68354142e-01 2.38571122e-01 8.73756707e-01 -9.19811428e-02
-1.97972625e-01 1.60177097e-01 -3.58563393e-01 -4.80981290e-01
-6.80836961e-02 5.85557222e-01 -8.66984501e-02 2.71693170e-01
1.55285269e-01 -6.76048577e-01 -5.82941175e-01 2.17814058e-01
-9.20035839e-01 1.90454960e-01 9.14946795e-01 9.83794987e-01
2.89671183e-01 6.03313409e-02 9.30983603e-01 -1.11600447e+00
6.09032631e-01 -3.61937433e-01 -5.02245426e-01 4.22215044e-01
-7.96345294e-01 1.81083158e-01 7.10638642e-01 -6.42371476e-01
-1.41687882e+00 1.79558635e-01 1.54868439e-01 -5.37120640e-01
9.70930904e-02 3.16912621e-01 8.11745599e-02 -3.30752820e-01
6.98348880e-01 1.69003934e-01 1.30082712e-01 -6.94704413e-01
2.55860060e-01 4.20423359e-01 1.16148695e-01 -1.89685598e-01
5.98307669e-01 3.48602951e-01 -1.42950833e-01 -5.09560585e-01
-8.09968412e-01 -3.73225123e-01 -1.01252329e+00 -1.77725255e-01
7.19518900e-01 -7.82091439e-01 -3.97444457e-01 7.59949088e-01
-1.12894607e+00 -6.01337314e-01 -6.16465688e-01 4.22091216e-01
-1.93843156e-01 4.51270677e-02 -4.62127149e-01 -5.32278180e-01
-2.58099854e-01 -8.97385418e-01 7.17400312e-01 1.23405561e-01
-1.12992756e-01 -1.04285657e+00 -1.56362578e-01 -6.21980708e-03
9.12104547e-01 -1.86025441e-01 9.28090096e-01 -7.02757239e-01
-2.59123474e-01 -4.06772792e-02 -4.79964375e-01 5.50190568e-01
1.94669634e-01 -4.84191298e-01 -1.11041498e+00 -6.19123757e-01
3.79448123e-02 -7.38912597e-02 1.12927902e+00 4.18225139e-01
1.14804912e+00 -8.04380774e-02 -6.25555933e-01 2.73163021e-01
1.28819454e+00 4.63904828e-01 5.56000352e-01 2.15800956e-01
8.09605300e-01 6.53379321e-01 2.66305506e-01 2.22986460e-01
2.48997927e-01 6.57799423e-01 3.51835430e-01 -3.78969759e-01
-6.73361659e-01 2.21485093e-01 1.30542144e-01 6.73177361e-01
-1.17975183e-01 1.39239582e-03 -9.10274625e-01 7.14832485e-01
-1.99065685e+00 -6.89311266e-01 3.68775040e-01 2.22730017e+00
6.62439823e-01 4.11426455e-01 1.83466682e-03 3.10141265e-01
7.71138310e-01 -6.93949759e-02 -7.38896549e-01 -3.73491973e-01
2.79724658e-01 1.33155674e-01 2.40994155e-01 4.95683819e-01
-8.59887302e-01 9.96579230e-01 6.50450468e+00 8.67608428e-01
-1.10383701e+00 4.42160130e-01 4.48567718e-01 -1.81572378e-01
1.33112157e-02 -2.07229882e-01 -9.39276338e-01 3.85091305e-01
7.16325700e-01 -1.20747415e-02 3.00282329e-01 8.40020716e-01
-2.02775419e-01 -1.86680704e-01 -1.05076528e+00 7.84107685e-01
2.64111847e-01 -1.20753896e+00 1.76022947e-01 -1.76118925e-01
7.63710797e-01 -1.33086547e-01 1.69130359e-02 5.19220769e-01
1.05802044e-01 -5.12358904e-01 6.10679805e-01 6.55996501e-01
7.76365161e-01 -8.38918686e-01 9.21504617e-01 4.32850242e-01
-1.55046725e+00 -6.48296237e-01 -2.37625808e-01 -7.80639350e-02
-7.47325495e-02 5.98367095e-01 -4.91531104e-01 5.69545388e-01
8.87561321e-01 5.29319286e-01 -8.92684042e-01 1.16269493e+00
2.21487321e-02 5.65777905e-02 -1.84842795e-01 2.26084769e-01
2.41593942e-01 2.71557391e-01 3.65156204e-01 1.02633941e+00
2.04571858e-01 -3.52993876e-01 1.54921979e-01 8.31564188e-01
-6.77157640e-02 -1.40366312e-02 -6.33016467e-01 4.56717819e-01
5.62861085e-01 1.26760375e+00 -6.95146859e-01 -4.71978039e-01
-4.14464712e-01 1.12452066e+00 7.20622122e-01 3.60930204e-01
-7.07627952e-01 -7.14209616e-01 2.11777464e-01 3.79284799e-01
5.29960454e-01 -1.00635558e-01 -1.87517971e-01 -7.45641768e-01
6.15678057e-02 -5.42429209e-01 2.09031895e-01 -5.46070635e-01
-1.21477139e+00 1.06951714e+00 1.24318302e-01 -9.63904440e-01
-8.05996954e-02 -5.63517451e-01 -4.59732533e-01 9.69277978e-01
-1.62890959e+00 -1.17015755e+00 -4.49742019e-01 8.67677271e-01
7.21627474e-01 -2.35348523e-01 6.92531288e-01 4.70268071e-01
-5.72472036e-01 7.94814408e-01 -5.01984954e-02 -3.07206929e-01
6.64936304e-01 -1.07950461e+00 2.88768113e-01 6.96096420e-01
-1.05389670e-01 5.80731094e-01 2.13248849e-01 -6.26053452e-01
-7.17029691e-01 -1.19653475e+00 9.84337628e-01 -2.25385427e-01
3.38366717e-01 -5.32735050e-01 -1.04516149e+00 6.74246788e-01
2.81496614e-01 4.98783402e-02 4.09861147e-01 3.16016138e-01
-3.21292788e-01 -4.93940562e-01 -1.12400770e+00 3.36531878e-01
1.05355525e+00 -3.51991266e-01 -5.96584737e-01 1.01072498e-01
8.80862236e-01 2.34827381e-02 -6.52407348e-01 3.58118147e-01
5.12090802e-01 -9.09586370e-01 8.07466924e-01 -2.81442940e-01
-1.45805985e-01 -2.37014025e-01 2.17410371e-01 -1.51621079e+00
-8.17509472e-01 -4.11766797e-01 -1.69082418e-01 1.09783423e+00
6.19941533e-01 -9.25552726e-01 5.95024288e-01 3.38769436e-01
-2.90832788e-01 -8.47266972e-01 -9.45441425e-01 -1.09716642e+00
-2.94231717e-02 -7.38673508e-02 3.95956218e-01 9.34659839e-01
1.28105292e-02 6.71428263e-01 -1.88177958e-01 -1.53127626e-01
2.83386886e-01 -3.14776212e-01 3.37577164e-01 -1.81850541e+00
-2.74762422e-01 -5.56030750e-01 -1.23073667e-01 -9.07114208e-01
1.04061797e-01 -8.80032241e-01 -9.61088315e-02 -1.60256398e+00
2.16285169e-01 -5.75114846e-01 -6.07714593e-01 7.72176504e-01
-2.02659294e-01 1.58781543e-01 3.93685102e-01 3.88357222e-01
-7.86197007e-01 5.47688365e-01 1.20248950e+00 1.48510084e-01
-5.04058480e-01 -1.11877188e-01 -6.14931345e-01 7.78123319e-01
7.58518815e-01 -5.86428523e-01 -6.29916608e-01 -6.92859590e-01
1.32538408e-01 -3.83805603e-01 1.27329737e-01 -1.45446074e+00
6.22950733e-01 4.93104517e-01 5.21457613e-01 -4.28018600e-01
3.91146928e-01 -1.18814802e+00 1.64237827e-01 7.34866917e-01
-3.39042217e-01 2.85006791e-01 5.17866135e-01 5.58390617e-01
-2.04111278e-01 -3.25877786e-01 8.49978507e-01 -1.75662726e-01
-6.25757694e-01 2.27529451e-01 -2.71185517e-01 -3.37853462e-01
1.29024398e+00 -4.24835384e-01 -2.37993002e-01 6.86848760e-02
-1.04294360e+00 2.35953957e-01 1.32810026e-01 3.69980633e-01
4.78016376e-01 -1.21699536e+00 -4.56716657e-01 3.88584942e-01
-5.70379496e-02 1.35827914e-01 4.97290999e-01 9.11630154e-01
-1.28918305e-01 4.09114838e-01 -4.48507637e-01 -3.55989277e-01
-1.09096897e+00 7.32939065e-01 3.04468185e-01 -8.13897312e-01
-4.51340020e-01 1.01946127e+00 4.42186773e-01 -4.08891410e-01
4.83599365e-01 -1.14805609e-01 -5.13973057e-01 4.06901449e-01
5.26770949e-01 5.71457207e-01 1.70616388e-01 -1.34393215e-01
-1.96881056e-01 5.97552001e-01 -5.56564748e-01 1.70051903e-01
1.35890186e+00 -1.30584583e-01 -2.26967633e-01 5.34576178e-01
7.62942612e-01 -2.83854544e-01 -1.48656940e+00 -5.96859992e-01
3.44010815e-02 -2.39524186e-01 1.82782814e-01 -9.76828098e-01
-1.22251439e+00 8.36843014e-01 7.38365412e-01 2.76566565e-01
1.32961130e+00 -5.47421388e-02 4.34381902e-01 3.30536872e-01
1.50847852e-01 -1.10249221e+00 5.31888962e-01 6.19820595e-01
1.02997553e+00 -1.06885779e+00 -9.71202031e-02 -4.73112166e-01
-6.07484460e-01 8.61302912e-01 9.45102036e-01 -1.92983091e-01
8.74194145e-01 2.00198442e-01 -4.29737985e-01 -2.24231973e-01
-8.10965896e-01 -1.89943820e-01 3.98442835e-01 4.65436935e-01
2.70192504e-01 -3.68024588e-01 -1.37767464e-01 6.30091190e-01
3.12844872e-01 1.46657959e-01 -2.89388709e-02 9.29360688e-01
-6.03945255e-01 -1.09210026e+00 -8.96890014e-02 5.11990905e-01
-2.28449166e-01 -3.12054724e-01 -1.08939931e-01 1.00502181e+00
5.15172541e-01 6.50779963e-01 3.07237417e-01 -2.70591170e-01
4.84381258e-01 3.77411246e-01 4.58884269e-01 -8.03055584e-01
-9.58525836e-01 -2.60164384e-02 -1.88004076e-01 -4.94647175e-01
-3.64719510e-01 -4.55838621e-01 -7.51698315e-01 -1.60252646e-01
-6.42098308e-01 1.37325123e-01 4.36499000e-01 7.47189224e-01
6.91398144e-01 1.13802075e+00 4.75143135e-01 -9.85120416e-01
-3.65101486e-01 -1.15738189e+00 -6.51547015e-01 1.19490355e-01
1.36287659e-01 -9.80651796e-01 -3.03982109e-01 -7.00815618e-02]
|
[9.629202842712402, 3.375495195388794]
|
5ba7da10-2c8e-4759-929f-a060667c8d89
|
few-shot-inductive-learning-on-temporal
|
2211.08169
| null |
https://arxiv.org/abs/2211.08169v1
|
https://arxiv.org/pdf/2211.08169v1.pdf
|
Few-Shot Inductive Learning on Temporal Knowledge Graphs using Concept-Aware Information
|
Knowledge graph completion (KGC) aims to predict the missing links among knowledge graph (KG) entities. Though various methods have been developed for KGC, most of them can only deal with the KG entities seen in the training set and cannot perform well in predicting links concerning novel entities in the test set. Similar problem exists in temporal knowledge graphs (TKGs), and no previous temporal knowledge graph completion (TKGC) method is developed for modeling newly-emerged entities. Compared to KGs, TKGs require temporal reasoning techniques for modeling, which naturally increases the difficulty in dealing with novel, yet unseen entities. In this work, we focus on the inductive learning of unseen entities' representations on TKGs. We propose a few-shot out-of-graph (OOG) link prediction task for TKGs, where we predict the missing entities from the links concerning unseen entities by employing a meta-learning framework and utilizing the meta-information provided by only few edges associated with each unseen entity. We construct three new datasets for TKG few-shot OOG link prediction, and we propose a model that mines the concept-aware information among entities. Experimental results show that our model achieves superior performance on all three datasets and our concept-aware modeling component demonstrates a strong effect.
|
['Volker Tresp', 'Zhen Han', 'Yunpu Ma', 'Bailan He', 'Jingpei Wu', 'Zifeng Ding']
|
2022-11-15
| null | null | null | null |
['temporal-knowledge-graph-completion']
|
['knowledge-base']
|
[-3.91574621e-01 6.87368512e-01 -6.55207634e-01 -1.02409758e-01
-1.75272897e-01 -1.21030383e-01 3.61796945e-01 4.49198633e-01
5.58528230e-02 9.81992900e-01 3.26165892e-02 -9.00314003e-02
-4.70337451e-01 -1.48838627e+00 -7.69125283e-01 -1.04832046e-01
-6.72302902e-01 6.64553165e-01 6.44981146e-01 -2.78420627e-01
-2.28516921e-01 1.23087339e-01 -1.36797702e+00 3.05488199e-01
1.22579217e+00 5.13289690e-01 4.67049964e-02 3.28856677e-01
-6.18079960e-01 1.25293720e+00 -1.57320946e-01 -1.07073116e+00
-1.14884488e-01 -1.21883936e-01 -1.03309381e+00 -1.89549774e-01
1.55381650e-01 7.35508204e-02 -8.33494723e-01 9.26074445e-01
1.92970589e-01 2.18024597e-01 5.54468751e-01 -1.71470213e+00
-8.51303101e-01 1.11319792e+00 -4.34379369e-01 3.46249074e-01
4.99608278e-01 -4.72369164e-01 1.19005477e+00 -1.06203640e+00
1.21803916e+00 1.11079836e+00 8.00427496e-01 3.74779373e-01
-7.35735953e-01 -5.80551267e-01 5.48156440e-01 8.13477099e-01
-1.61105585e+00 -1.36930019e-01 8.09111595e-01 -3.53803635e-01
1.13265049e+00 -1.49561957e-01 8.18050444e-01 8.68241549e-01
-6.98789209e-02 8.36520851e-01 3.55145425e-01 -4.97024238e-01
9.51743424e-02 1.73810363e-01 3.44992220e-01 9.77806270e-01
6.35601938e-01 -1.69643700e-01 -9.35771704e-01 -5.53354621e-02
2.64846742e-01 -6.77897828e-03 -2.85021842e-01 -3.99594009e-01
-8.44866574e-01 5.66169560e-01 4.89061207e-01 2.48178557e-01
-3.34693491e-01 1.59694120e-01 3.18934888e-01 2.62656391e-01
6.47164524e-01 -1.52544444e-02 -6.86324298e-01 2.21209899e-01
-5.23797095e-01 -3.67323756e-02 1.03397930e+00 1.61238384e+00
9.60673273e-01 -1.02931492e-01 -4.67846589e-03 5.91200471e-01
2.71159500e-01 1.33170575e-01 2.68908530e-01 -3.29885811e-01
7.91994214e-01 1.19354081e+00 7.88225383e-02 -1.49093819e+00
-4.02429581e-01 -5.87025881e-01 -5.16528904e-01 -6.98333800e-01
-2.18302399e-01 -8.62871259e-02 -9.08570111e-01 1.60256898e+00
6.32415891e-01 7.18415141e-01 2.19869062e-01 2.42026508e-01
1.28554106e+00 5.64296484e-01 4.18542892e-01 -5.11302233e-01
8.36824715e-01 -1.00817120e+00 -9.52243686e-01 -1.40926600e-01
1.04646719e+00 -1.56388819e-01 2.77858794e-01 -6.90056756e-02
-7.21603394e-01 -3.32412153e-01 -7.75620759e-01 1.42573178e-01
-1.01672733e+00 -2.51570880e-01 1.05511916e+00 2.40890548e-01
-8.73306751e-01 7.75903225e-01 -6.13649189e-01 -6.09694302e-01
2.65425086e-01 1.58134609e-01 -4.23091501e-01 -3.60037029e-01
-1.66870880e+00 9.56774533e-01 1.44188547e+00 1.28642559e-01
-7.24147856e-01 -8.49818945e-01 -1.03365111e+00 2.05757201e-01
9.94013667e-01 -6.96079850e-01 5.61431885e-01 -3.40373874e-01
-4.66267139e-01 5.66947758e-01 -1.98365495e-01 -5.08096099e-01
1.63768679e-01 -2.92100245e-03 -1.44534516e+00 2.19093747e-02
3.62897962e-01 4.98977900e-01 5.52665770e-01 -1.37008548e+00
-9.87452626e-01 -2.99395025e-01 1.89155191e-01 1.12257838e-01
-4.87401366e-01 -6.34581745e-01 -8.67044091e-01 -5.46531498e-01
2.14880213e-01 -5.74468017e-01 1.12222224e-01 -5.95359147e-01
-6.15038455e-01 -4.28018302e-01 1.01445675e+00 -5.71645260e-01
1.90177333e+00 -1.75923216e+00 2.03758385e-02 3.43740106e-01
3.73192072e-01 2.95935363e-01 -1.19489945e-01 9.29006517e-01
-1.39478117e-01 1.84584782e-01 3.23813170e-01 1.01065785e-02
-2.08311707e-01 4.74541396e-01 -3.01563770e-01 -2.50940323e-01
-7.20743537e-02 1.26302862e+00 -1.35522342e+00 -8.84398937e-01
-2.13479802e-01 -2.76580397e-02 -2.64245838e-01 -1.61838964e-01
-4.18669134e-01 -1.51912287e-01 -6.94295406e-01 1.05219841e+00
5.36254525e-01 -6.18177474e-01 6.29876316e-01 -4.29615557e-01
3.97379160e-01 -2.32020348e-01 -1.07708085e+00 1.54393387e+00
-1.27889529e-01 2.06262857e-01 -5.73942006e-01 -1.05012691e+00
8.86039436e-01 4.01469111e-01 3.47721487e-01 -4.72134441e-01
-2.83634275e-01 1.99219063e-01 -2.78057426e-01 -7.09155560e-01
5.69035769e-01 -7.94067457e-02 1.71474069e-01 -5.71172759e-02
5.96083581e-01 5.40492654e-01 6.89360201e-01 1.02712703e+00
1.33066225e+00 6.66849539e-02 3.98163766e-01 1.64586142e-01
3.60905945e-01 4.47200269e-01 9.46127713e-01 6.41621530e-01
-1.08723104e-01 -2.89828002e-01 2.95118034e-01 -4.63405222e-01
-5.18202126e-01 -9.91681159e-01 3.58246475e-01 8.60919774e-01
5.10503292e-01 -1.04440212e+00 -1.67178065e-01 -1.15384269e+00
4.35241014e-01 8.93059671e-01 -6.53534412e-01 -4.34476078e-01
-2.41264552e-01 -5.77445090e-01 2.58837759e-01 5.71919680e-01
3.35089415e-01 -8.91382873e-01 4.49321777e-01 5.28650880e-01
-3.40562880e-01 -1.35190439e+00 1.68984160e-02 -1.27481714e-01
-1.06081605e+00 -1.57485938e+00 -1.10270809e-02 -1.02789807e+00
7.73937702e-01 2.90246695e-01 1.16841352e+00 1.92648962e-01
-3.35221559e-01 7.66294003e-01 -7.39677727e-01 -2.99243331e-01
-3.19664031e-02 -6.44876361e-02 9.94865224e-02 -5.79754114e-02
5.78183651e-01 -8.31044674e-01 -2.13629797e-01 1.75975695e-01
-5.10028064e-01 2.40375668e-01 5.18543482e-01 7.13049471e-01
5.86143255e-01 8.64326358e-01 1.09639454e+00 -1.48463261e+00
3.77899855e-01 -9.41313863e-01 -3.02529365e-01 9.09404337e-01
-1.09284353e+00 1.40555138e-02 4.78426933e-01 -4.49300766e-01
-1.34038663e+00 -3.23953301e-01 4.86334145e-01 -5.88295817e-01
4.15109694e-01 1.40270114e+00 -1.24734499e-01 -4.65151034e-02
3.74814868e-01 1.70515314e-01 -7.10363865e-01 -4.26116407e-01
6.70162499e-01 -2.76905578e-02 5.30655503e-01 -6.33017004e-01
1.18638432e+00 3.80718172e-01 1.77663639e-01 -4.45663273e-01
-1.23436391e+00 -6.93111718e-01 -7.55164325e-01 -4.63749498e-01
5.18953741e-01 -1.19125974e+00 -5.01014054e-01 1.29188210e-01
-9.68657494e-01 9.09167454e-02 -3.02893370e-01 5.16909063e-01
-9.71761793e-02 4.24173743e-01 -4.87855941e-01 -7.85561144e-01
-2.65295416e-01 -9.21421498e-02 4.77123708e-01 2.24456355e-01
1.74572900e-01 -1.38891530e+00 1.05424359e-01 3.33987981e-01
6.95826020e-03 5.01121342e-01 1.21093822e+00 -7.98781991e-01
-9.81532216e-01 -4.23192888e-01 -1.49234727e-01 -2.08986357e-01
2.50691593e-01 -8.68257433e-02 -5.19280791e-01 -2.97349304e-01
-7.89641500e-01 -2.15119198e-01 9.78551745e-01 -6.72326237e-02
7.78434694e-01 -3.56183171e-01 -1.13949299e+00 1.69714734e-01
1.68991280e+00 2.86425591e-01 5.54299831e-01 1.64177492e-01
1.01826859e+00 5.15283048e-01 1.02239597e+00 4.02738273e-01
8.99221301e-01 3.57804984e-01 3.05381149e-01 3.38988453e-01
-1.01614170e-01 -9.36133802e-01 3.18903364e-02 1.15581107e+00
-2.74423093e-01 -5.47493398e-01 -9.01770294e-01 1.07024312e+00
-2.37559438e+00 -1.08298421e+00 -4.98510569e-01 1.76729441e+00
7.72792816e-01 2.31420100e-01 -4.29698199e-01 -1.38066858e-01
1.06108403e+00 -1.24294691e-01 -6.25869513e-01 2.53841609e-01
-1.14870653e-01 -4.48698215e-02 2.78138906e-01 3.59513402e-01
-9.79680657e-01 1.27983296e+00 5.55139017e+00 9.63044465e-01
-1.84308991e-01 1.40896469e-01 -6.03291392e-03 2.30296135e-01
-5.10009885e-01 5.02307236e-01 -1.07442999e+00 3.20345521e-01
7.83160090e-01 -7.87376821e-01 2.15239916e-02 1.08561063e+00
-1.56643659e-01 -1.64356995e-02 -1.00251448e+00 9.33031499e-01
6.98412508e-02 -1.43828011e+00 4.50860798e-01 -1.53996676e-01
1.09108233e+00 -2.28246257e-01 -5.36631346e-01 9.28438902e-01
5.69041789e-01 -5.21837831e-01 2.39563748e-01 1.00364864e+00
5.24522126e-01 -8.66242647e-01 6.97913766e-01 4.65357721e-01
-1.77429962e+00 -1.36824682e-01 -5.29211164e-01 2.62144595e-01
2.98808783e-01 8.93000364e-01 -1.18613064e+00 1.39321768e+00
6.90109074e-01 1.10377395e+00 -7.89247632e-01 1.06647444e+00
-4.64795947e-01 4.66671139e-01 1.43215284e-02 1.07155383e-01
-9.17028412e-02 -2.50341613e-02 4.70689446e-01 8.81201029e-01
4.61072534e-01 4.62425590e-01 2.36409754e-01 7.07658052e-01
-4.25000131e-01 2.25274116e-01 -7.31243014e-01 -3.39376718e-01
9.18736935e-01 1.27346015e+00 -6.61430418e-01 -5.41768074e-01
-6.34409428e-01 8.79545867e-01 7.19513118e-01 5.13399601e-01
-8.28420341e-01 -4.50870007e-01 4.16298099e-02 1.34341925e-01
4.38383371e-01 7.75029734e-02 3.71555418e-01 -1.47523224e+00
1.20451011e-01 -7.29601607e-02 1.12807655e+00 -9.35871840e-01
-1.58182085e+00 2.28643879e-01 1.56857267e-01 -1.11141694e+00
-1.23776995e-01 -2.17761144e-01 -7.43364811e-01 5.37348747e-01
-1.71478379e+00 -1.52751970e+00 -4.52194870e-01 7.99936593e-01
7.45429844e-02 -1.80575758e-01 5.47482669e-01 3.77309114e-01
-5.04347205e-01 3.56767654e-01 -4.35342863e-02 2.14020312e-01
4.97271299e-01 -1.18372428e+00 1.94448635e-01 9.47860122e-01
1.36827886e-01 6.70220137e-01 3.83173048e-01 -1.52394593e+00
-1.43184757e+00 -1.68573558e+00 1.35696948e+00 -3.41217965e-01
9.70348537e-01 1.08178079e-01 -1.02250397e+00 1.18723488e+00
-3.39615703e-01 3.85326236e-01 7.51858234e-01 7.22513556e-01
-4.12346244e-01 -1.03381030e-01 -8.98448586e-01 4.20332760e-01
1.59906495e+00 -5.01192987e-01 -8.35465968e-01 3.66706073e-01
1.08876467e+00 -6.63584173e-02 -1.26868284e+00 7.87429452e-01
1.12762295e-01 -3.92823189e-01 1.03672016e+00 -1.03166938e+00
1.07497789e-01 -5.67356825e-01 1.03956923e-01 -1.22617388e+00
-4.57662642e-01 -2.16835931e-01 -1.08267474e+00 1.65676153e+00
6.14343166e-01 -4.68488783e-01 1.10939276e+00 5.99545956e-01
-1.86264098e-01 -6.30063117e-01 -5.21292090e-01 -1.13382387e+00
-6.60296857e-01 -3.40070546e-01 4.94409472e-01 1.52410007e+00
4.56144720e-01 5.55965304e-01 -4.96841967e-01 5.58909595e-01
8.25383782e-01 3.43858093e-01 4.64800566e-01 -1.67234039e+00
-1.32122874e-01 3.00469071e-01 -7.27470696e-01 -3.66266727e-01
2.71465391e-01 -1.31751108e+00 -5.15556812e-01 -2.09337473e+00
4.39056635e-01 -4.45771903e-01 -1.96512297e-01 8.43334913e-01
-3.72673213e-01 -3.30582470e-01 -1.61672786e-01 1.51415557e-01
-1.20184362e+00 6.74398422e-01 1.17842782e+00 -4.06950831e-01
-2.78430045e-01 -2.73178697e-01 -5.08769870e-01 7.07830191e-01
4.09457266e-01 -4.10652101e-01 -9.56758857e-01 -7.22509250e-02
6.89452767e-01 2.92368948e-01 2.73774743e-01 -7.24655628e-01
8.86344790e-01 -7.43066967e-02 3.21095139e-01 -1.07447588e+00
8.39603767e-02 -7.41815209e-01 6.94603145e-01 1.93694159e-01
4.77716774e-02 -4.36039120e-01 5.40047921e-02 1.31852496e+00
-3.93056452e-01 -3.37463975e-01 1.10956147e-01 -2.14231417e-01
-1.66567755e+00 7.37310946e-01 2.60900527e-01 3.42330962e-01
1.16641402e+00 -2.16106698e-01 -6.00894749e-01 -1.98261172e-01
-1.46567583e+00 8.28042150e-01 -1.73581820e-02 4.33722347e-01
9.16986048e-01 -1.68291211e+00 -4.03196543e-01 -4.19186145e-01
7.20887005e-01 3.07985004e-02 5.25933146e-01 6.13533676e-01
-4.20490168e-02 3.53745997e-01 2.75295466e-01 4.28053886e-02
-1.10039020e+00 1.06273603e+00 -1.75692216e-02 -5.41319609e-01
-6.63948596e-01 8.14958155e-01 -8.21115002e-02 -3.08951169e-01
1.73745051e-01 1.20648682e-01 -4.35508728e-01 2.83685923e-01
1.62374109e-01 5.45033455e-01 -2.61000413e-02 -2.82500237e-01
-2.90341377e-01 1.80822745e-01 -4.47999954e-01 5.62024295e-01
1.40568531e+00 -3.24338228e-01 -1.92148462e-01 4.54116613e-01
8.50818634e-01 -1.39693394e-01 -4.94669616e-01 -6.04547024e-01
5.31252623e-01 -3.37713569e-01 -2.12752998e-01 -8.49801958e-01
-1.02834547e+00 2.31368139e-01 -3.77836935e-02 1.34460211e-01
9.03375387e-01 2.70543516e-01 8.12462330e-01 6.61019623e-01
9.60561574e-01 -1.22069800e+00 2.27117930e-02 4.37164605e-01
5.10651886e-01 -1.13400340e+00 1.21966980e-01 -1.13056612e+00
-5.58326662e-01 9.79348004e-01 9.87646878e-01 3.16839576e-01
1.03363574e+00 -2.60733455e-01 -7.51466274e-01 -8.33074927e-01
-1.12999249e+00 -5.79651773e-01 3.68653685e-01 7.84434795e-01
-8.35774839e-02 9.66221988e-02 -2.59925455e-01 7.61459529e-01
1.86747044e-01 2.15617895e-01 4.71347302e-01 9.68967438e-01
-5.48175097e-01 -1.16229665e+00 2.94173449e-01 6.11862659e-01
2.02844422e-02 -2.50866622e-01 -3.75928760e-01 1.09142470e+00
3.97197336e-01 8.38515162e-01 -3.91424447e-01 -6.31662965e-01
3.20624709e-01 3.30784649e-01 2.87875473e-01 -8.13054562e-01
2.22042754e-01 -4.89143431e-01 6.11155152e-01 -3.74208689e-01
-5.60901105e-01 -2.17339158e-01 -1.37898612e+00 -3.73963803e-01
-7.40734518e-01 5.19976676e-01 2.05529016e-02 8.74277592e-01
5.33488095e-01 5.43190956e-01 4.12927121e-01 -5.01052924e-02
2.51711667e-01 -7.13504910e-01 -1.16388750e+00 5.47882438e-01
-3.55592787e-01 -1.04146147e+00 -2.04881072e-01 1.02515504e-01]
|
[8.688430786132812, 7.9607768058776855]
|
578e44dd-2f68-4d0e-bb3c-0abe369798de
|
an-inter-and-intra-band-loss-for
|
2008.05133
| null |
https://arxiv.org/abs/2008.05133v1
|
https://arxiv.org/pdf/2008.05133v1.pdf
|
An Inter- and Intra-Band Loss for Pansharpening Convolutional Neural Networks
|
Pansharpening aims to fuse panchromatic and multispectral images from the satellite to generate images with both high spatial and spectral resolution. With the successful applications of deep learning in the computer vision field, a lot of scholars have proposed many convolutional neural networks (CNNs) to solve the pansharpening task. These pansharpening networks focused on various distinctive structures of CNNs, and most of them are trained by L2 loss between fused images and simulated desired multispectral images. However, L2 loss is designed to directly minimize the difference of spectral information of each band, which does not consider the inter-band relations in the training process. In this letter, we propose a novel inter- and intra-band (IIB) loss to overcome the drawback of original L2 loss. Our proposed IIB loss can effectively preserve both inter- and intra-band relations and can be directly applied to different pansharpening CNNs.
|
['Bo Huang', 'Jiajun Cai']
|
2020-08-12
| null | null | null | null |
['pansharpening']
|
['computer-vision']
|
[ 4.66021478e-01 -5.78590989e-01 -3.82733867e-02 -3.93330276e-01
-5.45731246e-01 -3.67918313e-01 3.05242062e-01 -3.98934603e-01
-4.88670081e-01 6.45548820e-01 3.19869444e-02 -1.53900370e-01
-4.23689157e-01 -1.24474955e+00 -6.49916947e-01 -9.45755661e-01
4.37225252e-01 -4.80273187e-01 1.17143579e-01 -6.05956852e-01
-2.92571396e-01 6.90137923e-01 -1.36459100e+00 2.11226344e-01
1.20743299e+00 1.14519131e+00 3.13533276e-01 4.47629213e-01
1.32361352e-01 5.66794813e-01 -4.37572271e-01 -2.01527268e-01
6.68321788e-01 -5.71941495e-01 -4.24879342e-01 1.51285559e-01
4.65184480e-01 -3.03333730e-01 -7.00594604e-01 1.68051207e+00
4.97724771e-01 1.74187109e-01 3.36570948e-01 -1.02300870e+00
-9.30356145e-01 5.80745935e-01 -1.16563308e+00 2.72104800e-01
-4.30472761e-01 1.14511639e-01 9.40360963e-01 -3.78432065e-01
4.50741351e-02 1.08679378e+00 9.14505064e-01 1.02378413e-01
-9.79798734e-01 -8.33993793e-01 -1.68083563e-01 2.54110873e-01
-1.33069503e+00 1.18529513e-01 1.01111579e+00 -1.74178079e-01
3.87163877e-01 3.61730933e-01 8.23841810e-01 3.94607186e-01
1.35855556e-01 6.71278000e-01 1.11104524e+00 -2.89019525e-01
-5.14028907e-01 -3.55824649e-01 -3.30415107e-02 3.55535865e-01
1.91510111e-01 5.72350144e-01 9.21298936e-02 3.31754088e-01
1.04906452e+00 1.13847986e-01 -7.21439362e-01 -1.21521458e-01
-1.02649903e+00 9.97478724e-01 1.13971806e+00 3.53483200e-01
-3.02970797e-01 -3.77176739e-02 3.41785461e-01 2.89880157e-01
4.92832452e-01 1.96647689e-01 -4.43830997e-01 5.54371059e-01
-9.70393777e-01 2.66017467e-01 1.83598585e-02 5.00173330e-01
1.11498249e+00 8.86153653e-02 -4.15101536e-02 1.25879288e+00
1.92140266e-01 4.72008824e-01 4.72353071e-01 -6.03444695e-01
5.78819573e-01 5.70692182e-01 -6.31942973e-02 -1.24629056e+00
-5.63827515e-01 -6.03053212e-01 -1.41745400e+00 8.18019807e-02
-2.79278625e-02 -3.73294026e-01 -9.59745586e-01 1.71632266e+00
1.39979407e-01 2.99634039e-02 -1.29271410e-02 1.38571835e+00
8.15835178e-01 1.17348111e+00 -5.70701174e-02 -1.96331605e-01
9.74658012e-01 -1.12550294e+00 -5.70857048e-01 -3.86810511e-01
6.73637465e-02 -9.27808166e-01 8.76523137e-01 8.05173442e-02
-7.64124691e-01 -7.95616567e-01 -1.29198968e+00 -2.07161993e-01
-2.69562334e-01 3.32259089e-01 7.56792307e-01 5.85841000e-01
-7.69216180e-01 5.62622964e-01 -4.00722474e-01 -5.67873418e-02
4.18002278e-01 1.06730916e-01 -1.50059432e-01 -1.22652709e-01
-1.59283400e+00 7.01825380e-01 8.96637499e-01 5.86415946e-01
-5.45395911e-01 -6.84609294e-01 -7.32141852e-01 2.51582891e-01
2.10740596e-01 -3.73694777e-01 8.64513755e-01 -1.48271048e+00
-1.38087630e+00 6.91401482e-01 5.34764171e-01 -2.71881104e-01
4.39968407e-01 -1.88696012e-01 -9.19955730e-01 5.88237271e-02
1.14859352e-02 5.60410738e-01 7.55428314e-01 -1.07996750e+00
-6.48788393e-01 -2.88192838e-01 -8.78057927e-02 3.01960945e-01
-3.26327205e-01 -6.06376901e-02 -3.95504445e-01 -1.15165257e+00
3.05557340e-01 -6.18526697e-01 -1.85562089e-01 1.65624112e-01
-2.04275027e-01 2.13147625e-01 1.07302129e+00 -1.08937800e+00
7.28297353e-01 -2.19087791e+00 2.34236047e-02 8.80569071e-02
-1.77950755e-01 7.91170537e-01 -4.93845820e-01 7.71249831e-02
-3.48667443e-01 -1.43666789e-01 -7.88721025e-01 4.26703006e-01
-3.53065223e-01 1.64984122e-01 -4.23299670e-01 5.67900658e-01
1.17436074e-01 7.07588255e-01 -6.65948987e-01 -3.98626268e-01
4.03453737e-01 7.63016641e-01 -2.43625507e-01 1.96852744e-01
-2.33635902e-01 2.27535501e-01 -3.80173594e-01 3.39785337e-01
1.39096057e+00 1.89924955e-01 -1.45635128e-01 -6.80630326e-01
-2.30785027e-01 -3.01558018e-01 -8.77408087e-01 1.59637833e+00
-5.38776755e-01 5.49375176e-01 3.68557960e-01 -1.14619339e+00
1.04905581e+00 4.66240086e-02 5.67512214e-01 -9.90653396e-01
8.59251171e-02 2.47249603e-01 -7.53780305e-02 -3.08577687e-01
5.21218419e-01 -3.85882199e-01 3.00187439e-01 9.56787169e-02
-1.63072959e-01 -4.86380786e-01 -2.86295097e-02 -4.05475169e-01
9.74557623e-02 3.50985788e-02 9.87842008e-02 -2.58260161e-01
8.40365589e-01 -3.76145393e-02 8.48443866e-01 4.07656759e-01
-1.07873507e-01 7.83381939e-01 6.40837178e-02 -6.15989864e-01
-1.15420055e+00 -8.30740392e-01 -1.61790445e-01 8.35403144e-01
5.49470425e-01 3.39992255e-01 -5.71264446e-01 -4.41090047e-01
-1.18385658e-01 3.59492600e-01 -4.20369655e-01 -3.85227472e-01
-4.97462153e-01 -1.35090768e+00 6.89600110e-01 3.35636526e-01
1.50569308e+00 -1.01231885e+00 -4.19648349e-01 2.44230822e-01
-3.85874510e-01 -8.31378400e-01 -6.86801314e-01 -8.64834487e-02
-8.11170101e-01 -1.07958329e+00 -1.10530436e+00 -7.35587716e-01
2.06465036e-01 7.16165185e-01 6.74278080e-01 -1.26444623e-01
-1.99331746e-01 -2.06552044e-01 -3.65881264e-01 -2.38815129e-01
-1.77860200e-01 1.88263133e-01 -5.17803252e-01 4.14265186e-01
5.61429150e-02 -7.58349359e-01 -6.32684410e-01 3.32189590e-01
-1.42217255e+00 3.49498808e-01 7.48267949e-01 9.54524755e-01
5.69839895e-01 7.21497536e-01 3.52983952e-01 -4.17166471e-01
5.00869572e-01 -1.09413788e-01 -9.68867898e-01 5.29309750e-01
-5.07199764e-01 -1.81717962e-01 6.27851784e-01 -2.38183260e-01
-1.33863294e+00 -6.48326799e-02 -4.04585034e-01 -3.28167826e-01
4.29916792e-02 7.51642048e-01 -4.44113523e-01 -4.41153646e-01
5.72333276e-01 5.05776525e-01 -2.12560207e-01 -6.29138350e-01
3.74911249e-01 6.58160329e-01 8.20559680e-01 -2.58511871e-01
7.84225702e-01 4.51039493e-01 1.25903502e-01 -7.99443901e-01
-1.17934728e+00 -2.86224633e-01 -2.94911414e-01 -1.12159990e-01
9.36011732e-01 -1.07242143e+00 -3.75186622e-01 1.06572032e+00
-1.02999282e+00 -2.57152766e-01 1.76784135e-02 4.42425758e-01
-1.53012305e-01 7.05595493e-01 -5.81565738e-01 -3.72651309e-01
-5.81177652e-01 -1.06657732e+00 7.50912488e-01 6.73655748e-01
8.64043534e-01 -7.29194999e-01 1.69193260e-02 2.51889437e-01
5.89799464e-01 2.70935565e-01 8.30991089e-01 2.27847099e-01
-2.40897179e-01 -1.25942051e-01 -1.00530148e+00 8.06206346e-01
3.41839254e-01 -3.57349999e-02 -8.26711297e-01 -3.81158799e-01
1.92022428e-01 -3.99010420e-01 1.43125427e+00 5.84676921e-01
1.36147499e+00 -3.56251985e-01 1.59588829e-01 1.19403827e+00
1.73599494e+00 2.34860048e-01 9.40520108e-01 6.11636996e-01
8.62091839e-01 3.96402806e-01 4.17351931e-01 6.21520728e-02
4.21672724e-02 5.36721766e-01 6.29153907e-01 -5.37291944e-01
-1.05147138e-01 -6.86094165e-02 6.86984807e-02 3.13110322e-01
-2.25960210e-01 -3.39533016e-02 -6.87531412e-01 4.52570081e-01
-1.81809831e+00 -1.02761900e+00 -2.12841138e-01 1.91023386e+00
1.02848840e+00 -1.92098841e-01 -1.79056481e-01 4.60453443e-02
1.15349913e+00 7.84236252e-01 -6.31812871e-01 4.07536142e-02
-7.90870845e-01 1.36277348e-01 9.87412870e-01 4.16613042e-01
-1.57137692e+00 6.98009014e-01 5.70954275e+00 1.11461830e+00
-1.55978942e+00 7.03045949e-02 6.24092996e-01 3.04003775e-01
-2.96854824e-01 -1.11474641e-01 -2.05201074e-01 3.74070883e-01
2.33385831e-01 -8.27793106e-02 5.77241957e-01 6.11102819e-01
3.82550471e-02 -1.38083417e-02 -1.81907415e-01 1.12790167e+00
-1.66908801e-01 -1.23721015e+00 1.00286871e-01 -1.95745185e-01
1.13537908e+00 2.88498282e-01 1.80860013e-01 -1.14080071e-01
1.75922617e-01 -8.41403246e-01 6.25531256e-01 3.55385661e-01
6.70650005e-01 -9.64605391e-01 7.26452649e-01 2.46766731e-01
-1.28793633e+00 -2.04775110e-01 -6.87583983e-01 1.81597725e-01
1.38785448e-02 1.00640297e+00 1.73091009e-01 1.06012702e+00
7.63494194e-01 9.25771475e-01 -4.20841396e-01 1.22282135e+00
-3.27630520e-01 3.91149521e-01 -2.02970058e-01 5.05037785e-01
5.42976201e-01 -6.62468255e-01 4.01850045e-01 1.09862077e+00
4.86089647e-01 1.52305245e-01 1.57899991e-01 9.79663968e-01
7.07912678e-03 -5.82549348e-03 -8.19621831e-02 -1.18376605e-01
-1.99754611e-02 1.29417813e+00 -3.86844695e-01 -1.28855109e-01
-3.80275518e-01 8.19697201e-01 -1.69279695e-01 4.09132957e-01
-9.08578753e-01 -6.59424245e-01 5.84026754e-01 -3.72060329e-01
2.74671733e-01 -1.05832875e-01 -3.05882573e-01 -1.19987035e+00
-2.26828843e-01 -8.89119864e-01 5.57164192e-01 -1.02827096e+00
-1.20671451e+00 5.32799780e-01 -1.86384752e-01 -1.23094964e+00
5.56925178e-01 -5.76457202e-01 -7.05329776e-01 1.24189079e+00
-1.98007631e+00 -1.52423620e+00 -6.82600260e-01 7.32448578e-01
2.36891329e-01 -4.24085297e-02 2.30134293e-01 4.22072440e-01
-5.72310746e-01 1.97528958e-01 3.57078195e-01 2.31908143e-01
5.55769980e-01 -7.51848817e-01 1.38836160e-01 1.21187949e+00
-1.27155483e-01 7.23850876e-02 5.71175456e-01 -2.87869364e-01
-7.04771042e-01 -1.48494589e+00 4.12639916e-01 7.61432350e-01
4.20608908e-01 4.49152529e-01 -1.04466236e+00 4.57739949e-01
2.67890245e-01 5.34097813e-02 1.63613513e-01 -4.02577639e-01
-5.31150043e-01 -6.59049451e-01 -1.07934678e+00 3.17295969e-01
6.30184054e-01 -4.80595529e-01 -3.83243114e-01 3.35207194e-01
6.14397943e-01 -3.60351086e-01 -5.32034099e-01 8.79592359e-01
4.56975251e-01 -1.16445088e+00 1.11034155e+00 -1.25146061e-01
6.95603907e-01 -4.27501470e-01 -1.16070822e-01 -1.49954367e+00
-5.23809791e-01 -2.13488489e-01 5.92871010e-01 8.65114808e-01
1.39523596e-01 -6.85339987e-01 4.02900010e-01 2.30697319e-02
-3.36577743e-01 -2.31033161e-01 -6.00388765e-01 -6.36525750e-01
2.28833705e-01 -1.01220548e-01 9.63302791e-01 1.12214434e+00
-7.29420125e-01 1.58358626e-02 -7.04496682e-01 4.52380598e-01
6.58654749e-01 5.57578683e-01 3.76057088e-01 -1.12414515e+00
-3.00688535e-01 -7.86068916e-01 6.04363270e-02 -9.21404600e-01
2.18202658e-02 -7.05683053e-01 6.33981451e-02 -1.55386400e+00
1.49342522e-01 -3.65351647e-01 -4.55980152e-01 6.33066714e-01
-2.86054134e-01 5.14064610e-01 1.30126476e-01 2.54106730e-01
1.67996794e-01 9.50824976e-01 1.66349018e+00 -6.01051152e-01
-1.59229562e-01 -5.96615747e-02 -5.81109643e-01 5.52268147e-01
9.41205800e-01 -2.19557077e-01 -2.48168021e-01 -8.60412419e-01
2.54792005e-01 1.35019973e-01 7.27121592e-01 -1.09615993e+00
1.27883330e-01 -4.60715890e-01 3.69781584e-01 -6.68064952e-01
1.71423420e-01 -7.79768705e-01 3.71464849e-01 4.64885503e-01
-2.11953446e-01 -4.36752647e-01 3.68703634e-01 2.91999072e-01
-6.38392985e-01 -4.90986705e-02 1.34267020e+00 -3.10854584e-01
-9.01590884e-01 5.91849446e-01 9.54167023e-02 -3.65867883e-01
7.99836814e-01 8.33331421e-02 -3.88534069e-01 -2.77016073e-01
-3.66356105e-01 1.59867868e-01 2.88958430e-01 2.63608158e-01
5.06395340e-01 -1.43918788e+00 -8.18550766e-01 3.39706033e-01
-1.07036382e-01 1.08731993e-01 8.46832931e-01 6.33198202e-01
-1.02540326e+00 2.89884031e-01 -6.99999988e-01 -2.22148240e-01
-9.51510429e-01 4.97759849e-01 1.02648032e+00 -2.13913798e-01
-5.59004128e-01 9.32478428e-01 4.46129531e-01 -6.38720691e-01
-2.06078857e-01 -3.58084291e-02 -2.18411550e-01 -3.30376215e-02
4.04985309e-01 2.92961210e-01 9.09215286e-02 -8.76721740e-01
-6.95978850e-02 7.66509235e-01 9.96982530e-02 7.90189654e-02
1.53270328e+00 -3.87444906e-02 -5.10428369e-01 -1.59961909e-01
1.31088901e+00 -2.64343351e-01 -1.50745082e+00 -5.06778240e-01
-4.69153523e-01 -6.12386167e-01 5.52667677e-01 -8.26389313e-01
-1.83473027e+00 1.06392384e+00 6.91370130e-01 2.11289614e-01
1.74361396e+00 -5.18242836e-01 1.18583512e+00 1.55776948e-01
2.92874011e-03 -1.13866913e+00 -1.42148599e-01 4.24225241e-01
9.33580220e-01 -1.06456208e+00 1.41988099e-01 -4.25852746e-01
-4.18876767e-01 1.26364613e+00 6.45114601e-01 -1.86382517e-01
6.77061379e-01 -2.01659501e-01 1.24498360e-01 -9.11219567e-02
1.65570900e-01 -2.48843253e-01 3.72693211e-01 5.13963163e-01
2.12879077e-01 1.05441891e-01 -4.84269142e-01 4.09954846e-01
-2.13771816e-02 1.83393970e-01 3.28601450e-01 3.44956964e-01
-5.70083618e-01 -7.97384858e-01 -5.39449871e-01 3.55346352e-01
-2.76466131e-01 -1.57005459e-01 -6.82305321e-02 5.85281909e-01
5.23154795e-01 7.79732764e-01 -1.67957097e-02 -3.29174191e-01
2.81495631e-01 -6.04562461e-01 3.29137981e-01 -3.10638119e-02
-3.48466814e-01 2.32574046e-01 -3.92741233e-01 -1.75029293e-01
-9.77036536e-01 -2.07166627e-01 -6.88739777e-01 -4.35172886e-01
-4.32136089e-01 8.55789781e-02 2.72174329e-01 8.00163865e-01
-9.43398029e-02 4.81222510e-01 8.79856586e-01 -8.83553088e-01
-5.15948415e-01 -9.31516171e-01 -1.12781286e+00 4.13101822e-01
4.86565650e-01 -4.16998148e-01 -1.97954684e-01 -9.31848958e-02]
|
[10.210060119628906, -1.8878494501113892]
|
f2914c63-b27d-4517-86c6-ade653710804
|
histogram-equalization-of-the-image
|
2108.12818
| null |
https://arxiv.org/abs/2108.12818v1
|
https://arxiv.org/pdf/2108.12818v1.pdf
|
Histogram Equalization Of The Image
|
The relevance and impact of probability distributions on image processing are the subject of this study.It may be characterized as a probability distribution function of brightness for a certain area, which might be a whole picture. To generate a histogram, the probability density function of the brightness is frequently calculated by counting how many times each brightness occurs in the picture region. The brightness average is defined as the sample mean of the brightness of pixels in a certain region. The frequency is shown by the histogram. The histogram has a wide range of uses in image processing. It could, for starters, be used for picture analysis. Second, the functions of an image's brightness and contrast, as well as the final two uses of equalizing and thresholding. Normalizing a histogram is one technique to convert the intensities of discrete distributions to the probability of discrete distribution functions. The technique to equalize the histogram is to control the image's contrast by altering their intensity distribution functions. The major goal of this procedure is to give the cumulative probability function a linear trend (CDF).A method of segmentation is to divide a section of the picture into constituent areas or objects.
|
['Ibraheem Shayea', 'W. T Al-Shaibani', 'Melih Gokdemir', 'Irem Doken']
|
2021-08-29
| null | null | null | null |
['local-color-enhancement']
|
['computer-vision']
|
[ 3.89788479e-01 -2.54305780e-01 -4.32082228e-02 -4.90396500e-01
-1.64652035e-01 -5.41287720e-01 4.50180441e-01 1.71588078e-01
-4.18390006e-01 4.05762583e-01 -9.82339084e-02 -4.01442558e-01
9.45159346e-02 -1.14494598e+00 -4.38045979e-01 -1.21308327e+00
1.26498282e-01 -9.48368087e-02 5.45856714e-01 1.67880997e-01
5.06609857e-01 8.11238706e-01 -1.91626406e+00 2.15877414e-01
4.74785477e-01 9.99946713e-01 1.49026632e-01 8.80326211e-01
-4.63023901e-01 5.96451879e-01 -1.08796728e+00 1.21098332e-01
4.51509953e-02 -8.19969594e-01 -6.57442093e-01 5.15079856e-01
2.25952342e-01 -2.55116493e-01 -3.86045277e-02 1.56125557e+00
-1.08957417e-01 1.66628614e-01 1.13512325e+00 -1.22887468e+00
-5.21056175e-01 1.13980539e-01 -7.27709353e-01 5.83899796e-01
3.29289734e-01 -1.88361123e-01 1.82162657e-01 -2.74252415e-01
2.16816559e-01 1.30185843e+00 3.26460868e-01 -1.48724556e-01
-1.19416881e+00 -1.35621279e-01 -5.10381460e-01 1.53081179e-01
-1.47801268e+00 2.77306795e-01 4.97352958e-01 -3.73328537e-01
5.55066943e-01 5.89848936e-01 8.47911537e-01 -1.89490259e-01
9.01395857e-01 4.79285985e-01 1.34144950e+00 -9.74856436e-01
1.38532296e-01 8.51613432e-02 1.77260250e-01 3.35211068e-01
7.47022852e-02 -3.54008883e-01 1.15150854e-01 5.13515882e-02
8.52217257e-01 2.07841069e-01 -1.19483516e-01 7.76460469e-02
-8.53200912e-01 6.23123109e-01 2.08129644e-01 5.01888156e-01
-4.14303988e-01 2.04545870e-01 3.26627381e-02 2.28185803e-01
3.20823163e-01 1.59929305e-01 -9.89972726e-02 -3.82104591e-02
-1.00570083e+00 1.47154808e-01 6.33064091e-01 3.76769066e-01
9.51921344e-01 -2.73383349e-01 -2.42212951e-01 6.11033082e-01
1.45498872e-01 9.07488704e-01 5.05843520e-01 -1.01114786e+00
-2.83621371e-01 6.27347171e-01 -3.82959209e-02 -9.91472661e-01
-3.56414467e-02 2.93356925e-01 -5.55888772e-01 7.07547009e-01
1.05086446e+00 1.10370845e-01 -1.20415914e+00 1.25719428e+00
2.57353067e-01 -5.53936899e-01 -3.18975210e-01 4.87250119e-01
5.40059566e-01 1.53170252e+00 2.38661334e-01 -5.92626929e-01
1.69618273e+00 -2.14541912e-01 -1.03936231e+00 2.97824174e-01
-7.04703629e-02 -1.06412828e+00 1.02630496e+00 4.50659543e-01
-1.11756206e+00 -6.99653745e-01 -8.21987092e-01 3.00654054e-01
-6.73304677e-01 -1.06977023e-01 9.30434838e-02 6.20625973e-01
-9.07705665e-01 6.14848018e-01 -4.57407862e-01 -3.43628705e-01
-5.82141764e-02 1.34223193e-01 -1.10711470e-01 2.41995469e-01
-9.45045114e-01 1.09373295e+00 4.33220476e-01 -1.85505509e-01
-3.13704967e-01 -1.89533249e-01 -8.15564692e-01 3.31618667e-01
-2.86582589e-01 -1.18707590e-01 9.98280823e-01 -1.31254387e+00
-1.12433362e+00 1.03452528e+00 -2.72941321e-01 9.60585102e-02
-1.29605932e-02 3.88650149e-01 -5.46527445e-01 4.35932308e-01
-6.26605153e-02 5.04766047e-01 1.15339196e+00 -1.23920202e+00
-9.94716883e-01 -1.31354541e-01 -4.68713671e-01 6.88533532e-03
-1.36916712e-01 1.34167612e-01 -4.51158494e-01 -3.36919785e-01
3.68184745e-01 -6.12255335e-01 -1.96018647e-02 -3.91454697e-01
-4.09632102e-02 -4.49317276e-01 1.00505853e+00 -8.29829514e-01
1.50318205e+00 -2.70199180e+00 -6.45469964e-01 7.42056966e-01
-1.60068095e-01 -1.79018110e-01 3.91226023e-01 1.64191023e-01
-3.37666929e-01 1.34282395e-01 -5.07667363e-01 6.63567662e-01
-7.14203343e-02 1.08716726e-01 -1.87426373e-01 8.08115602e-01
1.03424445e-01 2.15245903e-01 -6.05825722e-01 -8.04732263e-01
5.10349512e-01 5.54295480e-01 1.25230253e-01 2.40894929e-01
2.80045569e-01 -3.86151634e-02 -1.18276671e-01 3.55780184e-01
1.01623249e+00 1.72287986e-01 -4.21987753e-03 -2.54065484e-01
-5.40989459e-01 -2.41240859e-01 -1.02136445e+00 3.88491422e-01
2.18828440e-01 1.15988255e+00 -3.68060917e-01 -7.96616256e-01
1.21936643e+00 2.21665829e-01 4.53304410e-01 -8.12785029e-01
3.92077148e-01 3.41682397e-02 9.16631967e-02 -6.99520469e-01
6.75114274e-01 -3.27879846e-01 -9.27763805e-02 3.60467881e-01
-2.64864057e-01 -8.20599079e-01 5.25184929e-01 -8.16276595e-02
5.44539332e-01 -1.78535774e-01 6.34030759e-01 -5.17989457e-01
5.02007008e-01 2.15109602e-01 2.35780403e-01 3.49976599e-01
-2.75485694e-01 5.19689381e-01 8.81028473e-01 -4.23682868e-01
-1.25923097e+00 -1.36134481e+00 -6.45714879e-01 7.90253222e-01
5.14092624e-01 3.87394875e-01 -1.13173711e+00 -7.46105164e-02
-3.01881321e-02 9.63145316e-01 -7.19670773e-01 7.51559716e-03
-4.45423543e-01 -9.77002680e-01 -2.13301316e-01 1.40737416e-02
5.19671440e-01 -1.34702480e+00 -8.46505105e-01 -1.25642300e-01
-4.40788418e-02 -4.66774255e-01 -2.55073607e-01 2.41969526e-01
-9.81732488e-01 -8.96415412e-01 -8.47672462e-01 -8.78936887e-01
1.07431340e+00 3.75375062e-01 1.02470565e+00 3.34866694e-04
-3.56164068e-01 4.87051845e-01 -3.50487977e-01 -5.03351629e-01
-8.32797527e-01 -4.89167482e-01 -4.18984979e-01 -5.46618411e-03
5.43218315e-01 -3.30175012e-01 -7.39309311e-01 3.77007365e-01
-1.23284698e+00 -5.92169642e-01 5.51386893e-01 1.79944217e-01
1.01563764e+00 8.84206176e-01 8.97386111e-03 -5.32283485e-01
6.79170787e-01 -3.58317137e-01 -7.03318655e-01 3.55449826e-01
-2.76246965e-01 -1.05054446e-01 2.43228450e-01 -2.61913389e-01
-1.12443197e+00 -7.76830316e-02 2.93163806e-01 4.99536023e-02
-5.08445144e-01 9.51051563e-02 -1.07964789e-02 3.44769329e-01
5.26910424e-01 3.50210726e-01 4.64127481e-01 1.31339952e-03
2.64771074e-01 7.56065786e-01 9.90772486e-01 2.09259558e-02
4.21150297e-01 3.51150751e-01 1.69681653e-01 -1.08950377e+00
-3.40344369e-01 -8.02590430e-01 -5.39682567e-01 -8.40414226e-01
1.22409368e+00 -1.04166761e-01 -6.12041950e-01 5.68703711e-01
-9.15580630e-01 -1.77447885e-01 -3.77968729e-01 5.50177515e-01
-2.89917588e-01 5.05629003e-01 -3.01830977e-01 -9.78210986e-01
3.47415209e-02 -9.63620007e-01 5.26179254e-01 1.09403098e+00
-1.70123234e-01 -1.09652793e+00 -4.50404845e-02 -2.37801924e-01
2.12818265e-01 2.46095315e-01 1.18455243e+00 -2.60377020e-01
-5.07301092e-02 -6.31557405e-01 -3.23029220e-01 6.34897947e-01
3.50877851e-01 8.74005854e-01 -7.09798098e-01 9.32122767e-02
2.14899182e-01 1.76919818e-01 6.85696661e-01 1.03112149e+00
1.34280944e+00 7.22529693e-03 -2.10560873e-01 1.81878299e-01
1.54953659e+00 8.41761947e-01 1.30276537e+00 5.27549684e-01
-1.16530829e-03 6.58452392e-01 7.45941520e-01 3.23428392e-01
-2.33817518e-01 3.67186964e-01 3.65036845e-01 -4.54908133e-01
-1.76497728e-01 1.04275219e-01 2.49066576e-01 5.90259671e-01
-1.22203253e-01 -1.93552062e-01 -7.00640202e-01 5.32990634e-01
-1.09349871e+00 -1.35801709e+00 -6.88906074e-01 2.43595672e+00
7.61520803e-01 1.37871668e-01 3.07014912e-01 3.93018484e-01
1.30006039e+00 -7.16986656e-02 7.17170238e-02 -9.82974589e-01
3.90259810e-02 -2.45458502e-02 7.72596896e-01 5.39086401e-01
-1.17442131e+00 2.81957269e-01 7.92359352e+00 1.07495928e+00
-1.08585858e+00 -5.11103630e-01 9.71322000e-01 5.86599767e-01
-2.02233016e-01 1.25254346e-02 -5.51863790e-01 8.47261369e-01
6.25822484e-01 -1.83762401e-01 8.89954567e-02 6.82289362e-01
3.89054179e-01 -1.23742652e+00 -5.51072955e-01 8.02824736e-01
-1.08463272e-01 -4.71190602e-01 -4.18345369e-02 2.17336327e-01
7.25085557e-01 -7.79753566e-01 2.36065522e-01 -1.47860005e-01
-5.07202074e-02 -9.26702201e-01 8.18339407e-01 8.48228514e-01
8.02291989e-01 -7.81807959e-01 9.94946897e-01 -1.54576136e-03
-8.79653335e-01 1.71464548e-01 -5.56398273e-01 -1.49092391e-01
2.06906646e-01 7.53360271e-01 -6.68799877e-01 -1.75033167e-01
8.51223052e-01 -1.77505940e-01 -5.01361251e-01 1.44579923e+00
-1.30815208e-01 7.51760781e-01 -4.78906542e-01 -2.13233382e-01
2.52514511e-01 -6.74293816e-01 1.78608730e-01 1.43418300e+00
4.04287547e-01 1.73548684e-01 -2.27734357e-01 7.38757551e-01
3.41520995e-01 2.23295838e-01 -4.11168605e-01 1.59521326e-01
7.69520581e-01 1.17203081e+00 -1.51492214e+00 -6.49475873e-01
-3.59757066e-01 7.08528399e-01 -5.30288160e-01 3.98562938e-01
-8.91459584e-01 -8.84093165e-01 -4.19334695e-02 1.91797242e-01
3.71123850e-02 -1.61523316e-02 -6.21373475e-01 -2.72956099e-02
-2.74833888e-01 -4.84532505e-01 3.73780429e-01 -8.72598588e-01
-1.05618393e+00 4.11504686e-01 5.31258047e-01 -1.18542969e+00
4.05263528e-02 -5.96862555e-01 -9.09653068e-01 1.05089319e+00
-9.61987197e-01 -3.75265598e-01 -3.55176777e-01 6.38061583e-01
4.05373096e-01 -4.52962890e-02 4.30840790e-01 -3.81765850e-02
-2.83807456e-01 4.66956235e-02 2.37540856e-01 6.20154850e-02
3.02684337e-01 -1.52110565e+00 -2.33805150e-01 8.66744399e-01
-2.63532132e-01 1.77206531e-01 1.07576537e+00 -5.39094448e-01
-5.06918848e-01 -2.88026899e-01 1.13229501e+00 -3.16597484e-02
3.65171611e-01 2.35538393e-01 -1.01430357e+00 2.17833146e-01
3.24474156e-01 -3.35818321e-01 4.66644615e-01 -4.47888732e-01
2.97308654e-01 -2.53050953e-01 -1.22218740e+00 3.95617187e-01
-3.09818596e-01 -1.31887272e-01 -7.35499978e-01 -8.51830244e-02
5.74444644e-02 -4.08723317e-02 -8.77986431e-01 8.23043734e-02
8.12607944e-01 -1.28335822e+00 7.04950392e-01 3.62586349e-01
4.34556425e-01 -6.07731998e-01 3.81926373e-02 -1.04873228e+00
-4.97901738e-01 -2.03391343e-01 7.61566401e-01 1.15019357e+00
4.36192483e-01 -5.41808665e-01 4.59984511e-01 7.67366230e-01
8.04637372e-02 -3.14935178e-01 -7.48043597e-01 -6.23615503e-01
-1.19586423e-01 -1.83024183e-01 4.40928042e-01 6.69325173e-01
1.71202376e-01 -2.95818865e-01 2.48187035e-01 -3.30106057e-02
6.49454713e-01 -5.98172396e-02 4.57791388e-01 -8.47768843e-01
2.53244251e-01 -8.43403757e-01 -4.86779749e-01 -7.07586706e-01
-4.47890222e-01 -3.18903148e-01 2.74353653e-01 -1.69382977e+00
3.81576866e-01 -1.93890437e-01 9.77714434e-02 1.90744884e-02
-3.36034894e-01 2.65209228e-01 1.55667797e-01 5.28200924e-01
-8.77637267e-02 5.02218027e-03 1.37273586e+00 1.07480183e-01
-2.49348283e-01 1.45697325e-01 3.84835415e-02 8.86263132e-01
8.57390404e-01 -5.30090690e-01 -3.37269485e-01 1.05924033e-01
1.17340731e-02 -1.39834508e-01 1.91591844e-01 -8.34099770e-01
1.21427476e-02 -3.16244125e-01 6.20381057e-01 -8.24627936e-01
-1.86026588e-01 -1.01109195e+00 2.73811817e-01 3.95872235e-01
5.50109334e-03 2.84922689e-01 -7.69441808e-03 1.27405807e-01
-5.48572659e-01 -7.33461618e-01 1.22320020e+00 -1.42274380e-01
-1.01184201e+00 -4.10932720e-01 -1.08373165e+00 -5.08571148e-01
1.32731009e+00 -9.68373775e-01 -2.10095882e-01 -5.79897523e-01
-5.87753654e-01 -2.29377270e-01 5.96150517e-01 -3.33553761e-01
5.48686624e-01 -1.21109188e+00 -3.10929537e-01 1.68941632e-01
-2.44289815e-01 -2.75486410e-01 2.59033501e-01 8.70346844e-01
-1.28840876e+00 -1.18722282e-01 -5.08183956e-01 -6.23255551e-01
-1.34845340e+00 6.01067901e-01 4.59278405e-01 1.02072522e-01
-3.92543554e-01 6.10058486e-01 4.02956426e-01 5.02400935e-01
-8.61126259e-02 -4.23354089e-01 -6.48999155e-01 6.77247867e-02
7.44293451e-01 7.14344323e-01 -3.79682988e-01 -6.42564356e-01
-1.69850931e-01 7.32084990e-01 4.23322916e-01 -2.45796636e-01
8.41532767e-01 -2.76894122e-01 -8.36421013e-01 7.73764551e-01
1.30439746e+00 7.84707516e-02 -1.16690850e+00 5.48586965e-01
-2.31212795e-01 -6.20946050e-01 1.25140950e-01 -6.77951038e-01
-8.74999523e-01 5.39582253e-01 8.14900219e-01 1.19603550e+00
1.51409864e+00 1.56083837e-01 4.03537422e-01 -3.80609840e-01
-9.95570719e-02 -1.31754303e+00 -1.88545197e-01 2.84140021e-01
5.93212366e-01 -8.11006904e-01 1.92155212e-01 -5.15052676e-01
-7.97040701e-01 1.62634397e+00 -1.02175735e-01 -2.27192089e-01
9.38243091e-01 3.12753499e-01 1.44921616e-01 -3.55661571e-01
9.28953961e-02 -2.55151808e-01 3.81267637e-01 5.16507268e-01
6.39120221e-01 3.14011008e-01 -7.80969322e-01 -4.07055989e-02
-2.53346950e-01 -2.08404005e-01 5.69718063e-01 7.49173760e-01
-1.14207709e+00 -7.24259317e-01 -1.15321529e+00 4.60994810e-01
-7.91797936e-01 1.47595882e-01 -1.45389080e-01 8.65847647e-01
4.08880323e-01 1.02375364e+00 8.08775246e-01 -1.08251415e-01
2.87938267e-01 3.12837452e-04 4.67531890e-01 -3.01420316e-03
-1.33158043e-01 3.03217292e-01 -3.44392926e-01 -1.71902612e-01
-6.40901148e-01 -9.54409420e-01 -1.37149584e+00 -5.65037072e-01
-1.47057414e-01 2.43161395e-01 1.01191115e+00 7.38608122e-01
-7.11897254e-01 4.02459204e-01 7.08410084e-01 -6.83394670e-01
-3.67075354e-02 -1.06675637e+00 -1.25631332e+00 6.99601352e-01
6.24010526e-02 -4.66841787e-01 -7.78259397e-01 6.10198319e-01]
|
[10.862356185913086, -2.4072394371032715]
|
ad575bb3-785b-48d9-a9db-5d38ac165a08
|
latent-tree-learning-with-ordered-neurons
|
2010.04926
| null |
https://arxiv.org/abs/2010.04926v1
|
https://arxiv.org/pdf/2010.04926v1.pdf
|
Latent Tree Learning with Ordered Neurons: What Parses Does It Produce?
|
Recent latent tree learning models can learn constituency parsing without any exposure to human-annotated tree structures. One such model is ON-LSTM (Shen et al., 2019), which is trained on language modelling and has near-state-of-the-art performance on unsupervised parsing. In order to better understand the performance and consistency of the model as well as how the parses it generates are different from gold-standard PTB parses, we replicate the model with different restarts and examine their parses. We find that (1) the model has reasonably consistent parsing behaviors across different restarts, (2) the model struggles with the internal structures of complex noun phrases, (3) the model has a tendency to overestimate the height of the split points right before verbs. We speculate that both problems could potentially be solved by adopting a different training task other than unidirectional language modelling.
|
['Yian Zhang']
|
2020-10-10
| null |
https://aclanthology.org/2020.blackboxnlp-1.11
|
https://aclanthology.org/2020.blackboxnlp-1.11.pdf
|
emnlp-blackboxnlp-2020-11
|
['constituency-parsing']
|
['natural-language-processing']
|
[ 2.54891038e-01 7.63989508e-01 -1.27088591e-01 -5.14539778e-01
-1.06078041e+00 -8.13133180e-01 5.28941631e-01 2.99375236e-01
-3.03661913e-01 5.63567698e-01 6.54476583e-01 -8.85612011e-01
3.05696875e-01 -7.59949148e-01 -7.68021762e-01 -4.07839745e-01
-2.87404060e-02 5.69322288e-01 1.61500499e-01 4.43066191e-03
-6.17789738e-02 -6.51034713e-02 -1.02424395e+00 4.58396524e-01
6.47624910e-01 3.15907449e-01 3.94566774e-01 7.42118180e-01
-4.87073898e-01 1.01394022e+00 -5.33446252e-01 -5.30435562e-01
6.65592924e-02 -4.06641245e-01 -1.06023741e+00 -2.76557893e-01
3.36734593e-01 -2.95843303e-01 -1.22101143e-01 1.03342450e+00
3.18259522e-02 -3.19630742e-01 4.41518009e-01 -6.84936821e-01
-4.66978192e-01 1.18148434e+00 -3.51024956e-01 2.36023620e-01
-8.51474516e-03 1.49962947e-01 1.26458704e+00 -5.31972289e-01
7.82046616e-01 1.56000531e+00 7.88170993e-01 6.63480163e-01
-1.59213305e+00 -5.63724756e-01 4.53670740e-01 -3.66572499e-01
-7.00522959e-01 -4.37734276e-01 4.99271929e-01 -3.84715289e-01
1.26687634e+00 -3.34057771e-02 3.51791471e-01 1.23210609e+00
3.98507893e-01 8.27353597e-01 1.26969266e+00 -6.83989048e-01
1.45519361e-01 -1.78800628e-01 5.82238793e-01 6.93455815e-01
2.30737060e-01 1.23988196e-01 -4.56608325e-01 -3.78109813e-01
4.44936752e-01 -4.87353235e-01 1.81217089e-01 -1.38429552e-01
-1.00993443e+00 1.11416829e+00 2.33135507e-01 5.36615431e-01
-2.87445396e-01 2.09565848e-01 4.91987795e-01 -9.85315070e-03
5.69854319e-01 2.82177210e-01 -1.05705976e+00 -3.01576197e-01
-8.75260293e-01 1.24773443e-01 7.47267663e-01 8.65291238e-01
4.89154160e-01 2.01828375e-01 5.90202250e-02 9.24308956e-01
4.35399026e-01 -2.56101927e-03 5.54429889e-01 -1.07426250e+00
8.59884918e-01 3.97259325e-01 -9.65342745e-02 -3.47686380e-01
-5.56973696e-01 -2.43669897e-01 -2.37856030e-01 4.96786013e-02
8.92286897e-01 -3.86684418e-01 -1.06657064e+00 2.30314779e+00
-1.28110841e-01 -3.39938909e-01 1.52561128e-01 2.49806181e-01
6.51978791e-01 6.76806271e-01 9.00714219e-01 -2.34086677e-01
1.33246517e+00 -7.83057809e-01 -4.23696756e-01 -9.09726202e-01
9.47891891e-01 -6.41712964e-01 1.06232142e+00 1.85354594e-02
-1.34364820e+00 -5.03684342e-01 -9.47847426e-01 -3.20325941e-01
-1.33263811e-01 -1.01313107e-01 9.88849819e-01 7.58026183e-01
-1.00362957e+00 8.00905824e-01 -1.40119135e+00 -4.34579879e-01
3.65999974e-02 1.92159608e-01 -2.95846522e-01 2.17465714e-01
-9.74105000e-01 8.37255239e-01 5.83517432e-01 5.24217859e-02
-5.92903554e-01 -2.53678799e-01 -1.09677112e+00 2.32368961e-01
9.72122923e-02 -5.23941696e-01 1.58490121e+00 -9.61439073e-01
-1.39015698e+00 1.05472136e+00 -5.22622824e-01 -3.12788725e-01
1.54869899e-01 -2.48047307e-01 2.21388955e-02 -3.98314804e-01
3.58539402e-01 8.08934331e-01 4.15701985e-01 -1.27382100e+00
-6.22709036e-01 -5.65534532e-01 -2.79959798e-01 2.43857764e-02
1.61189407e-01 3.84691209e-01 -5.55679761e-02 -5.89851677e-01
7.01337636e-01 -8.57961595e-01 -2.08244994e-01 -6.79165483e-01
-4.13442045e-01 -3.32523465e-01 3.20018977e-01 -7.63132095e-01
1.12893355e+00 -1.95845580e+00 -6.58596307e-02 -2.80052423e-01
1.66235551e-01 9.65701416e-02 -2.16710001e-01 4.74882841e-01
-3.52641761e-01 5.11443615e-01 -3.20424706e-01 -6.44922733e-01
-1.30120339e-02 4.62913424e-01 -4.92521614e-01 8.25679749e-02
2.74259746e-01 1.04028857e+00 -9.01619136e-01 -3.84990782e-01
-1.64884150e-01 1.17653370e-01 -3.99239421e-01 1.03661492e-02
-4.00004953e-01 3.98559988e-01 -1.27835065e-01 5.18433571e-01
4.82556552e-01 -9.80503950e-03 8.08452785e-01 2.42568821e-01
-4.33322996e-01 1.23640537e+00 -6.72566593e-01 1.31746137e+00
-2.84246624e-01 6.48471653e-01 1.25526071e-01 -8.60720456e-01
5.57565629e-01 4.36020136e-01 6.78651109e-02 -5.12193620e-01
-1.44591900e-02 3.15587968e-01 4.70088929e-01 -2.02205539e-01
2.78840989e-01 -4.44642365e-01 -2.42145151e-01 5.66675961e-01
1.91522717e-01 2.66347658e-02 1.09616488e-01 9.59002897e-02
1.35379803e+00 2.58733779e-01 2.08084688e-01 -3.56001318e-01
-1.42210394e-01 1.45314410e-01 1.03357387e+00 9.47578609e-01
-1.48473680e-01 2.36279473e-01 8.39722037e-01 -6.17066443e-01
-1.09151709e+00 -1.30633318e+00 -1.64968446e-01 1.41210568e+00
-4.84892726e-01 -5.32127142e-01 -9.39048767e-01 -7.08445489e-01
-2.82813191e-01 1.11120844e+00 -4.27266419e-01 2.67242976e-02
-1.05946565e+00 -9.44089830e-01 7.31246591e-01 8.29622626e-01
1.85474649e-01 -1.53873754e+00 -8.03147018e-01 5.99024832e-01
-3.84789348e-01 -1.17766500e+00 4.80316356e-02 9.15588856e-01
-1.37448716e+00 -7.75866807e-01 -1.72184691e-01 -1.09062767e+00
5.27670801e-01 -1.78022757e-01 1.49168193e+00 -5.15122246e-03
3.13545525e-01 3.06507223e-03 -1.91889912e-01 -5.29006779e-01
-9.62805688e-01 2.33908772e-01 -1.86979249e-01 -7.84937799e-01
6.66148782e-01 -4.85474408e-01 -4.53700908e-02 -1.37125403e-01
-5.10369241e-01 1.70455351e-01 4.62521791e-01 9.50910628e-01
3.10281038e-01 2.02948153e-02 3.27579319e-01 -1.23392749e+00
8.14836025e-01 -4.79672253e-01 -6.62675917e-01 3.03903162e-01
-5.05987704e-01 3.10390204e-01 4.79469091e-01 -2.32918903e-01
-1.08559525e+00 1.33005351e-01 -2.07931027e-01 3.98513228e-01
-4.22257304e-01 3.51909399e-01 -8.60643163e-02 5.49490213e-01
4.08985257e-01 1.82234365e-02 -5.41898131e-01 -6.04362786e-01
3.41918439e-01 4.61835414e-01 3.32079977e-01 -9.02380049e-01
6.57563984e-01 1.01633966e-01 -3.35004419e-01 -6.01086497e-01
-9.37129140e-01 2.90504340e-02 -7.38266647e-01 3.86721224e-01
1.12750804e+00 -7.91924357e-01 -4.13470507e-01 4.13093567e-01
-1.53613615e+00 -9.37596262e-01 -1.33353487e-01 4.41388220e-01
-5.53406358e-01 3.32658380e-01 -1.29870808e+00 -9.30766404e-01
-3.27545166e-01 -1.11060190e+00 9.86951172e-01 -2.46760920e-02
-6.69156849e-01 -1.01439869e+00 1.29946783e-01 1.66222543e-01
1.59045562e-01 1.60795376e-02 1.79358768e+00 -6.64816260e-01
-4.48793858e-01 1.84812233e-01 -8.69387463e-02 1.77839741e-01
2.65470743e-01 -5.08415047e-03 -9.98278797e-01 -1.26733691e-01
3.07424873e-01 -3.73414040e-01 8.44719946e-01 6.11406088e-01
6.15899861e-01 -2.76281923e-01 -3.74503255e-01 4.85137373e-01
1.07747960e+00 3.33347231e-01 2.74834186e-01 3.78250152e-01
3.84895831e-01 9.53925610e-01 2.04603508e-01 -3.59482378e-01
7.12889314e-01 3.06938946e-01 1.43870249e-01 -1.06841028e-01
7.73169473e-02 -6.45977676e-01 5.26107013e-01 1.06732345e+00
3.51553172e-01 -1.91028610e-01 -1.20985854e+00 5.97570896e-01
-1.84629917e+00 -5.75204909e-01 -2.69968420e-01 1.88244164e+00
7.90295660e-01 8.44578922e-01 -1.46744484e-02 -7.92823061e-02
6.50055766e-01 2.20259994e-01 -4.72757101e-01 -1.00640440e+00
1.04885409e-02 3.92952263e-01 5.24584591e-01 6.59044027e-01
-9.42153931e-01 1.57815289e+00 7.73223591e+00 1.94100440e-01
-9.97043967e-01 2.62840062e-01 7.90469229e-01 2.72531062e-01
-2.47711852e-01 5.25877178e-01 -1.06477070e+00 3.84649098e-01
1.09821820e+00 2.91675180e-01 2.18238130e-01 6.93905771e-01
-4.93248180e-02 -1.52096108e-01 -1.26645350e+00 2.41737783e-01
-4.35342878e-01 -1.02467167e+00 -1.91221699e-01 1.72223628e-01
3.58463466e-01 4.52534586e-01 -2.75498271e-01 7.12400913e-01
9.54190969e-01 -1.02001524e+00 9.74128723e-01 -7.47310519e-02
4.05363917e-01 -2.70686090e-01 4.89736378e-01 8.31081033e-01
-1.03119838e+00 -1.56818286e-01 -3.99402112e-01 -4.05285090e-01
4.27072495e-01 2.27589399e-01 -7.55804121e-01 -1.43185601e-01
5.50706625e-01 9.84395519e-02 -6.81007445e-01 3.40246916e-01
-7.79559255e-01 1.25594938e+00 -4.86631632e-01 7.35015273e-02
3.82436752e-01 -3.66350673e-02 4.02825743e-01 1.35097849e+00
-7.73191899e-02 2.55457517e-02 5.43515049e-02 9.92384553e-01
1.04205526e-01 3.66112292e-02 -5.16421616e-01 -3.49315345e-01
5.62952340e-01 8.96339059e-01 -9.69704807e-01 -3.23321968e-01
-5.30003428e-01 6.67036057e-01 7.50119328e-01 4.26457793e-01
-4.90353465e-01 2.01118425e-01 3.85823876e-01 2.65857000e-02
2.28366256e-01 -6.09092891e-01 -6.53952003e-01 -1.06820965e+00
2.28488049e-03 -7.75569439e-01 5.29001117e-01 -7.74907470e-01
-1.14261401e+00 7.08327591e-01 5.95514067e-02 -3.90962690e-01
-5.13258815e-01 -7.95858860e-01 -7.56970465e-01 9.11838531e-01
-1.34695351e+00 -1.13950086e+00 4.12498772e-01 -3.66984047e-02
7.35243082e-01 3.08205690e-02 1.12170672e+00 -2.68184990e-01
-5.36053956e-01 4.57990646e-01 -1.20647930e-01 4.82775480e-01
3.10027778e-01 -1.41074169e+00 1.38109052e+00 1.00564873e+00
5.23343623e-01 1.04813802e+00 6.91180587e-01 -8.38820100e-01
-7.83551574e-01 -6.23723269e-01 1.35448480e+00 -8.99113953e-01
6.21222317e-01 -6.71073735e-01 -1.03108275e+00 1.41857064e+00
1.70029089e-01 -3.16487402e-01 6.86252356e-01 7.35749781e-01
-4.59527135e-01 4.76230711e-01 -8.21207941e-01 5.94852328e-01
1.02135253e+00 -3.74555826e-01 -9.40253198e-01 2.42017046e-01
7.76429057e-01 -3.85320157e-01 -5.46863079e-01 4.13244128e-01
4.43941742e-01 -1.02849841e+00 5.27356327e-01 -9.84126031e-01
5.03780007e-01 2.55745977e-01 -2.01307297e-01 -1.05612087e+00
-6.38786852e-01 -3.18480194e-01 6.71805739e-02 1.37532699e+00
7.01011002e-01 -6.14812434e-01 8.57472897e-01 9.27246273e-01
-9.42878425e-02 -7.06922710e-01 -9.04352248e-01 -6.53105140e-01
7.30413198e-01 -6.18040085e-01 3.75259966e-01 8.37307453e-01
4.88766320e-02 8.45464230e-01 5.73648885e-02 7.60706812e-02
4.36726362e-01 7.37931281e-02 3.98408234e-01 -1.22724426e+00
-4.85574394e-01 -3.40249002e-01 1.25268742e-01 -1.15184641e+00
5.22328734e-01 -9.00775552e-01 2.17139363e-01 -1.54699528e+00
7.57132098e-02 -5.81100225e-01 -4.76757251e-02 9.00832891e-01
-3.00222367e-01 -3.05319488e-01 3.09480131e-01 1.30581617e-01
-2.05891833e-01 8.50587413e-02 6.01037204e-01 3.27634633e-01
-2.66066402e-01 -8.23280513e-02 -8.66330087e-01 1.33383882e+00
9.24410999e-01 -1.01790226e+00 -1.20682947e-01 -9.51925337e-01
3.06028396e-01 1.45628542e-01 1.76692661e-02 -5.76868832e-01
2.64601987e-02 9.09806578e-04 3.87988061e-01 -4.15942520e-01
1.23877414e-01 -4.12303180e-01 -1.67023107e-01 5.59822559e-01
-4.40446675e-01 3.17066401e-01 3.83279294e-01 2.36516118e-01
5.63892610e-02 -5.54306924e-01 6.46568239e-01 -7.26404428e-01
-3.34068745e-01 -3.38005990e-01 -8.53750825e-01 2.42757842e-01
3.34557652e-01 -1.30621016e-01 -2.80344516e-01 -2.06676766e-01
-7.03841507e-01 9.18880329e-02 3.61665398e-01 4.60423470e-01
-2.04710469e-01 -8.36094081e-01 -5.49772859e-01 1.51744112e-01
-1.76328197e-01 9.83870476e-02 -2.66514689e-01 1.39690593e-01
-3.22131127e-01 5.35388529e-01 2.47470900e-01 -3.89915049e-01
-1.10216784e+00 2.17004314e-01 3.09484720e-01 -6.29858911e-01
-7.63403952e-01 9.65855122e-01 3.90891612e-01 -7.34493613e-01
2.12788492e-01 -5.38791537e-01 1.03504770e-01 -1.70937151e-01
1.11545220e-01 4.88239992e-03 -8.21930096e-02 -6.33047938e-01
-3.79071355e-01 3.13736886e-01 -2.69440204e-01 -1.34864643e-01
1.39646947e+00 7.56408051e-02 -5.53880893e-02 6.57944262e-01
7.81281114e-01 1.80240884e-01 -1.30499625e+00 -1.50770605e-01
6.75006688e-01 1.50770575e-01 -1.38025060e-01 -8.33549142e-01
-5.54314554e-01 1.01654589e+00 3.21021497e-01 2.81399548e-01
5.40047467e-01 1.94678694e-01 1.03662622e+00 1.83056206e-01
4.53346550e-01 -1.11992514e+00 -4.01155710e-01 8.41528177e-01
3.35503042e-01 -1.15476620e+00 -1.57941520e-01 -3.18155617e-01
-3.99969667e-01 1.17400169e+00 5.55743754e-01 4.61946093e-02
3.88612926e-01 5.73020458e-01 3.55108827e-01 -2.34969810e-01
-1.07493877e+00 -2.09791400e-02 -2.42481709e-01 3.56156230e-01
9.81952310e-01 3.24447840e-01 -4.50322688e-01 7.74271369e-01
-6.79101110e-01 -4.22099352e-01 2.51228720e-01 9.75151300e-01
-3.86161804e-01 -1.44394588e+00 -1.32544085e-01 1.59175828e-01
-7.61382639e-01 -5.04927039e-01 -5.00090480e-01 6.69238687e-01
1.33545429e-01 9.90992665e-01 2.28715651e-02 6.56491965e-02
2.28587776e-01 7.39465654e-01 5.02028584e-01 -1.22866726e+00
-6.68210566e-01 1.94540963e-01 2.47929841e-01 -4.31756556e-01
-5.61189279e-02 -8.57186675e-01 -1.40416682e+00 1.66843235e-01
-3.96565974e-01 9.17624459e-02 7.50121415e-01 1.09501505e+00
1.37118191e-01 5.03662109e-01 1.73176117e-02 -5.93696833e-01
-7.42507517e-01 -1.20430958e+00 -2.32215255e-01 1.86382681e-01
9.85867977e-02 -3.01871747e-01 -3.30627978e-01 5.10333106e-02]
|
[10.420512199401855, 9.518903732299805]
|
63b0195c-f121-45e5-9b43-3fdc486fc617
|
a-harmonic-based-fault-detection-algorithm
|
2303.15957
| null |
https://arxiv.org/abs/2303.15957v1
|
https://arxiv.org/pdf/2303.15957v1.pdf
|
A Harmonic-based Fault detection algorithm for Microgrids
|
The trend toward Microgrids (MGs) is significantly increasing by employing Distributed Generators (DGs) which leads to new challenges, especially in the fault detection. This paper proposes an algorithm based on the Total Harmonic Distortion (THD) of the grid voltages to detect the events of faults in MGs. The algorithm uses the THD together with the estimate amplitude voltages and the zero-sequence component for the detection and identification of the faults. The performance is evaluated by using MATLAB/Simulink simulations to validate the capability for detecting different fault types in the least possible time.
|
['Josep. M. Guerrero', 'Jorge. El mariachet', 'Jose Matas', 'Wael Al Hanaineh']
|
2023-03-28
| null | null | null | null |
['fault-detection']
|
['miscellaneous']
|
[-5.18901646e-01 -5.46459675e-01 5.22909403e-01 2.24887103e-01
-2.17145517e-01 -9.68309343e-01 5.65854013e-01 3.07551384e-01
5.97223878e-01 1.03171825e+00 -1.24833375e-01 -2.32362691e-02
-3.06391358e-01 -7.26748765e-01 1.36981621e-01 -1.13371551e+00
-6.07205451e-01 -1.35969277e-02 1.01584621e-01 -1.21849582e-01
3.08776796e-01 7.19015121e-01 -1.50814927e+00 -2.83410639e-01
9.86878872e-01 8.89716804e-01 1.54737800e-01 3.17891240e-01
7.28311241e-01 2.53933936e-01 -1.64764726e+00 5.24068654e-01
4.50211763e-01 -7.92551398e-01 -1.67328373e-01 1.01874642e-01
-6.67207956e-01 -5.57393789e-01 -3.07327867e-01 1.48948753e+00
7.41360545e-01 1.91428468e-01 8.69975448e-01 -1.75200295e+00
-9.64494348e-02 3.88941944e-01 -5.81074119e-01 8.39442730e-01
6.87986851e-01 2.01796144e-01 3.00730199e-01 -9.59062517e-01
2.47252315e-01 7.76066422e-01 2.77189106e-01 -3.96217674e-01
-7.86587417e-01 -3.87363791e-01 -4.74114001e-01 8.10647547e-01
-1.70812178e+00 4.98381913e-01 8.86481404e-01 -4.08960164e-01
1.12971246e+00 3.53390545e-01 7.80726612e-01 6.17919713e-02
6.16587520e-01 2.87401259e-01 1.23803937e+00 -4.53336805e-01
5.66238940e-01 -2.63764918e-01 1.89549297e-01 2.20644828e-02
7.85245657e-01 -7.71274865e-02 -1.88710228e-01 -6.29650801e-02
4.05578047e-01 -4.40239638e-01 -9.65919197e-01 2.36837611e-01
-6.87188029e-01 8.90272081e-01 -5.27972868e-03 9.16802585e-01
-4.58942950e-01 -4.23978955e-01 4.63508487e-01 4.29035306e-01
3.90028954e-01 1.91729307e-01 -2.00756550e-01 -2.11564139e-01
-9.47679639e-01 2.57015139e-01 7.60572791e-01 7.96128511e-01
4.16384898e-02 1.11201358e+00 3.86559635e-01 3.84261869e-02
4.00474310e-01 6.24040067e-01 7.39574194e-01 -1.79006428e-01
-2.39543617e-01 7.25870430e-01 2.78519243e-01 -9.56241965e-01
-4.36231941e-01 -8.34337294e-01 -7.44525075e-01 6.31360531e-01
1.26350418e-01 -5.18858314e-01 -5.14129877e-01 1.03256631e+00
2.63603419e-01 -4.07090671e-02 -8.10662135e-02 7.34587729e-01
2.09977239e-01 1.22491753e+00 -7.22046852e-01 -8.79256308e-01
1.23148501e+00 -3.25845122e-01 -1.32253325e+00 5.11462688e-01
3.48073363e-01 -1.02857816e+00 2.33813718e-01 8.00191760e-01
-8.86581540e-01 -4.96711284e-01 -1.57500446e+00 7.65785813e-01
-5.69723427e-01 2.54724711e-01 -2.83164501e-01 6.13895655e-01
-1.07901037e+00 4.40922529e-01 -3.85468185e-01 -1.93061560e-01
-4.91206825e-01 2.70913512e-01 -5.91469333e-02 5.76229274e-01
-1.48297095e+00 1.42672563e+00 6.41243935e-01 3.51133823e-01
-6.70284092e-01 -5.73512733e-01 -5.92273891e-01 -2.75915544e-02
-1.55324593e-01 -1.05776548e-01 8.77138853e-01 -7.68448830e-01
-1.10737228e+00 -1.43270969e-01 7.56433979e-02 -7.85500288e-01
3.48217905e-01 1.24774702e-01 -9.06335950e-01 5.65631568e-01
4.03130837e-02 -5.99397540e-01 6.52914047e-01 -7.63780475e-01
-8.21193099e-01 -3.56772333e-01 -5.11143625e-01 1.42444313e-01
3.60306017e-02 -1.83062479e-01 8.34706903e-01 -8.68952155e-01
2.38852859e-01 -1.51089177e-01 3.74973826e-02 -9.21110511e-01
-3.66541356e-01 -4.97356445e-01 1.47967398e+00 -1.04598773e+00
1.22833812e+00 -1.98837066e+00 -1.26534566e-01 5.75617075e-01
-3.87589037e-01 3.42937082e-01 6.28321111e-01 1.09858274e+00
-1.62259191e-01 -4.11658794e-01 9.93399769e-02 5.30912280e-01
-2.91777384e-02 3.12566847e-01 -2.33876124e-01 9.43112850e-01
1.57862708e-01 2.62143165e-01 -8.97461832e-01 3.94134134e-01
6.70569718e-01 3.31171602e-01 4.29351062e-01 3.00904423e-01
3.70982200e-01 4.55061495e-01 1.37239724e-01 2.12474629e-01
8.69397640e-01 3.77744585e-01 1.64967760e-01 -4.75487411e-01
-4.90076870e-01 2.53758758e-01 -1.77147853e+00 6.09639764e-01
1.47979185e-01 5.79439938e-01 4.52142060e-02 -1.24932075e+00
1.04175234e+00 7.81122446e-01 4.15146261e-01 -7.51822054e-01
1.02794781e-01 4.31412041e-01 2.34201223e-01 -3.11259240e-01
-1.95146412e-01 3.56262982e-01 3.24968040e-01 2.54309237e-01
6.20173812e-01 -2.69079149e-01 7.95822442e-01 1.20526798e-01
3.78299564e-01 -1.72393128e-01 6.16086006e-01 -1.04356480e+00
8.33497941e-01 -1.88136756e-01 7.16656268e-01 -1.06268547e-01
1.78927071e-02 -1.01194330e-01 5.11963546e-01 -1.28634706e-01
-6.61863029e-01 -8.63904178e-01 -3.43212575e-01 -4.28070009e-01
1.63723871e-01 -3.51180509e-02 -7.15252876e-01 -2.94668972e-01
1.08593322e-01 1.09532416e+00 -5.92967682e-02 -3.22482377e-01
-3.40442717e-01 -1.07163155e+00 1.22724041e-01 3.68604273e-01
5.03330290e-01 -3.75141740e-01 -7.37509489e-01 4.01877254e-01
2.92616546e-01 -6.14739656e-01 9.87077365e-04 5.33115625e-01
-6.04658365e-01 -1.31313837e+00 -7.24740982e-01 -8.74132335e-01
1.04247665e+00 3.43843177e-02 6.43739998e-01 -1.36929870e-01
-5.32213748e-01 1.30899057e-01 -4.96395141e-01 -2.43407220e-01
-3.68203253e-01 -5.80258369e-01 3.40626121e-01 -2.31421798e-01
1.16699710e-01 -7.39313126e-01 -4.76775557e-01 3.69013339e-01
-6.23725057e-01 -3.34064782e-01 1.94253579e-01 6.86170399e-01
-8.91324133e-02 1.26784182e+00 1.38598812e+00 -2.42002413e-01
8.09015274e-01 -7.72906184e-01 -1.43713272e+00 -1.87679648e-01
-9.65934873e-01 -5.06691039e-01 1.10260189e+00 -2.97278706e-02
-7.84068167e-01 -4.42726582e-01 -1.51976302e-01 1.47221208e-01
-3.73178989e-01 4.64968622e-01 -5.46358883e-01 -3.36439729e-01
1.14840925e-01 4.75383669e-01 -6.98091611e-02 -5.92931390e-01
-8.17202777e-02 7.15290487e-01 5.95377743e-01 3.43859732e-01
1.13359237e+00 -1.34957686e-01 2.99384356e-01 -8.18517447e-01
2.39550799e-01 -4.47062194e-01 -5.23356974e-01 -5.41204035e-01
1.47966862e-01 -8.80396008e-01 -8.36070836e-01 1.15378463e+00
-1.10794353e+00 2.31383264e-01 -5.51129356e-02 6.06596529e-01
4.90147956e-02 5.58275640e-01 -8.12219322e-01 -8.68967175e-01
-5.98085642e-01 -1.26944780e+00 1.37812108e-01 5.47585070e-01
6.20808154e-02 -1.25614154e+00 -3.36484522e-01 -4.57254857e-01
3.57361823e-01 8.02178085e-01 1.04293787e+00 -6.80135369e-01
-3.57692361e-01 -2.42715970e-01 3.85498047e-01 8.29580009e-01
6.00588083e-01 9.52629466e-03 -3.77414554e-01 -6.08346403e-01
7.96084404e-01 5.96564949e-01 -5.14682174e-01 1.86079860e-01
2.93797348e-02 -4.02745187e-01 5.33372834e-02 1.78164124e-01
2.26101351e+00 1.17467427e+00 4.54013497e-01 1.90899312e-01
1.48815960e-01 -3.92214581e-02 7.44055271e-01 7.51603067e-01
1.26004368e-01 4.64034140e-01 3.38805228e-01 9.81231257e-02
1.05381034e-01 3.22754890e-01 3.26875061e-01 1.03535342e+00
2.46037200e-01 -5.07912755e-01 -5.89429796e-01 7.65986741e-01
-1.36777997e+00 -6.06241167e-01 -8.66974950e-01 1.96854186e+00
5.22044718e-01 8.65242779e-02 6.71498030e-02 1.33189690e+00
8.65658879e-01 -1.42521128e-01 -1.67791024e-01 -5.97108483e-01
-4.78597909e-01 3.39920878e-01 5.63591897e-01 5.64863265e-01
-5.49634874e-01 -4.36225086e-01 6.22179651e+00 8.00440967e-01
-1.28327465e+00 -1.32904574e-01 2.64619619e-01 5.50597489e-01
1.75296366e-01 -2.31627956e-01 -4.91959959e-01 1.07878959e+00
1.14096618e+00 -1.21087646e+00 2.04473361e-01 4.82923895e-01
7.01775014e-01 -8.20459247e-01 -4.86400306e-01 6.20670974e-01
2.18798786e-01 -5.55027485e-01 2.03627273e-01 -2.01507613e-01
1.11102474e+00 -4.21937853e-01 -5.04531384e-01 -4.50688601e-01
-1.28071234e-01 -4.01833683e-01 6.82339251e-01 2.12287962e-01
-3.87393758e-02 -1.41710627e+00 1.34590626e+00 2.53832132e-01
-1.09942079e+00 -1.23534478e-01 -1.56062588e-01 -3.79521757e-01
6.79645419e-01 1.00049388e+00 -1.09526014e+00 1.39522851e+00
4.03830469e-01 4.49243635e-01 -3.52840900e-01 1.38934696e+00
-7.69089580e-01 9.31792438e-01 -3.67708772e-01 7.85743222e-02
2.11944547e-03 -5.89317858e-01 6.71895862e-01 8.18844080e-01
5.40935218e-01 9.31316391e-02 1.12123080e-01 5.64218879e-01
4.44820762e-01 -1.47306556e-02 -2.29217038e-01 4.47013587e-01
8.34958196e-01 1.18036270e+00 -7.40508854e-01 -4.22458857e-01
-3.74300838e-01 6.88676238e-01 -8.63956630e-01 4.14320737e-01
-4.64251429e-01 -9.34224308e-01 3.16558838e-01 -1.78202540e-01
1.40371080e-03 -3.08150917e-01 -3.92385334e-01 -5.34710467e-01
1.90727994e-01 -7.34747887e-01 4.39633310e-01 -8.04946482e-01
-1.16752064e+00 4.44866806e-01 1.64441302e-01 -1.56622076e+00
-9.38921392e-01 -2.60044634e-01 -9.81982172e-01 1.53444099e+00
-1.37440181e+00 -5.17631352e-01 -6.03417568e-02 6.73866332e-01
6.35771751e-01 -8.96772817e-02 7.63315201e-01 4.52813685e-01
-4.31472123e-01 1.22675836e-01 4.85221416e-01 4.62450385e-02
-2.40728352e-02 -1.59684432e+00 5.30264005e-02 1.50374949e+00
-4.26194400e-01 -7.80637637e-02 1.25614011e+00 -5.73231280e-01
-1.43193793e+00 -5.41517794e-01 7.20601797e-01 7.12386250e-01
7.03304589e-01 -2.73700744e-01 -6.13139808e-01 4.52059299e-01
1.06037390e+00 -6.80016056e-02 2.38415509e-01 -1.23900592e+00
3.10600817e-01 -2.78374672e-01 -1.79329431e+00 -2.06500739e-02
-1.68627843e-01 -3.33998799e-01 -8.32472980e-01 1.79781333e-01
-1.28649473e-01 -2.22763479e-01 -1.27625883e+00 4.67947483e-01
-2.97227204e-01 -6.73418880e-01 3.34251672e-01 4.42367524e-01
-6.45748675e-01 -1.16428626e+00 2.34103039e-01 -2.11106205e+00
-1.83196366e-01 -8.71082425e-01 -1.82201952e-01 1.39816654e+00
-7.38268048e-02 -1.20258391e+00 -6.20018393e-02 -2.97111779e-01
-7.77195916e-02 -3.15206259e-01 -1.02550900e+00 -8.81167054e-01
-3.06166440e-01 4.25416231e-01 6.17672145e-01 1.06173229e+00
7.10541010e-01 2.07355767e-02 6.44814074e-02 9.09634769e-01
8.99304271e-01 -4.64262106e-02 5.74853532e-02 -1.13579059e+00
2.04132095e-01 -1.77021891e-01 -1.06200075e+00 -1.80005118e-01
-4.80516315e-01 -1.72326311e-01 -2.54806429e-01 -1.74253440e+00
-8.77639651e-01 5.29236019e-01 -4.91117835e-02 2.15101421e-01
-1.59151286e-01 1.50836587e-01 -1.55972704e-01 -5.79202510e-02
3.09706926e-01 2.36672908e-01 3.93656403e-01 1.42658234e-01
4.32164341e-01 2.54471134e-02 3.09656471e-01 5.81806779e-01
1.10652959e+00 8.74600485e-02 -7.13200629e-01 3.27065922e-02
-3.85573983e-01 3.68099838e-01 -1.62682179e-02 -1.50309050e+00
2.96982408e-01 3.37949961e-01 5.61926961e-01 -1.31243575e+00
-4.70952213e-01 -9.35353458e-01 7.07891047e-01 1.10876143e+00
6.11783028e-01 6.94495320e-01 1.41092286e-01 1.67376399e-02
-6.43285096e-01 -3.84872705e-01 7.35447049e-01 1.27069592e-01
-7.01491475e-01 -4.82929111e-01 -1.16616225e+00 -5.43556809e-01
1.64303899e+00 -1.30274311e-01 -5.06144822e-01 -3.63996744e-01
-5.13755560e-01 3.32879037e-01 4.08395112e-01 2.51946986e-01
4.08195227e-01 -1.11631548e+00 -7.03788698e-01 5.20799041e-01
-5.37176788e-01 -4.53877300e-01 1.81425571e-01 6.31544113e-01
-7.88428783e-01 5.32229304e-01 -5.12740314e-01 -3.34134340e-01
-1.08656228e+00 4.31605339e-01 6.61414921e-01 1.71414271e-01
-2.50065118e-01 1.24119937e-01 -5.93504369e-01 7.54587293e-01
-5.72747625e-02 -1.42570913e-01 -4.23629224e-01 3.93336892e-01
8.11859667e-01 1.02955151e+00 6.77383840e-01 -5.82124233e-01
-1.92005366e-01 1.90093085e-01 4.75617766e-01 3.06893326e-02
1.16909790e+00 -3.07528466e-01 -5.98519981e-01 5.66422403e-01
1.08261478e+00 -8.47903490e-02 -9.27129507e-01 5.97139239e-01
1.89629108e-01 -3.44269186e-01 5.34094125e-02 -1.15740156e+00
-1.28680360e+00 2.84347981e-01 6.54762268e-01 1.06970370e+00
1.34883201e+00 -8.64067495e-01 4.64743137e-01 -5.39434433e-01
1.06296849e+00 -1.00989795e+00 -8.52050006e-01 5.88608347e-02
5.29716372e-01 -1.65748984e-01 5.14895879e-02 -5.27265131e-01
1.39939159e-01 1.48053133e+00 7.50258416e-02 -6.95430279e-01
8.99256110e-01 7.98972130e-01 -6.89885989e-02 2.15930387e-01
-6.64648890e-01 9.58528742e-02 -1.38685107e-01 9.94660497e-01
3.78462911e-01 1.00562580e-01 -1.40599215e+00 2.45740667e-01
-8.93337727e-02 -1.60954893e-01 1.17535317e+00 1.05067325e+00
-4.15662199e-01 -1.08932555e+00 -8.53472233e-01 4.08835649e-01
-8.98976862e-01 3.67211372e-01 2.52719611e-01 6.74946129e-01
4.77277994e-01 1.35946608e+00 7.09776953e-02 8.28691646e-02
4.89642471e-01 7.71405781e-03 2.67700791e-01 -5.38183525e-02
-6.70839369e-01 6.70899227e-02 -2.62876213e-01 4.84174900e-02
-1.45304903e-01 -6.76247716e-01 -1.54351473e+00 -7.04241395e-02
-7.23994970e-01 1.06993115e+00 8.73014808e-01 8.52290988e-01
-2.31016497e-03 5.75121641e-01 1.28673375e+00 -3.92343581e-01
-6.03029430e-01 -1.07153630e+00 -1.29626930e+00 3.54336113e-01
1.38495743e-01 -6.95898831e-01 -1.04922426e+00 -1.06568091e-01]
|
[5.917745113372803, 2.5414061546325684]
|
58033dc7-caa2-415e-8024-97cd7653b6e8
|
fairness-and-diversity-in-recommender-systems
|
2307.04644
| null |
https://arxiv.org/abs/2307.04644v1
|
https://arxiv.org/pdf/2307.04644v1.pdf
|
Fairness and Diversity in Recommender Systems: A Survey
|
Recommender systems are effective tools for mitigating information overload and have seen extensive applications across various domains. However, the single focus on utility goals proves to be inadequate in addressing real-world concerns, leading to increasing attention to fairness-aware and diversity-aware recommender systems. While most existing studies explore fairness and diversity independently, we identify strong connections between these two domains. In this survey, we first discuss each of them individually and then dive into their connections. Additionally, motivated by the concepts of user-level and item-level fairness, we broaden the understanding of diversity to encompass not only the item level but also the user level. With this expanded perspective on user and item-level diversity, we re-interpret fairness studies from the viewpoint of diversity. This fresh perspective enhances our understanding of fairness-related work and paves the way for potential future research directions. Papers discussed in this survey along with public code links are available at https://github.com/YuyingZhao/Awesome-Fairness-and-Diversity-Papers-in-Recommender-Systems .
|
['Tyler Derr', 'Charu Aggarwal', 'Xueqi Cheng', 'Yunchao Liu', 'Yu Wang', 'Yuying Zhao']
|
2023-07-10
| null | null | null | null |
['fairness', 'recommendation-systems', 'fairness']
|
['computer-vision', 'miscellaneous', 'miscellaneous']
|
[-3.92285854e-01 -1.37771308e-01 -7.30422676e-01 -6.27729952e-01
-1.14392184e-01 -6.04385257e-01 2.48959467e-01 2.52576381e-01
-2.84917772e-01 7.51455247e-01 5.17017841e-01 -4.36675400e-01
-4.18430179e-01 -6.42129421e-01 1.52241513e-01 -2.71988750e-01
1.12129360e-01 -1.26879085e-02 -2.98733175e-01 -6.27625227e-01
8.01619351e-01 1.76625699e-01 -1.65254104e+00 5.66181540e-02
1.51957643e+00 5.12699842e-01 -1.69536337e-01 3.58998269e-01
-1.55789196e-01 6.89578652e-01 -5.29397488e-01 -1.00974286e+00
3.76524985e-01 -7.40270734e-01 -7.70846426e-01 -3.47146392e-01
5.15497684e-01 -5.78743458e-01 -4.57440615e-01 9.51260448e-01
7.73841441e-01 5.76762855e-01 4.01829034e-01 -1.73317206e+00
-1.30352938e+00 6.51672721e-01 -4.56143826e-01 4.68111932e-01
3.41138601e-01 -7.64182061e-02 1.39432013e+00 -5.24365425e-01
1.86306909e-01 1.19438565e+00 3.53305608e-01 6.09504640e-01
-1.02978182e+00 -7.74314284e-01 5.83583891e-01 3.35579693e-01
-1.04754138e+00 -6.02649212e-01 2.68319994e-01 -3.55096102e-01
5.40830851e-01 7.61718452e-01 4.46053803e-01 5.29868364e-01
-1.00062199e-01 8.79653811e-01 1.03291178e+00 -4.39678878e-01
1.36573568e-01 3.90693247e-01 7.15934038e-01 4.83294278e-02
7.52219737e-01 1.67372510e-01 -5.81633627e-01 -3.00400764e-01
5.26009619e-01 2.65307873e-01 -2.29849741e-01 -2.52140760e-01
-5.83624184e-01 1.04674053e+00 2.99978435e-01 1.66649595e-01
-3.34563941e-01 -2.62760520e-01 4.97876823e-01 6.20961964e-01
6.20798588e-01 6.28554463e-01 -2.80328542e-02 -1.85117632e-01
-1.00639057e+00 5.22771358e-01 1.08507252e+00 8.15052927e-01
4.60066199e-01 5.53185493e-02 -7.03663349e-01 1.12749207e+00
2.05554336e-01 3.03371340e-01 1.43217117e-01 -1.30413985e+00
4.68642078e-02 3.00883353e-01 3.39007288e-01 -1.03128505e+00
-2.73156494e-01 -6.41996622e-01 -5.46904027e-01 2.14023143e-01
4.87417936e-01 -2.13801742e-01 1.12935035e-02 1.94171429e+00
2.59126306e-01 -4.71251935e-01 -4.56838995e-01 1.24694669e+00
7.67039120e-01 1.41260430e-01 3.92254353e-01 -4.65187222e-01
1.26990306e+00 -1.12155819e+00 -7.60878921e-01 1.26185820e-01
3.47087264e-01 -7.84807682e-01 1.14853179e+00 3.87699574e-01
-1.29096508e+00 -3.03466141e-01 -6.90921843e-01 -1.69115409e-01
-1.84801936e-01 -1.99927747e-01 6.91696167e-01 1.32264853e+00
-1.07654667e+00 4.74815428e-01 3.80320512e-02 -6.18346453e-01
4.51007396e-01 8.32842737e-02 2.74957240e-01 -1.29840314e-01
-1.41695631e+00 1.18421435e+00 -3.97030771e-01 -4.01324779e-01
-3.78671825e-01 -9.51666534e-01 -2.72735834e-01 3.04460615e-01
4.99995410e-01 -9.37640250e-01 1.40793371e+00 -8.04398954e-01
-1.42739701e+00 4.91388142e-01 -2.00320762e-02 -1.78622231e-01
6.06327832e-01 -3.64133656e-01 -4.38191563e-01 -2.91982830e-01
1.65397257e-01 1.50662884e-01 1.27305478e-01 -1.16663694e+00
-8.57202172e-01 -2.62052834e-01 5.63583672e-01 7.49662101e-01
-6.81672812e-01 4.52910244e-01 2.39963159e-02 -6.30820274e-01
-6.52930081e-01 -5.64914167e-01 -2.02125654e-01 7.95206055e-02
-1.09893635e-01 -3.54254186e-01 2.45024011e-01 -3.83272201e-01
1.98798859e+00 -1.96159506e+00 -4.24521297e-01 1.84638016e-02
4.89208132e-01 3.34985256e-01 -9.86608341e-02 7.86206186e-01
1.71571061e-01 4.18801814e-01 3.12217504e-01 -2.36621216e-01
4.34864312e-01 -1.60392016e-01 -8.65550116e-02 6.00023806e-01
-5.08034587e-01 6.03756368e-01 -9.83380377e-01 -3.13059725e-02
2.85060823e-01 3.89631003e-01 -8.52773070e-01 1.99719742e-01
2.03697413e-01 3.21888149e-01 -2.42087483e-01 4.28710461e-01
8.77471387e-01 -7.86970109e-02 1.23606451e-01 1.99661657e-01
-3.83205622e-01 5.12366712e-01 -9.98452485e-01 1.10540128e+00
-4.62766886e-01 2.45447919e-01 1.04810283e-01 -8.35453808e-01
6.58398271e-01 1.44631490e-01 5.07198751e-01 -1.06678450e+00
1.05972692e-01 7.84033462e-02 2.86133766e-01 -1.71413109e-01
1.00053096e+00 -2.38497853e-01 -8.13028514e-02 8.57986987e-01
-3.61931086e-01 3.68824691e-01 4.98210758e-01 5.38581967e-01
6.79588258e-01 -2.50989527e-01 6.51291013e-01 -6.46842062e-01
4.55931813e-01 -3.44633520e-01 4.99563605e-01 9.65512514e-01
-7.06586361e-01 1.41633779e-01 3.83315623e-01 -2.08184436e-01
-6.70859516e-01 -9.57065940e-01 -3.09926510e-01 1.95374274e+00
5.57289124e-01 -6.50957465e-01 -4.25162703e-01 -5.71681201e-01
5.12840271e-01 1.25407016e+00 -5.77542901e-01 -9.70204249e-02
9.69949514e-02 -6.14804029e-01 3.07770669e-01 1.09787375e-01
1.49828598e-01 -6.27678931e-01 -5.38931012e-01 -1.58720613e-01
-4.34339523e-01 -4.75176513e-01 -7.84035444e-01 -5.30442178e-01
-8.29508424e-01 -9.91947472e-01 -7.40466058e-01 -2.31361352e-02
2.79884249e-01 9.46087778e-01 1.37621820e+00 5.37374139e-01
-4.08523064e-03 6.81411743e-01 -4.91952926e-01 -4.40531462e-01
-8.56190920e-02 -2.67271131e-01 3.24766576e-01 -3.09013188e-01
5.54679573e-01 -4.97700423e-01 -9.44706321e-01 7.02309012e-01
-5.35859704e-01 -3.41402233e-01 1.47081733e-01 4.09799308e-01
-8.50856602e-02 -2.30888262e-01 1.15514708e+00 -1.11018991e+00
1.33420706e+00 -9.84722912e-01 -1.48130953e-01 2.17705160e-01
-1.20797896e+00 -5.64081252e-01 5.74028611e-01 -5.32631343e-03
-1.27619028e+00 -9.77828979e-01 5.36564440e-02 2.52363294e-01
-6.28230646e-02 3.70669872e-01 -6.22139089e-02 5.04976176e-02
9.13823187e-01 -2.70613015e-01 -9.11638979e-03 -3.91133249e-01
7.43462920e-01 7.34239459e-01 9.99734476e-02 -7.44836271e-01
3.86798561e-01 5.92443347e-02 -4.94529307e-01 -6.83108687e-01
-8.68975818e-01 -5.35696387e-01 -6.86434209e-02 -4.24676389e-01
2.25818738e-01 -8.10909867e-01 -9.37393188e-01 -7.14171818e-03
-5.09752572e-01 -2.08273977e-01 -3.24191958e-01 3.74393046e-01
-4.41784471e-01 6.05844200e-01 -6.93806291e-01 -1.31185079e+00
-4.53198165e-01 -7.23066449e-01 6.94665313e-02 7.14074910e-01
-4.55848783e-01 -9.49183404e-01 1.18507139e-01 6.78739071e-01
9.00680125e-01 -4.10145521e-01 6.46490693e-01 -7.09647834e-01
-2.52473980e-01 -1.54208273e-01 -3.12958181e-01 9.30139869e-02
-2.46254867e-03 -9.53573287e-02 -7.50158131e-01 -2.61729747e-01
-1.71760917e-01 -9.64789316e-02 5.44794261e-01 6.18511140e-01
9.62950051e-01 -2.36158699e-01 3.30805816e-02 1.04524821e-01
1.24188232e+00 -8.67824182e-02 5.13898253e-01 4.52414155e-01
1.69547796e-01 7.89389849e-01 8.70779932e-01 1.12261927e+00
8.07174802e-01 7.60168850e-01 2.39544854e-01 1.13880850e-01
-1.37296200e-01 5.33732250e-02 1.80482343e-01 4.28558350e-01
-3.79347146e-01 -6.12464070e-01 -4.53753859e-01 2.44514465e-01
-2.02169085e+00 -1.17539704e+00 -2.93287843e-01 2.69514537e+00
6.00522399e-01 -3.26345682e-01 8.81320894e-01 -4.08312939e-02
1.07105362e+00 1.40099963e-02 -4.53453660e-01 -7.85288751e-01
1.33782834e-01 -3.30595583e-01 2.73793995e-01 7.52346933e-01
-6.45496249e-01 6.93799496e-01 6.81791973e+00 5.24480820e-01
-5.98917723e-01 1.17015250e-01 7.04805672e-01 -5.51667035e-01
-7.22366035e-01 2.59778481e-02 -5.21019101e-01 6.64469659e-01
9.30913389e-01 -9.21679497e-01 6.35267079e-01 7.56114602e-01
5.75313866e-01 -3.40100855e-01 -8.61763716e-01 8.11214149e-01
4.73904703e-03 -7.02369869e-01 -1.18847191e-01 2.62214631e-01
6.04456723e-01 -2.38817945e-01 3.12456369e-01 5.65839112e-01
5.04101813e-01 -8.86681736e-01 4.97567266e-01 4.89247978e-01
5.04334450e-01 -7.65766203e-01 4.52820867e-01 2.70999521e-01
-7.61654019e-01 -1.65938631e-01 -5.21038592e-01 -7.85031557e-01
2.44290695e-01 8.90828133e-01 8.76592472e-02 6.16614878e-01
6.34867311e-01 6.20397091e-01 -2.26968274e-01 1.38245738e+00
-7.09436461e-02 4.84372914e-01 1.25905871e-01 4.47917031e-03
-3.67291957e-01 -3.36648971e-01 2.84634501e-01 1.13726485e+00
2.37166896e-01 3.12688529e-01 1.58646047e-01 7.29948401e-01
-9.62388292e-02 5.06418049e-01 -3.24597925e-01 6.59181699e-02
1.05160666e+00 1.37683964e+00 -2.98476130e-01 -2.95784891e-01
-7.12860525e-01 6.50258839e-01 3.62437934e-01 2.36014172e-01
-8.13228428e-01 -2.62266457e-01 1.15403175e+00 1.84213191e-01
-3.31506133e-01 -1.33565247e-01 -9.10073459e-01 -1.13115942e+00
-3.61261815e-01 -9.18478251e-01 7.19133139e-01 -2.05360964e-01
-1.61524642e+00 1.34363177e-03 9.69261155e-02 -1.16903102e+00
2.84846961e-01 -4.16089483e-02 -7.47451782e-01 1.18518674e+00
-1.77808964e+00 -5.17900944e-01 -4.49840605e-01 3.52330506e-01
3.66676778e-01 2.43912153e-02 7.98199236e-01 6.21806383e-01
-6.94557130e-01 1.03353250e+00 2.99835861e-01 -5.19232810e-01
1.36731803e+00 -1.00651145e+00 1.89526796e-01 7.50711203e-01
-3.38821113e-01 1.04046035e+00 8.20641279e-01 -4.15642351e-01
-1.03620517e+00 -6.16569459e-01 1.04864752e+00 -6.60039425e-01
2.06013829e-01 3.29708755e-02 -7.33592927e-01 2.18454123e-01
4.47647363e-01 -3.31149489e-01 1.57201946e+00 8.03045809e-01
-3.48061323e-01 6.53540716e-02 -1.39354777e+00 7.99576402e-01
1.38537538e+00 -2.43083909e-01 -1.92460731e-01 1.10745318e-01
1.16154954e-01 -2.44664893e-01 -9.74642932e-01 -1.35884717e-01
1.01995432e+00 -1.58174872e+00 9.23397958e-01 -5.84552944e-01
5.03823280e-01 4.15545739e-02 -1.69987604e-01 -1.27082205e+00
-1.12956715e+00 -5.95386863e-01 -1.34827480e-01 1.27135980e+00
1.87665597e-01 -7.06659913e-01 5.88750124e-01 1.34967732e+00
-1.24969654e-01 -6.29882276e-01 -2.22913399e-01 -7.87500143e-01
4.61975783e-01 -7.58146122e-02 5.45962691e-01 1.16097414e+00
6.87524676e-01 3.85643423e-01 -7.47614205e-01 -2.03506544e-01
6.72633410e-01 2.32572690e-01 7.99238741e-01 -1.38743281e+00
-8.49703252e-02 -1.00970542e+00 6.40044287e-02 -1.07641983e+00
-9.93724912e-02 -8.60864282e-01 -4.08124745e-01 -1.56593823e+00
4.27967012e-01 -6.06258214e-01 -6.29769266e-01 2.04517677e-01
-5.49941540e-01 3.65091056e-01 5.70972979e-01 2.87301451e-01
-1.00245655e+00 4.37699825e-01 1.06776977e+00 5.03121197e-01
-2.46913463e-01 2.88104326e-01 -1.95806539e+00 2.14705974e-01
1.24868846e+00 -1.48142725e-01 -6.36479676e-01 -1.96763262e-01
3.29390287e-01 -1.53968602e-01 -1.79755569e-01 -4.86405820e-01
1.77595958e-01 -8.10189724e-01 1.52247231e-02 5.66761056e-03
-3.69177647e-02 -4.90612090e-01 -8.57353210e-02 3.49473864e-01
-6.90838039e-01 -3.10221817e-02 -1.45754954e-02 4.87834871e-01
2.40430877e-01 -2.14648977e-01 8.01474571e-01 -1.29819393e-01
-5.57930470e-01 3.22870642e-01 -4.90872204e-01 3.22751760e-01
9.25686359e-01 -2.54024774e-01 -7.49552071e-01 -9.86519456e-01
-4.79593933e-01 5.95268786e-01 6.77727759e-01 5.39499104e-01
1.48672387e-01 -1.29682064e+00 -8.44635844e-01 -2.57530451e-01
2.94285923e-01 -1.04745007e+00 6.57754540e-01 9.97485518e-01
5.81575297e-02 4.20664966e-01 -5.56795001e-01 1.36996463e-01
-1.23302388e+00 5.52078426e-01 2.02404931e-01 1.32467732e-01
-5.28059229e-02 7.42038310e-01 3.84677917e-01 -4.29622173e-01
3.41943562e-01 4.54207748e-01 -3.41298014e-01 3.22178341e-02
7.30183363e-01 1.05128980e+00 -2.76592404e-01 -4.67509031e-01
-4.74852383e-01 8.49797390e-03 -4.92340624e-02 2.02712417e-01
9.58309233e-01 -7.26426959e-01 -1.25105113e-01 1.79924652e-01
6.42033458e-01 4.18381423e-01 -6.84123099e-01 -2.92556077e-01
-1.83404922e-01 -1.30181110e+00 1.20184338e-02 -1.10196543e+00
-9.00469303e-01 6.52133346e-01 2.86591351e-01 5.79649746e-01
1.12404490e+00 -4.85176742e-01 3.62003386e-01 -9.89622697e-02
2.85302013e-01 -1.20890915e+00 -1.05941288e-01 4.02139753e-01
6.18235290e-01 -1.20181453e+00 3.65831077e-01 -4.42687541e-01
-1.01869380e+00 8.46399605e-01 8.17025363e-01 -9.68614519e-02
7.32344866e-01 -1.43039301e-01 1.17511444e-01 2.04441622e-01
-7.32047617e-01 -3.16492051e-01 2.76881278e-01 7.52803624e-01
1.40083539e+00 2.90931135e-01 -1.14282560e+00 9.72698390e-01
-2.12973878e-01 1.06935278e-01 8.39186311e-01 6.23776793e-01
-6.91883206e-01 -1.61228752e+00 -2.27423623e-01 9.69850004e-01
-5.41774929e-01 -3.27059388e-01 -4.59130853e-01 3.33168745e-01
-1.48531616e-01 1.55972266e+00 -2.44111359e-01 -3.74192119e-01
4.73051310e-01 -2.64086336e-01 3.28285903e-01 -6.37402117e-01
-9.10524845e-01 -2.22530439e-01 4.19734508e-01 -6.01742387e-01
-3.17610860e-01 -7.46811748e-01 -6.47925794e-01 -1.35591793e+00
-4.14898604e-01 4.70459044e-01 3.55112523e-01 5.12083769e-01
5.98713517e-01 3.80714834e-01 6.17524624e-01 -4.37097162e-01
-7.16036797e-01 -7.06435680e-01 -8.93695772e-01 3.50026190e-01
9.70421582e-02 -8.28356445e-01 -2.03489542e-01 -6.03013813e-01]
|
[9.646844863891602, 5.660115718841553]
|
10148710-4e1b-42d3-92ba-ab8a83fc5655
|
enabling-surrogate-assisted-evolutionary
|
2301.13374
| null |
https://arxiv.org/abs/2301.13374v1
|
https://arxiv.org/pdf/2301.13374v1.pdf
|
Enabling surrogate-assisted evolutionary reinforcement learning via policy embedding
|
Evolutionary Reinforcement Learning (ERL) that applying Evolutionary Algorithms (EAs) to optimize the weight parameters of Deep Neural Network (DNN) based policies has been widely regarded as an alternative to traditional reinforcement learning methods. However, the evaluation of the iteratively generated population usually requires a large amount of computational time and can be prohibitively expensive, which may potentially restrict the applicability of ERL. Surrogate is often used to reduce the computational burden of evaluation in EAs. Unfortunately, in ERL, each individual of policy usually represents millions of weights parameters of DNN. This high-dimensional representation of policy has introduced a great challenge to the application of surrogates into ERL to speed up training. This paper proposes a PE-SAERL Framework to at the first time enable surrogate-assisted evolutionary reinforcement learning via policy embedding (PE). Empirical results on 5 Atari games show that the proposed method can perform more efficiently than the four state-of-the-art algorithms. The training process is accelerated up to 7x on tested games, comparing to its counterpart without the surrogate and PE.
|
['Ke Tang', 'Peng Yang', 'Guiying Li', 'Jinyuan Zhang', 'Xiaxi Li', 'Lan Tang']
|
2023-01-31
| null | null | null | null |
['atari-games']
|
['playing-games']
|
[-1.65348649e-01 -2.33904541e-01 1.40166461e-01 2.16366202e-02
-1.50839940e-01 -3.53879273e-01 4.39293534e-01 -6.77983239e-02
-1.11621559e+00 1.07062125e+00 -4.16443110e-01 -4.31338817e-01
-3.25931728e-01 -9.03822780e-01 -6.61183357e-01 -9.98748779e-01
1.82082672e-02 3.73932451e-01 7.04081804e-02 -5.11389017e-01
2.67242491e-01 5.53922355e-01 -1.90462303e+00 -3.80833894e-01
1.00826097e+00 1.02773702e+00 3.02328765e-01 3.70711923e-01
-2.80131787e-01 6.57826841e-01 -8.73187780e-01 -4.35307473e-01
2.35142186e-01 -4.34816718e-01 -1.06470630e-01 -3.78907442e-01
-1.51645675e-01 -3.05095941e-01 -4.39258307e-01 1.22644043e+00
8.98739278e-01 5.30506730e-01 5.33030629e-01 -1.26515293e+00
-5.25502086e-01 5.32746911e-01 -4.67122525e-01 2.71803647e-01
-3.37843984e-01 1.00138575e-01 6.26342773e-01 -5.34892857e-01
3.37410033e-01 1.12366045e+00 5.16585886e-01 9.51966763e-01
-9.97059643e-01 -8.29267502e-01 4.75950073e-03 4.24331725e-01
-9.64755058e-01 -1.14589989e-01 8.30042422e-01 6.59192652e-02
9.01859105e-01 -7.92527348e-02 1.23572958e+00 1.19427323e+00
-4.55979481e-02 9.68851984e-01 1.15086567e+00 -3.41198266e-01
7.55137503e-01 8.54852721e-02 -3.38746518e-01 6.81507885e-01
3.58671367e-01 7.45030463e-01 -3.85038108e-01 -1.35087818e-01
7.83514798e-01 -2.50871748e-01 -4.02550809e-02 -4.37216908e-01
-5.45557082e-01 1.15828395e+00 4.05464411e-01 8.18435028e-02
-5.57250619e-01 3.65319401e-01 6.74482942e-01 4.76924062e-01
2.20356271e-01 4.92005825e-01 -2.91392088e-01 -6.94439292e-01
-6.31609321e-01 3.60681713e-01 4.55111176e-01 3.24976027e-01
3.66337299e-01 7.70719111e-01 1.77855983e-01 9.54815209e-01
1.26423046e-01 4.49542493e-01 8.61932516e-01 -7.64562905e-01
2.41069078e-01 7.27389216e-01 1.13186225e-01 -7.80418873e-01
-2.36304045e-01 -5.48053026e-01 -8.45283210e-01 8.84125531e-01
2.19197780e-01 -4.40402180e-01 -5.08787155e-01 1.71454895e+00
4.46676612e-01 3.69355023e-01 2.94283956e-01 7.97682405e-01
3.30399752e-01 8.07428658e-01 1.92070529e-01 -2.37434313e-01
7.72648573e-01 -9.31650400e-01 -5.02377510e-01 -5.08756414e-02
5.27366698e-01 -2.67327994e-01 9.94623899e-01 3.87431383e-01
-8.09534729e-01 -5.30631006e-01 -1.09094679e+00 6.43843949e-01
-3.88702393e-01 3.32728833e-01 6.00440085e-01 8.93225729e-01
-1.03648698e+00 7.37792194e-01 -7.05869198e-01 7.73735046e-02
5.36525965e-01 6.07671857e-01 -5.55307381e-02 3.48656267e-01
-1.39641714e+00 9.84173059e-01 8.72313678e-01 1.95810139e-01
-6.84724331e-01 -6.84239566e-01 -6.03497088e-01 2.21907288e-01
4.78951663e-01 -3.58077019e-01 1.14678621e+00 -1.01250231e+00
-2.20890284e+00 1.08669959e-01 3.92933577e-01 -7.54417717e-01
5.71143806e-01 -1.01964943e-01 -3.58751893e-01 -3.35226804e-02
-5.65921068e-01 6.14693284e-01 1.03342783e+00 -8.75662386e-01
-7.31215358e-01 -1.62593514e-01 -1.73564777e-02 4.25183862e-01
-7.96031594e-01 -1.72665894e-01 -3.60058583e-02 -7.63843179e-01
-6.81656659e-01 -8.96218419e-01 -2.19272673e-01 -9.30349976e-02
5.05173564e-01 -4.61273074e-01 8.82359803e-01 -4.72106874e-01
1.35198128e+00 -2.09464049e+00 3.28061640e-01 1.72864199e-01
-3.22608687e-02 1.05706167e+00 -1.78201452e-01 2.01854587e-01
1.40291542e-01 -2.35680252e-01 -6.83729276e-02 4.61879112e-02
3.18265498e-01 4.55491334e-01 -1.98044017e-01 2.25033984e-01
-4.12600264e-02 8.84948611e-01 -9.96979475e-01 -2.59884983e-01
3.70185047e-01 5.55912137e-01 -7.11260557e-01 2.18479812e-01
-3.59814107e-01 2.44297713e-01 -5.10580719e-01 1.40900269e-01
2.93642581e-01 1.54785484e-01 1.58983111e-01 1.00132421e-01
-1.25965595e-01 -1.73308760e-01 -1.13974595e+00 1.26095140e+00
-5.17842710e-01 5.25585830e-01 -3.97528470e-01 -1.28662097e+00
1.26937997e+00 2.93836832e-01 4.58857447e-01 -8.52711916e-01
3.78859311e-01 2.98043102e-01 2.70597816e-01 -3.01379919e-01
3.08330417e-01 1.87864870e-01 8.21018144e-02 5.32296062e-01
1.67667076e-01 2.39007443e-01 3.67480695e-01 -4.69707936e-01
8.31674516e-01 4.74000782e-01 1.50714502e-01 1.51379537e-02
7.84912527e-01 -2.17917964e-01 7.23656654e-01 5.34326494e-01
-2.40680113e-01 -4.30327803e-01 3.48884374e-01 -5.85648835e-01
-1.18516243e+00 -6.61730647e-01 3.35619628e-01 9.45191920e-01
-8.25257897e-02 -5.22972718e-02 -7.48778403e-01 -6.90652192e-01
9.19840932e-02 9.09745097e-01 -6.04526520e-01 -6.12461090e-01
-6.28001273e-01 -9.32012200e-01 7.23688781e-01 5.20789087e-01
8.57333839e-01 -1.46979487e+00 -1.16055334e+00 5.46713948e-01
3.95006418e-01 -7.37202287e-01 -1.04639284e-01 1.05182402e-01
-1.04543269e+00 -7.68741369e-01 -8.24833810e-01 -5.32432079e-01
4.97220159e-01 -3.58370394e-01 7.28749335e-01 -2.30179317e-02
-1.07065439e-01 9.25728530e-02 -5.13969541e-01 -4.52154905e-01
-6.49387956e-01 4.36714292e-02 2.57485181e-01 -1.16044715e-01
4.66601998e-01 -6.38953507e-01 -5.94072104e-01 1.56702846e-01
-7.19919801e-01 -1.62522510e-01 7.11403489e-01 1.17105007e+00
3.37383777e-01 4.18218791e-01 8.65812480e-01 -5.08021176e-01
1.17396247e+00 -1.63348958e-01 -1.21151519e+00 4.75147367e-01
-8.28030705e-01 5.28369367e-01 1.06031942e+00 -7.65331626e-01
-1.05883944e+00 -2.61245459e-01 -2.12890342e-01 -6.65847421e-01
9.16485935e-02 4.53866929e-01 2.95846015e-01 -3.65620911e-01
4.40970272e-01 4.79642272e-01 2.42218539e-01 -3.67851526e-01
5.49511723e-02 5.25700808e-01 3.03971171e-01 -6.68800533e-01
4.57464695e-01 -8.70281234e-02 1.88045666e-01 -5.28500974e-01
-3.50210458e-01 1.13537908e-01 -1.11945122e-02 -4.30058360e-01
5.14522552e-01 -4.92608845e-01 -1.07765222e+00 5.36834598e-01
-7.40270913e-01 -5.48112512e-01 -4.34873879e-01 6.75151050e-01
-5.64407229e-01 1.49338260e-01 -3.81354481e-01 -1.08343840e+00
-6.11002266e-01 -1.26772916e+00 2.73445159e-01 5.50657392e-01
2.58078605e-01 -9.91607666e-01 2.04605386e-01 -3.88230719e-02
6.01933718e-01 1.71701953e-01 1.02575195e+00 -3.91364396e-01
1.69984758e-01 -4.20618318e-02 1.32169247e-01 5.30552864e-01
-1.42631620e-01 4.80662100e-02 -5.98501503e-01 -4.98173803e-01
7.21270815e-02 -5.16249478e-01 4.96555895e-01 3.13256413e-01
1.17525971e+00 -6.46047741e-02 1.95741996e-01 4.56400514e-01
1.62589598e+00 7.47331440e-01 5.36881030e-01 7.91810513e-01
1.88231230e-01 2.07276836e-01 7.49860048e-01 7.46115565e-01
-5.99616347e-03 6.28520906e-01 4.87906843e-01 1.88952431e-01
1.40651375e-01 -1.74363911e-01 6.65604293e-01 9.39270318e-01
-4.04270709e-01 -1.94490567e-01 -6.86099827e-01 2.57073849e-01
-1.94206023e+00 -1.04084802e+00 5.97022772e-01 2.07631779e+00
8.79790723e-01 3.88057940e-02 2.81368583e-01 2.83897817e-01
7.03612149e-01 8.55998485e-04 -1.06238532e+00 -7.97593594e-01
7.65642375e-02 5.04689991e-01 4.55646634e-01 1.21243112e-01
-8.40665638e-01 9.74214852e-01 5.23761415e+00 1.11642742e+00
-1.35181367e+00 2.50908993e-02 3.39962810e-01 -1.67717740e-01
8.81282613e-02 -4.70081389e-01 -8.31230640e-01 7.69897997e-01
1.09584892e+00 -3.57978761e-01 9.37332511e-01 1.02888680e+00
1.21268310e-01 2.07430273e-02 -5.00898659e-01 1.27593601e+00
-1.94723308e-01 -1.21057308e+00 9.68827903e-02 4.13338132e-02
8.96355033e-01 -5.47877438e-02 3.44473213e-01 7.25852430e-01
4.32554156e-01 -8.34873438e-01 5.26733518e-01 3.88839185e-01
6.39958441e-01 -1.43126595e+00 8.76719832e-01 4.57187712e-01
-8.28684568e-01 -6.19580507e-01 -8.81381691e-01 1.11616634e-01
-1.53423294e-01 6.22670688e-02 -4.47489977e-01 4.63836044e-01
8.05205941e-01 4.31771755e-01 -3.37604851e-01 1.04784429e+00
-3.46948594e-01 5.76229930e-01 -2.53863156e-01 -7.65129924e-01
7.15662122e-01 -6.01932883e-01 5.23263633e-01 4.95235771e-01
5.76834977e-01 -6.73863739e-02 -3.01808447e-01 6.53861642e-01
-1.74120709e-01 1.44645154e-01 -4.38432723e-01 -4.16375935e-01
5.50044537e-01 1.04107356e+00 -4.22610611e-01 -1.43351138e-01
-1.30982801e-01 7.72836983e-01 6.11645281e-01 1.67754248e-01
-1.12505150e+00 -3.04943144e-01 5.63360691e-01 -4.95704204e-01
7.17711508e-01 -1.27389729e-01 3.16140711e-01 -7.74232090e-01
-1.94659889e-01 -1.15247035e+00 1.22633874e-01 -3.95311445e-01
-1.05614209e+00 7.94171870e-01 -2.37220868e-01 -1.27936971e+00
-5.10481298e-01 -7.87895381e-01 -4.61944014e-01 5.92613280e-01
-1.44756544e+00 -4.13308412e-01 -4.70492616e-02 4.99153525e-01
5.07170200e-01 -8.58510375e-01 1.01163864e+00 3.75194579e-01
-8.34002078e-01 8.13394725e-01 6.88207030e-01 -9.21047777e-02
2.90397674e-01 -1.18440497e+00 1.49982512e-01 5.80606222e-01
1.19680531e-01 1.15424149e-01 6.10218644e-01 -2.58857369e-01
-1.30440271e+00 -8.83076310e-01 2.30090678e-01 3.94963324e-01
6.55294240e-01 -3.66273485e-02 -6.13498092e-01 2.54203100e-04
1.34008378e-01 -1.77959189e-01 6.23979867e-01 -3.62007558e-01
2.24838614e-01 -3.03245127e-01 -1.04424047e+00 9.64646876e-01
7.42085874e-01 -1.37681827e-01 -4.82996374e-01 -2.37096921e-01
2.36155748e-01 -2.42677271e-01 -8.73711526e-01 3.69654000e-01
5.75623453e-01 -9.41119909e-01 8.93479705e-01 -5.93677759e-01
2.13634625e-01 -2.18977928e-01 2.53059953e-01 -1.74561000e+00
-4.66564558e-02 -5.52096903e-01 -5.35452962e-01 9.56586480e-01
8.56583267e-02 -9.46283877e-01 8.76378715e-01 3.27974021e-01
1.95388511e-01 -1.18528068e+00 -1.04170728e+00 -1.01933038e+00
7.56484941e-02 -9.42360312e-02 8.71345043e-01 5.98300695e-01
-3.40029418e-01 -3.58804502e-02 -3.60224366e-01 -3.06965500e-01
5.77205122e-01 -2.53297482e-02 4.64593351e-01 -1.11168408e+00
-5.57526231e-01 -8.13056886e-01 -5.70410669e-01 -6.59194589e-01
4.19519395e-01 -6.67067587e-01 -3.49987179e-01 -1.16199172e+00
-4.57832605e-01 -7.02037215e-01 -6.80307388e-01 3.28660101e-01
-1.58731416e-01 -2.13418640e-02 2.14038938e-01 -2.66225845e-01
-3.03391337e-01 1.29158604e+00 1.26275349e+00 2.00237464e-02
-1.53170198e-01 2.14923434e-02 -2.65865982e-01 5.79297245e-01
1.42509413e+00 -6.72240555e-01 -8.54052007e-01 -2.94074893e-01
4.11276907e-01 -4.69244979e-02 1.86530009e-01 -1.16225231e+00
2.05641001e-01 4.07965034e-02 3.64527702e-01 -2.40919337e-01
3.03634793e-01 -9.96184051e-01 1.58390239e-01 8.60933006e-01
-2.08244622e-01 4.15123254e-01 4.78935212e-01 6.39721870e-01
-3.01919609e-01 -6.64850056e-01 6.89519942e-01 -1.82965547e-02
-9.05782223e-01 2.75409997e-01 -4.20488477e-01 -1.07486159e-01
1.17783785e+00 -2.16523200e-01 -6.61346689e-02 -7.47299865e-02
-4.24516380e-01 1.20036095e-01 2.20924690e-01 3.19168657e-01
7.60564148e-01 -1.36433697e+00 -4.22600418e-01 2.30400339e-01
-4.21467364e-01 -3.86652857e-01 2.50312060e-01 6.13589048e-01
-6.81446671e-01 1.12649538e-01 -7.19914615e-01 -1.16887785e-01
-1.27245414e+00 6.19246006e-01 4.54401135e-01 -5.48918366e-01
-6.83576465e-01 8.56121123e-01 -5.27169168e-01 -4.49643105e-01
4.90467489e-01 2.47762516e-01 -5.60488820e-01 1.66243520e-02
2.37696797e-01 6.25605345e-01 -1.89853549e-01 -2.19578490e-01
-5.35662435e-02 4.65883940e-01 -7.74064139e-02 -2.63953984e-01
1.56762838e+00 3.50360841e-01 9.32412297e-02 1.57680690e-01
9.06205535e-01 -5.67557573e-01 -1.53078687e+00 3.11748404e-02
-1.79599877e-02 -2.36921072e-01 3.37522477e-01 -6.95347607e-01
-1.31848443e+00 9.75196779e-01 9.02930200e-01 -1.74974486e-01
1.24042523e+00 -8.04729402e-01 8.00938606e-01 6.91933930e-01
5.79585493e-01 -1.57574391e+00 -2.98544802e-02 4.00484979e-01
7.37018049e-01 -1.05342484e+00 -1.90531611e-01 4.66146737e-01
-6.83262646e-01 1.38404870e+00 6.51031375e-01 -3.70548368e-01
4.43768889e-01 1.92755312e-01 -9.61089060e-02 1.00628130e-01
-8.34025502e-01 4.84530628e-02 -3.06668133e-02 3.87888640e-01
-1.13400735e-01 -1.33318409e-01 -5.27953088e-01 1.52154282e-01
-1.80762649e-01 1.22429423e-01 1.20885193e-01 1.00786114e+00
-2.80754000e-01 -1.46018374e+00 -1.29046351e-01 3.42123836e-01
-3.69991392e-01 1.41757891e-01 1.50513321e-01 8.63487005e-01
5.20969853e-02 6.16347790e-01 -4.34280075e-02 -4.09879565e-01
2.26471439e-01 -2.68837390e-03 7.02046812e-01 6.32334948e-02
-8.59286427e-01 -2.51412213e-01 -1.83675840e-01 -3.79594624e-01
-4.46023613e-01 -4.35275167e-01 -1.19503617e+00 -4.54935580e-01
-2.60784596e-01 4.00241762e-01 7.32129633e-01 6.42268896e-01
3.84283543e-01 5.82174778e-01 6.69188678e-01 -6.31136239e-01
-1.21371090e+00 -6.91113055e-01 -5.16237319e-01 1.55437723e-01
-2.17227995e-01 -1.08633173e+00 -1.58244714e-01 -3.85539830e-01]
|
[4.107669353485107, 2.1440775394439697]
|
ba46d17a-1408-40b3-8f12-528dd8f0fd0c
|
re-id-driven-localization-refinement-for
|
1909.08580
| null |
https://arxiv.org/abs/1909.08580v1
|
https://arxiv.org/pdf/1909.08580v1.pdf
|
Re-ID Driven Localization Refinement for Person Search
|
Person search aims at localizing and identifying a query person from a gallery of uncropped scene images. Different from person re-identification (re-ID), its performance also depends on the localization accuracy of a pedestrian detector. The state-of-the-art methods train the detector individually, and the detected bounding boxes may be sub-optimal for the following re-ID task. To alleviate this issue, we propose a re-ID driven localization refinement framework for providing the refined detection boxes for person search. Specifically, we develop a differentiable ROI transform layer to effectively transform the bounding boxes from the original images. Thus, the box coordinates can be supervised by the re-ID training other than the original detection task. With this supervision, the detector can generate more reliable bounding boxes, and the downstream re-ID model can produce more discriminative embeddings based on the refined person localizations. Extensive experimental results on the widely used benchmarks demonstrate that our proposed method performs favorably against the state-of-the-art person search methods.
|
['Jiacheng Ye', 'Xin Tan', 'Nong Sang', 'Chuchu Han', 'Changxin Gao', 'Yunshan Zhong', 'Chi Zhang']
|
2019-09-18
|
re-id-driven-localization-refinement-for-1
|
http://openaccess.thecvf.com/content_ICCV_2019/html/Han_Re-ID_Driven_Localization_Refinement_for_Person_Search_ICCV_2019_paper.html
|
http://openaccess.thecvf.com/content_ICCV_2019/papers/Han_Re-ID_Driven_Localization_Refinement_for_Person_Search_ICCV_2019_paper.pdf
|
iccv-2019-10
|
['person-search']
|
['computer-vision']
|
[-5.03742039e-01 -2.69377649e-01 1.30376682e-01 -3.43485206e-01
-8.73600006e-01 -4.69884932e-01 5.35313308e-01 -4.45062108e-02
-9.85886633e-01 4.74702060e-01 3.34138870e-01 4.00797129e-01
3.49285841e-01 -5.87038755e-01 -3.88608307e-01 -6.67131424e-01
2.98509926e-01 7.24576354e-01 5.21114469e-01 1.34668782e-01
2.81797294e-02 4.08756018e-01 -1.25539172e+00 -2.53702372e-01
7.50880897e-01 7.99411297e-01 2.15551332e-01 4.39849794e-01
2.30727926e-01 5.50818481e-02 -8.88378441e-01 -4.42771375e-01
3.27746451e-01 -1.51242346e-01 -3.80650610e-01 1.65564060e-01
4.98474181e-01 -5.40032327e-01 -5.70235550e-01 1.21645296e+00
6.60678685e-01 3.35513681e-01 6.32933378e-01 -1.13768995e+00
-9.96567428e-01 6.83521032e-02 -7.95003951e-01 3.44886452e-01
6.48942947e-01 2.42986292e-01 8.05058956e-01 -1.30165672e+00
2.41476983e-01 1.62468100e+00 6.69883788e-01 8.99937630e-01
-1.20026457e+00 -9.03239429e-01 4.78443772e-01 1.72876313e-01
-2.03047991e+00 -3.20682645e-01 4.82613385e-01 -5.53268373e-01
3.87532532e-01 2.89232153e-02 5.67553520e-01 9.75923061e-01
-4.76768315e-01 9.28963304e-01 6.65285528e-01 -3.31797272e-01
6.89676206e-04 5.52671552e-01 4.42182481e-01 9.12025273e-01
5.86932600e-01 1.86879933e-01 -3.34339261e-01 -2.08869383e-01
8.16065133e-01 3.62655997e-01 -2.27190360e-01 -6.02536201e-01
-1.22565401e+00 7.05331445e-01 8.15021038e-01 1.14474431e-01
-6.59171641e-02 7.38403946e-02 3.31753612e-01 -4.19507235e-01
3.39249521e-01 1.99373022e-01 -5.13796508e-03 3.96060050e-01
-9.95780051e-01 4.47045058e-01 3.11088622e-01 1.00408983e+00
7.91255772e-01 -4.27456737e-01 -7.81166494e-01 8.21713030e-01
4.62558866e-01 6.24529958e-01 3.02438289e-01 -2.65882194e-01
6.30846024e-01 8.80609989e-01 6.77837014e-01 -8.96285474e-01
-2.84819067e-01 -4.74869221e-01 -7.61053085e-01 -9.89147425e-02
7.29235828e-01 1.11731635e-02 -1.10390627e+00 1.60991824e+00
4.83036160e-01 7.67155886e-02 -1.50684565e-01 1.47297049e+00
8.03005517e-01 4.93746996e-01 2.88685858e-01 5.65102994e-01
1.74966252e+00 -1.36934459e+00 -2.82285690e-01 -6.20054126e-01
3.26216668e-01 -3.52405161e-01 7.98097014e-01 -1.76131979e-01
-7.04214513e-01 -1.08670390e+00 -9.29131031e-01 -1.13789916e-01
-3.83289963e-01 9.58878636e-01 -2.84119230e-02 7.41942048e-01
-9.18780267e-01 -6.99295402e-02 -5.91397524e-01 -6.45328522e-01
5.27037799e-01 3.02876711e-01 -4.69909370e-01 -2.26548553e-01
-1.00320113e+00 8.28842163e-01 4.03399050e-01 1.68846399e-01
-9.05107915e-01 -5.49215853e-01 -9.49669063e-01 7.54354894e-02
2.05328077e-01 -7.79274106e-01 7.76513815e-01 -2.94622779e-01
-7.12415993e-01 1.15031528e+00 -6.25887871e-01 -3.14300030e-01
7.96065927e-01 -3.77076656e-01 -2.94556081e-01 1.15327679e-01
8.07085037e-01 8.00052047e-01 9.77732837e-01 -1.20672297e+00
-1.13608956e+00 -6.47690892e-01 -1.36860177e-01 6.64679036e-02
-3.88624817e-01 3.54678690e-01 -1.04913282e+00 -6.56580567e-01
-2.41452884e-02 -8.84292066e-01 -2.93788642e-01 3.28295827e-01
-5.41391015e-01 -7.27231681e-01 7.83974826e-01 -8.03404748e-01
1.03516066e+00 -2.18828797e+00 1.43313438e-01 1.48337364e-01
3.41032177e-01 1.92374051e-01 1.74355228e-02 -1.44354641e-01
1.52622387e-01 3.72490473e-02 2.53617465e-01 -9.41663444e-01
1.09519176e-01 -3.59874159e-01 -1.25205338e-01 8.47884595e-01
1.66664705e-01 9.58912432e-01 -8.10997903e-01 -6.70939744e-01
3.44499201e-01 4.07193899e-01 -2.89595306e-01 3.12258035e-01
4.52290297e-01 5.58609545e-01 -5.58701098e-01 7.37756729e-01
6.61822021e-01 -3.17516714e-01 -5.08020401e-01 -2.69790500e-01
-1.51002094e-01 -2.92561024e-01 -1.23920214e+00 1.46569300e+00
1.29774241e-02 4.76983100e-01 -5.68204410e-02 -6.79027200e-01
1.03974450e+00 -1.51043147e-01 1.69271268e-02 -5.16203940e-01
-1.58205017e-01 1.05179437e-01 -4.02951241e-01 -1.54408872e-01
4.51832175e-01 3.97020906e-01 -3.70486289e-01 2.69216120e-01
-3.49655300e-02 6.37454391e-01 2.18081236e-01 2.61240453e-02
5.94784141e-01 2.37637013e-02 1.78749233e-01 -2.52187014e-01
1.07928240e+00 -4.56688628e-02 6.55713618e-01 9.58686888e-01
-8.00077915e-01 6.36290073e-01 -2.27578595e-01 -6.54255271e-01
-1.19709432e+00 -1.13664484e+00 -1.21935554e-01 1.29567564e+00
7.97606707e-01 -3.14564705e-01 -1.01357603e+00 -8.98737729e-01
2.02322483e-01 2.02988505e-01 -7.08005250e-01 -7.63222277e-02
-7.38020301e-01 -5.71251750e-01 7.16514230e-01 7.75931597e-01
9.56585288e-01 -7.91124284e-01 -3.79555255e-01 9.27671138e-03
-2.83346355e-01 -1.17249179e+00 -1.19151211e+00 -3.25590640e-01
-1.22498080e-01 -1.12363148e+00 -1.45978343e+00 -1.17905283e+00
1.14213169e+00 6.43589437e-01 6.30162120e-01 2.37029210e-01
-4.63010222e-01 4.50416386e-01 -1.18183024e-01 -2.38794193e-01
7.70280957e-02 -3.49972956e-02 4.32744950e-01 1.33516788e-01
8.76233101e-01 2.32771188e-01 -1.09769511e+00 7.23682225e-01
-1.42399147e-01 -2.34545112e-01 3.36597949e-01 8.64177525e-01
5.62730074e-01 1.16718531e-01 3.15130562e-01 -1.40026063e-01
4.74212527e-01 -7.31698871e-02 -6.61957026e-01 4.41301852e-01
-4.36967522e-01 9.96845365e-02 3.62824351e-01 -5.63159704e-01
-8.96658123e-01 3.29173833e-01 1.98111653e-01 -3.65274310e-01
-3.05947632e-01 -4.98084098e-01 -3.00446808e-01 -2.52332270e-01
5.58937252e-01 4.54262376e-01 -4.53365117e-01 -6.18827581e-01
2.86988407e-01 6.85158670e-01 7.78305054e-01 -5.03425956e-01
1.20854938e+00 7.63070285e-01 -4.95539695e-01 -4.61219251e-01
-8.08260679e-01 -1.04650998e+00 -9.21830416e-01 -2.07861900e-01
1.25475347e+00 -1.25499249e+00 -5.99436164e-01 4.12038773e-01
-1.36683416e+00 2.60828018e-01 1.16750114e-02 1.90988153e-01
-8.50530341e-04 4.90320385e-01 -3.66272777e-01 -1.03096795e+00
-5.10844052e-01 -1.18052948e+00 1.71635270e+00 9.16838408e-01
-1.04669884e-01 -6.50487423e-01 -7.32095465e-02 3.09605837e-01
1.46368863e-02 -1.15423560e-01 3.34441841e-01 -6.88019395e-01
-7.21630335e-01 -7.06247211e-01 -6.37271881e-01 4.39758673e-02
5.20335278e-03 -6.79397166e-01 -1.06030393e+00 -6.26299143e-01
-4.75861222e-01 -1.42730598e-03 1.13754618e+00 1.37127385e-01
9.05586660e-01 -9.32469070e-02 -1.04302895e+00 6.07733309e-01
1.13819385e+00 -1.06462233e-01 1.54969469e-01 6.02412760e-01
7.93063939e-01 5.07000089e-01 7.05534816e-01 3.84408146e-01
5.70036590e-01 9.25688505e-01 -4.05767784e-02 -3.72401774e-01
-2.53788233e-01 -7.81072021e-01 2.02374890e-01 -3.40495139e-01
6.49415180e-02 4.71409969e-02 -8.45694363e-01 7.27523565e-01
-2.03352141e+00 -9.06866133e-01 2.47532278e-01 2.19901538e+00
4.72980320e-01 -5.45741506e-02 6.33061469e-01 -1.29224464e-01
1.37450957e+00 1.07670829e-01 -5.51854312e-01 5.52107036e-01
5.31402752e-02 -3.24532837e-01 6.47279859e-01 3.59913141e-01
-1.57411349e+00 1.15146255e+00 5.64958572e+00 7.02551425e-01
-4.72857088e-01 3.05962890e-01 4.21046138e-01 1.13536879e-01
4.32931125e-01 -1.72207296e-01 -1.78076315e+00 7.91500330e-01
1.10358186e-01 -2.50640243e-01 1.38591588e-01 1.27279687e+00
1.90159038e-01 -7.62669221e-02 -1.33512366e+00 1.56432116e+00
3.89701486e-01 -9.40321684e-01 -1.80015966e-01 -9.86353587e-03
5.00025392e-01 -3.58620882e-01 6.30285293e-02 5.60620785e-01
1.90876633e-01 -8.57533932e-01 9.39901531e-01 5.24144650e-01
6.41916871e-01 -7.09545255e-01 7.41554797e-01 5.38413584e-01
-1.80908740e+00 -4.02844727e-01 -7.25826502e-01 2.96793729e-01
3.64494354e-01 1.15940057e-01 -7.15411603e-01 1.01957493e-01
1.11277986e+00 5.36336720e-01 -1.11744070e+00 1.38753569e+00
-3.40909570e-01 -7.08169583e-03 -3.24417174e-01 -1.78786203e-01
1.20688668e-02 -8.51957127e-02 5.81699073e-01 1.40684938e+00
1.81272551e-01 -1.35289118e-01 6.73487306e-01 1.26475084e+00
1.12875193e-01 -1.24746799e-01 -1.89946279e-01 5.23608267e-01
6.55324638e-01 1.18434393e+00 -6.15722954e-01 -4.41460282e-01
-3.90419483e-01 1.42891502e+00 5.81592679e-01 6.25751972e-01
-9.40322101e-01 -2.56942600e-01 8.78315449e-01 3.03119123e-01
4.23165441e-01 -2.13025525e-01 8.16601589e-02 -1.21331644e+00
2.77824908e-01 -4.74456906e-01 5.66235185e-01 -6.40260637e-01
-1.54398990e+00 5.44459283e-01 1.98352020e-02 -1.04081631e+00
-1.61461979e-02 -5.67838192e-01 -5.46011627e-01 1.24085057e+00
-1.44563186e+00 -1.51262343e+00 -7.97525525e-01 5.94297409e-01
5.21734059e-01 -3.62018317e-01 6.06830001e-01 4.52489913e-01
-9.63097513e-01 1.00879979e+00 -2.68991172e-01 8.86743724e-01
9.11673486e-01 -1.13493705e+00 7.05514789e-01 1.24776137e+00
-3.43561321e-02 9.00316715e-01 3.58603239e-01 -7.61751771e-01
-8.55811000e-01 -1.35741067e+00 6.46454155e-01 -8.50139976e-01
4.20689255e-01 -6.76265419e-01 -5.84871769e-01 5.35481453e-01
-4.81559873e-01 2.27013469e-01 3.19317311e-01 -6.90062642e-02
-4.42997396e-01 -2.17823267e-01 -1.06044984e+00 6.36540473e-01
1.27325261e+00 -5.99546790e-01 -6.08993649e-01 3.50633115e-01
4.68668610e-01 -2.84523636e-01 -2.71723181e-01 2.96930857e-02
3.51386189e-01 -6.55997992e-01 1.62997508e+00 -3.96848917e-01
-4.26444411e-01 -8.14027309e-01 1.13952748e-01 -1.00172484e+00
-6.31258607e-01 -7.70700425e-02 7.76891969e-03 1.48403132e+00
-6.93286434e-02 -5.30287743e-01 9.89414752e-01 9.59152758e-01
3.51550967e-01 -2.67415702e-01 -1.04962134e+00 -9.04500842e-01
-2.76091069e-01 5.62830642e-02 6.56348169e-01 1.04438983e-01
-2.53594488e-01 1.69735342e-01 -4.97630149e-01 7.22051144e-01
1.24686003e+00 -6.52129808e-03 1.20960498e+00 -1.28145850e+00
-1.01318106e-01 -2.68457651e-01 -6.90506697e-01 -1.68418264e+00
2.42577508e-01 -6.37372315e-01 3.13680559e-01 -1.51983988e+00
7.41769016e-01 -5.09613454e-01 -2.17654839e-01 3.70007306e-01
-7.54532099e-01 2.61290193e-01 3.80584389e-01 6.17866755e-01
-8.92784595e-01 5.15497506e-01 8.67102504e-01 -4.58425105e-01
-1.45911381e-01 1.19174257e-01 -7.08504975e-01 6.63655102e-01
3.67552459e-01 -4.18465644e-01 1.02809623e-01 -3.47939700e-01
-5.41669726e-01 -5.85176408e-01 1.14432275e+00 -1.15339160e+00
6.67063177e-01 2.52134442e-01 9.56610560e-01 -8.46133649e-01
4.44264978e-01 -6.82480633e-01 -3.48567754e-01 4.97583628e-01
-1.77580521e-01 -1.44867957e-01 -8.79850090e-02 8.24612141e-01
-5.32921590e-02 -2.64513373e-01 9.45256948e-01 -1.78306222e-01
-1.05488181e+00 5.37979007e-01 1.95370793e-01 2.49189548e-02
1.15772498e+00 -4.97877032e-01 -3.06400061e-01 -1.19315140e-01
-5.79379678e-01 5.12758076e-01 6.23545229e-01 5.66476047e-01
7.50240922e-01 -1.40936792e+00 -6.53882623e-01 2.86889493e-01
3.64908308e-01 6.69045299e-02 -7.00421631e-03 4.66849327e-01
-1.98799208e-01 6.41573191e-01 2.42940366e-01 -7.77757883e-01
-1.46099865e+00 8.67972255e-01 7.25178838e-01 -1.08081125e-01
-7.36110926e-01 1.03160059e+00 7.52820730e-01 -1.91067532e-01
5.50494313e-01 4.30653803e-02 -2.01377898e-01 -1.36481985e-01
1.03543174e+00 4.05012935e-01 -5.44268727e-01 -9.28992212e-01
-8.12621653e-01 8.41338217e-01 -2.59282202e-01 -1.02448918e-01
7.66000569e-01 -3.59334260e-01 2.80925483e-01 -1.81254610e-01
1.00297463e+00 -9.18333828e-02 -1.41055048e+00 -4.16250408e-01
4.24740948e-02 -7.24838018e-01 -3.69202733e-01 -3.44065994e-01
-6.89164102e-01 6.46689117e-01 9.06558692e-01 -2.63499200e-01
6.70728624e-01 3.46721023e-01 7.87737250e-01 3.36471617e-01
6.71966970e-01 -1.22703385e+00 2.83079952e-01 1.36135876e-01
7.35087335e-01 -1.50123823e+00 3.03108320e-02 -4.32369828e-01
-4.96096522e-01 8.20632815e-01 9.84506249e-01 -2.82708317e-01
4.87947494e-01 -3.82225737e-02 -9.55587029e-02 6.07175156e-02
2.81339377e-01 -5.30061722e-01 5.22842824e-01 9.70442832e-01
-1.43078759e-01 1.88455582e-02 1.68147326e-01 1.08000922e+00
-6.08078986e-02 -3.49127591e-01 -2.88020909e-01 4.10360157e-01
-5.92039585e-01 -8.29205990e-01 -1.01723671e+00 6.89529106e-02
3.62587459e-02 6.68627620e-02 -5.16972065e-01 7.70738244e-01
3.94459218e-01 1.00564253e+00 -1.84131917e-02 -2.58781165e-01
5.48589587e-01 -6.10190481e-02 2.04705045e-01 -7.25953877e-01
-2.84283459e-01 -1.20461330e-01 -2.33118966e-01 -4.26320136e-01
-1.59117952e-01 -6.64027393e-01 -1.18277001e+00 -1.43802911e-01
-5.06163895e-01 2.47002944e-01 2.36608699e-01 8.29646587e-01
2.24146605e-01 1.97570145e-01 4.00784582e-01 -1.02162421e+00
-6.07977927e-01 -8.86501908e-01 -3.31721634e-01 6.88204885e-01
3.94134194e-01 -1.04977524e+00 -2.68895984e-01 -9.47952718e-02]
|
[14.801182746887207, 0.7926192283630371]
|
524e24ed-350b-423a-be34-3ae13d32b1f2
|
fully-and-weakly-supervised-referring
|
2212.10278
| null |
https://arxiv.org/abs/2212.10278v1
|
https://arxiv.org/pdf/2212.10278v1.pdf
|
Fully and Weakly Supervised Referring Expression Segmentation with End-to-End Learning
|
Referring Expression Segmentation (RES), which is aimed at localizing and segmenting the target according to the given language expression, has drawn increasing attention. Existing methods jointly consider the localization and segmentation steps, which rely on the fused visual and linguistic features for both steps. We argue that the conflict between the purpose of identifying an object and generating a mask limits the RES performance. To solve this problem, we propose a parallel position-kernel-segmentation pipeline to better isolate and then interact the localization and segmentation steps. In our pipeline, linguistic information will not directly contaminate the visual feature for segmentation. Specifically, the localization step localizes the target object in the image based on the referring expression, and then the visual kernel obtained from the localization step guides the segmentation step. This pipeline also enables us to train RES in a weakly-supervised way, where the pixel-level segmentation labels are replaced by click annotations on center and corner points. The position head is fully-supervised and trained with the click annotations as supervision, and the segmentation head is trained with weakly-supervised segmentation losses. To validate our framework on a weakly-supervised setting, we annotated three RES benchmark datasets (RefCOCO, RefCOCO+ and RefCOCOg) with click annotations.Our method is simple but surprisingly effective, outperforming all previous state-of-the-art RES methods on fully- and weakly-supervised settings by a large margin. The benchmark code and datasets will be released.
|
['Yao Zhao', 'Eng Gee Lim', 'Jimin Xiao', 'MingJie Sun', 'Hui Li']
|
2022-12-17
| null | null | null | null |
['referring-expression', 'referring-expression-segmentation']
|
['computer-vision', 'computer-vision']
|
[ 2.56511927e-01 1.54874608e-01 -4.09255922e-01 -2.81323045e-01
-9.74484622e-01 -9.38895404e-01 4.53541577e-01 -1.01666292e-02
-6.21050537e-01 4.50097442e-01 -1.55082881e-01 -7.45537803e-02
4.87573445e-01 -4.46244776e-01 -8.35862994e-01 -7.08205581e-01
4.90294784e-01 3.22354257e-01 6.07681453e-01 9.37680602e-02
1.21430814e-01 4.09455776e-01 -1.28186345e+00 3.39589030e-01
8.03142369e-01 1.18615210e+00 4.05855179e-01 4.00834203e-01
-2.71735281e-01 6.92053139e-01 -4.63199228e-01 -4.99248952e-01
1.83496758e-01 -5.24125636e-01 -9.39193785e-01 2.62335449e-01
2.48836711e-01 8.16550404e-02 1.28210649e-01 1.26835632e+00
2.77275771e-01 -6.57931194e-02 5.03177762e-01 -1.03700387e+00
-5.80557525e-01 5.23104608e-01 -1.05612528e+00 -6.97873160e-02
1.86933950e-01 2.25366101e-01 1.26816881e+00 -1.04728734e+00
6.62941098e-01 1.02332783e+00 5.59615016e-01 5.25170028e-01
-1.40906441e+00 -4.61567819e-01 4.77428436e-01 -1.50466427e-01
-1.62280297e+00 -2.73239762e-01 8.22061837e-01 -5.73693752e-01
5.17576277e-01 2.64094472e-01 6.33836746e-01 9.68318522e-01
-3.66889209e-01 1.21061242e+00 1.12698901e+00 -4.14379537e-01
2.16617227e-01 3.28111827e-01 2.26705909e-01 8.33049119e-01
-4.08097506e-01 -1.27311513e-01 -1.66316256e-01 1.57058746e-01
7.20904768e-01 -1.65582478e-01 -2.39254087e-01 -3.22569907e-01
-1.10197294e+00 6.24996483e-01 6.03440642e-01 5.15093267e-01
-2.80172259e-01 2.41197020e-01 3.48981172e-01 -1.99616492e-01
6.13645077e-01 1.62843511e-01 -5.51791310e-01 9.96654015e-03
-1.19841468e+00 -8.16047639e-02 5.35696566e-01 1.07763469e+00
1.03933835e+00 -2.91936725e-01 -5.48937678e-01 9.17970598e-01
1.77942976e-01 2.05343679e-01 1.59645244e-01 -9.82906044e-01
3.44784439e-01 8.44045222e-01 7.42542073e-02 -5.67085803e-01
-2.95829713e-01 -5.61318099e-01 -4.88389611e-01 1.52167246e-01
6.82533503e-01 -1.82865515e-01 -1.19998741e+00 1.79820716e+00
4.66982573e-01 1.11968093e-01 -1.66924581e-01 1.05884564e+00
8.02506149e-01 3.99077505e-01 4.24743861e-01 7.94786289e-02
1.53218436e+00 -1.36693907e+00 -5.82372308e-01 -3.67664993e-01
7.82553136e-01 -7.28010654e-01 1.59482372e+00 1.14233784e-01
-1.13461602e+00 -5.33815026e-01 -7.94580102e-01 -3.39166254e-01
-3.47499251e-01 7.16687441e-01 4.63289171e-01 3.04939717e-01
-8.77566159e-01 3.02344799e-01 -9.14558411e-01 -1.57224447e-01
7.23645031e-01 1.71070099e-01 -1.81316316e-01 1.25385925e-01
-8.87769997e-01 6.78993881e-01 3.59113842e-01 2.09411174e-01
-6.90194726e-01 -6.04429185e-01 -9.44509923e-01 9.90932807e-02
5.62852681e-01 -4.34811801e-01 1.07413387e+00 -1.50781429e+00
-1.36225891e+00 1.50606251e+00 -2.94957101e-01 -3.01779181e-01
8.86910737e-01 -1.53127044e-01 1.14793330e-01 2.57048726e-01
3.94512892e-01 1.02444494e+00 8.48572314e-01 -1.42849171e+00
-8.00528169e-01 -3.52428317e-01 1.87924325e-01 6.82777241e-02
1.00837506e-01 1.52677238e-01 -1.19955182e+00 -7.03432322e-01
-7.35709723e-03 -9.62683320e-01 -8.34546611e-02 8.08463246e-02
-7.32017934e-01 -4.76992995e-01 8.55497479e-01 -5.84358335e-01
1.08073485e+00 -2.48443747e+00 1.61299631e-01 3.75561938e-02
1.67600229e-01 2.39470065e-01 -5.29229306e-02 -1.00191951e-01
-2.33193502e-01 2.83409923e-01 -3.94831628e-01 -7.31512666e-01
-1.57469921e-02 7.04220533e-02 -2.20587909e-01 6.13054574e-01
3.08043540e-01 1.35254443e+00 -8.29013228e-01 -8.04460287e-01
2.02318817e-01 2.84698784e-01 -3.80827397e-01 2.96756417e-01
-4.46423739e-01 7.02300489e-01 -4.40151304e-01 8.27113152e-01
6.78858042e-01 -5.13813794e-01 -1.22637942e-01 -4.09327090e-01
-1.22468285e-01 -1.40649781e-01 -9.37390447e-01 1.69350052e+00
-5.11609495e-01 4.67269421e-01 3.63433629e-01 -1.11440778e+00
6.99059963e-01 -2.16310397e-02 3.86389077e-01 -7.28798687e-01
1.62750542e-01 1.28773630e-01 -3.75766873e-01 -5.28575301e-01
1.58301249e-01 4.43625562e-02 -2.71143705e-01 2.70192951e-01
1.12601176e-01 -1.47265419e-01 3.42851490e-01 2.78070271e-01
7.07067549e-01 6.46149397e-01 2.13013947e-01 -9.04413387e-02
7.10295379e-01 7.71240517e-02 6.82085693e-01 6.03496373e-01
-3.53386045e-01 9.43153322e-01 8.27713728e-01 1.18612118e-01
-7.85613060e-01 -1.02758050e+00 -1.02199633e-02 1.47673631e+00
5.13966739e-01 -3.18239808e-01 -1.12557316e+00 -1.24633932e+00
-3.09907585e-01 8.45911443e-01 -8.34063590e-01 1.75505176e-01
-6.03013277e-01 -3.57365191e-01 4.71880704e-01 7.57415473e-01
5.67776740e-01 -1.13461995e+00 -4.22441900e-01 -1.17828138e-01
-1.87739268e-01 -1.35889649e+00 -7.57225454e-01 3.25529695e-01
-2.77351677e-01 -1.13518107e+00 -8.67265940e-01 -9.20521438e-01
1.04155600e+00 -4.65193354e-02 1.04221749e+00 2.60283828e-01
-1.48290515e-01 3.50532413e-01 -2.78518200e-01 3.55877578e-02
-1.89354748e-01 1.14138611e-01 -6.43193066e-01 2.48921081e-01
9.64680687e-02 -2.30652094e-01 -6.37622356e-01 4.17889684e-01
-8.33298087e-01 3.79685342e-01 5.15024483e-01 7.86127746e-01
1.03495181e+00 -3.92085910e-01 2.09496632e-01 -1.01841593e+00
1.61413416e-01 -2.41596445e-01 -8.13739359e-01 4.65114206e-01
-1.15744263e-01 -1.66613072e-01 6.02396071e-01 -3.70870531e-01
-9.56852078e-01 7.57889032e-01 -3.56430352e-01 -5.97657859e-01
-3.77387106e-01 2.61894852e-01 -3.89125049e-01 7.07833469e-02
3.96209031e-01 1.47959337e-01 -4.17267740e-01 -4.51692849e-01
7.24490225e-01 4.45828289e-01 6.19733930e-01 -5.73778391e-01
6.53320551e-01 6.09313607e-01 -3.16442668e-01 -4.62829888e-01
-1.25429761e+00 -5.10638058e-01 -8.75913262e-01 -1.94833770e-01
1.18896735e+00 -8.09350193e-01 -4.70339090e-01 3.33254784e-01
-1.25627267e+00 -6.09839320e-01 -5.14181614e-01 1.12381600e-01
-5.75857341e-01 2.37267211e-01 -5.40688515e-01 -6.45239592e-01
-1.49941906e-01 -1.40811670e+00 1.54698372e+00 4.20412004e-01
-5.03122061e-02 -9.91018713e-01 -3.29845041e-01 4.86899465e-01
8.17625821e-02 1.91943809e-01 6.73619568e-01 -7.97633290e-01
-5.86601138e-01 -1.99221298e-01 -6.63993001e-01 4.65002120e-01
-6.49391934e-02 2.02704351e-02 -1.20677972e+00 1.14508746e-02
-2.59419888e-01 -4.72656250e-01 1.12427855e+00 3.54470879e-01
1.47104037e+00 2.52586845e-02 -4.65176016e-01 7.73892581e-01
1.27325666e+00 -1.92510650e-01 4.50136960e-01 1.16280213e-01
8.93973529e-01 8.02675426e-01 8.14366877e-01 -3.03295068e-02
3.58294159e-01 7.43205309e-01 4.40344751e-01 -6.44532084e-01
-3.86977553e-01 -4.07612890e-01 1.93829611e-01 1.82952285e-01
1.49596170e-01 -2.65951008e-01 -8.09511662e-01 6.27129376e-01
-1.93064916e+00 -5.83151996e-01 -3.40396799e-02 1.88842046e+00
1.00112700e+00 1.07813627e-01 1.07527129e-01 -6.14683069e-02
8.61476064e-01 2.32882172e-01 -6.27073407e-01 -1.19021712e-02
-1.61467955e-01 1.88103452e-01 5.80658793e-01 5.16300678e-01
-1.38163114e+00 1.49292576e+00 5.36539221e+00 1.28160369e+00
-1.39484942e+00 2.61501610e-01 9.07217443e-01 1.59879446e-01
-1.59011453e-01 1.38110295e-01 -9.19467747e-01 4.93940353e-01
2.58344859e-01 2.95241237e-01 2.51116067e-01 9.79682863e-01
2.34983444e-01 -3.17192346e-01 -1.15013301e+00 8.80102515e-01
2.92600151e-02 -1.15022230e+00 -3.35696697e-01 -2.21060291e-01
6.40843570e-01 5.92909008e-02 1.20740704e-01 2.84602433e-01
2.30739504e-01 -9.88838553e-01 1.10046673e+00 3.85283440e-01
9.39919055e-01 -4.86848593e-01 5.40359735e-01 4.63710040e-01
-1.36513305e+00 1.57823741e-01 9.53233391e-02 3.16515088e-01
3.46382946e-01 5.67779124e-01 -5.83251357e-01 4.33835745e-01
6.70396924e-01 7.57095575e-01 -7.30248392e-01 7.86003947e-01
-8.87987852e-01 7.56544948e-01 -2.62826949e-01 2.79019624e-01
3.48764151e-01 -1.66385591e-01 3.87240261e-01 1.38773167e+00
-2.04907358e-01 -2.12334841e-01 5.26257038e-01 1.58017862e+00
-1.65142328e-01 1.27073228e-01 -6.02996126e-02 -6.81773992e-03
1.50757357e-01 1.57961857e+00 -1.15514326e+00 -3.47503334e-01
-4.56586927e-01 1.26206660e+00 4.98619825e-01 6.51833832e-01
-1.16575158e+00 -4.15085822e-01 1.18250452e-01 1.05207257e-01
5.86003661e-01 -6.47514313e-02 -4.74702686e-01 -1.09330010e+00
2.35311072e-02 -4.28332239e-01 3.48439634e-01 -8.36227953e-01
-1.07132387e+00 4.66738224e-01 -1.31632924e-01 -1.03625858e+00
-5.07601462e-02 -4.75046486e-01 -6.28135443e-01 9.21312153e-01
-1.77691793e+00 -1.65675461e+00 -4.38584059e-01 5.35043001e-01
5.88269830e-01 3.99175674e-01 5.01063466e-01 2.66497642e-01
-9.35899198e-01 7.05782294e-01 -4.81306702e-01 4.32640642e-01
5.97789466e-01 -1.41518331e+00 2.91738808e-02 9.47279274e-01
2.41975173e-01 4.75570202e-01 4.40875143e-01 -5.06859422e-01
-7.23939538e-01 -1.20392728e+00 6.33177042e-01 -3.63409191e-01
6.78091407e-01 -6.96497381e-01 -9.14725304e-01 6.96367502e-01
3.00344434e-02 5.62702715e-01 3.11214387e-01 -2.71760494e-01
-2.01046184e-01 1.76404402e-01 -1.04878712e+00 5.46526670e-01
1.01487422e+00 -6.08112514e-01 -5.53639352e-01 3.94002110e-01
7.66492724e-01 -6.27639055e-01 -4.08089697e-01 3.16321611e-01
1.99649572e-01 -8.44098330e-01 8.60355794e-01 -3.56459111e-01
3.98424625e-01 -5.36432147e-01 1.21827396e-02 -9.14074063e-01
4.24510911e-02 -4.94694948e-01 2.54261792e-01 1.61229038e+00
5.41162550e-01 -2.66788900e-01 7.81629384e-01 4.25028294e-01
-1.32524729e-01 -9.27061975e-01 -8.41661692e-01 -5.28630495e-01
4.04347666e-02 -5.38891912e-01 2.13529766e-01 7.30624259e-01
-2.97315449e-01 4.11987305e-01 -3.09946779e-02 1.37687996e-01
3.79052460e-01 3.49516153e-01 7.25726783e-01 -7.73106098e-01
-2.32625484e-01 -6.05232894e-01 -1.09263457e-01 -1.55333424e+00
5.57509899e-01 -9.96662080e-01 2.24646017e-01 -1.44771159e+00
2.62054205e-01 -5.35561264e-01 -1.73868060e-01 7.24276841e-01
-3.81627560e-01 4.69269067e-01 1.11535825e-01 1.85205385e-01
-8.91313970e-01 4.35868591e-01 1.27899027e+00 -2.22865224e-01
-2.31258869e-01 7.88533837e-02 -6.57760799e-01 1.10292780e+00
5.22320747e-01 -3.46035212e-01 -1.95185587e-01 -3.05560917e-01
4.46250709e-03 -2.25291371e-01 6.39651775e-01 -4.39608514e-01
3.49224508e-01 -1.54574076e-03 3.35973084e-01 -7.40384579e-01
1.88541517e-01 -8.54806960e-01 -4.19991106e-01 2.76218411e-02
-5.08142591e-01 -3.94465715e-01 2.08026394e-02 4.33009982e-01
-2.27042869e-01 -4.09915775e-01 1.09051287e+00 -2.00434208e-01
-7.05024719e-01 2.34391108e-01 -1.76677257e-02 4.33605283e-01
1.30800796e+00 -2.33894274e-01 -3.98428142e-02 -1.70139268e-01
-1.08439326e+00 4.68804359e-01 6.39843225e-01 2.40688771e-01
3.50493103e-01 -1.01289308e+00 -4.07532573e-01 2.06144974e-01
2.90518850e-01 3.16606939e-01 7.43757188e-02 1.28496778e+00
-3.31540793e-01 1.99203461e-01 3.15697253e-01 -8.62780154e-01
-1.25804281e+00 6.00710392e-01 5.89221478e-01 -3.77926856e-01
-5.25392771e-01 1.06993484e+00 6.53337181e-01 -2.97069550e-01
4.79152352e-01 -4.05790120e-01 -2.61059254e-01 7.86147788e-02
2.92934567e-01 1.19879238e-01 -2.24271879e-01 -9.61429894e-01
-3.54789287e-01 8.93230557e-01 8.55160691e-03 -1.10513434e-01
9.75081801e-01 -3.36370856e-01 -1.69300064e-01 2.82418668e-01
1.31308758e+00 3.48085672e-01 -1.45786858e+00 -2.34739408e-01
9.88321453e-02 -2.64970928e-01 1.64002813e-02 -9.41940546e-01
-1.39966750e+00 9.68903840e-01 4.46049362e-01 1.03934996e-01
1.09429491e+00 5.28055012e-01 6.18778825e-01 -1.98680758e-01
1.43139675e-01 -1.06965482e+00 1.75373361e-01 3.88965696e-01
7.58572936e-01 -1.23210740e+00 -2.76477814e-01 -7.09010899e-01
-8.56870055e-01 7.59359062e-01 8.34876120e-01 -1.01240538e-01
4.34254736e-01 3.96805555e-01 1.45966217e-01 -1.04751609e-01
-2.34906062e-01 -5.91044545e-01 3.80998909e-01 4.28169906e-01
4.32984531e-01 -7.05526173e-02 -3.60229015e-01 1.00478184e+00
1.53393030e-01 -7.25106224e-02 -9.52354167e-03 7.00896144e-01
-2.66550690e-01 -1.04899204e+00 -2.54766166e-01 7.78820068e-02
-6.51555598e-01 -9.64545012e-02 -6.49211943e-01 7.31878459e-01
4.76241589e-01 7.16773152e-01 7.30233938e-02 -8.90638903e-02
4.45501000e-01 -9.89184603e-02 2.14945361e-01 -6.85195863e-01
-6.05039418e-01 5.08427322e-01 -6.58742487e-02 -8.27227056e-01
-5.18825114e-01 -4.96794969e-01 -1.75753593e+00 3.57721329e-01
-5.05344152e-01 9.43512097e-02 4.82891530e-01 1.02434480e+00
1.58331037e-01 5.95100641e-01 4.86378491e-01 -9.29219484e-01
-2.56590158e-01 -6.71599746e-01 -4.46465641e-01 6.12642348e-01
2.51492113e-01 -6.38160646e-01 -5.23213923e-01 1.94829211e-01]
|
[9.961738586425781, 0.9575672745704651]
|
4b67ff2d-5a0e-419e-bd50-008416259c17
|
ganalyzer-analysis-and-manipulation-of-gans
|
2302.00908
| null |
https://arxiv.org/abs/2302.00908v1
|
https://arxiv.org/pdf/2302.00908v1.pdf
|
GANalyzer: Analysis and Manipulation of GANs Latent Space for Controllable Face Synthesis
|
Generative Adversarial Networks (GANs) are capable of synthesizing high-quality facial images. Despite their success, GANs do not provide any information about the relationship between the input vectors and the generated images. Currently, facial GANs are trained on imbalanced datasets, which generate less diverse images. For example, more than 77% of 100K images that we randomly synthesized using the StyleGAN3 are classified as Happy, and only around 3% are Angry. The problem even becomes worse when a mixture of facial attributes is desired: less than 1% of the generated samples are Angry Woman, and only around 2% are Happy Black. To address these problems, this paper proposes a framework, called GANalyzer, for the analysis, and manipulation of the latent space of well-trained GANs. GANalyzer consists of a set of transformation functions designed to manipulate latent vectors for a specific facial attribute such as facial Expression, Age, Gender, and Race. We analyze facial attribute entanglement in the latent space of GANs and apply the proposed transformation for editing the disentangled facial attributes. Our experimental results demonstrate the strength of GANalyzer in editing facial attributes and generating any desired faces. We also create and release a balanced photo-realistic human face dataset. Our code is publicly available on GitHub.
|
['Timothy Sweeny', 'Sarah Ariel Lamer', 'Mohammad H. Mahoor', 'Ali Pourramezan Fard']
|
2023-02-02
| null | null | null | null |
['face-generation']
|
['computer-vision']
|
[ 2.61977106e-01 5.01702785e-01 -4.32970859e-02 -6.15176857e-01
-4.16298360e-01 -5.54108620e-01 7.15929270e-01 -7.62781978e-01
1.69028059e-01 9.13712382e-01 2.00292289e-01 2.93802619e-01
4.24632430e-01 -1.04168546e+00 -6.08380020e-01 -1.13214707e+00
3.41265827e-01 5.31376421e-01 -8.40543687e-01 -2.25052908e-01
-3.54727864e-01 4.99678522e-01 -1.61456776e+00 2.03981712e-01
8.10460031e-01 9.50691104e-01 -7.11293280e-01 5.16082346e-01
1.05019949e-01 6.84793293e-01 -9.22909617e-01 -1.01008272e+00
5.52420497e-01 -1.00134027e+00 -3.22617322e-01 8.48420709e-02
6.32110357e-01 -4.67264414e-01 -3.32250714e-01 1.19242811e+00
4.86067295e-01 -3.57197851e-01 8.85369599e-01 -1.95509088e+00
-9.94125068e-01 4.19084311e-01 -5.38453877e-01 -5.08921325e-01
2.62425035e-01 3.72278631e-01 7.96628177e-01 -7.40598619e-01
6.76255643e-01 1.39811766e+00 1.66886836e-01 1.15550363e+00
-1.03644753e+00 -1.38554418e+00 -1.90850079e-01 -1.38303572e-02
-1.37912941e+00 -7.61103272e-01 9.98176515e-01 -3.22676033e-01
8.54940340e-03 5.47038853e-01 8.56861472e-01 1.68292427e+00
8.68019462e-02 3.98826957e-01 1.36837041e+00 -2.38000304e-01
1.17669865e-01 1.43320605e-01 -6.79769576e-01 6.27767384e-01
2.62041926e-01 9.50139537e-02 -4.92196977e-01 -3.12624753e-01
8.31317365e-01 -7.98667297e-02 -2.02472974e-02 -3.69167030e-01
-1.10583472e+00 9.96571302e-01 1.84324339e-01 -9.80854109e-02
-4.27756518e-01 7.65589178e-02 6.78316504e-02 2.70768285e-01
6.38599753e-01 2.80465901e-01 3.75619158e-02 1.88238416e-02
-7.51298189e-01 3.93581629e-01 5.09972513e-01 1.16350198e+00
7.96304405e-01 5.01489341e-01 -2.94875115e-01 7.74276435e-01
1.57863557e-01 9.03094471e-01 3.49361300e-01 -1.15559721e+00
2.22916171e-01 6.30641043e-01 1.41071342e-02 -1.18872726e+00
1.17099300e-01 -1.07105009e-01 -1.24755955e+00 4.18736100e-01
3.40957940e-01 -2.21328884e-01 -1.13653743e+00 2.24900341e+00
3.78261477e-01 -1.23109281e-01 2.45928355e-02 6.88119829e-01
9.31648731e-01 5.39713442e-01 7.86254629e-02 -2.12671325e-01
1.28871453e+00 -6.57351375e-01 -1.05750108e+00 -1.18296787e-01
1.69800743e-01 -8.19514453e-01 1.05830538e+00 -6.09894749e-03
-1.30122709e+00 -4.41675663e-01 -9.34346855e-01 6.41174912e-02
-1.87325284e-01 2.56156355e-01 6.31819248e-01 1.09378994e+00
-1.06852949e+00 2.23334223e-01 -5.68311512e-01 -1.72110721e-01
9.39127088e-01 3.71539444e-01 -6.91701591e-01 -4.92098890e-02
-1.06009471e+00 6.21017873e-01 -7.67512843e-02 2.78040290e-01
-1.13188732e+00 -6.48268342e-01 -8.60259652e-01 -5.28517179e-02
-4.62146886e-02 -9.26885247e-01 6.43658876e-01 -1.72994220e+00
-1.78923678e+00 1.15721607e+00 -2.42402274e-02 1.38002664e-01
7.97534704e-01 5.89633584e-02 -5.32039940e-01 -1.12045355e-01
-6.78678825e-02 8.85650873e-01 1.25423861e+00 -1.27458358e+00
-8.08630064e-02 -5.22413492e-01 4.09931540e-02 1.05579466e-01
-4.28843051e-01 6.49791509e-02 -1.40756160e-01 -8.25670660e-01
-2.91284859e-01 -1.09897542e+00 6.33992851e-02 8.97654444e-02
-7.67829955e-01 2.59725332e-01 6.92627072e-01 -6.72676086e-01
6.29923642e-01 -2.15974522e+00 3.30135375e-01 2.55771160e-01
3.93256932e-01 1.58110470e-01 -3.40958685e-01 1.64522722e-01
-4.08365041e-01 3.14599603e-01 -2.94521898e-01 -4.49109435e-01
-8.64316151e-03 2.01788113e-01 -3.39455456e-01 5.36249876e-01
2.59276956e-01 8.98424387e-01 -6.83389366e-01 -3.50492895e-01
-8.42853040e-02 7.23301291e-01 -7.16045499e-01 5.96090436e-01
8.09798092e-02 8.20894241e-01 -2.65419960e-01 8.66677403e-01
9.20269608e-01 2.27972284e-01 1.52038634e-01 -2.90288657e-01
4.55108106e-01 -1.28578305e-01 -6.50761187e-01 1.29744875e+00
-2.29065001e-01 6.06676996e-01 -1.13919087e-01 -5.74374318e-01
1.22329080e+00 3.49771857e-01 3.93756598e-01 -5.55711508e-01
3.68006349e-01 -4.83396165e-02 1.23015784e-01 -1.76264688e-01
1.50569484e-01 -4.08979297e-01 -8.96461457e-02 5.13756573e-01
1.53778404e-01 -2.14347124e-01 -8.41929242e-02 1.75985217e-01
8.59087229e-01 4.43023294e-02 1.56164140e-01 -2.67276391e-02
4.70297247e-01 -4.70307440e-01 7.10729897e-01 2.36143798e-01
-1.61047816e-01 8.23363125e-01 1.03918791e+00 -5.50382555e-01
-1.49929142e+00 -1.10011554e+00 1.21837944e-01 7.12774396e-01
-3.51951569e-01 -3.59269857e-01 -1.25222445e+00 -6.79619730e-01
-2.26958424e-01 7.07629502e-01 -1.05511081e+00 -5.86141944e-01
-4.08090889e-01 -9.19493735e-01 8.20988953e-01 2.56854832e-01
5.76775849e-01 -1.17370176e+00 -1.80124685e-01 -4.64936495e-01
-2.82646000e-01 -1.08455276e+00 -2.62230963e-01 -6.85759485e-01
-5.14250338e-01 -1.05800879e+00 -7.21587479e-01 -4.21115488e-01
1.26127934e+00 -3.73578817e-01 1.07640350e+00 8.68950635e-02
-1.28676921e-01 1.29151776e-01 -2.39532441e-01 -5.43971837e-01
-7.93729305e-01 -6.79522008e-02 1.81273326e-01 6.64215207e-01
1.32565722e-01 -7.77194977e-01 -7.83443749e-01 3.61803919e-01
-8.61086249e-01 4.31011200e-01 4.82270539e-01 8.67312133e-01
2.93237031e-01 -1.24560528e-01 5.54654181e-01 -1.28636205e+00
3.25993448e-01 -3.79000217e-01 -2.28479996e-01 1.34473875e-01
-3.45192999e-01 -1.98895946e-01 7.63810635e-01 -5.21667600e-01
-1.06064510e+00 -5.85409924e-02 -2.54691690e-01 -5.15559971e-01
-2.65388638e-01 -1.88920856e-01 -7.02183187e-01 -3.68837751e-02
4.88025397e-01 1.78992115e-02 2.87624031e-01 -1.01750150e-01
5.88167369e-01 4.93643492e-01 4.34320539e-01 -6.44685090e-01
1.19841707e+00 7.74483919e-01 4.99152252e-03 -4.97343838e-01
-5.39392889e-01 4.62637067e-01 -4.96647924e-01 -2.88383842e-01
8.03195596e-01 -9.64039207e-01 -6.58736110e-01 8.55782449e-01
-8.10354829e-01 -3.05420071e-01 -5.81576467e-01 7.95564428e-02
-8.20922852e-01 -2.01037318e-01 -3.35434705e-01 -5.61868250e-01
-3.80217165e-01 -1.18935001e+00 9.70993698e-01 2.93145180e-01
-2.98493057e-01 -5.18037379e-01 -5.65461665e-02 6.16033435e-01
5.11740267e-01 8.48833621e-01 1.10436118e+00 -2.88395524e-01
-4.28214252e-01 -2.74347752e-01 -5.27092405e-02 5.03320992e-01
5.96586406e-01 3.44032198e-01 -1.12421453e+00 -3.91445875e-01
-2.96874344e-02 -4.63456273e-01 2.55107582e-01 5.77841699e-02
1.23846805e+00 -7.41466641e-01 -1.62331446e-03 1.04383671e+00
9.69408691e-01 2.21967638e-01 1.01166475e+00 -2.53872693e-01
9.76359487e-01 7.11276054e-01 3.74618948e-01 3.87772679e-01
3.36891949e-01 6.73492968e-01 5.41487932e-01 -2.95479089e-01
-1.86371714e-01 -5.13986051e-01 5.06498933e-01 7.68406928e-01
-6.24440372e-01 -1.79113612e-01 -5.32864451e-01 2.82483131e-01
-1.29091680e+00 -1.04288566e+00 3.06153804e-01 2.01204967e+00
1.04689717e+00 -3.45209211e-01 1.07276745e-01 -6.58247322e-02
6.55285954e-01 2.36913830e-01 -6.45253062e-01 -3.85785788e-01
-3.21264029e-01 5.43475926e-01 3.14313054e-01 2.47346520e-01
-6.73184156e-01 1.04623342e+00 6.06455946e+00 7.24965632e-01
-1.17695832e+00 1.27118424e-01 1.16695333e+00 -4.14482683e-01
-7.20698774e-01 -7.95981959e-02 -4.62842882e-01 6.41012728e-01
6.76324964e-01 -2.66863465e-01 6.22536778e-01 8.51844251e-01
-1.87437013e-02 3.46154422e-01 -1.04337919e+00 1.16329026e+00
4.36535269e-01 -1.07029068e+00 5.65900147e-01 2.99782604e-01
9.60622489e-01 -6.62753046e-01 6.32480323e-01 1.03653699e-01
4.63483065e-01 -1.67731786e+00 7.34146178e-01 6.50215209e-01
1.48509049e+00 -9.69298303e-01 6.18219495e-01 -2.40823776e-02
-4.86772388e-01 2.18610525e-01 -3.77815306e-01 1.24378115e-01
-1.71199441e-01 4.46720988e-01 -5.07440925e-01 2.88846642e-01
6.66196227e-01 3.93000811e-01 -6.71029091e-01 1.96531400e-01
-5.81733644e-01 4.58523214e-01 -1.78431086e-02 3.53640676e-01
-3.09314430e-01 -4.78755653e-01 4.22336578e-01 4.66683805e-01
6.18496597e-01 2.74045765e-01 -5.85489452e-01 1.13210607e+00
-5.84429383e-01 1.30460024e-01 -8.92593980e-01 -2.13197574e-01
2.83431351e-01 1.49695718e+00 -3.62643063e-01 -2.42368490e-01
-4.00315737e-03 1.03854537e+00 1.43414065e-01 4.69313145e-01
-1.01503110e+00 -9.11314562e-02 1.27508497e+00 6.55465722e-02
-9.71517935e-02 1.25774384e-01 -2.38029972e-01 -1.26297319e+00
6.83632679e-03 -1.55446780e+00 4.74427901e-02 -9.75225806e-01
-1.19283092e+00 8.55580986e-01 -2.10783049e-01 -9.34351146e-01
-3.72079730e-01 -3.56997490e-01 -5.64699292e-01 9.94091034e-01
-1.07671225e+00 -1.57172275e+00 -6.55737042e-01 8.41236472e-01
1.08605362e-01 -5.74683249e-01 1.09212768e+00 2.66558975e-01
-5.86338460e-01 1.07273805e+00 -2.39907116e-01 3.14905047e-01
8.28773499e-01 -8.61737132e-01 3.51562142e-01 5.86820424e-01
8.97590443e-02 5.57892442e-01 7.25081682e-01 -4.75996614e-01
-1.23687029e+00 -1.07799649e+00 6.30899251e-01 -6.11418009e-01
7.40644708e-02 -8.08175445e-01 -5.05137801e-01 1.00847161e+00
4.02224511e-01 2.06913725e-01 9.19713199e-01 -1.83790594e-01
-5.23845196e-01 -3.68355662e-01 -1.39581370e+00 8.23572993e-01
1.22101629e+00 -4.50772196e-01 3.82476971e-02 2.57342607e-01
3.35913360e-01 -4.53048736e-01 -7.88945913e-01 5.39879382e-01
6.97939038e-01 -1.17749524e+00 8.88585746e-01 -8.26892793e-01
8.34518015e-01 -1.44747034e-01 -6.61679208e-02 -1.63816857e+00
-1.50650144e-01 -7.76716232e-01 1.00286171e-01 1.54494381e+00
7.52471313e-02 -6.70852423e-01 1.02556777e+00 6.60525918e-01
3.62041473e-01 -6.32423222e-01 -7.81090438e-01 -3.49331588e-01
1.90846592e-01 9.35749412e-02 1.22614348e+00 1.17423594e+00
-6.24155223e-01 -4.59211878e-03 -8.82253349e-01 -8.98047313e-02
7.18599737e-01 6.11804845e-03 1.26011491e+00 -8.31913531e-01
6.89125508e-02 -3.58300149e-01 -6.45922959e-01 -1.18908212e-01
4.31196392e-01 -7.61860132e-01 -4.31314588e-01 -8.34191620e-01
3.27147007e-01 -5.54775178e-01 -1.05088833e-03 7.10961819e-01
-4.61505391e-02 7.34983087e-01 1.99921593e-01 8.99587795e-02
1.53511763e-01 9.62173998e-01 1.57298839e+00 -1.66075096e-01
2.93209583e-01 -1.69398770e-01 -9.65193272e-01 6.89509511e-01
8.53880703e-01 -5.75935662e-01 -4.03048009e-01 -2.43619174e-01
4.48303342e-01 -5.81816658e-02 4.76184458e-01 -7.52681375e-01
-3.11757058e-01 -3.75644803e-01 7.23783135e-01 3.16066034e-02
5.37259817e-01 -7.81398535e-01 7.24820852e-01 2.10245624e-01
-3.33353400e-01 1.14604704e-01 -1.00961827e-01 7.49597698e-02
-2.92799294e-01 2.98599869e-01 9.88186955e-01 -2.56483983e-02
2.72340588e-02 6.51983321e-01 3.71436886e-02 1.37733042e-01
1.23141861e+00 -3.84030789e-02 -4.48048979e-01 -9.08356011e-01
-5.01207292e-01 -3.15693885e-01 8.86980474e-01 5.29122114e-01
4.74366248e-01 -1.86207879e+00 -1.00432634e+00 7.77621984e-01
6.00017011e-02 -1.18317194e-01 4.75807458e-01 4.82891172e-01
-5.07302046e-01 -1.43975988e-01 -9.80369329e-01 -2.58121997e-01
-1.27226448e+00 3.05638164e-01 3.38332027e-01 1.37500092e-01
-2.87712902e-01 8.36607873e-01 6.04379058e-01 -4.40825284e-01
-3.11192691e-01 2.14426085e-01 -1.17263578e-01 1.76278636e-01
3.36323142e-01 1.29252195e-01 -2.38398060e-01 -1.09611821e+00
-1.52306363e-01 5.71660280e-01 1.98060140e-01 -4.81698513e-02
1.24373376e+00 2.02991590e-01 -5.04304707e-01 1.80607259e-01
1.01300550e+00 3.78515244e-01 -1.21606588e+00 2.66003370e-01
-9.11387682e-01 -7.93526888e-01 -4.68797833e-01 -6.26203597e-01
-1.76498973e+00 8.22328746e-01 4.80723739e-01 -1.05339939e-04
1.38511395e+00 -1.28090233e-01 6.48632467e-01 -3.58565986e-01
2.37005606e-01 -7.16502666e-01 1.07128918e-01 -8.35254381e-04
1.07809734e+00 -1.01224673e+00 -1.49800986e-01 -4.83342499e-01
-7.97833264e-01 8.64372671e-01 7.98714101e-01 -2.17097253e-01
3.54260445e-01 1.36346236e-01 3.17645878e-01 -1.30992368e-01
-6.03716314e-01 3.43751729e-01 1.24443322e-01 6.87698662e-01
2.85402536e-01 3.38474959e-01 -1.78113580e-02 4.51625168e-01
-9.42769349e-01 -3.13692838e-01 6.30565524e-01 2.84742773e-01
4.94319379e-01 -1.31590426e+00 -3.49251479e-01 3.39848697e-01
-4.85618860e-01 7.22322464e-02 -5.24993300e-01 6.75243437e-01
4.20601547e-01 5.20788670e-01 2.12183237e-01 -4.51825589e-01
8.75013843e-02 2.49539778e-01 6.52704358e-01 -3.99253815e-01
-2.65469462e-01 -3.18867147e-01 -6.67522624e-02 -6.77015781e-01
-4.71490055e-01 -5.29200315e-01 -5.71952879e-01 -7.12856650e-01
2.41636470e-01 -1.28021436e-02 6.17608249e-01 5.87098897e-01
3.66466045e-01 3.35417122e-01 9.87496734e-01 -5.66946805e-01
-2.81820387e-01 -8.53782952e-01 -5.15388489e-01 8.56295466e-01
1.25935078e-01 -6.65093720e-01 -4.69504803e-01 1.91896707e-01]
|
[12.761621475219727, 0.2664283514022827]
|
9d690440-c733-4510-ba23-209787c87828
|
road-images-augmentation-with-synthetic
|
2101.04927
| null |
https://arxiv.org/abs/2101.04927v1
|
https://arxiv.org/pdf/2101.04927v1.pdf
|
Road images augmentation with synthetic traffic signs using neural networks
|
Traffic sign recognition is a well-researched problem in computer vision. However, the state of the art methods works only for frequent sign classes, which are well represented in training datasets. We consider the task of rare traffic sign detection and classification. We aim to solve that problem by using synthetic training data. Such training data is obtained by embedding synthetic images of signs in the real photos. We propose three methods for making synthetic signs consistent with a scene in appearance. These methods are based on modern generative adversarial network (GAN) architectures. Our proposed methods allow realistic embedding of rare traffic sign classes that are absent in the training set. We adapt a variational autoencoder for sampling plausible locations of new traffic signs in images. We demonstrate that using a mixture of our synthetic data with real data improves the accuracy of both classifier and detector.
|
['Vlad Shakhuro', 'Boris Faizov', 'Anton Konushin']
|
2021-01-13
| null | null | null | null |
['traffic-sign-recognition', 'traffic-sign-detection']
|
['computer-vision', 'computer-vision']
|
[ 4.97880578e-01 -1.63444862e-01 4.45330739e-02 -2.48203918e-01
-5.42200625e-01 -4.09223169e-01 8.83529007e-01 -1.18019724e+00
-1.35547638e-01 8.77930403e-01 -1.11147039e-01 -1.22703075e-01
3.06268066e-01 -6.99726939e-01 -1.06504595e+00 -8.28841150e-01
5.18900454e-01 5.62744141e-01 4.32605535e-01 -2.65204489e-01
1.63383394e-01 6.62928045e-01 -1.97999978e+00 3.37932557e-01
9.78127122e-01 6.21753514e-01 -2.36733764e-01 7.58998215e-01
-1.82911754e-01 9.40291703e-01 -9.25139725e-01 -6.05785549e-01
7.52931595e-01 -7.71611214e-01 -1.15453139e-01 4.15296853e-01
9.87279356e-01 -4.96999919e-01 -6.86676025e-01 1.25001192e+00
2.60106951e-01 -3.20849195e-02 1.09052002e+00 -1.79969907e+00
-7.80576944e-01 -2.44604275e-02 -3.58894169e-01 -1.56888455e-01
-1.13778980e-02 6.85326517e-01 5.67249298e-01 -7.80012965e-01
8.60795796e-01 1.07495713e+00 5.79478264e-01 9.99096334e-01
-1.06785214e+00 -7.19695747e-01 -2.49734297e-01 5.71634650e-01
-1.19838989e+00 -5.09918749e-01 1.24324834e+00 -5.82529962e-01
1.46403804e-01 2.96482414e-01 6.82176173e-01 1.55115569e+00
-2.22293094e-01 8.56028676e-01 1.44701183e+00 -6.22696936e-01
2.66537249e-01 2.36802056e-01 -6.37954995e-02 6.23030663e-01
3.71174514e-01 3.94673795e-01 -1.06364392e-01 2.68786103e-02
9.69974637e-01 8.00815448e-02 -2.11881533e-01 -6.26589000e-01
-1.07467926e+00 7.57976174e-01 3.35192889e-01 1.60247773e-01
-5.48206389e-01 3.92277092e-01 3.31957005e-02 2.60895550e-01
-3.66333015e-02 7.89544359e-02 -3.30824666e-02 5.62519245e-02
-9.42423582e-01 1.16781078e-01 6.75712705e-01 8.24697912e-01
4.60529089e-01 7.55042732e-01 -2.46804148e-01 7.85043120e-01
3.25463265e-01 1.22020447e+00 4.71395344e-01 -1.07348061e+00
2.12769419e-01 4.01129305e-01 1.68200046e-01 -6.47415340e-01
4.01843041e-01 -1.58806935e-01 -6.85262918e-01 8.55339289e-01
9.66643214e-01 -2.50966512e-02 -1.48894799e+00 1.54933977e+00
3.24294657e-01 5.07139325e-01 1.42607093e-01 7.69133627e-01
5.25227547e-01 6.38058782e-01 -1.72056571e-01 1.85999706e-01
1.00567925e+00 -8.88323367e-01 -5.34840763e-01 -8.69496092e-02
9.31552276e-02 -7.29384005e-01 1.02345824e+00 4.06525016e-01
-6.15221381e-01 -6.41153514e-01 -8.86063159e-01 2.09950566e-01
-5.71883678e-01 3.75961035e-01 3.55608761e-01 1.09332240e+00
-6.87201381e-01 3.77190143e-01 -4.81092811e-01 -3.53742301e-01
6.42051220e-01 -1.44884018e-02 -1.86079994e-01 -4.45623070e-01
-8.14384043e-01 9.42655444e-01 -1.10896401e-01 3.11938703e-01
-9.22131538e-01 -3.25051278e-01 -6.78017259e-01 -3.56933773e-01
1.58144668e-01 -6.63954735e-01 1.02935731e+00 -1.51804447e+00
-1.61638451e+00 1.01419711e+00 -1.63938299e-01 -2.05696315e-01
1.00705874e+00 1.93825081e-01 -6.74680054e-01 4.70462628e-02
-2.35673964e-01 5.83900928e-01 1.61029124e+00 -1.68791413e+00
-1.73843607e-01 7.88755789e-02 -2.99260318e-01 -4.67163622e-01
2.56325662e-01 -4.90694605e-02 3.69002298e-02 -5.53057253e-01
-2.19297722e-01 -9.80972469e-01 -5.91592714e-02 5.40773451e-01
-3.61328214e-01 1.67176485e-01 1.19137585e+00 -7.33095884e-01
3.56564015e-01 -1.92294884e+00 -3.83856028e-01 3.78998071e-01
-7.67967254e-02 7.20393002e-01 -3.03547382e-01 1.43910021e-01
-7.04408670e-03 -1.58564255e-01 -2.93105245e-01 -1.61179185e-01
2.56561190e-01 7.48337626e-01 -6.45261407e-01 5.19917190e-01
3.88345182e-01 1.08495796e+00 -7.04039037e-01 -4.96355355e-01
5.83856702e-01 5.93374133e-01 -3.95604938e-01 2.12812498e-01
-3.64077896e-01 6.82265937e-01 -3.48571360e-01 7.72860885e-01
9.85422492e-01 5.98645397e-02 -3.12483460e-01 -2.04401031e-01
1.60416707e-01 -1.43181890e-01 -1.04098785e+00 9.26944733e-01
-2.64332622e-01 8.67765069e-01 -2.46282607e-01 -1.10922110e+00
8.98061991e-01 -2.19354182e-02 2.36235246e-01 -6.86303556e-01
1.94828600e-01 3.60475838e-01 3.31305623e-01 -6.48883402e-01
1.99857011e-01 -3.18027169e-01 2.93170124e-01 3.92320842e-01
-9.38886032e-02 -3.99150640e-01 3.98252070e-01 -1.28358856e-01
9.58932638e-01 1.95499465e-01 -8.39364827e-02 3.80869389e-01
4.07451689e-01 9.67957750e-02 5.72787642e-01 7.56700873e-01
-3.82665932e-01 8.00027966e-01 3.77867937e-01 -4.35439289e-01
-1.53373706e+00 -1.23408389e+00 -1.26270130e-01 3.39164287e-01
-9.92666036e-02 3.81099522e-01 -6.28497839e-01 -1.04635072e+00
1.82749763e-01 8.67145121e-01 -7.98376560e-01 -4.53686938e-02
-8.16962719e-01 -6.38936758e-01 6.92183435e-01 5.17223954e-01
6.60731673e-01 -1.23262990e+00 -5.73255062e-01 -2.23854631e-01
4.14080732e-02 -1.12394381e+00 -2.16940686e-01 -2.77301937e-01
-5.85176647e-01 -1.42881715e+00 -1.20047414e+00 -6.64835632e-01
1.07088745e+00 4.91075590e-02 9.07122195e-01 -5.93121015e-02
-4.27531540e-01 4.49014723e-01 -3.79124761e-01 -3.90691847e-01
-1.03165901e+00 -7.88109303e-01 -1.12015732e-01 7.22350419e-01
2.22829074e-01 -3.03558290e-01 -6.55294478e-01 5.74628294e-01
-9.99108970e-01 -3.58601846e-02 7.15361357e-01 1.08831632e+00
3.22744012e-01 -2.89722115e-01 3.17975909e-01 -6.91012383e-01
1.68618992e-01 -1.81476504e-01 -6.01944447e-01 2.50650316e-01
-1.40660971e-01 2.33312204e-01 7.47270584e-01 -7.11619556e-01
-1.01090336e+00 1.73545614e-01 -1.91513702e-01 -8.89123619e-01
-5.76853991e-01 -3.61099333e-01 -1.07778721e-01 -4.56245661e-01
6.28822625e-01 6.51165128e-01 2.02722326e-01 -2.49471545e-01
5.20480335e-01 5.66602588e-01 4.80633318e-01 -3.72850060e-01
1.13806796e+00 7.96480119e-01 2.16750503e-01 -1.00093555e+00
-3.65839154e-01 -1.24880902e-01 -6.43713057e-01 -5.85709989e-01
5.84591091e-01 -3.80093992e-01 -3.39305639e-01 7.74737895e-01
-1.04096484e+00 -4.24325764e-01 -9.06901717e-01 3.78911138e-01
-7.96257496e-01 5.68423331e-01 -8.38259831e-02 -8.81612837e-01
1.09683417e-01 -1.14490926e+00 1.21025634e+00 1.35051712e-01
3.14599365e-01 -7.56416380e-01 1.76055565e-01 2.34523475e-01
5.38965166e-01 5.72528422e-01 5.92132449e-01 -3.80020201e-01
-9.61731851e-01 -5.76967359e-01 -2.47881547e-01 9.43006396e-01
1.15470245e-01 3.15066636e-01 -1.14011717e+00 2.36526266e-01
-2.27214202e-01 -3.25704694e-01 1.13114893e+00 4.80443895e-01
9.79356766e-01 -3.12879086e-01 -2.11160630e-01 5.98855376e-01
1.22467971e+00 3.38474959e-01 9.54095185e-01 -5.60355969e-02
7.56996036e-01 5.60866237e-01 3.65414828e-01 6.79122806e-02
-6.39848784e-02 6.25044942e-01 2.73640454e-01 -1.00089498e-01
-6.69119477e-01 -4.77614254e-01 4.33897108e-01 5.12310863e-01
-1.96599677e-01 -2.82693893e-01 -9.07000542e-01 5.56588769e-01
-1.79382288e+00 -1.44983482e+00 -3.54924381e-01 2.15814948e+00
3.25803310e-01 -3.98824811e-02 1.51135787e-01 4.64362681e-01
8.87261569e-01 1.57036353e-02 -6.40739560e-01 -1.75918430e-01
-3.83485943e-01 3.86985630e-01 6.31174445e-01 1.37973621e-01
-1.01351762e+00 7.70438552e-01 6.42212915e+00 6.97733164e-01
-1.28784835e+00 1.42238334e-01 3.52438331e-01 3.14631164e-01
-1.87761381e-01 -1.40333436e-02 -5.19315004e-01 7.17550516e-01
5.73636532e-01 2.13571832e-01 2.90010005e-01 8.09580564e-01
1.60308078e-01 1.04839534e-01 -9.78264630e-01 1.08866096e+00
4.53057289e-01 -1.14226496e+00 3.03777397e-01 6.29022298e-03
1.11767018e+00 -8.52516964e-02 4.32538390e-01 2.84037262e-01
5.26223004e-01 -9.83961105e-01 6.95100904e-01 9.99751449e-01
9.84553337e-01 -2.04355016e-01 5.89593828e-01 3.26435059e-01
-7.45378613e-01 -2.27605328e-02 -2.08318904e-01 3.33443969e-01
5.07342935e-01 4.15790409e-01 -6.88785732e-01 9.85489786e-02
1.01245660e-02 7.08747268e-01 -7.45807469e-01 1.42615616e+00
-4.57578838e-01 7.84191310e-01 -5.77112854e-01 -1.55195132e-01
8.00265893e-02 -4.19872433e-01 5.74000478e-01 8.13113093e-01
3.76949310e-01 -3.78637314e-01 -1.07397020e-01 1.27306759e+00
6.28945008e-02 -3.05590451e-01 -1.05428839e+00 6.89710602e-02
4.42228541e-02 6.20236337e-01 -5.96296668e-01 -5.45630276e-01
-2.54774868e-01 1.16491151e+00 -2.59840816e-01 5.91883063e-01
-9.99422908e-01 -2.92725950e-01 4.73330140e-01 8.20736140e-02
4.48498756e-01 -1.50094092e-01 -9.85908285e-02 -1.48712087e+00
2.35655740e-01 -9.35648918e-01 5.85936010e-02 -1.04392517e+00
-1.32335508e+00 3.65832239e-01 -8.20810720e-02 -1.76176238e+00
-5.21807253e-01 -9.87130702e-01 -9.78487849e-01 5.96729577e-01
-1.57068479e+00 -1.55844569e+00 -5.57421625e-01 7.11734653e-01
5.43782055e-01 -3.59600186e-01 5.22422910e-01 3.74063402e-01
-3.02267939e-01 5.74028552e-01 5.63424051e-01 3.32718760e-01
6.93879902e-01 -1.14484477e+00 4.39976454e-01 1.05438173e+00
4.14384484e-01 1.66393712e-01 6.45901740e-01 -5.00469565e-01
-1.04919767e+00 -1.12079883e+00 6.35493100e-01 -5.65170944e-01
5.29246867e-01 -1.62556887e-01 -6.73704386e-01 6.27598464e-01
-7.85850734e-02 5.27365148e-01 1.15651317e-01 -7.40360558e-01
-4.95014817e-01 -1.75568745e-01 -1.33637786e+00 6.13575101e-01
1.01034009e+00 -4.44672167e-01 -7.53280461e-01 3.33757907e-01
-2.41862431e-01 -1.99775472e-02 -1.74504712e-01 3.19521815e-01
7.17744648e-01 -7.55477011e-01 1.20342672e+00 -6.86343431e-01
3.35554034e-01 -5.88684797e-01 -2.43756875e-01 -1.45089626e+00
3.33548307e-01 -4.64085758e-01 -2.08980635e-01 1.03468025e+00
1.29375840e-02 -7.42128134e-01 1.00759864e+00 4.72508550e-01
8.25909376e-02 -1.54738605e-01 -9.35667396e-01 -1.14377141e+00
1.33429632e-01 -3.76479775e-01 3.73925209e-01 7.29296744e-01
-7.76543975e-01 -2.04310343e-01 -6.18107796e-01 -1.53180048e-01
1.18183911e+00 1.42478257e-01 1.35824645e+00 -1.16567087e+00
-2.85127848e-01 -4.65392232e-01 -8.45752895e-01 -9.45510030e-01
2.62334824e-01 -6.71025157e-01 3.80900390e-02 -1.28827679e+00
-1.69461891e-01 -4.66811895e-01 -6.33339398e-03 1.99588895e-01
1.60855785e-01 6.56217217e-01 2.51139492e-01 1.56548843e-01
-2.44779557e-01 8.76638651e-01 1.54394281e+00 -4.33738023e-01
2.59617597e-01 1.50260553e-01 2.72338111e-02 9.20642436e-01
7.76064754e-01 -4.58440453e-01 -1.49199693e-02 -6.20120689e-02
-3.17587197e-01 -2.72072285e-01 9.81989145e-01 -1.11546826e+00
6.43034726e-02 -2.58531213e-01 5.07682562e-01 -6.41707718e-01
5.24399817e-01 -9.61241603e-01 1.46019965e-01 5.83566010e-01
-3.20226438e-02 -3.32828194e-01 -7.75734484e-02 7.03861117e-01
-2.13403136e-01 -3.20070058e-01 1.17491913e+00 5.96040040e-02
-9.45241690e-01 1.17360897e-01 -3.63404840e-01 9.71599575e-03
1.12748539e+00 -5.77117205e-01 -3.93400580e-01 -5.17036438e-01
-6.14959776e-01 -2.67056584e-01 4.46853906e-01 3.75543743e-01
8.44885886e-01 -1.44433570e+00 -7.08569348e-01 6.68235719e-01
2.78723240e-01 -3.94327849e-01 1.34077594e-01 7.12001026e-01
-8.80941451e-01 1.51776507e-01 -6.52319133e-01 -5.73532403e-01
-1.25122154e+00 4.50365841e-01 6.30999565e-01 1.86616890e-02
-6.33893013e-01 3.80525291e-01 3.35985124e-02 -3.35778415e-01
5.99969029e-02 -4.94074553e-01 8.59677568e-02 -3.67304206e-01
2.15442628e-01 3.54636043e-01 -3.65462214e-01 -9.55027342e-01
-1.20280832e-01 9.00122464e-01 2.60353595e-01 -1.10250816e-01
1.06553388e+00 3.96981984e-01 4.32699323e-01 3.15723777e-01
1.00600111e+00 1.88583896e-01 -1.41477942e+00 -7.24662393e-02
-2.65000165e-01 -8.29608321e-01 -2.41719782e-01 -7.42715418e-01
-1.17559242e+00 9.97331798e-01 7.75837958e-01 -1.64580643e-01
8.56401205e-01 -1.80648759e-01 7.21249223e-01 4.63945359e-01
4.08074051e-01 -9.36342955e-01 1.64294139e-01 3.94005954e-01
9.03267682e-01 -1.50654578e+00 -4.18984681e-01 -2.31230319e-01
-6.91999435e-01 1.02304959e+00 4.41846102e-01 -7.47247636e-01
5.62740505e-01 -6.64879382e-02 2.99596131e-01 -6.50460348e-02
-3.45156044e-01 -7.08504438e-01 4.22119141e-01 9.72964168e-01
-2.97418028e-01 5.83039001e-02 -1.65002659e-01 4.25767154e-02
2.39507258e-02 1.83280241e-02 7.19163775e-01 7.91831493e-01
-1.79689422e-01 -1.18572366e+00 -6.85948253e-01 5.09356380e-01
-6.18418716e-02 1.12279765e-01 -6.44467175e-01 9.87691998e-01
4.14699882e-01 5.47132552e-01 -2.26460639e-02 -1.49258614e-01
3.63816738e-01 4.96858150e-01 7.90923893e-01 -1.29876554e-01
-3.98834124e-02 -4.29516226e-01 -3.56602408e-02 -3.15977514e-01
-4.82201666e-01 -9.35299814e-01 -5.81672370e-01 -3.15321870e-02
-2.40654081e-01 -1.76805139e-01 8.41613650e-01 7.73327470e-01
6.15755543e-02 4.32596892e-01 6.38072789e-01 -9.07177150e-01
-7.71795511e-01 -8.58271956e-01 -6.41513050e-01 9.31335270e-01
5.49369335e-01 -9.94347692e-01 -5.17627835e-01 4.12527293e-01]
|
[8.064159393310547, -0.8436663746833801]
|
d16b3732-5415-43a2-9bf0-43ace7f6a04b
|
inter-and-intra-patient-ecg-heartbeat
|
1812.07421
| null |
http://arxiv.org/abs/1812.07421v1
|
http://arxiv.org/pdf/1812.07421v1.pdf
|
Inter- and intra- patient ECG heartbeat classification for arrhythmia detection: a sequence to sequence deep learning approach
|
Electrocardiogram (ECG) signal is a common and powerful tool to study heart
function and diagnose several abnormal arrhythmia. While there have been
remarkable improvements in cardiac arrhythmia classification methods, they
still cannot offer an acceptable performance in detecting different heart
conditions, especially when dealing with imbalanced datasets. In this paper, we
propose a solution to address this limitation of current classification
approaches by developing an automatic heartbeat classification method using
deep convolutional neural networks and sequence to sequence models. We
evaluated the proposed method on the MIT-BIH arrhythmia database, considering
the intra-patient and inter-patient paradigms, and the AAMI EC57 standard. The
evaluation results for both paradigms show that our method achieves the best
performance in the literature (a positive predictive value of 96.46% and
sensitivity of 100% for the category S, and a positive predictive value of
98.68% and sensitivity of 97.40% for the category F for the intra-patient
scheme; a positive predictive value of 92.57% and sensitivity of 88.94% for the
category S, and a positive predictive value of 99.50% and sensitivity of 99.94%
for the category V for the inter-patient scheme.)
|
['Fatemeh Afghah', 'Sajad Mousavi']
|
2018-12-09
|
inter-and-intra-patient-ecg-heartbeat-1
| null | null |
arxiv181207421-2018-12
|
['arrhythmia-detection', 'heartbeat-classification']
|
['medical', 'medical']
|
[ 5.50471097e-02 -3.54786545e-01 -9.58676189e-02 -2.85883963e-01
-5.89801431e-01 -5.00125706e-01 -4.99438606e-02 3.47477525e-01
-3.58384490e-01 8.07282805e-01 -4.30529416e-01 -3.85887891e-01
-3.26425612e-01 -5.32219470e-01 -4.01362814e-02 -6.28573060e-01
-3.58108014e-01 5.19051373e-01 -1.61443666e-01 2.30148777e-01
1.17476873e-01 5.41721582e-01 -1.28707349e+00 4.15224880e-01
8.69403601e-01 1.49083722e+00 -4.58750099e-01 8.70915294e-01
2.87500352e-01 6.16128743e-01 -9.55822527e-01 -8.26940089e-02
2.24426299e-01 -8.59688222e-01 -5.86204827e-01 -2.17865095e-01
-1.71544328e-02 -7.10535496e-02 1.75142780e-01 6.55545652e-01
1.19518077e+00 -5.72492361e-01 5.36356091e-01 -1.18837953e+00
-2.28553042e-02 4.62392777e-01 -6.57006919e-01 6.29403353e-01
9.45421606e-02 5.26430458e-02 7.59738922e-01 -5.50980389e-01
4.78724182e-01 2.22866744e-01 1.16955972e+00 4.11001056e-01
-1.02415419e+00 -8.06982934e-01 -6.60729706e-01 -4.98256087e-03
-1.59896922e+00 -1.48234919e-01 7.35418618e-01 -5.15620828e-01
7.86677063e-01 4.47633654e-01 1.21975732e+00 6.12680614e-01
4.58605498e-01 3.39685708e-01 9.47570562e-01 -2.76295096e-01
1.81984186e-01 -4.25115265e-02 3.66415143e-01 4.28047180e-01
4.17884469e-01 1.43868491e-01 -2.71635503e-01 -4.94651407e-01
7.69527793e-01 4.97419424e-02 -4.02183473e-01 -1.90091338e-02
-1.41528141e+00 5.42921066e-01 1.85802802e-01 8.18251014e-01
-5.95233262e-01 -1.46469936e-01 8.83915067e-01 2.39695400e-01
1.22988172e-01 6.75988555e-01 -5.28789878e-01 -4.20604885e-01
-1.16973555e+00 4.27474797e-01 6.86565161e-01 3.29449952e-01
1.02046102e-01 3.21529478e-01 -4.08464372e-01 7.82692969e-01
-1.00048527e-01 4.37984288e-01 8.14534485e-01 -7.19814897e-01
1.93770915e-01 5.80106974e-01 1.72049299e-01 -9.19643402e-01
-1.07715380e+00 -1.15826023e+00 -1.39560521e+00 -2.18298107e-01
4.76210713e-01 -2.40609199e-01 -6.42259598e-01 1.53994977e+00
-5.22941798e-02 2.12869868e-01 1.47161677e-01 8.27651620e-01
9.52519119e-01 4.10277218e-01 -2.20853295e-02 -7.49563575e-01
1.25383914e+00 -3.47378910e-01 -7.24872708e-01 8.81654397e-02
8.01597416e-01 -5.47909319e-01 7.72011936e-01 7.48389602e-01
-9.70460117e-01 -6.84705198e-01 -1.16896975e+00 5.88243604e-01
2.11289376e-01 6.51912689e-01 4.34968412e-01 8.53621304e-01
-9.17935431e-01 4.82468188e-01 -6.47743762e-01 -3.40436101e-01
5.26058853e-01 3.45150381e-01 -1.11494802e-01 3.05030763e-01
-1.42818761e+00 5.07699192e-01 4.59705800e-01 1.50676414e-01
-6.07692957e-01 -8.53424668e-01 -3.07831794e-01 9.11344066e-02
-2.73031384e-01 -7.03361690e-01 9.14447248e-01 -8.22343171e-01
-1.10950577e+00 1.11903775e+00 1.83207065e-01 -7.56552100e-01
6.44579291e-01 -7.70197622e-03 -6.75413132e-01 1.96571663e-01
-6.08507320e-02 2.59899199e-01 3.97560030e-01 -7.21978128e-01
-5.96829474e-01 -5.75393438e-01 -3.22579980e-01 -4.23362255e-02
-5.28223291e-02 -5.43524884e-02 -3.59138325e-02 -6.38077915e-01
3.43963146e-01 -1.01929688e+00 -5.58294021e-02 -4.61551994e-01
-3.59330654e-01 4.54475023e-02 5.61953127e-01 -9.10700619e-01
1.66292417e+00 -2.08967185e+00 -1.94139197e-01 3.62016171e-01
4.92505640e-01 8.83037508e-01 3.83023083e-01 1.75201759e-01
-4.00309652e-01 2.49405921e-01 -5.55696130e-01 2.40062401e-01
-6.62008345e-01 -8.33468139e-02 -8.80035535e-02 3.65230769e-01
4.03589718e-02 8.26325715e-01 -5.66644549e-01 -1.49133384e-01
2.14293733e-01 5.30410051e-01 -3.03244501e-01 2.77244419e-01
4.77021217e-01 6.00787103e-01 -2.89812535e-01 5.54415703e-01
3.99324864e-01 -2.67606199e-01 4.56807345e-01 -1.26817048e-01
1.99111357e-01 3.86392064e-02 -1.19671905e+00 1.40388501e+00
-1.33977190e-01 5.39761841e-01 -3.45188349e-01 -1.19870269e+00
1.45206523e+00 6.60224617e-01 8.58651757e-01 -5.66841662e-01
2.46042222e-01 5.87117314e-01 6.75006628e-01 -5.94329834e-01
-7.59162083e-02 -3.53620708e-01 4.48420383e-02 3.16266567e-01
-1.57565430e-01 7.22164288e-02 6.66884556e-02 -2.60030944e-02
1.12264562e+00 -2.34697089e-01 6.31223679e-01 -3.89370382e-01
6.94116890e-01 -4.52072471e-01 7.99182594e-01 7.40745008e-01
-5.91316104e-01 9.65296924e-01 6.44930243e-01 -1.04127944e+00
-8.05306017e-01 -7.02772141e-01 -4.34745163e-01 6.36631101e-02
-1.74521849e-01 -3.48289549e-01 -7.20447004e-01 -4.38440889e-01
-9.84675959e-02 1.71760142e-01 -4.38777030e-01 -4.14154679e-01
-5.22195578e-01 -1.24903762e+00 1.04804397e+00 6.59591675e-01
7.43124902e-01 -1.06586492e+00 -1.15573323e+00 3.94259721e-01
-4.40636426e-01 -9.54039633e-01 2.59140313e-01 1.71732605e-01
-1.01694560e+00 -1.18979263e+00 -8.33476186e-01 -6.55455589e-01
-1.70314424e-02 -4.87948149e-01 1.27503026e+00 2.95751601e-01
-6.85052693e-01 -2.76104599e-01 -1.01605445e-01 -5.46095550e-01
-4.20417607e-01 3.92425470e-02 1.74307853e-01 1.97466880e-01
1.14814945e-01 -3.58702213e-01 -7.10325301e-01 2.71039486e-01
-4.80926186e-01 -1.51655748e-01 3.57812464e-01 9.70873892e-01
6.41179144e-01 -2.35111237e-01 1.29885638e+00 -8.09653878e-01
6.32305443e-01 -2.52181441e-01 -3.24572176e-01 5.84610216e-02
-1.06993568e+00 -5.95257580e-01 4.99524862e-01 -1.13084026e-01
-2.22157344e-01 7.97907934e-02 -5.16825914e-01 -4.41506714e-01
-1.43652603e-01 5.32284677e-01 9.30466354e-02 2.83179849e-01
7.20370352e-01 2.35049501e-01 3.42359319e-02 -1.94986999e-01
-5.44685006e-01 8.18852484e-01 5.34303367e-01 -2.01768294e-01
2.10592732e-01 1.70007646e-01 1.32205307e-01 -6.49948776e-01
-5.22429168e-01 -6.32084131e-01 -6.71155691e-01 -1.74669608e-01
8.56740415e-01 -8.85424674e-01 -6.42709315e-01 7.47267008e-01
-9.23125982e-01 5.93951009e-02 -3.12490910e-01 4.69988763e-01
-5.43127298e-01 4.28636670e-01 -5.09647965e-01 -8.45476210e-01
-1.01205134e+00 -1.11468387e+00 7.15716183e-01 -6.99366704e-02
-8.07980239e-01 -7.74820328e-01 4.91826124e-02 1.09713487e-01
6.00457609e-01 8.54514718e-01 7.62208343e-01 -9.95554507e-01
1.69891283e-01 -6.34574175e-01 -1.57428414e-01 4.59034413e-01
2.84826666e-01 3.78505401e-02 -1.03554428e+00 -3.44162166e-01
1.76620092e-02 5.95367365e-02 5.59676588e-01 6.11390054e-01
1.42201281e+00 1.87295690e-01 -3.28955084e-01 6.19312525e-01
1.31420469e+00 7.89886415e-01 8.75890136e-01 5.58579639e-02
5.98096013e-01 9.85178500e-02 3.70204598e-01 5.30146956e-01
4.90222462e-02 6.90767169e-01 9.84876975e-02 -2.66021609e-01
-2.00814828e-02 2.87307292e-01 -1.82061389e-01 6.56898260e-01
-2.35313356e-01 -1.97453305e-01 -1.51428807e+00 5.68999410e-01
-1.73991776e+00 -9.76038754e-01 -4.80746210e-01 2.43264341e+00
6.59699500e-01 1.61153600e-01 4.15374339e-01 1.01538348e+00
8.56358230e-01 -1.56956479e-01 -4.50302035e-01 -5.75578570e-01
-2.07039416e-01 2.49375001e-01 -1.41956389e-01 3.56579684e-02
-1.21121061e+00 1.38602406e-01 5.90517712e+00 3.54927331e-01
-1.50157845e+00 -7.89791271e-02 1.06252491e+00 1.53960392e-01
4.37750459e-01 -4.56335336e-01 -4.43097115e-01 5.90677381e-01
1.22264850e+00 -2.83966064e-02 -1.27547115e-01 4.89476293e-01
9.57697853e-02 2.09112063e-01 -9.71405029e-01 1.62531638e+00
1.12507388e-01 -1.47207308e+00 -2.77785778e-01 -1.21153235e-01
4.81737196e-01 -9.72071737e-02 -1.75777465e-01 1.67580858e-01
-8.34149718e-01 -1.05363262e+00 3.88684183e-01 4.04157072e-01
1.20463181e+00 -8.55496764e-01 1.27475107e+00 3.77641559e-01
-9.74096477e-01 -1.33767411e-01 1.92159787e-02 -1.82852089e-01
-1.21280514e-01 9.67524230e-01 -8.22106242e-01 6.31800950e-01
9.58032072e-01 9.22291160e-01 -3.49312752e-01 1.17572331e+00
1.02411643e-01 9.03708756e-01 -2.40997970e-01 1.37026936e-01
-2.60932058e-01 9.46033671e-02 5.30033052e-01 1.09759974e+00
3.35870534e-01 -1.20403375e-02 2.39024580e-01 5.42904437e-01
-1.59296431e-02 2.12218583e-01 -5.29702246e-01 1.58298582e-01
2.92975366e-01 1.11098528e+00 -7.58165538e-01 -5.17740786e-01
1.43570200e-01 5.16425312e-01 -1.09476402e-01 1.39328673e-01
-1.03593016e+00 -7.91516542e-01 8.81555751e-02 1.87179580e-01
-1.18043520e-01 4.68283474e-01 -8.71623456e-01 -9.95124698e-01
9.74618122e-02 -1.00850511e+00 6.54279053e-01 -5.36258101e-01
-9.34444964e-01 8.93066943e-01 -3.20297122e-01 -1.61709976e+00
-4.22515452e-01 -2.61876017e-01 -6.36362314e-01 1.07238615e+00
-1.10246551e+00 -7.91212916e-01 -4.47379529e-01 3.00728738e-01
8.86202082e-02 -3.45939666e-01 1.07264948e+00 5.19211471e-01
-4.77287203e-01 8.65978062e-01 -6.78408444e-02 2.60886490e-01
6.09368622e-01 -1.13749671e+00 9.62497145e-02 7.31923699e-01
-2.01774701e-01 4.01444137e-01 3.32809329e-01 -2.60488391e-01
-9.89433348e-01 -1.11005735e+00 1.16843665e+00 -3.28083664e-01
4.08435725e-02 1.40962228e-01 -9.81674850e-01 2.14322880e-01
-2.20570251e-01 2.20659643e-01 9.01114881e-01 1.10015295e-01
-1.41038850e-01 -3.32007885e-01 -1.37881422e+00 1.60768926e-02
4.82642055e-01 -3.02080542e-01 -2.53932148e-01 1.48368984e-01
-6.94331601e-02 -5.96643090e-01 -1.41189170e+00 9.54302847e-01
1.02263248e+00 -1.20355833e+00 8.61295581e-01 -3.27143937e-01
2.70589262e-01 -3.73941988e-01 1.41601309e-01 -1.03828049e+00
-3.63543779e-01 -4.93015349e-01 -8.08151215e-02 9.96762455e-01
5.24040282e-01 -8.06652725e-01 6.24239743e-01 1.90050438e-01
-2.48210013e-01 -1.27895784e+00 -9.59704220e-01 -6.10293090e-01
-7.41344839e-02 -4.48893040e-01 5.01965344e-01 1.21494496e+00
-1.52903453e-01 2.46093273e-01 -4.90590543e-01 -9.86808762e-02
2.06028149e-01 3.82255226e-01 4.78300720e-01 -1.61110568e+00
-1.35794595e-01 -3.47788334e-01 -6.45586312e-01 9.93481874e-02
-3.61725450e-01 -9.08393860e-01 -3.99049431e-01 -1.46794820e+00
3.35401028e-01 -6.17587268e-01 -8.51392627e-01 4.74484116e-01
-2.54170746e-01 6.73809171e-01 1.60497259e-02 3.60557705e-01
-6.81264773e-02 -2.21945375e-01 7.73991406e-01 6.31516203e-02
-4.85950261e-01 2.51087368e-01 -4.75393444e-01 6.13571167e-01
1.24995685e+00 -3.43509853e-01 -4.16570336e-01 8.85440335e-02
2.23753363e-01 3.78441602e-01 3.07478249e-01 -1.53519893e+00
-3.99951935e-01 4.25753295e-01 7.32862353e-01 -5.97697318e-01
8.90990999e-03 -4.75377381e-01 5.52838027e-01 1.20340586e+00
-3.29294324e-01 2.63437122e-01 2.36759797e-01 2.05812588e-01
-4.30768907e-01 2.25892544e-01 1.01809943e+00 1.88598678e-01
-2.55771279e-01 2.51796395e-01 -2.46701226e-01 2.05218285e-01
1.09381735e+00 -3.49298805e-01 -1.36303872e-01 -1.90246105e-01
-8.14202905e-01 -3.39691192e-02 6.63536936e-02 5.25346160e-01
7.05781281e-01 -1.18962669e+00 -8.32036674e-01 5.17539680e-01
4.45699394e-01 -4.25541729e-01 2.87362963e-01 1.16425705e+00
-7.18519866e-01 4.22393858e-01 -5.82589030e-01 -1.00237715e+00
-1.30204618e+00 1.57301933e-01 8.68960381e-01 -3.34087759e-01
-6.18730903e-01 6.76581800e-01 -1.06124036e-01 -7.05080107e-02
1.22835644e-01 -5.45406342e-01 -3.92777562e-01 1.92224130e-01
5.51698506e-01 6.05579019e-01 6.09531641e-01 -5.72852969e-01
-6.88311875e-01 7.67576158e-01 2.96648860e-01 4.21014488e-01
9.86688077e-01 2.71253556e-01 -1.65791601e-01 7.16053784e-01
9.88972247e-01 -3.29169959e-01 -2.54857361e-01 3.24003309e-01
-2.86610037e-01 -1.59245834e-01 -2.88836602e-02 -1.44355321e+00
-1.16920698e+00 1.33299911e+00 1.30622900e+00 3.71108860e-01
1.39636099e+00 -5.76477170e-01 8.14116180e-01 4.94895317e-02
3.31692368e-01 -7.18562007e-01 -3.98840576e-01 2.89841563e-01
6.35949790e-01 -9.30397391e-01 -7.57152066e-02 -1.85116902e-01
-8.27690244e-01 1.14721751e+00 3.06917578e-01 -6.44654129e-03
7.64279306e-01 7.75802433e-02 3.21287155e-01 -2.13339359e-01
-5.44525146e-01 3.74681443e-01 1.96589857e-01 5.21979630e-01
8.51774573e-01 3.33035320e-01 -8.80755782e-01 9.09571350e-01
2.11112518e-02 4.44549620e-01 4.76275831e-01 8.10746610e-01
-1.27372891e-01 -7.70424008e-01 -8.10659453e-02 7.47115552e-01
-1.02755892e+00 1.62919298e-01 -3.01853865e-01 7.08309829e-01
3.25345039e-01 8.11512887e-01 -1.16343508e-02 -4.87833500e-01
2.89006412e-01 5.68977892e-01 2.26952568e-01 -4.03011143e-01
-1.09093070e+00 1.34711042e-01 1.29478037e-01 -3.36944699e-01
-4.26976651e-01 -4.45223510e-01 -1.35065722e+00 1.74774513e-01
-1.19342357e-01 3.27354133e-01 5.37945807e-01 6.47397876e-01
6.44102037e-01 7.54526794e-01 6.04369104e-01 -3.78444612e-01
-4.00058657e-01 -9.86721337e-01 -7.82116830e-01 4.24635768e-01
4.16991293e-01 -3.54487300e-01 -1.41192973e-01 4.94301990e-02]
|
[14.323874473571777, 3.2819316387176514]
|
60d0904c-23b2-405d-814c-abc723c602bf
|
just-go-with-the-flow-self-supervised-scene
|
1912.00497
| null |
https://arxiv.org/abs/1912.00497v2
|
https://arxiv.org/pdf/1912.00497v2.pdf
|
Just Go with the Flow: Self-Supervised Scene Flow Estimation
|
When interacting with highly dynamic environments, scene flow allows autonomous systems to reason about the non-rigid motion of multiple independent objects. This is of particular interest in the field of autonomous driving, in which many cars, people, bicycles, and other objects need to be accurately tracked. Current state-of-the-art methods require annotated scene flow data from autonomous driving scenes to train scene flow networks with supervised learning. As an alternative, we present a method of training scene flow that uses two self-supervised losses, based on nearest neighbors and cycle consistency. These self-supervised losses allow us to train our method on large unlabeled autonomous driving datasets; the resulting method matches current state-of-the-art supervised performance using no real world annotations and exceeds state-of-the-art performance when combining our self-supervised approach with supervised learning on a smaller labeled dataset.
|
['David Held', 'Himangi Mittal', 'Brian Okorn']
|
2019-12-01
|
just-go-with-the-flow-self-supervised-scene-1
|
http://openaccess.thecvf.com/content_CVPR_2020/html/Mittal_Just_Go_With_the_Flow_Self-Supervised_Scene_Flow_Estimation_CVPR_2020_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2020/papers/Mittal_Just_Go_With_the_Flow_Self-Supervised_Scene_Flow_Estimation_CVPR_2020_paper.pdf
|
cvpr-2020-6
|
['scene-flow-estimation']
|
['computer-vision']
|
[-4.16784100e-02 4.02692147e-02 -3.93478096e-01 -5.25845110e-01
-3.95724207e-01 -5.16618907e-01 8.94224703e-01 -9.12665576e-02
-7.07131386e-01 6.79502070e-01 -6.93416670e-02 -2.33762950e-01
7.13025033e-02 -7.18069851e-01 -7.95372725e-01 -3.48047167e-01
-2.61765927e-01 7.49849796e-01 9.55323100e-01 -3.34082574e-01
2.30869383e-01 6.13660932e-01 -1.81444728e+00 9.06440488e-04
7.11271107e-01 6.33948088e-01 2.34087795e-01 9.77712870e-01
-4.96971384e-02 1.36374152e+00 1.09117543e-02 -1.93965569e-01
4.09146965e-01 -2.05081746e-01 -8.57617319e-01 -2.49591619e-02
7.40908504e-01 -3.52323085e-01 -8.42406631e-01 7.21072495e-01
-3.65505591e-02 4.84871686e-01 5.72482705e-01 -1.48490119e+00
4.64920439e-02 -5.51919593e-03 -2.29239076e-01 5.09830713e-01
9.79909524e-02 5.94372511e-01 7.97218442e-01 -5.72067082e-01
8.36696327e-01 1.25195706e+00 6.94809020e-01 4.95746583e-01
-1.19074762e+00 -5.32264650e-01 2.14673534e-01 5.18961310e-01
-9.57080066e-01 -6.23595953e-01 7.93571413e-01 -6.75528407e-01
1.01536846e+00 -3.13704312e-01 6.49772584e-01 7.30369866e-01
4.85709786e-01 7.44306684e-01 6.17666066e-01 -3.62149626e-02
4.29580569e-01 2.33477965e-01 3.10999826e-02 9.31059599e-01
1.67552143e-01 6.87532306e-01 -4.83973444e-01 3.89671117e-01
3.78740460e-01 -1.57455385e-01 1.78407565e-01 -9.83809471e-01
-1.29241931e+00 8.16291392e-01 5.36272824e-01 -1.18318517e-02
3.30218859e-02 5.24038315e-01 5.43949544e-01 7.49571398e-02
4.26409930e-01 1.95821434e-01 -4.06306475e-01 -3.42246801e-01
-9.51861739e-01 4.44211811e-01 8.86371315e-01 9.19366717e-01
1.24142313e+00 1.71140566e-01 1.58630028e-01 2.39707083e-01
3.13190222e-01 6.26852334e-01 3.50168228e-01 -1.45574689e+00
4.25521433e-01 3.48088205e-01 1.78022057e-01 -6.94375098e-01
-5.40057838e-01 -1.53438821e-01 -5.96883297e-01 7.72575796e-01
6.35811567e-01 -4.61529866e-02 -8.25386882e-01 1.53392267e+00
3.43306929e-01 5.85765600e-01 3.26266170e-01 8.21201980e-01
4.80034858e-01 5.96846700e-01 -1.03900731e-02 1.11036934e-01
7.96793103e-01 -1.34867620e+00 -5.02516985e-01 -5.49594998e-01
9.34315741e-01 -4.59980935e-01 8.39962184e-01 2.03038797e-01
-7.72383690e-01 -1.06248176e+00 -1.02146077e+00 -1.71217456e-01
-4.86678481e-01 -1.28851697e-01 8.05267274e-01 4.43729639e-01
-9.89591241e-01 6.03178740e-01 -1.27471149e+00 -4.45431650e-01
4.89020318e-01 1.81065604e-01 -5.32972276e-01 -2.10186765e-01
-8.85021508e-01 1.19557452e+00 3.17894638e-01 -2.84088939e-01
-1.17124224e+00 -9.56223488e-01 -1.34605920e+00 -3.57954413e-01
3.06394875e-01 -5.63576341e-01 1.34621167e+00 -5.58225870e-01
-1.60962498e+00 8.35469723e-01 -3.91410649e-01 -7.35143125e-01
9.09046054e-01 -4.24930215e-01 -3.47100645e-01 1.78063840e-01
3.15923542e-01 1.05529499e+00 6.47622645e-01 -1.28500676e+00
-9.71033573e-01 -1.69104010e-01 3.68185751e-02 1.09493598e-01
3.23101163e-01 -5.59017897e-01 -2.47672856e-01 -1.18256239e-02
-1.27059445e-01 -1.24557912e+00 -4.69750643e-01 3.67031813e-01
-5.08756526e-02 -1.87018082e-01 1.53570628e+00 -9.03191641e-02
3.99092257e-01 -2.08114338e+00 -9.49448645e-02 -2.12269008e-01
1.12366661e-01 1.89998671e-01 6.15026839e-02 -1.17753409e-01
1.41039610e-01 -4.79300499e-01 -3.89446884e-01 -5.94831407e-01
-1.51624352e-01 4.52429354e-01 -2.41501108e-01 6.97376907e-01
4.26993072e-01 1.02201486e+00 -1.32070124e+00 -5.92454016e-01
9.45965290e-01 3.76465648e-01 -5.27533054e-01 2.44470760e-01
-1.56420082e-01 7.67361164e-01 -3.27247232e-01 1.57178551e-01
4.44668651e-01 -6.15307875e-02 -2.69534647e-01 3.20072137e-02
-2.29399323e-01 1.98244214e-01 -1.16944897e+00 1.79366362e+00
-4.73536313e-01 1.22197902e+00 -2.24979252e-01 -1.20957649e+00
8.55625629e-01 -9.33992639e-02 8.37617755e-01 -7.47948766e-01
-1.24666631e-01 7.67756328e-02 -1.62405029e-01 -6.64869666e-01
6.72458887e-01 3.84726562e-02 -1.69132072e-02 2.08249599e-01
1.49107039e-01 -7.14485407e-01 4.54160362e-01 2.32933611e-01
1.26697278e+00 5.31729937e-01 -8.03846717e-02 -3.58243078e-01
7.54281878e-01 4.86987263e-01 4.38770145e-01 4.80038911e-01
-6.36083841e-01 4.45682943e-01 9.30253118e-02 -9.19625163e-01
-1.23306191e+00 -1.02457738e+00 -9.92729738e-02 8.10997128e-01
6.58241451e-01 -1.31743709e-02 -5.07028103e-01 -6.81669056e-01
1.51669115e-01 7.01015532e-01 -3.70980918e-01 -2.34874591e-01
-8.01922143e-01 -3.00667554e-01 3.27574313e-01 6.71776772e-01
5.87638736e-01 -1.09921038e+00 -1.08703101e+00 3.96213204e-01
-9.44910720e-02 -1.77710605e+00 -1.99678198e-01 2.18289137e-01
-7.51916051e-01 -1.22456074e+00 -2.36768231e-01 -5.01824319e-01
4.50473040e-01 4.19512659e-01 1.30410242e+00 -2.37554848e-01
-4.63044882e-01 4.84390169e-01 5.78430519e-02 -2.95973927e-01
-5.93927026e-01 1.62269190e-01 2.29038611e-01 -8.92806202e-02
2.54481852e-01 -3.60019326e-01 -4.85136807e-01 5.48010349e-01
-6.09615922e-01 -6.26374632e-02 2.49003217e-01 4.46870029e-01
5.30028403e-01 1.04019567e-01 3.49824429e-01 -7.89424241e-01
-2.31261730e-01 -3.23337317e-01 -9.74196374e-01 -1.90481588e-01
-5.25662899e-01 3.66495073e-01 6.78670585e-01 -3.38671118e-01
-1.14463007e+00 7.03343630e-01 6.17118776e-02 -5.36395848e-01
-4.23029482e-01 -2.06219524e-01 7.92679340e-02 -2.48949602e-01
7.38982797e-01 -6.99932724e-02 8.90585333e-02 2.07611948e-01
6.83367014e-01 1.13639645e-01 1.09502530e+00 -2.84030318e-01
8.79852295e-01 9.83085692e-01 4.18775827e-01 -7.90696144e-01
-7.35149503e-01 -8.95976424e-01 -1.05626619e+00 -5.47375560e-01
1.12701094e+00 -1.08583403e+00 -7.63129473e-01 5.14107227e-01
-1.04590619e+00 -8.57130051e-01 -6.57079339e-01 8.59983563e-01
-1.05346072e+00 2.07194209e-01 -3.91958326e-01 -7.27175236e-01
3.03680807e-01 -1.16552544e+00 1.14233971e+00 1.44499525e-01
-3.90333198e-02 -1.18137074e+00 4.04986620e-01 3.94092798e-01
3.11156183e-01 2.06803754e-01 2.15164363e-01 -3.03837270e-01
-9.13196623e-01 -1.91075638e-01 -1.64390028e-01 2.52636105e-01
9.62995961e-02 6.55666143e-02 -1.01960933e+00 -1.90812156e-01
-2.91555852e-01 -4.77638543e-01 1.10090470e+00 4.36426014e-01
8.77595305e-01 2.30760053e-01 -5.95530629e-01 6.78761125e-01
1.23036444e+00 1.17824646e-02 6.01406097e-01 2.50385940e-01
9.74781632e-01 8.72046888e-01 6.87915206e-01 1.29859626e-01
7.75778890e-01 5.87218761e-01 5.84814012e-01 -1.49786204e-01
-3.45661044e-01 -2.21789777e-01 4.18691546e-01 3.93358231e-01
2.42325827e-01 -1.33494079e-01 -1.00913000e+00 7.55992115e-01
-2.12541103e+00 -1.23213685e+00 -3.60109657e-01 2.11675787e+00
2.50298828e-01 6.53436065e-01 5.61009310e-02 1.39458910e-01
4.17381763e-01 3.32938969e-01 -9.58856285e-01 -1.42344505e-01
1.07077338e-01 -1.74888074e-01 1.00505459e+00 8.21914613e-01
-1.36466646e+00 1.14793968e+00 6.77865267e+00 2.46620432e-01
-1.18972921e+00 8.82172436e-02 6.31059110e-01 6.37441576e-02
3.84433270e-02 1.70805916e-01 -1.05835962e+00 3.50227088e-01
1.04589105e+00 -2.17666011e-02 2.29858294e-01 9.36411619e-01
4.80522722e-01 -4.18915063e-01 -1.31453621e+00 8.65712166e-01
7.74032921e-02 -1.46874523e+00 -6.22181237e-01 -1.10209072e-02
9.17983830e-01 5.31929731e-01 -3.33163798e-01 3.91286552e-01
8.37603033e-01 -7.55211771e-01 6.88714564e-01 6.34083867e-01
5.25368929e-01 -5.10332584e-01 5.87994397e-01 5.18189847e-01
-1.45767713e+00 -4.92245294e-02 -1.26146168e-01 -1.42508298e-01
6.28794014e-01 5.56830466e-01 -5.80839097e-01 3.39844912e-01
7.99319446e-01 1.24094450e+00 -5.44510961e-01 1.01873231e+00
-8.95987675e-02 4.75519121e-01 -4.50279593e-01 2.48343736e-01
3.98844272e-01 -1.03990138e-01 4.91901517e-01 1.18257034e+00
-7.82451183e-02 -1.78527698e-01 4.93684620e-01 7.97831059e-01
1.31394461e-01 -3.90351713e-01 -1.12273920e+00 5.88582635e-01
6.24817573e-02 1.21319938e+00 -8.81272316e-01 -5.08140504e-01
-3.90491724e-01 6.59031868e-01 3.29576492e-01 3.07341665e-01
-6.85699880e-01 -1.96879506e-01 7.53819704e-01 3.19744319e-01
2.53766686e-01 -6.92680120e-01 -1.40383333e-01 -1.10870564e+00
-5.78830466e-02 -2.11345199e-02 4.16393727e-02 -7.26603806e-01
-1.01032186e+00 4.00137395e-01 2.28564695e-01 -1.40997922e+00
-6.56893611e-01 -6.24141991e-01 -6.86198711e-01 3.69901270e-01
-1.94202209e+00 -9.93990958e-01 -7.36250103e-01 5.41001379e-01
7.30564415e-01 -2.91191459e-01 2.89863288e-01 1.44851536e-01
-1.89177468e-01 -1.48741780e-02 3.32877338e-02 2.90619992e-02
7.14910746e-01 -1.25731063e+00 6.75730407e-01 9.86768842e-01
7.48961046e-02 -2.52362221e-01 7.45241523e-01 -4.57639188e-01
-1.19691980e+00 -1.40597451e+00 6.37972295e-01 -8.79323840e-01
6.35154307e-01 -3.95170927e-01 -7.71127284e-01 6.53555334e-01
1.66351143e-02 9.39462841e-01 -4.91260067e-02 -3.86748374e-01
-1.32316828e-01 -3.94410312e-01 -1.02720594e+00 1.70941666e-01
1.09488034e+00 -5.15061438e-01 -4.07884061e-01 3.34236115e-01
6.95188999e-01 -4.61859226e-01 -2.51616299e-01 5.53911686e-01
4.42979157e-01 -9.84963119e-01 9.19608951e-01 -3.79583597e-01
2.91773617e-01 -5.49704194e-01 -2.75655203e-02 -1.16662669e+00
-6.67348653e-02 -5.97820878e-01 -1.58578455e-01 5.92528939e-01
2.00256214e-01 -4.80608761e-01 1.17055821e+00 5.79893053e-01
-3.85798991e-01 -1.10008344e-01 -9.22586381e-01 -9.69027221e-01
-4.22250852e-02 -6.96580350e-01 9.55866873e-02 8.32460999e-01
-3.72291267e-01 3.50981385e-01 -1.67889789e-01 -3.86707461e-03
9.44105566e-01 -1.61078721e-01 1.23149931e+00 -1.39127290e+00
-3.86969894e-02 -3.19297075e-01 -1.03342283e+00 -1.17067218e+00
8.49745214e-01 -6.95120096e-01 5.70402682e-01 -1.46217263e+00
-1.60328865e-01 -6.46306157e-01 1.16334669e-01 3.76335889e-01
1.00161389e-01 2.72154599e-01 8.42306837e-02 2.48328477e-01
-9.20219839e-01 6.62203848e-01 1.15074909e+00 -3.60792518e-01
-3.94225210e-01 1.07900888e-01 2.35345457e-02 7.57690787e-01
5.96975267e-01 -4.12077039e-01 -6.79442167e-01 -3.17398667e-01
-2.06242755e-01 -1.05493911e-01 6.22364461e-01 -1.54181457e+00
5.85146427e-01 -1.87509894e-01 2.25599930e-01 -7.12492704e-01
3.19711238e-01 -8.73921216e-01 -9.95244458e-02 6.91468000e-01
-2.91937768e-01 8.10516030e-02 4.62130368e-01 7.55605817e-01
-2.52424955e-01 3.85773778e-02 9.15858567e-01 -2.14531243e-01
-1.34453857e+00 3.81724983e-01 -5.99753678e-01 3.22967887e-01
1.22062147e+00 -3.38268816e-01 -3.02973092e-01 -4.87401336e-01
-5.58503032e-01 3.78715307e-01 3.57385427e-01 7.91644931e-01
4.42785561e-01 -1.13553929e+00 -6.80835187e-01 3.41140956e-01
3.22089106e-01 4.84098971e-01 2.76969969e-02 5.11460602e-01
-8.03604722e-01 5.41880012e-01 -3.09985518e-01 -1.15972996e+00
-7.90808916e-01 5.16508758e-01 5.40392697e-01 -2.07281560e-01
-8.96915078e-01 3.75016451e-01 3.19651932e-01 -7.61297166e-01
4.52096313e-02 -3.65630329e-01 -4.79477495e-02 -3.67757112e-01
2.19764963e-01 4.85054404e-01 1.13149047e-01 -9.38677013e-01
-3.74828458e-01 6.72840297e-01 1.06321096e-01 -1.73419505e-01
8.95924389e-01 -2.00219542e-01 3.93490195e-01 6.03721440e-01
1.37294054e+00 -3.40028048e-01 -1.98317945e+00 -2.63906747e-01
-1.37280658e-01 -4.41708356e-01 2.54987866e-01 -1.84797645e-01
-1.14112639e+00 9.13156271e-01 8.35880756e-01 -6.37632236e-02
5.73614001e-01 5.00019528e-02 7.35044301e-01 7.37918854e-01
5.76034486e-01 -9.94052947e-01 3.66958022e-01 8.02984536e-01
1.89716712e-01 -1.77650166e+00 6.78906078e-03 -3.18570495e-01
-6.21487916e-01 9.26540196e-01 9.39810395e-01 -3.68917912e-01
8.33184004e-01 3.99729162e-01 1.02428049e-01 -2.99507771e-02
-6.88820302e-01 -3.69971484e-01 1.50890097e-01 7.31707931e-01
-5.49611039e-02 -1.10268503e-01 4.00353372e-01 -3.69592786e-01
-2.76959956e-01 1.83995172e-01 5.21155417e-01 1.08209968e+00
-4.26918387e-01 -7.79751003e-01 5.37688844e-02 2.18529224e-01
6.62793815e-02 3.42688948e-01 1.15641125e-01 8.38798761e-01
1.48212671e-01 9.88031566e-01 4.64001715e-01 -1.41041234e-01
4.78951752e-01 -2.51816154e-01 3.59557062e-01 -3.91104579e-01
-2.27608178e-02 -4.98593867e-01 -1.51929215e-01 -9.86007333e-01
-7.52064586e-01 -8.23162198e-01 -1.50113952e+00 -3.47318262e-01
-4.13222946e-02 -2.61561751e-01 7.21080601e-01 1.16869974e+00
2.76805758e-01 5.63257277e-01 6.94544077e-01 -1.30636680e+00
-4.92740348e-02 -4.91169631e-01 -2.60458112e-01 6.63659871e-01
8.40173423e-01 -7.42475629e-01 -4.22856390e-01 5.57080805e-01]
|
[8.501389503479004, -1.87962007522583]
|
b00d16e9-181c-4e93-8df2-197a9e2da4ad
|
enhancing-egocentric-3d-pose-estimation-with
|
2201.02017
| null |
https://arxiv.org/abs/2201.02017v3
|
https://arxiv.org/pdf/2201.02017v3.pdf
|
Enhancing Egocentric 3D Pose Estimation with Third Person Views
|
In this paper, we propose a novel approach to enhance the 3D body pose estimation of a person computed from videos captured from a single wearable camera. The key idea is to leverage high-level features linking first- and third-views in a joint embedding space. To learn such embedding space we introduce First2Third-Pose, a new paired synchronized dataset of nearly 2,000 videos depicting human activities captured from both first- and third-view perspectives. We explicitly consider spatial- and motion-domain features, combined using a semi-Siamese architecture trained in a self-supervised fashion. Experimental results demonstrate that the joint multi-view embedded space learned with our dataset is useful to extract discriminatory features from arbitrary single-view egocentric videos, without needing domain adaptation nor knowledge of camera parameters. We achieve significant improvement of egocentric 3D body pose estimation performance on two unconstrained datasets, over three supervised state-of-the-art approaches. Our dataset and code will be available for research purposes.
|
['Francesc Moreno-Noguer', 'Albert Pumarola', 'Enric Corona', 'Mariella Dimiccoli', 'Ameya Dhamanaskar']
|
2022-01-06
| null | null | null | null |
['3d-pose-estimation']
|
['computer-vision']
|
[ 9.21441540e-02 -7.77827874e-02 -1.61514342e-01 -4.28509891e-01
-7.16626763e-01 -5.97962379e-01 5.21213233e-01 -3.43635261e-01
-3.95115376e-01 3.42378438e-01 7.59342790e-01 7.71804810e-01
5.88689893e-02 -2.65939295e-01 -7.08133519e-01 -4.30735171e-01
-3.43798697e-01 3.32442492e-01 -1.69334024e-01 1.07755633e-02
-2.52503455e-01 2.20055044e-01 -1.29155982e+00 1.89909507e-02
2.63402164e-01 7.18847513e-01 -2.90929049e-01 8.78104806e-01
8.79433215e-01 2.49512434e-01 -1.74342036e-01 -5.69950104e-01
5.54148853e-01 -4.30224061e-01 -5.04117310e-01 5.79186261e-01
1.08303130e+00 -5.28955042e-01 -7.87514627e-01 6.50494874e-01
7.30876327e-01 1.04334064e-01 3.75527650e-01 -1.15279222e+00
-1.98288620e-01 -1.25844225e-01 -5.28448284e-01 2.96633452e-01
1.17299008e+00 2.73574084e-01 9.04765308e-01 -8.06973755e-01
8.72811019e-01 1.08804369e+00 9.32910562e-01 4.41097707e-01
-1.27479386e+00 -3.60488802e-01 1.02907106e-01 9.21522751e-02
-1.42792356e+00 -4.39229339e-01 1.12973881e+00 -6.50817692e-01
6.33827507e-01 -3.04669328e-02 1.37292004e+00 1.49349666e+00
1.78005293e-01 9.18693364e-01 8.45935047e-01 -2.00660095e-01
-9.96784195e-02 -2.31763825e-01 -1.06208742e-01 8.59490633e-01
4.81794626e-01 -6.04748614e-02 -9.91777062e-01 -2.76348233e-01
9.37247932e-01 4.00756419e-01 -2.83916384e-01 -1.25487638e+00
-1.57911217e+00 7.76583791e-01 2.85711527e-01 4.89518866e-02
-4.01375562e-01 1.96451977e-01 4.27236676e-01 1.27141014e-01
5.55677891e-01 2.07504645e-01 -3.83859634e-01 -5.81211925e-01
-8.70106339e-01 6.70847893e-01 6.83817863e-01 9.58648503e-01
4.65576857e-01 -1.07439108e-01 1.62480190e-01 4.04726714e-01
4.01038527e-01 7.49772012e-01 4.24725205e-01 -9.87722635e-01
7.12668002e-01 5.29832125e-01 1.52178720e-01 -1.15725935e+00
-6.01657927e-01 -2.89976001e-01 -4.15133834e-01 -3.30711365e-01
5.78490555e-01 -2.79484123e-01 -4.76396143e-01 1.78080058e+00
7.28646457e-01 1.92696527e-01 -1.63609579e-01 1.30194843e+00
5.30160904e-01 7.39871338e-02 -2.71465033e-01 1.43194869e-01
1.60629690e+00 -1.01478386e+00 -5.07913053e-01 -3.86596382e-01
4.18969661e-01 -4.30716187e-01 6.94813848e-01 3.36236179e-01
-1.18783271e+00 -6.70687377e-01 -1.06869602e+00 -9.59666297e-02
-1.48637936e-01 2.75255114e-01 4.15761679e-01 7.31767416e-01
-6.21044397e-01 5.11598825e-01 -1.25777733e+00 -6.63098454e-01
1.49569526e-01 4.08414006e-01 -1.02340043e+00 -6.69370443e-02
-1.02465594e+00 6.14066243e-01 8.07996169e-02 1.03251562e-02
-8.17425311e-01 -5.36703825e-01 -1.34218609e+00 -3.74182224e-01
4.37704980e-01 -1.19377494e+00 9.39981222e-01 -5.91418386e-01
-1.48017275e+00 1.15102899e+00 -1.60830960e-01 -2.58931696e-01
5.98611057e-01 -9.39164102e-01 -3.03500891e-01 7.01666892e-01
9.44241881e-02 3.94298553e-01 1.12106276e+00 -8.33764613e-01
-1.64926335e-01 -1.10048270e+00 -1.95821058e-02 7.02837110e-01
-4.64316279e-01 -2.07206309e-01 -6.91900194e-01 -8.53496015e-01
2.23423675e-01 -1.40399683e+00 4.17202106e-03 2.31014848e-01
-2.00443864e-01 9.05734971e-02 7.58590162e-01 -8.13972414e-01
8.79173458e-01 -2.04386044e+00 8.13203275e-01 4.81200367e-02
3.16175640e-01 -6.12681210e-02 1.93160325e-01 3.47577751e-01
-3.26913930e-02 -4.78923619e-01 1.68841728e-03 -5.74547172e-01
2.58590002e-02 4.16093618e-02 2.31022030e-01 1.19298518e+00
1.36661539e-02 8.60026240e-01 -1.11310410e+00 -5.20133674e-01
4.01863366e-01 6.69739842e-01 -7.96263278e-01 5.30712783e-01
4.25788879e-01 7.84872174e-01 -4.92181361e-01 6.76837087e-01
3.58838111e-01 -1.80219263e-01 2.76230335e-01 -5.95983744e-01
4.18083936e-01 8.92900005e-02 -1.33911717e+00 2.71572638e+00
-1.22108251e-01 3.65318835e-01 -7.85034969e-02 -9.90485370e-01
5.05434096e-01 4.40470278e-01 8.86724591e-01 -9.51542854e-02
9.36999395e-02 -1.59403265e-01 -3.91811132e-01 -6.58365011e-01
3.07963580e-01 -7.35059530e-02 -4.32491034e-01 3.52523327e-01
5.36381364e-01 1.27077460e-01 7.83527046e-02 8.59658495e-02
9.68190908e-01 6.95901155e-01 5.18233180e-01 -3.08800731e-02
4.91679728e-01 -4.51889157e-01 5.62096715e-01 3.15904707e-01
-5.30471742e-01 8.33832026e-01 2.02525750e-01 -7.00669229e-01
-9.78809536e-01 -1.47934890e+00 1.29564747e-01 8.62854302e-01
1.39553159e-01 -8.59090567e-01 -6.33578420e-01 -7.56346226e-01
4.11222190e-01 -2.27643222e-01 -7.98508883e-01 -7.35582113e-02
-8.49853218e-01 -2.82216251e-01 4.01687473e-01 8.67276371e-01
4.61988926e-01 -3.38490725e-01 -9.76783454e-01 -7.52189606e-02
-3.62051308e-01 -1.41442072e+00 -9.66360509e-01 -2.89956123e-01
-9.97611523e-01 -1.13661671e+00 -1.16752064e+00 -4.33238983e-01
5.91505229e-01 3.93622130e-01 9.80278969e-01 -5.18069208e-01
-4.27531898e-01 1.23765409e+00 -2.59502739e-01 1.40755445e-01
5.71693659e-01 4.59490605e-02 7.23174155e-01 2.97318131e-01
4.67949033e-01 -8.51980090e-01 -9.18372273e-01 3.51554781e-01
-2.31184155e-01 -1.62370369e-01 3.39490920e-01 8.02923739e-01
3.95562887e-01 -7.11508870e-01 2.76822299e-02 -3.52659345e-01
1.67467132e-01 -2.76790053e-01 -5.05920500e-02 -1.18056506e-01
-1.24569967e-01 -2.11107060e-01 2.67666906e-01 -3.65449309e-01
-6.71634734e-01 5.42978466e-01 2.95820862e-01 -7.30782151e-01
-2.16249511e-01 2.45353356e-01 -4.33358163e-01 -8.64025205e-02
5.09952307e-01 2.17956394e-01 2.32446939e-01 -4.19541299e-01
5.45913100e-01 3.13242346e-01 9.21871781e-01 -6.45424068e-01
9.78015959e-01 8.25889528e-01 2.37116069e-02 -8.53975534e-01
-7.14977562e-01 -1.04869711e+00 -1.30794120e+00 -3.36124659e-01
1.08758426e+00 -1.58191955e+00 -7.04526544e-01 4.11184043e-01
-7.48698115e-01 8.88275430e-02 -1.88658118e-01 9.22786951e-01
-1.01670325e+00 7.27726221e-01 -3.69296491e-01 -6.54462397e-01
-2.10667625e-01 -7.55556226e-01 1.68008780e+00 7.71546131e-03
-6.00033164e-01 -1.02444935e+00 5.22749782e-01 7.09603846e-01
-3.52076799e-01 7.01932132e-01 -2.10023284e-01 -4.05885577e-01
-1.76702276e-01 -5.44778764e-01 3.06371212e-01 1.48555249e-01
2.74546087e-01 -5.78778207e-01 -8.41490209e-01 -8.20851803e-01
-6.45682141e-02 -4.74721581e-01 4.98741746e-01 4.08914566e-01
7.60968149e-01 -2.07168624e-01 -3.03407431e-01 9.82294440e-01
1.01344490e+00 -5.40836513e-01 3.49041879e-01 2.40155578e-01
1.03588498e+00 5.65478027e-01 5.80666721e-01 6.46144390e-01
8.57673824e-01 1.11878502e+00 1.46055236e-01 9.41501632e-02
1.10732704e-01 -6.02811337e-01 6.11100733e-01 6.98633015e-01
-5.12412667e-01 3.35203856e-01 -6.67244852e-01 6.44165635e-01
-1.74422812e+00 -1.21085238e+00 3.56239915e-01 2.28104377e+00
4.95612681e-01 -1.21086121e-01 8.27756941e-01 1.59454286e-01
4.57831562e-01 5.38447559e-01 -6.13609850e-01 2.59139895e-01
1.45635471e-01 -1.43477276e-01 3.79935145e-01 1.95967257e-01
-1.56802487e+00 5.87352157e-01 5.86905193e+00 -1.19648114e-01
-8.53247643e-01 1.72639340e-01 -1.03174865e-01 -6.41096175e-01
1.92647442e-01 -2.38895684e-01 -6.30929768e-01 4.24709320e-01
7.42195785e-01 8.62101167e-02 3.77953686e-02 8.91499400e-01
4.65589538e-02 -2.77621280e-02 -1.36330593e+00 1.55384552e+00
8.26745629e-01 -8.83892596e-01 -3.31310421e-01 3.58001560e-01
6.78379893e-01 -2.05309287e-01 9.42555070e-02 6.13229312e-02
-2.51872182e-01 -4.98147786e-01 6.80702031e-01 6.07968450e-01
6.42342091e-01 -5.86600244e-01 3.99552882e-01 1.64788768e-01
-1.46482480e+00 2.69729141e-02 -6.56264052e-02 -1.46128654e-01
4.44109112e-01 2.42469832e-01 -4.24527556e-01 6.92570686e-01
1.01748312e+00 1.34217644e+00 -5.28992057e-01 6.65477395e-01
-5.40127084e-02 2.63681680e-01 -3.66073281e-01 4.18540984e-01
-2.21021280e-01 -1.07765645e-01 7.84761310e-01 1.06231284e+00
2.14990571e-01 1.84779078e-01 3.66232187e-01 3.28503489e-01
9.63982195e-02 -1.22500725e-01 -9.01465595e-01 2.77922265e-02
-1.08148552e-01 1.28726494e+00 -2.70303160e-01 -3.45487326e-01
-5.68020821e-01 1.50471389e+00 1.42433554e-01 1.58545539e-01
-9.33205009e-01 -1.06909603e-01 8.93333733e-01 3.82673144e-01
5.35535336e-01 -6.18815362e-01 1.06455356e-01 -1.94971395e+00
4.17313695e-01 -8.36188138e-01 5.95454395e-01 -6.60998523e-01
-1.22796655e+00 1.49780497e-01 3.34731370e-01 -1.57282794e+00
-6.77343130e-01 -6.16182983e-01 -3.37133884e-01 4.54367459e-01
-9.61237967e-01 -1.63544595e+00 -5.91903448e-01 9.67754722e-01
4.51471418e-01 -2.49185488e-01 7.82469749e-01 1.79944843e-01
-3.72865409e-01 6.77406967e-01 -1.46778971e-01 3.26297671e-01
1.08316410e+00 -1.34542394e+00 3.57538313e-01 7.24716365e-01
4.08713609e-01 8.40196073e-01 6.47620499e-01 -5.56316435e-01
-2.01146364e+00 -7.67182171e-01 7.76817501e-01 -1.03465843e+00
3.27425689e-01 -6.63937986e-01 -1.82197869e-01 1.02530730e+00
-1.86919793e-02 5.07999778e-01 1.19947362e+00 3.64349812e-01
-3.27856630e-01 -2.27954209e-01 -1.14336073e+00 4.33376908e-01
1.51895320e+00 -6.84737682e-01 -1.02313232e+00 1.90782174e-01
1.78297833e-01 -6.95111156e-01 -1.30491364e+00 2.81528622e-01
1.19199061e+00 -8.42192233e-01 1.35599053e+00 -6.58039749e-01
1.89263403e-01 -2.93479443e-01 -3.51533920e-01 -1.17525434e+00
-3.51291806e-01 -7.56491065e-01 -6.71737373e-01 5.43448269e-01
-3.15051198e-01 -4.03341174e-01 1.09152389e+00 4.02111501e-01
9.09225419e-02 -5.91031373e-01 -9.27663922e-01 -8.26122642e-01
-2.58747876e-01 -8.79811645e-02 2.00614393e-01 7.29389191e-01
2.83514142e-01 2.36996368e-01 -8.87114942e-01 2.95665741e-01
1.05472326e+00 2.44738415e-01 1.47898650e+00 -1.17906785e+00
-4.26764250e-01 3.75751197e-01 -1.21800864e+00 -1.28087282e+00
7.68622244e-03 -6.12685978e-01 -4.64579463e-01 -1.01077020e+00
3.85794967e-01 4.51307893e-01 -1.34893209e-01 1.90458745e-01
-1.46220550e-01 6.85129881e-01 3.24298024e-01 1.49526909e-01
-8.39825809e-01 7.65073121e-01 9.80788708e-01 -5.91253750e-02
-7.56063871e-03 -2.27340207e-01 -4.93750632e-01 8.92116368e-01
3.18794787e-01 -2.31227830e-01 -4.54493940e-01 -3.41743767e-01
-9.78605822e-02 -9.46870819e-03 8.12146485e-01 -1.31502879e+00
1.36840463e-01 1.03331670e-01 8.93895328e-01 -5.63557863e-01
8.81942511e-01 -7.65636742e-01 1.96273759e-01 4.86437708e-01
-5.12520932e-02 3.11217099e-01 -6.53108060e-02 9.46495950e-01
-1.74177051e-01 3.87902468e-01 4.56427783e-01 -2.41166696e-01
-4.95659024e-01 4.76792276e-01 1.06371261e-01 3.26160818e-01
1.20585024e+00 -5.18183231e-01 3.06680024e-01 -6.55102372e-01
-1.16845846e+00 5.78846373e-02 7.83764839e-01 5.08005559e-01
6.81264997e-01 -1.64344132e+00 -5.79331279e-01 5.31065702e-01
4.43119913e-01 -3.10136259e-01 3.59317094e-01 9.71260548e-01
-3.06749552e-01 5.20795405e-01 -4.76042181e-01 -1.05452418e+00
-1.41049969e+00 3.86885911e-01 2.13684902e-01 -2.98508257e-01
-8.63777816e-01 4.93884116e-01 6.95335940e-02 -5.10268569e-01
-1.05791688e-01 -2.08333448e-01 1.14867382e-01 -1.46494228e-02
4.74716663e-01 5.61031282e-01 -2.27460787e-01 -1.10740542e+00
-5.98824382e-01 1.08038616e+00 4.27907854e-02 -2.53699839e-01
1.28402877e+00 -2.91298151e-01 4.99404013e-01 4.88496870e-01
1.49419951e+00 2.21290305e-01 -1.55786192e+00 -3.53513598e-01
-5.43071151e-01 -9.28654075e-01 -2.95389444e-01 -2.59435922e-01
-9.48070467e-01 7.99505055e-01 7.48693705e-01 -3.65481138e-01
8.90187681e-01 -2.29389761e-02 8.63083482e-01 3.62389177e-01
5.68362355e-01 -1.14135146e+00 5.35533667e-01 8.76559615e-02
8.17636430e-01 -1.35211337e+00 4.60202903e-01 -2.34463677e-01
-7.24679172e-01 9.85751569e-01 5.82606733e-01 -5.12582421e-01
7.00441837e-01 -2.90876567e-01 -1.92903519e-01 -3.34356993e-01
-3.22082222e-01 -1.44672334e-01 6.78573310e-01 7.87782729e-01
4.82008457e-01 1.08054012e-01 -1.66072860e-01 6.27517700e-01
-2.18532681e-01 -1.01777939e-02 1.45405307e-01 1.19904292e+00
7.21347928e-02 -9.73599970e-01 -3.79851133e-01 9.63033140e-02
-3.94103587e-01 5.60870647e-01 -3.79949272e-01 8.82036388e-01
9.11188126e-02 5.58182478e-01 -2.50630220e-03 -4.71392304e-01
5.02082646e-01 1.39292926e-01 1.06681001e+00 -5.83484650e-01
-1.84952646e-01 1.72053874e-01 6.16134554e-02 -1.04088163e+00
-7.89661765e-01 -1.16353858e+00 -6.28856122e-01 -7.64091462e-02
1.36419728e-01 -1.49982750e-01 3.33421290e-01 6.87101901e-01
5.20576537e-01 3.13873701e-02 5.09346545e-01 -1.46519756e+00
-6.11058354e-01 -7.93722689e-01 -7.15201735e-01 8.16165626e-01
4.42256540e-01 -1.16354346e+00 -2.07856640e-01 4.55228418e-01]
|
[7.069746494293213, -0.79920893907547]
|
7a497959-5831-4d02-b079-bf2b23cca4b6
|
robust-representation-learning-with-reliable
|
2305.16335
| null |
https://arxiv.org/abs/2305.16335v1
|
https://arxiv.org/pdf/2305.16335v1.pdf
|
Robust Representation Learning with Reliable Pseudo-labels Generation via Self-Adaptive Optimal Transport for Short Text Clustering
|
Short text clustering is challenging since it takes imbalanced and noisy data as inputs. Existing approaches cannot solve this problem well, since (1) they are prone to obtain degenerate solutions especially on heavy imbalanced datasets, and (2) they are vulnerable to noises. To tackle the above issues, we propose a Robust Short Text Clustering (RSTC) model to improve robustness against imbalanced and noisy data. RSTC includes two modules, i.e., pseudo-label generation module and robust representation learning module. The former generates pseudo-labels to provide supervision for the later, which contributes to more robust representations and correctly separated clusters. To provide robustness against the imbalance in data, we propose self-adaptive optimal transport in the pseudo-label generation module. To improve robustness against the noise in data, we further introduce both class-wise and instance-wise contrastive learning in the robust representation learning module. Our empirical studies on eight short text clustering datasets demonstrate that RSTC significantly outperforms the state-of-the-art models. The code is available at: https://github.com/hmllmh/RSTC.
|
['Xinting Liao', 'Chaochao Chen', 'Weiming Liu', 'Mengling Hu', 'Xiaolin Zheng']
|
2023-05-23
| null | null | null | null |
['pseudo-label', 'text-clustering', 'short-text-clustering']
|
['miscellaneous', 'natural-language-processing', 'natural-language-processing']
|
[ 2.01141126e-02 -3.05706203e-01 -1.82733849e-01 -5.63811481e-01
-1.24509943e+00 -3.92429829e-01 2.59879440e-01 3.80509973e-01
-2.14029863e-01 3.89187723e-01 3.90209436e-01 -3.11420858e-02
-7.21825585e-02 -5.70370257e-01 -4.62104529e-01 -8.45995545e-01
2.99891800e-01 6.04140460e-01 9.48727801e-02 -5.74940853e-02
4.27913308e-01 -1.36066573e-02 -1.70684707e+00 6.14047527e-01
1.29409111e+00 7.85026371e-01 -8.76733884e-02 3.60218406e-01
-4.14570391e-01 9.57986712e-01 -7.85240650e-01 -2.37731814e-01
1.62284911e-01 -4.35889006e-01 -8.58904123e-01 2.83271253e-01
-2.13502929e-01 1.38746440e-01 -1.62818775e-01 1.14535677e+00
7.88184762e-01 6.08036481e-02 7.93200314e-01 -1.62625253e+00
-5.36838949e-01 8.92589808e-01 -1.06938124e+00 -6.57914281e-02
1.18988857e-01 2.48065665e-02 9.47862864e-01 -7.82722890e-01
4.13810551e-01 1.40729558e+00 5.58167160e-01 6.26912236e-01
-1.02533662e+00 -1.03123021e+00 3.73801708e-01 1.89089105e-01
-1.51224470e+00 -5.31858385e-01 9.05807137e-01 -4.09424186e-01
3.77962530e-01 2.05834001e-01 4.10246067e-02 1.08658993e+00
-2.72839934e-01 1.06317937e+00 9.62873518e-01 -3.34605008e-01
2.84574896e-01 -6.29037991e-02 5.29261053e-01 2.11405843e-01
3.60846013e-01 -3.80651921e-01 -2.37817019e-01 -1.30179763e-01
4.03958708e-02 2.58939862e-01 -2.02489704e-01 -6.46725148e-02
-9.74061668e-01 7.90352941e-01 5.28517008e-01 3.44353139e-01
-6.84664473e-02 -8.55127051e-02 7.21786618e-01 2.48527035e-01
6.85328543e-01 2.88029999e-01 -4.83794451e-01 -6.21937448e-03
-8.68367195e-01 1.57975540e-01 5.27138114e-01 7.82686055e-01
5.90746522e-01 -1.98805034e-01 -1.51870787e-01 1.34595656e+00
4.39938486e-01 2.44076893e-01 1.01148129e+00 -7.02006578e-01
1.09467590e+00 1.01736617e+00 -4.44254726e-02 -1.24204636e+00
-6.60085797e-01 -5.82905233e-01 -1.18954360e+00 -2.21742243e-01
3.03772599e-01 -5.26302904e-02 -9.58037913e-01 1.57500899e+00
4.89062846e-01 1.43791735e-01 8.14765692e-02 7.79389739e-01
1.04637301e+00 8.34187210e-01 6.47320133e-03 -3.35722297e-01
1.22928870e+00 -1.05530679e+00 -9.54333544e-01 -2.34699279e-01
1.00604141e+00 -8.36970747e-01 1.09570110e+00 2.46749923e-01
-7.12118387e-01 -3.93440217e-01 -1.01338351e+00 -1.42781302e-01
-4.94500190e-01 2.38930300e-01 -5.07923216e-03 5.58529556e-01
-5.00253737e-01 3.18940610e-01 -7.68290401e-01 2.79008299e-02
6.12342775e-01 2.11642250e-01 -9.78114754e-02 -3.33156675e-01
-1.18267715e+00 4.13629770e-01 4.22527075e-01 1.06770739e-01
-4.00156975e-01 -4.29853022e-01 -8.09780478e-01 -1.07601732e-01
4.92812276e-01 -1.99292317e-01 9.66191053e-01 -9.66324449e-01
-1.13528979e+00 8.69879961e-01 -1.22798227e-01 -8.55394304e-02
5.95490098e-01 -1.32806078e-01 -3.39249372e-01 -8.52020234e-02
3.48574370e-01 4.76007223e-01 8.27915072e-01 -1.66730332e+00
-3.88475567e-01 -6.67616844e-01 -4.56818819e-01 3.77981305e-01
-3.09325784e-01 -4.52534482e-02 -7.64885008e-01 -1.13299561e+00
5.18366933e-01 -8.63752961e-01 -2.23703340e-01 -4.42046136e-01
-7.80047536e-01 -3.38893324e-01 9.21023548e-01 -4.99389082e-01
1.44035912e+00 -2.18847966e+00 1.20793097e-02 1.69148624e-01
1.01776518e-01 4.17310894e-01 -2.93616116e-01 2.77738750e-01
-4.15589720e-01 3.93709153e-01 -3.17190737e-01 -6.84542477e-01
4.75923531e-02 1.50244847e-01 -1.00774236e-01 4.66273040e-01
1.34385601e-01 6.63507342e-01 -8.20905030e-01 -5.60193956e-01
-5.10721244e-02 4.06282783e-01 -3.36341351e-01 2.41266266e-01
-1.20172299e-01 3.71705294e-01 -4.51631486e-01 7.31573582e-01
1.06975269e+00 -3.87152493e-01 2.07599208e-01 -6.64192662e-02
2.31927231e-01 3.22597593e-01 -1.62351513e+00 1.34863043e+00
-9.51251239e-02 3.03672962e-02 2.03305818e-02 -1.30389643e+00
9.62007761e-01 2.56234407e-01 5.06411672e-01 -6.78687990e-01
3.16167414e-01 3.02881330e-01 -2.99971074e-01 -5.28525472e-01
4.41810131e-01 -3.50789391e-02 -2.25975379e-01 7.77246177e-01
-2.71737427e-01 5.24973907e-02 3.00003976e-01 4.72745270e-01
9.66976941e-01 -1.28551021e-01 3.64006534e-02 -6.60905987e-02
5.45984745e-01 -2.99579352e-01 1.15339971e+00 3.86106133e-01
-3.01763773e-01 1.02116275e+00 7.09284246e-01 -3.04814249e-01
-6.04249835e-01 -4.04830158e-01 -6.93813562e-02 1.16662347e+00
1.97345451e-01 -6.75838172e-01 -8.96916270e-01 -9.73723829e-01
3.96486036e-02 3.05039793e-01 -6.69687212e-01 -3.81608695e-01
-3.54478508e-01 -1.25929594e+00 4.53273416e-01 4.91674900e-01
4.88413423e-01 -9.70769405e-01 9.78422463e-02 1.29024774e-01
-7.46820271e-01 -8.45144093e-01 -6.20400131e-01 2.81749010e-01
-8.77461970e-01 -1.28562999e+00 -4.29803342e-01 -6.27037108e-01
9.18840826e-01 6.23283148e-01 9.64220464e-01 5.69972277e-01
-2.32407644e-01 -1.96631551e-01 -8.14755440e-01 -2.46406823e-01
-3.85405928e-01 3.51487994e-01 7.95310587e-02 1.89242288e-01
5.00331402e-01 -2.91417807e-01 -5.02511382e-01 7.15213537e-01
-1.17712164e+00 -1.21462889e-01 3.27859223e-01 8.78047884e-01
8.20668280e-01 5.47755599e-01 8.50381732e-01 -1.32442904e+00
6.00052834e-01 -7.77800381e-01 -3.77975285e-01 1.83424205e-01
-6.65862381e-01 -4.16067354e-02 8.01956594e-01 -3.08112979e-01
-8.28784108e-01 1.41652524e-01 -2.45338306e-01 -2.05119550e-01
-1.27280220e-01 4.43910688e-01 -5.27187705e-01 4.26468879e-01
7.21644819e-01 4.84379055e-03 -1.02317326e-01 -7.39912868e-01
1.78035006e-01 1.25213861e+00 2.16263354e-01 -6.03178024e-01
7.61587620e-01 3.16078097e-01 -4.72226351e-01 -4.17817146e-01
-1.26420736e+00 -7.56884217e-01 -7.31825173e-01 4.80085462e-02
4.32849646e-01 -1.21533871e+00 -2.14034006e-01 8.28198731e-01
-7.80840695e-01 -3.21881920e-01 -2.81653404e-02 1.17637098e-01
-3.48848104e-01 4.73735213e-01 -6.85208917e-01 -6.61719024e-01
-4.67469305e-01 -1.33139575e+00 1.17419910e+00 2.77537137e-01
-4.25955318e-02 -6.55256629e-01 -1.69153124e-01 8.97273481e-01
1.53110906e-01 3.11708272e-01 8.25442910e-01 -7.88041472e-01
-1.90293521e-01 -3.45321000e-01 -3.47748131e-01 4.15834367e-01
2.16587916e-01 1.57487661e-01 -1.01212633e+00 -4.80033875e-01
-3.38255018e-02 -6.01827323e-01 1.13430810e+00 1.30605236e-01
1.56051850e+00 -3.50381017e-01 -2.89049089e-01 6.23919964e-01
1.25582218e+00 -8.35760087e-02 6.80729091e-01 3.35325092e-01
9.90952730e-01 7.28923678e-01 8.24170411e-01 5.13307452e-01
7.60230958e-01 5.53565860e-01 3.11161608e-01 -7.92389328e-05
-1.89116001e-01 -1.15190141e-01 1.02504306e-01 1.27781940e+00
3.22836697e-01 -3.50430816e-01 -1.06132245e+00 4.13322330e-01
-2.17483735e+00 -7.69167542e-01 -4.17019248e-01 1.98264587e+00
1.08832574e+00 1.31932512e-01 2.36182079e-01 8.15046310e-01
1.00691414e+00 1.13330431e-01 -7.02473402e-01 1.07265078e-01
-1.21053167e-01 -3.86433870e-01 1.14624433e-01 2.63326615e-01
-1.32206130e+00 7.76675940e-01 5.27482700e+00 1.07972801e+00
-7.21231937e-01 2.78761953e-01 1.03453457e+00 -1.31050840e-01
-3.37187469e-01 -2.91366786e-01 -7.67588496e-01 9.10723090e-01
6.64766073e-01 1.87166080e-01 1.51575029e-01 6.39626801e-01
1.26714990e-01 -4.77293581e-02 -6.09341204e-01 1.12537050e+00
2.08676323e-01 -9.64161396e-01 4.28405888e-02 -1.48725688e-01
1.03280592e+00 -9.28381160e-02 -9.50152278e-02 4.63893414e-01
3.87618601e-01 -9.07253385e-01 5.71480632e-01 2.09672213e-01
5.09634018e-01 -1.11806273e+00 9.93192613e-01 4.85551298e-01
-1.14726615e+00 -2.98575163e-01 -5.14162719e-01 2.11230874e-01
-3.10353816e-01 1.14722502e+00 -2.21220165e-01 6.00529671e-01
7.59048045e-01 9.98138607e-01 -9.36443985e-01 9.74896133e-01
-3.29715222e-01 6.51883304e-01 -2.06519216e-01 2.64912605e-01
1.43868495e-02 -1.27089828e-01 1.23666853e-01 1.22461629e+00
-1.79182831e-02 5.02049038e-03 4.62699264e-01 4.61748958e-01
-5.25269926e-01 3.83994520e-01 -2.35421419e-01 2.49424949e-01
7.64983416e-01 1.25378311e+00 -9.68592346e-01 -2.27798730e-01
2.35884287e-03 8.11201036e-01 5.16243160e-01 4.04801130e-01
-6.70554459e-01 -6.20448411e-01 4.32868034e-01 -1.05834845e-02
2.42048800e-02 1.30858257e-01 -6.69528663e-01 -1.32136738e+00
1.90335169e-01 -1.28060675e+00 8.88659656e-01 -4.01868463e-01
-1.50907338e+00 4.42479312e-01 -4.72133398e-01 -1.36934340e+00
1.56057566e-01 -1.78503916e-01 -4.75901991e-01 4.82624739e-01
-1.55687976e+00 -9.82479572e-01 -4.93589729e-01 6.48602426e-01
7.24522173e-01 -4.74039949e-02 5.72821021e-01 5.12063682e-01
-1.15809727e+00 8.88170004e-01 4.95088786e-01 3.13721508e-01
1.09551644e+00 -1.18046904e+00 2.36529663e-01 8.32779586e-01
-2.65147865e-01 4.38408017e-01 3.22383523e-01 -6.85683846e-01
-1.05680382e+00 -1.49807894e+00 6.75056517e-01 -4.83062923e-01
3.20917815e-01 -5.47540784e-01 -1.30749774e+00 4.06639516e-01
-2.83676624e-01 1.63462549e-01 9.47269201e-01 1.55102015e-01
-7.17258275e-01 -3.31269681e-01 -1.15292382e+00 2.90929079e-01
7.71578670e-01 -3.93552393e-01 -3.30662042e-01 5.67822278e-01
6.65845990e-01 -3.33761930e-01 -6.46244168e-01 3.49244356e-01
9.07680094e-02 -8.28839481e-01 7.51252651e-01 -3.16372573e-01
4.43805993e-01 -5.56800067e-01 -1.72991320e-01 -1.36202264e+00
-3.09384346e-01 -3.74617308e-01 -1.38863146e-01 1.75213838e+00
4.22239751e-01 -4.63179886e-01 6.04005754e-01 2.59070486e-01
4.58623320e-02 -8.30725610e-01 -7.73092508e-01 -6.16825819e-01
2.62741208e-01 -4.78119254e-01 8.89401495e-01 1.22789609e+00
5.61947003e-02 4.22666013e-01 -4.98457521e-01 6.20960332e-02
6.58077955e-01 1.95145950e-01 6.43194139e-01 -1.36487567e+00
2.00223505e-01 -3.94935310e-01 -1.29763842e-01 -6.56972885e-01
2.47965440e-01 -1.00131011e+00 3.11797231e-01 -1.55059969e+00
3.46483707e-01 -8.95299077e-01 -3.07568163e-01 6.41037226e-01
-7.74547696e-01 4.05300617e-01 1.12532772e-01 5.99665165e-01
-1.08719325e+00 6.49705946e-01 1.01386094e+00 -1.93668917e-01
-2.10599750e-01 2.26397946e-01 -1.25540793e+00 5.58695793e-01
1.15633607e+00 -9.36219871e-01 -4.34538990e-01 -3.73561233e-01
2.60905176e-01 -4.44482446e-01 -2.72109777e-01 -8.88007581e-01
2.62317777e-01 -1.16370264e-02 3.86532396e-01 -7.19669402e-01
-2.39954609e-02 -5.17558575e-01 -2.08321482e-01 3.54136467e-01
-4.35590714e-01 -3.32121029e-02 -1.13524668e-01 5.18132210e-01
-2.55083174e-01 -3.52048993e-01 1.09029436e+00 -1.39660522e-01
-1.09275974e-01 2.77360022e-01 -2.00354204e-01 4.68032241e-01
8.34835470e-01 1.91651747e-01 -6.23308778e-01 -3.67027491e-01
-3.36834371e-01 8.58473778e-01 6.01804793e-01 5.88945985e-01
4.69515920e-01 -1.43751204e+00 -7.87799656e-01 1.92189723e-01
2.76705980e-01 4.07087743e-01 3.65644634e-01 7.31424272e-01
-2.53041685e-01 -7.11544678e-02 2.85826743e-01 -4.90301341e-01
-1.21168244e+00 7.37278819e-01 1.86618775e-01 -4.73156303e-01
-4.29024041e-01 6.82350159e-01 -1.84959978e-01 -1.02324355e+00
6.70195222e-01 -2.89716218e-02 -3.08882177e-01 4.30850625e-01
7.83396065e-01 4.67296869e-01 3.19215059e-01 -7.96629548e-01
-4.60150540e-01 7.29711056e-01 -3.62921804e-01 3.46236080e-01
1.28160298e+00 -3.47792387e-01 -2.14312211e-01 4.76266950e-01
1.28985620e+00 -2.42582038e-01 -1.01855576e+00 -2.90667653e-01
1.44038439e-01 -4.66642201e-01 -2.92245988e-02 -8.24868917e-01
-1.29711676e+00 8.84384811e-01 4.85685587e-01 2.26387128e-01
1.23444068e+00 -1.73768654e-01 7.68192768e-01 2.04310372e-01
-7.83015192e-02 -1.51813447e+00 4.23051417e-01 5.40647268e-01
6.76015377e-01 -1.41029799e+00 1.14570566e-01 -4.57185566e-01
-8.55502844e-01 7.98615932e-01 6.64155364e-01 1.70883939e-01
6.61616325e-01 2.44638249e-01 4.98219818e-01 -1.62087992e-01
-7.36681342e-01 -1.73692390e-01 1.24581128e-01 6.57550454e-01
5.17529905e-01 4.94558327e-02 -2.90104538e-01 8.32918823e-01
4.28928658e-02 -4.00615931e-01 4.77022022e-01 8.83940935e-01
-3.66825044e-01 -1.25087094e+00 -6.89889014e-01 5.18950284e-01
-5.78758180e-01 3.02797943e-01 -4.46008116e-01 2.84610063e-01
2.93760508e-01 1.42093265e+00 -1.49642155e-01 -5.61581492e-01
3.45184892e-01 1.10407442e-01 -1.93731472e-01 -6.28097415e-01
-5.56891441e-01 2.04866201e-01 -1.78320318e-01 -4.81519490e-01
-5.19350946e-01 -7.13102162e-01 -1.42072010e+00 -2.46777773e-01
-5.83396554e-01 3.49137485e-01 5.46267748e-01 8.13151896e-01
5.47616422e-01 6.38587832e-01 1.10574973e+00 -7.31046975e-01
-5.83784699e-01 -1.09439909e+00 -6.93896770e-01 9.77200329e-01
2.19482437e-01 -5.96090972e-01 -5.57626545e-01 -7.49887899e-02]
|
[9.399024963378906, 3.9870641231536865]
|
f0f53bfc-a52d-42db-8775-24ebd246e538
|
portrait-a-hybrid-approach-to-create
|
2305.11536
| null |
https://arxiv.org/abs/2305.11536v1
|
https://arxiv.org/pdf/2305.11536v1.pdf
|
PORTRAIT: a hybrid aPproach tO cReate extractive ground-TRuth summAry for dIsaster evenT
|
Disaster summarization approaches provide an overview of the important information posted during disaster events on social media platforms, such as, Twitter. However, the type of information posted significantly varies across disasters depending on several factors like the location, type, severity, etc. Verification of the effectiveness of disaster summarization approaches still suffer due to the lack of availability of good spectrum of datasets along with the ground-truth summary. Existing approaches for ground-truth summary generation (ground-truth for extractive summarization) relies on the wisdom and intuition of the annotators. Annotators are provided with a complete set of input tweets from which a subset of tweets is selected by the annotators for the summary. This process requires immense human effort and significant time. Additionally, this intuition-based selection of the tweets might lead to a high variance in summaries generated across annotators. Therefore, to handle these challenges, we propose a hybrid (semi-automated) approach (PORTRAIT) where we partly automate the ground-truth summary generation procedure. This approach reduces the effort and time of the annotators while ensuring the quality of the created ground-truth summary. We validate the effectiveness of PORTRAIT on 5 disaster events through quantitative and qualitative comparisons of ground-truth summaries generated by existing intuitive approaches, a semi-automated approach, and PORTRAIT. We prepare and release the ground-truth summaries for 5 disaster events which consist of both natural and man-made disaster events belonging to 4 different countries. Finally, we provide a study about the performance of various state-of-the-art summarization approaches on the ground-truth summaries generated by PORTRAIT using ROUGE-N F1-scores.
|
['Sourav Kumar Dandapat', 'Roshni Chakraborty', 'Piyush Kumar Garg']
|
2023-05-19
| null | null | null | null |
['extractive-summarization']
|
['natural-language-processing']
|
[ 4.63459603e-02 2.82484561e-01 -3.96717116e-02 -1.48301795e-01
-1.25076902e+00 -8.20853949e-01 7.30103135e-01 1.10485315e+00
-2.76088864e-01 9.96246636e-01 1.06052160e+00 7.14303702e-02
5.06664962e-02 -9.26210403e-01 -2.42597654e-01 -4.64957356e-01
-1.22344017e-01 5.57820141e-01 -1.17781051e-01 -4.87848639e-01
4.49091315e-01 1.34984478e-01 -1.21319580e+00 3.75021785e-01
1.19706059e+00 4.26184058e-01 6.64930344e-02 6.92680717e-01
-2.61136174e-01 5.08581460e-01 -1.16956258e+00 -5.58339119e-01
-3.06837587e-03 -5.10497630e-01 -8.23943257e-01 1.72537372e-01
3.15677017e-01 -3.68572652e-01 -1.10698029e-01 8.88049722e-01
7.14428067e-01 1.53250054e-01 8.79652739e-01 -1.23523653e+00
-4.40469682e-01 1.07781529e+00 -5.10417163e-01 1.97627053e-01
8.08154166e-01 9.11364034e-02 9.50728714e-01 -8.67849410e-01
7.38418996e-01 1.12704408e+00 8.12558889e-01 1.96788684e-01
-8.64497662e-01 -4.47131306e-01 -7.84400776e-02 -3.20602059e-01
-1.35039902e+00 -3.97549748e-01 4.94722605e-01 -6.22770667e-01
6.29008472e-01 5.37212908e-01 4.63557631e-01 9.93420482e-01
3.19917500e-02 5.06039262e-01 6.35764182e-01 -2.53224194e-01
3.22351307e-01 1.15919553e-01 4.67630148e-01 3.42446715e-01
7.96030104e-01 -6.33772254e-01 -6.66250825e-01 -7.23179996e-01
8.45802054e-02 -4.03244086e-02 -5.31145275e-01 6.13491356e-01
-1.50622833e+00 8.71314824e-01 2.08895326e-01 3.40704113e-01
-8.31128240e-01 -3.78122866e-01 8.32732141e-01 -2.25966927e-02
8.78266752e-01 5.48829675e-01 1.63776189e-01 1.79372862e-01
-1.45154476e+00 5.97607374e-01 8.88456523e-01 9.24113095e-01
7.04680860e-01 -7.56729245e-02 -8.01134408e-01 4.23739851e-01
-1.21795960e-01 8.47201467e-01 2.80813903e-01 -5.06031096e-01
9.66254890e-01 6.96317792e-01 6.20831668e-01 -1.86938190e+00
-5.37162602e-01 -2.99282342e-01 -1.19726288e+00 -4.38124180e-01
1.19577155e-01 -5.92714429e-01 -5.10620236e-01 1.42995560e+00
2.98157930e-01 -2.97350883e-01 1.53273493e-01 8.01611245e-01
1.23990011e+00 8.56008768e-01 1.40257776e-01 -4.57552731e-01
1.49138951e+00 -6.50287151e-01 -1.04679954e+00 -2.10447520e-01
6.36583745e-01 -6.90115631e-01 9.16327000e-01 -1.56778470e-02
-9.21368480e-01 -2.22441614e-01 -9.44354355e-01 2.22376093e-01
-3.53880733e-01 3.70890111e-01 5.51197454e-02 3.90930355e-01
-7.74089038e-01 5.91780663e-01 -4.03067410e-01 -8.88405263e-01
2.87259102e-01 -9.56583545e-02 -5.98327518e-01 2.20748186e-01
-1.33519363e+00 7.83509731e-01 5.70870519e-01 -1.99501127e-01
-4.76972282e-01 -5.71487546e-01 -7.25638092e-01 2.23171599e-02
1.88701391e-01 -9.18941736e-01 1.21564794e+00 -4.17469472e-01
-7.70580053e-01 7.44044185e-01 -1.99565828e-01 -4.58232522e-01
5.67076981e-01 -3.58128130e-01 -1.07146129e-01 5.33216834e-01
8.12854230e-01 3.45177561e-01 3.43318969e-01 -1.41740179e+00
-6.94864035e-01 -1.25488847e-01 -1.35976076e-01 2.14488417e-01
-3.78133833e-01 1.26060665e-01 1.26818016e-01 -8.55124295e-01
-2.99717665e-01 -7.56035209e-01 -2.33395189e-01 -7.43286610e-01
-1.03385091e+00 -8.66455883e-02 7.52238989e-01 -9.37961638e-01
1.82225668e+00 -1.92045486e+00 -3.51853639e-01 -9.21115726e-02
1.82698354e-01 1.87609613e-01 2.87286229e-02 1.34516406e+00
2.20058501e-01 7.86694646e-01 -4.84186202e-01 -6.40534878e-01
9.42262542e-03 -1.14115246e-01 -9.19570148e-01 3.09728563e-01
9.27167535e-02 5.78760505e-01 -1.32922530e+00 -7.76109815e-01
-1.50511488e-01 4.57182713e-02 -8.88590589e-02 3.26515794e-01
-4.02195603e-02 5.94152868e-01 -6.20918453e-01 3.44428569e-01
6.15547359e-01 9.54188313e-03 -1.38980985e-01 -2.21166953e-01
-4.29286301e-01 3.24661314e-01 -9.73628402e-01 1.33858538e+00
-2.79047731e-02 5.88056207e-01 -3.00052166e-01 -5.41275561e-01
1.04661202e+00 6.68855667e-01 3.20701569e-01 4.42146622e-02
1.62098974e-01 2.53336996e-01 -8.24294925e-01 -4.82073665e-01
1.33027554e+00 -1.37533858e-01 -8.84581268e-01 1.08464789e+00
-2.43763670e-01 -4.31516439e-01 5.42140901e-01 6.68602169e-01
1.02116811e+00 -6.69829309e-01 6.59498513e-01 -1.64507985e-01
9.57355872e-02 7.29518473e-01 2.17100933e-01 9.93629277e-01
-4.58476059e-02 1.13881278e+00 4.96663779e-01 -5.24401009e-01
-1.07627857e+00 -5.13663709e-01 2.96440452e-01 5.25405407e-01
2.64239293e-02 -7.02433050e-01 -1.09038591e+00 -5.16923547e-01
-4.06480551e-01 1.11426485e+00 -7.78252304e-01 1.18856076e-02
-3.19188148e-01 -1.07078600e+00 9.48984087e-01 1.00379989e-01
7.14179456e-01 -9.78737414e-01 -8.76744628e-01 1.87936693e-01
-1.09032929e+00 -1.16132593e+00 -6.07158542e-01 -4.43883777e-01
-4.77045953e-01 -9.20781136e-01 -4.75249439e-01 -2.33731821e-01
7.12653816e-01 6.03159428e-01 1.05634332e+00 1.49663910e-01
2.88815260e-01 3.22178513e-01 -7.04007149e-01 -5.60883164e-01
-4.82872695e-01 4.20077115e-01 1.46401837e-01 5.51148504e-03
-6.29020929e-02 -4.03184205e-01 -4.18839365e-01 -1.23725012e-02
-1.23484886e+00 7.39043579e-02 1.60390809e-01 4.52645719e-01
3.22422683e-01 2.95724899e-01 9.18068349e-01 -8.43213379e-01
1.35857940e+00 -9.17632759e-01 2.30916977e-01 3.21529746e-01
-5.76729439e-02 -1.67781398e-01 5.83832026e-01 -1.48529813e-01
-1.00659633e+00 -3.31272960e-01 3.15661490e-01 2.54127651e-01
-8.99446309e-02 9.85967219e-01 2.70285428e-01 7.13281870e-01
1.03477824e+00 1.26197502e-01 -3.90309900e-01 -2.37493426e-01
1.95046902e-01 1.00607574e+00 8.64790201e-01 -2.82679826e-01
8.80825877e-01 4.90879357e-01 -4.13025051e-01 -8.72241437e-01
-1.29520261e+00 -4.83205497e-01 -5.73605955e-01 -3.05829078e-01
7.66742766e-01 -9.37372983e-01 -2.08173558e-01 5.12547791e-01
-1.55903888e+00 -1.16180152e-01 -2.94289410e-01 1.93385065e-01
-1.78224251e-01 5.89733124e-01 -1.12166598e-01 -8.63906085e-01
-1.16426623e+00 -5.70425928e-01 1.36783552e+00 2.22164661e-01
-9.23101544e-01 -9.31052029e-01 2.96740711e-01 2.54738539e-01
4.43533808e-01 1.16458499e+00 7.28442430e-01 -9.96535242e-01
-7.49828527e-04 -7.00846136e-01 -2.11213008e-01 -3.02905977e-01
3.91824305e-01 2.15456292e-01 -7.34811425e-01 2.98418850e-02
-2.40505636e-01 -1.13420337e-01 7.43955255e-01 3.11526448e-01
2.59362072e-01 -1.02191436e+00 -3.24313700e-01 -2.57385105e-01
1.19126248e+00 -4.34137106e-01 4.21781123e-01 4.66129959e-01
5.65781832e-01 8.95952404e-01 8.79737318e-01 1.02822697e+00
7.64354527e-01 4.14808124e-01 2.63339490e-01 9.13688689e-02
6.60065040e-02 -4.26682144e-01 4.03196901e-01 8.47923994e-01
-7.10829049e-02 -8.40363204e-01 -1.07045710e+00 8.96153808e-01
-1.94523478e+00 -1.42440808e+00 -5.13599217e-01 2.25091505e+00
8.89316797e-01 -1.33384585e-01 3.34396034e-01 3.78060669e-01
9.98137414e-01 3.85208637e-01 9.23385024e-02 -3.01425219e-01
-3.21696043e-01 -5.19323051e-01 3.04965466e-01 5.18456936e-01
-1.00808406e+00 7.08984137e-01 5.99600649e+00 8.01625371e-01
-1.05147171e+00 3.26279961e-02 4.55233485e-01 1.78792417e-01
-5.53034961e-01 -6.71242550e-02 -6.56069875e-01 4.63854581e-01
1.02095592e+00 -7.39422143e-01 -1.62163854e-01 5.16057372e-01
8.53209734e-01 -4.93236184e-01 -5.79234064e-01 6.39005780e-01
2.28534088e-01 -1.48658419e+00 4.18569446e-01 -2.14984432e-01
9.63962793e-01 -1.77487671e-01 -4.14443254e-01 6.75700158e-02
1.87769309e-01 -6.59400582e-01 1.10121822e+00 5.80113769e-01
7.05152035e-01 -6.88101947e-01 1.20343685e+00 5.64122379e-01
-1.06764317e+00 2.96454042e-01 -1.28562704e-01 -7.83426762e-02
6.88647747e-01 1.14902186e+00 -1.18862963e+00 9.64785576e-01
4.70183194e-01 5.08075178e-01 -4.87572908e-01 9.51909602e-01
-4.53244746e-01 6.75107419e-01 -1.92095265e-01 -1.75092573e-04
2.51742154e-01 1.06997661e-01 1.06369925e+00 1.70735919e+00
5.78069448e-01 4.23365355e-01 5.14090478e-01 4.96331483e-01
-7.01998398e-02 2.81001925e-01 -7.73877800e-01 -2.06669107e-01
7.82753348e-01 1.36708367e+00 -9.71705735e-01 -6.47044897e-01
2.45370299e-01 6.27905309e-01 1.41892452e-02 1.94028124e-01
-6.10021830e-01 -4.62952107e-01 1.25851095e-01 3.76863539e-01
-3.00397545e-01 -2.02384382e-01 -6.68459594e-01 -1.03035736e+00
-9.75838825e-02 -8.16052735e-01 3.37676048e-01 -7.92370200e-01
-1.09699953e+00 1.11658490e+00 5.47752380e-01 -1.35702896e+00
-2.69353390e-01 3.79923701e-01 -1.23225880e+00 4.57318574e-01
-1.01732934e+00 -1.01863503e+00 -8.49613130e-01 6.66372329e-02
6.30092502e-01 2.23059177e-01 7.59785116e-01 1.52596086e-01
-6.72718823e-01 2.35350370e-01 -3.93422425e-01 1.48961961e-01
7.86459744e-01 -1.18814075e+00 7.32124746e-01 9.80849028e-01
-3.73494387e-01 5.40945709e-01 1.31036174e+00 -1.12439501e+00
-7.20287681e-01 -1.48004353e+00 1.61197913e+00 -4.29590583e-01
6.01767063e-01 -3.43226418e-02 -8.19944322e-01 4.78593230e-01
4.75427389e-01 -6.58919573e-01 8.04086447e-01 -1.94518119e-01
8.12943354e-02 3.42965305e-01 -1.17201984e+00 7.28681326e-01
5.90478599e-01 -2.47762680e-01 -8.79469872e-01 6.66445673e-01
7.54596412e-01 -2.81629831e-01 -6.06203139e-01 1.45750985e-01
2.17587113e-01 -8.34491611e-01 4.16813731e-01 -1.09074190e-01
4.68874097e-01 -5.11464775e-01 7.19367415e-02 -1.53984928e+00
-2.23237020e-03 -8.85870159e-01 4.62840825e-01 1.77367055e+00
3.75524879e-01 -4.92324561e-01 2.76281573e-02 6.78485036e-01
-3.09430033e-01 -3.31854343e-01 -4.60481048e-01 -4.13935333e-01
-3.46725881e-01 -2.26261750e-01 7.22304165e-01 9.26556826e-01
4.07139093e-01 5.35536945e-01 -5.09740770e-01 2.43908703e-01
4.31409359e-01 1.44218221e-01 1.16333902e+00 -1.28760624e+00
5.42633712e-01 -2.58740455e-01 -2.43545827e-02 -2.57449478e-01
-2.03245819e-01 -5.25280774e-01 1.02593832e-01 -2.06706214e+00
4.96088058e-01 -1.24585509e-01 5.85491240e-01 6.04858398e-01
-4.72011983e-01 1.42768562e-01 4.88280617e-02 6.18459463e-01
-7.49413788e-01 5.10457218e-01 8.25283706e-01 -4.78396490e-02
-4.48351592e-01 -1.95202172e-01 -9.39535975e-01 6.05086267e-01
9.52972770e-01 -9.60813344e-01 -2.26147071e-01 -3.58337939e-01
4.90099549e-01 1.33558186e-02 3.24180663e-01 -1.00253499e+00
2.26667747e-01 -2.36070961e-01 -2.33678192e-01 -8.96492600e-01
-2.50232190e-01 -1.01627178e-01 4.01181251e-01 2.87735164e-01
-2.97521383e-01 3.30294400e-01 2.34532639e-01 5.21893978e-01
-4.59830463e-01 -2.31386825e-01 2.85550147e-01 -1.34906471e-01
-9.89898574e-04 1.40781179e-01 -6.49588048e-01 2.88301587e-01
7.18439937e-01 -2.64649957e-01 -7.23405659e-01 -8.22370589e-01
-3.17178577e-01 1.83165893e-01 4.75697726e-01 4.05365154e-02
2.19192281e-01 -1.13207674e+00 -1.36627316e+00 -5.72633743e-01
1.19976006e-01 6.20830841e-02 2.78422803e-01 1.01938570e+00
-6.16793096e-01 2.58946896e-01 -7.70719573e-02 -1.42132819e-01
-9.28548396e-01 2.38793090e-01 -2.08208770e-01 -6.28769517e-01
-5.69774210e-01 1.83737800e-01 -3.77325602e-02 -8.53020102e-02
-2.92081267e-01 -3.93392354e-01 -5.50712883e-01 7.58481085e-01
9.69257951e-01 6.61880910e-01 4.04945686e-02 -1.07884681e+00
-3.23877662e-01 3.83352101e-01 3.33083689e-01 -2.95865923e-01
1.34375823e+00 -4.25064236e-01 -1.26301646e-01 3.39274347e-01
7.05823720e-01 4.64894593e-01 -6.07386291e-01 7.10814819e-02
7.86710456e-02 -1.31550893e-01 -2.02063099e-01 -5.26038289e-01
-6.37720644e-01 4.62920636e-01 -3.19584280e-01 7.42058992e-01
9.84754086e-01 -2.38909230e-01 1.13371193e+00 2.87434220e-01
3.31426352e-01 -8.92918050e-01 7.37612471e-02 3.95337760e-01
1.40009141e+00 -1.30273938e+00 3.27820033e-01 -3.76737297e-01
-1.01501644e+00 1.10806894e+00 -7.34199360e-02 -8.32510144e-02
2.83232957e-01 -4.21710573e-02 2.34209783e-02 -4.78407294e-01
-5.77287376e-01 -7.16047063e-02 2.76217759e-01 4.43635911e-01
2.27994546e-01 2.45337963e-01 -6.35688782e-01 9.40404475e-01
-6.03094935e-01 -3.99037570e-01 9.66327369e-01 7.83080161e-01
-7.87930012e-01 -5.35322845e-01 -6.86862826e-01 2.18396053e-01
-5.75243175e-01 -8.60166028e-02 -8.53887916e-01 7.17853963e-01
-1.67673245e-01 1.59314811e+00 -2.34801501e-01 -4.37665910e-01
4.60533530e-01 -2.23091871e-01 -3.87260318e-01 -8.50034475e-01
-1.01840794e+00 -3.60890746e-01 5.54895282e-01 8.25876668e-02
-4.97348070e-01 -8.70533526e-01 -1.47228241e+00 -6.13103807e-01
-3.03128242e-01 6.57239258e-01 4.21478093e-01 1.06019664e+00
5.21906316e-01 3.25056468e-03 6.63830996e-01 -1.19606102e+00
-1.50190979e-01 -1.31469202e+00 -3.53232414e-01 5.56193233e-01
3.81591529e-01 -3.51836711e-01 -5.60718715e-01 3.38838279e-01]
|
[12.505966186523438, 9.4557466506958]
|
730a25df-2d78-4a7f-adfc-7a7876b7964b
|
efficient-few-shot-learning-for-pixel-precise
|
2210.15570
| null |
https://arxiv.org/abs/2210.15570v1
|
https://arxiv.org/pdf/2210.15570v1.pdf
|
Efficient few-shot learning for pixel-precise handwritten document layout analysis
|
Layout analysis is a task of uttermost importance in ancient handwritten document analysis and represents a fundamental step toward the simplification of subsequent tasks such as optical character recognition and automatic transcription. However, many of the approaches adopted to solve this problem rely on a fully supervised learning paradigm. While these systems achieve very good performance on this task, the drawback is that pixel-precise text labeling of the entire training set is a very time-consuming process, which makes this type of information rarely available in a real-world scenario. In the present paper, we address this problem by proposing an efficient few-shot learning framework that achieves performances comparable to current state-of-the-art fully supervised methods on the publicly available DIVA-HisDB dataset.
|
['Claudio Piciarelli', 'Emanuela Colombi', 'Gian Luca Foresti', 'Matteo Paier', 'Silvia Zottin', 'Axel De Nardin']
|
2022-10-27
| null | null | null | null |
['document-layout-analysis']
|
['computer-vision']
|
[ 4.55946505e-01 -5.06339610e-01 -5.78604117e-02 -2.11361453e-01
-8.24661314e-01 -5.61655641e-01 7.25902498e-01 2.97253877e-01
-6.90925896e-01 7.80975103e-01 -3.67063247e-02 -2.91460752e-01
-2.10287198e-02 -6.35024428e-01 -4.43847030e-01 -8.55912685e-01
4.66543615e-01 7.64793634e-01 5.52427232e-01 -4.83918339e-02
7.41274655e-01 7.45879173e-01 -1.58527493e+00 3.05849165e-01
5.60383439e-01 8.39741826e-01 4.29423869e-01 6.54907346e-01
-6.26738071e-01 7.61095166e-01 -6.24123275e-01 -4.47421789e-01
-8.82744789e-02 -4.95520353e-01 -8.08241069e-01 4.96937633e-01
5.79864144e-01 -1.82525307e-01 -1.96258768e-01 1.02636456e+00
6.05816603e-01 2.59919226e-01 7.69788861e-01 -4.12241727e-01
-2.96359479e-01 3.56329501e-01 -5.61218202e-01 2.21118286e-01
1.17978245e-01 -2.23659292e-01 1.06049132e+00 -8.62657070e-01
7.89967477e-01 4.69581336e-01 5.11822879e-01 3.39396387e-01
-1.23182738e+00 6.60054609e-02 -1.14973374e-01 5.29438019e-01
-1.25778449e+00 -6.55141234e-01 1.04048860e+00 -5.06137490e-01
9.38375711e-01 1.86756060e-01 4.35127050e-01 9.90588903e-01
-1.09357387e-01 1.09193099e+00 1.04012322e+00 -9.33689594e-01
6.30365551e-01 7.09742978e-02 3.70156378e-01 7.42329001e-01
9.05514881e-02 -6.54642403e-01 -6.46938324e-01 1.27499133e-01
6.36158347e-01 -1.40608415e-01 -3.10631424e-01 -5.30912697e-01
-9.33299005e-01 4.81349260e-01 -3.81384790e-02 7.46326685e-01
-3.27060252e-01 -1.53375939e-01 4.24617946e-01 6.82751760e-02
4.45980698e-01 5.36690474e-01 4.96446062e-03 -5.83053410e-01
-1.62305796e+00 8.58115777e-03 6.95501328e-01 8.29646289e-01
5.99215746e-01 5.05287871e-02 3.10583394e-02 1.10582018e+00
-6.96523935e-02 2.23854497e-01 5.38846135e-01 -5.36589026e-01
3.69802237e-01 6.49299085e-01 -6.25927970e-02 -7.58048475e-01
-2.17527077e-01 -3.25259179e-01 -7.37664461e-01 2.91066736e-01
8.53782833e-01 3.00259233e-01 -1.09626305e+00 7.86365211e-01
-4.65817936e-02 -3.98055941e-01 -1.62442937e-01 7.48652816e-01
4.96771991e-01 6.34927928e-01 -3.46362531e-01 -2.69639343e-01
1.27024925e+00 -1.18940437e+00 -8.72784078e-01 -2.78707922e-01
4.54521328e-01 -1.09450579e+00 1.09237993e+00 8.86821568e-01
-7.38102853e-01 -3.43303412e-01 -1.22417247e+00 -3.04705977e-01
-4.00436372e-01 3.67973030e-01 5.67535102e-01 8.00307333e-01
-5.97971559e-01 7.30271220e-01 -6.81297779e-01 -7.26464748e-01
5.96223533e-01 -5.42232543e-02 -4.49378401e-01 -3.52791041e-01
-5.23937166e-01 7.63389289e-01 3.62218350e-01 2.14664027e-01
-5.48022807e-01 -3.10500473e-01 -5.64369202e-01 5.09473830e-02
8.05565774e-01 1.49596050e-01 1.27871323e+00 -5.24359465e-01
-1.68718255e+00 9.98603344e-01 -2.80674577e-01 -4.19735193e-01
7.83503294e-01 -3.19762170e-01 -2.05202043e-01 1.78226724e-01
-3.54246467e-01 6.91149384e-02 1.07403743e+00 -1.11531663e+00
-4.21938658e-01 -4.55617368e-01 -2.61687428e-01 4.38546985e-02
-5.54064989e-01 -7.82215688e-03 -9.54425156e-01 -6.99630976e-01
2.52932180e-02 -6.27772570e-01 -1.41329184e-01 3.12159032e-01
-5.05149364e-01 -7.76148289e-02 9.80551720e-01 -7.63672054e-01
1.17775214e+00 -2.13406897e+00 1.99314490e-01 2.34578326e-02
-1.59054592e-01 8.56132686e-01 1.30648538e-01 7.16080725e-01
3.45391095e-01 -3.19410503e-01 -6.07279301e-01 -5.86621761e-01
-1.68026000e-01 1.97557315e-01 -4.15128171e-01 4.79555666e-01
-1.96479157e-01 7.19160438e-01 -7.58425295e-01 -7.23474264e-01
5.47477305e-01 2.88017005e-01 -1.15687493e-02 2.79572252e-02
-4.93267179e-01 2.03148782e-01 -2.70115137e-01 7.41879046e-01
2.54785687e-01 -1.16032124e-01 2.37617463e-01 -8.86871852e-03
-4.78346050e-01 -9.74896550e-02 -1.25001848e+00 2.06690550e+00
-1.57302454e-01 1.02637517e+00 -2.16822714e-01 -1.23086143e+00
1.03236783e+00 1.02735184e-01 3.55748951e-01 -6.88402236e-01
2.25551397e-01 2.69177169e-01 -2.31360465e-01 -4.18321162e-01
9.22547221e-01 -1.05776884e-01 1.13911726e-01 6.05733216e-01
7.19913617e-02 -1.61053911e-01 6.17471218e-01 1.35566313e-02
9.80526924e-01 4.52289909e-01 4.40817565e-01 -1.33491009e-01
6.00982130e-01 2.72800922e-01 2.69894540e-01 6.73863053e-01
-1.89477861e-01 1.09816885e+00 5.34518361e-01 -5.19999027e-01
-1.32490706e+00 -5.82187355e-01 -1.24414712e-01 6.85159028e-01
-1.00455672e-01 -4.44777101e-01 -9.10738051e-01 -5.02677500e-01
-3.36518884e-01 6.44362092e-01 -5.37296772e-01 3.09683353e-01
-4.46502030e-01 -6.26073539e-01 6.56986773e-01 5.22652864e-01
5.29158294e-01 -9.86881614e-01 -9.17064071e-01 3.34009975e-01
-1.50115907e-01 -1.21620798e+00 -1.85744260e-02 3.03103298e-01
-8.49564850e-01 -9.22020018e-01 -1.26320076e+00 -7.30623841e-01
6.60651684e-01 2.49555245e-01 7.77626157e-01 -1.11953141e-02
-7.81178236e-01 6.64695129e-02 -5.15788734e-01 -3.30195308e-01
-5.44079021e-02 1.91093698e-01 -4.58211094e-01 2.38682002e-01
3.97482902e-01 -1.18796095e-01 -1.73488230e-01 6.63907500e-03
-9.29731548e-01 -7.10049346e-02 6.53224885e-01 8.21726739e-01
7.36469328e-01 3.78215283e-01 5.62218800e-02 -1.18908405e+00
4.32004571e-01 3.01562160e-01 -7.55291104e-01 5.90943396e-01
-4.53102529e-01 1.39219552e-01 7.22853720e-01 -8.47098231e-02
-1.18642449e+00 3.63456368e-01 -3.02536458e-01 -1.64742798e-01
-3.97910446e-01 5.96622944e-01 -1.43692493e-01 -2.00833511e-02
4.43496287e-01 5.06435871e-01 -1.88022748e-01 -8.86537671e-01
1.17121585e-01 9.16208386e-01 7.90951788e-01 -3.35929722e-01
7.13325083e-01 6.18333042e-01 2.56389320e-01 -1.68083954e+00
-7.91350543e-01 -7.83037007e-01 -1.09527075e+00 -2.91683972e-01
7.90561020e-01 -3.04079443e-01 -3.43612805e-02 8.37662578e-01
-1.05079985e+00 -3.76356244e-01 -3.77300531e-01 2.01398700e-01
-6.17631614e-01 9.28442955e-01 -3.97542566e-01 -9.48153079e-01
-6.41210377e-02 -1.00513804e+00 9.68457103e-01 1.49472773e-01
-1.88136280e-01 -9.49979007e-01 2.41232291e-01 5.62235475e-01
2.27635324e-01 -6.38275966e-02 1.00248599e+00 -4.41639394e-01
-6.65225983e-01 -6.65867209e-01 -1.10607490e-01 3.30896646e-01
9.65160653e-02 3.45496148e-01 -1.13983071e+00 7.11091161e-02
-1.65257573e-01 -4.49179232e-01 1.06977916e+00 2.61569023e-01
1.12682593e+00 4.56227392e-01 -1.35484472e-01 2.90690094e-01
1.51708269e+00 7.81856701e-02 9.33056533e-01 3.71707886e-01
6.77774906e-01 6.69215262e-01 7.79893935e-01 4.94879603e-01
-1.34320125e-01 7.61519790e-01 -3.13807055e-02 7.32938480e-03
-3.65241170e-01 -1.13773562e-01 -2.23512664e-01 7.16127098e-01
-2.96741903e-01 -4.18975651e-01 -1.20698988e+00 5.48905969e-01
-2.06881642e+00 -1.14172196e+00 -2.27382064e-01 2.23353148e+00
7.13985980e-01 2.25318864e-01 -6.85860664e-02 7.96474397e-01
6.67966723e-01 3.90673488e-01 -1.54214278e-01 -3.16364318e-01
-3.77335310e-01 1.14854954e-01 3.32057148e-01 2.33175233e-01
-1.26149547e+00 9.36925411e-01 6.09509754e+00 1.05780625e+00
-1.16898751e+00 -1.38164476e-01 2.59735197e-01 1.29473552e-01
1.19075723e-01 -5.22416830e-02 -6.57551587e-01 3.54970396e-01
6.85774326e-01 2.08192527e-01 2.48396620e-01 7.77738392e-01
-1.64402253e-03 -6.48835897e-01 -1.08598077e+00 1.23606479e+00
5.71268141e-01 -1.46685541e+00 -6.17343597e-02 1.15602300e-01
6.48281932e-01 -3.50840569e-01 -1.47883296e-01 -7.35158771e-02
-3.56914878e-01 -8.65494668e-01 6.72514856e-01 5.78173220e-01
8.93680334e-01 -6.43095613e-01 6.19968176e-01 4.99102294e-01
-9.19375420e-01 7.69644454e-02 -5.43051660e-01 -1.05299242e-01
4.72776771e-01 9.01342630e-01 -5.50695419e-01 4.72007602e-01
2.84566849e-01 6.28371775e-01 -6.29065394e-01 1.51974201e+00
-4.40579504e-01 4.26720798e-01 -1.52306799e-02 -1.69881508e-01
2.86158979e-01 -8.75514746e-02 3.53746057e-01 1.20072961e+00
1.48919925e-01 -7.98092037e-02 -7.61389285e-02 4.38205481e-01
-2.18894601e-01 3.56377840e-01 -5.16679943e-01 -5.46022296e-01
8.49301219e-02 1.31678784e+00 -1.44423378e+00 -3.25628966e-01
-5.64820468e-01 1.37296295e+00 4.72912073e-01 1.18690684e-01
-4.62532908e-01 -6.77407742e-01 -7.43239820e-02 4.10198420e-02
5.85857332e-01 -5.19107997e-01 -3.69478971e-01 -1.12119985e+00
2.01749682e-01 -6.71528637e-01 1.17932558e-01 -5.05433023e-01
-8.90617430e-01 3.59690994e-01 -4.20457512e-01 -1.23764920e+00
-8.73055011e-02 -9.46387172e-01 -5.68495810e-01 6.10654354e-01
-1.48133373e+00 -1.23936617e+00 -2.49964640e-01 3.41174573e-01
9.70276415e-01 -2.18207926e-01 1.07180226e+00 4.91325021e-01
-7.02521145e-01 3.50888491e-01 5.95026433e-01 2.41613656e-01
9.29862618e-01 -1.25721455e+00 3.66223186e-01 1.13725877e+00
7.74413645e-01 3.47108632e-01 7.47292995e-01 -5.59732914e-01
-1.51059330e+00 -6.25647247e-01 1.00234878e+00 -2.30128288e-01
5.20236254e-01 -4.31577533e-01 -1.03999078e+00 2.32120693e-01
-5.92054203e-02 -2.63386909e-02 7.92589843e-01 1.92077562e-01
-2.49501944e-01 1.26867965e-01 -6.24505639e-01 4.37841594e-01
7.21498191e-01 -8.56518745e-01 -7.92526782e-01 2.86850423e-01
-2.76030779e-01 -1.89115584e-01 -5.01200795e-01 -1.00912109e-01
5.79762697e-01 -9.50629115e-01 5.56161463e-01 -2.52771348e-01
5.71811974e-01 -2.89350420e-01 -1.00751981e-01 -1.02650380e+00
-1.04282185e-01 -5.08533478e-01 -2.46253222e-01 1.33799732e+00
2.54588962e-01 -2.84112599e-02 1.15969765e+00 4.14312661e-01
-1.92090854e-01 -5.39798379e-01 -9.37224507e-01 -9.06952322e-01
-1.66359991e-01 -3.07611048e-01 -4.23684008e-02 8.00113916e-01
1.84744801e-02 3.25875252e-01 -6.69818342e-01 -4.06555891e-01
8.83489192e-01 3.69618416e-01 6.75884545e-01 -1.46349978e+00
-2.97424078e-01 -4.57771868e-01 -6.98715329e-01 -9.09257829e-01
4.92750779e-02 -4.95494008e-01 3.82658690e-01 -1.80355000e+00
2.55427301e-01 -6.77250847e-02 5.53836264e-02 2.46701181e-01
1.80711433e-01 5.19259155e-01 -5.20119555e-02 3.74652475e-01
-6.03957057e-01 3.83781642e-01 7.97831416e-01 -2.41261125e-01
6.60319328e-02 -2.06543744e-01 -1.85919866e-01 8.22494507e-01
6.51344359e-01 -2.10071415e-01 -3.01057398e-01 -5.61043978e-01
-8.60144347e-02 -3.48372042e-01 -2.98001878e-02 -1.12322426e+00
6.30593717e-01 -1.86720528e-02 3.59035730e-01 -9.17281985e-01
4.52016443e-01 -7.24547207e-01 -2.54141897e-01 3.17765415e-01
-1.28050774e-01 -4.12233084e-01 -4.82614040e-02 5.81359267e-01
-3.03047329e-01 -8.09383512e-01 7.76539803e-01 -2.30828717e-01
-1.28005123e+00 -4.49377410e-02 -5.66783488e-01 -1.39206395e-01
1.06698704e+00 -4.42567259e-01 -4.18361068e-01 -2.31728069e-02
-3.60111982e-01 -3.25321108e-01 7.97261357e-01 2.05104381e-01
6.31308496e-01 -8.09441030e-01 -3.92094344e-01 1.03330903e-01
3.61571819e-01 -7.74808377e-02 2.95414865e-01 7.10583806e-01
-9.04006541e-01 7.16224134e-01 -4.67590839e-01 -3.75458896e-01
-1.50635338e+00 5.43419063e-01 -5.92349917e-02 -2.51351327e-01
-8.95884871e-01 7.27784753e-01 -3.87285680e-01 1.35052383e-01
4.34137344e-01 2.11129099e-01 -3.38005573e-01 4.54552680e-01
8.42906594e-01 5.53865671e-01 6.13918543e-01 -7.42305279e-01
-1.48720443e-01 7.15059578e-01 -2.85917133e-01 -3.05110455e-01
1.39759839e+00 -1.76268618e-03 -1.26050161e-02 8.91301095e-01
8.29897404e-01 -5.14895916e-02 -1.03405225e+00 -3.66120338e-01
4.55110639e-01 -7.03553855e-01 2.92828619e-01 -5.49242795e-01
-6.42563105e-01 1.29590642e+00 2.26486415e-01 8.37118551e-03
8.37157607e-01 -2.05744132e-01 6.63324118e-01 1.00080812e+00
4.93081212e-01 -1.80291581e+00 5.69316559e-02 4.42542821e-01
5.72064102e-01 -1.18792522e+00 4.59331393e-01 -3.45771730e-01
-5.55364013e-01 1.39401877e+00 2.21663579e-01 9.05333236e-02
3.28861326e-01 2.84085989e-01 1.36790767e-01 -1.24157198e-01
-2.88114607e-01 -3.73491853e-01 3.70262891e-01 2.92064190e-01
6.14847660e-01 -3.04258138e-01 -4.80910748e-01 1.96403816e-01
2.84685969e-01 7.55328909e-02 7.02170789e-01 1.68085587e+00
-5.85334063e-01 -1.38192666e+00 -2.26623863e-01 3.75028938e-01
-5.68759918e-01 6.25191852e-02 -6.49581373e-01 4.64512676e-01
-2.61151671e-01 5.75842202e-01 -9.43588093e-03 1.93448812e-02
2.95957029e-01 4.38477993e-01 8.68386686e-01 -7.55639732e-01
-1.36072367e-01 3.15574437e-01 9.75107700e-02 -2.67579943e-01
-4.92065936e-01 -7.88479269e-01 -1.13018322e+00 -6.95576891e-02
-3.78367364e-01 -7.05402717e-02 1.01691079e+00 9.74019587e-01
-1.64967045e-01 5.38693786e-01 2.09834978e-01 -9.91419256e-01
-6.02840900e-01 -7.56138146e-01 -9.14789200e-01 4.74228561e-01
1.15352057e-01 -4.94589776e-01 8.05425569e-02 3.87803495e-01]
|
[11.794281959533691, 2.5499942302703857]
|
6a11d9a8-4c32-4c69-88e2-93796c2e3789
|
perceptual-grouping-in-vision-language-models
|
2210.09996
| null |
https://arxiv.org/abs/2210.09996v2
|
https://arxiv.org/pdf/2210.09996v2.pdf
|
Perceptual Grouping in Contrastive Vision-Language Models
|
Recent advances in zero-shot image recognition suggest that vision-language models learn generic visual representations with a high degree of semantic information that may be arbitrarily probed with natural language phrases. Understanding an image, however, is not just about understanding what content resides within an image, but importantly, where that content resides. In this work we examine how well vision-language models are able to understand where objects reside within an image and group together visually related parts of the imagery. We demonstrate how contemporary vision and language representation learning models based on contrastive losses and large web-based data capture limited object localization information. We propose a minimal set of modifications that results in models that uniquely learn both semantic and spatial information. We measure this performance in terms of zero-shot image recognition, unsupervised bottom-up and top-down semantic segmentations, as well as robustness analyses. We find that the resulting model achieves state-of-the-art results in terms of unsupervised segmentation, and demonstrate that the learned representations are uniquely robust to spurious correlations in datasets designed to probe the causal behavior of vision models.
|
['Jonathon Shlens', 'Alexander Toshev', 'Yinfei Yang', 'Sachin Ravi', 'Brandon McKinzie', 'Kanchana Ranasinghe']
|
2022-10-18
| null | null | null | null |
['unsupervised-semantic-segmentation-with', 'unsupervised-semantic-segmentation']
|
['computer-vision', 'computer-vision']
|
[ 4.94093716e-01 -3.04709598e-02 -2.70529062e-01 -5.10456085e-01
-6.85436487e-01 -5.38079321e-01 1.01002717e+00 2.90686309e-01
-3.34691972e-01 1.54046059e-01 2.90727466e-01 -6.97453544e-02
-3.27658802e-01 -7.86239147e-01 -1.08546817e+00 -6.14980817e-01
1.47104012e-02 4.49884146e-01 3.32485557e-01 -3.75696011e-02
4.65628535e-01 3.65820616e-01 -1.83097982e+00 2.36137316e-01
4.42055970e-01 9.54360485e-01 4.38082427e-01 7.45413661e-01
-2.26942286e-01 1.06428480e+00 -1.56611502e-01 -1.61586434e-01
1.28355101e-01 -4.19514418e-01 -9.51439619e-01 5.91774166e-01
9.86508429e-01 -1.56400949e-01 -3.76556635e-01 1.31089878e+00
-1.26747340e-01 6.56432956e-02 8.38532269e-01 -1.16623628e+00
-1.12860668e+00 3.17170352e-01 -6.99800551e-01 4.13362950e-01
7.05025792e-02 2.37348810e-01 1.33028066e+00 -8.05340886e-01
7.23828971e-01 1.34454584e+00 5.08206964e-01 2.85054922e-01
-1.64934802e+00 -4.58877236e-02 3.60925555e-01 2.28355899e-01
-1.23396838e+00 -4.37770963e-01 4.40072715e-01 -8.72342467e-01
8.63658547e-01 1.15636826e-01 3.63304466e-01 9.19469893e-01
1.03251845e-01 5.42100966e-01 1.11002421e+00 -5.36052942e-01
2.50273675e-01 1.49118796e-01 6.31733298e-01 9.76685643e-01
2.60879278e-01 1.93271898e-02 -7.19382703e-01 1.64020613e-01
6.94713950e-01 8.04927349e-02 -4.64595621e-03 -6.20655000e-01
-1.11466062e+00 9.23903346e-01 6.93290353e-01 2.87748933e-01
-2.00933531e-01 4.83301878e-01 3.90033498e-02 -1.67011395e-02
3.31327945e-01 3.87860894e-01 -2.93657303e-01 2.08728313e-01
-1.01074684e+00 -4.54815552e-02 7.25582302e-01 8.69763434e-01
1.26801181e+00 -1.36490494e-01 -1.68786451e-01 8.12987506e-01
3.07598978e-01 4.48132485e-01 3.46676499e-01 -1.23031628e+00
6.45378232e-03 4.24104899e-01 -7.98710659e-02 -1.17494345e+00
-2.65133560e-01 -2.95507967e-01 -6.00132823e-01 2.83830374e-01
3.78942430e-01 3.19854885e-01 -1.35533738e+00 1.96804571e+00
-2.37075180e-01 2.04388320e-01 -3.97454761e-02 8.82681787e-01
7.17765212e-01 4.29499269e-01 2.60808557e-01 -6.81900606e-02
1.50200093e+00 -8.83567572e-01 -3.22593242e-01 -7.87002325e-01
3.08926046e-01 -4.58976239e-01 1.20278156e+00 -5.42011596e-02
-9.38105822e-01 -6.60795629e-01 -9.11832452e-01 -3.22763354e-01
-5.91813326e-01 -1.43562019e-01 5.93045890e-01 5.92154324e-01
-1.18084824e+00 5.44183314e-01 -5.77974677e-01 -7.89396644e-01
7.19661653e-01 4.39400040e-03 -4.37479854e-01 -3.99848282e-01
-7.45913088e-01 7.63882339e-01 2.85006821e-01 -3.99883002e-01
-1.35879016e+00 -7.57288992e-01 -9.84810233e-01 1.28486425e-01
5.15237212e-01 -6.85294986e-01 9.64492381e-01 -1.04847348e+00
-6.11521780e-01 1.36558354e+00 -5.49782097e-01 -6.49646938e-01
1.73079431e-01 -4.67051519e-03 9.13481638e-02 3.97361577e-01
5.42281985e-01 8.96237016e-01 1.01568449e+00 -1.57210672e+00
-5.85519254e-01 -7.30488539e-01 1.57325521e-01 -1.51861161e-02
-3.83933559e-02 2.51029134e-02 -6.98176801e-01 -4.77122426e-01
3.42633992e-01 -8.05798948e-01 -2.51999170e-01 3.51730287e-01
-4.04192328e-01 1.41773336e-02 5.58932006e-01 -4.75572288e-01
4.92743999e-01 -2.06795669e+00 6.60484135e-02 6.50191307e-02
2.62979597e-01 -2.92778432e-01 -3.25173020e-01 2.39625707e-01
-1.23285539e-02 5.54994226e-01 -4.66037840e-01 -3.59709978e-01
-1.46918138e-02 5.76755702e-01 -6.38133228e-01 6.21989369e-01
4.22458202e-01 1.11571693e+00 -8.06734681e-01 -4.27141666e-01
3.60880494e-01 2.38552004e-01 -4.74512607e-01 -3.53601798e-02
-4.86937821e-01 2.12921351e-01 -2.99510270e-01 7.22068250e-01
3.57337147e-01 -7.26755321e-01 -1.67236790e-01 2.62134783e-02
-8.50839987e-02 -4.14907485e-02 -6.42695010e-01 1.87347472e+00
-1.68325990e-01 9.48455393e-01 -1.15979329e-01 -1.38280487e+00
5.77829421e-01 -2.58359998e-01 3.98631185e-01 -7.51392782e-01
-1.35638699e-01 -2.70004719e-01 -2.27020919e-01 -5.90886593e-01
4.31312710e-01 -1.76720291e-01 1.61797404e-02 4.51476008e-01
3.78643721e-01 -7.94415995e-02 9.93360355e-02 5.07027924e-01
9.82419491e-01 -1.20867819e-01 9.31554362e-02 -3.17711860e-01
1.21550076e-01 3.03690344e-01 1.43474758e-01 1.41457164e+00
-2.02138454e-01 8.78239036e-01 5.14293313e-01 -1.38962343e-01
-1.14613068e+00 -1.32498264e+00 -2.07136527e-01 1.45935512e+00
4.15195107e-01 -2.05826864e-01 -6.87319219e-01 -1.35488763e-01
-1.63728073e-02 8.84353995e-01 -1.01587331e+00 -2.07941622e-01
1.37374014e-01 -6.23276472e-01 3.15969437e-01 4.49476451e-01
3.12721848e-01 -9.91406798e-01 -6.52557790e-01 -2.47374997e-01
-6.35936484e-02 -1.35487247e+00 -1.13530636e-01 2.41798788e-01
-6.46873534e-01 -1.30395758e+00 -4.61868644e-01 -7.27378786e-01
7.18785226e-01 8.54091287e-01 1.18149197e+00 2.81605520e-03
-9.31366742e-01 1.06710362e+00 -2.48032674e-01 -2.08466068e-01
-1.49089828e-01 -4.16182250e-01 -1.15283087e-01 2.02777922e-01
5.84948838e-01 -4.42117274e-01 -4.06145453e-01 1.83804646e-01
-8.91947031e-01 -1.35038599e-01 6.14800394e-01 8.34280133e-01
7.61117935e-01 -7.24786818e-02 4.21814546e-02 -7.48096824e-01
2.83484280e-01 -6.56072378e-01 -4.36201960e-01 5.19016385e-01
-5.76327205e-01 3.08730274e-01 1.42866045e-01 -1.21415183e-01
-8.81943524e-01 6.89024404e-02 4.49785829e-01 -6.90488100e-01
-5.10272264e-01 3.26592237e-01 1.24118589e-01 -2.42849395e-01
8.41993928e-01 5.32337129e-01 8.12846571e-02 -3.70328814e-01
8.56472194e-01 2.31767684e-01 6.90064967e-01 -5.24235189e-01
6.39924765e-01 9.61110353e-01 -1.48586288e-01 -1.28689313e+00
-1.19950414e+00 -8.59952152e-01 -8.03086817e-01 -5.16291671e-02
1.22321117e+00 -1.03903699e+00 -4.73511934e-01 1.52222916e-01
-1.03257239e+00 -1.72669128e-01 -4.40732956e-01 1.64745718e-01
-9.58824217e-01 3.47641259e-01 -1.80478483e-01 -6.59323692e-01
2.79543132e-01 -1.01349056e+00 1.12721801e+00 1.87302127e-01
-4.78316173e-02 -8.65863979e-01 -1.76604763e-01 5.90140045e-01
5.97510189e-02 5.07660350e-03 1.05579257e+00 -4.73038584e-01
-9.96089101e-01 1.13642663e-01 -7.28541195e-01 2.90218055e-01
-4.03397419e-02 -1.53868049e-01 -1.17757785e+00 -9.49918851e-02
5.05051166e-02 -5.90747595e-01 1.51002717e+00 7.00406492e-01
1.30470800e+00 -2.18202606e-01 -1.37945250e-01 7.18558669e-01
1.80528820e+00 -3.12900513e-01 6.65937483e-01 2.55686104e-01
7.75493145e-01 9.54445243e-01 1.61290511e-01 2.88990378e-01
2.26489469e-01 3.35485488e-01 6.59306705e-01 7.85179734e-02
-1.87872291e-01 -4.32354867e-01 4.78065498e-02 1.45798251e-01
1.78990498e-01 -1.58960372e-02 -1.02967143e+00 1.00821745e+00
-1.98192811e+00 -1.07579458e+00 2.15697084e-02 2.00382710e+00
4.96983439e-01 1.27230596e-03 -1.36329591e-01 -3.73208672e-01
5.13455927e-01 4.96534646e-01 -8.55255425e-01 -1.39946744e-01
-3.73474896e-01 -1.42065184e-02 8.46730471e-01 4.09388602e-01
-1.08297396e+00 1.22232544e+00 7.24627399e+00 5.39520741e-01
-7.56619036e-01 1.85912088e-01 6.95599675e-01 1.43355969e-02
-3.15190613e-01 1.71415061e-01 -5.78865349e-01 6.02826364e-02
7.14931786e-01 -1.03024423e-01 6.12089574e-01 8.80453110e-01
1.96016744e-01 -3.72196227e-01 -1.24765480e+00 1.05158365e+00
3.92247647e-01 -1.60672474e+00 3.71214777e-01 1.19383782e-01
7.81350553e-01 4.01149064e-01 2.20866293e-01 -1.82248175e-01
4.58632886e-01 -1.36387455e+00 7.33862519e-01 8.74789834e-01
5.35605133e-01 -2.53628671e-01 1.21012211e-01 3.72457206e-01
-8.06794763e-01 -1.65966749e-01 -6.36095464e-01 -7.25284517e-02
-1.91144496e-01 1.97255343e-01 -4.49576169e-01 9.00526196e-02
8.43191743e-01 9.74741578e-01 -9.41771448e-01 8.80882978e-01
-2.60433376e-01 5.34659207e-01 -2.00165380e-02 1.97414801e-01
4.56809610e-01 -8.85608196e-02 4.45517480e-01 9.88087595e-01
-1.03704251e-01 5.57661131e-02 3.15048844e-01 1.50800312e+00
-1.31740242e-01 -1.43864766e-01 -8.50482702e-01 -2.38940090e-01
1.89392656e-01 8.64224970e-01 -8.08622777e-01 -2.30454654e-01
-5.69521368e-01 9.06578541e-01 4.50414211e-01 6.69319868e-01
-4.60871607e-01 1.25718862e-01 1.01414764e+00 9.28330943e-02
5.58644593e-01 -4.31701005e-01 -4.81266737e-01 -1.15040290e+00
-1.73886031e-01 -3.54782045e-01 1.49595767e-01 -1.13148022e+00
-1.49131656e+00 -6.85236827e-02 -4.75416817e-02 -7.35856295e-01
-1.50807783e-01 -8.59682858e-01 -4.46300060e-01 7.05114663e-01
-1.49855959e+00 -1.14906526e+00 -4.05011356e-01 6.71922505e-01
7.28354454e-01 -9.18554291e-02 8.12518120e-01 -3.54686171e-01
-2.27247223e-01 1.65511087e-01 2.18407556e-01 3.12048852e-01
3.10605735e-01 -1.16136026e+00 2.47338325e-01 9.97699678e-01
9.60558653e-01 7.47215152e-01 6.59261346e-01 -3.05561960e-01
-1.36014915e+00 -9.68641698e-01 4.33867365e-01 -5.99944055e-01
9.49197769e-01 -3.08228672e-01 -9.38541412e-01 7.44530439e-01
3.09768841e-02 3.63583684e-01 5.72888434e-01 3.23318839e-01
-9.26117063e-01 7.78701380e-02 -8.47724617e-01 4.94456977e-01
1.11301351e+00 -1.17936838e+00 -9.07229245e-01 4.57551956e-01
7.56469250e-01 2.97606021e-01 -4.54788476e-01 -1.16141271e-02
4.26395684e-01 -1.08198845e+00 1.25745785e+00 -9.79440868e-01
6.53998733e-01 -1.41698539e-01 -6.66244388e-01 -8.86237383e-01
-4.20078903e-01 -6.38961121e-02 2.82308996e-01 8.63632500e-01
3.50652009e-01 -2.79722244e-01 6.74933910e-01 7.64508247e-01
1.91510811e-01 -1.22450151e-01 -7.12943554e-01 -7.17557430e-01
9.59927365e-02 -6.18209302e-01 -1.77790359e-01 8.64255726e-01
-4.19997960e-01 3.98125887e-01 -2.20758706e-01 4.29358780e-01
1.18878174e+00 3.46817076e-01 4.93476987e-01 -1.36018634e+00
-1.17309429e-01 -4.61404324e-01 -8.08930576e-01 -1.00919187e+00
3.34127188e-01 -7.69401014e-01 1.65585876e-01 -1.68569458e+00
6.63478732e-01 -1.40359476e-01 -4.34983790e-01 5.17699659e-01
1.39552921e-01 3.68583292e-01 2.70871609e-01 5.41900039e-01
-7.44060099e-01 2.87629604e-01 8.03552628e-01 -5.12335598e-01
2.41499200e-01 -4.69761372e-01 -9.40262914e-01 7.87331581e-01
5.46282589e-01 -4.33776915e-01 -4.86575276e-01 -5.49413681e-01
1.28134698e-01 -3.50882649e-01 1.16239655e+00 -9.04746175e-01
4.60898221e-01 -2.84030467e-01 3.62415224e-01 -3.87856543e-01
3.64798158e-01 -6.25819445e-01 -4.23137963e-01 1.99516475e-01
-7.13758290e-01 -6.50036812e-01 1.13223404e-01 1.13044524e+00
-1.74252033e-01 -3.46891135e-01 8.51650834e-01 -6.20995581e-01
-1.43046367e+00 2.86810696e-01 -2.07548574e-01 3.55429173e-01
9.45539057e-01 -2.66954005e-01 -5.30637741e-01 -4.63336140e-01
-9.21173811e-01 1.68953225e-01 7.14938700e-01 5.83054960e-01
7.26179957e-01 -8.95998895e-01 -5.58487773e-01 1.61624655e-01
5.59859097e-01 -3.45622510e-01 2.82578856e-01 5.85072517e-01
-2.43044466e-01 4.63712662e-01 -2.22862601e-01 -9.64924872e-01
-1.11799347e+00 7.37566590e-01 3.87381285e-01 2.90549695e-01
-7.37512112e-01 1.03452730e+00 7.17574954e-01 -3.83955687e-02
1.88569695e-01 -3.31956506e-01 -5.56481481e-02 2.28218392e-01
4.96324450e-01 -1.68395638e-01 -4.63325918e-01 -9.28987026e-01
-3.52783084e-01 9.68841612e-01 3.27868313e-02 -1.81715578e-01
1.24790752e+00 -4.23509598e-01 -3.00756425e-01 8.49808633e-01
1.28064048e+00 -5.10684669e-01 -1.48094857e+00 -5.05105436e-01
1.49883255e-01 -6.90204263e-01 3.07785630e-01 -6.86206996e-01
-7.18557775e-01 1.17906952e+00 5.48933208e-01 2.29949862e-01
8.17826867e-01 6.50655091e-01 1.31215408e-01 4.57254529e-01
4.06882405e-01 -9.27420616e-01 4.32661384e-01 4.87467527e-01
7.47553110e-01 -1.78486598e+00 4.19687890e-02 -1.84303701e-01
-6.03710711e-01 8.04355323e-01 2.59667099e-01 -3.19812298e-01
7.38076627e-01 -2.73703009e-01 -2.79633641e-01 -5.17065942e-01
-6.60342634e-01 -8.20941389e-01 3.74386400e-01 8.39859784e-01
-1.93224195e-02 9.24069434e-02 3.33224803e-01 3.48878890e-01
7.50326784e-03 -2.92594790e-01 5.52583456e-01 5.71916521e-01
-9.20470536e-01 -3.51701319e-01 -2.64774412e-01 3.43854070e-01
-2.19226718e-01 -3.41526568e-01 -6.94964170e-01 6.90621734e-01
2.60885786e-02 1.01900613e+00 3.89523983e-01 -2.18813688e-01
-8.34316313e-02 5.17616086e-02 5.39918423e-01 -7.97922611e-01
1.35228887e-01 -2.75319576e-01 -2.61801243e-01 -7.41374373e-01
-6.55027330e-01 -5.58898151e-01 -1.20171440e+00 8.10395628e-02
9.53877196e-02 -3.22599828e-01 7.79895723e-01 1.16752708e+00
2.24945426e-01 4.88037258e-01 3.66966516e-01 -7.42472827e-01
-3.32875043e-01 -7.39356518e-01 -7.56469309e-01 5.64265490e-01
6.29835069e-01 -6.20819330e-01 -5.51539242e-01 4.41605449e-01]
|
[9.926176071166992, 1.9866281747817993]
|
9fb33e9a-1897-4e76-a68a-e6a97dfe0049
|
efficient-subtyping-of-ovarian-cancer
|
2302.08867
| null |
https://arxiv.org/abs/2302.08867v2
|
https://arxiv.org/pdf/2302.08867v2.pdf
|
Efficient subtyping of ovarian cancer histopathology whole slide images using active sampling in multiple instance learning
|
Weakly-supervised classification of histopathology slides is a computationally intensive task, with a typical whole slide image (WSI) containing billions of pixels to process. We propose Discriminative Region Active Sampling for Multiple Instance Learning (DRAS-MIL), a computationally efficient slide classification method using attention scores to focus sampling on highly discriminative regions. We apply this to the diagnosis of ovarian cancer histological subtypes, which is an essential part of the patient care pathway as different subtypes have different genetic and molecular profiles, treatment options, and patient outcomes. We use a dataset of 714 WSIs acquired from 147 epithelial ovarian cancer patients at Leeds Teaching Hospitals NHS Trust to distinguish the most common subtype, high-grade serous carcinoma, from the other four subtypes (low-grade serous, endometrioid, clear cell, and mucinous carcinomas) combined. We demonstrate that DRAS-MIL can achieve similar classification performance to exhaustive slide analysis, with a 3-fold cross-validated AUC of 0.8679 compared to 0.8781 with standard attention-based MIL classification. Our approach uses at most 18% as much memory as the standard approach, while taking 33% of the time when evaluating on a GPU and only 14% on a CPU alone. Reducing prediction time and memory requirements may benefit clinical deployment and the democratisation of AI, reducing the extent to which computational hardware limits end-user adoption.
|
['Nishant Ravikumar', 'Nicolas M. Orsi', 'Geoff Hall', 'Kieran Zucker', 'Katie Allen', 'Jack Breen']
|
2023-02-17
| null | null | null | null |
['whole-slide-images', 'multiple-instance-learning']
|
['computer-vision', 'methodology']
|
[ 4.94929165e-01 4.10413802e-01 -4.47730899e-01 -7.99497738e-02
-1.56494713e+00 -3.44313860e-01 1.75498221e-02 6.82898164e-01
-7.55023062e-01 5.66854417e-01 1.18645422e-01 -6.98984444e-01
-6.98320642e-02 -6.08085275e-01 -2.68161893e-01 -1.19333804e+00
-7.13694692e-02 7.62409687e-01 6.09717518e-02 2.83823937e-01
5.54165505e-02 6.60955608e-01 -9.87752974e-01 4.74634498e-01
7.15202749e-01 8.38192582e-01 -2.17919629e-02 1.15645576e+00
-8.60546455e-02 6.01224244e-01 -3.50453585e-01 -2.97444493e-01
-7.82119483e-02 -2.52292782e-01 -7.33155787e-01 -1.78697016e-02
6.89382851e-01 6.16415543e-03 1.95442438e-02 7.04170346e-01
6.24060810e-01 -2.78889686e-01 7.55872786e-01 -7.66133904e-01
2.17713550e-01 2.66182065e-01 -8.17568779e-01 3.69671196e-01
-1.04453512e-01 2.67454267e-01 1.05876946e+00 -7.28425264e-01
7.25187242e-01 6.44556105e-01 8.79014134e-01 4.67195898e-01
-1.26148593e+00 -5.06262422e-01 -1.99102700e-01 -4.39779609e-02
-1.50014520e+00 -4.42868263e-01 -3.85669246e-02 -3.89279783e-01
8.54593039e-01 7.29478121e-01 7.22967267e-01 5.27559817e-01
5.63045800e-01 9.03196037e-01 8.91175091e-01 -2.13966623e-01
2.63393372e-01 1.20082103e-01 2.44308531e-01 8.67617965e-01
4.83629704e-01 -5.84249139e-01 -2.22042397e-01 -6.36814117e-01
4.74008113e-01 2.67495126e-01 -1.97805539e-01 -2.05352202e-01
-1.06960416e+00 7.34987438e-01 2.63858706e-01 2.84601543e-02
-7.43111596e-02 3.58730392e-03 4.57019001e-01 8.07866380e-02
6.50527179e-01 4.64193821e-01 -1.43473402e-01 8.57155100e-02
-1.10917401e+00 -2.09799841e-01 5.88704109e-01 2.95942456e-01
4.25991982e-01 -5.08935273e-01 -1.29345700e-01 7.01318443e-01
2.11132690e-01 2.57742584e-01 7.29184151e-01 -4.37375486e-01
-6.57045022e-02 9.63213980e-01 -5.52920401e-02 -5.05243421e-01
-7.92987823e-01 -7.23007619e-01 -8.79792869e-01 9.26033780e-02
4.95792925e-01 -3.99892069e-02 -8.00608695e-01 1.10212553e+00
4.35122132e-01 2.78170079e-01 -4.57213521e-02 5.68091393e-01
6.61297619e-01 2.57618666e-01 2.82883495e-01 -1.98692843e-01
1.57486498e+00 -7.34835088e-01 -1.60022900e-01 -1.48474306e-01
1.29861641e+00 -4.31989670e-01 8.76229525e-01 1.84010476e-01
-9.59178150e-01 8.47489852e-03 -9.15395260e-01 -2.15946794e-01
-2.58906871e-01 1.66319177e-01 6.93688095e-01 7.44167864e-01
-1.02844179e+00 2.65238822e-01 -1.10943353e+00 -3.98676455e-01
9.89413261e-01 6.77002668e-01 -4.30117756e-01 -6.20445237e-02
-4.92796749e-01 7.11798549e-01 -2.32752234e-01 -2.69055665e-01
-5.37100971e-01 -1.36023796e+00 -7.30592132e-01 1.41815320e-01
8.96809474e-02 -4.14349973e-01 1.03955460e+00 -1.05699074e+00
-1.00351262e+00 1.16670930e+00 -4.62349147e-01 -4.46411729e-01
4.32348669e-01 4.04597938e-01 -7.29361922e-02 1.54262140e-01
-8.58794078e-02 6.25056267e-01 2.18675509e-01 -5.47072828e-01
-7.52337456e-01 -5.08835912e-01 -4.89592582e-01 4.34402257e-01
-2.82182842e-01 -2.96591669e-01 -5.91154575e-01 -1.76147267e-01
-1.29613683e-01 -1.24746001e+00 -6.79301977e-01 3.79819214e-01
-1.06247261e-01 -4.64745425e-02 6.67028725e-01 -5.89803517e-01
8.82695138e-01 -2.22308564e+00 -9.48044285e-03 4.30755466e-01
4.18807447e-01 2.37071887e-01 -4.10875957e-03 -2.08678782e-01
1.54261500e-01 2.44359747e-01 -1.13329589e-01 -2.80467421e-01
-2.59489328e-01 -9.75390300e-02 1.48383275e-01 8.12262774e-01
3.10279369e-01 1.08776677e+00 -1.00449002e+00 -8.31776857e-01
-8.18071812e-02 4.32521433e-01 -4.18145329e-01 -6.37794733e-02
3.21257934e-02 2.63019949e-01 -1.65315896e-01 9.37579334e-01
3.69782448e-01 -7.23151803e-01 6.25018835e-01 -4.16637473e-02
2.96057642e-01 6.79933056e-02 -6.69611990e-01 1.29877019e+00
-4.49089199e-01 9.54514444e-01 1.03391118e-01 -6.08019173e-01
4.66304988e-01 1.82386950e-01 5.94045937e-01 -4.30216312e-01
1.44063741e-01 3.57431203e-01 2.80596405e-01 -4.18668032e-01
2.94868141e-01 8.33355635e-02 -8.21664706e-02 3.57916534e-01
-4.01223838e-01 3.40412632e-02 2.34952107e-01 2.13524088e-01
1.64072382e+00 -3.23148817e-01 6.16448820e-01 -4.92487878e-01
2.81477123e-01 3.36948186e-01 5.77440560e-01 5.63365459e-01
-3.70485097e-01 5.25763035e-01 8.70796442e-01 -3.56390506e-01
-8.60767066e-01 -8.71550739e-01 -4.22780007e-01 9.12734866e-01
-5.78177050e-02 1.17906749e-01 -3.65717918e-01 -8.62009645e-01
2.08447382e-01 2.05451190e-01 -8.34958255e-01 1.30755514e-01
-4.57269400e-01 -1.27015495e+00 6.15357459e-01 4.49669152e-01
1.12610817e-01 -6.35124087e-01 -7.49406636e-01 1.11257687e-01
2.19223216e-01 -7.02899277e-01 -4.81884331e-01 3.08812886e-01
-9.26375031e-01 -1.27121985e+00 -1.04000783e+00 -7.97466457e-01
1.17533267e+00 -1.43446922e-02 1.15847731e+00 2.70376176e-01
-1.05630040e+00 1.62733406e-01 9.79525596e-02 -6.20942593e-01
-3.44327539e-01 2.31227025e-01 -3.87587398e-01 7.30878934e-02
5.23268282e-01 1.87235907e-01 -8.61525059e-01 2.23971736e-02
-5.59648931e-01 3.37059706e-01 1.03733325e+00 1.00288868e+00
9.37922776e-01 -3.95547062e-01 3.04338187e-01 -1.31851697e+00
4.68610302e-02 -6.59697771e-01 -3.56259197e-01 1.83008119e-01
-3.38386595e-01 -3.58592927e-01 3.66338342e-01 -4.32101429e-01
-8.05782199e-01 3.70059043e-01 8.98672342e-02 7.79124349e-03
1.14447236e-01 5.02020597e-01 4.22801644e-01 -4.03701067e-01
5.89778066e-01 6.72263801e-02 4.08934295e-01 1.45542219e-01
-4.77612734e-01 6.79317713e-01 1.55926630e-01 1.33662879e-01
2.41312414e-01 8.45311999e-01 2.18950108e-01 -1.03718531e+00
-6.39923513e-01 -9.70134974e-01 -1.16573714e-01 -8.85033384e-02
5.90317726e-01 -8.77961874e-01 -7.89417505e-01 3.13476264e-01
-4.80729342e-01 -7.04091191e-01 -2.26908296e-01 3.82677495e-01
-2.30843142e-01 1.41505957e-01 -9.41426218e-01 -3.47103238e-01
-7.41731822e-01 -1.24759924e+00 1.33253133e+00 2.84815937e-01
-6.43067718e-01 -1.11620116e+00 1.72868535e-01 5.26391923e-01
5.23461461e-01 3.40952694e-01 9.54787314e-01 -7.52589405e-01
-3.34733248e-01 -4.49354231e-01 -3.05714071e-01 -3.79660010e-01
1.62858088e-02 1.46360084e-01 -8.70750010e-01 -6.57887518e-01
-5.09381473e-01 -3.05561006e-01 8.19852591e-01 4.74151015e-01
1.19462800e+00 1.27671212e-02 -9.21860158e-01 6.42687619e-01
1.67008555e+00 1.34857252e-01 4.81249243e-01 1.26636595e-01
4.25580978e-01 4.04193074e-01 6.51072681e-01 2.44386181e-01
4.75428477e-02 2.96093792e-01 2.19953462e-01 -7.35486925e-01
-1.12755902e-01 3.61333311e-01 1.55072197e-01 3.27083588e-01
2.95342714e-01 -3.44460785e-01 -1.41360915e+00 7.76306927e-01
-1.30492640e+00 -6.02350056e-01 -1.13632195e-01 2.17868924e+00
9.33978498e-01 6.75387457e-02 -3.07626307e-01 7.86016136e-02
5.36966920e-01 -2.46456772e-01 -6.86721504e-01 -4.34403211e-01
1.10188045e-01 3.68131846e-01 7.67225742e-01 3.86842191e-01
-1.04447067e+00 1.94611475e-01 5.96952105e+00 1.08209121e+00
-1.36505735e+00 1.96775813e-02 1.50650477e+00 -5.07683635e-01
7.53969848e-02 -3.62664640e-01 -8.76546025e-01 4.07427102e-01
9.89787936e-01 2.81878025e-03 -4.89986867e-01 6.49080336e-01
8.68700966e-02 -5.10363162e-01 -1.10927773e+00 8.13553154e-01
1.52955055e-01 -1.78106344e+00 -1.97632462e-01 3.60602289e-01
7.66044378e-01 1.61012530e-01 9.91667658e-02 1.01621464e-01
1.36541858e-01 -1.16717911e+00 -1.93043262e-01 5.13128042e-01
1.29777968e+00 -5.71961761e-01 1.31289661e+00 1.69450104e-01
-7.94681251e-01 1.00036405e-01 -7.47483969e-02 4.61748660e-01
-5.15318036e-01 5.30006170e-01 -1.54282141e+00 -1.50876045e-02
5.20954370e-01 2.74464518e-01 -7.44130194e-01 1.08308649e+00
7.04387784e-01 9.13235366e-01 -4.24359083e-01 -3.36864471e-01
1.61185488e-01 4.15999740e-01 2.62698084e-01 1.48367143e+00
3.27457726e-01 1.49020195e-01 -1.22691073e-01 7.49013126e-02
3.90662476e-02 2.84658372e-01 -6.24081828e-02 5.06284423e-02
3.68302643e-01 1.53648412e+00 -1.20032322e+00 -4.00970250e-01
-4.55928832e-01 6.12409711e-01 1.79470524e-01 3.07784993e-02
-7.17192590e-01 -4.82950360e-01 3.61952364e-01 3.27043384e-01
8.47672224e-02 4.24644977e-01 -5.38432062e-01 -6.94261491e-01
-5.60793161e-01 -7.68684626e-01 6.80529535e-01 -3.28408808e-01
-1.00073814e+00 1.62881628e-01 -4.60108608e-01 -1.33652806e+00
1.06587611e-01 -5.86367607e-01 -5.86460769e-01 7.58959293e-01
-1.64120853e+00 -9.42484081e-01 -6.64082468e-01 -5.51963411e-02
4.32740897e-01 5.27862320e-03 1.03396523e+00 4.46446100e-03
-5.98487675e-01 1.05397677e+00 4.16517973e-01 2.55087435e-01
7.44803429e-01 -1.33227861e+00 -6.61094040e-02 1.77456602e-01
-4.56799388e-01 2.67021835e-01 2.08187923e-01 -4.03962314e-01
-1.57717633e+00 -1.39887106e+00 6.87335849e-01 -1.78441301e-01
4.74246323e-01 -1.04084961e-01 -7.68948019e-01 5.55073082e-01
-8.54569599e-02 3.99031818e-01 1.51279187e+00 -1.06885083e-01
9.87235680e-02 -1.69325903e-01 -1.38188171e+00 7.48846412e-01
3.59110504e-01 -1.35238722e-01 2.65466988e-01 3.57468903e-01
-2.78721806e-02 -7.35273719e-01 -1.11049712e+00 3.75310034e-01
6.43262267e-01 -6.23204887e-01 6.53252542e-01 -8.62043202e-02
1.90594569e-01 -7.36501366e-02 2.65022546e-01 -8.78893852e-01
-3.99886221e-01 -3.39068264e-01 2.21199259e-01 5.69149554e-01
7.71770716e-01 -7.20339060e-01 1.48515737e+00 5.48011124e-01
-2.28791125e-02 -1.38858402e+00 -9.53257084e-01 -7.06432834e-02
7.40758479e-02 5.96652441e-02 2.55399525e-01 7.90903270e-01
1.74636379e-01 -1.05922841e-01 5.94854653e-01 1.19422108e-01
5.97675681e-01 1.72660455e-01 5.54120600e-01 -1.03264987e+00
-3.19547445e-01 -6.26934648e-01 -8.68532240e-01 -4.14834648e-01
-1.82503268e-01 -9.79754329e-01 -1.96610063e-01 -1.31932259e+00
7.04222083e-01 -8.68254721e-01 -3.67042392e-01 6.67544246e-01
-4.24824297e-01 8.75872731e-01 -3.72985363e-01 2.64621526e-01
-7.35307395e-01 -3.04560959e-01 1.03813672e+00 -5.44742703e-01
-6.94146752e-02 -1.41304150e-01 -6.57463849e-01 7.25332797e-01
6.05020821e-01 -3.25411618e-01 -2.00945243e-01 -2.89323200e-02
2.70867527e-01 3.98142904e-01 1.50340214e-01 -9.65280890e-01
3.13036740e-01 -1.70135498e-01 7.96659291e-01 -5.00578880e-01
3.07436317e-01 -4.73466933e-01 2.30600625e-01 9.35790598e-01
-6.00758672e-01 -5.98816574e-01 3.95926744e-01 4.01441246e-01
-8.23990554e-02 7.56320544e-03 1.10788679e+00 -1.09730832e-01
-4.14752990e-01 4.12022412e-01 -6.66898310e-01 -1.26751110e-01
1.40545273e+00 -5.61641157e-01 -4.35300291e-01 8.46074969e-02
-7.71918714e-01 3.08212757e-01 5.33465922e-01 -2.30343133e-01
4.02628332e-01 -7.41978884e-01 -1.05855405e+00 1.18116416e-01
2.01646239e-01 3.34472716e-01 6.29015565e-01 1.16533220e+00
-1.00993264e+00 3.23168844e-01 1.56599432e-01 -7.78121531e-01
-1.76870608e+00 -1.42131090e-01 5.45175374e-01 -6.82927072e-01
-4.86050755e-01 1.15943205e+00 2.60210812e-01 2.75193783e-03
1.04110807e-01 -3.58581692e-02 -1.01524964e-01 2.24262953e-01
3.96755129e-01 4.00098532e-01 4.60030854e-01 -3.38555247e-01
-4.04062748e-01 2.92514980e-01 -5.58422506e-01 3.87141317e-01
9.80187118e-01 3.82188261e-01 -8.62964243e-02 3.98483276e-01
1.34431267e+00 1.40184015e-01 -1.00495040e+00 -1.56474933e-01
2.07729940e-03 -1.69264302e-01 4.46546853e-01 -6.14719272e-01
-9.19939518e-01 5.25600851e-01 9.35361147e-01 -1.29826605e-01
9.30240214e-01 -4.07954268e-02 6.01765931e-01 2.09767357e-01
1.32460862e-01 -6.72287047e-01 -2.08524659e-01 4.55455035e-02
3.73022199e-01 -1.30523539e+00 5.40335715e-01 -3.84666413e-01
-5.65244079e-01 1.02529848e+00 5.14338374e-01 -2.47843444e-01
3.04806173e-01 7.55514383e-01 2.11797282e-01 -3.63287516e-02
-1.22154212e+00 7.07943663e-02 -1.65997911e-02 1.52145773e-01
5.85521400e-01 2.53866822e-01 -4.31899950e-02 2.72367269e-01
1.47243291e-01 -1.40949920e-01 5.42283416e-01 9.40733910e-01
-3.29842538e-01 -5.60526848e-01 -1.72222689e-01 1.29350913e+00
-7.96131432e-01 -1.04799114e-01 -1.89967260e-01 8.56530070e-01
-7.74049833e-02 3.67831528e-01 8.06559920e-01 2.09128931e-01
-2.18354553e-01 -4.41720672e-02 2.92923957e-01 -7.37943172e-01
-7.48648107e-01 1.30208731e-01 4.58211415e-02 -1.86993927e-01
-1.20813176e-01 -7.87424028e-01 -1.30216300e+00 -1.20001942e-01
-3.36856842e-01 1.57112017e-01 5.29343724e-01 5.01456022e-01
5.31296432e-01 4.68724608e-01 4.82248574e-01 -6.37342632e-01
-2.66256839e-01 -7.86407471e-01 -5.26136696e-01 7.41720572e-02
5.85487545e-01 -1.92247659e-01 -5.43816924e-01 -8.94038305e-02]
|
[15.076101303100586, -3.033250570297241]
|
17d69e41-8df4-4d53-a36d-3d34e26e8107
|
changesim-towards-end-to-end-online-scene
|
2103.05368
| null |
https://arxiv.org/abs/2103.05368v2
|
https://arxiv.org/pdf/2103.05368v2.pdf
|
ChangeSim: Towards End-to-End Online Scene Change Detection in Industrial Indoor Environments
|
We present a challenging dataset, ChangeSim, aimed at online scene change detection (SCD) and more. The data is collected in photo-realistic simulation environments with the presence of environmental non-targeted variations, such as air turbidity and light condition changes, as well as targeted object changes in industrial indoor environments. By collecting data in simulations, multi-modal sensor data and precise ground truth labels are obtainable such as the RGB image, depth image, semantic segmentation, change segmentation, camera poses, and 3D reconstructions. While the previous online SCD datasets evaluate models given well-aligned image pairs, ChangeSim also provides raw unpaired sequences that present an opportunity to develop an online SCD model in an end-to-end manner, considering both pairing and detection. Experiments show that even the latest pair-based SCD models suffer from the bottleneck of the pairing process, and it gets worse when the environment contains the non-targeted variations. Our dataset is available at http://sammica.github.io/ChangeSim/.
|
['Jong-Hwan Kim', 'Ue-Hwan Kim', 'Sun-Kyung Lee', 'Sahng-Min Yoo', 'Jae-Hyuk Jang', 'Jin-Man Park']
|
2021-03-09
| null | null | null | null |
['scene-change-detection']
|
['computer-vision']
|
[ 4.70325977e-01 -5.09210289e-01 4.18176740e-01 -5.08058488e-01
-7.12035239e-01 -8.63798320e-01 4.75212306e-01 8.68946239e-02
-3.32859546e-01 4.99887347e-01 -1.93919405e-01 1.06679834e-01
1.42048791e-01 -6.94175124e-01 -1.04109573e+00 -8.52359772e-01
-1.87353902e-02 5.08205652e-01 4.99003321e-01 -1.31345123e-01
1.94048360e-02 4.27904278e-01 -1.89702404e+00 -4.23481055e-02
8.35847914e-01 1.02202523e+00 6.78447247e-01 9.55606282e-01
5.57791702e-02 3.18536341e-01 -2.80958503e-01 -1.66923832e-02
6.12991393e-01 -1.99658826e-01 -3.65672141e-01 3.45954597e-01
7.81617880e-01 -3.98150861e-01 -2.23760158e-01 1.38287103e+00
7.67625928e-01 5.11613041e-02 -3.90095934e-02 -1.35391104e+00
3.67045403e-03 1.92512333e-01 -4.54105288e-01 -6.64504096e-02
5.60554147e-01 6.38125122e-01 4.61222261e-01 -7.37109005e-01
8.43526602e-01 1.17963660e+00 5.94948053e-01 1.88611045e-01
-8.29259396e-01 -5.16593754e-01 2.61164069e-01 5.31863630e-01
-1.20157063e+00 -2.78815210e-01 7.92398095e-01 -3.72869253e-01
5.53398490e-01 3.96606803e-01 8.92547965e-01 1.28941464e+00
-6.83411881e-02 5.81809223e-01 1.26130664e+00 -8.73284936e-02
4.92131501e-01 -2.09950194e-01 -8.79815444e-02 3.84045184e-01
2.25769192e-01 3.38659346e-01 -5.39806247e-01 1.63211554e-01
5.45346260e-01 2.58360118e-01 -4.68230128e-01 -5.35304785e-01
-1.52164066e+00 1.42314220e-02 6.57936871e-01 -9.57734659e-02
-4.07371640e-01 8.38367343e-02 1.33091167e-01 3.10574830e-01
1.57257095e-01 3.60077247e-02 -7.81113863e-01 -3.42229217e-01
-6.08748317e-01 1.88240379e-01 5.14976442e-01 1.17370975e+00
1.19339037e+00 -2.20295012e-01 1.07432745e-01 4.70454603e-01
1.51493371e-01 1.07646060e+00 2.18033791e-01 -1.07007337e+00
6.21537328e-01 4.75542575e-01 3.72859746e-01 -1.00135255e+00
-5.45009613e-01 -2.95407891e-01 -7.92207837e-01 3.82634043e-03
3.75403881e-01 -2.80809794e-02 -1.06166756e+00 1.45514345e+00
8.85465145e-01 5.41138470e-01 -9.92487296e-02 1.10400116e+00
7.41164505e-01 4.95168269e-01 -4.39928770e-01 -2.02982709e-01
9.44538236e-01 -9.12440717e-01 -7.67544627e-01 -3.71146709e-01
3.30602109e-01 -7.01773465e-01 1.14190853e+00 4.42837268e-01
-6.67415500e-01 -5.64968944e-01 -7.94594347e-01 1.53129429e-01
-4.72988844e-01 -1.93746909e-01 2.82544464e-01 3.60572696e-01
-1.06222367e+00 2.85987675e-01 -9.66703057e-01 -6.64481580e-01
2.89320111e-01 -5.90571985e-02 -3.50300401e-01 -6.88876212e-01
-8.61835361e-01 5.28881192e-01 1.59981579e-01 4.53871906e-01
-1.12308896e+00 -6.78380787e-01 -8.49725604e-01 -6.28747761e-01
6.67455077e-01 -6.73509300e-01 1.10787666e+00 -8.23881388e-01
-1.26558995e+00 7.26731420e-01 -1.71378598e-01 -1.62832633e-01
1.06371546e+00 -2.12628081e-01 -2.05494925e-01 2.70308601e-03
7.42786527e-02 4.79097515e-01 6.20733082e-01 -1.60956216e+00
-6.65760696e-01 -6.92720830e-01 1.05870634e-01 3.23875904e-01
2.48041585e-01 -2.42434293e-01 -8.84514213e-01 -6.61419705e-02
3.20494264e-01 -1.20195699e+00 -4.10716981e-01 4.11515921e-01
-6.09701455e-01 5.30115902e-01 8.94324243e-01 -7.71647036e-01
5.98978817e-01 -2.13280225e+00 -6.68037757e-02 7.42680281e-02
-1.94004253e-01 5.79356775e-02 -1.87429041e-01 3.87830079e-01
1.32761806e-01 -1.63167357e-01 -6.29084527e-01 -5.36588907e-01
3.06443553e-02 5.19082069e-01 8.79518315e-02 6.83867991e-01
-1.43851280e-01 5.87188661e-01 -1.14553010e+00 -2.30512142e-01
6.80717826e-01 2.96585441e-01 -3.62440199e-01 1.69025555e-01
-3.22720975e-01 9.28953767e-01 -1.87172443e-01 1.00228798e+00
1.29094148e+00 2.08788104e-02 -3.89894694e-02 -2.65180588e-01
-2.28609696e-01 7.87528753e-02 -1.71820176e+00 1.94834113e+00
-4.12037343e-01 5.73226273e-01 3.25439095e-01 -4.70810801e-01
5.91293395e-01 -9.92219448e-02 5.67825019e-01 -1.01841557e+00
8.79802406e-02 5.13504073e-02 -4.38698620e-01 -6.24681234e-01
6.22825503e-01 4.02196646e-01 -4.14923131e-02 2.71701696e-03
-4.98765200e-01 -5.03705919e-01 2.03508914e-01 6.07942194e-02
1.10960972e+00 3.36276323e-01 1.27346858e-01 1.86667666e-01
2.54041612e-01 2.64579833e-01 8.30778360e-01 7.42643237e-01
-3.10147583e-01 1.04324532e+00 -1.58897400e-01 -3.37894052e-01
-9.18443441e-01 -1.25826597e+00 -9.55148935e-02 4.36450928e-01
8.19094121e-01 -4.21857141e-04 -5.49531817e-01 -2.11069465e-01
2.68395208e-02 5.45782328e-01 -4.02262866e-01 1.72163278e-01
-3.39246243e-01 -9.35950875e-01 5.53994514e-02 1.51949272e-01
8.82612169e-01 -7.25099087e-01 -8.87290776e-01 2.13489130e-01
-4.82534587e-01 -1.56906557e+00 -2.02137887e-01 1.78220034e-01
-6.72591090e-01 -1.41431952e+00 -2.13949800e-01 -3.93445313e-01
6.01011157e-01 4.62741971e-01 1.05555487e+00 -2.16499582e-01
-7.68277764e-01 5.18641412e-01 -4.29423571e-01 -4.25109416e-01
-2.45017901e-01 -5.20021558e-01 -8.88287425e-02 -2.60451646e-03
-3.15412253e-01 -4.97818708e-01 -8.94018710e-01 5.17993927e-01
-9.97178972e-01 1.47431016e-01 8.43572095e-02 4.80239987e-01
8.62300873e-01 7.42043480e-02 -3.05300038e-02 -6.79381132e-01
-3.29737991e-01 -3.45484108e-01 -7.72155166e-01 1.12902753e-01
-3.50319982e-01 -5.76828420e-01 3.71044546e-01 -1.30680472e-01
-1.09316552e+00 3.80964279e-01 -1.58059616e-02 -4.53149021e-01
-5.01909018e-01 1.25992656e-01 -5.25949895e-01 5.91035150e-02
5.85983872e-01 3.35019797e-01 -2.39675820e-01 -5.27549148e-01
3.31086725e-01 5.73134780e-01 8.65463495e-01 -3.09461653e-01
9.91494358e-01 8.89125526e-01 -2.50795007e-01 -7.83042789e-01
-5.13225853e-01 -7.70423293e-01 -7.45670021e-01 -6.12333298e-01
6.02197886e-01 -1.33353245e+00 -3.04303080e-01 1.16962397e+00
-1.04556191e+00 -4.90285397e-01 -3.60191196e-01 2.76544094e-01
-5.43230474e-01 4.25224900e-01 -4.32210833e-01 -7.50553310e-01
-1.17484527e-02 -1.09153378e+00 1.33682323e+00 2.18277469e-01
3.57026309e-01 -6.87155604e-01 -7.43678734e-02 5.46386778e-01
1.94344148e-01 7.86200583e-01 4.09159720e-01 5.99744059e-02
-1.10356283e+00 -3.46635580e-02 1.99147779e-02 3.32052648e-01
4.27054495e-01 1.84687167e-01 -1.10096669e+00 -3.15023363e-01
-1.93444729e-01 3.19959745e-02 5.81032574e-01 2.01282158e-01
1.06190073e+00 -7.70853832e-02 -1.98593900e-01 8.41198921e-01
1.71978700e+00 2.33871460e-01 6.77899718e-01 3.68913233e-01
1.03355014e+00 3.71773094e-01 1.10544133e+00 8.26324344e-01
8.84319782e-01 7.87509620e-01 1.13228989e+00 -2.53689617e-01
-2.48152167e-01 5.39056445e-03 3.97855371e-01 7.70099699e-01
2.57132947e-01 -4.15762573e-01 -1.06619000e+00 7.47183919e-01
-1.82874465e+00 -6.92707717e-01 -6.30646646e-01 2.33973241e+00
7.09567785e-01 -2.98495919e-01 -1.91754937e-01 5.85171059e-02
8.57921243e-01 1.47890285e-01 -1.02298427e+00 1.76892713e-01
-5.23762226e-01 -1.17022902e-01 7.82396555e-01 3.35695922e-01
-1.00972354e+00 7.69389331e-01 5.20808363e+00 3.38426650e-01
-1.19709373e+00 1.92765489e-01 3.12891185e-01 -1.93822905e-01
-2.52688468e-01 -1.13891453e-01 -4.79178131e-01 7.28987396e-01
5.99534690e-01 2.74697006e-01 5.79431057e-01 5.08111060e-01
5.53402126e-01 -6.04647100e-01 -8.47613990e-01 1.34531081e+00
-2.90721040e-02 -9.80759978e-01 -3.48303109e-01 -2.08097428e-01
1.06453669e+00 6.33500159e-01 -6.76358417e-02 -2.14329258e-01
5.13739467e-01 -3.44199061e-01 9.04470205e-01 6.71664298e-01
5.47049105e-01 -3.70810628e-01 7.12368011e-01 4.20870334e-01
-1.07445908e+00 -1.12220302e-01 -1.37007967e-01 -6.94973767e-02
3.42102349e-01 1.05262971e+00 -8.80124927e-01 9.97848630e-01
1.03690171e+00 1.07815707e+00 -8.59133482e-01 1.35237157e+00
-2.52322614e-01 3.63550127e-01 -6.59219086e-01 3.20029676e-01
-2.49369461e-02 -5.15312135e-01 7.05049515e-01 9.34766293e-01
5.96591651e-01 -1.01423740e-01 3.79386693e-01 5.57792425e-01
3.02054912e-01 -3.48818660e-01 -5.48865795e-01 3.83190900e-01
6.20652676e-01 1.20407379e+00 -5.75829923e-01 -1.93451747e-01
-7.08557814e-02 1.29035711e+00 -2.04856277e-01 4.92254257e-01
-9.85959411e-01 9.38065201e-02 1.05805981e+00 1.88165456e-02
3.38857830e-01 -4.59825933e-01 -2.10049823e-01 -1.21982408e+00
4.27516729e-01 -8.10082018e-01 1.63280725e-01 -1.13910949e+00
-1.02700889e+00 1.54897988e-01 -1.11257583e-01 -1.44039118e+00
-1.35224566e-01 -3.53300571e-01 -3.60284746e-01 4.45123106e-01
-1.76673460e+00 -1.04222155e+00 -1.13688910e+00 6.74222827e-01
4.38787192e-01 6.51565492e-01 4.23831165e-01 6.08926713e-01
-8.03033173e-01 7.52455816e-02 4.67255741e-01 -2.19533950e-01
7.26660013e-01 -1.15259135e+00 4.08161283e-01 1.18880439e+00
-6.66443482e-02 -1.15263648e-01 9.60989654e-01 -5.11212945e-01
-1.60433912e+00 -1.58440447e+00 2.43455335e-01 -3.92154485e-01
3.65933359e-01 -4.49517936e-01 -5.75500369e-01 4.69178528e-01
-2.28224650e-01 3.97718996e-01 -6.01065671e-03 -5.70863843e-01
5.79247735e-02 -3.89235109e-01 -1.35788417e+00 4.11606222e-01
1.68920684e+00 -4.01036680e-01 1.84585273e-01 6.21511936e-01
8.06135118e-01 -9.91165698e-01 -6.43257976e-01 5.61806977e-01
1.56873256e-01 -1.25413811e+00 9.88964021e-01 1.30777806e-01
1.05637334e-01 -9.06259775e-01 -5.80705404e-01 -1.36971653e+00
2.69115776e-01 -4.10060048e-01 3.35924655e-01 1.27436471e+00
1.32224321e-01 -6.64166272e-01 3.68148208e-01 3.42196733e-01
-2.98075527e-01 -2.07849935e-01 -1.09739721e+00 -9.48080420e-01
-5.50184309e-01 -7.18629777e-01 9.29215133e-01 9.72678721e-01
-8.24243844e-01 -2.58204341e-01 -2.62199342e-01 7.99510598e-01
7.98621356e-01 3.73996139e-01 1.15556765e+00 -9.63165283e-01
-7.76851028e-02 1.24611259e-01 -6.19468868e-01 -8.50660563e-01
-2.65060544e-01 -4.75976139e-01 4.95906621e-01 -1.80480182e+00
9.90840793e-02 -6.14757895e-01 -7.83134922e-02 3.73003364e-01
-1.25248432e-01 2.09608272e-01 1.72039822e-01 2.41087917e-02
-7.57201850e-01 6.02452457e-01 1.26342714e+00 -2.20267206e-01
-2.99869496e-02 -1.92008346e-01 -1.32801533e-01 5.60354710e-01
6.86085820e-01 -4.15859312e-01 -3.18842202e-01 -6.36272371e-01
2.02662200e-01 5.35328388e-02 7.54136324e-01 -1.36206245e+00
2.12355331e-01 -2.78769732e-01 1.45999238e-01 -9.27248716e-01
5.07533908e-01 -1.09304333e+00 6.61276877e-01 5.53148031e-01
2.19574794e-01 1.12900987e-01 8.73000771e-02 7.13517249e-01
-1.27877414e-01 2.66540140e-01 7.91455925e-01 -2.63381928e-01
-1.30353022e+00 5.47804415e-01 -9.86518618e-03 -1.19560165e-02
1.17687130e+00 -4.85274315e-01 -4.69828069e-01 -3.35615814e-01
-5.72252929e-01 4.93559510e-01 1.00394571e+00 5.68992794e-01
4.59645987e-01 -9.70483541e-01 -5.95775127e-01 2.88942009e-01
5.43984294e-01 7.93763816e-01 6.78679049e-01 8.79118264e-01
-7.12158084e-01 -3.54820788e-01 -1.09138042e-01 -1.12139153e+00
-1.26129997e+00 2.55502969e-01 4.59015042e-01 2.01353654e-01
-6.77450359e-01 6.95113242e-01 1.67724431e-01 -9.46377039e-01
9.08118635e-02 -7.24578977e-01 4.39841419e-01 1.85909867e-02
3.99793714e-01 4.38611180e-01 3.75296772e-01 -5.42021692e-01
-5.01478791e-01 6.83084607e-01 5.59324443e-01 1.37924552e-01
1.29285896e+00 -8.00306380e-01 -2.89270002e-02 7.16359317e-01
1.03247643e+00 -3.01611334e-01 -1.67865694e+00 -3.59344870e-01
-3.83944958e-01 -8.35141540e-01 -7.23639727e-02 -1.04017663e+00
-1.11556756e+00 6.26843750e-01 1.21879983e+00 -4.14095074e-01
1.25103152e+00 -1.53423637e-01 8.11862290e-01 3.66678059e-01
1.02235889e+00 -1.26971066e+00 -5.04253432e-02 6.93729520e-01
7.41842747e-01 -1.39251733e+00 -1.66829184e-01 -6.06579840e-01
-5.13546228e-01 5.73987365e-01 5.69937766e-01 2.70785123e-01
4.36145276e-01 3.98187965e-01 4.07898754e-01 -4.91583627e-03
-5.79674482e-01 -5.76637983e-01 -4.90736067e-01 7.87144423e-01
-5.16667843e-01 2.94641733e-01 4.08428580e-01 -6.12175129e-02
-2.53602237e-01 -1.77173436e-01 7.39354908e-01 1.04421413e+00
-2.47438207e-01 -7.20385551e-01 -6.47788107e-01 1.36896834e-01
3.09270173e-01 6.39999881e-02 -2.68916458e-01 6.80677593e-01
4.18194264e-01 1.06742942e+00 2.40589276e-01 -3.60014677e-01
5.79332054e-01 -3.31602842e-01 4.22801703e-01 -4.06578243e-01
-3.20459992e-01 -3.16992521e-01 1.19002081e-01 -1.05642903e+00
-4.88789856e-01 -1.10138559e+00 -1.38162506e+00 -2.17247590e-01
-1.68670252e-01 -4.47377175e-01 1.15907180e+00 7.67644882e-01
4.39557850e-01 6.43396139e-01 1.00345945e+00 -1.20202160e+00
-1.53588951e-01 -8.12859058e-01 -5.74567378e-01 6.41275704e-01
6.82143927e-01 -5.39696038e-01 -4.96501207e-01 3.03357661e-01]
|
[8.642467498779297, -2.201843023300171]
|
36fada08-d33d-4d05-b364-f6aed0832acc
|
evaluation-of-the-spatio-temporal-features
|
1904.01748
| null |
http://arxiv.org/abs/1904.01748v1
|
http://arxiv.org/pdf/1904.01748v1.pdf
|
Evaluation of the Spatio-Temporal features and GAN for Micro-expression Recognition System
|
Owing to the development and advancement of artificial intelligence, numerous
works were established in the human facial expression recognition system.
Meanwhile, the detection and classification of micro-expressions are attracting
attentions from various research communities in the recent few years. In this
paper, we first review the processes of a conventional optical-flow-based
recognition system, which comprised of facial landmarks annotations, optical
flow guided images computation, features extraction and emotion class
categorization. Secondly, a few approaches have been proposed to improve the
feature extraction part, such as exploiting GAN to generate more image samples.
Particularly, several variations of optical flow are computed in order to
generate optimal images to lead to high recognition accuracy. Next, GAN, a
combination of Generator and Discriminator, is utilized to generate new "fake"
images to increase the sample size. Thirdly, a modified state-of-the-art
Convolutional neural networks is proposed. To verify the effectiveness of the
the proposed method, the results are evaluated on spontaneous micro-expression
databases, namely SMIC, CASME II and SAMM. Both the F1-score and accuracy
performance metrics are reported in this paper.
|
['Kun-Hong Liu', 'Ran-Ke Lyu', 'Han-Zhe Zhang', 'Hao-Xuan Xua', 'Shu-Meng Lic', 'Sze-Teng Liong', 'Y. S. Gan', 'Danna Zheng']
|
2019-04-03
| null | null | null | null |
['micro-expression-recognition']
|
['computer-vision']
|
[ 2.51685202e-01 -2.28801608e-01 -1.86570063e-02 -4.98483211e-01
-1.43920541e-01 1.18395535e-03 5.39986968e-01 -5.06728113e-01
-4.38279003e-01 7.32903719e-01 -2.17137709e-02 4.35031831e-01
2.76671767e-01 -6.79521441e-01 -1.04938708e-01 -9.08540428e-01
2.24389195e-01 -3.74258816e-01 -3.47584516e-01 -2.76193976e-01
3.47525984e-01 6.93374038e-01 -1.68301427e+00 8.45058560e-02
7.57393479e-01 1.42459261e+00 -4.05231476e-01 2.32213035e-01
-3.56465906e-01 1.02886808e+00 -6.30639851e-01 -7.51314938e-01
5.95329180e-02 -8.93089294e-01 -4.98238236e-01 3.61643761e-01
1.28162980e-01 -2.73928642e-01 -2.27349788e-01 1.20139015e+00
6.95368469e-01 1.20572925e-01 5.63669980e-01 -1.57985616e+00
-5.69003880e-01 -2.20737178e-02 -7.03971863e-01 4.94745597e-02
3.61288846e-01 1.75039008e-01 4.40620959e-01 -9.98333156e-01
5.25582910e-01 1.25570703e+00 3.38745981e-01 6.90471590e-01
-7.19529867e-01 -1.10597670e+00 -2.26239592e-01 4.51213092e-01
-1.45958722e+00 -6.93156600e-01 1.42074430e+00 -5.07796645e-01
4.23562706e-01 -1.48137351e-02 8.73240948e-01 1.13552880e+00
-3.32605876e-02 8.31925213e-01 1.33498049e+00 -5.40384531e-01
5.78517653e-02 2.98871696e-01 -1.93335727e-01 9.58617747e-01
-1.50143638e-01 1.31722540e-01 -6.13282025e-01 1.51641145e-01
7.70778477e-01 -1.54013410e-01 -3.54826272e-01 2.12330446e-02
-6.23314857e-01 6.66654110e-01 3.86934251e-01 5.63035786e-01
-5.50950110e-01 -9.40261111e-02 6.82899594e-01 7.21620247e-02
6.04781687e-01 1.47752658e-01 9.29666683e-02 -4.20656651e-01
-9.45127785e-01 -8.16070065e-02 4.89452809e-01 6.06233299e-01
8.17452788e-01 4.42344457e-01 -1.68732882e-01 8.65586281e-01
1.92882404e-01 3.39580089e-01 7.13557839e-01 -7.55125880e-01
2.99167722e-01 7.79733956e-01 5.38656395e-03 -1.70991778e+00
-3.92551750e-01 -2.16791347e-01 -1.21306467e+00 2.34451130e-01
1.49685010e-01 -3.31076235e-01 -6.38965666e-01 1.61436725e+00
3.14463556e-01 5.26447713e-01 1.96782693e-01 1.14977276e+00
7.97384620e-01 5.98276854e-01 9.52337980e-02 -4.11973298e-01
1.15456688e+00 -9.02552545e-01 -1.06753349e+00 2.63413578e-01
5.55626988e-01 -7.43553698e-01 8.28501642e-01 3.70640010e-01
-8.19148540e-01 -8.06441784e-01 -9.29449379e-01 1.79192260e-01
-3.51462930e-01 5.26350737e-01 7.84608722e-01 8.80943060e-01
-6.89591110e-01 3.13668102e-01 -6.54830515e-01 -1.29996836e-01
1.01333880e+00 2.12705374e-01 -5.11070728e-01 2.02724069e-01
-1.18052101e+00 6.52906120e-01 2.08920538e-01 7.02347577e-01
-7.20216572e-01 -2.60783911e-01 -7.82851994e-01 -6.88311756e-02
-8.94806683e-02 -2.96391457e-01 7.86125362e-01 -1.78385973e+00
-1.96357214e+00 9.67552066e-01 -2.26999223e-01 -6.11871593e-02
5.32220423e-01 -1.17566679e-02 -6.48148775e-01 3.55717838e-01
-2.10080788e-01 6.57982588e-01 8.66501272e-01 -8.58857810e-01
-4.38369870e-01 -5.47604501e-01 -1.11185253e-01 1.13372251e-01
-4.75546986e-01 2.82947719e-01 -1.54448599e-01 -5.71812987e-01
-1.75894424e-01 -7.62898684e-01 7.82990977e-02 -6.33720458e-02
-3.91380429e-01 -1.46220967e-01 8.69406044e-01 -5.99632502e-01
1.09988165e+00 -2.11310005e+00 4.37784828e-02 2.61795908e-01
1.96205843e-02 5.71197450e-01 -1.39540151e-01 -2.23213751e-02
-1.22147761e-01 6.64609298e-02 -1.62600964e-01 -3.20260286e-01
-2.40051895e-01 7.87153933e-03 -6.45912439e-02 5.63905597e-01
5.41871071e-01 8.01540017e-01 -6.88490629e-01 -6.23169720e-01
3.57001871e-01 6.07774556e-01 -3.73240560e-01 3.71173292e-01
2.36231357e-01 8.89683127e-01 -5.57735384e-01 8.06321859e-01
9.55853581e-01 1.07797004e-01 -1.79108232e-01 -4.08298880e-01
-2.91865095e-02 -4.20907974e-01 -9.89791334e-01 1.51885259e+00
-5.65468073e-01 6.37543738e-01 7.25272074e-02 -1.19248700e+00
1.35018730e+00 3.75988126e-01 6.74385309e-01 -8.49633396e-01
6.96780980e-01 3.35097522e-01 -4.84178439e-02 -8.68715763e-01
3.29341054e-01 -1.15260512e-01 3.17362636e-01 1.04510829e-01
1.30744606e-01 1.98801056e-01 1.89834461e-01 -3.31764191e-01
5.20844102e-01 2.63339341e-01 1.66733444e-01 1.38858989e-01
1.22703159e+00 -2.96965271e-01 7.53302991e-01 3.29821445e-02
-4.31306392e-01 2.44237915e-01 4.98699933e-01 -6.61658347e-01
-7.05618918e-01 -4.42011505e-01 -3.43134329e-02 5.36901116e-01
3.34864914e-01 -8.75741392e-02 -1.01675153e+00 -6.49946451e-01
-3.96751910e-01 1.80824146e-01 -6.01879597e-01 -2.40801364e-01
-4.31948751e-01 -8.71475220e-01 8.14035356e-01 2.94885188e-01
1.32537770e+00 -1.49658418e+00 -7.38183796e-01 1.30593598e-01
-2.11728156e-01 -1.22479880e+00 -7.26014003e-02 -6.09690428e-01
-4.63657856e-01 -1.10403216e+00 -8.15098703e-01 -7.33446777e-01
8.69158089e-01 -1.95324644e-01 5.74555218e-01 2.26722077e-01
-5.03051877e-01 -4.26889025e-02 -5.15611708e-01 -3.26908976e-01
-2.30401859e-01 -6.21776506e-02 -1.55251190e-01 9.04425144e-01
5.19014537e-01 -5.11824965e-01 -6.35159969e-01 2.93798357e-01
-8.04087639e-01 1.26764283e-01 7.74010837e-01 1.02711546e+00
3.54728341e-01 -2.18984112e-01 5.23708165e-01 -7.03627348e-01
6.02383137e-01 -1.42657802e-01 -5.84253371e-01 1.11179918e-01
-3.73003781e-01 -3.05988908e-01 8.35308552e-01 -3.78656030e-01
-1.36825347e+00 9.95920748e-02 -3.80669832e-01 -4.66004997e-01
-3.77190024e-01 3.59304130e-01 -3.19903612e-01 -3.05130869e-01
4.72329795e-01 4.62094665e-01 4.14655775e-01 -1.05541646e-01
2.16440201e-01 1.02790368e+00 3.65108758e-01 -2.96457320e-01
4.45011944e-01 4.27686214e-01 1.61814958e-01 -9.79104578e-01
-6.32579803e-01 -1.78518996e-01 -5.38608789e-01 -5.70307553e-01
8.78855407e-01 -6.69273853e-01 -9.34493303e-01 1.11062968e+00
-1.34040594e+00 9.39922556e-02 1.75774753e-01 5.50894797e-01
-4.87122118e-01 1.22833923e-01 -5.97839713e-01 -9.48328376e-01
-5.65641880e-01 -1.35030544e+00 9.20667529e-01 7.06982017e-01
1.24703705e-01 -7.66582668e-01 -1.29281566e-01 2.49538019e-01
5.60345411e-01 5.74344039e-01 3.71716887e-01 -1.44140676e-01
-3.45204592e-01 -1.36865869e-01 -3.90121758e-01 6.20863855e-01
3.19859803e-01 9.33627412e-02 -9.94678020e-01 8.40666741e-02
-2.75439136e-02 -5.04886150e-01 5.05223393e-01 7.43673444e-02
1.34136093e+00 -4.82436746e-01 -2.07200035e-01 7.96045244e-01
1.13872349e+00 5.60180724e-01 1.00449812e+00 1.26506999e-01
6.46177232e-01 6.03090465e-01 6.73121691e-01 6.99050784e-01
1.95520204e-02 7.60966778e-01 3.49387020e-01 -2.65610039e-01
6.74996600e-02 -1.65028617e-01 2.70958126e-01 7.42512405e-01
-3.88583004e-01 1.73055995e-02 -3.68729889e-01 2.14795470e-01
-1.53045475e+00 -1.16580379e+00 2.31383488e-01 1.67948127e+00
6.36626661e-01 -2.19284460e-01 1.29848495e-02 2.39913478e-01
7.86668897e-01 2.92292774e-01 -4.56139177e-01 -3.94108951e-01
-1.52953401e-01 3.53238076e-01 -6.12706766e-02 1.80358246e-01
-1.07352686e+00 1.08518124e+00 4.93293953e+00 7.86069214e-01
-1.74028707e+00 2.20414833e-03 1.01805699e+00 2.80549884e-01
1.00677773e-01 -3.09042722e-01 -5.50546825e-01 6.91530108e-01
5.19458830e-01 7.32984245e-02 3.49758446e-01 8.05425286e-01
4.32515591e-01 7.14506358e-02 -4.24181461e-01 1.43571734e+00
3.09275776e-01 -1.19598579e+00 1.12379961e-01 -1.67096362e-01
6.20602131e-01 -5.99967957e-01 -4.18715626e-02 1.26856640e-01
-3.97257358e-01 -9.88658786e-01 3.11050504e-01 9.30324972e-01
8.03950906e-01 -9.67875183e-01 1.09934688e+00 1.11129008e-01
-9.28755522e-01 5.77032864e-02 -2.99971342e-01 -2.90235151e-02
1.40377641e-01 5.65686882e-01 -4.23988461e-01 6.84433937e-01
6.01560891e-01 8.11106443e-01 -4.44315314e-01 7.78046787e-01
-3.74309927e-01 6.16726398e-01 -1.47770301e-01 -3.59301776e-01
1.03545979e-01 -5.95232666e-01 2.13783175e-01 1.08216202e+00
3.75712335e-01 9.31191221e-02 -2.74893701e-01 1.02249825e+00
-1.36771590e-01 5.07123530e-01 -4.86668259e-01 -1.29831240e-01
2.28584781e-01 1.74717093e+00 -4.02838260e-01 -1.94433942e-01
-2.41460070e-01 1.09116209e+00 1.61120728e-01 3.15736264e-01
-9.10392344e-01 -6.74057782e-01 5.71077466e-01 -1.06453940e-01
-1.92060843e-01 -6.35377690e-02 1.59098387e-01 -1.18816125e+00
3.84766050e-02 -7.88963318e-01 -1.69894658e-02 -7.97038436e-01
-9.59359705e-01 7.82000005e-01 -3.54995459e-01 -1.25319827e+00
-3.41196150e-01 -6.32644415e-01 -6.81147277e-01 8.81094813e-01
-1.62371814e+00 -1.04342520e+00 -9.83587444e-01 6.52622581e-01
1.87694266e-01 -3.95269692e-01 8.19580078e-01 5.42457402e-01
-8.54477882e-01 8.04620445e-01 -3.26986760e-01 6.37325644e-01
4.88451779e-01 -4.77281690e-01 -2.72308886e-01 8.05881977e-01
-2.52073798e-02 3.33837599e-01 1.92049190e-01 -1.54765949e-01
-1.11101449e+00 -1.05503595e+00 6.29906595e-01 4.23324585e-01
3.11702996e-01 -1.64252460e-01 -6.22783124e-01 3.20922494e-01
9.22067314e-02 5.20746350e-01 5.00715137e-01 -6.03297591e-01
-1.62857827e-02 -5.70884466e-01 -1.29829407e+00 5.97043514e-01
8.57399523e-01 -3.11613172e-01 -9.73463058e-02 -1.07369170e-01
-2.34277407e-03 -3.37007821e-01 -7.88223386e-01 5.70728779e-01
7.48679936e-01 -1.24631274e+00 4.57179993e-01 -5.20181239e-01
5.93839169e-01 -3.31284910e-01 1.43674046e-01 -1.32337606e+00
9.93662849e-02 -6.62498713e-01 2.32268423e-01 1.44758475e+00
-4.94276360e-02 -7.38722980e-01 1.04479694e+00 2.42066085e-01
1.07931860e-01 -8.61452222e-01 -8.97907019e-01 -3.61517727e-01
-3.88030469e-01 -9.60189402e-02 6.27753675e-01 1.00605130e+00
-8.61896351e-02 2.96642751e-01 -5.06930709e-01 -3.66779923e-01
3.19467187e-01 1.19751915e-01 9.73904371e-01 -9.64856863e-01
1.25183120e-01 -5.47371149e-01 -9.54215825e-01 -8.61873388e-01
4.51591849e-01 -7.38643885e-01 -2.78797090e-01 -9.36821103e-01
-1.13664977e-01 -3.00240457e-01 -2.84069061e-01 3.71985644e-01
-1.70567427e-02 4.92842019e-01 3.37231010e-02 5.98120689e-02
-2.80784220e-01 1.01453733e+00 1.49062860e+00 -1.38395220e-01
-1.57215446e-01 -7.87706450e-02 -5.36936522e-01 6.71790957e-01
9.07019079e-01 -7.70919174e-02 -3.07088792e-01 -2.95460187e-02
-2.19216734e-01 1.70387834e-01 4.35610592e-01 -1.09984601e+00
-5.77473640e-02 -8.04792158e-03 6.72113359e-01 -2.50334591e-01
4.87768143e-01 -6.55338883e-01 1.52985649e-02 3.94622684e-01
-3.00266474e-01 8.41951221e-02 7.75734428e-03 1.44064859e-01
-7.64880002e-01 -1.24924153e-01 1.06622839e+00 -3.91012691e-02
-9.20836091e-01 5.21364510e-01 -1.62776455e-01 -1.92132443e-01
1.20345271e+00 -2.81615555e-01 -1.31446317e-01 -4.41039294e-01
-4.44177270e-01 -1.85054481e-01 4.25576717e-02 3.88342798e-01
8.32673848e-01 -1.48274338e+00 -6.46188557e-01 3.15147489e-01
2.01276675e-01 -2.55742699e-01 4.00705308e-01 1.07765567e+00
-7.08705962e-01 7.36208931e-02 -7.32858896e-01 -4.32593971e-01
-1.15582502e+00 1.88081801e-01 6.04725897e-01 -2.10373942e-02
-1.42590836e-01 5.95466197e-01 -5.53932600e-02 -1.54686093e-01
-1.12347551e-01 1.57090843e-01 -5.57458162e-01 1.29740730e-01
6.19701743e-01 2.96889454e-01 -1.34589702e-01 -1.09708369e+00
-3.11207086e-01 6.95923865e-01 1.58794031e-01 1.27504710e-02
1.07327378e+00 -1.06868565e-01 -3.06525111e-01 -6.53173327e-02
1.43338585e+00 -1.14554353e-01 -1.03397036e+00 4.99819368e-02
-1.99651986e-01 -5.81897676e-01 -5.60545921e-03 -5.96478760e-01
-1.58987999e+00 1.23482192e+00 8.47917676e-01 -4.93910350e-02
1.52897108e+00 -5.26730597e-01 7.11253166e-01 -8.05059448e-02
3.38410616e-01 -9.85016942e-01 2.19192475e-01 1.21325940e-01
8.03272784e-01 -1.10258389e+00 -3.48920196e-01 -4.15503860e-01
-6.73577905e-01 1.38081622e+00 9.01525259e-01 -7.59327710e-02
5.71029782e-01 1.84720799e-01 3.06112558e-01 -9.14498407e-04
-9.42316800e-02 -1.12728052e-01 1.81330591e-02 3.62644970e-01
4.93413866e-01 -1.77465037e-01 -5.89992523e-01 6.09517336e-01
-2.75237471e-01 5.82360923e-01 3.95526767e-01 5.89055777e-01
-9.70095098e-02 -8.27303648e-01 -1.79026484e-01 2.84232020e-01
-6.96100533e-01 2.51681894e-01 -3.34053755e-01 7.34996617e-01
2.75528550e-01 9.75791276e-01 1.14349633e-01 -6.13460302e-01
1.82909518e-01 7.61958165e-03 4.78089750e-01 -1.41318301e-02
-2.50354350e-01 -2.42915183e-01 -1.04561754e-01 -6.89443707e-01
-9.56041455e-01 -2.86866933e-01 -1.23479652e+00 -1.09543070e-01
-3.46651912e-01 2.72565007e-01 7.55278349e-01 9.31679785e-01
4.77764875e-01 2.36381203e-01 1.07158458e+00 -8.67810845e-01
-1.52969763e-01 -1.13638222e+00 -5.09930611e-01 8.19837332e-01
5.94754815e-02 -9.03944492e-01 -3.95633578e-01 9.68518332e-02]
|
[13.56795597076416, 1.7120131254196167]
|
a86a8dde-e704-452f-a861-b0c1d2ec6da5
|
uncertainty-aware-multi-view-co-training-for
|
2006.16806
| null |
https://arxiv.org/abs/2006.16806v1
|
https://arxiv.org/pdf/2006.16806v1.pdf
|
Uncertainty-aware multi-view co-training for semi-supervised medical image segmentation and domain adaptation
|
Although having achieved great success in medical image segmentation, deep learning-based approaches usually require large amounts of well-annotated data, which can be extremely expensive in the field of medical image analysis. Unlabeled data, on the other hand, is much easier to acquire. Semi-supervised learning and unsupervised domain adaptation both take the advantage of unlabeled data, and they are closely related to each other. In this paper, we propose uncertainty-aware multi-view co-training (UMCT), a unified framework that addresses these two tasks for volumetric medical image segmentation. Our framework is capable of efficiently utilizing unlabeled data for better performance. We firstly rotate and permute the 3D volumes into multiple views and train a 3D deep network on each view. We then apply co-training by enforcing multi-view consistency on unlabeled data, where an uncertainty estimation of each view is utilized to achieve accurate labeling. Experiments on the NIH pancreas segmentation dataset and a multi-organ segmentation dataset show state-of-the-art performance of the proposed framework on semi-supervised medical image segmentation. Under unsupervised domain adaptation settings, we validate the effectiveness of this work by adapting our multi-organ segmentation model to two pathological organs from the Medical Segmentation Decathlon Datasets. Additionally, we show that our UMCT-DA model can even effectively handle the challenging situation where labeled source data is inaccessible, demonstrating strong potentials for real-world applications.
|
['Zhuotun Zhu', 'Fengze Liu', 'Dong Yang', 'Zhiding Yu', 'Lequan Yu', 'Jinzheng Cai', 'Holger Roth', 'Daguang Xu', 'Alan Yuille', 'Yingda Xia']
|
2020-06-28
| null | null | null | null |
['semi-supervised-medical-image-segmentation', 'volumetric-medical-image-segmentation', 'pancreas-segmentation']
|
['computer-vision', 'medical', 'medical']
|
[ 7.00005144e-02 2.78807610e-01 -4.12267536e-01 -6.43140018e-01
-1.03722525e+00 -6.91426873e-01 1.04888581e-01 1.36855900e-01
-4.34453577e-01 6.83751166e-01 1.59880482e-02 -1.91889271e-01
8.49881470e-02 -7.08910048e-01 -8.03978264e-01 -8.21588218e-01
1.91828117e-01 1.02839625e+00 1.87178373e-01 2.98275471e-01
-3.16588849e-01 2.67795533e-01 -8.96716595e-01 6.42284900e-02
1.19871664e+00 8.46492052e-01 1.81503534e-01 3.27765018e-01
-1.51379570e-01 4.70130831e-01 -1.76118508e-01 -2.17909679e-01
3.53239357e-01 -3.78659040e-01 -9.95729983e-01 5.74230433e-01
2.10958302e-01 -4.85043615e-01 -4.42829058e-02 1.20723164e+00
5.23286700e-01 -1.07613601e-01 7.37015009e-01 -9.69706059e-01
-4.75198895e-01 5.49488902e-01 -6.19708776e-01 1.15464136e-01
-1.12570174e-01 -7.43530914e-02 6.19447410e-01 -5.56419730e-01
6.99091673e-01 7.73399234e-01 5.05980194e-01 5.86204171e-01
-1.29666960e+00 -4.74658281e-01 5.19186631e-02 -1.90834895e-01
-1.09431338e+00 -3.85922678e-02 7.94144750e-01 -6.39505744e-01
3.48900348e-01 -1.84464052e-01 5.89391530e-01 8.30540061e-01
2.76608497e-01 1.05805981e+00 1.18808019e+00 -1.79820701e-01
3.45235556e-01 2.72709616e-02 1.70561999e-01 7.57148325e-01
2.71738231e-01 -3.17798294e-02 5.89196756e-02 -9.13952813e-02
1.04440844e+00 2.57812530e-01 -4.09658998e-01 -1.03824508e+00
-1.51338184e+00 7.35963225e-01 5.74785769e-01 3.21622431e-01
-4.57628518e-01 -1.73607931e-01 5.52590013e-01 -2.67001651e-02
5.39395750e-01 1.38661429e-01 -4.28490072e-01 3.66378367e-01
-1.02214146e+00 -2.29200885e-01 6.45488977e-01 1.00544190e+00
4.75047886e-01 -2.32540146e-01 2.91215777e-02 8.34680736e-01
4.93953794e-01 3.71558517e-01 6.32292032e-01 -7.97669530e-01
2.52629071e-01 6.58325851e-01 -6.54303133e-02 -3.26569021e-01
-7.15679765e-01 -4.71383661e-01 -1.14928985e+00 -1.16441976e-02
8.11680913e-01 -1.25137538e-01 -1.42467773e+00 1.70334017e+00
5.87319076e-01 1.51780084e-01 2.10983008e-01 1.13256001e+00
7.97548175e-01 2.87763566e-01 1.53426588e-01 -3.28369260e-01
1.29593039e+00 -1.03600335e+00 -6.69592202e-01 -2.56677344e-02
6.97825789e-01 -4.11972344e-01 7.63333023e-01 3.70315075e-01
-8.97456765e-01 -2.85401911e-01 -9.11923945e-01 7.06175044e-02
3.84392962e-02 3.02885305e-02 5.62736034e-01 6.48947775e-01
-7.19684839e-01 3.08172554e-01 -1.40972376e+00 -1.71163648e-01
7.47521460e-01 5.05750895e-01 -5.69046557e-01 -3.45588624e-01
-9.64034140e-01 7.72108674e-01 6.18668735e-01 5.19636273e-02
-1.05140674e+00 -7.61658251e-01 -1.03191173e+00 -1.82854280e-01
5.72502077e-01 -7.51415610e-01 1.26419795e+00 -9.36830938e-01
-1.37023318e+00 1.12903082e+00 5.95242828e-02 -4.48425144e-01
6.97809756e-01 8.06569681e-02 -2.29427278e-01 5.34769714e-01
2.99029350e-01 7.39155293e-01 6.88136399e-01 -1.39841139e+00
-3.08296442e-01 -7.58958876e-01 -1.57173708e-01 2.13852048e-01
-9.88606066e-02 -4.87040818e-01 -6.63231075e-01 -6.05309546e-01
4.53858912e-01 -1.09950209e+00 -5.86985290e-01 -1.44357411e-02
-4.64947432e-01 8.17252770e-02 6.08717322e-01 -6.57686889e-01
7.97903478e-01 -1.97216475e+00 3.42431456e-01 2.83419967e-01
3.65039587e-01 1.14114076e-01 1.11474738e-01 -3.40241551e-01
-4.04097401e-02 1.89636983e-02 -6.82681918e-01 -2.28377923e-01
-2.32661024e-01 6.40648305e-01 3.01157504e-01 6.37178779e-01
1.37008261e-02 9.55919027e-01 -9.85933840e-01 -9.30792272e-01
3.27725410e-01 1.84676573e-01 -5.94549775e-01 2.15266630e-01
-3.55860859e-01 1.21075046e+00 -6.26554191e-01 6.70997024e-01
7.16946363e-01 -7.32484102e-01 4.11800891e-01 -2.24690080e-01
3.37736875e-01 -2.72267252e-01 -9.46361184e-01 2.20946074e+00
-4.10710067e-01 -1.42132968e-01 1.34845003e-01 -1.24823022e+00
5.37080467e-01 4.76256043e-01 9.81554329e-01 -4.91797566e-01
2.02762142e-01 3.49640548e-01 -6.01905547e-02 -4.96847332e-01
-1.07630789e-01 -5.71152449e-01 -6.90789819e-02 4.43430662e-01
4.28527862e-01 -2.18356073e-01 1.32149547e-01 1.79283693e-01
6.55479550e-01 1.83424175e-01 4.59485471e-01 -4.01531428e-01
5.76158702e-01 4.98091280e-02 8.50360930e-01 3.85796666e-01
-5.33396006e-01 8.11770797e-01 4.49279994e-01 -4.49062437e-01
-9.10744905e-01 -1.21681607e+00 -4.22876447e-01 6.24380589e-01
4.10579503e-01 9.51310322e-02 -8.62826347e-01 -1.36704397e+00
-6.16912171e-02 5.12594938e-01 -6.52941525e-01 1.90256730e-01
-4.79751170e-01 -1.05035806e+00 3.04970890e-01 7.09583700e-01
4.46898669e-01 -6.60218120e-01 -6.28266573e-01 2.70871937e-01
-3.85022670e-01 -1.32926846e+00 -4.69163954e-01 2.45091408e-01
-1.24852753e+00 -1.25940466e+00 -1.21831596e+00 -7.95417964e-01
1.03448415e+00 1.55412359e-02 1.22178090e+00 -3.71151604e-02
-7.82431811e-02 6.19199395e-01 -9.01811197e-02 -2.09839031e-01
-6.26418829e-01 5.75932637e-02 -5.84740341e-02 -1.97823882e-01
1.56417131e-01 -5.53104162e-01 -6.98906004e-01 5.18885374e-01
-1.17279446e+00 -5.04505122e-03 5.56432843e-01 1.15889406e+00
1.19634736e+00 -2.77406275e-01 6.72399879e-01 -1.29226339e+00
-2.48406772e-02 -4.27129328e-01 -7.11758912e-01 4.52578127e-01
-6.51909590e-01 1.73691943e-01 4.56467479e-01 -1.72847122e-01
-1.16456115e+00 5.02984405e-01 -2.25573823e-01 -6.42009795e-01
-3.00939798e-01 6.38784051e-01 -1.71216294e-01 -3.41380946e-02
5.37617564e-01 7.93339014e-02 2.14760929e-01 -3.44967395e-01
3.56560558e-01 3.45474392e-01 5.70015967e-01 -6.56005323e-01
4.20755059e-01 6.98886156e-01 2.10768590e-03 -4.23705012e-01
-9.92282152e-01 -6.05173767e-01 -9.84850287e-01 -1.31107315e-01
1.18456113e+00 -9.53899324e-01 -2.59067804e-01 4.30032432e-01
-7.96159506e-01 -3.16977978e-01 -2.72034317e-01 7.22633839e-01
-6.48838580e-01 7.32775927e-01 -6.59419179e-01 -2.47580066e-01
-2.99125642e-01 -1.58620203e+00 1.14653087e+00 2.43145451e-01
2.60724634e-01 -1.35379386e+00 8.13027024e-02 6.39917135e-01
4.27649496e-03 3.35165888e-01 9.72068965e-01 -1.10171902e+00
-5.92539787e-01 3.20522897e-02 -1.65334329e-01 3.93323898e-01
2.31671885e-01 -4.68669832e-01 -7.15774357e-01 -2.47227281e-01
1.80206478e-01 -5.18331826e-01 8.72950435e-01 7.63046205e-01
1.19073522e+00 2.89353848e-01 -4.20485437e-01 6.19541585e-01
1.41899502e+00 5.39473481e-02 3.04902196e-01 1.32730037e-01
7.97096074e-01 4.30769265e-01 6.53334856e-01 2.73339421e-01
5.28163910e-01 3.71700168e-01 6.51385665e-01 -4.26173419e-01
1.02578662e-01 3.66545399e-03 -2.47528285e-01 9.41219330e-01
2.10221913e-02 -1.28511012e-01 -1.27277589e+00 6.86935544e-01
-1.84264028e+00 -4.03872073e-01 -2.76784468e-02 2.10831857e+00
8.42066884e-01 1.23794507e-02 1.31926104e-01 -2.94455379e-01
7.05380440e-01 -1.30761608e-01 -8.13675821e-01 2.80088484e-01
1.52493387e-01 -3.52513567e-02 5.53529084e-01 2.59827703e-01
-1.40150654e+00 6.35829508e-01 5.87407923e+00 5.26616871e-01
-1.12758577e+00 3.26633662e-01 9.90216434e-01 1.83587655e-01
-3.10793102e-01 -3.32454711e-01 -3.15390825e-01 2.87300855e-01
5.36958158e-01 4.34290916e-02 1.40640318e-01 9.79860604e-01
-1.42003015e-01 -2.88269371e-01 -1.30075610e+00 8.89231324e-01
5.04466854e-02 -1.22591388e+00 -1.26828656e-01 6.53863773e-02
9.28373039e-01 2.13407964e-01 -9.81720015e-02 2.07650527e-01
3.34692657e-01 -8.10364962e-01 2.96879888e-01 2.18297794e-01
7.14328468e-01 -5.88085949e-01 9.45638120e-01 5.73425710e-01
-8.30227673e-01 3.15128565e-01 -1.75057843e-01 7.54434288e-01
3.11137408e-01 8.30833197e-01 -8.76209080e-01 9.57115233e-01
5.86861193e-01 7.27940083e-01 -2.35911027e-01 9.22537029e-01
-1.90078676e-01 3.58680338e-01 -3.81554037e-01 5.46270609e-01
1.63730845e-01 -3.17897320e-01 3.84809822e-01 9.19787586e-01
2.26376757e-01 1.86891913e-01 4.63927150e-01 8.66608858e-01
-2.62985677e-01 8.86865780e-02 -4.13811624e-01 1.36651129e-01
-5.51255383e-02 1.31556368e+00 -1.23434293e+00 -5.24682224e-01
-4.99886900e-01 8.87349308e-01 2.27161765e-01 2.21866056e-01
-9.31162536e-01 3.54715049e-01 7.28555098e-02 -2.62267619e-01
2.81166106e-01 -8.28959346e-02 -3.04569960e-01 -1.55240905e+00
-2.04919711e-01 -7.04267800e-01 7.06566155e-01 -3.24409038e-01
-1.53032124e+00 5.20211577e-01 3.64676304e-02 -1.45477724e+00
-3.25502783e-01 -5.90725899e-01 -1.41408667e-01 5.96648693e-01
-1.66324437e+00 -1.21168339e+00 -2.03938127e-01 5.96490145e-01
3.92043263e-01 1.51458919e-01 8.33078265e-01 4.29109097e-01
-4.69071865e-01 3.50004256e-01 2.15761736e-01 3.49482566e-01
8.16481709e-01 -1.45438802e+00 -1.33353129e-01 7.39691854e-01
7.97730908e-02 3.96731526e-01 1.88048199e-01 -7.01937616e-01
-1.24340463e+00 -1.19123793e+00 1.54658720e-01 -2.95042634e-01
3.76561314e-01 -5.85879199e-02 -9.71283376e-01 9.10809338e-01
8.90616104e-02 5.90778649e-01 1.04988694e+00 -5.27271368e-02
-1.22437373e-01 1.47394285e-01 -1.44416714e+00 2.43174002e-01
7.64645755e-01 -2.61450082e-01 -7.15043366e-01 5.14322400e-01
6.49524748e-01 -8.95807445e-01 -1.42771566e+00 6.98805034e-01
3.19004923e-01 -9.96899128e-01 8.91674161e-01 -4.95789051e-01
3.49276960e-01 -2.87222534e-01 -2.68728146e-03 -1.42760301e+00
4.31666635e-02 -1.54934347e-01 9.78822075e-03 9.76997495e-01
3.76279235e-01 -5.77127814e-01 9.15124357e-01 7.65341580e-01
-3.29269946e-01 -6.93765044e-01 -9.30199623e-01 -6.58622265e-01
3.90012264e-01 -2.95530409e-01 3.00199777e-01 1.21547639e+00
4.85888831e-02 2.22719818e-01 -1.41838042e-03 3.34272683e-01
9.10789669e-01 5.03669977e-01 4.23235297e-01 -1.15673256e+00
-3.01211834e-01 -1.00587815e-01 -1.90899074e-01 -8.19652379e-01
1.88492000e-01 -1.13208544e+00 1.64912507e-01 -1.71149909e+00
4.18029457e-01 -5.38843155e-01 -4.48510349e-01 4.59665239e-01
-2.68447578e-01 1.70220584e-01 1.55413728e-02 2.91766554e-01
-7.95594752e-01 4.61352944e-01 1.68599463e+00 -2.90043324e-01
-1.79660857e-01 1.32275552e-01 -4.07903999e-01 9.17685449e-01
6.24640822e-01 -4.40865397e-01 -5.17064154e-01 -5.06639779e-01
-3.59549135e-01 5.11488497e-01 2.28627011e-01 -8.56629908e-01
1.46824911e-01 1.13533475e-01 4.77963328e-01 -5.78562796e-01
1.07021618e-03 -1.17057967e+00 -1.14936596e-02 5.08773446e-01
-1.65203556e-01 -3.91363204e-01 2.43430421e-01 7.72428572e-01
-3.48086685e-01 -2.19149515e-01 1.03147149e+00 -4.51980263e-01
-5.50091445e-01 5.99448681e-01 -1.61009744e-01 3.08541059e-01
1.21277666e+00 3.49221304e-02 6.04509749e-02 -7.12100342e-02
-1.18787277e+00 5.95357776e-01 5.32220423e-01 -7.84763135e-03
5.36418259e-01 -1.15033329e+00 -4.82200652e-01 1.43465102e-01
1.52802393e-01 6.56471968e-01 5.46743155e-01 1.13309717e+00
-4.90992010e-01 2.84827828e-01 -2.32758105e-01 -1.30693543e+00
-8.42618644e-01 8.86604607e-01 4.66472358e-01 -6.87961638e-01
-6.11229420e-01 5.59086800e-01 5.95831871e-01 -7.66362071e-01
1.04411937e-01 -6.10092402e-01 -1.50023744e-01 -8.89872238e-02
1.44860134e-01 2.78190412e-02 1.52817816e-01 -5.47202289e-01
-3.66862983e-01 5.21534801e-01 -2.01439112e-01 -1.33952461e-02
1.30508673e+00 -1.67465180e-01 -6.86664507e-02 2.60470420e-01
1.12083304e+00 -2.88671404e-01 -1.33621860e+00 -4.34544563e-01
7.83944875e-02 -1.47510991e-01 1.17064580e-01 -8.24996352e-01
-1.44706130e+00 9.84427512e-01 5.84865153e-01 -9.29123685e-02
1.27276063e+00 1.17531516e-01 8.09244394e-01 1.90735489e-01
4.78174388e-01 -9.28273082e-01 2.61554238e-03 1.44082561e-01
4.14867222e-01 -1.66661263e+00 3.37965526e-02 -6.61861002e-01
-9.69422460e-01 1.04937363e+00 7.16752529e-01 7.56641105e-02
7.40481138e-01 2.56852150e-01 2.96304852e-01 -1.56882063e-01
-2.01388285e-01 -9.87333804e-02 2.90888041e-01 5.04017174e-01
3.06406349e-01 4.59324829e-02 -1.43011346e-01 7.84493268e-01
4.83720958e-01 2.82989383e-01 3.02445561e-01 9.93799806e-01
-8.58716816e-02 -1.14086044e+00 -4.13626015e-01 3.58834982e-01
-6.60557926e-01 1.99446172e-01 1.13336891e-01 8.91819060e-01
-3.83090563e-02 5.13750672e-01 -1.73688203e-01 5.35775796e-02
1.61413193e-01 1.33006081e-01 6.76120460e-01 -7.00793743e-01
-4.73362088e-01 5.91381371e-01 -2.82274872e-01 -4.04993951e-01
-7.59248257e-01 -6.81241572e-01 -1.64135289e+00 3.18081886e-01
-3.02356899e-01 1.70625597e-02 4.93985742e-01 1.04145217e+00
2.23589137e-01 6.88343585e-01 4.76496726e-01 -7.43705928e-01
-5.47337770e-01 -6.25472248e-01 -5.84964633e-01 5.43632567e-01
2.48098180e-01 -6.85277104e-01 2.12526843e-02 3.22287112e-01]
|
[14.602252960205078, -2.104022979736328]
|
5c16b06f-f59b-4fdb-a03a-63d7526a639a
|
revisiting-contrastive-methods-for
|
2106.05967
| null |
https://arxiv.org/abs/2106.05967v3
|
https://arxiv.org/pdf/2106.05967v3.pdf
|
Revisiting Contrastive Methods for Unsupervised Learning of Visual Representations
|
Contrastive self-supervised learning has outperformed supervised pretraining on many downstream tasks like segmentation and object detection. However, current methods are still primarily applied to curated datasets like ImageNet. In this paper, we first study how biases in the dataset affect existing methods. Our results show that current contrastive approaches work surprisingly well across: (i) object- versus scene-centric, (ii) uniform versus long-tailed and (iii) general versus domain-specific datasets. Second, given the generality of the approach, we try to realize further gains with minor modifications. We show that learning additional invariances -- through the use of multi-scale cropping, stronger augmentations and nearest neighbors -- improves the representations. Finally, we observe that MoCo learns spatially structured representations when trained with a multi-crop strategy. The representations can be used for semantic segment retrieval and video instance segmentation without finetuning. Moreover, the results are on par with specialized models. We hope this work will serve as a useful study for other researchers. The code and models are available at https://github.com/wvangansbeke/Revisiting-Contrastive-SSL.
|
['Luc van Gool', 'Stamatios Georgoulis', 'Simon Vandenhende', 'Wouter Van Gansbeke']
|
2021-06-10
| null |
http://proceedings.neurips.cc/paper/2021/hash/8757150decbd89b0f5442ca3db4d0e0e-Abstract.html
|
http://proceedings.neurips.cc/paper/2021/file/8757150decbd89b0f5442ca3db4d0e0e-Paper.pdf
|
neurips-2021-12
|
['video-instance-segmentation']
|
['computer-vision']
|
[ 3.11964214e-01 -5.86125962e-02 -6.45825624e-01 -4.32465881e-01
-7.41534173e-01 -7.69657731e-01 6.17716849e-01 -5.75364977e-02
-5.66828907e-01 4.77107942e-01 1.83227196e-01 -2.18683451e-01
1.45065328e-02 -5.98572910e-01 -9.80274737e-01 -5.02091706e-01
-8.18197150e-03 2.28346005e-01 5.73913395e-01 -2.92014509e-01
3.34419191e-01 4.78245676e-01 -1.66607845e+00 3.17029238e-01
6.58477545e-01 7.76082873e-01 4.45854694e-01 5.53996742e-01
2.68999990e-02 7.76798427e-01 -3.65077674e-01 -1.63434237e-01
4.72246557e-01 -4.03020352e-01 -1.08298969e+00 1.27634153e-01
8.90707195e-01 -1.56592265e-01 -3.58951807e-01 8.90351355e-01
4.19150442e-01 2.28428066e-01 6.37310624e-01 -1.07389641e+00
-8.05560529e-01 5.13653278e-01 -6.81728601e-01 4.92392212e-01
-1.53963370e-02 2.95165837e-01 1.09800529e+00 -8.50934684e-01
8.96769464e-01 1.18388557e+00 7.28691816e-01 6.93879545e-01
-1.33344781e+00 -5.98107994e-01 5.26902795e-01 2.79156625e-01
-1.18880486e+00 -5.91767728e-01 6.80610240e-01 -3.74013394e-01
7.88168192e-01 1.95616573e-01 3.71663988e-01 1.14235520e+00
-1.55625597e-01 1.24274552e+00 1.22636461e+00 -4.04642910e-01
-3.99800502e-02 1.28900751e-01 1.78735822e-01 8.08858991e-01
2.46893942e-01 2.05059111e-01 -2.40757793e-01 2.50956059e-01
8.76401603e-01 2.88595110e-02 -1.51943862e-01 -6.49589837e-01
-1.35633409e+00 8.48399401e-01 7.63175011e-01 5.17482340e-01
-9.22464207e-02 4.01882589e-01 3.98613691e-01 2.88528293e-01
6.71306849e-01 7.07424402e-01 -6.94866061e-01 4.38125022e-02
-1.17928410e+00 1.76576287e-01 4.44352388e-01 1.01830363e+00
9.46182072e-01 1.08510196e-01 -1.17326252e-01 9.46524501e-01
3.18790674e-02 4.22069132e-01 5.88305831e-01 -1.13585985e+00
2.52124488e-01 2.52029628e-01 -1.88604265e-01 -7.40560889e-01
-5.33001840e-01 -5.57893217e-01 -4.75061089e-01 2.49178916e-01
6.49668694e-01 -1.62744045e-01 -1.27012002e+00 1.88217115e+00
1.41330644e-01 2.38965720e-01 -1.98938280e-01 1.07410491e+00
8.16034734e-01 4.56390977e-01 3.20772618e-01 3.15888435e-01
1.39026499e+00 -1.38057661e+00 -4.92518097e-01 -5.07487953e-01
7.42678940e-01 -8.18407357e-01 1.34306717e+00 1.59470454e-01
-1.12239492e+00 -6.15362525e-01 -8.83685112e-01 -2.63475031e-01
-6.67313874e-01 1.76897183e-01 9.29065585e-01 5.93725085e-01
-1.16203415e+00 7.13320732e-01 -8.33491862e-01 -7.83177435e-01
8.69083941e-01 1.88371122e-01 -3.16211253e-01 -2.21437067e-01
-8.65045309e-01 8.41348052e-01 3.37819070e-01 -2.96550423e-01
-9.65600729e-01 -6.53273642e-01 -8.55223715e-01 -2.71589816e-01
4.01672363e-01 -6.53364599e-01 1.24267948e+00 -1.39899123e+00
-1.18257058e+00 1.20831728e+00 -2.57586896e-01 -5.19354463e-01
4.71388698e-01 -3.68204385e-01 -1.34093165e-01 3.24669838e-01
3.74028713e-01 1.31561375e+00 7.76074409e-01 -1.33559263e+00
-5.44188976e-01 -2.73982167e-01 2.50981092e-01 2.48648629e-01
-1.56702489e-01 5.30658290e-03 -5.97404420e-01 -9.28851902e-01
7.73480088e-02 -1.01377654e+00 -3.17622364e-01 2.92889532e-02
-4.26046163e-01 -1.30032212e-01 6.62962735e-01 -4.97126639e-01
8.20598781e-01 -2.24176383e+00 9.77215022e-02 -1.34319544e-01
2.80219633e-02 4.04672265e-01 -5.76604724e-01 2.82518625e-01
-3.87112826e-01 2.14251846e-01 -3.40037286e-01 -3.43152791e-01
-1.91182479e-01 1.81391165e-01 -9.88942757e-02 6.08130872e-01
3.88890684e-01 1.23159218e+00 -8.54356945e-01 -5.71697891e-01
2.55045533e-01 2.32889235e-01 -5.80745220e-01 -1.43955678e-01
-3.17196965e-01 3.75449717e-01 -2.20220044e-01 8.03466082e-01
6.31101429e-01 -3.68909627e-01 -2.07610831e-01 -2.20853135e-01
-1.40919939e-01 2.23689228e-01 -9.44843471e-01 2.02787185e+00
-3.99647713e-01 9.35380995e-01 2.44889855e-02 -1.26538467e+00
3.83740485e-01 -2.07675621e-01 3.00656021e-01 -8.50298107e-01
9.53979418e-03 7.81071112e-02 -1.00557707e-01 -4.70400244e-01
4.44051921e-01 -3.71565074e-02 1.40426204e-01 3.22539002e-01
3.72744769e-01 -1.58890545e-01 3.63467127e-01 2.96678603e-01
8.60618949e-01 5.06754994e-01 1.54928863e-01 -4.98587340e-01
9.64211747e-02 2.73554355e-01 4.10989612e-01 9.89004970e-01
-3.29706669e-01 9.30731893e-01 3.61568183e-01 -2.68478751e-01
-8.79127800e-01 -1.02511275e+00 -2.82030225e-01 1.68980360e+00
3.30952317e-01 -1.59905776e-01 -9.14318502e-01 -8.87979746e-01
4.26490828e-02 6.64034426e-01 -7.82832146e-01 -9.06557217e-03
-6.20205700e-01 -6.22125030e-01 6.75642312e-01 9.24741030e-01
6.02854788e-01 -1.18662620e+00 -5.19034922e-01 -1.51734546e-01
-5.69245666e-02 -1.05767047e+00 -4.84328657e-01 4.80446160e-01
-1.13994050e+00 -1.07261598e+00 -9.10265744e-01 -1.02733052e+00
6.06505692e-01 5.59592903e-01 1.16950727e+00 1.09788455e-01
-3.68785143e-01 6.27117693e-01 -4.81207430e-01 -4.07395810e-01
3.64793800e-02 5.27561069e-01 -2.79763311e-01 -2.52183437e-01
2.71006078e-01 -4.12062794e-01 -7.85103679e-01 4.44972217e-01
-1.11854005e+00 -1.71223059e-02 6.81940079e-01 6.84805095e-01
5.31224906e-01 -4.02621210e-01 5.19232273e-01 -1.27482688e+00
3.06075305e-01 -4.81407732e-01 -3.95462036e-01 5.54858558e-02
-4.82996613e-01 6.27691671e-02 2.55559742e-01 -5.46397150e-01
-9.28464413e-01 1.71588674e-01 -1.54092804e-01 -3.25160027e-01
-4.85105187e-01 2.00361148e-01 9.02064070e-02 -9.08645764e-02
9.52697635e-01 1.66334510e-01 -6.82664011e-03 -5.12449503e-01
6.77085936e-01 4.24413174e-01 3.48232448e-01 -5.65997839e-01
8.19134116e-01 7.58334041e-01 -2.94580847e-01 -8.26170862e-01
-9.87297475e-01 -5.82226276e-01 -7.52938151e-01 -8.56242550e-04
1.04972279e+00 -1.05662155e+00 -7.19395131e-02 4.07019079e-01
-8.72994959e-01 -9.67441022e-01 -5.45122266e-01 4.18521881e-01
-6.29060388e-01 3.13697875e-01 -7.25094020e-01 -4.26384121e-01
1.18089765e-01 -1.04269814e+00 1.10546970e+00 2.43193299e-01
-1.58742934e-01 -1.16123545e+00 -7.31540620e-02 5.19302309e-01
5.15380383e-01 -5.73177636e-02 5.40822864e-01 -7.30133653e-01
-7.28540957e-01 1.00640789e-01 -4.52406853e-01 4.37551409e-01
1.33781955e-01 -8.62407461e-02 -1.17528582e+00 -2.86602467e-01
-5.49009979e-01 -5.35057127e-01 1.47518837e+00 5.58984280e-01
1.38673460e+00 -6.96704462e-02 -3.30797762e-01 8.95295322e-01
1.26348352e+00 -2.84420490e-01 7.07584143e-01 5.86963534e-01
6.81986928e-01 6.56854868e-01 7.38461673e-01 5.45083657e-02
4.16099727e-01 4.51508164e-01 5.07696509e-01 -5.33020139e-01
-6.47831678e-01 -1.28497168e-01 1.94302410e-01 3.78267556e-01
-1.49590224e-01 -1.62270650e-01 -8.21692884e-01 9.46343124e-01
-1.90566444e+00 -9.39052701e-01 -1.03548896e-02 1.83349812e+00
6.69627607e-01 -9.90999490e-03 3.17666709e-01 -1.99642405e-01
6.84006631e-01 4.07720804e-01 -6.19836867e-01 -2.10866496e-01
-2.72156745e-01 3.10843199e-01 9.30171430e-01 3.22856367e-01
-1.48732674e+00 1.53079200e+00 6.26892138e+00 9.00312781e-01
-1.27052188e+00 2.59763867e-01 6.95896208e-01 -9.54341143e-02
-2.12634385e-01 3.69650051e-02 -5.94917238e-01 2.96018273e-01
5.90555012e-01 4.33287740e-01 3.90015841e-01 8.90061915e-01
-5.02075022e-03 -1.76734820e-01 -1.11056936e+00 8.61468554e-01
1.87452316e-01 -1.34805274e+00 -1.46887928e-01 -7.18814954e-02
9.17153418e-01 5.65125108e-01 3.72375488e-01 3.76222849e-01
3.59070629e-01 -1.03702259e+00 8.05296540e-01 3.04197967e-01
6.71671808e-01 -2.71494538e-01 5.24736583e-01 1.24981225e-01
-9.49980080e-01 -1.04741137e-02 -3.40106308e-01 -1.31647354e-02
-9.33687761e-02 3.11004758e-01 -5.01170874e-01 3.32018524e-01
8.84943008e-01 8.90036106e-01 -9.56837356e-01 1.07788372e+00
-2.47866154e-01 8.22087765e-01 -2.77581245e-01 3.28739285e-02
4.50910747e-01 8.57833773e-02 2.93370515e-01 1.62009656e+00
-5.06528765e-02 -2.90102661e-01 1.13784201e-01 7.33418763e-01
-7.44679943e-02 1.60103589e-01 -6.68605626e-01 -1.69671308e-02
1.65150076e-01 1.23660171e+00 -1.04877126e+00 -3.67864102e-01
-5.27703702e-01 1.08643854e+00 4.35793132e-01 6.50506377e-01
-7.58341730e-01 -3.01635653e-01 6.77239060e-01 2.92117774e-01
7.09230244e-01 -1.86816782e-01 -3.98049444e-01 -1.29614484e+00
-2.76443839e-01 -8.73197317e-01 4.69141394e-01 -7.99942613e-01
-1.23937488e+00 3.10060352e-01 1.15773655e-01 -9.19209182e-01
-5.83733805e-02 -7.63646841e-01 -4.77499604e-01 4.14460272e-01
-1.87583172e+00 -1.28814280e+00 -1.83509216e-01 6.58978462e-01
7.92886972e-01 -3.90937254e-02 5.97146630e-01 2.90499717e-01
-6.02402449e-01 5.99414349e-01 4.69247214e-02 4.25953954e-01
9.44479704e-01 -1.28760910e+00 3.68398339e-01 8.38094056e-01
5.00271738e-01 5.96274018e-01 5.22144973e-01 -3.93054307e-01
-1.00194240e+00 -1.14456928e+00 4.69037682e-01 -5.73965013e-01
5.76381624e-01 -3.75919998e-01 -7.68454432e-01 1.01790690e+00
4.80128646e-01 4.39795047e-01 5.08234978e-01 3.19307446e-01
-5.62724888e-01 -5.46367839e-02 -1.07813275e+00 6.06567740e-01
1.38269746e+00 -4.08214658e-01 -5.02322614e-01 5.28942406e-01
6.59230590e-01 -3.17790508e-01 -5.30518711e-01 4.49415416e-01
4.60745484e-01 -9.90982652e-01 1.00794137e+00 -8.73225212e-01
5.06113172e-01 -1.56472892e-01 -2.26761207e-01 -1.21754277e+00
-4.87842858e-01 -2.36119807e-01 1.65974140e-01 1.16123271e+00
5.29196501e-01 -6.61937416e-01 9.12390828e-01 1.42440498e-01
-2.28131145e-01 -6.59420729e-01 -5.71709633e-01 -9.30746913e-01
4.34967965e-01 -4.73571837e-01 3.02239895e-01 1.07731938e+00
-3.56753230e-01 3.13368082e-01 -1.39992937e-01 6.47005737e-02
5.47151983e-01 1.45667121e-01 8.43147159e-01 -9.49048758e-01
-2.31588185e-01 -5.60288608e-01 -4.38234776e-01 -1.29637575e+00
1.97181106e-01 -9.81126249e-01 -1.49588138e-02 -1.51894939e+00
2.13567004e-01 -5.16015768e-01 -4.60755140e-01 6.38291955e-01
-1.23339519e-01 6.07322097e-01 4.27362412e-01 1.79640323e-01
-9.51141238e-01 1.72253966e-01 1.33636951e+00 -8.47126693e-02
-4.54356335e-02 1.22058382e-02 -9.30976927e-01 7.50852048e-01
1.23459256e+00 -3.35756242e-01 -3.82952809e-01 -6.47042692e-01
-1.36637837e-01 -4.31967676e-01 4.89527196e-01 -9.89719272e-01
6.05594516e-02 -1.08777925e-01 3.90553743e-01 -3.61601681e-01
2.77779937e-01 -5.72241783e-01 -5.33506513e-01 3.27559590e-01
-5.10831416e-01 -1.27870321e-01 4.53312069e-01 5.07415891e-01
-1.71247184e-01 -4.10454988e-01 8.59084070e-01 -3.98321480e-01
-1.10642624e+00 1.87444717e-01 -3.42354268e-01 3.92107397e-01
9.26939189e-01 -3.16211224e-01 -4.33364838e-01 -4.09851730e-01
-6.76547885e-01 1.57245025e-01 6.24418914e-01 5.43507636e-01
4.14925456e-01 -9.40679669e-01 -6.60898447e-01 4.15614545e-02
2.91164279e-01 -2.98711956e-02 1.85829043e-01 8.65088582e-01
-6.32548034e-01 4.69751239e-01 -2.42178544e-01 -7.72904813e-01
-1.12884319e+00 5.55734038e-01 2.01828748e-01 -1.62637867e-02
-4.45213854e-01 1.06814921e+00 5.33343673e-01 -5.80404222e-01
2.48173103e-01 -4.46982741e-01 -2.90171020e-02 2.52794743e-01
2.12567300e-01 3.06488603e-01 -5.11394292e-02 -6.82206810e-01
-3.97423327e-01 7.21781790e-01 -2.99977690e-01 1.95118748e-02
1.54125965e+00 -1.36565939e-01 1.77244365e-01 4.35085326e-01
1.26048207e+00 2.08360050e-02 -1.52055097e+00 -2.57748395e-01
8.22480544e-02 -3.83103043e-01 1.17842533e-01 -7.78281808e-01
-1.27853143e+00 9.33101177e-01 7.03921080e-01 6.19870797e-02
1.08223259e+00 3.74119520e-01 5.62942982e-01 3.70268077e-01
1.55700341e-01 -1.15180612e+00 1.89677328e-01 3.63815665e-01
5.96514702e-01 -1.54233372e+00 1.26890123e-01 -3.86717558e-01
-8.10089767e-01 7.99681067e-01 6.44622445e-01 -3.60510617e-01
5.07742584e-01 9.38690305e-02 2.73990870e-01 -1.48380965e-01
-4.70502585e-01 -8.19967985e-01 3.63878071e-01 8.12577724e-01
5.86710513e-01 -9.90278795e-02 -3.65906917e-02 1.21915251e-01
-8.42304900e-03 -1.17999218e-01 3.13478172e-01 9.19829965e-01
-4.22287166e-01 -1.10796118e+00 -2.51715034e-01 4.38417822e-01
-5.06780148e-01 -2.49060810e-01 -4.68510568e-01 1.13611102e+00
1.93861470e-01 8.17542315e-01 1.11493379e-01 -3.62111591e-02
1.99098140e-01 8.41212645e-02 6.25526667e-01 -8.53787780e-01
-5.21744132e-01 6.95186406e-02 -2.29199715e-02 -8.20966005e-01
-7.38734066e-01 -8.07022631e-01 -1.06243896e+00 -6.02147821e-03
-3.44073892e-01 -2.64856249e-01 6.12549305e-01 6.59361005e-01
4.41193908e-01 4.98624682e-01 2.52959877e-01 -9.81844485e-01
-2.60021210e-01 -1.00049376e+00 -5.23105025e-01 5.72134197e-01
3.66639227e-01 -6.97247207e-01 -3.35917622e-01 2.15479046e-01]
|
[9.648971557617188, 1.5565307140350342]
|
d3a42b10-84e0-45b8-bf68-2681a2297f75
|
interaction-modeling-with-multiplex-attention
|
2208.10660
| null |
https://arxiv.org/abs/2208.10660v2
|
https://arxiv.org/pdf/2208.10660v2.pdf
|
Interaction Modeling with Multiplex Attention
|
Modeling multi-agent systems requires understanding how agents interact. Such systems are often difficult to model because they can involve a variety of types of interactions that layer together to drive rich social behavioral dynamics. Here we introduce a method for accurately modeling multi-agent systems. We present Interaction Modeling with Multiplex Attention (IMMA), a forward prediction model that uses a multiplex latent graph to represent multiple independent types of interactions and attention to account for relations of different strengths. We also introduce Progressive Layer Training, a training strategy for this architecture. We show that our approach outperforms state-of-the-art models in trajectory forecasting and relation inference, spanning three multi-agent scenarios: social navigation, cooperative task achievement, and team sports. We further demonstrate that our approach can improve zero-shot generalization and allows us to probe how different interactions impact agent behavior.
|
['Nick Haber', 'Jiajun Wu', 'Mykel Kochenderfer', 'Jiachen Li', 'Ruohan Zhang', 'Isaac Kauvar', 'Fan-Yun Sun']
|
2022-08-23
| null | null | null | null |
['trajectory-forecasting', 'social-navigation']
|
['computer-vision', 'robots']
|
[-3.14936340e-01 2.67758928e-02 -2.05961630e-01 -1.11443043e-01
-1.64660528e-01 -4.05561507e-01 1.25811481e+00 1.42207012e-01
-1.69742420e-01 7.38004804e-01 4.47499752e-01 -3.26627791e-01
-5.06530881e-01 -9.15171087e-01 -8.58524561e-01 -3.75606000e-01
-8.21049869e-01 1.25323844e+00 2.57945865e-01 -6.92297101e-01
-1.69420745e-02 3.30561846e-01 -1.35389912e+00 3.89473021e-01
8.07317138e-01 7.32237101e-02 -6.82184547e-02 1.16024756e+00
2.16328532e-01 1.60765553e+00 -4.86718088e-01 -2.56381661e-01
4.07884307e-02 -2.73072243e-01 -7.96980262e-01 -1.28614437e-02
-4.21801396e-03 -4.94891018e-01 -7.07154512e-01 5.14665306e-01
1.58712670e-01 6.19798720e-01 1.06205475e+00 -1.77058399e+00
-7.47193456e-01 9.04357612e-01 -3.44238698e-01 2.83753455e-01
2.87592560e-01 5.49508929e-01 1.19166064e+00 -2.42231309e-01
6.54050410e-01 1.65944278e+00 7.40532875e-01 4.89090681e-01
-1.44272089e+00 -4.79854643e-01 5.00550091e-01 4.25087750e-01
-7.08215952e-01 -2.73527443e-01 4.54619735e-01 -8.53934765e-01
1.60955000e+00 9.20888211e-04 6.97654605e-01 1.30568624e+00
3.22300047e-01 7.82594681e-01 6.62987947e-01 2.12505716e-03
8.73409882e-02 -2.64070541e-01 8.02435160e-01 9.16278720e-01
-2.38878801e-02 1.63521156e-01 -5.54572940e-01 -4.91845995e-01
6.65686071e-01 1.82223365e-01 3.01131576e-01 -2.68679321e-01
-1.04974759e+00 8.84856880e-01 4.80816364e-01 -1.90827683e-01
-6.65348291e-01 5.97925246e-01 1.38175011e-01 3.16636562e-01
4.35454637e-01 6.26636446e-01 -4.12943453e-01 -3.32138389e-01
-7.87170902e-02 7.96708941e-01 1.11345410e+00 7.21186101e-01
6.63097382e-01 -1.56196475e-01 -9.86657664e-02 8.10168684e-01
4.80412215e-01 1.31450981e-01 1.83268443e-01 -1.24468064e+00
3.74014825e-01 8.17709863e-01 3.67889374e-01 -7.34905303e-01
-8.67077887e-01 -2.42052287e-01 -5.22315145e-01 3.74432236e-01
6.65153623e-01 -5.27066946e-01 -6.77563906e-01 1.89019918e+00
1.72912151e-01 6.72301233e-01 3.05375278e-01 4.51983750e-01
7.61124134e-01 6.29512250e-01 2.14765742e-01 8.99875909e-02
1.02398157e+00 -1.64316809e+00 -4.51205969e-01 -3.28446984e-01
9.42490995e-01 -4.50554192e-02 8.59071791e-01 -1.63277447e-01
-1.32386792e+00 -2.08448604e-01 -6.96017563e-01 4.42915261e-02
-6.31882191e-01 -4.92134631e-01 1.12444413e+00 1.91464573e-01
-9.68716323e-01 6.91228986e-01 -1.27820909e+00 -4.58324045e-01
4.19710368e-01 5.89511037e-01 3.85860614e-02 2.71930486e-01
-1.20277071e+00 1.13458323e+00 -1.27640307e-01 -2.66971558e-01
-1.18574464e+00 -8.39868307e-01 -7.23952174e-01 3.80760342e-01
4.22077209e-01 -9.70307469e-01 1.34532118e+00 -4.65051681e-01
-1.65074170e+00 2.45629624e-01 -1.01133764e-01 -6.92983031e-01
3.54805052e-01 -4.72382419e-02 -1.17504098e-01 -3.13921183e-01
-8.25023577e-02 4.77934241e-01 2.21771941e-01 -1.06641984e+00
-5.93026996e-01 -1.40633076e-01 6.01178527e-01 5.41740656e-01
-8.99591073e-02 9.13877264e-02 -8.75891820e-02 -1.61327302e-01
-6.37206674e-01 -1.32692540e+00 -6.04076862e-01 -4.86280680e-01
-3.78075987e-01 -4.71096426e-01 4.93317187e-01 -2.41604641e-01
1.01263106e+00 -1.50062501e+00 8.14472914e-01 1.15706272e-01
6.36754513e-01 3.88643891e-02 -4.81502563e-01 1.01247466e+00
2.52310634e-01 9.78344157e-02 3.42561193e-02 -8.35789323e-01
4.65129435e-01 3.21549028e-01 -1.08686551e-01 8.73815864e-02
-7.49208033e-02 1.30554307e+00 -1.02585149e+00 2.34064218e-02
1.77025482e-01 3.96495014e-01 -7.29850471e-01 2.43799627e-01
-6.60932004e-01 4.07512009e-01 -3.81785274e-01 2.24597052e-01
1.24952421e-01 -7.50039935e-01 4.87162203e-01 3.72824371e-01
1.02478318e-01 3.59420776e-01 -8.55168998e-01 1.21201062e+00
-2.54752755e-01 6.62920356e-01 -9.59857404e-02 -5.97202361e-01
2.62001723e-01 5.14152125e-02 6.73625052e-01 -2.35326961e-01
1.06064111e-01 -3.13942224e-01 5.84484458e-01 -5.46838641e-01
4.40503180e-01 3.19378227e-01 -1.43139809e-01 9.22718465e-01
2.73827836e-03 2.53547877e-01 5.59455514e-01 6.05627954e-01
1.42260921e+00 -1.12923481e-01 2.05887422e-01 -2.49603372e-02
5.51067777e-02 -1.26985263e-03 3.49491060e-01 1.17875230e+00
-1.92335144e-01 -3.19091082e-01 7.78072536e-01 -7.11926877e-01
-9.87014294e-01 -9.74813998e-01 5.08469641e-01 1.58213437e+00
1.16818875e-01 -4.95800853e-01 -4.17906851e-01 -4.60329503e-01
2.93878078e-01 6.46808565e-01 -9.18229580e-01 -1.84079170e-01
-6.82788789e-01 -1.23816049e+00 4.42504704e-01 4.90090191e-01
5.64327650e-02 -1.14451051e+00 -3.54123771e-01 3.61772031e-01
-7.55991638e-02 -9.65218961e-01 -3.18692088e-01 -1.58980936e-01
-2.50418723e-01 -1.32026172e+00 -2.58823007e-01 -4.94391799e-01
3.49517256e-01 1.34511262e-01 1.35081685e+00 3.30944449e-01
-1.32983893e-01 5.47692537e-01 1.10400636e-02 -2.40001246e-01
-7.43340492e-01 2.48114496e-01 4.00343180e-01 -1.80318579e-01
3.71832281e-01 -8.38610947e-01 -2.49365062e-01 1.81849986e-01
-4.31503803e-01 2.83826441e-01 1.30555376e-01 7.78551102e-01
-3.73084605e-01 -1.05871677e-01 5.87945640e-01 -1.04242718e+00
1.18164134e+00 -1.03854132e+00 -5.57074487e-01 3.51157278e-01
-3.34619015e-01 -5.76725751e-02 5.57965815e-01 -7.82443821e-01
-8.25509906e-01 -2.95718491e-01 1.99502915e-01 -1.40748248e-01
-2.83041358e-01 6.09299541e-01 3.65164012e-01 -1.65397942e-01
6.38327599e-01 -3.35298963e-02 3.08517456e-01 -8.77908245e-02
5.27774334e-01 2.55527556e-01 -8.74552038e-03 -5.23825407e-01
5.44805586e-01 3.27762216e-01 2.79127479e-01 -8.65951061e-01
-2.91206002e-01 -4.68657799e-02 -5.26290953e-01 -4.32225168e-01
7.84556210e-01 -8.54907870e-01 -1.68400800e+00 8.94347250e-01
-1.15338147e+00 -1.31077588e+00 1.08928971e-01 5.70328414e-01
-7.16862619e-01 5.66715049e-03 -1.46734524e+00 -1.01007998e+00
1.26003936e-01 -1.25520217e+00 5.40517926e-01 2.04802379e-01
-3.80646646e-01 -1.46800601e+00 5.04052043e-01 2.95342684e-01
4.09587711e-01 -2.46407720e-03 1.03720295e+00 -7.82287896e-01
-9.86121893e-01 2.73662150e-01 4.82746214e-02 -6.34802938e-01
-2.44122166e-02 1.23301081e-01 -4.24825370e-01 -2.93900281e-01
-8.39242816e-01 -3.98172677e-01 7.95493066e-01 6.58056557e-01
3.88478398e-01 -4.44850951e-01 -7.77152181e-01 3.21324617e-01
7.79523849e-01 2.54680693e-01 3.78403693e-01 2.68336564e-01
9.51140523e-01 7.00173855e-01 3.41865304e-03 3.57158840e-01
1.38750136e+00 8.87195110e-01 3.66882801e-01 3.52970734e-02
1.43169969e-01 -8.41101781e-02 5.08562565e-01 6.05529606e-01
-3.74735177e-01 -6.82211816e-01 -1.16847467e+00 3.79118800e-01
-2.60664344e+00 -1.37275159e+00 -2.69123495e-01 1.63850284e+00
4.16130871e-01 6.42452836e-02 7.87907898e-01 -7.01943815e-01
3.05775195e-01 1.64127097e-01 -8.80035758e-01 -1.77736029e-01
2.91536804e-02 -5.14599800e-01 3.12564224e-01 1.29314756e+00
-1.00128496e+00 1.19343066e+00 7.56682253e+00 2.81948596e-01
-3.16128582e-01 4.34855856e-02 4.08633083e-01 -3.61891270e-01
-1.54402822e-01 -2.37970933e-01 -7.91867495e-01 4.34603363e-01
1.10830212e+00 -2.04877093e-01 1.02237320e+00 3.16756248e-01
1.44025430e-01 8.54313150e-02 -1.29161441e+00 4.29365993e-01
-4.23528813e-02 -1.65216720e+00 -4.75637289e-03 3.78658712e-01
8.55934799e-01 3.86679739e-01 1.52171835e-01 8.03655922e-01
1.67371941e+00 -1.09615803e+00 1.97567061e-01 7.70549774e-01
-1.42641351e-01 -5.82830012e-01 2.56258994e-01 7.20842659e-01
-1.20112562e+00 -5.43019772e-01 -2.60048565e-02 -8.24722111e-01
5.16694903e-01 -3.60628396e-01 -8.27901065e-01 2.29572326e-01
3.52630585e-01 9.85204637e-01 -3.36244762e-01 7.29288518e-01
1.65712819e-01 3.90451431e-01 -2.57545650e-01 -2.95166433e-01
3.53600800e-01 -3.54157060e-01 7.81074226e-01 8.15153241e-01
-6.56758770e-02 3.14991504e-01 5.25128722e-01 7.74015546e-01
3.65819931e-02 -4.15043771e-01 -8.06146741e-01 -1.71118826e-01
5.51356554e-01 8.80327582e-01 -5.33865452e-01 -6.08610570e-01
-6.22313917e-01 7.19643414e-01 1.07500136e+00 7.15077221e-01
-9.51580763e-01 3.28966558e-01 1.42009878e+00 -1.86029002e-01
8.28949660e-02 -6.22121871e-01 1.70779690e-01 -1.31780624e+00
-6.31574750e-01 -8.32538486e-01 4.47404027e-01 -4.75111604e-01
-1.55742538e+00 4.83357728e-01 1.35722384e-02 -6.69203162e-01
-7.78761446e-01 -3.54481310e-01 -8.62550616e-01 5.99295974e-01
-1.15276015e+00 -1.59167695e+00 -7.56047517e-02 3.54292929e-01
5.51894963e-01 -4.87397492e-01 9.11369026e-01 1.99218407e-01
-9.85769033e-01 3.80953580e-01 1.30308315e-01 -5.15673589e-03
2.54473656e-01 -1.29442132e+00 8.93173516e-01 5.31007826e-01
5.13653159e-02 6.39313042e-01 6.94497228e-01 -9.17823017e-01
-1.28808856e+00 -9.71983254e-01 5.86787820e-01 -8.66208971e-01
9.83733296e-01 -5.30698001e-01 -8.13148916e-01 1.51337063e+00
3.28196406e-01 -4.46825832e-01 9.15916502e-01 7.09624410e-01
-2.77929783e-01 5.23234606e-01 -6.45392835e-01 1.23787391e+00
1.23832595e+00 -4.75269467e-01 -3.61574352e-01 5.93496680e-01
8.38955104e-01 -2.15826094e-01 -7.64012814e-01 -5.91205843e-02
6.51902854e-01 -8.42255712e-01 1.33156919e+00 -1.38189840e+00
6.26446128e-01 2.56666485e-02 6.10191412e-02 -1.78258407e+00
-8.45125794e-01 -7.15604901e-01 -6.78901732e-01 6.76363885e-01
6.49024725e-01 -9.25850034e-01 8.01546633e-01 1.10327351e+00
-1.14293173e-01 -6.44970477e-01 -3.90175641e-01 -5.85513234e-01
2.12261558e-01 -9.20705963e-03 8.24909747e-01 1.08560252e+00
5.82552850e-01 6.78628922e-01 -8.35636675e-01 4.08479899e-01
7.74202168e-01 5.15009798e-02 1.07803881e+00 -1.35533512e+00
-7.38378644e-01 -8.79240513e-01 -1.79148450e-01 -1.23293018e+00
6.09623373e-01 -7.82590628e-01 -3.99266630e-01 -1.71582246e+00
3.10201138e-01 -4.34001237e-01 -4.04647328e-02 3.01465362e-01
-2.78244644e-01 -1.85265206e-02 3.01753402e-01 2.65016288e-01
-9.41398144e-01 6.63404465e-01 9.52679515e-01 -2.30720088e-01
-5.75755000e-01 4.70698811e-02 -2.74486244e-01 8.12080860e-01
6.97651625e-01 -1.46130383e-01 -5.22132397e-01 -4.89280075e-01
2.47167706e-01 2.63867080e-01 4.98173654e-01 -7.83271134e-01
6.63587391e-01 -4.47073817e-01 -9.18746740e-03 -3.11934173e-01
9.52204168e-01 -4.13139403e-01 2.19387889e-01 6.39471769e-01
-7.67311513e-01 1.56402767e-01 9.78041887e-02 8.22297037e-01
3.99555326e-01 2.93715745e-01 3.11700910e-01 -2.53807306e-01
-4.78624284e-01 6.40257239e-01 -9.25320387e-01 -1.11152805e-01
1.23628783e+00 3.68455291e-01 -8.41108799e-01 -9.13590312e-01
-1.06124520e+00 9.77403998e-01 3.38415414e-01 4.33830619e-01
2.17018038e-01 -1.24403024e+00 -9.11921680e-01 -2.08253134e-02
-1.72347575e-01 -6.25293016e-01 6.45422041e-01 6.50041401e-01
-2.53791571e-01 2.37199232e-01 -3.87875259e-01 -2.22155556e-01
-1.31454110e+00 5.82040668e-01 5.85943460e-01 -6.73933744e-01
-5.48234820e-01 8.37921262e-01 2.25617990e-01 -8.88416350e-01
1.60104319e-01 -1.86473683e-01 -5.54671586e-01 -2.44517699e-02
6.05728030e-01 7.11731076e-01 -6.85630023e-01 -6.48193061e-01
-1.72645554e-01 2.15633124e-01 -3.55445683e-01 -2.22508922e-01
1.51472116e+00 -1.80800021e-01 -6.80547431e-02 6.56184793e-01
7.20637083e-01 -3.99162233e-01 -1.60252607e+00 -3.18177909e-01
-2.31269166e-01 -1.71256989e-01 -3.53118569e-01 -7.60270357e-01
-4.26332295e-01 6.16512418e-01 -8.29837024e-02 8.43129337e-01
2.21343383e-01 8.20536762e-02 6.76007032e-01 6.31864786e-01
4.31451470e-01 -7.72931516e-01 2.04241246e-01 9.87694383e-01
6.60305679e-01 -1.39263165e+00 -1.12355851e-01 -3.17447901e-01
-6.71127021e-01 7.64150679e-01 9.22265589e-01 -3.17582816e-01
8.50236535e-01 4.13656652e-01 -1.95252076e-01 -4.57490087e-01
-1.56829071e+00 -3.78808707e-01 1.05397187e-01 5.95961869e-01
1.24607027e-01 3.01846147e-01 2.30539992e-01 4.57849026e-01
6.44390434e-02 -2.64208347e-01 7.36809433e-01 5.29012263e-01
-1.53952196e-01 -1.15374553e+00 1.05220273e-01 7.47647822e-01
5.38863130e-02 -3.04405391e-02 -3.90728563e-01 7.20180869e-01
-4.01578069e-01 1.03971434e+00 3.65993381e-01 -5.68080366e-01
9.95477736e-02 1.12510480e-01 5.17324328e-01 -5.50287664e-01
-7.96584427e-01 -3.74484032e-01 4.12432790e-01 -5.50142109e-01
-2.35001385e-01 -7.33367980e-01 -1.20699775e+00 -9.01919603e-01
5.25643937e-02 -4.65447791e-02 2.05033664e-02 9.86064374e-01
6.63191557e-01 1.02935338e+00 2.47126833e-01 -1.22372341e+00
-2.28642836e-01 -1.14002311e+00 -3.01481426e-01 5.06682694e-01
5.54169953e-01 -1.07228959e+00 -1.81149855e-01 -2.63660610e-01]
|
[5.819774627685547, 0.8505988717079163]
|
9141c5a4-1f16-4454-952b-62847e0bcede
|
arabisc-context-sensitive-neural-spelling
| null | null |
https://aclanthology.org/2020.nlptea-1.2
|
https://aclanthology.org/2020.nlptea-1.2.pdf
|
Arabisc: Context-Sensitive Neural Spelling Checker
|
Traditional statistical approaches to spelling correction usually consist of two consecutive processes — error detection and correction — and they are generally computationally intensive. Current state-of-the-art neural spelling correction models usually attempt to correct spelling errors directly over an entire sentence, which, as a consequence, lacks control of the process, e.g. they are prone to overcorrection. In recent years, recurrent neural networks (RNNs), in particular long short-term memory (LSTM) hidden units, have proven increasingly popular and powerful models for many natural language processing (NLP) problems. Accordingly, we made use of a bidirectional LSTM language model (LM) for our context-sensitive spelling detection and correction model which is shown to have much control over the correction process. While the use of LMs for spelling checking and correction is not new to this line of NLP research, our proposed approach makes better use of the rich neighbouring context, not only from before the word to be corrected, but also after it, via a dual-input deep LSTM network. Although in theory our proposed approach can be applied to any language, we carried out our experiments on Arabic, which we believe adds additional value given the fact that there are limited linguistic resources readily available in Arabic in comparison to many languages. Our experimental results demonstrate that the proposed methods are effective in both improving the quality of correction suggestions and minimising overcorrection.
|
['Andy Way', 'Rejwanul Haque', 'Yasmin Moslem']
|
2020-12-01
| null | null | null | null |
['spelling-correction']
|
['natural-language-processing']
|
[ 4.97360021e-01 -1.83985934e-01 3.32568549e-02 -1.06988572e-01
-5.62937498e-01 -3.59948158e-01 5.74897945e-01 7.53240526e-01
-8.75286460e-01 7.62407184e-01 1.64697453e-01 -6.71228468e-01
4.47999090e-02 -7.65824080e-01 -7.32389688e-01 -4.50675577e-01
2.58836001e-01 1.61477536e-01 1.53404489e-01 -4.25289571e-01
7.86620975e-01 5.06630063e-01 -1.40311956e+00 2.51739502e-01
1.24283576e+00 5.06519973e-01 4.98589218e-01 6.57960832e-01
-4.32761759e-01 9.43186402e-01 -7.05836177e-01 -3.89887601e-01
-2.89472759e-01 -5.26798069e-01 -8.88108790e-01 -2.48652831e-01
2.07165644e-01 6.23335242e-02 3.55229340e-02 1.26000643e+00
3.64686280e-01 3.71114701e-01 3.94746780e-01 -4.77929056e-01
-1.05332160e+00 6.48619056e-01 -2.88788825e-01 4.45024341e-01
2.78568476e-01 -2.61065304e-01 8.79005790e-01 -1.24894547e+00
3.68818581e-01 8.23319972e-01 8.18667650e-01 7.02536702e-01
-6.76278055e-01 -3.47837895e-01 2.08931431e-01 3.30299705e-01
-1.17634857e+00 -3.62047344e-01 4.44176108e-01 -1.54162392e-01
1.38675821e+00 3.61672938e-01 4.52737629e-01 8.69861484e-01
6.29558682e-01 8.69780481e-01 1.05189204e+00 -1.18915772e+00
3.09813581e-03 2.58946776e-01 2.66123444e-01 4.41296607e-01
1.94482237e-01 -7.26245046e-02 -5.59146881e-01 2.36368448e-01
5.38040519e-01 -1.54439928e-02 -2.54524499e-01 2.17143774e-01
-1.07366979e+00 6.77919149e-01 3.73832524e-01 1.00386095e+00
-5.13368011e-01 2.27748286e-02 4.76549685e-01 2.73910791e-01
5.86141407e-01 3.82369787e-01 -4.77868974e-01 -2.56546408e-01
-1.29276061e+00 9.11077932e-02 5.46027720e-01 6.42498136e-01
4.34972942e-01 1.74781561e-01 -1.89949214e-01 9.97153580e-01
2.24356979e-01 4.19569075e-01 8.91707063e-01 -1.93727061e-01
6.67304575e-01 5.71813881e-01 2.86331654e-01 -1.33122694e+00
-3.47958148e-01 -5.54409385e-01 -9.47936475e-01 -3.88449281e-02
4.27189857e-01 2.08651032e-02 -9.10654247e-01 1.48052692e+00
-2.75789827e-01 -4.53487411e-02 3.78723145e-02 7.59738445e-01
3.70578825e-01 8.00590694e-01 1.34838775e-01 -4.70611840e-01
1.09624708e+00 -9.41581309e-01 -1.05003202e+00 -5.66417158e-01
8.66769850e-01 -1.06181288e+00 1.11233985e+00 4.32855159e-01
-1.30446458e+00 -3.79238605e-01 -1.00873661e+00 -1.90151230e-01
-6.43001139e-01 2.78884590e-01 3.18742692e-01 5.50409734e-01
-1.07158911e+00 9.06930387e-01 -7.13227272e-01 -4.83417779e-01
-1.04229070e-01 2.29412317e-01 -1.12128563e-01 3.47726862e-03
-1.57218814e+00 1.35780835e+00 3.50120217e-01 8.03849220e-01
-2.22367391e-01 -2.13862300e-01 -7.48674393e-01 2.24395856e-01
3.27714175e-01 -3.01041573e-01 1.48169231e+00 -1.31870878e+00
-1.53085196e+00 6.48977220e-01 -5.88989437e-01 -6.51886582e-01
3.23346525e-01 -4.47879761e-01 -6.56405449e-01 -2.98477262e-01
-2.84610420e-01 1.00048579e-01 7.39486158e-01 -8.74803424e-01
-7.65188754e-01 -2.55466878e-01 -3.57392460e-01 1.54855371e-01
-5.11314392e-01 5.46348751e-01 -7.48946592e-02 -1.04685760e+00
1.40721440e-01 -7.55479693e-01 -1.62305608e-01 -5.88861525e-01
-1.18256286e-01 -2.52356559e-01 3.35338414e-01 -1.04886639e+00
1.91817605e+00 -1.78066838e+00 1.99804112e-01 2.15238571e-01
-3.22836369e-01 1.02283120e+00 -7.63175786e-02 6.83587015e-01
-1.09513417e-01 2.80865282e-01 -4.44223911e-01 -4.67770845e-01
-1.18882455e-01 1.53455973e-01 -4.69611883e-01 3.55123460e-01
1.38483956e-01 7.62894511e-01 -9.49823737e-01 -1.28210321e-01
1.86752066e-01 5.55643976e-01 -1.96246102e-01 5.47238346e-03
-1.29402474e-01 -2.09442042e-02 5.87385744e-02 3.47658932e-01
3.90044004e-01 1.15087077e-01 -2.25865133e-02 4.60661292e-01
-4.70590353e-01 7.08441019e-01 -1.18504512e+00 1.37054491e+00
-7.63783574e-01 6.61942184e-01 -3.15450996e-01 -7.48882771e-01
9.59604859e-01 4.76655334e-01 -1.99041978e-01 -9.30335939e-01
8.08785632e-02 6.04611695e-01 -7.88932443e-02 -3.28228503e-01
1.08585799e+00 -5.98783381e-02 2.12099463e-01 6.33605182e-01
-1.64635912e-01 2.36967906e-01 3.10217917e-01 9.13058296e-02
7.06177890e-01 4.33745570e-02 6.17174685e-01 -6.83635473e-02
8.88945282e-01 -2.04998881e-01 5.51836014e-01 8.57128084e-01
-6.38194829e-02 4.09404993e-01 1.97290927e-02 -3.08470279e-01
-1.05253792e+00 -4.48067039e-01 1.24932863e-01 1.15964639e+00
-3.31358731e-01 -5.05938947e-01 -9.67808425e-01 -3.33179653e-01
-4.33811605e-01 1.08826983e+00 -4.81458247e-01 -1.34564102e-01
-9.22984004e-01 -6.34240925e-01 6.64010346e-01 5.72060049e-01
2.43303716e-01 -1.64108598e+00 -5.12865901e-01 6.42363548e-01
-1.22301936e-01 -7.28892803e-01 -5.24461746e-01 2.61127651e-01
-1.10913289e+00 -7.83626974e-01 -8.72405052e-01 -8.24932575e-01
5.82925439e-01 3.64376843e-01 8.47549260e-01 5.53092957e-01
2.89755404e-01 2.76308064e-03 -6.13776326e-01 -5.85304558e-01
-6.89387858e-01 1.96064755e-01 1.20399982e-01 -6.57720044e-02
7.14304805e-01 -2.65020818e-01 -2.30381161e-01 -5.68147451e-02
-1.00026488e+00 -6.68165684e-02 7.41437197e-01 8.23054850e-01
4.05434817e-01 -1.33780703e-01 4.45361942e-01 -1.05208349e+00
1.04035997e+00 -1.82695940e-01 -5.31162322e-01 4.78505224e-01
-9.07154739e-01 -4.35862727e-02 7.84508169e-01 -6.82468340e-02
-1.09269941e+00 -3.42240155e-01 -4.04820681e-01 1.17598839e-01
-2.03205198e-01 9.45868254e-01 2.68627107e-01 -1.51897192e-01
5.86341858e-01 5.18873036e-01 -2.81904280e-01 -6.41914070e-01
2.49718572e-03 8.09325159e-01 5.00054717e-01 -2.35640211e-03
4.95043159e-01 -1.68029830e-01 -2.85286248e-01 -8.66978943e-01
-7.20573783e-01 -2.97977924e-01 -8.06545436e-01 -1.14982128e-01
5.27763426e-01 -4.29443836e-01 -5.36527038e-01 7.61688650e-01
-1.41843569e+00 -6.30621836e-02 1.43132567e-01 3.56904507e-01
-1.95711270e-01 6.51685297e-01 -9.44221854e-01 -9.70049739e-01
-6.50340259e-01 -9.45291162e-01 5.60426474e-01 3.08311939e-01
-3.63333225e-01 -1.27806234e+00 -2.94200704e-02 2.00049509e-03
6.64974630e-01 -2.49129236e-01 1.05167592e+00 -6.03765368e-01
-1.38727993e-01 -3.67283762e-01 -5.24643734e-02 6.45374179e-01
1.48170725e-01 7.17046931e-02 -7.97111094e-01 -2.19737917e-01
2.13478237e-01 1.16247937e-01 9.22098756e-01 1.95502058e-01
9.94015872e-01 -4.18900549e-01 -3.46767344e-02 5.81845939e-02
1.31638908e+00 1.74846411e-01 8.11717689e-01 7.08862126e-01
6.81713581e-01 4.67602283e-01 7.10239828e-01 3.04384917e-01
1.84484884e-01 6.18747413e-01 2.42987216e-01 3.37495841e-02
-2.47471333e-02 -2.83764839e-01 5.93190134e-01 1.54952943e+00
-1.68485045e-01 -5.58884025e-01 -1.08350492e+00 6.70248389e-01
-1.82589960e+00 -9.37192857e-01 -4.39261317e-01 2.42064309e+00
9.56225395e-01 1.43527344e-01 -3.95615190e-01 6.24076724e-01
7.81414807e-01 8.12965557e-02 8.59429985e-02 -1.22513449e+00
-7.23838061e-02 3.11202943e-01 5.85490167e-01 7.54395902e-01
-9.14916813e-01 1.22399414e+00 5.86548328e+00 8.45671833e-01
-1.38603222e+00 1.14249792e-02 1.03407651e-01 2.04886854e-01
-3.86693448e-01 -2.33190760e-01 -8.52800965e-01 5.05791008e-01
1.19957101e+00 4.44028489e-02 5.16378760e-01 4.90984321e-01
6.39701128e-01 -3.76793504e-01 -8.23306024e-01 6.85145199e-01
2.61145383e-01 -1.21447086e+00 2.01513186e-01 -3.69355023e-01
6.13217711e-01 -1.84396639e-01 -1.26901856e-02 7.50510246e-02
-1.85142104e-02 -1.09265804e+00 8.63959193e-01 6.19466722e-01
3.59740943e-01 -1.07316649e+00 1.00325644e+00 6.41004086e-01
-6.71734750e-01 7.99664780e-02 -4.99719858e-01 -4.59084034e-01
1.54580474e-01 6.63037479e-01 -7.31981575e-01 4.01311636e-01
5.07566571e-01 6.61900640e-01 -6.77079141e-01 1.11461675e+00
-7.17795491e-01 7.43605614e-01 1.60698250e-01 -3.70246917e-01
4.65794444e-01 -2.72373436e-03 4.93782282e-01 1.55349016e+00
4.75118130e-01 -2.11268999e-02 -3.50635082e-01 4.72092867e-01
1.05737969e-01 4.46781576e-01 -2.77495205e-01 -2.89726645e-01
3.99370313e-01 7.40785480e-01 -6.90238893e-01 -2.63870597e-01
-2.98507601e-01 1.10182929e+00 4.77755070e-01 3.08638364e-01
-4.75914538e-01 -7.07252443e-01 3.47151309e-01 -2.38066509e-01
3.56758595e-01 -4.75190967e-01 -6.03385687e-01 -1.13666391e+00
6.24149628e-02 -9.63833869e-01 1.51388496e-01 -6.51923001e-01
-1.01048052e+00 6.57754362e-01 -4.57260400e-01 -9.56119120e-01
-3.71884406e-01 -5.70469022e-01 -4.63564456e-01 1.25139260e+00
-1.79900610e+00 -8.65308344e-01 1.22544318e-01 3.03404093e-01
7.86813796e-01 2.73271240e-02 1.05162764e+00 4.25977558e-01
-5.34825623e-01 7.10643888e-01 2.09455788e-01 1.19034946e-02
7.59103835e-01 -1.12277377e+00 4.31954503e-01 1.38895786e+00
3.90092075e-01 1.11494529e+00 8.77591729e-01 -7.09297061e-01
-9.80496347e-01 -9.76849675e-01 2.00941706e+00 -2.00528860e-01
6.12235248e-01 -2.50171274e-01 -1.30974519e+00 7.18957603e-01
3.69827747e-01 -4.84177530e-01 4.90778923e-01 8.66828263e-02
5.61447069e-02 8.58377889e-02 -8.32455516e-01 7.23202646e-01
5.95242441e-01 -6.10565305e-01 -9.43532467e-01 2.04026088e-01
4.97317702e-01 -3.67656261e-01 -3.70750010e-01 2.00435087e-01
2.07649752e-01 -9.49467838e-01 4.10009861e-01 -4.75637287e-01
4.49153155e-01 -2.95636177e-01 1.17874600e-01 -1.61501181e+00
-3.38592678e-01 -4.77404773e-01 2.03605201e-02 1.42683721e+00
4.85606939e-01 -4.80878979e-01 2.67637730e-01 3.90730679e-01
-3.44984829e-01 -5.71103334e-01 -6.95237398e-01 -4.63570595e-01
2.24465117e-01 -7.92945504e-01 4.23699707e-01 8.71109605e-01
1.21182561e-01 -6.70087943e-03 -6.99512243e-01 8.75953063e-02
1.11769736e-01 -1.57341927e-01 1.16460532e-01 -9.28623915e-01
-1.19393341e-01 -5.45008063e-01 1.09307356e-01 -7.61657536e-01
1.83028251e-01 -7.44957745e-01 3.96841884e-01 -1.49094284e+00
-1.83591291e-01 -3.14011544e-01 -5.33156455e-01 5.47530890e-01
-3.45631033e-01 2.09538475e-01 2.84063607e-01 2.13727474e-01
-2.41636246e-01 3.78493428e-01 9.25530374e-01 1.00618601e-02
-2.88408250e-01 2.79770315e-01 -3.98105949e-01 8.90293777e-01
8.16169739e-01 -6.56011522e-01 8.35139528e-02 -9.27628040e-01
6.81466341e-01 -2.06568897e-01 3.71103436e-02 -8.33779335e-01
5.28293371e-01 -1.37311384e-01 2.33644545e-01 -3.75535339e-01
-1.19533874e-01 -6.61759496e-01 -3.09618890e-01 4.90123481e-01
-4.82062489e-01 5.07914901e-01 3.90072793e-01 2.38996297e-01
-3.43218148e-01 -7.57445991e-01 7.14462638e-01 -2.75827825e-01
-8.88859093e-01 -1.58722416e-01 -9.60116506e-01 -3.10605884e-01
5.40582061e-01 -2.26752535e-01 3.85628901e-02 -1.88294381e-01
-5.66145539e-01 -1.68123201e-01 3.78624052e-01 5.49977779e-01
7.24959373e-01 -1.06546903e+00 -5.47304034e-01 2.28802696e-01
-8.35417435e-02 -3.35843623e-01 1.23454802e-01 7.97841847e-01
-7.15295613e-01 9.36830699e-01 -9.58077535e-02 -6.81863278e-02
-1.38299131e+00 5.74085772e-01 2.44707525e-01 -3.26583892e-01
-5.62044084e-01 9.53775287e-01 -5.08087754e-01 -2.95251936e-01
4.57487941e-01 -5.85884750e-01 -5.47734797e-01 1.70808211e-01
9.07211065e-01 4.17902082e-01 6.35272503e-01 -7.19267428e-01
-3.51341784e-01 3.75407785e-01 -3.32575381e-01 -6.26409054e-02
1.24701524e+00 -3.92334908e-01 -5.20461977e-01 7.21575797e-01
5.44789791e-01 2.90878326e-01 -3.92287433e-01 -3.45171183e-01
6.24599576e-01 -4.44378048e-01 2.02310815e-01 -1.05931711e+00
-6.48554027e-01 1.15063787e+00 2.04336256e-01 1.11780599e-01
1.09740150e+00 -8.01358283e-01 1.10057175e+00 4.86036658e-01
2.98545867e-01 -1.37462473e+00 -3.24971229e-01 1.11387336e+00
8.15877438e-01 -9.52191055e-01 -2.88690180e-01 -2.95590907e-02
-4.25250500e-01 1.41625750e+00 3.25407267e-01 -1.40690967e-01
1.29477724e-01 1.10571511e-01 2.14603335e-01 2.05813780e-01
-6.56303883e-01 1.59956232e-01 2.37678185e-01 2.68015712e-01
9.15320456e-01 -7.24285394e-02 -8.80517185e-01 4.84230280e-01
-8.77268240e-02 2.27370620e-01 8.91086459e-01 1.00597715e+00
-6.47909760e-01 -1.40310788e+00 -5.84237814e-01 1.55845657e-01
-6.74354732e-01 -5.20091534e-01 -3.98702681e-01 5.04575491e-01
9.59843844e-02 1.03064954e+00 -1.24778643e-01 -1.23008847e-01
2.60140985e-01 1.50590897e-01 3.33877683e-01 -6.72330618e-01
-1.08648503e+00 -2.53681868e-01 -1.99990407e-01 -3.93579215e-01
-3.24796736e-01 -7.49576151e-01 -1.26196253e+00 -4.83288199e-01
-4.85866547e-01 2.96479106e-01 8.19878876e-01 1.38606894e+00
1.45077869e-01 5.58129370e-01 2.87014931e-01 -7.06313550e-01
-6.75415635e-01 -1.19798279e+00 -5.12300670e-01 1.88019313e-02
3.97265255e-01 -2.76875317e-01 -1.46135554e-01 -1.84926778e-01]
|
[10.950892448425293, 10.736434936523438]
|
7413abbe-3753-47b3-ad1c-79ca6fd5fd4c
|
cilex-an-investigation-of-context-information
| null | null |
https://aclanthology.org/2022.coling-1.362
|
https://aclanthology.org/2022.coling-1.362.pdf
|
CILex: An Investigation of Context Information for Lexical Substitution Methods
|
Lexical substitution, which aims to generate substitutes for a target word given a context, is an important natural language processing task useful in many applications. Due to the paucity of annotated data, existing methods for lexical substitution tend to rely on manually curated lexical resources and contextual word embedding models. Methods based on lexical resources are likely to miss relevant substitutes whereas relying only on contextual word embedding models fails to provide adequate information on the impact of a substitute in the entire context and the overall meaning of the input. We proposed CILex, which uses contextual sentence embeddings along with methods that capture additional context information complimenting contextual word embeddings for lexical substitution. This ensured the semantic consistency of a substitute with the target word while maintaining the overall meaning of the sentence. Our experimental comparisons with previously proposed methods indicated that our solution is now the state-of-the-art on both the widely used LS07 and CoInCo datasets with P@1 scores of 55.96% and 57.25% for lexical substitution. The implementation of the proposed approach is available at https://github.com/sandaruSen/CILex under the MIT license.
|
['Hanna Suominen', 'Artem Lenskiy', 'Elena Daskalaki', 'Sandaru Seneviratne']
| null | null | null | null |
coling-2022-10
|
['sentence-embeddings', 'sentence-embeddings']
|
['methodology', 'natural-language-processing']
|
[ 2.19028875e-01 -3.34013142e-02 -2.04864830e-01 -1.95246562e-01
-4.92526025e-01 -2.93335348e-01 6.46084189e-01 6.06265247e-01
-8.48164439e-01 7.18432248e-01 7.27273941e-01 -2.61868060e-01
1.87187240e-01 -7.26498485e-01 -4.41047579e-01 -4.09207493e-01
5.91383755e-01 1.71429351e-01 2.81100720e-01 -6.69540584e-01
5.05929649e-01 1.64908335e-01 -1.61431599e+00 5.07946074e-01
7.80206025e-01 6.07505560e-01 5.23180962e-01 3.36281329e-01
-6.17434680e-01 2.14400977e-01 -6.42259061e-01 -7.13569939e-01
1.92633688e-01 -3.86767626e-01 -6.60254657e-01 -4.67566848e-01
5.31467021e-01 2.20973313e-01 -1.84341475e-01 1.09227157e+00
7.86019981e-01 1.85726881e-01 3.14034849e-01 -9.00239885e-01
-1.21559179e+00 6.94949210e-01 -1.13041224e-02 4.21058029e-01
5.53613484e-01 8.81345570e-02 1.49634862e+00 -1.52372301e+00
7.36735284e-01 1.09600532e+00 7.53213525e-01 6.70203269e-01
-1.14820957e+00 -4.71781045e-01 1.53733790e-01 4.42325681e-01
-1.26138747e+00 -3.92106831e-01 6.98182881e-01 -1.33986682e-01
1.57717907e+00 5.13406634e-01 7.48565078e-01 1.22085416e+00
4.62802321e-01 4.18829381e-01 7.91655302e-01 -9.67567861e-01
-3.89451124e-02 3.18666011e-01 3.78568232e-01 2.36372486e-01
3.69402170e-01 -6.43271506e-02 -8.05161655e-01 -2.09096313e-01
3.54171768e-02 -7.80488551e-02 -2.02690676e-01 -5.57852238e-02
-1.03495526e+00 7.90277898e-01 4.40485716e-01 5.60620308e-01
-5.14268875e-01 2.37775162e-01 6.96336210e-01 2.44591787e-01
6.49629772e-01 6.03930473e-01 -4.82392907e-01 -5.94009347e-02
-7.00961173e-01 5.40945649e-01 4.14754897e-01 9.67337847e-01
5.96318781e-01 7.48443529e-02 -2.40427271e-01 9.13969755e-01
2.24129319e-01 2.98462570e-01 8.88692498e-01 -4.14412588e-01
4.10041273e-01 6.74691737e-01 1.96893066e-01 -9.57227290e-01
-1.26097202e-01 -2.61783808e-01 -3.55852932e-01 -3.06705892e-01
2.46019028e-02 1.65350541e-01 -7.79862881e-01 1.81048214e+00
3.18819851e-01 2.48709574e-01 1.03149928e-01 7.60060549e-01
1.00506806e+00 6.31349564e-01 3.18665147e-01 -1.01654507e-01
1.41375697e+00 -1.01677597e+00 -8.97864819e-01 -5.66871941e-01
8.05213988e-01 -1.32482255e+00 1.73805821e+00 -1.52219206e-01
-1.02580404e+00 -6.15728736e-01 -1.20491552e+00 -3.71001840e-01
-7.51075089e-01 -1.33474708e-01 2.36253768e-01 7.32751548e-01
-1.09546876e+00 4.34231251e-01 -3.15424591e-01 -6.32351696e-01
-6.83034584e-02 4.22861204e-02 -4.08269972e-01 -2.50675470e-01
-1.61196041e+00 1.39820325e+00 4.33701098e-01 -3.53635708e-03
2.48604082e-02 -8.42149198e-01 -9.70199645e-01 6.77750185e-02
2.96919256e-01 -7.08564103e-01 1.07589066e+00 -8.45228732e-01
-1.12564576e+00 7.21404493e-01 -5.47663808e-01 -5.09021163e-01
2.27187842e-01 -5.49215496e-01 -5.23762941e-01 -1.81834877e-01
3.52713138e-01 6.38941228e-01 5.92369616e-01 -8.24489057e-01
-4.96191353e-01 -1.23157948e-01 -5.52855395e-02 3.63291264e-01
-4.57179815e-01 1.41524345e-01 -1.83989003e-01 -1.11103570e+00
-1.36822641e-01 -9.03237998e-01 -1.12349063e-01 -1.62239835e-01
-2.34606769e-02 -3.73046517e-01 5.82850993e-01 -7.03306913e-01
1.66375041e+00 -2.21353936e+00 2.30731472e-01 -1.68080017e-01
-1.87789217e-01 5.58518589e-01 -2.45203599e-01 1.13030744e+00
-2.78584182e-01 3.42852354e-01 -3.08434159e-01 -4.08764929e-01
1.77898690e-01 1.98562816e-01 -5.57571828e-01 2.67997354e-01
3.07943404e-01 8.75878870e-01 -9.67438817e-01 -1.59048110e-01
3.55364591e-01 5.51343322e-01 -6.23762012e-01 1.49113327e-01
-6.58664778e-02 -6.93852305e-02 2.40699708e-01 2.47123450e-01
3.90754908e-01 5.01001179e-01 2.01328009e-01 -2.37763777e-01
-7.45151043e-02 7.19925165e-01 -1.11185861e+00 1.50355542e+00
-7.58316398e-01 4.02188122e-01 -4.42084700e-01 -6.70920789e-01
8.88410509e-01 4.59744126e-01 -3.55339199e-02 -8.05306315e-01
3.46106961e-02 6.11265659e-01 5.35390601e-02 -4.87688601e-01
1.01878655e+00 -4.22856718e-01 -3.15302968e-01 2.95443475e-01
1.36275500e-01 -3.12521726e-01 2.41636604e-01 1.60104871e-01
1.06317925e+00 1.09855637e-01 7.90289402e-01 -4.78622198e-01
5.85288107e-01 -2.30629053e-02 5.18682599e-01 4.82410789e-01
-2.24165812e-01 5.87386727e-01 1.17817819e-01 -4.73807395e-01
-1.33078885e+00 -9.83257174e-01 -1.64232813e-02 1.14294314e+00
1.54584333e-01 -9.23083127e-01 -5.30318081e-01 -4.02077258e-01
1.19497376e-02 1.44302380e+00 -6.02815092e-01 -4.17384416e-01
-6.92516267e-01 -5.05811572e-01 4.36079115e-01 3.96471173e-01
1.17491744e-02 -1.44442904e+00 -5.23102343e-01 4.04216081e-01
-1.29518226e-01 -8.74414504e-01 -7.42503881e-01 2.46905908e-01
-5.75006664e-01 -9.27696466e-01 -2.59208173e-01 -9.29524124e-01
4.75978374e-01 2.78775513e-01 1.11930525e+00 3.89024585e-01
-2.49104217e-01 3.68764587e-02 -7.34674335e-01 -3.14837098e-01
-4.36373889e-01 -4.79870923e-02 2.83053458e-01 -2.72274822e-01
7.28604376e-01 -3.91554624e-01 -3.67498994e-01 -4.96926643e-02
-1.19832265e+00 -5.81883453e-02 2.91451156e-01 1.23159289e+00
5.78748703e-01 -3.50330681e-01 7.27283776e-01 -1.19139814e+00
9.83904183e-01 -6.30207062e-01 3.32503244e-02 1.47737876e-01
-8.04900587e-01 1.10389151e-01 7.23525405e-01 -3.23396981e-01
-9.61440384e-01 -1.90905780e-01 -5.62624454e-01 7.14120567e-02
2.48774588e-02 5.92072606e-01 -7.63356015e-02 5.68976581e-01
6.98309362e-01 1.26123220e-01 -2.95779914e-01 -4.23130125e-01
5.76187313e-01 6.31258428e-01 3.37435842e-01 -5.40815055e-01
3.73281449e-01 4.34986874e-02 -3.90340358e-01 -7.91412473e-01
-6.02027774e-01 -5.62857151e-01 -6.13985240e-01 1.15972362e-01
7.06548929e-01 -6.32498980e-01 2.74197608e-01 7.60605559e-02
-1.40790510e+00 8.93216804e-02 -4.36879545e-01 2.12054074e-01
-1.81766465e-01 5.56140304e-01 -2.73748249e-01 -5.10541141e-01
-5.51146984e-01 -1.04320037e+00 8.61625493e-01 -1.84261531e-01
-9.67774034e-01 -1.04449606e+00 1.24157779e-01 1.65940225e-01
5.82587123e-01 -1.26939833e-01 1.16772723e+00 -7.44534552e-01
1.31537691e-01 -4.44753110e-01 4.59344536e-02 3.61980915e-01
3.78259301e-01 -2.99938601e-02 -7.40348399e-01 9.50395502e-03
-6.14653639e-02 -7.65834143e-03 7.61663020e-01 5.06913709e-03
6.10829949e-01 -2.95688838e-01 2.22805678e-03 4.67980430e-02
1.43760860e+00 -7.66687244e-02 7.02300251e-01 4.46935922e-01
4.92348135e-01 6.80808723e-01 7.32539237e-01 3.60752761e-01
1.83055520e-01 7.83989072e-01 3.77031147e-01 3.72221619e-01
-4.34466481e-01 -4.03032035e-01 3.76745373e-01 1.11010551e+00
6.53724313e-01 -3.78474087e-01 -1.00073862e+00 1.01572955e+00
-1.89347434e+00 -8.11621130e-01 -3.00894409e-01 2.14979148e+00
1.09727705e+00 3.38306189e-01 -3.81469727e-01 2.76091993e-01
8.42421234e-01 2.43880436e-01 -1.07433319e-01 -1.00923145e+00
-2.53094375e-01 6.49149537e-01 3.39692265e-01 9.08472121e-01
-6.99000955e-01 1.36108112e+00 5.57856131e+00 8.78438771e-01
-9.41769779e-01 3.76746923e-01 4.56056222e-02 -1.20721936e-01
-7.47125328e-01 8.49396214e-02 -7.93203771e-01 5.33904910e-01
1.05970263e+00 -4.86759722e-01 3.11754435e-01 5.67877948e-01
2.95165509e-01 -6.21122643e-02 -9.85574067e-01 8.15213859e-01
3.36234003e-01 -1.34245014e+00 1.92660004e-01 -4.88524467e-01
5.86522162e-01 -1.54686302e-01 1.20284699e-01 3.05040270e-01
-9.64979231e-02 -9.08685148e-01 1.01204336e+00 2.59682387e-01
5.30378282e-01 -7.27723181e-01 1.07940412e+00 2.24162862e-01
-1.09718251e+00 1.90217152e-01 -4.44589794e-01 -5.03261149e-01
1.49960071e-01 4.56682801e-01 -7.82079041e-01 3.22870195e-01
5.86503685e-01 5.61704814e-01 -6.49657607e-01 6.48521006e-01
-6.31990194e-01 4.78808403e-01 -1.33090347e-01 -3.45089465e-01
2.28962034e-01 1.14232838e-01 6.64828897e-01 1.45966160e+00
3.87288541e-01 -3.41183752e-01 -4.16083299e-02 4.59218800e-01
-1.18097760e-01 7.00673580e-01 -7.97688961e-01 3.76466587e-02
7.98266053e-01 8.61787677e-01 -4.15142745e-01 -3.52664083e-01
-4.83077317e-01 1.12658453e+00 4.77485687e-01 -1.13874516e-02
-8.62713754e-01 -4.29671913e-01 9.29543853e-01 1.79691657e-01
4.12566602e-01 -1.48497969e-01 -6.03532016e-01 -1.00573981e+00
3.93361598e-01 -7.49845564e-01 2.09027350e-01 -7.32759774e-01
-1.40360749e+00 7.88877904e-01 9.34855938e-02 -1.08101797e+00
6.49380591e-03 -6.43770039e-01 -7.28001297e-01 1.18805540e+00
-1.39351273e+00 -1.01560044e+00 1.49964783e-02 1.06237214e-02
1.00609863e+00 -1.72446281e-01 1.08426690e+00 3.33746850e-01
-3.65539581e-01 6.46116912e-01 -1.29742324e-01 -4.47799206e-01
9.74906027e-01 -1.09015501e+00 5.68866730e-01 1.18822110e+00
2.56249785e-01 9.72510755e-01 1.10020638e+00 -7.90471733e-01
-1.13847971e+00 -1.10161281e+00 1.76693308e+00 -4.52950537e-01
8.94659936e-01 -2.79350907e-01 -1.07079756e+00 4.48619097e-01
7.28012025e-01 -1.01924747e-01 9.97182488e-01 -1.12789884e-01
-4.60081220e-01 1.34536773e-01 -1.03462613e+00 1.08733785e+00
1.12858474e+00 -5.75032294e-01 -1.13172412e+00 1.07535094e-01
1.09240115e+00 -2.88574606e-01 -3.59639496e-01 1.45231992e-01
3.90684992e-01 -7.50356674e-01 7.78686047e-01 -6.22509003e-01
5.66541791e-01 -2.88248569e-01 -5.89018285e-01 -1.52180886e+00
-3.25317323e-01 -1.75130203e-01 1.36550769e-01 1.28720582e+00
5.65874517e-01 -6.55820251e-01 3.28600794e-01 5.92366755e-01
-4.39893395e-01 -8.45771670e-01 -1.18954635e+00 -6.24700963e-01
2.11904049e-01 -7.32163072e-01 7.02657998e-01 7.89429128e-01
2.14199582e-03 5.75866282e-01 -2.71824270e-01 -1.55083109e-02
6.23877570e-02 -2.40024745e-01 4.29162920e-01 -6.69473827e-01
-3.34883742e-02 -4.98364776e-01 -3.85652184e-01 -4.96981144e-01
6.12710297e-01 -1.16703415e+00 -8.85483399e-02 -1.53318810e+00
-1.79257900e-01 -2.06636474e-01 -3.85546744e-01 3.77482712e-01
-5.95215559e-01 4.76760685e-01 2.87128776e-01 -3.07720266e-02
-8.94318819e-02 8.05109978e-01 6.00897014e-01 -1.12592623e-01
-5.10001257e-02 -5.60758173e-01 -8.06727946e-01 5.11956632e-01
1.00384414e+00 -6.73588932e-01 -3.69105935e-01 -6.87364161e-01
4.88231510e-01 -7.39845335e-01 1.97567597e-01 -6.33592188e-01
2.54025478e-02 -1.43279836e-01 -2.91216876e-02 -2.03303263e-01
4.52438861e-01 -9.37672138e-01 2.41866767e-01 5.84529221e-01
-4.33423579e-01 6.64498210e-01 4.18220878e-01 3.72765005e-01
-2.78040826e-01 -6.77694619e-01 6.60231233e-01 -3.13120112e-02
-1.04450047e+00 -2.74152964e-01 -3.54580402e-01 1.35181591e-01
9.76964533e-01 -1.41332448e-01 -1.75754666e-01 -4.96352389e-02
-5.21261871e-01 -1.11989744e-01 5.49019873e-01 7.81563342e-01
9.51655507e-01 -1.66392934e+00 -6.79727256e-01 1.99626505e-01
3.94459635e-01 -4.92014974e-01 2.09315140e-02 4.78258193e-01
-7.39368021e-01 3.52126181e-01 -1.55177936e-01 -3.22050564e-02
-1.42357576e+00 6.24175847e-01 1.05248988e-02 -6.85566217e-02
-4.65709209e-01 8.92762184e-01 -2.46608928e-01 -6.30575061e-01
-1.04706474e-01 -4.68260676e-01 -1.60742588e-02 1.14960074e-01
4.39197510e-01 1.34092242e-01 3.31331819e-01 -1.01197755e+00
-5.40645659e-01 2.67282814e-01 -2.45811529e-02 -8.75308365e-02
1.26561654e+00 -1.93113521e-01 -3.40296298e-01 6.20200396e-01
1.07851291e+00 3.31885785e-01 -3.48132759e-01 -4.30030435e-01
3.14685702e-01 -8.01616192e-01 4.66905981e-02 -6.07297838e-01
-5.08712113e-01 7.82662272e-01 2.59165257e-01 3.71983573e-02
9.31343913e-01 -2.61685044e-01 1.01671922e+00 2.11911410e-01
1.72775686e-01 -1.18592477e+00 -3.80167440e-02 6.65512085e-01
1.18748176e+00 -9.85340834e-01 -1.27678970e-02 -5.38045824e-01
-8.11359465e-01 1.06260633e+00 6.27956510e-01 -3.67859215e-01
6.19446993e-01 -4.60889265e-02 1.83405563e-01 2.09549010e-01
-9.64275122e-01 -1.08534560e-01 1.93069592e-01 3.66609573e-01
8.26621950e-01 1.34555727e-01 -9.63270724e-01 7.44648516e-01
-3.37734669e-01 -4.36346859e-01 5.37659883e-01 9.97658312e-01
-3.74010593e-01 -1.70069289e+00 -1.47025660e-01 3.14889193e-01
-3.98551017e-01 -8.32784235e-01 -6.35092735e-01 6.29945934e-01
3.37853998e-01 7.62886405e-01 -1.29278556e-01 -2.76476592e-01
5.29244244e-01 6.05712771e-01 2.83943981e-01 -1.18084264e+00
-8.70876551e-01 -9.96909663e-02 2.29599431e-01 -4.37811404e-01
-2.32059479e-01 -6.49711072e-01 -1.23193026e+00 -4.19735193e-01
-2.21228480e-01 7.75920376e-02 5.36434889e-01 8.65704536e-01
4.63449597e-01 4.02042598e-01 2.08791628e-01 -6.50622070e-01
-4.87079978e-01 -1.13671052e+00 -2.29154840e-01 6.27088249e-01
-1.45761482e-03 -6.51397109e-01 -2.47970149e-01 -7.43968263e-02]
|
[10.69241714477539, 9.201778411865234]
|
0d420517-4213-4ed6-bdd7-dd188e7351ce
|
learning-geometry-aware-representations-by
|
2304.08204
| null |
https://arxiv.org/abs/2304.08204v1
|
https://arxiv.org/pdf/2304.08204v1.pdf
|
Learning Geometry-aware Representations by Sketching
|
Understanding geometric concepts, such as distance and shape, is essential for understanding the real world and also for many vision tasks. To incorporate such information into a visual representation of a scene, we propose learning to represent the scene by sketching, inspired by human behavior. Our method, coined Learning by Sketching (LBS), learns to convert an image into a set of colored strokes that explicitly incorporate the geometric information of the scene in a single inference step without requiring a sketch dataset. A sketch is then generated from the strokes where CLIP-based perceptual loss maintains a semantic similarity between the sketch and the image. We show theoretically that sketching is equivariant with respect to arbitrary affine transformations and thus provably preserves geometric information. Experimental results show that LBS substantially improves the performance of object attribute classification on the unlabeled CLEVR dataset, domain transfer between CLEVR and STL-10 datasets, and for diverse downstream tasks, confirming that LBS provides rich geometric information.
|
['Byoung-Tak Zhang', 'Kibeom Kim', 'Won-Seok Choi', 'Hyunsung Go', 'Inwoo Hwang', 'Hyundo Lee']
|
2023-04-17
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Lee_Learning_Geometry-Aware_Representations_by_Sketching_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Lee_Learning_Geometry-Aware_Representations_by_Sketching_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['semantic-textual-similarity', 'semantic-similarity']
|
['natural-language-processing', 'natural-language-processing']
|
[ 2.67633528e-01 1.03503941e-02 -1.50953576e-01 -7.05646276e-01
-3.28222960e-01 -1.04842389e+00 8.82693410e-01 -3.40541676e-02
-1.56492919e-01 2.99537033e-01 -2.12354716e-02 -9.22702923e-02
-7.81766772e-02 -8.89782250e-01 -1.07171965e+00 -4.04849738e-01
3.41821522e-01 6.14743173e-01 1.59273878e-01 -1.03777079e-02
3.85454178e-01 1.05331433e+00 -1.21486616e+00 4.36966181e-01
5.45056462e-01 7.48726010e-01 2.08007857e-01 8.89970958e-01
-2.72206694e-01 7.33787596e-01 -1.92544609e-01 -6.00233972e-01
5.44609964e-01 -3.23817730e-01 -6.90133333e-01 2.39374876e-01
1.29167366e+00 -4.75801021e-01 -4.78653818e-01 8.42726946e-01
-4.22241688e-02 3.28625709e-01 1.06942892e+00 -1.40520704e+00
-1.05907834e+00 5.82422838e-02 -4.90019798e-01 -3.57967079e-01
-2.55928822e-02 1.33975223e-01 1.17962897e+00 -9.82201576e-01
1.13620234e+00 1.63899064e+00 6.24595940e-01 6.22150064e-01
-1.71369171e+00 -5.55345833e-01 3.45097065e-01 -9.26999003e-02
-1.07207477e+00 -1.72029659e-01 1.09687197e+00 -5.69999516e-01
3.21327716e-01 1.55505553e-01 5.99038541e-01 1.06187117e+00
-1.40198335e-01 9.37273860e-01 8.47502708e-01 -3.11775655e-01
4.40129668e-01 -5.09174354e-02 9.33722258e-02 1.09524977e+00
2.35986724e-01 2.06428692e-02 -5.10276139e-01 -2.22901385e-02
1.17636740e+00 3.91807050e-01 1.58044010e-01 -1.19493139e+00
-1.19660342e+00 8.24369550e-01 7.59011924e-01 -1.04665183e-01
-2.02781465e-02 6.07515335e-01 1.52568057e-01 3.25984865e-01
1.74903348e-01 6.06382310e-01 -1.38132945e-01 3.79276544e-01
-6.95104659e-01 4.00734454e-01 6.34002149e-01 1.22109520e+00
7.13615000e-01 2.63857353e-03 -1.82269797e-01 7.97634959e-01
-1.18669057e-02 8.02907467e-01 -1.46055043e-01 -1.46711397e+00
3.58120948e-01 6.30536497e-01 6.13275953e-02 -1.06163633e+00
-9.30587947e-02 -8.41309652e-02 -7.38218725e-01 7.73882270e-01
6.77009761e-01 4.48615909e-01 -9.43170488e-01 1.93053985e+00
-9.50246006e-02 1.74043253e-01 -2.09519073e-01 7.74223924e-01
5.96266448e-01 3.91646981e-01 2.09388807e-01 5.70344567e-01
1.09232509e+00 -8.68383646e-01 -1.98780119e-01 -2.04746738e-01
2.47195780e-01 -5.30913234e-01 1.41192245e+00 4.00264323e-01
-1.11452198e+00 -7.24667490e-01 -1.24754870e+00 -7.17363417e-01
-5.38318276e-01 4.60684121e-01 5.44223070e-01 5.00152349e-01
-8.50684404e-01 7.00677335e-01 -6.92496359e-01 -5.31528473e-01
1.01022673e+00 -3.22456211e-02 -5.50142050e-01 -4.42224801e-01
-4.26653206e-01 6.71195269e-01 2.46607251e-02 -4.12202507e-01
-8.74586999e-01 -1.03602827e+00 -8.96522820e-01 7.70276263e-02
5.67008667e-02 -9.54759657e-01 1.01737893e+00 -9.33262467e-01
-1.24249601e+00 1.15798104e+00 -3.07391286e-01 -4.31414157e-01
9.68813837e-01 -1.76403254e-01 8.49766359e-02 4.04132247e-01
-1.21642917e-01 1.14506555e+00 1.30140841e+00 -1.59177125e+00
-3.76031458e-01 -4.48370755e-01 2.30482638e-01 1.89676151e-01
-5.92086613e-02 -5.56336343e-01 -3.23976427e-01 -7.79205561e-01
1.33784175e-01 -1.02843833e+00 -6.74350113e-02 1.13623810e+00
-1.79546148e-01 -9.81615707e-02 1.12530804e+00 -4.15208668e-01
3.00555885e-01 -2.11057591e+00 1.69191062e-01 3.32012445e-01
3.18195134e-01 2.66169101e-01 -5.35515308e-01 3.36581200e-01
1.61815122e-01 -1.50637478e-02 -5.45038760e-01 -4.73280996e-01
1.05599836e-01 3.85261804e-01 -1.00680649e+00 4.55963731e-01
3.76815408e-01 1.37291586e+00 -1.14679313e+00 -1.92713425e-01
5.60832322e-01 4.70424473e-01 -7.90520072e-01 3.64051610e-01
-5.86724162e-01 4.62507993e-01 -4.32041317e-01 3.63627940e-01
7.93456018e-01 -1.29356116e-01 -1.88132785e-02 -3.15769494e-01
2.97159076e-01 -1.70881510e-01 -8.82513821e-01 2.14990330e+00
-6.01619005e-01 8.76796782e-01 -1.46313190e-01 -9.36977386e-01
1.08326507e+00 -2.60921210e-01 4.19433005e-02 -6.79385245e-01
-1.98941916e-01 -1.02802567e-01 -4.36197042e-01 -2.55507827e-01
2.31941268e-01 -1.43442422e-01 -6.49773749e-04 5.41193247e-01
7.31930584e-02 -6.74220026e-01 1.39105627e-02 5.80722332e-01
7.70042717e-01 4.59294915e-01 9.18687582e-02 -2.14921281e-01
2.80154020e-01 -1.79636791e-01 1.10587977e-01 9.51093912e-01
-2.84390524e-02 8.85768473e-01 4.77147579e-01 -6.50151849e-01
-1.41092706e+00 -1.85863698e+00 -3.03280801e-02 1.06953979e+00
3.88229907e-01 -7.15381950e-02 -7.10865378e-01 -8.55098486e-01
5.21412969e-01 9.33396101e-01 -8.69472027e-01 -2.97050625e-01
-5.09472728e-01 2.05975845e-01 6.19104922e-01 8.41590524e-01
5.18466830e-01 -9.79068875e-01 -5.93992352e-01 -1.06329747e-01
1.01904884e-01 -1.21605122e+00 -7.82148004e-01 -2.41511181e-01
-9.14213002e-01 -1.22110367e+00 -6.42456889e-01 -7.61520326e-01
1.08013260e+00 5.08757353e-01 1.12011218e+00 1.56341285e-01
-7.92282581e-01 6.97936058e-01 2.13728491e-02 -2.46638328e-01
-3.50795537e-01 -2.15052679e-01 -3.14621061e-01 2.28480846e-01
6.72951415e-02 -6.53533995e-01 -5.15284479e-01 1.61233887e-01
-8.64111304e-01 3.33903044e-01 5.43736994e-01 8.54231358e-01
6.31429672e-01 -4.75680143e-01 4.87442553e-01 -1.15434456e+00
2.72389650e-01 4.77192402e-02 -6.80647790e-01 4.09755528e-01
-3.37267518e-01 5.89710355e-01 7.82644749e-01 -3.87569726e-01
-1.24487233e+00 4.33148235e-01 2.98340559e-01 -7.35705853e-01
-4.05905157e-01 -3.46266359e-01 -1.50532752e-01 -2.53934264e-01
5.43212712e-01 2.46475697e-01 5.15248161e-03 -5.44950604e-01
1.21791816e+00 1.67079166e-01 9.89218831e-01 -1.06358755e+00
1.08986580e+00 1.17263126e+00 4.59519088e-01 -9.48337436e-01
-9.05033350e-01 -3.23004395e-01 -1.21092188e+00 -3.83492559e-02
9.61786687e-01 -5.64560115e-01 -7.25711048e-01 1.32062465e-01
-1.24590671e+00 -3.81287485e-01 -5.22296786e-01 9.25833061e-02
-9.95680213e-01 1.65573880e-01 -2.63258100e-01 -6.39763713e-01
-1.61352642e-02 -7.90964901e-01 1.21217179e+00 1.84457228e-02
-2.89696455e-01 -1.01829123e+00 -6.31439686e-02 2.17158467e-01
2.27171436e-01 3.90231162e-01 1.42933381e+00 -1.96341783e-01
-9.50884044e-01 -4.88914400e-02 -7.11202919e-01 3.42836618e-01
1.13908341e-02 -3.65034342e-02 -1.12409449e+00 -2.28410244e-01
-5.72215676e-01 -6.69969201e-01 1.06613958e+00 1.73085675e-01
1.61317468e+00 -2.07713872e-01 -1.65557131e-01 9.24342215e-01
1.42219627e+00 4.51457798e-02 5.44709861e-01 -1.56487346e-01
9.62956727e-01 6.15765929e-01 2.26539671e-01 4.91908938e-02
1.12462521e-01 4.85480845e-01 3.28341752e-01 -2.57257044e-01
-6.40421450e-01 -9.12285745e-01 1.77878942e-02 2.90645063e-01
8.56062993e-02 2.85507794e-02 -4.42144006e-01 4.45974499e-01
-1.70444429e+00 -1.17209244e+00 1.70490518e-01 2.26412988e+00
5.73544800e-01 -2.63251420e-02 -1.28875270e-01 -1.20728515e-01
4.31869000e-01 2.05239788e-01 -9.57281351e-01 -4.54549789e-01
-1.32374316e-01 3.46374035e-01 4.48912948e-01 6.07642293e-01
-9.75953639e-01 1.11810446e+00 6.28058577e+00 4.59785193e-01
-1.01833999e+00 -3.52291673e-01 5.97259700e-01 9.90283638e-02
-3.53367805e-01 -1.05271459e-01 -3.19001019e-01 -2.87636276e-02
1.61210805e-01 -1.37080029e-01 7.38643527e-01 9.72792864e-01
-2.66478267e-02 3.36813122e-01 -1.72571683e+00 1.01181912e+00
2.20467687e-01 -1.67285538e+00 7.48466611e-01 -1.08676657e-01
8.63902271e-01 -2.45808586e-01 4.64600623e-01 -1.19085051e-02
6.76364779e-01 -1.15425599e+00 8.93802881e-01 6.90605044e-01
1.07604063e+00 -6.66892111e-01 -3.37210149e-02 1.33285359e-01
-1.21925795e+00 1.11212879e-01 -5.49546659e-01 1.04391791e-01
-1.89234927e-01 8.24671686e-02 -8.26902747e-01 2.25716144e-01
3.43627155e-01 1.09240782e+00 -7.40605414e-01 8.65785420e-01
-3.32984030e-01 3.91441494e-01 -6.57087415e-02 2.43894219e-01
2.48182744e-01 -4.84214574e-01 4.61236835e-01 1.28886247e+00
2.94462265e-03 1.12143293e-01 1.02624089e-01 1.45428276e+00
-5.37842631e-01 -2.04055667e-01 -9.43134487e-01 6.63620373e-03
5.31457841e-01 1.14265907e+00 -7.22632766e-01 -4.20967937e-01
-1.97841063e-01 1.27205372e+00 5.65821826e-01 6.56983733e-01
-4.58474100e-01 -4.63000417e-01 8.63961101e-01 8.81105438e-02
4.50355440e-01 -4.27782416e-01 -6.33726478e-01 -1.08734763e+00
-5.34772053e-02 -4.66301084e-01 1.32312492e-01 -1.24012315e+00
-1.36672008e+00 1.63083926e-01 -1.23461366e-01 -1.14927828e+00
1.00493133e-01 -1.01430833e+00 -8.33259046e-01 6.02211714e-01
-1.48370516e+00 -1.53595471e+00 -6.19043529e-01 6.15042210e-01
8.34661245e-01 -7.16800839e-02 6.94259644e-01 -1.52409360e-01
1.32416207e-02 6.11801863e-01 6.54908344e-02 3.49267721e-01
8.43061566e-01 -1.55747354e+00 8.11074734e-01 5.52689552e-01
7.43435919e-01 6.98576212e-01 4.01564896e-01 -3.45045209e-01
-1.43756986e+00 -1.37774491e+00 3.89652163e-01 -8.49082589e-01
4.21851575e-01 -6.79546475e-01 -9.41149414e-01 7.54214108e-01
-1.28948823e-01 3.81582290e-01 2.33588114e-01 -2.83100933e-01
-1.32126248e+00 -3.87574844e-02 -1.25802732e+00 8.01483631e-01
1.59670985e+00 -9.24658656e-01 -8.41355443e-01 1.44949108e-01
6.76781118e-01 -1.48932710e-01 -5.17377734e-01 -3.28979380e-02
8.90367270e-01 -6.53671980e-01 1.53693783e+00 -1.22774351e+00
5.71225584e-01 -3.57056201e-01 -4.51109499e-01 -1.38166070e+00
-3.58917505e-01 -3.70824009e-01 1.21901877e-01 9.41395819e-01
3.53437513e-02 -2.66072154e-01 8.00268114e-01 6.49219632e-01
1.18176825e-01 -3.22170943e-01 -4.66226429e-01 -1.00940454e+00
4.45347100e-01 -2.11987987e-01 4.15670156e-01 9.43372846e-01
-5.16491175e-01 1.99875370e-01 -1.86217993e-01 -2.21846905e-02
1.12221646e+00 5.05611658e-01 9.95933652e-01 -1.57044339e+00
-1.60270393e-01 -6.38820887e-01 -5.17401099e-01 -1.20223987e+00
5.10272324e-01 -1.23894656e+00 -2.28268772e-01 -1.42630601e+00
2.50952512e-01 -6.11414611e-01 -4.18500155e-02 3.87392163e-01
1.70367703e-01 4.97324944e-01 4.56334323e-01 1.06603086e-01
-5.75275004e-01 5.49640417e-01 1.44334722e+00 -4.37013507e-01
9.79357958e-02 -2.86050290e-01 -5.46093643e-01 1.00090086e+00
5.07898629e-01 -2.33879834e-01 -6.79989755e-01 -5.98717570e-01
-1.11110806e-01 -2.78291792e-01 8.06899607e-01 -6.94197774e-01
9.07552242e-02 -4.09458727e-01 7.07393885e-01 -4.75316703e-01
4.06592280e-01 -8.89984906e-01 -2.38256171e-01 3.92059207e-01
-1.03373826e+00 -1.72657773e-01 6.06774762e-02 9.37736630e-01
2.11339191e-01 9.74909291e-02 1.18363345e+00 5.18527366e-02
-7.97587693e-01 5.55334151e-01 2.06156418e-01 4.42313731e-01
8.98698926e-01 -9.38502550e-02 -4.10701722e-01 -3.37010205e-01
-6.11018062e-01 1.51495980e-02 8.16801190e-01 5.59899092e-01
8.09967577e-01 -1.49913919e+00 -6.45990193e-01 5.37487388e-01
4.84420866e-01 -1.22881472e-01 -3.79395373e-02 -2.61371639e-02
-7.14932323e-01 3.03616226e-01 -5.14709294e-01 -5.25909662e-01
-1.32870877e+00 6.08201921e-01 3.77581447e-01 2.99265623e-01
-1.11650598e+00 6.44338489e-01 8.97011638e-01 -5.61472833e-01
4.48256493e-01 -5.32094181e-01 3.09972614e-01 -3.74812990e-01
6.00149930e-01 3.01863939e-01 -2.63045341e-01 -3.97167534e-01
-5.79634346e-02 1.06980491e+00 9.75911412e-03 -3.22610766e-01
1.14144802e+00 6.19756095e-02 1.96111351e-01 3.21363956e-01
1.48889554e+00 -2.62485072e-02 -1.86796153e+00 -4.27016586e-01
1.65509470e-02 -8.76211166e-01 -9.64618176e-02 -8.10810089e-01
-9.41199839e-01 1.25903499e+00 3.24605942e-01 -3.41612369e-01
5.98214269e-01 1.34267092e-01 5.76994777e-01 1.02568626e+00
4.11901802e-01 -7.68673480e-01 6.66632652e-01 2.04153612e-01
1.24229157e+00 -1.09328270e+00 -1.47907257e-01 -4.59269762e-01
-7.28709698e-01 1.19221520e+00 3.72813374e-01 -6.61843717e-01
5.57444692e-01 -4.47726026e-02 -4.32636663e-02 -6.72468394e-02
-5.87835968e-01 2.59220712e-02 5.51919937e-01 9.39119160e-01
-3.43273347e-03 1.82502389e-01 3.51744592e-01 1.91711530e-01
-1.73038796e-01 -7.51469955e-02 3.43192518e-01 5.40404975e-01
-4.78640556e-01 -1.08218849e+00 -2.58781649e-02 3.82644624e-01
2.96978027e-01 1.48860216e-01 -8.47256958e-01 6.75039053e-01
2.81479489e-02 3.20035130e-01 3.57410610e-01 1.58038624e-02
3.49574387e-01 1.82046130e-01 9.54553545e-01 -4.17187095e-01
7.51934573e-02 -5.70309699e-01 -3.75075966e-01 -7.19436169e-01
-1.33846067e-02 -5.36869824e-01 -1.41337764e+00 -2.35954121e-01
3.48917961e-01 -3.04980874e-01 5.79068542e-01 6.75155818e-01
3.28248948e-01 2.44846866e-01 5.02187192e-01 -8.26419294e-01
-5.66478193e-01 -2.42353976e-01 -5.93946457e-01 1.09860766e+00
4.81195301e-01 -7.71826863e-01 -2.97271997e-01 5.81490397e-01]
|
[11.699152946472168, 0.3149799406528473]
|
05800d6c-547d-4b34-a563-99d3cd586e1a
|
pointinst3d-segmenting-3d-instances-by-points
|
2204.11402
| null |
https://arxiv.org/abs/2204.11402v2
|
https://arxiv.org/pdf/2204.11402v2.pdf
|
PointInst3D: Segmenting 3D Instances by Points
|
The current state-of-the-art methods in 3D instance segmentation typically involve a clustering step, despite the tendency towards heuristics, greedy algorithms, and a lack of robustness to the changes in data statistics. In contrast, we propose a fully-convolutional 3D point cloud instance segmentation method that works in a per-point prediction fashion. In doing so it avoids the challenges that clustering-based methods face: introducing dependencies among different tasks of the model. We find the key to its success is assigning a suitable target to each sampled point. Instead of the commonly used static or distance-based assignment strategies, we propose to use an Optimal Transport approach to optimally assign target masks to the sampled points according to the dynamic matching costs. Our approach achieves promising results on both ScanNet and S3DIS benchmarks. The proposed approach removes intertask dependencies and thus represents a simpler and more flexible 3D instance segmentation framework than other competing methods, while achieving improved segmentation accuracy.
|
['Chunhua Shen', 'Wei Yin', 'Anton Van Den Hengel', 'Tong He']
|
2022-04-25
| null | null | null | null |
['3d-instance-segmentation-1']
|
['computer-vision']
|
[ 8.35490152e-02 -5.57386465e-02 -2.30923042e-01 -4.50874448e-01
-8.32664192e-01 -5.44546783e-01 7.25461125e-01 3.17044526e-01
-5.24500489e-01 3.71852130e-01 -4.44689691e-01 -2.38501668e-01
-2.11495548e-01 -7.74132907e-01 -7.44949281e-01 -5.57496548e-01
1.84153467e-01 1.11453652e+00 8.55701029e-01 1.29007280e-01
6.72155976e-01 9.77435946e-01 -1.62309492e+00 -1.23655191e-02
1.01541829e+00 1.17747879e+00 3.37077051e-01 3.28329712e-01
-6.53153777e-01 1.52316973e-01 -4.51900810e-01 -2.04374641e-01
5.83701789e-01 -1.56776905e-01 -9.53311265e-01 3.59948575e-01
4.90795940e-01 1.10331997e-01 2.75880396e-01 8.05741012e-01
4.07011092e-01 1.34139061e-01 6.95633888e-01 -1.06719291e+00
6.02315590e-02 2.61056244e-01 -8.85685563e-01 5.62113635e-02
3.15573290e-02 1.32417455e-01 7.88929343e-01 -8.20420563e-01
6.13891721e-01 9.76944029e-01 9.06206369e-01 3.31577450e-01
-1.46077776e+00 -4.08517867e-01 3.72236937e-01 -2.92202961e-02
-1.46606195e+00 -3.32781374e-01 7.15434313e-01 -5.21310747e-01
9.94845867e-01 2.59805530e-01 6.61391795e-01 5.26685894e-01
-1.75183311e-01 7.13195562e-01 1.08133090e+00 -3.79497230e-01
4.97392803e-01 -2.02083774e-02 1.46943361e-01 4.31601763e-01
1.41746402e-01 -2.68082738e-01 -2.96376139e-01 -5.75312972e-02
7.51363337e-01 -4.39472049e-02 -8.50323886e-02 -9.63562489e-01
-1.31803107e+00 5.98317146e-01 3.94891679e-01 4.36402559e-01
-5.22618532e-01 1.97268769e-01 3.93272877e-01 -3.61840576e-02
7.69184947e-01 3.84048373e-01 -7.31928349e-01 -1.24456793e-01
-1.65982640e+00 4.88317132e-01 6.74326062e-01 9.75140810e-01
9.38406408e-01 -3.23236555e-01 -1.90001935e-01 6.40339971e-01
2.51267403e-01 1.76166788e-01 4.53910492e-02 -1.13793302e+00
4.18212742e-01 8.44214559e-01 3.61292288e-02 -9.56350863e-01
-5.45072973e-01 -5.39487541e-01 -3.75764608e-01 4.62313324e-01
4.15306121e-01 2.64759213e-01 -1.36998749e+00 1.14173114e+00
6.68139040e-01 1.89023897e-01 -4.15125549e-01 8.34700346e-01
4.03974742e-01 3.32852215e-01 -3.60667519e-02 6.57134280e-02
9.03725266e-01 -8.56661141e-01 -2.25401759e-01 -1.25207052e-01
5.37770152e-01 -8.74436915e-01 9.70032871e-01 4.81518239e-01
-1.18633926e+00 -5.71378469e-01 -8.02907348e-01 3.84899043e-02
-4.00664955e-01 -1.59174576e-01 5.18620610e-01 6.97525799e-01
-1.12919736e+00 7.49304414e-01 -1.02054119e+00 -3.53587806e-01
6.88436687e-01 6.51995957e-01 -5.57557642e-02 1.01447567e-01
-3.87124717e-01 6.55443668e-01 4.09245104e-01 -4.66662645e-02
-4.15803462e-01 -9.60996628e-01 -5.23499906e-01 -1.04885578e-01
5.78625619e-01 -5.99534571e-01 1.29937375e+00 -8.39267373e-01
-1.42447579e+00 1.02143562e+00 -1.93697244e-01 -5.26577890e-01
7.38955975e-01 -1.95177943e-01 3.62477869e-01 1.88331842e-01
1.31434366e-01 9.77559805e-01 7.78995037e-01 -1.52620196e+00
-9.04956102e-01 -4.39534366e-01 -8.37289467e-02 1.18928663e-01
1.37541324e-01 -2.91032583e-01 -9.39648628e-01 -2.65575498e-01
4.77969259e-01 -9.83565509e-01 -7.07743347e-01 -7.40061095e-03
-5.00504196e-01 -2.55655050e-01 1.01347232e+00 -8.09169114e-02
9.70762134e-01 -1.93366456e+00 2.28973895e-01 4.77389008e-01
2.34799460e-01 2.72230864e-01 2.09506303e-01 2.73138523e-01
2.69095153e-01 1.59022570e-01 -5.05695403e-01 -7.58789480e-01
8.76339152e-02 1.14412129e-01 1.48496822e-01 4.63757753e-01
1.24791019e-01 7.28849292e-01 -7.81510592e-01 -6.88986123e-01
7.16514766e-01 4.30772245e-01 -6.43952489e-01 -1.20129026e-01
-5.77463865e-01 5.63517332e-01 -6.49714410e-01 7.14644849e-01
9.62396026e-01 -2.71522731e-01 9.97129083e-03 -6.16722107e-02
-4.26563174e-01 3.30683440e-01 -1.32557321e+00 2.01245832e+00
-2.96893746e-01 3.10173780e-01 1.61212817e-01 -1.17226732e+00
1.10896194e+00 7.23896846e-02 9.88729835e-01 -5.88061213e-01
7.13941967e-03 5.08446276e-01 -7.69293606e-02 -9.37906578e-02
5.74711919e-01 1.26238331e-01 -2.25079879e-02 1.77003890e-01
-1.33634612e-01 -4.81030941e-01 1.19207010e-01 1.07190898e-03
9.62114334e-01 5.84005177e-01 1.21374473e-01 -4.97055382e-01
5.10668099e-01 4.87387985e-01 5.32743454e-01 8.26124728e-01
-1.33834392e-01 8.95851314e-01 3.84911180e-01 -4.61208045e-01
-9.52008188e-01 -8.51729155e-01 -3.22793722e-01 4.28668469e-01
3.15382242e-01 -1.96067974e-01 -1.01053154e+00 -7.80549347e-01
1.16814256e-01 5.79307616e-01 -3.86809498e-01 3.67535770e-01
-7.24941254e-01 -6.22211337e-01 2.00633600e-01 3.57563794e-01
4.70657259e-01 -8.24516356e-01 -8.22388411e-01 4.74911392e-01
4.91131321e-02 -1.13405776e+00 -2.19313264e-01 4.34795946e-01
-1.29750764e+00 -1.01276684e+00 -5.96319377e-01 -5.17033577e-01
6.86041713e-01 3.80004525e-01 1.27693605e+00 1.55351326e-01
-1.36651278e-01 2.66277522e-01 -4.58101273e-01 -2.96770453e-01
-9.47136953e-02 8.01476359e-01 -4.07621562e-01 -7.35924095e-02
5.40632010e-01 -4.57904637e-01 -8.15091550e-01 4.26304311e-01
-7.36220777e-01 -3.93533520e-02 5.04873335e-01 4.53133345e-01
1.10667121e+00 -4.18714471e-02 2.50160456e-01 -1.15441298e+00
1.78744644e-01 -2.82728851e-01 -6.56714380e-01 -1.27276838e-01
-8.23092937e-01 -9.82094184e-02 5.09364128e-01 -1.09973572e-01
-7.41694093e-01 5.86867392e-01 -1.69820160e-01 -5.57574272e-01
-6.72207057e-01 1.67371422e-01 -1.28399670e-01 -3.44706982e-01
4.49690044e-01 1.02408007e-02 -2.08105110e-02 -7.09010482e-01
3.80731642e-01 4.13213342e-01 2.25489289e-01 -7.01418638e-01
8.82872641e-01 7.41256952e-01 9.03216377e-02 -7.36166179e-01
-5.64790428e-01 -7.46515632e-01 -1.27266777e+00 -3.02169532e-01
1.00018132e+00 -5.42664051e-01 -3.71598840e-01 4.55200076e-01
-1.29265022e+00 -4.13069487e-01 -3.61115009e-01 1.12106182e-01
-7.10892618e-01 2.96543330e-01 -3.01665246e-01 -6.53320968e-01
-9.50976759e-02 -1.58550990e+00 1.42216229e+00 5.61375059e-02
-1.99581042e-01 -8.73560667e-01 -1.41631007e-01 3.95422429e-01
3.29705358e-01 5.44428468e-01 7.82740355e-01 -5.96038699e-01
-1.05738938e+00 -1.98435143e-01 -2.32782900e-01 1.39384598e-01
8.00242871e-02 1.79517642e-01 -8.48256350e-01 2.41869092e-02
-1.55827850e-01 2.20041871e-01 6.87657177e-01 7.01068759e-01
1.43733776e+00 2.86756903e-01 -6.62173152e-01 6.72832489e-01
1.63722157e+00 1.95631400e-01 5.24534702e-01 5.77006280e-01
7.25580633e-01 5.76259017e-01 8.34226608e-01 2.53626555e-01
4.92963791e-01 9.05705094e-01 7.25009203e-01 -3.40399861e-01
-1.59315225e-02 4.28093709e-02 -3.40556532e-01 4.85306978e-01
-2.00412527e-01 -2.26337478e-01 -1.19967461e+00 7.03941047e-01
-2.02480412e+00 -7.18981624e-01 -6.30823910e-01 2.27291894e+00
4.82337981e-01 4.74464685e-01 3.70862007e-01 2.87659436e-01
5.45797825e-01 1.82148665e-01 -3.97667050e-01 -3.95339608e-01
2.83690929e-01 4.63721305e-01 7.74662077e-01 3.83154511e-01
-1.17664206e+00 1.12212324e+00 6.10874653e+00 8.86708260e-01
-1.14813972e+00 2.66913325e-02 6.27226293e-01 -2.88517028e-01
-1.35151312e-01 2.13372800e-02 -8.28880370e-01 4.87989366e-01
5.04239023e-01 3.65339845e-01 1.18788213e-01 7.71702826e-01
3.83093476e-01 -2.80747801e-01 -1.11682916e+00 7.32810318e-01
-1.44530848e-01 -1.49250281e+00 -5.55736162e-02 2.63728648e-01
6.51963711e-01 2.40791604e-01 -1.66684657e-01 -1.15606762e-01
1.41179329e-02 -7.93160379e-01 9.65093136e-01 5.76636553e-01
4.20248598e-01 -9.20510948e-01 4.88922209e-01 5.22412658e-01
-1.12412310e+00 1.75843194e-01 -1.96746022e-01 1.34580106e-01
2.07427815e-01 8.59079957e-01 -8.16647410e-01 6.26665533e-01
9.12738144e-01 4.78620410e-01 -4.53463942e-01 1.32043374e+00
1.84598461e-01 3.59545529e-01 -4.70795304e-01 1.35233521e-01
6.13678753e-01 -3.22537899e-01 5.29443979e-01 1.35073447e+00
2.00689062e-01 -3.32089752e-01 4.13864166e-01 8.57626617e-01
2.06777498e-01 1.41027287e-01 -4.66591626e-01 4.34526116e-01
5.14861405e-01 1.18496263e+00 -1.42985737e+00 -1.37024269e-01
-3.82587224e-01 8.53840590e-01 2.05567837e-01 1.07388139e-01
-7.35006213e-01 -1.83716103e-01 6.35696888e-01 5.73365390e-01
7.84738779e-01 -6.42139316e-01 -7.78969586e-01 -5.87158978e-01
1.90123647e-01 -6.87234998e-01 8.72166902e-02 -3.24328631e-01
-1.06761694e+00 4.73603308e-01 1.94326252e-01 -1.32382870e+00
-3.38136032e-02 -4.53353018e-01 -4.91783142e-01 7.10497141e-01
-1.79278803e+00 -1.12269104e+00 -2.63715386e-01 4.99508291e-01
8.40030491e-01 2.08165973e-01 3.68641436e-01 4.47286606e-01
-2.77523965e-01 1.95107296e-01 1.41826007e-04 -2.12132633e-01
4.27781969e-01 -1.44221187e+00 7.19079673e-01 7.77594030e-01
5.13571128e-02 3.68295938e-01 5.31200647e-01 -6.37981057e-01
-1.02912307e+00 -9.49098289e-01 8.26727986e-01 -4.94516432e-01
3.53085101e-01 -4.48734492e-01 -9.37738061e-01 3.33538234e-01
-8.44718441e-02 -7.28722364e-02 3.91388893e-01 5.92796654e-02
4.61280905e-02 -5.50724827e-02 -1.22145903e+00 3.71591985e-01
1.31063902e+00 -2.93037035e-02 -2.71207571e-01 2.15344042e-01
6.64561272e-01 -6.85530484e-01 -7.90308416e-01 4.85152483e-01
3.23823363e-01 -1.24224401e+00 1.09415340e+00 -1.78330302e-01
3.36626917e-01 -5.31647205e-01 2.13015173e-02 -1.12629032e+00
-3.29034001e-01 -4.77727473e-01 1.25006244e-01 1.15381622e+00
5.14595568e-01 -3.95413756e-01 1.25470519e+00 6.35846019e-01
-5.60399652e-01 -9.44047034e-01 -1.14544499e+00 -7.24757373e-01
6.03086688e-02 -7.20020056e-01 8.78199458e-01 8.53888869e-01
-6.90953970e-01 -2.21708089e-01 1.14900224e-01 2.36450866e-01
8.60284805e-01 3.53301853e-01 1.02207422e+00 -1.57822156e+00
-1.08930759e-01 -7.78059661e-01 -3.83224338e-01 -1.21478665e+00
-1.54265268e-02 -8.83732438e-01 1.86908960e-01 -1.67331934e+00
-3.04623336e-01 -1.12024927e+00 -1.00531660e-01 2.63604879e-01
4.05490696e-02 3.54794502e-01 1.96924105e-01 3.88660491e-01
-6.99907660e-01 8.95173848e-02 1.16797924e+00 7.41653144e-02
-5.43451130e-01 1.89315483e-01 -3.85519087e-01 6.58543944e-01
7.42767990e-01 -6.27335250e-01 -3.17517877e-01 -6.38782501e-01
-1.51810318e-01 -1.96715400e-01 4.22154218e-01 -1.10508037e+00
3.60400856e-01 -2.13060528e-01 3.09186816e-01 -1.11626530e+00
3.13455284e-01 -1.12164438e+00 2.74367720e-01 2.51834452e-01
1.33456793e-02 -8.33086595e-02 1.26257300e-01 4.33387399e-01
-2.81918831e-02 -2.48320147e-01 9.56895769e-01 -4.21114385e-01
-6.99784458e-01 5.86531103e-01 -1.50857359e-01 -2.17637196e-01
1.19009674e+00 -7.83662915e-01 1.73466206e-01 2.70151615e-01
-5.58088064e-01 2.79195368e-01 8.65581751e-01 3.35844576e-01
3.94022018e-01 -9.10049438e-01 -3.99153054e-01 1.12809137e-01
-1.43728964e-02 7.01685607e-01 1.11455165e-01 1.03725719e+00
-6.90821648e-01 3.23064923e-01 3.49494964e-02 -1.10454869e+00
-1.00028479e+00 2.75623947e-01 4.89701509e-01 -1.97198093e-01
-8.17381084e-01 6.58956051e-01 -1.44664362e-01 -5.36120057e-01
2.86156684e-01 -3.89052540e-01 9.40874293e-02 1.07599922e-01
-7.02609718e-02 3.87666851e-01 5.68315506e-01 -5.83964884e-01
-5.10702968e-01 7.50200987e-01 -1.19788356e-01 3.13758887e-02
1.39152455e+00 -1.34102911e-01 6.61929846e-02 4.11311090e-01
8.71281922e-01 -1.49786696e-01 -1.36832404e+00 -1.02374554e-01
4.35436219e-01 -6.58088267e-01 3.20733517e-01 -5.69917798e-01
-1.24561429e+00 7.99266577e-01 5.29512107e-01 4.86563951e-01
8.94280791e-01 1.25508517e-01 9.44859982e-01 -7.30057731e-02
4.83252734e-01 -1.21037817e+00 -3.95220935e-01 3.48421961e-01
3.45060676e-01 -1.15533388e+00 8.76166523e-02 -5.58111727e-01
-2.90852278e-01 1.05064130e+00 6.03790700e-01 -2.46846288e-01
6.59393549e-01 2.66465425e-01 1.19595699e-01 -4.54484046e-01
-3.31896126e-01 -3.78021121e-01 2.66285449e-01 7.99117386e-01
4.03134674e-01 -9.87048596e-02 -3.89527619e-01 -5.15711075e-03
-2.06691748e-03 -8.51234049e-02 1.19345412e-01 8.79671574e-01
-4.45752829e-01 -1.43810248e+00 -4.58665609e-01 5.90962052e-01
-2.88447201e-01 2.18578324e-01 -4.07417893e-01 1.21832216e+00
2.41339922e-01 6.34293258e-01 3.12203735e-01 -2.21085146e-01
6.06261313e-01 9.61671621e-02 3.53313357e-01 -8.00944269e-01
-8.02147090e-01 2.73104548e-01 -1.95585221e-01 -7.29754925e-01
-8.31397355e-01 -1.05219626e+00 -1.39941204e+00 -2.04825446e-01
-4.56067801e-01 -7.19101876e-02 9.00309205e-01 1.00210857e+00
5.95026374e-01 3.26496780e-01 6.45243227e-01 -1.44432175e+00
-1.44775853e-01 -4.87482131e-01 -3.33271444e-01 2.58358032e-01
2.73799505e-02 -7.78989494e-01 -1.73933297e-01 -2.11926460e-01]
|
[8.043991088867188, -3.0657503604888916]
|
e26bde70-dd00-4fa3-9cdd-03dffe5972dc
|
attentive-and-contrastive-learning-for-joint-1
|
2110.06853
| null |
https://arxiv.org/abs/2110.06853v1
|
https://arxiv.org/pdf/2110.06853v1.pdf
|
Attentive and Contrastive Learning for Joint Depth and Motion Field Estimation
|
Estimating the motion of the camera together with the 3D structure of the scene from a monocular vision system is a complex task that often relies on the so-called scene rigidity assumption. When observing a dynamic environment, this assumption is violated which leads to an ambiguity between the ego-motion of the camera and the motion of the objects. To solve this problem, we present a self-supervised learning framework for 3D object motion field estimation from monocular videos. Our contributions are two-fold. First, we propose a two-stage projection pipeline to explicitly disentangle the camera ego-motion and the object motions with dynamics attention module, called DAM. Specifically, we design an integrated motion model that estimates the motion of the camera and object in the first and second warping stages, respectively, controlled by the attention module through a shared motion encoder. Second, we propose an object motion field estimation through contrastive sample consensus, called CSAC, taking advantage of weak semantic prior (bounding box from an object detector) and geometric constraints (each object respects the rigid body motion model). Experiments on KITTI, Cityscapes, and Waymo Open Dataset demonstrate the relevance of our approach and show that our method outperforms state-of-the-art algorithms for the tasks of self-supervised monocular depth estimation, object motion segmentation, monocular scene flow estimation, and visual odometry.
|
['In So Kweon', 'Fei Pan', 'Francois Rameau', 'Seokju Lee']
|
2021-10-13
|
attentive-and-contrastive-learning-for-joint
|
http://openaccess.thecvf.com//content/ICCV2021/html/Lee_Attentive_and_Contrastive_Learning_for_Joint_Depth_and_Motion_Field_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Lee_Attentive_and_Contrastive_Learning_for_Joint_Depth_and_Motion_Field_ICCV_2021_paper.pdf
|
iccv-2021-1
|
['motion-segmentation', 'scene-flow-estimation']
|
['computer-vision', 'computer-vision']
|
[-5.09482436e-02 -2.27339670e-01 -2.31484875e-01 -1.02122240e-01
-1.40300155e-01 -6.95208192e-01 6.50405586e-01 -7.57947624e-01
-4.03743297e-01 3.11251789e-01 1.10782102e-01 1.36036262e-01
2.95586079e-01 -3.21415871e-01 -8.51840019e-01 -7.35977948e-01
5.23415506e-01 5.79003394e-01 5.16770780e-01 3.00344050e-01
3.97155941e-01 6.22876465e-01 -1.37377417e+00 5.80693930e-02
5.30249596e-01 6.95919454e-01 6.69220209e-01 9.92779732e-01
8.04327205e-02 1.20676374e+00 7.69082755e-02 3.08634620e-02
3.44226360e-01 -2.29459047e-01 -1.02951813e+00 6.94320917e-01
7.84420550e-01 -8.52635384e-01 -8.59807551e-01 9.80444312e-01
4.99349236e-02 1.63571477e-01 5.62897980e-01 -1.26183712e+00
-1.31351054e-01 -2.22933535e-02 -6.58598900e-01 1.80306241e-01
4.74714220e-01 4.66540426e-01 7.55555749e-01 -9.09664571e-01
1.23363078e+00 1.35916531e+00 1.62089750e-01 5.14892042e-01
-1.14130402e+00 -2.53318042e-01 3.85386646e-01 3.63560706e-01
-1.28530967e+00 -5.62231421e-01 1.01043952e+00 -1.02936137e+00
8.28066468e-01 -1.67112321e-01 7.66489744e-01 9.28167045e-01
2.46580780e-01 9.31707025e-01 6.70548201e-01 -2.13439614e-01
3.26481014e-01 -2.18274146e-02 -7.49022141e-02 7.84063220e-01
1.48672760e-01 2.14871272e-01 -4.31335986e-01 1.75172493e-01
1.07603085e+00 7.09745958e-02 -4.73861814e-01 -1.18602538e+00
-1.30758929e+00 5.43494761e-01 1.55713499e-01 4.66750227e-02
-2.06528798e-01 3.64461780e-01 6.24621175e-02 -1.69669092e-01
3.25324535e-01 -6.23884276e-02 -3.66672456e-01 -1.81305911e-02
-7.81159222e-01 3.00274551e-01 6.91881657e-01 1.15987003e+00
8.25831950e-01 6.73971027e-02 2.67078936e-01 1.67706594e-01
5.67683935e-01 5.71782649e-01 3.40580553e-01 -1.48821604e+00
4.76287782e-01 4.87486482e-01 4.19344693e-01 -9.40794051e-01
-3.59407634e-01 -7.89531022e-02 -4.97035772e-01 3.04168463e-01
6.70445621e-01 7.44534060e-02 -8.31113279e-01 1.70451164e+00
8.08245838e-01 3.61350924e-01 -7.30381161e-02 1.33880770e+00
6.46998942e-01 4.24937546e-01 -3.33739042e-01 -1.35856137e-01
1.08859301e+00 -1.26873672e+00 -6.20314538e-01 -5.89207351e-01
5.23534179e-01 -6.94438517e-01 5.99736810e-01 1.35463819e-01
-1.34430695e+00 -6.23351634e-01 -8.72586250e-01 -4.87132549e-01
5.89004122e-02 1.30354166e-01 4.62347150e-01 1.64020047e-01
-6.24456227e-01 2.56254524e-01 -1.25087285e+00 -3.04758340e-01
3.19278181e-01 1.96630120e-01 -5.98546088e-01 -2.99924314e-01
-4.97780234e-01 9.00254846e-01 4.31695700e-01 8.38876069e-02
-1.15053844e+00 -5.96316397e-01 -1.25058317e+00 -1.58417806e-01
4.14649218e-01 -1.22234583e+00 1.07499433e+00 -9.05076027e-01
-1.67880249e+00 9.35573459e-01 -4.23317075e-01 -1.84903890e-01
8.00611854e-01 -4.76896673e-01 3.54698062e-01 4.56861347e-01
5.50314561e-02 8.11465323e-01 8.95585954e-01 -1.16379714e+00
-6.35599077e-01 -4.98257458e-01 7.02494830e-02 6.62141860e-01
3.62936407e-01 -2.13089138e-01 -7.92554319e-01 -2.24491358e-01
4.08566236e-01 -1.28257179e+00 -2.32546419e-01 3.22902709e-01
-3.40053082e-01 1.25551343e-01 9.50902641e-01 -6.47730052e-01
7.46795774e-01 -2.12664413e+00 7.21154392e-01 -4.15364891e-01
1.37836128e-01 6.31165504e-02 1.33138195e-01 -1.08903669e-01
8.17065537e-02 -5.89214742e-01 -2.22869277e-01 -6.36602461e-01
-3.43948990e-01 2.08926588e-01 -2.74675906e-01 1.05300963e+00
1.68469310e-01 1.03832984e+00 -1.03571463e+00 -3.90968084e-01
7.46215045e-01 3.99917334e-01 -7.68407524e-01 4.00376141e-01
-3.02176386e-01 8.99229348e-01 -2.99629420e-01 5.54540336e-01
8.47848594e-01 -3.63717258e-01 8.15191343e-02 -2.60686874e-01
-3.03036839e-01 3.04791033e-01 -1.67152917e+00 1.98945904e+00
-3.02993625e-01 6.69856131e-01 2.07093418e-01 -7.07014143e-01
2.62788296e-01 1.19634494e-01 5.97491086e-01 -1.71685755e-01
2.08767265e-01 8.03579316e-02 -9.28261504e-02 -7.21772134e-01
3.61908853e-01 1.76267102e-01 3.05902958e-01 3.68715554e-01
2.31371641e-01 -6.68240786e-01 7.05918074e-02 1.99522018e-01
6.99017406e-01 7.37892449e-01 3.90664130e-01 -5.36487699e-02
6.80457532e-01 3.58487442e-02 6.60242260e-01 3.14647406e-01
-3.98793578e-01 8.20237815e-01 2.29832187e-01 -5.75076759e-01
-1.20878839e+00 -9.32019234e-01 1.19962282e-01 3.68960857e-01
5.18036008e-01 4.36441153e-02 -6.64148629e-01 -5.42302489e-01
7.50077143e-02 3.28866482e-01 -4.72332120e-01 9.88160633e-03
-8.19921672e-01 -3.77710938e-01 -1.87989309e-01 5.28800726e-01
5.59782922e-01 -7.85007358e-01 -1.09987986e+00 3.29849049e-02
-5.89927971e-01 -1.99223149e+00 -9.43383336e-01 -1.69551685e-01
-8.73776793e-01 -1.13291979e+00 -5.44106185e-01 -5.64307153e-01
5.95567346e-01 8.20859075e-01 7.33624816e-01 -4.12816733e-01
-3.16384494e-01 5.64231634e-01 2.04489604e-01 1.30931020e-01
-1.51913971e-01 -1.61645263e-01 3.97874489e-02 3.77065837e-01
1.48922205e-01 -5.36533058e-01 -8.47286105e-01 3.71795446e-01
-7.55666018e-01 4.67251152e-01 2.41363689e-01 4.91508693e-01
4.71215278e-01 -2.87521183e-01 -3.46712291e-01 -4.23769474e-01
-7.16807365e-01 -4.24586654e-01 -1.07687771e+00 -2.13778958e-01
4.66759391e-02 5.38812857e-03 2.99660116e-01 -4.73887205e-01
-1.16615868e+00 9.79114830e-01 3.60495090e-01 -9.07070816e-01
-2.56141096e-01 -1.53748661e-01 -4.57085103e-01 -1.06645994e-01
1.92846701e-01 3.76612037e-01 -2.04431098e-02 -3.00560474e-01
5.94373882e-01 3.02882075e-01 7.34545231e-01 -1.26145795e-01
9.67818141e-01 1.38296819e+00 1.74161434e-01 -9.66488838e-01
-8.52147222e-01 -8.94559979e-01 -1.30758786e+00 -2.37337500e-01
1.41263103e+00 -1.38305223e+00 -7.87782490e-01 7.55933940e-01
-1.60576046e+00 -3.07136595e-01 -1.54704586e-01 8.33455563e-01
-9.86676455e-01 7.39779353e-01 -5.82690358e-01 -7.17005074e-01
-3.52468677e-02 -1.28775632e+00 1.36843526e+00 3.65153775e-02
-8.71774554e-02 -1.02875257e+00 9.79342759e-02 6.57329917e-01
-1.84701368e-01 2.99406022e-01 4.82746989e-01 2.10067153e-01
-1.40841675e+00 2.33321264e-01 -1.72700882e-01 2.70016313e-01
-6.04397655e-02 4.26090555e-03 -1.09876323e+00 -1.88012257e-01
2.35263109e-01 -1.95207763e-02 8.44392061e-01 7.06087470e-01
6.71433210e-01 -7.93255493e-02 -2.63414621e-01 1.18889463e+00
1.44980204e+00 6.58741295e-02 4.79491711e-01 1.64826915e-01
1.36842120e+00 8.47841501e-01 5.34631133e-01 3.48253638e-01
5.99621773e-01 1.07923472e+00 7.50090063e-01 2.25109816e-01
-3.01287025e-01 -2.74479181e-01 6.01177096e-01 7.14632869e-01
1.00245088e-01 8.05607215e-02 -7.52653539e-01 6.84062660e-01
-1.86332262e+00 -9.13888574e-01 -3.63971740e-01 2.16443968e+00
3.49522352e-01 -9.32933092e-02 1.44404834e-02 -8.90382752e-02
3.81277531e-01 1.14040010e-01 -8.17889452e-01 1.21310942e-01
-1.71461791e-01 -4.60541874e-01 5.15252233e-01 1.07458878e+00
-1.09336591e+00 1.26852655e+00 5.13082552e+00 1.10816807e-01
-1.19999290e+00 1.09764062e-01 1.83813989e-01 -3.16085458e-01
4.40448150e-02 3.25517505e-01 -9.83493030e-01 3.96176934e-01
2.81588465e-01 6.37121573e-02 3.18268210e-01 8.08986425e-01
2.39673182e-01 -2.62739003e-01 -1.43406653e+00 1.20013928e+00
3.03570360e-01 -1.23152483e+00 -1.43153518e-01 2.38217577e-01
9.67901707e-01 3.08779210e-01 -1.38658985e-01 -3.79865110e-01
1.99837327e-01 -4.69359696e-01 1.10809410e+00 4.04855192e-01
4.58140731e-01 -2.53154814e-01 3.63251835e-01 5.80996037e-01
-1.28452432e+00 -1.67057529e-01 -3.09202194e-01 -2.14263380e-01
6.07902646e-01 4.79930490e-01 -5.55185139e-01 5.36219358e-01
5.51272988e-01 1.10745633e+00 -1.62951604e-01 9.11703408e-01
-2.24249572e-01 2.20227540e-02 -2.39670828e-01 6.62399232e-01
2.03590661e-01 -3.81683081e-01 8.64713550e-01 8.91728759e-01
-3.09051778e-02 2.78312653e-01 1.54239848e-01 1.15054381e+00
2.10057244e-01 -3.70273709e-01 -7.11878359e-01 3.22106034e-01
-6.51031919e-03 1.16418958e+00 -5.69267035e-01 -3.36182982e-01
-6.32832944e-01 1.25508499e+00 2.29259759e-01 4.94096279e-01
-7.68838286e-01 2.63805926e-01 8.44008684e-01 1.70494929e-01
6.08637989e-01 -6.20955884e-01 -2.05842048e-01 -1.75538456e+00
1.83574274e-01 -3.95177156e-01 6.87712207e-02 -1.08846653e+00
-6.33070946e-01 2.51522809e-01 2.64601540e-02 -1.32964110e+00
-3.30123514e-01 -7.02178836e-01 -3.83960605e-01 7.68541038e-01
-1.71383393e+00 -1.06934881e+00 -6.52402520e-01 5.89228988e-01
8.11504722e-01 2.67018706e-01 1.51831463e-01 1.37475669e-01
-3.98574024e-01 -2.25271523e-01 -1.76773682e-01 4.42621149e-02
4.51900691e-01 -9.80473101e-01 4.70814884e-01 1.04773712e+00
1.23097010e-01 2.33448297e-01 4.68724936e-01 -5.13220310e-01
-1.70475781e+00 -1.12755060e+00 8.56408715e-01 -1.03139520e+00
4.80182797e-01 -5.38300633e-01 -8.38627815e-01 8.63372505e-01
-1.03287287e-01 3.81957024e-01 1.63098965e-02 -7.88507402e-01
-1.24629907e-01 2.33938217e-01 -6.71655357e-01 5.92118859e-01
1.25623441e+00 -5.78933299e-01 -5.10395348e-01 3.00000608e-01
7.02840686e-01 -7.59738564e-01 -3.78009319e-01 1.75123900e-01
6.66601539e-01 -9.02105629e-01 1.12569451e+00 -3.70145857e-01
5.99307239e-01 -5.74246705e-01 -2.80043632e-01 -7.90227175e-01
-1.89063326e-01 -7.96922505e-01 -5.01109958e-01 6.84807420e-01
-3.10547650e-01 -1.63603514e-01 9.98264730e-01 5.63033402e-01
-4.49609570e-02 -2.83120632e-01 -9.30729032e-01 -5.98978400e-01
-1.79632723e-01 -3.78136218e-01 -2.33240034e-02 9.77446854e-01
-2.79734939e-01 4.80575353e-01 -4.95465577e-01 4.09253597e-01
8.00248265e-01 2.47657776e-01 1.25012648e+00 -1.04873896e+00
-4.98692334e-01 -2.59511799e-01 -5.52817285e-01 -1.79338062e+00
4.34356302e-01 -6.11152112e-01 5.69471158e-02 -1.24361670e+00
2.50999629e-01 1.75302282e-01 4.18461889e-01 -1.24489427e-01
-7.85964951e-02 5.16295508e-02 3.40884030e-01 4.43224669e-01
-4.99415427e-01 6.89895630e-01 1.46979463e+00 1.17376409e-01
-3.54199648e-01 -7.29819685e-02 3.51127461e-02 1.11239517e+00
8.28163400e-02 -3.94953609e-01 -3.68276387e-01 -7.79692411e-01
3.14202812e-03 3.40739340e-01 7.12131739e-01 -7.53597617e-01
4.31489289e-01 -4.02000308e-01 2.84707367e-01 -8.06273460e-01
5.82552075e-01 -1.01400304e+00 1.89821988e-01 6.87511146e-01
4.62501384e-02 -6.56938404e-02 2.20669135e-02 7.06167638e-01
1.12011330e-02 -3.65670882e-02 1.02379382e+00 -2.77907223e-01
-8.32611620e-01 5.70540726e-01 -1.98107883e-01 1.49275750e-01
1.07705975e+00 -3.37259084e-01 -9.59563777e-02 -3.50319177e-01
-6.51264608e-01 2.07586750e-01 7.42493987e-01 5.25594413e-01
7.38493264e-01 -1.13507664e+00 -6.60349965e-01 5.15832841e-01
1.05045184e-01 4.64494765e-01 4.10394669e-01 1.02342713e+00
-7.98129380e-01 5.84402800e-01 -8.63701701e-02 -1.08732235e+00
-1.16015685e+00 6.39395058e-01 5.89246035e-01 -2.59409323e-02
-8.36618125e-01 6.09813809e-01 9.69852567e-01 -3.54966313e-01
4.98149991e-02 -5.30352116e-01 3.42971943e-02 -3.12620729e-01
4.71120238e-01 6.13598943e-01 -2.33691946e-01 -1.02840471e+00
-3.83898735e-01 1.22627413e+00 8.00580606e-02 -3.07771355e-01
9.78776991e-01 -6.63686037e-01 7.09613711e-02 5.49740195e-01
1.31879914e+00 -2.50249863e-01 -1.97957504e+00 -3.92293453e-01
-2.67535031e-01 -7.72526741e-01 1.15726322e-01 -1.32781923e-01
-9.86284971e-01 1.10216641e+00 2.89785922e-01 -5.78431010e-01
8.26985180e-01 3.35552618e-02 6.57112300e-01 1.36411071e-01
3.45071018e-01 -8.94292355e-01 1.65092528e-01 6.91171825e-01
6.64153636e-01 -1.42257428e+00 -1.10625969e-02 -7.42949486e-01
-5.91847956e-01 1.01023912e+00 7.27160990e-01 -2.94398785e-01
5.60707927e-01 -1.85284056e-02 -4.65338081e-02 7.71827549e-02
-8.80959392e-01 -1.73609510e-01 4.42711741e-01 4.33920830e-01
-3.43224593e-02 -3.18459749e-01 2.74540722e-01 -9.76626873e-02
1.83607444e-01 1.65100780e-03 7.40902901e-01 8.09300780e-01
-2.55266190e-01 -5.69216311e-01 -2.25666374e-01 -5.06897449e-01
-1.70519277e-01 1.41494185e-01 -3.29634964e-01 8.53940606e-01
1.36724830e-01 6.89251184e-01 2.56839365e-01 -1.50618181e-02
3.01305711e-01 -1.05405249e-01 8.46894503e-01 -6.58532023e-01
1.51867783e-02 2.41122514e-01 -3.18800241e-01 -9.64559078e-01
-6.49188459e-01 -9.15348589e-01 -1.12444723e+00 -2.63960864e-02
-2.37849861e-01 -3.81284297e-01 7.83560157e-01 1.14750612e+00
2.73038715e-01 1.91615820e-01 5.64072549e-01 -1.28766847e+00
-4.04201925e-01 -6.11303866e-01 -3.95255417e-01 5.49456894e-01
6.81015611e-01 -7.56718338e-01 -6.04027987e-01 5.13099134e-01]
|
[8.51708984375, -2.040933609008789]
|
a8dddf1d-0159-4a0f-8bdf-02e21eddf9ef
|
lqvsumm-a-corpus-of-linguistic-quality
| null | null |
https://aclanthology.org/L14-1467
|
https://aclanthology.org/L14-1467.pdf
|
LQVSumm: A Corpus of Linguistic Quality Violations in Multi-Document Summarization
|
We present LQVSumm, a corpus of about 2000 automatically created extractive multi-document summaries from the TAC 2011 shared task on Guided Summarization, which we annotated with several types of linguistic quality violations. Examples for such violations include pronouns that lack antecedents or ungrammatical clauses. We give details on the annotation scheme and show that inter-annotator agreement is good given the open-ended nature of the task. The annotated summaries have previously been scored for Readability on a numeric scale by human annotators in the context of the TAC challenge; we show that the number of instances of violations of linguistic quality of a summary correlates with these intuitively assigned numeric scores. On a system-level, the average number of violations marked in a system{'}s summaries achieves higher correlation with the Readability scores than current supervised state-of-the-art methods for assigning a single readability score to a summary. It is our hope that our corpus facilitates the development of methods that not only judge the linguistic quality of automatically generated summaries as a whole, but which also allow for detecting, labeling, and fixing particular violations in a text.
|
['Annemarie Friedrich', 'Marina Valeeva', 'Alexis Palmer']
|
2014-05-01
| null | null | null |
lrec-2014-5
|
['sentence-compression']
|
['natural-language-processing']
|
[ 3.18958193e-01 7.10745692e-01 -1.27660885e-01 -5.64509273e-01
-1.66286051e+00 -1.00835991e+00 7.89575875e-01 9.95189428e-01
-3.86864066e-01 9.40887809e-01 1.22872913e+00 -1.41421920e-02
-3.54883224e-01 -4.69450593e-01 -4.70687121e-01 6.65381402e-02
3.72243166e-01 6.41306043e-01 1.08056180e-01 -2.41991028e-01
8.49248707e-01 -5.63706085e-02 -1.26536715e+00 7.19863117e-01
1.43409836e+00 4.99303311e-01 -4.12384309e-02 9.50962007e-01
-8.91419649e-02 8.71220887e-01 -1.27432930e+00 -7.45363712e-01
-2.54737020e-01 -4.45784986e-01 -1.30714500e+00 8.11317489e-02
1.10357904e+00 -6.27664253e-02 9.99605283e-02 1.11830986e+00
3.99038702e-01 9.53319818e-02 8.94207716e-01 -8.74469042e-01
-7.97480464e-01 1.07695329e+00 -4.14976068e-02 4.27350163e-01
8.82413685e-01 1.48416892e-01 1.65015996e+00 -4.95502651e-01
7.60799944e-01 1.13093650e+00 4.82301861e-01 3.98608029e-01
-1.08761859e+00 -7.02549443e-02 -1.74440727e-01 -5.21097332e-02
-7.21210063e-01 -8.69593084e-01 2.95160949e-01 -7.50401616e-01
1.39235222e+00 5.86084545e-01 2.49960169e-01 8.09838772e-01
2.24325821e-01 5.64189672e-01 7.66034484e-01 -6.76354885e-01
1.48850679e-01 -5.01245596e-02 5.62377512e-01 6.34162128e-01
6.69274807e-01 -6.24152064e-01 -8.13120663e-01 -2.73951054e-01
-1.23819679e-01 -1.02383912e+00 -5.49417436e-01 3.21756333e-01
-1.34119523e+00 6.91071033e-01 -8.88417810e-02 6.68988407e-01
-1.59534708e-01 -7.45467842e-02 8.87403905e-01 2.11177677e-01
7.40155876e-01 1.32713354e+00 -5.55098653e-01 -6.80311203e-01
-1.29216671e+00 4.07410353e-01 1.04792309e+00 1.00404799e+00
3.70654970e-01 -2.05267370e-01 -7.88090706e-01 7.44901299e-01
-2.09430277e-01 3.64510566e-01 5.24972975e-01 -1.37159419e+00
9.37555969e-01 9.00857091e-01 5.04427075e-01 -9.67400312e-01
-4.62678850e-01 -2.45477781e-01 -3.34283829e-01 -2.64626235e-01
4.11667556e-01 4.28388454e-02 -1.86346874e-01 1.58441234e+00
-3.80670249e-01 -7.64451265e-01 2.77432472e-01 4.11493659e-01
1.19174397e+00 6.23783112e-01 -1.65210709e-01 -6.85777664e-01
1.31044352e+00 -8.12488794e-01 -1.11069679e+00 -2.67775655e-01
1.02405536e+00 -8.61608326e-01 1.66428018e+00 4.52210516e-01
-1.42656732e+00 -4.27818507e-01 -1.12184048e+00 -3.11185628e-01
-2.88096666e-02 4.42557067e-01 -2.66554649e-03 4.58715409e-01
-1.14253485e+00 7.50354171e-01 -4.21638191e-01 -4.22064781e-01
5.70609346e-02 -2.98503991e-02 -3.08402002e-01 3.28252256e-01
-9.29992855e-01 1.23221660e+00 4.15506244e-01 -2.89971918e-01
-2.39008456e-01 -4.27324831e-01 -1.02958822e+00 2.42896169e-01
3.28191310e-01 -4.25172091e-01 1.66108549e+00 -6.78075194e-01
-1.10688114e+00 1.18188286e+00 -2.25671813e-01 -1.55052453e-01
3.67911994e-01 -4.83827561e-01 -3.98950249e-01 3.80173922e-02
6.24719977e-01 2.84125358e-01 5.78106157e-02 -8.10208976e-01
-7.00289726e-01 -2.39317585e-02 6.02448136e-02 3.67822081e-01
-4.47180003e-01 3.60484689e-01 6.31207153e-02 -3.72396111e-01
-2.95924515e-01 -4.21219528e-01 2.95987159e-01 -7.28866160e-01
-7.53692091e-01 -8.24847043e-01 1.51339561e-01 -1.03793299e+00
1.83645225e+00 -1.82131147e+00 2.69196630e-01 -3.01931560e-01
2.02183202e-01 2.56075233e-01 -1.98042721e-01 6.86826229e-01
3.41603607e-01 6.83214962e-01 -3.18228185e-01 -3.41966897e-01
4.19369280e-01 -1.65996343e-01 -2.67374903e-01 2.17128128e-01
1.46720946e-01 8.23139310e-01 -1.22333503e+00 -6.17378116e-01
-1.46864861e-01 -2.87235081e-01 -3.41116518e-01 3.12163323e-01
-4.81042564e-01 -5.83882220e-02 -2.06157386e-01 1.25654072e-01
-3.46207842e-02 -1.05097055e-01 -6.17456138e-02 -2.79004544e-01
-4.75183398e-01 1.08349597e+00 -7.05659091e-01 1.57488263e+00
-1.79492310e-01 8.55952144e-01 -3.83012444e-01 -3.22971374e-01
8.69948030e-01 4.28204417e-01 -2.11393714e-01 -4.94015515e-01
-9.94280875e-02 3.96294028e-01 4.05337997e-02 -7.39949763e-01
1.18675900e+00 1.21777214e-01 -6.89737380e-01 6.41345978e-01
2.23108128e-01 -6.75318837e-01 9.78776813e-01 6.69046760e-01
1.32074761e+00 -5.84076941e-02 7.67955542e-01 -5.30540526e-01
5.65670013e-01 2.18902126e-01 5.78376114e-01 8.45971048e-01
-1.02246679e-01 6.31342530e-01 9.38361645e-01 -7.77375326e-02
-1.34643328e+00 -6.44490182e-01 -2.41109058e-02 9.63145733e-01
-3.87694597e-01 -1.08295763e+00 -9.66354787e-01 -6.67733133e-01
-3.32927316e-01 1.27684271e+00 -4.36503798e-01 -2.99329758e-02
-5.09927750e-01 -2.99232811e-01 8.83544266e-01 3.93781602e-01
2.67001361e-01 -1.25065863e+00 -7.15237200e-01 2.50242442e-01
-7.63050020e-01 -1.24429619e+00 -6.20057762e-01 -1.72183052e-01
-4.36442584e-01 -1.16685474e+00 -1.21652693e-01 -6.22161806e-01
3.99536163e-01 -4.29666489e-01 1.57666278e+00 3.38948041e-01
3.49733055e-01 2.70153612e-01 -5.25700748e-01 -3.60414922e-01
-1.09273350e+00 3.51764292e-01 7.26109296e-02 -7.15443671e-01
2.92695224e-01 4.72371699e-03 -1.06015071e-01 -1.56728432e-01
-8.16835761e-01 4.67907004e-02 2.77252138e-01 6.13518417e-01
1.92628771e-01 -1.86624601e-01 7.83829212e-01 -1.00086415e+00
1.32503283e+00 8.89246389e-02 -3.38992804e-01 6.67380452e-01
-4.60032284e-01 4.78603132e-02 5.98742604e-01 1.10739045e-01
-9.40156758e-01 -5.39069414e-01 -3.09363566e-02 6.72040522e-01
-1.48881197e-01 7.80679166e-01 -1.39310062e-01 6.05985940e-01
9.17061985e-01 -1.48308918e-01 -4.17063057e-01 -1.96339428e-01
3.23183000e-01 8.83682966e-01 9.34565783e-01 -7.44459629e-01
4.43572134e-01 -3.79779428e-01 -3.07730943e-01 -7.61361718e-01
-1.61020756e+00 -3.93162400e-01 -6.00620985e-01 -2.63914257e-01
8.44488263e-01 -6.86188936e-01 -3.79245579e-01 1.64270595e-01
-1.47664225e+00 -3.75016063e-01 -4.32223141e-01 2.32467093e-02
-7.34405994e-01 6.52166486e-01 -6.78281128e-01 -6.70933604e-01
-7.45572448e-01 -8.49887669e-01 1.12098038e+00 1.79448962e-01
-1.32877183e+00 -1.01377547e+00 2.24679306e-01 6.49923503e-01
1.35212511e-01 4.44685310e-01 1.26699567e+00 -9.70735729e-01
1.87771782e-01 -2.36700580e-01 -4.08462919e-02 4.74656940e-01
2.95604438e-01 5.01047909e-01 -5.91199994e-01 -1.73879400e-01
-6.00340217e-02 -6.02117479e-01 4.74722594e-01 2.39314422e-01
6.48673296e-01 -1.02318144e+00 1.11291304e-01 -2.21267045e-01
1.11605430e+00 -4.12186533e-01 4.16298896e-01 5.26562631e-01
4.49568003e-01 6.96907103e-01 7.32176483e-01 2.65890539e-01
4.22355056e-01 5.57648480e-01 5.66616021e-02 4.12658304e-01
-9.35602635e-02 -1.91743299e-01 5.44010341e-01 1.18357444e+00
2.58815110e-01 -7.12854981e-01 -1.05411363e+00 7.91370630e-01
-1.87944865e+00 -1.09318280e+00 -4.60960299e-01 1.97580492e+00
1.21571767e+00 5.26001811e-01 1.14889696e-01 2.52736151e-01
6.78991616e-01 2.70641655e-01 3.64673100e-02 -9.34081912e-01
-3.14943522e-01 -3.69665213e-02 6.86265007e-02 8.14428031e-01
-7.58847117e-01 9.08903301e-01 6.79840326e+00 4.86450911e-01
-3.78276139e-01 6.97921379e-04 3.91961247e-01 -4.17387486e-02
-5.78148365e-01 2.83720661e-02 -7.30187356e-01 4.64261651e-01
1.16569579e+00 -7.12656319e-01 9.86345951e-03 3.76841426e-01
4.75422323e-01 -3.80723923e-01 -1.50025439e+00 3.57497513e-01
3.85307670e-01 -1.24682164e+00 1.31024912e-01 -2.20142975e-01
8.81234586e-01 -1.73145279e-01 -3.68306339e-01 2.60321051e-01
3.20858479e-01 -9.07033563e-01 1.06052101e+00 5.37063837e-01
9.27918792e-01 -7.18019962e-01 9.37689722e-01 6.50795996e-01
-5.81374109e-01 1.96887553e-01 -2.58168668e-01 -3.29553038e-01
3.55302036e-01 8.20116818e-01 -6.82003677e-01 4.52016532e-01
3.81444722e-01 4.98513401e-01 -1.02638876e+00 8.49020839e-01
-6.15159035e-01 7.34579086e-01 1.66346803e-02 -3.29380363e-01
1.50426909e-01 -3.69316377e-02 8.61750662e-01 1.62686455e+00
2.45399535e-01 8.12026188e-02 1.05964877e-01 8.87339950e-01
-3.29811156e-01 1.87001735e-01 -5.07826805e-01 -2.77406245e-01
6.83836460e-01 1.10013032e+00 -4.80936319e-01 -6.63911521e-01
-6.17726892e-02 7.83099174e-01 6.27460420e-01 -6.51486665e-02
-3.70869696e-01 -5.76029301e-01 3.82382348e-02 -1.41289541e-02
-9.49445292e-02 -1.29664943e-01 -7.00255334e-01 -1.13325691e+00
3.13474953e-01 -1.05709779e+00 4.30552244e-01 -8.52764964e-01
-1.16899395e+00 6.54597700e-01 -4.54036407e-02 -9.05350447e-01
-3.27093422e-01 -2.66962051e-01 -7.33788550e-01 6.50281250e-01
-1.11478746e+00 -5.58228493e-01 -3.74643207e-01 -2.38497645e-01
6.84248507e-01 -1.02783978e-01 9.31417346e-01 -1.70890138e-01
-4.22123462e-01 5.03164589e-01 -3.21813464e-01 4.95262966e-02
1.06052816e+00 -1.66402209e+00 5.21619916e-01 1.17971337e+00
3.59228738e-02 5.07028580e-01 1.28983033e+00 -8.02580833e-01
-5.42693377e-01 -1.01355684e+00 1.69741559e+00 -9.02466774e-01
6.94926202e-01 1.30928811e-02 -1.01445174e+00 7.26591349e-01
7.32441783e-01 -7.56318867e-01 7.84498990e-01 3.07504565e-01
-2.33827934e-01 2.22451955e-01 -9.89684641e-01 3.72400314e-01
9.58795428e-01 -5.21536827e-01 -1.36325479e+00 7.78715730e-01
7.54849434e-01 -5.81709564e-01 -8.67798507e-01 3.27596962e-01
4.80404450e-03 -8.86687815e-01 1.28571376e-01 -5.99619389e-01
9.47271466e-01 -3.13007772e-01 -2.01359272e-01 -1.66127825e+00
-4.22880352e-01 -7.03224123e-01 4.50909473e-02 1.75123274e+00
7.44249105e-01 -8.13486576e-02 9.72528979e-02 8.48050058e-01
-6.51261449e-01 -3.44020277e-01 -8.88519526e-01 -7.87407935e-01
3.23327601e-01 -1.00426190e-01 3.78186971e-01 7.82064795e-01
8.47762048e-01 7.15872109e-01 3.18711922e-02 -1.85160920e-01
3.98081869e-01 -2.40332529e-01 5.94730794e-01 -1.33044219e+00
-2.23399606e-02 -8.12977076e-01 -2.35238299e-01 -5.37674129e-01
4.05473053e-01 -8.60143781e-01 2.41643950e-01 -2.06239223e+00
4.59002942e-01 7.39026591e-02 2.90401787e-01 4.79585677e-01
-4.83668834e-01 -9.01271105e-02 7.75666237e-02 2.22394437e-01
-1.16600537e+00 2.91720092e-01 7.66450822e-01 -1.32266343e-01
-2.18320310e-01 -4.17424440e-01 -1.14864612e+00 6.81529522e-01
6.62554383e-01 -3.90098184e-01 -4.58356086e-03 -5.76144040e-01
7.04426467e-01 -4.85830801e-03 -7.72865862e-02 -1.14840364e+00
2.16573045e-01 -1.45488009e-01 -6.75798878e-02 -4.43225235e-01
-1.98712200e-01 -4.54530008e-02 -2.21970230e-01 1.96021885e-01
-9.35546279e-01 3.53986084e-01 1.48537382e-01 -4.97474195e-03
-2.60873735e-01 -5.83422422e-01 5.75716913e-01 -7.51568675e-02
-1.71700537e-01 -4.75519776e-01 -4.88194048e-01 7.37850189e-01
6.10974669e-01 -4.22450565e-02 -9.78053391e-01 -4.50998157e-01
-2.06752568e-01 3.63709778e-01 7.65557945e-01 2.56856769e-01
2.32368931e-01 -1.04396224e+00 -1.18309832e+00 -6.09897077e-01
4.25266296e-01 -6.09277710e-02 -2.00554356e-01 5.00899553e-01
-4.67804432e-01 5.49777031e-01 -3.75218913e-02 -3.24938267e-01
-1.30701613e+00 4.68912013e-02 1.00775905e-01 -5.91650128e-01
-5.14975011e-01 5.36591768e-01 -3.58408928e-01 -2.85641164e-01
2.07620248e-01 -4.86382961e-01 -3.54553103e-01 2.14694321e-01
7.21792221e-01 6.23249471e-01 3.90653998e-01 -6.72980368e-01
-2.62405217e-01 3.31851780e-01 -8.15543458e-02 -3.10799956e-01
1.36472726e+00 -1.71957970e-01 -4.05875206e-01 8.80856216e-01
8.22356403e-01 4.13293213e-01 -7.02150524e-01 -5.74078038e-02
4.87820059e-01 -1.09457850e-01 -2.33137488e-01 -1.24916530e+00
-1.68982670e-01 4.22166765e-01 -3.72348964e-01 5.12160420e-01
7.12462127e-01 1.43235922e-01 6.68259144e-01 5.79755902e-01
-1.16907507e-02 -1.56413221e+00 2.18434662e-01 8.76562119e-01
1.30083656e+00 -1.06933093e+00 3.60751390e-01 -3.54521513e-01
-7.88877189e-01 1.24897254e+00 6.05004668e-01 9.29493234e-02
-4.17772293e-01 7.84853473e-02 3.77060175e-02 -3.67641479e-01
-8.65930617e-01 1.97892457e-01 5.62036037e-01 3.71551454e-01
9.70922589e-01 2.70137846e-01 -8.70498300e-01 6.88779533e-01
-9.45408165e-01 -3.63320827e-01 1.32779837e+00 6.42523587e-01
-8.95128191e-01 -7.95975447e-01 -8.32674205e-02 7.54135966e-01
-5.29662430e-01 -1.30332515e-01 -9.19064999e-01 5.12253046e-01
-3.51532727e-01 1.37025750e+00 1.21226460e-02 8.42252001e-03
7.11193502e-01 1.72933966e-01 5.76794624e-01 -1.20917118e+00
-9.55170453e-01 -3.67837608e-01 9.49372411e-01 -2.21744746e-01
-3.72322768e-01 -9.70752120e-01 -1.21762800e+00 -2.95828789e-01
-3.38633716e-01 4.98114288e-01 2.57657200e-01 1.23056054e+00
9.57473963e-02 5.16182959e-01 6.94335476e-02 -4.86874163e-01
-7.26883471e-01 -1.39002967e+00 -2.89924651e-01 6.65950119e-01
4.12587881e-01 -1.68170020e-01 -6.95875287e-01 3.41393314e-02]
|
[12.133143424987793, 9.405505180358887]
|
78974d55-3b90-43e9-af35-5b896392a39f
|
g-tuna-a-corpus-of-referring-expressions-in
| null | null |
https://aclanthology.org/W17-3522
|
https://aclanthology.org/W17-3522.pdf
|
G-TUNA: a corpus of referring expressions in German, including duration information
|
Corpora of referring expressions elicited from human participants in a controlled environment are an important resource for research on automatic referring expression generation. We here present G-TUNA, a new corpus of referring expressions for German. Using the furniture stimuli set developed for the TUNA and D-TUNA corpora, our corpus extends on these corpora by providing data collected in a simulated driving dual-task setting, and additionally provides exact duration annotations for the spoken referring expressions. This corpus will hence allow researchers to analyze the interaction between referring expression length and speech rate, under conditions where the listener is under high vs. low cognitive load.
|
['Jorrig Vogels', 'David Howcroft', 'Vera Demberg']
|
2017-09-01
| null | null | null |
ws-2017-9
|
['referring-expression-generation']
|
['computer-vision']
|
[-5.36350161e-02 6.67776391e-02 -3.02995052e-02 -6.04472697e-01
-9.72652495e-01 -6.63069606e-01 5.63904047e-01 1.85318198e-02
-4.30018783e-01 7.04904974e-01 5.02935886e-01 -3.02331746e-01
-2.42107511e-01 -4.07909483e-01 -1.95816662e-02 -3.50821614e-01
2.36197829e-01 2.83083797e-01 -1.49039268e-01 -7.16500401e-01
2.69847274e-01 6.66307867e-01 -1.72490561e+00 8.32479522e-02
1.63526043e-01 5.17096400e-01 4.00656790e-01 6.14011586e-01
1.56979844e-01 7.71712303e-01 -1.13901389e+00 -3.82632375e-01
1.57531556e-02 -7.11601555e-01 -1.06078494e+00 2.78348736e-02
2.40255177e-01 2.84444958e-01 -4.46387559e-01 4.93702620e-01
9.72966015e-01 7.34820604e-01 1.25782475e-01 -1.32057035e+00
-6.63745880e-01 6.67613685e-01 8.03433061e-02 7.63160825e-01
1.04415691e+00 1.79514900e-01 9.39555109e-01 -3.23303223e-01
7.67925024e-01 1.40813315e+00 2.29598150e-01 8.61583054e-01
-1.09625435e+00 -6.53752148e-01 4.87159416e-02 1.73768997e-01
-1.44752324e+00 -9.13650692e-01 9.34412241e-01 -2.43470088e-01
1.33505058e+00 4.90373760e-01 4.94176298e-01 1.49944580e+00
-3.38958018e-02 4.62532252e-01 1.27391112e+00 -4.02846158e-01
1.02224857e-01 1.33106820e-02 1.88160881e-01 -3.04382499e-02
-3.21044177e-01 3.33887488e-01 -9.20578837e-01 -8.61994028e-02
4.70563322e-01 -8.38388264e-01 -4.08914059e-01 3.69325876e-01
-1.21696961e+00 7.74797857e-01 9.82845724e-02 8.19480777e-01
-5.08994460e-01 2.57866144e-01 8.09561729e-01 5.32295823e-01
3.71860087e-01 4.97082531e-01 -1.41266704e-01 -7.84553528e-01
-3.89819562e-01 7.73433626e-01 6.70416594e-01 1.40244567e+00
1.04145609e-01 3.13830733e-01 -5.07539809e-01 1.04592335e+00
1.43893570e-01 4.32134658e-01 3.24245214e-01 -1.04590428e+00
5.10287821e-01 -9.69718471e-02 5.27740121e-01 -7.43113339e-01
-8.55700195e-01 7.61544108e-02 2.51873314e-01 -1.54394388e-01
4.44817483e-01 -4.58276480e-01 -2.93525845e-01 2.19395494e+00
7.17726797e-02 -5.59388399e-01 3.10870886e-01 1.16630912e+00
8.85326564e-01 2.88475096e-01 4.26394761e-01 -4.64647621e-01
1.91680467e+00 -3.57146353e-01 -1.62090921e+00 -3.58209223e-01
9.25576329e-01 -9.35700178e-01 1.45920444e+00 5.72759174e-02
-1.19028246e+00 -7.57329524e-01 -1.00784290e+00 -2.84717649e-01
-2.48284414e-01 -8.35675895e-02 8.15346301e-01 8.14481556e-01
-1.07489717e+00 3.85016352e-02 -3.88687968e-01 -6.57311916e-01
1.57234892e-02 6.30833879e-02 -3.39684069e-01 3.58796716e-01
-1.56655180e+00 1.34549630e+00 1.34468572e-02 2.47797593e-01
-2.50330985e-01 -1.64850876e-01 -1.11576843e+00 -6.04560614e-01
1.76459268e-01 -4.33369398e-01 1.76649833e+00 -6.67112172e-01
-1.53220940e+00 1.38214397e+00 -5.10184586e-01 -2.93453515e-01
2.20064908e-01 -2.52007157e-01 -8.43867481e-01 7.46610388e-02
3.18063408e-01 6.56213820e-01 1.71729833e-01 -9.51175928e-01
-1.95408434e-01 -4.01698500e-01 1.30961806e-01 2.77752548e-01
7.78830588e-01 7.10380673e-01 2.46322334e-01 -5.70858777e-01
3.78124453e-02 -9.31572378e-01 2.07610771e-01 -6.81705058e-01
-1.29638836e-01 -7.97724783e-01 6.79918885e-01 -3.75684738e-01
1.16168606e+00 -2.30742025e+00 3.77529189e-02 -1.96001992e-01
-4.70766351e-02 -2.18834713e-01 -9.58016887e-02 5.43310881e-01
-4.98883933e-01 1.05180338e-01 1.09769493e-01 -1.63161904e-01
4.15142775e-01 4.95632470e-01 -2.74099678e-01 4.33434874e-01
2.76314199e-01 9.52182353e-01 -8.94598842e-01 -5.50710618e-01
1.30968332e-01 2.02293336e-01 -2.38083616e-01 3.84031981e-01
1.94892988e-01 6.34541810e-01 -3.90751779e-01 2.77653486e-01
7.81989172e-02 8.32365274e-01 -3.26424241e-01 6.93601966e-02
-4.86747235e-01 6.67777300e-01 -7.35526919e-01 1.75897872e+00
-8.14842880e-01 1.10364425e+00 7.42015662e-03 -4.73180205e-01
1.23429418e+00 7.20520973e-01 -1.11078657e-01 -1.18943274e+00
7.69449294e-01 -5.84642543e-03 3.79299581e-01 -9.97530043e-01
6.41207516e-01 -7.25607634e-01 -8.93440723e-01 3.55530828e-01
-2.10284024e-01 -8.59506428e-01 5.10905266e-01 -8.87956396e-02
8.80056143e-01 2.50102729e-01 1.87888891e-01 -3.63621950e-01
5.02682626e-01 -1.11705139e-01 2.62303919e-01 4.38989639e-01
-6.43678308e-01 2.20074192e-01 6.29096806e-01 -4.14611876e-01
-7.33017266e-01 -9.45811093e-01 -4.66654688e-01 1.24493873e+00
-5.90338707e-02 -3.16717476e-01 -8.80421519e-01 1.69791296e-01
-5.85519969e-01 1.61741436e+00 -6.84623718e-01 -3.70649338e-01
-6.87409103e-01 -2.29969025e-01 9.14864838e-01 6.35700285e-01
2.59883523e-01 -1.64364719e+00 -1.07936537e+00 2.79308677e-01
-8.31810474e-01 -1.67770576e+00 -6.18558347e-01 3.53572100e-01
-2.08122954e-01 -6.77433014e-01 -2.06943169e-01 -6.94066703e-01
-7.27456212e-02 5.68592362e-02 1.27167344e+00 -2.07141817e-01
-2.04682365e-01 5.84124386e-01 -4.64648098e-01 -7.76277184e-01
-3.98050278e-01 -4.17106688e-01 6.55267686e-02 -4.93254155e-01
7.36300349e-01 -4.65666324e-01 7.32583255e-02 5.38132787e-01
-5.56412578e-01 -9.13233683e-02 -3.92810591e-02 3.87880415e-01
3.58189255e-01 -5.90696514e-01 8.71893883e-01 -4.26023811e-01
1.30712867e+00 -2.37459183e-01 -2.42062882e-01 -4.45753455e-01
2.33196616e-01 -2.00352624e-01 -7.07777590e-02 -4.54609692e-01
-1.09293306e+00 -3.36452603e-01 -4.11619931e-01 2.80976593e-02
-6.27812266e-01 2.14361787e-01 -3.15775365e-01 2.92692274e-01
8.08495641e-01 -3.71592134e-01 -2.18069106e-01 2.11122572e-01
4.91362125e-01 8.28677297e-01 8.57948422e-01 -8.59062433e-01
2.27125585e-01 -2.70464011e-02 -6.93974197e-02 -1.03458560e+00
-8.42918992e-01 -2.12822199e-01 -7.43052065e-01 -6.10277295e-01
1.22483242e+00 -7.61560738e-01 -1.10099316e+00 9.56068635e-02
-1.32581532e+00 -6.99253082e-01 -3.98519337e-01 5.95715463e-01
-1.26597035e+00 -2.54859239e-01 -3.20428640e-01 -1.13730955e+00
7.87003860e-02 -1.17102873e+00 1.23454452e+00 1.41715452e-01
-1.25275016e+00 -8.32261682e-01 -1.27161175e-01 4.71969008e-01
3.98967981e-01 4.29848969e-01 7.98987746e-01 -6.40277922e-01
2.17689916e-01 -3.84864658e-02 5.00610918e-02 -1.95908323e-01
1.68541342e-01 -2.90292054e-01 -1.05961573e+00 4.11226660e-01
3.17802399e-01 -6.08974218e-01 -1.11552447e-01 2.24326968e-01
4.95346576e-01 1.35387123e-01 1.27384454e-01 3.18726525e-02
8.15782368e-01 5.90891600e-01 7.03971803e-01 2.23538443e-01
1.53755546e-01 1.22023487e+00 8.43970537e-01 3.68685067e-01
5.19087791e-01 1.09589827e+00 9.13303420e-02 2.87542969e-01
-8.54003578e-02 1.12400450e-01 3.01313728e-01 6.16377711e-01
8.50440376e-03 -5.28500140e-01 -7.78055847e-01 7.79682398e-01
-1.39759076e+00 -1.10194325e+00 -4.88522470e-01 1.96180928e+00
7.11080909e-01 1.14820153e-01 4.27532107e-01 3.02218765e-01
6.96755350e-01 1.72585651e-01 -2.74028014e-02 -1.16510975e+00
-2.12587744e-01 5.13468802e-01 3.21893245e-02 5.47676206e-01
-7.55892098e-01 1.12458658e+00 7.27788496e+00 3.70034635e-01
-9.16514814e-01 3.85776423e-02 8.77094939e-02 -1.74980342e-01
-1.29046785e-02 -3.47683609e-01 -4.84905720e-01 -1.03185542e-01
1.70456350e+00 -6.26557648e-01 3.94183546e-01 5.31929255e-01
8.21350098e-01 -4.38810080e-01 -1.29985368e+00 1.10342884e+00
1.17712386e-01 -3.95572960e-01 -5.44948637e-01 -3.52298558e-01
-1.07180871e-01 -2.61738479e-01 -4.00214922e-03 4.89362121e-01
-7.50900507e-02 -8.76376271e-01 1.18022907e+00 4.03095007e-01
5.67775428e-01 -8.21471393e-01 8.36489379e-01 2.49322299e-02
-7.93806016e-01 2.33825237e-01 -5.68290167e-02 -5.08437574e-01
7.76636958e-01 -1.39581561e-01 -4.37306046e-01 3.68696958e-01
5.17245293e-01 6.81861266e-02 -3.18336397e-01 4.19712126e-01
-4.65306222e-01 5.96551180e-01 -1.51700407e-01 -4.87334132e-01
2.96551943e-01 -1.42292678e-01 6.88982725e-01 1.31909204e+00
1.59904420e-01 4.73002791e-01 -1.20089740e-01 1.10122871e+00
4.03964669e-01 2.94022501e-01 -1.17703104e+00 -1.56899244e-01
5.49486458e-01 1.04596114e+00 -7.03342259e-01 7.26401955e-02
-2.62959361e-01 7.84155130e-01 2.28790045e-02 3.37923080e-01
-1.08960330e+00 -6.51617646e-01 8.21147144e-01 2.20009610e-02
-3.15000266e-01 -5.55654466e-01 -1.46252006e-01 -1.94626793e-01
2.41615460e-03 -6.93114519e-01 7.31152594e-02 -1.54547215e+00
-9.98742819e-01 1.08565843e+00 6.68052435e-01 -1.06402850e+00
-5.89892685e-01 -3.55224013e-01 -2.97949255e-01 1.45736790e+00
-1.17408729e+00 -6.65607750e-01 -4.12642330e-01 4.74789500e-01
7.34394729e-01 1.98292717e-01 1.09125805e+00 2.65607357e-01
-6.14404202e-01 4.73275363e-01 -1.20762622e+00 -9.86793786e-02
8.38655770e-01 -9.32219446e-01 3.64883006e-01 3.13782632e-01
2.39105411e-02 6.93639398e-01 1.14589560e+00 -1.91500857e-01
-1.05495155e+00 -5.10323822e-01 1.11722088e+00 -7.22595155e-01
7.95958936e-01 -2.88207471e-01 -7.33857512e-01 7.50590563e-01
5.80070972e-01 -1.69484600e-01 9.65773225e-01 -1.08483899e-02
-1.10628111e-02 3.99054080e-01 -1.10702908e+00 6.99009120e-01
1.33505130e+00 -8.54554176e-01 -1.10551810e+00 3.66937071e-01
8.84638190e-01 -8.67869437e-01 -7.99089491e-01 -2.83744410e-02
1.90537289e-01 -9.30873752e-01 4.09932345e-01 -7.22622931e-01
1.65706530e-01 -1.20796852e-01 -3.64114136e-01 -1.26371157e+00
-1.76970288e-01 -1.01779962e+00 7.90628254e-01 1.29068327e+00
3.78156543e-01 -5.38719416e-01 -2.24400703e-02 7.80587733e-01
-6.47887111e-01 -1.76148504e-01 -1.23131800e+00 -6.10204399e-01
-8.72802362e-02 -1.02559650e+00 3.92451406e-01 6.21758699e-01
4.05967057e-01 7.70610750e-01 1.18453026e-01 -2.53553331e-01
-1.98108703e-01 -2.88233489e-01 6.93448722e-01 -1.06462145e+00
2.85068423e-01 -5.31508267e-01 -6.45528436e-01 -7.19493389e-01
9.17824090e-01 -7.99684465e-01 3.95975918e-01 -1.16574872e+00
-3.92048925e-01 -1.35641024e-01 3.63199234e-01 4.27918643e-01
-7.11809769e-02 -8.05807766e-03 2.32370734e-01 -4.04563516e-01
-1.90003335e-01 6.25692070e-01 1.42620337e+00 3.31209093e-01
-2.88683712e-01 7.21039921e-02 -7.72597551e-01 6.30133390e-01
8.30298066e-01 -3.43593776e-01 -5.15535593e-01 -1.63264900e-01
3.31510194e-02 4.32466894e-01 2.31455535e-01 -8.23769867e-01
-2.07447574e-01 -2.23963961e-01 -4.32831198e-01 -5.34644008e-01
6.28590107e-01 -6.54498041e-01 1.34044796e-01 -7.97289163e-02
-5.92087030e-01 7.01065302e-01 1.04088461e+00 4.53485623e-02
-2.73543149e-01 -3.09308678e-01 7.11945415e-01 4.89111319e-02
-6.11206412e-01 -5.35265148e-01 -9.90173459e-01 3.92594784e-01
1.04929054e+00 -4.42665219e-01 -1.05655245e-01 -6.76611900e-01
-1.05903208e+00 2.06532344e-01 -7.93081000e-02 8.88216436e-01
4.66086835e-01 -1.63499641e+00 -7.20248878e-01 -2.47698687e-02
3.83272201e-01 -3.63923609e-01 2.18996048e-01 9.49469745e-01
-2.08763748e-01 6.24681830e-01 -4.44289982e-01 -3.74179065e-01
-1.31721163e+00 6.27622366e-01 3.50066334e-01 3.37497354e-01
-6.26977563e-01 7.43870437e-01 -6.93215132e-02 -1.19618595e-01
-6.45473925e-03 -3.04499000e-01 -3.41953814e-01 3.87386799e-01
5.09136558e-01 1.03105746e-01 1.29879370e-01 -1.41625834e+00
-2.79941142e-01 1.90289155e-01 5.64702094e-01 -6.87141955e-01
7.75665522e-01 -4.46002066e-01 7.74483057e-03 1.40253878e+00
9.17368591e-01 2.71564186e-01 -3.45244825e-01 1.70539036e-01
1.60528734e-01 -1.97581291e-01 8.30805376e-02 -4.33742881e-01
-5.86889982e-01 5.85044980e-01 4.64492023e-01 1.90978944e-01
9.25895333e-01 3.17302704e-01 4.52147573e-01 5.32873690e-01
5.76277375e-01 -1.18708968e+00 1.51900845e-02 5.00304878e-01
1.56464791e+00 -6.39189959e-01 -4.46431369e-01 -4.81706321e-01
-1.01918042e+00 9.50076997e-01 5.75394630e-01 1.71519667e-01
1.63307697e-01 4.86402124e-01 7.28194237e-01 -5.48422098e-01
-8.78060460e-01 -6.86361611e-01 -1.67978302e-01 8.66624773e-01
1.11404634e+00 1.30179331e-01 -7.54524708e-01 7.92377591e-01
-9.94748414e-01 -2.88232952e-01 8.25066268e-01 8.42273057e-01
-3.95927988e-02 -9.37124729e-01 -6.09143555e-01 6.45940155e-02
-4.75364447e-01 6.17322586e-02 -7.01332688e-01 1.33047354e+00
-6.46136850e-02 1.48679507e+00 4.33672756e-01 -2.13497996e-01
1.40349150e+00 5.69833696e-01 6.53126180e-01 -6.74006402e-01
-7.53422678e-01 8.44218582e-02 8.21460724e-01 -4.68801469e-01
-8.87221694e-01 -1.09718037e+00 -1.56513739e+00 -5.00759594e-02
-4.02892888e-01 3.94339889e-01 4.21431690e-01 9.89394784e-01
3.55092064e-02 1.00444531e+00 4.64150131e-01 -9.98606384e-01
1.37422651e-01 -1.42112291e+00 -6.48572624e-01 5.05035937e-01
-8.29500780e-02 -9.97195959e-01 -4.95791644e-01 1.22299418e-01]
|
[10.390190124511719, 9.032584190368652]
|
ca30f4e4-8506-402f-88fd-74a16dc37a56
|
practical-algorithms-for-orientations-of
|
2302.14386
| null |
https://arxiv.org/abs/2302.14386v1
|
https://arxiv.org/pdf/2302.14386v1.pdf
|
Practical Algorithms for Orientations of Partially Directed Graphical Models
|
In observational studies, the true causal model is typically unknown and needs to be estimated from available observational and limited experimental data. In such cases, the learned causal model is commonly represented as a partially directed acyclic graph (PDAG), which contains both directed and undirected edges indicating uncertainty of causal relations between random variables. The main focus of this paper is on the maximal orientation task, which, for a given PDAG, aims to orient the undirected edges maximally such that the resulting graph represents the same Markov equivalent DAGs as the input PDAG. This task is a subroutine used frequently in causal discovery, e. g., as the final step of the celebrated PC algorithm. Utilizing connections to the problem of finding a consistent DAG extension of a PDAG, we derive faster algorithms for computing the maximal orientation by proposing two novel approaches for extending PDAGs, both constructed with an emphasis on simplicity and practical effectiveness.
|
['Maciej Liśkiewicz', 'Marcel Wienöbst', 'Malte Luttermann']
|
2023-02-28
| null | null | null | null |
['causal-discovery']
|
['knowledge-base']
|
[ 3.29529852e-01 4.59758788e-01 -5.18423200e-01 -2.55023360e-01
-3.08803052e-01 -7.48253465e-01 6.13153815e-01 2.90206432e-01
8.17574039e-02 1.12700069e+00 1.30344808e-01 -7.53418326e-01
-7.94721127e-01 -8.16778421e-01 -8.62149119e-01 -6.83770120e-01
-7.13477194e-01 5.38593471e-01 1.28107473e-01 4.37685341e-01
3.27606618e-01 6.45361543e-01 -1.06757760e+00 -3.04989457e-01
8.93590629e-01 2.63530493e-01 9.64972004e-02 7.79528797e-01
3.80503805e-03 5.21989584e-01 -6.83656111e-02 -3.52951556e-01
-1.15351930e-01 -7.12572098e-01 -1.07458234e+00 2.09932685e-01
-2.68353336e-02 4.37932797e-02 -4.89469796e-01 1.11586368e+00
1.06567405e-01 1.08508669e-01 8.15273285e-01 -1.61155415e+00
-1.02275707e-01 1.03868270e+00 -6.17666602e-01 3.58620942e-01
3.44519526e-01 -3.06218207e-01 1.25597072e+00 -5.41208625e-01
6.29277229e-01 1.43315482e+00 1.02733821e-01 2.85706818e-01
-1.66753650e+00 -6.65009975e-01 4.21473980e-01 3.85816097e-01
-1.23485601e+00 -2.55404077e-02 6.96594417e-01 -4.38770682e-01
3.69138390e-01 4.57055956e-01 5.70102155e-01 1.12822664e+00
3.81090492e-01 3.96048069e-01 1.19793200e+00 -4.65057373e-01
5.78040302e-01 -4.07116622e-01 3.90355825e-01 7.40714431e-01
7.56762385e-01 5.23633480e-01 -6.42619312e-01 -6.47815108e-01
6.75085008e-01 -2.80291378e-01 -4.17209327e-01 -6.93822443e-01
-1.03765929e+00 8.71469259e-01 9.74589810e-02 2.87471600e-02
-3.37739259e-01 3.15685630e-01 -1.04974462e-02 9.15922225e-02
1.86817989e-01 2.96621859e-01 -4.59918886e-01 1.80812746e-01
-4.57531959e-01 4.83284712e-01 1.05938923e+00 9.13731337e-01
5.63471138e-01 -4.66130763e-01 1.88930199e-01 9.62610468e-02
4.65476155e-01 4.01647866e-01 -3.28621686e-01 -7.54236341e-01
3.30905020e-01 6.46018267e-01 3.33784312e-01 -9.32187855e-01
-6.05168998e-01 -2.57854849e-01 -9.20096397e-01 -1.61969647e-01
6.04832828e-01 -2.15215489e-01 -6.82812393e-01 2.02718377e+00
7.16651201e-01 5.55301666e-01 -1.89034045e-01 6.23864889e-01
4.44443405e-01 5.65926492e-01 3.57889414e-01 -5.83073795e-01
1.18044913e+00 -2.49514729e-01 -6.16968989e-01 -2.38705441e-01
3.86455297e-01 -4.72029805e-01 5.60471356e-01 3.62633049e-01
-7.54565239e-01 5.58589920e-02 -9.87072587e-01 4.19346333e-01
-7.53572211e-02 -2.60529131e-01 1.02437103e+00 5.33604026e-01
-7.61822939e-01 8.01912308e-01 -8.77242327e-01 -4.30000424e-01
-1.23667978e-02 3.75932932e-01 -3.99356335e-01 -3.04884553e-01
-1.29058146e+00 5.89751363e-01 6.33290410e-01 3.36033970e-01
-1.27454138e+00 -7.24259853e-01 -6.95209563e-01 1.47348955e-01
8.82500708e-01 -8.15691590e-01 1.07586896e+00 -4.24340010e-01
-9.82821107e-01 5.68436980e-01 -3.31962347e-01 -4.12339032e-01
4.84965712e-01 1.68311477e-01 -4.71844614e-01 1.08029786e-02
3.11529934e-02 8.15009400e-02 8.17014933e-01 -1.33412790e+00
-6.45819187e-01 -7.58364558e-01 1.42224148e-01 -1.77114084e-01
1.93952650e-01 5.97901642e-02 -2.61026442e-01 -5.18499732e-01
2.94064760e-01 -1.19700038e+00 -6.95148110e-01 -4.79461312e-01
-8.13800693e-01 -5.22666574e-01 5.26391268e-01 -3.51876795e-01
1.49620199e+00 -1.76678693e+00 3.40349525e-01 5.61499000e-01
5.43500364e-01 -4.94116038e-01 1.16750952e-02 6.35394871e-01
-6.01266980e-01 2.24086747e-01 -5.01380861e-01 3.03928435e-01
-2.15548024e-01 4.00470644e-01 -3.76876086e-01 7.95996368e-01
3.54447812e-01 5.72789371e-01 -1.26914454e+00 -5.21112382e-01
-1.13941142e-02 -2.30864152e-01 -3.40104550e-01 3.24916989e-01
-4.56828713e-01 6.03578687e-01 -7.63315797e-01 1.31881922e-01
6.71918631e-01 -3.14142674e-01 9.85542774e-01 1.56768337e-01
-3.93701494e-01 2.90899128e-01 -1.57072783e+00 1.25021589e+00
-1.15825042e-01 2.96832293e-01 -1.40428632e-01 -1.30807972e+00
7.11643934e-01 4.30412918e-01 4.73592281e-01 -2.81419959e-02
6.60401955e-02 7.82912672e-02 1.59479544e-01 -3.31011713e-01
1.17968939e-01 -1.74800619e-01 -2.67450124e-01 4.20423776e-01
-7.05745146e-02 2.80425400e-01 4.07915115e-01 6.30047441e-01
1.23243833e+00 -3.82517464e-02 7.06672966e-01 -7.49156415e-01
3.49502474e-01 6.50570076e-03 7.17260063e-01 9.74543869e-01
1.53965816e-01 4.72073741e-02 1.28832936e+00 -2.40661129e-01
-8.96650314e-01 -1.21198618e+00 -1.84215099e-01 5.51605701e-01
1.24557659e-01 -5.53709865e-01 -5.63075721e-01 -7.26628125e-01
4.23911810e-02 6.77397251e-01 -9.46546316e-01 -2.31152266e-01
-4.48441714e-01 -9.32486951e-01 2.71894876e-02 3.36615980e-01
8.85169581e-02 -5.86419702e-01 -4.19103891e-01 3.48963916e-01
-3.34728584e-02 -9.02712345e-01 -2.87551343e-01 4.58344251e-01
-1.12302017e+00 -1.73250449e+00 -2.91316807e-01 -3.82801950e-01
8.26159000e-01 2.05426931e-01 1.13683033e+00 -2.07166255e-01
-1.09488413e-01 1.47066519e-01 -1.77414656e-01 -4.58653390e-01
-3.10538232e-01 -2.60450661e-01 1.24735691e-01 -6.38906285e-02
4.75235917e-02 -7.79544175e-01 -3.30071270e-01 3.61953229e-01
-8.20926964e-01 1.46327958e-01 5.11848807e-01 7.95392811e-01
6.13941908e-01 4.66230154e-01 6.31625891e-01 -1.04630804e+00
4.55892116e-01 -7.45416582e-01 -1.16602647e+00 2.93555558e-01
-7.04361558e-01 4.26110089e-01 6.83316767e-01 -3.19978267e-01
-1.11039305e+00 1.21549606e-01 3.20520431e-01 4.07538330e-03
-3.52453850e-02 9.81028318e-01 -5.44689298e-01 1.58944875e-01
3.81837428e-01 -1.62905604e-01 -4.07511652e-01 -4.96500075e-01
4.17510092e-01 1.97281405e-01 4.81151551e-01 -7.46532142e-01
4.93811160e-01 3.06746811e-01 8.45044196e-01 -5.45114040e-01
-7.91271567e-01 -3.43059570e-01 -5.76905012e-01 -2.06965849e-01
5.43205559e-01 -2.92928576e-01 -1.04189849e+00 1.04648639e-02
-1.23513997e+00 -2.66840786e-01 1.12869106e-01 8.45349848e-01
-4.36189711e-01 3.47757250e-01 -1.52422130e-01 -1.00044537e+00
2.62183696e-01 -8.39605212e-01 6.21369958e-01 -3.60674746e-02
-2.20528841e-01 -1.02404797e+00 5.10833621e-01 -1.41036376e-01
-5.13782084e-01 4.11531717e-01 1.52539134e+00 -6.23212218e-01
-6.14928424e-01 -2.31586069e-01 -2.73253947e-01 -3.11980546e-01
2.39076130e-02 2.65864372e-01 -5.59160471e-01 -5.35684936e-02
-2.90369034e-01 2.33601213e-01 7.70932674e-01 7.18326569e-01
1.14487886e+00 -3.36700678e-01 -7.45469570e-01 6.01737835e-02
1.34099019e+00 3.38820070e-01 1.96793288e-01 -1.64936349e-01
5.38561821e-01 7.59242058e-01 5.55537164e-01 4.79860991e-01
2.76950061e-01 4.46303338e-01 5.84135652e-01 1.95537955e-01
2.21572548e-01 -5.21155536e-01 -3.16419974e-02 5.04827321e-01
-2.06493333e-01 -4.17652607e-01 -8.31792235e-01 4.37209904e-01
-2.11332226e+00 -7.53463924e-01 -8.34454000e-01 2.69217920e+00
7.26538360e-01 3.80972028e-02 1.75077617e-01 8.80540088e-02
1.01224911e+00 -1.07498080e-01 -4.92385298e-01 -3.63706231e-01
6.88813180e-02 1.46441028e-01 6.05178773e-01 6.55698717e-01
-8.69767725e-01 5.44278622e-01 6.90331888e+00 5.16257942e-01
-4.48404759e-01 -2.10838407e-01 3.96022737e-01 4.13134515e-01
-4.01776522e-01 6.60650134e-01 -6.80537224e-01 2.89855987e-01
1.04024875e+00 -6.05808914e-01 3.36611599e-01 5.29646754e-01
6.50626600e-01 -4.18213308e-01 -1.37914586e+00 3.46716106e-01
-6.84471250e-01 -1.04758883e+00 3.08894552e-02 3.31728578e-01
8.01962018e-01 -5.28014421e-01 -3.57558697e-01 -1.37533084e-01
9.79178488e-01 -9.11403179e-01 4.17523950e-01 3.44262719e-01
6.06463015e-01 -8.89260769e-01 3.70105058e-01 2.54047453e-01
-1.23918879e+00 7.39927739e-02 -2.29051948e-01 -1.12828933e-01
2.41960809e-01 9.86354768e-01 -8.34029317e-01 9.60374475e-01
4.08255488e-01 4.98068154e-01 -6.18083626e-02 1.09007275e+00
-6.50987387e-01 9.58034396e-01 -2.49248192e-01 -3.00485138e-02
1.61329299e-01 -3.65423918e-01 7.85561621e-01 1.01104748e+00
1.66319281e-01 4.02274728e-01 7.41195977e-02 8.73418510e-01
-1.24629885e-01 -1.15337580e-01 -6.33858740e-01 -3.31577480e-01
5.08136034e-01 9.57076013e-01 -9.21225369e-01 -1.67644009e-01
-2.50278234e-01 4.75027055e-01 3.43321770e-01 3.83442551e-01
-8.52711260e-01 6.55203983e-02 4.27563161e-01 -1.26891434e-01
-3.63894626e-02 -2.15930626e-01 -2.02462688e-01 -9.41662014e-01
-3.13613057e-01 -6.62865281e-01 9.54930246e-01 -4.67396677e-01
-1.09792531e+00 8.69457647e-02 4.21058774e-01 -7.44704783e-01
-3.16842496e-01 -3.71642560e-01 -5.33289433e-01 8.81683350e-01
-9.64029193e-01 -6.13295257e-01 -1.71641335e-01 6.23151779e-01
1.52135059e-01 5.95178187e-01 7.87574589e-01 -8.05778727e-02
-8.58480334e-01 2.18511224e-02 1.06901079e-01 -4.56270456e-01
3.51621479e-01 -1.51361859e+00 1.25397742e-01 1.16833973e+00
-1.95169430e-02 5.88957965e-01 1.16300356e+00 -1.11049640e+00
-1.60570431e+00 -1.01823282e+00 1.09269845e+00 -1.57808855e-01
1.29564321e+00 -2.66802818e-01 -7.69030213e-01 8.96441817e-01
-1.68126538e-01 -1.49447352e-01 3.96152139e-01 6.60612345e-01
-1.84461355e-01 1.83004677e-01 -7.09451377e-01 8.15726280e-01
1.34154356e+00 -2.19671298e-02 -4.68506932e-01 2.48008236e-01
6.00490928e-01 -1.62038952e-01 -8.27898681e-01 3.53023797e-01
4.44210500e-01 -4.61803079e-01 8.30627859e-01 -1.24995697e+00
5.24162769e-01 -3.65899324e-01 1.84772432e-01 -1.43534696e+00
-4.19783235e-01 -7.94335842e-01 -2.44655430e-01 1.09396064e+00
4.12539423e-01 -5.10644794e-01 6.74715221e-01 4.93677258e-01
1.62558749e-01 -4.79946584e-01 -8.82205427e-01 -7.70683825e-01
-2.23900288e-01 -5.90148747e-01 4.69626278e-01 1.07860267e+00
2.01465711e-01 6.32003188e-01 -4.60776746e-01 6.21966898e-01
1.08100581e+00 4.71089542e-01 4.84435290e-01 -1.64293706e+00
-1.92367122e-01 -1.56923920e-01 -2.01530129e-01 -7.10746288e-01
2.70733804e-01 -8.84295344e-01 -3.30948085e-02 -1.29521489e+00
6.56286836e-01 -6.31474435e-01 8.06435663e-03 9.76477116e-02
-3.55212808e-01 -4.66505140e-01 -1.79357022e-01 -1.43113077e-01
-2.13577271e-01 3.84195268e-01 1.18537104e+00 4.16256562e-02
-3.44053894e-01 3.58652323e-01 -7.64492810e-01 8.18148077e-01
7.39009857e-01 -9.08246040e-01 -5.45122623e-01 1.21698089e-01
4.54544812e-01 7.09526837e-01 7.30992079e-01 -1.72731131e-01
2.34115958e-01 -6.41542137e-01 -1.87011272e-01 -5.33414543e-01
-3.24206322e-01 -7.35307455e-01 7.31073678e-01 6.05479538e-01
-4.94933099e-01 -1.75455101e-02 8.61529447e-03 1.09366155e+00
2.44436152e-02 -4.60672379e-01 3.93361896e-01 2.67327446e-02
-5.42907894e-01 3.65665883e-01 -3.68994147e-01 -6.46875277e-02
9.11177874e-01 3.00439239e-01 -1.96685582e-01 -1.80145577e-01
-9.55056369e-01 3.77873927e-01 -9.20797884e-02 7.47295246e-02
3.68993491e-01 -1.10815227e+00 -7.23500431e-01 -4.18758094e-01
1.07508026e-01 -5.19730337e-02 2.90386558e-01 1.00659108e+00
2.45617609e-02 7.47336864e-01 3.01379412e-01 -2.67935812e-01
-1.26748741e+00 1.02045608e+00 5.53985946e-02 -6.02264822e-01
-6.12548828e-01 4.84313846e-01 6.68095052e-01 -9.58852936e-03
5.44290580e-02 -2.74355948e-01 -1.84968188e-01 -2.03595996e-01
4.99987423e-01 7.00123250e-01 -1.25212729e-01 -3.87878455e-02
-4.40815866e-01 -2.71214023e-02 1.26847297e-01 -1.60103083e-01
1.14535320e+00 -2.35002264e-01 -5.04872739e-01 3.93829346e-01
8.44037771e-01 1.29522473e-01 -1.09715819e+00 -3.92728783e-02
3.75690401e-01 -4.20483649e-01 -1.27893938e-02 -5.76521814e-01
-7.73517013e-01 2.98392147e-01 9.72131863e-02 4.92656738e-01
9.16794717e-01 4.38760787e-01 -3.27312946e-02 1.49435356e-01
4.90828067e-01 -4.43724096e-01 -3.83582354e-01 6.07929491e-02
8.90771210e-01 -8.12366068e-01 4.94883209e-02 -9.23839808e-01
-1.07469611e-01 1.17832732e+00 1.85624838e-01 -9.99817811e-03
6.84280694e-01 2.49112532e-01 -7.96649337e-01 -3.78963709e-01
-9.00063336e-01 -5.33223897e-02 2.73036718e-01 4.91599381e-01
3.10892284e-01 4.88368094e-01 -9.68632817e-01 6.64929211e-01
-9.79512930e-03 1.45584628e-01 7.51746655e-01 6.99415565e-01
-1.38965964e-01 -1.13398314e+00 -4.71047938e-01 4.83692408e-01
-3.70216489e-01 -1.50554150e-01 -4.05479014e-01 8.91510367e-01
-2.81202257e-01 1.17441154e+00 -8.49942714e-02 -1.28505692e-01
2.06150115e-01 -2.80296415e-01 6.55037463e-01 -3.17360282e-01
1.29973978e-01 -7.17303976e-02 3.66384447e-01 -5.93104601e-01
-4.27358687e-01 -8.42925012e-01 -1.08800530e+00 -5.30512691e-01
-4.46725130e-01 4.87307906e-01 3.34188461e-01 1.11080468e+00
8.21091700e-03 5.84075749e-01 7.70319581e-01 -2.41173208e-01
-4.25693065e-01 -6.33143723e-01 -9.81336176e-01 -1.35063410e-01
5.35178892e-02 -9.31285083e-01 -4.20534879e-01 6.70278668e-02]
|
[7.716587066650391, 5.2976155281066895]
|
416885cb-9bad-429f-8a25-4b7792620919
|
tm2d-bimodality-driven-3d-dance-generation
|
2304.02419
| null |
https://arxiv.org/abs/2304.02419v1
|
https://arxiv.org/pdf/2304.02419v1.pdf
|
TM2D: Bimodality Driven 3D Dance Generation via Music-Text Integration
|
We propose a novel task for generating 3D dance movements that simultaneously incorporate both text and music modalities. Unlike existing works that generate dance movements using a single modality such as music, our goal is to produce richer dance movements guided by the instructive information provided by the text. However, the lack of paired motion data with both music and text modalities limits the ability to generate dance movements that integrate both. To alleviate this challenge, we propose to utilize a 3D human motion VQ-VAE to project the motions of the two datasets into a latent space consisting of quantized vectors, which effectively mix the motion tokens from the two datasets with different distributions for training. Additionally, we propose a cross-modal transformer to integrate text instructions into motion generation architecture for generating 3D dance movements without degrading the performance of music-conditioned dance generation. To better evaluate the quality of the generated motion, we introduce two novel metrics, namely Motion Prediction Distance (MPD) and Freezing Score, to measure the coherence and freezing percentage of the generated motion. Extensive experiments show that our approach can generate realistic and coherent dance movements conditioned on both text and music while maintaining comparable performance with the two single modalities. Code will be available at: https://garfield-kh.github.io/TM2D/.
|
['Xinchao Wang', 'Zihang Jiang', 'Xinxin Zuo', 'Chuan Guo', 'Heng Chang', 'Dongze Lian', 'Kehong Gong']
|
2023-04-05
| null | null | null | null |
['motion-prediction']
|
['computer-vision']
|
[ 4.98884395e-02 -3.32909912e-01 -1.94787443e-01 5.90633117e-02
-1.01373148e+00 -8.36170673e-01 7.76682973e-01 -6.26601398e-01
-1.22865602e-01 3.93375844e-01 7.43541598e-01 6.75803749e-03
2.18593538e-01 -7.57729828e-01 -6.64481997e-01 -6.66329861e-01
4.30510491e-01 2.52083391e-01 -1.68283097e-02 -2.36413211e-01
8.89550969e-02 -5.54342642e-02 -1.67223704e+00 4.95219678e-01
8.37561786e-01 5.29716492e-01 4.81425226e-01 1.06324577e+00
4.24107723e-03 7.21857429e-01 -7.27254868e-01 -5.75480685e-02
2.96332449e-01 -1.11649680e+00 -6.39714539e-01 -6.18762672e-02
3.43521744e-01 -4.45612282e-01 -3.10387284e-01 5.56786120e-01
8.57719779e-01 5.43630719e-01 6.54232502e-01 -1.23864436e+00
-7.48729765e-01 6.12736046e-01 -3.24352592e-01 -1.96239412e-01
8.19347560e-01 6.54414952e-01 9.63600695e-01 -7.00788319e-01
8.67743552e-01 1.16610599e+00 3.84184211e-01 1.08499682e+00
-1.35127425e+00 -7.09727943e-01 -4.02916282e-01 1.36075258e-01
-1.21934199e+00 -5.59210718e-01 8.34357679e-01 -3.60578388e-01
6.52556002e-01 5.87463379e-01 8.47875297e-01 1.66377223e+00
5.62579418e-03 9.30664241e-01 7.43144453e-01 -3.77472341e-01
2.10056901e-01 -5.18126369e-01 -7.55943537e-01 4.42931414e-01
-3.73264939e-01 1.46701202e-01 -1.11677194e+00 7.59764016e-02
9.66781735e-01 -4.47842032e-01 -3.89449537e-01 -1.30417511e-01
-1.73620725e+00 7.02655733e-01 1.49389356e-01 3.96406740e-01
-2.91365087e-01 4.23317522e-01 1.50886521e-01 -8.19418505e-02
1.23763688e-01 5.03768921e-01 2.71746106e-02 -8.17792714e-01
-1.19590676e+00 5.90863168e-01 4.72886086e-01 8.40230942e-01
3.82804036e-01 2.74987012e-01 -7.32834220e-01 4.85634655e-01
4.99395132e-01 8.27276587e-01 7.11142480e-01 -1.36588264e+00
6.33828998e-01 3.54521036e-01 1.04179680e-01 -9.76001382e-01
-1.83945671e-01 1.56900048e-01 -7.46132135e-01 1.23837464e-01
5.14969110e-01 -1.49876699e-01 -7.19595611e-01 1.98034298e+00
3.80116463e-01 4.34224784e-01 -1.38423946e-02 1.21704686e+00
9.06941235e-01 9.09231901e-01 -5.07131852e-02 1.24214524e-02
1.02694166e+00 -1.08977568e+00 -8.43582392e-01 1.19504340e-01
7.07239270e-01 -1.04617715e+00 1.65934038e+00 1.17798477e-01
-1.27796805e+00 -8.59485090e-01 -9.56480742e-01 -4.08589363e-01
3.53746831e-01 2.17806667e-01 3.51828068e-01 4.75271553e-01
-9.45726454e-01 7.48693943e-01 -1.24587238e+00 -8.96532312e-02
1.11878887e-01 -3.94240543e-02 -1.00889035e-01 5.71436882e-02
-9.49891090e-01 4.00677562e-01 2.06675574e-01 3.15634608e-02
-8.06748092e-01 -6.17738605e-01 -9.80418921e-01 -1.88412681e-01
-1.03744261e-01 -9.63238716e-01 1.20036352e+00 -6.60514534e-01
-2.03621125e+00 4.33625221e-01 -2.32716814e-01 2.45407104e-01
6.19555056e-01 -3.72178137e-01 -4.21591960e-02 1.80010274e-01
2.46108741e-01 1.17268872e+00 5.44920325e-01 -1.21422052e+00
-2.66398132e-01 6.90880045e-02 -1.80948004e-01 6.52935445e-01
-1.37333855e-01 -4.00027931e-01 -7.70695865e-01 -9.57054734e-01
-1.21572830e-01 -1.22084177e+00 8.57917517e-02 -3.62811200e-02
-7.24289596e-01 8.30305442e-02 6.51469946e-01 -5.61457396e-01
1.24459743e+00 -2.21146917e+00 7.39082396e-01 -1.17268637e-02
-2.37485483e-01 -1.20013922e-01 -5.16987920e-01 5.06730020e-01
3.23146194e-01 -4.00131149e-03 -1.58220574e-01 -7.00034499e-01
3.23215514e-01 2.50628531e-01 -4.68958378e-01 8.84612128e-02
7.73006827e-02 1.14590919e+00 -1.04951847e+00 -3.56216758e-01
2.27014825e-01 7.88510859e-01 -8.60821545e-01 3.47375929e-01
-4.77511585e-01 1.29415858e+00 -4.46648806e-01 3.34060490e-01
1.83049843e-01 -1.83841527e-01 2.05716595e-01 -1.78314731e-01
1.26424566e-01 4.11483228e-01 -1.25549543e+00 2.47591853e+00
-3.80308539e-01 5.60952783e-01 -4.58085537e-01 -2.04679757e-01
7.48604178e-01 6.16657555e-01 4.66328412e-01 -7.79890716e-01
1.99614584e-01 1.87568879e-03 -1.73568875e-01 -4.74505752e-01
8.13893914e-01 -2.35513195e-01 -2.71777779e-01 8.82686555e-01
-2.45612934e-02 -4.77482498e-01 2.55155295e-01 1.90886483e-02
9.11169171e-01 8.43415320e-01 -2.60214329e-01 3.19020063e-01
5.38239665e-02 1.71124354e-01 3.35325599e-01 4.95670050e-01
1.14257261e-01 9.84859109e-01 2.50559330e-01 1.29917234e-01
-1.07490230e+00 -1.32363319e+00 4.63065803e-01 1.14890146e+00
3.22177351e-01 -6.58935189e-01 -7.01227129e-01 -3.69075328e-01
-4.24240857e-01 8.14358175e-01 -5.09137869e-01 -1.65671691e-01
-7.92242527e-01 -5.96861959e-01 9.51224864e-01 5.09401739e-01
3.47349614e-01 -1.28612554e+00 -7.48793364e-01 1.88768543e-02
-9.84782219e-01 -9.82311785e-01 -1.09137607e+00 -4.97835010e-01
-7.98715889e-01 -7.97843754e-01 -8.58339787e-01 -6.48786366e-01
1.90627947e-01 6.24260940e-02 9.61361766e-01 1.83397178e-02
-9.22200456e-02 4.31814760e-01 -6.08676136e-01 1.08965479e-01
-6.94762051e-01 9.86518189e-02 7.41956979e-02 -2.66308099e-01
-1.80042628e-02 -6.55908823e-01 -8.22457075e-01 2.73896128e-01
-1.06412983e+00 7.37073064e-01 3.43351454e-01 6.45752668e-01
6.87906981e-01 -4.61497068e-01 1.81993991e-01 -1.47363737e-01
6.01869583e-01 -2.76494712e-01 -1.01331966e-02 -2.01828763e-01
6.79222196e-02 1.65758967e-01 4.18475747e-01 -8.08583915e-01
-8.97651672e-01 1.36266321e-01 -2.77011991e-01 -5.19548357e-01
-1.78161591e-01 2.60762274e-01 -1.88483298e-01 6.10141993e-01
7.49537766e-01 4.04434174e-01 -1.13383278e-01 -3.30820054e-01
8.71632457e-01 4.13726181e-01 7.83491433e-01 -6.29479825e-01
9.39645708e-01 4.02996778e-01 -7.95105025e-02 -5.46164215e-01
-4.66581494e-01 -9.57306400e-02 -4.70995426e-01 -3.63245577e-01
1.31217277e+00 -8.51612866e-01 -7.98876345e-01 4.07402039e-01
-1.06721926e+00 -9.53802466e-01 -4.36577171e-01 7.72507966e-01
-1.05376959e+00 3.06592196e-01 -8.16774964e-01 -4.47922289e-01
-3.19210052e-01 -1.25726140e+00 1.38659966e+00 1.53268158e-01
-8.41935098e-01 -7.55901694e-01 5.77406108e-01 4.85218555e-01
2.32232600e-01 5.97248077e-01 5.96409798e-01 1.16530642e-01
-6.19630635e-01 1.34639204e-01 4.22399551e-01 -5.67168668e-02
4.72872376e-01 8.02141726e-02 -8.16238403e-01 -1.92234829e-01
-2.71067619e-01 -5.43415189e-01 4.60291862e-01 3.73941392e-01
6.24771357e-01 -1.93854302e-01 1.05925597e-01 7.12458134e-01
1.00931275e+00 -3.59247439e-02 6.40933454e-01 2.38718197e-01
9.95013058e-01 3.34813535e-01 5.78229845e-01 6.03672683e-01
4.63604033e-01 1.08059418e+00 1.30447999e-01 2.38199979e-01
-5.87380767e-01 -6.79636061e-01 7.45370805e-01 1.31812608e+00
-3.16545606e-01 -4.05299336e-01 -7.59511232e-01 3.52610856e-01
-1.77744997e+00 -1.26231670e+00 -3.16047110e-02 1.93564510e+00
1.17966056e+00 -2.14659363e-01 2.47059241e-01 1.39025390e-01
3.70947063e-01 1.97911873e-01 -3.54261160e-01 -7.77858421e-02
-3.70992832e-02 3.26330751e-01 -1.88481703e-01 5.47150254e-01
-7.88302124e-01 9.18622136e-01 5.70941687e+00 9.30021465e-01
-1.34556639e+00 -2.34321244e-02 2.14568600e-01 -7.44766831e-01
-6.63512886e-01 -2.50790149e-01 -4.31966126e-01 6.41574502e-01
8.63415837e-01 3.66993714e-03 5.28864264e-01 2.98525989e-01
7.56008029e-01 1.35409907e-01 -1.17839384e+00 9.50768173e-01
2.73928624e-02 -1.39283323e+00 4.19846356e-01 -4.62617800e-02
9.50922668e-01 -3.41847479e-01 4.10253108e-01 2.61668891e-01
1.87929645e-01 -1.12538183e+00 1.15832961e+00 7.44692922e-01
1.08984268e+00 -4.96080309e-01 1.47965387e-01 4.21744734e-01
-1.33386302e+00 4.60187078e-01 1.81389093e-01 -7.26003945e-02
4.56467628e-01 -6.29101396e-02 -4.98582453e-01 6.43519282e-01
4.74796534e-01 6.67302191e-01 -3.03650945e-01 7.28627920e-01
-3.88482392e-01 6.74201608e-01 -2.75592864e-01 1.10715859e-01
9.18750837e-02 -1.94783688e-01 6.65887594e-01 8.55395257e-01
8.62671375e-01 -3.13083567e-02 9.58799049e-02 1.13221991e+00
7.72646815e-02 4.36786823e-02 -4.87711489e-01 -2.44431481e-01
6.08028412e-01 9.50747669e-01 -6.26101315e-01 -1.40098274e-01
3.07339523e-02 1.43529892e+00 -1.20043896e-01 4.30231869e-01
-1.00605786e+00 -4.73570302e-02 5.79605699e-01 -1.61155522e-01
2.84975350e-01 -6.41168773e-01 -3.24918836e-01 -1.24634469e+00
7.58019229e-03 -8.94931376e-01 -3.64887249e-03 -9.46051240e-01
-9.93243754e-01 4.44267184e-01 -1.16357811e-01 -1.77103341e+00
-6.78022921e-01 -1.57325596e-01 -6.65026963e-01 8.69181693e-01
-8.75202894e-01 -1.23851860e+00 -5.00323772e-01 6.09433472e-01
5.22945166e-01 8.57522562e-02 9.00940955e-01 2.82654434e-01
-3.62630993e-01 6.51434898e-01 -5.18633574e-02 1.01585984e-01
9.95465338e-01 -1.13137031e+00 3.78705800e-01 5.84935725e-01
5.66326618e-01 3.56866747e-01 7.84932435e-01 -6.19796157e-01
-1.45089626e+00 -1.02236259e+00 4.82958913e-01 -8.33044529e-01
1.83266699e-01 -2.85625905e-01 -7.11109817e-01 4.81114984e-01
2.20875159e-01 -4.57044184e-01 1.06907666e+00 -3.90256912e-01
-7.59423971e-02 4.72174048e-01 -6.48475707e-01 1.02240098e+00
1.14435327e+00 -4.46826071e-01 -5.10471582e-01 -7.57783651e-02
7.46913791e-01 -7.41287231e-01 -9.27343845e-01 2.56010383e-01
6.51686132e-01 -8.95122170e-01 8.00674915e-01 -3.62530798e-01
9.53318954e-01 -5.68585873e-01 -2.62078673e-01 -1.37673879e+00
-1.00085005e-01 -7.32141435e-01 -2.17879862e-01 1.10599411e+00
2.76486069e-01 3.10080737e-01 9.02814806e-01 2.43451148e-01
-1.42012909e-01 -4.61704254e-01 -9.00709689e-01 -6.81761205e-01
1.90912053e-01 -6.87128246e-01 6.23696744e-01 9.78526592e-01
5.90990037e-02 5.58845818e-01 -7.14267850e-01 -8.06918368e-02
2.81371415e-01 4.01134700e-01 1.21705544e+00 -4.51907605e-01
-7.11432278e-01 -5.52921295e-01 -8.91815946e-02 -1.36331224e+00
-9.98316184e-02 -1.10070848e+00 4.45471525e-01 -1.65651810e+00
2.85073016e-02 -1.67214170e-01 1.91916168e-01 4.71797377e-01
-3.65448058e-01 6.99237466e-01 6.21587515e-01 4.30660099e-01
-5.03232300e-01 1.23589361e+00 1.83839202e+00 9.20899808e-02
-7.34792113e-01 -3.28340560e-01 -3.52647573e-01 5.53048432e-01
7.02055931e-01 -3.77742529e-01 -6.49583995e-01 -7.93399751e-01
1.03799731e-01 4.53966200e-01 3.29772413e-01 -1.21656156e+00
5.11307903e-02 -3.13116640e-01 4.26258147e-01 -5.21821916e-01
6.73370361e-01 -1.70919254e-01 4.64388967e-01 3.70168954e-01
-4.92145091e-01 3.88494171e-02 3.90347183e-01 4.55130994e-01
-9.74914581e-02 3.53167474e-01 4.08695877e-01 1.45987079e-01
-3.09866518e-01 6.63122013e-02 -5.12454987e-01 1.07808329e-01
7.72999406e-01 -2.09132135e-01 -2.04552263e-01 -6.87755585e-01
-8.13817859e-01 9.34015960e-02 6.71076298e-01 6.99200094e-01
6.35960519e-01 -1.96489596e+00 -7.95810401e-01 -5.75072318e-02
2.39999387e-02 -2.27314159e-02 5.08074224e-01 7.84043908e-01
-6.40879512e-01 1.30816489e-01 -3.86837631e-01 -7.39740908e-01
-1.05498016e+00 6.91378787e-02 7.43082352e-03 -1.32839903e-01
-6.69237196e-01 5.41263163e-01 -1.29799405e-02 -5.88621795e-01
-5.83342537e-02 -3.31744701e-01 1.83718111e-02 -2.10326508e-01
5.27937531e-01 3.16817015e-01 -4.41282213e-01 -8.39485168e-01
-6.63708821e-02 6.66960657e-01 6.77619934e-01 -7.99745381e-01
1.05625534e+00 -1.18881010e-01 3.71673435e-01 7.24544048e-01
8.48496079e-01 3.59625012e-01 -1.51210821e+00 2.34492198e-01
-3.79343778e-01 -4.99919057e-01 -4.18248951e-01 -6.17515683e-01
-8.15907478e-01 8.39447498e-01 5.29801369e-01 -2.27595344e-01
1.09424055e+00 -6.60310611e-02 1.22283864e+00 8.26951563e-02
1.02748841e-01 -6.97503746e-01 6.20550454e-01 4.91086423e-01
9.78723049e-01 -8.29277694e-01 -4.04248595e-01 -1.02271684e-01
-9.47225690e-01 9.54055846e-01 6.69129252e-01 5.62418159e-03
1.53720170e-01 6.59527406e-02 3.55356663e-01 1.30187780e-01
-7.42873788e-01 -1.89643651e-01 6.11379147e-01 5.97312510e-01
6.23333275e-01 -2.06844695e-02 -1.67721063e-01 6.03003442e-01
-7.90231407e-01 1.96784973e-01 5.73161960e-01 8.21564078e-01
-9.34862904e-03 -1.20851004e+00 -4.51341361e-01 4.05374262e-03
-7.12150633e-02 -4.20274287e-02 -2.80087471e-01 5.54868102e-01
2.24880219e-01 8.93981099e-01 2.93162048e-01 -7.90256321e-01
2.90865630e-01 7.37023652e-02 7.24391699e-01 -6.03315353e-01
-3.26478034e-01 3.90320987e-01 -4.58541811e-02 -7.71968603e-01
-6.10415757e-01 -4.77434009e-01 -1.45783329e+00 -5.27294457e-01
1.67322114e-01 -7.63323903e-03 2.38893375e-01 7.43025243e-01
4.27512556e-01 6.46550179e-01 3.88151377e-01 -1.47740078e+00
-1.78065911e-01 -1.03960431e+00 -4.00643259e-01 8.04817975e-01
1.01118259e-01 -4.89284128e-01 -5.79639301e-02 5.19497812e-01]
|
[5.799209117889404, -0.17836861312389374]
|
1522420b-7596-4514-a6e4-4381f8ce9210
|
a-simple-lstm-model-for-transition-based
|
1708.08959
| null |
http://arxiv.org/abs/1708.08959v2
|
http://arxiv.org/pdf/1708.08959v2.pdf
|
A Simple LSTM model for Transition-based Dependency Parsing
|
We present a simple LSTM-based transition-based dependency parser. Our model
is composed of a single LSTM hidden layer replacing the hidden layer in the
usual feed-forward network architecture. We also propose a new initialization
method that uses the pre-trained weights from a feed-forward neural network to
initialize our LSTM-based model. We also show that using dropout on the input
layer has a positive effect on performance. Our final parser achieves a 93.06%
unlabeled and 91.01% labeled attachment score on the Penn Treebank. We
additionally replace LSTMs with GRUs and Elman units in our model and explore
the effectiveness of our initialization method on individual gates constituting
all three types of RNN units.
|
['Mohab El-karef', 'Bernd Bohnet']
|
2017-08-29
| null | null | null | null |
['transition-based-dependency-parsing']
|
['natural-language-processing']
|
[-1.34340763e-01 7.75465667e-01 -8.94108117e-02 -8.25835824e-01
-8.21640968e-01 -4.02518749e-01 -8.10903590e-03 -8.96872729e-02
-7.41486132e-01 7.49958038e-01 2.43412405e-01 -8.87600005e-01
6.77435875e-01 -7.95480013e-01 -8.20298076e-01 -4.94562358e-01
-1.07787542e-01 4.68145519e-01 4.10143197e-01 -2.34131277e-01
-1.49651870e-01 3.16017568e-01 -2.04337791e-01 4.85438436e-01
5.62636316e-01 5.80249369e-01 1.79789409e-01 8.30347478e-01
-3.75804365e-01 1.20740914e+00 -9.18738961e-01 -7.20831335e-01
-4.35111187e-02 -4.27084833e-01 -1.11388612e+00 -4.19727176e-01
6.96854219e-02 -5.08039176e-01 -4.57716763e-01 7.95963883e-01
2.34354734e-01 2.50116944e-01 8.88207927e-02 -5.36736012e-01
-6.95519745e-01 1.54619265e+00 -3.02050054e-01 3.39756608e-01
-1.78129114e-02 -2.03840896e-01 1.07458866e+00 -6.12832010e-01
4.45326984e-01 1.14878702e+00 7.64523745e-01 6.12941504e-01
-1.10802245e+00 -6.26145124e-01 4.09668058e-01 -4.52169701e-02
-7.19520211e-01 -5.04897475e-01 5.45434773e-01 5.42606749e-02
1.66701317e+00 -4.84261066e-01 2.41894051e-01 1.10137463e+00
5.84714293e-01 7.36336648e-01 6.94252372e-01 -7.06768274e-01
6.73130993e-03 -2.05429539e-01 7.33751833e-01 1.01175129e+00
2.01257333e-01 -3.00835907e-01 -1.93188593e-01 4.08054776e-02
1.02375531e+00 -1.13826737e-01 1.56106576e-01 1.20922878e-01
-6.75019324e-01 1.03110218e+00 8.84048462e-01 5.49465597e-01
-2.71940440e-01 4.20794547e-01 4.45946217e-01 4.29076433e-01
3.81523639e-01 3.84564847e-01 -9.30295169e-01 5.84628172e-02
-7.49810874e-01 -6.16277277e-01 8.94612968e-01 9.59742665e-01
6.41030371e-01 5.44560552e-01 -1.47761360e-01 8.59401047e-01
2.82942176e-01 1.03321180e-01 5.07180750e-01 -9.03603256e-01
7.80022144e-01 3.52513880e-01 -1.31857526e-02 -1.19838119e-01
-5.89534998e-01 -5.11627376e-01 -3.65695924e-01 1.07492348e-02
2.72559613e-01 -9.39729154e-01 -1.34878957e+00 1.78330696e+00
-1.27487972e-01 1.17236622e-01 4.34075177e-01 3.77716690e-01
8.12073290e-01 8.05015385e-01 5.04059434e-01 -4.02860083e-02
1.18890131e+00 -1.28135109e+00 -8.47462475e-01 -5.78823388e-01
1.38575029e+00 -4.51933146e-01 7.50480473e-01 -1.74913462e-02
-1.46195662e+00 -3.56361985e-01 -1.10367608e+00 -4.20292437e-01
-1.97384357e-01 2.74643958e-01 6.45362914e-01 6.69692278e-01
-1.35121536e+00 7.67905474e-01 -1.35614157e+00 -2.31617257e-01
1.68181226e-01 8.50696683e-01 -4.52186167e-01 9.43561569e-02
-1.30496252e+00 1.25536239e+00 6.36602581e-01 4.59894598e-01
-4.97854829e-01 -6.10026941e-02 -1.12078655e+00 3.71892154e-01
-1.88569933e-01 -5.46153247e-01 1.81741893e+00 -7.60248601e-01
-2.07030988e+00 5.39397776e-01 -4.19159710e-01 -8.70229363e-01
-1.99243098e-01 -2.84926116e-01 -2.33693287e-01 -6.59687668e-02
-3.22549254e-01 7.39201486e-01 3.29625666e-01 -9.14220273e-01
-6.30492866e-01 1.68627277e-02 1.27798274e-01 -4.26059216e-02
-1.86855569e-01 3.64324868e-01 -1.61003664e-01 -3.13368052e-01
3.07324469e-01 -8.34425449e-01 -5.22961318e-01 -9.33006883e-01
-4.41844046e-01 -2.57877737e-01 5.99262893e-01 -9.29696739e-01
1.15707433e+00 -1.89781964e+00 -1.97482437e-01 -1.47705927e-01
-1.50013000e-01 5.96409738e-01 -4.03203726e-01 3.66217375e-01
-2.10471988e-01 3.00454646e-01 -3.55173737e-01 -8.34761739e-01
-2.55359739e-01 7.87062466e-01 -1.66809291e-01 1.20967500e-01
5.61567426e-01 1.09625483e+00 -8.34318519e-01 -4.51154917e-01
3.01589053e-02 4.96651798e-01 -4.83353496e-01 3.67124200e-01
-5.28394692e-02 3.29748154e-01 -2.14547396e-01 3.97232890e-01
4.33921158e-01 1.61798924e-01 6.07473910e-01 2.36943305e-01
-1.62323251e-01 1.14481401e+00 -2.57492542e-01 1.95643997e+00
-6.97661996e-01 6.92604959e-01 -2.42722128e-02 -9.49920654e-01
1.09102356e+00 7.34324515e-01 -2.79672623e-01 -5.14922082e-01
5.05326986e-01 2.33622313e-01 6.55475080e-01 -1.91268191e-01
4.48572576e-01 -3.92694712e-01 -2.41273835e-01 3.66412997e-01
7.16628730e-01 4.19117928e-01 1.02006435e-01 1.54659659e-01
1.42868006e+00 2.86919266e-01 -9.59350839e-02 -6.98548779e-02
1.48947477e-01 -1.75618157e-01 8.54767203e-01 6.88423336e-01
-1.32123195e-02 4.32001173e-01 6.78545713e-01 -4.63006884e-01
-7.99501419e-01 -8.61905932e-01 -1.07304282e-01 1.42158675e+00
-8.25289607e-01 -1.67492792e-01 -8.21922183e-01 -8.97254288e-01
-5.98265111e-01 9.71739352e-01 -7.53442228e-01 -1.16332762e-01
-1.29417706e+00 -8.57320011e-01 8.14811170e-01 1.02207315e+00
4.38945234e-01 -1.50508237e+00 -4.30808127e-01 5.90591967e-01
-9.89186019e-02 -1.41368771e+00 -7.08564073e-02 1.06521571e+00
-1.42642570e+00 -5.21202624e-01 -3.93471807e-01 -1.41381359e+00
7.89637446e-01 -4.38808829e-01 1.07407629e+00 -3.83770913e-02
7.39745915e-01 -4.80523974e-01 -5.56944966e-01 -2.04815611e-01
-5.48462927e-01 7.84339070e-01 -4.47626233e-01 -4.08725858e-01
4.34966832e-01 -5.90012789e-01 -8.82371441e-02 -2.37638980e-01
-3.09138447e-01 -1.54573783e-01 7.90662587e-01 9.08789814e-01
-2.64282320e-02 -4.45409983e-01 4.54291523e-01 -1.45451772e+00
4.33014512e-01 -3.40875417e-01 -3.29315305e-01 -5.39151207e-02
-1.17417678e-01 4.73702669e-01 7.48073578e-01 -2.81839848e-01
-1.38284099e+00 2.97238201e-01 -8.49250436e-01 -1.54478788e-01
-8.83900896e-02 7.30904818e-01 1.28640831e-02 3.90534624e-02
2.44417801e-01 -2.76122183e-01 -6.08380675e-01 -7.43004024e-01
2.91406423e-01 4.37793434e-01 6.75958455e-01 -3.41444522e-01
3.66348565e-01 -1.43612534e-01 -3.92388701e-01 -4.72643912e-01
-9.89836931e-01 -2.96151871e-03 -1.10850739e+00 4.29113835e-01
9.08886552e-01 -7.66627014e-01 -5.21585763e-01 4.41899061e-01
-1.58009982e+00 -9.96886194e-01 -2.80829698e-01 6.09266877e-01
-1.01724848e-01 1.54607212e-02 -1.67905307e+00 -5.90614557e-01
-6.22721374e-01 -9.48732555e-01 2.79357225e-01 1.90402493e-01
-1.23963229e-01 -1.39262331e+00 2.15443209e-01 -1.93656817e-01
3.08491766e-01 -8.97059664e-02 9.45925117e-01 -1.05233526e+00
-1.69208627e-02 -2.67620891e-01 4.58694175e-02 4.64531720e-01
-3.02830264e-02 1.68730929e-01 -1.12087893e+00 -1.94432288e-01
1.65152147e-01 -1.87927693e-01 1.23225951e+00 7.57557750e-01
4.36935425e-01 -2.22995713e-01 -2.83902645e-01 6.73820138e-01
1.24751878e+00 5.70036709e-01 7.75779545e-01 4.24467862e-01
7.52120018e-01 3.02479953e-01 1.82883721e-02 -1.10131480e-01
4.01924431e-01 -3.19834538e-02 2.20390782e-01 -3.33524436e-01
1.49542645e-01 -3.20899457e-01 8.02167833e-01 1.42051876e+00
1.90058835e-02 -2.85366595e-01 -9.80268419e-01 5.92925668e-01
-1.71462035e+00 -5.33013999e-01 -4.27666336e-01 1.89555597e+00
9.28446531e-01 7.05473006e-01 -1.20155670e-01 -5.54797240e-02
7.78295815e-01 -2.02612821e-02 -8.56605545e-02 -1.48569643e+00
1.90243855e-01 1.05937052e+00 8.33453715e-01 8.91652644e-01
-1.06742918e+00 1.63847864e+00 7.35653162e+00 -1.64759636e-01
-1.07381988e+00 4.36978966e-01 5.59013724e-01 1.15617119e-01
3.31173018e-02 2.16591299e-01 -1.05360889e+00 1.19228959e-01
1.75790739e+00 5.20164371e-01 -3.23267043e-01 6.31715536e-01
-4.38338555e-02 -1.04791798e-01 -1.10335755e+00 8.77255797e-02
-4.05757517e-01 -8.85893345e-01 -1.89581700e-02 -1.31605327e-01
4.30741400e-01 5.31538188e-01 -1.63576752e-01 7.12098718e-01
1.27731979e+00 -9.67054069e-01 4.47713107e-01 -5.42571917e-02
5.30319214e-01 -8.09352040e-01 1.07605970e+00 1.34003147e-01
-1.05782747e+00 -1.23062156e-01 -7.50263214e-01 -5.35439193e-01
4.37158495e-01 2.82572210e-01 -8.84426415e-01 2.14477643e-01
4.27836448e-01 3.70811254e-01 -3.17628264e-01 8.67356181e-01
-9.29344952e-01 1.18716633e+00 -3.32155406e-01 2.36362845e-01
7.25258708e-01 -6.46046773e-02 -1.07404599e-02 1.68992257e+00
8.28741714e-02 2.30234653e-01 -1.31644189e-01 4.60092664e-01
-4.61628854e-01 -3.09727818e-01 -1.98273003e-01 1.10049084e-01
4.48491663e-01 1.22348249e+00 -7.30574012e-01 -5.20714939e-01
-5.30175447e-01 9.60077822e-01 9.33892369e-01 5.36472857e-01
-7.47163713e-01 -7.11035013e-01 4.10628200e-01 -3.45077872e-01
7.17636824e-01 -4.05212194e-01 -3.87488902e-01 -1.02252424e+00
-3.24549705e-01 -1.73782229e-01 4.71897036e-01 -4.81547207e-01
-7.32062995e-01 1.05599666e+00 -2.83460677e-01 -4.05736119e-01
-4.24463838e-01 -6.89028144e-01 -1.15110815e+00 1.12458122e+00
-1.66554606e+00 -9.70368683e-01 3.91423494e-01 2.92001456e-01
4.52920705e-01 3.58010173e-01 1.24247909e+00 1.47260815e-01
-1.24582720e+00 7.60569990e-01 -1.36989459e-01 1.06694520e+00
5.67814469e-01 -1.40939963e+00 1.43782830e+00 1.11322308e+00
3.45984995e-02 9.29806888e-01 3.68409932e-01 -5.11981726e-01
-8.73521566e-01 -1.10389841e+00 1.45062339e+00 -2.76448399e-01
7.46252954e-01 -5.57747364e-01 -1.07096457e+00 1.94594133e+00
7.44635940e-01 8.17143023e-02 5.34279585e-01 5.08383214e-01
-2.02535182e-01 2.42868170e-01 -9.78064299e-01 3.08213502e-01
6.60026371e-01 -3.09401572e-01 -1.07351696e+00 -1.63338080e-01
1.04853654e+00 -6.21177495e-01 -9.38028991e-01 2.18958229e-01
3.03494155e-01 -6.51637733e-01 5.32783091e-01 -8.14546525e-01
4.15526569e-01 4.24447715e-01 2.45014265e-01 -1.45448577e+00
-8.00052822e-01 -4.93153989e-01 -1.08401842e-01 1.23128581e+00
1.00888741e+00 -8.98471534e-01 9.91363108e-01 9.04066205e-01
-7.18006015e-01 -5.91732323e-01 -7.33710349e-01 -5.55411637e-01
6.44526601e-01 -1.27436563e-01 2.11894125e-01 8.92082155e-01
4.26517159e-01 8.37671459e-01 -1.35583326e-01 1.40314475e-01
2.16379717e-01 -3.43149573e-01 2.32858777e-01 -1.09476340e+00
-1.79457799e-01 -2.05222405e-02 1.08644344e-01 -1.20295608e+00
2.12198764e-01 -8.09699118e-01 3.09650332e-01 -1.94450414e+00
-1.13977447e-01 -4.48747605e-01 -8.09876084e-01 1.12485695e+00
-8.82437751e-02 -6.89893216e-02 2.06611767e-01 -1.03626221e-01
-1.71273500e-01 1.03214920e-01 8.97152007e-01 2.03561351e-01
-4.32331055e-01 -5.76087423e-02 -4.60042477e-01 8.70362043e-01
1.24346817e+00 -1.08704329e+00 -2.12859493e-02 -1.13950205e+00
-5.63468747e-02 3.10329050e-01 -3.15931320e-01 -8.44218254e-01
2.79078484e-01 3.69456947e-01 3.08707684e-01 -3.60217184e-01
3.02416712e-01 -2.79162645e-01 -4.43077773e-01 7.25585818e-01
-4.47447807e-01 3.33966523e-01 3.83450001e-01 -1.71128973e-01
-9.34303403e-02 -6.96970701e-01 7.10116565e-01 -4.65432256e-01
-3.08227271e-01 -8.20220262e-03 -8.19822252e-01 -1.16686098e-01
2.14062676e-01 4.50513624e-02 -2.73505896e-01 8.83446336e-02
-1.02784133e+00 2.02639312e-01 1.05240673e-01 1.24672987e-01
2.60208040e-01 -8.38318706e-01 -6.41183496e-01 1.43157616e-01
-6.93768561e-01 -2.52828412e-02 -2.13744998e-01 4.54266608e-01
-6.41996741e-01 5.92705846e-01 -3.53955477e-01 -5.43809421e-02
-9.54238296e-01 2.18158484e-01 3.85229379e-01 -6.48527861e-01
-6.54947937e-01 1.34444070e+00 -4.03123647e-02 -5.70481360e-01
2.46069402e-01 -9.93149102e-01 -2.80876905e-01 -1.87722757e-01
2.24906072e-01 -9.83305126e-02 2.71149397e-01 -3.79162908e-01
-3.26296061e-01 -1.57882169e-01 -3.47040147e-01 -3.23544353e-01
1.48346722e+00 2.99126387e-01 -6.25263453e-02 7.30336607e-01
1.15030181e+00 -3.17656159e-01 -1.07830155e+00 -2.25886315e-01
3.90408158e-01 4.51295465e-01 1.29991084e-01 -7.35557377e-01
-1.21129727e+00 1.24084580e+00 -2.16026139e-02 -1.58584848e-01
7.61385322e-01 -2.28903145e-01 1.28925407e+00 7.22827196e-01
9.21549350e-02 -9.44146216e-01 -3.79954547e-01 1.13301909e+00
1.50266483e-01 -8.61799419e-01 -4.05838370e-01 -2.41262883e-01
-3.66109908e-01 1.14111733e+00 6.34299219e-01 -5.64938605e-01
4.37316686e-01 9.43739891e-01 2.59228319e-01 6.44946694e-02
-9.90735292e-01 -1.58575907e-01 -3.82842869e-01 3.31789553e-01
1.06536555e+00 -6.64149672e-02 -3.43640745e-01 7.55045235e-01
-3.11396241e-01 -1.25069082e-01 8.28843951e-01 1.24478829e+00
-6.66814506e-01 -1.54884887e+00 -3.43069830e-03 2.57464677e-01
-8.74444783e-01 -5.81068933e-01 -3.10406834e-01 6.60237551e-01
-3.05205941e-01 9.46294546e-01 3.05115223e-01 -3.85422587e-01
2.82910198e-01 6.76504970e-01 5.14839828e-01 -1.06065905e+00
-1.22142637e+00 3.16982903e-02 4.59893972e-01 -1.15321085e-01
-2.10441023e-01 -3.32907200e-01 -1.86385095e+00 -2.44291559e-01
-6.19727731e-01 4.28617477e-01 5.68335772e-01 1.06815338e+00
8.19470435e-02 7.64252782e-01 4.11721528e-01 -7.11975276e-01
-3.23658168e-01 -1.34342420e+00 -3.24260831e-01 -2.50014365e-01
2.26829678e-01 -1.43505976e-01 2.67496724e-02 -4.73289862e-02]
|
[10.265020370483398, 9.748156547546387]
|
1fc86719-478c-4e57-8d87-5d6f96b5d8f3
|
hgnet-learning-hierarchical-geometry-from
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Yao_HGNet_Learning_Hierarchical_Geometry_From_Points_Edges_and_Surfaces_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Yao_HGNet_Learning_Hierarchical_Geometry_From_Points_Edges_and_Surfaces_CVPR_2023_paper.pdf
|
HGNet: Learning Hierarchical Geometry From Points, Edges, and Surfaces
|
Parsing an unstructured point set into constituent local geometry structures (e.g., edges or surfaces) would be helpful for understanding and representing point clouds. This motivates us to design a deep architecture to model the hierarchical geometry from points, edges, surfaces (triangles), to super-surfaces (adjacent surfaces) for the thorough analysis of point clouds. In this paper, we present a novel Hierarchical Geometry Network (HGNet) that integrates such hierarchical geometry structures from super-surfaces, surfaces, edges, to points in a top-down manner for learning point cloud representations. Technically, we first construct the edges between every two neighbor points. A point-level representation is learnt with edge-to-point aggregation, i.e., aggregating all connected edges into the anchor point. Next, as every two neighbor edges compose a surface, we obtain the edge-level representation of each anchor edge via surface-to-edge aggregation over all neighbor surfaces. Furthermore, the surface-level representation is achieved through super-surface-to-surface aggregation by transforming all super-surfaces into the anchor surface. A Transformer structure is finally devised to unify all the point-level, edge-level, and surface-level features into the holistic point cloud representations. Extensive experiments on four point cloud analysis datasets demonstrate the superiority of HGNet for 3D object classification and part/semantic segmentation tasks. More remarkably, HGNet achieves the overall accuracy of 89.2% on ScanObjectNN, improving PointNeXt-S by 1.5%.
|
['Tao Mei', 'Yingwei Pan', 'Yehao Li', 'Ting Yao']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['3d-object-classification']
|
['computer-vision']
|
[-7.64154410e-03 3.28988254e-01 8.44437033e-02 -6.08350277e-01
-7.85806954e-01 -5.64629495e-01 4.13993895e-01 4.22281981e-01
4.09972608e-01 -5.40540405e-02 -3.73009562e-01 -2.65018523e-01
-2.04109214e-02 -1.38187695e+00 -1.12484586e+00 -4.92533565e-01
-2.60658920e-01 4.80024606e-01 5.16964078e-01 -1.56890899e-01
7.13507831e-02 1.07150519e+00 -1.64079416e+00 2.92113513e-01
9.65404510e-01 1.41045594e+00 1.89749282e-02 9.68120322e-02
-7.16097116e-01 -1.19667530e-01 -2.33101144e-01 -2.23599389e-01
4.11965579e-01 1.70953259e-01 -5.95257998e-01 3.45932215e-01
6.96892440e-01 -3.14462483e-01 1.94296286e-01 1.01988661e+00
-5.60091063e-02 -3.22665907e-02 5.63700795e-01 -1.23935997e+00
-6.35288537e-01 1.32665977e-01 -8.98817122e-01 -4.90394950e-01
1.16621852e-02 -5.01878336e-02 1.19581866e+00 -1.29345560e+00
3.73993576e-01 1.42216468e+00 6.99891686e-01 6.52884096e-02
-9.08688426e-01 -6.68276072e-01 5.61800241e-01 -3.14137012e-01
-1.50757480e+00 1.74153998e-01 1.14147222e+00 -6.28257036e-01
8.16189647e-01 3.27808678e-01 7.23110199e-01 2.64605314e-01
-8.57916996e-02 6.62454605e-01 5.41978598e-01 3.33182812e-02
1.43748522e-01 -3.93629313e-01 4.38891292e-01 7.39090681e-01
4.48296338e-01 -4.06258643e-01 -1.88357756e-01 -1.85179353e-01
1.34090710e+00 4.69499439e-01 7.74216354e-02 -5.86447120e-01
-8.85914087e-01 7.58709013e-01 1.13420486e+00 1.46626264e-01
-5.99545658e-01 3.60837877e-01 9.83089060e-02 -8.06306526e-02
6.22862279e-01 1.22402385e-01 -5.74036837e-01 4.89655226e-01
-7.56191790e-01 2.57557988e-01 3.29439700e-01 1.37148058e+00
1.49789107e+00 -7.60437623e-02 -2.53193732e-02 8.82265329e-01
5.02946854e-01 4.73764271e-01 -3.14354956e-01 -9.88116086e-01
6.46385074e-01 1.45114625e+00 -3.79227027e-02 -1.14404225e+00
-4.78914768e-01 -6.04090631e-01 -9.34315562e-01 2.57203788e-01
-1.67218581e-01 3.28147739e-01 -1.18238497e+00 1.35136271e+00
6.93624973e-01 3.75901997e-01 -3.62432063e-01 8.07853818e-01
1.15728247e+00 5.89647889e-01 8.03520977e-02 2.95467317e-01
1.73469889e+00 -6.84944868e-01 -5.31032383e-02 -6.23200759e-02
4.92867053e-01 -2.30539039e-01 9.91711736e-01 1.81220192e-02
-1.14186192e+00 -6.87368751e-01 -8.64740014e-01 -4.22934860e-01
-3.08667690e-01 8.71834457e-02 7.23163307e-01 6.45268261e-02
-1.01652002e+00 4.41930473e-01 -8.46094072e-01 -3.37197743e-02
7.07920492e-01 4.99040753e-01 -4.01207328e-01 1.37095779e-01
-8.03608179e-01 4.89838310e-02 1.58555359e-01 1.09269172e-02
-5.18117666e-01 -9.80003774e-01 -1.03251672e+00 3.52281034e-01
1.54631138e-01 -1.06905186e+00 1.06595886e+00 -5.41062772e-01
-9.88694489e-01 1.02380860e+00 -4.32764769e-01 -1.09842978e-01
1.76482067e-01 4.06385474e-02 1.53458503e-04 2.18425140e-01
3.46160114e-01 8.20821285e-01 6.90009117e-01 -1.85096145e+00
-8.23691308e-01 -8.26384604e-01 2.27473944e-01 1.38557255e-01
5.32908030e-02 -5.28747499e-01 -8.28621268e-01 -6.27017856e-01
1.17621255e+00 -7.02974260e-01 -1.00070611e-01 1.94181412e-01
-7.01528132e-01 -8.12418401e-01 9.41519380e-01 -4.06513244e-01
1.06419563e+00 -2.34220099e+00 1.49835479e-02 4.76049453e-01
6.02238417e-01 -2.94302076e-01 -6.04514703e-02 2.37363651e-01
-2.93643475e-01 4.51701701e-01 -5.60167313e-01 -5.08987308e-01
-1.12277304e-03 1.22684255e-01 -2.10219741e-01 1.21342361e-01
5.57686567e-01 1.12985480e+00 -7.34516084e-01 -2.70537913e-01
3.44893575e-01 5.41823924e-01 -6.95962548e-01 -9.74291004e-04
-3.73430848e-01 1.76188678e-01 -9.77209091e-01 1.06497467e+00
1.09904468e+00 -4.01198149e-01 -4.44842637e-01 -3.90411496e-01
-1.36412784e-01 2.88804829e-01 -9.93567169e-01 1.58115947e+00
-4.14522380e-01 6.96600527e-02 3.19487005e-01 -6.83984101e-01
1.32234883e+00 -2.39357390e-02 8.25733602e-01 -3.93833458e-01
-9.44984406e-02 3.23112339e-01 -4.29097027e-01 2.82045337e-03
3.17678303e-01 -1.68559343e-01 -2.92502284e-01 -9.73668993e-02
-2.34589711e-01 -4.65009928e-01 -4.14203674e-01 1.08476356e-01
7.16081858e-01 -8.48251581e-03 -9.70700681e-02 -2.01856762e-01
5.68842053e-01 2.65215524e-03 5.61954200e-01 3.06151211e-01
2.94059724e-01 9.18728948e-01 3.86041075e-01 -6.57084584e-01
-9.05146241e-01 -1.35756719e+00 -2.81557620e-01 6.97727263e-01
5.76160610e-01 -4.37402993e-01 -9.20962751e-01 -3.86977315e-01
4.84016120e-01 4.87394661e-01 -5.03684282e-01 -5.59299588e-02
-7.22350597e-01 -1.95776805e-01 5.26308380e-02 7.45387197e-01
5.17745078e-01 -8.55545938e-01 -2.93893814e-01 1.94400072e-01
6.56423420e-02 -1.02577877e+00 -3.25589299e-01 1.68831721e-02
-1.28117001e+00 -8.64859223e-01 -4.15025830e-01 -8.49491835e-01
8.23873639e-01 5.78757882e-01 1.24439120e+00 3.92681956e-01
1.50121942e-01 7.73032531e-02 -3.14995408e-01 -3.82198960e-01
2.70327508e-01 1.16419699e-02 -2.21028537e-01 1.62256032e-01
2.87531614e-01 -7.84362316e-01 -7.08736420e-01 3.88061762e-01
-8.52630019e-01 2.46474281e-01 6.07340991e-01 4.29140747e-01
1.37028849e+00 3.49839479e-02 2.67860323e-01 -7.98172832e-01
-2.00967975e-02 -5.98135889e-01 -5.01771271e-01 -7.78609738e-02
9.85163525e-02 -2.06362382e-01 4.17420030e-01 1.03782259e-01
-5.97148716e-01 1.80996120e-01 -3.04950178e-01 -9.03882802e-01
-3.88680935e-01 3.97232741e-01 -6.58380747e-01 -1.53519273e-01
1.51033416e-01 2.45752200e-01 -3.34906608e-01 -7.53685296e-01
5.28162420e-01 5.51636100e-01 3.87242764e-01 -7.92073846e-01
1.02737689e+00 8.17249715e-01 2.31152266e-01 -8.97844672e-01
-7.48762012e-01 -3.97705853e-01 -8.52104425e-01 -3.19420248e-02
1.09700847e+00 -1.01931071e+00 -7.42664278e-01 4.49563146e-01
-1.36283827e+00 -8.86331722e-02 -4.14707392e-01 -7.74938762e-02
-4.47009712e-01 2.14655235e-01 -4.44814771e-01 -5.51685333e-01
-3.77568364e-01 -1.15998852e+00 1.89681554e+00 1.92202359e-01
8.82890001e-02 -7.49481738e-01 -4.89214689e-01 3.26245070e-01
-1.18586808e-01 6.07191503e-01 1.19231260e+00 -3.00711542e-01
-1.18581820e+00 -1.76046550e-01 -5.73560834e-01 2.11895674e-01
2.58750468e-01 1.44398153e-01 -6.96959078e-01 -5.26544377e-02
-7.32804649e-03 2.45276809e-01 7.84006774e-01 4.08232570e-01
1.69152558e+00 5.12744039e-02 -5.96969128e-01 1.13438499e+00
1.41139078e+00 1.13876455e-01 4.06597406e-01 2.59435922e-01
1.29658508e+00 5.27573705e-01 3.55285466e-01 2.83379376e-01
7.18526781e-01 5.21844923e-01 8.56989682e-01 -3.02243531e-01
-7.92623088e-02 -5.15026450e-01 -2.31050953e-01 6.24665201e-01
-1.23187210e-02 -1.54150590e-01 -1.10251760e+00 4.43548054e-01
-1.44534969e+00 -4.03523475e-01 -8.66055012e-01 2.06520772e+00
2.01689154e-01 1.84924141e-01 -9.84810814e-02 -6.08081259e-02
8.48274350e-01 1.74310192e-01 -7.80959666e-01 -1.09624034e-02
-5.01724593e-02 1.73094824e-01 3.67895275e-01 3.19060922e-01
-1.12882769e+00 1.08945906e+00 4.76424694e+00 8.77199113e-01
-1.00068045e+00 -1.31110892e-01 6.40640140e-01 3.50959212e-01
-7.75467217e-01 5.68364859e-02 -8.89517128e-01 4.19499457e-01
1.68776184e-01 1.64551288e-01 1.29773878e-02 1.06867957e+00
8.09119493e-02 4.06433821e-01 -1.06704748e+00 1.15396035e+00
-4.14726794e-01 -1.32330656e+00 7.55118787e-01 1.85314104e-01
7.62574673e-01 1.20278515e-01 -8.84479061e-02 1.56114832e-01
1.83516741e-01 -8.24285209e-01 8.83658111e-01 3.59622955e-01
1.12224317e+00 -6.82997108e-01 4.11596447e-01 4.58016366e-01
-1.90278161e+00 1.71329841e-01 -4.32411641e-01 1.54588103e-01
2.25506008e-01 7.82473862e-01 -1.87295884e-01 9.96976376e-01
8.28935385e-01 8.49449098e-01 -2.74939269e-01 8.65911067e-01
-6.94878474e-02 2.00933948e-01 -7.63946056e-01 5.02113163e-01
5.19726992e-01 -6.31420970e-01 4.94198114e-01 8.07249010e-01
4.38630909e-01 6.40457034e-01 2.52174586e-01 1.25714588e+00
-3.63153040e-01 -2.75974423e-02 -5.47153711e-01 3.83508891e-01
8.81433129e-01 1.28696287e+00 -8.45540762e-01 -4.01052505e-01
-4.93333966e-01 6.00555658e-01 4.69267070e-01 2.51543790e-01
-6.40361965e-01 -4.50485736e-01 1.11164212e+00 4.73741204e-01
3.62063795e-01 -4.10767496e-01 -8.12242448e-01 -8.02626669e-01
4.69598442e-01 -2.76166171e-01 1.76219136e-01 -8.95089447e-01
-1.45777071e+00 6.51316702e-01 -8.24593659e-03 -1.42085826e+00
5.21620333e-01 -4.84537244e-01 -9.20670450e-01 1.09688556e+00
-1.48840678e+00 -1.41293097e+00 -8.02121460e-01 5.35225809e-01
4.92475003e-01 3.92275155e-01 3.30118537e-01 1.44047230e-01
-3.36788118e-01 3.30208719e-01 -3.77234966e-01 3.08867514e-01
-1.90965444e-01 -1.19525087e+00 9.39272881e-01 4.30171162e-01
-1.28338151e-02 8.56561303e-01 -1.64006114e-01 -8.40721965e-01
-1.29605901e+00 -1.55161226e+00 5.87356687e-01 -4.73535657e-01
4.34506387e-01 -6.54543221e-01 -1.37352121e+00 6.26553953e-01
-5.57328582e-01 1.04671501e-01 2.87212789e-01 -2.41957903e-02
-4.47131336e-01 -2.18299523e-01 -1.09748614e+00 3.40866297e-01
1.41419411e+00 -5.85733056e-01 -6.22349083e-01 3.96033168e-01
1.12727702e+00 -5.74543893e-01 -1.10372424e+00 8.37632477e-01
1.58606499e-01 -8.87151241e-01 1.24845529e+00 -3.45165312e-01
4.14010167e-01 -5.63913167e-01 -1.39364153e-01 -1.05694675e+00
-5.54275811e-01 -2.04163879e-01 -4.02111672e-02 1.21343958e+00
2.88264155e-01 -6.21409655e-01 1.07733524e+00 3.11912328e-01
-8.69172156e-01 -1.18187475e+00 -9.93293226e-01 -6.31333649e-01
3.26221794e-01 -7.62499094e-01 1.21869528e+00 9.02714014e-01
-7.50194430e-01 9.51071531e-02 5.04711211e-01 6.28791153e-01
7.44275272e-01 6.17097855e-01 7.76297390e-01 -1.73491502e+00
2.59951383e-01 -6.96891665e-01 -5.59063494e-01 -1.64204299e+00
6.50700694e-03 -1.14327967e+00 -1.60765618e-01 -2.06195164e+00
-1.62639722e-01 -1.00429511e+00 -1.17574558e-01 4.20535117e-01
-2.66002089e-01 2.21354380e-01 1.35003656e-01 4.15220350e-01
-2.10405320e-01 7.08962739e-01 1.53630078e+00 -1.40657589e-01
-3.72176200e-01 -3.28573510e-02 -7.16057062e-01 9.66798425e-01
6.34387672e-01 -2.75052667e-01 -1.55114964e-01 -8.18493783e-01
6.56253844e-02 -4.03486975e-02 5.07267356e-01 -9.58894789e-01
1.36871561e-01 4.64821421e-02 2.83519745e-01 -1.21657455e+00
5.64779341e-01 -1.00693643e+00 2.84043133e-01 3.93738486e-02
2.76278347e-01 -1.96905866e-01 1.80836320e-01 4.74524617e-01
-4.23383206e-01 2.38798320e-01 6.81448996e-01 -2.11622372e-01
-5.55887640e-01 8.90512764e-01 5.49010158e-01 -2.46749416e-01
1.25009465e+00 -5.78120112e-01 -6.31314330e-03 8.04690942e-02
-8.89698446e-01 5.80260992e-01 6.78870320e-01 4.45008248e-01
8.12451899e-01 -1.44452107e+00 -5.52352846e-01 6.30266428e-01
3.04270536e-01 1.17992878e+00 4.53688592e-01 4.08423036e-01
-8.07054162e-01 3.12157959e-01 5.66395782e-02 -1.15700877e+00
-9.72574353e-01 2.46727064e-01 4.53304470e-01 2.11336374e-01
-9.00983274e-01 1.29663682e+00 9.75463390e-01 -5.77452540e-01
-8.80413949e-02 -9.03872013e-01 7.18143582e-02 -2.13292763e-02
4.82428074e-02 1.83675215e-01 2.15860158e-01 -8.48885417e-01
-4.49794054e-01 1.42641473e+00 2.11329311e-01 4.61975008e-01
1.40366483e+00 1.53734952e-01 -4.66767877e-01 3.88711661e-01
1.18571949e+00 -1.89656332e-01 -1.21411514e+00 -1.22616574e-01
-6.68076575e-02 -4.16012526e-01 4.38840911e-02 -1.35130733e-01
-1.27214766e+00 1.18399560e+00 6.88928440e-02 3.69772196e-01
9.51932371e-01 5.53892255e-01 9.64791656e-01 1.70436814e-01
8.70243192e-01 -3.27626407e-01 -1.56770036e-01 4.97024000e-01
9.24334109e-01 -9.32403266e-01 -2.35376894e-01 -1.23439646e+00
-1.12059921e-01 1.03072786e+00 7.23959446e-01 -4.78688300e-01
8.24853659e-01 -6.48494391e-03 -3.44852269e-01 -8.82433712e-01
-3.95187289e-01 -2.14215070e-01 5.45604706e-01 2.88995653e-01
1.17241792e-01 5.22178531e-01 2.72309691e-01 7.35272050e-01
-3.84175152e-01 -3.27318877e-01 -9.54840556e-02 7.48548448e-01
-6.93552494e-01 -8.11443090e-01 -4.31299120e-01 6.51672244e-01
4.79830429e-02 2.23323815e-02 -4.85673815e-01 8.64580989e-01
5.37876010e-01 5.98835349e-01 7.76131511e-01 -4.98942316e-01
8.37852061e-01 -1.50963470e-01 -6.13224097e-02 -1.09533966e+00
-4.45628703e-01 1.94260664e-02 -4.89063472e-01 -6.61939979e-01
-4.77354899e-02 -5.65033376e-01 -1.79473722e+00 -1.83180884e-01
-2.18072385e-01 1.89458519e-01 7.09855795e-01 5.71767509e-01
7.30804265e-01 5.50138474e-01 8.01746964e-01 -1.20042610e+00
3.93143333e-02 -6.86505973e-01 -6.09479427e-01 5.06312549e-01
3.35048050e-01 -8.27764571e-01 -4.11530405e-01 -2.22871542e-01]
|
[7.99152946472168, -3.539184808731079]
|
88d483dd-b484-4ebc-ae9a-af3c0658975f
|
goca-guided-online-cluster-assignment-for
|
2207.10158
| null |
https://arxiv.org/abs/2207.10158v1
|
https://arxiv.org/pdf/2207.10158v1.pdf
|
GOCA: Guided Online Cluster Assignment for Self-Supervised Video Representation Learning
|
Clustering is a ubiquitous tool in unsupervised learning. Most of the existing self-supervised representation learning methods typically cluster samples based on visually dominant features. While this works well for image-based self-supervision, it often fails for videos, which require understanding motion rather than focusing on background. Using optical flow as complementary information to RGB can alleviate this problem. However, we observe that a naive combination of the two views does not provide meaningful gains. In this paper, we propose a principled way to combine two views. Specifically, we propose a novel clustering strategy where we use the initial cluster assignment of each view as prior to guide the final cluster assignment of the other view. This idea will enforce similar cluster structures for both views, and the formed clusters will be semantically abstract and robust to noisy inputs coming from each individual view. Additionally, we propose a novel regularization strategy to address the feature collapse problem, which is common in cluster-based self-supervised learning methods. Our extensive evaluation shows the effectiveness of our learned representations on downstream tasks, e.g., video retrieval and action recognition. Specifically, we outperform the state of the art by 7% on UCF and 4% on HMDB for video retrieval, and 5% on UCF and 6% on HMDB for video classification
|
['Chen Wang', 'Federico Tombari', 'Joshua L. Moore', 'Alireza Zareian', 'Huseyin Coskun']
|
2022-07-20
| null | null | null | null |
['video-classification']
|
['computer-vision']
|
[-1.52009046e-02 -3.05717587e-01 -3.97419602e-01 -3.36038619e-01
-5.55783629e-01 -4.87726718e-01 4.60217834e-01 1.34981573e-01
-3.48077565e-01 3.77786160e-01 3.36981177e-01 1.22432016e-01
3.33957411e-02 -6.09834075e-01 -7.72251129e-01 -9.67598021e-01
2.64424354e-01 1.40156731e-01 3.41803312e-01 1.78621307e-01
1.21087007e-01 2.99716830e-01 -1.79886103e+00 4.83497024e-01
7.77754545e-01 9.80733514e-01 2.89282858e-01 3.06004971e-01
-6.40827790e-02 1.00776088e+00 -3.99722546e-01 2.84024421e-02
2.32612014e-01 -4.89024401e-01 -7.01883972e-01 6.97468042e-01
6.55664623e-01 -2.97882676e-01 -5.61580598e-01 1.06506193e+00
3.32323641e-01 4.83633459e-01 7.04850435e-01 -1.27730143e+00
-4.26602095e-01 3.08040112e-01 -7.89920866e-01 1.62189633e-01
2.58054167e-01 7.72957727e-02 9.02282000e-01 -8.17815304e-01
6.51106894e-01 1.25397229e+00 1.50890544e-01 6.38594449e-01
-1.12450492e+00 -5.43910086e-01 5.72374344e-01 3.93241256e-01
-1.39037967e+00 -6.71545625e-01 9.13429737e-01 -4.64631379e-01
4.09225464e-01 6.70921281e-02 5.83707571e-01 9.61058080e-01
-2.89159447e-01 1.15835357e+00 1.04518592e+00 -3.85173112e-01
3.78591657e-01 1.65443569e-01 1.95815161e-01 8.02170575e-01
1.60690755e-01 -7.63599649e-02 -7.31427789e-01 1.61220580e-01
6.88394129e-01 4.48344916e-01 -5.62561393e-01 -8.55346203e-01
-1.24284410e+00 8.07637334e-01 4.65565175e-01 1.85508102e-01
-1.54516265e-01 2.39594311e-01 3.08672696e-01 1.86782926e-01
4.72519100e-01 8.69766530e-03 -1.30044729e-01 1.09164096e-01
-1.05233657e+00 -2.26539627e-01 5.18141866e-01 9.84591484e-01
8.78311157e-01 1.10339560e-02 -1.22560963e-01 7.84271479e-01
4.89625275e-01 3.66389245e-01 5.30885100e-01 -1.17426229e+00
3.16468030e-01 6.51511014e-01 5.03811538e-02 -1.23659515e+00
-1.81969747e-01 -2.51264960e-01 -8.99774015e-01 1.71402216e-01
3.68301332e-01 1.60169289e-01 -1.07365441e+00 1.66372538e+00
2.51209110e-01 2.62983561e-01 1.17869206e-01 1.14315486e+00
9.60841477e-01 5.53271532e-01 -1.51516989e-01 -5.17456055e-01
9.22661483e-01 -1.27610981e+00 -7.19877005e-01 -1.07935943e-01
5.77159822e-01 -5.94909847e-01 9.39736485e-01 5.03124475e-01
-9.43152130e-01 -6.78049326e-01 -9.45081592e-01 1.50326252e-01
-1.20118700e-01 2.95017213e-01 6.88011825e-01 5.55293024e-01
-9.25630450e-01 5.54348052e-01 -1.07422447e+00 -5.04221380e-01
5.28371513e-01 1.87038437e-01 -5.93887091e-01 -4.42390203e-01
-6.91518962e-01 4.55376744e-01 2.39115521e-01 -1.10326260e-01
-8.60594511e-01 -3.03206682e-01 -8.02119195e-01 -9.49137658e-02
5.28398395e-01 -4.25587386e-01 6.29855812e-01 -1.07151186e+00
-1.31784248e+00 8.10906291e-01 -4.80477929e-01 -3.73617709e-01
4.76987034e-01 -5.07127523e-01 -2.92592317e-01 7.31385827e-01
1.61176622e-01 8.68231952e-01 1.12240005e+00 -1.56053650e+00
-6.51657581e-01 -4.27043915e-01 -1.90045144e-02 4.01176035e-01
-7.12166488e-01 -3.23484391e-01 -1.13358748e+00 -6.35381103e-01
4.49955672e-01 -9.70234811e-01 -2.69051671e-01 2.32773378e-01
-2.52268463e-01 -1.35722816e-01 1.06433702e+00 -3.22357655e-01
1.13020444e+00 -2.34465981e+00 3.52590710e-01 2.38192633e-01
1.59001872e-01 1.57898217e-01 4.00952548e-02 1.60870329e-01
-6.32909015e-02 -7.61014223e-02 -1.58882409e-01 -4.01549101e-01
-3.57201755e-01 3.15514028e-01 -2.23451614e-01 6.77931488e-01
1.04468301e-01 5.69309592e-01 -1.06260359e+00 -6.23676538e-01
4.55748945e-01 4.06652689e-01 -7.68585742e-01 2.46763304e-01
9.15111508e-03 6.10166907e-01 -2.33139753e-01 5.95730305e-01
4.90440547e-01 -4.31550264e-01 1.43620595e-01 -4.55990046e-01
1.19522236e-01 -8.86872783e-02 -1.41963804e+00 2.00617862e+00
-2.29957014e-01 5.41695774e-01 -2.34254599e-02 -1.32138252e+00
6.86377108e-01 1.92980334e-01 7.91960657e-01 -4.10350621e-01
-1.79596201e-01 -1.54885978e-01 -1.88183069e-01 -4.97932613e-01
4.73482907e-02 6.81050271e-02 2.50313461e-01 5.16279936e-01
2.41599619e-01 2.54327923e-01 3.68382543e-01 4.05446202e-01
9.67469513e-01 2.67193586e-01 6.29345030e-02 -7.48917758e-02
5.13450444e-01 -9.74998400e-02 7.35504925e-01 6.84145093e-01
-2.60702342e-01 9.83301580e-01 2.46886104e-01 -2.83802181e-01
-5.13146579e-01 -1.03377914e+00 1.10070752e-02 9.99575496e-01
5.28668523e-01 -7.48221993e-01 -5.64574242e-01 -9.43970382e-01
-1.54223830e-01 3.28452468e-01 -4.52018291e-01 -3.57021809e-01
-3.97404701e-01 -6.69353187e-01 9.18943733e-02 6.24758005e-01
6.66179597e-01 -7.14771688e-01 -5.69165409e-01 -1.17388321e-02
-4.31294560e-01 -1.25260675e+00 -4.72143918e-01 2.78113246e-01
-1.08725846e+00 -1.29801404e+00 -6.71070516e-01 -7.38668263e-01
1.02151990e+00 8.86166930e-01 8.55678797e-01 2.53805101e-01
-1.74718425e-01 9.38990116e-01 -6.69773281e-01 1.33153901e-01
-7.13717565e-02 -2.55205899e-01 3.39998811e-01 4.50581878e-01
3.40999573e-01 -4.08483088e-01 -8.37815464e-01 5.30228972e-01
-1.00597131e+00 -7.98006728e-02 4.42961037e-01 8.87469530e-01
6.81427181e-01 1.97850987e-01 2.03323781e-01 -7.54753709e-01
-1.80134028e-02 -3.06253552e-01 -2.06353128e-01 2.15112194e-01
-4.03101116e-01 -1.02835476e-01 6.19568825e-01 -4.44348216e-01
-8.75367284e-01 6.36329830e-01 3.13012362e-01 -1.17594516e+00
-3.31706583e-01 2.44042814e-01 -4.02763605e-01 -1.89653970e-02
4.77183849e-01 2.21088976e-01 1.92230508e-01 -3.84594351e-01
4.96098340e-01 5.24159968e-01 4.46112275e-01 -4.21985090e-01
7.65142500e-01 9.29822028e-01 -2.68321306e-01 -9.33869183e-01
-9.14248824e-01 -9.47132289e-01 -9.39460278e-01 -4.13999021e-01
1.01986969e+00 -1.18623722e+00 -5.14498949e-01 1.25592455e-01
-7.87637949e-01 -1.58891559e-01 -2.12861001e-01 6.04654253e-01
-7.12438107e-01 8.18727851e-01 -3.72350156e-01 -6.27348185e-01
7.97440484e-03 -1.13719690e+00 1.01355720e+00 7.92524293e-02
-1.88461449e-02 -8.90293479e-01 -2.69718319e-01 5.64795017e-01
2.53670309e-02 3.72406133e-02 5.26248455e-01 -5.05679488e-01
-6.04915559e-01 -6.91805482e-02 -1.87676430e-01 6.18151188e-01
4.32236671e-01 1.77984871e-02 -1.04364514e+00 -5.27585983e-01
8.07142258e-02 -4.02177036e-01 1.40599763e+00 2.81117618e-01
1.22959852e+00 -1.22126266e-01 -4.58876073e-01 5.93313158e-01
1.32654774e+00 1.24575458e-01 6.30161822e-01 1.53234616e-01
1.04223657e+00 7.84093976e-01 6.80713654e-01 3.41001838e-01
1.03304386e-01 6.66519761e-01 3.81009996e-01 -1.24490194e-01
-2.72708058e-01 -6.30837604e-02 5.67154765e-01 9.02400196e-01
-1.95518032e-01 -7.25155920e-02 -7.65039027e-01 4.55081671e-01
-2.25706363e+00 -1.00938630e+00 3.26726586e-02 2.37153316e+00
5.64058185e-01 9.47236642e-02 1.91282436e-01 2.61027545e-01
7.57843316e-01 1.45953044e-01 -4.20417190e-01 4.38707560e-01
-1.88423768e-01 -1.68591380e-01 2.83373982e-01 2.73588777e-01
-1.54807627e+00 1.03523338e+00 5.75353909e+00 6.89174652e-01
-1.06941760e+00 2.86094975e-02 5.58765709e-01 -2.46566013e-01
-1.56484209e-02 5.56680486e-02 -5.75738728e-01 5.22490799e-01
3.59109521e-01 2.18825027e-01 1.81370363e-01 7.59824276e-01
2.58224189e-01 -2.53586590e-01 -1.28364146e+00 1.22494102e+00
4.46199626e-01 -1.13486338e+00 3.25995892e-01 -7.02927783e-02
7.43319571e-01 -2.32430205e-01 -3.58291827e-02 4.21790555e-02
4.04920615e-02 -8.65168273e-01 5.23174942e-01 5.66747606e-01
4.01816249e-01 -7.33487248e-01 5.48989236e-01 2.91259348e-01
-1.10293722e+00 -1.54299606e-02 -4.76634890e-01 1.04784735e-01
-1.28790615e-02 5.48464417e-01 -1.85240746e-01 5.03195405e-01
9.43878651e-01 1.26269352e+00 -6.96344435e-01 1.08897305e+00
-1.96178108e-01 7.06891596e-01 -3.29744220e-01 4.38794911e-01
1.72986493e-01 -2.92735815e-01 4.17731702e-01 1.13050497e+00
9.46059674e-02 1.10241234e-01 5.54199576e-01 4.64604408e-01
1.25617951e-01 8.00413638e-02 -7.75325775e-01 9.87954140e-02
1.80208683e-01 1.19419730e+00 -9.68957245e-01 -5.09071171e-01
-6.85706198e-01 1.39026833e+00 2.94377834e-01 6.36727750e-01
-6.09698296e-01 -3.87947895e-02 4.81842488e-01 1.28209293e-01
6.58611536e-01 -2.13755980e-01 1.59922555e-01 -1.56178200e+00
3.16856131e-02 -7.88183153e-01 5.60780108e-01 -5.98210931e-01
-1.26783943e+00 2.42876336e-01 -4.06455621e-02 -1.72105169e+00
-1.59707904e-01 -5.08903444e-01 -4.96708423e-01 1.55820206e-01
-1.36975574e+00 -8.40123117e-01 -4.77343112e-01 9.74641383e-01
6.03219271e-01 -2.37263605e-01 7.36476839e-01 4.85334396e-01
-6.80544972e-01 4.58571792e-01 2.79309332e-01 3.18364024e-01
1.02294588e+00 -1.20506871e+00 -3.95227253e-01 9.26170647e-01
6.56921268e-01 6.39981747e-01 4.19687599e-01 -5.05591571e-01
-1.47539270e+00 -1.18299711e+00 2.16885880e-01 -3.05163056e-01
4.13416952e-01 -1.82873741e-01 -9.47352827e-01 4.89803046e-01
9.81156081e-02 4.32953298e-01 7.96259344e-01 -3.96448486e-02
-4.61240977e-01 -2.98642099e-01 -7.70053744e-01 5.14286518e-01
1.15244579e+00 -4.07302022e-01 -3.75872642e-01 5.09225070e-01
5.26795268e-01 -1.67353407e-01 -7.68453777e-01 4.90630746e-01
3.08343917e-01 -1.13441050e+00 1.00036561e+00 -6.73993349e-01
5.44169247e-01 -6.40248418e-01 -4.37862366e-01 -1.23554885e+00
-3.16478163e-01 -4.75366056e-01 -3.98848534e-01 1.22926772e+00
6.62334962e-03 -1.19728364e-01 1.04946375e+00 2.40328461e-01
-4.49965522e-02 -5.62224388e-01 -7.44800329e-01 -8.43027651e-01
-1.86953932e-01 -4.35635716e-01 -2.36677349e-01 1.13238728e+00
1.18819200e-01 3.75013679e-01 -4.89397526e-01 1.80919856e-01
8.40357840e-01 3.72621149e-01 9.03337598e-01 -9.75550354e-01
-3.73133689e-01 -2.72386611e-01 -7.23937571e-01 -1.28331637e+00
1.35386303e-01 -7.87657261e-01 1.08330995e-01 -1.58190727e+00
4.01746660e-01 -2.14042455e-01 -4.83350933e-01 4.03738797e-01
-2.25455299e-01 3.96785140e-01 4.38339800e-01 5.66641748e-01
-1.09627211e+00 5.69531322e-01 1.11655307e+00 -3.09548974e-01
-2.28714690e-01 -4.21501920e-02 -6.22221231e-01 9.26338136e-01
6.08641267e-01 -3.08955997e-01 -5.48679650e-01 -3.81597191e-01
-2.75651097e-01 -1.32543460e-01 4.04709101e-01 -1.11497104e+00
4.61225241e-01 -1.07161468e-02 5.86685240e-01 -5.93527496e-01
3.17319453e-01 -9.56823111e-01 -2.30131149e-01 3.55435163e-01
-2.28272051e-01 -3.18031013e-01 -9.92958546e-02 1.11627626e+00
-5.37134111e-01 -5.63785210e-02 7.34030604e-01 -2.67212033e-01
-9.08075094e-01 3.26566905e-01 -3.58900785e-01 -3.90448011e-02
1.08259261e+00 -2.91009814e-01 -6.06657602e-02 -6.35784328e-01
-9.58327055e-01 4.43455577e-01 5.88943124e-01 4.27110761e-01
8.44347715e-01 -1.39396369e+00 -3.98072422e-01 1.86426267e-01
3.12304288e-01 1.02224939e-01 9.68324244e-02 9.97386932e-01
-2.76141256e-01 2.17860818e-01 -1.61419347e-01 -1.01117110e+00
-1.30822825e+00 7.89279580e-01 4.44689475e-04 8.00436586e-02
-7.70752370e-01 7.07358420e-01 5.73391557e-01 -3.93690281e-02
7.64898717e-01 -6.89177811e-02 -4.05380607e-01 3.20808411e-01
5.34752429e-01 2.35694200e-01 -6.85612261e-02 -7.47421265e-01
-4.27839160e-01 7.59688795e-01 -2.42343530e-01 1.06585491e-02
1.25679946e+00 -3.14460814e-01 9.45325568e-02 6.91439509e-01
1.36502719e+00 -1.61384732e-01 -1.42677736e+00 -4.00848299e-01
-9.52105671e-02 -6.18739069e-01 1.15541354e-01 -2.67755955e-01
-1.53932619e+00 1.03765655e+00 6.25059009e-01 3.74593437e-02
1.19087374e+00 1.57683462e-01 5.34065604e-01 5.74558794e-01
3.42936546e-01 -1.16855872e+00 4.98613864e-01 1.87162742e-01
6.83751166e-01 -1.62618399e+00 2.75081664e-01 -5.75150847e-01
-8.15405190e-01 9.91067588e-01 7.51486123e-01 -2.52625495e-01
7.22237527e-01 -1.67664289e-01 7.03082234e-02 -1.98074013e-01
-6.71801209e-01 -5.93745887e-01 3.91983211e-01 6.01476371e-01
4.04321998e-01 -2.54528493e-01 -6.77319104e-03 2.03615054e-01
2.88840055e-01 -2.42466718e-01 3.08235645e-01 1.05736005e+00
-4.55725968e-01 -8.60823750e-01 -4.41166818e-01 4.47682828e-01
-2.84140229e-01 7.11153448e-02 -4.95602459e-01 6.95691943e-01
3.63166397e-03 1.09559071e+00 2.22604603e-01 -4.74835247e-01
1.85110182e-01 -7.12514445e-02 5.86462200e-01 -8.07343245e-01
-1.19549513e-01 6.19526327e-01 -1.87732220e-01 -8.27452660e-01
-1.09047234e+00 -7.17924297e-01 -1.24831867e+00 1.24414638e-01
-3.62797171e-01 -2.35330816e-02 2.03943655e-01 8.57756734e-01
1.70640573e-01 2.92444438e-01 7.96267092e-01 -9.01878297e-01
-1.86998919e-01 -6.24451160e-01 -5.54618537e-01 8.16803098e-01
2.79354393e-01 -9.51213121e-01 -5.71569920e-01 4.49673921e-01]
|
[8.771631240844727, 0.7204731702804565]
|
6ae14a6a-17af-49ab-94fe-dd2af16b56ea
|
measuring-gender-bias-in-word-embeddings
| null | null |
https://aclanthology.org/W19-3804
|
https://aclanthology.org/W19-3804.pdf
|
Measuring Gender Bias in Word Embeddings across Domains and Discovering New Gender Bias Word Categories
|
Prior work has shown that word embeddings capture human stereotypes, including gender bias. However, there is a lack of studies testing the presence of specific gender bias categories in word embeddings across diverse domains. This paper aims to fill this gap by applying the WEAT bias detection method to four sets of word embeddings trained on corpora from four different domains: news, social networking, biomedical and a gender-balanced corpus extracted from Wikipedia (GAP). We find that some domains are definitely more prone to gender bias than others, and that the categories of gender bias present also vary for each set of word embeddings. We detect some gender bias in GAP. We also propose a simple but novel method for discovering new bias categories by clustering word embeddings. We validate this method through WEAT{'}s hypothesis testing mechanism and find it useful for expanding the relatively small set of well-known gender bias word categories commonly used in the literature.
|
['Alfredo Maldonado', 'Kaytlin Chaloner']
|
2019-08-01
| null | null | null |
ws-2019-8
|
['gender-bias-detection', 'gender-bias-detection']
|
['miscellaneous', 'natural-language-processing']
|
[-3.67086202e-01 1.78219810e-01 -7.77508557e-01 -6.36760414e-01
1.52066723e-02 -7.45578408e-01 9.13485885e-01 8.30805838e-01
-9.94687557e-01 6.34218752e-01 7.58313477e-01 -4.72367853e-01
-1.38134763e-01 -9.73160267e-01 -2.42637217e-01 -3.49663556e-01
-2.73062736e-02 6.55454397e-01 6.73443079e-02 -4.95598048e-01
8.19579065e-01 7.61674643e-02 -1.53310049e+00 -2.92157441e-01
7.32579052e-01 4.28829581e-01 -3.08798850e-01 1.67137861e-01
-1.32319108e-01 -1.01643063e-01 -5.89146554e-01 -7.36607075e-01
-1.82028692e-02 -1.19295143e-01 -7.49446571e-01 -4.31266814e-01
5.73536336e-01 -1.31021053e-01 -7.03872964e-02 1.31713867e+00
9.44611073e-01 -1.49662429e-02 1.03173995e+00 -1.33644259e+00
-1.00427663e+00 5.29152930e-01 -6.29432023e-01 7.12811112e-01
2.12947905e-01 -2.77738035e-01 1.36428106e+00 -8.54634047e-01
1.00792551e+00 1.61440468e+00 6.98489487e-01 9.73691761e-01
-1.26101565e+00 -1.18971455e+00 1.10682636e-01 9.83439982e-02
-1.20401645e+00 -3.11200947e-01 6.42532408e-01 -6.37194395e-01
7.37826765e-01 1.84712678e-01 6.33633316e-01 1.61352837e+00
4.21844244e-01 1.54437527e-01 1.38711369e+00 -4.61038172e-01
2.57523954e-01 5.81763089e-01 5.89024127e-01 4.46309775e-01
1.11926091e+00 3.53287719e-02 -6.94816113e-01 -5.50899148e-01
3.28113258e-01 -1.28352836e-01 1.52598977e-01 -4.57854718e-01
-1.18368304e+00 1.42535746e+00 -4.33439426e-02 6.15744948e-01
1.39324144e-01 7.79356062e-02 7.70762861e-01 2.58363336e-01
1.07529402e+00 9.19323385e-01 -5.37555814e-01 -1.50323719e-01
-7.26459920e-01 7.50108898e-01 6.17468774e-01 5.27378678e-01
6.53532863e-01 -1.88871682e-01 -9.10215005e-02 1.11823809e+00
5.85033774e-01 5.07330954e-01 8.78823280e-01 -4.02033865e-01
1.84094563e-01 4.75064158e-01 6.76642433e-02 -1.60642457e+00
-4.99183536e-01 -3.60230580e-02 -1.95318714e-01 -1.43929094e-01
5.84667206e-01 -1.15445279e-01 -6.72169805e-01 2.06241250e+00
3.63855392e-01 -4.16284233e-01 -3.11357707e-01 7.92671680e-01
1.09315515e+00 1.41851455e-01 4.63182032e-01 1.98923305e-01
1.93108499e+00 -1.21378243e-01 -8.69259894e-01 -5.23483455e-01
7.61436641e-01 -6.49795532e-01 9.78515148e-01 4.96968962e-02
-6.16367459e-01 -4.11051244e-01 -1.06792736e+00 -1.26252443e-01
-9.71669972e-01 -3.95874530e-01 9.08110023e-01 1.37836432e+00
-6.04697704e-01 3.42937946e-01 -2.84278572e-01 -9.46744263e-01
5.69549620e-01 2.09785715e-01 -3.19099933e-01 -4.42048013e-02
-1.55490077e+00 1.13422656e+00 8.21401924e-02 -7.02811062e-01
-3.92640769e-01 -9.78817165e-01 -1.06053483e+00 -4.07173932e-01
-1.89850137e-01 -4.09049481e-01 7.60499179e-01 -8.17719519e-01
-4.08185601e-01 1.62408054e+00 -4.34914589e-01 -1.28348708e-01
8.42370316e-02 5.00722490e-02 -9.01126981e-01 -2.74470448e-01
6.57166660e-01 7.53637195e-01 6.34394228e-01 -1.17722487e+00
-6.22333169e-01 -7.71910608e-01 2.50719695e-05 -2.22153410e-01
-1.13114321e+00 4.04097348e-01 3.74148309e-01 -8.64983737e-01
-2.66292214e-01 -7.40775943e-01 -7.12430403e-02 -1.07329212e-01
-6.82424307e-02 -8.75938475e-01 4.22490269e-01 -5.22670627e-01
1.57639492e+00 -2.09146881e+00 -1.47202685e-01 2.50940025e-01
5.11965632e-01 -6.92002252e-02 -3.36015262e-02 4.06379610e-01
-2.35115483e-01 5.88945806e-01 1.07453270e-02 1.97322257e-02
4.58529979e-01 5.68567455e-01 -3.31089377e-01 7.82941103e-01
2.50598609e-01 5.48654497e-01 -1.17113614e+00 -6.98461354e-01
-1.86974794e-01 1.71286151e-01 -9.88492548e-01 -2.21593738e-01
2.31392726e-01 -8.99046063e-02 -9.72295180e-02 6.84771478e-01
8.62806678e-01 4.83029515e-01 5.06418407e-01 -5.21919802e-02
-4.19861823e-01 4.90871400e-01 -9.14958179e-01 1.13275528e+00
-1.78610146e-01 9.26775157e-01 -3.01437289e-01 -1.02428627e+00
1.14289510e+00 -1.54057562e-01 1.32059887e-01 -7.47744262e-01
5.12811303e-01 4.42170084e-01 5.53548396e-01 -3.96690398e-01
1.07652426e+00 -6.80574059e-01 -5.97319126e-01 5.51810920e-01
4.48384315e-01 3.97162139e-02 5.25442362e-01 -2.52049919e-02
7.25466490e-01 -4.60294664e-01 2.49417067e-01 -1.08419371e+00
6.40358627e-02 9.21004713e-02 7.10209966e-01 4.78601068e-01
-6.52990639e-01 4.57617939e-01 9.36670065e-01 -4.46482301e-01
-9.96153235e-01 -9.85565543e-01 -9.02421296e-01 1.60379720e+00
9.73182172e-02 -5.56753576e-01 -2.98389882e-01 -7.42935419e-01
5.31513751e-01 9.08801198e-01 -1.27435648e+00 -1.70404121e-01
-2.74077475e-01 -1.33054960e+00 5.47838271e-01 4.63137954e-01
-4.62333262e-02 -6.35877192e-01 -4.60970312e-01 -2.60991454e-01
1.80268556e-01 -7.76047468e-01 -2.51209080e-01 1.72635838e-01
-6.56840980e-01 -1.17739618e+00 -6.27151012e-01 -9.29603279e-01
6.05509341e-01 -1.24865517e-01 1.48294425e+00 -3.23760584e-02
-3.01292986e-01 4.26509619e-01 -3.50981593e-01 -1.02347434e+00
-2.06519723e-01 2.70964414e-01 5.85757256e-01 -4.56580043e-01
1.27226639e+00 -3.32412064e-01 -6.20857716e-01 3.33417833e-01
-7.84132600e-01 -8.65225315e-01 8.09384733e-02 7.75130630e-01
-8.96442831e-02 -1.51348010e-01 8.45519423e-01 -1.09936237e+00
1.01210523e+00 -1.14174819e+00 4.07588966e-02 -5.71988463e-01
-1.12433958e+00 -2.13038385e-01 -4.14204970e-02 -5.04658401e-01
-7.48696566e-01 -9.85196829e-01 -2.14738488e-01 4.07599092e-01
-7.55132437e-02 2.25706235e-01 1.46973848e-01 3.17553133e-01
8.84738266e-01 -4.16397840e-01 1.79794952e-01 -3.27293217e-01
1.64883584e-01 8.42111647e-01 -1.17294796e-01 -7.67377257e-01
6.84951067e-01 4.76420730e-01 -3.14597517e-01 -8.61058235e-01
-6.01567805e-01 -4.63816375e-01 -3.53608012e-01 -4.40835170e-02
1.12700677e+00 -8.12379599e-01 -4.47699547e-01 -1.25945151e-01
-9.76860404e-01 8.31869766e-02 -8.73613283e-02 5.11392355e-01
1.08422264e-01 2.42808849e-01 -1.94542259e-01 -5.62022865e-01
4.15732525e-02 -5.15266895e-01 8.66402149e-01 1.33659348e-01
-1.19804668e+00 -1.45299387e+00 5.99885404e-01 4.24433723e-02
2.01080918e-01 5.33053130e-02 1.13205481e+00 -1.06470108e+00
6.17793918e-01 -9.57222208e-02 -3.63872707e-01 1.26064315e-01
2.27864906e-01 1.20044217e-01 -8.96339238e-01 -2.35825449e-01
-4.01568592e-01 -2.26719424e-01 1.03474224e+00 2.04191461e-01
9.27402973e-01 -1.74965803e-02 -5.22136152e-01 1.03635430e-01
1.47462988e+00 7.94343948e-02 3.67425859e-01 6.26073897e-01
5.82175195e-01 8.87339234e-01 6.91835642e-01 3.91195357e-01
5.53190172e-01 2.97309905e-01 2.85892367e-01 1.35811337e-03
2.54116178e-01 -1.29888386e-01 2.49876052e-01 5.52399993e-01
-1.55179366e-01 3.27212922e-02 -1.18563116e+00 1.34431636e+00
-1.39187276e+00 -9.84519482e-01 -1.69271052e-01 1.92635536e+00
9.04754162e-01 1.82627514e-01 4.13862109e-01 2.34213367e-01
7.28890717e-01 5.33285081e-01 -5.88322319e-02 -1.26213205e+00
1.13488324e-01 7.26230443e-01 6.83266282e-01 1.94280371e-01
-1.09448612e+00 6.58673823e-01 7.02874374e+00 6.15329802e-01
-8.12662482e-01 9.84755680e-02 4.92490828e-01 6.61924928e-02
-6.79929793e-01 -3.12239170e-01 -8.52176309e-01 4.96786445e-01
8.25317562e-01 -4.19484943e-01 -3.72431993e-01 8.36490452e-01
-5.95216788e-02 -5.08028716e-02 -1.16148615e+00 8.32914591e-01
4.05611277e-01 -9.53771651e-01 -6.91159964e-02 2.43804440e-01
5.41817307e-01 -2.24052846e-01 2.15274334e-01 5.37441790e-01
2.84416199e-01 -1.21444821e+00 5.90245783e-01 -1.22438893e-01
7.44462132e-01 -9.39653814e-01 8.54600072e-01 -2.27027982e-01
-3.86261940e-01 -2.87255108e-01 -6.77237689e-01 -5.55006027e-01
-2.85246283e-01 8.34779263e-01 -7.19500899e-01 1.51669621e-01
9.76723492e-01 7.21599460e-01 -8.59008968e-01 4.44152594e-01
6.26898557e-02 7.17144608e-01 2.08710998e-01 -4.27750200e-01
5.52721582e-02 5.20839309e-03 3.60410541e-01 1.37874055e+00
3.21607918e-01 -5.00405550e-01 -2.47066721e-01 6.22928083e-01
3.25584337e-02 2.32610092e-01 -9.36270714e-01 -4.21563148e-01
5.75376689e-01 1.30643022e+00 -8.40953112e-01 -1.75325513e-01
-5.32251656e-01 4.74584043e-01 -1.17208526e-01 -7.41868913e-02
-6.23891413e-01 -5.33265471e-01 1.43967116e+00 2.42714792e-01
4.61901166e-02 -2.54640937e-01 -5.79221308e-01 -9.72674012e-01
-5.20113111e-01 -8.25497866e-01 4.72247988e-01 -1.11765683e-01
-1.77608061e+00 -1.05180986e-01 2.76741177e-01 -8.70001078e-01
-3.32746096e-02 -1.06440127e+00 -3.77111971e-01 6.57982230e-01
-1.37624049e+00 -7.87973940e-01 2.47779824e-02 1.23774819e-01
2.28311777e-01 -2.96181381e-01 1.02900624e+00 6.15948796e-01
-4.98589426e-01 7.56864488e-01 -2.21520007e-01 2.75388151e-01
1.42934823e+00 -1.48512304e+00 2.70167410e-01 1.49380848e-01
-2.88535029e-01 1.07566094e+00 1.22003675e+00 -6.06783032e-01
-9.68573928e-01 -7.82218337e-01 1.64544511e+00 -8.63029540e-01
8.60728025e-01 -7.58590102e-01 -3.66480947e-01 6.54237807e-01
5.18227756e-01 -3.07427883e-01 1.49833584e+00 7.55858541e-01
-8.56427670e-01 5.72898202e-02 -1.47787225e+00 7.65857279e-01
1.19868267e+00 -3.28336328e-01 -1.14597750e+00 9.71274227e-02
3.88184100e-01 9.26385447e-02 -8.08635175e-01 1.31007090e-01
8.43744397e-01 -8.94738138e-01 9.56607938e-01 -9.02718484e-01
1.06795776e+00 2.23947212e-01 -4.04881001e-01 -1.71679604e+00
-4.31086600e-01 1.52631402e-01 4.59606558e-01 1.58805537e+00
2.60702461e-01 -1.05560076e+00 5.68129182e-01 4.44863319e-01
3.46443057e-02 -5.99251568e-01 -1.05300057e+00 -6.29553139e-01
9.70715582e-01 -2.76993126e-01 7.53961265e-01 1.68502951e+00
1.99102700e-01 4.14925247e-01 5.53544946e-02 -2.45571420e-01
5.44349492e-01 -2.34210920e-02 7.12093532e-01 -1.63694227e+00
2.86926508e-01 -6.13596320e-01 -7.26796389e-01 -2.69582361e-01
4.58401293e-01 -1.00192511e+00 -3.21113765e-01 -1.41811192e+00
4.31201488e-01 -2.89492041e-01 -2.55752236e-01 1.12488337e-01
-1.45958632e-01 6.03067935e-01 -3.15737247e-01 -4.42612648e-01
-1.56345621e-01 3.57242197e-01 9.94194388e-01 -1.61886454e-01
3.77720267e-01 -7.07313359e-01 -1.48672962e+00 8.80083919e-01
8.30447197e-01 -5.31392694e-01 -3.26043218e-01 -2.77157456e-01
6.59197807e-01 -1.08997703e+00 2.39930496e-01 -4.03926373e-01
-4.94507194e-01 -2.02489898e-01 5.78277230e-01 -6.81339353e-02
-1.11013591e-01 -5.30514598e-01 -7.09131598e-01 4.80229050e-01
-1.55583158e-01 2.84647137e-01 2.97201723e-01 3.17709506e-01
-8.37911740e-02 -2.59583741e-01 6.22578919e-01 3.75448056e-02
-8.35698545e-01 2.96363160e-02 -7.75735617e-01 5.79540610e-01
7.50645161e-01 -2.85900950e-01 -4.80755210e-01 8.67264122e-02
-3.15603733e-01 2.06946746e-01 5.40789127e-01 9.04987216e-01
2.51420707e-01 -1.63805950e+00 -5.36687315e-01 1.55087665e-01
6.75615072e-01 -8.50588322e-01 -1.41737252e-01 6.03264570e-01
-9.76820439e-02 6.98834360e-02 -5.90939879e-01 -2.40340218e-01
-1.35412908e+00 3.92422020e-01 -1.98837802e-01 1.90396860e-01
1.40745476e-01 8.77069592e-01 -1.99978799e-03 -8.84955645e-01
-2.04652235e-01 -3.62681508e-01 -6.07116282e-01 1.10660982e+00
2.30102330e-01 6.36028707e-01 -3.03841621e-01 -7.31007040e-01
-8.74509394e-01 5.23143649e-01 4.44380082e-02 -2.06485733e-01
1.48728669e+00 1.96395721e-02 -4.82848942e-01 7.49167264e-01
1.27898109e+00 4.72873092e-01 -2.49752775e-02 3.99278879e-01
1.75421104e-01 -8.79814506e-01 -2.16522172e-01 -3.01227897e-01
-6.64815485e-01 7.09500194e-01 9.41349089e-01 4.39441442e-01
4.27758127e-01 6.59860820e-02 5.39972365e-01 -1.95963353e-01
1.95043698e-01 -1.62691259e+00 3.03292135e-03 5.59814036e-01
5.31434894e-01 -1.21129560e+00 1.92849100e-01 -2.15037331e-01
-3.70566547e-01 7.60906458e-01 6.82893276e-01 -4.63961661e-01
9.69885647e-01 -1.76479772e-01 1.87880322e-01 -4.61802274e-01
-4.82362092e-01 -2.40198955e-01 5.43514043e-02 9.08866286e-01
1.05083406e+00 1.50011644e-01 -1.49888325e+00 9.77968097e-01
-5.19464135e-01 -2.85436451e-01 6.12461746e-01 6.77470922e-01
-3.15048575e-01 -1.41248250e+00 -3.73606086e-01 7.44866848e-01
-8.06267262e-01 -5.48136495e-02 -5.97798049e-01 1.05989742e+00
6.91167116e-01 8.55861843e-01 7.03511655e-01 -3.98161858e-01
2.82299191e-01 3.63911986e-01 6.81441128e-01 -8.08158398e-01
-5.45896947e-01 -5.75733125e-01 6.49770439e-01 -2.65565008e-01
-5.56734145e-01 -8.40231836e-01 -8.37296605e-01 -5.93221903e-01
1.64915115e-01 9.50793177e-02 4.83436048e-01 6.54400170e-01
1.13837443e-01 2.48148248e-01 2.43137419e-01 -5.60578108e-01
-1.74606845e-01 -1.07609391e+00 -9.34618831e-01 9.13676083e-01
1.09420620e-01 -1.23774791e+00 -5.56542695e-01 -4.69590753e-01]
|
[9.365493774414062, 10.202165603637695]
|
f267b2f9-4088-48c9-a5cc-f378f587c037
|
casenet-deep-category-aware-semantic-edge
|
1705.09759
| null |
http://arxiv.org/abs/1705.09759v1
|
http://arxiv.org/pdf/1705.09759v1.pdf
|
CASENet: Deep Category-Aware Semantic Edge Detection
|
Boundary and edge cues are highly beneficial in improving a wide variety of
vision tasks such as semantic segmentation, object recognition, stereo, and
object proposal generation. Recently, the problem of edge detection has been
revisited and significant progress has been made with deep learning. While
classical edge detection is a challenging binary problem in itself, the
category-aware semantic edge detection by nature is an even more challenging
multi-label problem. We model the problem such that each edge pixel can be
associated with more than one class as they appear in contours or junctions
belonging to two or more semantic classes. To this end, we propose a novel
end-to-end deep semantic edge learning architecture based on ResNet and a new
skip-layer architecture where category-wise edge activations at the top
convolution layer share and are fused with the same set of bottom layer
features. We then propose a multi-label loss function to supervise the fused
activations. We show that our proposed architecture benefits this problem with
better performance, and we outperform the current state-of-the-art semantic
edge detection methods by a large margin on standard data sets such as SBD and
Cityscapes.
|
['Ming-Yu Liu', 'Zhiding Yu', 'Srikumar Ramalingam', 'Chen Feng']
|
2017-05-27
|
casenet-deep-category-aware-semantic-edge-1
|
http://openaccess.thecvf.com/content_cvpr_2017/html/Yu_CASENet_Deep_Category-Aware_CVPR_2017_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2017/papers/Yu_CASENet_Deep_Category-Aware_CVPR_2017_paper.pdf
|
cvpr-2017-7
|
['object-proposal-generation']
|
['computer-vision']
|
[ 3.01660210e-01 7.71335931e-03 5.22171780e-02 -6.20887578e-01
-4.60897774e-01 -3.30480427e-01 4.70839620e-01 2.66768664e-01
-7.46554792e-01 4.54926819e-01 -9.14966390e-02 -1.74758181e-01
2.37239882e-01 -8.92269909e-01 -7.33509839e-01 -5.07022560e-01
-6.90179924e-03 2.87812173e-01 6.69101894e-01 5.51000684e-02
3.63427028e-02 2.77174205e-01 -1.56379545e+00 2.51593113e-01
8.35402608e-01 1.57539511e+00 1.22334749e-01 4.57694620e-01
-1.79702356e-01 6.46587491e-01 -9.06857550e-02 -3.94154817e-01
3.05446506e-01 -3.54867697e-01 -9.02596235e-01 2.15051532e-01
6.99228287e-01 -2.91955043e-02 -9.64400768e-02 1.41455400e+00
4.65946019e-01 3.30520213e-01 6.61878109e-01 -1.20899653e+00
-7.34196723e-01 2.14794755e-01 -8.75977814e-01 1.72295943e-01
-2.08838224e-01 -3.07645589e-01 1.12674642e+00 -9.61352646e-01
5.41221082e-01 1.06330526e+00 7.92595088e-01 4.44580019e-01
-1.01035023e+00 -3.84253025e-01 4.60652649e-01 4.52474654e-01
-1.35126293e+00 -1.60653107e-02 7.92928100e-01 -4.12972301e-01
7.01183856e-01 -1.40295938e-01 5.51839530e-01 7.00874925e-01
-1.37857929e-01 1.03633261e+00 1.03863156e+00 -3.50240171e-01
2.47219145e-01 -3.05869821e-02 2.73767263e-01 9.22248363e-01
2.68322706e-01 -1.81612566e-01 -2.00455397e-01 3.30712616e-01
6.97545648e-01 1.34015247e-01 -2.22478479e-01 -4.93311733e-01
-7.76114404e-01 8.77359211e-01 9.76118803e-01 2.12592319e-01
-4.52616394e-01 3.47418159e-01 2.72962600e-01 -3.00704408e-02
7.13130176e-01 -4.62803431e-03 -5.55649161e-01 3.39338243e-01
-9.12379861e-01 2.61523500e-02 6.03540361e-01 6.21704757e-01
9.09093380e-01 -1.64081439e-01 -1.96498200e-01 1.02480650e+00
3.26059312e-01 9.70196277e-02 3.88651311e-01 -7.65479028e-01
2.35310197e-01 8.62264335e-01 1.31604761e-01 -8.31329882e-01
-7.32850492e-01 -6.56626403e-01 -7.98793733e-01 6.65322721e-01
4.52606916e-01 -1.31508350e-01 -1.46870077e+00 1.80761051e+00
5.56448102e-01 5.43428004e-01 -1.60054713e-01 1.12681675e+00
9.53337193e-01 4.32918847e-01 2.62172461e-01 3.54230165e-01
1.69763529e+00 -1.35596442e+00 -4.76040542e-01 -7.25587666e-01
5.05237401e-01 -6.05596721e-01 5.91741204e-01 8.14434811e-02
-9.01326656e-01 -4.75198895e-01 -9.33367610e-01 -3.75053465e-01
-5.97280085e-01 7.33098537e-02 9.09932017e-01 4.77953076e-01
-1.11565709e+00 4.56001937e-01 -8.96331728e-01 -2.30200335e-01
8.63316000e-01 2.10980460e-01 -3.62405062e-01 -2.15343475e-01
-1.15896940e+00 7.69990861e-01 5.69098592e-01 1.22539207e-01
-4.47770149e-01 -4.44902569e-01 -1.16513431e+00 2.61659205e-01
4.67734188e-01 -1.03480196e+00 1.03250492e+00 -1.08381569e+00
-1.08909035e+00 1.35706997e+00 -7.76109099e-02 -5.07865608e-01
7.73411632e-01 -2.51725495e-01 -3.39559436e-01 7.62546882e-02
3.00850570e-01 1.02782750e+00 6.32577300e-01 -1.16592133e+00
-1.18999624e+00 -4.06485319e-01 9.68712792e-02 2.92526066e-01
-2.15113759e-01 -5.98945022e-02 -7.66985536e-01 -7.85126150e-01
2.52097219e-01 -8.84180486e-01 -4.24010634e-01 3.16302747e-01
-4.34759468e-01 -4.25560057e-01 7.41305351e-01 -5.78997076e-01
9.94350255e-01 -2.03657079e+00 -1.20271780e-02 -1.91523712e-02
2.11575851e-01 3.09326082e-01 -6.59035370e-02 -2.94210494e-01
-7.95915723e-02 6.70698136e-02 -7.41994023e-01 -5.21353424e-01
-5.32326251e-02 -1.00652650e-01 1.88547224e-01 3.82011652e-01
3.30451429e-01 9.70984459e-01 -9.78218615e-01 -3.59110832e-01
2.27230534e-01 6.18615866e-01 -5.61420381e-01 -3.04479748e-02
-1.39099672e-01 3.30845505e-01 -3.88133645e-01 6.34110272e-01
6.79173350e-01 -5.38012028e-01 -3.60402524e-01 -1.75100237e-01
1.93131547e-02 -3.97119019e-03 -1.39002168e+00 1.89594781e+00
-5.15917182e-01 7.22682774e-01 1.20376378e-01 -1.17465746e+00
7.86713839e-01 1.48076102e-01 3.95797163e-01 -6.59599483e-01
2.58869559e-01 2.99784034e-01 -3.63410860e-02 -3.41667384e-01
4.97919768e-01 -1.69796228e-01 -8.38025287e-02 1.40286252e-01
1.01964949e-02 8.58930126e-02 2.26415113e-01 -6.85857087e-02
8.92531753e-01 -3.60052288e-02 9.76032764e-02 -2.32298806e-01
5.15040934e-01 -1.69718936e-01 8.14328909e-01 7.26815403e-01
-3.33554745e-01 8.63999069e-01 2.49999449e-01 -2.66349167e-01
-7.93435395e-01 -8.11085641e-01 -3.13974410e-01 1.13765395e+00
5.39002419e-01 1.19652756e-01 -8.73652995e-01 -8.03592801e-01
1.77374005e-01 4.92152900e-01 -7.40230083e-01 -2.05186397e-01
-4.65664476e-01 -8.37611735e-01 1.31283164e-01 8.44736278e-01
1.07897472e+00 -1.09480083e+00 -5.60042977e-01 3.14954132e-01
-4.29025218e-02 -1.40107381e+00 -7.10413933e-01 3.04609120e-01
-6.12219095e-01 -1.00282252e+00 -1.00323856e+00 -1.34933007e+00
6.68095112e-01 2.06836998e-01 1.12459242e+00 -1.09517634e-01
-5.72940052e-01 1.53414980e-02 -3.00641090e-01 -3.03524613e-01
3.39290462e-02 1.24709919e-01 -4.64126170e-01 4.54680681e-01
4.19732809e-01 -2.25507274e-01 -9.78060901e-01 2.32224047e-01
-9.88422930e-01 1.09357566e-01 4.97537702e-01 9.26621556e-01
6.74286366e-01 3.15467343e-02 5.81180871e-01 -1.11256206e+00
2.39840686e-01 -3.13991159e-01 -6.23744249e-01 2.87048101e-01
-3.36519212e-01 1.33377507e-01 4.15595740e-01 1.69980284e-02
-1.10030305e+00 2.40087301e-01 -4.32739615e-01 -3.39061111e-01
-2.98081905e-01 4.62125361e-01 -1.91178411e-01 -2.31712326e-01
3.24815512e-01 -2.06808835e-01 -4.05240864e-01 -4.58874106e-01
5.49553156e-01 6.36187613e-01 6.98466539e-01 -1.05424635e-01
3.28589708e-01 7.26674199e-01 1.12334117e-01 -6.30691648e-01
-1.15563881e+00 -7.93444812e-01 -4.50752288e-01 -1.75581247e-01
1.29881382e+00 -1.05533624e+00 -3.75485897e-01 9.72775757e-01
-1.11951983e+00 -4.43025470e-01 -1.39031500e-01 3.72676224e-01
-3.87315631e-01 3.18706155e-01 -5.24435341e-01 -5.39667308e-01
-2.42632523e-01 -1.15063286e+00 1.34257674e+00 6.60457730e-01
2.28838652e-01 -1.22473645e+00 -2.95034796e-01 3.15715492e-01
2.62184799e-01 4.15307432e-01 7.03030944e-01 -7.06768274e-01
-5.00447452e-01 -3.38442534e-01 -7.10610271e-01 4.73186582e-01
-3.91863957e-02 -4.30047035e-01 -9.48580563e-01 -1.96254537e-01
-2.05966786e-01 -2.97689170e-01 1.60279679e+00 5.87664902e-01
1.23360407e+00 2.62989908e-01 -4.76829529e-01 8.17547977e-01
1.62179172e+00 -2.26958729e-02 3.44583660e-01 3.32829297e-01
9.67552900e-01 5.10345817e-01 2.85202295e-01 1.67049721e-01
4.64102775e-01 6.59911871e-01 6.61781132e-01 -5.99685729e-01
-4.58973110e-01 3.90830822e-02 -1.71379760e-01 1.32114604e-01
9.85102281e-02 -3.35751742e-01 -9.21923816e-01 8.30150068e-01
-1.96447217e+00 -6.68680608e-01 -2.98651397e-01 2.09684491e+00
5.45325398e-01 3.86484951e-01 2.34019179e-02 -5.39746806e-02
1.12066972e+00 5.00280634e-02 -7.81410813e-01 -1.26872316e-01
-2.19303325e-01 3.06019008e-01 7.35753596e-01 4.94859159e-01
-1.73636818e+00 1.24148941e+00 5.28566599e+00 7.09955096e-01
-1.18409693e+00 1.65580764e-01 1.00447917e+00 4.11655605e-01
5.90503402e-03 -9.80925709e-02 -8.08726013e-01 3.74751389e-01
1.97872266e-01 2.38950446e-01 3.23958807e-02 9.96757686e-01
-1.47650614e-01 -2.16750368e-01 -9.30774450e-01 1.00944149e+00
1.05064526e-01 -1.17032003e+00 -1.40507355e-01 -1.71124518e-01
1.13911831e+00 2.28177547e-01 6.05627112e-02 1.30764231e-01
4.59846824e-01 -8.87573659e-01 7.69720972e-01 2.16721341e-01
6.22097909e-01 -5.01059353e-01 6.44982517e-01 1.39614880e-01
-1.42240787e+00 -1.60025552e-01 -2.99790233e-01 6.01339824e-02
4.80282456e-01 6.88826919e-01 -3.19993854e-01 4.68494296e-01
6.88066006e-01 8.45952332e-01 -3.92061323e-01 1.70705271e+00
-4.60641652e-01 4.08325672e-01 -4.10834938e-01 2.21941248e-01
6.80171311e-01 -2.02324212e-01 3.21228147e-01 1.26169991e+00
2.28308350e-01 7.52531886e-02 3.87097776e-01 8.52540910e-01
-4.83068675e-01 -1.06650896e-01 -2.83864617e-01 3.62010539e-01
2.42870465e-01 1.57223177e+00 -1.27724278e+00 -4.08118814e-01
-7.79536307e-01 1.34699535e+00 4.59568352e-01 4.60266411e-01
-7.46867180e-01 -6.93836391e-01 7.72942364e-01 -2.30086222e-01
5.88767052e-01 4.73872535e-02 -4.22764212e-01 -1.10494947e+00
-6.52339235e-02 -5.05102836e-02 6.01081371e-01 -5.79247653e-01
-1.34899366e+00 5.49043000e-01 -4.90610152e-01 -8.24360609e-01
1.41097635e-01 -1.02003193e+00 -7.35915482e-01 7.67562211e-01
-2.03385663e+00 -1.09811687e+00 -5.26972651e-01 3.85591418e-01
5.58816373e-01 2.89651990e-01 4.44269836e-01 6.51813149e-01
-5.88218749e-01 4.96779144e-01 1.48887113e-01 5.74363530e-01
4.07189012e-01 -1.46831584e+00 6.10975325e-01 1.01113272e+00
2.43854418e-01 -9.63335857e-02 3.86993736e-01 -4.77364451e-01
-4.68527913e-01 -1.46391964e+00 9.04721081e-01 1.96114816e-02
4.80963886e-01 -3.82736742e-01 -8.66163731e-01 5.50254047e-01
-2.33874992e-02 6.42958999e-01 2.12220803e-01 -1.28016695e-01
-1.62586942e-01 9.48540866e-02 -9.54436719e-01 4.29652214e-01
1.33277857e+00 -4.58709240e-01 -4.61537272e-01 2.69598961e-01
5.04496574e-01 -3.49889278e-01 -4.85413969e-01 5.72712302e-01
2.35493347e-01 -9.36620831e-01 1.07450116e+00 -6.74480379e-01
2.73565382e-01 -3.52161288e-01 7.56781325e-02 -1.38067842e+00
-3.56439799e-01 -1.33440971e-01 2.16110215e-01 1.13588154e+00
4.02653009e-01 -5.65681219e-01 1.05117476e+00 5.40508270e-01
-6.31042242e-01 -8.05870414e-01 -1.01193833e+00 -7.66298771e-01
4.61911075e-02 -3.06667089e-01 2.01511353e-01 8.62064838e-01
-3.73073280e-01 4.70533013e-01 -1.42178699e-01 1.52270079e-01
6.09932065e-01 4.11029905e-01 2.20230341e-01 -1.46811903e+00
-1.02332607e-01 -9.48882520e-01 -7.83038795e-01 -1.28093338e+00
1.74009845e-01 -1.22615647e+00 2.66647428e-01 -2.10437632e+00
1.51129588e-01 -5.49057901e-01 -5.00684261e-01 5.46405435e-01
-7.82329142e-01 5.37638843e-01 2.63311602e-02 -2.16255635e-01
-8.39255273e-01 5.37058711e-01 1.11592472e+00 -2.60726184e-01
-5.29465601e-02 1.45086143e-02 -5.63996494e-01 9.42092240e-01
5.62919080e-01 -3.88558239e-01 -1.14501350e-01 -5.67489147e-01
2.31306463e-01 -2.99505740e-01 5.95124066e-01 -1.12166822e+00
3.90004158e-01 2.81975329e-01 2.25863993e-01 -1.75050721e-01
1.86634749e-01 -9.40303504e-01 -2.83490509e-01 3.70640278e-01
-2.25512147e-01 -3.71960014e-01 1.16194703e-01 6.97892547e-01
-3.90225142e-01 -2.62262166e-01 1.01888371e+00 -1.74512550e-01
-1.35040176e+00 6.29171073e-01 7.38529442e-03 4.76642370e-01
1.18853390e+00 -3.24239105e-01 -8.61731470e-02 1.68433711e-02
-8.64454925e-01 4.90608931e-01 3.69847506e-01 4.78988916e-01
3.63463521e-01 -1.17787313e+00 -7.14759231e-01 1.05264120e-01
2.20316947e-01 3.31388205e-01 2.87125796e-01 7.12517262e-01
-4.69671100e-01 1.67519376e-01 -1.89491004e-01 -6.96456730e-01
-9.33870077e-01 3.76716882e-01 5.36707699e-01 -1.58201247e-01
-6.19673133e-01 1.38670099e+00 6.91327572e-01 -1.72392875e-01
3.49532366e-01 -3.06745380e-01 -3.84867847e-01 1.32288530e-01
2.10274175e-01 2.79933393e-01 2.33923510e-01 -7.63982356e-01
-3.98088723e-01 8.44560981e-01 -9.24353227e-02 2.27178290e-01
1.19106376e+00 -1.35880336e-01 1.63365483e-01 2.28906348e-01
1.35704744e+00 -5.72090089e-01 -1.65817416e+00 -4.14809048e-01
2.32439592e-01 -2.04877645e-01 4.00732189e-01 -8.41813147e-01
-1.46817708e+00 9.09471869e-01 8.15769613e-01 1.38607040e-01
1.25370884e+00 1.85855433e-01 1.08089781e+00 3.40339132e-02
1.68506190e-01 -1.34651291e+00 4.55641672e-02 5.31837940e-01
2.63933301e-01 -1.69654679e+00 -4.94790345e-01 -6.42604113e-01
-5.76768577e-01 7.92864978e-01 5.62895179e-01 -2.31448889e-01
8.01802933e-01 4.91920263e-02 3.60627212e-02 -2.18942821e-01
-1.46505773e-01 -8.89278054e-01 5.41689694e-01 3.55910271e-01
4.21585351e-01 6.24218099e-02 -4.27212059e-01 5.03857613e-01
3.83684605e-01 2.10453346e-02 1.07312396e-01 8.14879298e-01
-7.62243450e-01 -9.53787565e-01 -1.04902551e-01 5.89223325e-01
-5.46741784e-01 -2.09798768e-01 -1.51269317e-01 5.64417958e-01
3.21591765e-01 7.76914537e-01 4.33154136e-01 1.59727428e-02
2.62046754e-01 1.74587682e-01 2.29034126e-01 -6.22373164e-01
-3.20251733e-01 -6.89655766e-02 -4.90386710e-02 -3.19589555e-01
-4.51740801e-01 -5.82677901e-01 -1.55993760e+00 2.23584652e-01
-4.82639134e-01 -6.71825111e-02 6.47507906e-01 9.08506215e-01
3.36740911e-01 6.91040754e-01 4.69337821e-01 -8.26477528e-01
-1.98167741e-01 -7.08527148e-01 -6.56262457e-01 8.20550680e-01
2.81872362e-01 -9.00140524e-01 -1.84934795e-01 1.31465673e-01]
|
[9.555041313171387, 0.4696494936943054]
|
dfe08db9-d95e-45db-b278-4ad1f55f0be4
|
attention-lstm-for-multivariate-traffic-state
|
2301.02731
| null |
https://arxiv.org/abs/2301.02731v1
|
https://arxiv.org/pdf/2301.02731v1.pdf
|
Attention-LSTM for Multivariate Traffic State Prediction on Rural Roads
|
Accurate traffic volume and speed prediction have a wide range of applications in transportation. It can result in useful and timely information for both travellers and transportation decision-makers. In this study, an Attention based Long Sort-Term Memory model (A-LSTM) is proposed to simultaneously predict traffic volume and speed in a critical rural road segmentation which connects Tehran to Chalus, the most tourist destination city in Iran. Moreover, this study compares the results of the A-LSTM model with the Long Short-Term Memory (LSTM) model. Both models show acceptable performance in predicting speed and flow. However, the A-LSTM model outperforms the LSTM in 5 and 15-minute intervals. In contrast, there is no meaningful difference between the two models for the 30-minute time interval. By comparing the performance of the models based on different time horizons, the 15-minute horizon model outperforms the others by reaching the lowest Mean Square Error (MSE) loss of 0.0032, followed by the 30 and 5-minutes horizons with 0.004 and 0.0051, respectively. In addition, this study compares the results of the models based on two transformations of temporal categorical input variables, one-hot or cyclic, for the 15-minute time interval. The results demonstrate that both LSTM and A-LSTM with cyclic feature encoding outperform those with one-hot feature encoding.
|
['Seyedehsan Seyedabrishami', 'Amir Hossein Karbasi', 'Bilal Farooq', 'Elahe Sherafat']
|
2023-01-06
| null | null | null | null |
['road-segementation']
|
['computer-vision']
|
[-1.90976530e-01 -3.96788329e-01 -4.22010601e-01 -2.99980104e-01
-4.40654069e-01 -6.69712499e-02 5.27797639e-01 -1.48679558e-02
-5.69008112e-01 9.93659556e-01 5.50677255e-02 -8.26331019e-01
-5.92356265e-01 -1.10764170e+00 -4.16113853e-01 -7.20320046e-01
-3.28715563e-01 1.49844512e-01 2.68288046e-01 -3.80528450e-01
2.80085772e-01 4.92311120e-01 -1.64980936e+00 1.96409613e-01
1.20654297e+00 1.26225972e+00 2.98067808e-01 3.93985122e-01
-3.31520170e-01 6.60730183e-01 -3.67361277e-01 -2.27597326e-01
1.18912227e-01 -1.11088142e-01 -7.29417503e-01 -4.57564980e-01
1.63670164e-04 -2.00731382e-01 -6.49666965e-01 6.25505805e-01
3.81777465e-01 8.25999081e-01 4.14960563e-01 -1.37834609e+00
-4.39266324e-01 4.04106557e-01 -4.91021127e-01 6.08285844e-01
2.70497636e-04 1.65085047e-01 3.95032376e-01 -6.06942296e-01
6.71779783e-03 1.34758365e+00 6.26388252e-01 4.89592217e-02
-6.69828653e-01 -6.06653512e-01 3.09809178e-01 6.78921223e-01
-1.40151823e+00 -1.61082193e-01 4.38059211e-01 -5.72381020e-01
1.22014821e+00 2.34484851e-01 5.66378593e-01 5.34524202e-01
8.87709975e-01 5.21563172e-01 8.60587776e-01 -8.32828730e-02
-8.12473819e-02 5.81303388e-02 2.18524858e-01 4.19687361e-01
-3.75156514e-02 4.79370117e-01 7.92160407e-02 3.93953413e-01
5.35129249e-01 3.32302809e-01 2.57160455e-01 8.10438037e-01
-9.69217956e-01 8.37249637e-01 5.65298617e-01 5.28013945e-01
-8.46589386e-01 4.63883691e-02 5.88059008e-01 3.75615269e-01
7.51497388e-01 -1.57765687e-01 -5.11422396e-01 -4.98658299e-01
-9.60479081e-01 -1.41090276e-02 3.20530355e-01 6.24534309e-01
4.83941317e-01 6.74878001e-01 -6.10319614e-01 7.94048905e-01
1.46910146e-01 5.66210806e-01 5.20681262e-01 -6.83577299e-01
7.39536762e-01 2.13292837e-01 5.98249957e-02 -1.37332416e+00
-5.84111154e-01 -5.08278430e-01 -8.54930043e-01 -2.03791946e-01
4.74337399e-01 -3.63919377e-01 -1.02883863e+00 1.47676909e+00
-1.36708170e-01 5.01936376e-01 7.12524774e-03 6.91806197e-01
7.58557618e-01 1.24905443e+00 3.88006300e-01 -3.09042066e-01
9.61199045e-01 -1.30260158e+00 -1.01596880e+00 -2.17653885e-01
8.10259283e-01 -5.95169365e-01 8.56474042e-01 -1.72633156e-02
-9.70433652e-01 -7.61307418e-01 -4.38738018e-01 1.80815279e-01
-9.18464541e-01 -4.43544686e-02 5.31226337e-01 4.51195627e-01
-8.65182579e-01 6.85722709e-01 -5.82334399e-01 -5.72031796e-01
1.03487529e-01 1.39572337e-01 -1.23380478e-02 1.86383314e-02
-1.88512480e+00 1.26967096e+00 2.74669945e-01 5.71347773e-01
-5.49369574e-01 -6.63078845e-01 -7.95423150e-01 1.31165102e-01
5.16247936e-02 -2.47527003e-01 1.07628191e+00 -8.33042800e-01
-1.32909620e+00 1.34228230e-01 -5.65377116e-01 -6.76479518e-01
4.50657576e-01 1.67556815e-02 -1.06928611e+00 -2.28552774e-01
1.27508223e-01 6.09437168e-01 2.55109638e-01 -6.68932199e-01
-1.18316603e+00 -1.93004400e-01 -6.94994703e-02 3.19116376e-02
-1.30966231e-02 -6.60527349e-02 -9.26325321e-02 -4.90681380e-01
-1.51954383e-01 -7.66446650e-01 -3.26789945e-01 -3.79764050e-01
-8.20043609e-02 -4.62006092e-01 9.99421239e-01 -9.70757902e-01
1.72132039e+00 -1.93295634e+00 -7.62848854e-01 2.30475202e-01
-4.07501042e-01 5.05469680e-01 -1.73997283e-01 4.32240814e-01
-7.14435726e-02 3.16647351e-01 6.02556020e-02 8.86055455e-02
-1.80378363e-01 5.11238217e-01 -2.47313514e-01 2.52503932e-01
7.82903582e-02 1.11512244e+00 -9.12524998e-01 -3.97307217e-01
6.60502255e-01 4.39079642e-01 2.44015977e-01 -1.46635309e-01
3.60227436e-01 2.91494250e-01 -2.63094902e-01 4.04884785e-01
5.74548304e-01 1.25907034e-01 -5.81258655e-01 2.14286909e-01
-7.29818404e-01 2.20431551e-01 -7.29534268e-01 1.02593887e+00
-6.97753727e-01 1.18622720e+00 -5.69528580e-01 -1.08071268e+00
1.14995849e+00 6.43944144e-01 5.16067684e-01 -1.47362137e+00
1.06098138e-01 2.65543222e-01 7.66874254e-02 -6.86848581e-01
6.70196056e-01 -2.22914338e-01 8.34941044e-02 9.60718989e-02
-5.57414532e-01 5.14873326e-01 3.57331991e-01 -2.41122261e-01
4.14484084e-01 -3.33435565e-01 -3.04428250e-01 -1.54707909e-01
5.74569404e-01 -2.64066011e-01 6.38699412e-01 4.25616384e-01
-3.63990784e-01 1.04364924e-01 3.91563892e-01 -7.21710920e-01
-6.63308382e-01 -7.51474440e-01 -1.35931283e-01 9.32204008e-01
-3.43523473e-02 3.41179281e-01 -3.00859600e-01 -2.32871532e-01
1.61501005e-01 1.23191714e+00 -7.79882312e-01 -3.38679582e-01
-6.13289952e-01 -4.01223302e-01 5.11334062e-01 6.47348762e-01
9.04867709e-01 -1.10254526e+00 -5.62070310e-01 4.43123728e-01
-4.05021071e-01 -1.04439533e+00 -3.53529334e-01 -3.26599926e-01
-9.42260742e-01 -4.78362501e-01 -8.80373001e-01 -7.03610778e-01
2.85442472e-01 4.77323830e-01 6.64612472e-01 -1.26738325e-01
2.31594622e-01 -1.06335685e-01 -6.55797124e-02 -5.51578641e-01
1.65262818e-01 1.50318682e-01 -1.52910963e-01 1.74146056e-01
5.02223909e-01 -2.87778467e-01 -4.98458177e-01 4.95172650e-01
-3.78784597e-01 -5.82208820e-02 3.98083925e-01 5.89297950e-01
3.41741592e-01 4.18430448e-01 1.14394033e+00 -3.55279744e-01
6.01819813e-01 -1.01757360e+00 -3.67577136e-01 2.49513969e-01
-1.04211807e+00 -3.31058055e-01 6.83232188e-01 -2.83409148e-01
-1.11859608e+00 -6.19571209e-01 -2.38344491e-01 -3.16889286e-01
-2.42284443e-02 9.84277725e-01 1.85759023e-01 2.50348449e-01
-2.16325670e-02 2.25887641e-01 -1.79190278e-01 -3.83021623e-01
6.31336123e-02 6.89831018e-01 3.05452853e-01 -1.22149102e-01
2.37599596e-01 2.51828849e-01 2.55485810e-02 -1.13491118e+00
-3.64605814e-01 -3.76672894e-01 -5.36551893e-01 -6.78957641e-01
8.34446490e-01 -6.64942443e-01 -5.83293557e-01 6.63772047e-01
-1.01794434e+00 -3.39120686e-01 -2.80424505e-01 8.81563187e-01
-4.81222242e-01 -5.36240861e-02 -5.58495462e-01 -1.03513360e+00
-2.34538600e-01 -9.87563252e-01 2.67944872e-01 4.33435947e-01
-1.40308633e-01 -1.45899177e+00 -4.48365927e-01 1.06294118e-02
8.29875648e-01 3.96448284e-01 9.86747742e-01 -4.72628206e-01
-1.76609769e-01 -3.48110199e-01 -3.55725557e-01 1.50968209e-01
1.91155881e-01 2.34462261e-01 -6.80647969e-01 -1.48150623e-01
-4.29578692e-01 4.18259412e-01 9.70315337e-01 1.11703110e+00
1.12238884e+00 -4.91648704e-01 -3.50790650e-01 3.40820432e-01
1.25433517e+00 1.13776004e+00 1.11393583e+00 6.07734084e-01
4.47262734e-01 9.40574884e-01 9.14710581e-01 1.93590283e-01
5.76179206e-01 3.47919047e-01 2.07918912e-01 -3.94301623e-01
1.80520937e-01 -3.57195884e-01 2.18910426e-01 7.02125311e-01
-1.07615024e-01 -4.21609908e-01 -1.11822104e+00 1.01486123e+00
-1.88813293e+00 -1.35863674e+00 -6.19744182e-01 2.31311727e+00
1.13822021e-01 3.94531995e-01 2.48570889e-01 4.29844975e-01
8.15920353e-01 2.37003133e-01 -4.08520520e-01 -1.18860590e+00
-1.94033943e-02 -7.97716603e-02 8.14703882e-01 6.18203938e-01
-7.85487354e-01 9.63140368e-01 6.11656189e+00 1.00337923e+00
-1.46356797e+00 7.97627214e-03 8.46478879e-01 1.81192830e-02
-1.81252599e-01 -1.02869295e-01 -5.02820194e-01 9.94475245e-01
1.92279482e+00 -5.34654319e-01 1.28631383e-01 3.62168670e-01
1.17620337e+00 -4.00887072e-01 -4.91874486e-01 8.00420165e-01
-3.75501603e-01 -1.05338538e+00 2.12981775e-01 9.53149498e-02
7.38263071e-01 -2.79716440e-02 2.83508360e-01 6.45028472e-01
-1.54755771e-01 -1.21268237e+00 4.89248306e-01 1.05197811e+00
7.16363192e-01 -1.34689295e+00 1.05614567e+00 3.38243395e-01
-1.49123502e+00 -4.63815510e-01 -2.39891723e-01 -5.43621182e-01
7.08276689e-01 5.92751145e-01 -3.09347630e-01 5.08757889e-01
7.18914628e-01 8.95607769e-01 -2.52025843e-01 1.31202400e+00
2.55905211e-01 8.89281332e-01 -5.96336462e-02 9.43829566e-02
8.60176504e-01 -3.85491371e-01 3.67647767e-01 1.22314107e+00
5.41886568e-01 1.26445115e-01 -1.14201382e-02 3.70492488e-01
4.33306366e-01 -9.57523137e-02 -8.33015859e-01 1.63549613e-02
6.27098322e-01 6.92181349e-01 -4.46087092e-01 -5.38999856e-01
-5.84557652e-01 3.34849656e-01 -3.40414464e-01 8.41292799e-01
-1.23137331e+00 -9.70265985e-01 5.93756318e-01 2.92274058e-01
1.69778958e-01 -1.43797606e-01 -5.05659342e-01 -3.27806950e-01
-1.39332652e-01 -5.29127494e-02 4.00953233e-01 -7.30928838e-01
-9.28123236e-01 5.62047541e-01 2.05123201e-01 -1.24755490e+00
-2.62895882e-01 -1.31470323e-01 -9.83688772e-01 1.20235384e+00
-1.85675323e+00 -1.00234926e+00 -1.72065035e-01 5.39266467e-01
8.75142276e-01 2.51517706e-02 3.60230267e-01 6.76993728e-01
-1.00326300e+00 5.61264634e-01 2.71111220e-01 2.28028875e-02
1.21464178e-01 -6.37052298e-01 5.04935980e-01 7.02217042e-01
-7.90261686e-01 3.56999665e-01 6.04117155e-01 -5.01167715e-01
-6.07468486e-01 -1.56867528e+00 1.62153602e+00 2.17627287e-01
3.32958490e-01 5.48228562e-01 -9.75903988e-01 8.15393806e-01
9.48431790e-02 -3.69697064e-01 4.04713959e-01 -1.17415011e-01
3.82890403e-01 -4.70974177e-01 -1.23780787e+00 4.59047765e-01
5.65499425e-01 -3.61480653e-01 -3.71607006e-01 2.27998905e-02
7.53403008e-01 3.09418552e-02 -1.02310824e+00 4.72679347e-01
8.01303566e-01 -5.94869316e-01 7.20080614e-01 -5.23896575e-01
1.09130688e-01 6.29562046e-03 -5.36889099e-02 -1.23399949e+00
-5.55658460e-01 -3.02275240e-01 5.37566282e-02 1.09850192e+00
6.73221707e-01 -1.29657257e+00 3.40511262e-01 7.78881907e-01
-4.53680545e-01 -8.97480488e-01 -1.15824425e+00 -1.11576498e+00
2.89836198e-01 -7.60441124e-01 8.98822963e-01 7.39884377e-01
-2.67494798e-01 4.36627492e-02 -4.09475237e-01 -2.04764679e-01
1.82663277e-01 -9.70715433e-02 2.96458662e-01 -1.14462900e+00
7.55798280e-01 -6.63674355e-01 -3.13355774e-01 -1.06070590e+00
1.54556200e-01 -5.82596123e-01 -2.26169318e-01 -2.09747434e+00
-1.60517111e-01 -2.87889391e-01 -6.28221989e-01 5.42703748e-01
3.41626294e-02 -2.65978426e-01 -9.19072889e-03 -1.02376774e-01
-8.18673968e-02 7.13272631e-01 1.20183432e+00 -1.35119379e-01
-4.65285301e-01 4.51822639e-01 -3.69221568e-01 4.60813940e-01
1.15310955e+00 -2.31689990e-01 -4.70067203e-01 -5.20444870e-01
-1.24797054e-01 4.18366015e-01 1.95671216e-01 -9.36693013e-01
3.63279045e-01 -5.24406135e-01 2.23175779e-01 -1.24015856e+00
2.80290186e-01 -6.46785438e-01 8.53881538e-02 7.06440985e-01
-3.13325018e-01 6.55048072e-01 6.48201108e-01 4.52185631e-01
-4.04673994e-01 3.36710006e-01 6.58990264e-01 3.83878089e-02
-1.13278306e+00 4.69516426e-01 -9.68618691e-01 -2.94080794e-01
1.25108850e+00 -9.35721397e-01 -2.19996929e-01 -6.50528729e-01
-7.68976927e-01 8.35454583e-01 -3.20842057e-01 7.64759421e-01
8.20782065e-01 -1.48491657e+00 -7.01592326e-01 3.25239807e-01
-3.71169567e-01 -3.68313432e-01 7.38834977e-01 1.27350295e+00
-2.98002273e-01 1.14329982e+00 -3.36144179e-01 -4.30400878e-01
-8.88547421e-01 4.00678605e-01 3.08160961e-01 -6.04442284e-02
-2.71434307e-01 4.46838170e-01 -1.93109736e-01 -2.37912789e-01
1.02618657e-01 -4.92473215e-01 -6.12242639e-01 3.64969879e-01
4.30124193e-01 1.31512344e+00 9.38316658e-02 -1.03308809e+00
-2.86205500e-01 6.31776810e-01 3.14267546e-01 -1.54682487e-01
1.01666820e+00 -5.90890884e-01 1.40496030e-01 9.84368980e-01
1.38030338e+00 -7.28066981e-01 -1.03886461e+00 -6.50698692e-02
2.05780104e-01 -5.09587586e-01 3.12667102e-01 -7.35598803e-01
-1.28431487e+00 1.17025089e+00 7.40621388e-01 1.52836561e-01
1.19353366e+00 -6.61271214e-01 1.40190804e+00 -7.07438439e-02
2.27826476e-01 -1.21603739e+00 -4.76483077e-01 1.06541979e+00
7.52232373e-01 -1.02234149e+00 -5.52608490e-01 2.26862982e-01
-6.40664339e-01 1.05503130e+00 5.85466504e-01 2.59813935e-01
8.25931609e-01 -4.16480333e-01 -3.74904722e-02 8.83187950e-02
-1.02711594e+00 -4.32265371e-01 4.46342021e-01 3.10577124e-01
2.68332630e-01 3.48244935e-01 -5.80199778e-01 2.62617320e-01
-2.30572924e-01 1.60176262e-01 3.15756798e-01 6.38538182e-01
-5.43704927e-01 -4.31709468e-01 -9.77757573e-02 6.87085867e-01
-3.08169901e-01 -3.08868252e-02 2.23410949e-01 1.04711890e+00
5.92939667e-02 1.53284979e+00 7.14121044e-01 -6.38477206e-01
4.93435651e-01 -2.81123701e-03 -3.07875901e-01 -1.22373849e-01
-3.49026650e-01 -1.89020082e-01 1.88088462e-01 -5.37415802e-01
-1.56726778e-01 -6.87572658e-01 -1.36659336e+00 -1.04822481e+00
-3.83093685e-01 4.05365109e-01 6.09729707e-01 1.04938543e+00
3.50140959e-01 6.14589870e-01 1.00059915e+00 -5.74626625e-01
2.28463523e-02 -1.08886743e+00 -6.03184104e-01 2.01913100e-02
5.46369255e-01 -7.65970349e-01 -3.55248719e-01 -4.41474974e-01]
|
[6.305049419403076, 1.9650936126708984]
|
6c539706-83ca-4ba1-9519-be2fd29fc6a9
|
maximum-mean-discrepancy-kernels-for
|
2301.09624
| null |
https://arxiv.org/abs/2301.09624v1
|
https://arxiv.org/pdf/2301.09624v1.pdf
|
Maximum Mean Discrepancy Kernels for Predictive and Prognostic Modeling of Whole Slide Images
|
How similar are two images? In computational pathology, where Whole Slide Images (WSIs) of digitally scanned tissue samples from patients can be multi-gigapixels in size, determination of degree of similarity between two WSIs is a challenging task with a number of practical applications. In this work, we explore a novel strategy based on kernelized Maximum Mean Discrepancy (MMD) analysis for determination of pairwise similarity between WSIs. The proposed approach works by calculating MMD between two WSIs using kernels over deep features of image patches. This allows representation of an entire dataset of WSIs as a kernel matrix for WSI level clustering, weakly-supervised prediction of TP-53 mutation status in breast cancer patients from their routine WSIs as well as survival analysis with state of the art prediction performance. We believe that this work will open up further avenues for application of WSI-level kernels for predictive and prognostic tasks in computational pathology.
|
['Fayyaz ul Amir Afsar Minhas', 'Muhammad Dawood', 'Piotr Keller']
|
2023-01-23
| null | null | null | null |
['whole-slide-images', 'survival-analysis']
|
['computer-vision', 'miscellaneous']
|
[ 4.58733797e-01 -7.81746674e-03 -1.94279365e-02 -3.72100741e-01
-1.17782140e+00 -6.05620623e-01 2.97733307e-01 8.27956855e-01
-5.37252486e-01 4.58616585e-01 -1.49749801e-01 -3.54748696e-01
-6.43864274e-01 -6.98134661e-01 -3.73263717e-01 -1.30649900e+00
-3.38899314e-01 5.36400020e-01 4.85839754e-01 1.80503622e-01
2.98290730e-01 8.92739534e-01 -1.00274396e+00 5.55517077e-01
6.18932307e-01 9.20614302e-01 1.78081751e-01 1.16278231e+00
-1.68399513e-01 4.30165410e-01 -2.67405927e-01 -3.95309001e-01
9.93620828e-02 -1.51042879e-01 -8.71452749e-01 -1.34890661e-01
5.85926473e-01 3.46861601e-01 -5.42656407e-02 1.10514271e+00
5.38765192e-01 -2.06011042e-01 1.24430954e+00 -8.76377642e-01
-3.22141320e-01 1.25900492e-01 -9.45104122e-01 5.00317454e-01
-8.90847743e-02 -1.35538697e-01 8.09971511e-01 -5.50088346e-01
8.84803355e-01 6.74451828e-01 9.67319131e-01 2.41743937e-01
-1.64818406e+00 -3.18862706e-01 -7.66773283e-01 4.25180972e-01
-1.61945403e+00 -2.84922756e-02 5.02126634e-01 -6.04555607e-01
8.98671985e-01 6.69208586e-01 5.54118514e-01 2.79180646e-01
4.97243464e-01 6.05344236e-01 1.21453428e+00 -4.83883739e-01
3.37790787e-01 1.49183989e-01 4.01569903e-01 8.73688221e-01
1.91244707e-01 -4.56196517e-01 -2.98138559e-01 -7.16898978e-01
5.97268522e-01 7.82589391e-02 -2.58213103e-01 -4.59825307e-01
-1.49674881e+00 6.66668296e-01 2.18541354e-01 6.76158845e-01
5.45295812e-02 -1.92602351e-01 5.46111107e-01 4.31510270e-01
5.46092927e-01 2.37406701e-01 -3.32078904e-01 1.12653233e-01
-1.07916844e+00 -5.58398850e-02 5.75272679e-01 2.28432298e-01
6.28078043e-01 -9.21787798e-01 8.19435939e-02 8.28674018e-01
-1.46777993e-02 2.03585520e-01 7.99821973e-01 -3.87671262e-01
-1.25203356e-01 8.45412076e-01 -3.31782699e-01 -1.07414174e+00
-6.01853788e-01 -8.85174498e-02 -1.17116237e+00 2.08419397e-01
8.54079247e-01 3.00908327e-01 -5.49874246e-01 1.22441804e+00
6.53644204e-01 4.01660532e-01 6.00304268e-02 6.64639294e-01
6.75116599e-01 4.11990643e-01 -5.54653667e-02 -8.58704001e-02
1.55681860e+00 -3.06580901e-01 -2.03227490e-01 4.75117117e-01
1.21080172e+00 -8.23892593e-01 7.10770309e-01 2.47720778e-01
-8.22257519e-01 -9.74046886e-02 -7.47662485e-01 -4.95554172e-02
-3.86959761e-01 6.45463288e-01 4.98037785e-01 3.67173523e-01
-1.19969440e+00 7.29879737e-01 -1.10517871e+00 -8.92710865e-01
6.59581661e-01 5.67564189e-01 -1.06697881e+00 5.76681234e-02
-5.49673200e-01 7.23271430e-01 2.34905005e-01 -7.61176124e-02
-1.81690782e-01 -1.24692738e+00 -5.94021797e-01 -1.78433046e-01
-2.54490644e-01 -1.60212740e-01 6.02096856e-01 -7.88008094e-01
-1.06801856e+00 1.70948243e+00 -1.00313365e-01 -2.73203999e-01
4.04190391e-01 7.11827338e-01 -3.04500729e-01 3.67660522e-01
-1.67504936e-01 1.29447997e-01 5.28577209e-01 -7.45021939e-01
-5.14255464e-01 -8.24197888e-01 -6.78696334e-01 -2.15959884e-02
-4.56903636e-01 -6.86352104e-02 -3.47598493e-02 -3.75267357e-01
3.13623518e-01 -9.67928052e-01 -3.05157363e-01 5.33912718e-01
-5.15728295e-01 -1.62815392e-01 7.80115366e-01 -7.08740234e-01
7.70268679e-01 -2.36140108e+00 6.19049333e-02 5.24760783e-01
3.35618585e-01 4.05024707e-01 -2.39540517e-01 4.07267690e-01
-3.68729919e-01 -8.04847553e-02 -1.09997943e-01 3.43077481e-02
-2.18499392e-01 -1.61683187e-01 1.65800050e-01 1.12570322e+00
2.18419373e-01 8.83977473e-01 -6.98490977e-01 -7.68133938e-01
6.69424050e-03 4.40873086e-01 4.23090532e-02 1.89342141e-01
4.63332474e-01 2.31873006e-01 -3.59369218e-01 5.14352202e-01
1.01502979e+00 -4.82890815e-01 2.49068737e-01 -5.54301143e-01
2.32073173e-01 -3.33345801e-01 -9.61878538e-01 1.52529621e+00
4.99123782e-02 8.83767843e-01 6.95849732e-02 -1.37614918e+00
6.96055055e-01 8.11323375e-02 6.49558604e-01 -4.56829779e-02
-6.32192492e-02 1.03591606e-01 2.00608388e-01 -5.26574016e-01
-7.88226649e-02 -5.10508537e-01 2.18757108e-01 4.32476580e-01
1.74841210e-02 -7.73113668e-02 1.30703002e-01 1.88005313e-01
1.72378600e+00 -7.27312267e-01 7.22354949e-01 -7.93288648e-01
7.88126469e-01 1.57338172e-01 2.03844413e-01 2.94643909e-01
-5.22289336e-01 6.23317540e-01 9.16789234e-01 -5.79454243e-01
-1.14810181e+00 -1.28831077e+00 -7.71159530e-01 4.42266911e-01
-1.78263798e-01 1.08487293e-01 -6.48307800e-01 -7.62276173e-01
3.83033961e-01 -8.15460086e-02 -1.00251091e+00 -5.55729195e-02
-3.59971941e-01 -1.25416648e+00 7.98432052e-01 3.21484268e-01
5.31727523e-02 -4.15923685e-01 -2.72949427e-01 -2.55790859e-01
2.16547370e-01 -8.25884223e-01 -4.68474060e-01 2.67721087e-01
-9.68590260e-01 -1.43687725e+00 -1.00631189e+00 -1.11785102e+00
1.12695670e+00 2.95855194e-01 7.78968036e-01 1.07741363e-01
-1.27923441e+00 2.60303289e-01 -1.98166743e-01 -8.06447566e-02
-5.86699903e-01 -1.99005112e-01 -8.25103298e-02 1.30674034e-01
6.27990544e-01 -4.52752054e-01 -8.54837537e-01 3.25392902e-01
-1.28116429e+00 -1.40752256e-01 9.12355363e-01 9.56108809e-01
9.95843530e-01 1.44944236e-01 2.30279461e-01 -9.85326290e-01
5.00251055e-01 -4.76568073e-01 -4.39786494e-01 5.46182036e-01
-1.57080963e-01 -7.50782192e-02 5.58623552e-01 -4.23643500e-01
-8.29348087e-01 6.17278814e-02 2.09672540e-01 -5.09584844e-02
-3.75204384e-01 4.43245053e-01 3.95792753e-01 -6.16021574e-01
8.49616289e-01 3.91351789e-01 4.45651084e-01 -2.78957561e-02
-2.30764151e-01 7.47464776e-01 2.14669317e-01 -1.25294060e-01
5.93542159e-01 9.49005783e-01 4.68468666e-01 -1.13303924e+00
-3.64636928e-01 -1.07699001e+00 -8.31110656e-01 1.69066533e-01
8.12887549e-01 -5.00594795e-01 -8.76199901e-01 5.04656553e-01
-5.80854356e-01 -4.79417503e-01 -4.89439107e-02 4.96588141e-01
-4.17092651e-01 7.82290399e-01 -1.05239093e+00 -2.91250795e-01
-3.51443678e-01 -9.53846574e-01 1.28049052e+00 1.53582886e-01
-2.00240299e-01 -1.31564891e+00 6.37226999e-01 3.71354282e-01
2.01873511e-01 3.01803738e-01 1.24703252e+00 -9.17325318e-01
-2.25570537e-02 -8.04270387e-01 -4.70005006e-01 1.49717107e-01
3.15642118e-01 1.98092133e-01 -7.01736033e-01 -3.94101471e-01
-2.55964816e-01 -3.13896924e-01 7.88771629e-01 5.52980781e-01
1.16047251e+00 1.03726901e-01 -7.51569569e-01 6.08519614e-01
1.75652921e+00 -2.59660333e-01 5.64463675e-01 1.69681028e-01
3.35322827e-01 7.73490965e-01 5.41848540e-01 3.91382873e-01
-5.60370684e-02 5.35860419e-01 2.90922318e-02 -2.59250194e-01
-2.67979726e-02 3.93385768e-01 -1.23600267e-01 4.83691096e-01
7.36718103e-02 1.44558281e-01 -1.14120722e+00 6.16051078e-01
-1.61872554e+00 -9.12842095e-01 -4.52563465e-01 2.31941199e+00
8.60877335e-01 -3.88621271e-01 -1.77878410e-01 2.70326138e-01
8.22633922e-01 -1.34699196e-01 -4.79844630e-01 -2.30800450e-01
-7.01498911e-02 1.52627707e-01 4.60550338e-01 1.05473146e-01
-1.02199519e+00 4.34263140e-01 5.92464733e+00 1.41672337e+00
-1.08150971e+00 -7.76595548e-02 1.18305361e+00 2.98908772e-03
1.13446608e-01 -2.54879355e-01 -6.72836602e-01 3.52016002e-01
7.07907915e-01 -6.46971539e-03 -1.35590568e-01 5.03418922e-01
3.32978442e-02 -6.89640164e-01 -1.04483664e+00 9.15221751e-01
-1.11885309e-01 -1.60289097e+00 -2.57597148e-01 3.43441486e-01
7.42451906e-01 -1.55356973e-01 2.14867041e-01 -3.25312138e-01
-3.42478999e-03 -1.03751779e+00 -3.57301325e-01 5.57161093e-01
9.51307297e-01 -7.44809389e-01 1.04208791e+00 2.76617259e-01
-9.77295697e-01 3.45011503e-01 -7.43565381e-01 4.39918190e-01
-5.14012694e-01 1.11924481e+00 -1.36935687e+00 3.21882427e-01
5.40895343e-01 4.60433066e-01 -7.26928532e-01 1.00006139e+00
6.81796432e-01 5.79012513e-01 -2.94659555e-01 -1.24433085e-01
-3.22974660e-02 -2.00918138e-01 1.66561007e-01 1.40615332e+00
4.00813609e-01 1.09193243e-01 -2.27751419e-01 5.04915476e-01
4.46589708e-01 4.16920722e-01 -3.53450537e-01 2.42193993e-02
1.90911382e-01 1.82768333e+00 -1.25701964e+00 -2.26885587e-01
-2.64254153e-01 1.15532303e+00 7.02587664e-01 -8.98732245e-02
-3.47600996e-01 -3.64073664e-01 8.50708008e-01 2.63008714e-01
1.32952988e-01 6.30068257e-02 -1.61302760e-01 -9.19788957e-01
-3.27264406e-02 -4.70577210e-01 6.32159352e-01 -4.16010737e-01
-1.68050313e+00 3.97181392e-01 -2.54940271e-01 -1.24557042e+00
1.72986120e-01 -1.13149238e+00 -9.41247344e-01 6.12101257e-01
-1.55011082e+00 -1.08888519e+00 -2.65982091e-01 6.22253478e-01
-1.50731876e-01 -4.35525812e-02 1.18856597e+00 -1.47988811e-01
-3.65180075e-01 5.96448898e-01 7.96964049e-01 3.75296891e-01
9.36632633e-01 -1.42636514e+00 -2.17475757e-01 1.12791084e-01
-9.35444310e-02 3.56450617e-01 5.46786249e-01 -4.71140325e-01
-1.40247130e+00 -9.86758411e-01 6.84846759e-01 -1.75208449e-01
9.93662119e-01 -6.90147877e-02 -1.01468456e+00 2.74024367e-01
-8.92710537e-02 7.48217762e-01 1.65684462e+00 -3.03575218e-01
-1.45710126e-01 -2.17517778e-01 -1.54246986e+00 4.21369255e-01
3.56291234e-01 -3.79619479e-01 2.01584194e-02 6.63075864e-01
-2.51842231e-01 -2.62352765e-01 -1.47483718e+00 1.41430885e-01
4.56647843e-01 -1.10567129e+00 1.01975870e+00 -4.92888093e-01
2.56394356e-01 -3.17123264e-01 4.62040864e-02 -1.24226272e+00
-3.26499939e-01 -4.76821102e-02 3.59201342e-01 7.72008479e-01
2.49627441e-01 -5.25881052e-01 1.19258273e+00 4.60297793e-01
1.98832989e-01 -1.06146407e+00 -1.25994539e+00 -5.75492203e-01
2.40151539e-01 3.76441237e-03 1.40202940e-01 9.45679605e-01
5.09535849e-01 -5.10547698e-01 4.54230696e-01 3.04967403e-01
8.67768764e-01 1.80049501e-02 5.06307304e-01 -1.21470451e+00
-3.38765770e-01 -4.89162415e-01 -1.42575443e+00 1.20782331e-01
2.27578476e-01 -1.16464603e+00 -1.98259249e-01 -1.06564677e+00
7.76530683e-01 -4.27318186e-01 -3.58375460e-01 1.99075580e-01
-1.47187158e-01 6.79397583e-01 -4.08985794e-01 2.97990263e-01
-4.21099216e-01 -9.04566646e-02 1.21661031e+00 -2.49427333e-01
1.62298918e-01 -1.21574700e-01 -2.57575840e-01 6.11003041e-01
8.98452222e-01 -3.43576401e-01 -3.53240259e-02 2.17054307e-01
1.87593757e-03 1.30450562e-01 4.40335959e-01 -1.07731223e+00
3.42822433e-01 -2.26691395e-01 6.93698525e-01 -4.18048292e-01
1.37148932e-01 -6.82789862e-01 2.67927945e-01 6.14597857e-01
-5.34228265e-01 -2.73827046e-01 1.30497515e-01 7.19485939e-01
-2.86651611e-01 -3.44042897e-01 1.03936493e+00 -9.34515521e-02
-3.65025818e-01 3.52104634e-01 -6.27721786e-01 -3.83463204e-01
1.63082612e+00 -5.33897221e-01 -2.57318437e-01 2.69881517e-01
-7.96583891e-01 -2.67024457e-01 5.59516549e-01 -4.23971713e-01
7.22036123e-01 -1.22163010e+00 -7.87791371e-01 7.80066922e-02
4.91208702e-01 -1.73183501e-01 7.24010885e-01 1.48643219e+00
-8.69590700e-01 3.20967793e-01 -1.55580193e-01 -7.95438886e-01
-1.95126557e+00 3.63076389e-01 2.85150856e-01 -8.11052203e-01
-4.69691753e-01 1.14967573e+00 3.03076923e-01 -4.94598031e-01
-1.08533561e-01 -1.86621696e-01 -6.51070923e-02 -5.23534371e-03
7.47338235e-01 2.17678070e-01 4.49367732e-01 -5.97424209e-01
-3.33455443e-01 7.75345504e-01 -6.00196779e-01 3.11679393e-01
1.32117295e+00 1.87451139e-01 -5.56692243e-01 3.88301134e-01
1.84195173e+00 -1.20182723e-01 -8.69294405e-01 -2.53291458e-01
8.35697874e-02 -5.30560613e-01 5.60202710e-02 -6.06290638e-01
-9.03519630e-01 7.57908285e-01 9.15337026e-01 4.16833572e-02
1.08290553e+00 4.59202886e-01 5.75188637e-01 4.26626652e-01
-2.74817809e-03 -1.01820409e+00 1.76293422e-02 -1.28526971e-01
4.24237311e-01 -1.46484315e+00 1.94348022e-02 -4.65156943e-01
-4.85102773e-01 1.53906345e+00 1.69079870e-01 -4.60179448e-01
9.59952533e-01 5.15694439e-01 2.34752774e-01 -3.65033686e-01
-7.07298696e-01 2.43594274e-02 4.95982677e-01 7.49326050e-01
7.41179645e-01 2.04477906e-01 -5.45584738e-01 2.94792324e-01
2.13983819e-01 -1.07452616e-01 3.25260848e-01 8.67528975e-01
-4.13553894e-01 -1.02872753e+00 -3.58449161e-01 1.00115716e+00
-4.26869363e-01 1.04279079e-01 -2.96495885e-01 6.14963830e-01
-1.23361059e-01 3.30999583e-01 1.80603251e-01 -1.41672641e-01
-1.31913841e-01 -8.56671482e-02 6.25928164e-01 -3.42818856e-01
-3.30986977e-01 -2.28485450e-01 -4.20550913e-01 -3.71051788e-01
-3.05290490e-01 -8.11785996e-01 -1.37424064e+00 -3.09931457e-01
-3.01121891e-01 8.62682760e-02 6.31159961e-01 7.77429044e-01
3.02936673e-01 7.24938661e-02 5.29991329e-01 -5.81582069e-01
-4.63986039e-01 -7.26596236e-01 -1.31944454e+00 4.06124473e-01
3.17187726e-01 -2.10098192e-01 -3.28369528e-01 1.55926511e-01]
|
[15.002342224121094, -2.912241220474243]
|
b25e4ec4-aa81-4c0c-b8c8-12008911f0da
|
icfvr-2017-3rd-international-competition-on
|
1801.01262
| null |
http://arxiv.org/abs/1801.01262v1
|
http://arxiv.org/pdf/1801.01262v1.pdf
|
ICFVR 2017: 3rd International Competition on Finger Vein Recognition
|
In recent years, finger vein recognition has become an important sub-field in
biometrics and been applied to real-world applications. The development of
finger vein recognition algorithms heavily depends on large-scale real-world
data sets. In order to motivate research on finger vein recognition, we
released the largest finger vein data set up to now and hold finger vein
recognition competitions based on our data set every year. In 2017,
International Competition on Finger Vein Recognition(ICFVR) is held jointly
with IJCB 2017. 11 teams registered and 10 of them joined the final evaluation.
The winner of this year dramatically improved the EER from 2.64% to 0.483%
compared to the winner of last year. In this paper, we introduce the process
and results of ICFVR 2017 and give insights on development of state-of-art
finger vein recognition algorithms.
|
['Yingjie Chen', 'Wei Xu', 'Nasir Uddin Ahmed', 'Md. Shakil Ahmed', 'Liao Ni', 'Yilun Jin', 'Jingxuan Wen', 'Houjun Huang', 'Yi Zhang', 'Wenxin Li', 'Haifeng Zhang']
|
2018-01-04
| null | null | null | null |
['finger-vein-recognition']
|
['computer-vision']
|
[ 2.70708978e-01 -3.48059952e-01 -1.92210823e-01 -4.09094155e-01
-1.80589780e-01 -8.80062819e-01 4.27184403e-01 -4.39456284e-01
-5.78793883e-01 6.24481261e-01 2.15750694e-01 3.01759224e-02
2.55390018e-01 -8.85244429e-01 1.84540913e-01 -3.40323657e-01
7.09409192e-02 2.50526756e-01 3.01313311e-01 1.07443318e-01
3.60760629e-01 1.10644734e+00 -8.16626787e-01 3.73179793e-01
3.45995069e-01 1.02630198e+00 -7.10252166e-01 9.29252505e-01
-2.66851127e-01 1.70597032e-01 -5.96037984e-01 -1.17104483e+00
6.61776364e-01 -4.67989057e-01 -7.32712030e-01 -1.70835435e-01
7.55303860e-01 -5.43918312e-01 -1.03568542e+00 5.98796904e-01
1.12586486e+00 -4.89401698e-01 6.44946098e-01 -8.21177423e-01
-4.61683154e-01 3.24195266e-01 -8.37722361e-01 2.90294200e-01
2.63661414e-01 3.35206598e-01 6.59721434e-01 -7.47410834e-01
6.66496813e-01 1.01672006e+00 6.11970007e-01 1.04429460e+00
-1.14838517e+00 -6.95227265e-01 -2.36126319e-01 -9.16842744e-03
-1.14219069e+00 -3.78579795e-01 7.37642288e-01 -2.05761924e-01
6.87280059e-01 5.29286504e-01 1.01554203e+00 1.45719993e+00
6.55615255e-02 1.04316115e+00 1.52835548e+00 -2.53878415e-01
-2.88123220e-01 -2.59369671e-01 4.87495542e-01 7.18311131e-01
6.55090392e-01 3.09616894e-01 -6.36299491e-01 -1.97327062e-01
1.54155767e+00 -5.52272685e-02 1.67041436e-01 1.17910266e-01
-1.07852256e+00 6.13882005e-01 2.58433014e-01 2.70031035e-01
-4.37587976e-01 1.36543781e-01 3.40654612e-01 4.68480587e-01
-1.51587173e-01 3.30018222e-01 -1.23219639e-01 -4.70674336e-01
-9.34315920e-01 3.37426305e-01 1.25077963e+00 6.95435584e-01
-1.30227178e-01 -2.89579630e-01 -8.24860334e-01 9.97355402e-01
4.53480899e-01 7.02075779e-01 5.89357466e-02 -6.66937172e-01
3.02838027e-01 4.10121083e-01 1.36784688e-02 -8.16622138e-01
-1.70242637e-02 -3.70984733e-01 -1.22148407e+00 -5.38253784e-02
1.29918575e+00 -1.28743142e-01 -1.20040572e+00 8.23762178e-01
-4.76147413e-01 1.39372006e-01 -4.68904406e-01 7.66821980e-01
1.10437655e+00 -2.49101266e-01 -1.43712476e-01 1.38791010e-01
1.46665311e+00 -5.98104417e-01 -5.73107421e-01 1.12158567e-01
-4.64736760e-01 -1.09194958e+00 4.53387856e-01 8.57225001e-01
-1.01307786e+00 -5.80044925e-01 -8.11376035e-01 2.61199743e-01
-1.45489648e-01 4.04310264e-02 1.00806904e+00 1.61532021e+00
-8.51548553e-01 4.24155802e-01 -6.20242536e-01 -6.79203510e-01
1.00727141e+00 4.79181767e-01 -4.11557645e-01 -1.74823031e-01
-9.41093504e-01 7.96606719e-01 -4.57638323e-01 4.72437114e-01
-3.62404734e-01 -5.48358619e-01 6.41895905e-02 -5.95222175e-01
-3.20349983e-03 -3.92133087e-01 7.64920235e-01 2.49118917e-02
-1.51965261e+00 1.36713755e+00 -4.20011371e-01 -1.82313606e-01
1.12644470e+00 -3.65799814e-01 -8.73583972e-01 -5.54705085e-03
-3.64793152e-01 2.08312392e-01 8.29428911e-01 -7.12214649e-01
-4.84607071e-01 -7.23722279e-01 -4.90770549e-01 -6.70663893e-01
-2.03298509e-01 1.44674942e-01 -8.73337507e-01 -6.97316825e-01
1.97295025e-01 -6.73520803e-01 -2.38762394e-01 1.69897154e-01
-4.18689191e-01 -4.01467144e-01 4.92130786e-01 -7.83798933e-01
1.31129777e+00 -1.71459794e+00 -4.09210294e-01 9.39684868e-01
6.47246361e-01 4.71913040e-01 -3.33562225e-01 4.14836913e-01
1.15378849e-01 1.24125585e-01 1.14278860e-01 1.84197545e-01
-1.10273547e-01 -1.53574556e-01 -4.48448747e-01 4.43741232e-01
1.90590814e-01 1.24285769e+00 -6.49060845e-01 -2.42206424e-01
3.19640607e-01 5.74631274e-01 -2.00609192e-01 2.16635942e-01
7.42123902e-01 5.06785572e-01 -5.96948743e-01 1.16664767e+00
9.65624630e-01 -4.76802550e-02 1.72905102e-01 -7.79718161e-01
4.71253432e-02 -4.00139838e-01 -1.04995871e+00 1.29089189e+00
2.71167278e-01 6.83209360e-01 -3.83351505e-01 -4.56097484e-01
1.52817249e+00 2.33052269e-01 8.48039865e-01 -6.99327290e-01
2.29214668e-01 5.03706396e-01 1.97181061e-01 -2.77645916e-01
1.57289341e-01 1.54562190e-01 3.03379476e-01 1.98974520e-01
1.16491407e-01 1.63635150e-01 2.99028426e-01 -1.07901707e-01
1.24678802e+00 2.49199278e-04 7.29641914e-02 -5.27378842e-02
6.64736450e-01 -5.70523202e-01 2.26111799e-01 1.27510774e+00
-8.64567816e-01 7.92595923e-01 4.02044356e-01 -7.77635276e-01
-8.64425123e-01 -1.53021479e+00 -5.04705131e-01 1.56420052e-01
-3.15355398e-02 -4.94691700e-01 -6.56381369e-01 -7.71985769e-01
4.41833586e-01 -2.97895283e-01 -8.02879214e-01 2.70112932e-01
-8.78319502e-01 -7.48078406e-01 1.33947861e+00 6.60544872e-01
1.10472965e+00 -1.18427014e+00 -4.37226802e-01 2.11364105e-01
9.72043276e-02 -9.91379738e-01 -4.96802539e-01 -3.76540065e-01
-7.07090914e-01 -1.37789416e+00 -1.49505889e+00 -5.03761947e-01
3.44321519e-01 -3.32659811e-01 1.09292710e+00 7.11533725e-02
-1.28632522e+00 5.13706386e-01 -1.67799458e-01 -3.51268619e-01
3.50315809e-01 1.87871188e-01 -2.62332022e-01 3.27889919e-01
1.16490316e+00 -1.51408851e-01 -8.53886843e-01 4.11546052e-01
-3.45640838e-01 -6.53976679e-01 7.59704232e-01 9.92329895e-01
3.21386874e-01 -5.64300358e-01 4.27423507e-01 -9.86112177e-01
9.26010668e-01 3.20801139e-01 -4.59670454e-01 5.98363996e-01
-5.43067873e-01 -1.47027373e-01 -2.35606264e-02 -1.96765468e-01
-7.63468742e-01 -1.16036069e-02 -2.38575175e-01 3.61168325e-01
-5.45136519e-02 9.09004360e-03 9.96895283e-02 -6.08439326e-01
5.54754555e-01 2.36323819e-01 1.02694668e-01 -5.02575099e-01
9.81081724e-02 1.04817462e+00 6.44136608e-01 -6.74424052e-01
8.57377470e-01 3.58154416e-01 2.65866190e-01 -1.03899324e+00
-1.90935731e-01 -5.36393523e-01 -8.80143762e-01 -4.04919565e-01
6.01192832e-01 -3.27679247e-01 -1.22826982e+00 1.21403646e+00
-9.02729273e-01 9.29590836e-02 -6.09629415e-02 2.37009645e-01
-5.95841855e-02 7.42317140e-01 -6.50410175e-01 -8.75439763e-01
-6.74781680e-01 -8.24451745e-01 5.52023351e-01 6.03276908e-01
-3.65770668e-01 -7.81154335e-01 3.56518105e-02 3.42951506e-01
1.02452600e+00 3.79358381e-01 1.02550223e-01 -1.70787126e-01
-3.75992507e-01 -7.37729430e-01 -6.72493815e-01 1.77740112e-01
5.75489819e-01 8.41604322e-02 -1.14802563e+00 -3.10138464e-01
-5.84677756e-01 -5.34702502e-02 1.12548184e+00 3.23799103e-01
1.38581395e+00 6.44619823e-01 -4.56082672e-01 5.28117597e-01
1.21417809e+00 2.46313229e-01 1.32311380e+00 -4.76323292e-02
5.18197596e-01 5.32614410e-01 2.11199701e-01 6.23219013e-01
-8.07955116e-02 5.17813265e-01 -2.37329364e-01 -2.81940132e-01
-4.59462494e-01 -1.18923143e-01 -2.56359160e-01 6.71785921e-02
-9.90685046e-01 2.70588808e-02 -6.93795443e-01 6.43353239e-02
-1.36886024e+00 -1.01038027e+00 -2.77516335e-01 2.31949091e+00
7.16911733e-01 -1.91004400e-03 7.29149699e-01 1.16422139e-02
5.92290401e-01 7.54481927e-03 -6.96060777e-01 -5.04722707e-02
-3.72174621e-01 1.25622404e+00 8.25231373e-01 2.50319868e-01
-1.35913348e+00 1.07685447e+00 6.94555950e+00 4.25944686e-01
-1.14553833e+00 -4.59035844e-01 6.92800343e-01 2.70958602e-01
7.02809691e-02 -5.25591850e-01 -9.20101345e-01 2.81594515e-01
3.04471910e-01 4.01489362e-02 7.50752330e-01 2.36550570e-01
-3.48540992e-01 2.56360412e-01 -9.68625367e-01 1.71989036e+00
-7.55639654e-03 -1.25104654e+00 -2.57455204e-02 3.84311050e-01
2.10571393e-01 -9.04271752e-02 6.55281618e-02 -4.81157415e-02
-5.28923944e-02 -1.71333992e+00 -2.33282879e-01 1.01686037e+00
1.24602771e+00 -5.21108568e-01 9.55418229e-01 -6.15451694e-01
-1.49401033e+00 3.52360159e-01 -4.38506693e-01 3.48894536e-01
-1.96230672e-02 3.14407259e-01 -4.87298012e-01 3.61448050e-01
5.94022691e-01 8.40618670e-01 -9.92929757e-01 1.42238271e+00
-8.19358528e-02 7.46483266e-01 -3.45670849e-01 -4.82151002e-01
-2.58688539e-01 -1.18037470e-01 2.79888183e-01 1.44465852e+00
-1.82581499e-01 6.91133514e-02 -2.17787966e-01 7.17981994e-01
-1.25334352e-01 -1.10181376e-01 -2.41493717e-01 -4.16942000e-01
4.50007766e-01 1.41006792e+00 -7.25391209e-01 -9.46085006e-02
-5.06778538e-01 1.40571034e+00 -2.26076230e-01 4.26965922e-01
-3.47491175e-01 -6.92449093e-01 5.09431779e-01 1.87715609e-02
-5.01824059e-02 -4.19776179e-02 -7.90841401e-01 -1.31640208e+00
2.77499408e-01 -7.12641776e-01 4.15921837e-01 2.94354379e-01
-2.20218635e+00 6.78039491e-01 -6.36830151e-01 -8.91978920e-01
9.06886533e-02 -1.15559685e+00 -4.29061890e-01 1.49676967e+00
-1.42124319e+00 -1.29158771e+00 -7.77156830e-01 7.13393450e-01
-4.10161316e-02 -8.50456119e-01 1.34888506e+00 3.08125585e-01
-3.01903337e-01 1.35728288e+00 -2.92423517e-01 1.00624418e+00
1.15780711e+00 -1.26580191e+00 1.11607146e+00 6.63948298e-01
8.70401487e-02 1.13120520e+00 9.34610739e-02 -7.38081753e-01
-1.81838214e+00 -6.50062442e-01 9.26244557e-01 -8.34754646e-01
3.37178260e-01 -1.91793188e-01 -1.98950261e-01 3.63429785e-01
1.69257835e-01 4.36055601e-01 1.02563894e+00 1.47551537e-01
-7.08523154e-01 -4.28345859e-01 -1.63585603e+00 6.48016870e-01
1.15039611e+00 -3.77310753e-01 -2.82581121e-01 -2.11014092e-01
-6.76134169e-01 -3.50905061e-01 -1.32180250e+00 4.51400578e-01
1.65672970e+00 -7.52017260e-01 1.14311779e+00 -8.06251407e-01
2.31871363e-02 -2.70329863e-01 -1.13982074e-01 -5.13747334e-01
-5.11460543e-01 -8.58738899e-01 -2.71806151e-01 1.09533429e+00
3.50449272e-02 -8.77127767e-01 1.50577700e+00 8.01502824e-01
7.40290582e-01 -7.96802998e-01 -7.23673999e-01 -6.30729914e-01
1.06816031e-01 -2.85958033e-02 3.58231932e-01 5.62682867e-01
-2.18113184e-01 -9.68072191e-02 -6.58425212e-01 -4.73587453e-01
1.36845267e+00 4.44581965e-04 9.72959280e-01 -1.58453798e+00
-4.54565614e-01 -7.61034846e-01 -8.51409376e-01 -1.24867392e+00
-6.46188021e-01 -6.43989623e-01 -5.98201454e-01 -1.40700603e+00
2.95755088e-01 -9.47536528e-02 -5.90252221e-01 4.63298827e-01
-2.89083749e-01 1.08336949e+00 3.18674803e-01 -7.98330382e-02
5.97555081e-05 -4.14174557e-01 1.45739579e+00 -2.80225694e-01
-8.11326206e-02 2.41548419e-01 -8.62607121e-01 -4.87002470e-02
9.44567084e-01 4.32319164e-01 1.53218701e-01 3.73340771e-02
-4.34198320e-01 -2.69957840e-01 2.49086633e-01 -5.86943865e-01
1.35454386e-01 1.29230306e-01 1.37404263e+00 -6.22178853e-01
2.06064671e-01 -6.32445812e-01 -1.68735638e-01 9.55937445e-01
-1.54032305e-01 -4.74883914e-01 -7.93467462e-02 8.93069729e-02
-4.86724488e-02 3.30438823e-01 7.37678587e-01 -1.74523607e-01
-7.80022860e-01 8.69299054e-01 -2.60775417e-01 4.85234670e-02
7.71190584e-01 -7.45552599e-01 -3.84702444e-01 1.54500897e-03
-8.41166019e-01 4.93549630e-02 -1.74546003e-01 5.68101883e-01
9.85642433e-01 -1.20223355e+00 -1.24571943e+00 6.36233389e-01
3.11442673e-01 -1.00220525e+00 -3.14296293e-03 5.77217996e-01
-8.05344582e-01 5.29513538e-01 -6.37389123e-01 -3.75395954e-01
-1.55034983e+00 -3.34872454e-01 3.85671079e-01 -2.57737160e-01
-6.35902345e-01 1.02758372e+00 -6.77707613e-01 -1.55938268e-01
5.01439273e-01 1.93622142e-01 -3.96554977e-01 -2.17108414e-01
9.19132054e-01 4.63216931e-01 -1.18122771e-01 3.71404327e-02
-6.58088803e-01 7.85459697e-01 -3.09724152e-01 -6.27236441e-03
1.28620160e+00 5.99034607e-01 -2.93234020e-01 -7.23781362e-02
8.74566376e-01 1.24420553e-01 -8.17033231e-01 -8.87596011e-02
1.90753415e-01 -9.38741922e-01 -4.51749206e-01 -1.49364662e+00
-1.37785363e+00 8.05796802e-01 1.20414519e+00 -1.44214064e-01
9.42425907e-01 -2.96781242e-01 9.88120377e-01 3.41627896e-01
6.81777954e-01 -9.16683793e-01 -8.42133239e-02 4.63221788e-01
8.25713694e-01 -1.02543211e+00 -1.13921709e-01 -7.82338440e-01
-3.45490932e-01 1.83089328e+00 5.66047907e-01 -5.37720859e-01
9.25444007e-01 3.87573749e-01 3.69610369e-01 -1.15987612e-02
2.56714046e-01 1.45291854e-02 6.56216919e-01 1.16018701e+00
1.09984314e+00 2.81238019e-01 -8.50744545e-01 7.25882292e-01
1.72304027e-02 4.75120038e-01 -1.23544835e-01 7.53788829e-01
-2.64715264e-03 -2.04981375e+00 -9.35254693e-02 1.14018881e+00
-5.90672970e-01 6.67402893e-03 -9.76332605e-01 5.33364236e-01
-2.11264506e-01 9.83998775e-01 -2.17228934e-01 -6.26199305e-01
5.20017564e-01 -2.59697020e-01 1.51362324e+00 -6.71387613e-02
-8.17680061e-01 9.65921432e-02 3.76985185e-02 -7.41336346e-01
-1.62253022e-01 -5.35932124e-01 -7.13484168e-01 -5.51349521e-01
1.50721237e-01 -6.82241172e-02 3.00859720e-01 5.82477570e-01
-9.16686002e-03 3.35115701e-01 5.48163354e-01 -2.76236743e-01
-6.65877998e-01 -1.00817704e+00 -9.51280653e-01 2.18098313e-01
7.58470148e-02 -3.74878913e-01 2.55934775e-01 -1.94552317e-01]
|
[13.030921936035156, 1.019033670425415]
|
8d82fe59-3c86-4a4d-b21c-7d9b162b6839
|
faceqan-face-image-quality-assessment-through
|
2212.02127
| null |
https://arxiv.org/abs/2212.02127v1
|
https://arxiv.org/pdf/2212.02127v1.pdf
|
FaceQAN: Face Image Quality Assessment Through Adversarial Noise Exploration
|
Recent state-of-the-art face recognition (FR) approaches have achieved impressive performance, yet unconstrained face recognition still represents an open problem. Face image quality assessment (FIQA) approaches aim to estimate the quality of the input samples that can help provide information on the confidence of the recognition decision and eventually lead to improved results in challenging scenarios. While much progress has been made in face image quality assessment in recent years, computing reliable quality scores for diverse facial images and FR models remains challenging. In this paper, we propose a novel approach to face image quality assessment, called FaceQAN, that is based on adversarial examples and relies on the analysis of adversarial noise which can be calculated with any FR model learned by using some form of gradient descent. As such, the proposed approach is the first to link image quality to adversarial attacks. Comprehensive (cross-model as well as model-specific) experiments are conducted with four benchmark datasets, i.e., LFW, CFP-FP, XQLFW and IJB-C, four FR models, i.e., CosFace, ArcFace, CurricularFace and ElasticFace, and in comparison to seven state-of-the-art FIQA methods to demonstrate the performance of FaceQAN. Experimental results show that FaceQAN achieves competitive results, while exhibiting several desirable characteristics.
|
['Vitomir Štruc', 'Peter Peer', 'Žiga Babnik']
|
2022-12-05
| null | null | null | null |
['face-image-quality', 'face-image-quality-assessment']
|
['computer-vision', 'computer-vision']
|
[ 1.30964160e-01 -3.97875965e-01 1.70658574e-01 -5.95727980e-01
-9.39940274e-01 -3.62422168e-01 5.47485650e-01 -5.94678819e-01
-9.75849666e-03 5.96374869e-01 -1.78851530e-01 -8.02205503e-02
-2.55660236e-01 -6.11186922e-01 -7.21978605e-01 -6.81092203e-01
-1.38272241e-01 1.25203058e-01 -4.23066586e-01 -3.00522417e-01
1.26217857e-01 9.21534359e-01 -1.59835100e+00 3.45581234e-01
9.04014349e-01 1.52807271e+00 -6.04106069e-01 5.48749149e-01
5.33648245e-02 7.09085226e-01 -7.53774107e-01 -1.16169858e+00
4.28992122e-01 -3.94914001e-01 -6.74635410e-01 9.52451229e-02
9.37843561e-01 -3.96564335e-01 -4.60762620e-01 1.21724665e+00
6.57509327e-01 -1.10273011e-01 6.10636055e-01 -1.53115594e+00
-7.64436007e-01 -3.13845538e-02 -4.78161454e-01 1.13938421e-01
5.94221532e-01 5.12947440e-01 6.89625025e-01 -1.29099655e+00
3.84926051e-01 1.86287081e+00 6.40917838e-01 8.88535440e-01
-1.24063838e+00 -9.93839860e-01 -1.04841143e-01 4.20724571e-01
-1.40595186e+00 -8.13229680e-01 8.68908525e-01 -2.39573240e-01
3.77767593e-01 2.49670193e-01 2.61597335e-01 1.15180147e+00
1.26506716e-01 6.82143807e-01 1.52937174e+00 -3.20240796e-01
1.97170690e-01 -1.39848426e-01 -2.30662107e-01 8.34255934e-01
-8.94084126e-02 5.20101249e-01 -7.18998551e-01 -2.21719772e-01
5.56584895e-01 -2.77090549e-01 -4.34960693e-01 -1.33472696e-01
-5.58610618e-01 6.65523052e-01 6.50962830e-01 2.31685385e-01
-2.54225194e-01 7.25115314e-02 2.16030672e-01 6.09727263e-01
4.64765459e-01 4.14577015e-02 -1.14034928e-01 2.04566449e-01
-9.98436511e-01 2.65849799e-01 6.51436210e-01 4.03300017e-01
6.52206302e-01 4.99234200e-01 -3.46539348e-01 8.26424420e-01
5.34763277e-01 9.40515161e-01 2.23579139e-01 -9.80153859e-01
3.53274614e-01 3.54941756e-01 3.58981849e-03 -1.22309983e+00
7.82869831e-02 -1.26369148e-01 -8.35674107e-01 7.96166718e-01
3.79482716e-01 1.60571367e-01 -1.09350514e+00 1.65103352e+00
2.67344534e-01 5.49162686e-01 1.37007743e-01 7.95914769e-01
1.00380409e+00 4.85751927e-01 1.46446675e-01 -2.44482234e-01
1.10994887e+00 -7.00787723e-01 -7.03248203e-01 1.61101192e-01
-2.58240074e-01 -8.98913622e-01 1.01314378e+00 7.32989192e-01
-1.06348670e+00 -9.57139492e-01 -1.00397110e+00 4.77476269e-01
-1.12876274e-01 -1.54996905e-02 3.24031204e-01 1.25303423e+00
-1.19629788e+00 6.73137605e-01 -5.78584969e-01 2.64792114e-01
9.22610700e-01 5.50413132e-01 -7.77132630e-01 -5.21955490e-01
-1.10187769e+00 7.50928581e-01 9.62219853e-03 2.21250281e-01
-1.47648394e+00 -9.44885135e-01 -7.40022659e-01 -4.21875995e-03
2.40129530e-01 -3.12042892e-01 9.78444636e-01 -1.37520802e+00
-1.63018405e+00 7.46518314e-01 -1.19121328e-01 -1.60234138e-01
5.82391262e-01 -2.72829145e-01 -9.74177063e-01 4.41813856e-01
-4.03000712e-01 5.72410047e-01 1.37424994e+00 -1.59166574e+00
-2.06328079e-01 -6.94348931e-01 2.25557778e-02 -2.21832573e-01
-2.96859831e-01 4.85651433e-01 -3.50416929e-01 -6.22860551e-01
-3.63724202e-01 -6.49505079e-01 2.01458663e-01 5.82098007e-01
-2.41317168e-01 -2.18383759e-01 9.78447318e-01 -8.23323667e-01
1.00879252e+00 -2.20633078e+00 8.09226483e-02 5.25904655e-01
-1.63416062e-02 7.62831092e-01 -5.58737099e-01 1.02737896e-01
-2.04246253e-01 2.55876392e-01 -2.64916360e-01 -3.82029980e-01
-3.20879146e-02 1.25996932e-01 -2.14965656e-01 6.19016588e-01
7.73732901e-01 9.15814519e-01 -6.48737133e-01 -4.50263858e-01
1.34508699e-01 9.19973314e-01 -5.05883813e-01 5.32616019e-01
8.39459710e-03 5.25121927e-01 -2.20691651e-01 1.07166076e+00
1.04014945e+00 1.62814572e-01 -4.48147319e-02 -5.19085407e-01
4.76693362e-01 -4.89487082e-01 -1.31851733e+00 1.20493102e+00
-4.31005388e-01 2.50889540e-01 1.55968130e-01 -7.16187358e-01
9.27040815e-01 4.29894716e-01 2.53070742e-01 -8.33505630e-01
2.53950268e-01 2.60668337e-01 1.70949548e-02 -4.19418693e-01
-1.05133109e-01 -8.39141607e-02 4.36434984e-01 1.59716710e-01
4.49805856e-01 2.38443743e-02 2.24355966e-01 -1.10753000e-01
8.78719926e-01 3.06377281e-02 1.88663621e-02 -1.78638220e-01
1.10604608e+00 -7.15590298e-01 4.43108112e-01 3.70314002e-01
-5.27320504e-01 5.34780979e-01 2.61167347e-01 -3.54444385e-01
-7.32110977e-01 -1.25241041e+00 -3.17580581e-01 6.05821729e-01
-5.46351122e-03 -1.95782274e-01 -1.10444129e+00 -9.35013950e-01
1.91398963e-01 2.18269706e-01 -7.98609316e-01 -2.67068952e-01
-5.14629543e-01 -4.86961633e-01 8.04812133e-01 2.89522648e-01
7.26451457e-01 -1.32216787e+00 7.25688273e-03 -1.15745366e-01
1.54216170e-01 -1.05954218e+00 -3.91995847e-01 -5.42014956e-01
-5.06116807e-01 -1.31845379e+00 -6.59406900e-01 -4.41488206e-01
7.23925054e-01 -4.42332812e-02 1.32428074e+00 5.09562314e-01
-4.85654205e-01 5.35583973e-01 -3.42129320e-01 -1.86777055e-01
-5.67933798e-01 -7.21402168e-01 2.33965009e-01 6.25153303e-01
1.18518136e-01 -2.65319109e-01 -6.19635880e-01 6.20037854e-01
-1.16952157e+00 -7.23776340e-01 5.62558293e-01 1.20236075e+00
6.70261919e-01 1.36530161e-01 5.64231873e-01 -8.47701013e-01
5.50120652e-01 -3.62438381e-01 -5.52413523e-01 5.78184783e-01
-6.24863088e-01 -8.59912932e-02 7.78197229e-01 -2.99426347e-01
-1.24108422e+00 -2.61700124e-01 -5.75134039e-01 -8.33368719e-01
-2.61841327e-01 2.84706414e-01 -7.21604288e-01 -8.56124163e-01
7.67074168e-01 3.29010114e-02 2.58288324e-01 -3.06811869e-01
3.16828847e-01 4.46601063e-01 6.65386915e-01 -6.46112621e-01
1.15286899e+00 3.68044347e-01 1.17654055e-01 -6.88840032e-01
-5.98475397e-01 2.50634607e-02 -3.25509727e-01 -6.57822907e-01
3.79642993e-01 -7.34384179e-01 -9.54621017e-01 7.38281012e-01
-8.55244458e-01 5.57681657e-02 -5.61784999e-03 5.91894910e-02
-5.51905930e-01 4.69223797e-01 -5.96718848e-01 -9.61705327e-01
-6.08418345e-01 -1.40652382e+00 1.05160415e+00 2.76320875e-01
3.55668873e-01 -7.17709541e-01 -2.05099314e-01 5.70822597e-01
5.70939481e-01 5.03424704e-01 6.38934791e-01 -2.38570228e-01
-4.98786479e-01 -2.08459809e-01 -2.34942347e-01 8.50779772e-01
1.17054328e-01 2.61144876e-01 -1.17065871e+00 -6.35869920e-01
-5.48002012e-02 -8.48053753e-01 6.45652413e-01 2.02467833e-02
1.42291176e+00 -4.44799095e-01 1.50049105e-01 7.71873832e-01
1.48245907e+00 6.82087764e-02 1.18285751e+00 -1.44901440e-01
3.38647425e-01 4.70835358e-01 6.37130678e-01 3.11866462e-01
-1.95527121e-01 7.92193532e-01 7.64795303e-01 -5.23626916e-02
-3.59623015e-01 -8.18589702e-02 6.58904493e-01 3.99365038e-01
-1.16133898e-01 -1.33636340e-01 -5.96120179e-01 -5.71248680e-02
-1.20740306e+00 -1.19470370e+00 4.31924671e-01 1.94019878e+00
8.07168067e-01 -5.48604727e-02 2.48366985e-02 6.79989934e-01
5.01606941e-01 4.18047905e-02 -7.02670991e-01 -3.29937190e-01
-1.32247448e-01 7.23090768e-01 -2.46035591e-01 3.26401711e-01
-1.02415824e+00 8.34842324e-01 6.29753017e+00 1.10359037e+00
-1.16248214e+00 -1.69759747e-02 1.15879428e+00 2.13597551e-01
-1.11571461e-01 -5.79285800e-01 -7.01412380e-01 4.57162410e-01
8.63231719e-01 6.96255593e-03 8.72133195e-01 8.48927379e-01
-7.41614774e-02 3.59826088e-01 -9.71349537e-01 1.26682603e+00
3.87831420e-01 -8.87632370e-01 2.46043548e-01 -5.60585298e-02
7.32554317e-01 -4.07292336e-01 5.11917710e-01 2.08586961e-01
1.34646460e-01 -1.55874610e+00 5.69920123e-01 7.08559036e-01
1.27869952e+00 -1.08877957e+00 7.69678414e-01 -1.81275100e-01
-1.13417244e+00 -1.99170291e-01 -4.21229273e-01 6.05613887e-01
-2.16674358e-01 5.79827011e-01 -2.56012619e-01 8.60824466e-01
9.93747234e-01 3.96364957e-01 -6.64779007e-01 8.79538357e-01
-2.39926815e-01 6.95095003e-01 -1.74338967e-02 4.19268280e-01
-1.54665187e-01 8.78438726e-03 3.22401196e-01 8.12348902e-01
3.46013069e-01 5.77478670e-02 -5.53735234e-02 8.66668940e-01
-5.50958991e-01 2.47854486e-01 -2.75848567e-01 -2.54207347e-02
3.04823577e-01 1.49025381e+00 -2.21363813e-01 6.66400045e-02
-3.69082689e-01 8.45614195e-01 2.78100401e-01 2.18175858e-01
-7.73605168e-01 -9.21515003e-02 9.67899263e-01 -1.18119873e-01
2.14174956e-01 1.00438982e-01 2.59714723e-01 -9.49521065e-01
1.77514344e-01 -1.60547805e+00 3.53104442e-01 -5.35440028e-01
-1.63987827e+00 1.07711208e+00 -3.81994009e-01 -1.05185401e+00
-1.60207152e-01 -9.32921350e-01 -6.19027495e-01 9.20362592e-01
-1.90068913e+00 -1.22528827e+00 -5.52728176e-01 1.00466275e+00
2.93597221e-01 -5.88512719e-01 8.44996572e-01 5.09556532e-01
-5.29397190e-01 1.33185959e+00 -7.06752064e-03 2.24244535e-01
6.02344036e-01 -8.59331250e-01 2.74976373e-01 8.38115156e-01
2.53873914e-01 4.12504733e-01 4.77577418e-01 -3.20149571e-01
-1.75406659e+00 -1.14783895e+00 2.37279281e-01 -3.27102840e-01
2.79846579e-01 -1.54042067e-02 -1.00464785e+00 -1.08746767e-01
1.05745189e-01 7.23373413e-01 7.80754387e-01 -3.78453851e-01
-7.56836891e-01 -5.31039715e-01 -1.81877291e+00 3.41406494e-01
1.08370090e+00 -6.09319389e-01 -7.74548724e-02 -6.16997704e-02
2.05432013e-01 -1.63197726e-01 -1.10714698e+00 7.53808737e-01
6.43657982e-01 -1.15157974e+00 1.25739634e+00 -7.33183384e-01
2.94061959e-01 -3.91281575e-01 -3.22275728e-01 -1.37196887e+00
-1.33365601e-01 -4.17996854e-01 -3.15497041e-01 1.47519338e+00
6.74952045e-02 -3.17760259e-01 8.06057274e-01 4.28593397e-01
1.27373740e-01 -7.86105514e-01 -1.14978004e+00 -8.98244023e-01
2.13967147e-03 -3.29276413e-01 9.55045283e-01 6.38423860e-01
-7.10343659e-01 -3.52784425e-01 -5.53736687e-01 2.27276042e-01
9.09701705e-01 -1.99427366e-01 8.12172294e-01 -1.19388711e+00
-2.61790991e-01 -3.79867613e-01 -7.17442393e-01 -3.87038141e-01
5.37487507e-01 -7.68369377e-01 -6.18710369e-02 -8.75047445e-01
8.71861354e-02 -3.04253012e-01 -4.60767448e-01 4.35371786e-01
-4.61656868e-01 8.69490445e-01 3.59676301e-01 8.56929272e-02
-3.70625079e-01 7.51907468e-01 1.31733406e+00 -3.85129303e-01
4.87948418e-01 -7.77597353e-02 -4.64206874e-01 3.36222857e-01
5.97816408e-01 -2.03067020e-01 -1.90984696e-01 -1.42896637e-01
-1.55429944e-01 7.86103457e-02 4.63917673e-01 -1.31612909e+00
-9.36606303e-02 -7.65539780e-02 8.25908840e-01 -2.04628482e-02
4.58506346e-01 -9.98350918e-01 3.47167790e-01 5.54552913e-01
-1.48625091e-01 2.69700475e-02 2.21859798e-01 3.80062282e-01
-5.54878592e-01 -1.69283137e-01 1.21263194e+00 1.27449796e-01
-6.65199280e-01 7.59611964e-01 3.23961437e-01 8.21423680e-02
8.74464512e-01 -8.23078007e-02 -3.27433378e-01 -3.83287907e-01
-5.53326070e-01 -3.34922820e-02 2.85132140e-01 4.61339504e-01
1.05520236e+00 -1.64578533e+00 -1.09982777e+00 5.32645762e-01
1.39944941e-01 -6.74101472e-01 3.85787308e-01 3.00734222e-01
-4.47234064e-01 3.25158648e-02 -6.19084060e-01 -4.80380297e-01
-1.53852165e+00 6.33386433e-01 5.15160918e-01 -2.48379052e-01
9.20007005e-03 9.36924338e-01 -3.59707139e-02 -2.73903251e-01
3.33806127e-01 5.20183921e-01 -2.36424431e-01 -2.91365981e-01
1.03020632e+00 3.11275423e-01 3.44456464e-01 -1.09045160e+00
-3.74624789e-01 6.92394614e-01 7.27660134e-02 2.06234887e-01
1.15282393e+00 3.08823198e-01 -4.99771982e-02 -3.38309646e-01
1.17244077e+00 -1.40310898e-01 -1.36527264e+00 -2.09636599e-01
-9.60153639e-02 -1.04388034e+00 -2.27040425e-02 -1.04740393e+00
-1.72203803e+00 9.23303127e-01 1.26584923e+00 -1.40497699e-01
1.58990538e+00 -3.13224405e-01 5.44792712e-01 9.07906368e-02
5.50821483e-01 -6.50278270e-01 6.32323265e-01 3.91678736e-02
1.32190585e+00 -1.37907338e+00 -1.53613374e-01 -5.01340628e-01
-3.54952008e-01 1.15461218e+00 7.26224065e-01 2.87305992e-02
7.95306385e-01 1.18684895e-01 1.59725502e-01 1.01066809e-02
-4.87558454e-01 -4.39072400e-02 6.85513616e-01 6.85442686e-01
2.39532188e-01 -1.49783179e-01 3.20226997e-02 3.81154895e-01
4.24636938e-02 1.59944281e-01 -1.13845512e-01 6.79623365e-01
-1.37659788e-01 -1.43505895e+00 -6.13065183e-01 4.30168420e-01
-8.97719026e-01 1.45050243e-01 -2.66806573e-01 5.86647391e-01
1.80133954e-01 1.15275979e+00 -4.39555943e-01 -4.87309664e-01
4.78659928e-01 -1.74481969e-03 8.08001518e-01 -1.43342838e-01
-7.30342567e-01 -3.69974583e-01 -1.78888574e-01 -8.01004767e-01
-4.54179376e-01 -4.49755162e-01 -7.73563862e-01 -5.93811393e-01
-4.17275935e-01 1.12520643e-01 7.28595018e-01 6.92721486e-01
2.00368792e-01 2.94914812e-01 1.12534189e+00 -8.75291765e-01
-8.09469521e-01 -7.77850807e-01 -4.31734860e-01 8.21106732e-01
1.97509542e-01 -8.92103136e-01 -3.55899543e-01 4.86945137e-02]
|
[13.06754207611084, 0.6579708456993103]
|
32bf9aff-b813-4290-b430-704488430398
|
sasmu-boost-the-performance-of-generalized
|
2306.01449
| null |
https://arxiv.org/abs/2306.01449v1
|
https://arxiv.org/pdf/2306.01449v1.pdf
|
SASMU: boost the performance of generalized recognition model using synthetic face dataset
|
Nowadays, deploying a robust face recognition product becomes easy with the development of face recognition techniques for decades. Not only profile image verification but also the state-of-the-art method can handle the in-the-wild image almost perfectly. However, the concern of privacy issues raise rapidly since mainstream research results are powered by tons of web-crawled data, which faces the privacy invasion issue. The community tries to escape this predicament completely by training the face recognition model with synthetic data but faces severe domain gap issues, which still need to access real images and identity labels to fine-tune the model. In this paper, we propose SASMU, a simple, novel, and effective method for face recognition using a synthetic dataset. Our proposed method consists of spatial data augmentation (SA) and spectrum mixup (SMU). We first analyze the existing synthetic datasets for developing a face recognition system. Then, we reveal that heavy data augmentation is helpful for boosting performance when using synthetic data. By analyzing the previous frequency mixup studies, we proposed a novel method for domain generalization. Extensive experimental results have demonstrated the effectiveness of SASMU, achieving state-of-the-art performance on several common benchmarks, such as LFW, AgeDB-30, CA-LFW, CFP-FP, and CP-LFW.
|
['Chinson Yeh', 'Haoyuan He', 'Yong-Sheng Chen', 'Pei-Chun Chang', 'Chia-Chun Chung']
|
2023-06-02
| null | null | null | null |
['robust-face-recognition', 'face-recognition', 'domain-generalization']
|
['computer-vision', 'computer-vision', 'methodology']
|
[ 3.31598133e-01 -1.94676429e-01 -1.63278908e-01 -7.07133293e-01
-6.35993898e-01 -4.29057956e-01 5.35409868e-01 -6.49595797e-01
-1.24845974e-01 7.95540810e-01 -2.39791870e-01 -2.39445150e-01
-1.61175460e-01 -5.97364545e-01 -5.26292086e-01 -8.00027490e-01
1.53308451e-01 5.04722260e-02 -2.20229417e-01 -2.56683290e-01
3.54152322e-02 4.35699791e-01 -1.77466583e+00 4.73583132e-01
9.20663357e-01 1.50690663e+00 -2.79872686e-01 -3.77240628e-02
-2.77491231e-02 3.14794481e-01 -6.39965832e-01 -9.92380083e-01
7.58656383e-01 -2.72450149e-01 -6.39725268e-01 6.62200227e-02
5.74177682e-01 -4.34192032e-01 -3.10252011e-01 1.29123187e+00
7.16229022e-01 -1.99029431e-01 3.46986234e-01 -1.81294847e+00
-7.67610550e-01 2.63900131e-01 -8.37945879e-01 -7.32159540e-02
3.68245959e-01 8.45780522e-02 2.22433850e-01 -1.03064895e+00
6.59204125e-01 1.32547164e+00 8.19749415e-01 8.90165627e-01
-1.17931008e+00 -1.48009551e+00 -1.26912892e-01 4.56247389e-01
-1.76674759e+00 -8.25771749e-01 7.94476569e-01 -2.63904601e-01
2.77446717e-01 3.35111529e-01 1.04268067e-01 1.55290663e+00
-4.82570499e-01 4.65407580e-01 1.48594213e+00 -4.74714607e-01
5.03354408e-02 5.08091688e-01 6.92337900e-02 4.06884700e-01
3.18289995e-01 2.08996445e-01 -8.11341703e-01 -3.02134365e-01
3.11719418e-01 -3.79971117e-02 -4.82010216e-01 -2.75337785e-01
-6.09959066e-01 4.84006137e-01 -4.03869152e-02 1.71310548e-03
-5.12407720e-02 -5.59785306e-01 4.60330874e-01 4.65627760e-01
5.21519065e-01 -7.06528947e-02 -3.36287946e-01 1.17645554e-01
-1.04187977e+00 1.27480507e-01 6.37324810e-01 8.90359700e-01
4.84068453e-01 -5.74277937e-02 -4.31228280e-02 1.05141068e+00
2.14944884e-01 7.99491286e-01 6.23951435e-01 -5.77910304e-01
3.80407274e-01 5.40361226e-01 -3.64070944e-02 -1.02973068e+00
-2.43717432e-02 -2.51681179e-01 -1.06125176e+00 5.81648722e-02
5.56661189e-01 -6.51432723e-02 -9.06360686e-01 1.79781783e+00
4.37091857e-01 3.89053673e-01 8.72504190e-02 6.12813294e-01
6.26741588e-01 4.34721887e-01 3.18671949e-02 -2.88213760e-01
1.37052631e+00 -6.46326005e-01 -8.04972112e-01 7.96946287e-02
1.64982244e-01 -7.12788463e-01 1.12194371e+00 4.85899687e-01
-4.64487523e-01 -6.78077459e-01 -1.06585288e+00 3.29309702e-01
-5.20146906e-01 2.45642662e-01 6.03227615e-01 1.54605091e+00
-9.90591645e-01 4.27747160e-01 -3.06441844e-01 -4.49163228e-01
1.10833311e+00 5.24941027e-01 -9.20996070e-01 -6.49507999e-01
-1.22876537e+00 3.53802741e-01 2.50257313e-01 3.33382227e-02
-7.52726436e-01 -9.17431712e-01 -6.14979863e-01 -1.55163512e-01
3.98344636e-01 -7.66900629e-02 8.84667635e-01 -1.12404895e+00
-1.39015341e+00 1.05550587e+00 -3.21376249e-02 -3.41278791e-01
5.26933849e-01 4.92695831e-02 -1.21081579e+00 -6.60103327e-03
-4.38428260e-02 4.87369508e-01 1.19156611e+00 -1.20083892e+00
-4.83869433e-01 -8.09150338e-01 -3.54959399e-01 -4.27240044e-01
-8.52845013e-01 2.42917046e-01 -2.50885397e-01 -6.07897401e-01
-1.35817468e-01 -6.89818084e-01 3.41255009e-01 1.38779134e-01
-3.77097815e-01 3.06028817e-02 1.34265029e+00 -8.45555961e-01
1.16991079e+00 -2.62382627e+00 -5.00155509e-01 4.47694868e-01
-1.78785264e-01 7.50934958e-01 -2.91605920e-01 1.29819393e-01
-4.18876767e-01 1.76326618e-01 -1.87445551e-01 -3.86451662e-01
1.96885150e-02 -6.86685294e-02 -6.20574057e-01 4.58862871e-01
2.48827413e-01 4.62327361e-01 -4.54346240e-01 -3.34843189e-01
-1.37191400e-01 6.03209555e-01 -5.95905066e-01 1.57261536e-01
6.25435859e-02 3.66467625e-01 -1.49925306e-01 1.05605245e+00
1.58193815e+00 -7.21984729e-02 3.60074341e-01 -2.79443353e-01
1.55208021e-01 -2.65456259e-01 -1.35483873e+00 1.31219077e+00
-1.03254333e-01 3.84765595e-01 1.42681256e-01 -8.60939980e-01
9.81762946e-01 2.62854636e-01 3.32250923e-01 -7.34892190e-01
1.99530706e-01 3.70292783e-01 -1.37631685e-01 -4.93426561e-01
1.17401846e-01 9.66990665e-02 2.45748386e-01 2.58356482e-01
1.95432454e-01 5.27336717e-01 -5.02767898e-02 -1.91227064e-01
7.13568985e-01 3.36994231e-02 -5.84133156e-02 -3.00571948e-01
7.46884823e-01 -3.57025743e-01 7.24973559e-01 3.76157850e-01
-4.03918713e-01 4.39102083e-01 3.85596007e-01 -2.93092132e-01
-6.12985671e-01 -1.00884295e+00 -4.71540213e-01 9.74943042e-01
4.72930782e-02 -3.88123840e-01 -1.08688247e+00 -9.03705299e-01
1.97206646e-01 3.66615534e-01 -5.74800193e-01 -2.99227536e-01
-1.62913516e-01 -9.66943145e-01 9.47349489e-01 2.21870646e-01
1.29006338e+00 -6.31430387e-01 7.22638518e-02 -3.55290174e-01
-1.45645544e-01 -1.27730286e+00 -5.30475199e-01 -4.97965604e-01
-3.55102569e-01 -1.09681559e+00 -7.45919585e-01 -6.65883183e-01
8.28609765e-01 3.11808258e-01 5.31455457e-01 2.84826867e-02
-5.81275761e-01 1.83572084e-01 -2.32580349e-01 -4.86652732e-01
-1.52900606e-01 -5.72999679e-02 4.20745343e-01 7.32007265e-01
7.30251312e-01 -5.71009815e-01 -6.36710584e-01 7.26844192e-01
-9.33223784e-01 -3.98336709e-01 3.91852170e-01 1.04024553e+00
2.77415484e-01 1.22100167e-01 9.56903875e-01 -9.19601917e-01
5.23169637e-01 -3.62354785e-01 -5.93282223e-01 4.05007660e-01
-7.56948769e-01 -2.84328312e-01 5.20966470e-01 -6.30824447e-01
-1.35745025e+00 1.28327915e-02 -1.46453157e-01 -4.46716934e-01
-1.31551877e-01 4.37952094e-02 -6.94692671e-01 -5.58350146e-01
7.69700468e-01 3.43438029e-01 3.95328075e-01 -4.96189266e-01
2.06048831e-01 1.25143719e+00 6.14810586e-01 -5.18072724e-01
1.01175427e+00 5.69494128e-01 -7.83058107e-02 -1.05508065e+00
-4.86609459e-01 -2.33807027e-01 -3.26690674e-01 -1.70469165e-01
3.68933141e-01 -1.03653777e+00 -1.04761732e+00 1.03395784e+00
-8.92063379e-01 2.49459758e-01 1.40572321e-02 1.75625339e-01
-4.09845337e-02 5.10809600e-01 -2.63659835e-01 -9.23386693e-01
-2.15738371e-01 -1.05920708e+00 8.71336222e-01 3.78073961e-01
1.81536272e-01 -4.35638458e-01 -2.52078801e-01 7.29937911e-01
7.65187979e-01 2.45823786e-01 5.78078568e-01 -6.93333566e-01
-4.94091183e-01 -3.18047881e-01 -4.76483196e-01 7.38462090e-01
2.97957242e-01 -2.10959256e-01 -1.56511867e+00 -4.46368933e-01
8.18803683e-02 -3.98200154e-01 6.28203571e-01 -1.26528010e-01
1.61434436e+00 -4.88962799e-01 -4.21780407e-01 8.83490920e-01
1.15691209e+00 1.86393678e-01 9.00526941e-01 2.08657887e-02
3.18773806e-01 6.43748403e-01 6.33011997e-01 5.53176522e-01
1.44292146e-01 7.55127430e-01 1.75070733e-01 6.10275269e-02
-5.76108769e-02 -2.96668857e-01 3.98148537e-01 9.87970531e-02
3.94160487e-03 7.07990825e-02 -7.80636728e-01 2.26972982e-01
-1.41127288e+00 -1.09744346e+00 3.65033001e-01 2.34590054e+00
9.77696419e-01 -2.36900687e-01 1.20890327e-01 4.29393977e-01
6.53315425e-01 -2.80804131e-02 -4.57873017e-01 2.35634279e-02
-2.42311686e-01 4.99826789e-01 5.10281384e-01 -4.13378254e-02
-1.24588168e+00 7.74529040e-01 5.51505947e+00 1.17245102e+00
-1.44071913e+00 1.95796341e-01 1.00297570e+00 -2.73766629e-02
1.52056903e-01 -4.90793198e-01 -1.04717910e+00 6.61312401e-01
8.38539898e-01 -1.69010103e-01 6.38315797e-01 1.22367096e+00
-2.79231220e-01 1.85278073e-01 -1.02245080e+00 1.55703211e+00
3.94797385e-01 -1.26146448e+00 7.42863268e-02 1.92652553e-01
6.92175806e-01 -3.56500685e-01 5.68405688e-01 3.32456976e-01
-1.80091441e-01 -1.23639631e+00 3.94573450e-01 1.68237820e-01
1.27593482e+00 -7.00244486e-01 7.04120457e-01 1.46503240e-01
-9.28257048e-01 -2.73331583e-01 -4.75500256e-01 2.62523919e-01
-3.97917509e-01 5.91943800e-01 -6.58058584e-01 7.52195776e-01
8.61523688e-01 3.57187688e-01 -6.89375162e-01 8.03371072e-01
2.36626074e-01 4.84253496e-01 -3.57694238e-01 2.90098011e-01
-4.01154399e-01 -6.42974600e-02 6.33474812e-02 8.60617399e-01
6.08052552e-01 1.13203764e-01 -8.15599486e-02 6.82366252e-01
-4.73847061e-01 2.41403118e-01 -7.34084725e-01 -2.18647987e-01
6.88946545e-01 1.17573357e+00 -2.44215250e-01 1.35797456e-01
-3.63895506e-01 8.98104966e-01 -4.74771857e-02 3.28497916e-01
-8.72709274e-01 -3.41734469e-01 1.03542590e+00 2.79685318e-01
1.10514984e-01 5.58002181e-02 -4.54440936e-02 -1.06782138e+00
3.19520414e-01 -1.37563109e+00 4.98525053e-01 -3.49172533e-01
-1.36061668e+00 8.39431643e-01 -1.28086627e-01 -1.18318617e+00
-4.33774246e-03 -8.43438804e-01 -2.57990003e-01 8.67818654e-01
-1.60553372e+00 -1.32201397e+00 -4.33352828e-01 9.89091635e-01
1.14422590e-01 -7.29163706e-01 1.02200460e+00 9.18903530e-01
-8.36728930e-01 1.33797431e+00 1.09240294e-01 3.21135461e-01
9.79894578e-01 -3.65214318e-01 1.51942670e-01 7.24153757e-01
3.79688130e-03 7.71941483e-01 2.80683398e-01 -4.29411173e-01
-1.52880430e+00 -1.16396618e+00 5.83622098e-01 -2.92615712e-01
2.68206865e-01 -6.83393419e-01 -8.96322966e-01 4.43892986e-01
3.42673734e-02 3.39574724e-01 9.93902802e-01 -4.07718234e-02
-9.03994560e-01 -7.24023521e-01 -1.85538065e+00 3.81474912e-01
1.12544286e+00 -6.97847128e-01 -1.35707945e-01 3.83178771e-01
4.23121035e-01 -8.28398615e-02 -6.84377968e-01 5.67978323e-01
7.24176705e-01 -1.06824791e+00 1.04204595e+00 -6.08595312e-01
-2.86142738e-03 -2.80833244e-01 -4.75521863e-01 -1.01357269e+00
2.21495911e-01 -7.96439528e-01 1.38863504e-01 1.78617704e+00
3.06277245e-01 -9.56055343e-01 9.48557854e-01 6.29102468e-01
5.07647574e-01 -5.32996595e-01 -1.05973494e+00 -1.08781469e+00
-2.93328881e-01 -3.15704346e-01 1.11656082e+00 1.19932663e+00
-1.16356365e-01 -2.31759980e-01 -5.58613956e-01 3.37039143e-01
9.48218882e-01 -9.66891423e-02 8.97591174e-01 -1.20930970e+00
-2.57089850e-03 -1.30104631e-01 -5.65783620e-01 -5.00886083e-01
2.71226496e-01 -8.31756353e-01 -3.95877361e-01 -4.91497576e-01
1.49511218e-01 -2.21263722e-01 -1.86007872e-01 5.30759335e-01
1.48882717e-01 5.58805645e-01 1.16812997e-01 -1.68699473e-01
-9.02203247e-02 5.27295172e-01 9.14553702e-01 -2.34176621e-01
1.82024181e-01 2.41600797e-02 -8.52217615e-01 4.67496216e-01
7.29781449e-01 -2.33447447e-01 -6.32381201e-01 -5.28741744e-04
-3.59404147e-01 -3.63078713e-01 4.04840082e-01 -1.21484697e+00
1.83271021e-01 -9.76905376e-02 6.01879299e-01 -2.23730937e-01
4.69336748e-01 -9.73105431e-01 2.51457334e-01 2.49424800e-01
7.02535221e-03 -3.28323543e-01 4.25533265e-01 3.53725225e-01
-3.04533631e-01 1.71397254e-01 1.04202306e+00 3.78956556e-01
-8.17485332e-01 5.76730430e-01 4.20719743e-01 1.61692668e-02
1.22248340e+00 -3.32066596e-01 -6.61158919e-01 -1.85062006e-01
-4.06895608e-01 1.35213494e-01 2.85765052e-01 5.51715314e-01
6.55101776e-01 -1.49512780e+00 -6.41533971e-01 9.72397327e-01
3.60047460e-01 -5.70141613e-01 5.24959981e-01 5.14991939e-01
-1.88641414e-01 1.37815908e-01 -5.30682266e-01 -4.88152385e-01
-1.56956196e+00 6.79841518e-01 2.31644690e-01 1.52108684e-01
-1.67525366e-01 9.33079720e-01 1.08730897e-01 -4.26753551e-01
5.20987630e-01 3.50112095e-02 1.38264611e-01 1.29409626e-01
1.01704824e+00 2.74416417e-01 1.46388069e-01 -6.50032163e-01
-5.42561054e-01 5.24378359e-01 -2.18064576e-01 9.30723324e-02
1.17031896e+00 5.69758937e-03 -1.66804492e-01 -2.89622456e-01
1.26588869e+00 -1.07438184e-01 -1.02832401e+00 -2.14244023e-01
-4.22592312e-02 -1.04188108e+00 -3.87749285e-01 -9.32795644e-01
-1.26906991e+00 8.34600031e-01 1.13463235e+00 1.29087716e-01
1.46743834e+00 -4.34512436e-01 7.13396907e-01 1.77456692e-01
6.77333593e-01 -1.11059248e+00 2.64826533e-03 7.38130808e-02
8.09524059e-01 -1.37773073e+00 -1.71260238e-01 -8.31453085e-01
-5.41904688e-01 7.67008603e-01 7.12160468e-01 5.69622278e-01
1.10468841e+00 1.89654887e-01 9.45008472e-02 2.32385412e-01
-3.47129583e-01 2.16046602e-01 1.58069193e-01 9.00816739e-01
5.87319545e-02 -6.92178980e-02 -4.95542623e-02 9.74664330e-01
-9.72378775e-02 3.13373208e-01 9.38952565e-02 7.06506670e-01
-5.66009320e-02 -1.57219863e+00 -5.76682687e-01 3.67636412e-01
-5.41599631e-01 1.44459322e-01 -3.97499889e-01 7.82149374e-01
3.34708333e-01 9.67896760e-01 -1.98480681e-01 -7.43934989e-01
4.28700835e-01 4.95349228e-01 4.19270784e-01 -2.06389397e-01
-3.24149758e-01 -5.16870677e-01 -8.15218315e-02 -6.16971135e-01
-3.00824046e-01 -6.61302805e-01 -6.75996006e-01 -6.39132261e-01
-1.69009298e-01 1.12567646e-02 7.95204937e-01 5.87939978e-01
6.90151751e-01 6.39542863e-02 1.02479422e+00 -3.73345435e-01
-9.99939978e-01 -8.57659519e-01 -7.81803727e-01 5.28585792e-01
2.74046451e-01 -8.01347613e-01 -2.05805525e-01 8.52643028e-02]
|
[13.065349578857422, 0.8576679825782776]
|
532e5bbc-3165-4646-8119-b82e8facdcb3
|
on-efficient-real-time-semantic-segmentation
|
2206.08605
| null |
https://arxiv.org/abs/2206.08605v2
|
https://arxiv.org/pdf/2206.08605v2.pdf
|
On Efficient Real-Time Semantic Segmentation: A Survey
|
Semantic segmentation is the problem of assigning a class label to every pixel in an image, and is an important component of an autonomous vehicle vision stack for facilitating scene understanding and object detection. However, many of the top performing semantic segmentation models are extremely complex and cumbersome, and as such are not suited to deployment onboard autonomous vehicle platforms where computational resources are limited and low-latency operation is a vital requirement. In this survey, we take a thorough look at the works that aim to address this misalignment with more compact and efficient models capable of deployment on low-memory embedded systems while meeting the constraint of real-time inference. We discuss several of the most prominent works in the field, placing them within a taxonomy based on their major contributions, and finally we evaluate the inference speed of the discussed models under consistent hardware and software setups that represent a typical research environment with high-end GPU and a realistic deployed scenario using low-memory embedded GPU hardware. Our experimental results demonstrate that many works are capable of real-time performance on resource-constrained hardware, while illustrating the consistent trade-off between latency and accuracy.
|
['Muhammad Shafique', 'Christopher J. Holder']
|
2022-06-17
| null | null | null | null |
['real-time-semantic-segmentation']
|
['computer-vision']
|
[ 3.68538290e-01 -7.77154565e-02 -3.11342269e-01 -4.63950962e-01
-2.98039138e-01 -6.50256276e-01 5.94753802e-01 5.33995070e-02
-6.03389680e-01 1.72161892e-01 -8.49580228e-01 -6.50237918e-01
1.88355863e-01 -9.64702427e-01 -7.64689445e-01 -5.16099155e-01
5.01322076e-02 7.84147263e-01 9.50072944e-01 -6.11593835e-02
3.91661704e-01 6.21770263e-01 -2.52320719e+00 -2.02916935e-02
6.30155981e-01 1.29755890e+00 3.92874479e-01 8.60036671e-01
-6.84306175e-02 3.74302179e-01 -4.23846394e-01 -2.36559883e-01
3.24641317e-01 1.97784662e-01 -5.48187613e-01 2.55886763e-01
8.63386512e-01 -3.64558518e-01 -4.93579619e-02 1.20218897e+00
1.07816808e-01 -1.86211228e-01 4.13735867e-01 -1.68261337e+00
4.08637822e-01 -2.49408633e-01 -4.72243845e-01 4.19052869e-01
-5.87404184e-02 2.81281233e-01 5.79835773e-01 -7.34928668e-01
4.31659013e-01 1.05579662e+00 8.01649928e-01 2.19158813e-01
-7.75493562e-01 -6.58537626e-01 1.85467526e-01 5.62964678e-01
-1.12579644e+00 -5.41832149e-01 3.35192293e-01 -3.92291427e-01
1.21140790e+00 3.73461902e-01 8.00746739e-01 6.01672649e-01
3.79067302e-01 7.65583813e-01 8.01319063e-01 -5.29567413e-02
6.27065539e-01 1.73158303e-01 6.56266809e-01 8.21382761e-01
6.81036770e-01 -6.00701608e-02 -5.00695586e-01 -1.89040944e-01
3.21795106e-01 -1.98855370e-01 3.49652976e-01 -3.66893858e-01
-9.28583324e-01 8.91512454e-01 1.94774404e-01 -1.63970008e-01
-1.12958580e-01 4.05477464e-01 7.33052194e-01 -3.02984770e-02
2.80425906e-01 -6.74716756e-02 -3.78981590e-01 -1.84972301e-01
-1.15761209e+00 3.35126162e-01 8.42939019e-01 1.27836096e+00
1.04339373e+00 1.22097870e-02 1.84632942e-01 3.21001023e-01
4.54783469e-01 6.98176920e-01 1.90796673e-01 -1.14573324e+00
4.75284196e-02 4.61448342e-01 -9.06275511e-02 -9.00721908e-01
-5.93118310e-01 -4.17793006e-01 -3.92782539e-01 7.10512221e-01
1.43555403e-01 -1.39973722e-02 -9.85814631e-01 9.95818675e-01
6.04768813e-01 5.25346875e-01 -7.50341490e-02 7.99091160e-01
9.38959241e-01 4.91593003e-01 4.33981061e-01 1.33555844e-01
1.97348690e+00 -1.19712412e+00 -4.16276395e-01 -8.63631725e-01
6.03676379e-01 -8.04563284e-01 8.48357797e-01 2.91998684e-01
-7.01904118e-01 -6.53907597e-01 -1.56748164e+00 -3.31741333e-01
-5.93941271e-01 1.46238476e-01 1.19721675e+00 1.05953288e+00
-1.14124560e+00 3.01291823e-01 -1.39995193e+00 -6.03439510e-01
4.12161946e-01 3.56237322e-01 6.50722682e-02 1.04957238e-01
-5.96046388e-01 8.05128574e-01 3.63859951e-01 9.39404145e-02
-6.70506477e-01 -6.50673687e-01 -6.73286617e-01 -2.44202554e-01
4.24746037e-01 -6.58026576e-01 1.35879207e+00 -8.46053302e-01
-1.37237358e+00 1.11923087e+00 -3.00138265e-01 -7.45777309e-01
5.70043266e-01 -1.42863229e-01 -2.93551236e-01 9.60877910e-02
1.18134782e-01 8.85516644e-01 7.01821327e-01 -1.17879736e+00
-1.21130371e+00 -5.14283657e-01 1.22560203e-01 1.76740751e-01
-6.45911961e-04 3.49439457e-02 -8.42141747e-01 1.28862917e-01
4.80591133e-02 -1.22719228e+00 -3.40873539e-01 1.15585648e-01
-2.73807179e-02 -1.69819608e-01 1.51935220e+00 -2.63307184e-01
7.54579067e-01 -2.13277578e+00 -5.75202823e-01 6.64097220e-02
1.55533955e-01 7.33233511e-01 4.34374839e-01 1.30716786e-01
5.35720408e-01 -2.50775725e-01 -2.62851924e-01 -3.76235813e-01
4.34286371e-02 3.98308039e-01 -3.24035406e-01 6.20979249e-01
5.77742755e-02 7.34487891e-01 -7.74051011e-01 -6.73215866e-01
6.48487151e-01 5.52240312e-01 -4.19992715e-01 2.52667628e-03
-4.40053582e-01 1.65869951e-01 -5.37547052e-01 6.15856588e-01
8.78385007e-01 -1.00706331e-01 1.40483275e-01 -1.67091087e-01
-4.88518983e-01 -2.04305016e-02 -1.30215585e+00 1.35241222e+00
-3.40002239e-01 1.07812560e+00 3.07555616e-01 -9.12373841e-01
7.57844448e-01 -2.44265914e-01 2.19479576e-01 -8.25612903e-01
4.55376059e-01 3.00265372e-01 -1.86156407e-01 -3.70190769e-01
1.03609395e+00 5.06871998e-01 1.67897213e-02 2.08836943e-02
-2.87899107e-01 -2.90796071e-01 4.59485620e-01 6.78023323e-02
8.66391182e-01 1.46873251e-01 2.25014523e-01 -5.29287219e-01
4.35456455e-01 6.12995267e-01 3.66979003e-01 7.79240251e-01
-4.34832871e-01 2.68064171e-01 3.15776348e-01 -4.44299787e-01
-1.01732445e+00 -6.98075950e-01 -5.40359616e-01 1.15011072e+00
6.34786844e-01 -4.11862373e-01 -1.05520892e+00 -1.13501690e-01
-9.06747431e-02 5.94225824e-01 -2.68805534e-01 2.60038257e-01
-5.75803101e-01 -9.71751273e-01 4.67597812e-01 5.73166192e-01
8.64457309e-01 -6.82052493e-01 -1.77187705e+00 1.38464898e-01
3.13166678e-01 -1.53269422e+00 3.87583882e-01 5.12897447e-02
-8.48519981e-01 -1.12898684e+00 -4.18853771e-04 -1.01282597e+00
5.36726356e-01 7.97262073e-01 1.29588401e+00 4.80977505e-01
-6.06683016e-01 2.89784610e-01 -1.32097006e-01 -7.12436676e-01
-5.89000359e-02 -2.77354196e-02 -3.08599740e-01 -1.34259582e-01
6.34820819e-01 -2.18188360e-01 -7.45340347e-01 3.57530832e-01
-7.73962975e-01 3.10636789e-01 4.05656099e-01 4.24975723e-01
7.82206953e-01 3.81430984e-02 6.96781697e-03 -9.07906353e-01
2.02648547e-02 -4.31951970e-01 -1.27380753e+00 -7.35479519e-02
-6.18145823e-01 -3.31067085e-01 4.99785572e-01 9.58421007e-02
-8.69934082e-01 4.17443901e-01 -3.95465225e-01 8.31695423e-02
-3.01582992e-01 5.47190756e-03 -4.51769419e-02 -3.81088346e-01
2.94087470e-01 1.18430257e-01 6.68167695e-02 -9.66430679e-02
3.36090684e-01 7.88530469e-01 7.52589107e-01 -3.87606144e-01
3.69473606e-01 1.02900064e+00 2.59093076e-01 -1.20899749e+00
-5.95313311e-01 -7.56379247e-01 -5.58475435e-01 -3.88860136e-01
8.75025451e-01 -9.84997094e-01 -8.79284859e-01 6.82198763e-01
-1.05477810e+00 -4.36277121e-01 1.16575733e-01 2.44767860e-01
-7.62346804e-01 3.60569477e-01 -2.98717141e-01 -6.50780857e-01
-3.74159634e-01 -1.74810970e+00 1.49514961e+00 5.45633197e-01
8.08721557e-02 -8.23186219e-01 -3.91192973e-01 5.63010514e-01
5.08363187e-01 1.70291767e-01 6.12248003e-01 -3.10389280e-01
-8.91912997e-01 -2.66516775e-01 -6.87361658e-01 -3.89615372e-02
-4.94554192e-01 3.89503181e-01 -1.17132246e+00 -2.71171033e-01
-6.47712685e-03 -4.60944958e-02 6.63320363e-01 3.88434619e-01
1.19277239e+00 3.54019314e-01 -7.42134631e-01 7.95286179e-01
1.69112730e+00 7.66372010e-02 4.35733646e-01 5.41592300e-01
6.08630359e-01 5.14387846e-01 1.06945908e+00 2.59870857e-01
5.50507724e-01 9.28341091e-01 7.23289132e-01 -6.14846796e-02
-3.85159492e-01 2.42289871e-01 1.70991682e-02 5.83005011e-01
1.14711642e-01 -2.65224427e-01 -1.22885823e+00 4.60202128e-01
-2.07107019e+00 -5.54004669e-01 -8.54360104e-01 2.14001465e+00
1.43313706e-01 4.35642153e-01 8.01942647e-02 3.01747173e-02
5.80388129e-01 1.63827999e-03 -6.25029147e-01 -4.52762216e-01
1.51007175e-01 4.16137353e-02 1.11129820e+00 5.09009719e-01
-1.26805389e+00 1.11791658e+00 6.68486977e+00 8.96025538e-01
-1.29408813e+00 2.73536950e-01 4.97868866e-01 1.25798166e-01
2.59830594e-01 7.34416544e-02 -1.17494178e+00 5.28170109e-01
1.37791586e+00 -3.46031077e-02 -2.71969642e-02 1.34295440e+00
1.45454183e-01 -7.22145140e-01 -9.78503525e-01 1.04553783e+00
1.38856351e-01 -1.32893419e+00 -4.07530755e-01 5.46759963e-02
4.88359094e-01 6.03417158e-01 -8.65367055e-02 2.18104869e-02
1.16393037e-01 -8.07045341e-01 9.73209381e-01 -1.01240352e-02
6.18097067e-01 -7.33177483e-01 9.17539358e-01 3.74666870e-01
-1.28850245e+00 2.03809455e-01 -4.49707508e-01 -2.20006406e-01
1.83507279e-01 4.56360072e-01 -7.66829669e-01 1.72027126e-01
1.04122543e+00 4.23879802e-01 -7.38527656e-01 1.13441777e+00
2.40046009e-02 5.76063395e-01 -5.62307239e-01 -1.79489344e-01
4.01261181e-01 -1.66180655e-01 2.59226084e-01 1.55104423e+00
1.17206872e-01 -7.65295103e-02 4.88235712e-01 2.71456420e-01
4.52636778e-01 -1.93204895e-01 -3.74966025e-01 4.81568754e-01
5.26219904e-01 1.45967686e+00 -1.62875438e+00 -6.17928803e-01
-5.81912816e-01 7.19764411e-01 2.92180683e-02 -1.22835882e-01
-1.27453995e+00 -2.45360747e-01 1.07043552e+00 1.80415493e-02
2.82753825e-01 -6.58558607e-01 -9.62407768e-01 -5.36118031e-01
-5.82982562e-02 -4.46716994e-01 1.22524738e-01 -5.06633580e-01
-5.48438847e-01 4.92889196e-01 1.58297345e-02 -1.09573543e+00
-6.04071952e-02 -9.73262966e-01 -2.05279931e-01 4.33271050e-01
-1.84615219e+00 -1.09139204e+00 -7.96636283e-01 3.59258920e-01
8.75814199e-01 -1.37640322e-02 6.96681440e-01 5.41906536e-01
-5.81679642e-01 3.81033361e-01 2.88559850e-02 -3.02464545e-01
7.42752478e-02 -8.34843993e-01 7.73387372e-01 1.01407814e+00
-6.73207343e-02 3.75652343e-01 9.37245548e-01 -5.20779848e-01
-1.94326091e+00 -1.24867618e+00 3.62800807e-01 -3.80600512e-01
4.80201423e-01 -5.98399520e-01 -7.70251274e-01 5.67286253e-01
7.26770833e-02 1.85899138e-01 3.76592726e-01 -1.72578126e-01
-8.89133513e-02 2.30513215e-02 -1.12721384e+00 5.79439998e-01
1.08130455e+00 -2.19886705e-01 -2.10390016e-01 5.82607031e-01
4.32118773e-01 -8.49088252e-01 -2.84996301e-01 3.57672125e-01
5.33644199e-01 -1.19551027e+00 8.61554027e-01 -4.03219052e-02
-7.95069262e-02 -6.77424073e-01 -9.41512287e-02 -4.82505113e-01
-3.08609977e-02 -3.62400889e-01 5.66351674e-02 8.42063785e-01
1.21190771e-01 -7.71026492e-01 1.12609208e+00 4.83158767e-01
-5.08074760e-01 -4.51199532e-01 -9.95650113e-01 -7.73510873e-01
-5.08545697e-01 -9.42378044e-01 3.99795055e-01 3.33414227e-01
-5.71164131e-01 2.39100933e-01 2.07808092e-01 4.53289807e-01
8.05548370e-01 1.40882447e-01 1.03814590e+00 -1.15609288e+00
1.54668540e-01 -4.72151965e-01 -1.08479559e+00 -1.09210706e+00
8.77650306e-02 -5.22671878e-01 3.65038931e-01 -1.50177240e+00
1.29032791e-01 -6.07854009e-01 4.35627133e-01 1.82322830e-01
1.36478871e-01 7.82527924e-01 1.27539774e-02 2.02609435e-01
-8.22553217e-01 1.41487503e-02 6.10310316e-01 -1.35717511e-01
2.33879939e-01 1.64167769e-02 -3.16147357e-01 1.16308975e+00
6.41886532e-01 -4.21256453e-01 -6.14151418e-01 -7.37134576e-01
1.53509423e-01 -4.44753855e-01 5.41074514e-01 -1.42704821e+00
6.74095452e-01 -5.69223017e-02 -1.84707716e-01 -7.72887886e-01
5.79631507e-01 -1.01696789e+00 2.41654456e-01 5.99413276e-01
4.66914266e-01 1.06436662e-01 4.46868420e-01 4.50579554e-01
-1.32595897e-01 -2.75299728e-01 1.00528181e+00 3.30126029e-03
-1.72566760e+00 1.97674796e-01 -5.09979844e-01 -1.67532176e-01
1.62743044e+00 -8.41600597e-01 -4.63851660e-01 2.80322701e-01
-1.22205049e-01 2.01894999e-01 6.70042455e-01 5.01237214e-01
2.76562065e-01 -7.70638943e-01 -4.56218272e-01 2.80078322e-01
1.42782450e-01 1.30138025e-01 2.49154463e-01 6.06718481e-01
-1.35644734e+00 7.18068302e-01 -3.51840198e-01 -1.31968808e+00
-1.48788846e+00 3.03738236e-01 6.44150153e-02 2.91595548e-01
-8.29288960e-01 8.10086787e-01 1.55924484e-01 -8.41892511e-02
2.40323514e-01 -2.26030275e-01 -5.59135079e-02 -4.85466346e-02
6.17707670e-01 5.11402309e-01 8.15222383e-01 -7.67988622e-01
-5.38395643e-01 6.73805654e-01 1.64773613e-01 2.49376953e-01
8.60211134e-01 -2.24597231e-01 -8.47008973e-02 1.66472197e-01
9.33294833e-01 -4.02068794e-01 -1.47151887e+00 1.00422196e-01
1.86967812e-02 -3.63874644e-01 4.48928088e-01 -1.90268129e-01
-1.09150147e+00 7.50086069e-01 8.56686950e-01 1.54987693e-01
1.02809596e+00 -8.63069668e-02 9.31196570e-01 2.73373753e-01
9.78529751e-01 -1.33142769e+00 -4.53109324e-01 6.69405580e-01
1.83570653e-01 -1.31168783e+00 2.73697793e-01 -1.16148758e+00
-2.40833476e-01 1.05294228e+00 7.19678938e-01 -2.71099895e-01
7.05147743e-01 7.15938807e-01 1.45710379e-01 -3.54615241e-01
-4.32153195e-01 -3.78221810e-01 -1.55276000e-01 7.84691274e-01
1.20994272e-02 2.07306489e-01 -5.07922769e-01 -5.21469042e-02
-5.26997328e-01 -2.62596279e-01 4.51742828e-01 1.09733021e+00
-7.12142766e-01 -7.27193654e-01 -3.88909876e-01 3.91964287e-01
-2.23943278e-01 2.53938884e-01 4.36865687e-02 6.48073614e-01
3.65486324e-01 1.27664471e+00 5.10123074e-01 -2.24291414e-01
2.12750226e-01 -2.56479174e-01 2.41918102e-01 -4.97857809e-01
-4.54864442e-01 -4.42125201e-01 2.26126984e-01 -8.66054773e-01
-4.12692457e-01 -6.09869957e-01 -1.40163958e+00 -6.05051041e-01
-1.53646857e-01 -2.44222447e-01 1.42188108e+00 9.73727345e-01
6.20721579e-01 5.71604133e-01 1.33566782e-01 -1.22710991e+00
1.77013744e-02 -4.10663158e-01 -3.67130339e-01 4.25344817e-02
1.56683177e-01 -8.91952157e-01 1.88443251e-02 1.54716074e-01]
|
[8.314438819885254, -1.2517366409301758]
|
82b20b94-29ad-47ab-815a-ca4b7186369f
|
deep-attention-q-network-for-personalized
|
2307.01519
| null |
https://arxiv.org/abs/2307.01519v1
|
https://arxiv.org/pdf/2307.01519v1.pdf
|
Deep Attention Q-Network for Personalized Treatment Recommendation
|
Tailoring treatment for individual patients is crucial yet challenging in order to achieve optimal healthcare outcomes. Recent advances in reinforcement learning offer promising personalized treatment recommendations; however, they rely solely on current patient observations (vital signs, demographics) as the patient's state, which may not accurately represent the true health status of the patient. This limitation hampers policy learning and evaluation, ultimately limiting treatment effectiveness. In this study, we propose the Deep Attention Q-Network for personalized treatment recommendations, utilizing the Transformer architecture within a deep reinforcement learning framework to efficiently incorporate all past patient observations. We evaluated the model on real-world sepsis and acute hypotension cohorts, demonstrating its superiority to state-of-the-art models. The source code for our model is available at https://github.com/stevenmsm/RL-ICU-DAQN.
|
['Shihao Yang', 'Nicoleta Serban', 'Junghwan Lee', 'Simin Ma']
|
2023-07-04
| null | null | null | null |
['deep-attention', 'deep-attention']
|
['computer-vision', 'natural-language-processing']
|
[-4.26429249e-02 -9.57410634e-02 -5.45804560e-01 -3.06785643e-01
-5.46814561e-01 -1.79795459e-01 -1.58200800e-01 4.87355560e-01
-3.70450884e-01 1.06923282e+00 5.62546194e-01 -5.99658787e-01
-3.95090580e-01 -5.94797015e-01 -3.88916731e-01 -6.01989806e-01
-4.63214628e-02 7.89434195e-01 -3.81459266e-01 -1.66654214e-01
1.11807726e-01 2.80259877e-01 -8.33552361e-01 3.12024802e-01
9.86287832e-01 7.87832379e-01 8.28100890e-02 7.54154265e-01
1.66490734e-01 1.03276813e+00 -2.44356245e-01 -2.28942987e-02
1.25635996e-01 -6.00930572e-01 -6.97156131e-01 -1.70497403e-01
-5.44450954e-02 -9.98004496e-01 -8.03688943e-01 5.65497041e-01
8.01719069e-01 2.32667655e-01 4.27469462e-01 -8.99063349e-01
-5.58351815e-01 6.16671622e-01 9.79041383e-02 3.45581800e-01
8.14793631e-02 7.35778213e-01 8.13596785e-01 -1.69399217e-01
5.99683933e-02 8.83842289e-01 4.53569889e-01 9.96200800e-01
-1.15564847e+00 -4.98799890e-01 2.68174678e-01 3.54474992e-01
-8.05765033e-01 -2.85767347e-01 3.83529961e-01 -3.65569681e-01
9.51680541e-01 -1.74173325e-01 1.00430632e+00 1.44621778e+00
5.56680202e-01 6.54818952e-01 8.35677564e-01 2.36672506e-01
3.60226870e-01 -3.38582695e-01 -7.74533376e-02 4.76639509e-01
-6.11744933e-02 4.74908352e-01 -2.40831688e-01 -3.22220683e-01
1.02649534e+00 8.16172063e-01 -5.21663904e-01 -4.61266279e-01
-1.05159807e+00 1.01502609e+00 4.49590683e-01 -1.12035245e-01
-1.14795494e+00 4.29547906e-01 4.76817816e-01 1.25274554e-01
8.96226838e-02 6.25762343e-01 -9.10728991e-01 -5.84032178e-01
-3.98704171e-01 3.00637960e-01 5.15407205e-01 4.15883482e-01
2.04858139e-01 2.21927553e-01 -6.36669517e-01 6.06455863e-01
2.73613520e-02 6.61855936e-01 6.22689009e-01 -1.30412543e+00
1.59532234e-01 3.23393524e-01 5.38645089e-01 -4.28522080e-01
-8.10602665e-01 -8.40790749e-01 -7.13213384e-01 -2.09345877e-01
3.09716791e-01 -7.11085677e-01 -8.36208105e-01 1.90006518e+00
1.25865892e-01 4.71179068e-01 2.16401681e-01 1.03222537e+00
4.70248431e-01 3.64762694e-01 4.73094136e-01 -2.93756783e-01
1.11720490e+00 -8.01944315e-01 -6.93321705e-01 -2.30544865e-01
6.21074736e-01 -2.14531079e-01 8.69035840e-01 5.90584397e-01
-1.12045062e+00 -1.92601487e-01 -4.95679110e-01 5.07923305e-01
1.58265561e-01 -4.84941490e-02 6.00102425e-01 8.53403136e-02
-8.28193665e-01 1.06259882e+00 -1.19654202e+00 -4.12667066e-01
6.18194818e-01 5.50830901e-01 1.69633493e-01 -1.95540413e-01
-1.51693153e+00 8.75143349e-01 2.66121507e-01 3.13699506e-02
-1.38933170e+00 -9.89128649e-01 -5.06232262e-01 4.82867301e-01
6.10425532e-01 -1.27677906e+00 1.87887001e+00 -3.19862068e-01
-1.83569217e+00 -4.21204753e-02 -5.81536582e-03 -7.04571128e-01
5.06650507e-01 -5.33034325e-01 -2.51271576e-01 3.15772921e-01
-4.63283002e-01 3.44506502e-01 6.24043286e-01 -8.20670724e-01
-6.99762762e-01 -2.34228238e-01 2.17031151e-01 4.48241651e-01
-1.13566399e-01 -5.24518549e-01 1.86514348e-01 -3.63198519e-01
-3.39586645e-01 -9.11413431e-01 -6.66715205e-01 -2.39263892e-01
-4.25989479e-02 9.58925039e-02 3.88588607e-01 -5.97034574e-01
1.20083952e+00 -1.91708338e+00 3.68526876e-02 -1.33583233e-01
4.11891848e-01 3.79830599e-01 -1.24287315e-01 8.41554582e-01
-8.69372413e-02 -1.69372201e-01 -1.63230836e-01 -5.91644123e-02
2.31998749e-02 3.93259019e-01 -3.61472547e-01 3.80212754e-01
1.78262502e-01 1.04452050e+00 -1.22554302e+00 -1.65294543e-01
6.01683259e-01 4.82823312e-01 -8.45664978e-01 6.85070753e-01
-3.94481212e-01 1.05458641e+00 -8.58693123e-01 2.54414856e-01
8.38637054e-02 -5.23122370e-01 2.72830933e-01 3.30675481e-04
2.72234768e-01 2.90815175e-01 -6.28687382e-01 1.39038086e+00
-5.54579735e-01 -7.47958347e-02 -1.50681779e-01 -1.14028394e+00
5.49780190e-01 6.67518377e-01 1.08031940e+00 -8.10940325e-01
2.89195806e-01 -9.44063738e-02 3.03950280e-01 -7.67384708e-01
-7.03090876e-02 -4.03702766e-01 2.84162551e-01 4.83417153e-01
-1.90694049e-01 4.74801436e-02 3.51964720e-02 1.31344199e-01
1.29401982e+00 1.23712763e-01 4.37710524e-01 -2.64516231e-02
1.55983567e-01 -1.12593941e-01 1.06030488e+00 9.45672512e-01
-5.12012064e-01 3.07919800e-01 5.98358750e-01 -4.34073687e-01
-8.11915994e-01 -1.14875543e+00 -1.43659309e-01 8.30138981e-01
-2.18297288e-01 -2.04105869e-01 -4.11360413e-01 -5.68388700e-01
3.63038599e-01 8.90274763e-01 -7.05453813e-01 -6.56719744e-01
-5.61276436e-01 -7.38608360e-01 3.64786908e-02 7.26319373e-01
-9.69454646e-02 -1.29658329e+00 -9.83130336e-01 8.29605043e-01
-3.17398369e-01 -7.77638912e-01 -4.81891632e-01 2.25772858e-01
-1.10906029e+00 -1.23589075e+00 -7.37985015e-01 -1.46126390e-01
3.49060357e-01 -2.06073001e-01 1.14151096e+00 1.41435638e-01
-1.82683811e-01 6.30522430e-01 -3.49139422e-01 -3.91459167e-01
-3.99357408e-01 -9.33860838e-02 3.99039686e-01 -2.42952257e-01
1.97419718e-01 -5.60262382e-01 -1.51558912e+00 2.03004386e-02
-6.60865664e-01 -1.05655424e-01 5.31935990e-01 1.18582106e+00
6.00037396e-01 -4.13601577e-01 9.64014709e-01 -1.06067324e+00
7.31956482e-01 -6.75976038e-01 -4.37062651e-01 6.88074455e-02
-9.89673972e-01 -9.12985485e-03 1.07009006e+00 -3.59811813e-01
-8.42106760e-01 -2.06154183e-01 -2.00470373e-01 -6.13489568e-01
-2.22845078e-01 5.57376146e-01 4.66468841e-01 5.24247169e-01
2.43266046e-01 2.52365679e-01 9.82585698e-02 -4.13722873e-01
7.51689672e-02 5.46883702e-01 3.94739419e-01 -5.82130849e-01
3.75308171e-02 1.12616494e-01 4.21588793e-02 -8.27690288e-02
-1.00530970e+00 -2.78624594e-01 -1.64188400e-01 2.45373547e-02
6.84781075e-01 -8.03826094e-01 -1.43099761e+00 1.42133906e-01
-6.01587474e-01 -9.39438581e-01 -3.63972306e-01 8.95401001e-01
-9.36390340e-01 1.42505661e-01 -9.83375013e-01 -6.74925923e-01
-6.78141296e-01 -1.28193069e+00 6.01544380e-01 2.49904677e-01
-2.14203730e-01 -1.15655816e+00 3.65849793e-01 3.10136050e-01
6.63816214e-01 9.69364494e-02 1.11220348e+00 -5.91836691e-01
-2.35120445e-01 4.77541611e-02 1.16663508e-01 3.71971101e-01
4.28116590e-01 -1.08305685e-01 -5.26239812e-01 -6.19026780e-01
-1.20157622e-01 -3.72146070e-01 6.23062909e-01 8.06358516e-01
1.52764285e+00 -3.06879759e-01 -1.92838132e-01 6.53661430e-01
1.31076705e+00 5.78931332e-01 3.26348782e-01 -1.04610585e-01
5.26255012e-01 1.08356111e-01 5.98600209e-01 1.32501304e+00
5.34134686e-01 2.47589931e-01 7.52134264e-01 -7.67291784e-02
3.31430256e-01 -1.66932255e-01 8.55321959e-02 6.67915940e-01
1.46726802e-01 -4.33123291e-01 -1.13994229e+00 4.85731572e-01
-2.07306147e+00 -9.13956046e-01 3.07349056e-01 2.05341935e+00
7.24054277e-01 -4.63742055e-02 1.02938123e-01 -4.42265451e-01
2.81547099e-01 -1.81680948e-01 -1.26006162e+00 -5.79024851e-01
4.22316968e-01 2.78654188e-01 4.70487207e-01 2.87928492e-01
-8.58722925e-01 6.64551914e-01 6.14782858e+00 1.14771095e-03
-1.42140090e+00 -9.21327621e-02 6.86569393e-01 -3.16453964e-01
1.15880407e-02 -2.55667090e-01 -3.60592186e-01 4.66038376e-01
1.27209306e+00 -2.46011913e-01 7.83496439e-01 4.43008214e-01
8.52854729e-01 1.48073658e-01 -1.14612484e+00 7.84053564e-01
-5.10102451e-01 -1.10877120e+00 -2.83069432e-01 -3.45971510e-02
6.42389953e-01 4.08182353e-01 2.58207709e-01 6.01417780e-01
7.74378419e-01 -9.62775826e-01 -2.58358140e-02 8.58942866e-01
6.20542467e-01 -7.88904965e-01 8.75010133e-01 2.59886563e-01
-5.77789724e-01 -6.67088747e-01 -1.78141922e-01 -6.51718304e-03
2.72712886e-01 3.32195014e-01 -1.16196442e+00 3.69329572e-01
5.20651877e-01 8.63331974e-01 -8.74048993e-02 9.44810212e-01
-1.07380420e-01 9.71077859e-01 5.97584667e-03 2.60440737e-01
3.63546431e-01 -2.14625418e-01 2.57497698e-01 7.75617361e-01
3.50193083e-01 6.85061336e-01 4.85196948e-01 4.36242878e-01
-1.36039844e-02 1.67409763e-01 -2.58118361e-01 -3.17992061e-01
3.19781870e-01 1.00701964e+00 -1.43865660e-01 -5.06400108e-01
-3.57239842e-01 5.30812919e-01 4.13089007e-01 4.36346889e-01
-8.39599967e-01 1.28744587e-01 1.03767478e+00 -7.91099817e-02
3.38096768e-01 8.63776281e-02 -4.13352661e-02 -1.19915819e+00
-5.58440030e-01 -1.19609272e+00 6.66120529e-01 -6.11045420e-01
-1.43319190e+00 3.77616614e-01 -3.48930180e-01 -1.26412809e+00
-5.39447010e-01 -4.39986736e-01 -4.75548267e-01 8.53804410e-01
-1.49652910e+00 -4.73297209e-01 -2.11636126e-01 5.57517588e-01
3.14614356e-01 1.15701452e-01 9.95704830e-01 2.17183396e-01
-8.55323434e-01 4.10842448e-01 5.13643563e-01 -2.48122569e-02
9.26479280e-01 -1.24864542e+00 1.81272835e-03 2.82813072e-01
-6.53363883e-01 5.27434111e-01 7.57501006e-01 -6.62612855e-01
-1.48049891e+00 -1.15375543e+00 1.17303342e-01 -3.66179973e-01
7.12667227e-01 2.51053303e-01 -1.12763250e+00 6.80051208e-01
1.42640471e-01 -2.21271012e-02 8.06442976e-01 2.84984503e-02
3.03620934e-01 -2.50886440e-01 -1.06312633e+00 6.55503929e-01
8.30263734e-01 -1.51257023e-01 -3.87360871e-01 3.52810651e-01
6.82124436e-01 -5.50963938e-01 -1.39591730e+00 5.99613905e-01
4.77645427e-01 -5.85967660e-01 7.63612330e-01 -1.19879365e+00
7.03107595e-01 1.80620134e-01 7.92433918e-02 -1.87037838e+00
-6.21645689e-01 -5.47605634e-01 -4.04846013e-01 3.90678316e-01
1.81643128e-01 -1.01214778e+00 6.70579314e-01 7.42587626e-01
-1.48365617e-01 -1.36301267e+00 -6.23946726e-01 -2.73954928e-01
2.09758297e-01 -8.50515589e-02 9.34259534e-01 9.00433540e-01
2.14887440e-01 1.18513077e-01 -6.15422487e-01 5.72580583e-02
5.36345005e-01 2.23169774e-01 3.26798856e-01 -8.14649105e-01
-7.36391366e-01 -4.11009848e-01 1.20787546e-01 -6.81850970e-01
1.89279597e-02 -5.85419178e-01 -3.74588482e-02 -1.86852741e+00
2.69761503e-01 -3.67317826e-01 -1.13791573e+00 7.80940235e-01
-4.28991109e-01 -2.48872086e-01 1.88700080e-01 -1.47086993e-01
-5.38770258e-01 1.07608616e+00 1.59840965e+00 1.00314282e-01
-2.66587466e-01 1.89726934e-01 -7.27581799e-01 2.83544451e-01
1.42773831e+00 -5.64144194e-01 -6.80563807e-01 -3.48631561e-01
-1.53283536e-01 9.76895332e-01 2.02724874e-01 -8.36674511e-01
-1.02935381e-01 -7.79171586e-01 4.64792132e-01 -3.21965277e-01
3.41718167e-01 -7.03684509e-01 -1.08214654e-02 1.20912778e+00
-5.66280961e-01 3.85858744e-01 2.61686713e-01 5.17037809e-01
1.01377867e-01 2.20263213e-01 7.61482477e-01 -1.83419675e-01
-3.55165750e-01 1.01765072e+00 -5.09680092e-01 2.60607868e-01
7.43560791e-01 3.83747816e-01 -5.02104163e-01 -6.90556526e-01
-9.54698980e-01 7.85751581e-01 3.22019398e-01 2.39512205e-01
7.50299513e-01 -8.81835401e-01 -9.91512120e-01 -1.38324142e-01
-9.70398542e-03 -1.84249893e-01 8.90106022e-01 1.01957929e+00
-5.02376199e-01 6.46949351e-01 -3.35821450e-01 -2.83584595e-01
-7.27737486e-01 8.66680086e-01 7.96897948e-01 -3.82709891e-01
-7.58909702e-01 1.20445758e-01 4.12374020e-01 -2.77878731e-01
1.87190160e-01 -2.88785189e-01 -2.76514143e-01 -4.60990816e-01
4.51361686e-01 2.53660768e-01 -1.53223589e-01 6.46632016e-02
-3.14810187e-01 9.27782431e-02 -2.69378424e-01 2.16136083e-01
1.48904717e+00 -1.53261200e-01 2.12589130e-01 2.76294082e-01
8.00724268e-01 -6.73971951e-01 -1.58946157e+00 -3.82427424e-01
-4.94115263e-01 -2.26157427e-01 2.36754224e-01 -1.12411177e+00
-1.20180845e+00 9.08014357e-01 8.19234788e-01 -2.96478480e-01
1.16337335e+00 -3.87742162e-01 1.13529336e+00 5.78390360e-01
1.40561029e-01 -9.56258535e-01 1.12403259e-02 4.50607002e-01
6.93555057e-01 -1.31707215e+00 -2.57356768e-03 3.83692533e-01
-9.78519380e-01 9.40152347e-01 7.80605257e-01 -1.44377649e-01
7.09977865e-01 -1.10545583e-01 5.54470658e-01 -4.68444079e-03
-1.43035483e+00 1.46538960e-02 -2.01636732e-01 3.26448828e-01
4.00079340e-01 4.33883369e-01 -2.04864696e-01 5.90905011e-01
9.12688524e-02 4.17564213e-01 6.77068472e-01 9.06738281e-01
-1.58198103e-01 -1.06049335e+00 -9.44574475e-02 1.01571012e+00
-6.79565847e-01 -1.47819623e-01 2.92828798e-01 3.18559825e-01
-2.98506230e-01 8.37242067e-01 -2.08751168e-02 -7.50856325e-02
5.06510317e-01 7.26926252e-02 4.66895610e-01 -6.03119671e-01
-6.41391337e-01 -3.34097855e-02 -1.98506340e-01 -7.31001258e-01
1.22639909e-01 -6.86674893e-01 -1.46015847e+00 -3.89546961e-01
4.02306587e-01 1.70214012e-01 3.36878240e-01 7.16412902e-01
7.55520284e-01 1.05532610e+00 7.89529145e-01 -4.05725121e-01
-1.31193697e+00 -9.17366445e-01 -4.79090005e-01 5.10312438e-01
7.85276532e-01 -6.30591810e-01 -1.37654945e-01 -3.01936269e-01]
|
[4.011046409606934, 2.722944736480713]
|
9898e350-7e5b-411e-bd6d-fd990508c80b
|
millimeter-wave-communications-with-an
|
2002.10572
| null |
https://arxiv.org/abs/2002.10572v3
|
https://arxiv.org/pdf/2002.10572v3.pdf
|
Millimeter Wave Communications with an Intelligent Reflector: Performance Optimization and Distributional Reinforcement Learning
|
In this paper, a novel framework is proposed to optimize the downlink multi-user communication of a millimeter wave base station, which is assisted by a reconfigurable intelligent reflector (IR). In particular, a channel estimation approach is developed to measure the channel state information (CSI) in real-time. First, for a perfect CSI scenario, the precoding transmission of the BS and the reflection coefficient of the IR are jointly optimized, via an iterative approach, so as to maximize the sum of downlink rates towards multiple users. Next, in the imperfect CSI scenario, a distributional reinforcement learning (DRL) approach is proposed to learn the optimal IR reflection and maximize the expectation of downlink capacity. In order to model the transmission rate's probability distribution, a learning algorithm, based on quantile regression (QR), is developed, and the proposed QR-DRL method is proved to converge to a stable distribution of downlink transmission rate. Simulation results show that, in the error-free CSI scenario, the proposed approach yields over 30% and 2-fold increase in the downlink sum-rate, compared with a fixed IR reflection scheme and direct transmission scheme, respectively. Simulation results also show that by deploying more IR elements, the downlink sum-rate can be significantly improved. However, as the number of IR components increases, more time is required for channel estimation, and the slope of increase in the IR-aided transmission rate will become smaller. Furthermore, under limited knowledge of CSI, simulation results show that the proposed QR-DRL method, which learns a full distribution of the downlink rate, yields a better prediction accuracy and improves the downlink rate by 10% for online deployments, compared with a Q-learning baseline.
|
['Qianqian Zhang', 'Walid Saad', 'Mehdi Bennis']
|
2020-02-24
| null | null | null | null |
['distributional-reinforcement-learning']
|
['methodology']
|
[-6.60585016e-02 3.10855746e-01 -1.61507502e-02 1.02019534e-01
-8.99387360e-01 -3.03452015e-01 -2.85244972e-01 -1.08352974e-01
-1.58666998e-01 1.04765463e+00 -4.92169484e-02 -4.72388923e-01
-4.24737543e-01 -1.11626017e+00 -6.64536476e-01 -1.22750533e+00
-3.23475629e-01 -9.38571990e-02 -4.95515257e-01 -2.77472943e-01
-5.23025952e-02 2.95766383e-01 -9.05175805e-01 -6.20764971e-01
7.53215551e-01 1.37686789e+00 4.32152867e-01 5.81819475e-01
4.07468349e-01 3.91407669e-01 -6.35548890e-01 -1.24062859e-01
2.42453784e-01 -3.97743195e-01 -5.04494049e-02 -1.70560896e-01
-3.90065998e-01 -5.78191280e-01 -6.14627182e-01 7.43343890e-01
1.00592875e+00 -2.47289926e-01 5.68420470e-01 -7.72936046e-01
2.75826026e-02 6.62252486e-01 -6.96413100e-01 -4.01565671e-01
2.44912669e-01 -4.31841344e-01 7.73739994e-01 -3.46537590e-01
-1.35848802e-02 8.28270257e-01 6.07378900e-01 2.15079769e-01
-7.93429673e-01 -1.06255627e+00 -4.87875231e-02 -3.98395568e-01
-1.63068175e+00 -1.88822657e-01 4.67819303e-01 -4.37381752e-02
2.67122120e-01 1.14131793e-01 8.66867125e-01 3.34580600e-01
5.09914100e-01 6.16597295e-01 5.99267423e-01 -6.71922147e-01
4.57032084e-01 1.61965653e-01 -4.87100959e-01 6.43556535e-01
5.61044157e-01 1.58445537e-01 -1.55857190e-01 -2.94351846e-01
1.03560567e+00 -1.10002555e-01 -4.74887669e-01 -2.25994736e-01
-6.66502059e-01 5.19566000e-01 2.78337151e-01 1.72282189e-01
-5.34508884e-01 2.56077260e-01 -3.26949090e-01 4.48808759e-01
3.72050494e-01 2.28955209e-01 -2.55863190e-01 3.17622013e-02
-7.96052814e-01 8.85115713e-02 1.06059468e+00 1.15684724e+00
7.39326179e-01 9.27135199e-02 -4.47765678e-01 5.72232783e-01
6.86194301e-01 1.34997594e+00 -1.20296888e-01 -7.13129997e-01
8.15976262e-01 -7.74431927e-03 6.27554059e-01 -7.64049768e-01
-5.98209441e-01 -1.32516706e+00 -1.00408268e+00 -3.15165281e-01
2.27784514e-01 -1.19170022e+00 -4.24931049e-01 1.74904239e+00
1.03631042e-01 2.61328429e-01 6.84637070e-01 7.52712786e-01
1.96637422e-01 9.04476464e-01 -4.40953761e-01 -9.17338967e-01
7.68428624e-01 -1.31166935e-01 -7.90778041e-01 -9.68434736e-02
6.98912621e-01 -4.59362179e-01 3.60094696e-01 4.58907962e-01
-1.08360827e+00 -7.85924867e-02 -1.27974355e+00 1.10692370e+00
4.53747392e-01 4.06396627e-01 3.93475026e-01 1.06919444e+00
-8.15948427e-01 -5.64900273e-03 -5.85500836e-01 -1.14835173e-01
1.82746723e-01 4.59894061e-01 4.85454172e-01 -3.43145847e-01
-1.18394709e+00 3.50501835e-01 2.23293111e-01 1.34736463e-01
-7.62965441e-01 -7.61287987e-01 -5.68837821e-01 1.74070418e-01
1.63271874e-01 -7.72728801e-01 1.22924638e+00 -6.46079063e-01
-1.93650007e+00 -1.66460738e-01 -5.50545081e-02 -3.46445322e-01
3.08125228e-01 -7.92508647e-02 -2.39745215e-01 4.41514403e-02
-3.23265195e-01 -2.66856879e-01 5.31695127e-01 -1.48140669e+00
-8.16161036e-01 -4.11586404e-01 2.65724361e-01 5.25729299e-01
-5.44525504e-01 -6.07292116e-01 -4.18460518e-01 -3.24443638e-01
2.11851701e-01 -9.92694914e-01 -4.92680609e-01 -4.06358719e-01
-3.01283091e-01 3.70841593e-01 3.68735969e-01 -4.40869749e-01
1.43429589e+00 -2.17248726e+00 6.17173165e-02 7.36713648e-01
-3.57528627e-01 -1.14774533e-01 1.23839185e-01 6.19190753e-01
5.79900682e-01 -2.01659009e-01 -1.07671812e-01 -4.19561416e-02
-4.49403197e-01 -5.02998233e-02 -1.02498502e-01 4.99600798e-01
-5.80338299e-01 3.23705226e-01 -9.61612046e-01 2.04721447e-02
-8.51238072e-02 2.85418361e-01 -6.18300736e-01 4.09325123e-01
1.62670076e-01 6.69116735e-01 -8.87890995e-01 5.59520900e-01
9.90542710e-01 -1.50962740e-01 6.05210304e-01 -2.16794834e-01
-3.57970208e-01 -4.08995509e-01 -1.19868016e+00 1.46249139e+00
-1.23506033e+00 1.09232411e-01 2.76811242e-01 -1.14291120e+00
1.26675761e+00 4.22262967e-01 9.00606751e-01 -1.08226085e+00
2.04325825e-01 1.63234040e-01 -2.32178003e-01 -4.11668330e-01
1.95811138e-01 -3.20939869e-01 -2.64831454e-01 4.79843169e-01
-3.13689649e-01 -1.02351059e-03 -2.99124330e-01 1.80828303e-01
1.09242511e+00 -2.21031368e-01 2.19279677e-01 -1.65330887e-01
5.15402913e-01 -5.84664345e-01 5.81033647e-01 8.21024060e-01
3.39902699e-01 2.83485129e-02 1.65418446e-01 1.25571758e-01
-6.60451531e-01 -1.03202820e+00 -2.10745767e-01 6.26874685e-01
8.20709169e-01 1.84121102e-01 -5.25717795e-01 1.46688774e-01
1.96398437e-01 7.92020202e-01 8.07544813e-02 -2.35963106e-01
6.44553676e-02 -1.25142217e+00 2.54929394e-01 -7.21120536e-02
6.39125466e-01 -2.00669333e-01 -3.49993795e-01 5.44245601e-01
-3.15320313e-01 -1.13847756e+00 2.13263422e-01 4.03289795e-02
-7.85166800e-01 -5.21187544e-01 -1.05739558e+00 -4.27580863e-01
6.29685938e-01 5.03314853e-01 4.54959601e-01 -4.92716804e-02
4.24715020e-02 7.93004334e-01 -4.27744508e-01 -4.81141627e-01
4.50928882e-02 6.54841438e-02 9.45013091e-02 1.97379962e-01
-6.20519102e-01 -5.79384983e-01 -9.14010227e-01 4.36830640e-01
-4.80121285e-01 -8.58566836e-02 9.92778122e-01 5.63293815e-01
4.20865655e-01 3.12964499e-01 9.82719064e-01 -5.03124654e-01
4.75932866e-01 -8.50026548e-01 -8.89041245e-01 4.45726663e-01
-5.51683307e-01 1.31486580e-01 6.53885126e-01 5.81823699e-02
-1.37109518e+00 -4.29416485e-02 -1.39880091e-01 1.62947401e-01
5.34045875e-01 7.78759420e-01 -2.86463499e-01 -4.89399731e-01
3.38142604e-01 2.75829941e-01 -4.85298596e-02 7.89811313e-02
1.22638429e-02 1.06885898e+00 -4.76221442e-02 -6.30240798e-01
9.29352403e-01 4.28158760e-01 3.20274919e-01 -1.12990105e+00
-1.04439557e+00 -5.19803047e-01 5.24449274e-02 -5.02306104e-01
1.65027291e-01 -1.53042483e+00 -9.97447431e-01 3.35832477e-01
-7.88383603e-01 -4.81860578e-01 4.56083566e-01 9.48998213e-01
-8.10703993e-01 2.04860583e-01 -2.78851360e-01 -1.52106261e+00
-4.78003114e-01 -7.75783539e-01 8.13379586e-01 4.60401267e-01
5.75421214e-01 -8.64467263e-01 -5.40844016e-02 -1.05257044e-02
5.19847393e-01 1.75733745e-01 6.11944020e-01 2.48671606e-01
-7.45732844e-01 -2.51370937e-01 -1.47197843e-01 1.60867050e-02
5.43691330e-02 -6.44068718e-01 -6.11862242e-01 -8.26900423e-01
-2.40937471e-01 -4.23556678e-02 1.25157312e-01 8.11292231e-01
1.14538646e+00 -4.96396720e-01 -5.62267900e-01 8.68333042e-01
1.95870304e+00 3.47013921e-01 8.44725668e-01 1.64172620e-01
1.05562098e-02 -1.70423374e-01 1.06289840e+00 1.40013945e+00
5.03484130e-01 5.94081938e-01 8.39441955e-01 -1.09539174e-01
5.03773808e-01 7.70531893e-02 3.98335099e-01 5.67231297e-01
-1.77629873e-01 -7.08413720e-01 -3.76897037e-01 4.77135833e-03
-1.68217897e+00 -6.95371509e-01 1.17714405e-01 2.81140757e+00
5.86668670e-01 -1.59532353e-01 -8.29483790e-04 -5.27487658e-02
5.28679430e-01 -3.39226723e-01 -5.79651594e-01 8.80957097e-02
1.11270189e-01 -1.07954122e-01 1.25805151e+00 4.91361409e-01
-5.74595749e-01 3.70436907e-01 6.08211184e+00 5.38562477e-01
-1.19373822e+00 -2.52300799e-01 4.61015642e-01 8.40294138e-02
-6.47764683e-01 -1.87517405e-01 -9.27257419e-01 4.26358432e-01
9.78487253e-01 -4.73030806e-01 6.13141596e-01 5.12220383e-01
6.89936161e-01 -3.86114359e-01 -4.53603685e-01 1.07091653e+00
-1.64751351e-01 -9.96178150e-01 -1.52809501e-01 3.99267703e-01
7.79381633e-01 -2.67444342e-01 -5.51872849e-02 2.76801974e-01
1.18582234e-01 -3.69289756e-01 4.65808272e-01 1.09281445e+00
7.94614732e-01 -1.26254404e+00 9.92231488e-01 6.70689702e-01
-1.11895275e+00 -7.13882804e-01 -4.90954667e-01 -7.92797431e-02
1.72048002e-01 1.20027435e+00 -7.50479937e-01 1.13910997e+00
4.00246441e-01 2.89341480e-01 3.33821565e-01 1.50062203e+00
-9.25035849e-02 8.07968974e-01 -3.21300596e-01 -4.07286704e-01
-1.00145862e-02 -4.70436811e-01 5.41023672e-01 9.25677598e-01
1.05387354e+00 4.39820737e-01 2.74884403e-01 1.19287178e-01
9.43585020e-03 2.04827741e-01 -4.45517540e-01 4.47611570e-01
8.34597886e-01 1.34769714e+00 -1.00465976e-01 6.65468052e-02
-3.52891207e-01 5.07493019e-01 -1.15813047e-01 8.71004939e-01
-8.58437717e-01 -5.62067211e-01 7.16767609e-01 8.55007023e-02
2.35846847e-01 -5.09674013e-01 -2.35776693e-01 -5.55115640e-01
-8.91285166e-02 -1.60820588e-01 1.31302416e-01 -3.99739772e-01
-6.42952800e-01 4.39594835e-02 -3.08603883e-01 -1.55329573e+00
-2.81717062e-01 -1.96936131e-01 -4.09882218e-01 8.24188769e-01
-1.90259600e+00 -7.72848010e-01 -3.97226363e-01 4.59163576e-01
-1.39645353e-01 -1.91544577e-01 7.91966379e-01 4.59800571e-01
-4.39208269e-01 9.46858525e-01 8.96966577e-01 -1.14273414e-01
3.94697815e-01 -7.49100506e-01 -8.08419049e-01 3.73053938e-01
-4.51960325e-01 2.88595438e-01 7.65056312e-01 -3.28397572e-01
-1.99994814e+00 -1.13850784e+00 1.72216557e-02 5.69685280e-01
4.58890826e-01 -1.26670942e-01 -1.74125895e-01 1.89007148e-01
-1.20998010e-01 -1.43148839e-01 9.38633919e-01 5.24807461e-02
4.10623163e-01 -7.34206557e-01 -1.16698420e+00 4.36794966e-01
7.20596731e-01 4.91768382e-02 4.77691561e-01 3.31489801e-01
4.62067634e-01 -5.24827182e-01 -1.12163460e+00 4.12728697e-01
9.43450689e-01 -5.07032335e-01 8.90307426e-01 2.55790442e-01
-7.91168511e-02 -1.67029545e-01 -3.75390798e-01 -1.51794100e+00
-2.74992436e-01 -7.99204767e-01 -1.84656784e-01 9.56647635e-01
6.25451088e-01 -6.22123480e-01 7.12564170e-01 7.61231482e-02
-6.61470518e-02 -8.87769639e-01 -1.01292038e+00 -9.07113791e-01
-2.29354948e-01 -2.68020600e-01 2.33954847e-01 6.70926943e-02
9.08920467e-02 1.61959931e-01 -5.35463393e-01 9.70082581e-01
8.82373214e-01 4.10075635e-02 8.51626813e-01 -1.02342772e+00
-4.79389042e-01 2.32811600e-01 -2.06279546e-01 -1.52269840e+00
-1.63105249e-01 -6.46616459e-01 2.46798903e-01 -1.81534040e+00
1.32157840e-03 -1.07689869e+00 -5.53000867e-01 4.25328724e-02
7.97110870e-02 -9.99626070e-02 -1.45408273e-01 -5.32163270e-02
-5.21845162e-01 8.96022379e-01 1.35724819e+00 -1.65796578e-02
-5.45509398e-01 1.04583585e+00 -8.01329315e-01 2.61715651e-01
7.17398882e-01 -2.84576476e-01 -4.37029064e-01 -3.03500861e-01
5.51846266e-01 9.77898836e-01 -9.22485441e-02 -1.37628293e+00
1.89318582e-01 -2.58971423e-01 2.80064076e-01 -4.60780978e-01
2.77767688e-01 -9.59918141e-01 -2.95566097e-02 6.79783762e-01
1.43798336e-01 -7.86197126e-01 9.49750692e-02 9.98159230e-01
1.57397851e-01 -1.41323030e-01 6.72582388e-01 3.05476397e-01
-1.29974872e-01 5.40420949e-01 -5.49836636e-01 -3.59980285e-01
1.03993452e+00 -9.40025711e-05 2.09867999e-01 -1.13986146e+00
-3.24207962e-01 4.84082907e-01 -1.90859854e-01 -1.63104609e-01
4.87950861e-01 -1.16107571e+00 -6.45699561e-01 -1.52044386e-01
-1.46991690e-03 -2.06684366e-01 3.18561196e-01 7.52372205e-01
-1.33012444e-01 3.77540469e-01 3.27054352e-01 -4.80495781e-01
-6.97966635e-01 -2.45624185e-01 4.67848897e-01 -2.07803741e-01
-1.93831161e-01 5.56096792e-01 -5.49472086e-02 -5.14029115e-02
3.08839619e-01 1.27813742e-01 -1.04944795e-01 -1.92547739e-01
4.51995313e-01 9.88142788e-02 1.66500628e-01 -8.29817951e-02
3.42978537e-02 6.84416771e-01 4.81758803e-01 -1.33208007e-01
1.39260888e+00 -6.25314176e-01 1.93861589e-01 -2.52680406e-02
8.42967749e-01 3.04390818e-01 -1.34288049e+00 -3.43057394e-01
-5.16140878e-01 -4.29871410e-01 3.86416852e-01 -7.19002604e-01
-1.13372624e+00 3.05938244e-01 8.17360878e-01 1.46551520e-01
1.29608166e+00 -1.90190703e-01 6.00091279e-01 7.17261016e-01
1.09779501e+00 -1.04940128e+00 1.31050214e-01 5.20477235e-01
6.26310110e-01 -8.31659019e-01 1.25765085e-01 -3.21781725e-01
-2.41110072e-01 1.23793983e+00 4.05143261e-01 2.64721155e-01
8.66247356e-01 6.03556335e-01 -1.16441831e-01 2.64206529e-01
-2.21571282e-01 -2.81568527e-01 -2.35582784e-01 3.66383255e-01
5.20560324e-01 3.37348461e-01 -4.51746017e-01 6.12458050e-01
-1.73438713e-01 -4.30067591e-02 6.14668012e-01 7.62832403e-01
-9.17716146e-01 -1.02490163e+00 -3.54042917e-01 5.00016928e-01
-2.47176945e-01 6.61606416e-02 5.03608167e-01 5.13451397e-01
-3.13214540e-01 1.16219270e+00 3.17185111e-02 -2.90151805e-01
3.67683440e-01 -6.75910592e-01 5.94893157e-01 -3.55161905e-01
4.53546405e-01 3.60862389e-02 3.77406515e-02 -2.12761015e-01
-1.87433973e-01 -2.96131343e-01 -1.49972594e+00 7.07347691e-02
-6.26511276e-01 4.70445842e-01 9.43808556e-01 1.14532733e+00
2.65854418e-01 6.05088115e-01 1.49911785e+00 -4.61344630e-01
-6.38202965e-01 -6.49445653e-01 -9.57370639e-01 -4.97858644e-01
2.07316875e-01 -5.96250772e-01 -3.16995203e-01 -6.12101972e-01]
|
[6.074460506439209, 1.4476983547210693]
|
5fc159d7-0bb3-4580-8f52-b9edf019a320
|
lip-to-speech-synthesis-for-arbitrary
|
2209.00642
| null |
https://arxiv.org/abs/2209.00642v1
|
https://arxiv.org/pdf/2209.00642v1.pdf
|
Lip-to-Speech Synthesis for Arbitrary Speakers in the Wild
|
In this work, we address the problem of generating speech from silent lip videos for any speaker in the wild. In stark contrast to previous works, our method (i) is not restricted to a fixed number of speakers, (ii) does not explicitly impose constraints on the domain or the vocabulary and (iii) deals with videos that are recorded in the wild as opposed to within laboratory settings. The task presents a host of challenges, with the key one being that many features of the desired target speech, like voice, pitch and linguistic content, cannot be entirely inferred from the silent face video. In order to handle these stochastic variations, we propose a new VAE-GAN architecture that learns to associate the lip and speech sequences amidst the variations. With the help of multiple powerful discriminators that guide the training process, our generator learns to synthesize speech sequences in any voice for the lip movements of any person. Extensive experiments on multiple datasets show that we outperform all baselines by a large margin. Further, our network can be fine-tuned on videos of specific identities to achieve a performance comparable to single-speaker models that are trained on $4\times$ more data. We conduct numerous ablation studies to analyze the effect of different modules of our architecture. We also provide a demo video that demonstrates several qualitative results along with the code and trained models on our website: \url{http://cvit.iiit.ac.in/research/projects/cvit-projects/lip-to-speech-synthesis}}
|
['C. V. Jawahar', 'Vinay P Namboodiri', 'Rudrabha Mukhopadhyay', 'K R Prajwal', 'Sindhu B Hegde']
|
2022-09-01
| null | null | null | null |
['lip-to-speech-synthesis']
|
['computer-vision']
|
[ 2.16212064e-01 9.14875790e-02 -1.32432655e-01 -3.59565914e-01
-1.15616035e+00 -7.73313284e-01 4.15399969e-01 -8.51671040e-01
-4.25700285e-02 6.15004182e-01 2.41700962e-01 -2.11491331e-01
4.29080784e-01 -1.10796079e-01 -7.61430740e-01 -6.94824815e-01
1.76223025e-01 1.56557336e-01 -2.91026812e-02 -3.22193727e-02
-5.02868555e-02 4.54147965e-01 -1.62195039e+00 3.21863860e-01
4.45336938e-01 9.90897119e-01 2.43987203e-01 8.12274754e-01
1.91354364e-01 3.78757417e-01 -6.36111319e-01 -3.95949483e-01
4.20991302e-01 -8.98251235e-01 -4.81462717e-01 2.40107000e-01
7.58558571e-01 -4.79889691e-01 -3.07958513e-01 9.87197757e-01
8.96358192e-01 -7.47977421e-02 5.08119226e-01 -1.35411870e+00
-3.29902709e-01 4.67568845e-01 -4.05993789e-01 1.04114376e-01
4.29755747e-01 6.12595439e-01 7.16422319e-01 -9.34664130e-01
5.00949383e-01 1.17090905e+00 4.91385460e-01 1.08877695e+00
-1.20547378e+00 -9.51096296e-01 5.71033694e-02 -1.58802584e-01
-1.41436982e+00 -1.38021326e+00 6.91678047e-01 -3.98957491e-01
5.24686933e-01 1.99587539e-01 4.27014500e-01 1.69288135e+00
-3.10787946e-01 5.65786779e-01 9.78931904e-01 -3.28683645e-01
1.47599518e-01 1.30839989e-01 -5.64024329e-01 5.85491896e-01
-2.28003308e-01 1.99472293e-01 -7.39714622e-01 -4.98609096e-02
7.72789180e-01 -5.44778883e-01 -5.38143516e-01 -2.35526860e-01
-1.26345909e+00 7.66250491e-01 -4.84679192e-02 2.20650524e-01
-9.73473489e-02 2.18608946e-01 2.98740625e-01 2.02673420e-01
2.41843790e-01 1.70387506e-01 -4.54847336e-01 -3.00260216e-01
-1.03816938e+00 2.02130303e-01 7.19687045e-01 1.01478326e+00
4.49546546e-01 4.39357847e-01 -1.23496644e-01 9.94443178e-01
2.76746362e-01 6.65246546e-01 7.07852900e-01 -1.25623190e+00
4.81635869e-01 -1.93022475e-01 -7.21188858e-02 -4.68037993e-01
-2.47539915e-02 -1.56479493e-01 -4.81518865e-01 3.12391847e-01
5.67740262e-01 -6.75949514e-01 -9.65814114e-01 2.20848298e+00
1.79591596e-01 4.19094473e-01 -7.88371116e-02 7.27777421e-01
8.77930641e-01 4.74192917e-01 -8.70818794e-02 -3.19090605e-01
1.15969110e+00 -8.02358150e-01 -7.30900347e-01 -3.77741039e-01
1.40054360e-01 -9.89338577e-01 1.15356719e+00 1.78713277e-01
-1.30466950e+00 -5.27523875e-01 -7.58653164e-01 1.71715826e-01
-3.88264889e-03 2.99375564e-01 3.68966460e-01 8.10576499e-01
-1.41122723e+00 1.70681730e-01 -6.06099665e-01 -4.17277634e-01
4.35314804e-01 4.51441228e-01 -3.04555774e-01 2.20307499e-01
-9.16804790e-01 4.42602605e-01 -1.70206696e-01 1.19962422e-02
-1.15988779e+00 -5.72472870e-01 -9.81146634e-01 -9.16954204e-02
1.91402122e-01 -6.82777882e-01 1.51862764e+00 -1.32355213e+00
-2.03190541e+00 8.41906011e-01 -5.50334334e-01 -2.69239247e-01
6.24909580e-01 1.00169145e-01 -4.07432914e-01 2.20790967e-01
1.03145968e-02 1.07876253e+00 1.16708398e+00 -1.11139286e+00
-4.73663032e-01 4.06498695e-03 -1.28020152e-01 1.73166111e-01
-8.93128291e-02 2.19351307e-01 -5.35715997e-01 -8.20630014e-01
-2.22296789e-01 -1.14199340e+00 2.65130550e-01 -8.47110227e-02
-3.33486408e-01 -3.22639309e-02 8.12247574e-01 -5.53912997e-01
7.87569225e-01 -2.42526960e+00 4.51344764e-03 -1.53825864e-01
-9.93380025e-02 3.38966936e-01 -3.80867809e-01 2.39149302e-01
-1.02793969e-01 2.91098535e-01 -1.37443125e-01 -5.69035351e-01
-7.93465525e-02 -2.40137473e-01 -4.53010738e-01 4.50323969e-01
2.04725370e-01 7.42572188e-01 -5.49401581e-01 -3.79427761e-01
-7.80756548e-02 7.43374109e-01 -6.83231652e-01 3.60267371e-01
-1.98982999e-01 7.36314118e-01 -1.42283112e-01 6.59160972e-01
4.44744438e-01 4.22962271e-02 2.01327994e-01 -9.91188213e-02
8.04780200e-02 4.77523744e-01 -1.03978968e+00 1.76173425e+00
-5.51276028e-01 8.05139780e-01 4.88159865e-01 -7.96565413e-01
6.60018802e-01 8.50575447e-01 3.38071287e-01 -3.97149861e-01
1.64601788e-01 3.00926089e-01 1.46954730e-01 -6.75068498e-01
-8.95017087e-02 -1.03281073e-01 1.38414711e-01 4.60971355e-01
3.61806422e-01 -2.66847581e-01 1.67655826e-01 -1.22051917e-01
7.98245370e-01 2.02555880e-01 -6.43349439e-02 -1.13686666e-01
4.22403425e-01 -6.38276935e-01 6.61177397e-01 3.98677856e-01
-3.91140997e-01 8.97832036e-01 5.28271616e-01 8.07545781e-02
-9.15199697e-01 -1.16946626e+00 -1.98686495e-01 1.17915761e+00
-1.73226431e-01 -2.06148937e-01 -1.01796234e+00 -5.60990095e-01
-2.16663718e-01 5.00719845e-01 -4.11937863e-01 9.86483097e-02
-4.92977232e-01 -2.71710843e-01 7.57743001e-01 3.18316013e-01
3.44123185e-01 -1.18931496e+00 -3.17050636e-01 -1.79249331e-01
-3.54148656e-01 -1.43488193e+00 -9.73801076e-01 -2.20779806e-01
-4.34776962e-01 -7.74424076e-01 -7.69216716e-01 -1.04176569e+00
5.40965855e-01 3.47193703e-02 8.92516315e-01 -2.12657869e-01
-2.18472049e-01 4.18321908e-01 -2.87899040e-02 -5.44620872e-01
-6.12173855e-01 1.56600811e-02 2.15126827e-01 3.13156098e-01
1.89861745e-01 -7.12805569e-01 -5.81128895e-01 3.47630531e-01
-5.63505530e-01 -3.92445736e-02 4.35035735e-01 7.58440614e-01
3.91399384e-01 -7.39802569e-02 9.41204667e-01 -5.79616487e-01
6.17214203e-01 -3.50435078e-01 -5.87606668e-01 -1.98362060e-02
-3.10916658e-02 -1.76741958e-01 4.68458891e-01 -7.21362174e-01
-8.62560034e-01 2.80056179e-01 -2.91769505e-01 -5.73416531e-01
-4.82713401e-01 -1.26988530e-01 -5.74561417e-01 6.34078309e-02
5.09769917e-01 2.91997403e-01 3.29927415e-01 -2.79925942e-01
2.55022258e-01 1.05021429e+00 5.36493421e-01 -4.34558600e-01
7.25153804e-01 3.67106944e-01 -2.73987800e-01 -1.11727107e+00
-4.62059975e-01 2.13390887e-02 -3.41210067e-01 -2.22789600e-01
7.43638992e-01 -1.20774937e+00 -9.03841555e-01 6.39790595e-01
-9.37669575e-01 -7.37691402e-01 -9.54408348e-02 5.74011922e-01
-9.30173695e-01 1.10208087e-01 -5.74283779e-01 -9.12024975e-01
-1.75655037e-01 -1.46738422e+00 1.05723202e+00 2.88900942e-01
-3.54026437e-01 -7.34403968e-01 -1.60763964e-01 6.58024609e-01
5.10387182e-01 9.03386623e-02 4.47926521e-01 -5.87370753e-01
-4.68983650e-01 6.37102202e-02 3.60913515e-01 6.12745762e-01
3.84337068e-01 1.86374262e-01 -1.34970474e+00 -3.76510918e-01
3.79336253e-02 -6.05807304e-01 6.11138046e-01 5.20412326e-01
1.05763578e+00 -4.99396652e-01 -4.76878658e-02 7.04890788e-01
8.84936690e-01 2.61123717e-01 5.03562391e-01 -2.35202268e-01
3.45599294e-01 6.58535063e-01 2.14621246e-01 1.78851590e-01
1.23413600e-01 8.68236482e-01 2.17399135e-01 -3.79042737e-02
-5.26419878e-01 -4.32914019e-01 7.78271377e-01 3.87544453e-01
1.32361293e-01 -3.39150637e-01 -5.69355190e-01 6.49314582e-01
-1.27464426e+00 -1.14038694e+00 5.16843855e-01 2.24809408e+00
1.00681090e+00 -5.42389788e-02 6.03298545e-01 -5.16071878e-02
9.45104182e-01 5.63268811e-02 -6.45905197e-01 -3.02606672e-01
-3.81040908e-02 3.40994895e-01 1.71594784e-01 7.28622079e-01
-9.03575659e-01 1.03808844e+00 6.29135323e+00 6.16792917e-01
-1.65700364e+00 2.57590134e-02 7.37505853e-01 -5.35130143e-01
-8.99169371e-02 -2.71888405e-01 -8.34048629e-01 6.30556703e-01
1.04379261e+00 -7.04931170e-02 8.35129023e-01 6.30484164e-01
4.71586019e-01 3.23714525e-01 -1.23458147e+00 1.02501285e+00
2.06938758e-01 -1.10137308e+00 -1.09540224e-01 1.76570579e-01
5.08193135e-01 1.65898785e-01 5.88072956e-01 1.06638178e-01
1.70306846e-01 -1.31459081e+00 9.56535339e-01 1.70821160e-01
1.26021779e+00 -5.26121378e-01 2.45406732e-01 2.76028812e-01
-8.97900343e-01 -7.69468397e-02 3.89339477e-02 3.09450090e-01
1.18582524e-01 7.19553456e-02 -1.03759158e+00 4.31097001e-02
6.80079341e-01 2.75487900e-01 -2.25227922e-01 7.59143293e-01
-3.68519932e-01 8.64873469e-01 -2.95960367e-01 1.94418401e-01
-2.02246740e-01 8.87157395e-02 6.17677331e-01 1.07621324e+00
4.83810961e-01 -1.56298324e-01 -2.01592296e-01 9.24732566e-01
-3.30081999e-01 -1.58721022e-02 -7.20393121e-01 -1.73731402e-01
6.39225185e-01 1.18458867e+00 -1.75618872e-01 2.13740952e-02
-3.92463207e-01 6.64022684e-01 4.29954156e-02 6.32731557e-01
-8.07037234e-01 -2.80883789e-01 9.73006070e-01 4.27741349e-01
4.93487090e-01 -1.08597651e-01 1.29673645e-01 -1.09902036e+00
2.02952385e-01 -1.29060066e+00 -2.26494092e-02 -8.12046409e-01
-1.07692099e+00 9.03676510e-01 -1.74646974e-01 -1.01875544e+00
-8.07674229e-01 -4.73730981e-01 -6.93657696e-01 1.12690330e+00
-1.37095284e+00 -1.13045621e+00 -8.13904330e-02 7.65786052e-01
8.42403591e-01 -4.92063165e-01 8.51392746e-01 2.85457939e-01
-6.44707978e-01 9.76692319e-01 -2.71417648e-01 4.28933978e-01
9.18374002e-01 -8.25317383e-01 3.55516344e-01 8.86412799e-01
2.53322035e-01 5.37014186e-01 7.54563391e-01 -2.23820016e-01
-1.09105670e+00 -8.40380907e-01 8.06517899e-01 -3.92918497e-01
5.45592010e-01 -8.06038141e-01 -5.48589945e-01 8.10362816e-01
4.30050880e-01 1.88578069e-01 8.37930977e-01 -1.51523873e-01
-3.57796520e-01 -2.85735279e-01 -1.13876605e+00 6.31356776e-01
1.01829445e+00 -6.02172673e-01 -1.67884231e-01 2.29281649e-01
6.11893117e-01 -5.29821515e-01 -5.00984192e-01 3.92111391e-01
6.97114289e-01 -1.07981157e+00 7.47391999e-01 -5.16994655e-01
1.45608902e-01 -2.07676440e-01 -1.95604399e-01 -1.21453893e+00
1.48527339e-01 -1.12184811e+00 -4.54620458e-02 1.70808041e+00
6.29790664e-01 -6.88551605e-01 6.44429386e-01 4.70554799e-01
4.23910916e-02 -6.42848492e-01 -1.01372445e+00 -8.13645363e-01
2.29660526e-01 -3.42664748e-01 7.12112606e-01 5.51593006e-01
-1.83139130e-01 3.96460444e-01 -3.92880708e-01 1.80042788e-01
4.13257897e-01 -6.12563044e-02 9.31738675e-01 -7.66082466e-01
-3.97834092e-01 -4.33847666e-01 -2.04353347e-01 -1.03275526e+00
5.70229948e-01 -7.16331840e-01 2.23438919e-01 -1.03123057e+00
-1.11269780e-01 -3.81808907e-01 1.89505778e-02 4.15088475e-01
2.49474272e-02 3.57833475e-01 2.08993524e-01 5.35373986e-02
-7.48027638e-02 4.25112814e-01 1.07148790e+00 -4.28883024e-02
-1.65503442e-01 3.20608050e-01 -8.46155763e-01 6.44704163e-01
9.77946460e-01 -3.33987087e-01 -6.95963979e-01 -4.33431000e-01
-2.23302200e-01 2.28057787e-01 3.79710346e-01 -9.79735553e-01
1.16196074e-01 -8.45625699e-02 2.46765211e-01 8.48527551e-02
6.34615898e-01 -5.48812985e-01 2.66359150e-02 1.30281553e-01
-4.74263519e-01 -1.71592116e-01 2.32392043e-01 2.35573009e-01
-9.53023136e-02 -8.41931924e-02 1.13552439e+00 -2.53843367e-02
-1.55209929e-01 4.16744471e-01 -4.11255032e-01 3.28283638e-01
9.19383943e-01 -1.30636869e-02 -2.31398806e-01 -8.22438896e-01
-7.01776266e-01 -5.46455272e-02 5.29856920e-01 5.13177633e-01
2.85696447e-01 -1.24244595e+00 -8.73415232e-01 4.98660326e-01
-3.97822373e-02 -4.65819128e-02 1.45945430e-01 6.32863820e-01
-2.16627970e-01 3.75536382e-01 -8.91066641e-02 -5.91097236e-01
-1.42539108e+00 4.05573368e-01 6.30422592e-01 3.35142463e-01
-3.53503823e-01 8.47616494e-01 4.14696515e-01 -2.25580886e-01
4.47384715e-01 5.25843874e-02 1.06348924e-01 -1.22991778e-01
4.75815892e-01 -5.02010882e-02 -9.04109105e-02 -8.85571063e-01
-2.89518982e-01 4.72163171e-01 1.96399689e-01 -4.76594925e-01
1.19579554e+00 -2.04493448e-01 3.16007793e-01 3.77379596e-01
1.14511025e+00 5.35397530e-01 -1.63846993e+00 6.50654361e-02
-4.92500126e-01 -4.32649583e-01 -2.39664301e-01 -7.96839058e-01
-1.33261180e+00 9.39898014e-01 5.45347691e-01 -8.84523094e-02
1.16620350e+00 1.73023656e-01 6.12038076e-01 -5.74940117e-03
3.70444208e-02 -8.71203244e-01 2.28522733e-01 2.03405291e-01
1.10056984e+00 -1.16310346e+00 -5.89665532e-01 -3.86040151e-01
-8.55954766e-01 1.02368164e+00 5.11001348e-01 1.57566801e-01
5.47775149e-01 4.62148190e-01 5.11211336e-01 1.77539587e-01
-8.20011437e-01 -2.04539388e-01 1.43347844e-01 8.64155591e-01
6.75265253e-01 -1.34018824e-01 1.88055947e-01 2.97945917e-01
-6.23058081e-01 6.42037168e-02 5.00686884e-01 5.24908602e-01
-1.17117107e-01 -1.16688073e+00 -2.19732493e-01 5.71625009e-02
-7.65412748e-01 -7.11144060e-02 -3.69878918e-01 5.68989933e-01
1.92945525e-01 1.25441158e+00 3.17188390e-02 -2.74217457e-01
1.05486304e-01 3.57272565e-01 4.49110895e-01 -6.73343301e-01
-2.64367551e-01 2.98983485e-01 5.44212386e-03 -4.87861156e-01
-2.09286049e-01 -1.01371253e+00 -1.01065445e+00 -2.33594745e-01
-1.11487105e-01 6.65605813e-02 7.31874287e-01 6.19929194e-01
5.57287693e-01 3.19862306e-01 8.30194116e-01 -9.05128419e-01
-5.12913764e-01 -1.07652628e+00 -5.47924459e-01 3.15640748e-01
6.28881276e-01 -4.59487289e-01 -7.89243758e-01 3.18745762e-01]
|
[13.304154396057129, -0.2822752296924591]
|
a9e9c9ca-099d-4df6-8ae2-93c7ef45425a
|
a-generative-map-for-image-based-camera
|
1902.11124
| null |
http://arxiv.org/abs/1902.11124v4
|
http://arxiv.org/pdf/1902.11124v4.pdf
|
A Generative Map for Image-based Camera Localization
|
In image-based camera localization systems, information about the environment
is usually stored in some representation, which can be referred to as a map.
Conventionally, most maps are built upon hand-crafted features. Recently,
neural networks have attracted attention as a data-driven map representation,
and have shown promising results in visual localization. However, these neural
network maps are generally hard to interpret by human. A readable map is not
only accessible to humans, but also provides a way to be verified when the
ground truth pose is unavailable. To tackle this problem, we propose Generative
Map, a new framework for learning human-readable neural network maps, by
combining a generative model with the Kalman filter, which also allows it to
incorporate additional sensor information such as stereo visual odometry. For
evaluation, we use real world images from the 7-Scenes and Oxford RobotCar
datasets. We demonstrate that our Generative Map can be queried with a pose of
interest from the test sequence to predict an image, which closely resembles
the true scene. For localization, we show that Generative Map achieves
comparable performance with current regression models. Moreover, our framework
is trained completely from scratch, unlike regression models which rely on
large ImageNet pretrained networks.
|
['Stefan Matthes', 'Mingpan Guo', 'Jiaojiao Ye', 'Hao Shen']
|
2019-02-18
| null | null | null | null |
['camera-localization']
|
['computer-vision']
|
[ 1.12508230e-01 5.32325543e-03 -1.31992817e-01 -6.34494901e-01
-6.32442355e-01 -6.38984919e-01 7.15841293e-01 8.34970027e-02
-5.38841605e-01 7.44870961e-01 -3.00968718e-02 -1.97519585e-02
2.39436086e-02 -9.51913774e-01 -1.30063796e+00 -6.48737192e-01
2.70864338e-01 7.36549020e-01 3.32269818e-01 -1.20948002e-01
1.25688106e-01 3.96552354e-01 -1.45207107e+00 -2.06372038e-01
6.73674941e-01 1.04472339e+00 8.85431588e-01 4.68648195e-01
1.87266722e-01 9.97671783e-01 -3.82633597e-01 1.09046243e-01
1.15259081e-01 -2.68092722e-01 -5.33360064e-01 6.23784587e-02
5.06406248e-01 -4.81743008e-01 -5.22871137e-01 1.22874010e+00
5.26832379e-02 1.95434511e-01 5.33662260e-01 -1.28291011e+00
-6.03667080e-01 2.95140713e-01 5.93819767e-02 -2.61996567e-01
5.12719154e-01 6.45462275e-02 8.29686284e-01 -7.41020322e-01
7.56284118e-01 1.17211282e+00 5.16659796e-01 3.00559551e-01
-1.31601691e+00 -4.31095332e-01 6.47872761e-02 4.17012930e-01
-1.56874347e+00 -2.99011469e-01 6.91074014e-01 -5.00554621e-01
8.52576137e-01 -1.13101110e-01 6.52334213e-01 1.05300915e+00
3.68456364e-01 6.38898075e-01 1.07640243e+00 -3.01271677e-01
4.60918218e-01 1.96881980e-01 -2.64995694e-01 9.20007348e-01
6.20938353e-02 2.37625584e-01 -6.32238567e-01 1.76113144e-01
1.04058230e+00 3.38809520e-01 -5.62885940e-01 -1.11276996e+00
-1.42616582e+00 9.22001481e-01 1.22270739e+00 -1.00569084e-01
-5.18999219e-01 4.47280794e-01 -6.87417910e-02 5.94009459e-02
2.03392029e-01 4.75402325e-01 -2.51603216e-01 -5.98324910e-02
-6.58792794e-01 3.09757553e-02 8.73045146e-01 1.11542308e+00
1.38304055e+00 -1.15077183e-01 3.43185991e-01 5.10948062e-01
4.93684173e-01 1.11645186e+00 5.71459651e-01 -1.00535536e+00
3.46618891e-01 5.47981441e-01 2.74271816e-01 -1.28144491e+00
-4.12950844e-01 -2.46223956e-01 -8.82003188e-01 2.38541633e-01
1.94182694e-01 9.79698896e-02 -1.12696314e+00 1.84328210e+00
-2.47368831e-02 1.93035737e-01 1.28085390e-02 1.21378291e+00
6.28287911e-01 7.72265613e-01 -4.43258017e-01 3.62327546e-01
9.94837582e-01 -9.42947447e-01 -3.50421637e-01 -7.31593788e-01
2.66610354e-01 -2.51586825e-01 7.36905336e-01 3.18690985e-01
-5.01864016e-01 -6.06482506e-01 -1.21648550e+00 -4.16607261e-02
-5.77877045e-01 9.85126123e-02 5.19285321e-01 2.84136504e-01
-1.46569359e+00 3.30141038e-01 -1.16678905e+00 -6.91459358e-01
4.75225188e-02 3.71043652e-01 -8.41825664e-01 -2.39296824e-01
-8.71913254e-01 1.21058190e+00 7.09825933e-01 2.08603427e-01
-1.40626180e+00 3.03385146e-02 -1.49182737e+00 -2.51934621e-02
3.72290105e-01 -6.61394000e-01 1.15806568e+00 -6.33640826e-01
-1.47627044e+00 5.99643826e-01 -3.86978805e-01 -5.86321712e-01
2.77661949e-01 -2.81861901e-01 -2.99569536e-02 7.14114234e-02
2.76546210e-01 1.11563611e+00 9.64054585e-01 -1.36288059e+00
-5.26867688e-01 -2.35668197e-01 3.01518232e-01 9.22102928e-02
-1.12777874e-02 -4.71620947e-01 -5.49722016e-01 -4.37699556e-02
5.00942409e-01 -1.25828648e+00 -4.50287670e-01 3.87545042e-02
-4.94302392e-01 2.38184735e-01 5.95991135e-01 -4.73664314e-01
5.47367096e-01 -1.98131657e+00 3.46200556e-01 3.24747533e-01
2.48817578e-01 -7.16637373e-02 -5.98503975e-03 2.76887804e-01
2.05326587e-01 -3.14523429e-01 -1.40738830e-01 -2.85155207e-01
4.28044423e-03 5.82195699e-01 -4.52691644e-01 6.93546355e-01
1.00380301e-01 9.98177826e-01 -1.08086669e+00 -1.43204480e-01
6.31247103e-01 5.43852627e-01 -4.45219636e-01 3.13331902e-01
-3.11607212e-01 7.64236510e-01 -3.58132422e-01 3.86238188e-01
3.85284424e-01 -4.71432537e-01 2.54235566e-02 -5.50899878e-02
-7.42601091e-03 3.01905274e-01 -9.97006416e-01 2.17094064e+00
-6.65501177e-01 9.08425748e-01 -2.69410253e-01 -9.28680182e-01
1.22001326e+00 -4.73836064e-02 9.58019048e-02 -6.48390651e-01
6.48934096e-02 1.16741426e-01 -4.07854825e-01 -5.00475168e-02
6.45655870e-01 1.92662567e-01 -2.83025891e-01 2.33327538e-01
2.72329748e-01 -4.94491458e-01 -1.73758529e-02 2.92219203e-02
1.23430085e+00 4.27215308e-01 3.46252650e-01 6.86660036e-02
4.37163234e-01 3.17291021e-01 4.22892183e-01 9.71118927e-01
1.26309633e-01 7.88419664e-01 8.55584741e-02 -6.59074903e-01
-9.93998826e-01 -1.30942869e+00 7.59762898e-02 7.04871178e-01
6.98593140e-01 -3.59316051e-01 -5.75527608e-01 -4.97244149e-01
-1.01622209e-01 4.41389769e-01 -5.88194251e-01 -1.80088177e-01
-4.12408173e-01 -2.38692537e-01 1.73821852e-01 7.01617062e-01
6.33538783e-01 -1.09545243e+00 -8.59739602e-01 3.59040052e-01
-2.03229681e-01 -1.26847458e+00 -1.85761124e-01 5.67430019e-01
-7.08954513e-01 -1.07874656e+00 -5.20775199e-01 -8.14728558e-01
1.07460225e+00 4.65530038e-01 1.02019572e+00 -3.02638322e-01
4.21171598e-02 4.96624976e-01 -2.63354957e-01 -3.99424136e-01
-3.49912703e-01 1.25289425e-01 2.05487818e-01 -1.04605809e-01
2.78639317e-01 -4.66884762e-01 -3.42141896e-01 5.88117003e-01
-6.89552188e-01 2.11167768e-01 7.00762808e-01 1.01786947e+00
8.26100826e-01 -1.23638585e-01 1.75996333e-01 -6.10446334e-01
1.46244809e-01 -2.02812746e-01 -1.01023233e+00 9.37778726e-02
-5.12463868e-01 4.50113177e-01 4.61365670e-01 -2.59756923e-01
-6.76088452e-01 6.56567276e-01 5.20148538e-02 -6.04644895e-01
-2.74225265e-01 7.46301055e-01 -9.99834239e-02 -3.05601954e-01
6.72003627e-01 5.09964347e-01 1.51324257e-01 -3.25253636e-01
4.95875895e-01 4.70095485e-01 1.03092384e+00 -3.70723695e-01
1.07810509e+00 5.02395630e-01 1.98999643e-02 -5.69477379e-01
-8.01844835e-01 -5.33420920e-01 -9.32248652e-01 -1.71716243e-01
8.13469291e-01 -1.23805439e+00 -6.18882298e-01 2.78870791e-01
-1.30913353e+00 -4.70045656e-01 2.61968393e-02 7.66466618e-01
-9.25803959e-01 -1.68587286e-02 -4.13181692e-01 -4.65218306e-01
2.00920254e-01 -1.30400729e+00 1.31657350e+00 2.92442977e-01
-6.04667887e-02 -1.01963139e+00 5.55668958e-02 -7.21021667e-02
3.73574555e-01 2.89566576e-01 4.52580035e-01 -5.37990391e-01
-1.19318807e+00 -4.46736753e-01 -2.64995873e-01 2.90558487e-01
1.37351573e-01 -6.21972978e-01 -9.33555961e-01 -3.79908323e-01
-8.89289379e-02 -2.88266271e-01 9.01692212e-01 2.75563419e-01
7.68517137e-01 -1.44268766e-01 -5.80057144e-01 7.48498321e-01
1.46626294e+00 1.31452709e-01 5.00556052e-01 5.27111828e-01
8.17438185e-01 1.35854632e-01 5.68145394e-01 2.44530350e-01
6.94247186e-01 7.59316385e-01 8.21697533e-01 -1.29298851e-01
2.81446189e-01 -7.73730099e-01 3.09334010e-01 7.08224535e-01
1.49183264e-02 -2.40218282e-01 -1.08438754e+00 3.97896588e-01
-2.18771386e+00 -6.17937088e-01 3.41177315e-01 2.18621445e+00
4.81150120e-01 1.48620859e-01 -3.24616581e-01 -1.96478963e-01
5.90613484e-01 1.08365022e-01 -6.01705670e-01 1.50443673e-01
2.14874130e-02 -1.09868839e-01 8.31232905e-01 5.32731652e-01
-1.20512414e+00 1.04947901e+00 6.02669048e+00 2.68544227e-01
-1.34178829e+00 6.24462627e-02 1.83497220e-01 3.54848474e-01
-2.30506901e-02 1.33077785e-01 -8.18420351e-01 2.77465314e-01
8.49563062e-01 -3.28889899e-02 6.10950053e-01 1.23849022e+00
2.44092569e-02 -3.45020622e-01 -1.47724450e+00 1.24130988e+00
4.56542790e-01 -1.22818148e+00 -2.48417839e-01 1.22495748e-01
5.49887121e-01 5.03403187e-01 -2.23257300e-02 3.87418032e-01
7.14574039e-01 -1.12287736e+00 9.50402319e-01 6.60088003e-01
6.06564581e-01 -5.15065849e-01 9.78975117e-01 7.69610882e-01
-1.03015316e+00 1.05259806e-01 -7.53787160e-01 -1.61047652e-01
2.28318706e-01 2.70772636e-01 -1.30633688e+00 4.14605647e-01
6.08242095e-01 1.03026545e+00 -7.06156731e-01 1.23691249e+00
-7.68160045e-01 2.28837207e-01 -4.68916684e-01 -1.28059611e-01
2.97447860e-01 -8.65650102e-02 3.09634537e-01 7.28787184e-01
5.33996820e-01 -3.73127401e-01 5.11581361e-01 1.02441907e+00
8.50842446e-02 -3.29800278e-01 -1.09063339e+00 2.41622657e-01
3.72629791e-01 1.03676271e+00 -6.52571738e-01 -2.58053482e-01
-3.42215121e-01 1.02498138e+00 5.15155256e-01 4.64331210e-01
-7.04002142e-01 -2.75885910e-01 3.94670129e-01 -1.84870362e-01
3.00874203e-01 -6.48254395e-01 1.94216028e-01 -1.22299969e+00
-7.90135935e-02 -4.97254729e-01 -2.64987826e-01 -1.26799107e+00
-8.14276218e-01 8.32035065e-01 2.01176088e-02 -1.35553908e+00
-8.11655819e-01 -8.38520348e-01 -1.84013456e-01 8.90951157e-01
-1.45965767e+00 -1.10417616e+00 -8.45367849e-01 4.76647049e-01
2.38211975e-01 -1.45500585e-01 9.40911770e-01 2.14121900e-02
-1.29187137e-01 5.81921116e-02 2.05002412e-01 4.20629740e-01
7.57470250e-01 -1.40135419e+00 4.53281999e-01 7.59238005e-01
6.78908169e-01 6.08983338e-01 7.39219487e-01 -4.70091730e-01
-1.53877747e+00 -1.25940490e+00 6.52694881e-01 -8.16959441e-01
5.53737819e-01 -5.85990667e-01 -7.62903988e-01 1.09836650e+00
-1.11434102e-01 2.10086986e-01 -1.15936045e-02 1.02479599e-01
-2.85505116e-01 -2.95451671e-01 -7.70792246e-01 2.44990304e-01
8.50809455e-01 -7.90594161e-01 -5.07571459e-01 2.24476889e-01
7.02488601e-01 -7.90369570e-01 -5.16698241e-01 2.82937646e-01
3.16303581e-01 -9.28857923e-01 9.81344759e-01 -9.72926319e-02
1.45064726e-01 -5.77609301e-01 -3.82702202e-01 -1.70431983e+00
-2.73482263e-01 -1.32354394e-01 1.08154252e-01 7.65342772e-01
4.23061848e-01 -6.96780562e-01 9.26004529e-01 4.86162186e-01
-1.30965218e-01 -2.06926197e-01 -8.78294528e-01 -9.91004229e-01
-4.58249718e-01 -4.43899661e-01 7.03887939e-01 5.83597898e-01
-1.32588089e-01 3.46461058e-01 -5.00969410e-01 4.95075166e-01
4.90525335e-01 7.29525983e-02 1.17846918e+00 -1.19859791e+00
-1.60145834e-01 -4.10183817e-02 -1.12491655e+00 -1.53500009e+00
4.00524884e-01 -8.99249852e-01 7.83626556e-01 -1.66485834e+00
1.01053901e-01 -3.49093735e-01 -2.00469434e-01 6.88282967e-01
1.78738788e-01 3.06783527e-01 1.69934243e-01 2.86138028e-01
-8.58909726e-01 5.41781425e-01 9.40252900e-01 -3.62906903e-01
-8.14489424e-02 -1.72560871e-01 -1.83396563e-01 8.16387236e-01
6.01429641e-01 -4.96735752e-01 -4.89912719e-01 -5.98104119e-01
3.27894270e-01 2.27392584e-01 7.99189866e-01 -1.33609235e+00
7.32012630e-01 -5.76650798e-02 4.95136023e-01 -6.78786039e-01
5.97953856e-01 -1.03183722e+00 3.91477317e-01 3.29360276e-01
-1.16603345e-01 1.97212994e-02 -8.11692402e-02 8.57111692e-01
-5.33674955e-01 -1.16481185e-01 3.62513393e-01 -1.68935090e-01
-1.20986974e+00 3.52469295e-01 -1.33803666e-01 -4.11018074e-01
7.99002647e-01 -1.27583042e-01 -2.90740848e-01 -7.26317227e-01
-6.18032455e-01 2.17361763e-01 7.84681618e-01 5.66642880e-01
7.56913960e-01 -1.39981794e+00 -3.58970314e-01 5.08856177e-01
4.83365864e-01 4.05375183e-01 -2.01796114e-01 6.09532714e-01
-7.45668352e-01 7.03445971e-01 -2.20165178e-01 -1.17422748e+00
-6.71908140e-01 7.53148377e-01 2.87514687e-01 8.34382400e-02
-6.62244499e-01 5.64010918e-01 4.94881004e-01 -6.44009650e-01
1.54296130e-01 -6.02484584e-01 9.02034417e-02 -4.06279981e-01
4.90374058e-01 -2.38417432e-01 -5.06345741e-02 -8.61705303e-01
-3.97408962e-01 5.21334887e-01 2.46509850e-01 -1.55870259e-01
1.21169829e+00 -2.90647775e-01 -4.61908579e-02 5.50833642e-01
1.05006933e+00 -3.55960220e-01 -1.55131757e+00 -6.93317771e-01
2.89384946e-02 -4.77598488e-01 2.10256670e-02 -6.22468174e-01
-8.96562934e-01 8.10579717e-01 4.71132904e-01 -7.41053447e-02
8.17153156e-01 1.33254245e-01 3.91413152e-01 1.03233254e+00
1.16046941e+00 -7.60838032e-01 4.99385335e-02 6.62925601e-01
8.31501603e-01 -1.64317048e+00 -2.13100940e-01 -1.25428826e-01
-5.29384315e-01 1.04341054e+00 5.79511285e-01 -2.24824011e-01
5.86731315e-01 9.66537818e-02 1.97245210e-01 -2.62455158e-02
-5.07181227e-01 -3.44986320e-01 2.85857975e-01 8.26419830e-01
-1.80502325e-01 1.29420698e-01 7.27429271e-01 3.35131079e-01
-4.52575147e-01 1.78876985e-02 3.66223752e-01 8.54873717e-01
-7.93595672e-01 -9.54780757e-01 -4.67123955e-01 2.00583756e-01
2.21135467e-01 1.92525610e-02 -2.26999298e-01 7.39693344e-01
-9.76036265e-02 7.68848479e-01 1.21716127e-01 -5.29462457e-01
2.17021465e-01 -1.96220979e-01 4.62866217e-01 -7.18300939e-01
1.90854132e-01 -2.97959298e-01 -1.93964720e-01 -7.78787851e-01
-3.95965129e-01 -3.89992386e-01 -1.07126737e+00 -1.17341191e-01
-2.65501708e-01 1.37395978e-01 1.08458400e+00 1.10120606e+00
1.87914833e-01 3.46021801e-01 4.34898853e-01 -1.34360468e+00
-2.55469114e-01 -8.72568011e-01 -5.53353369e-01 2.54075617e-01
6.02035046e-01 -8.93367112e-01 -1.39555246e-01 2.04495847e-01]
|
[7.5865302085876465, -2.0597851276397705]
|
8c5452a0-d94e-4367-a4bd-5a33db01f525
|
scaling-through-abstractions-high-performance
|
2004.10519
| null |
https://arxiv.org/abs/2004.10519v1
|
https://arxiv.org/pdf/2004.10519v1.pdf
|
Scaling through abstractions -- high-performance vectorial wave simulations for seismic inversion with Devito
|
[Devito] is an open-source Python project based on domain-specific language and compiler technology. Driven by the requirements of rapid HPC applications development in exploration seismology, the language and compiler have evolved significantly since inception. Sophisticated boundary conditions, tensor contractions, sparse operations and features such as staggered grids and sub-domains are all supported; operators of essentially arbitrary complexity can be generated. To accommodate this flexibility whilst ensuring performance, data dependency analysis is utilized to schedule loops and detect computational-properties such as parallelism. In this article, the generation and simulation of MPI-parallel propagators (along with their adjoints) for the pseudo-acoustic wave-equation in tilted transverse isotropic media and the elastic wave-equation are presented. Simulations are carried out on industry scale synthetic models in a HPC Cloud system and reach a performance of 28TFLOP/s, hence demonstrating Devito's suitability for production-grade seismic inversion problems.
|
['Rhodri Nelson', 'Philipp Witte', 'Felix J. Herrmann', 'Mathias Louboutin', 'Jan Thorbecke', 'Gerard Gorman', 'Fabio Luporini', 'George Bisbas']
|
2020-04-22
| null | null | null | null |
['seismic-inversion']
|
['miscellaneous']
|
[-1.10219263e-01 -3.89604092e-01 8.45727623e-01 2.39082752e-03
-4.20599073e-01 -2.92899340e-01 5.03069222e-01 -2.34683380e-01
-3.88252586e-01 6.94138944e-01 2.38883927e-01 -8.12560380e-01
-2.53871053e-01 -7.86550701e-01 -1.23110816e-01 -1.02017808e+00
-1.10490310e+00 5.08059561e-01 4.14873034e-01 -4.29820716e-01
4.92240310e-01 7.01666415e-01 -1.47298825e+00 3.29301417e-01
6.02920592e-01 6.81942642e-01 9.46886688e-02 1.06861401e+00
1.94623083e-01 9.66440022e-01 -3.88013944e-02 4.43697214e-01
2.07879812e-01 -1.91309676e-01 -1.06990325e+00 -2.68354475e-01
-3.68726701e-01 -2.55039692e-01 6.23693056e-02 4.80149984e-01
1.02516925e+00 1.74540579e-01 4.59058285e-01 -8.81106377e-01
5.61995447e-01 2.60028541e-01 -5.61137438e-01 3.96601707e-01
4.44222093e-01 3.43285590e-01 5.50189316e-01 -1.36982954e+00
5.89861631e-01 7.66487956e-01 1.30695450e+00 -1.46616712e-01
-1.64144790e+00 -3.85846794e-01 -1.05109167e+00 -3.12676132e-01
-1.51085198e+00 -5.93993008e-01 5.38807154e-01 -8.08791459e-01
1.40389979e+00 8.92901957e-01 8.88405681e-01 5.09972274e-01
7.59666145e-01 -5.94745390e-03 1.22301388e+00 -4.15218860e-01
7.29072154e-01 -4.11158830e-01 -2.63230443e-01 2.41716608e-01
1.78132877e-01 3.45303535e-01 -7.30668604e-01 -9.39389706e-01
1.00545895e+00 -8.18246663e-01 -1.53095186e-01 -2.90553331e-01
-1.43996167e+00 8.88902783e-01 -3.45692895e-02 2.80077577e-01
-6.93889558e-01 1.94729745e-01 9.93331373e-01 2.99005032e-01
6.86008930e-01 7.20860302e-01 -3.97135913e-01 -5.00880003e-01
-1.41402137e+00 9.57433224e-01 1.18706691e+00 6.40605807e-01
6.92363203e-01 6.10144079e-01 2.84144819e-01 5.86257041e-01
4.18072999e-01 4.65654641e-01 2.47086644e-01 -1.19246423e+00
1.75896242e-01 -5.94652407e-02 1.08790025e-01 -9.43532348e-01
-8.72488558e-01 -4.77517724e-01 -9.53811109e-01 5.07395446e-01
6.88826339e-03 -6.36711657e-01 -6.12668812e-01 9.00147021e-01
6.79070830e-01 1.92927808e-01 2.97523350e-01 1.00986660e+00
5.43229282e-01 7.79727042e-01 5.83289154e-02 -1.08931594e-01
1.33254826e+00 -3.80714148e-01 -1.94625005e-01 -1.46117121e-01
1.31580770e+00 -1.22589386e+00 4.13809985e-01 2.66777873e-01
-1.27539670e+00 -1.70837641e-01 -6.09631777e-01 1.33287966e-01
1.16506852e-01 -6.93067968e-01 8.56108367e-01 4.70117599e-01
-1.38290226e+00 6.50566936e-01 -1.23059475e+00 7.62806684e-02
-2.90164262e-01 2.31492668e-01 -1.93466753e-01 3.00454766e-01
-1.02686048e+00 5.45417130e-01 1.90084621e-01 2.76583344e-01
-5.39691806e-01 -1.15214705e+00 -6.88659906e-01 -1.90584653e-03
-5.16257346e-01 -8.66160333e-01 1.07287312e+00 -4.12453800e-01
-1.25883663e+00 7.04378128e-01 1.63495645e-01 -2.78892249e-01
7.16235042e-01 3.24748576e-01 -2.68819958e-01 5.62283807e-02
4.65300411e-01 -5.11328541e-02 4.01858002e-01 -6.93387210e-01
-2.52398938e-01 1.88388884e-01 -4.98734772e-01 1.85889155e-01
2.47942403e-01 4.77789402e-01 2.99568176e-01 -6.89974606e-01
4.87831801e-01 -1.02964473e+00 -5.30407310e-01 -4.72355843e-01
-2.60787845e-01 4.83969808e-01 6.52992547e-01 -9.00827646e-01
1.02797365e+00 -2.19920826e+00 7.83400238e-02 6.34925187e-01
-1.37107456e-02 -1.72365278e-01 3.86098027e-01 1.19004679e+00
-2.92979211e-01 -2.08120689e-01 -7.97093987e-01 1.44330710e-01
-1.03167638e-01 1.42383099e-01 -2.95138955e-01 6.65943503e-01
-2.26193503e-03 1.56069947e-02 -8.18524599e-01 -4.74809647e-01
-1.15438245e-01 4.41991538e-01 -1.04591537e+00 1.42396882e-01
2.36596972e-01 8.44450295e-01 -6.62835658e-01 5.44120252e-01
1.18472958e+00 1.25161827e-01 6.29996359e-02 2.73126066e-01
-1.22203922e+00 5.04464805e-01 -1.50688148e+00 1.42476141e+00
-7.20809996e-01 5.24429619e-01 1.01816583e+00 -9.23697650e-01
8.41283441e-01 6.63319588e-01 6.22607052e-01 -3.79190952e-01
-3.28094780e-01 9.16826308e-01 4.01295573e-01 -9.09927130e-01
7.23322988e-01 -3.49355638e-01 7.74711296e-02 5.11644959e-01
-4.57299948e-01 -7.65825808e-01 3.35169613e-01 9.64221060e-02
1.43739569e+00 2.60831475e-01 -2.36235365e-01 -1.43971658e+00
3.63827169e-01 5.97407758e-01 3.99816602e-01 4.53291416e-01
3.72716427e-01 5.61335564e-01 5.22543073e-01 -6.52795494e-01
-1.51265109e+00 -6.43325329e-01 -8.72519672e-01 1.11618137e+00
-4.82915044e-01 -2.77225405e-01 -4.04915214e-01 8.07842016e-01
-6.98775947e-02 4.43739057e-01 -2.24027440e-01 5.05123317e-01
-9.19275224e-01 -1.24484980e+00 8.22399020e-01 1.92849830e-01
5.12283087e-01 -1.13784063e+00 -1.20447874e+00 8.54311466e-01
1.70184523e-01 -8.60473692e-01 3.92780840e-01 3.15926552e-01
-1.12191272e+00 -7.14202225e-01 -6.03096783e-01 -6.87292993e-01
3.77734572e-01 -2.53654033e-01 1.19303429e+00 7.86228254e-02
-4.50845152e-01 7.23431259e-02 -7.62482733e-02 -1.00777782e-01
-5.40141702e-01 -1.09839447e-01 -9.18727964e-02 -2.13066339e-01
-4.05861586e-01 -1.13811958e+00 -6.72795832e-01 1.34681702e-01
-9.93551433e-01 5.13047159e-01 1.51822910e-01 8.24707270e-01
3.54402177e-02 -2.04251036e-02 2.42371708e-01 -6.68550551e-01
6.30226195e-01 -7.75382936e-01 -6.29587829e-01 -5.43909431e-01
-1.50511369e-01 2.72663981e-02 3.36275727e-01 7.28353113e-02
-1.23212397e+00 -1.75831735e-01 -6.21427357e-01 3.46280664e-01
5.65916523e-02 1.25249374e+00 6.04577959e-01 -6.05642915e-01
8.35720241e-01 2.47600049e-01 5.20832166e-02 -4.19373989e-01
-4.30996746e-01 4.81492430e-01 4.03617024e-01 -1.13433647e+00
4.92084891e-01 5.30618072e-01 4.91275132e-01 -1.47925091e+00
2.24697724e-01 -5.43395460e-01 -1.05155468e-01 -3.17068964e-01
6.37171209e-01 -8.59344244e-01 -5.19621372e-01 8.56777906e-01
-1.21366870e+00 -6.50708616e-01 3.41264009e-02 7.50911951e-01
-3.43640745e-01 2.42102548e-01 -8.60283494e-01 -9.05818403e-01
-3.64723831e-01 -1.18757284e+00 9.25841153e-01 -2.40919441e-01
-4.29048359e-01 -1.27392030e+00 5.59094787e-01 -2.10348472e-01
1.22192299e+00 8.45412791e-01 5.79399049e-01 -1.18539423e-01
-3.78720284e-01 2.01825667e-02 -2.23971307e-01 -2.48359844e-01
-6.54291987e-01 1.83960125e-01 -8.97163987e-01 -4.53367323e-01
5.05167365e-01 3.43000628e-02 2.27852896e-01 4.00297344e-01
6.89685822e-01 -2.49291614e-01 -1.64607167e-01 1.00314701e+00
1.46365488e+00 -3.80376458e-01 6.27862930e-01 4.95995671e-01
3.82731855e-01 5.14631450e-01 1.35272294e-01 1.12686741e+00
-1.26616478e-01 4.35842693e-01 1.30904123e-01 -9.00380462e-02
4.33977634e-01 5.81194997e-01 3.75594385e-02 1.07502770e+00
-8.21142495e-01 3.95677865e-01 -1.73233092e+00 4.92525756e-01
-1.31266773e+00 -8.95679295e-01 -1.16351497e+00 1.95911753e+00
8.79899263e-01 9.67533141e-02 -2.22837344e-01 4.06889170e-01
2.59015769e-01 1.36090696e-01 1.38131693e-01 -9.15816247e-01
1.18136786e-01 6.85163438e-01 8.93986344e-01 5.56836486e-01
-8.35932195e-01 2.96733171e-01 6.40667105e+00 3.18209171e-01
-1.48959529e+00 1.86210349e-01 6.74867406e-02 3.19076926e-01
-3.85836273e-01 3.46722454e-01 -1.81647539e-01 3.27693641e-01
1.22980762e+00 -2.08964020e-01 2.68604755e-01 5.88658631e-01
6.54369712e-01 -4.45239723e-01 -4.03118193e-01 2.68035620e-01
-7.08657861e-01 -1.65918207e+00 -8.19581270e-01 1.40409395e-01
6.94072843e-01 6.91113949e-01 -4.73786354e-01 5.45196496e-02
6.40516728e-02 -7.55796731e-01 7.36400723e-01 3.55787009e-01
7.59150684e-01 -7.18030572e-01 7.55227983e-01 3.25318068e-01
-1.20429969e+00 3.87208760e-01 -1.11925945e-01 -7.85220206e-01
5.17419755e-01 9.12443399e-01 -7.82942295e-01 7.81629205e-01
8.79056334e-01 2.58982867e-01 2.70890146e-02 1.22051203e+00
4.24126983e-01 1.03398144e+00 -6.75845623e-01 2.13998228e-01
5.61421990e-01 -2.97227293e-01 9.10260439e-01 1.63469291e+00
6.96449697e-01 3.48652184e-01 4.94385548e-02 5.66469908e-01
8.33391547e-01 6.47730380e-02 -4.25348401e-01 5.47754169e-01
2.04769686e-01 1.17698586e+00 -6.91908538e-01 -6.00784980e-02
-1.80314213e-01 2.81570226e-01 -3.63749385e-01 4.86864150e-01
-6.91228449e-01 -5.54493487e-01 7.30268955e-01 6.12713575e-01
1.27751946e-01 -8.03839028e-01 -7.08957314e-01 -6.79414690e-01
-2.10318178e-01 -5.68389177e-01 7.79189989e-02 -7.61418641e-01
-7.44556069e-01 5.93602359e-01 2.65488595e-01 -1.07176125e+00
-4.58254486e-01 -5.77007532e-01 -9.97166932e-01 1.33267796e+00
-1.10007095e+00 -7.87211180e-01 -2.07722355e-02 3.28135997e-01
-9.85976607e-02 5.10325357e-02 1.07596695e+00 4.38662589e-01
-2.03192197e-02 -3.47310394e-01 4.90064919e-01 2.98336893e-02
1.26778930e-01 -8.48529458e-01 7.33372271e-01 9.14539635e-01
-1.00165677e+00 6.37306333e-01 1.37605906e+00 -8.81266713e-01
-1.80088699e+00 -6.43232644e-01 9.26862717e-01 3.57961833e-01
1.26512706e+00 -2.07935467e-01 -1.23033309e+00 3.54816139e-01
2.93676436e-01 2.81710446e-01 4.25234318e-01 1.35064395e-02
1.77609384e-01 1.92762956e-01 -1.01026309e+00 2.42626175e-01
5.94403088e-01 -2.28197008e-01 -1.97829857e-01 6.16289675e-01
-6.00691587e-02 -9.83381331e-01 -1.21925163e+00 5.86816370e-01
2.30118603e-01 -1.36345172e+00 9.22965705e-01 -1.19676284e-01
4.38955516e-01 -2.08577886e-01 1.69108942e-01 -1.04162323e+00
-2.52601981e-01 -1.20917439e+00 5.04829705e-01 7.04489410e-01
1.60767227e-01 -1.10780120e+00 4.76006150e-01 6.91300631e-01
-7.99584568e-01 -4.80275691e-01 -1.51985836e+00 -5.65931976e-01
2.05948547e-01 -6.76512063e-01 2.80218661e-01 1.03967714e+00
2.58926302e-01 -4.04057279e-02 -1.97564542e-01 4.64348674e-01
7.82250524e-01 -1.07883856e-01 5.22449613e-01 -1.13518822e+00
-3.86289597e-01 -2.36944422e-01 -4.15450364e-01 -2.69616425e-01
-2.74425924e-01 -6.92713499e-01 1.27056107e-01 -1.16310585e+00
-4.02192414e-01 -1.10941803e+00 4.87147093e-01 2.71793067e-01
4.26888883e-01 4.21792448e-01 -5.57513833e-01 3.40929180e-01
4.96422499e-01 -1.22200390e-02 8.36066067e-01 5.49422681e-01
-1.36222839e-01 -1.91161022e-01 3.42326909e-01 5.82181692e-01
6.85358524e-01 -5.71973324e-01 1.26796756e-02 -6.68727756e-01
7.21839726e-01 5.02550125e-01 7.60753870e-01 -1.31876838e+00
3.53263021e-01 -1.19473353e-01 -1.43653333e-01 -3.60051930e-01
1.37720183e-01 -3.16561997e-01 9.29520190e-01 7.84214139e-01
9.02185962e-02 4.92855400e-01 4.40305561e-01 -1.98961288e-01
-3.21866155e-01 -2.15253010e-01 7.90876627e-01 -1.74725860e-01
-6.06788337e-01 -1.92604348e-01 -9.38329458e-01 2.89349765e-01
7.70713389e-01 -1.67771101e-01 1.41847998e-01 8.63502026e-02
-6.22254789e-01 7.32105877e-03 4.99342918e-01 -6.18009567e-01
3.62070233e-01 -1.00462496e+00 -1.38358045e+00 6.52161121e-01
-2.23617554e-01 3.30107838e-01 5.29297650e-01 1.20568335e+00
-1.89861584e+00 9.88021214e-03 -3.15937042e-01 -7.64855683e-01
-8.89145613e-01 -1.99397355e-01 4.15351003e-01 -2.05718994e-01
-8.55842054e-01 9.78174210e-01 -1.26977831e-01 -4.09452975e-01
-7.03481257e-01 -2.31776938e-01 4.48418498e-01 -2.21407816e-01
4.42053169e-01 6.19962275e-01 5.05550623e-01 -4.65203792e-01
-3.70393336e-01 2.73952156e-01 7.42349863e-01 -5.65819860e-01
1.71881866e+00 2.42999226e-01 -6.68980122e-01 2.64231533e-01
1.04557431e+00 1.35230392e-01 -1.03882647e+00 3.26380461e-01
5.92725649e-02 -3.41026664e-01 3.86232048e-01 -1.38391346e-01
-6.23760581e-01 7.53750324e-01 -8.67812708e-03 6.60283506e-01
8.50222111e-01 -3.56048018e-01 6.22018993e-01 -6.88620955e-02
3.82193774e-01 -9.80451584e-01 -7.86693215e-01 8.78001213e-01
1.08608007e+00 -5.39720476e-01 4.12189811e-01 -4.15410876e-01
-2.04029649e-01 1.36899781e+00 2.42946334e-02 -1.99551031e-01
7.36731589e-01 1.10011649e+00 7.81737193e-02 -3.42358381e-01
-5.38610578e-01 4.21387732e-01 -3.37025940e-01 1.86297387e-01
9.42892611e-01 -1.23379985e-02 -4.99434829e-01 -3.74550551e-01
-4.00893688e-01 -5.68306781e-02 6.22842968e-01 1.52681339e+00
-2.52098054e-01 -9.62917984e-01 -1.00334418e+00 2.43590057e-01
-5.26573122e-01 -4.54421610e-01 6.66440189e-01 7.74770319e-01
-1.84222683e-01 4.10459816e-01 2.01722056e-01 2.39664793e-01
-1.15278838e-02 3.65602667e-03 -3.67827229e-02 -4.88067657e-01
-9.29455221e-01 2.14459985e-01 7.89314866e-01 -1.80768952e-01
-3.46977264e-01 -9.84761953e-01 -1.56053960e+00 -7.26425946e-01
3.60994227e-02 7.94928670e-01 8.45850825e-01 7.18662620e-01
4.23073322e-01 6.54064298e-01 4.95199502e-01 -1.71759188e+00
-3.40771466e-01 -1.06336498e+00 -8.79006326e-01 -1.31666198e-01
2.15611923e-02 -4.34946567e-01 -5.83007336e-01 4.83782291e-02]
|
[6.521066188812256, 3.148137331008911]
|
e0161ddf-5a71-4ce8-9bdc-215d9e2e0e4d
|
universal-adversarial-perturbation-for-text
|
1910.04618
| null |
https://arxiv.org/abs/1910.04618v1
|
https://arxiv.org/pdf/1910.04618v1.pdf
|
Universal Adversarial Perturbation for Text Classification
|
Given a state-of-the-art deep neural network text classifier, we show the existence of a universal and very small perturbation vector (in the embedding space) that causes natural text to be misclassified with high probability. Unlike images on which a single fixed-size adversarial perturbation can be found, text is of variable length, so we define the "universality" as "token-agnostic", where a single perturbation is applied to each token, resulting in different perturbations of flexible sizes at the sequence level. We propose an algorithm to compute universal adversarial perturbations, and show that the state-of-the-art deep neural networks are highly vulnerable to them, even though they keep the neighborhood of tokens mostly preserved. We also show how to use these adversarial perturbations to generate adversarial text samples. The surprising existence of universal "token-agnostic" adversarial perturbations may reveal important properties of a text classifier.
|
['Hang Gao', 'Tim Oates']
|
2019-10-10
| null | null | null | null |
['adversarial-text']
|
['adversarial']
|
[ 6.02153301e-01 2.27531835e-01 2.24989668e-01 -2.59274542e-01
-4.80877429e-01 -1.05764341e+00 8.14672410e-01 3.32133770e-02
-3.60964984e-01 6.29203618e-01 6.85878769e-02 -3.96611542e-01
4.88987356e-01 -1.10481071e+00 -1.36639571e+00 -9.25158799e-01
5.91947176e-02 4.89556849e-01 1.27403125e-01 -4.83963102e-01
7.51297846e-02 6.08778536e-01 -1.34950113e+00 5.31086206e-01
6.14980936e-01 4.54511821e-01 -3.98754716e-01 1.12794733e+00
-2.93795675e-01 7.00226545e-01 -1.19949925e+00 -9.47560608e-01
6.49235129e-01 -4.40744072e-01 -9.51522291e-01 -9.36003998e-02
8.09397936e-01 -4.27834362e-01 -8.75338733e-01 1.43233919e+00
5.79826057e-01 2.06622496e-01 8.98418605e-01 -1.41389656e+00
-1.37501276e+00 1.16438186e+00 -5.25012761e-02 -3.49113792e-02
2.14378759e-01 4.07784104e-01 7.93070078e-01 -5.95282018e-01
6.36531234e-01 1.36518300e+00 7.83304036e-01 9.76234376e-01
-1.20482635e+00 -6.65343165e-01 2.04151139e-01 -8.60198289e-02
-1.13949418e+00 -2.40431890e-01 6.10833406e-01 -5.05926073e-01
7.17497110e-01 4.63201225e-01 2.94024855e-01 1.73288739e+00
6.91563368e-01 7.59045959e-01 7.06405699e-01 -5.92706025e-01
3.34544659e-01 8.85571688e-02 -1.35180131e-01 6.77995801e-01
3.77585888e-01 9.06891897e-02 -1.74251676e-01 -2.48131007e-01
2.72956520e-01 1.76779404e-01 -2.44028479e-01 -8.69326517e-02
-1.19037604e+00 9.74004686e-01 3.73889893e-01 3.11588913e-01
4.84087206e-02 5.30347109e-01 9.18793440e-01 7.77153730e-01
4.19815063e-01 6.26086533e-01 -3.91911656e-01 6.21249303e-02
-4.37300354e-01 5.47248662e-01 8.25798869e-01 9.04797852e-01
6.70094013e-01 4.47422981e-01 -5.52365601e-01 6.77454054e-01
-4.41443861e-01 7.03031600e-01 9.38799560e-01 -4.40271884e-01
3.88590693e-01 2.63631672e-01 -5.09315841e-02 -8.63880277e-01
-9.33634117e-02 -1.97637320e-01 -1.11288416e+00 5.91770053e-01
6.35360479e-01 -3.27954680e-01 -1.16347551e+00 1.86823285e+00
2.28767954e-02 -2.17604294e-01 1.48421854e-01 3.73628855e-01
3.99996847e-01 4.27574635e-01 -1.26592770e-01 2.93377519e-01
1.11934197e+00 -5.97383380e-01 -4.91793185e-01 -2.57910788e-01
7.25820720e-01 -7.44074762e-01 1.36949217e+00 1.63588211e-01
-8.95592928e-01 -2.40585595e-01 -1.14914405e+00 1.44285977e-01
-1.05177784e+00 -4.83631790e-01 2.66010135e-01 1.08983922e+00
-8.52973640e-01 8.84007990e-01 -4.36350584e-01 -1.83920667e-01
3.81347567e-01 1.35901466e-01 -4.46874559e-01 -1.40153393e-01
-1.56255293e+00 9.35710669e-01 4.45415109e-01 -2.20710710e-01
-1.15502465e+00 -8.08545530e-01 -7.40700543e-01 -2.70794779e-02
-2.92128213e-02 -6.68424249e-01 1.19241941e+00 -1.57864392e+00
-1.32880604e+00 9.83298182e-01 -2.08696760e-02 -5.78110635e-01
1.21142352e+00 1.09710470e-01 -3.25025856e-01 1.99983731e-01
-8.68590996e-02 4.45667565e-01 1.51594603e+00 -1.22404921e+00
-2.46766269e-01 -2.36380234e-01 1.94938272e-01 -1.21439017e-01
-6.79258287e-01 1.01829633e-01 1.73267126e-01 -1.28310812e+00
-3.57472926e-01 -9.81014371e-01 -2.21993521e-01 2.45102718e-01
-8.06845844e-01 6.46811277e-02 9.91677046e-01 -2.86420047e-01
7.21738160e-01 -1.94927037e+00 -2.05627270e-02 -2.96079498e-02
3.44521940e-01 2.94585586e-01 -4.01039958e-01 4.04140830e-01
-5.92150390e-01 6.26032293e-01 -4.76974159e-01 -1.79969788e-01
3.35260332e-01 2.46960372e-01 -6.89331114e-01 7.49715149e-01
1.15224868e-01 1.06650293e+00 -9.45821762e-01 -8.34821686e-02
7.33026862e-02 2.11475834e-01 -5.01604557e-01 1.81199834e-02
-5.40727973e-01 -3.48042190e-01 -1.96060061e-01 4.02080327e-01
8.51494789e-01 2.62532920e-01 -2.20445588e-01 1.13389939e-01
5.66818774e-01 -1.40352741e-01 -8.85835111e-01 1.04075956e+00
-2.23144859e-01 1.02417350e+00 -2.87618458e-01 -1.10419846e+00
5.71242988e-01 2.52702773e-01 4.50251959e-02 -1.30260870e-01
2.62458354e-01 3.04277204e-02 -3.90824750e-02 -3.11773121e-01
6.82206571e-01 -2.90287554e-01 -1.91084921e-01 8.07640672e-01
1.02239326e-01 -2.86074877e-02 5.98280989e-02 4.07044113e-01
1.51570427e+00 -5.48286974e-01 -8.48693624e-02 -2.21958786e-01
4.08740461e-01 -1.98503971e-01 6.54679313e-02 1.35734344e+00
-3.62659633e-01 7.25041330e-01 6.34914994e-01 -6.51942074e-01
-1.55784297e+00 -1.08586466e+00 -1.69021085e-01 1.27780640e+00
-2.21052542e-01 -1.26168251e-01 -1.08160412e+00 -1.02489972e+00
3.80146414e-01 4.86384839e-01 -1.21470535e+00 -7.33146071e-01
-4.46546674e-01 -7.80875444e-01 1.28740370e+00 5.00143528e-01
4.37541753e-01 -1.26214814e+00 9.38472804e-03 2.12203581e-02
1.21763110e-01 -1.00137973e+00 -8.04280460e-01 4.28436935e-01
-4.58316207e-01 -9.69186544e-01 -7.77538419e-01 -6.96314991e-01
9.49328065e-01 -5.85758463e-02 1.10602093e+00 1.74653307e-01
-3.12566847e-01 3.01611006e-01 -4.34153467e-01 -5.52339137e-01
-1.27568030e+00 3.81991565e-02 1.86025068e-01 -5.47977872e-02
2.73584127e-01 -5.01366735e-01 -2.10239276e-01 -3.78778111e-03
-1.45002079e+00 -5.85105538e-01 3.41730058e-01 1.06944621e+00
-1.15227199e-03 3.62955242e-01 4.13987249e-01 -1.04158223e+00
9.14879858e-01 -4.08327907e-01 -2.97046304e-01 1.00173183e-01
-2.28261307e-01 1.68935791e-01 1.64071918e+00 -1.01388657e+00
-5.51904976e-01 -9.62594226e-02 -2.29967922e-01 -6.73262894e-01
-3.49466503e-01 1.35091497e-02 -2.82965630e-01 -5.01968265e-01
1.30856562e+00 5.25287628e-01 -8.77804905e-02 -2.41793524e-02
7.54424214e-01 4.52662468e-01 6.36099100e-01 -8.13535810e-01
1.39341903e+00 5.38396657e-01 8.52942914e-02 -8.12706709e-01
-5.16607761e-01 3.18646014e-01 -5.64792514e-01 -1.67147741e-02
5.69406092e-01 -5.12473702e-01 -7.23861516e-01 1.00059581e+00
-1.24057949e+00 -4.72367942e-01 -7.21687257e-01 -1.09791309e-01
-5.44487655e-01 5.79106271e-01 -7.47435689e-01 -5.01322508e-01
-4.46928084e-01 -1.05912960e+00 6.72814369e-01 -3.67217779e-01
-1.29397050e-01 -1.05763590e+00 -7.34259188e-02 -2.61279643e-01
2.25958019e-01 3.56982350e-01 1.21037352e+00 -1.20118797e+00
-3.17784280e-01 -5.57372987e-01 8.32273290e-02 8.10083449e-01
7.06508309e-02 3.49548608e-01 -9.92791057e-01 -3.69028956e-01
-1.68966591e-01 -2.94429988e-01 8.09342325e-01 1.34905249e-01
1.35327327e+00 -9.73419845e-01 -9.81975347e-02 8.76493275e-01
1.34332907e+00 2.41286941e-02 7.80483842e-01 4.75912720e-01
9.11495149e-01 2.15441048e-01 -1.57968611e-01 2.73816198e-01
-3.50256592e-01 1.32678717e-01 5.56963205e-01 3.17744538e-02
1.31256148e-01 -2.22177818e-01 6.44003272e-01 3.20670247e-01
4.87953961e-01 -6.61249220e-01 -7.31783032e-01 6.82252944e-01
-1.54591382e+00 -1.27551079e+00 2.97878869e-02 2.03072143e+00
1.15185416e+00 4.36252296e-01 -5.16513437e-02 4.87068862e-01
8.73249650e-01 3.66802305e-01 -6.92859888e-01 -8.13315034e-01
-4.24529374e-01 5.18408775e-01 9.40360069e-01 4.54204172e-01
-1.17555928e+00 1.11613500e+00 7.53816271e+00 1.04569626e+00
-1.19519210e+00 -6.88214153e-02 4.87506598e-01 -1.61058918e-01
-4.97012109e-01 -5.35064638e-01 -5.18643916e-01 6.86918557e-01
1.04036641e+00 -6.60136402e-01 4.80458289e-01 1.05589676e+00
-2.26082489e-01 5.85626125e-01 -1.26145101e+00 5.35818875e-01
1.10085353e-01 -1.62136984e+00 8.48355353e-01 -1.27230510e-01
9.68873858e-01 -5.42880818e-02 3.08463663e-01 2.26369247e-01
1.05344117e+00 -1.19512546e+00 7.90730536e-01 2.07540825e-01
8.71417522e-01 -1.12402296e+00 5.18064678e-01 1.06346324e-01
-8.07500124e-01 -5.90314418e-02 -7.99143612e-01 2.87481934e-01
-5.41171372e-01 7.09708393e-01 -7.39520192e-01 1.42210484e-01
5.35315692e-01 2.76802540e-01 -6.08465195e-01 3.25487286e-01
-8.51491019e-02 3.29484969e-01 -8.18543807e-02 -1.75977558e-01
3.56752455e-01 1.04456924e-01 6.83360457e-01 1.43240893e+00
1.66812271e-01 -2.63114303e-01 -3.83075178e-02 9.41258490e-01
-8.61274004e-01 -1.01528972e-01 -1.14504337e+00 -1.56612098e-01
4.10391659e-01 9.16264355e-01 -4.60405171e-01 -5.28582513e-01
7.47494400e-02 1.53975379e+00 2.62313634e-01 4.91530091e-01
-6.16904140e-01 -9.65503693e-01 1.16832030e+00 -2.41859376e-01
3.93328637e-01 1.63661227e-01 -4.02670085e-01 -1.23734188e+00
1.89688683e-01 -1.34991860e+00 1.14988498e-01 -6.74166143e-01
-1.59585345e+00 6.16862297e-01 -5.20553708e-01 -9.88105714e-01
-2.51730770e-01 -9.08538997e-01 -8.61648619e-01 9.59257007e-01
-8.59149098e-01 -1.04081082e+00 -1.13686316e-01 9.59993720e-01
4.22172129e-01 -5.61032116e-01 9.65522349e-01 -1.80005372e-01
-4.60861117e-01 1.50164473e+00 7.08993435e-01 8.15083921e-01
7.10452616e-01 -1.45896077e+00 1.06308734e+00 1.16778541e+00
-2.04342842e-01 5.81495166e-01 1.12102008e+00 -5.83823144e-01
-1.38488305e+00 -1.67025304e+00 6.60228193e-01 -7.12165654e-01
9.45736945e-01 -7.60981917e-01 -1.06710684e+00 1.22143269e+00
8.28243047e-02 2.55090773e-01 3.44291955e-01 -3.00895482e-01
-1.00174284e+00 7.50200823e-02 -1.64962685e+00 1.26121044e+00
9.76699173e-01 -9.96798575e-01 -5.50892174e-01 7.65551448e-01
1.23901606e+00 -4.04169768e-01 -5.95970571e-01 -7.13310465e-02
5.57475567e-01 -7.99745083e-01 9.91555035e-01 -1.15290403e+00
8.67723763e-01 3.73000167e-02 -1.12585500e-01 -1.66988218e+00
-1.77484825e-01 -8.62372577e-01 -4.08456996e-02 9.03825879e-01
2.28601098e-01 -9.91123915e-01 9.24254417e-01 2.26449877e-01
-1.32122502e-01 -4.11388755e-01 -8.99261892e-01 -1.21179354e+00
9.77264225e-01 -4.25618738e-01 6.89835787e-01 1.32868803e+00
-6.27863780e-02 -2.18799323e-01 -2.65113145e-01 1.72660068e-01
5.93112350e-01 -5.75613976e-01 7.71257102e-01 -8.96605968e-01
-7.90324882e-02 -6.52088702e-01 -7.25495279e-01 -6.83904648e-01
6.77972555e-01 -9.93407249e-01 2.25866571e-01 -6.76972151e-01
3.17605436e-02 -1.20255493e-01 -8.71364251e-02 4.80574906e-01
-4.32885021e-01 3.12198818e-01 9.80014279e-02 -1.27012879e-01
8.58793408e-02 4.29634243e-01 9.46477354e-01 -6.82444632e-01
3.96756440e-01 -1.70417219e-01 -7.96528220e-01 6.52402699e-01
8.31637919e-01 -8.64675701e-01 -1.59140155e-01 -3.29957962e-01
2.79861420e-01 -5.48826039e-01 3.08020741e-01 -8.12056601e-01
1.73195153e-02 -1.55639410e-01 5.39294422e-01 -2.27667034e-01
4.66869620e-04 -6.13554299e-01 -2.53119379e-01 5.38438201e-01
-7.42644966e-01 1.65151000e-01 3.22709024e-01 4.95179594e-01
1.74147308e-01 -6.22608185e-01 1.18884850e+00 -4.00033087e-01
-4.53378186e-02 5.00817895e-01 -8.67671847e-01 3.26142013e-01
9.35492635e-01 -1.22288270e-02 -6.84540749e-01 -3.79855275e-01
-4.23479915e-01 -3.41242224e-01 7.66291678e-01 2.80310899e-01
4.12836224e-01 -1.25393593e+00 -8.62551749e-01 2.17510805e-01
1.32403597e-01 -2.95958012e-01 1.74803257e-01 -1.95111230e-01
-6.72512531e-01 1.15732588e-01 -2.50461400e-01 -2.85974890e-01
-1.04996037e+00 1.05136836e+00 8.59753549e-01 -1.14258260e-01
-7.12228119e-01 9.58825648e-01 1.77363276e-01 -5.56513846e-01
2.91607559e-01 -2.30521068e-01 5.79812646e-01 -1.89132810e-01
5.99538743e-01 1.85966007e-02 3.42702836e-01 -2.74124146e-01
-2.35099383e-02 3.03555578e-01 -4.09005553e-01 -1.80291831e-02
7.37241149e-01 3.20220858e-01 -6.62693381e-02 2.38045469e-01
1.56030595e+00 9.29155871e-02 -9.01825607e-01 -1.69092163e-01
-4.58832920e-01 -4.88063037e-01 -3.07171911e-01 -5.93646348e-01
-9.73714650e-01 8.69910836e-01 3.87484998e-01 5.54293215e-01
8.35480869e-01 -4.96931136e-01 8.05667400e-01 9.82657790e-01
-1.95026919e-01 -9.22761142e-01 1.97228655e-01 9.08824623e-01
9.65481877e-01 -1.00095117e+00 -2.32011154e-01 1.04371712e-01
-4.10880506e-01 1.12865758e+00 4.80086029e-01 -2.54268169e-01
4.15798903e-01 5.52734375e-01 1.60124332e-01 2.53415316e-01
-8.12879801e-01 1.83300361e-01 -1.44035161e-01 1.07160914e+00
3.15823369e-02 1.09936623e-02 1.08895138e-01 8.73154551e-02
-6.36627376e-01 -4.30212349e-01 1.06361532e+00 8.86472344e-01
-3.53378862e-01 -1.09070122e+00 -7.04382598e-01 5.09101391e-01
-7.99058199e-01 -4.60774928e-01 -6.34232044e-01 7.82894909e-01
-5.78268431e-02 6.40223145e-01 1.07238062e-01 -6.06897414e-01
1.94485724e-01 4.11511362e-01 4.66610372e-01 -3.07965845e-01
-1.02271795e+00 -9.37272191e-01 -2.19535545e-01 -1.74470067e-01
3.21244687e-01 -3.77747983e-01 -9.89674866e-01 -1.15976286e+00
1.53502434e-01 -1.02027126e-01 5.81316173e-01 9.35655057e-01
7.78682977e-02 5.48894882e-01 1.02074265e+00 -7.30723739e-01
-1.01368737e+00 -9.23968673e-01 -6.00406110e-01 8.57034981e-01
7.72335827e-01 -7.94254020e-02 -1.04305291e+00 3.67578208e-01]
|
[5.942464351654053, 8.073801040649414]
|
f351b9ec-8328-4b6a-b2c5-33876ac8f0f9
|
face-alignment-in-full-pose-range-a-3d-total
|
1804.01005
| null |
http://arxiv.org/abs/1804.01005v1
|
http://arxiv.org/pdf/1804.01005v1.pdf
|
Face Alignment in Full Pose Range: A 3D Total Solution
|
Face alignment, which fits a face model to an image and extracts the semantic
meanings of facial pixels, has been an important topic in the computer vision
community. However, most algorithms are designed for faces in small to medium
poses (yaw angle is smaller than 45 degrees), which lack the ability to align
faces in large poses up to 90 degrees. The challenges are three-fold. Firstly,
the commonly used landmark face model assumes that all the landmarks are
visible and is therefore not suitable for large poses. Secondly, the face
appearance varies more drastically across large poses, from the frontal view to
the profile view. Thirdly, labelling landmarks in large poses is extremely
challenging since the invisible landmarks have to be guessed. In this paper, we
propose to tackle these three challenges in an new alignment framework termed
3D Dense Face Alignment (3DDFA), in which a dense 3D Morphable Model (3DMM) is
fitted to the image via Cascaded Convolutional Neural Networks. We also utilize
3D information to synthesize face images in profile views to provide abundant
samples for training. Experiments on the challenging AFLW database show that
the proposed approach achieves significant improvements over the
state-of-the-art methods.
|
['Stan Z. Li', 'Zhen Lei', 'Xiaoming Liu', 'Xiangyu Zhu']
|
2018-04-02
| null | null | null | null |
['depth-image-estimation']
|
['computer-vision']
|
[-6.63327724e-02 8.64549354e-02 -8.95669162e-02 -6.75162911e-01
-3.73179197e-01 -3.25663865e-01 4.61028188e-01 -7.92791843e-01
-5.70288338e-02 2.82253265e-01 2.13370323e-02 1.83458790e-01
2.10951954e-01 -5.64290702e-01 -6.95724487e-01 -5.43697536e-01
2.46943519e-01 5.01994133e-01 -2.33123794e-01 -9.72817838e-02
3.53319533e-02 9.36861992e-01 -1.53700185e+00 -1.39001787e-01
3.84603590e-01 9.94067848e-01 -1.35696167e-02 7.40132406e-02
-2.58692622e-01 6.92697614e-02 -3.66762519e-01 -6.17324591e-01
6.87598169e-01 -3.08242291e-01 -5.39896250e-01 5.40256143e-01
1.17196345e+00 -4.43192393e-01 -5.81972599e-02 1.29318988e+00
5.69513679e-01 -1.21084809e-01 4.10764724e-01 -1.40436482e+00
-4.03897107e-01 -3.44510600e-02 -1.05990422e+00 -1.86828926e-01
3.67691666e-01 -1.80452004e-01 5.34431756e-01 -1.39962423e+00
6.56912506e-01 1.71541452e+00 6.76438332e-01 8.59305620e-01
-1.05982852e+00 -9.22620654e-01 2.45956942e-01 9.95509848e-02
-1.62101340e+00 -9.61906731e-01 1.04537058e+00 -4.20306593e-01
4.90942657e-01 5.87523170e-02 5.70436299e-01 8.80070925e-01
8.82679503e-03 2.14618817e-01 1.07878208e+00 -3.61573339e-01
3.11751124e-02 -8.03791136e-02 -5.05630851e-01 8.75744820e-01
1.29230618e-01 -2.18763247e-01 -5.22564471e-01 4.43005674e-02
1.00391245e+00 2.55200326e-01 -1.31296441e-01 -4.90223616e-01
-8.55272233e-01 6.03159964e-01 4.52450544e-01 3.50268707e-02
-3.64206612e-01 -6.91425949e-02 -3.05907149e-02 1.44573867e-01
6.25683665e-01 2.93702260e-02 -3.19471568e-01 3.82547408e-01
-9.82855201e-01 1.98107272e-01 5.84289253e-01 1.17202747e+00
1.05128598e+00 2.05246434e-01 3.53708267e-01 9.34104323e-01
7.06686497e-01 7.00121522e-01 1.17019430e-01 -9.35212553e-01
4.76371944e-01 6.82148278e-01 -5.16372658e-02 -1.29598200e+00
-4.94360834e-01 -1.86296254e-01 -9.32472467e-01 2.67184526e-01
3.73235345e-01 2.64345575e-02 -1.00914967e+00 1.85844028e+00
8.14991295e-01 3.45644504e-01 -2.17006758e-01 9.88543749e-01
9.92696404e-01 4.45266932e-01 -2.21128017e-01 -3.28503460e-01
1.51125121e+00 -9.28528726e-01 -8.87691379e-01 -6.31601036e-01
-1.20451963e-02 -1.06046748e+00 7.59897947e-01 -3.51265632e-02
-1.12076306e+00 -7.70892560e-01 -8.93330932e-01 -7.77637772e-03
-1.13636345e-01 1.12431973e-01 3.65899831e-01 7.10009456e-01
-1.28860676e+00 1.03353329e-01 -6.98385894e-01 -3.94981205e-01
6.94109082e-01 6.65363371e-01 -9.16928411e-01 -3.82002532e-01
-7.87654579e-01 8.72648776e-01 -1.48127347e-01 7.51110613e-01
-8.36551130e-01 -3.42905879e-01 -1.13464034e+00 -1.51429310e-01
3.87092948e-01 -4.70634073e-01 9.20442283e-01 -1.02353680e+00
-1.60959983e+00 1.11174512e+00 -6.17163837e-01 2.88828671e-01
5.13391912e-01 -1.87551379e-01 -3.77986372e-01 3.83881270e-03
1.19131431e-01 8.69842052e-01 1.23413563e+00 -1.31443095e+00
-1.06372714e-01 -8.52567613e-01 3.03394701e-02 2.26448089e-01
-2.62449622e-01 2.86806196e-01 -8.91741037e-01 -4.05294895e-01
3.61022145e-01 -1.08927524e+00 -1.02380320e-01 2.03163251e-01
-2.44222119e-01 -1.10677794e-01 1.11267292e+00 -7.70559788e-01
7.70885348e-01 -1.87637532e+00 2.85195559e-01 2.13630959e-01
1.71176538e-01 2.98946828e-01 -1.96910456e-01 7.64525309e-02
-1.97935849e-01 -1.00944474e-01 -4.14599888e-02 -6.97897494e-01
-7.17882365e-02 2.18734950e-01 4.55345176e-02 6.64791822e-01
3.29118699e-01 8.26024652e-01 -4.42725122e-01 -6.16821527e-01
1.97522745e-01 7.28331149e-01 -6.16223514e-01 4.01394337e-01
5.84135279e-02 6.26942635e-01 -3.20020705e-01 9.88754094e-01
1.28990054e+00 -9.97991338e-02 2.73216635e-01 -4.71771359e-01
-3.86574604e-02 -2.41809338e-01 -1.25572646e+00 1.77733696e+00
-4.80679303e-01 2.57548362e-01 2.99584568e-01 -7.35057592e-01
1.11460006e+00 4.16123688e-01 4.52522933e-01 -4.82961297e-01
1.76491946e-01 1.41936973e-01 -1.47706762e-01 -5.21691442e-01
-3.81896459e-02 -1.63081288e-01 2.81073540e-01 3.38475704e-01
2.04942018e-01 -1.75687730e-01 -1.00987449e-01 -3.74141276e-01
4.67750728e-01 3.46705139e-01 3.10847938e-01 -1.82674408e-01
8.34632277e-01 -6.24208152e-01 9.06437874e-01 -2.14671776e-01
-1.48197263e-01 7.57763207e-01 3.19357574e-01 -8.36610019e-01
-8.45736921e-01 -8.17839205e-01 -4.38995287e-02 5.91186523e-01
1.99371159e-01 -3.35949957e-01 -1.12449121e+00 -7.93991685e-01
-1.57361656e-01 -1.30974203e-01 -5.99434257e-01 9.16627347e-02
-8.33552003e-01 -5.70798934e-01 1.83456451e-01 4.27781284e-01
7.28302777e-01 -9.67645168e-01 -2.32035711e-01 -1.79725394e-01
-2.34573737e-01 -1.40408719e+00 -7.37654507e-01 -4.74900842e-01
-6.53271496e-01 -1.07605922e+00 -6.29520535e-01 -1.02176499e+00
1.42176795e+00 3.28308165e-01 9.55079257e-01 1.37698799e-01
-2.72576123e-01 1.36532754e-01 4.13499363e-02 -3.85678351e-01
-3.00124083e-02 -2.44705230e-01 3.34441096e-01 5.14497042e-01
5.42316139e-01 -3.76112342e-01 -6.13363206e-01 7.23746717e-01
-6.22142792e-01 2.80946214e-02 6.01582348e-01 7.35575736e-01
6.64122343e-01 -9.38877687e-02 5.05756199e-01 -8.04467976e-01
-6.69467971e-02 -2.52999961e-01 -6.74863100e-01 1.80037454e-01
-2.29891971e-01 -2.84193128e-01 4.59767610e-01 -2.42440090e-01
-9.92087543e-01 4.72385168e-01 -3.86487663e-01 -6.46874845e-01
-2.05882996e-01 -1.16339751e-01 -9.52081144e-01 -5.32859325e-01
2.01433957e-01 -8.71245563e-02 3.37002575e-01 -5.19496441e-01
2.73738682e-01 5.62453985e-01 3.74387801e-01 -5.12376308e-01
1.17607212e+00 6.37066066e-01 3.20212275e-01 -9.17292714e-01
-9.14366722e-01 -2.09138617e-01 -1.19289541e+00 -4.43473130e-01
8.76585186e-01 -9.77677763e-01 -6.35134101e-01 6.08475983e-01
-1.22494090e+00 3.10204532e-02 1.80728063e-01 3.35037917e-01
-5.26772320e-01 3.24591547e-01 -2.25248948e-01 -6.23905480e-01
-3.18434477e-01 -1.54222775e+00 1.31722116e+00 5.21338165e-01
-1.22258045e-01 -8.70217502e-01 -2.60363489e-01 5.03717840e-01
3.73444200e-01 3.28513920e-01 6.21882081e-01 -1.44176796e-01
-5.56962907e-01 -1.10716134e-01 -1.51147306e-01 3.85997236e-01
4.95738328e-01 -2.96515953e-02 -1.17558610e+00 -3.52251977e-01
1.40524864e-01 -3.91739190e-01 2.55707383e-01 3.15432757e-01
1.09636974e+00 -3.06865871e-01 -2.19281062e-01 8.91276181e-01
1.14211774e+00 8.44047517e-02 6.05081379e-01 6.11732304e-02
9.11397994e-01 8.72337222e-01 6.37082636e-01 1.86663643e-01
4.44883376e-01 1.09136522e+00 6.26788795e-01 -3.20105284e-01
-1.88978657e-01 -2.29632989e-01 3.03971708e-01 7.84096003e-01
-2.87233144e-01 1.93558961e-01 -6.70372009e-01 2.02381387e-01
-1.47284544e+00 -7.59433448e-01 1.83059886e-01 2.14001918e+00
7.72817910e-01 -3.23112249e-01 -1.05642207e-01 -5.55567592e-02
8.11906874e-01 2.28651792e-01 -4.01377201e-01 -1.15583144e-01
8.40935931e-02 2.36931562e-01 1.97869688e-02 5.05520165e-01
-1.09926045e+00 1.10849774e+00 5.84463120e+00 5.21859348e-01
-1.17819500e+00 8.11419711e-02 6.82018042e-01 1.36493295e-01
-3.85987721e-02 -1.27040586e-02 -1.25815499e+00 3.99923146e-01
4.35424000e-01 3.90633643e-01 3.72275442e-01 8.52835178e-01
1.07312530e-01 2.77255625e-01 -1.18783081e+00 1.35566485e+00
5.00867963e-01 -7.90789604e-01 1.17577352e-01 2.48272792e-01
8.35874140e-01 -4.79865342e-01 2.52394468e-01 -9.13134888e-02
-2.40012497e-01 -1.27726269e+00 6.13786399e-01 4.79499936e-01
1.02783263e+00 -8.04753423e-01 6.76356852e-01 1.44896418e-01
-1.29963148e+00 3.57210636e-01 -7.23394811e-01 1.81209132e-01
1.61267892e-01 3.88504893e-01 -7.94207335e-01 4.40191925e-01
7.39334047e-01 5.97136199e-01 -4.95164394e-01 6.88988090e-01
-2.12147623e-01 4.49297801e-02 -3.56761843e-01 5.46492934e-01
-6.11873604e-02 -4.03689295e-01 2.06853211e-01 7.71702647e-01
5.81652522e-01 -1.25370920e-01 1.76479623e-01 7.86772549e-01
-3.88824612e-01 5.52279055e-02 -7.07015276e-01 2.45161727e-01
4.89565074e-01 1.64342916e+00 -6.92907751e-01 1.37224391e-01
-6.31461799e-01 1.03650880e+00 3.13758463e-01 2.53590763e-01
-6.75577641e-01 1.32894158e-01 8.61229539e-01 2.25568667e-01
1.67800933e-01 -1.62842959e-01 1.49883941e-01 -1.04146934e+00
2.04112470e-01 -9.38287795e-01 5.80042936e-02 -5.21109581e-01
-1.19517398e+00 8.03268015e-01 -1.04064934e-01 -1.21923685e+00
-1.95894986e-01 -6.51731253e-01 -4.74750906e-01 1.04336739e+00
-1.71351814e+00 -1.52372444e+00 -6.87800884e-01 7.24541724e-01
6.63549483e-01 -2.72163272e-01 8.65147650e-01 4.75567967e-01
-7.39068627e-01 7.53079772e-01 -5.01454651e-01 3.11630994e-01
8.36764216e-01 -7.74571061e-01 6.33928418e-01 7.55616128e-01
2.52572358e-01 7.72405863e-01 2.81980425e-01 -5.12012184e-01
-1.74452496e+00 -1.28543043e+00 9.57677901e-01 -5.10434031e-01
-2.09382328e-04 -5.46594977e-01 -7.40224957e-01 7.87499249e-01
1.09797306e-01 6.22677922e-01 6.47905171e-01 -1.75303042e-01
-4.59581286e-01 -3.97578537e-01 -1.31593800e+00 5.98655522e-01
1.23490751e+00 -3.59729052e-01 -2.16160193e-01 2.06037611e-01
2.25212008e-01 -6.35701239e-01 -7.51744092e-01 4.94720042e-01
7.00092256e-01 -7.78629720e-01 1.06948316e+00 -4.11167979e-01
1.52496397e-01 -5.72242379e-01 -2.38116875e-01 -1.22507215e+00
-6.80667162e-02 -6.33169293e-01 -1.00059152e-01 1.46758783e+00
1.12082134e-03 -6.35942996e-01 8.03547502e-01 4.44063991e-01
9.53775123e-02 -7.85644233e-01 -1.10987425e+00 -4.47033644e-01
-3.36371243e-01 -4.51830029e-02 9.41031098e-01 9.50662851e-01
-5.91162860e-01 3.17367107e-01 -5.23246646e-01 4.23652917e-01
6.90000057e-01 1.62612036e-01 1.12633324e+00 -1.36971867e+00
2.23431081e-01 -1.65223107e-01 -7.53621817e-01 -1.03592408e+00
5.97627401e-01 -7.96715200e-01 -8.16754028e-02 -1.27426028e+00
1.17256500e-01 -4.24250782e-01 1.96194604e-01 5.63981950e-01
-1.46648526e-01 6.89415574e-01 8.65899846e-02 8.87496248e-02
-2.40957797e-01 5.63216567e-01 1.40821767e+00 -8.74619260e-02
2.53503621e-01 -1.36303222e-02 -6.40050650e-01 1.04419816e+00
5.02372980e-01 -2.45026961e-01 -4.17566031e-01 -6.82161450e-01
1.71962306e-02 -1.65844142e-01 1.72414064e-01 -7.27179527e-01
8.93632919e-02 -2.34139189e-01 8.76744151e-01 -5.49443007e-01
6.80948079e-01 -1.16910434e+00 3.67114455e-01 1.14173919e-01
2.08788261e-01 3.79217774e-01 1.78142205e-01 2.65845507e-01
-3.04488629e-01 -8.38408992e-02 1.02337766e+00 -1.87803790e-01
-5.80476105e-01 8.50848973e-01 3.79924715e-01 -1.03222802e-01
1.25064862e+00 -2.95638472e-01 2.69761942e-02 -2.36245722e-01
-4.86960948e-01 1.43540576e-01 6.40740335e-01 6.95158720e-01
8.80372584e-01 -1.69392109e+00 -7.61222005e-01 8.12244117e-01
-7.02301413e-03 3.84422779e-01 3.71460855e-01 7.73141146e-01
-6.31132543e-01 3.08316767e-01 -3.88679951e-01 -7.79659152e-01
-1.50975978e+00 3.83591205e-01 4.60309088e-01 2.73181558e-01
-4.88546342e-01 9.70411599e-01 4.92669553e-01 -6.20549321e-01
1.56114325e-01 1.93703264e-01 -3.19520265e-01 1.53241381e-01
6.51102960e-01 5.05676121e-02 1.80291697e-01 -1.32876575e+00
-4.66946393e-01 1.41496348e+00 -9.44788754e-02 2.54212588e-01
1.33876836e+00 -1.88017890e-01 -4.18508679e-01 -1.54429138e-01
1.30291903e+00 1.65298924e-01 -1.36513877e+00 -3.32554162e-01
-2.56071776e-01 -9.32889342e-01 -2.05794841e-01 -1.73452809e-01
-1.64630401e+00 1.06176257e+00 5.30640662e-01 -4.71665055e-01
1.15118313e+00 -1.45528659e-01 6.11776531e-01 1.17387459e-01
5.17070830e-01 -8.66796374e-01 9.65645015e-02 2.69545823e-01
1.07904780e+00 -1.21041119e+00 -9.07287281e-03 -6.83490694e-01
-4.37704802e-01 1.30309784e+00 9.96473372e-01 1.93073869e-01
7.21596718e-01 3.07289332e-01 3.29863578e-01 -3.13603908e-01
-2.76402056e-01 -1.45700912e-03 4.22587723e-01 5.25910854e-01
3.44709516e-01 -2.33944625e-01 1.80762008e-01 1.27534062e-01
-2.17243627e-01 -3.06157082e-01 1.18972607e-01 6.88516140e-01
-1.43233627e-01 -1.24211383e+00 -6.30350351e-01 5.18380255e-02
-4.73641574e-01 2.42657453e-01 -4.64125156e-01 7.49044597e-01
3.76727194e-01 8.30254853e-01 1.39351740e-01 -2.26490840e-01
3.51835281e-01 2.39442721e-01 7.10508227e-01 -6.97911859e-01
-3.74588519e-02 2.95180053e-01 -3.20705444e-01 -7.34808147e-01
-6.79069400e-01 -4.78379220e-01 -1.03585207e+00 -2.61738688e-01
-3.38903308e-01 -1.12837277e-01 8.10070634e-01 8.76576781e-01
3.32899570e-01 1.60559893e-01 9.35178280e-01 -1.29808378e+00
-3.65998626e-01 -1.00160301e+00 -4.32129085e-01 2.88753569e-01
2.79898196e-01 -1.05938017e+00 -2.23167196e-01 1.40559703e-01]
|
[13.312356948852539, 0.30389782786369324]
|
5a56913f-dbc8-4808-93c0-dff3ebc3cd66
|
low-resource-style-transfer-via-domain
| null | null |
https://openreview.net/forum?id=p_-ZgMkRD3
|
https://openreview.net/pdf?id=p_-ZgMkRD3
|
Low Resource Style Transfer via Domain Adaptive Meta Learning
|
Text style transfer (TST) without parallel data has achieved some practical success. However, most of the existing unsupervised text style transfer methods suffer from (i) requiring massive amounts of nonparallel data to guide transferring different text styles. (ii) colossal performance degradation when fine-tuning the model in new domains. In this work, we propose DAML-ATM(Domain Adaptive Meta-Learning with Adversarial Transfer Model), which consists of two parts, DAML and ATM. DAML is a domain adaptive meta-learning approach to refine general knowledge in multi-heterogeneous source domains, capable of adapting to new unseen domains with a small amount of data. Moreover, we propose a new unsupervised TST approach Adversarial Transfer Model (ATM), composed of a sequence-to-sequence pre-trained language model and uses adversarial style training for better content preservation and style transfer. Results on multi-domain datasets demonstrate that our approach generalizes well on unseen low-resource domains, achieving state-of-the-art results against ten-strong baselines.
|
['Anonymous']
|
2022-01-16
| null | null | null |
acl-arr-january-2022-1
|
['text-style-transfoer']
|
['natural-language-processing']
|
[ 6.25923812e-01 -3.05659384e-01 5.10449968e-02 -5.22678673e-01
-8.80191147e-01 -7.98893332e-01 8.52776945e-01 -2.70278424e-01
-5.05247116e-01 8.79003942e-01 2.85546869e-01 -1.02664337e-01
3.99891496e-01 -7.42673934e-01 -9.20394659e-01 -5.12441814e-01
5.12667060e-01 9.88619745e-01 6.24207675e-01 -7.63304412e-01
1.43282264e-01 1.95691079e-01 -9.23063219e-01 7.90201247e-01
9.62319911e-01 7.09171116e-01 2.53047943e-01 5.73180020e-01
-6.18037939e-01 5.73040307e-01 -6.73969388e-01 -6.57634377e-01
3.62283558e-01 -6.77614272e-01 -1.17685866e+00 -5.77275790e-02
5.87681472e-01 -4.48401004e-01 -3.64000559e-01 8.13559592e-01
6.93865776e-01 1.41967878e-01 9.11656141e-01 -1.13049507e+00
-9.27446008e-01 4.49288607e-01 -5.76285064e-01 6.15871660e-02
8.28563124e-02 1.19336851e-01 5.55802047e-01 -9.97950017e-01
8.29892516e-01 1.54710066e+00 6.11322880e-01 1.03820252e+00
-1.42138588e+00 -7.59769678e-01 1.82330191e-01 3.35117988e-02
-9.00898159e-01 -3.29955131e-01 8.96640480e-01 -1.93973064e-01
7.87783146e-01 -8.69581923e-02 1.16236240e-01 1.71688950e+00
2.08009318e-01 9.38215673e-01 1.41868317e+00 -5.21152735e-01
3.25798035e-01 3.57289284e-01 -3.44076484e-01 1.55803949e-01
-2.27257192e-01 -3.44191864e-02 -7.64905751e-01 -1.70090407e-01
7.05986381e-01 -1.41093254e-01 1.77029982e-01 -3.54981124e-01
-1.34919524e+00 7.40921259e-01 2.45451272e-01 2.93935210e-01
-9.56884250e-02 -2.15549886e-01 1.01469004e+00 1.00347281e+00
9.36706841e-01 1.83217019e-01 -7.21973538e-01 7.22502917e-03
-9.15254951e-01 3.77322763e-01 5.70517123e-01 1.39127374e+00
6.35393620e-01 2.09986329e-01 -3.56851488e-01 1.23757553e+00
-1.20183051e-01 7.16532707e-01 8.87148678e-01 -3.56370002e-01
9.25241470e-01 3.54005724e-01 -9.22996998e-02 -4.53634828e-01
-8.77127871e-02 -1.89668015e-01 -1.04939258e+00 2.56151378e-01
3.47110003e-01 -4.71569836e-01 -1.05485618e+00 1.74972022e+00
1.98477998e-01 3.41788828e-02 1.94598258e-01 5.78939736e-01
5.15788734e-01 8.63976061e-01 2.90404469e-01 2.56124914e-01
1.08494163e+00 -1.15687966e+00 -3.96085024e-01 -5.14956117e-01
4.85949725e-01 -9.46707487e-01 1.40253162e+00 4.45096284e-01
-1.12538731e+00 -8.87513340e-01 -8.15146327e-01 -2.70005018e-01
-6.11135006e-01 -2.93964416e-01 -2.19866429e-02 5.23222685e-01
-1.11745036e+00 5.75097501e-01 -5.03934681e-01 -7.34409451e-01
5.48715174e-01 2.50362873e-01 -3.51380438e-01 -2.46422723e-01
-1.25659204e+00 7.58780658e-01 7.96633184e-01 -5.07448971e-01
-9.71078873e-01 -9.12714362e-01 -6.56294823e-01 -2.32441261e-01
2.09689125e-01 -8.58671367e-01 1.25984561e+00 -1.73660028e+00
-1.97323239e+00 1.00962973e+00 -7.98463225e-02 -4.01134610e-01
9.34724092e-01 -5.61582923e-01 -6.29025042e-01 1.11040957e-01
9.54656005e-02 9.52996433e-01 1.31968677e+00 -1.31447077e+00
-7.23867059e-01 -3.13741773e-01 -3.03782046e-01 3.05312306e-01
-8.66704583e-01 1.55305946e-02 -2.96370775e-01 -1.43516052e+00
-4.93462682e-01 -1.00759721e+00 4.33416441e-02 2.37149317e-02
-6.64274469e-02 -2.69281507e-01 1.17802560e+00 -8.99702370e-01
9.21250761e-01 -2.09228325e+00 6.20568931e-01 -1.91859677e-01
-8.26787874e-02 7.34107614e-01 -5.98680437e-01 6.33241475e-01
-2.61653308e-02 -7.69892558e-02 -4.01677191e-01 -4.41117316e-01
2.25170013e-02 2.86786675e-01 -5.64003646e-01 -8.89502540e-02
4.38132674e-01 8.26399922e-01 -9.27936912e-01 -4.67015356e-01
6.74137771e-02 1.93393633e-01 -6.56452239e-01 4.83880430e-01
-5.73869944e-01 7.55756319e-01 -5.01359761e-01 3.23909640e-01
8.24301839e-01 -3.85189392e-02 2.16325879e-01 1.86165664e-02
2.24885568e-01 9.23726410e-02 -8.49973142e-01 2.39706779e+00
-5.45703948e-01 3.76382947e-01 -1.99820682e-01 -1.02797115e+00
1.13018572e+00 2.81568438e-01 1.04980737e-01 -8.28764558e-01
4.26240787e-02 2.73897409e-01 -4.97128218e-01 -1.67980060e-01
5.93450129e-01 -5.59020042e-01 -3.28927428e-01 4.41446573e-01
4.75779414e-01 -2.26405621e-01 -4.97103259e-02 2.46439725e-01
9.53700900e-01 4.23640966e-01 -5.92443310e-02 -3.45731556e-01
7.11140633e-01 9.42278877e-02 4.46577162e-01 5.87856054e-01
-7.35334903e-02 6.80859089e-01 1.75964922e-01 -5.75975418e-01
-1.49331617e+00 -1.07975996e+00 2.31408685e-01 1.75497353e+00
2.23869807e-03 -5.44662029e-02 -8.75311494e-01 -1.21852005e+00
3.38916816e-02 8.85436714e-01 -6.78368688e-01 -2.80230999e-01
-8.13745320e-01 -4.18396920e-01 6.66372538e-01 7.26925015e-01
9.42552507e-01 -1.23293936e+00 1.24772690e-01 3.81320179e-01
-2.55531073e-01 -1.11099541e+00 -9.02220011e-01 6.07091095e-03
-1.03423774e+00 -3.64847302e-01 -1.33527112e+00 -1.04456127e+00
4.89461064e-01 1.86820179e-01 1.32387269e+00 -4.67168629e-01
2.18028113e-01 3.21624160e-01 -5.93190789e-01 -3.85790318e-01
-9.85276818e-01 4.71923202e-01 4.65111509e-02 1.71314657e-01
3.26440334e-01 -6.75420761e-01 -3.33852142e-01 5.08658826e-01
-1.30176759e+00 1.91274211e-01 7.23603010e-01 1.15980780e+00
4.78889018e-01 -3.13879520e-01 9.19170022e-01 -1.33361900e+00
5.27821481e-01 -5.49573720e-01 -3.05235147e-01 4.28329647e-01
-4.93886471e-01 2.31897935e-01 1.11629140e+00 -6.88855350e-01
-1.74653172e+00 -2.88956851e-01 -6.56893477e-02 -4.26170886e-01
-3.97601277e-01 -6.87144026e-02 -2.25150973e-01 4.46086749e-02
9.96058345e-01 5.63847125e-01 -1.52588531e-01 -7.40308404e-01
5.40308297e-01 7.60237098e-01 4.31004524e-01 -9.99506354e-01
1.08440816e+00 4.92049336e-01 -3.29504371e-01 -6.19453073e-01
-6.98339224e-01 -1.80010036e-01 -1.03609860e+00 2.11187944e-01
7.56728172e-01 -1.05714571e+00 2.85029411e-01 9.63004231e-01
-9.53534961e-01 -9.45816576e-01 -3.06171179e-01 -1.08370267e-01
-6.49250746e-01 4.60643560e-01 -7.43834555e-01 -6.16583601e-02
-6.00742698e-01 -7.35861540e-01 1.11184371e+00 -1.06207229e-01
-1.60665452e-01 -1.31543624e+00 3.07175487e-01 3.60138923e-01
7.25898743e-01 1.98127463e-01 9.06739533e-01 -8.25397015e-01
-5.88289872e-02 2.29370579e-01 -2.34815478e-01 7.38498092e-01
1.93025917e-01 -6.87131047e-01 -9.40528572e-01 -8.10308814e-01
-2.03363016e-01 -8.24598372e-01 7.39888132e-01 -2.91531682e-01
1.13172400e+00 -2.99915701e-01 -1.52496636e-01 6.93606615e-01
1.37595785e+00 1.55131459e-01 6.42663240e-01 4.92806882e-01
8.54539812e-01 4.55875427e-01 5.69424808e-01 3.36032271e-01
1.76045164e-01 6.75397396e-01 -1.55341715e-01 -2.50274301e-01
-5.23404241e-01 -4.32323635e-01 6.34387791e-01 9.15500998e-01
1.04101844e-01 -4.41171825e-01 -6.76822484e-01 6.67890906e-01
-1.62123263e+00 -7.05626309e-01 2.54958749e-01 1.90155137e+00
1.19421899e+00 3.51799279e-01 4.97258186e-01 -2.49125093e-01
6.85959756e-01 4.26408164e-02 -9.13757741e-01 -6.06312752e-01
-2.06838414e-01 4.89146799e-01 4.15769160e-01 1.01835087e-01
-1.01301396e+00 1.34076405e+00 5.56806946e+00 1.15002418e+00
-1.16560853e+00 4.01637018e-01 5.56339622e-01 1.19185910e-01
-3.41008604e-01 -3.27586710e-01 -7.31293201e-01 6.42240882e-01
9.72909331e-01 -2.27410793e-01 4.82856870e-01 7.60297537e-01
-3.13106596e-01 5.47987759e-01 -1.05857074e+00 6.57914877e-01
1.93657607e-01 -1.01230824e+00 8.03487539e-01 -3.58405203e-01
1.02106380e+00 1.51901573e-01 1.82931632e-01 6.53905928e-01
5.66041410e-01 -4.55052078e-01 7.59521484e-01 1.63166821e-01
1.27395856e+00 -7.00887263e-01 4.71675664e-01 2.08451137e-01
-9.25112545e-01 2.70386547e-01 -4.94283438e-01 3.38097394e-01
-6.16715178e-02 1.87701970e-01 -7.69673228e-01 7.96665311e-01
6.56085670e-01 8.69357765e-01 -7.21865177e-01 3.54114562e-01
-7.59702921e-02 6.87178195e-01 -2.46063918e-02 1.91532314e-01
2.61757225e-01 1.31082296e-01 6.24584138e-01 1.59705567e+00
2.55134851e-01 -7.49420002e-02 1.34454980e-01 7.39224494e-01
-3.93927932e-01 1.48450479e-01 -5.59448242e-01 7.72417560e-02
2.63064414e-01 9.30291951e-01 -4.45134431e-01 -6.59923077e-01
-6.58056140e-01 1.87334073e+00 3.84484172e-01 4.58105028e-01
-5.99265635e-01 -4.46762949e-01 5.24701357e-01 -4.00059158e-03
3.51695746e-01 -1.17076807e-01 -3.50787580e-01 -1.45798850e+00
-6.70642555e-02 -1.27686501e+00 6.25706255e-01 -5.75323701e-01
-1.81323171e+00 7.74828315e-01 -1.58124138e-02 -1.44306254e+00
-1.30666867e-01 -6.62111759e-01 -4.61755246e-01 1.03939295e+00
-1.63199270e+00 -1.65039003e+00 -1.48437783e-01 1.08395672e+00
1.23911834e+00 -6.71078801e-01 8.84193063e-01 2.51199603e-01
-2.05731258e-01 1.08835590e+00 5.61186671e-01 2.84550518e-01
1.52852607e+00 -1.44612837e+00 1.06436217e+00 7.33219087e-01
-2.76763707e-01 1.66942522e-01 5.38307726e-01 -8.18666637e-01
-1.26569319e+00 -1.53633130e+00 5.85910201e-01 -5.19583404e-01
7.02848196e-01 -6.32441401e-01 -1.26813519e+00 9.56564903e-01
6.14232957e-01 -1.09166995e-01 6.52701855e-01 -1.02723405e-01
-6.34243965e-01 -2.33339071e-01 -1.39607704e+00 6.29355073e-01
1.15971196e+00 -3.59836042e-01 -9.59332526e-01 2.70233750e-01
7.79604137e-01 -3.88246387e-01 -8.75802398e-01 1.64804474e-01
3.52172613e-01 -7.66430795e-01 9.46765006e-01 -8.13618004e-01
6.42865777e-01 1.37676103e-02 -1.05106153e-01 -1.65652812e+00
-4.82566029e-01 -6.69335604e-01 2.52558976e-01 1.57744575e+00
1.93063095e-01 -7.21348763e-01 6.56296194e-01 1.85842037e-01
-1.17774665e-01 -1.36018740e-02 -7.18574524e-01 -1.06357586e+00
8.51162672e-01 1.16875552e-01 6.05309725e-01 1.23015630e+00
-2.70436078e-01 7.08195388e-01 -6.11239493e-01 -1.39883667e-01
7.24493623e-01 -1.05809793e-01 1.03158724e+00 -9.65389967e-01
-4.31065023e-01 -2.31423303e-01 3.28995660e-02 -1.14644551e+00
2.87461281e-01 -9.91055608e-01 -1.91395208e-01 -1.10671055e+00
7.51607716e-02 -4.53691572e-01 -4.31174308e-01 4.35513616e-01
-2.59020984e-01 1.25614077e-01 2.53154755e-01 2.81557858e-01
-5.83313286e-01 8.25649559e-01 1.36085927e+00 -3.94154191e-01
-1.37105659e-01 -1.49091154e-01 -5.20499825e-01 6.17506385e-01
8.84642184e-01 -5.79145670e-01 -4.96288449e-01 -8.60703170e-01
-2.49502793e-01 -1.39979392e-01 7.62705877e-02 -9.84681845e-01
-1.37250423e-01 -1.99925870e-01 6.10482216e-01 -4.46339786e-01
6.63536415e-02 -8.67547810e-01 -1.90585583e-01 3.33497554e-01
-5.84769905e-01 9.49847177e-02 7.64062047e-01 6.58758342e-01
-1.56210139e-01 -4.19181697e-02 1.12048888e+00 -7.90267438e-02
-1.06017399e+00 2.12637484e-01 -1.48120210e-01 3.86712492e-01
7.82091737e-01 1.00714184e-01 -2.90392876e-01 -8.18009675e-02
-6.85520113e-01 1.18771389e-01 5.87539554e-01 8.24226856e-01
3.48740548e-01 -1.65525270e+00 -1.11710632e+00 1.80398166e-01
3.55626315e-01 5.77957705e-02 3.69738102e-01 -3.59536633e-02
-4.65589345e-01 6.88550025e-02 -6.96506143e-01 -5.07922292e-01
-1.04682505e+00 7.89934337e-01 1.07885450e-01 -5.36458731e-01
-6.83944762e-01 7.52333224e-01 5.42819619e-01 -7.43480623e-01
-1.22788332e-01 7.49268830e-02 1.56212971e-01 -7.83068612e-02
6.25423193e-01 4.13129747e-01 7.75182471e-02 -3.56986314e-01
-1.04142856e-02 5.30760884e-01 -7.15722084e-01 -2.99034685e-01
1.38541853e+00 -2.95522600e-01 6.10054471e-02 4.03311014e-01
1.17984903e+00 -3.06439310e-01 -1.54349148e+00 -9.30527091e-01
5.10430895e-02 -3.51160109e-01 -2.83432007e-01 -1.17155159e+00
-7.46170819e-01 1.07657969e+00 8.24759901e-01 -1.85922101e-01
1.41509473e+00 -3.46470267e-01 1.40561581e+00 4.04104680e-01
4.31319624e-01 -1.26949763e+00 5.82981586e-01 7.23253310e-01
1.07029867e+00 -1.36135209e+00 -4.13417429e-01 -1.81329884e-02
-9.03586447e-01 1.15630448e+00 8.16670477e-01 -1.37228221e-01
3.84662002e-01 -1.68055035e-02 1.92578360e-01 3.79726738e-01
-6.01468146e-01 2.14444995e-01 3.13382059e-01 7.89224446e-01
2.67476141e-01 -5.46440780e-02 1.80108324e-02 4.56629694e-01
5.62912636e-02 1.71328425e-01 5.58497831e-02 1.07957768e+00
-3.34586889e-01 -1.70471299e+00 -3.50695491e-01 9.16098356e-02
-3.87525380e-01 -2.25026906e-01 -5.06082237e-01 6.79630220e-01
-6.71304017e-02 7.18642056e-01 -1.13966480e-01 -2.91748077e-01
6.39263749e-01 4.27092016e-01 5.13179660e-01 -7.36198902e-01
-7.75941074e-01 8.63917992e-02 -1.37755156e-01 -2.03636304e-01
-1.82796866e-01 -6.41862631e-01 -6.96232557e-01 -4.82080579e-01
2.85531014e-01 -1.66103601e-01 2.91680962e-01 9.64215994e-01
5.07295966e-01 5.59601963e-01 6.79386973e-01 -8.03804994e-01
-6.51463985e-01 -1.24288213e+00 -5.55714965e-01 8.99959028e-01
2.02316284e-01 -3.36766660e-01 1.39268488e-01 6.83222771e-01]
|
[11.70520305633545, 9.577610969543457]
|
7eb8ccc8-c157-4b59-83b4-07de2b73a776
|
incremental-self-supervised-learning-based-on
|
2303.17354
| null |
https://arxiv.org/abs/2303.17354v4
|
https://arxiv.org/pdf/2303.17354v4.pdf
|
ISSTAD: Incremental Self-Supervised Learning Based on Transformer for Anomaly Detection and Localization
|
In the realm of machine learning, the study of anomaly detection and localization within image data has gained substantial traction, particularly for practical applications such as industrial defect detection. While the majority of existing methods predominantly use Convolutional Neural Networks (CNN) as their primary network architecture, we introduce a novel approach based on the Transformer backbone network. Our method employs a two-stage incremental learning strategy. During the first stage, we train a Masked Autoencoder (MAE) model solely on normal images. In the subsequent stage, we apply pixel-level data augmentation techniques to generate corrupted normal images and their corresponding pixel labels. This process allows the model to learn how to repair corrupted regions and classify the status of each pixel. Ultimately, the model generates a pixel reconstruction error matrix and a pixel anomaly probability matrix. These matrices are then combined to produce an anomaly scoring matrix that effectively detects abnormal regions. When benchmarked against several state-of-the-art CNN-based methods, our approach exhibits superior performance on the MVTec AD dataset, achieving an impressive 97.6% AUC.
|
['Li Zhu', 'Fei Guo', 'Wenping Jin']
|
2023-03-30
| null | null | null | null |
['defect-detection']
|
['computer-vision']
|
[ 7.64698029e-01 6.45001084e-02 1.27947003e-01 -2.61456609e-01
-5.78602433e-01 -8.54725540e-02 3.99367332e-01 4.01670933e-01
-2.93407857e-01 2.53344029e-01 -2.58967578e-01 -4.10956174e-01
3.30070078e-01 -8.70338678e-01 -6.84646547e-01 -8.65209103e-01
1.24039754e-01 1.03403546e-01 3.75014573e-01 -8.01019650e-03
2.66308337e-01 6.07101202e-01 -1.60595679e+00 4.65934545e-01
9.14211094e-01 1.48591053e+00 -2.80809104e-01 4.71147418e-01
-1.66422784e-01 9.66994882e-01 -6.41528726e-01 -2.00115889e-01
3.50233883e-01 -3.95107597e-01 -6.57039881e-01 5.14960170e-01
3.91941100e-01 -4.42620277e-01 -3.72496217e-01 1.16995525e+00
2.67271817e-01 -1.37219427e-03 5.12805700e-01 -1.15220439e+00
-5.24886727e-01 2.70787686e-01 -5.50802112e-01 3.80742997e-01
-5.26736071e-03 3.50572914e-01 9.14885461e-01 -1.00978220e+00
2.60864973e-01 7.67695248e-01 6.86950803e-01 4.41618949e-01
-1.38996124e+00 -3.58417302e-01 7.08781555e-02 3.69953007e-01
-1.08376646e+00 -1.42229423e-01 1.01692259e+00 -4.73413646e-01
7.47921467e-01 -3.67593467e-02 6.42460465e-01 9.56442058e-01
3.22525084e-01 7.96588123e-01 9.69309032e-01 -4.57168639e-01
4.70336199e-01 -2.08925515e-01 -1.86923906e-01 7.75617421e-01
7.90037587e-02 1.49217984e-02 -3.16584557e-01 1.07703097e-04
7.37239122e-01 3.73857230e-01 -1.75989717e-01 -5.10149479e-01
-1.09748971e+00 8.20270717e-01 6.45760238e-01 1.61585853e-01
-8.75393629e-01 -1.04329892e-01 4.76855963e-01 4.10100669e-01
5.37547410e-01 2.96272129e-01 -2.38506153e-01 2.30453953e-01
-8.81774843e-01 1.29469946e-01 3.36867571e-01 1.31578550e-01
6.91381991e-01 2.32703567e-01 -2.12167621e-01 9.91036892e-01
2.55306870e-01 -1.45969279e-02 5.63029528e-01 -7.35703886e-01
2.55763739e-01 1.01417184e+00 -2.20337778e-01 -9.62902248e-01
-2.62250692e-01 -5.93474925e-01 -1.03254855e+00 7.49517560e-01
4.10340726e-01 7.56425261e-02 -1.26072538e+00 1.33119774e+00
2.93721676e-01 3.05414379e-01 1.52321175e-01 6.77534223e-01
2.92143643e-01 4.77948666e-01 -4.38251123e-02 1.87818229e-01
1.05814028e+00 -8.38477254e-01 -5.01108527e-01 -4.51014698e-01
5.70980251e-01 -5.33329725e-01 9.11667228e-01 5.85519135e-01
-9.96536374e-01 -4.40837622e-01 -1.33529615e+00 1.77209646e-01
-3.65351200e-01 2.66002446e-01 2.60644257e-01 4.05985087e-01
-8.87324691e-01 7.01100767e-01 -1.10811925e+00 -1.81385487e-01
8.16701055e-01 2.54240930e-01 -4.24598575e-01 -3.25734913e-01
-7.70734489e-01 6.49656057e-01 3.80771935e-01 3.47025067e-01
-1.12815642e+00 -5.69989502e-01 -9.66144383e-01 7.62081519e-02
2.91688621e-01 -3.04101139e-01 1.17009950e+00 -1.04108930e+00
-1.27362347e+00 7.42415309e-01 1.19077109e-01 -7.22393811e-01
3.75349462e-01 -1.18044212e-01 -5.72426856e-01 2.00429127e-01
1.03838138e-01 5.93453109e-01 1.07590818e+00 -1.24221981e+00
-9.75698471e-01 -4.00405139e-01 -1.52245477e-01 -1.92045480e-01
-4.09551471e-01 -1.93865329e-01 -3.65236104e-01 -8.65089893e-01
5.53617835e-01 -6.67692244e-01 -3.94870073e-01 2.30609983e-01
-5.89740574e-01 -5.80589026e-02 1.05836427e+00 -7.48815060e-01
1.11850214e+00 -2.40636301e+00 4.57360521e-02 5.87719321e-01
2.90314138e-01 3.41706663e-01 -1.56910326e-02 -8.23481902e-02
-4.51301545e-01 -3.27620327e-01 -8.51732671e-01 -3.21374714e-01
-3.68995517e-01 5.33169061e-02 -1.01997271e-01 4.49673682e-01
8.23185563e-01 7.25065291e-01 -6.61385238e-01 -2.34788299e-01
4.28787649e-01 2.98544466e-01 -5.92262208e-01 3.04257989e-01
-2.53240854e-01 4.36957717e-01 -2.33704999e-01 9.49504435e-01
5.09564042e-01 -2.85304070e-01 -9.27430987e-02 -1.64848670e-01
1.68030381e-01 -1.08957618e-01 -1.01978552e+00 1.49016380e+00
-2.20288441e-01 5.85754871e-01 1.06380263e-03 -1.32635748e+00
9.84024763e-01 2.09567294e-01 6.82663262e-01 -7.70164013e-01
1.76269710e-01 3.33135635e-01 1.42315596e-01 -4.59671319e-01
2.41444156e-01 6.29332364e-02 1.23784974e-01 3.59501302e-01
8.17378750e-04 3.40683043e-01 9.36331868e-04 -1.00406259e-01
1.64422667e+00 -7.97184110e-02 -2.96618175e-02 1.44037306e-01
8.54385853e-01 6.17002212e-02 5.44159651e-01 4.93139505e-01
-3.65404904e-01 8.09089839e-01 5.99153697e-01 -7.49130309e-01
-1.13358665e+00 -1.06547868e+00 -1.37003735e-02 5.70447445e-01
-1.65941179e-01 1.02766193e-02 -8.69591594e-01 -1.03794813e+00
3.80628556e-02 4.87647831e-01 -8.15406322e-01 -5.33930480e-01
-5.57539225e-01 -8.75594735e-01 2.88659215e-01 7.38290548e-01
7.10476279e-01 -1.38746691e+00 -5.46667278e-01 2.81415105e-01
1.33080932e-03 -1.05693734e+00 -5.41169792e-02 4.14104521e-01
-9.60967422e-01 -1.12198007e+00 -4.75926340e-01 -9.14486349e-01
1.07031620e+00 -1.57806143e-01 9.27867353e-01 3.01141351e-01
-6.17030025e-01 1.25156984e-01 -2.99347788e-01 -3.25894356e-01
-5.30580103e-01 -2.03653529e-01 -1.92125767e-01 5.87679386e-01
4.07783180e-01 -5.20374954e-01 -7.37560630e-01 1.16906844e-01
-1.14224827e+00 -4.02604789e-01 1.02464390e+00 9.87801790e-01
9.31336880e-01 1.94584161e-01 4.92225885e-01 -9.01810884e-01
4.80321914e-01 -5.83180010e-01 -5.99622428e-01 2.15377137e-02
-7.67272353e-01 -6.47420809e-02 6.41840458e-01 -2.74709344e-01
-8.57642829e-01 3.30882370e-01 -4.86109406e-01 -6.53774619e-01
-3.83986562e-01 5.59754431e-01 -9.01050121e-02 -2.54906006e-02
6.74947858e-01 3.78987014e-01 2.21336618e-01 -5.11760354e-01
-4.07726727e-02 5.67656100e-01 1.01701927e+00 -8.46275836e-02
8.26518953e-01 4.95070159e-01 -3.75677459e-02 -6.28565848e-01
-7.62152255e-01 -3.70231807e-01 -6.68787181e-01 -2.77579546e-01
9.83939111e-01 -7.12580144e-01 -2.49097809e-01 8.94552588e-01
-7.96425283e-01 -3.80864769e-01 -5.50012648e-01 1.71734855e-01
-2.97497094e-01 2.96786845e-01 -6.00026250e-01 -6.27799511e-01
-2.58038223e-01 -1.20147622e+00 9.83534932e-01 1.97775066e-02
1.05601326e-01 -7.64091194e-01 -1.16480745e-01 3.07509184e-01
4.02733445e-01 5.08776307e-01 1.09572184e+00 -8.51293504e-01
-5.88638246e-01 -7.38980293e-01 -1.92397624e-01 9.67166781e-01
1.09265149e-01 -1.73112258e-01 -9.17230248e-01 -1.74388885e-01
-8.80766585e-02 -2.28172868e-01 1.01558232e+00 2.51197577e-01
1.77617896e+00 -4.97217923e-02 -2.21975073e-01 5.09201705e-01
1.31299806e+00 2.72298545e-01 8.74263525e-01 5.73527515e-01
7.34115124e-01 3.42661649e-01 4.38788116e-01 2.37897620e-01
7.01945089e-03 4.64466333e-01 8.92649651e-01 -3.44282359e-01
-1.13961570e-01 -3.08192670e-02 3.19898665e-01 4.48765963e-01
2.32065886e-01 -8.75662938e-02 -9.91756141e-01 6.13984942e-01
-1.71707428e+00 -7.36451685e-01 8.37784186e-02 2.04649186e+00
4.55492586e-01 4.79541242e-01 -1.41818956e-01 7.27669597e-01
5.98139286e-01 9.76926759e-02 -7.07041204e-01 -2.05244392e-01
-6.70498461e-02 2.86975265e-01 2.47103006e-01 -3.39153525e-03
-1.31190205e+00 6.07434094e-01 6.23877668e+00 4.54250485e-01
-1.14049077e+00 -1.22154877e-01 8.79882932e-01 1.95502982e-01
6.94620758e-02 -3.76061082e-01 -2.59664416e-01 5.22017419e-01
8.59327435e-01 5.36876917e-01 1.83377638e-01 8.55756462e-01
8.86752363e-03 -9.41023454e-02 -9.81926501e-01 7.07474649e-01
1.94545165e-01 -1.21096849e+00 2.01472156e-02 9.41972062e-02
7.33735681e-01 7.02387020e-02 1.66703820e-01 1.80825427e-01
-6.94631180e-03 -9.59693134e-01 5.19794881e-01 4.33331817e-01
6.14935100e-01 -7.98403621e-01 1.02591193e+00 6.96787313e-02
-7.66725719e-01 -4.75116193e-01 -8.13896358e-02 1.81437448e-01
1.30679235e-02 8.03829432e-01 -7.49231100e-01 2.82426983e-01
8.63850355e-01 6.99937642e-01 -7.06803024e-01 1.11379230e+00
-1.97799519e-01 7.36778080e-01 -1.61123089e-02 6.31660700e-01
1.55834451e-01 -7.71117806e-02 3.84202242e-01 7.69625783e-01
2.66705275e-01 -2.21048698e-01 1.80286989e-01 9.24762189e-01
-2.32386246e-01 -3.62735130e-02 -4.76363838e-01 2.96398643e-02
1.20808482e-01 1.33670485e+00 -8.10925245e-01 -1.78917557e-01
-7.25181639e-01 1.30330765e+00 2.32621536e-01 2.03298524e-01
-5.44135869e-01 -4.27792072e-01 7.84268856e-01 1.14878632e-01
5.30968308e-01 6.88968673e-02 -3.56800526e-01 -8.94475877e-01
2.79616982e-01 -9.90737438e-01 4.55027193e-01 -4.68999177e-01
-1.34395468e+00 6.81894124e-01 -3.68900239e-01 -1.34562886e+00
-2.43078396e-01 -8.57611001e-01 -9.48401034e-01 6.36722386e-01
-1.50946200e+00 -9.60802078e-01 -5.27395904e-01 5.70581794e-01
5.94215989e-01 -4.84190702e-01 7.64975488e-01 4.76293683e-01
-1.08013654e+00 5.83781838e-01 7.44195059e-02 5.11498153e-01
3.74465376e-01 -1.36327899e+00 6.11012816e-01 1.39480865e+00
9.42517910e-03 1.02655657e-01 5.14254332e-01 -6.11211896e-01
-8.54231238e-01 -1.51623881e+00 3.96942496e-01 -2.72017866e-01
4.91181582e-01 -1.35927081e-01 -1.24744308e+00 6.80281222e-01
-1.69204488e-01 6.86226368e-01 4.43091363e-01 -3.05304319e-01
-2.18628109e-01 -1.84534028e-01 -1.26818025e+00 4.23478842e-01
5.94951510e-01 -4.47204471e-01 -3.14703524e-01 9.75795835e-02
3.77548724e-01 -4.56009328e-01 -8.58245254e-01 6.08576596e-01
2.89387822e-01 -1.07873774e+00 8.25958312e-01 -4.71452922e-01
7.26022542e-01 -5.90090871e-01 -9.16220918e-02 -1.28587210e+00
-1.86073691e-01 -8.51211697e-03 -3.42391551e-01 9.39175725e-01
6.25037909e-01 -5.77260435e-01 1.01466489e+00 2.46891290e-01
-4.86731172e-01 -1.13823366e+00 -7.87911654e-01 -4.35029179e-01
-2.29899958e-01 -5.49523175e-01 4.92690325e-01 6.18756533e-01
-4.84323829e-01 -1.53208628e-01 -1.07447885e-01 3.88837665e-01
5.39546847e-01 -1.63566977e-01 4.59901243e-01 -1.35408938e+00
-7.99116343e-02 -3.75896126e-01 -7.28152096e-01 -5.62295079e-01
6.47579655e-02 -7.64234602e-01 3.26260954e-01 -1.38003886e+00
-3.15789059e-02 -3.68823171e-01 -7.76001334e-01 7.54444063e-01
-2.93487102e-01 7.40302920e-01 -3.08852226e-01 1.34419337e-01
-4.16141808e-01 5.17546773e-01 7.78826833e-01 -2.61708081e-01
-1.41582549e-01 8.14076886e-02 -5.72279453e-01 6.99453652e-01
9.07639325e-01 -2.87771434e-01 -1.56893671e-01 -2.82997638e-01
-1.07839078e-01 -4.93470162e-01 7.10470498e-01 -1.49860978e+00
5.86487651e-02 2.41683111e-01 8.00291896e-01 -5.12308598e-01
3.26198749e-02 -9.44006920e-01 -2.24329993e-01 5.78840613e-01
-1.85072243e-01 2.81366169e-01 1.87675759e-01 7.38781154e-01
-5.02294779e-01 -6.75870925e-02 8.56495261e-01 -2.77211238e-02
-8.44520092e-01 4.98283356e-01 -4.55791444e-01 -4.03119892e-01
1.22978842e+00 -2.90254802e-01 -8.48553330e-02 -8.77424404e-02
-7.74345458e-01 9.01341997e-03 4.66578990e-01 5.08668303e-01
8.83904159e-01 -1.31415510e+00 -5.33059537e-01 7.75727153e-01
3.81777346e-01 2.85879254e-01 2.92340219e-01 9.23822820e-01
-5.78724861e-01 -4.42251414e-02 -3.96388680e-01 -8.94975245e-01
-9.97557342e-01 4.31985676e-01 5.38209558e-01 -2.22521469e-01
-8.72184634e-01 7.33430445e-01 -1.02379955e-02 -1.07697122e-01
3.02749872e-01 -2.66824961e-01 -1.87155932e-01 -2.93550879e-01
5.37952363e-01 2.51915425e-01 4.68127221e-01 -5.63840985e-01
-1.43882528e-01 1.90870449e-01 -2.27970421e-01 1.10191047e-01
1.39871466e+00 1.14404738e-01 -1.54112354e-01 1.81836948e-01
1.11720741e+00 -3.39026809e-01 -1.36704290e+00 -3.72944921e-01
2.04215050e-01 -3.83809030e-01 3.87464523e-01 -8.39601994e-01
-1.50436616e+00 7.34001458e-01 1.07642639e+00 2.52814591e-01
1.48688626e+00 -4.00657579e-02 8.99218738e-01 2.35430941e-01
-1.83783814e-01 -9.64279115e-01 4.08558488e-01 2.78974235e-01
6.26509845e-01 -1.49833775e+00 -2.69438446e-01 -2.57777452e-01
-5.09045541e-01 1.02326179e+00 7.56096125e-01 -1.45374283e-01
5.69739997e-01 2.08181947e-01 2.81202406e-01 -3.94478232e-01
-4.93670791e-01 -6.47152290e-02 3.01350296e-01 5.78341126e-01
1.74444437e-01 -2.66990364e-01 2.45521545e-01 3.86919677e-01
1.38093501e-01 -1.45164415e-01 3.42235625e-01 1.11110437e+00
-4.37916130e-01 -1.12373734e+00 -3.35474044e-01 8.69107425e-01
-6.86002254e-01 1.31824479e-01 -2.05567300e-01 4.75722283e-01
3.28916222e-01 7.53335059e-01 4.00698066e-01 -4.92039114e-01
5.45536935e-01 2.91911960e-01 5.84846362e-02 -5.73524058e-01
-4.68567044e-01 -2.08928496e-01 -3.89609337e-01 -8.52003396e-01
-2.01889753e-01 -8.74465108e-01 -1.12794840e+00 4.80810106e-02
-1.78300694e-01 -1.48712680e-01 6.44802094e-01 9.94120181e-01
2.67943650e-01 1.03973436e+00 7.03235984e-01 -8.13421369e-01
-4.32802409e-01 -8.87186229e-01 -4.63989854e-01 6.70204341e-01
7.17717648e-01 -5.61876953e-01 -2.76899636e-01 7.41504654e-02]
|
[7.629889965057373, 2.0210838317871094]
|
4bd1378e-1b98-4f3f-a909-c383fbfea358
|
prompt-learning-for-fine-grained-entity-1
| null | null |
https://openreview.net/forum?id=7EemgCzGXAN
|
https://openreview.net/pdf?id=7EemgCzGXAN
|
Prompt-Learning for Fine-Grained Entity Typing
|
As an effective approach to tune pre-trained language models (PLMs) for specific tasks, prompt-learning has recently attracted much attention from researchers. By using cloze-style language prompts to stimulate the versatile knowledge of PLMs, prompt-learning can achieve promising results on a series of NLP tasks, such as natural language inference, sentiment classification, and knowledge probing. In this work, we investigate the application of prompt-learning on fine-grained entity typing in fully supervised, few-shot, and zero-shot scenarios.
We first develop a simple and effective prompt-learning pipeline by constructing entity-oriented verbalizer and templates and conducting masked language modeling. Further, to tackle the zero-shot regime, we propose a self-supervised strategy that carries out distribution-level optimization in prompt-learning to automatically summarize the information of entity types. Extensive experiments on three fine-grained entity typing benchmarks (with up to 86 classes) under fully supervised, few-shot and zero-shot settings show that prompt-learning methods significantly outperform fine-tuning baselines, especially when the training data is insufficient.
|
['Anonymous']
|
2021-11-16
| null | null | null |
acl-arr-november-2021-11
|
['entity-typing']
|
['natural-language-processing']
|
[-1.43832508e-02 -1.10037643e-02 -7.54108071e-01 -5.86717248e-01
-1.05898631e+00 -7.12856710e-01 6.60091698e-01 4.12991524e-01
-7.56714702e-01 7.81509876e-01 4.13259596e-01 -3.26805264e-01
1.47802368e-01 -7.03408957e-01 -7.39285469e-01 -3.73679936e-01
3.27925831e-01 4.73662317e-01 2.68139690e-01 -2.61199087e-01
1.71693668e-01 5.70006482e-02 -1.46197128e+00 5.17485619e-01
1.10182047e+00 6.75976396e-01 3.16394478e-01 6.19858325e-01
-7.35400021e-01 1.08092880e+00 -5.04777312e-01 -6.38095438e-01
-3.46568555e-01 -1.13281183e-01 -9.62262452e-01 -3.15762132e-01
1.99770838e-01 -2.03482434e-01 -3.96345137e-03 8.10475051e-01
7.73752868e-01 2.90027827e-01 7.14974463e-01 -9.65023458e-01
-6.79379225e-01 9.12869215e-01 -3.67570192e-01 3.38329524e-01
4.67683226e-01 2.64217526e-01 1.30790317e+00 -1.15102041e+00
5.60003340e-01 1.36313808e+00 6.14906132e-01 8.52344334e-01
-1.12327003e+00 -6.10745311e-01 5.53085446e-01 2.66262919e-01
-1.16525543e+00 -4.73424524e-01 5.71998656e-01 -4.39247400e-01
1.19356012e+00 1.02606900e-01 -2.14054007e-02 1.46795630e+00
-2.26356402e-01 1.18084788e+00 9.56201494e-01 -7.32676029e-01
3.25923741e-01 5.36101341e-01 6.09816074e-01 6.80973947e-01
8.31389651e-02 -6.99718446e-02 -8.44203293e-01 -4.70126271e-01
2.25947991e-01 -1.32203177e-01 -5.00481576e-02 3.82696115e-03
-9.97987866e-01 8.28147531e-01 -4.49066460e-02 1.63653389e-01
-4.62831616e-01 -2.25064129e-01 5.13695896e-01 2.01842830e-01
6.37775004e-01 7.96414256e-01 -9.48590577e-01 -3.03282529e-01
-6.88547313e-01 2.80498952e-01 1.07861912e+00 1.09974754e+00
8.50961447e-01 -5.48876114e-02 -9.81083572e-01 1.10260487e+00
1.88601762e-01 5.39276600e-01 5.60463488e-01 -5.28013945e-01
6.41096532e-01 6.17571473e-01 2.66601235e-01 -4.55195844e-01
-4.00643677e-01 -2.87439108e-01 -6.63512349e-01 -5.51798344e-01
2.62073725e-01 -5.38834333e-01 -6.81618333e-01 1.82509577e+00
3.89688075e-01 3.65228444e-01 1.00240827e-01 5.12677193e-01
1.14460468e+00 6.94666624e-01 6.29309297e-01 -3.13266158e-01
1.60204995e+00 -1.09539366e+00 -7.95519114e-01 -2.64989555e-01
9.30401146e-01 -4.51190412e-01 1.71755600e+00 1.09563217e-01
-8.25522423e-01 -3.25593054e-01 -5.85984468e-01 -1.57102570e-01
-5.16654313e-01 4.96248975e-02 6.73018217e-01 4.51804519e-01
-5.15304506e-01 3.04881006e-01 -6.67963624e-01 -2.69641042e-01
4.53064024e-01 -1.97672565e-02 2.15166863e-02 -1.03659838e-01
-1.45280182e+00 7.09452510e-01 4.33994621e-01 -3.50417405e-01
-7.28369355e-01 -1.10568070e+00 -9.88191307e-01 3.57634962e-01
6.77930713e-01 -8.06319475e-01 1.61421001e+00 -3.35786104e-01
-1.63024879e+00 8.44044805e-01 -6.10413790e-01 -3.65533412e-01
2.03876495e-01 -4.51188982e-01 -2.13297039e-01 -2.45674610e-01
1.53344586e-01 3.87588054e-01 6.68928504e-01 -9.76971090e-01
-6.89419389e-01 -1.39974803e-02 1.15541771e-01 2.77699977e-01
-6.33484364e-01 2.84560978e-01 -3.47249866e-01 -6.34130955e-01
-6.55528665e-01 -5.18087447e-01 -2.98505992e-01 -6.25168443e-01
-5.78918040e-01 -8.83045435e-01 2.68149823e-01 -1.95170432e-01
1.67649019e+00 -2.19283080e+00 -1.52291387e-01 -1.02453671e-01
7.31933042e-02 5.78716636e-01 -3.82128924e-01 3.60504210e-01
2.55479395e-01 5.50761111e-02 7.66806901e-02 -5.77215254e-01
3.10780555e-01 1.86190084e-01 -5.61182141e-01 -1.25387818e-01
3.60984385e-01 1.27428579e+00 -1.18847442e+00 -5.50843894e-01
-6.63250834e-02 1.11265957e-01 -7.46914029e-01 7.94741869e-01
-6.98215365e-01 2.36643106e-01 -5.84793925e-01 7.05785751e-01
2.45720774e-01 -6.08313918e-01 3.44792716e-02 4.04411787e-03
2.75268834e-02 5.80878735e-01 -9.16610718e-01 1.60473037e+00
-7.02184737e-01 1.39634669e-01 -2.94613242e-01 -6.95404708e-01
6.41756594e-01 2.88957506e-01 3.86815108e-02 -7.76465714e-01
-5.99297509e-03 1.07392520e-01 -3.88642967e-01 -8.89115036e-01
6.72297537e-01 -1.23071238e-01 -4.94536579e-01 6.60390019e-01
4.22510177e-01 2.35644072e-01 1.96574762e-01 3.69074672e-01
1.08290899e+00 -8.68510455e-02 5.41932583e-01 -3.86686064e-02
3.84502262e-01 -1.37562841e-01 4.65464890e-01 1.13563621e+00
-8.76206346e-03 1.81376725e-01 4.16360170e-01 -1.43310368e-01
-5.49616933e-01 -8.28763008e-01 3.63351256e-02 2.20668459e+00
9.49674994e-02 -5.99851012e-01 -7.43581176e-01 -8.33996594e-01
6.86893091e-02 9.52409029e-01 -5.19175649e-01 -2.86518067e-01
-4.11248863e-01 -8.61889064e-01 6.06958568e-01 6.87643349e-01
1.78805724e-01 -1.44709861e+00 -2.21529797e-01 3.19893241e-01
-2.78829813e-01 -1.30939734e+00 -5.91587245e-01 5.11210561e-01
-4.99035180e-01 -7.95249343e-01 -6.34195387e-01 -9.34735298e-01
4.96891528e-01 4.46860492e-02 1.32021987e+00 -1.40498891e-01
-1.03220977e-01 3.86378288e-01 -5.23292840e-01 -4.60493475e-01
-9.70445275e-02 6.44212246e-01 1.12304114e-01 2.81256791e-02
8.11279416e-01 -2.79480278e-01 -2.18570977e-01 2.22762525e-02
-5.45296788e-01 -1.79445297e-01 6.14318848e-01 1.08928657e+00
4.06633824e-01 -4.62459326e-01 8.75144899e-01 -1.50948381e+00
9.98430967e-01 -6.53511763e-01 -3.99027169e-01 7.23788619e-01
-4.65260476e-01 4.22538489e-01 7.94526160e-01 -7.10357606e-01
-1.44771254e+00 -2.57213742e-01 -3.31222802e-01 -1.54999375e-01
-3.56928468e-01 6.27430856e-01 -2.50814378e-01 2.52442777e-01
9.17359054e-01 2.95561165e-01 -7.08819151e-01 -6.46581471e-01
7.32531905e-01 9.10158634e-01 5.28809905e-01 -1.01089513e+00
6.20711207e-01 3.67471874e-02 -8.12402129e-01 -8.33100319e-01
-1.59569359e+00 -7.98126161e-01 -3.60184401e-01 2.34588280e-01
6.86426103e-01 -1.15481651e+00 -7.78328121e-01 6.04208589e-01
-1.20159912e+00 -7.76558578e-01 -2.67656088e-01 1.85487792e-01
-3.38540941e-01 1.24269068e-01 -7.06656277e-01 -8.03815722e-01
-6.18366182e-01 -6.83043599e-01 1.20949566e+00 4.83859986e-01
-3.88852417e-01 -1.24839580e+00 2.37555012e-01 8.78776088e-02
4.59942639e-01 -5.19419372e-01 1.04656565e+00 -1.21719575e+00
-4.60894763e-01 1.36950120e-01 -1.22032844e-01 3.35982665e-02
7.09784701e-02 -4.08197939e-01 -1.06851315e+00 -5.82148023e-02
-3.53874266e-01 -9.07584667e-01 8.54330122e-01 1.28697157e-01
1.21855652e+00 -5.41416943e-01 -4.17756617e-01 7.67301142e-01
1.06036747e+00 -2.29237407e-01 1.72153339e-01 2.66904414e-01
8.48971963e-01 5.21425426e-01 8.83867741e-01 6.44035399e-01
9.08111751e-01 6.28506124e-01 -1.76204309e-01 2.66329804e-03
8.67437944e-02 -6.52778268e-01 2.34770790e-01 7.68705189e-01
3.26579541e-01 -2.43309557e-01 -7.09425569e-01 5.59099972e-01
-1.83927786e+00 -8.69129121e-01 3.02582443e-01 1.92194641e+00
1.56048524e+00 1.18670858e-01 1.04207039e-01 -4.15211082e-01
5.44077456e-01 3.44152480e-01 -7.27405608e-01 -3.60555828e-01
-1.25548705e-01 3.03122938e-01 2.40403131e-01 5.83828032e-01
-1.08868349e+00 1.55585718e+00 5.99267578e+00 1.05875289e+00
-1.15758848e+00 1.81097791e-01 4.44072276e-01 6.49122819e-02
-5.86066842e-01 -2.89550424e-01 -1.53337586e+00 6.12658501e-01
1.04487705e+00 -3.58579457e-01 3.49361658e-01 9.14180160e-01
2.02681031e-02 1.05765769e-02 -1.07926822e+00 9.29637253e-01
-2.39930744e-03 -1.41178751e+00 -5.93576171e-02 -4.58899856e-01
8.29031467e-01 1.32952243e-01 -9.94226038e-02 9.86732543e-01
6.88394606e-01 -7.79304028e-01 3.84889573e-01 4.10023838e-01
1.07182860e+00 -5.42214394e-01 5.59687853e-01 8.64531040e-01
-1.05878460e+00 -5.88262230e-02 -2.40441442e-01 -1.47401839e-02
2.76624143e-01 5.37951231e-01 -8.88932168e-01 2.50623882e-01
4.95156407e-01 5.29268980e-01 -3.52691323e-01 8.40134084e-01
-5.49557090e-01 9.93008494e-01 -1.34956881e-01 -5.36463559e-01
3.83269005e-02 4.07495916e-01 2.08423004e-01 1.72810507e+00
-1.82971746e-01 4.30934817e-01 3.44619215e-01 7.25875676e-01
-3.54971766e-01 1.77110583e-01 -2.30479926e-01 -2.16337070e-01
9.65844095e-01 1.31572461e+00 -2.81364918e-01 -6.43307447e-01
-4.50337946e-01 8.24171603e-01 8.17653835e-01 6.27672315e-01
-6.36750996e-01 -5.12045741e-01 5.08715153e-01 -1.05034783e-01
3.47737938e-01 2.28016600e-02 -3.07679325e-01 -1.42450523e+00
-1.24770254e-01 -9.00685668e-01 5.57414055e-01 -4.00745571e-01
-1.55255663e+00 4.91127819e-01 1.63374133e-02 -6.89060807e-01
-6.01429343e-01 -5.43899953e-01 -7.52595007e-01 8.71176720e-01
-1.89046240e+00 -9.57862198e-01 -2.39973590e-01 6.02335930e-01
8.01285088e-01 -3.51039588e-01 1.03577507e+00 2.67231792e-01
-7.12588549e-01 1.17086184e+00 -8.27201083e-02 2.43422776e-01
1.05815756e+00 -1.46933842e+00 5.19956052e-01 5.91186345e-01
1.45057678e-01 9.20910716e-01 5.88739157e-01 -5.52542627e-01
-1.25138950e+00 -1.30357301e+00 1.32263613e+00 -5.43400943e-01
6.53407931e-01 -6.85611904e-01 -1.23847735e+00 7.08677232e-01
1.44502625e-01 8.81904811e-02 1.00986099e+00 6.51424706e-01
-6.18816793e-01 8.22781771e-02 -8.12356651e-01 5.06936550e-01
9.24361348e-01 -8.36791039e-01 -9.18063521e-01 4.14287537e-01
1.04115796e+00 -6.12271309e-01 -5.47196090e-01 2.72291005e-01
3.70516896e-01 -4.33976799e-01 8.22399080e-01 -1.10324526e+00
3.59212905e-01 1.59017727e-01 1.04352258e-01 -1.57094991e+00
-2.99667835e-01 -7.67883658e-01 -6.42425597e-01 1.50562775e+00
5.35704911e-01 -5.10849714e-01 7.27007329e-01 5.35673022e-01
-4.98521850e-02 -9.57433701e-01 -2.90957659e-01 -6.10046089e-01
7.45323068e-03 -2.96830833e-01 5.17162621e-01 9.70971823e-01
2.05150470e-01 8.08064580e-01 -4.55008715e-01 1.06596835e-02
4.06901300e-01 4.70842384e-02 8.65400195e-01 -1.04445422e+00
-5.24725735e-01 -3.32034558e-01 3.52200806e-01 -1.48980498e+00
7.07988918e-01 -8.89854074e-01 2.92480409e-01 -1.27543068e+00
4.21095788e-01 -5.99954247e-01 -2.77912706e-01 7.03677952e-01
-9.69719529e-01 -3.24660897e-01 -1.88872680e-01 -1.40291736e-01
-1.16302872e+00 5.37504733e-01 8.24399292e-01 4.04036343e-02
-2.63495386e-01 1.86874568e-01 -1.03517187e+00 6.13041043e-01
3.95082533e-01 -3.99813354e-01 -2.99193233e-01 -3.67262512e-01
3.16789418e-01 -1.30803064e-01 -2.66522355e-02 -1.57195538e-01
6.06458664e-01 -3.39978218e-01 -1.11025006e-01 -5.31826496e-01
1.41135037e-01 -1.61085606e-01 -7.33781636e-01 4.56798077e-03
-8.38926554e-01 -2.93859959e-01 1.69438824e-01 4.94691551e-01
-1.43090740e-01 -2.96633929e-01 5.74696481e-01 -1.33742049e-01
-9.86725569e-01 4.57336485e-01 -2.16088891e-01 8.20337772e-01
6.60570264e-01 3.22207868e-01 -5.21048427e-01 -2.09641859e-01
-5.04154563e-01 6.14523172e-01 1.22613437e-01 5.33587813e-01
4.04737175e-01 -1.20075917e+00 -7.05036938e-01 1.63246155e-01
5.57875872e-01 2.13537320e-01 3.90227199e-01 6.43477261e-01
2.16354832e-01 5.73263109e-01 2.65347958e-01 -5.86280525e-01
-9.80802000e-01 5.76885402e-01 1.26695316e-02 -6.48493409e-01
-3.15324068e-01 1.31642163e+00 3.63700002e-01 -8.02260220e-01
6.66622341e-01 -2.49786109e-01 -3.20954204e-01 2.42110267e-01
8.12315583e-01 7.18583912e-02 6.64018746e-03 -1.05666235e-01
-3.21123034e-01 3.04882348e-01 -5.07480204e-01 1.53494269e-01
1.23473132e+00 -9.48842615e-02 2.08946139e-01 7.24756062e-01
1.04764521e+00 1.92216024e-01 -1.17995703e+00 -8.43473434e-01
4.87149745e-01 -1.13076337e-01 -2.65707314e-01 -9.03904080e-01
-3.86772126e-01 9.10685301e-01 -8.26932341e-02 -5.48710255e-03
7.84771919e-01 2.37172455e-01 9.45054770e-01 6.92839622e-01
4.49586570e-01 -1.02527070e+00 2.72906959e-01 1.01853049e+00
3.33563030e-01 -1.41202104e+00 -4.17321712e-01 -2.94416189e-01
-8.78515601e-01 5.76253176e-01 8.75826001e-01 2.28680134e-01
4.55107957e-01 4.35386807e-01 1.37061819e-01 2.65222378e-02
-1.24511409e+00 -2.83511311e-01 5.13299942e-01 4.53952551e-01
6.24398410e-01 7.53462315e-02 -1.44195691e-01 1.31292999e+00
-6.89868852e-02 -4.80083227e-02 1.50841698e-01 8.03666115e-01
-6.75381124e-01 -1.06628537e+00 5.22786453e-02 5.36202133e-01
-4.39258277e-01 -5.24808347e-01 -2.36513227e-01 3.38072181e-01
-2.76358306e-01 7.41279006e-01 -1.87571034e-01 -2.25388214e-01
4.28462625e-01 2.61956066e-01 3.41203749e-01 -1.25364971e+00
-7.01784551e-01 -1.42994702e-01 2.71131575e-01 -5.22001207e-01
-1.30262807e-01 -5.31460226e-01 -1.18461001e+00 1.03389183e-02
-4.70665455e-01 3.28003526e-01 2.00449213e-01 1.36544335e+00
4.15915161e-01 5.53836584e-01 6.48012102e-01 -5.41636765e-01
-9.73363936e-01 -1.10376203e+00 -3.69232327e-01 4.45927173e-01
3.33047748e-01 -6.30106926e-01 -3.01566929e-01 -9.69464481e-02]
|
[10.63659381866455, 8.204693794250488]
|
26ae115c-1966-4b6a-90aa-b847e2ed01c5
|
an-adaptive-threshold-for-the-canny-edge
|
2209.08699
| null |
https://arxiv.org/abs/2209.08699v1
|
https://arxiv.org/pdf/2209.08699v1.pdf
|
An Adaptive Threshold for the Canny Edge Detection with Actor-Critic Algorithm
|
Visual surveillance aims to perform robust foreground object detection regardless of the time and place. Object detection shows good results using only spatial information, but foreground object detection in visual surveillance requires proper temporal and spatial information processing. In deep learning-based foreground object detection algorithms, the detection ability is superior to classical background subtraction (BGS) algorithms in an environment similar to training. However, the performance is lower than that of the classical BGS algorithm in the environment different from training. This paper proposes a spatio-temporal fusion network (STFN) that could extract temporal and spatial information using a temporal network and a spatial network. We suggest a method using a semi-foreground map for stable training of the proposed STFN. The proposed algorithm shows excellent performance in an environment different from training, and we show it through experiments with various public datasets. Also, STFN can generate a compliant background image in a semi-supervised method, and it can operate in real-time on a desktop with GPU. The proposed method shows 11.28% and 18.33% higher FM than the latest deep learning method in the LASIESTA and SBI dataset, respectively.
|
['Jong-Eun Ha', 'Keong-Hun Choi']
|
2022-09-19
| null | null | null | null |
['edge-detection']
|
['computer-vision']
|
[ 5.23564875e-01 -6.22916043e-01 2.71596223e-01 -2.94861495e-01
1.12259440e-01 -2.05559403e-01 6.41396344e-01 -2.86506683e-01
-8.20917964e-01 6.30356610e-01 -3.75315577e-01 -4.98838007e-01
1.83077961e-01 -8.98625135e-01 -7.26776302e-01 -1.30767453e+00
-7.30585381e-02 -2.56756358e-02 1.26410985e+00 1.14034861e-01
-1.15873061e-01 5.69292128e-01 -1.67074943e+00 5.35614491e-01
7.23897457e-01 1.03698814e+00 5.13692260e-01 8.88203800e-01
-2.01823875e-01 1.32637727e+00 -1.12795305e+00 1.73691344e-02
5.44213831e-01 -3.58376205e-01 -2.75074452e-01 1.22862704e-01
5.30186355e-01 -6.70086086e-01 -3.91207337e-01 1.04732442e+00
4.81744438e-01 2.25419868e-02 2.77055621e-01 -1.28842354e+00
-1.35044381e-01 1.75581604e-01 -8.80466759e-01 9.88216519e-01
-2.75272638e-01 1.85381174e-01 -4.19288129e-01 -4.03649867e-01
4.68561202e-01 1.29963732e+00 7.05763936e-01 5.34117639e-01
-6.21042609e-01 -8.14973235e-01 3.83765608e-01 2.86276519e-01
-1.24055409e+00 -5.18447384e-02 4.76172179e-01 -4.68426853e-01
5.45575738e-01 3.08781296e-01 9.94228423e-01 8.76627624e-01
4.14639175e-01 9.50952590e-01 1.07773554e+00 -2.39070877e-01
5.12377322e-02 -3.44152860e-02 3.58508736e-01 7.83747613e-01
7.92263746e-01 3.79314348e-02 -2.35986114e-01 3.95619869e-02
7.88450778e-01 5.42652845e-01 -3.18094432e-01 -2.20256043e-03
-1.28035760e+00 6.71738863e-01 3.31711471e-01 5.08338749e-01
-2.88389087e-01 -4.05097827e-02 3.35321218e-01 -8.95270333e-02
4.52384859e-01 -4.22996849e-01 -3.90625864e-01 2.44230792e-01
-1.35770428e+00 4.06871773e-02 5.65694988e-01 7.60959625e-01
3.68469298e-01 5.77975273e-01 -4.36754495e-01 2.14140296e-01
3.07678521e-01 1.16820192e+00 3.57686490e-01 -8.11522722e-01
1.49321854e-01 5.18452644e-01 3.67675722e-01 -1.25736523e+00
-6.60342157e-01 -3.29616308e-01 -1.16203249e+00 6.77454412e-01
7.06565619e-01 -6.01041317e-01 -1.33245099e+00 1.36071670e+00
7.74871707e-01 6.17326140e-01 2.17358813e-01 8.90312016e-01
1.26570189e+00 1.22974420e+00 1.54180243e-03 -5.40046513e-01
1.25849223e+00 -8.72115493e-01 -1.06149495e+00 -2.56651431e-01
1.66922942e-01 -6.70785069e-01 2.31815368e-01 7.81383693e-01
-5.60963154e-01 -8.25280190e-01 -1.00018156e+00 4.44664747e-01
-4.50029254e-01 4.83414263e-01 5.37009776e-01 9.58230317e-01
-1.11466300e+00 2.53537953e-01 -9.95284081e-01 -5.30017495e-01
3.52642566e-01 3.51202488e-01 -1.68433189e-01 -1.07785342e-02
-1.02389646e+00 7.23448396e-01 8.80286157e-01 5.53043365e-01
-1.12622964e+00 -2.46572360e-01 -6.14088833e-01 -4.18407172e-01
4.50020462e-01 -5.37158728e-01 8.78427744e-01 -1.35102320e+00
-1.12102103e+00 5.46575725e-01 -3.18128377e-01 -7.26322711e-01
7.70196140e-01 -2.44778935e-02 -6.36702538e-01 2.79666781e-01
4.42182608e-02 5.56200981e-01 9.55299437e-01 -1.08528554e+00
-1.33628595e+00 -1.82806447e-01 -8.69179203e-04 -1.26778528e-01
-1.21683720e-02 4.81317818e-01 -4.11567748e-01 -4.23541456e-01
3.93680558e-02 -5.30818999e-01 -1.88012064e-01 1.78488463e-01
5.22383973e-02 -9.42033082e-02 1.62065995e+00 -1.08769011e+00
9.41102803e-01 -2.00339937e+00 -4.58618730e-01 -9.00792852e-02
2.03172952e-01 1.12490475e+00 1.88739300e-01 -2.48047143e-01
2.86562026e-01 -3.66804779e-01 -1.16227314e-01 2.01146752e-01
-7.64193892e-01 2.95502961e-01 8.49627107e-02 6.22136295e-01
1.66410550e-01 4.12375867e-01 -1.00085771e+00 -9.09081340e-01
4.54969138e-01 5.09106457e-01 -5.52131683e-02 1.20032042e-01
9.22303367e-03 3.72968674e-01 -1.34187311e-01 6.93315268e-01
1.14252198e+00 -1.25234202e-01 -1.89934112e-02 -2.25271180e-01
-5.30474484e-01 -6.19136155e-01 -1.47803271e+00 9.02357876e-01
2.86174595e-01 1.11934543e+00 4.26278621e-01 -8.78230453e-01
8.41770828e-01 1.96629122e-01 3.43522310e-01 -6.88241601e-01
3.32596242e-01 -5.08038402e-02 1.12820879e-01 -6.38814926e-01
3.16611290e-01 2.77479619e-01 7.29144692e-01 -5.45900464e-02
-1.02443993e-01 6.17943704e-01 5.51276326e-01 1.40854985e-01
1.00092506e+00 1.43106371e-01 -1.14254784e-02 -4.75048423e-01
6.84802830e-01 2.90688843e-01 9.72899675e-01 1.00921130e+00
-5.49185038e-01 1.98182806e-01 6.78748116e-02 -9.50091779e-01
-5.25348365e-01 -9.28172529e-01 -1.36562109e-01 8.48213911e-01
6.66956604e-01 2.56436497e-01 -7.57876813e-01 -4.81123447e-01
-3.23087275e-01 4.04254019e-01 -6.47361577e-01 4.97763827e-02
-8.17150414e-01 -1.42363441e+00 5.14540374e-01 4.54370797e-01
1.31756222e+00 -1.02875721e+00 -1.08988392e+00 1.02542996e-01
-2.96047986e-01 -1.18370187e+00 2.20149290e-02 4.38194424e-02
-7.91849911e-01 -1.27129304e+00 -8.60120416e-01 -9.78865921e-01
5.45027018e-01 9.40291703e-01 7.70324767e-01 2.27552857e-02
-3.41773808e-01 8.03751498e-02 -2.54847527e-01 -8.88926804e-01
-4.11074132e-01 -5.65645218e-01 -4.04019468e-02 3.37529778e-01
4.97253239e-01 1.26661807e-01 -7.00025260e-01 4.78176087e-01
-1.03455532e+00 3.70482147e-01 5.17613590e-01 6.45314753e-01
3.24359179e-01 4.36818510e-01 1.16782121e-01 -4.94377524e-01
-5.40673323e-02 -3.13228846e-01 -1.25581300e+00 4.39981610e-01
1.80658661e-02 -5.13025463e-01 3.51785034e-01 -6.50363982e-01
-1.48143160e+00 7.37795532e-02 3.45893681e-01 -2.95477808e-01
-2.98922390e-01 -2.20466763e-01 -2.42532894e-01 -3.19081545e-01
6.19405985e-01 4.32580918e-01 -6.94068670e-02 -2.50985473e-01
-1.98446989e-01 5.00191152e-01 6.32376850e-01 1.03950381e-01
7.49013901e-01 9.12380457e-01 4.73760888e-02 -1.06001377e+00
-5.53679883e-01 -5.86912215e-01 -7.37743080e-01 -7.11203456e-01
1.32417965e+00 -1.02734697e+00 -7.15883791e-01 1.09256935e+00
-1.37123942e+00 -3.92969370e-01 3.44300240e-01 5.27941167e-01
1.35808140e-01 5.61643004e-01 -4.48238701e-01 -1.23675954e+00
-3.84431720e-01 -1.01710641e+00 9.25479352e-01 5.91406047e-01
5.95811963e-01 -8.75395000e-01 -3.65661293e-01 2.83859640e-01
4.68072891e-01 7.58698463e-01 -1.69976801e-02 -3.50548774e-01
-9.93263423e-01 -1.04069166e-01 -6.26569808e-01 4.12650228e-01
3.80039752e-01 4.63989258e-01 -9.36254203e-01 -2.73168176e-01
3.03334028e-01 5.38625658e-01 1.15470457e+00 8.75123799e-01
7.54743993e-01 -2.49619156e-01 -5.42263031e-01 7.66606867e-01
1.40925527e+00 1.03800797e+00 4.75680977e-01 5.80629706e-01
7.64865816e-01 3.48085940e-01 7.75590181e-01 3.17889392e-01
-7.48309270e-02 2.30019540e-01 5.48730075e-01 -6.45995677e-01
-1.84759259e-01 5.53593099e-01 5.45955420e-01 1.31272838e-01
-2.36773074e-01 -3.05945367e-01 -1.06307220e+00 5.17595172e-01
-2.06866193e+00 -1.52802706e+00 -7.22035766e-01 1.92980480e+00
5.28731048e-01 1.27792835e-01 2.97799498e-01 3.90029401e-02
1.10245943e+00 9.96462554e-02 -3.94749254e-01 6.75970763e-02
-5.89174211e-01 -2.80857354e-01 1.01187360e+00 2.17914030e-01
-1.79000366e+00 7.75884569e-01 5.84689999e+00 8.65651608e-01
-1.36113441e+00 3.62446070e-01 6.95335627e-01 -2.45274305e-01
7.12938070e-01 -5.91114342e-01 -9.91849184e-01 8.44626606e-01
8.18343043e-01 1.49402365e-01 5.13385907e-02 8.69797945e-01
3.89784306e-01 -5.95471680e-01 -3.79968822e-01 1.05690622e+00
2.15230301e-01 -1.34794712e+00 2.60566063e-02 -3.59846056e-01
8.09714139e-01 4.48993966e-02 -2.67162770e-01 2.14169156e-02
1.03402883e-01 -6.31893694e-01 7.78648615e-01 4.53652859e-01
2.11692110e-01 -6.47009969e-01 1.30811942e+00 5.73913515e-01
-1.27851200e+00 -1.82568550e-01 -5.03726065e-01 1.91496238e-02
5.78070339e-03 4.28795367e-01 -7.32638001e-01 5.70320308e-01
1.10266507e+00 3.82324487e-01 -8.54977667e-01 1.36501324e+00
1.72537565e-01 7.63816535e-01 -4.94812876e-01 -2.70085067e-01
5.51496327e-01 -2.86641896e-01 4.50360000e-01 1.74212456e+00
2.29630142e-01 1.11668490e-01 3.52989614e-01 3.48082244e-01
5.76683879e-01 -1.74902156e-01 -5.78938782e-01 3.76834750e-01
-2.84940954e-02 1.19057500e+00 -1.20628810e+00 -1.01530933e+00
-4.73174930e-01 8.26393306e-01 -5.36790967e-01 3.63344193e-01
-1.32043576e+00 -4.29542214e-01 2.41656512e-01 1.19463503e-01
4.57454413e-01 -1.26261041e-01 3.72480527e-02 -9.47634220e-01
5.29687665e-02 -7.49120235e-01 4.99513328e-01 -7.90592849e-01
-6.93036616e-01 7.20282197e-01 4.46752757e-01 -1.12665796e+00
1.93663493e-01 -1.07953525e+00 -7.53444135e-01 5.98363817e-01
-1.39465117e+00 -1.21903670e+00 -9.50606585e-01 8.59273314e-01
4.33629721e-01 -3.84691894e-01 8.49583596e-02 4.78742033e-01
-8.83515775e-01 2.29015741e-02 2.40877628e-01 4.95496482e-01
5.02498507e-01 -1.00009227e+00 1.44740656e-01 1.83930802e+00
-2.44759530e-01 2.29361281e-01 6.09233260e-01 -9.62429285e-01
-1.13955009e+00 -1.59318066e+00 3.08769524e-01 -2.28386655e-01
1.91714972e-01 -1.85933039e-01 -7.98709095e-01 4.83633041e-01
4.74069208e-01 1.30120263e-01 3.04795831e-01 -7.34940171e-01
2.27145597e-01 -4.15826797e-01 -1.36146498e+00 4.49936330e-01
9.15374935e-01 3.06624830e-01 -4.14695174e-01 5.31602442e-01
8.02038610e-01 -4.07502860e-01 -1.78621861e-03 6.45232618e-01
2.97273576e-01 -1.09858954e+00 8.45458150e-01 -2.09805965e-01
-3.57144982e-01 -1.13885581e+00 -1.31183192e-01 -4.09799278e-01
-4.74677294e-01 -5.55383086e-01 -5.68187200e-02 1.05304706e+00
-1.56408399e-01 -6.05824649e-01 7.60923326e-01 3.05067062e-01
3.42824124e-02 -1.26368761e-01 -9.14573312e-01 -1.02521729e+00
-5.48879385e-01 -2.31676981e-01 2.80728430e-01 6.86437309e-01
-1.13527822e+00 -2.02146858e-01 -6.42670333e-01 7.17462838e-01
8.50199997e-01 4.44481224e-02 9.23115075e-01 -1.31604755e+00
9.39755291e-02 -2.45026097e-01 -7.18002141e-01 -6.52493954e-01
-3.41344655e-01 -2.08690941e-01 6.83429539e-02 -1.79066980e+00
2.53319710e-01 1.48852959e-01 -3.72520059e-01 4.17172372e-01
-4.14766520e-01 3.77034366e-01 1.53002501e-01 -1.59676030e-01
-8.32851827e-01 5.44530153e-02 9.91700172e-01 -4.31347996e-01
-1.06370740e-01 1.89086586e-01 6.49120212e-02 9.88804936e-01
9.64657545e-01 -3.69410992e-01 -2.54092336e-01 -4.90256310e-01
-6.98176503e-01 -2.09962487e-01 6.57962382e-01 -1.42245519e+00
4.18378979e-01 -4.56628770e-01 1.06669283e+00 -1.24029899e+00
2.60214746e-01 -9.55661893e-01 3.74167442e-01 1.02965212e+00
5.93884468e-01 7.07492083e-02 7.00450003e-01 6.46491349e-01
-7.67725408e-02 -1.15859114e-01 1.01164925e+00 -1.52117223e-01
-1.26344085e+00 9.31737944e-02 -8.64362419e-01 -2.45794401e-01
1.32065153e+00 -6.91053391e-01 -5.32431185e-01 6.26537055e-02
-3.64098698e-01 5.10501936e-02 2.39268899e-01 1.92206904e-01
5.64974844e-01 -1.13832521e+00 -6.02435470e-01 3.17903578e-01
-3.04980904e-01 5.87782152e-02 4.15063530e-01 9.32923019e-01
-1.20471776e+00 4.04959351e-01 -4.93354499e-01 -9.86163259e-01
-1.83616793e+00 7.65518188e-01 4.62612391e-01 7.67553449e-02
-6.04633927e-01 7.72590518e-01 4.86339390e-01 1.90243974e-01
5.29110074e-01 -6.10394537e-01 -1.83128402e-01 3.68344262e-02
1.12624872e+00 5.73397696e-01 -1.39678434e-01 -5.51966548e-01
-5.42344809e-01 3.57350796e-01 3.45943093e-01 2.01189548e-01
1.01557958e+00 3.12835604e-01 -2.55822778e-01 1.74773648e-01
6.66894794e-01 -2.43344560e-01 -1.39436495e+00 -2.10860848e-01
-1.25136212e-01 -6.88093305e-01 3.99318859e-02 -7.10491478e-01
-1.34398556e+00 7.65962780e-01 1.26343536e+00 2.75057375e-01
1.53533399e+00 -5.51570177e-01 5.84638476e-01 5.87517023e-01
2.93012977e-01 -9.37492251e-01 -1.03380822e-01 3.25322866e-01
5.55944681e-01 -1.53228712e+00 -6.32592961e-02 -2.50073493e-01
-5.09404242e-01 1.32375288e+00 9.02870059e-01 1.17366984e-02
6.84568286e-01 5.05098403e-01 3.17768961e-01 2.18637772e-02
-4.54349339e-01 -4.10064936e-01 2.01757252e-01 1.01428938e+00
1.99958626e-02 -8.12963843e-02 2.37357654e-02 1.54282838e-01
4.44875717e-01 -4.03691083e-02 4.96503085e-01 1.14778805e+00
-8.62837374e-01 -4.64610189e-01 -9.80893731e-01 3.38236272e-01
-6.14331424e-01 1.24521792e-01 -2.62064129e-01 1.01063967e+00
8.12770367e-01 1.32664847e+00 2.42695764e-01 -1.67689860e-01
1.03638053e-01 -8.69220421e-02 2.48355016e-01 -7.28927627e-02
-5.05012870e-01 1.84497029e-01 -8.90733600e-02 -5.85518718e-01
-9.62501705e-01 -6.14480436e-01 -1.02152872e+00 -4.48193699e-01
-3.85928452e-01 -3.08436323e-02 5.55349112e-01 6.84092581e-01
-8.77990574e-02 7.55289316e-01 2.04636112e-01 -9.39370155e-01
3.06756854e-01 -8.14925790e-01 -4.21284080e-01 2.59653300e-01
4.99759823e-01 -6.83944046e-01 -1.76373884e-01 3.47263277e-01]
|
[8.811417579650879, -0.7954463362693787]
|
c5989a2f-a3ee-4f3a-9102-27a6bacadca5
|
evolving-tsukamoto-neuro-fuzzy-model-for
|
2305.10421
| null |
https://arxiv.org/abs/2305.10421v1
|
https://arxiv.org/pdf/2305.10421v1.pdf
|
Evolving Tsukamoto Neuro Fuzzy Model for Multiclass Covid 19 Classification with Chest X Ray Images
|
Du e to rapid population growth and the need to use artificial intelligence to make quick decisions, developing a machine learning-based disease detection model and abnormality identification system has greatly improved the level of medical diagnosis Since COVID-19 has become one of the most severe diseases in the world, developing an automatic COVID-19 detection framework helps medical doctors in the diagnostic process of disease and provides correct and fast results. In this paper, we propose a machine lear ning based framework for the detection of Covid 19. The proposed model employs a Tsukamoto Neuro Fuzzy Inference network to identify and distinguish Covid 19 disease from normal and pneumonia cases. While the traditional training methods tune the parameters of the neuro-fuzzy model by gradient-based algorithms and recursive least square method, we use an evolutionary-based optimization, the Cat swarm algorithm to update the parameters. In addition, six texture features extracted from chest X-ray images are give n as input to the model. Finally, the proposed model is conducted on the chest X-ray dataset to detect Covid 19. The simulation results indicate that the proposed model achieves an accuracy of 98.51%, sensitivity of 98.35%, specificity of 98.08%, and F1 score of 98.17%.
|
['Maysam Orouskhani', 'Farzan Vahedifard', 'Hossein Abbasi', 'Negar Firoozeh', 'Sevda Molani', 'Marziyeh Rezaei']
|
2023-05-17
| null | null | null | null |
['medical-diagnosis', 'specificity']
|
['medical', 'natural-language-processing']
|
[-4.17883834e-03 -5.95175087e-01 3.59842256e-02 8.98124948e-02
1.02743484e-01 -6.63043037e-02 -1.07218616e-01 1.46206588e-01
-5.31634152e-01 7.29068577e-01 -4.46544111e-01 -7.20520839e-02
-5.28838038e-01 -7.04433441e-01 1.39619291e-01 -8.85220647e-01
1.87055707e-01 8.45661223e-01 2.06956267e-01 -3.72682279e-03
4.31562215e-01 6.57218874e-01 -1.57052743e+00 4.75493520e-02
1.22979546e+00 1.02380645e+00 5.06730974e-01 9.61186647e-01
2.67972976e-01 6.58291101e-01 -7.63761342e-01 1.86414808e-01
1.63990427e-02 -4.79477286e-01 -2.69633502e-01 -9.64298472e-02
-4.87078995e-01 -2.67374903e-01 1.69758543e-01 9.35094237e-01
5.39515734e-01 2.90657431e-01 1.21241713e+00 -9.23387349e-01
-4.63884652e-01 -1.01803914e-01 -6.09372377e-01 6.22115076e-01
-4.91298884e-02 -1.11445621e-01 1.01899706e-01 -7.59557962e-01
4.42969292e-01 1.08851862e+00 8.30881655e-01 6.15869224e-01
-4.90992278e-01 -5.40828824e-01 -4.15178120e-01 3.56263161e-01
-1.49095345e+00 2.86406249e-01 3.43484253e-01 -6.89170539e-01
9.33675170e-01 4.39178705e-01 8.20021093e-01 2.68729270e-01
6.18266463e-01 3.66842806e-01 1.00588620e+00 -6.01069689e-01
1.42878458e-01 2.48396605e-01 2.20182285e-01 9.25104141e-01
8.13593149e-01 1.85703654e-02 2.18706474e-01 -2.93455154e-01
7.38284469e-01 6.99827492e-01 -7.65211955e-02 3.57040942e-01
-8.08317900e-01 1.11188817e+00 2.26981968e-01 6.16090655e-01
-6.31419599e-01 -4.14954841e-01 3.13614607e-01 -1.79389387e-01
-2.01583561e-02 5.97606540e-01 -4.01328236e-01 -1.64063219e-02
-5.21797419e-01 1.69177845e-01 4.52162862e-01 3.86977568e-02
-5.73164336e-02 5.68893962e-02 -3.21012996e-02 9.29092288e-01
5.70150256e-01 1.03283167e+00 7.93367267e-01 -7.32402623e-01
-1.41894659e-02 9.32442307e-01 4.82175872e-02 -1.37991607e+00
-5.11899650e-01 -1.38747841e-01 -1.11557066e+00 -1.08635545e-01
-1.28530830e-01 -5.03295898e-01 -1.06477439e+00 1.03733242e+00
4.95174915e-01 2.78995663e-01 1.79403976e-01 9.15881217e-01
5.25108218e-01 7.90718675e-01 -1.23295933e-01 -5.17435670e-01
1.41973042e+00 -5.80649912e-01 -8.84270310e-01 3.80395859e-01
3.64953518e-01 -6.24031305e-01 4.12242621e-01 5.99336803e-01
-6.98180616e-01 -4.95678008e-01 -9.85865533e-01 7.61475801e-01
-3.09995353e-01 4.70542341e-01 2.24923193e-01 6.75862610e-01
-6.42957568e-01 6.68036938e-02 -1.11765993e+00 -6.93061590e-01
1.93633169e-01 5.22734523e-01 2.33563378e-01 9.43103712e-03
-1.09601068e+00 8.69329214e-01 5.93979418e-01 2.27025867e-01
-3.50563765e-01 -2.80989379e-01 -3.61194849e-01 -2.61557400e-01
1.30100429e-01 -9.44070637e-01 7.43961811e-01 -4.77577537e-01
-1.04767418e+00 5.08464158e-01 -1.15181327e-01 -3.01927567e-01
1.83927283e-01 1.04453182e-02 -6.32889450e-01 4.89012748e-01
-1.03625759e-01 2.19496086e-01 6.41367733e-01 -1.06429183e+00
-1.04480863e+00 -4.32300031e-01 -4.22123522e-01 1.46192253e-01
-2.78724283e-01 3.92685056e-01 -1.19367406e-01 -6.71841681e-01
2.10295375e-02 -9.49752152e-01 -2.34429657e-01 -2.61294454e-01
-1.65553376e-01 -2.45446026e-01 1.17754304e+00 -6.91154599e-01
1.54406857e+00 -1.85388839e+00 -2.89054573e-01 6.65803790e-01
3.44808042e-01 9.25418735e-01 5.33384800e-01 5.92167340e-02
2.79194027e-01 -1.34547183e-03 -3.54683131e-01 5.59449136e-01
-5.46468019e-01 3.47303510e-01 4.84704375e-01 1.60184637e-01
2.49759868e-01 2.47178361e-01 -5.57327449e-01 -7.28513837e-01
4.05143172e-01 6.72064960e-01 -6.47064686e-01 5.29000580e-01
1.17712110e-01 2.56052464e-01 -9.62828994e-01 6.66926742e-01
5.60060620e-01 -5.49048364e-01 -3.32733132e-02 1.01083472e-01
8.55605900e-02 -6.48374319e-01 -1.35415554e+00 4.91522342e-01
-1.07734822e-01 1.59226894e-01 7.04680160e-02 -1.07781804e+00
1.01380336e+00 5.41872084e-01 6.57112002e-01 -3.83748710e-02
7.16138422e-01 2.54821867e-01 2.09724337e-01 -1.36000729e+00
4.27336320e-02 4.05078009e-03 3.88126075e-01 4.25898403e-01
-6.74106538e-01 -5.42864539e-02 2.68225044e-01 -2.55453080e-01
7.76658475e-01 -5.94545066e-01 4.48712707e-01 -1.74359426e-01
1.02469385e+00 4.72546756e-01 6.50987029e-01 5.44493675e-01
-1.12785034e-01 3.20953250e-01 -2.41634324e-01 -4.07764137e-01
-5.96841753e-01 -7.58086145e-01 -4.60874796e-01 3.43013108e-01
1.38260394e-01 3.80680561e-01 -1.02657425e+00 -1.99889988e-01
2.38932576e-02 4.01683569e-01 -4.95514244e-01 -2.63181835e-01
-6.86525285e-01 -1.26834404e+00 4.24968034e-01 2.96009213e-01
7.47396290e-01 -9.96983528e-01 -1.06949294e+00 2.61958778e-01
-9.03248340e-02 -6.48543239e-01 -1.09935038e-01 -1.51508182e-01
-8.53095412e-01 -1.17650044e+00 -6.23652816e-01 -1.16701663e+00
8.79852831e-01 2.89680287e-02 4.79671687e-01 6.65719569e-01
-9.24434662e-01 1.60475418e-01 -4.13356513e-01 -8.60639155e-01
-4.45232451e-01 -2.36552164e-01 2.76451230e-01 -2.04566628e-01
5.95004797e-01 1.09466612e-01 -6.98638260e-01 3.37205529e-01
-6.31080031e-01 -3.04012895e-01 4.83583927e-01 9.44502890e-01
7.00894237e-01 6.98678315e-01 6.24578476e-01 -6.44172072e-01
8.96444976e-01 -5.52362621e-01 -4.80664939e-01 3.51796180e-01
-1.00140965e+00 -4.21407431e-01 6.81421757e-01 -3.21544707e-01
-1.13798213e+00 -1.56825438e-01 -6.32985234e-02 -3.72506320e-01
-2.22133502e-01 4.78373379e-01 3.87868434e-01 1.00252785e-01
5.91498733e-01 7.66768828e-02 2.60919303e-01 -3.23134661e-01
-6.17013872e-01 1.16873062e+00 5.10619164e-01 -4.29624133e-02
5.15021861e-01 1.89359620e-01 -2.92830579e-02 -8.09235394e-01
-5.34707129e-01 -6.18961930e-01 -3.53214890e-01 -3.13518822e-01
1.43577266e+00 -6.18234932e-01 -9.34450328e-01 6.72669232e-01
-8.55692804e-01 3.83488595e-01 2.73875266e-01 8.57512116e-01
-1.25459433e-01 6.81478754e-02 -6.44600272e-01 -9.86985624e-01
-7.90168047e-01 -1.21458209e+00 4.89608526e-01 5.69871247e-01
-2.50776678e-01 -1.00725961e+00 1.58364967e-01 4.70054358e-01
3.81109685e-01 4.27267492e-01 1.14223588e+00 -7.21481264e-01
-6.46021068e-02 -3.56221199e-01 -1.75744027e-01 3.48638147e-01
5.78786373e-01 2.87367284e-01 -4.79664981e-01 -1.57471329e-01
4.67377245e-01 6.15923852e-02 4.27797258e-01 8.29862595e-01
1.05405724e+00 -2.22248897e-01 -6.93628430e-01 6.02859735e-01
1.58864641e+00 1.28959811e+00 9.88469049e-02 2.26727635e-01
6.81481302e-01 2.73028553e-01 9.18444097e-01 7.50961483e-01
3.98099333e-01 1.53280228e-01 2.10551038e-01 -7.87404850e-02
2.31661364e-01 3.59650671e-01 -2.66245812e-01 1.11629117e+00
-5.16679704e-01 -4.73963320e-01 -1.21650219e+00 2.80713975e-01
-1.57097197e+00 -9.74360347e-01 -3.38238657e-01 1.76209807e+00
6.45465791e-01 -9.91530642e-02 -1.14764664e-02 4.15128708e-01
1.04169583e+00 -7.72982597e-01 -5.02508163e-01 -5.70885301e-01
1.74783528e-01 1.78179890e-01 1.42808989e-01 2.15150028e-01
-1.12795675e+00 2.86238641e-01 6.10210705e+00 5.45081496e-01
-1.56820977e+00 -2.02352673e-01 4.84213829e-01 1.92360371e-01
3.06477010e-01 -7.97962606e-01 -7.04867482e-01 6.50249183e-01
5.18007874e-01 -6.15635365e-02 4.87715006e-01 6.19577706e-01
6.28948212e-01 -3.68998170e-01 -2.64055490e-01 1.13532221e+00
1.53701007e-01 -1.05118096e+00 -8.29459727e-02 -1.97374359e-01
9.51767802e-01 -1.90313295e-01 -4.42164543e-04 -1.56277865e-01
2.51621693e-01 -9.36235547e-01 -1.08132884e-01 7.67712295e-01
5.13343692e-01 -1.00327885e+00 1.17792010e+00 5.88807166e-01
-1.13126218e+00 -3.13351512e-01 -3.51123869e-01 7.59733021e-02
1.78608932e-02 4.63406801e-01 -1.26359940e+00 3.13524604e-01
8.88278842e-01 2.09407374e-01 -3.74849796e-01 1.20221496e+00
2.77044386e-01 7.20636845e-01 -5.04664838e-01 -5.88472605e-01
6.54260488e-03 -2.09951296e-01 5.28723478e-01 9.64873672e-01
5.37741601e-01 5.07735372e-01 2.88271815e-01 3.95843327e-01
5.62474847e-01 4.22496051e-01 -3.22428912e-01 7.55344108e-02
5.56226075e-01 7.85772979e-01 -9.30534184e-01 -5.80535650e-01
1.37002423e-01 3.93265665e-01 -2.98000634e-01 9.48346127e-03
-9.15577710e-01 -7.66645849e-01 1.68458149e-01 5.39340600e-02
3.38691682e-01 3.70443314e-01 -4.15856987e-01 -7.23117471e-01
-4.11768526e-01 -9.47250068e-01 7.08810747e-01 -6.05325401e-01
-1.13577652e+00 7.09435880e-01 1.40321538e-01 -9.38425243e-01
-4.33495641e-01 -6.49320185e-01 -7.05747426e-01 8.18593204e-01
-1.05294812e+00 -5.09448230e-01 -3.97112846e-01 5.56690574e-01
3.90139133e-01 -5.16013741e-01 8.48323345e-01 9.57682356e-02
-9.03194964e-01 3.14785779e-01 5.08682787e-01 -2.97774132e-02
4.28547919e-01 -8.92482460e-01 -4.70135629e-01 8.19821417e-01
-8.18202496e-01 6.83983028e-01 5.99195182e-01 -9.37960088e-01
-9.94954467e-01 -1.04388177e+00 7.75101960e-01 -3.91354673e-02
8.87556523e-02 3.92852068e-01 -8.60289574e-01 -2.62164548e-02
-8.49835947e-02 -1.49808720e-01 5.06259084e-01 -8.75780761e-01
5.72353005e-01 9.32022855e-02 -1.68643999e+00 2.22664997e-01
3.96834940e-01 2.71276534e-02 -7.39953041e-01 3.27660173e-01
2.35323504e-01 -3.27518195e-01 -1.14369941e+00 6.21359885e-01
6.57999754e-01 -7.31399179e-01 9.39211488e-01 -1.92098334e-01
-1.07431374e-01 -4.92984474e-01 6.89622387e-02 -8.25235546e-01
-3.35153669e-01 1.63033471e-01 2.38407865e-01 6.89057529e-01
2.56078094e-01 -9.13791955e-01 6.64818525e-01 2.94687122e-01
2.56811827e-01 -1.12935722e+00 -4.28376794e-01 -4.26518828e-01
-4.14540619e-01 2.79044807e-02 5.23838639e-01 9.89618719e-01
-3.43523264e-01 1.97594501e-02 -2.64407881e-02 1.23780549e-01
4.36014563e-01 -3.14790756e-02 2.16699913e-01 -1.62995768e+00
-2.90243477e-01 -2.48063952e-01 -1.83803543e-01 -3.49388033e-01
-4.65142220e-01 -4.94065613e-01 4.72011790e-02 -1.67583418e+00
2.75437236e-01 -6.01070583e-01 -3.89084965e-01 2.30733141e-01
-5.21663308e-01 1.55522183e-01 -3.87817286e-02 3.08630377e-01
-5.87435551e-02 4.98563983e-02 1.42656481e+00 8.91545340e-02
-4.75315541e-01 3.55044365e-01 -2.15602309e-01 9.44301724e-01
1.16209042e+00 -6.08587503e-01 -6.30273700e-01 -2.02936694e-01
-1.81875341e-02 4.00773644e-01 -1.12994969e-01 -9.67420280e-01
5.00532873e-02 -5.94976485e-01 5.56726813e-01 -8.30179036e-01
1.90833285e-01 -1.03873229e+00 2.72803485e-01 1.40048718e+00
7.82489255e-02 3.47135663e-01 3.82620730e-02 4.00950760e-01
-1.75135389e-01 -5.72085500e-01 8.37739348e-01 -1.07713409e-01
-5.08398533e-01 5.38089685e-02 -8.28436613e-01 -1.97243355e-02
1.37503970e+00 -6.00171387e-01 -1.74007013e-01 1.80797651e-01
-5.14467359e-01 2.32921436e-01 2.44025677e-01 1.44761935e-01
7.56237447e-01 -8.93004060e-01 -6.82142675e-01 3.92611772e-01
-1.87637150e-01 8.84787440e-02 3.26669782e-01 1.01019406e+00
-1.24205852e+00 4.53389019e-01 -2.62192011e-01 -8.07981789e-01
-1.66530478e+00 3.93991500e-01 4.19330269e-01 -2.94091821e-01
-1.42436519e-01 8.31489861e-01 -7.23642707e-02 -1.83236361e-01
-7.51418844e-02 -2.49088421e-01 -8.72660339e-01 -2.26146042e-01
4.91920441e-01 8.06648374e-01 -4.62343171e-02 -7.62866616e-01
-6.21487975e-01 9.66267049e-01 1.88869443e-02 3.70269954e-01
1.20402396e+00 3.32708098e-02 -3.68621022e-01 1.32747680e-01
8.52596223e-01 -2.36988872e-01 -4.00842637e-01 2.54049093e-01
-1.90001473e-01 -3.41718137e-01 9.76881236e-02 -1.05254698e+00
-8.95491183e-01 5.72872162e-01 1.22784328e+00 9.77944210e-02
1.41783309e+00 -4.19788241e-01 8.51100922e-01 6.10060990e-01
5.17571308e-02 -1.21042764e+00 -1.59070507e-01 3.06504220e-01
3.99394780e-01 -1.23636889e+00 1.14807740e-01 -3.28177541e-01
-7.45635748e-01 9.91270721e-01 6.80715680e-01 -2.96503574e-01
8.93362224e-01 4.74022567e-01 3.64168435e-01 -2.02208653e-01
-7.63183296e-01 2.41298631e-01 1.37402132e-01 7.17181742e-01
6.24586605e-02 2.29405209e-01 -6.00979149e-01 4.98075962e-01
9.27642360e-03 3.19949985e-01 2.38646969e-01 1.09259677e+00
-9.83917832e-01 -6.42589331e-01 -1.08078563e+00 1.01763606e+00
-9.09639955e-01 3.09198350e-01 5.08561842e-02 6.51491702e-01
4.58177745e-01 1.17493033e+00 1.48423046e-01 -3.67213517e-01
6.03272691e-02 -2.77859300e-01 3.02953094e-01 -3.68651509e-01
-4.49483782e-01 1.33336768e-01 -2.69135773e-01 2.66636074e-01
-5.31963587e-01 -6.40156865e-01 -1.86477923e+00 -1.67224094e-01
-7.25552857e-01 4.25577670e-01 6.61474943e-01 1.02350330e+00
9.45687816e-02 5.75452864e-01 6.77439153e-01 -1.65389359e-01
-3.46146703e-01 -7.16414392e-01 -4.26306456e-01 7.36413300e-02
2.11560443e-01 -7.78522909e-01 -2.63493299e-01 1.59359783e-01]
|
[15.580198287963867, -1.6967624425888062]
|
5dde2ef6-2cc5-4f98-ba2c-494afd0317fd
|
msr-gcn-multi-scale-residual-graph
|
2108.07152
| null |
https://arxiv.org/abs/2108.07152v2
|
https://arxiv.org/pdf/2108.07152v2.pdf
|
MSR-GCN: Multi-Scale Residual Graph Convolution Networks for Human Motion Prediction
|
Human motion prediction is a challenging task due to the stochasticity and aperiodicity of future poses. Recently, graph convolutional network has been proven to be very effective to learn dynamic relations among pose joints, which is helpful for pose prediction. On the other hand, one can abstract a human pose recursively to obtain a set of poses at multiple scales. With the increase of the abstraction level, the motion of the pose becomes more stable, which benefits pose prediction too. In this paper, we propose a novel Multi-Scale Residual Graph Convolution Network (MSR-GCN) for human pose prediction task in the manner of end-to-end. The GCNs are used to extract features from fine to coarse scale and then from coarse to fine scale. The extracted features at each scale are then combined and decoded to obtain the residuals between the input and target poses. Intermediate supervisions are imposed on all the predicted poses, which enforces the network to learn more representative features. Our proposed approach is evaluated on two standard benchmark datasets, i.e., the Human3.6M dataset and the CMU Mocap dataset. Experimental results demonstrate that our method outperforms the state-of-the-art approaches. Code and pre-trained models are available at https://github.com/Droliven/MSRGCN.
|
['Guiqing Li', 'Qing Zhang', 'Chengjiang Long', 'Yongwei Nie', 'Lingwei Dang']
|
2021-08-16
| null |
http://openaccess.thecvf.com//content/ICCV2021/html/Dang_MSR-GCN_Multi-Scale_Residual_Graph_Convolution_Networks_for_Human_Motion_Prediction_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Dang_MSR-GCN_Multi-Scale_Residual_Graph_Convolution_Networks_for_Human_Motion_Prediction_ICCV_2021_paper.pdf
|
iccv-2021-1
|
['human-pose-forecasting']
|
['computer-vision']
|
[-3.08788307e-02 -2.12838605e-01 -8.35596099e-02 -2.01948628e-01
-4.02446330e-01 7.25667924e-02 2.77733624e-01 -3.43158036e-01
-3.09704840e-01 6.13670945e-01 4.49038327e-01 4.34379667e-01
2.10483521e-02 -5.94279945e-01 -6.95143580e-01 -6.06128871e-01
-5.93555085e-02 3.00439596e-01 4.43081766e-01 -3.10025513e-01
-2.00303361e-01 3.68753463e-01 -1.41815329e+00 3.22352052e-02
5.47926784e-01 9.52588499e-01 3.10577154e-01 5.35933435e-01
4.41097796e-01 6.71090186e-01 -2.68799365e-01 -1.85865909e-01
1.51755452e-01 -2.47593224e-01 -5.30756176e-01 5.14502265e-02
2.18012363e-01 -3.53281379e-01 -9.22062814e-01 9.43540990e-01
6.03111148e-01 5.81840038e-01 3.51936638e-01 -1.04717600e+00
-2.80562669e-01 2.75834382e-01 -5.80246925e-01 1.10137098e-01
3.22884053e-01 2.82513142e-01 9.19924200e-01 -9.66409504e-01
7.81319797e-01 1.49785924e+00 4.63904172e-01 6.67558253e-01
-7.92215407e-01 -6.28743470e-01 3.65621686e-01 4.43752289e-01
-1.41153824e+00 -1.16225466e-01 9.66897070e-01 -3.76746148e-01
7.53924131e-01 7.95767754e-02 8.12511384e-01 1.19862390e+00
4.86318260e-01 9.24416363e-01 5.51378906e-01 9.69975144e-02
-1.04125805e-01 -7.21765161e-01 -1.21617712e-01 9.95080948e-01
9.94415954e-02 1.07454248e-01 -7.28958309e-01 9.61089209e-02
1.01964974e+00 3.61879706e-01 -5.82283497e-01 -4.17383313e-01
-1.39573634e+00 6.15289211e-01 1.00891292e+00 1.08307578e-01
-5.65327764e-01 4.99792576e-01 3.68426055e-01 -1.93742141e-01
3.45344454e-01 -1.34278044e-01 -4.48949307e-01 -1.59713611e-01
-5.07983923e-01 3.92529994e-01 4.14917558e-01 7.23258495e-01
5.97368956e-01 1.48399714e-02 -1.48046568e-01 6.68443859e-01
4.80896384e-01 3.65763634e-01 4.93133724e-01 -8.13976467e-01
7.52745926e-01 6.32780313e-01 4.90626320e-02 -1.38428330e+00
-7.71315932e-01 -6.29899383e-01 -1.01238310e+00 -3.42633277e-02
2.52682775e-01 -1.48405716e-01 -1.00986028e+00 1.74034679e+00
5.91613770e-01 4.03882861e-01 -1.79848447e-01 1.37426770e+00
8.49071264e-01 8.40797722e-01 -4.63788509e-02 9.60271806e-02
1.22866404e+00 -1.15042233e+00 -6.02592111e-01 -4.13092077e-01
3.83502185e-01 -4.89036769e-01 8.81986976e-01 2.52353340e-01
-7.67449617e-01 -9.55315173e-01 -9.34046626e-01 -9.39101353e-02
1.15537159e-01 3.97750080e-01 5.32835186e-01 -5.55188023e-02
-7.20764399e-01 8.86627197e-01 -1.34574258e+00 -1.78177580e-01
2.56476402e-01 3.65325063e-01 -3.43531042e-01 -1.86863784e-02
-1.29619730e+00 7.13742614e-01 5.17455757e-01 6.83040977e-01
-7.36017644e-01 -2.91043460e-01 -1.01611423e+00 -1.18004970e-01
4.25024718e-01 -7.80027330e-01 1.01567602e+00 -5.48072159e-01
-1.36093092e+00 3.42039317e-01 -9.72137321e-03 -2.65559137e-01
7.45613039e-01 -6.31352365e-01 -3.04778427e-01 1.64392501e-01
1.57565996e-01 7.65005529e-01 8.03939700e-01 -9.20832872e-01
-6.57796264e-01 -4.08910841e-01 4.80876624e-04 5.85168600e-01
-2.15444520e-01 -3.75764459e-01 -1.01028478e+00 -7.42176473e-01
2.51851708e-01 -1.22712076e+00 -4.58194643e-01 -1.59196720e-01
-4.27287251e-01 -2.60169208e-01 6.82150960e-01 -9.80832040e-01
1.31478083e+00 -2.08002043e+00 8.68369102e-01 7.28673562e-02
2.31025383e-01 9.91918147e-02 -6.77054301e-02 2.41665155e-01
9.15315840e-03 -3.78733337e-01 -1.07354507e-01 -3.79721820e-01
-1.34076253e-01 1.03727907e-01 1.32508978e-01 5.70288301e-01
1.08406276e-01 1.04100120e+00 -8.64691973e-01 -4.89794582e-01
3.67002517e-01 6.80013418e-01 -5.21524429e-01 1.84340134e-01
-1.94931135e-01 8.01271737e-01 -7.29346991e-01 4.19080347e-01
2.89422214e-01 -3.96261722e-01 -3.79490890e-02 -3.97092521e-01
2.91674763e-01 1.23343661e-01 -1.22371113e+00 2.08087516e+00
-2.80254275e-01 2.48006076e-01 -3.12628269e-01 -5.98133266e-01
8.33073854e-01 1.97237983e-01 4.70617920e-01 -3.73779565e-01
3.20479453e-01 3.44570205e-02 7.16688037e-02 -3.76002550e-01
3.62291873e-01 2.38993630e-01 -2.64974564e-01 -1.34995371e-01
-2.85527129e-02 3.55085969e-01 1.81422412e-01 -3.71970609e-02
8.94313812e-01 5.83151937e-01 2.25584283e-01 3.41140628e-02
7.51659393e-01 -2.16787055e-01 9.42403734e-01 8.52459967e-02
-2.96627760e-01 6.06268406e-01 2.23484844e-01 -6.06897414e-01
-6.87819004e-01 -1.10717261e+00 4.38835531e-01 9.29960132e-01
3.50655526e-01 -3.91809553e-01 -7.48960853e-01 -6.06498361e-01
-1.02166355e-01 1.88830584e-01 -5.63028276e-01 -3.91048074e-01
-1.03641498e+00 -3.84469569e-01 1.37928754e-01 8.25350165e-01
6.59862697e-01 -1.36903369e+00 -7.02576995e-01 3.03745180e-01
-5.55207491e-01 -1.08692920e+00 -7.61699557e-01 -3.73819679e-01
-8.03158522e-01 -9.45248067e-01 -8.11127841e-01 -9.12265122e-01
5.32805026e-01 2.87574083e-02 7.70105183e-01 1.85389772e-01
-2.78941989e-01 1.76820591e-01 -2.81698316e-01 3.72608453e-02
1.75442677e-02 1.84591874e-01 2.53116786e-01 2.24289447e-01
6.41645715e-02 -5.89160860e-01 -9.65627551e-01 3.03698391e-01
-6.04315400e-01 1.66276932e-01 7.43297994e-01 8.30703020e-01
8.27989638e-01 -1.43633988e-02 3.17629069e-01 -4.39354718e-01
3.63526136e-01 -2.30950072e-01 -3.77555430e-01 -1.82336895e-03
4.38350998e-02 7.69950598e-02 6.65263593e-01 -5.25547624e-01
-9.98588979e-01 4.55372840e-01 -7.35469535e-02 -7.27365434e-01
-1.34287000e-01 7.02452481e-01 -3.05663675e-01 1.92030653e-01
4.46971089e-01 2.89189756e-01 -6.13515526e-02 -5.38837552e-01
2.13682622e-01 2.43476525e-01 7.62082398e-01 -3.71505082e-01
9.37307656e-01 3.25497061e-01 2.76227415e-01 -7.16931999e-01
-7.62957454e-01 -4.86822158e-01 -7.75553644e-01 -6.28573895e-01
1.14471197e+00 -1.02155852e+00 -6.07275963e-01 6.21818423e-01
-1.14606726e+00 -3.10284078e-01 1.91271812e-01 7.09062874e-01
-6.35718226e-01 4.02215570e-01 -8.15603614e-01 -5.20904720e-01
-4.13634419e-01 -1.18458247e+00 1.20361531e+00 3.92865062e-01
-3.02005529e-01 -8.02922010e-01 -6.14468344e-02 3.56711328e-01
-1.61947995e-01 6.86299026e-01 3.97603720e-01 -1.85216501e-01
-6.95808053e-01 -4.09357905e-01 1.57914862e-01 3.04776654e-02
9.04585347e-02 -2.60696411e-01 -4.28078622e-01 -5.45586526e-01
-1.70045629e-01 -3.51354718e-01 8.92867446e-01 3.95648688e-01
1.06269431e+00 -1.45880759e-01 -4.63380277e-01 6.24872983e-01
1.11414135e+00 -4.83022518e-02 5.97552776e-01 2.49556795e-01
1.23682177e+00 5.61380088e-01 9.31541562e-01 3.73062640e-01
4.33825433e-01 8.24200869e-01 4.23947722e-01 1.01646617e-01
-1.47566229e-01 -5.16719699e-01 4.01800901e-01 1.01443315e+00
-4.28530872e-01 6.16013929e-02 -9.49589252e-01 2.58685142e-01
-2.22942519e+00 -7.87373185e-01 -4.16264609e-02 1.92896771e+00
5.38013875e-01 3.67406696e-01 1.59776941e-01 -8.71078148e-02
8.06922257e-01 4.41647798e-01 -6.57153487e-01 2.47978717e-01
3.88085693e-01 3.47784981e-02 2.59175658e-01 3.56633842e-01
-1.26754963e+00 1.03296816e+00 4.16735172e+00 7.31854856e-01
-1.04803467e+00 -1.49638161e-01 3.85942847e-01 -1.97598621e-01
3.07812989e-01 -2.84908742e-01 -7.24944413e-01 3.93089890e-01
4.84711885e-01 -5.62520325e-02 4.16319102e-01 8.13287854e-01
2.05759734e-01 1.29583910e-01 -1.03969538e+00 1.02366495e+00
-1.23901732e-01 -9.57763433e-01 1.20123327e-01 -1.35066614e-01
5.34698963e-01 -2.96069589e-03 -3.05296388e-02 2.92071134e-01
2.54453328e-02 -9.11908448e-01 7.79762864e-01 7.96966910e-01
5.51045477e-01 -1.04520750e+00 6.53089583e-01 5.25817215e-01
-1.73407006e+00 3.77374627e-02 -3.78287911e-01 -1.47297606e-01
2.90729940e-01 2.50722408e-01 -3.68560851e-01 8.19430172e-01
8.04910600e-01 1.03580642e+00 -6.39793038e-01 8.21401417e-01
-4.89119083e-01 2.50259161e-01 -1.44534871e-01 -2.19610080e-01
1.66351736e-01 -6.96131065e-02 5.78982651e-01 8.06757808e-01
2.87095010e-01 1.07195362e-01 4.90616590e-01 4.87244606e-01
2.62971222e-02 -5.05903102e-02 -3.55671734e-01 1.70003131e-01
2.61671692e-01 1.37010455e+00 -6.54783607e-01 -1.30899698e-01
-3.10189784e-01 1.20722044e+00 4.99214709e-01 3.84108514e-01
-9.94244874e-01 -2.85156995e-01 6.51251853e-01 -1.54763997e-01
3.67449999e-01 -4.39831138e-01 1.25105932e-01 -1.26415908e+00
3.01104277e-01 -7.27161407e-01 4.40294832e-01 -5.60071290e-01
-1.07229555e+00 5.58726966e-01 -1.13306291e-01 -1.51549089e+00
-4.11358505e-01 -4.87490445e-01 -5.13474464e-01 7.57667363e-01
-9.45733666e-01 -1.13581645e+00 -5.14377117e-01 6.42012954e-01
5.10336339e-01 1.67202607e-01 4.84024107e-01 2.05326557e-01
-5.95948100e-01 4.56832826e-01 -3.37007344e-01 5.93796909e-01
4.52205360e-01 -9.32875752e-01 6.59783065e-01 7.90622115e-01
1.75228156e-02 5.10065436e-01 6.65198386e-01 -1.00817716e+00
-1.35027969e+00 -1.41464245e+00 6.36038721e-01 -3.20012540e-01
5.89281023e-01 -3.27735215e-01 -7.32757032e-01 6.99854016e-01
-3.14257473e-01 1.66266352e-01 1.70255095e-01 -2.07662582e-01
-9.38850343e-02 -8.61255229e-02 -5.69620550e-01 9.23815429e-01
1.37070370e+00 -2.37015411e-01 -5.14555871e-01 2.35475317e-01
9.16782260e-01 -7.47218013e-01 -8.86184573e-01 6.50681376e-01
6.60451770e-01 -7.66144156e-01 1.01884997e+00 -4.75845516e-01
6.53683960e-01 -4.33155358e-01 -7.68112093e-02 -1.33690155e+00
-6.37679458e-01 -4.65002239e-01 -2.60052234e-01 8.30354810e-01
1.82820365e-01 -3.78515095e-01 9.31872725e-01 2.52139628e-01
-1.27491385e-01 -1.11708117e+00 -1.03609657e+00 -7.21865714e-01
-2.26743326e-01 -1.84069842e-01 3.62595439e-01 4.16756928e-01
-1.55809090e-01 5.07084429e-01 -6.99243069e-01 3.70168865e-01
7.15338349e-01 1.93525329e-01 9.05791521e-01 -1.05759668e+00
-4.39714193e-01 -2.74850756e-01 -7.81654298e-01 -1.27663159e+00
9.26701650e-02 -7.62355924e-01 2.29883775e-01 -1.63253570e+00
1.03547595e-01 1.54347464e-01 -3.38859260e-01 2.44295552e-01
-5.70310950e-01 1.31709278e-01 3.71591151e-01 2.57115901e-01
-6.87623620e-01 8.10836554e-01 1.58124542e+00 -7.64477998e-02
-3.21723998e-01 1.16026849e-01 -3.01195439e-02 9.36766028e-01
9.83334720e-01 -2.75554031e-01 -3.17285866e-01 -2.27525607e-01
-1.76107347e-01 3.15867633e-01 5.69636524e-01 -1.55022752e+00
1.38477027e-01 -2.91069839e-02 7.73674905e-01 -9.35986936e-01
6.35021746e-01 -6.14542723e-01 3.00186485e-01 9.17049527e-01
-3.12869906e-01 5.08965291e-02 -2.97786109e-02 8.84392560e-01
-1.42103299e-01 2.81728983e-01 6.25676215e-01 -1.17122211e-01
-9.51467156e-01 8.56077313e-01 1.23786755e-01 -1.16371280e-02
9.21990752e-01 4.06021215e-02 6.78767413e-02 -4.16927874e-01
-9.90760028e-01 4.53835547e-01 3.44185114e-01 7.73140430e-01
8.68876398e-01 -1.74829841e+00 -5.58484733e-01 -1.64078653e-01
7.90568441e-02 2.91779637e-01 5.60862839e-01 8.74248028e-01
-5.30655324e-01 3.48175943e-01 -2.94264644e-01 -6.83781624e-01
-1.25803590e+00 5.37896991e-01 2.62014985e-01 -4.30222839e-01
-8.20918322e-01 7.90552318e-01 2.93900549e-01 -3.65324318e-01
4.17980582e-01 -3.56286645e-01 -3.52256209e-01 -1.77618876e-01
4.22163248e-01 3.99095476e-01 -2.61471897e-01 -1.05263841e+00
-5.67026436e-01 7.39495635e-01 7.04809055e-02 4.69322354e-02
1.33365655e+00 -6.97702095e-02 4.25249897e-02 4.25749451e-01
1.34066105e+00 -3.27989519e-01 -1.58088052e+00 -3.35131466e-01
-8.76869261e-02 -3.39080274e-01 -3.42963517e-01 -3.35679740e-01
-1.22794569e+00 8.61606956e-01 4.83347088e-01 -3.86931032e-01
1.14751744e+00 -2.87582092e-02 1.10791504e+00 3.67556214e-01
6.05827272e-01 -1.07607543e+00 4.09377933e-01 5.27446210e-01
1.15328336e+00 -8.80216479e-01 4.40352038e-02 -5.10861278e-01
-7.27855921e-01 1.07170486e+00 9.50951636e-01 -3.61314118e-01
5.50172806e-01 -2.47417375e-01 -7.41009191e-02 -1.29717946e-01
-5.00872254e-01 -2.38365144e-01 7.22673953e-01 4.50296998e-01
3.81313920e-01 2.04063326e-01 -3.37396711e-01 7.23826230e-01
-2.50484765e-01 1.08311586e-01 1.03505850e-01 9.55955386e-01
-4.55469191e-01 -7.81816900e-01 -3.93014342e-01 3.29298407e-01
-2.72541672e-01 2.27107272e-01 -3.61969501e-01 8.03450286e-01
1.10796534e-01 7.14083970e-01 -2.72451997e-01 -8.38236392e-01
5.00577092e-01 -8.20824727e-02 3.97496253e-01 -5.75832903e-01
-2.39912421e-01 3.02355319e-01 1.10849768e-01 -1.02859390e+00
-3.86389703e-01 -7.53933251e-01 -1.70960462e+00 -3.69850881e-02
-5.82749359e-02 -1.14307739e-01 4.16880809e-02 7.40355909e-01
2.68587798e-01 8.43651354e-01 3.89064103e-01 -1.30998266e+00
-6.08296573e-01 -9.96538579e-01 -3.59905124e-01 6.66329026e-01
2.00412109e-01 -9.84533131e-01 -6.98828995e-02 1.07549159e-02]
|
[7.29472017288208, -0.38634833693504333]
|
083f69cf-1664-4c2d-a46a-c55e91654ba4
|
scenehgn-hierarchical-graph-networks-for-3d
|
2302.10237
| null |
https://arxiv.org/abs/2302.10237v1
|
https://arxiv.org/pdf/2302.10237v1.pdf
|
SceneHGN: Hierarchical Graph Networks for 3D Indoor Scene Generation with Fine-Grained Geometry
|
3D indoor scenes are widely used in computer graphics, with applications ranging from interior design to gaming to virtual and augmented reality. They also contain rich information, including room layout, as well as furniture type, geometry, and placement. High-quality 3D indoor scenes are highly demanded while it requires expertise and is time-consuming to design high-quality 3D indoor scenes manually. Existing research only addresses partial problems: some works learn to generate room layout, and other works focus on generating detailed structure and geometry of individual furniture objects. However, these partial steps are related and should be addressed together for optimal synthesis. We propose SCENEHGN, a hierarchical graph network for 3D indoor scenes that takes into account the full hierarchy from the room level to the object level, then finally to the object part level. Therefore for the first time, our method is able to directly generate plausible 3D room content, including furniture objects with fine-grained geometry, and their layout. To address the challenge, we introduce functional regions as intermediate proxies between the room and object levels to make learning more manageable. To ensure plausibility, our graph-based representation incorporates both vertical edges connecting child nodes with parent nodes from different levels, and horizontal edges encoding relationships between nodes at the same level. Extensive experiments demonstrate that our method produces superior generation results, even when comparing results of partial steps with alternative methods that can only achieve these. We also demonstrate that our method is effective for various applications such as part-level room editing, room interpolation, and room generation by arbitrary room boundaries.
|
['Jie Yang', 'Leonidas J. Guibas', 'Yu-Kun Lai', 'Kaichun Mo', 'Jia-Mu Sun', 'Lin Gao']
|
2023-02-16
| null | null | null | null |
['scene-generation']
|
['computer-vision']
|
[ 3.10209841e-01 1.65667951e-01 3.96024525e-01 -2.85804600e-01
-3.09264034e-01 -6.49602711e-01 4.00967270e-01 2.68817216e-01
4.18535143e-01 5.62579036e-01 3.32862526e-01 -4.19111818e-01
-9.11608711e-02 -1.35080945e+00 -8.01042557e-01 -4.45317686e-01
-4.52528447e-02 4.56198990e-01 2.05729097e-01 -3.88254851e-01
-5.41742556e-02 8.52089703e-01 -1.68262827e+00 1.38594463e-01
1.20642745e+00 5.83011270e-01 4.51124907e-01 6.40792251e-01
-3.60127747e-01 3.84158909e-01 -4.94345367e-01 -7.70290345e-02
4.84308541e-01 -4.45923746e-01 -4.25911248e-01 4.49055582e-01
6.95189774e-01 -4.03470337e-01 -1.66564777e-01 6.62488639e-01
4.95351970e-01 3.82996380e-01 4.28357035e-01 -1.12284195e+00
-6.20667338e-01 4.84129339e-01 -2.94906169e-01 -5.98678470e-01
7.27320969e-01 -1.87076069e-02 8.59450996e-01 -7.74343193e-01
3.94668639e-01 1.32572675e+00 5.23116112e-01 5.53348243e-01
-1.14093304e+00 -4.74699408e-01 6.73464656e-01 -3.06168467e-01
-1.36843479e+00 -1.88126639e-01 1.05606675e+00 -2.92782933e-01
5.15513062e-01 6.02826238e-01 1.10889077e+00 7.72274315e-01
-9.61788073e-02 6.70635283e-01 9.90468323e-01 -3.63371670e-01
4.04652178e-01 3.88013795e-02 -1.92738786e-01 1.03927207e+00
2.25677803e-01 -1.72128215e-01 -1.28876656e-01 2.76852511e-02
1.44396925e+00 2.99510270e-01 -4.65455830e-01 -8.34988713e-01
-1.30186307e+00 5.20496249e-01 9.19429004e-01 1.80283472e-01
-8.45267922e-02 1.67010397e-01 -2.42842510e-01 -1.45074069e-01
1.56110615e-01 6.69322312e-01 -3.13483834e-01 3.87567103e-01
-7.49260187e-01 4.31671530e-01 6.83045030e-01 1.64690614e+00
8.76708865e-01 -1.46040441e-02 -4.42061961e-01 8.41220677e-01
2.62418747e-01 3.49827230e-01 -3.53868693e-01 -1.13191390e+00
5.95762432e-01 7.10179687e-01 2.48428375e-01 -1.14414144e+00
-4.25623983e-01 -3.62440795e-01 -1.18060374e+00 8.21720064e-02
2.20466733e-01 7.35728517e-02 -1.10967124e+00 1.55455816e+00
6.11643553e-01 3.92983347e-01 -4.08391088e-01 7.80176818e-01
1.19486785e+00 7.99681604e-01 -1.23975299e-01 3.91219929e-02
1.28925860e+00 -1.25699174e+00 -5.43380141e-01 -2.74656802e-01
4.44314718e-01 -6.08244538e-01 1.36014009e+00 1.05283044e-01
-1.25324667e+00 -7.03459442e-01 -9.35768247e-01 -2.31236503e-01
-2.62113750e-01 -2.00447477e-02 9.38461125e-01 7.77742028e-01
-1.11995697e+00 3.96829724e-01 -5.38847625e-01 -1.26855761e-01
3.16204011e-01 3.55851740e-01 -1.74372867e-01 -4.01680559e-01
-7.22918987e-01 4.17135328e-01 -1.09040521e-01 3.36707592e-01
-6.24855042e-01 -8.72530758e-01 -1.33292472e+00 2.18726471e-01
4.70950037e-01 -1.21226656e+00 1.03225672e+00 -1.85628787e-01
-1.31347275e+00 4.80826080e-01 -2.49100685e-01 2.97605425e-01
3.15923959e-01 1.17628291e-01 -6.81451783e-02 -3.73044193e-01
2.87852921e-02 6.04668498e-01 4.12823290e-01 -1.87981987e+00
-4.58077431e-01 -3.45753431e-01 5.34090042e-01 3.87521476e-01
7.98272621e-03 -7.63224602e-01 -7.09848106e-01 -5.05949020e-01
7.88551450e-01 -8.10867131e-01 -7.29121268e-01 1.50525644e-01
-6.55588865e-01 -3.54942195e-02 5.05863369e-01 -5.21959662e-01
1.18857789e+00 -2.16345263e+00 7.03671426e-02 5.12446702e-01
2.37768129e-01 -3.27731788e-01 -3.78857227e-03 3.44716609e-01
1.40977830e-01 3.32056731e-01 -2.55141079e-01 -6.91799879e-01
1.88342988e-01 2.30889201e-01 -3.31810787e-02 -9.07404348e-03
-2.34351873e-01 7.72031903e-01 -1.05513096e+00 -3.85596633e-01
7.12846875e-01 6.78462625e-01 -9.03439641e-01 1.85264304e-01
-1.59052610e-01 6.53113365e-01 -5.53037643e-01 5.78437209e-01
8.16749454e-01 -1.85359597e-01 1.01585388e-01 -3.33617866e-01
-9.97161940e-02 4.53676313e-01 -1.50455844e+00 1.96576154e+00
-1.02409101e+00 1.47028461e-01 2.92018026e-01 -2.18769789e-01
1.05000377e+00 9.95269641e-02 2.80718148e-01 -4.44795996e-01
-1.10767007e-01 6.76270276e-02 -4.14967149e-01 -9.01228637e-02
7.99640834e-01 1.89459287e-02 -3.36721033e-01 6.43433481e-02
-4.98974323e-01 -9.56970572e-01 -1.59672990e-01 2.14383766e-01
8.39892030e-01 2.78136373e-01 2.17990920e-01 -4.55157943e-02
3.21334541e-01 -3.01530600e-01 3.27739000e-01 7.97664344e-01
3.22383106e-01 8.67600799e-01 9.90342721e-02 -3.78499180e-01
-1.00836253e+00 -1.24893057e+00 3.89380567e-02 6.66455150e-01
5.98216712e-01 -5.34575820e-01 -8.47859561e-01 -2.81567574e-01
-1.07620046e-01 8.73158932e-01 -5.73938966e-01 9.29637179e-02
-7.35912502e-01 -3.10429782e-01 -1.85309619e-01 5.98731816e-01
4.95919794e-01 -9.25759912e-01 -4.49196190e-01 1.24576591e-01
-2.13594764e-01 -1.20322537e+00 -6.89671397e-01 6.97979331e-02
-9.66125011e-01 -7.02499509e-01 -5.48832715e-01 -9.38408494e-01
1.12369454e+00 6.22673810e-01 1.31672323e+00 4.93748665e-01
-3.81006092e-01 4.00759220e-01 -1.50112465e-01 -8.52787048e-02
-3.19488943e-01 8.46903473e-02 -2.88425684e-01 -2.22710416e-01
-5.79402626e-01 -8.95243526e-01 -7.38629699e-01 4.76539075e-01
-7.08056748e-01 7.23614275e-01 1.99296921e-01 4.03040320e-01
8.30184340e-01 3.03339124e-01 -5.46493530e-02 -9.11910415e-01
3.67892206e-01 1.80047490e-02 -6.12985790e-01 2.30799437e-01
-1.28538758e-01 -7.02266023e-02 7.50832617e-01 1.04413452e-02
-1.09835505e+00 1.76856026e-01 -2.12869763e-01 -2.25829884e-01
-4.06176090e-01 -1.32626683e-01 -8.84943962e-01 5.39273992e-02
3.67335677e-01 -1.33819701e-02 -6.84470952e-01 -3.64433050e-01
4.91204560e-01 1.17826127e-01 3.18534672e-01 -9.26286519e-01
1.02946854e+00 3.45775664e-01 2.13446453e-01 -7.40667343e-01
-7.75125384e-01 -1.48907542e-01 -7.75146008e-01 -2.28461862e-01
7.90646076e-01 -8.39793861e-01 -7.11002231e-01 1.06541947e-01
-1.20051455e+00 -6.41699672e-01 -4.79670525e-01 6.53282031e-02
-4.66316044e-01 -3.85492593e-02 -4.77639288e-01 -8.99983108e-01
1.31682605e-01 -1.16286421e+00 1.40933871e+00 1.89200893e-01
-8.31317529e-02 -1.02607024e+00 -3.30377370e-01 3.96619976e-01
1.58812329e-01 5.93471706e-01 1.16549778e+00 3.47686768e-01
-1.41074979e+00 1.56901747e-01 4.85750064e-02 -2.16543391e-01
6.04565322e-01 -2.40009222e-02 -8.58294487e-01 -9.39969197e-02
-4.49087530e-01 2.44741678e-01 4.59563345e-01 4.42210734e-01
1.63001704e+00 -2.93990970e-01 -4.82101858e-01 8.29386532e-01
1.33710086e+00 8.74527842e-02 7.34455884e-01 9.00178105e-02
1.32756138e+00 8.30737710e-01 3.37952197e-01 3.61704409e-01
6.81970596e-01 7.50537038e-01 6.69026077e-01 -3.58213723e-01
-3.14773530e-01 -7.34744847e-01 -1.88760355e-01 8.16176057e-01
-2.95995057e-01 -4.05657470e-01 -5.81939995e-01 3.64921093e-01
-1.58271289e+00 -8.17987263e-01 -3.75454038e-01 2.49675894e+00
4.19127166e-01 1.63218305e-02 -1.28996372e-01 2.19219893e-01
4.64008749e-01 2.20127866e-01 -3.46286952e-01 -3.79616559e-01
8.30392465e-02 1.65696338e-01 2.11645037e-01 7.61090279e-01
-7.61381984e-01 9.35735226e-01 5.86599064e+00 4.59568590e-01
-5.34507453e-01 -2.26232022e-01 6.87425971e-01 8.20253626e-04
-1.09310019e+00 -8.68256837e-02 -6.87066019e-01 7.77616948e-02
1.28684312e-01 2.64854848e-01 7.60425985e-01 8.24760616e-01
3.51117224e-01 -1.63213797e-02 -1.26963735e+00 1.06313515e+00
-8.33907872e-02 -1.39596391e+00 2.12539747e-01 8.46762955e-02
1.10507226e+00 -7.89868355e-01 -4.68136221e-02 2.69593179e-01
3.90713155e-01 -1.14651072e+00 8.48635018e-01 2.58615166e-01
6.99855983e-01 -8.89406562e-01 1.75409645e-01 4.28701937e-01
-1.84095693e+00 2.08671018e-01 -2.15696305e-01 -1.75406300e-02
3.17283005e-01 5.85244954e-01 -8.88871968e-01 6.76516652e-01
6.13079607e-01 3.66027027e-01 -3.14880848e-01 9.90530849e-01
-4.62892711e-01 2.57082470e-02 -2.95508236e-01 1.26584861e-02
4.98862267e-02 -3.81311744e-01 2.00694472e-01 7.42146671e-01
6.26106679e-01 2.64544189e-01 5.26148796e-01 1.12797260e+00
-1.26152843e-01 1.45587120e-02 -8.79446983e-01 6.91903651e-01
7.07595468e-01 1.30331075e+00 -1.06411815e+00 -2.36088842e-01
-8.84541124e-02 1.10700488e+00 1.93487689e-01 4.61687952e-01
-9.77590084e-01 -1.31384179e-01 5.91770291e-01 5.80481887e-01
1.90676287e-01 -6.44943297e-01 -6.59678340e-01 -8.35989177e-01
4.90502343e-02 -5.40784538e-01 -1.40016317e-01 -1.01383936e+00
-9.44659233e-01 5.70412755e-01 6.06392138e-02 -1.05294681e+00
1.18746854e-01 -3.36372524e-01 -4.32895988e-01 7.98006117e-01
-1.33071637e+00 -1.27518487e+00 -9.03139412e-01 6.51260257e-01
6.63066864e-01 6.26729786e-01 8.70330632e-01 2.96512276e-01
-3.82111967e-01 4.85887349e-01 -3.82360399e-01 -2.27605216e-02
1.51623800e-01 -1.31229079e+00 6.35420978e-01 6.19534791e-01
1.32363632e-01 7.42598593e-01 5.21511674e-01 -6.36538506e-01
-1.41262615e+00 -1.26342964e+00 6.94829226e-01 -5.87270617e-01
-1.26490027e-01 -8.64918888e-01 -6.08053982e-01 5.49650311e-01
-2.64399469e-01 -1.12257503e-01 4.84600246e-01 1.93335250e-01
-8.25022012e-02 -9.15265828e-02 -1.15896773e+00 1.09996498e+00
1.78250241e+00 -2.08762914e-01 -9.19728652e-02 3.47124845e-01
1.00262511e+00 -8.47881436e-01 -6.90719783e-01 5.64449131e-01
3.30193907e-01 -1.21039140e+00 1.41141450e+00 9.45437327e-02
4.33403969e-01 -5.88270307e-01 -2.07266599e-01 -1.34842324e+00
-5.24148703e-01 -5.61831832e-01 -6.99076504e-02 1.05489612e+00
4.48869854e-01 -2.40735322e-01 8.84706318e-01 9.11941350e-01
-7.28939950e-01 -7.80476570e-01 -5.20318866e-01 -4.91345555e-01
-4.05883461e-01 -4.48994577e-01 1.28101957e+00 7.92309642e-01
-6.45165443e-01 3.37576568e-01 -3.17314088e-01 4.87898976e-01
6.39342546e-01 6.47171021e-01 1.08082044e+00 -1.34700310e+00
-2.26618141e-01 -4.76156086e-01 1.19112156e-01 -1.61939311e+00
-1.29619867e-01 -6.90166593e-01 3.14872831e-01 -2.43769598e+00
-1.31267488e-01 -1.20410538e+00 3.50854248e-01 2.97556072e-01
-2.19506770e-01 1.00526772e-01 1.34476826e-01 -2.26989150e-01
-3.79327387e-01 6.71327770e-01 2.15613699e+00 -7.56673589e-02
-7.42393315e-01 2.50543296e-01 -7.59348154e-01 7.76430011e-01
6.04359150e-01 7.13414475e-02 -7.15517521e-01 -6.51876330e-01
4.19126684e-03 1.15564406e-01 4.48964983e-01 -9.99852479e-01
-1.00450609e-02 -2.95241922e-01 5.40671170e-01 -7.50557303e-01
7.17904210e-01 -1.15782857e+00 4.56711590e-01 1.57849461e-01
-1.81911588e-02 -1.54826045e-02 1.22273922e-01 4.73123729e-01
2.34236196e-01 1.54368326e-01 4.21972752e-01 -4.61585611e-01
-4.49665308e-01 7.05985546e-01 -3.14204916e-02 -3.81669641e-01
9.59803581e-01 -6.97533429e-01 1.11962989e-01 -4.90020692e-01
-7.11843491e-01 2.63574541e-01 8.63423526e-01 4.97109413e-01
7.99850583e-01 -1.46897411e+00 -3.58736843e-01 4.74337280e-01
1.18062012e-02 9.28132236e-01 5.36139786e-01 1.29905492e-01
-5.57019591e-01 2.09865764e-01 1.63066059e-01 -6.44916534e-01
-1.15090811e+00 5.16928315e-01 2.40324289e-01 -2.04628706e-01
-6.82301164e-01 8.22657108e-01 6.75809920e-01 -9.20498669e-01
3.23888838e-01 -8.45188260e-01 -5.41194752e-02 -3.31194729e-01
1.30198509e-01 2.44818836e-01 3.78959179e-02 -3.95493597e-01
-1.62649721e-01 9.02704597e-01 6.07399344e-01 1.94638088e-01
1.12631989e+00 -2.45492741e-01 -1.82599694e-01 4.42110956e-01
6.60293877e-01 4.54534918e-01 -1.26613533e+00 9.85645801e-02
-6.04722381e-01 -7.25826263e-01 -1.80783138e-01 -5.01130223e-01
-9.21672344e-01 8.01627517e-01 1.28473818e-01 1.49091139e-01
1.07549059e+00 5.21166809e-02 7.73505270e-01 1.68770984e-01
1.05425084e+00 -7.28451550e-01 1.00192353e-01 4.15926695e-01
9.00736451e-01 -7.85639167e-01 7.48138875e-02 -1.31264675e+00
-4.92710844e-02 8.80357623e-01 9.08048451e-01 3.76212820e-02
5.55901527e-01 2.65030921e-01 -4.04358864e-01 -1.39164537e-01
-1.88984931e-01 -6.27208203e-02 4.26591575e-01 6.08892798e-01
5.14327347e-01 3.98382723e-01 3.10451031e-01 3.18072170e-01
-6.44224048e-01 -4.98637915e-01 3.09161276e-01 8.99522424e-01
-4.35223728e-01 -1.23874831e+00 -6.78716779e-01 2.23158643e-01
3.27921689e-01 -1.82866544e-01 -4.78709154e-02 7.63707459e-01
4.13244724e-01 8.75641227e-01 7.11410269e-02 -2.72850782e-01
7.90286422e-01 -4.52245712e-01 8.27305794e-01 -1.03138328e+00
-4.44043070e-01 4.03428935e-02 5.47377355e-02 -5.22040904e-01
-2.84546435e-01 -3.43468249e-01 -1.33056355e+00 -4.49076653e-01
-2.21694350e-01 1.52903005e-01 6.15807176e-01 4.96571064e-01
3.20159674e-01 1.07765436e+00 7.29113221e-01 -1.31379437e+00
2.70748943e-01 -3.99955153e-01 -5.91729105e-01 3.08305591e-01
2.61578739e-01 -5.79969764e-01 -1.36948854e-01 -1.19775042e-01]
|
[9.147400856018066, -3.0214970111846924]
|
77591232-80b7-455e-9d07-7e70d66d4a45
|
weakly-supervised-temporal-action
| null | null |
http://openaccess.thecvf.com/content_ICCV_2019/html/Liu_Weakly_Supervised_Temporal_Action_Localization_Through_Contrast_Based_Evaluation_Networks_ICCV_2019_paper.html
|
http://openaccess.thecvf.com/content_ICCV_2019/papers/Liu_Weakly_Supervised_Temporal_Action_Localization_Through_Contrast_Based_Evaluation_Networks_ICCV_2019_paper.pdf
|
Weakly Supervised Temporal Action Localization Through Contrast Based Evaluation Networks
|
Weakly-supervised temporal action localization (WS-TAL) is a promising but challenging task with only video-level action categorical labels available during training. Without requiring temporal action boundary annotations in training data, WS-TAL could possibly exploit automatically retrieved video tags as video-level labels. However, such coarse video-level supervision inevitably incurs confusions, especially in untrimmed videos containing multiple action instances. To address this challenge, we propose the Contrast-based Localization EvaluAtioN Network (CleanNet) with our new action proposal evaluator, which provides pseudo-supervision by leveraging the temporal contrast in snippet-level action classification predictions. Essentially, the new action proposal evaluator enforces an additional temporal contrast constraint so that high-evaluation-score action proposals are more likely to coincide with true action instances. Moreover, the new action localization module is an integral part of CleanNet which enables end-to-end training. This is in contrast to many existing WS-TAL methods where action localization is merely a post-processing step. Experiments on THUMOS14 and ActivityNet datasets validate the efficacy of CleanNet against existing state-ofthe- art WS-TAL algorithms.
|
[' Gang Hua', ' Nanning Zheng', ' Zhenxing Niu', ' Zhanning Gao', ' Qilin Zhang', ' Le Wang', 'Ziyi Liu']
|
2019-10-01
| null | null | null |
iccv-2019-10
|
['weakly-supervised-action-localization', 'weakly-supervised-temporal-action']
|
['computer-vision', 'computer-vision']
|
[ 4.35426176e-01 -2.44500395e-02 -8.22199166e-01 -3.20265621e-01
-8.33663523e-01 -3.63987029e-01 6.34480000e-01 -1.46422118e-01
-7.17685521e-01 6.38768971e-01 4.63253140e-01 1.16863959e-01
3.98751907e-02 -2.62743860e-01 -7.20022619e-01 -6.52413905e-01
-3.49380463e-01 -5.65672293e-02 8.31103444e-01 1.33355543e-01
-3.76463942e-02 6.91744462e-02 -1.32946718e+00 7.27426648e-01
6.25636995e-01 1.29520273e+00 8.62855762e-02 3.64198834e-01
1.32200971e-01 1.32996655e+00 -3.82018507e-01 -1.96282849e-01
3.60658616e-01 -5.35190642e-01 -9.05634880e-01 3.25245053e-01
7.91281998e-01 -4.23373520e-01 -3.52624089e-01 8.62684071e-01
1.76591426e-01 4.38146442e-01 2.18606248e-01 -1.43327534e+00
-3.81155699e-01 5.08559465e-01 -5.11166096e-01 4.03514236e-01
5.19187748e-01 4.27992404e-01 1.16755474e+00 -1.04823244e+00
8.52238655e-01 1.05228221e+00 7.46628702e-01 5.68263292e-01
-1.21252418e+00 -4.16450441e-01 6.21027350e-01 6.19611382e-01
-1.29845631e+00 -5.09178579e-01 7.67640948e-01 -3.23401779e-01
8.22519720e-01 -4.32408229e-02 7.57549942e-01 1.50739932e+00
-5.57680055e-02 1.41310632e+00 8.93308103e-01 -4.98126596e-02
3.26034278e-01 -3.55816334e-01 -1.40652612e-01 7.68207490e-01
-2.85532326e-01 -2.82583423e-02 -9.47279274e-01 2.20886469e-01
7.20805287e-01 2.14186415e-01 -1.96810558e-01 -6.16787970e-01
-1.68618858e+00 3.28864068e-01 4.66008037e-01 4.59039092e-01
-5.95635056e-01 4.36619729e-01 6.00940704e-01 9.14724767e-02
4.64286536e-01 2.62014747e-01 -5.51781893e-01 -5.22734463e-01
-1.13682771e+00 -5.00800200e-02 1.85773075e-01 7.83842683e-01
5.48130751e-01 1.18118867e-01 -7.77655900e-01 6.85087919e-01
1.70610379e-02 3.15421410e-02 4.90109652e-01 -1.22593451e+00
4.88896638e-01 7.02996671e-01 6.96026906e-02 -5.17645121e-01
-1.73329175e-01 -4.94182080e-01 -4.49475080e-01 5.16282879e-02
3.63087863e-01 1.29903182e-01 -9.20805514e-01 1.72888625e+00
3.98365945e-01 7.16650903e-01 -1.73022002e-01 1.05165768e+00
5.49420893e-01 2.57441074e-01 5.24399519e-01 -1.92234859e-01
1.11887026e+00 -1.50280905e+00 -7.72545576e-01 -1.55664638e-01
9.16218460e-01 -2.69811928e-01 1.32711267e+00 1.82964548e-01
-8.78306508e-01 -7.56388009e-01 -8.38626146e-01 7.34699368e-02
-9.02023464e-02 5.05398214e-01 6.78628564e-01 6.38735592e-02
-8.67060363e-01 6.79015279e-01 -1.15928042e+00 -3.60529989e-01
7.72664368e-01 8.69112760e-02 -7.45351553e-01 -7.66488016e-02
-1.09354806e+00 6.44445896e-01 5.30968606e-01 1.64358824e-01
-1.15067804e+00 -5.04277587e-01 -1.06305194e+00 -1.02311142e-01
9.88867342e-01 -2.99085468e-01 1.45272779e+00 -1.33722246e+00
-1.35219514e+00 5.79590559e-01 -2.18426198e-01 -6.72428429e-01
7.68174231e-01 -2.65655011e-01 -4.42357600e-01 5.15740931e-01
4.41609591e-01 9.74693954e-01 8.25204134e-01 -9.50501919e-01
-9.15383101e-01 3.84470746e-02 3.78183216e-01 2.51613259e-01
-3.03017497e-01 -1.86361298e-01 -7.72539854e-01 -7.98928976e-01
6.02317117e-02 -8.25578213e-01 -1.60423264e-01 2.97745138e-01
-1.53611109e-01 -5.49944580e-01 8.37392509e-01 -4.32400227e-01
1.28756523e+00 -2.31187630e+00 -3.59570086e-02 -2.17001900e-01
8.65390822e-02 2.56993115e-01 -4.29853916e-01 2.10469991e-01
-1.47328719e-01 -6.47716373e-02 -9.81038883e-02 -5.06577075e-01
-2.31237733e-04 1.71482831e-01 2.13554483e-02 5.20921350e-01
5.03544331e-01 1.05627167e+00 -1.39570451e+00 -7.43491888e-01
4.01806265e-01 1.72624528e-01 -5.70608258e-01 3.64176258e-02
-3.70323390e-01 6.58496916e-01 -4.81884539e-01 1.04557812e+00
1.37914523e-01 -3.80590737e-01 1.39453933e-01 -3.84096116e-01
-1.33424103e-01 3.93972725e-01 -9.51344192e-01 2.12453794e+00
-2.35429004e-01 6.51253521e-01 -2.74902076e-01 -1.12342703e+00
2.56687611e-01 4.67432141e-01 1.08088875e+00 -8.96381915e-01
-1.07096374e-01 -7.84858465e-02 -9.49126780e-02 -6.51120961e-01
3.50951135e-01 2.83676200e-02 -4.10560220e-02 2.92398363e-01
4.49470431e-01 5.53367138e-01 5.45421302e-01 3.31249833e-01
1.58027732e+00 1.00227070e+00 2.10458815e-01 1.04657561e-01
5.30408621e-01 1.71817224e-02 1.15665960e+00 7.26100624e-01
-8.35665226e-01 4.84480202e-01 5.77360094e-01 -3.94696921e-01
-6.34452283e-01 -9.28720057e-01 4.52010445e-02 1.43618739e+00
1.61410257e-01 -7.49386668e-01 -5.65155745e-01 -1.42216885e+00
-2.38894790e-01 3.81824911e-01 -8.10596585e-01 -2.34159961e-01
-6.20456874e-01 -1.04117781e-01 4.76925880e-01 8.96950662e-01
7.58705378e-01 -1.27334261e+00 -4.52242434e-01 3.66153121e-01
-4.35836017e-01 -1.55721188e+00 -7.40693331e-01 2.37203345e-01
-7.84809291e-01 -1.26370943e+00 -6.00803614e-01 -6.39892399e-01
7.45048106e-01 3.55920076e-01 9.39075768e-01 -6.61857724e-02
3.10601927e-02 4.26510900e-01 -7.27880836e-01 1.19584598e-01
2.17488967e-03 -1.87102705e-01 6.44078702e-02 2.37862378e-01
6.00476623e-01 -5.24007380e-01 -8.56571674e-01 5.97975492e-01
-7.89917231e-01 1.56403080e-01 6.94995701e-01 7.98140585e-01
7.57040977e-01 -1.87882364e-01 5.50370038e-01 -4.52983201e-01
-4.75945957e-02 -2.63906062e-01 -3.14437985e-01 3.09687972e-01
-1.82572827e-01 -1.09975144e-01 4.39379215e-01 -6.48831487e-01
-9.88511503e-01 4.88313138e-01 1.16642751e-02 -8.30883324e-01
-1.71919629e-01 4.42971200e-01 -8.63482654e-02 1.29049182e-01
5.66322505e-01 3.06162119e-01 -2.11288467e-01 -4.14919376e-01
2.17836112e-01 1.84205949e-01 6.40889823e-01 -3.98470968e-01
5.95756233e-01 5.98506331e-01 -2.28443161e-01 -4.71585333e-01
-1.29520607e+00 -7.39302874e-01 -9.64541554e-01 -6.19335413e-01
1.10049772e+00 -1.12834311e+00 -6.61610484e-01 2.86619902e-01
-8.55485559e-01 -6.59006298e-01 -5.09890795e-01 7.10453570e-01
-7.04259038e-01 3.48528296e-01 -4.71720159e-01 -5.58398664e-01
1.85331792e-01 -1.05815411e+00 1.31215012e+00 -1.57698229e-01
-2.84526229e-01 -8.63451540e-01 -1.26588434e-01 5.55657864e-01
-4.98473532e-02 3.55380356e-01 2.54430920e-01 -5.63961327e-01
-5.98742664e-01 -3.49207848e-01 -1.10157430e-01 5.30889034e-01
2.55540401e-01 -1.64194748e-01 -8.40408444e-01 -1.72677234e-01
-3.25598240e-01 -6.78767443e-01 1.16434598e+00 3.20861101e-01
1.40680289e+00 -1.11090578e-01 -2.91084200e-01 3.92952293e-01
9.86395895e-01 4.40650396e-02 4.03483063e-01 2.57492542e-01
8.23705256e-01 3.61813724e-01 1.20217407e+00 3.32507193e-01
2.53884971e-01 9.77968514e-01 4.54457998e-01 -1.88759536e-01
-3.16626847e-01 -5.04403055e-01 8.35234702e-01 3.54417384e-01
-2.76519269e-01 -2.39408389e-01 -6.23433650e-01 5.08356988e-01
-2.27033257e+00 -1.34777343e+00 1.00218236e-01 2.05259299e+00
8.09359193e-01 3.15839618e-01 3.47119302e-01 2.30722222e-02
5.19095242e-01 5.24543941e-01 -6.47714853e-01 4.42240953e-01
4.13102321e-02 5.43541799e-04 4.56478417e-01 1.29777834e-01
-1.61502695e+00 1.12677109e+00 5.53984642e+00 9.52042401e-01
-9.05021250e-01 4.74560440e-01 3.21724534e-01 -3.91587079e-01
2.52921999e-01 7.57206157e-02 -6.04442477e-01 5.59190154e-01
6.03130817e-01 3.19111139e-01 -7.99177289e-02 9.89178538e-01
6.25954449e-01 -3.58016491e-01 -1.45371068e+00 8.28772545e-01
1.07723251e-01 -1.24245203e+00 -1.72046602e-01 -1.28053620e-01
6.40122235e-01 1.10014088e-01 -2.27822930e-01 6.80552602e-01
2.13668868e-01 -4.86492932e-01 9.82524753e-01 3.02823663e-01
7.82609701e-01 -3.21717471e-01 6.74089372e-01 1.99303046e-01
-1.54452515e+00 -2.63660848e-01 2.68701967e-02 4.82217595e-02
4.80469644e-01 2.60949314e-01 -6.49639428e-01 4.44962651e-01
7.03310549e-01 1.50584567e+00 -6.09794617e-01 1.09307754e+00
-6.28610432e-01 7.32956946e-01 -5.59214652e-02 3.55612606e-01
7.17761815e-01 1.20179936e-01 4.49860156e-01 1.18061030e+00
2.17396785e-02 7.32959136e-02 6.93049788e-01 4.83219862e-01
-1.27428398e-01 -2.15377256e-01 -4.49061364e-01 -1.67527705e-01
1.57137647e-01 1.15507758e+00 -8.03515434e-01 -4.57095653e-01
-5.70285201e-01 1.24423897e+00 2.78482527e-01 4.94058698e-01
-1.16205132e+00 1.08632021e-01 7.40275860e-01 1.20441251e-01
4.08335268e-01 -2.08879635e-01 2.45115101e-01 -1.44137239e+00
3.10814977e-01 -7.11946011e-01 5.04730642e-01 -7.12212384e-01
-1.02355862e+00 2.63343304e-01 -2.98779812e-02 -1.99407601e+00
-1.33725479e-01 -3.78065079e-01 -4.83463079e-01 4.88599837e-02
-1.43670702e+00 -1.24981129e+00 -3.40678185e-01 6.67695701e-01
1.02473390e+00 8.28564689e-02 4.31350976e-01 5.28946817e-01
-7.38334119e-01 5.97876072e-01 -3.74077737e-01 3.43157709e-01
9.01899695e-01 -1.15424013e+00 9.58011486e-03 1.06218266e+00
3.73537600e-01 3.16151828e-01 3.65732819e-01 -7.55134344e-01
-1.02745306e+00 -1.36231863e+00 6.83816314e-01 -5.82281709e-01
8.31028998e-01 -3.64144087e-01 -8.57565165e-01 7.90104032e-01
6.51179999e-02 6.35130942e-01 5.69205046e-01 1.37401000e-01
-3.36886048e-01 -1.56638533e-01 -6.76758051e-01 5.71572304e-01
1.67878687e+00 -6.57813013e-01 -5.45810223e-01 6.12727344e-01
7.23043144e-01 -2.24628866e-01 -7.53785014e-01 5.42932570e-01
5.30986965e-01 -8.61976445e-01 9.23018396e-01 -6.83445215e-01
5.69449425e-01 -5.73980033e-01 4.26430963e-02 -9.48641479e-01
-3.40970188e-01 -5.09428918e-01 -3.57012540e-01 1.07879555e+00
2.69955844e-01 -5.16405180e-02 8.34189057e-01 3.42423946e-01
-4.90001887e-01 -8.32885861e-01 -1.04089475e+00 -1.19788194e+00
-6.94269717e-01 -7.21585393e-01 9.97767672e-02 1.02311683e+00
1.48404241e-01 8.14908966e-02 -6.08167529e-01 -1.03908189e-01
4.65254456e-01 -2.24937931e-01 6.32732570e-01 -8.84710133e-01
-2.99337417e-01 -3.93377930e-01 -5.89731514e-01 -1.30042410e+00
4.32444990e-01 -7.74590850e-01 4.17803168e-01 -1.43570256e+00
1.23505488e-01 -2.30160922e-01 -8.64369154e-01 1.12940609e+00
-1.23594873e-01 6.09062731e-01 1.03985220e-01 1.99227273e-01
-1.51997614e+00 7.84391701e-01 1.31330538e+00 -1.63134009e-01
-2.19933540e-01 -1.21091165e-01 -1.33774906e-01 8.82789075e-01
4.30963755e-01 -5.68561018e-01 -5.53373754e-01 -2.40449652e-01
-1.19130030e-01 -6.03528507e-02 6.09816611e-01 -1.14142370e+00
1.99656621e-01 -4.04961437e-01 2.94180185e-01 -6.57086909e-01
4.39618379e-01 -8.56362820e-01 -1.42740935e-01 3.23975772e-01
-6.08947635e-01 -2.62953311e-01 -1.02029428e-01 8.17080498e-01
-5.65559983e-01 8.64515230e-02 6.15826726e-01 -5.07989079e-02
-1.38480341e+00 6.41057014e-01 -3.65450501e-01 -7.69813657e-02
1.33452880e+00 -5.43235898e-01 -2.22385213e-01 -8.75886604e-02
-9.24075603e-01 4.18573052e-01 3.67336601e-01 6.14659071e-01
5.18148363e-01 -1.55392706e+00 -3.43035370e-01 -3.20699662e-02
5.87198615e-01 -6.21755607e-02 3.65487963e-01 1.40876508e+00
5.14103100e-02 2.85725176e-01 -1.46182820e-01 -7.31202245e-01
-1.23863685e+00 3.70181978e-01 2.18484864e-01 -2.40993813e-01
-7.87875295e-01 8.69940519e-01 2.33527079e-01 -3.81886028e-02
5.09075284e-01 -3.90801072e-01 -2.11945757e-01 7.78676122e-02
4.60879058e-01 1.78878129e-01 -1.60570294e-01 -6.57856286e-01
-5.87725282e-01 1.78379744e-01 -5.48741557e-02 -1.39190346e-01
1.25310528e+00 1.10859513e-01 2.14870423e-01 4.79307652e-01
1.04482329e+00 -3.23465705e-01 -1.95377064e+00 -4.46711004e-01
2.16754019e-01 -5.71397305e-01 1.18131660e-01 -8.23113799e-01
-1.07471204e+00 7.71017730e-01 4.41500932e-01 -3.02959055e-01
1.08529866e+00 4.30430807e-02 6.78319454e-01 4.15390402e-01
4.44817722e-01 -1.38801885e+00 7.60363460e-01 3.46793592e-01
7.06226230e-01 -1.49992704e+00 -5.39815091e-02 -2.07403734e-01
-9.13450360e-01 6.63365781e-01 1.05832064e+00 6.19689152e-02
3.45576644e-01 -1.40552908e-01 -1.63318530e-01 -9.03178379e-02
-9.50464189e-01 -5.02072453e-01 4.81557906e-01 3.91561896e-01
3.21694136e-01 -1.86663955e-01 -4.55687940e-01 4.14510101e-01
6.24557257e-01 3.52312624e-01 2.07349777e-01 1.27245677e+00
-3.36332858e-01 -1.02294993e+00 1.83365390e-01 3.70253354e-01
-4.68779415e-01 7.19878003e-02 -2.69673526e-01 6.64508641e-01
3.56222183e-01 7.74431467e-01 3.00640259e-02 -6.11075580e-01
2.64887273e-01 1.66862190e-01 2.41185620e-01 -7.03794956e-01
-4.56814915e-01 2.03485101e-01 2.77860552e-01 -1.21354127e+00
-9.69408572e-01 -8.07133794e-01 -1.38145661e+00 2.70918757e-01
-2.38039374e-01 7.86502510e-02 3.08435977e-01 1.16131330e+00
4.51798856e-01 6.90359056e-01 4.85443771e-01 -8.86702001e-01
-3.00083935e-01 -1.01371360e+00 -4.12908137e-01 6.38548970e-01
3.77214134e-01 -9.73437667e-01 -2.04019055e-01 4.97142732e-01]
|
[8.501032829284668, 0.6233439445495605]
|
53f9fce3-0fd7-4ce6-936c-17ead02b97b7
|
unsupervised-text-embedding-space-generation
|
2306.17181
| null |
https://arxiv.org/abs/2306.17181v2
|
https://arxiv.org/pdf/2306.17181v2.pdf
|
Unsupervised Text Embedding Space Generation Using Generative Adversarial Networks for Text Synthesis
|
Generative Adversarial Networks (GAN) is a model for data synthesis, which creates plausible data through the competition of generator and discriminator. Although GAN application to image synthesis is extensively studied, it has inherent limitations to natural language generation. Because natural language is composed of discrete tokens, a generator has difficulty updating its gradient through backpropagation; therefore, most text-GAN studies generate sentences starting with a random token based on a reward system. Thus, the generators of previous studies are pre-trained in an autoregressive way before adversarial training, causing data memorization that synthesized sentences reproduce the training data. In this paper, we synthesize sentences using a framework similar to the original GAN. More specifically, we propose Text Embedding Space Generative Adversarial Networks (TESGAN) which generate continuous text embedding spaces instead of discrete tokens to solve the gradient backpropagation problem. Furthermore, TESGAN conducts unsupervised learning which does not directly refer to the text of the training data to overcome the data memorization issue. By adopting this novel method, TESGAN can synthesize new sentences, showing the potential of unsupervised learning for text synthesis. We expect to see extended research combining Large Language Models with a new perspective of viewing text as an continuous space.
|
['Tae-Bin Ha', 'Jun-Min Lee']
|
2023-06-19
| null | null | null | null |
['image-generation', 'memorization', 'text-generation']
|
['computer-vision', 'natural-language-processing', 'natural-language-processing']
|
[ 6.65262043e-01 5.02486646e-01 7.68773854e-02 -1.06075957e-01
-4.97992843e-01 -4.21403915e-01 1.02928245e+00 -4.68534827e-01
-1.50175303e-01 9.83710170e-01 4.23862338e-01 -1.69388652e-01
6.87228858e-01 -1.32731724e+00 -9.65537906e-01 -7.70547211e-01
6.59290016e-01 3.31403583e-01 -2.45343104e-01 -3.49348158e-01
1.50978118e-01 1.62696183e-01 -1.13751721e+00 2.97933817e-01
1.09920287e+00 4.87529993e-01 3.24554384e-01 7.05915630e-01
-4.90902781e-01 1.11527550e+00 -1.28341901e+00 -5.61669648e-01
2.00534329e-01 -1.19654441e+00 -4.35618848e-01 4.58713174e-02
7.63545707e-02 -4.84922558e-01 -4.34216082e-01 1.01190448e+00
5.11389554e-01 -9.76108313e-02 7.98795342e-01 -1.35770333e+00
-1.53301334e+00 8.46536756e-01 3.13847954e-03 -4.09102470e-01
2.96400189e-01 4.14578319e-01 7.54218876e-01 -8.45808387e-01
5.67282617e-01 1.14461434e+00 3.21627200e-01 1.12376416e+00
-1.14200640e+00 -6.13746881e-01 1.66010708e-02 -2.65272200e-01
-1.00318110e+00 -2.92175591e-01 1.13721609e+00 -4.18759197e-01
7.99795628e-01 3.60958546e-01 8.92819941e-01 1.65269661e+00
6.23232663e-01 7.42621839e-01 9.84622657e-01 -5.21791399e-01
3.87539208e-01 2.75089145e-01 -7.85771132e-01 4.47253466e-01
-2.11024471e-02 2.33765244e-01 -4.24036205e-01 2.41107509e-01
9.31920648e-01 2.54327785e-02 -6.31049797e-02 -4.68698964e-02
-1.21660566e+00 1.14952886e+00 5.54977953e-01 2.93682545e-01
-3.56883258e-01 4.30777937e-01 2.74671406e-01 4.92964774e-01
4.18027401e-01 8.62625957e-01 1.65032148e-01 3.21840052e-03
-8.21229458e-01 2.39213437e-01 4.88352746e-01 1.05883729e+00
6.34961605e-01 1.00018537e+00 -5.86705804e-01 6.00253761e-01
2.64248610e-01 8.64833295e-01 1.11318743e+00 -4.93366808e-01
5.34582913e-01 4.97864038e-01 -3.11644971e-01 -9.40325558e-01
2.16469228e-01 -2.26667494e-01 -1.34271276e+00 3.57982159e-01
6.31045103e-02 -4.77707684e-01 -1.19988716e+00 1.81296599e+00
-2.22534120e-01 7.79439360e-02 4.07110333e-01 4.72450793e-01
9.42207456e-01 1.01640558e+00 -2.10640244e-02 -6.60449341e-02
7.83315003e-01 -1.06784379e+00 -9.38413322e-01 -3.39921087e-01
3.40982676e-01 -6.61793292e-01 1.27260530e+00 1.65601656e-01
-1.19600499e+00 -7.66704619e-01 -1.08120036e+00 -8.06165412e-02
-6.11608863e-01 -6.36368245e-02 4.44356143e-01 7.26290107e-01
-1.14489996e+00 2.47794509e-01 -5.05048871e-01 4.97893021e-02
4.03153569e-01 9.39774439e-02 -2.85651237e-02 3.58722508e-01
-1.50483382e+00 8.11721921e-01 3.83588195e-01 2.48232231e-01
-9.89248931e-01 -4.19343114e-01 -1.05450785e+00 8.44744354e-05
-2.68979389e-02 -1.09454620e+00 1.00131941e+00 -1.46923077e+00
-2.13780022e+00 5.25230169e-01 -1.13790035e-02 -6.85707688e-01
8.26825142e-01 1.74580306e-01 -3.53118986e-01 -2.34629229e-01
2.02559996e-02 9.39257503e-01 1.39714861e+00 -1.30305696e+00
7.93851912e-02 2.12327585e-01 -3.10024824e-02 1.22315519e-01
-3.56270224e-01 -4.72752929e-01 1.27609804e-01 -1.30471921e+00
-2.41937891e-01 -7.14023650e-01 -3.75660747e-01 -4.09695804e-01
-6.36957228e-01 -2.76954062e-02 7.05352187e-01 -4.46610063e-01
1.10045362e+00 -2.13583541e+00 1.71783790e-01 1.83888692e-02
3.02938581e-01 2.99943149e-01 -4.12912399e-01 8.48607302e-01
-1.08386807e-01 6.32745206e-01 -4.44617301e-01 -3.07628602e-01
3.73889543e-02 2.21933126e-01 -9.61787879e-01 -1.69453859e-01
5.90159535e-01 1.46891689e+00 -9.80574310e-01 -3.07660401e-01
2.72957146e-01 3.37148458e-01 -6.66153550e-01 4.20612395e-01
-5.28756678e-01 6.57499611e-01 -4.86540824e-01 1.97168097e-01
5.19805431e-01 3.52057666e-02 -8.25792849e-02 1.62748083e-01
4.67453972e-02 1.71210438e-01 -6.13442779e-01 1.61208677e+00
-6.44300222e-01 8.41608584e-01 -7.08803594e-01 -1.04756558e+00
1.22660971e+00 4.12979424e-01 1.68556035e-01 -6.79070532e-01
-1.22307008e-02 8.38215724e-02 8.62991624e-03 -3.32377493e-01
6.81745946e-01 -4.04077619e-01 -1.75652951e-01 5.00117660e-01
-3.05444859e-02 -9.33909297e-01 2.51272470e-02 2.77907550e-01
9.06177044e-01 7.01203942e-02 3.94272320e-02 3.88171703e-01
2.90830135e-01 -2.32637301e-01 2.21101955e-01 9.62703586e-01
3.86978328e-01 9.29956019e-01 6.36642754e-01 -2.00094864e-01
-1.47111058e+00 -1.30273187e+00 2.60938287e-01 5.80021381e-01
-1.85050219e-01 -4.78503257e-01 -9.12136078e-01 -6.64157689e-01
-3.01228255e-01 1.01715517e+00 -8.66538107e-01 -5.33502400e-01
-5.02978981e-01 -5.01580596e-01 8.12062740e-01 5.02845824e-01
7.12515593e-01 -1.75441492e+00 -2.01162532e-01 3.08063596e-01
-1.33795097e-01 -6.99564219e-01 -6.47102118e-01 -8.52844343e-02
-7.36156404e-01 -4.78770375e-01 -1.01448917e+00 -9.30360019e-01
9.95636761e-01 -2.11367190e-01 1.04945230e+00 -1.70885231e-02
2.76976917e-02 1.05245270e-01 -5.45196891e-01 -6.45585001e-01
-1.17221141e+00 -2.10941900e-02 -2.56791592e-01 -7.65431523e-02
-1.04509950e-01 -5.32844305e-01 -4.22113627e-01 -1.20622218e-01
-1.38442945e+00 5.18435299e-01 7.58502543e-01 1.20457125e+00
4.52521414e-01 6.25346825e-02 1.09832275e+00 -1.10579908e+00
1.15731382e+00 -4.03688431e-01 -3.22780371e-01 2.19714507e-01
-6.08687103e-01 1.59140661e-01 1.23403525e+00 -6.83019280e-01
-1.11061049e+00 -2.40458041e-01 -3.16306144e-01 -3.23784679e-01
3.20572243e-03 4.74535465e-01 -2.08971903e-01 3.11331064e-01
6.99143589e-01 8.37840855e-01 1.08756378e-01 5.13133444e-02
7.13948905e-01 5.72222233e-01 3.45203698e-01 -3.66297424e-01
1.13769329e+00 8.50159302e-02 -2.33658195e-01 -5.96435666e-01
-2.41742432e-01 6.57726705e-01 -3.36871803e-01 -1.13868840e-01
1.05199194e+00 -8.16468596e-01 -2.72780359e-01 7.90430665e-01
-1.27059913e+00 -5.64822197e-01 -9.34554875e-01 2.86288977e-01
-7.22597957e-01 1.40556931e-01 -5.25247991e-01 -5.08036673e-01
-3.99360478e-01 -1.04880166e+00 8.58854473e-01 2.34889835e-01
-1.62463889e-01 -1.13293946e+00 1.25422850e-01 -9.41070821e-03
6.82109594e-01 4.60785657e-01 8.49346578e-01 -4.33226824e-01
-6.62274539e-01 -1.73925757e-01 1.98754594e-01 7.59386241e-01
5.55894792e-01 1.10686637e-01 -7.74629951e-01 -2.26467416e-01
1.91338241e-01 -3.22260678e-01 7.64415205e-01 1.86643377e-01
1.28168094e+00 -6.86132193e-01 1.27741680e-01 5.38237274e-01
1.10576296e+00 4.58786100e-01 1.07175207e+00 1.63529977e-01
7.86213875e-01 1.92651972e-01 2.79601552e-02 2.93028682e-01
1.23639114e-01 2.66344041e-01 3.04124445e-01 -4.22898978e-01
-3.54712337e-01 -8.92100573e-01 7.04632223e-01 9.46096301e-01
1.31747648e-01 -6.12127900e-01 -5.17072558e-01 3.11977834e-01
-1.49433875e+00 -1.21440589e+00 1.46563604e-01 1.92763162e+00
1.13772666e+00 3.25876594e-01 -2.00149074e-01 8.18513930e-02
5.99472940e-01 3.66909474e-01 -7.13611841e-01 -6.49586320e-01
-3.29809248e-01 5.73467612e-01 2.31300473e-01 3.22705328e-01
-6.06315017e-01 1.09407854e+00 6.43170834e+00 8.02111804e-01
-1.49370432e+00 -1.62340358e-01 5.92728019e-01 8.64293873e-02
-8.99749339e-01 -7.19269738e-02 -3.62232327e-01 8.42022002e-01
7.42429137e-01 -5.19948661e-01 5.64229608e-01 6.91394508e-01
2.37580657e-01 2.70468980e-01 -1.07484841e+00 7.81393647e-01
2.15423971e-01 -1.55412531e+00 7.77604163e-01 -8.36434737e-02
1.27250969e+00 -4.87482309e-01 4.89329904e-01 4.16807771e-01
4.12045449e-01 -1.31346631e+00 7.68196106e-01 7.31254578e-01
1.06691599e+00 -5.60042083e-01 5.70847154e-01 3.89144093e-01
-7.26059198e-01 1.23113275e-01 -4.03087646e-01 -4.06327359e-02
1.56199306e-01 5.28764665e-01 -1.12434816e+00 4.77304816e-01
2.74343640e-02 7.03516364e-01 -7.11789012e-01 4.66644287e-01
-6.72783077e-01 6.66797817e-01 1.61935464e-01 -2.37379849e-01
1.65095970e-01 -2.37733752e-01 4.57428336e-01 9.31820035e-01
5.21719098e-01 -2.93736070e-01 -7.54983574e-02 1.35088491e+00
-4.43662912e-01 5.39487861e-02 -1.08029580e+00 -5.79634011e-01
3.45736831e-01 7.65258193e-01 -4.68772143e-01 -4.39825922e-01
-1.26222789e-01 1.25826430e+00 3.09837032e-02 6.10591173e-01
-9.10606265e-01 -6.42679095e-01 9.74400267e-02 -1.90383121e-02
2.06029993e-02 -1.67441472e-01 -6.69088125e-01 -1.21456063e+00
5.47091998e-02 -1.03518653e+00 -3.88008595e-01 -1.02853703e+00
-1.38973546e+00 8.20164144e-01 -3.11635792e-01 -1.36511731e+00
-7.60216892e-01 -1.56033218e-01 -1.01664901e+00 1.02973843e+00
-1.12718439e+00 -1.23051906e+00 -1.88507318e-01 4.59866136e-01
8.25566053e-01 -5.57899535e-01 1.00313175e+00 -1.80027828e-01
-3.63985002e-01 8.31911922e-01 2.12222204e-01 3.91920716e-01
6.23017132e-01 -1.30373454e+00 9.04221356e-01 9.38819110e-01
2.20397949e-01 6.10194445e-01 5.92035413e-01 -8.79759908e-01
-1.16443717e+00 -1.25046396e+00 6.92180336e-01 -3.64272892e-01
4.74103838e-01 -7.20288455e-01 -7.94106305e-01 8.19734514e-01
6.71852231e-01 -4.50059116e-01 5.21865964e-01 -5.30354679e-01
-2.71212161e-02 3.60678919e-02 -1.13215017e+00 9.15000439e-01
7.44122624e-01 -5.12185872e-01 -5.86609244e-01 2.63029993e-01
1.04213524e+00 -4.53759789e-01 -5.89331388e-01 1.29960999e-01
2.24765152e-01 -7.06555545e-01 6.76931560e-01 -5.12282491e-01
1.05713761e+00 -1.19279765e-01 2.04890385e-01 -1.78060949e+00
-4.34930697e-02 -6.65884197e-01 4.40054722e-02 1.43533468e+00
4.85471308e-01 -9.78588045e-01 7.22850978e-01 2.14563042e-01
-2.01185420e-01 -5.68067670e-01 -6.31874800e-01 -7.41798759e-01
4.97511625e-01 -1.33544371e-01 1.00062346e+00 9.27251577e-01
-2.62606531e-01 4.38811541e-01 -7.03817189e-01 -4.29090947e-01
2.44663835e-01 -2.02108040e-01 1.01297867e+00 -6.12109184e-01
-2.47668356e-01 -5.40334463e-01 -1.52354091e-01 -1.14233184e+00
2.84938753e-01 -1.09713197e+00 2.96578079e-01 -1.61298430e+00
-2.04144880e-01 -4.10929948e-01 1.40707701e-01 2.53498048e-01
-2.18723267e-01 2.46582136e-01 4.14179444e-01 1.58485875e-01
1.10579073e-01 1.05381179e+00 1.98486483e+00 -4.69901592e-01
-2.36292571e-01 -8.13329685e-03 -7.59860575e-01 3.28049451e-01
9.62238252e-01 -3.74156892e-01 -7.94245601e-01 -5.83690464e-01
3.07934254e-01 3.76083027e-03 2.69260913e-01 -8.12176108e-01
1.08604021e-01 -3.43097210e-01 5.99333942e-01 -2.83548445e-01
2.00054362e-01 -4.67999101e-01 3.58039975e-01 4.42192525e-01
-6.11974537e-01 5.12215346e-02 -3.21951583e-02 4.60938245e-01
-3.75728548e-01 -3.85140061e-01 5.71845233e-01 -3.65329117e-01
-2.20219001e-01 1.94301605e-01 -7.78762937e-01 1.22698374e-01
9.26710308e-01 -2.52391249e-01 -2.12257057e-01 -6.59232974e-01
-5.85050046e-01 -1.35641009e-01 5.31993866e-01 5.56479096e-01
9.30840373e-01 -1.57269251e+00 -8.55813920e-01 5.85355818e-01
-2.05865294e-01 1.96813434e-01 1.75248563e-01 3.57713103e-02
-4.97729689e-01 3.21194738e-01 -3.15740615e-01 -2.54016697e-01
-4.70574439e-01 6.08768284e-01 2.83225328e-01 -3.30780089e-01
-5.37567437e-01 6.03129506e-01 3.62969935e-01 -4.19815451e-01
-1.56336457e-01 -3.57931882e-01 -4.47892025e-02 -1.04949512e-01
3.42164814e-01 -3.19712348e-02 -2.54059553e-01 -1.39211908e-01
2.96560317e-01 2.89461970e-01 -2.11644340e-02 -2.95312792e-01
1.11267745e+00 1.67055950e-01 -1.31084725e-01 4.62720096e-01
1.00170493e+00 3.14411879e-01 -1.11549199e+00 3.48385014e-02
-5.71798086e-01 -2.80548155e-01 -4.20222402e-01 -7.05908000e-01
-1.02594793e+00 8.83700013e-01 2.58894265e-01 5.48901439e-01
1.16935432e+00 -2.72114486e-01 8.12853217e-01 3.38293165e-01
-6.95603639e-02 -9.88145709e-01 6.67200625e-01 4.67932552e-01
1.31163871e+00 -1.10885525e+00 -4.01440978e-01 -4.79208753e-02
-7.68239081e-01 1.14025235e+00 7.85705030e-01 -3.94513965e-01
2.95076728e-01 2.62360185e-01 4.85300645e-02 1.81880265e-01
-6.71395719e-01 8.55425820e-02 9.52034667e-02 7.98552990e-01
4.10571277e-01 8.41794387e-02 -3.99702311e-01 3.06323320e-01
-7.99210012e-01 3.67556959e-02 8.78059626e-01 7.29653955e-01
-1.03920810e-01 -1.49046075e+00 -2.20937401e-01 4.57637995e-01
-1.94328085e-01 -4.26050723e-01 -4.74372625e-01 5.71771383e-01
9.77338776e-02 8.74534249e-01 1.61571920e-01 -4.79098737e-01
1.34295210e-01 9.29481462e-02 3.93381983e-01 -8.94659936e-01
-5.59683144e-01 -2.31050804e-01 -4.08014327e-01 8.04511085e-02
-1.64769292e-01 -3.00098658e-01 -1.27243066e+00 -2.47893766e-01
-1.82596773e-01 1.98246568e-01 6.84758663e-01 7.51316249e-01
1.85987666e-01 1.01021791e+00 1.08563268e+00 -5.45586348e-01
-5.94108224e-01 -1.12561345e+00 -3.27063590e-01 4.51647639e-01
2.74014801e-01 -1.48785636e-01 -3.22169721e-01 3.86718690e-01]
|
[11.895118713378906, 9.354439735412598]
|
1bf8d3e2-728e-440e-a012-ad3b949e0287
|
figure-descriptive-text-extraction-using
|
2208.06040
| null |
https://arxiv.org/abs/2208.06040v1
|
https://arxiv.org/pdf/2208.06040v1.pdf
|
Figure Descriptive Text Extraction using Ontological Representation
|
Experimental research publications provide figure form resources including graphs, charts, and any type of images to effectively support and convey methods and results. To describe figures, authors add captions, which are often incomplete, and more descriptions reside in body text. This work presents a method to extract figure descriptive text from the body of scientific articles. We adopted ontological semantics to aid concept recognition of figure-related information, which generates human- and machine-readable knowledge representations from sentences. Our results show that conceptual models bring an improvement in figure descriptive sentence classification over word-based approaches.
|
['Line Pouchard', 'Julia Rayz', 'Gilchan Park']
|
2022-08-11
| null | null | null | null |
['sentence-classification']
|
['natural-language-processing']
|
[ 4.32422347e-02 4.00955319e-01 -1.40588343e-01 -3.21205348e-01
-5.08335173e-01 -7.28497565e-01 6.60171151e-01 9.08909917e-01
4.99822944e-02 8.55452418e-01 5.23843944e-01 -5.69166005e-01
5.97871898e-04 -1.01897168e+00 -6.12853944e-01 7.95482695e-02
4.27886061e-02 4.31106575e-02 1.27043933e-01 -1.57476082e-01
1.11536837e+00 8.02341402e-01 -1.57925582e+00 7.32275784e-01
8.03599477e-01 8.05280626e-01 4.66113463e-02 6.25054359e-01
-1.15882421e+00 1.13976300e+00 -1.24621952e+00 -5.45014143e-01
-6.42612576e-01 -3.07566047e-01 -6.73431337e-01 2.04176486e-01
3.31930906e-01 -4.64286171e-02 -3.15566212e-01 1.02491355e+00
2.78842151e-01 -2.62798935e-01 1.13393676e+00 -1.33074200e+00
-1.41267753e+00 6.21933401e-01 -2.63195843e-01 2.93922961e-01
1.18676662e+00 -4.40958202e-01 4.21805680e-01 -8.33375990e-01
1.06476521e+00 1.60309875e+00 3.38608593e-01 3.77814263e-01
-6.58163965e-01 -4.02037323e-01 1.55778840e-01 1.91821486e-01
-1.36275828e+00 -3.41257125e-01 1.02334630e+00 -5.03655493e-01
1.00760710e+00 4.43010718e-01 9.78700876e-01 1.00506222e+00
4.38700676e-01 5.42896688e-01 9.41485107e-01 -9.01458263e-01
3.67541075e-01 4.58143800e-01 6.40108049e-01 1.08451653e+00
8.07385266e-01 -8.19568157e-01 -7.51228929e-01 1.16892964e-01
8.76753092e-01 6.02252148e-02 -8.06545094e-02 1.24962054e-01
-8.37613702e-01 4.17922378e-01 5.08971274e-01 4.71786976e-01
-1.49868160e-01 5.28250821e-03 5.44984519e-01 -2.21757039e-01
4.56596285e-01 5.65148413e-01 2.20432833e-01 -6.45207465e-02
-6.19660735e-01 6.39822930e-02 6.79499328e-01 1.71060336e+00
5.46827972e-01 -6.82246238e-02 -2.78762609e-01 4.38530236e-01
3.73472780e-01 8.41113627e-01 4.12524074e-01 -5.55188060e-01
5.29674947e-01 1.32116079e+00 6.85554445e-02 -1.72862864e+00
-4.42492217e-01 -2.22944677e-01 -4.74945992e-01 -3.37176323e-01
-1.54167965e-01 5.13844252e-01 -1.12875879e+00 6.31321371e-01
-1.08130284e-01 -4.96872067e-01 2.08987221e-01 6.98082447e-01
1.87125540e+00 7.44121373e-01 5.55781364e-01 -1.61758140e-01
1.84027731e+00 -5.95581830e-01 -1.50916266e+00 1.56867579e-01
8.03363740e-01 -7.30794311e-01 1.02352834e+00 9.68948528e-02
-9.07520592e-01 -3.28191549e-01 -1.35463071e+00 -4.74201620e-01
-1.51548731e+00 2.77809739e-01 5.83454549e-01 6.56224191e-01
-7.09519565e-01 4.46339935e-01 -3.03495139e-01 -3.94283712e-01
8.94945383e-01 -4.83654857e-01 -4.16520298e-01 -5.73346503e-02
-1.04561770e+00 1.07705712e+00 4.17966872e-01 2.94940900e-02
-9.13259163e-02 -7.36757755e-01 -1.13090420e+00 -4.13639061e-02
3.60737950e-01 -7.77573168e-01 7.15084970e-01 -4.52241033e-01
-1.14758933e+00 1.16807783e+00 -1.44890830e-01 -1.65816262e-01
7.33751580e-02 1.78672671e-02 -7.20970213e-01 8.77052784e-01
1.24977671e-01 4.20109272e-01 5.68463206e-01 -1.54934657e+00
1.83383767e-02 -6.19484961e-01 2.71213293e-01 -7.80365290e-03
-3.07929784e-01 3.17390829e-01 -5.00092685e-01 -7.83369422e-01
1.21689782e-01 -2.00501978e-01 2.83729762e-01 2.41947994e-01
-4.56775367e-01 -4.97160614e-01 8.12968314e-01 -8.63700986e-01
1.59126449e+00 -1.96311963e+00 -2.80902296e-01 1.80067539e-01
4.55205083e-01 -3.25244479e-02 3.19345623e-01 7.48952270e-01
2.13243276e-01 1.06289697e+00 -4.76778746e-02 1.69002965e-01
1.02700874e-01 1.97903126e-01 -2.42277056e-01 -2.67185755e-02
2.45830014e-01 1.09197974e+00 -7.95794189e-01 -1.25082648e+00
2.89889634e-01 5.81378162e-01 2.19615340e-01 -8.11360702e-02
-2.52885789e-01 -3.79438192e-01 -9.82742965e-01 9.69880283e-01
5.55669129e-01 -4.45238650e-01 1.25230685e-01 -3.75695735e-01
-2.34400049e-01 1.74566090e-01 -9.25063312e-01 1.58121979e+00
-4.25591111e-01 9.77111578e-01 -5.89925408e-01 -7.40427554e-01
1.18856144e+00 2.50767171e-01 -2.39054486e-01 -8.08758438e-01
2.33747691e-01 1.66027620e-02 -7.45198786e-01 -9.45809424e-01
4.20984477e-01 1.08111054e-02 -6.92193136e-02 -2.60736383e-02
-2.63094932e-01 -4.77272540e-01 5.17869353e-01 6.80749834e-01
6.35169148e-01 4.47411567e-01 4.97682959e-01 -2.23995045e-01
6.07954621e-01 3.89667004e-01 -2.42583379e-01 6.30755246e-01
2.11616382e-01 4.59560126e-01 8.32971752e-01 -6.14441574e-01
-8.92753363e-01 -1.13362825e+00 -1.82574138e-01 4.97484565e-01
1.19056240e-01 -9.69228208e-01 -7.99724340e-01 -3.85580868e-01
-1.76918387e-01 9.84345436e-01 -7.37737894e-01 2.03026444e-01
-3.67987514e-01 -1.99319124e-01 1.94551185e-01 6.38781846e-01
1.60008296e-01 -1.10403466e+00 -7.74268627e-01 -7.86399543e-02
8.34211782e-02 -9.41219270e-01 8.43806565e-02 -2.15502128e-01
-9.06931162e-01 -1.34953976e+00 -7.97923863e-01 -8.67620766e-01
1.22806931e+00 9.79122072e-02 1.25470483e+00 4.10333395e-01
-7.10622907e-01 4.46718782e-01 -6.77582860e-01 -1.02676427e+00
-2.61208862e-01 -3.97627592e-01 -5.58484614e-01 -8.19337845e-01
2.60219067e-01 -3.83505486e-02 -1.63857490e-01 -3.14835578e-01
-1.12906349e+00 3.59301865e-01 3.18967313e-01 1.81397930e-01
4.25637513e-01 -1.50950596e-01 4.69520956e-01 -1.02552724e+00
9.55731869e-01 -1.54754147e-01 -2.75534689e-01 7.07007766e-01
-6.47323072e-01 3.07459623e-01 5.79623938e-01 8.08358788e-02
-8.34565639e-01 -6.88424170e-01 2.67044604e-01 -1.90996647e-01
-3.08019549e-01 7.45386958e-01 -2.24400774e-01 1.84446514e-01
5.44334769e-01 2.04844192e-01 1.20983519e-01 -4.03368562e-01
5.30554593e-01 9.10344243e-01 2.53217161e-01 -6.27486825e-01
3.26240480e-01 2.01709762e-01 2.58505285e-01 -1.07316673e+00
-7.44952679e-01 -3.97529423e-01 -6.51108503e-01 -7.51010239e-01
8.66546512e-01 -6.37118280e-01 -7.58694947e-01 -2.82692790e-01
-1.49046659e+00 5.83341837e-01 -7.44910613e-02 1.93396345e-01
-3.37422580e-01 3.19186330e-01 -3.70087385e-01 -1.03384554e+00
-4.16611791e-01 -5.43468177e-01 1.19549847e+00 5.49611568e-01
-1.14090167e-01 -9.89243090e-01 -5.81483126e-01 2.82732964e-01
1.02223828e-01 6.45717740e-01 1.27097535e+00 -5.73806345e-01
-3.43674809e-01 -3.62243712e-01 -5.76773345e-01 -2.68277556e-01
-5.06144203e-02 5.95399559e-01 -5.54165721e-01 4.45614427e-01
-6.24006033e-01 -1.49561331e-01 6.01403534e-01 4.84824888e-02
1.37539041e+00 -7.70742655e-01 -7.56171107e-01 4.35008183e-02
1.54409528e+00 6.37090325e-01 1.00580919e+00 5.33429205e-01
5.92112362e-01 8.48140538e-01 3.06802958e-01 3.63952607e-01
3.59815329e-01 2.47117028e-01 1.24352232e-01 1.87216513e-02
-1.64511219e-01 -4.38494474e-01 -3.34991455e-01 9.31696236e-01
-1.42541632e-01 -6.13345385e-01 -1.01688015e+00 2.12224171e-01
-1.50540864e+00 -1.00723350e+00 -4.56984937e-01 1.38119030e+00
6.09397292e-01 1.61571711e-01 -5.55321686e-02 1.85543180e-01
7.54605770e-01 -2.07074750e-02 1.80214690e-03 -5.92183888e-01
-2.32530907e-01 6.42306404e-03 2.49580860e-01 1.25150949e-01
-6.90258205e-01 7.56551802e-01 7.11058283e+00 7.92853951e-01
-9.16111350e-01 -3.99276316e-01 4.99439150e-01 4.32404101e-01
-6.45072579e-01 -2.80387878e-01 -3.99377495e-01 4.14810181e-01
1.01856983e+00 -7.85109401e-01 -2.74572819e-01 9.00436282e-01
2.61694342e-01 -1.33198336e-01 -8.18750799e-01 1.22250974e+00
5.13328850e-01 -2.02901578e+00 9.29860771e-01 -4.82318848e-01
9.12931040e-02 -1.21293390e+00 -3.41118127e-01 3.97185460e-02
-3.73511136e-01 -1.21031654e+00 9.37753856e-01 1.21967661e+00
7.26437747e-01 -6.64508522e-01 9.06466007e-01 -1.61468163e-01
-1.02731383e+00 4.80042547e-01 -5.63305795e-01 -2.15745449e-01
-1.92790210e-01 4.34891969e-01 -7.89461195e-01 1.14351571e+00
5.96105516e-01 9.06836092e-01 -1.11452818e+00 8.12130213e-01
-4.04709190e-01 2.07586840e-01 6.28872067e-02 -1.05575776e+00
-1.16396755e-01 -1.00989826e-01 1.56210825e-01 1.60885072e+00
2.42174983e-01 3.05637330e-01 -1.58318162e-01 8.61719787e-01
-1.46929935e-01 6.70502663e-01 -9.53927338e-01 -6.72642887e-01
5.32442153e-01 1.04375172e+00 -1.44790483e+00 -7.13816226e-01
-3.92813355e-01 5.86335421e-01 9.01166499e-02 2.31655553e-01
-6.02861106e-01 -9.19825554e-01 -2.61347443e-01 3.12987715e-01
-1.94150373e-01 -2.69014597e-01 -6.49483144e-01 -8.83697331e-01
1.80899352e-01 -3.31580997e-01 3.69850785e-01 -1.65308344e+00
-1.13647926e+00 6.52979255e-01 4.12386507e-01 -1.24110043e+00
2.01017797e-01 -1.11524224e+00 -2.83585578e-01 5.78309536e-01
-1.00675881e+00 -1.33563209e+00 -3.08295846e-01 3.19107715e-03
5.19034743e-01 -1.30259469e-01 1.11465991e+00 -5.42379320e-02
-4.97504592e-01 -6.73368424e-02 -1.33467644e-01 4.56491053e-01
3.80853415e-01 -1.22944140e+00 -1.93990827e-01 4.42682952e-01
1.93991423e-01 1.02568030e+00 8.06985080e-01 -8.85035753e-01
-1.45963144e+00 -5.62073290e-01 1.06210017e+00 -5.22474110e-01
6.83182240e-01 -3.13930124e-01 -7.14892566e-01 2.05091134e-01
4.99239236e-01 -2.16294035e-01 1.04795206e+00 -8.04529935e-02
-5.99619448e-01 2.38946170e-01 -9.68298316e-01 8.41688216e-01
9.66341078e-01 -5.40312052e-01 -1.21392238e+00 3.85718316e-01
8.02734256e-01 -4.38605130e-01 -8.53832483e-01 -7.80689567e-02
6.00501001e-01 -3.95342141e-01 1.00308609e+00 -8.32063675e-01
8.62448096e-01 -3.82125169e-01 1.20825768e-01 -8.50008011e-01
8.99245515e-02 -8.04901421e-02 -2.07867339e-01 1.23568857e+00
7.05768764e-01 -2.77883410e-01 4.37191844e-01 6.19237244e-01
-3.60206395e-01 -4.85614806e-01 -6.34234369e-01 -6.74316049e-01
-4.73479852e-02 -3.05499792e-01 4.81050372e-01 8.76967967e-01
6.39087319e-01 4.81058538e-01 3.52213532e-01 -2.61561751e-01
4.46912676e-01 1.34511605e-01 2.14604363e-01 -1.24495256e+00
8.56173217e-01 -6.81519687e-01 -9.14493322e-01 -3.72044891e-01
1.64771736e-01 -1.01315308e+00 -4.98377800e-01 -2.69708562e+00
2.68679857e-01 2.25997165e-01 1.13126688e-01 4.36745673e-01
-2.91747060e-02 -5.46455849e-03 3.22110504e-01 1.04002066e-01
-8.61948431e-01 3.16976190e-01 1.58364630e+00 -4.38827515e-01
2.56606281e-01 -7.06713259e-01 -8.81622851e-01 5.79838097e-01
5.90527713e-01 -4.04647231e-01 -4.21505481e-01 9.39051993e-03
5.91804385e-01 6.74679726e-02 4.29175913e-01 -9.79905009e-01
1.39530301e-01 -2.59266019e-01 1.00418532e+00 -9.17572200e-01
-1.20147392e-01 -8.39786410e-01 -5.77757992e-02 3.37127894e-01
-5.17399371e-01 2.58822590e-01 4.77935582e-01 6.15762174e-01
-1.81136653e-01 -4.58089888e-01 1.53579906e-01 -4.05876875e-01
-4.65648741e-01 -3.41335773e-01 -6.85723722e-01 4.83823195e-02
8.25640559e-01 -4.67235267e-01 -9.24869418e-01 -3.27654630e-01
-6.48758113e-01 3.26220691e-02 4.36083317e-01 5.12720704e-01
1.10752010e+00 -1.24668157e+00 -2.87610352e-01 -1.66609287e-01
3.89263123e-01 -1.49496973e-01 1.98468864e-01 2.56346226e-01
-1.32169318e+00 8.21974754e-01 -5.52758515e-01 -8.97619799e-02
-1.18987536e+00 1.03733885e+00 -2.97966719e-01 2.37171844e-01
-9.25006449e-01 3.40966135e-01 -1.62250489e-01 1.77707881e-01
2.75554597e-01 -6.49155080e-01 -9.51931834e-01 3.44702005e-01
8.31848681e-01 4.54274118e-01 -1.04779139e-01 -5.31769693e-01
-7.65758336e-01 6.21522665e-01 1.54506937e-01 1.09795816e-01
1.23174727e+00 -1.14434116e-01 -3.02606463e-01 8.36395800e-01
1.15916836e+00 1.38676882e-01 -2.55941629e-01 3.62668037e-01
2.60479152e-01 -2.76049882e-01 -1.16362236e-01 -1.06412685e+00
-3.10594797e-01 8.36432874e-01 1.04110986e-01 7.43694007e-01
8.77989769e-01 2.15768069e-01 5.56865968e-02 4.85648870e-01
2.59907871e-01 -1.21763253e+00 1.53325692e-01 1.66256055e-01
1.50495660e+00 -9.64793205e-01 7.33826637e-01 -1.07421207e+00
-4.83560055e-01 2.04130292e+00 4.59782869e-01 2.99542278e-01
6.05769098e-01 3.65643650e-01 1.18401647e-01 -7.57947028e-01
-4.92594630e-01 -7.82140270e-02 5.32967389e-01 8.62123728e-01
8.54725063e-01 -4.28320952e-02 -7.04337597e-01 9.59449947e-01
-2.46188059e-01 2.21927725e-02 6.75662220e-01 1.44742429e+00
-5.46148777e-01 -7.42210746e-01 -6.50078237e-01 5.53785086e-01
-4.48128611e-01 -1.85333282e-01 -7.74060726e-01 1.00636029e+00
-3.08236390e-01 9.70895231e-01 3.66434962e-01 1.88878402e-01
5.27073801e-01 3.67377460e-01 6.18059754e-01 -5.70326805e-01
-1.91637442e-01 -2.04738751e-01 4.58449125e-01 -1.81254342e-01
-7.69758523e-01 -1.30453318e-01 -1.83875358e+00 -2.44613796e-01
1.21540099e-01 4.11643714e-01 9.91520047e-01 8.12516689e-01
4.16727215e-01 1.10206556e+00 -4.22631465e-02 -3.71315390e-01
5.11950195e-01 -9.60141480e-01 -4.61091012e-01 4.69176829e-01
3.01038567e-02 -7.12889791e-01 -3.29465657e-01 5.52228332e-01]
|
[11.339344024658203, 2.2110841274261475]
|
151ef7a9-22de-40a4-87da-23dd9556bdac
|
efficient-and-safe-exploration-in
|
1904.01068
| null |
http://arxiv.org/abs/1904.01068v1
|
http://arxiv.org/pdf/1904.01068v1.pdf
|
Efficient and Safe Exploration in Deterministic Markov Decision Processes with Unknown Transition Models
|
We propose a safe exploration algorithm for deterministic Markov Decision
Processes with unknown transition models. Our algorithm guarantees safety by
leveraging Lipschitz-continuity to ensure that no unsafe states are visited
during exploration. Unlike many other existing techniques, the provided safety
guarantee is deterministic. Our algorithm is optimized to reduce the number of
actions needed for exploring the safe space. We demonstrate the performance of
our algorithm in comparison with baseline methods in simulation on navigation
tasks.
|
['Erdem Biyik', 'Shahrouz Ryan Alimo', 'Jonathan Margoliash', 'Dorsa Sadigh']
|
2019-04-01
| null | null | null | null |
['safe-exploration']
|
['robots']
|
[ 1.54047042e-01 4.80886906e-01 -5.17944634e-01 -2.62706578e-01
-9.60315585e-01 -8.03944170e-01 5.04726470e-01 2.08096072e-01
-6.27858281e-01 1.01028490e+00 1.92321092e-01 -1.01243675e+00
-1.27263278e-01 -8.14463735e-01 -8.02313149e-01 -5.37835181e-01
-8.73765171e-01 2.22062126e-01 5.07079244e-01 -9.52708349e-02
3.59954238e-01 5.46980023e-01 -1.02427626e+00 -3.43843907e-01
7.57655263e-01 8.11236501e-01 -4.08164382e-01 8.11957181e-01
5.23459613e-01 7.03376830e-01 -1.36703566e-01 3.13739210e-01
5.40518701e-01 -2.71411479e-01 -1.09415293e+00 -3.70696902e-01
-2.80942053e-01 -7.09340394e-01 -4.43899602e-01 1.20205951e+00
-1.54408310e-02 7.33043551e-01 4.52691495e-01 -1.50829339e+00
3.01678061e-01 7.56818771e-01 -2.81924218e-01 1.08015426e-01
3.88679504e-01 4.19981748e-01 5.69535434e-01 4.70253127e-03
6.33592308e-01 1.41722214e+00 4.81947571e-01 8.08695257e-01
-1.39643300e+00 -6.65440023e-01 7.36908138e-01 -6.12680763e-02
-1.15385079e+00 -4.16540593e-01 7.71167874e-02 -1.40160903e-01
1.06915307e+00 1.71026021e-01 5.81230521e-01 1.26353610e+00
8.46390307e-01 6.31984293e-01 1.23950124e+00 -5.58256246e-02
8.27662110e-01 -4.00078237e-01 2.53729701e-01 6.55090570e-01
4.20001566e-01 1.09726822e+00 -2.99260676e-01 -5.85166335e-01
7.13839054e-01 -1.79274276e-01 3.57660428e-02 -7.20707417e-01
-1.03980386e+00 8.81357610e-01 -1.73856363e-01 -5.97179592e-01
-2.25448906e-01 7.41765559e-01 5.70333064e-01 3.87178481e-01
-1.26636013e-01 2.27522865e-01 -1.80092290e-01 -7.84525990e-01
-5.03278971e-01 7.39745677e-01 1.04494619e+00 1.18400133e+00
1.85319126e-01 1.51437044e-01 -1.68171853e-01 -3.60094398e-01
3.78132045e-01 6.89210951e-01 -2.59988517e-01 -1.51848269e+00
3.71704072e-01 -2.59821545e-02 9.89736676e-01 -3.16047341e-01
-2.79780447e-01 -2.04201080e-02 -3.86890888e-01 9.90175843e-01
3.16344917e-01 -5.16552150e-01 -9.19905424e-01 1.79020572e+00
5.55455983e-01 8.04860070e-02 3.40608001e-01 5.92238426e-01
-5.36845028e-01 6.27981126e-01 6.40502274e-02 -3.68092984e-01
5.35591602e-01 -7.00503230e-01 -8.79026949e-01 -1.94823354e-01
4.04679120e-01 5.23390621e-02 9.18621719e-01 6.67233586e-01
-1.08014238e+00 1.03667580e-01 -1.08376014e+00 5.01585960e-01
1.77740231e-01 -9.45912659e-01 5.39127231e-01 8.65159810e-01
-8.77807558e-01 7.95575440e-01 -1.68702316e+00 4.90097515e-03
3.67362559e-01 2.82690376e-01 -1.43388659e-01 3.69421065e-01
-1.11593544e+00 1.02339554e+00 4.66621131e-01 3.67243998e-02
-2.10643053e+00 -2.46162370e-01 -1.04043555e+00 -1.20337345e-01
7.40503550e-01 -2.47496516e-01 1.64156723e+00 9.42194685e-02
-1.76363444e+00 -1.46938026e-01 -9.44974869e-02 -1.12789834e+00
9.30083394e-01 -5.93259573e-01 -7.42128119e-02 4.04864103e-02
-2.45415326e-02 1.31084695e-01 5.60281038e-01 -1.23687816e+00
-7.96378136e-01 3.11016683e-02 9.62622762e-02 3.07857871e-01
2.93084651e-01 -1.21949613e-01 1.37260497e-01 -1.08129889e-01
1.59833375e-02 -1.24395609e+00 -1.12527204e+00 6.45572171e-02
-4.76028353e-01 1.70618519e-01 5.68376958e-01 -2.22270727e-01
1.23986125e+00 -1.91692853e+00 -7.19228461e-02 6.82180643e-01
-2.17375040e-01 -2.72320002e-01 2.51803815e-01 5.40590107e-01
4.42373216e-01 8.77188519e-02 -4.41936523e-01 -1.24542281e-01
2.67901272e-01 4.17580694e-01 -1.01504600e+00 8.76345456e-01
-3.87406945e-01 4.08351719e-01 -1.17531633e+00 -4.00265902e-01
2.28891522e-01 4.30141278e-02 -5.11251450e-01 2.36928210e-01
-2.80812591e-01 6.07722700e-01 -7.07305908e-01 3.93153042e-01
5.13464153e-01 4.85365927e-01 2.26369187e-01 1.04644299e+00
-4.17164534e-01 6.85900569e-01 -1.38565266e+00 1.41387153e+00
-4.56168264e-01 8.33293051e-02 3.37011814e-01 -4.03680354e-01
4.40956712e-01 2.12408990e-01 2.76762933e-01 -3.09337348e-01
1.73507065e-01 -1.28281564e-01 -2.91842371e-01 2.19555553e-02
3.08153600e-01 -3.54000509e-01 -6.38959467e-01 7.24964082e-01
-6.58252656e-01 -1.61750555e-01 -3.58814150e-01 2.79738098e-01
1.38699412e+00 5.29167771e-01 3.74883413e-01 -7.50580251e-01
1.10345602e-01 2.98041612e-01 7.68331647e-01 1.21271849e+00
-8.07202101e-01 -4.13118869e-01 7.98346102e-01 -3.37621719e-01
-5.34304738e-01 -1.37238955e+00 -1.12824924e-01 7.86797881e-01
7.62508273e-01 -3.58298540e-01 -7.50541449e-01 -7.78243065e-01
5.20332530e-02 1.27981913e+00 -9.64099646e-01 -3.45646352e-01
-4.82178807e-01 7.22898915e-03 7.70914972e-01 7.00595498e-01
2.42711484e-01 -7.39759147e-01 -1.53396297e+00 2.13387072e-01
1.12649940e-01 -6.57973111e-01 -4.15365249e-01 5.95030308e-01
-1.04206622e+00 -9.94217932e-01 3.08286160e-01 -1.34989500e-01
5.99372327e-01 -5.04513867e-02 3.89113486e-01 -3.48699063e-01
2.19271645e-01 3.20573539e-01 5.96236363e-02 -3.85533720e-01
-4.94345158e-01 -2.93839216e-01 3.07073742e-01 -6.52090967e-01
-1.51831746e-01 -2.75936335e-01 -4.95063365e-01 4.21771139e-01
-4.67778921e-01 -1.32465452e-01 -3.19996066e-02 6.63025618e-01
8.41443360e-01 2.85709977e-01 1.08273216e-01 -6.47413552e-01
7.22533047e-01 -4.32605028e-01 -1.21634948e+00 5.42832874e-02
-7.98387945e-01 4.88416702e-01 4.70502526e-01 -3.49931508e-01
-1.24393213e+00 2.30365664e-01 1.39266416e-01 -7.29647577e-02
-1.76115870e-01 2.94261035e-02 -1.25017971e-01 -1.50041252e-01
5.09242356e-01 2.04711378e-01 1.17103621e-01 -1.18047491e-01
3.45905364e-01 -6.18632771e-02 4.17614460e-01 -9.94113564e-01
7.33326614e-01 8.39839995e-01 3.21135521e-01 -1.20957665e-01
-3.50681990e-01 8.64002481e-02 -1.81781977e-01 -1.43039092e-01
5.78013301e-01 -5.89387953e-01 -1.30185735e+00 1.02610387e-01
-5.62752485e-01 -1.09379339e+00 -4.20675308e-01 4.43937272e-01
-9.95168626e-01 4.12714332e-01 -6.85012400e-01 -1.66635334e+00
-1.33861974e-01 -1.28167415e+00 8.04780245e-01 8.64267424e-02
-4.90237117e-01 -7.89627254e-01 5.37495494e-01 -5.08179128e-01
3.02554220e-01 6.11599267e-01 3.71494681e-01 -4.95268673e-01
-6.29974186e-01 -3.70324701e-02 4.52611864e-01 -3.98507118e-01
-1.28845692e-01 -2.04259917e-01 -4.38101739e-01 -5.68508863e-01
6.34845048e-02 -1.92927420e-01 6.47926748e-01 3.86171162e-01
1.08690035e+00 -8.32027495e-01 -7.66732216e-01 5.40889502e-01
1.20121622e+00 6.89934790e-01 3.79198581e-01 7.91998684e-01
4.01210375e-02 3.86768132e-01 1.32294786e+00 7.40987360e-01
2.34865814e-01 1.83237061e-01 8.89132023e-01 4.43630725e-01
8.04374278e-01 -5.46445549e-01 7.99558938e-01 -4.22332466e-01
1.16440602e-01 -1.52958352e-02 -1.18485630e+00 8.39247286e-01
-2.10019636e+00 -9.25039768e-01 1.39712274e-01 2.60215116e+00
8.35785151e-01 8.25398326e-01 2.14941472e-01 -7.25133047e-02
2.03358725e-01 -1.04434066e-01 -7.73818791e-01 -1.01410139e+00
5.46872497e-01 1.89256042e-01 1.06052077e+00 1.29819763e+00
-1.23807120e+00 9.09912050e-01 8.28885937e+00 3.19264889e-01
-4.12341177e-01 7.48219416e-02 1.46215647e-01 -5.37339568e-01
-4.14047509e-01 5.22451758e-01 -8.67754638e-01 1.33957416e-01
1.19428229e+00 -4.64686275e-01 3.71275485e-01 1.21580756e+00
3.20874989e-01 -5.20792484e-01 -1.10612750e+00 1.02030933e-01
-6.73362076e-01 -9.74031746e-01 -5.27428865e-01 7.79033527e-02
6.66099548e-01 -2.91611791e-01 1.45823330e-01 3.55233103e-01
1.46440995e+00 -1.11984396e+00 1.00079060e+00 2.91842222e-01
4.83000934e-01 -1.34477222e+00 3.79329294e-01 4.20372099e-01
-1.15545881e+00 -2.39058763e-01 1.26624048e-01 -2.83564568e-01
8.49514723e-01 1.05110064e-01 -4.78658706e-01 2.07955196e-01
6.76493764e-01 1.56488240e-01 1.73262462e-01 7.46695161e-01
-6.39709413e-01 6.15712285e-01 -5.53563297e-01 -6.36071898e-03
6.96299493e-01 -1.61895856e-01 8.99859488e-01 9.55896616e-01
-5.41726649e-02 1.48461089e-01 6.08700216e-01 7.00309932e-01
8.50067377e-01 -6.32375777e-01 -8.93842816e-01 2.25371033e-01
7.41771638e-01 4.41743404e-01 -7.20180213e-01 -2.94884086e-01
1.72770575e-01 6.98495984e-01 4.98419665e-02 4.88192946e-01
-1.20285308e+00 -5.95752954e-01 1.11424685e+00 -1.81838781e-01
1.58341140e-01 -7.75062740e-01 -5.11334598e-01 -4.91511822e-01
-1.92712098e-01 -5.48318148e-01 5.85533321e-01 -4.15958017e-02
-5.96467078e-01 5.41785061e-01 3.93829942e-01 -9.74611282e-01
-4.20926213e-01 -3.71360034e-02 -5.98147988e-01 7.14378417e-01
-1.17410994e+00 -5.64507186e-01 1.62966982e-01 5.61620295e-01
3.44749868e-01 3.85684311e-01 6.52473688e-01 -5.57487428e-01
-4.77805704e-01 3.64245564e-01 -5.62820956e-03 -4.48417187e-01
4.38616991e-01 -1.27515888e+00 9.27088678e-01 1.37410939e+00
-7.33088434e-01 9.55120146e-01 1.19946170e+00 -1.20973682e+00
-1.50137746e+00 -1.06968546e+00 4.57663573e-02 -5.86431682e-01
8.53164852e-01 -3.40263456e-01 -8.83324802e-01 1.22165215e+00
1.00700699e-01 -2.96146125e-01 3.20918322e-01 6.34498820e-02
-2.36354738e-01 3.85689557e-01 -1.15976596e+00 9.51985657e-01
1.26173198e+00 -2.52477467e-01 -5.03093958e-01 9.07355174e-03
8.90327990e-01 -7.38606572e-01 -6.52586460e-01 2.94958442e-01
6.27059221e-01 -5.48520386e-01 6.12283409e-01 -8.48572373e-01
-2.41456106e-01 -4.44506258e-01 -1.06323287e-01 -1.09574056e+00
-1.31827340e-01 -1.48828471e+00 -6.37300968e-01 3.48830104e-01
4.30084199e-01 -8.42133939e-01 7.08734691e-01 1.09025812e+00
-1.75397426e-01 -7.55187988e-01 -1.36677134e+00 -1.34649193e+00
1.56381041e-01 -5.74471712e-01 5.32779992e-01 3.43777359e-01
6.91207170e-01 -3.88707310e-01 -5.07063985e-01 5.14270246e-01
1.14582479e+00 1.69385239e-01 4.83270377e-01 -4.59840596e-01
-6.06479757e-02 -2.21665204e-01 1.26330093e-01 -7.84698486e-01
4.47142988e-01 -4.14018303e-01 7.29681373e-01 -1.41324365e+00
-5.39624281e-02 -6.53309703e-01 -2.70871997e-01 6.20749116e-01
1.00782681e-02 -7.17597902e-01 -1.34023532e-01 -2.26912037e-01
-8.99116278e-01 7.48979568e-01 9.41657841e-01 2.58038163e-01
-5.03781140e-01 1.82029039e-01 -4.92639095e-01 6.03745282e-01
1.06983292e+00 -5.57243407e-01 -6.51949346e-01 -4.06494318e-03
1.06096953e-01 4.42666560e-01 1.87552229e-01 -8.95791888e-01
8.99806321e-02 -1.01681483e+00 -3.38599950e-01 -7.39546120e-01
1.75302848e-01 -8.86849761e-01 3.99375826e-01 1.34928918e+00
-8.19162965e-01 3.72002423e-02 4.70206410e-01 1.21127403e+00
3.02185774e-01 1.63912177e-01 8.64732027e-01 2.74040997e-01
-7.49018013e-01 3.29375863e-01 -1.23091042e+00 6.58663437e-02
1.55704713e+00 -1.28642142e-01 -1.24490477e-01 -5.15820265e-01
-9.26241755e-01 8.48215342e-01 7.32229829e-01 1.05618186e-01
6.87545538e-01 -1.03898549e+00 -4.02018577e-02 1.09065333e-02
-1.70717001e-01 7.34998062e-02 1.68010801e-01 6.92722857e-01
-4.54649895e-01 3.57519329e-01 -3.85986239e-01 -1.43207923e-01
-1.06964934e+00 1.00659907e+00 3.95214796e-01 -3.33356351e-01
-9.32597399e-01 6.74018204e-01 -4.14011478e-02 -9.37787816e-02
6.48385584e-01 -4.61555481e-01 3.20035130e-01 -6.79750979e-01
6.58776104e-01 7.63838112e-01 -5.14129877e-01 1.65410250e-01
-7.08251119e-01 -1.18153254e-02 2.37055942e-02 -1.03886771e+00
9.56196666e-01 -2.87787110e-01 2.41116926e-01 4.98993605e-01
3.75568360e-01 -5.64914010e-02 -1.98521662e+00 3.36519748e-01
1.57729581e-01 -6.07973695e-01 -6.32126704e-02 -5.87880075e-01
-2.11977676e-01 6.67326212e-01 5.15501380e-01 -2.48781383e-01
6.84053183e-01 -4.89147544e-01 7.41798997e-01 5.81035078e-01
1.08353162e+00 -1.17594063e+00 -5.39201736e-01 6.78546906e-01
5.25492907e-01 -8.12690794e-01 -5.96391074e-02 -3.39354396e-01
-9.29559112e-01 6.33315504e-01 7.02286243e-01 -3.75021100e-01
5.35637736e-01 9.13667142e-01 -2.39965767e-01 1.77858233e-01
-1.08171856e+00 2.51867697e-02 -5.66311955e-01 6.83990002e-01
-4.61897939e-01 3.18817109e-01 -2.40458280e-01 4.89928693e-01
-3.38151269e-02 -3.76207009e-02 7.87131727e-01 1.72053289e+00
-8.00690234e-01 -9.59306002e-01 -5.37587702e-01 9.55917165e-02
-3.58478427e-01 2.43441164e-01 -6.50751814e-02 7.27186203e-01
-6.64158404e-01 1.05717099e+00 -1.25042096e-01 -2.15707794e-01
1.34472474e-01 -4.50593531e-02 4.95294690e-01 -3.51873428e-01
-4.07738119e-01 -4.95100766e-02 3.96842927e-01 -1.36493421e+00
2.00488806e-01 -8.79333138e-01 -1.60578215e+00 -4.11104769e-01
-6.94716349e-02 3.75341296e-01 3.15500975e-01 5.87090492e-01
1.19297929e-01 4.59001094e-01 6.15095496e-01 -3.55594814e-01
-1.12550473e+00 -1.68823153e-01 -4.36083257e-01 -1.95910588e-01
7.80822694e-01 -9.40449774e-01 -3.68719339e-01 -3.00562948e-01]
|
[4.5413408279418945, 2.1394903659820557]
|
c3d2b3cf-eaea-43e1-a139-4e5d982c40f1
|
recovering-the-unbiased-scene-graphs-from-the
|
2107.02112
| null |
https://arxiv.org/abs/2107.02112v1
|
https://arxiv.org/pdf/2107.02112v1.pdf
|
Recovering the Unbiased Scene Graphs from the Biased Ones
|
Given input images, scene graph generation (SGG) aims to produce comprehensive, graphical representations describing visual relationships among salient objects. Recently, more efforts have been paid to the long tail problem in SGG; however, the imbalance in the fraction of missing labels of different classes, or reporting bias, exacerbating the long tail is rarely considered and cannot be solved by the existing debiasing methods. In this paper we show that, due to the missing labels, SGG can be viewed as a "Learning from Positive and Unlabeled data" (PU learning) problem, where the reporting bias can be removed by recovering the unbiased probabilities from the biased ones by utilizing label frequencies, i.e., the per-class fraction of labeled, positive examples in all the positive examples. To obtain accurate label frequency estimates, we propose Dynamic Label Frequency Estimation (DLFE) to take advantage of training-time data augmentation and average over multiple training iterations to introduce more valid examples. Extensive experiments show that DLFE is more effective in estimating label frequencies than a naive variant of the traditional estimate, and DLFE significantly alleviates the long tail and achieves state-of-the-art debiasing performance on the VG dataset. We also show qualitatively that SGG models with DLFE produce prominently more balanced and unbiased scene graphs.
|
['Jiashi Feng', 'Roger Zimmermann', 'Changhu Wang', 'Hanshu Yan', 'Henghui Ding', 'Meng-Jiun Chiou']
|
2021-07-05
| null | null | null | null |
['visual-relationship-detection', 'unbiased-scene-graph-generation']
|
['computer-vision', 'computer-vision']
|
[ 3.89464110e-01 1.85212702e-01 -5.51035404e-01 -3.18536103e-01
-9.00010645e-01 -5.64010739e-01 4.55820471e-01 -1.96115933e-02
-2.61301585e-02 9.52428997e-01 2.52356887e-01 -1.42062873e-01
6.58445507e-02 -5.30618906e-01 -9.63747323e-01 -9.11646307e-01
2.32157260e-01 5.04726291e-01 9.07445922e-02 3.40866804e-01
1.44077733e-01 2.13205472e-01 -1.52756691e+00 8.83862898e-02
1.06410491e+00 7.39862025e-01 1.48052260e-01 3.41855139e-01
-2.00677559e-01 1.07170630e+00 -6.81396723e-01 -3.91367614e-01
1.25393435e-01 -5.70309639e-01 -7.34284043e-01 4.86951500e-01
1.01629710e+00 -3.52020830e-01 -3.82772475e-01 1.44693804e+00
3.23061138e-01 1.63717955e-01 1.00017333e+00 -1.65568030e+00
-8.45409453e-01 5.36859214e-01 -1.24129176e+00 1.58662528e-01
-7.65175447e-02 -1.53911896e-02 1.09587586e+00 -9.50213075e-01
8.45269561e-01 1.42674398e+00 5.07810652e-01 6.10836923e-01
-1.63233304e+00 -8.48028183e-01 4.36010152e-01 2.36163199e-01
-1.33211374e+00 -5.80689833e-02 9.67689812e-01 -5.73175371e-01
2.72384673e-01 3.45296204e-01 5.09228528e-01 1.12910759e+00
-2.02170148e-01 1.08852816e+00 1.16467905e+00 -4.63534832e-01
2.66052157e-01 3.04056942e-01 5.44020414e-01 7.61326790e-01
6.23014629e-01 2.13293731e-02 -6.61898375e-01 -3.26247185e-01
6.50980413e-01 -2.13184562e-02 -5.27066350e-01 -8.60793591e-01
-9.85220730e-01 8.17081273e-01 6.37720227e-01 -1.54455811e-01
-6.83094636e-02 3.13003182e-01 2.40096465e-01 -1.65244207e-01
8.70822012e-01 3.85010302e-01 -1.53020665e-01 3.47705871e-01
-8.96733940e-01 2.41517127e-01 5.25555968e-01 1.06838608e+00
1.00011468e+00 7.82588869e-02 -5.18653512e-01 9.50245202e-01
1.49983531e-02 7.12531805e-01 2.23113954e-01 -1.03033912e+00
2.19926938e-01 6.61220014e-01 9.56496298e-02 -9.88819599e-01
-7.27297589e-02 -7.10768163e-01 -1.02667487e+00 1.42414793e-01
5.80764472e-01 -1.54156303e-02 -1.28345561e+00 2.23198462e+00
3.84846538e-01 1.92945153e-01 -2.24245906e-01 8.89869630e-01
6.71308279e-01 8.19471240e-01 3.70834261e-01 -3.86332005e-01
1.20670283e+00 -9.90708172e-01 -7.94706762e-01 -5.17795205e-01
5.24590492e-01 -5.33443809e-01 1.24821389e+00 3.58620256e-01
-7.36741841e-01 -2.92767256e-01 -9.52250957e-01 -1.91185951e-01
-1.34649947e-01 2.89250582e-01 7.64482737e-01 5.31857252e-01
-8.03470731e-01 3.87434304e-01 -3.26585889e-01 -1.95641518e-01
8.31024289e-01 -4.73813852e-03 -2.20039010e-01 -4.79802728e-01
-8.75115395e-01 6.07283175e-01 2.96074867e-01 -3.72042239e-01
-1.03573167e+00 -1.00086856e+00 -8.91642988e-01 6.31788969e-02
7.29604125e-01 -6.45739734e-01 9.70659614e-01 -1.06210017e+00
-6.12590134e-01 9.81297433e-01 -4.59159315e-01 -2.56155849e-01
5.14387369e-01 -1.68732584e-01 2.44381893e-02 1.55155538e-02
3.50100935e-01 9.44685340e-01 1.03074324e+00 -1.88554657e+00
-5.59931278e-01 -5.09917200e-01 -1.67504445e-01 2.51630813e-01
-2.01431364e-01 -4.71846104e-01 -4.48110223e-01 -9.27021503e-01
1.13570258e-01 -1.00939786e+00 -7.60708051e-03 1.30375206e-01
-6.07106805e-01 -2.31878534e-01 9.19500828e-01 -6.06725216e-01
1.17832422e+00 -2.24813151e+00 2.57199481e-02 1.12544745e-01
5.45367539e-01 1.64332837e-01 2.95744743e-03 7.86454380e-02
-2.19382063e-01 5.21300873e-03 -2.64979064e-01 -4.39061642e-01
-6.72104806e-02 3.49982351e-01 -6.18263543e-01 5.27832747e-01
8.04892648e-03 7.82241523e-01 -1.27778995e+00 -6.10246599e-01
1.50472596e-01 1.81966588e-01 -4.30504858e-01 1.21856086e-01
-3.63118261e-01 2.16912419e-01 -1.86070055e-02 5.66415668e-01
8.96820426e-01 -7.68596828e-01 1.52156010e-01 -3.59632164e-01
4.69844937e-01 -1.98846549e-01 -1.14756393e+00 1.19800401e+00
-1.48638546e-01 7.16672361e-01 -1.97519436e-01 -7.62258112e-01
7.28547037e-01 -1.68311745e-01 1.16294786e-01 -3.49495977e-01
4.55983169e-03 9.32868756e-03 -4.93541211e-01 -1.75760761e-01
6.49901927e-01 -2.73453683e-01 6.63079321e-03 4.53484148e-01
1.23863742e-01 -1.01881638e-01 2.94169873e-01 7.67751098e-01
7.19804585e-01 -1.16977341e-01 3.61713946e-01 -2.39286274e-01
1.12752989e-02 2.73221936e-02 5.85315287e-01 9.84080315e-01
-2.39280879e-01 6.53409898e-01 6.89875424e-01 -1.97378442e-01
-1.01632142e+00 -1.09564555e+00 -4.14010286e-02 1.17561269e+00
4.93242323e-01 -2.96275467e-01 -8.03487897e-01 -1.09703827e+00
2.44160771e-01 9.13037777e-01 -8.22849154e-01 -4.96825725e-01
-1.75842583e-01 -1.06042981e+00 1.06834628e-01 4.33071971e-01
4.57380891e-01 -8.50291133e-01 -1.52631208e-01 -1.67617768e-01
-4.07688588e-01 -7.95987248e-01 -6.58064604e-01 1.97620586e-01
-6.45888746e-01 -1.09123695e+00 -9.04936373e-01 -7.42846251e-01
1.11948884e+00 7.54764736e-01 1.27129638e+00 -7.60156438e-02
-2.90814966e-01 1.71241298e-01 -1.76193416e-01 -3.44795883e-01
-3.36347461e-01 -1.99003056e-01 -2.25529507e-01 2.38374788e-02
1.40071392e-01 -3.23543847e-01 -4.95108843e-01 1.95672572e-01
-9.16244328e-01 3.85666341e-01 4.50482607e-01 1.17339253e+00
7.29496896e-01 -7.76434392e-02 7.11978614e-01 -1.44610715e+00
3.67226571e-01 -4.16046143e-01 -5.72371125e-01 3.81465077e-01
-8.37818563e-01 2.60201097e-01 3.71430159e-01 -7.58300662e-01
-1.18909311e+00 -7.63773099e-02 4.01326060e-01 -7.12528467e-01
4.93989773e-02 1.87982693e-01 -2.07790300e-01 -2.64037289e-02
7.76310742e-01 1.67857125e-01 -1.64371759e-01 -2.78477699e-01
7.39102006e-01 4.53717828e-01 7.04702497e-01 -5.00407815e-01
6.60562754e-01 6.65228844e-01 2.15115026e-02 -5.98599553e-01
-1.43940496e+00 -5.10674059e-01 -1.08917750e-01 -4.40312922e-01
4.66495752e-01 -9.38060045e-01 -2.50541419e-01 5.37551284e-01
-9.46084678e-01 -3.29324573e-01 -6.12620354e-01 1.50797158e-01
-3.02240819e-01 5.36189198e-01 -4.13548321e-01 -9.82383966e-01
-2.41275981e-01 -8.58386457e-01 1.09592605e+00 3.46972495e-01
-1.90836608e-01 -8.23580384e-01 -2.59323381e-02 3.44466239e-01
8.20453465e-02 1.68327123e-01 1.28061128e+00 -3.05933148e-01
-6.24540389e-01 -1.89416558e-01 -6.91420138e-01 4.56139833e-01
1.47321448e-01 -1.47003382e-01 -1.13219869e+00 -4.79221493e-01
-2.81364292e-01 -4.92467225e-01 1.18607068e+00 6.74148142e-01
1.38077295e+00 -3.25269312e-01 -5.48217177e-01 3.58774811e-01
1.40881896e+00 -2.92144697e-02 4.74440426e-01 -1.75512433e-01
1.05408192e+00 6.00008726e-01 6.93841696e-01 5.46903610e-01
3.44745219e-01 5.09335637e-01 5.17992735e-01 -3.74589801e-01
-6.09756351e-01 -5.96300721e-01 -4.56411503e-02 3.27269703e-01
4.52945858e-01 -4.04623240e-01 -6.92861199e-01 7.72809625e-01
-1.90057874e+00 -8.52804661e-01 -1.90077394e-01 2.32858729e+00
1.01506495e+00 5.28990198e-03 4.77665626e-02 7.84858689e-02
1.21738899e+00 2.85851061e-01 -8.80725205e-01 2.99244672e-01
-2.31860951e-01 -2.06142634e-01 7.23728895e-01 4.91598636e-01
-1.19815230e+00 9.80034590e-01 6.53307962e+00 1.24065614e+00
-8.53997886e-01 2.39825193e-02 1.14592791e+00 -9.69443619e-02
-4.35920775e-01 1.47956967e-01 -7.61305809e-01 5.35782695e-01
1.91103697e-01 -3.01850498e-01 4.58749980e-01 1.12732720e+00
-1.18008800e-01 -3.65224212e-01 -9.57506835e-01 1.21185791e+00
3.08371872e-01 -1.30500817e+00 3.54690701e-01 5.35446666e-02
1.21030891e+00 -3.71936500e-01 2.79092491e-01 2.95900553e-01
7.16563225e-01 -6.94865763e-01 9.53206897e-01 2.71243453e-01
9.09056008e-01 -6.95743084e-01 5.16588807e-01 4.38072741e-01
-8.81035745e-01 -8.34169798e-03 -5.57026327e-01 2.12287098e-01
1.18181191e-01 1.10063112e+00 -7.56553650e-01 4.27331418e-01
5.95247149e-01 7.55561113e-01 -7.57410467e-01 1.05470657e+00
-3.37930918e-01 7.44466960e-01 -6.72762468e-02 3.05482582e-03
-5.76757975e-02 -1.53251186e-01 5.32197237e-01 9.41189766e-01
2.02528425e-02 -1.17181398e-01 1.64342910e-01 1.04772139e+00
-4.07888204e-01 -1.22022144e-01 -5.68898678e-01 1.59725398e-01
5.23056328e-01 1.22818744e+00 -7.53106058e-01 -6.85551405e-01
-1.29050136e-01 8.12862456e-01 6.59498334e-01 5.59276402e-01
-8.94066453e-01 -4.01084274e-02 3.09879392e-01 6.52425140e-02
1.59141839e-01 2.77993202e-01 -4.27762359e-01 -1.24512088e+00
-1.03825681e-01 -7.32123256e-01 5.93346238e-01 -1.10375702e+00
-1.61484349e+00 1.18818507e-01 1.75024122e-01 -1.05368793e+00
-6.64016157e-02 -2.86332130e-01 -3.01882029e-01 7.01803207e-01
-1.35601342e+00 -1.04591703e+00 -5.21807194e-01 4.23273742e-01
4.88797963e-01 4.73133087e-01 3.13455999e-01 2.24508911e-01
-6.19847536e-01 5.62394142e-01 1.18978001e-01 -1.95297077e-01
9.87440586e-01 -1.55164862e+00 5.07907718e-02 9.59455252e-01
2.31349006e-01 3.56783926e-01 8.47400129e-01 -8.79090607e-01
-8.54619324e-01 -1.33506882e+00 7.41069734e-01 -2.76860237e-01
3.99143398e-01 -3.70002180e-01 -1.15672529e+00 7.04638183e-01
-2.90143490e-02 1.19804643e-01 4.17992264e-01 4.50981222e-02
-6.39306426e-01 -1.91790350e-02 -1.14672732e+00 7.22811401e-01
1.18929374e+00 -3.07713181e-01 -3.75571132e-01 4.84986633e-01
6.53876543e-01 -3.24081361e-01 -1.78841695e-01 3.68739456e-01
1.33712962e-01 -7.62720406e-01 8.88752937e-01 -4.19881225e-01
4.97020274e-01 -3.57828677e-01 3.77019718e-02 -1.32601666e+00
-3.92387450e-01 -1.62161484e-01 -1.94984227e-01 1.36458945e+00
2.26311594e-01 -4.04083610e-01 1.01306069e+00 5.52054226e-01
1.40389070e-01 -4.15434539e-01 -7.92939544e-01 -7.68490195e-01
-6.68378398e-02 -7.76469335e-02 3.12924951e-01 1.14548707e+00
-1.88397512e-01 5.11150122e-01 -8.67996216e-01 7.67690986e-02
1.01303267e+00 4.25942034e-01 8.13355207e-01 -1.22248006e+00
-2.67258495e-01 -1.67953461e-01 -1.45683601e-01 -1.11837280e+00
1.56854346e-01 -8.76591146e-01 2.41188169e-01 -1.55210400e+00
9.03669596e-01 -3.47874850e-01 1.77086126e-02 4.93822545e-01
-7.64617205e-01 2.80168056e-01 1.38063639e-01 2.87407815e-01
-7.93967903e-01 6.08115375e-01 1.41011596e+00 -3.57315570e-01
9.28966627e-02 -3.44939530e-01 -8.92230868e-01 7.25001633e-01
4.23621476e-01 -6.17118299e-01 -5.60993254e-01 -3.77839319e-02
2.08473802e-01 -1.40968740e-01 4.90036756e-01 -7.11299658e-01
2.36660652e-02 -1.91658512e-01 5.04130483e-01 -7.52911150e-01
1.60574213e-01 -6.25338554e-01 2.84055322e-01 2.76651114e-01
-5.26841760e-01 -3.61328006e-01 -4.88806665e-02 1.00505114e+00
1.51317343e-02 -2.13742629e-01 9.97762084e-01 -1.21968940e-01
-7.22293496e-01 1.42934978e-01 -1.55897379e-01 3.73188913e-01
9.99341369e-01 1.51945511e-04 -8.13098848e-01 -4.95798022e-01
-5.79285681e-01 2.67172098e-01 6.68638766e-01 5.90070114e-02
4.14648771e-01 -1.46336031e+00 -4.62617010e-01 -4.05334774e-03
3.57925147e-01 1.40804932e-01 5.16547918e-01 5.55530190e-01
-1.37641072e-01 -8.97212848e-02 1.31499261e-01 -6.12787127e-01
-1.34834814e+00 8.40876281e-01 8.99838433e-02 -5.71318269e-01
-4.21913177e-01 9.74127412e-01 8.79837573e-01 -3.48624401e-02
3.99315894e-01 3.48214991e-02 1.19095563e-03 1.73204362e-01
3.82332057e-01 5.32889426e-01 -2.18662858e-01 -4.50449586e-01
-1.08737200e-01 3.74172628e-01 -2.90104210e-01 9.24089253e-02
9.13518250e-01 -2.30975464e-01 1.24715120e-02 3.59756827e-01
1.05348730e+00 -1.04749175e-02 -1.53773928e+00 -3.88657570e-01
-2.29288846e-01 -8.18259180e-01 1.68111265e-01 -9.09832776e-01
-1.24812675e+00 7.50227094e-01 4.10675675e-01 2.07385704e-01
1.04549587e+00 3.35649610e-01 4.53150958e-01 -1.11456685e-01
4.18401271e-01 -8.72512877e-01 3.35113108e-01 1.16417840e-01
6.82159364e-01 -1.26089287e+00 1.07808873e-01 -9.18985188e-01
-8.66832316e-01 5.83539307e-01 7.32808709e-01 -1.05916023e-01
3.19018543e-01 -3.16504128e-02 -9.50511843e-02 -2.54741520e-01
-5.92342734e-01 -1.46526903e-01 3.43658209e-01 6.17548883e-01
-6.72596842e-02 1.76333040e-01 -1.99106753e-01 3.66790295e-01
1.37773067e-01 -9.17012468e-02 5.39749980e-01 7.51307309e-01
-4.14686620e-01 -6.99148059e-01 -3.84927005e-01 7.48443484e-01
-1.87900305e-01 -1.52947456e-01 -4.76897746e-01 7.22488701e-01
4.59238663e-02 7.71787345e-01 1.92647346e-03 -2.14911968e-01
1.68628514e-01 3.62416916e-02 4.95107412e-01 -5.71666777e-01
-2.80353501e-02 1.88855398e-02 -3.35595980e-02 -3.72738212e-01
-4.23774004e-01 -4.96197909e-01 -8.92983496e-01 -3.66105229e-01
-7.18773425e-01 -1.83875603e-03 1.41739771e-01 6.46175563e-01
2.63847888e-01 5.68583190e-01 4.70588177e-01 -7.77391732e-01
-6.42778516e-01 -8.56396556e-01 -9.83850121e-01 8.14643145e-01
2.88524121e-01 -1.02087188e+00 -9.13366139e-01 1.83734760e-01]
|
[10.139814376831055, 1.9870936870574951]
|
0bd224c8-f1c4-44a4-b041-32915a0f4faf
|
ris-aided-joint-localization-and-1
|
2204.13484
| null |
https://arxiv.org/abs/2204.13484v1
|
https://arxiv.org/pdf/2204.13484v1.pdf
|
RIS-aided Joint Localization and Synchronization with a Single-Antenna Receiver: Beamforming Design and Low-Complexity Estimation
|
Reconfigurable intelligent surfaces (RISs) have attracted enormous interest thanks to their ability to overcome line-of-sight blockages in mmWave systems, enabling in turn accurate localization with minimal infrastructure. Less investigated are however the benefits of exploiting RIS with suitably designed beamforming strategies for optimized localization and synchronization performance. In this paper, a novel low-complexity method for joint localization and synchronization based on an optimized design of the base station (BS) active precoding and RIS passive phase profiles is proposed, for the challenging case of a single-antenna receiver. The theoretical position error bound is first derived and used as metric to jointly optimize the BS-RIS beamforming, assuming a priori knowledge of the user position. By exploiting the low-dimensional structure of the solution, a novel codebook-based robust design strategy with optimized beam power allocation is then proposed, which provides low-complexity while taking into account the uncertainty on the user position. Finally, a reduced-complexity maximum-likelihood based estimation procedure is devised to jointly recover the user position and the synchronization offset. Extensive numerical analysis shows that the proposed joint BS-RIS beamforming scheme provides enhanced localization and synchronization performance compared to existing solutions, with the proposed estimator attaining the theoretical bounds even at low signal-to-noise-ratio and in the presence of additional uncontrollable multipath propagation.
|
['Gonzalo Seco-Granados', 'Henk Wymeersch', 'Angelo Coluccia', 'Musa Furkan Keskin', 'Alessio Fascista']
|
2022-04-28
| null | null | null | null |
['robust-design']
|
['miscellaneous']
|
[ 1.61886707e-01 2.94610769e-01 3.88709933e-01 -1.94644053e-02
-8.46953869e-01 -4.86578822e-01 3.00952852e-01 1.27579004e-01
-2.72938877e-01 6.68574035e-01 -1.82783958e-02 -2.72980362e-01
-8.10188949e-01 -6.40793025e-01 -5.35303831e-01 -1.39311254e+00
-2.22050563e-01 1.28256202e-01 -1.64006442e-01 -1.18761934e-01
2.04375148e-01 7.84447491e-01 -1.37647724e+00 -8.38715196e-01
8.09480906e-01 1.13045311e+00 4.42996830e-01 6.23312116e-01
4.36045676e-01 7.16045871e-02 -6.48186326e-01 6.69241650e-03
1.15378000e-01 -6.04722239e-02 1.99267700e-01 -8.27321932e-02
-1.07251950e-01 -1.12254955e-01 -2.72227645e-01 7.14784265e-01
8.74036908e-01 -9.97459590e-02 6.34979725e-01 -6.93519652e-01
3.63659978e-01 3.88432294e-01 -4.97876853e-01 -1.08674787e-01
3.55103612e-01 -4.42374676e-01 5.59362113e-01 -5.63709497e-01
2.48006195e-01 5.21367729e-01 7.12743640e-01 -2.25017354e-01
-8.49573314e-01 -5.66074848e-01 -4.34284031e-01 3.90200131e-02
-1.83356273e+00 -6.46074712e-01 7.05442727e-01 -2.07985967e-01
2.91685730e-01 5.76073468e-01 3.87066692e-01 4.14104551e-01
3.59294444e-01 -6.44906685e-02 6.67685926e-01 -9.57309902e-01
3.24808538e-01 7.81250671e-02 -2.69766748e-01 7.64791727e-01
1.01775277e+00 4.97155301e-02 -3.46624345e-01 -2.62141138e-01
7.39127576e-01 -3.71965915e-01 -5.28361142e-01 -7.44097829e-01
-1.29769099e+00 2.24758282e-01 4.51283336e-01 8.97292793e-01
-6.00372970e-01 1.42305791e-01 -4.95761514e-01 -4.51905467e-02
1.00551076e-01 4.92231548e-01 -8.09553862e-02 2.18652576e-01
-8.79338324e-01 -1.16966106e-01 9.44746315e-01 1.15171480e+00
5.85044086e-01 2.10363910e-01 -1.48697242e-01 5.35576880e-01
7.45605350e-01 1.13152266e+00 -1.97508246e-01 -5.82481980e-01
4.75537866e-01 4.73309085e-02 6.17506146e-01 -1.21496332e+00
-9.80758727e-01 -1.68121922e+00 -9.46250379e-01 -2.96651632e-01
4.26164001e-01 -5.68956912e-01 -4.53306973e-01 1.64031672e+00
5.01136839e-01 2.19393462e-01 4.25119996e-01 4.62957174e-01
3.45987827e-01 5.27296364e-01 -7.47276485e-01 -6.25340402e-01
1.17485189e+00 -1.56807229e-01 -6.77551866e-01 -2.86593854e-01
5.36570668e-01 -7.62131155e-01 5.74805960e-02 3.81668389e-01
-8.19190264e-01 6.57954663e-02 -1.53118932e+00 8.16859245e-01
2.69101322e-01 6.64503813e-01 1.34699225e-01 1.11895382e+00
-9.04358149e-01 -9.81616676e-02 -8.12421918e-01 -2.40640536e-01
7.28621930e-02 4.94551450e-01 1.85513739e-02 -2.80465055e-02
-7.68802047e-01 6.39719069e-01 -1.33584008e-01 5.31482756e-01
-3.27380113e-02 -6.24665797e-01 -6.33451104e-01 1.02187939e-01
1.18524835e-01 -7.47418344e-01 1.04942346e+00 -3.36611003e-01
-1.76036441e+00 -4.14874740e-02 -4.42657828e-01 -3.70518476e-01
2.32014120e-01 5.19203320e-02 -4.40265387e-01 4.06782061e-01
1.18912145e-01 -4.52259809e-01 7.02316940e-01 -1.31872082e+00
-4.80779767e-01 -3.86273503e-01 -2.79466808e-01 1.24687083e-01
-2.65844971e-01 -5.50794423e-01 -2.90650159e-01 -3.89931589e-01
1.01303899e+00 -1.00831759e+00 -6.59234047e-01 -5.48002303e-01
-4.07071471e-01 5.34947693e-01 2.21195459e-01 -3.40021342e-01
1.12663257e+00 -2.08001995e+00 -3.80947702e-02 8.20283175e-01
-2.64468163e-01 4.88169901e-02 8.40459988e-02 7.75171995e-01
2.48521566e-01 -6.69314384e-01 -6.82221726e-02 -2.25241557e-01
-3.02563548e-01 -2.44870424e-01 1.46887377e-01 1.06091619e+00
-4.95088041e-01 2.89337099e-01 -7.48992026e-01 3.95923108e-02
2.61546284e-01 5.41797042e-01 -5.44620216e-01 3.05630416e-02
4.67069477e-01 9.17233169e-01 -8.56956959e-01 5.32669842e-01
8.61393094e-01 9.75633785e-03 2.87226260e-01 -3.39158088e-01
-6.04084611e-01 -1.67292744e-01 -1.57591677e+00 1.29018724e+00
-7.96219885e-01 5.59848785e-01 6.04572892e-01 -9.76868570e-01
1.04795504e+00 4.33635563e-01 7.41613209e-01 -8.67826581e-01
3.28268975e-01 5.18649817e-01 -4.22478169e-02 -3.94447356e-01
1.41168073e-01 -7.34814927e-02 -3.17544103e-01 2.09584594e-01
-2.16988191e-01 9.46505368e-03 -2.12106690e-01 1.87617466e-02
1.24795270e+00 -1.92992523e-01 5.52179337e-01 -4.92705882e-01
9.78198349e-01 -4.45121109e-01 3.18616748e-01 8.91461194e-01
5.47328889e-01 2.50238627e-01 -1.20300055e-01 2.29897097e-01
-3.90580058e-01 -6.99018061e-01 -3.37894559e-01 1.89757839e-01
7.01786637e-01 -1.36777222e-01 -6.27207994e-01 1.67532593e-01
4.28791195e-02 8.51798415e-01 -1.49213418e-01 9.96692572e-03
-6.04367077e-01 -9.27693605e-01 2.43504420e-01 -2.44780689e-01
4.46314633e-01 2.71909814e-02 -9.00770307e-01 4.23655391e-01
-8.31601843e-02 -1.09512496e+00 4.63752747e-01 1.38090685e-01
-5.84376037e-01 -8.50670159e-01 -7.15135813e-01 -5.91562331e-01
1.02113402e+00 7.16424227e-01 3.30638617e-01 -1.27704084e-01
-1.07390642e-01 9.56884742e-01 -2.52512217e-01 -2.76282132e-01
-9.41290557e-02 4.59518060e-02 1.92082062e-01 3.11990350e-01
-4.89936411e-01 -6.50626361e-01 -7.77450025e-01 7.88592517e-01
-2.88048893e-01 -2.18795165e-02 9.09122825e-01 6.01019144e-01
2.66158432e-01 1.95753485e-01 7.18331635e-01 -1.98795587e-01
-9.01224837e-02 -4.90365326e-01 -1.00131524e+00 1.04708076e-01
-3.62190217e-01 3.97625528e-02 3.93517733e-01 2.36954778e-01
-1.18538725e+00 1.49496511e-01 -2.71326274e-01 5.93165219e-01
8.51879269e-02 5.73571980e-01 -5.66355228e-01 -8.66365731e-01
4.25285816e-01 1.65356562e-01 -2.82706738e-01 -2.34418347e-01
2.84603894e-01 8.43813598e-01 3.49967986e-01 -2.28513733e-01
1.10228884e+00 6.35971189e-01 7.02405930e-01 -1.59846961e+00
-6.22329772e-01 -6.64367378e-01 -3.96353215e-01 -3.59256387e-01
1.40681580e-01 -9.77254510e-01 -8.06629121e-01 3.82382423e-01
-1.01275110e+00 2.26528600e-01 3.08275729e-01 9.74586368e-01
-4.71257150e-01 4.15580660e-01 2.57215917e-01 -1.22356880e+00
-2.58690566e-01 -8.58413815e-01 7.48103619e-01 1.49757043e-01
-2.78988145e-02 -7.47376382e-01 -8.43719691e-02 1.83193833e-01
7.35639215e-01 3.58325511e-01 3.84734660e-01 -1.08947061e-01
-9.33200717e-01 -7.77239680e-01 -1.31298155e-02 -3.93859208e-01
1.90546185e-01 -7.83435464e-01 -5.88396251e-01 -4.95561481e-01
2.61053473e-01 7.28569269e-01 1.49678096e-01 7.42952049e-01
4.20338005e-01 -2.86859870e-01 -7.61533678e-01 7.84118056e-01
1.66447902e+00 2.22095594e-01 4.56739843e-01 4.02352571e-01
2.94575244e-01 3.36177677e-01 8.20843399e-01 8.58401775e-01
3.21483672e-01 8.46411467e-01 6.26733184e-01 1.44646212e-01
1.92935675e-01 5.28281808e-01 -9.88898352e-02 7.04350412e-01
9.53446180e-02 -6.69589520e-01 -5.44446826e-01 3.33141357e-01
-1.70051670e+00 -5.94301522e-01 -5.33055604e-01 2.54771757e+00
2.54842173e-02 -1.88384786e-01 -5.11957228e-01 2.47109294e-01
4.78273690e-01 -8.11365843e-02 -2.10408121e-01 2.39599645e-01
-1.72820285e-01 4.93260324e-02 1.25721157e+00 7.55485058e-01
-9.15635586e-01 7.60026202e-02 4.76401091e+00 5.56745052e-01
-9.80665326e-01 9.70437080e-02 -2.21391648e-01 1.12238526e-01
-4.87365305e-01 -1.34806350e-01 -8.20781767e-01 2.51285255e-01
7.92994142e-01 -6.16513602e-02 -1.05793945e-01 3.79496157e-01
3.88365209e-01 -5.57099998e-01 -4.22973275e-01 1.06208658e+00
4.89001572e-02 -1.13266957e+00 -5.57424426e-01 2.13047445e-01
6.97300315e-01 -3.69488686e-01 -3.11214384e-02 -4.46216762e-01
-4.64855403e-01 -1.14306971e-01 7.62676775e-01 8.01135600e-01
4.18420851e-01 -7.81679094e-01 8.20508480e-01 5.39971948e-01
-1.22121215e+00 -4.18663561e-01 -1.32293878e-02 4.48959507e-02
3.84901077e-01 1.09661400e+00 -8.73292625e-01 1.21043110e+00
2.39643365e-01 1.98084593e-01 -1.01515567e-02 1.52143121e+00
-2.39068121e-01 5.66826105e-01 -8.94466639e-01 -4.19131756e-01
-2.98403315e-02 -3.07333469e-01 1.06181443e+00 9.25912976e-01
1.28393924e+00 2.45995626e-01 -4.78071839e-01 1.21498883e-01
4.86342281e-01 2.66616106e-01 -3.54085594e-01 5.42383730e-01
7.45038986e-01 1.25044680e+00 -7.98448205e-01 3.32603663e-01
-2.61128485e-01 4.28246826e-01 -3.82267803e-01 6.02291584e-01
-5.00823498e-01 -6.66739762e-01 4.47184771e-01 4.27189052e-01
5.86034894e-01 -8.97405386e-01 -5.76897264e-01 -6.04663849e-01
-3.24264169e-02 -1.40116557e-01 -8.52201283e-02 -2.52818048e-01
-2.01809034e-01 3.25493246e-01 -1.35447547e-01 -1.58325589e+00
-4.27013189e-01 -7.95207843e-02 -2.93200105e-01 6.90222085e-01
-1.44039845e+00 -1.03075230e+00 -3.10688466e-01 8.59258845e-02
-2.63212025e-01 -1.84495226e-01 7.72591233e-01 4.85233396e-01
-4.86343801e-01 5.49131453e-01 9.83468711e-01 -4.45233971e-01
3.23644668e-01 -6.10468328e-01 -4.33540612e-01 1.02390301e+00
-9.36046392e-02 7.60883510e-01 1.16367161e+00 -4.05135602e-01
-1.93549156e+00 -8.39517474e-01 6.71974361e-01 2.03233942e-01
3.65752369e-01 -4.37439978e-01 -3.25181347e-04 1.66783020e-01
3.75655107e-02 -3.17839175e-01 8.10132205e-01 -1.79524869e-01
4.95558828e-01 -4.66037691e-01 -1.15789306e+00 6.40401721e-01
7.52514780e-01 2.44132504e-01 3.82350273e-02 3.74366432e-01
6.44172505e-02 -5.85735500e-01 -4.60260302e-01 5.85715234e-01
7.79299736e-01 -6.31957173e-01 1.18408608e+00 6.55861437e-01
-6.17398798e-01 -5.82773328e-01 -4.76827502e-01 -1.09135234e+00
-5.20400047e-01 -9.04858530e-01 9.47740376e-02 1.15791321e+00
3.55482459e-01 -1.01235950e+00 8.82391751e-01 -4.60213423e-02
-1.45206362e-01 -6.59800529e-01 -1.39478612e+00 -7.44047940e-01
-8.12490284e-01 -3.24457943e-01 2.97495782e-01 2.25330472e-01
-1.62617788e-01 2.13567376e-01 -3.00818920e-01 1.26893997e+00
1.08326399e+00 5.80452494e-02 9.11903799e-01 -1.25346196e+00
-4.01193231e-01 -1.45472726e-02 -5.45500755e-01 -1.25200844e+00
-1.92285359e-01 -4.65794176e-01 3.44212353e-01 -1.75097859e+00
-6.70713723e-01 -9.47466314e-01 -6.59982339e-02 -1.30273730e-01
3.63831550e-01 4.39189464e-01 -4.09479707e-01 -1.03375658e-01
-3.79611701e-01 5.91024518e-01 8.04489434e-01 2.07311794e-01
-3.38673294e-01 9.09349620e-01 -5.90333283e-01 5.89734852e-01
5.87280869e-01 -3.30198616e-01 -3.42764944e-01 -2.10832819e-01
4.74414140e-01 4.13513482e-01 1.97888300e-01 -1.48345482e+00
4.38154787e-01 2.28847668e-01 3.03148031e-01 -5.92696309e-01
5.39595425e-01 -1.18046689e+00 6.36229753e-01 6.72538340e-01
3.50622028e-01 -7.23029017e-01 7.26733655e-02 9.58659470e-01
2.88288444e-01 -4.53321934e-01 7.88084388e-01 5.13560951e-01
-1.29624456e-01 -1.51475638e-01 -7.09708154e-01 -7.26024926e-01
1.25855231e+00 -2.53013581e-01 -8.46205428e-02 -7.34396398e-01
-5.03346801e-01 6.14845268e-02 3.58623043e-02 -1.38255239e-01
2.27781132e-01 -1.07368660e+00 -5.13342261e-01 2.54682332e-01
1.01375198e-02 -2.72966623e-01 4.90337729e-01 1.15237486e+00
-4.73859429e-01 8.46037328e-01 1.85786143e-01 -7.76164591e-01
-1.21845865e+00 -2.40627080e-01 2.56204695e-01 1.48004964e-01
-2.72856981e-01 7.66810715e-01 -1.64268896e-01 -1.29057422e-01
1.60372347e-01 -1.68053191e-02 -2.43019849e-01 4.03105170e-02
3.05916905e-01 4.21340287e-01 6.21386647e-01 -8.07483196e-01
-5.56014538e-01 1.06183028e+00 5.70677221e-01 -2.41280377e-01
1.28904438e+00 -6.56081736e-01 -1.45481870e-01 -8.27922598e-02
9.19682503e-01 8.55513215e-01 -7.31417000e-01 -1.87200606e-01
5.53033128e-02 -5.38275838e-01 1.68666974e-01 -4.82751638e-01
-6.85936809e-01 2.71565318e-01 7.59303689e-01 1.85185075e-01
1.00525916e+00 -1.36809021e-01 3.29317659e-01 6.69491291e-01
1.18790793e+00 -6.64201260e-01 -2.73136944e-01 2.79060781e-01
7.12634623e-01 -4.68691915e-01 3.02374899e-01 -6.38551950e-01
3.24732929e-01 1.15769792e+00 1.92657541e-02 3.68784070e-02
9.25777674e-01 3.48052830e-01 -3.25265937e-02 7.17806593e-02
1.72440842e-01 -1.87509328e-01 1.89749196e-01 5.84229529e-01
1.56101897e-01 5.20743430e-02 -6.24622166e-01 7.56626189e-01
-1.83921561e-01 -4.98604834e-01 4.72228348e-01 1.06924212e+00
-8.65441740e-01 -1.15638638e+00 -8.53827596e-01 3.28089774e-01
-2.88556010e-01 3.51941139e-01 4.35486853e-01 5.20374298e-01
-4.63685133e-02 1.20899820e+00 -2.10887760e-01 -1.40743569e-01
5.98194659e-01 -4.82424229e-01 7.04058945e-01 -3.07847768e-01
2.57420875e-02 3.28365117e-01 2.13459074e-01 -3.80980611e-01
-3.19688976e-01 -7.70980418e-01 -1.12109423e+00 4.03519660e-01
-7.06712306e-01 5.81065714e-01 1.18611145e+00 1.09162116e+00
6.14978790e-01 5.47576785e-01 1.01911986e+00 -1.05195284e+00
-2.79115438e-01 -5.42774200e-01 -6.90644324e-01 -7.09580421e-01
4.39137876e-01 -7.92353749e-01 -4.01074260e-01 -7.35803127e-01]
|
[6.286422252655029, 1.3048655986785889]
|
07b63485-f8e2-447e-912a-5ffb48345a23
|
cross-domain-deep-feature-combination-for
|
1811.10199
| null |
http://arxiv.org/abs/1811.10199v1
|
http://arxiv.org/pdf/1811.10199v1.pdf
|
Cross-domain Deep Feature Combination for Bird Species Classification with Audio-visual Data
|
In recent decade, many state-of-the-art algorithms on image classification as
well as audio classification have achieved noticeable successes with the
development of deep convolutional neural network (CNN). However, most of the
works only exploit single type of training data. In this paper, we present a
study on classifying bird species by exploiting the combination of both visual
(images) and audio (sounds) data using CNN, which has been sparsely treated so
far. Specifically, we propose CNN-based multimodal learning models in three
types of fusion strategies (early, middle, late) to settle the issues of
combining training data cross domains. The advantage of our proposed method
lies on the fact that We can utilize CNN not only to extract features from
image and audio data (spectrogram) but also to combine the features across
modalities. In the experiment, we train and evaluate the network structure on a
comprehensive CUB-200-2011 standard data set combing our originally collected
audio data set with respect to the data species. We observe that a model which
utilizes the combination of both data outperforms models trained with only an
either type of data. We also show that transfer learning can significantly
increase the classification performance.
|
['Takuya Akashi', 'Chao Zhang', 'Bold Naranchimeg']
|
2018-11-26
| null | null | null | null |
['bird-species-classification-with-audio-visual']
|
['audio']
|
[ 1.34304062e-01 -5.76767147e-01 -4.05484699e-02 -2.50329792e-01
-5.01399577e-01 -5.68101108e-01 7.14294732e-01 2.70829409e-01
-7.88183093e-01 4.42899883e-01 1.41774878e-01 9.32416543e-02
-8.64898786e-02 -7.46279418e-01 -6.14636958e-01 -5.71177721e-01
-2.58892715e-01 -1.15129113e-01 2.02217132e-01 -3.56151640e-01
5.54524362e-02 3.03847998e-01 -2.34422231e+00 5.63303530e-01
5.44905663e-01 1.42427671e+00 -3.87764461e-02 6.55642867e-01
-7.35921711e-02 6.44637108e-01 -5.82401097e-01 -2.54779518e-01
3.23998302e-01 -2.85580248e-01 -7.05566764e-01 -3.39459181e-02
6.24091744e-01 -2.21680671e-01 -2.44767264e-01 9.32302117e-01
7.24811256e-01 -1.03988357e-01 5.31037152e-01 -1.51666749e+00
-3.56207967e-01 6.44208431e-01 -5.54871738e-01 1.92174941e-01
1.68864876e-01 1.36132911e-01 9.98410106e-01 -6.30247593e-01
2.88228601e-01 9.22195435e-01 7.67119348e-01 3.91997516e-01
-1.07140076e+00 -8.44574869e-01 -1.10628211e-03 5.95900297e-01
-1.52140701e+00 -3.19329709e-01 1.10263741e+00 -7.31222510e-01
7.33163655e-01 7.79436231e-02 7.83704698e-01 9.86696005e-01
-2.44321525e-01 7.70737767e-01 1.10010886e+00 -5.12487829e-01
1.25288263e-01 1.92292124e-01 4.87727206e-03 5.51379144e-01
-1.53187305e-01 2.67584413e-01 -8.41575682e-01 -9.03908983e-02
3.93135041e-01 -1.57756999e-01 -1.95861012e-01 -2.20500544e-01
-1.06225181e+00 8.26158345e-01 6.94171071e-01 6.19774580e-01
-3.00072312e-01 3.53766568e-02 6.42986119e-01 4.55413818e-01
5.04164636e-01 3.27333391e-01 -4.01606143e-01 -1.32387489e-01
-1.10257161e+00 1.47926643e-01 6.08218551e-01 4.74736094e-01
7.81892776e-01 2.24469870e-01 4.97068912e-02 1.15823364e+00
1.07107311e-01 3.22710067e-01 5.61040044e-01 -5.30778050e-01
3.14359516e-01 5.09753585e-01 -3.97333890e-01 -9.61131155e-01
-4.69021708e-01 -6.10393345e-01 -9.72507477e-01 1.22668296e-01
4.57070887e-01 -1.62085086e-01 -8.95829022e-01 1.92651081e+00
1.71320349e-01 2.71724969e-01 8.94696862e-02 8.42649043e-01
1.37810242e+00 4.63872671e-01 1.29874870e-01 1.57042518e-01
1.26181686e+00 -7.85675704e-01 -4.04301345e-01 2.14469939e-01
4.50392842e-01 -7.02419758e-01 6.27416670e-01 5.78258216e-01
-7.80885756e-01 -9.35467482e-01 -1.20883560e+00 1.25272810e-01
-8.46713543e-01 5.60879052e-01 5.75412512e-01 5.74248135e-01
-1.15908146e+00 3.98013413e-01 -4.37100857e-01 -5.67141235e-01
4.12255764e-01 4.93930072e-01 -6.12806916e-01 2.63725817e-01
-1.25831175e+00 6.41080737e-01 4.83866930e-01 1.53534085e-01
-1.07498312e+00 -5.54484844e-01 -7.27671027e-01 1.03037253e-01
2.24545047e-01 -4.32353407e-01 1.06823206e+00 -1.42262256e+00
-1.38580573e+00 7.71640897e-01 4.08017695e-01 -6.27223849e-01
2.42722929e-01 -4.94900160e-02 -4.26252365e-01 2.02485889e-01
-2.31416076e-01 1.15859985e+00 7.54979730e-01 -1.22830510e+00
-9.69716489e-01 -2.92248398e-01 3.09970796e-01 -2.04995006e-01
-7.65918016e-01 -7.32804462e-02 4.03607823e-02 -5.19322038e-01
-3.51135761e-01 -8.18296313e-01 1.97100878e-01 6.87059239e-02
-1.98541999e-01 -1.96382508e-01 7.06368566e-01 -4.67798144e-01
1.10103822e+00 -2.36879182e+00 3.92473876e-01 6.92465110e-03
7.52268657e-02 5.15812874e-01 -4.28251177e-01 5.66517949e-01
-1.90595984e-01 1.06790856e-01 -3.53901178e-01 -4.58250552e-01
-7.73116872e-02 1.73021361e-01 -2.43610978e-01 2.83926606e-01
2.70215839e-01 4.55229104e-01 -7.15519071e-01 -3.72467816e-01
2.40252316e-01 5.87672412e-01 -6.14920259e-01 1.84872270e-01
3.05249020e-02 3.99283439e-01 5.11012040e-02 6.22753322e-01
7.18005896e-01 2.48934433e-01 3.31827924e-02 -4.21371579e-01
-3.68513107e-01 -5.04511148e-02 -1.03078580e+00 1.81963551e+00
-4.54895198e-01 7.74457216e-01 1.80215344e-01 -1.23323882e+00
7.17987955e-01 4.33675289e-01 6.58298433e-01 -5.22199571e-01
2.38873631e-01 2.76971787e-01 3.19380492e-01 -5.62128127e-01
4.11161244e-01 -1.03485644e-01 1.21027799e-02 1.88047886e-01
7.92532682e-01 3.76157798e-02 3.23450953e-01 -1.73045531e-01
7.56555140e-01 -1.49749443e-01 1.62063241e-01 -1.26585569e-02
6.78160965e-01 -1.12484552e-01 2.41322085e-01 6.23775780e-01
-2.18688652e-01 4.55450505e-01 2.76400775e-01 -4.05104756e-01
-9.03843582e-01 -7.82071888e-01 -2.24820599e-01 1.21778810e+00
-5.79742976e-02 -5.05209208e-01 -6.01607084e-01 -5.28200805e-01
2.39085909e-02 5.71670383e-02 -8.42174709e-01 -1.29726604e-01
-1.50839344e-01 -7.34026551e-01 1.07088971e+00 5.83004594e-01
8.41180444e-01 -1.02649558e+00 -7.34940350e-01 -1.28227308e-01
-1.16432220e-01 -1.03011930e+00 2.08381861e-02 3.40871960e-01
-5.28656840e-01 -1.01446795e+00 -5.34787655e-01 -7.53213525e-01
8.94493684e-02 3.08112532e-01 8.02668095e-01 8.92214626e-02
-3.47659707e-01 6.74606502e-01 -7.06403732e-01 -5.19713104e-01
-1.14529073e-01 3.73301029e-01 -1.18871152e-01 4.80334193e-01
2.70092964e-01 -8.67303669e-01 -2.55758256e-01 3.56807075e-02
-1.18988252e+00 8.22517350e-02 5.24721086e-01 8.15880895e-01
1.40812606e-01 -7.02042058e-02 6.50102913e-01 -1.98167548e-01
4.38685000e-01 -5.31435847e-01 -3.88492823e-01 1.60305843e-01
1.35720856e-02 -3.14917117e-02 5.31373024e-01 -6.22520030e-01
-7.13857949e-01 2.84392744e-01 -2.61227250e-01 -4.71326441e-01
-6.14167929e-01 9.69686329e-01 1.07525706e-01 -2.11530715e-01
5.27431726e-01 1.35692984e-01 5.17781861e-02 -6.23642981e-01
2.52047449e-01 1.03922784e+00 4.33505237e-01 -5.41562617e-01
5.84804118e-01 5.47528088e-01 -6.36029756e-03 -1.11078668e+00
-6.76486194e-01 -5.13055384e-01 -7.50824273e-01 -5.23753822e-01
1.00151873e+00 -1.00781369e+00 -7.60973632e-01 7.16459513e-01
-1.03997004e+00 2.21035406e-02 -1.06294297e-01 7.31603980e-01
-2.37152994e-01 1.71756595e-01 -4.06468481e-01 -8.67716253e-01
-5.22808693e-02 -1.10108900e+00 1.06837165e+00 7.87279010e-02
1.63135305e-01 -8.01311016e-01 2.88067728e-01 8.02658126e-02
5.40308237e-01 2.48068646e-01 7.34259427e-01 -7.34396219e-01
-2.81927645e-01 -1.68373615e-01 -1.80875063e-01 6.12457573e-01
9.10697356e-02 1.54764280e-01 -1.45069206e+00 -2.12023050e-01
-1.34391159e-01 -5.93843699e-01 1.34278774e+00 2.42817253e-01
1.24932575e+00 9.79439616e-02 1.16509162e-01 5.17876983e-01
1.32056236e+00 1.42147169e-01 4.33213353e-01 2.71880060e-01
6.11605287e-01 6.78205788e-01 2.30712727e-01 4.97426212e-01
3.20021957e-01 8.16391051e-01 7.57849276e-01 -1.77948505e-01
-1.27327174e-01 -8.14372301e-02 3.67922068e-01 8.99229825e-01
-3.90594155e-01 -5.27048223e-02 -8.36884081e-01 6.96902096e-01
-1.81376874e+00 -1.11522245e+00 1.73866943e-01 2.09357905e+00
6.89056754e-01 -2.75860935e-01 5.44850886e-01 4.89729673e-01
6.75876617e-01 7.31197372e-02 -8.17032903e-03 -1.40308157e-01
-1.93874300e-01 3.52379948e-01 2.36985058e-01 1.74456283e-01
-1.49943078e+00 5.25679708e-01 6.11620951e+00 9.86485243e-01
-1.64679623e+00 6.08269684e-02 1.56328395e-01 -2.01431975e-01
1.22740090e-01 -2.73753762e-01 -5.18559098e-01 3.24158430e-01
8.10559928e-01 3.12745214e-01 5.32571137e-01 6.73649609e-01
-2.97162056e-01 -1.32774681e-01 -1.12723529e+00 1.21216476e+00
2.36712828e-01 -1.05492306e+00 1.23651117e-01 2.91475952e-02
3.75556409e-01 6.13952391e-02 1.94557056e-01 4.03891146e-01
-2.22695038e-01 -9.53950644e-01 9.94963825e-01 4.14265603e-01
5.15922010e-01 -7.32818365e-01 9.51204300e-01 3.19480538e-01
-1.48042786e+00 -3.14785659e-01 -1.33549049e-01 -2.36387372e-01
-2.39364728e-01 2.14800492e-01 -5.39933205e-01 9.77023005e-01
9.92784798e-01 1.02683103e+00 -9.27700102e-01 1.36578214e+00
1.03283279e-01 7.77702153e-01 -4.91540700e-01 1.38307353e-02
2.89344341e-01 1.43711537e-01 4.59736049e-01 1.25842428e+00
4.78399813e-01 -3.51707131e-01 3.25214714e-01 7.49086797e-01
4.20175493e-02 1.66194394e-01 -6.34851813e-01 -2.08416685e-01
2.10905343e-01 1.38482869e+00 -5.27772427e-01 -3.15455556e-01
-5.21717906e-01 2.82899618e-01 2.89044380e-01 1.60495490e-01
-8.40870678e-01 -3.40817153e-01 4.83823270e-01 -1.43407136e-01
5.33396006e-01 -2.32183769e-01 -7.17736706e-02 -1.12068617e+00
-1.29460886e-01 -6.66913152e-01 3.67474556e-01 -6.91830575e-01
-1.44551969e+00 8.34698796e-01 3.37407857e-01 -1.59177947e+00
-1.62200108e-01 -7.27773488e-01 -4.59983170e-01 5.88171661e-01
-1.59910429e+00 -1.44656730e+00 -4.39623475e-01 7.21228421e-01
2.68597364e-01 -4.89607304e-01 7.34956443e-01 7.81633973e-01
-3.85917127e-01 5.15013874e-01 -1.70575544e-01 3.95434648e-01
7.36834407e-01 -9.69564378e-01 -4.64709669e-01 5.14011562e-01
4.22810525e-01 3.41307074e-01 4.35712606e-01 -4.83845174e-02
-1.00828481e+00 -8.73528838e-01 5.10735035e-01 8.10859576e-02
7.02109337e-01 -3.24343622e-01 -7.00298488e-01 2.45877177e-01
5.75160146e-01 2.58345027e-02 9.67301011e-01 1.77750960e-01
-6.94718599e-01 -4.50203925e-01 -1.00732005e+00 1.20197095e-01
6.27942443e-01 -8.12205553e-01 -4.53978270e-01 -1.93133220e-01
4.06730741e-01 -7.22858459e-02 -7.91736066e-01 6.90894783e-01
7.44513750e-01 -1.15291882e+00 8.07387531e-01 -5.64262211e-01
6.58539534e-01 -5.17457902e-01 -4.77599651e-01 -1.44950294e+00
8.52944329e-02 2.09339652e-02 2.80524105e-01 1.45129228e+00
2.36886352e-01 -4.27699029e-01 2.50570029e-01 -2.41935402e-01
-2.65839934e-01 -4.12195593e-01 -1.11734867e+00 -6.96113884e-01
1.10977897e-02 -5.56178033e-01 5.47355354e-01 9.54528153e-01
4.36913893e-02 3.63816768e-01 -5.38422763e-01 1.14163496e-02
2.48314992e-01 2.60481298e-01 8.27034712e-01 -1.42513132e+00
-2.83687890e-01 -5.45639455e-01 -8.05321515e-01 -5.20491004e-01
1.64199293e-01 -9.77343440e-01 -8.88227893e-04 -1.17248070e+00
3.46092433e-01 -9.83176902e-02 -5.47223330e-01 8.43143582e-01
2.43356749e-01 7.11936414e-01 4.60950255e-01 -4.61785831e-02
-5.36772132e-01 5.76397717e-01 8.62778902e-01 -3.55068117e-01
-6.41632825e-02 -2.34843493e-01 -3.94741118e-01 6.67201281e-01
6.97340846e-01 -3.63695562e-01 -1.81258798e-01 -4.57138956e-01
1.75161704e-01 -1.56892851e-01 8.03948641e-01 -1.38970995e+00
2.62342304e-01 3.27017665e-01 3.29268396e-01 -6.16858661e-01
6.01851761e-01 -9.47970271e-01 1.72646940e-01 3.11336398e-01
-5.57136953e-01 -1.84587941e-01 5.35234153e-01 3.66592139e-01
-8.28254819e-01 -1.91415802e-01 7.86152422e-01 1.21745870e-01
-7.44076133e-01 8.70252177e-02 -4.71830994e-01 -3.67554575e-01
7.42224574e-01 -3.00348196e-02 -2.70921081e-01 -4.16902602e-01
-8.82931888e-01 -3.86267863e-02 9.89308581e-02 6.06992722e-01
3.74604613e-01 -1.48678243e+00 -6.54908419e-01 2.04981014e-01
2.75204539e-01 -5.14184475e-01 4.63904887e-01 1.01290643e+00
-2.97053933e-01 4.37914252e-01 -7.61734843e-01 -8.78163218e-01
-1.49115336e+00 6.24139369e-01 3.71905893e-01 3.86951468e-03
1.10681608e-01 6.37504935e-01 7.39122182e-02 -4.83798474e-01
4.73604649e-01 -2.56612599e-01 -5.69036365e-01 5.36702216e-01
4.17325497e-01 1.71627417e-01 1.36159807e-01 -9.05454099e-01
-3.37104917e-01 6.46804988e-01 2.44278267e-01 -2.22028807e-01
1.52784586e+00 3.03568244e-01 -1.68792188e-01 6.96697056e-01
1.23167002e+00 -2.00570583e-01 -7.91660190e-01 -2.26347312e-01
-3.22458953e-01 -3.97331268e-01 1.82377994e-01 -7.70070910e-01
-1.35736132e+00 1.42004085e+00 9.26055968e-01 6.59580112e-01
1.38351703e+00 -1.55674011e-01 4.36101586e-01 1.81561485e-01
2.38108516e-01 -8.96963537e-01 5.07539250e-02 4.63399112e-01
8.55778158e-01 -1.19562495e+00 -2.61644393e-01 -1.62818044e-01
-5.35084426e-01 1.28216326e+00 5.26315033e-01 -1.02548100e-01
9.18115973e-01 1.26886711e-01 -4.56067175e-02 -4.34241444e-02
-8.16365659e-01 -7.96636224e-01 5.76917470e-01 5.70613146e-01
5.19160390e-01 -6.85503036e-02 -1.46028876e-01 6.98363006e-01
-1.37484521e-01 4.61862572e-02 3.03598583e-01 9.60198462e-01
-4.51823473e-01 -1.20692241e+00 -3.31012249e-01 2.11151510e-01
-4.59629267e-01 -2.57969685e-02 -7.21661627e-01 8.59905660e-01
7.31446564e-01 1.03951716e+00 1.52434498e-01 -7.86035001e-01
1.76105782e-01 -5.60529865e-02 6.54096484e-01 -3.80149901e-01
-9.62517798e-01 2.13242635e-01 -1.56787261e-02 -1.92289785e-01
-1.19953036e+00 -4.59280938e-01 -6.03415787e-01 -1.41534597e-01
-3.52283895e-01 1.48470342e-01 8.16627264e-01 8.76108468e-01
2.50761002e-01 5.97196817e-01 4.88477975e-01 -1.31127775e+00
-3.04169327e-01 -1.18112874e+00 -6.64099038e-01 2.59953499e-01
5.43613553e-01 -9.61408317e-01 -4.48726326e-01 1.15190670e-01]
|
[15.15077018737793, 5.092846393585205]
|
74ddce97-d48c-419b-bf2c-9440c98a4c0c
|
disconnected-emerging-knowledge-graph
|
2209.01397
| null |
https://arxiv.org/abs/2209.01397v1
|
https://arxiv.org/pdf/2209.01397v1.pdf
|
Disconnected Emerging Knowledge Graph Oriented Inductive Link Prediction
|
Inductive link prediction (ILP) is to predict links for unseen entities in emerging knowledge graphs (KGs), considering the evolving nature of KGs. A more challenging scenario is that emerging KGs consist of only unseen entities, called as disconnected emerging KGs (DEKGs). Existing studies for DEKGs only focus on predicting enclosing links, i.e., predicting links inside the emerging KG. The bridging links, which carry the evolutionary information from the original KG to DEKG, have not been investigated by previous work so far. To fill in the gap, we propose a novel model entitled DEKG-ILP (Disconnected Emerging Knowledge Graph Oriented Inductive Link Prediction) that consists of the following two components. (1) The module CLRM (Contrastive Learning-based Relation-specific Feature Modeling) is developed to extract global relation-based semantic features that are shared between original KGs and DEKGs with a novel sampling strategy. (2) The module GSM (GNN-based Subgraph Modeling) is proposed to extract the local subgraph topological information around each link in KGs. The extensive experiments conducted on several benchmark datasets demonstrate that DEKG-ILP has obvious performance improvements compared with state-of-the-art methods for both enclosing and bridging link prediction. The source code is available online.
|
['Lei Zhao', 'Wei Chen', 'Pengpeng Zhao', 'Hongzhi Yin', 'Weiqing Wang', 'Yufeng Zhang']
|
2022-09-03
| null | null | null | null |
['inductive-link-prediction']
|
['graphs']
|
[-2.18185648e-01 7.02683985e-01 -5.81107020e-01 5.94909079e-02
7.84042701e-02 -3.24132472e-01 4.40015405e-01 4.97609168e-01
4.50215518e-01 9.29304481e-01 -6.28756657e-02 -2.57122189e-01
-6.56258345e-01 -1.47918952e+00 -7.59937942e-01 -4.02853042e-01
-7.26997495e-01 4.85857576e-01 8.53755355e-01 -4.18491423e-01
-1.20229967e-01 2.67104268e-01 -1.36744404e+00 8.81786421e-02
1.48742199e+00 7.64880478e-01 5.60090132e-02 6.93165436e-02
-3.34359527e-01 8.60838175e-01 -2.96874661e-02 -7.23188102e-01
4.37682308e-02 -2.03518748e-01 -1.07922256e+00 -3.39414597e-01
1.36915758e-01 2.61863440e-01 -5.27108371e-01 8.92034531e-01
2.15785459e-01 -2.87709475e-01 4.42853421e-01 -1.63451517e+00
-5.00373423e-01 1.01487255e+00 -5.52415311e-01 2.11099252e-01
4.33305770e-01 -3.59580010e-01 1.32041013e+00 -1.04661083e+00
1.27915251e+00 1.03907025e+00 8.29426885e-01 -1.30332664e-01
-8.74980748e-01 -7.16257453e-01 4.30394590e-01 6.98671281e-01
-1.80407381e+00 -5.01398593e-02 9.79962587e-01 -4.43816513e-01
7.38703907e-01 9.44130197e-02 1.05505431e+00 7.14252234e-01
2.63572961e-01 6.77240372e-01 5.88374555e-01 -2.16023296e-01
-7.78956115e-02 1.55550435e-01 2.47300938e-01 9.61178243e-01
7.40482926e-01 -1.21528439e-01 -7.78168082e-01 -6.95839375e-02
4.62909192e-01 -4.05301183e-01 -4.20856595e-01 -9.04546618e-01
-1.06883490e+00 6.24246299e-01 8.36449683e-01 3.28387380e-01
-2.90946335e-01 -2.50661761e-01 2.78307766e-01 4.22218740e-01
7.77928293e-01 -2.32700258e-02 -8.14984500e-01 3.13154817e-01
-6.01736248e-01 5.01668714e-02 1.01930022e+00 1.20951164e+00
9.90008473e-01 -3.02596033e-01 1.51222169e-01 6.02970719e-01
4.86218452e-01 4.58095707e-02 7.79407099e-02 1.74733959e-02
7.52682269e-01 1.34143567e+00 -2.34258145e-01 -1.61214125e+00
-6.86146200e-01 -1.02666354e+00 -6.88693941e-01 -5.95537126e-01
-8.59645382e-02 5.88979051e-02 -4.80565131e-01 1.56300735e+00
8.40380728e-01 5.85487306e-01 2.23600328e-01 4.26596493e-01
1.49868023e+00 4.85923916e-01 1.84347704e-01 -4.41172779e-01
9.96483743e-01 -9.71206725e-01 -5.14564931e-01 3.95883922e-04
1.04157662e+00 -3.37940276e-01 2.75640339e-01 -1.35396034e-01
-7.59031177e-01 -4.87835526e-01 -1.02905285e+00 3.22245419e-01
-9.02621090e-01 -6.67597353e-02 8.53002191e-01 8.64931941e-02
-1.03966439e+00 4.81270760e-01 -4.06191319e-01 -7.93914318e-01
3.00374389e-01 3.41784537e-01 -5.34119129e-01 5.48463129e-02
-1.68541920e+00 6.37954175e-01 1.29719603e+00 7.18729496e-02
-2.96221137e-01 -1.05714869e+00 -8.86048496e-01 5.60808145e-02
8.79251480e-01 -9.45186734e-01 2.54805118e-01 -5.42866528e-01
-7.98726559e-01 7.30732262e-01 -2.68670563e-02 -1.55036882e-01
3.44926625e-01 5.12704514e-02 -1.18686712e+00 1.67877734e-01
2.20189020e-01 4.78015870e-01 4.81000960e-01 -1.55144000e+00
-7.26966321e-01 -4.04933512e-01 -1.59653634e-01 3.10122341e-01
-6.73638821e-01 -6.41038537e-01 -5.81502438e-01 -7.54707694e-01
5.65827727e-01 -7.09986866e-01 3.67618233e-01 -5.16853392e-01
-8.80405188e-01 -5.42422354e-01 1.09313583e+00 -6.66681170e-01
1.85456765e+00 -1.69815779e+00 3.44408244e-01 6.84376299e-01
5.81285834e-01 3.51225972e-01 1.04335174e-02 9.28988218e-01
-2.96305984e-01 4.32994626e-02 2.70777106e-01 3.99928868e-01
-3.36188406e-01 1.31855570e-02 -1.50825694e-01 -1.36398748e-02
-6.05577826e-02 1.40018630e+00 -1.30758452e+00 -9.48213816e-01
-1.09440513e-01 -9.83330682e-02 -8.05798918e-02 -1.24759801e-01
-1.31571814e-01 3.71423900e-01 -5.26738882e-01 1.03843582e+00
8.57594311e-01 -4.39564109e-01 6.67777061e-01 -5.74498236e-01
1.00878783e-01 -4.76086326e-03 -1.35376859e+00 1.48246408e+00
1.25439689e-01 4.58460823e-02 -2.74238616e-01 -1.06775081e+00
1.07012558e+00 -1.24055343e-02 6.01882517e-01 -3.46113324e-01
-2.61369586e-01 4.23244059e-01 -3.07022464e-02 -5.27990937e-01
5.05715489e-01 3.60456139e-01 4.83656004e-02 -1.60484090e-01
3.33171427e-01 5.80464602e-01 4.37805295e-01 8.66681933e-01
1.35610092e+00 2.08505139e-01 3.96623909e-01 -5.05440295e-01
7.00549722e-01 1.12675332e-01 8.64578247e-01 3.02511126e-01
1.25090599e-01 -1.84831619e-01 5.34482181e-01 -3.94303948e-01
-4.32750225e-01 -1.38978922e+00 -3.53350714e-02 5.28389633e-01
7.48930335e-01 -8.93343627e-01 -2.85040885e-01 -1.07808244e+00
5.64810753e-01 4.03275907e-01 -4.11698550e-01 -5.03846228e-01
-3.48433226e-01 -6.71404719e-01 2.81572580e-01 4.92414534e-01
5.70158958e-01 -8.67073894e-01 3.33956748e-01 3.10188621e-01
-2.42075622e-01 -1.25263774e+00 -9.36656352e-03 -7.86772296e-02
-9.71530795e-01 -1.65143418e+00 -8.07069764e-02 -1.09014606e+00
8.59441698e-01 1.70976520e-01 1.26435125e+00 2.73235351e-01
-1.77352294e-01 5.45888841e-01 -6.35033786e-01 -1.28339082e-01
-2.82567088e-02 5.79905510e-01 2.51004118e-02 6.85121268e-02
3.15020114e-01 -8.13717902e-01 -5.31697392e-01 4.44075823e-01
-4.47677791e-01 4.19781625e-01 7.61101186e-01 6.76033258e-01
5.89547753e-01 6.76625252e-01 1.03506625e+00 -1.24897718e+00
4.16096240e-01 -1.13352108e+00 -2.13237122e-01 6.40115321e-01
-9.83353734e-01 -2.04452172e-01 5.07030547e-01 -1.85892910e-01
-1.03195989e+00 -3.83884698e-01 2.04426140e-01 -4.35244069e-02
2.14639544e-01 1.09822917e+00 -3.43638539e-01 -2.63994664e-01
3.89586948e-02 1.92176223e-01 -4.83800352e-01 -4.37737018e-01
3.11662227e-01 3.85000050e-01 4.01294857e-01 -5.19086659e-01
1.27657306e+00 3.13774705e-01 3.33241999e-01 -6.04717016e-01
-6.47987008e-01 -4.61510301e-01 -8.66921484e-01 -5.63540220e-01
3.00366849e-01 -9.88026500e-01 -4.92010653e-01 4.45090294e-01
-7.57596433e-01 1.32926300e-01 -2.29798913e-01 3.67324114e-01
-1.31249517e-01 3.68604392e-01 -5.34561336e-01 -5.42497516e-01
-4.00261283e-01 -2.58629888e-01 6.83975160e-01 2.93514252e-01
-1.13932818e-01 -1.24320793e+00 5.75114042e-02 3.70253444e-01
-8.62590894e-02 6.00719690e-01 1.33431065e+00 -7.84770191e-01
-8.57957363e-01 -1.74881436e-03 -2.81824231e-01 -2.78833508e-01
3.27035338e-01 -5.49064986e-02 -2.22798392e-01 -3.54475409e-01
-9.84030724e-01 2.74177361e-02 8.26886594e-01 -1.57355189e-01
7.42937505e-01 -2.54111946e-01 -1.21969187e+00 4.58484024e-01
1.54601371e+00 1.40929997e-01 6.18967652e-01 4.33902532e-01
1.02980173e+00 6.89543128e-01 1.02054095e+00 5.74806593e-02
9.23685670e-01 6.60386860e-01 3.95502508e-01 -2.50640083e-02
-1.23300068e-01 -8.42068553e-01 4.75859903e-02 1.37036836e+00
-2.05985576e-01 -3.70573193e-01 -9.00221646e-01 7.34618604e-01
-2.33462667e+00 -6.31106913e-01 -6.78164423e-01 1.81885731e+00
6.25021160e-01 2.97804654e-01 -8.34547654e-02 1.42756760e-01
1.08851755e+00 5.48071414e-02 -7.80911744e-01 1.34792820e-01
-2.03180790e-01 -1.91146377e-02 1.59490585e-01 2.48778164e-01
-1.04262269e+00 1.19022548e+00 4.43375683e+00 1.01573110e+00
-3.97115439e-01 -1.07832747e-02 2.23561063e-01 4.11622822e-01
-5.02814353e-01 3.98082733e-01 -9.03063118e-01 6.18700087e-01
4.78058130e-01 -3.71845037e-01 -1.20220613e-03 7.73283064e-01
-1.75541475e-01 -1.72647536e-02 -7.35144794e-01 6.67991281e-01
-1.65260702e-01 -1.26846254e+00 3.54782581e-01 6.94987848e-02
8.80971491e-01 -1.86120436e-01 -3.79779220e-01 4.01952773e-01
2.88041562e-01 -4.39557761e-01 3.85513484e-01 9.37425256e-01
5.77620924e-01 -8.93324196e-01 8.60308230e-01 2.64748394e-01
-2.15975308e+00 -4.91633192e-02 -1.42183587e-01 3.56272966e-01
9.99605507e-02 1.07369554e+00 -7.66881168e-01 1.88088441e+00
8.17178190e-01 1.20822001e+00 -8.37952256e-01 9.50058162e-01
-3.61944437e-01 5.84532201e-01 -1.15236402e-01 2.78377265e-01
-8.68425369e-02 -3.06076318e-01 8.06667566e-01 8.93461645e-01
4.71383989e-01 -3.40700336e-03 2.74246484e-01 6.62406087e-01
-3.13416213e-01 4.72070068e-01 -6.41379297e-01 1.59246296e-01
8.50067675e-01 1.48970807e+00 -9.39505994e-01 -2.61426829e-02
-4.51579869e-01 6.57181084e-01 5.68032265e-01 4.15242195e-01
-7.38153338e-01 -4.66624230e-01 3.17278415e-01 5.30577064e-01
5.46382427e-01 1.11689344e-01 3.24945927e-01 -1.24075663e+00
1.97653681e-01 -1.76842973e-01 9.27942932e-01 -7.34229863e-01
-1.48549056e+00 3.62859488e-01 1.55819550e-01 -1.21244383e+00
2.83113718e-01 -8.47076476e-02 -7.14711547e-01 3.62263232e-01
-1.65872359e+00 -1.66963851e+00 -5.49759090e-01 5.21219075e-01
-1.60875352e-05 -2.11063921e-01 3.46355319e-01 3.56296539e-01
-5.76776624e-01 5.19647896e-01 6.32824227e-02 1.74332969e-02
3.84177148e-01 -1.12495792e+00 3.53104286e-02 6.81867063e-01
8.85327682e-02 5.31618357e-01 2.49506876e-01 -1.38764906e+00
-1.37597048e+00 -1.53135717e+00 1.06207645e+00 -9.83322710e-02
9.08665657e-01 -1.95068464e-01 -9.93417203e-01 8.25991690e-01
-2.46453762e-01 3.27639282e-01 6.68548882e-01 4.63766634e-01
-1.88992321e-01 -4.53919500e-01 -9.23813701e-01 3.63137573e-01
1.90047324e+00 -1.23377457e-01 -5.14051437e-01 1.67645723e-01
7.96082437e-01 -2.57896632e-01 -1.47974575e+00 1.15736473e+00
4.27282006e-01 -6.79394722e-01 1.18086970e+00 -4.61335748e-01
6.93645850e-02 -6.97916150e-01 3.14067781e-01 -1.37190187e+00
-4.54396218e-01 -2.76510656e-01 -1.06448710e+00 1.76448965e+00
3.06290597e-01 -9.88004923e-01 8.02963674e-01 -2.07216695e-01
-7.91891590e-02 -1.29662037e+00 -6.72513723e-01 -1.04591155e+00
-4.29645211e-01 6.99617714e-02 7.01311827e-01 1.35148013e+00
2.74176687e-01 5.31959116e-01 -1.08140513e-01 4.95719045e-01
6.97851658e-01 4.23395544e-01 6.10175550e-01 -1.90960264e+00
9.83125344e-02 -7.24590495e-02 -9.46053267e-01 -7.46568799e-01
3.21782023e-01 -1.49723268e+00 -6.66287780e-01 -1.91303146e+00
2.85794288e-01 -7.34829247e-01 -3.95974040e-01 4.75166798e-01
-3.93505096e-01 -7.82287642e-02 -1.78374454e-01 3.16207200e-01
-1.03654075e+00 7.39091635e-01 1.06229973e+00 -2.24660531e-01
-3.36616933e-01 -2.60075182e-01 -6.90216243e-01 8.04452002e-01
5.48085809e-01 -2.53758788e-01 -8.87110651e-01 2.19314039e-01
6.77360535e-01 -1.16183795e-01 4.27191317e-01 -9.76021051e-01
4.48482811e-01 8.11752230e-02 3.46509129e-01 -1.06472206e+00
-2.75671482e-01 -8.22970986e-01 8.21411133e-01 5.99271297e-01
1.53758347e-01 -3.22404414e-01 -2.50266362e-02 1.03480768e+00
-3.24516833e-01 5.23750335e-02 1.52007118e-01 7.60997012e-02
-1.23925829e+00 5.31037986e-01 3.49749625e-01 1.18348010e-01
1.57318425e+00 -3.25509340e-01 -6.51419997e-01 1.98080074e-02
-9.88565505e-01 7.86130309e-01 1.43560573e-01 5.93100548e-01
6.75485313e-01 -1.55965114e+00 -6.68855190e-01 -1.35752663e-01
5.63329518e-01 1.17037602e-01 3.32325369e-01 9.15976346e-01
-3.04753453e-01 3.78498226e-01 3.01724762e-01 -2.83648014e-01
-1.30836105e+00 6.26220345e-01 -5.49632050e-02 -7.59769559e-01
-8.29128981e-01 7.66601622e-01 -5.26752137e-02 -4.91076469e-01
-9.94138494e-02 2.86081582e-01 -5.40646553e-01 3.59663844e-01
-2.45501816e-01 6.37996972e-01 9.77356657e-02 -7.37680256e-01
-6.13840461e-01 6.42639220e-01 -1.85781315e-01 9.26841259e-01
1.44904733e+00 -2.58041054e-01 -5.12883365e-01 4.11079526e-01
9.02860522e-01 1.02674931e-01 -6.07965112e-01 -4.41910058e-01
5.16073167e-01 -2.29220688e-01 -3.12504292e-01 -7.46873379e-01
-1.30226135e+00 5.62336929e-02 1.86977744e-01 3.39263231e-01
1.14845908e+00 4.94293392e-01 1.05797374e+00 1.47300959e-01
9.04719532e-01 -1.15308785e+00 -2.83228718e-02 1.99227184e-01
5.75348020e-01 -9.08952355e-01 2.39739388e-01 -1.44226182e+00
-1.53806731e-01 1.01276195e+00 1.06849790e+00 2.85277367e-01
1.06972337e+00 -2.64748484e-01 -8.07475984e-01 -7.25718379e-01
-6.60174370e-01 -4.32214737e-01 4.62762713e-01 5.34132600e-01
-1.58700719e-01 1.15243874e-01 -6.67719543e-01 7.94791222e-01
-5.10890931e-02 -1.07191456e-02 -2.44943518e-02 8.46449196e-01
-3.16716880e-01 -1.27239966e+00 7.43879974e-02 7.38279223e-01
1.61035568e-01 -1.88372850e-01 -6.21543884e-01 9.98530328e-01
7.03970194e-01 8.32835257e-01 -1.74062192e-01 -8.70567143e-01
2.64622629e-01 4.31839637e-02 1.91400811e-01 -5.29194415e-01
-1.73790365e-01 -5.98257005e-01 3.58439326e-01 -3.83450985e-01
-4.91729409e-01 -3.68066162e-01 -1.33042300e+00 -3.50412965e-01
-7.42689490e-01 5.36245108e-01 4.33900543e-02 7.65680969e-01
7.33708501e-01 6.88804865e-01 6.69521213e-01 -1.32872462e-01
3.81121814e-01 -7.17024922e-01 -9.74349260e-01 4.36297417e-01
-4.19176698e-01 -1.12194669e+00 -2.28380069e-01 -2.02039808e-01]
|
[8.664121627807617, 7.917446613311768]
|
a7e2361c-b40b-4018-be31-e18518e8d47d
|
in-vitro-micropropagation-and-apocarotenoid
|
2208.13292
| null |
https://arxiv.org/abs/2208.13292v1
|
https://arxiv.org/pdf/2208.13292v1.pdf
|
In vitro micropropagation and apocarotenoid gene expression in saffron
|
Saffron (Crocus sativus L.) is a triploid, sterile, monocot plant belonging to the family Iridaceae, sub-family Crocoideae. C.sativus only blooms once a year and should be collected within a very short duration, the stigmas of Saffron flowers are harvested manually and subjected to desiccation then have been used as a spice. It has been also used as a drug to treat tumors, cancer, chronic uterine hemorrhage, insomnia, scarlet fever, smallpox, colds, and cardiovascular disorders. It has been shown that saffron is a protective agent against chromosomal damage. Saffron has been vegetatively propagated by corm, each mother corm produces 7-8 cormlets each year. The main colors of saffron, crocetin, and crocetin glycosides, and the main flavors, picrocrocin, Safranal is the main component of aroma and Its bitter taste is related to Glycosidepicrocrocin that is derived from the oxidative cleavage of the carotenoid, zeaxanthin.by zeaxanthin cleavage dioxygenase (ZCD). We investigated gene expression of ZCD in vitro by using tissue culture of perianth obtained from immature flora buds of Ghaen of Khorasan province, Iran, on MS medium supplemented by 10 mg/L NAA and BAP. RNA of each sample was extracted using RNx method; followed by RT-PCR techniques. The results indicated that ZCD was present in perianth of all cultured samples of mentioned areas. Investigation of this pathway which controls the saffron apocarotenoid pigments in perianth is important to produce saffron with high quality and quantity. The results showed that the expression of ZCD in perianth whether through mevalonic acid or non-mevalonic acid needs more investigation
|
['Mandana Mirbakhsh']
|
2022-08-28
| null | null | null | null |
['culture']
|
['speech']
|
[ 2.00970009e-01 -1.05847999e-01 -3.41758043e-01 1.84235454e-01
2.96085507e-01 -1.24056840e+00 3.86187822e-01 5.36917210e-01
1.61584422e-01 8.24837983e-01 2.14404374e-01 -4.48627949e-01
-3.59004512e-02 -8.24473679e-01 6.12568706e-02 -1.10095918e+00
-3.34318668e-01 1.01940237e-01 1.58642437e-02 -3.14337909e-01
2.18801960e-01 1.01198590e+00 -1.50813437e+00 -1.87357903e-01
1.11393809e+00 4.72183079e-01 2.90295571e-01 5.26726604e-01
-5.76780252e-02 1.22712210e-01 -4.72437263e-01 -5.46830148e-02
1.56952813e-01 -9.32159662e-01 -5.56716740e-01 1.95303708e-01
-3.86650652e-01 -1.67519510e-01 6.38625741e-01 1.13113010e+00
-1.63780242e-01 -3.40303838e-01 8.40806663e-01 -1.38098323e+00
-6.66118622e-01 6.09860718e-01 -8.89657021e-01 -2.85726786e-01
-3.62375416e-02 6.14274219e-02 3.37703228e-01 -8.08824003e-01
4.56628263e-01 1.28376591e+00 -2.20290665e-02 4.24334645e-01
-8.47217500e-01 -8.76214862e-01 -7.67214537e-01 -1.14414971e-02
-1.36198986e+00 -1.26996974e-03 1.21849246e-01 -1.60402328e-01
3.15028399e-01 4.67849642e-01 1.06776977e+00 -6.81711435e-02
9.53676164e-01 4.09514993e-01 1.35739207e+00 -3.64199758e-01
3.89653504e-01 -1.52894258e-01 -2.24713713e-01 6.19750500e-01
8.78199518e-01 1.23425096e-01 3.60787325e-02 2.04030573e-01
4.14010555e-01 -5.16391322e-02 -3.15061063e-01 2.16554537e-01
-5.07484078e-01 7.64364183e-01 3.61787409e-01 6.30282581e-01
2.34404076e-02 -3.73257399e-01 3.39784503e-01 1.78462174e-02
3.29095200e-02 2.28932247e-01 -5.43019891e-01 -2.93829031e-02
-5.37250161e-01 -1.05649695e-01 8.44942510e-01 4.77525115e-01
2.28508770e-01 1.57770216e-01 3.77281606e-01 5.95675409e-01
7.79448211e-01 6.94557548e-01 2.90814310e-01 -8.36081922e-01
-7.75587261e-01 6.97545111e-01 -3.04973591e-03 -5.81140757e-01
-1.32381665e-02 -1.52071770e-02 -7.08173335e-01 4.61330503e-01
6.38734460e-01 -3.61744672e-01 -7.06820071e-01 1.40497446e+00
5.28586745e-01 -1.46144971e-01 1.14937723e-01 7.70829141e-01
7.00111985e-01 9.84800637e-01 4.73868668e-01 -7.60724247e-01
1.76827347e+00 -4.52251375e-01 -9.95974541e-01 5.24739385e-01
3.10616285e-01 -1.32405496e+00 6.46378219e-01 8.81648898e-01
-1.08034861e+00 -1.64046943e-01 -1.31454146e+00 5.06904721e-01
-5.13228357e-01 2.39933059e-01 6.30816877e-01 1.25738513e+00
-4.27517295e-01 5.98476589e-01 -7.62121439e-01 -6.92892492e-01
1.85077623e-01 2.81652898e-01 -3.10795665e-01 -9.32948738e-02
-6.55276120e-01 8.10496807e-01 2.57446408e-01 3.12814832e-01
-6.23154104e-01 -4.06827152e-01 -7.03317463e-01 1.29568189e-01
1.68968961e-01 -3.34473461e-01 9.44648147e-01 -1.10308444e+00
-1.90100384e+00 6.20628715e-01 -2.36026004e-01 -8.54039639e-02
-5.67361474e-01 -2.58689541e-02 -6.84375107e-01 2.20236048e-01
9.76385921e-02 1.48491576e-01 2.62364447e-01 -1.08362114e+00
-4.48693454e-01 -6.01736963e-01 -3.24206114e-01 -3.84568959e-01
2.09109277e-01 5.29649854e-01 4.88957912e-01 -5.38898587e-01
1.67969540e-01 -8.92277837e-01 -2.74779469e-01 5.11801466e-02
-3.48315418e-01 -1.52573109e-01 1.10494745e+00 -6.60610557e-01
6.93518043e-01 -1.98554277e+00 -4.16052014e-01 2.66154081e-01
-3.11338007e-01 5.75417161e-01 6.70718029e-02 6.44746304e-01
-4.11671326e-02 8.91867816e-01 -2.07362510e-02 1.01660562e+00
-4.18843627e-01 2.08346143e-01 2.84781754e-01 7.15709984e-01
4.67043549e-01 2.41543323e-01 -1.19650018e+00 -4.94481325e-01
-4.52526025e-02 6.52428269e-01 3.59308660e-01 -4.52429444e-01
-3.90407327e-03 -1.46131366e-02 -4.57572132e-01 1.39567447e+00
1.41086268e+00 7.06178173e-02 3.19056988e-01 -1.77497730e-01
-3.94483179e-01 -3.17945838e-01 -1.06863332e+00 1.21498406e+00
1.66543573e-01 2.97388613e-01 5.21378160e-01 -1.41337678e-01
8.16900373e-01 2.71390676e-01 6.39831945e-02 -3.52805704e-01
6.30635440e-01 1.74372405e-01 3.11784595e-01 -1.16791911e-01
3.13041538e-01 -3.01458567e-01 4.35351431e-01 1.61405593e-01
-2.30718434e-01 -5.33537194e-03 1.02705348e+00 4.81618047e-01
3.37049931e-01 3.69211197e-01 7.41485894e-01 -6.91392601e-01
8.91800702e-01 -1.11964159e-01 8.45897019e-01 -6.29085422e-01
-3.78129371e-02 -4.43379283e-02 8.01971674e-01 2.99890876e-01
-6.41584098e-01 -1.24422383e+00 -5.91343999e-01 3.37094605e-01
1.48653269e-01 -5.37111461e-01 -6.46984458e-01 -1.65821567e-01
-1.87924236e-01 8.70546818e-01 -5.38658857e-01 -1.93140566e-01
1.50319949e-01 -7.57086754e-01 3.74555618e-01 -1.40809134e-01
8.94916773e-01 -7.11849868e-01 -5.30026436e-01 -7.62187913e-02
4.24686670e-01 -1.71604127e-01 -1.18752599e-01 -1.17871419e-01
-1.02791572e+00 -1.17688489e+00 -4.93374705e-01 -3.22144806e-01
4.27260131e-01 3.11503023e-01 5.45221508e-01 3.07660580e-01
-3.77615631e-01 -3.95473093e-01 -6.95469856e-01 -1.02885830e+00
-6.16056442e-01 -6.46030247e-01 -2.59147555e-01 -3.41155440e-01
4.79053617e-01 -6.41093254e-01 -4.33101058e-01 -8.60756934e-02
-8.19749236e-01 -4.38116372e-01 5.84093809e-01 5.81107080e-01
4.53662097e-01 2.29355216e-01 5.00842333e-01 -1.26144671e+00
3.27113479e-01 -7.15506911e-01 -8.53645921e-01 -2.08783634e-02
-7.38390505e-01 -6.69175833e-02 4.90366876e-01 1.24407336e-02
-8.73078406e-01 -2.31995061e-01 2.72241950e-01 3.95723373e-01
-1.08822055e-01 5.76228023e-01 -6.55955613e-01 -8.35199952e-02
5.33628285e-01 -9.74915177e-02 4.51499969e-01 -3.33644599e-01
1.78481504e-01 5.24827302e-01 2.71271020e-01 -1.68472365e-01
9.87506509e-01 5.03425956e-01 8.30720961e-01 -1.12900674e+00
-5.07926226e-01 -4.72902715e-01 -4.81757730e-01 -2.75898874e-01
1.09190416e+00 -5.67895830e-01 -1.26247573e+00 1.86001390e-01
-7.48726189e-01 2.18847752e-01 2.69034117e-01 4.78158444e-01
2.22029760e-01 5.39932191e-01 -5.95957100e-01 -1.05413747e+00
-6.97217524e-01 -2.86261201e-01 2.17796549e-01 1.03000736e+00
1.25033244e-01 -6.08395696e-01 2.04783440e-01 -6.52845427e-02
4.23111469e-02 1.09321904e+00 1.19658077e+00 -1.84638366e-01
-3.60293448e-01 -1.08386371e-02 -1.60007462e-01 1.92770809e-01
4.84360009e-01 1.02207947e+00 -6.13522470e-01 -3.41033079e-02
-2.19059810e-01 -8.02533403e-02 4.52976525e-01 4.30813402e-01
4.63561207e-01 -2.20097527e-01 -3.24979484e-01 5.15826166e-01
2.11700416e+00 7.12349415e-01 7.78423965e-01 3.27085704e-01
5.86205721e-02 3.45063001e-01 1.33949947e+00 4.89389092e-01
-5.12329936e-01 -3.06016475e-01 8.73111844e-01 -1.86710700e-01
3.30022909e-02 7.67880604e-02 4.75139916e-01 4.04779047e-01
-5.68427801e-01 -2.17265621e-01 -3.57073098e-01 3.34852785e-01
-8.33262444e-01 -9.32617724e-01 -1.13390815e+00 2.29218340e+00
8.97176445e-01 -3.47951800e-01 1.59940124e-01 3.57174844e-01
3.98416191e-01 -8.06992233e-01 -2.43531182e-01 -9.01805639e-01
-1.97947264e-01 1.08204889e+00 9.86553848e-01 4.03281868e-01
-6.40012860e-01 7.15668142e-01 4.92340517e+00 6.30888939e-02
-1.28710687e+00 -1.67905673e-01 3.11947227e-01 1.57346338e-01
-2.04011321e-01 7.88334966e-01 -5.16477764e-01 2.96923190e-01
9.58031893e-01 -4.93778855e-01 1.68857858e-01 4.09533739e-01
6.65707767e-01 -8.87055635e-01 -4.16362822e-01 5.22734344e-01
-1.21514216e-01 -1.12074757e+00 -3.29095334e-01 2.91440487e-01
6.08737886e-01 -5.39056480e-01 -4.57028240e-01 -1.08285084e-01
2.20693171e-01 -1.02001250e+00 2.32925832e-01 2.43539035e-01
5.50681949e-01 -1.53235388e+00 6.37975335e-01 1.23967238e-01
-1.21503472e+00 2.34381035e-02 -4.00190860e-01 -4.13255468e-02
-1.50887109e-02 6.38710618e-01 -9.26412523e-01 6.40375316e-01
6.81578159e-01 5.98417044e-01 -8.35105419e-01 1.47556531e+00
-5.90560675e-01 1.00413501e+00 -9.78441536e-02 -2.41622090e-01
3.99189182e-02 -8.14838409e-01 3.23736072e-01 5.56048512e-01
4.32914048e-01 2.21663713e-01 -9.34077278e-02 7.41445422e-01
2.36001268e-01 6.17359102e-01 -3.04254711e-01 -5.97036719e-01
4.42244768e-01 1.61765754e+00 -1.50110149e+00 -2.83247214e-02
-3.51300925e-01 4.36443627e-01 -8.05915475e-01 2.35913679e-01
-2.56494671e-01 -6.70594871e-01 5.00935197e-01 3.44380736e-01
1.01426698e-01 -6.54483167e-03 -8.00615102e-02 -5.40186584e-01
-8.82274032e-01 -7.44601667e-01 5.82384527e-01 -7.00437367e-01
-7.13732779e-01 -2.61513382e-01 -7.98190683e-02 -9.23331380e-01
6.87321648e-02 -5.90472877e-01 -6.71939850e-01 1.30697608e+00
-1.32155061e+00 -1.32856154e+00 -6.94186687e-02 8.03871527e-02
2.29071170e-01 3.33161712e-01 1.34830439e+00 -3.90210837e-01
-7.21337199e-01 -1.01256691e-01 3.42935294e-01 -1.40680671e-01
9.78832424e-01 -1.17728043e+00 -5.70777953e-01 9.18096662e-01
-2.77772337e-01 6.16514504e-01 7.86854386e-01 -7.27245331e-01
-1.41954553e+00 -9.35016572e-01 1.10695863e+00 3.58832151e-01
1.26121178e-01 4.79257815e-02 -4.64466602e-01 3.51205796e-01
7.86228180e-01 -5.68683088e-01 1.35865366e+00 -7.48525411e-02
2.05397725e-01 8.29744712e-02 -1.33320808e+00 6.13482177e-01
-1.92124933e-01 1.67830646e-01 -3.51209119e-02 4.24639404e-01
5.52975237e-02 2.24967062e-01 -1.17840290e+00 -1.55426428e-01
6.43657744e-01 -8.12460065e-01 4.12630260e-01 -1.85200870e-01
3.44965637e-01 -1.12182057e+00 2.94699252e-01 -9.30162787e-01
-4.78785843e-01 -7.87209153e-01 4.00036633e-01 1.46483588e+00
1.64085045e-01 -1.80469364e-01 3.19567919e-01 -2.82852128e-02
3.91277745e-02 -3.23418118e-02 -1.21208318e-01 -5.11725426e-01
-1.31821394e-01 8.45177829e-01 4.38954294e-01 1.04204369e+00
1.38336062e-01 8.77753496e-01 1.34235039e-01 7.24467263e-02
6.56209588e-01 2.75023282e-01 3.96560937e-01 -1.80733752e+00
-2.73232497e-02 7.63346106e-02 -1.28260151e-01 1.85774907e-01
-2.29601383e-01 -9.66492951e-01 -7.30026186e-01 -1.46173418e+00
7.39965290e-02 1.82978064e-01 3.14371020e-01 3.53569776e-01
1.95090950e-01 1.54866576e-01 4.00599897e-01 -3.94919485e-01
5.29579401e-01 -1.02010690e-01 1.51232874e+00 1.15512937e-01
-4.72075939e-01 -4.73217554e-02 -1.08750629e+00 6.76859915e-01
1.10281754e+00 -4.76116002e-01 -9.24598932e-01 2.33047590e-01
5.17319083e-01 1.36758834e-01 -1.62095204e-01 -5.17931283e-01
-3.54780138e-01 -6.21992290e-01 5.05991340e-01 -5.42431474e-01
1.38526484e-01 -8.21088493e-01 5.39252520e-01 1.10098410e+00
4.24050838e-01 -3.82704705e-01 2.87601918e-01 1.94369838e-01
2.70142794e-01 -6.48770392e-01 1.17732096e+00 -3.10902894e-01
-4.06894833e-01 -1.54435426e-01 -1.29547298e+00 -4.33793187e-01
1.31495345e+00 -6.77677751e-01 -1.25664677e-02 -1.37955680e-01
-5.83868146e-01 -2.75442451e-02 8.06655705e-01 -1.27717137e-01
2.76842952e-01 -1.21855211e+00 -8.57696950e-01 2.15271451e-02
3.22436430e-02 -2.95846909e-01 1.12713594e-02 1.18033040e+00
-1.10999250e+00 2.68598665e-02 -9.77430046e-01 -3.85535449e-01
-1.71378744e+00 6.74086869e-01 -1.61430240e-01 2.90584445e-01
2.05465838e-01 6.24574602e-01 -1.65189669e-01 4.17755753e-01
-4.61455911e-01 -4.77546483e-01 -8.20336759e-01 -4.14815024e-02
4.87082660e-01 5.49103796e-01 -6.29458278e-02 -6.46488726e-01
-1.93448305e-01 1.49659663e-01 1.91367984e-01 2.16267243e-01
1.03414285e+00 2.24113241e-01 -8.31421196e-01 4.67397332e-01
6.46944702e-01 6.03421926e-01 -3.97338301e-01 8.90174925e-01
-1.22541137e-01 -6.90808892e-01 1.68303117e-01 -1.16597247e+00
-6.66671872e-01 4.92999732e-01 6.86636329e-01 1.99505836e-02
1.54450321e+00 -3.36327463e-01 5.67658246e-01 -3.16194624e-01
1.05720842e-02 -9.46512938e-01 -5.13099432e-01 1.27108619e-01
6.21324241e-01 -4.27071869e-01 2.72924751e-02 -1.04204226e+00
-1.83132619e-01 1.34652126e+00 2.49181628e-01 -3.28675807e-01
1.00824189e+00 3.01807106e-01 -5.18222675e-02 1.99641958e-01
-1.06611919e+00 -1.54325932e-01 -1.42985344e-01 6.62959635e-01
1.28305483e+00 1.56878144e-01 -1.70494342e+00 6.60924852e-01
-2.19881848e-01 9.02761519e-02 9.34037983e-01 1.19115186e+00
-6.34406388e-01 -1.51943517e+00 -7.05976844e-01 4.86652315e-01
-7.63956308e-01 1.01117089e-01 -6.97274029e-01 1.12759197e+00
7.39535630e-01 1.08915126e+00 -2.77732134e-01 2.68800378e-01
-5.03451154e-02 -8.41190666e-02 8.04183662e-01 -4.01704401e-01
-7.33610570e-01 9.40640986e-01 1.18644059e-01 -1.09731391e-01
-6.57949686e-01 -6.93018854e-01 -1.77362871e+00 -4.98848021e-01
-4.36919689e-01 6.67878866e-01 8.69464338e-01 7.37178862e-01
-3.33088100e-01 1.68439448e-02 8.15654635e-01 1.89884618e-01
1.89799611e-02 -1.02634394e+00 -1.35411680e+00 -7.97303021e-02
-2.52856106e-01 -4.90913451e-01 -2.77787834e-01 5.52737892e-01]
|
[4.662353038787842, 5.089835166931152]
|
4ce58e23-992c-4a23-851f-45e105db5353
|
multi-modal-attention-network-for-stock
|
2112.13593
| null |
https://arxiv.org/abs/2112.13593v5
|
https://arxiv.org/pdf/2112.13593v5.pdf
|
Multi-modal Attention Network for Stock Movements Prediction
|
Stock prices move as piece-wise trending fluctuation rather than a purely random walk. Traditionally, the prediction of future stock movements is based on the historical trading record. Nowadays, with the development of social media, many active participants in the market choose to publicize their strategies, which provides a window to glimpse over the whole market's attitude towards future movements by extracting the semantics behind social media. However, social media contains conflicting information and cannot replace historical records completely. In this work, we propose a multi-modality attention network to reduce conflicts and integrate semantic and numeric features to predict future stock movements comprehensively. Specifically, we first extract semantic information from social media and estimate their credibility based on posters' identity and public reputation. Then we incorporate the semantic from online posts and numeric features from historical records to make the trading strategy. Experimental results show that our approach outperforms previous methods by a significant margin in both prediction accuracy (61.20\%) and trading profits (9.13\%). It demonstrates that our method improves the performance of stock movements prediction and informs future research on multi-modality fusion towards stock prediction.
|
['Shi Gu', 'Shwai He']
|
2021-12-27
| null | null | null | null |
['stock-prediction']
|
['time-series']
|
[-6.65802002e-01 -2.97324508e-01 -5.77790499e-01 -2.91548878e-01
-5.87260783e-01 -7.40529001e-01 7.28919327e-01 1.07994080e-01
-4.78946030e-01 7.65541911e-01 7.51869023e-01 -7.89166912e-02
7.24038631e-02 -1.18921471e+00 -6.01990521e-01 -3.46339375e-01
-1.10901065e-01 2.60516167e-01 4.72830445e-01 -6.64780259e-01
5.83891451e-01 -1.91891700e-01 -1.30285800e+00 2.12422729e-01
5.72079062e-01 1.69459116e+00 -1.01494864e-02 -1.90059900e-01
-6.17350936e-01 1.28693187e+00 -7.05662966e-01 -1.03842211e+00
4.44357127e-01 -2.15784669e-01 -3.23462099e-01 -1.25273839e-01
-2.05614924e-01 -3.99931878e-01 -1.17320187e-01 1.50252807e+00
2.28397876e-01 -3.16756904e-01 2.16983035e-01 -1.24764359e+00
-1.24070609e+00 1.56057513e+00 -9.95609999e-01 3.75258207e-01
3.61970393e-03 -2.51810133e-01 1.63379276e+00 -8.09374809e-01
6.42295659e-01 8.42264175e-01 7.25870132e-01 -6.19759858e-02
-7.25141943e-01 -1.16647983e+00 6.31114900e-01 1.78245902e-01
-9.53626633e-01 -2.19483841e-02 1.02294588e+00 -3.59925836e-01
3.08064997e-01 2.98145652e-01 1.09509015e+00 8.39628279e-01
4.59173881e-02 1.08959794e+00 1.28197372e+00 1.85280338e-01
1.08030975e-01 3.70880783e-01 1.89940378e-01 1.73322409e-02
5.29324532e-01 -1.25470355e-01 -8.32430363e-01 -2.32144117e-01
3.98621321e-01 3.89228910e-01 -1.21882282e-01 2.40946636e-01
-1.33716464e+00 9.52441037e-01 6.49566770e-01 5.66690683e-01
-6.10539019e-01 -2.95912032e-03 1.07794441e-01 5.04254758e-01
1.01967120e+00 3.49403888e-01 -6.01964891e-01 -3.07696909e-01
-1.00518012e+00 3.84823948e-01 8.75643134e-01 5.01997232e-01
5.46685278e-01 4.09113429e-02 2.40207687e-01 3.69299233e-01
7.42546618e-01 7.62038708e-01 9.17463422e-01 -7.45454550e-01
5.36824763e-01 7.67678261e-01 5.51601410e-01 -1.55570579e+00
-2.18703434e-01 -7.37274170e-01 -4.67228740e-01 1.17356867e-01
3.13179344e-01 -4.28889483e-01 -3.84150058e-01 1.35070932e+00
6.35963753e-02 2.92284757e-01 1.02933787e-01 1.00773692e+00
5.17087936e-01 7.51245558e-01 2.72673685e-02 -2.78544188e-01
1.27105844e+00 -6.48062706e-01 -1.03739619e+00 2.05790158e-02
2.49698967e-01 -8.01216960e-01 4.11035448e-01 2.76870817e-01
-1.03901136e+00 -1.48801133e-01 -1.14337921e+00 5.28007627e-01
-5.00325859e-01 -2.49751762e-01 6.04422688e-01 4.24453259e-01
-7.17090428e-01 7.56170213e-01 -5.65600753e-01 5.77355146e-01
4.78252143e-01 2.23996848e-01 2.01863304e-01 8.05419087e-01
-1.86118114e+00 7.52400517e-01 3.12644899e-01 6.21802248e-02
1.02870025e-01 -5.55178404e-01 -4.42243844e-01 -6.87872171e-02
3.61274034e-01 -3.02353829e-01 1.17022467e+00 -1.10919070e+00
-1.37768185e+00 3.78058255e-01 3.39167297e-01 -8.77273083e-01
8.48720491e-01 -1.51245281e-01 -7.74619997e-01 -2.28088617e-01
2.25865722e-01 3.10555458e-01 6.30611837e-01 -9.44334805e-01
-1.17734110e+00 -4.07533139e-01 1.31238624e-01 2.38830566e-01
-6.00658774e-01 4.23378013e-02 -1.82133451e-01 -1.13571680e+00
2.20436722e-01 -7.50371516e-01 -5.84915467e-02 -2.56153315e-01
-1.29728913e-01 -2.57834554e-01 7.50573277e-01 -6.66126668e-01
1.46489000e+00 -1.91123927e+00 -2.81203240e-01 4.31346565e-01
3.83534819e-01 -5.40521964e-02 3.35499376e-01 3.46249998e-01
3.88386458e-01 2.18713015e-01 2.85406655e-04 -1.06916741e-01
4.58969325e-01 -2.87048012e-01 -8.91406775e-01 3.12240720e-01
-2.98360586e-01 1.32984757e+00 -6.58054352e-01 -2.06417203e-01
-1.43183410e-01 1.24127053e-01 -1.87929869e-01 -4.49872255e-01
-3.34868491e-01 5.73079176e-02 -9.20292258e-01 7.63432741e-01
6.18728697e-01 -8.79311025e-01 9.45011750e-02 -8.51517543e-02
-2.28255615e-01 3.28260928e-01 -1.22647369e+00 1.14004922e+00
-7.05286190e-02 4.61375147e-01 -3.52833748e-01 -6.00319445e-01
1.09407759e+00 4.38301899e-02 6.61324322e-01 -7.88649380e-01
3.42338473e-01 5.16147554e-01 -3.26914757e-01 6.87164590e-02
7.31289864e-01 -4.12790805e-01 -4.72130716e-01 8.25423658e-01
-5.71773887e-01 4.32830691e-01 1.16263225e-03 1.58927515e-01
4.36820477e-01 -9.02609974e-02 8.40764344e-02 -4.49689552e-02
3.29563349e-01 4.37661447e-02 6.71255410e-01 3.70094031e-01
-2.23982200e-01 1.32298484e-01 3.27894211e-01 -6.23431921e-01
-5.54068387e-01 -7.31816411e-01 1.05421402e-01 8.31828058e-01
6.26021087e-01 -2.12781519e-01 -4.31073219e-01 -6.10405326e-01
4.02231663e-01 7.36504734e-01 -6.37211680e-01 2.56831408e-01
-1.06931448e-01 -1.13043261e+00 5.29916212e-02 6.19584620e-01
7.74531126e-01 -1.01803637e+00 -1.94663808e-01 4.01251197e-01
-2.96559393e-01 -9.04184580e-01 -4.95386690e-01 -5.07786036e-01
-5.66650271e-01 -9.20100868e-01 -9.17349815e-01 -3.70126247e-01
1.78947121e-01 1.09686300e-01 8.12483490e-01 -8.97122025e-02
6.81131661e-01 4.80870940e-02 -3.68977487e-01 -8.02721143e-01
9.16553438e-02 -2.82508116e-02 1.55261988e-02 4.10165757e-01
8.08141291e-01 -4.47542787e-01 -7.95709431e-01 3.64612460e-01
-7.13104606e-01 -1.07131116e-01 5.05859911e-01 3.69534075e-01
3.32191169e-01 4.21080858e-01 9.89142478e-01 -8.42417419e-01
8.23129654e-01 -8.97035658e-01 -9.85143840e-01 7.88845718e-02
-1.13514018e+00 -2.10265383e-01 4.90428247e-02 -3.36794972e-01
-1.18871462e+00 -5.14373243e-01 1.86518833e-01 -1.40432686e-01
6.56879246e-01 1.13230515e+00 2.94543177e-01 2.05391735e-01
-3.13607603e-02 3.53735685e-01 1.12176705e-02 -4.98015225e-01
2.43168756e-01 7.56208956e-01 1.34332657e-01 1.99929908e-01
1.00865030e+00 7.84229338e-01 -6.20178401e-01 -1.17985517e-01
-1.11977005e+00 -1.82875663e-01 5.42318039e-02 -3.22704583e-01
5.87579966e-01 -1.25248921e+00 -9.89784181e-01 1.01639485e+00
-7.05003500e-01 2.78404802e-01 -1.81696922e-01 8.12322557e-01
-4.96930555e-02 2.51283616e-01 -9.91491497e-01 -1.03028393e+00
-3.70583594e-01 -7.00430095e-01 5.56211114e-01 1.82880387e-01
-1.66870639e-01 -1.10293806e+00 2.97744181e-02 7.05688596e-01
6.75005615e-01 1.35659531e-01 7.42008761e-02 -1.08085990e+00
-1.05559301e+00 -4.91902560e-01 -3.03815573e-01 1.65345788e-01
1.90774366e-01 -3.23965847e-01 -6.62742972e-01 1.27777129e-01
2.52128899e-01 -1.04058482e-01 1.11570632e+00 3.20888638e-01
4.63327855e-01 -5.65632164e-01 -9.81445312e-02 1.13333641e-02
1.17644513e+00 3.53979468e-01 3.90985757e-01 8.02131474e-01
4.75272894e-01 6.19204044e-01 6.64645255e-01 8.58870685e-01
9.31636930e-01 2.01697037e-01 3.70007992e-01 3.40652198e-01
5.30431867e-01 -5.98225892e-01 6.90358341e-01 1.15423560e+00
-1.38725743e-01 -1.11998931e-01 -5.47951341e-01 3.55468869e-01
-1.93753386e+00 -1.31899810e+00 1.30581260e-01 1.70570672e+00
8.02585840e-01 6.38327897e-01 3.19064528e-01 8.68918076e-02
1.04555774e+00 5.73575556e-01 -6.10815346e-01 5.60724318e-01
-5.83914757e-01 -5.70675611e-01 9.43645060e-01 1.68293983e-01
-1.03307438e+00 6.66563630e-01 5.66792774e+00 9.77288306e-01
-1.17647624e+00 -6.11937896e-04 8.69906366e-01 -9.74909663e-02
-1.01085901e+00 -2.34236166e-01 -1.07808888e+00 1.10582161e+00
7.39498138e-01 -7.91171134e-01 1.41809151e-01 6.15588427e-01
4.61924635e-02 2.13059962e-01 -1.85438186e-01 6.78517342e-01
2.40586028e-02 -1.94881809e+00 1.08199872e-01 2.57242709e-01
8.55158925e-01 2.10550711e-01 4.00638372e-01 6.87509626e-02
6.32078469e-01 -2.72708803e-01 1.37683666e+00 8.29962134e-01
-3.67559493e-02 -7.81976759e-01 1.04967177e+00 3.22918773e-01
-1.44966924e+00 -1.73704028e-01 4.62525599e-02 -2.01158926e-01
5.16093910e-01 6.62529230e-01 -1.70329869e-01 5.53704560e-01
7.66694427e-01 1.27642572e+00 -2.24401668e-01 6.71557009e-01
-5.73062226e-02 5.12644231e-01 -3.74231935e-01 -6.16235256e-01
6.03216290e-01 -4.96260613e-01 4.23209786e-01 3.38671029e-01
7.13272929e-01 2.57366896e-01 -2.35569868e-02 8.23763490e-01
-4.31286544e-01 2.44400874e-01 -1.73505470e-01 -5.80640256e-01
4.21361029e-01 8.44999135e-01 -9.65323448e-01 -3.98743391e-01
-9.29763854e-01 5.42865932e-01 -1.83245122e-01 1.47309810e-01
-9.20182288e-01 -1.62789628e-01 5.43269515e-01 1.95386976e-01
6.56453311e-01 1.23225369e-01 -3.85492951e-01 -1.45817649e+00
2.43335843e-01 -2.61233807e-01 4.59354281e-01 -6.90024257e-01
-1.74819326e+00 5.71076572e-01 -4.41660643e-01 -1.69378209e+00
4.84548397e-02 -3.90762299e-01 -6.59015238e-01 5.38385570e-01
-1.80391800e+00 -9.54364121e-01 4.65932608e-01 3.31673831e-01
1.72028482e-01 -5.52114189e-01 1.75548673e-01 2.56471753e-01
-1.47849873e-01 1.17638081e-01 2.14953631e-01 5.07863641e-01
3.97473127e-01 -1.23827124e+00 6.17227018e-01 4.74799484e-01
4.67486501e-01 3.11240315e-01 4.85716373e-01 -1.08392334e+00
-9.54880893e-01 -8.80261958e-01 1.24795425e+00 -3.48893940e-01
1.69242537e+00 2.50938684e-01 -8.39543879e-01 6.81486428e-01
1.91415563e-01 -2.72580475e-01 9.66552794e-01 -1.92589536e-02
-2.93486893e-01 -2.00621277e-01 -9.13840175e-01 5.82260191e-01
7.21734524e-01 -3.58006656e-01 -9.64766681e-01 2.05660369e-02
8.68511796e-01 -8.73534307e-02 -1.01186001e+00 9.28557664e-02
6.93323135e-01 -9.84428644e-01 7.34840810e-01 -1.51231140e-01
2.10543245e-01 -4.31645900e-01 -1.57323033e-01 -1.08092797e+00
-2.10701182e-01 -6.63848460e-01 -1.01411186e-01 1.11602342e+00
9.16499257e-01 -1.06853712e+00 9.48053241e-01 8.04420412e-01
4.24960524e-01 -4.85902399e-01 -7.86942244e-01 -4.91184384e-01
-1.32196054e-01 -6.21489704e-01 1.14858997e+00 1.23561919e+00
2.37759411e-01 5.12944460e-02 -5.87018967e-01 -8.35622568e-03
5.27903378e-01 8.88700545e-01 1.64425969e-01 -1.42135608e+00
-3.41916159e-02 -7.31489956e-01 -4.61593628e-01 -1.02428412e+00
1.46214873e-01 -7.32110083e-01 -7.65971780e-01 -1.17169654e+00
3.17486785e-02 -2.97169387e-01 -9.37860668e-01 2.21981466e-01
5.52484021e-03 5.05791843e-01 3.58595997e-01 6.66092932e-01
-8.39343369e-01 7.27044463e-01 1.36503649e+00 -3.03726315e-01
-1.12164140e-01 3.88265759e-01 -1.11961353e+00 8.82482052e-01
7.47535467e-01 -2.91889012e-01 -8.91647190e-02 -6.71957955e-02
1.03265488e+00 4.26742919e-02 1.45098329e-01 -4.51439947e-01
4.71446544e-01 -1.18634857e-01 4.18471664e-01 -8.86192560e-01
2.62291461e-01 -8.70533824e-01 2.81107724e-01 4.91300046e-01
-3.96808535e-01 2.04093918e-01 -1.99213535e-01 1.07878006e+00
-6.94009900e-01 1.33118615e-01 1.82386264e-01 -1.25113145e-01
-8.27451587e-01 3.88381481e-01 -2.93372005e-01 -9.86755732e-03
1.10214496e+00 -1.90679103e-01 -2.51089633e-01 -8.27348590e-01
-9.53762412e-01 4.95187759e-01 2.18379766e-01 7.36233175e-01
4.83039290e-01 -1.73732209e+00 -7.73195207e-01 2.22217049e-02
-9.97414514e-02 -6.82229161e-01 2.64901787e-01 8.14043462e-01
-3.57172787e-01 2.81516105e-01 -2.96833571e-02 4.46057096e-02
-7.97001243e-01 3.69470954e-01 1.03072181e-01 -8.30822214e-02
-3.72494996e-01 8.09298813e-01 -2.26175278e-01 1.29162326e-01
8.90986845e-02 -4.10912395e-01 -7.35064626e-01 1.06310046e+00
7.50030637e-01 3.71212721e-01 -3.65440339e-01 -8.71522069e-01
-1.38050646e-01 5.14799714e-01 -7.00986385e-02 -4.45054412e-01
1.74006498e+00 -5.91567159e-01 -1.33883238e-01 8.37047637e-01
1.02830362e+00 3.29266965e-01 -1.14897490e+00 -6.92399502e-01
4.85177815e-01 -6.02549016e-01 2.55075306e-01 -8.30163479e-01
-1.64336050e+00 1.12555236e-01 1.15673646e-01 8.82901967e-01
8.49941254e-01 6.78693652e-02 1.40716815e+00 2.57756889e-01
5.80361366e-01 -1.40857065e+00 -2.61314958e-01 7.82552809e-02
6.70128703e-01 -1.39476800e+00 1.75936520e-02 -2.10936982e-02
-1.16031325e+00 7.57370353e-01 8.41995850e-02 -2.06936926e-01
1.46014059e+00 3.34171355e-02 5.28959394e-01 -2.73071557e-01
-7.37726331e-01 -1.56766757e-01 3.35825503e-01 -2.35093951e-01
1.44646749e-01 1.55852184e-01 -2.88354248e-01 1.32455337e+00
-4.72128063e-01 -4.18120762e-03 4.78739053e-01 7.59556472e-01
-6.45378292e-01 -1.03962827e+00 -2.16640159e-01 6.74287081e-01
-9.84108925e-01 -1.24361709e-01 -4.63880032e-01 5.04655182e-01
4.57592905e-02 8.13622236e-01 2.82640517e-01 -5.51650882e-01
1.89989164e-01 -1.27325878e-01 -3.09643120e-01 -9.84049961e-02
-7.21819401e-01 4.64523196e-01 6.70400932e-02 -2.65466213e-01
-8.52191687e-01 -9.77477252e-01 -1.18310702e+00 -6.79577589e-01
-4.83199030e-01 4.83115464e-01 5.33085465e-01 8.28927994e-01
3.43693197e-01 1.71074808e-01 1.01622748e+00 -3.93916130e-01
-8.28473926e-01 -7.06261992e-01 -1.08392751e+00 2.85846651e-01
2.01534912e-01 -7.23042071e-01 -6.37066603e-01 -1.81248263e-01]
|
[4.383845806121826, 4.278177261352539]
|
83b23b7d-6fca-4537-a6c9-ab862867a60e
|
using-data-augmentations-and-vtln-to-reduce
|
2307.02009
| null |
https://arxiv.org/abs/2307.02009v1
|
https://arxiv.org/pdf/2307.02009v1.pdf
|
Using Data Augmentations and VTLN to Reduce Bias in Dutch End-to-End Speech Recognition Systems
|
Speech technology has improved greatly for norm speakers, i.e., adult native speakers of a language without speech impediments or strong accents. However, non-norm or diverse speaker groups show a distinct performance gap with norm speakers, which we refer to as bias. In this work, we aim to reduce bias against different age groups and non-native speakers of Dutch. For an end-to-end (E2E) ASR system, we use state-of-the-art speed perturbation and spectral augmentation as data augmentation techniques and explore Vocal Tract Length Normalization (VTLN) to normalise for spectral differences due to differences in anatomy. The combination of data augmentation and VTLN reduced the average WER and bias across various diverse speaker groups by 6.9% and 3.9%, respectively. The VTLN model trained on Dutch was also effective in improving performance of Mandarin Chinese child speech, thus, showing generalisability across languages
|
['Odette Scharenborg', 'Tanvina Patel']
|
2023-07-05
| null | null | null | null |
['anatomy', 'speech-recognition']
|
['miscellaneous', 'speech']
|
[ 2.03707933e-01 4.43711251e-01 1.95985273e-01 -2.72925586e-01
-9.09917355e-01 -6.35028839e-01 3.37818056e-01 -5.57553880e-02
-6.52965903e-01 3.47160071e-01 6.39701605e-01 -5.11283636e-01
1.41236708e-01 -1.75662488e-01 -5.18862128e-01 -5.68836331e-01
2.25532129e-01 9.04922709e-02 -6.84613362e-02 -3.33744824e-01
-1.25447884e-01 4.04486001e-01 -1.47153640e+00 -2.48079235e-03
9.30344343e-01 1.96358576e-01 2.10652441e-01 7.33364522e-01
-1.35813169e-02 9.09809247e-02 -1.07610834e+00 -3.17394167e-01
4.27663438e-02 -4.04774398e-01 -4.84915465e-01 -2.73607612e-01
8.39801192e-01 -1.51767179e-01 -3.56817096e-01 9.67867494e-01
1.25459313e+00 1.79775417e-01 3.50053072e-01 -5.62995553e-01
-6.66044891e-01 1.32390392e+00 -4.65938717e-01 2.80951619e-01
3.31435025e-01 1.11836396e-01 5.82017899e-01 -6.84549153e-01
2.85625249e-01 1.44086170e+00 5.74435532e-01 1.18083560e+00
-1.32947445e+00 -9.17589247e-01 2.39049643e-01 -3.51408087e-02
-1.17633402e+00 -1.05006981e+00 6.53728783e-01 -3.82094711e-01
1.06669247e+00 4.85801667e-01 6.40835702e-01 1.35359061e+00
-4.70330864e-01 4.75360930e-01 1.00488031e+00 -6.76626325e-01
-4.00503241e-02 -1.90271679e-02 3.96384150e-02 5.49629964e-02
-1.16232283e-01 2.63692498e-01 -9.33496654e-01 2.39204690e-01
4.38337743e-01 -7.80477405e-01 -3.44985604e-01 4.10864770e-01
-1.26445842e+00 5.30950844e-01 -3.45269106e-02 6.08747542e-01
-1.93242878e-01 -1.42953023e-01 6.72954202e-01 4.33296323e-01
5.80558240e-01 4.91571426e-01 -7.63274491e-01 -6.67654634e-01
-7.88345873e-01 -7.40004145e-03 5.92531919e-01 8.59792531e-01
-1.64724246e-01 7.48526752e-01 -1.88941285e-01 1.65607750e+00
1.65290847e-01 8.05265605e-01 8.34417403e-01 -7.13244498e-01
6.05855227e-01 1.14895245e-02 -5.52567542e-01 -1.94807917e-01
-5.63140392e-01 -4.91966516e-01 -5.19738615e-01 1.29531771e-01
7.24638939e-01 -4.21907574e-01 -1.31058860e+00 2.38862610e+00
2.83259720e-01 -3.31164241e-01 2.73859620e-01 6.73394144e-01
8.92781734e-01 5.63094318e-01 2.33615711e-01 -4.56102908e-01
1.20373380e+00 -8.25177550e-01 -9.63052392e-01 -3.87926161e-01
4.92070198e-01 -1.15282762e+00 1.40456486e+00 4.68555748e-01
-1.38115966e+00 -5.28074741e-01 -9.87480998e-01 1.25692561e-01
-1.15650788e-01 2.54286509e-02 1.41623005e-01 1.49219227e+00
-1.16124618e+00 3.82893473e-01 -1.03449535e+00 -3.80845487e-01
-2.23356578e-03 5.23527861e-01 -4.16542560e-01 4.21391785e-01
-1.04412735e+00 8.16967130e-01 2.80060619e-01 -1.23921402e-01
-5.38000822e-01 -1.23249209e+00 -8.72898579e-01 -2.42672816e-01
-6.86868727e-02 -3.76579836e-02 1.40480447e+00 -7.45077789e-01
-2.18243217e+00 9.16968703e-01 1.32488355e-01 -2.50777245e-01
5.16742766e-01 -5.88837147e-01 -8.30845952e-01 -2.55380988e-01
-2.22199634e-01 5.31263411e-01 6.79949582e-01 -7.03877330e-01
-4.55412745e-01 -4.72700775e-01 -6.39846265e-01 3.95405084e-01
-4.74076658e-01 6.59808099e-01 -1.04456708e-01 -9.70672190e-01
2.85171270e-01 -1.04689884e+00 1.79000750e-01 -5.51799238e-01
-3.12504143e-01 -2.23014548e-01 5.29300094e-01 -1.40937340e+00
1.30640411e+00 -2.38948584e+00 -2.97224778e-03 2.14322731e-01
-8.40581805e-02 5.81117511e-01 -2.00235114e-01 2.99969465e-02
-3.52958709e-01 2.89928705e-01 -5.94333597e-02 -1.76798493e-01
-1.08079072e-02 -7.75907515e-03 2.15914458e-01 5.32186866e-01
1.15130715e-01 4.32211131e-01 -6.42473578e-01 1.23791926e-01
2.87125885e-01 6.68624222e-01 -5.83833814e-01 1.78100348e-01
2.13805676e-01 7.22055852e-01 3.57581288e-01 4.51937437e-01
5.95094502e-01 1.00980234e+00 -1.84973359e-01 2.59478569e-01
-4.47048962e-01 8.31619680e-01 -9.44121182e-01 1.72314394e+00
-6.04024768e-01 8.13549161e-01 5.02331913e-01 -6.22976184e-01
9.80533421e-01 6.13683701e-01 2.16698274e-03 -7.00646937e-01
8.56767669e-02 7.05634534e-01 1.03738248e+00 -2.75315434e-01
1.26049578e-01 -1.57835186e-01 2.17657909e-01 -1.03079408e-01
6.21709339e-02 -2.14061916e-01 -4.22034739e-03 -4.31811780e-01
8.35301995e-01 -3.30862522e-01 -3.60486917e-02 -6.69651151e-01
4.61539954e-01 -7.07635880e-01 5.69651067e-01 5.53897858e-01
-2.30795920e-01 7.01039672e-01 2.07605258e-01 2.41496876e-01
-1.08901739e+00 -1.36000085e+00 -2.36220568e-01 1.47414827e+00
-7.96616554e-01 -1.87152520e-01 -1.07176077e+00 -1.38707533e-01
-2.58226097e-01 1.00749743e+00 -2.01553628e-01 -8.28450471e-02
-1.00461960e+00 -5.47787666e-01 9.67228055e-01 5.26469827e-01
1.49409056e-01 -1.09619725e+00 -1.60186350e-01 3.92007977e-01
1.32358670e-01 -1.20365942e+00 -7.64075398e-01 9.87150595e-02
-4.34802324e-01 -4.59028602e-01 -1.19683552e+00 -7.90384948e-01
1.85450941e-01 -3.94815564e-01 6.88413084e-01 -3.79273564e-01
5.22421151e-02 2.84930915e-01 -4.31229264e-01 -8.22491288e-01
-1.04783094e+00 4.12312567e-01 7.23011851e-01 -3.91029865e-01
1.78528816e-01 -6.47620857e-01 -4.42539662e-01 2.69208461e-01
-3.89486283e-01 -1.61540881e-01 4.45602745e-01 5.63298106e-01
6.61715046e-02 -6.10222936e-01 8.71881843e-01 -5.88294625e-01
5.34453809e-01 -3.01267020e-02 -5.09092629e-01 -1.35165393e-01
-5.12310147e-01 -7.60583356e-02 4.54143733e-01 -9.60018337e-01
-1.07888973e+00 -2.31706589e-01 -8.66307795e-01 1.73030570e-02
-3.69221658e-01 1.66929021e-01 -5.61140597e-01 1.97514802e-01
6.93962932e-01 -1.22727349e-01 2.58094996e-01 -6.26224458e-01
2.94839382e-01 1.16231072e+00 8.22467208e-01 -5.38649976e-01
7.98275888e-01 -2.51443326e-01 -5.03798068e-01 -1.49879515e+00
-3.42706800e-01 -2.50789255e-01 -5.18825948e-01 6.81790058e-03
8.74352515e-01 -7.28724003e-01 -4.36814785e-01 1.00592494e+00
-1.06556237e+00 -5.35029888e-01 -2.98128068e-01 1.00204742e+00
-3.72597486e-01 8.99428949e-02 -4.92353439e-01 -8.91356766e-01
-8.01506221e-01 -1.14886534e+00 5.75088680e-01 2.79215604e-01
-5.66295981e-01 -6.69182122e-01 -7.82135278e-02 4.76866722e-01
6.55668557e-01 1.22580998e-01 7.69932985e-01 -1.13867426e+00
3.73525918e-01 1.08493149e-01 3.12046915e-01 9.77675259e-01
3.04153234e-01 2.82177404e-02 -1.33557034e+00 -4.48351443e-01
-7.27149621e-02 9.40631181e-02 1.55233115e-01 5.74224114e-01
5.71229696e-01 -9.67498720e-02 3.04159760e-01 5.80052197e-01
5.21127164e-01 6.18460357e-01 4.13635969e-01 7.21260607e-02
7.29590237e-01 9.60799277e-01 1.55598253e-01 -7.88965379e-04
-1.23000547e-01 6.98213875e-01 -1.77382216e-01 -2.17063680e-01
-6.44529462e-01 -6.24050274e-02 7.75695741e-01 1.56565976e+00
-1.39041156e-01 3.70842009e-03 -1.07096803e+00 8.75056028e-01
-8.98632050e-01 -5.83573341e-01 -2.82410860e-01 2.34390235e+00
1.17782629e+00 3.49170178e-01 4.67313141e-01 5.12152135e-01
1.03476620e+00 5.22602051e-02 -4.70489889e-01 -1.00632012e+00
-1.99942559e-01 5.06417871e-01 5.26225746e-01 4.50055063e-01
-5.76914966e-01 1.08698618e+00 6.24625397e+00 9.58482325e-01
-1.33639729e+00 1.61841005e-01 4.00412440e-01 -3.29250246e-01
-8.65314007e-02 -5.88964343e-01 -8.07626367e-01 1.83661208e-01
1.44728100e+00 -3.29590440e-01 7.82257617e-01 5.21037817e-01
4.09431487e-01 3.94160777e-01 -1.07786059e+00 9.60074961e-01
-1.23958131e-02 -5.18482327e-01 -5.02295434e-01 -6.08677929e-03
5.67593217e-01 4.83887285e-01 1.99717522e-01 3.48452568e-01
-4.72951010e-02 -9.46111262e-01 1.09047914e+00 -1.85379088e-01
1.18928516e+00 -7.49903679e-01 4.94492143e-01 1.16222903e-01
-8.75360847e-01 2.17498109e-01 9.22922119e-02 -1.01468097e-02
5.27067855e-03 1.80851042e-01 -9.23916996e-01 7.98334330e-02
6.18082166e-01 -5.64605631e-02 -4.07057583e-01 8.04211080e-01
-3.19836318e-01 1.30804241e+00 -6.70894504e-01 -8.58317595e-03
-6.10644668e-02 1.98296994e-01 8.58119369e-01 1.50698555e+00
4.41635817e-01 -5.47255464e-02 -4.19636846e-01 4.67681646e-01
-8.92246608e-03 4.91522431e-01 -1.20888606e-01 -1.40912205e-01
7.54947722e-01 7.94095993e-01 -3.71615857e-01 1.44778356e-01
-6.19611263e-01 6.93486929e-01 -7.10140308e-03 2.09888041e-01
-4.29097235e-01 -6.39876783e-01 1.00095081e+00 1.50799662e-01
-7.22632855e-02 -1.23098880e-01 -3.24712813e-01 -6.46421492e-01
6.01704642e-02 -1.27063191e+00 -3.82660590e-02 -1.20568149e-01
-9.44715559e-01 6.17384374e-01 -4.88480646e-03 -7.38595128e-01
-4.33377355e-01 -7.06860483e-01 -5.80541253e-01 1.31402993e+00
-1.13022304e+00 -9.64896560e-01 1.35677114e-01 4.14349020e-01
8.28779221e-01 -4.49909657e-01 7.45412886e-01 5.77288568e-01
-5.62701821e-01 1.30104256e+00 1.67082879e-03 1.85209408e-01
7.76963770e-01 -1.45690417e+00 8.36781740e-01 1.05324793e+00
-3.37079801e-02 6.68285131e-01 8.83526742e-01 -3.02076191e-01
-9.16892648e-01 -8.96876693e-01 7.65315056e-01 -1.01545610e-01
7.05695450e-01 -7.30856955e-01 -9.67449605e-01 4.04217899e-01
3.48363519e-01 -3.22396070e-01 7.85194755e-01 3.34257424e-01
-3.10111284e-01 -1.02630340e-01 -1.00017738e+00 8.61716628e-01
1.30237472e+00 -4.99534249e-01 -6.28963172e-01 -6.68386370e-02
1.20442116e+00 -6.46120608e-01 -1.04200125e+00 5.47836840e-01
7.11702287e-01 -5.80083489e-01 7.65721202e-01 -2.52638072e-01
-1.32363960e-01 6.81302994e-02 -1.35471478e-01 -1.86284375e+00
-6.10056110e-02 -1.26061904e+00 2.21458212e-01 2.00161839e+00
7.33761787e-01 -7.88777590e-01 4.07542735e-01 5.09097159e-01
-7.04841256e-01 -2.32701197e-01 -1.26088798e+00 -9.77682471e-01
5.46288550e-01 -6.66076481e-01 5.21851540e-01 7.23469138e-01
-8.11167806e-02 2.56403834e-01 -6.97285756e-02 2.72439778e-01
1.70647949e-01 -8.97293508e-01 5.15069604e-01 -1.08016288e+00
-1.51580244e-01 -7.82872021e-01 -4.52681035e-01 -5.14703453e-01
1.83993220e-01 -7.83975720e-01 1.28953189e-01 -1.07143724e+00
-4.14877474e-01 -1.97704777e-01 -2.27654338e-01 3.65119576e-01
-3.92092913e-01 -1.20149679e-01 2.62177467e-01 -4.79911774e-01
4.43714052e-01 3.85125577e-01 1.03055215e+00 9.56561044e-02
-6.60567760e-01 3.91014546e-01 -7.72776365e-01 7.91154146e-01
9.90742922e-01 -3.98193121e-01 -4.21961725e-01 -5.89557409e-01
-2.64446348e-01 -1.04778804e-01 -3.33360344e-01 -1.05655241e+00
-1.57442749e-01 2.11637422e-01 5.67628667e-02 -3.38037252e-01
7.53473788e-02 -3.38486046e-01 -2.07452565e-01 5.16411185e-01
-3.48835289e-01 2.76646949e-03 6.33002937e-01 -2.51255065e-01
9.59326606e-03 -1.17166601e-01 1.05726933e+00 3.95279258e-01
-1.27819285e-01 -1.16426051e-01 -6.56490445e-01 4.71336871e-01
4.82662052e-01 -2.00102538e-01 -2.44573757e-01 -3.06343704e-01
-7.98117995e-01 -1.90574247e-02 -2.94088740e-02 7.09743619e-01
1.42282590e-01 -1.11341822e+00 -1.25171113e+00 4.63639528e-01
-1.12477699e-02 -1.80770710e-01 2.44473889e-01 6.83611512e-01
-4.20447707e-01 3.09864581e-01 -5.57706282e-02 -3.79874736e-01
-1.74936247e+00 9.48239341e-02 3.39827746e-01 2.81079769e-01
-3.91427338e-01 1.26139820e+00 8.05118084e-02 -9.04273450e-01
5.66533566e-01 -5.46465874e-01 -2.29390904e-01 1.44804791e-01
4.54514802e-01 7.06211448e-01 4.43709522e-01 -7.26653278e-01
-4.57316816e-01 3.74918014e-01 -2.22616062e-01 -5.51180482e-01
1.05097175e+00 -6.90808892e-02 2.57657439e-01 5.63685060e-01
1.00863731e+00 7.58584142e-01 -8.65330219e-01 -4.24693748e-02
1.69482268e-02 -1.12416834e-01 1.62066996e-01 -9.54005241e-01
-9.57497895e-01 8.27308118e-01 9.70628977e-01 2.19706357e-01
1.14012086e+00 3.23409848e-02 6.83277428e-01 -1.55648589e-01
-1.73228800e-01 -1.29508233e+00 -5.31001151e-01 6.65328085e-01
1.16292119e+00 -7.33317733e-01 -5.18810868e-01 -4.86854762e-01
-4.69570428e-01 8.98412049e-01 5.72505116e-01 3.85990202e-01
4.63499963e-01 4.44759637e-01 6.26855195e-01 3.79294425e-01
-3.43699723e-01 -3.65924388e-01 4.36974645e-01 9.31423783e-01
1.05814528e+00 3.82461220e-01 -6.56675041e-01 6.02566957e-01
-1.23117006e+00 -6.69433832e-01 3.30265731e-01 4.61254209e-01
-8.85999128e-02 -1.12882328e+00 -6.93492770e-01 3.01158756e-01
-1.00368857e+00 -3.14918995e-01 -2.07051530e-01 8.70427489e-01
1.85166985e-01 1.28605044e+00 1.13992274e-01 -3.98158491e-01
8.97236109e-01 3.76440108e-01 3.58469546e-01 -7.46417284e-01
-1.02469206e+00 5.94418466e-01 4.18210000e-01 -1.22019120e-01
-2.96704890e-03 -1.10297155e+00 -1.23841679e+00 -1.40357569e-01
-4.16707128e-01 -5.04043549e-02 1.10816669e+00 7.48531044e-01
-2.34357089e-01 9.62875903e-01 3.75615180e-01 -4.28053081e-01
-6.30463719e-01 -1.57858682e+00 -4.96770620e-01 1.62396580e-01
6.08994961e-01 -1.98441714e-01 -5.78247070e-01 -8.21084604e-02]
|
[14.459238052368164, 6.4495320320129395]
|
47c5c8d5-5334-4414-80ed-54e527017657
|
a-25d-cascaded-convolutional-neural-network
|
1806.01018
| null |
http://arxiv.org/abs/1806.01018v2
|
http://arxiv.org/pdf/1806.01018v2.pdf
|
A 2.5D Cascaded Convolutional Neural Network with Temporal Information for Automatic Mitotic Cell Detection in 4D Microscopic Images
|
In recent years, intravital skin imaging has been increasingly used in
mammalian skin research to investigate cell behaviors. A fundamental step of
the investigation is mitotic cell (cell division) detection. Because of the
complex backgrounds (normal cells), the majority of the existing methods cause
several false positives. In this paper, we proposed a 2.5D cascaded end-to-end
convolutional neural network (CasDetNet) with temporal information to
accurately detect automatic mitotic cell in 4D microscopic images with few
training data. The CasDetNet consists of two 2.5D networks. The first one is
used for detecting candidate cells with only volume information and the second
one, containing temporal information, for reducing false positive and adding
mitotic cells that were missed in the first step. The experimental results show
that our CasDetNet can achieve higher precision and recall compared to other
state-of-the-art methods.
|
['Yen-Wei Chen', 'Satoko Takemoto', 'Xian-Hau Han', 'Titinunt Kitrungrotsakul', 'Yutaro Iwamoto', 'Tomomi Nemoto', 'Hideo Yokota', 'Xiong Wei', 'Sari Ipponjima']
|
2018-06-04
| null | null | null | null |
['cell-detection']
|
['computer-vision']
|
[ 1.08497046e-01 -5.09336948e-01 2.35704258e-02 2.33848870e-01
-3.58678550e-01 -3.74001294e-01 4.31119740e-01 2.19865859e-01
-7.85873652e-01 8.76062989e-01 -5.93497038e-01 -1.00366235e-01
4.91574705e-01 -9.04174626e-01 -4.40158308e-01 -1.10709774e+00
8.59748647e-02 1.70776710e-01 9.93365467e-01 1.25397697e-01
3.78281742e-01 4.74384516e-01 -1.23934817e+00 1.47577062e-01
6.97527111e-01 1.01662374e+00 1.07166961e-01 9.17620957e-01
-3.06163102e-01 5.59411108e-01 -3.27108771e-01 -9.37519670e-02
-1.39731988e-01 -3.41684163e-01 -4.68492240e-01 -2.26395369e-01
-4.14611623e-02 -4.26208824e-01 -2.35460490e-01 8.85881841e-01
8.44708383e-01 -4.17425901e-01 7.48841703e-01 -8.87937546e-01
-4.61543232e-01 -2.90111601e-02 -9.99556482e-01 6.78997636e-01
-5.03646396e-02 2.32543275e-01 6.26800582e-02 -8.84678423e-01
8.39956403e-01 8.82535636e-01 7.45071888e-01 7.88153708e-01
-9.14962292e-01 -4.62247044e-01 -1.21825837e-01 4.17836905e-02
-1.51222181e+00 -1.66954815e-01 4.78982806e-01 -4.89393800e-01
7.25819767e-01 -9.73473713e-02 8.86650145e-01 9.98371482e-01
6.01333857e-01 6.80344105e-01 1.17724812e+00 -2.52523303e-01
3.62596333e-01 -1.93679668e-02 4.18543629e-02 6.49888158e-01
2.69692481e-01 1.74313970e-02 -1.60452366e-01 9.81117934e-02
1.37750399e+00 -3.13279852e-02 -6.38723671e-02 -2.38558333e-02
-9.96572733e-01 4.29963589e-01 2.89680988e-01 6.48748398e-01
-3.95411700e-02 4.14289117e-01 5.52928865e-01 -3.62002939e-01
4.85516727e-01 -2.50129819e-01 -3.78806770e-01 -1.77054033e-02
-9.27014291e-01 1.41992956e-01 4.35645610e-01 3.08626860e-01
3.62210542e-01 -2.11162984e-01 -3.44293058e-01 6.31102145e-01
2.67237574e-01 1.55844539e-01 6.64287806e-01 -5.77591419e-01
-2.88343072e-01 9.63134170e-01 -4.86486703e-02 -6.75819397e-01
-6.96665764e-01 -3.87969226e-01 -1.11605346e+00 3.62841129e-01
8.63210320e-01 -2.10479647e-01 -1.31634867e+00 1.34759760e+00
6.24673367e-01 3.43657076e-01 -2.58196265e-01 1.11056387e+00
8.65552306e-01 5.91907859e-01 1.23827882e-01 -4.55921590e-01
1.22996521e+00 -5.78642905e-01 -6.28415346e-01 4.92200814e-02
6.36547208e-01 -6.50698543e-01 6.93181396e-01 2.25280970e-01
-8.31233442e-01 -2.79904425e-01 -8.62750530e-01 -1.78495154e-01
-7.95271456e-01 3.50522012e-01 6.34189487e-01 2.61977404e-01
-1.17970455e+00 4.36216086e-01 -9.47837472e-01 -6.78217471e-01
7.35243261e-01 4.06628460e-01 -1.81208134e-01 2.75893182e-01
-9.21518207e-01 4.67702329e-01 8.15312117e-02 9.17919949e-02
-1.05272603e+00 -5.55637658e-01 -4.15433437e-01 -1.06658779e-01
1.15242153e-01 -6.63802624e-01 1.21055818e+00 -5.10025680e-01
-1.43101323e+00 1.26238859e+00 -3.52425635e-01 -2.00991765e-01
8.06941926e-01 3.85267854e-01 1.36993274e-01 2.78686106e-01
1.19010337e-01 7.37011671e-01 1.35503471e-01 -1.11835539e+00
-6.92300737e-01 -6.84436858e-01 -2.61824548e-01 -1.92520134e-02
-5.10153994e-02 -9.37523767e-02 -8.79720569e-01 -3.98471922e-01
-1.63327008e-02 -5.64292550e-01 -1.46930486e-01 5.81952751e-01
-3.33909422e-01 -3.15849662e-01 1.10114503e+00 -5.19713402e-01
1.03924739e+00 -1.92435193e+00 -9.69933942e-02 -1.77900895e-01
3.57436538e-01 5.32371938e-01 6.91349506e-02 1.73548058e-01
1.56587213e-01 3.33278000e-01 -2.32616514e-02 -3.88330132e-01
-5.02536058e-01 -2.93020636e-01 3.75423133e-01 6.21873736e-01
9.61803794e-02 9.59600031e-01 -7.77063727e-01 -8.21176231e-01
2.14234784e-01 7.70884812e-01 1.24416508e-01 -7.87495151e-02
-1.54789373e-01 5.67323148e-01 -2.56116241e-01 1.13720453e+00
9.84620750e-01 -5.21487117e-01 -9.65723544e-02 -8.58930424e-02
-3.79650146e-01 -4.46240991e-01 -7.98067868e-01 1.26422977e+00
1.05457148e-02 7.42670000e-01 1.60772488e-01 -5.09686053e-01
7.14590907e-01 1.53480247e-01 6.04177475e-01 -7.90390849e-01
6.84196234e-01 3.35088611e-01 -2.93702960e-01 -5.47212899e-01
-1.25557957e-02 -1.65782347e-01 3.39223951e-01 1.94984432e-02
-1.22782484e-01 1.17091738e-01 4.01408345e-01 3.96212786e-02
9.88959849e-01 -1.32310212e-01 1.98416412e-01 -1.46978766e-01
7.53238738e-01 -1.54512510e-01 8.62463474e-01 3.24002147e-01
-7.71955907e-01 6.73971713e-01 7.32664526e-01 -7.43074954e-01
-9.80994999e-01 -1.03647208e+00 -1.36175022e-01 3.68614018e-01
5.04084349e-01 1.83248326e-01 -9.50865865e-01 -6.05369985e-01
-5.25071062e-02 -2.22909197e-01 -8.55945706e-01 1.18556157e-01
-2.95038670e-01 -9.75432992e-01 7.58382082e-01 6.20337546e-01
9.78826165e-01 -6.54200017e-01 -5.14349520e-01 1.93899676e-01
8.75659660e-02 -1.01446462e+00 -2.62531668e-01 8.80459175e-02
-7.83445776e-01 -1.30992484e+00 -1.06831777e+00 -9.94098544e-01
8.65200937e-01 3.34049106e-01 5.55372536e-01 3.95808578e-01
-8.87470961e-01 -2.00345427e-01 -1.76665023e-01 -4.76883620e-01
-6.42528906e-02 1.61171202e-02 -7.48897195e-02 -1.98568806e-01
6.38309002e-01 -4.23659861e-01 -1.02044237e+00 3.04926902e-01
-9.20001447e-01 1.30712211e-01 6.07876897e-01 8.31762969e-01
1.14025581e+00 6.29566237e-02 6.31050885e-01 -8.41201305e-01
2.82044888e-01 -2.02187151e-01 -7.94288456e-01 1.12579636e-01
-3.59806150e-01 -5.93774080e-01 7.68983841e-01 -5.85618556e-01
-8.88434768e-01 2.90224940e-01 -1.82500675e-01 -2.37176627e-01
-2.34433874e-01 3.76672268e-01 2.39252612e-01 -4.95074600e-01
5.19295990e-01 4.41489995e-01 1.66275322e-01 -2.23288804e-01
-5.61643839e-01 5.79114854e-01 4.63510275e-01 9.81214084e-03
2.99281329e-01 8.21299791e-01 3.56897026e-01 -7.82835245e-01
-5.01250684e-01 -4.48908925e-01 -5.26831090e-01 -6.27169371e-01
1.03156233e+00 -8.71753812e-01 -9.40745592e-01 1.29743707e+00
-1.17714679e+00 -4.82279688e-01 1.40175730e-01 2.40082756e-01
-2.71148056e-01 4.51188922e-01 -1.19528866e+00 -1.05947864e+00
-5.62707782e-01 -9.43867385e-01 1.23090160e+00 1.03166521e+00
3.04689735e-01 -1.07176661e+00 2.90534854e-01 2.31344074e-01
3.90095055e-01 5.93742311e-01 7.58481860e-01 -2.64389247e-01
-6.98049128e-01 -3.81997585e-01 -4.43340212e-01 -1.45811439e-01
1.71675131e-01 7.30333626e-01 -1.16032183e+00 -1.99579224e-01
-4.11197424e-01 -2.00819209e-01 1.12183177e+00 9.87701893e-01
1.31259966e+00 1.99740946e-01 -8.20437908e-01 6.95939362e-01
1.76312041e+00 4.61251408e-01 9.23723638e-01 1.97312862e-01
3.85850012e-01 2.35194594e-01 5.22342563e-01 2.32679561e-01
1.45001918e-01 1.59207180e-01 5.02637148e-01 -6.64512634e-01
-1.21359237e-01 -5.79214096e-02 -5.08000292e-02 5.20710886e-01
-1.79288492e-01 -4.13875282e-01 -9.69263792e-01 8.30592215e-01
-1.83723819e+00 -6.21796906e-01 -2.49297336e-01 2.00312281e+00
7.42912352e-01 2.35547543e-01 2.64636129e-01 1.30347595e-01
1.03300011e+00 -2.77074724e-01 -9.37245548e-01 1.31117580e-02
-2.14817718e-01 4.56744023e-02 5.24392068e-01 1.82429656e-01
-1.17778027e+00 8.69882345e-01 6.34202576e+00 9.63599920e-01
-1.24887192e+00 -1.39987215e-01 1.22294414e+00 1.67566910e-01
1.83068827e-01 -2.08054200e-01 -9.40673411e-01 6.30474746e-01
2.49134481e-01 -4.80428003e-02 2.83490424e-03 4.91884083e-01
2.78884888e-01 -7.42624402e-01 -8.66985977e-01 8.73796046e-01
-2.40848452e-01 -1.63773358e+00 -3.83030362e-02 1.16196036e-01
6.01300001e-01 -2.91222692e-01 -4.78207320e-02 5.25846593e-02
-1.63594231e-01 -1.03447258e+00 2.33866349e-02 6.75311685e-01
1.04635167e+00 -7.88064063e-01 1.35801530e+00 5.93677163e-01
-1.28218353e+00 2.35810041e-01 -4.06538486e-01 -8.91918242e-02
1.30594522e-01 7.69912541e-01 -8.70494783e-01 9.63013172e-02
6.51294827e-01 5.34917951e-01 -5.34496963e-01 1.43446839e+00
3.40522021e-01 3.05668116e-01 -4.15080637e-01 -5.80759764e-01
1.14017591e-01 3.79275642e-02 2.68833607e-01 1.24281490e+00
3.57588619e-01 -6.46893829e-02 -1.51853397e-01 9.41020072e-01
-1.60905123e-01 -3.64766009e-02 -6.05522156e-01 -2.51626708e-02
5.18103123e-01 1.58363342e+00 -1.47780085e+00 -1.72695424e-02
-1.18067876e-01 8.75421107e-01 2.82705426e-01 1.38494328e-01
-8.12587380e-01 -5.28048098e-01 2.40898177e-01 3.45246315e-01
1.19921044e-01 -7.62574822e-02 -3.68411511e-01 -7.60329187e-01
-3.66725415e-01 -8.08536112e-02 3.27923656e-01 -5.99924862e-01
-1.43605947e+00 6.32028952e-02 -5.10929942e-01 -8.18354487e-01
3.44065547e-01 -8.11127722e-01 -9.88832414e-01 6.69219553e-01
-1.49153829e+00 -1.11778104e+00 -7.17367828e-01 3.45919341e-01
7.48176634e-01 1.86644584e-01 5.11198759e-01 3.09311122e-01
-9.33211267e-01 6.83062017e-01 -8.07015076e-02 2.67621458e-01
6.19688451e-01 -1.27629471e+00 1.14691474e-01 7.80593634e-01
-7.21307933e-01 3.77942801e-01 3.16608191e-01 -8.46546292e-01
-9.99803483e-01 -9.97723103e-01 7.06873298e-01 -1.04248896e-01
2.43988797e-01 -2.92499095e-01 -7.67316103e-01 1.03052579e-01
2.60634050e-02 2.47331768e-01 7.48289764e-01 -5.92570901e-01
2.30649173e-01 -2.26511419e-01 -1.43008661e+00 7.29975760e-01
7.23263144e-01 -1.19231336e-01 2.39171043e-01 2.09983617e-01
5.16074359e-01 -7.08951592e-01 -7.40076065e-01 4.44427043e-01
7.07058609e-01 -9.40519869e-01 6.55200303e-01 1.15313664e-01
4.80371326e-01 -6.44632697e-01 4.58926350e-01 -8.44181538e-01
-3.66061658e-01 -1.34606272e-01 -1.61829159e-01 1.09483635e+00
2.00607255e-01 -3.01042438e-01 1.23088121e+00 1.51365614e-02
8.35665539e-02 -1.17808068e+00 -1.27902031e+00 -5.83752573e-01
2.26482496e-01 3.78830492e-01 1.55453861e-01 7.02238142e-01
-1.31922886e-01 3.72010082e-01 2.26108693e-02 -1.22443534e-01
6.29278719e-01 -3.05885494e-01 5.70468426e-01 -1.30542517e+00
2.31741220e-01 -4.76316839e-01 -7.01475680e-01 -8.48334670e-01
-2.66092688e-01 -4.07529384e-01 -6.38084486e-02 -1.61943173e+00
5.36175668e-01 4.55550030e-02 -2.26163387e-01 2.72235990e-01
-2.36052945e-01 4.21480954e-01 -3.37871313e-01 -1.13649398e-01
-5.85204720e-01 1.86631545e-01 1.59860730e+00 -1.39200613e-01
-9.42317545e-02 -2.07556739e-01 -3.99535686e-01 7.63042212e-01
9.24591899e-01 -1.12717956e-01 -2.20308095e-01 -1.73589930e-01
1.45239532e-01 -6.95057884e-02 5.34658909e-01 -1.25340462e+00
7.61591077e-01 -7.43857920e-02 1.15944290e+00 -1.03160083e+00
1.73738182e-01 -3.92534435e-01 2.78380862e-03 6.93909943e-01
-4.29611243e-02 -3.13645244e-01 3.24474126e-01 7.10121572e-01
-1.63759679e-01 3.48795019e-02 1.26659453e+00 -4.57823128e-01
-5.10367334e-01 5.06791294e-01 -6.45196497e-01 -2.41289824e-01
1.53646469e+00 -6.42449617e-01 -7.63952553e-01 8.12398493e-02
-4.61235970e-01 4.49700117e-01 7.51995444e-01 -2.37131983e-01
8.16023111e-01 -1.26096213e+00 -3.89548212e-01 -2.19395719e-02
2.13913769e-02 3.10946882e-01 6.28685057e-01 9.60817933e-01
-1.07863295e+00 2.25027144e-01 -3.73859078e-01 -8.14198732e-01
-1.31155801e+00 2.73751616e-01 7.79699028e-01 -5.56364775e-01
-1.38293132e-01 1.11928737e+00 2.59376913e-01 -1.08394735e-01
2.26096138e-01 -2.22977966e-01 -4.35265541e-01 -2.45761424e-01
4.99732643e-01 3.79544646e-01 -1.11174114e-01 -2.08047137e-01
-5.44283450e-01 8.48988712e-01 -4.42179263e-01 3.04491937e-01
1.17831099e+00 -4.68142442e-02 -3.04893702e-01 6.04401529e-01
9.05503333e-01 -4.06636924e-01 -1.39114559e+00 3.03469867e-01
-4.39254850e-01 -3.33312303e-01 9.96589959e-02 -9.28675413e-01
-1.14163649e+00 9.78148580e-01 9.68918920e-01 1.19138204e-01
1.26049602e+00 -4.02238697e-01 1.01798570e+00 2.91679669e-02
4.21560258e-01 -1.46902394e+00 1.21761963e-01 4.55194831e-01
1.86930224e-01 -1.31108367e+00 -7.10885674e-02 -6.37437820e-01
-1.91083446e-01 1.36835921e+00 1.28930449e+00 -1.27980769e-01
6.08157277e-01 5.66648543e-01 -1.68505020e-03 -2.96636283e-01
-9.29506063e-01 -9.95425582e-02 -3.42494279e-01 7.22799420e-01
4.67436641e-01 -3.38799238e-01 -7.67619014e-01 7.14902520e-01
6.44776464e-01 5.94131291e-01 6.28777146e-01 8.86851311e-01
-5.43945909e-01 -6.37782812e-01 -1.48120020e-02 4.67894912e-01
-8.09289217e-01 2.91902393e-01 -8.78611445e-01 9.55593050e-01
3.76549453e-01 5.28948843e-01 2.23027810e-01 -4.34389979e-01
-2.87029557e-02 -3.41530859e-01 5.07174850e-01 -2.09492549e-01
-3.81493956e-01 3.19215238e-01 -3.47627252e-01 -3.84128571e-01
-3.05573672e-01 -3.27664256e-01 -1.62875772e+00 -3.67931485e-01
-4.05801177e-01 -2.83213407e-01 5.32373726e-01 6.49780095e-01
3.42583597e-01 5.68873823e-01 4.92593944e-01 -7.52678692e-01
1.29966870e-01 -7.98698485e-01 -7.40925968e-01 -1.31536916e-01
4.31095004e-01 -6.40674472e-01 -4.12529409e-01 2.06411421e-01]
|
[14.724593162536621, -3.2014238834381104]
|
812afb71-82ae-4d91-b8f3-d4aec7579b18
|
two-headed-eye-segmentation-approach-for
|
2209.15471
| null |
https://arxiv.org/abs/2209.15471v1
|
https://arxiv.org/pdf/2209.15471v1.pdf
|
Two-headed eye-segmentation approach for biometric identification
|
Iris-based identification systems are among the most popular approaches for person identification. Such systems require good-quality segmentation modules that ideally identify the regions for different eye components. This paper introduces the new two-headed architecture, where the eye components and eyelashes are segmented using two separate decoding modules. Moreover, we investigate various training scenarios by adopting different training losses. Thanks to the two-headed approach, we were also able to examine the quality of the model with the convex prior, which enforces the convexity of the segmented shapes. We conducted an extensive evaluation of various learning scenarios on real-life conditions high-resolution near-infrared iris images.
|
['Christian Brendel', 'Tobias Zillig', 'Tanguy Jeanneau', 'Maciej Zieba', 'Wiktor Lazarski']
|
2022-09-30
| null | null | null | null |
['person-identification']
|
['computer-vision']
|
[ 1.79193512e-01 1.17035776e-01 -1.26715779e-01 -5.37447333e-01
-3.97210121e-01 -4.69462484e-01 4.28784519e-01 -1.94937170e-01
-5.68709433e-01 6.16011977e-01 -1.99526995e-01 -1.94181383e-01
-2.36791492e-01 -3.29025030e-01 -4.87293482e-01 -7.18602657e-01
1.89302176e-01 3.57243985e-01 -2.58095503e-01 1.43263534e-01
3.37541640e-01 5.28679013e-01 -1.87594151e+00 -7.34250844e-02
1.24528801e+00 9.44964528e-01 -5.78198433e-01 8.99397433e-01
2.31906280e-01 -9.39342603e-02 -5.47850966e-01 -7.82173336e-01
6.98045373e-01 -2.27639630e-01 -6.41771615e-01 4.84789312e-01
1.12967503e+00 -5.31551600e-01 1.02927171e-01 1.20096636e+00
5.70872664e-01 -9.65196863e-02 4.39349025e-01 -8.51747692e-01
-2.51482368e-01 8.26352835e-02 -8.17293346e-01 1.37313128e-01
4.97401148e-01 4.15108562e-01 7.51282692e-01 -4.04614300e-01
2.60325164e-01 1.01478994e+00 6.83735549e-01 7.89080679e-01
-1.27830434e+00 -5.06591737e-01 -1.03412330e-01 9.39748250e-03
-1.50859046e+00 -5.78289509e-01 3.88114125e-01 -5.14343977e-01
4.91174698e-01 3.53844941e-01 5.81555426e-01 6.12489045e-01
-1.90701053e-01 7.95736015e-01 1.60725427e+00 -6.52311146e-01
-2.19080985e-01 6.45626485e-01 4.16731834e-01 6.65413737e-01
3.72335732e-01 5.37783325e-01 -2.24930957e-01 1.54929850e-02
7.86715388e-01 -3.26597214e-01 -3.37933719e-01 -8.60100687e-02
-6.80358350e-01 4.00269538e-01 1.81890517e-01 2.71620035e-01
-1.71825781e-01 -4.01273072e-01 5.14312796e-02 -5.39076701e-03
4.28547651e-01 4.18362319e-01 -1.27531677e-01 1.54300347e-01
-1.35754967e+00 -9.06922147e-02 6.29194736e-01 7.92931259e-01
5.61848402e-01 -3.48680615e-01 -4.46026713e-01 5.76532900e-01
5.30745089e-01 2.78070569e-01 2.39139795e-01 -4.96403843e-01
2.74122417e-01 7.94623256e-01 2.71916687e-01 -4.25186604e-01
-4.37281042e-01 -6.64418280e-01 -7.19648421e-01 3.88813704e-01
9.13576305e-01 -4.61732864e-01 -1.18762112e+00 1.19988966e+00
3.12976748e-01 4.26729113e-01 -1.26385808e-01 1.08694899e+00
7.33450294e-01 -4.14653793e-02 -3.46817658e-03 4.03038375e-02
1.42689216e+00 -7.62703180e-01 -4.79877204e-01 1.05091184e-01
1.94205418e-01 -8.50614607e-01 8.07498276e-01 5.98798633e-01
-1.13026452e+00 -7.19588995e-01 -7.87728846e-01 -4.30169627e-02
-1.86138242e-01 8.45608294e-01 4.86268699e-01 1.44851744e+00
-1.25096452e+00 3.88940871e-01 -5.32876372e-01 -3.22472095e-01
4.23316061e-01 8.65795314e-01 -2.88055360e-01 2.99606830e-01
-7.01001585e-01 6.88810050e-01 1.60132647e-01 3.64623576e-01
-2.19409361e-01 -4.99761164e-01 -8.09878945e-01 -8.54144990e-02
-1.71081554e-02 -5.67470193e-01 9.18970346e-01 -1.16814625e+00
-1.66154051e+00 1.57192528e+00 -5.05848646e-01 -4.84822929e-01
6.13057137e-01 -1.55922547e-01 -2.63224065e-01 2.04609066e-01
-4.73926246e-01 3.85198504e-01 1.11035132e+00 -1.25626469e+00
-5.73730946e-01 -7.01743543e-01 -8.45687091e-03 1.07085750e-01
-1.46036997e-01 4.73760784e-01 -6.29679143e-01 -3.67787123e-01
-1.99277610e-01 -8.44067633e-01 4.12153117e-02 -1.57454282e-01
-8.11055422e-01 -2.67724488e-02 2.47318611e-01 -8.90469134e-01
1.17343974e+00 -2.21981621e+00 -4.37498093e-02 5.85273266e-01
2.69841015e-01 6.85397089e-01 1.44888684e-01 -4.15136129e-01
-1.29300848e-01 8.05780515e-02 -1.01318471e-01 -8.67895842e-01
6.96426705e-02 -8.08906257e-02 2.69476529e-02 6.50606573e-01
1.15828097e-01 6.54254496e-01 -2.02856690e-01 -4.66740578e-01
4.02100086e-01 6.54744029e-01 -2.88371295e-01 2.10650146e-01
1.54684171e-01 7.41693974e-01 -1.44665465e-01 7.92894185e-01
9.30528462e-01 -7.64344409e-02 -1.53444842e-01 -4.98799756e-02
-1.62937000e-01 -2.06078663e-01 -1.30718267e+00 1.08423448e+00
-1.43423542e-01 5.19540608e-01 3.52312148e-01 -5.12049615e-01
7.34247446e-01 3.44255239e-01 4.36979532e-01 -6.21304452e-01
3.37874949e-01 1.52291998e-01 4.50846627e-02 -2.77892143e-01
4.39076692e-01 -2.02381715e-01 4.06164855e-01 6.92194328e-02
-1.25200916e-02 2.07296148e-01 1.61569819e-01 -3.24750602e-01
1.46313608e-01 1.14514627e-01 1.77072987e-01 -1.60209224e-01
9.81434226e-01 -3.22342783e-01 3.75391692e-01 5.16805947e-01
-4.64773089e-01 7.98326552e-01 3.49188328e-01 -3.77179921e-01
-7.80216932e-01 -7.50063181e-01 -8.02639127e-01 5.53812444e-01
1.88607126e-01 1.05706073e-01 -1.22661138e+00 -5.07067919e-01
3.71004385e-03 3.41739833e-01 -5.76840580e-01 3.42284650e-01
-3.11875701e-01 -1.04389513e+00 7.35901713e-01 7.26137757e-02
4.28095907e-01 -5.25597095e-01 -5.06151855e-01 -2.95101643e-01
2.15044886e-01 -1.11570930e+00 -4.24736202e-01 -2.92057544e-01
-7.12585330e-01 -1.54530644e+00 -1.00229692e+00 -6.69059277e-01
1.05495608e+00 -3.63394290e-01 9.68926311e-01 3.68461788e-01
-5.84085584e-01 4.18264836e-01 8.12956616e-02 -4.08965439e-01
-1.30905449e-01 -4.29798514e-02 1.75830990e-01 6.96042240e-01
6.62641823e-01 1.39890924e-01 -7.30367362e-01 4.19727743e-01
-4.98788953e-01 -2.59070039e-01 4.73849833e-01 6.20952725e-01
5.97911656e-01 1.04947262e-01 -8.80383030e-02 -7.17574239e-01
4.72394943e-01 4.43604141e-02 -1.16039515e+00 4.95949298e-01
-7.34854281e-01 -5.50484955e-02 4.51002240e-01 -2.62586802e-01
-1.10942709e+00 1.50544673e-01 -2.55410105e-01 -2.31415346e-01
-7.00853527e-01 -2.88112819e-01 -1.19748145e-01 -6.79253221e-01
5.31364202e-01 1.14142582e-01 1.11998051e-01 -6.17784798e-01
2.26829290e-01 1.04262853e+00 5.71628213e-01 -5.19091427e-01
8.19507480e-01 4.56244588e-01 -1.92332305e-02 -8.96829545e-01
-6.32585585e-01 -7.84956157e-01 -8.77967417e-01 -1.89506248e-01
9.05700088e-01 -7.13547707e-01 -1.24165094e+00 1.09512722e+00
-9.29288983e-01 -1.47586986e-01 -1.41810045e-01 6.45512879e-01
-3.83283824e-01 5.46801686e-01 -5.92289686e-01 -1.12122297e+00
-4.78385806e-01 -1.25724697e+00 1.13735545e+00 9.97466326e-01
1.77465230e-01 -1.12630248e+00 3.39271687e-02 6.43715680e-01
2.30609924e-01 5.53477108e-02 5.03420830e-01 -4.87068534e-01
-4.95626032e-01 -2.69690484e-01 -4.21904474e-01 3.90472084e-01
2.13654935e-02 3.13823283e-01 -1.41506696e+00 -5.27488232e-01
-1.57840699e-01 -2.98477765e-02 8.73300254e-01 7.67899394e-01
1.13769042e+00 -1.68659136e-01 -2.94026107e-01 1.13766861e+00
1.48501885e+00 -7.08027035e-02 8.65798056e-01 1.73567563e-01
6.69911921e-01 8.36707771e-01 4.12413985e-01 2.90796965e-01
2.16955826e-01 7.89972246e-01 1.88829422e-01 -7.72695720e-01
-8.86509195e-02 1.98281288e-01 -1.75091587e-02 -6.13694042e-02
-4.33555663e-01 -5.91631047e-02 -8.99535179e-01 5.24824321e-01
-1.35915995e+00 -7.68816292e-01 -2.13482276e-01 2.66589499e+00
9.08171356e-01 -1.03817217e-01 4.86364067e-01 -9.23760002e-04
8.71777952e-01 -3.18726897e-01 -6.04498208e-01 -3.37715209e-01
-2.13189214e-01 4.96661723e-01 7.09996104e-01 6.98120832e-01
-1.37261367e+00 7.50159860e-01 7.34529972e+00 4.49738592e-01
-1.16505158e+00 -3.71618271e-01 8.89913261e-01 -1.97506398e-01
1.90477580e-01 -2.07158327e-01 -1.36993992e+00 5.27015388e-01
9.01894212e-01 1.26531139e-01 3.69465709e-01 2.90322572e-01
1.80043429e-01 -4.25011724e-01 -9.32376742e-01 9.13058758e-01
5.70153147e-02 -9.53028798e-01 -3.52421135e-01 3.01641077e-01
6.20427430e-01 -1.87335730e-01 4.38681215e-01 -3.00630212e-01
-9.28353071e-02 -1.33933902e+00 1.34530723e-01 9.99033034e-01
1.10588241e+00 -5.30574858e-01 7.77702689e-01 9.57163349e-02
-9.51847970e-01 -4.93134484e-02 -1.60640433e-01 2.64781058e-01
-8.49713385e-02 3.96589905e-01 -7.09606290e-01 6.23614430e-01
4.55342203e-01 4.71020550e-01 -9.16305900e-01 1.67562544e+00
-7.10328147e-02 5.78029215e-01 -4.57493126e-01 4.20965552e-01
-1.99991927e-01 -4.93685693e-01 6.17506802e-01 1.05900180e+00
-6.63194060e-02 -1.12967134e-01 -1.56451046e-01 1.07162678e+00
1.31096601e-01 8.25752020e-02 -1.69494450e-01 2.55689323e-01
-8.28614160e-02 1.17865837e+00 -3.42731655e-01 -3.12067270e-02
-5.68906188e-01 7.31444538e-01 -1.26670986e-01 4.59321588e-01
-5.27247608e-01 -2.47256309e-01 8.45157266e-01 3.89571667e-01
2.18147654e-02 7.59716108e-02 -5.48216045e-01 -1.03727913e+00
4.21588495e-02 -9.82719183e-01 5.09405434e-01 -3.64112705e-01
-1.13097620e+00 5.74060380e-01 -1.72251970e-01 -9.99979913e-01
-1.21099383e-01 -8.74475360e-01 -6.16217375e-01 1.50523913e+00
-2.02024436e+00 -1.21973813e+00 -2.64352083e-01 7.30981529e-01
-1.50848059e-02 -2.55745858e-01 6.04825675e-01 4.04591590e-01
-1.18829811e+00 1.18851781e+00 2.63154767e-02 3.90027910e-01
8.04899514e-01 -1.45505822e+00 2.79790401e-01 1.10363829e+00
1.09348729e-01 9.55542386e-01 4.77394134e-01 -3.30572307e-01
-8.45226765e-01 -6.70256972e-01 7.08427548e-01 -3.99735391e-01
9.36820135e-02 1.90896839e-02 -8.02922428e-01 5.67672014e-01
3.09115082e-01 -1.96765009e-02 8.04107904e-01 2.67938226e-01
-6.36860430e-02 -8.09237510e-02 -1.41728115e+00 2.15960607e-01
3.73561293e-01 -5.44095337e-01 -2.70395368e-01 2.19698817e-01
-9.83246714e-02 -6.26096904e-01 -9.83666599e-01 3.37553829e-01
5.17692804e-01 -1.22000015e+00 1.00311089e+00 -5.48002124e-01
-6.03890345e-02 -3.23213071e-01 4.68781501e-01 -8.10017765e-01
1.85660020e-01 -8.43274355e-01 -5.06722406e-02 1.15932858e+00
4.19356018e-01 -9.09783304e-01 1.00255859e+00 1.11096382e+00
4.41267043e-01 -6.85666323e-01 -7.82048225e-01 -5.25324523e-01
-1.39257396e-02 9.33107585e-02 5.25289595e-01 5.06472707e-01
-2.85213470e-01 -1.90687358e-01 -3.09654087e-01 5.34676790e-01
9.44673955e-01 4.52767164e-02 7.49267280e-01 -1.36412716e+00
-3.63124549e-01 -5.72091639e-01 -5.29204547e-01 -9.52307105e-01
1.50356606e-01 -4.07237083e-01 -2.78548270e-01 -9.81542885e-01
2.29023978e-01 -4.10154134e-01 -2.19642162e-01 4.64222878e-01
-4.45114523e-01 5.54487586e-01 -2.09768396e-02 1.45223245e-01
-2.07501754e-01 -1.11140355e-01 1.06966770e+00 -5.07249832e-02
-3.45546782e-01 7.09232032e-01 -6.07763588e-01 7.33530283e-01
8.53652894e-01 9.75476503e-02 -1.46516159e-01 -2.37248376e-01
-5.79970628e-02 -2.92085171e-01 5.94924271e-01 -9.08103526e-01
4.50584888e-01 1.76683545e-01 4.97297347e-01 -3.16477984e-01
2.28263333e-01 -7.54693210e-01 -1.48099467e-01 1.97313532e-01
-1.32759050e-01 -5.23892403e-01 4.09870774e-01 1.31873280e-01
-2.43746966e-01 -2.16894791e-01 1.30475044e+00 1.07081242e-01
-4.36451733e-01 3.84637892e-01 2.48167709e-01 -8.67883712e-02
1.06170225e+00 -8.38523448e-01 -2.11883336e-01 -1.50706321e-02
-9.35567677e-01 2.65459865e-01 7.60254920e-01 3.68212722e-02
4.14034635e-01 -5.26203573e-01 -8.06102276e-01 8.98856521e-01
-1.56743992e-02 -1.67024016e-01 -8.45157076e-03 1.14258623e+00
-4.47434694e-01 5.26514947e-01 -2.77483106e-01 -7.50456274e-01
-1.87459648e+00 4.74580705e-01 1.14044452e+00 -1.28674194e-01
-4.61378515e-01 1.05327702e+00 -2.52765883e-02 -3.98479939e-01
5.02951443e-01 -1.74417719e-01 -5.42363703e-01 2.40077674e-02
6.82786644e-01 4.59301740e-01 8.07900280e-02 -9.16837871e-01
-3.08450669e-01 1.02216482e+00 -1.96576640e-01 2.01260239e-01
6.95081592e-01 -3.05635214e-01 -1.21319279e-01 -1.94514066e-01
6.83275282e-01 1.98467076e-01 -1.31280637e+00 -1.69643611e-01
-6.42754212e-02 -7.40683079e-01 1.53164998e-01 -8.83330584e-01
-1.05861020e+00 8.30703020e-01 9.66596603e-01 7.24493638e-02
1.46949613e+00 -5.05978703e-01 5.27748346e-01 -2.93845087e-01
2.40044698e-01 -1.07589650e+00 -6.59030199e-01 2.90639829e-02
2.97412097e-01 -1.41456342e+00 7.67191276e-02 -6.17841005e-01
-4.51233119e-01 1.07372320e+00 4.57997292e-01 1.89303026e-01
4.84307259e-01 1.17873773e-01 4.43283647e-01 -5.77218533e-02
1.91540748e-03 -6.68297768e-01 1.14237976e+00 6.12934947e-01
5.06086290e-01 6.09179325e-02 -2.49265090e-01 2.99703360e-01
-2.02448338e-01 -9.50696543e-02 2.89529771e-01 1.80279240e-01
-3.96229178e-02 -1.39177895e+00 -6.17778540e-01 5.16249597e-01
-7.51409113e-01 -9.80802476e-02 -6.87680542e-01 5.96497715e-01
2.45629430e-01 1.06427717e+00 9.01370049e-02 -1.83039069e-01
2.39721835e-01 2.72514485e-02 6.31630301e-01 -5.37803054e-01
-9.90321517e-01 -9.99262847e-05 -1.20669469e-01 -6.40298903e-01
-5.72488964e-01 -7.68202126e-01 -9.67632353e-01 -4.32510853e-01
-5.36703408e-01 -1.90875769e-01 5.43794811e-01 8.86214733e-01
3.43283683e-01 1.45495102e-01 6.15172982e-01 -4.84004617e-01
-5.08316755e-01 -7.26157308e-01 -8.53462756e-01 3.61294150e-01
8.03171575e-01 -3.29480886e-01 -3.48648906e-01 6.90690279e-02]
|
[3.7445249557495117, -3.6309854984283447]
|
f69acabb-d075-4d5d-a13a-a8a082d4ce01
|
low-light-image-and-video-enhancement-via
|
2203.04889
| null |
https://arxiv.org/abs/2203.04889v1
|
https://arxiv.org/pdf/2203.04889v1.pdf
|
Low-light Image and Video Enhancement via Selective Manipulation of Chromaticity
|
Image acquisition in low-light conditions suffers from poor quality and significant degradation in visual aesthetics. This affects the visual perception of the acquired image and the performance of various computer vision and image processing algorithms applied after acquisition. Especially for videos, the additional temporal domain makes it more challenging, wherein we need to preserve quality in a temporally coherent manner. We present a simple yet effective approach for low-light image and video enhancement. To this end, we introduce "Adaptive Chromaticity", which refers to an adaptive computation of image chromaticity. The above adaptivity allows us to avoid the costly step of low-light image decomposition into illumination and reflectance, employed by many existing techniques. All stages in our method consist of only point-based operations and high-pass or low-pass filtering, thereby ensuring that the amount of temporal incoherence is negligible when applied on a per-frame basis for videos. Our results on standard lowlight image datasets show the efficacy of our algorithm and its qualitative and quantitative superiority over several state-of-the-art techniques. For videos captured in the wild, we perform a user study to demonstrate the preference for our method in comparison to state-of-the-art approaches.
|
['Matthias Trapp', 'Jürgen Döllner', 'Sebastian Pasewaldt', 'Amir Semmo', 'Max Reimann', 'Sumit Shekhar']
|
2022-03-09
| null | null | null | null |
['video-enhancement']
|
['computer-vision']
|
[ 5.53378105e-01 -6.11454487e-01 3.48464221e-01 -2.02399239e-01
-2.31877670e-01 -5.83302438e-01 5.15432537e-01 1.32298559e-01
-7.02430010e-01 5.94575107e-01 -2.27664575e-01 -4.56341803e-02
1.43061783e-02 -7.07684815e-01 -5.51962495e-01 -1.08843315e+00
1.10751070e-01 -5.31877637e-01 3.48613828e-01 -2.45619193e-01
2.15172991e-01 5.33106387e-01 -2.02478242e+00 2.02076644e-01
7.11699545e-01 1.00179303e+00 3.20866287e-01 6.72040403e-01
3.09736207e-02 6.22002959e-01 -3.76412213e-01 -4.22776341e-01
4.96989608e-01 -4.74299639e-01 -4.06057149e-01 4.09686089e-01
5.72092891e-01 -4.71952975e-01 -1.51465014e-01 1.30076396e+00
4.95466143e-01 3.19206774e-01 4.15929914e-01 -8.72206151e-01
-4.00995106e-01 -2.77420461e-01 -7.22161233e-01 2.30463833e-01
5.53104997e-01 5.35932720e-01 6.44031763e-01 -8.78281474e-01
5.16493976e-01 7.35222280e-01 4.31651652e-01 1.91261545e-01
-1.41635752e+00 -4.23503458e-01 -1.66974410e-01 4.15219158e-01
-1.25381553e+00 -6.58012271e-01 8.99010837e-01 -3.25846523e-01
6.98626280e-01 2.94587553e-01 8.81658077e-01 5.72396815e-01
2.97269732e-01 4.72221933e-02 1.47732496e+00 -7.37171888e-01
2.66520411e-01 1.29176125e-01 -2.46129811e-01 6.11737907e-01
3.78549874e-01 4.06713307e-01 -5.38812816e-01 2.60112047e-01
5.99047780e-01 2.21517142e-02 -5.45088887e-01 -3.16987693e-01
-1.10651803e+00 3.00916731e-01 3.56386840e-01 3.75204653e-01
-4.78514642e-01 1.79280207e-01 8.62378627e-02 1.67199612e-01
4.84185934e-01 3.19018573e-01 1.04697615e-01 -8.28720853e-02
-1.06611764e+00 -1.78217113e-01 3.03629488e-01 5.41322649e-01
8.08146954e-01 -7.69835263e-02 -1.37143105e-01 7.91970134e-01
1.14381276e-01 5.61643362e-01 7.55254179e-02 -1.08650172e+00
-3.90898734e-02 2.04054609e-01 4.22228634e-01 -1.03423536e+00
-3.18998575e-01 -3.46658081e-01 -8.09335887e-01 8.35737944e-01
5.85430026e-01 1.71138406e-01 -6.46087825e-01 1.46469200e+00
3.77333462e-01 7.47449249e-02 2.53128242e-02 1.15194690e+00
5.21986842e-01 7.90517032e-01 -5.21302819e-02 -8.41221750e-01
1.25712740e+00 -7.17024744e-01 -9.27434087e-01 3.10847163e-02
-3.57728302e-02 -1.29255688e+00 1.35896385e+00 7.06544876e-01
-1.53968608e+00 -6.16948605e-01 -1.00384820e+00 -1.02687880e-01
-1.06208853e-01 2.47526884e-01 4.41461653e-01 8.44082355e-01
-9.88615155e-01 5.64923108e-01 -4.41751420e-01 -3.14216197e-01
5.63985631e-02 1.42979264e-01 -3.11986685e-01 -2.72429824e-01
-7.98483491e-01 9.02210534e-01 1.58617601e-01 2.59609669e-01
-4.30677384e-01 -6.17696941e-01 -6.04991198e-01 -8.57609510e-03
4.57389116e-01 -5.82265675e-01 9.72405016e-01 -1.09332478e+00
-1.82008898e+00 9.81747389e-01 -2.58397639e-01 -2.14718163e-01
4.23522681e-01 -1.65125728e-01 -4.30770934e-01 6.37516916e-01
-3.28163981e-01 2.29143098e-01 1.21530068e+00 -1.53322136e+00
-4.53816891e-01 7.86876157e-02 3.84975880e-01 2.88067430e-01
-3.59240294e-01 1.33256346e-01 -7.62158453e-01 -5.50345540e-01
2.12703869e-02 -7.90879726e-01 2.84882467e-02 3.39542240e-01
9.49635282e-02 2.71944433e-01 7.77432323e-01 -5.03009617e-01
1.28076875e+00 -2.28973365e+00 -1.71298429e-01 1.25607953e-03
1.06812723e-01 3.50897640e-01 -5.84963225e-02 3.31676990e-01
-7.63171464e-02 -3.54508460e-01 -3.55769694e-01 -2.45971277e-01
-3.80837440e-01 -1.01832617e-02 1.77477747e-02 8.67503762e-01
8.29988420e-02 4.19998795e-01 -9.88070488e-01 -5.46415091e-01
7.82898068e-01 7.87498295e-01 -5.19850254e-01 2.82615751e-01
2.99319476e-02 5.39369166e-01 1.13084845e-01 6.31762028e-01
8.89637113e-01 1.37219772e-01 -9.93379019e-03 -7.92326987e-01
-5.85927784e-01 -1.54112473e-01 -1.16982079e+00 1.65378928e+00
-7.16898739e-01 7.19377935e-01 1.64889395e-01 -4.77592647e-01
6.71380103e-01 1.19359761e-01 6.01369917e-01 -1.08694172e+00
2.08089605e-01 2.77134299e-01 -1.49185926e-01 -6.83154762e-01
7.16126621e-01 -4.03154671e-01 4.47073519e-01 3.87621164e-01
-2.65151054e-01 -2.54961282e-01 5.03072202e-01 3.28547992e-02
6.48425102e-01 3.75611931e-01 3.33474904e-01 -2.50311613e-01
7.25185335e-01 -2.99149305e-01 2.41239443e-01 4.83924657e-01
-1.14970505e-01 7.01109469e-01 3.31900939e-02 -1.22515611e-01
-9.70267951e-01 -1.04846764e+00 -1.06843717e-01 8.40048254e-01
5.11452794e-01 -3.19662780e-01 -7.73600578e-01 -1.19791508e-01
-4.85109091e-01 6.11868978e-01 -3.36105585e-01 -1.68097958e-01
-4.61237729e-01 -6.98025703e-01 -1.42107338e-01 -3.47862579e-02
6.86169684e-01 -8.71978283e-01 -1.14764559e+00 7.01271966e-02
-3.19503397e-01 -1.27292931e+00 -3.62802625e-01 -1.11832149e-01
-7.29327559e-01 -1.09830701e+00 -7.23535240e-01 -3.64475697e-01
7.52054393e-01 7.79713392e-01 1.10483432e+00 1.28674746e-01
-6.13729298e-01 5.03548563e-01 -4.74971026e-01 -4.96834479e-02
-1.95618495e-01 -5.99900723e-01 -4.09598388e-02 4.39148694e-01
6.27869964e-02 -5.11064649e-01 -1.16487288e+00 2.97646582e-01
-1.16749561e+00 8.78146514e-02 5.90626597e-01 7.57517159e-01
4.81140196e-01 5.22738814e-01 -1.49851471e-01 -5.18483698e-01
4.25155818e-01 2.75172323e-01 -8.39256644e-01 1.47228315e-01
-6.22124732e-01 -2.45888904e-01 7.29016125e-01 -3.63200098e-01
-1.34343314e+00 4.76968735e-02 1.06624141e-01 -3.37780744e-01
-9.10889581e-02 2.52220929e-02 -5.39818034e-02 -5.52421033e-01
6.82809889e-01 2.56007642e-01 -2.27969196e-02 -3.11805815e-01
4.26093847e-01 4.49309617e-01 6.90836370e-01 -2.64262557e-01
9.26153660e-01 8.83660376e-01 4.35477972e-01 -1.19545221e+00
-6.45397365e-01 -4.97908354e-01 -6.40324712e-01 -8.11826229e-01
9.85107958e-01 -6.46527231e-01 -1.14962757e+00 5.04627824e-01
-1.06548321e+00 -1.94492847e-01 -4.30239677e-01 5.11735618e-01
-4.52486813e-01 6.53693140e-01 -4.09324914e-01 -9.64635015e-01
-1.56884357e-01 -1.22082746e+00 8.27545762e-01 3.05519551e-01
1.57569572e-01 -9.10853028e-01 -8.14581588e-02 2.99623340e-01
6.10121012e-01 1.96552917e-01 6.18720412e-01 7.09788382e-01
-6.98132932e-01 1.21029764e-01 -4.24792737e-01 6.07481301e-01
1.72970116e-01 2.37263516e-01 -1.24157465e+00 -4.11552101e-01
1.24363303e-01 -6.97731674e-02 7.59131491e-01 4.95451748e-01
1.08925653e+00 3.61665227e-02 2.08954200e-01 7.37803996e-01
1.90152514e+00 1.70398384e-01 8.81314993e-01 4.81636941e-01
4.84211802e-01 8.66103470e-01 9.90763843e-01 4.01270777e-01
-1.28318503e-01 1.02044165e+00 6.28097236e-01 -7.49369860e-01
-4.81963128e-01 3.73032689e-01 3.78611296e-01 3.80261242e-01
-5.16672671e-01 -1.96423650e-01 -4.76247191e-01 2.72339642e-01
-1.49943531e+00 -1.16534042e+00 -3.81956369e-01 2.65593791e+00
8.05502057e-01 -1.64296672e-01 -4.20321245e-03 4.03472334e-01
5.76981127e-01 1.37320161e-01 -1.21623874e-01 -3.48761976e-01
-3.05784374e-01 2.87893832e-01 5.83091199e-01 5.75776339e-01
-9.49182689e-01 5.49012959e-01 6.36398411e+00 7.80610502e-01
-1.44821084e+00 1.42501712e-01 3.92826259e-01 -4.46945310e-01
-1.69629499e-01 4.52735797e-02 -1.37649313e-01 5.64359128e-01
5.34641683e-01 8.64065811e-02 5.87757766e-01 2.27932140e-01
7.35056341e-01 -7.93701649e-01 -8.30862880e-01 1.34924912e+00
1.68177918e-01 -9.15575147e-01 -2.38519132e-01 -5.57142906e-02
5.67260981e-01 -4.17122722e-01 2.21488953e-01 -4.71813679e-01
-4.20384973e-01 -7.24990427e-01 7.76431561e-01 5.97063601e-01
9.71543550e-01 -7.67106712e-01 3.88579249e-01 -1.90178141e-01
-1.20987499e+00 -7.52583891e-03 -3.24172437e-01 5.48694730e-02
4.44753021e-01 8.63186479e-01 -2.56244272e-01 6.50391221e-01
8.75585616e-01 6.39214873e-01 -5.11521161e-01 1.11426485e+00
-3.08946908e-01 2.82871395e-01 -2.59831309e-01 2.53886402e-01
1.17054924e-01 -6.49046481e-01 5.39832115e-01 1.28298891e+00
2.93263257e-01 3.46561104e-01 -5.87394945e-02 7.07817912e-01
2.55419552e-01 2.13797674e-01 -4.47028667e-01 2.07346052e-01
-1.63343281e-01 1.53635561e+00 -8.01403821e-01 -2.11259484e-01
-6.54748440e-01 1.23473370e+00 -3.98072422e-01 4.99474108e-01
-7.74088562e-01 -5.11275232e-01 4.06001985e-01 4.40065920e-01
1.69381380e-01 -3.90419573e-01 -2.23918706e-01 -1.14629424e+00
3.42074960e-01 -7.14462578e-01 1.60306826e-01 -1.14761007e+00
-8.71769011e-01 6.52327836e-01 -5.26102893e-02 -1.70056736e+00
1.38750136e-01 -5.55708945e-01 -4.56959367e-01 8.03567410e-01
-1.98147345e+00 -9.92518961e-01 -6.82114959e-01 9.45567131e-01
3.79850715e-01 3.28074604e-01 4.95507419e-01 5.62147498e-01
-3.11012179e-01 2.18253538e-01 1.92610353e-01 -4.75898087e-01
9.80109930e-01 -8.37515473e-01 -3.52638096e-01 1.46960974e+00
-1.39471024e-01 5.11004984e-01 9.92327631e-01 -2.56650060e-01
-1.42147100e+00 -8.23827744e-01 6.23542130e-01 1.70524027e-02
3.68776411e-01 -1.53678119e-01 -7.18671858e-01 1.95136461e-02
4.41057414e-01 1.12209588e-01 5.22225380e-01 -3.32197726e-01
-8.29601511e-02 -4.57705736e-01 -1.19277501e+00 6.22455597e-01
8.36823463e-01 -4.86349106e-01 -2.96207160e-01 1.02021903e-01
2.90169001e-01 -8.99871066e-02 -6.62005186e-01 3.34675759e-01
8.02675724e-01 -1.63544643e+00 1.03771245e+00 3.39627892e-01
4.41902906e-01 -7.45595217e-01 -1.22325860e-01 -1.12728393e+00
-3.04608405e-01 -7.46648610e-01 1.66188210e-01 1.19126022e+00
1.57937557e-02 -5.77916384e-01 4.17426229e-01 4.56726074e-01
1.60131380e-01 -3.75987232e-01 -5.78372896e-01 -6.13666892e-01
-6.67834699e-01 -3.81944925e-01 5.19334376e-02 7.38696098e-01
-3.20843726e-01 8.81670136e-03 -4.98953432e-01 1.41546845e-01
1.04176402e+00 3.11916798e-01 6.73915446e-01 -9.08807874e-01
-4.54985678e-01 -2.75438905e-01 -2.67782092e-01 -7.03793406e-01
-3.06962371e-01 -3.64605427e-01 6.43531606e-02 -1.39360213e+00
1.94931746e-01 -2.32293263e-01 -2.10714385e-01 1.42427683e-01
-7.52601400e-02 9.28797305e-01 3.75184178e-01 2.44873792e-01
-4.32093441e-01 4.67936188e-01 1.26051569e+00 1.34805739e-02
-3.59206080e-01 -1.69413567e-01 -3.35144430e-01 6.70089900e-01
4.90579724e-01 -2.27447763e-01 -4.23351616e-01 -3.55745286e-01
3.11831623e-01 -3.48149776e-01 5.35921693e-01 -1.06343865e+00
2.15780720e-01 -1.88651353e-01 3.70534837e-01 -3.22013199e-01
6.04225457e-01 -1.13952923e+00 2.98324317e-01 5.58364511e-01
7.06010982e-02 -2.15849280e-01 1.15086116e-01 4.15660530e-01
-3.14599335e-01 -2.31103614e-01 1.32633460e+00 -1.07040532e-01
-7.17989206e-01 4.26647291e-02 -3.15400749e-01 -4.46960270e-01
1.18048048e+00 -5.04309833e-01 -3.34915787e-01 -3.97118032e-01
-4.55750883e-01 -4.55027044e-01 9.29839671e-01 7.40522891e-02
6.40535295e-01 -1.01694643e+00 -4.78636086e-01 2.85132408e-01
2.23334268e-01 -5.09822428e-01 6.10568047e-01 1.29579079e+00
-8.31062019e-01 -3.54416511e-06 -4.42158997e-01 -6.78644896e-01
-1.64163840e+00 6.31599367e-01 2.57302195e-01 4.74215448e-02
-7.54291296e-01 4.11986917e-01 3.09519976e-01 5.40340960e-01
3.55272330e-02 -2.45965883e-01 -1.51993915e-01 -3.25552896e-02
5.77708125e-01 5.07677436e-01 2.53368527e-01 -6.42272592e-01
-2.64407158e-01 1.06205535e+00 2.73289114e-01 -1.86135441e-01
1.26052153e+00 -5.68425357e-01 -2.88146853e-01 4.18020874e-01
1.02361524e+00 3.23224425e-01 -1.32363558e+00 2.62545012e-02
-6.50847256e-01 -1.11233461e+00 4.32075351e-01 -6.16716623e-01
-1.08798850e+00 9.83577013e-01 8.69080603e-01 3.19132715e-01
1.91421461e+00 -4.27207232e-01 5.36244571e-01 3.05663086e-02
4.39990431e-01 -1.16995764e+00 1.16840161e-01 -1.31796990e-02
7.11492419e-01 -1.22861648e+00 2.97829568e-01 -6.19036257e-01
-3.73471141e-01 1.23054743e+00 2.25528479e-01 2.23226905e-01
4.03695732e-01 2.38838539e-01 2.08539903e-01 9.83014610e-03
-4.83381271e-01 -5.97291529e-01 3.12538117e-01 4.60116446e-01
5.43731451e-01 -3.71725410e-01 -6.04449689e-01 -2.71892250e-01
1.76384658e-01 -1.22404527e-02 7.26259768e-01 8.37590516e-01
-5.14072478e-01 -1.04338586e+00 -6.35102332e-01 -1.06391877e-01
-6.58889771e-01 -1.16123617e-01 5.65588884e-02 5.62690854e-01
2.53962487e-01 1.22577727e+00 -1.08813100e-01 -1.21625207e-01
4.06415164e-01 -3.64428222e-01 7.87750065e-01 -2.12406710e-01
-5.34181535e-01 4.97961074e-01 -7.62843639e-02 -9.52338994e-01
-1.06211388e+00 -4.92286980e-01 -8.48192096e-01 -2.55009145e-01
-2.95344502e-01 -1.98502645e-01 9.63931918e-01 5.03780663e-01
6.59479424e-02 5.91245651e-01 7.17268586e-01 -1.14078772e+00
1.22906022e-01 -5.37669420e-01 -7.13487029e-01 8.12175095e-01
4.30941641e-01 -7.47415364e-01 -4.29403424e-01 4.33053583e-01]
|
[10.715812683105469, -2.4876012802124023]
|
a1376457-1b22-48f2-89e5-cacc88f02ebd
|
learning-classifier-synthesis-for-generalized
|
1906.02944
| null |
https://arxiv.org/abs/1906.02944v5
|
https://arxiv.org/pdf/1906.02944v5.pdf
|
Learning Adaptive Classifiers Synthesis for Generalized Few-Shot Learning
|
Object recognition in the real-world requires handling long-tailed or even open-ended data. An ideal visual system needs to recognize the populated head visual concepts reliably and meanwhile efficiently learn about emerging new tail categories with a few training instances. Class-balanced many-shot learning and few-shot learning tackle one side of this problem, by either learning strong classifiers for head or learning to learn few-shot classifiers for the tail. In this paper, we investigate the problem of generalized few-shot learning (GFSL) -- a model during the deployment is required to learn about tail categories with few shots and simultaneously classify the head classes. We propose the ClAssifier SynThesis LEarning (CASTLE), a learning framework that learns how to synthesize calibrated few-shot classifiers in addition to the multi-class classifiers of head classes with a shared neural dictionary, shedding light upon the inductive GFSL. Furthermore, we propose an adaptive version of CASTLE (ACASTLE) that adapts the head classifiers conditioned on the incoming tail training examples, yielding a framework that allows effective backward knowledge transfer. As a consequence, ACASTLE can handle GFSL with classes from heterogeneous domains effectively. CASTLE and ACASTLE demonstrate superior performances than existing GFSL algorithms and strong baselines on MiniImageNet as well as TieredImageNet datasets. More interestingly, they outperform previous state-of-the-art methods when evaluated with standard few-shot learning criteria.
|
['De-Chuan Zhan', 'Han-Jia Ye', 'Hexiang Hu']
|
2019-06-07
| null | null | null | null |
['generalized-few-shot-learning']
|
['methodology']
|
[ 1.55854762e-01 1.37256682e-02 -4.15048331e-01 -5.36948383e-01
-8.40058208e-01 -3.66524935e-01 8.27348113e-01 -1.21931449e-01
-3.68163973e-01 5.66405833e-01 3.83687764e-02 1.59706220e-01
1.22871466e-01 -7.07948983e-01 -8.43317866e-01 -7.21592486e-01
1.26336487e-02 6.30322993e-01 6.32842481e-01 -2.73569196e-01
-2.33355284e-01 2.58253276e-01 -2.35545826e+00 6.65556133e-01
5.98640144e-01 1.44199717e+00 3.96153092e-01 5.99921525e-01
-4.23309833e-01 1.31560862e+00 -2.00847238e-01 -4.80105728e-01
2.30926901e-01 -3.27067196e-01 -6.55354142e-01 2.57214695e-01
9.02479768e-01 -2.71734923e-01 -3.60772014e-02 9.51389968e-01
4.98867303e-01 3.62563550e-01 8.51855695e-01 -1.52047348e+00
-6.30813599e-01 5.18463314e-01 -4.96534944e-01 3.82584810e-01
3.10054421e-02 4.93458003e-01 9.66140985e-01 -1.55854046e+00
1.15948594e+00 1.07555151e+00 6.25192881e-01 8.11644375e-01
-1.09195256e+00 -7.88855255e-01 3.76200944e-01 8.86050999e-01
-1.27983987e+00 -6.66058302e-01 6.09001517e-01 -8.17115307e-01
7.83745468e-01 -6.37714267e-02 6.33064091e-01 1.42983949e+00
-1.92609690e-02 1.02642119e+00 1.21369934e+00 -4.09549177e-01
8.24815452e-01 2.55129844e-01 6.47961080e-01 6.07298791e-01
4.83442061e-02 3.27867508e-01 -7.34527290e-01 1.19892679e-01
-1.13887591e-02 3.08392197e-01 -1.13005109e-01 -9.06848073e-01
-8.44923496e-01 9.03074622e-01 6.11218512e-01 1.67249188e-01
-3.87491643e-01 -4.07384103e-03 5.20017326e-01 4.94640976e-01
5.06739557e-01 5.73558807e-02 -4.76001143e-01 1.53188378e-01
-1.18440366e+00 -1.20056961e-02 8.51128578e-01 1.49168396e+00
1.16174412e+00 2.39319295e-01 -5.62610745e-01 8.34328115e-01
-9.92876068e-02 4.20046896e-01 5.94539404e-01 -5.74950337e-01
1.00082442e-01 3.98253441e-01 -2.97664642e-01 -1.30996734e-01
-1.88012764e-01 -5.55886030e-01 -7.14552104e-01 3.97927999e-01
8.24948996e-02 -7.55318478e-02 -1.59554732e+00 1.57107329e+00
4.13627237e-01 5.95047653e-01 6.48862198e-02 6.64711773e-01
1.16439950e+00 7.10154653e-01 3.39394152e-01 -3.53170305e-01
1.22407925e+00 -1.23238993e+00 -3.84133369e-01 -7.22582936e-01
3.46279055e-01 -1.88352704e-01 1.20899713e+00 2.33611494e-01
-6.48410618e-01 -8.95832062e-01 -1.09725130e+00 1.39064819e-01
-8.31881464e-01 -2.96970785e-01 4.41629976e-01 3.29477608e-01
-8.73928666e-01 3.23487461e-01 -5.11988103e-01 -5.53896725e-01
8.63167405e-01 -3.30053680e-02 -3.35905015e-01 -6.22266591e-01
-8.91798854e-01 8.37581277e-01 5.70791423e-01 -4.53379452e-01
-1.50849092e+00 -1.07679248e+00 -1.08070242e+00 1.58275858e-01
6.27935231e-01 -6.22458816e-01 1.45735538e+00 -1.10571098e+00
-1.23840213e+00 9.18591261e-01 1.07274182e-01 -4.99575049e-01
3.73930335e-01 -1.80378444e-02 -4.27534819e-01 1.16892092e-01
2.77255327e-01 8.07270467e-01 1.30168891e+00 -1.27796030e+00
-9.79549587e-01 -3.24939370e-01 -2.43130103e-01 5.10474592e-02
-4.49556172e-01 -1.92604259e-01 -3.45405668e-01 -4.79962349e-01
-5.05335629e-01 -6.57300055e-01 -9.03825983e-02 3.55472147e-01
6.49552569e-02 -2.02025384e-01 9.05220449e-01 -7.05614239e-02
8.44162524e-01 -2.27437234e+00 3.14359546e-01 -2.00950891e-01
2.96335757e-01 4.95053560e-01 -2.89107829e-01 1.68141335e-01
-8.42783600e-02 -6.66673243e-01 -1.78240061e-01 -2.81029075e-01
-6.18990744e-03 4.78815913e-01 -5.63678503e-01 2.72195786e-01
1.02285460e-01 1.16420996e+00 -1.07463396e+00 -4.93564606e-01
4.33650970e-01 1.38627440e-01 -3.52451324e-01 5.07237852e-01
-3.08270872e-01 -9.78609361e-03 3.70248854e-02 9.13051546e-01
5.29030561e-01 -4.29068387e-01 -6.19018339e-02 -1.60646021e-01
2.32309364e-02 -7.11836040e-01 -1.21509552e+00 1.73899794e+00
-5.48088491e-01 4.60139155e-01 -1.57961175e-01 -9.85314071e-01
8.71274829e-01 1.26697766e-02 1.81321595e-02 -6.68846071e-01
2.73574471e-01 1.90346643e-01 -3.15311641e-01 -4.93754148e-01
-2.98650265e-02 -6.35119855e-01 -5.43021038e-02 8.77969116e-02
1.16494501e+00 1.64258555e-02 2.83779502e-01 2.53802031e-01
9.54461932e-01 1.63931809e-02 7.05443084e-01 -1.01924613e-01
2.20258951e-01 -1.34929242e-02 7.09042490e-01 1.02704895e+00
-3.88251454e-01 6.83351040e-01 1.11958189e-02 -7.13597834e-01
-8.58005464e-01 -1.18622661e+00 -4.68247645e-02 1.83501256e+00
1.46264717e-01 -2.08305120e-01 -4.91839200e-01 -8.14202607e-01
1.55968934e-01 9.82971728e-01 -9.43550587e-01 -4.42163736e-01
-2.24827603e-02 -5.26741743e-01 -3.60340811e-02 7.57134855e-01
3.22783321e-01 -1.22450340e+00 -9.77990448e-01 3.00342947e-01
2.35111490e-01 -1.17784250e+00 -2.13738784e-01 8.19314063e-01
-4.41459894e-01 -1.19771802e+00 -1.02636468e+00 -9.41047728e-01
3.51401746e-01 3.54004294e-01 1.16222131e+00 -3.48137408e-01
-7.08497524e-01 4.57910389e-01 -6.61416888e-01 -5.29554725e-01
-8.25047567e-02 -5.89357503e-02 5.53421639e-02 4.59626287e-01
6.46142542e-01 -5.69625139e-01 -3.87138754e-01 8.68600681e-02
-8.04743409e-01 1.64423138e-01 6.67394042e-01 1.11345482e+00
6.27994239e-01 -6.08267784e-01 6.76694214e-01 -1.15975738e+00
1.39152244e-01 -8.06081116e-01 -4.11834389e-01 5.84287524e-01
-7.19903708e-01 5.37352636e-02 6.82031512e-01 -7.52274632e-01
-1.18402851e+00 1.47923604e-01 -1.02373958e-02 -9.74543631e-01
-3.21521103e-01 1.66511133e-01 -5.60467355e-02 -1.80727929e-01
1.13022780e+00 3.45188558e-01 -3.35201532e-01 -5.05042255e-01
9.13729489e-01 6.80114210e-01 7.65532315e-01 -4.13357377e-01
8.60828102e-01 6.25083089e-01 -3.62972289e-01 -1.02260244e+00
-1.30643940e+00 -9.76395249e-01 -9.22215402e-01 -5.23281813e-01
7.25378275e-01 -1.20008802e+00 8.10018033e-02 6.31162882e-01
-8.49655151e-01 -4.99455005e-01 -9.20678675e-01 5.47772311e-02
-8.07698727e-01 -8.09085667e-02 -3.05403888e-01 -5.03343523e-01
-4.08912659e-01 -8.56637597e-01 1.05342197e+00 3.26193213e-01
1.10687047e-01 -7.07367063e-01 2.28765473e-01 -1.60800330e-02
4.18681145e-01 2.40557671e-01 6.86915278e-01 -8.14706385e-01
-3.43203276e-01 -1.03286721e-01 -3.07682455e-01 2.49849737e-01
-2.78151631e-01 -3.64138931e-01 -1.42787886e+00 -5.29288113e-01
-2.36344159e-01 -1.01153934e+00 1.29595673e+00 2.18888164e-01
1.01346874e+00 -1.33924678e-01 -3.71452868e-01 1.02637887e+00
1.62200701e+00 -9.16411076e-03 3.30023855e-01 1.51381165e-01
6.07469738e-01 3.61690998e-01 7.03489780e-01 6.81905389e-01
3.01966548e-01 5.44781446e-01 3.65202516e-01 1.31834239e-01
-7.58061826e-01 -2.60847598e-01 2.51090109e-01 7.95928180e-01
1.30300790e-01 8.51017162e-02 -9.99729693e-01 7.60043263e-01
-1.90078163e+00 -1.09755898e+00 3.65275085e-01 1.81025624e+00
8.22942317e-01 9.50124934e-02 1.14004053e-01 -1.53691515e-01
6.31186485e-01 2.78392076e-01 -8.67848217e-01 -1.80099867e-02
-8.30538273e-02 5.29244363e-01 2.63899714e-01 1.41891941e-01
-1.18715060e+00 1.10674500e+00 5.58765221e+00 9.59508896e-01
-9.68397141e-01 6.24189258e-01 3.85644078e-01 -1.96662486e-01
1.37693688e-01 -5.99565636e-03 -1.23342609e+00 2.79761940e-01
8.12751830e-01 -4.42408860e-01 2.74400681e-01 1.51451182e+00
-7.41551936e-01 1.19186662e-01 -1.20641255e+00 1.20454049e+00
5.57904363e-01 -1.59167600e+00 1.76890790e-01 -3.53094012e-01
9.96822655e-01 4.49314743e-01 6.34694621e-02 1.03563535e+00
5.31664073e-01 -6.00663841e-01 9.52489138e-01 6.40564442e-01
1.21205366e+00 -4.51759845e-01 5.04754901e-01 6.25138223e-01
-1.34263194e+00 -7.26135612e-01 -5.90067387e-01 2.11134497e-02
2.36584600e-02 3.47294152e-01 -6.65612042e-01 2.74874121e-01
8.81586611e-01 9.97227609e-01 -8.00003707e-01 1.40547216e+00
-3.81605506e-01 5.19611478e-01 5.49435476e-03 1.27972290e-01
4.12425399e-01 3.07846159e-01 3.80029529e-01 1.21243560e+00
1.82219923e-01 1.87128872e-01 4.06546026e-01 6.68163061e-01
-6.31267428e-02 -4.16118605e-03 -6.16077960e-01 2.83390492e-01
4.14173543e-01 1.44905913e+00 -6.15282595e-01 -8.51963103e-01
-7.08867371e-01 9.92166162e-01 7.55105793e-01 3.70623201e-01
-5.29134870e-01 -3.92536461e-01 4.10872489e-01 1.17732391e-01
7.10780501e-01 4.16100115e-01 2.11518794e-01 -1.49508381e+00
-3.17873091e-01 -8.52044821e-01 8.74963939e-01 -8.73606145e-01
-1.69353056e+00 8.18567753e-01 1.70441642e-01 -1.39129329e+00
-2.93138564e-01 -7.43687987e-01 -7.13699937e-01 2.15793148e-01
-1.80485415e+00 -1.57067573e+00 -7.34011054e-01 9.74294066e-01
1.15156817e+00 -5.83541930e-01 6.87285125e-01 1.77146137e-01
-2.96466827e-01 4.60868537e-01 1.91370890e-01 -1.18136108e-01
8.32592547e-01 -1.19977772e+00 3.12981665e-01 7.00414181e-01
4.75001663e-01 6.27677441e-02 6.92854822e-01 -5.02964139e-01
-1.20120692e+00 -1.48970866e+00 5.23587465e-01 -3.73121530e-01
8.45142126e-01 -6.43001854e-01 -1.04908693e+00 7.46035933e-01
1.39443710e-01 8.54705632e-01 6.51775658e-01 1.49193615e-01
-8.25150609e-01 -3.19994926e-01 -8.44055295e-01 2.11322114e-01
1.14622533e+00 -5.56981206e-01 -1.03674781e+00 3.00451338e-01
6.59462869e-01 -8.79467353e-02 -4.78204459e-01 2.69726276e-01
4.40998495e-01 -8.52474988e-01 9.63061631e-01 -8.34995210e-01
2.18062371e-01 -1.32110387e-01 -4.42964822e-01 -1.51589441e+00
-5.61456800e-01 -1.92946419e-01 -6.74244225e-01 9.23814058e-01
6.21983968e-02 -1.02912515e-01 6.29863620e-01 2.31348664e-01
-2.81557113e-01 -4.39063072e-01 -9.94194388e-01 -1.17145717e+00
9.55666155e-02 -3.34459484e-01 3.23817581e-01 8.85272682e-01
-1.05203018e-01 5.69754958e-01 -5.94803214e-01 -2.86881834e-01
1.13543487e+00 3.89618248e-01 8.32259417e-01 -1.50447118e+00
-3.23909461e-01 -8.27290416e-02 -6.73255801e-01 -3.97845924e-01
3.24704051e-01 -1.20057833e+00 2.65032887e-01 -1.35686028e+00
6.65407956e-01 -2.98797823e-02 -4.73024130e-01 8.29670489e-01
-1.47947416e-01 3.01007152e-01 5.27388930e-01 1.68637738e-01
-1.10207617e+00 8.01783085e-01 8.21539342e-01 -5.34741580e-01
8.06902200e-02 -7.94796497e-02 -4.78115767e-01 7.49417901e-01
2.59608030e-01 -4.60773677e-01 -6.00271881e-01 7.01173544e-02
-1.43366814e-01 -3.59009802e-01 5.44470966e-01 -1.43225658e+00
5.69610596e-01 -1.74737453e-01 5.18075347e-01 -6.01736963e-01
3.54948968e-01 -7.80109107e-01 -4.29618135e-02 3.34476829e-01
-2.27752894e-01 -6.91818535e-01 -2.80330610e-02 8.70751679e-01
-1.79280639e-01 -2.05808893e-01 1.27370584e+00 -2.91004807e-01
-1.98145890e+00 7.51157403e-01 -1.67432562e-01 4.38442826e-01
1.61185193e+00 -1.85682565e-01 -4.87489253e-01 -7.58459345e-02
-1.13417029e+00 2.85660893e-01 3.48667026e-01 6.81498766e-01
8.83071005e-01 -1.43665957e+00 -5.50619364e-01 3.43565613e-01
1.05948496e+00 -2.88494706e-01 4.22078580e-01 5.31425476e-01
1.91978049e-02 -2.96820849e-02 -6.14144683e-01 -5.85615754e-01
-1.13964975e+00 1.31153858e+00 2.16488868e-01 4.71061654e-02
-9.56576586e-01 1.16181779e+00 4.60560769e-01 -4.25800532e-01
4.61176068e-01 9.24236849e-02 -2.26048827e-01 6.23652577e-01
9.75984216e-01 2.38845408e-01 2.85718609e-02 -5.82540393e-01
-3.23291510e-01 4.15838987e-01 -3.26016068e-01 1.98492497e-01
1.63745582e+00 -3.92622948e-02 4.49432105e-01 8.84997070e-01
1.21477485e+00 -8.90222549e-01 -1.66019940e+00 -7.33619571e-01
1.31871328e-01 -3.05632412e-01 7.87155479e-02 -8.75305057e-01
-9.25736129e-01 1.16561985e+00 9.49703515e-01 -4.49638903e-01
1.00001955e+00 4.00290519e-01 5.21644711e-01 4.97383237e-01
7.33929336e-01 -1.35325348e+00 3.62320751e-01 4.68682289e-01
6.43613219e-01 -1.55853558e+00 -1.95622981e-01 -6.23714216e-02
-1.00806391e+00 1.04821658e+00 8.22399437e-01 -3.01463939e-02
8.86714399e-01 2.74906784e-01 -1.81202721e-02 -3.90551955e-01
-1.25932646e+00 -8.51396322e-01 4.92319256e-01 9.19530451e-01
-2.97175705e-01 -5.49373366e-02 3.68395120e-01 7.99474716e-01
3.50319624e-01 4.48103309e-01 2.77191341e-01 1.06708038e+00
-9.46319044e-01 -6.40312552e-01 -2.78801285e-02 6.44684970e-01
2.77160496e-01 -1.43546075e-01 -1.50386333e-01 5.39212167e-01
5.05636454e-01 6.58643961e-01 -9.00859758e-03 -3.52242678e-01
3.47146034e-01 5.07119894e-01 3.47491026e-01 -1.10167587e+00
-2.81455070e-01 -2.49042392e-01 -1.94985867e-01 -5.95539749e-01
-2.17159912e-01 -5.82856178e-01 -6.78587854e-01 1.73994064e-01
-3.17286134e-01 -1.07619338e-01 2.50946254e-01 1.06171358e+00
2.12435424e-01 4.94614631e-01 5.69379866e-01 -1.05750382e+00
-7.94540942e-01 -7.90311098e-01 -9.59830642e-01 6.06371820e-01
4.06783283e-01 -1.09063506e+00 -3.55409741e-01 3.09818029e-01]
|
[9.998780250549316, 2.684119939804077]
|
af84a4d3-6fde-4a5f-8949-f3a9143451ba
|
a-resource-light-method-for-cross-lingual
|
1801.06436
| null |
http://arxiv.org/abs/1801.06436v1
|
http://arxiv.org/pdf/1801.06436v1.pdf
|
A Resource-Light Method for Cross-Lingual Semantic Textual Similarity
|
Recognizing semantically similar sentences or paragraphs across languages is
beneficial for many tasks, ranging from cross-lingual information retrieval and
plagiarism detection to machine translation. Recently proposed methods for
predicting cross-lingual semantic similarity of short texts, however, make use
of tools and resources (e.g., machine translation systems, syntactic parsers or
named entity recognition) that for many languages (or language pairs) do not
exist. In contrast, we propose an unsupervised and a very resource-light
approach for measuring semantic similarity between texts in different
languages. To operate in the bilingual (or multilingual) space, we project
continuous word vectors (i.e., word embeddings) from one language to the vector
space of the other language via the linear translation model. We then align
words according to the similarity of their vectors in the bilingual embedding
space and investigate different unsupervised measures of semantic similarity
exploiting bilingual embeddings and word alignments. Requiring only a
limited-size set of word translation pairs between the languages, the proposed
approach is applicable to virtually any pair of languages for which there
exists a sufficiently large corpus, required to learn monolingual word
embeddings. Experimental results on three different datasets for measuring
semantic textual similarity show that our simple resource-light approach
reaches performance close to that of supervised and resource intensive methods,
displaying stability across different language pairs. Furthermore, we evaluate
the proposed method on two extrinsic tasks, namely extraction of parallel
sentences from comparable corpora and cross lingual plagiarism detection, and
show that it yields performance comparable to those of complex
resource-intensive state-of-the-art models for the respective tasks.
|
['Marc Franco-Salvador', 'Goran Glavaš', 'Simone Paolo Ponzetto', 'Paolo Rosso']
|
2018-01-19
| null | null | null | null |
['cross-lingual-information-retrieval', 'cross-lingual-semantic-textual-similarity']
|
['natural-language-processing', 'natural-language-processing']
|
[ 7.56970495e-02 -3.63479197e-01 -3.64303589e-01 -2.10396618e-01
-9.82257783e-01 -9.05969679e-01 9.69682097e-01 6.40444338e-01
-8.26457262e-01 4.86834288e-01 3.37334812e-01 -5.30738294e-01
4.75536101e-02 -6.95629954e-01 -4.75024194e-01 -4.06314790e-01
4.24501717e-01 4.97835606e-01 6.77787438e-02 -4.28801268e-01
5.69553256e-01 4.00353044e-01 -1.31937051e+00 2.07328185e-01
9.51010227e-01 3.57085854e-01 2.32773200e-01 2.88240820e-01
-7.45113730e-01 2.15468660e-01 -3.19910020e-01 -6.87752783e-01
1.17263220e-01 -4.40641612e-01 -1.14383221e+00 -1.06611274e-01
5.24956524e-01 5.34274578e-01 -2.11714413e-02 1.35699213e+00
3.82667780e-01 -5.05507141e-02 7.18123436e-01 -8.37184370e-01
-9.85511899e-01 5.30623138e-01 -2.60317773e-01 3.09982210e-01
7.39854693e-01 -3.35099339e-01 1.18305480e+00 -1.28018880e+00
9.44696486e-01 1.20192063e+00 5.81671894e-01 2.25492075e-01
-1.33669758e+00 -3.32324058e-01 -2.51174778e-01 3.03970933e-01
-1.27293682e+00 -4.80344713e-01 6.74362004e-01 -6.47418618e-01
1.16356349e+00 1.97187066e-01 6.11630417e-02 1.19371581e+00
2.33914644e-01 1.82618558e-01 1.37760997e+00 -8.54157031e-01
-3.55308093e-02 7.49502957e-01 2.70181417e-01 5.50829113e-01
4.30120766e-01 -3.17289948e-01 -5.09782910e-01 -2.96768367e-01
-6.44289255e-02 -2.06970543e-01 -2.70906687e-01 -4.67214316e-01
-1.36688268e+00 1.12175381e+00 -5.64810336e-02 1.14467931e+00
-1.66238412e-01 -5.50286055e-01 8.97592425e-01 6.41004562e-01
6.54289782e-01 7.39477634e-01 -3.43844593e-01 9.11253691e-02
-8.30080092e-01 -4.63431366e-02 1.04653060e+00 1.06977999e+00
9.39137936e-01 -3.58718127e-01 1.62388667e-01 9.05663967e-01
7.77614117e-02 5.27956009e-01 1.16596079e+00 -1.21000819e-01
8.25200796e-01 6.35674059e-01 -1.29576489e-01 -1.39326346e+00
-2.71529168e-01 5.35239577e-02 -5.37444949e-01 -3.30238879e-01
3.76250982e-01 2.94307560e-01 -5.75146154e-02 1.63872480e+00
2.81202674e-01 -2.43295670e-01 6.32326126e-01 5.47682106e-01
7.08093405e-01 5.80795884e-01 -6.47538155e-02 -3.62288713e-01
1.75787604e+00 -7.75397301e-01 -6.39146447e-01 -3.50687623e-01
1.11831164e+00 -1.48088765e+00 1.35256028e+00 -8.85901302e-02
-9.05995131e-01 -5.02554238e-01 -1.00571191e+00 -2.22081035e-01
-8.45141351e-01 6.29463419e-02 9.26215723e-02 8.18978965e-01
-8.11144114e-01 7.54255950e-01 -4.54740822e-01 -1.05335736e+00
-9.96208414e-02 -1.15728781e-01 -8.04796100e-01 -1.59019679e-01
-1.34385681e+00 1.34383476e+00 5.24384379e-01 -3.14868748e-01
-2.02792987e-01 -4.14994031e-01 -1.05605102e+00 -5.04831076e-02
8.69078338e-02 -4.13409203e-01 6.80721700e-01 -1.02254999e+00
-1.19234943e+00 1.40701675e+00 -2.23355234e-01 -2.40731090e-01
2.62328148e-01 -2.16494337e-01 -5.56396127e-01 2.04207674e-02
4.41774189e-01 5.10599464e-02 6.87470019e-01 -8.12331200e-01
-3.12163472e-01 -5.05226731e-01 -1.55273080e-01 1.78752333e-01
-1.02876353e+00 6.10980570e-01 -1.35167316e-01 -6.36153877e-01
5.19731082e-02 -9.71370459e-01 7.91344345e-02 -3.63637686e-01
-1.81506634e-01 -3.18560809e-01 6.16099536e-01 -1.03154469e+00
1.17903757e+00 -2.01729679e+00 3.73334020e-01 3.44985574e-02
-2.15548709e-01 2.97939807e-01 -3.63505274e-01 9.98544693e-01
-2.71711975e-01 2.18478680e-01 -4.50375795e-01 -2.68156171e-01
4.73740734e-02 2.76714951e-01 -2.11963519e-01 7.31062531e-01
2.18746230e-01 6.39610767e-01 -9.73374426e-01 -6.47609770e-01
1.73078775e-01 2.98914015e-01 -2.60684013e-01 1.71052068e-01
3.79912049e-01 2.12973282e-01 -2.98894167e-01 1.48693800e-01
4.70087022e-01 2.03838393e-01 5.81207812e-01 -1.91157356e-01
-2.35222191e-01 7.92463899e-01 -9.48550284e-01 1.84069860e+00
-1.11074054e+00 7.70492673e-01 -4.94430780e-01 -1.31773031e+00
1.29976022e+00 5.42207479e-01 8.51292387e-02 -6.25979483e-01
1.40492618e-01 6.42116666e-01 -2.00664654e-01 -6.81582153e-01
6.79299533e-01 -4.00632739e-01 -3.85987967e-01 6.28638387e-01
3.76085579e-01 -7.16122016e-02 3.48005354e-01 -1.23324782e-01
1.01925588e+00 -6.69184104e-02 7.65651941e-01 -6.64853573e-01
1.19343257e+00 8.69445875e-02 2.04745844e-01 2.53800660e-01
-1.47800678e-02 1.74473017e-01 3.75614941e-01 -3.20783854e-01
-1.39019299e+00 -9.13999379e-01 -3.17030340e-01 1.06037366e+00
8.30810666e-02 -5.12934089e-01 -8.24739039e-01 -6.12918198e-01
-1.57534018e-01 8.36132646e-01 -1.87661767e-01 -1.15033805e-01
-7.50308931e-01 -5.83747625e-01 7.45469749e-01 1.17409930e-01
-7.89075866e-02 -1.14468765e+00 -1.81219995e-01 2.12326184e-01
-3.09790820e-01 -1.41008282e+00 -3.79100829e-01 -1.03554264e-01
-9.53086019e-01 -1.02055812e+00 -4.57748979e-01 -1.21215403e+00
3.94234538e-01 2.54936695e-01 1.21018839e+00 -3.03543836e-01
-2.20668152e-01 3.50412071e-01 -5.19686759e-01 1.81949481e-01
-1.01541245e+00 2.47575313e-01 4.68758702e-01 4.49253619e-02
8.28094661e-01 -5.98082960e-01 -2.12295563e-04 1.91329673e-01
-1.14146328e+00 -3.59913677e-01 4.19667244e-01 8.58114183e-01
2.56023735e-01 -3.62443298e-01 3.79682571e-01 -8.95990729e-01
8.90581667e-01 -4.99946207e-01 -4.27853912e-01 5.19427061e-01
-6.57302976e-01 4.96495187e-01 1.03770149e+00 -5.20330906e-01
-7.95963049e-01 -3.35440129e-01 -8.13221484e-02 -2.51223445e-02
-2.82169163e-01 4.98988152e-01 -2.00339064e-01 -6.15896238e-03
7.69508898e-01 4.42026675e-01 -1.57702446e-01 -6.86689377e-01
7.05471694e-01 1.01014280e+00 4.11331803e-01 -4.75569159e-01
9.89014328e-01 2.49924392e-01 -2.70133615e-01 -1.06265640e+00
-4.56536055e-01 -8.53286862e-01 -9.69726562e-01 2.97622621e-01
8.84566724e-01 -6.76699698e-01 -1.43932015e-01 4.06422019e-02
-1.60029352e+00 6.81374669e-01 -1.24446966e-01 7.58418977e-01
-5.18839478e-01 8.05659175e-01 -4.25533623e-01 -3.07774961e-01
-4.61350620e-01 -9.38982010e-01 9.06060398e-01 -2.60447621e-01
-5.30026019e-01 -1.43637300e+00 6.70647204e-01 3.49127442e-01
2.00667515e-01 -4.41712737e-02 1.28088808e+00 -1.18381739e+00
1.48467824e-01 -3.52406472e-01 -2.46354993e-02 6.22963965e-01
3.47380310e-01 -3.27812016e-01 -8.01450670e-01 -4.19858754e-01
2.64892817e-01 -2.12427348e-01 4.11403269e-01 -4.05507922e-01
3.24317187e-01 -3.98229331e-01 -1.59047738e-01 1.77717298e-01
1.63988161e+00 -1.93091854e-01 4.75490153e-01 5.94594479e-01
6.72010899e-01 1.01979220e+00 4.66215640e-01 3.17260548e-02
1.17954157e-01 8.35591793e-01 -1.78565815e-01 1.64507627e-01
5.48625179e-02 -2.28929922e-01 8.02626252e-01 1.62321222e+00
3.48668098e-01 1.91830508e-02 -9.51331735e-01 8.76277030e-01
-1.56550097e+00 -9.29359257e-01 -3.31976116e-01 2.47275662e+00
8.40199053e-01 -3.58869359e-02 -2.10381418e-01 6.45998269e-02
8.80116045e-01 2.59863347e-01 1.38339430e-01 -1.00646675e+00
-2.63136685e-01 3.81171674e-01 4.51644063e-01 5.52578509e-01
-8.94923151e-01 1.13364196e+00 4.83398008e+00 9.21991825e-01
-1.03583074e+00 6.11920297e-01 -1.36203796e-01 4.18009877e-01
-4.76730675e-01 1.98634058e-01 -6.91378295e-01 3.21469188e-01
1.22970474e+00 -4.96064186e-01 2.45225742e-01 7.54908562e-01
1.07402071e-01 1.99053124e-01 -1.25628185e+00 9.66364264e-01
5.73826194e-01 -1.10724926e+00 2.76357513e-02 -1.96260452e-01
6.19380236e-01 1.70452401e-01 -1.99707359e-01 9.01166946e-02
-1.38373226e-01 -6.78240895e-01 4.61575538e-01 7.74347931e-02
6.49676859e-01 -7.34911323e-01 9.32778835e-01 3.28873634e-01
-1.14782703e+00 3.58067155e-01 -7.03790069e-01 1.60783678e-01
8.24448988e-02 4.08241928e-01 -6.22781515e-01 7.47117043e-01
3.97771329e-01 7.72153020e-01 -6.76657915e-01 4.70056981e-01
-2.52585709e-01 2.38005608e-01 2.31754500e-02 -1.82805210e-01
4.38381702e-01 -5.82235098e-01 7.03578055e-01 1.60496259e+00
5.96763313e-01 -5.22204101e-01 -1.49668811e-03 7.37970948e-01
-6.70481548e-02 1.06277335e+00 -1.18544793e+00 -4.05380815e-01
3.31432015e-01 1.23664665e+00 -5.08851588e-01 -4.38796431e-01
-7.87059724e-01 1.23202562e+00 5.40026546e-01 -7.18150437e-02
-4.71765876e-01 -6.14669144e-01 8.07952344e-01 -8.65208581e-02
1.86870307e-01 -2.74438798e-01 -1.04031585e-01 -1.48141503e+00
4.10095692e-01 -9.21221972e-01 2.52401143e-01 -3.64726216e-01
-1.64236522e+00 9.65069592e-01 -1.93210855e-01 -1.41890645e+00
-4.14985150e-01 -7.86380768e-01 -4.73370016e-01 1.01964855e+00
-1.54089057e+00 -1.03564334e+00 2.77045637e-01 6.90103412e-01
5.93917310e-01 -4.32905734e-01 1.17338812e+00 4.38049883e-01
-2.58009821e-01 5.84511995e-01 5.47077537e-01 1.08525887e-01
9.99900997e-01 -9.02521014e-01 5.45036972e-01 8.25126469e-01
7.02988446e-01 7.91208327e-01 6.46709085e-01 -4.57964301e-01
-1.41992474e+00 -9.49856579e-01 2.04129553e+00 -4.60443377e-01
1.33133221e+00 -5.87391257e-01 -1.18034947e+00 2.94330478e-01
4.00620073e-01 -2.63467312e-01 9.48128581e-01 2.30740130e-01
-8.04659367e-01 1.66403174e-01 -8.56119037e-01 6.03921831e-01
8.45882535e-01 -1.18545282e+00 -1.20429742e+00 8.13120604e-01
7.22141385e-01 2.52261400e-01 -9.80961740e-01 -1.29392877e-01
3.69044870e-01 -7.99553096e-01 8.28149796e-01 -9.62891579e-01
5.29434323e-01 -1.46842569e-01 -3.52004021e-01 -1.23438668e+00
-3.13449539e-02 -4.06845093e-01 6.21119916e-01 1.46137393e+00
3.74247968e-01 -7.89707959e-01 1.89505517e-01 2.17130091e-02
1.27689103e-02 -2.67750829e-01 -1.05649984e+00 -1.14697194e+00
4.49594945e-01 -4.30648237e-01 3.09197754e-01 1.36736906e+00
3.05615604e-01 6.86639011e-01 -6.35228306e-02 1.90484077e-01
3.95290643e-01 3.66027117e-01 5.63583076e-01 -1.03932881e+00
-2.18620390e-01 -4.65179056e-01 -7.25091934e-01 -5.04798532e-01
8.15278888e-01 -1.48049045e+00 -1.21989779e-01 -1.01765752e+00
2.96032101e-01 -1.45205721e-01 -7.63647184e-02 7.08312541e-02
-1.39631614e-01 2.68879682e-01 1.46774054e-01 4.03667301e-01
-1.59430608e-01 4.98555958e-01 4.02402997e-01 -1.25542104e-01
1.79161821e-02 -4.13566411e-01 -4.09754157e-01 8.40288699e-01
6.16647422e-01 -9.25540805e-01 1.03187561e-01 -3.65617931e-01
3.00746918e-01 -2.19199032e-01 1.99597538e-01 -7.92141378e-01
-2.55822968e-02 -4.41128984e-02 -3.62265289e-01 -1.89935346e-03
-1.48239300e-01 -8.46029580e-01 -1.88213661e-01 5.17871082e-01
-3.87998790e-01 5.98045886e-01 2.86415648e-02 3.65473986e-01
-5.36354125e-01 -8.50170910e-01 7.47981787e-01 2.37697493e-02
-5.72151601e-01 -1.88872069e-01 -4.48242009e-01 3.16483527e-01
8.60673428e-01 -6.34203404e-02 -1.64013907e-01 8.27378258e-02
-3.99593979e-01 -4.09485132e-01 7.40914643e-01 9.07898068e-01
3.93809438e-01 -1.54995036e+00 -8.49523365e-01 5.58812395e-02
6.13737106e-01 -9.24121082e-01 -1.46601617e-01 9.20684278e-01
-4.74602014e-01 6.75774395e-01 -3.37641269e-01 -4.14659232e-01
-1.48174524e+00 9.21538055e-01 -1.09252386e-01 -4.85844940e-01
-4.01124328e-01 2.49075100e-01 1.29292548e-01 -1.00713861e+00
-3.94419163e-01 -5.20920120e-02 -3.62436235e-01 3.16799104e-01
3.67076457e-01 2.91326582e-01 4.43451583e-01 -1.31297421e+00
-5.05864620e-01 8.92670631e-01 5.84164970e-02 -1.72233582e-01
1.17812800e+00 -3.55139196e-01 -5.06698489e-01 6.82013273e-01
1.90080130e+00 1.97196603e-01 9.11232904e-02 -5.97489834e-01
6.71493351e-01 -4.56479430e-01 -2.42948443e-01 7.07633942e-02
-5.03691614e-01 1.05846560e+00 5.07993996e-01 1.57879695e-01
8.10618103e-01 1.70830771e-01 8.26197326e-01 6.07849002e-01
3.49582285e-01 -1.12115622e+00 -1.72407031e-01 5.93532801e-01
7.55317688e-01 -1.31963491e+00 -9.13457870e-02 -4.11960751e-01
-5.10990918e-01 1.31158888e+00 -9.66520573e-04 -1.39825329e-01
3.72479886e-01 -1.54437348e-01 1.79498687e-01 -1.10890470e-01
-3.62900048e-01 -8.36122185e-02 5.35363197e-01 4.09761310e-01
7.53337502e-01 -2.03620028e-02 -1.07492685e+00 1.64452657e-01
-2.83880234e-01 -5.36233544e-01 4.63514686e-01 7.79966772e-01
-1.95267811e-01 -1.60295081e+00 -5.63848257e-01 -7.09601715e-02
-4.40927953e-01 -5.29224038e-01 -4.07806635e-01 8.43365312e-01
-2.83676106e-02 9.56970751e-01 6.17242381e-02 -1.13853656e-01
2.93101549e-01 4.35734451e-01 3.51287335e-01 -7.66691923e-01
-7.89037228e-01 -2.34890506e-01 1.46654442e-01 -5.36453843e-01
-6.17696643e-01 -7.52019823e-01 -7.84417808e-01 -3.16455901e-01
-3.44241589e-01 2.90366769e-01 9.22979116e-01 1.17394006e+00
1.45028561e-01 -1.35007858e-01 8.17402899e-01 -5.63459694e-01
-7.03520954e-01 -9.48802352e-01 -5.38062871e-01 8.59160602e-01
-1.67226151e-01 -3.66943955e-01 -3.48626584e-01 1.27465859e-01]
|
[11.017614364624023, 9.902726173400879]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.