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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
00a69dbb-22af-47c5-8bad-3c00c1e5deca
|
ernie-enhanced-representation-through
|
1904.09223
| null |
http://arxiv.org/abs/1904.09223v1
|
http://arxiv.org/pdf/1904.09223v1.pdf
|
ERNIE: Enhanced Representation through Knowledge Integration
|
We present a novel language representation model enhanced by knowledge called
ERNIE (Enhanced Representation through kNowledge IntEgration). Inspired by the
masking strategy of BERT, ERNIE is designed to learn language representation
enhanced by knowledge masking strategies, which includes entity-level masking
and phrase-level masking. Entity-level strategy masks entities which are
usually composed of multiple words.Phrase-level strategy masks the whole phrase
which is composed of several words standing together as a conceptual
unit.Experimental results show that ERNIE outperforms other baseline methods,
achieving new state-of-the-art results on five Chinese natural language
processing tasks including natural language inference, semantic similarity,
named entity recognition, sentiment analysis and question answering. We also
demonstrate that ERNIE has more powerful knowledge inference capacity on a
cloze test.
|
['Hao Tian', 'Yu Sun', 'Xuyi Chen', 'Hua Wu', 'Yukun Li', 'Xin Tian', 'Shikun Feng', 'Han Zhang', 'Danxiang Zhu', 'Shuohuan Wang']
|
2019-04-19
| null | null | null | null |
['cloze-test', 'chinese-named-entity-recognition']
|
['natural-language-processing', 'natural-language-processing']
|
[-1.70777757e-02 -6.70100003e-02 -1.19636856e-01 -1.15255013e-01
-6.61733687e-01 -5.40144145e-01 3.50351274e-01 5.18812597e-01
-8.74744534e-01 8.72977734e-01 6.83984876e-01 -4.06137675e-01
-3.13218087e-02 -1.08662999e+00 -4.77299333e-01 -1.27688631e-01
-5.36455996e-02 1.13760695e-01 1.65121615e-01 -6.35429978e-01
3.68162662e-01 1.92629308e-01 -1.18451083e+00 7.96289802e-01
9.83281434e-01 6.79959595e-01 2.74569869e-01 3.16417933e-01
-7.53347218e-01 1.14495564e+00 -6.88815653e-01 -5.13566732e-01
-2.05405012e-01 -7.62894824e-02 -1.09508395e+00 -3.06470573e-01
-1.58366650e-01 2.19043627e-01 8.00005868e-02 1.13658607e+00
3.73752534e-01 4.85523164e-01 5.93970835e-01 -5.96782088e-01
-1.44847536e+00 1.12345529e+00 -3.50682199e-01 3.39009672e-01
6.46887481e-01 -4.33083355e-01 1.26112127e+00 -1.36628568e+00
4.75432456e-01 1.36654305e+00 6.08798802e-01 3.84386688e-01
-7.62184441e-01 -8.83311629e-01 6.37878835e-01 3.42750043e-01
-1.72516906e+00 4.12487872e-02 2.61807323e-01 -1.70955613e-01
1.64588559e+00 6.70684921e-03 -6.59868270e-02 6.46000564e-01
5.95413754e-03 9.71459985e-01 1.02139890e+00 -7.86631823e-01
8.70090798e-02 4.46068794e-01 8.14183474e-01 7.57469058e-01
4.14062709e-01 -2.01394305e-01 -5.45464337e-01 -2.34844252e-01
5.31027853e-01 -2.89380699e-01 -2.85138160e-01 3.38456959e-01
-1.29236400e+00 9.54981983e-01 3.54841858e-01 6.59828842e-01
-5.20558894e-01 -2.00831622e-01 3.70028406e-01 4.25848246e-01
1.88459367e-01 7.76151538e-01 -9.68084276e-01 1.67838514e-01
-4.02351320e-01 1.14275873e-01 8.74993026e-01 1.21951568e+00
9.19856548e-01 1.90678641e-01 -5.35604358e-01 8.51537645e-01
1.18795343e-01 5.42457104e-01 1.01131463e+00 -2.91587204e-01
5.38270235e-01 1.14642823e+00 1.50753394e-01 -9.14752185e-01
-3.37602377e-01 -4.69919831e-01 -1.02845144e+00 -5.04041553e-01
-2.74723679e-01 -3.59028190e-01 -8.13516974e-01 1.75637782e+00
-4.25063260e-02 2.90078372e-01 7.14337170e-01 3.07013214e-01
1.35768473e+00 6.48180962e-01 5.03934383e-01 -6.16157800e-02
2.03921151e+00 -1.09535444e+00 -9.97905433e-01 -3.84731770e-01
6.43560469e-01 -6.35257900e-01 8.51108849e-01 2.29246363e-01
-9.19196665e-01 -7.41049170e-01 -7.86573768e-01 -9.89994705e-02
-1.25859058e+00 3.95436496e-01 7.68556058e-01 7.93626368e-01
-7.34626114e-01 1.34576425e-01 -1.78586587e-01 -1.61775276e-01
9.94670093e-02 1.90529063e-01 -6.41014755e-01 -2.22598612e-02
-1.97403324e+00 1.16172874e+00 1.11357749e+00 -5.55191515e-03
-2.80517042e-01 -7.43750155e-01 -1.37010431e+00 2.55537093e-01
6.82899237e-01 -8.37465703e-01 9.80743229e-01 -4.94123399e-01
-1.21143854e+00 7.85678566e-01 -6.37264729e-01 -5.24768531e-01
-4.26146239e-01 -5.31927943e-01 -9.88409460e-01 -1.43111959e-01
2.03129038e-01 4.16610181e-01 5.11569023e-01 -1.01519203e+00
-4.80945289e-01 -3.88828665e-02 1.05483338e-01 3.21670055e-01
-4.02416915e-01 6.17309093e-01 -4.37734276e-01 -9.54884827e-01
-1.16343424e-01 -2.39298284e-01 -2.37747192e-01 -7.34675944e-01
-2.65278161e-01 -4.66615587e-01 4.08453822e-01 -8.58962119e-01
1.83835602e+00 -1.88690460e+00 -1.18318118e-01 1.70155182e-01
7.87234679e-02 5.66161156e-01 -3.14968050e-01 6.41100407e-01
-4.92622167e-01 5.08149683e-01 -1.54602006e-01 -4.44662459e-02
1.04016669e-01 1.81072474e-01 -7.83240974e-01 -4.37671959e-01
5.29735804e-01 1.39416528e+00 -1.07744539e+00 -2.74335116e-01
-1.00227579e-01 2.07366556e-01 -4.36702192e-01 1.55862302e-01
-2.00528085e-01 -6.50881901e-02 -3.97079915e-01 7.46227384e-01
7.31028557e-01 -4.09963310e-01 -7.79691860e-02 -7.97208250e-02
-2.39203162e-02 5.34065306e-01 -1.49785185e+00 1.51156032e+00
-5.17862856e-01 2.02385732e-03 -2.78407991e-01 -6.90931022e-01
1.05619860e+00 4.58511353e-01 -3.87852132e-01 -4.10863549e-01
1.30545527e-01 -1.67311922e-01 -1.75956354e-01 -4.66459602e-01
9.51149642e-01 -4.14216250e-01 -4.51678932e-01 3.83389682e-01
1.54557154e-01 7.85997882e-02 1.79733157e-01 4.47541744e-01
1.06012189e+00 -1.77849725e-01 1.09386563e+00 -3.16397876e-01
1.17144561e+00 -2.20986366e-01 6.11070096e-01 9.77755249e-01
-4.30454090e-02 6.20641969e-02 7.44505674e-02 6.43817633e-02
-2.55914688e-01 -1.24301994e+00 1.74917430e-01 1.40145624e+00
5.84320836e-02 -8.20811749e-01 -3.54876906e-01 -7.08326280e-01
5.22073358e-02 1.03089035e+00 -7.50560582e-01 -2.85030931e-01
-4.55438018e-01 -5.82661927e-01 9.20578003e-01 1.05042565e+00
9.26875651e-01 -1.44108784e+00 1.48006171e-01 3.75490636e-01
-2.82774270e-01 -1.36781859e+00 -4.55399334e-01 1.54151350e-01
-5.30428886e-01 -7.94333816e-01 -2.81090230e-01 -1.14585447e+00
6.01675332e-01 1.57340378e-01 1.43111086e+00 8.81328359e-02
-1.70591787e-01 3.20306391e-01 -7.41684139e-01 -6.71328247e-01
-2.33486928e-02 -1.61873594e-01 7.34090665e-03 -2.20776588e-01
1.07294440e+00 -2.72575527e-01 -1.53331146e-01 8.50738958e-02
-1.03375006e+00 -3.21315467e-01 7.98365891e-01 8.77968788e-01
4.28165376e-01 2.49569103e-01 9.02279198e-01 -9.31681097e-01
9.92583454e-01 -6.15789473e-01 -2.44270772e-01 7.74908602e-01
-1.70642912e-01 2.49835163e-01 6.61807895e-01 -4.65982229e-01
-1.25503910e+00 -4.02839541e-01 -1.82149231e-01 -1.03130834e-02
-4.44534630e-01 1.06143486e+00 -2.90511370e-01 2.17697933e-01
5.75709343e-01 6.41818941e-01 -5.89468181e-01 -5.85075140e-01
8.32496762e-01 7.21262395e-01 6.50337338e-01 -8.36540043e-01
7.61770964e-01 2.49241322e-01 -7.18464851e-01 -8.68535757e-01
-1.10713863e+00 -7.78361440e-01 -6.79393470e-01 4.66879368e-01
1.04809535e+00 -1.32624900e+00 -8.04927230e-01 1.80018529e-01
-1.34574819e+00 4.08810526e-01 -1.83790922e-01 4.92817998e-01
1.03466816e-01 5.08359253e-01 -6.42167926e-01 -7.98649430e-01
-7.89211452e-01 -4.13481832e-01 9.78000462e-01 2.89611340e-01
-3.03401887e-01 -1.21690428e+00 -6.66583776e-02 3.65096360e-01
4.32235032e-01 -1.95401803e-01 1.40213931e+00 -1.17400730e+00
-2.77077079e-01 -3.50170918e-02 -3.11418653e-01 7.33759940e-01
2.83527076e-01 -6.46992505e-01 -6.81214273e-01 -3.62742431e-02
-5.38418591e-02 -3.86201411e-01 1.05951071e+00 -2.70281464e-01
1.22360933e+00 -2.09361374e-01 -1.25275567e-01 6.73390627e-02
1.42836082e+00 3.68587039e-02 7.71480560e-01 2.79719025e-01
5.11848748e-01 4.49171275e-01 3.86850893e-01 1.98556378e-01
6.79522693e-01 1.27001703e-01 -3.09531182e-01 1.73101917e-01
-1.35575840e-02 -3.46873641e-01 3.96698058e-01 1.21322167e+00
1.44742966e-01 6.41509518e-03 -8.95120263e-01 8.37460339e-01
-1.82979095e+00 -1.05142951e+00 1.38037413e-01 1.87385440e+00
1.17668605e+00 1.40602021e-02 -3.30690324e-01 -1.51006550e-01
8.40362966e-01 -1.64580151e-01 -1.95149168e-01 -4.84083325e-01
-5.60247123e-01 7.73962915e-01 1.72109514e-01 5.78967631e-01
-1.07102811e+00 1.67988098e+00 6.73217916e+00 1.13995457e+00
-5.05313694e-01 3.90090123e-02 6.40401756e-03 7.19999671e-01
-3.48294586e-01 2.50973403e-02 -1.28279936e+00 2.16999017e-02
5.73336840e-01 -6.43562734e-01 1.00040197e-01 7.03736246e-01
-3.09792399e-01 -2.70296574e-01 -7.92593658e-01 8.90878022e-01
4.51413363e-01 -1.37796199e+00 6.03255808e-01 -6.12305105e-01
8.29876900e-01 -1.71997800e-01 -3.01926136e-01 9.46548462e-01
4.82492208e-01 -1.28195858e+00 1.07349686e-01 5.97545147e-01
3.86389315e-01 -8.02252531e-01 8.98434758e-01 5.22377610e-01
-1.65308356e+00 -2.49230023e-03 -4.78524685e-01 -3.26934993e-01
1.30972818e-01 2.38337904e-01 -6.39354289e-01 1.13852942e+00
4.50198650e-01 5.75940430e-01 -6.45690501e-01 6.23318553e-01
-7.92314172e-01 6.50433481e-01 7.53948763e-02 -2.87464619e-01
2.46071160e-01 -1.39759583e-02 2.52881616e-01 1.73102272e+00
-7.77850971e-02 4.86788332e-01 1.53455153e-01 9.56301391e-01
-2.40931734e-01 4.16622162e-01 -5.09607732e-01 -1.32957950e-01
6.23777151e-01 1.14123750e+00 -4.31374341e-01 -7.85488129e-01
-6.28004789e-01 9.19647336e-01 6.41104400e-01 4.57141936e-01
-5.23614943e-01 -9.76431012e-01 6.49286151e-01 -6.34816170e-01
8.98204803e-01 -1.92179456e-01 -2.54736423e-01 -1.36785769e+00
-9.62417573e-02 -8.37998807e-01 6.32527888e-01 -9.21490192e-01
-1.79163444e+00 6.57935679e-01 1.18869729e-01 -9.13113892e-01
1.27092442e-02 -9.45961773e-01 -6.41045511e-01 1.12214959e+00
-1.84649777e+00 -1.00941074e+00 2.26275790e-02 9.24333990e-01
3.54027033e-01 -4.52297449e-01 1.43209743e+00 7.01314509e-02
-4.26613629e-01 7.21669555e-01 -2.74107605e-01 6.76594317e-01
5.62935114e-01 -1.12141848e+00 3.25836122e-01 8.21732521e-01
3.60134512e-01 1.53846049e+00 1.01719936e-02 -8.89648438e-01
-1.05867934e+00 -1.18951344e+00 1.73457026e+00 -6.37291849e-01
9.51852143e-01 -2.36030966e-01 -1.25578272e+00 8.85669231e-01
4.05079186e-01 -3.15585881e-01 1.38797915e+00 2.62352139e-01
-8.49781811e-01 3.16254675e-01 -1.01191819e+00 6.47523046e-01
8.37571979e-01 -9.61043894e-01 -1.89671457e+00 1.14952512e-01
1.06847405e+00 -8.98274109e-02 -8.32304239e-01 6.75679386e-01
3.53622064e-02 -2.25695446e-01 9.47596908e-01 -1.17514586e+00
9.37828496e-02 -6.37920380e-01 -2.63426960e-01 -1.18400562e+00
-6.26096725e-01 -4.86141205e-01 -1.77305192e-01 1.30686712e+00
6.86985910e-01 -6.77724361e-01 3.56471151e-01 2.74152726e-01
8.13170969e-02 -5.04943967e-01 -6.30037189e-01 -1.11645567e+00
2.43597180e-01 -5.52356601e-01 7.51270294e-01 1.20871484e+00
3.72475326e-01 9.26398337e-01 -7.26739317e-02 4.35449928e-01
-2.58370861e-03 1.57195315e-01 2.53144622e-01 -8.08150113e-01
-3.12677085e-01 -4.15350288e-01 -2.61741161e-01 -1.19280899e+00
6.85821235e-01 -1.09181857e+00 -1.43298656e-01 -1.63333774e+00
2.86453485e-01 5.72549514e-02 -6.28405273e-01 8.15010369e-01
-8.39241207e-01 3.52923013e-02 1.73494220e-01 -4.52773362e-01
-8.40529978e-01 6.55562699e-01 8.23332548e-01 -5.42532615e-02
-1.11082174e-01 -2.88529366e-01 -1.29500389e+00 9.85123634e-01
7.48144984e-01 -2.17709467e-01 -1.78940803e-01 -4.42157000e-01
5.69157064e-01 -5.33040226e-01 -6.43989295e-02 -6.63388789e-01
5.36079824e-01 2.26544309e-02 2.92330116e-01 -7.27404892e-01
1.51663736e-01 -6.82228625e-01 -5.42756557e-01 3.30009043e-01
-3.80045980e-01 3.84719342e-01 7.04757690e-01 5.70456982e-01
-6.26220882e-01 -4.96907115e-01 2.40042880e-01 -4.57570285e-01
-1.30707467e+00 -1.15135990e-01 -6.77957654e-01 4.17801499e-01
7.19393790e-01 7.53597021e-02 -5.29513299e-01 -1.71767548e-01
-6.74658954e-01 4.57367241e-01 -5.33468187e-01 7.77632833e-01
9.32258010e-01 -1.38262093e+00 -8.65470111e-01 1.88921496e-01
5.84249198e-01 -2.22638890e-01 -2.21232567e-02 2.77108073e-01
-7.85662606e-02 8.06963086e-01 2.40207225e-01 2.88298547e-01
-1.32332087e+00 7.21711993e-01 -1.58961415e-02 -7.85422862e-01
-1.21794954e-01 1.23726380e+00 1.96149215e-01 -8.69248390e-01
2.65551746e-01 -6.70603335e-01 -7.94264317e-01 6.32904768e-02
1.07201874e+00 1.88260913e-01 1.21240839e-01 -4.61698979e-01
-6.40382588e-01 6.51235163e-01 -4.50846851e-01 2.45797075e-02
8.47835839e-01 5.68359941e-02 -6.54398918e-01 4.18319464e-01
8.43208313e-01 3.93063009e-01 6.56940639e-02 -8.58987927e-01
5.27001023e-01 -1.14341006e-01 -1.57222852e-01 -1.24205720e+00
-3.49414915e-01 5.01357079e-01 -4.32536937e-02 -2.55895913e-01
1.08417273e+00 -9.42100883e-02 8.20245504e-01 1.08093154e+00
2.99558789e-01 -1.11423373e+00 1.03863887e-01 1.40876853e+00
9.96846437e-01 -1.15205097e+00 2.71048304e-03 -6.77155614e-01
-9.43640292e-01 8.28197956e-01 8.18712354e-01 -1.05762362e-01
9.51905847e-01 3.03026944e-01 8.65312964e-02 -1.02437837e-02
-9.25452769e-01 -7.76758134e-01 6.62457883e-01 5.25571167e-01
5.21512449e-01 6.15478158e-02 -5.34805417e-01 1.41510558e+00
-2.91259140e-01 -7.93334618e-02 6.38561770e-02 1.07145011e+00
-6.89931750e-01 -1.21754420e+00 -3.19459081e-01 1.29865617e-01
-4.50665206e-01 -9.48790133e-01 -4.38664794e-01 6.58560932e-01
3.25053573e-01 1.07236838e+00 -2.27121845e-01 -3.86256844e-01
4.86619860e-01 5.98765910e-01 1.59328222e-01 -1.23139226e+00
-1.23546827e+00 -3.47953171e-01 2.15091243e-01 -1.93258077e-01
-4.48286206e-01 -9.45214406e-02 -1.61286497e+00 7.78128579e-02
-4.80608433e-01 6.07836723e-01 2.72939205e-01 1.20494926e+00
4.89254892e-01 4.73241568e-01 2.15680420e-01 1.54291049e-01
-4.49589103e-01 -1.14195120e+00 -5.83658934e-01 3.95043463e-01
1.05792485e-01 -3.61228317e-01 -1.52153522e-01 -2.53122523e-02]
|
[9.837624549865723, 9.442853927612305]
|
4e5025a2-3fb3-4aeb-a7ff-bfd88b218d4a
|
known-plaintext-attack-and-ciphertext-only
|
1905.13594
| null |
https://arxiv.org/abs/1905.13594v1
|
https://arxiv.org/pdf/1905.13594v1.pdf
|
Known-plaintext attack and ciphertext-only attack for encrypted single-pixel imaging
|
In many previous works, a single-pixel imaging (SPI) system is constructed as an optical image encryption system. Unauthorized users are not able to reconstruct the plaintext image from the ciphertext intensity sequence without knowing the illumination pattern key. However, little cryptanalysis about encrypted SPI has been investigated in the past. In this work, we propose a known-plaintext attack scheme and a ciphertext-only attack scheme to an encrypted SPI system for the first time. The known-plaintext attack is implemented by interchanging the roles of illumination patterns and object images in the SPI model. The ciphertext-only attack is implemented based on the statistical features of single-pixel intensity values. The two schemes can crack encrypted SPI systems and successfully recover the key containing correct illumination patterns.
|
['Xiaocong Yuan', 'Zhenwei Xie', 'Yang Gao', 'Shuming Jiao', 'Ting Lei']
|
2019-05-31
| null | null | null | null |
['cryptanalysis']
|
['miscellaneous']
|
[ 1.10278141e+00 -3.07707071e-01 3.67518127e-01 -1.85158879e-01
-2.98989862e-01 -7.46401608e-01 3.81402701e-01 -4.45433885e-01
-6.87392175e-01 4.38948900e-01 -4.53550190e-01 -4.97016281e-01
-3.62254456e-02 -1.00628936e+00 -6.39230430e-01 -1.32793319e+00
2.21866980e-01 -2.92051792e-01 1.26086518e-01 1.62324697e-01
5.76633990e-01 5.08678436e-01 -1.40986431e+00 3.54485452e-01
5.92408717e-01 1.01551795e+00 1.53401092e-01 1.03162360e+00
2.61393692e-02 6.49384081e-01 -6.61950827e-01 -4.88595277e-01
7.92446971e-01 -3.99652421e-01 -6.76272213e-01 2.59628203e-02
2.11568519e-01 -7.81640172e-01 -6.36462688e-01 1.55524731e+00
2.32673600e-01 -9.10634518e-01 3.00489426e-01 -1.52043402e+00
-3.04033399e-01 1.95103437e-01 -3.13863337e-01 -3.80027175e-01
3.52355152e-01 3.95551831e-01 4.37666267e-01 -1.78694159e-01
4.50200826e-01 7.10841477e-01 1.45308018e-01 1.23460732e-01
-1.03878939e+00 -8.94539118e-01 -5.53042710e-01 5.54260373e-01
-1.43776727e+00 5.37836440e-02 8.15649390e-01 -1.14603071e-02
3.80794823e-01 4.78781015e-01 5.92193305e-01 5.07067144e-01
6.38459504e-01 4.90487903e-01 2.04585505e+00 -6.22408688e-01
-2.78917938e-01 4.03426826e-01 6.29275069e-02 6.17501378e-01
5.74252248e-01 7.42344141e-01 -2.70145625e-01 -4.37925875e-01
6.28094792e-01 9.35247168e-02 -8.56743515e-01 -1.11778311e-01
-1.25979769e+00 2.98452169e-01 7.11368620e-02 6.11824682e-03
-1.72905609e-01 3.00499666e-02 2.87338439e-02 6.04769826e-01
-3.89099270e-01 5.56929827e-01 -2.18039721e-01 2.87744969e-01
-2.23334178e-01 -8.73783603e-03 9.62764025e-01 6.78693533e-01
1.02388227e+00 -8.04778337e-02 2.91708320e-01 -1.17998958e-01
7.26351887e-02 1.26608825e+00 1.61240160e-01 -7.64062047e-01
2.08877906e-01 2.29627937e-01 1.33237764e-01 -8.53747189e-01
1.46853417e-01 2.35880569e-01 -7.61649013e-01 7.92399168e-01
4.93670642e-01 -3.77750963e-01 -6.23900294e-01 1.39241624e+00
1.85597688e-01 3.39416444e-01 7.38417029e-01 5.09488702e-01
4.84877646e-01 7.93964505e-01 -3.70189101e-01 -3.30759645e-01
1.42043972e+00 -2.56457627e-01 -6.81498945e-01 3.05450588e-01
1.83395565e-01 -1.28765213e+00 5.85736394e-01 7.10078061e-01
-8.37513387e-01 -5.04588723e-01 -1.38317013e+00 2.19211996e-01
-1.35939419e-01 4.17840481e-02 4.32484090e-01 1.21348822e+00
-6.89966023e-01 2.78810620e-01 -3.21333170e-01 3.50154072e-01
1.36270955e-01 7.84208477e-01 -5.38180053e-01 -3.02296042e-01
-1.09826052e+00 4.39584553e-01 4.28932041e-01 -5.16067632e-02
-4.54105884e-01 -5.88060021e-01 -5.51745713e-01 6.59153378e-03
3.44178170e-01 -4.91142720e-01 8.84749293e-01 -9.44396973e-01
-1.64849579e+00 7.62599111e-01 -3.60666774e-02 -4.01016563e-01
2.31524467e-01 4.01951045e-01 -5.66912532e-01 6.04312718e-01
-3.81273270e-01 2.41738245e-01 1.23223805e+00 -1.62886786e+00
-6.24991417e-01 -2.96416968e-01 -8.86486098e-02 -3.87092791e-02
-8.48147646e-03 -4.79201647e-03 8.38287100e-02 -1.54200360e-01
1.78220108e-01 -1.07829010e+00 4.36125994e-02 7.51869455e-02
-6.79028034e-01 7.33830214e-01 1.42089641e+00 -2.25664109e-01
7.16190040e-01 -2.45324183e+00 -5.22555530e-01 4.85340923e-01
-3.55147012e-03 5.45938969e-01 -5.74894212e-02 6.14722550e-01
-2.72746444e-01 -1.09292045e-02 -2.03190938e-01 1.55841291e-01
-2.73968160e-01 3.40612888e-01 -8.37212443e-01 6.42229319e-01
-4.34844429e-03 6.19311750e-01 -5.92697620e-01 -2.78948635e-01
3.71488661e-01 4.24350083e-01 -1.24899983e-01 3.48968357e-01
2.34421268e-01 2.94001907e-01 -2.41939545e-01 3.63800585e-01
1.58725381e+00 6.35471568e-02 4.79053438e-01 -6.70639098e-01
-4.42784429e-01 -1.28685847e-01 -1.31530166e+00 5.87599993e-01
-1.06956899e-01 5.54054141e-01 4.11464274e-02 -4.43221748e-01
6.10413313e-01 5.65346718e-01 4.71900880e-01 -5.02808750e-01
2.45134607e-01 1.75817028e-01 2.54104257e-01 -6.31099582e-01
4.15065050e-01 -1.38625637e-01 6.95855319e-02 8.61248195e-01
-4.38677430e-01 -3.32590461e-01 -3.32636178e-01 1.07497294e-02
7.79231369e-01 -2.80983120e-01 2.55780369e-01 -3.89140435e-02
7.41484106e-01 -2.40719602e-01 3.38192999e-01 7.94167578e-01
2.22542256e-01 6.00469887e-01 4.45703328e-01 -2.95910686e-01
-8.54589939e-01 -9.40121233e-01 -2.69497484e-01 -1.06659979e-01
8.34002078e-01 -3.61785620e-01 -5.42394459e-01 -6.36811018e-01
-1.93687320e-01 2.69682229e-01 -8.81784782e-03 -2.10788306e-02
-3.46794248e-01 -5.65933406e-01 8.30335855e-01 -1.43022940e-01
1.32506990e+00 -8.21227014e-01 -5.19863486e-01 -1.55182943e-01
-3.27617787e-02 -1.49843121e+00 -2.32973501e-01 -2.34096035e-01
-3.46577764e-01 -1.59994364e+00 7.33526945e-02 -8.37993205e-01
1.15447509e+00 6.50463104e-01 2.40631968e-01 4.33440566e-01
-3.85555714e-01 4.01297808e-01 -8.18549842e-02 -6.68233395e-01
-6.74079776e-01 -6.85719550e-01 -1.60618156e-01 5.45757592e-01
1.63670570e-01 -2.44795516e-01 -7.34684825e-01 3.13284457e-01
-1.35663629e+00 2.44559988e-01 7.41254032e-01 8.14863265e-01
4.19329971e-01 7.43650019e-01 -3.49554539e-01 -1.05831409e+00
1.69943660e-01 2.95466751e-01 -9.73019779e-01 4.58988309e-01
-6.89087629e-01 2.40443245e-01 9.27112520e-01 -5.24326861e-01
-1.08535457e+00 1.80062443e-01 -1.58056721e-01 1.18669324e-01
-2.94729799e-01 8.89180079e-02 -5.08794010e-01 -1.00826788e+00
4.53880541e-02 7.85279810e-01 2.54657120e-01 1.85787175e-02
-2.11193543e-02 8.87779117e-01 8.03989053e-01 -3.60669911e-01
1.30595148e+00 1.00997663e+00 5.95374048e-01 -9.06861901e-01
-1.23161562e-01 -2.35137254e-01 -2.89412200e-01 3.35422903e-02
6.11817181e-01 -5.61246037e-01 -1.32603025e+00 1.46273148e+00
-1.22557294e+00 1.82988197e-01 -5.40075116e-02 6.14646196e-01
-2.92153567e-01 1.13964474e+00 -4.24343616e-01 -5.73094904e-01
-3.08203906e-01 -1.56642020e+00 8.06599379e-01 2.23862231e-01
6.13162637e-01 -5.58946073e-01 -3.64046305e-01 1.18156850e-01
2.99648762e-01 2.08560333e-01 9.26171303e-01 -1.12250887e-01
-1.29529417e+00 -5.98950803e-01 -5.57368338e-01 6.92180574e-01
3.33278179e-01 1.00440979e-01 -1.10781837e+00 -3.67192358e-01
7.84096122e-01 2.68182606e-02 4.60393995e-01 4.81759990e-03
1.33871377e+00 -5.41396677e-01 -3.68873104e-02 9.51773345e-01
1.86527991e+00 4.88032371e-01 1.36292470e+00 2.74350464e-01
4.13745612e-01 1.43217370e-01 2.87923843e-01 3.49718779e-01
6.02545477e-02 3.83782059e-01 4.79215175e-01 -2.66755670e-01
2.16455132e-01 3.71398181e-02 3.18302929e-01 3.56107056e-01
1.89149886e-01 -4.37684387e-01 -3.55839103e-01 -1.88140705e-01
-1.15154421e+00 -1.23812366e+00 -5.72486699e-01 2.44466949e+00
8.77066553e-01 -5.82727604e-02 -7.86152720e-01 3.97637248e-01
4.93800640e-01 1.72647089e-01 -5.35210408e-02 -1.92935839e-01
-4.46672589e-01 4.08378541e-01 1.03212273e+00 6.25716865e-01
-9.69427288e-01 5.55981457e-01 5.94114733e+00 5.89215457e-01
-1.50907528e+00 -3.55558008e-01 2.04674140e-01 6.41228855e-01
-3.45701158e-01 7.04706967e-01 -7.34655440e-01 5.86809695e-01
2.11598918e-01 -8.30717981e-02 4.41612601e-01 2.41286397e-01
-2.90505975e-01 -4.96493220e-01 -6.51377618e-01 1.05666459e+00
1.28705576e-01 -1.03975391e+00 1.32569775e-01 4.90191072e-01
5.31853437e-01 -6.36321902e-01 4.23451096e-01 -4.60965961e-01
-1.34985030e-01 -5.44167161e-01 2.40223601e-01 4.94573683e-01
9.11882937e-01 -4.33225960e-01 8.33046377e-01 2.03897879e-01
-8.82787406e-01 -5.40684611e-02 -3.66748840e-01 -3.69874574e-02
2.59783156e-02 2.32922539e-01 -7.19128311e-01 7.62400627e-01
3.73180836e-01 1.07396111e-01 -1.50388628e-01 9.63102877e-01
-4.89054471e-01 6.55361176e-01 -3.81906778e-01 6.97918385e-02
6.93806335e-02 -5.06217718e-01 5.91418862e-01 6.06698096e-01
3.35428149e-01 5.49810410e-01 6.89242855e-02 4.92686361e-01
4.42149490e-02 -1.53460860e-01 -6.70239866e-01 2.34141439e-01
3.05303007e-01 1.16204774e+00 -3.94199431e-01 -1.42116725e-01
-5.14854312e-01 1.15093815e+00 -7.47768402e-01 3.27266157e-01
-4.18017477e-01 -8.16205502e-01 7.11293757e-01 -5.07200100e-02
3.11348349e-01 -5.47705591e-01 -2.69357532e-01 -9.87285197e-01
-1.48520708e-01 -1.01805067e+00 -2.09775548e-02 -8.67374122e-01
-1.18357015e+00 2.84281135e-01 -5.10787815e-02 -1.39696562e+00
2.27182716e-01 -8.45784426e-01 -6.90221071e-01 9.73698497e-01
-2.06271219e+00 -1.10903394e+00 -2.62625098e-01 1.01055253e+00
-4.38000202e-01 -1.38926893e-01 1.04008043e+00 -1.79862902e-01
-1.02891564e-01 5.08724511e-01 3.80811036e-01 3.78400952e-01
6.82399809e-01 -8.67199421e-01 -6.45019934e-02 1.11896801e+00
4.29273397e-02 7.44746029e-01 5.04626274e-01 -4.33342904e-01
-1.76329696e+00 -4.90473777e-01 7.11040676e-01 2.43770912e-01
3.89280677e-01 -9.21607986e-02 -6.99493349e-01 3.12719494e-01
7.79655099e-01 -8.95828530e-02 6.62926853e-01 -9.60638046e-01
-5.21496356e-01 -4.19691414e-01 -1.41510999e+00 5.79573393e-01
1.75471470e-01 -9.06969309e-01 -1.75916284e-01 2.89289713e-01
4.02101249e-01 -3.72765303e-01 -5.79164267e-01 3.86830539e-01
7.18618691e-01 -8.08819711e-01 1.10116994e+00 1.09235406e-01
-1.33500621e-02 -7.50657380e-01 -1.71783760e-01 -5.40812790e-01
3.33547741e-01 -9.86852646e-01 5.11516631e-01 7.19947875e-01
-9.32684466e-02 -1.18388867e+00 4.96919632e-01 4.64835197e-01
3.57338250e-01 -1.26980349e-01 -7.16851115e-01 -7.02368379e-01
-5.59386492e-01 -3.60184699e-01 9.88246024e-01 5.24795830e-01
-3.88760865e-01 -3.19633454e-01 -5.95474124e-01 1.08720315e+00
1.02279806e+00 4.29217786e-01 1.02013123e+00 -8.88436794e-01
-2.89313048e-01 7.44917244e-02 -6.98918819e-01 -9.32881117e-01
3.12851332e-02 -4.56176519e-01 -7.66104981e-02 -6.17578804e-01
3.25126290e-01 -5.21136403e-01 -6.01086691e-02 4.21906501e-01
-1.86450094e-01 5.64300537e-01 3.02443951e-01 2.70808697e-01
1.19524345e-01 1.05057597e-01 1.22030759e+00 -2.18281880e-01
1.18357576e-01 1.55554563e-01 -5.70811570e-01 3.55890870e-01
1.04977179e+00 -6.65101171e-01 -5.36657691e-01 -2.32159525e-01
2.87745923e-01 -7.26812705e-02 7.92312801e-01 -1.21359217e+00
4.62187290e-01 -2.67237663e-01 2.44028363e-02 -4.04887259e-01
3.06761503e-01 -1.26601446e+00 5.02279937e-01 7.29090512e-01
1.57396212e-01 -1.05982020e-01 -8.33029076e-02 3.83382559e-01
-2.19721794e-01 -5.79854488e-01 8.24798882e-01 -8.21442157e-02
-5.93779266e-01 3.02804679e-01 -4.05278862e-01 -4.86656636e-01
1.03828371e+00 -5.75322628e-01 -7.66570151e-01 -2.56113410e-01
-9.36318040e-02 -3.82423460e-01 7.79205561e-01 -2.83796400e-01
1.06565380e+00 -7.61741757e-01 -5.09445131e-01 9.45999563e-01
-1.18219070e-01 -4.22828615e-01 3.12642634e-01 4.60351944e-01
-1.07590330e+00 1.56663507e-01 -2.27548838e-01 -5.24764597e-01
-1.94820559e+00 5.16135037e-01 3.70973736e-01 -1.50029466e-01
-3.55512887e-01 3.30186844e-01 7.61457607e-02 3.58482264e-02
-2.21731246e-01 1.38659878e-02 2.78497279e-01 -7.20018566e-01
7.97085822e-01 -8.24267268e-02 -1.95383325e-01 -5.45941949e-01
1.89310879e-01 9.82569337e-01 -1.11161262e-01 -2.20460862e-01
9.21650529e-01 -3.51447433e-01 -5.23661375e-01 -3.97522837e-01
1.51239240e+00 3.95043850e-01 -9.13318276e-01 -2.88755000e-01
-6.40518904e-01 -1.12265253e+00 1.74554512e-02 -5.31086862e-01
-1.03174853e+00 7.28908539e-01 7.86752284e-01 3.03538620e-01
1.47998321e+00 -7.07889020e-01 9.36794341e-01 4.70417529e-01
6.23993576e-01 -5.97409189e-01 -1.51666537e-01 2.71543443e-01
1.88085079e-01 -1.03152514e+00 2.26661265e-01 -8.22204530e-01
-6.96721435e-01 1.41817725e+00 1.59943104e-01 1.25435024e-01
8.71634781e-01 6.31274283e-01 2.27546737e-01 -1.26535550e-01
-3.35788101e-01 -2.13241111e-02 -6.80644363e-02 8.20135891e-01
-4.14939940e-01 -7.53276574e-04 -2.62217581e-01 7.01670647e-02
-1.03889070e-01 -4.69954982e-02 8.85221004e-01 1.15215445e+00
-7.92920366e-02 -1.82628882e+00 -6.50271833e-01 6.78373054e-02
-5.72151303e-01 -1.98550195e-01 -1.77806690e-01 5.57009220e-01
3.02507013e-01 8.52352977e-01 -1.33688360e-01 -6.75484776e-01
2.28550658e-03 -1.71266347e-01 5.25880933e-01 -1.81029737e-01
-4.55741286e-01 -2.97060400e-01 -4.40476745e-01 -3.44268888e-01
-6.41488671e-01 -4.97535348e-01 -1.27945483e+00 -5.26228189e-01
-5.47320664e-01 8.67545009e-02 8.12177598e-01 7.05654979e-01
1.53793469e-01 -1.82373837e-01 1.45362747e+00 -3.02377820e-01
-5.00203609e-01 5.89288659e-02 -7.88291693e-01 2.35913724e-01
4.72602934e-01 1.60252258e-01 -4.85743850e-01 9.86902639e-02]
|
[4.41495418548584, 7.989650249481201]
|
35d3b1fc-99c6-4299-be40-8863e55386bd
|
robust-uncertainty-estimation-for
|
2307.01325
| null |
https://arxiv.org/abs/2307.01325v1
|
https://arxiv.org/pdf/2307.01325v1.pdf
|
Robust Uncertainty Estimation for Classification of Maritime Objects
|
We explore the use of uncertainty estimation in the maritime domain, showing the efficacy on toy datasets (CIFAR10) and proving it on an in-house dataset, SHIPS. We present a method joining the intra-class uncertainty achieved using Monte Carlo Dropout, with recent discoveries in the field of outlier detection, to gain more holistic uncertainty measures. We explore the relationship between the introduced uncertainty measures and examine how well they work on CIFAR10 and in a real-life setting. Our work improves the FPR95 by 8% compared to the current highest-performing work when the models are trained without out-of-distribution data. We increase the performance by 77% compared to a vanilla implementation of the Wide ResNet. We release the SHIPS dataset and show the effectiveness of our method by improving the FPR95 by 44.2% with respect to the baseline. Our approach is model agnostic, easy to implement, and often does not require model retraining.
|
['Lazaros Nalpantidis', 'Evangelos Boukas', 'Frederik Scholler', 'Jonathan Becktor']
|
2023-07-03
| null | null | null | null |
['classification-1', 'outlier-detection']
|
['methodology', 'methodology']
|
[-3.51315588e-01 1.36416659e-01 2.82367945e-01 -5.53986669e-01
-1.25904596e+00 -5.13209879e-01 7.08505273e-01 6.58311099e-02
-9.43210363e-01 1.11276078e+00 1.88949093e-01 -1.00931570e-01
-2.71459579e-01 -5.84921658e-01 -1.11561465e+00 -4.71927971e-01
-3.66883546e-01 7.33808458e-01 4.39985275e-01 5.74447177e-02
9.61146355e-02 3.18401426e-01 -1.25597727e+00 1.62158340e-01
8.36141407e-01 1.10525787e+00 -6.05279207e-01 4.89407718e-01
2.87854463e-01 6.61251366e-01 -8.85001779e-01 -5.88078439e-01
4.04394567e-01 2.13533435e-02 -6.28669560e-01 -2.97689438e-01
6.65467560e-01 -4.04338062e-01 -4.03264910e-01 9.98148799e-01
5.05845845e-01 4.27258939e-01 7.21853495e-01 -1.20137429e+00
-4.80898023e-01 1.02451801e+00 -3.91376048e-01 6.06617630e-01
-2.60577321e-01 1.10873379e-01 8.33163023e-01 -6.10117376e-01
4.00436968e-01 1.17416000e+00 1.05172944e+00 2.77253896e-01
-1.41643393e+00 -9.91174996e-01 1.23351187e-01 5.57052940e-02
-1.41804695e+00 -3.68209958e-01 -2.11011968e-03 -3.75238001e-01
1.26340330e+00 -1.90444484e-01 2.90281773e-01 1.49369693e+00
2.93469429e-03 6.61643147e-01 1.02694595e+00 -1.42396137e-01
8.39278162e-01 -8.99801627e-02 3.49685431e-01 2.60503978e-01
4.71050233e-01 3.98022711e-01 -3.06631625e-01 -2.35563502e-01
4.95047837e-01 -1.91878945e-01 -7.95386359e-02 -4.51896749e-02
-7.41050899e-01 8.58943284e-01 4.49800164e-01 3.17207575e-01
-4.84553576e-02 7.71136940e-01 4.05506372e-01 3.60762894e-01
8.85430276e-01 7.09962845e-01 -6.63497210e-01 -3.07145596e-01
-1.39246404e+00 4.82525080e-01 8.68911982e-01 9.61060464e-01
3.64463627e-01 2.15830058e-01 -2.72486001e-01 5.44982016e-01
2.67787218e-01 2.78230458e-01 4.07543659e-01 -1.07277966e+00
4.59923595e-01 3.01432493e-03 2.53769875e-01 -2.76788086e-01
-5.01976669e-01 -7.74985909e-01 -4.00678009e-01 5.91104329e-01
8.13860536e-01 -4.23706055e-01 -1.35420966e+00 1.76230562e+00
-1.29421368e-01 8.02440524e-01 5.20982035e-03 4.40489680e-01
4.77392226e-01 3.83232832e-01 2.16227695e-01 3.15395951e-01
9.35003757e-01 -8.40626240e-01 -3.20719987e-01 -2.93324023e-01
6.48682117e-01 -4.24421161e-01 8.90594006e-01 6.54843271e-01
-8.20012748e-01 -2.17453733e-01 -1.25251949e+00 3.34884346e-01
-3.57366204e-01 -2.00819522e-01 5.10403216e-01 8.38209748e-01
-8.88609767e-01 1.25164485e+00 -1.24121284e+00 -9.52472724e-03
7.55002260e-01 4.60251242e-01 -3.73763263e-01 -2.83611212e-02
-1.35235429e+00 9.54216301e-01 6.82577491e-01 -3.01279217e-01
-1.07674003e+00 -1.12311459e+00 -8.30714703e-01 1.10008948e-01
3.25880349e-01 -4.72534180e-01 1.54054451e+00 -6.95277214e-01
-1.27193713e+00 3.38256210e-01 4.16865587e-01 -1.19909251e+00
1.05177379e+00 -5.83621740e-01 -3.11604559e-01 -1.99787691e-01
-1.44905105e-01 5.61246276e-01 4.96888191e-01 -8.25799406e-01
-4.87672806e-01 -1.14245325e-01 2.21348237e-02 -2.41497427e-01
3.15412357e-02 -6.30598962e-02 -3.48893911e-01 -7.27584243e-01
-2.95782894e-01 -8.59497368e-01 -3.29354823e-01 -3.74762505e-01
-3.20104629e-01 -9.25376266e-02 3.05569261e-01 -5.84086061e-01
9.35972571e-01 -2.19753098e+00 -3.68025512e-01 3.81135762e-01
-3.62116173e-02 2.35371023e-01 -5.13717420e-02 2.40154743e-01
-1.45486206e-01 3.35623085e-01 -8.13447177e-01 -6.19284630e-01
2.38453373e-01 2.60711253e-01 -3.02411646e-01 5.65893352e-01
4.98141348e-01 7.17483282e-01 -8.24482620e-01 -1.90366119e-01
1.75572380e-01 5.45552909e-01 -7.52082527e-01 -2.25552410e-01
-2.57767767e-01 3.12296361e-01 -1.57160386e-01 2.77164221e-01
7.76340961e-01 8.02685544e-02 -3.45816255e-01 2.67518044e-01
1.42288715e-01 3.56006056e-01 -1.32910252e+00 1.94413137e+00
-3.90501171e-01 7.15135872e-01 -3.39745253e-01 -5.48376262e-01
6.48126364e-01 1.94165632e-01 -4.66926657e-02 -5.48357725e-01
1.86914191e-01 2.87904531e-01 2.35091105e-01 -3.88480537e-02
2.33875886e-01 -2.11555541e-01 -1.55142188e-01 1.92396253e-01
3.94006610e-01 -2.16232911e-01 4.05306756e-01 2.14955926e-01
1.44545329e+00 3.05572748e-01 -7.03136325e-02 -5.36569953e-01
-2.00672626e-01 -2.57250100e-01 2.29923204e-01 1.06177342e+00
-3.00839454e-01 9.96913433e-01 6.06783092e-01 -2.74459660e-01
-1.05307317e+00 -1.34292984e+00 -6.70321822e-01 6.82522893e-01
-3.58513951e-01 -2.77900249e-01 -9.30766642e-01 -8.79126310e-01
2.63760477e-01 1.39258182e+00 -8.43353570e-01 -1.11006483e-01
-2.79827565e-01 -1.14157009e+00 9.22605574e-01 8.34724367e-01
5.25298476e-01 -8.58352840e-01 -3.68979365e-01 2.59568363e-01
2.83054769e-01 -1.15306377e+00 -9.45351645e-02 2.83038080e-01
-9.35372829e-01 -9.49626386e-01 -5.81869543e-01 1.64840333e-02
2.02418491e-01 -8.10985625e-01 1.43557703e+00 -2.98227727e-01
-2.23830253e-01 1.25395544e-02 -1.22383796e-01 -4.58591878e-01
-3.33316654e-01 2.01492697e-01 1.83650821e-01 -3.60325873e-01
2.79022127e-01 -8.03275108e-01 -4.27828699e-01 7.23691424e-04
-1.06473672e+00 -6.26852155e-01 5.31276882e-01 7.26868451e-01
3.19455951e-01 1.91187531e-01 7.71425962e-01 -1.30056655e+00
4.31765616e-01 -6.83750987e-01 -8.06889057e-01 -1.56119645e-01
-6.92688525e-01 5.44670999e-01 3.47097576e-01 -2.05846965e-01
-1.13483429e+00 -2.08011001e-01 -4.81442660e-01 -5.42728364e-01
-2.20547527e-01 2.22565621e-01 1.47406071e-01 1.97668985e-01
1.00695157e+00 -5.90523243e-01 -4.70105886e-01 -6.13209903e-01
5.04372954e-01 4.03712213e-01 6.61314249e-01 -5.24434447e-01
6.14456117e-01 4.25301760e-01 -2.52168745e-01 -3.77366275e-01
-9.12971973e-01 -1.03211917e-01 -3.46523076e-01 3.77153397e-01
5.87979138e-01 -1.10450065e+00 -4.63106632e-01 3.47815573e-01
-8.89451325e-01 -4.22729641e-01 -7.16679096e-01 8.44375849e-01
-4.16667908e-01 7.67193511e-02 -7.05257177e-01 -8.26558769e-01
-1.59409642e-01 -1.04121840e+00 9.78029728e-01 8.75154510e-02
-1.34928495e-01 -9.20889616e-01 2.87153512e-01 -1.78506747e-01
7.26671219e-01 4.32686538e-01 6.33011043e-01 -1.35032034e+00
-4.22357619e-01 -2.36430869e-01 -2.61702240e-01 5.60601830e-01
-5.04064500e-01 -8.24881811e-03 -1.42390835e+00 -1.38612255e-01
-1.51810169e-01 -3.55206519e-01 1.47070205e+00 4.40105408e-01
1.17461944e+00 1.07901752e-01 1.48474202e-02 4.70312566e-01
1.38934207e+00 -1.87733725e-01 8.81003737e-01 4.86108601e-01
2.63826579e-01 2.87576288e-01 1.58585429e-01 3.17533761e-01
1.41958222e-02 4.80626643e-01 5.59419513e-01 2.38766253e-01
-1.11631826e-01 -1.51687399e-01 2.98827618e-01 3.27878892e-01
1.15127154e-01 -2.04309344e-01 -8.99049222e-01 5.57896972e-01
-1.86719608e+00 -9.58721817e-01 3.50479037e-02 2.25484872e+00
8.02434087e-01 6.90859914e-01 -1.27163818e-02 -1.48809087e-02
4.69071656e-01 -7.14823678e-02 -5.18038630e-01 -2.88412690e-01
7.24219680e-02 5.26539028e-01 7.63912618e-01 6.49315834e-01
-1.31967843e+00 8.16721678e-01 7.03308678e+00 1.08402634e+00
-4.66003329e-01 2.42205963e-01 9.43589687e-01 -5.39968371e-01
-8.67961049e-02 2.04061512e-02 -7.60418236e-01 5.59741557e-01
1.58171690e+00 8.72148275e-02 2.87761927e-01 9.92009044e-01
-2.19177857e-01 -3.06338638e-01 -1.48416078e+00 6.25569999e-01
-6.25944957e-02 -1.23645437e+00 -5.74530900e-01 -1.27343222e-01
8.23754549e-01 6.27252758e-01 1.81307865e-03 8.39385986e-01
7.22916961e-01 -1.31131637e+00 7.51604259e-01 5.76695502e-01
5.03105879e-01 -1.02061236e+00 1.00167191e+00 1.94571987e-01
-4.87324089e-01 7.73601085e-02 -4.21395302e-01 -2.83766799e-02
1.32960185e-01 9.89510298e-01 -8.04871678e-01 4.30960506e-01
9.87882197e-01 4.72687691e-01 -7.38947093e-01 1.41439593e+00
-2.95423806e-01 9.71967995e-01 -9.95714903e-01 2.58082926e-01
4.16048795e-01 8.79821777e-02 4.58256632e-01 1.36862516e+00
1.61480308e-01 -4.66561049e-01 -9.83736590e-02 1.05170274e+00
-5.16184747e-01 -2.40974307e-01 -4.28899169e-01 2.18732178e-01
3.55784357e-01 8.91379952e-01 -6.07536077e-01 -4.00489926e-01
-1.04767881e-01 7.94039249e-01 3.95391196e-01 3.22884113e-01
-1.25697601e+00 -6.68276429e-01 4.81736690e-01 -1.11169912e-01
4.92743641e-01 1.50900543e-01 -1.04369670e-01 -1.17385745e+00
2.76811309e-02 -6.31014347e-01 3.32337379e-01 -7.17909753e-01
-1.61930442e+00 9.11333382e-01 4.38890696e-01 -8.14304829e-01
-5.30317783e-01 -6.19379938e-01 -5.17354906e-01 9.48766649e-01
-1.51631045e+00 -5.42273283e-01 1.21377304e-01 1.72532603e-01
4.37665939e-01 2.98850182e-02 7.57778168e-01 4.93050724e-01
-4.46955889e-01 7.61867046e-01 5.47239959e-01 1.26078829e-01
7.07045078e-01 -1.37506223e+00 9.94141638e-01 1.00184083e+00
5.19270301e-01 4.12357658e-01 8.51962984e-01 -5.92710853e-01
-4.65633899e-01 -1.13730431e+00 4.69246596e-01 -9.47935581e-01
9.03525352e-01 -6.04706168e-01 -9.29678142e-01 9.89292622e-01
1.19170703e-01 3.52510720e-01 3.01910818e-01 2.30758935e-01
-6.41843498e-01 1.59611881e-01 -1.55537689e+00 3.29266280e-01
1.08625412e+00 -2.64397353e-01 -8.51092458e-01 2.78998345e-01
9.76816773e-01 -5.18581688e-01 -8.13583434e-01 3.86306763e-01
3.58752102e-01 -1.22053945e+00 8.01691234e-01 -8.41648757e-01
2.51422495e-01 -2.16463104e-01 -2.20538169e-01 -1.54667222e+00
8.56328979e-02 -4.09441143e-01 -1.00313656e-01 1.30732310e+00
8.50519538e-01 -7.74648011e-01 6.53046906e-01 7.85481989e-01
-2.53808796e-01 -6.00012362e-01 -1.32998598e+00 -9.61069643e-01
4.26322132e-01 -1.07802868e+00 5.48869312e-01 5.52104533e-01
-4.36029285e-01 6.16809949e-02 -1.52943388e-01 2.96904624e-01
6.06759727e-01 -4.81410891e-01 3.32388788e-01 -1.30620050e+00
-6.59872770e-01 -2.26835772e-01 -7.24200308e-01 -4.21082765e-01
2.81117052e-01 -8.70400846e-01 -6.49102964e-03 -1.22107160e+00
5.08401319e-02 -3.16903144e-01 -5.00341535e-01 4.79356319e-01
8.28139111e-02 5.29861629e-01 1.09664090e-01 1.72559731e-02
-6.99354768e-01 5.75381041e-01 3.04542243e-01 -3.10416147e-02
5.70835955e-02 8.26186910e-02 -5.93010902e-01 9.84411955e-01
8.16504240e-01 -9.65105355e-01 -2.16588870e-01 -3.13546002e-01
2.00164840e-01 -5.73946238e-01 3.38479042e-01 -1.52387023e+00
2.40789689e-02 5.12797236e-01 5.71284413e-01 -2.58225471e-01
2.30125114e-01 -8.34654748e-01 1.02855109e-01 3.80423665e-01
-3.66299808e-01 1.22700438e-01 7.67971814e-01 8.37044477e-01
-1.94013957e-03 -4.47785258e-01 7.38482475e-01 6.90958500e-02
-5.14791429e-01 1.16710238e-01 -5.64117096e-02 6.36398733e-01
9.95077789e-01 4.08025414e-01 -4.85429078e-01 -3.30516130e-01
-1.06765807e+00 2.17580363e-01 3.91288787e-01 1.78300649e-01
1.54807478e-01 -9.75669742e-01 -9.01574552e-01 2.16490611e-01
1.32649183e-01 -4.79170643e-02 1.16874002e-01 7.25578964e-01
-5.54227352e-01 1.03492297e-01 9.86160785e-02 -6.78500235e-01
-5.55772722e-01 2.15243235e-01 7.01139748e-01 -5.32552540e-01
-7.05899358e-01 9.83993769e-01 -6.33436516e-02 -4.47231799e-01
4.66441154e-01 -7.57588267e-01 2.17121035e-01 -1.89506859e-01
5.99895000e-01 4.81104046e-01 5.10753095e-01 6.41756803e-02
-5.17768502e-01 9.00570080e-02 -1.44891396e-01 -6.27448738e-01
1.41228080e+00 2.60775685e-01 3.54217678e-01 5.94373286e-01
1.10671997e+00 1.07771963e-01 -1.72808683e+00 -2.44605150e-02
3.35157663e-01 -4.03950512e-01 1.44491300e-01 -1.17780399e+00
-9.46561277e-01 7.52893329e-01 7.89048374e-01 -1.16671342e-02
7.37241983e-01 -1.57107450e-02 3.13241690e-01 6.10190153e-01
3.16482514e-01 -1.03182936e+00 -2.78761178e-01 6.00396454e-01
6.25125647e-01 -1.24814498e+00 8.58237520e-02 1.48169607e-01
-7.32864380e-01 7.53361702e-01 3.35225970e-01 -7.22659647e-01
8.86687875e-01 6.71612740e-01 -1.67861670e-01 -1.06326398e-02
-7.04797268e-01 -1.64280221e-01 2.43814945e-01 4.37199593e-01
3.31600785e-01 -2.07070887e-01 8.29427391e-02 7.41500497e-01
-2.49950141e-01 1.69794649e-01 6.75152481e-01 5.86879253e-01
-3.09502184e-01 -8.09675992e-01 -1.68835726e-02 5.89262545e-01
-1.00505269e+00 -4.76462126e-01 1.14475317e-01 1.27696824e+00
-8.69059861e-02 8.40580761e-01 3.45237970e-01 -9.35563259e-03
5.16597986e-01 5.03431380e-01 4.53885287e-01 -5.27244866e-01
-6.84689403e-01 -2.62370914e-01 2.99746335e-01 -6.61597192e-01
-1.73257917e-01 -5.46790004e-01 -1.31349623e+00 -2.95639545e-01
-4.16187525e-01 2.64570355e-01 8.21970701e-01 1.12571108e+00
4.64967132e-01 7.76299536e-01 9.88569204e-03 -8.05683672e-01
-9.21448767e-01 -1.29813254e+00 -6.66875243e-01 3.89602482e-01
2.55889863e-01 -7.06165433e-01 -8.81423891e-01 -4.60857600e-01]
|
[7.536930561065674, 3.692758798599243]
|
b4ac9cd5-3fa6-4e40-b6c0-9c56317ce2bf
|
for-women-life-freedom-a-participatory-ai
|
2307.03764
| null |
https://arxiv.org/abs/2307.03764v1
|
https://arxiv.org/pdf/2307.03764v1.pdf
|
For Women, Life, Freedom: A Participatory AI-Based Social Web Analysis of a Watershed Moment in Iran's Gender Struggles
|
In this paper, we present a computational analysis of the Persian language Twitter discourse with the aim to estimate the shift in stance toward gender equality following the death of Mahsa Amini in police custody. We present an ensemble active learning pipeline to train a stance classifier. Our novelty lies in the involvement of Iranian women in an active role as annotators in building this AI system. Our annotators not only provide labels, but they also suggest valuable keywords for more meaningful corpus creation as well as provide short example documents for a guided sampling step. Our analyses indicate that Mahsa Amini's death triggered polarized Persian language discourse where both fractions of negative and positive tweets toward gender equality increased. The increase in positive tweets was slightly greater than the increase in negative tweets. We also observe that with respect to account creation time, between the state-aligned Twitter accounts and pro-protest Twitter accounts, pro-protest accounts are more similar to baseline Persian Twitter activity.
|
['Ashiqur R. KhudaBukhsh', 'Sujan Dutta', 'Adel Khorramrouz']
|
2023-07-07
| null | null | null | null |
['active-learning', 'active-learning']
|
['methodology', 'natural-language-processing']
|
[ 6.16862578e-03 1.04493523e+00 -6.78084314e-01 -5.09854019e-01
-9.95838404e-01 -6.49150848e-01 1.38452268e+00 9.56924796e-01
-7.74487853e-01 1.08655870e+00 1.03611386e+00 -3.34301800e-01
1.48540899e-01 -7.66657948e-01 -2.79061764e-01 -7.59857476e-01
1.03743032e-01 1.27697611e+00 -1.53471276e-01 -7.92301059e-01
3.79185915e-01 1.69076711e-01 -9.32541132e-01 2.82954037e-01
7.58631110e-01 1.05151525e-02 -5.73260486e-01 3.07888061e-01
-1.19887330e-01 1.40409827e+00 -9.58137333e-01 -6.86454475e-01
-9.45055783e-02 -2.66200542e-01 -1.20057750e+00 -2.04100519e-01
2.15289935e-01 -2.43531000e-02 -6.14174120e-02 6.18007302e-01
5.35648823e-01 -1.26812486e-02 8.90810907e-01 -1.24741161e+00
1.40229106e-01 1.71421874e+00 -7.48701632e-01 4.66221720e-01
3.12596530e-01 -4.06213962e-02 1.02704358e+00 -7.79654741e-01
1.21381295e+00 1.45691907e+00 7.41546452e-01 2.90734857e-01
-9.90173340e-01 -9.17397618e-01 -7.46179894e-02 2.14100063e-01
-9.64453280e-01 -9.02171731e-01 8.12516093e-01 -7.95356870e-01
4.81302947e-01 3.78157794e-01 8.23977411e-01 1.25525272e+00
-1.52638987e-01 6.62138820e-01 1.00138152e+00 -3.18455935e-01
-2.73433030e-02 3.70549321e-01 4.90775198e-01 6.61591738e-02
5.43348730e-01 -6.29674017e-01 -5.84731460e-01 -7.39286959e-01
-3.11671793e-01 -4.52381283e-01 2.38446653e-01 5.46625614e-01
-1.04577386e+00 1.31634247e+00 -6.19035400e-03 4.94203120e-01
-4.24212724e-01 -2.61147350e-01 8.18277419e-01 -1.66581497e-01
1.18054116e+00 4.89784420e-01 -1.70020327e-01 -5.58437467e-01
-1.02098596e+00 4.64947313e-01 1.00082409e+00 2.09082201e-01
6.10556424e-01 -4.60808054e-02 -6.96303025e-02 6.64179623e-01
2.62384653e-01 5.81791282e-01 -5.15732542e-03 -9.84917879e-01
8.63908887e-01 9.96119916e-01 1.60574332e-01 -1.55668104e+00
-4.09430385e-01 -1.14363074e-01 -3.95428538e-01 -1.38673112e-01
7.97157884e-01 -3.95383805e-01 -8.30255896e-02 1.47019720e+00
4.25047010e-01 -6.36237741e-01 5.09998202e-02 4.11420256e-01
9.10967529e-01 6.45482779e-01 5.49809813e-01 -8.22183192e-01
1.39330912e+00 -2.29434714e-01 -9.55196023e-01 -4.93405193e-01
5.99976659e-01 -7.83525288e-01 5.12647867e-01 -4.43760715e-02
-1.11865532e+00 5.43148704e-02 -6.98407233e-01 1.44469753e-01
-1.82876348e-01 -3.49474132e-01 2.65784591e-01 5.76614201e-01
-4.20171201e-01 4.34961736e-01 -7.89964557e-01 -4.21616614e-01
6.21093273e-01 8.31021145e-02 -1.34638399e-01 7.30791211e-01
-1.34262156e+00 1.04940677e+00 5.56058466e-01 -8.14766362e-02
-1.86585441e-01 -4.79503036e-01 -9.89856362e-01 -4.44602877e-01
4.23924476e-01 2.24428058e-01 1.17957795e+00 -1.13952243e+00
-7.52396226e-01 1.48615503e+00 -2.12498024e-01 -7.02903807e-01
7.36780405e-01 8.29255730e-02 -3.61057639e-01 4.10052836e-02
5.43140650e-01 3.42531800e-01 3.04640263e-01 -1.04967546e+00
-6.11989081e-01 -5.25505006e-01 7.10326955e-02 1.78557843e-01
-3.24278861e-01 7.90345490e-01 6.19785845e-01 -3.28738421e-01
-1.39515102e-01 -9.80604231e-01 3.03254481e-02 -9.50425506e-01
-5.81902921e-01 -6.30791068e-01 1.05292261e+00 -7.30299771e-01
1.56537890e+00 -1.90621257e+00 -3.85328174e-01 1.90461293e-01
5.41467965e-01 -4.09950987e-02 8.93437207e-01 7.43191779e-01
-1.19171038e-01 2.89848059e-01 -2.82875240e-01 -3.44753861e-01
-1.81283578e-02 2.68676251e-01 -5.22783458e-01 8.24038744e-01
1.65981874e-01 6.35584772e-01 -1.15706885e+00 -9.70324218e-01
-4.33858424e-01 5.50561510e-02 -1.46045506e-01 -4.00834262e-01
1.58699855e-01 7.42170095e-01 7.52343750e-03 6.34010375e-01
5.07952154e-01 -3.67526300e-02 7.28855550e-01 2.02243719e-02
-8.56903076e-01 8.01957965e-01 -4.60903883e-01 4.79946107e-01
2.34573171e-01 1.21809649e+00 2.47403800e-01 -9.52123344e-01
1.01237488e+00 2.90515363e-01 7.80691385e-01 -6.01015329e-01
6.53516531e-01 5.83714128e-01 3.97181332e-01 -1.37162790e-01
1.05186009e+00 -2.11816058e-02 -6.47149384e-01 8.29860270e-01
-5.07532954e-01 1.06785342e-01 6.07693851e-01 4.72441465e-01
3.94174784e-01 -2.59016305e-01 6.52261794e-01 -7.81569421e-01
5.45337737e-01 4.28068668e-01 6.99962497e-01 2.00505987e-01
-3.14915061e-01 2.03349307e-01 9.69398677e-01 -4.77415442e-01
-1.00289512e+00 -4.21071559e-01 -2.36493006e-01 1.26929533e+00
-7.97219798e-02 -6.21669531e-01 -5.11044979e-01 -5.32938600e-01
-3.36736500e-01 1.04611254e+00 -7.79949069e-01 2.41810873e-01
-1.40795600e+00 -1.21501839e+00 6.58028901e-01 5.42014278e-02
5.25617123e-01 -1.09685159e+00 -9.78087246e-01 1.88001290e-01
-8.89013469e-01 -8.90266716e-01 -2.24180371e-02 -8.52489769e-02
-4.79573160e-01 -1.27663076e+00 -3.06020617e-01 -4.50011283e-01
4.54545021e-01 -5.76022327e-01 1.03008604e+00 -1.73667446e-01
1.57222793e-01 -6.38435781e-02 -2.00495109e-01 -1.24584067e+00
-1.05203354e+00 3.37972492e-01 -1.03760324e-01 -1.34384871e-01
3.87922347e-01 -1.64059937e-01 -3.38300943e-01 -1.72019973e-01
-3.97640407e-01 -2.30677396e-01 -3.39493483e-01 4.18470234e-01
-8.44948292e-02 -4.65850532e-01 4.92410868e-01 -1.58448005e+00
6.81460559e-01 -1.00172722e+00 5.07107452e-02 -5.21501005e-01
-7.27619946e-01 -5.30201852e-01 -2.01826282e-02 -2.37154931e-01
-1.08091021e+00 -4.46334928e-01 -8.63276049e-02 6.78289711e-01
3.11111152e-01 6.64082468e-01 2.97329813e-01 9.51634884e-01
9.79658246e-01 -4.40709144e-01 2.60146528e-01 -3.87297459e-02
-2.21591607e-01 9.39281464e-01 2.64834791e-01 -3.25397015e-01
6.55702412e-01 5.84260762e-01 -2.73562312e-01 -8.87217700e-01
-8.74186337e-01 -3.96814793e-01 -6.90391123e-01 -6.14095747e-01
7.54095078e-01 -9.54026401e-01 -8.97469640e-01 5.36212325e-01
-1.21363342e+00 -4.70791191e-01 -3.54676574e-01 1.15568228e-01
-2.84202486e-01 -1.91622705e-03 -5.55695057e-01 -1.27723241e+00
-6.75805867e-01 -6.72414303e-01 6.72058165e-01 2.15816125e-01
-1.40806341e+00 -1.28257823e+00 5.67750514e-01 6.53898895e-01
2.15616524e-01 7.97251999e-01 8.62516046e-01 -1.49810958e+00
5.06722987e-01 4.36675325e-02 8.15363675e-02 -4.01326448e-01
2.50582963e-01 3.26616496e-01 -7.91172147e-01 -1.94194883e-01
-2.70660430e-01 -3.06108445e-01 3.29904854e-01 1.52092099e-01
7.69426078e-02 -1.10200560e+00 -4.44361180e-01 -3.61230999e-01
8.36497664e-01 2.82518804e-01 5.35470784e-01 8.51524830e-01
2.96933472e-01 8.98628652e-01 7.52614975e-01 6.17716610e-01
6.52288437e-01 4.59308684e-01 7.17466697e-02 -1.19562848e-02
9.05090049e-02 4.34301645e-02 6.34848475e-01 6.98075056e-01
-4.26182896e-01 -2.21577268e-02 -1.61782050e+00 6.53710127e-01
-1.86385572e+00 -1.41934037e+00 -6.86764956e-01 1.87554812e+00
1.17695272e+00 4.19222444e-01 7.30322659e-01 4.73439366e-01
7.82274127e-01 5.74113250e-01 8.44528750e-02 -4.86336261e-01
-3.22509080e-01 5.18765040e-02 4.60820973e-01 8.48265529e-01
-1.26716018e+00 7.03419924e-01 6.19595051e+00 5.05093873e-01
-9.96994555e-01 4.11564440e-01 8.43803108e-01 1.85955931e-02
-1.20610535e-01 1.10414989e-01 -8.63567710e-01 6.28713250e-01
1.27163148e+00 -6.78473592e-01 -4.43484664e-01 5.98699570e-01
3.31486344e-01 -3.46142679e-01 -5.76517820e-01 7.15127230e-01
1.85253978e-01 -1.49721718e+00 -4.82660025e-01 4.84359175e-01
4.94291276e-01 -3.86874788e-02 -3.00670207e-01 3.06617498e-01
3.51408780e-01 -7.43403614e-01 1.26125813e+00 3.14182132e-01
7.07144618e-01 -9.17106569e-01 9.62118089e-01 5.27433276e-01
-5.90123415e-01 -1.81001231e-01 3.11110705e-01 -6.22067213e-01
4.21326071e-01 4.39153999e-01 -1.59580135e+00 1.70597777e-01
3.95541519e-01 6.15987420e-01 -3.78744334e-01 2.83861667e-01
-5.92937097e-02 1.13235617e+00 -1.96315184e-01 -8.20803344e-02
2.56529629e-01 -2.13567078e-01 1.07733178e+00 1.40213835e+00
-1.41315192e-01 1.49689436e-01 2.17962280e-01 4.18775946e-01
-1.07785009e-01 3.47508192e-01 -6.06283605e-01 -4.12926257e-01
7.61596680e-01 1.37126696e+00 -1.10050201e+00 -4.80590254e-01
1.96709841e-01 8.33430514e-02 1.86545223e-01 -6.78272024e-02
-7.68735290e-01 -1.18560150e-01 3.54377985e-01 9.42358077e-01
-6.18108273e-01 -8.02540183e-02 -4.58473891e-01 -6.21131957e-01
-6.88248158e-01 -8.78555894e-01 4.43616539e-01 -7.78232887e-02
-8.68417859e-01 4.35879707e-01 4.65364724e-01 -9.39247072e-01
-6.53397620e-01 1.40108004e-01 -7.90358484e-01 5.92832446e-01
-7.22216845e-01 -1.13075662e+00 -1.08608887e-01 3.62597257e-02
4.80147243e-01 -3.34792823e-01 7.66699135e-01 4.02056694e-01
-5.41449308e-01 4.49806571e-01 -4.13751364e-01 7.17936575e-01
8.69842708e-01 -1.00168836e+00 2.24152535e-01 5.31872332e-01
-3.60609263e-01 4.73296165e-01 1.26839840e+00 -1.07664311e+00
-5.13955653e-01 -6.51986361e-01 1.69403660e+00 -7.00647056e-01
9.74751413e-01 -2.35650584e-01 -4.38189864e-01 8.63252223e-01
5.91173351e-01 -8.40115368e-01 9.64564264e-01 2.01270074e-01
-2.77784336e-02 2.77479947e-01 -1.11314404e+00 5.50141692e-01
7.01892495e-01 -1.56442434e-01 -8.40239108e-01 3.95383179e-01
2.29568675e-01 -3.62555504e-01 -4.32659596e-01 1.82485148e-01
4.08023596e-01 -7.73151517e-01 5.84936798e-01 -6.95456088e-01
6.35588467e-01 2.52829701e-01 1.74817428e-01 -9.24249589e-01
-3.02907348e-01 -6.88719988e-01 2.24463552e-01 1.65113091e+00
7.71223426e-01 -6.95829272e-01 6.51396930e-01 4.97252613e-01
-7.97317699e-02 -3.65552396e-01 -9.91698444e-01 7.31549971e-03
4.41356122e-01 -2.24479556e-01 1.60040721e-01 1.47862017e+00
4.61721599e-01 7.04577744e-01 -2.10124835e-01 -5.90945065e-01
5.47899365e-01 -1.28191531e-01 7.62126505e-01 -1.76625085e+00
2.74516910e-01 -4.87123996e-01 -1.07343152e-01 1.12040579e-01
3.46930265e-01 -8.66707742e-01 -1.48165688e-01 -1.48104167e+00
5.57063162e-01 -4.26425219e-01 6.25240564e-01 6.47679687e-01
1.70773233e-03 5.61763525e-01 2.31677219e-01 7.27627158e-01
-3.17207038e-01 -2.98571810e-02 6.54269636e-01 -3.56626481e-01
-4.25396115e-01 -9.20198187e-02 -1.02317595e+00 1.17103791e+00
7.15722084e-01 -6.84663892e-01 1.77532919e-02 2.72704959e-01
9.40087855e-01 -2.25321755e-01 5.42659275e-02 -4.08613324e-01
-6.64065257e-02 -2.83809453e-01 8.87565911e-02 -7.89069295e-01
2.54425138e-01 -2.85934210e-01 2.26894751e-01 6.42346740e-01
-3.08019429e-01 2.21154671e-02 9.46959108e-02 -6.22050613e-02
-1.23082802e-01 -1.32753953e-01 7.05868781e-01 -1.18172430e-01
-4.92767543e-02 -1.33117437e-01 -1.02025342e+00 4.21961784e-01
1.03701138e+00 -3.05952340e-01 -7.55256474e-01 -5.66374660e-01
-7.68948138e-01 1.03417896e-01 3.29923749e-01 6.31678849e-02
-1.80606470e-01 -1.15936553e+00 -1.38713419e+00 -4.07893568e-01
-8.73233229e-02 -3.17290574e-01 -1.49067700e-01 1.44796860e+00
-6.39123023e-01 8.58801156e-02 -1.15067750e-01 -1.44588917e-01
-1.65290999e+00 -2.18319386e-01 -1.69511303e-01 -4.15810883e-01
-3.11571568e-01 4.51603353e-01 -3.44024748e-01 -5.70207238e-01
-2.33658686e-01 5.18370748e-01 -9.07870829e-01 1.47710776e+00
4.73787695e-01 9.45572495e-01 -1.36180565e-01 -1.52300870e+00
-6.24397635e-01 -1.89944968e-01 1.16552472e-01 -3.76231194e-01
1.47230077e+00 -3.17223847e-01 -5.51107049e-01 1.14201677e+00
9.55393672e-01 8.68832529e-01 -3.55307341e-01 1.78080738e-01
5.28114200e-01 -2.99876958e-01 -5.73027611e-01 -5.57566583e-01
-4.65682894e-01 1.97571129e-01 -7.78804794e-02 7.09433138e-01
3.23455393e-01 4.78924513e-01 4.72011983e-01 3.24588493e-02
1.53323293e-01 -1.39581513e+00 -1.56321079e-01 9.62465763e-01
9.54345644e-01 -1.15369344e+00 4.56325561e-01 -4.32044834e-01
-8.36913466e-01 1.17311263e+00 3.91405493e-01 5.79898097e-02
3.01642805e-01 3.20432514e-01 3.66042554e-01 -5.32929182e-01
-4.95532066e-01 1.64062351e-01 -1.54276520e-01 1.76853195e-01
1.01787806e+00 1.38471603e-01 -1.17770290e+00 5.82489610e-01
-9.37244534e-01 -7.37247229e-01 9.51407969e-01 6.97788596e-01
-4.37812388e-01 -9.20935750e-01 -7.06861913e-01 2.50003338e-01
-8.17248940e-01 1.92217156e-01 -9.74052072e-01 1.25135720e+00
2.76509672e-01 9.18183625e-01 8.34712327e-01 -1.67938501e-01
-6.50670193e-03 3.65329325e-01 1.45249650e-01 -6.31560445e-01
-9.75971460e-01 1.22359917e-01 1.17361546e+00 1.14783153e-01
-9.02795911e-01 -1.21226096e+00 -1.48987472e+00 -7.80619025e-01
-4.23539057e-02 5.19356906e-01 2.72276074e-01 1.05171645e+00
-1.35904908e-01 3.51538621e-02 5.13274789e-01 -7.75709927e-01
-1.08409457e-01 -1.47048438e+00 -3.46898109e-01 5.08998334e-01
1.40900627e-01 -4.60416526e-01 -5.66213906e-01 -2.36341506e-02]
|
[8.966416358947754, 10.068723678588867]
|
2a38b44b-acbb-4ece-9fd2-435e33a6b215
|
practical-transformer-based-multilingual-text
| null | null |
https://aclanthology.org/2021.naacl-industry.16
|
https://aclanthology.org/2021.naacl-industry.16.pdf
|
Practical Transformer-based Multilingual Text Classification
|
Transformer-based methods are appealing for multilingual text classification, but common research benchmarks like XNLI (Conneau et al., 2018) do not reflect the data availability and task variety of industry applications. We present an empirical comparison of transformer-based text classification models in a variety of practical monolingual and multilingual pretraining and fine-tuning settings. We evaluate these methods on two distinct tasks in five different languages. Departing from prior work, our results show that multilingual language models can outperform monolingual ones in some downstream tasks and target languages. We additionally show that practical modifications such as task- and domain-adaptive pretraining and data augmentation can improve classification performance without the need for additional labeled data.
|
['Michele Banko', 'Cindy Wang']
|
2021-06-01
| null | null | null |
naacl-2021-4
|
['multilingual-text-classification']
|
['miscellaneous']
|
[-5.55405058e-02 -3.41470510e-01 -5.82139969e-01 -6.18672013e-01
-9.73494351e-01 -9.15687323e-01 9.48373497e-01 3.19258869e-01
-8.16145778e-01 9.08750117e-01 2.49544710e-01 -9.51203644e-01
6.15519173e-02 -3.39002252e-01 -4.22898024e-01 -1.42860964e-01
3.89185846e-01 7.26307929e-01 -1.40249148e-01 -5.82089841e-01
-5.37975924e-03 5.73545741e-03 -1.07286441e+00 5.68818927e-01
1.01972258e+00 8.04386556e-01 1.17807731e-01 6.23107255e-01
-3.10361087e-01 7.60120511e-01 -4.85553086e-01 -7.63053715e-01
3.04453164e-01 -7.13676587e-02 -9.57984865e-01 -6.62014782e-02
8.62526178e-01 -2.36524660e-02 -1.88830495e-01 8.44736695e-01
4.05988246e-01 -6.23884797e-02 5.92410147e-01 -1.11362851e+00
-8.52145016e-01 1.33212793e+00 -3.93826485e-01 2.86700070e-01
2.36769572e-01 -1.29651383e-01 1.21623194e+00 -1.13086128e+00
5.72142601e-01 1.37378931e+00 9.91656065e-01 1.54583439e-01
-1.21114957e+00 -9.23712790e-01 5.35034060e-01 1.79377615e-01
-1.20726109e+00 -5.51838934e-01 5.67938745e-01 -4.18547571e-01
1.06409895e+00 1.20182103e-02 1.54951692e-01 1.50863791e+00
4.03709203e-01 1.05106688e+00 1.39844334e+00 -6.99710488e-01
-4.15451199e-01 4.80295122e-01 3.90738815e-01 5.42516172e-01
1.28033295e-01 -1.68292239e-01 -6.53490722e-01 -1.22558355e-01
3.19057435e-01 -2.69712001e-01 5.03597707e-02 -3.26521814e-01
-1.53317261e+00 9.56670642e-01 7.64810592e-02 4.83441472e-01
9.06793624e-02 -2.50852197e-01 9.83687222e-01 8.09928954e-01
9.65251029e-01 2.97101766e-01 -1.05714333e+00 -7.13746101e-02
-1.01248896e+00 5.15580140e-02 7.34746277e-01 1.18623114e+00
6.47377193e-01 2.44815171e-01 -4.76442650e-02 1.26169479e+00
7.30678290e-02 5.71113944e-01 8.86069119e-01 -3.76319349e-01
1.06101418e+00 4.50724989e-01 -3.56474191e-01 -4.23059076e-01
-4.55819190e-01 -6.30112827e-01 -8.54652345e-01 -3.57246667e-01
7.47814298e-01 -2.33557969e-01 -6.55726314e-01 1.45101380e+00
-9.95245799e-02 -4.57831115e-01 1.93674788e-01 4.01484758e-01
6.52487278e-01 4.75262880e-01 3.22714090e-01 4.13348563e-02
1.43577945e+00 -1.23754430e+00 -7.44822562e-01 -5.63292980e-01
1.24444354e+00 -1.20585513e+00 1.59226441e+00 4.76212114e-01
-7.31870711e-01 -4.97094095e-01 -8.24076712e-01 -4.80741054e-01
-6.86709702e-01 5.33509910e-01 7.98847139e-01 8.36228371e-01
-8.33556771e-01 5.25150299e-02 -5.85174799e-01 -6.60570562e-01
2.02850625e-01 1.06232747e-01 -4.66389477e-01 -2.31171831e-01
-1.43780935e+00 1.05646145e+00 4.28528696e-01 1.35131739e-02
-6.47321939e-01 -7.89892316e-01 -8.29155982e-01 -3.02186221e-01
2.05111951e-01 -4.64065492e-01 1.52319908e+00 -9.58545685e-01
-1.39364326e+00 1.07462525e+00 -2.73994580e-02 -4.67061758e-01
7.11256742e-01 -3.27984422e-01 -4.35916662e-01 -4.75069046e-01
3.28559339e-01 6.04615390e-01 5.05845368e-01 -8.64967763e-01
-9.59224224e-01 -4.21725512e-01 -2.09073946e-02 5.51053584e-01
-8.95415187e-01 3.11877102e-01 -2.60786116e-01 -1.00801361e+00
-2.72131652e-01 -8.20926666e-01 -1.29861206e-01 -4.71030176e-01
-4.23060775e-01 -3.85251194e-01 7.86375880e-01 -6.97545588e-01
9.57966745e-01 -1.59939659e+00 -1.22111998e-01 -1.52568042e-01
-1.11185992e-02 5.76051064e-02 -2.96918243e-01 4.91264611e-01
-2.49645654e-02 1.71051145e-01 6.99326247e-02 -5.90789914e-01
1.78667560e-01 5.85492514e-02 -4.16987747e-01 4.47335452e-01
-1.21411741e-01 9.55153286e-01 -7.20818222e-01 -5.18081248e-01
2.14066550e-01 3.95357490e-01 -4.32889640e-01 -3.06984574e-01
-1.90403953e-01 5.78177154e-01 -1.47600412e-01 5.29340744e-01
3.23466510e-01 -8.78470542e-04 5.01226306e-01 -2.05804765e-01
-1.02157004e-01 9.50659573e-01 -8.18995833e-01 1.81291759e+00
-1.05295360e+00 9.92729664e-01 8.49943906e-02 -1.05373836e+00
6.99255049e-01 4.51324999e-01 2.23997280e-01 -8.24744940e-01
1.82674542e-01 2.77138591e-01 -5.67335375e-02 -5.74114099e-02
7.41308570e-01 -1.99613303e-01 -4.46802080e-01 6.95148408e-01
2.58410543e-01 -3.90060917e-02 4.10738438e-01 1.41853139e-01
5.35055339e-01 7.44838342e-02 4.29656118e-01 -8.24726582e-01
5.53413332e-01 1.67140782e-01 2.78669596e-01 7.49678671e-01
-1.21080726e-01 7.09498161e-03 7.86133632e-02 -4.93593216e-01
-1.25406742e+00 -6.79445505e-01 -6.17481649e-01 2.05694127e+00
-4.53810811e-01 -5.08602202e-01 -3.89511228e-01 -1.05851626e+00
5.88653386e-02 5.91302574e-01 -3.76094162e-01 1.63979024e-01
-5.41799664e-01 -1.06625843e+00 8.71497810e-01 5.20369172e-01
1.88375488e-01 -7.72103727e-01 1.59999505e-01 2.07541853e-01
-4.03163135e-01 -1.55303812e+00 -7.07404315e-01 6.63692474e-01
-8.98566723e-01 -8.53325546e-01 -4.56051737e-01 -1.08019400e+00
3.82674605e-01 1.24006994e-01 1.46496665e+00 -4.52715635e-01
1.17121123e-01 3.25128466e-01 -3.62976968e-01 -3.85117888e-01
-6.27796352e-01 8.88026893e-01 1.46458060e-01 -3.86406422e-01
5.34976006e-01 -2.51022339e-01 1.00879274e-01 3.69467527e-01
-5.98725498e-01 5.54586947e-02 4.31657463e-01 1.06534910e+00
3.08231115e-01 -1.22229718e-01 7.10698545e-01 -1.28260458e+00
9.72804606e-01 -4.17567015e-01 -5.30240238e-01 4.20704305e-01
-8.75552356e-01 1.86178982e-01 9.41581130e-01 -6.28391922e-01
-9.03187096e-01 -3.50688994e-01 -9.12281796e-02 -8.56036842e-02
-1.18456647e-01 8.12809885e-01 -7.91391581e-02 9.75600109e-02
8.56729329e-01 1.98743105e-01 -4.02296126e-01 -6.10697865e-01
4.65032071e-01 8.45185876e-01 2.48852491e-01 -9.46759224e-01
6.84899509e-01 1.13216586e-01 -4.83367175e-01 -5.56416810e-01
-1.12806106e+00 -3.55438560e-01 -1.05467606e+00 2.09722221e-01
4.70560223e-01 -1.28221917e+00 -3.39642137e-01 3.11308026e-01
-9.55225885e-01 -6.26699924e-01 -1.04514219e-01 5.84605992e-01
-3.16590875e-01 2.72880852e-01 -9.30626571e-01 -3.87898207e-01
-4.76108432e-01 -1.28347373e+00 1.04901707e+00 -5.34100354e-01
-2.18202710e-01 -1.71673453e+00 -2.33773831e-02 5.25398016e-01
4.24829721e-01 -4.35685188e-01 9.08539474e-01 -9.84099627e-01
1.05364881e-01 -1.44252807e-01 -5.72506338e-02 4.27783102e-01
2.49429584e-01 -1.68794245e-01 -8.94154191e-01 -5.54540157e-01
-2.01151058e-01 -8.20356607e-01 6.56143069e-01 1.87899917e-01
7.84050703e-01 -3.11539650e-01 -2.53172278e-01 5.34990668e-01
1.02668118e+00 -1.20399512e-01 -4.69742008e-02 6.59322381e-01
1.08360445e+00 7.11331904e-01 6.22249782e-01 5.88048361e-02
8.26640785e-01 7.30401516e-01 -2.05444336e-01 -2.32649967e-01
-1.00775532e-01 -3.12460959e-01 5.88019550e-01 1.26334441e+00
4.26858872e-01 -3.96637738e-01 -1.18584716e+00 5.84535837e-01
-1.66145909e+00 -5.66031635e-01 -2.26635844e-01 2.06078839e+00
1.29509485e+00 1.02928698e-01 2.64141619e-01 5.89678846e-02
6.05392694e-01 -8.33973736e-02 -3.42975289e-01 -4.39125478e-01
-4.18543518e-01 3.14162940e-01 7.44856000e-01 7.06953466e-01
-1.25674272e+00 1.48523962e+00 7.22948551e+00 9.15703416e-01
-1.18035901e+00 4.78036195e-01 6.72495961e-01 -3.64033990e-02
-2.61139452e-01 -6.62680492e-02 -1.20453978e+00 3.78165394e-02
1.15885043e+00 -2.67839700e-01 2.90848374e-01 7.30853498e-01
-1.09620400e-01 4.85329926e-01 -1.39520216e+00 8.72582614e-01
1.45323679e-01 -8.94378364e-01 1.42389998e-01 -7.86665305e-02
7.39902258e-01 4.93022114e-01 2.00961038e-01 6.91948950e-01
9.10591304e-01 -9.66839612e-01 1.01628029e+00 -2.44253442e-01
9.12447274e-01 -6.65953994e-01 5.81291497e-01 3.58860314e-01
-1.19363463e+00 -1.16729587e-02 -2.58545548e-01 -3.63416933e-02
-1.29236996e-01 5.70282221e-01 -8.04493368e-01 6.10298514e-01
7.07683563e-01 1.05733490e+00 -8.15971136e-01 2.92809278e-01
-1.80410221e-01 8.66764188e-01 -2.72327244e-01 2.10041329e-01
3.29063058e-01 -5.60449325e-02 2.71135688e-01 1.62376904e+00
1.61507837e-02 -7.80812204e-01 5.13303757e-01 1.86987266e-01
-3.45240295e-01 7.48300612e-01 -7.23929346e-01 -6.22708946e-02
4.77649361e-01 1.22002828e+00 -4.71662015e-01 -3.40678811e-01
-6.81266963e-01 7.01394558e-01 5.85266948e-01 3.31901878e-01
-4.73728687e-01 -2.08332986e-01 5.39769232e-01 -2.25896910e-02
5.84917478e-02 -4.41109955e-01 -4.74378586e-01 -1.54158008e+00
-8.30485893e-04 -1.42599261e+00 6.21523857e-01 -2.84792125e-01
-1.60665274e+00 7.59003401e-01 4.40131761e-02 -1.04277849e+00
-4.69561189e-01 -9.28985178e-01 -9.14252996e-02 9.22820628e-01
-1.73803437e+00 -1.68206561e+00 1.98072106e-01 7.65546501e-01
8.12482119e-01 -6.71600461e-01 7.64747441e-01 6.22235298e-01
-7.01818526e-01 1.11863124e+00 4.54237729e-01 3.31724107e-01
1.31354249e+00 -1.27126098e+00 5.78885913e-01 7.33786523e-01
4.18552697e-01 5.16440570e-01 4.49838668e-01 -3.64649177e-01
-1.18521500e+00 -1.22630823e+00 1.27372837e+00 -8.68551314e-01
1.22313571e+00 -9.56443965e-01 -6.55719221e-01 1.51087999e+00
6.05811536e-01 -1.91207707e-01 7.45044947e-01 8.20101321e-01
-7.03197181e-01 -1.73441365e-01 -6.56939805e-01 6.84149802e-01
9.12183881e-01 -8.30773234e-01 -3.46488655e-01 7.12522686e-01
4.46826071e-01 -2.68402725e-01 -1.16917598e+00 3.55941713e-01
5.05462945e-01 -5.72776139e-01 8.38572085e-01 -8.54223669e-01
2.56489187e-01 3.00351411e-01 -2.09907249e-01 -1.48952484e+00
-1.61859915e-01 -3.06626260e-01 4.47875202e-01 1.35179007e+00
6.60849333e-01 -9.32019472e-01 3.86962771e-01 2.41266396e-02
-2.64089316e-01 -2.63439864e-01 -8.08699131e-01 -9.37203467e-01
1.05468369e+00 -6.92848861e-01 2.42832944e-01 1.55477953e+00
2.42863998e-01 9.38310742e-01 -2.28509501e-01 -2.21750990e-01
5.60502052e-01 -4.62083407e-02 7.63009012e-01 -1.27956057e+00
1.87782235e-02 -6.54000759e-01 2.02486906e-02 -1.12729633e+00
8.03174555e-01 -1.52683616e+00 -1.78669855e-01 -1.25735629e+00
1.38899818e-01 -8.08336914e-01 -2.31718555e-01 8.95044386e-01
-2.36831501e-01 2.67022938e-01 -1.38451150e-02 3.16057146e-01
-4.73704994e-01 5.02698421e-01 9.37820733e-01 -4.44628894e-01
6.93295151e-02 -1.77114248e-01 -8.26964498e-01 3.94830465e-01
7.92473078e-01 -4.58480150e-01 -6.03087485e-01 -8.58996689e-01
2.53834277e-01 -2.81750649e-01 -1.25387356e-01 -6.10400558e-01
8.09843689e-02 -5.27845584e-02 1.83442742e-01 -3.10054332e-01
-1.24221593e-01 -7.67612159e-01 -4.00469929e-01 1.69483751e-01
-6.78460300e-01 5.24360597e-01 4.75044817e-01 1.77152231e-01
-3.27498436e-01 -4.13254686e-02 6.95898354e-01 1.10286959e-01
-3.06335390e-01 1.92546040e-01 -6.27973616e-01 3.79973233e-01
6.17530227e-01 8.43803957e-02 -4.74148452e-01 -2.36583203e-01
-4.45439398e-01 4.15645361e-01 2.30193362e-01 8.67049396e-01
-1.07145235e-01 -1.34238756e+00 -1.11647427e+00 1.67693526e-01
3.97882879e-01 -4.37424332e-01 -2.84288257e-01 8.59929383e-01
-2.23159999e-01 1.01235390e+00 -1.33620918e-01 -6.91500723e-01
-1.14833665e+00 3.94052953e-01 4.43059891e-01 -8.56028259e-01
-3.76621097e-01 5.09126782e-01 3.26260477e-01 -1.31087732e+00
2.45986775e-01 -5.19889534e-01 -1.31179318e-01 1.08312771e-01
1.43620044e-01 8.22450444e-02 5.58676183e-01 -5.96940994e-01
-3.23052853e-01 2.07374096e-01 -6.31508529e-01 -2.79453814e-01
9.72882271e-01 -2.97891110e-01 8.22481327e-03 9.03512239e-01
1.07346344e+00 1.62215084e-01 -5.82549810e-01 -6.48144960e-01
2.24830568e-01 1.70741864e-02 2.72470146e-01 -8.62239420e-01
-9.61109042e-01 9.48898196e-01 3.45088989e-01 8.17867890e-02
8.69106829e-01 -1.52099326e-01 6.41428351e-01 6.88447952e-01
3.73881280e-01 -1.13314104e+00 -2.79703319e-01 1.08972752e+00
6.64565504e-01 -1.38980603e+00 -8.14929157e-02 -2.87603348e-01
-7.46992946e-01 9.58922923e-01 5.55677652e-01 4.44605440e-01
8.14423263e-01 5.43295383e-01 5.70989966e-01 1.28940165e-01
-1.07618475e+00 -3.27865011e-03 4.30474639e-01 4.82750416e-01
1.23010266e+00 -2.73051355e-02 -3.58565032e-01 3.79598588e-01
-7.40897477e-01 -3.27970415e-01 1.59568653e-01 9.00867939e-01
3.83270346e-02 -1.52960062e+00 -3.40077221e-01 5.57168245e-01
-6.70820177e-01 -6.40532613e-01 -3.79769951e-01 9.17299807e-01
-2.26028547e-01 9.04246569e-01 9.62877274e-02 -2.53089160e-01
1.96945518e-01 4.78531778e-01 6.25344217e-01 -8.84317935e-01
-9.39920962e-01 8.01754519e-02 3.76669586e-01 2.11147945e-02
-3.14638555e-01 -8.59296560e-01 -8.15180123e-01 -4.24646914e-01
-3.96048605e-01 1.86099336e-01 6.44509614e-01 1.14594710e+00
1.10592648e-01 3.92446965e-01 3.26260805e-01 -5.60333550e-01
-5.75027704e-01 -1.39836526e+00 -3.12032342e-01 3.47679257e-01
1.18476383e-01 -4.58952636e-01 -2.82289952e-01 3.58113170e-01]
|
[10.936624526977539, 9.980938911437988]
|
6b4b0bbb-0cae-4057-9f6a-13a38aa9169c
|
reasoning-on-knowledge-graphs-with-debate
|
2001.00461
| null |
https://arxiv.org/abs/2001.00461v1
|
https://arxiv.org/pdf/2001.00461v1.pdf
|
Reasoning on Knowledge Graphs with Debate Dynamics
|
We propose a novel method for automatic reasoning on knowledge graphs based on debate dynamics. The main idea is to frame the task of triple classification as a debate game between two reinforcement learning agents which extract arguments -- paths in the knowledge graph -- with the goal to promote the fact being true (thesis) or the fact being false (antithesis), respectively. Based on these arguments, a binary classifier, called the judge, decides whether the fact is true or false. The two agents can be considered as sparse, adversarial feature generators that present interpretable evidence for either the thesis or the antithesis. In contrast to other black-box methods, the arguments allow users to get an understanding of the decision of the judge. Since the focus of this work is to create an explainable method that maintains a competitive predictive accuracy, we benchmark our method on the triple classification and link prediction task. Thereby, we find that our method outperforms several baselines on the benchmark datasets FB15k-237, WN18RR, and Hetionet. We also conduct a survey and find that the extracted arguments are informative for users.
|
['Yunpu Ma', 'Jorge Andres Quintero Serna', 'Mitchell Joblin', 'Martin Ringsquandl', 'Marcel Hildebrandt', 'Volker Tresp']
|
2020-01-02
| null | null | null | null |
['triple-classification']
|
['graphs']
|
[ 8.91029835e-02 1.04195619e+00 -7.02648997e-01 -2.00300142e-01
-6.24274850e-01 -8.21035266e-01 8.11505198e-01 1.96929932e-01
1.12542957e-02 1.18117154e+00 3.28357279e-01 -9.19727027e-01
-3.17411453e-01 -1.26431477e+00 -9.65953887e-01 -4.37939644e-01
6.61404729e-02 7.91857421e-01 5.88693842e-02 -4.62593615e-01
5.87148070e-02 -1.34282321e-01 -1.27054954e+00 6.35516822e-01
9.68838155e-01 1.07075977e+00 -7.85269797e-01 5.99268258e-01
5.29060252e-02 1.61356115e+00 -7.97006905e-01 -1.47364068e+00
1.49192601e-01 -6.38936877e-01 -1.47224569e+00 -4.20600325e-01
2.84858853e-01 -1.15388215e-01 -3.25499624e-01 1.18817663e+00
1.15435481e-01 -2.84595281e-01 9.94875908e-01 -1.99223304e+00
-7.62136519e-01 1.45463634e+00 -3.59897345e-01 1.13031873e-03
7.13960409e-01 -6.70495629e-02 1.69284940e+00 -4.72801387e-01
9.18459237e-01 1.35148180e+00 4.45393384e-01 4.23331559e-01
-1.04190814e+00 -5.55902600e-01 2.02767700e-01 8.51273000e-01
-7.89498925e-01 -2.51791865e-01 9.42558110e-01 -4.16233927e-01
5.63655555e-01 5.51463485e-01 8.62899899e-01 1.48577762e+00
2.38840431e-01 9.31102395e-01 1.29871094e+00 -4.39745694e-01
4.88406152e-01 1.73484132e-01 4.00229067e-01 1.03262103e+00
4.55487281e-01 1.57637611e-01 -7.01509416e-01 -5.04503965e-01
1.51553333e-01 -7.27298379e-01 -3.72822523e-01 -4.70643431e-01
-1.03447592e+00 1.35643387e+00 6.09403849e-01 -9.30551514e-02
-3.13535571e-01 1.11570157e-01 2.56524831e-01 5.44288099e-01
2.79036283e-01 4.01144385e-01 -4.11531895e-01 -1.34005472e-01
-4.59392369e-01 5.73272109e-01 1.52529073e+00 5.09978592e-01
2.72776365e-01 -3.26122642e-01 -4.15427685e-01 5.41179962e-02
5.79527199e-01 2.60198206e-01 1.11397803e-01 -9.65065718e-01
5.75520813e-01 8.25686812e-01 1.92773923e-01 -1.37603426e+00
-1.20942265e-01 -5.63601255e-01 -5.01806617e-01 4.18115258e-01
6.98181808e-01 -3.46826196e-01 -4.73374873e-01 1.54326856e+00
4.77317333e-01 2.80777156e-01 5.61532199e-01 8.01780343e-01
1.23840940e+00 4.25933987e-01 -7.19177723e-03 5.89348748e-02
1.62143731e+00 -1.07365263e+00 -7.90370703e-01 6.23419043e-03
6.29300535e-01 -1.86763898e-01 6.81966484e-01 3.91488135e-01
-8.99886429e-01 3.99178267e-02 -1.17808914e+00 9.53545198e-02
-4.34954017e-01 -1.28877461e-01 9.04888391e-01 5.95301807e-01
-5.60515523e-01 5.96723258e-01 -2.06257850e-01 3.01028371e-01
7.36920655e-01 6.30967021e-02 -1.59145519e-01 1.30484045e-01
-1.86086547e+00 1.05013347e+00 4.67224389e-01 -8.57928470e-02
-8.97336006e-01 -5.26273310e-01 -7.13182449e-01 2.83482641e-01
7.85935283e-01 -9.20777202e-01 1.17326617e+00 -8.83436918e-01
-1.39062095e+00 1.00700772e+00 1.60389662e-01 -8.12300205e-01
1.08854914e+00 -6.66454136e-02 -4.08745080e-01 -1.40552188e-03
1.77982956e-01 3.34406465e-01 7.91832507e-01 -1.40751266e+00
-8.78088892e-01 -1.74027041e-01 7.95628488e-01 1.49945050e-01
1.48801953e-01 -3.66637498e-01 3.49349752e-02 -4.43525791e-01
-2.12186635e-01 -1.00083697e+00 9.90119502e-02 -2.36006409e-01
-1.22987354e+00 -6.34067774e-01 8.93284619e-01 -6.13825917e-01
1.04764771e+00 -1.62166786e+00 3.75951916e-01 5.60963869e-01
6.20275855e-01 6.61961585e-02 2.14235455e-01 1.02100745e-01
-1.13565952e-01 4.71455723e-01 5.99437281e-02 3.78379673e-01
1.40167758e-01 3.89336884e-01 -8.03034306e-01 3.24887156e-01
1.05877362e-01 1.23049986e+00 -1.08882678e+00 -4.92357582e-01
-3.10646892e-01 1.54714473e-02 -3.66533399e-01 1.14481591e-01
-6.02920413e-01 2.15367973e-01 -7.22521424e-01 5.09639978e-01
3.60849768e-01 -6.05252445e-01 4.66026634e-01 -2.66768605e-01
5.89168727e-01 5.36756694e-01 -1.18614519e+00 8.89546871e-01
-4.45411950e-02 5.99877715e-01 -5.00187129e-02 -1.16415429e+00
6.19591594e-01 2.51153857e-01 -8.18414092e-02 -5.46890616e-01
3.74906421e-01 6.62085414e-02 1.77830622e-01 -3.31445605e-01
2.05880657e-01 1.81535073e-03 -1.84605822e-01 6.11107767e-01
-1.78384632e-01 -7.84301534e-02 3.23621154e-01 7.99606323e-01
1.17068052e+00 2.27346152e-01 3.49635661e-01 -1.36678562e-01
5.99842429e-01 1.68259919e-01 4.58501101e-01 9.87262368e-01
-2.71059517e-02 -2.46421158e-01 1.37237859e+00 -5.66052258e-01
-5.22347331e-01 -1.02680147e+00 3.78828675e-01 8.39442492e-01
2.72072792e-01 -5.86228132e-01 -5.13392687e-01 -1.59750569e+00
4.43136245e-01 9.93673563e-01 -9.64870334e-01 -1.84976682e-01
-1.94043517e-01 -2.40237147e-01 6.34031057e-01 2.38255903e-01
6.89867854e-01 -1.11596262e+00 -4.38113451e-01 -4.27972563e-02
-7.32293606e-01 -1.01931906e+00 4.60161604e-02 2.28450567e-01
-2.81775683e-01 -1.77179992e+00 4.12203595e-02 -4.45334107e-01
5.77041566e-01 -2.57951707e-01 1.42392468e+00 4.70082253e-01
3.30114029e-02 3.60409796e-01 -3.71349633e-01 -4.66216594e-01
-6.80106223e-01 -6.00318015e-02 -1.68354005e-01 9.78807807e-02
5.52986860e-02 -3.50557923e-01 -3.77210021e-01 3.53369564e-01
-5.27784288e-01 1.41665339e-01 2.62131929e-01 8.09372485e-01
2.56553113e-01 3.22894841e-01 5.98959982e-01 -1.43554175e+00
9.95411098e-01 -6.90829754e-01 -4.49447066e-01 5.77887058e-01
-6.55384839e-01 4.57239002e-01 6.00881398e-01 -1.29187405e-01
-1.11431384e+00 -3.08650494e-01 2.35157877e-01 8.74683186e-02
1.96402967e-01 7.36079335e-01 -2.34905720e-01 7.18459710e-02
9.58522141e-01 -1.38279334e-01 -3.02762657e-01 2.20486328e-01
8.81676733e-01 3.29032034e-01 5.12565970e-01 -9.02262509e-01
1.15781581e+00 3.77289772e-01 9.58371013e-02 -1.24314167e-01
-1.47064304e+00 1.77887246e-01 -1.93263203e-01 -2.45825157e-01
6.74406350e-01 -8.08663309e-01 -1.34302139e+00 -3.94731462e-02
-1.11186016e+00 -2.84254432e-01 -1.97063044e-01 9.46816504e-02
-5.17525434e-01 1.51340917e-01 -4.94166881e-01 -7.89231300e-01
-2.44758055e-01 -9.20069993e-01 4.79095399e-01 3.34796727e-01
-5.48305988e-01 -1.11616540e+00 -2.40096916e-02 8.81137908e-01
4.93324064e-02 4.68152016e-01 1.35812819e+00 -1.05518818e+00
-4.64842796e-01 -1.44239143e-01 -9.94135439e-02 -9.06142145e-02
-6.46195337e-02 1.87647387e-01 -8.64798367e-01 1.42691165e-01
-4.24920142e-01 -8.53298485e-01 7.49738693e-01 3.99601944e-02
1.11462951e+00 -8.29065025e-01 -5.02727926e-01 -5.13215996e-02
8.05736303e-01 -8.88972580e-02 7.73203373e-01 4.89824146e-01
3.01239610e-01 5.56331158e-01 4.97831434e-01 1.23863898e-01
7.85894632e-01 6.43770397e-01 7.62350142e-01 7.12136403e-02
-5.38377417e-03 -5.23457825e-01 2.56489158e-01 2.35258671e-03
-3.52014720e-01 -5.85120618e-01 -5.93291581e-01 2.20928565e-01
-2.15228319e+00 -1.41542685e+00 -3.03246289e-01 1.65159643e+00
1.18347919e+00 5.11871338e-01 2.19864771e-01 4.09200966e-01
5.29966235e-01 2.45151758e-01 -4.01257873e-01 -3.89883637e-01
-2.51176685e-01 -4.21912689e-03 1.17338300e-01 7.61898637e-01
-1.04883325e+00 7.18275249e-01 5.15590763e+00 8.61647189e-01
-6.89743340e-01 -1.59756973e-01 9.72717166e-01 1.55686170e-01
-7.24478006e-01 3.38779628e-01 -6.49332702e-01 2.78337687e-01
7.00858474e-01 -3.79100978e-01 4.29199815e-01 8.60353768e-01
-3.40386480e-01 -7.35822842e-02 -1.33114326e+00 4.87142116e-01
-1.00071713e-01 -1.58549881e+00 2.01826751e-01 1.19661085e-01
6.26682758e-01 -5.70517182e-01 5.84474318e-02 6.86926484e-01
1.18330312e+00 -1.24563360e+00 9.38629448e-01 4.45960313e-01
2.73386866e-01 -6.49287760e-01 6.47715688e-01 4.29007560e-01
-7.93107152e-01 -1.72705874e-01 2.45532263e-02 -1.70805901e-01
-6.45116568e-02 7.97519505e-01 -9.74298239e-01 9.71894681e-01
3.98782521e-01 5.56924284e-01 -4.07928437e-01 4.96325135e-01
-1.26651978e+00 8.27552974e-01 2.11997330e-02 -2.58895576e-01
9.39981714e-02 5.64716943e-02 6.13452494e-01 7.56406963e-01
-2.11725786e-01 3.10442001e-01 2.80938983e-01 9.35558140e-01
-6.56305254e-01 -1.30248964e-01 -5.20801961e-01 -2.32926160e-01
2.50649184e-01 1.49907839e+00 -6.06506348e-01 -5.07595301e-01
-2.49169037e-01 6.96506977e-01 6.50115430e-01 2.72531301e-01
-1.06595755e+00 -2.03603745e-01 1.79813758e-01 -1.21287972e-01
2.67806768e-01 3.87868643e-01 -2.34972835e-01 -1.24842632e+00
5.09811491e-02 -1.22514892e+00 8.21432173e-01 -8.17381799e-01
-1.32587063e+00 5.60762525e-01 -2.52519220e-01 -7.65565097e-01
-4.46759790e-01 -6.86722457e-01 -8.59876633e-01 5.77171385e-01
-1.53238857e+00 -1.17845213e+00 -1.33723974e-01 5.48922002e-01
-1.06137045e-01 -1.20970331e-01 8.20604026e-01 -3.57269645e-01
-2.96074778e-01 4.13429320e-01 -4.91458386e-01 4.41644788e-01
2.66886324e-01 -1.46580637e+00 7.97412619e-02 4.63548273e-01
5.63533962e-01 3.67983937e-01 8.69087040e-01 -9.57231343e-01
-1.22733760e+00 -7.45478272e-01 9.11626101e-01 -8.32925200e-01
9.46562529e-01 -2.13389173e-01 -7.69946456e-01 6.15132987e-01
4.09145266e-01 7.29405284e-02 7.59133697e-01 5.21456778e-01
-7.93996096e-01 2.50586092e-01 -1.10415912e+00 7.70924032e-01
9.89922047e-01 -2.89596468e-01 -9.23472583e-01 5.17529666e-01
4.65732604e-01 -4.92500663e-01 -6.82157099e-01 1.59061715e-01
7.10249424e-01 -7.87105441e-01 8.70492995e-01 -1.46063626e+00
1.12365675e+00 -2.23385483e-01 7.57874176e-02 -1.65219009e+00
-1.91969499e-01 -6.11197293e-01 -7.16939211e-01 9.58994389e-01
9.87394214e-01 -5.50665736e-01 9.05436695e-01 6.81320488e-01
4.74421382e-01 -8.80471766e-01 -1.10045528e+00 -4.60739225e-01
-5.60061894e-02 -6.38818294e-02 4.49015468e-01 1.29753375e+00
5.25460780e-01 8.45947623e-01 -4.30214256e-01 1.45000041e-01
6.81419194e-01 6.99328423e-01 6.53760195e-01 -1.67473555e+00
-4.22309399e-01 -4.74572718e-01 -6.54424801e-02 -4.93182123e-01
6.77150249e-01 -1.15437651e+00 -3.17176372e-01 -1.78898716e+00
2.16631621e-01 -2.11415753e-01 -1.31014511e-01 9.66631591e-01
-3.67438942e-01 -1.63711533e-01 -3.89607251e-02 -1.40947834e-01
-7.05783367e-01 5.19446373e-01 1.42026877e+00 -4.91667509e-01
9.56953242e-02 2.61202693e-01 -1.19640684e+00 8.46165121e-01
6.31877244e-01 -4.12584007e-01 -4.06084448e-01 1.21163808e-01
9.99035180e-01 3.74188393e-01 7.52285898e-01 -3.24650377e-01
3.85505021e-01 -1.83153227e-01 3.62299681e-01 -2.71293163e-01
1.05354950e-01 -7.09959745e-01 -1.43014789e-01 7.25402415e-01
-7.00273454e-01 -1.40420005e-01 -1.40398860e-01 8.99723291e-01
-4.76117395e-02 -1.49546400e-01 2.10236773e-01 -3.76625359e-02
-3.51340413e-01 -3.44139487e-02 -6.37811944e-02 4.96005714e-01
1.05033863e+00 3.11515778e-01 -1.03186858e+00 -9.56485152e-01
-9.16824937e-01 4.44959730e-01 -1.58242121e-01 3.70520264e-01
5.22215426e-01 -1.37831974e+00 -1.13705254e+00 -3.81491661e-01
2.94038430e-02 -3.03363770e-01 -1.54341474e-01 5.78706205e-01
-1.66602567e-01 2.72864848e-01 1.97943114e-03 -1.08312685e-02
-1.22995853e+00 3.85106355e-01 3.79939020e-01 -7.83465385e-01
-3.36902320e-01 7.74578810e-01 -3.06587815e-01 -3.38292360e-01
2.22489044e-01 -2.84574311e-02 -5.26369870e-01 2.73962528e-01
2.70879716e-01 3.90242487e-01 -2.65713502e-02 -3.73887062e-01
-3.63214612e-01 -1.82276830e-01 -1.60346851e-01 3.88774015e-02
1.31006801e+00 4.67015266e-01 -1.58248007e-01 2.30025779e-02
5.35433590e-01 4.59864974e-01 -7.73056388e-01 -1.89723372e-01
-6.32130057e-02 -4.20084924e-01 -1.23025775e-01 -1.43182671e+00
-1.01549923e+00 4.51404661e-01 -5.31284958e-02 8.08357596e-01
3.82150501e-01 3.10944796e-01 1.91309258e-01 6.44271195e-01
2.58361340e-01 -9.13128555e-01 1.12274505e-01 3.85272831e-01
1.17043424e+00 -1.17405736e+00 1.37279078e-01 -7.56688058e-01
-8.57015848e-01 1.06571937e+00 6.54586256e-01 1.10068157e-01
2.99427897e-01 5.86713739e-02 -7.24348873e-02 -5.84483027e-01
-9.83726561e-01 1.18557952e-01 4.36246544e-01 5.14667749e-01
2.08623290e-01 3.04734379e-01 -2.94332445e-01 7.65908837e-01
-6.11590743e-01 -1.69814631e-01 5.99208117e-01 5.55362523e-01
-2.01110557e-01 -1.04685593e+00 -2.34442785e-01 5.78811765e-01
-3.96295398e-01 5.51735191e-03 -1.02288938e+00 8.15398097e-01
-2.13334441e-01 1.29197371e+00 -4.77622569e-01 -3.89262795e-01
3.10916871e-01 9.63177159e-02 4.13155049e-01 -2.40410656e-01
-5.81491172e-01 -7.10887432e-01 9.21256959e-01 -4.33867455e-01
-3.99450541e-01 -2.14372963e-01 -1.08414650e+00 -5.31026483e-01
-3.94422203e-01 6.88556910e-01 3.05847317e-01 1.24787271e+00
2.04555690e-01 6.76603854e-01 4.58946228e-01 1.62729959e-03
-8.17100406e-01 -4.63333279e-01 -3.33256125e-01 5.94017744e-01
3.06546278e-02 -9.31924880e-01 -4.36616540e-01 -7.94397965e-02]
|
[9.636714935302734, 7.974894046783447]
|
dee52f23-e176-4c4b-9b59-f026f7afd127
|
global-context-enhanced-graph-convolutional
| null | null |
https://aclanthology.org/2020.coling-main.461
|
https://aclanthology.org/2020.coling-main.461.pdf
|
Global Context-enhanced Graph Convolutional Networks for Document-level Relation Extraction
|
Document-level Relation Extraction (RE) is particularly challenging due to complex semantic interactions among multiple entities in a document. Among exiting approaches, Graph Convolutional Networks (GCN) is one of the most effective approaches for document-level RE. However, traditional GCN simply takes word nodes and adjacency matrix to represent graphs, which is difficult to establish direct connections between distant entity pairs. In this paper, we propose Global Context-enhanced Graph Convolutional Networks (GCGCN), a novel model which is composed of entities as nodes and context of entity pairs as edges between nodes to capture rich global context information of entities in a document. Two hierarchical blocks, Context-aware Attention Guided Graph Convolution (CAGGC) for partially connected graphs and Multi-head Attention Guided Graph Convolution (MAGGC) for fully connected graphs, could take progressively more global context into account. Meantime, we leverage a large-scale distantly supervised dataset to pre-train a GCGCN model with curriculum learning, which is then fine-tuned on the human-annotated dataset for further improving document-level RE performance. The experimental results on DocRED show that our model could effectively capture rich global context information in the document, leading to a state-of-the-art result. Our code is available at https://github.com/Huiweizhou/GCGCN.
|
['Haibin Jiang', 'Chengkun Lang', 'Zhe Liu', 'Weihong Yao', 'Yibin Xu', 'Huiwei Zhou']
|
2020-12-01
| null | null | null |
coling-2020-8
|
['document-level-relation-extraction']
|
['natural-language-processing']
|
[-1.62848115e-01 1.82109177e-01 -1.80580899e-01 -2.95003682e-01
-3.41045380e-01 -5.90108514e-01 6.53233826e-01 3.97404581e-01
-1.30931452e-01 4.89679098e-01 4.30878937e-01 -4.80873853e-01
-2.19859987e-01 -1.21702504e+00 -5.96281648e-01 -3.22026759e-01
-1.40136167e-01 3.80520254e-01 9.22747403e-02 -3.19440871e-01
8.79860595e-02 3.97005260e-01 -8.82190824e-01 2.46212184e-01
1.01343369e+00 6.20030046e-01 3.02383304e-01 7.57007360e-01
-5.60923100e-01 9.36992526e-01 -4.74687546e-01 -6.87098622e-01
-2.32511833e-01 -3.45719248e-01 -1.13417530e+00 -4.77750320e-03
2.87337363e-01 -1.66886151e-01 -6.63412213e-01 1.07459188e+00
3.29101115e-01 2.40677208e-01 4.20078814e-01 -1.05108011e+00
-1.29675448e+00 1.03689218e+00 -5.34451783e-01 2.97405243e-01
2.61659592e-01 -2.23126262e-01 1.57191467e+00 -8.51777732e-01
6.32155299e-01 1.13761425e+00 4.33021605e-01 1.85999021e-01
-6.95289612e-01 -6.68729901e-01 7.77896702e-01 3.39772582e-01
-1.47437155e+00 2.36437142e-01 8.58823180e-01 -1.51042849e-01
1.35186195e+00 1.52648836e-01 7.10565686e-01 8.89421344e-01
1.62621345e-02 7.70426095e-01 3.99873048e-01 -4.10339892e-01
-2.95890957e-01 -5.05681634e-01 3.71031553e-01 8.72118652e-01
4.61952955e-01 -4.14297372e-01 -1.81780793e-02 1.26388878e-01
8.09377611e-01 2.96794236e-01 -5.16321480e-01 -1.22280363e-02
-1.06338072e+00 7.72581875e-01 1.14857244e+00 6.47674143e-01
-2.83157319e-01 3.36637765e-01 4.71053541e-01 6.28112331e-02
4.82513368e-01 4.90794599e-01 -3.85874122e-01 2.48194233e-01
-4.49543536e-01 5.86880967e-02 6.77999556e-01 1.39156377e+00
8.05956900e-01 -3.10310334e-01 -4.57360566e-01 7.91054547e-01
2.44700268e-01 2.32175648e-01 3.35467190e-01 -1.91718668e-01
9.38789427e-01 1.19701052e+00 -4.72687483e-01 -1.48767936e+00
-4.03085083e-01 -8.16503763e-01 -1.07487035e+00 -7.48100221e-01
-1.19806200e-01 -2.04690099e-01 -1.01020658e+00 1.45038545e+00
2.27932453e-01 3.92494887e-01 2.93660760e-02 8.18248749e-01
1.37905145e+00 7.53661811e-01 2.68177778e-01 2.18942776e-01
1.45183086e+00 -1.40292001e+00 -8.43382716e-01 -3.89647096e-01
1.00194025e+00 -4.10939544e-01 1.00362670e+00 -1.04828440e-01
-6.17461801e-01 -4.65326250e-01 -8.47563565e-01 -4.62346941e-01
-8.36263478e-01 3.57110314e-02 8.48324776e-01 -1.22605879e-02
-1.06095207e+00 4.48914111e-01 -6.13184929e-01 -4.01282340e-01
4.88891393e-01 2.19557196e-01 -3.31786782e-01 -3.31476808e-01
-1.50273025e+00 4.99110758e-01 7.66849458e-01 5.45566320e-01
-5.49974144e-01 -3.62246066e-01 -1.19149554e+00 5.67182958e-01
7.17019975e-01 -6.56878948e-01 1.01090860e+00 -5.01115441e-01
-8.19187343e-01 7.44616926e-01 5.13334088e-02 -1.09564550e-01
-8.80690888e-02 -1.22452140e-01 -7.35380590e-01 1.16820283e-01
7.60348281e-03 2.93265790e-01 2.98403293e-01 -1.10615420e+00
-5.33395827e-01 -2.72906363e-01 3.13174367e-01 2.95526505e-01
-3.88809919e-01 8.28498080e-02 -1.05546772e+00 -9.37873363e-01
-1.17533840e-01 -7.01431274e-01 -1.90199986e-01 -6.75412953e-01
-8.61652553e-01 -6.04308486e-01 7.30061650e-01 -8.87473345e-01
1.86794102e+00 -1.82226717e+00 2.04871327e-01 2.50890672e-01
6.34360611e-01 4.23115224e-01 -4.96668518e-01 7.63187230e-01
-2.30516896e-01 2.98745215e-01 -1.94046587e-01 -2.05490455e-01
-5.59555404e-02 1.85487792e-01 -8.93646851e-02 -6.99168593e-02
4.06574249e-01 1.70392954e+00 -1.22257090e+00 -4.18572426e-01
-2.06007719e-01 5.59653103e-01 -4.23127383e-01 2.91679353e-01
-3.57936293e-01 1.60628095e-01 -7.20257699e-01 6.19994223e-01
5.67721188e-01 -8.03318918e-01 5.49357772e-01 -3.27435374e-01
5.03317058e-01 4.28889334e-01 -8.85414124e-01 1.61552119e+00
-5.11810184e-01 4.86076981e-01 -2.13909313e-01 -9.19306338e-01
1.08034754e+00 6.96213543e-02 3.92386280e-02 -6.06039882e-01
1.44548789e-01 3.64163257e-02 -6.98627010e-02 -4.30156618e-01
6.91414595e-01 3.43171865e-01 -1.40829951e-01 2.72401989e-01
2.44585767e-01 1.46921501e-01 2.11029321e-01 7.25893438e-01
1.24431431e+00 -1.10696971e-01 4.05470580e-01 -3.90918441e-02
6.30536437e-01 -2.27186233e-01 4.64010477e-01 4.51685220e-01
1.87310234e-01 5.99465370e-01 6.90842509e-01 -2.67320663e-01
-5.89190483e-01 -5.32410741e-01 4.38452065e-01 8.61429513e-01
2.57309437e-01 -1.04903769e+00 -6.03556693e-01 -1.09707606e+00
-2.44765473e-03 5.43130934e-01 -6.17365539e-01 -1.62551388e-01
-7.70975292e-01 -5.89676380e-01 1.82046115e-01 8.40412319e-01
7.02436626e-01 -1.16352069e+00 3.67039651e-01 3.08992237e-01
-2.09324241e-01 -1.34933317e+00 -8.93885970e-01 -5.26178740e-02
-6.27366424e-01 -1.33297515e+00 -7.14402974e-01 -1.08561862e+00
8.24860930e-01 4.13200200e-01 1.38028026e+00 7.37706304e-01
-1.15380965e-01 3.59141767e-01 -6.70457721e-01 -6.08326197e-02
4.11740392e-02 5.10046840e-01 -6.42594278e-01 -1.81086168e-01
5.88852227e-01 -5.86720407e-01 -7.11556792e-01 4.79273871e-02
-7.64530778e-01 1.10327199e-01 6.32652879e-01 9.04056907e-01
7.06530988e-01 2.67717928e-01 5.45425057e-01 -1.52068758e+00
9.69575763e-01 -6.92879915e-01 -4.74384189e-01 6.52210653e-01
-7.45004714e-01 -1.33276656e-01 7.68428624e-01 -1.16087861e-01
-8.87497962e-01 -5.43387592e-01 -3.15540209e-02 -3.50486726e-01
7.90152103e-02 1.19631088e+00 -4.31896567e-01 2.34385297e-01
2.09454998e-01 9.01935548e-02 -6.20146811e-01 -3.80670160e-01
6.09763801e-01 6.09371066e-01 6.21594548e-01 -5.64134061e-01
6.45742595e-01 -5.47902361e-02 -7.97632057e-03 -3.28080595e-01
-1.12487888e+00 -5.69235146e-01 -6.42642796e-01 8.36320594e-02
9.25286770e-01 -1.01099873e+00 -4.02971953e-01 3.90948296e-01
-1.11881804e+00 -6.12568378e-01 7.47050196e-02 2.24241525e-01
1.23760566e-01 4.32284474e-01 -8.71185601e-01 -2.89449096e-01
-6.35840952e-01 -7.56337523e-01 1.01536167e+00 4.30121422e-01
1.93021834e-01 -1.49806356e+00 -4.71022092e-02 2.64261127e-01
2.29587644e-01 3.52855682e-01 1.02386427e+00 -8.52877080e-01
-7.55820513e-01 -4.07862246e-01 -7.78123736e-01 1.55193552e-01
3.83115798e-01 -2.17362512e-02 -5.18058896e-01 -3.35363209e-01
-7.91298211e-01 -4.31561172e-02 9.41112757e-01 -9.43731517e-03
1.50599241e+00 -4.47843522e-01 -7.23618269e-01 6.42584264e-01
1.49972022e+00 -5.76941483e-02 6.01721048e-01 3.30427974e-01
1.57089198e+00 4.41390783e-01 5.88502765e-01 1.26579344e-01
8.85159731e-01 4.01471853e-01 4.31502432e-01 -2.37302482e-01
-3.77941847e-01 -6.13623023e-01 -1.18502527e-01 1.23463953e+00
-1.71282768e-01 -8.69751692e-01 -1.09113491e+00 7.13004410e-01
-1.94663632e+00 -6.42587721e-01 -3.34392607e-01 1.53966427e+00
5.97176611e-01 8.59901980e-02 -2.02009767e-01 -2.44167313e-01
1.07477641e+00 3.10187757e-01 -2.94108629e-01 -2.35014692e-01
9.37210955e-03 1.45837963e-01 2.97414631e-01 4.79419142e-01
-1.11443436e+00 1.20513511e+00 4.51892996e+00 8.47765625e-01
-7.68497467e-01 -7.00259879e-02 5.02051175e-01 3.37896973e-01
-6.34239674e-01 1.29438639e-01 -8.17087591e-01 5.29214919e-01
8.02294970e-01 -3.07347834e-01 3.07558149e-01 8.03649247e-01
-3.09228897e-01 5.19389212e-01 -8.67703021e-01 8.34924221e-01
8.28703791e-02 -1.37027252e+00 3.56687635e-01 4.84900847e-02
8.40014279e-01 6.59685656e-02 -3.39404672e-01 6.32855833e-01
6.83202744e-01 -1.09513712e+00 2.65801679e-02 4.85274822e-01
7.55963564e-01 -1.02077878e+00 9.61376429e-01 2.83800829e-02
-1.80251133e+00 1.61984172e-02 -4.01462555e-01 2.35046446e-01
-2.44369935e-02 7.09206343e-01 -6.64190769e-01 1.12508416e+00
5.92372179e-01 1.16067827e+00 -7.86753833e-01 7.79573500e-01
-7.43118405e-01 5.75871587e-01 1.05470076e-01 -2.59203315e-01
5.41285574e-01 -2.60617316e-01 3.31400454e-01 1.61514843e+00
2.39758164e-01 4.41500217e-01 1.88086003e-01 7.88737833e-01
-6.22943878e-01 1.31439343e-01 -4.61802572e-01 -4.31539327e-01
5.97357273e-01 1.70942807e+00 -7.34335005e-01 -4.12728667e-01
-6.53081894e-01 1.10039401e+00 1.00545096e+00 5.34498632e-01
-5.97368896e-01 -1.00728261e+00 3.65059495e-01 -1.56950340e-01
5.81049621e-01 -2.86194235e-01 2.93733627e-01 -1.36882174e+00
7.40126334e-03 -6.35253489e-01 9.62329209e-01 -9.05465901e-01
-1.55374849e+00 8.86103094e-01 -1.54411405e-01 -9.63355541e-01
7.46005550e-02 -6.63469195e-01 -8.90681386e-01 9.79052186e-01
-1.73203123e+00 -1.55523324e+00 -7.15550423e-01 6.73745036e-01
1.82672188e-01 -8.55185837e-02 6.47628427e-01 2.89480865e-01
-7.53293872e-01 8.52487922e-01 -8.45057443e-02 7.47307837e-01
4.70657289e-01 -1.53771484e+00 7.64267266e-01 1.03486097e+00
2.58071095e-01 9.58059013e-01 -8.16378519e-02 -9.77492034e-01
-1.16283607e+00 -1.61632729e+00 1.18339443e+00 -3.10933501e-01
8.47737551e-01 -4.45879817e-01 -1.22222352e+00 1.11767709e+00
2.57161230e-01 4.03209269e-01 6.37029052e-01 3.98149610e-01
-4.60752696e-01 1.18239559e-01 -6.51699424e-01 6.78691983e-01
1.45639551e+00 -7.56765604e-01 -4.41117138e-01 4.60608900e-01
1.24048054e+00 -6.56283259e-01 -1.15118420e+00 3.63016427e-01
1.78932641e-02 -4.64427948e-01 7.46712148e-01 -7.23658681e-01
7.24308074e-01 -2.51550198e-01 5.47408573e-02 -1.41407108e+00
-6.42696559e-01 -4.41415012e-01 -5.78099072e-01 1.61357033e+00
4.15409505e-01 -7.37954021e-01 5.72827578e-01 3.02616000e-01
-3.68744433e-01 -1.01127148e+00 -1.87709987e-01 -6.49526000e-01
9.32511985e-02 -1.82498142e-01 1.04007697e+00 1.34348524e+00
4.45210263e-02 7.52311647e-01 1.03206359e-01 4.13345873e-01
1.60497665e-01 4.48268235e-01 5.62696338e-01 -1.00663483e+00
-2.62877822e-01 -5.22707880e-01 -4.67902064e-01 -1.15667701e+00
3.84538949e-01 -1.49898863e+00 -3.16873282e-01 -2.22723579e+00
3.50565046e-01 -4.34620440e-01 -4.61757332e-01 6.45554125e-01
-8.05208087e-01 -1.05351873e-01 5.54675870e-02 8.95346701e-03
-8.74162495e-01 6.91764891e-01 1.60418010e+00 -3.36992323e-01
-6.35155886e-02 -3.29907924e-01 -9.95511591e-01 3.05323839e-01
6.28041923e-01 -2.33245432e-01 -6.50590897e-01 -6.62206590e-01
5.33741176e-01 -1.65628314e-01 3.50732744e-01 -5.08188546e-01
4.16742921e-01 -2.25314070e-02 3.26881766e-01 -6.41389430e-01
-3.33103597e-01 -7.46768177e-01 -1.07921228e-01 1.01848152e-02
-3.46684664e-01 1.17907844e-01 9.04325470e-02 8.03378463e-01
-4.90089178e-01 -6.89746961e-02 7.78135136e-02 -2.20151901e-01
-8.87123108e-01 8.09449017e-01 2.08635435e-01 1.89409494e-01
6.37118518e-01 2.21657902e-01 -6.90494955e-01 -4.04606998e-01
-6.51422918e-01 8.10881376e-01 5.46864532e-02 5.84014297e-01
5.90811491e-01 -1.50993264e+00 -7.18443274e-01 -2.14354560e-01
4.14766371e-01 6.58489883e-01 3.61587554e-01 4.66807663e-01
-4.96890187e-01 5.55832565e-01 3.93089354e-01 -2.17050210e-01
-1.24959755e+00 8.02619636e-01 4.03962076e-01 -8.04973900e-01
-8.46970856e-01 1.16550136e+00 2.94536382e-01 -7.20518947e-01
4.57412526e-02 -5.22071481e-01 -4.79133904e-01 -2.43531927e-01
4.88242537e-01 3.77655178e-02 1.00269362e-01 -4.91257429e-01
-2.91649938e-01 5.91497838e-01 -5.66521823e-01 5.98050475e-01
1.29319918e+00 -1.90124005e-01 -4.42090809e-01 -6.11253120e-02
1.28766525e+00 6.60864636e-02 -9.38482761e-01 -5.86848438e-01
1.40359595e-01 -2.47810170e-01 1.79026529e-01 -7.46626139e-01
-1.39108181e+00 8.16736341e-01 -1.42430723e-01 2.40718603e-01
1.25549054e+00 2.95330763e-01 9.91071701e-01 4.69164342e-01
1.01124085e-01 -7.54528046e-01 1.72942922e-01 6.20357871e-01
1.09044981e+00 -1.14359415e+00 1.00990519e-01 -8.36495161e-01
-5.81427455e-01 1.21767390e+00 8.45072091e-01 -1.41760483e-01
7.78926253e-01 -8.06821957e-02 -2.65418172e-01 -6.75999582e-01
-5.33715665e-01 -4.72881943e-01 6.87619090e-01 6.18673027e-01
5.47591746e-01 1.56633556e-01 -2.19050571e-01 1.12480211e+00
-1.07271694e-01 -1.65284500e-01 2.40122139e-01 9.16455567e-01
4.57045548e-02 -1.10514307e+00 3.11421216e-01 5.16613185e-01
-2.95503974e-01 -5.72455466e-01 -6.53703868e-01 8.33638370e-01
1.77884083e-02 8.56217563e-01 1.39426619e-01 -5.53838313e-01
4.44243878e-01 -2.68245578e-01 3.17881763e-01 -9.99357283e-01
-6.80998087e-01 -1.32896304e-01 2.84674704e-01 -4.57374185e-01
-2.00533971e-01 -2.51966566e-01 -1.58194911e+00 -3.07793528e-01
-5.66412389e-01 1.78336516e-01 1.55324414e-01 7.91537523e-01
6.25498116e-01 9.89223897e-01 4.78202909e-01 -3.66534173e-01
2.00866580e-01 -1.02135766e+00 -5.93538344e-01 4.16632444e-01
1.56377509e-01 -4.26495761e-01 -1.44988179e-01 -2.34303996e-01]
|
[9.030948638916016, 8.229461669921875]
|
dca0db50-6616-4203-a209-2f3fee11cee2
|
introducing-anisotropic-minkowski-functionals
|
2004.01029
| null |
https://arxiv.org/abs/2004.01029v1
|
https://arxiv.org/pdf/2004.01029v1.pdf
|
Introducing Anisotropic Minkowski Functionals for Local Structure Analysis and Prediction of Biomechanical Strength of Proximal Femur Specimens
|
Bone fragility and fracture caused by osteoporosis or injury are prevalent in adults over the age of 50 and can reduce their quality of life. Hence, predicting the biomechanical bone strength, specifically of the proximal femur, through non-invasive imaging-based methods is an important goal for the diagnosis of Osteoporosis as well as estimating fracture risk. Dual X-ray absorptiometry (DXA) has been used as a standard clinical procedure for assessment and diagnosis of bone strength and osteoporosis through bone mineral density (BMD) measurements. However, previous studies have shown that quantitative computer tomography (QCT) can be more sensitive and specific to trabecular bone characterization because it reduces the overlap effects and interferences from the surrounding soft tissue and cortical shell. This study proposes a new method to predict the bone strength of proximal femur specimens from quantitative multi-detector computer tomography (MDCT) images. Texture analysis methods such as conventional statistical moments (BMD mean), Isotropic Minkowski Functionals (IMF) and Anisotropic Minkowski Functionals (AMF) are used to quantify BMD properties of the trabecular bone micro-architecture. Combinations of these extracted features are then used to predict the biomechanical strength of the femur specimens using sophisticated machine learning techniques such as multiregression (MultiReg) and support vector regression with linear kernel (SVRlin). The prediction performance achieved with these feature sets is compared to the standard approach that uses the mean BMD of the specimens and multiregression models using root mean square error (RMSE).
|
['Titas De']
|
2020-04-02
| null | null | null | null |
['texture-classification']
|
['computer-vision']
|
[-1.38690561e-01 -2.23771095e-01 -2.53307223e-01 -2.62012661e-01
-7.87341952e-01 3.98539037e-01 4.49153669e-02 4.36657786e-01
-5.60354054e-01 8.28900397e-01 7.47895241e-02 9.02175754e-02
-5.01454294e-01 -1.30724049e+00 -3.72757196e-01 -6.86372161e-01
-2.94391155e-01 1.18168747e+00 6.06156230e-01 -2.89662123e-01
2.43679270e-01 6.28700793e-01 -1.65833557e+00 3.40701610e-01
6.59888685e-01 8.46074224e-01 2.38139883e-01 3.24910104e-01
1.41475707e-01 2.66699970e-01 8.37308988e-02 -1.69566467e-01
-2.46724278e-01 -3.02250385e-01 -5.62227607e-01 -1.20243557e-01
-2.21752942e-01 -4.08242792e-01 3.08428913e-01 6.40782595e-01
6.78940892e-01 -4.36725803e-02 9.35272038e-01 -2.87887484e-01
-3.34706604e-01 3.55381936e-01 -8.71272147e-01 3.40351224e-01
3.76670033e-01 -1.11332580e-01 4.14692491e-01 -1.23121142e+00
2.14137107e-01 1.34165215e+00 6.75368607e-01 5.05457819e-01
-1.39794087e+00 -5.67018151e-01 -9.18631315e-01 5.44890225e-01
-1.26336098e+00 -3.02840114e-01 5.53074360e-01 -6.62281632e-01
5.62199831e-01 3.79787028e-01 8.96290839e-01 5.72512150e-01
9.44999635e-01 1.89640701e-01 1.44172919e+00 -5.71475506e-01
2.85782665e-01 -4.01065052e-01 -9.87673029e-02 8.36958408e-01
5.88419974e-01 2.67636389e-01 -3.37898552e-01 -2.69722015e-01
7.78047919e-01 7.43185356e-02 7.05793053e-02 -3.12954575e-01
-8.27416241e-01 8.46079051e-01 1.10029541e-01 5.65291464e-01
-5.39358020e-01 9.17430371e-02 5.02306938e-01 -7.31333811e-03
4.86164927e-01 -3.43506098e-01 -2.40795493e-01 -1.10201307e-01
-1.13055599e+00 2.88052946e-01 2.11585835e-01 -1.14172198e-01
6.56340003e-01 -1.00956850e-01 1.79677665e-01 1.27877581e+00
8.54957104e-01 9.39393938e-01 8.90650392e-01 -7.16581523e-01
-1.62633494e-01 6.13165677e-01 -3.48863780e-01 -9.31062043e-01
-3.66053164e-01 -1.14047028e-01 -8.94923270e-01 4.74873394e-01
3.04156780e-01 5.05955458e-01 -6.27331614e-01 1.01991570e+00
5.45084417e-01 -2.84939438e-01 -5.34561455e-01 1.04848909e+00
5.75463533e-01 1.77905172e-01 1.60705552e-01 -2.99078047e-01
1.52340734e+00 -2.09693313e-01 -2.64372587e-01 3.78392860e-02
3.18558455e-01 -7.01807737e-01 9.42192256e-01 3.93513888e-01
-1.27325749e+00 -2.60520518e-01 -8.63543630e-01 1.91682115e-01
2.70672347e-02 -2.02448189e-01 2.99716204e-01 9.28120017e-01
-5.27671278e-01 8.48961890e-01 -1.38893056e+00 -1.96455806e-01
1.58404157e-01 5.13201237e-01 -6.88104630e-01 -1.98549047e-01
-9.26018655e-01 1.15962267e+00 -8.51664022e-02 -1.17837511e-01
-4.62093085e-01 -6.06096923e-01 -6.09624028e-01 -1.21012233e-01
-2.00554863e-01 -8.59286427e-01 6.97502136e-01 -5.30234933e-01
-1.71344817e+00 8.51467848e-01 -1.00930154e-01 -9.99312401e-02
2.14776218e-01 -2.26262391e-01 -7.11634010e-02 5.22867084e-01
5.46521783e-01 -1.71661451e-01 3.92317057e-01 -9.53552842e-01
-1.33897901e-01 -1.14679956e+00 -8.43991578e-01 -2.60191590e-01
6.74490035e-02 2.94079930e-01 -5.24735451e-03 -4.60506260e-01
8.04636657e-01 -7.08956718e-01 -2.35275418e-01 -1.26518831e-01
-2.43049413e-01 -7.86352083e-02 3.10598314e-01 -9.98744667e-01
9.89693224e-01 -1.97604847e+00 5.02825789e-02 8.69726598e-01
-1.58148892e-02 -3.41717780e-01 5.07739842e-01 2.93789446e-01
-1.96266565e-02 2.10240129e-02 -2.90048629e-01 1.31138220e-01
-5.35871029e-01 4.13275659e-01 7.80917645e-01 7.89170265e-01
-2.03218952e-01 5.58717549e-01 -5.64634621e-01 -8.91285717e-01
2.53557324e-01 6.20101035e-01 -4.38999146e-01 -3.95409971e-01
4.29018348e-01 3.96111637e-01 -5.64952075e-01 9.00181115e-01
6.29257262e-01 -6.89741448e-02 2.69686639e-01 -2.23553091e-01
1.25325378e-02 -1.44159824e-01 -1.02683592e+00 1.19159579e+00
-4.49693203e-01 -2.37358421e-01 5.64857833e-02 -9.99458373e-01
1.12734354e+00 3.30406636e-01 1.06939697e+00 -1.15833354e+00
4.16675955e-02 7.21552432e-01 1.81279629e-01 -6.53038681e-01
-4.42696810e-01 -9.28213656e-01 5.23527801e-01 3.11503679e-01
-3.25930864e-01 1.27617136e-01 1.48701400e-01 -2.11210519e-01
8.11125517e-01 1.14203282e-01 4.94472951e-01 -7.11624324e-01
7.02913642e-01 -2.75014341e-01 2.95858085e-01 1.50579354e-02
5.22044227e-02 3.92446071e-01 1.57797948e-01 -4.20146853e-01
-1.22709012e+00 -1.58983099e+00 -8.08231950e-01 5.34162164e-01
-1.56726018e-01 4.03391793e-02 -4.14199382e-01 2.36889526e-01
6.29607260e-01 1.78664908e-01 -6.16045713e-01 -4.90403585e-02
-8.12535226e-01 -1.21823299e+00 1.73944220e-01 5.21185219e-01
1.44708157e-01 -6.39638126e-01 -5.66420615e-01 4.15414155e-01
-9.27682891e-02 -4.49100673e-01 3.56167287e-01 1.84262112e-01
-1.77145040e+00 -1.17451155e+00 -8.12214732e-01 -3.44920367e-01
3.43044370e-01 -1.30635381e-01 9.75550175e-01 4.71021503e-01
-6.11369312e-01 3.70306104e-01 -2.51258343e-01 2.01073271e-04
-4.08772796e-01 -4.60812092e-01 2.46362001e-01 -6.19736798e-02
3.04903030e-01 -8.23111594e-01 -9.99509692e-01 4.84737307e-01
-6.52270138e-01 -1.05984583e-01 9.45595264e-01 6.41886711e-01
1.28095424e+00 -8.27301741e-02 4.25964087e-01 -8.24241102e-01
3.13435316e-01 -6.13462329e-01 -6.29140213e-02 -2.03096792e-02
-1.06684816e+00 2.16983315e-02 1.67343676e-01 7.75539177e-03
-9.81679618e-01 -4.42859530e-01 -3.53255689e-01 2.74692744e-01
9.18340385e-02 8.91711354e-01 2.60816980e-02 -1.92690596e-01
7.63782501e-01 1.87734291e-01 5.73892474e-01 -7.95877218e-01
-4.46667731e-01 5.59562147e-01 4.77926821e-01 -8.94846439e-01
1.84727073e-01 7.04041243e-01 7.79059470e-01 -1.11054206e+00
-2.06140071e-01 -7.66077936e-01 -6.13928020e-01 -6.18011475e-01
7.90773869e-01 -7.05545306e-01 -8.83435369e-01 1.55740857e-01
-4.92777258e-01 2.47000121e-02 1.86324805e-01 1.09811068e+00
-6.50954783e-01 7.75317729e-01 -8.29727471e-01 -6.46363616e-01
-5.08361578e-01 -1.26751924e+00 8.48208308e-01 -1.80945456e-01
-1.75795004e-01 -9.57592368e-01 3.73063117e-01 6.74812913e-01
3.86699080e-01 7.79147923e-01 1.29890347e+00 1.23152614e-01
1.00329574e-02 -2.13011190e-01 -1.98797788e-02 2.69834995e-01
2.28779633e-02 1.02714926e-01 -2.44543880e-01 1.48143873e-01
2.94318020e-01 -8.03274885e-02 6.34216607e-01 8.21250319e-01
7.57261455e-01 1.96369439e-01 -3.90563339e-01 2.70188034e-01
2.08407068e+00 1.01405576e-01 9.53052402e-01 8.40804160e-01
3.00625980e-01 5.33039749e-01 4.72744584e-01 7.21820891e-01
4.63956408e-02 7.68185496e-01 5.35384297e-01 3.77448171e-01
-1.91330761e-01 3.42398971e-01 1.77628770e-01 8.23405683e-01
-1.07356858e+00 5.04685462e-01 -1.16122866e+00 4.41649616e-01
-1.39721906e+00 -1.06129754e+00 -8.30747187e-01 2.21163034e+00
8.87821257e-01 1.91724792e-01 3.32932591e-01 6.34926617e-01
6.27264082e-01 -7.46547937e-01 -1.13277704e-01 -3.01884353e-01
-2.26163864e-03 1.03192794e+00 4.74251449e-01 3.12707722e-01
-2.49054000e-01 -4.56092358e-02 6.25498581e+00 5.80394804e-01
-1.09672999e+00 2.94253767e-01 4.04936641e-01 6.03523962e-02
-2.43551657e-01 7.73655102e-02 -3.05344969e-01 6.74630702e-01
1.22742033e+00 1.32384926e-01 -1.32924020e-02 8.10821474e-01
7.10378528e-01 -9.77810621e-01 -6.09754086e-01 5.20924747e-01
-3.79161805e-01 -9.91028011e-01 -7.37295970e-02 2.63497531e-01
8.75155106e-02 4.45442311e-02 -1.26907840e-01 -2.96827495e-01
-3.27256173e-01 -9.58548844e-01 5.57010293e-01 1.08525133e+00
8.17135930e-01 -6.34214163e-01 1.03338587e+00 1.02093637e-01
-1.02510822e+00 9.96263102e-02 -3.71191144e-01 -2.03661770e-01
3.31641167e-01 1.34973228e+00 -5.55298030e-01 4.50798690e-01
1.05056131e+00 1.60697758e-01 -4.35812324e-01 1.09710741e+00
3.10451686e-01 6.54859900e-01 -5.44657052e-01 1.46075159e-01
-3.16679418e-01 -4.57242668e-01 2.05942199e-01 6.82247281e-01
4.52184647e-01 3.90699767e-02 -3.17137510e-01 5.67411125e-01
5.08327425e-01 7.31451869e-01 1.15630142e-01 3.73888552e-01
1.15897931e-01 6.49138391e-01 -9.89245236e-01 -2.18476709e-02
-7.73892343e-01 3.01691949e-01 -2.11367831e-02 -2.72144645e-01
-3.58407408e-01 3.60418707e-01 1.62259787e-02 1.09828699e+00
-4.97851521e-02 -2.07575843e-01 -6.77159131e-01 -6.69315338e-01
-1.13261923e-01 -5.29212117e-01 4.08445418e-01 -4.84140486e-01
-1.27425551e+00 -1.48093030e-01 3.74994934e-01 -6.79414988e-01
-9.49546397e-02 -6.20557606e-01 -3.06878865e-01 9.89482522e-01
-9.86447096e-01 -9.62369084e-01 -1.26261309e-01 4.49498236e-01
6.17328249e-02 5.81643544e-02 8.80411744e-01 5.06071746e-01
-3.87118220e-01 -6.13280162e-02 5.13307512e-01 -8.08428228e-02
3.36808056e-01 -1.32137883e+00 -7.52047956e-01 1.41688436e-01
-6.85011089e-01 4.84036595e-01 9.70179796e-01 -1.12293756e+00
-1.19817603e+00 -4.24987853e-01 8.72808278e-01 1.75024912e-01
4.05795306e-01 5.01079798e-01 -6.95107996e-01 1.68546692e-01
-6.18765891e-01 1.99386701e-02 1.28033721e+00 -4.39051874e-02
1.85219586e-01 -1.91894308e-01 -1.19901502e+00 9.88082886e-02
2.70008624e-01 -1.09810069e-01 -4.49019969e-01 7.00579137e-02
-4.54791397e-01 2.71867421e-02 -1.70069265e+00 9.23478901e-01
1.24012876e+00 -1.51194787e+00 1.46712351e+00 -2.38428544e-02
6.96226418e-01 7.88604766e-02 -3.28412682e-01 -3.91810596e-01
-5.42233348e-01 6.32776022e-01 8.59682858e-02 7.49319494e-01
1.23822190e-01 -2.05087259e-01 1.00976086e+00 2.92666703e-01
-2.68739492e-01 -1.27858758e+00 -1.21401465e+00 -7.91525245e-01
2.69248456e-01 -2.85049856e-01 6.70455471e-02 3.66762400e-01
-4.53332663e-02 -7.72904754e-02 9.67242196e-02 -1.81792125e-01
8.48165452e-01 -1.33013651e-01 3.28773499e-01 -1.72733569e+00
-4.30259436e-01 -1.79397121e-01 -7.66153336e-01 1.79687172e-01
-3.06212723e-01 -7.45680869e-01 -5.06900787e-01 -1.86356878e+00
4.85402972e-01 -4.15382296e-01 -1.56590760e-01 1.08059598e-02
2.52777219e-01 5.11509240e-01 -4.22728777e-01 6.69207692e-01
4.38241333e-01 3.88829291e-01 1.45628202e+00 -1.64780132e-02
-1.70706570e-01 3.42435569e-01 -4.01947722e-02 6.77112460e-01
5.85336566e-01 -6.55787528e-01 -2.02508382e-02 -1.46907764e-02
4.31533366e-01 4.29665804e-01 5.52315533e-01 -1.17850232e+00
-3.73464614e-01 -3.54756415e-01 7.15638638e-01 -5.71712494e-01
2.70620525e-01 -8.08573484e-01 7.48960853e-01 1.21979082e+00
3.38830471e-01 -1.30983487e-01 -2.08424181e-01 3.68737340e-01
-3.21189463e-01 -5.00077307e-01 1.25595641e+00 -4.71459627e-01
-2.37625897e-01 -6.00751340e-02 -6.86438024e-01 -5.09070337e-01
9.18217480e-01 -6.89032972e-01 2.29956225e-01 1.51636899e-01
-1.20693314e+00 -4.21209872e-01 2.84613132e-01 -4.90845531e-01
8.80886793e-01 -1.36646080e+00 -8.76904368e-01 -8.19200501e-02
-2.23147571e-02 -2.22948194e-01 5.15948176e-01 1.53408706e+00
-1.20716560e+00 2.35529140e-01 -7.79804647e-01 -7.50866354e-01
-1.24467337e+00 3.17364752e-01 4.79268521e-01 -4.53919798e-01
-5.68004191e-01 5.82287908e-01 -2.44405195e-01 3.74228805e-01
-5.58194578e-01 -1.11727849e-01 -4.03303564e-01 -1.31780416e-01
2.83325195e-01 9.94922400e-01 2.03221023e-01 -9.21980917e-01
-5.67671657e-01 7.83654451e-01 1.86992928e-01 -1.04774892e-01
1.53435063e+00 -3.12574148e-01 -5.74935913e-01 6.24356031e-01
1.06847835e+00 1.27183408e-01 -4.25894618e-01 3.75789665e-02
1.41004845e-01 -2.84869820e-01 2.42936924e-01 -5.26851892e-01
-9.02705252e-01 7.47199059e-01 1.04741800e+00 -1.03521988e-01
9.94639039e-01 3.01500142e-01 7.53808737e-01 -6.25542104e-02
4.15348262e-01 -1.20098841e+00 1.26507849e-01 -3.62148792e-01
8.07294250e-01 -9.79723036e-01 5.08194685e-01 -6.55331910e-01
1.45588413e-01 1.58167851e+00 2.04322249e-01 -3.24630737e-01
1.22285306e+00 2.77094036e-01 -1.21723920e-01 -4.28636312e-01
-1.34302184e-01 -8.59252587e-02 1.15194127e-01 5.33722341e-01
8.23879659e-01 3.74853462e-01 -1.24953544e+00 6.59352362e-01
-1.91164538e-01 2.15534717e-01 7.22786561e-02 1.18298423e+00
-9.19143558e-01 -1.20584631e+00 -8.70657682e-01 8.79572332e-01
-9.72063720e-01 2.22289070e-01 1.63347766e-01 9.17480171e-01
2.44727090e-01 5.57835937e-01 -2.05628261e-01 -5.37769832e-02
1.10672496e-01 4.32945043e-02 7.77204692e-01 -6.89274311e-01
-1.72609210e-01 3.91002566e-01 -1.08716987e-01 -4.51690167e-01
-9.93190467e-01 -9.58674312e-01 -1.50368643e+00 -2.99320251e-01
-3.98050696e-01 -1.25541911e-02 8.82914901e-01 1.32739353e+00
-4.26871121e-01 2.63203830e-01 3.98282856e-01 -6.22672856e-01
-2.80021816e-01 -9.30171251e-01 -1.29904747e+00 3.43164623e-01
-2.74895847e-01 -1.17494094e+00 -1.75470591e-01 9.14802775e-02]
|
[14.266946792602539, -1.999971628189087]
|
20ed48f7-a7f4-4d2e-9fef-630207d96f6e
|
contrastive-trajectory-similarity-learning
|
2210.05155
| null |
https://arxiv.org/abs/2210.05155v3
|
https://arxiv.org/pdf/2210.05155v3.pdf
|
Contrastive Trajectory Similarity Learning with Dual-Feature Attention
|
Trajectory similarity measures act as query predicates in trajectory databases, making them the key player in determining the query results. They also have a heavy impact on the query efficiency. An ideal measure should have the capability to accurately evaluate the similarity between any two trajectories in a very short amount of time. Towards this aim, we propose a contrastive learning-based trajectory modeling method named TrajCL. We present four trajectory augmentation methods and a novel dual-feature self-attention-based trajectory backbone encoder. The resultant model can jointly learn both the spatial and the structural patterns of trajectories. Our model does not involve any recurrent structures and thus has a high efficiency. Besides, our pre-trained backbone encoder can be fine-tuned towards other computationally expensive measures with minimal supervision data. Experimental results show that TrajCL is consistently and significantly more accurate than the state-of-the-art trajectory similarity measures. After fine-tuning, i.e., to serve as an estimator for heuristic measures, TrajCL can even outperform the state-of-the-art supervised method by up to 56% in the accuracy for processing trajectory similarity queries.
|
['Egemen Tanin', 'Yuxuan Liang', 'Jianzhong Qi', 'Yanchuan Chang']
|
2022-10-11
| null | null | null | null |
['trajectory-modeling']
|
['time-series']
|
[-3.65689009e-01 -5.04520297e-01 -6.83980465e-01 -3.73984337e-01
-1.15805364e+00 -6.00945890e-01 7.90401518e-01 7.62808502e-01
-6.71432137e-01 4.94844913e-01 3.53917956e-01 -3.01146656e-01
-2.80039489e-01 -1.17155623e+00 -7.43015110e-01 -3.92795801e-01
-2.18206108e-01 6.34969890e-01 5.39108515e-01 -2.16684371e-01
1.93199858e-01 5.58176041e-01 -1.60427582e+00 9.98519138e-02
9.34972942e-01 1.04504740e+00 1.39243260e-01 4.79996681e-01
-1.35847703e-01 9.28993046e-01 -2.88408279e-01 -5.87480426e-01
-1.03078067e-01 -2.23722354e-01 -9.39552665e-01 -3.18038583e-01
2.13725522e-01 -4.46483821e-01 -8.71408105e-01 7.25986183e-01
3.27142477e-01 4.75625306e-01 6.17482960e-01 -1.19765794e+00
-3.72688204e-01 5.25829375e-01 4.85028997e-02 5.70269585e-01
5.02348661e-01 -2.18947586e-02 1.23768854e+00 -7.87295341e-01
5.34629643e-01 9.55816865e-01 6.86958432e-01 -1.51862018e-03
-1.06691325e+00 -4.16712135e-01 8.41146111e-02 7.32318103e-01
-1.59762669e+00 -3.82379979e-01 5.51258862e-01 -2.77812749e-01
1.02366233e+00 5.50386369e-01 4.05558437e-01 9.26912129e-01
-2.64437109e-01 9.76147771e-01 3.90336961e-01 4.94450144e-02
1.75665021e-01 1.01818973e-02 -1.33223655e-02 6.45903945e-01
-7.34623894e-02 5.81419654e-02 -2.04897746e-01 -3.59705955e-01
4.92266059e-01 3.40208769e-01 -4.74859215e-02 -2.04135980e-02
-1.24556351e+00 8.12662721e-01 5.91343224e-01 6.01735592e-01
-4.48614806e-01 3.03177565e-01 6.21527851e-01 3.11627775e-01
4.32053268e-01 4.73992229e-01 -1.23721004e-01 -7.13342011e-01
-9.78161514e-01 5.15722811e-01 5.30668318e-01 1.06405711e+00
7.52900839e-01 -2.74907887e-01 -7.03347981e-01 5.39887667e-01
-6.95132464e-02 6.07045949e-01 4.99627739e-01 -7.49019444e-01
5.97256362e-01 8.02023828e-01 2.64649987e-01 -1.09386182e+00
-2.04541415e-01 -3.94373983e-01 -7.02509344e-01 -5.14161825e-01
2.74393350e-01 2.69471705e-01 -5.49766898e-01 1.64875591e+00
2.80005604e-01 2.43652344e-01 -2.72798717e-01 8.41987729e-01
4.49860364e-01 9.19247270e-01 1.43622840e-02 1.07387435e-02
1.15095508e+00 -9.84878182e-01 -5.09407699e-01 2.66221195e-01
1.05603898e+00 -6.59022450e-01 1.24789727e+00 -1.63191587e-01
-1.20090079e+00 -6.44088030e-01 -7.93070674e-01 -1.03348114e-01
-5.25343657e-01 3.12572986e-01 5.69049656e-01 3.01750511e-01
-7.46791005e-01 8.47553194e-01 -1.01712263e+00 -3.12867880e-01
2.04573736e-01 1.95926711e-01 -7.36591071e-02 -4.21734876e-04
-1.27690411e+00 7.91584790e-01 4.85196918e-01 -4.04499382e-01
-6.84535682e-01 -8.32855880e-01 -6.68367147e-01 4.04725999e-01
2.97787994e-01 -5.69174707e-01 1.29699111e+00 -1.95275187e-01
-1.24489272e+00 4.88635421e-01 -5.74650526e-01 -5.80494106e-01
5.03049433e-01 -1.93260327e-01 -5.11120915e-01 1.45208552e-01
2.96366066e-01 3.41546714e-01 5.47743559e-01 -6.72528625e-01
-9.20693994e-01 -1.23681515e-01 7.86129162e-02 2.76752021e-02
-5.07674813e-01 3.90208773e-02 -9.01107192e-01 -7.00491667e-01
-2.71217018e-01 -9.74219620e-01 -9.76109281e-02 -1.73088670e-01
-2.56725073e-01 -6.98638022e-01 6.51222050e-01 -4.49733138e-01
1.89241791e+00 -1.87294853e+00 -6.62083998e-02 3.05912733e-01
1.03556581e-01 6.61136329e-01 -1.14650115e-01 9.32034314e-01
3.76957953e-01 -2.39050332e-02 -3.28719765e-01 -3.16333890e-01
2.11389497e-01 2.14547589e-02 -4.61235136e-01 4.46110159e-01
2.68929392e-01 1.17352676e+00 -1.26721454e+00 -5.19549668e-01
1.37061983e-01 2.53396451e-01 -5.60238421e-01 4.78213191e-01
-2.76787877e-01 3.84868413e-01 -6.16759658e-01 3.70571494e-01
3.14219326e-01 -4.04935360e-01 -7.28489831e-02 -3.02800089e-02
-3.34388912e-01 6.10140443e-01 -7.14026511e-01 1.79309034e+00
-6.00708663e-01 6.55770659e-01 -6.77550852e-01 -1.02948427e+00
9.13190544e-01 1.47477657e-01 7.77378142e-01 -1.10625613e+00
-1.29391983e-01 3.72086614e-01 -2.31784537e-01 -5.43118179e-01
7.45289683e-01 4.51206148e-01 -2.38251969e-01 6.12635255e-01
-2.16529414e-01 4.03174460e-01 5.69147170e-01 2.12557271e-01
1.23197448e+00 -2.92317629e-01 -9.68183577e-03 -2.77076792e-02
8.19833934e-01 3.12098172e-02 5.20545661e-01 6.16629004e-01
7.00089037e-02 2.19859689e-01 2.71694720e-01 -4.51007158e-01
-1.26493692e+00 -8.89615417e-01 4.95229065e-02 1.18618679e+00
2.53505796e-01 -6.96214855e-01 -6.45209372e-01 -5.73430240e-01
2.64529854e-01 7.15271115e-01 -3.43668312e-01 -3.99986774e-01
-7.60001302e-01 -3.54753524e-01 9.20466125e-01 8.25635731e-01
6.36520326e-01 -7.77946711e-01 -5.35809696e-01 3.16818297e-01
-4.31477368e-01 -1.19973636e+00 -9.44908202e-01 -4.83208477e-01
-7.47686148e-01 -1.05595767e+00 -7.58797765e-01 -4.97331649e-01
2.95802981e-01 4.09136355e-01 1.05892038e+00 8.71242061e-02
4.56584767e-02 2.57471114e-01 -5.11910796e-01 -3.49480174e-02
-2.28457451e-01 6.52085602e-01 -7.65601024e-02 1.74045548e-01
6.45566523e-01 -6.40081227e-01 -7.38779247e-01 5.24773657e-01
-8.93087804e-01 -3.35198909e-01 5.25810003e-01 6.83160365e-01
6.43261194e-01 -6.40444979e-02 7.32821584e-01 -6.35210037e-01
8.25331986e-01 -6.86107993e-01 -4.78053570e-01 3.66393983e-01
-7.98156977e-01 3.30953330e-01 8.21565330e-01 -3.40837777e-01
-7.84023046e-01 -2.49671265e-01 -3.92967343e-01 -5.53612471e-01
-6.90194368e-02 5.88804424e-01 8.43885168e-02 1.57388568e-01
4.49748218e-01 5.39716125e-01 -3.49330813e-01 -6.12569809e-01
4.11493838e-01 7.26535976e-01 5.25598466e-01 -4.74316388e-01
7.54005015e-01 4.69428062e-01 -7.27651194e-02 -8.65565479e-01
-6.20922506e-01 -1.03549564e+00 -7.00421929e-01 -1.41141880e-02
6.07419074e-01 -7.38638937e-01 -1.02971816e+00 2.29406565e-01
-1.12985384e+00 -2.92607605e-01 -2.11308613e-01 5.97892046e-01
-7.30237246e-01 4.15529579e-01 -6.19516790e-01 -6.15468502e-01
-3.87687773e-01 -1.04359090e+00 1.18144822e+00 -9.25327092e-02
-2.12341592e-01 -1.00397408e+00 2.98513979e-01 4.68448848e-02
6.36224449e-01 8.30359757e-02 1.01206374e+00 -8.47873747e-01
-8.20334315e-01 -3.14961135e-01 -5.45736372e-01 -1.71588615e-01
1.03637710e-01 -2.30302051e-01 -7.36874521e-01 -3.51581782e-01
-3.96491379e-01 -1.72503278e-01 1.03909802e+00 -1.51956454e-01
1.31382215e+00 -5.01951277e-01 -4.95117694e-01 5.25895834e-01
1.21542764e+00 1.99942932e-01 4.96153444e-01 1.75568357e-01
6.50355399e-01 3.28361601e-01 8.83326530e-01 5.02749741e-01
5.82161188e-01 1.20874274e+00 2.41518915e-01 1.81829691e-01
-1.60870925e-02 -6.38502479e-01 1.51741967e-01 1.05976617e+00
5.01327775e-02 -3.64085197e-01 -1.02671516e+00 8.60889733e-01
-2.22090721e+00 -1.23782790e+00 -9.21747833e-02 2.39886832e+00
6.75904632e-01 -4.52568270e-02 5.30535758e-01 1.73440591e-01
4.55417693e-01 3.97017747e-01 -6.66994333e-01 -2.59314269e-01
3.06987911e-01 -4.33153696e-02 5.89749277e-01 3.48368943e-01
-1.21041632e+00 1.03137910e+00 5.69191551e+00 1.12412500e+00
-9.96080041e-01 1.04719356e-01 1.11820087e-01 2.08615679e-02
-4.92437899e-01 -2.46918514e-01 -7.04117954e-01 8.55848432e-01
1.26808071e+00 -4.55264300e-01 4.14604843e-01 7.98647165e-01
2.89144903e-01 1.53415397e-01 -1.23025608e+00 1.15995312e+00
-1.94277883e-01 -1.50295925e+00 1.15989402e-01 7.47189717e-03
4.80165362e-01 2.33524755e-01 -6.49461150e-02 4.82151359e-01
-7.14531168e-02 -8.69459152e-01 5.48955798e-01 8.02721322e-01
9.06730533e-01 -1.12919331e+00 6.74686074e-01 5.93631446e-01
-1.63491392e+00 -7.81397000e-02 -4.80685234e-01 2.34146982e-01
3.27885717e-01 3.66206914e-01 -6.63846731e-01 6.03470862e-01
4.04618561e-01 8.33448231e-01 -3.97565156e-01 1.27601111e+00
1.21443316e-01 6.87696755e-01 -3.45794350e-01 -2.04458207e-01
6.51489675e-01 -1.63732424e-01 5.63140988e-01 1.56408596e+00
3.43313545e-01 1.13994390e-01 3.57974410e-01 8.00190032e-01
-2.78776705e-01 1.73759371e-01 -6.89646363e-01 -3.30197811e-01
8.73133659e-01 6.83255553e-01 -3.40628326e-01 -5.46247244e-01
-3.63943994e-01 1.12695682e+00 5.44665396e-01 2.59032279e-01
-1.05321789e+00 -6.20954037e-01 1.09724963e+00 2.41198957e-01
4.64566767e-01 -2.61300772e-01 2.30286211e-01 -1.11089766e+00
2.93119967e-01 -5.83101213e-01 2.66529560e-01 -1.12298608e-01
-1.16844642e+00 6.36081636e-01 -2.24785376e-02 -1.44654334e+00
-6.11341715e-01 -5.91444895e-02 -6.86408579e-01 7.17040896e-01
-1.63716948e+00 -9.73730385e-01 -3.33039522e-01 6.54825151e-01
4.85572487e-01 -9.37173441e-02 8.37430596e-01 8.75645161e-01
-5.93604684e-01 8.71968210e-01 4.36425865e-01 2.36055970e-01
5.07145643e-01 -1.07146180e+00 9.53911304e-01 6.02557898e-01
3.60818118e-01 5.41211069e-01 3.06887358e-01 -5.36421955e-01
-1.41418791e+00 -1.49879551e+00 1.25592196e+00 -3.49686205e-01
6.04313850e-01 -9.26930532e-02 -1.13070667e+00 4.28740710e-01
-2.31876493e-01 -1.19256459e-01 6.90278172e-01 1.02949925e-01
-3.01216990e-01 -2.88604677e-01 -7.64252901e-01 4.28912699e-01
1.26907921e+00 -9.95990515e-01 -3.95951837e-01 3.49156678e-01
8.04040253e-01 -2.64823109e-01 -1.07394087e+00 3.77054185e-01
5.37381709e-01 -9.82281148e-01 1.07628143e+00 -8.60247493e-01
1.08289935e-01 -3.91146354e-02 2.61709373e-03 -1.03587413e+00
-4.81794566e-01 -5.96660078e-01 -5.29917300e-01 1.17278326e+00
3.20322275e-01 -5.52382231e-01 9.06462848e-01 4.69090372e-01
-9.65189561e-02 -1.19338179e+00 -7.58123517e-01 -1.22329235e+00
-1.25912547e-01 -6.59210265e-01 1.03899932e+00 7.06595182e-01
1.11118466e-01 3.19661275e-02 -2.74551213e-01 8.27270597e-02
4.62947875e-01 3.76856029e-01 7.89539933e-01 -1.02061737e+00
5.62547054e-03 -6.09005332e-01 -6.09346032e-01 -1.54683638e+00
2.32898876e-01 -1.06516480e+00 -3.32975626e-01 -1.32822561e+00
-1.31908362e-03 -7.29091287e-01 -3.36937398e-01 1.44870609e-01
-1.67662323e-01 -1.37913838e-01 1.48375392e-01 6.13092959e-01
-8.69140744e-01 9.76643145e-01 8.22014034e-01 -1.70136288e-01
-3.34447682e-01 4.91729468e-01 -2.48070493e-01 3.80550921e-01
8.61870766e-01 -5.95370352e-01 -5.94287455e-01 -4.66466665e-01
1.56214848e-01 -1.73606211e-04 1.96785435e-01 -1.02245593e+00
5.73469758e-01 -5.91073968e-02 -1.50800273e-01 -8.59843850e-01
3.73322219e-01 -6.71555817e-01 -8.84049684e-02 3.28665435e-01
-5.82661211e-01 4.54507530e-01 -5.98060973e-02 8.43176067e-01
-4.90690053e-01 -1.14162683e-01 3.73455077e-01 9.39059034e-02
-6.51800990e-01 7.28636384e-01 -3.52361083e-01 2.24116296e-01
9.23096061e-01 1.09880723e-01 -1.12879299e-01 -5.89685559e-01
-2.68865317e-01 5.63795149e-01 4.45828915e-01 5.22061050e-01
6.98926628e-01 -1.61107147e+00 -5.81026375e-01 1.36753783e-01
3.63594055e-01 -2.31831849e-01 4.33511771e-02 8.28770399e-01
-3.34808975e-01 1.18302774e+00 1.23779587e-01 -5.33279181e-01
-9.20374155e-01 7.71757483e-01 8.49920884e-02 -6.26730859e-01
-5.25880992e-01 4.77528542e-01 -3.45635355e-01 -2.55433887e-01
5.07049143e-01 -4.98287022e-01 -3.98419127e-02 1.73247620e-01
7.90028274e-01 9.71915126e-01 1.57005280e-01 -6.35680735e-01
-3.71937573e-01 4.48539644e-01 -4.43770923e-02 3.11264750e-02
1.12813675e+00 -6.70376644e-02 1.81482360e-01 3.66411477e-01
1.63371468e+00 -2.07999185e-01 -9.64338958e-01 -5.26483953e-01
4.36057806e-01 -5.62639773e-01 -1.19869597e-02 -3.01652610e-01
-8.25090826e-01 8.60339820e-01 3.17043960e-01 2.70924777e-01
9.60316896e-01 4.29472467e-03 1.57241559e+00 6.91283643e-01
5.05472660e-01 -9.64798689e-01 6.48132190e-02 4.41998988e-01
6.81315064e-01 -1.27066743e+00 -2.46520251e-01 -2.13543847e-01
-5.30891776e-01 8.93918097e-01 1.73901156e-01 -1.19694941e-01
5.85438907e-01 -2.04643711e-01 -3.81893039e-01 -2.54988879e-01
-8.00367177e-01 -6.57809198e-01 6.89130068e-01 3.62483650e-01
3.14799547e-01 -2.43188329e-02 -2.89912522e-01 1.59444213e-01
2.85149571e-02 -5.50057404e-02 -1.23136394e-01 5.90921998e-01
-3.81744385e-01 -1.19051707e+00 2.05529839e-01 4.48441625e-01
-3.02103341e-01 1.31879196e-01 -2.11706936e-01 6.40659094e-01
-2.58938700e-01 9.00475144e-01 2.13130444e-01 -6.44562185e-01
4.88566011e-01 -3.93103696e-02 6.56426996e-02 -2.92149305e-01
-6.63276255e-01 -5.28746545e-01 1.57687128e-01 -1.06176937e+00
-2.29240820e-01 -6.27378225e-01 -1.20172572e+00 -6.40822172e-01
-9.82919559e-02 4.38331842e-01 3.14041108e-01 9.29316878e-01
7.84529746e-01 2.33483613e-01 9.07968581e-01 -5.78319430e-01
-7.58256197e-01 -8.85488629e-01 -1.55953422e-01 5.77357531e-01
3.34502876e-01 -5.79747617e-01 1.02059640e-01 -3.62302542e-01]
|
[6.5940656661987305, 1.9927822351455688]
|
490bf6fa-61a8-426e-8578-c2173f11669b
|
a-field-test-of-bandit-algorithms-for
|
2304.09088
| null |
https://arxiv.org/abs/2304.09088v1
|
https://arxiv.org/pdf/2304.09088v1.pdf
|
A Field Test of Bandit Algorithms for Recommendations: Understanding the Validity of Assumptions on Human Preferences in Multi-armed Bandits
|
Personalized recommender systems suffuse modern life, shaping what media we read and what products we consume. Algorithms powering such systems tend to consist of supervised learning-based heuristics, such as latent factor models with a variety of heuristically chosen prediction targets. Meanwhile, theoretical treatments of recommendation frequently address the decision-theoretic nature of the problem, including the need to balance exploration and exploitation, via the multi-armed bandits (MABs) framework. However, MAB-based approaches rely heavily on assumptions about human preferences. These preference assumptions are seldom tested using human subject studies, partly due to the lack of publicly available toolkits to conduct such studies. In this work, we conduct a study with crowdworkers in a comics recommendation MABs setting. Each arm represents a comic category, and users provide feedback after each recommendation. We check the validity of core MABs assumptions-that human preferences (reward distributions) are fixed over time-and find that they do not hold. This finding suggests that any MAB algorithm used for recommender systems should account for human preference dynamics. While answering these questions, we provide a flexible experimental framework for understanding human preference dynamics and testing MABs algorithms with human users. The code for our experimental framework and the collected data can be found at https://github.com/HumainLab/human-bandit-evaluation.
|
['Alan L. Montgomery', 'Zachary C. Lipton', 'Fatma Kılınç-Karzan', 'Giulio Zhou', 'Liu Leqi']
|
2023-04-16
| null | null | null | null |
['multi-armed-bandits']
|
['miscellaneous']
|
[-2.69944161e-01 -1.11144967e-01 -8.19720685e-01 -1.65564761e-01
-2.89381117e-01 -7.72618115e-01 5.19891381e-01 -2.09359944e-01
-3.85582864e-01 6.09767437e-01 4.28385615e-01 -6.67641401e-01
-5.58410645e-01 -5.80403030e-01 -6.21006668e-01 -6.18570745e-01
1.33592263e-01 7.27330327e-01 -1.12501867e-01 -1.36658251e-01
3.67802441e-01 6.80012256e-02 -1.51900578e+00 3.46187949e-01
8.25674534e-01 6.99849546e-01 1.87138513e-01 7.45167375e-01
3.79956007e-01 6.38730168e-01 5.94662353e-02 -8.57521594e-01
6.96499527e-01 -2.43618548e-01 -7.03814924e-01 -3.17073688e-02
4.44303565e-02 -6.09523654e-01 -4.36656594e-01 9.21954989e-01
5.32893062e-01 5.57462990e-01 6.45767093e-01 -1.34115982e+00
-1.09177649e+00 9.40218508e-01 -3.76736462e-01 2.69129217e-01
2.43138343e-01 2.82991469e-01 1.59589076e+00 -3.95621598e-01
5.77012002e-01 1.25064898e+00 4.63261455e-01 4.70504493e-01
-1.25014889e+00 -4.60449606e-01 5.20565212e-01 1.21253416e-01
-7.69879282e-01 -5.07049203e-01 5.25872409e-01 -7.52631843e-01
5.48839092e-01 6.98696315e-01 8.36678505e-01 1.55839968e+00
-7.55016059e-02 1.05173814e+00 1.30098701e+00 -3.28057021e-01
3.42331320e-01 3.82525206e-01 5.49438834e-01 3.38279843e-01
5.90897322e-01 4.07603562e-01 -6.01775110e-01 -6.01402164e-01
8.28411877e-01 4.01415080e-01 -2.07974643e-01 -4.44212735e-01
-8.17529738e-01 1.13699722e+00 1.12670235e-01 8.47233981e-02
-6.43179893e-01 -3.27020474e-02 8.54792818e-02 3.74247491e-01
7.07485855e-01 6.29638851e-01 -4.48290795e-01 -2.25369066e-01
-7.69440830e-01 5.75834632e-01 8.91761482e-01 8.46907258e-01
3.59475493e-01 -4.44240153e-01 -4.70368505e-01 9.08485472e-01
4.68910396e-01 5.17089188e-01 7.63448894e-01 -1.24871278e+00
1.74028892e-02 2.41530240e-01 8.74681056e-01 -1.07212067e+00
-2.68492639e-01 -4.55500007e-01 -1.96272731e-01 -2.14672506e-01
7.55616486e-01 -5.08134305e-01 -4.17332321e-01 1.48641527e+00
3.11765879e-01 -1.25848055e-01 -5.59264123e-01 1.30790758e+00
6.31670952e-02 3.48842710e-01 2.64700651e-02 -3.54314417e-01
1.24156487e+00 -1.07424164e+00 -3.11549872e-01 -2.22801387e-01
4.71661687e-01 -5.83604515e-01 1.36988151e+00 6.53929234e-01
-1.12075329e+00 -1.17948465e-01 -5.38411140e-01 2.53369480e-01
-3.87771660e-03 -1.27988413e-01 1.10934329e+00 1.02359581e+00
-8.15730095e-01 6.28769696e-01 -7.26335406e-01 -5.45806587e-01
2.85368800e-01 2.39087626e-01 3.16036493e-01 -6.56293407e-02
-1.00073898e+00 6.61582410e-01 -3.83667946e-01 -1.83129534e-02
-7.58155763e-01 -5.17294645e-01 -6.95732832e-02 1.38627663e-01
4.94639575e-01 -8.69323194e-01 1.65900743e+00 -1.45448852e+00
-1.59867740e+00 5.03598869e-01 5.63042276e-02 -4.04487431e-01
6.80844128e-01 -4.44959193e-01 -1.78540260e-01 -2.36998945e-01
-5.07684387e-02 1.07160203e-01 6.77644253e-01 -1.18104863e+00
-8.07094574e-01 -3.55994791e-01 4.64524627e-01 2.34985337e-01
-3.39144021e-01 1.79900035e-01 -2.41327688e-01 -4.80609596e-01
-3.72303724e-01 -1.54470575e+00 -5.21266699e-01 -7.08580017e-01
-2.35215738e-01 -1.53116018e-01 -1.73067763e-01 -3.86839628e-01
1.34133875e+00 -1.94661522e+00 -2.06462413e-01 3.49976897e-01
4.47166450e-02 -1.66194543e-01 -2.38512471e-01 4.65714306e-01
2.84917980e-01 3.38826597e-01 4.98817712e-01 -2.03136280e-01
5.55144489e-01 -1.61140829e-01 -6.87528610e-01 6.81883693e-01
-7.76681066e-01 8.27087045e-01 -8.86337936e-01 -1.16728082e-01
-2.62165308e-01 4.33081165e-02 -1.09888053e+00 1.53744936e-01
-3.76236260e-01 1.19920969e-01 -7.08629489e-01 5.79755604e-01
2.84496188e-01 -4.76699501e-01 5.98690093e-01 2.81681150e-01
-2.03806698e-01 3.61524671e-01 -8.93462420e-01 1.05713272e+00
-2.98094809e-01 2.19845414e-01 6.44074902e-02 -6.37415707e-01
2.87853688e-01 7.50792176e-02 5.45086980e-01 -6.22105896e-01
2.48719499e-01 3.68609950e-02 2.31535107e-01 -5.27754664e-01
5.29321611e-01 1.02891121e-02 1.72569677e-01 8.95799994e-01
-3.33219349e-01 4.62591201e-01 1.49814963e-01 2.66445190e-01
1.15679824e+00 1.88219190e-01 8.27986002e-02 -5.34297943e-01
-2.40873948e-01 1.91633388e-01 6.85067713e-01 1.20640814e+00
-1.87439814e-01 2.57712424e-01 4.56116229e-01 -4.64551389e-01
-9.51548934e-01 -7.45345235e-01 7.83351511e-02 2.01300311e+00
1.77516341e-01 -1.03116125e-01 -6.24232531e-01 -5.49741685e-01
3.43585819e-01 1.01325417e+00 -7.16406047e-01 2.00612918e-01
-3.67979147e-02 -9.04502332e-01 3.92426401e-02 2.13967398e-01
-1.76159278e-01 -1.02086473e+00 -7.65057683e-01 1.84744038e-02
-2.02982560e-01 -4.36873674e-01 -7.51401722e-01 -7.94452950e-02
-7.64591694e-01 -1.08676946e+00 -6.78613484e-01 -5.34888320e-02
5.34565568e-01 6.36392653e-01 1.07972455e+00 2.83057928e-01
1.90787211e-01 8.57773781e-01 -6.35583103e-01 -3.34696978e-01
1.46492064e-01 1.00133114e-01 4.20643151e-01 6.89209849e-02
5.06557167e-01 -6.00228965e-01 -1.15018702e+00 7.16043293e-01
-5.17363012e-01 -1.37494337e-02 5.78705192e-01 7.71394134e-01
2.15253025e-01 -3.29401523e-01 3.88404697e-01 -1.32981157e+00
9.99962568e-01 -1.04980588e+00 -7.43632555e-01 3.58880281e-01
-9.25275564e-01 -1.80857927e-01 2.92864442e-01 -7.95020461e-01
-1.08286607e+00 -1.79800496e-01 3.42928469e-01 -2.71256268e-01
-3.56765464e-02 4.73399967e-01 1.50681540e-01 3.27178657e-01
8.83126616e-01 -3.29547167e-01 -2.24410146e-01 -5.06488621e-01
6.25303864e-01 8.04807186e-01 -9.32674576e-03 -1.08732295e+00
5.26770592e-01 3.59050691e-01 -7.44861841e-01 -5.29860914e-01
-1.00069737e+00 -5.05156517e-01 -1.21078379e-01 -2.58828729e-01
4.82710540e-01 -7.69721687e-01 -1.08323753e+00 -2.37194151e-01
-7.27029800e-01 -9.68812943e-01 5.41767180e-02 6.83133066e-01
-7.97018468e-01 7.06352144e-02 -7.83479452e-01 -1.35414517e+00
-1.81161076e-01 -1.03528357e+00 4.10743088e-01 2.26437375e-01
-5.73931694e-01 -7.89048731e-01 2.23400325e-01 7.98806190e-01
5.16358316e-01 -5.16659856e-01 6.42131627e-01 -8.73652816e-01
-5.09671807e-01 -2.49727964e-02 1.58299014e-01 -2.56624281e-01
-2.34963879e-01 5.45626432e-02 -8.44838500e-01 -4.07173306e-01
-1.57591045e-01 -3.53638291e-01 6.22229099e-01 7.11169064e-01
1.05001438e+00 -7.52962947e-01 -3.66712570e-01 3.02513063e-01
8.82900000e-01 1.41086519e-01 2.46880874e-01 4.48482752e-01
3.00548464e-01 8.35944533e-01 6.84061587e-01 8.41253579e-01
4.76968437e-01 6.57146215e-01 2.82260001e-01 4.16852027e-01
5.24518251e-01 -4.29243892e-01 4.86485690e-01 3.68358552e-01
-3.77165437e-01 -6.33052528e-01 -7.49779165e-01 5.96195877e-01
-2.34067559e+00 -1.04558182e+00 1.23806477e-01 2.31760311e+00
5.77766716e-01 3.56094502e-02 8.19868863e-01 -3.64079386e-01
6.58277512e-01 -3.15605849e-01 -6.67486370e-01 -3.13748091e-01
2.49741003e-01 -4.78114486e-01 8.15916598e-01 5.59173346e-01
-7.55946875e-01 8.52548182e-01 6.38383484e+00 5.41977465e-01
-9.14473474e-01 3.21158558e-01 9.09105182e-01 -6.42016172e-01
-6.62656784e-01 8.92426670e-02 -7.26548851e-01 6.17664397e-01
1.06188464e+00 -2.43297011e-01 1.09611571e+00 9.34962928e-01
4.69819129e-01 -1.44637749e-01 -1.10424864e+00 8.62112224e-01
-2.61902332e-01 -9.65643764e-01 -3.32820237e-01 4.17852432e-01
8.93204510e-01 1.39565557e-01 5.13429284e-01 3.98485482e-01
1.25459146e+00 -7.88325369e-01 1.01202536e+00 6.62795782e-01
1.78303719e-01 -5.51245511e-01 3.70986164e-01 5.35765231e-01
-1.71233892e-01 -6.88954830e-01 -4.94243830e-01 -4.61944252e-01
-3.17594968e-02 6.51545882e-01 -4.88068432e-01 -8.24377500e-03
7.75226831e-01 3.04121733e-01 -3.19641352e-01 1.06161594e+00
-7.39454851e-02 9.79019165e-01 -1.75454572e-01 -1.75993890e-01
1.35947600e-01 -4.61206794e-01 3.32417756e-01 9.82610881e-01
5.31818867e-01 3.84156197e-01 1.25342220e-01 7.19322503e-01
1.23061396e-01 3.29286128e-01 -4.96684223e-01 -3.56107235e-01
6.67146862e-01 1.28251290e+00 -8.16199362e-01 3.21219303e-02
-5.99693120e-01 6.54681981e-01 2.62412369e-01 4.83194917e-01
-8.10123920e-01 4.18644220e-01 7.52523959e-01 4.97060657e-01
1.71113983e-01 -1.32419527e-01 -4.59480673e-01 -1.16603005e+00
-4.24706817e-01 -1.26583970e+00 5.25086641e-01 -5.94534874e-01
-1.64397740e+00 9.00778323e-02 5.97954867e-03 -6.91084146e-01
-9.31239203e-02 -4.26868796e-01 -4.66505080e-01 4.74492341e-01
-9.38845396e-01 -8.89352262e-01 -1.61258560e-02 5.00333428e-01
5.25348783e-01 -5.62645458e-02 3.96851927e-01 1.40892729e-01
-7.81504810e-01 5.64108610e-01 2.39820018e-01 -4.24143761e-01
1.01094472e+00 -1.05250895e+00 1.70299053e-01 5.52936971e-01
1.33290514e-01 1.26082623e+00 1.02483320e+00 -8.23301375e-01
-1.60453486e+00 -6.09188974e-01 4.92888957e-01 -7.44708776e-01
8.13980460e-01 -2.79177010e-01 -3.47303629e-01 9.02395070e-01
2.14940414e-01 -2.65747339e-01 1.02380788e+00 8.81462455e-01
-2.34115481e-01 1.54577643e-01 -8.51434529e-01 8.67818713e-01
1.34502733e+00 -1.45293564e-01 -1.18235484e-01 4.72670764e-01
5.48036039e-01 -2.35843211e-02 -6.51682675e-01 -1.03347018e-01
1.10439086e+00 -1.16204691e+00 7.22309113e-01 -9.37008619e-01
4.76427317e-01 8.78073126e-02 -3.65651280e-01 -1.46959627e+00
-8.91595662e-01 -7.88921416e-01 -1.30224079e-01 1.00513220e+00
6.11241639e-01 -6.50696456e-01 7.96869576e-01 1.21775639e+00
1.27001539e-01 -6.87554061e-01 -2.98310697e-01 -6.41459823e-01
3.01372819e-02 -2.93833911e-01 5.78125894e-01 1.13427281e+00
2.98475116e-01 2.61919290e-01 -6.90289617e-01 1.22156693e-02
4.50434417e-01 3.33596975e-01 8.93784344e-01 -1.23513615e+00
-8.99850368e-01 -6.41513646e-01 6.31226718e-01 -1.10445833e+00
1.14554457e-01 -5.53044617e-01 7.63945356e-02 -1.27478409e+00
4.78448719e-01 -6.99504852e-01 -4.00724918e-01 3.08951914e-01
-4.23906036e-02 -2.09704190e-02 2.12711722e-01 4.85430449e-01
-9.41827297e-01 2.38865376e-01 9.45488930e-01 1.86932757e-01
-3.46388251e-01 3.60489577e-01 -1.38506925e+00 7.78034389e-01
8.74834895e-01 -4.42573845e-01 -6.31983340e-01 -2.82647341e-01
8.30608547e-01 1.07193798e-01 1.87036648e-01 -2.61583239e-01
3.03568423e-01 -9.46870625e-01 1.66214153e-01 -5.16342260e-02
-2.87472946e-03 -7.65659392e-01 4.31269318e-01 2.65530646e-01
-6.31230354e-01 1.04536228e-01 -3.58113229e-01 6.73770368e-01
5.20608306e-01 -8.82843286e-02 4.87247378e-01 -2.33667299e-01
-2.35254973e-01 3.23015541e-01 -4.37259823e-01 -2.17722416e-01
7.02939272e-01 1.09611243e-01 -4.91793364e-01 -7.37368882e-01
-5.53142309e-01 2.64344633e-01 6.84833646e-01 3.72817725e-01
-2.25574896e-03 -1.03677642e+00 -5.50096035e-01 -2.33142495e-01
-1.95058491e-02 -7.86696017e-01 2.95082003e-01 1.02073324e+00
-7.04903901e-02 4.64264482e-01 -1.92545369e-01 1.28162280e-01
-9.12189662e-01 9.26219106e-01 2.09321305e-01 -2.40592703e-01
-1.90340877e-01 7.16469765e-01 4.42616940e-01 -2.30474517e-01
1.80987924e-01 5.79744540e-02 -1.08795010e-01 2.30595663e-01
3.94086182e-01 8.50820780e-01 -3.15831602e-01 -3.08570772e-01
-8.64645839e-02 -1.75531745e-01 -3.82128954e-02 -2.59467095e-01
1.42724407e+00 -3.92635822e-01 1.04271688e-01 2.91395962e-01
3.96318227e-01 4.63923693e-01 -1.27946293e+00 -1.37205884e-01
2.08931137e-02 -7.26324975e-01 1.06275961e-01 -9.03328538e-01
-9.22055900e-01 1.31454408e-01 3.75104457e-01 5.23669302e-01
7.01648772e-01 -2.22591594e-01 4.18185145e-01 3.70820850e-01
4.69424725e-01 -1.21665847e+00 -5.32211252e-02 1.89804018e-01
7.02867985e-01 -1.11573470e+00 6.63055703e-02 -3.40758562e-02
-1.00879455e+00 5.07133186e-01 5.60383439e-01 -1.70142367e-01
8.10076535e-01 -2.86025733e-01 -1.40585244e-01 -1.08011499e-01
-1.20396924e+00 -1.67402789e-01 2.68453062e-01 1.99772999e-01
7.58637786e-01 4.64749664e-01 -6.08001351e-01 1.37347424e+00
-4.74097818e-01 8.08631778e-02 4.92159635e-01 4.09408122e-01
-3.87072474e-01 -1.14875221e+00 -4.04231220e-01 8.65518212e-01
-6.11039817e-01 -2.32123826e-02 -4.31090832e-01 3.68999958e-01
-1.63971156e-01 1.30280674e+00 -6.28418624e-02 -4.27307039e-01
1.92431286e-01 -6.54382445e-03 4.31757867e-01 -5.56426346e-01
-6.40771329e-01 3.25995028e-01 2.65872806e-01 -7.17142761e-01
-2.57474184e-01 -1.00943100e+00 -4.86428380e-01 -7.85784364e-01
-3.91451359e-01 4.08728719e-01 2.37436339e-01 6.56600416e-01
4.04191196e-01 -6.64214939e-02 5.96144617e-01 -8.07244956e-01
-7.99825013e-01 -9.14177477e-01 -8.88086379e-01 5.64579844e-01
5.97642623e-02 -8.67052257e-01 -4.21268702e-01 -7.62744471e-02]
|
[9.692510604858398, 5.591029644012451]
|
07f11079-905b-4546-8b68-679ec3d3dab1
|
byte-level-grammatical-error-correction-using
|
2305.17906
| null |
https://arxiv.org/abs/2305.17906v1
|
https://arxiv.org/pdf/2305.17906v1.pdf
|
Byte-Level Grammatical Error Correction Using Synthetic and Curated Corpora
|
Grammatical error correction (GEC) is the task of correcting typos, spelling, punctuation and grammatical issues in text. Approaching the problem as a sequence-to-sequence task, we compare the use of a common subword unit vocabulary and byte-level encoding. Initial synthetic training data is created using an error-generating pipeline, and used for finetuning two subword-level models and one byte-level model. Models are then finetuned further on hand-corrected error corpora, including texts written by children, university students, dyslexic and second-language writers, and evaluated over different error types and origins. We show that a byte-level model enables higher correction quality than a subword approach, not only for simple spelling errors, but also for more complex semantic, stylistic and grammatical issues. In particular, initial training on synthetic corpora followed by finetuning on a relatively small parallel corpus of real-world errors helps the byte-level model correct a wide range of commonly occurring errors. Our experiments are run for the Icelandic language but should hold for other similar languages, particularly morphologically rich ones.
|
['Vésteinn Snæbjarnarson', 'Vilhjálmur Þorsteinsson', 'Haukur Barri Símonarson', 'Haukur Páll Jónsson', 'Pétur Orri Ragnarsson', 'Svanhvít Lilja Ingólfsdóttir']
|
2023-05-29
| null | null | null | null |
['grammatical-error-correction']
|
['natural-language-processing']
|
[ 4.24619943e-01 -8.99379477e-02 4.59365398e-01 -2.95466840e-01
-8.70363832e-01 -5.48387051e-01 3.43647301e-01 9.45074141e-01
-8.66841376e-01 8.91401470e-01 3.41758639e-01 -6.04555845e-01
2.23500147e-01 -6.63377106e-01 -7.89912462e-01 1.58721767e-02
2.71163613e-01 6.03415430e-01 3.95496666e-01 -6.13973856e-01
8.13742161e-01 4.50855792e-02 -1.80615282e+00 3.97886574e-01
1.58531225e+00 -2.09486838e-02 4.09984469e-01 9.34006572e-01
-3.72262150e-01 6.44713819e-01 -1.03707135e+00 -7.83715665e-01
-9.08343568e-02 -3.78451824e-01 -9.42372024e-01 -2.80544549e-01
8.83354068e-01 1.39622092e-01 3.49451870e-01 1.41718221e+00
4.85048234e-01 1.08093329e-01 5.43865860e-01 -4.36952919e-01
-7.68625259e-01 8.51095319e-01 9.93865356e-02 2.99611211e-01
5.30674219e-01 2.45527864e-01 8.88468504e-01 -8.21608901e-01
7.41813183e-01 1.36509812e+00 9.71035421e-01 8.84000838e-01
-1.00853896e+00 -3.37214857e-01 -5.60314069e-03 1.10461608e-01
-1.02921283e+00 -1.84210882e-01 -4.82843770e-03 -5.73887229e-01
1.71886134e+00 1.78238913e-01 4.17045563e-01 9.88961220e-01
2.45710582e-01 4.05699432e-01 1.13108015e+00 -1.11986458e+00
1.11237064e-01 -2.16417480e-02 4.08253938e-01 7.02333152e-01
6.36373281e-01 -1.12772234e-01 -3.41019720e-01 1.51396662e-01
1.76900223e-01 -5.43494582e-01 -3.65964234e-01 6.58580184e-01
-1.10609758e+00 6.41678929e-01 -3.33744168e-01 4.32579666e-01
-4.31694277e-02 1.62389055e-01 7.19090044e-01 6.70326352e-01
5.15646338e-01 8.60708594e-01 -8.37057829e-01 -8.02859306e-01
-1.07712519e+00 5.34907997e-01 8.15356851e-01 1.14195347e+00
4.61551577e-01 2.10017487e-01 -1.45969406e-01 1.12564969e+00
9.62790474e-02 5.40780187e-01 1.03263628e+00 -4.44884509e-01
8.79645288e-01 6.50533795e-01 7.69392997e-02 -7.07013071e-01
-3.95552874e-01 -2.79446006e-01 -3.95799249e-01 2.79698581e-01
8.95396948e-01 1.02771975e-01 -9.74278927e-01 1.54354084e+00
-1.27360402e-02 -3.35729957e-01 -1.44929588e-01 5.15765548e-01
7.93551505e-01 4.55890954e-01 2.28701666e-01 -1.53610036e-01
1.44050801e+00 -9.41030502e-01 -5.91954708e-01 -3.81712347e-01
1.21464550e+00 -1.06504011e+00 1.55325866e+00 6.15272343e-01
-1.43429315e+00 -5.60814202e-01 -9.19805288e-01 -3.98001999e-01
-5.07583380e-01 8.34832713e-03 -9.93786976e-02 7.79245555e-01
-1.07217968e+00 9.19637084e-01 -5.04719675e-01 -4.14425820e-01
-2.33203551e-04 -3.12227495e-02 -2.81042546e-01 -9.73872021e-02
-1.19362581e+00 1.21715033e+00 5.73018193e-01 -7.41202712e-01
-2.29461893e-01 -8.73284459e-01 -9.68183041e-01 -1.77795619e-01
-1.29957095e-01 -2.52714872e-01 1.47696352e+00 -1.01700306e+00
-1.22965300e+00 1.28060114e+00 -1.96290955e-01 -3.83464903e-01
5.02003312e-01 -1.13014281e-01 -6.63697183e-01 -3.95141840e-01
2.64789671e-01 1.77144200e-01 7.48092473e-01 -7.30992377e-01
-9.73829925e-01 -2.07070589e-01 -3.60275149e-01 1.92729756e-02
2.44032155e-04 6.20612800e-01 3.76220495e-02 -1.25970626e+00
-1.18771397e-01 -6.75233364e-01 1.13191633e-02 -6.70109570e-01
-8.25863257e-02 -4.14983720e-01 -1.40692800e-01 -1.33630085e+00
1.81123531e+00 -1.86635697e+00 1.53248727e-01 9.97096524e-02
-1.57548949e-01 6.55344188e-01 -2.86333621e-01 3.66705298e-01
-3.19507390e-01 3.35771412e-01 -5.09625554e-01 -5.65442085e-01
2.90815085e-02 1.30846024e-01 -1.91886798e-02 9.54997838e-02
2.36833960e-01 7.48415291e-01 -1.19640815e+00 -9.84684527e-02
-1.50664151e-01 -2.80730911e-02 -8.54372919e-01 -8.20211917e-02
-4.26074535e-01 2.91072667e-01 4.35596317e-01 5.78425407e-01
4.35201705e-01 2.10880160e-01 -1.50871336e-01 7.17222631e-01
-2.33830675e-01 1.04094088e+00 -1.18514848e+00 1.48775423e+00
-7.99364150e-01 5.09035707e-01 -4.77685034e-01 -5.25295734e-01
9.74344373e-01 3.90897924e-03 -4.81463432e-01 -8.84233117e-01
5.78100830e-02 9.85481501e-01 2.89937198e-01 -5.87788582e-01
1.06298006e+00 -1.05480164e-01 -2.58468181e-01 6.43729329e-01
2.75471449e-01 -3.56319457e-01 7.27891922e-01 8.65119696e-02
1.30223846e+00 -5.98447062e-02 4.87451434e-01 -4.40967202e-01
6.41567051e-01 1.58983245e-01 4.18489844e-01 7.60299623e-01
1.19762495e-02 9.08946931e-01 3.04073930e-01 -3.22174042e-01
-1.44945371e+00 -5.03160834e-01 -2.21716821e-01 1.26369357e+00
-4.23574090e-01 -9.94264722e-01 -1.18719888e+00 -7.11114764e-01
1.25285894e-01 1.26650047e+00 -3.43908280e-01 -9.15209651e-02
-9.57546532e-01 -5.22650778e-01 8.76066029e-01 4.10740107e-01
-8.67926329e-03 -1.36000979e+00 -4.56840485e-01 6.47104561e-01
-1.19565614e-01 -8.29984307e-01 -5.91031134e-01 3.30024422e-03
-6.55521333e-01 -1.05015159e+00 -4.10897195e-01 -8.15751314e-01
6.01150453e-01 -4.01674390e-01 1.58088100e+00 9.08631623e-01
-3.16080868e-01 6.08567744e-02 -7.00626910e-01 -6.24968112e-01
-1.24552608e+00 -5.03935385e-03 2.10031211e-01 -6.29799247e-01
5.92578113e-01 -2.20377848e-01 1.37192786e-01 -1.89222232e-01
-8.08628976e-01 -1.21855088e-01 4.62367758e-02 9.84750628e-01
3.47993165e-01 -2.47156650e-01 2.19361886e-01 -1.29794669e+00
8.00799549e-01 -1.74099669e-01 -6.40002608e-01 2.88225442e-01
-7.36565828e-01 2.34744921e-01 9.04750228e-01 -2.20202833e-01
-8.51943910e-01 -3.46897572e-01 -7.19270229e-01 4.09703583e-01
-4.08490032e-01 2.95396090e-01 4.75959200e-03 7.94351101e-02
1.14549220e+00 2.45717585e-01 -2.02593535e-01 -7.08829403e-01
2.32102647e-01 8.74197483e-01 7.22772539e-01 -8.68767858e-01
6.16445959e-01 -4.28989828e-01 -4.32729453e-01 -6.90994084e-01
-5.70753217e-01 -1.88816354e-01 -7.28484869e-01 1.54729918e-01
4.17247355e-01 -6.41748846e-01 -1.55268595e-01 9.60758567e-01
-1.57676327e+00 -5.59746623e-01 -4.90338296e-01 2.16008678e-01
-3.98116738e-01 4.08007830e-01 -7.83327579e-01 -2.85777897e-01
-4.46920246e-01 -1.20737505e+00 1.05736768e+00 1.14194036e-01
-6.19843483e-01 -1.23185158e+00 2.56924152e-01 1.08245999e-01
2.80590028e-01 -2.06691101e-01 1.37090027e+00 -8.39509249e-01
-4.82060127e-02 9.37028509e-03 -5.53211756e-03 6.22623801e-01
-2.41248727e-01 1.79572105e-01 -5.31880379e-01 -2.64508575e-01
-3.58122200e-01 -2.79241413e-01 8.09677184e-01 -2.35754937e-01
8.81850064e-01 -3.82225245e-01 3.03507000e-01 5.84094346e-01
1.32852447e+00 -1.44755905e-02 8.25355768e-01 6.65264547e-01
8.18198562e-01 5.10748029e-01 4.50400233e-01 3.38189423e-01
5.94725013e-01 4.60953027e-01 8.05290714e-02 5.98037899e-01
-5.39104998e-01 -3.50468516e-01 4.46642369e-01 1.26284039e+00
-1.43814966e-01 -2.52825588e-01 -1.33701146e+00 8.74028504e-01
-1.44556844e+00 -7.62783051e-01 -7.54795969e-01 2.36760736e+00
1.44150996e+00 1.75608203e-01 7.35940486e-02 5.13476670e-01
7.38728285e-01 -3.62446547e-01 5.26462570e-02 -1.18814242e+00
-2.29174867e-01 8.15115750e-01 4.84033793e-01 7.56887078e-01
-7.01608598e-01 1.40677595e+00 6.34435987e+00 1.02094817e+00
-9.96630907e-01 2.12041065e-01 1.40013695e-01 9.58933756e-02
-6.86277092e-01 -7.35884011e-02 -1.18955028e+00 9.10972118e-01
1.27675593e+00 -5.54705262e-02 5.97746968e-01 4.80350673e-01
-1.70195717e-02 -2.33253017e-01 -9.73034322e-01 9.35769022e-01
3.41358006e-01 -1.19827688e+00 -5.10590523e-02 -4.65448499e-01
9.61141407e-01 2.16746572e-02 -4.55119908e-01 5.01018822e-01
4.73751932e-01 -1.08868933e+00 1.23789024e+00 4.07504410e-01
1.09685731e+00 -7.21982658e-01 6.08800650e-01 3.93141598e-01
-5.45730531e-01 1.14696451e-01 -5.11730731e-01 -4.72477347e-01
-1.13074630e-01 5.88024497e-01 -5.05841374e-01 1.68756783e-01
6.10108733e-01 6.48529172e-01 -1.15304422e+00 9.89186645e-01
-7.66216576e-01 9.26808417e-01 -5.06365635e-02 -2.56555855e-01
7.39134848e-02 2.46939845e-02 6.42541230e-01 1.88655174e+00
4.84886855e-01 2.04780363e-02 -3.16516012e-01 3.87125224e-01
3.93091738e-02 5.71634054e-01 -2.31117517e-01 6.10530861e-02
6.53880775e-01 8.64735901e-01 -4.29808587e-01 -4.73810166e-01
-3.17530513e-01 1.18900108e+00 7.86929548e-01 -7.20773190e-02
-4.39485908e-01 -8.28193843e-01 6.75635159e-01 2.39444003e-01
1.67616829e-01 -2.19930351e-01 -7.57392526e-01 -1.34222460e+00
7.25771040e-02 -1.41245413e+00 1.53779373e-01 -5.63074231e-01
-1.09025729e+00 5.22571862e-01 -4.36475307e-01 -9.70305800e-01
-3.23530793e-01 -9.44195449e-01 -7.45524287e-01 1.25600708e+00
-1.43011236e+00 -6.13954902e-01 -1.34126693e-01 2.38591224e-01
8.50749612e-01 -5.25446057e-01 7.97797978e-01 3.91569853e-01
-4.53252643e-01 1.05958009e+00 1.81341469e-01 1.22251078e-01
9.35790777e-01 -1.69196844e+00 1.07027745e+00 1.23249209e+00
6.08541071e-02 7.19911635e-01 7.83112407e-01 -9.25848305e-01
-6.36621952e-01 -1.39771461e+00 1.84185076e+00 -6.96109474e-01
7.97219992e-01 -2.93593377e-01 -1.26986337e+00 4.43993896e-01
2.07133852e-02 -2.50028878e-01 3.95122021e-01 9.56631452e-03
-4.06647712e-01 4.12528843e-01 -1.07719946e+00 5.38499534e-01
1.20521915e+00 -5.14492989e-01 -1.05674589e+00 4.38422143e-01
7.30842054e-01 -8.30051482e-01 -6.81837797e-01 -1.47989288e-01
2.51887739e-01 -6.87449157e-01 3.88581514e-01 -9.96427357e-01
9.96329546e-01 -2.37637535e-01 -5.25824167e-02 -2.08833361e+00
-3.51202309e-01 -6.31283343e-01 4.21671659e-01 1.40341127e+00
5.91643333e-01 -4.48998958e-01 9.30398144e-03 3.19198400e-01
-6.49576843e-01 -3.15917701e-01 -7.51615286e-01 -9.30094302e-01
6.43195987e-01 -8.04134607e-01 7.34918177e-01 9.13626850e-01
2.31749833e-01 5.28171845e-03 -7.28209317e-02 -8.41146335e-02
3.35239619e-01 -5.83994567e-01 5.59260607e-01 -1.30333138e+00
-2.59037882e-01 -5.73536754e-01 -3.87017757e-01 -4.56183195e-01
2.77518213e-01 -1.19988310e+00 4.35230970e-01 -1.16939223e+00
-3.46769571e-01 -6.22890294e-01 3.10533047e-01 4.26024824e-01
-6.86141133e-01 2.20387712e-01 1.86457485e-01 -2.88395416e-02
-3.29578310e-01 1.39176205e-01 9.36135054e-01 5.55438176e-02
3.27932909e-02 -2.42982894e-01 -5.14194846e-01 9.11588192e-01
7.74240017e-01 -7.32299805e-01 2.47334659e-01 -8.63203168e-01
8.57244253e-01 -3.41542810e-01 1.22280188e-01 -1.10331404e+00
1.77974716e-01 -1.10059977e-01 -1.22012757e-01 -9.50618163e-02
-6.37856126e-01 -1.91909894e-01 -3.34662616e-01 5.25200725e-01
-3.22967261e-01 5.37035644e-01 3.60907435e-01 1.01785064e-01
-1.55843850e-02 -1.00032902e+00 1.02161407e+00 -2.97415912e-01
-8.70102704e-01 -1.76168680e-01 -5.32992482e-01 7.73996592e-01
6.45482302e-01 -2.94402361e-01 -5.23173392e-01 1.70097217e-01
-4.89811689e-01 -2.20913794e-02 8.22405875e-01 5.08870542e-01
3.99669349e-01 -1.12897694e+00 -1.16518128e+00 7.23768890e-01
4.64896202e-01 -1.59766525e-01 -1.73868626e-01 4.39027578e-01
-1.12822104e+00 3.13541621e-01 -2.12182134e-01 -2.40499467e-01
-1.35313356e+00 1.62022248e-01 3.46686393e-01 -2.54190624e-01
-5.05876780e-01 1.18729436e+00 -6.10677898e-01 -8.52594316e-01
1.55018076e-01 -7.13568568e-01 -1.72829658e-01 -2.36864928e-02
1.00296438e+00 5.27827680e-01 7.96837866e-01 -7.63426185e-01
-2.48097986e-01 6.23492181e-01 -1.56368077e-01 -8.34516808e-02
1.04607391e+00 -1.19013853e-01 -2.94630915e-01 4.16992068e-01
6.87017083e-01 4.04624641e-01 -5.94298244e-01 -3.00075442e-01
3.77052546e-01 -4.16007042e-01 -2.55182594e-01 -1.13150346e+00
-3.65275294e-01 9.44474041e-01 1.90060273e-01 -7.98516497e-02
8.05734813e-01 -5.06077707e-01 8.41159821e-01 2.18249410e-01
4.90133524e-01 -1.53817379e+00 -2.84535140e-01 1.46982503e+00
7.74762332e-01 -1.10395718e+00 -4.33797479e-01 -1.91553667e-01
-5.05300701e-01 1.31628549e+00 6.31305099e-01 -2.69404173e-01
1.53239384e-01 2.44579986e-01 -6.77733030e-03 2.32795030e-01
-7.22624540e-01 -1.85078621e-01 2.77157009e-01 5.81891000e-01
8.45397711e-01 9.08227414e-02 -1.04437232e+00 6.72575116e-01
-7.47447968e-01 -2.25681946e-01 9.92838323e-01 7.01672018e-01
-5.75754046e-01 -1.52514279e+00 -5.27148128e-01 5.17118335e-01
-5.64378977e-01 -7.83001840e-01 -1.07818879e-01 6.45811558e-01
5.85819244e-01 7.50136733e-01 3.07621092e-01 -1.54817283e-01
5.65970004e-01 3.39169770e-01 7.42725313e-01 -1.24979043e+00
-1.28105903e+00 -5.82380593e-01 3.42683971e-01 -3.24751496e-01
2.61801243e-01 -1.00997388e+00 -1.40858638e+00 -5.46682060e-01
-9.07116607e-02 -1.26565602e-02 7.29713142e-01 1.27452612e+00
7.45546771e-03 5.85501730e-01 -2.00673658e-02 -3.84778500e-01
-7.59909034e-01 -1.30357742e+00 -5.02211988e-01 8.58120859e-01
2.82790273e-01 -2.25080788e-01 -3.75575900e-01 1.32530108e-01]
|
[11.044946670532227, 10.667569160461426]
|
33d1af8c-edb5-4061-9961-0283b0455b74
|
transformer-based-unet-with-multi-headed
|
2306.02815
| null |
https://arxiv.org/abs/2306.02815v1
|
https://arxiv.org/pdf/2306.02815v1.pdf
|
Transformer-Based UNet with Multi-Headed Cross-Attention Skip Connections to Eliminate Artifacts in Scanned Documents
|
The extraction of text in high quality is essential for text-based document analysis tasks like Document Classification or Named Entity Recognition. Unfortunately, this is not always ensured, as poor scan quality and the resulting artifacts lead to errors in the Optical Character Recognition (OCR) process. Current approaches using Convolutional Neural Networks show promising results for background removal tasks but fail correcting artifacts like pixelation or compression errors. For general images, Transformer backbones are getting integrated more frequently in well-known neural network structures for denoising tasks. In this work, a modified UNet structure using a Swin Transformer backbone is presented to remove typical artifacts in scanned documents. Multi-headed cross-attention skip connections are used to more selectively learn features in respective levels of abstraction. The performance of this approach is examined regarding compression errors, pixelation and random noise. An improvement in text extraction quality with a reduced error rate of up to 53.9% on the synthetic data is archived. The pretrained base-model can be easily adapted to new artifacts. The cross-attention skip connections allow to integrate textual information extracted from the encoder or in form of commands to more selectively control the models outcome. The latter is shown by means of an example application.
|
['Michael Munz', 'David Kreuzer']
|
2023-06-05
| null | null | null | null |
['optical-character-recognition', 'document-classification']
|
['computer-vision', 'natural-language-processing']
|
[ 8.45805585e-01 -5.01393564e-02 3.72933149e-01 -2.66374528e-01
-5.13501823e-01 -2.14818016e-01 6.78470731e-01 2.57130086e-01
-7.41832197e-01 6.36739969e-01 7.01230243e-02 9.94418748e-03
-1.45538986e-01 -7.85195351e-01 -8.36584032e-01 -7.53625393e-01
2.37701952e-01 1.54327795e-01 1.69342548e-01 -1.30420670e-01
6.77942276e-01 8.58870089e-01 -1.57308650e+00 8.90158534e-01
9.70369935e-01 8.08567286e-01 5.69646299e-01 9.83087182e-01
-4.71619904e-01 8.10803711e-01 -1.07454157e+00 -5.26537657e-01
9.41771194e-02 -2.84799635e-01 -6.10605180e-01 2.69927531e-01
6.93446696e-01 -3.47863108e-01 -1.97219968e-01 1.05575705e+00
5.51045597e-01 -9.21051055e-02 6.41796052e-01 -5.84472299e-01
-6.70718551e-01 7.47814476e-01 -3.86023730e-01 1.37569651e-01
1.87038258e-01 1.05263457e-01 6.56582654e-01 -9.72320676e-01
7.43009329e-01 1.08560228e+00 6.55368328e-01 3.75846714e-01
-1.47715104e+00 -3.44294339e-01 -1.71340466e-01 5.19513130e-01
-1.10346329e+00 -3.95528704e-01 5.46929777e-01 -2.20595926e-01
1.24145174e+00 4.66094226e-01 4.75797474e-01 1.17913985e+00
4.59000796e-01 7.34263122e-01 9.26423490e-01 -7.43186116e-01
-6.34108856e-02 4.08985227e-01 1.73286408e-01 4.31639522e-01
4.96246904e-01 -3.89103413e-01 -7.51063466e-01 6.40843391e-01
4.54918504e-01 -5.07086888e-02 -4.22361791e-01 7.72793293e-02
-9.87986863e-01 4.81283814e-01 4.33872670e-01 8.71358752e-01
-4.34699714e-01 1.93260014e-01 4.60430801e-01 3.81076694e-01
3.95009845e-01 5.11480093e-01 -3.47273171e-01 1.18845655e-02
-1.50765717e+00 -9.70235169e-02 5.58499277e-01 9.76090848e-01
4.10813272e-01 4.19833124e-01 -5.79276860e-01 9.05816436e-01
-9.59154442e-02 4.38158184e-01 5.33718526e-01 -4.73037779e-01
6.43385351e-01 6.23689592e-01 -1.00035317e-01 -9.99632299e-01
-3.59673649e-01 -6.16329849e-01 -1.01955545e+00 5.92436731e-01
3.55791032e-01 2.18023628e-01 -1.16262484e+00 9.53935921e-01
-1.86562732e-01 -4.17201132e-01 -6.10854477e-02 5.32416880e-01
5.34589589e-01 8.82047892e-01 -2.00307712e-01 -1.40791625e-01
1.43246531e+00 -8.36768448e-01 -1.17083633e+00 -2.28704780e-01
5.33007383e-01 -9.68369961e-01 1.08661461e+00 9.93578494e-01
-1.18153346e+00 -8.00142825e-01 -1.27704203e+00 -3.87767375e-01
-8.04211199e-01 4.75503564e-01 -2.97884673e-01 8.97099376e-01
-9.46052134e-01 9.90526199e-01 -5.97150981e-01 -3.77924860e-01
6.11192763e-01 4.53678101e-01 -4.17389721e-01 -7.04177842e-02
-8.84480357e-01 1.02189493e+00 4.40204859e-01 2.63072044e-01
-4.21782643e-01 -4.00361776e-01 -5.23258150e-01 4.84754443e-01
2.61973828e-01 -1.70476496e-01 8.10907125e-01 -1.17242873e+00
-1.58045697e+00 6.83741987e-01 4.25505079e-02 -7.27318108e-01
7.69152999e-01 -6.46455288e-01 -4.27932739e-01 2.92855501e-01
-2.70724356e-01 4.26377654e-01 1.35558963e+00 -1.02011704e+00
-4.68514472e-01 -3.07481617e-01 -5.52518010e-01 -1.51435688e-01
-7.59004593e-01 1.51162863e-01 -4.58751887e-01 -1.14249945e+00
4.43896651e-02 -3.50489587e-01 2.52996325e-01 1.06046423e-01
-5.84078789e-01 2.89947331e-01 1.15803552e+00 -1.09882522e+00
1.33496487e+00 -1.82899475e+00 1.76777139e-01 1.26012415e-01
-1.99227348e-01 4.92474556e-01 -2.42570072e-01 3.22754413e-01
-2.66830474e-01 1.84891000e-01 -3.46726388e-01 -5.24939537e-01
-2.67148852e-01 -1.13980435e-01 -1.02752656e-01 3.22105587e-01
6.52187467e-01 6.29481733e-01 -2.42819339e-01 -4.71448004e-01
4.19184476e-01 8.80412519e-01 -4.49865848e-01 -3.45538370e-02
-2.24307418e-01 1.15200244e-01 3.95335145e-02 4.44976807e-01
7.07159817e-01 1.00172527e-01 -8.92311186e-02 -5.32605886e-01
-2.65794843e-01 2.11170048e-01 -1.23902583e+00 1.56012583e+00
-5.73936582e-01 1.17203689e+00 1.57556549e-01 -8.84576797e-01
1.06679666e+00 2.97324568e-01 -8.03541318e-02 -9.84966338e-01
2.71721870e-01 2.18801513e-01 -8.51554275e-02 -6.63520575e-01
9.24465954e-01 3.22351396e-01 4.46860731e-01 1.22364566e-01
1.23120636e-01 -1.89225003e-01 4.21753168e-01 1.26285786e-02
9.42486167e-01 1.98153824e-01 -2.06108503e-02 -1.95056260e-01
8.49914432e-01 -1.39739588e-01 -1.23692240e-04 8.46407413e-01
3.28549773e-01 9.41223323e-01 5.43747544e-01 -3.30829322e-01
-1.39596534e+00 -3.27322811e-01 -2.26504385e-01 8.25292945e-01
-5.16612470e-01 -3.84777963e-01 -9.68562186e-01 -4.98154163e-01
-4.14580315e-01 1.05893660e+00 -5.61132789e-01 -1.37051076e-01
-8.46203268e-01 -6.48172021e-01 4.56582934e-01 4.18780208e-01
6.28898680e-01 -1.40928137e+00 -8.61820340e-01 4.04098630e-01
6.50539100e-02 -1.02009749e+00 -1.61192372e-01 6.97602212e-01
-1.11562204e+00 -7.14691818e-01 -8.77603471e-01 -5.79398215e-01
6.85120344e-01 -1.11740299e-01 8.50595891e-01 2.52676755e-01
-5.56115508e-01 1.10540234e-01 -4.85470027e-01 -3.32700640e-01
-7.49598026e-01 2.68228173e-01 -3.72558624e-01 3.04622233e-01
2.84763902e-01 -2.22568214e-01 -3.44162494e-01 -4.64324430e-02
-1.36685574e+00 -3.21770236e-02 7.02662289e-01 1.09823692e+00
3.29111308e-01 1.45795241e-01 1.14333183e-01 -9.19849932e-01
6.49880290e-01 2.63952851e-01 -7.26358056e-01 1.45533323e-01
-6.74036443e-01 3.89270693e-01 9.58297968e-01 -4.14311081e-01
-1.18059754e+00 -3.62910852e-02 -2.29079172e-01 -1.55657306e-01
-4.08802658e-01 2.57763714e-01 -3.10329467e-01 1.00705968e-02
7.28276074e-01 4.51490015e-01 -2.96359003e-01 -6.45304978e-01
-9.39981937e-02 8.53882492e-01 3.91611993e-01 -1.73954144e-02
8.54423702e-01 3.40262264e-01 1.63762271e-02 -1.17477691e+00
-4.37274098e-01 7.42760897e-02 -8.00602436e-01 -2.15820402e-01
9.15302455e-01 -4.80645925e-01 -2.41113558e-01 7.60640085e-01
-1.45934367e+00 -2.38914832e-01 -3.46695989e-01 2.51758695e-01
-2.39653647e-01 3.67122412e-01 -6.60189092e-01 -8.25945973e-01
-5.58647990e-01 -1.39842629e+00 9.58320379e-01 2.55423874e-01
-3.62155028e-02 -5.78349769e-01 -4.56463456e-01 2.97862828e-01
5.68969011e-01 -2.14708030e-01 9.67974663e-01 -6.21877432e-01
-5.75890541e-01 -4.46355283e-01 -2.39321947e-01 6.90320492e-01
-1.86933111e-02 3.31603169e-01 -1.20241618e+00 -3.49355251e-01
2.25711334e-02 8.75830576e-02 1.19798660e+00 2.79051423e-01
1.38246441e+00 -2.64232665e-01 -4.38055955e-02 4.97118831e-01
1.56244540e+00 2.77843952e-01 1.28916049e+00 5.90802908e-01
5.65570235e-01 7.21919239e-01 2.78164804e-01 4.47156340e-01
-6.64701283e-01 6.17022753e-01 5.15852511e-01 2.73008980e-02
-4.26567793e-01 2.32880816e-01 4.64117616e-01 5.77934086e-01
-6.52713627e-02 -6.14883542e-01 -7.41690218e-01 3.13597143e-01
-1.12652719e+00 -1.11854315e+00 -5.68791330e-01 2.17979407e+00
7.63323545e-01 5.01862526e-01 -4.39421773e-01 6.62341356e-01
8.20010841e-01 1.57458372e-02 -1.97867945e-01 -7.48405159e-01
-4.53943193e-01 5.34981132e-01 6.13764286e-01 4.51534361e-01
-9.25682127e-01 6.58412576e-01 5.02571440e+00 9.36341941e-01
-1.17432857e+00 -3.96431088e-02 4.93421972e-01 -2.48546183e-01
2.34427843e-02 -3.65614414e-01 -9.27695155e-01 5.41580677e-01
1.01817024e+00 5.55515945e-01 1.22629248e-01 4.55579191e-01
4.21145111e-01 -2.31458798e-01 -8.68552625e-01 1.02222204e+00
4.01390374e-01 -1.34607208e+00 3.58920038e-01 -7.35914484e-02
4.13039714e-01 -4.21233892e-01 9.29648206e-02 -4.97826524e-02
-4.92623627e-01 -1.12722075e+00 1.02578759e+00 8.67208958e-01
8.56344223e-01 -8.09387743e-01 6.99757636e-01 2.70643625e-02
-7.54055858e-01 -2.63735741e-01 -2.81571358e-01 3.96632701e-01
1.26103446e-01 7.66365588e-01 -6.31488144e-01 4.09461260e-01
8.28964949e-01 4.93734002e-01 -8.35331619e-01 1.13383412e+00
-1.33402169e-01 4.50643510e-01 -3.22656512e-01 -1.15135707e-01
1.26771614e-01 -2.46987551e-01 6.97017372e-01 1.74475873e+00
5.27338922e-01 -4.73028839e-01 -8.98596287e-01 9.47861373e-01
-2.97906339e-01 2.78586060e-01 -4.51349705e-01 -1.37418330e-01
-5.58875725e-02 1.28443372e+00 -9.03521001e-01 -3.88939857e-01
-3.95298809e-01 1.50072801e+00 -7.89259747e-02 2.54374087e-01
-6.37563050e-01 -9.07964408e-01 1.62755266e-01 3.97158742e-01
8.40470970e-01 -1.04852282e-02 -3.78729433e-01 -9.89538014e-01
1.55418679e-01 -9.68126476e-01 -1.01181790e-01 -1.00099313e+00
-7.40652978e-01 7.72738993e-01 -4.00044560e-01 -1.05437422e+00
1.75226927e-01 -1.08864450e+00 -5.07044792e-01 9.28466678e-01
-1.31928635e+00 -8.59672606e-01 -4.34735060e-01 3.48847240e-01
1.08575094e+00 -3.46663862e-01 6.95963502e-01 4.73359704e-01
-7.82571435e-01 6.11423790e-01 4.46031064e-01 8.95519555e-02
7.74556696e-01 -1.24546158e+00 2.63031065e-01 1.24135387e+00
2.39937112e-01 2.63979912e-01 8.99924815e-01 -6.66942418e-01
-1.11933994e+00 -9.90301907e-01 9.29536283e-01 -4.83331308e-02
2.45446220e-01 -5.35947740e-01 -1.23447669e+00 3.31464618e-01
6.52911067e-01 -2.90926188e-01 7.61224702e-02 -4.74344343e-01
-3.07420552e-01 -2.92839706e-01 -1.18726897e+00 6.30678475e-01
3.98575395e-01 -3.16063493e-01 -3.59587193e-01 2.79805690e-01
3.76682609e-01 -1.42880142e-01 -5.61022580e-01 -7.98806772e-02
4.37492818e-01 -1.30438411e+00 7.88127005e-01 -3.36856633e-01
7.21494198e-01 -2.28460670e-01 1.61837582e-02 -9.50286388e-01
-1.70664549e-01 -5.17645299e-01 -4.53250818e-02 1.33262062e+00
5.29300749e-01 -1.18480735e-01 8.89569163e-01 2.08740145e-01
-2.67426181e-03 -2.46156737e-01 -8.86576295e-01 -4.88548428e-01
-1.94254462e-02 -3.87779057e-01 1.04522921e-01 5.25479794e-01
-4.79666114e-01 3.46922159e-01 -5.03250718e-01 8.29388201e-03
4.51704413e-01 -3.57859910e-01 4.01633739e-01 -1.24046433e+00
-1.98295638e-01 -5.73481262e-01 -3.37349802e-01 -7.68240452e-01
-2.85384208e-01 -6.07402623e-01 1.24760531e-02 -1.43783867e+00
-7.00293183e-02 9.84267518e-02 2.94050034e-02 1.87199384e-01
1.11732371e-01 3.34034175e-01 3.66851836e-01 -1.16913237e-01
-5.96255548e-02 3.91455293e-01 9.82323110e-01 -5.35294652e-01
-1.40240073e-01 -7.61445463e-02 -1.35863394e-01 4.92620647e-01
8.87741208e-01 -6.25844061e-01 6.03082553e-02 -6.56026721e-01
4.31563437e-01 -2.76314706e-01 3.05155247e-01 -1.40559125e+00
1.91850901e-01 4.64153618e-01 9.19527829e-01 -8.76013994e-01
2.82988787e-01 -1.26672029e+00 -2.66658999e-02 6.32531285e-01
-4.96551275e-01 1.19571485e-01 3.78542811e-01 4.44818050e-01
-2.44545102e-01 -9.39216256e-01 1.00693166e+00 -9.40786675e-02
-3.30645621e-01 -3.53999197e-01 -7.75834203e-01 -5.86224079e-01
6.02695525e-01 -6.95760906e-01 -1.27599537e-01 -2.96568513e-01
-7.48114705e-01 -3.90199095e-01 2.46945545e-01 2.05621362e-01
7.26075411e-01 -7.96423793e-01 -7.76131213e-01 3.91267508e-01
-1.23990417e-01 -1.60032094e-01 3.20763230e-01 8.67912531e-01
-9.24757600e-01 4.59645152e-01 -4.04302955e-01 -4.62406427e-01
-1.55901229e+00 5.20414948e-01 5.15033662e-01 -3.98989111e-01
-7.13637412e-01 7.22211063e-01 -4.18711543e-01 1.50085941e-01
4.77963835e-01 -5.18996656e-01 -4.07215297e-01 3.88149410e-01
8.17020476e-01 5.21946609e-01 6.99947059e-01 -2.94069707e-01
1.13454340e-02 5.18369615e-01 -4.93739098e-01 -5.07331155e-02
1.54288936e+00 -6.56075850e-02 -8.89675319e-02 2.37125456e-01
1.13856971e+00 8.41121525e-02 -1.15209568e+00 -2.64293142e-02
2.18267739e-01 -2.65169472e-01 2.84586936e-01 -9.81969297e-01
-1.25062025e+00 1.22686887e+00 9.54490006e-01 3.45101267e-01
1.37962747e+00 -6.75548315e-01 3.80056173e-01 5.89281857e-01
-1.33627817e-01 -1.42665482e+00 1.63129851e-01 4.28942710e-01
1.23364556e+00 -9.82714951e-01 1.57292292e-01 -7.62798414e-02
-3.24599206e-01 1.73724043e+00 3.83393586e-01 -1.49727277e-02
1.30784363e-01 5.93329012e-01 -1.98291957e-01 5.46796545e-02
-5.18517256e-01 -8.32693931e-03 3.51298377e-02 5.51608443e-01
6.26309216e-01 -6.16360426e-01 -2.67337888e-01 2.14427948e-01
1.64872423e-01 -1.42431498e-01 9.57951725e-01 9.86421704e-01
-4.52815473e-01 -1.12070751e+00 -8.52134407e-01 7.18303621e-01
-6.92038536e-01 -5.06455839e-01 -3.93296659e-01 6.78088069e-01
2.09096789e-01 7.00282753e-01 2.43852854e-01 -8.73902515e-02
4.97407049e-01 3.83789748e-01 4.58881587e-01 -4.15444911e-01
-1.17142165e+00 2.47317433e-01 -2.15481240e-02 -4.45466727e-01
-4.08013731e-01 -5.92039347e-01 -8.44770908e-01 -1.58095166e-01
-6.24238849e-01 -2.39791900e-01 9.85569060e-01 6.68626249e-01
1.80108547e-01 1.03132963e+00 8.64068791e-02 -8.76958072e-01
-3.30757976e-01 -1.26970637e+00 -5.46682239e-01 3.60964686e-01
3.62695634e-01 -1.88492358e-01 -2.94312596e-01 5.61416268e-01]
|
[11.79328727722168, 2.5644261837005615]
|
3da9af02-e726-4ec7-ab21-2bef6ae1f4e6
|
improving-event-causality-identification-via
|
2106.01654
| null |
https://arxiv.org/abs/2106.01654v1
|
https://arxiv.org/pdf/2106.01654v1.pdf
|
Improving Event Causality Identification via Self-Supervised Representation Learning on External Causal Statement
|
Current models for event causality identification (ECI) mainly adopt a supervised framework, which heavily rely on labeled data for training. Unfortunately, the scale of current annotated datasets is relatively limited, which cannot provide sufficient support for models to capture useful indicators from causal statements, especially for handing those new, unseen cases. To alleviate this problem, we propose a novel approach, shortly named CauSeRL, which leverages external causal statements for event causality identification. First of all, we design a self-supervised framework to learn context-specific causal patterns from external causal statements. Then, we adopt a contrastive transfer strategy to incorporate the learned context-specific causal patterns into the target ECI model. Experimental results show that our method significantly outperforms previous methods on EventStoryLine and Causal-TimeBank (+2.0 and +3.4 points on F1 value respectively).
|
['Yuguang Chen', 'Weihua Peng', 'Jun Zhao', 'Kang Liu', 'Yubo Chen', 'Pengfei Cao', 'Xinyu Zuo']
|
2021-06-03
| null |
https://aclanthology.org/2021.findings-acl.190
|
https://aclanthology.org/2021.findings-acl.190.pdf
|
findings-acl-2021-8
|
['event-causality-identification']
|
['natural-language-processing']
|
[ 1.65007368e-01 3.04754764e-01 -8.18603337e-01 -5.61957359e-01
-8.08068931e-01 -5.07146299e-01 8.79623234e-01 2.13998958e-01
-1.40224949e-01 1.09829235e+00 6.31717086e-01 -3.58151406e-01
-1.73586294e-01 -7.20340133e-01 -8.04643512e-01 -2.67046988e-01
-2.60552585e-01 2.10520968e-01 3.60285699e-01 1.96934074e-01
-8.70314613e-03 3.08210272e-02 -8.65208685e-01 4.50167567e-01
8.07974458e-01 5.71177959e-01 -4.22964571e-03 3.01146388e-01
-1.98127359e-01 1.47366500e+00 -5.69277048e-01 -3.57295454e-01
-4.61206466e-01 -7.09263027e-01 -9.48769629e-01 -5.37184596e-01
-2.19352201e-01 -4.45964277e-01 -4.70896661e-01 6.98689938e-01
1.55308574e-01 -1.35114774e-01 7.46900678e-01 -1.47901821e+00
-4.01239127e-01 1.45929074e+00 -4.36392784e-01 5.90760469e-01
5.64743757e-01 -1.99306160e-01 1.35074091e+00 -7.41134226e-01
6.79941654e-01 1.46831000e+00 6.01227283e-01 5.21319747e-01
-1.01082337e+00 -1.08879650e+00 6.22469306e-01 4.43999171e-01
-8.12689483e-01 -3.95259887e-01 8.79361451e-01 -2.67615318e-01
9.52647090e-01 1.57989755e-01 3.65220785e-01 1.58464611e+00
-1.41347826e-01 1.05732715e+00 7.60430396e-01 -3.98401976e-01
1.32715106e-01 -1.97909206e-01 1.69608697e-01 3.17893803e-01
-5.88267064e-03 4.39310193e-01 -9.69581187e-01 -2.64243215e-01
6.94955766e-01 -2.30144262e-01 -2.09809080e-01 2.95384824e-01
-1.41123867e+00 7.85427690e-01 3.15724999e-01 2.80123472e-01
-3.59594166e-01 5.08161902e-01 5.85415959e-01 1.66611359e-01
5.13996005e-01 4.85210381e-02 -8.02593410e-01 -3.05026889e-01
-4.78809774e-01 4.07925576e-01 7.23349333e-01 9.56261396e-01
2.58922458e-01 -7.39220530e-02 -1.40192017e-01 5.67137837e-01
5.21235466e-01 3.29513133e-01 3.95338416e-01 -6.63587511e-01
6.42133236e-01 6.50748014e-01 1.30477071e-01 -8.41708064e-01
-4.10463572e-01 -2.33867750e-01 -5.34892976e-01 -6.03418827e-01
3.64181221e-01 -3.81967396e-01 -5.41170597e-01 1.92749298e+00
3.56017590e-01 7.15221405e-01 3.51564884e-02 7.22748220e-01
8.27446818e-01 7.04243302e-01 5.68363726e-01 -4.72875088e-01
1.06671560e+00 -6.88986778e-01 -9.77527857e-01 -3.12055051e-01
6.81923449e-01 -5.59948444e-01 1.01499510e+00 1.38542548e-01
-8.55778754e-01 -2.38216981e-01 -7.90442348e-01 4.10961658e-01
5.10566831e-02 1.64793544e-02 8.82521987e-01 1.99822336e-01
-2.90078044e-01 2.51565516e-01 -9.38869298e-01 -1.71700135e-01
2.18717888e-01 3.33008505e-02 -1.37864709e-01 1.86413571e-01
-1.83478749e+00 5.90203106e-01 9.11722481e-01 -1.35883808e-01
-1.23502946e+00 -1.17710638e+00 -7.83946216e-01 -4.98034060e-02
6.05243862e-01 -3.51366341e-01 1.47685790e+00 -6.98223233e-01
-1.14828634e+00 3.77463907e-01 -3.50791395e-01 -6.57601297e-01
5.22706330e-01 -6.32955432e-01 -7.56366611e-01 4.66176383e-02
3.80891442e-01 2.26690620e-01 6.34501576e-01 -1.25298154e+00
-9.81976151e-01 9.71680135e-03 1.28291011e-01 -1.01267174e-01
-4.36565846e-01 7.42530584e-01 -2.21207187e-01 -9.44270134e-01
-2.50517875e-01 -6.20506346e-01 -1.66167021e-01 -6.07009709e-01
-6.16684139e-01 -6.73851073e-01 9.93011653e-01 -5.38264215e-01
1.70711398e+00 -2.09231830e+00 -1.20087184e-01 -1.04163721e-01
1.10143106e-02 -1.20794838e-02 1.64348811e-01 3.73669028e-01
-6.13694787e-01 2.84347147e-01 -1.45509541e-01 7.45283961e-02
-1.49873465e-01 3.59223157e-01 -9.11764562e-01 1.48814572e-02
6.66128337e-01 8.03920686e-01 -1.42656636e+00 -7.46418774e-01
6.55378327e-02 1.23514775e-02 -5.63209772e-01 4.93547529e-01
-4.31897432e-01 6.81102991e-01 -7.57330596e-01 3.22030216e-01
6.04972616e-02 -3.80617738e-01 4.93130714e-01 -1.42649889e-01
-1.66969627e-01 9.72606957e-01 -1.14526451e+00 1.38768375e+00
-4.51711178e-01 2.23664001e-01 -5.49459934e-01 -1.17965055e+00
6.96350038e-01 9.39466834e-01 4.93330717e-01 -3.29683989e-01
-1.17867440e-01 1.71003729e-01 -1.35288879e-01 -5.26669800e-01
-1.97021052e-01 -1.88781396e-01 -2.51727611e-01 4.27673757e-01
1.37808561e-01 5.18280447e-01 2.96437979e-01 4.50942725e-01
1.32612526e+00 3.53085339e-01 4.59364891e-01 -1.69349629e-02
3.85923266e-01 1.48251966e-01 1.17231452e+00 4.96730059e-01
-9.99303460e-02 4.25766647e-01 8.68204951e-01 -3.52210552e-01
-5.20722926e-01 -9.19706702e-01 -1.02005877e-01 9.25958812e-01
-3.03614829e-02 -6.55382037e-01 -4.26360726e-01 -1.39806390e+00
-1.88997865e-01 1.12344837e+00 -5.68793833e-01 -1.32823572e-01
-1.07947612e+00 -1.12722480e+00 6.48670793e-01 1.02252114e+00
4.28463966e-01 -1.22286320e+00 -4.97831434e-01 6.07627451e-01
-6.38227999e-01 -1.11212015e+00 -2.73463666e-01 2.54430145e-01
-7.56278574e-01 -1.11893547e+00 -1.19563811e-01 -3.89257014e-01
2.03871682e-01 -1.01399362e-01 1.14662564e+00 -1.90523610e-01
2.20679551e-01 -1.07807964e-01 -5.17996252e-01 -6.06175303e-01
-5.48156977e-01 6.03259206e-02 4.87843677e-02 -2.90723443e-02
3.92215014e-01 -6.14688098e-01 -3.26833397e-01 3.26351345e-01
-5.94381511e-01 2.33198218e-02 3.66712838e-01 9.40015793e-01
1.58301353e-01 2.21208557e-01 1.25451612e+00 -1.14145792e+00
3.86794627e-01 -8.90409589e-01 -3.15223813e-01 2.98436224e-01
-6.15312397e-01 -3.62503789e-02 6.96086407e-01 -4.78197306e-01
-1.84724450e+00 -1.23006403e-01 8.62390101e-02 -1.32801384e-01
-3.28974366e-01 9.47267890e-01 -3.58660668e-01 8.26079488e-01
6.68230236e-01 -3.18458170e-01 -7.88536966e-01 -4.23909932e-01
4.52748179e-01 4.41466600e-01 7.10585535e-01 -6.76906347e-01
7.40306854e-01 3.24456245e-01 -4.55353677e-01 -2.97991484e-02
-1.31758726e+00 -3.59393656e-01 -5.67083478e-01 -2.38920271e-01
7.23164380e-01 -1.06644559e+00 -3.18903983e-01 1.58903494e-01
-1.42822623e+00 -5.48753202e-01 -7.95527995e-02 7.70133734e-01
-3.38306963e-01 6.16622791e-02 -7.41204977e-01 -7.95445323e-01
-9.37515274e-02 -5.42964876e-01 8.22587252e-01 1.01416960e-01
-7.70330906e-01 -1.26131260e+00 1.64572969e-01 9.01932195e-02
-9.58471149e-02 3.09916109e-01 1.16200483e+00 -6.72300637e-01
-2.95866102e-01 -2.21364405e-02 -3.48627597e-01 -8.66257474e-02
3.16540033e-01 1.57526731e-01 -9.65322733e-01 3.00484300e-01
-2.29101360e-01 -3.69494200e-01 7.63039708e-01 1.48397073e-01
1.07079077e+00 -4.60911721e-01 -8.13603401e-01 -1.11732644e-03
9.14496303e-01 3.25007468e-01 2.54641443e-01 2.92116374e-01
7.55007148e-01 5.46943784e-01 9.15772021e-01 6.38047576e-01
6.60808861e-01 4.84255522e-01 3.05526912e-01 -9.64358598e-02
-2.53149986e-01 -8.52930903e-01 4.96533662e-01 6.53950810e-01
-1.15762606e-01 -2.14815900e-01 -1.05926406e+00 9.48375881e-01
-2.13776016e+00 -1.05716002e+00 -4.79899615e-01 1.81416655e+00
1.50207281e+00 4.87303317e-01 4.98008132e-02 3.81398052e-01
7.47319877e-01 -9.82768647e-03 -3.87660265e-01 1.00301839e-01
1.11675389e-01 -3.02573536e-02 2.39378929e-01 2.06176311e-01
-1.31485713e+00 1.02229559e+00 6.24661493e+00 7.23962665e-01
-8.85847509e-01 2.34081835e-01 4.34207648e-01 1.83092833e-01
-4.85957444e-01 3.95405233e-01 -8.67738366e-01 6.09024644e-01
1.32238007e+00 -3.92201304e-01 -1.26546726e-01 6.46052659e-01
4.98979986e-01 3.06574345e-01 -1.33802831e+00 5.85725546e-01
-4.63226587e-01 -1.31690061e+00 -4.13063355e-02 -2.70572811e-01
7.21303105e-01 -2.32718408e-01 -3.64016265e-01 3.01859200e-01
9.01930928e-01 -7.15852976e-01 8.81398022e-01 2.46043012e-01
5.94803095e-01 -5.50543725e-01 5.14174700e-01 2.94525295e-01
-1.17554784e+00 -1.43172920e-01 5.05826361e-02 -1.48225278e-01
4.83625799e-01 9.33083057e-01 -9.30641592e-01 8.41369450e-01
7.99865484e-01 1.20138252e+00 -1.46133468e-01 5.99239171e-01
-9.62877750e-01 1.69559419e+00 -2.17830122e-01 1.48227051e-01
1.09339664e-02 5.27111530e-01 4.87077862e-01 1.44359553e+00
7.23355711e-02 2.64308214e-01 2.09425479e-01 8.51266444e-01
-1.71965972e-01 -1.61747783e-01 -4.66399550e-01 -1.66481629e-03
7.64965832e-01 9.03094530e-01 -4.75610763e-01 -4.90125567e-01
-6.18475080e-01 7.19600201e-01 2.82927096e-01 3.56981814e-01
-1.32823861e+00 -1.09206051e-01 8.29898492e-02 -1.59837946e-01
-8.28434601e-02 2.74779592e-02 -2.87759274e-01 -1.33982027e+00
-1.35074615e-01 -7.12446034e-01 1.03361607e+00 -5.26243150e-01
-1.46292269e+00 2.69312292e-01 4.85776752e-01 -9.92027164e-01
-6.48748040e-01 -1.66144952e-01 -9.73994374e-01 5.05654037e-01
-1.59985006e+00 -1.26071680e+00 1.25188346e-03 5.66135883e-01
4.90917921e-01 3.61800641e-02 7.83123314e-01 4.87594008e-01
-8.26141536e-01 5.06088853e-01 -5.45511544e-01 3.72252047e-01
1.11816323e+00 -1.39871669e+00 4.11672771e-01 9.94629741e-01
1.31748587e-01 6.61322832e-01 6.45140588e-01 -9.10605490e-01
-1.04279041e+00 -1.38052571e+00 1.43574977e+00 -6.30127370e-01
1.12037683e+00 -1.40582800e-01 -9.84936237e-01 1.11663854e+00
1.53270900e-01 1.68079026e-02 6.86389625e-01 6.09630585e-01
-5.79937994e-01 -1.01302259e-01 -7.21174598e-01 5.27049005e-01
1.30437124e+00 -2.27046967e-01 -1.03500915e+00 2.74697304e-01
1.09354174e+00 -1.20544560e-01 -9.10329163e-01 6.93422735e-01
2.17545971e-01 -4.90371585e-01 8.66692245e-01 -9.41196084e-01
7.98726201e-01 -3.16835403e-01 9.42033008e-02 -1.24857771e+00
-1.53113827e-01 -6.62686288e-01 -3.39125335e-01 1.88333893e+00
7.55266070e-01 -5.70330381e-01 3.96333158e-01 5.55498481e-01
-8.87734592e-02 -1.64735347e-01 -6.79645836e-01 -7.24607289e-01
-8.67702663e-02 -8.98283124e-01 6.18181884e-01 1.46212351e+00
4.10076290e-01 6.96301222e-01 -2.38263533e-01 3.29881907e-01
6.50527596e-01 2.64738590e-01 4.40649420e-01 -1.09168768e+00
-2.59027183e-01 -1.63724303e-01 1.64461195e-01 -6.65004671e-01
5.17061889e-01 -7.46442020e-01 4.61408310e-02 -1.29041696e+00
3.60836864e-01 -8.35306048e-01 -6.32630229e-01 9.64585304e-01
-7.81866372e-01 -1.51361257e-01 -1.62840784e-01 3.03027630e-01
-4.43978012e-01 5.11903465e-01 7.16479421e-01 2.10003592e-02
-2.83998609e-01 -6.44176900e-02 -5.49888670e-01 1.06480610e+00
8.73134792e-01 -7.09297240e-01 -6.28031850e-01 -4.32128906e-01
3.36028934e-01 3.65893424e-01 6.69066727e-01 -5.90636194e-01
1.47229701e-01 -5.62247992e-01 7.60425553e-02 -5.13515949e-01
-4.69470441e-01 -4.36462671e-01 9.30868238e-02 3.01493287e-01
-7.69044518e-01 -1.81812242e-01 1.99903756e-01 8.66370857e-01
-5.51096141e-01 -1.02501452e-01 2.32774317e-01 4.86385748e-02
-8.02406609e-01 4.77823280e-02 -2.64622658e-01 3.50542009e-01
8.06411684e-01 6.13418579e-01 -3.22326273e-01 -1.95994079e-01
-5.46977222e-01 3.37182283e-01 -3.64164501e-01 5.83788693e-01
4.00698483e-01 -1.48931849e+00 -1.05213404e+00 -1.86977014e-01
2.61637926e-01 5.92631325e-02 -4.48291413e-02 7.62069464e-01
2.62551427e-01 4.61918116e-01 3.47163647e-01 -2.19634473e-01
-9.04763460e-01 5.70106208e-01 -4.19796705e-02 -5.61253428e-01
-6.68593764e-01 7.91649282e-01 5.09291589e-01 -2.41463438e-01
1.96052149e-01 -3.82728606e-01 -2.79522389e-01 9.26940218e-02
5.89814842e-01 2.04977363e-01 -1.32094666e-01 -5.05042747e-02
-5.08395851e-01 -1.34543031e-01 -2.51326319e-02 -2.82298297e-01
1.35561657e+00 6.56739995e-02 -1.49444446e-01 6.73939705e-01
7.79631376e-01 -3.68172079e-02 -1.28426623e+00 -4.10897911e-01
5.80874979e-01 -1.94590926e-01 -4.27163728e-02 -1.13265896e+00
-7.89059758e-01 7.53366053e-01 -6.48650229e-02 1.55690730e-01
1.14702094e+00 2.77499944e-01 7.92989969e-01 -7.22443461e-02
2.41073996e-01 -7.54363000e-01 6.31186739e-02 3.43772620e-01
1.07405293e+00 -1.03208160e+00 -4.34693158e-01 -7.22607672e-01
-4.42503422e-01 8.78875375e-01 5.91620862e-01 7.29861856e-02
6.24942183e-01 5.69123030e-01 -7.67391920e-02 -7.29926229e-02
-1.23564851e+00 3.86755466e-02 7.30485320e-02 3.37211818e-01
7.92235792e-01 1.98043570e-01 -4.95775998e-01 1.01411998e+00
6.63626194e-02 2.06756800e-01 2.53931046e-01 6.82784498e-01
-2.38799080e-02 -1.28358877e+00 -3.10259014e-01 2.73944348e-01
-7.30721056e-01 -2.85430789e-01 -2.83633709e-01 8.36420238e-01
1.81218963e-02 1.14023483e+00 -1.63484842e-01 -2.36649767e-01
4.06606436e-01 2.83469528e-01 2.11358324e-01 -5.49385965e-01
-3.69532943e-01 2.10536242e-01 5.56361914e-01 -5.72539687e-01
-8.02541077e-01 -9.78253543e-01 -1.65454376e+00 8.80460665e-02
-2.77879745e-01 3.12313028e-02 8.48481432e-02 1.10735536e+00
4.56853323e-02 8.25209439e-01 5.90586543e-01 -1.46864533e-01
-3.43137830e-01 -1.05067432e+00 -1.02573588e-01 5.06102800e-01
7.69516006e-02 -8.42476726e-01 -2.74767131e-01 6.40528917e-01]
|
[9.104774475097656, 9.104517936706543]
|
b3eb3d85-de02-4a35-81be-d0fc51ce7a9f
|
influence-of-initialization-on-the
|
2003.03789
| null |
https://arxiv.org/abs/2003.03789v1
|
https://arxiv.org/pdf/2003.03789v1.pdf
|
Influence of Initialization on the Performance of Metaheuristic Optimizers
|
All metaheuristic optimization algorithms require some initialization, and the initialization for such optimizers is usually carried out randomly. However, initialization can have some significant influence on the performance of such algorithms. This paper presents a systematic comparison of 22 different initialization methods on the convergence and accuracy of five optimizers: differential evolution (DE), particle swarm optimization (PSO), cuckoo search (CS), artificial bee colony (ABC) algorithm and genetic algorithm (GA). We have used 19 different test functions with different properties and modalities to compare the possible effects of initialization, population sizes and the numbers of iterations. Rigorous statistical ranking tests indicate that 43.37\% of the functions using the DE algorithm show significant differences for different initialization methods, while 73.68\% of the functions using both PSO and CS algorithms are significantly affected by different initialization methods. The simulations show that DE is less sensitive to initialization, while both PSO and CS are more sensitive to initialization. In addition, under the condition of the same maximum number of function evaluations (FEs), the population size can also have a strong effect. Particle swarm optimization usually requires a larger population, while the cuckoo search needs only a small population size. Differential evolution depends more heavily on the number of iterations, a relatively small population with more iterations can lead to better results. Furthermore, ABC is more sensitive to initialization, while such initialization has little effect on GA. Some probability distributions such as the beta distribution, exponential distribution and Rayleigh distribution can usually lead to better performance. The implications of this study and further research topics are also discussed in detail.
|
['Xin-She Yang', 'San-Yang Liu', 'Qian Li']
|
2020-03-08
| null | null | null | null |
['metaheuristic-optimization']
|
['methodology']
|
[-2.72666723e-01 -5.84650457e-01 3.73683199e-02 1.98920071e-01
3.08243811e-01 -5.27260661e-01 2.44695529e-01 1.88480228e-01
-6.99782968e-01 1.16504717e+00 -2.86892444e-01 -1.90731943e-01
-5.46200931e-01 -1.16703629e+00 -2.43173867e-01 -1.24947536e+00
-5.41289672e-02 2.97867239e-01 2.80828446e-01 -4.71495152e-01
7.33072579e-01 5.29003859e-01 -1.90847266e+00 -4.09766525e-01
1.17785656e+00 7.21232891e-01 3.06037396e-01 4.44891095e-01
-1.52787387e-01 9.14336368e-02 -1.41649938e+00 -5.73281460e-02
1.66114330e-01 -5.37182748e-01 -2.92422950e-01 -1.78179532e-01
-9.12397504e-01 4.16286767e-01 3.44496012e-01 1.12540114e+00
9.57803786e-01 3.25670242e-01 6.59172118e-01 -1.32023871e+00
-4.35970724e-01 5.24088979e-01 -6.54267967e-01 2.96270758e-01
3.25007468e-01 4.25374091e-01 3.36929500e-01 -2.63996720e-01
2.79229641e-01 9.59173858e-01 6.95573390e-01 1.77439153e-01
-8.83821845e-01 -8.01956654e-01 -3.09383869e-01 9.55180917e-03
-1.73094165e+00 1.75338596e-01 5.23761272e-01 1.51723757e-01
7.00988710e-01 7.49393523e-01 9.90916193e-01 4.51586038e-01
5.25112629e-01 1.85541704e-01 1.05379617e+00 -5.60429335e-01
4.86231685e-01 2.40041763e-01 -3.44989374e-02 2.39761323e-01
1.22296214e+00 3.15258712e-01 8.21211934e-02 -2.41946056e-01
4.53322142e-01 -2.34518796e-01 -5.92624903e-01 -1.52563723e-02
-8.76994848e-01 9.90240335e-01 2.10109308e-01 8.79586637e-01
-3.12924385e-01 -2.10029006e-01 2.31269583e-01 1.76948279e-01
-1.40373394e-01 1.24025726e+00 -5.11663139e-01 -3.64142895e-01
-6.49608731e-01 1.63082659e-01 1.07513559e+00 3.94834131e-01
4.02577698e-01 1.54102415e-01 -1.56877398e-01 8.60811770e-01
2.65118480e-01 6.94263399e-01 8.95080686e-01 -4.70722258e-01
2.12242216e-01 7.71102428e-01 4.14683461e-01 -1.26520586e+00
-6.69030488e-01 -6.56857908e-01 -5.85975230e-01 2.59497046e-01
4.07450378e-01 -5.03871202e-01 -5.71290553e-01 1.21656632e+00
1.91496894e-01 -2.71116704e-01 2.51945525e-01 7.31998444e-01
7.68080175e-01 9.91749465e-01 -2.54643500e-01 -7.56922543e-01
1.10098648e+00 -8.59219074e-01 -9.58086789e-01 1.04001738e-01
4.52712148e-01 -9.94939268e-01 9.52290893e-01 3.37542713e-01
-1.07904816e+00 -3.46400410e-01 -1.28945863e+00 9.00907934e-01
-6.22984588e-01 1.52450517e-01 5.06557822e-01 1.42048085e+00
-7.42173970e-01 5.71620822e-01 -7.57579863e-01 -3.40600908e-01
-1.33547902e-01 6.72270000e-01 1.37924597e-01 2.57953763e-01
-9.70819771e-01 9.72068548e-01 8.11093092e-01 1.67441294e-01
1.01827726e-01 -3.29722375e-01 -6.08271778e-01 1.75318539e-01
2.26473153e-01 -4.48446155e-01 4.16243076e-01 -8.72719705e-01
-1.72550881e+00 1.86464578e-01 -2.07130000e-01 -1.16771050e-01
2.07360759e-01 2.37835780e-01 -6.05587780e-01 -2.11000428e-01
-1.61013082e-01 1.29923984e-01 3.48755062e-01 -1.40059054e+00
-6.40210688e-01 -1.32255957e-01 -1.31293654e-01 2.92814881e-01
-2.83538371e-01 1.82838663e-01 -3.38134527e-01 -4.33726281e-01
-6.93640411e-02 -9.64607596e-01 -2.45608881e-01 -7.33511746e-01
-1.45035267e-01 -1.31561309e-01 5.96805274e-01 -3.21235240e-01
1.90043712e+00 -1.97017086e+00 1.21100552e-01 6.42569244e-01
-3.70238006e-01 4.54794466e-01 5.63737340e-02 3.93332630e-01
6.16240762e-02 5.21419346e-01 -2.04596758e-01 5.30549586e-01
-5.21394350e-02 2.58699387e-01 6.06445432e-01 3.39102060e-01
-2.02419192e-01 3.84665042e-01 -7.26643801e-01 -2.56443858e-01
1.79369152e-01 3.14461380e-01 -5.26589155e-01 -1.73675660e-02
2.98437327e-01 1.33196265e-01 -5.21858811e-01 8.02227259e-01
8.16965401e-01 -7.78537095e-02 -2.44697668e-02 -2.50366628e-01
-6.18432045e-01 -4.42820579e-01 -1.60662806e+00 6.14154339e-01
-3.19973916e-01 4.49107260e-01 1.97404046e-02 -9.57693756e-01
1.02932477e+00 2.50086188e-01 5.07728457e-01 -7.32886016e-01
6.47166491e-01 4.19633090e-01 6.88143730e-01 -3.66970897e-01
5.49676180e-01 2.55514055e-01 3.74257386e-01 4.89014417e-01
-5.19928575e-01 -3.37789059e-01 9.05651510e-01 -4.51950163e-01
6.36775017e-01 -3.96637380e-01 3.38439703e-01 -4.04080689e-01
7.43383169e-01 -3.73005122e-02 6.76166773e-01 6.18089795e-01
-2.36163195e-02 3.47092479e-01 3.38946790e-01 -2.31351018e-01
-7.68306077e-01 -3.74979496e-01 -5.13137162e-01 3.78486216e-01
7.99133122e-01 -3.75713676e-01 -6.08947873e-01 -1.86475143e-01
-9.88531858e-02 1.16085243e+00 -4.02159601e-01 -4.69139248e-01
-5.38275838e-01 -1.86766171e+00 2.85867065e-01 -1.72166228e-02
8.09670866e-01 -1.07232785e+00 -8.19757760e-01 1.16267100e-01
6.60941470e-03 -5.66057384e-01 1.84873492e-02 6.12991042e-02
-8.68889689e-01 -1.06535625e+00 -7.24981368e-01 -4.66846049e-01
8.43800724e-01 9.80627164e-02 9.25707400e-01 8.02345037e-01
-1.70174539e-01 2.49400079e-01 -7.49062419e-01 -6.04355931e-01
-5.71373284e-01 -2.26761792e-02 -4.58463235e-03 -3.27326834e-01
1.28651083e-01 -2.93770641e-01 -4.02403980e-01 8.72680545e-01
-8.64704788e-01 -6.01774395e-01 3.71480852e-01 9.91681755e-01
4.59335417e-01 1.27935123e+00 1.54972568e-01 -2.99706787e-01
1.05743754e+00 -4.23698574e-01 -6.81757748e-01 4.33535129e-01
-9.05329287e-01 1.33098945e-01 5.92918634e-01 -6.27399623e-01
-7.79534578e-01 -6.24750674e-01 1.49479508e-01 5.80461808e-02
6.09836802e-02 5.70207179e-01 -2.37282276e-01 -5.69207668e-01
6.22998059e-01 1.53583616e-01 2.04821542e-01 -2.25844249e-01
-6.23954415e-01 5.95782101e-01 -2.54421413e-01 -4.03531343e-01
8.73222589e-01 7.63508901e-02 5.51897027e-02 -8.57808769e-01
-1.39121087e-02 2.20116153e-02 -3.72459106e-02 -2.29919806e-01
7.31034338e-01 -3.76653373e-02 -8.02514791e-01 6.19536519e-01
-6.37932181e-01 -7.60137588e-02 4.31269519e-02 7.73772180e-01
1.47717446e-01 3.00780505e-01 -2.50408035e-02 -8.38667810e-01
-3.28665823e-01 -1.56823218e+00 2.17585400e-01 1.03755403e+00
-1.89210922e-01 -1.13174248e+00 -3.12389255e-01 1.26235113e-01
6.89734042e-01 3.56360644e-01 8.26402545e-01 -6.40421629e-01
-3.05920057e-02 -2.79325455e-01 3.32169890e-01 4.35324870e-02
4.90842074e-01 6.25534892e-01 -1.28961101e-01 -5.35938203e-01
3.15540373e-01 3.36953998e-01 6.94371909e-02 7.14233398e-01
9.03525054e-01 -1.27152652e-01 -6.19783103e-01 8.01956654e-01
1.61390674e+00 1.18591142e+00 9.27702367e-01 1.12269211e+00
1.87826335e-01 2.28188559e-01 9.84240055e-01 5.61863482e-01
-5.60552925e-02 4.26493585e-01 4.33189213e-01 -1.61809415e-01
2.79419452e-01 4.33135808e-01 3.34833980e-01 6.36691570e-01
-5.40193915e-01 -7.82493353e-01 -9.70273614e-01 1.02090776e-01
-1.21500349e+00 -8.59851837e-01 -2.21297503e-01 2.46731615e+00
5.61884522e-01 1.71883442e-02 6.72617033e-02 6.95008099e-01
1.08032537e+00 -2.60158092e-01 -3.46195966e-01 -6.21611536e-01
-4.08949614e-01 2.77704567e-01 7.70167768e-01 2.33509511e-01
-6.03751242e-01 3.77298236e-01 6.31209660e+00 9.05705452e-01
-1.45603371e+00 -4.59021538e-01 4.25156385e-01 -2.11598396e-01
-1.90443337e-01 -1.59569345e-02 -7.81045735e-01 1.15507591e+00
6.84689701e-01 -5.42772353e-01 4.93963867e-01 3.43287587e-01
1.87278569e-01 -8.15158904e-01 -2.73529053e-01 9.72078204e-01
-3.90416794e-02 -9.57990944e-01 -1.73191592e-01 2.78576642e-01
9.54989612e-01 -4.92572665e-01 -6.87991008e-02 1.18852839e-01
-2.11122241e-02 -1.14056647e+00 2.63957113e-01 2.97993332e-01
6.27879351e-02 -1.22496152e+00 1.47725320e+00 2.80733466e-01
-9.28336978e-01 -3.51023287e-01 -5.71707904e-01 -6.23962618e-02
1.21208511e-01 5.16949177e-01 -2.60974854e-01 7.56117582e-01
9.23754930e-01 -4.49608229e-02 -6.55232847e-01 1.76083016e+00
-2.30096579e-02 8.19506645e-01 -8.42753351e-01 -8.46091151e-01
7.82598630e-02 -5.88616312e-01 9.70886588e-01 4.83872265e-01
6.63896501e-01 2.07170904e-01 -1.65532082e-01 6.68923020e-01
7.40912735e-01 3.69929820e-01 2.01217607e-02 -2.59153873e-01
1.02419305e+00 8.66261899e-01 -1.29992187e+00 -3.86707783e-02
-2.05350250e-01 5.58222115e-01 -5.44112802e-01 4.33384210e-01
-1.12845933e+00 -7.84364522e-01 3.18771064e-01 -1.15564026e-01
4.50594753e-01 5.47946505e-02 -4.72683936e-01 -6.71281397e-01
-2.53478289e-01 -8.58834982e-01 4.19546902e-01 -6.04577303e-01
-8.80709469e-01 5.05711854e-01 1.23383731e-01 -1.23529899e+00
3.85083742e-02 -5.68489671e-01 -8.70639324e-01 8.17439854e-01
-1.02764976e+00 -1.12946987e-01 -3.48961353e-01 2.14361399e-01
1.03735052e-01 -4.77695525e-01 5.56845129e-01 1.85838088e-01
-1.08486199e+00 7.08463669e-01 4.63784963e-01 -3.65634203e-01
3.63491654e-01 -8.64714742e-01 -5.36730349e-01 7.11743653e-01
-4.08548772e-01 7.07430065e-01 1.06794727e+00 -5.69475234e-01
-1.38108110e+00 -2.45699629e-01 3.94323111e-01 1.40182450e-02
1.52271792e-01 4.25452679e-01 -6.00099683e-01 -8.34912211e-02
2.26148590e-01 -5.26856780e-01 7.28604972e-01 7.83054009e-02
8.50228906e-01 1.77016798e-02 -1.44521153e+00 7.44368255e-01
4.23791349e-01 5.56340218e-01 -1.92960292e-01 2.93532778e-02
2.61211574e-01 -4.85427350e-01 -1.12347472e+00 6.44301534e-01
3.91477466e-01 -1.14305484e+00 9.82308924e-01 1.25469819e-01
-4.59580384e-02 -6.49925590e-01 2.20902160e-01 -1.44384599e+00
-4.62841719e-01 -3.87876511e-01 3.55468988e-01 1.33509779e+00
6.26691043e-01 -1.28552926e+00 5.15736997e-01 8.33238661e-01
8.79037976e-02 -8.76721144e-01 -5.82681477e-01 -1.09904051e+00
-1.48877492e-02 1.50500819e-01 1.10376239e+00 9.51122582e-01
-2.57418901e-01 -5.04112951e-02 1.02076925e-01 2.30932906e-01
1.91960737e-01 8.79708230e-02 7.18587816e-01 -1.16729856e+00
-1.12819366e-01 -8.38579655e-01 -3.75458598e-01 -2.37787962e-02
-3.01546216e-01 -4.29017961e-01 -4.90852505e-01 -1.62219465e+00
-5.82215451e-02 -6.62316561e-01 -1.65003568e-01 1.53708786e-01
-4.57409739e-01 1.71288233e-02 1.45662770e-01 1.20526969e-01
2.39312276e-02 5.34751654e-01 1.36108279e+00 9.90328267e-02
-7.89508581e-01 3.37164521e-01 -4.72325951e-01 6.83360398e-01
1.01448011e+00 -5.15362740e-01 -3.59298885e-01 -1.23438388e-01
2.98189670e-01 -9.90426764e-02 -3.34938943e-01 -1.04223776e+00
1.75916880e-01 -3.03421825e-01 5.35407901e-01 -3.14480692e-01
-1.04630262e-01 -9.62051749e-01 5.77154100e-01 8.93677592e-01
4.97719616e-01 5.49503267e-01 5.48764884e-01 -2.06653811e-02
-2.81514078e-01 -9.18315053e-01 7.23785877e-01 1.20956726e-01
-3.87054741e-01 -3.27071577e-01 -6.01449072e-01 -1.39581159e-01
1.26419818e+00 -1.14050090e+00 1.24848792e-02 -3.41763318e-01
-4.42514360e-01 3.68145436e-01 8.05615902e-01 1.47064045e-01
9.55101252e-02 -1.05524755e+00 -4.29024786e-01 1.52349740e-01
-4.72816318e-01 -3.23413163e-01 -9.80774686e-02 1.02629960e+00
-9.99351561e-01 1.80449605e-01 -3.41474771e-01 -3.12594026e-01
-1.38947284e+00 2.67172813e-01 4.78113413e-01 -7.68361092e-02
1.05136536e-01 9.34503138e-01 -4.47423846e-01 -2.80527830e-01
9.38530415e-02 -1.58791155e-01 -7.14426577e-01 1.89996466e-01
2.33092889e-01 1.02529097e+00 -9.11119580e-02 -5.99298239e-01
-4.78569537e-01 8.53478849e-01 2.54659444e-01 6.88931122e-02
1.16176963e+00 8.05011541e-02 -4.50315595e-01 1.78526063e-02
7.67156065e-01 4.74974066e-01 -4.69830960e-01 7.16442883e-01
-3.26415777e-01 -6.78232312e-01 9.26978588e-02 -1.08180845e+00
-1.23179352e+00 2.01444805e-01 7.13748038e-01 4.62947398e-01
1.52178895e+00 -6.17019355e-01 5.06795347e-01 1.62570760e-01
5.88898063e-01 -1.20257688e+00 -2.66127527e-01 5.57874680e-01
6.89285457e-01 -1.02510190e+00 3.38686675e-01 -4.02290195e-01
-5.37000954e-01 1.24548042e+00 7.14683652e-01 1.54676050e-01
6.36107922e-01 4.41633463e-01 -1.99385226e-01 -7.68714622e-02
-3.57996225e-01 -5.36533482e-02 2.37769276e-01 3.77839446e-01
4.14293766e-01 -2.04133734e-01 -1.30080926e+00 7.23586857e-01
-6.37671828e-01 -2.96060652e-01 5.55214822e-01 1.00288415e+00
-3.97885799e-01 -1.26944494e+00 -1.05646956e+00 3.07256192e-01
-4.52350497e-01 7.56650120e-02 -3.38855177e-01 1.10967219e+00
3.08093399e-01 1.14590549e+00 1.00799337e-01 -3.15663397e-01
1.77254111e-01 -3.61270398e-01 4.37780261e-01 -4.23773192e-02
-9.77624655e-01 -4.01312001e-02 1.57834943e-02 -2.54352331e-01
-4.01944369e-01 -7.38994420e-01 -1.51016581e+00 -6.35316372e-01
-1.06449175e+00 9.35822010e-01 7.24356472e-01 7.81643927e-01
1.27184346e-01 5.37032962e-01 5.07509053e-01 -6.92394674e-01
-8.57901871e-02 -8.11303735e-01 -5.37198842e-01 -1.85490102e-02
-4.05873567e-01 -1.05533969e+00 -9.87451375e-01 -5.95181048e-01]
|
[5.683096885681152, 3.5018553733825684]
|
952b1fe0-5978-43c2-af8f-57a7675b90e6
|
unsupervised-ehr-based-phenotyping-via-matrix
|
2209.00322
| null |
https://arxiv.org/abs/2209.00322v1
|
https://arxiv.org/pdf/2209.00322v1.pdf
|
Unsupervised EHR-based Phenotyping via Matrix and Tensor Decompositions
|
Computational phenotyping allows for unsupervised discovery of subgroups of patients as well as corresponding co-occurring medical conditions from electronic health records (EHR). Typically, EHR data contains demographic information, diagnoses and laboratory results. Discovering (novel) phenotypes has the potential to be of prognostic and therapeutic value. Providing medical practitioners with transparent and interpretable results is an important requirement and an essential part for advancing precision medicine. Low-rank data approximation methods such as matrix (e.g., non-negative matrix factorization) and tensor decompositions (e.g., CANDECOMP/PARAFAC) have demonstrated that they can provide such transparent and interpretable insights. Recent developments have adapted low-rank data approximation methods by incorporating different constraints and regularizations that facilitate interpretability further. In addition, they offer solutions for common challenges within EHR data such as high dimensionality, data sparsity and incompleteness. Especially extracting temporal phenotypes from longitudinal EHR has received much attention in recent years. In this paper, we provide a comprehensive review of low-rank approximation-based approaches for computational phenotyping. The existing literature is categorized into temporal vs. static phenotyping approaches based on matrix vs. tensor decompositions. Furthermore, we outline different approaches for the validation of phenotypes, i.e., the assessment of clinical significance.
|
['Evrim Acar', 'Age K. Smilde', 'Florian Becker']
|
2022-09-01
| null | null | null | null |
['computational-phenotyping']
|
['medical']
|
[ 1.82483788e-03 -3.36421579e-01 -2.93087602e-01 -3.90622139e-01
-4.39817697e-01 -5.88913798e-01 -2.03980491e-01 4.53486443e-01
1.68503806e-01 6.96530819e-01 4.76552337e-01 -2.12951243e-01
-8.00077915e-01 -3.99599433e-01 -1.05246007e-01 -7.90406585e-01
-5.06942987e-01 6.67293847e-01 -7.89385438e-01 2.03508094e-01
-3.60718727e-01 5.04639328e-01 -1.24010003e+00 6.02081299e-01
1.13128686e+00 8.12393904e-01 -1.52772412e-01 5.10576487e-01
1.10342704e-01 6.82110131e-01 -2.30182871e-01 -2.80297428e-01
9.80030671e-02 -9.15764272e-02 -4.83248293e-01 1.19286381e-01
-5.70592172e-02 -1.75945193e-01 -3.55816811e-01 8.11067939e-01
4.05549824e-01 -2.63604909e-01 4.49042737e-01 -1.28890276e+00
-7.13590145e-01 4.45718914e-01 -5.16137123e-01 -2.77589099e-03
5.38453221e-01 8.04866627e-02 8.31627190e-01 -8.84858429e-01
5.55426717e-01 1.03805411e+00 8.96650314e-01 3.23952317e-01
-1.67441261e+00 -3.22174788e-01 -1.86651964e-02 2.43624792e-01
-1.58423519e+00 -4.01537389e-01 5.16385376e-01 -9.56682205e-01
5.57965159e-01 9.91353214e-01 7.72714913e-01 9.22123492e-01
3.55866961e-02 7.77570546e-01 1.10832274e+00 2.09234029e-01
7.05378726e-02 -2.44705454e-01 4.10452694e-01 4.45845902e-01
6.94321036e-01 4.05730084e-02 -4.88209009e-01 -1.15106893e+00
4.74008650e-01 7.52615392e-01 -3.29254180e-01 -2.12153047e-01
-1.68762791e+00 4.29116696e-01 -2.39690632e-01 1.59191549e-01
-1.00448310e+00 -2.45513856e-01 5.35628736e-01 2.06033349e-01
3.17693084e-01 4.28106397e-01 -6.85372770e-01 -1.72456861e-01
-9.34656680e-01 5.37400320e-02 5.90486825e-01 7.14362204e-01
3.84940863e-01 1.25516504e-01 -4.01593208e-01 7.88867652e-01
1.78879201e-01 6.80027187e-01 3.70703876e-01 -8.72317255e-01
1.52058810e-01 9.53323007e-01 1.34779990e-01 -1.29182041e+00
-8.41146052e-01 -4.58422661e-01 -1.57549870e+00 -7.58550346e-01
1.50345609e-01 -1.44241229e-01 -5.71635187e-01 1.58438206e+00
5.98147511e-01 4.00512427e-01 5.71997836e-02 7.92250395e-01
7.26348341e-01 1.30772248e-01 1.20251834e-01 -6.53975010e-01
1.67958117e+00 -1.68870315e-01 -1.04437983e+00 6.29416525e-01
6.53544664e-01 -6.15802586e-01 6.69623435e-01 6.67119980e-01
-7.96514094e-01 -5.91727979e-02 -3.76624644e-01 1.51398465e-01
-1.43728498e-02 5.96909463e-01 1.15012884e+00 5.28721392e-01
-7.42198348e-01 6.52390599e-01 -1.28686917e+00 -3.88270319e-01
2.84390986e-01 3.58854800e-01 -5.77619255e-01 -3.08115900e-01
-8.15366447e-01 1.81907583e-02 3.43135819e-02 2.25161090e-01
-6.04318500e-01 -1.15452969e+00 -6.61754906e-01 -4.74135242e-02
4.94716018e-02 -1.25662208e+00 5.87771654e-01 -1.45575926e-01
-9.95427072e-01 5.93957901e-01 -5.45437753e-01 -1.46259904e-01
7.25481585e-02 -1.13644466e-01 -7.75833011e-01 2.62854517e-01
1.71930835e-01 -1.16371460e-01 5.86604953e-01 -7.04823315e-01
-3.68675411e-01 -7.86519229e-01 -5.39985657e-01 -2.21371993e-01
-5.09432614e-01 1.21870898e-01 -2.17873380e-01 -8.20766389e-01
5.54983854e-01 -1.05646992e+00 -5.22748649e-01 -2.80717582e-01
-6.77728593e-01 -8.71427506e-02 4.89360183e-01 -1.07205033e+00
1.77951241e+00 -2.13502002e+00 2.67693609e-01 1.73698664e-01
9.48412478e-01 -1.68521807e-01 1.59931436e-01 6.18455112e-01
-5.11867821e-01 2.05475554e-01 -6.09300882e-02 -1.85147002e-01
-4.09208208e-01 1.32290125e-01 -1.52183503e-01 6.90757751e-01
1.79978237e-01 8.31257820e-01 -1.01474547e+00 -1.45004511e-01
-7.61136711e-02 5.39506435e-01 -6.61677003e-01 1.73856214e-01
2.44209886e-01 9.28250194e-01 -6.59630775e-01 1.02935338e+00
7.17752993e-01 -8.53249609e-01 6.60222113e-01 -4.99029070e-01
-1.53754100e-01 -1.10558555e-01 -1.13346839e+00 1.34170091e+00
2.13406488e-01 6.53005391e-02 1.33884653e-01 -9.15309370e-01
6.19630575e-01 7.88085639e-01 1.36786592e+00 1.20477311e-01
-2.13464826e-01 1.45212933e-01 1.92370892e-01 -8.43669951e-01
9.77210179e-02 -6.94051608e-02 2.45766580e-01 3.55656654e-01
-5.39283395e-01 8.26967120e-01 2.74943948e-01 2.39159331e-01
1.43766713e+00 -2.98746705e-01 4.64641392e-01 -2.21068725e-01
4.99147266e-01 1.55605406e-01 1.15266669e+00 2.57399529e-01
-1.24237814e-03 4.61081654e-01 5.13119519e-01 -7.38641560e-01
-9.10399556e-01 -9.70748484e-01 -5.20665050e-01 5.00722110e-01
-5.16112983e-01 -8.53618681e-01 -2.01592222e-02 -2.38462299e-01
5.88828444e-01 2.44875289e-02 -7.89398432e-01 -3.73343043e-02
-2.65241176e-01 -1.39178729e+00 5.72764933e-01 4.88116145e-01
-1.03143886e-01 -2.53723025e-01 -1.38925880e-01 2.90517718e-01
-5.18422127e-01 -1.08134675e+00 -1.96023375e-01 -2.38819510e-01
-1.42129517e+00 -1.38657153e+00 -5.91716886e-01 -1.78290710e-01
1.01648355e+00 3.69771808e-01 9.94992793e-01 -3.41817439e-02
-5.88315547e-01 6.55968547e-01 -3.67005676e-01 -1.33295909e-01
-2.47400794e-02 -4.22148675e-01 7.71003008e-01 6.11953616e-01
4.97050136e-01 -7.10094035e-01 -9.49589610e-01 3.85724515e-01
-8.77252162e-01 -1.64171439e-02 7.12239563e-01 1.01165593e+00
1.07235253e+00 5.47549091e-02 5.21901309e-01 -1.08987379e+00
8.04884613e-01 -9.28215504e-01 -3.02360862e-01 3.88395697e-01
-8.99080575e-01 -4.23909314e-02 8.05976391e-01 -4.54895228e-01
-7.50828326e-01 9.59935263e-02 3.52994889e-01 -7.95606554e-01
-1.13779463e-01 1.08586073e+00 1.02034554e-01 2.07611024e-01
5.08604944e-01 2.60209203e-01 1.60601407e-01 -9.38987434e-01
1.12214781e-01 5.43056726e-01 4.53882106e-03 -6.88583255e-01
4.77651030e-01 8.31834614e-01 4.46671575e-01 -8.68134022e-01
-4.88581747e-01 -9.10331845e-01 -4.71181422e-01 8.90112519e-02
7.02501059e-01 -9.70217526e-01 -1.18873715e+00 1.90748662e-01
-7.00899243e-01 2.76778698e-01 -2.51181394e-01 8.95723403e-01
-2.33735368e-01 7.19690323e-01 -7.90099621e-01 -7.64312387e-01
-4.49780464e-01 -9.40765381e-01 9.04702544e-01 -3.98313314e-01
-4.74271864e-01 -8.62634838e-01 2.95620978e-01 4.64334816e-01
2.59503722e-01 8.41502488e-01 1.19639373e+00 -4.68592942e-01
-3.34003776e-01 -2.97136575e-01 -1.63802549e-01 3.06179374e-02
3.99146557e-01 1.12793900e-01 -6.59863532e-01 -4.16024894e-01
5.31701818e-02 1.03091843e-01 3.93538862e-01 7.03973949e-01
1.21979702e+00 -5.34174562e-01 -6.08037174e-01 8.79825354e-01
1.05118108e+00 9.60999206e-02 3.29976946e-01 -3.88925135e-01
9.60407913e-01 7.07337737e-01 5.92192829e-01 1.09497917e+00
5.54654181e-01 5.44632971e-01 -1.92110077e-01 -1.12336673e-01
2.14478239e-01 4.50796038e-02 -6.83822930e-02 1.15535724e+00
-3.11194420e-01 3.07501107e-01 -1.09134293e+00 3.41721565e-01
-2.11441898e+00 -7.55208373e-01 -8.62341166e-01 2.30208278e+00
6.40486002e-01 -8.64100754e-01 3.40648651e-01 1.34169862e-01
6.00975275e-01 -4.74545479e-01 -8.08898091e-01 9.39896777e-02
-2.92046845e-01 -2.92258441e-01 2.16197535e-01 -5.88439703e-02
-8.78641129e-01 3.03735852e-01 6.46654892e+00 2.18564466e-01
-9.03719842e-01 1.25886992e-01 6.19471729e-01 -1.70346275e-01
-3.95503044e-01 2.20462633e-03 -3.16205144e-01 4.19508427e-01
8.81470919e-01 -1.21469013e-01 6.20648324e-01 6.60458624e-01
8.26430559e-01 2.05055296e-01 -1.25076377e+00 1.47586656e+00
-2.49536067e-01 -1.23743355e+00 -4.66427170e-02 5.20829201e-01
7.91418672e-01 -7.57931024e-02 1.70474708e-01 -9.64925736e-02
1.00962423e-01 -8.22663844e-01 -6.16854951e-02 9.38560128e-01
1.22666419e+00 -3.87874216e-01 6.27439737e-01 -2.25206129e-02
-1.25141668e+00 -2.27459565e-01 -3.76439840e-01 -6.03355058e-02
2.34042376e-01 1.49096656e+00 -8.79186809e-01 9.02002215e-01
7.24887669e-01 1.17615855e+00 -2.72269875e-01 1.00961447e+00
2.24495113e-01 7.40056813e-01 -2.26746053e-01 5.09502411e-01
-4.19735312e-01 -6.58994257e-01 6.47362947e-01 8.54753315e-01
4.17354137e-01 4.72573727e-01 3.54543418e-01 7.24361837e-01
3.74476373e-01 3.31934363e-01 -5.05752563e-01 -6.46521151e-01
2.76608109e-01 1.24943817e+00 -3.46490949e-01 -4.04173225e-01
-5.49154282e-01 6.61978066e-01 1.25478595e-01 5.52097857e-01
-4.16942865e-01 1.42712191e-01 1.23923922e+00 2.65614063e-01
-2.67190725e-01 -2.81784266e-01 -4.97108102e-01 -1.69015121e+00
6.20172136e-02 -1.28741169e+00 9.96923387e-01 -2.48650178e-01
-1.52513492e+00 3.07823896e-01 -1.78715453e-01 -1.44127977e+00
-1.05540782e-01 -4.76418585e-01 5.61167784e-02 8.96074355e-01
-1.08617413e+00 -8.54931414e-01 -2.73688614e-01 8.07818592e-01
-2.01562271e-01 -1.48055702e-01 1.16123331e+00 6.51476204e-01
-1.15743029e+00 3.78625304e-01 4.38912928e-01 3.11874971e-02
6.43403769e-01 -1.11762905e+00 -1.53863937e-01 6.46519125e-01
-4.07529205e-01 1.38078201e+00 5.02125025e-01 -9.95921910e-01
-2.06767154e+00 -1.23367596e+00 8.84081900e-01 -5.87011039e-01
6.53855860e-01 -1.66260861e-02 -8.26690912e-01 6.66418314e-01
-5.55919111e-01 3.37353610e-02 1.68575716e+00 6.08686924e-01
-3.17679673e-01 -3.73249739e-01 -1.22106552e+00 5.97446263e-01
9.63245213e-01 -3.70184332e-01 -5.04144914e-02 6.07952654e-01
3.24014425e-01 -1.36111360e-02 -1.60094631e+00 6.80995822e-01
7.60741174e-01 -6.26488090e-01 1.00211406e+00 -1.25739777e+00
-6.02753125e-02 -5.84536135e-01 -9.31317583e-02 -9.06332552e-01
-9.40488517e-01 -9.08183336e-01 -6.92211270e-01 8.01388323e-01
1.42015010e-01 -7.24890530e-01 8.48415375e-01 1.04178631e+00
8.29811487e-03 -9.38570201e-01 -6.00022674e-01 -7.41644561e-01
-6.67103827e-01 -3.79717529e-01 7.54776537e-01 1.40349567e+00
2.34998703e-01 1.67254671e-01 -6.51845157e-01 4.45896447e-01
7.72865653e-01 4.29612637e-01 7.70600677e-01 -1.67080128e+00
-4.11309242e-01 -1.40893519e-01 -6.12379849e-01 -5.74529648e-01
-2.83007681e-01 -1.04192483e+00 -8.51808846e-01 -1.39964569e+00
5.62560856e-01 -6.28946066e-01 -4.14174289e-01 5.72340488e-01
-3.26526046e-01 -8.59297663e-02 -3.28917921e-01 6.42015159e-01
-3.42315584e-01 2.68542677e-01 1.10327017e+00 -5.58545180e-02
-4.19406474e-01 3.01765986e-02 -9.18819606e-01 4.32475239e-01
8.86963964e-01 -4.94129121e-01 -5.10180712e-01 -1.94829434e-01
6.08138919e-01 4.19035107e-01 1.34249225e-01 -4.56401855e-01
6.00463487e-02 -3.85274589e-01 3.96411479e-01 -5.78859091e-01
2.64396012e-01 -7.70282805e-01 9.42311287e-01 7.32821763e-01
4.01995964e-02 5.26687980e-01 -1.48917690e-01 9.08652842e-01
-2.45406345e-01 4.56003815e-01 1.42635196e-01 -3.26717831e-02
-9.64097232e-02 5.43388486e-01 -3.54482889e-01 -5.07655740e-02
9.20889497e-01 -3.19199376e-02 -1.38814092e-01 -2.66636193e-01
-1.19182777e+00 3.91394049e-01 1.76619217e-01 1.10519215e-01
8.47284079e-01 -1.38495910e+00 -1.01617539e+00 2.35927701e-01
4.07184839e-01 -3.24047446e-01 8.61822546e-01 1.79378033e+00
-2.76135087e-01 6.91998661e-01 1.28880575e-01 -7.83059716e-01
-1.34845746e+00 1.03209305e+00 -1.73795044e-01 -4.44646388e-01
-6.15151882e-01 3.39935452e-01 3.72546852e-01 -3.01049232e-01
-4.90331277e-02 -4.93298799e-01 -2.05565721e-01 7.93264583e-02
7.50976145e-01 7.62385786e-01 8.78965855e-03 -4.80900109e-01
-5.33174038e-01 2.70927250e-01 2.12342609e-02 4.09981251e-01
1.48659849e+00 -3.53893042e-02 -8.47841263e-01 4.54419166e-01
1.04793072e+00 1.14528686e-01 -5.51764488e-01 -6.78050518e-02
1.25107393e-01 -4.31109607e-01 -2.21500486e-01 -6.74950719e-01
-9.43040669e-01 4.47475553e-01 3.54700178e-01 2.40167215e-01
1.27101576e+00 -1.42066389e-01 6.40442848e-01 3.30263346e-01
3.99718881e-01 -5.94603181e-01 -4.31032091e-01 9.20927525e-02
7.97790527e-01 -9.46586013e-01 2.03515515e-01 -6.35807097e-01
-5.64651608e-01 1.01480699e+00 2.75451001e-02 3.31362039e-01
8.92658412e-01 -2.61297841e-02 4.05942975e-03 -5.22457480e-01
-9.60276425e-01 1.06623694e-01 5.07040322e-01 6.54608369e-01
7.46742845e-01 7.25840211e-01 -6.52251780e-01 9.67516303e-01
-8.57187808e-03 6.13381751e-02 2.26378992e-01 7.04148293e-01
3.79386127e-01 -1.21373129e+00 -6.86174154e-01 1.14699078e+00
-5.96571565e-01 -2.51687527e-01 -2.93251634e-01 1.17995791e-01
6.13002554e-02 9.91713047e-01 -4.31298882e-01 -5.37846148e-01
2.82630533e-01 1.43911511e-01 1.49841219e-01 -6.76897824e-01
-3.67077917e-01 2.67459869e-01 9.11717191e-02 -8.34516108e-01
-2.77127296e-01 -1.12429833e+00 -8.98101509e-01 -2.55113423e-01
8.49489644e-02 2.77680665e-01 3.90238792e-01 4.48642254e-01
1.10583842e+00 5.04633367e-01 5.61471045e-01 -5.41998707e-02
-5.38041711e-01 -7.05708683e-01 -8.68395150e-01 6.63123488e-01
2.54352421e-01 -5.66916585e-01 -1.69992030e-01 1.63671792e-01]
|
[6.472021102905273, 5.8940229415893555]
|
d42c02d9-2307-43fa-8420-ed638f77398e
|
prompting-electra-few-shot-learning-with
|
2205.15223
| null |
https://arxiv.org/abs/2205.15223v3
|
https://arxiv.org/pdf/2205.15223v3.pdf
|
Prompting ELECTRA: Few-Shot Learning with Discriminative Pre-Trained Models
|
Pre-trained masked language models successfully perform few-shot learning by formulating downstream tasks as text infilling. However, as a strong alternative in full-shot settings, discriminative pre-trained models like ELECTRA do not fit into the paradigm. In this work, we adapt prompt-based few-shot learning to ELECTRA and show that it outperforms masked language models in a wide range of tasks. ELECTRA is pre-trained to distinguish if a token is generated or original. We naturally extend that to prompt-based few-shot learning by training to score the originality of the target options without introducing new parameters. Our method can be easily adapted to tasks involving multi-token predictions without extra computation overhead. Analysis shows that ELECTRA learns distributions that align better with downstream tasks.
|
['Ves Stoyanov', 'Danqi Chen', 'Jingfei Du', 'Mikel Artetxe', 'Mengzhou Xia']
|
2022-05-30
| null | null | null | null |
['text-infilling']
|
['natural-language-processing']
|
[ 2.54309654e-01 1.32742912e-01 -2.71316320e-01 -3.33388478e-01
-1.08627105e+00 -3.63449007e-01 9.50589240e-01 3.07676882e-01
-7.02287436e-01 5.26038527e-01 5.84945023e-01 -5.10877192e-01
2.35281020e-01 -7.92863071e-01 -4.96815056e-01 -4.83363479e-01
1.13167368e-01 5.06970644e-01 6.02745354e-01 -3.06971610e-01
1.28458947e-01 -6.63963780e-02 -1.40329993e+00 6.42722845e-01
4.83800620e-01 2.92656004e-01 4.15063202e-01 7.05806792e-01
-4.29953456e-01 8.05237889e-01 -4.47698802e-01 -3.12220424e-01
1.62297994e-01 -6.81141675e-01 -6.92715228e-01 -1.64772198e-01
8.76960009e-02 -2.79113591e-01 -2.26390690e-01 6.54958069e-01
7.04488337e-01 4.78340060e-01 8.44321728e-01 -8.64701867e-01
-6.66325986e-01 1.13839281e+00 -2.76388824e-01 5.03602386e-01
1.30269974e-01 4.06433821e-01 1.34346545e+00 -1.50848889e+00
7.27938354e-01 1.21384192e+00 8.75656128e-01 8.98238957e-01
-1.51998031e+00 -3.28882873e-01 1.16438910e-01 1.97004706e-01
-1.04922307e+00 -8.92615020e-01 3.45376104e-01 -4.82022434e-01
1.51442277e+00 -1.04428701e-01 1.16504572e-01 1.41302097e+00
-1.04184272e-02 9.28101361e-01 9.23939109e-01 -8.06664646e-01
6.03691816e-01 5.09973839e-02 4.63575691e-01 3.75626147e-01
5.03915846e-02 -3.21428850e-02 -8.02070320e-01 -1.25487074e-01
3.63857038e-02 4.80407961e-02 1.22040786e-01 -1.14573687e-01
-1.08064139e+00 1.00964022e+00 -1.69542044e-01 4.72606212e-01
-2.41938069e-01 1.87218025e-01 5.49692154e-01 3.71657431e-01
7.27629840e-01 4.57989454e-01 -3.84642988e-01 -4.68226910e-01
-1.36525857e+00 1.20785885e-01 5.08128524e-01 1.02807605e+00
9.56261456e-01 2.60051548e-01 -9.34730053e-01 8.12150598e-01
1.27152726e-01 -4.64534350e-02 9.90487516e-01 -5.89431226e-01
3.05420101e-01 -2.49009654e-02 2.37751588e-01 -8.08401965e-03
-1.53649464e-01 2.70203948e-02 -2.33148605e-01 2.82905489e-01
5.24087727e-01 -3.62003088e-01 -1.25541592e+00 1.71856630e+00
-1.28141791e-02 2.33153179e-01 1.06650390e-01 4.81612712e-01
3.09995621e-01 8.58005762e-01 3.60736281e-01 -2.19289199e-01
1.22069275e+00 -1.02182877e+00 -6.73218250e-01 -7.19880044e-01
1.06952333e+00 -7.05792725e-01 1.49802577e+00 2.33208369e-02
-9.98321414e-01 -5.18627822e-01 -9.62056816e-01 -2.22116053e-01
-4.42238957e-01 -2.61764407e-01 4.52308118e-01 6.41070962e-01
-1.09052503e+00 8.59690964e-01 -7.06997454e-01 -6.16152525e-01
3.91790122e-01 -7.29651563e-03 6.86288774e-02 -4.00053849e-03
-1.40774107e+00 1.05448043e+00 4.62330818e-01 -5.35492301e-01
-1.02046907e+00 -7.99265444e-01 -9.88121510e-01 4.02023166e-01
5.12731016e-01 -4.67157304e-01 1.83976996e+00 -5.67302883e-01
-1.51975274e+00 8.38387609e-01 -4.81776029e-01 -7.01161623e-01
4.63895917e-01 -1.41642407e-01 -7.92334080e-02 -2.40722224e-01
2.30620831e-01 6.08670592e-01 9.42514956e-01 -7.97485471e-01
-5.01881540e-01 -8.33827779e-02 -1.21054202e-01 -9.60780010e-02
-4.64643985e-01 3.49387646e-01 -1.48667052e-01 -6.47281528e-01
-5.04625559e-01 -4.25138742e-01 -2.84980953e-01 -5.07533140e-02
-1.02332853e-01 -4.29824680e-01 6.08875215e-01 -1.95272446e-01
1.36506426e+00 -2.12321329e+00 -2.26927504e-01 -2.74663836e-01
-1.25130760e-02 5.33694685e-01 -3.92028511e-01 8.61006439e-01
2.12965682e-02 1.21333994e-01 -2.53914744e-01 -8.52974713e-01
2.77618378e-01 2.41128251e-01 -7.12372184e-01 3.41877699e-01
3.03617686e-01 1.04147398e+00 -1.20276570e+00 -5.74390948e-01
1.95560813e-01 1.03040203e-01 -5.07870197e-01 1.96388066e-01
-4.50738490e-01 -1.64412007e-01 4.62550484e-02 4.31101352e-01
2.18560204e-01 -8.15234631e-02 6.18992932e-02 6.21335089e-01
-1.38878524e-01 5.03511727e-01 -9.06345129e-01 1.69235110e+00
-6.10796094e-01 8.32735300e-01 -1.65579006e-01 -9.13622677e-01
7.18286455e-01 4.33447301e-01 4.91780415e-02 -4.75355893e-01
4.09310535e-02 3.55119631e-02 1.40556619e-01 -4.34438974e-01
6.11830056e-01 -7.18573928e-01 -2.51138240e-01 7.44890630e-01
4.57655609e-01 1.91887289e-01 2.68514305e-01 4.23711032e-01
1.36487663e+00 1.60283506e-01 6.90798998e-01 -5.70021160e-02
-1.79500937e-01 -1.09619454e-01 5.34090579e-01 1.23478568e+00
-3.59651387e-01 6.28681481e-01 5.13241768e-01 4.85105552e-02
-1.15539110e+00 -1.15793538e+00 2.62818690e-02 1.80052781e+00
-3.97481263e-01 -7.07114816e-01 -5.34005761e-01 -6.82764053e-01
-2.43800413e-02 1.23244119e+00 -6.09026194e-01 -2.67104566e-01
-3.13663274e-01 -4.97652799e-01 4.68090385e-01 7.79192269e-01
-3.27469885e-01 -1.16109800e+00 -5.59569418e-01 3.96297008e-01
1.75517306e-01 -8.25876057e-01 -6.78636789e-01 7.23565280e-01
-6.41247630e-01 -5.05194962e-01 -9.28544462e-01 -8.84662271e-01
1.91187769e-01 4.09608871e-01 9.42159832e-01 -1.86430171e-01
-2.70938933e-01 7.27022514e-02 -5.29842854e-01 -4.57673997e-01
-6.18733108e-01 1.16491029e-02 6.66486323e-02 -1.69908240e-01
7.44737208e-01 -4.97283548e-01 -1.95867106e-01 8.66288096e-02
-6.37948990e-01 -1.05266511e-01 5.09686232e-01 1.06070852e+00
1.46527201e-01 -3.95218819e-01 6.88233316e-01 -1.12621641e+00
7.47233748e-01 -7.58433223e-01 -1.87951177e-01 2.56836653e-01
-6.52553439e-01 2.49614954e-01 6.95400894e-01 -7.33761072e-01
-1.07613301e+00 -3.22581716e-02 -1.56448781e-01 -5.51753104e-01
-2.37609938e-01 4.23482865e-01 6.21393770e-02 5.97219110e-01
8.84778559e-01 1.82808235e-01 -2.54557729e-01 -6.42935932e-01
6.33107245e-01 7.62819052e-01 2.17578530e-01 -3.49935085e-01
7.16588140e-01 2.68290013e-01 -4.58841980e-01 -9.23612595e-01
-9.83952880e-01 -8.99017632e-01 -5.76036096e-01 -1.14183631e-02
8.67857814e-01 -9.21888888e-01 -1.84813902e-01 3.31806809e-01
-1.20827222e+00 -8.89264464e-01 -7.02245414e-01 4.41569358e-01
-5.95063210e-01 3.77113581e-01 -7.20555007e-01 -9.92026508e-01
-3.07291001e-01 -7.07205832e-01 9.44117486e-01 -8.29125643e-02
-6.09811306e-01 -1.01502275e+00 2.47943491e-01 -2.10947648e-01
5.08776248e-01 -5.81759751e-01 1.01572621e+00 -1.22903991e+00
-1.67128786e-01 -1.28504127e-01 8.04457515e-02 -6.18615858e-02
-9.12073650e-04 -2.15499207e-01 -1.22567022e+00 -1.22208290e-01
-7.78864473e-02 -5.47963202e-01 1.40477777e+00 3.03681731e-01
7.63166010e-01 -1.57183394e-01 -3.43481690e-01 3.17699969e-01
1.14047611e+00 -3.22038047e-02 4.32674259e-01 8.84324908e-02
3.13694358e-01 6.12807930e-01 6.53307736e-01 5.87180793e-01
-4.50844504e-02 6.63177669e-01 -6.77527264e-02 4.17418659e-01
-2.67877489e-01 -6.32850528e-01 6.88887656e-01 5.94398379e-01
3.06595355e-01 -4.35722917e-01 -1.00351048e+00 8.49832833e-01
-1.87840557e+00 -1.54674113e+00 7.46310428e-02 1.90484834e+00
1.12158811e+00 5.35301805e-01 2.26436257e-01 -4.91097122e-02
7.54484057e-01 4.04681802e-01 -4.02633399e-01 -5.80144942e-01
1.53659001e-01 6.35499179e-01 2.39223436e-01 6.65998459e-01
-9.87846494e-01 1.36431909e+00 6.86192703e+00 1.01539505e+00
-8.80739987e-01 6.47617877e-01 3.44292760e-01 -3.78193885e-01
-5.40383518e-01 4.05933827e-01 -1.04721475e+00 5.27989984e-01
1.27764714e+00 -5.18589258e-01 1.89440444e-01 1.03919351e+00
3.26367348e-01 -3.89827006e-02 -1.38366520e+00 5.26904166e-01
7.42439479e-02 -1.45910692e+00 -6.97604567e-02 -4.67175096e-02
6.86022341e-01 1.65450752e-01 -9.62871164e-02 1.07941055e+00
5.42529404e-01 -8.63132417e-01 7.58513033e-01 3.11324269e-01
7.71082163e-01 -3.26501042e-01 3.49080920e-01 7.03657925e-01
-9.45359647e-01 -6.81777671e-02 -6.17097020e-01 -4.98123318e-01
3.44773024e-01 4.18756366e-01 -1.26531875e+00 -6.07220083e-02
1.74813434e-01 5.90937912e-01 -4.15612042e-01 9.02759194e-01
-2.73647517e-01 8.55115652e-01 9.15286541e-02 -3.09837908e-01
4.48969781e-01 2.49774545e-01 5.30885160e-01 1.58794367e+00
3.87966603e-01 -2.63543371e-02 2.61825770e-01 8.45219672e-01
-5.03117070e-02 2.19246104e-01 -8.11304390e-01 -3.03175896e-01
6.41171336e-01 9.80825305e-01 -7.02788651e-01 -7.08370268e-01
-6.70724511e-01 1.11765921e+00 5.32933116e-01 2.85808563e-01
-5.69440603e-01 -5.46914518e-01 6.47866189e-01 1.58155605e-01
7.51701951e-01 -1.94999903e-01 -2.35245377e-01 -1.11234546e+00
-4.21834826e-01 -4.50058937e-01 3.07459533e-01 -7.36042976e-01
-1.45820570e+00 3.45033824e-01 5.28324060e-02 -1.00860167e+00
-7.13755667e-01 -7.03832626e-01 -1.36163330e+00 8.86055887e-01
-1.34418035e+00 -9.53974307e-01 2.34130010e-01 3.40533048e-01
1.24272025e+00 -1.75523952e-01 9.14545298e-01 -8.81650373e-02
-4.78919059e-01 6.07808173e-01 1.64037287e-01 1.99892074e-01
1.06723797e+00 -1.40808105e+00 8.91234219e-01 1.21912360e+00
4.27374780e-01 6.88653290e-01 9.37589824e-01 -7.23878980e-01
-1.02368438e+00 -1.15814292e+00 1.37370718e+00 -2.29739860e-01
1.10301638e+00 -6.98230803e-01 -1.06043732e+00 8.21979582e-01
4.88657326e-01 1.16364874e-01 8.66220295e-01 5.22578716e-01
-5.09501874e-01 2.59350359e-01 -6.03434265e-01 7.93097079e-01
9.67702329e-01 -8.89216721e-01 -1.01743829e+00 3.24311942e-01
7.89879501e-01 1.48423195e-01 -3.81123930e-01 -1.58362925e-01
1.47519201e-01 -7.56183565e-01 5.83231091e-01 -9.04228687e-01
4.32601631e-01 2.39173844e-01 -1.89025447e-01 -1.35222578e+00
-3.59813362e-01 -8.59275222e-01 -3.40305775e-01 1.29191124e+00
5.47821164e-01 -2.93633223e-01 6.39337003e-01 4.90436018e-01
-3.32423776e-01 -6.19701266e-01 -1.03482366e+00 -1.08085787e+00
2.73900479e-01 -5.78284919e-01 1.95501149e-01 8.36984277e-01
6.34901226e-01 5.90071261e-01 -5.58962762e-01 -4.18814391e-01
3.78726125e-01 -1.22891441e-01 5.76588929e-01 -1.05625427e+00
-7.74420977e-01 -5.59211373e-01 -1.26817867e-01 -8.96638095e-01
3.21542203e-01 -1.20600414e+00 5.33269107e-01 -1.46365809e+00
3.63605052e-01 -1.62439302e-01 -3.65202278e-01 8.63406599e-01
-4.12267745e-01 1.13824509e-01 2.39259481e-01 1.74071267e-01
-5.71614385e-01 5.36622465e-01 4.85371321e-01 -1.24417223e-01
-2.03936830e-01 2.16030359e-01 -6.34076655e-01 6.89906538e-01
8.36064398e-01 -5.97398698e-01 -5.01347959e-01 -2.17078924e-01
-1.31580774e-02 -8.34275112e-02 7.19071627e-02 -8.07901442e-01
2.31955588e-01 -1.62405431e-01 5.47769032e-02 -3.40082228e-01
4.33624953e-01 -7.44146630e-02 -5.69305539e-01 3.11377466e-01
-7.88242459e-01 -1.41568676e-01 1.59269739e-02 7.39757895e-01
1.18960030e-01 -8.62542272e-01 7.66745865e-01 -4.16675329e-01
-9.57965314e-01 1.46360636e-01 -1.00038576e+00 3.97280872e-01
1.05811977e+00 -2.58444160e-01 -1.33304924e-01 -3.45889598e-01
-8.91592503e-01 -4.28550467e-02 5.09769559e-01 3.07897538e-01
5.67905962e-01 -1.02055693e+00 -5.55197001e-01 1.50829166e-01
3.68497103e-01 -4.68560815e-01 1.41829208e-01 6.96530282e-01
1.69571280e-01 4.14424896e-01 1.58667192e-01 -2.25647509e-01
-1.15294719e+00 8.34214628e-01 -3.15041058e-02 -3.19570720e-01
-8.35671127e-01 8.19606662e-01 -4.00864636e-04 -1.05526499e-01
2.72229433e-01 -2.50042796e-01 1.30706578e-01 4.75885749e-01
8.40134025e-01 2.33969197e-01 -6.38628751e-02 -2.18095496e-01
-2.63325691e-01 -1.43163830e-01 -2.69538879e-01 -6.40255272e-01
1.27698398e+00 -8.75237957e-02 4.39547002e-01 1.03529429e+00
1.09324098e+00 1.03123046e-01 -1.27265549e+00 -5.87394953e-01
4.33746040e-01 -2.63145000e-01 7.87350610e-02 -6.14008188e-01
-1.67593539e-01 1.34312785e+00 9.82222930e-02 -9.29453522e-02
5.28701127e-01 2.09443301e-01 7.12355912e-01 5.21358013e-01
1.39071435e-01 -1.32658970e+00 3.63687396e-01 8.80231082e-01
2.18625486e-01 -1.31089318e+00 -3.80403847e-01 2.20130637e-01
-8.39820504e-01 1.02351713e+00 5.77045321e-01 5.70240915e-02
6.09104097e-01 4.35200304e-01 -2.53488477e-02 1.32169962e-01
-1.26678550e+00 -5.87096393e-01 -4.87396084e-02 6.79763913e-01
5.38868845e-01 -1.69895127e-01 -3.42421234e-01 6.90690219e-01
8.59965160e-02 -1.19677134e-01 7.15386033e-01 1.15294886e+00
-1.03897166e+00 -1.07608831e+00 -1.10466937e-02 4.00093526e-01
-3.54805559e-01 -6.94293499e-01 -1.82029456e-01 4.59070385e-01
-3.60884257e-02 1.00543487e+00 2.78413862e-01 -3.04205656e-01
-9.23539922e-02 9.20138597e-01 2.15096682e-01 -1.43385208e+00
-5.42775333e-01 1.75710261e-01 5.69844916e-02 -1.09485768e-01
2.25066058e-02 -7.66789138e-01 -1.24405718e+00 -6.22908920e-02
-2.19651625e-01 1.08071774e-01 3.20260555e-01 1.14822578e+00
2.69836605e-01 3.83289874e-01 4.31040943e-01 -9.72684622e-01
-1.07668614e+00 -1.25474262e+00 -4.35498923e-01 2.55704284e-01
2.58217543e-01 -4.28801537e-01 -3.85215849e-01 -2.92432830e-02]
|
[10.892176628112793, 8.182221412658691]
|
d34ac7e3-3b8b-4bc7-8cc6-14a076aeaa27
|
fast-and-correct-gradient-based-optimisation
|
2301.03415
| null |
https://arxiv.org/abs/2301.03415v1
|
https://arxiv.org/pdf/2301.03415v1.pdf
|
Fast and Correct Gradient-Based Optimisation for Probabilistic Programming via Smoothing
|
We study the foundations of variational inference, which frames posterior inference as an optimisation problem, for probabilistic programming. The dominant approach for optimisation in practice is stochastic gradient descent. In particular, a variant using the so-called reparameterisation gradient estimator exhibits fast convergence in a traditional statistics setting. Unfortunately, discontinuities, which are readily expressible in programming languages, can compromise the correctness of this approach. We consider a simple (higher-order, probabilistic) programming language with conditionals, and we endow our language with both a measurable and a smoothed (approximate) value semantics. We present type systems which establish technical pre-conditions. Thus we can prove stochastic gradient descent with the reparameterisation gradient estimator to be correct when applied to the smoothed problem. Besides, we can solve the original problem up to any error tolerance by choosing an accuracy coefficient suitably. Empirically we demonstrate that our approach has a similar convergence as a key competitor, but is simpler, faster, and attains orders of magnitude reduction in work-normalised variance.
|
['Dominik Wagner', 'C. -H. Luke Ong', 'Basim Khajwal']
|
2023-01-09
| null | null | null | null |
['probabilistic-programming']
|
['methodology']
|
[ 1.01410225e-01 2.52201051e-01 5.22899255e-02 -4.19464767e-01
-1.07873046e+00 -6.61484122e-01 7.45478094e-01 6.26192689e-02
-6.64184690e-01 8.40996027e-01 -1.20149009e-01 -4.89826232e-01
-3.26436937e-01 -7.51531780e-01 -8.88496101e-01 -1.09415233e+00
1.30870042e-03 5.18067896e-01 2.46597737e-01 -7.91703016e-02
4.38406259e-01 5.12858093e-01 -1.41995239e+00 -3.49632531e-01
7.14103997e-01 7.86516070e-01 -1.39793335e-02 1.10219014e+00
-7.06657171e-02 5.00277340e-01 -2.05508143e-01 -6.60894156e-01
1.37682557e-01 -1.99947208e-01 -8.06128025e-01 -2.43551567e-01
3.41288149e-01 -6.02324493e-02 1.03258580e-01 1.27773726e+00
4.75253314e-01 2.32771531e-01 8.77569258e-01 -1.26771736e+00
-3.73479903e-01 6.78243995e-01 -5.85552216e-01 2.07367018e-02
3.70826215e-01 -2.35788986e-01 1.08482468e+00 -8.22220743e-01
2.72867739e-01 1.35813951e+00 1.04988921e+00 4.68727410e-01
-1.89554060e+00 8.16021115e-02 -1.35092065e-01 -4.47548270e-01
-1.46309948e+00 -4.09107476e-01 5.45335770e-01 -4.52518582e-01
5.97422838e-01 6.40398204e-01 2.73964882e-01 1.00977337e+00
2.21753284e-01 8.24152827e-01 1.18086791e+00 -6.40306950e-01
5.24844646e-01 5.03517747e-01 2.48110458e-01 6.62616968e-01
1.31490752e-01 4.31021638e-02 -2.89840400e-01 -6.03884399e-01
7.68040538e-01 -5.14341928e-02 -1.62628576e-01 -5.25113583e-01
-1.13173997e+00 1.11260295e+00 -1.49505168e-01 1.79526821e-01
-2.63445020e-01 6.09054208e-01 3.25572461e-01 2.09569395e-01
6.06828690e-01 3.48481052e-02 -3.95450145e-01 -3.55500907e-01
-1.12666523e+00 6.11258328e-01 1.32603574e+00 9.09360766e-01
6.46758139e-01 -1.78231150e-01 -3.38717043e-01 7.27610707e-01
6.67164624e-01 6.73699021e-01 1.53619096e-01 -1.33525288e+00
1.19987801e-01 -3.23259681e-01 3.99494201e-01 -7.81900346e-01
-2.20045254e-01 -8.82056355e-02 -7.00100780e-01 2.21837416e-01
7.38316059e-01 -1.70616746e-01 -4.02252823e-01 1.88497233e+00
4.36782032e-01 1.18609928e-01 -1.50396481e-01 4.68681395e-01
-1.15028493e-01 6.96401656e-01 -8.73942301e-03 -5.89771569e-01
1.15991914e+00 -3.16396087e-01 -5.43054879e-01 2.44322345e-01
4.24684048e-01 -6.42250299e-01 1.19850266e+00 6.15962625e-01
-1.32831895e+00 -2.24590525e-01 -9.09036636e-01 -1.70724735e-01
-1.19973674e-01 -2.47788519e-01 5.83508790e-01 1.20910263e+00
-1.31589162e+00 1.04990613e+00 -1.14639211e+00 -1.64538339e-01
-2.95433160e-02 1.98001042e-01 1.20744638e-01 5.09897530e-01
-8.73812914e-01 8.77752006e-01 4.69650209e-01 1.61531597e-01
-6.61143780e-01 -8.74754608e-01 -8.13538313e-01 -6.96668774e-02
2.53962815e-01 -6.14922941e-01 1.43382955e+00 -5.07029176e-01
-2.01717687e+00 8.02210808e-01 -3.73162836e-01 -4.32460815e-01
1.04167914e+00 -1.75603360e-01 1.18129008e-01 -1.83172613e-01
-3.59259956e-02 6.20761812e-02 9.52637672e-01 -1.09108937e+00
-5.04678607e-01 -4.76049304e-01 1.83175877e-01 -1.41423315e-01
-9.39749088e-03 1.90634012e-01 -2.05877587e-01 -4.04629201e-01
1.31558627e-01 -8.41098368e-01 -3.56956691e-01 1.06909890e-02
-4.29675043e-01 -4.77212220e-01 1.49025813e-01 -4.33326393e-01
1.11101234e+00 -2.12158847e+00 1.90429002e-01 5.91314495e-01
4.65989336e-02 -2.90413409e-01 4.21571076e-01 2.52456248e-01
5.99104054e-02 3.75130743e-01 -8.31976950e-01 -7.20490098e-01
6.39749408e-01 2.89604396e-01 -4.77527201e-01 9.76015389e-01
5.86952455e-02 6.36463940e-01 -9.81853127e-01 -6.91975296e-01
2.15118334e-01 3.65459561e-01 -7.76086748e-01 4.65567708e-02
-3.14624369e-01 7.50463083e-02 -5.32330394e-01 2.73271531e-01
7.61160553e-01 1.63792446e-02 1.57768503e-01 2.28447959e-01
-3.30278873e-01 6.80838898e-02 -1.65585530e+00 1.61679316e+00
-6.95836127e-01 5.56807339e-01 3.16252768e-01 -1.12785578e+00
6.46967351e-01 1.76375970e-01 3.29153150e-01 2.12640658e-01
-1.36040330e-01 3.85443598e-01 -6.91611707e-01 -2.74787724e-01
5.89203358e-01 -5.26782036e-01 -8.90996158e-02 3.91193211e-01
-1.11133814e-01 -4.33709979e-01 1.56165093e-01 8.39145184e-02
6.89464331e-01 5.17705977e-01 8.11708719e-02 -8.24707031e-01
5.07316709e-01 -3.81032765e-01 2.47457415e-01 1.32139301e+00
-6.17658831e-02 6.78532422e-01 8.80577326e-01 -9.47243795e-02
-1.17888999e+00 -1.41974413e+00 -7.66400397e-01 1.36070561e+00
-3.35008562e-01 -2.67958432e-01 -8.90049934e-01 -3.57249141e-01
1.33455634e-01 9.17865515e-01 -5.16200244e-01 1.78761512e-01
-3.88550013e-01 -1.11447799e+00 7.07562208e-01 5.02282321e-01
1.35774627e-01 -5.82690001e-01 -5.64467072e-01 3.84636223e-01
1.75983265e-01 -6.40039682e-01 -3.56018066e-01 1.37657225e-01
-9.20148313e-01 -5.42846799e-01 -8.63482058e-01 -3.40877950e-01
2.49541238e-01 -5.14451921e-01 1.13762474e+00 -1.58277988e-01
1.59197941e-01 5.99512875e-01 2.65247554e-01 -3.37656349e-01
-6.44500434e-01 -2.42899150e-01 2.24480197e-01 2.79559195e-02
2.33849898e-01 -6.96491599e-01 -1.55416429e-01 -1.43834829e-01
-9.51896429e-01 -4.09041554e-01 1.72829598e-01 8.73215616e-01
5.48284531e-01 1.27649054e-01 -5.63856922e-02 -9.93286371e-01
7.04475284e-01 -3.58213395e-01 -1.07503188e+00 2.15394035e-01
-6.85885966e-01 6.62428916e-01 3.82897824e-01 -1.84705570e-01
-1.18028772e+00 -5.25172204e-02 -3.21020633e-01 -7.46845547e-03
2.28444897e-02 6.49584889e-01 -4.54120280e-04 6.31096587e-02
7.51984835e-01 2.63023585e-01 -2.28375625e-02 -5.11081040e-01
6.66921675e-01 6.32497251e-01 6.51663005e-01 -1.28882992e+00
6.16659164e-01 7.14721978e-01 3.39408934e-01 -9.88551319e-01
-6.70190215e-01 -3.43519837e-01 -4.65241075e-01 1.49538338e-01
6.05952740e-01 -4.79482561e-01 -1.15022111e+00 4.11149591e-01
-1.23412704e+00 -2.71901697e-01 -3.94457132e-01 3.60130876e-01
-1.04752862e+00 6.51261330e-01 -5.64654887e-01 -1.59636796e+00
2.97194012e-02 -9.91693914e-01 1.25157177e+00 -4.87992428e-02
-1.47700325e-01 -1.37515545e+00 3.74476522e-01 -2.50404090e-01
2.51942605e-01 3.15387607e-01 6.97045505e-01 -5.98074436e-01
-1.67312294e-01 -5.17259240e-02 -1.04919322e-01 5.34725845e-01
-4.88925815e-01 6.19814694e-01 -9.54833567e-01 -4.59828973e-02
5.68538904e-01 9.44018587e-02 7.64142692e-01 5.65094650e-01
1.18486798e+00 -3.04155052e-01 -1.72332913e-01 5.77163398e-01
1.56641603e+00 -5.67384183e-01 4.91188645e-01 1.72756106e-01
6.43267870e-01 6.37788773e-01 1.98000863e-01 6.14222467e-01
3.06765914e-01 6.24091685e-01 -9.72940028e-02 4.79136944e-01
6.27494752e-01 -1.18198782e-01 6.13588631e-01 7.25365341e-01
-3.83440375e-01 2.55975127e-01 -9.36707437e-01 4.45089459e-01
-2.09796667e+00 -1.04593074e+00 -4.78563130e-01 2.59893394e+00
1.37866938e+00 3.35965127e-01 2.60433465e-01 1.53565153e-01
6.56048477e-01 -1.44018941e-02 -1.84436768e-01 -7.39744365e-01
1.20118134e-01 3.97404432e-01 9.19044852e-01 8.87689590e-01
-9.82626021e-01 4.04869556e-01 7.53781080e+00 1.07462776e+00
-6.09831810e-01 4.06447351e-01 3.64846021e-01 6.20027520e-02
-6.42992496e-01 1.25747323e-01 -7.85293996e-01 6.64168775e-01
1.35949075e+00 -1.73317209e-01 5.28289139e-01 1.07319021e+00
2.22792462e-01 -1.39915496e-01 -1.44942307e+00 8.03076506e-01
-3.77262324e-01 -1.03552866e+00 -4.90366101e-01 1.73096329e-01
6.33571744e-01 -6.93622902e-02 3.58908437e-02 3.09847742e-01
5.96037924e-01 -1.00614417e+00 9.23902810e-01 7.78348505e-01
4.28840667e-01 -9.51296031e-01 6.55133545e-01 5.22296488e-01
-8.29824448e-01 3.16268176e-01 -4.76536214e-01 -9.10646543e-02
2.60118008e-01 1.03189671e+00 -1.44353092e-01 4.25606132e-01
4.81310219e-01 2.09283516e-01 -7.24966750e-02 8.49648237e-01
-2.93110847e-01 7.00742245e-01 -1.10077775e+00 -2.13852540e-01
1.67742416e-01 -6.54205918e-01 7.32829809e-01 1.57966626e+00
2.00325012e-01 -2.74804026e-01 -1.17822702e-03 1.15713394e+00
1.95857227e-01 -7.30930790e-02 -3.32898408e-01 1.59830198e-01
2.85466492e-01 8.40598643e-01 -5.16593099e-01 -4.57945585e-01
-3.20999444e-01 7.92610407e-01 3.39652866e-01 5.09948373e-01
-8.45850885e-01 -3.78372341e-01 5.25280237e-01 -1.90176100e-01
3.14886957e-01 -3.47700030e-01 -5.84691346e-01 -1.13332474e+00
2.60975927e-01 -5.00434637e-01 2.58984953e-01 -4.61890876e-01
-1.35234988e+00 -3.56297530e-02 4.95941967e-01 -3.61503929e-01
-6.42021894e-01 -8.60765874e-01 -5.60947955e-01 1.14899528e+00
-1.17768919e+00 -7.30590165e-01 4.94048208e-01 3.83774579e-01
8.49037245e-02 4.48105514e-01 8.11846375e-01 -1.49626762e-01
-4.24247801e-01 5.38379610e-01 6.42848432e-01 -2.82545596e-01
1.63300827e-01 -2.03093982e+00 1.86257273e-01 9.62099850e-01
1.91987455e-01 8.39438260e-01 1.35044193e+00 -2.72686809e-01
-1.62917614e+00 -6.31045461e-01 1.01592946e+00 -8.65411639e-01
1.11035073e+00 -4.26697731e-01 -8.80526006e-01 8.43192399e-01
-1.85362071e-01 -2.14516535e-01 2.40370303e-01 5.66211045e-01
-2.80117959e-01 8.20150077e-02 -1.28815675e+00 6.34334087e-01
7.26689875e-01 -9.09837723e-01 -8.88186038e-01 5.36540329e-01
4.44318682e-01 -4.42759991e-01 -1.18188179e+00 5.10206819e-02
7.76894450e-01 -9.93264794e-01 9.52237606e-01 -4.81614500e-01
2.14761525e-01 -3.18067610e-01 -2.89289683e-01 -8.82910967e-01
2.57166605e-02 -1.14992499e+00 -2.91815639e-01 1.27643859e+00
3.93321782e-01 -9.34851348e-01 5.96101582e-01 1.11570489e+00
2.39353716e-01 -6.34400308e-01 -1.16116416e+00 -9.63168979e-01
5.58621228e-01 -1.06511915e+00 2.56841183e-01 4.90730494e-01
1.77504748e-01 -1.01236090e-01 -2.58489579e-01 2.53088504e-01
9.77057815e-01 -1.43795431e-01 5.70910871e-01 -1.21750712e+00
-6.95482254e-01 -7.69098103e-01 -3.79772067e-01 -1.26883352e+00
3.01493645e-01 -7.23965943e-01 3.97425324e-01 -9.27858770e-01
1.56256482e-01 -3.62775564e-01 -2.76705652e-01 2.86627393e-02
-2.15525985e-01 2.86254976e-02 -2.40134522e-01 9.49782804e-02
-2.62473285e-01 3.39902550e-01 5.70518970e-01 1.95860356e-01
-2.97754794e-01 3.76548320e-01 -4.68049467e-01 1.00849020e+00
6.03707731e-01 -6.49212956e-01 -1.27821192e-01 -1.41334161e-01
6.41833842e-01 -8.08499157e-02 6.40827298e-01 -5.15611827e-01
3.18037778e-01 -2.27432221e-01 -1.35811232e-02 -3.77262831e-01
1.53309464e-01 -4.62038666e-01 2.52600253e-01 1.31680965e-01
-5.34788132e-01 -1.49843559e-01 7.44546801e-02 5.64215660e-01
2.42116079e-01 -9.39285815e-01 7.62476504e-01 -1.20316749e-03
-1.63063988e-01 7.59877488e-02 -3.75329494e-01 1.52011648e-01
4.78157401e-01 -6.75779209e-02 3.23448628e-01 -1.30133659e-01
-8.17771077e-01 -2.27133445e-02 6.04793370e-01 -3.50638688e-01
1.93343475e-01 -1.16305864e+00 -8.75863731e-01 -1.45510510e-01
-2.78354436e-01 1.62747875e-02 -6.89243600e-02 1.11986327e+00
-5.19255877e-01 2.35533223e-01 5.39018095e-01 -7.96551585e-01
-7.67711461e-01 5.55514038e-01 2.70380646e-01 -5.43304943e-02
-4.83318299e-01 8.94268215e-01 -8.05435553e-02 -3.93141896e-01
2.01204479e-01 -5.39660096e-01 4.37775433e-01 -1.87673554e-01
6.10176980e-01 6.53398693e-01 -1.14216514e-01 -3.29335988e-01
-3.61036927e-01 4.51128483e-01 1.99490607e-01 -8.52109075e-01
1.04636312e+00 -3.40692997e-01 -4.06854451e-01 1.04305303e+00
1.25857830e+00 2.80771196e-01 -1.22070682e+00 -1.74231365e-01
2.41940811e-01 -2.87411541e-01 1.88334793e-01 -1.63598731e-01
-4.35616642e-01 7.28983521e-01 2.86270171e-01 7.87680686e-01
6.41688824e-01 1.16096757e-01 1.81366280e-01 6.22032702e-01
3.19575250e-01 -1.21924210e+00 -8.66368473e-01 3.64491612e-01
7.65746772e-01 -1.11649382e+00 1.13570988e-01 -2.51867980e-01
-1.17172927e-01 1.10419750e+00 -2.72841990e-01 -3.04633468e-01
9.08344150e-01 2.03033596e-01 -3.07463229e-01 -3.42787392e-02
-4.29728389e-01 -1.82231925e-02 8.08352903e-02 4.78836417e-01
3.99936259e-01 2.56130040e-01 -4.37534630e-01 2.74720460e-01
-4.59007621e-01 -1.11425750e-01 4.40968603e-01 9.05875504e-01
-3.44592780e-01 -8.67277443e-01 -5.01772583e-01 2.82876790e-01
-7.67898142e-01 -2.84846663e-01 2.87055463e-01 7.86575317e-01
-3.96088243e-01 8.82722259e-01 4.93933037e-02 3.13858032e-01
1.71932399e-01 2.70658672e-01 7.46918678e-01 -4.04498011e-01
-1.53106451e-01 -5.84423803e-02 -2.50624213e-03 -5.69400966e-01
-6.15852058e-01 -8.84587348e-01 -9.79776800e-01 -6.94268227e-01
-3.17477852e-01 3.83740574e-01 1.11457884e+00 1.20715582e+00
-9.42075029e-02 5.88820502e-02 5.01694441e-01 -7.84949124e-01
-1.43805528e+00 -8.54209363e-01 -1.03467298e+00 4.81154993e-02
4.03464347e-01 -5.12315094e-01 -7.30747283e-01 6.56838194e-02]
|
[6.961679935455322, 4.0986433029174805]
|
d8c7e29f-c276-4f8f-996f-767bf23f2261
|
point-cloud-instance-segmentation-using
|
1912.00145
| null |
https://arxiv.org/abs/1912.00145v2
|
https://arxiv.org/pdf/1912.00145v2.pdf
|
Point Cloud Instance Segmentation using Probabilistic Embeddings
|
In this paper we propose a new framework for point cloud instance segmentation. Our framework has two steps: an embedding step and a clustering step. In the embedding step, our main contribution is to propose a probabilistic embedding space for point cloud embedding. Specifically, each point is represented as a tri-variate normal distribution. In the clustering step, we propose a novel loss function, which benefits both the semantic segmentation and the clustering. Our experimental results show important improvements to the SOTA, i.e., 3.1% increased average per-category mAP on the PartNet dataset.
|
['Biao Zhang', 'Peter Wonka']
|
2019-11-30
| null |
http://openaccess.thecvf.com//content/CVPR2021/html/Zhang_Point_Cloud_Instance_Segmentation_Using_Probabilistic_Embeddings_CVPR_2021_paper.html
|
http://openaccess.thecvf.com//content/CVPR2021/papers/Zhang_Point_Cloud_Instance_Segmentation_Using_Probabilistic_Embeddings_CVPR_2021_paper.pdf
|
cvpr-2021-1
|
['3d-instance-segmentation-1']
|
['computer-vision']
|
[-2.62412708e-03 2.19144419e-01 1.87126026e-02 -3.63981545e-01
-8.53131652e-01 -3.91092449e-01 3.39010596e-01 2.51253456e-01
-3.13903719e-01 2.29232311e-01 -2.69073218e-01 -3.15601751e-02
5.05967885e-02 -9.97998536e-01 -8.23354661e-01 -7.15697408e-01
1.06029369e-01 6.53564453e-01 7.13029563e-01 3.24061096e-01
8.64343643e-02 7.23694503e-01 -1.22461104e+00 -1.12464868e-01
8.64089310e-01 1.12289095e+00 1.09260008e-01 3.47448826e-01
-5.17809689e-01 7.94448853e-02 -4.35510546e-01 -3.88824165e-01
2.50554830e-01 4.48666625e-02 -5.97050011e-01 2.44559482e-01
1.80296183e-01 -1.75686210e-01 -2.07129732e-01 9.94920254e-01
3.44232827e-01 6.82868659e-02 8.25764596e-01 -1.59526742e+00
-2.68765837e-01 5.37995875e-01 -8.26409400e-01 -1.94633424e-01
-3.10985595e-01 -1.58760935e-01 1.01779640e+00 -1.00414515e+00
2.97195315e-01 1.21388578e+00 7.33725786e-01 3.25120568e-01
-1.18067634e+00 -7.15715766e-01 1.88864946e-01 2.20145136e-01
-1.59235036e+00 -8.11026022e-02 1.18048024e+00 -6.11033618e-01
6.07568800e-01 3.56172360e-02 7.69170403e-01 4.73670632e-01
-6.66121542e-02 9.45040941e-01 7.61721611e-01 -2.41392404e-01
4.68874902e-01 1.77933216e-01 8.38997588e-02 2.34530479e-01
3.93040299e-01 -3.30209613e-01 9.89776179e-02 -2.93306887e-01
7.43318021e-01 2.57923692e-01 6.40650690e-02 -7.83901572e-01
-9.31299210e-01 1.07098472e+00 6.16421998e-01 8.86803344e-02
-3.72333497e-01 4.29431945e-01 2.12724358e-01 -3.12813193e-01
7.72877932e-01 -2.32641143e-03 -2.59647220e-01 -2.06489004e-02
-1.00076759e+00 1.04428075e-01 6.88032508e-01 1.03778112e+00
8.71932447e-01 -3.96784484e-01 1.04031004e-01 8.94711018e-01
7.38814354e-01 6.75536036e-01 -4.73121181e-02 -8.93536210e-01
4.84485060e-01 6.64993286e-01 -1.99718727e-03 -1.13599300e+00
-1.08835235e-01 -3.01424682e-01 -5.81835091e-01 1.51381329e-01
-1.74595676e-02 1.28951129e-02 -9.30311203e-01 1.36577749e+00
5.29839337e-01 6.51340008e-01 -2.32399926e-01 7.14472950e-01
4.94630188e-01 6.47873819e-01 2.19664708e-01 3.05423707e-01
1.28076375e+00 -1.03153813e+00 -6.97706401e-01 4.34153974e-02
2.79360324e-01 -7.74850965e-01 8.39026332e-01 6.33153319e-02
-1.04545891e+00 -3.08075249e-01 -1.07946086e+00 -3.45463045e-02
-3.39036793e-01 3.26363780e-02 5.27135730e-01 5.54042339e-01
-1.00992203e+00 4.68361557e-01 -1.15263915e+00 -1.53759167e-01
6.75183594e-01 3.06306958e-01 -2.25461468e-01 -4.21144851e-02
-6.86042726e-01 2.81805634e-01 3.85581851e-01 -1.00482091e-01
-3.44394505e-01 -7.31215775e-01 -9.36357498e-01 2.91190088e-01
2.47390360e-01 -7.65020907e-01 9.71698999e-01 -2.80578166e-01
-1.45809960e+00 7.23316729e-01 -2.90181071e-01 -3.64775002e-01
4.16446388e-01 -7.40262493e-02 -5.52655011e-02 4.06336993e-01
1.90895125e-01 7.54293501e-01 7.95546114e-01 -1.62310112e+00
-7.23416805e-01 -5.27640581e-01 -1.55629233e-01 -2.01889053e-02
-2.50889421e-01 -2.92877555e-01 -1.00656545e+00 -7.71627724e-01
4.98667240e-01 -1.11337650e+00 -2.12478429e-01 2.60386169e-01
-4.50392902e-01 -3.35838437e-01 1.00112951e+00 -6.74391806e-01
9.49257493e-01 -2.37397790e+00 1.56027243e-01 3.62906247e-01
4.17135626e-01 -3.06446254e-02 1.93776160e-01 1.64497823e-01
4.66481633e-02 4.15280998e-01 -7.13278651e-01 -7.87864208e-01
3.41177940e-01 1.98128805e-01 -2.97072262e-01 4.77244169e-01
3.89671654e-01 7.95291781e-01 -7.90995419e-01 -4.97256756e-01
5.06054938e-01 8.40651035e-01 -6.49892867e-01 7.16003403e-02
5.73048219e-02 -1.13822937e-01 -4.52916443e-01 6.81339145e-01
1.23537898e+00 -1.74227804e-01 -2.23010182e-01 -1.46788031e-01
3.56498100e-02 -1.67385563e-02 -1.21816540e+00 1.71612835e+00
-2.53850609e-01 4.10671353e-01 1.31775171e-01 -7.25663245e-01
9.68743980e-01 1.36685416e-01 9.13580239e-01 -3.56428809e-02
3.33027430e-02 9.06711742e-02 -3.54250491e-01 7.92031139e-02
5.11068881e-01 -9.56282765e-02 2.86579821e-02 2.85820991e-01
4.52455096e-02 -3.08822751e-01 -1.80072486e-01 1.46626487e-01
9.37407315e-01 -9.21687931e-02 -2.79028386e-01 -1.75933555e-01
4.33385760e-01 8.02901238e-02 7.92755008e-01 2.09769696e-01
-2.44819552e-01 8.64198387e-01 6.23605847e-01 -4.23837043e-02
-1.02470791e+00 -1.36586940e+00 -3.01197022e-01 3.27120990e-01
5.47808230e-01 -4.96041119e-01 -8.65909457e-01 -9.58614886e-01
3.76311094e-01 6.76670194e-01 -3.95258427e-01 -1.83493048e-01
-5.05114973e-01 -6.31658614e-01 2.16772959e-01 8.60658109e-01
4.88726318e-01 -6.42938733e-01 -2.51486808e-01 1.08399890e-01
-1.85661037e-02 -1.25824928e+00 -5.85960388e-01 -9.02865753e-02
-1.28533983e+00 -9.16784704e-01 -5.90192020e-01 -6.79374337e-01
7.55697548e-01 4.46086764e-01 8.13248575e-01 -1.69535756e-01
1.74806751e-02 4.06071752e-01 -3.74242067e-01 -4.81499404e-01
1.87533662e-01 2.54241973e-01 -2.49377728e-01 -7.35269412e-02
6.14657760e-01 -6.53951347e-01 -8.71182084e-01 4.52272117e-01
-8.72085512e-01 -2.85123497e-01 7.48706996e-01 5.24684668e-01
1.13933301e+00 2.68275946e-01 -8.18849728e-03 -7.08543658e-01
2.58253813e-01 -6.38231814e-01 -7.09965527e-01 -6.46917969e-02
-6.92509651e-01 -1.21546425e-01 2.75353372e-01 -1.50668427e-01
-5.42392373e-01 3.39522898e-01 -3.47554475e-01 -7.53338337e-01
-2.46253371e-01 1.40347615e-01 -5.13026178e-01 -2.29538441e-01
-6.02720566e-02 9.96375456e-02 1.18018277e-02 -7.79808402e-01
5.26230216e-01 6.89570725e-01 3.54951680e-01 -5.86252272e-01
1.23789012e+00 8.68925810e-01 -9.05658156e-02 -9.47542310e-01
-1.55924410e-01 -9.49591219e-01 -8.58768761e-01 -8.01045354e-03
1.08105230e+00 -8.65971148e-01 -4.67656285e-01 2.86992997e-01
-1.21818542e+00 -1.25386361e-02 -5.28337896e-01 5.41246533e-01
-5.82024992e-01 4.99958366e-01 -4.00069296e-01 -5.93464315e-01
-3.91640037e-01 -1.27444613e+00 1.43185830e+00 2.19443664e-01
3.94928992e-01 -9.92874920e-01 1.90571204e-01 3.48149210e-01
5.58991954e-02 3.48155111e-01 6.94629610e-01 -5.81194341e-01
-7.45287716e-01 -3.23771894e-01 -5.51824689e-01 5.57239175e-01
3.04264762e-02 9.65751261e-02 -1.03538311e+00 -1.75437123e-01
9.97918025e-02 3.20736438e-01 8.08745801e-01 3.50485593e-01
1.48998499e+00 2.40491703e-01 -6.21238530e-01 8.96964014e-01
1.60010493e+00 2.33712211e-01 7.04209149e-01 2.90624142e-01
9.74481046e-01 4.50097471e-01 6.35633171e-01 2.08282992e-01
5.89012444e-01 7.75897682e-01 4.09442514e-01 -1.60886437e-01
-1.56188235e-01 -3.24384540e-01 -1.02194846e-02 1.16011381e+00
2.44515672e-01 8.96046124e-03 -1.17581487e+00 8.81458580e-01
-1.95677733e+00 -6.48349047e-01 -2.70488322e-01 2.05284691e+00
4.08200055e-01 1.75161436e-01 6.68936446e-02 2.21752837e-01
8.44187260e-01 4.23932932e-02 -4.91821736e-01 -6.17564209e-02
1.94363907e-01 3.21014524e-01 7.52612174e-01 5.95554173e-01
-1.33081174e+00 9.14001167e-01 6.08724356e+00 8.86141241e-01
-9.07057166e-01 2.34114662e-01 1.94830179e-01 2.17674494e-01
-3.95386159e-01 -4.27884571e-02 -8.06433916e-01 7.53377795e-01
6.56913280e-01 -1.44742563e-01 4.92288284e-02 1.10482860e+00
4.86687087e-02 1.86019406e-01 -7.06109285e-01 1.12101984e+00
6.19914234e-02 -1.10785174e+00 4.92816642e-02 3.41874838e-01
4.22316611e-01 5.86401597e-02 3.42574231e-02 1.77905008e-01
-8.47463682e-02 -7.54746497e-01 6.57641649e-01 3.51454139e-01
5.65432370e-01 -1.20179367e+00 8.08026969e-01 1.74142197e-01
-1.48169410e+00 3.77810121e-01 -5.31064630e-01 4.82564896e-01
2.25219056e-01 9.63304937e-01 -7.25068092e-01 6.91042364e-01
6.69304848e-01 7.72479892e-01 -4.63235825e-01 1.51688540e+00
-3.81358624e-01 6.66320682e-01 -5.51036298e-01 3.56964886e-01
1.77771926e-01 -5.67530155e-01 7.83676505e-01 1.18466020e+00
1.90176710e-01 -3.29905115e-02 2.83194214e-01 1.07727873e+00
-1.62055746e-01 1.28325671e-01 -4.92751956e-01 1.35670841e-01
6.54014349e-01 1.36120248e+00 -9.47327197e-01 -1.57328948e-01
-2.74066687e-01 1.08241749e+00 1.40561089e-01 2.58153707e-01
-9.76796508e-01 -9.36043978e-01 9.49063063e-01 9.67953280e-02
7.13974535e-01 -4.20685410e-01 -7.26470411e-01 -1.03883779e+00
3.24924767e-01 -4.09366935e-02 3.55132036e-02 -3.51625711e-01
-1.24990535e+00 4.32404339e-01 1.79709837e-01 -1.05772638e+00
1.56915292e-01 -5.54946244e-01 -7.30676234e-01 9.99703526e-01
-1.75540233e+00 -1.28327954e+00 -3.33284944e-01 3.23939949e-01
4.14928913e-01 1.63334787e-01 5.31102955e-01 6.31688476e-01
-5.03916919e-01 5.55647492e-01 2.90961772e-01 2.52828926e-01
3.49562883e-01 -1.69792306e+00 6.54880941e-01 7.75965035e-01
1.54901445e-01 5.90058804e-01 4.32356328e-01 -5.72865188e-01
-1.10531485e+00 -1.37185180e+00 6.64331377e-01 -4.22663808e-01
5.56479216e-01 -6.67012453e-01 -9.95220602e-01 6.69181764e-01
-2.47103721e-01 1.32455938e-02 7.69330621e-01 -7.88196828e-03
-1.34175003e-01 -9.58677083e-02 -1.56004477e+00 2.93546081e-01
7.16677129e-01 -4.39712435e-01 -6.36322916e-01 2.45028436e-01
1.05368102e+00 -2.05845550e-01 -1.17209589e+00 5.45341194e-01
3.13140333e-01 -4.84332740e-01 1.21233392e+00 1.50347296e-02
1.41611710e-01 -6.12782836e-01 -2.35427037e-01 -1.16085970e+00
-4.46856320e-01 -2.19613031e-01 -3.13525677e-01 1.30412829e+00
3.28181535e-01 -6.64667964e-01 1.05591273e+00 4.28516477e-01
-1.75544649e-01 -9.48485374e-01 -1.04381156e+00 -9.72591102e-01
2.46151358e-01 -7.86578774e-01 7.81236589e-01 7.10867345e-01
-5.44377744e-01 9.90024060e-02 1.72425151e-01 4.37600493e-01
1.11102426e+00 2.21663062e-02 7.21508145e-01 -1.36010897e+00
-1.82077270e-02 -5.04509568e-01 -7.37626910e-01 -1.11669016e+00
1.29144222e-01 -8.94827783e-01 7.84903765e-02 -1.69772923e+00
2.12914124e-01 -6.35535657e-01 -4.33126241e-01 6.92128912e-02
-1.52732164e-01 9.00922492e-02 3.33793730e-01 3.04099381e-01
-5.00692785e-01 8.90846252e-01 7.56699145e-01 -3.54766667e-01
-2.02202231e-01 2.03697652e-01 -4.12134439e-01 7.03527510e-01
9.87049639e-01 -7.52432346e-01 -3.10618013e-01 -6.12268984e-01
-3.52145791e-01 -6.14661098e-01 3.99588108e-01 -1.07268345e+00
7.42670894e-02 1.46669433e-01 9.60991159e-02 -1.21649611e+00
5.63710511e-01 -1.20523894e+00 7.79542848e-02 3.68560523e-01
3.18059683e-01 -1.16186157e-01 3.44978243e-01 9.43980098e-01
-2.73484081e-01 -9.66070443e-02 8.01678598e-01 2.98232794e-01
-4.70884025e-01 5.15981793e-01 2.48883605e-01 -1.87287673e-01
1.35728848e+00 -2.56455332e-01 -6.85476065e-02 5.47836944e-02
-7.19580531e-01 4.66944903e-01 7.40622580e-01 3.67947161e-01
8.73726189e-01 -1.57094932e+00 -5.42305470e-01 6.01460449e-02
1.82207227e-01 4.23347354e-01 1.35168642e-01 8.57284307e-01
-6.23025417e-01 2.92983741e-01 7.51611441e-02 -9.40259516e-01
-1.02689373e+00 4.66762722e-01 1.55134043e-02 3.77319045e-02
-7.96190619e-01 7.94203222e-01 6.21138252e-02 -6.01553798e-01
1.85978964e-01 -4.76866215e-01 -1.19278416e-01 -7.31836185e-02
3.29786003e-01 5.76108754e-01 -1.06746126e-02 -7.94579387e-01
-4.85350132e-01 9.06981766e-01 -2.23714020e-02 -2.68006861e-01
1.36917782e+00 7.59888813e-02 -3.17846507e-01 6.79216743e-01
1.45132947e+00 -3.51686403e-02 -1.12295556e+00 -9.98823494e-02
8.88717026e-02 -7.22051263e-01 3.77995074e-01 -4.76911992e-01
-1.19539869e+00 9.92840469e-01 7.82451093e-01 1.62760600e-01
9.60105956e-01 2.58914977e-01 1.26044500e+00 -6.00071363e-02
3.57523531e-01 -9.73420501e-01 -3.84049565e-01 4.08946902e-01
6.68449044e-01 -1.15881371e+00 -1.04661793e-01 -7.65875816e-01
-5.51411092e-01 9.56979811e-01 3.16856652e-01 -3.99712592e-01
1.01336122e+00 4.82231677e-02 -1.46354377e-01 -3.95449340e-01
-3.34448040e-01 -1.89554170e-01 4.57927942e-01 5.06758273e-01
6.56915829e-02 3.20293367e-01 -3.27675372e-01 7.14665711e-01
-1.68219432e-01 -2.81294137e-01 3.00332136e-03 8.16244483e-01
-5.55845380e-01 -1.13993156e+00 -4.35757190e-01 3.06763798e-01
-1.78658947e-01 2.25385293e-01 -4.15556908e-01 8.54183018e-01
1.35269433e-01 8.35955203e-01 3.90002638e-01 -7.10394442e-01
5.96491814e-01 5.99608347e-02 1.08966976e-01 -7.98347652e-01
-1.20407544e-01 1.93511099e-01 -5.43840706e-01 -6.62612915e-01
-1.65767118e-01 -7.59979248e-01 -1.43692994e+00 -2.54704446e-01
-3.97402793e-01 3.37061524e-01 1.31387734e+00 4.37727690e-01
4.27585691e-01 5.31920493e-01 8.30503345e-01 -9.05690670e-01
-3.38889986e-01 -7.84972608e-01 -6.12458408e-01 2.61948973e-01
3.80785577e-02 -8.46895099e-01 -6.44315004e-01 -2.19687521e-01]
|
[7.983331203460693, -3.2629058361053467]
|
c108bc57-33d6-45b4-80d2-2708edca0fd7
|
measuring-intersectional-biases-in-historical
|
2305.12376
| null |
https://arxiv.org/abs/2305.12376v1
|
https://arxiv.org/pdf/2305.12376v1.pdf
|
Measuring Intersectional Biases in Historical Documents
|
Data-driven analyses of biases in historical texts can help illuminate the origin and development of biases prevailing in modern society. However, digitised historical documents pose a challenge for NLP practitioners as these corpora suffer from errors introduced by optical character recognition (OCR) and are written in an archaic language. In this paper, we investigate the continuities and transformations of bias in historical newspapers published in the Caribbean during the colonial era (18th to 19th centuries). Our analyses are performed along the axes of gender, race, and their intersection. We examine these biases by conducting a temporal study in which we measure the development of lexical associations using distributional semantics models and word embeddings. Further, we evaluate the effectiveness of techniques designed to process OCR-generated data and assess their stability when trained on and applied to the noisy historical newspapers. We find that there is a trade-off between the stability of the word embeddings and their compatibility with the historical dataset. We provide evidence that gender and racial biases are interdependent, and their intersection triggers distinct effects. These findings align with the theory of intersectionality, which stresses that biases affecting people with multiple marginalised identities compound to more than the sum of their constituents.
|
['Isabelle Augenstein', 'Natacha Klein Käfer', 'Natália da Silva Perez', 'Thea Rolskov', 'Karolina Stańczak', 'Nadav Borenstein']
|
2023-05-21
| null | null | null | null |
['optical-character-recognition']
|
['computer-vision']
|
[ 5.02354838e-02 -2.42632985e-01 -3.86755526e-01 -4.42845613e-01
-3.39469343e-01 -9.08115923e-01 1.36284912e+00 7.19222009e-01
-1.00308311e+00 4.20795351e-01 1.24085093e+00 -4.70087141e-01
-1.33325443e-01 -7.86248386e-01 -4.07137305e-01 -4.69090521e-01
1.86969087e-01 2.29080230e-01 -2.93655276e-01 -5.88693380e-01
6.91145062e-01 2.58554399e-01 -1.28148496e+00 -4.26454470e-02
8.61753404e-01 1.69079453e-01 -2.96430349e-01 4.73535746e-01
-2.59657115e-01 4.37604517e-01 -7.49926686e-01 -8.90764654e-01
2.09611608e-03 -1.90295815e-01 -4.84827191e-01 -2.95742750e-01
9.20146406e-01 -5.47677167e-02 -3.96753579e-01 1.12525773e+00
6.87578022e-01 -1.13577895e-01 7.49269009e-01 -5.19364417e-01
-1.40860033e+00 7.04567015e-01 -3.51704657e-01 7.39278674e-01
2.02526554e-01 -1.20751336e-01 1.17184591e+00 -9.56244051e-01
1.01079726e+00 1.58685207e+00 6.10481262e-01 4.14926887e-01
-1.26417255e+00 -5.03795385e-01 2.63810426e-01 2.75539365e-02
-1.15433514e+00 -8.27937126e-01 6.06596291e-01 -8.62304866e-01
4.07608777e-01 2.79552191e-01 8.29213798e-01 1.41592681e+00
2.50519276e-01 1.57364979e-01 1.32435691e+00 -5.06299019e-01
-1.51614435e-02 3.26791972e-01 3.45452815e-01 1.66888282e-01
5.94575763e-01 2.33362406e-01 -7.21548378e-01 -4.15951699e-01
2.44053468e-01 -2.71538645e-01 -1.79249793e-01 3.00599307e-01
-1.24931300e+00 1.08831871e+00 1.66573673e-01 5.95737517e-01
6.06584363e-02 -4.55905609e-02 3.67539406e-01 1.87727198e-01
8.96830380e-01 7.69633949e-01 -2.93255746e-02 -2.28741989e-01
-9.26462054e-01 3.28281701e-01 4.67257380e-01 2.49515131e-01
1.65459648e-01 -1.48103565e-01 1.24686845e-01 1.03287899e+00
2.72436529e-01 7.26195991e-01 5.74478030e-01 -4.15424824e-01
3.34687144e-01 2.68992603e-01 9.69865993e-02 -1.56302357e+00
-2.41272524e-01 -2.35670596e-01 -1.59681812e-01 -1.31923586e-01
6.64522767e-01 -8.63844231e-02 -8.62688541e-01 1.89857364e+00
2.41306946e-01 -7.31678545e-01 -2.07195967e-01 9.22556043e-01
3.60045761e-01 6.48410380e-01 5.74908555e-01 6.06480278e-02
1.41826165e+00 -1.22398570e-01 -7.16762424e-01 -5.62077284e-01
4.37469780e-01 -9.71846521e-01 1.18845725e+00 -6.57828851e-03
-7.41725862e-01 -2.36348391e-01 -1.13544369e+00 -3.57270777e-01
-6.78878963e-01 -3.21963936e-01 3.91794890e-01 1.04251826e+00
-6.40111506e-01 5.57957947e-01 -4.26126003e-01 -6.00731730e-01
3.84471864e-01 -2.27195427e-01 -5.29430322e-02 5.01840338e-02
-1.14129293e+00 1.26843977e+00 -1.14852473e-01 -1.26558740e-03
-3.04005146e-01 -6.40005589e-01 -8.06321800e-01 -6.54165208e-01
-8.92020091e-02 9.57400650e-02 6.49773240e-01 -1.21485519e+00
-6.32619143e-01 1.17122185e+00 -1.05872780e-01 4.48678359e-02
4.36936468e-01 -4.97296840e-01 -1.03169584e+00 -1.78864226e-01
3.28714311e-01 2.23734990e-01 5.21547377e-01 -1.21557641e+00
-4.47719842e-01 -6.32575572e-01 -1.83364317e-01 4.75208424e-02
-6.86267674e-01 4.74101692e-01 2.63772994e-01 -1.11750615e+00
-3.50946598e-02 -7.73807704e-01 7.67938420e-02 -1.06160976e-01
1.06023349e-01 -3.19857821e-02 2.74964571e-01 -1.05871856e+00
1.51506460e+00 -2.34878731e+00 3.32819745e-02 3.98171782e-01
9.80332568e-02 1.00382091e-02 1.29656732e-01 6.00987315e-01
1.39025584e-01 5.62496781e-01 -3.04506421e-01 2.06987336e-02
1.57249212e-01 3.36649954e-01 -5.36502719e-01 9.45225477e-01
6.14943691e-02 6.11102045e-01 -1.05564499e+00 -4.32341576e-01
-2.26186827e-01 5.51190257e-01 -3.71103793e-01 -6.13032877e-01
9.37365890e-02 -1.24493435e-01 5.70599623e-02 7.67270446e-01
4.31617469e-01 4.45231676e-01 5.28471887e-01 3.05687133e-02
-7.02029288e-01 6.19792700e-01 -4.93520617e-01 1.15720892e+00
-1.45247415e-01 1.23053885e+00 -6.73723519e-02 -2.52059489e-01
9.60075796e-01 -1.22889623e-01 -3.09433222e-01 -1.03784728e+00
3.13538283e-01 4.01223481e-01 4.12360907e-01 -4.99595791e-01
1.19086790e+00 -7.01579094e-01 -3.14237207e-01 5.47884464e-01
-2.69643575e-01 -3.08146868e-02 2.85963476e-01 1.21627726e-01
4.78256285e-01 -1.63996279e-01 6.22595847e-02 -7.83159912e-01
-8.18059687e-03 2.52570361e-01 5.56241095e-01 5.14905035e-01
-1.78269655e-01 5.41864514e-01 5.37994027e-01 -4.90355521e-01
-1.44434714e+00 -1.14590633e+00 -8.74522030e-01 1.19634664e+00
-1.62545836e-03 -2.14525759e-01 -3.40097368e-01 -2.90194452e-01
2.87737846e-01 1.10963202e+00 -1.03550112e+00 -1.87489003e-01
-6.60777450e-01 -1.31687486e+00 6.84754133e-01 2.39825711e-01
-1.56573191e-01 -5.83010554e-01 -7.81688154e-01 6.17918968e-02
-6.59184307e-02 -8.07670116e-01 -3.17789465e-01 -3.53737324e-01
-4.71074045e-01 -9.55336809e-01 -5.25217175e-01 -4.61055964e-01
6.63869977e-01 -1.79870769e-01 9.58266616e-01 1.05443574e-01
-2.35172153e-01 3.31826657e-01 -4.13064897e-01 -6.57526016e-01
-5.95777512e-01 4.37270775e-02 1.11742280e-01 -1.42026484e-01
6.82543635e-01 -2.92726636e-01 -3.38504970e-01 -1.04725331e-01
-1.08475006e+00 -5.28504312e-01 3.20614010e-01 4.50718760e-01
-1.57729700e-01 -4.01160032e-01 4.72480059e-01 -9.37081754e-01
8.55611920e-01 -8.37579072e-01 -1.99718922e-01 2.56062418e-01
-6.42841280e-01 -9.73416418e-02 1.12082891e-01 -5.50412953e-01
-1.07885921e+00 -9.06566322e-01 9.10700634e-02 6.38197899e-01
2.07928166e-01 5.57592869e-01 2.90221602e-01 3.28813076e-01
9.71172094e-01 -2.36519724e-01 -1.30665153e-01 -3.42195898e-01
4.44866180e-01 9.55730498e-01 4.56506401e-01 -7.77298868e-01
6.64614320e-01 7.40425825e-01 -6.49764717e-01 -1.24899161e+00
-7.79380918e-01 3.83595414e-02 -4.73095119e-01 -4.02295917e-01
8.41585755e-01 -9.90528286e-01 9.69816074e-02 3.47168088e-01
-8.28024507e-01 -4.64094430e-02 -1.43162087e-01 6.09686255e-01
3.34249198e-01 1.90854296e-01 -5.01991391e-01 -8.01344693e-01
-1.31062558e-02 -5.73232770e-01 4.26209420e-01 3.70668173e-02
-8.39589059e-01 -1.07121241e+00 5.27661324e-01 1.28861368e-01
2.80081242e-01 6.36390090e-01 1.25488901e+00 -7.73614049e-01
2.25717112e-01 6.46735122e-03 -8.76987875e-02 1.36151165e-01
2.58287966e-01 4.28691536e-01 -8.01534057e-01 -9.59930792e-02
-4.61541526e-02 -5.34067526e-02 7.30708063e-01 -8.39080811e-02
1.71574280e-01 -3.23224336e-01 -1.26884848e-01 1.64462388e-01
1.61221218e+00 1.49722219e-01 4.90206033e-01 7.69256353e-01
5.72385788e-01 1.00870705e+00 1.25501841e-01 1.94166109e-01
4.60458785e-01 1.22036666e-01 -2.78780103e-01 2.92214692e-01
9.66340117e-03 -2.82895476e-01 5.30932963e-01 9.20821488e-01
1.40234694e-01 -4.16324735e-02 -1.36105525e+00 1.26858640e+00
-1.21671736e+00 -9.40810919e-01 -3.09405655e-01 2.17820287e+00
1.03020000e+00 2.44163737e-01 8.11383650e-02 3.16729918e-02
7.73519993e-01 7.13929117e-01 5.49721196e-02 -8.86503637e-01
-4.98915076e-01 2.71780550e-01 6.21678054e-01 6.82129264e-01
-6.75137639e-01 7.65824378e-01 7.37943363e+00 2.83299685e-01
-1.11561990e+00 4.92312573e-02 5.26109278e-01 -4.98099238e-01
-9.56560075e-01 -6.61155954e-02 -4.48827416e-01 6.40032768e-01
8.89290929e-01 -2.15803310e-01 1.32075116e-01 3.32371086e-01
-2.89778020e-02 -2.31660530e-01 -8.28397572e-01 3.56280655e-01
4.10376579e-01 -1.01875818e+00 2.69806813e-02 2.01145887e-01
6.94053292e-01 9.16082487e-02 4.24489021e-01 -1.98363096e-01
4.03641999e-01 -1.01935840e+00 1.39477921e+00 2.99970180e-01
6.89794242e-01 -7.42372811e-01 5.34838915e-01 -3.85015190e-01
-2.19277248e-01 -2.96329618e-01 -3.82919550e-01 -3.79293829e-01
2.62917787e-01 8.18382204e-01 -3.86432737e-01 -1.24015950e-01
5.42040706e-01 4.19331402e-01 -7.96261549e-01 2.93225437e-01
-2.73621708e-01 5.25891244e-01 -1.17033467e-01 -1.69611424e-01
2.86367923e-01 -3.19306016e-01 6.10121787e-01 1.34811103e+00
2.58671165e-01 -4.90816645e-02 -6.28553629e-01 6.85721099e-01
5.55722788e-02 2.29811355e-01 -5.58206677e-01 -6.61134422e-01
8.07337940e-01 9.12224352e-01 -7.96747625e-01 -2.47454301e-01
-5.12784719e-01 4.52814400e-01 3.64344597e-01 2.49088630e-01
-5.63056707e-01 -2.47438043e-01 9.56002235e-01 3.53373945e-01
-3.81529890e-02 -6.18070543e-01 -5.83555162e-01 -9.39492226e-01
-8.65198020e-03 -1.00865769e+00 2.36620590e-01 -2.03398257e-01
-1.36193180e+00 1.84890404e-01 -3.09419744e-02 -1.91239923e-01
4.19721663e-01 -5.14052510e-01 -2.33063459e-01 6.92905605e-01
-1.15830588e+00 -8.58549535e-01 2.74666905e-01 5.76544143e-02
1.12854004e-01 3.33151668e-02 5.55654645e-01 2.91611165e-01
-5.62070131e-01 3.51140887e-01 4.23953086e-01 4.05032694e-01
9.25334334e-01 -1.06151056e+00 6.74686849e-01 8.79001677e-01
1.97209641e-01 8.39058578e-01 7.95783579e-01 -9.80914354e-01
-1.00115180e+00 -5.63787460e-01 1.53064656e+00 -9.04529452e-01
1.07535386e+00 -5.37227690e-01 -6.27191126e-01 6.77362084e-01
3.54710519e-01 -6.61782980e-01 1.12870109e+00 5.68249941e-01
-7.22396255e-01 2.04434708e-01 -9.87639010e-01 9.62082446e-01
1.05919540e+00 -9.68780518e-01 -1.11323357e+00 -1.51286185e-01
4.35558677e-01 5.13983443e-02 -8.12894285e-01 1.41905760e-02
9.59213972e-01 -7.40162194e-01 6.16621912e-01 -5.11445045e-01
8.70019019e-01 6.85925186e-02 -4.13758993e-01 -1.37624407e+00
-4.08480138e-01 -2.49759004e-01 7.45317459e-01 1.52701759e+00
5.74490845e-01 -6.51727438e-01 9.05249715e-02 7.22474992e-01
1.13956422e-01 -2.35431120e-01 -8.53230178e-01 -2.43795648e-01
5.50689936e-01 -2.36590058e-01 5.63645899e-01 1.48962379e+00
4.83712293e-02 1.69881716e-01 8.37934241e-02 8.98111537e-02
4.85970080e-01 -2.28736073e-01 3.04258078e-01 -1.17470336e+00
4.08264935e-01 -5.51706970e-01 -2.93384254e-01 -1.10983439e-02
1.94823612e-02 -8.34583342e-01 -1.99293658e-01 -1.15718627e+00
-4.23340350e-02 -5.49842477e-01 1.27650872e-02 -1.96476385e-01
-2.60370731e-01 1.65775627e-01 2.98142165e-01 2.70034581e-01
1.83620945e-01 3.13593596e-01 7.24628210e-01 8.62981286e-03
-1.34259760e-01 -9.31037366e-01 -1.15184426e+00 6.97928965e-01
8.16608608e-01 -6.08992696e-01 7.80046582e-02 -9.72418487e-01
1.01678431e+00 -7.78810203e-01 4.60801333e-01 -7.51294792e-01
-1.02339968e-01 -2.61962652e-01 4.95948672e-01 -9.40702781e-02
8.08560625e-02 -5.11571586e-01 -2.91505381e-02 2.75390297e-01
-4.67685223e-01 4.81160402e-01 1.53124735e-01 2.83445477e-01
-2.61276728e-03 -1.55964822e-01 5.55171907e-01 -1.13019697e-01
-3.99862021e-01 -3.53364468e-01 -7.11902261e-01 5.72765589e-01
6.19702756e-01 -6.42827526e-02 -8.17307234e-01 9.53422785e-02
-3.03936154e-01 -1.03214137e-01 8.79258752e-01 6.99347794e-01
2.54606634e-01 -1.42160583e+00 -9.21378911e-01 -2.66983621e-02
8.09840709e-02 -7.04522371e-01 -1.85827956e-01 5.72779179e-01
-6.40852273e-01 -4.94973101e-02 -3.04031610e-01 1.29942566e-01
-9.38840926e-01 4.96648282e-01 2.74068397e-02 5.32739878e-01
-2.10282087e-01 8.08463931e-01 -6.23633750e-02 -3.24018329e-01
-2.41997197e-01 -7.81597272e-02 -2.10143298e-01 1.05656815e+00
4.48642135e-01 6.72912002e-01 -4.01287645e-01 -1.00802910e+00
-4.53437388e-01 5.40754020e-01 -5.93623705e-02 -6.38045967e-01
1.39698017e+00 -2.55523324e-01 -4.08996284e-01 8.37040961e-01
1.12561905e+00 8.02869618e-01 -6.26912534e-01 -1.00335374e-01
2.44583338e-01 -7.95046151e-01 -7.58291110e-02 -6.52789354e-01
-5.58857024e-01 4.26128536e-01 3.61478865e-01 2.25996837e-01
3.13129216e-01 1.35892741e-02 6.18241310e-01 -2.75161445e-01
-2.61272341e-01 -1.76816237e+00 -3.93207103e-01 2.60716707e-01
8.45658839e-01 -8.45659077e-01 4.89928097e-01 5.29150926e-02
-5.47409773e-01 9.19118047e-01 1.85368046e-01 -1.20331429e-01
5.50823808e-01 -2.89017092e-02 3.42541695e-01 -4.66172993e-01
-4.16267097e-01 -9.06255990e-02 1.37968302e-01 4.46342289e-01
7.11734414e-01 2.52645612e-01 -1.07351673e+00 2.74937212e-01
-8.46566916e-01 -6.49587810e-01 6.69867218e-01 8.70307267e-01
-3.31733763e-01 -1.00779438e+00 -7.34409511e-01 2.63529092e-01
-8.14657271e-01 -3.64740878e-01 -8.69317651e-01 9.64300036e-01
4.71071005e-01 8.42312455e-01 6.18020773e-01 -2.61147112e-01
1.09744295e-01 3.30198169e-01 4.72853243e-01 -2.92741418e-01
-7.66039968e-01 -2.48702571e-01 6.08345211e-01 6.41357675e-02
-5.56528032e-01 -1.11868560e+00 -6.78422272e-01 -6.53740764e-01
1.22630503e-03 1.21956497e-01 8.50060523e-01 8.04704130e-01
9.06264484e-02 1.23516239e-01 2.71873981e-01 -2.79598087e-01
-3.98286194e-01 -8.40447545e-01 -4.18217361e-01 4.36913520e-01
5.07722199e-01 -3.51646483e-01 -6.08870447e-01 -1.17585637e-01]
|
[9.33457088470459, 10.146484375]
|
207ceb1e-af9f-4976-a9a3-4d2138edfde3
|
detecting-word-level-adversarial-text-attacks
| null | null |
https://aclanthology.org/2022.repl4nlp-1.16
|
https://aclanthology.org/2022.repl4nlp-1.16.pdf
|
Detecting Word-Level Adversarial Text Attacks via SHapley Additive exPlanations
|
State-of-the-art machine learning models are prone to adversarial attacks”:" Maliciously crafted inputs to fool the model into making a wrong prediction, often with high confidence. While defense strategies have been extensively explored in the computer vision domain, research in natural language processing still lacks techniques to make models resilient to adversarial text inputs. We adapt a technique from computer vision to detect word-level attacks targeting text classifiers. This method relies on training an adversarial detector leveraging Shapley additive explanations and outperforms the current state-of-the-art on two benchmarks. Furthermore, we prove the detector requires only a low amount of training samples and, in some cases, generalizes to different datasets without needing to retrain.
|
['Georg Groh', 'Marc Alexander Kühn', 'Lukas Huber', 'Edoardo Mosca']
| null | null | null | null |
repl4nlp-acl-2022-5
|
['adversarial-text']
|
['adversarial']
|
[ 4.87077683e-01 4.18213278e-01 -9.33464840e-02 -2.22271487e-01
-8.16858470e-01 -1.12466872e+00 8.76589894e-01 2.89336801e-01
-2.71889418e-01 4.75338668e-01 -1.06572755e-01 -8.02890718e-01
4.21948969e-01 -7.63891995e-01 -1.16497183e+00 -4.70815331e-01
3.05789918e-01 4.37747061e-01 2.70125687e-01 -2.05306143e-01
2.97006547e-01 4.62321192e-01 -8.69232059e-01 5.67570627e-01
5.40056229e-01 5.26439726e-01 -6.31956518e-01 9.75394905e-01
2.75825281e-02 9.56945956e-01 -7.15417087e-01 -1.14836180e+00
6.03295922e-01 -3.44537675e-01 -8.60406935e-01 -1.15499496e-01
6.28549159e-01 -3.78096104e-01 -7.30218768e-01 1.59426093e+00
9.43928957e-02 -1.68177068e-01 6.42783105e-01 -1.37839592e+00
-1.13457930e+00 9.51579630e-01 -3.40368062e-01 1.92129031e-01
2.17110023e-01 4.83487874e-01 1.00330937e+00 -6.41825020e-01
4.83373314e-01 1.34938085e+00 6.55656397e-01 1.17951322e+00
-1.40078497e+00 -7.81490028e-01 2.57667631e-01 6.25832304e-02
-8.70791972e-01 -6.68154284e-02 7.43003368e-01 -2.94386268e-01
7.95305610e-01 5.53340971e-01 6.78223297e-02 1.90466785e+00
4.48009998e-01 9.06757116e-01 1.05891371e+00 -4.64961797e-01
3.91597748e-01 4.71310258e-01 2.14167349e-02 6.56637311e-01
5.32780528e-01 3.47478122e-01 -7.27270916e-02 -4.80657697e-01
7.30018765e-02 4.03891243e-02 -1.31828278e-01 -2.93052882e-01
-7.36401379e-01 1.40202498e+00 4.86658424e-01 3.02607685e-01
-8.80845785e-02 3.28801423e-01 6.61531150e-01 4.35881078e-01
5.26750863e-01 7.96799719e-01 -4.55947667e-01 2.40199745e-01
-8.23783159e-01 2.11293325e-01 9.64736760e-01 6.34496689e-01
1.67440891e-01 1.91495419e-01 1.85905714e-02 1.66027382e-01
9.45317298e-02 5.86099029e-01 4.93657291e-01 -5.27265012e-01
4.20498133e-01 4.95861739e-01 -1.98003218e-01 -9.11487937e-01
-6.80800453e-02 -4.28837776e-01 -7.71638334e-01 5.60503006e-01
4.63016808e-01 -2.23809481e-01 -8.85776460e-01 1.61196721e+00
9.56279114e-02 2.79019386e-01 2.10157916e-01 7.05213726e-01
1.83036566e-01 5.34200490e-01 8.96520689e-02 1.81147590e-01
1.13740265e+00 -9.04891253e-01 -3.26913953e-01 -6.87299430e-01
5.67029297e-01 -5.70189238e-01 1.13304436e+00 6.26308560e-01
-9.39573884e-01 -4.19962257e-02 -1.39207661e+00 1.39959365e-01
-6.20076597e-01 -7.79048681e-01 5.59569895e-01 1.17345095e+00
-5.07028043e-01 6.99323654e-01 -6.11837685e-01 1.91438664e-03
6.74446404e-01 2.63681263e-01 -5.37603080e-01 -1.11786604e-01
-1.49822962e+00 1.25696707e+00 3.19587886e-01 -5.44715226e-01
-1.36152864e+00 -7.71652937e-01 -6.93041861e-01 5.65197617e-02
4.92346734e-01 -7.25740254e-01 1.41083062e+00 -1.50231910e+00
-1.21803641e+00 9.61464405e-01 2.30858296e-01 -1.16973746e+00
1.13985670e+00 -4.53231514e-01 -3.33340347e-01 2.03590259e-01
-1.82478160e-01 1.69168994e-01 1.37986922e+00 -1.23938763e+00
-2.61114150e-01 -3.80961567e-01 3.42066020e-01 -4.06081200e-01
-6.84823811e-01 1.61412939e-01 2.19452828e-01 -9.85969722e-01
-3.86677086e-01 -1.07908607e+00 -5.39606273e-01 1.81838905e-03
-7.30873227e-01 3.93670425e-02 1.07704127e+00 -1.90134957e-01
7.32062519e-01 -1.92770684e+00 -1.36770234e-01 9.57609490e-02
1.67498797e-01 7.17461705e-01 -1.13385066e-01 3.44099313e-01
-2.14404762e-01 7.10018396e-01 -3.46047401e-01 -3.75128657e-01
2.99875051e-01 2.06482917e-01 -1.06817603e+00 7.70797491e-01
2.05571428e-01 7.48673856e-01 -1.00230610e+00 -1.47131771e-01
2.27561057e-01 2.60269612e-01 -7.25954652e-01 1.36125833e-01
-6.28438175e-01 1.59082964e-01 -3.48110706e-01 3.52827281e-01
7.19898283e-01 -1.16376780e-01 2.70541180e-02 4.79136974e-01
6.71591401e-01 2.33201191e-01 -7.74633408e-01 1.16405141e+00
-3.27115059e-01 7.20470786e-01 -1.73043877e-01 -1.14335394e+00
5.76800585e-01 3.30289334e-01 -1.79801896e-01 -1.54838115e-01
4.28613335e-01 5.02421800e-03 -1.23749435e-01 -3.03187519e-01
1.54903024e-01 -1.92878842e-01 -3.29994529e-01 5.62463284e-01
-2.12366711e-02 -1.02042779e-01 -3.78202051e-01 4.60604876e-01
1.60889173e+00 -3.89770001e-01 2.03264192e-01 8.03678557e-02
7.13780582e-01 9.30216983e-02 2.97682881e-01 1.24111307e+00
-2.27982700e-01 2.60463864e-01 5.40641963e-01 -6.17085934e-01
-1.27494717e+00 -1.08652663e+00 1.87457830e-01 1.24549890e+00
-1.95962682e-01 -2.17377260e-01 -1.16445613e+00 -1.41363978e+00
3.61838251e-01 1.15929031e+00 -1.03785706e+00 -6.79747045e-01
-2.45538145e-01 -3.38755459e-01 1.27126586e+00 5.42632163e-01
2.51039296e-01 -1.01637864e+00 -3.68027389e-01 -3.96538526e-02
3.10056210e-01 -1.24605632e+00 -4.22808200e-01 2.52412319e-01
-5.86999059e-01 -1.30952740e+00 -3.01786631e-01 -3.62811506e-01
7.08439589e-01 1.04365014e-01 1.15365243e+00 4.23170120e-01
-3.48349154e-01 2.72322327e-01 -3.97343367e-01 -8.05088699e-01
-1.27244091e+00 1.98549293e-02 1.94399968e-01 -3.56475636e-02
5.86032629e-01 -2.83824027e-01 -2.44717732e-01 6.58449680e-02
-1.16239846e+00 -3.45853537e-01 5.51637709e-01 9.62305248e-01
1.14946425e-01 4.28024717e-02 5.37110984e-01 -1.38370824e+00
6.62737429e-01 -5.52845359e-01 -5.87013006e-01 1.09283261e-01
-6.92445576e-01 2.23383963e-01 1.60157847e+00 -9.46922123e-01
-5.77309608e-01 2.50372849e-02 -1.46059752e-01 -8.29593956e-01
-3.58339101e-01 4.49506799e-03 -3.29214424e-01 -2.30790347e-01
1.10697603e+00 2.59802788e-01 -1.97100237e-01 -3.06750219e-02
6.99425340e-01 4.13043916e-01 7.27030098e-01 -4.21994388e-01
1.40459764e+00 5.33188820e-01 2.04122812e-01 -3.88697684e-01
-1.03835547e+00 1.08657591e-01 -3.92959535e-01 2.23319381e-01
6.68857038e-01 -3.33682865e-01 -7.50617027e-01 3.44757259e-01
-1.21960008e+00 -9.89234224e-02 -1.69050619e-01 2.29108095e-01
-4.49146867e-01 4.62106824e-01 -7.39100814e-01 -8.07146847e-01
-5.75047076e-01 -8.97738159e-01 4.13891792e-01 -1.16361976e-01
-2.64598578e-01 -1.15649915e+00 -1.97253134e-02 3.45774204e-01
3.45680684e-01 3.00772697e-01 9.35861051e-01 -1.62093794e+00
-2.52924889e-01 -8.54458511e-01 3.05814117e-01 6.92392945e-01
-3.35999489e-01 2.50344630e-02 -1.21230137e+00 -2.08967239e-01
2.80903429e-01 -3.44450921e-01 1.01826191e+00 -1.98221922e-01
1.23973465e+00 -9.77998316e-01 -1.42914206e-01 4.10930485e-01
1.26778698e+00 -1.51928052e-01 4.61341351e-01 3.57462347e-01
5.95466733e-01 4.50072438e-01 1.84814468e-01 2.10420042e-01
-1.99228689e-01 2.71062553e-01 1.00847352e+00 1.08178511e-01
3.66658241e-01 -7.34164238e-01 4.73719150e-01 -1.02865167e-01
6.12095296e-01 -4.13525134e-01 -7.43142188e-01 3.38488489e-01
-1.76432180e+00 -1.23203480e+00 -9.18919221e-02 2.00824952e+00
7.64983416e-01 7.63628244e-01 -1.98769078e-01 4.05784607e-01
6.08244717e-01 4.88243662e-02 -9.00410354e-01 -9.78850603e-01
-1.12452395e-01 4.48389858e-01 1.00798237e+00 4.67141479e-01
-1.41797721e+00 1.33818555e+00 6.37704182e+00 7.20649004e-01
-1.04666734e+00 1.27397582e-01 4.75496769e-01 3.46215703e-02
-5.09063303e-01 2.34607741e-01 -4.50476199e-01 3.55544150e-01
9.95777667e-01 -3.40842545e-01 4.73580599e-01 1.30539346e+00
-1.09319091e-01 4.85516846e-01 -1.23844123e+00 3.34910810e-01
2.85472900e-01 -1.36447680e+00 3.03724736e-01 -3.81145924e-02
5.87675035e-01 -8.86527076e-02 3.21972817e-01 3.84568065e-01
1.08543634e+00 -1.40301728e+00 7.14818060e-01 1.77699737e-02
1.59539133e-01 -9.30017769e-01 7.23409653e-01 8.38409603e-01
-4.13993746e-01 -2.62901247e-01 -5.53779125e-01 -7.73841962e-02
-2.57255316e-01 3.67532253e-01 -1.00341642e+00 1.79824650e-01
4.15705591e-01 1.35823563e-01 -8.11450839e-01 4.24239963e-01
-6.95312202e-01 1.04412818e+00 2.91018300e-02 -2.85446703e-01
5.90380967e-01 4.77386475e-01 7.91596472e-01 1.22434485e+00
-1.49837613e-01 -3.64839882e-02 3.12114537e-01 8.46396029e-01
-5.15113413e-01 -1.21528774e-01 -1.01785553e+00 -1.97966516e-01
4.66383040e-01 9.90735829e-01 -3.42875689e-01 -4.06642765e-01
-2.43951425e-01 1.24405301e+00 2.52087802e-01 -1.05101084e-02
-9.11046207e-01 -1.92198083e-01 5.63326240e-01 1.21095711e-02
1.96766719e-01 -5.42341843e-02 -6.01530194e-01 -1.26814902e+00
-1.72009841e-01 -1.20466399e+00 4.78340149e-01 -5.14226258e-01
-1.71472502e+00 5.39999664e-01 -5.62884271e-01 -9.38444436e-01
-3.56056750e-01 -8.00115287e-01 -8.31321418e-01 6.71105683e-01
-9.71717536e-01 -1.17985499e+00 2.67318577e-01 8.24296951e-01
6.00651145e-01 -4.65834230e-01 1.24152875e+00 -2.58211255e-01
-4.14571792e-01 1.17863297e+00 -4.99617457e-02 6.44536912e-01
6.90020442e-01 -1.34306538e+00 1.07922304e+00 1.32007706e+00
4.00185734e-01 5.88035166e-01 1.32812476e+00 -5.29683948e-01
-1.52416146e+00 -1.02052283e+00 7.79573917e-01 -7.02842355e-01
1.05542409e+00 -5.80493093e-01 -1.22289133e+00 9.18254137e-01
2.51517117e-01 3.10811788e-01 6.73951864e-01 -1.10159196e-01
-1.26919937e+00 3.04004192e-01 -1.61888587e+00 8.30468357e-01
6.53474689e-01 -7.52451062e-01 -7.97000170e-01 5.22537470e-01
8.50682616e-01 -2.59385556e-01 -2.47308552e-01 1.28607109e-01
4.48191553e-01 -8.22616458e-01 9.03322220e-01 -1.51433194e+00
5.15781641e-01 5.12260050e-02 -1.84740841e-01 -1.35988724e+00
-1.40837267e-01 -8.00780892e-01 -3.17679852e-01 1.01749885e+00
4.66471881e-01 -8.06778073e-01 8.71044397e-01 7.17728436e-01
2.82502733e-02 -3.69198501e-01 -9.58928704e-01 -9.09465432e-01
7.88447142e-01 -7.91653097e-01 3.46971929e-01 1.24353814e+00
5.97604774e-02 2.34295785e-01 -7.36524403e-01 5.34776866e-01
7.71017373e-01 -3.62807393e-01 9.98776913e-01 -1.18177938e+00
-4.27702367e-01 -5.77619672e-01 -5.72076797e-01 -5.12298703e-01
8.08688223e-01 -1.09686100e+00 -1.92806050e-01 -8.46401095e-01
5.57536781e-02 3.41122113e-02 -1.61385939e-01 6.33005798e-01
-3.05804968e-01 2.72609115e-01 3.71887356e-01 -5.29986843e-02
-2.93114841e-01 1.73314407e-01 5.37979841e-01 -6.30317867e-01
4.05065536e-01 2.70722151e-01 -1.00189650e+00 1.22894168e+00
8.82244170e-01 -1.06480384e+00 -2.35680714e-01 -2.24035785e-01
2.52414048e-01 -3.07854205e-01 7.06470132e-01 -8.23176086e-01
2.46823102e-01 -2.94265330e-01 4.64545697e-01 -1.54695651e-02
-3.93846706e-02 -9.46931005e-01 -2.54049420e-01 8.23683918e-01
-8.79517615e-01 3.28739211e-02 1.29962474e-01 7.73071289e-01
3.22687954e-01 -5.90517998e-01 1.20037878e+00 -2.33727887e-01
-1.54349416e-01 3.45374882e-01 -5.31369865e-01 1.98868915e-01
1.34390068e+00 3.46887231e-01 -6.34400070e-01 -4.06520158e-01
-4.92864549e-01 -9.00382698e-02 6.40071929e-01 6.40049636e-01
5.05193293e-01 -9.08772647e-01 -8.08273673e-01 1.20491177e-01
-6.31513521e-02 -5.57614923e-01 2.31679436e-02 4.86278906e-02
-3.78131628e-01 2.68532306e-01 5.08293323e-02 -1.00368276e-01
-1.33820200e+00 1.33133554e+00 4.09623146e-01 -4.90682244e-01
-6.97479486e-01 8.61494780e-01 -7.98810497e-02 -2.41605431e-01
2.78210729e-01 1.60683692e-01 3.25889558e-01 -4.31772053e-01
7.07070649e-01 6.44507483e-02 1.26748562e-01 -3.62563133e-01
-4.23906952e-01 -1.20611325e-01 -4.56318557e-01 -1.91350982e-01
7.37219572e-01 3.99144948e-01 3.41928363e-01 1.46446347e-01
1.12274909e+00 2.12865412e-01 -9.38089550e-01 -3.45850796e-01
1.52724996e-01 -5.82965791e-01 -1.46145999e-01 -9.04542804e-01
-7.23972201e-01 1.22925687e+00 4.35864747e-01 6.43695772e-01
8.05470824e-01 -1.93254128e-01 8.57222795e-01 6.97610557e-01
2.16719195e-01 -8.29064608e-01 3.22195709e-01 6.53000772e-01
7.43984282e-01 -1.27172112e+00 -8.61158222e-02 -1.01400942e-01
-9.20080781e-01 1.05497921e+00 4.10466045e-01 -5.84449351e-01
6.19420409e-01 5.10223448e-01 2.02242374e-01 1.90116435e-01
-9.60477054e-01 3.87450397e-01 1.75944939e-01 5.64708114e-01
2.04668511e-02 1.49370447e-01 -8.75811428e-02 7.13860691e-01
-3.33626032e-01 -3.09130818e-01 7.96317637e-01 7.86612153e-01
-4.17085618e-01 -1.18399012e+00 -4.79711652e-01 2.56079078e-01
-1.31770337e+00 -3.67944002e-01 -9.22797441e-01 6.44282579e-01
-1.48318157e-01 1.25532448e+00 -3.45816791e-01 -7.01031744e-01
2.56482571e-01 2.51451701e-01 8.51546302e-02 -5.41620374e-01
-1.12010276e+00 -5.52563190e-01 -1.43344432e-01 -4.17276502e-01
1.14091501e-01 -3.50539804e-01 -1.14503801e+00 -7.23177373e-01
-2.99802065e-01 -2.37361938e-02 7.17335403e-01 1.06785393e+00
2.12883845e-01 2.66814172e-01 9.35172558e-01 -5.39482892e-01
-1.35915256e+00 -8.76639843e-01 -2.60478675e-01 5.25705218e-01
4.34694171e-01 -2.38772497e-01 -8.56934190e-01 -7.13570565e-02]
|
[5.877494812011719, 7.952759265899658]
|
ecdab4cc-d6d4-43fa-bc9b-d2ea64ce1bbb
|
does-entity-abstraction-help-generative-1
|
2201.01787
| null |
https://arxiv.org/abs/2201.01787v2
|
https://arxiv.org/pdf/2201.01787v2.pdf
|
Does Entity Abstraction Help Generative Transformers Reason?
|
We study the utility of incorporating entity type abstractions into pre-trained Transformers and test these methods on four NLP tasks requiring different forms of logical reasoning: (1) compositional language understanding with text-based relational reasoning (CLUTRR), (2) abductive reasoning (ProofWriter), (3) multi-hop question answering (HotpotQA), and (4) conversational question answering (CoQA). We propose and empirically explore three ways to add such abstraction: (i) as additional input embeddings, (ii) as a separate sequence to encode, and (iii) as an auxiliary prediction task for the model. Overall, our analysis demonstrates that models with abstract entity knowledge performs better than without it. The best abstraction aware models achieved an overall accuracy of 88.8% and 91.8% compared to the baseline model achieving 62.9% and 89.8% on CLUTRR and ProofWriter respectively. However, for HotpotQA and CoQA, we find that F1 scores improve by only 0.5% on average. Our results suggest that the benefit of explicit abstraction is significant in formally defined logical reasoning settings requiring many reasoning hops, but point to the notion that it is less beneficial for NLP tasks having less formal logical structure.
|
['Christopher Pal', 'Siva Reddy', 'Nicolas Gontier']
|
2022-01-05
|
does-entity-abstraction-help-generative
|
https://openreview.net/forum?id=rSI-tyrv-ni
|
https://openreview.net/pdf?id=rSI-tyrv-ni
| null |
['multi-hop-question-answering', 'relational-reasoning']
|
['knowledge-base', 'natural-language-processing']
|
[-1.36187330e-01 8.90514731e-01 -8.85593519e-02 -2.71355122e-01
-9.91454482e-01 -7.42701888e-01 8.37408245e-01 4.04998839e-01
-2.61069566e-01 8.30504954e-01 6.12082124e-01 -1.01219261e+00
-2.28892371e-01 -1.10731423e+00 -8.26175630e-01 8.15971196e-02
1.14045598e-01 8.69074464e-01 2.41183430e-01 -5.10721684e-01
2.09866449e-01 3.27837795e-01 -1.14137864e+00 5.81657708e-01
1.16323280e+00 8.03349197e-01 -3.85731578e-01 8.06823671e-01
-4.76983756e-01 1.91824722e+00 -6.77166879e-01 -9.51886773e-01
3.71140167e-02 -4.84234020e-02 -1.50982904e+00 -6.04345024e-01
5.52448213e-01 -5.03190935e-01 -3.02702695e-01 6.18626595e-01
-5.05169071e-02 2.63730697e-02 5.91138363e-01 -1.41931272e+00
-1.13643754e+00 1.08834445e+00 -2.12112576e-01 -3.83157507e-02
7.54482150e-01 3.09847742e-01 1.60942066e+00 -7.34847486e-01
6.53425753e-01 1.40614915e+00 7.94450819e-01 5.15652955e-01
-1.37530637e+00 -4.57394481e-01 -1.66511722e-02 3.60459685e-01
-1.18146288e+00 -3.66520166e-01 3.90643626e-01 -2.97499627e-01
1.66253471e+00 3.42765749e-01 2.56049514e-01 6.64860129e-01
6.07304983e-02 6.86424375e-01 1.00506580e+00 -4.12384689e-01
1.00100420e-01 1.85716733e-01 6.66640520e-01 9.73644376e-01
3.64749849e-01 -4.94355172e-01 -4.15551633e-01 -4.73210096e-01
3.15735638e-01 -3.02368850e-01 -1.29705697e-01 -2.88811810e-02
-1.03212130e+00 9.82041657e-01 3.67038071e-01 1.86095551e-01
-4.41119760e-01 3.43558729e-01 6.15249455e-01 7.33365476e-01
1.93153828e-01 1.16252124e+00 -8.76796603e-01 -2.46662453e-01
-5.09014547e-01 5.71167767e-01 1.51444638e+00 9.08147573e-01
6.54024899e-01 -1.52276486e-01 -1.03440732e-01 8.78542423e-01
1.38669923e-01 3.25269401e-01 1.77786767e-01 -1.43059444e+00
8.42953086e-01 8.88338625e-01 1.82965145e-01 -7.06399441e-01
-1.64103970e-01 1.81379933e-02 -1.63356379e-01 -1.49912357e-01
6.41341090e-01 -1.08649187e-01 -6.57350838e-01 1.86172974e+00
3.52070332e-02 -3.21291208e-01 6.57033741e-01 5.46913385e-01
9.98307526e-01 9.36920524e-01 3.08843017e-01 3.47335458e-01
1.77214527e+00 -1.13931024e+00 -5.86742342e-01 -1.77284986e-01
1.05961299e+00 -3.68910342e-01 1.44444489e+00 3.31787825e-01
-1.38011718e+00 -1.88871980e-01 -9.27936256e-01 -8.74321699e-01
-6.01889789e-01 -1.28951177e-01 1.01785064e+00 4.29622024e-01
-1.21674478e+00 4.95782271e-02 -5.04319012e-01 -3.62773031e-01
2.09830806e-01 8.78727809e-02 -3.90359133e-01 -4.46819007e-01
-1.53006709e+00 1.22812963e+00 2.60810852e-01 -1.96171597e-01
-5.49411774e-01 -1.17428315e+00 -1.06715751e+00 4.97577161e-01
4.78383780e-01 -8.08236420e-01 1.43492925e+00 -3.24786037e-01
-1.43249762e+00 7.55524695e-01 -2.03685090e-01 -7.65798926e-01
2.84511060e-01 -5.14512300e-01 -1.57294333e-01 2.85170078e-01
1.84031382e-01 6.12078965e-01 5.94702139e-02 -1.00793171e+00
-4.55965102e-01 -6.33389503e-02 1.09529245e+00 1.95025504e-01
-4.61556250e-03 -3.43602896e-02 -4.26132083e-02 -1.67415008e-01
-2.10358888e-01 -5.58076143e-01 3.46900851e-01 7.56863877e-02
-2.57385135e-01 -9.01328206e-01 6.58068120e-01 -1.10345435e+00
9.61219728e-01 -1.69054687e+00 1.32903516e-01 -7.19908625e-02
4.18492049e-01 2.96212643e-01 -1.87589675e-01 8.14108908e-01
8.06333646e-02 4.30834293e-01 -1.58067688e-01 1.28338128e-01
3.60317439e-01 4.13006991e-01 -8.12924564e-01 -3.85977000e-01
6.16890848e-01 1.32427526e+00 -8.62196505e-01 -4.82420832e-01
-3.79259661e-02 3.62165533e-02 -1.02358770e+00 1.35657832e-01
-8.16790044e-01 -2.46793613e-01 -2.82288820e-01 7.01509774e-01
3.54113340e-01 -4.45695996e-01 3.91882271e-01 -1.49505883e-01
1.86064944e-01 1.20586717e+00 -7.42998719e-01 1.32269633e+00
-8.84361267e-01 7.94937372e-01 -4.08184454e-02 -7.73144901e-01
8.04727018e-01 3.91907573e-01 -5.46374619e-02 -6.69322729e-01
-3.57890129e-01 1.59739226e-01 2.42765471e-01 -6.59576297e-01
6.35766804e-01 -3.38533252e-01 -1.24122374e-01 7.09012210e-01
4.82889898e-02 -3.73697400e-01 2.47845888e-01 6.32785559e-01
1.63150716e+00 -8.37239623e-02 2.54758060e-01 -2.91349649e-01
7.01002538e-01 2.72638977e-01 2.68944502e-01 8.05826545e-01
1.01602986e-01 -7.41456002e-02 9.84540403e-01 -4.16197747e-01
-9.53766465e-01 -1.15469205e+00 3.73130813e-02 1.08989060e+00
-7.06753805e-02 -7.06853330e-01 -4.44718361e-01 -7.85537839e-01
3.47086906e-01 1.41685057e+00 -2.99921036e-01 -1.96172059e-01
-8.08986664e-01 -1.61126584e-01 1.20048463e+00 9.35705960e-01
6.45998657e-01 -9.97169316e-01 -4.18854743e-01 1.01751210e-02
-6.44968927e-01 -1.34791017e+00 1.46100774e-01 8.63151252e-02
-7.37249970e-01 -1.03510606e+00 -8.05521831e-02 -3.79838079e-01
3.39092225e-01 -1.35824218e-01 1.34278882e+00 4.99427855e-01
2.77217209e-01 6.87287450e-01 -4.14910555e-01 -2.29822904e-01
-3.85182142e-01 1.66454419e-01 -2.68673092e-01 -6.00753903e-01
5.61781228e-01 -4.65303779e-01 -2.40066171e-01 -1.03051789e-01
-7.65813351e-01 -3.96485627e-03 5.46626329e-01 8.02925885e-01
-1.33432895e-01 -4.14960943e-02 4.84687448e-01 -1.10397124e+00
8.47557008e-01 -5.65426290e-01 -2.52646178e-01 5.38098752e-01
-5.44975162e-01 3.35830808e-01 9.20659482e-01 -1.98060229e-01
-1.29034603e+00 -8.80867243e-01 -2.61137217e-01 5.37436903e-02
7.48782605e-02 7.18237460e-01 1.26643181e-01 2.56639391e-01
8.26358497e-01 -3.35960612e-02 -7.14715868e-02 -1.34751245e-01
5.66370785e-01 6.13366961e-01 3.96902472e-01 -1.34864891e+00
9.16475236e-01 2.14146465e-01 -1.62399411e-01 -7.93056011e-01
-1.03022587e+00 -4.71564345e-02 -1.05530262e-01 4.11853760e-01
9.64359164e-01 -9.02956903e-01 -9.59243834e-01 -7.30447620e-02
-1.34587157e+00 -9.69188929e-01 -4.71278191e-01 2.07355410e-01
-4.08730090e-01 3.48918676e-01 -1.01911199e+00 -8.16189587e-01
-4.66480404e-01 -8.16670537e-01 9.72403824e-01 -2.33861501e-03
-7.32718170e-01 -1.07212448e+00 -1.64138243e-01 1.01935756e+00
3.82327706e-01 -7.15498924e-02 1.71786320e+00 -8.94459963e-01
-8.63219023e-01 -3.39548327e-02 -6.70131385e-01 4.41121697e-01
-7.61812553e-02 -1.13408700e-01 -7.74087250e-01 2.70635873e-01
-2.65391856e-01 -9.63286519e-01 3.81618470e-01 -2.69784540e-01
1.00929749e+00 -6.52219713e-01 -3.74025144e-02 1.24156050e-01
1.19935691e+00 1.62798882e-01 1.01481807e+00 1.68627903e-01
4.53656107e-01 5.03155887e-01 3.08452874e-01 -1.90908238e-01
1.10667527e+00 4.45351213e-01 -1.00426078e-02 3.76343638e-01
-1.86934620e-01 -6.51177466e-01 2.91801542e-01 7.57750988e-01
2.47839093e-02 -2.01784030e-01 -1.40979612e+00 6.22163236e-01
-1.79719579e+00 -1.00143862e+00 -5.50689325e-02 1.55907500e+00
1.17773807e+00 1.53268442e-01 -2.35991105e-01 1.20464884e-01
2.31736880e-02 6.49094582e-02 -3.72191578e-01 -1.15887570e+00
7.85600469e-02 7.01228261e-01 3.23174447e-02 8.89054596e-01
-4.10436153e-01 9.82527196e-01 6.13103199e+00 2.24259123e-01
-7.08296895e-01 3.31607759e-02 2.04905972e-01 2.65279710e-01
-7.31480181e-01 3.61435801e-01 -5.45226872e-01 1.66580677e-01
1.02203250e+00 -5.42166159e-02 7.34921277e-01 6.63342595e-01
-4.73446339e-01 -1.16793297e-01 -1.48304784e+00 4.08722281e-01
5.55860437e-03 -1.40290129e+00 4.14697260e-01 -2.20497996e-01
4.21236485e-01 -1.61047250e-01 -3.34752083e-01 1.04506052e+00
8.16572309e-01 -1.25266087e+00 4.98830557e-01 4.65063363e-01
5.04411161e-01 -3.73738915e-01 8.68117034e-01 3.93732935e-01
-9.34926867e-01 -2.83494145e-01 -2.52089590e-01 -4.72680807e-01
8.30282718e-02 2.08759740e-01 -8.78097653e-01 6.95970595e-01
5.01609147e-01 2.22737491e-01 -4.84024405e-01 2.63958126e-01
-6.76096439e-01 7.45422900e-01 -4.46687579e-01 -2.84058481e-01
3.77530754e-01 2.95048151e-02 2.68610895e-01 1.13867915e+00
-1.72428817e-01 4.58453327e-01 -2.95681268e-01 1.17624092e+00
-3.96005660e-01 -3.64485353e-01 -3.63425523e-01 -3.92206252e-01
7.87059069e-01 8.75612795e-01 -9.12477374e-02 -5.67436039e-01
-6.62921250e-01 7.05079734e-01 7.85340607e-01 5.09229958e-01
-8.51626635e-01 -7.09986389e-01 7.26288259e-01 1.03507839e-01
3.08291078e-01 -1.81497470e-01 -4.07325834e-01 -1.29580522e+00
2.38544345e-01 -1.08039391e+00 6.02387309e-01 -1.31642842e+00
-1.33703637e+00 1.91319451e-01 2.12117955e-01 -1.89989418e-01
-3.99565190e-01 -7.89514899e-01 -6.05923831e-01 1.05438209e+00
-1.70048201e+00 -1.22216332e+00 -2.20786646e-01 2.05574781e-01
3.23588610e-01 3.67511101e-02 9.81556177e-01 2.22471401e-01
-9.56681147e-02 6.88683808e-01 -4.97799695e-01 3.95125121e-01
3.54823887e-01 -1.52462089e+00 2.83536613e-01 5.36353588e-01
6.38353601e-02 1.27543294e+00 5.53429067e-01 -3.71500015e-01
-1.73493707e+00 -8.75988185e-01 1.51175404e+00 -1.02146602e+00
8.47253561e-01 -3.89168710e-01 -1.21604300e+00 1.34683502e+00
3.08297604e-01 -3.24662417e-01 5.85588694e-01 6.47649050e-01
-1.04878891e+00 -1.65774047e-01 -1.37176907e+00 9.11284029e-01
9.94937360e-01 -1.05553329e+00 -1.48244941e+00 2.70391792e-01
1.20523584e+00 -3.81942242e-01 -1.23481953e+00 2.82474548e-01
5.99183977e-01 -6.16734982e-01 9.40642059e-01 -9.50181842e-01
1.00707197e+00 -3.58400315e-01 -3.53055865e-01 -9.62959826e-01
-1.89412966e-01 -3.25454921e-01 -5.58270633e-01 1.05939007e+00
7.09489405e-01 -1.13642704e+00 4.74713922e-01 1.16708493e+00
-2.50616111e-02 -9.73640263e-01 -5.94091654e-01 -6.10931695e-01
5.49354672e-01 -5.00266790e-01 6.48531735e-01 1.03154659e+00
4.49894845e-01 7.56768644e-01 3.67405057e-01 2.89930671e-01
1.71880469e-01 1.92041084e-01 9.50595617e-01 -1.05244923e+00
-3.54005724e-01 -2.87761241e-01 -1.75063968e-01 -1.20083880e+00
6.65188074e-01 -1.10066962e+00 -3.74469310e-01 -2.16757488e+00
3.89707424e-02 -5.70626199e-01 2.73888130e-02 8.67014408e-01
-1.17200017e-01 -1.40853941e-01 1.91028535e-01 -2.50447188e-02
-4.35612351e-01 3.35956097e-01 8.78358901e-01 -2.29993805e-01
1.72723517e-01 -5.74669778e-01 -1.16800773e+00 5.13360620e-01
7.12998271e-01 -3.41903828e-02 -6.14345074e-01 -8.07025611e-01
6.02089703e-01 1.93796754e-01 6.15761638e-01 -6.05937958e-01
2.53446579e-01 -1.26277125e-02 -1.25951663e-01 -2.05111668e-01
4.51478511e-01 -6.04824662e-01 -2.98014671e-01 4.76199776e-01
-5.85099936e-01 1.78291380e-01 3.60872924e-01 8.18830580e-02
-2.70016760e-01 -3.10726881e-01 4.01011407e-01 -1.13827020e-01
-6.43594623e-01 -4.13975030e-01 -3.82601857e-01 6.18153811e-01
5.73435426e-01 4.98072207e-02 -9.00865674e-01 -5.14992177e-01
-2.43333519e-01 5.95997810e-01 1.78561509e-01 3.24565470e-01
3.90746742e-01 -9.67729628e-01 -5.46446562e-01 -2.54345328e-01
1.71923563e-01 1.50593817e-01 -1.21560834e-01 6.79346681e-01
-7.76607037e-01 9.07253325e-01 1.12456664e-01 -1.89186156e-01
-9.66135085e-01 2.23747805e-01 3.72557044e-01 -6.58970296e-01
-7.03046083e-01 8.04320753e-01 7.70009756e-02 -1.08486259e+00
9.21827778e-02 -8.27500641e-01 1.42414331e-01 -2.65274465e-01
3.00408691e-01 3.40952516e-01 2.32177842e-02 2.75269132e-02
-4.56460088e-01 2.21196637e-01 -2.64735878e-01 -1.17145248e-01
1.18268335e+00 2.38398314e-01 -4.30705875e-01 2.78458953e-01
9.33285415e-01 1.52675316e-01 -5.15686631e-01 -3.63952011e-01
2.72860467e-01 -1.58285573e-02 -3.92897964e-01 -1.33532596e+00
-3.71982485e-01 8.13077033e-01 -5.02852321e-01 4.32195902e-01
6.89424813e-01 2.29403198e-01 9.51383352e-01 1.06950712e+00
4.71854120e-01 -6.93323374e-01 5.46892881e-02 9.17991936e-01
1.13870096e+00 -7.43835330e-01 7.67535865e-02 -3.20849121e-01
-6.02512360e-01 9.24193442e-01 7.34804153e-01 -3.03138811e-02
2.23652571e-01 4.06547159e-01 -6.31181449e-02 -3.92528504e-01
-1.33934081e+00 1.54050868e-02 -8.50516930e-02 5.05298853e-01
5.82233012e-01 -2.94834469e-02 -1.83642492e-01 6.23375297e-01
-5.58165312e-01 1.49451673e-01 4.74066406e-01 8.43530416e-01
-7.90675953e-02 -8.12604904e-01 -6.09766208e-02 5.98883569e-01
-3.96585166e-01 -2.75930911e-01 -6.01550877e-01 1.26719916e+00
-2.19971418e-01 9.04766083e-01 1.66407764e-01 -5.09588718e-02
3.84821534e-01 7.26418376e-01 5.55155933e-01 -7.31552064e-01
-6.87836409e-01 -1.19405103e+00 8.25661957e-01 -4.60714906e-01
1.01036467e-01 -3.08844328e-01 -1.52398181e+00 -6.68331921e-01
-1.15211569e-01 4.38573062e-01 9.07183141e-02 1.07468534e+00
4.31847036e-01 5.38488984e-01 -1.37044132e-01 9.63405892e-02
-8.50832701e-01 -9.29255009e-01 -1.78528894e-02 3.34582329e-01
2.92575032e-01 -4.28909659e-01 -3.96614611e-01 -1.51489571e-01]
|
[9.740581512451172, 7.530217170715332]
|
a2645a02-87b6-440e-94bb-9c344778f053
|
improved-probabilistic-image-text
|
2305.18171
| null |
https://arxiv.org/abs/2305.18171v1
|
https://arxiv.org/pdf/2305.18171v1.pdf
|
Improved Probabilistic Image-Text Representations
|
Image-Text Matching (ITM) task, a fundamental vision-language (VL) task, suffers from the inherent ambiguity arising from multiplicity and imperfect annotations. Deterministic functions are not sufficiently powerful to capture ambiguity, prompting the exploration of probabilistic embeddings to tackle the challenge. However, the existing probabilistic ITM approach encounters two key shortcomings; the burden of heavy computations due to the Monte Carlo approximation, and the loss saturation issue in the face of abundant false negatives. To overcome the issues, this paper presents an improved Probabilistic Cross-Modal Embeddings (named PCME++) by introducing a new probabilistic distance with a closed-form solution. In addition, two optimization techniques are proposed to enhance PCME++ further; first, the incorporation of pseudo-positives to prevent the loss saturation problem under massive false negatives; second, mixed sample data augmentation for probabilistic matching. Experimental results on MS-COCO Caption and two extended benchmarks, CxC and ECCV Caption, demonstrate the effectiveness of PCME++ compared to state-of-the-art ITM methods. The robustness of PCME++ is also evaluated under noisy image-text correspondences. In addition, the potential applicability of PCME++ in automatic prompt tuning for zero-shot classification is shown. The code is available at https://naver-ai.github.io/pcmepp/.
|
['Sanghyuk Chun']
|
2023-05-29
| null | null | null | null |
['text-matching']
|
['natural-language-processing']
|
[ 3.71774524e-01 -2.50029981e-01 -6.82856366e-02 -2.96797693e-01
-1.26313138e+00 -1.15092769e-01 8.04696739e-01 -7.52973929e-02
-5.39148331e-01 4.35423464e-01 3.62198830e-01 -1.84341557e-02
-4.85172495e-02 -3.46681327e-01 -6.83211327e-01 -6.46085024e-01
3.41048777e-01 5.31244159e-01 3.22264284e-01 6.42098859e-02
1.90522924e-01 -3.57721597e-02 -1.94432163e+00 3.77139479e-01
9.65297163e-01 1.19867504e+00 5.07329106e-01 3.57536554e-01
-3.32998842e-01 4.90958422e-01 -2.96702206e-01 -7.18775511e-01
2.86263376e-01 -1.99399032e-02 -3.25453848e-01 -4.03072014e-02
6.59171522e-01 -2.24009678e-01 -5.18214583e-01 1.25593865e+00
7.30894029e-01 1.41064823e-01 6.78259492e-01 -1.75570107e+00
-9.84834909e-01 3.52820039e-01 -6.55624688e-01 4.18201201e-02
2.20745951e-01 3.14796269e-01 1.08112466e+00 -1.56523621e+00
3.70310694e-01 1.32995951e+00 9.01174366e-01 6.45478427e-01
-1.08614242e+00 -7.23026872e-01 -1.41412213e-01 4.52432662e-01
-1.71463966e+00 -3.52665573e-01 6.69974744e-01 -5.73721349e-01
8.05134594e-01 1.80497348e-01 1.98012859e-01 1.37968373e+00
1.71484344e-03 1.04047847e+00 1.22770607e+00 -4.35939848e-01
5.62193990e-02 3.17660391e-01 2.55224049e-01 7.08601058e-01
8.68678018e-02 4.98439699e-01 -7.27762640e-01 -3.07082593e-01
3.10927778e-01 -1.53181970e-01 -3.74123067e-01 -3.48590940e-01
-1.19879293e+00 7.93455362e-01 2.58355528e-01 -4.81948517e-02
-1.20787643e-01 2.35166267e-01 4.25585836e-01 -2.06380844e-01
2.15752348e-01 1.97299853e-01 -1.57533631e-01 -4.19981658e-01
-1.17480147e+00 2.43602157e-01 4.47882533e-01 1.19247675e+00
5.73032916e-01 -9.94445160e-02 -5.26907444e-01 9.89901483e-01
5.21994174e-01 7.25079179e-01 4.17580843e-01 -7.96710968e-01
6.19860470e-01 4.00004387e-01 -8.42319278e-05 -9.41199660e-01
-2.29744948e-02 -2.00190499e-01 -9.16443408e-01 1.02171198e-01
1.82753265e-01 2.91680098e-01 -1.14871299e+00 1.61635029e+00
1.54534772e-01 3.29244345e-01 -1.15897069e-02 1.04841220e+00
1.00800812e+00 7.96735406e-01 -2.01870818e-02 6.86198175e-02
1.43881989e+00 -1.04873919e+00 -8.19944561e-01 -1.97603002e-01
2.23546103e-01 -9.59222376e-01 1.18205667e+00 -1.29193766e-02
-7.09369540e-01 -5.32261372e-01 -9.33837593e-01 -1.43932536e-01
-3.28508914e-01 2.55297750e-01 4.48291034e-01 5.86098254e-01
-9.28723276e-01 2.98290461e-01 -5.48364758e-01 -1.94231078e-01
6.10275805e-01 1.81031391e-01 -2.49862894e-01 -3.68129641e-01
-1.24856389e+00 9.48696136e-01 3.63425821e-01 1.93976089e-01
-4.92013097e-01 -8.70943785e-01 -1.02247810e+00 -1.16079584e-01
2.78364629e-01 -6.78161561e-01 1.05632782e+00 -6.44570947e-01
-1.20109737e+00 7.81340599e-01 -2.82279909e-01 -4.39561456e-01
8.31123471e-01 -2.63784528e-01 -1.50367364e-01 -1.23819418e-01
2.07764655e-01 1.19920707e+00 8.86899233e-01 -1.26151764e+00
-4.89977360e-01 -1.91445783e-01 -3.16610247e-01 2.67193675e-01
-2.23809525e-01 -3.48448679e-02 -9.31496680e-01 -7.95064092e-01
-6.91122338e-02 -9.52362180e-01 -1.35795370e-01 2.70766139e-01
-2.49291897e-01 -2.75547892e-01 8.00591707e-01 -4.61664826e-01
1.08996880e+00 -2.22432160e+00 -5.03181778e-02 -1.29508272e-01
-6.19945535e-03 4.67682749e-01 -2.87086159e-01 3.20001960e-01
1.38303831e-01 3.04848012e-02 -5.02844751e-01 -7.47323453e-01
3.17833155e-01 3.68438989e-01 -4.33873475e-01 2.62191206e-01
4.97602910e-01 1.07935619e+00 -7.74672210e-01 -8.93278420e-01
6.04901612e-01 6.23842180e-01 -3.34712893e-01 3.15836482e-02
-2.63029665e-01 9.76391137e-02 -1.03752732e-01 8.63593817e-01
8.56820166e-01 -3.15384924e-01 -3.17025959e-01 -3.68507445e-01
-1.04715833e-02 -8.57453197e-02 -1.40572023e+00 1.52053535e+00
-2.11404890e-01 6.79915190e-01 -2.73789346e-01 -7.34914541e-01
7.48253882e-01 1.61652371e-01 4.01951373e-01 -7.26868570e-01
3.33390594e-01 2.07915887e-01 -3.52748603e-01 -4.44907159e-01
7.76499689e-01 4.46411408e-02 2.64460444e-02 -1.20970406e-01
3.71285751e-02 3.91027294e-02 3.01910974e-02 2.35802963e-01
6.88458264e-01 1.94838643e-01 4.02777409e-03 -1.98226616e-01
4.12778795e-01 1.98755562e-02 7.08662689e-01 8.68124783e-01
-5.79039633e-01 9.88973320e-01 3.16713154e-02 7.63469702e-03
-1.08470666e+00 -1.22473037e+00 -4.98830169e-01 6.84669077e-01
5.18547118e-01 -4.78591055e-01 -4.36736315e-01 -4.17892605e-01
3.51679809e-02 6.66390419e-01 -4.63424027e-01 -6.24080822e-02
-8.22995529e-02 -1.06904435e+00 6.16008341e-01 6.16463006e-01
6.02224767e-01 -8.38724434e-01 -3.80780607e-01 6.49622530e-02
-6.21285319e-01 -1.53745830e+00 -8.10614824e-01 -2.08631665e-01
-5.25581241e-01 -8.45173061e-01 -7.34569252e-01 -7.57433712e-01
3.61481518e-01 4.71559227e-01 9.67441738e-01 -1.54041737e-01
-5.53847969e-01 5.78522921e-01 -3.26725304e-01 -4.84104604e-01
-4.11684930e-01 -4.42795366e-01 1.25673369e-01 4.30016145e-02
5.39795637e-01 -2.78517991e-01 -5.51330924e-01 3.08916241e-01
-7.91460454e-01 2.63253331e-01 6.33583963e-01 1.22511554e+00
8.23313534e-01 -9.43346024e-02 3.55948061e-01 -4.00461823e-01
2.58525431e-01 -2.96356291e-01 -7.00309396e-01 3.44795763e-01
-8.55278850e-01 1.15079552e-01 2.28251576e-01 -6.52129889e-01
-9.65155184e-01 1.39443442e-01 -1.67341739e-01 -8.29626381e-01
1.48587003e-01 2.48142108e-01 -2.70115137e-01 -2.06179661e-03
4.43058461e-01 4.49751258e-01 1.56403944e-01 -2.85888523e-01
5.06680727e-01 8.31583858e-01 7.42015362e-01 -6.77746415e-01
9.81959045e-01 4.72533971e-01 -1.96136996e-01 -8.90513182e-01
-7.95708179e-01 -7.66509712e-01 -1.55096605e-01 -2.52816349e-01
7.29572594e-01 -1.16093826e+00 -3.84631664e-01 6.65614367e-01
-1.23306859e+00 9.26590338e-02 -2.25299463e-01 6.26521826e-01
-5.24758339e-01 7.72506654e-01 -4.12700742e-01 -9.09377635e-01
-5.17233074e-01 -1.36187685e+00 1.14489174e+00 2.87708998e-01
5.82810724e-03 -5.90134561e-01 -9.48356837e-02 7.77389467e-01
3.16133380e-01 -3.48033290e-03 6.61527872e-01 -4.64988828e-01
-8.14853489e-01 -1.39858320e-01 -6.65567815e-01 4.01777267e-01
-2.53231972e-01 8.60172734e-02 -1.22004998e+00 -1.60039485e-01
-2.55216062e-01 -3.44491303e-01 9.86034632e-01 3.51528883e-01
1.07406819e+00 -1.58797964e-01 -1.95230171e-01 6.97142541e-01
1.57001686e+00 -2.31455222e-01 7.64902771e-01 3.37076306e-01
7.13668525e-01 3.35553735e-01 8.54593694e-01 4.31521684e-01
5.60469687e-01 8.33593607e-01 4.51152831e-01 1.20428503e-01
-4.89159644e-01 -3.95660281e-01 3.83986354e-01 7.94049680e-01
3.36458981e-01 -3.11787635e-01 -9.22206461e-01 6.69783771e-01
-2.00162554e+00 -8.57023716e-01 -3.86987299e-01 2.26474762e+00
8.47852468e-01 1.07664622e-01 -2.53544539e-01 9.44029987e-02
8.67580473e-01 2.03844011e-01 -3.65145415e-01 4.90096696e-02
-4.13519472e-01 1.69071093e-01 6.27675653e-01 4.38184351e-01
-1.18651569e+00 7.76670873e-01 4.99666929e+00 1.23149121e+00
-7.85802305e-01 2.48271912e-01 3.04239303e-01 1.69507474e-01
-2.42317066e-01 3.24112549e-02 -1.00731194e+00 7.57405162e-01
7.22560048e-01 4.58296910e-02 2.53179729e-01 7.11407125e-01
-4.84271608e-02 -1.29614025e-01 -9.21728253e-01 1.59503365e+00
2.90878445e-01 -1.44630790e+00 2.49733537e-01 -1.14279009e-01
6.51224017e-01 4.33533579e-01 2.71729648e-01 3.32277954e-01
-1.15734460e-02 -9.14683700e-01 8.15226793e-01 5.21155655e-01
9.35977995e-01 -4.42059278e-01 8.49524200e-01 2.41525799e-01
-1.10771513e+00 7.85208344e-02 -4.30030316e-01 3.11515361e-01
2.31342390e-01 7.91757762e-01 -6.90978229e-01 6.53341830e-01
7.91552901e-01 5.44528961e-01 -6.28542960e-01 1.19100320e+00
8.33587814e-03 4.08835232e-01 -4.83813077e-01 -1.87614970e-02
2.03769833e-01 -8.12327713e-02 8.30103993e-01 1.45904255e+00
3.75861526e-01 -1.50679991e-01 1.18078195e-01 9.19349372e-01
-6.81944489e-02 -5.42797856e-02 -4.35460359e-01 8.45970307e-03
7.33903289e-01 1.11775005e+00 -2.69403994e-01 -1.30482599e-01
-4.61083382e-01 1.09565437e+00 1.89957395e-01 2.41315097e-01
-1.09465539e+00 -9.42964256e-02 6.72069252e-01 -1.07849352e-01
1.81672335e-01 -1.46017268e-01 -3.41140151e-01 -1.16685390e+00
4.24554139e-01 -8.17636311e-01 2.83675104e-01 -7.62936234e-01
-1.63330722e+00 4.08766389e-01 -1.04497582e-01 -1.42056763e+00
1.46533832e-01 -5.91546476e-01 -3.45572561e-01 8.48205566e-01
-1.67694485e+00 -1.41648924e+00 -4.57556933e-01 4.35474515e-01
6.64587796e-01 -7.31353983e-02 6.60283089e-01 5.81163645e-01
-4.51869279e-01 1.01543581e+00 2.69153029e-01 1.85197182e-02
7.89788842e-01 -1.11484396e+00 3.70578110e-01 8.42980683e-01
2.17315301e-01 1.75961733e-01 8.11689258e-01 -4.22389090e-01
-1.34589005e+00 -1.31402969e+00 1.05519807e+00 -7.67030120e-01
6.14908099e-01 -4.27396744e-01 -1.06371868e+00 2.47146234e-01
-5.95546886e-02 6.69652596e-02 5.26843548e-01 -3.10137033e-01
-6.32687986e-01 1.84692349e-02 -1.05675602e+00 7.16209531e-01
9.15204227e-01 -6.25800967e-01 -6.59360766e-01 3.10731888e-01
8.61648738e-01 -4.45293784e-01 -7.67185450e-01 6.15151823e-01
5.29108644e-01 -8.33049297e-01 1.04342508e+00 -3.79844122e-02
3.78245950e-01 -4.47707683e-01 -6.24094546e-01 -8.29592168e-01
-9.38540101e-02 -5.39873719e-01 -1.94568679e-01 1.62671459e+00
2.15558514e-01 -6.07663333e-01 6.07405901e-01 5.94762146e-01
-7.49693364e-02 -8.45669150e-01 -1.38899481e+00 -9.13617969e-01
-8.23982880e-02 -7.95902252e-01 2.34932646e-01 7.84841359e-01
-3.71769130e-01 2.51478255e-01 -6.31021917e-01 5.03453791e-01
9.94861007e-01 -1.07966401e-01 6.46795928e-01 -1.04133725e+00
-2.94683039e-01 -4.12649751e-01 -6.50873840e-01 -1.08708942e+00
-4.19421308e-02 -9.12529826e-01 3.12022418e-01 -1.46224332e+00
5.58181643e-01 -4.22981530e-01 -2.60811120e-01 1.30550116e-01
-4.85168010e-01 3.06474328e-01 4.92413342e-01 3.96524698e-01
-5.74284494e-01 1.01368678e+00 8.62966239e-01 -2.79158533e-01
6.64599091e-02 -2.06416532e-01 -3.87886107e-01 5.91958344e-01
6.62248254e-01 -5.56392074e-01 -2.67588675e-01 -3.62492055e-01
-6.22700155e-02 -1.29824758e-01 7.46177077e-01 -1.02434456e+00
3.51679116e-01 4.71776053e-02 1.51503161e-01 -9.89386797e-01
7.18149185e-01 -7.57514656e-01 1.03900097e-01 2.80976474e-01
-1.49716392e-01 5.40749058e-02 2.18977854e-01 9.41807330e-01
-2.83429563e-01 -2.41075695e-01 9.10321295e-01 2.77629226e-01
-9.47497427e-01 4.02945131e-01 -2.96164937e-02 3.21592480e-01
8.27152610e-01 -3.93929332e-01 -4.84663904e-01 -8.94981921e-02
-2.00792789e-01 6.82537019e-01 4.91410285e-01 8.25549722e-01
8.91052127e-01 -1.45010662e+00 -7.26910293e-01 -2.83362921e-02
6.53325081e-01 -4.41606790e-02 4.41250145e-01 9.97569084e-01
-1.92254364e-01 2.89270014e-01 1.80194363e-01 -8.99964690e-01
-1.31666362e+00 6.01703644e-01 2.94857293e-01 -1.44198313e-01
-7.22989559e-01 8.53494942e-01 1.56230554e-01 -3.95163298e-01
5.83796561e-01 -1.72548175e-01 1.70688137e-01 -1.46928713e-01
4.65073615e-01 2.18235016e-01 -1.45074040e-01 -6.66993380e-01
-5.28254211e-01 6.06369793e-01 -6.81226999e-02 -1.86859503e-01
8.44110370e-01 -2.69959211e-01 1.44680098e-01 4.00004506e-01
1.17909765e+00 -3.97089303e-01 -1.31582797e+00 -6.04557395e-01
1.23056211e-01 -6.84427559e-01 8.06393325e-02 -8.15267563e-01
-6.72636509e-01 1.00453544e+00 9.91502881e-01 -5.89694083e-01
9.34855461e-01 -5.64989112e-02 1.01510322e+00 1.41171336e-01
1.66673571e-01 -1.14542150e+00 5.95288724e-02 4.99189883e-01
8.01656425e-01 -1.69812989e+00 -1.89394817e-01 -3.15215528e-01
-8.62371087e-01 8.50145459e-01 5.78690410e-01 2.60731637e-01
6.26016796e-01 2.12994114e-01 5.11995666e-02 6.56777695e-02
-7.06358135e-01 -3.21103305e-01 4.59948868e-01 7.07241535e-01
4.51467484e-02 -3.83586064e-02 -1.77613616e-01 5.05745471e-01
2.62486506e-02 1.28834799e-01 4.59153831e-01 7.08972692e-01
-2.33116746e-01 -8.93288910e-01 -4.37663406e-01 4.43490684e-01
-3.49482566e-01 -4.53934371e-01 -1.34754404e-01 7.23419607e-01
1.65110111e-01 8.59462559e-01 6.08399026e-02 -4.45928425e-01
3.01363170e-01 1.12233713e-01 3.06017011e-01 -2.97402233e-01
-1.24278158e-01 -1.69099987e-01 -9.94436815e-02 -4.36719418e-01
-4.85333055e-01 -5.97837210e-01 -1.14619350e+00 -2.07883015e-01
-4.73684013e-01 -7.43697435e-02 7.12005734e-01 7.70557046e-01
4.85567778e-01 2.88049996e-01 3.52055639e-01 -7.25297511e-01
-9.80260134e-01 -8.98739874e-01 -2.33746842e-01 4.56495792e-01
2.67707825e-01 -9.01497900e-01 -4.68005687e-01 -1.36445060e-01]
|
[10.508296012878418, 1.184250831604004]
|
c6a54472-5e0d-4cb1-866d-0e1f2aba5404
|
omnivore-a-single-model-for-many-visual
|
2201.08377
| null |
https://arxiv.org/abs/2201.08377v2
|
https://arxiv.org/pdf/2201.08377v2.pdf
|
Omnivore: A Single Model for Many Visual Modalities
|
Prior work has studied different visual modalities in isolation and developed separate architectures for recognition of images, videos, and 3D data. Instead, in this paper, we propose a single model which excels at classifying images, videos, and single-view 3D data using exactly the same model parameters. Our 'Omnivore' model leverages the flexibility of transformer-based architectures and is trained jointly on classification tasks from different modalities. Omnivore is simple to train, uses off-the-shelf standard datasets, and performs at-par or better than modality-specific models of the same size. A single Omnivore model obtains 86.0% on ImageNet, 84.1% on Kinetics, and 67.1% on SUN RGB-D. After finetuning, our models outperform prior work on a variety of vision tasks and generalize across modalities. Omnivore's shared visual representation naturally enables cross-modal recognition without access to correspondences between modalities. We hope our results motivate researchers to model visual modalities together.
|
['Ishan Misra', 'Armand Joulin', 'Laurens van der Maaten', 'Nikhila Ravi', 'Mannat Singh', 'Rohit Girdhar']
|
2022-01-20
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Girdhar_Omnivore_A_Single_Model_for_Many_Visual_Modalities_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Girdhar_Omnivore_A_Single_Model_for_Many_Visual_Modalities_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['scene-recognition']
|
['computer-vision']
|
[-1.23339161e-01 -5.36437213e-01 -5.50649405e-01 -4.03974146e-01
-6.50548995e-01 -8.66186261e-01 7.68028438e-01 -4.34345514e-01
-4.35728967e-01 1.57855749e-01 4.25221711e-01 -3.78256470e-01
3.08850169e-01 -3.92749578e-01 -8.47410381e-01 -2.66472727e-01
1.54106542e-01 1.32719576e-01 9.46376473e-02 -2.50073671e-02
1.52950659e-01 5.12379527e-01 -1.75747418e+00 6.77970350e-01
5.64262271e-02 1.23875213e+00 1.59303769e-01 9.79598939e-01
4.80727665e-02 1.05338347e+00 -4.06837044e-03 -2.29287475e-01
3.83888960e-01 1.94879770e-02 -1.04945803e+00 2.84367234e-01
1.16066670e+00 -7.02478588e-01 -8.49592149e-01 5.42647541e-01
5.63907325e-01 -1.10794269e-01 5.07172167e-01 -1.46896899e+00
-1.20717263e+00 1.21112786e-01 -6.21084690e-01 4.54836160e-01
5.72148621e-01 4.80930537e-01 9.01003182e-01 -9.79983330e-01
5.99151850e-01 1.23163450e+00 6.15578949e-01 7.88863838e-01
-1.27522051e+00 -3.99367929e-01 3.07987392e-01 2.90382445e-01
-1.15429795e+00 -8.09776664e-01 1.99784562e-01 -5.70196211e-01
1.55454445e+00 1.65288880e-01 7.04308450e-01 1.48540330e+00
9.68863219e-02 8.89518321e-01 1.26008224e+00 -1.57278031e-01
-2.35705033e-01 -1.48767337e-01 -2.04974730e-02 6.69951677e-01
-1.46342352e-01 7.91948885e-02 -9.59774137e-01 2.50376642e-01
9.23214853e-01 3.47874761e-01 -2.67103821e-01 -6.26156092e-01
-1.62942195e+00 3.41777444e-01 3.87961954e-01 1.58575639e-01
1.52816568e-02 4.80066478e-01 4.50762868e-01 6.53651774e-01
2.20510855e-01 2.62735605e-01 -5.94286799e-01 -3.85726929e-01
-5.55744350e-01 -1.50857523e-01 5.16310751e-01 1.03820324e+00
6.46387339e-01 1.54245406e-01 1.77287921e-01 8.20488095e-01
2.97674090e-01 8.35326552e-01 4.57316756e-01 -1.22541773e+00
4.68783617e-01 4.72049952e-01 -1.41600251e-01 -5.95200956e-01
-5.37551761e-01 -2.72651583e-01 -6.38497889e-01 2.51267672e-01
3.28693300e-01 9.50773358e-02 -1.33772111e+00 1.73516452e+00
-7.61993676e-02 1.94945455e-01 5.79247922e-02 1.19859433e+00
1.25623035e+00 4.68447626e-01 1.07518248e-01 5.59595048e-01
1.35219705e+00 -1.00252569e+00 -5.24180718e-02 -3.57401758e-01
4.28772420e-01 -9.31851268e-01 1.19428873e+00 4.22419250e-01
-1.19020331e+00 -7.41478026e-01 -8.87906075e-01 -4.83601034e-01
-4.35454011e-01 8.74456763e-02 8.63849282e-01 5.35762310e-01
-1.59063482e+00 3.32667589e-01 -8.70110750e-01 -8.40885460e-01
3.19059372e-01 2.99172580e-01 -9.39924896e-01 -2.66456097e-01
-6.12047791e-01 1.00693572e+00 4.09744009e-02 -2.33517617e-01
-1.33584285e+00 -8.60677481e-01 -8.22844684e-01 -2.22634599e-01
1.96459740e-01 -1.03391814e+00 1.33801007e+00 -9.42305207e-01
-1.39679742e+00 1.43115675e+00 -2.14788526e-01 -2.37504199e-01
1.14469983e-01 -3.09456140e-01 -6.79905593e-01 3.97583723e-01
-6.45947978e-02 1.00029981e+00 7.74397135e-01 -1.05486619e+00
-5.33303499e-01 -2.95610607e-01 6.22574568e-01 3.05670351e-01
-3.35380614e-01 -1.22963808e-01 -8.67683291e-01 -2.17329711e-01
5.57884537e-02 -9.97854114e-01 3.28217149e-01 3.41418147e-01
-2.15626583e-01 -3.54950354e-02 8.39688361e-01 -3.08373183e-01
4.29925859e-01 -2.48152113e+00 3.05746794e-01 -1.60080701e-01
4.76169288e-01 -1.00720108e-01 -5.90093195e-01 3.28621447e-01
-1.00307718e-01 3.38655896e-02 3.43027443e-01 -3.35835665e-01
1.41813785e-01 2.04840526e-01 -2.30443746e-01 6.04792774e-01
9.40630436e-02 1.00205028e+00 -6.26735687e-01 -3.19251060e-01
5.25303364e-01 5.71376264e-01 -6.57755852e-01 1.99115887e-01
-4.78242934e-02 3.02545041e-01 -3.14221717e-02 9.52711940e-01
5.94487011e-01 -8.57806265e-01 2.31464982e-01 -6.51146412e-01
-1.87737495e-02 1.75154999e-01 -7.99362957e-01 2.24777722e+00
-6.08980060e-01 8.07699084e-01 1.23647459e-01 -9.05497253e-01
5.01129746e-01 3.15536767e-01 3.77949774e-01 -1.10479367e+00
-6.55618906e-02 2.33256131e-01 -2.10263848e-01 -6.54360890e-01
4.65470970e-01 1.37514606e-01 -3.34302373e-02 7.05510616e-01
6.04154944e-01 -2.89501762e-03 1.60730034e-02 3.24370623e-01
1.05721593e+00 3.47196579e-01 -4.93196584e-02 6.06838241e-02
5.99954687e-02 -1.88425615e-01 5.16126081e-02 8.85460973e-01
-2.73149163e-01 8.99117947e-01 1.00742251e-01 -6.25564516e-01
-1.13727093e+00 -1.44571197e+00 -2.76625641e-02 1.43419540e+00
3.31261754e-01 -4.64130282e-01 -8.94422978e-02 -5.28685927e-01
2.57418931e-01 1.66852444e-01 -6.29295766e-01 3.20518054e-02
-1.32295132e-01 -3.41594249e-01 6.34490192e-01 8.14906657e-01
6.19177282e-01 -5.30790508e-01 -6.20369196e-01 -3.13033998e-01
-2.68658429e-01 -1.31895888e+00 -4.44207639e-01 2.03049377e-01
-7.73527741e-01 -1.21110046e+00 -5.88223994e-01 -6.20045781e-01
1.50978953e-01 1.00893295e+00 1.46547484e+00 -9.00992081e-02
-2.09062427e-01 1.09129071e+00 -1.71183422e-01 -1.41293211e-02
-6.24417234e-03 6.24761544e-02 1.95679441e-01 -8.41115490e-02
5.22055626e-01 -6.79042518e-01 -6.83116436e-01 3.66175324e-01
-8.14022660e-01 2.22035199e-01 5.26611626e-01 7.23163962e-01
4.12606478e-01 -7.69793034e-01 1.40726417e-01 -3.74512017e-01
-1.14451274e-02 -6.15423560e-01 -2.15147287e-01 2.91299075e-01
-2.48879284e-01 -1.48440644e-01 4.27782208e-01 -4.21089351e-01
-7.24236727e-01 -1.08123541e-01 1.04350954e-01 -9.85317230e-01
-4.12603617e-01 2.25059643e-01 7.93539733e-02 -2.92411447e-01
6.28920853e-01 3.64887506e-01 1.30589664e-01 -5.63558161e-01
5.32736003e-01 5.51697016e-01 6.70281351e-01 -3.53654027e-01
5.04077673e-01 8.09747338e-01 -2.46932089e-01 -6.82634473e-01
-7.55304873e-01 -4.01209205e-01 -5.55101275e-01 -1.68086678e-01
9.28998888e-01 -1.50465083e+00 -9.58506227e-01 7.65943706e-01
-8.88955534e-01 -4.03820276e-01 -8.21201205e-02 6.24580145e-01
-5.97729027e-01 3.08987886e-01 -7.58275747e-01 -2.43325159e-01
9.52017214e-03 -1.00729001e+00 1.22134709e+00 1.86110064e-01
-7.87119418e-02 -9.26331997e-01 -8.32079649e-02 5.98081470e-01
5.55064976e-01 -1.50658622e-01 5.22121608e-01 -1.75554186e-01
-7.90482104e-01 2.03782916e-01 -7.14419365e-01 3.41102779e-01
8.80298540e-02 -2.76036784e-02 -1.27281928e+00 -4.90159690e-01
-5.10533631e-01 -9.94808078e-01 1.02312827e+00 2.59845436e-01
1.29528856e+00 2.84070019e-02 -2.45031983e-01 1.03180575e+00
1.29642701e+00 -1.13862030e-01 5.21943152e-01 3.96706879e-01
8.42604935e-01 1.07433647e-01 7.54652992e-02 3.07993621e-01
7.51124680e-01 7.07215428e-01 6.62126303e-01 -1.75195411e-01
-3.64364684e-01 -1.30130813e-01 4.54647452e-01 6.80499494e-01
-2.47099116e-01 -1.65998012e-01 -9.22928929e-01 6.64551795e-01
-1.58034182e+00 -1.26006186e+00 3.33180666e-01 1.71372962e+00
5.27134717e-01 -2.57285330e-02 3.00819337e-01 -4.13139164e-01
4.28467155e-01 3.37735057e-01 -7.91778147e-01 -3.58639538e-01
-3.79857361e-01 1.38794305e-02 5.14482439e-01 2.44164020e-01
-1.26451552e+00 7.04658449e-01 7.56027508e+00 3.37119579e-01
-1.49427176e+00 2.18150765e-01 3.74336869e-01 -6.60356998e-01
-3.94623905e-01 -1.48532903e-02 -5.58671772e-01 2.69622892e-01
8.54921401e-01 1.60657749e-01 6.12173319e-01 7.29761899e-01
-2.08752140e-01 2.89463755e-02 -1.46558547e+00 1.50755596e+00
4.26608771e-01 -1.49756622e+00 2.22854450e-01 2.09114149e-01
5.76489687e-01 7.77113795e-01 4.02368784e-01 9.70109627e-02
3.70721757e-01 -1.31940615e+00 8.92277598e-01 6.60735726e-01
1.12961578e+00 -2.11222842e-01 3.25033158e-01 -3.59240547e-02
-1.08148360e+00 -1.31191239e-01 -1.81633219e-01 -2.29472950e-01
-3.08094155e-02 1.58740595e-01 -3.59578907e-01 3.06169391e-01
1.15172565e+00 1.32889891e+00 -8.15770030e-01 7.86949039e-01
2.93189228e-01 2.79748499e-01 -4.62017894e-01 3.16204816e-01
2.59328753e-01 2.79865235e-01 1.95217595e-01 1.29336727e+00
2.21642464e-01 -1.16812266e-01 2.30475560e-01 4.35744077e-01
-2.54813880e-01 -4.10817623e-01 -8.93022597e-01 -1.26797631e-01
4.85263884e-01 1.07266700e+00 -2.34322295e-01 -3.54421914e-01
-1.00210238e+00 1.06659973e+00 3.53632510e-01 4.05627102e-01
-7.82957673e-01 -5.04389964e-02 7.92524874e-01 -9.15845782e-02
4.90697294e-01 -4.35843259e-01 -1.57902509e-01 -1.74228334e+00
1.77791733e-02 -1.09831464e+00 4.98200774e-01 -1.31588984e+00
-1.48538184e+00 6.04523599e-01 -1.79289967e-01 -1.39532757e+00
-2.59198129e-01 -1.02871144e+00 -9.54900756e-02 7.17224002e-01
-1.63909853e+00 -1.45501888e+00 -6.77902520e-01 1.24236155e+00
2.70858914e-01 -3.19453180e-01 8.78002703e-01 3.62132758e-01
-3.05655360e-01 5.99328995e-01 4.95199449e-02 1.83306903e-01
9.14889038e-01 -1.09371972e+00 5.24987280e-01 5.89432299e-01
3.88181269e-01 5.59259057e-01 2.22517103e-01 -4.51381616e-02
-2.09448957e+00 -6.90399289e-01 5.60474277e-01 -8.03258777e-01
7.32954264e-01 -1.70854479e-01 -4.43399817e-01 1.02641606e+00
4.89415109e-01 3.60972792e-01 8.14370751e-01 4.35385942e-01
-1.30967629e+00 -3.44155490e-01 -8.25130045e-01 5.79008341e-01
1.52763474e+00 -1.17469954e+00 -4.15781409e-01 1.91477105e-01
3.79677773e-01 -5.43562472e-01 -1.10669672e+00 2.61301607e-01
9.94758666e-01 -1.19064415e+00 1.29389513e+00 -9.05622423e-01
6.80981636e-01 -1.51353642e-01 -6.70915306e-01 -1.09792960e+00
-3.74646097e-01 -1.80578649e-01 -3.98910910e-01 7.24792719e-01
2.18788475e-01 -5.85607350e-01 5.96435547e-01 3.71817410e-01
-4.91080321e-02 -5.27677298e-01 -8.30270290e-01 -7.76066363e-01
-1.27540752e-01 -5.41407824e-01 5.53501308e-01 9.68499124e-01
-1.08543612e-01 4.27079916e-01 -5.57687104e-01 -7.20535358e-03
6.01203680e-01 5.19869626e-01 1.06261826e+00 -8.47323477e-01
-6.22039378e-01 -3.93560886e-01 -6.87192500e-01 -1.47102523e+00
-5.70872193e-03 -9.47663069e-01 -4.90953833e-01 -1.38276815e+00
5.82721889e-01 -3.65159005e-01 -5.64546347e-01 1.01126242e+00
3.46926004e-01 8.78915727e-01 6.27147377e-01 5.27597070e-01
-8.80828321e-01 2.78418750e-01 1.22490370e+00 -4.01401311e-01
2.91227579e-01 -5.88268876e-01 -9.91070092e-01 6.09768093e-01
6.79875076e-01 2.77699381e-01 -5.52315652e-01 -1.20003819e+00
1.06932767e-01 6.49569836e-03 9.98778045e-01 -1.06684256e+00
5.16191386e-02 -2.20625520e-01 7.11166024e-01 -4.74476904e-01
6.71780825e-01 -7.14415491e-01 1.88522413e-01 -5.20150773e-02
-2.44389504e-01 2.76191682e-01 3.85653734e-01 3.93100798e-01
-1.31621882e-01 4.54399437e-01 8.05661738e-01 -3.19972068e-01
-1.32204425e+00 3.25150996e-01 -3.07530016e-01 1.65589109e-01
8.05269301e-01 -4.17972684e-01 -8.80249619e-01 -4.60257858e-01
-8.65474582e-01 1.58701569e-01 8.73265803e-01 8.79425108e-01
6.58642054e-01 -1.41062617e+00 -5.56628942e-01 1.33077472e-01
4.02826726e-01 -5.89012742e-01 5.91256261e-01 8.26499701e-01
-4.16402549e-01 5.43107927e-01 -6.24383152e-01 -1.18931293e+00
-1.22498930e+00 6.17825449e-01 5.07208169e-01 2.04793811e-01
-6.08098209e-01 7.25203812e-01 2.41445601e-01 -4.99648035e-01
1.42987177e-01 -5.88615350e-02 2.02432811e-01 -1.98686391e-01
4.37957793e-01 8.64100307e-02 -2.48096678e-02 -6.22201800e-01
-6.86500788e-01 8.67486537e-01 -1.97514176e-01 -1.82527319e-01
1.37060630e+00 -3.37738723e-01 9.28329825e-02 7.09444404e-01
1.44087768e+00 -2.91813821e-01 -1.48351586e+00 -2.97866225e-01
-6.93423271e-01 -5.78201950e-01 3.02621350e-02 -1.04374421e+00
-1.22716582e+00 1.05074406e+00 7.44471490e-01 5.33367321e-02
1.26088738e+00 2.96006203e-01 6.41887128e-01 4.08283055e-01
4.97246176e-01 -8.89545143e-01 3.91810954e-01 6.42927706e-01
6.94695115e-01 -1.34059310e+00 -1.24589065e-02 -7.11937919e-02
-6.15577340e-01 9.04240668e-01 8.95380259e-01 4.69015166e-02
5.26665330e-01 3.07710707e-01 2.97294229e-01 -1.48285553e-01
-1.10291183e+00 -3.30603480e-01 2.73500681e-01 7.64124393e-01
5.46971798e-01 -1.05601951e-01 5.77653885e-01 -7.64761940e-02
1.27812952e-01 1.68451399e-01 4.19752210e-01 1.06410289e+00
-8.07227716e-02 -8.52544785e-01 -2.59060115e-01 3.93636197e-01
-1.97519884e-01 -1.52031168e-01 -2.81410635e-01 8.85883987e-01
6.89785182e-02 8.72088134e-01 3.56549084e-01 -7.19423771e-01
1.93970636e-01 -2.43756231e-02 9.98640418e-01 -3.89600873e-01
-3.93197298e-01 -2.13096082e-01 8.83385986e-02 -1.04640949e+00
-9.35336888e-01 -6.25671446e-01 -6.54830635e-01 -8.38360190e-01
2.13379711e-01 -3.36631447e-01 6.09007955e-01 7.18564987e-01
9.64608014e-01 1.45850375e-01 5.66531360e-01 -1.11410403e+00
-3.16343039e-01 -7.09333062e-01 -4.91250277e-01 6.22077048e-01
6.27502441e-01 -8.03842425e-01 -1.77328572e-01 1.88472986e-01]
|
[10.064738273620605, 1.3994168043136597]
|
4a17e8fd-d3de-4324-adbd-4d53dfc5a97b
|
optimal-power-flow-for-integrated-primary
|
2306.13287
| null |
https://arxiv.org/abs/2306.13287v1
|
https://arxiv.org/pdf/2306.13287v1.pdf
|
Optimal Power Flow for Integrated Primary-Secondary Distribution Networks with Service Transformers
|
Secondary distribution networks (SDNets) play an increasingly important role in smart grids due to a high proliferation of distributed energy resources (DERs) in SDNets. However, most existing optimal power flow (OPF) problems do not take into account SDNets with service transformers. Handling the nonlinear and nonconvex SDNet power flow constraints is still an outstanding problem. To meet this gap, we first utilize the second-order cone programming relaxation and linearization to make service transformer constraints convex, respectively. Then, the linearized triplex service line power flow model, including its compact matrix-vector form, is further developed to compose the SDNet OPF model with our proposed service transformer model. This proposed SDNet OPF model can be easily embedded into existing primary distribution network (PDNet) OPF models, resulting in a holistic power system decision-making solution for integrated primary-secondary distribution networks. A case study is presented for an integrated primary-secondary distribution network that demonstrates the practical effectiveness of this model.
|
['Zixiao Ma', 'Zhaoyu Wang', 'Naihao Shi', 'Rui Cheng']
|
2023-06-23
| null | null | null | null |
['decision-making']
|
['reasoning']
|
[-4.96242285e-01 -2.81179120e-04 -3.43258440e-01 -7.17332587e-02
-2.53661811e-01 -8.95753086e-01 4.14780527e-02 -1.12186097e-01
3.91337305e-01 1.02469313e+00 1.57890528e-01 -4.17749196e-01
-7.15737998e-01 -7.79544055e-01 7.35140666e-02 -1.03004634e+00
-7.71294301e-03 4.82178271e-01 -3.89069736e-01 -4.60080385e-01
-1.09752826e-01 7.82348633e-01 -5.37043214e-01 -7.08291173e-01
1.54729187e+00 1.25489628e+00 8.83053839e-02 -3.25788140e-01
1.06897525e-01 6.25921428e-01 -6.17542922e-01 -1.10527121e-01
4.79399920e-01 -2.68677711e-01 -4.38746601e-01 4.26290482e-01
-7.30448186e-01 -4.72739428e-01 -3.53815794e-01 1.47597814e+00
4.90491003e-01 2.69937873e-01 3.29732776e-01 -2.31374478e+00
-3.52375925e-01 8.18612874e-01 -7.77260303e-01 2.19324097e-01
1.46661639e-01 7.92179704e-02 1.38202226e+00 -5.00383079e-01
9.61762890e-02 8.96150529e-01 1.76889986e-01 -7.75924623e-02
-1.20003176e+00 -6.00254595e-01 4.38143224e-01 5.60166597e-01
-1.33532846e+00 1.22556537e-01 1.02867687e+00 -5.34006953e-02
1.19913673e+00 3.62141728e-01 1.57620978e+00 2.63406225e-02
4.09222841e-02 9.22910154e-01 8.23108137e-01 2.34289750e-01
5.04474163e-01 -2.27537844e-02 -1.01507194e-01 -5.19278571e-02
6.00678980e-01 -4.50367272e-01 1.92944154e-01 -4.81133386e-02
4.95661348e-01 5.87989129e-02 -6.54805779e-01 -7.44302869e-01
-7.05999792e-01 9.08222497e-01 6.64371908e-01 3.97399604e-01
-2.73547947e-01 -3.91689599e-01 4.19716001e-01 3.62422802e-02
4.43678558e-01 3.46800596e-01 -4.15411681e-01 -1.03181377e-01
-1.01072979e+00 3.07896771e-02 1.18686175e+00 1.23311508e+00
2.71926999e-01 8.52393866e-01 2.01848835e-01 8.08931589e-01
6.60975695e-01 7.34302580e-01 -1.08229585e-01 -7.40270495e-01
4.68141615e-01 7.32878327e-01 2.66673360e-02 -9.16983128e-01
-6.25568867e-01 -1.05458391e+00 -1.03588808e+00 -6.70621246e-02
1.28718000e-02 -4.29432243e-01 4.16947119e-02 1.26493573e+00
2.16921791e-01 -3.42749864e-01 -9.84154344e-02 1.08777738e+00
1.04152068e-01 1.06578171e+00 -2.33543143e-01 -8.46641898e-01
1.26054907e+00 -8.66387606e-01 -1.12958181e+00 1.83929518e-01
2.01193586e-01 -4.35440153e-01 2.91829020e-01 3.65233988e-01
-1.27535522e+00 3.67707521e-01 -1.25886655e+00 1.41918898e-01
-3.45416516e-02 6.27419204e-02 3.93046111e-01 4.20224786e-01
-1.03290176e+00 1.45008445e-01 -5.84890187e-01 -2.86704689e-01
6.43634021e-01 4.61831808e-01 1.16664082e-01 6.04969151e-02
-1.04159665e+00 1.05501437e+00 3.24412078e-01 7.37725079e-01
-8.70896459e-01 -1.28699183e+00 -7.46764004e-01 7.08261192e-01
7.41087258e-01 -7.01214790e-01 1.09180593e+00 -5.35590708e-01
-1.52345037e+00 -9.13757905e-02 2.63574898e-01 -1.43231452e-01
4.88186955e-01 3.09461802e-01 -6.77297115e-01 1.28592104e-01
3.22772078e-02 -3.13355207e-01 2.69889027e-01 -1.27364457e+00
-7.36180246e-01 -2.32639670e-01 1.80350631e-01 3.79241824e-01
-1.96142152e-01 -2.25237846e-01 3.52353215e-01 -7.37490654e-01
1.34637713e-01 -2.98166901e-01 -6.60405159e-01 -9.70429629e-02
-7.95553923e-01 -4.40150589e-01 1.17359900e+00 -7.88907409e-01
1.29604542e+00 -1.78775954e+00 3.50540578e-01 7.81036615e-01
2.57690936e-01 -5.77546982e-03 1.48573816e-01 7.24601150e-01
-2.56392032e-01 -1.82311550e-01 -2.91554004e-01 2.69042198e-02
6.90265417e-01 5.20279348e-01 -2.15646967e-01 6.15344763e-01
2.90229172e-02 7.76926696e-01 -8.53531897e-01 1.26956463e-01
4.11136299e-01 2.53778815e-01 -6.70546830e-01 -7.93589279e-02
5.88675030e-03 1.54098317e-01 -6.19933844e-01 6.61043704e-01
1.11321366e+00 -1.59273118e-01 5.05949199e-01 -7.30299413e-01
-4.01201785e-01 7.18467217e-03 -1.36831200e+00 1.35232770e+00
-4.61819857e-01 3.55102539e-01 7.93387055e-01 -1.55429149e+00
7.06442356e-01 1.53500214e-01 1.15889192e+00 -8.17589581e-01
7.65007660e-02 3.37056488e-01 2.48647854e-01 -2.23497644e-01
1.94993392e-01 -3.11898202e-01 9.41020325e-02 4.14944232e-01
1.09913260e-01 -3.13677430e-01 7.90181875e-01 3.72691602e-01
6.42891705e-01 -2.93307304e-01 3.79435003e-01 -9.86066163e-01
1.03165758e+00 -9.19733122e-02 1.27807128e+00 -4.24331278e-01
-9.05219764e-02 -1.18456639e-01 1.21731198e+00 7.59043023e-02
-1.15952408e+00 -1.18905818e+00 -1.92512035e-01 1.17661797e-01
2.30150133e-01 -2.97325402e-01 -4.18564975e-01 -5.82052469e-01
1.03428438e-01 9.32451189e-01 3.20756644e-01 9.52368081e-02
-3.35222185e-01 -1.07527208e+00 -2.16265440e-01 4.62289810e-01
7.24241436e-01 -1.62688643e-01 1.85801357e-01 4.71105427e-01
-1.47455288e-02 -1.03673530e+00 -5.71326137e-01 2.07373396e-01
-4.60233092e-01 -1.18888438e+00 -7.83608377e-01 -7.73835719e-01
1.07641745e+00 -1.42774396e-02 9.57250118e-01 -2.85551906e-01
-6.36431500e-02 3.40635151e-01 -1.55281812e-01 5.95666394e-02
1.79334655e-01 -4.18846272e-02 3.57477874e-01 -8.59612450e-02
-5.40097840e-02 -9.55027699e-01 -6.29294038e-01 4.33248788e-01
-7.34952331e-01 -1.09438635e-01 1.69849262e-01 7.92612612e-01
4.43236023e-01 8.99476290e-01 1.10009146e+00 -1.94912940e-01
7.91295171e-01 -7.84842253e-01 -1.37632394e+00 2.30632827e-01
-7.32432961e-01 -4.26469117e-01 1.13077569e+00 1.68346420e-01
-9.41773474e-01 -3.56151640e-01 -1.85885042e-01 -2.70488858e-01
6.16103649e-01 6.63923502e-01 -1.12239480e+00 -2.39608884e-01
-7.41272032e-01 3.08160216e-01 1.01858094e-01 -4.62919056e-01
-6.08572848e-02 4.40698773e-01 -9.48295966e-02 -4.78906482e-01
1.41960573e+00 3.70901614e-01 4.06857312e-01 -5.21424472e-01
-4.52672184e-01 -1.86517864e-01 -2.89250761e-01 -1.26862869e-01
4.44755882e-01 -8.20713460e-01 -1.38885164e+00 4.12914991e-01
-8.64548743e-01 -9.69832018e-03 -5.52379251e-01 4.44152832e-01
-2.66179860e-01 3.08796287e-01 -5.26469827e-01 -6.67986333e-01
-2.36851558e-01 -1.35215068e+00 2.00350285e-01 3.79349589e-01
3.99379820e-01 -1.40667832e+00 -3.03054869e-01 1.77574918e-01
7.13581383e-01 1.86049834e-01 1.17121351e+00 -4.41352844e-01
-8.29300284e-01 3.52923959e-01 -3.47891390e-01 7.60586739e-01
2.25278288e-01 -2.95375995e-02 -1.10649072e-01 -8.44877958e-01
2.09252700e-01 3.53473991e-01 -2.74304897e-01 3.93366128e-01
8.76584411e-01 -8.41514230e-01 -2.29517356e-01 7.34676063e-01
2.29983377e+00 5.52681863e-01 3.91385585e-01 2.46902838e-01
5.98045588e-01 3.54343921e-01 4.39450234e-01 9.56898391e-01
8.68448436e-01 4.99647468e-01 7.38218725e-01 -2.57653952e-01
5.81254959e-01 -8.33479688e-02 2.79086351e-01 1.22898996e+00
1.80989236e-01 -4.34400082e-01 -5.78669846e-01 5.32361567e-01
-1.68391466e+00 -7.40698695e-01 -2.43395254e-01 1.53507733e+00
4.91963252e-02 -2.74794966e-01 1.18095264e-01 4.43831712e-01
4.58314359e-01 -3.80376130e-02 -7.86964655e-01 -4.55579311e-01
-3.14809263e-01 -9.19015706e-02 6.93125784e-01 2.12902531e-01
-5.44370890e-01 -9.51381102e-02 5.08881617e+00 1.08537722e+00
-9.57959235e-01 9.76489633e-02 4.15528327e-01 -2.21650720e-01
-7.55675793e-01 1.73512012e-01 -4.65637714e-01 7.64522970e-01
2.90810913e-01 -1.22021759e+00 8.33289027e-01 7.03174829e-01
1.13782012e+00 -5.72345890e-02 -7.64868736e-01 1.07117510e+00
-2.73256928e-01 -9.82469678e-01 -1.74654961e-01 3.23845863e-01
1.02465355e+00 -1.86486751e-01 -3.73876393e-01 1.11149155e-01
4.41954643e-01 -6.27622485e-01 6.61915898e-01 8.00661668e-02
3.77708852e-01 -1.23406374e+00 6.49293303e-01 9.06029046e-02
-1.46609962e+00 -5.74893236e-01 -4.03481275e-02 -4.07459289e-02
1.06137931e+00 9.56879258e-01 -2.87561685e-01 1.24294317e+00
7.22798347e-01 1.35082912e+00 5.53115346e-02 1.38358259e+00
-4.81582791e-01 4.91914630e-01 -4.64449286e-01 2.72183102e-02
2.93779820e-01 -1.00285327e+00 7.01018214e-01 2.14485586e-01
5.29125750e-01 2.16428384e-01 4.06767368e-01 1.04496884e+00
-9.46581364e-02 1.99067906e-01 -1.02255918e-01 1.11720659e-01
5.48499882e-01 1.69905448e+00 -5.05964518e-01 -3.39114852e-02
-7.53492355e-01 2.43986115e-01 -3.87288183e-01 8.23471367e-01
-8.45919788e-01 -5.49007833e-01 1.30970061e+00 -3.30773741e-02
1.37160301e-01 -2.03814566e-01 -4.17874396e-01 -1.50824380e+00
1.56102911e-01 -7.53847182e-01 3.31194609e-01 -8.26905608e-01
-1.72240973e+00 2.41647840e-01 4.72417139e-02 -1.56875813e+00
-5.43795759e-03 -3.51427287e-01 -8.46861184e-01 9.66745853e-01
-2.03035688e+00 -9.74010706e-01 5.67951947e-02 5.84155381e-01
3.38764012e-01 -4.88461182e-02 1.84927315e-01 8.85026336e-01
-1.18891513e+00 1.59801617e-01 7.21724153e-01 4.52347653e-04
-3.62048090e-01 -1.29847205e+00 -6.80946589e-01 1.17680597e+00
-8.37750137e-01 4.97881323e-02 6.51521623e-01 -2.53337532e-01
-1.78588307e+00 -8.40667188e-01 5.71114242e-01 6.68210864e-01
1.26543212e+00 -4.28039640e-01 -5.84802151e-01 6.43971384e-01
8.67813706e-01 -7.16047809e-02 5.73298216e-01 -5.30190468e-01
4.77015108e-01 -7.97919869e-01 -1.60010529e+00 4.76775467e-01
6.48322642e-01 -9.55160260e-02 -1.89718828e-01 3.43443543e-01
2.59393036e-01 -1.80965647e-01 -1.34911680e+00 4.11995590e-01
-1.15562096e-01 -7.35353231e-02 7.54133224e-01 -1.19212329e-01
3.43398340e-02 -8.56591046e-01 -1.10958725e-01 -2.10723519e+00
-3.42411548e-01 -8.98248911e-01 -2.18160331e-01 1.50176394e+00
1.70110717e-01 -1.27236784e+00 4.63568777e-01 4.73024905e-01
-2.65088022e-01 -8.41552913e-01 -1.29003131e+00 -9.30171072e-01
8.57907459e-02 7.82482028e-02 1.09862626e+00 1.14875638e+00
8.25243771e-01 2.73097277e-01 8.77498910e-02 3.57771397e-01
9.97489572e-01 -1.77808143e-02 -3.03212199e-02 -9.45879281e-01
7.24945515e-02 -6.12679839e-01 -3.10015470e-01 -1.13609803e+00
2.11296201e-01 -1.08894050e+00 -3.84478390e-01 -2.12937999e+00
4.17194329e-02 -7.08304644e-01 -1.76852271e-01 3.53454590e-01
4.45763171e-01 -3.34109329e-02 5.76211333e-01 -2.62895525e-02
-2.30809510e-01 1.25598192e+00 1.53963017e+00 -4.12857860e-01
2.58635491e-01 -5.72875813e-02 -7.50902295e-01 4.69306767e-01
7.39499509e-01 1.01704754e-01 -7.21871912e-01 -2.57836759e-01
3.60553503e-01 4.73815382e-01 -1.31547347e-01 -6.17010832e-01
4.27833885e-01 -2.75909036e-01 -4.92537692e-02 -9.28057015e-01
5.78889512e-02 -1.61379504e+00 4.04421449e-01 7.15161979e-01
4.26730454e-01 2.63971865e-01 3.99119519e-02 9.46015194e-02
-4.48147446e-01 1.31886685e-02 5.70744932e-01 4.67541456e-01
-5.38018048e-01 6.25628889e-01 -7.11248755e-01 -5.46990447e-02
1.50551689e+00 -2.46174559e-02 -5.93098760e-01 -3.79991740e-01
-8.04576516e-01 1.38963449e+00 9.13648084e-02 3.70552629e-01
3.41560096e-01 -1.51645100e+00 -5.44850230e-01 1.53904423e-01
-6.89292789e-01 3.80339958e-02 3.52662802e-01 1.41334438e+00
-6.23334229e-01 7.71114230e-01 -2.21198499e-01 -5.37234128e-01
-5.95737457e-01 2.95726329e-01 6.47640347e-01 -5.21090925e-01
-3.40547442e-01 6.43456951e-02 -2.02398404e-01 -2.55791456e-01
-2.18361542e-01 -4.61670011e-01 -1.98611096e-01 3.72928977e-01
-4.88835014e-02 1.10573757e+00 -1.51069602e-02 -5.42371929e-01
-4.23377275e-01 3.51953030e-01 4.24433380e-01 3.71790200e-01
1.77199268e+00 -6.51854873e-01 -7.49741733e-01 -2.67209858e-01
1.27916634e+00 -1.39078274e-01 -1.01389039e+00 2.68288463e-01
-3.09325159e-01 -3.78387272e-01 3.09841096e-01 -7.68229067e-01
-2.04900646e+00 3.18825155e-01 6.30075410e-02 7.00410306e-01
1.59027088e+00 -3.30646724e-01 7.96284258e-01 -1.19037516e-01
7.60084093e-01 -1.10351515e+00 -5.12086570e-01 3.50248814e-01
7.63460219e-01 -5.70719957e-01 1.56954765e-01 -7.63203919e-01
-2.18693808e-01 1.15066981e+00 5.90113044e-01 -1.29553065e-01
1.00991917e+00 5.53353131e-01 -4.83052939e-01 1.04317158e-01
-4.90484804e-01 8.22047740e-02 -2.93823540e-01 3.58521730e-01
-1.67740539e-01 1.78429186e-01 -9.35350657e-01 8.77270579e-01
-1.57924324e-01 1.35244727e-02 8.99242043e-01 8.49700034e-01
1.34266704e-01 -1.18619752e+00 -3.09495389e-01 7.08634257e-01
-3.65924895e-01 1.09535865e-01 6.82434559e-01 5.49420655e-01
-3.08718346e-02 1.08591866e+00 2.38072753e-01 2.88114607e-01
7.08987892e-01 -6.30569637e-01 2.40050510e-01 -1.33684888e-01
-3.70891958e-01 2.13532448e-01 -1.19062066e-01 -3.93835187e-01
-2.01201513e-01 -8.20921183e-01 -1.30322754e+00 -7.57328629e-01
-4.83991265e-01 6.81157708e-01 5.80520689e-01 1.03507113e+00
-1.36851877e-01 8.19877744e-01 1.24625862e+00 -5.89715540e-01
-6.73977852e-01 -4.55053449e-01 -1.44133568e+00 -2.68413007e-01
2.28281468e-01 -4.70506519e-01 -6.64899707e-01 -6.18662000e-01]
|
[5.669195175170898, 2.561311721801758]
|
7ba92acc-a73f-4057-8b91-9c87fa3cf327
|
is-chatgpt-the-ultimate-programming-assistant
|
2304.11938
| null |
https://arxiv.org/abs/2304.11938v1
|
https://arxiv.org/pdf/2304.11938v1.pdf
|
Is ChatGPT the Ultimate Programming Assistant -- How far is it?
|
The recent progress in generative AI techniques has significantly influenced software engineering, as AI-driven methods tackle common developer challenges such as code synthesis from descriptions, program repair, and natural language summaries for existing programs. Large-scale language models (LLMs), like OpenAI's Codex, are increasingly adopted in AI-driven software engineering. ChatGPT, another LLM, has gained considerable attention for its potential as a bot for discussing source code, suggesting changes, providing descriptions, and generating code. To evaluate the practicality of LLMs as programming assistant bots, it is essential to examine their performance on unseen problems and various tasks. In our paper, we conduct an empirical analysis of ChatGPT's potential as a fully automated programming assistant, emphasizing code generation, program repair, and code summarization. Our study assesses ChatGPT's performance on common programming problems and compares it to state-of-the-art approaches using two benchmarks. Our research indicates that ChatGPT effectively handles typical programming challenges. However, we also discover the limitations in its attention span: comprehensive descriptions can restrict ChatGPT's focus and impede its ability to utilize its extensive knowledge for problem-solving. Surprisingly, we find that ChatGPT's summary explanations of incorrect code provide valuable insights into the developer's original intentions. This insight can be served as a foundation for future work addressing the oracle problem. Our study offers valuable perspectives on the development of LLMs for programming assistance, specifically by highlighting the significance of prompt engineering and enhancing our comprehension of ChatGPT's practical applications in software engineering.
|
['Tegawendé F. Bissyandé', 'Jacques Klein', 'Shing-Chi Cheung', 'Xunzhu Tang', 'Tsz On Li', 'Weiqi Lu', 'Haoye Tian']
|
2023-04-24
| null | null | null | null |
['program-repair', 'prompt-engineering', 'program-repair']
|
['computer-code', 'natural-language-processing', 'reasoning']
|
[ 1.93356141e-01 6.26938760e-01 -2.56221682e-01 -1.51054189e-01
-8.39529634e-01 -6.58493757e-01 3.02719593e-01 1.57600209e-01
4.31478381e-01 1.52742922e-01 1.71768755e-01 -6.96722209e-01
1.77159123e-02 -6.46430433e-01 -5.86387873e-01 -8.44735727e-02
1.11305609e-01 2.14606419e-01 1.15940839e-01 -3.87676746e-01
6.23509288e-01 -1.20027244e-01 -1.42162597e+00 4.04245257e-01
1.21767306e+00 -7.74618387e-02 4.94194955e-01 6.43290937e-01
-5.94014466e-01 1.42702329e+00 -7.94131398e-01 -7.48610258e-01
-7.29708523e-02 -5.13521731e-01 -1.09911203e+00 -2.32006431e-01
9.48208123e-02 -2.65367687e-01 2.52407081e-02 1.19982314e+00
2.73072600e-01 -3.25915962e-01 2.14869410e-01 -1.76698685e+00
-1.07132471e+00 1.36686659e+00 -5.13637602e-01 -2.35387683e-02
6.35705829e-01 6.13459945e-01 1.22933328e+00 -6.17529392e-01
7.11179197e-01 1.05306232e+00 7.45946646e-01 1.00153840e+00
-1.15651846e+00 -5.86996973e-01 1.16199456e-01 6.37612417e-02
-1.12863266e+00 -3.35272402e-01 8.90061498e-01 -9.08040941e-01
1.46234155e+00 4.05231267e-01 5.94787896e-01 1.11643231e+00
2.66637087e-01 1.07150888e+00 6.16928875e-01 -6.86829984e-01
1.01682469e-01 3.78284663e-01 2.91140825e-01 9.48043406e-01
2.43046865e-01 -9.03772265e-02 -2.92114675e-01 -5.48878074e-01
4.32820231e-01 -3.58903795e-01 -1.25708550e-01 -4.15312909e-02
-1.05256021e+00 8.72094929e-01 7.97804222e-02 5.18741786e-01
-6.99596182e-02 6.30717874e-01 3.88458639e-01 3.36442083e-01
3.48556668e-01 1.12437272e+00 -2.49531895e-01 -9.01656210e-01
-6.36071980e-01 5.58172822e-01 1.25963831e+00 1.59791422e+00
6.13027871e-01 2.61769503e-01 -5.26199862e-02 5.14374733e-01
3.69486153e-01 2.75877357e-01 3.52922738e-01 -1.07836545e+00
6.25281811e-01 1.16795552e+00 -1.13587849e-01 -1.12839258e+00
5.73613979e-02 -3.11391979e-01 -1.44883186e-01 6.33674115e-02
1.59472048e-01 -1.96762905e-01 -2.46566728e-01 1.45791006e+00
-2.31226534e-01 -2.54078120e-01 5.98873980e-02 5.73328316e-01
7.86842465e-01 7.16472328e-01 -8.65270123e-02 1.01895511e-01
1.13596416e+00 -1.20900869e+00 -3.40057611e-01 -7.08178282e-01
1.15301859e+00 -6.46325350e-01 1.29978192e+00 2.50053436e-01
-1.07024932e+00 -1.89929530e-01 -7.52152085e-01 -1.08684646e-02
5.90490177e-02 2.63540894e-01 9.71693695e-01 5.82663000e-01
-1.05236626e+00 3.63951832e-01 -8.90720725e-01 -5.36097884e-01
4.22216117e-01 2.29034275e-02 1.29379883e-01 -1.40825719e-01
-5.54414690e-01 8.18824887e-01 5.42627200e-02 -1.63731232e-01
-9.07248974e-01 -8.47261786e-01 -9.22730267e-01 2.62012213e-01
6.26324117e-01 -6.61024153e-01 1.79618549e+00 -8.74064684e-01
-1.22671258e+00 6.70882225e-01 -6.45858720e-02 -2.81331807e-01
2.18941316e-01 -6.99007213e-02 -4.93705980e-02 -4.81757849e-01
3.52489680e-01 3.88680428e-01 5.65941036e-01 -1.16429186e+00
-4.60147977e-01 1.96730241e-01 5.20277500e-01 -3.46811324e-01
-2.23799869e-01 4.47747469e-01 -1.52777269e-01 -6.48275673e-01
-3.09164524e-01 -9.84589636e-01 -4.27737176e-01 -2.27374479e-01
-5.70143402e-01 -5.48574090e-01 5.10827065e-01 -6.06301188e-01
1.64394462e+00 -2.17359328e+00 4.25195903e-01 -1.47709817e-01
5.38780868e-01 2.38758892e-01 -4.30101126e-01 1.02353704e+00
1.69934541e-01 6.92300081e-01 -2.80284107e-01 -1.23129515e-02
3.69998574e-01 2.23976802e-02 -5.03420115e-01 -2.56623238e-01
5.52431703e-01 1.31448853e+00 -1.01461124e+00 -3.46028835e-01
-2.27718860e-01 -7.90191963e-02 -9.43011761e-01 2.38151744e-01
-8.27750504e-01 9.71000716e-02 -6.28676414e-01 7.12528169e-01
4.67016660e-02 -3.82828355e-01 4.51399060e-03 3.19017828e-01
-3.35603952e-01 3.69495392e-01 -5.40333867e-01 1.61943531e+00
-4.79147583e-01 1.07033420e+00 -1.48175880e-01 -5.96744657e-01
1.09763682e+00 1.85322583e-01 -9.77126509e-02 -2.83843577e-01
-9.31028873e-02 2.04663619e-01 4.28216040e-01 -1.02041042e+00
4.92010087e-01 3.74258906e-01 -2.45155051e-01 1.07716048e+00
-2.75679022e-01 -4.68435019e-01 3.63215536e-01 5.13133645e-01
1.49877954e+00 2.04428673e-01 4.04480368e-01 -2.82861888e-01
2.87308544e-01 6.32965386e-01 3.03911626e-01 1.02104568e+00
2.12059349e-01 2.47637734e-01 9.83035982e-01 -4.22163188e-01
-1.03095961e+00 -3.59711438e-01 4.54943925e-01 1.18724775e+00
-1.60170034e-01 -1.13477468e+00 -9.41685736e-01 -8.61614704e-01
-1.95992917e-01 1.22741890e+00 -3.82358551e-01 -3.24637562e-01
-7.86043525e-01 -4.44368690e-01 8.95627558e-01 5.99233270e-01
1.10900491e-01 -1.43480623e+00 -8.70094061e-01 3.06815714e-01
-2.80870050e-01 -6.55013561e-01 -3.70198131e-01 -5.64884059e-02
-6.25108540e-01 -1.15429688e+00 -2.34316468e-01 -8.46839845e-01
8.73529077e-01 3.24797094e-01 1.29123139e+00 6.34947777e-01
-3.35552096e-01 5.92782378e-01 -6.29656196e-01 -6.34491444e-01
-1.37273943e+00 4.36103344e-01 -5.33323705e-01 -8.48708153e-01
4.90778387e-01 -7.02578366e-01 -1.01558175e-02 2.22961143e-01
-7.45086312e-01 4.38240677e-01 6.57407522e-01 5.37995338e-01
-3.32324684e-01 -7.09083751e-02 3.87169808e-01 -1.29272485e+00
1.07764077e+00 -6.26078129e-01 -6.21198356e-01 3.62869650e-01
-6.74462736e-01 1.53761879e-01 5.03268301e-01 -4.61110234e-01
-1.08195329e+00 -3.74931008e-01 -5.82146607e-02 1.01525873e-01
-4.62051928e-02 1.04001832e+00 2.82182574e-01 -2.04582840e-01
1.05850446e+00 4.02637899e-01 -1.24982990e-01 -3.92883688e-01
2.70040959e-01 6.83162332e-01 3.74600589e-01 -1.07543015e+00
1.03053403e+00 -3.11717510e-01 -7.35165358e-01 -5.66139102e-01
-4.73073930e-01 -1.05189912e-01 -7.39037171e-02 9.39524546e-02
3.86078358e-01 -5.93480885e-01 -5.47381818e-01 1.29813418e-01
-1.46972203e+00 -5.92266500e-01 -2.34791324e-01 -1.40173718e-01
-4.14221346e-01 2.65386313e-01 -6.27091169e-01 -8.31661880e-01
-3.85759652e-01 -1.48005116e+00 8.76891255e-01 1.62488982e-01
-9.20505047e-01 -7.62172282e-01 2.39561722e-01 5.60073614e-01
7.03722835e-01 5.24926977e-03 1.58171940e+00 -6.80196702e-01
-9.02815580e-01 -1.18150890e-01 -9.10691991e-02 8.76711085e-02
-1.64829530e-02 5.14540851e-01 -6.68169320e-01 -7.75589049e-02
-9.18400958e-02 -1.67888135e-01 1.20928667e-01 -1.15728945e-01
9.61802065e-01 -6.38023198e-01 -3.43960643e-01 2.77075768e-01
1.16408026e+00 4.73074526e-01 4.59884793e-01 4.06594604e-01
7.43554533e-01 6.31031692e-01 5.61860442e-01 5.22539318e-01
5.39908111e-01 3.99568319e-01 5.22324383e-01 3.57011795e-01
3.81848142e-02 -4.25774485e-01 5.95742702e-01 9.64106262e-01
2.39933692e-02 -2.37006098e-01 -1.46929240e+00 7.05266476e-01
-2.03527355e+00 -7.89104521e-01 -2.48477802e-01 1.49167085e+00
8.60196292e-01 -8.63253102e-02 -2.89196000e-02 -2.08739296e-01
5.19622266e-01 -6.64645061e-02 -4.63298559e-01 -6.38567328e-01
5.48163950e-01 -5.19260541e-02 -1.87530890e-01 1.12418905e-01
-5.26811302e-01 1.02792597e+00 6.02563763e+00 6.30329072e-01
-9.54837322e-01 3.56641449e-02 3.95905711e-02 3.40341747e-01
-6.91903651e-01 4.58734542e-01 -6.79178298e-01 3.15796375e-01
6.84664965e-01 -8.01481009e-01 8.33574295e-01 1.48084426e+00
2.62970594e-03 9.44035492e-05 -1.59342110e+00 5.75950682e-01
8.76229927e-02 -1.43025362e+00 -5.90295240e-04 -4.93748300e-02
6.56196535e-01 -2.95325518e-02 -1.33332565e-01 7.14315236e-01
6.62083626e-01 -8.38331938e-01 9.16096807e-01 1.22146003e-01
3.28580946e-01 -5.21180034e-01 6.28707051e-01 8.12691510e-01
-8.27987254e-01 -2.51242906e-01 -2.80150980e-01 -5.27508080e-01
-1.46809459e-01 9.65694711e-02 -1.24982357e+00 1.72312647e-01
3.46451789e-01 7.91355908e-01 -8.29325736e-01 7.97207057e-01
-4.36415881e-01 8.18414509e-01 2.00686678e-01 -5.26354492e-01
1.54148668e-01 1.51114419e-01 7.93123305e-01 1.24919987e+00
3.79062861e-01 1.13152049e-01 3.81179482e-01 1.80788255e+00
2.07320288e-01 -7.15830475e-02 -7.79443562e-01 -8.61379743e-01
5.90865970e-01 1.11484718e+00 -4.92654830e-01 -2.06386536e-01
-5.35158753e-01 4.38848555e-01 2.68805414e-01 2.75654465e-01
-8.36631954e-01 -6.38692021e-01 5.59475422e-01 1.66036904e-01
-3.60992402e-02 -2.28305683e-01 -3.73069853e-01 -1.07265866e+00
2.10314929e-01 -1.38609529e+00 -1.33704334e-01 -1.01132691e+00
-9.34078872e-01 7.56354094e-01 1.70491263e-01 -9.82450366e-01
-4.96167958e-01 -1.13378592e-01 -1.02316475e+00 5.36973655e-01
-9.47914064e-01 -1.25534379e+00 -3.51377517e-01 -4.62963767e-02
9.22206283e-01 -4.48373973e-01 7.65131295e-01 -1.37633923e-02
-6.54908419e-01 5.56356192e-01 -5.54027379e-01 1.86101913e-01
2.64669299e-01 -1.13419139e+00 1.09627330e+00 1.13647223e+00
1.15745828e-01 1.15319347e+00 8.97918344e-01 -7.24877477e-01
-1.83296525e+00 -1.04229844e+00 8.59443307e-01 -7.84120679e-01
9.88056898e-01 -2.83100218e-01 -9.45923030e-01 1.01102018e+00
1.03747934e-01 -6.07293427e-01 3.34818393e-01 7.82972947e-02
-3.00210685e-01 3.29645872e-01 -7.95049012e-01 7.82067776e-01
1.06894290e+00 -5.79154611e-01 -6.06034875e-01 4.13240254e-01
9.47774827e-01 -5.08047283e-01 -6.63825631e-01 -1.46380231e-01
3.07871222e-01 -8.55961680e-01 4.01539832e-01 -5.03585875e-01
1.22868085e+00 -1.89366266e-01 1.60788655e-01 -1.35925698e+00
-2.66636521e-01 -1.01891255e+00 1.10593274e-01 1.61225748e+00
4.88002300e-01 -5.66790164e-01 5.60877800e-01 1.08028388e+00
-4.53490704e-01 -6.91476762e-01 -1.52951449e-01 -5.35063922e-01
-5.78171834e-02 -7.30831802e-01 5.89114249e-01 9.88471985e-01
6.70827866e-01 2.09383816e-01 -1.97534934e-01 2.68740989e-02
1.25263616e-01 3.42657357e-01 1.23079777e+00 -1.05021250e+00
-6.83737993e-01 -5.61337650e-01 -9.77488086e-02 -7.09863663e-01
4.59215730e-01 -1.07740712e+00 2.19760060e-01 -1.43964052e+00
4.67656970e-01 -4.62452739e-01 6.13343477e-01 8.81689787e-01
1.48951588e-02 -3.22841257e-01 4.05904800e-01 2.43191943e-01
-4.08128649e-01 3.73244211e-02 1.01172364e+00 -4.75399703e-01
-4.54091996e-01 1.39058232e-01 -1.24394059e+00 7.66116619e-01
7.42339134e-01 -8.79206598e-01 -5.68925440e-01 -8.14123333e-01
7.90842533e-01 2.55478054e-01 3.20735723e-01 -8.62937331e-01
4.51581210e-01 -4.96947676e-01 -7.58337438e-01 1.82594821e-01
-3.81095737e-01 -6.26155674e-01 3.48038971e-01 5.66779375e-01
-5.66282332e-01 2.08699644e-01 2.57881403e-01 1.49989694e-01
-1.27657175e-01 -7.75845110e-01 2.03620002e-01 -4.88249719e-01
-8.81016731e-01 -5.59361503e-02 -9.35734093e-01 2.12763727e-01
1.01146603e+00 -2.44206607e-01 -7.65526652e-01 -2.92471230e-01
-2.80806363e-01 2.32940346e-01 7.51929700e-01 7.80125439e-01
5.92347741e-01 -8.44148099e-01 -6.63926780e-01 2.58391917e-01
4.73393887e-01 -3.18260603e-02 -1.37991413e-01 6.76319480e-01
-6.96762085e-01 3.53550017e-01 -6.17819987e-02 -5.77494562e-01
-1.36885285e+00 4.79874045e-01 1.16240140e-02 -1.29620239e-01
-7.97376335e-01 7.48176336e-01 3.46763670e-01 -4.44163620e-01
5.98216802e-02 -6.26765430e-01 1.38774946e-01 -5.06731093e-01
4.66993898e-01 3.14876884e-01 -2.71858156e-01 5.18106110e-02
-2.74262965e-01 2.48576120e-01 -2.80478835e-01 1.52668610e-01
1.62083542e+00 1.40419275e-01 -5.91926932e-01 4.08592075e-01
5.38073778e-01 1.43166989e-01 -7.54654408e-01 -5.18794581e-02
4.66718018e-01 -3.68396550e-01 -3.34973544e-01 -9.40537393e-01
-6.69449151e-01 8.20794225e-01 -2.51274139e-01 3.81668031e-01
6.93091094e-01 3.08203846e-01 5.69857299e-01 7.71734297e-01
7.59568334e-01 -3.34850580e-01 2.06625611e-01 5.95346153e-01
1.10637307e+00 -1.19731057e+00 -2.50936300e-01 -5.16846180e-01
-7.83766925e-01 1.32493544e+00 1.07904828e+00 4.25048023e-02
-1.21383771e-01 6.83272779e-01 -7.89607875e-03 -5.04495025e-01
-1.24089515e+00 3.05011868e-01 3.92493494e-02 7.12012231e-01
7.50048578e-01 -1.17039092e-01 3.58586200e-02 7.47634828e-01
-3.66739839e-01 8.72488916e-02 1.08761322e+00 1.24252760e+00
-3.87203246e-01 -1.18106878e+00 -1.47995189e-01 2.99650818e-01
-2.49414846e-01 -3.96860391e-01 -6.87983334e-01 9.22597349e-01
-1.57750711e-01 8.69241416e-01 -3.87177825e-01 -4.56564665e-01
2.62247205e-01 8.01126063e-02 4.47564721e-01 -1.26217592e+00
-9.12500381e-01 -3.82563382e-01 2.66104728e-01 -5.50587356e-01
-6.05233498e-02 -3.85576457e-01 -1.06720340e+00 -4.44452375e-01
-3.61452341e-01 3.80621582e-01 5.57844996e-01 9.22535360e-01
5.42855382e-01 5.81837416e-01 9.03528631e-02 -4.97854799e-01
-6.56253517e-01 -7.77347147e-01 7.20545053e-02 -1.36909589e-01
1.29270330e-01 -1.69154614e-01 -1.42486855e-01 2.91164517e-01]
|
[7.91172456741333, 7.698004722595215]
|
539f5b98-6531-41f9-8774-9fc4492fe52e
|
a-data-driven-approach-for-motion-planning-of
|
1904.08784
| null |
https://arxiv.org/abs/1904.08784v4
|
https://arxiv.org/pdf/1904.08784v4.pdf
|
Efficient Motion Planning for Automated Lane Change based on Imitation Learning and Mixed-Integer Optimization
|
Intelligent motion planning is one of the core components in automated vehicles, which has received extensive interests. Traditional motion planning methods suffer from several drawbacks in terms of optimality, efficiency and generalization capability. Sampling based methods cannot guarantee the optimality of the generated trajectories. Whereas the optimization-based methods are not able to perform motion planning in real-time, and limited by the simplified formalization. In this work, we propose a learning-based approach to handle those shortcomings. Mixed Integer Quadratic Problem based optimization (MIQP) is used to generate the optimal lane-change trajectories which served as the training dataset for learning-based action generation algorithms. A hierarchical supervised learning model is devised to make the fast lane-change decision. Numerous experiments have been conducted to evaluate the optimality, efficiency, and generalization capability of the proposed approach. The experimental results indicate that the proposed model outperforms several commonly used motion planning baselines.
|
['Chenyang Xi', 'Yuankai Wu', 'Tianyu Shi', 'Lijun Sun']
|
2019-04-18
| null | null | null | null |
['action-generation']
|
['computer-vision']
|
[ 8.93010050e-02 5.04941028e-03 -6.92460477e-01 -3.15718055e-01
-6.48486972e-01 -9.37851667e-02 7.79930472e-01 -8.76200497e-02
-3.87193292e-01 1.01714325e+00 -2.44728057e-03 -5.51585495e-01
-4.51300204e-01 -8.22819948e-01 -2.79239506e-01 -8.28023970e-01
-1.67746156e-01 4.19151723e-01 4.50969696e-01 -2.55013734e-01
6.60915315e-01 6.57088161e-01 -1.52785575e+00 -1.25113100e-01
1.29966044e+00 5.77199578e-01 5.80649316e-01 3.36359113e-01
-2.15811312e-01 5.75317204e-01 -9.64393392e-02 2.00515941e-01
3.52645487e-01 -3.54712754e-01 -5.70744157e-01 3.50798160e-01
-1.64977834e-01 -4.90445197e-01 -3.48266810e-01 8.54011476e-01
1.93268374e-01 5.16898155e-01 7.11119771e-01 -1.83880103e+00
1.87565148e-01 2.98071913e-02 -2.48595551e-01 5.16340486e-04
1.57023042e-01 3.93018752e-01 7.19336689e-01 -5.71083724e-01
5.97817481e-01 1.34530759e+00 2.76007324e-01 4.22204167e-01
-6.91006362e-01 -2.92984635e-01 2.15539128e-01 6.18744969e-01
-1.33079243e+00 -2.85356939e-01 7.02964842e-01 -4.81713355e-01
8.06397676e-01 1.84579380e-02 6.55735433e-01 3.67172569e-01
6.13486409e-01 8.80885839e-01 8.92040074e-01 -1.92979217e-01
3.07865977e-01 1.58220530e-01 -1.09342694e-01 9.12574828e-01
2.87418753e-01 2.57557750e-01 2.08173484e-01 -1.37988910e-01
5.44544280e-01 -3.04514151e-02 2.17143446e-02 -4.58763927e-01
-1.10977829e+00 9.90116715e-01 2.50168204e-01 2.23096367e-02
-5.83434701e-01 -1.18018702e-01 5.23247063e-01 -1.49142757e-01
-4.27468680e-02 2.63646007e-01 -5.57378232e-02 -2.57589519e-01
-9.22342777e-01 7.67426908e-01 4.39630032e-01 1.14070880e+00
8.62200022e-01 3.97850811e-01 -2.91632414e-01 3.51759970e-01
4.16653067e-01 5.38987935e-01 3.90532017e-01 -1.18982267e+00
6.92235589e-01 7.21309602e-01 5.92526257e-01 -1.41094697e+00
-4.73153770e-01 7.36979842e-02 -6.93933427e-01 4.94265586e-01
2.05152154e-01 -3.26533884e-01 -7.42954493e-01 1.09145761e+00
4.22265291e-01 2.50567555e-01 2.37615019e-01 9.09723818e-01
2.84440309e-01 1.23951244e+00 1.30875379e-01 -6.21711731e-01
5.82692087e-01 -1.26224768e+00 -8.80145252e-01 -1.81046198e-03
8.43852222e-01 -5.23264170e-01 7.87161410e-01 3.09538871e-01
-8.16175520e-01 -6.41292155e-01 -1.19199061e+00 4.03853536e-01
-1.72905818e-01 2.41278797e-01 4.46421087e-01 5.68264067e-01
-7.96548128e-01 5.38187563e-01 -7.18611002e-01 -2.12299839e-01
2.21531630e-01 2.97614038e-01 -1.38829559e-01 -6.55598715e-02
-1.00917482e+00 1.07245159e+00 8.72906983e-01 2.91496247e-01
-9.52816188e-01 -1.34538636e-01 -7.09301412e-01 -1.86567470e-01
4.08634484e-01 -2.70508558e-01 1.24069583e+00 -6.34060144e-01
-1.51500654e+00 1.32919580e-01 -3.88599485e-01 -4.45516109e-01
6.66281044e-01 3.70820947e-02 -3.80491227e-01 1.15099857e-02
1.23950429e-01 5.88734031e-01 6.27512813e-01 -1.17359793e+00
-1.40855217e+00 1.10521443e-01 3.71019319e-02 5.15746117e-01
5.33063300e-02 -3.82167935e-01 -2.96571344e-01 -2.78895557e-01
1.26677439e-01 -1.11319530e+00 -6.39465749e-01 -3.43351692e-01
-2.47894377e-01 -3.70545328e-01 1.05296588e+00 -5.14458060e-01
1.55178404e+00 -1.63744688e+00 -3.28042917e-02 2.34031409e-01
-4.22823936e-01 6.65788472e-01 4.82883789e-02 7.38640249e-01
4.35130090e-01 -1.51835233e-01 -2.72400588e-01 4.21852665e-03
2.64485590e-02 4.40082818e-01 -2.68236667e-01 3.44814181e-01
1.35276124e-01 5.12003183e-01 -9.90435183e-01 -6.52363420e-01
5.67051888e-01 -8.59568194e-02 -3.74135077e-01 2.61789352e-01
-5.40751219e-01 4.68859106e-01 -9.69271541e-01 6.03079259e-01
6.00727260e-01 3.37147534e-01 1.51337132e-01 -1.61015987e-02
-6.44890368e-01 -3.87331024e-02 -1.36095238e+00 1.18361557e+00
-4.71769810e-01 5.39290428e-01 -7.06287250e-02 -1.08199167e+00
1.03240323e+00 2.35917032e-01 7.43054092e-01 -5.88478625e-01
2.36154124e-02 3.71018767e-01 -6.73534200e-02 -8.61521065e-01
8.64095807e-01 1.05375119e-01 -7.22531676e-02 1.85423493e-01
-6.42245293e-01 -1.88766345e-01 2.79810786e-01 -3.17794234e-01
7.02720284e-01 4.16107595e-01 5.23016930e-01 -8.28149244e-02
1.08571899e+00 7.79310882e-01 8.45252633e-01 3.33759606e-01
-3.87171328e-01 -3.29675945e-03 2.06540301e-01 -5.87766945e-01
-1.05314755e+00 -5.36605954e-01 2.29040742e-01 4.67718750e-01
5.43323874e-01 -9.43527445e-02 -6.24001682e-01 -5.80795825e-01
-1.56022608e-01 1.03413939e+00 1.17374025e-01 1.88263543e-02
-8.37287545e-01 -4.95474637e-01 2.30269819e-01 1.26013115e-01
7.76118457e-01 -9.86556172e-01 -8.51759851e-01 4.96731937e-01
-2.43859023e-01 -1.11943042e+00 -2.57567555e-01 -6.65980220e-01
-9.47817087e-01 -1.06997359e+00 -3.60763222e-01 -7.26363182e-01
5.92436790e-01 6.47188783e-01 2.11812183e-01 9.06775817e-02
8.63336250e-02 1.63494274e-02 -2.81071275e-01 -4.73816097e-01
-5.75976193e-01 1.28941908e-01 9.06942189e-02 1.59756094e-01
8.33103210e-02 -1.80196494e-01 -4.76897418e-01 5.80995023e-01
-6.94918573e-01 1.09519489e-01 1.00520897e+00 6.49974883e-01
5.86948395e-01 5.65027475e-01 8.10591817e-01 -6.45302176e-01
6.48429871e-01 -6.53731048e-01 -9.48614955e-01 9.22812149e-02
-8.24862838e-01 6.36433586e-02 7.26972938e-01 -6.24761209e-02
-1.22067010e+00 2.30215624e-01 -2.21560802e-02 -6.52792528e-02
-2.94624507e-01 5.18407166e-01 -4.27313566e-01 -8.87837335e-02
1.54248431e-01 3.92594486e-01 6.99553266e-02 -1.00513540e-01
1.97152704e-01 7.09459126e-01 3.88970047e-01 -3.08521867e-01
8.24683309e-01 3.09438795e-01 3.41402024e-01 -8.84924650e-01
-1.23940565e-01 -6.39125228e-01 -5.55889487e-01 -6.22276127e-01
8.77108097e-01 -6.91023827e-01 -5.40581703e-01 3.56805891e-01
-1.20580018e+00 -1.03153199e-01 2.99569517e-01 6.38714671e-01
-9.73380625e-01 4.48388517e-01 1.02551416e-01 -1.12771857e+00
-5.96396327e-02 -1.41254425e+00 5.94079137e-01 3.47866237e-01
3.91582847e-02 -9.09524381e-01 -1.15653083e-01 3.56038630e-01
2.93782413e-01 5.98814368e-01 9.49876428e-01 -3.52028638e-01
-1.00796032e+00 -3.76800209e-01 -1.01011425e-01 1.04135936e-02
2.05129236e-01 4.83727343e-02 -3.53128612e-01 -1.00480348e-01
-1.95528582e-01 -4.63910913e-03 3.10312688e-01 4.18428808e-01
6.39302671e-01 -6.89116597e-01 -7.13786662e-01 3.00776303e-01
1.58648849e+00 8.54993343e-01 7.97968030e-01 6.18570983e-01
5.34749925e-01 7.47683465e-01 1.72207403e+00 5.60454786e-01
5.00389159e-01 7.48529196e-01 5.94319403e-01 2.57525623e-01
4.08752441e-01 -4.10920471e-01 4.76423234e-01 5.54057539e-01
-7.54004493e-02 -2.46289238e-01 -1.04812860e+00 7.26026475e-01
-2.33033657e+00 -1.20181441e+00 -3.68493408e-01 2.06528807e+00
3.22597057e-01 3.28390002e-01 2.92341739e-01 2.71636248e-01
6.28740549e-01 2.32357875e-01 -3.69372815e-01 -6.11254632e-01
3.36483300e-01 -5.90156019e-01 7.57002831e-01 6.75260186e-01
-1.19047701e+00 9.89266694e-01 5.99175358e+00 1.03063536e+00
-1.17869282e+00 -1.00925848e-01 1.31977379e-01 3.86786431e-01
-2.46029586e-01 1.45677269e-01 -9.78160441e-01 5.07332683e-01
8.06634605e-01 -3.69506001e-01 2.07654387e-01 9.60414469e-01
1.17473304e+00 -3.48654091e-01 -6.62422001e-01 7.08421946e-01
-2.80484229e-01 -1.41073859e+00 4.32519615e-02 -1.93242859e-02
9.91340816e-01 -4.20166850e-01 -2.19071791e-01 3.02484065e-01
1.04385968e-02 -9.83172119e-01 5.58753908e-01 7.71487772e-01
2.99532205e-01 -1.06062388e+00 8.52137089e-01 9.81002450e-01
-1.20080507e+00 -2.34960854e-01 -3.95101219e-01 -1.89374387e-01
5.75891137e-01 1.74942002e-01 -1.28732693e+00 8.74345124e-01
-3.64809446e-02 4.98565257e-01 -9.42631960e-02 1.33928168e+00
-4.34531569e-02 4.47699368e-01 -1.05382621e-01 -4.60628837e-01
8.28316391e-01 -6.22608185e-01 7.92963564e-01 9.64528918e-01
3.30274612e-01 4.19653170e-02 6.67756200e-01 5.62694192e-01
7.49977648e-01 2.80319333e-01 -9.34094071e-01 -1.60261914e-02
6.16442621e-01 9.06907976e-01 -4.97723222e-01 -1.42574638e-01
-4.02592391e-01 4.45516080e-01 -1.41863808e-01 2.76582092e-01
-9.76108551e-01 -4.33280826e-01 4.13965940e-01 1.98484629e-01
3.09392244e-01 -6.92135870e-01 -2.23406643e-01 -3.88920665e-01
-1.00171097e-01 -7.06885517e-01 -2.13644635e-02 -3.86592239e-01
-5.55124938e-01 2.83899367e-01 5.61195791e-01 -1.65574086e+00
-6.01410747e-01 -3.68527472e-01 -8.01141441e-01 7.76553512e-01
-1.58199906e+00 -1.17323232e+00 -2.62461543e-01 1.90143168e-01
1.05622482e+00 -5.88535070e-01 3.80025834e-01 1.92977011e-01
-5.46013713e-01 1.72169119e-01 2.18609408e-01 -3.90163004e-01
5.08468114e-02 -7.09436834e-01 -7.88515955e-02 9.40755785e-01
-6.15812004e-01 1.04889318e-01 9.29004073e-01 -8.71147513e-01
-1.55283689e+00 -1.36029375e+00 9.99736011e-01 1.73997506e-01
3.18715036e-01 4.90985870e-01 -6.86459959e-01 3.30542237e-01
-4.35903072e-02 -4.19614762e-01 5.21479174e-02 -6.53973043e-01
6.77214563e-01 -1.94743812e-01 -1.08347797e+00 7.63275385e-01
7.00610936e-01 1.97201774e-01 -4.83520925e-01 2.42155492e-01
4.20541137e-01 -2.36659303e-01 -5.05175114e-01 5.57622075e-01
4.35586125e-01 -9.03539121e-01 5.82775354e-01 -4.93434519e-01
1.00330375e-01 -6.09896660e-01 -4.90740687e-03 -1.09734261e+00
-2.62096822e-01 -6.70980155e-01 1.25851721e-01 1.03405237e+00
4.02551532e-01 -5.86426318e-01 9.02650595e-01 5.40952206e-01
-4.07737702e-01 -1.07246101e+00 -9.13567007e-01 -9.07591105e-01
-1.17292061e-01 -3.78130734e-01 6.08487248e-01 4.98239726e-01
-1.16830252e-01 8.10882747e-02 -5.14504552e-01 3.12633008e-01
4.78410006e-01 5.78342155e-02 1.04900932e+00 -6.63578868e-01
1.77706808e-01 -3.82984489e-01 -4.85883683e-01 -1.03486323e+00
3.14229190e-01 -5.54223478e-01 4.34708327e-01 -1.91727769e+00
-3.35867554e-01 -6.23075247e-01 2.18704268e-01 1.50621369e-01
-3.78038846e-02 -6.48804784e-01 1.42067745e-01 4.03385788e-01
-3.50920051e-01 7.22984970e-01 1.29418612e+00 -8.13818201e-02
-5.35976410e-01 5.56085527e-01 8.28823075e-02 7.80567110e-01
1.01161754e+00 -1.90745205e-01 -9.72246289e-01 -1.35564148e-01
-2.87720263e-01 5.50758898e-01 3.46125700e-02 -1.06216788e+00
3.78899306e-01 -1.03894174e+00 -4.32705343e-01 -1.08630538e+00
2.15463802e-01 -9.75013733e-01 1.84551984e-01 6.46353304e-01
-1.82079002e-01 2.80784100e-01 6.64484128e-02 6.90528512e-01
-2.99005836e-01 -4.53281939e-01 6.18570983e-01 2.68134121e-02
-1.45515311e+00 4.72584844e-01 -7.55655766e-01 -2.00925961e-01
1.57129741e+00 -5.24395227e-01 1.48020446e-01 -3.47256333e-01
-8.09828192e-02 6.21270478e-01 2.08366156e-01 3.78826976e-01
8.36396456e-01 -1.44115531e+00 -6.07038617e-01 -9.58129913e-02
-1.59218743e-01 1.49815681e-03 1.39038727e-01 8.70176256e-01
-8.67349088e-01 9.03190553e-01 -4.67651963e-01 -4.52126861e-01
-9.77146208e-01 5.51460803e-01 2.23531932e-01 -3.08443427e-01
-4.00227189e-01 -7.82407448e-02 -3.67179602e-01 -2.76094615e-01
1.73236504e-01 -2.21990421e-01 -5.75388789e-01 -1.39909044e-01
2.09131196e-01 1.01241553e+00 -2.35636175e-01 -7.99041927e-01
-1.99062258e-01 4.28698808e-01 4.24933553e-01 -3.97558481e-01
8.38927567e-01 -4.80364859e-01 2.53898412e-01 3.34910721e-01
9.03615475e-01 -2.75215626e-01 -1.38888204e+00 1.86852105e-02
4.39892322e-01 -6.29946589e-01 5.30987279e-03 -2.75491536e-01
-5.84424019e-01 6.60755575e-01 3.67449880e-01 7.19923526e-02
8.76807988e-01 -8.32374036e-01 1.10547435e+00 4.95531678e-01
7.76468158e-01 -1.33682239e+00 -2.48630941e-01 5.57166040e-01
7.78587341e-01 -1.24100745e+00 1.05766892e-01 -6.43772364e-01
-7.58631408e-01 1.18674588e+00 8.35190654e-01 -7.63473064e-02
3.80935490e-01 -1.55640364e-01 -1.21011667e-01 2.82465756e-01
-7.97680974e-01 -3.27796876e-01 2.63305038e-01 7.62797773e-01
-4.45286706e-02 -5.71930129e-03 -8.94773006e-01 2.35194102e-01
7.10447878e-02 1.01773940e-01 4.71058965e-01 1.09832370e+00
-1.06549823e+00 -1.01814950e+00 -4.33430225e-01 2.04304084e-01
1.82877526e-01 6.58966422e-01 1.47812888e-01 9.87954557e-01
3.33172768e-01 1.17780912e+00 -1.96204782e-01 -3.64494443e-01
4.32221055e-01 -4.24161479e-02 1.06544770e-01 -4.57901686e-01
1.09630577e-01 -2.41611943e-01 4.13394928e-01 -5.82544446e-01
-5.34158826e-01 -8.10361743e-01 -1.64017403e+00 -1.90901816e-01
-8.91327932e-02 1.85783938e-01 5.81899285e-01 1.26632977e+00
2.38799050e-01 1.47447780e-01 1.02192521e+00 -8.63416731e-01
-7.03932226e-01 -5.75333536e-01 -3.35189253e-02 6.00943305e-02
2.10914969e-01 -8.72790039e-01 -2.51791421e-02 -4.61132899e-02]
|
[5.364114761352539, 1.4910842180252075]
|
7a2b6460-a647-4af1-b102-ba1ab07a71be
|
which-neural-network-to-choose-for-post-fault
|
2104.03115
| null |
https://arxiv.org/abs/2104.03115v1
|
https://arxiv.org/pdf/2104.03115v1.pdf
|
Which Neural Network to Choose for Post-Fault Localization, Dynamic State Estimation and Optimal Measurement Placement in Power Systems?
|
We consider a power transmission system monitored with Phasor Measurement Units (PMUs) placed at significant, but not all, nodes of the system. Assuming that a sufficient number of distinct single-line faults, specifically pre-fault state and (not cleared) post-fault state, are recorded by the PMUs and are available for training, we, first, design a comprehensive sequence of Neural Networks (NNs) locating the faulty line. Performance of different NNs in the sequence, including Linear Regression, Feed-Forward NN, AlexNet, Graphical Convolutional NN, Neural Linear ODE and Neural Graph-based ODE, ordered according to the type and amount of the power flow physics involved, are compared for different levels of observability. Second, we build a sequence of advanced Power-System-Dynamics-Informed and Neural-ODE based Machine Learning schemes trained, given pre-fault state, to predict the post-fault state and also, in parallel, to estimate system parameters. Finally, third, and continuing to work with the first (fault localization) setting we design a (NN-based) algorithm which discovers optimal PMU placement.
|
['Michael Chertkov', 'Andrei Afonin']
|
2021-04-07
| null | null | null | null |
['fault-localization']
|
['computer-code']
|
[-3.64600956e-01 -1.30197965e-02 -2.01749988e-02 1.70382023e-01
-8.63645822e-02 -7.02775657e-01 3.17805886e-01 3.45120639e-01
7.07308114e-01 8.01423430e-01 -2.26002514e-01 -7.21588790e-01
-6.49682343e-01 -7.37539291e-01 -5.93225956e-01 -6.43603802e-01
-1.01362920e+00 5.82964420e-01 -5.73637523e-02 -2.09064350e-01
4.67423759e-02 1.08407664e+00 -7.76928961e-01 -2.48089224e-01
7.57865071e-01 1.06254733e+00 -2.33576372e-01 1.04248977e+00
6.62034929e-01 1.27654576e+00 -1.11672390e+00 5.07389843e-01
2.49402329e-01 -4.21047568e-01 -8.54913116e-01 1.90480158e-01
-1.26923740e-01 -4.86229599e-01 -1.13630760e+00 9.95402575e-01
3.45219880e-01 -6.38813153e-03 6.53514445e-01 -1.59226394e+00
-4.76285100e-01 7.88394988e-01 -1.33383885e-01 5.43612659e-01
4.90125388e-01 4.79600400e-01 7.80560315e-01 -3.57976884e-01
3.34955484e-01 7.20921397e-01 7.75587738e-01 -6.97866678e-02
-9.70867038e-01 -1.24679446e-01 2.67679337e-02 5.22910595e-01
-1.09462810e+00 -5.85322864e-02 9.88556027e-01 -3.42452824e-01
1.41853774e+00 1.02714553e-01 8.61052692e-01 8.19969118e-01
5.72530508e-01 5.96957147e-01 6.33996427e-01 -2.68644005e-01
5.15887976e-01 -4.20064867e-01 3.34269077e-01 7.22776473e-01
9.84951779e-02 1.03691682e-01 -1.96687520e-01 -3.09615284e-01
7.88290381e-01 3.14203687e-02 -9.43280756e-01 -2.59639889e-01
-7.55539298e-01 3.91048551e-01 6.70225859e-01 6.98980033e-01
-4.47266519e-01 2.77549565e-01 4.95547533e-01 9.36458707e-01
1.17082901e-01 7.15618253e-01 -8.58981848e-01 -1.32724181e-01
-8.96711171e-01 -1.47392809e-01 1.31554306e+00 7.09015310e-01
5.94433129e-01 8.94435406e-01 9.91918296e-02 2.65029043e-01
1.40539661e-01 6.77273190e-03 6.13741338e-01 -3.95391285e-01
2.88966924e-01 7.45104194e-01 7.13936090e-02 -7.87706316e-01
-1.05386519e+00 -9.03393686e-01 -1.27040994e+00 1.98157698e-01
2.97479242e-01 -7.83195436e-01 -9.11002159e-01 1.37614214e+00
-3.42371017e-02 6.42829299e-01 -1.56973794e-01 6.79565847e-01
-7.41722584e-02 1.05860794e+00 -5.41063130e-01 -4.28658247e-01
9.83620465e-01 -8.88492882e-01 -6.89235568e-01 1.27615169e-01
9.82003808e-01 -1.69881493e-01 3.07001531e-01 5.69412470e-01
-1.09549451e+00 -2.93753445e-01 -1.54326093e+00 4.81802613e-01
-6.40846908e-01 2.21957430e-01 1.42098173e-01 1.49681449e-01
-1.37110043e+00 1.37127650e+00 -9.71247554e-01 -2.97641426e-01
1.83397576e-01 5.41724086e-01 -2.36826554e-01 2.50843078e-01
-1.49129426e+00 1.49516559e+00 4.62981820e-01 7.86653936e-01
-1.47399127e+00 -1.01795506e+00 -7.09878922e-01 5.96898019e-01
3.17906231e-01 -6.21901989e-01 9.53274250e-01 -6.37575686e-01
-1.32416487e+00 -2.73090452e-01 5.03346145e-01 -8.10739756e-01
3.22910637e-01 3.25474143e-02 -7.83396840e-01 2.56607383e-01
-3.52306306e-01 -3.30633610e-01 8.23674917e-01 -1.17776942e+00
-4.73002553e-01 -1.59109429e-01 1.36720866e-01 -3.15099908e-03
-7.85245374e-02 -3.19255918e-01 2.62619346e-01 -5.94755299e-02
-1.36826247e-01 -3.63887370e-01 -4.67286617e-01 -2.11741865e-01
-1.02367330e+00 -2.60726690e-01 1.43422472e+00 -1.05539739e+00
1.23961008e+00 -1.77788949e+00 2.46088371e-01 7.09246099e-01
3.30510587e-01 5.89351475e-01 2.74596840e-01 1.15293157e+00
-6.66521966e-01 -2.84134388e-01 -3.77182424e-01 -5.58824688e-02
1.31250814e-01 4.11803871e-01 -1.38585761e-01 7.89588749e-01
4.03865337e-01 9.53734577e-01 -8.53192031e-01 2.82872409e-01
7.80750692e-01 2.03643799e-01 1.69744655e-01 3.67006689e-01
-4.14112993e-02 4.46403414e-01 -3.24331403e-01 5.31182528e-01
4.07098681e-01 -4.42997873e-01 2.22602308e-01 -3.74321520e-01
2.17076987e-02 5.03809378e-02 -1.13394308e+00 1.24939334e+00
-5.70606768e-01 9.43565309e-01 1.13267742e-01 -1.44732654e+00
5.65517008e-01 6.87696636e-01 9.06271517e-01 -2.81524628e-01
3.37090522e-01 1.13183215e-01 2.00242788e-01 -4.27047253e-01
-1.15606217e-02 4.92780000e-01 1.31317332e-01 5.88417768e-01
5.81355155e-01 1.10201821e-01 6.20584905e-01 5.96253797e-02
1.70698380e+00 -1.64803281e-01 4.11824554e-01 -5.56885719e-01
6.59424245e-01 1.13037713e-02 2.52280056e-01 5.34925342e-01
-8.50756913e-02 3.63570094e-01 1.30565929e+00 -5.28754473e-01
-1.05969787e+00 -7.11550653e-01 -1.93074703e-01 1.37102067e-01
-5.17429151e-02 -7.88216665e-02 -7.47359157e-01 -6.86956406e-01
1.48355469e-01 7.47713447e-01 -6.20013118e-01 -4.86866832e-01
-7.25665689e-01 -8.31397176e-01 3.97342980e-01 6.15070641e-01
2.54447162e-01 -1.08394110e+00 -4.01004761e-01 6.24987483e-01
7.09346354e-01 -8.63430202e-01 3.37511152e-02 9.62567270e-01
-5.90299606e-01 -1.49041140e+00 -4.01506782e-01 -7.60124564e-01
1.10589683e+00 -5.67286015e-01 1.00269413e+00 5.21618843e-01
-2.66864419e-01 3.49926353e-01 -1.89523295e-01 3.82001787e-01
-4.82765913e-01 -1.01637945e-01 1.08131357e-01 -1.25080779e-01
-3.69158208e-01 -1.04101980e+00 -3.36803198e-01 1.49010401e-02
-8.33313107e-01 -4.91086602e-01 4.89620149e-01 8.63656342e-01
-2.24993214e-01 8.05012822e-01 4.74646717e-01 -6.12593532e-01
8.10113668e-01 -7.10071027e-01 -8.90195787e-01 3.64933193e-01
-6.59775376e-01 -2.25780487e-01 1.59804809e+00 -2.72757500e-01
-3.67305100e-01 -2.26243064e-01 -1.50916710e-01 -7.90083885e-01
-2.85054684e-01 8.54054630e-01 -2.77196348e-01 -3.10120165e-01
3.77276421e-01 9.12484527e-02 -2.75596112e-01 -4.37501669e-01
2.54238009e-01 4.47464943e-01 6.33250654e-01 -8.49717781e-02
1.26828241e+00 3.25778462e-02 4.33990687e-01 -8.52819920e-01
-3.80407214e-01 -6.35553226e-02 -1.10990357e+00 -2.23246619e-01
3.38600129e-01 -5.01276493e-01 -9.35620368e-01 8.62094522e-01
-1.07251775e+00 -6.08159006e-01 -4.97277766e-01 6.63899183e-02
-1.96298406e-01 4.12241489e-01 -1.12269723e+00 -6.38174355e-01
-3.94459337e-01 -9.81751800e-01 5.56818008e-01 1.87565669e-01
-5.45181930e-02 -1.74218035e+00 -9.31051597e-02 -4.93054599e-01
3.71274471e-01 3.65123093e-01 1.24595189e+00 -9.28992629e-01
-5.48462749e-01 -6.07071102e-01 -1.02670252e-01 5.49035370e-01
2.14049295e-01 3.04913163e-01 -6.02959394e-01 -8.38575482e-01
1.17033124e-01 2.13749498e-01 -2.68764999e-02 6.37919903e-01
8.63749802e-01 -5.04717767e-01 -5.27520239e-01 5.20762444e-01
1.62421000e+00 4.80054766e-01 4.02521551e-01 4.50126082e-02
8.39738846e-01 -7.53076673e-02 -2.43330106e-01 4.27760243e-01
3.46473664e-01 2.91549623e-01 6.08982384e-01 -4.02073085e-01
6.49478436e-02 -1.89705975e-02 4.60762054e-01 9.50729251e-01
4.30977732e-01 -6.29899681e-01 -7.45467484e-01 6.35791302e-01
-1.71017218e+00 -4.65006143e-01 -2.15894595e-01 1.93000996e+00
1.84249118e-01 4.38312709e-01 -3.03919874e-02 7.90468156e-01
6.06422007e-01 7.91432261e-02 -7.34468460e-01 -5.70632696e-01
-6.75226450e-02 6.77442402e-02 5.88526845e-01 5.63260138e-01
-8.39876533e-01 1.34557605e-01 6.34769011e+00 4.98888165e-01
-1.21506095e+00 -1.33694127e-01 4.82332766e-01 3.85263622e-01
2.11101949e-01 1.98237658e-01 -1.63708508e-01 5.33644676e-01
1.45741653e+00 -2.18149275e-01 5.96101105e-01 4.80650812e-01
4.03136939e-01 -1.91581637e-01 -1.48619580e+00 3.96111012e-01
-1.16318144e-01 -1.11997211e+00 -3.10785383e-01 -1.35205775e-01
8.92897487e-01 1.72808528e-01 -5.84474623e-01 1.28547251e-01
5.47724366e-01 -8.19479704e-01 3.31262529e-01 4.95292544e-01
1.36547446e-01 -7.28727520e-01 8.85911763e-01 2.96789825e-01
-1.10361493e+00 -5.98693788e-01 6.59205997e-03 -2.70638108e-01
5.17717421e-01 8.58050942e-01 -6.62450790e-01 1.27801526e+00
4.81944323e-01 1.08077371e+00 -4.16311145e-01 1.11710179e+00
-5.84978044e-01 7.51895130e-01 -4.18954611e-01 9.60807949e-02
2.82842845e-01 -6.14706129e-02 5.23267388e-01 5.85043609e-01
1.78833604e-01 -3.08046907e-01 3.71720880e-01 9.38196361e-01
1.42943755e-01 -8.23559582e-01 -4.35374498e-01 7.70075917e-02
5.96230149e-01 1.49987841e+00 -6.77096546e-01 -3.50461543e-01
-2.52393931e-01 7.17096686e-01 9.13752764e-02 7.05141842e-01
-5.61459064e-01 -5.11154771e-01 4.79031146e-01 -1.91465333e-01
-1.62557662e-02 -3.63220274e-01 -2.27992550e-01 -1.00005591e+00
-5.34411222e-02 -4.40505147e-01 3.69290680e-01 -1.13619328e+00
-1.37758625e+00 4.75561053e-01 -1.08874179e-01 -1.28224599e+00
-8.30001712e-01 -6.85042679e-01 -1.19203174e+00 9.08938348e-01
-1.65102410e+00 -8.95104170e-01 1.26491517e-01 4.74364698e-01
3.46687645e-01 6.00780584e-02 6.04448557e-01 4.36219066e-01
-1.30408049e+00 3.55291963e-02 2.31591135e-01 5.37370622e-01
-3.75350058e-01 -1.67845774e+00 4.29380387e-01 1.10208583e+00
4.16226611e-02 1.03423774e-01 7.37180233e-01 -5.77706754e-01
-1.94599581e+00 -1.02615452e+00 5.41757405e-01 -2.52998173e-01
1.36582208e+00 -1.76171377e-01 -1.04989898e+00 1.14625549e+00
6.70343101e-01 2.17311502e-01 -1.84563965e-01 -3.13246161e-01
6.61998987e-01 1.60103280e-03 -1.19131374e+00 4.56712157e-01
3.48346621e-01 -4.89308178e-01 -5.32696843e-01 6.07464969e-01
1.27614498e-01 -7.69294322e-01 -1.41634178e+00 2.04262257e-01
-1.67119443e-01 -5.29649436e-01 7.63991654e-01 -5.26940346e-01
1.95164025e-01 -5.69373727e-01 5.99613845e-01 -2.11621022e+00
-3.92366678e-01 -9.94549334e-01 -9.05643821e-01 8.23748291e-01
5.05058706e-01 -1.05758548e+00 5.93789279e-01 1.67545557e-01
-7.34124660e-01 -9.98258770e-01 -1.10506022e+00 -6.26278400e-01
-6.16353974e-02 -8.44803527e-02 8.03477287e-01 1.20466840e+00
4.48755354e-01 2.70671278e-01 -2.84992188e-01 1.01191878e+00
2.67052740e-01 1.00859255e-02 1.89793304e-01 -1.12276590e+00
-1.54281452e-01 -5.77643812e-01 -5.54138064e-01 -7.76740313e-01
1.09602608e-01 -5.35065591e-01 -2.89856941e-01 -1.75564158e+00
-6.51133239e-01 -5.71849495e-02 -5.61269224e-01 7.98458815e-01
1.33624852e-01 -3.42509717e-01 -1.82045341e-01 1.62179217e-01
-2.26318762e-01 3.02358985e-01 1.07238209e+00 -1.68545227e-02
7.92520642e-02 -2.32208911e-02 2.20841512e-01 3.68414700e-01
6.14098191e-01 -2.27735937e-02 -1.35262743e-01 -2.05231905e-01
4.27383324e-03 8.06791604e-01 4.12687629e-01 -1.44895971e+00
5.70932925e-01 4.39661682e-01 7.37209141e-01 -7.73378968e-01
-6.08759932e-02 -1.27157676e+00 1.51873097e-01 9.60355520e-01
4.77455482e-02 4.69524413e-01 1.50367975e-01 1.42962843e-01
-1.84250772e-01 -4.34942693e-01 4.07009035e-01 1.50624007e-01
-6.71360672e-01 4.31042075e-01 -7.46789277e-01 -4.45377052e-01
9.11243498e-01 -6.46455958e-02 -5.87005198e-01 -3.77440512e-01
-9.14388716e-01 7.71825254e-01 1.32123172e-01 3.62794906e-01
7.32583255e-02 -1.06580579e+00 -3.39242876e-01 4.90339071e-01
-5.67095935e-01 -8.12017098e-02 3.07628274e-01 1.01188564e+00
-6.35300934e-01 5.85197508e-01 -2.57253230e-01 -5.88402808e-01
-3.12346727e-01 4.46498066e-01 1.01090372e+00 -4.98400092e-01
-7.61691272e-01 2.55060405e-01 -7.64616966e-01 -2.79186875e-01
-7.94641078e-02 -5.08647442e-01 -1.54738486e-01 -7.77935609e-02
-2.09078584e-02 6.87202513e-01 5.93402922e-01 -1.90825194e-01
-5.61668761e-02 3.09736580e-01 3.13046277e-01 6.44413650e-01
1.34465373e+00 -1.29934121e-02 -3.38607460e-01 4.07534063e-01
1.10603333e+00 -8.48758578e-01 -1.41505563e+00 1.64577380e-01
1.25807822e-01 1.18021034e-01 2.22178683e-01 -9.40431535e-01
-1.67799175e+00 7.08875954e-01 3.74205619e-01 1.27307367e+00
1.12423873e+00 -5.26512563e-01 5.24052978e-01 3.65857571e-01
3.07372421e-01 -9.06384945e-01 -4.29614365e-01 4.41948473e-01
4.68816727e-01 -5.32765210e-01 4.20406424e-02 -1.07610546e-01
2.30631590e-01 1.57455635e+00 5.84689915e-01 -5.42269886e-01
9.61884558e-01 8.61668408e-01 -3.56699824e-02 -5.01302361e-01
-1.02814794e+00 2.46517584e-01 1.30881861e-01 6.13598347e-01
-1.22478493e-01 -1.77072704e-01 3.33772898e-01 5.05412742e-03
-1.73896164e-01 -1.09768696e-01 8.89676571e-01 1.22655082e+00
-1.29441604e-01 -6.29467666e-01 -2.29708403e-01 9.15569544e-01
-1.17334798e-01 3.31387669e-01 -4.57838885e-02 1.10878706e+00
-2.52524972e-01 8.21546674e-01 2.64493763e-01 -2.75787041e-02
6.00521743e-01 -2.24395514e-01 2.35943899e-01 -2.53691167e-01
-7.88042307e-01 -3.47564429e-01 -1.63633823e-02 -3.67580175e-01
4.06389594e-01 -4.05055702e-01 -1.31530201e+00 -3.31977963e-01
-7.10160732e-01 3.23314667e-01 4.52464104e-01 1.40146196e+00
5.69264777e-02 1.16408312e+00 1.13260078e+00 -1.06861281e+00
-6.61387503e-01 -1.20173132e+00 -1.07850683e+00 -1.23221362e-02
7.04879642e-01 -5.00355721e-01 -8.69864821e-01 -4.77423757e-01]
|
[6.118586540222168, 2.618344783782959]
|
22630cc7-2969-4b19-8839-de2d98c2e672
|
improving-sequential-determinantal-point
|
1807.10957
| null |
http://arxiv.org/abs/1807.10957v2
|
http://arxiv.org/pdf/1807.10957v2.pdf
|
Improving Sequential Determinantal Point Processes for Supervised Video Summarization
|
It is now much easier than ever before to produce videos. While the
ubiquitous video data is a great source for information discovery and
extraction, the computational challenges are unparalleled. Automatically
summarizing the videos has become a substantial need for browsing, searching,
and indexing visual content. This paper is in the vein of supervised video
summarization using sequential determinantal point process (SeqDPP), which
models diversity by a probabilistic distribution. We improve this model in two
folds. In terms of learning, we propose a large-margin algorithm to address the
exposure bias problem in SeqDPP. In terms of modeling, we design a new
probabilistic distribution such that, when it is integrated into SeqDPP, the
resulting model accepts user input about the expected length of the summary.
Moreover, we also significantly extend a popular video summarization dataset by
1) more egocentric videos, 2) dense user annotations, and 3) a refined
evaluation scheme. We conduct extensive experiments on this dataset (about 60
hours of videos in total) and compare our approach to several competitive
baselines.
|
['Tianbao Yang', 'Ali Borji', 'Aidean Sharghi', 'Boqing Gong', 'Chengtao Li']
|
2018-07-28
|
improving-sequential-determinantal-point-1
|
http://openaccess.thecvf.com/content_ECCV_2018/html/Aidean_Sharghi_Improving_Sequential_Determinantal_ECCV_2018_paper.html
|
http://openaccess.thecvf.com/content_ECCV_2018/papers/Aidean_Sharghi_Improving_Sequential_Determinantal_ECCV_2018_paper.pdf
|
eccv-2018-9
|
['supervised-video-summarization']
|
['computer-vision']
|
[ 3.62189144e-01 5.46660833e-02 -4.13018823e-01 -2.51147985e-01
-1.03474593e+00 -6.86704397e-01 6.10330343e-01 2.30007902e-01
-3.69101673e-01 6.29776239e-01 9.96069908e-01 -4.79968414e-02
1.19284175e-01 -4.18287754e-01 -9.47846055e-01 -6.57304585e-01
4.38171625e-02 2.27349564e-01 3.74001116e-01 1.71847284e-01
5.52217007e-01 1.38186440e-01 -1.42974162e+00 2.80687660e-01
9.61603522e-01 8.98574889e-01 2.70572692e-01 7.09590495e-01
1.35580763e-01 1.02441037e+00 -3.43598932e-01 -5.17866135e-01
3.06012541e-01 -5.26189923e-01 -7.91754127e-01 5.22869647e-01
6.89127028e-01 -4.67209935e-01 -5.88900983e-01 9.04865563e-01
6.30621791e-01 4.26415652e-01 6.22069895e-01 -1.27965689e+00
-3.79658222e-01 6.64232433e-01 -9.24178004e-01 1.07245976e-02
5.56903958e-01 4.89864424e-02 1.20823944e+00 -7.08272099e-01
8.25497508e-01 1.19970763e+00 5.00493586e-01 2.78554857e-01
-1.17409086e+00 -3.51793766e-01 3.69383067e-01 2.89091319e-01
-1.11684084e+00 -6.33215129e-01 8.75322163e-01 -6.00371122e-01
5.15553296e-01 3.55613321e-01 6.06516838e-01 1.24054408e+00
-1.66089013e-02 1.24325132e+00 6.28095329e-01 -2.19256312e-01
3.36486697e-01 1.27415285e-01 2.09753551e-02 6.14013374e-01
1.64188862e-01 -5.04370391e-01 -6.96857035e-01 -2.59212643e-01
5.70715785e-01 1.57066420e-01 -4.44308609e-01 -7.15042412e-01
-1.27447927e+00 7.02670217e-01 -7.02185854e-02 -1.16078593e-01
-4.26100761e-01 1.36474282e-01 6.02571666e-01 2.07385924e-02
4.65114534e-01 4.05355334e-01 -1.84089735e-01 -3.21588784e-01
-1.33619010e+00 5.66898584e-01 8.88572752e-01 1.04798901e+00
5.28202236e-01 -2.83477426e-01 -4.11208034e-01 7.88113892e-01
4.78880331e-02 3.90051514e-01 4.64118421e-01 -1.31476104e+00
6.70214117e-01 2.61684746e-01 1.57730773e-01 -1.14900935e+00
2.08246000e-02 -2.16403008e-01 -7.53782809e-01 -3.38891119e-01
1.37742132e-01 -1.84766889e-01 -4.77914333e-01 1.76572180e+00
2.42649555e-01 9.01944861e-02 -2.04046175e-01 8.36271644e-01
5.39240301e-01 1.01417387e+00 -5.36828861e-02 -5.87806404e-01
1.26805103e+00 -1.26952708e+00 -5.60539544e-01 7.04508200e-02
4.46181118e-01 -5.72533846e-01 8.74842167e-01 5.28457522e-01
-1.26743054e+00 -3.82215500e-01 -8.68185341e-01 -3.19036543e-01
4.09966826e-01 2.75346674e-02 3.92832547e-01 3.48762065e-01
-1.12351859e+00 6.51162803e-01 -8.41121793e-01 -5.44435561e-01
5.13169885e-01 -4.83012870e-02 -4.13677663e-01 -1.46100968e-01
-7.04493880e-01 4.09482181e-01 4.54217374e-01 -4.12867785e-01
-6.46694720e-01 -5.15548289e-01 -9.92173433e-01 1.34188563e-01
8.42778981e-01 -8.32382262e-01 1.22412252e+00 -9.68931735e-01
-1.42929769e+00 4.94122446e-01 -4.06192690e-01 -4.74226832e-01
7.79743433e-01 -4.65264678e-01 2.50176102e-01 5.00862777e-01
1.45447075e-01 7.92917013e-01 9.81300771e-01 -1.17388749e+00
-6.85041487e-01 -2.22340643e-01 -3.10634617e-02 4.68352288e-01
-4.80024606e-01 2.09011827e-02 -8.27126324e-01 -9.40647900e-01
-1.99045241e-01 -1.00589943e+00 -3.54476511e-01 -6.39352277e-02
-5.18490613e-01 -3.52835506e-01 6.43662930e-01 -9.69317257e-01
1.54141366e+00 -2.28805184e+00 6.61809742e-01 3.63540836e-02
3.92196327e-01 8.85619596e-03 -3.20650190e-02 5.00301421e-01
1.23496249e-01 1.18060164e-01 -1.06996961e-01 -5.75236559e-01
1.10470243e-01 -1.07251205e-01 -5.14132321e-01 3.19847345e-01
-7.81336427e-02 7.30426013e-01 -9.83043671e-01 -7.98744798e-01
-1.15629181e-01 1.46464020e-01 -1.02900660e+00 1.73457250e-01
-3.09849083e-01 2.75888771e-01 -3.55259031e-01 5.38050652e-01
4.32962000e-01 -2.39089787e-01 2.88608044e-01 -3.57868522e-01
-2.24857807e-01 3.76254767e-02 -1.19716346e+00 1.89837825e+00
-5.33035882e-02 7.27227211e-01 -2.61330426e-01 -9.89251018e-01
3.34666908e-01 2.83930123e-01 6.64150953e-01 -2.55806446e-01
1.55610025e-01 -2.82842219e-01 -5.27254045e-01 -6.27546787e-01
1.00177288e+00 2.36780494e-01 -2.79128134e-01 5.48410654e-01
1.37742430e-01 7.79110193e-02 6.47875667e-01 6.51658952e-01
1.06123435e+00 4.21711922e-01 4.28159505e-01 -5.20496108e-02
1.68320715e-01 -1.99434340e-01 7.93947518e-01 7.47021317e-01
-2.81965315e-01 7.50202537e-01 1.06890249e+00 -1.70149148e-01
-1.11727428e+00 -1.03198922e+00 3.02361131e-01 1.13525236e+00
2.33952284e-01 -8.04913163e-01 -8.55001628e-01 -7.26676404e-01
-2.52792597e-01 5.67082942e-01 -2.57723629e-01 8.19698051e-02
-5.22246301e-01 -5.57708323e-01 1.59955963e-01 5.78146338e-01
3.86076570e-01 -7.65464306e-01 -5.26786387e-01 8.47860649e-02
-6.73816681e-01 -1.24504185e+00 -1.07418871e+00 -3.36593479e-01
-8.41242731e-01 -8.64208221e-01 -8.81416202e-01 -6.59336090e-01
4.30482954e-01 6.53540194e-01 9.62714732e-01 -4.16441798e-01
-7.29708001e-02 5.39553463e-01 -3.90670955e-01 -2.50789881e-01
-1.39267072e-01 1.07203133e-01 9.88891497e-02 1.15002662e-01
7.03351758e-03 -6.01442218e-01 -7.59534836e-01 5.63907102e-02
-1.02334082e+00 3.97842109e-01 7.68223345e-01 5.36354184e-01
6.13418818e-01 1.63376536e-02 4.74202335e-01 -9.08039629e-01
4.30446297e-01 -7.11403072e-01 -3.38755846e-01 2.43900388e-01
-3.09366345e-01 2.05218717e-01 3.80195856e-01 -4.25270885e-01
-1.16382909e+00 2.62066305e-01 5.65234311e-02 -4.88900691e-01
1.50943035e-02 3.69126439e-01 -2.90727586e-01 4.05779928e-01
4.52255607e-01 2.37220272e-01 6.49956428e-03 -3.41114879e-01
4.89968807e-01 5.89463413e-01 4.92683709e-01 -4.35084701e-01
6.90864980e-01 4.90444899e-01 -2.26009458e-01 -1.05368197e+00
-8.38248432e-01 -6.33082807e-01 -4.39869225e-01 -2.36260101e-01
8.83293927e-01 -1.12725461e+00 -5.20785928e-01 2.72988290e-01
-1.00757742e+00 -1.80268660e-01 -3.04827482e-01 4.57395434e-01
-6.87944293e-01 9.72650588e-01 -5.46289563e-01 -6.35288835e-01
-2.18699351e-01 -1.03997445e+00 1.04651570e+00 2.04766154e-01
-3.29533845e-01 -6.86189890e-01 2.24861115e-01 5.30850768e-01
1.09963216e-01 2.68027604e-01 7.35192895e-01 -7.23286927e-01
-7.52010763e-01 -1.83468133e-01 -2.88452625e-01 4.23516989e-01
-1.64883971e-01 3.61168832e-02 -6.37905300e-01 -2.30813995e-01
-8.43387768e-02 -3.50219011e-01 1.05055857e+00 5.12812495e-01
1.40352261e+00 -6.27387583e-01 -3.90281081e-01 4.24592763e-01
1.24681079e+00 -1.17839828e-01 6.31803572e-01 1.80741176e-01
7.85843372e-01 6.41375244e-01 5.45252085e-01 7.10952699e-01
7.39728868e-01 5.90261936e-01 1.85114115e-01 2.21838996e-01
7.76158497e-02 -5.72376311e-01 6.01786315e-01 9.17815328e-01
-2.38478437e-01 -5.76279879e-01 -4.46114033e-01 6.12478793e-01
-2.07586098e+00 -1.32490349e+00 2.02811673e-01 2.20526481e+00
8.38308215e-01 1.73359830e-02 4.98200148e-01 1.88691318e-02
6.69840455e-01 5.04028261e-01 -3.43058556e-01 -7.18216784e-03
1.04475645e-02 -3.02144259e-01 4.30250943e-01 3.05786371e-01
-1.27282000e+00 6.32016838e-01 5.72139931e+00 9.13564920e-01
-8.66478205e-01 -4.97542061e-02 5.58956146e-01 -4.05532241e-01
-2.92731166e-01 5.91674708e-02 -6.06950939e-01 7.46077299e-01
5.69955528e-01 -1.98273242e-01 2.84569293e-01 9.63726819e-01
3.58802468e-01 -2.64378101e-01 -1.29360068e+00 1.25895977e+00
5.61796367e-01 -1.29206145e+00 4.03450608e-01 5.68879135e-02
9.28874195e-01 -2.03874320e-01 -8.01069736e-02 2.72344857e-01
4.13825735e-02 -4.94889438e-01 8.32928777e-01 5.41848481e-01
4.93299127e-01 -7.03766286e-01 4.87251401e-01 3.37258667e-01
-9.57208097e-01 -1.04170397e-01 -2.42007524e-01 2.59210378e-01
4.00342584e-01 5.15597641e-01 -5.47585964e-01 5.77634096e-01
6.08241856e-01 1.03302097e+00 -5.26051879e-01 1.24928796e+00
-1.64208099e-01 6.58564091e-01 -5.26920930e-02 6.86613172e-02
5.63874096e-02 -2.27869853e-01 8.60526562e-01 1.39215207e+00
3.71240824e-01 7.66214356e-02 4.38033491e-01 1.98412374e-01
-4.72411245e-01 2.19264537e-01 -4.52388346e-01 -2.05820009e-01
3.53305489e-01 1.18234706e+00 -7.05732644e-01 -4.60595101e-01
-4.04468328e-01 1.21881008e+00 1.92286223e-01 3.61637592e-01
-8.82915080e-01 -3.93437088e-01 3.61065447e-01 2.41506144e-01
3.67103398e-01 -1.03048831e-01 1.79493502e-01 -1.47779894e+00
2.31384382e-01 -8.56877148e-01 3.28951746e-01 -7.10252225e-01
-1.10629833e+00 3.42469007e-01 1.80644602e-01 -1.20647216e+00
-3.23756367e-01 -1.47877768e-01 -3.90208691e-01 1.93060130e-01
-1.34777832e+00 -9.69340205e-01 -3.05458248e-01 3.38126034e-01
9.84404385e-01 1.68643638e-01 2.44359434e-01 3.01705837e-01
-6.01746321e-01 4.00587261e-01 1.88998625e-01 -1.01423182e-01
1.08429170e+00 -1.20723462e+00 1.57615274e-01 8.99207652e-01
1.72431499e-01 2.09012762e-01 9.34818566e-01 -5.94660163e-01
-1.43529618e+00 -1.04126477e+00 8.47276151e-01 -4.39823270e-01
6.64906323e-01 -3.49570900e-01 -7.38276005e-01 8.58556926e-01
3.19176257e-01 -5.91662228e-01 9.27039206e-01 4.17697877e-02
-2.77792811e-01 -2.30784975e-02 -7.05931485e-01 7.95808613e-01
1.11501729e+00 -1.87805995e-01 -7.23324955e-01 4.33132589e-01
7.97803700e-01 -1.86064005e-01 -4.38628286e-01 1.83798566e-01
6.26809061e-01 -9.32305753e-01 6.72945082e-01 -3.93419445e-01
7.86389709e-01 -2.46933520e-01 -1.57711625e-01 -1.23094225e+00
-3.22426587e-01 -9.53574121e-01 -2.94281363e-01 1.35132492e+00
1.37752190e-01 -5.80305196e-02 7.60288894e-01 4.85907018e-01
-1.24313489e-01 -7.87278235e-01 -5.10503769e-01 -5.51524699e-01
-4.16641623e-01 -2.45430395e-01 5.84841752e-03 6.20158792e-01
2.24438295e-01 5.13221443e-01 -8.84082735e-01 -1.43079370e-01
6.94376230e-01 2.47915581e-01 1.05346441e+00 -1.03198695e+00
-4.76744235e-01 -3.93479347e-01 -4.51724529e-01 -1.59535182e+00
-1.74249597e-02 -6.80997610e-01 1.24058761e-01 -1.60877776e+00
9.45993543e-01 1.95934922e-01 -1.30853429e-02 8.25882424e-03
-2.94818044e-01 3.75150815e-02 3.32410514e-01 3.81317496e-01
-1.32674813e+00 6.37300014e-01 9.66557205e-01 9.37223211e-02
-2.65430897e-01 -3.61857451e-02 -8.97412300e-01 8.87402117e-01
5.23136675e-01 -2.62484938e-01 -4.56066728e-01 -4.80257690e-01
3.24529260e-01 2.39700690e-01 1.60967886e-01 -8.31667364e-01
3.66690874e-01 -4.08796929e-02 2.00632885e-01 -8.97419214e-01
1.91643536e-01 -4.49160933e-01 -8.25250745e-02 1.05490446e-01
-5.30927181e-01 -1.15860656e-01 -2.98834622e-01 9.58969355e-01
-2.78086185e-01 -8.38448480e-02 5.66870749e-01 1.81230009e-02
-6.86590552e-01 4.42056119e-01 -4.39702958e-01 1.66104957e-01
1.12618232e+00 -1.29521355e-01 -2.57481355e-02 -6.96796298e-01
-4.76956755e-01 4.38492030e-01 7.03726709e-01 3.18205446e-01
4.97402340e-01 -1.14062512e+00 -6.55793607e-01 -2.25926876e-01
5.81976660e-02 5.33464439e-02 4.20050263e-01 8.51933658e-01
-4.80977923e-01 2.44679689e-01 -1.41261546e-02 -5.46498299e-01
-1.38750792e+00 6.29749238e-01 -4.09369379e-01 -2.99558938e-01
-5.44129729e-01 6.71948671e-01 4.43688333e-01 1.38346791e-01
4.42926377e-01 -1.03075214e-01 -2.68378079e-01 3.48785818e-01
7.32149541e-01 5.78179657e-01 -4.75538254e-01 -4.78949547e-01
4.79378998e-02 2.57905692e-01 -3.80968779e-01 -1.45707816e-01
1.44041634e+00 -4.46500689e-01 3.83395585e-03 3.73795837e-01
1.06381226e+00 3.93096864e-01 -1.55744946e+00 -3.17365855e-01
4.77032065e-02 -5.29221952e-01 -2.62691349e-01 -2.90534049e-01
-7.66660631e-01 6.03770554e-01 1.00805067e-01 9.77813378e-02
1.14388645e+00 2.08412439e-01 1.01947522e+00 3.48483771e-01
1.99668393e-01 -1.11240697e+00 4.20277596e-01 2.57451028e-01
7.82432079e-01 -1.22545910e+00 2.51766413e-01 -3.28878403e-01
-1.00989938e+00 8.45801890e-01 1.95923135e-01 -2.98569292e-01
3.07332903e-01 -2.29849368e-02 -5.30322433e-01 9.39871520e-02
-9.28014338e-01 2.23749146e-01 2.20338836e-01 1.83460176e-01
2.54518151e-01 -1.43690243e-01 -4.12144691e-01 7.19215989e-01
-1.63277257e-02 7.01460168e-02 7.18706131e-01 1.01759470e+00
-5.95724642e-01 -8.57358456e-01 -1.04033835e-01 4.13085014e-01
-5.78931391e-01 1.33971276e-03 -1.77877933e-01 4.16622937e-01
-2.41142243e-01 7.28589714e-01 8.72170669e-04 -2.83401579e-01
2.93267190e-01 9.53661744e-03 4.78436768e-01 -5.40024161e-01
3.29098441e-02 2.28951842e-01 -5.72405271e-02 -6.83674395e-01
-4.90706712e-01 -9.58312094e-01 -7.63810039e-01 -2.45178252e-01
-1.51992396e-01 1.82943657e-01 5.74309826e-01 8.61047506e-01
4.16508585e-01 3.10024977e-01 7.47801900e-01 -1.14543903e+00
-6.27441525e-01 -8.28848243e-01 -3.79691780e-01 3.88101101e-01
2.12826908e-01 -4.14070934e-01 -3.19807768e-01 5.34045935e-01]
|
[10.439117431640625, 0.48687297105789185]
|
943db0f3-501d-4c2a-974b-a5f4c629a6ef
|
learning-inner-group-relations-on-point
|
2108.12468
| null |
https://arxiv.org/abs/2108.12468v1
|
https://arxiv.org/pdf/2108.12468v1.pdf
|
Learning Inner-Group Relations on Point Clouds
|
The prevalence of relation networks in computer vision is in stark contrast to underexplored point-based methods. In this paper, we explore the possibilities of local relation operators and survey their feasibility. We propose a scalable and efficient module, called group relation aggregator. The module computes a feature of a group based on the aggregation of the features of the inner-group points weighted by geometric relations and semantic relations. We adopt this module to design our RPNet. We further verify the expandability of RPNet, in terms of both depth and width, on the tasks of classification and segmentation. Surprisingly, empirical results show that wider RPNet fits for classification, while deeper RPNet works better on segmentation. RPNet achieves state-of-the-art for classification and segmentation on challenging benchmarks. We also compare our local aggregator with PointNet++, with around 30% parameters and 50% computation saving. Finally, we conduct experiments to reveal the robustness of RPNet with regard to rigid transformation and noises.
|
['Li Lu', 'Jun Liu', 'Wei Zhuo', 'Haoxi Ran']
|
2021-08-27
| null |
http://openaccess.thecvf.com//content/ICCV2021/html/Ran_Learning_Inner-Group_Relations_on_Point_Clouds_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Ran_Learning_Inner-Group_Relations_on_Point_Clouds_ICCV_2021_paper.pdf
|
iccv-2021-1
|
['3d-classification']
|
['computer-vision']
|
[-1.30579034e-02 4.64039475e-01 -1.43522277e-01 -3.21453691e-01
-3.75474304e-01 -4.78096426e-01 6.72850192e-01 1.18940160e-01
-2.18170539e-01 1.58185542e-01 3.12101822e-02 -2.21523866e-01
-3.91121209e-01 -9.82752085e-01 -5.58813989e-01 -5.62320769e-01
-1.12322204e-01 5.86488307e-01 7.67512262e-01 -1.46905512e-01
1.05946593e-01 8.87476027e-01 -1.54515791e+00 1.80570543e-01
8.43001485e-01 1.42598808e+00 -6.07364550e-02 3.18161249e-01
4.21194620e-02 4.90180999e-01 -4.93807286e-01 -4.68482733e-01
5.29999554e-01 1.03644572e-01 -9.24617290e-01 5.97807467e-02
7.49641895e-01 4.73323427e-02 -2.56354690e-01 8.00165117e-01
4.48666871e-01 1.95436180e-01 4.65298921e-01 -1.14739680e+00
-5.55941701e-01 6.48568869e-01 -5.63414454e-01 4.68450459e-03
4.14043039e-01 4.07005660e-02 1.35827875e+00 -1.01227784e+00
8.30421805e-01 1.25541413e+00 1.05083489e+00 9.62608308e-02
-1.32268536e+00 -3.42383325e-01 2.31703103e-01 1.80809408e-01
-1.50584376e+00 -2.32162058e-01 7.13294029e-01 -1.82229549e-01
1.12530458e+00 3.55352402e-01 7.41198897e-01 5.00516117e-01
1.02906361e-01 6.25878453e-01 7.04989254e-01 -1.88041493e-01
2.12150991e-01 -3.77093524e-01 3.79675686e-01 9.08984303e-01
1.19502991e-01 -1.69318154e-01 -5.60806632e-01 -1.16973691e-01
1.01261210e+00 -1.64250925e-01 -2.42560431e-01 -6.86518013e-01
-1.34290206e+00 6.25200689e-01 9.55896556e-01 3.81873250e-02
-2.17879310e-01 1.32687420e-01 1.74227178e-01 1.48178294e-01
6.94438457e-01 5.73816478e-01 -4.58854437e-01 2.66099781e-01
-6.37021661e-01 1.16201162e-01 9.42768574e-01 1.19384027e+00
7.95854390e-01 -5.90995550e-01 5.63577414e-02 1.01003158e+00
1.84875607e-01 2.19797298e-01 -7.08250627e-02 -1.23149383e+00
3.95029068e-01 9.66708779e-01 -5.41691899e-01 -1.41394901e+00
-7.42084801e-01 -4.93358225e-01 -8.56629908e-01 5.44203259e-02
2.51482815e-01 1.57826886e-01 -8.66259336e-01 1.09558558e+00
5.85363925e-01 2.89306402e-01 -2.00176015e-01 7.61035323e-01
1.19027054e+00 2.45072559e-01 -4.65400331e-02 6.07147999e-02
1.27277410e+00 -1.29848623e+00 -1.67360052e-01 1.86646450e-02
5.85341156e-01 -6.76327288e-01 8.44853163e-01 2.90536433e-01
-1.09777498e+00 -6.56079173e-01 -8.55716705e-01 -3.47495705e-01
-1.78448886e-01 4.16959412e-02 1.02268720e+00 3.01269650e-01
-1.18874955e+00 9.69170809e-01 -9.36599433e-01 -7.16293871e-01
7.45225549e-01 6.15186036e-01 -5.03033042e-01 3.36862862e-01
-5.17031729e-01 7.57381439e-01 2.51026332e-01 5.41281402e-02
-1.33245245e-01 -9.42697585e-01 -8.37871134e-01 -1.41556069e-01
3.94061089e-01 -1.00942409e+00 9.86994684e-01 -1.03133768e-01
-1.28214693e+00 9.41767454e-01 2.10818779e-02 -6.12674117e-01
5.44430912e-01 -2.12422088e-01 6.60358146e-02 4.05496895e-01
1.27567798e-01 1.04233015e+00 3.73393893e-01 -1.09019792e+00
-8.94874752e-01 -4.68388677e-01 2.19673648e-01 3.08081537e-01
-6.25384748e-02 -1.20260850e-01 -7.69046128e-01 -3.81149441e-01
8.66156340e-01 -9.29334283e-01 -4.80159551e-01 4.09913749e-01
-6.18214011e-01 -7.07040370e-01 1.04441285e+00 -1.09251209e-01
7.98146367e-01 -2.22812462e+00 9.44302380e-02 4.67633694e-01
6.10628009e-01 2.16705009e-01 -4.17528823e-02 1.74322855e-02
-3.30783427e-02 2.39827991e-01 -2.66680986e-01 -5.68992853e-01
-7.31470808e-02 5.77035904e-01 -1.54096916e-01 5.20104706e-01
2.80123830e-01 1.15879059e+00 -7.29915082e-01 -7.81537712e-01
3.90332818e-01 4.32377249e-01 -5.18321753e-01 -2.98696160e-01
-1.08267151e-01 2.40528986e-01 -5.09301126e-01 9.02942777e-01
7.11518109e-01 -2.72742927e-01 -1.00968204e-01 -6.31497681e-01
-4.29905243e-02 3.93986970e-01 -1.08919322e+00 1.82543135e+00
-3.47048678e-02 5.38627863e-01 -9.66277495e-02 -8.44356894e-01
1.15302205e+00 1.57972500e-02 7.71687448e-01 -4.70121831e-01
1.41048580e-01 7.11057708e-02 -3.53485383e-02 -2.02473253e-01
4.75380212e-01 3.67729902e-01 2.63277203e-01 1.80974707e-01
1.54568538e-01 -3.66969615e-01 1.73996910e-01 1.69784978e-01
1.19945848e+00 3.50855887e-01 4.41234112e-01 -4.92399544e-01
3.50688249e-01 5.32129779e-02 4.39451128e-01 6.66332543e-01
-1.90989748e-01 8.85780573e-01 6.45702183e-01 -5.98339677e-01
-6.80031598e-01 -1.05102146e+00 -4.54312325e-01 9.68097329e-01
5.12830555e-01 -6.09069765e-01 -6.20018780e-01 -8.44300985e-01
1.07665658e-02 2.67982155e-01 -5.33469260e-01 2.03656524e-01
-6.68352425e-01 -6.46835923e-01 4.31446552e-01 8.43773425e-01
8.36080015e-01 -9.55090284e-01 -6.03796899e-01 -1.48567259e-02
4.45286045e-03 -1.48484981e+00 -1.75346568e-01 2.32502799e-02
-1.03237641e+00 -1.21999764e+00 -2.28858963e-01 -8.98462474e-01
6.81939781e-01 2.48798028e-01 1.21814263e+00 1.70633078e-01
-8.12309533e-02 1.87654540e-01 -3.91199350e-01 -2.24210083e-01
1.24074504e-01 6.06371939e-01 -2.58043647e-01 -2.26944536e-01
1.19221956e-01 -7.77300894e-01 -7.36408949e-01 7.04624891e-01
-4.58051026e-01 3.43769938e-02 4.21229690e-01 2.88780928e-01
9.25923765e-01 -1.61469914e-02 -1.45710749e-03 -8.85237455e-01
4.25529003e-01 -1.78877618e-02 -3.54311645e-01 1.24437004e-01
-4.42566007e-01 -1.48824081e-01 2.10087746e-01 -2.00245932e-01
-7.56687403e-01 3.43759775e-01 -1.42811552e-01 -2.80411303e-01
-1.27111316e-01 2.89704055e-01 -7.38255084e-02 -5.09207904e-01
8.00071359e-01 -4.20531392e-01 -1.00980522e-02 -3.76553774e-01
7.10916162e-01 1.98314875e-01 8.28150272e-01 -7.64996886e-01
6.00479901e-01 9.25943434e-01 4.41578895e-01 -6.83133185e-01
-9.13457036e-01 -7.13732243e-01 -8.90683711e-01 -4.07165550e-02
8.98837328e-01 -6.26752138e-01 -8.82645786e-01 3.42783749e-01
-1.31037283e+00 -2.52713233e-01 -6.44042075e-01 1.23928130e-01
-6.84893548e-01 1.55136347e-01 -7.06131220e-01 -2.41736621e-01
-3.22665900e-01 -1.16717577e+00 1.24596393e+00 2.14584917e-01
-9.86860693e-02 -1.09348643e+00 -2.30431035e-01 3.47074240e-01
-3.60635705e-02 4.01757449e-01 4.64830935e-01 -6.16837740e-01
-7.42490292e-01 2.15227846e-02 -6.55566573e-01 2.92463005e-01
3.61858830e-02 1.94497958e-01 -8.86805952e-01 2.24060267e-01
-8.01114589e-02 2.05461383e-01 1.01552892e+00 2.98926711e-01
1.22329843e+00 3.36543880e-02 -5.97442448e-01 9.96625185e-01
1.30257595e+00 -1.83113843e-01 7.46530592e-01 2.47216433e-01
1.20482469e+00 7.40669906e-01 5.83532035e-01 1.57655239e-01
5.56673050e-01 6.96474433e-01 5.47111154e-01 -3.09502631e-01
-4.39131826e-01 -3.88226397e-02 -2.63906121e-01 6.53651059e-01
-5.38966179e-01 9.30926483e-03 -1.22682035e+00 3.60358238e-01
-2.11133170e+00 -5.25272131e-01 -4.88574505e-01 1.71412075e+00
4.88898039e-01 2.34116420e-01 1.15130648e-01 6.71621859e-02
6.30385399e-01 1.28767431e-01 -2.57570267e-01 -1.80770725e-01
-2.20591873e-01 4.46458250e-01 7.00571299e-01 4.00992453e-01
-1.34341776e+00 1.26350725e+00 6.56790018e+00 8.34583163e-01
-9.46186662e-01 -1.05670452e-01 6.59802437e-01 1.52308702e-01
3.93449180e-02 7.43737370e-02 -7.92212903e-01 -4.99011166e-02
2.50613242e-01 1.97231531e-01 1.57514140e-01 8.70533884e-01
-1.06869832e-01 -4.39904779e-02 -1.39168811e+00 1.01282012e+00
3.88516001e-02 -1.51565456e+00 -1.37943607e-02 1.31852105e-01
7.15951920e-01 3.43237370e-01 -9.68544558e-02 -1.52308896e-01
3.47740144e-01 -8.85437071e-01 6.12490356e-01 4.11449045e-01
4.38380986e-01 -6.17441118e-01 7.41480947e-01 9.88523513e-02
-1.45602405e+00 1.53915688e-01 -5.49626529e-01 -1.16206281e-01
1.09344147e-01 6.75687313e-01 -8.39038968e-01 8.09978902e-01
8.75873744e-01 1.09216440e+00 -7.94224322e-01 1.04759049e+00
-4.60889906e-01 3.57128650e-01 -8.75123262e-01 2.83500105e-01
1.83306575e-01 -3.03655833e-01 4.94509876e-01 1.07227314e+00
7.61283934e-02 1.89768881e-01 4.03692186e-01 7.51336932e-01
-5.53589780e-03 6.50642738e-02 -4.76052672e-01 5.68645298e-01
5.09638250e-01 1.51866937e+00 -1.19209766e+00 -1.66773871e-01
-1.99470252e-01 6.32737339e-01 4.40675020e-01 2.18684852e-01
-7.31174886e-01 -7.03328401e-02 6.62432015e-01 1.47703707e-01
4.55907971e-01 -3.70549172e-01 -9.13945556e-01 -8.41672540e-01
2.60979086e-01 -4.47777957e-01 3.42255086e-01 -6.67401254e-01
-1.31884670e+00 6.07944071e-01 7.81281069e-02 -1.03456020e+00
1.72634721e-01 -6.68249369e-01 -4.42526311e-01 5.05703151e-01
-1.15347636e+00 -1.41365647e+00 -4.78406549e-01 4.55382854e-01
4.01167005e-01 2.49450311e-01 6.03298008e-01 1.24063969e-01
-4.81233120e-01 3.76346767e-01 -6.94245458e-01 3.41323614e-01
3.45754534e-01 -1.38369632e+00 8.57391536e-01 5.37689567e-01
3.55956882e-01 6.80584848e-01 4.50440198e-01 -5.34060001e-01
-8.72984111e-01 -1.12101877e+00 8.95944357e-01 -6.41634166e-01
7.39190459e-01 -2.91142166e-01 -7.15304911e-01 6.65365458e-01
-1.01599760e-01 6.73895001e-01 2.51749486e-01 4.36768472e-01
-5.42788863e-01 -1.90155759e-01 -1.30699456e+00 7.69812882e-01
1.79542899e+00 -3.98515135e-01 -5.22399426e-01 3.70881617e-01
1.05959129e+00 -7.29465485e-01 -1.28436244e+00 7.33510971e-01
3.98716062e-01 -1.17798555e+00 1.31909585e+00 -8.42118636e-02
3.87493789e-01 -4.02757317e-01 -6.39663711e-02 -1.06590104e+00
-4.76919472e-01 -5.30724287e-01 3.11515648e-02 1.14819491e+00
4.22745883e-01 -9.23440278e-01 1.01155829e+00 3.81077200e-01
-4.12082046e-01 -1.29294384e+00 -8.50915611e-01 -8.07905197e-01
-8.81384015e-02 -6.95995152e-01 7.12128162e-01 7.71244228e-01
-1.51981935e-01 4.10353124e-01 3.99091780e-01 2.35740662e-01
6.07337534e-01 1.52188942e-01 8.18516910e-01 -1.48381817e+00
-4.86144908e-02 -6.20413184e-01 -9.81145859e-01 -1.22944117e+00
4.13638540e-02 -8.78065884e-01 -5.16681001e-02 -1.89465666e+00
-8.22016150e-02 -8.24155211e-01 -4.60733958e-02 7.03469515e-01
-2.34617982e-02 6.66245103e-01 3.76065254e-01 4.42255318e-01
-8.74464393e-01 2.10127175e-01 1.49002469e+00 4.93466808e-03
-3.11800301e-01 -4.96799387e-02 -6.45448565e-01 1.05368912e+00
7.53494561e-01 -2.75417715e-01 -4.11982179e-01 -4.72644866e-01
1.37145042e-01 -4.24960881e-01 4.98526424e-01 -1.21642566e+00
5.20469189e-01 9.63667855e-02 -3.29447575e-02 -8.34501445e-01
3.35826784e-01 -7.01002896e-01 8.09028521e-02 -2.12093182e-02
-2.12547675e-01 5.21029569e-02 -1.11323886e-01 3.76975983e-01
-1.59100294e-01 1.31078064e-01 5.33701956e-01 -6.29093200e-02
-6.87509894e-01 4.80084002e-01 4.95593667e-01 -1.16851628e-01
1.09284031e+00 -5.96593559e-01 -5.40841103e-01 -3.87444277e-03
-8.57742250e-01 3.36470157e-01 4.62911487e-01 2.31894925e-01
5.50686717e-01 -1.09531176e+00 -4.99488115e-01 5.24727292e-02
6.80607632e-02 7.84155607e-01 -1.55031383e-01 9.45783317e-01
-9.80722785e-01 2.90444762e-01 3.81001681e-02 -1.11815953e+00
-1.34340739e+00 3.85158420e-01 3.55127722e-01 -1.34643763e-01
-9.82307673e-01 1.02859712e+00 2.76053339e-01 -4.92575645e-01
2.52481371e-01 -7.60631263e-01 -1.05318345e-01 1.19777903e-01
7.07099363e-02 5.78977406e-01 3.59691083e-01 -6.05189264e-01
-5.81575155e-01 9.33634937e-01 -1.13994852e-02 1.03967860e-01
1.27064312e+00 1.32467598e-01 -4.91411328e-01 1.26059562e-01
1.04635608e+00 -1.34935185e-01 -1.26261747e+00 -3.48025441e-01
5.16596213e-02 -2.60614008e-01 -5.04560322e-02 -4.24314767e-01
-1.33224308e+00 4.55692828e-01 1.40705183e-01 4.94518191e-01
1.03240085e+00 6.22197449e-01 6.95324063e-01 3.98511946e-01
5.98956108e-01 -9.44646358e-01 -1.60771564e-01 5.88885546e-01
8.73167813e-01 -8.69290352e-01 4.11135584e-01 -1.46460414e+00
-3.33903164e-01 9.04040337e-01 4.52939063e-01 -5.92332304e-01
9.68579233e-01 2.37198025e-01 7.66590089e-02 -5.35640776e-01
-5.78008890e-01 -4.04424727e-01 5.42118192e-01 8.12096953e-01
2.11146206e-01 1.53755501e-01 -3.49070668e-01 3.51296403e-02
-7.54740119e-01 -2.89887369e-01 -6.49078488e-02 7.28609085e-01
-2.42304936e-01 -1.02596486e+00 -2.92349428e-01 4.27798748e-01
-1.26872286e-01 9.36312750e-02 -5.94964623e-01 7.99857914e-01
6.00105166e-01 9.17043328e-01 5.08796155e-01 -4.63678926e-01
5.87007403e-01 -4.69901323e-01 5.35792470e-01 -7.38776267e-01
-7.51256764e-01 -6.50094301e-02 2.47244135e-01 -1.05837059e+00
-6.80870533e-01 -7.76074469e-01 -1.47038031e+00 -2.78731614e-01
-4.02592778e-01 -3.16758484e-01 6.55458272e-01 9.24172223e-01
4.58371162e-01 5.00868678e-01 2.31758863e-01 -8.42944980e-01
-3.34110916e-01 -6.64777040e-01 -2.17371792e-01 3.72609109e-01
-2.01814938e-02 -6.24534547e-01 -2.15062127e-01 -4.04582433e-02]
|
[7.94253396987915, -3.346328020095825]
|
1a889270-cbe7-443a-8322-a31bd472431c
|
revisiting-few-shot-relation-classification
|
2104.08481
| null |
https://arxiv.org/abs/2104.08481v1
|
https://arxiv.org/pdf/2104.08481v1.pdf
|
Revisiting Few-shot Relation Classification: Evaluation Data and Classification Schemes
|
We explore Few-Shot Learning (FSL) for Relation Classification (RC). Focusing on the realistic scenario of FSL, in which a test instance might not belong to any of the target categories (none-of-the-above, aka NOTA), we first revisit the recent popular dataset structure for FSL, pointing out its unrealistic data distribution. To remedy this, we propose a novel methodology for deriving more realistic few-shot test data from available datasets for supervised RC, and apply it to the TACRED dataset. This yields a new challenging benchmark for FSL RC, on which state of the art models show poor performance. Next, we analyze classification schemes within the popular embedding-based nearest-neighbor approach for FSL, with respect to constraints they impose on the embedding space. Triggered by this analysis we propose a novel classification scheme, in which the NOTA category is represented as learned vectors, shown empirically to be an appealing option for FSL.
|
['Ido Dagan', 'Yoav Goldberg', 'Yanai Elazar', 'Ofer Sabo']
|
2021-04-17
| null | null | null | null |
['few-shot-relation-classification', 'few-shot-relation-classification']
|
['methodology', 'natural-language-processing']
|
[ 4.13925052e-01 5.04489005e-01 -3.30528408e-01 -2.62157351e-01
-5.31928241e-01 -5.00728667e-01 9.23876345e-01 3.47294331e-01
-2.21906602e-01 9.37950850e-01 1.98130295e-01 -3.48351210e-01
-8.41571212e-01 -9.30541217e-01 -3.35814744e-01 -8.10372412e-01
-6.14908598e-02 5.66168070e-01 4.30489272e-01 -3.15247297e-01
7.13327900e-02 7.62323856e-01 -1.84533286e+00 3.33794028e-01
6.18718624e-01 8.71541023e-01 -3.36046845e-01 4.07436788e-01
-1.53902043e-02 8.29349935e-01 -6.23724937e-01 -6.86508238e-01
-1.58123448e-02 -5.16810477e-01 -1.30334198e+00 -1.66845843e-01
2.50663996e-01 6.41576573e-02 -1.52489439e-01 7.37016082e-01
3.84338140e-01 4.47317928e-01 1.04464889e+00 -1.45096779e+00
-4.86984789e-01 7.13680685e-01 1.01172224e-01 1.88821852e-02
5.37355006e-01 -1.55719876e-01 1.30349696e+00 -8.13657224e-01
1.08388245e+00 9.86628830e-01 6.43853903e-01 5.41666806e-01
-1.52041221e+00 -3.42129320e-02 -3.91294435e-02 8.21098328e-01
-1.34896171e+00 -5.39062440e-01 6.82623565e-01 -3.98910999e-01
1.01772130e+00 6.50197864e-01 5.20915508e-01 1.48857105e+00
-2.70758599e-01 6.81689620e-01 1.05957651e+00 -8.61981094e-01
7.20483541e-01 3.15888762e-01 5.33335626e-01 1.56490773e-01
3.56014162e-01 1.54325962e-01 -3.82506430e-01 -3.16803664e-01
-6.74390197e-02 -3.20297748e-01 -2.99329251e-01 -1.00639319e+00
-9.84073579e-01 8.20485115e-01 2.41222620e-01 7.51941979e-01
-4.87819649e-02 -4.40076023e-01 4.77352768e-01 5.51828802e-01
5.16544044e-01 5.87324917e-01 -5.43404162e-01 -1.70071572e-01
-7.44244814e-01 1.84043229e-01 1.02135897e+00 7.13127971e-01
5.74730277e-01 -3.66140217e-01 -6.11866832e-01 8.64243090e-01
-1.10692613e-01 -3.56689632e-01 3.54899406e-01 -4.41212595e-01
2.37896547e-01 7.50717640e-01 -1.07355863e-02 -7.03132272e-01
-3.33731502e-01 -3.84100437e-01 -6.35215402e-01 1.39730737e-01
4.77010220e-01 2.69283146e-01 -4.36515898e-01 1.49855077e+00
4.74074632e-01 4.55630213e-01 5.68057120e-01 5.33908904e-01
7.24654615e-01 2.01841205e-01 -5.14630228e-02 -4.48246270e-01
1.10772765e+00 -6.96199179e-01 -5.88619053e-01 2.11870521e-01
1.34492648e+00 -3.33348900e-01 1.02031803e+00 5.36306918e-01
-4.71886873e-01 -4.56805557e-01 -1.14136994e+00 1.21154204e-01
-9.60292637e-01 8.11991170e-02 6.54853284e-01 7.54235148e-01
-7.85643578e-01 9.67913091e-01 -4.46649969e-01 -7.62827218e-01
3.73681664e-01 4.19239402e-02 -6.46983385e-01 -2.09644333e-01
-1.35336876e+00 1.13794136e+00 7.08771825e-01 -9.68578532e-02
-5.09598434e-01 -6.24319315e-01 -9.08028781e-01 1.76535979e-01
8.31437707e-01 -1.29638672e-01 7.94965208e-01 -4.11689013e-01
-1.36520386e+00 9.47046041e-01 1.16263434e-01 -5.99754632e-01
6.02395892e-01 -2.26585925e-01 -6.89807415e-01 -5.03671728e-02
-1.94973260e-01 2.12566987e-01 6.69320107e-01 -1.31007457e+00
-1.41967356e-01 -9.98845175e-02 3.21929783e-01 -3.46001297e-01
-4.79085594e-01 -2.62786653e-02 1.37120232e-01 -4.07426536e-01
-1.50864854e-01 -7.46930540e-01 8.28241855e-02 -9.46363807e-02
-6.23719871e-01 -6.64226592e-01 8.09460878e-01 -5.78067563e-02
1.53372765e+00 -2.18973589e+00 3.62435877e-01 3.04206342e-01
2.53002942e-02 5.85788190e-01 -2.28288323e-01 7.24940777e-01
-6.17087066e-01 -6.15657642e-02 -2.75129616e-01 -1.02756083e-01
2.00226337e-01 4.45948362e-01 -2.88226932e-01 4.95700091e-01
4.11479115e-01 9.67680812e-01 -1.21224046e+00 -6.28643513e-01
2.93020219e-01 2.83417761e-01 -2.59849578e-01 2.60058701e-01
-1.75974488e-01 1.37263998e-01 -2.94550210e-01 6.10887587e-01
5.05461931e-01 -3.48866284e-02 4.18654978e-01 -4.23546791e-01
3.32222506e-02 5.03532253e-02 -1.28309035e+00 1.49778175e+00
-3.75925213e-01 3.26368958e-01 -8.09408486e-01 -1.35010207e+00
1.12202227e+00 3.40294361e-01 4.32714701e-01 -2.90675610e-01
1.33212626e-01 2.12086797e-01 2.16281101e-01 -5.72926044e-01
1.32605240e-01 7.57863373e-02 -3.63476723e-02 2.98635930e-01
4.46282208e-01 1.22735217e-01 4.00366336e-01 -5.50266020e-02
1.52606893e+00 4.04833674e-01 8.36837947e-01 -2.69232959e-01
7.52772331e-01 -2.16685876e-01 3.29211235e-01 7.67931104e-01
-2.07882047e-01 4.37803507e-01 8.25310767e-01 -4.75862831e-01
-7.58382320e-01 -1.01115406e+00 -3.73931766e-01 8.91920149e-01
-2.36586601e-01 -6.93847418e-01 -4.52174395e-01 -1.05317795e+00
5.69095723e-02 1.01580298e+00 -9.42349494e-01 -5.13157845e-01
-2.47540444e-01 -6.60115123e-01 4.79461968e-01 2.46544272e-01
1.21174879e-01 -1.04674339e+00 -7.25906551e-01 7.04739094e-02
1.81053475e-01 -8.52337897e-01 3.51171076e-01 5.61777115e-01
-5.48043966e-01 -1.46803010e+00 -5.40962219e-01 -5.41368365e-01
2.22211465e-01 -3.16700459e-01 1.31757021e+00 -6.98785782e-02
-4.31079119e-01 3.38923931e-01 -9.39046144e-01 1.77562237e-02
-4.63813007e-01 2.75174081e-01 5.52891046e-02 3.56288493e-01
7.50823677e-01 -6.40850842e-01 -1.04317345e-01 2.13010460e-01
-8.81716967e-01 -4.80402410e-01 3.80149454e-01 8.69347572e-01
2.76726991e-01 -1.51394438e-02 6.69989526e-01 -1.34231460e+00
5.01984119e-01 -6.51830554e-01 -2.64284462e-01 7.27817416e-01
-6.38775170e-01 2.00036824e-01 6.34256899e-01 -6.30396903e-01
-6.22319520e-01 -6.67067245e-02 -5.52275553e-02 -4.65136915e-01
-4.45041925e-01 4.00995851e-01 -4.18998033e-01 -1.52933463e-01
9.28744674e-01 -1.84531522e-03 -2.67348945e-01 -5.49554408e-01
4.29733753e-01 6.91593647e-01 2.89203078e-01 -5.21531999e-01
8.02222073e-01 2.64958233e-01 3.39393586e-01 -8.41287613e-01
-1.00318873e+00 -4.24827754e-01 -1.10622311e+00 -1.52874604e-01
6.94019198e-01 -3.06793332e-01 -7.12886155e-01 -7.56462440e-02
-1.11389935e+00 -2.22259566e-01 -7.90950894e-01 5.34367025e-01
-7.76411593e-01 2.78255254e-01 -3.61637533e-01 -9.10727620e-01
1.18891299e-01 -7.81189740e-01 8.50447297e-01 -2.65084773e-01
-6.16862953e-01 -1.06140745e+00 3.13048512e-01 -1.17349401e-01
2.18485355e-01 4.32044864e-01 1.35829127e+00 -1.37208974e+00
-9.43991616e-02 -1.95822507e-01 -2.32774876e-02 2.94722289e-01
2.56099999e-01 -3.50878946e-02 -1.25975525e+00 -2.17574611e-01
-1.10667162e-01 -4.05693531e-01 8.89977276e-01 -2.07551017e-01
1.14841104e+00 1.76894362e-03 -5.09239674e-01 2.94431657e-01
1.38905752e+00 1.63391475e-02 7.88325131e-01 3.81187469e-01
4.20901209e-01 7.51687229e-01 9.82432485e-01 3.97731513e-01
-7.90244639e-02 1.03261399e+00 3.68745863e-01 2.45902672e-01
-1.17082819e-01 -1.65940002e-02 1.48604214e-01 6.26455247e-01
-1.02643430e-01 -3.55063558e-01 -7.43813276e-01 5.86257517e-01
-1.87440789e+00 -9.25058067e-01 -8.29031840e-02 2.36276388e+00
6.76512301e-01 2.57979125e-01 1.01693429e-01 7.76709437e-01
4.88796860e-01 4.40815985e-01 -2.11424008e-01 -6.35684013e-01
-2.14717209e-01 5.32537580e-01 -1.41697107e-02 3.96674186e-01
-1.27420402e+00 8.19111824e-01 5.76825333e+00 1.13810921e+00
-6.77565575e-01 1.86135605e-01 2.06377342e-01 2.30818838e-01
-2.53621459e-01 2.31363505e-01 -7.74600267e-01 2.34364524e-01
1.17265677e+00 -1.19889773e-01 2.23557189e-01 8.40498805e-01
-2.91332483e-01 -1.00214735e-01 -1.66570520e+00 6.81831896e-01
3.60178322e-01 -1.23778784e+00 1.63753852e-01 -5.38556501e-02
3.91648173e-01 -5.33838987e-01 -2.70555228e-01 6.29766703e-01
5.82967326e-02 -9.50394034e-01 2.92666107e-01 5.86649597e-01
7.81425714e-01 -6.39029026e-01 8.20922792e-01 2.38564655e-01
-9.88438487e-01 -1.42606273e-01 -5.53205252e-01 3.35975997e-02
-2.58240700e-01 5.80466509e-01 -6.37115240e-01 1.03015292e+00
2.78602421e-01 8.87607753e-01 -8.82810950e-01 9.43818092e-01
-1.34972557e-01 5.18700421e-01 1.75002590e-02 -9.48489681e-02
4.72167991e-02 -7.08358660e-02 5.56897342e-01 1.25373888e+00
2.94129312e-01 -1.50355190e-01 2.77931094e-02 6.93963885e-01
2.39658430e-01 2.78358340e-01 -1.12193966e+00 5.18158525e-02
4.08242434e-01 1.15171778e+00 -7.64822304e-01 -2.16534510e-01
-4.38788414e-01 9.21547949e-01 5.91939747e-01 2.04539791e-01
-6.69787467e-01 -5.46095252e-01 4.91547018e-01 -9.54614133e-02
3.97554874e-01 3.40472907e-01 2.84890950e-01 -1.11070526e+00
7.24754333e-02 -7.64991701e-01 5.74555039e-01 -5.49091578e-01
-1.43228364e+00 7.06755042e-01 3.65968406e-01 -1.55512011e+00
-4.04051811e-01 -8.07441473e-01 -5.72737217e-01 4.53966945e-01
-1.42352760e+00 -1.17186344e+00 -5.17232977e-02 4.96269614e-01
4.56218988e-01 -3.03776324e-01 1.38645005e+00 3.20745945e-01
-4.09229785e-01 6.73527598e-01 9.51076075e-02 -1.39963090e-01
6.68738186e-01 -1.25344706e+00 1.11250512e-01 5.86405277e-01
5.60845435e-01 4.02807206e-01 8.26244056e-01 -4.53257024e-01
-1.13028371e+00 -1.12960720e+00 1.21976638e+00 -6.38466477e-01
1.03772664e+00 -5.31966150e-01 -1.16097772e+00 5.40818095e-01
-2.41779745e-01 3.91775370e-01 8.66930544e-01 4.05083120e-01
-5.45362353e-01 -1.19770184e-01 -1.10728788e+00 5.05840123e-01
1.24314106e+00 -7.11336553e-01 -8.87623131e-01 3.91802639e-01
6.00938916e-01 1.84318736e-01 -1.03884113e+00 5.79287112e-01
3.56886595e-01 -9.74196017e-01 9.29647565e-01 -1.00569630e+00
3.36531430e-01 -8.41919705e-02 -4.17403787e-01 -1.33100379e+00
-2.33370095e-01 -3.53842199e-01 -5.06364584e-01 1.47552240e+00
3.03417236e-01 -5.88685453e-01 7.36485004e-01 1.29236400e-01
9.17348638e-02 -9.52604055e-01 -1.36382413e+00 -1.15351045e+00
-6.67558052e-03 -4.47488338e-01 6.69442296e-01 1.14100075e+00
1.81672573e-01 2.51782477e-01 -2.71868378e-01 -1.25108406e-01
4.19426739e-01 2.08035752e-01 6.33839726e-01 -1.67518604e+00
-4.99721557e-01 -2.23865166e-01 -9.78906214e-01 -2.40378797e-01
5.58723092e-01 -1.06330538e+00 -2.68396407e-01 -1.17488396e+00
5.90813439e-03 -3.68045986e-01 -6.09064817e-01 4.70951051e-01
1.33088946e-01 3.13323885e-01 1.05308652e-01 -1.41341239e-01
-6.30227447e-01 5.10866582e-01 6.97298706e-01 -1.92670435e-01
4.22868095e-02 1.08734667e-01 -2.60150224e-01 6.02673590e-01
4.27704245e-01 -3.77918214e-01 -5.51203310e-01 3.00117522e-01
3.46788585e-01 -1.41203851e-01 4.31231350e-01 -1.11506808e+00
-6.59291968e-02 -5.60928024e-02 2.62132119e-02 -3.64641517e-01
4.89393026e-01 -1.00195324e+00 1.51678607e-01 3.96361262e-01
-5.75052202e-01 -6.09918475e-01 -1.55419827e-01 4.72968876e-01
-2.42950976e-01 -6.22104168e-01 4.99312043e-01 2.81670839e-01
-7.76297033e-01 -1.26742527e-01 6.21100888e-04 4.83735017e-02
1.33252478e+00 -3.52402031e-01 -4.76302564e-01 5.93594052e-02
-1.21015000e+00 -1.73789576e-01 2.71867037e-01 4.00597692e-01
4.71247822e-01 -1.40929008e+00 -6.32873595e-01 2.72013187e-01
8.71524096e-01 -5.95816851e-01 1.20304011e-01 8.92788649e-01
-9.86419246e-02 4.00081992e-01 -4.36506905e-02 -3.20752144e-01
-1.23337853e+00 1.02711415e+00 9.81409848e-02 -4.68687564e-01
-6.21747613e-01 5.72940528e-01 -3.10004413e-01 -4.58873391e-01
1.37360096e-01 -3.28731656e-01 -7.13885069e-01 4.74140465e-01
2.98902214e-01 4.51667935e-01 4.30144101e-01 -4.82140124e-01
-4.83125865e-01 4.07478243e-01 8.93752202e-02 2.36208051e-01
1.44558132e+00 2.43621111e-01 1.89083833e-02 1.04577363e+00
1.20338762e+00 2.29537264e-02 -6.81530893e-01 -2.01168805e-01
6.07383251e-01 -4.99275595e-01 -4.08170372e-01 -6.16076410e-01
-4.58546221e-01 8.06857169e-01 5.01596928e-01 5.63887835e-01
8.84602487e-01 2.36528918e-01 1.12788849e-01 6.41434312e-01
6.33406103e-01 -9.02068317e-01 -2.08429784e-01 4.77894634e-01
7.65507877e-01 -1.04012489e+00 1.50688171e-01 -6.28410399e-01
-3.47934306e-01 1.23714542e+00 3.14846933e-01 -1.82665542e-01
7.00085580e-01 -1.56304482e-02 -3.72488767e-01 -1.83112368e-01
-9.27177608e-01 -4.69494402e-01 2.73033768e-01 8.73824716e-01
3.77133220e-01 1.60657242e-02 -5.95816255e-01 3.95208806e-01
-2.92391013e-02 -3.01524717e-03 4.17839229e-01 8.03591967e-01
8.36847629e-03 -1.40317786e+00 2.84589026e-02 7.14525759e-01
-2.64060982e-02 2.67100390e-02 -5.84766686e-01 9.98039484e-01
3.34044367e-01 9.18934584e-01 -1.26145944e-01 -3.85155410e-01
4.67963219e-01 5.07554114e-01 4.72005129e-01 -7.84397781e-01
-4.81290758e-01 -4.96940881e-01 4.63786036e-01 -5.33421218e-01
-5.03886998e-01 -5.63386977e-01 -4.93261516e-01 1.80084631e-01
-4.47478056e-01 2.58556902e-01 2.07445323e-01 1.16545463e+00
6.69312850e-02 4.89589453e-01 5.61826169e-01 -7.07399368e-01
-6.59701407e-01 -9.93243814e-01 -8.41061532e-01 6.75790787e-01
1.03823498e-01 -1.15933669e+00 -5.69435179e-01 -3.42648357e-01]
|
[9.82834243774414, 3.0138676166534424]
|
8c665b90-c72e-47f5-8407-4937afd6474c
|
exploiting-unlabelled-photos-for-stronger
|
2303.13779
| null |
https://arxiv.org/abs/2303.13779v1
|
https://arxiv.org/pdf/2303.13779v1.pdf
|
Exploiting Unlabelled Photos for Stronger Fine-Grained SBIR
|
This paper advances the fine-grained sketch-based image retrieval (FG-SBIR) literature by putting forward a strong baseline that overshoots prior state-of-the-arts by ~11%. This is not via complicated design though, but by addressing two critical issues facing the community (i) the gold standard triplet loss does not enforce holistic latent space geometry, and (ii) there are never enough sketches to train a high accuracy model. For the former, we propose a simple modification to the standard triplet loss, that explicitly enforces separation amongst photos/sketch instances. For the latter, we put forward a novel knowledge distillation module can leverage photo data for model training. Both modules are then plugged into a novel plug-n-playable training paradigm that allows for more stable training. More specifically, for (i) we employ an intra-modal triplet loss amongst sketches to bring sketches of the same instance closer from others, and one more amongst photos to push away different photo instances while bringing closer a structurally augmented version of the same photo (offering a gain of ~4-6%). To tackle (ii), we first pre-train a teacher on the large set of unlabelled photos over the aforementioned intra-modal photo triplet loss. Then we distill the contextual similarity present amongst the instances in the teacher's embedding space to that in the student's embedding space, by matching the distribution over inter-feature distances of respective samples in both embedding spaces (delivering a further gain of ~4-5%). Apart from outperforming prior arts significantly, our model also yields satisfactory results on generalising to new classes. Project page: https://aneeshan95.github.io/Sketch_PVT/
|
['Yi-Zhe Song', 'Tao Xiang', 'Soumitri Chattopadhyay', 'Pinaki Nath Chowdhury', 'Subhadeep Koley', 'Ayan Kumar Bhunia', 'Aneeshan Sain']
|
2023-03-24
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Sain_Exploiting_Unlabelled_Photos_for_Stronger_Fine-Grained_SBIR_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Sain_Exploiting_Unlabelled_Photos_for_Stronger_Fine-Grained_SBIR_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['sketch-based-image-retrieval']
|
['computer-vision']
|
[ 3.90842795e-01 4.64970618e-02 -1.08603299e-01 -1.68228388e-01
-1.25353837e+00 -8.29867423e-01 8.93277407e-01 -1.02371596e-01
-2.61588573e-01 4.93463784e-01 1.76303312e-01 -1.17807016e-01
-3.78480613e-01 -5.72699666e-01 -8.50057781e-01 -8.02726984e-01
1.46381646e-01 4.78395998e-01 2.27278858e-01 -1.05832063e-01
3.09621483e-01 6.30930305e-01 -1.60953271e+00 3.93801391e-01
5.68044007e-01 1.02910852e+00 6.26367182e-02 4.79327738e-01
-1.05910122e-01 3.00290018e-01 -3.00854355e-01 -6.15927219e-01
6.38568223e-01 -1.88339248e-01 -7.69839525e-01 1.26826033e-01
1.18632674e+00 -2.71544307e-01 -4.13173616e-01 6.96324885e-01
4.95674163e-01 2.00684771e-01 8.25590193e-01 -1.30839992e+00
-8.72721374e-01 1.28888533e-01 -5.85200429e-01 -1.48377344e-01
3.01063359e-01 1.47865057e-01 1.24918854e+00 -1.05170453e+00
6.73066080e-01 1.23124230e+00 6.11034274e-01 5.53283930e-01
-1.43066525e+00 -7.48911798e-01 2.62605876e-01 -4.27494757e-02
-1.49675953e+00 -5.24119496e-01 9.07163739e-01 -3.43112499e-01
6.27497792e-01 4.60171103e-01 5.00003040e-01 1.19600511e+00
-5.51383257e-01 9.24293399e-01 1.00794315e+00 -4.92855430e-01
-1.98438503e-02 3.46231967e-01 -2.65322685e-01 4.68697131e-01
-1.40525267e-01 3.64287458e-02 -4.57178026e-01 -1.18974477e-01
7.92361498e-01 1.62412852e-01 -2.42049143e-01 -8.31114709e-01
-1.00198174e+00 7.55403817e-01 8.08352709e-01 3.04774135e-01
-2.16314986e-01 2.50869155e-01 7.42284060e-02 2.83775359e-01
3.37734789e-01 4.70218956e-01 -3.50683153e-01 2.01523174e-02
-1.14317703e+00 4.04207498e-01 5.66908002e-01 8.66304100e-01
1.08839655e+00 -4.05615717e-01 -1.38605908e-01 9.56193686e-01
2.55991966e-01 2.85745233e-01 1.33269787e-01 -9.23891008e-01
5.19838095e-01 6.05640650e-01 -9.22067836e-02 -1.05469978e+00
3.38360459e-01 -2.77529091e-01 -5.58566034e-01 2.21384585e-01
4.80202556e-01 2.44956553e-01 -9.13602829e-01 1.77789664e+00
3.70766848e-01 2.64980972e-01 -3.35987806e-01 8.02635908e-01
4.70921457e-01 5.32746255e-01 1.22457156e-02 2.61888355e-01
1.27702141e+00 -1.09316146e+00 -5.99971786e-02 -2.43046910e-01
3.56487215e-01 -9.48703229e-01 1.41895556e+00 3.43308836e-01
-1.03457582e+00 -4.45889950e-01 -9.96137440e-01 -4.00391012e-01
-5.76643467e-01 1.29319966e-01 5.34223974e-01 4.30181205e-01
-1.20467389e+00 7.57219613e-01 -4.40606475e-01 -4.77266997e-01
6.55612469e-01 3.05436343e-01 -5.89528143e-01 -5.30633032e-01
-8.17658722e-01 6.80562437e-01 5.42866401e-02 -1.26552179e-01
-7.33036935e-01 -1.22100616e+00 -6.41479731e-01 3.92256640e-02
5.09187818e-01 -8.55591774e-01 8.12322557e-01 -1.07096469e+00
-1.45395911e+00 8.83556783e-01 9.33178142e-02 1.43845100e-02
7.71858335e-01 -2.20632464e-01 -5.31101190e-02 1.75211817e-01
6.53557479e-02 1.02608287e+00 1.16499197e+00 -1.53714573e+00
-2.54947782e-01 -3.38164032e-01 2.33540311e-01 2.59925276e-01
-5.36297619e-01 -2.25364625e-01 -8.10440183e-01 -8.38538647e-01
-9.44765136e-02 -1.12309766e+00 8.00609514e-02 5.31332612e-01
-2.73142546e-01 -3.44325364e-01 8.97555113e-01 -3.14502150e-01
1.01984739e+00 -2.30332923e+00 1.81805074e-01 3.04821253e-01
1.17016606e-01 3.95816267e-01 -5.85331559e-01 7.74496973e-01
-1.85731098e-01 9.74601582e-02 -2.60959476e-01 -7.23489702e-01
2.56085575e-01 3.31606776e-01 -5.07219315e-01 2.18271390e-01
4.33646321e-01 9.81527030e-01 -8.82102132e-01 -3.73765439e-01
2.94304490e-01 8.03657293e-01 -6.98992431e-01 7.61739016e-02
-1.16487384e-01 1.86082527e-01 -2.06239775e-01 4.37195718e-01
7.28027523e-01 -3.64527434e-01 5.67503944e-02 -3.40112418e-01
7.11837709e-02 2.18932301e-01 -1.30700672e+00 2.01696825e+00
-4.52125162e-01 4.64933962e-01 -8.28518346e-02 -8.49124134e-01
7.57522821e-01 7.62618929e-02 3.70971560e-01 -6.04879975e-01
-2.60672301e-01 1.05695993e-01 -3.93969268e-01 -2.73334265e-01
4.22691584e-01 -2.97856092e-01 1.77842051e-01 4.36517626e-01
1.36934772e-01 -1.42957062e-01 -2.16266271e-02 3.31090033e-01
9.83457267e-01 3.31655115e-01 -2.65537560e-01 -8.72650221e-02
3.97098690e-01 -4.70270991e-01 7.99245387e-02 5.89776576e-01
7.53410254e-03 9.03045833e-01 3.10715675e-01 -3.62862229e-01
-1.11568534e+00 -1.28708482e+00 -8.82737860e-02 1.24899578e+00
5.67861311e-02 -4.79647636e-01 -4.32143331e-01 -8.79379213e-01
3.33802700e-01 2.76624888e-01 -8.30539346e-01 3.44479010e-02
-3.86035860e-01 -2.35754132e-01 5.29634655e-01 4.51079488e-01
3.12027276e-01 -9.08880055e-01 -1.55128673e-01 -1.98359042e-01
-6.71594124e-03 -8.77283931e-01 -6.32803559e-01 -6.62889257e-02
-6.35010421e-01 -9.41654682e-01 -9.00817037e-01 -6.61701381e-01
7.45439827e-01 4.25872207e-01 1.05824399e+00 3.22381318e-01
-3.79215449e-01 4.96250957e-01 -2.54980296e-01 -8.15411061e-02
4.39000390e-02 1.88921794e-01 -1.80127770e-01 1.51288912e-01
1.12002373e-01 -7.98395157e-01 -9.11302209e-01 3.13852340e-01
-1.11967432e+00 -2.27662455e-02 7.15078413e-01 8.12447429e-01
4.11679953e-01 -2.83278793e-01 2.99374551e-01 -7.19239295e-01
3.31565470e-01 -2.43149891e-01 -3.46508265e-01 4.85584259e-01
-6.08525872e-01 2.15464801e-01 4.41322774e-01 -6.80612087e-01
-6.95319235e-01 4.31908183e-02 6.27338588e-02 -7.94429064e-01
-1.08723514e-01 1.65257126e-01 -1.40067890e-01 -2.09156126e-01
5.79413354e-01 9.38756615e-02 5.25684506e-02 -6.84843838e-01
7.90865064e-01 4.50397432e-01 4.91689801e-01 -8.75936210e-01
1.17702520e+00 5.69978595e-01 -1.14060350e-01 -7.89025605e-01
-6.42098904e-01 -5.08812428e-01 -6.63486004e-01 1.00702323e-01
5.76394141e-01 -7.99307168e-01 -6.30923986e-01 6.87960237e-02
-7.66285956e-01 -3.82206976e-01 -3.80825251e-01 1.43964171e-01
-4.93141621e-01 5.03903925e-01 -3.55410397e-01 -6.83816552e-01
-1.82795718e-01 -1.05092633e+00 1.45230985e+00 8.86937603e-02
-1.61229178e-01 -9.88002956e-01 2.20231861e-01 4.83664781e-01
3.55506301e-01 9.55250040e-02 9.10297334e-01 -4.87633586e-01
-8.30058873e-01 -3.97734046e-01 -5.66587508e-01 4.18830901e-01
5.95029183e-02 1.35751203e-01 -1.20304239e+00 -4.87294465e-01
-5.22860885e-01 -4.51762646e-01 9.17653143e-01 -6.33034334e-02
1.07545698e+00 -3.89297783e-01 -2.88079172e-01 6.01015031e-01
1.47661912e+00 -3.05063605e-01 8.50668550e-01 3.51897508e-01
7.94606864e-01 5.93770683e-01 3.42762262e-01 1.44432560e-01
4.48464721e-01 9.46844101e-01 2.88502634e-01 -1.71829313e-01
-3.96579623e-01 -6.25678658e-01 2.29307815e-01 5.87764442e-01
7.90968612e-02 -4.34304215e-02 -5.82600892e-01 7.60384202e-01
-1.76395071e+00 -8.77436638e-01 3.16330135e-01 2.38073564e+00
8.61258149e-01 -1.67448461e-01 2.87664443e-01 5.42851239e-02
3.96325767e-01 2.92943329e-01 -3.79896849e-01 -1.97362721e-01
8.88956115e-02 4.41302091e-01 2.49051511e-01 5.77998042e-01
-9.95204449e-01 9.72535253e-01 5.24495983e+00 1.17939949e+00
-1.21863127e+00 -1.00838631e-01 3.59921217e-01 -2.68110991e-01
-5.37491918e-01 2.87591219e-01 -6.18485332e-01 4.49109107e-01
5.62246621e-01 1.63326904e-01 6.92123473e-01 6.00856066e-01
-2.71986812e-01 1.17574953e-01 -1.29855871e+00 1.01389647e+00
1.30192980e-01 -1.30309308e+00 4.59605187e-01 2.98021257e-01
7.56903350e-01 -6.55325800e-02 3.93930912e-01 2.72468388e-01
1.09426327e-01 -1.09241366e+00 6.14667714e-01 5.15943587e-01
1.00180197e+00 -5.64358950e-01 2.76270628e-01 2.43562549e-01
-1.10100794e+00 -3.11458688e-02 -3.77319932e-01 6.65777028e-02
-2.50945717e-01 3.07178408e-01 -6.25969052e-01 6.91130698e-01
5.72926521e-01 6.29749119e-01 -7.21711278e-01 9.32408035e-01
-1.72398239e-01 2.13201806e-01 -4.47137177e-01 3.08036208e-01
4.50213045e-01 -2.30307624e-01 4.07293826e-01 1.12659311e+00
2.44524807e-01 -8.60363692e-02 9.02812183e-02 7.98134685e-01
-3.01938653e-01 -5.96400648e-02 -6.88630223e-01 -1.82957612e-02
5.37529767e-01 1.35114968e+00 -3.20443034e-01 -2.55668402e-01
-3.82039428e-01 1.22803259e+00 4.66286749e-01 4.66900617e-01
-5.48562944e-01 -3.76430839e-01 7.91591585e-01 2.56633848e-01
6.77321315e-01 9.66350809e-02 -7.80429989e-02 -1.10005713e+00
2.36241505e-01 -8.10448766e-01 3.81529331e-01 -8.44380796e-01
-1.64058733e+00 4.14066374e-01 -1.20818458e-01 -1.14318287e+00
-7.38929734e-02 -5.43307483e-01 -5.87918222e-01 9.79965866e-01
-1.73279047e+00 -1.70464790e+00 -1.89511865e-01 6.43504798e-01
3.33076090e-01 1.35715067e-01 8.33339572e-01 5.83797455e-01
-4.14150566e-01 1.00601339e+00 3.11763119e-02 5.39222285e-02
1.11295784e+00 -1.33270872e+00 3.37257415e-01 5.21848083e-01
5.13598144e-01 8.29170346e-01 4.58946973e-01 -2.91080624e-01
-1.36052215e+00 -9.52000082e-01 9.59800363e-01 -9.31478858e-01
7.53584921e-01 -5.52250326e-01 -9.84278917e-01 6.40876710e-01
3.89133655e-02 3.09729010e-01 5.09675145e-01 2.13349327e-01
-9.51276958e-01 -1.60921887e-01 -1.03087187e+00 6.79928124e-01
1.04008985e+00 -9.61411655e-01 -4.04836684e-01 1.89859822e-01
4.34472233e-01 -1.19142190e-01 -8.91230583e-01 2.84252942e-01
9.61461782e-01 -9.23230708e-01 1.35952020e+00 -6.70746803e-01
5.95736980e-01 -3.30738366e-01 -3.61365676e-01 -1.10545218e+00
-2.47342527e-01 -6.98340356e-01 7.37833455e-02 1.39843071e+00
2.06642389e-01 -3.84540498e-01 9.95821893e-01 6.97522521e-01
7.17761591e-02 -9.87078846e-01 -8.80355477e-01 -8.05186570e-01
4.51292515e-01 -3.44657451e-01 5.40278614e-01 1.01716113e+00
-2.62781560e-01 3.16474080e-01 -3.70736539e-01 3.93048897e-02
5.84086180e-01 3.28986645e-02 1.05229247e+00 -1.03122330e+00
-3.63780618e-01 -6.67148411e-01 -3.51868689e-01 -1.22257078e+00
-7.75282681e-02 -9.20212865e-01 -2.05725208e-01 -1.44539809e+00
2.99300283e-01 -7.66735852e-01 -3.86259943e-01 6.71584308e-01
-2.20326707e-01 7.19333112e-01 6.07735097e-01 3.56995463e-01
-5.97426414e-01 5.42124033e-01 1.16773641e+00 -2.94566955e-02
4.29536169e-03 -1.97595730e-01 -7.51402438e-01 4.16139036e-01
3.41143966e-01 -3.31512779e-01 -5.62444866e-01 -4.84538108e-01
2.45551795e-01 -2.51651675e-01 7.48623431e-01 -8.05862486e-01
1.71127632e-01 7.61525258e-02 3.81983072e-01 -3.61398667e-01
6.02019370e-01 -1.02273214e+00 2.39199251e-01 -7.93239474e-03
-3.69446337e-01 -1.98038742e-01 1.83504999e-01 6.16151810e-01
4.65106033e-02 4.71285218e-03 6.18342102e-01 2.09804680e-02
-3.63332540e-01 4.48326766e-01 2.55055487e-01 2.52056308e-02
8.03883255e-01 -4.32275325e-01 -3.67620379e-01 -2.43518025e-01
-5.32044113e-01 1.07832015e-01 7.28695035e-01 5.43954730e-01
3.69676083e-01 -1.52171743e+00 -5.25041938e-01 3.91967535e-01
4.26738381e-01 -2.82213558e-02 3.96870017e-01 7.87653804e-01
-1.57927439e-01 2.96109736e-01 -3.73770334e-02 -5.96776903e-01
-1.28989172e+00 5.31948864e-01 6.99489862e-02 -2.29109898e-01
-6.09583557e-01 9.77872968e-01 4.92424339e-01 -6.83192253e-01
3.28571975e-01 -7.27690905e-02 4.19772893e-01 1.75834998e-01
4.42459077e-01 2.05920517e-01 8.85121077e-02 -4.74499315e-01
-3.83817941e-01 7.64624715e-01 -2.57407844e-01 -1.37087807e-01
1.52937090e+00 -8.65341208e-05 1.57497227e-01 2.70090520e-01
1.69108498e+00 1.35423601e-01 -1.58846056e+00 -3.56197357e-01
-6.49960712e-02 -7.97829628e-01 -1.51331738e-01 -1.01910210e+00
-8.79372358e-01 9.98672187e-01 4.99483228e-01 7.22107738e-02
9.50184226e-01 1.93752006e-01 6.95048034e-01 1.57776445e-01
9.32434052e-02 -8.07783842e-01 4.29979056e-01 1.61909893e-01
9.76999342e-01 -1.16368496e+00 1.46692514e-01 -1.44694000e-01
-5.45057058e-01 9.32434320e-01 2.45929778e-01 -3.58248025e-01
5.37706971e-01 -1.30127549e-01 9.59995762e-02 -3.59335750e-01
-6.19446635e-01 -5.54596782e-02 6.43503368e-01 4.91395354e-01
2.36278653e-01 -9.73405242e-02 1.32230356e-01 5.69499768e-02
-6.98904172e-02 -1.46112535e-02 -8.40826929e-02 7.29585171e-01
-1.78933278e-01 -1.43729758e+00 -1.81758642e-01 3.58487964e-01
-1.51686192e-01 -1.75134197e-01 -5.63449502e-01 9.20084894e-01
1.13567196e-01 5.95327854e-01 7.91442171e-02 -2.58339942e-01
4.28684324e-01 1.69129670e-01 6.75116360e-01 -5.07524490e-01
-4.86561090e-01 9.96764004e-02 -2.04744026e-01 -4.97199357e-01
-3.11401039e-01 -6.11536503e-01 -6.27241671e-01 -4.03552890e-01
-2.65638858e-01 3.25356089e-02 7.34466136e-01 8.29688609e-01
4.97209579e-01 1.29591569e-01 6.67188287e-01 -1.16641641e+00
-6.03430510e-01 -7.10347354e-01 -3.45298529e-01 6.16425633e-01
4.42253947e-01 -6.94906533e-01 -5.79831541e-01 -1.03295885e-01]
|
[11.605019569396973, 0.5884585976600647]
|
0e04ebea-6990-40da-a190-507655811a3d
|
learning-to-blindly-assess-image-quality-in
|
1907.00516
| null |
https://arxiv.org/abs/1907.00516v3
|
https://arxiv.org/pdf/1907.00516v3.pdf
|
Learning to Blindly Assess Image Quality in the Laboratory and Wild
|
Computational models for blind image quality assessment (BIQA) are typically trained in well-controlled laboratory environments with limited generalizability to realistically distorted images. Similarly, BIQA models optimized for images captured in the wild cannot adequately handle synthetically distorted images. To face the cross-distortion-scenario challenge, we develop a BIQA model and an approach of training it on multiple IQA databases (of different distortion scenarios) simultaneously. A key step in our approach is to create and combine image pairs within individual databases as the training set, which effectively bypasses the issue of perceptual scale realignment. We compute a continuous quality annotation for each pair from the corresponding human opinions, indicating the probability of one image having better perceptual quality. We train a deep neural network for BIQA over the training set of massive image pairs by minimizing the fidelity loss. Experiments on six IQA databases demonstrate that the optimized model by the proposed training strategy is effective in blindly assessing image quality in the laboratory and wild, outperforming previous BIQA methods by a large margin.
|
['Xiaokang Yang', 'Kede Ma', 'Weixia Zhang', 'Guangtao Zhai']
|
2019-07-01
| null | null | null | null |
['blind-image-quality-assessment']
|
['computer-vision']
|
[ 1.53470039e-01 -4.76544976e-01 5.43213367e-01 -4.69068378e-01
-1.19883478e+00 -7.26007342e-01 3.68138641e-01 -1.10964112e-01
-4.42829728e-01 4.94093686e-01 2.15067908e-01 -2.92219102e-01
-2.31227711e-01 -6.09313250e-01 -8.35540891e-01 -4.68389004e-01
8.87626708e-02 1.93346068e-01 -1.39962867e-01 -9.59137753e-02
3.22275609e-01 4.25197452e-01 -1.63209772e+00 2.29823381e-01
1.25495744e+00 1.20985818e+00 3.08961738e-02 9.86635566e-01
5.26233852e-01 5.07646918e-01 -1.02984834e+00 -9.51490879e-01
9.71565068e-01 -4.96035457e-01 -5.94942510e-01 3.50543231e-01
1.07850397e+00 -8.31664801e-01 -3.33083242e-01 1.38335907e+00
1.03803813e+00 -1.44707501e-01 5.12810230e-01 -1.25667310e+00
-9.69909966e-01 -9.30285826e-02 -2.66144931e-01 2.89013118e-01
6.33628309e-01 7.42570639e-01 9.27109361e-01 -9.68223453e-01
1.34806916e-01 1.13018668e+00 5.60093641e-01 3.66747499e-01
-1.14770794e+00 -4.99442875e-01 -1.91946492e-01 2.82198966e-01
-1.25587201e+00 -6.50784016e-01 3.51588488e-01 -4.32147563e-01
8.31739366e-01 1.57733485e-01 6.74893618e-01 8.16188693e-01
-1.00684643e-01 5.32034278e-01 1.53099501e+00 -3.53413939e-01
2.66273767e-01 4.61650416e-02 -5.03757834e-01 4.48377192e-01
2.14675322e-01 4.94633675e-01 -5.42500973e-01 -4.51380834e-02
8.20031822e-01 -3.14184904e-01 -5.54527879e-01 -6.40773118e-01
-1.50626516e+00 4.50354338e-01 6.68990612e-01 -9.75993574e-02
-4.00200576e-01 -1.24173872e-01 8.42143297e-02 7.77868629e-01
4.53144312e-02 8.56561303e-01 -1.63902357e-01 3.70973423e-02
-1.00152946e+00 4.70810294e-01 3.62203747e-01 8.80386293e-01
5.32262266e-01 1.65610351e-02 -4.32542056e-01 8.86730552e-01
1.96330026e-01 7.01650321e-01 5.03115416e-01 -1.22496891e+00
5.76011837e-01 3.86524737e-01 7.61921644e-01 -8.35811973e-01
-1.85939912e-02 -4.33195829e-01 -7.06205785e-01 7.88149953e-01
6.09857857e-01 5.91105558e-02 -1.17357802e+00 1.61405480e+00
1.65717937e-02 -1.59584224e-01 1.52921855e-01 1.23477173e+00
5.05425036e-01 4.80160505e-01 -1.45299807e-02 -1.12180002e-01
1.15598965e+00 -8.34742248e-01 -6.13287926e-01 -2.05472499e-01
-7.78204203e-02 -8.70849609e-01 1.37024212e+00 7.52730429e-01
-1.43860185e+00 -7.99255252e-01 -1.38748145e+00 -7.75219128e-02
-2.55518317e-01 1.90840989e-01 1.50680527e-01 9.84311223e-01
-1.57592046e+00 3.71894807e-01 -1.98753178e-01 -2.19731227e-01
6.03809774e-01 3.88123244e-01 -4.73981291e-01 -4.44665343e-01
-9.68889654e-01 1.01441252e+00 -8.24741367e-03 1.98035538e-01
-1.33620453e+00 -6.70888662e-01 -8.33294213e-01 -3.42453420e-02
-1.06662154e-01 -1.09345818e+00 1.21431434e+00 -1.20471525e+00
-1.51564646e+00 1.02825117e+00 -5.59041202e-02 -4.59556103e-01
8.49497378e-01 -2.42727846e-01 -6.19754076e-01 2.56709248e-01
1.18610300e-01 7.75286615e-01 1.27165914e+00 -1.61208642e+00
-4.77515489e-01 -4.75429744e-01 3.98386419e-01 5.09773731e-01
-1.83945626e-01 2.08511427e-01 -3.66185129e-01 -6.82461560e-01
-1.51331410e-01 -5.33033013e-01 -1.91309471e-02 3.20893884e-01
-1.39608577e-01 3.74198586e-01 1.45010069e-01 -9.28918183e-01
8.66572797e-01 -2.03548408e+00 5.47313765e-02 1.25396371e-01
2.33455777e-01 5.48637867e-01 -4.57747459e-01 -2.39582565e-02
3.24025117e-02 -2.53487438e-01 -4.65057731e-01 -4.25987273e-01
8.71620178e-02 -1.39876619e-01 -1.73810527e-01 4.79662269e-01
4.18091983e-01 7.90446520e-01 -8.89768064e-01 -3.32539648e-01
2.02195153e-01 3.46326351e-01 -6.49197280e-01 8.66625190e-01
1.07152753e-01 3.83307576e-01 2.67421156e-01 6.71369970e-01
1.01793206e+00 -1.14450634e-01 -1.38052881e-01 -3.68100286e-01
2.22434342e-01 5.19629084e-02 -1.19355381e+00 1.72293079e+00
-5.49225569e-01 5.68251193e-01 5.77495098e-02 -6.54959857e-01
7.96538174e-01 4.21798468e-01 5.63115664e-02 -1.14964879e+00
8.76868609e-03 2.88459122e-01 1.52756020e-01 -5.06581306e-01
3.43677431e-01 -1.76745370e-01 1.95842236e-01 4.70341623e-01
4.17384833e-01 -4.49022621e-01 5.43212853e-02 -1.42528890e-02
8.43785942e-01 -3.07352066e-01 1.36479050e-01 1.43792793e-01
5.81108987e-01 -4.30050969e-01 4.12266523e-01 8.34453881e-01
-8.93710971e-01 1.08101833e+00 1.65977404e-01 -3.44968468e-01
-1.49412632e+00 -1.57206380e+00 -1.21776201e-01 7.27725506e-01
3.29123110e-01 3.03252991e-02 -8.81193042e-01 -5.82921982e-01
-9.34768766e-02 3.01735908e-01 -4.74067718e-01 -1.82869062e-01
-1.81650400e-01 -7.34755099e-01 3.31090868e-01 2.39078537e-01
8.34610701e-01 -9.53957558e-01 -4.88082647e-01 -2.16894358e-01
-3.21702957e-01 -1.14038229e+00 -5.82789183e-01 -5.09809256e-01
-4.24431831e-01 -1.03247619e+00 -1.09692240e+00 -8.39504540e-01
6.81157470e-01 4.94086415e-01 1.36779797e+00 -4.79117781e-02
-2.21431121e-01 4.77081656e-01 -8.92324299e-02 -2.09550202e-01
-5.02724767e-01 -7.78178692e-01 2.50385314e-01 2.52650321e-01
1.05053134e-01 -3.83192390e-01 -1.18381727e+00 5.29193878e-01
-9.10767496e-01 -1.98055938e-01 6.89345837e-01 9.14984107e-01
4.91216004e-01 1.24747291e-01 7.05169320e-01 -8.34176466e-02
9.71054316e-01 -5.81530817e-02 -7.26378977e-01 4.53656673e-01
-6.59248531e-01 -2.67904460e-01 5.02577245e-01 -2.91013390e-01
-9.97189641e-01 -2.01797158e-01 -4.55005001e-03 -3.51623386e-01
-2.27178931e-01 -4.49916497e-02 -6.08932614e-01 -5.14991283e-01
8.65108550e-01 2.32561335e-01 1.92893744e-02 -1.16262697e-01
5.38236737e-01 7.77241468e-01 1.02191305e+00 -3.81159097e-01
8.42633665e-01 3.28366607e-01 -2.64278203e-01 -3.00464988e-01
-5.01953006e-01 -4.06077534e-01 -4.53291535e-01 -4.08975571e-01
7.72122145e-01 -1.17945063e+00 -6.89212620e-01 9.47578967e-01
-1.00237930e+00 -2.81186402e-01 -2.27586463e-01 4.90623027e-01
-8.07465076e-01 6.22148871e-01 -3.99091452e-01 -6.33823037e-01
-3.45172405e-01 -1.51969266e+00 1.02613437e+00 2.32026443e-01
2.22280964e-01 -5.91314018e-01 4.76232693e-02 7.85889328e-01
4.55042750e-01 -7.48090073e-02 5.87364614e-01 -1.47179663e-01
-5.94928205e-01 -2.91146964e-01 -6.61584020e-01 8.63998115e-01
1.56763569e-01 -2.86099255e-01 -1.18309414e+00 -5.92906952e-01
1.05111003e-01 -6.14458263e-01 4.61031646e-01 4.43905920e-01
1.03353071e+00 -2.48930767e-01 6.30002737e-01 7.41389394e-01
1.47102654e+00 1.28268063e-01 7.95632482e-01 3.45667332e-01
3.78294349e-01 4.82930869e-01 4.33410347e-01 1.83074474e-01
3.90662104e-01 7.10635662e-01 5.76652765e-01 -2.90229768e-01
-3.81313026e-01 -2.11939231e-01 4.11254436e-01 5.02183199e-01
1.08335741e-01 -3.48219991e-01 -7.31252968e-01 7.72202492e-01
-1.30120301e+00 -9.15000498e-01 3.81799966e-01 2.54521513e+00
8.68442774e-01 4.82725278e-02 3.65792722e-01 2.55740523e-01
6.92689896e-01 -1.18129238e-01 -6.41744673e-01 -3.67312819e-01
-3.61220598e-01 9.15379822e-02 3.38822544e-01 5.01848876e-01
-1.10690761e+00 4.81204867e-01 7.30379438e+00 2.79128134e-01
-8.58861268e-01 -6.54984638e-02 7.95352459e-01 -1.86683640e-01
-1.07218370e-01 -2.64179647e-01 -8.01104158e-02 4.14649963e-01
9.00744677e-01 -2.11700976e-01 8.49844694e-01 4.53103304e-01
2.29832098e-01 -9.67065617e-02 -1.14458764e+00 1.30649209e+00
3.51862222e-01 -9.19028461e-01 3.66251916e-01 -8.76290202e-02
1.03562677e+00 -1.06920496e-01 7.88375795e-01 -7.73070306e-02
4.96498942e-01 -1.31788731e+00 8.35616589e-01 5.15082896e-01
1.11296272e+00 -4.41708148e-01 8.19216907e-01 -5.20937257e-02
-6.03713512e-01 -3.07966769e-01 -4.50827152e-01 8.07581544e-02
-8.22229963e-03 3.54299963e-01 -6.79199696e-01 4.87556130e-01
9.87612307e-01 4.33649004e-01 -1.00520611e+00 1.57443726e+00
-8.21007118e-02 2.44855434e-01 1.13772027e-01 5.70187271e-01
-5.58629781e-02 -1.99578092e-01 4.73091304e-01 8.87434125e-01
5.02494097e-01 -1.09783478e-01 -3.90218049e-01 9.42883193e-01
-2.33042538e-01 -1.21530779e-01 -4.28860426e-01 2.00655967e-01
4.34637636e-01 8.93343985e-01 3.54586053e-03 -3.03181946e-01
-5.33537090e-01 1.26811576e+00 1.30939841e-01 5.34678698e-01
-4.97396022e-01 -4.43686247e-01 8.32289577e-01 -1.80643216e-01
2.15780437e-01 -2.03985665e-02 -3.65817130e-01 -1.21341825e+00
4.36913997e-01 -1.44630349e+00 2.28452548e-01 -1.36417317e+00
-1.53780997e+00 8.21578741e-01 -3.85828286e-01 -1.71226394e+00
-7.78981596e-02 -7.92747080e-01 -4.94451076e-01 1.29964268e+00
-1.74913061e+00 -1.05641711e+00 -5.73478639e-01 9.94016409e-01
2.47437909e-01 -2.95809031e-01 6.79130256e-01 4.56409812e-01
-2.77191728e-01 9.30203974e-01 2.47708019e-02 1.57921165e-01
1.18576062e+00 -1.52757454e+00 5.10928869e-01 1.33805549e+00
3.89662385e-02 4.88402009e-01 7.09375739e-01 -7.97474161e-02
-1.10242200e+00 -9.38450396e-01 5.03326774e-01 -5.69257617e-01
4.23591733e-01 -1.11136205e-01 -9.82464135e-01 2.31454521e-01
3.67992163e-01 1.90103605e-01 5.72353184e-01 -3.37853938e-01
-7.73029208e-01 -3.07901651e-01 -1.49008834e+00 3.75744402e-01
9.36066985e-01 -9.79624033e-01 -6.17018521e-01 3.11068356e-01
6.41124606e-01 -2.85897046e-01 -9.69353795e-01 2.81055421e-01
4.77422446e-01 -1.42908752e+00 1.26618063e+00 -4.34991509e-01
4.32923436e-01 -6.05597138e-01 -2.94104576e-01 -1.71613789e+00
-1.10479958e-01 -5.45089006e-01 1.39102146e-01 1.00365615e+00
2.58159995e-01 -4.26085263e-01 2.96279430e-01 5.44180036e-01
7.45544955e-02 -1.37269139e-01 -9.51071799e-01 -8.18970025e-01
1.82309121e-01 -2.82797873e-01 9.71002936e-01 5.79214036e-01
-2.79972523e-01 -1.85834914e-01 -4.51354235e-01 5.22653520e-01
9.67218757e-01 -5.96804246e-02 8.29851270e-01 -7.68585205e-01
-3.91576976e-01 -5.39462805e-01 -8.20904434e-01 -8.24421287e-01
-2.80871302e-01 -4.11128193e-01 2.81995654e-01 -1.52561975e+00
2.08594665e-01 -2.42065877e-01 -3.31798971e-01 -1.81212038e-01
-4.98362690e-01 6.35774374e-01 1.44321159e-01 1.83027893e-01
-6.67664587e-01 5.69419265e-01 1.53760576e+00 -3.58872086e-01
5.20253852e-02 -1.16522118e-01 -7.75591373e-01 2.84978956e-01
5.69110036e-01 1.23920895e-01 -5.11463284e-01 -8.29575360e-01
3.50420326e-01 6.83881491e-02 6.66472077e-01 -1.39854085e+00
1.77918062e-01 7.65982345e-02 6.13762736e-01 9.77558270e-02
2.70259708e-01 -8.28924239e-01 -4.10680808e-02 9.15776640e-02
-3.94389957e-01 2.17322677e-01 1.25317320e-01 5.35302341e-01
-5.27806818e-01 1.56249970e-01 1.13655579e+00 -1.94958504e-02
-5.37724912e-01 4.89348739e-01 1.06749587e-01 5.76435328e-02
7.51793683e-01 -2.14754283e-01 -4.40210432e-01 -6.61161721e-01
-5.67265213e-01 1.13734640e-01 7.68284023e-01 3.60667378e-01
9.26845551e-01 -1.47177660e+00 -9.19556975e-01 4.76355374e-01
5.44284403e-01 -2.43901923e-01 4.45286334e-01 3.39575976e-01
-5.69676757e-01 1.06573112e-01 -6.98961079e-01 -5.00800788e-01
-1.05674946e+00 7.80812204e-01 8.98302853e-01 2.29516048e-02
-1.78041607e-01 8.89035285e-01 2.25615844e-01 -4.80336964e-01
2.47950807e-01 -1.84480444e-01 -1.06437042e-01 -3.88457030e-01
8.17328691e-01 1.81135386e-01 4.24167991e-01 -8.98887217e-01
4.38344516e-02 5.21234751e-01 1.01784021e-01 -4.07976270e-01
9.85648870e-01 -4.48716462e-01 -6.22849539e-03 7.43886409e-03
1.12596226e+00 -9.44672376e-02 -1.49398077e+00 -3.23272258e-01
-4.74743366e-01 -1.03954339e+00 1.03985101e-01 -1.38160741e+00
-1.01910257e+00 9.77388501e-01 1.22338104e+00 2.80228443e-02
1.61739349e+00 -3.61276865e-01 5.73104739e-01 2.34076437e-02
3.09157521e-01 -8.35356236e-01 4.08055454e-01 -1.09451368e-01
1.22693169e+00 -1.70072985e+00 -2.21056715e-01 1.42244995e-01
-7.59293854e-01 7.16021776e-01 6.10609591e-01 -3.01459935e-02
3.81241590e-01 -2.52426177e-01 5.21286368e-01 3.59212048e-03
-3.28062743e-01 -6.98964372e-02 5.61127663e-01 1.14157104e+00
7.59205967e-02 2.34160960e-01 2.73694366e-01 1.78572908e-01
-4.74157274e-01 -9.22306031e-02 5.72960019e-01 4.50310081e-01
-1.81027144e-01 -8.42721879e-01 -6.99337304e-01 4.13512975e-01
-3.73122603e-01 -8.62745047e-02 -2.38142908e-01 2.10458323e-01
1.81975618e-01 1.37978733e+00 6.38257861e-02 -4.11078662e-01
6.37428164e-01 -2.74729878e-01 6.84922278e-01 -1.28626302e-01
-5.46273172e-01 -3.14847440e-01 -2.45035484e-01 -7.75222301e-01
-4.92385209e-01 -5.50406516e-01 -3.93618643e-01 -3.14316511e-01
4.83888648e-02 -1.90624937e-01 5.63919723e-01 6.49588227e-01
2.78442115e-01 3.10484111e-01 8.94734919e-01 -6.56907201e-01
-6.53106809e-01 -8.69043708e-01 -5.81599057e-01 8.23070049e-01
8.69936585e-01 -4.67143625e-01 -5.26473224e-01 1.86324552e-01]
|
[11.899995803833008, -1.7972522974014282]
|
8d1e2f5d-f8c3-48d4-872e-36f226840ab2
|
soft-prompt-decoding-for-multilingual-dense
|
2305.09025
| null |
https://arxiv.org/abs/2305.09025v1
|
https://arxiv.org/pdf/2305.09025v1.pdf
|
Soft Prompt Decoding for Multilingual Dense Retrieval
|
In this work, we explore a Multilingual Information Retrieval (MLIR) task, where the collection includes documents in multiple languages. We demonstrate that applying state-of-the-art approaches developed for cross-lingual information retrieval to MLIR tasks leads to sub-optimal performance. This is due to the heterogeneous and imbalanced nature of multilingual collections -- some languages are better represented in the collection and some benefit from large-scale training data. To address this issue, we present KD-SPD, a novel soft prompt decoding approach for MLIR that implicitly "translates" the representation of documents in different languages into the same embedding space. To address the challenges of data scarcity and imbalance, we introduce a knowledge distillation strategy. The teacher model is trained on rich English retrieval data, and by leveraging bi-text data, our distillation framework transfers its retrieval knowledge to the multilingual document encoder. Therefore, our approach does not require any multilingual retrieval training data. Extensive experiments on three MLIR datasets with a total of 15 languages demonstrate that KD-SPD significantly outperforms competitive baselines in all cases. We conduct extensive analyses to show that our method has less language bias and better zero-shot transfer ability towards new languages.
|
['James Allan', 'Hamed Zamani', 'Hansi Zeng', 'Zhiqi Huang']
|
2023-05-15
| null | null | null | null |
['cross-lingual-information-retrieval']
|
['natural-language-processing']
|
[-2.55108505e-01 -2.57878274e-01 -7.26833999e-01 -1.55649453e-01
-1.69617689e+00 -6.87160313e-01 7.30223656e-01 1.51981071e-01
-8.19654822e-01 7.36496925e-01 5.43417931e-01 -3.29422265e-01
-1.08686283e-01 -4.65383053e-01 -7.79665530e-01 -3.38931710e-01
2.85519511e-01 6.92776740e-01 -9.42523852e-02 -5.62042594e-01
-7.59279728e-02 6.77158684e-02 -1.10946178e+00 4.96508628e-01
1.00774693e+00 5.63990951e-01 3.44931573e-01 2.55597442e-01
-2.25552082e-01 6.42633021e-01 -4.51198786e-01 -6.95106328e-01
3.18502843e-01 -2.25661579e-03 -8.31485093e-01 -3.47748995e-01
5.27483821e-01 -4.07979488e-01 -4.92013931e-01 8.75442624e-01
7.98033118e-01 -7.28357062e-02 8.29206288e-01 -7.61582136e-01
-1.29634774e+00 1.02605498e+00 -4.49103296e-01 8.25673863e-02
2.80378908e-01 -4.67910141e-01 1.45551014e+00 -1.32587099e+00
8.95297468e-01 1.41640055e+00 4.62398291e-01 3.76620919e-01
-1.06987476e+00 -7.53162265e-01 1.26161188e-01 1.76921889e-01
-1.80217636e+00 -6.01609170e-01 5.46740532e-01 -2.92217761e-01
1.22892964e+00 4.11559157e-02 8.81501213e-02 1.00630653e+00
3.88519727e-02 1.18976748e+00 7.78105855e-01 -8.66060376e-01
-3.17621589e-01 6.98087752e-01 2.64852673e-01 3.26612324e-01
2.64889151e-01 -1.30909458e-01 -4.95669752e-01 6.18370883e-02
1.81496099e-01 -2.58782476e-01 -3.19592059e-01 -1.69752508e-01
-1.17716694e+00 9.11281049e-01 3.97959381e-01 4.51623529e-01
-2.03123704e-01 -2.21328959e-01 6.52493000e-01 6.42645657e-01
8.97332966e-01 4.75808233e-01 -6.39076173e-01 3.52574944e-01
-8.05874407e-01 8.79002586e-02 7.69351840e-01 1.18209147e+00
8.23134601e-01 -3.46353173e-01 -1.93006590e-01 1.31547081e+00
4.19888914e-01 9.61878896e-01 6.64205730e-01 -4.80564833e-01
8.03298175e-01 4.20236886e-01 -2.08623737e-01 -6.18200839e-01
9.23059415e-03 -5.09566188e-01 -5.78753054e-01 -6.53660238e-01
-1.10802419e-01 4.03824821e-02 -6.35511458e-01 1.65598977e+00
1.67822931e-02 -5.75481892e-01 6.23835266e-01 7.80018270e-01
8.79626691e-01 8.77410233e-01 1.31715974e-02 -5.31913564e-02
1.40628338e+00 -1.13739038e+00 -8.78860176e-01 -4.34326053e-01
9.89791811e-01 -9.23138082e-01 1.13443244e+00 5.37966844e-03
-9.02841449e-01 -3.18428487e-01 -1.07735348e+00 -6.40682220e-01
-7.89042175e-01 5.14345646e-01 3.22531044e-01 3.47973108e-01
-1.15988326e+00 -8.91503096e-02 -3.92401665e-01 -3.46778959e-01
-1.03438772e-01 -2.18185043e-04 -5.05423367e-01 -6.70734048e-01
-1.64322925e+00 1.21936119e+00 6.65603638e-01 -4.10979986e-02
-5.49593210e-01 -7.46161461e-01 -1.08286905e+00 -8.55255723e-02
2.15531483e-01 -4.21661496e-01 1.07106757e+00 -7.56630421e-01
-1.08784556e+00 9.45790052e-01 -2.59972990e-01 -8.55137259e-02
2.15260267e-01 -3.98363173e-01 -4.23538834e-01 -7.44796693e-02
2.95967907e-01 6.27585411e-01 3.38618875e-01 -1.32969260e+00
-4.40619916e-01 -3.69929194e-01 -1.72580838e-01 6.37056351e-01
-6.87327087e-01 4.57082868e-01 -1.09886992e+00 -7.35061824e-01
-2.00121179e-01 -9.24546003e-01 2.36879274e-01 -4.86092985e-01
-3.16488534e-01 -4.37839329e-01 3.80022377e-01 -1.08880556e+00
1.52754688e+00 -2.07564592e+00 3.55394691e-01 6.33898601e-02
-2.09805831e-01 2.59253561e-01 -5.14318168e-01 7.29000390e-01
2.50938565e-01 1.73968554e-01 1.46991417e-01 -5.23610115e-01
3.29126775e-01 2.80387819e-01 -6.63886428e-01 1.59285396e-01
1.87408835e-01 1.07253218e+00 -8.90273809e-01 -6.20181501e-01
-1.90789923e-01 8.14700127e-01 -4.76712525e-01 1.46323979e-01
2.19270233e-02 1.73802584e-01 -3.26842666e-01 8.06211710e-01
4.94914979e-01 -1.20058730e-01 4.04473901e-01 -3.28289777e-01
3.21573839e-02 7.09002733e-01 -6.77448750e-01 2.08741546e+00
-7.92844892e-01 6.29090250e-01 -1.94501281e-01 -7.64009356e-01
7.11619377e-01 4.48254079e-01 2.48654291e-01 -1.34092915e+00
-7.15447962e-02 6.94219768e-01 -2.80235231e-01 -3.45997036e-01
1.03195322e+00 -1.78333484e-02 -5.34788787e-01 6.11170352e-01
3.45803857e-01 1.54444845e-02 3.58546495e-01 3.66631925e-01
6.88892126e-01 -6.40603378e-02 1.55518606e-01 -3.08242947e-01
4.81626540e-01 3.61941047e-02 3.91332984e-01 7.99588442e-01
1.66455537e-01 1.12419352e-01 -1.24911696e-01 3.37723754e-02
-1.15341949e+00 -1.05862689e+00 -4.55364406e-01 1.63506365e+00
4.19502780e-02 -3.72880667e-01 -4.23050553e-01 -5.90610087e-01
9.69580710e-02 7.17468202e-01 -2.72407323e-01 -2.55530208e-01
-6.42321944e-01 -8.46664071e-01 6.55121446e-01 1.34744748e-01
1.23307109e-01 -9.05279756e-01 2.64800727e-01 1.82594284e-01
-5.01953423e-01 -1.31415808e+00 -6.21075511e-01 1.29348576e-01
-4.54473138e-01 -6.10366046e-01 -9.73122358e-01 -1.16105282e+00
5.60762584e-01 4.26502794e-01 1.47280538e+00 -2.15581402e-01
-1.47074535e-01 5.62905908e-01 -5.75992048e-01 -3.52550834e-01
-5.53522348e-01 6.97244644e-01 1.06363036e-01 -4.05619740e-01
9.01929677e-01 1.74413219e-01 -2.00187162e-01 -4.68924232e-02
-1.08609927e+00 -2.11248115e-01 6.97701931e-01 8.23053539e-01
6.39078557e-01 -1.25072747e-01 7.52275527e-01 -7.33557463e-01
9.91936147e-01 -6.77157164e-01 -4.91537213e-01 9.78401065e-01
-6.70109272e-01 4.75390822e-01 3.90984565e-01 -3.82644534e-01
-1.00849295e+00 -5.85853815e-01 4.98487391e-02 -1.77725837e-01
5.40664315e-01 9.99643981e-01 -3.10204886e-02 1.64836735e-01
4.70567256e-01 3.25680733e-01 -3.09819281e-01 -7.62792289e-01
6.73029125e-01 1.29264092e+00 1.40965089e-01 -8.14185798e-01
4.71521765e-01 2.14174818e-02 -7.93565094e-01 -6.03743792e-01
-1.10275757e+00 -6.11656368e-01 -6.48860455e-01 6.35320097e-02
6.80603385e-01 -1.61829937e+00 5.26884990e-03 2.87020773e-01
-1.17027009e+00 -2.10060358e-01 -8.19273740e-02 6.10293448e-01
5.92622627e-03 7.71975145e-02 -8.95713687e-01 -5.91520548e-01
-6.85580432e-01 -1.32061064e+00 1.44062448e+00 -3.00056130e-01
1.00081787e-01 -1.23660946e+00 4.51468319e-01 3.93401712e-01
4.76348013e-01 -6.49900079e-01 1.11188745e+00 -8.89309347e-01
-3.87977362e-01 -1.82158515e-01 -4.26858187e-01 5.55979848e-01
1.66906089e-01 -2.99095482e-01 -8.53750348e-01 -7.24149704e-01
-3.94065559e-01 -8.48371804e-01 9.45667386e-01 -1.77481085e-01
7.06137896e-01 -3.72579485e-01 -2.40698323e-01 3.87595057e-01
1.60923803e+00 -3.74391451e-02 3.29246491e-01 5.62936246e-01
6.52464092e-01 7.53793061e-01 5.31901479e-01 1.32140657e-02
1.08379602e+00 9.39704597e-01 -2.07617208e-01 -2.39465013e-01
-4.02904093e-01 -4.35456783e-01 6.91766441e-01 1.86434960e+00
4.76740718e-01 -2.94959068e-01 -9.76558506e-01 7.64041722e-01
-1.61125481e+00 -5.56648910e-01 4.27866250e-01 2.26068854e+00
1.57601798e+00 -4.43759412e-01 -3.94577980e-01 -4.66229320e-01
3.47467452e-01 -6.32468760e-02 -3.82199645e-01 -2.71888018e-01
-4.70651627e-01 2.70387113e-01 6.70143485e-01 8.61860931e-01
-9.42037463e-01 1.29454756e+00 6.52812433e+00 9.72002804e-01
-1.12169492e+00 3.69183391e-01 2.23920539e-01 -9.09573808e-02
-6.85093820e-01 -2.34716475e-01 -1.25969470e+00 3.19427490e-01
8.54984105e-01 -4.48446631e-01 5.15760958e-01 5.21549106e-01
-3.74461681e-01 3.33140343e-01 -1.15190315e+00 8.41016948e-01
5.57667255e-01 -9.00153399e-01 3.96689594e-01 -9.45721343e-02
8.23759317e-01 7.11947918e-01 1.26666352e-01 9.25733924e-01
6.07231200e-01 -8.56169105e-01 5.84944189e-01 2.67619252e-01
1.09371626e+00 -6.33923054e-01 7.77918756e-01 3.41478735e-01
-9.94728923e-01 1.20745689e-01 -6.26980603e-01 5.57765067e-01
9.33026001e-02 4.99200135e-01 -8.90141726e-01 8.38024378e-01
6.54407024e-01 9.17363822e-01 -6.93711102e-01 4.50234324e-01
-2.13696599e-01 1.73005611e-01 -1.75974667e-01 2.19314247e-01
1.38476521e-01 -1.36575297e-01 2.31850550e-01 1.66943359e+00
3.94758731e-01 -5.22451162e-01 2.32592434e-01 4.64943200e-01
-5.12445569e-01 5.90267181e-01 -8.85776818e-01 -1.24149188e-01
6.31561399e-01 9.04806495e-01 3.01741958e-01 -5.11304796e-01
-6.47818565e-01 1.01648378e+00 8.32490683e-01 6.22008443e-01
-3.67254376e-01 -4.82112110e-01 5.27620494e-01 -3.14671189e-01
8.68798122e-02 -1.55741990e-01 3.01769704e-01 -1.74520755e+00
2.72001296e-01 -1.10577703e+00 5.62687457e-01 -6.71604931e-01
-1.58110785e+00 7.18292117e-01 1.67864058e-02 -1.06363201e+00
-5.63382566e-01 -5.34026146e-01 4.35187042e-01 1.16322005e+00
-2.37395191e+00 -1.32042062e+00 3.47842932e-01 6.23363733e-01
4.98004526e-01 -5.23813367e-01 9.74878132e-01 9.39722002e-01
-4.27736193e-01 1.05384481e+00 7.22323596e-01 3.92325372e-01
1.29656637e+00 -9.39529181e-01 -1.92917662e-03 7.34688759e-01
2.62943745e-01 8.16218913e-01 1.53115660e-01 -5.38801968e-01
-1.76022732e+00 -1.17660844e+00 1.61395264e+00 -6.00419939e-01
9.51776803e-01 -5.05918860e-01 -1.07763767e+00 9.69343662e-01
6.08872294e-01 -2.64393091e-01 1.02396607e+00 3.93467963e-01
-8.61424148e-01 -2.37954929e-01 -7.25547373e-01 6.14810884e-01
5.39671779e-01 -1.14901888e+00 -8.26117873e-01 5.00884950e-01
1.03692818e+00 -1.94876745e-01 -1.21304321e+00 4.62687165e-01
5.07949829e-01 -1.26182929e-01 1.16702521e+00 -6.55036092e-01
3.73918951e-01 -8.08011815e-02 -5.31273663e-01 -1.54538631e+00
-1.50444895e-01 4.55810279e-02 -2.96184402e-02 1.27691734e+00
5.68582654e-01 -6.28239810e-01 -2.01038852e-01 2.89938986e-01
1.07114585e-02 -3.47985059e-01 -9.57187951e-01 -8.39174747e-01
7.59355485e-01 -3.63346696e-01 3.70610654e-01 1.24482989e+00
2.46576622e-01 7.35866725e-01 -4.04615372e-01 3.98511849e-02
5.55526793e-01 -4.80418094e-02 6.04661226e-01 -8.13361049e-01
-2.25818112e-01 -3.08838248e-01 1.09349921e-01 -1.16992879e+00
7.18546510e-01 -1.64111435e+00 1.66104138e-01 -1.53909361e+00
5.49918890e-01 -7.32870936e-01 -6.29007936e-01 6.74684882e-01
-3.96112531e-01 2.50382125e-01 1.34914011e-01 5.10291934e-01
-8.45324993e-01 6.65963829e-01 1.02086556e+00 -4.52501982e-01
-7.45495409e-02 -7.60231078e-01 -8.78065526e-01 4.14520130e-02
3.89059722e-01 -5.35230935e-01 -3.34137261e-01 -1.28479505e+00
5.72258353e-01 -1.23374633e-01 -2.48404339e-01 -3.31566036e-01
2.65310258e-01 2.97970742e-01 -1.77992415e-02 -3.69094580e-01
2.27676064e-01 -8.37506771e-01 -1.91772684e-01 1.46605685e-01
-6.61503613e-01 1.18487604e-01 3.55592430e-01 1.63543612e-01
-6.12369835e-01 -2.10946247e-01 2.70603091e-01 9.02082399e-02
-6.08896613e-01 1.78515956e-01 -2.45962113e-01 3.40647876e-01
4.21485990e-01 5.96833110e-01 -6.11264825e-01 -1.90226957e-01
-2.15299472e-01 4.71972436e-01 3.94107133e-01 9.44334388e-01
2.09816188e-01 -1.65418351e+00 -1.26584697e+00 1.97784126e-01
7.99067020e-01 -1.77234098e-01 -1.32962465e-01 6.04651630e-01
-2.03264475e-01 9.99027193e-01 2.98010975e-01 -4.69693094e-01
-9.67113018e-01 5.01553953e-01 1.14355698e-01 -7.35679388e-01
-3.04473817e-01 5.80896914e-01 2.23699152e-01 -1.20618153e+00
2.86912471e-01 7.26274997e-02 -1.94087669e-01 3.66095930e-01
5.63165843e-01 -5.80924265e-02 4.34673250e-01 -7.97626495e-01
-3.66994470e-01 6.14485741e-01 -6.32031500e-01 -3.61240178e-01
1.20025921e+00 -4.14172739e-01 -4.17019486e-01 5.63159466e-01
1.70577037e+00 2.13322431e-01 -4.62386489e-01 -1.01400268e+00
2.35204458e-01 -2.72057891e-01 3.23399544e-01 -9.66650248e-01
-8.90703917e-01 7.77154148e-01 3.77951980e-01 -3.30229580e-01
1.07778072e+00 2.13834345e-01 7.34494269e-01 7.09836960e-01
5.20645559e-01 -1.32879210e+00 -2.50771642e-01 1.02061760e+00
8.53466988e-01 -1.62003541e+00 -8.51181373e-02 9.40556750e-02
-6.45106316e-01 7.98928022e-01 2.61312187e-01 2.42392763e-01
6.83211684e-01 9.95091200e-02 4.40814465e-01 -5.17007820e-02
-9.05926764e-01 -2.43665680e-01 6.83098733e-01 1.90826356e-01
8.47232997e-01 2.83709224e-02 -4.24493730e-01 4.65553015e-01
-2.20420659e-01 -6.13727309e-02 -2.58663995e-03 9.31804299e-01
-1.40024900e-01 -1.43172801e+00 -1.31589130e-01 1.14434175e-01
-6.39114201e-01 -7.06883967e-01 -2.43089095e-01 7.12313116e-01
-2.93387532e-01 7.60977507e-01 1.43621698e-01 -1.46416262e-01
1.29139423e-01 2.65319496e-01 4.17089611e-01 -6.45763040e-01
-5.38679540e-01 1.76256686e-01 1.11054130e-01 -2.23470598e-01
-3.84513378e-01 -4.39840674e-01 -7.19797313e-01 -4.54364456e-02
-3.80770475e-01 3.11332971e-01 1.02637553e+00 8.39049339e-01
5.55096030e-01 4.58045930e-01 5.89075387e-01 -3.96810234e-01
-6.19792223e-01 -1.08557773e+00 -3.84379476e-01 2.73802251e-01
2.72233009e-01 -2.82692641e-01 -3.68958563e-01 -1.27160683e-01]
|
[11.369599342346191, 9.803342819213867]
|
8c129263-b0df-46dd-8eb2-2a4e1fe57d9f
|
generalized-difference-in-differences-models
|
2211.06710
| null |
https://arxiv.org/abs/2211.06710v4
|
https://arxiv.org/pdf/2211.06710v4.pdf
|
Generalized Difference-in-differences Models: Robust Bounds
|
The difference-in-differences (DID) method identifies the average treatment effects on the treated (ATT) under mainly the so-called parallel trends (PT) assumption. The most common and widely used approach to justify the PT assumption is the pre-treatment period examination. If a null hypothesis of the same trend in the outcome means for both treatment and control groups in the pre-treatment periods is rejected, researchers believe less in PT and the DID results. This paper develops a robust generalized DID method that utilizes all the information available not only from the pre-treatment periods but also from multiple data sources. Our approach interprets PT in a different way using a notion of selection bias, which enables us to generalize the standard DID estimand by defining an information set that may contain multiple pre-treatment periods or other baseline covariates. Our main assumption states that the selection bias in the post-treatment period lies within the convex hull of all selection biases in the pre-treatment periods. We provide a sufficient condition for this assumption to hold. Based on the baseline information set we construct, we first provide an identified set for the ATT that always contains the true ATT under our identifying assumption, and also the standard DID estimand. Secondly, we propose a class of criteria on the selection biases from the perspective of policymakers that can achieve a point identification of the ATT. Finally, we illustrate our methodology through some numerical and empirical examples.
|
['Désiré Kédagni', 'Kyunghoon Ban']
|
2022-11-12
| null | null | null | null |
['selection-bias']
|
['natural-language-processing']
|
[ 1.71215832e-01 2.10721418e-01 -1.02354741e+00 -2.32311368e-01
-5.71417689e-01 -5.96197844e-01 6.39407814e-01 3.44544470e-01
-4.14596349e-01 7.71847248e-01 6.27373278e-01 -6.59871578e-01
-5.77168584e-01 -7.70690322e-01 -6.65817440e-01 -7.81776607e-01
2.21793026e-01 2.65430748e-01 -1.73234805e-01 1.43002570e-01
3.60291451e-01 4.12469327e-01 -1.32355738e+00 6.31406009e-02
1.31102931e+00 5.44183493e-01 -1.90581322e-01 -2.89291386e-02
4.40710969e-02 3.75747293e-01 -3.81612688e-01 -3.94393325e-01
5.19737065e-01 -4.19329256e-01 -6.64719760e-01 1.35082766e-01
4.49249685e-01 -3.46967727e-01 1.50812771e-02 1.13069940e+00
3.89003694e-01 -1.66383421e-03 1.05130720e+00 -1.51518786e+00
-5.49258709e-01 9.81007338e-01 -6.73314691e-01 2.22650971e-02
2.73359120e-01 1.28320232e-01 7.73684680e-01 -5.79800546e-01
4.83272940e-01 1.36252844e+00 6.39665663e-01 2.61890709e-01
-1.38481510e+00 -6.71629846e-01 2.68971384e-01 -1.53693944e-01
-1.15989542e+00 -2.74460107e-01 4.80983168e-01 -8.75549078e-01
2.49893609e-02 4.36029494e-01 6.56890333e-01 9.90015507e-01
5.34694672e-01 4.03959572e-01 1.52474809e+00 -7.72679090e-01
3.14907372e-01 3.59634280e-01 7.75238812e-01 -9.74449441e-02
7.78755009e-01 5.44459403e-01 2.68097948e-02 -5.80123544e-01
7.14017093e-01 2.10576296e-01 -4.09087270e-01 -2.58961499e-01
-9.02846396e-01 1.16457427e+00 1.63554192e-01 4.27282214e-01
-7.88951218e-01 -1.97357714e-01 3.87812287e-01 2.95380592e-01
5.87929428e-01 2.26473778e-01 -3.90728980e-01 5.01343548e-01
-7.45323062e-01 2.17427850e-01 5.82488477e-01 4.58751261e-01
5.22490084e-01 -1.68621674e-01 -4.49127614e-01 2.79330999e-01
1.54164717e-01 7.56897986e-01 5.44578016e-01 -7.12018073e-01
3.51276696e-01 6.36442363e-01 4.30221617e-01 -7.62220383e-01
-2.68497825e-01 -4.71069396e-01 -8.19911897e-01 1.38277471e-01
7.66796470e-01 -4.31224614e-01 -5.96237540e-01 1.97361672e+00
4.98040497e-01 -9.85019058e-02 -2.84762606e-02 4.54028279e-01
2.85943508e-01 3.35083902e-01 2.56305099e-01 -1.00395966e+00
1.34637499e+00 -1.93707928e-01 -9.96858001e-01 1.18270911e-01
8.32493246e-01 -4.19678658e-01 9.36906576e-01 1.28262728e-01
-9.66785967e-01 -2.04135582e-01 -3.66390318e-01 4.21680719e-01
-1.31954262e-02 4.29468267e-02 4.76522565e-01 8.76943052e-01
-8.33417177e-01 4.37703252e-01 -4.02058929e-01 -5.14130831e-01
8.37628171e-02 3.31073493e-01 -1.11947566e-01 2.22298563e-01
-9.37190771e-01 7.22147644e-01 6.56564608e-02 -1.91351101e-01
-6.95348740e-01 -1.04608750e+00 -5.51063716e-01 3.16219926e-01
6.21846974e-01 -9.87454474e-01 1.08858001e+00 -1.43791080e+00
-1.02166903e+00 6.11153364e-01 -2.60660321e-01 -2.61215687e-01
8.38999450e-01 3.41554359e-02 -1.95296615e-01 -2.37093851e-01
3.68762553e-01 -1.24425843e-01 5.09627819e-01 -1.28005075e+00
-6.91109002e-01 -7.89473474e-01 -1.42075896e-01 -1.36167873e-02
-3.25402468e-01 4.83861119e-01 3.48690957e-01 -6.02149725e-01
2.15788260e-02 -8.08705032e-01 -3.53227884e-01 -5.45359015e-01
-6.07100666e-01 -5.63750029e-01 2.71677732e-01 -3.10512185e-01
1.49006593e+00 -2.20553303e+00 -2.58690774e-01 5.09872794e-01
9.46829766e-02 -2.28744715e-01 2.04276890e-01 6.50442541e-01
-7.46814728e-01 4.55484182e-01 -2.22492144e-02 -8.31453353e-02
5.00378795e-02 -1.02750540e-01 -6.55755162e-01 9.79035795e-01
-3.02779406e-01 3.60712647e-01 -5.23247063e-01 -2.42258117e-01
6.11496046e-02 -1.03595387e-03 -5.13170719e-01 4.35807407e-02
3.42418134e-01 5.09342790e-01 -4.49982643e-01 -1.13953784e-01
8.05304170e-01 1.55407354e-01 3.13277602e-01 7.63872936e-02
-8.06029499e-01 3.66971940e-01 -1.41101158e+00 6.02942348e-01
-1.18713632e-01 2.40149945e-01 -1.85122550e-01 -1.29669273e+00
7.15604126e-01 6.69073284e-01 7.14101195e-01 -2.99575597e-01
3.09864670e-01 3.78183126e-01 2.17476875e-01 -4.14105207e-01
7.34678051e-03 -5.11983156e-01 -1.03438765e-01 4.69652236e-01
-2.88647950e-01 2.84351408e-01 5.68702035e-02 -1.34352192e-01
6.43514633e-01 -6.03276968e-01 7.81767368e-01 -1.02909636e+00
4.64123964e-01 -7.00521171e-02 9.83582854e-01 8.66560042e-01
-5.52139292e-03 2.05907613e-01 8.24154496e-01 -2.19568998e-01
-9.07000840e-01 -8.40429783e-01 -5.24052143e-01 5.51231503e-01
-2.91683197e-01 2.20756501e-01 -5.72204351e-01 -7.25346804e-01
2.73410469e-01 1.01748562e+00 -1.08391154e+00 -9.18057859e-02
-2.30516180e-01 -8.07065427e-01 1.03758134e-01 2.25159064e-01
4.44283247e-01 -4.57634419e-01 -4.29481208e-01 -1.12479545e-01
1.50943756e-01 -3.90573978e-01 -4.57430810e-01 -1.80370077e-01
-1.02063477e+00 -1.24119282e+00 -7.72733748e-01 -5.06421566e-01
7.34550178e-01 1.47939339e-01 6.33030534e-01 -1.94434583e-01
5.41495979e-01 2.95457363e-01 -9.95297432e-02 -9.70538855e-01
-4.76260424e-01 -3.57194692e-01 3.26540649e-01 1.50657445e-01
4.52317297e-01 -3.58843565e-01 -6.23979509e-01 2.68435150e-01
-7.50204504e-01 -2.22663000e-01 2.69612312e-01 5.65825522e-01
2.30027720e-01 3.02418798e-01 9.13331389e-01 -8.23037207e-01
4.11001295e-01 -6.34917676e-01 -8.21794093e-01 3.10815752e-01
-9.28312361e-01 -1.62506297e-01 3.77129346e-01 -7.02811897e-01
-1.19923162e+00 -4.49847400e-01 4.06905264e-01 -7.68996477e-02
-3.40678900e-01 7.47802854e-01 -3.88274461e-01 4.33911771e-01
5.96979976e-01 -3.40724617e-01 3.77953798e-02 -5.51504850e-01
-1.60284694e-02 7.00527787e-01 1.55765995e-01 -5.06046653e-01
6.32903814e-01 5.72374642e-01 6.02355301e-02 -4.31564212e-01
-7.73097754e-01 -3.32828224e-01 -3.99375975e-01 -2.54545826e-02
7.58159757e-01 -7.11948216e-01 -1.07687163e+00 3.00979078e-01
-8.71458054e-01 -3.41098964e-01 -6.04955852e-01 1.12390482e+00
-4.59722996e-01 1.71786264e-01 -2.15266258e-01 -1.32647192e+00
-1.61423162e-01 -1.20857513e+00 4.84714001e-01 1.68560877e-01
-2.30839834e-01 -1.39342058e+00 1.85209498e-01 -8.81936848e-02
1.44343423e-02 4.55190867e-01 1.11622107e+00 -9.49119568e-01
-9.11688507e-02 -2.29038298e-01 2.46467553e-02 1.07544534e-01
3.98954004e-01 1.75026178e-01 -5.48805952e-01 -3.21368128e-01
5.36624014e-01 5.32332659e-01 5.59533060e-01 1.45221400e+00
8.86468112e-01 -7.33329356e-01 -5.68617463e-01 1.90406427e-01
1.69343257e+00 6.28181398e-01 4.98892486e-01 5.03287971e-01
3.63699228e-01 1.04240680e+00 5.67707181e-01 4.34098393e-01
2.93873608e-01 7.01205850e-01 1.25529930e-01 -3.17918062e-01
4.36155021e-01 -1.17270701e-01 4.20501858e-01 2.31553197e-01
7.17217615e-03 -1.70598641e-01 -8.81655633e-01 8.27215731e-01
-1.88187873e+00 -1.13708007e+00 -7.79492199e-01 3.08928728e+00
8.57896984e-01 -1.34945497e-01 6.61976278e-01 2.14067504e-01
1.04866219e+00 -4.92795080e-01 -1.22747114e-02 -7.45362222e-01
-6.33928850e-02 -2.29552668e-03 8.13529789e-01 6.03827238e-01
-8.03685844e-01 7.46288821e-02 7.11367464e+00 6.01378620e-01
-9.03508127e-01 -8.17527156e-03 8.29611182e-01 2.90770531e-01
-5.71673274e-01 3.57304841e-01 -9.13885415e-01 4.96928602e-01
1.01641631e+00 -9.03473437e-01 -3.62466156e-01 5.82678080e-01
9.76778269e-01 -4.56113487e-01 -1.34216535e+00 2.30393574e-01
-3.10257226e-01 -7.58050919e-01 -1.72970340e-01 7.37975538e-01
1.10475051e+00 -7.22895443e-01 3.79288912e-01 4.77741845e-02
5.89261174e-01 -7.64758170e-01 6.79157019e-01 2.31070429e-01
8.03696990e-01 -6.50527179e-01 9.89608169e-01 5.11790991e-01
-7.24241972e-01 -4.88981962e-01 -1.47105098e-01 -3.47836018e-01
-8.80350992e-02 1.11064601e+00 -6.41395032e-01 1.00058484e+00
2.93834299e-01 5.61009109e-01 -4.54197871e-03 1.35306203e+00
-1.94360137e-01 8.90722454e-01 -1.26802891e-01 4.67558891e-01
9.61671993e-02 -4.22315836e-01 5.25874078e-01 7.43541121e-01
6.70159698e-01 1.61571786e-01 6.45075887e-02 7.40932465e-01
2.45739698e-01 4.45510149e-01 -8.47475529e-01 4.81813997e-01
4.32736307e-01 5.05937457e-01 -3.87921780e-01 -5.32540083e-01
-5.26007593e-01 2.89323274e-02 -3.79501194e-01 5.28045535e-01
-4.74354208e-01 1.13246180e-01 4.70383704e-01 2.66385853e-01
-2.07261518e-01 3.49016368e-01 -7.27079928e-01 -1.02406418e+00
-2.24439621e-01 -9.63792443e-01 8.53736401e-01 -2.42470086e-01
-1.33388841e+00 -2.48873815e-01 8.29937518e-01 -1.08378923e+00
-1.23116635e-01 -3.54595065e-01 -7.82221854e-01 1.18860149e+00
-1.03610075e+00 -7.07223475e-01 8.39665681e-02 5.24335802e-01
2.62766093e-01 2.74824023e-01 4.95559216e-01 1.05856128e-01
-9.94791985e-01 4.74026263e-01 2.69294620e-01 -1.00255661e-01
7.41221011e-01 -1.19867992e+00 -4.13696885e-01 8.38866532e-01
-7.18638182e-01 8.41989636e-01 1.00979173e+00 -8.96190464e-01
-7.54062176e-01 -7.90723324e-01 1.19487858e+00 -2.21774906e-01
6.82425201e-01 1.13235578e-01 -7.36841321e-01 1.16479182e+00
2.51487106e-01 -7.61105597e-01 6.59327388e-01 3.60067397e-01
-9.43083614e-02 -2.30067745e-01 -1.26628959e+00 6.93597436e-01
6.04173064e-01 -6.96680695e-02 -9.19119000e-01 3.79874289e-01
4.24397886e-01 6.87608272e-02 -9.90386188e-01 5.92831552e-01
4.85790581e-01 -8.66943240e-01 6.56947196e-01 -6.60591304e-01
3.43283027e-01 1.74108207e-01 2.90935040e-02 -1.37733686e+00
-3.45132202e-01 -5.32154262e-01 7.43342936e-01 1.53506482e+00
2.69055456e-01 -1.28822541e+00 3.30952376e-01 8.87068868e-01
4.93344478e-02 -3.77719373e-01 -9.43656623e-01 -1.05098879e+00
6.35978281e-01 -1.23438358e-01 8.05894911e-01 1.36685276e+00
6.95304349e-02 1.07981198e-01 -9.95519012e-02 1.61482483e-01
7.49695659e-01 6.88966364e-02 6.69731438e-01 -1.55580628e+00
-4.09203358e-02 -4.82043833e-01 -9.05678868e-02 -3.85712683e-01
2.77381152e-01 -5.10917783e-01 -2.55170852e-01 -1.53226984e+00
7.02653110e-01 -5.09804487e-01 -1.01661295e-01 4.35615718e-01
-3.32953483e-01 -5.08022189e-01 -5.99047691e-02 4.05772805e-01
5.35451114e-01 3.79629076e-01 1.05520511e+00 2.00649545e-01
-5.57789326e-01 4.27185476e-01 -8.39531958e-01 7.85407543e-01
8.02657843e-01 -6.93247020e-01 -3.32582533e-01 4.57954407e-02
-4.64386940e-02 2.07820952e-01 4.03246045e-01 -4.09461200e-01
-2.61096150e-01 -8.59216213e-01 2.96410336e-03 -4.25232857e-01
-4.06027764e-01 -1.08796370e+00 5.89652598e-01 8.42984617e-01
-5.44080257e-01 9.12120640e-02 1.21894673e-01 9.83935967e-02
1.24200091e-01 -4.74134207e-01 5.32411039e-01 1.80316687e-01
2.00150445e-01 4.44660038e-02 -4.62591887e-01 -1.90216392e-01
1.09307837e+00 -1.65614069e-01 -5.33948898e-01 -4.74303663e-01
-5.66816628e-01 2.17676029e-01 5.11775851e-01 -2.04179794e-01
-8.03009346e-02 -1.48114812e+00 -9.95745718e-01 -1.08730160e-01
-1.19369611e-01 -5.20245016e-01 2.26627231e-01 1.53551412e+00
2.64911860e-01 5.77857494e-01 -8.35712925e-02 -4.95552987e-01
-1.31894898e+00 1.01466489e+00 2.84087002e-01 -3.02034408e-01
-5.05187154e-01 1.02823801e-01 9.19988453e-01 1.15703031e-01
-6.02490269e-02 -3.38524818e-01 -4.20398831e-01 3.55994582e-01
5.47951519e-01 5.50515771e-01 -2.69283682e-01 -6.07433736e-01
-2.38010734e-01 5.42187452e-01 1.97096005e-01 -3.77402216e-01
1.22673059e+00 -3.23383689e-01 -2.97824025e-01 7.55612671e-01
1.12489140e+00 3.50968599e-01 -7.53310382e-01 -9.66480747e-02
-2.48152651e-02 -6.85741544e-01 -2.66618058e-02 -4.14195061e-01
-8.88775229e-01 4.22271609e-01 5.69447517e-01 5.89627862e-01
1.28777826e+00 -1.01977274e-01 -3.19327474e-01 -2.77074963e-01
1.08037971e-01 -8.28640878e-01 -4.45966393e-01 -1.01469919e-01
9.96775389e-01 -9.91368532e-01 1.20721407e-01 -7.15436876e-01
-4.91555423e-01 7.13791132e-01 1.44750342e-01 -1.45017773e-01
6.49184227e-01 -1.75969422e-01 -5.40864132e-02 -9.76805985e-02
-5.90027690e-01 -8.45651180e-02 4.56754893e-01 3.90668064e-01
5.76530159e-01 2.29317591e-01 -1.43550026e+00 5.18135309e-01
-1.29834741e-01 2.61889566e-02 8.39707017e-01 5.60711026e-01
-2.03252003e-01 -1.07071733e+00 -8.74561131e-01 4.40615386e-01
-8.81666839e-01 1.27424300e-01 -2.48602912e-01 1.16111720e+00
1.95570253e-02 1.18904555e+00 3.59667242e-01 8.81725475e-02
5.83304644e-01 1.21652238e-01 8.01954940e-02 -5.53360939e-01
-7.07068503e-01 4.87538427e-01 6.90380335e-02 -1.61396205e-01
-7.25703001e-01 -9.85667467e-01 -7.68722892e-01 -6.97389245e-01
-7.93515146e-01 4.84310865e-01 3.18241507e-01 9.88976419e-01
-3.48869376e-02 3.75723362e-01 1.19688106e+00 -3.40408385e-01
-8.74731719e-01 -9.81120229e-01 -7.33734012e-01 4.59020466e-01
3.37567508e-01 -7.79000401e-01 -9.29460287e-01 -2.43618935e-01]
|
[7.980116844177246, 5.189881801605225]
|
2f3f48d9-c7cb-46e6-a187-cd45b8e0d879
|
bapgan-gan-based-bone-age-progression-of
|
2110.08509
| null |
https://arxiv.org/abs/2110.08509v1
|
https://arxiv.org/pdf/2110.08509v1.pdf
|
BAPGAN: GAN-based Bone Age Progression of Femur and Phalange X-ray Images
|
Convolutional Neural Networks play a key role in bone age assessment for investigating endocrinology, genetic, and growth disorders under various modalities and body regions. However, no researcher has tackled bone age progression/regression despite its valuable potential applications: bone-related disease diagnosis, clinical knowledge acquisition, and museum education. Therefore, we propose Bone Age Progression Generative Adversarial Network (BAPGAN) to progress/regress both femur/phalange X-ray images while preserving identity and realism. We exhaustively confirm the BAPGAN's clinical potential via Frechet Inception Distance, Visual Turing Test by two expert orthopedists, and t-Distributed Stochastic Neighbor Embedding.
|
['Toshifumi Ozaki', 'Ryuichi Nakahara', 'Joe Hasei', 'Changhee Han', 'Shinji Nakazawa']
|
2021-10-16
| null | null | null | null |
['clinical-knowledge']
|
['miscellaneous']
|
[ 1.06459828e-02 4.18718606e-01 -9.74084660e-02 -3.12093068e-02
-6.96127534e-01 6.14907816e-02 2.19375789e-01 -8.72428194e-02
-3.82234931e-01 8.98854792e-01 2.89432049e-01 -3.81993800e-01
-2.13504463e-01 -8.83513391e-01 -6.25413060e-01 -6.20035529e-01
-3.59385848e-01 6.82277203e-01 -1.87356144e-01 -1.54138550e-01
-6.79455027e-02 3.26070130e-01 -8.91385436e-01 -3.10152829e-01
9.88938630e-01 6.41521990e-01 -4.77862060e-01 8.17743361e-01
3.28111947e-01 7.77751029e-01 -4.01712894e-01 -8.69129062e-01
-2.49487627e-02 -4.74680454e-01 -7.78000593e-01 -3.37606162e-01
2.12938786e-01 -5.68609655e-01 -5.70396602e-01 8.45764577e-01
1.06493986e+00 -2.94939935e-01 9.28779662e-01 -1.12076354e+00
-1.44050407e+00 6.96529627e-01 -7.53016770e-01 1.80459514e-01
1.53633639e-01 2.94884235e-01 4.32358593e-01 -5.52831411e-01
4.72357154e-01 1.11354601e+00 9.36079562e-01 1.14964867e+00
-1.05488276e+00 -7.50795841e-01 -6.74255490e-01 2.06505552e-01
-1.34693182e+00 1.86264262e-01 7.20442593e-01 -5.20810723e-01
8.38078409e-02 1.92174628e-01 1.19140971e+00 1.52672493e+00
8.86823058e-01 6.42356396e-01 1.13739014e+00 -1.68622017e-01
2.38746449e-01 -6.89277351e-01 -4.02099222e-01 9.38108385e-01
1.41782373e-01 4.35018450e-01 -2.90134400e-01 1.62437603e-01
1.39103723e+00 -2.09436074e-01 1.62941039e-01 -5.01633286e-02
-9.35599744e-01 7.32370973e-01 6.59687221e-01 8.68405104e-02
-2.14995310e-01 7.72871792e-01 2.74251997e-01 1.72335610e-01
2.16894567e-01 2.40056962e-01 9.03157443e-02 -3.36424671e-02
-7.67860770e-01 2.52370089e-01 -4.20722924e-02 3.16524595e-01
-1.22082189e-01 4.81418699e-01 4.67501879e-02 6.69876456e-01
3.98395330e-01 8.72817755e-01 7.85611629e-01 -8.62339020e-01
-2.22790420e-01 4.76062387e-01 -6.04409933e-01 -8.53326440e-01
-5.82214355e-01 -5.24211228e-01 -1.26056683e+00 3.79155129e-01
3.70310396e-01 1.55203938e-01 -1.25101006e+00 1.66274333e+00
5.31250715e-01 1.63824037e-01 -1.11018360e-01 8.54787946e-01
9.81417298e-01 -2.49678921e-02 4.54129964e-01 2.38285214e-01
1.49180233e+00 -5.29398799e-01 -2.09447727e-01 -9.62311849e-02
2.19376266e-01 -3.79102677e-01 1.01002121e+00 1.57877296e-01
-1.62742472e+00 -2.85520434e-01 -9.51877475e-01 -3.27497035e-01
1.45903137e-02 -2.16844812e-01 8.46387029e-01 8.89306724e-01
-7.64976561e-01 6.98569059e-01 -1.25675404e+00 -1.41218767e-01
7.83162236e-01 5.78364551e-01 -4.36947703e-01 -3.73842125e-03
-1.43931711e+00 8.85070920e-01 -1.75070509e-01 1.02481298e-01
-1.03971481e+00 -1.11074114e+00 -8.25993955e-01 -2.12896973e-01
-3.05063546e-01 -1.49841213e+00 9.43825364e-01 -4.85577583e-01
-1.39919055e+00 1.18848670e+00 6.14320576e-01 -6.18977129e-01
7.85521686e-01 -3.30829352e-01 -5.90147436e-01 1.39821112e-01
1.29544288e-02 7.74068296e-01 6.89670563e-01 -6.87207997e-01
-5.01169823e-03 -9.54828918e-01 -3.46594095e-01 1.37118816e-01
-2.16150042e-02 -9.00715888e-02 -8.24999586e-02 -1.13606322e+00
2.65477866e-01 -7.85183251e-01 -3.75498593e-01 6.32720590e-01
-3.35471064e-01 1.04000263e-01 2.92816728e-01 -1.17372632e+00
9.13161218e-01 -1.99343920e+00 3.38409722e-01 1.94247708e-01
5.19403815e-01 -1.43804729e-01 5.45763671e-02 5.91505431e-02
-1.39935717e-01 2.73721814e-01 -3.81082058e-01 1.75164357e-01
-1.58868894e-01 1.57130718e-01 4.77074057e-01 6.75338030e-01
2.36289557e-02 1.58289647e+00 -9.39200282e-01 -8.23493838e-01
6.67624250e-02 6.04125917e-01 -5.63074410e-01 -2.87289381e-01
1.08215220e-01 7.20034182e-01 -3.32045943e-01 1.06564760e+00
4.51823682e-01 -1.60734311e-01 -8.56813192e-02 -4.27810512e-02
6.05493546e-01 -4.22213912e-01 -6.15109921e-01 1.74343193e+00
-3.19312125e-01 3.07541251e-01 -1.91232175e-01 -4.81275558e-01
6.73959851e-01 1.61121577e-01 6.06976390e-01 -9.35921550e-01
2.04986915e-01 1.73243418e-01 2.37502322e-01 -3.75477731e-01
4.36913110e-02 -7.05743313e-01 -1.13454806e-02 3.41657251e-01
-2.68692672e-01 -1.44897282e-01 -2.97807693e-01 1.35244429e-01
1.11817789e+00 1.25947177e-01 9.57178697e-02 -8.66690278e-03
2.90638238e-01 -4.94422108e-01 2.98636019e-01 3.15718532e-01
-5.09138882e-01 9.18315530e-01 4.30496663e-01 -5.48093617e-01
-1.31176615e+00 -1.83800972e+00 -9.61950943e-02 6.37763321e-01
-6.13887906e-02 2.88403064e-01 -5.75792670e-01 -6.50497317e-01
1.24301709e-01 3.30334932e-01 -1.10367572e+00 -5.46453476e-01
-6.27266645e-01 -5.17464459e-01 1.23482502e+00 9.53996241e-01
2.18555644e-01 -9.30039883e-01 -2.51985341e-01 -6.01153448e-02
2.24984527e-01 -1.88163713e-01 -4.14136440e-01 -2.87629843e-01
-9.81914103e-01 -1.15515208e+00 -1.45927358e+00 -7.92917609e-01
6.39444470e-01 -9.25020158e-01 9.25013840e-01 -1.15644820e-01
-7.75755942e-01 5.15742958e-01 8.22873600e-03 -2.65549213e-01
-7.23623276e-01 -2.75953989e-02 2.95372128e-01 -6.38678432e-01
-8.60165730e-02 -1.01467574e+00 -1.31533098e+00 1.74489796e-01
-7.92304933e-01 -3.08891125e-02 8.19086313e-01 7.38784254e-01
5.41154265e-01 -1.13937706e-01 5.29659331e-01 -6.94952130e-01
6.11484826e-01 -4.21586812e-01 5.91260241e-03 1.87816650e-01
-1.02299285e+00 8.58440101e-02 1.23950802e-01 -5.08274138e-01
-5.43759465e-01 -4.04337555e-01 -5.15950382e-01 -1.37726799e-01
1.56797737e-01 2.97754139e-01 1.63061589e-01 -8.86060968e-02
9.27272439e-01 2.22865060e-01 3.83276969e-01 -2.00244382e-01
3.69391292e-01 1.46339908e-01 1.17000985e+00 -5.97681940e-01
7.51854241e-01 3.47685218e-01 4.55856740e-01 -5.26299238e-01
-1.29512474e-01 4.37790841e-01 -1.51693836e-01 -3.35310757e-01
1.02066374e+00 -7.69682229e-01 -8.12381506e-01 5.61832607e-01
-4.99980271e-01 -7.29749277e-02 -5.43324471e-01 6.45989299e-01
-5.71199834e-01 5.36484361e-01 -9.60565388e-01 -2.26028070e-01
-9.63447392e-01 -9.07965541e-01 8.31706762e-01 3.71500105e-01
-3.89022678e-01 -8.69517386e-01 3.83916169e-01 4.01121080e-01
4.07361239e-01 7.28892565e-01 1.38735592e+00 -7.55072385e-02
-1.41262949e-01 -4.07684952e-01 -9.91963744e-02 3.28807443e-01
-6.05688468e-02 4.32209298e-02 -4.71549809e-01 -1.39954016e-01
-3.75402451e-01 -3.51240426e-01 4.52849656e-01 7.38935888e-01
1.03837085e+00 -7.77878985e-02 -1.88237093e-02 7.36963511e-01
1.20665967e+00 2.80862898e-01 1.17480469e+00 2.90418088e-01
7.64043570e-01 1.43054262e-01 -5.75024560e-02 3.66011143e-01
4.32671249e-01 1.71566084e-02 6.30274355e-01 -2.33382061e-01
-4.81177211e-01 -4.39587653e-01 -1.29658086e-02 8.06337655e-01
-4.29927170e-01 6.80002198e-02 -1.09668422e+00 7.06296861e-01
-9.33313727e-01 -7.21554279e-01 2.23729480e-02 1.75823724e+00
8.77910852e-01 3.15176785e-01 2.11628482e-01 1.20464809e-01
7.31209099e-01 -2.05118358e-01 -8.04407537e-01 -2.18911916e-01
-6.12858944e-02 8.59235346e-01 5.41265905e-01 1.59693137e-02
-4.76758182e-01 5.05741775e-01 7.05750704e+00 5.39047837e-01
-1.06876826e+00 3.96818481e-02 8.19572330e-01 -2.76271105e-02
-7.86196828e-01 -3.00591707e-01 4.82163467e-02 5.02958834e-01
6.81655347e-01 -2.20773682e-01 1.72803283e-01 7.37950504e-01
-2.19001621e-01 2.02048030e-02 -1.01424646e+00 8.17023218e-01
-2.39121631e-01 -1.12063622e+00 -2.08239287e-01 -2.03501265e-02
5.23828924e-01 -3.38275075e-01 7.62418687e-01 2.91629463e-01
4.04654145e-01 -1.46373248e+00 2.85135061e-01 7.67233133e-01
1.31046510e+00 -8.44687343e-01 5.58352888e-01 -2.77695060e-01
-6.52090073e-01 2.01727480e-01 1.18205152e-01 2.30225161e-01
9.64201167e-02 2.62499690e-01 -8.17883134e-01 4.09501284e-01
6.59968555e-01 2.12980323e-02 -6.92967832e-01 9.49903905e-01
-3.06834012e-01 6.34845614e-01 -2.50567138e-01 1.34420738e-01
-1.56429470e-01 1.33460596e-01 2.48176545e-01 3.06419760e-01
3.76560301e-01 8.91264603e-02 -5.25188506e-01 7.42098272e-01
-1.01221733e-01 2.92659551e-01 -1.92928031e-01 -2.06643209e-01
9.55920294e-02 5.51244736e-01 -6.75455749e-01 3.09528291e-01
-1.05630286e-01 9.48271036e-01 -1.63126618e-01 1.37342572e-01
-9.23478603e-01 -2.68043965e-01 4.56530988e-01 4.75891203e-01
-1.83280841e-01 -4.52513359e-02 -6.19416296e-01 -6.44459188e-01
-3.21087152e-01 -8.38719368e-01 5.34762621e-01 -9.92646873e-01
-1.12925553e+00 1.41113237e-01 -2.97384799e-01 -7.87223876e-01
-2.79907256e-01 -3.32518995e-01 -5.25815606e-01 6.28757119e-01
-6.32641971e-01 -1.76400113e+00 3.18137929e-02 3.75344664e-01
2.18326375e-01 -2.24803388e-01 6.80863440e-01 4.34481651e-01
-5.48494816e-01 9.74124968e-01 9.20906514e-02 2.66892970e-01
6.48876071e-01 -1.51195014e+00 5.68044186e-01 4.46140677e-01
-4.77960855e-01 5.86630583e-01 7.14955270e-01 -1.10984910e+00
-1.18822694e+00 -8.11203778e-01 5.63881695e-01 -1.31897807e-01
6.42897069e-01 2.18893692e-01 -5.25257766e-01 6.01672888e-01
-5.48502430e-02 2.95423884e-02 6.58805072e-01 -8.42648558e-03
-1.90110952e-01 -3.32638882e-02 -1.34168422e+00 1.02280498e+00
8.66091251e-01 -2.81058043e-01 -4.09611702e-01 1.12250075e-01
6.47618473e-01 -6.30553126e-01 -1.52863145e+00 7.80553460e-01
1.16195011e+00 -5.49209714e-01 1.36307418e+00 -6.36140704e-01
1.23788929e+00 1.56461909e-01 1.66275874e-01 -8.04389000e-01
-2.07264677e-01 -4.45232809e-01 -1.64081767e-01 9.56257761e-01
4.46732044e-02 -2.32240811e-01 1.27960324e+00 5.46420455e-01
-7.08843246e-02 -9.75472569e-01 -1.10507023e+00 -5.45400202e-01
7.73326993e-01 -4.53538060e-01 5.80976129e-01 7.79610455e-01
-4.26803946e-01 5.01611922e-03 -5.32549262e-01 1.91385522e-01
7.09043503e-01 -4.84003752e-01 6.73416436e-01 -1.01577711e+00
-2.34920561e-01 -6.36061668e-01 -1.03685057e+00 -2.21643656e-01
-3.23601484e-01 -9.28928256e-01 -6.03086472e-01 -1.58711481e+00
1.14598662e-01 -3.05836886e-01 -3.20853323e-01 2.02063262e-01
-1.33340359e-01 4.72959250e-01 -3.34659338e-01 -1.60026744e-01
6.85877502e-02 5.48748016e-01 1.66166449e+00 -4.10150230e-01
8.00568238e-02 1.06556073e-01 -6.53322160e-01 5.78368306e-01
6.68910086e-01 -5.40898621e-01 -3.03872198e-01 -1.75075099e-01
5.74922681e-01 2.95873433e-01 6.13714516e-01 -1.14776301e+00
-2.72809435e-02 -1.72187805e-01 7.52789497e-01 -5.96192420e-01
1.87526494e-02 -4.31183398e-01 6.43779337e-01 1.15655708e+00
-2.56376445e-01 1.66728407e-01 -5.25607280e-02 3.29144359e-01
2.21116617e-01 -1.19493283e-01 9.42937315e-01 -3.18022728e-01
-4.53366935e-01 4.68961716e-01 -4.26023662e-01 1.96935534e-01
9.72407758e-01 -4.05435503e-01 -1.67039052e-01 -4.53180552e-01
-1.37751555e+00 2.03678906e-01 6.15117908e-01 1.58179671e-01
9.85142052e-01 -1.52609229e+00 -1.09675908e+00 2.49117568e-01
1.23741645e-02 -6.62697256e-02 7.49921322e-01 8.46140206e-01
-1.35837245e+00 -2.36407638e-01 -6.70965493e-01 -4.86181617e-01
-1.12337887e+00 4.74454999e-01 3.30029279e-01 -3.42235804e-01
-6.88685238e-01 1.12307370e+00 1.56078145e-01 -2.49403983e-01
8.72109532e-02 1.21168336e-02 1.68303415e-01 -4.47013766e-01
1.58615068e-01 8.16995740e-01 -3.36812615e-01 -2.24082828e-01
-2.11269379e-01 6.66952193e-01 -1.36470541e-01 -2.09504038e-01
1.21171212e+00 -4.32805941e-02 3.42242606e-02 2.12431654e-01
9.36110020e-01 -1.00955851e-01 -8.07772398e-01 3.26532423e-02
-4.74511385e-01 -1.32071868e-01 -1.95762292e-01 -9.79120493e-01
-1.15488517e+00 7.67307281e-01 1.37631047e+00 -2.73172319e-01
1.13996720e+00 2.57228225e-01 1.15666902e+00 -2.42981076e-01
8.17091670e-03 -8.38893890e-01 3.62132192e-01 -1.07783139e-01
9.37173784e-01 -9.53690231e-01 2.88781643e-01 4.84546162e-02
-4.66209650e-01 7.94703066e-01 6.84588492e-01 -1.20939553e-01
5.44128776e-01 1.90834075e-01 3.40412587e-01 -2.59613454e-01
-9.13314968e-02 1.05178162e-01 3.11373591e-01 1.00732553e+00
3.86483788e-01 2.52895504e-01 -7.26507306e-01 5.56671739e-01
-5.12195885e-01 1.45540044e-01 1.82514310e-01 8.48661840e-01
-9.85486507e-02 -1.04116702e+00 -4.47451532e-01 5.53628564e-01
-7.48067975e-01 -2.71616597e-02 -1.49853528e-01 1.06507552e+00
1.64465904e-01 -1.56016603e-01 -7.48222843e-02 -3.45868677e-01
-3.26100155e-03 8.46202821e-02 8.51018250e-01 -1.48465768e-01
-4.38985139e-01 -8.59179497e-02 -3.01167309e-01 -2.68660635e-01
-2.41874643e-02 -5.70082486e-01 -1.35867155e+00 -4.25955921e-01
-2.89545469e-02 -3.65879387e-01 9.02253687e-01 6.36894166e-01
-1.15543514e-01 7.42712915e-01 4.11527485e-01 -5.56367859e-02
-3.74611288e-01 -9.56013262e-01 -7.76101947e-01 4.20651346e-01
-6.82167038e-02 -5.30130744e-01 1.03931516e-01 -1.25733450e-01]
|
[14.132866859436035, -1.9663728475570679]
|
7ae060eb-aeed-48f1-a7fd-560dd098324a
|
yolo-facev2-a-scale-and-occlusion-aware-face
|
2208.02019
| null |
https://arxiv.org/abs/2208.02019v2
|
https://arxiv.org/pdf/2208.02019v2.pdf
|
YOLO-FaceV2: A Scale and Occlusion Aware Face Detector
|
In recent years, face detection algorithms based on deep learning have made great progress. These algorithms can be generally divided into two categories, i.e. two-stage detector like Faster R-CNN and one-stage detector like YOLO. Because of the better balance between accuracy and speed, one-stage detectors have been widely used in many applications. In this paper, we propose a real-time face detector based on the one-stage detector YOLOv5, named YOLO-FaceV2. We design a Receptive Field Enhancement module called RFE to enhance receptive field of small face, and use NWD Loss to make up for the sensitivity of IoU to the location deviation of tiny objects. For face occlusion, we present an attention module named SEAM and introduce Repulsion Loss to solve it. Moreover, we use a weight function Slide to solve the imbalance between easy and hard samples and use the information of the effective receptive field to design the anchor. The experimental results on WiderFace dataset show that our face detector outperforms YOLO and its variants can be find in all easy, medium and hard subsets. Source code in https://github.com/Krasjet-Yu/YOLO-FaceV2
|
['Xiuying Wang', 'Yahui Liu', 'YongXin Su', 'Weijun Chen', 'Hongbo Huang', 'Ziping Yu']
|
2022-08-03
| null | null | null | null |
['face-detection']
|
['computer-vision']
|
[-4.68128234e-01 -2.68445641e-01 -9.64339450e-02 -2.71031380e-01
-8.18943530e-02 5.05490554e-03 7.42549002e-02 -6.31949246e-01
-4.08380538e-01 1.35906070e-01 -6.31410480e-02 1.85102895e-01
2.34000877e-01 -6.66937649e-01 -4.38556015e-01 -8.53541315e-01
3.31843227e-01 6.53951541e-02 3.83492023e-01 -1.47015825e-01
3.81018110e-02 4.95192289e-01 -1.41081333e+00 6.12514652e-02
5.13679504e-01 1.14421344e+00 2.61465967e-01 1.58833697e-01
-2.12984428e-01 4.93443668e-01 -4.35425431e-01 -2.27728650e-01
4.69116390e-01 -3.86770099e-01 -5.57807125e-02 3.17648575e-02
1.76301435e-01 -6.23358130e-01 -6.06545508e-01 1.16325402e+00
1.14781272e+00 -2.88732387e-02 5.69704592e-01 -1.34045064e+00
-8.52901518e-01 4.07942861e-01 -1.23482621e+00 4.97655630e-01
2.14643776e-01 2.09131703e-01 4.82990772e-01 -1.37247610e+00
3.35868299e-01 1.82995296e+00 8.16398501e-01 9.28335190e-01
-7.37960637e-01 -1.10864687e+00 2.16880769e-01 3.59701604e-01
-1.65507436e+00 -7.73692608e-01 8.19756031e-01 -2.86083341e-01
5.32027543e-01 -1.61741123e-01 1.74065307e-01 9.26058233e-01
6.89205760e-03 1.01126444e+00 8.15124869e-01 -3.96725863e-01
-1.67226017e-01 3.33428442e-01 -3.91125642e-02 1.01949918e+00
2.89114565e-01 1.14003524e-01 -9.37236026e-02 1.27078310e-01
9.65778291e-01 3.58360648e-01 -3.61044973e-01 -1.09064639e-01
-3.66697520e-01 8.90986621e-01 8.01632524e-01 2.57896453e-01
3.36678624e-02 -7.52816275e-02 3.63061875e-01 7.25305527e-02
3.33027601e-01 -2.63671756e-01 -2.60114610e-01 6.23564303e-01
-5.66317737e-01 -1.67688262e-02 2.67392248e-01 8.80363882e-01
5.42790830e-01 1.01384446e-01 -2.88159877e-01 1.20459282e+00
6.61492705e-01 5.65086067e-01 8.75114858e-01 -4.08873618e-01
4.84623700e-01 8.15811694e-01 -2.83213824e-01 -9.38125849e-01
-4.79325831e-01 -2.82288700e-01 -8.74768317e-01 3.33380252e-01
1.71609759e-01 -2.83553451e-01 -9.85425711e-01 1.51003408e+00
6.02444649e-01 9.98042673e-02 -2.89321512e-01 1.09005463e+00
1.48268962e+00 6.67942822e-01 -1.28769577e-02 -1.38143063e-01
1.52517247e+00 -1.07182574e+00 -6.02649271e-01 -1.25236407e-01
3.68019938e-01 -7.72534251e-01 7.63634443e-01 1.62644967e-01
-8.73012364e-01 -9.70579684e-01 -9.37201977e-01 -1.71870410e-01
-3.27717215e-01 8.49639833e-01 4.36318398e-01 7.42064834e-01
-9.27411616e-01 2.13070229e-01 -5.20702183e-01 -3.87371212e-01
8.09359550e-01 6.26684666e-01 -3.81515145e-01 2.56919563e-02
-1.07757306e+00 4.98643637e-01 2.79197574e-01 4.71560150e-01
-8.00189734e-01 -2.63073325e-01 -7.81397402e-01 1.47829086e-01
4.90544111e-01 -3.18294823e-01 9.44349706e-01 -8.45807910e-01
-1.37138319e+00 8.95437777e-01 -9.83154997e-02 1.67187899e-02
4.82646048e-01 -7.40914643e-02 -4.98570293e-01 5.07222814e-03
1.15678072e-01 7.76934028e-01 1.06862509e+00 -8.51457775e-01
-6.41128123e-01 -6.92631364e-01 -2.08199665e-01 7.16105923e-02
-5.23140132e-01 4.25472409e-01 -7.41923332e-01 -5.34974813e-01
-6.71908110e-02 -6.70026600e-01 2.81088520e-02 3.85345697e-01
-3.75102729e-01 -7.91442692e-01 9.96768832e-01 -5.84428966e-01
1.35383749e+00 -2.33307290e+00 -1.90505221e-01 -1.54908216e-02
3.41601908e-01 7.84362435e-01 -1.98642373e-01 -1.29269481e-01
-1.52093768e-01 -7.08990768e-02 8.14150125e-02 -3.32347840e-01
-2.81033725e-01 -2.23774046e-01 -4.92151715e-02 6.09363735e-01
4.13377762e-01 7.19992042e-01 -4.56631273e-01 -7.20442176e-01
1.83296278e-01 6.33824050e-01 -4.95297462e-01 2.01394245e-01
9.47726145e-02 2.01984689e-01 -7.46721148e-01 9.04216170e-01
1.17432237e+00 -2.16575507e-02 -2.67149329e-01 -1.82944387e-01
-1.88120231e-01 -2.82951891e-01 -1.43222606e+00 1.08441365e+00
-2.63384372e-01 2.85170317e-01 3.37278426e-01 -7.12912977e-01
1.13103485e+00 1.57476783e-01 8.61785039e-02 -6.64324760e-01
6.80770457e-01 3.91936079e-02 9.26848352e-02 -5.16808569e-01
-1.89058095e-01 8.11581016e-02 5.14239550e-01 4.65632603e-02
1.32615432e-01 6.19695485e-01 -1.87290125e-02 -1.25738755e-01
6.39514744e-01 1.20426886e-01 2.14925170e-01 -1.38783693e-01
9.46589649e-01 -7.01242566e-01 1.01310742e+00 4.16732043e-01
-6.01390004e-01 6.99010313e-01 4.49431360e-01 -4.71652448e-01
-6.20773971e-01 -5.79918742e-01 -3.74360472e-01 1.24650216e+00
3.13643485e-01 -1.42557889e-01 -9.53178227e-01 -7.58018076e-01
2.32933909e-01 -1.70533344e-01 -7.80395925e-01 -1.98142827e-01
-8.04516673e-01 -9.01485741e-01 4.90906388e-01 7.68327653e-01
9.35589314e-01 -1.49562430e+00 -2.58234739e-01 9.27195512e-03
3.71443152e-01 -8.30792010e-01 -6.93461180e-01 -1.51751503e-01
-4.89064753e-01 -1.11230195e+00 -9.96735275e-01 -1.17634296e+00
8.64228666e-01 5.02129793e-01 5.47183216e-01 3.57548267e-01
-7.22322047e-01 -1.39398966e-02 -2.98552901e-01 -4.82402056e-01
1.31354347e-01 -1.02970250e-01 3.08469176e-01 2.11701334e-01
6.26918256e-01 -1.37208730e-01 -9.07047927e-01 6.33851528e-01
-6.97799325e-01 -4.92917955e-01 7.13271618e-01 7.44276285e-01
3.49012196e-01 -3.80853489e-02 5.82412064e-01 -6.79841816e-01
3.34592998e-01 -3.75570148e-01 -7.63139606e-01 2.38626495e-01
-4.56854105e-01 -2.12912578e-02 7.38410592e-01 -5.24238169e-01
-1.13257575e+00 2.73019493e-01 -3.94832969e-01 -6.49556875e-01
9.34680849e-02 -2.26038352e-01 -6.00867212e-01 -2.99710870e-01
3.82504910e-01 1.82200596e-01 2.02345084e-02 -7.20800817e-01
2.70567060e-01 1.00780296e+00 3.02578598e-01 -1.85527116e-01
7.65477121e-01 5.22099555e-01 -1.97871402e-01 -8.60193849e-01
-6.18885338e-01 -4.92567003e-01 -3.13664109e-01 -2.31513754e-01
8.48000824e-01 -9.95530605e-01 -1.07322717e+00 6.57867312e-01
-1.04035711e+00 -2.16422714e-02 1.00474879e-01 3.04518491e-01
5.68411499e-02 2.86484212e-01 -6.58565402e-01 -8.42782974e-01
-6.33478224e-01 -1.19171214e+00 1.04799676e+00 8.29346180e-01
5.39038002e-01 -6.05623782e-01 -2.01389596e-01 -3.95431295e-02
2.13437319e-01 -1.14513509e-01 1.98520362e-01 -3.34723085e-01
-3.24699014e-01 -1.37831926e-01 -6.32586420e-01 4.76160139e-01
-4.08460721e-02 1.16275772e-01 -1.21097827e+00 -5.32534778e-01
1.20523803e-01 -3.55914623e-01 1.40156031e+00 5.95978498e-01
1.37884808e+00 -1.31624684e-01 -6.67984605e-01 8.55989397e-01
1.27772307e+00 2.98101276e-01 6.35797679e-01 1.32580131e-01
7.98316300e-01 4.85603094e-01 5.83802760e-01 3.30106765e-01
1.12219542e-01 7.92457998e-01 1.45069495e-01 -1.74800262e-01
-3.87323916e-01 -3.46867502e-01 4.85575318e-01 4.15829420e-01
-1.03692450e-01 -1.72047973e-01 -5.57801545e-01 1.67309046e-01
-1.66826606e+00 -9.75285709e-01 4.02032211e-02 1.87458384e+00
4.40265596e-01 2.23473147e-01 3.20461810e-01 -7.56111927e-03
1.18161905e+00 3.34375575e-02 -6.45617783e-01 -1.17686160e-01
-1.56138986e-01 -3.03098820e-02 4.50838387e-01 2.94621825e-01
-1.21068966e+00 1.07490480e+00 5.18195391e+00 1.28330541e+00
-1.27800155e+00 2.62236416e-01 9.70112562e-01 -3.57956924e-02
3.64358395e-01 -3.79878312e-01 -1.60409510e+00 7.39243209e-01
1.67865425e-01 3.08124721e-01 2.56766349e-01 1.18938720e+00
2.30293378e-01 7.33542889e-02 -8.04213583e-01 1.44039381e+00
2.83115268e-01 -9.41314459e-01 -1.30898505e-01 -1.40818730e-01
3.74371350e-01 -1.82903215e-01 2.76794553e-01 4.50009048e-01
-1.02086112e-01 -9.92946684e-01 4.01284248e-01 2.47060731e-01
9.77158725e-01 -6.63944483e-01 7.60425746e-01 1.39365137e-01
-1.57053828e+00 -5.36997855e-01 -9.62776721e-01 1.54409423e-01
-3.42588663e-01 4.14090186e-01 -4.13329959e-01 8.25314671e-02
8.59158039e-01 6.46732330e-01 -6.21973038e-01 1.14798450e+00
-2.99290657e-01 3.46381307e-01 -4.83833432e-01 -7.63011947e-02
-3.04847886e-03 -3.91577408e-02 3.28488529e-01 1.00543606e+00
1.64546490e-01 1.26947701e-01 1.26872540e-01 7.36161947e-01
-3.60738784e-01 2.20497295e-01 -4.16821688e-01 4.66479927e-01
6.38426244e-01 1.58264947e+00 -7.25469291e-01 -1.20016783e-01
-4.84350055e-01 8.59426260e-01 3.20540339e-01 7.73307607e-02
-7.94841468e-01 -6.05666518e-01 3.51136953e-01 1.63499251e-01
6.72931015e-01 2.91349292e-01 7.39968792e-02 -9.48170781e-01
1.79409146e-01 -6.04242384e-01 5.33755958e-01 -5.57484150e-01
-1.15992296e+00 7.41777420e-01 -3.27589393e-01 -1.12452495e+00
3.50154996e-01 -9.28320765e-01 -9.54357088e-01 7.47071564e-01
-1.45783401e+00 -1.10788178e+00 -4.06663299e-01 5.62807202e-01
6.33929551e-01 -4.88089889e-01 2.10896850e-01 6.73371494e-01
-1.35199440e+00 1.10421693e+00 6.72187433e-02 5.89155197e-01
8.29590976e-01 -7.14782357e-01 2.29287177e-01 5.85504413e-01
-1.86513171e-01 6.57511771e-01 1.32751554e-01 -4.15543765e-01
-1.17238438e+00 -1.15428388e+00 4.57061410e-01 -1.53111935e-01
1.77940011e-01 -5.38171411e-01 -8.15221250e-01 4.31925088e-01
-1.51880831e-01 4.67356324e-01 2.08159879e-01 -3.04190367e-01
-3.69707912e-01 -5.28615415e-01 -1.42908287e+00 2.96103179e-01
1.12284899e+00 -7.67741129e-02 -3.14253211e-01 2.47493386e-01
4.34685767e-01 -2.17512190e-01 -3.71251255e-01 4.54898655e-01
5.79715848e-01 -9.99064505e-01 1.02105474e+00 -9.12271887e-02
7.01620355e-02 -4.36562985e-01 4.62005734e-01 -8.51841927e-01
-6.32720828e-01 -3.85166973e-01 1.02671251e-01 1.55151951e+00
-8.41189846e-02 -7.46842027e-01 8.28578651e-01 1.76128298e-02
1.10215627e-01 -8.80822539e-01 -9.24491346e-01 -6.48023009e-01
-1.39271736e-01 2.22078398e-01 5.05936146e-01 5.33564568e-01
-3.97215068e-01 4.25841600e-01 -2.98952281e-01 5.26196770e-02
5.53375244e-01 -7.42692798e-02 5.64683318e-01 -1.16549647e+00
-1.68652490e-01 -6.53345108e-01 -4.26406801e-01 -1.22995746e+00
-1.13858678e-03 -6.12596571e-01 -2.20375112e-03 -1.09790111e+00
4.26368207e-01 -5.16732216e-01 -2.55669981e-01 5.69166064e-01
-3.63949239e-01 6.08376086e-01 6.58839522e-03 3.22100073e-01
-6.21078908e-01 7.05648124e-01 1.19221377e+00 3.78083847e-02
-3.24466914e-01 -1.53148416e-02 -6.66386306e-01 8.77554476e-01
7.75055051e-01 -3.58770072e-01 -3.21321487e-02 -3.61524463e-01
-1.94715366e-01 -2.98636764e-01 2.08565742e-01 -1.04098427e+00
2.78770506e-01 2.09655538e-01 9.88678038e-01 -5.37962556e-01
2.16686383e-01 -5.61741769e-01 -3.58233511e-01 6.49558246e-01
-1.61884744e-02 -8.31330493e-02 2.10420802e-01 3.77560526e-01
-6.69821352e-02 -2.98294872e-01 1.18170786e+00 -6.29939511e-02
-9.11226451e-01 7.52648711e-01 8.11284259e-02 -9.68593508e-02
1.14570940e+00 -2.02622399e-01 -3.21519434e-01 4.27212752e-02
-5.50679922e-01 5.37394881e-01 6.77771121e-02 5.19761920e-01
8.21502209e-01 -1.40789044e+00 -8.58254492e-01 5.33896983e-01
3.86723466e-02 -2.00196117e-01 4.58312005e-01 6.68069243e-01
-4.64759499e-01 3.81653637e-01 -2.89715588e-01 -3.79689068e-01
-1.49848378e+00 6.29115164e-01 5.41652858e-01 5.75019047e-02
-5.70411026e-01 1.30916023e+00 6.52036369e-01 -2.12790444e-01
4.53683406e-01 7.66576380e-02 -5.52075863e-01 -4.58945595e-02
9.79784250e-01 4.24692094e-01 -2.50642598e-01 -8.04226577e-01
-6.94984615e-01 8.69513035e-01 -2.65616357e-01 5.44232547e-01
1.00557125e+00 -6.16843030e-02 -1.14529237e-01 -1.53379038e-01
1.25755334e+00 4.87612523e-02 -1.26722753e+00 -1.01935253e-01
-4.03772265e-01 -4.37667906e-01 1.46761268e-01 -3.72843623e-01
-1.50490701e+00 1.00336337e+00 1.19386888e+00 -3.94624472e-02
1.14205122e+00 9.13061798e-02 8.38129222e-01 3.38970087e-02
2.25443784e-02 -9.16357577e-01 2.14548886e-01 3.27927619e-01
9.50825036e-01 -1.44859779e+00 1.16607435e-02 -6.28673077e-01
-3.77081454e-01 1.11821616e+00 1.10317445e+00 -3.48736852e-01
7.75104105e-01 1.35287538e-01 7.59181902e-02 2.59482432e-02
-2.67328292e-01 -4.80652094e-01 1.72238857e-01 5.50916076e-01
3.62450004e-01 -1.55395746e-01 -3.77860516e-01 6.37492180e-01
2.07806617e-01 3.89601439e-02 -3.30881067e-02 4.81543660e-01
-7.51160920e-01 -1.00057316e+00 -6.05511844e-01 5.02457798e-01
-4.67251629e-01 1.11188725e-01 -1.83696046e-01 7.42648602e-01
7.02163339e-01 8.91203225e-01 4.02460843e-02 -4.87311363e-01
2.43206888e-01 -3.77136976e-01 5.04216909e-01 -5.34081101e-01
-3.84860456e-01 2.36741781e-01 -5.90494275e-01 -4.20054495e-01
-3.02516147e-02 -2.98452884e-01 -1.22746837e+00 -2.61738479e-01
-6.82576776e-01 4.31495570e-02 3.54428768e-01 6.37020230e-01
2.26253271e-01 2.60966331e-01 9.07128155e-01 -7.74757743e-01
-6.73289299e-01 -1.09018624e+00 -5.93662262e-01 1.12890609e-01
2.79515624e-01 -9.68373001e-01 -3.27334613e-01 -3.68715972e-01]
|
[13.3101167678833, 0.6427334547042847]
|
6cda27b9-717e-4a41-898f-5978594b5cc1
|
the-design-of-stratega-a-general-strategy
|
2009.05643
| null |
https://arxiv.org/abs/2009.05643v1
|
https://arxiv.org/pdf/2009.05643v1.pdf
|
The Design Of "Stratega": A General Strategy Games Framework
|
Stratega, a general strategy games framework, has been designed to foster research on computational intelligence for strategy games. In contrast to other strategy game frameworks, Stratega allows to create a wide variety of turn-based and real-time strategy games using a common API for agent development. While the current version supports the development of turn-based strategy games and agents, we will add support for real-time strategy games in future updates. Flexibility is achieved by utilising YAML-files to configure tiles, units, actions, and levels. Therefore, the user can design and run a variety of games to test developed agents without specifically adjusting it to the game being generated. The framework has been built with a focus of statistical forward planning (SFP) agents. For this purpose, agents can access and modify game-states and use the forward model to simulate the outcome of their actions. While SFP agents have shown great flexibility in general game-playing, their performance is limited in case of complex state and action-spaces. Finally, we hope that the development of this framework and its respective agents helps to better understand the complex decision-making process in strategy games. Stratega can be downloaded at: https://github.research.its.qmul.ac.uk/eecsgameai/Stratega
|
['Alexander Dockhorn', 'Diego Perez-Liebana', 'Jorge Hurtado Grueso', 'Dominik Jeurissen']
|
2020-09-11
| null | null | null | null |
['real-time-strategy-games']
|
['playing-games']
|
[-2.59816498e-01 1.60595104e-01 1.95729300e-01 9.15283710e-02
-2.37614661e-01 -9.03138220e-01 9.65727031e-01 -2.45511711e-01
-4.14447635e-01 7.67712414e-01 1.92470595e-01 -8.31033409e-01
-4.36576068e-01 -1.28380930e+00 -1.12110330e-02 -4.10162419e-01
-1.78837869e-02 9.64563608e-01 7.10196733e-01 -8.19019437e-01
2.78590143e-01 2.68610239e-01 -1.69557106e+00 1.61634013e-01
2.29298502e-01 2.44183660e-01 6.26994312e-01 1.07066298e+00
-9.13247317e-02 1.06086278e+00 -4.92146343e-01 -8.84367377e-02
4.46794152e-01 -8.39069724e-01 -8.69261861e-01 -2.73688436e-01
-8.99470925e-01 -1.18722022e-01 3.82483564e-02 6.50265515e-01
5.21159530e-01 3.36172611e-01 2.02613950e-01 -1.33314240e+00
2.87154764e-01 6.70494497e-01 2.43411437e-01 1.94957435e-01
8.58381093e-01 5.92814028e-01 6.78810239e-01 -6.33393899e-02
8.17904413e-01 1.11117947e+00 3.70931000e-01 5.84966421e-01
-7.78566539e-01 -5.09719670e-01 -3.83054763e-02 2.15913475e-01
-1.22128975e+00 -1.90079764e-01 5.07684469e-01 -1.78089038e-01
1.36377406e+00 6.56154335e-01 1.43563509e+00 8.94720614e-01
3.48704368e-01 5.75445592e-01 1.29211056e+00 -5.82909524e-01
6.03026748e-01 -2.15396821e-01 -3.79950374e-01 5.57565808e-01
-1.79002538e-01 4.84377295e-01 -1.19187295e-01 -1.25781938e-01
1.08134520e+00 -4.88468021e-01 3.79311085e-01 -3.79231453e-01
-1.04857361e+00 9.41052318e-01 -5.59998974e-02 5.30671537e-01
-6.43726707e-01 2.00441077e-01 4.48387057e-01 4.65913385e-01
-8.75257850e-02 8.27892661e-01 -3.87992948e-01 -1.13689888e+00
-5.95337808e-01 1.04265106e+00 1.24886286e+00 5.41572154e-01
3.37786674e-01 1.34897217e-01 1.23622008e-01 4.57182735e-01
5.12351751e-01 2.95249131e-02 4.33372974e-01 -1.53672802e+00
-2.78963596e-02 6.55825973e-01 3.31411868e-01 -6.91618443e-01
-9.06693518e-01 -1.50379285e-01 -1.15885906e-01 9.49944437e-01
5.49205065e-01 -2.19315231e-01 -6.56192183e-01 1.57833123e+00
4.52353895e-01 -4.43311408e-02 3.01517904e-01 7.02549458e-01
7.18667984e-01 6.26003623e-01 3.84695083e-02 -2.23330319e-01
1.32928312e+00 -6.95562840e-01 -4.91916835e-01 -2.16480821e-01
8.41422081e-01 -6.33530140e-01 8.42641294e-01 2.95506120e-01
-1.64308012e+00 -7.98638836e-02 -8.16381574e-01 4.30561036e-01
-5.99125683e-01 -7.34508514e-01 9.94404793e-01 7.55203724e-01
-1.29986429e+00 2.44047508e-01 -1.12006974e+00 -6.40256047e-01
-4.79814969e-02 4.31381971e-01 -3.81125063e-02 2.04621911e-01
-1.37045789e+00 1.23872936e+00 8.21038663e-01 -3.24679852e-01
-6.64604723e-01 -3.12469035e-01 -8.43585610e-01 -3.88360247e-02
6.31286800e-01 -8.61316562e-01 1.67080355e+00 -6.92226946e-01
-2.08196449e+00 6.34688735e-01 5.04962623e-01 -4.79341269e-01
6.50611579e-01 6.30049586e-01 -1.03634655e-01 -4.08763140e-01
2.43889794e-01 5.38920760e-01 -1.81261465e-01 -6.99034870e-01
-9.64357853e-01 -6.99826330e-02 8.77090752e-01 9.04392898e-01
6.67302608e-01 6.07078612e-01 -3.90769124e-01 -1.74191535e-01
-1.75588429e-01 -8.83663237e-01 -6.01411879e-01 -6.94102645e-01
1.15531050e-01 -5.45451529e-02 2.47597396e-01 -1.38897881e-01
1.38869381e+00 -1.61488879e+00 3.37253302e-01 1.14682220e-01
-1.52677849e-01 1.99000105e-01 -5.97877130e-02 1.07819569e+00
1.54821619e-01 -2.52963305e-02 2.15460151e-01 2.15342775e-01
3.46730679e-01 5.79386055e-01 2.90778637e-01 -4.86582071e-02
-4.93915349e-01 7.60747015e-01 -9.17637169e-01 -2.58241951e-01
6.55907929e-01 -4.66005206e-02 -7.04228103e-01 -1.19647495e-02
-4.03180450e-01 3.46474230e-01 -6.91500425e-01 2.97483116e-01
2.58706182e-01 1.78172722e-01 6.53454959e-01 8.28875124e-01
-8.63080323e-01 7.31409848e-01 -1.54921222e+00 1.61496556e+00
-4.78386611e-01 2.48088107e-01 1.87895834e-01 -7.42425799e-01
5.22444785e-01 4.03095633e-01 4.40714508e-01 -7.20785737e-01
2.86008686e-01 1.65507764e-01 6.86003029e-01 -1.82116076e-01
5.86748064e-01 -2.13360354e-01 -4.69042599e-01 8.09091270e-01
-2.86397159e-01 -7.85227716e-01 8.20481062e-01 1.14837930e-01
1.34333396e+00 4.04723138e-01 7.38288879e-01 -3.29912841e-01
4.43187475e-01 5.37653983e-01 3.58126909e-01 8.49068224e-01
-1.09554000e-01 1.23676769e-01 6.94657743e-01 -5.25015354e-01
-9.13206756e-01 -8.45998824e-01 2.01335758e-01 1.41136730e+00
7.05262460e-03 -8.78572404e-01 -8.19123864e-01 -5.66417053e-02
-5.25874019e-01 1.13283610e+00 -4.85628843e-01 2.05374643e-01
-6.00624681e-01 -6.07368469e-01 5.39012909e-01 1.32023841e-01
5.19707263e-01 -1.69197178e+00 -1.34027827e+00 7.15784431e-01
9.15495753e-02 -5.03538132e-01 4.69224378e-02 8.85903314e-02
-3.76540810e-01 -1.26145935e+00 -4.56772298e-02 -3.92844886e-01
-6.31189495e-02 4.40527871e-03 1.04486585e+00 1.45900145e-01
-6.29799217e-02 6.82380855e-01 -5.47989607e-01 -6.76581025e-01
-9.14456546e-01 2.66783893e-01 -2.35052466e-01 -1.04995251e+00
1.70146432e-02 -5.86119771e-01 -4.04501945e-01 4.52069521e-01
-9.29302454e-01 7.86088467e-01 -1.03431784e-01 6.20726764e-01
1.41635925e-01 4.25286025e-01 2.39629954e-01 -6.03304386e-01
1.20057750e+00 -3.79576266e-01 -9.74801600e-01 -5.87992668e-02
-3.68062973e-01 -2.70935118e-01 2.68926203e-01 -8.38074386e-02
-8.40480268e-01 -2.73661882e-01 -4.17386174e-01 4.00183558e-01
-2.41691396e-01 8.95129561e-01 -1.73176482e-01 2.35583052e-01
5.50107002e-01 2.19959810e-01 5.04181564e-01 3.46145779e-02
2.58148521e-01 3.09173703e-01 -5.12487628e-02 -5.97427130e-01
4.17134345e-01 6.18536957e-02 -1.35904178e-01 -4.44995672e-01
6.61071986e-02 -1.18889824e-01 -2.21876144e-01 -6.55944645e-01
6.80962026e-01 -4.89571661e-01 -9.23157096e-01 8.03646803e-01
-5.43535054e-01 -1.10737002e+00 -5.43417931e-01 2.62974858e-01
-1.06758893e+00 -2.92521417e-01 -3.36252630e-01 -8.11520517e-01
8.29973519e-02 -1.46975434e+00 3.88539016e-01 6.52107775e-01
-6.72947764e-01 -1.23357809e+00 4.50386405e-01 1.40996754e-01
6.39472961e-01 2.08252206e-01 5.56266963e-01 -7.12016702e-01
-5.00226140e-01 -1.40617147e-01 5.47962546e-01 -3.55059892e-01
1.41939864e-01 2.06319034e-01 -4.94433753e-02 -1.12017188e-02
-1.31154791e-01 -9.95842833e-03 -1.39192075e-01 5.39189935e-01
1.78401768e-01 -6.18625954e-02 -2.51197457e-01 1.99831083e-01
1.18053353e+00 9.47959483e-01 9.08404768e-01 1.20620239e+00
-3.32077891e-02 3.83718818e-01 8.70425045e-01 5.76966703e-01
7.76161551e-01 1.12014580e+00 4.79727179e-01 2.67959982e-01
1.58427671e-01 -7.29370043e-02 4.50368613e-01 4.05547917e-01
-5.87697566e-01 -3.16104889e-01 -1.18856859e+00 1.50483251e-01
-2.11212444e+00 -1.45584977e+00 -7.52547160e-02 1.80532026e+00
6.20381951e-01 3.71098429e-01 6.28590643e-01 1.56792283e-01
3.23312759e-01 2.48124108e-01 -1.50108024e-01 -1.00568247e+00
2.79123038e-01 4.43289757e-01 2.82254159e-01 9.33951974e-01
-6.75996780e-01 1.47007000e+00 6.26464510e+00 8.45034540e-01
-7.96637297e-01 1.58646822e-01 4.59374078e-02 -4.23569471e-01
-2.42636576e-01 2.67865419e-01 -3.90931219e-01 3.87492836e-01
1.00269938e+00 -7.84986317e-01 9.90094721e-01 6.25889003e-01
8.75160158e-01 -4.86855388e-01 -3.38557482e-01 3.47966015e-01
-6.23079240e-01 -1.75750136e+00 -4.96962994e-01 3.17070365e-01
2.33126387e-01 2.37199236e-02 -2.73412943e-01 6.77806020e-01
1.17521775e+00 -9.35495138e-01 1.05689442e+00 3.02824050e-01
3.31589371e-01 -9.96437550e-01 7.13646710e-01 7.00829446e-01
-1.14263344e+00 -2.39386201e-01 7.72870928e-02 -1.12085295e+00
4.28859562e-01 -4.64971781e-01 -9.35439169e-01 5.45321524e-01
4.81549263e-01 2.44777516e-01 -2.18813211e-01 9.71302748e-01
-2.36052096e-01 4.00278568e-01 -5.40604174e-01 -3.43612373e-01
5.60528934e-01 -5.28795063e-01 6.37940645e-01 7.07147181e-01
2.84795165e-01 5.64189553e-01 3.35963130e-01 5.98045409e-01
9.54859674e-01 -3.24135949e-03 -6.30168676e-01 -1.96801350e-02
6.03064716e-01 9.53142762e-01 -1.06869471e+00 -2.79194415e-02
-1.28620669e-01 5.18287241e-01 -1.52596101e-01 7.69777000e-02
-9.17911351e-01 8.99582542e-03 1.06962776e+00 4.28915828e-01
8.55535865e-02 -3.44368130e-01 -1.47346124e-01 -5.62463999e-01
-5.68864703e-01 -1.33631909e+00 4.02379364e-01 -1.06366646e+00
-3.51374149e-01 4.56863403e-01 5.26235759e-01 -6.99479818e-01
-1.06637716e+00 -4.71916050e-01 -1.02897024e+00 7.68520653e-01
-4.28866714e-01 -1.19117522e+00 7.99640194e-02 3.91201556e-01
5.99003434e-01 -4.38189626e-01 1.07042611e+00 -2.42770761e-01
-5.31610668e-01 7.53632002e-03 -1.57136753e-01 -1.64487287e-01
-5.66705130e-02 -1.22627234e+00 6.52357996e-01 6.13184690e-01
-2.36974090e-01 3.65172982e-01 1.09982562e+00 -5.79874694e-01
-1.27806044e+00 -3.00642669e-01 3.53563845e-01 -4.56776023e-01
8.62480640e-01 -1.57353073e-01 -2.39427224e-01 7.25129485e-01
4.89152551e-01 -6.31076872e-01 4.40239817e-01 5.57972267e-02
4.62610096e-01 4.39075261e-01 -1.04483664e+00 1.03729463e+00
1.10868275e+00 -7.41458684e-02 -4.95292246e-01 2.50130203e-02
1.32379740e-01 -9.11054194e-01 -5.84380746e-01 -3.42065915e-02
6.36483550e-01 -1.53918827e+00 8.62172067e-01 -4.42407042e-01
8.89538452e-02 -5.77372491e-01 2.80518401e-02 -1.42733085e+00
-5.32210112e-01 -8.38227093e-01 4.23600078e-01 8.97447646e-01
3.16833496e-01 -1.02640760e+00 8.17037404e-01 7.16846108e-01
-1.37027487e-01 -6.02703333e-01 -1.15180063e+00 -5.99201202e-01
1.80227816e-01 -1.01539302e+00 9.87062156e-01 5.56830585e-01
5.24244308e-01 5.58824800e-02 2.04317737e-02 -8.74583721e-02
-3.45765129e-02 -1.24352075e-01 9.75102484e-01 -9.20853794e-01
-6.46768749e-01 -8.95282388e-01 -5.73581755e-01 -3.34454149e-01
-7.93038309e-02 -6.76639199e-01 -1.22842006e-01 -1.93683803e+00
-2.17030331e-01 -5.39778709e-01 1.19062871e-01 5.27620435e-01
2.24974081e-01 -5.96940629e-02 6.00068510e-01 -5.81691489e-02
-5.34432113e-01 1.61244050e-02 1.26072979e+00 4.20080692e-01
-6.13952100e-01 3.34679276e-01 -7.60888040e-01 8.11370075e-01
1.21600664e+00 -3.49557936e-01 -6.30955696e-01 8.57087746e-02
6.26486003e-01 3.07523906e-01 2.55017757e-01 -1.14678133e+00
2.32877776e-01 -7.60340452e-01 -3.39605749e-01 -2.06799909e-01
3.86596799e-01 -4.77813542e-01 1.02209520e+00 7.83667505e-01
1.36829019e-02 4.17428583e-01 4.25453037e-01 -2.97282040e-02
3.11943330e-02 -4.86002326e-01 4.81485873e-01 -7.56617785e-01
-8.23988557e-01 -2.48663828e-01 -1.43903208e+00 -9.84295085e-02
1.42591071e+00 -7.33505964e-01 -7.03772753e-02 -7.26687670e-01
-8.02690685e-01 3.43957096e-01 9.64259684e-01 2.64704555e-01
1.82285339e-01 -1.06773496e+00 -4.74572182e-01 -3.57218757e-02
-2.29295254e-01 -1.99454367e-01 2.70873964e-01 5.39047897e-01
-1.08814096e+00 3.34318042e-01 -7.22751200e-01 -2.40731463e-02
-1.35129094e+00 2.72622019e-01 6.73685968e-01 -7.94289470e-01
-6.43509984e-01 5.93413830e-01 -8.18567574e-02 -6.22140467e-01
-3.54212791e-01 -2.30901584e-01 -3.84446084e-01 5.42147942e-02
6.13066852e-01 2.77654886e-01 -2.18395874e-01 -4.91278589e-01
-3.81109416e-01 1.76384244e-02 2.92700499e-01 -7.88225234e-01
1.50095069e+00 5.55397011e-03 -1.56483993e-01 2.42533028e-01
6.07031696e-02 -6.49366826e-02 -1.15511084e+00 4.87490207e-01
-7.54200444e-02 -2.58359045e-01 -2.29034461e-02 -1.05228388e+00
-5.76922476e-01 7.63933435e-02 3.22034299e-01 7.09684730e-01
1.03763771e+00 -1.67716652e-01 1.29337251e-01 -1.27240634e-02
9.43562329e-01 -1.08080959e+00 -3.84630948e-01 1.07059598e+00
9.13876295e-01 -4.59291309e-01 -3.06169037e-02 -5.82396761e-02
-8.14570487e-01 9.55483615e-01 4.20395464e-01 1.65585720e-03
3.32277894e-01 6.95060730e-01 2.14948282e-01 -4.68521506e-01
-1.10750008e+00 -6.15926325e-01 -5.38426578e-01 9.60682631e-01
1.06598943e-01 3.88887554e-01 -6.61941588e-01 5.37576318e-01
-7.65841961e-01 3.55622262e-01 9.79443014e-01 1.25744629e+00
-4.25323039e-01 -1.72763240e+00 -5.46305954e-01 2.25480497e-01
-2.22531527e-01 -4.51925350e-03 -2.95719922e-01 1.37863219e+00
1.21928014e-01 9.04199064e-01 1.53972447e-01 -3.10329437e-01
3.65592778e-01 5.56746013e-02 6.66138053e-01 -8.08040917e-01
-1.08007836e+00 -1.04148306e-01 7.94661939e-01 -5.84742427e-01
-2.61379391e-01 -1.08202755e+00 -1.37166297e+00 -6.95809782e-01
9.71381217e-02 3.68782520e-01 5.97668648e-01 8.48609507e-01
1.49115905e-01 6.36274934e-01 4.06258106e-02 -1.08665645e+00
-1.26913428e-01 -8.48511398e-01 -5.26656151e-01 -3.01156282e-01
-6.26564503e-01 -8.63660574e-01 1.20009826e-02 -5.43676376e-01]
|
[3.4513096809387207, 1.4707000255584717]
|
35a708db-4a12-42be-867a-b2fcff5a893b
|
spatiotemporal-networks-for-video-emotion
|
1704.00570
| null |
http://arxiv.org/abs/1704.00570v3
|
http://arxiv.org/pdf/1704.00570v3.pdf
|
Spatiotemporal Networks for Video Emotion Recognition
|
Our experiment adapts several popular deep learning methods as well as some
traditional methods on the problem of video emotion recognition. In our
experiment, we use the CNN-LSTM architecture for visual information extraction
and classification and utilize traditional methods such as for audio feature
classification. For multimodal fusion, we use the traditional Support Vector
Machine. Our experiment yields a good result on the AFEW 6.0 Dataset.
|
['Lijie Fan', 'Yunjie Ke']
|
2017-04-03
| null | null | null | null |
['video-emotion-recognition']
|
['computer-vision']
|
[-3.86558115e-01 -7.27050960e-01 -1.53479442e-01 -3.35037857e-01
-5.04309595e-01 -2.31067747e-01 1.37003839e-01 -9.86031070e-02
-6.26962364e-01 6.21288657e-01 2.44312853e-01 -5.70845418e-02
3.13319325e-01 -4.86469328e-01 -4.41607147e-01 -6.15477622e-01
-2.66693264e-01 -3.86752099e-01 -2.67304450e-01 -3.48532230e-01
2.26625860e-01 4.25195754e-01 -1.59052885e+00 9.52011406e-01
1.62860051e-01 1.94207513e+00 -2.86453992e-01 8.61128271e-01
-3.04303020e-01 1.07680786e+00 -8.79086912e-01 -4.30746198e-01
-7.65804574e-02 -8.13586712e-02 -6.38888240e-01 -1.38471154e-02
1.58345327e-01 -4.55960035e-01 -7.64648080e-01 7.91877389e-01
8.21024597e-01 4.82429504e-01 6.70499265e-01 -1.72795630e+00
-6.26940668e-01 4.03427869e-01 -6.47857666e-01 3.80667001e-01
6.03370845e-01 -3.11583966e-01 4.55925733e-01 -1.18717623e+00
4.06288892e-01 1.05194342e+00 7.03929007e-01 4.97161478e-01
-3.49302709e-01 -9.09632266e-01 3.03688291e-02 9.98529017e-01
-1.54182386e+00 -6.67821407e-01 1.01355827e+00 -3.46410483e-01
1.24720347e+00 1.06044590e-01 8.49205494e-01 1.49546587e+00
4.38674271e-01 1.17992353e+00 8.57517481e-01 -3.99024874e-01
-6.27193321e-03 1.44842686e-02 6.65273666e-02 9.48488414e-01
-3.73149544e-01 -1.72182173e-01 -9.82700288e-01 -2.06040651e-01
5.76402366e-01 2.83893257e-01 -3.61005574e-01 2.42945448e-01
-1.06040394e+00 8.70012820e-01 2.83820540e-01 6.16689801e-01
-3.62519056e-01 3.46413374e-01 1.03742397e+00 8.45262289e-01
4.40151036e-01 -8.87026358e-03 -6.07793450e-01 -1.73245072e-01
-8.61102998e-01 -1.05156630e-01 5.87064385e-01 6.38405800e-01
3.36068839e-01 4.77296889e-01 -3.43587846e-01 8.14086378e-01
3.96040320e-01 7.70277306e-02 8.39799881e-01 -1.14350009e+00
1.93737432e-01 2.42007360e-01 -6.50760084e-02 -1.27566957e+00
-1.97717831e-01 -4.86484952e-02 -1.08103371e+00 4.00316268e-01
-1.97398782e-01 -5.40729761e-01 -1.01277184e+00 1.43293357e+00
-2.84637123e-01 2.76716620e-01 3.37249309e-01 9.34217691e-01
1.60950160e+00 8.90027523e-01 1.07883431e-01 -2.39117563e-01
1.15496635e+00 -1.08063221e+00 -1.15401411e+00 1.69039190e-01
4.21662122e-01 -7.49657691e-01 5.94040453e-01 1.01595294e+00
-8.00101757e-01 -6.60738528e-01 -1.03200138e+00 -4.90281358e-02
-7.63614595e-01 5.94202697e-01 7.08654583e-01 4.03003842e-01
-1.29810190e+00 3.79739255e-01 -5.19040406e-01 -3.91937405e-01
3.13990921e-01 4.09979165e-01 -8.42269242e-01 3.05512935e-01
-1.34691060e+00 8.77897918e-01 6.73567235e-01 3.75197977e-01
-7.96244800e-01 2.34246895e-01 -1.06218565e+00 2.11095557e-01
1.55875802e-01 -5.46750546e-01 1.26557994e+00 -1.43362415e+00
-1.79601753e+00 6.21715784e-01 -1.87704116e-01 -2.36495897e-01
-1.29073232e-01 -1.71313852e-01 -7.76852310e-01 3.91450763e-01
-6.37068748e-01 6.93266153e-01 1.01408195e+00 -8.56363177e-01
-4.31750447e-01 -1.65065527e-01 -2.66058087e-01 1.40655339e-01
-8.47737730e-01 4.25964981e-01 -1.24077752e-01 -7.09478557e-01
-1.69725388e-01 -3.48577321e-01 3.95361893e-02 6.72855750e-02
-2.00202629e-01 -3.69450748e-01 1.37125731e+00 -8.45642269e-01
1.14577520e+00 -2.34928012e+00 2.29636014e-01 2.04944566e-01
5.29739261e-01 2.11333454e-01 -1.30947664e-01 4.64786112e-01
-2.86851019e-01 2.74661742e-02 2.79499561e-01 -5.48244119e-01
1.14666365e-01 9.56724435e-02 -2.79789090e-01 1.81953892e-01
1.08149141e-01 7.91735232e-01 -3.96077424e-01 -9.35665369e-01
2.02398270e-01 5.94113410e-01 -8.90142769e-02 4.79231745e-01
3.61379147e-01 1.27538042e-02 -2.75521874e-01 9.85429525e-01
2.30596304e-01 7.36232698e-02 -9.14357007e-02 -5.24753392e-01
-1.05177619e-01 -4.61890697e-01 -9.59683180e-01 1.95890093e+00
-1.98813573e-01 1.18656659e+00 1.82184204e-01 -1.46450484e+00
1.06009459e+00 9.98437881e-01 4.41032022e-01 -3.56266677e-01
7.46528745e-01 2.26712693e-02 -2.26925582e-01 -1.14331388e+00
4.18237895e-01 -4.73822206e-02 5.31777442e-02 1.23080112e-01
8.27122271e-01 4.96274263e-01 -1.25747994e-01 1.80391055e-02
8.39018106e-01 -5.74165024e-02 3.89585644e-01 1.67041019e-01
2.82933593e-01 -1.83932394e-01 5.77461123e-01 3.62589836e-01
-4.49408829e-01 2.54173785e-01 4.05759931e-01 -6.53201938e-01
-5.69849551e-01 -6.58122540e-01 2.44627640e-01 1.27959061e+00
-2.47099087e-01 -5.99446774e-01 -3.65612417e-01 -7.37998962e-01
-2.97985345e-01 4.79097851e-02 -7.11086631e-01 -2.43122146e-01
-5.29596815e-04 -5.08156121e-01 8.58781040e-01 9.43944037e-01
6.50067806e-01 -1.31545401e+00 -6.31600678e-01 7.74582028e-02
-4.06646460e-01 -9.86243367e-01 7.33798146e-02 2.54225552e-01
-6.12655282e-01 -7.89220452e-01 -7.59062052e-01 -1.03787434e+00
-7.31243342e-02 1.53523192e-01 8.93831015e-01 1.01749115e-01
-2.60272712e-01 5.68396986e-01 -6.82533443e-01 -5.87879300e-01
2.35909164e-01 -3.94708008e-01 1.21973991e-01 2.21858650e-01
5.83019376e-01 -4.26570177e-01 -4.84664440e-01 -2.38976941e-01
-7.36187696e-01 -2.98142582e-01 4.54305589e-01 8.48754108e-01
1.81250408e-01 1.76829144e-01 5.36290646e-01 1.93768367e-01
8.87736380e-01 -4.57991332e-01 1.38875186e-01 3.54880303e-01
-2.46522054e-02 -2.62968034e-01 4.06774312e-01 -7.74585187e-01
-8.98171961e-01 1.57985892e-02 -3.12031806e-01 -1.04120648e+00
-4.17144120e-01 1.01594317e+00 4.51259799e-02 -4.67614502e-01
3.18346411e-01 3.54198031e-02 -1.43677741e-01 -4.63995755e-01
1.20428065e-02 1.13763189e+00 4.79198694e-01 -2.48673096e-01
-1.98626578e-01 2.21905366e-01 -3.24324727e-01 -8.66697550e-01
-2.77774900e-01 -2.74594933e-01 -2.21948594e-01 -6.83803022e-01
1.08442760e+00 -9.88775194e-01 -1.12628102e+00 7.12608337e-01
-1.40756738e+00 9.59214941e-02 2.08687678e-01 8.78466368e-01
-3.26663971e-01 3.54516804e-01 -1.25993133e+00 -1.00954449e+00
-5.31192839e-01 -1.00424910e+00 9.07534182e-01 2.81775296e-01
-7.98073784e-02 -8.78764093e-01 1.63586259e-01 -2.00634643e-01
5.13724983e-01 4.40680891e-01 5.38515210e-01 -7.18337715e-01
3.18899274e-01 -2.56656706e-01 -9.01156142e-02 5.83809912e-01
-2.88340181e-01 6.46930814e-01 -1.22394490e+00 -7.76284337e-02
1.92942813e-01 -1.02547038e+00 1.46891809e+00 3.92404795e-01
1.62282348e+00 -1.25234455e-01 -2.64667422e-01 4.11285967e-01
1.30411685e+00 5.47387838e-01 7.21130729e-01 1.14762627e-01
5.90782344e-01 5.32444954e-01 3.90223235e-01 5.84387481e-01
3.08294088e-01 3.37795407e-01 3.35880607e-01 -4.11338329e-01
5.90897858e-01 1.71109021e-01 5.94913125e-01 1.13173783e+00
-4.19868976e-01 -4.76814836e-01 -7.94248044e-01 9.29589421e-02
-2.20260930e+00 -1.23088503e+00 2.39161059e-01 1.53816736e+00
4.62551534e-01 -3.56410742e-01 7.86285698e-02 4.07344580e-01
6.37112081e-01 1.94215447e-01 6.79318085e-02 -8.12300146e-01
-2.68930733e-01 3.91693175e-01 -5.64160123e-02 -3.69932465e-02
-1.43199039e+00 8.99897635e-01 7.42644882e+00 8.50508332e-01
-1.65195024e+00 3.04848522e-01 6.14015996e-01 -2.70422906e-01
3.40230495e-01 -7.30899453e-01 -1.49133369e-01 2.15873584e-01
8.74282181e-01 2.58126371e-02 2.55221814e-01 7.89318442e-01
-1.99091852e-01 -1.58842459e-01 -9.13618207e-01 1.89287090e+00
5.28430820e-01 -1.13033199e+00 -1.56639926e-02 -4.35673326e-01
1.90083593e-01 -1.78674832e-01 5.78096136e-02 6.50042355e-01
-3.09628636e-01 -1.38394618e+00 5.88113964e-01 9.55503583e-01
6.01088583e-01 -9.29142892e-01 1.04434347e+00 -6.92999735e-02
-1.32166886e+00 -2.93236732e-01 -2.94608504e-01 -1.90260500e-01
1.34688199e-01 2.17643246e-01 -3.86389974e-03 4.93178010e-01
1.20449638e+00 8.70765567e-01 -4.11309808e-01 1.09226096e+00
1.38742074e-01 3.28842014e-01 -1.10996313e-01 -1.89325660e-02
2.67172813e-01 1.33227527e-01 3.60085696e-01 1.58725810e+00
6.05989695e-01 1.32540286e-01 2.35725775e-01 2.77330905e-01
-2.25084186e-01 3.20276707e-01 -1.06967711e+00 -2.18994617e-01
2.03731153e-02 1.50467885e+00 -5.86195767e-01 -8.78865182e-01
-5.32134891e-01 1.03630579e+00 2.28050470e-01 6.74496531e-01
-1.02822614e+00 -8.49987149e-01 4.60044384e-01 -8.13177347e-01
3.37225586e-01 -1.97230250e-01 1.58308223e-01 -1.39494145e+00
-2.30007306e-01 -8.78197074e-01 5.23263931e-01 -1.26528454e+00
-1.30243611e+00 9.11908209e-01 -9.06813517e-02 -1.20873129e+00
-2.19636470e-01 -1.07097483e+00 -1.02441072e+00 2.27780476e-01
-1.07576728e+00 -9.22474980e-01 -1.92806900e-01 1.24725282e+00
4.11477655e-01 -5.32834232e-01 1.01902568e+00 6.25831902e-01
-6.85157657e-01 4.06288981e-01 -2.57897466e-01 3.92549515e-01
9.44905102e-01 -8.17040265e-01 -3.80114764e-01 5.49180150e-01
3.36144924e-01 7.36974701e-02 4.38774377e-01 -2.35080689e-01
-1.29761255e+00 -5.52002311e-01 6.72645032e-01 1.17481634e-01
5.45352995e-01 -1.37398466e-01 -7.09019125e-01 7.43116736e-01
1.05245066e+00 -3.02933175e-02 1.12290657e+00 -3.66501845e-02
-3.29536021e-01 -1.34567052e-01 -1.18759286e+00 3.69783014e-01
6.35586739e-01 -7.23854840e-01 -6.44850850e-01 -1.62005842e-01
6.14657938e-01 -7.44618922e-02 -8.92415166e-01 8.08249950e-01
8.02800298e-01 -8.08134079e-01 7.29617298e-01 -1.10698080e+00
5.33055544e-01 -1.44596398e-01 -6.47165060e-01 -1.45270860e+00
-2.87189573e-01 -2.56685704e-01 -4.75619435e-01 1.01300025e+00
1.78731099e-01 -2.94747293e-01 2.68978506e-01 3.32537413e-01
-9.01137963e-02 -7.44341850e-01 -9.33676183e-01 -3.52248430e-01
-6.46354198e-01 -5.50960839e-01 2.04832643e-01 1.35946453e+00
4.00368690e-01 4.06873673e-01 -6.23123288e-01 -2.49072209e-01
2.38160118e-01 6.88334927e-02 3.20234835e-01 -1.11445999e+00
-2.36485377e-01 -6.69702470e-01 -8.81527841e-01 -2.80630291e-01
5.05346060e-01 -4.24540013e-01 -1.08635552e-01 -1.36955512e+00
2.62673527e-01 6.46167636e-01 -1.05356646e+00 8.60180676e-01
4.71222579e-01 7.50251412e-01 1.21495575e-01 -4.00079846e-01
-1.02870798e+00 8.61137569e-01 7.27478266e-01 -4.59136575e-01
-7.20780119e-02 -4.45414752e-01 -4.52361345e-01 1.00994229e+00
1.02725196e+00 -2.19518587e-01 -6.03481345e-02 -3.89767319e-01
2.47758552e-01 3.17250133e-01 4.81026530e-01 -9.82990801e-01
2.23850384e-01 2.05124244e-02 1.07398343e+00 -5.72180450e-01
7.38275051e-01 -8.39792430e-01 -1.90537542e-01 2.32607618e-01
-1.59157574e-01 5.57473063e-01 4.78787839e-01 2.19737247e-01
-9.39657867e-01 3.37010585e-02 2.42485613e-01 -7.12787956e-02
-1.24815977e+00 2.64356911e-01 -1.03013706e+00 -5.33899188e-01
8.51657271e-01 -7.42974952e-02 -4.34399277e-01 -7.27396846e-01
-1.23573732e+00 1.76824361e-01 -8.15629289e-02 4.56032276e-01
1.34708500e+00 -1.83844543e+00 -4.29207593e-01 4.78897728e-02
1.88699260e-01 -9.30196881e-01 2.07324117e-01 1.12454426e+00
-1.60019204e-01 1.97097242e-01 -8.20806980e-01 -2.99504519e-01
-1.69824255e+00 5.51696420e-01 4.18852478e-01 3.66604656e-01
-8.43708366e-02 9.38484848e-01 -9.83734801e-02 1.46214142e-02
7.22297847e-01 -1.97217345e-01 -7.76844680e-01 4.17737424e-01
6.35617971e-01 4.36403483e-01 -4.13153507e-02 -5.63840032e-01
-4.47154492e-01 3.58853042e-01 2.74882644e-01 -4.12189871e-01
1.29158282e+00 3.18859406e-02 -3.06582749e-01 9.34221029e-01
1.40844131e+00 -4.17926043e-01 -4.67549592e-01 -1.94732472e-02
-2.71347135e-01 -2.90766597e-01 4.31923479e-01 -6.70625985e-01
-1.46624696e+00 1.26841128e+00 9.70034540e-01 2.80477256e-01
1.48996854e+00 -2.68469214e-01 6.94563866e-01 8.98438632e-01
2.22648934e-01 -1.40830994e+00 1.09595619e-01 5.28560996e-01
1.01509166e+00 -1.35113156e+00 -3.43908817e-01 3.29681247e-01
-8.71029139e-01 1.53184521e+00 8.55196714e-01 -1.08055338e-01
8.87181222e-01 5.16420126e-01 2.83557504e-01 -3.55536908e-01
-1.18091583e+00 -1.74120322e-01 4.85933840e-01 2.42670417e-01
7.33806491e-01 -5.51714823e-02 -2.28699103e-01 8.53348911e-01
2.14918256e-01 3.89713228e-01 2.31032804e-01 1.09410930e+00
-3.97719204e-01 -7.37952352e-01 -3.63296181e-01 3.79884750e-01
-7.50843585e-01 -3.18261385e-02 -7.54119754e-01 5.84326446e-01
1.68077335e-01 1.00990450e+00 2.55330324e-01 -1.09781051e+00
-4.40008851e-04 4.88765806e-01 6.54984772e-01 -2.33811773e-02
-6.24504864e-01 3.00526440e-01 1.60949901e-01 -7.75692761e-01
-8.84063721e-01 -1.30334988e-01 -1.08025432e+00 -4.27012891e-01
-1.95939526e-01 5.07106446e-02 7.28000104e-01 1.03944480e+00
2.81163633e-01 5.35023987e-01 4.46436197e-01 -1.17469132e+00
6.94862083e-02 -1.20867825e+00 -6.45197034e-01 -5.60610853e-02
4.19245601e-01 -7.69284546e-01 -4.80505526e-02 7.66877457e-02]
|
[13.328503608703613, 5.156844139099121]
|
19ce0b0f-6f99-432b-8db6-6a3100fd1ded
|
unify-a-unified-policy-designing-framework
|
2210.14030
| null |
https://arxiv.org/abs/2210.14030v1
|
https://arxiv.org/pdf/2210.14030v1.pdf
|
UNIFY: a Unified Policy Designing Framework for Solving Constrained Optimization Problems with Machine Learning
|
The interplay between Machine Learning (ML) and Constrained Optimization (CO) has recently been the subject of increasing interest, leading to a new and prolific research area covering (e.g.) Decision Focused Learning and Constrained Reinforcement Learning. Such approaches strive to tackle complex decision problems under uncertainty over multiple stages, involving both explicit (cost function, constraints) and implicit knowledge (from data), and possibly subject to execution time restrictions. While a good degree of success has been achieved, the existing methods still have limitations in terms of both applicability and effectiveness. For problems in this class, we propose UNIFY, a unified framework to design a solution policy for complex decision-making problems. Our approach relies on a clever decomposition of the policy in two stages, namely an unconstrained ML model and a CO problem, to take advantage of the strength of each approach while compensating for its weaknesses. With a little design effort, UNIFY can generalize several existing approaches, thus extending their applicability. We demonstrate the method effectiveness on two practical problems, namely an Energy Management System and the Set Multi-cover with stochastic coverage requirements. Finally, we highlight some current challenges of our method and future research directions that can benefit from the cross-fertilization of the two fields.
|
['Michela Milano', 'Michele Lombardi', 'Allegra De Filippo', 'Mattia Silvestri']
|
2022-10-25
| null | null | null | null |
['energy-management']
|
['time-series']
|
[ 6.06580615e-01 2.62758046e-01 -9.67548430e-01 -2.22623408e-01
-9.17354286e-01 -4.02631968e-01 4.29186016e-01 3.85821849e-01
-3.10995638e-01 1.22178543e+00 -1.00345939e-01 -3.52227479e-01
-4.79406387e-01 -8.80689561e-01 -6.51305556e-01 -8.35053205e-01
-7.00324997e-02 4.55041021e-01 -1.70128569e-02 -3.76834646e-02
9.46478620e-02 2.07590282e-01 -1.54171085e+00 -2.02135086e-01
1.11200023e+00 1.37071764e+00 3.13475341e-01 1.47164658e-01
3.04015398e-01 7.26085544e-01 -3.91115934e-01 -1.45316824e-01
2.02200845e-01 -1.50441363e-01 -8.49077046e-01 3.51136506e-01
-3.05443347e-01 4.85259145e-02 4.45010997e-02 1.01595557e+00
4.49326456e-01 1.18081950e-01 4.61922854e-01 -1.41399240e+00
-1.10797755e-01 7.14523554e-01 -6.64387882e-01 -9.67742354e-02
1.26766324e-01 4.44270484e-02 1.15206492e+00 -3.13193202e-01
1.97726905e-01 1.00019753e+00 4.40354377e-01 5.51862538e-01
-1.21156597e+00 -3.82028401e-01 7.34320045e-01 4.77960914e-01
-1.17212796e+00 -3.01320344e-01 6.65898025e-01 -2.76529998e-01
1.04734051e+00 2.44037390e-01 6.89807713e-01 8.34344745e-01
1.59717500e-01 1.05623972e+00 1.35045266e+00 -8.20949972e-01
8.60051930e-01 1.34759635e-01 -9.93996635e-02 3.50693315e-01
4.20056313e-01 3.95088643e-01 -8.41360986e-02 -1.26529276e-01
2.29914919e-01 1.19739883e-01 -2.60263681e-01 -7.90880322e-01
-7.82751024e-01 1.17928767e+00 2.28685990e-01 1.59177676e-01
-3.33008051e-01 7.34421164e-02 3.26075226e-01 2.16913491e-01
5.04818439e-01 5.40905058e-01 -5.43285549e-01 6.89992756e-02
-1.02449882e+00 2.64759243e-01 9.47619557e-01 1.07432294e+00
6.50057495e-01 3.41721028e-02 -2.90777296e-01 6.60722017e-01
2.78055698e-01 2.80736357e-01 5.36619574e-02 -6.64575458e-01
4.79029089e-01 4.52460021e-01 2.85921127e-01 -5.48278928e-01
-4.47398871e-01 -6.23390317e-01 -8.52402210e-01 3.11209917e-01
9.85301211e-02 -5.88781595e-01 -6.64301813e-01 2.01182103e+00
3.01367819e-01 -2.70659514e-02 4.60861549e-02 6.89143598e-01
-1.17096707e-01 6.57106757e-01 1.69909298e-02 -8.59571040e-01
1.17520332e+00 -8.65686476e-01 -9.05990064e-01 -4.02200371e-01
3.95925134e-01 -2.57283151e-01 5.31650782e-01 5.84486544e-01
-1.27354968e+00 -4.58924435e-02 -1.41252744e+00 6.72622800e-01
-3.43777984e-01 -8.69832411e-02 8.27547133e-01 7.42037892e-01
-8.17982435e-01 5.03321707e-01 -8.21578622e-01 -2.56510019e-01
5.32579720e-01 6.83094144e-01 3.59617680e-01 -2.61730313e-01
-1.21017671e+00 1.25470626e+00 5.74147284e-01 4.34348136e-02
-8.25678587e-01 -3.72563452e-01 -7.99847901e-01 1.48283839e-01
1.50365460e+00 -6.46245599e-01 1.43294632e+00 -8.57232571e-01
-1.50905180e+00 1.83096811e-01 3.27855527e-01 -6.73430562e-01
5.88408232e-01 -1.42197818e-01 -3.78815919e-01 -2.05603376e-01
-2.07560897e-01 2.14301243e-01 6.69932425e-01 -1.17084563e+00
-1.14683497e+00 -3.16592634e-01 4.71775323e-01 3.38274151e-01
-3.80475432e-01 -2.36888856e-01 -2.56902665e-01 -5.81249118e-01
-3.60159546e-01 -8.47366095e-01 -7.11441278e-01 -4.31207210e-01
-4.80006278e-01 -3.67089242e-01 6.28752708e-01 -1.93946779e-01
1.73001683e+00 -1.50990045e+00 5.73981524e-01 2.07194984e-01
-2.61247724e-01 1.71160564e-01 1.54045597e-01 6.72511160e-01
1.43849999e-01 1.16481997e-01 -6.62006974e-01 -3.72541517e-01
9.62145030e-02 5.29637396e-01 -6.84027597e-02 4.76196557e-01
3.64090145e-01 7.79091716e-01 -8.97760808e-01 -2.23388121e-01
2.58909822e-01 -1.08250603e-01 -4.73668545e-01 9.53670591e-02
-8.99784744e-01 2.07267404e-01 -6.46520853e-01 9.22761202e-01
4.71171230e-01 -1.54897287e-01 4.84576046e-01 1.69814736e-01
-6.91761449e-02 -1.20152265e-01 -1.58564997e+00 1.45374024e+00
-6.77628696e-01 1.16663270e-01 4.17957783e-01 -1.65209603e+00
4.10962969e-01 2.89506406e-01 8.03405344e-01 -4.85916138e-01
1.87022746e-01 2.48185560e-01 -1.05106652e-01 -4.55784053e-01
2.64058799e-01 -3.29275936e-01 -2.80954927e-01 4.10042375e-01
-1.92477033e-01 -2.68208027e-01 2.40046710e-01 -3.42732400e-01
9.65887725e-01 -2.39947299e-03 8.90018165e-01 -4.03088450e-01
6.42916739e-01 -6.97586909e-02 7.85626233e-01 7.07065225e-01
-8.32395628e-02 2.56676623e-03 5.17717838e-01 -1.82008624e-01
-6.76924586e-01 -3.55188459e-01 -3.02048087e-01 7.84389913e-01
2.01248169e-01 -1.14119187e-01 -6.99859023e-01 -8.81108344e-01
2.55159110e-01 8.67455006e-01 -5.46881616e-01 1.90339610e-02
-4.40529704e-01 -1.01642764e+00 -7.19255731e-02 6.31542981e-01
1.91304207e-01 -9.10416245e-01 -8.55671108e-01 3.18241119e-01
1.32007688e-01 -1.10229683e+00 -2.29157060e-01 6.61572814e-01
-7.61516273e-01 -9.97778356e-01 -6.31638765e-01 -4.54109341e-01
3.70781511e-01 1.32478932e-02 1.14347756e+00 -2.14632422e-01
-2.32994244e-01 6.55754805e-01 -4.37134445e-01 -8.29765320e-01
-3.00860763e-01 2.67189562e-01 1.91040069e-01 3.97479683e-02
8.87996256e-02 -4.01541144e-01 -2.80175030e-01 4.15706545e-01
-1.22418022e+00 7.55322806e-04 8.67692649e-01 9.51326251e-01
8.59412432e-01 4.06678289e-01 8.94141555e-01 -9.13605690e-01
5.49425006e-01 -7.49232829e-01 -9.43668306e-01 5.08258641e-01
-1.12303364e+00 1.26964420e-01 4.62558597e-01 -2.75083214e-01
-9.05679464e-01 -4.09092456e-02 8.50428566e-02 -3.72704536e-01
1.02374041e-02 7.92977631e-01 -4.91989970e-01 5.18413596e-02
2.28065819e-01 -7.64585212e-02 -4.97364104e-02 -1.46535635e-01
3.12830210e-01 5.72979033e-01 1.59323350e-01 -8.38139474e-01
7.06791341e-01 1.95127591e-01 1.54384077e-01 -4.25563246e-01
-9.91845131e-01 -2.46111661e-01 -2.84008294e-01 -1.26827598e-01
6.69984102e-01 -7.84699380e-01 -5.66753328e-01 8.77969190e-02
-7.24417746e-01 -3.02653104e-01 -6.19864821e-01 4.23975408e-01
-8.96185279e-01 2.65052408e-01 -1.41271442e-01 -1.31234264e+00
-1.86672911e-01 -1.18742204e+00 6.63906634e-01 3.28629404e-01
1.39410034e-01 -1.06791055e+00 1.53001733e-02 2.01382384e-01
3.89256686e-01 6.21004939e-01 1.17446232e+00 -4.98042941e-01
-7.42335737e-01 -2.17509523e-01 1.26767799e-01 3.72487992e-01
2.19159037e-01 -3.66567194e-01 -8.29629183e-01 -8.51540327e-01
2.37708196e-01 -6.42284930e-01 8.18284810e-01 6.02588117e-01
1.35342038e+00 -5.03254414e-01 -5.36025763e-01 2.89596200e-01
1.90087414e+00 4.01094198e-01 1.85243323e-01 3.89706254e-01
1.48115844e-01 6.63010478e-01 1.06130445e+00 7.36588001e-01
3.81153941e-01 9.96654689e-01 9.78710890e-01 -5.33004329e-02
3.51429343e-01 8.78148004e-02 2.44720861e-01 3.20666462e-01
-2.06200451e-01 -4.25962090e-01 -6.64372385e-01 4.93623227e-01
-2.35410333e+00 -8.05205047e-01 4.56807911e-01 2.48359156e+00
7.45822370e-01 2.85643607e-01 2.06866309e-01 4.06659424e-01
6.31024003e-01 2.53871650e-01 -1.08244777e+00 -5.26432037e-01
-4.43553589e-02 1.05066404e-01 5.78722298e-01 5.01977801e-01
-1.21263778e+00 4.94110316e-01 6.07178402e+00 1.12477803e+00
-8.45203459e-01 -7.75733218e-02 8.20621550e-01 -2.16139719e-01
-3.52317333e-01 -2.43623778e-02 -7.79853344e-01 4.06517655e-01
8.67342174e-01 -1.18898556e-01 7.44006574e-01 8.20265055e-01
1.42241076e-01 -3.71126950e-01 -1.40768528e+00 7.10311234e-01
-8.65360424e-02 -1.29434383e+00 -3.53522897e-01 3.06634635e-01
9.69112694e-01 -8.38009045e-02 -5.40499985e-02 5.33171535e-01
3.33840638e-01 -1.16735339e+00 7.49508381e-01 3.45665038e-01
7.77157843e-01 -1.00164235e+00 8.98749411e-01 6.83585525e-01
-1.10741901e+00 -6.40716672e-01 -1.02722429e-01 -2.94073403e-01
9.89312977e-02 8.49564970e-01 -2.72235662e-01 1.07037389e+00
3.83280486e-01 5.29364347e-01 1.46149606e-01 1.12161720e+00
-1.81189194e-01 4.99754101e-01 -2.61731774e-01 -2.63840079e-01
2.73135960e-01 -1.02475092e-01 5.03892601e-01 1.01431715e+00
-6.69386685e-02 8.23986381e-02 7.49573708e-01 7.32977092e-01
1.82816267e-01 -2.71555111e-02 -3.64231348e-01 7.75020570e-02
5.58807373e-01 1.17813766e+00 -6.14050031e-01 -3.95829082e-02
-6.90936983e-01 3.58054727e-01 3.31472933e-01 2.86098927e-01
-8.81296635e-01 -1.05451290e-02 5.44669390e-01 -2.40809262e-01
4.08512414e-01 -1.13837272e-01 -5.40008783e-01 -9.91816819e-01
1.94465026e-01 -1.10124385e+00 6.61463618e-01 9.91655663e-02
-1.20393085e+00 2.15968966e-01 4.26579207e-01 -1.17455387e+00
-4.30688769e-01 -5.99899650e-01 -4.96599823e-01 6.73081994e-01
-2.06503201e+00 -8.49230111e-01 -2.51763966e-03 2.67674088e-01
7.78787732e-01 -1.33798423e-03 8.19574416e-01 8.05931091e-02
-9.65813100e-01 3.72962683e-01 3.24942440e-01 -6.12957180e-01
3.31378132e-01 -1.40802598e+00 -1.64646089e-01 6.31328523e-01
-2.38483697e-01 -8.94754827e-02 5.67957342e-01 -3.61755878e-01
-1.64996183e+00 -1.09080136e+00 4.10438120e-01 -1.45333469e-01
4.16007996e-01 -3.61978948e-01 -4.16494161e-01 4.24553841e-01
2.78929025e-01 -1.75798595e-01 6.00776434e-01 1.07223861e-01
2.21516266e-01 -2.06657499e-01 -1.34748530e+00 4.52151597e-01
7.09593952e-01 6.04231209e-02 -2.51436591e-01 4.04588550e-01
6.26456678e-01 -2.94478446e-01 -8.72962117e-01 7.59999633e-01
2.51405716e-01 -7.57166088e-01 6.22178257e-01 -4.12084728e-01
2.60345161e-01 -1.24447174e-01 -4.90255296e-01 -1.30707181e+00
-2.71194220e-01 -8.25570643e-01 -8.49682093e-01 1.20514405e+00
4.35295194e-01 -5.68892956e-01 7.32380390e-01 6.42360389e-01
-1.45316407e-01 -1.71970332e+00 -1.03648162e+00 -8.18863213e-01
2.69225955e-01 -5.81547916e-01 6.14977777e-01 6.95876479e-01
1.85647875e-01 1.97044015e-01 -5.69267929e-01 2.64512122e-01
4.51014310e-01 3.30720156e-01 2.36610934e-01 -1.18575346e+00
-6.15676403e-01 -4.66174662e-01 1.57807052e-01 -9.81163383e-01
-1.00085223e-02 -6.62416995e-01 1.64966822e-01 -1.46215677e+00
2.45903000e-01 -6.90016747e-01 -6.21499360e-01 6.42824829e-01
-8.79688114e-02 -3.85563701e-01 3.19552213e-01 5.42848073e-02
-8.21766078e-01 8.18135560e-01 1.03861129e+00 -3.57210249e-01
-2.90903807e-01 6.28478348e-01 -9.18036938e-01 6.24630213e-01
7.95774221e-01 -2.82336861e-01 -5.57043374e-01 -2.15329394e-01
1.92498729e-01 4.31069046e-01 -1.47803470e-01 -1.08750808e+00
1.96720719e-01 -7.92830884e-01 7.65715241e-02 -3.92200470e-01
2.01478764e-01 -1.12494397e+00 -1.79041892e-01 7.42893338e-01
-2.65382171e-01 -9.90672931e-02 2.20259950e-01 9.68627274e-01
-1.59018449e-02 -5.27392209e-01 8.96497130e-01 -6.98385760e-02
-7.18239069e-01 4.86842185e-01 -2.13567287e-01 -1.86166875e-02
1.55706334e+00 -1.04678817e-01 7.74909183e-02 -4.05668706e-01
-6.07172549e-01 9.18057799e-01 2.70904034e-01 3.44758153e-01
2.53291190e-01 -1.01279616e+00 -6.46690190e-01 -1.93379745e-01
-2.26665866e-02 2.06745297e-01 -1.94218252e-02 8.01594973e-01
3.57145399e-01 6.96125925e-01 1.01896390e-01 -4.94358569e-01
-7.48552620e-01 1.00781047e+00 3.89118552e-01 -9.49491084e-01
-2.65115589e-01 4.00619775e-01 -2.05452546e-01 -1.15712740e-01
7.41822600e-01 -4.11162525e-01 -4.13772345e-01 2.23113120e-01
2.49091998e-01 5.59217393e-01 1.70710549e-01 -3.67089286e-02
-3.68209779e-01 5.00582099e-01 2.80802022e-03 4.82189357e-02
1.36359060e+00 -2.65541166e-01 3.58821511e-01 2.40356058e-01
4.93047327e-01 -2.00174391e-01 -1.40578914e+00 -4.79917914e-01
2.82291323e-01 -2.58725137e-02 3.05252343e-01 -1.25724077e+00
-1.05415130e+00 5.78410327e-01 5.43001473e-01 5.36620021e-01
1.51523018e+00 -1.20253973e-01 4.34147596e-01 3.95146489e-01
8.48186195e-01 -1.33753073e+00 -1.40629515e-01 3.43483865e-01
7.30404675e-01 -1.34220672e+00 3.46453756e-01 -4.05539930e-01
-5.67401171e-01 9.65328872e-01 4.83880252e-01 3.20307799e-02
6.11224473e-01 4.60899144e-01 -4.89537120e-01 1.79530129e-01
-1.04739118e+00 -4.27064151e-01 1.82397649e-01 6.16651654e-01
8.94428119e-02 2.25543231e-01 -5.69763303e-01 8.23402166e-01
3.07261139e-01 -3.44118662e-02 3.27366918e-01 1.31705976e+00
-5.19513845e-01 -1.36193931e+00 -3.90864193e-01 5.82236826e-01
-4.94647235e-01 2.56485641e-01 -5.89049570e-02 8.17416728e-01
3.74934793e-01 1.25175893e+00 -2.91209638e-01 -1.01000629e-01
3.55970293e-01 -1.43008977e-01 4.26149875e-01 -7.63543487e-01
-3.10640424e-01 7.83905685e-02 1.45335913e-01 -4.81897712e-01
-5.84914804e-01 -8.25572252e-01 -8.19063962e-01 2.42318392e-01
-7.64139354e-01 2.21464515e-01 6.49288356e-01 1.15809882e+00
6.43836111e-02 7.98724353e-01 9.34010804e-01 -8.34226906e-01
-1.27485657e+00 -6.92154169e-01 -6.74137414e-01 -2.62174249e-01
4.45228815e-01 -9.12206829e-01 -7.82390684e-02 -5.34484327e-01]
|
[4.607207298278809, 2.568974494934082]
|
483790d5-195b-4d21-a69d-5f49b76a6a52
|
spatial-moment-pooling-improves-neural-image
|
2209.14583
| null |
https://arxiv.org/abs/2209.14583v1
|
https://arxiv.org/pdf/2209.14583v1.pdf
|
Spatial Moment Pooling Improves Neural Image Assessment
|
In recent years, there has been widespread attention drawn to convolutional neural network (CNN) based blind image quality assessment (IQA). A large number of works start by extracting deep features from CNN. Then, those features are processed through spatial average pooling (SAP) and fully connected layers to predict quality. Inspired by full reference IQA and texture features, in this paper, we extend SAP ($1^{st}$ moment) into spatial moment pooling (SMP) by incorporating higher order moments (such as variance, skewness). Moreover, we provide learning friendly normalization to circumvent numerical issue when computing gradients of higher moments. Experimental results suggest that simply upgrading SAP to SMP significantly enhances CNN-based blind IQA methods and achieves state of the art performance.
|
['Hongwei Qin', 'Yan Wang', 'Yifan Shao', 'Tongda Xu']
|
2022-09-29
| null | null | null | null |
['blind-image-quality-assessment']
|
['computer-vision']
|
[-1.24788262e-01 -5.74761808e-01 -3.73597853e-02 -3.74000400e-01
-7.54693985e-01 -2.79569387e-01 5.03696740e-01 -4.28001443e-03
-5.61262310e-01 5.48849046e-01 4.31718320e-01 -3.23756039e-01
-2.90631205e-01 -9.08510029e-01 -4.27187890e-01 -5.83466649e-01
-1.55228540e-01 -6.02121055e-01 -4.43873135e-03 -2.97360271e-02
5.72717190e-01 7.63887048e-01 -1.42701733e+00 3.79765809e-01
1.02701676e+00 1.54223871e+00 -3.59037608e-01 5.77236295e-01
-1.54484272e-01 6.73399985e-01 -7.53770709e-01 -9.22147810e-01
4.35551584e-01 -3.25468421e-01 -8.03704262e-01 -2.04019114e-01
5.11652350e-01 -5.36003470e-01 -5.32164335e-01 1.26707530e+00
9.08589244e-01 6.09205253e-02 5.62096238e-01 -1.07123089e+00
-1.18390608e+00 -3.35533312e-03 -5.66446364e-01 4.78866667e-01
1.74852014e-01 5.47914624e-01 8.95382702e-01 -1.02969360e+00
-4.86365035e-02 1.07954144e+00 9.15337443e-01 2.99912632e-01
-7.02692807e-01 -5.35473049e-01 -2.14021906e-01 5.19834161e-01
-1.18797743e+00 -2.30969697e-01 7.47667849e-01 -2.98369080e-01
1.13538206e+00 2.78402716e-01 6.94808960e-01 6.74254179e-01
1.43448025e-01 7.62509465e-01 1.36089945e+00 -1.56603023e-01
1.76750705e-01 -2.83921272e-01 -2.01443493e-01 7.13874936e-01
2.71874070e-01 -6.10540286e-02 -5.32355607e-01 8.33305717e-02
9.23322141e-01 -8.53247289e-03 -2.30461389e-01 1.19299456e-01
-1.27751601e+00 6.05869293e-01 8.45175087e-01 4.50909436e-02
-5.67673922e-01 5.35816327e-02 3.75789464e-01 3.91823590e-01
3.18740398e-01 5.38439929e-01 -4.63712752e-01 -1.40172392e-01
-1.16482997e+00 2.76758432e-01 3.15575987e-01 5.49267590e-01
8.01168323e-01 -3.70050706e-02 -7.88917422e-01 1.12032497e+00
1.72867134e-01 5.92298031e-01 5.57007432e-01 -1.07320678e+00
3.04726392e-01 6.62112772e-01 -3.14966328e-02 -1.08872807e+00
-4.69872922e-01 -6.50954008e-01 -1.28050828e+00 6.18513286e-01
5.60334623e-01 7.76793212e-02 -1.04973340e+00 1.21078646e+00
-3.53018731e-01 -2.41656601e-01 -2.79604495e-01 1.19917190e+00
8.32140803e-01 4.06017482e-01 2.47364249e-02 7.79915005e-02
1.31048548e+00 -8.35929155e-01 -5.17200530e-01 1.19751170e-01
-2.33519007e-03 -9.68593240e-01 1.31147158e+00 5.70321083e-01
-1.22654653e+00 -7.65307188e-01 -9.84474599e-01 -2.04749107e-01
-6.78781927e-01 4.35248107e-01 7.19813645e-01 7.67113864e-01
-1.55297267e+00 9.04660940e-01 -4.42495018e-01 -1.67123511e-01
1.01933336e+00 5.57603121e-01 -4.13350105e-01 -7.95039982e-02
-1.09181416e+00 4.63774949e-01 -8.10976475e-02 1.87445134e-01
-5.18640220e-01 -5.52870810e-01 -6.81314111e-01 2.59244461e-02
-3.64518046e-01 -8.98047686e-01 1.05769956e+00 -8.67801547e-01
-1.78907919e+00 7.10297287e-01 -3.84550154e-01 -4.03185844e-01
3.97195220e-01 -1.94453627e-01 -3.94439757e-01 2.77788639e-01
-6.00144193e-02 6.14716530e-01 1.19550419e+00 -8.50813150e-01
-3.25652897e-01 -4.69690651e-01 1.57398582e-01 -1.35581389e-01
-7.30584979e-01 5.17056227e-01 -4.18883294e-01 -9.70362008e-01
3.31772059e-01 -2.34623104e-01 1.47279100e-02 2.59819955e-01
-4.98620003e-01 -3.30123127e-01 3.41592520e-01 -1.08879352e+00
1.42090857e+00 -2.05826712e+00 -1.66807458e-01 3.39044213e-01
4.16000605e-01 5.13123214e-01 -2.88745433e-01 1.20935328e-01
2.47534718e-02 2.03092992e-01 -2.72006452e-01 -3.85272712e-01
1.87884733e-01 -2.38482669e-01 -8.43429714e-02 4.92419213e-01
5.94307423e-01 1.08611321e+00 -5.50740600e-01 -3.72798443e-01
1.99758306e-01 6.92408681e-01 -7.51882255e-01 4.61216830e-02
2.17782021e-01 3.31078261e-01 -1.15888074e-01 9.62593019e-01
1.11698520e+00 -3.88833106e-01 -4.93929625e-01 -3.94932389e-01
-2.38466248e-01 3.27367067e-01 -7.99354434e-01 1.48642242e+00
-5.47391176e-01 7.24261522e-01 -1.81590199e-01 -7.04094529e-01
1.01983809e+00 1.23172201e-01 2.12304056e-01 -1.06294453e+00
2.57403374e-01 8.52066725e-02 5.72215812e-03 -5.48928857e-01
4.82776225e-01 1.05089054e-01 3.57478142e-01 1.17661297e-01
3.11034828e-01 1.35451034e-01 -7.51526952e-02 -1.04298219e-01
8.64216030e-01 -3.35787117e-01 -5.73818199e-02 -9.95776579e-02
7.19728053e-01 -6.42511606e-01 3.28429729e-01 5.72180808e-01
-7.27150679e-01 1.13938117e+00 6.56897366e-01 -4.19177115e-01
-1.20916438e+00 -1.29937208e+00 -2.44853437e-01 7.88843989e-01
-7.19888285e-02 -3.61663878e-01 -7.63853669e-01 -4.87609863e-01
-3.23106200e-02 -3.01246524e-01 -6.07155085e-01 8.08084086e-02
-4.19030547e-01 -1.08665943e+00 5.84738016e-01 7.82102346e-01
1.29838848e+00 -1.12396479e+00 -7.25026056e-02 7.52862841e-02
-7.92844146e-02 -7.57702351e-01 -3.98877919e-01 -3.15184683e-01
-8.59314144e-01 -8.28706980e-01 -1.47681391e+00 -7.63462961e-01
5.67570031e-01 1.69468850e-01 8.86690199e-01 1.54170813e-03
-3.35254252e-01 5.58998585e-02 -2.60601461e-01 -2.99320936e-01
3.48005503e-01 -1.87488139e-01 8.59887451e-02 2.15555161e-01
3.40391338e-01 -6.50221884e-01 -1.18105567e+00 -2.75406409e-02
-8.82810235e-01 -3.84476721e-01 9.23561990e-01 6.26562119e-01
5.02749622e-01 8.11863840e-02 5.92522085e-01 1.31983086e-01
1.07489061e+00 1.28833607e-01 -3.91637146e-01 1.98635440e-02
-3.94684464e-01 -1.27204821e-01 6.67567372e-01 -1.79032326e-01
-7.96928763e-01 -5.03039241e-01 -3.08897793e-01 -2.88653523e-01
-1.29264817e-01 3.40135247e-01 -2.27616206e-01 -5.27568281e-01
4.76821601e-01 2.76744813e-01 -5.26812784e-02 -6.49299979e-01
2.74603724e-01 7.91277826e-01 6.94051921e-01 -2.93383658e-01
7.80182779e-01 5.18202484e-01 -2.46587917e-02 -5.46512783e-01
-6.14198506e-01 -3.47261816e-01 -5.29217243e-01 -2.15576857e-01
9.67056394e-01 -8.51047397e-01 -1.09272730e+00 1.14718866e+00
-1.19817090e+00 -8.79550725e-02 -8.95671248e-02 6.19993687e-01
-2.72551298e-01 4.94738668e-01 -8.26578021e-01 -7.52985597e-01
-6.86873734e-01 -1.20035040e+00 8.74231398e-01 5.85892200e-01
1.12766936e-01 -5.94822109e-01 -1.50043935e-01 3.45758438e-01
9.91525471e-01 -5.93898669e-02 8.66321146e-01 -4.74799983e-02
-5.09080946e-01 -3.39944720e-01 -1.09881330e+00 8.50024939e-01
6.66259304e-02 -9.98240486e-02 -1.17971241e+00 -4.07455951e-01
-2.94288307e-01 1.41438060e-02 1.13419127e+00 7.32417524e-01
1.50061893e+00 -5.29686868e-01 4.20661628e-01 1.04866159e+00
1.33151376e+00 -2.26474572e-02 9.78704095e-01 6.98159873e-01
4.99579370e-01 2.26915792e-01 -1.87560841e-01 5.69900990e-01
2.98805714e-01 4.14866060e-01 5.03633976e-01 -2.83488035e-01
-4.82684761e-01 -2.70884018e-02 2.31295779e-01 7.14098394e-01
-4.88456190e-01 3.21360111e-01 -7.52116323e-01 6.76057637e-01
-1.20846808e+00 -7.95413256e-01 1.29038608e-02 2.14551234e+00
7.68935978e-01 9.72858630e-03 1.60810485e-01 4.66747999e-01
6.16056681e-01 2.62102246e-01 -4.54492360e-01 -3.12385768e-01
-5.17342448e-01 7.48576462e-01 6.04424775e-01 2.75845617e-01
-1.32854080e+00 6.92346275e-01 6.27606726e+00 7.85927653e-01
-1.22650087e+00 4.33777869e-02 7.73300886e-01 -4.03325027e-03
-5.29764518e-02 -3.26079339e-01 -1.46811470e-01 4.90742087e-01
4.38553452e-01 1.36914194e-01 7.53912508e-01 5.03346801e-01
5.67345694e-02 6.65310025e-02 -2.93663949e-01 1.31281233e+00
-3.20598879e-03 -1.10049629e+00 3.55696470e-01 4.70045656e-02
9.67779458e-01 5.95252179e-02 8.12427104e-01 6.72012120e-02
-1.07814439e-01 -1.35219574e+00 4.04023290e-01 8.84881198e-01
9.19334471e-01 -8.90071809e-01 1.14634049e+00 -3.88309926e-01
-1.10822344e+00 -3.33858848e-01 -6.22555494e-01 -2.29867637e-01
-2.50087291e-01 9.66226399e-01 -3.35006565e-01 5.50918758e-01
1.15362859e+00 6.85201585e-01 -1.10037470e+00 1.63906407e+00
-1.85037822e-01 4.28564727e-01 3.83680463e-02 7.48324618e-02
2.08854705e-01 -4.21969369e-02 3.43443841e-01 1.06565499e+00
4.14718628e-01 -5.77239878e-02 -6.84852183e-01 8.59953165e-01
-4.05068338e-01 2.54961580e-01 -1.48303553e-01 1.10107306e-02
5.03759496e-02 1.11441720e+00 -5.26133955e-01 -3.46437663e-01
-5.86361408e-01 1.22545040e+00 -2.53911205e-02 5.93047678e-01
-3.74291778e-01 -1.09186876e+00 1.07350028e+00 -4.14906032e-02
3.00985307e-01 -2.93950617e-01 -8.15372169e-01 -1.23235929e+00
3.70564461e-01 -6.71814740e-01 5.16622514e-02 -8.92316103e-01
-1.59793603e+00 6.08829975e-01 -6.43415630e-01 -1.39121783e+00
3.38703245e-01 -9.82317805e-01 -8.28317642e-01 1.42776787e+00
-1.97959924e+00 -1.03791261e+00 -4.18138534e-01 8.96280944e-01
3.71499322e-02 -3.04444671e-01 4.76054102e-01 4.92510885e-01
-5.45245767e-01 1.00373757e+00 -4.45883116e-03 7.13368952e-01
7.64435887e-01 -1.22903192e+00 5.42648852e-01 1.01858187e+00
-2.54193693e-01 8.34341228e-01 1.68081313e-01 -3.07435989e-01
-1.12324786e+00 -1.06168270e+00 1.06395304e+00 -1.83917999e-01
5.04212856e-01 1.60569057e-01 -8.56355727e-01 1.56962182e-04
1.33074522e-01 3.26215118e-01 4.25312310e-01 -1.00664265e-01
-7.68302083e-01 -3.65166545e-01 -1.34991789e+00 5.02179086e-01
9.09857273e-01 -1.06081998e+00 -1.90739602e-01 4.63398173e-02
4.26252723e-01 7.85094574e-02 -8.59960973e-01 7.12772489e-01
6.87392116e-01 -1.42682195e+00 1.18032026e+00 -5.31244457e-01
4.60603833e-01 -3.27546507e-01 -8.51135626e-02 -1.15027249e+00
-4.78055835e-01 -4.42378938e-01 5.50472885e-02 8.71152937e-01
2.48942241e-01 -7.23710537e-01 6.15765154e-01 3.14118803e-01
-1.75049126e-01 -8.09465468e-01 -1.02414453e+00 -7.46867955e-01
3.09522897e-01 -5.42554498e-01 9.69636440e-01 7.21188903e-01
-2.12799069e-02 -5.81045806e-01 -2.29589671e-01 3.56920809e-01
5.68929911e-01 -3.10143214e-02 3.40612531e-01 -1.17130792e+00
-9.37870741e-02 -1.02486575e+00 -8.70726109e-01 -8.24606061e-01
-3.33096743e-01 -8.20183694e-01 -2.79499650e-01 -1.86304104e+00
1.42398030e-01 1.17711909e-01 -8.71242821e-01 4.12133425e-01
-1.90035924e-01 9.44692791e-01 1.44998372e-01 4.51119654e-02
-5.55699885e-01 5.45354009e-01 1.58673799e+00 -2.74153233e-01
-8.52111280e-02 -8.39858726e-02 -5.56489348e-01 5.63059449e-01
9.79453504e-01 1.58487350e-01 2.38735080e-01 -6.00283504e-01
3.50046545e-01 -2.38788679e-01 8.30697000e-01 -1.48730290e+00
1.72385350e-01 3.03396434e-01 9.68741477e-01 -2.43765622e-01
1.26152232e-01 -2.43404552e-01 -5.84316671e-01 2.79158473e-01
-1.61880851e-01 1.25219718e-01 1.94640979e-01 1.09665394e-01
-6.06728554e-01 2.63105214e-01 8.62404704e-01 6.65859506e-02
-5.03475428e-01 7.11963236e-01 -7.23570958e-02 -2.96879143e-01
2.58598655e-01 -5.84329367e-02 -4.61147338e-01 -3.90309304e-01
-5.97308397e-01 -3.48315537e-01 3.12445611e-01 2.59747654e-01
1.01246774e+00 -1.52140915e+00 -6.28744066e-01 5.41299522e-01
8.77041444e-02 -4.06857759e-01 5.08456409e-01 9.60963428e-01
-6.53918326e-01 4.48498338e-01 -5.90924680e-01 -1.05819777e-01
-8.13185751e-01 3.34877938e-01 4.06519681e-01 5.57185113e-02
-4.13517714e-01 1.08704519e+00 -2.46406317e-01 -2.25354478e-01
3.73604536e-01 -4.58915234e-01 -3.97636324e-01 3.56847905e-02
8.76141369e-01 5.64353287e-01 3.83318037e-01 -5.50711393e-01
-3.01013827e-01 6.79958582e-01 1.14944309e-01 -1.45277217e-01
1.26436388e+00 8.83876681e-02 -4.82641429e-01 -2.01893687e-01
1.48732948e+00 2.37955842e-02 -1.28756261e+00 -1.89770281e-01
-2.06780255e-01 -6.97691977e-01 2.19391376e-01 -1.07338679e+00
-1.36977160e+00 1.35931456e+00 1.01650393e+00 1.95393756e-01
1.53058946e+00 -1.47086486e-01 1.01929414e+00 1.83333814e-01
1.78485706e-01 -7.92791247e-01 2.63443083e-01 5.47074556e-01
9.52354550e-01 -1.24852002e+00 -2.17450842e-01 1.53215200e-01
-1.40319437e-01 1.19877076e+00 4.32245582e-01 -2.87328303e-01
6.62059605e-01 -1.88737378e-01 2.20624104e-01 2.50703432e-02
1.11397155e-01 -4.17152464e-01 6.81794643e-01 7.78709173e-01
3.88195962e-01 6.72181696e-02 -2.30464473e-01 8.80575478e-01
-4.59964156e-01 8.18590913e-03 1.26595765e-01 5.11789203e-01
-4.29475039e-01 -7.89799690e-01 -3.55546206e-01 8.44932675e-01
-7.82986224e-01 -6.45351171e-01 -1.16797134e-01 2.17533529e-01
2.71121144e-01 9.89720643e-01 6.58567175e-02 -6.33657157e-01
3.61625344e-01 -6.01122677e-02 4.44290638e-01 -1.19737685e-02
-5.37560642e-01 -3.50163400e-01 -5.06443799e-01 -8.09091389e-01
-3.83675963e-01 -4.55979645e-01 -8.72570634e-01 -3.81408721e-01
1.68757424e-01 -1.93884194e-01 7.65729487e-01 7.55565763e-01
2.64550298e-01 5.47143161e-01 6.02278888e-01 -7.73898780e-01
-2.26227045e-01 -1.27712369e+00 -6.17022574e-01 2.77062058e-01
7.60225654e-01 -3.33119571e-01 -4.92414922e-01 -1.05246365e-01]
|
[11.870079040527344, -1.8123410940170288]
|
0f262caa-fa31-4d42-84a9-6a49acdc6cbc
|
learning-with-neighbor-consistency-for-noisy-1
|
2202.02200
| null |
https://arxiv.org/abs/2202.02200v2
|
https://arxiv.org/pdf/2202.02200v2.pdf
|
Learning with Neighbor Consistency for Noisy Labels
|
Recent advances in deep learning have relied on large, labelled datasets to train high-capacity models. However, collecting large datasets in a time- and cost-efficient manner often results in label noise. We present a method for learning from noisy labels that leverages similarities between training examples in feature space, encouraging the prediction of each example to be similar to its nearest neighbours. Compared to training algorithms that use multiple models or distinct stages, our approach takes the form of a simple, additional regularization term. It can be interpreted as an inductive version of the classical, transductive label propagation algorithm. We thoroughly evaluate our method on datasets evaluating both synthetic (CIFAR-10, CIFAR-100) and realistic (mini-WebVision, WebVision, Clothing1M, mini-ImageNet-Red) noise, and achieve competitive or state-of-the-art accuracies across all of them.
|
['Cordelia Schmid', 'Anurag Arnab', 'Jack Valmadre', 'Ahmet Iscen']
|
2022-02-04
|
learning-with-neighbor-consistency-for-noisy
|
http://openaccess.thecvf.com//content/CVPR2022/html/Iscen_Learning_With_Neighbor_Consistency_for_Noisy_Labels_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Iscen_Learning_With_Neighbor_Consistency_for_Noisy_Labels_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['learning-with-noisy-labels', 'learning-with-noisy-labels']
|
['computer-vision', 'natural-language-processing']
|
[ 2.88801491e-01 4.03441153e-02 -1.97347393e-03 -8.58886480e-01
-1.22883010e+00 -5.93527019e-01 7.58773446e-01 1.91552863e-01
-7.67052233e-01 8.11412871e-01 9.48711578e-03 -1.81876160e-02
4.75730039e-02 -6.87898040e-01 -9.17029083e-01 -6.86388075e-01
8.96551460e-02 6.03271008e-01 2.56832600e-01 5.39078303e-02
-6.01954609e-02 1.44133836e-01 -1.44122005e+00 6.31572604e-01
3.89739394e-01 1.29085433e+00 -2.21218616e-01 4.61192787e-01
1.01206183e-01 1.30205560e+00 -3.54488075e-01 -6.31233633e-01
1.80335864e-01 -1.81273311e-01 -1.19936645e+00 9.64971557e-02
6.99592233e-01 -3.77933234e-02 -4.32677455e-02 1.03248644e+00
5.17983317e-01 1.86811388e-01 8.20702970e-01 -1.02658165e+00
-7.06624508e-01 6.10670328e-01 -3.14683229e-01 -1.76656306e-01
5.12400009e-02 9.34463516e-02 9.60723698e-01 -9.66142595e-01
7.88987637e-01 1.21936142e+00 1.25146699e+00 6.86643541e-01
-1.87802231e+00 -6.18757129e-01 1.05497167e-01 -3.93543094e-02
-1.20532227e+00 -3.34957898e-01 4.94936705e-01 -4.47160244e-01
7.38787830e-01 1.56974778e-01 4.25407141e-01 1.63500202e+00
-3.64048034e-01 8.53934884e-01 1.45940852e+00 -5.55531621e-01
4.86303031e-01 3.18274081e-01 3.67097557e-01 6.20320916e-01
-9.50121582e-02 3.14025074e-01 -3.93533677e-01 -2.44544163e-01
3.49481851e-01 -1.63038313e-01 -9.51312296e-03 -3.38951647e-01
-8.86005163e-01 9.62243140e-01 7.44191825e-01 6.99169561e-02
-1.91473469e-01 4.94918793e-01 4.79817748e-01 4.16955680e-01
8.62091899e-01 4.11072075e-01 -5.73669195e-01 -2.51254588e-02
-8.30450892e-01 1.70600981e-01 9.49627578e-01 8.25336695e-01
1.08856034e+00 -1.28290325e-01 -1.84869424e-01 1.12171948e+00
4.50218976e-01 1.80909470e-01 4.02358174e-01 -1.15384233e+00
1.26341552e-01 4.06767577e-01 6.49321303e-02 -7.22979903e-01
-5.38113296e-01 -6.56178892e-01 -8.82792234e-01 4.45450336e-01
3.73974949e-01 -1.70005441e-01 -1.34660172e+00 1.73681045e+00
1.39109686e-01 4.32496011e-01 -2.35014223e-03 9.36700463e-01
9.46687222e-01 4.51165676e-01 6.71762109e-01 8.52041170e-02
7.96409428e-01 -1.31574297e+00 -2.51850218e-01 -4.29582566e-01
1.04295683e+00 -6.47164941e-01 1.01445174e+00 5.51393449e-01
-7.62759149e-01 -6.35251045e-01 -6.59772754e-01 -2.13038981e-01
-6.03712797e-01 1.99343101e-03 6.83180928e-01 5.45812070e-01
-1.13053620e+00 8.33806515e-01 -4.39797848e-01 -4.06870484e-01
7.20000863e-01 3.79048496e-01 -4.57649022e-01 -4.50040340e-01
-1.13784468e+00 8.52340162e-01 3.16280663e-01 3.11670601e-02
-1.33258307e+00 -6.06414974e-01 -6.75274551e-01 -2.00628519e-01
2.08574548e-01 -4.60402429e-01 1.48216355e+00 -1.39058208e+00
-1.22887027e+00 1.35285556e+00 2.11313397e-01 -5.00678182e-01
5.90930879e-01 -3.85368347e-01 -2.82677501e-01 -2.01375052e-01
-4.77672257e-02 1.01698279e+00 6.23888969e-01 -1.84550309e+00
-5.92034459e-01 -1.22661337e-01 8.94633774e-03 -5.71193583e-02
-2.53908277e-01 -3.54340672e-02 -1.10549122e-01 -5.58407009e-01
4.28408235e-02 -1.11785972e+00 -5.12269080e-01 -6.40810579e-02
-6.56350136e-01 -3.73456866e-01 5.09515405e-01 -2.32754856e-01
7.52347827e-01 -1.94165075e+00 -9.74650607e-02 2.56527364e-01
2.07388490e-01 5.55740178e-01 -2.75448859e-01 2.96222150e-01
-4.00556475e-02 3.10775727e-01 -2.71313190e-01 -6.57446563e-01
4.31087129e-02 4.31902051e-01 -1.01517275e-01 4.07313704e-01
2.70640552e-01 9.64913368e-01 -1.16805291e+00 -4.35640514e-01
1.01988599e-01 5.13081908e-01 -3.16321999e-01 5.46995699e-02
-3.28197956e-01 3.33211213e-01 -1.83021799e-01 6.12831712e-01
4.51115847e-01 -5.47799110e-01 -6.55541122e-02 -1.41630292e-01
2.31821403e-01 1.01770103e-01 -9.21726465e-01 1.75414598e+00
-4.91949260e-01 5.27939379e-01 -2.20036611e-01 -1.13971949e+00
7.56310165e-01 3.31881523e-01 2.25357905e-01 -6.52767539e-01
1.24003537e-01 3.36483806e-01 -4.78591532e-01 -4.62666154e-01
1.95885096e-02 -2.00656638e-01 -1.54214334e-02 4.24119979e-01
5.24938285e-01 1.06163554e-01 9.27737951e-02 8.95020142e-02
1.21715331e+00 1.82357997e-01 -1.18385412e-01 9.15198214e-03
2.54304796e-01 -4.62831790e-03 4.25629556e-01 1.25436950e+00
-3.17406923e-01 8.35200191e-01 2.43425205e-01 -7.46872783e-01
-9.38759089e-01 -1.07578957e+00 -1.88120753e-01 1.38784277e+00
-2.02664182e-01 -3.00082892e-01 -6.34062409e-01 -1.19919860e+00
-8.07656534e-03 7.60055304e-01 -9.04119611e-01 -2.51685917e-01
-2.54736036e-01 -8.90943646e-01 7.01122105e-01 5.66733778e-01
3.24498147e-01 -1.30893028e+00 -6.64610937e-02 1.93219334e-01
-9.90199223e-02 -9.83986676e-01 9.37267095e-02 7.60521948e-01
-7.45493412e-01 -8.78853202e-01 -5.00404000e-01 -1.11813867e+00
7.33302832e-01 -1.86405867e-01 1.61867166e+00 4.25637327e-02
-2.58415699e-01 2.77383685e-01 -4.48640525e-01 -3.18749577e-01
-3.57346773e-01 9.85273123e-02 -1.90573052e-01 8.41661245e-02
5.51235080e-01 -3.39426041e-01 -4.45839763e-01 1.88092723e-01
-8.10991049e-01 -1.83676966e-02 6.16394103e-01 1.09364331e+00
7.83507466e-01 -1.64143503e-01 6.79707348e-01 -1.47764480e+00
4.84213263e-01 -6.47275388e-01 -3.37434679e-01 2.86554605e-01
-7.00131834e-01 -4.82188687e-02 7.95310199e-01 -6.03290379e-01
-9.73956466e-01 4.87810969e-01 -3.39593530e-01 -3.48230004e-01
-6.47025287e-01 4.25171107e-01 2.44007260e-01 -2.98385888e-01
1.09114695e+00 -5.76984920e-02 -5.72869301e-01 -7.00332999e-01
7.32502937e-01 6.49860322e-01 5.10474324e-01 -6.28975570e-01
6.68710887e-01 5.49174666e-01 -1.53759599e-01 -3.15866947e-01
-1.50644767e+00 -5.26979327e-01 -6.97255433e-01 -3.06937426e-01
6.83802128e-01 -8.26386034e-01 -4.29148436e-01 7.17964172e-01
-9.71225381e-01 -6.17187858e-01 -5.48256874e-01 4.02829528e-01
-5.31598330e-01 -1.40481308e-01 -9.48433459e-01 -5.94846368e-01
-1.82149231e-01 -1.02310729e+00 1.03317356e+00 1.20419122e-01
-1.39965102e-01 -1.24710393e+00 3.75085860e-01 2.75963664e-01
4.60851252e-01 3.48807275e-01 7.20383346e-01 -1.02062654e+00
-2.63819128e-01 -4.35184807e-01 -4.67596591e-01 8.20724666e-01
-1.77446425e-01 -2.10734352e-01 -1.55431581e+00 -1.31659642e-01
-3.01063359e-01 -1.45005071e+00 1.28014243e+00 1.48254305e-01
1.22842824e+00 -2.16494247e-01 -3.18351895e-01 5.68479776e-01
1.61772430e+00 -3.31711441e-01 5.65976977e-01 4.36889857e-01
7.92868555e-01 6.36700332e-01 3.55716527e-01 7.17452243e-02
2.82475203e-01 4.24611211e-01 4.46646839e-01 -4.93257910e-01
-1.07413836e-01 -3.33316363e-02 2.20433641e-02 6.13206327e-01
4.59024385e-02 -1.45373732e-01 -9.44146037e-01 5.07594526e-01
-1.89856935e+00 -7.06866741e-01 -2.37565055e-01 2.12263703e+00
1.10341573e+00 2.12999985e-01 -1.21346593e-01 1.13743000e-01
4.26770359e-01 -6.53531104e-02 -5.55022657e-01 -3.06628853e-01
-1.82293177e-01 2.35529110e-01 4.81023997e-01 3.07478130e-01
-1.54998982e+00 1.11624908e+00 7.11310244e+00 1.08634388e+00
-1.06306207e+00 3.55222017e-01 1.23981631e+00 -1.55968428e-01
-2.14733165e-02 -1.32178098e-01 -6.07394516e-01 3.76873434e-01
1.15739954e+00 4.92064238e-01 4.21246946e-01 9.61460471e-01
-1.76766649e-01 -1.28018543e-01 -1.32520020e+00 8.62254679e-01
-7.45653808e-02 -1.17911208e+00 -2.83185333e-01 -1.92248031e-01
1.04253721e+00 5.71071148e-01 1.47926539e-01 5.42306960e-01
9.46219742e-01 -1.21800518e+00 6.94230676e-01 5.59370995e-01
7.96204150e-01 -6.78914189e-01 9.86974001e-01 3.40327859e-01
-5.49078941e-01 -1.26792997e-01 -5.09619832e-01 5.49104996e-02
-1.14493314e-02 8.79192173e-01 -5.48639357e-01 1.66996494e-01
9.63102579e-01 6.90668464e-01 -8.33107352e-01 1.10029078e+00
-3.83309603e-01 1.07295418e+00 -4.00185049e-01 2.12452784e-01
5.42229533e-01 1.12209640e-01 -1.54064551e-01 1.50723541e+00
-1.69359758e-01 -8.97662342e-02 4.92854059e-01 6.69422090e-01
-4.97616410e-01 5.56094684e-02 -5.64245462e-01 1.92339912e-01
1.85856581e-01 1.54734397e+00 -6.67804956e-01 -3.85867745e-01
-6.35106087e-01 8.80929649e-01 8.71760249e-01 4.29558873e-01
-6.96123064e-01 1.89969614e-02 2.52948433e-01 -5.87402396e-02
1.11864895e-01 3.34056616e-01 -3.56300354e-01 -7.77943254e-01
-2.42896184e-01 -7.25505888e-01 2.34416917e-01 -7.82095850e-01
-1.83732414e+00 7.65150368e-01 -4.72336948e-01 -1.03648424e+00
-2.98846327e-02 -4.80647534e-01 -2.36457780e-01 6.04566872e-01
-1.61583221e+00 -1.50718081e+00 -3.31607789e-01 5.59375346e-01
2.63411045e-01 1.06385529e-01 1.27750611e+00 4.70902860e-01
-3.73038918e-01 5.74263573e-01 4.51444000e-01 3.21818888e-01
8.86287451e-01 -1.44035375e+00 2.89122105e-01 2.61976421e-01
4.63804215e-01 1.55126974e-01 4.25350189e-01 -4.47719663e-01
-4.41988170e-01 -1.52239394e+00 8.76660943e-01 -5.66666007e-01
6.96790576e-01 -5.97757399e-01 -9.52575684e-01 8.57768416e-01
9.43375155e-02 6.16455913e-01 8.19420934e-01 3.44081461e-01
-6.82597160e-01 -9.08738226e-02 -1.24621308e+00 2.51634002e-01
1.16307414e+00 -5.22595048e-01 -2.31660843e-01 8.67053032e-01
5.50557852e-01 -8.40683207e-02 -8.68690133e-01 4.56309557e-01
4.50507849e-01 -1.05467939e+00 1.01462162e+00 -9.33703423e-01
4.63488519e-01 4.40819003e-02 -2.11344227e-01 -1.54143000e+00
-6.85981810e-01 -2.38750458e-01 6.02346994e-02 1.29315221e+00
6.49281085e-01 -2.83033699e-01 8.97289038e-01 6.42694175e-01
-6.35139495e-02 -8.45544279e-01 -7.71795154e-01 -7.75698245e-01
1.27205640e-01 -6.29358828e-01 -2.07146965e-02 1.11254442e+00
-4.24769849e-01 4.21510398e-01 -5.73843360e-01 -1.01283669e-01
7.92133570e-01 -4.82396781e-01 3.51720542e-01 -1.57495081e+00
-9.48999748e-02 -1.10652946e-01 -3.83818567e-01 -7.13596761e-01
3.22264642e-01 -9.63062346e-01 3.36633801e-01 -1.56550407e+00
2.48272747e-01 -1.00154030e+00 -7.03997791e-01 8.25900435e-01
5.72467260e-02 9.80002999e-01 -1.70794442e-01 4.67685908e-01
-1.19602048e+00 2.92397082e-01 8.65048289e-01 -1.35043129e-01
2.04897270e-01 -9.66944639e-03 -4.56174970e-01 1.00788283e+00
6.96620882e-01 -9.95432556e-01 -2.09630206e-01 -5.45302331e-01
3.88471127e-01 -5.86104333e-01 5.24825990e-01 -1.27389777e+00
2.28615955e-01 5.47138639e-02 5.30010879e-01 -2.26962447e-01
4.24255162e-01 -1.00138617e+00 -1.93610676e-02 2.45741844e-01
-9.90264237e-01 -4.22101766e-01 -1.15558878e-01 7.76506186e-01
-2.08931476e-01 -5.36732435e-01 1.06727517e+00 -3.83010864e-01
-8.45704377e-01 1.22035913e-01 -3.92747462e-01 2.00970545e-01
9.14245605e-01 1.59824908e-01 -4.65782464e-01 -2.62397647e-01
-8.78132045e-01 1.21862531e-01 3.17915231e-01 3.86341006e-01
4.35077846e-01 -1.34398627e+00 -7.79347062e-01 5.38669750e-02
3.17679226e-01 8.01238567e-02 -1.67440791e-02 6.58286631e-01
-4.81633723e-01 2.77521104e-01 1.30105078e-01 -7.06439435e-01
-9.82037663e-01 6.09522700e-01 4.44629401e-01 -6.18051052e-01
-3.53114605e-01 1.34080577e+00 1.17786638e-01 -8.77801836e-01
4.85316724e-01 4.62150574e-02 -1.12116918e-01 -6.99467212e-02
3.75973344e-01 2.99794257e-01 1.78260893e-01 -6.67705357e-01
-1.98885307e-01 4.29184079e-01 -2.04895377e-01 4.98279184e-03
1.46266544e+00 -1.45857148e-02 -5.67915216e-02 7.22569883e-01
1.43257737e+00 -4.62471277e-01 -1.35165572e+00 -6.17548525e-01
3.65519583e-01 -2.00264916e-01 1.98423371e-01 -1.35886395e+00
-1.14528584e+00 7.56390691e-01 9.63538289e-01 2.12484226e-01
9.40835238e-01 1.99460328e-01 6.44485652e-01 6.44402802e-01
4.40062970e-01 -1.29640186e+00 2.12278500e-01 5.89571834e-01
5.22358418e-01 -1.75967205e+00 -3.02714825e-01 -2.14406177e-01
-5.88595331e-01 8.35215747e-01 6.99696839e-01 -2.27455139e-01
9.15172040e-01 2.49270633e-01 5.96414626e-01 -3.06423426e-01
-8.47883105e-01 -2.44906411e-01 1.57100901e-01 7.36733019e-01
4.79169130e-01 -8.75584781e-02 8.26070979e-02 4.97670054e-01
2.61721730e-01 2.35863417e-01 7.85112903e-02 9.05081749e-01
-5.37671745e-01 -9.91034627e-01 -1.97476879e-01 6.09982908e-01
-6.09837770e-01 -2.71300286e-01 -4.34109122e-01 5.90692163e-01
4.88500476e-01 1.15079165e+00 -7.44849592e-02 -5.04940093e-01
3.03629369e-01 2.44749889e-01 2.58805335e-01 -7.07567692e-01
-8.82645667e-01 7.94838089e-03 3.20886075e-01 -5.76896846e-01
-6.54358745e-01 -2.39770755e-01 -1.05547023e+00 -1.16008162e-01
-4.68015552e-01 -5.22598065e-03 7.01029778e-01 9.76234078e-01
7.82271251e-02 2.62144834e-01 4.77891892e-01 -8.60560596e-01
-5.23670733e-01 -1.17461288e+00 -5.92020869e-01 1.01363707e+00
6.92121983e-02 -6.37658775e-01 -4.87364143e-01 2.28940547e-01]
|
[9.409334182739258, 3.801140785217285]
|
0f9d419b-9e00-4e6b-a9d8-4ee13feb9810
|
asymptotically-unbiased-off-policy-policy
|
2302.11725
| null |
https://arxiv.org/abs/2302.11725v1
|
https://arxiv.org/pdf/2302.11725v1.pdf
|
Asymptotically Unbiased Off-Policy Policy Evaluation when Reusing Old Data in Nonstationary Environments
|
In this work, we consider the off-policy policy evaluation problem for contextual bandits and finite horizon reinforcement learning in the nonstationary setting. Reusing old data is critical for policy evaluation, but existing estimators that reuse old data introduce large bias such that we can not obtain a valid confidence interval. Inspired from a related field called survey sampling, we introduce a variant of the doubly robust (DR) estimator, called the regression-assisted DR estimator, that can incorporate the past data without introducing a large bias. The estimator unifies several existing off-policy policy evaluation methods and improves on them with the use of auxiliary information and a regression approach. We prove that the new estimator is asymptotically unbiased, and provide a consistent variance estimator to a construct a large sample confidence interval. Finally, we empirically show that the new estimator improves estimation for the current and future policy values, and provides a tight and valid interval estimation in several nonstationary recommendation environments.
|
['Martha White', 'Philip Thomas', 'Yash Chandak', 'Vincent Liu']
|
2023-02-23
| null | null | null | null |
['multi-armed-bandits']
|
['miscellaneous']
|
[ 8.45679939e-02 -1.88923642e-01 -1.14749193e+00 -2.46115372e-01
-1.11895096e+00 -7.21527815e-01 3.20950180e-01 3.39521058e-02
-4.36542511e-01 1.65790975e+00 2.34328076e-01 -8.96664321e-01
-3.68508220e-01 -7.64432907e-01 -1.06025636e+00 -6.76404715e-01
-1.02445096e-01 3.70986551e-01 1.00543343e-01 1.85548827e-01
1.48007154e-01 8.06411207e-02 -1.30402601e+00 -4.35426861e-01
1.15592957e+00 1.16953564e+00 -8.40642378e-02 7.13428259e-01
3.22765112e-01 6.48504257e-01 -7.47831345e-01 -1.60064399e-01
2.19771937e-01 -4.52132910e-01 -3.94697815e-01 -1.99939713e-01
1.86642051e-01 -9.08354461e-01 2.00262502e-01 1.07055962e+00
4.55176622e-01 5.24157226e-01 5.47940433e-01 -1.07008314e+00
-6.04901254e-01 9.77500021e-01 -7.79748380e-01 3.51723820e-01
2.80916095e-01 -3.14171761e-01 9.17823374e-01 -1.86020777e-01
1.94120213e-01 1.46200562e+00 8.00857544e-01 3.54770422e-01
-1.28884947e+00 -8.22405815e-01 6.37921870e-01 -1.08065881e-01
-5.80029488e-01 -2.49052405e-01 4.62511152e-01 -1.38143107e-01
1.99565843e-01 3.29926819e-01 8.52656662e-01 1.61792123e+00
-1.45279452e-01 9.88343775e-01 1.48364747e+00 -5.90198517e-01
8.25823009e-01 8.13192725e-02 4.77757066e-01 2.23128319e-01
8.85152638e-01 7.91063964e-01 -3.96405905e-02 -7.32815564e-01
9.51576412e-01 3.15734982e-01 -2.15941504e-01 -4.41381931e-01
-8.54088485e-01 1.02431643e+00 -3.91644239e-01 -2.08282426e-01
-6.33520663e-01 2.61860579e-01 2.41112217e-01 4.58581924e-01
8.33054304e-01 2.74972375e-02 -5.52133977e-01 -4.34185475e-01
-8.39933217e-01 4.12965417e-01 1.07436323e+00 9.19860661e-01
3.76119286e-01 2.52706140e-01 -8.25439394e-01 6.94063485e-01
-1.33502362e-02 1.16532075e+00 2.50315130e-01 -1.33625376e+00
3.57830197e-01 -4.41010475e-01 1.09102333e+00 -1.99211910e-01
-5.13191558e-02 -7.49394119e-01 -5.22973061e-01 -4.69418755e-03
6.66758597e-01 -7.50697434e-01 -6.33855641e-01 2.09696054e+00
5.29881775e-01 2.93810159e-01 -2.97338534e-02 5.18304586e-01
-3.21301401e-01 4.02514100e-01 -9.65612754e-02 -9.50638771e-01
8.73244822e-01 -7.96577632e-01 -9.89643991e-01 -1.03014037e-02
1.21109009e-01 -3.36675078e-01 9.61351871e-01 6.21795952e-01
-1.16882324e+00 -1.97043613e-01 -9.51379776e-01 7.22410381e-01
-3.70793939e-02 -1.87345028e-01 6.53820097e-01 9.29484427e-01
-6.33519828e-01 6.76763833e-01 -6.66092813e-01 -1.29517734e-01
3.14318597e-01 -9.33749676e-02 5.25197566e-01 1.21509656e-02
-1.16062963e+00 6.75132275e-01 1.37796268e-01 -3.26700032e-01
-1.00505221e+00 -7.63391495e-01 -5.94281614e-01 4.73598838e-02
1.07863081e+00 -6.95150793e-01 1.92880380e+00 -1.05129230e+00
-2.01280141e+00 -5.15939742e-02 -1.68859527e-01 -9.09034908e-01
7.32034504e-01 -5.59725106e-01 -4.42667812e-01 -5.23054302e-02
1.12263501e-01 -3.61386031e-01 1.16583323e+00 -9.88500118e-01
-9.68011856e-01 -4.47551012e-01 2.06327483e-01 2.55567506e-02
-1.04983598e-01 -2.90257990e-01 1.05320670e-01 -1.01160657e+00
-5.39011121e-01 -1.00434053e+00 -4.33401257e-01 -5.85791588e-01
-1.29595697e-01 -1.50305718e-01 4.71608073e-01 -5.16808867e-01
1.42907906e+00 -1.69563115e+00 -4.19529796e-01 4.02850956e-01
-3.27369362e-01 -4.19830494e-02 2.24842429e-01 2.17277110e-01
1.89065814e-01 3.27579640e-02 -1.70389295e-01 1.75870016e-01
1.04199350e-01 2.96551138e-01 -6.78503573e-01 6.00325823e-01
-6.33341610e-01 6.08614624e-01 -1.16908646e+00 -8.31462368e-02
1.14330150e-01 -3.11072797e-01 -6.38877094e-01 1.51534140e-01
-4.71972615e-01 4.41594929e-01 -5.96931994e-01 4.04066294e-01
5.76643109e-01 -3.07918817e-01 3.47435802e-01 4.24705029e-01
-9.60464478e-02 6.10099733e-03 -1.18516672e+00 1.14111292e+00
-6.15619957e-01 5.06292284e-02 -2.75829155e-02 -1.43284452e+00
6.62604630e-01 3.11418563e-01 6.63672447e-01 -3.57020855e-01
6.31969199e-02 2.19566151e-01 -5.15198708e-01 -2.60543138e-01
3.88338804e-01 -2.56369263e-01 -1.28226891e-01 7.22767532e-01
-1.64686814e-01 4.44161415e-01 1.17570087e-02 -2.41267309e-01
8.14276993e-01 3.15005273e-01 8.85228455e-01 -4.06715184e-01
1.70932189e-01 -4.85858828e-01 6.70398951e-01 1.64213037e+00
-2.91563511e-01 -1.60829857e-01 4.91366982e-01 -1.33477852e-01
-7.44576871e-01 -1.24818993e+00 -1.12262405e-01 1.48503351e+00
-2.30904277e-02 2.90035427e-01 -3.85609150e-01 -8.45689833e-01
6.34761333e-01 9.36777055e-01 -7.58348286e-01 -1.57080069e-01
-5.32143563e-02 -7.03112662e-01 4.09339406e-02 5.92304170e-01
3.63414139e-01 -4.62223053e-01 -4.04342324e-01 5.42133868e-01
-1.35293856e-01 -7.68731415e-01 -6.15814447e-01 9.14135575e-03
-9.99965250e-01 -1.03644764e+00 -1.15980589e+00 1.00497425e-01
1.39748916e-01 4.21150535e-01 9.85597908e-01 -6.43072188e-01
7.65471756e-01 8.37638557e-01 -2.83914477e-01 -8.69124591e-01
-3.69159847e-01 -1.80881843e-01 3.04246098e-01 2.96655744e-02
1.51067302e-01 -5.43495715e-01 -7.88728237e-01 3.74950886e-01
-5.88732541e-01 -5.16900778e-01 2.66839117e-01 1.15253890e+00
5.92490852e-01 -4.62452978e-01 1.30914867e+00 -1.00333273e+00
8.67445648e-01 -6.93033874e-01 -1.25533330e+00 4.54579353e-01
-1.12103426e+00 4.07587528e-01 5.48772156e-01 -8.22095692e-01
-1.33402932e+00 -5.51767051e-01 2.81150311e-01 -3.53034258e-01
1.49806097e-01 4.58378702e-01 2.55810231e-01 3.98476124e-01
5.02932847e-01 -6.37680069e-02 1.21149771e-01 -6.66699111e-01
4.00664836e-01 6.93094075e-01 4.29979682e-01 -1.14202583e+00
2.33266070e-01 5.97384453e-01 -1.01864576e-01 -3.62016559e-01
-1.43020535e+00 -4.98199016e-01 1.87071010e-01 -7.88614154e-02
1.99703336e-01 -7.54136801e-01 -9.92161453e-01 -6.92024603e-02
-5.37931979e-01 -6.92443669e-01 -6.97356462e-01 1.10134101e+00
-1.10261953e+00 3.56986195e-01 -2.49588847e-01 -1.65744650e+00
-2.37577617e-01 -7.18744457e-01 7.85228968e-01 1.56431898e-01
3.05738300e-02 -1.00633562e+00 5.07344067e-01 -1.47039816e-01
2.63168424e-01 2.99343914e-01 5.46358109e-01 -5.44634938e-01
-6.07835948e-02 -1.02940798e-01 -1.19765280e-02 3.99811506e-01
5.65399043e-02 -1.74810126e-01 -6.45842671e-01 -7.06971645e-01
-7.04459697e-02 -4.68527675e-02 9.77337360e-01 1.14696550e+00
1.47822154e+00 -7.23723233e-01 -2.07974181e-01 3.28709692e-01
1.18698347e+00 4.34942037e-01 2.11551473e-01 4.32305127e-01
-1.55186400e-01 4.89046276e-02 9.51625645e-01 1.07644045e+00
1.97372973e-01 4.10748959e-01 1.08383209e-01 2.51907915e-01
5.35392940e-01 -5.11588871e-01 5.69690228e-01 1.57396406e-01
-3.32124740e-01 -1.77059993e-01 -2.34068379e-01 5.52292883e-01
-2.33055210e+00 -1.32316601e+00 4.40052003e-01 3.07506466e+00
9.92378712e-01 9.82795730e-02 7.50538588e-01 -1.51971295e-01
9.63477671e-01 -1.85329262e-02 -1.13401258e+00 -4.69676346e-01
5.69831319e-02 4.58398551e-01 1.11408627e+00 4.82810646e-01
-9.03375208e-01 4.68217582e-01 7.49255228e+00 7.66328573e-01
-6.92185104e-01 2.69145370e-01 5.54375112e-01 -1.83484256e-01
-2.69419342e-01 1.49065778e-01 -8.23837757e-01 6.27547622e-01
1.38798296e+00 -5.27635455e-01 5.63501656e-01 1.11807859e+00
5.19478202e-01 -4.46177393e-01 -8.35494041e-01 6.91458046e-01
-4.45968360e-01 -1.14281964e+00 -3.05570990e-01 2.69134730e-01
1.04121101e+00 -9.68295559e-02 1.94415495e-01 6.67867839e-01
1.11615884e+00 -4.27242309e-01 7.24871159e-01 7.12301254e-01
8.59219611e-01 -9.96252298e-01 5.69751978e-01 3.43275219e-01
-6.15699351e-01 -5.46235859e-01 -5.00600815e-01 -1.81785166e-01
4.71022613e-02 8.32520783e-01 -3.88842344e-01 4.55167443e-01
4.58420545e-01 7.02207088e-01 6.91392720e-02 1.16965783e+00
-1.30441412e-01 1.08910084e+00 -3.25843483e-01 -1.97910726e-01
2.25496665e-01 -3.69711012e-01 7.75011063e-01 7.77320862e-01
7.41120815e-01 -2.08214656e-01 2.95102388e-01 4.75418091e-01
6.85075894e-02 -3.64083238e-02 -5.47842324e-01 3.90629739e-01
7.60796309e-01 5.43692231e-01 -1.01200052e-01 -6.49861336e-01
-5.04907727e-01 4.15637523e-01 9.53955352e-02 8.44305277e-01
-1.02885687e+00 7.22061619e-02 6.31468952e-01 -2.81379223e-01
7.23898411e-01 1.71370342e-01 6.46476969e-02 -1.22828639e+00
-1.42928779e-01 -9.36288774e-01 7.77245820e-01 -2.72167295e-01
-1.30464935e+00 -3.33613485e-01 4.67898816e-01 -1.15059185e+00
-7.00865984e-01 -1.84536591e-01 -3.44946235e-01 4.82163042e-01
-1.56431437e+00 -4.58369732e-01 3.80443573e-01 5.18033862e-01
3.10561776e-01 1.76723879e-02 6.62892640e-01 -1.61024734e-01
-7.42058814e-01 6.15379989e-01 1.14473140e+00 -4.61439848e-01
8.43041122e-01 -1.49832571e+00 -1.51724875e-01 6.81869984e-01
-3.12846869e-01 5.29879630e-01 8.63909423e-01 -6.45288408e-01
-1.12227714e+00 -1.11920166e+00 -2.48276189e-01 -1.27328068e-01
7.67630160e-01 2.11400315e-01 -4.50700313e-01 8.60294282e-01
-1.32333398e-01 6.29785657e-02 7.56557643e-01 6.65437043e-01
-3.31961960e-01 -3.76635760e-01 -1.14973080e+00 5.64079642e-01
9.53846753e-01 -1.45795509e-01 -6.36707067e-01 1.91413417e-01
7.85523415e-01 -1.47356555e-01 -9.06629086e-01 4.03750300e-01
1.23894727e+00 -8.26062024e-01 8.97634208e-01 -1.08589923e+00
-2.30550528e-01 4.25695539e-01 -2.12294087e-01 -1.48343611e+00
-2.17971385e-01 -1.33454049e+00 -8.84128511e-01 8.64969730e-01
3.33364099e-01 -1.18382299e+00 5.45859635e-01 3.07509452e-01
2.92140812e-01 -4.18627501e-01 -9.69872534e-01 -1.56782246e+00
2.38382936e-01 -5.60553133e-01 1.02267611e+00 6.60384238e-01
7.61558414e-02 9.89318267e-02 -8.62702191e-01 5.45282364e-02
1.08256233e+00 5.66465616e-01 7.42334664e-01 -1.22146881e+00
-7.27633774e-01 -3.76845390e-01 5.02510250e-01 -1.41713583e+00
2.50377059e-01 -1.12731010e-01 3.34133953e-03 -9.72733557e-01
3.41374665e-01 -3.84108514e-01 -8.64031136e-01 -5.50253205e-02
-3.87908936e-01 -4.55435395e-01 -2.63478786e-01 -1.69240043e-01
-7.88562953e-01 6.44740760e-01 1.17710865e+00 1.48052856e-01
-3.02295655e-01 8.93580914e-01 -9.47850883e-01 6.46532238e-01
9.32582855e-01 -5.30529082e-01 -5.44152796e-01 3.01719069e-01
7.88826197e-02 6.53256834e-01 2.65481293e-01 -5.87453306e-01
-3.40682536e-01 -6.77303493e-01 2.10143924e-01 -7.28537381e-01
-1.59961864e-01 -6.56781316e-01 -5.08271195e-02 4.32022691e-01
-5.54199159e-01 -1.24002539e-01 -4.22244280e-04 1.46561563e+00
3.42341632e-01 -2.31750369e-01 7.58622408e-01 -1.76602721e-01
-1.07892871e-01 4.90902573e-01 -3.87403160e-01 3.15363795e-01
8.60755920e-01 8.39658678e-02 -3.28808874e-01 -1.06406319e+00
-6.73474967e-01 3.81732076e-01 9.00340751e-02 6.56573921e-02
2.16702282e-01 -1.39845228e+00 -5.14482021e-01 -1.98337093e-01
-1.56775817e-01 -6.02298737e-01 -6.87634423e-02 8.70883226e-01
4.97590780e-01 5.64474285e-01 2.66503423e-01 -3.75619024e-01
-7.97302902e-01 8.07887614e-01 2.30338216e-01 -5.89810789e-01
-2.94850618e-01 1.03991486e-01 -3.58815081e-02 -2.70327684e-02
4.13342088e-01 -6.51220024e-01 -4.19421494e-02 4.12865123e-03
7.99979031e-01 8.67388427e-01 -3.40784639e-01 2.88578480e-01
1.51886567e-01 8.52117613e-02 1.24486625e-01 -5.65571070e-01
1.20483959e+00 -5.35408676e-01 3.25576991e-01 7.92162836e-01
7.50380695e-01 1.17753066e-01 -1.54654169e+00 -6.68530762e-01
1.52064770e-01 -5.74370801e-01 1.62737593e-01 -6.86725676e-01
-6.26273096e-01 2.93567300e-01 5.20331800e-01 6.47667766e-01
8.54449987e-01 -3.62680048e-01 6.50217712e-01 4.83109206e-01
5.90077877e-01 -1.44098914e+00 -3.30907375e-01 3.66192579e-01
6.06992304e-01 -1.27035677e+00 2.66407937e-01 2.32251227e-01
-4.00347352e-01 8.34091961e-01 9.14713964e-02 -5.90195656e-02
8.16907585e-01 -2.82906312e-02 -5.12605488e-01 4.83515501e-01
-8.05251420e-01 -5.18364847e-01 4.74382080e-02 8.10601532e-01
4.66002710e-02 5.49653113e-01 -5.95707953e-01 7.92589903e-01
-3.42141166e-02 4.08189446e-01 6.12557471e-01 8.69218647e-01
-5.82293749e-01 -9.08865452e-01 -6.10822737e-01 8.43970954e-01
-8.62438917e-01 2.06536651e-01 2.57813036e-01 6.41095340e-01
-6.88193738e-01 1.11526799e+00 8.95864144e-03 2.62049794e-01
1.94972351e-01 4.10578437e-02 6.73764408e-01 -2.71993905e-01
1.66975185e-01 3.82850647e-01 3.27395678e-01 -4.78122324e-01
-6.31998777e-01 -7.98579991e-01 -3.19671988e-01 -4.67693478e-01
-6.57433569e-01 5.44983983e-01 2.51442134e-01 9.85506237e-01
1.04900390e-01 3.59143972e-01 1.10922730e+00 -5.00177085e-01
-1.65535986e+00 -1.08828020e+00 -8.82423222e-01 -5.17534763e-02
8.80366325e-01 -1.08788550e+00 -4.09531444e-01 -4.62931961e-01]
|
[4.464090347290039, 3.1358742713928223]
|
b9acc6b9-ea6a-4c66-8257-5310b5d4d9a0
|
divergence-based-quadrangle-and-applications
|
2306.16525
| null |
https://arxiv.org/abs/2306.16525v1
|
https://arxiv.org/pdf/2306.16525v1.pdf
|
Divergence Based Quadrangle and Applications
|
This paper introduces a novel framework for assessing risk and decision-making in the presence of uncertainty, the \emph{$\varphi$-Divergence Quadrangle}. This approach expands upon the traditional Risk Quadrangle, a model that quantifies uncertainty through four key components: \emph{risk, deviation, regret}, and \emph{error}. The $\varphi$-Divergence Quadrangle incorporates the $\varphi$-divergence as a measure of the difference between probability distributions, thereby providing a more nuanced understanding of risk. Importantly, the $\varphi$-Divergence Quadrangle is closely connected with the distributionally robust optimization based on the $\varphi$-divergence approach through the duality theory of convex functionals. To illustrate its practicality and versatility, several examples of the $\varphi$-Divergence Quadrangle are provided, including the Quantile Quadrangle. The final portion of the paper outlines a case study implementing regression with the Entropic Value-at-Risk Quadrangle. The proposed $\varphi$-Divergence Quadrangle presents a refined methodology for understanding and managing risk, contributing to the ongoing development of risk assessment and management strategies.
|
['Stan Uryasev', 'Cheng Peng', 'Siddhartha Gupte', 'Anton Malandii']
|
2023-06-28
| null | null | null | null |
['management', 'decision-making']
|
['miscellaneous', 'reasoning']
|
[-1.76167339e-01 4.44144040e-01 -3.65189128e-02 -4.63986695e-01
-1.09955287e+00 -6.40455067e-01 1.44643426e-01 5.02669036e-01
-4.60626423e-01 7.17247903e-01 1.57408878e-01 -6.95180416e-01
-1.04988384e+00 -9.74116623e-01 -2.95271307e-01 -6.76539481e-01
-4.25281733e-01 7.13132992e-02 -5.08564055e-01 -1.54414937e-01
6.85176730e-01 2.78711349e-01 -1.28656638e+00 -3.53379250e-01
1.18358386e+00 1.79483294e+00 -4.52468663e-01 -3.42921205e-02
5.30575179e-02 3.12283993e-01 -5.78595817e-01 -1.02294874e+00
6.45070016e-01 -2.55269468e-01 -2.37631306e-01 -7.74240196e-01
-7.79327303e-02 -2.05821112e-01 6.45416617e-01 1.36071444e+00
2.89992452e-01 2.90090740e-01 1.12207198e+00 -1.25591218e+00
-3.77832353e-01 6.42720580e-01 -6.29798830e-01 4.59735274e-01
3.11076880e-01 -8.40682816e-03 1.44910455e+00 -7.55222678e-01
-5.76467020e-03 1.23859251e+00 5.65119207e-01 -4.95091043e-02
-1.01065922e+00 -6.28710866e-01 3.19968224e-01 -3.83021265e-01
-1.39151633e+00 4.36292142e-01 2.85605401e-01 -8.53300750e-01
5.17594516e-01 5.79819798e-01 6.01339102e-01 2.46896431e-01
8.92371774e-01 3.80941719e-01 1.23256743e+00 -2.20480189e-01
4.82800126e-01 2.26614311e-01 -1.29275203e-01 1.64119437e-01
5.14450729e-01 7.66064167e-01 -1.58946052e-01 -4.33966011e-01
4.66278702e-01 1.11305580e-01 -4.29672636e-02 -3.19438607e-01
-5.27260423e-01 1.15389264e+00 1.38829723e-01 -1.84697717e-01
-2.92927980e-01 6.02537990e-02 2.13351712e-01 3.27041209e-01
7.69024968e-01 4.00676340e-01 -1.75195903e-01 -2.48129755e-01
-8.67018044e-01 4.02511835e-01 8.39687407e-01 5.04768908e-01
2.58846611e-01 1.31798282e-01 -5.19487560e-01 4.01649773e-01
8.39311421e-01 6.52335703e-01 -4.03184086e-01 -1.01857698e+00
7.91999578e-01 4.37871724e-01 5.45468867e-01 -1.08183324e+00
-3.06420535e-01 -5.47263145e-01 -5.35132647e-01 8.15192819e-01
4.51620311e-01 -5.63470125e-01 -3.12721282e-01 1.81594801e+00
1.49046853e-01 -5.39734364e-01 -4.66634221e-02 6.73187733e-01
-1.64151773e-01 4.95699108e-01 2.05827624e-01 -7.23798275e-01
1.12312305e+00 -1.82135046e-01 -5.20541787e-01 5.47105595e-02
4.25392017e-03 -5.17871797e-01 7.92814314e-01 5.81231296e-01
-1.41765594e+00 7.57249668e-02 -8.67524147e-01 6.52216196e-01
-2.31793955e-01 -6.54239774e-01 2.31881052e-01 1.26175702e+00
-7.47910142e-01 6.43277049e-01 -2.83933133e-01 3.59254062e-01
2.67406106e-01 1.12476602e-01 1.84662223e-01 2.49836296e-01
-1.38153279e+00 1.14441907e+00 1.24352641e-01 3.09171557e-01
-6.51681364e-01 -1.14599741e+00 -8.67256522e-01 3.48329782e-01
4.35159653e-01 -3.80825639e-01 9.34450865e-01 -4.92950559e-01
-1.35958219e+00 3.78670782e-01 6.94725215e-01 -1.86234534e-01
1.04774213e+00 -2.65134811e-01 -2.21762493e-01 -2.37561136e-01
1.30580410e-01 -2.99022555e-01 5.17051876e-01 -9.91514862e-01
-6.53970540e-01 -5.15410781e-01 3.26711871e-02 4.28357869e-01
2.37505928e-01 2.94202805e-01 5.56284428e-01 -9.35667098e-01
-1.61849216e-01 -4.33919489e-01 -4.05210167e-01 -2.88068682e-01
-2.64521450e-01 -1.09879248e-01 -2.99637735e-01 -4.71344650e-01
1.81862473e+00 -1.97020233e+00 -2.65183412e-02 9.41353321e-01
8.05184990e-03 -1.49103045e-01 4.59986269e-01 8.94867182e-01
-1.99664950e-01 5.01878977e-01 -7.05989897e-01 1.33347824e-01
6.06253028e-01 -2.31615752e-01 -2.93240935e-01 5.77917337e-01
1.27629414e-01 3.10610473e-01 -6.23813987e-01 1.89797148e-01
5.20883091e-02 7.73371905e-02 -3.81581336e-01 -2.47890651e-02
-3.36603038e-02 -7.79749900e-02 -4.31633919e-01 6.39726818e-01
8.00491750e-01 3.87359798e-01 -6.61871880e-02 2.45769575e-01
-5.78247726e-01 -2.15335295e-01 -1.48303473e+00 8.82991433e-01
-3.69205862e-01 4.85777594e-02 3.00576866e-01 -9.95621502e-01
9.54958200e-01 3.37590873e-02 6.19118273e-01 -4.55291152e-01
4.38677996e-01 3.84544909e-01 -1.99204251e-01 1.42285572e-02
4.23134893e-01 -8.51910233e-01 -7.56085932e-01 8.33654702e-01
-3.75886351e-01 -4.68404859e-01 1.94449741e-02 -9.74045098e-02
6.53295636e-01 -1.42582998e-01 5.65330267e-01 -9.83077765e-01
2.68563360e-01 -4.80942667e-01 6.79360569e-01 5.44887304e-01
-5.60163438e-01 3.41285408e-01 1.24073303e+00 -2.60334581e-01
-5.34020782e-01 -1.73031020e+00 -5.45316219e-01 7.43391573e-01
8.85817483e-02 -1.94454473e-02 -6.40107036e-01 -5.99556565e-01
6.55768812e-01 1.23892832e+00 -9.12752390e-01 -2.08169624e-01
1.01219431e-01 -1.09207428e+00 3.47940773e-01 3.92030627e-01
7.46233389e-02 -5.46758175e-01 -8.49234164e-01 9.35198180e-03
1.32874131e-01 -6.01123795e-02 -4.57982481e-01 2.30058432e-01
-5.01119673e-01 -1.01655841e+00 -6.21443510e-01 2.23064348e-01
4.31530088e-01 -4.13912356e-01 1.19473994e+00 -6.64758563e-01
-3.09342623e-01 6.20636106e-01 -1.44801393e-01 -1.05900645e+00
5.67881167e-02 -1.04875028e+00 -6.40176982e-03 -8.63081142e-02
1.11847579e-01 -5.08538008e-01 -9.74026501e-01 3.56772482e-01
-8.57467711e-01 -9.16474700e-01 3.18731338e-01 7.33323991e-01
8.06671381e-01 2.33783618e-01 8.56189013e-01 -6.99757099e-01
1.12906706e+00 -9.38809872e-01 -1.08529043e+00 2.40051195e-01
-1.15299940e+00 3.86584643e-03 1.19645871e-01 1.32143319e-01
-1.16093373e+00 -7.36502111e-01 -8.65144506e-02 2.52379999e-02
5.72834134e-01 9.70979154e-01 -1.20868064e-01 2.77692050e-01
4.39296335e-01 -3.43522459e-01 -1.12880997e-01 -2.57098913e-01
4.54596221e-01 2.58028150e-01 6.52332827e-02 -6.35835648e-01
4.36189324e-01 2.27253258e-01 1.76845610e-01 2.68296134e-02
-9.08751845e-01 -7.25553036e-02 -8.15621205e-03 -4.58636075e-01
7.55632162e-01 -5.46974063e-01 -1.23166394e+00 1.01897515e-01
-4.83334184e-01 6.53158175e-03 -9.30890083e-01 7.09011376e-01
-8.24396491e-01 8.88077244e-02 -1.72319874e-01 -1.59473443e+00
-3.19841087e-01 -1.19486225e+00 7.19389856e-01 2.43894875e-01
5.64744882e-02 -1.25946653e+00 9.81413126e-02 3.35184932e-01
5.42314053e-01 9.86367762e-01 1.01045382e+00 -5.36342919e-01
-1.11831114e-01 -5.12006581e-01 -2.12079033e-01 6.64470613e-01
-1.94939703e-01 4.67412698e-04 -3.79074037e-01 -2.95706183e-01
4.88924384e-01 -7.60124698e-02 5.29705226e-01 7.47025490e-01
9.42682445e-01 -4.91336495e-01 2.39235386e-01 3.25428665e-01
1.73174465e+00 5.37862182e-01 4.52012926e-01 2.97207534e-01
-1.31081507e-01 9.87467766e-01 1.00091839e+00 1.06215119e+00
5.50903797e-01 2.56123304e-01 9.20202494e-01 4.53777343e-01
9.13828850e-01 3.66243790e-03 4.70594466e-01 3.10655206e-01
-7.41752833e-02 -1.60141245e-01 -1.05966008e+00 2.78052658e-01
-1.43480825e+00 -9.60956514e-01 2.93050170e-01 2.50740790e+00
4.95974213e-01 2.97546387e-01 3.00506592e-01 -1.00453924e-02
6.94369972e-01 2.00697631e-01 -4.16222215e-01 -1.23430836e+00
2.50901312e-01 8.83834958e-02 7.55745053e-01 8.24506223e-01
-8.54431808e-01 2.30991155e-01 6.64388657e+00 1.04807162e+00
-6.40990734e-01 -2.52861142e-01 9.57148790e-01 -2.02391416e-01
-8.75650346e-01 9.81108472e-02 -6.18083417e-01 8.39080155e-01
8.17618430e-01 -8.58592689e-01 -1.15636535e-01 8.48083794e-01
3.37726235e-01 -6.06691360e-01 -9.13561165e-01 4.99544114e-01
-1.90431863e-01 -9.00350332e-01 -4.75170434e-01 4.29856867e-01
7.23461926e-01 -3.51929277e-01 6.92416668e-01 8.99914280e-02
8.36733878e-01 -1.23530602e+00 1.01941121e+00 8.40628982e-01
9.87990677e-01 -1.27652669e+00 9.21994269e-01 1.46574810e-01
-1.06995451e+00 -5.38366020e-01 6.66296110e-02 -1.50466770e-01
5.73170662e-01 1.17492712e+00 -1.82957307e-01 9.74982679e-01
8.66416991e-01 2.64250096e-02 2.61020184e-01 8.98857534e-01
-2.86664646e-02 7.91220888e-02 -4.04261470e-01 -3.64424214e-02
2.13051900e-01 -7.94190943e-01 7.10449338e-01 1.00148165e+00
7.44512141e-01 1.55130059e-01 -9.45321172e-02 1.11653006e+00
3.48389626e-01 1.83911368e-01 -3.97705019e-01 2.04214320e-01
4.57215905e-01 7.52552330e-01 -3.67020100e-01 3.93130153e-01
-1.46278813e-01 4.77263005e-03 -9.64179114e-02 6.14644773e-02
-6.10091388e-01 -5.89631021e-01 8.76818419e-01 -1.19316667e-01
5.70387691e-02 6.05586767e-02 -6.87364876e-01 -7.69675553e-01
2.79649943e-01 -5.78658760e-01 8.83213401e-01 -2.52278954e-01
-1.68164420e+00 4.62941647e-01 6.83226943e-01 -1.14900935e+00
-1.51595145e-01 -6.02383077e-01 -7.14207411e-01 1.23409915e+00
-1.48729074e+00 -3.40169847e-01 1.16928071e-01 4.10820156e-01
-1.55484915e-01 -2.10231822e-02 5.37819743e-01 -1.04215741e-01
-5.67819536e-01 7.20539272e-01 7.24724710e-01 -4.95482206e-01
2.99395740e-01 -1.56128609e+00 -2.07239047e-01 6.02030575e-01
-7.36137211e-01 4.74422216e-01 8.42927694e-01 -6.21452928e-01
-9.27869618e-01 -7.18969166e-01 4.26232725e-01 -5.50041378e-01
8.02976012e-01 -5.72371557e-02 -3.27789962e-01 3.60907137e-01
-9.92058888e-02 -2.86631733e-01 1.04119933e+00 -6.83374032e-02
-2.14484155e-01 -4.64667737e-01 -1.88841128e+00 4.94620293e-01
5.68092585e-01 -2.08773345e-01 -6.49764717e-01 4.05376293e-02
5.06909490e-01 2.95953359e-02 -1.47131014e+00 5.92616498e-01
7.87513435e-01 -1.43927681e+00 7.47998238e-01 -1.17055587e-01
4.89467196e-02 3.72702390e-01 -5.35230637e-01 -1.28682506e+00
-8.17619413e-02 -9.08565938e-01 2.81751871e-01 1.10368359e+00
5.87922871e-01 -9.79990840e-01 4.02259976e-01 1.08268154e+00
-1.22634992e-02 -1.19489026e+00 -1.54088008e+00 -9.41773117e-01
9.07300234e-01 -6.74953222e-01 5.92546761e-01 5.08164942e-01
3.90126765e-01 -6.54179513e-01 -1.46245092e-01 -2.68519163e-01
1.01155066e+00 -1.52712315e-01 -1.06832638e-01 -1.25267589e+00
-2.51690328e-01 -8.36033106e-01 -2.57712036e-01 -2.88098216e-01
-3.21521431e-01 -6.58748388e-01 -3.65352519e-02 -1.38166749e+00
-5.28470566e-03 -5.10936379e-01 -7.90124416e-01 2.74477340e-02
-1.26550198e-01 -3.72661352e-01 4.35560495e-01 -1.01369567e-01
-1.50963143e-01 7.16980457e-01 9.57245350e-01 1.99658573e-01
-7.96146318e-02 4.48383629e-01 -1.16471541e+00 9.49773252e-01
7.30820119e-01 -5.39118052e-01 -4.78085369e-01 1.01797000e-01
8.28319371e-01 6.24986827e-01 1.03757806e-01 -3.92582566e-01
-1.38615176e-01 -9.23318148e-01 -1.37835639e-02 -3.74059051e-01
1.74568236e-01 -8.71032298e-01 3.20446163e-01 4.99670058e-01
-4.05965328e-01 6.71946049e-01 -3.42504517e-03 7.42843211e-01
-2.48521760e-01 -1.60808846e-01 8.41180801e-01 -2.55340990e-02
5.49474768e-02 1.60206646e-01 -3.82695615e-01 4.45377052e-01
1.62155366e+00 -1.51767939e-01 -2.91994840e-01 -5.38000643e-01
-7.45339811e-01 5.30982971e-01 5.42649813e-02 -1.39845014e-02
4.35608298e-01 -1.17191279e+00 -8.00316513e-01 -1.04336284e-01
-1.79157332e-01 -8.40600207e-02 3.85733843e-01 8.97506297e-01
-4.27148849e-01 3.20343196e-01 -1.16415761e-01 -4.01054733e-02
-5.18504202e-01 2.29985416e-01 6.46405995e-01 -7.46127963e-01
8.66618231e-02 1.08442533e+00 3.55845660e-01 -1.45059198e-01
2.49102548e-01 -1.91952243e-01 -7.03288149e-03 4.59806323e-01
5.25056720e-01 9.82536197e-01 -3.09273124e-01 -3.79731536e-01
-5.83043098e-01 5.42522848e-01 3.46864969e-01 -7.98305929e-01
1.20741034e+00 -3.18655223e-01 -3.75170916e-01 4.92532134e-01
7.03283727e-01 9.72406492e-02 -1.44159210e+00 2.40860716e-01
2.02680752e-01 -6.93116844e-01 -1.36299610e-01 -1.36006033e+00
-8.41133833e-01 8.14231753e-01 5.41799545e-01 5.75280607e-01
1.22632396e+00 -3.71737599e-01 1.26692176e-01 -3.05079162e-01
3.47889692e-01 -1.44396961e+00 -2.28043228e-01 2.89462835e-01
1.24681270e+00 -9.73579526e-01 2.38389030e-01 -2.08969206e-01
-1.03827262e+00 7.88635075e-01 2.08209172e-01 -2.08907053e-01
1.48695195e+00 2.16649443e-01 -6.05825055e-03 1.54381488e-02
-2.75209039e-01 3.29799384e-01 5.05415559e-01 3.64178091e-01
1.37630925e-01 4.49981660e-01 -1.13432193e+00 1.34044063e+00
-3.73187244e-01 -3.76970649e-01 4.98653293e-01 9.02563810e-01
-3.39781761e-01 -1.03398418e+00 -3.22581708e-01 5.85336149e-01
-9.09749448e-01 -5.84158637e-02 6.81915507e-02 6.69430673e-01
1.36919498e-01 1.16281605e+00 7.44631886e-02 -2.31879815e-01
7.02478945e-01 -2.11181805e-01 5.28741218e-02 -3.78528506e-01
-8.69262516e-01 -7.37104192e-02 -1.96381714e-02 -7.68468618e-01
2.45851800e-02 -8.60606968e-01 -8.18010509e-01 -4.05844837e-01
-1.39157251e-01 6.05758727e-01 7.57615745e-01 8.23366642e-01
4.48682271e-02 2.42584601e-01 9.88197923e-01 -5.70384502e-01
-1.39740145e+00 -6.17635787e-01 -1.09409404e+00 -1.56282231e-01
1.61361724e-01 -9.15610790e-01 -8.37090313e-01 -8.75974536e-01]
|
[5.043295383453369, 3.940107583999634]
|
f32894db-3031-45a5-9184-5b81101db213
|
in-search-of-deep-learning-architectures-for
|
2302.13046
| null |
https://arxiv.org/abs/2302.13046v1
|
https://arxiv.org/pdf/2302.13046v1.pdf
|
In Search of Deep Learning Architectures for Load Forecasting: A Comparative Analysis and the Impact of the Covid-19 Pandemic on Model Performance
|
In power grids, short-term load forecasting (STLF) is crucial as it contributes to the optimization of their reliability, emissions, and costs, while it enables the participation of energy companies in the energy market. STLF is a challenging task, due to the complex demand of active and reactive power from multiple types of electrical loads and their dependence on numerous exogenous variables. Amongst them, special circumstances, such as the COVID-19 pandemic, can often be the reason behind distribution shifts of load series. This work conducts a comparative study of Deep Learning (DL) architectures, namely Neural Basis Expansion Analysis Time Series Forecasting (N-BEATS), Long Short-Term Memory (LSTM), and Temporal Convolutional Networks (TCN), with respect to forecasting accuracy and training sustainability, meanwhile examining their out-of-distribution generalization capabilities during the COVID-19 pandemic era. A Pattern Sequence Forecasting (PSF) model is used as baseline. The case study focuses on day-ahead forecasts for the Portuguese national 15-minute resolution net load time series. The results can be leveraged by energy companies and network operators (i) to reinforce their forecasting toolkit with state-of-the-art DL models; (ii) to become aware of the serious consequences of crisis events on model performance; (iii) as a high-level model evaluation, deployment, and sustainability guide within a smart grid context.
|
['John Psarras', 'Nuno Amaro', 'Georgios Kormpakis', 'Spiros Mouzakitis', 'Vasileios Schoinas', 'Francisco Silva', 'Evangelos Karakolis', 'Sotiris Pelekis']
|
2023-02-25
| null | null | null | null |
['load-forecasting']
|
['miscellaneous']
|
[-2.75457740e-01 -4.19719815e-01 -5.28720655e-02 -2.85404734e-02
-2.39714131e-01 -6.85434461e-01 7.66329169e-01 2.00147688e-01
1.70609161e-01 7.40326822e-01 3.48371208e-01 -8.75338495e-01
-4.21778619e-01 -9.89176869e-01 -2.88189709e-01 -9.97638643e-01
-6.72295809e-01 4.70295221e-01 -5.77562690e-01 -2.86879182e-01
-6.74880669e-02 9.37081754e-01 -1.28261769e+00 2.43083552e-01
9.78438258e-01 1.32113969e+00 2.39087462e-01 2.87949830e-01
-3.19536701e-02 8.52687597e-01 -7.84023404e-01 2.00276151e-01
-2.02764962e-02 -1.79733604e-01 -4.09709722e-01 -3.80992472e-01
-6.80276334e-01 -2.38668412e-01 1.76921651e-01 8.62535954e-01
7.46412992e-01 2.47020558e-01 5.69437563e-01 -1.53355646e+00
-4.95606959e-01 6.48261666e-01 -1.99943528e-01 6.24881506e-01
8.97348598e-02 4.46063191e-01 5.12318373e-01 -5.54604888e-01
1.12795211e-01 7.10958004e-01 9.00346458e-01 -7.76185989e-02
-9.78383362e-01 -7.20625520e-01 1.78375319e-02 6.20093465e-01
-9.98351634e-01 -5.34892827e-02 1.03319764e+00 -8.97119641e-01
1.73602855e+00 1.18518881e-01 8.15238774e-01 1.00278032e+00
6.34795845e-01 3.73610049e-01 8.66111517e-01 -2.64785290e-01
5.16480327e-01 -2.06066519e-01 -1.80859089e-01 -4.37957913e-01
-1.90973058e-01 4.28898811e-01 -7.25743100e-02 8.69212672e-02
-3.11913192e-02 -1.98261999e-03 -2.85956681e-01 3.31883758e-01
-7.96423495e-01 8.79507303e-01 3.94855827e-01 1.07006240e+00
-1.03950775e+00 -1.61604866e-01 6.88968062e-01 3.67161453e-01
9.25405145e-01 2.43857801e-01 -8.00208390e-01 -3.56017590e-01
-1.08092093e+00 4.44861725e-02 7.84479618e-01 3.56148332e-01
3.35457742e-01 1.01816738e+00 -8.92214924e-02 3.15100700e-01
-1.12571746e-01 6.72184050e-01 8.78931582e-01 -3.26506048e-01
2.49239579e-01 3.24814528e-01 1.77719444e-01 -1.06266320e+00
-9.86004591e-01 -1.06557608e+00 -1.38290238e+00 2.15662062e-01
-1.22280352e-01 -3.92999083e-01 -2.94614434e-01 1.46107376e+00
-1.05447248e-01 1.31059930e-01 -1.59289539e-02 2.13851199e-01
1.49133548e-01 1.19464362e+00 4.92972769e-02 -6.57593727e-01
8.75274360e-01 -6.54442668e-01 -7.95238197e-01 8.54218826e-02
8.37106287e-01 -4.60874647e-01 3.63867372e-01 4.34232950e-01
-7.83173680e-01 -6.64780915e-01 -7.82886446e-01 5.39271951e-01
-1.05781376e+00 5.26327454e-02 1.00158222e-01 4.34913188e-01
-9.81531918e-01 5.73132753e-01 -7.09741116e-01 -3.24158907e-01
-5.06043527e-03 -5.19707762e-02 4.17836547e-01 5.66577017e-01
-1.54273403e+00 1.43468618e+00 7.58883417e-01 5.55482805e-01
-6.61211669e-01 -1.17255616e+00 -5.32804728e-01 4.80056494e-01
-1.53224334e-01 -3.37685317e-01 1.04522324e+00 -7.93973863e-01
-1.14518487e+00 -5.25337160e-02 7.35503584e-02 -1.05941021e+00
3.30434948e-01 2.22690448e-01 -1.23311007e+00 -3.76698196e-01
-2.21010312e-01 -1.68981299e-01 4.93085176e-01 -7.34895766e-01
-7.92176127e-01 -2.38370061e-01 -5.22906959e-01 -7.11312294e-02
-2.57479817e-01 -1.95240658e-02 1.07526696e+00 -8.41437042e-01
-5.68095803e-01 -5.22003233e-01 -9.54262763e-02 -1.01479864e+00
-1.75998151e-01 -6.68454707e-01 1.32179356e+00 -1.27048659e+00
1.61047208e+00 -1.94315064e+00 -1.19062103e-01 4.44604218e-01
-3.08514059e-01 4.15159166e-01 1.63507402e-01 9.89792407e-01
-7.43744433e-01 6.38675876e-03 -2.88180672e-02 -1.52542880e-02
3.50786150e-01 3.88108671e-01 -1.03260660e+00 2.79723078e-01
-1.33360494e-02 1.07513475e+00 -7.49282539e-01 6.53911948e-01
8.21345270e-01 5.50708473e-01 3.50364208e-01 1.17640961e-02
-5.32696128e-01 6.97152555e-01 -2.95330733e-02 1.04578517e-01
5.06527960e-01 -2.17202336e-01 2.72880048e-02 -3.67498815e-01
-6.51175320e-01 1.05674930e-01 -7.20768154e-01 1.00939345e+00
-8.52488458e-01 6.80532992e-01 -4.60262626e-01 -1.26101589e+00
9.52016890e-01 6.84202969e-01 9.48272586e-01 -1.18805659e+00
1.01552883e-04 4.43465829e-01 3.71137485e-02 -3.71722609e-01
2.22896427e-01 5.98515943e-02 2.43750662e-01 8.23077142e-01
-1.50364295e-01 2.61600643e-01 5.41474760e-01 -4.39765871e-01
4.71643478e-01 -2.66318619e-01 1.56438455e-01 -5.24654210e-01
6.63296878e-01 -3.89257342e-01 4.04720336e-01 -1.24302665e-02
3.55569601e-01 -2.36627951e-01 3.66687745e-01 -1.11444068e+00
-1.13175344e+00 -6.74614489e-01 -2.50648618e-01 7.59031475e-01
-7.77259767e-01 1.17922954e-01 -2.38850877e-01 -2.31154993e-01
8.99698138e-02 1.78517973e+00 -5.75589299e-01 -6.15198165e-02
-6.75480723e-01 -1.18720007e+00 2.62097687e-01 5.64589918e-01
2.76279241e-01 -1.26141512e+00 -9.09158826e-01 7.68179238e-01
-9.10363067e-03 -8.08558702e-01 4.48101637e-04 5.22629738e-01
-3.96353394e-01 -8.06338906e-01 -6.85771883e-01 -3.29898506e-01
2.46709362e-01 -4.25201952e-01 1.37257791e+00 -3.89120877e-01
2.47635156e-01 9.01087001e-02 -1.31552309e-01 -6.20238960e-01
-7.09647357e-01 2.96427041e-01 1.50903046e-01 -1.37892231e-01
4.41222608e-01 -1.03724015e+00 -4.12858665e-01 2.07938820e-01
-9.40722287e-01 -1.46416612e-02 3.57548557e-02 5.52339017e-01
2.43301958e-01 7.50487626e-01 1.31622458e+00 -3.05430353e-01
8.61038744e-01 -8.93127203e-01 -1.15241265e+00 2.94968516e-01
-1.23674214e+00 -5.42391360e-01 1.16264987e+00 -1.13119289e-01
-8.83696556e-01 -4.48480695e-01 -1.80078968e-01 -3.49765182e-01
-1.34917185e-01 1.01812065e+00 2.31341451e-01 1.98069617e-01
2.75569499e-01 6.31275594e-01 -3.29134941e-01 -6.33977234e-01
1.39286578e-01 5.13395667e-01 4.77887422e-01 -1.54408902e-01
7.68675447e-01 2.88217906e-02 8.95715132e-02 -6.40227854e-01
-4.72514689e-01 -9.31462422e-02 -5.82759023e-01 -4.04558152e-01
6.20589614e-01 -8.64934444e-01 -7.72338569e-01 8.26810300e-01
-1.08777678e+00 -5.49018502e-01 -5.39876461e-01 2.90350437e-01
-2.62629241e-01 -1.09526128e-01 -5.24563193e-01 -9.08703327e-01
-6.38088703e-01 -9.49080467e-01 2.79619902e-01 9.52424854e-02
-1.14755578e-01 -1.72077274e+00 8.68270919e-02 -3.87107909e-01
1.33788264e+00 7.34261572e-01 1.42731655e+00 -8.24332714e-01
5.71319498e-02 -1.13895804e-01 2.51938086e-02 8.05689216e-01
2.61981368e-01 -3.72578241e-02 -9.95443225e-01 -6.85488284e-01
2.38153383e-01 3.05042833e-01 6.46785572e-02 6.77962840e-01
1.22130072e+00 -6.69623792e-01 -1.20444730e-01 5.13041854e-01
1.40963590e+00 8.52971375e-01 4.59117204e-01 3.91104192e-01
2.97505170e-01 2.98687309e-01 -6.41492903e-02 7.86964297e-01
5.87212086e-01 3.62817347e-01 4.49058682e-01 -2.60192126e-01
1.71948776e-01 3.63610089e-02 3.57704580e-01 1.26779377e+00
1.39367655e-01 -4.35790330e-01 -1.15415895e+00 6.58150554e-01
-1.52990985e+00 -1.11803532e+00 8.08429811e-03 1.90136015e+00
2.42162511e-01 5.73114343e-02 2.27242827e-01 5.27661204e-01
4.44121867e-01 3.68833154e-01 -7.51437843e-01 -6.97234869e-01
-4.75757331e-01 6.53840601e-02 3.08595568e-01 1.57301798e-01
-7.44742155e-01 -1.35970995e-01 5.33955288e+00 9.62854087e-01
-1.63077152e+00 7.49522299e-02 1.11233687e+00 1.60417810e-01
-3.28734964e-01 -5.40038645e-01 -5.85382283e-01 9.99439955e-01
1.60725844e+00 -8.92972589e-01 7.90896535e-01 5.86739600e-01
7.58796930e-01 1.01290688e-01 -9.98790801e-01 6.80943906e-01
-1.21147156e-01 -1.47288430e+00 -2.05377936e-01 1.09245867e-01
1.11472130e+00 6.55603409e-01 5.27878515e-02 4.38587874e-01
1.03383236e-01 -1.11247790e+00 5.11718094e-01 7.07143247e-01
4.03220594e-01 -1.10216951e+00 1.18073273e+00 5.92211008e-01
-1.38050711e+00 -6.41456664e-01 1.29895836e-01 -2.11612731e-01
7.36701429e-01 1.04685545e+00 -5.83853304e-01 8.99813056e-01
8.05843294e-01 9.19549227e-01 -1.77683920e-01 6.46195829e-01
-2.03753673e-02 8.97532940e-01 -5.37242651e-01 2.90641189e-01
3.93102556e-01 -2.64668226e-01 4.88025516e-01 9.70053017e-01
7.28567541e-01 -2.69881457e-01 1.40481621e-01 7.41941273e-01
1.68732330e-01 -1.96313560e-01 -5.37592530e-01 -1.36702746e-01
4.33734506e-01 1.08394706e+00 -4.40672994e-01 -2.46046469e-01
-3.25844616e-01 3.31553996e-01 -5.01323104e-01 8.51741612e-01
-7.00062513e-01 -5.99258989e-02 3.59693289e-01 -4.09464687e-02
2.78919756e-01 -6.71067685e-02 -3.96121234e-01 -7.08874226e-01
-7.54407421e-02 -6.93539023e-01 4.80909705e-01 -9.11115110e-01
-1.61200595e+00 7.19130099e-01 -5.94470613e-02 -1.17289400e+00
-1.06626487e+00 -2.10009322e-01 -1.01943219e+00 1.38423216e+00
-1.90254664e+00 -1.20329797e+00 7.73745775e-02 4.15283054e-01
4.42810208e-01 -3.94357055e-01 8.96768034e-01 5.14742792e-01
-5.58216989e-01 3.80361527e-02 7.88575053e-01 -9.31011364e-02
-3.75262558e-01 -1.10605109e+00 4.61253524e-01 8.92149508e-01
-2.95188963e-01 -2.42242947e-01 6.14976883e-01 -3.84727418e-01
-1.11038697e+00 -1.28257442e+00 1.14505601e+00 1.98090240e-01
9.59179282e-01 -1.09170359e-02 -9.45295513e-01 7.14783013e-01
6.87607825e-01 -2.54618317e-01 5.39038956e-01 -2.81861097e-01
2.23202273e-01 -4.26853925e-01 -1.15261149e+00 -4.24391702e-02
4.62017674e-03 -5.50284386e-01 -2.57608473e-01 3.98809016e-01
3.19237977e-01 1.64465643e-02 -1.28352010e+00 6.76069975e-01
3.03334326e-01 -8.73133481e-01 6.51793242e-01 -3.26982856e-01
-5.82538061e-02 -2.92040855e-01 -1.92836627e-01 -1.92975950e+00
-5.75209022e-01 -6.56886041e-01 -4.72593069e-01 1.23667073e+00
3.17225069e-01 -1.17481625e+00 1.56416506e-01 3.70281965e-01
-3.20589572e-01 -8.10806096e-01 -1.35604775e+00 -9.27400589e-01
3.95300180e-01 -6.48381412e-01 1.43569303e+00 1.28979325e+00
-7.89991990e-02 -7.63362497e-02 -3.41679066e-01 3.19700003e-01
2.34925091e-01 3.93253505e-01 1.65252015e-01 -1.04490232e+00
8.49908367e-02 -9.22464073e-01 -3.43131125e-02 -2.18465120e-01
6.12480082e-02 -7.59768188e-01 -5.41301668e-01 -1.45120919e+00
-5.98258018e-01 -2.33723924e-01 -7.30671763e-01 5.77143252e-01
5.76668501e-01 -1.58018589e-01 4.25285161e-01 3.57894748e-02
1.57578811e-01 8.52631330e-01 4.45610464e-01 -1.70071647e-01
6.44114241e-02 2.40813404e-01 1.82919025e-01 5.68877459e-01
1.16464365e+00 -6.75820105e-04 -3.51223767e-01 -4.08210695e-01
7.26200819e-01 2.25920066e-01 1.51083022e-01 -1.25632834e+00
2.58949041e-01 3.83484885e-02 5.53057313e-01 -1.27560318e+00
-2.83455282e-01 -1.16308224e+00 1.00026226e+00 8.68035197e-01
4.40860316e-02 9.16067421e-01 6.01412594e-01 1.42539451e-02
-2.27910697e-01 7.57176355e-02 2.88827986e-01 2.44850993e-01
-7.20847249e-01 2.82134771e-01 -6.78509474e-01 -2.87187725e-01
1.04112005e+00 1.15645073e-01 -5.15524745e-01 -4.66992646e-01
-8.05541813e-01 5.73582947e-01 -2.23138276e-02 5.82471132e-01
-1.23359576e-01 -1.29012382e+00 -9.27143276e-01 4.66141284e-01
-4.17119563e-01 -3.07528198e-01 5.14904916e-01 8.63636553e-01
-1.04619876e-01 7.92109609e-01 -4.20631468e-02 -4.80469167e-01
-3.17854971e-01 6.58414602e-01 8.16213071e-01 -6.80356979e-01
-4.40200031e-01 8.00165087e-02 -1.47614300e-01 -3.48410100e-01
8.76048505e-02 -5.40796697e-01 -3.81628871e-01 7.38160014e-01
6.06811643e-01 8.93365324e-01 6.20737374e-01 -5.80271304e-01
-1.57703131e-01 1.83643207e-01 5.70872903e-01 5.89412689e-01
1.76607537e+00 -1.89425781e-01 -3.43991905e-01 8.71817410e-01
1.02626157e+00 -6.74991429e-01 -8.70052576e-01 -8.74012522e-03
2.14085445e-01 2.41442934e-01 2.76526749e-01 -1.47178555e+00
-1.40859318e+00 8.59989822e-01 9.05529380e-01 1.01130354e+00
1.48104215e+00 -5.52803338e-01 9.51236367e-01 -6.27375692e-02
3.12215388e-01 -9.58172739e-01 -7.67782688e-01 6.14809394e-01
1.05136752e+00 -5.86793542e-01 -3.18854213e-01 5.93261003e-01
-2.21653227e-02 1.41954744e+00 -1.01881742e-01 2.74527937e-01
1.06725800e+00 5.77982426e-01 -1.94586039e-01 4.71442081e-02
-9.88984525e-01 2.21478611e-01 1.75336272e-01 5.30070186e-01
1.86561659e-01 3.82863432e-01 -7.87694007e-03 6.22188866e-01
-2.70658135e-01 3.16610426e-01 2.19536513e-01 4.54336256e-01
1.09663092e-01 -6.65184975e-01 -1.69851512e-01 6.37974977e-01
-4.35526729e-01 -7.56741911e-02 5.65542400e-01 4.01948333e-01
3.71024936e-01 8.80476177e-01 3.68557185e-01 -1.38926625e-01
2.41817579e-01 1.90327331e-01 -3.01325738e-01 1.27060711e-01
-9.86659288e-01 -7.55146220e-02 -1.33619919e-01 -1.11742631e-01
-1.44036129e-01 -6.16959453e-01 -9.22451854e-01 -5.80641925e-01
-1.25071108e-01 3.53799373e-01 9.41897333e-01 1.20806277e+00
3.73852968e-01 9.21932518e-01 1.08141816e+00 -9.99304652e-01
-5.89347184e-01 -1.33760834e+00 -6.87634587e-01 1.21711902e-01
3.01540464e-01 -1.48799136e-01 -5.98625481e-01 -1.86606392e-01]
|
[6.124459266662598, 2.7681636810302734]
|
9919ddde-5565-46e4-85c1-7a929979bdad
|
maskcon-masked-contrastive-learning-for
|
2303.12756
| null |
https://arxiv.org/abs/2303.12756v1
|
https://arxiv.org/pdf/2303.12756v1.pdf
|
MaskCon: Masked Contrastive Learning for Coarse-Labelled Dataset
|
Deep learning has achieved great success in recent years with the aid of advanced neural network structures and large-scale human-annotated datasets. However, it is often costly and difficult to accurately and efficiently annotate large-scale datasets, especially for some specialized domains where fine-grained labels are required. In this setting, coarse labels are much easier to acquire as they do not require expert knowledge. In this work, we propose a contrastive learning method, called $\textbf{Mask}$ed $\textbf{Con}$trastive learning~($\textbf{MaskCon}$) to address the under-explored problem setting, where we learn with a coarse-labelled dataset in order to address a finer labelling problem. More specifically, within the contrastive learning framework, for each sample our method generates soft-labels with the aid of coarse labels against other samples and another augmented view of the sample in question. By contrast to self-supervised contrastive learning where only the sample's augmentations are considered hard positives, and in supervised contrastive learning where only samples with the same coarse labels are considered hard positives, we propose soft labels based on sample distances, that are masked by the coarse labels. This allows us to utilize both inter-sample relations and coarse labels. We demonstrate that our method can obtain as special cases many existing state-of-the-art works and that it provides tighter bounds on the generalization error. Experimentally, our method achieves significant improvement over the current state-of-the-art in various datasets, including CIFAR10, CIFAR100, ImageNet-1K, Standford Online Products and Stanford Cars196 datasets. Code and annotations are available at https://github.com/MrChenFeng/MaskCon_CVPR2023.
|
['Ioannis Patras', 'Chen Feng']
|
2023-03-22
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Feng_MaskCon_Masked_Contrastive_Learning_for_Coarse-Labelled_Dataset_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Feng_MaskCon_Masked_Contrastive_Learning_for_Coarse-Labelled_Dataset_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['learning-with-coarse-labels']
|
['computer-vision']
|
[ 2.09727958e-01 3.11760008e-01 -3.10082257e-01 -7.94952631e-01
-9.22956765e-01 -6.20688856e-01 3.79010826e-01 1.94275662e-01
-5.92489541e-01 8.30541193e-01 -1.96741313e-01 -1.25045717e-01
-1.63579077e-01 -7.16694832e-01 -8.39192808e-01 -6.58923328e-01
1.77481212e-02 8.24092805e-01 2.71674007e-01 -2.55774781e-02
-2.02281296e-01 3.00557524e-01 -1.51579046e+00 4.15219605e-01
7.11778164e-01 1.36204624e+00 5.57880513e-02 2.14402080e-01
4.67979014e-02 8.29590261e-01 -4.16577339e-01 -6.30857527e-01
4.83718336e-01 -1.45917088e-01 -1.20382118e+00 1.77387998e-01
7.45512187e-01 -1.36261567e-01 2.31182799e-02 1.23232412e+00
2.58590609e-01 6.04715720e-02 5.39540172e-01 -1.26805282e+00
-6.98967159e-01 7.14672327e-01 -8.22079062e-01 1.18383743e-01
-8.45890939e-02 7.10617229e-02 1.33665872e+00 -1.03917825e+00
4.23308223e-01 1.20315516e+00 5.46323001e-01 4.40828353e-01
-1.34903252e+00 -1.04931092e+00 4.50644433e-01 2.45095089e-01
-1.56782806e+00 -2.54275858e-01 7.07321703e-01 -4.61634785e-01
5.02827406e-01 1.59525558e-01 1.92552581e-01 9.63440657e-01
-4.10107881e-01 8.28378737e-01 1.38548791e+00 -4.71093327e-01
2.18099326e-01 2.80665457e-01 6.62660778e-01 9.38612223e-01
1.10691495e-01 7.11374059e-02 -1.68331116e-01 2.50538508e-03
5.39295375e-01 5.77573068e-02 -1.69342890e-01 -3.58010143e-01
-1.11140704e+00 9.19571698e-01 6.79331601e-01 2.73305386e-01
-2.56780088e-02 8.88024122e-02 2.28830785e-01 3.13416988e-01
6.15887225e-01 4.17792767e-01 -7.41968691e-01 2.63722360e-01
-7.11540878e-01 7.85619766e-02 7.13424504e-01 1.00180340e+00
1.23364055e+00 -2.30241567e-01 -5.10933511e-02 9.24812436e-01
9.25741866e-02 2.21096545e-01 3.68484706e-01 -1.02014041e+00
5.24167180e-01 6.66253686e-01 7.01396614e-02 -7.08384931e-01
-5.04924178e-01 -7.87097275e-01 -1.04277205e+00 2.32206479e-01
5.37118554e-01 -5.15080318e-02 -9.71809626e-01 1.92094576e+00
3.84318948e-01 3.52262139e-01 -1.56430721e-01 7.87608564e-01
6.85921907e-01 3.63934696e-01 8.60676169e-02 8.95703025e-03
1.31613338e+00 -1.20567977e+00 -3.62040371e-01 -3.68061841e-01
9.03132081e-01 -4.67245638e-01 1.45095193e+00 4.88014311e-01
-7.01200306e-01 -7.66034126e-01 -8.53289843e-01 -6.15597814e-02
-5.66143513e-01 2.89184421e-01 4.81360018e-01 4.64808226e-01
-9.70991492e-01 4.93592292e-01 -6.73465669e-01 -1.47597536e-01
7.54705489e-01 5.00761688e-01 -3.95760298e-01 -2.57503241e-01
-1.09813857e+00 7.33009577e-01 6.03412509e-01 1.00769103e-01
-9.05660629e-01 -4.84752059e-01 -8.78040910e-01 1.54683518e-03
8.18612933e-01 -1.71199024e-01 1.29466927e+00 -1.01179647e+00
-1.13646829e+00 1.15198421e+00 3.24550876e-03 -4.34388787e-01
4.99904782e-01 -2.36753061e-01 -2.67412931e-01 -5.18486015e-02
2.22045943e-01 1.04092419e+00 6.37539089e-01 -1.31971979e+00
-7.79888213e-01 -1.86106667e-01 4.13764596e-01 -5.56248240e-02
-3.23442370e-01 -1.12769037e-01 -4.66035247e-01 -6.16476178e-01
-1.70282573e-01 -9.78325844e-01 -2.62324065e-01 9.61781219e-02
-5.73518574e-01 -5.46051621e-01 7.87595093e-01 -1.74121872e-01
9.38524783e-01 -2.27532601e+00 -1.12725027e-01 2.77200729e-01
5.15686572e-01 5.81628621e-01 -3.67342442e-01 -4.53480007e-03
-1.18758947e-01 2.14326873e-01 -4.73973870e-01 -4.66165930e-01
1.59667879e-01 3.75564992e-01 -9.07403603e-02 4.07358050e-01
3.84289652e-01 9.14245546e-01 -9.37437117e-01 -5.30303061e-01
1.03934497e-01 9.84114036e-02 -4.21751618e-01 9.94998887e-02
-1.68070033e-01 3.74218851e-01 -3.67829472e-01 6.24545932e-01
5.86967766e-01 -8.48169982e-01 2.52658799e-02 -2.74382621e-01
2.31070459e-01 4.63214926e-02 -1.29798353e+00 1.36952174e+00
-3.98334771e-01 3.28830868e-01 9.73239690e-02 -1.45048118e+00
8.79368007e-01 8.77242815e-03 1.80743828e-01 -5.84353089e-01
1.75330430e-01 2.38155514e-01 -6.52434304e-02 -2.34041914e-01
2.17813388e-01 -1.72315955e-01 -1.64589658e-01 3.51660103e-01
1.27115592e-01 8.63120109e-02 3.04433078e-01 1.24961846e-01
9.91908133e-01 -7.57957227e-04 4.35239315e-01 -3.25580835e-01
5.52621603e-01 -1.72606692e-01 7.29297817e-01 7.61137605e-01
-3.05697232e-01 4.71793562e-01 4.19120431e-01 -4.16202575e-01
-6.91664934e-01 -9.03950572e-01 -2.55746901e-01 1.35198116e+00
1.73615128e-01 -2.73412079e-01 -7.55697370e-01 -1.23315656e+00
2.89440318e-03 4.07018393e-01 -7.05007255e-01 -4.84145731e-02
-5.47065377e-01 -6.97217345e-01 5.48423290e-01 6.69922709e-01
7.23764658e-01 -1.09581161e+00 -2.13094488e-01 8.14844444e-02
-1.06281206e-01 -1.18162143e+00 -4.03670162e-01 4.75750983e-01
-4.80300069e-01 -1.15811133e+00 -3.87868732e-01 -9.82813895e-01
7.65372753e-01 3.32643352e-02 1.25583601e+00 6.86393604e-02
-1.23774901e-01 -2.34627407e-02 -5.02232254e-01 -3.11658740e-01
-1.77386194e-01 2.74012387e-01 2.54955143e-02 2.44612113e-01
4.48186517e-01 -3.78316402e-01 -3.81153375e-01 6.39317334e-01
-8.88210893e-01 -6.83308542e-02 7.27670491e-01 1.07874227e+00
9.35497105e-01 2.14632511e-01 8.57557416e-01 -1.44067883e+00
1.48359671e-01 -5.75016260e-01 -7.34013855e-01 3.16594571e-01
-6.25357747e-01 3.39481756e-02 8.53843808e-01 -6.07520580e-01
-8.07829916e-01 9.96000171e-02 -2.62695134e-01 -4.53195632e-01
-4.95499372e-01 4.37388897e-01 -3.28259021e-01 -6.19276315e-02
7.54585385e-01 -2.41783023e-01 -4.10791457e-01 -6.03605926e-01
3.98789227e-01 6.98111832e-01 5.16542435e-01 -7.97192752e-01
7.97913313e-01 4.84108627e-01 -1.08106911e-01 -4.16956127e-01
-1.41147089e+00 -4.88833308e-01 -7.34934270e-01 5.68193048e-02
8.00268352e-01 -8.65207434e-01 -5.79413116e-01 4.06164110e-01
-6.73209727e-01 -6.59436464e-01 -5.66833317e-01 3.93668771e-01
-2.90177405e-01 2.64880240e-01 -6.99326932e-01 -4.73806083e-01
-5.49297072e-02 -1.27376497e+00 1.05740607e+00 1.19395934e-01
-1.46209836e-01 -9.39929664e-01 -2.55329281e-01 4.82160449e-01
5.68321869e-02 2.25734383e-01 8.65620613e-01 -1.04555941e+00
-5.48752785e-01 -1.82158589e-01 -4.44383979e-01 7.74524987e-01
1.75990224e-01 -4.24902439e-01 -1.12852359e+00 -3.96959245e-01
-1.79803446e-01 -9.37920213e-01 9.01748776e-01 8.89007822e-02
1.48144436e+00 -3.42453837e-01 -3.16227704e-01 4.81665194e-01
1.25379920e+00 -3.13752145e-02 2.91548908e-01 1.16766743e-01
9.25537825e-01 6.73280776e-01 7.82996953e-01 2.26690426e-01
4.71505433e-01 6.00009978e-01 4.66351718e-01 -3.56577754e-01
-3.46908458e-02 -6.43274561e-03 -1.82100721e-02 5.72704136e-01
1.02494486e-01 -2.39169717e-01 -9.72451866e-01 4.99600798e-01
-1.78133154e+00 -6.40345275e-01 4.48233038e-02 2.01072073e+00
1.08114123e+00 3.74769181e-01 2.05804035e-02 1.45801857e-01
7.61514485e-01 1.17233649e-01 -6.36690795e-01 4.34746146e-02
-9.66942981e-02 5.24082184e-01 4.79681194e-01 4.97187167e-01
-1.44052863e+00 1.01384389e+00 5.12599230e+00 1.10600984e+00
-9.27151620e-01 4.14572626e-01 1.12425733e+00 -5.64495735e-02
-4.36500348e-02 -1.68374509e-01 -1.04890585e+00 4.24779296e-01
7.25603223e-01 4.53852713e-01 3.00157517e-01 9.92422223e-01
-2.65910208e-01 -6.36793382e-04 -1.52832258e+00 9.12222862e-01
-7.28710964e-02 -1.26269126e+00 -3.59546453e-01 1.55573577e-01
7.86818564e-01 2.46978596e-01 1.77635461e-01 6.14461780e-01
7.92592287e-01 -9.76342499e-01 7.88339198e-01 1.28666312e-01
1.03775728e+00 -6.67939425e-01 9.60938573e-01 6.40783966e-01
-1.30416250e+00 -1.80132091e-01 -2.75628716e-01 -9.79293063e-02
-1.10950209e-01 9.34480846e-01 -5.92023790e-01 4.53708857e-01
9.20332968e-01 7.17517793e-01 -6.57131433e-01 6.04158580e-01
-4.08178806e-01 7.44843662e-01 -2.80800641e-01 2.44180486e-01
3.38729292e-01 1.87791977e-02 -4.01184745e-02 1.19181824e+00
1.71806980e-02 1.51170194e-01 7.75568962e-01 8.68011415e-01
-6.15889728e-01 -7.46285096e-02 -2.28683352e-01 1.72651142e-01
5.20726204e-01 1.44645035e+00 -7.25925207e-01 -5.61872184e-01
-5.13359070e-01 6.39692426e-01 8.33451033e-01 3.75047714e-01
-8.04272830e-01 -2.36350060e-01 3.75708818e-01 -2.57715899e-02
4.24114197e-01 1.15824856e-01 -8.94933343e-02 -1.16770446e+00
7.09015578e-02 -9.06357348e-01 5.47745645e-01 -4.30313021e-01
-1.70778644e+00 6.37264013e-01 6.08171411e-02 -1.14295244e+00
-1.02604024e-01 -7.27016687e-01 -2.55143404e-01 8.37443888e-01
-1.70177662e+00 -1.17710149e+00 -2.51745015e-01 7.22437739e-01
4.99746472e-01 4.81472798e-02 6.97086751e-01 4.75271016e-01
-6.04132414e-01 9.07464743e-01 -6.45337030e-02 3.72750908e-01
8.38511705e-01 -1.34986711e+00 2.13721707e-01 7.26911068e-01
2.59433955e-01 3.03459883e-01 2.56795377e-01 -4.43850309e-01
-5.75010538e-01 -1.47255802e+00 6.86045110e-01 -3.40708613e-01
6.91669524e-01 -5.62179685e-01 -1.05407345e+00 9.41264272e-01
-1.71773927e-03 7.31805503e-01 6.41536474e-01 2.33759105e-01
-6.49649024e-01 -3.76922697e-01 -1.33237386e+00 1.43830374e-01
1.18028188e+00 -5.56335747e-01 -4.99282330e-01 4.24408287e-01
7.15964437e-01 -3.31158906e-01 -8.59850228e-01 5.90948939e-01
2.03701690e-01 -7.28307545e-01 8.18487167e-01 -3.99699271e-01
1.85061425e-01 -3.66720557e-01 -2.68333524e-01 -1.22439563e+00
-4.62993681e-01 -1.74323440e-01 -7.73520395e-02 1.37017870e+00
4.67246771e-01 -7.57602453e-01 8.52838278e-01 4.53145087e-01
-3.01204205e-01 -9.88893092e-01 -8.22974145e-01 -9.48197007e-01
1.09097160e-01 -3.19440812e-01 5.37647128e-01 1.31601095e+00
-3.79119545e-01 3.33630353e-01 -2.80735970e-01 2.30821818e-01
6.50387526e-01 8.54535550e-02 5.27838647e-01 -1.48961341e+00
-5.18941045e-01 -2.52673239e-01 -2.03556940e-01 -1.09173214e+00
4.96862561e-01 -1.00407171e+00 2.16834486e-01 -1.16508698e+00
2.60072201e-01 -1.01642561e+00 -5.84406316e-01 1.01726007e+00
-2.29947880e-01 6.59763098e-01 1.58765838e-01 2.20037296e-01
-8.35558891e-01 2.79351383e-01 1.01949668e+00 -2.11646855e-01
3.81923057e-02 -3.66500404e-04 -7.66014040e-01 9.75360155e-01
7.74059117e-01 -5.98025203e-01 -4.83166486e-01 -3.87687624e-01
1.61053628e-01 -3.59384716e-01 4.99069005e-01 -9.14691627e-01
6.90224841e-02 -1.02407180e-01 2.75450468e-01 -5.20153403e-01
3.03324133e-01 -8.37975919e-01 -6.44220486e-02 3.10768723e-01
-6.75625801e-01 -1.64563015e-01 -7.31838867e-02 5.31623483e-01
-2.95902073e-01 -2.87564456e-01 1.00337708e+00 -2.23126024e-01
-7.84684181e-01 5.20699263e-01 6.19849227e-02 3.85003686e-01
9.90787506e-01 1.12229548e-01 -5.35450518e-01 -6.04404919e-02
-7.84947693e-01 3.88580412e-01 2.92112827e-01 1.56067848e-01
3.78885865e-01 -1.27783716e+00 -6.40726686e-01 4.17706855e-02
3.38040769e-01 4.77104038e-01 1.60857230e-01 7.66015172e-01
-6.31852821e-02 2.18639895e-01 1.02002807e-02 -7.23072886e-01
-1.05764949e+00 7.27752030e-01 1.94802940e-01 -6.19630158e-01
-4.06837434e-01 1.16571033e+00 6.98183775e-01 -6.70422137e-01
3.89501065e-01 -5.10200083e-01 -1.34396017e-01 -4.35912944e-02
4.90821689e-01 1.73385352e-01 1.11294784e-01 -5.88368773e-01
-2.99919426e-01 5.13372421e-01 -4.30966586e-01 1.83115989e-01
1.20336282e+00 -5.07459864e-02 6.93834871e-02 4.06136990e-01
1.39108813e+00 -4.80469801e-02 -1.38088787e+00 -7.47204483e-01
1.20163769e-01 -3.44970942e-01 -1.74230143e-01 -9.30460453e-01
-1.28044665e+00 8.77319098e-01 6.78789616e-01 1.76336423e-01
1.23356092e+00 4.71131682e-01 5.99911690e-01 6.03645563e-01
3.91858697e-01 -9.70037460e-01 2.08859891e-01 4.68203485e-01
4.57480282e-01 -1.54366493e+00 -2.28720441e-01 -5.67621768e-01
-5.98850012e-01 6.43753409e-01 8.14738929e-01 -1.38822526e-01
6.31369472e-01 1.88884914e-01 -3.36790793e-02 -1.91205770e-01
-5.38550317e-01 -3.19798321e-01 2.91068465e-01 4.41167086e-01
2.55232930e-01 2.24760965e-01 4.35608178e-02 6.83920085e-01
-5.43451533e-02 -2.88907979e-02 2.61680096e-01 9.20828164e-01
-4.47974354e-01 -1.27513850e+00 -1.92187324e-01 6.85685456e-01
-4.31302726e-01 -2.08684251e-01 -2.43538871e-01 8.87966156e-01
6.72219992e-01 9.24398959e-01 2.20828457e-03 -4.17925417e-01
1.82739317e-01 7.46542728e-03 2.19730988e-01 -9.60977793e-01
-4.94776785e-01 -1.65058617e-02 2.41470888e-01 -4.86263871e-01
-7.24012256e-01 -4.76910859e-01 -1.43210280e+00 -1.17124811e-01
-4.16303605e-01 1.85410708e-01 1.89815655e-01 1.19989777e+00
1.89825624e-01 3.77913654e-01 6.37989640e-01 -7.02601910e-01
-5.68614900e-01 -9.36412632e-01 -6.19905233e-01 5.39122880e-01
3.76443744e-01 -8.48912418e-01 -3.80306989e-01 1.49044320e-01]
|
[9.508539199829102, 3.188457727432251]
|
f3bc6365-141b-4087-8fa2-79db15b175e8
|
adversarial-attack-by-limited-point-cloud
|
2110.03745
| null |
https://arxiv.org/abs/2110.03745v1
|
https://arxiv.org/pdf/2110.03745v1.pdf
|
Adversarial Attack by Limited Point Cloud Surface Modifications
|
Recent research has revealed that the security of deep neural networks that directly process 3D point clouds to classify objects can be threatened by adversarial samples. Although existing adversarial attack methods achieve high success rates, they do not restrict the point modifications enough to preserve the point cloud appearance. To overcome this shortcoming, two constraints are proposed. These include applying hard boundary constraints on the number of modified points and on the point perturbation norms. Due to the restrictive nature of the problem, the search space contains many local maxima. The proposed method addresses this issue by using a high step-size at the beginning of the algorithm to search the main surface of the point cloud fast and effectively. Then, in order to converge to the desired output, the step-size is gradually decreased. To evaluate the performance of the proposed method, it is run on the ModelNet40 and ScanObjectNN datasets by employing the state-of-the-art point cloud classification models; including PointNet, PointNet++, and DGCNN. The obtained results show that it can perform successful attacks and achieve state-of-the-art results by only a limited number of point modifications while preserving the appearance of the point cloud. Moreover, due to the effective search algorithm, it can perform successful attacks in just a few steps. Additionally, the proposed step-size scheduling algorithm shows an improvement of up to $14.5\%$ when adopted by other methods as well. The proposed method also performs effectively against popular defense methods.
|
['Shohreh Kasaei', 'Hanieh Naderi', 'Atrin Arya']
|
2021-10-07
|
adversarial-attack-by-limited-point-cloud-1
|
https://openreview.net/forum?id=MACKPM_haAu
|
https://openreview.net/pdf?id=MACKPM_haAu
| null |
['point-cloud-classification']
|
['computer-vision']
|
[-1.21250831e-01 -2.21709251e-01 9.14146751e-02 1.26245897e-02
-4.19827312e-01 -6.56256437e-01 6.17116332e-01 1.14873871e-01
-5.12834251e-01 4.23514128e-01 -7.88199902e-01 -4.35132086e-01
-8.01556781e-02 -1.15100527e+00 -9.24761534e-01 -9.41689134e-01
-1.31852478e-01 3.34893823e-01 4.49053824e-01 -3.38746637e-01
5.38307130e-01 1.34791720e+00 -1.42764437e+00 -1.55237064e-01
7.56656170e-01 1.19163775e+00 -2.44033054e-01 2.23628297e-01
-1.30363896e-01 1.29292369e-01 -8.46472085e-01 -3.79189730e-01
8.38820279e-01 3.63813847e-01 -4.55399752e-01 -3.74437392e-01
5.79107523e-01 -3.74624640e-01 -4.89374220e-01 1.59122860e+00
5.43797255e-01 2.24443555e-01 3.89454514e-01 -1.52821374e+00
-3.25955153e-01 1.73676327e-01 -6.55839622e-01 1.99761033e-01
-1.84251457e-01 3.23858052e-01 5.80411136e-01 -6.22662008e-01
3.35187674e-01 1.13222241e+00 6.19165897e-01 6.58701003e-01
-9.77900267e-01 -1.40831339e+00 1.13883197e-01 1.36892363e-01
-1.49974060e+00 1.00319915e-01 1.16204417e+00 -1.45818114e-01
8.62271786e-01 4.53911632e-01 4.45259154e-01 1.04192507e+00
2.84222305e-01 1.96149528e-01 7.54058361e-01 -5.39171370e-03
4.75506634e-01 -5.53213060e-02 8.03992711e-03 3.41560572e-01
3.60104352e-01 4.66445208e-01 -1.01919123e-03 -5.51736593e-01
6.27612174e-01 1.22579299e-01 -1.95176244e-01 -3.62763166e-01
-8.04123402e-01 9.58610356e-01 9.59942579e-01 2.73201704e-01
-2.77513564e-01 1.65952653e-01 4.72456723e-01 8.14933330e-02
3.33598584e-01 5.56639135e-01 -1.32351652e-01 2.64722258e-01
-7.59952128e-01 4.85919446e-01 4.84241873e-01 8.13608468e-01
5.45570195e-01 2.88253695e-01 3.78352046e-01 3.20223004e-01
1.85210854e-01 6.80602610e-01 4.00775224e-02 -6.13822222e-01
6.72288358e-01 4.84782279e-01 6.95596263e-02 -1.57070017e+00
-2.46467590e-01 -5.31697392e-01 -1.08513999e+00 1.01478100e+00
2.15995982e-01 -6.08873293e-02 -1.01197958e+00 1.67167580e+00
6.39215052e-01 5.69866478e-01 -3.34572680e-02 9.13125575e-01
5.15581965e-01 8.08608234e-01 1.40918717e-01 1.25954390e-01
1.00990021e+00 -3.31664324e-01 -2.44393155e-01 -9.14810002e-02
2.29609489e-01 -6.78559601e-01 7.72350132e-01 4.31120634e-01
-9.11264718e-01 -6.07808352e-01 -1.46323097e+00 6.33857071e-01
-4.18377668e-01 -3.56787920e-01 4.08229947e-01 8.39381158e-01
-5.32923222e-01 9.03591335e-01 -1.14138007e+00 5.35650514e-02
6.32655978e-01 6.27843678e-01 -3.82219076e-01 2.19897673e-01
-1.23094463e+00 7.30989993e-01 4.15385991e-01 2.44572937e-01
-7.76033223e-01 -9.93445456e-01 -3.80026966e-01 1.57058939e-01
7.85692334e-02 -3.51383716e-01 5.73883474e-01 -6.76951885e-01
-1.26590633e+00 6.02524996e-01 4.05596673e-01 -7.25582898e-01
7.83598840e-01 -2.51874447e-01 -3.75196248e-01 3.90700400e-01
-2.22402766e-01 5.67168415e-01 1.03976345e+00 -1.39894950e+00
-6.39297307e-01 -5.88867307e-01 2.84298509e-01 9.26235020e-02
-4.74248379e-01 4.27382365e-02 -2.24384978e-01 -6.81184828e-01
2.84357786e-01 -1.19342518e+00 -3.38367701e-01 3.18995804e-01
-4.40871298e-01 -6.27779141e-02 1.29441464e+00 -1.05208576e-01
5.71135044e-01 -2.43941331e+00 -6.95141330e-02 5.46528399e-01
2.66857445e-01 6.61585748e-01 -6.81680366e-02 1.15253247e-01
-3.20552737e-01 4.05835778e-01 -2.76877075e-01 -1.93971619e-01
-1.65502593e-01 8.42809230e-02 -7.33775616e-01 9.51556027e-01
1.03651926e-01 3.54207993e-01 -5.21934330e-01 -7.98581019e-02
5.47410488e-01 4.80669618e-01 -5.59279263e-01 -3.55379395e-02
-5.07590696e-02 2.25694031e-01 -6.72922194e-01 4.85451639e-01
1.48724782e+00 2.44457692e-01 -4.26875979e-01 -1.17087223e-01
-1.11285029e-02 -1.36863038e-01 -1.37683880e+00 1.15733659e+00
-2.19771460e-01 3.47081751e-01 2.66001314e-01 -8.53203833e-01
1.09593606e+00 1.87617585e-01 6.65497243e-01 -1.89132407e-01
3.45754862e-01 2.42598340e-01 9.01877731e-02 8.53222907e-02
1.92927554e-01 -1.05409421e-01 7.97995105e-02 5.90288006e-02
-4.64622110e-01 -4.17750180e-01 -4.71645415e-01 -1.33735463e-01
1.07939303e+00 -3.36528182e-01 -1.72990799e-01 -1.67054623e-01
8.02496791e-01 8.00686404e-02 5.94154537e-01 7.08521843e-01
-3.10258150e-01 5.34789443e-01 1.39819577e-01 -5.26274085e-01
-1.05077934e+00 -1.14023507e+00 -2.31740907e-01 3.50815713e-01
6.61853373e-01 4.47064936e-02 -7.20262349e-01 -8.19547296e-01
3.24653864e-01 8.68895888e-01 -5.39391935e-01 -5.06684482e-01
-7.41369188e-01 -5.97594500e-01 1.05296183e+00 3.86689276e-01
9.08415854e-01 -1.05729222e+00 -5.74322164e-01 -1.08954042e-01
2.84368902e-01 -1.09921992e+00 1.84707418e-02 -1.05318747e-01
-9.21750426e-01 -1.03170121e+00 -3.07809740e-01 -6.38331592e-01
7.07054317e-01 2.06153557e-01 6.06705725e-01 4.15869623e-01
1.09238133e-01 -2.15276062e-01 -2.28450179e-01 -5.98523915e-01
-4.14290816e-01 1.22006066e-01 3.47270250e-01 -1.13582447e-01
2.95454413e-01 -7.82583177e-01 -4.74985778e-01 4.75102365e-01
-1.07021749e+00 -6.00729108e-01 3.49105686e-01 6.52413964e-01
4.84441251e-01 5.38808048e-01 2.38624424e-01 -3.81083786e-01
3.85826409e-01 -2.31099129e-01 -1.03457880e+00 -2.52090842e-01
-4.07856733e-01 -3.18125576e-01 9.00052130e-01 -6.61072552e-01
-5.57308495e-01 -5.95550798e-02 -4.46852028e-01 -1.20065570e+00
-3.86141866e-01 3.16831730e-02 -3.11207861e-01 -8.87100518e-01
6.75319493e-01 4.30463739e-02 9.87667292e-02 -3.18675667e-01
6.25562444e-02 3.25703412e-01 5.31716168e-01 -3.57205957e-01
1.65728104e+00 8.08786213e-01 4.85204697e-01 -7.69588768e-01
-1.72588065e-01 -1.45898312e-01 -4.10563350e-01 -9.90101248e-02
6.71320558e-01 -5.09319484e-01 -1.19219053e+00 5.87033093e-01
-1.18843579e+00 3.01850259e-01 -6.94432259e-02 3.92573923e-01
-2.07905829e-01 6.52852833e-01 -4.71819669e-01 -6.32907987e-01
-6.58608735e-01 -1.29348218e+00 6.14680827e-01 -1.64730679e-02
1.92658231e-01 -5.75969219e-01 -2.68896669e-01 -1.33922382e-03
3.80090952e-01 7.46954143e-01 9.88988996e-01 -8.89923871e-01
-6.99273586e-01 -6.40844345e-01 1.06146513e-02 4.70611870e-01
-1.05575658e-01 2.06540987e-01 -7.52774775e-01 -5.82129776e-01
5.39232373e-01 8.85061175e-03 4.76349980e-01 7.15743601e-02
1.37536621e+00 -2.31505990e-01 -3.73993099e-01 1.06357551e+00
1.46911848e+00 4.50027257e-01 7.61090040e-01 7.60740817e-01
7.12765992e-01 1.21554315e-01 6.47736490e-01 1.99979022e-01
-5.53436875e-01 5.76137543e-01 1.33560240e+00 4.11682837e-02
4.52746034e-01 -4.10486087e-02 -1.95313506e-02 2.21949607e-01
9.10566654e-03 -9.18210968e-02 -9.92447078e-01 2.86460489e-01
-1.40477681e+00 -9.26621675e-01 3.03526931e-02 2.19786906e+00
2.58755535e-01 7.12248564e-01 -2.41828397e-01 4.59032059e-01
8.30990732e-01 4.26960498e-01 -6.70462072e-01 -2.94682354e-01
3.46153788e-02 3.70754331e-01 9.46213186e-01 2.68825054e-01
-1.27943957e+00 9.53325748e-01 5.22757339e+00 1.04037654e+00
-1.65894282e+00 -1.73104107e-01 1.50929168e-01 -1.92910388e-01
1.16183683e-02 -1.71836331e-01 -6.40914142e-01 5.47410667e-01
5.22977233e-01 -9.13303494e-02 3.11017454e-01 1.20713592e+00
-9.75523144e-02 5.38289905e-01 -7.19873011e-01 8.68540704e-01
-1.79206133e-01 -1.41041338e+00 2.26272032e-01 2.06438720e-01
3.27167600e-01 2.01412171e-01 3.76841933e-01 7.29854181e-02
3.41778882e-02 -8.78041089e-01 6.89061403e-01 -4.67560887e-02
5.80817759e-01 -1.37883472e+00 8.20889831e-01 6.14286363e-01
-9.13890064e-01 -1.09534837e-01 -6.81753337e-01 1.61556095e-01
-6.09524548e-02 3.15639257e-01 -6.82426631e-01 6.48160160e-01
9.35139120e-01 1.51416957e-01 -3.56862724e-01 1.08048534e+00
-2.11635187e-01 5.12028873e-01 -8.82973492e-01 -8.79965723e-02
5.20354927e-01 -8.04723725e-02 1.24035871e+00 6.05970204e-01
3.55072290e-01 1.35546029e-01 5.53240888e-02 9.50626373e-01
-9.50392988e-03 -1.55925572e-01 -6.97522640e-01 3.04910421e-01
8.26106191e-01 1.01445913e+00 -5.96164286e-01 1.19758032e-01
-3.88939911e-03 4.68422949e-01 1.28943697e-02 2.02866375e-01
-1.11915994e+00 -5.98710895e-01 1.04761922e+00 7.75522590e-02
3.47137868e-01 -3.73417139e-01 -6.34340107e-01 -7.15312958e-01
1.25338510e-01 -9.83049095e-01 7.97439367e-02 -5.04007041e-01
-1.25740457e+00 9.56779480e-01 8.48615658e-04 -1.59111404e+00
3.22685316e-02 -6.98011994e-01 -8.88296008e-01 8.46412659e-01
-1.14726961e+00 -1.08634996e+00 -3.20850760e-01 7.19781458e-01
2.27744207e-01 -4.85736996e-01 6.70124412e-01 3.35170954e-01
-4.76484507e-01 8.51749420e-01 3.09048332e-02 2.61645824e-01
2.83951521e-01 -8.16597879e-01 8.05164814e-01 1.09913909e+00
-1.06121555e-01 7.72475421e-01 8.62903118e-01 -8.38607013e-01
-1.26414084e+00 -9.97656286e-01 1.54272199e-01 -2.57171720e-01
4.88964885e-01 -4.35538054e-01 -1.12037885e+00 3.60739619e-01
-1.44133136e-01 2.11398676e-01 9.24324244e-02 -3.20322245e-01
-2.86690772e-01 -2.53581405e-01 -1.71386456e+00 6.66981816e-01
6.38532817e-01 -2.06485182e-01 -5.78695476e-01 2.23286510e-01
8.94237697e-01 -7.09491551e-01 -7.14324236e-01 8.13460827e-01
6.06162287e-02 -8.63480031e-01 1.27626288e+00 -5.34395576e-01
1.49215445e-01 -4.92703199e-01 -1.24663040e-01 -1.13756156e+00
-3.57912004e-01 -5.91370821e-01 5.33085428e-02 1.02867389e+00
1.15078025e-01 -9.47211802e-01 1.24534321e+00 5.87564886e-01
-1.32748052e-01 -7.67501116e-01 -1.54918456e+00 -8.15030634e-01
3.26410294e-01 -4.65461940e-01 9.64913428e-01 1.02066696e+00
-6.53590441e-01 -3.21789771e-01 -4.43456881e-02 1.08640170e+00
9.67055261e-01 -1.30184770e-01 9.43555355e-01 -1.23821115e+00
1.47833899e-01 -3.98672044e-01 -8.29951286e-01 -5.33288658e-01
3.25465292e-01 -6.42651320e-01 -2.87022412e-01 -7.86431134e-01
-5.79064190e-01 -7.93980539e-01 -3.88149858e-01 4.17405844e-01
-5.42135239e-02 1.78024754e-01 5.51143229e-01 3.45769018e-01
2.98697084e-01 4.61172819e-01 8.87350202e-01 -3.76095563e-01
-1.08296372e-01 5.05301714e-01 -1.65147960e-01 8.58812809e-01
9.01742458e-01 -6.97377384e-01 -2.74856001e-01 -3.87470543e-01
-5.07648885e-02 -2.66359568e-01 5.93318164e-01 -1.33729458e+00
2.39923701e-01 -2.22939596e-01 3.78949523e-01 -9.55422223e-01
5.91457188e-01 -1.42552590e+00 2.45848507e-01 7.30772495e-01
4.36115861e-02 2.38720864e-01 6.08613670e-01 5.66341877e-01
-1.43402070e-01 -3.86531025e-01 1.23755944e+00 9.41298604e-02
-3.33148092e-01 5.07414162e-01 1.04880128e-02 -3.29408288e-01
1.41627538e+00 -2.83479780e-01 -3.12765807e-01 -2.19774209e-02
-3.76370817e-01 1.25722185e-01 6.13751113e-01 5.37925839e-01
7.40129888e-01 -1.38121617e+00 -4.58628297e-01 4.03122574e-01
-3.01170051e-01 2.13803396e-01 1.99124753e-01 2.11162418e-01
-1.01313126e+00 1.44761860e-01 -4.81583029e-01 -7.67175496e-01
-1.37801683e+00 1.00818813e+00 5.63642859e-01 9.98920798e-02
-8.30989063e-01 7.25054085e-01 -3.68242376e-02 -3.92902732e-01
4.01319683e-01 -8.98719355e-02 1.56369135e-02 -4.03054297e-01
2.66263068e-01 3.95782828e-01 2.10271239e-01 -5.17954051e-01
-5.77073276e-01 5.88281393e-01 -1.55910760e-01 8.87718648e-02
1.30804098e+00 3.64274293e-01 -1.21841051e-01 -3.09138656e-01
1.15172386e+00 2.39954248e-01 -1.04756391e+00 1.30095601e-01
-3.73654872e-01 -7.58740127e-01 1.06510788e-01 -3.27657133e-01
-1.37810993e+00 8.68577063e-01 8.53243291e-01 2.82134295e-01
1.05447555e+00 -6.25652552e-01 9.18740332e-01 4.29218352e-01
4.55540419e-01 -6.21596217e-01 -2.12756157e-01 4.48925376e-01
8.27039599e-01 -8.99559438e-01 1.03428952e-01 -6.11862838e-01
-1.43579572e-01 9.09967601e-01 8.75598431e-01 -6.93179667e-01
7.72044659e-01 2.41675258e-01 6.85016736e-02 -7.89832622e-02
-2.27638066e-01 4.41445738e-01 1.31256491e-01 6.02249086e-01
-5.71141601e-01 -2.91811675e-01 -1.14406552e-02 1.67067826e-01
-3.62869114e-01 -3.79577726e-01 2.07813755e-01 8.78361464e-01
-3.41657460e-01 -7.80883670e-01 -8.10290933e-01 5.86918779e-02
-7.14174271e-01 1.39412627e-01 -3.08454424e-01 1.06355262e+00
5.72975054e-02 8.31192791e-01 9.25492123e-02 -6.57408357e-01
5.24227738e-01 -3.38462085e-01 5.85606843e-02 -1.33725047e-01
-9.28472042e-01 -3.06193113e-01 -3.95986080e-01 -6.50513172e-01
5.49225174e-02 -3.27663064e-01 -1.23653817e+00 -6.97904706e-01
-4.49830651e-01 2.70760268e-01 8.55191708e-01 6.11645520e-01
2.73571253e-01 3.31497133e-01 9.61225629e-01 -1.08052289e+00
-1.11655152e+00 -6.86159492e-01 -3.02859008e-01 4.71430838e-01
3.05830240e-01 -7.45811343e-01 -8.58909488e-01 -6.88688993e-01]
|
[7.700366973876953, -4.470816612243652]
|
13b165ba-7444-46f2-a7ec-18f5e3a6af61
|
theory-of-minds-understanding-behavior-in
|
1901.06085
| null |
http://arxiv.org/abs/1901.06085v1
|
http://arxiv.org/pdf/1901.06085v1.pdf
|
Theory of Minds: Understanding Behavior in Groups Through Inverse Planning
|
Human social behavior is structured by relationships. We form teams, groups,
tribes, and alliances at all scales of human life. These structures guide
multi-agent cooperation and competition, but when we observe others these
underlying relationships are typically unobservable and hence must be inferred.
Humans make these inferences intuitively and flexibly, often making rapid
generalizations about the latent relationships that underlie behavior from just
sparse and noisy observations. Rapid and accurate inferences are important for
determining who to cooperate with, who to compete with, and how to cooperate in
order to compete. Towards the goal of building machine-learning algorithms with
human-like social intelligence, we develop a generative model of multi-agent
action understanding based on a novel representation for these latent
relationships called Composable Team Hierarchies (CTH). This representation is
grounded in the formalism of stochastic games and multi-agent reinforcement
learning. We use CTH as a target for Bayesian inference yielding a new
algorithm for understanding behavior in groups that can both infer hidden
relationships as well as predict future actions for multiple agents interacting
together. Our algorithm rapidly recovers an underlying causal model of how
agents relate in spatial stochastic games from just a few observations. The
patterns of inference made by this algorithm closely correspond with human
judgments and the algorithm makes the same rapid generalizations that people
do.
|
['Max Kleiman-Weiner', 'Michael Shum', 'Michael L. Littman', 'Joshua B. Tenenbaum']
|
2019-01-18
| null | null | null | null |
['action-understanding']
|
['computer-vision']
|
[-2.73358762e-01 3.77176970e-01 2.79269278e-01 -3.89096588e-01
1.47240786e-02 -5.00515223e-01 8.83276701e-01 2.22872362e-01
-2.35012919e-01 9.53263879e-01 4.06328619e-01 -1.86174046e-02
-7.38411844e-01 -1.05838168e+00 -2.99540132e-01 -5.93733132e-01
-4.46009755e-01 1.31343699e+00 2.25504115e-01 -5.78804672e-01
1.42920837e-01 4.63326313e-02 -1.40103126e+00 2.23219767e-01
6.44244790e-01 1.46187440e-01 2.08377242e-01 1.00793290e+00
2.43203208e-01 1.61256552e+00 -6.54649794e-01 -5.19521236e-01
5.75480685e-02 -7.71563828e-01 -9.23767209e-01 3.72564822e-01
-5.14635623e-01 -4.26972389e-01 -1.37331873e-01 6.98717535e-01
-9.12222117e-02 3.38525206e-01 1.04411590e+00 -1.43436658e+00
-6.08342528e-01 1.14257646e+00 -4.18857455e-01 1.38807148e-01
7.04083502e-01 1.50737971e-01 1.31523466e+00 -3.18410732e-02
5.93426943e-01 1.63461185e+00 4.74771231e-01 4.85573173e-01
-1.36946344e+00 -2.61993706e-01 3.78325313e-01 2.41916567e-01
-1.22279644e+00 -1.81509197e-01 3.85871351e-01 -6.30574346e-01
7.13307798e-01 2.65923083e-01 9.62854922e-01 1.01029682e+00
4.77144331e-01 6.45149469e-01 1.24902439e+00 -2.12369800e-01
4.67267871e-01 -2.47595832e-01 2.90819798e-02 8.22153687e-01
2.58763712e-02 1.34113982e-01 -8.16840529e-01 -6.11162066e-01
1.08114457e+00 2.53498256e-01 2.70019203e-01 -1.48892596e-01
-1.09164333e+00 9.60615337e-01 1.72015205e-01 6.22246452e-02
-6.63470209e-01 4.34636861e-01 -2.65219778e-01 7.02941775e-01
3.00768048e-01 4.69784468e-01 -3.14113379e-01 -3.10662389e-01
-1.80256277e-01 6.29245400e-01 1.19366693e+00 4.95979428e-01
9.11869228e-01 -3.56383502e-01 3.29766482e-01 6.53066635e-01
4.46839482e-01 3.88455570e-01 2.02557240e-02 -1.56371760e+00
-5.71260112e-04 5.72810352e-01 3.81719202e-01 -1.18753350e+00
-6.02364302e-01 -2.84824818e-01 -6.55368328e-01 3.65717769e-01
5.91079593e-01 -3.79660785e-01 -1.90214187e-01 1.99160767e+00
3.83761764e-01 1.48547277e-01 5.92039749e-02 5.09854138e-01
1.82859153e-01 4.79207903e-01 -4.38245796e-02 -2.83240348e-01
1.00531685e+00 -4.74245846e-01 -4.00554717e-01 -1.54822096e-01
5.88030040e-01 -2.23166078e-01 5.25140464e-01 4.20585901e-01
-1.08569098e+00 -2.39839897e-01 -6.42416418e-01 3.90660316e-01
1.42390490e-01 -7.09568202e-01 1.20716071e+00 4.76690501e-01
-1.27150083e+00 7.46612549e-01 -1.04065621e+00 -3.37812394e-01
4.57184277e-02 5.41739821e-01 -5.85042015e-02 3.49400401e-01
-9.83636260e-01 7.99545884e-01 6.70833066e-02 9.65046212e-02
-1.25706685e+00 -8.79112706e-02 -5.46710193e-01 -6.54916689e-02
7.99045265e-01 -9.93073940e-01 1.41376352e+00 -8.30588341e-01
-1.53064883e+00 6.20146751e-01 3.64390314e-02 -4.32010531e-01
2.93931186e-01 1.25606939e-01 -1.16294622e-01 -1.03706896e-01
3.71947199e-01 4.46486026e-01 6.09689713e-01 -1.19829226e+00
-8.09442699e-01 -6.82631969e-01 4.56151009e-01 6.34836733e-01
1.66165993e-01 1.41974613e-01 3.03062379e-01 -2.10171014e-01
9.44267139e-02 -1.19358408e+00 -6.29179597e-01 -5.24714768e-01
-1.38967037e-01 -7.08574712e-01 1.35725901e-01 -1.26306176e-01
7.01891482e-01 -1.81473696e+00 7.19851077e-01 3.74661386e-01
1.00123906e+00 -7.73932517e-01 -1.13986738e-01 8.78279090e-01
4.51033771e-01 2.25413833e-02 1.99161053e-01 -2.06633911e-01
3.94939840e-01 5.74206412e-01 -1.41554371e-01 4.93340880e-01
-2.73684293e-01 8.27710211e-01 -1.08183098e+00 -2.38625079e-01
9.62329879e-02 -1.61533773e-01 -7.04378188e-01 5.66753805e-01
-4.47553545e-01 5.58508873e-01 -9.07929957e-01 1.72929063e-01
-2.76511200e-02 -6.65937424e-01 8.72401714e-01 8.38243246e-01
6.31740019e-02 2.40056440e-01 -1.25185287e+00 1.10354149e+00
-1.73648685e-01 3.83469671e-01 1.08777083e-01 -8.64932954e-01
8.38604391e-01 3.91003609e-01 4.97187108e-01 -9.20699015e-02
2.48580292e-01 -9.14461091e-02 5.19366384e-01 -3.71517271e-01
1.52097791e-01 -2.54647940e-01 -5.27099133e-01 1.30404377e+00
-1.24985859e-01 -4.31340158e-01 2.87556738e-01 6.32130086e-01
1.34464586e+00 -7.47653237e-03 5.68421960e-01 -3.11650574e-01
-7.22135045e-03 -1.29067525e-03 6.88563406e-01 1.28528821e+00
-1.23620490e-02 -1.11434862e-01 8.45084906e-01 -9.07576084e-01
-8.36883426e-01 -1.34338140e+00 5.67100167e-01 1.44853950e+00
1.70637697e-01 -4.32099044e-01 -5.50505161e-01 -7.31006637e-02
-5.59718832e-02 5.17557561e-01 -1.00487149e+00 3.16569023e-02
-3.59322786e-01 -7.60197759e-01 1.62617579e-01 3.71774763e-01
3.47504973e-01 -1.21704078e+00 -6.02850556e-01 5.17807484e-01
-9.64086726e-02 -9.38138187e-01 -1.20618209e-01 7.75529221e-02
-5.84556341e-01 -1.22994971e+00 9.85141192e-03 -2.78654397e-01
4.04770702e-01 1.20661527e-01 1.28452671e+00 4.56149876e-01
-2.12051287e-01 7.64019012e-01 -3.53355855e-01 -4.66898918e-01
-7.55355000e-01 -3.10327053e-01 5.81739843e-01 -3.62405963e-02
3.90397757e-01 -1.17357767e+00 -2.13704392e-01 6.23013675e-01
-4.49476182e-01 1.96670070e-01 2.45653138e-01 6.28272533e-01
-1.85254693e-01 3.21245104e-01 1.79771513e-01 -8.25704396e-01
1.04544663e+00 -7.43630528e-01 -5.94071388e-01 3.51792336e-01
1.44439312e-02 1.49164319e-01 5.28767884e-01 -4.35206145e-01
-1.16078746e+00 -4.39982235e-01 5.11170387e-01 2.83338040e-01
-3.68726283e-01 4.25221056e-01 2.08044499e-01 4.08141166e-01
7.17515051e-01 -1.61272548e-02 1.61116228e-01 -1.11442909e-01
3.32647949e-01 5.62113702e-01 1.10021882e-01 -1.02450621e+00
6.13418698e-01 5.25983334e-01 4.61041788e-03 -8.81432891e-01
-7.91193306e-01 -5.79944812e-02 -7.42773056e-01 -5.61697662e-01
9.15582538e-01 -7.95586586e-01 -1.67145026e+00 5.80565691e-01
-1.02318299e+00 -8.27512324e-01 -9.91540849e-02 5.16288757e-01
-9.04518425e-01 7.07418472e-02 -8.34367692e-01 -1.13230062e+00
6.19506836e-01 -9.25888658e-01 7.20336318e-01 1.45244047e-01
-8.67502928e-01 -1.23772538e+00 4.86005396e-01 6.20065093e-01
1.24270044e-01 -1.85104851e-02 7.86803126e-01 -7.09138930e-01
-9.52848256e-01 4.53602254e-01 3.64793986e-01 -3.22959781e-01
2.30318457e-01 6.06197640e-02 -2.00650051e-01 -1.48603469e-01
1.49573386e-03 -5.56410193e-01 3.68903935e-01 5.60391366e-01
4.83657151e-01 -1.28117755e-01 -2.77412623e-01 -4.15461920e-02
7.31040955e-01 4.06824738e-01 2.90038884e-01 -4.47948612e-02
3.55759442e-01 9.67961788e-01 1.83758438e-01 7.76596308e-01
9.46977615e-01 6.51259422e-01 2.93608069e-01 4.04842764e-01
3.98053199e-01 -2.77374774e-01 4.23477709e-01 8.45360696e-01
-6.84211016e-01 -2.68564284e-01 -9.93526280e-01 2.56607443e-01
-2.33327937e+00 -1.29845428e+00 -1.43431902e-01 1.72150218e+00
7.36930609e-01 1.26406714e-01 4.44179475e-01 -3.58665526e-01
5.99699378e-01 2.49726884e-02 -5.09650350e-01 -3.95645380e-01
-2.79352181e-02 -1.52175143e-01 5.83686642e-02 9.88650024e-01
-5.16066074e-01 1.00569117e+00 7.42570066e+00 4.63083655e-01
1.55984368e-02 9.28857774e-02 7.28611767e-01 2.41912249e-02
-4.36978847e-01 2.60857731e-01 -7.04697728e-01 -3.40558924e-02
8.40667248e-01 -1.40964970e-01 1.08445954e+00 2.70369977e-01
4.41903561e-01 -4.24306184e-01 -1.29167974e+00 7.17785180e-01
-2.14395091e-01 -1.22669351e+00 -1.76765904e-01 5.15875697e-01
8.68527591e-01 -1.84502974e-01 -2.35384986e-01 1.43635064e-01
2.07557821e+00 -1.18027377e+00 6.50884509e-01 5.31875551e-01
-9.01003480e-02 -5.61756313e-01 3.09968770e-01 9.07074809e-01
-9.10050929e-01 -4.35186177e-01 -3.19884300e-01 -1.19766212e+00
1.09768279e-01 2.12283909e-01 -8.09326172e-01 1.33060124e-02
5.10208488e-01 6.78314805e-01 -1.45820260e-01 2.92965651e-01
-5.78684509e-01 3.55394542e-01 -4.09449548e-01 -5.41039526e-01
2.66286343e-01 -6.93746150e-01 5.19601345e-01 1.64795429e-01
-6.45506084e-02 7.69683957e-01 4.73431319e-01 9.51122701e-01
2.88367540e-01 -2.89399981e-01 -6.90350652e-01 -1.35991156e-01
4.44941759e-01 8.24771166e-01 -9.73051250e-01 -2.66084105e-01
-1.46969169e-01 7.83860564e-01 6.98613763e-01 3.52748901e-01
-5.94590902e-01 4.59478825e-01 1.00450456e+00 -7.58302808e-02
-5.97715750e-02 -6.14179850e-01 1.20933704e-01 -1.21918643e+00
-5.05357742e-01 -1.07340264e+00 4.07140702e-01 -7.30497122e-01
-1.46501708e+00 3.39107871e-01 2.65170515e-01 -5.31220555e-01
-8.69486392e-01 -3.03941399e-01 -5.63663244e-01 4.09251809e-01
-4.00741279e-01 -8.88594091e-01 9.78905782e-02 8.00212443e-01
3.91615301e-01 -4.04523432e-01 8.93505216e-01 -6.65758014e-01
-3.49590600e-01 -3.06090653e-01 -7.44876489e-02 8.72790590e-02
1.59854084e-01 -1.43195343e+00 3.57196212e-01 3.08245271e-01
5.20499170e-01 7.11695731e-01 8.20897937e-01 -8.54249716e-01
-1.07635236e+00 -1.85649991e-01 4.55973446e-01 -8.91300976e-01
1.18687773e+00 -5.95181942e-01 -4.94453430e-01 1.05863011e+00
3.03436577e-01 -5.50608516e-01 9.79433239e-01 9.83069837e-01
-2.00007021e-01 9.55786183e-02 -6.46499038e-01 1.01864183e+00
1.32044375e+00 -5.84106326e-01 -8.42691958e-01 5.06370187e-01
4.54171032e-01 1.25732765e-01 -7.19653666e-01 -2.27274761e-01
5.95710814e-01 -1.48998094e+00 6.82664037e-01 -8.70947838e-01
3.78517032e-01 -2.30552316e-01 -1.85288951e-01 -1.60649371e+00
-6.15022004e-01 -9.99549270e-01 3.50359380e-01 5.90422332e-01
4.35390323e-01 -8.19984794e-01 9.38640773e-01 8.54443491e-01
4.41419125e-01 -2.47387409e-01 -6.47414088e-01 -6.37951791e-01
-3.39153744e-02 -3.24933708e-01 6.65403366e-01 9.63831365e-01
2.91294873e-01 5.06009161e-01 -5.90026498e-01 1.82494476e-01
1.21045423e+00 2.39423320e-01 1.06164408e+00 -1.48990583e+00
-1.07617092e+00 -2.71058321e-01 -4.45710868e-01 -1.10223639e+00
2.56689459e-01 -2.98811883e-01 -1.77031681e-02 -1.34974039e+00
5.56383669e-01 -5.26520371e-01 2.70374179e-01 3.05679053e-01
4.67814133e-02 -1.84491649e-01 1.50922582e-01 4.14765924e-01
-1.08546054e+00 4.25294995e-01 1.10920906e+00 1.69415236e-01
-2.93211073e-01 2.77106941e-01 -7.44286597e-01 1.20293391e+00
7.14008033e-01 -5.70447326e-01 -5.32342494e-01 -2.79104978e-01
7.95430183e-01 7.00712502e-01 4.97832417e-01 -5.98227262e-01
6.98293626e-01 -7.69008040e-01 1.35911465e-01 -9.57402885e-02
3.78993750e-01 -4.40720648e-01 6.40365899e-01 6.19455755e-01
-5.73144317e-01 1.28058195e-01 -5.48376858e-01 7.53236771e-01
9.13528875e-02 2.45965011e-02 1.99989319e-01 -6.19150937e-01
-5.77944219e-01 8.90913606e-02 -1.06955540e+00 1.24184214e-01
1.09017563e+00 -6.40238523e-02 -2.18483210e-01 -1.25824022e+00
-1.38716662e+00 5.03801703e-01 3.91377538e-01 2.08350942e-01
5.12672544e-01 -8.72127712e-01 -1.07077944e+00 -1.52524382e-01
-3.43234479e-01 -4.44123119e-01 3.13265651e-01 5.11073649e-01
-3.99020255e-01 -4.37048674e-02 -3.47959787e-01 -4.83948022e-01
-1.19199681e+00 2.42689133e-01 3.94496351e-01 -3.79510224e-01
-3.42884272e-01 9.98394430e-01 4.73291934e-01 -5.09896934e-01
-1.14300504e-01 -1.29191115e-01 -2.61557549e-01 -1.82517618e-01
6.45492077e-01 2.47393146e-01 -7.25237310e-01 -3.40622902e-01
-4.35741395e-02 2.85140097e-01 -1.03430368e-01 -3.91213417e-01
1.63585138e+00 -4.27806109e-01 -6.34051740e-01 8.50093663e-01
3.42077643e-01 -7.66099840e-02 -1.34199178e+00 -2.42450133e-01
-9.21717733e-02 -4.60138708e-01 -5.49382746e-01 -3.94504845e-01
-3.84444624e-01 3.74569952e-01 -3.52572113e-01 1.05996263e+00
5.30245602e-01 5.22526085e-01 8.84605572e-02 6.41368330e-01
9.47753072e-01 -9.39311922e-01 4.69138205e-01 5.20587862e-01
6.64658904e-01 -1.11261785e+00 1.05874524e-01 -3.99165213e-01
-7.17252195e-01 8.97316456e-01 4.24245954e-01 -3.10031533e-01
7.37541318e-01 4.48879302e-01 -1.34262085e-01 -6.76380098e-01
-1.45914054e+00 -2.06554994e-01 -3.02343786e-01 7.54546225e-01
-2.96072736e-02 6.01604819e-01 1.69821844e-01 1.81221619e-01
-4.59951192e-01 -3.18633080e-01 9.00694728e-01 6.21717811e-01
-6.68561280e-01 -1.28057158e+00 -4.67303872e-01 4.52713847e-01
-7.95484856e-02 1.02882430e-01 -7.30513453e-01 6.82039261e-01
-7.20170066e-02 1.39407861e+00 1.67411923e-01 -4.13076431e-01
-1.80162519e-01 -4.37722236e-01 7.86698699e-01 -8.71076524e-01
-4.59227324e-01 -1.49988145e-01 3.17490667e-01 -5.97102463e-01
-7.56226063e-01 -1.00610340e+00 -1.03896797e+00 -8.37814867e-01
1.27443701e-01 5.05372643e-01 -2.85039507e-02 1.37434721e+00
-1.38332576e-01 3.19756150e-01 6.12512648e-01 -6.93948984e-01
-6.89852178e-01 -7.97901273e-01 -1.03811789e+00 3.66070628e-01
-3.75940539e-02 -8.28671992e-01 -4.25654709e-01 -1.03896521e-01]
|
[3.917890787124634, 1.5873730182647705]
|
974d25a5-3b77-4f20-9718-634f172cce03
|
camera-pose-voting-for-large-scale-image
| null | null |
http://openaccess.thecvf.com/content_iccv_2015/html/Zeisl_Camera_Pose_Voting_ICCV_2015_paper.html
|
http://openaccess.thecvf.com/content_iccv_2015/papers/Zeisl_Camera_Pose_Voting_ICCV_2015_paper.pdf
|
Camera Pose Voting for Large-Scale Image-Based Localization
|
Image-based localization approaches aim to determine the camera pose from which an image was taken. Finding correct 2D-3D correspondences between query image features and 3D points in the scene model becomes harder as the size of the model increases. Current state-of-the-art methods therefore combine elaborate matching schemes with camera pose estimation techniques that are able to handle large fractions of wrong matches. In this work we study the benefits and limitations of spatial verification compared to appearance-based filtering. We propose a voting-based pose estimation strategy that exhibits O(n) complexity in the number of matches and thus facilitates to consider much more matches than previous approaches - whose complexity grows at least quadratically. This new outlier rejection formulation enables us to evaluate pose estimation for 1-to-many matches and to surpass the state-of-the-art. At the same time, we show that using more matches does not automatically lead to a better performance.
|
['Bernhard Zeisl', 'Marc Pollefeys', 'Torsten Sattler']
|
2015-12-01
| null | null | null |
iccv-2015-12
|
['image-based-localization']
|
['computer-vision']
|
[ 4.74015549e-02 -3.49496901e-01 1.72123894e-01 -1.83699504e-01
-1.06336904e+00 -6.43502414e-01 5.50284207e-01 5.00185311e-01
-7.92494833e-01 2.44568944e-01 -4.06594127e-01 -1.29005954e-01
4.41458225e-02 -5.04587531e-01 -8.45312476e-01 -2.61100262e-01
8.19209665e-02 7.00630069e-01 8.11945558e-01 -6.32369965e-02
6.00608826e-01 1.03874195e+00 -1.69362235e+00 -2.63616413e-01
4.17962372e-01 1.09614766e+00 -1.62648447e-02 8.90794098e-01
1.81363374e-01 5.73716573e-02 -4.73472804e-01 -4.48767543e-01
6.18639171e-01 -1.82676330e-01 -4.82647687e-01 2.41064280e-01
1.34314418e+00 -2.85704970e-01 -1.97150409e-01 1.18912625e+00
4.13673908e-01 -3.49423587e-02 4.16706353e-01 -1.12324607e+00
2.39436459e-02 -2.44892314e-01 -5.81560671e-01 9.45008546e-02
8.08538556e-01 -1.03222176e-01 7.35145152e-01 -1.20930040e+00
8.07445407e-01 9.07242537e-01 1.14845717e+00 2.62679849e-02
-1.29089558e+00 -2.80430347e-01 8.70935544e-02 1.82965308e-01
-1.85017335e+00 -4.00595009e-01 5.50941646e-01 -3.73694330e-01
8.70998681e-01 3.91808569e-01 6.17317736e-01 4.81232941e-01
6.87924493e-03 2.34360322e-01 8.94627988e-01 -6.54118061e-01
9.14081186e-02 -4.62042280e-02 -5.54240718e-02 8.40986013e-01
6.66945994e-01 4.57267649e-02 -5.62936723e-01 -3.67711902e-01
9.30658281e-01 1.01930968e-01 -1.14098422e-01 -9.27137434e-01
-1.27108991e+00 5.39989054e-01 4.13286775e-01 2.85165101e-01
-3.01399976e-01 3.62286478e-01 4.12005782e-02 3.01838249e-01
2.02401400e-01 6.21452153e-01 -4.14024830e-01 -4.03490961e-02
-1.24515271e+00 3.03169399e-01 6.29393041e-01 1.12275517e+00
1.02089274e+00 -3.64919722e-01 3.69540423e-01 3.23418140e-01
1.36466682e-01 6.57928407e-01 4.87666354e-02 -1.02458501e+00
2.01202109e-01 6.90776408e-01 4.31167841e-01 -1.37802088e+00
-4.54877436e-01 -3.55487376e-01 -5.13420701e-01 4.90731418e-01
7.49143243e-01 3.07599366e-01 -7.97678292e-01 1.27201295e+00
4.68466431e-01 -2.20855549e-02 -3.55933815e-01 8.69729638e-01
1.61232904e-01 1.24532253e-01 -5.96044838e-01 -1.22290298e-01
1.24569321e+00 -5.63377559e-01 -2.87520379e-01 -2.51265317e-01
5.21294773e-01 -1.25274217e+00 6.31473780e-01 4.88132805e-01
-1.00405693e+00 -5.51148236e-01 -1.08027124e+00 6.67292923e-02
-3.06940466e-01 2.97578663e-01 1.72441319e-01 6.49055541e-01
-1.14884210e+00 6.38118267e-01 -7.03741908e-01 -6.56443834e-01
-3.54251377e-02 8.09191644e-01 -7.33449101e-01 -4.87160459e-02
-4.83964652e-01 1.20587373e+00 1.34917110e-01 -6.32889047e-02
-3.23744476e-01 -4.24474746e-01 -8.54106009e-01 -2.64515787e-01
5.86267948e-01 -6.38953626e-01 1.26523650e+00 -7.19031990e-01
-1.02643752e+00 1.06768811e+00 -5.38601995e-01 -3.52272600e-01
9.31486011e-01 -3.06401610e-01 -7.30652064e-02 2.78106123e-01
1.29693881e-01 3.94335687e-01 8.99030566e-01 -1.28889334e+00
-6.91708565e-01 -6.29395902e-01 -2.88435798e-02 1.47869438e-01
1.14044450e-01 -6.75329566e-02 -8.33963454e-01 -3.75301301e-01
8.45504522e-01 -1.14272213e+00 -4.83169407e-01 4.78921860e-01
-1.33754134e-01 1.10470876e-01 5.80908120e-01 -3.51618171e-01
9.71288383e-01 -2.05544686e+00 -8.93037021e-02 6.34292841e-01
1.53863996e-01 1.91987649e-01 1.50966179e-02 5.68721056e-01
9.47680846e-02 -2.29542285e-01 1.96044102e-01 -6.20652974e-01
-1.53036922e-01 5.02157249e-02 5.67912422e-02 1.17565358e+00
-1.07450396e-01 4.66422677e-01 -7.55178630e-01 -5.54151416e-01
5.70269704e-01 2.90261477e-01 -4.62775975e-01 -2.33172979e-02
1.64575115e-01 2.56782174e-01 -1.76723227e-01 6.02569044e-01
9.81926262e-01 -7.90931433e-02 -5.92716783e-02 -4.34745938e-01
-2.96915799e-01 -5.49837539e-04 -1.71239340e+00 1.82922971e+00
-2.83123761e-01 5.66865683e-01 -4.95974300e-03 -4.88768488e-01
9.42140162e-01 1.90642215e-02 5.32743752e-01 -5.08569121e-01
1.50488034e-01 6.28964186e-01 -2.47523829e-01 -3.78400348e-02
7.27250516e-01 2.52341833e-02 -1.45135045e-01 5.66275790e-02
-1.09033957e-01 -2.35570773e-01 3.46625596e-02 -8.09162632e-02
1.14753687e+00 9.14650187e-02 6.53173327e-01 -1.42962158e-01
7.18137145e-01 2.00419366e-01 3.87008458e-01 9.36290979e-01
-6.82872832e-02 7.67793357e-01 1.26916900e-01 -5.39294899e-01
-1.29326987e+00 -9.68349218e-01 -4.03925404e-02 5.47537804e-01
6.88289762e-01 -4.54967469e-01 -7.08948612e-01 -5.32652318e-01
2.67689824e-01 1.51045248e-01 -4.82525975e-01 1.42932117e-01
-5.90387940e-01 -1.44586861e-01 2.53317535e-01 3.75243813e-01
2.26449460e-01 -2.97672331e-01 -9.58948374e-01 3.48033197e-03
2.64829457e-01 -1.20013666e+00 -5.26424706e-01 6.34014932e-03
-9.23248112e-01 -1.30666101e+00 -5.61918914e-01 -5.49680054e-01
1.21151268e+00 3.92718494e-01 9.79491413e-01 2.90975720e-01
-3.48013282e-01 6.17181659e-01 -4.00480986e-01 -1.79890215e-01
-1.90426409e-01 3.09121404e-02 3.39131534e-01 -8.86023343e-02
4.72329259e-01 -3.70566398e-01 -5.40819287e-01 6.64740324e-01
-7.08688319e-01 -4.55688119e-01 6.23773813e-01 5.97262383e-01
8.26729119e-01 -2.09944114e-01 -3.39794070e-01 -6.01941228e-01
5.19070961e-02 3.16434234e-01 -1.25483060e+00 2.28615969e-01
-5.95649540e-01 1.68528706e-01 4.04528230e-01 -4.54231769e-01
-3.89422446e-01 9.02849317e-01 1.16973994e-02 -6.51679456e-01
-1.72407374e-01 7.41206259e-02 1.24289215e-01 -8.91294956e-01
7.19228685e-01 1.13534465e-01 -9.30014998e-02 -4.69333082e-01
2.85103112e-01 3.66325170e-01 8.44334483e-01 -3.04350615e-01
1.17758822e+00 6.53439045e-01 5.41143179e-01 -7.35860944e-01
-5.97813725e-01 -1.13497448e+00 -1.22813606e+00 -2.57800996e-01
3.97325754e-01 -8.50659192e-01 -7.89142251e-01 2.82946408e-01
-1.40592432e+00 3.91381502e-01 -1.57357544e-01 6.28578007e-01
-6.60899758e-01 6.86125457e-01 -1.94747686e-01 -9.53520119e-01
1.38929561e-01 -1.19835711e+00 1.36332214e+00 -1.50469422e-01
-2.80226231e-01 -4.27247167e-01 1.83476448e-01 -1.14970542e-02
1.44402266e-01 1.99852705e-01 9.04962048e-02 -5.07189095e-01
-8.94629180e-01 -8.60889614e-01 -1.51566267e-01 -9.75318253e-02
-3.43195736e-01 -3.37547362e-02 -8.68417144e-01 -4.43290681e-01
-2.68383082e-02 3.22123796e-01 5.38384497e-01 1.89710692e-01
5.52496195e-01 -2.01782901e-02 -3.31467777e-01 4.99717712e-01
1.82627523e+00 -1.00584537e-01 6.35710239e-01 5.12274802e-01
5.36682308e-01 4.14511621e-01 1.01420772e+00 4.06056106e-01
1.01538517e-01 1.18910348e+00 6.60114765e-01 -1.54204041e-01
1.11349232e-01 -1.97651103e-01 1.28388464e-01 5.86466908e-01
-1.69014297e-02 1.69100821e-01 -9.17815983e-01 5.11792541e-01
-1.96142542e+00 -7.34162807e-01 -4.65647042e-01 2.80763698e+00
3.38775039e-01 1.74404547e-01 2.06052676e-01 1.21181950e-01
6.80555403e-01 -2.03421548e-01 -2.48087659e-01 -7.89613053e-02
9.03112814e-02 -4.49201912e-02 1.14734077e+00 7.10465491e-01
-1.14082611e+00 7.94359446e-01 6.66605520e+00 5.78032434e-01
-8.23208928e-01 -2.90774405e-02 -1.02372197e-02 6.74327612e-02
7.26182759e-02 2.91944474e-01 -9.99958932e-01 2.11668923e-01
5.28542578e-01 1.59861535e-01 1.58312589e-01 9.76665139e-01
-1.03686638e-01 -6.32593393e-01 -1.26080799e+00 1.42299628e+00
5.57794631e-01 -1.15030921e+00 -2.37977058e-01 3.13925207e-01
6.08839929e-01 8.72736946e-02 -2.54586399e-01 -3.85387331e-01
-2.15804294e-01 -6.91156089e-01 9.64086890e-01 4.91415322e-01
6.77877188e-01 -6.86170638e-01 8.92346859e-01 4.76967216e-01
-1.34410274e+00 5.64075597e-02 -5.65764427e-01 4.25267629e-02
1.60439134e-01 5.36106825e-01 -7.71001279e-01 5.75027883e-01
5.91810882e-01 2.50372767e-01 -8.76509190e-01 1.66146863e+00
-3.67775769e-03 -2.44498804e-01 -8.25303018e-01 -6.23665452e-02
1.01709686e-01 -1.25502676e-01 5.85969388e-01 1.01964176e+00
7.11768985e-01 -1.16927288e-01 3.27790260e-01 4.09209281e-01
1.86409838e-02 2.01321840e-01 -8.75156879e-01 6.97984517e-01
5.85404694e-01 9.74708259e-01 -9.43451881e-01 -2.70674109e-01
-4.42100406e-01 1.24127889e+00 2.65941828e-01 -1.43619761e-01
-6.42130494e-01 -2.83335894e-01 4.30510670e-01 5.28731823e-01
5.38906336e-01 -7.38093972e-01 -1.10274173e-01 -1.14119637e+00
3.91910404e-01 -6.91544533e-01 1.14490271e-01 -6.26992047e-01
-9.70076323e-01 6.02506042e-01 7.93526918e-02 -1.65621483e+00
-2.74262398e-01 -7.48153508e-01 -1.33078083e-01 6.21765792e-01
-1.17755055e+00 -1.02260375e+00 -3.86115223e-01 4.09068316e-01
2.51531482e-01 2.57418364e-01 7.80980468e-01 4.16833878e-01
-3.74663882e-02 5.23473501e-01 2.06514612e-01 9.89864618e-02
8.06166410e-01 -1.05619502e+00 4.55645531e-01 1.17353296e+00
5.28882802e-01 6.29041970e-01 9.65296865e-01 -5.25555372e-01
-1.44588387e+00 -6.40242517e-01 1.22056782e+00 -8.86575997e-01
5.12845814e-01 -2.89647847e-01 -4.33737785e-01 4.92695481e-01
-2.92715967e-01 3.39976043e-01 1.75972059e-01 1.30136073e-01
-4.34014171e-01 -2.85082638e-01 -1.09258318e+00 5.03493965e-01
9.37343240e-01 -6.44198358e-01 -4.13253218e-01 1.44952148e-01
1.24731556e-01 -7.09285736e-01 -7.13528872e-01 3.28051686e-01
5.21380663e-01 -1.15343511e+00 1.19700360e+00 -4.10787575e-02
-4.93730634e-01 -9.18909132e-01 -3.32178712e-01 -7.56316841e-01
-5.78590110e-02 -6.62030637e-01 1.96585417e-01 7.81114757e-01
2.33947501e-01 -4.90198910e-01 9.64876413e-01 5.14375269e-01
1.90364406e-01 -4.74178255e-01 -1.34789252e+00 -1.13911557e+00
-5.02828658e-01 -3.91140968e-01 3.35179389e-01 6.05283678e-01
-2.26903513e-01 3.31639089e-02 -3.49561214e-01 5.21647751e-01
7.77894437e-01 1.79360375e-01 1.16453850e+00 -1.44291389e+00
-1.07552156e-01 -3.55454057e-01 -1.21589434e+00 -1.15025747e+00
-2.24424973e-01 -2.84577101e-01 1.32696614e-01 -1.08199477e+00
2.45076921e-02 -3.67147386e-01 1.07274212e-01 2.28180885e-01
7.40686879e-02 6.36889935e-01 2.58668840e-01 2.51497120e-01
-9.19320703e-01 -1.04690112e-01 6.01628602e-01 2.42903620e-01
3.80508080e-02 1.10333130e-01 -1.14853904e-01 9.23811376e-01
5.24557531e-01 -6.65720463e-01 8.14568549e-02 -3.75226676e-01
3.36630464e-01 -8.07027668e-02 6.53050840e-01 -1.35735607e+00
6.62120402e-01 9.39000696e-02 6.10555947e-01 -8.61646235e-01
5.66838920e-01 -1.33625102e+00 3.85578901e-01 4.16458219e-01
3.97934541e-02 6.25495493e-01 -7.08702952e-03 6.19682550e-01
-2.11280227e-01 -6.19273186e-01 7.68098831e-01 -2.81659096e-01
-6.94094837e-01 2.79875666e-01 -1.14926085e-01 -4.20967430e-01
1.19880080e+00 -6.84936225e-01 5.50229736e-02 -4.30769265e-01
-6.74048126e-01 -2.01396927e-01 1.19166017e+00 1.74690604e-01
5.52527130e-01 -1.27123249e+00 -3.98146033e-01 2.13926047e-01
3.62247795e-01 3.04994676e-02 -1.28853142e-01 1.15878093e+00
-1.03701735e+00 3.04100066e-01 1.16828158e-01 -9.15559769e-01
-1.83533919e+00 6.28806412e-01 2.82428831e-01 -1.59821749e-01
-3.79963577e-01 7.35345304e-01 -4.43285972e-01 -3.37767363e-01
3.77201259e-01 -3.26756030e-01 3.82332444e-01 1.47664836e-02
2.43649229e-01 4.40260082e-01 3.99205029e-01 -1.00464809e+00
-7.03072846e-01 1.22140825e+00 1.19648099e-01 -1.40566468e-01
8.79226446e-01 -1.55292168e-01 -2.27577854e-02 1.77288577e-01
1.27239621e+00 4.18400705e-01 -1.10874212e+00 -2.37113550e-01
3.18868160e-01 -1.06016409e+00 -5.71836792e-02 -2.35857815e-01
-6.56209111e-01 7.08789527e-01 8.42417121e-01 9.24894679e-03
9.22866762e-01 6.54732436e-02 4.40730125e-01 6.44129038e-01
8.68447661e-01 -1.18082750e+00 -8.28735530e-02 3.40237856e-01
5.85038126e-01 -1.29735672e+00 5.55082023e-01 -5.48518777e-01
-1.28250450e-01 1.25511420e+00 4.06950653e-01 -4.62696135e-01
4.53777134e-01 2.74708003e-01 7.15211406e-02 -1.57103449e-01
-1.38777465e-01 -4.98519033e-01 4.97905105e-01 5.12865186e-01
1.03831150e-01 -3.78101200e-01 -1.01250872e-01 -2.94930905e-01
-2.72426661e-02 -2.41012007e-01 2.60954440e-01 9.87721920e-01
-6.13036156e-01 -1.47553885e+00 -8.39832008e-01 1.78244859e-01
-2.69761384e-01 3.04380357e-02 -5.75204909e-01 1.01645112e+00
1.87751487e-01 7.33487606e-01 2.79727802e-02 -2.64432013e-01
7.59140432e-01 -1.78247958e-01 7.92894602e-01 -3.72989476e-01
-5.75375319e-01 2.84697413e-01 -2.80825924e-02 -1.00704217e+00
-6.41239285e-01 -8.77551317e-01 -8.61911237e-01 -2.87323773e-01
-7.57394433e-01 -1.26591548e-02 7.82853186e-01 6.05013251e-01
3.81982535e-01 -1.97163790e-01 6.17288411e-01 -1.15833926e+00
-7.33842432e-01 -4.98684853e-01 -4.18609053e-01 3.94405872e-01
5.82020283e-01 -6.63551152e-01 -3.98424000e-01 -1.12738006e-01]
|
[7.763842582702637, -2.2073795795440674]
|
72069244-5b7b-46b4-8d62-7d507dae34a5
|
show-control-and-tell-a-framework-for
|
1811.10652
| null |
https://arxiv.org/abs/1811.10652v3
|
https://arxiv.org/pdf/1811.10652v3.pdf
|
Show, Control and Tell: A Framework for Generating Controllable and Grounded Captions
|
Current captioning approaches can describe images using black-box architectures whose behavior is hardly controllable and explainable from the exterior. As an image can be described in infinite ways depending on the goal and the context at hand, a higher degree of controllability is needed to apply captioning algorithms in complex scenarios. In this paper, we introduce a novel framework for image captioning which can generate diverse descriptions by allowing both grounding and controllability. Given a control signal in the form of a sequence or set of image regions, we generate the corresponding caption through a recurrent architecture which predicts textual chunks explicitly grounded on regions, following the constraints of the given control. Experiments are conducted on Flickr30k Entities and on COCO Entities, an extended version of COCO in which we add grounding annotations collected in a semi-automatic manner. Results demonstrate that our method achieves state of the art performances on controllable image captioning, in terms of caption quality and diversity. Code and annotations are publicly available at: https://github.com/aimagelab/show-control-and-tell.
|
['Lorenzo Baraldi', 'Rita Cucchiara', 'Marcella Cornia']
|
2018-11-26
|
show-control-and-tell-a-framework-for-1
|
http://openaccess.thecvf.com/content_CVPR_2019/html/Cornia_Show_Control_and_Tell_A_Framework_for_Generating_Controllable_and_CVPR_2019_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2019/papers/Cornia_Show_Control_and_Tell_A_Framework_for_Generating_Controllable_and_CVPR_2019_paper.pdf
|
cvpr-2019-6
|
['controllable-image-captioning']
|
['computer-vision']
|
[ 3.63228858e-01 6.27206028e-01 -3.43905210e-01 -3.48995328e-01
-6.83073401e-01 -9.13889289e-01 8.53259027e-01 -8.25842619e-02
2.20277742e-01 7.31764019e-01 6.30988598e-01 3.45101207e-02
4.04365808e-01 -6.28286541e-01 -1.29457676e+00 -4.17482764e-01
2.15729475e-01 6.43422484e-01 -1.73142955e-01 -3.78376901e-01
-1.51474968e-01 6.84050620e-02 -1.41008782e+00 7.27329850e-01
7.27642119e-01 8.82570684e-01 5.39802194e-01 7.03808606e-01
4.45165858e-03 1.04997599e+00 -2.45334655e-01 -4.05250788e-01
1.09330341e-01 -6.80948913e-01 -7.37777948e-01 4.96554524e-01
5.48082769e-01 -1.46544566e-02 -5.04649818e-01 8.01965535e-01
2.94205785e-01 -2.08948106e-01 5.00880301e-01 -1.33539164e+00
-1.26278448e+00 1.04529774e+00 -1.12340450e-01 -3.48639190e-01
4.85543579e-01 4.00155127e-01 1.17346215e+00 -6.97937131e-01
1.11496496e+00 1.03031039e+00 2.75315106e-01 9.96917188e-01
-1.60057163e+00 -4.90890235e-01 3.80628020e-01 2.81539578e-02
-1.31263804e+00 -4.79168177e-01 6.55588806e-01 -5.46889901e-01
5.85088015e-01 3.07685196e-01 6.57631457e-01 1.39958072e+00
-2.38620654e-01 1.03442478e+00 8.40379000e-01 -3.23800951e-01
1.82935223e-01 4.09928143e-01 -2.58509666e-01 4.73963052e-01
-9.47622135e-02 -1.56400502e-01 -4.35954243e-01 1.92136586e-01
7.69784212e-01 -2.98590869e-01 -6.41576946e-01 -6.48443341e-01
-1.77473032e+00 8.17140162e-01 9.34853852e-01 2.28709653e-01
-3.39160055e-01 4.84551758e-01 1.73663929e-01 -4.33769077e-02
3.84785295e-01 8.84161592e-01 -2.55611777e-01 1.50187477e-01
-9.01610434e-01 3.22688490e-01 6.61139369e-01 1.50538850e+00
5.81263363e-01 -1.32673621e-01 -7.92431355e-01 4.47571993e-01
-2.60946732e-02 7.84865141e-01 3.39908719e-01 -7.38811135e-01
7.17648745e-01 3.50152791e-01 5.83211839e-01 -9.63355482e-01
-9.28236842e-02 -4.12482142e-01 -9.09476876e-01 -3.49649400e-01
1.53274402e-01 -1.24831043e-01 -1.26883173e+00 2.00675535e+00
-1.80481859e-02 2.67415911e-01 9.27235484e-02 1.18665063e+00
9.41043496e-01 1.06310821e+00 3.23288701e-02 3.03652287e-02
1.34541512e+00 -1.31403971e+00 -8.30968797e-01 -4.02726859e-01
4.52201366e-01 -4.09606993e-01 1.21279371e+00 -1.25078097e-01
-1.07876325e+00 -5.17537296e-01 -8.45976710e-01 -9.48587209e-02
-2.56087095e-01 2.41719961e-01 2.41199598e-01 3.75456549e-03
-1.17304730e+00 1.98401302e-01 -4.35965598e-01 -2.51371473e-01
4.25764650e-01 1.05659030e-01 -3.90459716e-01 -1.94384325e-02
-1.31001496e+00 7.65565932e-01 5.36612272e-01 1.52343258e-01
-1.14377034e+00 -7.18979955e-01 -1.07899785e+00 2.00759843e-01
2.70154566e-01 -8.54058385e-01 1.19967723e+00 -1.35133779e+00
-1.09087276e+00 9.38762546e-01 -3.95770408e-02 -7.14380741e-01
7.16696501e-01 -1.35429233e-01 -1.17683314e-01 2.96730369e-01
2.10525006e-01 1.42600131e+00 8.18788588e-01 -1.61285746e+00
-1.16056092e-01 1.85825616e-01 3.41905832e-01 2.33857334e-01
-9.66026932e-02 -2.22931713e-01 -6.56501293e-01 -5.89719951e-01
-3.80119443e-01 -1.34537256e+00 -3.62190843e-01 -1.35916583e-02
-8.68242085e-01 2.20176101e-01 4.16944325e-01 -5.28066874e-01
1.12492156e+00 -2.09558368e+00 5.73183417e-01 -8.88753682e-02
1.56870216e-01 5.07874154e-02 -2.93400973e-01 5.46227098e-01
-2.51626968e-01 5.74736238e-01 -4.07441109e-01 -4.25422698e-01
2.09788278e-01 1.08523160e-01 -6.68749213e-01 1.52017534e-01
6.13774538e-01 1.34025347e+00 -1.07555509e+00 -6.29625559e-01
2.02314079e-01 5.93266547e-01 -4.42734301e-01 4.52482402e-01
-1.02651477e+00 5.92998326e-01 -5.07875323e-01 3.70002866e-01
2.44917557e-01 -8.34321022e-01 1.53094605e-01 -2.49681368e-01
1.01831881e-02 -9.39751863e-02 -7.77438223e-01 1.80272520e+00
-6.98769033e-01 8.35001171e-01 -1.94615424e-01 -4.92263228e-01
8.43932092e-01 5.78325391e-01 7.56834522e-02 -3.99502873e-01
-8.97177085e-02 2.81927995e-02 -4.51674581e-01 -4.09696341e-01
6.22709095e-01 7.16600642e-02 -4.29165423e-01 2.25954190e-01
-4.36556339e-02 -3.19620699e-01 3.18192720e-01 4.67655063e-01
8.88935149e-01 3.46135736e-01 1.22881152e-01 -8.25699642e-02
2.54200727e-01 3.43158931e-01 1.25927955e-01 8.44342887e-01
1.35013938e-01 1.40948927e+00 6.89208388e-01 -4.44108903e-01
-1.68106735e+00 -8.40279758e-01 -1.10679008e-01 7.15629041e-01
2.19534814e-01 -4.28042442e-01 -8.29266667e-01 -4.09786165e-01
-1.98273003e-01 8.33853185e-01 -1.06803012e+00 1.66779950e-01
-5.08592367e-01 -2.68050790e-01 3.26012760e-01 3.16210836e-01
4.40727502e-01 -1.42236662e+00 -4.45494950e-01 1.49898931e-01
-6.66119576e-01 -1.49888468e+00 -7.60564923e-01 -1.44450411e-01
-2.97568142e-01 -5.59847474e-01 -1.05801344e+00 -8.84961843e-01
8.97599578e-01 8.95564109e-02 1.50379276e+00 6.56380579e-02
1.21038165e-02 3.86115223e-01 -4.70250994e-01 -2.54457772e-01
-6.54917777e-01 3.69553477e-01 -4.20969993e-01 2.35733092e-01
-2.88711816e-01 -3.07781100e-01 -5.57779968e-01 2.04462633e-01
-1.19332016e+00 9.14844275e-01 7.03684270e-01 7.92681873e-01
7.38932014e-01 -6.70605838e-01 4.31773841e-01 -8.93974900e-01
3.78916442e-01 -6.96334302e-01 -4.94778663e-01 4.67589080e-01
-1.53489709e-01 3.79885763e-01 6.55479610e-01 -5.93991220e-01
-8.30674231e-01 6.31451130e-01 4.09283549e-01 -8.07625473e-01
-1.67439103e-01 3.93043846e-01 -2.07022801e-01 4.41593617e-01
7.71268725e-01 4.98658746e-01 -1.59753829e-01 -7.13749751e-02
9.46769655e-01 4.72906917e-01 6.34146512e-01 -4.77312863e-01
8.91053915e-01 3.83147150e-01 -2.46767104e-01 -4.67430890e-01
-1.06777275e+00 -1.44528747e-01 -6.31486297e-01 -4.23133373e-01
1.26492286e+00 -1.17830110e+00 -2.02135727e-01 7.92843010e-03
-1.40185654e+00 -4.57271039e-01 -2.30844632e-01 -6.11340404e-02
-9.20489132e-01 -2.10536703e-01 -2.82931089e-01 -4.13163513e-01
-3.36343855e-01 -1.21374488e+00 1.39703238e+00 1.99604213e-01
-2.44578242e-01 -7.14691579e-01 -1.06239200e-01 3.75422776e-01
4.00148571e-01 7.62293100e-01 5.05948246e-01 -3.77434134e-01
-1.04357529e+00 -1.37071162e-01 -2.83182174e-01 -2.60169208e-02
-1.76107347e-01 -4.80543971e-02 -8.39601159e-01 -8.32241848e-02
-5.51200509e-01 -6.77704751e-01 7.29001760e-01 1.96414545e-01
1.05767739e+00 -8.52566481e-01 -3.27763140e-01 4.69126999e-01
1.67037165e+00 -3.18932176e-01 8.44463110e-01 4.36151147e-01
8.73641729e-01 6.27138257e-01 5.17837107e-01 3.91019642e-01
4.50422674e-01 9.03283596e-01 8.09639215e-01 -1.76561415e-01
-1.84767947e-01 -7.31370032e-01 2.13741571e-01 4.22195315e-01
2.86249787e-01 -7.27364779e-01 -9.62324739e-01 8.97083461e-01
-2.14434361e+00 -1.16403282e+00 -9.14252102e-02 1.75957513e+00
1.00046849e+00 -8.34564716e-02 -3.46669555e-02 -5.01166582e-01
1.13371503e+00 3.82297307e-01 -5.80247402e-01 -2.07869589e-01
-3.49716455e-01 -4.40425962e-01 4.33099449e-01 5.25519550e-01
-9.92078304e-01 1.08998871e+00 5.21730566e+00 5.25146008e-01
-1.17457533e+00 -4.99061495e-02 9.26587105e-01 -1.69523761e-01
-7.54865944e-01 -3.45505513e-02 -6.20749772e-01 6.16022170e-01
9.13270533e-01 -2.38058463e-01 4.34015721e-01 7.18173027e-01
2.96180934e-01 3.22821200e-01 -1.26011562e+00 9.26884770e-01
1.91047519e-01 -1.89751470e+00 4.39961880e-01 -3.63169983e-02
1.04044175e+00 9.88856331e-03 2.91686028e-01 1.88303858e-01
1.63408369e-01 -1.25087059e+00 1.32060361e+00 7.36111403e-01
9.44061220e-01 -2.90819913e-01 5.30298114e-01 1.64307073e-01
-9.20932174e-01 1.50023133e-01 -1.16538785e-01 1.35085091e-01
4.16343421e-01 3.25928003e-01 -1.02576244e+00 4.40405935e-01
3.81324828e-01 7.94648945e-01 -6.06021285e-01 7.47372925e-01
-4.75944877e-01 4.57117736e-01 -7.88036268e-03 -3.26746285e-01
5.21570861e-01 2.25171335e-02 5.73906302e-01 1.25916874e+00
2.81849831e-01 1.46162599e-01 3.69468033e-02 1.37983680e+00
-3.27993512e-01 2.59954389e-02 -7.97012210e-01 -2.14150280e-01
4.82590318e-01 1.24082148e+00 -6.37530386e-01 -4.87432837e-01
-7.24861920e-02 1.07944858e+00 3.34823817e-01 4.02417243e-01
-1.19870627e+00 3.48648801e-02 3.46640885e-01 3.58222663e-01
5.34057140e-01 5.27385846e-02 -1.58049002e-01 -1.58008003e+00
1.90911382e-01 -8.72125745e-01 -4.14698012e-02 -1.54919338e+00
-9.93702292e-01 1.07978535e+00 1.01608224e-01 -1.38421679e+00
-3.79993707e-01 -1.91179857e-01 -3.57070595e-01 6.16532624e-01
-1.26143444e+00 -1.49767721e+00 -6.18727922e-01 2.29137227e-01
5.98757803e-01 2.95227379e-01 7.36982942e-01 -5.26702516e-02
-3.66846442e-01 2.64921755e-01 -8.39891583e-02 1.31782562e-01
6.26297235e-01 -1.28867304e+00 5.40298879e-01 7.39713907e-01
4.12841350e-01 1.98516592e-01 1.31344521e+00 -4.09184515e-01
-1.11467314e+00 -1.48367548e+00 1.00497055e+00 -6.07047975e-01
6.62837386e-01 -8.08509111e-01 -7.59595037e-01 8.77736747e-01
6.11226201e-01 1.18720397e-01 2.39629775e-01 -3.28254253e-01
-4.27864254e-01 1.57359421e-01 -8.07306409e-01 8.23169470e-01
9.28757846e-01 -4.80669260e-01 -3.63202482e-01 7.27584302e-01
1.28985441e+00 -5.78740478e-01 -6.31414175e-01 1.71398818e-01
3.44615221e-01 -6.74733758e-01 7.44501054e-01 -5.43429613e-01
1.04784822e+00 -5.63266873e-01 -1.15386255e-01 -1.26928031e+00
-3.60580266e-01 -7.92059362e-01 9.33924690e-02 1.34071827e+00
9.45810676e-01 4.46607359e-02 7.51109183e-01 6.96707129e-01
-1.78203851e-01 -6.75040543e-01 -4.46701914e-01 -5.62714338e-01
-1.25054166e-01 -1.67627707e-01 7.89695859e-01 8.60731483e-01
-2.41942573e-02 5.78559637e-01 -7.79782653e-01 2.16970250e-01
3.95241529e-01 3.74903023e-01 7.52595961e-01 -5.62515676e-01
-1.93984807e-01 -1.15030885e-01 -3.34750146e-01 -1.03864443e+00
2.88263500e-01 -9.62619424e-01 3.07409197e-01 -1.76735246e+00
3.68969709e-01 -2.45735019e-01 1.42069533e-01 6.49690211e-01
-9.59491543e-03 5.08494139e-01 7.02908695e-01 4.46164280e-01
-9.55826581e-01 6.31996810e-01 1.38810098e+00 -3.92146021e-01
-4.20711339e-02 -4.68005091e-01 -7.11108208e-01 2.14833081e-01
8.56475651e-01 -2.67652988e-01 -4.63972121e-01 -6.89592242e-01
4.57677573e-01 3.08998704e-01 4.40317690e-01 -9.17882204e-01
-3.33097875e-02 -9.53177959e-02 2.31872857e-01 -3.33975315e-01
4.37951863e-01 -6.65177405e-01 5.69818616e-01 2.03634545e-01
-9.42266166e-01 1.28592744e-01 1.71077728e-01 6.23016894e-01
-3.55252862e-01 -5.81286848e-02 6.24874830e-01 -3.16855669e-01
-6.07153118e-01 3.41732085e-01 -5.09788245e-02 1.18238986e-01
1.20065308e+00 1.49592370e-01 -4.77980524e-01 -9.43820894e-01
-9.76097524e-01 4.50002640e-01 8.67596090e-01 7.03137159e-01
5.62553287e-01 -1.57486522e+00 -1.05674720e+00 -2.46404782e-01
6.05796695e-01 9.48931798e-02 1.67459428e-01 3.04543734e-01
-6.57895386e-01 6.94345415e-01 -2.28494123e-01 -7.63169527e-01
-8.78660679e-01 8.27220738e-01 3.52902979e-01 -2.46631190e-01
-5.46576679e-01 6.32443607e-01 5.31164944e-01 -1.49757430e-01
-5.83097376e-02 -3.56451124e-01 -1.72403276e-01 1.24823228e-02
4.89344895e-01 -5.32121062e-01 -4.44985300e-01 -9.30705309e-01
-9.18171033e-02 3.44598144e-01 7.62473792e-02 -2.74715900e-01
1.30195081e+00 -3.10621887e-01 6.07167296e-02 3.70940983e-01
1.09950900e+00 -3.60630661e-01 -1.49888134e+00 -1.06510604e-02
-2.16663256e-01 -2.63262451e-01 -2.33701766e-01 -9.87376034e-01
-8.80154550e-01 6.42700970e-01 1.98525876e-01 3.20492268e-01
9.24040616e-01 4.94101137e-01 4.86044914e-01 1.11075066e-01
1.64457321e-01 -5.51152885e-01 2.82242060e-01 2.53529459e-01
1.52016652e+00 -1.28049934e+00 -4.85075265e-01 -3.47737640e-01
-1.13117480e+00 8.14839363e-01 6.38399363e-01 -1.87610567e-01
8.06783140e-02 -6.22193031e-02 -6.26893193e-02 2.70778760e-02
-1.09526038e+00 -2.21806884e-01 3.37949514e-01 4.85133588e-01
3.49211663e-01 1.85990989e-01 9.00521353e-02 3.73685479e-01
-3.19132417e-01 -7.58775920e-02 8.77372980e-01 3.55346620e-01
-2.24372417e-01 -7.83280730e-01 -3.60904753e-01 1.44205689e-01
-3.04538786e-01 -3.93106520e-01 -4.89303708e-01 8.46794665e-01
1.10619949e-04 7.58722544e-01 1.00598603e-01 -2.00969666e-01
1.58188134e-01 -9.47013795e-02 2.24132538e-01 -8.47003281e-01
-3.60528857e-01 -1.90979898e-01 2.78473020e-01 -5.38801789e-01
-5.01495302e-01 -6.53893709e-01 -1.09515417e+00 1.44757360e-01
-1.95232585e-01 2.77533591e-01 6.34135485e-01 6.11325741e-01
6.78948760e-01 4.49473053e-01 5.79368830e-01 -9.44173574e-01
-1.93824142e-01 -9.88480270e-01 -1.63064286e-01 6.96523190e-01
4.94435102e-01 -1.59907088e-01 -3.64330173e-01 5.66311479e-01]
|
[10.998027801513672, 0.9797388315200806]
|
d814b3d6-c5d1-46ad-9c79-8881e71f4465
|
scene-text-telescope-text-focused-scene-image
| null | null |
http://openaccess.thecvf.com//content/CVPR2021/html/Chen_Scene_Text_Telescope_Text-Focused_Scene_Image_Super-Resolution_CVPR_2021_paper.html
|
http://openaccess.thecvf.com//content/CVPR2021/papers/Chen_Scene_Text_Telescope_Text-Focused_Scene_Image_Super-Resolution_CVPR_2021_paper.pdf
|
Scene Text Telescope: Text-Focused Scene Image Super-Resolution
|
Image super-resolution, which is often regarded as a preprocessing procedure of scene text recognition, aims to recover the realistic features from a low-resolution text image. It has always been challenging due to large variations in text shapes, fonts, backgrounds, etc. However, most existing methods employ generic super-resolution frameworks to handle scene text images while ignoring text-specific properties such as text-level layouts and character-level details. In this paper, we establish a text-focused super-resolution framework, called Scene Text Telescope (STT). In terms of text-level layouts, we propose a Transformer-Based Super-Resolution Network (TBSRN) containing a Self-Attention Module to extract sequential information, which is robust to tackle the texts in arbitrary orientations. In terms of character-level details, we propose a Position-Aware Module and a Content-Aware Module to highlight the position and the content of each character. By observing that some characters look indistinguishable in low-resolution conditions, we use a weighted cross-entropy loss to tackle this problem. We conduct extensive experiments, including text recognition with pre-trained recognizers and image quality evaluation, on TextZoom and several scene text recognition benchmarks to assess the super-resolution images. The experimental results show that our STT can indeed generate text-focused super-resolution images and outperform the existing methods in terms of recognition accuracy.
|
['xiangyang xue', 'Bin Li', 'Jingye Chen']
|
2021-06-19
| null | null | null |
cvpr-2021-1
|
['scene-text-recognition']
|
['computer-vision']
|
[ 8.78324807e-01 -6.43514872e-01 1.82051152e-01 -3.36556405e-01
-7.51187921e-01 -4.14846897e-01 8.54607880e-01 -3.72497946e-01
-1.26363244e-02 4.65722352e-01 4.82417554e-01 7.93635622e-02
-1.17564902e-01 -7.97680736e-01 -7.51772523e-01 -9.05262232e-01
6.88197494e-01 2.76325166e-01 4.49318856e-01 -4.37447727e-01
5.44924438e-01 4.53597605e-01 -1.59586549e+00 7.79934227e-01
1.14925122e+00 6.55836761e-01 5.86105406e-01 7.58010387e-01
-3.61359358e-01 7.95793593e-01 -6.74960494e-01 -4.14216816e-01
1.82512239e-01 -4.24038827e-01 -5.51952481e-01 5.24590671e-01
8.75656962e-01 -5.19581556e-01 -5.65466762e-01 1.19252253e+00
4.33678359e-01 1.97043940e-02 5.29837012e-01 -3.32271188e-01
-9.83072758e-01 7.28313446e-01 -9.23448443e-01 4.78251845e-01
3.28106970e-01 -8.19563195e-02 8.11773539e-01 -1.05225015e+00
5.79842746e-01 1.37940872e+00 4.23628718e-01 2.45904654e-01
-1.22914457e+00 -4.73607570e-01 3.12121600e-01 1.92801386e-01
-1.39500654e+00 -4.15127695e-01 8.51626754e-01 -1.72569931e-01
5.38621247e-01 4.07573372e-01 -1.11645631e-01 1.38008142e+00
2.42315829e-01 7.49618709e-01 1.19943511e+00 -4.39512819e-01
-4.28775214e-02 2.39094868e-01 3.10525633e-02 3.28675866e-01
2.70689130e-01 -1.93428025e-01 -6.88498855e-01 4.00410801e-01
1.21954513e+00 1.63676038e-01 -4.12024528e-01 -3.20969731e-01
-1.30054736e+00 4.08595920e-01 3.27289343e-01 5.29068351e-01
-8.29422697e-02 -5.10512173e-01 2.68519521e-01 -1.02321193e-01
5.72014391e-01 1.26458049e-01 -8.20884407e-02 1.39567107e-01
-1.04088843e+00 -6.88444227e-02 3.16094518e-01 1.10281014e+00
3.56670707e-01 4.42867041e-01 -5.56295097e-01 1.16815794e+00
-2.25390792e-01 5.78032792e-01 6.59289598e-01 -3.02985162e-01
8.50448430e-01 6.19330883e-01 6.30701706e-02 -1.02376401e+00
-1.73015103e-01 -6.54425442e-01 -1.29916787e+00 -8.32839683e-02
1.45488143e-01 2.20825240e-01 -9.65710104e-01 1.23904538e+00
2.11021751e-01 1.32812291e-01 1.39523029e-01 1.16066217e+00
8.55085135e-01 6.25728190e-01 -4.40337062e-01 -2.64532149e-01
1.45920932e+00 -1.00797486e+00 -7.74274349e-01 -2.16902524e-01
-1.54237542e-02 -9.72355604e-01 1.29587317e+00 3.38396817e-01
-1.03237605e+00 -8.08322012e-01 -1.20598555e+00 -3.18792164e-01
-3.19369167e-01 3.09277147e-01 2.18507405e-02 6.13422751e-01
-7.83681750e-01 4.98764187e-01 -4.59222704e-01 -3.16703022e-01
3.84957999e-01 -9.28971451e-03 -1.85635701e-01 -3.82015586e-01
-9.71766174e-01 5.30717909e-01 3.62245232e-01 7.28542209e-02
-6.02608562e-01 -4.93158549e-01 -7.16448545e-01 2.93480158e-01
5.43235660e-01 -5.15550852e-01 7.81479716e-01 -9.30482686e-01
-1.61539614e+00 5.57347000e-01 -4.28878665e-01 -5.39593361e-02
6.12590373e-01 -1.50497869e-01 -5.92158973e-01 3.02729160e-01
3.01370258e-03 -7.58822039e-02 1.56641459e+00 -1.25873387e+00
-7.17225909e-01 -5.69098890e-01 -2.31629148e-01 5.72380245e-01
-4.51273412e-01 1.75845489e-01 -6.17375255e-01 -9.84902620e-01
1.93640560e-01 -4.15015757e-01 2.99259666e-02 -3.32632095e-01
-5.67368925e-01 2.28327215e-01 1.09222949e+00 -8.04574072e-01
1.05307376e+00 -2.03999925e+00 3.07189792e-01 -3.65557075e-01
2.35429525e-01 2.19398528e-01 -1.65173754e-01 3.63259055e-02
-6.03316277e-02 1.35901913e-01 -6.37923554e-02 -3.81464809e-01
-1.93602353e-01 -1.92714468e-01 -5.40391386e-01 1.97104439e-01
2.63721466e-01 7.12371230e-01 -5.38255155e-01 -6.41139448e-01
6.91226602e-01 9.67669904e-01 -1.67066187e-01 -6.51373565e-02
-2.30401844e-01 4.48721081e-01 -6.10358775e-01 4.97800857e-01
1.13658047e+00 -3.83223772e-01 5.08177541e-02 -4.52229768e-01
-3.31833273e-01 1.22144841e-01 -1.13815880e+00 1.52609038e+00
-4.75806743e-01 7.24512875e-01 5.67841940e-02 -6.65606916e-01
9.36063647e-01 -7.77237564e-02 1.80220291e-01 -1.10852706e+00
2.81797647e-02 -1.97462216e-01 -5.36825776e-01 -3.24528337e-01
1.09105778e+00 1.61515459e-01 1.04234561e-01 2.06908107e-01
-3.56926203e-01 1.81481332e-01 5.44245206e-02 1.64072648e-01
7.91264355e-01 1.85769290e-01 -1.18103974e-01 -8.67193788e-02
6.75008714e-01 -2.01649860e-01 2.09435582e-01 8.15305889e-01
2.39762783e-01 1.12740052e+00 2.56149501e-01 -3.48296463e-01
-1.47037709e+00 -8.41978669e-01 -4.36640561e-01 1.25753856e+00
4.36072797e-01 -2.76079744e-01 -8.11455369e-01 -3.30537558e-01
-4.33079004e-01 5.63634634e-01 -5.20404279e-01 1.33815393e-01
-6.71988904e-01 -9.58145440e-01 2.79181808e-01 5.13617396e-01
1.01972234e+00 -7.50693083e-01 -4.26151991e-01 -8.49511400e-02
-3.53884339e-01 -1.71132839e+00 -8.20653915e-01 7.53922462e-02
-7.80444145e-01 -6.33836508e-01 -8.75862479e-01 -7.46103525e-01
6.87122226e-01 7.53491938e-01 7.92961836e-01 -2.87567466e-01
-4.71491784e-01 1.44076850e-02 -5.70919871e-01 2.24037737e-01
-1.20746829e-01 1.24525182e-01 -2.64199585e-01 4.21435654e-01
1.10473432e-01 -4.20149416e-01 -5.13781786e-01 4.87127841e-01
-1.05069566e+00 5.20651639e-01 8.86204183e-01 9.04811025e-01
7.37991929e-01 7.14502692e-01 -9.74504203e-02 -8.39581668e-01
5.33796787e-01 3.47432122e-02 -6.42181396e-01 4.09901053e-01
-3.93734992e-01 1.77407473e-01 1.13747847e+00 -5.03803492e-01
-1.57152021e+00 -6.95121661e-02 2.96972424e-01 -5.14017284e-01
-3.16000581e-01 9.66012105e-02 -5.45943022e-01 -9.66672134e-03
5.49271286e-01 1.02145350e+00 -5.76186001e-01 -7.36321449e-01
2.07541883e-01 8.32401991e-01 6.97521806e-01 -4.71044600e-01
1.20057809e+00 6.37664557e-01 -1.74977370e-02 -1.13951886e+00
-1.04727507e+00 -2.38448068e-01 -8.80996346e-01 2.64946759e-01
8.09351921e-01 -9.14866328e-01 -3.89021218e-01 5.35999179e-01
-9.20192838e-01 -8.56678188e-02 -2.31702961e-02 1.21945359e-01
-2.96844184e-01 7.24660218e-01 -5.51748514e-01 -6.66972578e-01
-3.86809200e-01 -1.11958218e+00 1.52272475e+00 4.30814654e-01
6.40747368e-01 -7.75749385e-01 -2.56797820e-01 5.81557870e-01
6.26790404e-01 1.16891272e-01 8.98469746e-01 -3.17381352e-01
-9.00626898e-01 1.01088487e-01 -8.15448105e-01 1.96046293e-01
2.52688050e-01 -5.38282916e-02 -1.01827872e+00 -3.12533528e-01
1.28566707e-02 -8.40253476e-03 9.30732846e-01 2.65613645e-01
1.15212178e+00 -3.91944081e-01 -5.13876230e-02 1.04233122e+00
1.55604172e+00 -1.30476639e-01 8.42118740e-01 3.80237013e-01
1.10081673e+00 5.30678332e-01 4.88279969e-01 5.68223298e-01
2.25747794e-01 9.10964906e-01 2.20646575e-01 -1.05761588e-01
-2.87689000e-01 -3.58728528e-01 3.32648188e-01 6.01614773e-01
-4.52498123e-02 -3.83231223e-01 -4.59704071e-01 1.56576335e-01
-1.67956388e+00 -1.14302278e+00 -1.74503163e-01 2.17011523e+00
6.72266126e-01 1.70648798e-01 -1.64320827e-01 7.04692155e-02
1.18930149e+00 4.56715018e-01 -8.82165790e-01 -1.91194922e-01
-7.69352376e-01 7.54590780e-02 4.77037370e-01 2.94216692e-01
-1.06701815e+00 1.27498996e+00 5.63041687e+00 1.24329579e+00
-1.18433845e+00 -1.51158795e-01 5.48079967e-01 -1.10016719e-01
-2.32682735e-01 -3.44352841e-01 -1.03631139e+00 4.80062425e-01
5.74834466e-01 -2.17922822e-01 7.92773366e-01 5.84683836e-01
2.00632781e-01 2.30584890e-01 -8.00740063e-01 1.12347579e+00
4.72830832e-01 -1.21333206e+00 5.93300164e-01 -1.64626718e-01
7.64171362e-01 -2.66327620e-01 4.42873776e-01 1.44801572e-01
1.02316707e-01 -1.21191013e+00 5.87804139e-01 4.12148416e-01
1.23634696e+00 -7.73193598e-01 3.91620964e-01 2.51475781e-01
-1.56914544e+00 4.76598181e-02 -7.16881394e-01 3.49497437e-01
-2.67338932e-01 5.40190458e-01 -4.93656397e-01 9.73752081e-01
8.74761164e-01 9.27373648e-01 -8.32497418e-01 5.96709788e-01
-5.64286187e-02 1.87425196e-01 -7.08869919e-02 1.40800804e-01
-4.48813252e-02 -2.87856698e-01 6.32805169e-01 1.21289229e+00
1.84274659e-01 -1.78995598e-02 -4.81700748e-02 1.17208338e+00
-1.35028824e-01 1.92936525e-01 -2.46587858e-01 1.69227656e-03
3.30944955e-01 1.38236666e+00 -7.61026621e-01 -1.73897251e-01
-3.40230286e-01 1.52468967e+00 -7.24150008e-03 5.52808762e-01
-7.56034374e-01 -5.81240714e-01 2.57399529e-01 1.05452575e-01
7.36365497e-01 9.52745788e-03 -5.01331985e-01 -1.61947250e+00
3.98302287e-01 -1.14884913e+00 1.17585674e-01 -1.10970080e+00
-9.92600381e-01 9.43650663e-01 -2.52659231e-01 -1.34750915e+00
9.61738527e-02 -6.03597164e-01 -4.32239980e-01 9.42920506e-01
-1.67376626e+00 -1.43703151e+00 -7.26549804e-01 8.04671407e-01
1.21801865e+00 5.44910319e-04 2.83484429e-01 6.28206134e-02
-9.21666026e-01 7.57135153e-01 5.91924012e-01 1.63145766e-01
6.89613461e-01 -1.15027630e+00 4.56267595e-01 1.11838984e+00
-6.82489621e-03 3.96773428e-01 7.53088415e-01 -6.28791630e-01
-1.63673067e+00 -1.18998563e+00 2.80890703e-01 -6.47072375e-01
4.29632306e-01 -6.65138602e-01 -1.17602479e+00 4.88860667e-01
1.06500901e-01 -1.51721731e-01 8.86525586e-03 -1.36443570e-01
-5.89098155e-01 -1.07057199e-01 -1.04617238e+00 7.09803700e-01
1.06074715e+00 -5.89412808e-01 -5.22021651e-01 2.66158015e-01
8.72083306e-01 -5.92314541e-01 -7.07698882e-01 3.76989156e-01
4.45988029e-01 -1.24888456e+00 1.13205028e+00 -1.85709134e-01
6.60551786e-01 -3.37359488e-01 -2.16750994e-01 -9.05076742e-01
-6.21304035e-01 -4.82131630e-01 1.75691366e-01 1.41996539e+00
4.69924286e-02 -3.65225971e-01 6.90460742e-01 2.16798916e-01
3.68727557e-02 -2.28021175e-01 -8.49402845e-01 -6.15332842e-01
8.32041129e-02 -1.54160950e-02 6.90552056e-01 9.48690593e-01
-2.74182796e-01 6.57458007e-01 -7.09461331e-01 4.44181502e-01
8.59633565e-01 4.51425076e-01 6.21765375e-01 -1.06677759e+00
-5.04989773e-02 -6.06276035e-01 -1.16655029e-01 -1.22955394e+00
-3.18022817e-02 -4.49107558e-01 1.35196308e-02 -1.41022229e+00
6.28379285e-01 1.06572367e-01 -1.12699211e-01 -8.70150477e-02
-3.42515737e-01 1.89593881e-01 4.38813958e-03 2.88450420e-01
-8.43347430e-01 6.70986891e-01 1.53442597e+00 -2.73694158e-01
-1.36436895e-01 -2.10386738e-01 -8.09395134e-01 5.72296023e-01
6.29869461e-01 -2.42077447e-02 -2.73242533e-01 -6.27428293e-01
9.72195044e-02 6.42073825e-02 3.06170940e-01 -7.88579404e-01
3.32572520e-01 -2.22251788e-01 8.02722394e-01 -1.04243243e+00
2.85527796e-01 -7.44762838e-01 -1.94086760e-01 -8.38258862e-02
-4.80683237e-01 -3.26659501e-01 2.49020204e-01 6.15747690e-01
-9.58950445e-02 -8.15224275e-02 9.69722271e-01 4.65015173e-02
-4.74756360e-01 2.43218362e-01 -1.53696150e-01 6.20074794e-02
6.20510340e-01 -4.83045548e-01 -7.99845994e-01 -1.95275649e-01
-9.11062136e-02 5.28071560e-02 8.39949310e-01 5.69068730e-01
9.25882638e-01 -1.03477669e+00 -9.43366468e-01 3.11218709e-01
1.32302999e-01 2.49090731e-01 7.45384276e-01 5.76818049e-01
-2.67440319e-01 5.98109066e-01 -1.23986386e-01 -7.08493531e-01
-1.32743144e+00 9.19123113e-01 3.09858203e-01 -4.10472482e-01
-1.07213080e+00 4.57567841e-01 6.86722815e-01 -1.03647120e-01
2.56445944e-01 -2.66996205e-01 -4.26364213e-01 -2.93325216e-01
9.60169375e-01 2.86595762e-01 -1.36899546e-01 -7.61535525e-01
-4.25203517e-03 1.22893929e+00 -5.65723300e-01 1.90997198e-01
1.16118443e+00 -5.59207976e-01 6.72435537e-02 2.28299409e-01
7.61331618e-01 9.58417207e-02 -1.50176930e+00 -7.30584383e-01
-1.75938338e-01 -7.92772233e-01 2.92404324e-01 -7.58866012e-01
-1.03993821e+00 9.67237294e-01 5.52617252e-01 5.25111072e-02
1.40389597e+00 -3.02208096e-01 6.16267562e-01 3.22168469e-01
2.30678409e-01 -1.06348157e+00 3.19971025e-01 5.58734179e-01
8.56186628e-01 -1.27821779e+00 -2.08225008e-02 -5.11946738e-01
-6.93951905e-01 1.23335338e+00 6.90175772e-01 1.44469246e-01
-6.24030344e-02 4.17683482e-01 -2.86242098e-01 1.69229031e-01
-7.72486091e-01 -2.19556555e-01 3.23312968e-01 5.51647127e-01
2.15967402e-01 -2.44408280e-01 3.30103338e-01 5.49831390e-01
-1.42799899e-01 -2.84387976e-01 7.00084031e-01 5.20432949e-01
-6.02140367e-01 -7.11275220e-01 -7.12018549e-01 2.87827343e-01
-4.47979510e-01 -4.65852231e-01 -5.26510239e-01 4.49519366e-01
-2.59369016e-01 8.41234446e-01 7.20268115e-02 -5.31699598e-01
3.77068579e-01 -2.23261327e-01 4.79489714e-01 -4.01942939e-01
-2.99412429e-01 4.00247931e-01 -3.29878151e-01 -3.37929994e-01
-2.51727730e-01 -5.10147929e-01 -9.08203125e-01 -4.86220986e-01
-5.09555697e-01 -1.58668756e-01 6.54634595e-01 6.80796385e-01
3.75641286e-01 8.89852226e-01 9.74847555e-01 -8.86609674e-01
-4.48123395e-01 -9.94040132e-01 -7.04084337e-01 3.88357788e-01
4.23468590e-01 -2.79409200e-01 -5.17343819e-01 1.88132748e-01]
|
[11.424978256225586, -1.8380905389785767]
|
f0c80d2c-bb93-410a-a03d-3580bf3da920
|
unsupervised-extractive-summarization-of
|
2306.01444
| null |
https://arxiv.org/abs/2306.01444v1
|
https://arxiv.org/pdf/2306.01444v1.pdf
|
Unsupervised Extractive Summarization of Emotion Triggers
|
Understanding what leads to emotions during large-scale crises is important as it can provide groundings for expressed emotions and subsequently improve the understanding of ongoing disasters. Recent approaches trained supervised models to both detect emotions and explain emotion triggers (events and appraisals) via abstractive summarization. However, obtaining timely and qualitative abstractive summaries is expensive and extremely time-consuming, requiring highly-trained expert annotators. In time-sensitive, high-stake contexts, this can block necessary responses. We instead pursue unsupervised systems that extract triggers from text. First, we introduce CovidET-EXT, augmenting (Zhan et al. 2022)'s abstractive dataset (in the context of the COVID-19 crisis) with extractive triggers. Second, we develop new unsupervised learning models that can jointly detect emotions and summarize their triggers. Our best approach, entitled Emotion-Aware Pagerank, incorporates emotion information from external sources combined with a language understanding module, and outperforms strong baselines. We release our data and code at https://github.com/tsosea2/CovidET-EXT.
|
['Cornelia Caragea', 'Junyi Jessy Li', 'Hongli Zhan', 'Tiberiu Sosea']
|
2023-06-02
| null | null | null | null |
['unsupervised-extractive-summarization', 'abstractive-text-summarization', 'extractive-summarization']
|
['natural-language-processing', 'natural-language-processing', 'natural-language-processing']
|
[ 2.55812287e-01 3.15120310e-01 -3.10850382e-01 -4.93664324e-01
-1.26038313e+00 -7.20390439e-01 6.88184261e-01 1.03485727e+00
-5.10981321e-01 9.12042558e-01 1.41335261e+00 7.45971575e-02
1.82248697e-01 -5.60977936e-01 -2.59757489e-01 -2.22988784e-01
-2.56617695e-01 5.10064423e-01 -4.86442000e-01 -2.76642591e-01
3.05391610e-01 1.95444435e-01 -1.21102178e+00 7.46013284e-01
7.68357277e-01 7.22502351e-01 -2.92412043e-01 8.94101858e-01
-2.91814774e-01 1.43684566e+00 -8.07407737e-01 -3.12075496e-01
-2.44081527e-01 -5.14380753e-01 -1.03497505e+00 -1.80315211e-01
-2.34153762e-01 -2.87481606e-01 -7.19482973e-02 5.41933417e-01
6.35964930e-01 1.90508351e-01 8.93126130e-01 -1.23498511e+00
-3.96000266e-01 8.49806964e-01 -4.76898193e-01 3.40619773e-01
7.23606646e-01 1.73150916e-02 1.19890583e+00 -9.35324550e-01
6.78744674e-01 1.19444227e+00 5.74758828e-01 3.39121938e-01
-1.03540468e+00 -4.15815949e-01 3.41847837e-01 1.24579377e-01
-7.13138938e-01 -6.31535530e-01 8.15904319e-01 -5.26928186e-01
1.36410415e+00 4.46411580e-01 4.35643524e-01 1.41026616e+00
-2.27048278e-01 1.12429142e+00 1.16261339e+00 -2.03820080e-01
3.62039447e-01 -2.61323035e-01 4.23712492e-01 2.88683832e-01
3.07749026e-02 -6.04315817e-01 -1.01071954e+00 -4.93913025e-01
5.00955358e-02 1.82862371e-01 -4.07898605e-01 6.96080923e-01
-1.26920950e+00 8.42381954e-01 2.87788928e-01 1.98675990e-01
-1.20913482e+00 -9.80603509e-03 8.16050529e-01 2.47415558e-01
1.23667753e+00 7.91032612e-01 -6.41071737e-01 -5.88686705e-01
-9.58473206e-01 1.51554644e-01 1.08028746e+00 2.57761806e-01
7.21961319e-01 -1.15135245e-01 -2.78192163e-01 7.79828787e-01
-3.41762155e-01 6.05704129e-01 6.64000437e-02 -9.96006489e-01
5.45535862e-01 6.56792402e-01 4.46503818e-01 -1.37456906e+00
-8.36316168e-01 -1.55149341e-01 -8.70091379e-01 -4.27187681e-01
1.28672924e-02 -8.55571151e-01 -6.01410449e-01 1.65745795e+00
2.66241252e-01 5.81847839e-02 4.01412427e-01 9.38118100e-01
1.00778866e+00 1.04527819e+00 3.53720218e-01 -6.75020039e-01
1.44576764e+00 -5.59781551e-01 -1.02467215e+00 -6.08603179e-01
8.65861297e-01 -6.30244315e-01 1.09025645e+00 5.17333329e-01
-9.53591645e-01 3.85660678e-01 -5.77090740e-01 -6.51948899e-03
-3.81986141e-01 4.37338389e-02 6.19901538e-01 -2.36586750e-01
-8.70137930e-01 2.45551869e-01 -5.63395381e-01 -7.29909241e-01
2.88129240e-01 -2.68628806e-01 -2.72078395e-01 2.88886279e-01
-1.69313896e+00 9.34952796e-01 2.61699587e-01 -2.22291816e-02
-6.89504325e-01 -4.73986596e-01 -7.77149558e-01 9.63783916e-03
6.16449535e-01 -3.74648035e-01 1.40708411e+00 -7.89019942e-01
-1.16928637e+00 7.35291064e-01 -4.51393306e-01 -3.47368836e-01
1.04306497e-01 -7.59768784e-01 -2.61450738e-01 5.87773263e-01
3.56261104e-01 4.93282527e-01 3.67624164e-01 -1.10537994e+00
-4.15328771e-01 -1.12461336e-01 -1.06397785e-01 4.47357714e-01
-6.71893835e-01 6.58261240e-01 1.96854901e-02 -6.13437891e-01
-3.62050653e-01 -3.88342798e-01 -4.42467451e-01 -8.53752792e-01
-5.79647541e-01 -4.25332725e-01 3.93552274e-01 -1.14987290e+00
1.43797171e+00 -1.75413597e+00 1.65244907e-01 7.37107173e-03
2.70998448e-01 -1.63670659e-01 -1.29241943e-01 1.17495191e+00
-2.42072120e-02 3.84455979e-01 -3.62072855e-01 -3.70195836e-01
7.49387294e-02 1.85862228e-01 -9.41763520e-01 -9.82003659e-02
6.20701015e-01 8.57152045e-01 -1.17747283e+00 -4.38327581e-01
-2.31247276e-01 2.25556925e-01 -4.42703158e-01 3.35862726e-01
-3.20888638e-01 3.04073602e-01 -6.11449540e-01 6.91721678e-01
-4.14198674e-02 -3.23343158e-01 7.43223801e-02 1.04539078e-02
-2.11783454e-01 5.83762765e-01 -6.32524431e-01 1.13934350e+00
-4.61830527e-01 7.10017025e-01 1.43563017e-01 -1.05844414e+00
8.34940016e-01 6.30006194e-01 6.56131268e-01 -4.07956421e-01
-7.30428547e-02 -8.09618607e-02 -1.04048491e+00 -8.44364047e-01
5.86973608e-01 -1.85074404e-01 -8.35715234e-01 9.93068278e-01
-2.71739334e-01 -2.74513453e-01 1.55586705e-01 8.81746411e-01
1.44391131e+00 -5.31053782e-01 5.07713258e-01 1.34188488e-01
-8.02088231e-02 3.72513354e-01 8.28580678e-01 6.36740446e-01
-9.06955972e-02 4.99318302e-01 1.16659892e+00 -5.31393170e-01
-6.10426009e-01 -6.12972319e-01 4.73506361e-01 1.19784141e+00
-4.19754356e-01 -8.11227918e-01 -4.28793043e-01 -5.72994411e-01
-2.96356469e-01 9.33363438e-01 -6.71058893e-01 1.10176064e-01
-4.03879255e-01 -8.38743269e-01 6.23705149e-01 4.60651129e-01
1.87279969e-01 -1.38035882e+00 -9.79971766e-01 3.04713011e-01
-1.23058712e+00 -1.12103939e+00 -9.20136720e-02 3.42026591e-01
-3.91207784e-01 -1.01408994e+00 -3.74724090e-01 -1.69701084e-01
4.14809525e-01 -1.31117418e-01 1.40329754e+00 -6.59142286e-02
-7.70038813e-02 6.35293126e-01 -6.93270564e-01 -8.41353178e-01
1.31666940e-02 3.18874456e-02 2.22825497e-01 -7.29384422e-02
5.68709671e-01 -5.68473279e-01 -4.63207722e-01 -3.94050866e-01
-1.03659463e+00 3.93643379e-02 2.67605662e-01 5.10413647e-01
2.12148204e-01 -2.58636981e-01 1.16796255e+00 -7.93806732e-01
1.25346267e+00 -1.03247762e+00 1.47844374e-01 1.23315014e-01
-2.94200778e-01 -2.58639067e-01 4.30312097e-01 -7.21728355e-02
-1.00269318e+00 -1.96876034e-01 1.82533771e-01 -1.41600817e-02
-3.52974802e-01 1.16104388e+00 2.33704716e-01 1.07401466e+00
7.44874120e-01 -1.55890524e-01 -4.70410973e-01 -3.31612170e-01
5.53219795e-01 9.74121571e-01 7.24799037e-01 -5.42267621e-01
4.54716504e-01 3.80966842e-01 -6.56084180e-01 -8.50472271e-01
-1.36049044e+00 -6.68252051e-01 -1.84307933e-01 -6.62182868e-01
7.80884564e-01 -1.28119302e+00 -4.16503400e-01 2.90920228e-01
-1.58812726e+00 -5.89711249e-01 -2.05080479e-01 4.75290924e-01
-3.80335569e-01 1.34884343e-01 -8.83226871e-01 -1.09201562e+00
-7.56560326e-01 -2.50072151e-01 1.07448649e+00 2.04668432e-01
-1.01498604e+00 -9.17744696e-01 4.06669557e-01 3.28811437e-01
3.75023723e-01 8.13882530e-01 5.54445148e-01 -1.06156290e+00
1.49976835e-01 -2.80271351e-01 -8.24756473e-02 -1.57818750e-01
1.79494455e-01 -7.74679109e-02 -9.23262239e-01 2.08624616e-01
-8.29695314e-02 -1.14798987e+00 1.10784245e+00 1.84523165e-01
7.72553682e-01 -1.06922162e+00 -5.61218522e-02 -1.09417044e-01
9.11149919e-01 -2.65524447e-01 2.19346404e-01 2.62665093e-01
3.15722793e-01 1.14973235e+00 7.11204529e-01 1.14550519e+00
8.64182174e-01 4.99139773e-04 1.14880547e-01 -3.66106331e-01
5.39014041e-01 -1.13787442e-01 7.02115417e-01 1.00295782e+00
1.79366004e-02 -3.11098874e-01 -1.26367879e+00 9.75526035e-01
-2.06253290e+00 -1.20294428e+00 -7.46184215e-02 1.36260545e+00
1.19480491e+00 -5.49570546e-02 9.11508203e-02 4.71681133e-02
4.54349577e-01 5.45747519e-01 -4.11278933e-01 -7.10312009e-01
-4.17556763e-01 -9.67911538e-03 -1.43420175e-01 7.22431242e-01
-1.01256585e+00 1.03674984e+00 5.84813023e+00 3.20905387e-01
-8.84446442e-01 1.20867841e-01 8.99168611e-01 -5.82246840e-01
-5.52183509e-01 3.70751992e-02 -1.22378364e-01 8.15753639e-02
1.11003399e+00 -3.10642630e-01 2.02111125e-01 4.94034171e-01
7.12783694e-01 -3.64081085e-01 -7.97151506e-01 6.31538570e-01
3.22562367e-01 -1.11423588e+00 -2.70216018e-01 -4.13800985e-01
6.62491500e-01 2.29681432e-01 -3.53342861e-01 4.45874333e-01
5.33647120e-01 -7.78808057e-01 4.57431197e-01 6.67357147e-01
6.93868518e-01 -7.64685810e-01 7.95916080e-01 4.07519668e-01
-8.54082227e-01 1.19619288e-01 -3.79694663e-02 -5.97114623e-01
5.78088820e-01 1.09452307e+00 -8.45674098e-01 4.53794360e-01
7.12584794e-01 1.01626766e+00 -2.99769312e-01 3.39075655e-01
-8.29767048e-01 1.05779338e+00 -3.70369494e-01 -6.84670210e-02
2.52181023e-01 2.45810419e-01 7.54338562e-01 1.64191008e+00
4.09999825e-02 9.88306284e-01 4.76983666e-01 3.82267296e-01
-2.78787047e-01 3.22531879e-01 -7.65394807e-01 -4.58507061e-01
5.83898723e-01 1.51986647e+00 -7.42299974e-01 -8.42773616e-01
5.95825911e-03 8.01169932e-01 5.45008242e-01 6.21259809e-01
-4.21353251e-01 -4.00221437e-01 4.95702773e-01 -2.62687653e-01
-2.17611849e-01 4.99286614e-02 -3.83684129e-01 -1.46368706e+00
-1.11970358e-01 -1.08368564e+00 7.55983114e-01 -1.00287974e+00
-1.34290290e+00 6.29548728e-01 1.03639804e-01 -6.30119562e-01
-6.09165132e-01 2.30125133e-02 -1.03502166e+00 3.67008507e-01
-1.37151122e+00 -7.59066880e-01 -2.83271670e-01 4.30354804e-01
6.73682392e-01 3.70900929e-01 9.67900217e-01 -1.64038777e-01
-7.43542731e-01 -1.14522062e-01 -5.11842787e-01 1.84206456e-01
1.13315284e+00 -1.24489462e+00 2.09287390e-01 8.79351020e-01
-1.35403067e-01 3.92852426e-01 1.05757856e+00 -1.02952170e+00
-9.77473199e-01 -1.00965834e+00 1.58421540e+00 -7.37735152e-01
8.84696722e-01 -3.19582313e-01 -1.00042903e+00 6.73185110e-01
7.36658454e-01 -5.95604181e-01 1.01649427e+00 5.19036233e-01
-3.50829035e-01 3.17868561e-01 -8.08015227e-01 6.44008994e-01
7.03673124e-01 -6.70676947e-01 -8.47692907e-01 7.09260225e-01
9.04828906e-01 1.59379020e-01 -8.10689270e-01 3.38300407e-01
1.23871788e-01 -6.82621360e-01 4.64353263e-01 -9.02326465e-01
7.82382309e-01 1.14605181e-01 4.33359854e-02 -1.57916176e+00
7.12251440e-02 -9.03666198e-01 -5.48945144e-02 1.42543936e+00
4.98916179e-01 -4.83782172e-01 1.50973275e-01 9.93718922e-01
1.66721328e-03 -6.22159481e-01 -3.96684676e-01 -1.52622059e-01
-2.57066488e-01 -6.92873776e-01 4.78384733e-01 1.44463551e+00
8.82692575e-01 5.13218820e-01 -6.32095158e-01 1.71230584e-01
3.73568743e-01 7.14222565e-02 7.49092102e-01 -1.18158436e+00
1.96188733e-01 -3.34798783e-01 3.05028796e-01 -3.35936725e-01
2.71920234e-01 -5.39881527e-01 1.94274083e-01 -1.95559371e+00
4.07326669e-01 1.67758271e-01 -4.74863231e-01 1.03758562e+00
-4.95616615e-01 3.96885984e-02 1.11877866e-01 2.15244666e-01
-1.04106092e+00 5.94738603e-01 2.78652400e-01 5.29861078e-02
-3.84629577e-01 -4.84797567e-01 -9.16819930e-01 9.83528256e-01
1.22782350e+00 -6.71097219e-01 -5.34661934e-02 -4.26249981e-01
9.29109275e-01 2.29234189e-01 2.47330248e-01 -5.55252969e-01
3.31044853e-01 -5.64603150e-01 1.16948761e-01 -8.62137973e-01
5.06579168e-02 -4.27066162e-02 -2.03002661e-01 1.09754167e-01
-8.24451208e-01 3.39126915e-01 2.57655740e-01 4.47357506e-01
-4.19183761e-01 1.69255674e-01 -1.21077579e-02 -1.37076125e-01
-3.51868391e-01 -9.04902294e-02 -9.84080553e-01 5.26933014e-01
5.23078024e-01 4.71485138e-01 -5.89332163e-01 -1.13819134e+00
-5.01602411e-01 6.77193880e-01 3.15852553e-01 3.09062153e-01
7.10015595e-01 -9.72056627e-01 -1.27440178e+00 -5.01121998e-01
1.43426031e-01 1.37861976e-02 2.50462055e-01 9.72774565e-01
1.39806289e-02 2.23648965e-01 1.63149923e-01 5.91691099e-02
-8.86363447e-01 2.44695038e-01 -3.13601047e-01 -5.87871373e-01
-4.60139155e-01 7.55032599e-01 4.66719568e-02 -3.24585497e-01
2.14841336e-01 -2.67902970e-01 -6.49704039e-01 8.19537938e-01
8.02693367e-01 3.01974475e-01 -3.07188570e-01 -4.09507871e-01
-4.97819722e-01 1.66863173e-01 1.33691519e-01 -6.47286236e-01
1.78361142e+00 -2.91214913e-01 -4.55009401e-01 6.30192399e-01
8.49388838e-01 5.04905619e-02 -7.87040651e-01 -3.47434461e-01
5.52716196e-01 1.72452629e-01 4.90656495e-02 -1.08367968e+00
-5.18039942e-01 7.95911908e-01 -4.63689893e-01 3.90257865e-01
1.42707527e+00 1.34970501e-01 9.42309201e-01 8.02494407e-01
-1.68395445e-01 -1.33740866e+00 4.18967277e-01 7.41670787e-01
1.40551388e+00 -1.27570641e+00 4.86374609e-02 1.56181991e-01
-1.22847283e+00 9.25435364e-01 5.09842336e-01 2.20155135e-01
1.20011427e-01 2.28286982e-01 4.90075737e-01 -6.95446074e-01
-1.56661534e+00 -2.04883114e-01 4.80677001e-02 8.91805813e-02
4.41932410e-01 2.77851343e-01 -3.06171507e-01 1.00344586e+00
-1.35315694e-02 -2.55295694e-01 8.50614071e-01 9.72859144e-01
-5.83063900e-01 -5.66508293e-01 -4.49412495e-01 4.92551506e-01
-7.32744873e-01 -3.00759435e-01 -1.16444266e+00 2.21116513e-01
-5.32799900e-01 1.40625417e+00 -2.00102359e-01 -2.88005859e-01
2.88016409e-01 4.05650198e-01 -5.18143117e-01 -5.94574511e-01
-6.91020250e-01 1.86179474e-01 7.34754980e-01 -7.48438358e-01
-5.76159358e-01 -8.46078277e-01 -1.40788054e+00 1.17620630e-02
2.05159739e-01 5.45690656e-01 4.29500550e-01 1.00758481e+00
5.99530578e-01 3.04348171e-01 8.08997452e-01 -8.13826859e-01
-1.37653217e-01 -1.15854359e+00 1.84533522e-02 4.87540603e-01
3.52454364e-01 -1.95826977e-01 -5.79390287e-01 1.39076591e-01]
|
[12.760396003723145, 6.376925945281982]
|
800b9cb7-c793-479c-aebc-c5ccf3215fd9
|
a-semi-paired-approach-for-label-to-image
|
2306.13585
| null |
https://arxiv.org/abs/2306.13585v2
|
https://arxiv.org/pdf/2306.13585v2.pdf
|
A Semi-Paired Approach For Label-to-Image Translation
|
Data efficiency, or the ability to generalize from a few labeled data, remains a major challenge in deep learning. Semi-supervised learning has thrived in traditional recognition tasks alleviating the need for large amounts of labeled data, yet it remains understudied in image-to-image translation (I2I) tasks. In this work, we introduce the first semi-supervised (semi-paired) framework for label-to-image translation, a challenging subtask of I2I which generates photorealistic images from semantic label maps. In the semi-paired setting, the model has access to a small set of paired data and a larger set of unpaired images and labels. Instead of using geometrical transformations as a pretext task like previous works, we leverage an input reconstruction task by exploiting the conditional discriminator on the paired data as a reverse generator. We propose a training algorithm for this shared network, and we present a rare classes sampling algorithm to focus on under-represented classes. Experiments on 3 standard benchmarks show that the proposed model outperforms state-of-the-art unsupervised and semi-supervised approaches, as well as some fully supervised approaches while using a much smaller number of paired samples.
|
['Bin Yang', 'Diandian Guo', 'Mark Youssef', 'Mohamed Abdelsamad', 'Shuai Zhang', 'George Eskandar']
|
2023-06-23
| null | null | null | null |
['image-to-image-translation', 'image-to-image-translation']
|
['computer-vision', 'miscellaneous']
|
[ 9.47684467e-01 3.75740647e-01 -2.07013264e-01 -6.01808965e-01
-1.13341951e+00 -7.82877505e-01 1.00226724e+00 -5.04685283e-01
-2.83118367e-01 8.13180447e-01 -1.30328834e-02 -1.87487096e-01
2.75859207e-01 -5.76740444e-01 -1.03574252e+00 -9.10237908e-01
6.10185683e-01 8.21150362e-01 -6.99763820e-02 7.11128935e-02
7.90827498e-02 2.89567411e-01 -1.48776031e+00 4.35889095e-01
7.29643285e-01 1.07294774e+00 1.82135075e-01 2.29776114e-01
-1.47633534e-02 9.88678753e-01 -3.69182795e-01 -3.75407964e-01
5.15445590e-01 -9.01370108e-01 -9.16650116e-01 2.68893808e-01
1.01801074e+00 -2.29854256e-01 -2.01279551e-01 9.51743364e-01
6.53574705e-01 3.89221869e-02 1.02642179e+00 -1.40470338e+00
-9.17269111e-01 5.11347890e-01 -4.97218490e-01 -4.39528912e-01
8.92231092e-02 1.24187700e-01 8.20174277e-01 -1.28960073e+00
9.21134353e-01 1.08705842e+00 4.47722405e-01 6.38166368e-01
-1.54563034e+00 -7.91266978e-01 -3.07746470e-01 -2.69098487e-02
-1.40941501e+00 -5.77480257e-01 1.05060470e+00 -5.55023670e-01
3.80391657e-01 6.46614209e-02 3.18228185e-01 1.25608206e+00
-4.04128462e-01 9.18621123e-01 1.68693924e+00 -6.66289032e-01
2.79536963e-01 4.35706496e-01 -4.11455989e-01 6.08644962e-01
-1.74173817e-01 4.65390116e-01 -5.90274274e-01 9.28636119e-02
7.55341291e-01 -1.15393072e-01 -2.70228773e-01 -7.24656880e-01
-1.53867781e+00 9.16612208e-01 6.68009818e-01 -6.78182766e-02
3.79582234e-02 -3.60542573e-02 2.25518197e-01 3.78608346e-01
6.30342782e-01 4.90668535e-01 -2.82053202e-01 2.94575095e-01
-1.13207042e+00 6.98437691e-02 6.68999672e-01 1.35705435e+00
1.17922819e+00 2.09258310e-03 -4.11870509e-01 8.45664740e-01
-2.67824799e-01 7.02174246e-01 4.02964890e-01 -8.29982042e-01
5.63851297e-01 5.18430889e-01 6.45466447e-02 -6.74894750e-01
-1.80875659e-02 -3.98095399e-01 -1.18092370e+00 2.21774608e-01
4.98356044e-01 2.59071589e-01 -1.04968107e+00 1.86088133e+00
3.82746160e-01 3.41489874e-02 -1.44218765e-02 7.44547307e-01
7.65654743e-01 5.05037665e-01 -2.02031091e-01 -1.18990876e-01
7.51234710e-01 -1.48230398e+00 -2.30456531e-01 -2.19806120e-01
7.31582046e-01 -9.09189284e-01 1.21465278e+00 2.71434963e-01
-8.56333613e-01 -7.93354213e-01 -6.81773961e-01 -3.74841422e-01
-3.99425834e-01 5.83046556e-01 4.35587853e-01 3.69249403e-01
-1.12923419e+00 6.58236802e-01 -2.66703993e-01 -4.05331492e-01
7.99113393e-01 2.68832058e-01 -6.05956912e-01 -4.39610511e-01
-8.12397420e-01 7.50492752e-01 3.24486613e-01 -7.08233863e-02
-1.13756680e+00 -8.05239737e-01 -8.97525012e-01 -3.83743107e-01
3.33967626e-01 -5.55377364e-01 1.02863705e+00 -1.32315278e+00
-1.63254380e+00 1.47373617e+00 4.57334965e-02 -1.54312760e-01
7.39335835e-01 1.87873691e-01 1.13200927e-02 2.58673370e-01
2.90368557e-01 1.31236541e+00 1.08618450e+00 -1.57729685e+00
-3.10198188e-01 -2.44691685e-01 -1.28431365e-01 1.24935679e-01
-1.70746103e-01 -1.71135738e-01 -6.72327504e-02 -7.28046477e-01
6.61912262e-02 -1.48437905e+00 3.95611711e-02 3.20061713e-01
-5.91700315e-01 -6.01932369e-02 7.26602554e-01 -4.14462507e-01
3.90713900e-01 -2.09991241e+00 3.04057509e-01 -9.27467644e-02
2.40096860e-02 2.53176093e-01 -3.56821537e-01 3.19734544e-01
-3.79676402e-01 -2.03143388e-01 -5.36261141e-01 -6.17213905e-01
-4.84125763e-02 1.38315275e-01 -5.87816954e-01 5.96247196e-01
4.38768387e-01 1.14660907e+00 -1.10933220e+00 -6.31004870e-01
2.35278159e-01 2.67048448e-01 -4.37420040e-01 4.66700554e-01
-1.82434604e-01 1.05627096e+00 8.50762799e-03 5.70378482e-01
7.33064592e-01 -4.14978862e-01 -5.53472750e-02 -4.02133405e-01
4.81305346e-02 8.27015638e-02 -7.97903597e-01 1.98785245e+00
-7.14613259e-01 5.17357171e-01 -3.19599837e-01 -1.38334405e+00
1.09813702e+00 1.77058876e-01 2.54192829e-01 -6.39375627e-01
1.33969098e-01 4.50587362e-01 -3.68397713e-01 -3.93322855e-01
1.40894204e-01 -3.11575979e-01 -1.36540562e-01 8.52038741e-01
3.99142116e-01 -5.73505521e-01 4.64134142e-02 1.65893316e-01
6.82333946e-01 6.44379139e-01 2.54412174e-01 -2.13535324e-01
3.74597967e-01 1.01776391e-01 5.37752926e-01 6.49570823e-01
-1.25441402e-02 1.19792843e+00 1.14485122e-01 -5.08878112e-01
-1.36209762e+00 -1.01228917e+00 -1.63208265e-02 9.56296742e-01
1.73205093e-01 1.47604436e-01 -7.95413673e-01 -8.84519696e-01
-2.42618576e-01 7.09082067e-01 -7.86058128e-01 -2.37066686e-01
-4.73776668e-01 -5.53508639e-01 4.44990486e-01 5.41192770e-01
7.05857396e-01 -1.13069189e+00 -2.21355736e-01 -1.86146915e-01
-2.79484361e-01 -1.30356860e+00 -6.01602852e-01 2.97465712e-01
-6.61643863e-01 -7.52823949e-01 -9.20248806e-01 -1.27484691e+00
1.11942935e+00 4.35471922e-01 1.22745073e+00 -2.43546277e-01
-2.86907017e-01 8.88171569e-02 -3.31405252e-01 -1.53698489e-01
-6.25108123e-01 1.13157265e-01 -9.65138897e-02 3.43596876e-01
2.94837672e-02 -7.62927651e-01 -5.23285389e-01 5.64873457e-01
-9.76662040e-01 6.92482233e-01 8.52592409e-01 1.17711210e+00
8.31236601e-01 -4.87455964e-01 4.90826428e-01 -1.21605039e+00
-1.11068189e-01 -1.78557083e-01 -4.25389767e-01 3.46745163e-01
-4.57570702e-01 2.35398218e-01 9.25766289e-01 -6.83056831e-01
-1.12469685e+00 5.54646611e-01 3.48659098e-01 -5.74179649e-01
-3.12786639e-01 2.88386224e-03 -4.74505603e-01 -4.27114844e-01
7.35613585e-01 3.83796304e-01 1.33120120e-01 -3.35639209e-01
7.92885125e-01 7.31908143e-01 6.85606956e-01 -6.95148647e-01
1.05131721e+00 7.58463979e-01 9.17869285e-02 -5.41674018e-01
-1.21017897e+00 -3.23265284e-01 -1.01353514e+00 -1.64555401e-01
7.07569063e-01 -9.93472397e-01 -1.41824096e-01 5.79187930e-01
-1.10522306e+00 -6.46124482e-01 -7.08867371e-01 3.08600992e-01
-9.84919608e-01 2.63264567e-01 -3.19834858e-01 -2.60744423e-01
-1.69848934e-01 -1.15560687e+00 1.50028193e+00 -9.00077261e-03
2.00995747e-02 -7.33775973e-01 1.10163532e-01 6.13415360e-01
3.57949406e-01 3.27355295e-01 7.56027341e-01 -6.98907256e-01
-5.94833791e-01 -1.58747688e-01 -4.74086612e-01 6.74457073e-01
2.13672996e-01 -4.34595764e-01 -1.19108319e+00 -3.70262146e-01
-2.89082434e-02 -1.17040074e+00 8.43502343e-01 -1.48929015e-01
1.28876185e+00 -2.34889999e-01 -2.29417071e-01 8.35563064e-01
1.18681419e+00 -2.94779032e-01 5.70406079e-01 -1.17287219e-01
1.06559741e+00 8.01207602e-01 5.58311224e-01 -3.18459310e-02
3.27159762e-01 7.67483056e-01 2.69849092e-01 -4.06326175e-01
-6.73341513e-01 -6.42719567e-01 2.71750718e-01 8.75737369e-01
1.83894336e-01 -8.42164233e-02 -6.88384295e-01 4.83314991e-01
-1.59347391e+00 -7.77368307e-01 9.40306336e-02 2.32175398e+00
1.18827093e+00 -2.25970298e-01 -1.08518541e-01 3.99601599e-03
8.89372766e-01 2.75289509e-02 -7.19963491e-01 1.92193568e-01
-3.28021765e-01 3.39688152e-01 5.11800885e-01 3.67967337e-01
-1.24394619e+00 1.24190593e+00 5.89929676e+00 1.06390679e+00
-1.28512311e+00 2.72192240e-01 9.54501092e-01 1.26099512e-02
-3.07229608e-01 2.73643136e-01 -6.64927244e-01 3.24452162e-01
6.28522754e-01 2.12519631e-01 5.60280800e-01 8.87050748e-01
-5.41338623e-02 3.90064009e-02 -1.47919643e+00 1.17881429e+00
5.56633711e-01 -1.32120609e+00 3.61569554e-01 1.13438107e-01
1.32516658e+00 -8.46603811e-02 1.64703846e-01 1.63594618e-01
3.45356911e-01 -9.66035008e-01 1.00547504e+00 4.25388306e-01
1.51422048e+00 -3.18649977e-01 3.67432982e-01 3.45896035e-01
-7.35154092e-01 5.36595844e-02 -4.48216289e-01 6.78137988e-02
1.34665325e-01 7.36903429e-01 -6.20460749e-01 5.07083535e-01
3.27393562e-01 8.47949088e-01 -6.90641344e-01 6.27235591e-01
-5.73943794e-01 6.43120766e-01 -3.15483987e-01 3.25348884e-01
1.36849806e-01 -4.11860168e-01 3.35033648e-02 8.62435460e-01
3.10083538e-01 -1.76874340e-01 3.00406754e-01 1.05020380e+00
-4.28264856e-01 6.31787926e-02 -7.95483589e-01 -2.69634686e-02
3.79451513e-01 1.39723921e+00 -7.67369211e-01 -4.99311894e-01
-2.68440962e-01 1.33287477e+00 5.23420036e-01 3.15242916e-01
-6.80563152e-01 -1.92201808e-01 -2.49248490e-01 9.02901962e-02
2.15897709e-01 3.44172702e-03 -2.84691900e-01 -1.30967605e+00
2.17730319e-03 -9.50496793e-01 3.64423394e-02 -1.09606600e+00
-1.26885867e+00 5.88306308e-01 -6.56168815e-03 -1.65570080e+00
-3.51319581e-01 -6.03002608e-01 -5.04181981e-01 7.24616170e-01
-1.59294510e+00 -1.80903339e+00 -5.32588482e-01 6.00792170e-01
4.15647775e-01 -2.84183532e-01 8.92999530e-01 2.88565665e-01
-2.68237442e-01 7.68083334e-01 1.91199884e-01 2.27521047e-01
1.16111457e+00 -1.06501746e+00 2.62166411e-01 7.19090343e-01
5.22019088e-01 2.88015664e-01 3.43085766e-01 -2.78503478e-01
-1.14019942e+00 -1.40519607e+00 9.76519823e-01 -4.53515261e-01
3.23359698e-01 -7.02213287e-01 -5.32177269e-01 6.76127434e-01
1.49759606e-01 5.75804710e-01 5.99826694e-01 -5.56100547e-01
-6.09331548e-01 -1.66088760e-01 -1.10959470e+00 4.76835579e-01
1.36595404e+00 -7.72907972e-01 -2.61772722e-01 7.23665118e-01
7.33399689e-01 -2.56437898e-01 -5.03969908e-01 3.60652447e-01
3.16787541e-01 -8.11739743e-01 8.89663398e-01 -3.12988997e-01
8.12048912e-01 -3.16868305e-01 -1.73910841e-01 -1.49188113e+00
-5.26978821e-02 -7.12266862e-01 3.53036970e-01 1.40124643e+00
3.73558015e-01 -3.95525962e-01 8.66443574e-01 2.98424810e-01
-1.56269118e-01 -6.28427029e-01 -7.52306223e-01 -9.59916115e-01
2.68320054e-01 -1.55948207e-03 4.33948338e-01 1.03283906e+00
-4.86635745e-01 5.80667138e-01 -6.57242060e-01 -2.97861874e-01
9.34024990e-01 6.08645320e-01 1.06934261e+00 -9.91377056e-01
-3.36447656e-01 -1.07772566e-01 -2.70863205e-01 -1.12899590e+00
4.40155119e-01 -1.35905182e+00 2.33278394e-01 -1.13840079e+00
3.97754431e-01 -6.42585039e-01 1.14192985e-01 6.68925583e-01
6.85347393e-02 9.33627963e-01 -4.21964191e-02 5.13919175e-01
-4.96680498e-01 9.45348442e-01 1.40108848e+00 -4.14733261e-01
1.36136338e-01 -2.71473199e-01 -6.19467318e-01 5.24968147e-01
6.25706375e-01 -6.60892367e-01 -5.90365589e-01 -4.54587311e-01
1.60130739e-01 -2.42993429e-01 5.28617382e-01 -8.62301648e-01
1.60209745e-01 -1.88564926e-01 2.62633234e-01 -4.24974948e-01
1.19617239e-01 -5.42588115e-01 1.11665621e-01 -2.67932680e-03
-7.20850170e-01 -3.50284368e-01 -4.35223728e-01 4.77510810e-01
-2.92116612e-01 -2.08350539e-01 1.02270532e+00 -7.21843615e-02
-5.18355429e-01 4.40440923e-01 3.10833842e-01 3.74227196e-01
9.52245414e-01 -1.63837910e-01 -2.11396188e-01 -3.27521801e-01
-5.53892672e-01 -1.58943400e-01 7.90777743e-01 3.78191680e-01
4.50094968e-01 -1.60378098e+00 -6.71064138e-01 2.91635841e-01
6.40016377e-01 3.62472028e-01 9.47975218e-02 7.10993588e-01
-3.59859228e-01 1.63371861e-01 -3.19515973e-01 -8.13838363e-01
-8.54094148e-01 7.94155419e-01 6.09087981e-02 -1.79454595e-01
-5.57889700e-01 9.11124945e-01 5.67061722e-01 -1.01882613e+00
2.31494159e-02 2.66198404e-02 2.05616772e-01 2.27682441e-02
3.75404239e-01 4.10408191e-02 8.27555805e-02 -9.22535241e-01
-2.58731889e-03 7.10717142e-01 -4.66746502e-02 -1.72202021e-01
1.27198195e+00 -3.12078390e-02 -1.70905724e-01 4.89421159e-01
1.56858325e+00 -3.69191825e-01 -1.50369203e+00 -6.54397666e-01
-3.12668473e-01 -4.81772661e-01 -4.57413942e-02 -7.58026004e-01
-1.19250715e+00 1.11936617e+00 4.80858058e-01 -4.16127682e-01
9.94168460e-01 1.13541752e-01 7.18583822e-01 4.17253226e-01
5.95482647e-01 -1.01600277e+00 4.42396164e-01 2.36381248e-01
1.02908671e+00 -1.56437182e+00 -1.00111566e-01 -5.33556879e-01
-7.82014310e-01 8.54890645e-01 4.87297177e-01 -2.03250438e-01
4.58291262e-01 -1.88618572e-03 2.10694373e-01 9.10179839e-02
-3.49537641e-01 -6.33871257e-02 2.03940243e-01 7.40490794e-01
2.54902869e-01 6.84793368e-02 4.76942323e-02 1.74124628e-01
-3.01083863e-01 1.31059870e-01 3.53208959e-01 9.31338727e-01
2.47690231e-02 -1.24806225e+00 -1.54115707e-01 2.66598165e-01
1.83137089e-01 -1.85063347e-01 -5.61532736e-01 6.39091730e-01
1.70509234e-01 7.09900320e-01 2.22455314e-03 -4.76182044e-01
2.55550686e-02 5.50292395e-02 7.10787296e-01 -8.09751928e-01
-2.60263711e-01 -2.08892852e-01 -1.49879918e-01 -5.69617033e-01
-7.38067985e-01 -5.22092462e-01 -7.03268766e-01 4.52407598e-02
-3.30045998e-01 -1.87084720e-01 5.28353095e-01 9.45455670e-01
4.21514243e-01 2.55274092e-04 1.02407026e+00 -1.38343143e+00
-7.29056954e-01 -1.07876611e+00 -5.19759893e-01 9.28298771e-01
6.68444335e-02 -7.91648567e-01 -5.01458347e-01 4.30864960e-01]
|
[11.472912788391113, -0.03469494357705116]
|
e88fd109-7676-4e58-9d34-abd811c1fd0e
|
mesoscopic-structure-of-the-stock-market-and
|
2112.06544
| null |
https://arxiv.org/abs/2112.06544v1
|
https://arxiv.org/pdf/2112.06544v1.pdf
|
Mesoscopic Structure of the Stock Market and Portfolio Optimization
|
The idiosyncratic (microscopic) and systemic (macroscopic) components of market structure have been shown to be responsible for the departure of the optimal mean-variance allocation from the heuristic `equally-weighted' portfolio. In this paper, we exploit clustering techniques derived from Random Matrix Theory (RMT) to study a third, intermediate (mesoscopic) market structure that turns out to be the most stable over time and provides important practical insights from a portfolio management perspective. First, we illustrate the benefits, in terms of predicted and realized risk profiles, of constructing portfolios by filtering out both random and systemic co-movements from the correlation matrix. Second, we redefine the portfolio optimization problem in terms of stock clusters that emerge after filtering. Finally, we propose a new wealth allocation scheme that attaches equal importance to stocks belonging to the same community and show that it further increases the reliability of the constructed portfolios. Results are robust across different time spans, cross-sectional dimensions and set of constraints defining the optimization problem
|
['Diego Garlaschelli', 'Tiziano Squartini', 'Giorgio Fagiolo', 'Sebastiano Michele Zema']
|
2021-12-13
| null | null | null | null |
['portfolio-optimization']
|
['time-series']
|
[-2.38485768e-01 -2.08538794e-03 1.18660867e-01 4.80835050e-01
-1.44014180e-01 -9.13945317e-01 8.84691238e-01 1.20651171e-01
2.34517939e-02 7.93691635e-01 3.65862936e-01 -2.41653442e-01
-1.12926686e+00 -1.05842400e+00 -1.70652479e-01 -9.12614644e-01
-5.38362801e-01 4.45157290e-01 2.70364106e-01 -1.28060609e-01
3.37835580e-01 6.25781298e-01 -1.40010250e+00 8.12093075e-03
7.29613960e-01 1.16477013e+00 -1.27623215e-01 2.28742793e-01
2.41833963e-02 9.34977412e-01 -4.95165020e-01 -7.82080650e-01
8.85002255e-01 -4.78500068e-01 -3.72516692e-01 3.06936145e-01
-3.79746228e-01 2.40039349e-01 1.95810303e-01 1.02308416e+00
2.96294183e-01 9.58465189e-02 9.55717921e-01 -1.05112410e+00
-5.60451865e-01 7.95881748e-01 -5.72885036e-01 2.33725518e-01
-1.93415403e-01 3.45923491e-02 1.38362336e+00 -5.66509902e-01
7.86160231e-01 9.03781354e-01 7.54506052e-01 1.12487316e-01
-1.53661549e+00 -3.14269364e-01 1.08422257e-01 -2.47044027e-01
-1.17107618e+00 -1.38696939e-01 7.37458289e-01 -9.54742432e-01
7.79946327e-01 4.81387466e-01 7.23319650e-01 5.80390811e-01
4.52786505e-01 1.02363765e-01 1.36673021e+00 -2.01549023e-01
4.67658013e-01 1.74469411e-01 -5.51770665e-02 -3.94572914e-02
9.66190457e-01 3.57784778e-01 -3.72119814e-01 -3.58715683e-01
7.40505219e-01 -1.23970695e-01 -1.63949803e-01 -7.74261773e-01
-1.16143501e+00 8.06754410e-01 4.19057086e-02 6.57769680e-01
-6.91193104e-01 -4.89639770e-03 -4.15160395e-02 5.69489002e-01
5.49727380e-01 7.31039405e-01 -4.74048227e-01 1.59230128e-01
-8.52288544e-01 1.81402057e-01 9.86396611e-01 5.16199648e-01
5.88353693e-01 7.84532279e-02 -1.22861765e-01 2.69003302e-01
4.18545008e-01 7.26741493e-01 1.77376583e-01 -1.17799330e+00
4.58024204e-01 5.44173539e-01 3.72377008e-01 -1.27227569e+00
-3.96093905e-01 -8.68389368e-01 -9.83431637e-01 4.19635743e-01
6.16783023e-01 -5.81249774e-01 8.59751701e-02 1.60395896e+00
1.17617197e-01 7.91785419e-02 9.11941230e-02 4.83317673e-01
-1.97968781e-01 3.21099758e-01 -2.67736971e-01 -5.50976217e-01
9.68585193e-01 -6.61378443e-01 -1.88394383e-01 2.40675122e-01
1.06895417e-01 -6.04340434e-01 1.63711742e-01 4.31699157e-01
-1.21285117e+00 -9.24791172e-02 -6.77970290e-01 1.09503293e+00
-2.13888392e-01 -4.39505696e-01 5.51526189e-01 8.00831497e-01
-1.31219983e+00 1.24559772e+00 -5.61379373e-01 -1.05564989e-01
1.25339642e-01 3.75251472e-01 1.55481547e-01 7.65125930e-01
-9.81854618e-01 8.81640911e-01 -1.78903356e-01 -4.50657457e-02
-3.77344847e-01 -7.85804868e-01 1.52322343e-02 3.38008136e-01
4.41095918e-01 -7.51598299e-01 7.28494763e-01 -1.13506079e+00
-1.28875494e+00 2.07513437e-01 4.84268516e-01 -4.73612934e-01
7.42045760e-01 3.03055614e-01 -1.33248717e-01 3.09100300e-01
1.36890382e-01 -3.84108514e-01 6.73496783e-01 -1.23815191e+00
-3.52743834e-01 -1.62380785e-01 -1.37983754e-01 -8.57874155e-02
-3.14235628e-01 1.93141580e-01 4.73402947e-01 -9.94780898e-01
-9.22594145e-02 -8.41753304e-01 -5.11101127e-01 -7.19328642e-01
-4.38953519e-01 3.28083128e-01 -1.11989737e-01 -4.65607375e-01
1.30314851e+00 -1.72815824e+00 5.32145917e-01 9.53215480e-01
2.68811136e-01 -4.12489027e-01 6.92159832e-02 1.12755287e+00
-3.90217334e-01 4.48948711e-01 -1.94707692e-01 -3.03678751e-01
4.45001096e-01 -8.84174630e-02 -4.08451140e-01 4.90839869e-01
7.38726109e-02 7.25618660e-01 -6.31486297e-01 2.36163773e-02
7.70166516e-02 -7.22723976e-02 -4.16846931e-01 -1.61780462e-01
4.36264873e-02 5.73559105e-01 -5.57094216e-01 2.73199916e-01
6.10665083e-01 -2.62020022e-01 4.74918038e-01 2.74110913e-01
-3.48425716e-01 -2.81204027e-03 -1.50902760e+00 7.10863888e-01
-7.79730231e-02 2.29551822e-01 2.24943519e-01 -9.79655027e-01
7.16377914e-01 2.92122155e-01 1.01153338e+00 -2.58057624e-01
6.54219985e-02 2.69514501e-01 2.29220495e-01 -4.83125560e-02
1.66659459e-01 -5.67256808e-01 -2.72754505e-02 1.07184064e+00
-7.28253871e-02 3.11371446e-01 2.59304583e-01 2.95147877e-02
1.18891752e+00 -3.09482545e-01 3.05062711e-01 -1.06548536e+00
4.70755190e-01 -2.88791269e-01 6.18775666e-01 7.04621673e-01
3.81896272e-02 3.70865315e-01 1.03868508e+00 7.76266530e-02
-1.04864681e+00 -1.18257070e+00 8.67564511e-03 4.69575495e-01
-1.09883007e-02 -2.20425576e-02 -6.23459339e-01 -3.08907598e-01
5.50482929e-01 3.74425590e-01 -8.31331849e-01 8.32793564e-02
-2.45704383e-01 -1.44517255e+00 -1.16849728e-01 1.25959441e-01
1.02547847e-01 -8.93972695e-01 -6.86827600e-01 3.61373454e-01
3.30801755e-01 -5.00044823e-01 -2.70102471e-01 6.62371591e-02
-8.86320174e-01 -1.22266078e+00 -7.47528136e-01 1.47875443e-01
4.96454120e-01 -2.77211443e-02 1.22901237e+00 5.56080267e-02
4.83161956e-02 7.55468786e-01 -2.72835910e-01 -1.62376478e-01
-2.75350451e-01 -3.07773822e-03 3.18250120e-01 6.40114307e-01
-5.46325976e-03 -8.86698723e-01 -6.77980542e-01 3.64276648e-01
-7.50918031e-01 -4.34052259e-01 3.94492954e-01 4.81949389e-01
4.29228581e-02 6.88331127e-01 9.21530724e-01 -5.39219439e-01
8.69177163e-01 -7.26608038e-01 -9.11902308e-01 3.25918049e-01
-9.72863495e-01 1.13603756e-01 3.07630569e-01 -3.18051547e-01
-1.16855526e+00 -5.28898835e-01 5.23948312e-01 1.20770581e-01
3.15655917e-01 6.29301488e-01 -2.07824130e-02 -1.33506134e-01
1.14594370e-01 -2.85193957e-02 2.44475856e-01 -6.87715948e-01
2.06926867e-01 2.51723111e-01 -1.00572839e-01 -7.02724159e-01
1.16068053e+00 6.85804844e-01 3.64003032e-01 -5.26459813e-01
-4.01610106e-01 -1.57635018e-01 -7.13130832e-01 -4.02776927e-01
7.59850264e-01 -6.88355923e-01 -8.11519086e-01 5.06352842e-01
-7.17961550e-01 -3.76932174e-01 -7.18912303e-01 5.15271842e-01
-6.19608104e-01 2.54846931e-01 -7.34903991e-01 -1.44197595e+00
-6.23807050e-02 -6.98021412e-01 4.58091497e-01 -2.36800276e-02
-2.74346739e-01 -1.45331979e+00 5.96996546e-01 -1.16772726e-01
7.75855124e-01 6.08793199e-01 8.35183084e-01 -7.00124741e-01
-9.15787101e-01 2.00541705e-01 1.67358890e-01 1.57965913e-01
2.29438901e-01 1.92001909e-01 -5.60391605e-01 -2.21620291e-01
4.57243592e-01 7.12383151e-01 8.15494537e-01 4.77364659e-01
-2.01421287e-02 -5.59827864e-01 -9.38626304e-02 2.51099437e-01
1.70183384e+00 2.13620067e-01 3.77479315e-01 6.42607749e-01
1.66991010e-01 1.46046090e+00 1.21841945e-01 7.67481804e-01
1.23873182e-01 5.22314370e-01 1.93079561e-01 1.91095650e-01
4.04071003e-01 2.20029458e-01 4.70680952e-01 1.05259407e+00
-4.24307436e-01 8.30303207e-02 -7.50673652e-01 5.08378685e-01
-2.05320477e+00 -1.57977355e+00 -8.70776847e-02 2.39322472e+00
4.55370307e-01 1.69950724e-01 7.06822574e-01 -1.64851785e-01
6.94719136e-01 -2.16630157e-02 -1.91536069e-01 -6.70301961e-03
-4.67166841e-01 9.29389298e-02 8.30417156e-01 4.45540398e-01
-7.82717705e-01 2.52355278e-01 6.57709551e+00 6.39444590e-01
-6.73834801e-01 6.84966668e-02 7.09363580e-01 -3.68935794e-01
-6.04722440e-01 5.13955094e-02 -4.89150733e-01 6.72423244e-01
1.17158854e+00 -7.08111346e-01 5.69109619e-01 3.47901225e-01
3.79699171e-01 -1.17782183e-01 -7.26685643e-01 2.68172264e-01
-3.24015379e-01 -1.32945013e+00 -2.48461559e-01 6.72104716e-01
1.03381622e+00 -7.68170208e-02 1.97962821e-01 -2.79240519e-01
7.68310547e-01 -7.59681821e-01 1.01077199e+00 1.03276718e+00
6.22109696e-02 -9.60944057e-01 7.15801537e-01 2.56741941e-01
-1.36699200e+00 -4.09526795e-01 -2.46522039e-01 -2.21695498e-01
2.30416536e-01 9.34825480e-01 8.13294575e-02 7.50613689e-01
5.62935531e-01 5.88019729e-01 -6.09013081e-01 9.27123547e-01
2.90781051e-01 1.95076734e-01 -1.56893864e-01 4.23855931e-02
3.08270659e-02 -9.50538576e-01 6.46450341e-01 5.90224445e-01
6.95464909e-01 -6.38503954e-02 -4.09809768e-01 1.10012412e+00
3.52363855e-01 4.79329042e-02 -7.43377507e-01 -1.72967970e-01
2.50934005e-01 1.08911347e+00 -1.21111023e+00 -3.84134836e-02
-5.43432534e-01 4.36048895e-01 -1.16799720e-01 4.11900222e-01
-3.71819884e-01 -1.06902733e-01 8.88718784e-01 3.23568285e-01
7.10600913e-01 -1.91844881e-01 -4.20661420e-01 -1.38800776e+00
1.87283084e-02 -5.55405259e-01 1.22458346e-01 -1.62995338e-01
-1.85182559e+00 3.45956832e-01 1.23628683e-01 -1.30124056e+00
-3.06513250e-01 -5.66236794e-01 -9.08402085e-01 8.48245442e-01
-1.25078845e+00 -4.28435326e-01 4.86054003e-01 2.26050571e-01
-2.04226866e-01 -3.96469414e-01 3.78015816e-01 4.72166725e-02
-5.92067122e-01 -7.84646422e-02 9.15341794e-01 -1.55040488e-01
1.23963900e-01 -1.69376266e+00 3.90255779e-01 8.25416148e-01
-1.18791036e-01 8.40310037e-01 6.52165294e-01 -9.94557500e-01
-1.17407501e+00 -1.01441085e+00 7.07143307e-01 -8.55731249e-01
1.58199811e+00 -3.10423732e-01 -5.60685694e-01 4.34724420e-01
5.22901535e-01 -6.56487942e-01 8.38020504e-01 -1.23414584e-01
4.47489768e-02 -2.41741940e-01 -1.17268336e+00 6.12251401e-01
7.38388300e-01 -3.89297247e-01 -4.51825798e-01 1.18940741e-01
3.66739988e-01 7.15967298e-01 -1.40131056e+00 4.51400056e-02
5.11139929e-01 -1.53740954e+00 1.14521122e+00 -2.52240062e-01
1.04886003e-01 -5.40609136e-02 -1.54212654e-01 -1.23069310e+00
-5.14794290e-01 -1.08079600e+00 9.28252265e-02 1.46276426e+00
5.99555373e-01 -1.24018276e+00 8.03770900e-01 6.92267716e-01
6.15945935e-01 -6.14504933e-01 -1.18220878e+00 -1.17418838e+00
6.56600475e-01 2.08741389e-02 8.68878901e-01 8.66310716e-01
2.32171208e-01 5.81534654e-02 -3.13108861e-01 -8.02917406e-02
1.07564008e+00 5.64721048e-01 3.66736710e-01 -1.70550036e+00
-7.78156698e-01 -9.71131146e-01 -1.24140032e-01 -3.23631316e-01
-1.64886609e-01 -6.27317607e-01 -3.58583182e-01 -1.04000413e+00
1.31775260e-01 -4.31250572e-01 -4.26697910e-01 -4.32961762e-01
2.51616091e-01 -8.10657814e-03 7.23628938e-01 5.89423537e-01
-4.90536958e-01 3.59636992e-01 7.67407954e-01 1.85565323e-01
-1.57256499e-01 2.12906435e-01 -1.04881823e+00 5.87961435e-01
7.79519320e-01 -3.46546829e-01 -2.39218414e-01 1.85709015e-01
7.42511213e-01 2.97624320e-01 2.32026160e-01 -7.28953421e-01
1.21068405e-02 -5.27264416e-01 4.60929386e-02 -2.95910537e-01
-8.57579634e-02 -9.48713243e-01 9.71814692e-01 5.71083665e-01
-1.05860449e-01 3.77877831e-01 -1.05598718e-01 6.72817230e-01
1.32794119e-02 -5.41498542e-01 5.98066807e-01 -3.73145729e-01
2.43887484e-01 -7.58080464e-03 -6.47281349e-01 -6.17453903e-02
1.28996897e+00 -3.48508656e-02 -2.69027859e-01 -3.47417265e-01
-9.27200615e-01 1.33252203e-01 7.81545937e-01 -1.12184696e-01
-8.42829868e-02 -1.36353803e+00 -6.38720095e-01 -2.76797473e-01
-4.21045631e-01 -7.35567868e-01 1.16387114e-01 1.22556245e+00
-3.70140314e-01 5.58489382e-01 -6.82686642e-02 1.46625880e-02
-7.31700182e-01 6.97278917e-01 5.23085773e-01 -7.11448908e-01
-3.67695093e-01 4.07394558e-01 2.79535413e-01 -8.37580562e-02
-1.90239325e-01 -2.26031274e-01 -2.77782559e-01 7.69124329e-01
3.33655030e-01 7.82664120e-01 -1.79354936e-01 -7.50410676e-01
-2.77919471e-01 9.48671937e-01 6.09870315e-01 -3.91327322e-01
1.67744899e+00 -5.52650511e-01 -5.94393909e-01 5.44317842e-01
7.76324213e-01 5.21004736e-01 -1.31040490e+00 6.19907193e-02
7.65828907e-01 -2.51949698e-01 -5.21923542e-01 -3.92762184e-01
-1.41210842e+00 5.13262689e-01 1.51517853e-01 1.10892189e+00
8.72746885e-01 -1.93778276e-01 2.72036642e-01 1.14624374e-01
5.33801913e-01 -1.05827785e+00 1.65617391e-02 6.76740557e-02
7.97160804e-01 -4.89019871e-01 1.18827440e-01 -2.91780829e-01
-5.82013011e-01 8.70652735e-01 -1.71251133e-01 -5.97535849e-01
1.17605567e+00 1.62780628e-01 -2.18984798e-01 -3.53071749e-01
-8.86011064e-01 -3.35347027e-01 1.41694486e-01 5.36178887e-01
-1.51042178e-01 2.19960451e-01 -2.76871592e-01 8.38301599e-01
-1.56670332e-01 -6.09487593e-01 8.39034200e-01 6.94259107e-01
-3.53180200e-01 -1.18074524e+00 -4.61425602e-01 6.06012404e-01
-7.19971478e-01 -1.54137462e-01 -7.46252894e-01 7.87192702e-01
7.05091804e-02 8.92740607e-01 2.29398757e-01 -2.33324006e-01
2.20137224e-01 -1.06253223e-02 1.17569558e-01 -6.72833741e-01
-1.01648295e+00 4.97374803e-01 9.40520409e-03 -2.86655933e-01
-6.24518335e-01 -1.19221807e+00 -5.55984020e-01 -4.86607671e-01
-4.71232295e-01 5.24181902e-01 4.41425443e-02 7.90704668e-01
4.45309550e-01 4.30080295e-01 9.57706153e-01 -8.57302547e-01
-8.57370734e-01 -6.43584907e-01 -1.41167545e+00 2.44416952e-01
1.72849566e-01 -8.45784903e-01 -1.00421524e+00 -1.74191102e-01]
|
[5.012136459350586, 4.052708625793457]
|
288f3e3d-88b4-429b-8a3a-8184e55b3070
|
causal-fault-localisation-in-dataflow-systems
|
2304.11987
| null |
https://arxiv.org/abs/2304.11987v1
|
https://arxiv.org/pdf/2304.11987v1.pdf
|
Causal fault localisation in dataflow systems
|
Dataflow computing was shown to bring significant benefits to multiple niches of systems engineering and has the potential to become a general-purpose paradigm of choice for data-driven application development. One of the characteristic features of dataflow computing is the natural access to the dataflow graph of the entire system. Recently it has been observed that these dataflow graphs can be treated as complete graphical causal models, opening opportunities to apply causal inference techniques to dataflow systems. In this demonstration paper we aim to provide the first practical validation of this idea with a particular focus on causal fault localisation. We provide multiple demonstrations of how causal inference can be used to detect software bugs and data shifts in multiple scenarios with three modern dataflow engines.
|
['Neil D. Lawrence', 'Andrei Paleyes']
|
2023-04-24
| null | null | null | null |
['causal-inference', 'causal-inference']
|
['knowledge-base', 'miscellaneous']
|
[ 8.51177499e-02 1.17646813e-01 -4.27921236e-01 -3.20382625e-01
2.28967547e-01 -3.94806027e-01 6.97290897e-01 2.62733936e-01
5.38788795e-01 3.57023001e-01 1.18885845e-01 -8.98468018e-01
-5.66539109e-01 -8.28650296e-01 -5.65064192e-01 -1.31686509e-01
-7.23054349e-01 4.19200882e-02 3.49569023e-01 -2.16380864e-01
5.78904688e-01 6.57831848e-01 -2.02717400e+00 3.16847563e-01
6.21606231e-01 3.80131751e-01 -7.73411170e-02 9.42052186e-01
-1.31488636e-01 8.92109513e-01 -8.59688580e-01 -1.51499603e-02
-1.68563902e-01 -3.27297539e-01 -7.02088177e-01 -3.26031327e-01
3.55556786e-01 -2.14829862e-01 -9.48901847e-02 7.80279160e-01
8.63355547e-02 -7.50648797e-01 -1.75871536e-01 -2.04572821e+00
-4.03361954e-02 8.58030558e-01 -7.58034468e-01 4.16845590e-01
5.58793366e-01 3.86861712e-01 8.98423791e-01 -4.26265359e-01
7.28633523e-01 1.50853693e+00 2.70525128e-01 3.74368392e-02
-1.62552810e+00 -7.83625960e-01 3.98564339e-02 1.92778304e-01
-8.76224518e-01 -3.42878044e-01 3.94909084e-01 -7.04354763e-01
1.14516473e+00 4.70993519e-01 6.58934772e-01 6.96444273e-01
7.47101903e-01 2.26077080e-01 1.20484769e+00 -7.22014308e-01
2.81829476e-01 -2.71309853e-01 6.52922034e-01 6.17772698e-01
6.75798416e-01 5.54825962e-01 -8.86985004e-01 -5.95750690e-01
7.36189663e-01 7.69996345e-02 -1.86184589e-02 -3.57775420e-01
-1.13140440e+00 4.58553493e-01 3.48772258e-01 5.83342791e-01
4.74551283e-02 9.74546790e-01 4.24448013e-01 5.42308927e-01
-7.89119825e-02 3.94153893e-01 -4.58561242e-01 -4.88095254e-01
-4.91939247e-01 5.48858166e-01 9.84051943e-01 9.47281599e-01
5.20188451e-01 9.05851498e-02 7.15484694e-02 -1.68352321e-01
8.80203366e-01 3.14008594e-01 -1.09906688e-01 -7.92591095e-01
1.05495786e-03 1.04051256e+00 -2.20697314e-01 -7.94222414e-01
-2.79172271e-01 -1.45694643e-01 -1.05358459e-01 8.24603379e-01
3.93728167e-01 1.44160716e-02 -4.40657407e-01 1.25552630e+00
2.53613710e-01 2.50781942e-02 -3.41833830e-01 3.13624769e-01
-1.04097478e-01 2.15432599e-01 -8.28663632e-02 -2.62837648e-01
1.08768642e+00 1.91100672e-01 -8.76303077e-01 2.72125453e-01
6.42702758e-01 -8.18616450e-01 7.98828840e-01 7.71802783e-01
-6.18346512e-01 -2.52226204e-01 -1.37225115e+00 5.96180499e-01
-2.08956432e-02 -5.62719405e-01 1.02432358e+00 1.03664386e+00
-9.18197632e-01 5.42362094e-01 -1.24497056e+00 -4.66984391e-01
3.79104942e-01 2.67663717e-01 9.01206359e-02 -9.99719575e-02
-7.04261661e-01 4.62301433e-01 1.54927611e-01 -6.64982796e-02
-1.25693905e+00 -1.15660858e+00 -2.48256654e-01 5.69489785e-02
4.42393273e-01 -6.07466280e-01 1.12906182e+00 -3.14914942e-01
-7.79632747e-01 3.78424525e-01 -1.90422609e-01 -2.33046904e-01
2.46618599e-01 -2.71706194e-01 -9.36599016e-01 -5.49507737e-01
5.64697869e-02 -2.56923586e-01 7.34759748e-01 -1.01976967e+00
-7.99751103e-01 -5.50920606e-01 1.65310606e-01 -9.65558290e-01
-2.34852448e-01 4.80554610e-01 1.84832945e-01 -9.21645984e-02
-3.74442697e-01 -6.44115865e-01 -3.24122369e-01 -3.42050642e-01
-5.90375781e-01 -1.87140465e-01 1.22564697e+00 -1.67491779e-01
1.66091752e+00 -1.97753954e+00 -2.28841111e-01 3.50752592e-01
4.72691923e-01 -9.89650190e-02 3.81400615e-01 9.50337470e-01
-4.40888822e-01 5.50674319e-01 -6.65261075e-02 6.58308387e-01
3.74770835e-02 1.43879116e-01 -3.60674471e-01 5.42335331e-01
6.08954132e-01 6.37813568e-01 -9.81846988e-01 -2.15293646e-01
3.60611320e-01 1.30559966e-01 -5.93188763e-01 1.55877650e-01
-4.82545465e-01 1.47092566e-01 -3.11501563e-01 5.11716008e-01
5.48754513e-01 -1.30678579e-01 6.52216852e-01 2.38890365e-01
-7.43768096e-01 4.83712465e-01 -1.37077129e+00 1.32130027e+00
-3.57219249e-01 7.59407282e-01 -2.42489517e-01 -4.60511088e-01
7.69117475e-01 3.85135502e-01 4.55672055e-01 -6.98669732e-01
-2.66237020e-01 -8.20866507e-03 4.96512592e-01 -5.51633000e-01
1.51977912e-01 3.97429317e-02 -3.16608876e-01 7.56945610e-01
-2.42782250e-01 4.39952612e-02 4.75171745e-01 3.06514978e-01
1.99058735e+00 3.16558301e-01 3.31545889e-01 -7.02096224e-01
1.57555461e-01 4.23164159e-01 3.93246710e-01 4.06238973e-01
2.18707353e-01 -7.62939975e-02 1.28272605e+00 -3.77992004e-01
-9.05291975e-01 -1.16403294e+00 -3.69991660e-01 6.17211282e-01
-8.24832246e-02 -1.20581508e+00 -5.36345899e-01 -7.24463046e-01
2.72128552e-01 9.91402805e-01 -5.91911614e-01 -8.59316438e-02
-6.03665054e-01 -8.23544681e-01 5.21112978e-01 4.75131333e-01
-1.79648459e-01 -7.86719918e-01 -1.18542778e+00 2.21530467e-01
5.73094130e-01 -4.74555403e-01 3.48292083e-01 3.24961156e-01
-1.20281780e+00 -1.59697413e+00 4.17906851e-01 3.43760382e-03
4.62321669e-01 1.30174654e-02 1.48226476e+00 3.70867163e-01
-1.04654670e+00 1.99191540e-01 -1.05880097e-01 -4.98835057e-01
-9.42356825e-01 -4.79350388e-01 -1.47427514e-01 -3.95041257e-01
2.51540303e-01 -7.72645831e-01 -1.66632473e-01 6.26088321e-01
-9.88426864e-01 -3.78646880e-01 1.66646913e-01 5.03041863e-01
3.23978066e-02 4.30803537e-01 3.91206443e-01 -1.19444633e+00
8.27229738e-01 -7.19996035e-01 -1.17973757e+00 1.16955131e-01
-1.18738449e+00 3.88171524e-01 3.97552490e-01 8.12943876e-02
-1.04827929e+00 -6.37720823e-02 2.78155833e-01 -1.55006126e-01
-3.00446570e-01 8.28720748e-01 -3.07376355e-01 4.01243001e-01
8.56338561e-01 -6.88176453e-01 -5.81306405e-03 -3.64270151e-01
5.13156295e-01 2.56216049e-01 1.97378442e-01 -6.77437425e-01
6.76732361e-01 5.16478419e-01 4.31249171e-01 -6.99799657e-01
3.02055418e-01 -2.95009494e-01 -5.92317104e-01 -5.18457294e-01
4.15248424e-01 -2.57991284e-01 -1.06647336e+00 6.38809577e-02
-1.16314244e+00 -1.55255780e-01 -3.30811709e-01 -8.70321840e-02
-2.34149009e-01 -1.48250580e-01 -8.04523826e-02 -9.50502992e-01
4.29325134e-01 -8.80763948e-01 7.12124944e-01 -4.47331108e-02
-5.95201433e-01 -9.49612975e-01 3.74174178e-01 -2.14461297e-01
4.17322934e-01 5.55865169e-01 1.25530016e+00 -1.00036889e-01
-7.92309582e-01 -3.45431894e-01 -1.46817729e-01 -1.15794882e-01
3.25643927e-01 1.07772362e+00 -9.70511198e-01 -1.04713246e-01
-2.21443936e-01 3.01009446e-01 1.65017679e-01 9.85516757e-02
7.80810237e-01 9.08684954e-02 -6.79171801e-01 -9.13898572e-02
1.71101558e+00 1.64019465e-01 8.49745572e-01 -9.93016884e-02
7.19783247e-01 6.62749648e-01 6.88104570e-01 6.27319098e-01
-1.40024960e-01 6.96748376e-01 8.23667228e-01 8.63605961e-02
-3.17726701e-01 -3.21625829e-01 5.21742940e-01 3.62662822e-01
4.39017452e-02 -7.98631757e-02 -1.44938827e+00 4.26902175e-01
-1.90120971e+00 -9.58972156e-01 -1.44737792e+00 2.30776715e+00
3.15436631e-01 3.31289828e-01 1.64307103e-01 2.80424118e-01
8.01806092e-01 -3.99708390e-01 -3.45396787e-01 -4.42300618e-01
4.70531702e-01 3.07369322e-01 5.01846075e-01 1.78484038e-01
-5.35931528e-01 2.30894729e-01 7.36493301e+00 1.20018229e-01
-8.39298844e-01 -5.03148325e-02 1.61176533e-01 6.20534308e-02
-5.67054391e-01 7.77479529e-01 -6.87083304e-01 3.40990901e-01
1.57841671e+00 -7.75762439e-01 1.03135139e-01 6.43855453e-01
6.91112518e-01 -4.46044594e-01 -1.57722759e+00 3.04968566e-01
-6.50082111e-01 -1.46001971e+00 -2.34456241e-01 4.55922514e-01
6.13852620e-01 -3.44382301e-02 -4.24655139e-01 -3.20888191e-01
9.62840736e-01 -1.15571105e+00 5.88852584e-01 5.84475577e-01
5.88439584e-01 -6.45892322e-01 3.48962814e-01 -1.81717221e-02
-8.66087496e-01 -3.26001495e-01 2.59854823e-01 -6.27602875e-01
1.66049540e-01 1.08279192e+00 -1.23414409e+00 7.36019671e-01
1.03521287e+00 6.68641388e-01 -5.17334223e-01 1.28090012e+00
-3.78810793e-01 9.44826722e-01 -2.44436249e-01 2.91655902e-02
-4.77180004e-01 3.01052749e-01 5.08260071e-01 1.03792572e+00
1.77584574e-01 -7.51388311e-01 -3.25223237e-01 1.33079076e+00
4.80415076e-01 -4.92432058e-01 -7.45390475e-01 -3.31755966e-01
3.88706982e-01 9.94241655e-01 -7.55890369e-01 -2.37637330e-02
-5.29931128e-01 8.55859295e-02 -3.04829895e-01 -8.79957825e-02
-6.60328388e-01 -2.66690075e-01 8.51218939e-01 7.45918036e-01
-8.31615552e-02 -4.30474758e-01 -6.25137985e-01 -5.69802701e-01
-3.97888198e-02 -7.84319222e-01 3.59740168e-01 -4.96711671e-01
-1.09940398e+00 -6.47679344e-03 3.33830893e-01 -9.48099494e-01
-4.73689377e-01 -4.94300783e-01 -8.13873291e-01 1.01300073e+00
-8.46097171e-01 -6.79175794e-01 -2.81977057e-01 4.01669741e-01
-1.82110384e-01 1.79271415e-01 7.62093782e-01 2.08882958e-01
-6.95701361e-01 -4.76876013e-02 -3.43908161e-01 -3.45947623e-01
6.63986146e-01 -1.50654161e+00 9.06036377e-01 1.36197388e+00
1.51111633e-01 1.05414307e+00 1.21831429e+00 -9.25437748e-01
-2.00925803e+00 -8.42973948e-01 6.52403712e-01 -6.64070308e-01
1.20586026e+00 -5.63334048e-01 -6.38528824e-01 7.84543335e-01
4.24009144e-01 1.09799337e-02 3.55820477e-01 5.39844811e-01
-4.03919190e-01 -2.46818155e-01 -7.55873382e-01 4.16442424e-01
8.08925211e-01 -2.35209554e-01 -2.73745924e-01 1.86684132e-01
8.46340835e-01 1.37505099e-01 -1.20623899e+00 1.95049122e-01
4.17643905e-01 -1.30979657e+00 4.47206259e-01 -4.90020126e-01
6.05346978e-01 -7.97617495e-01 8.22475180e-02 -9.94289756e-01
-3.79217905e-03 -1.16781425e+00 -3.22762489e-01 1.63620687e+00
3.91449153e-01 -6.56665444e-01 4.35440183e-01 6.07323468e-01
-9.40879434e-02 2.47177128e-02 -6.89149022e-01 -7.20495403e-01
-2.61439413e-01 -8.52345109e-01 7.21446514e-01 1.01222539e+00
6.17011786e-01 2.72126198e-01 1.39117166e-01 3.33797932e-01
7.33158410e-01 1.98666692e-01 9.39822376e-01 -1.59229183e+00
-6.06430471e-01 -4.41206872e-01 -6.41032159e-01 -1.98157892e-01
-3.76634836e-01 -6.85967147e-01 -1.72663197e-01 -1.15052068e+00
3.23623642e-02 -6.77364528e-01 -5.35946004e-02 4.54105675e-01
7.97038674e-02 -2.57027209e-01 -2.79625263e-02 -4.20532301e-02
3.55207324e-02 -3.55217248e-01 7.54615545e-01 1.95698813e-01
-1.64561868e-01 1.45224869e-01 -6.50462806e-01 1.43214107e-01
6.45807147e-01 -7.68043458e-01 -4.86941516e-01 -3.86834741e-02
8.67794991e-01 4.69364971e-01 6.80031359e-01 -9.85336959e-01
2.39489824e-01 -3.03894937e-01 -4.01658416e-02 -3.79650146e-01
-6.15196705e-01 -1.01946163e+00 7.32245445e-01 8.03185761e-01
-2.03379929e-01 3.31408381e-01 5.44762135e-01 7.15810716e-01
-2.56658316e-01 -8.38127136e-02 1.08776435e-01 1.82765648e-01
-8.59028280e-01 -1.25907376e-01 -5.35382390e-01 -2.06562728e-01
1.29391050e+00 2.49141559e-01 -7.53236651e-01 2.59878665e-01
-4.57283914e-01 1.76894292e-01 6.15131080e-01 9.30030465e-01
3.60630274e-01 -9.29394126e-01 -5.13191462e-01 3.17313492e-01
4.08915430e-01 -4.45459247e-01 -9.97080430e-02 8.37390661e-01
-4.42795128e-01 4.34690744e-01 -3.93770844e-01 -9.75955009e-01
-1.25831318e+00 9.06971812e-01 1.00736819e-01 -3.30821015e-02
-6.11862421e-01 2.25741476e-01 -1.34196475e-01 -3.13209593e-02
-3.19715261e-01 -5.23714066e-01 4.41354513e-01 -2.08255187e-01
6.91474378e-01 6.48755729e-01 4.52346325e-01 2.44831458e-01
-7.08092332e-01 -1.35962009e-01 8.39564651e-02 -1.18533194e-01
1.38729751e+00 1.80601478e-01 -6.71195865e-01 1.02936530e+00
6.12276435e-01 5.66401705e-02 -1.15834999e+00 5.26654899e-01
6.61317766e-01 -6.50550485e-01 2.22329810e-01 -1.01248336e+00
-7.65055120e-01 9.33292150e-01 6.54452085e-01 1.02968752e+00
9.52185154e-01 5.66375591e-02 -2.58454531e-01 -3.63575906e-01
7.27343619e-01 -4.08461094e-01 -2.65584171e-01 -1.63911656e-01
7.49727011e-01 -5.66547275e-01 9.17992964e-02 -7.54029155e-01
1.89718083e-01 1.52625120e+00 2.82360733e-01 -2.05963209e-01
8.01141024e-01 1.19320393e+00 -3.35492581e-01 -6.57078326e-01
-1.32408369e+00 7.95293599e-02 -3.42819333e-01 7.22855270e-01
8.53105783e-01 2.91111559e-01 -2.89097011e-01 -4.53137346e-02
1.35297686e-01 5.19977808e-01 1.00245714e+00 1.35553515e+00
-9.74829346e-02 -1.81060922e+00 -6.39701486e-01 5.98814368e-01
-3.33937913e-01 4.29667309e-02 -4.20257151e-01 1.24067676e+00
-6.92024976e-02 1.08409047e+00 2.77005695e-02 -4.18160975e-01
5.21467626e-01 5.54104848e-03 5.45255780e-01 -6.20810866e-01
-8.36568296e-01 -2.44882986e-01 2.41595462e-01 -1.13318729e+00
-1.51325598e-01 -1.12848878e+00 -1.19760895e+00 -5.06816387e-01
-2.27896482e-01 -1.70975268e-01 8.73726666e-01 7.87490606e-01
5.84109485e-01 1.30093288e+00 4.62023199e-01 -1.47899747e-01
-6.26549125e-02 -5.64349592e-01 -5.38562655e-01 6.69724047e-02
2.57781237e-01 -6.26842737e-01 -2.11928859e-01 3.90926808e-01]
|
[7.8822832107543945, 5.577849388122559]
|
d3273ad7-316a-4430-8428-22e5d1b3d2d3
|
neuralangelo-high-fidelity-neural-surface-1
|
2306.03092
| null |
https://arxiv.org/abs/2306.03092v2
|
https://arxiv.org/pdf/2306.03092v2.pdf
|
Neuralangelo: High-Fidelity Neural Surface Reconstruction
|
Neural surface reconstruction has been shown to be powerful for recovering dense 3D surfaces via image-based neural rendering. However, current methods struggle to recover detailed structures of real-world scenes. To address the issue, we present Neuralangelo, which combines the representation power of multi-resolution 3D hash grids with neural surface rendering. Two key ingredients enable our approach: (1) numerical gradients for computing higher-order derivatives as a smoothing operation and (2) coarse-to-fine optimization on the hash grids controlling different levels of details. Even without auxiliary inputs such as depth, Neuralangelo can effectively recover dense 3D surface structures from multi-view images with fidelity significantly surpassing previous methods, enabling detailed large-scale scene reconstruction from RGB video captures.
|
['Chen-Hsuan Lin', 'Ming-Yu Liu', 'Mathias Unberath', 'Russell H. Taylor', 'Alex Evans', 'Thomas Müller', 'Zhaoshuo Li']
|
2023-06-05
|
neuralangelo-high-fidelity-neural-surface
|
http://openaccess.thecvf.com//content/CVPR2023/html/Li_Neuralangelo_High-Fidelity_Neural_Surface_Reconstruction_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Li_Neuralangelo_High-Fidelity_Neural_Surface_Reconstruction_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['neural-rendering']
|
['computer-vision']
|
[ 3.72147232e-01 1.90801814e-01 5.34709394e-01 -1.34864137e-01
-1.15206504e+00 -3.06431115e-01 4.23378915e-01 -1.94256678e-02
-1.02124199e-01 5.85721433e-01 2.08466396e-01 -8.86754543e-02
1.98848829e-01 -1.25789058e+00 -9.62314725e-01 -4.66850132e-01
-1.47211671e-01 6.66784346e-01 3.43505561e-01 -2.74192572e-01
4.06763911e-01 9.96062458e-01 -2.03943801e+00 4.79501843e-01
4.19739515e-01 1.23367441e+00 5.30687831e-02 7.64673531e-01
-3.01661491e-01 4.26229179e-01 -1.27373725e-01 -4.55273800e-02
5.24119198e-01 -2.24558905e-01 -5.82821608e-01 6.94259182e-02
1.04640853e+00 -8.22214961e-01 -2.21094340e-01 8.08882654e-01
4.34062392e-01 8.32987353e-02 5.70541561e-01 -5.84782481e-01
-3.21715325e-01 -4.60360914e-01 -5.76922059e-01 -2.25189656e-01
5.12490928e-01 1.39114901e-01 7.87366033e-01 -1.19985592e+00
8.72351885e-01 1.48934257e+00 1.01753151e+00 4.12673652e-01
-1.55423343e+00 -3.70652139e-01 -3.87121916e-01 -3.01357299e-01
-1.35930336e+00 -3.98455232e-01 9.26812530e-01 -4.41153735e-01
1.18137932e+00 2.97439873e-01 9.39163029e-01 3.75835091e-01
2.14457944e-01 4.91366982e-01 1.43859982e+00 -7.54376948e-02
2.52901763e-01 -3.57814848e-01 -3.43095690e-01 9.27753270e-01
-7.30975345e-02 2.86353141e-01 -7.57109225e-01 -4.09251183e-01
1.73163116e+00 -2.54010912e-02 -4.23332274e-01 -2.23198533e-01
-1.05419862e+00 7.28986859e-01 4.67753053e-01 -2.38551676e-01
-6.32413626e-01 4.10711795e-01 9.90919396e-02 1.38326406e-01
8.18636000e-01 4.39229459e-01 -3.41222048e-01 -3.27027924e-02
-9.47586179e-01 5.14175117e-01 6.81225419e-01 7.52120793e-01
1.32674921e+00 1.72359288e-01 2.86625713e-01 4.25562471e-01
1.90942466e-01 5.26533842e-01 -1.44025654e-01 -1.53612053e+00
1.60183236e-01 5.94674468e-01 8.24625194e-02 -1.06284463e+00
-2.52595693e-01 1.23773292e-01 -9.52820241e-01 9.32424307e-01
2.19151333e-01 3.73836488e-01 -1.01132381e+00 1.10686445e+00
6.26537323e-01 1.59960195e-01 -2.11035997e-01 7.74171352e-01
7.66139746e-01 8.58445108e-01 -4.21575636e-01 1.08700231e-01
1.07135952e+00 -5.28522134e-01 -4.10576642e-01 2.49599572e-02
1.60850972e-01 -4.59813088e-01 1.31530857e+00 5.42743802e-01
-1.66449428e+00 -2.40056708e-01 -8.63688588e-01 -7.61833310e-01
-6.00998178e-02 -5.01576364e-01 8.04480076e-01 6.88180923e-02
-1.48249769e+00 9.16684031e-01 -9.70591486e-01 1.04332596e-01
7.38183618e-01 4.90532696e-01 -6.49562120e-01 -2.25182489e-01
-5.49059868e-01 3.09631526e-01 -7.69306198e-02 -5.51533476e-02
-8.88156712e-01 -1.02607810e+00 -8.92647505e-01 -8.86884183e-02
5.62934242e-02 -1.05330718e+00 1.00295937e+00 -5.24233222e-01
-1.49525714e+00 1.15174603e+00 -2.86943883e-01 1.02186669e-03
4.91693497e-01 -2.37128273e-01 4.43389982e-01 4.87218738e-01
-1.67668983e-01 6.04279280e-01 7.95629263e-01 -1.51288509e+00
-3.61418664e-01 -7.81456232e-01 -1.29289795e-02 4.01719600e-01
1.62882552e-01 -4.45005298e-01 -4.25456583e-01 -4.74997550e-01
4.88890827e-01 -3.45268995e-01 -4.53176737e-01 7.95545459e-01
-2.96316832e-01 1.19751692e-01 8.36593807e-01 -8.80214870e-01
5.96641243e-01 -2.09735298e+00 3.09179634e-01 2.06291318e-01
5.20532429e-01 -2.21295848e-01 -6.03794456e-02 2.31056675e-01
3.34025174e-01 2.65539903e-02 -5.12277424e-01 -7.88986087e-01
-2.04730809e-01 5.48416153e-02 -2.82877117e-01 5.64416230e-01
1.22785725e-01 8.67903769e-01 -5.36005139e-01 -2.12800130e-01
3.88485372e-01 1.24867713e+00 -8.79667521e-01 4.07767743e-01
-3.31549108e-01 6.74451888e-01 -4.37671155e-01 8.83754134e-01
7.66986012e-01 -5.41943729e-01 -2.97095299e-01 -3.91190201e-01
-2.52135962e-01 4.83645320e-01 -1.05131650e+00 1.95270681e+00
-4.01230901e-01 4.86854106e-01 7.92502820e-01 -3.50933313e-01
6.71990812e-01 2.17942446e-01 5.14676571e-01 -8.36296737e-01
-7.12720454e-02 2.08534732e-01 -9.39364135e-01 1.46762945e-03
6.42424524e-01 -6.16413534e-01 2.74427891e-01 3.05042744e-01
-4.45602626e-01 -9.16832983e-01 -5.38366914e-01 1.00000009e-01
1.16614950e+00 3.87442708e-01 6.30965084e-02 -2.14845687e-01
3.40833552e-02 1.50378034e-01 2.20340505e-01 3.61604512e-01
3.97311896e-01 1.11916935e+00 2.52485335e-01 -6.88396752e-01
-1.50770223e+00 -1.12859416e+00 -2.02701151e-01 4.85242546e-01
4.30679768e-02 -2.99328983e-01 -7.18980134e-01 -5.53744175e-02
3.07054967e-01 2.33348876e-01 -7.96464086e-01 3.84378105e-01
-8.31920028e-01 -3.97196263e-01 2.14991093e-01 3.98249596e-01
3.33537251e-01 -1.07715356e+00 -9.48559582e-01 1.59002706e-01
1.41044855e-01 -9.61518705e-01 -1.21310003e-01 2.21440271e-01
-1.48270881e+00 -1.05269349e+00 -6.61659598e-01 -3.07258964e-01
7.46399939e-01 3.87914121e-01 1.59038258e+00 4.58808661e-01
-4.13468331e-01 4.12905991e-01 3.35538208e-01 1.56261712e-01
-2.18488783e-01 -3.37089300e-01 -3.06197137e-01 -3.82517755e-01
-4.01211232e-01 -1.25298762e+00 -6.99310482e-01 6.18342385e-02
-1.10009885e+00 5.48253000e-01 3.89592409e-01 6.18431807e-01
1.19762063e+00 6.07213788e-02 -2.00703263e-01 -8.11592519e-01
1.80352747e-01 -1.53611317e-01 -8.45917225e-01 -2.68784255e-01
-3.10765952e-01 1.64201856e-02 4.44190234e-01 -1.11209080e-01
-7.45049953e-01 8.68225619e-02 -5.71905494e-01 -5.83629131e-01
-1.64319485e-01 1.36934102e-01 -2.97075771e-02 -3.67909342e-01
4.58241463e-01 2.24776730e-01 1.19694509e-01 -6.04126096e-01
1.91558957e-01 6.01032265e-02 6.28995240e-01 -6.52369440e-01
7.76689291e-01 1.16483784e+00 2.78247416e-01 -1.01018107e+00
-7.01715648e-01 -2.19507217e-02 -7.43142426e-01 -2.12439038e-02
1.06059182e+00 -1.01961553e+00 -9.21028495e-01 6.04926229e-01
-1.20203161e+00 -8.09134364e-01 -5.90354860e-01 -8.16356018e-02
-9.13186252e-01 1.23501956e-01 -1.11310661e+00 -7.87584782e-01
-4.10427511e-01 -1.06810176e+00 1.72835934e+00 -1.37420207e-01
1.11938291e-03 -8.53903472e-01 5.52862100e-02 2.31429949e-01
4.06592548e-01 7.94927776e-01 1.07863426e+00 5.84672034e-01
-1.33880675e+00 6.27659708e-02 -4.07986581e-01 1.56040832e-01
-1.24551788e-01 -9.87036899e-02 -1.18957698e+00 -2.42410496e-01
1.86935738e-01 -5.47911108e-01 5.26673555e-01 4.95662421e-01
1.13075960e+00 -3.48541647e-01 -8.32983851e-03 1.45633137e+00
1.73978543e+00 -2.08816081e-01 1.03274250e+00 3.22389901e-01
1.08264339e+00 5.31547785e-01 1.03000633e-01 4.27306235e-01
6.00688934e-01 5.80401003e-01 6.23503029e-01 -4.29516107e-01
-3.64219904e-01 -4.26976889e-01 -6.11018650e-02 8.22196901e-01
-3.05876940e-01 2.38360196e-01 -8.56765330e-01 1.82513803e-01
-1.22425902e+00 -6.06459677e-01 -3.41115236e-01 2.18699241e+00
7.01804101e-01 -2.45741438e-02 -1.16185173e-01 7.09236562e-02
1.01563275e-01 3.28656912e-01 -7.76222765e-01 -3.72075737e-02
-2.94922799e-01 4.59202915e-01 3.79784018e-01 8.80950034e-01
-6.26421154e-01 9.35173571e-01 7.12744474e+00 6.26099944e-01
-1.01245320e+00 1.16092421e-01 6.45564616e-01 -1.87219530e-01
-9.39770222e-01 -1.64715186e-01 -5.77642918e-01 -6.06476627e-02
5.39518774e-01 3.20636094e-01 7.34906971e-01 5.60413957e-01
-1.67807013e-01 -1.46784678e-01 -8.54565203e-01 1.15288854e+00
4.07273229e-03 -1.68323064e+00 3.28196764e-01 4.68699306e-01
7.05541372e-01 3.16748142e-01 -1.04658410e-01 -2.83961445e-01
4.37571853e-01 -1.29942250e+00 8.18102062e-01 5.82789958e-01
1.35734630e+00 -7.69880831e-01 6.04908466e-02 3.15031648e-01
-1.13783395e+00 3.61335307e-01 -3.91407192e-01 -1.33587345e-01
4.86354321e-01 7.95150161e-01 -3.04089099e-01 3.64732414e-01
8.70496511e-01 6.46983922e-01 -1.63111269e-01 5.36548078e-01
9.71964747e-03 6.13771677e-02 -6.78887546e-01 5.31364501e-01
6.05952851e-02 -3.49966437e-01 3.49353224e-01 7.41121531e-01
5.02927542e-01 5.58636785e-01 -4.89376225e-02 1.06016099e+00
4.36335220e-04 -2.05915481e-01 -9.27175879e-01 5.02670780e-02
1.62557170e-01 1.00402474e+00 -8.02514076e-01 -3.35232228e-01
-2.25649282e-01 1.24992406e+00 5.95621109e-01 3.67506802e-01
-1.71892628e-01 4.18760069e-02 7.39424765e-01 7.19887137e-01
2.35884145e-01 -8.20018947e-01 -7.06018388e-01 -1.00639141e+00
-6.85430020e-02 -7.09206045e-01 -6.00684695e-02 -1.11547816e+00
-1.03924227e+00 6.97958887e-01 -3.56958508e-01 -7.76223600e-01
3.65249962e-02 -6.07118189e-01 -1.80463016e-01 1.06230843e+00
-1.73043621e+00 -1.18481147e+00 -4.81845886e-01 6.96583092e-01
3.39692056e-01 4.36269909e-01 9.50102687e-01 1.58987373e-01
9.07213464e-02 -6.75218999e-02 -1.48090854e-01 -3.27530235e-01
1.04580209e-01 -1.15486574e+00 8.63952041e-01 4.20556307e-01
-1.02887280e-01 4.91787761e-01 4.73618150e-01 -8.50440025e-01
-1.89155984e+00 -5.64312160e-01 3.42177361e-01 -6.09087229e-01
1.37799144e-01 -4.38112229e-01 -1.30161393e+00 5.88981569e-01
-1.86904579e-01 1.92392975e-01 4.32052553e-01 -1.42415509e-01
-4.77935940e-01 3.01869780e-01 -1.45138311e+00 3.87606472e-01
1.22420728e+00 -9.21566188e-01 -1.42212600e-01 8.27534795e-02
6.05875432e-01 -7.96447873e-01 -1.10411251e+00 5.97831666e-01
7.14512885e-01 -1.59290946e+00 1.36806154e+00 -1.39471903e-01
6.92082942e-01 -3.24258775e-01 -6.10035896e-01 -8.72500300e-01
-2.34563917e-01 -7.32564807e-01 -3.02953809e-01 5.51189244e-01
-1.40765235e-01 -4.45905417e-01 9.94347334e-01 8.27879488e-01
-3.81129265e-01 -1.03534186e+00 -9.27815616e-01 -2.48126462e-01
3.69946398e-02 -4.64616448e-01 7.18274593e-01 8.03899944e-01
-8.16315591e-01 7.65837505e-02 -1.68952376e-01 2.02282831e-01
1.03512084e+00 1.77781463e-01 9.65627849e-01 -1.41417873e+00
-2.88442135e-01 -3.68628383e-01 -3.74136031e-01 -1.41163099e+00
-1.90537572e-02 -6.82403088e-01 8.62646401e-02 -1.84153426e+00
-1.55226551e-02 -6.85589671e-01 2.90294051e-01 2.99975038e-01
1.27301931e-01 7.63983667e-01 -1.10384926e-01 3.55971068e-01
-1.41291976e-01 7.41407871e-01 1.61870289e+00 3.75541896e-01
-7.66260922e-02 -5.33922553e-01 -4.80989039e-01 8.95330489e-01
2.43749291e-01 -2.47211143e-01 -8.92588720e-02 -7.55816758e-01
6.06794715e-01 4.67071623e-01 7.04556286e-01 -9.17446792e-01
5.53152151e-02 -4.51043881e-02 5.22719502e-01 -9.98040259e-01
9.37722087e-01 -7.46786773e-01 4.46255088e-01 1.34574026e-01
9.13034379e-03 2.78576642e-01 3.53938848e-01 4.71714199e-01
-4.53406610e-02 4.11052108e-01 9.13817644e-01 -5.16350150e-01
-4.76620674e-01 6.67523563e-01 -2.00581867e-02 -1.81026030e-02
4.02573973e-01 -4.96231288e-01 7.27924630e-02 -3.11566114e-01
-6.09308720e-01 -3.36870492e-01 1.38270092e+00 -2.19957978e-01
1.06807029e+00 -1.36615133e+00 -5.73879004e-01 6.18841231e-01
-3.68732184e-01 9.02515531e-01 3.96245241e-01 4.43212837e-01
-1.00483811e+00 -6.52800724e-02 -1.86118335e-01 -7.22551823e-01
-9.64915097e-01 9.31081548e-02 3.82847309e-01 -1.06415659e-01
-1.43069994e+00 9.92173135e-01 6.07957304e-01 -5.64859033e-01
7.57022724e-02 -3.24417323e-01 4.75139976e-01 -3.26206863e-01
6.89424932e-01 3.18468422e-01 1.21885560e-01 -5.37464261e-01
-9.71165765e-03 1.00479460e+00 3.58333886e-01 -3.46086264e-01
1.73999298e+00 -1.87062193e-02 -4.04664695e-01 6.23486042e-01
1.40115297e+00 -2.52661705e-02 -1.93306494e+00 8.44323728e-03
-4.30592924e-01 -8.10506344e-01 4.23286080e-01 -3.91781211e-01
-1.03018916e+00 1.17586553e+00 3.51208240e-01 5.08701317e-02
1.08613205e+00 4.85685701e-03 1.33047545e+00 1.74852952e-01
8.35350573e-01 -6.09275699e-01 4.98100370e-02 4.69273269e-01
1.01767266e+00 -9.65974987e-01 2.44800389e-01 -5.00733972e-01
7.75630176e-02 1.08550334e+00 2.36723185e-01 -4.62756246e-01
6.41040564e-01 7.58303881e-01 -2.36734115e-02 -8.22469950e-01
-5.61490715e-01 1.14289969e-01 3.04777801e-01 5.36979556e-01
2.92954922e-01 -1.52658924e-01 5.10159910e-01 -2.90903211e-01
-2.71831334e-01 -1.26306996e-01 1.84003949e-01 8.72354746e-01
-3.58601004e-01 -7.07860768e-01 -4.55233455e-01 4.25544828e-01
-2.93866396e-01 -2.60917753e-01 -8.78694504e-02 6.36378467e-01
-3.93571138e-01 2.49368683e-01 4.56692904e-01 -2.41260558e-01
5.78206539e-01 -2.04097837e-01 8.02091718e-01 -5.95613003e-01
-4.83743668e-01 2.29609728e-01 -7.50309080e-02 -1.30670917e+00
-2.11562663e-01 -6.16942585e-01 -1.40223718e+00 -4.14690018e-01
1.79730773e-01 -2.42813855e-01 8.04072201e-01 6.29955649e-01
5.45379341e-01 3.10528308e-01 3.77845585e-01 -1.81988227e+00
-1.11481011e-01 -4.48452652e-01 -7.73361683e-01 4.35093999e-01
7.98789263e-01 -4.99931961e-01 -6.17648900e-01 -6.24684840e-02]
|
[9.135470390319824, -3.1236929893493652]
|
4b9421cb-c8ed-45c3-8cae-73f119776751
|
echovest-real-time-sound-classification-and
|
2307.04604
| null |
https://arxiv.org/abs/2307.04604v1
|
https://arxiv.org/pdf/2307.04604v1.pdf
|
EchoVest: Real-Time Sound Classification and Depth Perception Expressed through Transcutaneous Electrical Nerve Stimulation
|
Over 1.5 billion people worldwide live with hearing impairment. Despite various technologies that have been created for individuals with such disabilities, most of these technologies are either extremely expensive or inaccessible for everyday use in low-medium income countries. In order to combat this issue, we have developed a new assistive device, EchoVest, for blind/deaf people to intuitively become more aware of their environment. EchoVest transmits vibrations to the user's body by utilizing transcutaneous electric nerve stimulation (TENS) based on the source of the sounds. EchoVest also provides various features, including sound localization, sound classification, noise reduction, and depth perception. We aimed to outperform CNN-based machine-learning models, the most commonly used machine learning model for classification tasks, in accuracy and computational costs. To do so, we developed and employed a novel audio pipeline that adapts the Audio Spectrogram Transformer (AST) model, an attention-based model, for our sound classification purposes, and Fast Fourier Transforms for noise reduction. The application of Otsu's Method helped us find the optimal thresholds for background noise sound filtering and gave us much greater accuracy. In order to calculate direction and depth accurately, we applied Complex Time Difference of Arrival algorithms and SOTA localization. Our last improvement was to use blind source separation to make our algorithms applicable to multiple microphone inputs. The final algorithm achieved state-of-the-art results on numerous checkpoints, including a 95.7\% accuracy on the ESC-50 dataset for environmental sound classification.
|
['Ryan Park', 'Siddhant Sood', 'Jesse Choe']
|
2023-07-10
| null | null | null | null |
['environmental-sound-classification', 'sound-classification', 'classification-1']
|
['audio', 'audio', 'methodology']
|
[-8.39887932e-02 -3.83478582e-01 5.79603255e-01 1.26620233e-01
-1.11998475e+00 -4.46694374e-01 7.49596134e-02 -1.04964741e-01
-6.52393281e-01 4.79688406e-01 6.05760992e-01 -1.74909592e-01
-5.14099700e-03 -7.24901497e-01 -2.40961447e-01 -6.36028171e-01
-1.11065611e-01 1.30851820e-01 2.88668454e-01 -2.79178947e-01
2.55343139e-01 3.73213261e-01 -2.18020606e+00 3.58075172e-01
8.60282362e-01 1.24277937e+00 3.88095826e-01 9.61676776e-01
4.96340506e-02 2.30445981e-01 -8.86203587e-01 -2.20892459e-01
3.70362811e-02 -2.21389309e-01 -5.36843717e-01 -8.02431047e-01
4.36611712e-01 -4.69905704e-01 -1.71408921e-01 8.55185568e-01
1.43072271e+00 -7.52184093e-02 4.69274104e-01 -9.11564291e-01
-2.81559408e-01 5.04077554e-01 -4.81361970e-02 2.05546990e-01
8.28876972e-01 1.87363505e-01 7.55455375e-01 -7.76286662e-01
-1.62951685e-02 1.06731331e+00 1.02986836e+00 6.41692281e-01
-9.73364353e-01 -9.32095826e-01 -3.29796135e-01 5.85102379e-01
-1.19884074e+00 -9.15083051e-01 6.75952315e-01 -3.92262191e-01
1.45194399e+00 3.99800956e-01 1.04837787e+00 1.16013420e+00
5.07101715e-02 4.16569948e-01 1.02797461e+00 -6.85283303e-01
3.88193756e-01 -6.03976250e-02 -2.65359670e-01 2.82960773e-01
-1.83191553e-01 1.72562018e-01 -8.43237162e-01 -2.17931345e-01
4.85064149e-01 -5.31111658e-01 -6.71882212e-01 4.38539714e-01
-8.98153305e-01 4.09128398e-01 3.46370786e-01 5.56223571e-01
-3.49397242e-01 2.26111501e-01 3.54285628e-01 3.23484600e-01
5.74815452e-01 4.88694459e-01 -5.60247719e-01 -6.35415494e-01
-6.63162768e-01 1.88950107e-01 7.54489362e-01 1.69055641e-01
1.58664063e-01 1.74417228e-01 -7.17280507e-02 1.28767240e+00
4.80672449e-01 8.01850140e-01 6.70536518e-01 -8.13504934e-01
3.57392758e-01 1.10004917e-01 -1.64043512e-02 -8.24782193e-01
-7.20153928e-01 -6.99942470e-01 -7.16562510e-01 2.91922420e-01
2.24811152e-01 -1.88986734e-01 -1.04868317e+00 1.64286709e+00
5.43801524e-02 4.14377630e-01 -2.34768435e-01 1.02033198e+00
6.80871308e-01 3.45888138e-01 -2.90101934e-02 2.97180153e-02
1.19907081e+00 -4.96265620e-01 -7.24879801e-01 -1.23207167e-01
3.97024095e-01 -8.40546429e-01 1.24951196e+00 8.38963151e-01
-9.38183963e-01 -4.47472185e-01 -7.77603388e-01 1.32813409e-01
-4.01392788e-01 2.20826194e-02 5.57607710e-01 1.23667634e+00
-1.36484301e+00 4.52993214e-01 -7.96302974e-01 -3.81075263e-01
3.96104217e-01 5.08645356e-01 -2.72064328e-01 2.13593751e-01
-1.07070363e+00 7.44597018e-01 -2.93763876e-01 9.53902528e-02
-5.73933959e-01 -7.93392956e-01 -5.23184538e-01 6.17688708e-02
-3.16664904e-01 -8.40824425e-01 1.40826201e+00 -5.55044532e-01
-1.87848198e+00 6.72788918e-01 -1.74934000e-01 -3.94474715e-01
2.71419913e-01 -5.81074536e-01 -6.60582125e-01 1.36240289e-01
8.66713971e-02 3.55205387e-01 1.07851946e+00 -8.05593431e-01
-6.99761689e-01 -4.95338380e-01 -5.89228809e-01 8.92610475e-03
-7.46303678e-01 3.02447379e-01 -1.15159623e-01 -6.60918117e-01
2.91943014e-01 -5.45867085e-01 2.15723753e-01 -1.36704817e-01
-3.00037682e-01 -9.63804573e-02 5.73189914e-01 -1.21211433e+00
1.03389955e+00 -2.23883557e+00 -6.99108019e-02 1.18077807e-01
1.00283481e-01 5.13093293e-01 -1.01777650e-01 3.20952237e-01
-1.44003322e-02 5.19717531e-03 -2.34137341e-01 -5.92877746e-01
-7.42636696e-02 -2.96139449e-01 -2.77025759e-01 3.12365115e-01
-2.63512403e-01 3.21854115e-01 -7.46864915e-01 -4.96916007e-03
3.93383056e-01 8.70226681e-01 -8.10549438e-01 2.83951432e-01
2.34003097e-01 2.75951982e-01 1.68027785e-02 7.51901805e-01
7.12634087e-01 6.42729163e-01 -6.01346135e-01 -7.03237280e-02
-2.54030108e-01 5.63605905e-01 -1.12194979e+00 1.74903798e+00
-8.66561949e-01 7.17497051e-01 4.87891793e-01 -5.71689606e-01
7.31460035e-01 7.24946439e-01 4.06310350e-01 -7.71389902e-01
2.17115685e-01 7.34305203e-01 -1.92838628e-02 -1.04952002e+00
-5.16032502e-02 -1.70629084e-01 4.81354743e-01 -4.27654246e-03
2.97230445e-02 -4.37752485e-01 -4.88197148e-01 -2.09669992e-01
1.28875434e+00 -2.63096184e-01 -3.07996888e-02 -2.43465155e-02
2.95239419e-01 -5.50898373e-01 1.66520253e-01 5.79360843e-01
-2.19324335e-01 9.75479543e-01 -1.76682174e-01 -1.55837893e-01
-3.15734893e-01 -1.26050162e+00 -1.42585382e-01 1.04317820e+00
-4.69873458e-01 -4.06303734e-01 -8.89089406e-01 -6.60516173e-02
-1.72732268e-02 7.78043270e-01 -7.00040609e-02 -1.48787841e-01
-1.84769675e-01 -4.47975367e-01 9.13618684e-01 4.67867255e-01
6.74732685e-01 -1.05306304e+00 -6.65296078e-01 4.33376104e-01
-5.52940667e-01 -7.87760079e-01 -5.38053252e-02 2.31114328e-01
-4.32133317e-01 -7.37365246e-01 -9.01250422e-01 -7.47182369e-01
-8.78139734e-02 4.18837816e-02 6.29455388e-01 -1.90028071e-01
-5.63567340e-01 7.40918338e-01 -4.71171141e-01 -9.94901299e-01
-1.14718765e-01 -8.75151902e-02 5.22357881e-01 -1.03349105e-01
1.71139300e-01 -1.22640157e+00 -8.04174006e-01 5.80856949e-03
-5.10360837e-01 -5.57365596e-01 4.81731266e-01 4.99450833e-01
7.23056272e-02 1.96518660e-01 8.45778883e-01 1.94886297e-01
9.58567917e-01 -1.40954211e-01 -1.39302075e-01 -3.93119633e-01
-2.19795048e-01 -3.61638218e-01 5.10603607e-01 -4.84980434e-01
-8.55870068e-01 -9.72113907e-02 -9.92937684e-01 -5.38589060e-02
-3.86840791e-01 2.17479840e-01 -2.62607396e-01 -1.28778905e-01
8.55153978e-01 1.16245598e-01 -1.31598696e-01 -8.72703850e-01
-5.26142567e-02 1.57885098e+00 8.98211598e-01 -1.61793515e-01
4.40572858e-01 3.52386326e-01 -2.61962414e-01 -1.27152824e+00
-3.12474370e-01 -6.09832287e-01 -2.42844850e-01 -3.39026362e-01
7.68660784e-01 -8.58644724e-01 -8.23455691e-01 1.29931009e+00
-1.36039793e+00 -2.29453355e-01 4.04574238e-02 7.41208017e-01
-3.79197717e-01 9.52175558e-02 -3.90780389e-01 -1.27563500e+00
-6.30218029e-01 -8.37707698e-01 1.03761220e+00 9.70489010e-02
-3.05283040e-01 -5.24634600e-01 3.86291146e-01 3.46145272e-01
8.89650285e-01 -2.29794025e-01 7.64621675e-01 -2.24780932e-01
1.07370652e-02 -1.69906318e-01 7.17129335e-02 6.27426088e-01
1.92688793e-01 -3.31411958e-01 -1.75432706e+00 -1.48620009e-01
4.40304764e-02 -1.18238643e-01 1.06181145e+00 7.37925589e-01
1.10783947e+00 -1.80229060e-02 -2.96911359e-01 5.55071354e-01
9.92635548e-01 2.89243639e-01 1.00553083e+00 2.91209817e-01
4.04001713e-01 5.69629073e-01 8.58801603e-02 3.82919163e-01
2.28352100e-01 8.32771540e-01 7.04543173e-01 -1.04820997e-01
-5.83458960e-01 -1.94641411e-01 4.58284080e-01 6.78653896e-01
-3.95048589e-01 -1.82079807e-01 -9.68305111e-01 6.87300742e-01
-1.11581409e+00 -7.27241039e-01 -2.15194389e-01 2.21417403e+00
6.57950163e-01 -1.26092667e-02 1.26392692e-01 8.59498441e-01
4.28548038e-01 -3.45590442e-01 -2.77618527e-01 -2.89988667e-01
-1.19688138e-02 8.50591063e-01 2.11047351e-01 3.42236608e-01
-9.02177334e-01 6.72366917e-01 6.36676979e+00 6.14132583e-01
-1.51315951e+00 1.65238842e-01 -4.56510764e-03 -1.40771478e-01
-8.44918657e-03 -6.74207509e-01 -3.80817384e-01 4.57447052e-01
1.01090562e+00 6.31534278e-01 7.48679578e-01 5.79079270e-01
4.11852479e-01 -2.11482719e-01 -7.87068009e-01 1.41555095e+00
1.05354227e-02 -8.36903870e-01 -3.70189428e-01 -2.72322506e-01
1.48277134e-02 3.02403688e-01 1.17092021e-01 -3.52631649e-03
-4.28698957e-01 -9.91230965e-01 9.68542159e-01 5.66658199e-01
9.50672507e-01 -6.34362698e-01 7.87451088e-01 1.04334377e-01
-1.23171031e+00 -5.25892377e-01 -1.90037210e-02 -4.56921190e-01
1.49885014e-01 7.65434563e-01 -9.24046755e-01 1.40184954e-01
1.43422663e+00 2.49765381e-01 -3.55353057e-01 1.56110477e+00
-2.42726833e-01 1.01233983e+00 -6.65393770e-01 -1.00036427e-01
-4.84242082e-01 3.62366349e-01 9.29284990e-01 1.16555357e+00
7.97777474e-01 -2.34558582e-02 -5.42029202e-01 4.97772813e-01
3.12831670e-01 6.54475838e-02 -4.88786250e-01 2.91966319e-01
5.83535075e-01 8.34388733e-01 -3.66420150e-01 3.36086124e-01
-2.33725175e-01 1.01768303e+00 -3.00087541e-01 4.06370789e-01
-4.73575175e-01 -9.20666873e-01 1.01700234e+00 3.29532772e-01
-5.56349941e-02 -2.40447700e-01 -3.78113836e-01 -7.15012789e-01
3.15047592e-01 -8.29358757e-01 2.25179400e-02 -1.07180321e+00
-1.14625478e+00 6.37793005e-01 -4.01367307e-01 -1.10457420e+00
-2.01631188e-01 -8.77345264e-01 -7.58107960e-01 1.05935276e+00
-1.32527471e+00 -9.48890388e-01 -2.88670808e-01 7.94398725e-01
3.40338826e-01 -2.54584223e-01 1.28570247e+00 6.57145798e-01
-2.39523098e-01 6.27012968e-01 -1.70939207e-01 3.22660878e-02
8.92156482e-01 -1.09567010e+00 3.33793342e-01 6.79729581e-01
-4.23177481e-02 4.61292416e-01 8.58268142e-01 -4.06596243e-01
-1.26210296e+00 -7.25426614e-01 8.48845959e-01 -1.20040536e-01
5.20177007e-01 -4.81556237e-01 -6.26996875e-01 1.81031719e-01
-7.66221285e-02 -2.94663817e-01 6.35185778e-01 1.11582972e-01
-4.15603012e-01 -4.37252522e-01 -1.25507832e+00 3.53314668e-01
1.27264154e+00 -9.01246607e-01 -5.39288759e-01 1.77402884e-01
6.91551268e-01 -1.02520689e-01 -4.23773378e-01 3.34335506e-01
8.26376736e-01 -1.15709317e+00 1.03568935e+00 -6.14193492e-02
-7.45356902e-02 -1.52590662e-01 -3.49751621e-01 -1.79289305e+00
-8.36324170e-02 -7.98990428e-01 3.19659300e-02 1.39821136e+00
3.22556168e-01 -1.17245889e+00 2.62495011e-01 1.52366221e-01
-5.91391563e-01 -3.60269725e-01 -1.43625045e+00 -7.33477533e-01
-1.47225946e-01 -1.19564283e+00 5.62124252e-01 3.55499059e-01
1.85394034e-01 1.42252207e-01 -1.20295227e-01 4.84911263e-01
3.47667605e-01 -5.33225715e-01 3.93581182e-01 -1.49114633e+00
-3.35232198e-01 -4.46990669e-01 -8.04593861e-01 -6.77079976e-01
-6.27110451e-02 -5.42389870e-01 1.37551099e-01 -1.83351219e+00
-4.46576267e-01 -2.67379075e-01 -2.37288743e-01 6.13472462e-01
1.48006082e-01 4.00297761e-01 -1.91795826e-01 -1.71854883e-01
2.37637147e-01 4.89392251e-01 7.80169845e-01 -1.79042265e-01
-5.48163533e-01 4.32370991e-01 -6.96051538e-01 7.79026926e-01
8.52547884e-01 -3.22569072e-01 -1.34536624e-01 -6.85979903e-01
3.08001071e-01 -1.31746501e-01 7.54728138e-01 -1.64555418e+00
2.70246714e-01 4.96430814e-01 2.69948155e-01 -3.78626585e-01
7.45765030e-01 -6.89396918e-01 -1.34817809e-01 5.58045268e-01
1.12846673e-01 -5.83461285e-01 3.72267783e-01 1.01564988e-01
-1.84936523e-01 1.29055455e-01 7.76587903e-01 1.52352512e-01
-3.24918270e-01 -2.17183203e-01 -7.57223904e-01 -3.02882105e-01
3.77019495e-01 5.55595458e-02 -3.39979768e-01 -6.59325898e-01
-6.77992940e-01 -3.32447082e-01 -1.04352318e-01 3.61360282e-01
7.83894122e-01 -9.64791954e-01 -7.55953193e-01 4.20040369e-01
-5.17244674e-02 -2.42562592e-01 3.15109193e-01 8.45510006e-01
-3.94832283e-01 3.78394037e-01 -2.69239247e-01 -6.14950478e-01
-1.33864272e+00 -4.58497740e-02 7.37568259e-01 5.30455589e-01
-6.13336027e-01 1.38959086e+00 -4.29274887e-01 -3.71021390e-01
6.77265346e-01 -7.64019370e-01 -2.88107723e-01 1.77781224e-01
8.46649289e-01 8.43295872e-01 6.34980917e-01 -2.63020545e-01
-4.09844756e-01 7.82351553e-01 5.03412485e-01 -5.66575050e-01
1.41359794e+00 -1.27428085e-01 -2.04435438e-01 2.73000628e-01
1.11781442e+00 4.96594697e-01 -5.91343462e-01 3.58483851e-01
-4.71612126e-01 -5.03560543e-01 5.75640857e-01 -1.24596274e+00
-8.49610567e-01 1.26563942e+00 1.33596361e+00 4.55229580e-01
1.76343322e+00 -5.08945063e-02 9.79876578e-01 3.30875009e-01
6.34577930e-01 -9.30817544e-01 2.86183674e-02 4.80086684e-01
1.09979367e+00 -7.16879308e-01 -6.74345136e-01 -4.46819335e-01
4.38584015e-02 1.04850662e+00 2.66482919e-01 3.52266103e-01
7.85064578e-01 6.11419797e-01 4.58522826e-01 1.85356423e-01
-1.57474771e-01 -3.96468639e-01 2.00661108e-01 1.27516150e+00
4.14706409e-01 2.48353615e-01 1.85372874e-01 1.04165721e+00
-8.66406381e-01 2.19630264e-02 3.67387868e-02 5.51473141e-01
-7.03518331e-01 -8.33244383e-01 -8.09071183e-01 5.20026147e-01
-4.53673840e-01 -2.90410161e-01 -4.89211798e-01 1.38829842e-01
4.35255736e-01 1.54303563e+00 -5.19785881e-02 -8.94533217e-01
5.72697997e-01 2.80158013e-01 4.51641083e-01 -5.36392689e-01
-6.29297018e-01 1.34902254e-01 3.14507447e-02 -5.55197954e-01
-8.57519060e-02 -6.49544656e-01 -1.28784406e+00 6.29278496e-02
-2.77295828e-01 -1.77961871e-01 1.07529199e+00 7.87402391e-01
3.33260030e-01 7.43331671e-01 4.97092426e-01 -1.17467439e+00
-3.38165879e-01 -1.38641572e+00 -7.80778170e-01 -1.16483182e-01
6.67037547e-01 -6.55872703e-01 -7.63092339e-01 -2.25493446e-01]
|
[15.038809776306152, 5.759720325469971]
|
939b9661-6aa3-45a7-a15f-658a815f9959
|
temporal-relation-classification-in-persian
| null | null |
https://aclanthology.org/R13-1034
|
https://aclanthology.org/R13-1034.pdf
|
Temporal Relation Classification in Persian and English contexts
| null |
['Yadollah Yaghoobzadeh', 'Gholamreza Ghassem-Sani', 'Negin Karimi Hosseini', 'Mirrosh', 'Seyed Abolghasem el', 'Mahbaneh Eshaghzadeh Torbati']
|
2013-09-01
|
temporal-relation-classification-in-persian-1
|
https://aclanthology.org/R13-1034
|
https://aclanthology.org/R13-1034.pdf
|
ranlp-2013-9
|
['temporal-relation-classification']
|
['natural-language-processing']
|
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
|
[-7.381738662719727, 3.5931644439697266]
|
b3f2d3fb-66f7-4e7d-8bd8-e7a3f3035aa5
|
medical-diagnosis-with-large-scale-multimodal
|
2212.09162
| null |
https://arxiv.org/abs/2212.09162v2
|
https://arxiv.org/pdf/2212.09162v2.pdf
|
Medical Diagnosis with Large Scale Multimodal Transformers: Leveraging Diverse Data for More Accurate Diagnosis
|
Multimodal deep learning has been used to predict clinical endpoints and diagnoses from clinical routine data. However, these models suffer from scaling issues: they have to learn pairwise interactions between each piece of information in each data type, thereby escalating model complexity beyond manageable scales. This has so far precluded a widespread use of multimodal deep learning. Here, we present a new technical approach of "learnable synergies", in which the model only selects relevant interactions between data modalities and keeps an "internal memory" of relevant data. Our approach is easily scalable and naturally adapts to multimodal data inputs from clinical routine. We demonstrate this approach on three large multimodal datasets from radiology and ophthalmology and show that it outperforms state-of-the-art models in clinically relevant diagnosis tasks. Our new approach is transferable and will allow the application of multimodal deep learning to a broad set of clinically relevant problems.
|
['Daniel Truhn', 'Jakob Nikolas Kather', 'Sven Nebelung', 'Christiane Kuhl', 'Keno Bressem', 'Johannes Stegmaier', 'Christoph Haarburger', 'Soroosh Tayebi Arasteh', 'Tianyu Han', 'Tianci Wang', 'Gustav Mueller-Franzes', 'Firas Khader']
|
2022-12-18
| null | null | null | null |
['multimodal-deep-learning']
|
['natural-language-processing']
|
[ 3.89707267e-01 -1.44959893e-02 -4.95507509e-01 -4.12213534e-01
-1.17599952e+00 -4.84826535e-01 4.01925892e-01 6.95355117e-01
-4.27152187e-01 8.17544937e-01 4.86413419e-01 -3.37616742e-01
-7.43824720e-01 -4.37642872e-01 -4.98776495e-01 -7.52432048e-01
-4.48617101e-01 8.52846384e-01 -1.07400179e-01 -1.83642268e-01
-2.70745069e-01 4.91641760e-01 -1.12626243e+00 8.04989934e-01
3.30134541e-01 8.09043825e-01 -6.09490275e-02 6.91326916e-01
9.19275954e-02 9.16670740e-01 -2.01442271e-01 -1.95829168e-01
-9.84857455e-02 -3.71110022e-01 -1.00802422e+00 -4.18509953e-02
6.03706539e-01 -3.59242439e-01 -3.90355498e-01 4.52423394e-01
9.21386778e-01 -2.27164969e-01 6.56305790e-01 -9.82145011e-01
-3.99166167e-01 4.71034765e-01 -4.78759438e-01 6.48632497e-02
2.20163807e-01 1.82570651e-01 1.16748595e+00 -5.63413501e-01
8.50489974e-01 1.13907027e+00 7.13262320e-01 7.43292630e-01
-1.54149401e+00 -3.22296917e-01 -1.70728806e-02 -8.09495971e-02
-8.22426796e-01 -2.81110525e-01 3.24777991e-01 -6.39228702e-01
9.54768598e-01 5.00710011e-01 5.81730962e-01 1.25300980e+00
4.49947476e-01 7.50661135e-01 9.54992175e-01 -1.34325638e-01
7.14159682e-02 -2.11273685e-01 2.10948095e-01 8.07178319e-01
-4.47404431e-03 2.16819033e-01 -6.33946240e-01 -6.13061726e-01
5.81013322e-01 1.66374892e-01 -1.28638357e-01 -4.44838673e-01
-1.61163437e+00 8.97545815e-01 3.28310311e-01 2.74397731e-01
-3.68528128e-01 3.49121630e-01 4.52755094e-01 4.29340512e-01
1.06852546e-01 5.80283344e-01 -7.35256791e-01 1.51368141e-01
-6.14732563e-01 2.26006404e-01 7.40531027e-01 4.46554810e-01
3.25173408e-01 -6.26101434e-01 -2.38146037e-01 9.73238826e-01
2.37332329e-01 4.11316097e-01 3.76893610e-01 -1.02684200e+00
4.46418464e-01 6.73436701e-01 -1.39686512e-02 -7.50127435e-01
-1.17378449e+00 -3.88123333e-01 -9.88189697e-01 -6.63658557e-03
3.68011773e-01 -4.00533140e-01 -9.93660212e-01 1.70710218e+00
2.45915651e-01 -9.37604979e-02 -8.56118277e-02 7.45689869e-01
1.18640006e+00 1.11769907e-01 4.78744507e-01 -9.96386111e-02
1.26364541e+00 -5.50509512e-01 -6.50972784e-01 -5.92414849e-02
9.02938902e-01 -6.72929287e-01 7.13856936e-01 5.21003246e-01
-1.19601429e+00 -8.94076601e-02 -7.30839908e-01 -1.62184462e-01
-3.76192987e-01 8.75339061e-02 1.05148923e+00 2.52117008e-01
-1.17105174e+00 5.75063765e-01 -9.05683637e-01 -4.74746823e-01
4.88155693e-01 1.16391563e+00 -6.67648137e-01 -5.63808419e-02
-1.14622390e+00 1.05389571e+00 1.74443588e-01 3.89985554e-02
-8.81418228e-01 -8.82451832e-01 -6.48049831e-01 -1.67290177e-02
2.05507964e-01 -1.33349121e+00 1.22514868e+00 -7.36787319e-01
-1.18593931e+00 9.57853079e-01 2.79584229e-02 -1.48621604e-01
4.60386038e-01 1.85485054e-02 -2.41398439e-01 2.02008992e-01
-2.22008750e-01 1.00126851e+00 4.41341221e-01 -1.17207134e+00
-5.53638458e-01 -4.37813580e-01 -1.30654499e-01 2.07167536e-01
-4.34897304e-01 2.25213226e-02 -2.95796186e-01 -3.50064158e-01
-2.12862603e-02 -1.01602852e+00 -7.51589298e-01 3.33618931e-02
-3.26575488e-01 -1.34207070e-01 4.50571835e-01 -2.90968090e-01
1.05285728e+00 -1.90540886e+00 6.33057356e-01 2.44742617e-01
7.00636268e-01 1.49437608e-02 -5.23978174e-01 7.28723168e-01
-2.78674901e-01 1.51276931e-01 -8.02357122e-02 -5.42587459e-01
-1.64049715e-01 2.84045637e-01 1.94521900e-02 3.89693111e-01
3.14255983e-01 1.25409460e+00 -1.00257885e+00 -5.94357133e-01
2.80036539e-01 5.66924572e-01 -5.65918982e-01 1.32199481e-01
-2.74025232e-01 6.02114618e-01 -5.37387609e-01 7.51856446e-01
3.60164583e-01 -8.36534321e-01 5.50692916e-01 -2.40981296e-01
3.35449427e-01 5.54185137e-02 -7.16488540e-01 1.62166870e+00
-3.27027678e-01 3.43050927e-01 -5.50071038e-02 -8.67599487e-01
4.25711066e-01 4.47987616e-01 1.12138212e+00 -4.95300889e-01
3.34069818e-01 2.22674876e-01 1.84917897e-01 -8.28224242e-01
-1.53164834e-01 -4.55469996e-01 -1.58527177e-02 2.67116666e-01
2.43621767e-01 1.01605080e-01 -1.08399659e-01 4.82972004e-02
1.44053471e+00 -3.71448874e-01 3.75054121e-01 -2.22446099e-02
2.96953648e-01 7.94635713e-02 9.42568853e-02 8.37897301e-01
-4.45621181e-03 6.61718369e-01 7.55311608e-01 -6.51831388e-01
-7.09465325e-01 -1.09400427e+00 -3.04070443e-01 1.16208482e+00
-1.91621646e-01 -1.90947607e-01 -1.83939431e-02 -8.08377206e-01
2.64372915e-01 -3.89458947e-02 -1.02595091e+00 -1.24639109e-01
-3.90838444e-01 -1.15373647e+00 4.46349412e-01 5.77758849e-01
-2.44174808e-01 -8.11577976e-01 -2.52199411e-01 3.10404658e-01
-1.51357040e-01 -9.09548879e-01 -1.07468292e-01 3.61741871e-01
-1.09243333e+00 -1.18980944e+00 -6.96887076e-01 -6.41236782e-01
5.36012590e-01 -1.11353688e-01 1.16751552e+00 8.55920166e-02
-5.43297708e-01 5.86681068e-01 5.64555544e-03 -3.09239358e-01
-4.79361266e-01 3.04039180e-01 -1.50525123e-01 4.77164015e-02
3.07007551e-01 -2.53021628e-01 -8.79603446e-01 1.59491569e-01
-9.92926300e-01 -9.55356434e-02 8.26500595e-01 1.26980841e+00
6.16661489e-01 -5.27262986e-01 7.90165544e-01 -1.08471251e+00
7.96025038e-01 -7.22400010e-01 -2.51233608e-01 3.16523135e-01
-3.20562810e-01 8.50339830e-02 2.30651274e-01 -4.47206795e-01
-7.45785594e-01 3.05003345e-01 -2.28718985e-02 -2.68301815e-01
-1.95085213e-01 1.02366507e+00 2.42265463e-01 -1.69772610e-01
6.42111480e-01 -2.60900021e-01 2.86373913e-01 -4.44476426e-01
2.85273254e-01 5.25906444e-01 3.92109871e-01 -4.72979575e-01
7.83590600e-02 6.03971064e-01 7.25250840e-01 -5.72919011e-01
-6.41562462e-01 -4.41039622e-01 -5.45970321e-01 7.72942370e-03
9.27210033e-01 -8.15536737e-01 -1.04598129e+00 2.90022105e-01
-9.54048872e-01 -2.74577677e-01 -1.17149547e-01 7.24632978e-01
-5.41780889e-01 1.75485343e-01 -8.37929308e-01 -2.06911519e-01
-1.81721866e-01 -1.28273022e+00 1.40795946e+00 -2.14811862e-01
-4.95237380e-01 -1.49493647e+00 5.95350325e-01 4.37282920e-01
2.91380346e-01 4.94873047e-01 1.28535187e+00 -6.20420575e-01
-5.13373911e-01 -4.14148599e-01 -1.76174611e-01 -3.01833097e-02
3.67021173e-01 -8.27740952e-02 -9.49495733e-01 -2.79503495e-01
-4.99691069e-01 -7.55006671e-01 1.05808139e+00 5.28265238e-01
1.23760557e+00 6.90043345e-02 -7.28804767e-01 3.46539587e-01
1.10962892e+00 -4.20437343e-02 4.24157411e-01 7.43965358e-02
6.77694917e-01 6.99496567e-01 2.38310531e-01 3.78851324e-01
4.14670199e-01 5.35292447e-01 5.66860974e-01 -7.79051483e-01
2.29124408e-02 1.74878389e-01 -1.24792390e-01 4.53235269e-01
-1.97393283e-01 -3.69440168e-01 -1.15646410e+00 4.49302703e-01
-2.11153960e+00 -6.86996639e-01 -6.89474866e-02 2.12675786e+00
9.30758178e-01 -3.06030482e-01 1.16676800e-01 -4.65133965e-01
2.11601898e-01 -2.06610307e-01 -5.42038381e-01 -2.69545048e-01
-1.52874991e-01 2.46373177e-01 3.77615899e-01 4.93445665e-01
-1.34831131e+00 5.63250244e-01 7.98356342e+00 4.86432552e-01
-1.27750957e+00 4.02231216e-02 8.38141143e-01 -4.79172021e-01
-3.04617465e-01 -4.60343391e-01 -4.10244405e-01 1.04492337e-01
9.98816729e-01 3.86637479e-01 1.12594455e-01 1.84128836e-01
2.21072793e-01 8.36590829e-04 -1.73222530e+00 9.21890438e-01
-2.66690552e-01 -1.62111425e+00 1.98271230e-01 4.01042402e-01
6.76891804e-01 3.15019369e-01 4.35078889e-01 -1.34642376e-02
5.96430063e-01 -1.50206542e+00 -2.43656650e-01 6.80317581e-01
7.66662419e-01 -3.89460474e-01 8.56767535e-01 9.19026230e-03
-5.05885482e-01 -2.02904344e-01 -2.35840008e-02 3.16866815e-01
-3.54210176e-02 3.58856201e-01 -9.93608117e-01 5.02794743e-01
5.24434805e-01 7.95570433e-01 -6.56768858e-01 1.07753754e+00
4.38488483e-01 3.32154781e-01 -2.19777003e-01 1.19953275e-01
3.50344002e-01 3.00062567e-01 1.53690293e-01 1.30072439e+00
2.11440191e-01 1.75413918e-02 2.05716908e-01 2.38441974e-01
-1.45413578e-01 2.83130646e-01 -8.03223670e-01 -8.87063816e-02
-3.50747332e-02 1.27660525e+00 -4.05451089e-01 -1.36581421e-01
-5.77221692e-01 4.15215015e-01 4.75197047e-01 2.65490860e-01
-6.08052850e-01 3.31162661e-01 5.45041621e-01 9.02155973e-03
3.48494090e-02 1.20183058e-01 -3.63177210e-01 -1.10368049e+00
-4.12755102e-01 -1.08066356e+00 9.58631694e-01 -4.47647810e-01
-1.76583230e+00 4.57552612e-01 -9.50374920e-03 -1.38668334e+00
-1.74404070e-01 -8.72520208e-01 -1.36815190e-01 6.66652858e-01
-1.17381477e+00 -1.45804584e+00 -9.60041210e-02 6.78215027e-01
-8.38477612e-02 -2.58600801e-01 1.14371765e+00 4.13927615e-01
-3.77332300e-01 5.25779843e-01 2.04036176e-01 -1.04431704e-01
1.08377755e+00 -1.26889396e+00 -3.48842144e-01 -1.18732132e-01
-2.21629292e-01 6.76660836e-01 4.89384830e-01 -4.72763389e-01
-1.54309916e+00 -8.57677460e-01 9.30724442e-01 -8.48557770e-01
9.44952071e-01 -1.44996308e-02 -6.63374007e-01 7.92144775e-01
4.71904606e-01 2.47109979e-02 1.40766287e+00 4.74711955e-01
-2.78083652e-01 -2.39114746e-01 -9.26534891e-01 6.27542973e-01
8.44503045e-01 -3.64792645e-01 -2.76154160e-01 6.69177115e-01
4.03795391e-01 -3.96075070e-01 -1.28656840e+00 7.66480327e-01
6.39946043e-01 -6.42899632e-01 1.09705591e+00 -1.36214614e+00
7.05961525e-01 3.73445861e-02 -8.16213489e-02 -1.21785522e+00
-3.02926421e-01 -5.35071969e-01 -2.12631732e-01 5.30242562e-01
7.81078041e-01 -6.22798502e-01 4.75622177e-01 8.13668013e-01
2.25529801e-02 -1.16675365e+00 -1.02591407e+00 -1.69842705e-01
3.80167663e-01 -1.55809015e-01 3.81834328e-01 1.08254194e+00
3.91429484e-01 4.18210596e-01 -4.27495956e-01 7.31690824e-02
4.51597363e-01 1.57412440e-01 5.47528386e-01 -1.24922705e+00
-5.28634369e-01 -6.21281862e-01 -3.85786593e-01 -5.81345737e-01
-8.00847188e-02 -9.69105959e-01 -4.33012694e-01 -1.53445065e+00
6.15484118e-01 -4.26395774e-01 -8.16924214e-01 7.79922247e-01
-1.17546842e-01 3.09618741e-01 1.11353826e-02 1.03239007e-01
-7.15501070e-01 2.12731749e-01 1.46283913e+00 -4.52522248e-01
-3.38079900e-01 -2.13646278e-01 -8.59921873e-01 5.13951957e-01
6.80838108e-01 -3.74073565e-01 -4.68920738e-01 -4.22815323e-01
4.73441601e-01 3.87783229e-01 5.43091118e-01 -5.61676204e-01
1.70956805e-01 -2.96450794e-01 6.03453934e-01 -4.38477427e-01
3.98810536e-01 -8.29545438e-01 2.93529481e-01 5.56259930e-01
-6.76551044e-01 -1.14039056e-01 3.82461518e-01 5.48972547e-01
-1.97045475e-01 2.54593611e-01 5.58495522e-01 -9.63663682e-02
-2.56768644e-01 4.92607832e-01 -4.06737715e-01 -1.60170659e-01
9.26906824e-01 3.39936763e-01 -5.53727448e-01 -3.16759080e-01
-1.36448181e+00 5.17080843e-01 3.67581025e-02 3.83691251e-01
5.50341368e-01 -1.33681250e+00 -8.26152086e-01 -2.59935074e-02
2.69703329e-01 -2.79003382e-01 4.88154352e-01 1.25762784e+00
-4.37461734e-01 5.65001905e-01 -1.18184201e-01 -8.98827434e-01
-1.44278538e+00 6.22159660e-01 6.26858294e-01 -5.79325974e-01
-2.70215899e-01 4.93193060e-01 3.50537419e-01 -5.16243517e-01
2.19856128e-01 -3.67261291e-01 -2.71036059e-01 2.43058130e-01
3.28231722e-01 1.00405231e-01 1.76304922e-01 -3.27685446e-01
-4.41476017e-01 5.46501637e-01 -3.51899385e-01 -1.97118986e-02
1.46833706e+00 3.42444442e-02 -3.36247534e-01 4.15537179e-01
1.36339617e+00 -3.85839462e-01 -8.74253333e-01 -1.30936161e-01
3.69241424e-02 -3.88505980e-02 3.38523909e-02 -1.22093630e+00
-1.03456569e+00 9.64268327e-01 7.51887321e-01 3.12547423e-02
1.12838745e+00 3.74008745e-01 6.20984435e-01 6.93395078e-01
1.18690416e-01 -6.70986652e-01 2.22604841e-01 2.23112285e-01
8.55700612e-01 -1.58687353e+00 3.29894796e-02 -2.23496631e-01
-7.49166250e-01 1.12812853e+00 3.16102833e-01 2.09585056e-01
8.79462063e-01 3.48921388e-01 3.95559043e-01 -5.91491163e-01
-1.16125309e+00 -2.36131802e-01 5.50676882e-01 4.02245373e-01
7.30100334e-01 1.91762000e-01 -2.80548871e-01 3.55915785e-01
4.67777371e-01 4.13863629e-01 2.68239021e-01 8.49783421e-01
-2.70650163e-02 -1.40558290e+00 -1.93163007e-01 6.27123535e-01
-5.82017064e-01 -1.16851851e-01 -6.25774384e-01 9.91626441e-01
2.04593942e-01 6.90767109e-01 -2.56304771e-01 -3.85816723e-01
2.17726380e-01 -9.35479812e-03 6.92661285e-01 -5.47103345e-01
-7.38184750e-01 4.28751081e-01 1.60120219e-01 -6.83652639e-01
-5.71089447e-01 -7.98037231e-01 -1.15372431e+00 -1.34923175e-01
1.50435967e-02 -4.09004241e-01 2.58377463e-01 9.06907499e-01
6.66988611e-01 5.84420979e-01 3.95856112e-01 -6.48239970e-01
-5.86678267e-01 -6.29582524e-01 -3.42444330e-01 4.20846611e-01
7.43210971e-01 -6.66721046e-01 6.58911169e-02 2.05861665e-02]
|
[15.072677612304688, -2.631326675415039]
|
bee3c20e-ed0f-463b-b624-de4102da1183
|
multi-view-azimuth-stereo-via-tangent-space
|
2303.16447
| null |
https://arxiv.org/abs/2303.16447v1
|
https://arxiv.org/pdf/2303.16447v1.pdf
|
Multi-View Azimuth Stereo via Tangent Space Consistency
|
We present a method for 3D reconstruction only using calibrated multi-view surface azimuth maps. Our method, multi-view azimuth stereo, is effective for textureless or specular surfaces, which are difficult for conventional multi-view stereo methods. We introduce the concept of tangent space consistency: Multi-view azimuth observations of a surface point should be lifted to the same tangent space. Leveraging this consistency, we recover the shape by optimizing a neural implicit surface representation. Our method harnesses the robust azimuth estimation capabilities of photometric stereo methods or polarization imaging while bypassing potentially complex zenith angle estimation. Experiments using azimuth maps from various sources validate the accurate shape recovery with our method, even without zenith angles.
|
['Yasuyuki Matsushita', 'Fumio Okura', 'Hiroaki Santo', 'Xu Cao']
|
2023-03-29
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Cao_Multi-View_Azimuth_Stereo_via_Tangent_Space_Consistency_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Cao_Multi-View_Azimuth_Stereo_via_Tangent_Space_Consistency_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['3d-shape-reconstruction']
|
['computer-vision']
|
[ 4.39956158e-01 -1.24691248e-01 2.24258319e-01 -6.19601786e-01
-9.72898483e-01 -1.03054893e+00 8.27558398e-01 -7.59638309e-01
1.47556722e-01 5.70441484e-01 3.37544113e-01 -2.58938462e-01
-1.94093749e-01 -6.09798431e-01 -6.09834552e-01 -8.63333821e-01
5.64124465e-01 6.64832711e-01 -9.81232896e-02 -3.86408240e-01
4.47088867e-01 6.57282591e-01 -1.43672466e+00 1.94432825e-01
4.84305322e-01 8.73158514e-01 -2.08935905e-02 5.62377155e-01
2.43926495e-01 9.80673432e-02 7.86517039e-02 -1.99483559e-01
7.00340629e-01 3.40869635e-01 -5.44873536e-01 6.24555983e-02
1.40284860e+00 -6.27651215e-01 2.16884054e-02 9.26547408e-01
3.01055640e-01 -1.19856238e-01 5.89428961e-01 -5.72228014e-01
-2.93270022e-01 -7.38917768e-01 -6.60476446e-01 -1.65262416e-01
5.99141479e-01 -3.05797726e-01 1.07236123e+00 -1.14727342e+00
7.96832442e-01 9.71549749e-01 1.00534260e+00 1.48369521e-01
-1.21426022e+00 -1.68637976e-01 -3.79967532e-04 -1.54105082e-01
-1.29590464e+00 -9.30158496e-01 9.22980487e-01 -4.91327882e-01
8.24693918e-01 5.39872169e-01 3.80966663e-01 1.04048014e+00
3.20276320e-01 1.17058186e-02 1.64146125e+00 -3.44119042e-01
-2.20285714e-01 2.10250467e-02 -1.11789159e-01 4.85971302e-01
3.47658157e-01 2.60634214e-01 -7.77566850e-01 -2.60749876e-01
1.04465735e+00 -1.22577827e-02 -5.17188430e-01 -1.04433274e+00
-1.03615475e+00 5.89464545e-01 -8.74050781e-02 -2.86274016e-01
-3.07003617e-01 -2.19986945e-01 -9.67059508e-02 2.61971086e-01
9.15512860e-01 5.25808632e-01 -5.58094561e-01 2.66702235e-01
-5.34668446e-01 1.23867437e-01 7.83779323e-01 8.58989596e-01
9.84692991e-01 2.19794810e-01 8.12952936e-01 7.66197026e-01
4.28605407e-01 1.41061974e+00 -1.84312731e-01 -1.47188759e+00
4.76541162e-01 7.52611160e-02 4.96179014e-01 -7.99516678e-01
-4.09798354e-01 -1.99148864e-01 -6.02178693e-01 5.97567260e-01
3.05660427e-01 -2.79949494e-02 -6.68663681e-01 1.47457218e+00
6.81598961e-01 -2.01290518e-01 2.82828122e-01 9.11550760e-01
3.33922923e-01 2.67413318e-01 -9.66083646e-01 -1.22930087e-01
1.21342504e+00 -6.03836834e-01 -2.77976006e-01 -4.53381628e-01
3.06149870e-02 -1.14590406e+00 6.39086545e-01 6.02346241e-01
-9.43361700e-01 -3.49973552e-02 -8.99392724e-01 3.52691226e-02
-9.45383757e-02 -1.70297250e-01 6.19553685e-01 5.54982841e-01
-1.25628710e+00 5.01093268e-01 -6.84885621e-01 -2.34331325e-01
-2.60304451e-01 3.01438272e-02 -7.90327072e-01 -1.59452394e-01
-6.26073658e-01 8.18729520e-01 -4.60417360e-01 8.27574059e-02
-3.36731434e-01 -7.42294252e-01 -9.64774191e-01 -3.46428394e-01
1.72589004e-01 -9.76218581e-01 1.18125486e+00 -7.70389378e-01
-1.49176180e+00 1.11722231e+00 -7.47083187e-01 1.19869702e-01
3.45844567e-01 -6.01584077e-01 -6.18426263e-01 3.64464104e-01
1.82212129e-01 -5.79809509e-02 8.18383932e-01 -1.77044713e+00
-2.55193319e-02 -8.66817951e-01 -1.68473452e-01 8.29683542e-01
3.08083773e-01 -3.77593130e-01 -3.30949605e-01 2.46344618e-02
9.78403986e-01 -9.84553933e-01 -3.85593884e-02 -4.00728956e-02
-4.27357525e-01 7.93529749e-01 8.32877398e-01 -7.55913913e-01
2.36678496e-01 -1.93361557e+00 -5.17321378e-02 4.44536418e-01
8.15404505e-02 -4.18334216e-01 4.32507992e-02 5.82297027e-01
1.23161664e-02 -3.18712473e-01 2.73501761e-02 -4.41714078e-01
-3.08885843e-01 8.13662931e-02 -7.46399760e-01 9.73618746e-01
-2.87049681e-01 3.67914110e-01 -4.37537551e-01 1.50996715e-01
4.78540152e-01 6.69232905e-01 -5.83611429e-01 1.42090291e-01
1.45598620e-01 9.44608867e-01 -3.34507287e-01 8.11309934e-01
1.18115997e+00 -4.71533865e-01 4.17460293e-01 -5.47200024e-01
-6.04987085e-01 5.19081593e-01 -1.20094252e+00 1.65396011e+00
-8.63635778e-01 5.56174278e-01 6.75651550e-01 -4.40611988e-01
9.23672855e-01 2.58643389e-01 4.65111107e-01 -9.93414164e-01
-5.88730633e-01 4.59936649e-01 -6.88564777e-01 -2.10536882e-01
6.60386622e-01 -5.41806221e-01 2.88790792e-01 1.59702748e-01
-3.87520254e-01 -7.19480336e-01 -8.12890291e-01 -2.38666341e-01
4.54686344e-01 5.40137053e-01 3.44441950e-01 -4.17950422e-01
2.81756490e-01 -1.89777613e-01 3.60263735e-01 6.36988938e-01
3.52003217e-01 1.09283340e+00 -8.52985457e-02 -8.30380201e-01
-1.48808551e+00 -1.46494019e+00 -7.05996752e-01 5.93461692e-01
2.99799919e-01 -5.83777651e-02 -1.93229914e-01 -1.87412918e-01
1.04097970e-01 4.26282495e-01 -3.77166003e-01 5.45187116e-01
-6.51063263e-01 -7.72245407e-01 1.21342586e-02 9.32141989e-02
4.12833780e-01 -2.29031414e-01 -4.68603879e-01 -2.55562931e-01
-2.92047381e-01 -1.31276929e+00 -1.92668378e-01 -6.81429207e-02
-1.05515206e+00 -1.11655533e+00 -6.60803199e-01 -1.11942448e-01
5.77920020e-01 9.31754708e-01 1.32411206e+00 -5.63157141e-01
1.71007887e-01 9.87504423e-01 1.13654651e-01 -1.04728445e-01
-1.83529612e-02 -5.01625180e-01 3.22442055e-01 1.34313807e-01
4.04866524e-02 -1.02419937e+00 -7.04682112e-01 6.68886840e-01
-2.86621511e-01 3.87657434e-01 1.65691346e-01 7.93255389e-01
9.14426148e-01 -7.42212892e-01 -8.44173506e-02 -9.67643678e-01
-2.66371042e-01 -3.38101834e-01 -1.06684184e+00 7.50543848e-02
-8.50202680e-01 4.21163999e-02 4.52662915e-01 4.27834809e-01
-1.55398607e+00 -1.29163191e-01 -1.08743884e-01 -2.99395800e-01
-1.99088797e-01 1.41910419e-01 -1.21448651e-01 -6.01506591e-01
6.38506711e-01 2.93457270e-01 9.96462777e-02 -7.73715794e-01
1.83469668e-01 4.02440012e-01 5.74648499e-01 -5.38226128e-01
7.37864435e-01 1.51509511e+00 6.25735283e-01 -1.25864077e+00
-1.09895766e+00 -7.50263035e-01 -7.22627342e-01 -8.23077261e-02
7.70632684e-01 -1.25410080e+00 -5.47740936e-01 4.50042397e-01
-1.07359028e+00 -2.36181945e-01 1.76590130e-01 5.89450181e-01
-7.14413702e-01 6.47389233e-01 -1.49022520e-01 -8.64317834e-01
-1.92560241e-01 -7.74038255e-01 1.95900524e+00 -3.34196538e-02
5.80155775e-02 -1.34822190e+00 4.80869204e-01 6.18670106e-01
5.01607537e-01 2.74351954e-01 4.81487721e-01 9.45577174e-02
-9.06389594e-01 6.77516460e-02 -2.79020488e-01 9.63504799e-03
2.46287763e-01 -2.21796304e-01 -1.52807164e+00 -3.31892759e-01
6.06039584e-01 -3.32031369e-01 7.55780220e-01 6.86940432e-01
7.66017854e-01 -2.63747633e-01 -2.33367041e-01 1.65941811e+00
1.88303363e+00 -1.83299392e-01 5.09680688e-01 6.22234404e-01
8.47845495e-01 6.48893476e-01 3.67723644e-01 5.52802503e-01
5.05397320e-01 1.10141444e+00 6.64918661e-01 -1.91258967e-01
1.36883616e-01 -1.06117725e-01 5.57443127e-02 8.19557428e-01
-3.70299220e-01 1.86409559e-02 -8.63900781e-01 2.60071009e-01
-1.26897442e+00 -9.39777076e-01 -2.79991925e-01 2.63774061e+00
3.39074045e-01 -3.78711402e-01 -6.81844234e-01 -4.62461740e-01
1.88621879e-01 6.03774369e-01 -5.12575924e-01 -3.70967418e-01
-6.95209801e-01 -1.93824992e-02 1.01381540e+00 1.20960104e+00
-9.64148700e-01 7.45938897e-01 6.97642422e+00 1.79836657e-02
-1.44994664e+00 -7.22523257e-02 1.53003752e-01 -8.46289936e-03
-1.10968077e+00 1.93472564e-01 -9.68574584e-01 -2.44442284e-01
6.61797047e-01 3.97601724e-01 6.86668217e-01 5.40561914e-01
1.55631170e-01 -1.36723772e-01 -7.86748409e-01 1.11761689e+00
3.73615712e-01 -1.48107350e+00 -7.62578845e-03 9.72573608e-02
9.20146585e-01 6.60463035e-01 1.75103888e-01 -4.99032885e-01
3.09600115e-01 -7.62595057e-01 5.16172528e-01 7.17918754e-01
1.27731228e+00 -1.60719514e-01 2.21859083e-01 1.67503923e-01
-9.82823849e-01 4.72171485e-01 -1.25667036e-01 -8.95586796e-03
6.39975727e-01 6.48371637e-01 -8.09636593e-01 1.12437892e+00
7.59510398e-01 7.41580129e-01 -1.08259417e-01 5.47482491e-01
-2.02679634e-01 9.43083912e-02 -6.49733663e-01 8.20328295e-01
5.39794639e-02 -8.11382830e-01 9.09575880e-01 5.30045509e-01
6.32346749e-01 -3.99225727e-02 -2.69070387e-01 4.17637199e-01
4.98398900e-01 -2.73298144e-01 -1.38503647e+00 5.99542260e-01
3.32050681e-01 1.27661228e+00 -1.65422354e-02 -7.44545311e-02
-6.90322042e-01 1.09233546e+00 6.02457151e-02 7.27452636e-01
-1.12707324e-01 3.70227844e-01 7.71109521e-01 2.09334388e-01
2.81546801e-01 -5.35895824e-01 -7.13619292e-01 -1.57263434e+00
4.19065893e-01 -7.39168584e-01 6.41776025e-02 -1.39668643e+00
-1.12647545e+00 3.43000501e-01 1.05647780e-01 -1.33051813e+00
-4.09846514e-01 -1.03537977e+00 -4.25129980e-01 1.40591323e+00
-1.78530025e+00 -1.37784266e+00 -4.29151267e-01 4.77594227e-01
8.21640193e-02 9.01654884e-02 1.21405625e+00 -1.27494246e-01
2.80977815e-01 -6.48623258e-02 6.99208319e-01 -6.58590794e-01
9.78360474e-01 -1.49127579e+00 6.07843280e-01 5.86217940e-01
1.54223904e-01 8.02470684e-01 7.37572432e-01 -5.45810878e-01
-1.89721358e+00 -5.91791213e-01 5.84622979e-01 -8.47532928e-01
3.90350968e-01 -2.97391385e-01 -7.58782685e-01 9.97303486e-01
2.29971766e-01 -2.60061193e-02 6.88039422e-01 4.79865432e-01
-8.41848254e-01 -1.53592244e-01 -8.93665016e-01 4.27603126e-01
9.31241095e-01 -1.04225254e+00 -4.19878781e-01 4.46787328e-01
2.58466989e-01 -7.84312725e-01 -7.77361989e-01 7.56395161e-01
1.08212340e+00 -1.36437833e+00 1.29333961e+00 -3.58373135e-01
4.41268794e-02 -3.01432431e-01 -8.38891089e-01 -1.48265493e+00
-1.59486279e-01 -7.30280399e-01 2.46650293e-01 6.45215273e-01
2.15269208e-01 -1.13333440e+00 7.72012234e-01 2.97238082e-01
-4.53906655e-01 -2.79673010e-01 -1.09510899e+00 -8.52025270e-01
-2.49897558e-02 -1.61609069e-01 4.19006974e-01 1.15898430e+00
-5.50625682e-01 3.42646182e-01 -5.48415542e-01 8.99125338e-01
1.15601456e+00 7.98033834e-01 9.61746871e-01 -1.34472656e+00
-5.97627461e-01 5.09967245e-02 1.52800843e-01 -1.34515142e+00
3.06606218e-02 -5.02834082e-01 -9.21975747e-02 -1.26499236e+00
8.80664587e-02 -5.41766346e-01 2.67059207e-01 1.41347134e-02
2.77522713e-01 3.89504373e-01 -3.26898277e-01 5.54068685e-01
-1.24344110e-01 4.47753012e-01 1.25692701e+00 3.29283446e-01
1.32511750e-01 -9.08885002e-02 -5.09149253e-01 1.21701622e+00
5.03369510e-01 -6.69813082e-02 -2.24632502e-01 -9.79532301e-01
7.87357688e-01 3.91034931e-01 6.18772686e-01 -5.67584991e-01
-1.81170274e-02 -3.14677835e-01 3.16915810e-01 -7.11228013e-01
1.04899526e+00 -8.60621810e-01 6.24673188e-01 -1.90784663e-01
2.98743457e-01 2.04913661e-01 1.22378342e-01 5.90938330e-01
-1.70836285e-01 1.57722890e-01 7.27252662e-01 -1.99611768e-01
-7.76162744e-01 1.77055433e-01 2.02094302e-01 1.03920072e-01
2.67646790e-01 -3.09143186e-01 -5.30955672e-01 -5.63598335e-01
-4.41258520e-01 -1.65932819e-01 1.28550351e+00 3.36726219e-03
5.69436669e-01 -1.27182758e+00 -5.22255838e-01 6.86238289e-01
3.41663480e-01 6.60307929e-02 3.10467154e-01 1.04754972e+00
-7.73283005e-01 5.33719540e-01 -6.16890639e-02 -9.26554620e-01
-1.23739839e+00 6.61502453e-03 7.26021588e-01 5.95906451e-02
-8.20865512e-01 5.39961696e-01 8.94247711e-01 -9.81948972e-01
-8.40224564e-01 1.91589653e-01 5.10589890e-02 -3.50313276e-01
3.83990616e-01 2.44689122e-01 3.33255708e-01 -7.21976280e-01
-4.40073758e-01 1.20444155e+00 2.11021483e-01 -5.75687408e-01
1.47538602e+00 -5.51266253e-01 -1.13604859e-01 5.56328952e-01
1.18880343e+00 9.70419288e-01 -1.55440974e+00 -1.13389529e-02
-5.40121198e-01 -6.92945361e-01 5.29454052e-02 -5.34176290e-01
-5.85130751e-01 9.04394388e-01 2.77757496e-01 -1.12723343e-01
8.25292408e-01 -1.12732358e-01 3.16181660e-01 6.89642072e-01
5.86689770e-01 -7.06509054e-01 -3.69862914e-01 7.87684858e-01
8.86335373e-01 -1.31923926e+00 2.37697467e-01 -6.43956482e-01
-3.68903875e-01 1.31362426e+00 2.68256605e-01 -3.70935500e-02
5.67508936e-01 1.99194804e-01 3.89699936e-01 -6.03435695e-01
-4.49347228e-01 3.63438815e-01 4.58833486e-01 5.81766665e-01
2.88620114e-01 -6.01831973e-02 3.01662952e-01 -3.29102844e-01
-1.80785835e-01 -6.03486598e-01 5.93657613e-01 6.77659810e-01
-4.04228687e-01 -8.88343513e-01 -7.05904126e-01 7.90844932e-02
-3.47065806e-01 -3.11261326e-01 -3.01956892e-01 7.10432708e-01
-5.30762255e-01 4.80313390e-01 2.91748554e-01 -1.16639145e-01
2.18270689e-01 1.09992057e-01 7.34587014e-01 -4.45561409e-01
2.43959919e-01 2.30006665e-01 6.74142063e-01 -9.11064744e-01
-5.62006593e-01 -1.07883418e+00 -6.02107346e-01 -2.15100989e-01
-2.57041544e-01 -1.31856948e-01 9.04071271e-01 8.64580274e-01
4.30807143e-01 -3.96691412e-01 9.80768144e-01 -9.72734809e-01
-5.93357861e-01 -4.79213595e-01 -7.75233328e-01 3.31205249e-01
7.89217412e-01 -8.08968723e-01 -8.63191664e-01 -5.89181930e-02]
|
[9.57697868347168, -2.9189815521240234]
|
765b3839-d263-4eb0-aad0-a881a3b5e38f
|
a-3d-probabilistic-deep-learning-system-for
|
1902.03233
| null |
https://arxiv.org/abs/1902.03233v3
|
https://arxiv.org/pdf/1902.03233v3.pdf
|
A 3D Probabilistic Deep Learning System for Detection and Diagnosis of Lung Cancer Using Low-Dose CT Scans
|
We introduce a new computer aided detection and diagnosis system for lung cancer screening with low-dose CT scans that produces meaningful probability assessments. Our system is based entirely on 3D convolutional neural networks and achieves state-of-the-art performance for both lung nodule detection and malignancy classification tasks on the publicly available LUNA16 and Kaggle Data Science Bowl challenges. While nodule detection systems are typically designed and optimized on their own, we find that it is important to consider the coupling between detection and diagnosis components. Exploiting this coupling allows us to develop an end-to-end system that has higher and more robust performance and eliminates the need for a nodule detection false positive reduction stage. Furthermore, we characterize model uncertainty in our deep learning systems, a first for lung CT analysis, and show that we can use this to provide well-calibrated classification probabilities for both nodule detection and patient malignancy diagnosis. These calibrated probabilities informed by model uncertainty can be used for subsequent risk-based decision making towards diagnostic interventions or disease treatments, as we demonstrate using a probability-based patient referral strategy to further improve our results.
|
['Rebecca L. Russell', 'Onur Ozdemir', 'Andrew A. Berlin']
|
2019-02-08
| null | null | null | null |
['probabilistic-deep-learning', 'lung-nodule-detection']
|
['computer-vision', 'medical']
|
[ 5.33435643e-02 6.37670338e-01 -7.18153298e-01 -4.86246407e-01
-1.40897763e+00 -4.24088389e-01 3.85039926e-01 2.24927932e-01
-2.27692381e-01 1.50059626e-01 3.14380974e-01 -9.73671496e-01
-3.81384194e-01 -7.87913203e-01 -7.86437094e-01 -5.73190451e-01
-1.58523083e-01 1.22902107e+00 6.33822739e-01 4.57840025e-01
-6.27423525e-01 4.81537521e-01 -1.07385838e+00 4.39697742e-01
4.61080551e-01 1.06781423e+00 1.07948042e-01 1.09857631e+00
3.17874819e-01 1.02011073e+00 1.30463049e-01 -2.92697102e-01
3.62423927e-01 -2.77354658e-01 -8.88289690e-01 -2.13580146e-01
3.51552099e-01 -7.78555155e-01 -4.57243711e-01 6.64743841e-01
6.36347473e-01 -5.69306016e-01 1.09655213e+00 -7.16596007e-01
-1.79173574e-01 9.63143587e-01 -2.75958955e-01 3.31141204e-01
-2.24751323e-01 4.40988839e-01 7.14554429e-01 -3.98289561e-01
2.30355054e-01 8.74041915e-01 1.11990321e+00 7.74069726e-01
-8.68282378e-01 -3.83115113e-01 -4.80671972e-01 -3.01746130e-01
-9.57882702e-01 7.04457536e-02 -2.37219438e-01 -6.52990639e-01
6.66851044e-01 3.28521818e-01 8.73281360e-01 1.07491946e+00
6.55985057e-01 8.12961280e-01 5.85623085e-01 -3.81448448e-01
1.70676321e-01 8.12742189e-02 4.24806848e-02 1.15963447e+00
8.39849412e-01 6.14373744e-01 9.07140300e-02 -3.77589136e-01
1.05986881e+00 8.83350149e-02 -1.19483203e-01 -6.75281525e-01
-1.33041680e+00 8.45773697e-01 8.90099406e-01 -8.57463852e-02
-3.40476125e-01 7.77682781e-01 3.38056386e-01 -4.85124230e-01
8.12741965e-02 3.27065468e-01 -1.94067419e-01 1.98024139e-01
-9.92438197e-01 2.36073062e-01 1.06452978e+00 6.17816925e-01
-1.58999022e-02 -5.92530012e-01 -1.10286438e+00 3.95682245e-01
6.84481382e-01 5.70386648e-01 2.64980912e-01 -8.43685329e-01
-1.04152195e-01 3.27397615e-01 1.12329662e-01 -1.55050725e-01
-9.16583180e-01 -7.16342568e-01 -7.10826457e-01 1.27665251e-01
4.04851556e-01 -2.38618046e-01 -1.61822331e+00 1.37626159e+00
2.26922020e-01 2.76005685e-01 -1.94037333e-02 6.71766758e-01
9.79911566e-01 -4.96585220e-02 3.93012851e-01 3.19599211e-01
1.73727047e+00 -7.70691216e-01 -1.41560704e-01 -2.68762439e-01
1.12886047e+00 -3.33502918e-01 6.48304820e-01 -5.44093028e-02
-1.16804612e+00 -1.14691034e-02 -9.05974984e-01 6.82376651e-03
3.46774817e-01 4.27707493e-01 8.76575291e-01 1.05059409e+00
-9.85594869e-01 5.29785454e-01 -1.69369626e+00 -5.59432149e-01
7.95085371e-01 5.30423522e-01 1.03777923e-01 -1.41455501e-01
-9.99732912e-01 1.21621442e+00 3.39952707e-01 -2.20082849e-01
-1.45835996e+00 -1.34035087e+00 -7.30187535e-01 2.68644124e-01
3.89518499e-01 -1.27244890e+00 2.25972486e+00 -1.81110054e-01
-9.60988462e-01 7.96920955e-01 1.77909270e-01 -9.41642165e-01
8.28728735e-01 2.51464784e-01 1.14576980e-01 1.49662539e-01
3.16388249e-01 8.70917261e-01 4.00863588e-01 -7.42403150e-01
-9.43903804e-01 -3.28884833e-02 -3.87754202e-01 1.63901702e-01
1.31852031e-02 -1.94774479e-01 -7.44498014e-01 -1.60050482e-01
7.42258579e-02 -1.09531999e+00 -6.05164587e-01 4.91677463e-01
-4.71977144e-01 4.75550219e-02 2.41421878e-01 -4.74558294e-01
8.81565452e-01 -1.61814976e+00 -4.11230654e-01 4.03072834e-01
4.95422184e-01 2.56660376e-02 3.04639697e-01 -4.69279200e-01
-4.82922755e-02 4.13725466e-01 -2.14918554e-01 -8.69056210e-02
-1.06491096e-01 1.30297199e-01 4.51952845e-01 2.37092718e-01
5.70103645e-01 1.24130368e+00 -8.99893284e-01 -7.65326023e-01
5.30598819e-01 3.70885998e-01 -6.88488066e-01 1.83762938e-01
-2.26283237e-01 2.24136323e-01 -5.52403092e-01 7.73053765e-01
3.83309811e-01 -7.33231068e-01 1.56840160e-01 -3.52517068e-01
3.54214936e-01 3.77604246e-01 -7.49912202e-01 1.15073979e+00
-5.66139340e-01 6.71401173e-02 1.15357507e-02 -3.72597814e-01
7.84086138e-02 4.11547422e-01 7.61733532e-01 -1.56964362e-01
3.81890357e-01 3.70398879e-01 5.15604138e-01 -4.02919382e-01
-8.18498880e-02 -2.88213283e-01 1.31015629e-02 4.64813232e-01
1.02107273e-02 -5.46087801e-01 -9.91011113e-02 2.22650841e-01
1.74658477e+00 -2.57932484e-01 4.98665959e-01 -5.50753593e-01
-1.87999770e-01 4.79066402e-01 1.61985204e-01 1.18225038e+00
-3.53972226e-01 8.50222349e-01 6.06643617e-01 -2.61577249e-01
-1.06976330e+00 -1.43639112e+00 -7.11743116e-01 5.83088100e-01
-2.51526356e-01 2.09584266e-01 -4.19710666e-01 -8.86377513e-01
3.20014864e-01 7.24688292e-01 -9.47503388e-01 -3.76931995e-01
-2.83118188e-01 -1.26097918e+00 6.45617723e-01 1.10138738e+00
2.20312491e-01 -6.83780909e-01 -6.21895969e-01 2.12092549e-01
1.84735253e-01 -7.00350165e-01 -2.35233158e-01 8.24747086e-01
-8.63821030e-01 -1.42551613e+00 -1.02872717e+00 -6.62149131e-01
4.53894436e-01 1.13250725e-02 1.49830842e+00 -3.99003699e-02
-7.71307886e-01 6.60380900e-01 3.69359832e-03 -6.88827515e-01
-1.10843408e+00 2.66490012e-01 -1.75656214e-01 -8.75433028e-01
3.00343096e-01 9.14258212e-02 -9.82584536e-01 2.60688066e-01
-6.52840257e-01 1.15535513e-01 1.10615396e+00 8.99143040e-01
7.92612970e-01 -1.31255984e-01 2.37380620e-02 -1.09761357e+00
3.35571438e-01 -7.14272320e-01 -5.78495443e-01 3.19495648e-01
-6.11443758e-01 3.32798958e-01 -1.39320880e-01 -1.96993917e-01
-1.02661908e+00 4.33597535e-01 -2.40120724e-01 -3.16702247e-01
4.53209579e-02 4.84390825e-01 5.19073963e-01 -1.54383212e-01
1.21179497e+00 -3.46266329e-01 1.03100650e-02 -1.20889932e-01
2.16467321e-01 5.83312929e-01 6.15992904e-01 -2.82064497e-01
5.72934091e-01 5.16180515e-01 3.42941016e-01 1.54279685e-02
-1.26121652e+00 -5.32275438e-01 -3.83555979e-01 3.54641676e-02
1.15558577e+00 -1.19067824e+00 -7.01358080e-01 1.83511022e-02
-6.04141414e-01 -5.06263852e-01 -6.94275439e-01 7.26352692e-01
-4.44416225e-01 2.11409982e-02 -7.52132893e-01 -4.92044657e-01
-4.96863097e-01 -1.35863984e+00 1.38150430e+00 2.74940550e-01
-2.45570123e-01 -1.03362751e+00 -1.06516153e-01 9.40440595e-02
5.42187750e-01 -1.91403627e-02 1.09848964e+00 -7.33918726e-01
-8.92020226e-01 -5.81435323e-01 -6.08755052e-01 2.47299355e-02
1.79530457e-01 1.88297629e-02 -8.10834467e-01 -1.02018349e-01
-2.64545739e-01 -4.04950857e-01 1.19984186e+00 1.16244686e+00
1.47896135e+00 3.71651620e-01 -9.66250181e-01 6.64574802e-01
1.04239035e+00 -3.59271228e-01 2.62066096e-01 2.39470489e-02
7.24151194e-01 8.36635157e-02 3.01474452e-01 2.18108341e-01
3.96625102e-01 6.12408295e-02 7.78760552e-01 -3.60442638e-01
-5.58326483e-01 -1.85047761e-01 -2.93003142e-01 -4.84426692e-02
2.56832272e-01 -3.80609781e-01 -1.45967662e+00 6.65292501e-01
-1.60934818e+00 -5.58498263e-01 -2.91977376e-01 2.23283076e+00
8.94288898e-01 4.11884665e-01 -1.47111848e-01 -4.56194639e-01
4.56335872e-01 -3.85215372e-01 -7.33811200e-01 1.24402247e-01
5.38534522e-01 3.97736549e-01 1.22459841e+00 3.34977239e-01
-1.39704347e+00 3.30501795e-01 7.70624018e+00 8.25663447e-01
-8.30037415e-01 1.23229340e-01 1.14825118e+00 -1.83629483e-01
-4.03916776e-01 -2.84611881e-01 -7.85126150e-01 -1.91968605e-02
9.75753963e-01 8.45280662e-02 -1.87515602e-01 1.02061594e+00
-3.55673432e-02 -3.96616250e-01 -1.67418277e+00 5.11752248e-01
-4.40751612e-01 -1.63864148e+00 -2.05129161e-01 -2.71847751e-02
6.18385494e-01 6.12055182e-01 1.29760712e-01 5.16227245e-01
9.47296977e-01 -1.32888126e+00 2.24227950e-01 4.14782345e-01
9.24512506e-01 -3.68306011e-01 1.02587342e+00 2.28229985e-01
-8.64980161e-01 1.67792857e-01 -1.37294129e-01 6.44452751e-01
-1.02385163e-01 7.90385664e-01 -1.81387758e+00 3.15561831e-01
7.38417745e-01 3.16900462e-01 -8.10404480e-01 1.37825131e+00
-4.17591305e-03 9.83378410e-01 -7.95865238e-01 -7.82402679e-02
8.94219950e-02 7.51326561e-01 1.25980303e-01 1.11206257e+00
5.21218538e-01 -1.18129425e-01 1.65933594e-01 1.01636064e+00
-2.04097927e-01 -5.32685459e-01 -2.25028694e-01 9.84227881e-02
3.21644157e-01 1.38398504e+00 -8.58073711e-01 -3.45617503e-01
-2.62410551e-01 4.87505734e-01 6.14796281e-02 -3.35143685e-01
-1.05958271e+00 4.58606988e-01 1.81989029e-01 3.38099599e-01
2.62374610e-01 2.91620880e-01 -4.09977913e-01 -8.50375652e-01
-4.24602807e-01 -2.84038723e-01 6.44531786e-01 -5.51715612e-01
-1.53456855e+00 2.17069238e-01 2.28308246e-01 -1.04518211e+00
-7.17848599e-01 -1.08202052e+00 -6.77361369e-01 8.42305660e-01
-1.46161997e+00 -1.11801231e+00 -4.88973171e-01 1.31723747e-01
9.53911617e-03 1.36238232e-01 8.07285070e-01 1.07885018e-01
-2.82529533e-01 5.95369756e-01 6.58636838e-02 2.41846532e-01
6.68285191e-01 -1.58795178e+00 5.04758000e-01 5.49269855e-01
-3.83429587e-01 -1.40220493e-01 3.17118704e-01 -8.90325189e-01
-1.24596941e+00 -1.48121572e+00 3.29175621e-01 -9.81650233e-01
5.84744155e-01 1.44608393e-01 -4.86069351e-01 8.54880214e-01
-4.09374088e-01 5.36562026e-01 5.37608266e-01 1.87531742e-03
1.22622840e-01 2.31440589e-01 -1.35583389e+00 5.85747063e-01
8.43209684e-01 1.54526280e-02 -2.54266709e-01 7.74892628e-01
7.46582747e-01 -9.51478362e-01 -8.81234765e-01 1.04940951e+00
4.78866428e-01 -8.50827515e-01 1.05751956e+00 -3.98135662e-01
3.61083895e-01 9.89376605e-02 1.17606945e-01 -9.12528336e-01
-5.92234254e-01 7.39614144e-02 1.78673081e-02 4.25400347e-01
8.57013702e-01 -1.89004436e-01 1.39459920e+00 9.95237589e-01
-1.68002859e-01 -8.85748267e-01 -9.07809019e-01 -4.97152537e-01
3.79893094e-01 -6.43090844e-01 2.22015053e-01 3.25713366e-01
-4.45089132e-01 -2.83659488e-01 2.26276293e-01 4.81975883e-01
7.36132681e-01 -1.81656227e-01 1.09800607e-01 -1.14515316e+00
-5.69314957e-01 -5.53380311e-01 -1.79469809e-01 -6.23232067e-01
-5.92303574e-01 -1.12252045e+00 3.40940595e-01 -1.98957157e+00
8.02531540e-01 -5.77948272e-01 -3.17936331e-01 5.27984679e-01
-3.72690946e-01 1.89988628e-01 -2.64047682e-01 1.27199620e-01
-4.19747323e-01 6.82960600e-02 1.26829457e+00 -2.07995698e-02
1.15856804e-01 6.24425232e-01 -8.48863900e-01 8.88824284e-01
5.04324377e-01 -7.10770011e-01 -1.51564851e-01 -3.06892306e-01
-8.61720834e-03 3.18830252e-01 7.90276408e-01 -1.29562616e+00
7.49616921e-02 1.27791474e-02 8.02981257e-01 -7.50970185e-01
7.62761161e-02 -7.39852846e-01 -1.01880193e-01 1.17075980e+00
-3.60784054e-01 -6.98052287e-01 3.34650934e-01 7.45656133e-01
3.57660562e-01 -3.54520291e-01 1.03004861e+00 -3.59844834e-01
-2.16590703e-01 5.83856583e-01 -6.46918654e-01 -9.85352620e-02
1.09577620e+00 1.03795655e-01 -1.55991897e-01 -2.68647820e-01
-9.00278091e-01 6.41534865e-01 9.20130461e-02 -3.34778652e-02
1.40171915e-01 -1.32244611e+00 -9.74506378e-01 -1.75490975e-02
3.21035683e-01 5.88734746e-01 1.60604954e-01 8.19761097e-01
-9.33079243e-01 5.64595103e-01 2.51812905e-01 -1.12243140e+00
-7.75674164e-01 1.83200523e-01 1.16409969e+00 -8.32315981e-01
-3.91434610e-01 1.17494261e+00 2.42564380e-01 -5.95115006e-01
3.89728010e-01 -1.06277442e+00 2.21759751e-01 -5.03060162e-01
1.25577196e-01 9.67419986e-03 2.45086148e-01 4.02523786e-01
-4.45980847e-01 1.55272549e-02 -2.73289263e-01 1.76136028e-02
9.93604422e-01 3.14272881e-01 3.98704499e-01 1.70834120e-02
6.58368707e-01 -4.63276833e-01 -1.20008683e+00 -1.69684052e-01
-1.93499103e-02 6.49031019e-03 6.04905784e-01 -1.37074661e+00
-8.94128442e-01 6.53166354e-01 1.09807241e+00 8.95726383e-02
8.06680024e-01 6.13609433e-01 4.54161793e-01 5.80522418e-01
-1.49903014e-01 -3.04478288e-01 -9.85792652e-02 1.27667055e-01
4.55993682e-01 -1.91313815e+00 2.34451771e-01 -6.16894186e-01
-5.30668080e-01 1.02232504e+00 7.58054435e-01 -1.80773944e-01
9.56121564e-01 7.97403276e-01 5.61244134e-03 -3.06394547e-01
-8.41977060e-01 -4.04976815e-01 5.91813326e-01 6.10312641e-01
5.87065220e-01 4.21554714e-01 2.65442848e-01 8.01515996e-01
7.02928454e-02 2.56863415e-01 4.78573620e-01 8.56676042e-01
-5.74625432e-01 -1.01506281e+00 -3.39202523e-01 1.31416881e+00
-5.59199631e-01 -1.96686894e-01 -7.37692118e-02 1.03234482e+00
-8.78513902e-02 2.83283472e-01 1.55819163e-01 2.64833495e-02
1.55045763e-01 -4.47393768e-02 5.59058964e-01 -1.15940893e+00
-5.95425308e-01 -4.50050049e-02 2.01710209e-01 -4.72537011e-01
-5.25214262e-02 -6.12579226e-01 -1.29490983e+00 3.78234759e-02
-4.18104410e-01 -2.65803248e-01 5.94974339e-01 7.44529068e-01
2.47700691e-01 1.12162745e+00 9.06903371e-02 -7.11507618e-01
-1.08494186e+00 -8.77651036e-01 -3.35405886e-01 -1.64889783e-01
2.69667774e-01 -4.40526277e-01 -1.30536288e-01 -2.78243184e-01]
|
[15.296072006225586, -2.1978023052215576]
|
529202f3-855a-4d1b-bdca-77ef702b2330
|
online-low-rank-matrix-completion
|
2209.03997
| null |
https://arxiv.org/abs/2209.03997v2
|
https://arxiv.org/pdf/2209.03997v2.pdf
|
Online Low Rank Matrix Completion
|
We study the problem of {\em online} low-rank matrix completion with $\mathsf{M}$ users, $\mathsf{N}$ items and $\mathsf{T}$ rounds. In each round, the algorithm recommends one item per user, for which it gets a (noisy) reward sampled from a low-rank user-item preference matrix. The goal is to design a method with sub-linear regret (in $\mathsf{T}$) and nearly optimal dependence on $\mathsf{M}$ and $\mathsf{N}$. The problem can be easily mapped to the standard multi-armed bandit problem where each item is an {\em independent} arm, but that leads to poor regret as the correlation between arms and users is not exploited. On the other hand, exploiting the low-rank structure of reward matrix is challenging due to non-convexity of the low-rank manifold. We first demonstrate that the low-rank structure can be exploited using a simple explore-then-commit (ETC) approach that ensures a regret of $O(\mathsf{polylog} (\mathsf{M}+\mathsf{N}) \mathsf{T}^{2/3})$. That is, roughly only $\mathsf{polylog} (\mathsf{M}+\mathsf{N})$ item recommendations are required per user to get a non-trivial solution. We then improve our result for the rank-$1$ setting which in itself is quite challenging and encapsulates some of the key issues. Here, we propose \textsc{OCTAL} (Online Collaborative filTering using iterAtive user cLustering) that guarantees nearly optimal regret of $O(\mathsf{polylog} (\mathsf{M}+\mathsf{N}) \mathsf{T}^{1/2})$. OCTAL is based on a novel technique of clustering users that allows iterative elimination of items and leads to a nearly optimal minimax rate.
|
['Soumyabrata Pal', 'Prateek Jain']
|
2022-09-08
| null | null | null | null |
['low-rank-matrix-completion', 'matrix-completion']
|
['methodology', 'methodology']
|
[ 1.07069448e-01 5.59947751e-02 -1.58080518e-01 -2.85104394e-01
-1.34011614e+00 -1.12818670e+00 -2.93304324e-01 8.49292427e-02
-8.09524536e-01 7.37866640e-01 -1.23443276e-01 -7.71670759e-01
-1.08841348e+00 -6.94613159e-01 -1.04780960e+00 -6.67302430e-01
-5.05140722e-01 7.31174588e-01 -3.02767217e-01 -9.29943696e-02
1.54301152e-01 2.36805439e-01 -1.33247888e+00 4.42148112e-02
8.52146029e-01 1.38410628e+00 1.94546565e-01 7.84263849e-01
-4.77664024e-02 4.36351746e-01 -2.27442637e-01 -6.04585886e-01
7.76539862e-01 -5.42464733e-01 -7.36154556e-01 7.44260922e-02
2.99120069e-01 -1.67943284e-01 -4.17670935e-01 1.27340615e+00
1.93279386e-01 5.56601226e-01 5.89363396e-01 -7.35587239e-01
-3.53898793e-01 8.67949009e-01 -1.05780804e+00 4.52520289e-02
1.72781050e-01 -2.03837469e-01 1.38664579e+00 -6.68082714e-01
1.70125991e-01 8.90015125e-01 3.53139162e-01 9.96844172e-02
-1.33931172e+00 -9.84771430e-01 4.49393511e-01 -4.23914015e-01
-1.45600295e+00 -2.51821905e-01 1.91871822e-01 -2.73471028e-01
1.76360011e-01 7.96392202e-01 1.97773904e-01 1.62619457e-01
-4.32112664e-01 7.95030475e-01 1.18441916e+00 -3.56877565e-01
4.57431644e-01 3.57447155e-02 4.90913242e-01 7.64579117e-01
4.80758816e-01 -1.17120855e-01 -3.49664927e-01 -2.81248808e-01
9.26256120e-01 4.22156006e-01 -2.56313175e-01 -2.10922398e-02
-8.40324700e-01 9.89894986e-01 2.55199462e-01 9.35345739e-02
-2.61942118e-01 5.56749105e-01 -1.34859085e-01 5.32940149e-01
1.82673708e-01 2.65916020e-01 -7.10230291e-01 -1.73727244e-01
-1.25258172e+00 2.89112240e-01 7.48311460e-01 1.22703099e+00
9.15066242e-01 -1.84495181e-01 -1.77543864e-01 7.12104678e-01
1.00952588e-01 8.93773317e-01 -3.11680585e-01 -1.44832098e+00
9.22413588e-01 2.54420906e-01 7.62571096e-01 -6.87184215e-01
-3.10323864e-01 -6.93980932e-01 -9.27188158e-01 1.63489282e-01
8.82387877e-01 -5.15994430e-01 -3.91962558e-01 1.87814915e+00
1.50199786e-01 -3.84181827e-01 -4.61712003e-01 9.76131976e-01
-6.80143684e-02 5.82982004e-01 -5.30264795e-01 -7.38403320e-01
1.34970200e+00 -5.96606851e-01 -2.05404416e-01 -2.15947896e-01
5.90100408e-01 -7.78094113e-01 1.15326631e+00 7.15573967e-01
-1.68163383e+00 1.66777167e-02 -6.68305576e-01 4.43647146e-01
1.74205601e-01 -1.55627385e-01 1.02604210e+00 1.21113920e+00
-8.36957216e-01 6.65907681e-01 -5.37635922e-01 4.44475442e-01
3.21908951e-01 9.37966228e-01 -8.42016861e-02 -4.65941101e-01
-5.93818128e-01 -5.20411059e-02 -1.55654028e-01 -3.81913111e-02
-6.24976575e-01 -6.65314555e-01 -4.20160264e-01 1.27831176e-01
7.02780962e-01 -4.23327059e-01 9.36258316e-01 -8.65591526e-01
-1.01662266e+00 3.77338439e-01 -1.52996793e-01 -9.19338241e-02
5.06255925e-01 -1.30215526e-01 2.02821344e-02 5.51283434e-02
4.60961573e-02 -6.17637597e-02 6.84309244e-01 -9.80339050e-01
-8.93384099e-01 -7.72247851e-01 4.96249914e-01 2.03002561e-02
-4.96166110e-01 1.17776431e-01 -5.61377287e-01 -6.54655874e-01
5.46728075e-01 -1.10718048e+00 -5.09403527e-01 -3.95062178e-01
-1.00545265e-01 1.17259830e-01 -2.86740065e-01 -3.02719921e-01
1.36253726e+00 -2.12096906e+00 2.33140677e-01 8.79945695e-01
3.00918400e-01 -1.54822245e-01 -1.55429263e-02 3.13089371e-01
1.76525086e-01 2.86064297e-01 -5.73749468e-02 -4.08335716e-01
1.40868068e-01 1.76164448e-01 -1.89694747e-01 7.11176455e-01
-8.53974521e-01 3.69151235e-01 -6.96701765e-01 1.49334684e-01
-2.10514605e-01 -9.58517939e-02 -9.36188281e-01 -1.17049389e-01
-2.46143937e-01 3.38460207e-01 -6.41518176e-01 3.83793235e-01
1.02217078e+00 -6.13980532e-01 3.68435830e-01 1.46931812e-01
4.61906753e-02 -3.75073515e-02 -2.10003376e+00 1.46284950e+00
-3.81760955e-01 -2.22256690e-01 7.59967744e-01 -9.10070360e-01
2.30757415e-01 3.90635580e-02 7.70099759e-01 -5.18144906e-01
2.06051931e-01 5.18709660e-01 -3.11108261e-01 5.33857755e-02
3.02042603e-01 -4.19393331e-01 -2.62358814e-01 1.13863587e+00
-2.58264720e-01 4.96337950e-01 2.22314805e-01 4.62785304e-01
1.44185233e+00 -3.17661077e-01 -3.99237514e-01 -4.14684474e-01
2.52047271e-01 -3.58502239e-01 5.85562825e-01 1.27687478e+00
3.50914448e-01 3.05116475e-01 4.43337023e-01 -2.10665718e-01
-7.52216041e-01 -1.15836453e+00 4.25683036e-02 1.57790577e+00
2.49117285e-01 -2.53756911e-01 -7.04569519e-01 -6.02001607e-01
8.99278000e-02 5.48549771e-01 -6.42045736e-01 3.58520359e-01
-3.62207144e-01 -9.70957637e-01 3.18481088e-01 3.72587264e-01
4.69827875e-02 -4.42528039e-01 -2.37035081e-01 2.44374961e-01
-1.49372607e-01 -6.23194933e-01 -1.03073037e+00 5.01743793e-01
-8.97306204e-01 -8.55608642e-01 -7.37467587e-01 -2.43065760e-01
1.03624618e+00 6.14265501e-01 7.73346364e-01 6.42432505e-03
2.63474565e-02 4.59701598e-01 -4.26167518e-01 -1.79953650e-01
3.96390319e-01 -1.25626490e-01 2.97241479e-01 1.65029705e-01
3.38776857e-02 -8.25829089e-01 -1.02826405e+00 4.88086671e-01
-1.01176667e+00 -3.21268469e-01 5.97396076e-01 6.86513960e-01
7.72805631e-01 2.03195155e-01 3.61003369e-01 -1.17056322e+00
3.40378702e-01 -3.53464186e-01 -1.11615849e+00 -1.33268395e-02
-7.71703184e-01 2.38829732e-01 8.85352552e-01 -4.06539977e-01
-6.40612841e-01 1.59113318e-01 1.07988164e-01 -3.89287502e-01
3.28684360e-01 5.39786577e-01 1.29820079e-01 3.96622717e-03
6.89433813e-01 1.54956356e-01 -2.14220047e-01 -8.83599758e-01
5.81841648e-01 4.99419928e-01 3.10527384e-01 -9.17490005e-01
9.26079214e-01 5.80864370e-01 2.34654456e-01 -2.90235162e-01
-1.30675709e+00 -3.79219383e-01 -5.48011288e-02 1.58484459e-01
1.83615685e-01 -8.62473071e-01 -1.66080534e+00 -4.69233572e-01
-4.18460697e-01 -4.40154880e-01 -4.30123448e-01 7.42808998e-01
-6.98238730e-01 4.99371916e-01 -6.18270934e-01 -1.58176422e+00
-2.83054948e-01 -9.06638384e-01 5.20329475e-01 -8.20422545e-02
-2.28073336e-02 -5.11450827e-01 -4.48425621e-01 9.44270730e-01
2.64945626e-01 -2.69910276e-01 8.59401822e-01 -5.03129542e-01
-9.44955766e-01 -4.51326311e-01 -4.30895448e-01 3.24875563e-01
-1.54459700e-01 -9.35314178e-01 -3.00146669e-01 -7.59730875e-01
5.82845844e-02 -2.97297090e-01 7.13728189e-01 4.38486248e-01
1.40641415e+00 -9.21588600e-01 -1.99898984e-02 4.47994649e-01
1.41414046e+00 -7.26985782e-02 3.65218282e-01 -1.66618139e-01
4.21762705e-01 2.13905931e-01 6.77376509e-01 1.03007412e+00
2.70165294e-01 4.67823744e-01 4.16682422e-01 1.45420909e-01
5.76718271e-01 1.39065951e-01 4.52130258e-01 4.49085087e-01
-2.39250526e-01 -2.09398344e-02 -3.82696778e-01 4.15859371e-01
-1.99828386e+00 -8.49068344e-01 -1.63753152e-01 3.03076482e+00
9.29063022e-01 1.34994343e-01 4.38715667e-01 2.52437204e-01
4.75702137e-01 -3.50573957e-01 -3.47243935e-01 -2.85339743e-01
2.55303923e-03 7.20970869e-01 1.17685187e+00 6.42925203e-01
-7.06928790e-01 6.15824103e-01 3.96606255e+00 1.33484221e+00
-4.37547266e-01 2.10334584e-01 7.31079638e-01 -9.12018657e-01
-5.45929372e-01 1.87440410e-01 -7.71081030e-01 5.60701430e-01
9.20312345e-01 2.47034393e-02 1.28572392e+00 7.82931447e-01
2.26420805e-01 -3.76456022e-01 -1.26090634e+00 1.24389052e+00
-2.16503456e-01 -1.30064082e+00 -4.19087380e-01 5.81965268e-01
9.14265037e-01 -2.22422376e-01 3.66007626e-01 9.02142003e-02
9.47895765e-01 -1.10369730e+00 6.95942283e-01 1.67993933e-01
1.21463406e+00 -1.12764275e+00 3.36399674e-01 6.59076929e-01
-1.18195474e+00 -5.78074038e-01 -4.21415269e-01 -1.70028150e-01
1.48198768e-01 1.01113141e+00 -7.51834875e-03 6.46995544e-01
1.01256096e+00 -1.97798237e-01 2.33104214e-01 7.97762394e-01
2.81896591e-01 4.60441858e-01 -8.20519507e-01 3.56313810e-02
3.44302058e-01 -6.17155969e-01 3.60164464e-01 1.00670707e+00
5.11777341e-01 7.82952487e-01 6.00816965e-01 3.72887433e-01
-4.80977595e-01 3.43137890e-01 -2.73291580e-02 2.41729245e-02
4.63259906e-01 1.08029222e+00 -7.74022579e-01 -1.09070078e-01
-1.68593675e-01 8.65954578e-01 2.99860030e-01 3.85290116e-01
-8.34698558e-01 -4.19022292e-01 6.51432037e-01 5.34653723e-01
6.83017433e-01 -3.37738007e-01 -5.46569169e-01 -9.34177935e-01
2.26631552e-01 -7.11509168e-01 7.86263943e-01 -7.26125836e-02
-1.31351364e+00 4.76185046e-02 -2.48084843e-01 -9.03794765e-01
1.07635178e-01 -4.41488653e-01 2.04632536e-01 9.45968986e-01
-1.01950300e+00 -5.51748693e-01 2.14402273e-01 1.01444376e+00
2.66701318e-02 -1.59906689e-02 7.74022520e-01 6.08946323e-01
-2.62282968e-01 1.16765738e+00 7.77259707e-01 -1.12245910e-01
4.63213533e-01 -1.09061563e+00 -4.99780864e-01 5.13694704e-01
1.15635268e-01 8.83067787e-01 6.39099658e-01 -3.17074269e-01
-1.98986328e+00 -1.01024461e+00 6.02768719e-01 -4.68738139e-01
5.11403799e-01 -4.19616967e-01 -1.96911648e-01 7.15590417e-01
-3.80327433e-01 3.03060021e-02 1.15269089e+00 4.48747069e-01
-4.87387836e-01 -5.07014573e-01 -1.34260261e+00 7.84769535e-01
1.29756701e+00 -2.49825120e-01 2.80923277e-01 6.49236441e-01
4.69324470e-01 -2.58022070e-01 -1.08229268e+00 1.06638424e-01
5.33493102e-01 -8.33575487e-01 8.64447832e-01 -5.24340153e-01
-1.09138548e-01 -3.34253728e-01 -6.65000677e-01 -5.16205072e-01
-4.61174220e-01 -1.28580010e+00 -2.58442551e-01 7.80412316e-01
8.92361224e-01 -4.74968225e-01 1.08150172e+00 1.02140939e+00
2.42164537e-01 -7.27602780e-01 -1.10841286e+00 -7.84074545e-01
2.62162238e-01 -6.55435443e-01 1.38688877e-01 6.83514774e-01
5.96409142e-02 1.32454053e-01 -7.21505761e-01 1.14836484e-01
8.07162821e-01 3.70035052e-01 6.87367916e-01 -1.11582804e+00
-1.22920120e+00 -3.06531549e-01 4.66260284e-01 -1.52888489e+00
-5.61630309e-01 -9.26180661e-01 -2.35514253e-01 -1.21763301e+00
5.17421424e-01 -1.23233771e+00 -6.13006115e-01 4.95176584e-01
1.43267512e-01 2.00209990e-01 2.85354555e-01 1.75635427e-01
-1.16511822e+00 8.63826349e-02 9.30480123e-01 8.36411640e-02
-3.43165725e-01 4.78771240e-01 -1.27388287e+00 4.60399270e-01
3.98720860e-01 -6.34945810e-01 -4.33684677e-01 -5.90196550e-01
1.00401509e+00 7.72933483e-01 -7.98151866e-02 -5.34490883e-01
3.26108664e-01 -3.40380043e-01 2.30228245e-01 -5.26519060e-01
4.06850815e-01 -8.49543452e-01 2.36593172e-01 3.08443248e-01
-2.86486745e-01 -3.21366787e-02 -2.31659308e-01 8.57519209e-01
4.93093699e-01 -3.18397939e-01 7.14112937e-01 -3.04280967e-01
5.61370373e-01 5.46213031e-01 -2.95374155e-01 1.39618844e-01
8.71393144e-01 -1.15157746e-01 3.03474218e-02 -7.55988419e-01
-7.67311394e-01 4.18290168e-01 3.44901323e-01 -2.66862273e-01
3.20535511e-01 -1.06181359e+00 -5.39492905e-01 -1.29045740e-01
-3.41173947e-01 1.68596745e-01 6.38875306e-01 1.03952229e+00
-2.72980109e-02 2.76068330e-01 4.23027813e-01 -2.27610528e-01
-9.10481870e-01 7.84107745e-01 -7.64484257e-02 -5.61656415e-01
-5.45177497e-02 1.21683645e+00 2.50840932e-02 -1.80053458e-01
3.79454285e-01 1.53700411e-01 5.37332773e-01 -1.38407927e-02
6.01232588e-01 7.65664518e-01 7.94588495e-03 -1.59811154e-01
-4.17140499e-02 2.62031794e-01 -4.10078257e-01 -5.78038275e-01
1.42634594e+00 -3.41494530e-01 -3.18092912e-01 5.38096093e-02
1.16807353e+00 4.88508999e-01 -1.19175637e+00 -3.92631590e-01
-8.09845552e-02 -7.15257287e-01 -2.10637167e-01 -8.17893088e-01
-1.19623935e+00 7.39749491e-01 6.98690176e-01 2.12541178e-01
1.12225699e+00 -3.30794826e-02 8.23226988e-01 4.94967490e-01
1.07383943e+00 -1.45741987e+00 1.97958305e-01 5.10863781e-01
4.05120909e-01 -8.52449954e-01 -2.90182279e-03 -2.58908659e-01
-4.34045821e-01 7.49735117e-01 -1.88847110e-02 -2.44788334e-01
8.27179193e-01 4.72270884e-02 -5.87251484e-01 -1.25107393e-01
-2.15708315e-01 -1.18567266e-01 3.60642225e-02 -1.45024210e-01
3.70515853e-01 3.71986061e-01 -4.58124667e-01 1.39294219e+00
-1.96792603e-01 -1.95130214e-01 3.58668089e-01 9.11719799e-01
-6.61916733e-01 -1.47218430e+00 -5.19682467e-01 9.58395898e-01
-9.91021454e-01 -1.37013078e-01 2.02533931e-01 3.40219550e-02
5.30082248e-02 1.30116761e+00 -2.83043146e-01 -2.06135496e-01
1.75251603e-01 -1.00808911e-01 6.32257342e-01 -5.84232926e-01
-6.91039145e-01 6.41925097e-01 -2.73911003e-02 -5.27405143e-01
1.80584967e-01 -5.92275858e-01 -1.34469175e+00 -7.39425123e-01
-2.50600547e-01 5.04608512e-01 4.16813344e-01 8.60901594e-01
3.78941149e-01 4.44690622e-02 1.07697022e+00 -4.82088238e-01
-8.72508466e-01 -6.20734632e-01 -1.11129653e+00 3.98245275e-01
-8.56514126e-02 -3.70132118e-01 -3.89936119e-01 -4.00995612e-01]
|
[4.920475006103516, 3.6521778106689453]
|
b2a8a9fe-f3a6-4dee-9386-d637c76ce968
|
improving-transformer-based-image-matching-by
|
2303.02885
| null |
https://arxiv.org/abs/2303.02885v1
|
https://arxiv.org/pdf/2303.02885v1.pdf
|
Improving Transformer-based Image Matching by Cascaded Capturing Spatially Informative Keypoints
|
Learning robust local image feature matching is a fundamental low-level vision task, which has been widely explored in the past few years. Recently, detector-free local feature matchers based on transformers have shown promising results, which largely outperform pure Convolutional Neural Network (CNN) based ones. But correlations produced by transformer-based methods are spatially limited to the center of source views' coarse patches, because of the costly attention learning. In this work, we rethink this issue and find that such matching formulation degrades pose estimation, especially for low-resolution images. So we propose a transformer-based cascade matching model -- Cascade feature Matching TRansformer (CasMTR), to efficiently learn dense feature correlations, which allows us to choose more reliable matching pairs for the relative pose estimation. Instead of re-training a new detector, we use a simple yet effective Non-Maximum Suppression (NMS) post-process to filter keypoints through the confidence map, and largely improve the matching precision. CasMTR achieves state-of-the-art performance in indoor and outdoor pose estimation as well as visual localization. Moreover, thorough ablations show the efficacy of the proposed components and techniques.
|
['Yanwei Fu', 'Chenjie Cao']
|
2023-03-06
| null | null | null | null |
['visual-localization']
|
['computer-vision']
|
[-8.99802148e-02 -5.70577800e-01 -3.90027538e-02 -3.10374439e-01
-9.75957811e-01 -2.32828960e-01 5.07325351e-01 -1.01736739e-01
-5.00922918e-01 3.17725420e-01 1.98177561e-01 2.26232275e-01
-3.74849230e-01 -8.17042291e-01 -9.11364079e-01 -5.97031832e-01
2.61372268e-01 1.20734654e-01 5.97903073e-01 -3.15449864e-01
2.08797038e-01 6.83261395e-01 -1.65660334e+00 4.49372008e-02
8.00667524e-01 1.36012328e+00 3.51720124e-01 1.18573308e-01
2.45234773e-01 4.64894652e-01 -4.04695898e-01 -4.00030941e-01
4.15907681e-01 -4.14724834e-03 -2.67441452e-01 -1.08671725e-01
1.06472743e+00 -3.22982281e-01 -6.22123241e-01 1.11611664e+00
6.93998754e-01 5.52501045e-02 4.16324854e-01 -1.08086133e+00
-4.65575933e-01 8.23710784e-02 -7.50562966e-01 -2.39146361e-03
4.16623503e-01 -1.23672411e-02 9.13574517e-01 -1.43169498e+00
4.75186765e-01 1.20879829e+00 1.06743312e+00 1.54766947e-01
-8.03541183e-01 -1.06957960e+00 2.61373252e-01 3.76219541e-01
-1.78785157e+00 -3.94071370e-01 8.40399623e-01 -2.29290560e-01
7.85776496e-01 7.68517479e-02 7.30020821e-01 8.89267504e-01
3.71962935e-01 5.70183873e-01 1.01860213e+00 -2.59493351e-01
-1.09027848e-01 -2.06249878e-01 -4.38130856e-01 1.06005251e+00
9.90071669e-02 3.85428190e-01 -1.00106609e+00 -3.23077887e-02
1.14351523e+00 4.89558488e-01 -4.00905252e-01 -7.20898151e-01
-1.43881619e+00 7.38260806e-01 1.17281413e+00 2.31390551e-01
-2.03045666e-01 2.74282485e-01 -1.57096758e-02 9.77907106e-02
4.57437277e-01 3.27460796e-01 -3.20387512e-01 2.68599749e-01
-9.55037177e-01 2.30685636e-01 6.22803010e-02 9.64815438e-01
9.77063775e-01 -3.04328144e-01 -3.39475334e-01 7.92241335e-01
3.71588200e-01 7.22015917e-01 2.70416528e-01 -5.70572078e-01
6.05167270e-01 6.73592985e-01 2.37551495e-01 -1.39641261e+00
-4.92212683e-01 -8.88142824e-01 -9.39955890e-01 1.06005117e-01
3.19705218e-01 2.38747537e-01 -8.39805782e-01 1.41262710e+00
3.90158445e-01 1.05912752e-01 -3.93580437e-01 1.23098028e+00
7.71383464e-01 3.43607545e-01 -2.76081651e-01 9.40942690e-02
1.20536840e+00 -1.09261549e+00 -5.06291091e-01 -3.32098961e-01
2.62706399e-01 -1.06668735e+00 7.69207358e-01 2.87157208e-01
-7.07535505e-01 -8.25581372e-01 -1.22896993e+00 -2.12080941e-01
-2.79336959e-01 4.62080896e-01 7.64783800e-01 2.89414823e-01
-8.80340040e-01 7.42286623e-01 -7.92765319e-01 -3.58452171e-01
5.53761601e-01 4.27862853e-01 -5.88110924e-01 -4.34173614e-01
-1.02156508e+00 9.15274441e-01 -6.23556562e-02 5.21201968e-01
-6.82905555e-01 -5.80383420e-01 -9.71909463e-01 -9.43295062e-02
4.59218234e-01 -9.62933779e-01 9.77333903e-01 -5.43331861e-01
-1.25423777e+00 7.02878535e-01 -4.24692929e-01 -2.51893103e-01
6.10564291e-01 -5.76689661e-01 -2.14988336e-01 8.55246000e-03
2.64171481e-01 6.42190516e-01 1.05924547e+00 -9.82901454e-01
-8.06629956e-01 -5.42901576e-01 1.70656405e-02 3.16305250e-01
-3.04547876e-01 -4.27055396e-02 -6.42567337e-01 -6.87709033e-01
7.65742064e-01 -7.44336307e-01 -3.27024639e-01 3.92099202e-01
-2.07875222e-01 -2.31695965e-01 6.23334944e-01 -4.04608399e-01
1.12546158e+00 -2.01383328e+00 -1.01642393e-01 2.43691131e-01
3.77158344e-01 3.84633839e-02 -5.50616682e-02 3.08861554e-01
1.31990105e-01 -3.96199405e-01 1.69482529e-01 -4.70212221e-01
-1.99445814e-01 -1.67102873e-01 -2.66299933e-01 7.92647302e-01
1.45712912e-01 1.03844023e+00 -8.29720616e-01 -3.77568513e-01
5.11408150e-01 5.21524251e-01 -4.92570996e-01 5.69817685e-02
1.12956263e-01 3.98688495e-01 -5.77194870e-01 8.79695237e-01
1.03975642e+00 -2.45102599e-01 -3.91599327e-01 -9.66956258e-01
-4.12239194e-01 5.11266887e-02 -1.27267766e+00 2.21669412e+00
-6.56547010e-01 4.76528645e-01 -1.86760619e-01 -7.53854692e-01
1.11980712e+00 -2.24889293e-01 3.65353167e-01 -9.75483537e-01
1.23484321e-01 4.10565853e-01 -2.62506872e-01 1.00622959e-02
3.12408566e-01 1.80102184e-01 -4.55251262e-02 -2.80008793e-01
1.70868903e-01 -6.13230355e-02 -3.07379216e-01 -4.37884443e-02
9.87454116e-01 3.20945799e-01 4.25560713e-01 -7.30044991e-02
6.43325567e-01 -1.34524256e-01 6.42868876e-01 7.14129150e-01
-3.33529748e-02 9.39135730e-01 -2.28842959e-01 -5.06439924e-01
-7.53124714e-01 -1.16491294e+00 -2.29544908e-01 6.48741543e-01
7.31096148e-01 -7.08793759e-01 -4.43328679e-01 -5.56053519e-01
6.91133365e-02 -3.79921407e-01 -4.49185342e-01 -2.60981590e-01
-7.62773454e-01 -3.77201170e-01 1.46927014e-01 7.48871624e-01
1.06991100e+00 -6.24687374e-01 -5.96649945e-01 2.09756598e-01
-2.09872410e-01 -1.15723121e+00 -6.09598637e-01 1.06450394e-01
-6.06250465e-01 -9.81808364e-01 -8.99414420e-01 -7.26123571e-01
6.40599310e-01 6.43963218e-01 8.46100986e-01 3.74534540e-02
-3.78693730e-01 1.82449177e-01 -3.00823927e-01 -3.17123197e-02
4.65595812e-01 3.57117988e-02 2.33555809e-01 1.00422636e-01
2.20750555e-01 -5.39683104e-01 -1.00581384e+00 6.94586873e-01
-2.51488835e-01 -3.00423931e-02 1.08211470e+00 1.09525335e+00
7.29536831e-01 -2.14598775e-01 1.60128459e-01 -2.28795856e-01
5.32293841e-02 1.49285674e-01 -8.41193974e-01 3.29476506e-01
-4.87422913e-01 8.04685131e-02 6.11297846e-01 -3.74475181e-01
-7.54213333e-01 3.61239791e-01 -3.28005731e-01 -8.06212068e-01
1.10314451e-01 2.52843291e-01 -2.04248920e-01 -8.76379192e-01
6.83682382e-01 2.89870441e-01 -2.60156840e-01 -4.95977372e-01
1.47851929e-01 4.55787718e-01 6.12105668e-01 -3.41528058e-01
1.32752109e+00 6.10926569e-01 2.97684014e-01 -6.07894540e-01
-1.09004986e+00 -7.52806783e-01 -7.01323807e-01 -1.17708847e-01
6.45348072e-01 -1.38098967e+00 -8.68031919e-01 5.71515441e-01
-1.13779998e+00 1.30117118e-01 3.14788893e-02 6.28957570e-01
-3.46723229e-01 2.04174280e-01 -3.66514415e-01 -4.05634075e-01
-3.28675777e-01 -1.20919490e+00 1.43671501e+00 4.09634084e-01
9.78038236e-02 -5.48117638e-01 -2.94464808e-02 2.57160246e-01
6.43607616e-01 8.27669278e-02 1.69583708e-01 -7.08032399e-02
-1.20974493e+00 -3.33249182e-01 -4.76535201e-01 1.27137721e-01
9.12904516e-02 -3.32742184e-01 -1.16672146e+00 -5.58056831e-01
-2.10824981e-01 -1.70241535e-01 1.04924929e+00 4.32182461e-01
1.11232507e+00 9.63947475e-02 -6.11394465e-01 1.14219201e+00
1.46810889e+00 -1.87911436e-01 4.87224132e-01 4.41436827e-01
8.92838299e-01 3.61686647e-01 1.04293013e+00 3.26787353e-01
3.43096286e-01 1.12036037e+00 4.35665876e-01 -3.87167096e-01
-1.73361123e-01 -6.16091073e-01 2.38174811e-01 4.48095858e-01
-8.31389204e-02 2.00061709e-01 -6.08504117e-01 2.58645266e-01
-2.03156304e+00 -6.99617684e-01 1.62096992e-01 2.35633421e+00
4.86488283e-01 9.86404121e-02 -2.28526622e-01 -1.80990860e-01
5.51269770e-01 3.18503499e-01 -4.03981060e-01 5.34848154e-01
-1.15139268e-01 4.94766712e-01 7.25090206e-01 4.07182187e-01
-1.40033269e+00 9.95654643e-01 5.36715651e+00 1.36558807e+00
-1.13179815e+00 1.36175334e-01 2.54088312e-01 7.76657909e-02
-6.07948527e-02 5.75126521e-02 -1.00172698e+00 2.10233837e-01
1.71851113e-01 2.33910531e-01 1.93812355e-01 1.00381279e+00
-8.03894326e-02 -2.23270699e-01 -1.01806355e+00 1.53012431e+00
1.22971460e-01 -1.41994798e+00 -5.48137501e-02 -1.13335319e-01
6.78336740e-01 8.39691237e-02 2.09065676e-01 2.13069022e-01
-3.00015628e-01 -9.95338380e-01 7.14835107e-01 5.17414570e-01
8.06715548e-01 -9.37470555e-01 8.42427254e-01 3.08259398e-01
-1.86620986e+00 -2.89753288e-01 -7.53406465e-01 2.67839190e-02
-1.08174808e-01 7.85778821e-01 -2.71329284e-01 7.88909614e-01
1.05941999e+00 9.65842426e-01 -9.03538048e-01 1.40597808e+00
-2.52193570e-01 -9.70112309e-02 -5.08958519e-01 1.59750894e-01
9.39912871e-02 -3.81762497e-02 2.75269806e-01 9.46394324e-01
4.79498953e-01 -1.41170651e-01 3.90092462e-01 8.97758186e-01
1.50944833e-02 -1.62432436e-02 -5.14933050e-01 5.21413863e-01
5.89906991e-01 1.52420509e+00 -6.29073441e-01 -4.17060629e-02
-4.29039329e-01 1.19326043e+00 4.15589780e-01 1.23741776e-01
-7.12023318e-01 -4.58727956e-01 7.21892178e-01 2.85760522e-01
3.36173594e-01 -3.51434976e-01 1.18583679e-01 -1.38497567e+00
4.50357974e-01 -7.39760816e-01 5.08369952e-02 -6.20409548e-01
-1.22842789e+00 4.60750818e-01 -1.71544567e-01 -1.90531933e+00
-1.30020261e-01 -5.61648726e-01 -4.59133893e-01 9.32442725e-01
-1.64813340e+00 -1.47734034e+00 -8.56355786e-01 7.99834311e-01
3.33278626e-01 2.85903364e-02 5.00480831e-01 4.94607240e-01
-2.67900348e-01 8.73440921e-01 -1.33246273e-01 2.72168368e-01
1.01321685e+00 -7.80888796e-01 4.09341186e-01 1.00189960e+00
2.94413805e-01 8.47507894e-01 3.50480735e-01 -5.13100803e-01
-1.42871273e+00 -1.00612295e+00 7.61561036e-01 -4.32029873e-01
3.25794280e-01 -6.77638590e-01 -5.71729839e-01 2.69467503e-01
-1.56177521e-01 5.43900371e-01 2.12243907e-02 2.69227382e-02
-5.61671913e-01 -6.29459262e-01 -9.53152418e-01 5.43216527e-01
1.53372359e+00 -6.76854253e-01 -2.98267335e-01 3.18987548e-01
6.84513330e-01 -7.16101646e-01 -6.64591432e-01 7.85762250e-01
6.18982077e-01 -1.35088372e+00 1.38327312e+00 1.84894413e-01
8.85412842e-02 -6.96330249e-01 -2.28829324e-01 -1.05262816e+00
-5.15498817e-01 -4.78993326e-01 5.88659896e-03 1.08030796e+00
-6.72898665e-02 -6.55433297e-01 7.92708814e-01 8.11665598e-03
-2.21652284e-01 -9.31700706e-01 -1.05876184e+00 -8.52388203e-01
-3.64617079e-01 -1.81887016e-01 4.24474567e-01 5.50653160e-01
-3.96072268e-01 2.61695892e-01 -4.46129173e-01 4.81161803e-01
8.00139487e-01 4.36673194e-01 8.88692915e-01 -1.13424814e+00
-2.12751791e-01 -2.14295521e-01 -8.14874232e-01 -1.58477247e+00
-1.21878505e-01 -4.13231760e-01 1.51019588e-01 -1.44593275e+00
1.16368122e-01 -5.69808364e-01 -1.65402994e-01 2.95415372e-01
-9.44292694e-02 5.61891675e-01 6.39803112e-02 2.56273866e-01
-6.86469495e-01 7.83820331e-01 1.30742002e+00 -4.30052839e-02
1.22049183e-01 2.25340411e-01 -3.80597025e-01 7.02375054e-01
3.24464619e-01 -3.43568116e-01 -2.07686201e-01 -2.28282884e-01
3.11939716e-01 -1.86270028e-01 6.95291936e-01 -1.47963524e+00
6.67809546e-01 1.27344476e-02 1.02472460e+00 -9.74718153e-01
6.45016849e-01 -9.80692446e-01 5.11442535e-02 5.20734727e-01
6.94339424e-02 2.35769764e-01 -5.46106361e-02 5.22304952e-01
-5.85973561e-01 1.77743644e-01 7.45876551e-01 -1.04087651e-01
-9.60302532e-01 8.52755904e-01 3.10519427e-01 -2.46316284e-01
8.28078210e-01 -3.69295210e-01 -3.04519087e-01 -2.99771786e-01
-2.70521402e-01 1.14993587e-01 6.09664679e-01 4.61291939e-01
9.56487715e-01 -1.58426464e+00 -4.24085319e-01 4.57919151e-01
4.23246622e-01 1.94816247e-01 4.34534192e-01 1.03466201e+00
-3.48353356e-01 3.76559854e-01 -1.57805681e-01 -8.72008502e-01
-1.14323306e+00 5.01740634e-01 4.87438351e-01 -3.91683020e-02
-7.57402718e-01 9.75302696e-01 4.51600164e-01 -3.63385111e-01
2.71771044e-01 -4.50699806e-01 -9.03112590e-02 -5.70981242e-02
5.35352349e-01 1.62590802e-01 3.54670972e-01 -6.24360204e-01
-7.47676373e-01 1.48580909e+00 -1.02723442e-01 2.21780628e-01
9.96292293e-01 -1.42588705e-01 9.52149555e-02 -1.10265799e-01
1.41298282e+00 2.41438091e-01 -1.35485601e+00 -5.11295021e-01
-2.61940807e-01 -9.55596745e-01 1.31616771e-01 -3.80837947e-01
-1.00579965e+00 9.94900882e-01 9.96907234e-01 -5.53025723e-01
1.16519439e+00 1.19028045e-02 7.50958085e-01 4.17108387e-01
7.71966815e-01 -8.22509646e-01 3.21125031e-01 5.02634943e-01
9.86167014e-01 -1.43675458e+00 2.14803085e-01 -5.96833348e-01
-1.13815323e-01 1.23257494e+00 9.13967073e-01 -3.70588124e-01
6.81132317e-01 1.82881847e-01 3.68501469e-02 -1.76728487e-01
-3.08084399e-01 -4.94602889e-01 6.61940515e-01 6.59653723e-01
1.76584423e-01 -2.61326373e-01 8.16062316e-02 3.69134575e-01
-1.65432200e-01 -2.67168671e-01 -2.37221181e-01 7.74362147e-01
-4.35996950e-01 -9.00214076e-01 -4.66075152e-01 2.14254498e-01
-3.57372090e-02 -1.91507667e-01 -1.84550032e-01 7.80381382e-01
3.65497559e-01 6.71071291e-01 3.93180400e-02 -7.28774846e-01
5.60653806e-01 -5.46392322e-01 8.19698036e-01 -3.68357778e-01
-5.27666986e-01 2.53202915e-01 -4.08767641e-01 -1.17295992e+00
-5.30798078e-01 -3.73364419e-01 -8.36465836e-01 -1.03451528e-01
-6.72277093e-01 -1.40963584e-01 4.21845198e-01 9.41250980e-01
3.47783387e-01 3.41213912e-01 7.21292973e-01 -1.33377743e+00
-6.66851759e-01 -7.97754109e-01 -3.04412901e-01 1.54978648e-01
5.15653193e-01 -1.04758191e+00 -3.02132875e-01 -5.83619833e-01]
|
[7.870059013366699, -2.151845932006836]
|
0f8a4324-cd9b-4b79-8811-fadc924283ff
|
photorealistic-monocular-3d-reconstruction-of
|
2204.08906
| null |
https://arxiv.org/abs/2204.08906v1
|
https://arxiv.org/pdf/2204.08906v1.pdf
|
Photorealistic Monocular 3D Reconstruction of Humans Wearing Clothing
|
We present PHORHUM, a novel, end-to-end trainable, deep neural network methodology for photorealistic 3D human reconstruction given just a monocular RGB image. Our pixel-aligned method estimates detailed 3D geometry and, for the first time, the unshaded surface color together with the scene illumination. Observing that 3D supervision alone is not sufficient for high fidelity color reconstruction, we introduce patch-based rendering losses that enable reliable color reconstruction on visible parts of the human, and detailed and plausible color estimation for the non-visible parts. Moreover, our method specifically addresses methodological and practical limitations of prior work in terms of representing geometry, albedo, and illumination effects, in an end-to-end model where factors can be effectively disentangled. In extensive experiments, we demonstrate the versatility and robustness of our approach. Our state-of-the-art results validate the method qualitatively and for different metrics, for both geometric and color reconstruction.
|
['Cristian Sminchisescu', 'Mihai Zanfir', 'Thiemo Alldieck']
|
2022-04-19
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Alldieck_Photorealistic_Monocular_3D_Reconstruction_of_Humans_Wearing_Clothing_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Alldieck_Photorealistic_Monocular_3D_Reconstruction_of_Humans_Wearing_Clothing_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['3d-human-reconstruction']
|
['computer-vision']
|
[ 1.29678145e-01 1.01698525e-01 5.21927774e-01 -2.85070270e-01
-4.83099788e-01 -5.47010958e-01 4.28333461e-01 -3.25928420e-01
-2.25274742e-01 4.63620871e-01 2.11131111e-01 2.23814454e-02
2.84403712e-01 -5.02865732e-01 -9.02979612e-01 -4.09071356e-01
1.95618331e-01 3.69325250e-01 -1.91360891e-01 -4.99529280e-02
-1.38783619e-01 8.38068187e-01 -1.66553974e+00 9.31974426e-02
5.81083894e-01 9.90969241e-01 -3.68700117e-01 8.77673745e-01
4.75829929e-01 8.20790350e-01 -2.98959464e-01 -4.89674032e-01
8.57612669e-01 -4.01443392e-01 -4.24931169e-01 4.30911332e-01
1.20258999e+00 -7.47287333e-01 -6.24175847e-01 6.54518485e-01
5.51255703e-01 2.25982994e-01 4.87359732e-01 -1.02234781e+00
-6.32726789e-01 -4.48005378e-01 -6.72105312e-01 -5.23255706e-01
6.54678524e-01 6.19139075e-01 8.69332552e-01 -8.56904149e-01
6.78232253e-01 1.33928943e+00 7.18786776e-01 5.15034676e-01
-1.50137687e+00 -2.40462974e-01 2.37126142e-01 -2.86314189e-01
-1.22577846e+00 -3.58468831e-01 1.03349805e+00 -4.91803646e-01
7.51742661e-01 2.80210555e-01 1.04564714e+00 9.73396122e-01
2.73650475e-02 6.99419558e-01 1.35948277e+00 -5.14745176e-01
1.47644162e-01 -3.44088748e-02 -3.98783267e-01 1.03076732e+00
1.73913121e-01 6.67857587e-01 -6.64990366e-01 1.16444543e-01
1.37130105e+00 -2.67700385e-03 -4.33315188e-01 -9.43739772e-01
-1.15690899e+00 2.98085839e-01 7.89192379e-01 -3.44188333e-01
-4.51146871e-01 4.51894879e-01 -2.40340561e-01 -4.36219722e-02
6.74081981e-01 2.52140939e-01 -3.91264677e-01 6.29710853e-02
-7.00419366e-01 3.01122397e-01 6.09547496e-01 8.80082250e-01
7.47963846e-01 8.72496963e-02 2.87387427e-02 3.82048786e-01
2.82064021e-01 7.53256619e-01 -5.39579391e-01 -1.58838069e+00
1.81593895e-01 5.01534700e-01 4.54646111e-01 -9.27919388e-01
-5.51747978e-01 -4.27803010e-01 -8.43228877e-01 1.05212820e+00
6.87923968e-01 -1.34689555e-01 -9.81381953e-01 1.76046383e+00
6.45738482e-01 -1.53968796e-01 -1.97302759e-01 1.41669381e+00
6.69407964e-01 1.75957456e-01 -1.68371022e-01 2.72548765e-01
1.18142807e+00 -8.83949697e-01 -1.23352863e-01 -4.45946187e-01
-8.44903514e-02 -6.21372640e-01 1.22004414e+00 6.06464267e-01
-1.47563529e+00 -5.44876516e-01 -8.57858658e-01 -6.31903589e-01
8.09538588e-02 2.67681360e-01 7.21067727e-01 6.28504395e-01
-1.26442373e+00 6.00845695e-01 -9.07067299e-01 -2.70101726e-01
2.53320754e-01 6.96423873e-02 -5.31341076e-01 -4.40760940e-01
-6.38883054e-01 8.27528119e-01 -2.12953746e-01 3.59716445e-01
-7.91992247e-01 -8.24630380e-01 -9.17213261e-01 6.32583797e-02
2.71144927e-01 -1.22929299e+00 1.05903506e+00 -1.04652929e+00
-1.58674419e+00 1.12974167e+00 9.79847386e-02 6.08406886e-02
1.07435834e+00 -4.27808672e-01 3.31812292e-01 4.49479312e-01
-4.28790480e-01 7.37904906e-01 7.09699392e-01 -1.72173846e+00
-1.41258538e-01 -6.32117927e-01 3.13453168e-01 6.33218586e-01
2.32267082e-01 -3.94314528e-01 -6.58050358e-01 -4.36974764e-01
1.12729825e-01 -8.49877000e-01 -1.97752476e-01 9.92792487e-01
-5.90970099e-01 5.31332731e-01 7.14344010e-02 -1.00996995e+00
1.07629441e-01 -2.02493477e+00 4.63514656e-01 1.61382645e-01
4.52005804e-01 -1.59948826e-01 -1.26989633e-01 2.61198312e-01
1.02011645e-02 -2.90874153e-01 -2.64654994e-01 -9.84916806e-01
2.29472294e-01 -1.17479838e-01 9.67002362e-02 7.20830262e-01
2.11707994e-01 9.17512655e-01 -6.86686635e-01 -2.06249624e-01
7.18503952e-01 1.23325431e+00 -4.86460209e-01 4.54813659e-01
-2.41215229e-01 7.77248204e-01 -5.31149376e-03 6.83007181e-01
8.66412163e-01 -1.80468246e-01 7.12032095e-02 -4.33574051e-01
6.55406117e-02 1.02303125e-01 -1.00632215e+00 2.05858684e+00
-6.43682480e-01 6.19388282e-01 3.42431813e-01 -2.76851326e-01
5.21577477e-01 2.78091673e-02 4.16207075e-01 -1.01386440e+00
1.08660772e-01 -6.35675564e-02 -5.24932146e-01 -2.12551042e-01
4.25564259e-01 -1.50887147e-01 1.97431222e-01 3.95531118e-01
-2.40991771e-01 -1.92803562e-01 -5.05536616e-01 1.99515551e-01
8.15161288e-01 8.49693894e-01 6.92434162e-02 8.14816207e-02
-5.15247509e-02 -1.35686040e-01 1.75181091e-01 4.36454803e-01
-1.29392460e-01 1.36345947e+00 3.60609591e-01 -5.47927141e-01
-1.39578795e+00 -1.26246572e+00 2.60601431e-01 7.73420334e-01
2.07093850e-01 9.57258418e-02 -6.94245756e-01 -3.49433780e-01
1.59287900e-01 5.80321729e-01 -9.37476158e-01 1.86510712e-01
-4.71716583e-01 -3.04817975e-01 2.52469808e-01 4.98067468e-01
3.75584066e-01 -7.53678679e-01 -1.11904895e+00 -1.70503646e-01
-1.78322107e-01 -1.32787108e+00 -2.84705251e-01 -1.96763620e-01
-7.02129364e-01 -1.30723536e+00 -8.32073987e-01 -2.00716645e-01
8.28918934e-01 5.89958787e-01 1.39076066e+00 1.88965335e-01
-6.65929735e-01 6.81816518e-01 2.40547620e-02 1.51452571e-01
-1.05559997e-01 -6.24227345e-01 -2.93635696e-01 5.17342947e-02
-2.59015560e-01 -6.70377553e-01 -1.09363258e+00 1.50221348e-01
-5.65113306e-01 6.70801044e-01 4.31946784e-01 4.60914105e-01
5.56206584e-01 -5.08906126e-01 -5.67103088e-01 -7.04663217e-01
-4.21328507e-02 2.60760158e-01 -7.36643434e-01 2.52070934e-01
-3.42220277e-01 -5.32192551e-02 4.09371346e-01 -1.73655450e-01
-1.26070917e+00 4.16146576e-01 3.66412848e-02 -6.94295287e-01
-4.10319686e-01 -1.16193861e-01 -1.04606338e-01 -4.09402549e-01
7.91183293e-01 6.68036565e-02 9.86903310e-02 -4.75327849e-01
6.68629706e-01 -1.04063598e-03 7.64145792e-01 -5.09950101e-01
8.87228668e-01 9.90040243e-01 3.86353523e-01 -6.89680457e-01
-7.15069413e-01 -1.76471755e-01 -9.34377253e-01 -3.97665709e-01
8.49521339e-01 -1.31413209e+00 -9.82810497e-01 6.10510886e-01
-1.29357636e+00 -6.42064631e-01 -2.62778848e-01 2.47930184e-01
-6.65151834e-01 4.01250988e-01 -6.42724574e-01 -1.10773599e+00
-2.87554979e-01 -8.60182106e-01 1.57607651e+00 1.44024286e-02
3.23607326e-02 -8.72506738e-01 -1.35511056e-01 5.43839812e-01
1.76567763e-01 9.51846957e-01 7.02438116e-01 4.51244682e-01
-1.03488135e+00 -1.86744407e-01 -5.47395587e-01 2.81726778e-01
-1.77103788e-01 -8.75365883e-02 -1.45934367e+00 -2.62200356e-01
-3.23196054e-01 -4.81083333e-01 7.43940830e-01 3.59999001e-01
8.43388081e-01 -1.05720088e-01 1.84324682e-01 1.04806495e+00
1.58313477e+00 -5.78599274e-01 7.40185738e-01 5.67851588e-02
1.35629797e+00 7.88085461e-01 1.78044513e-01 3.52369308e-01
6.64235771e-01 7.18620121e-01 5.73696554e-01 -7.57159352e-01
-6.25700653e-01 -3.44458193e-01 1.52209967e-01 6.44545630e-02
-3.86309981e-01 -1.59502283e-01 -5.23509145e-01 1.10905886e-01
-1.52702248e+00 -6.52331471e-01 -1.24903291e-01 2.45686030e+00
4.76072162e-01 -2.43096948e-01 2.74151832e-01 8.78963023e-02
1.85957000e-01 -4.76611257e-02 -7.37805367e-01 -4.40722331e-02
-3.25451374e-01 1.57891914e-01 4.87095743e-01 7.23915458e-01
-7.77100861e-01 8.08629036e-01 6.12299109e+00 6.60586357e-02
-9.54062283e-01 -1.79973960e-01 7.67161012e-01 -5.93419254e-01
-6.25562668e-01 -3.79010029e-02 -9.95587707e-02 -6.66241422e-02
2.28497982e-01 6.43712461e-01 8.69562685e-01 5.34477532e-01
1.68403298e-01 -2.86558837e-01 -1.18137646e+00 1.19809914e+00
3.25488329e-01 -9.81449127e-01 -1.66023180e-01 2.39851654e-01
7.14896321e-01 -5.20033911e-02 1.96456119e-01 -2.64871925e-01
3.41714442e-01 -9.89143729e-01 1.15114355e+00 7.40367413e-01
1.01243484e+00 -6.52974427e-01 2.16257706e-01 5.81420660e-02
-8.30322266e-01 1.63996220e-01 -9.93268937e-02 -1.34129390e-01
1.85727596e-01 6.14088655e-01 -2.23215923e-01 6.34194791e-01
6.85569286e-01 5.95554352e-01 -6.33433461e-01 8.61606359e-01
-5.76063991e-01 -2.79254857e-02 -4.46272641e-01 2.79670566e-01
-8.41660798e-02 -3.05690169e-01 3.21940809e-01 8.42465222e-01
-1.00181783e-02 3.28547657e-01 3.02576758e-02 1.28306615e+00
1.22329509e-02 -1.30445629e-01 -3.95343632e-01 2.82180101e-01
-6.96943700e-03 1.12068725e+00 -5.37349403e-01 1.06763124e-01
-2.59145796e-01 1.43022645e+00 5.73876441e-01 8.38412106e-01
-5.78652620e-01 -6.17535226e-02 6.49147630e-01 2.81484723e-01
2.94825137e-01 -5.08805990e-01 -6.41569495e-01 -1.36464500e+00
4.55435127e-01 -8.12369823e-01 2.06206534e-02 -1.28648770e+00
-1.22053945e+00 4.88412410e-01 -1.02757163e-01 -1.11798239e+00
-1.99394032e-01 -8.65937293e-01 -1.87446013e-01 1.06195855e+00
-1.69479263e+00 -1.57674563e+00 -7.11339116e-01 6.04089916e-01
-1.67776849e-02 3.00874561e-01 7.66767323e-01 5.31417457e-03
-2.40848601e-01 4.91701633e-01 -1.40829206e-01 3.27595361e-02
5.96950054e-01 -1.39719737e+00 7.70306468e-01 1.05410242e+00
4.71010245e-02 5.70985436e-01 6.70799673e-01 -3.21588099e-01
-1.89067137e+00 -7.31675506e-01 4.18937713e-01 -6.38684213e-01
-1.20817021e-01 -7.71030724e-01 -5.22697866e-01 4.54260558e-01
3.75710092e-02 1.53836265e-01 2.58815467e-01 1.59146875e-01
-8.56137395e-01 -8.55327956e-03 -1.32495916e+00 7.91238070e-01
1.23814273e+00 -6.57454431e-01 -5.87600619e-02 2.85612315e-01
6.68522179e-01 -8.19750965e-01 -6.04291856e-01 -1.40511687e-03
1.01640320e+00 -1.59059978e+00 1.28141403e+00 -2.05936641e-01
6.44361794e-01 -3.64163965e-01 -3.44439566e-01 -1.25727129e+00
-1.67439669e-01 -6.49857342e-01 -1.51850119e-01 6.11758351e-01
6.45819604e-02 -2.97163129e-01 9.03893769e-01 1.17153549e+00
7.83822834e-02 -4.71026093e-01 -8.14912856e-01 -4.03097063e-01
2.99858116e-02 -4.02599186e-01 4.06052291e-01 5.80221355e-01
-5.29064715e-01 -4.23435085e-02 -7.88101017e-01 3.00272197e-01
1.13717055e+00 2.69765168e-01 1.12696779e+00 -1.31189358e+00
-6.35038495e-01 -2.39549473e-01 -2.80708194e-01 -1.06636727e+00
-1.86754882e-01 -3.78289908e-01 1.45187661e-01 -1.55128229e+00
1.15931235e-01 -2.23004669e-01 1.49215087e-01 4.39801037e-01
-9.93866846e-02 6.70436382e-01 4.17832255e-01 -4.97217067e-02
-3.32953215e-01 6.63272560e-01 1.40338564e+00 1.45882934e-01
-5.99151701e-02 -3.78095001e-01 -7.16294229e-01 5.48860312e-01
3.38301837e-01 4.10848297e-02 -2.39015400e-01 -9.53013361e-01
4.05321866e-01 1.15364812e-01 1.12667429e+00 -9.57697272e-01
-6.69956878e-02 -7.13890046e-02 1.10011530e+00 -3.52897286e-01
9.23972189e-01 -9.20160949e-01 3.19724143e-01 1.88945904e-01
-3.00446630e-01 -1.35442153e-01 2.37956077e-01 5.94730079e-01
5.04643857e-01 5.83680034e-01 8.78645539e-01 -2.98490912e-01
-3.66790682e-01 4.43499476e-01 3.57476711e-01 1.35988474e-01
4.96748149e-01 -3.24411601e-01 -8.68684426e-02 -6.83053315e-01
-3.59761238e-01 -1.33313850e-01 1.13568962e+00 1.62622273e-01
8.54219973e-01 -1.18560421e+00 -7.83246934e-01 3.92748356e-01
1.05213240e-01 1.42654538e-01 5.35675645e-01 4.17450786e-01
-1.06252038e+00 -7.92059079e-02 -2.55632818e-01 -6.15864217e-01
-1.17963517e+00 3.78066570e-01 8.02376270e-01 2.72675663e-01
-1.03601122e+00 8.04910243e-01 3.96338969e-01 -5.48132837e-01
5.05949259e-01 -3.62526834e-01 6.09799385e-01 -6.87393427e-01
4.12981719e-01 2.05125153e-01 2.58563757e-02 -6.96667910e-01
-2.38254145e-01 9.33065772e-01 5.08814037e-01 -4.54184175e-01
1.33695281e+00 -3.25842619e-01 3.48374918e-02 3.10881943e-01
1.16788805e+00 -1.46023603e-02 -1.98268807e+00 -1.31816417e-01
-1.05511212e+00 -8.21320355e-01 2.62652189e-01 -1.26334810e+00
-1.27620912e+00 1.14438391e+00 6.21428132e-01 -2.92712480e-01
1.20941281e+00 -3.21450979e-01 6.81488037e-01 3.45937520e-01
4.97900039e-01 -7.45607913e-01 -2.92514767e-02 6.25695735e-02
9.30812001e-01 -1.32231855e+00 3.76276821e-01 -6.09758139e-01
-5.73314965e-01 9.62928176e-01 5.96692026e-01 -2.12944910e-01
2.90488124e-01 -1.33927315e-01 3.08086514e-01 -2.04726860e-01
-4.32471931e-01 -2.35578418e-01 6.43281281e-01 8.76472056e-01
3.38244259e-01 4.87168171e-02 5.67138374e-01 -1.64520532e-01
-2.52474487e-01 -2.43963674e-01 4.17808563e-01 6.82318747e-01
3.30409035e-02 -6.27486110e-01 -3.65940958e-01 -2.63756186e-01
4.59609069e-02 -6.22821935e-02 -6.94668591e-01 9.08597410e-01
5.86052053e-02 8.07879567e-01 5.99290505e-02 -2.95483083e-01
5.74881673e-01 -3.32863957e-01 1.12227428e+00 -2.17204720e-01
-4.71513718e-01 2.01141551e-01 1.16768681e-01 -1.07388079e+00
-3.92115146e-01 -5.34599721e-01 -9.47573245e-01 -3.66616935e-01
1.69063687e-01 -5.28305650e-01 8.00577581e-01 7.61779308e-01
5.05686760e-01 2.74325490e-01 6.28556728e-01 -1.41508138e+00
-1.54401630e-01 -5.55910826e-01 -5.90253890e-01 8.22702825e-01
6.92659974e-01 -6.12235785e-01 -4.10225481e-01 -6.85881451e-02]
|
[9.492656707763672, -3.0228922367095947]
|
7699a2cd-72b4-4d0f-b0f8-bbb2713a8e6a
|
single-image-lens-flare-removal
|
2011.12485
| null |
https://arxiv.org/abs/2011.12485v4
|
https://arxiv.org/pdf/2011.12485v4.pdf
|
How to Train Neural Networks for Flare Removal
|
When a camera is pointed at a strong light source, the resulting photograph may contain lens flare artifacts. Flares appear in a wide variety of patterns (halos, streaks, color bleeding, haze, etc.) and this diversity in appearance makes flare removal challenging. Existing analytical solutions make strong assumptions about the artifact's geometry or brightness, and therefore only work well on a small subset of flares. Machine learning techniques have shown success in removing other types of artifacts, like reflections, but have not been widely applied to flare removal due to the lack of training data. To solve this problem, we explicitly model the optical causes of flare either empirically or using wave optics, and generate semi-synthetic pairs of flare-corrupted and clean images. This enables us to train neural networks to remove lens flare for the first time. Experiments show our data synthesis approach is critical for accurate flare removal, and that models trained with our technique generalize well to real lens flares across different scenes, lighting conditions, and cameras.
|
['Jonathan T. Barron', 'Ashok Veeraraghavan', 'Jiawen Chen', 'Rahul Garg', 'Tianfan Xue', 'Qiurui He', 'Yicheng Wu']
|
2020-11-25
| null |
http://openaccess.thecvf.com//content/ICCV2021/html/Wu_How_To_Train_Neural_Networks_for_Flare_Removal_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Wu_How_To_Train_Neural_Networks_for_Flare_Removal_ICCV_2021_paper.pdf
|
iccv-2021-1
|
['flare-removal']
|
['computer-vision']
|
[ 6.15045071e-01 -6.12589300e-01 5.76402903e-01 -1.57019421e-01
-3.28171998e-01 -1.10099363e+00 5.03237963e-01 -4.43498462e-01
3.15195054e-01 7.33002782e-01 2.49415889e-01 2.30969843e-02
-5.13040200e-02 -5.23815334e-01 -6.66900277e-01 -6.19082093e-01
2.13477686e-01 -7.13632256e-02 4.14694160e-01 -9.16687474e-02
3.78199369e-01 7.21578121e-01 -1.93050826e+00 -4.06198278e-02
9.34936523e-01 6.20691240e-01 1.07725002e-01 1.05702889e+00
1.34883314e-01 1.12077534e+00 -8.91524017e-01 -1.74230114e-01
4.35056716e-01 -6.90228164e-01 -8.25239420e-02 4.79688704e-01
1.21301734e+00 -5.79326928e-01 -2.51322418e-01 1.10068345e+00
3.35636258e-01 1.72131844e-02 5.24842918e-01 -8.88541877e-01
-7.32636750e-01 -1.64753661e-01 -7.07737803e-01 2.25367785e-01
5.96639514e-01 4.25958455e-01 4.11763906e-01 -6.88095212e-01
4.80232179e-01 9.04692769e-01 9.93867993e-01 4.79948223e-01
-1.16300011e+00 -6.04660749e-01 -3.22903156e-01 -1.71688050e-01
-9.71071184e-01 -6.83209240e-01 8.12395513e-01 -6.09911799e-01
9.08020854e-01 4.03632402e-01 8.73644292e-01 6.22877002e-01
5.11844933e-01 3.95209879e-01 1.43737721e+00 -5.12440860e-01
7.79349431e-02 3.39742526e-02 1.92127988e-01 4.94189858e-01
6.24638438e-01 1.64870337e-01 -2.67034501e-01 -1.19699270e-01
7.21846521e-01 -7.47703388e-02 -1.05461872e+00 -1.86071575e-01
-7.76373804e-01 4.94510233e-01 1.70976818e-01 8.18819627e-02
2.81122699e-02 2.31882989e-01 -3.55507016e-01 3.97262216e-01
3.87312472e-01 1.02306664e+00 -3.22742522e-01 1.17450319e-01
-9.41762269e-01 5.13898373e-01 5.67239761e-01 5.56277394e-01
8.18982899e-01 2.42077306e-01 4.10133421e-01 8.78096163e-01
8.10730681e-02 9.29030001e-01 9.81530920e-03 -1.18979633e+00
-1.46634087e-01 2.87418097e-01 5.96181989e-01 -9.88985598e-01
-4.51803178e-01 -2.06342548e-01 -5.46619296e-01 9.43183184e-01
4.66831625e-01 -3.91107142e-01 -1.28173149e+00 1.34504700e+00
-1.74224317e-01 2.89347649e-01 3.56152952e-02 1.25834370e+00
6.56145513e-01 6.06544733e-01 -6.18091464e-01 -5.50134540e-01
6.84806347e-01 -6.57610059e-01 -1.14059269e+00 -7.09060252e-01
1.69097066e-01 -1.31245863e+00 9.51152146e-01 7.68386543e-01
-1.46522605e+00 -2.45733529e-01 -1.05238557e+00 4.29977570e-03
-1.12637855e-01 -2.35558152e-01 6.29925370e-01 5.76857805e-01
-1.12585807e+00 3.96151572e-01 -5.07393956e-01 -2.07648486e-01
1.69169232e-02 9.28615183e-02 -6.45717159e-02 -2.97721148e-01
-5.78836024e-01 9.55671430e-01 -3.84661406e-01 2.48955488e-01
-3.42806876e-01 -6.85868025e-01 -7.05239773e-01 -1.35315612e-01
2.27534696e-01 -6.53296649e-01 1.18658566e+00 -1.41109371e+00
-1.27726865e+00 6.20201170e-01 -3.57307494e-01 -1.52775571e-01
1.68775097e-01 -3.32921952e-01 -5.79224110e-01 -1.80408023e-02
-4.62978214e-01 -2.11592056e-02 1.25743079e+00 -1.78140259e+00
-6.00730300e-01 -1.37798518e-01 1.71691537e-01 1.23674214e-01
3.00394028e-01 2.89832324e-01 -1.53830618e-01 -3.65786016e-01
-1.20768644e-01 -9.71639752e-01 9.86061618e-02 -1.77118361e-01
-1.08266540e-01 5.66962361e-01 8.17731619e-01 -5.33352137e-01
1.21066833e+00 -2.14225054e+00 -9.18226391e-02 -1.20565832e-01
1.12063564e-01 4.87738013e-01 7.23598599e-02 4.35392946e-01
-8.34881961e-02 -9.81597155e-02 -5.25295734e-01 -4.90970612e-02
-4.46564019e-01 2.09476069e-01 -4.15847927e-01 3.61842543e-01
2.91696548e-01 3.85888070e-01 -7.11406291e-01 1.35075778e-01
3.39621514e-01 5.05758345e-01 -3.52838993e-01 2.91298360e-01
-3.71179104e-01 2.87327915e-01 4.04978842e-02 4.41878229e-01
9.82006431e-01 9.94450524e-02 -2.60779798e-01 -1.08970918e-01
-4.81328636e-01 -1.35412663e-01 -1.25295889e+00 1.05673099e+00
-4.96188730e-01 1.17905426e+00 3.34319562e-01 8.84280801e-02
7.38331258e-01 4.36141081e-02 6.69491589e-02 -5.68564534e-01
3.25857885e-02 3.16846788e-01 -2.69439854e-02 -9.06796634e-01
4.12994951e-01 -4.56049293e-01 5.55484116e-01 3.10449809e-01
-4.69745517e-01 -5.99691093e-01 1.38535216e-01 -1.36497334e-01
1.31197155e+00 5.30733243e-02 -1.64140850e-01 -7.95317739e-02
1.55068204e-01 1.41358271e-01 4.78361547e-01 6.34129226e-01
2.15355366e-01 1.26158261e+00 -1.69342496e-02 -6.48389935e-01
-8.11969697e-01 -9.03594553e-01 -1.87546179e-01 4.81073022e-01
2.48671845e-01 -8.37796405e-02 -7.81183183e-01 1.83410440e-02
1.80698708e-02 9.16975617e-01 -1.92218527e-01 -1.90006867e-01
-6.64606512e-01 -9.17393506e-01 -1.38784051e-01 1.54531464e-01
3.08177531e-01 -1.04843676e+00 -1.00329363e+00 -1.22781973e-02
2.13209838e-01 -8.96411121e-01 -2.83773065e-01 -1.24000348e-01
-5.48415363e-01 -1.45441830e+00 -2.53358364e-01 -5.09959221e-01
6.16352201e-01 9.20934975e-01 1.33019137e+00 3.02964121e-01
-7.73646355e-01 5.77074170e-01 -2.81914175e-01 -6.44552767e-01
-3.62682104e-01 -9.52507436e-01 -8.27755854e-02 1.09524764e-01
2.85440922e-01 -4.85481471e-01 -6.88394308e-01 -2.60235090e-02
-1.12776041e+00 -5.09257838e-02 3.12972873e-01 4.29050148e-01
2.09016293e-01 5.32675147e-01 9.48466882e-02 -9.37024653e-01
6.92220032e-01 -1.72092319e-02 -1.01574373e+00 -1.26622751e-01
-3.73136222e-01 -2.96332061e-01 5.83639205e-01 -4.42620426e-01
-1.31596887e+00 -3.00325900e-02 2.37252176e-01 -5.33540308e-01
-4.60044652e-01 1.94830701e-01 -1.62118860e-02 -4.95567501e-01
9.59165394e-01 -1.94676921e-01 -1.33778751e-01 -6.47022724e-01
-1.30267935e-02 5.65019906e-01 7.22156107e-01 1.14402413e-01
1.12358451e+00 6.70668125e-01 1.22017324e-01 -1.18835509e+00
-1.10090280e+00 -3.22028041e-01 -2.26845041e-01 -3.00815225e-01
7.40728676e-01 -8.35907817e-01 -3.03676963e-01 8.68001997e-01
-9.31837559e-01 -5.74809313e-01 -8.06325600e-02 5.61071634e-01
-2.70448327e-01 1.84427530e-01 -4.02910888e-01 -1.03807783e+00
3.14577855e-02 -8.71548057e-01 9.11772549e-01 6.65537000e-01
2.44163692e-01 -8.12190890e-01 4.45236921e-01 2.81265140e-01
4.71338660e-01 4.61920947e-01 7.52450764e-01 5.62736332e-01
-9.32693481e-01 -2.82642514e-01 -1.70734331e-01 5.16511679e-01
4.96207267e-01 6.44746840e-01 -1.21532500e+00 -3.16992611e-01
2.83004314e-01 -8.95529389e-02 1.04787326e+00 7.45672882e-01
6.02884769e-01 -3.14374380e-02 -6.52570650e-02 8.60505581e-01
1.85835278e+00 3.36862445e-01 8.93781424e-01 3.46185923e-01
6.86219096e-01 5.27481914e-01 3.26098412e-01 1.44909710e-01
-2.09215000e-01 4.72965658e-01 6.28327012e-01 -5.78767061e-01
-6.41739666e-01 4.79486644e-01 2.04280362e-01 2.79275477e-01
-4.49637659e-02 -5.97052634e-01 -7.06722498e-01 5.51735222e-01
-1.50569963e+00 -1.18688309e+00 -7.21281588e-01 2.29943681e+00
6.69268787e-01 -5.02811410e-02 -1.82249621e-01 3.05794203e-03
3.77814680e-01 7.40268975e-02 -3.81788284e-01 -3.89766753e-01
-4.60435688e-01 1.54583395e-01 7.56059587e-01 1.01634407e+00
-9.26553428e-01 7.82038271e-01 7.05439949e+00 -2.27829784e-01
-1.43683612e+00 -2.90268391e-01 -1.81365654e-01 -5.20839572e-01
-5.07478058e-01 7.46316044e-03 -4.44147438e-01 3.12102646e-01
5.73538363e-01 1.21415623e-01 8.86519432e-01 -3.78598794e-02
4.68102217e-01 -4.92910206e-01 -7.04481602e-01 1.08916044e+00
5.88193655e-01 -9.94469285e-01 -2.51345605e-01 -2.66846597e-01
8.79996836e-01 7.99107999e-02 -1.78199410e-01 -4.00438279e-01
2.92861283e-01 -9.50962305e-01 4.12472636e-01 1.09913278e+00
5.10024965e-01 -6.35283589e-01 5.13515234e-01 -1.69292837e-03
-4.71545249e-01 -2.77006123e-02 -5.32891572e-01 -3.38490039e-01
2.80068874e-01 8.51592302e-01 -5.03373265e-01 9.09146965e-02
7.26677060e-01 5.72440624e-01 -7.84989715e-01 1.57847857e+00
-2.55526334e-01 6.23074353e-01 -5.28438330e-01 4.29444879e-01
-1.15177073e-01 -5.48430979e-01 6.78679585e-01 1.05001771e+00
4.54415888e-01 1.48770437e-01 -1.46740705e-01 6.02566183e-01
1.91734567e-01 -3.88671815e-01 -8.92729402e-01 1.20289810e-01
-1.56345386e-02 1.23803365e+00 -3.14504147e-01 2.51883745e-01
-7.92717278e-01 8.01049948e-01 -2.64633805e-01 7.07626462e-01
-5.09003818e-01 -4.41682756e-01 1.10892904e+00 5.09713650e-01
4.43937331e-02 -2.82866091e-01 -9.33202915e-03 -1.07778084e+00
2.26268128e-01 -8.74073982e-01 -1.20317854e-01 -1.65665305e+00
-1.16248298e+00 5.15252173e-01 -1.57769322e-01 -1.07811534e+00
-1.66905895e-01 -8.17543447e-01 -8.33224595e-01 8.81519377e-01
-1.87063920e+00 -1.06964171e+00 -6.28173828e-01 5.46261311e-01
4.06451970e-01 1.00703761e-01 7.08502650e-01 1.45976618e-01
-3.97236675e-01 -2.17017680e-01 4.59839284e-01 -3.42509478e-01
1.22277343e+00 -1.49655795e+00 1.11030191e-01 1.24592364e+00
-2.04053149e-02 4.02811199e-01 1.45811605e+00 -5.88575542e-01
-1.81709695e+00 -1.04478037e+00 4.85580295e-01 -6.83373511e-01
6.01913333e-01 -1.22412935e-01 -1.08450389e+00 8.17768037e-01
4.68794912e-01 -9.09853503e-02 3.49696457e-01 -8.20999742e-02
-2.02563345e-01 -1.91638008e-01 -7.45780230e-01 6.88791215e-01
7.72411644e-01 -2.75992274e-01 -6.00229204e-01 6.32397175e-01
2.71529794e-01 -3.76866192e-01 -3.64487976e-01 4.77116436e-01
4.01588231e-01 -1.61327207e+00 8.05816948e-01 -2.37033412e-01
3.47700149e-01 -6.12308443e-01 1.54205784e-01 -1.54979408e+00
-4.18925315e-01 -9.97744262e-01 1.68706980e-02 1.15642691e+00
3.11702788e-01 -5.03874600e-01 3.15177619e-01 5.02672076e-01
-3.90662283e-01 3.69382128e-02 -1.60930287e-02 -5.12033105e-01
-1.68422639e-01 -5.54690808e-02 2.90493250e-01 8.57563376e-01
-5.97815871e-01 2.90344566e-01 -5.77296853e-01 6.19879305e-01
7.77838647e-01 5.80930948e-01 1.06049764e+00 -1.58732414e+00
-3.16644818e-01 -3.11804444e-01 -5.50852977e-02 -5.02383471e-01
-7.35131055e-02 -2.99424022e-01 4.15257156e-01 -1.72358155e+00
-3.04319244e-02 -1.25591293e-01 1.99148521e-01 1.59377977e-01
-3.86377305e-01 6.55649066e-01 1.17283784e-01 3.89667213e-01
-2.29084510e-02 -7.16576278e-02 1.18098509e+00 2.02637911e-01
-5.27103424e-01 -1.10434175e-01 -5.79244196e-01 1.23947263e+00
7.05938101e-01 -9.17580649e-02 -5.43394506e-01 -9.62825418e-01
7.38927007e-01 -1.33291453e-01 4.18171942e-01 -1.26262128e+00
-1.45020969e-02 -2.85142213e-01 5.49185216e-01 -2.15320319e-01
4.29497749e-01 -8.85820508e-01 6.04885340e-01 1.74976252e-02
6.38387352e-02 -2.42173806e-01 3.85733634e-01 4.43479866e-01
-1.36689693e-01 -4.95524496e-01 1.01824403e+00 -1.84908852e-01
-3.46973777e-01 -9.72430408e-02 -5.56592286e-01 -4.85635214e-02
6.62048340e-01 -2.74591476e-01 -8.07110488e-01 -5.65878034e-01
-4.82317984e-01 -1.09061323e-01 1.25392616e+00 2.39015520e-01
4.70359147e-01 -8.13306987e-01 -6.52569294e-01 5.04218102e-01
-3.70685458e-02 -1.64495677e-01 6.71757907e-02 5.28673112e-01
-1.08834827e+00 2.14869124e-05 -4.46261577e-02 -2.48637319e-01
-1.48058486e+00 6.28355443e-01 5.96690178e-01 4.47579831e-01
-8.83249223e-01 7.40085959e-01 4.90866572e-01 3.59806210e-01
-4.89568748e-02 -1.92061797e-01 -1.54360786e-01 -1.89856008e-01
9.18748558e-01 1.86150968e-01 1.30275071e-01 -5.21931708e-01
1.07593156e-01 1.23668718e+00 2.15193480e-01 3.19940031e-01
1.27782738e+00 -3.30233455e-01 -2.55858660e-01 5.67199707e-01
6.68355525e-01 7.27620959e-01 -1.40773010e+00 5.81977591e-02
-5.88742018e-01 -9.06090915e-01 3.95825356e-01 -8.52399170e-01
-1.15888655e+00 7.74666429e-01 5.07040083e-01 5.66652954e-01
1.66768694e+00 -2.98007160e-01 7.16085553e-01 1.91130728e-01
1.25314981e-01 -5.57695866e-01 -1.81431100e-01 3.48354280e-01
8.61763835e-01 -9.19598877e-01 -3.04961465e-02 -5.23749352e-01
-3.65963995e-01 1.21241212e+00 3.16708118e-01 -3.15860242e-01
4.57069248e-01 7.00737476e-01 5.34610212e-01 -3.50152075e-01
-6.27567947e-01 -4.40053254e-01 1.56216964e-01 7.23381817e-01
2.75830388e-01 -4.48268026e-01 1.27450749e-02 -2.64857024e-01
-2.89193898e-01 1.15171164e-01 1.28209615e+00 1.02694905e+00
-8.40004683e-01 -7.00100243e-01 -8.11732888e-01 5.46728551e-01
-5.40198803e-01 -2.17207638e-03 -8.35001945e-01 5.51803231e-01
2.54025310e-01 1.21082497e+00 1.26599863e-01 1.23856723e-01
4.42635804e-01 6.48163036e-02 8.84035766e-01 -4.75891680e-01
-5.50939023e-01 6.18894100e-01 1.33414894e-01 -4.18189675e-01
-6.63312137e-01 -6.27877116e-01 -8.10054719e-01 -1.58804834e-01
-4.35525090e-01 -3.04293513e-01 5.20469546e-01 6.52251482e-01
8.18660632e-02 4.86666143e-01 5.03432274e-01 -8.55741978e-01
1.91639036e-01 -7.32170820e-01 -8.70718479e-01 7.72148609e-01
1.02229381e+00 -6.63635969e-01 -1.10769355e+00 5.13839602e-01]
|
[10.665538787841797, -2.977074384689331]
|
04175314-55a2-4061-8e52-db1543731489
|
dynamic-structural-brain-network-construction
|
2305.10077
| null |
https://arxiv.org/abs/2305.10077v1
|
https://arxiv.org/pdf/2305.10077v1.pdf
|
Dynamic Structural Brain Network Construction by Hierarchical Prototype Embedding GCN using T1-MRI
|
Constructing structural brain networks using T1-weighted magnetic resonance imaging (T1-MRI) presents a significant challenge due to the lack of direct regional connectivity information. Current methods with T1-MRI rely on predefined regions or isolated pretrained location modules to obtain atrophic regions, which neglects individual specificity. Besides, existing methods capture global structural context only on the whole-image-level, which weaken correlation between regions and the hierarchical distribution nature of brain connectivity.We hereby propose a novel dynamic structural brain network construction method based on T1-MRI, which can dynamically localize critical regions and constrain the hierarchical distribution among them for constructing dynamic structural brain network. Specifically, we first cluster spatially-correlated channel and generate several critical brain regions as prototypes. Further, we introduce a contrastive loss function to constrain the prototypes distribution, which embed the hierarchical brain semantic structure into the latent space. Self-attention and GCN are then used to dynamically construct hierarchical correlations of critical regions for brain network and explore the correlation, respectively. Our method is evaluated on ADNI-1 and ADNI-2 databases for mild cognitive impairment (MCI) conversion prediction, and acheive the state-of-the-art (SOTA) performance. Our source code is available at http://github.com/*******.
|
['Jian Zheng', 'Zheng Yanyan', 'Chen Bai', 'Wenju Cui', 'Yilin Leng']
|
2023-05-17
| null | null | null | null |
['specificity']
|
['natural-language-processing']
|
[ 5.66074206e-03 7.20008016e-02 -6.34590164e-02 -4.41320866e-01
-2.16051087e-01 -3.98620993e-01 4.44689006e-01 -3.17265451e-01
-2.94806004e-01 5.22741437e-01 4.43967879e-01 4.99103330e-02
-6.54457569e-01 -8.32441092e-01 -5.16371667e-01 -6.23046219e-01
-2.99170107e-01 7.00307012e-01 3.44067812e-01 3.28588076e-02
-9.45258960e-02 3.13636661e-01 -9.36470687e-01 3.21363509e-01
1.13291478e+00 8.04087460e-01 7.07811236e-01 -6.58125058e-02
-5.21291569e-02 4.08890337e-01 -2.36284986e-01 -1.09996974e-01
5.51377684e-02 -4.76633519e-01 -8.18518460e-01 -1.51909128e-01
1.63041249e-01 -6.63675070e-02 -4.73602682e-01 1.39425123e+00
5.13129234e-01 1.25269279e-01 5.81834137e-01 -9.53024805e-01
-9.38588679e-01 8.81879449e-01 -5.19201159e-01 9.14049983e-01
-1.93286389e-01 1.57151192e-01 1.01870370e+00 -8.90963733e-01
6.44137919e-01 1.24622166e+00 5.00151038e-01 3.31727862e-01
-1.34465230e+00 -9.30889070e-01 4.81360108e-01 6.72009051e-01
-1.54596663e+00 -8.99020955e-02 9.16804731e-01 -5.92871010e-01
8.61123681e-01 -2.40533218e-01 1.19942224e+00 1.11774504e+00
4.42519516e-01 4.02763188e-01 1.37681627e+00 1.91941381e-01
-3.06087937e-02 -4.98174787e-01 3.70442837e-01 7.00834811e-01
1.19474074e-02 2.75163203e-02 -1.68589838e-02 5.16933203e-02
1.08461893e+00 2.95970678e-01 -3.40849340e-01 -3.10836613e-01
-1.62769341e+00 7.50676870e-01 1.22156096e+00 7.56773412e-01
-6.03182375e-01 -1.91915318e-01 3.44338030e-01 1.38358369e-01
2.58418381e-01 6.00087233e-02 -1.73685879e-01 5.07848322e-01
-1.01895940e+00 -2.40685299e-01 7.55124865e-03 6.62451625e-01
6.07115626e-01 -8.63914490e-02 -2.72288203e-01 1.19545341e+00
3.70021611e-01 1.48770213e-01 9.17294979e-01 -8.47270608e-01
4.63833392e-01 6.41569197e-01 -7.40030527e-01 -1.06636214e+00
-7.57623792e-01 -8.55885386e-01 -1.22535634e+00 -2.43416667e-01
-9.06058922e-02 2.40228977e-03 -9.22723651e-01 1.92145824e+00
6.45733699e-02 1.94009945e-01 -4.03886497e-01 1.08352149e+00
6.80479825e-01 2.59092659e-01 1.96960583e-01 -2.81426430e-01
1.36253643e+00 -1.13608301e+00 -6.14738464e-01 -2.27370709e-01
3.32681090e-01 -9.90045071e-02 1.22892940e+00 1.93124749e-02
-9.85465109e-01 -5.47912300e-01 -7.33531177e-01 1.76112838e-02
-2.18362823e-01 -1.26717478e-01 6.00541234e-01 3.73893023e-01
-1.51846743e+00 2.67067015e-01 -1.05630767e+00 -1.71895891e-01
7.60145843e-01 4.46742654e-01 -4.95198250e-01 -1.86983809e-01
-1.40799737e+00 7.81168878e-01 4.93639559e-01 2.83779085e-01
-1.01391125e+00 -9.24807549e-01 -4.39664215e-01 1.70482829e-01
2.28418723e-01 -9.11721706e-01 3.78433824e-01 -9.29751039e-01
-1.23150897e+00 7.75335431e-01 4.41664234e-02 -2.31668919e-01
4.10758615e-01 1.53460503e-01 -4.83758599e-01 6.02725923e-01
4.10466880e-01 1.04996693e+00 6.66722953e-01 -9.82401788e-01
-8.98910239e-02 -6.68513179e-01 -1.36664525e-01 2.71306753e-01
-4.18841422e-01 7.62183666e-02 -2.88213670e-01 -7.57172108e-01
4.94821310e-01 -8.27970266e-01 -3.58650327e-01 -1.56751245e-01
-4.55597937e-01 -9.44776833e-02 6.08751118e-01 -1.00712657e+00
1.01416266e+00 -1.85643232e+00 3.82809520e-01 6.42976224e-01
7.05796540e-01 -2.64933228e-01 -3.08866382e-01 -5.11409521e-01
-6.08041286e-01 1.41332179e-01 -5.71081281e-01 1.77758604e-01
-2.11747438e-01 5.15989400e-02 2.58376688e-01 3.94457579e-01
-1.46795437e-01 1.25068545e+00 -8.30591679e-01 -5.44582367e-01
4.05537039e-02 6.53014958e-01 -8.33711624e-01 -2.37664729e-02
4.10645604e-01 8.63742530e-01 -3.68330002e-01 3.51035506e-01
6.00007534e-01 -3.65340412e-01 3.62863302e-01 -4.97656435e-01
-1.53612271e-02 1.32090122e-01 -6.01507187e-01 1.84114504e+00
-1.82936504e-01 1.08587436e-01 1.49467021e-01 -1.46735132e+00
8.81483734e-01 1.86588466e-01 7.75804639e-01 -8.50808620e-01
2.20680133e-01 -4.75134403e-02 6.69650674e-01 -3.17788124e-01
-5.37173152e-01 -1.85424775e-01 2.83544123e-01 3.94093454e-01
2.87426323e-01 4.70822960e-01 -3.71279903e-02 1.10229559e-01
1.20850599e+00 -2.40240380e-01 -2.62294970e-02 -6.86531067e-01
5.16033649e-01 -4.48339254e-01 6.52670145e-01 4.97728914e-01
-4.15953189e-01 5.48791707e-01 5.87551653e-01 -1.84404820e-01
-8.99933994e-01 -1.48914504e+00 -2.24935040e-01 8.65103781e-01
-1.05015203e-01 1.92180816e-02 -9.14769411e-01 -6.51273251e-01
-2.62811512e-01 4.19875085e-01 -8.53621423e-01 -3.93164426e-01
-6.73279166e-01 -1.16288066e+00 3.43142897e-01 5.91893792e-01
8.68150711e-01 -1.12727654e+00 -1.11665025e-01 1.76594108e-01
-5.02172112e-01 -9.11983848e-01 -7.11829007e-01 -5.96154900e-03
-1.36147821e+00 -1.02080023e+00 -9.25472617e-01 -1.00911200e+00
9.35263216e-01 3.00805598e-01 9.24152851e-01 1.61354855e-01
-3.45926911e-01 3.26304644e-01 -1.43796846e-01 4.59298611e-01
1.57172084e-01 3.22057039e-01 -3.48357968e-02 8.46651271e-02
7.13103637e-02 -1.26107430e+00 -1.07911861e+00 5.46269596e-01
-6.60049975e-01 1.16736121e-01 7.97550857e-01 8.40221405e-01
8.01799536e-01 1.58330351e-01 6.87080264e-01 -6.92141950e-01
6.46113873e-01 -8.34705114e-01 -2.37376943e-01 3.64725173e-01
-6.71719491e-01 -1.56531140e-01 4.50852573e-01 -4.77902234e-01
-1.06345522e+00 -1.04966030e-01 -6.25060275e-02 -4.68896598e-01
-1.63245291e-01 7.30219126e-01 -5.60393274e-01 6.05055504e-02
3.70899022e-01 4.80186462e-01 -6.13253340e-02 -5.40558577e-01
3.67700726e-01 6.45680800e-02 5.93657315e-01 -5.31509995e-01
4.94789243e-01 5.09032905e-01 -1.96893632e-01 -4.02154118e-01
-6.79877400e-01 -2.41593644e-01 -1.31838810e+00 -2.83607334e-01
9.73433673e-01 -8.20631623e-01 -4.00127083e-01 7.97687843e-02
-9.01859462e-01 -5.04887283e-01 1.05951644e-01 6.72343433e-01
-2.29028478e-01 4.29253340e-01 -7.08151996e-01 1.27597243e-01
-4.23810959e-01 -1.25326097e+00 4.31225926e-01 -8.68690386e-02
4.57218625e-02 -1.07664859e+00 -7.14049190e-02 5.33975720e-01
4.88319725e-01 -2.09559631e-02 1.36729801e+00 -2.26360232e-01
-4.59860265e-01 1.39125302e-01 -6.78133011e-01 2.74921149e-01
8.59746709e-02 -5.54188132e-01 -4.42861348e-01 -2.59837866e-01
4.50253263e-02 2.01464906e-01 1.08988452e+00 7.07511842e-01
1.55256927e+00 -3.65280271e-01 -3.59058142e-01 7.33757496e-01
1.07667518e+00 1.37621328e-01 6.86339021e-01 2.54102260e-01
1.05151939e+00 7.88981736e-01 -2.68986076e-01 6.24969639e-02
8.13345909e-01 4.98326242e-01 1.20738253e-01 -2.52179150e-02
-3.59516054e-01 7.74874091e-02 2.73431689e-01 1.30681121e+00
-4.78626370e-01 4.36054409e-01 -1.12960386e+00 6.09545112e-01
-1.68550229e+00 -1.01292551e+00 -4.07412201e-02 1.86902690e+00
9.02687192e-01 6.27901480e-02 1.09262645e-01 -3.07972670e-01
1.17085063e+00 1.03787743e-02 -7.52615571e-01 3.95861804e-01
-2.27879912e-01 2.20046520e-01 1.88075706e-01 4.60501999e-01
-8.65056515e-01 9.05164003e-01 5.85191584e+00 6.64456844e-01
-1.13885057e+00 9.29050386e-01 7.93662012e-01 -3.18876594e-01
-3.27650905e-01 -1.86462685e-01 -3.25799286e-01 6.44237161e-01
8.44927609e-01 -7.37553686e-02 7.25367367e-01 6.06584370e-01
3.15667719e-01 3.03918272e-01 -6.96792543e-01 1.00060725e+00
1.45816170e-02 -1.19833302e+00 2.46303111e-01 1.06405184e-01
6.43913031e-01 4.02706206e-01 1.07417233e-01 2.19676733e-01
1.35442168e-01 -1.03061640e+00 6.36284113e-01 8.92166257e-01
6.90433025e-01 -6.94383383e-01 4.40906733e-01 -2.91686691e-02
-1.34164453e+00 -2.54942149e-01 -4.65028524e-01 3.16367418e-01
5.35182245e-02 7.63530672e-01 -2.53826201e-01 3.01284939e-01
9.82795715e-01 9.13383603e-01 -7.54193902e-01 1.09286046e+00
-4.13067937e-01 6.11630142e-01 -1.25991136e-01 3.65842104e-01
1.39281452e-01 -3.99671733e-01 3.66970569e-01 1.05122221e+00
4.31039661e-01 2.88426816e-01 3.67409885e-01 1.33257496e+00
-1.42433003e-01 1.92046985e-01 -1.99932441e-01 3.93444955e-01
3.25015783e-01 1.37228525e+00 -1.26758564e+00 -3.09874028e-01
-3.85072917e-01 9.24871564e-01 5.44769526e-01 5.11710167e-01
-8.74415934e-01 -9.66732949e-02 1.33655429e-01 3.52045625e-01
7.46774673e-02 -4.74042654e-01 -4.74727809e-01 -1.34193695e+00
-4.47126366e-02 -5.00170767e-01 4.79028463e-01 -7.58651614e-01
-1.44505739e+00 9.32018280e-01 1.62165537e-01 -7.82346845e-01
3.52685034e-01 -1.95941254e-01 -6.82911634e-01 7.58789062e-01
-1.43698049e+00 -1.03629017e+00 -4.80047226e-01 1.06111979e+00
1.40621081e-01 -2.89420158e-01 5.51525474e-01 6.52945459e-01
-8.22269976e-01 3.92133772e-01 -5.69275320e-02 3.47585618e-01
4.28390115e-01 -8.43832254e-01 9.30974446e-03 7.32762933e-01
-1.62138239e-01 9.48450446e-01 -7.04642981e-02 -1.00465298e+00
-6.44264817e-01 -1.29035497e+00 4.84276444e-01 -2.68253326e-01
9.92741406e-01 -5.46935618e-01 -1.15344107e+00 6.95531905e-01
2.61494238e-02 2.52405733e-01 5.74825585e-01 5.92355430e-02
-4.27492529e-01 -1.44858912e-01 -1.00509536e+00 5.17188132e-01
1.54278898e+00 -7.32059658e-01 -7.45315313e-01 5.17571390e-01
6.73815787e-01 2.10025609e-01 -1.18902385e+00 3.69028866e-01
4.09177005e-01 -6.69613361e-01 1.25064278e+00 -2.54229099e-01
3.26568037e-01 -1.06766418e-01 1.38463438e-01 -1.37563050e+00
-1.02153683e+00 -1.12503141e-01 2.14180738e-01 9.96922791e-01
4.18004751e-01 -7.71164179e-01 6.83273315e-01 6.05278373e-01
-3.42926770e-01 -5.89892566e-01 -1.06469882e+00 -8.35448980e-01
3.68168980e-01 -1.34051964e-01 5.14528513e-01 1.31347525e+00
-1.53127953e-01 3.55203867e-01 -3.98499286e-03 1.17574342e-01
8.96869183e-01 -3.84947769e-02 -2.48247653e-01 -1.29395473e+00
-8.53748992e-02 -8.28531146e-01 -3.27615887e-01 -5.28604984e-01
4.21827585e-01 -1.66972780e+00 -3.38143200e-01 -1.67738366e+00
6.72934949e-01 -4.52978075e-01 -9.34960842e-01 7.98890889e-01
-3.53768580e-02 3.08582962e-01 -4.29288559e-02 5.79760730e-01
-5.74369311e-01 8.44039917e-01 1.52331829e+00 -2.94890910e-01
-2.28276312e-01 -5.54160237e-01 -6.61157310e-01 5.86402297e-01
1.05197036e+00 -5.04083157e-01 -6.77267671e-01 -4.92937535e-01
-2.10207105e-01 -1.40951216e-01 8.47810090e-01 -1.18682921e+00
1.34181470e-01 6.61331862e-02 8.07170093e-01 -2.63597161e-01
-3.02515235e-02 -7.29732037e-01 5.66047356e-02 5.25948703e-01
-3.74726057e-01 2.25621179e-01 -1.81789950e-01 3.55873436e-01
-2.02379823e-01 8.49109665e-02 1.07170415e+00 -4.90908116e-01
-5.09898603e-01 9.24068153e-01 -3.67593139e-01 9.35437083e-02
7.36545265e-01 -1.25872806e-01 -2.58635789e-01 1.37341306e-01
-1.23731816e+00 1.06360264e-01 -1.23694800e-01 4.36169386e-01
8.03258896e-01 -1.53959429e+00 -6.01880252e-01 1.73932210e-01
-3.07064950e-01 -4.06723619e-01 7.83568144e-01 1.50429761e+00
-1.69090211e-01 4.19821799e-01 -9.05893087e-01 -6.28668308e-01
-9.48505402e-01 5.53951740e-01 5.48768103e-01 -2.50748396e-01
-1.08693564e+00 7.04715669e-01 7.45179057e-01 -5.02439559e-01
-9.35749859e-02 -2.72532731e-01 -5.05879462e-01 2.22369641e-01
4.26629454e-01 4.04534861e-02 -3.37675475e-02 -8.10857236e-01
-5.27272105e-01 4.43359941e-01 -2.71520376e-01 -4.92447764e-02
1.57781935e+00 -3.53753597e-01 -6.72702134e-01 1.00270070e-01
1.31565285e+00 -6.35684967e-01 -1.12837720e+00 -4.65837866e-01
8.08804575e-03 -1.34907380e-01 4.48562115e-01 -8.23006213e-01
-1.75417686e+00 1.08844757e+00 1.02331698e+00 -3.26377004e-01
1.14567804e+00 2.66936332e-01 7.91726172e-01 1.97529092e-01
2.89974600e-01 -9.38647151e-01 2.63380200e-01 5.77835202e-01
1.14995885e+00 -8.36708844e-01 -2.48907283e-01 -2.50621080e-01
-5.30103922e-01 8.64548385e-01 7.51294553e-01 -3.38304967e-01
1.17737901e+00 -2.45307828e-03 -2.77046472e-01 -5.45821667e-01
-3.20247084e-01 -6.32080436e-02 3.99999201e-01 6.21921301e-01
3.55617255e-01 3.48347843e-01 -2.52456427e-01 1.15486014e+00
-3.49908203e-01 -1.95221424e-01 -1.30577415e-01 3.24355721e-01
-4.19026256e-01 -8.43775570e-01 -2.99656034e-01 8.36986065e-01
-2.40897968e-01 -4.17368472e-01 -2.04381451e-01 4.58036125e-01
3.10468853e-01 4.13403392e-01 1.06888689e-01 -3.28727841e-01
2.38577455e-01 1.01143114e-01 4.94799793e-01 -5.75057864e-01
-4.84854877e-01 3.28716457e-01 -3.72749060e-01 -5.46647131e-01
-3.67084742e-01 -8.39921474e-01 -1.44168758e+00 -9.30891782e-02
1.57589540e-01 -1.54561922e-02 2.19585642e-01 8.84274542e-01
5.02878487e-01 8.09242785e-01 4.45363998e-01 -8.67801964e-01
1.88041016e-01 -1.03250194e+00 -6.19925678e-01 3.22102576e-01
-1.59009263e-01 -1.04343247e+00 1.71958413e-02 3.00576091e-02]
|
[12.467683792114258, 3.3576953411102295]
|
25998e78-c4d8-4661-b733-f4dd050c9d0c
|
facial-expression-video-generation-based-on-1
|
2210.11182
| null |
https://arxiv.org/abs/2210.11182v1
|
https://arxiv.org/pdf/2210.11182v1.pdf
|
Facial Expression Video Generation Based-On Spatio-temporal Convolutional GAN: FEV-GAN
|
Facial expression generation has always been an intriguing task for scientists and researchers all over the globe. In this context, we present our novel approach for generating videos of the six basic facial expressions. Starting from a single neutral facial image and a label indicating the desired facial expression, we aim to synthesize a video of the given identity performing the specified facial expression. Our approach, referred to as FEV-GAN (Facial Expression Video GAN), is based on Spatio-temporal Convolutional GANs, that are known to model both content and motion in the same network. Previous methods based on such a network have shown a good ability to generate coherent videos with smooth temporal evolution. However, they still suffer from low image quality and low identity preservation capability. In this work, we address this problem by using a generator composed of two image encoders. The first one is pre-trained for facial identity feature extraction and the second for spatial feature extraction. We have qualitatively and quantitatively evaluated our model on two international facial expression benchmark databases: MUG and Oulu-CASIA NIR&VIS. The experimental results analysis demonstrates the effectiveness of our approach in generating videos of the six basic facial expressions while preserving the input identity. The analysis also proves that the use of both identity and spatial features enhances the decoder ability to better preserve the identity and generate high-quality videos. The code and the pre-trained model will soon be made publicly available.
|
['Lahoucine Ballihi', 'Hamza Bouzid']
|
2022-10-20
|
facial-expression-video-generation-based-on
|
https://www.sciencedirect.com/science/article/pii/S266730532200076X
|
https://doi.org/10.1016/j.iswa.2022.200139
|
intelligent-systems-with-applications-2022-11
|
['video-generation', 'facial-expression-generation']
|
['computer-vision', 'computer-vision']
|
[ 3.38007540e-01 1.00286298e-01 1.03914939e-01 -3.37970674e-01
-4.47887868e-01 -3.59490305e-01 6.65729403e-01 -7.95310974e-01
-8.84310976e-02 9.28322077e-01 1.64283112e-01 2.59986728e-01
2.99006581e-01 -6.69867337e-01 -7.10515618e-01 -1.14856279e+00
4.57205027e-02 -2.70621255e-02 -3.15820336e-01 -2.62738347e-01
-9.77898166e-02 7.08116710e-01 -1.81132519e+00 2.78094053e-01
5.89779675e-01 1.17493761e+00 -1.12465106e-01 5.32771647e-01
1.22528963e-01 1.14653695e+00 -6.15707457e-01 -7.11528480e-01
2.42923692e-01 -9.79211390e-01 -7.43435144e-01 3.19657832e-01
3.57290268e-01 -2.87926018e-01 -2.08601475e-01 1.01644671e+00
6.73082411e-01 7.75544811e-03 6.20721281e-01 -1.46248281e+00
-5.71182311e-01 8.83091465e-02 -6.47834241e-01 -1.99553519e-01
5.20005167e-01 1.95116907e-01 5.48714280e-01 -6.49619401e-01
1.01364660e+00 1.13522732e+00 4.72000510e-01 1.07297325e+00
-1.09994912e+00 -8.89636099e-01 -3.04519653e-01 2.15904161e-01
-1.59159994e+00 -7.17647672e-01 9.81400430e-01 -3.70083690e-01
3.75274628e-01 1.80263460e-01 8.29091966e-01 1.37914419e+00
5.94868213e-02 6.10495508e-01 1.26244187e+00 -4.84367013e-01
1.12963051e-01 1.28422543e-01 -7.38250434e-01 6.50277317e-01
-3.20280075e-01 2.34150708e-01 -6.16971016e-01 1.16397679e-01
8.68212819e-01 -3.77942950e-01 -4.31993783e-01 -2.12409362e-01
-7.44365215e-01 8.28989387e-01 2.45974749e-01 5.41752994e-01
-7.07693040e-01 3.53691012e-01 4.12427694e-01 2.68549055e-01
5.30672312e-01 7.47483000e-02 2.57724077e-01 -2.69334435e-01
-1.19477737e+00 2.88481593e-01 5.70539355e-01 7.90264845e-01
6.77875936e-01 5.28525293e-01 -3.44958425e-01 7.33332872e-01
9.23140869e-02 4.54542071e-01 3.45491230e-01 -1.13288462e+00
-1.36435121e-01 3.33135903e-01 1.00922778e-01 -1.20712781e+00
-1.49495024e-02 -1.96316987e-01 -9.42754626e-01 5.51105499e-01
1.10077575e-01 -3.95846784e-01 -9.73291636e-01 2.12629414e+00
3.91088843e-01 4.60141033e-01 2.08793849e-01 9.22020555e-01
7.81612098e-01 8.19868922e-01 1.09453492e-01 -3.50074261e-01
1.26541305e+00 -8.23043525e-01 -1.12611496e+00 3.90064418e-01
2.77905554e-01 -7.64502347e-01 6.78776503e-01 2.73026347e-01
-1.37734842e+00 -5.87724745e-01 -9.65415180e-01 1.66785926e-01
-2.45512407e-02 3.79829228e-01 4.25887704e-01 8.25111091e-01
-1.43066168e+00 3.62782776e-01 -5.37818491e-01 -4.01023775e-01
6.28572524e-01 4.53866333e-01 -7.18946040e-01 1.16496935e-01
-1.13647974e+00 7.82591701e-01 8.26839358e-02 2.48525932e-01
-9.50103879e-01 -4.21452671e-01 -7.93979406e-01 2.32093390e-02
-1.44398019e-01 -6.31351769e-01 1.10118163e+00 -2.04651165e+00
-2.08409309e+00 1.06650066e+00 -3.47183526e-01 -2.91002691e-01
5.88799059e-01 3.84131745e-02 -5.52158296e-01 3.72241050e-01
-9.11500826e-02 9.89942789e-01 1.10556197e+00 -1.37375772e+00
-4.03431326e-01 -2.97392190e-01 -1.22758806e-01 7.66517594e-02
-3.83076698e-01 3.40912879e-01 -4.97815311e-01 -7.36424267e-01
-4.02483553e-01 -1.00169265e+00 1.26601517e-01 1.89446677e-02
-1.59776479e-01 1.73347488e-01 1.07428598e+00 -7.75840521e-01
9.88034964e-01 -2.15225410e+00 2.71920890e-01 2.53922433e-01
-1.70169339e-01 3.54579359e-01 -2.01727986e-01 3.69340062e-01
-2.27362022e-01 -8.52948949e-02 -2.03406110e-01 -4.11544114e-01
-2.94881493e-01 1.33206561e-01 -1.69722110e-01 5.95769644e-01
4.00032610e-01 8.87898982e-01 -7.54542172e-01 -5.07779002e-01
5.61325699e-02 9.78270650e-01 -5.12623250e-01 3.94937932e-01
-6.60473630e-02 9.30308402e-01 -2.34551072e-01 6.94170833e-01
8.49371612e-01 1.50880367e-01 1.24945946e-01 -3.18506777e-01
3.14530879e-02 -4.51980740e-01 -8.92451108e-01 1.66819930e+00
-4.42756861e-01 7.41061807e-01 3.01671773e-01 -8.00462723e-01
1.04980290e+00 7.57037759e-01 7.50122547e-01 -8.50217342e-01
3.93060654e-01 3.33646268e-01 -9.81978625e-02 -6.25791490e-01
3.27186793e-01 -4.21791464e-01 1.92632020e-01 2.77957618e-01
3.59687805e-01 -6.74839392e-02 2.34089181e-01 -1.27757028e-01
7.07762897e-01 5.03563344e-01 7.85184428e-02 -1.40898362e-01
8.86855721e-01 -3.46983999e-01 4.70372826e-01 1.00006066e-01
-4.86721881e-02 7.13777065e-01 6.66967452e-01 -3.42538297e-01
-1.22935438e+00 -7.81379282e-01 1.40952498e-01 6.61442101e-01
-1.23959057e-01 -1.81840047e-01 -1.12148714e+00 -5.03758311e-01
-4.69701052e-01 4.18999374e-01 -8.52895379e-01 -1.43034711e-01
-5.56403875e-01 -4.11935091e-01 7.57332623e-01 3.10713053e-01
6.92456901e-01 -1.31287396e+00 -5.32470822e-01 2.55396347e-02
-3.22929114e-01 -1.24277174e+00 -2.87587404e-01 -4.25451577e-01
-3.67459267e-01 -6.74209595e-01 -1.13751173e+00 -7.87291408e-01
8.44203651e-01 -2.66043335e-01 8.68151486e-01 2.37858780e-02
-2.48867244e-01 2.72661924e-01 -4.57439721e-01 -3.43998790e-01
-6.42226577e-01 -2.39901274e-01 -8.84035230e-02 8.15905869e-01
8.20081607e-02 -7.62427509e-01 -6.58294320e-01 2.16518506e-01
-1.10608733e+00 1.29562542e-01 5.26641548e-01 9.70211804e-01
4.12254184e-01 -1.71962827e-01 6.57104731e-01 -6.82328820e-01
4.14034247e-01 -3.55903476e-01 -5.05402386e-01 1.50635643e-02
-1.30425096e-01 -9.41282511e-02 6.27226949e-01 -2.46334389e-01
-1.26579404e+00 4.68909532e-01 -5.12250781e-01 -7.49975979e-01
-1.61052912e-01 1.23908781e-01 -3.17204565e-01 -2.71229237e-01
5.37526190e-01 4.35750514e-01 4.22938973e-01 -3.94872464e-02
3.32577884e-01 5.97353637e-01 6.98906839e-01 -4.96596485e-01
6.71394348e-01 7.47116387e-01 2.10178569e-01 -7.96445370e-01
-3.47474992e-01 -4.28384766e-02 -5.24881721e-01 -7.25796700e-01
1.08307505e+00 -9.12461162e-01 -8.30100417e-01 8.02800953e-01
-1.19489431e+00 -2.01900631e-01 -2.10737497e-01 2.14126319e-01
-9.53796387e-01 1.05728656e-01 -3.99016351e-01 -9.15058017e-01
-4.72185284e-01 -1.14960563e+00 1.19424427e+00 3.61339897e-01
-1.56864986e-01 -8.61481011e-01 1.58486053e-01 5.52874506e-02
6.82203114e-01 8.45221817e-01 5.25553763e-01 6.04412481e-02
-5.06694555e-01 -1.24527728e-02 -7.04517141e-02 5.19595981e-01
3.57367992e-01 2.15274975e-01 -1.21999025e+00 -2.52512693e-01
-1.64762717e-02 -3.64995182e-01 5.71624994e-01 2.84976155e-01
1.02996194e+00 -4.39512581e-01 -7.40559958e-03 8.32341850e-01
1.46059799e+00 3.76707107e-01 1.08937657e+00 8.28872174e-02
4.74902421e-01 8.43366325e-01 5.16775548e-01 4.52393860e-01
-1.06617145e-01 1.07982230e+00 4.59695727e-01 -4.67265964e-01
-2.65937269e-01 -2.32877612e-01 5.78242123e-01 3.63655508e-01
-5.96237361e-01 -2.92927116e-01 -4.67158407e-01 5.54664195e-01
-1.68922651e+00 -1.37626946e+00 2.36932054e-01 1.84238505e+00
7.81534910e-01 -5.90339065e-01 2.16007158e-01 5.04671708e-02
6.18621290e-01 5.09366132e-02 -2.89027482e-01 -5.34174323e-01
-3.63503635e-01 5.70624113e-01 1.01218641e-01 3.78746301e-01
-8.72913599e-01 9.96669590e-01 5.81642008e+00 7.95574784e-01
-1.71943414e+00 1.30284697e-01 8.24930370e-01 -2.35743925e-01
-1.47865102e-01 -2.19637066e-01 -4.69855577e-01 5.39182246e-01
1.01781034e+00 -2.15432987e-01 2.43894786e-01 7.65997171e-01
4.92717832e-01 5.62892891e-02 -6.36249959e-01 1.11670828e+00
3.69565040e-01 -1.43162298e+00 3.43788743e-01 -2.17132680e-02
9.88884389e-01 -6.13475263e-01 1.36885464e-01 -6.25055954e-02
-9.11562592e-02 -1.40689027e+00 8.95196795e-01 7.26926148e-01
1.10108936e+00 -1.02693963e+00 7.89526343e-01 -9.19582471e-02
-9.75523591e-01 1.12114228e-01 -3.72800394e-03 1.96387574e-01
4.29603696e-01 1.62371919e-01 -5.15459359e-01 6.71752512e-01
5.85094094e-01 7.39229739e-01 -2.50885874e-01 5.86829245e-01
-3.07803631e-01 4.65876490e-01 1.67592545e-03 9.65186581e-02
2.05169320e-01 -3.64778012e-01 3.87166709e-01 1.10762477e+00
6.90607548e-01 4.50966991e-02 -4.25185531e-01 9.80517507e-01
-6.55066967e-02 3.17679316e-01 -8.26485813e-01 -4.71693128e-02
-1.91979464e-02 1.31434882e+00 -4.22225982e-01 -2.13144526e-01
-3.24096918e-01 1.40880096e+00 3.74319479e-02 3.03652704e-01
-1.10090816e+00 -1.47204429e-01 8.22069705e-01 2.25348353e-01
3.30011368e-01 3.80795263e-02 1.53572947e-01 -9.25115883e-01
1.97513271e-02 -1.05043244e+00 -8.73390469e-04 -9.97372448e-01
-8.19151342e-01 1.09613383e+00 -1.70807913e-01 -1.17901182e+00
-6.93175793e-01 -5.28243124e-01 -5.14634788e-01 9.29241598e-01
-1.40238786e+00 -1.56140649e+00 -6.42503738e-01 9.53563452e-01
2.19143689e-01 -2.61370718e-01 9.17229116e-01 5.64864933e-01
-4.72410381e-01 7.28351176e-01 -1.63449541e-01 1.32903114e-01
5.52390337e-01 -6.78349316e-01 -9.58689749e-02 7.41455376e-01
7.64811337e-02 3.21323305e-01 7.37724602e-01 -3.22462797e-01
-1.22166705e+00 -1.10459197e+00 8.21258008e-01 1.94494035e-02
1.76734999e-01 -3.56613785e-01 -5.93129158e-01 5.95150590e-01
6.35871828e-01 1.73311494e-02 6.10760152e-01 -7.02029288e-01
4.71428186e-02 -2.62149394e-01 -1.31662452e+00 5.70546389e-01
9.27755952e-01 -3.72396260e-01 6.84199110e-02 1.04405046e-01
1.40826896e-01 -3.53686661e-01 -7.86621869e-01 4.89981502e-01
7.35110104e-01 -1.27830410e+00 5.61541736e-01 -3.85752052e-01
6.99696839e-01 -3.86881649e-01 -1.31994396e-01 -1.20757663e+00
8.23066384e-03 -9.48912621e-01 2.30232447e-01 1.41418886e+00
9.70848277e-02 -3.69495332e-01 9.04337406e-01 2.15909734e-01
1.17155187e-01 -7.05888093e-01 -8.52406979e-01 -6.96711779e-01
-1.69427767e-01 -3.79511826e-02 6.59023225e-01 8.11711967e-01
-3.57642412e-01 1.78570017e-01 -8.84854078e-01 -1.49651989e-01
3.04519504e-01 2.80364268e-02 1.02546096e+00 -7.13347733e-01
1.76969487e-02 -4.47240323e-01 -8.24634671e-01 -5.21092892e-01
5.31529844e-01 -6.12000704e-01 -1.06371090e-01 -1.09462667e+00
8.51932317e-02 -7.51854703e-02 9.72313359e-02 4.49731946e-01
1.88616648e-01 7.20976353e-01 1.04018278e-01 -5.07607274e-02
-9.58962180e-03 7.68865347e-01 1.36608386e+00 9.93379578e-02
7.47224549e-03 -1.08414881e-01 -4.93379742e-01 4.71334875e-01
6.35545433e-01 -2.27272958e-01 -5.35625339e-01 -1.04438707e-01
-9.51466411e-02 6.10780343e-02 4.39993262e-01 -1.08335507e+00
-2.19800826e-02 -1.47829540e-02 4.15643156e-01 -8.46480802e-02
7.07472503e-01 -8.60924184e-01 7.49102771e-01 4.31151152e-01
-1.50437325e-01 1.76017299e-01 2.01015949e-01 1.84294447e-01
-7.63726652e-01 -3.41862924e-02 1.22314501e+00 -7.60853942e-03
-7.81782508e-01 3.47359180e-01 -3.52604479e-01 -4.04877365e-01
1.37240851e+00 -2.73736268e-01 8.97734687e-02 -9.15417671e-01
-7.74627686e-01 -3.51657033e-01 6.59710824e-01 3.96149874e-01
5.71254849e-01 -1.69503284e+00 -9.85756159e-01 4.75733072e-01
1.13019173e-03 -6.02060258e-01 4.18907434e-01 8.85426044e-01
-7.61063516e-01 1.57131061e-01 -8.19100320e-01 -5.77117383e-01
-1.42680836e+00 3.29051971e-01 5.95478415e-01 -3.67727056e-02
-3.18868250e-01 6.92186654e-01 2.12625429e-01 -1.37331628e-03
-7.35732019e-02 2.30710283e-01 -3.15152764e-01 -2.44535301e-02
5.89993656e-01 1.68117823e-03 -6.48157373e-02 -1.39456785e+00
-1.29995346e-01 7.64650345e-01 3.96015674e-01 -3.17150384e-01
1.25962961e+00 3.36020589e-02 -3.38052332e-01 2.32412685e-02
1.33902192e+00 2.22074419e-01 -1.22639322e+00 2.84798086e-01
-3.99519086e-01 -6.28619611e-01 -1.99368700e-01 -5.20253241e-01
-1.53299701e+00 7.06991911e-01 7.13791907e-01 -1.14530571e-01
1.69930863e+00 -2.40473285e-01 7.01876760e-01 -4.12430495e-01
3.72666746e-01 -7.65078783e-01 1.94870666e-01 2.29974076e-01
1.06069016e+00 -7.77549624e-01 -3.91393602e-01 -4.59524781e-01
-7.58890748e-01 1.10651481e+00 5.65823972e-01 -8.32051262e-02
3.97020429e-01 3.28500092e-01 3.08616817e-01 -1.23577565e-01
-5.94242811e-01 -2.39669800e-01 1.73839971e-01 7.17878878e-01
5.72247207e-01 -2.15658844e-01 -4.48153853e-01 2.26939633e-01
-3.36846501e-01 4.30270702e-01 4.33966428e-01 6.64135575e-01
1.22590348e-01 -1.19243884e+00 -2.53721416e-01 -1.02350809e-01
-7.53965437e-01 2.01594770e-01 -4.16060567e-01 8.67231607e-01
2.73865163e-01 6.93790257e-01 3.57540734e-02 -3.57207447e-01
2.71524131e-01 1.71446159e-01 7.32791066e-01 -1.56551510e-01
-5.53560972e-01 -2.35100631e-02 9.86749977e-02 -6.97839439e-01
-9.43440020e-01 -5.83857477e-01 -1.06052542e+00 -4.13379192e-01
5.08483648e-02 1.79202214e-01 7.67231107e-01 6.27343357e-01
4.92663205e-01 3.31538767e-01 8.31492186e-01 -9.07163560e-01
-1.12165026e-01 -8.78598392e-01 -6.67164266e-01 7.43500531e-01
3.44363034e-01 -6.73879981e-01 -1.67844072e-01 5.37263274e-01]
|
[12.840998649597168, -0.07432319223880768]
|
4610ab3d-3d6b-45f2-89a0-0eec97816d53
|
extracting-temporal-event-relation-with
|
2104.09570
| null |
https://arxiv.org/abs/2104.09570v2
|
https://arxiv.org/pdf/2104.09570v2.pdf
|
Extracting Temporal Event Relation with Syntax-guided Graph Transformer
|
Extracting temporal relations (e.g., before, after, and simultaneous) among events is crucial to natural language understanding. One of the key challenges of this problem is that when the events of interest are far away in text, the context in-between often becomes complicated, making it challenging to resolve the temporal relationship between them. This paper thus proposes a new Syntax-guided Graph Transformer network (SGT) to mitigate this issue, by (1) explicitly exploiting the connection between two events based on their dependency parsing trees, and (2) automatically locating temporal cues between two events via a novel syntax-guided attention mechanism. Experiments on two benchmark datasets, MATRES and TB-Dense, show that our approach significantly outperforms previous state-of-the-art methods on both end-to-end temporal relation extraction and temporal relation classification; This improvement also proves to be robust on the contrast set of MATRES. The code is publicly available at https://github.com/VT-NLP/Syntax-Guided-Graph-Transformer.
|
['Qiang Ning', 'Lifu Huang', 'Shuaicheng Zhang']
|
2021-04-19
|
extracting-temporal-event-relation-with-1
|
https://aclanthology.org/2022.findings-naacl.29
|
https://aclanthology.org/2022.findings-naacl.29.pdf
|
findings-naacl-2022-7
|
['temporal-relation-extraction', 'temporal-relation-classification']
|
['natural-language-processing', 'natural-language-processing']
|
[ 9.95916035e-03 1.33964811e-02 -2.85978585e-01 -4.17093366e-01
-7.82770157e-01 -8.28873098e-01 7.36427724e-01 6.83578610e-01
-3.64272356e-01 4.68701184e-01 3.87823194e-01 -4.50112224e-01
-2.63320804e-01 -7.42452025e-01 -5.08313298e-01 -4.59177166e-01
-5.18059850e-01 5.55513322e-01 4.92846906e-01 -2.21996948e-01
9.68909077e-03 2.70105243e-01 -1.17119634e+00 3.28040421e-01
5.59906065e-01 9.19976830e-01 9.30285975e-02 4.76726294e-01
-1.78833947e-01 1.04078591e+00 -2.01679260e-01 -4.56787109e-01
-5.31021245e-02 -4.92228359e-01 -1.20844364e+00 -1.89411968e-01
-1.13655655e-02 1.28417298e-01 -6.05675817e-01 7.92908072e-01
7.24197179e-02 3.78046662e-01 2.03625605e-01 -1.22491992e+00
-4.20529366e-01 9.44418728e-01 -7.95471311e-01 8.49655986e-01
4.20081168e-01 -2.88057566e-01 1.45159853e+00 -6.46254420e-01
8.08061182e-01 1.14837945e+00 4.05280292e-01 1.65757149e-01
-1.20674956e+00 -5.18098533e-01 5.39410055e-01 6.48413181e-01
-1.37584019e+00 -3.93809438e-01 9.57752407e-01 -3.89201581e-01
1.38076723e+00 1.93228647e-01 5.37916601e-01 1.09902167e+00
3.03772092e-01 6.92442775e-01 7.94241369e-01 -3.98823261e-01
2.89950110e-02 -5.83108962e-01 5.88831186e-01 4.86039281e-01
-2.52654791e-01 9.58506688e-02 -7.55044103e-01 7.22834170e-02
3.77450883e-01 -1.27200365e-01 -2.66762078e-01 -6.35119081e-02
-1.35267913e+00 6.07515574e-01 3.58740121e-01 7.09837556e-01
-3.27769756e-01 3.17893662e-02 7.47413993e-01 2.54840761e-01
7.45079815e-01 1.21465631e-01 -6.84128106e-01 -3.44276279e-01
-6.90253854e-01 7.24372268e-02 6.27185404e-01 8.39162409e-01
3.90932024e-01 -5.03837645e-01 -1.13397583e-01 6.46612585e-01
8.48928466e-02 -4.67306972e-02 2.50930369e-01 -5.35228491e-01
8.46073270e-01 6.81084871e-01 -7.88601637e-02 -1.15729725e+00
-6.07308567e-01 -2.02601671e-01 -7.04891086e-01 -4.28088605e-01
6.30446792e-01 -1.10389479e-03 -7.60873556e-01 2.02114725e+00
5.83149016e-01 4.47488934e-01 -4.34873439e-02 8.92305434e-01
1.04373121e+00 7.99364269e-01 2.53998101e-01 -5.17085373e-01
1.59850109e+00 -9.09097254e-01 -9.73676085e-01 -5.51623285e-01
7.75345147e-01 -6.26437485e-01 7.65943766e-01 1.69755146e-03
-9.55073535e-01 -2.92376161e-01 -7.52606571e-01 -4.14293915e-01
-3.06470126e-01 -1.20543361e-01 9.49754536e-01 -2.09589437e-01
-7.01665103e-01 6.55444503e-01 -1.16972494e+00 -6.11414254e-01
1.22004628e-01 2.78124273e-01 -5.47594607e-01 1.03410490e-01
-1.41230857e+00 6.28327489e-01 6.02032006e-01 2.99951047e-01
-4.10599679e-01 -6.16132319e-01 -1.11167920e+00 9.92749631e-02
7.53719270e-01 -3.71039122e-01 1.37118840e+00 -6.52860701e-01
-9.91478741e-01 9.50376272e-01 -5.65101147e-01 -4.36132371e-01
2.11271167e-01 -4.42052543e-01 -6.22053564e-01 1.50502890e-01
5.02425432e-01 2.66080558e-01 3.19276839e-01 -6.84606194e-01
-5.87046444e-01 -3.94869596e-01 6.33047968e-02 1.29265592e-01
1.51615897e-02 5.56055844e-01 -8.82920146e-01 -7.68444300e-01
2.14342698e-01 -9.44030166e-01 -1.47308707e-01 -3.93811107e-01
-5.95643282e-01 -6.63442850e-01 8.11248660e-01 -8.24810922e-01
1.50061500e+00 -2.17953539e+00 2.22112954e-01 -5.25288545e-02
2.39933491e-01 4.68233004e-02 2.13380102e-02 7.45206773e-01
-6.71044469e-01 5.98886609e-02 -2.73877025e-01 -3.42416793e-01
-2.48449787e-01 3.34081918e-01 -3.93491238e-01 4.53013122e-01
2.71932811e-01 8.99857879e-01 -1.27560723e+00 -5.69050252e-01
1.62626997e-01 3.26144427e-01 -4.76014726e-02 2.64414907e-01
-2.27721542e-01 7.48760879e-01 -5.50517797e-01 4.57589090e-01
2.78367907e-01 -5.20486295e-01 5.59966564e-01 -2.75591195e-01
-1.48749337e-01 1.05309761e+00 -1.06506646e+00 1.69987333e+00
-4.34645176e-01 8.20912898e-01 -3.00748199e-01 -1.11807156e+00
7.23457992e-01 5.77550948e-01 5.51135480e-01 -8.87583792e-01
2.80923456e-01 -1.07796386e-01 8.14691335e-02 -5.13804972e-01
2.79861659e-01 8.58558565e-02 -3.27204376e-01 4.41756129e-01
-2.14224346e-02 4.02668715e-01 8.01259041e-01 5.06567538e-01
1.38437402e+00 2.04782695e-01 7.52906561e-01 -1.50429711e-01
4.42851990e-01 -1.09930083e-01 1.10084736e+00 2.79421061e-01
-2.95920968e-01 3.75362933e-01 9.40045953e-01 -6.82156026e-01
-5.14289498e-01 -9.44687128e-01 1.82219893e-01 9.14602101e-01
1.66134059e-01 -9.24508810e-01 -1.45126775e-01 -9.35218215e-01
-2.72613823e-01 7.14236796e-01 -9.08229411e-01 1.28059953e-01
-9.80888426e-01 -5.12361467e-01 3.23460579e-01 8.40718806e-01
2.39503518e-01 -9.70473230e-01 -4.93130326e-01 4.56616491e-01
-8.11318398e-01 -1.69567239e+00 -6.53186858e-01 3.46822947e-01
-8.02071393e-01 -1.23722768e+00 -4.86768037e-02 -7.73891330e-01
3.53435814e-01 2.60422617e-01 1.44744539e+00 -8.79821740e-03
-1.41546102e-02 -2.66354233e-02 -6.24510646e-01 -7.37325381e-03
-3.09954826e-02 -6.00676890e-03 -3.93983752e-01 7.97513649e-02
4.42654014e-01 -9.93894041e-01 -3.89533311e-01 3.24566931e-01
-6.52846217e-01 2.43251592e-01 1.54391676e-01 5.58250070e-01
6.45309567e-01 2.75473654e-01 2.96027392e-01 -9.63606358e-01
2.65760005e-01 -6.47354364e-01 -5.85062623e-01 2.48353302e-01
-2.54117101e-01 5.35175987e-02 5.36111295e-01 -4.29710329e-01
-1.00947237e+00 -3.30511993e-03 8.80569741e-02 -2.41507381e-01
-2.43613452e-01 9.04667318e-01 -1.98314749e-02 5.72777152e-01
3.21915388e-01 1.05975224e-02 -7.10651577e-01 -2.73500830e-01
2.95965344e-01 -1.49464021e-02 6.44727290e-01 -5.93224108e-01
5.54307699e-01 5.64012766e-01 1.52458489e-01 -6.54908955e-01
-1.20457029e+00 -6.99637890e-01 -8.42592478e-01 -1.32779881e-01
1.00520623e+00 -8.09873223e-01 -4.31023926e-01 2.44208738e-01
-1.45809793e+00 -5.54998457e-01 -3.24747637e-02 4.63868439e-01
-2.42750406e-01 4.82022166e-01 -9.45885360e-01 -6.03022039e-01
-1.83074102e-01 -8.36796343e-01 9.70035434e-01 2.34109402e-01
-5.99041641e-01 -1.09822786e+00 2.63014466e-01 3.86676550e-01
-1.90314978e-01 5.37246466e-01 9.12427962e-01 -6.60212100e-01
-5.14911592e-01 -1.29470736e-01 -2.58546501e-01 -4.82002676e-01
2.78594136e-01 8.01847428e-02 -5.65101445e-01 9.35068578e-02
-3.46010357e-01 -7.34130144e-02 7.98898280e-01 3.22186828e-01
7.70056903e-01 -1.40680373e-01 -6.28752589e-01 3.33679169e-01
1.01921201e+00 3.84558916e-01 4.47570473e-01 2.35834658e-01
7.52805829e-01 7.03230441e-01 8.36706281e-01 2.98034281e-01
8.02686453e-01 9.97696400e-01 1.86795861e-01 1.10054679e-01
-1.69204071e-01 -1.77475452e-01 2.79282749e-01 8.54029715e-01
-8.69575217e-02 -5.01359999e-01 -1.10011852e+00 1.03134155e+00
-2.34665227e+00 -8.31838012e-01 -5.94437063e-01 1.85653710e+00
9.14418280e-01 2.80819237e-01 1.09014913e-01 2.70143449e-01
8.70043874e-01 4.27469283e-01 -1.99143440e-01 -2.41903678e-01
1.96486544e-02 2.23511100e-01 4.36890908e-02 4.41035986e-01
-1.42197382e+00 1.16715145e+00 4.61789274e+00 5.91063440e-01
-1.00257277e+00 2.06225634e-01 6.18636489e-01 6.79023266e-02
6.03711307e-02 3.51142317e-01 -6.85279787e-01 2.90907949e-01
1.03391898e+00 -2.30653018e-01 3.03399622e-01 3.20267767e-01
3.47934961e-01 -1.61659956e-01 -1.34520161e+00 7.62770116e-01
-2.99082309e-01 -1.14976311e+00 -4.38002050e-01 -3.79519433e-01
3.29715908e-01 4.32069274e-03 -4.41873580e-01 1.96964592e-01
3.32526058e-01 -5.93839228e-01 8.98651242e-01 1.21074714e-01
4.44921702e-01 -6.54749811e-01 6.59942865e-01 4.60060984e-02
-2.00543141e+00 3.07950944e-01 2.67486036e-01 -2.37086505e-01
6.43682659e-01 9.77056861e-01 -5.26328385e-01 1.00903082e+00
9.37345684e-01 1.13284993e+00 -3.97156179e-01 7.64761209e-01
-9.21444178e-01 7.90282428e-01 -3.22918445e-01 1.85690343e-01
1.35112375e-01 -6.28667697e-02 8.44653845e-01 1.27824008e+00
1.00016996e-01 4.28417534e-01 1.98787063e-01 7.65092909e-01
4.02798951e-02 4.02657799e-02 -5.34528136e-01 -4.32727337e-01
4.72298801e-01 1.28363764e+00 -1.34158039e+00 -1.20256774e-01
-4.97886360e-01 8.60476911e-01 7.17611015e-01 3.16088051e-01
-1.06711018e+00 -2.21949801e-01 3.91420245e-01 -1.77524090e-01
4.55267251e-01 -5.97478509e-01 -1.68493435e-01 -1.18644702e+00
2.79687762e-01 -5.20481050e-01 1.04174674e+00 -6.47640109e-01
-1.18553579e+00 7.80745864e-01 6.22653104e-02 -9.62050557e-01
-1.84110463e-01 -1.47678241e-01 -7.02978134e-01 6.74051046e-01
-1.20918465e+00 -1.22569060e+00 -1.31645828e-01 6.16190612e-01
6.66393161e-01 5.46234071e-01 5.37259102e-01 5.15889108e-01
-8.60787392e-01 3.70889932e-01 -6.12400234e-01 5.08751392e-01
5.83538055e-01 -1.21493185e+00 6.34858727e-01 1.39623690e+00
4.66068298e-01 5.58810055e-01 6.89745307e-01 -7.39372909e-01
-1.26897061e+00 -1.09922242e+00 1.61194301e+00 -3.40531886e-01
1.18182325e+00 -5.97630799e-01 -1.07082617e+00 1.04269183e+00
3.40898454e-01 1.62785679e-01 5.35249889e-01 5.94531715e-01
-6.27994478e-01 -6.92285001e-02 -4.97851938e-01 7.46020019e-01
1.31842387e+00 -5.82131088e-01 -7.09460974e-01 4.86506581e-01
8.93151164e-01 -6.14873469e-01 -8.64502549e-01 5.47324061e-01
1.34933457e-01 -9.00280416e-01 7.58939564e-01 -5.38926542e-01
5.23893118e-01 -3.60933542e-01 -1.49241416e-02 -1.03251803e+00
-3.76694292e-01 -9.12486851e-01 -3.62362206e-01 1.65488100e+00
5.89391530e-01 -6.58353925e-01 4.64364797e-01 4.29350734e-01
-1.32191569e-01 -6.43656492e-01 -9.87950265e-01 -9.04762745e-01
-3.51921141e-01 -7.35158622e-01 2.28622451e-01 1.34624457e+00
2.85855502e-01 8.34876359e-01 -2.31895983e-01 4.60253537e-01
2.48132229e-01 5.29819191e-01 4.12187338e-01 -1.15341771e+00
-4.19103175e-01 -5.61523616e-01 -2.15500399e-01 -8.96629691e-01
3.20413798e-01 -8.32693338e-01 1.48224726e-01 -1.68545389e+00
1.05895810e-01 -3.00698638e-01 -1.73845738e-01 8.39505613e-01
-3.32911223e-01 -1.40069336e-01 -2.74553560e-02 4.22908030e-02
-8.17167878e-01 4.43546325e-01 1.00886142e+00 -4.99633998e-02
-3.32625389e-01 -1.29468828e-01 -4.21671927e-01 6.13060594e-01
7.69814372e-01 -7.59997487e-01 -4.40862268e-01 -3.13228816e-01
4.03852373e-01 4.97211665e-01 2.87299812e-01 -5.61093688e-01
3.31432551e-01 -2.95729667e-01 -2.43026525e-01 -7.48844624e-01
1.64796799e-01 -6.83710456e-01 1.57497063e-01 1.98109031e-01
-3.09637547e-01 3.90351176e-01 4.71086651e-01 6.31826639e-01
-4.16769803e-01 2.25700021e-01 4.75559771e-01 1.35823280e-01
-9.83257413e-01 3.38252187e-01 -3.07462931e-01 4.82404053e-01
1.01618230e+00 4.22227025e-01 -4.38865483e-01 -3.88167858e-01
-8.91817451e-01 2.84098417e-01 -2.26344746e-02 5.54631293e-01
4.05016810e-01 -1.31349397e+00 -5.55070281e-01 -2.65167713e-01
2.28666604e-01 1.83389172e-01 2.76614904e-01 1.24611557e+00
-1.87357157e-01 4.44453865e-01 3.73271555e-01 -5.45698643e-01
-1.56042695e+00 7.72890627e-01 7.66050145e-02 -7.93913662e-01
-9.67234552e-01 1.01502824e+00 3.71547312e-01 -6.90061599e-02
5.08191921e-02 -6.32761180e-01 -2.07244486e-01 1.20978884e-01
4.98580933e-01 1.86316594e-02 1.42357543e-01 -6.65206313e-01
-7.68256485e-01 5.02696097e-01 -2.08231911e-01 -1.09917581e-01
1.31989622e+00 -1.78067252e-01 -4.83614206e-01 6.21461630e-01
9.86658871e-01 -4.58737761e-02 -1.03871322e+00 -5.01872718e-01
4.30060029e-01 -2.44357079e-01 4.31497255e-03 -6.27385020e-01
-1.21881783e+00 6.79210544e-01 -1.24105737e-01 5.96655607e-01
1.32464790e+00 4.49691892e-01 9.96278286e-01 1.40649199e-01
1.73159644e-01 -6.37100935e-01 -1.60548147e-02 8.01692009e-01
9.02687371e-01 -1.12642026e+00 -1.13900878e-01 -9.04777825e-01
-4.08971548e-01 1.00329328e+00 4.31147844e-01 -8.46661907e-03
7.79556751e-01 3.66788924e-01 -3.22927088e-02 -4.49660569e-01
-9.71795917e-01 -4.99591053e-01 5.70601225e-01 1.85138166e-01
7.75454223e-01 -1.65850986e-02 -4.52950984e-01 6.21808410e-01
6.87542260e-02 -2.53516763e-01 1.93046853e-01 1.12059879e+00
2.84974128e-01 -1.26672077e+00 -1.38273195e-03 -3.66660208e-02
-6.46058142e-01 -3.58503044e-01 -4.68354255e-01 9.43316281e-01
1.18490523e-02 1.26372433e+00 7.00128600e-02 -2.93620557e-01
3.01715970e-01 -9.05885249e-02 4.78886068e-01 -5.33142924e-01
-5.27945101e-01 2.83114791e-01 5.91257513e-01 -9.11524117e-01
-5.45148849e-01 -9.85527337e-01 -1.60993826e+00 -1.16773181e-01
-6.58218712e-02 1.66907296e-01 7.19587132e-02 1.14003289e+00
4.26513284e-01 8.29290807e-01 3.59279811e-01 -6.13244772e-01
3.60326558e-01 -6.99826896e-01 -2.65300244e-01 5.91513157e-01
2.27887362e-01 -7.79982388e-01 -2.53406495e-01 1.51471511e-01]
|
[9.072601318359375, 9.124466896057129]
|
e675773f-fa0b-480c-8e19-ca4179de5568
|
idisc-internal-discretization-for-monocular
|
2304.06334
| null |
https://arxiv.org/abs/2304.06334v1
|
https://arxiv.org/pdf/2304.06334v1.pdf
|
iDisc: Internal Discretization for Monocular Depth Estimation
|
Monocular depth estimation is fundamental for 3D scene understanding and downstream applications. However, even under the supervised setup, it is still challenging and ill-posed due to the lack of full geometric constraints. Although a scene can consist of millions of pixels, there are fewer high-level patterns. We propose iDisc to learn those patterns with internal discretized representations. The method implicitly partitions the scene into a set of high-level patterns. In particular, our new module, Internal Discretization (ID), implements a continuous-discrete-continuous bottleneck to learn those concepts without supervision. In contrast to state-of-the-art methods, the proposed model does not enforce any explicit constraints or priors on the depth output. The whole network with the ID module can be trained end-to-end, thanks to the bottleneck module based on attention. Our method sets the new state of the art with significant improvements on NYU-Depth v2 and KITTI, outperforming all published methods on the official KITTI benchmark. iDisc can also achieve state-of-the-art results on surface normal estimation. Further, we explore the model generalization capability via zero-shot testing. We observe the compelling need to promote diversification in the outdoor scenario. Hence, we introduce splits of two autonomous driving datasets, DDAD and Argoverse. Code is available at http://vis.xyz/pub/idisc .
|
['Fisher Yu', 'Christos Sakaridis', 'Luigi Piccinelli']
|
2023-04-13
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Piccinelli_iDisc_Internal_Discretization_for_Monocular_Depth_Estimation_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Piccinelli_iDisc_Internal_Discretization_for_Monocular_Depth_Estimation_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['surface-normals-estimation', 'monocular-depth-estimation']
|
['computer-vision', 'computer-vision']
|
[ 2.69985646e-01 2.75485724e-01 -1.32231697e-01 -3.78092170e-01
-6.29203022e-01 -3.16620499e-01 5.58645248e-01 -2.57023335e-01
-3.78921568e-01 4.81306314e-01 -7.85187483e-02 -2.09712535e-01
3.29966210e-02 -9.98741627e-01 -9.92338896e-01 -7.25943148e-01
1.60339683e-01 5.53123176e-01 4.14781421e-01 -1.75893173e-01
2.30862856e-01 3.73027503e-01 -1.85203111e+00 2.68552870e-01
1.13456070e+00 1.23913336e+00 4.96635139e-01 4.37026381e-01
-1.09934114e-01 6.39702320e-01 -5.48928827e-02 -1.64671168e-01
4.99419689e-01 -1.18694611e-01 -4.40931439e-01 1.90805778e-01
8.91550601e-01 -4.93550926e-01 -4.84611124e-01 9.63739634e-01
4.93536592e-01 1.48680523e-01 6.66075289e-01 -1.00186455e+00
-3.36820394e-01 -4.27919114e-03 -7.08275676e-01 -2.36510523e-02
7.32934773e-02 5.03675938e-01 8.53002846e-01 -1.05885971e+00
7.17768192e-01 1.14496005e+00 4.10083205e-01 4.75274116e-01
-1.18317509e+00 -5.16189277e-01 6.89745545e-01 3.09106886e-01
-1.17780972e+00 -3.55167121e-01 8.24848771e-01 -5.20181239e-01
9.69886303e-01 -2.30602786e-01 6.73124015e-01 1.24030864e+00
-4.62167822e-02 9.72565889e-01 9.87360775e-01 -1.77406117e-01
4.69898790e-01 -5.75095266e-02 2.06078589e-01 6.31324649e-01
2.35226035e-01 1.72112331e-01 -6.35129511e-01 2.57870883e-01
7.17352331e-01 1.17279090e-01 -2.99245059e-01 -7.71758080e-01
-1.04204428e+00 7.81302989e-01 6.02512479e-01 -2.28573978e-01
-2.50299871e-01 2.44157702e-01 1.70218855e-01 1.84234813e-01
6.52539313e-01 2.70240098e-01 -5.14388263e-01 -4.25212532e-02
-6.89509869e-01 3.37734520e-01 5.52925587e-01 1.05086493e+00
1.19469953e+00 -7.19949976e-02 -5.04484661e-02 6.92593634e-01
1.33567974e-01 4.51482773e-01 9.61429775e-02 -1.15749764e+00
7.45036066e-01 5.63579917e-01 5.15312627e-02 -5.75759113e-01
-3.80867034e-01 -6.90143049e-01 -8.25136423e-01 5.20540714e-01
3.85473728e-01 -1.20303303e-01 -1.19336200e+00 1.62381804e+00
4.36538309e-01 4.21271533e-01 -4.19855490e-03 1.00114429e+00
7.14856446e-01 6.16064429e-01 -3.67956996e-01 1.26130432e-01
9.21733201e-01 -1.29440570e+00 -3.33659798e-01 -6.51344836e-01
6.28591895e-01 -2.39576429e-01 1.28556085e+00 5.82268298e-01
-9.05292809e-01 -6.33315861e-01 -1.12649858e+00 -4.53265548e-01
-3.12315613e-01 -1.61276162e-02 8.30694914e-01 2.91988790e-01
-9.74346161e-01 5.29955506e-01 -9.86511350e-01 -2.87416160e-01
7.34790921e-01 1.12811491e-01 -2.44921416e-01 -4.95075852e-01
-8.90864313e-01 5.52389145e-01 1.01072535e-01 5.90704679e-02
-1.12786293e+00 -9.70941365e-01 -1.18591475e+00 -1.98303297e-01
5.63378334e-01 -9.77794349e-01 1.14249945e+00 -6.82258248e-01
-1.65021229e+00 9.67795253e-01 -2.25025028e-01 -4.59455252e-01
8.19199622e-01 -4.61283743e-01 1.98821500e-01 2.18811527e-01
1.31558061e-01 9.87444758e-01 7.57680774e-01 -1.24091327e+00
-7.56396472e-01 -5.09093285e-01 3.74652892e-01 3.28665316e-01
-1.18347332e-01 -7.20633209e-01 -7.46301055e-01 -3.42533827e-01
3.96967292e-01 -7.98463821e-01 -3.97655785e-01 3.68218601e-01
-3.15060318e-01 -3.56161147e-02 6.53458416e-01 -1.67605296e-01
8.43869686e-01 -2.32158399e+00 3.66164684e-01 -1.79147854e-01
1.92885473e-01 -4.73512299e-02 -4.88762856e-02 2.83185393e-01
2.73308992e-01 -1.56775445e-01 -5.64920068e-01 -8.08081150e-01
-6.40350999e-03 2.68132299e-01 -2.79731452e-01 5.86712301e-01
1.72737926e-01 7.60989964e-01 -8.56171608e-01 -1.21639743e-01
5.92929184e-01 4.63596612e-01 -9.85570014e-01 1.32944122e-01
-5.51143408e-01 6.99340820e-01 -3.75811130e-01 6.87453210e-01
9.96228874e-01 -2.85860896e-01 -1.86173230e-01 4.19535302e-02
-3.28339785e-01 4.48069841e-01 -1.13190651e+00 2.35451484e+00
-5.63006401e-01 5.92822611e-01 -4.74508442e-02 -9.84824300e-01
6.76633894e-01 -2.45736346e-01 1.50667608e-01 -7.59451807e-01
-6.73945546e-02 1.62457496e-01 -2.17321351e-01 -5.12467206e-01
2.89008170e-01 5.71386516e-02 -3.63033563e-02 -6.43921196e-02
-5.01791164e-02 -4.09409970e-01 3.64060253e-02 8.84643048e-02
1.07888269e+00 4.81483340e-01 -3.49379554e-02 -2.29632884e-01
2.92807072e-01 -2.19356529e-02 6.91769183e-01 6.63630426e-01
-1.09773442e-01 9.33287501e-01 5.90970993e-01 -3.29330444e-01
-9.20446455e-01 -1.18684578e+00 -3.23424101e-01 6.97150052e-01
6.13689303e-01 -2.65477926e-01 -5.39927483e-01 -5.08684754e-01
2.56086946e-01 6.74603939e-01 -8.08272004e-01 1.40942577e-02
-3.00818086e-01 -5.02219617e-01 1.98193818e-01 5.76004267e-01
6.69759095e-01 -7.83936858e-01 -8.71425748e-01 3.72640081e-02
3.23694721e-02 -1.39379716e+00 -1.77123189e-01 2.89745182e-01
-9.55077052e-01 -9.91868734e-01 -5.80490291e-01 -5.15573919e-01
4.89734203e-01 3.92717063e-01 1.07068002e+00 -2.34068811e-01
-2.70376772e-01 2.14184389e-01 -3.06702495e-01 -3.33366662e-01
1.83612883e-01 3.22237313e-01 -2.43289888e-01 -2.73118597e-02
2.82387555e-01 -8.10280979e-01 -9.61037099e-01 2.72011459e-01
-7.19804525e-01 4.48848456e-01 6.33341372e-01 7.65161872e-01
8.53639960e-01 -2.20840782e-01 2.59622335e-01 -9.74850118e-01
-3.21187437e-01 -4.97706741e-01 -7.66751349e-01 -2.70705849e-01
-4.49098200e-01 1.37605658e-02 5.13048887e-01 -1.50375873e-01
-1.07497633e+00 2.17968121e-01 -1.64927512e-01 -7.32996702e-01
-4.39523786e-01 2.45580941e-01 -3.87025505e-01 2.04279181e-02
5.76291919e-01 1.15764096e-01 -1.63625747e-01 -5.64011872e-01
3.47611487e-01 4.06647295e-01 5.03973842e-01 -5.12136400e-01
7.42716193e-01 9.19839323e-01 -9.59602743e-03 -1.08379352e+00
-1.07493615e+00 -5.89862943e-01 -7.13874817e-01 -7.99935088e-02
8.94725025e-01 -1.30022228e+00 -4.05502319e-01 8.00687730e-01
-9.95707393e-01 -9.00914669e-01 -2.68057525e-01 3.56969029e-01
-7.50405729e-01 3.17326427e-01 -5.09130657e-01 -6.93922102e-01
6.05330616e-02 -1.09321332e+00 1.34228468e+00 -8.67047533e-03
2.23862633e-01 -7.60724068e-01 -5.14367446e-02 4.18105334e-01
2.37226620e-01 3.77669215e-01 7.64322758e-01 8.65175128e-02
-1.09735262e+00 1.35703981e-01 -2.52482265e-01 3.12274754e-01
-1.85462739e-02 -1.81354076e-01 -1.45668828e+00 -2.49912947e-01
-3.71511355e-02 -5.42942941e-01 1.52674627e+00 5.18992662e-01
1.43804431e+00 7.95552954e-02 -2.49675930e-01 1.24110878e+00
1.57685840e+00 -9.34682563e-02 8.54588747e-01 3.69888961e-01
9.85980332e-01 7.87500024e-01 6.60030127e-01 4.98397559e-01
6.37577415e-01 6.27331853e-01 8.40450048e-01 -1.43628389e-01
-1.28856912e-01 -2.95149267e-01 2.54273713e-01 3.40366364e-01
4.11398970e-02 -2.91584998e-01 -8.88910115e-01 5.58018804e-01
-1.88638449e+00 -7.99141645e-01 -5.06448261e-02 2.14718533e+00
5.17600954e-01 4.68415260e-01 -1.46999806e-01 1.27045378e-01
3.03697139e-01 5.07286370e-01 -1.00753105e+00 -4.54576984e-02
-2.09266454e-01 1.06402680e-01 6.16652071e-01 7.26922929e-01
-1.14487517e+00 1.12932646e+00 4.95897055e+00 7.02315629e-01
-1.19466531e+00 3.01095396e-02 8.28956902e-01 -3.47088516e-01
-4.12946731e-01 -4.56554070e-02 -1.05993605e+00 3.39931369e-01
4.85947937e-01 2.88369834e-01 2.81949490e-01 9.07746434e-01
1.97302297e-01 -4.03746247e-01 -1.29716027e+00 1.02525997e+00
4.53072898e-02 -1.32764494e+00 -9.42812487e-02 1.57743186e-01
9.39544141e-01 4.63175446e-01 5.98008409e-02 3.79585773e-01
4.21171039e-02 -1.00042796e+00 9.02408957e-01 4.76331860e-01
1.04162586e+00 -6.51963294e-01 4.20768917e-01 5.92555404e-01
-1.14219654e+00 -1.08430579e-01 -6.77325964e-01 -4.67948258e-01
7.77126476e-02 7.90734768e-01 -2.54163116e-01 5.76409161e-01
7.39248633e-01 1.18526590e+00 -4.40450847e-01 9.83592868e-01
-3.40307176e-01 3.53930175e-01 -4.66360211e-01 2.69275814e-01
3.76767486e-01 -2.54468143e-01 4.06715900e-01 9.72193897e-01
2.13974595e-01 9.96056795e-02 2.72612691e-01 8.50842953e-01
-6.22285269e-02 -2.13642389e-01 -8.24948072e-01 4.54773903e-01
2.87417084e-01 9.66278732e-01 -4.73552674e-01 -3.18706095e-01
-5.95811784e-01 1.05495548e+00 6.25986636e-01 5.21211028e-01
-7.15565681e-01 -1.94437981e-01 1.07307172e+00 3.57976019e-01
6.04009390e-01 -3.64905953e-01 -6.37555122e-01 -1.30277908e+00
3.04911077e-01 -6.75049067e-01 1.24213986e-01 -5.50379455e-01
-1.27119482e+00 3.82420748e-01 5.69067560e-02 -1.31789267e+00
-2.54857279e-02 -8.62954736e-01 -5.25201082e-01 7.57131994e-01
-2.08528996e+00 -9.49895501e-01 -8.31491649e-01 4.25899506e-01
8.75418544e-01 1.94167435e-01 4.85915214e-01 3.37332428e-01
-6.03580713e-01 4.20944512e-01 -1.00114895e-03 -2.03495279e-01
6.74380839e-01 -1.14345169e+00 5.84949791e-01 7.65697420e-01
-1.04063995e-01 1.24243900e-01 5.03234744e-01 -4.68736827e-01
-1.38312137e+00 -1.25143254e+00 4.55839247e-01 -4.85726863e-01
4.59304124e-01 -7.91548371e-01 -9.25401092e-01 6.06403410e-01
-6.11744523e-02 3.31702948e-01 1.85158283e-01 -3.77352312e-02
-3.50519359e-01 -2.18840435e-01 -8.51317763e-01 4.86958563e-01
1.65674615e+00 -3.63564610e-01 -2.41388053e-01 2.89998442e-01
1.04926050e+00 -6.25795901e-01 -3.61428469e-01 6.44092143e-01
3.93862844e-01 -1.39879024e+00 9.75377977e-01 -2.14781404e-01
7.96402752e-01 -2.30437905e-01 -3.81155550e-01 -1.05855441e+00
-1.73783422e-01 -3.29732299e-01 -1.98334172e-01 7.81220794e-01
4.67483193e-01 -7.90065348e-01 1.14052725e+00 3.62394899e-01
-6.34761989e-01 -1.06563485e+00 -1.01167440e+00 -8.56215656e-01
2.49490738e-01 -8.08337748e-01 3.33878338e-01 6.87543809e-01
-4.11731064e-01 3.63208324e-01 -2.66433269e-01 3.86312753e-01
8.70312631e-01 2.67590493e-01 1.08534491e+00 -1.15942943e+00
-6.18729472e-01 -4.09912378e-01 -4.65495020e-01 -1.68128359e+00
1.31128579e-01 -6.90478444e-01 1.09525636e-01 -1.64223468e+00
-9.21844915e-02 -5.14503539e-01 -3.71552655e-03 3.23983848e-01
-7.68278092e-02 2.10840598e-01 -1.37786074e-02 1.15856119e-01
-5.10323644e-01 1.07887578e+00 1.43559289e+00 -1.62977755e-01
-2.26432770e-01 -8.75920504e-02 -4.75339025e-01 8.36243272e-01
7.05395699e-01 -2.40662262e-01 -6.54596567e-01 -7.44913518e-01
1.65475249e-01 -1.39798105e-01 5.13138175e-01 -1.21864510e+00
2.07882509e-01 -1.80723578e-01 3.01823974e-01 -8.27713370e-01
6.49468958e-01 -4.89241272e-01 -2.18910992e-01 3.46671790e-01
-5.16986810e-02 -4.02435243e-01 1.81172386e-01 6.45277679e-01
-1.85513973e-01 8.49026442e-02 9.80792642e-01 -1.73641384e-01
-9.82594728e-01 6.71612024e-01 -1.79304823e-03 3.03519636e-01
9.74825919e-01 -4.43419367e-01 -3.53451312e-01 -1.28448427e-01
-3.86386156e-01 5.03526211e-01 8.43344927e-01 2.03454986e-01
6.26172781e-01 -1.01397026e+00 -6.22331619e-01 3.97029817e-01
4.69996512e-01 9.40888524e-01 4.82924223e-01 7.38974988e-01
-6.10789061e-01 2.10869581e-01 -4.69936840e-02 -9.52018023e-01
-6.08898520e-01 4.31647122e-01 4.98783886e-01 -4.37677316e-02
-1.07838976e+00 9.65107799e-01 9.45644677e-01 -4.54491317e-01
4.79691207e-01 -6.02747858e-01 1.78560149e-02 -7.39681944e-02
3.95047635e-01 2.87149042e-01 5.15910983e-02 -1.57078207e-01
-2.30958849e-01 9.54161763e-01 9.03627351e-02 3.26150917e-02
1.33708370e+00 -1.73411191e-01 2.39862114e-01 6.31436825e-01
1.23359632e+00 -1.73979402e-01 -2.05319524e+00 -1.44476354e-01
-3.26022059e-01 -7.04041839e-01 1.47651613e-01 -4.97345626e-01
-1.05911207e+00 1.22897077e+00 5.19282699e-01 -4.15371597e-01
1.01551604e+00 3.63210477e-02 6.85086548e-01 6.20513380e-01
6.00342751e-01 -9.43207085e-01 1.75677523e-01 7.76407421e-01
8.73947322e-01 -1.67451739e+00 -1.77059263e-01 -6.18850529e-01
-4.96356845e-01 7.50442743e-01 9.33702946e-01 -3.41441482e-01
7.79426754e-01 2.16579765e-01 6.23326339e-02 -2.37325430e-01
-9.06165421e-01 -3.17923248e-01 9.16585624e-02 5.37341118e-01
8.73209089e-02 -2.11263314e-01 1.26211494e-01 3.56829435e-01
-1.84004530e-01 -1.20559052e-01 4.19979036e-01 7.30058968e-01
-5.36299944e-01 -8.00805330e-01 -1.87238269e-02 3.35096210e-01
1.35637550e-02 -1.47367671e-01 -5.32683842e-02 8.04713070e-01
2.76549220e-01 7.06975937e-01 2.30219200e-01 -2.67962009e-01
5.04539073e-01 -1.95240572e-01 5.63800395e-01 -9.41394925e-01
4.92927730e-02 -1.53383702e-01 -2.42313109e-02 -9.00889099e-01
-1.67544469e-01 -7.83719242e-01 -1.14673519e+00 -2.12525159e-01
-2.48086266e-02 -3.81254822e-01 5.48435628e-01 8.29458594e-01
4.12319213e-01 5.85164428e-01 5.94301581e-01 -1.42739213e+00
-3.03132445e-01 -9.24675643e-01 -4.73218143e-01 2.94760942e-01
5.48477411e-01 -9.33477759e-01 -5.97185731e-01 -2.35870793e-01]
|
[8.575189590454102, -2.5651438236236572]
|
90f4e4e7-45f7-4182-b4d9-197177a391ab
|
proof-of-swarm-based-ensemble-learning-for
|
2212.14050
| null |
https://arxiv.org/abs/2212.14050v2
|
https://arxiv.org/pdf/2212.14050v2.pdf
|
Proof of Swarm Based Ensemble Learning for Federated Learning Applications
|
Ensemble learning combines results from multiple machine learning models in order to provide a better and optimised predictive model with reduced bias, variance and improved predictions. However, in federated learning it is not feasible to apply centralised ensemble learning directly due to privacy concerns. Hence, a mechanism is required to combine results of local models to produce a global model. Most distributed consensus algorithms, such as Byzantine fault tolerance (BFT), do not normally perform well in such applications. This is because, in such methods predictions of some of the peers are disregarded, so a majority of peers can win without even considering other peers' decisions. Additionally, the confidence score of the result of each peer is not normally taken into account, although it is an important feature to consider for ensemble learning. Moreover, the problem of a tie event is often left un-addressed by methods such as BFT. To fill these research gaps, we propose PoSw (Proof of Swarm), a novel distributed consensus algorithm for ensemble learning in a federated setting, which was inspired by particle swarm based algorithms for solving optimisation problems. The proposed algorithm is theoretically proved to always converge in a relatively small number of steps and has mechanisms to resolve tie events while trying to achieve sub-optimum solutions. We experimentally validated the performance of the proposed algorithm using ECG classification as an example application in healthcare, showing that the ensemble learning model outperformed all local models and even the FL-based global model. To the best of our knowledge, the proposed algorithm is the first attempt to make consensus over the output results of distributed models trained using federated learning.
|
['Shujun Li', 'Ludovic Koehl', 'Kim Phuc Tran', 'Ali Raza']
|
2022-12-28
| null | null | null | null |
['ecg-classification']
|
['medical']
|
[ 9.38334912e-02 8.34130123e-02 1.95242181e-01 -3.53743911e-01
-3.92745733e-01 -3.87528062e-01 2.81422943e-01 7.05390275e-01
-3.51255685e-01 1.27842617e+00 -4.21090543e-01 -6.34466186e-02
-8.53510201e-01 -8.28106701e-01 -5.59971631e-01 -1.12127888e+00
-2.34385520e-01 8.15810740e-01 -3.94922346e-02 2.37791408e-02
1.34012729e-01 3.37789416e-01 -1.67351604e+00 4.01975930e-01
1.31322718e+00 1.08575630e+00 -1.33287162e-01 4.49041605e-01
3.88996601e-02 7.52301991e-01 -9.55649436e-01 -3.61951113e-01
2.86503524e-01 -6.53128445e-01 -4.80847865e-01 -2.74177104e-01
1.23766996e-01 2.35677168e-01 4.96308982e-01 8.09970677e-01
6.96673334e-01 1.92280650e-01 3.69838268e-01 -1.34045768e+00
2.90373266e-01 6.04413867e-01 -6.86385185e-02 -1.47774935e-01
1.58196718e-01 -1.36766165e-01 7.54258275e-01 -3.29795569e-01
6.53876126e-01 6.69070780e-01 7.20299423e-01 3.40989381e-01
-1.18395007e+00 -4.76562262e-01 5.01199700e-02 3.27968627e-01
-1.26698387e+00 -2.11870044e-01 4.41811323e-01 -2.70089097e-02
5.96474469e-01 8.01150262e-01 6.69873834e-01 7.16526926e-01
6.19949460e-01 3.82690907e-01 1.37023950e+00 -3.79116565e-01
7.91399837e-01 4.41983461e-01 -7.62181878e-02 2.44286776e-01
5.73541403e-01 -5.17717041e-02 -6.40560091e-01 -7.55565166e-01
-1.32412151e-01 1.74495056e-01 -4.54889476e-01 -4.20914292e-01
-9.71501172e-01 5.63755035e-01 4.35527027e-01 5.16630411e-01
-8.76765490e-01 -1.93247601e-01 4.11641330e-01 4.91062492e-01
6.33739114e-01 2.10366726e-01 -5.42710006e-01 -1.01747058e-01
-1.09385943e+00 6.83659241e-02 1.25422263e+00 1.26754805e-01
4.96619552e-01 -1.93809196e-01 9.74343643e-02 5.16391695e-01
1.96530789e-01 3.17224979e-01 5.23942292e-01 -8.65473986e-01
9.87996608e-02 8.06393683e-01 9.48274732e-02 -1.18276918e+00
-3.12839806e-01 -8.54882360e-01 -1.21497881e+00 5.80010116e-01
4.13746446e-01 -3.11190456e-01 -1.71641022e-01 1.52479529e+00
7.22896397e-01 2.73088068e-01 3.50141048e-01 8.97029400e-01
3.90433073e-01 3.98005784e-01 -4.73945178e-02 -7.30601609e-01
9.37464714e-01 -7.10743129e-01 -7.17035592e-01 2.86679447e-01
6.40510321e-01 -6.10619605e-01 -2.04290211e-01 9.72501695e-01
-7.57706583e-01 -1.88310638e-01 -1.01415634e+00 9.12147403e-01
-2.60197669e-01 -9.95164961e-02 2.69984663e-01 7.09394276e-01
-8.99314702e-01 1.03218937e+00 -8.33853960e-01 -4.65894550e-01
3.42794955e-01 7.15004802e-01 -4.27034289e-01 -1.24106281e-01
-1.07823193e+00 1.06797051e+00 3.87799084e-01 2.02536061e-01
-3.09380472e-01 -3.62556726e-01 -1.80035934e-01 5.80524802e-02
3.03777486e-01 -9.91629362e-01 7.46036232e-01 -1.32148552e+00
-1.30544841e+00 1.96070954e-01 -1.71896622e-01 -7.55567431e-01
1.07434464e+00 1.42415881e-01 -3.40837419e-01 -1.38911024e-01
-3.63077104e-01 -2.77391411e-02 7.17961311e-01 -1.17978299e+00
-7.18597412e-01 -5.85449815e-01 -3.20628464e-01 2.29816794e-01
-3.70400220e-01 -3.16961706e-01 4.14120913e-01 -2.30952591e-01
-2.88803279e-02 -8.82061124e-01 -4.57191497e-01 -1.30367413e-01
-2.76149046e-02 -2.91936636e-01 8.09490323e-01 -5.02351344e-01
1.14069283e+00 -1.75323462e+00 1.23301134e-01 7.51088023e-01
2.64789790e-01 3.81144136e-01 1.91021264e-01 6.53953969e-01
1.84701979e-01 2.32177065e-03 -2.15387672e-01 -3.19335550e-01
-2.75795788e-01 4.43512619e-01 -5.05123548e-02 5.28047264e-01
-3.99358422e-01 3.23576808e-01 -7.86334515e-01 -4.44481611e-01
2.21371204e-01 6.18003845e-01 -2.98194110e-01 -4.74128984e-02
-1.70589581e-01 5.75672686e-01 -5.45865774e-01 2.75114477e-01
6.46935761e-01 -1.97865784e-01 6.65796280e-01 5.33196665e-02
3.47498693e-02 -1.31412253e-01 -1.61388958e+00 1.25223517e+00
-4.29810226e-01 -2.73860060e-02 1.71476915e-01 -1.20934129e+00
1.20471740e+00 8.07662904e-01 8.61352980e-01 -2.81416714e-01
1.91575393e-01 6.86421037e-01 1.77990884e-01 -1.51831716e-01
9.61120501e-02 1.03741810e-01 2.88863987e-01 7.64189899e-01
-1.90204874e-01 2.55722761e-01 2.66657968e-04 -6.27742037e-02
1.20374596e+00 -6.64853007e-02 5.19770861e-01 -1.65300086e-01
7.60975242e-01 8.38962123e-02 8.30470800e-01 9.25311744e-01
3.11001260e-02 3.68100911e-01 2.40863889e-01 -6.82986259e-01
-6.66831613e-01 -5.28052151e-01 -7.85031691e-02 6.75742447e-01
9.44967493e-02 -4.82986391e-01 -7.89580762e-01 -9.24757004e-01
2.24321540e-02 6.43713772e-01 -3.97849500e-01 -9.48051549e-03
-2.45197281e-01 -9.19779718e-01 2.61584938e-01 -4.83303629e-02
3.88816088e-01 -9.70360398e-01 -1.05920529e+00 7.17652500e-01
-6.48492053e-02 -4.71557796e-01 1.09216072e-01 3.26078773e-01
-1.08667839e+00 -1.28006208e+00 -3.76431853e-01 -8.20717216e-02
7.34250188e-01 -2.26704001e-01 9.48736012e-01 4.00371522e-01
-1.17231235e-01 1.62317529e-01 -4.77543563e-01 -6.73581898e-01
-7.55609393e-01 4.17384468e-02 2.35662982e-01 4.56865996e-01
9.11175087e-02 -6.91802382e-01 -6.28126681e-01 6.34212434e-01
-7.62397468e-01 -3.27418596e-01 4.24925715e-01 1.14836013e+00
4.00425404e-01 4.74547029e-01 9.88151968e-01 -8.38830650e-01
6.59071624e-01 -4.43759203e-01 -4.97712225e-01 5.74093044e-01
-9.75536942e-01 -7.97297284e-02 9.51519668e-01 -3.30793977e-01
-8.52739632e-01 1.08897328e-01 1.62391424e-01 -3.84601980e-01
-2.38104343e-01 6.01486623e-01 2.70359684e-02 -2.68589854e-01
7.07580030e-01 2.18574554e-01 5.62767148e-01 -2.75630295e-01
-1.44065455e-01 9.65339720e-01 -3.26103368e-03 -4.94494945e-01
2.23904461e-01 3.50070834e-01 1.44683301e-01 -4.22636151e-01
-3.25477421e-01 -2.73727775e-01 -3.41331184e-01 -4.81548607e-01
2.78235853e-01 -6.04469478e-01 -1.03043306e+00 3.03679377e-01
-1.17020583e+00 2.94150084e-01 -4.00480032e-01 5.13533771e-01
-3.73162299e-01 2.86808342e-01 1.47294262e-02 -1.27329123e+00
-8.51735830e-01 -9.34810162e-01 3.86024237e-01 2.10663483e-01
-4.57652211e-01 -1.00908303e+00 1.24597661e-01 4.25977975e-01
7.13227153e-01 4.95134801e-01 3.23318303e-01 -1.26877880e+00
-2.72471279e-01 -5.60948789e-01 5.43062329e-01 4.46863323e-01
1.93563908e-01 -9.51269583e-04 -8.17427158e-01 -5.15643179e-01
2.62737215e-01 6.77499473e-02 4.54679519e-01 1.93180829e-01
7.53343701e-01 -4.20191705e-01 -5.09229183e-01 1.05571738e-02
1.37059259e+00 1.93226531e-01 3.89982700e-01 2.96394646e-01
1.32832397e-02 6.67584121e-01 6.57006204e-01 8.33139479e-01
1.25061467e-01 6.22457504e-01 5.30169845e-01 1.90717697e-01
3.15832376e-01 3.80832016e-01 2.10648447e-01 6.37291133e-01
-4.84227061e-01 -3.90829831e-01 -7.76640236e-01 2.78384447e-01
-2.33805203e+00 -8.87969494e-01 -2.74908334e-01 2.53331923e+00
4.94412363e-01 -9.63514373e-02 -1.38784692e-01 4.58253920e-01
8.67308259e-01 -3.27057481e-01 -6.26170158e-01 -6.82720304e-01
-1.10323533e-01 2.99557954e-01 2.89515555e-01 7.31900781e-02
-7.05247164e-01 5.09512722e-02 4.56814909e+00 7.36216784e-01
-1.21821737e+00 3.06250125e-01 4.95043010e-01 -1.92465588e-01
-8.03601667e-02 1.51142001e-01 -2.44095087e-01 7.12845147e-01
1.04963815e+00 -3.42965513e-01 3.33160847e-01 8.41004372e-01
3.36794645e-01 -3.75768423e-01 -9.14270580e-01 7.97016323e-01
-1.76782776e-02 -1.17918122e+00 -4.64529932e-01 3.12764376e-01
9.05477405e-01 6.58098310e-02 -3.65475386e-01 -2.19816327e-01
2.27480188e-01 -9.48665559e-01 3.74043882e-01 8.05598259e-01
5.28982878e-02 -9.00600195e-01 1.26853359e+00 8.43397856e-01
-6.76129043e-01 -3.86981905e-01 -1.56919792e-01 -5.30557148e-02
4.84572425e-02 1.08343542e+00 -8.81622434e-01 1.27166784e+00
7.75075316e-01 3.56218666e-01 -2.37965927e-01 1.43714821e+00
1.72231615e-01 5.97396433e-01 -7.78963268e-01 -2.77837157e-01
-8.32658410e-02 -1.16171405e-01 7.80277669e-01 5.33122540e-01
5.92659056e-01 -6.13765530e-02 3.70630957e-02 3.64486545e-01
2.04366967e-01 5.10962665e-01 -4.04475987e-01 4.31079984e-01
5.24918377e-01 1.30710292e+00 -5.82744122e-01 -2.66785264e-01
3.67940515e-02 6.54015183e-01 -3.53919752e-02 -1.00252472e-01
-5.24923503e-01 -2.65664235e-02 3.75532895e-01 4.66310270e-02
2.18892172e-01 2.60570914e-01 -3.40611279e-01 -9.25597191e-01
1.15401223e-01 -1.13937616e+00 7.93955803e-01 -4.24361855e-01
-1.51100552e+00 9.01441395e-01 -3.53161573e-01 -1.47444189e+00
-5.09327769e-01 -3.95260379e-02 -6.67746186e-01 7.78882861e-01
-1.00034428e+00 -7.98844576e-01 -2.93237776e-01 5.12943387e-01
3.37928347e-02 -2.06131831e-01 1.18212247e+00 4.11586873e-02
-3.46188307e-01 3.86005282e-01 6.74205959e-01 -5.78779221e-01
9.50459063e-01 -1.01588690e+00 -5.61796486e-01 5.91487408e-01
1.35611901e-02 4.10238117e-01 8.18049431e-01 -6.96734607e-01
-1.14161456e+00 -9.53323245e-01 1.03885722e+00 -2.43495256e-01
1.51623607e-01 1.76805064e-01 -9.18160856e-01 8.08254182e-02
2.71114141e-01 2.11976975e-01 6.74346983e-01 1.16002969e-01
2.86422789e-01 -6.91497922e-01 -1.57936847e+00 -1.70696620e-02
4.65091467e-01 2.92912245e-01 -2.98777521e-01 3.76230597e-01
-1.44050196e-01 -2.75827825e-01 -1.30155075e+00 5.85722864e-01
5.05485535e-01 -1.27520335e+00 5.04323423e-01 -1.82517499e-01
-5.79677634e-02 -3.80118221e-01 2.97945868e-02 -1.56112385e+00
1.84180483e-01 -5.38477659e-01 -1.85609877e-01 1.24590433e+00
2.54524231e-01 -1.33910418e+00 8.11755657e-01 7.24420071e-01
2.90799439e-01 -8.27690780e-01 -1.50656223e+00 -6.47677004e-01
-2.18146697e-01 -2.09203824e-01 8.19009066e-01 1.00108111e+00
2.45796498e-02 -3.06801528e-01 -3.50167722e-01 1.41907185e-01
9.54865277e-01 3.49958360e-01 7.02461898e-01 -1.56605208e+00
-3.36838722e-01 -2.02142477e-01 -5.61023414e-01 4.84230034e-02
4.83788410e-03 -7.58235991e-01 -1.40363589e-01 -1.38669717e+00
-2.68322323e-02 -9.11216736e-01 -5.62256813e-01 6.24734998e-01
4.91711125e-02 1.45644456e-01 3.23799312e-01 2.47455776e-01
-6.13177538e-01 2.51942545e-01 9.30915892e-01 -2.37536791e-04
-1.94137484e-01 4.69911933e-01 -4.45265830e-01 3.35729152e-01
1.00187933e+00 -8.73260498e-01 -2.69631147e-01 9.75802839e-02
1.61176443e-01 3.87533933e-01 2.57343858e-01 -1.20792139e+00
8.66813779e-01 -4.74700471e-03 3.87473583e-01 -3.12800258e-01
6.26731738e-02 -1.18982530e+00 1.02704835e+00 6.78751528e-01
-7.68298507e-02 -1.15526415e-01 -2.27673858e-01 6.50052309e-01
-2.98144490e-01 -2.65337944e-01 4.92591470e-01 -1.04891183e-02
-1.35704815e-01 -2.55577769e-02 -2.36257702e-01 -6.29152656e-01
1.47984207e+00 -3.83450985e-01 -1.89802706e-01 -2.95750409e-01
-9.31586146e-01 5.35697162e-01 4.72789884e-01 2.00806692e-01
3.12978715e-01 -7.59162903e-01 -9.86397922e-01 4.52685654e-02
-2.31051162e-01 1.18361324e-01 5.21795213e-01 1.34808016e+00
-2.59867638e-01 1.66434273e-01 -1.70364920e-02 -7.61141181e-01
-1.62851560e+00 1.78401381e-01 5.08321047e-01 -4.12543207e-01
-4.20698702e-01 2.69191653e-01 -5.95925212e-01 -6.70370221e-01
8.49443972e-02 1.82330638e-01 -6.24010041e-02 1.77429259e-01
3.68429571e-01 6.38683021e-01 7.85685122e-01 -4.26491827e-01
-6.24972761e-01 3.13755095e-01 1.69899955e-01 2.41251707e-01
1.54043031e+00 5.07250428e-03 -6.41501069e-01 1.18417613e-01
6.75460219e-01 -1.49936050e-01 -8.10774326e-01 4.40726951e-02
-1.28845572e-01 -2.54448891e-01 1.04855865e-01 -1.31164455e+00
-9.92229760e-01 3.14544737e-01 7.60086417e-01 2.91207522e-01
1.37642527e+00 -4.70896035e-01 4.71495926e-01 3.75709802e-01
7.74383068e-01 -9.72749412e-01 -6.61908031e-01 4.81734015e-02
5.56155920e-01 -1.06035781e+00 1.34299293e-01 3.05918306e-02
-5.42050898e-01 1.35405195e+00 3.60440969e-01 1.10977173e-01
5.62483013e-01 2.96137214e-01 1.03146665e-01 1.68973878e-01
-1.25147653e+00 3.35678875e-01 5.59563749e-02 2.94419885e-01
-1.70960110e-02 4.04638387e-02 -9.06623244e-01 5.28380454e-01
6.76063150e-02 3.82519066e-01 3.28614980e-01 1.06766713e+00
-2.20288128e-01 -1.47949874e+00 -8.27180564e-01 5.71914792e-01
-5.75591505e-01 4.26849097e-01 -2.39843041e-01 4.52613443e-01
4.87440616e-01 1.20205390e+00 -1.71894595e-01 -2.89882421e-01
1.50511622e-01 2.88764447e-01 2.72376418e-01 -2.88544577e-02
-1.17012489e+00 -1.81495488e-01 3.97920050e-02 -4.21019286e-01
-7.30552077e-01 -6.61139488e-01 -1.04270697e+00 -4.08600599e-01
-6.10619605e-01 8.62291515e-01 1.13396168e+00 9.04648840e-01
6.21673226e-01 4.08065856e-01 8.57561409e-01 -6.72403276e-01
-8.12817752e-01 -6.72047675e-01 -5.65408766e-01 2.91594982e-01
-3.50988843e-02 -3.42605680e-01 -5.84162533e-01 -2.61523753e-01]
|
[6.322873592376709, 6.287884712219238]
|
e823675b-70fc-460f-b4e5-e0aee6295347
|
cross-modal-information-fusion-for-voice
| null | null |
https://www.sciencedirect.com/science/article/pii/S0167639323000109
|
https://www.sciencedirect.com/science/article/pii/S0167639323000109
|
Cross-modal information fusion for voice spoofing detection
|
In recent years, speaker verification systems have been used in many production scenarios. Unfortunately, they are still very vulnerable to different kinds of spoofing attacks, such as speech synthesis attacks, replay attacks, etc. Researchers have proposed many methods to defend against these attacks, but in the existing methods, researchers just focus on speech features. In recent studies, researchers have found that speech contains a large amount of face information. In fact, we can determine the speaker's gender, age, mouth shape, and other information by voice. These information can help us distinguish spoofing attacks. Inspired by this phenomenon, we propose a generalized framework named GACMNet. To cope with different attack scenarios, we instantiated two different models. Our framework is mainly divided into data pre-processing phase, feature extraction phase, feature fusion phase, and classification phase. Specifically, our framework consists of two branches. On the one hand, we extract face features in speech by a convolutional neural network. On the other hand, we use a densely connected network to extract speech features. For the more, we designed a global attention-based information fusion mechanism to distinguish the importance of each part of the features. Our solution was proven to be effective in two large scenarios. Compared to the existing methods, our model improves the tandem decision cost function (t-DCF) and equal error rate (EER) scores by 9% and 11% in the logical access scenario, respectively, our model improves the EER score by 10% in the physical access scenario.
|
['Lei Shi', 'Bin Wu', 'Huawei Song', 'Hao Zhou', 'Junxiao Xue']
|
2023-02-01
| null | null | null |
journal-2023-2
|
['fake-voice-detection', 'voice-anti-spoofing', 'speaker-verification', 'speech-synthesis']
|
['audio', 'audio', 'speech', 'speech']
|
[ 3.45499218e-02 -3.09593767e-01 -4.06359211e-02 -2.99463212e-01
-3.33058655e-01 -2.27107838e-01 4.28102046e-01 6.86366707e-02
-3.13182741e-01 2.94193953e-01 2.77750671e-01 -3.36631298e-01
5.01971580e-02 -8.58232498e-01 -1.13097161e-01 -7.50758648e-01
1.42331734e-01 -2.06624418e-01 2.95732349e-01 -3.50330561e-01
2.64577478e-01 6.00591838e-01 -1.43120372e+00 4.98423539e-02
7.62814939e-01 1.29724836e+00 -1.53069541e-01 2.26225942e-01
-3.42182606e-01 3.88807297e-01 -9.82969046e-01 -5.11350095e-01
-2.57619303e-02 -3.65664333e-01 -4.72453237e-01 -5.00710979e-02
-3.27737451e-01 -4.50431198e-01 -5.26846945e-01 1.22730160e+00
9.27337945e-01 -2.34600082e-01 1.67050496e-01 -1.47162926e+00
-1.84195116e-01 7.37099588e-01 -6.35459363e-01 2.19381988e-01
4.92740542e-01 1.49850547e-01 4.73381817e-01 -7.42369711e-01
-6.82265684e-02 1.58562064e+00 5.36206484e-01 5.12896955e-01
-5.62089920e-01 -1.22034788e+00 6.35465011e-02 4.18255180e-01
-1.37225783e+00 -8.45129073e-01 1.01811945e+00 -3.37551385e-02
3.27026576e-01 3.41922283e-01 2.64753670e-01 1.09863758e+00
-9.68997702e-02 7.49258757e-01 9.75993454e-01 -3.38040084e-01
1.31677128e-02 2.48337939e-01 1.28623575e-01 4.28878188e-01
1.28811851e-01 2.32893869e-01 -2.42913291e-01 -2.80844510e-01
3.44488442e-01 1.60798296e-01 -5.05028427e-01 3.11653018e-01
-9.03709948e-01 8.03747714e-01 2.71492302e-01 4.88923639e-01
-1.74302921e-01 -3.24815720e-01 4.95709866e-01 2.65487254e-01
8.81767422e-02 -6.11854978e-02 -2.44301260e-01 1.01110697e-01
-6.46781206e-01 -2.67468113e-02 8.89130890e-01 4.32538092e-01
5.99919915e-01 4.60114883e-04 -1.34347990e-01 8.74798536e-01
7.03808069e-01 5.75846016e-01 7.69780695e-01 -2.82134414e-01
6.60615087e-01 2.49272510e-01 -2.16528431e-01 -1.42285681e+00
-4.23667490e-01 -4.07166272e-01 -9.94274735e-01 -1.77861467e-01
2.79165953e-02 -4.26276892e-01 -7.65365660e-01 1.75577736e+00
3.90263289e-01 3.14991891e-01 8.96635372e-03 7.40442157e-01
9.34138656e-01 5.00042617e-01 -5.57714216e-02 -5.13824999e-01
1.51331174e+00 -6.44243479e-01 -1.05010223e+00 6.79688230e-02
6.38821209e-03 -9.58785295e-01 4.93079275e-01 3.30267131e-01
-7.17628419e-01 -5.92146158e-01 -1.26422155e+00 4.75467891e-01
-4.20197308e-01 1.08641209e-02 4.95139897e-01 1.43588054e+00
-7.87920117e-01 3.60678434e-01 -5.25164902e-01 -2.29020983e-01
4.72529113e-01 5.76377332e-01 -2.71600276e-01 9.56850722e-02
-1.53725171e+00 4.48142976e-01 2.89223999e-01 2.10403368e-01
-7.71623373e-01 -6.73481822e-02 -7.24103630e-01 3.38204294e-01
4.48968232e-01 -2.49854311e-01 1.02683210e+00 -8.45214963e-01
-1.69984484e+00 3.33168805e-01 -1.48035914e-01 -3.12901139e-01
2.13232353e-01 7.61118233e-02 -1.20276082e+00 3.66152301e-02
-3.07429999e-01 1.01535402e-01 1.06426501e+00 -9.55874205e-01
-5.96979499e-01 -5.63914537e-01 3.95793542e-02 -1.20847687e-01
-7.09598064e-01 5.26757538e-01 -3.75243396e-01 -6.51844740e-01
1.86785951e-01 -5.86830020e-01 4.10781205e-02 -2.12808758e-01
-3.91781420e-01 -1.18031368e-01 1.28551555e+00 -8.53392959e-01
1.52964723e+00 -2.42244315e+00 -2.57789493e-01 4.26460385e-01
2.19811887e-01 8.06302011e-01 1.34043038e-01 1.93504304e-01
-7.01501220e-02 4.56614107e-01 -1.49060309e-01 -2.72874922e-01
-1.02298699e-01 -1.09996609e-01 -3.87148708e-01 3.19519877e-01
1.10004827e-01 2.80538172e-01 -5.90782762e-01 -5.47314286e-01
2.09852576e-01 6.90477788e-01 -4.56972122e-01 2.59362042e-01
3.03563327e-01 4.10421938e-01 -6.10504627e-01 6.20679677e-01
1.09446299e+00 2.79662788e-01 1.24454364e-01 -1.30842820e-01
7.39774182e-02 3.37621808e-01 -1.44796467e+00 1.24978745e+00
-4.38697636e-01 3.82372975e-01 5.91913879e-01 -9.39921975e-01
1.02499139e+00 6.75634861e-01 3.27060103e-01 -5.42618871e-01
6.49593174e-01 2.59416968e-01 2.20693663e-01 -5.57686150e-01
8.59192088e-02 2.20717378e-02 7.43554384e-02 3.37072402e-01
-6.27574623e-02 3.45907092e-01 -3.51170450e-01 -7.85354525e-02
9.90244985e-01 -5.59582293e-01 2.32513234e-01 8.11411142e-02
1.32339895e+00 -8.37610483e-01 7.77778149e-01 2.50754505e-01
-4.85848099e-01 3.46326619e-01 3.49562407e-01 -2.58611202e-01
-3.38086843e-01 -6.98928118e-01 -1.45467162e-01 7.11061478e-01
4.83643353e-01 -5.32399952e-01 -8.61095250e-01 -8.26211810e-01
-1.17470115e-01 2.54528463e-01 -1.83541074e-01 -4.64517772e-01
-5.79524994e-01 -8.24002266e-01 9.45397556e-01 1.61911830e-01
1.21000254e+00 -1.08764398e+00 -1.64255783e-01 1.00030228e-01
-2.68842548e-01 -1.12899971e+00 -4.84896064e-01 -2.55490661e-01
-3.63159657e-01 -9.00539696e-01 -5.35003304e-01 -5.76447964e-01
3.16403687e-01 4.17034984e-01 4.02482599e-01 4.18706328e-01
-8.53605792e-02 -1.68009564e-01 -5.19436777e-01 -5.71535587e-01
-4.39033836e-01 1.01247117e-01 3.07496846e-01 6.54628813e-01
3.38757843e-01 -6.14857972e-01 -5.37116826e-01 5.69254816e-01
-8.86418581e-01 -5.48558533e-01 4.96880084e-01 6.19811654e-01
-2.58002579e-01 4.03766900e-01 6.83157861e-01 -3.96584958e-01
8.39897335e-01 -5.28591275e-01 -3.12086821e-01 1.88107267e-01
-4.00909454e-01 1.82750151e-02 6.67501390e-01 -4.16577220e-01
-9.35802102e-01 -2.31745392e-01 -7.17894912e-01 -2.06967562e-01
-3.00351560e-01 2.60274559e-01 -8.09077322e-01 -2.67171204e-01
2.39427775e-01 3.00440103e-01 1.74498066e-01 -5.90000451e-01
-1.67205676e-01 1.46502352e+00 1.97311014e-01 -3.47474366e-01
9.95078146e-01 2.63413519e-01 -2.69348323e-01 -8.68890107e-01
-2.39631459e-01 -1.84724763e-01 -9.34978947e-02 -3.50554623e-02
6.83831334e-01 -8.02761555e-01 -1.16030765e+00 8.76623094e-01
-1.13352573e+00 6.09382153e-01 4.57615346e-01 5.67775786e-01
9.94311497e-02 8.41995656e-01 -5.16528189e-01 -1.07528698e+00
-6.15245581e-01 -1.36006129e+00 8.20666254e-01 5.67500174e-01
2.74459302e-01 -4.66883034e-01 -2.97595471e-01 3.18980873e-01
7.82500088e-01 -2.10574847e-02 6.66158080e-01 -9.75682020e-01
-1.59887940e-01 -2.08156601e-01 -2.55619556e-01 3.69687647e-01
4.44250911e-01 -1.27740249e-01 -1.20664382e+00 -3.57987970e-01
4.00279164e-01 1.23525456e-01 7.77427197e-01 3.59543748e-02
1.34240198e+00 -3.57437462e-01 -5.05617499e-01 6.10887825e-01
8.88446987e-01 6.55771017e-01 8.00514758e-01 -1.31638631e-01
4.82835442e-01 5.02403557e-01 3.35422128e-01 4.45281625e-01
1.07161604e-01 8.76015604e-01 4.74661976e-01 1.48484603e-01
-1.58076342e-02 -8.40669945e-02 6.44607127e-01 6.88018441e-01
9.74680930e-02 -2.63827562e-01 -6.67167783e-01 2.04630464e-01
-1.44404471e+00 -1.09079540e+00 3.56346905e-01 2.32085109e+00
5.63419521e-01 3.22258115e-01 2.52921104e-01 6.67184770e-01
1.13057184e+00 3.35542142e-01 -3.38914841e-01 -3.33979875e-01
3.54060233e-02 1.62605628e-01 2.30181843e-01 2.90025085e-01
-1.14825165e+00 6.85594678e-01 5.40428925e+00 1.05931830e+00
-1.60837984e+00 1.56592786e-01 5.81001759e-01 1.60027191e-01
-5.47188520e-02 -9.02475715e-02 -8.44769955e-01 9.27112937e-01
8.46462607e-01 -3.90719324e-02 4.41215485e-01 5.47258079e-01
-1.98956430e-02 3.10972303e-01 -4.12593007e-01 1.30915105e+00
3.32104653e-01 -8.99212897e-01 -6.32131398e-02 2.11072624e-01
-8.10902864e-02 -3.01315695e-01 -3.19242142e-02 3.02671313e-01
8.12918544e-02 -9.96687233e-01 4.15838212e-01 8.48063678e-02
6.24630034e-01 -9.88712788e-01 9.92495596e-01 3.10911089e-01
-1.32501972e+00 -3.87758613e-01 -1.14562303e-01 1.10300682e-01
1.94013312e-01 6.83368087e-01 -4.81877118e-01 9.41133261e-01
5.38324654e-01 2.14056522e-01 -3.61031651e-01 1.04562211e+00
-2.72877693e-01 5.87445140e-01 -4.05148536e-01 -3.69768366e-02
-1.44829512e-01 1.96057990e-01 4.75589365e-01 8.52591753e-01
4.18597847e-01 7.63466731e-02 2.26612687e-01 3.96914959e-01
-1.19280592e-01 1.84663147e-01 -3.69432628e-01 1.07480310e-01
7.69763350e-01 1.26050162e+00 -4.19049919e-01 -1.33467600e-01
-2.58275568e-01 7.95448065e-01 -2.51896560e-01 1.21015765e-01
-9.42373574e-01 -9.68867183e-01 6.79137230e-01 -1.77112836e-02
1.79522112e-01 -8.51293504e-02 4.75289039e-02 -1.16650844e+00
1.01319715e-01 -1.11157465e+00 2.24294901e-01 -2.61239946e-01
-9.02797401e-01 8.20880055e-01 -4.19889778e-01 -1.04480863e+00
-6.11411706e-02 -4.16035056e-01 -7.72948265e-01 9.70694959e-01
-1.49492383e+00 -8.47942770e-01 -3.82708877e-01 8.15704048e-01
3.83192718e-01 -4.11254257e-01 7.86252260e-01 7.44178712e-01
-1.06040204e+00 1.00512516e+00 -2.18982339e-01 6.53279364e-01
6.37491107e-01 -5.19653380e-01 3.36757809e-01 8.14018250e-01
3.81562077e-02 7.98393250e-01 3.76194566e-01 -3.00503880e-01
-1.13996947e+00 -6.95664763e-01 7.73049235e-01 2.81662285e-01
3.04664642e-01 -2.69037664e-01 -7.51986861e-01 1.86539993e-01
1.00641847e-01 5.38278818e-02 5.66454113e-01 -7.01095015e-02
-5.53015530e-01 -3.89503717e-01 -1.47394371e+00 4.56751376e-01
7.71373272e-01 -5.65237284e-01 -4.05308455e-01 -3.96416672e-02
6.84665203e-01 -1.39339000e-01 -6.12181365e-01 5.30244887e-01
5.87488711e-01 -9.69909668e-01 9.00397897e-01 -2.95834601e-01
-5.37600555e-02 -4.43252712e-01 -1.84183955e-01 -1.00180197e+00
-1.28611878e-01 -8.92476261e-01 -1.25106037e-01 1.61611712e+00
1.59989342e-01 -9.72558439e-01 5.10269284e-01 1.13662153e-01
8.67303237e-02 -5.82045555e-01 -1.00853634e+00 -6.43825352e-01
-2.96235949e-01 -3.10938388e-01 1.20756269e+00 8.77008259e-01
7.83912167e-02 3.56660664e-01 -4.32111144e-01 4.58120376e-01
3.18511903e-01 -1.44967049e-01 6.31380558e-01 -1.35062945e+00
-2.42890924e-01 -6.31891847e-01 -4.88904953e-01 -9.57860827e-01
1.25965968e-01 -5.30333221e-01 -2.66153902e-01 -9.89825010e-01
-3.41950580e-02 -4.84566450e-01 -5.29433668e-01 4.02675658e-01
-2.70428210e-01 -8.60771835e-02 2.79109329e-01 -3.01024076e-02
-9.85259116e-02 4.45074350e-01 7.52826929e-01 -2.86777079e-01
-1.57634705e-01 2.93894291e-01 -7.17294991e-01 6.17954493e-01
9.59172070e-01 -2.77530342e-01 -1.58999979e-01 -1.65159434e-01
-2.61191159e-01 1.66308358e-01 1.51014760e-01 -1.29805923e+00
2.82338321e-01 -1.66075560e-03 2.42267817e-01 -3.37562829e-01
3.54206532e-01 -9.90331233e-01 -1.17651701e-01 7.12117314e-01
1.19383834e-01 -1.46510527e-01 1.46408275e-01 3.43244195e-01
-4.38122034e-01 -8.82379487e-02 7.09871590e-01 1.57877371e-01
-3.49614888e-01 3.98758560e-01 -2.53053427e-01 -4.42482173e-01
9.14372802e-01 -3.04998853e-03 -4.47902411e-01 -5.36299944e-01
-4.07218724e-01 1.39742106e-01 6.59824535e-02 6.36479020e-01
6.31125212e-01 -1.24605024e+00 -5.87725401e-01 7.33308494e-01
-1.90513000e-01 -4.52585787e-01 2.94580698e-01 9.18516874e-01
-2.89212137e-01 3.55517536e-01 -2.14257360e-01 -2.62724370e-01
-1.61536217e+00 7.80917943e-01 2.97815770e-01 -1.64136037e-01
-1.40090570e-01 6.19119465e-01 -1.80317760e-01 -1.78626150e-01
3.52301300e-01 7.40522286e-03 -5.32410741e-01 1.89019069e-01
9.35340226e-01 1.90397963e-01 1.26734987e-01 -8.74223232e-01
-5.92707872e-01 5.89460254e-01 -1.78168565e-01 -1.10149004e-01
8.38315308e-01 -7.84701928e-02 -9.62603316e-02 -2.22376511e-01
1.22752964e+00 4.99141961e-01 -5.29531002e-01 -3.16068292e-01
-1.63450435e-01 -4.03683335e-01 1.19470209e-01 -6.53331697e-01
-1.48991275e+00 1.08280206e+00 7.61500120e-01 6.08293533e-01
1.40439010e+00 -4.13748682e-01 1.13108408e+00 3.87963802e-02
4.97380495e-01 -6.87355876e-01 -2.20600069e-01 4.09888059e-01
6.05357349e-01 -1.15649116e+00 -2.32912883e-01 -6.04144335e-01
-3.21334690e-01 1.04070866e+00 4.25879091e-01 2.87502646e-01
1.06040692e+00 2.62036741e-01 6.87725171e-02 1.23243622e-01
-2.20326900e-01 -1.93955466e-01 4.65884507e-02 5.29744327e-01
1.49321988e-01 -1.68553658e-03 -4.52269375e-01 9.22516823e-01
-2.50737488e-01 -2.53742039e-01 1.72322407e-01 6.52457297e-01
-5.24351478e-01 -1.36647153e+00 -7.07555115e-01 1.15130827e-01
-7.95012534e-01 2.50084307e-02 -3.28539073e-01 2.93120593e-01
3.05706590e-01 1.59536994e+00 -1.14946432e-01 -1.10826850e+00
1.85311601e-01 -2.98053063e-02 -4.45221178e-02 -2.41688639e-01
-6.94991648e-01 -6.72296956e-02 -3.56838368e-02 -4.42517579e-01
-3.58553618e-01 -4.49210495e-01 -1.11983931e+00 -5.94173908e-01
-5.84478259e-01 4.25370187e-01 8.47032368e-01 1.05974042e+00
4.24854547e-01 5.88492334e-01 1.25623691e+00 -4.58221436e-01
-5.26789665e-01 -1.09005177e+00 -4.10760164e-01 2.70414799e-01
4.86438870e-01 -7.18071878e-01 -4.86832976e-01 -4.97773677e-01]
|
[13.974494934082031, 5.748691082000732]
|
718bb9dd-6a39-4075-9ec5-5413ac7b7922
|
learning-to-learn-with-compound-hd-models
| null | null |
http://papers.nips.cc/paper/4474-learning-to-learn-with-compound-hd-models
|
http://papers.nips.cc/paper/4474-learning-to-learn-with-compound-hd-models.pdf
|
Learning to Learn with Compound HD Models
|
We introduce HD (or ``Hierarchical-Deep'') models, a new compositional learning architecture that integrates deep learning models with structured hierarchical Bayesian models. Specifically we show how we can learn a hierarchical Dirichlet process (HDP) prior over the activities of the top-level features in a Deep Boltzmann Machine (DBM). This compound HDP-DBM model learns to learn novel concepts from very few training examples, by learning low-level generic features, high-level features that capture correlations among low-level features, and a category hierarchy for sharing priors over the high-level features that are typical of different kinds of concepts. We present efficient learning and inference algorithms for the HDP-DBM model and show that it is able to learn new concepts from very few examples on CIFAR-100 object recognition, handwritten character recognition, and human motion capture datasets.
|
['Antonio Torralba', 'Ruslan R. Salakhutdinov', 'Joshua B. Tenenbaum']
|
2011-12-01
| null | null | null |
neurips-2011-12
|
['novel-concepts']
|
['reasoning']
|
[-2.21644446e-01 2.23173738e-01 -1.54791594e-01 -7.99110115e-01
-5.26462853e-01 -1.66561473e-02 1.09612012e+00 2.13632241e-01
-6.68418467e-01 5.20654500e-01 4.39584523e-01 2.95275390e-01
-2.39765465e-01 -1.02659011e+00 -8.20070446e-01 -9.49097037e-01
-6.14852965e-01 9.48468089e-01 6.40459657e-01 3.17865849e-01
1.67057648e-01 4.85842675e-01 -1.51968050e+00 6.71549499e-01
1.80441570e-02 9.53450382e-01 3.78603429e-01 7.88628876e-01
-1.25403687e-01 9.59352851e-01 -3.30046177e-01 -1.92388177e-01
-1.80325419e-01 -5.92258796e-02 -8.78946602e-01 -9.29132178e-02
4.38270450e-01 -6.35998666e-01 -4.05924171e-01 6.82125390e-01
1.00378208e-01 4.18341905e-01 1.28759801e+00 -9.70596731e-01
-7.71500051e-01 4.40181226e-01 -3.34907562e-01 4.57755923e-01
-8.35978091e-02 2.86707412e-02 1.29418898e+00 -1.08766913e+00
2.92545438e-01 1.53594530e+00 8.32619429e-01 7.54387021e-01
-1.52877724e+00 -2.48512506e-01 2.98057824e-01 4.29904073e-01
-1.23517311e+00 5.22926040e-02 4.88616079e-01 -9.47610497e-01
1.25610077e+00 -3.55177075e-01 8.15686226e-01 1.45077443e+00
3.51957679e-01 1.03490579e+00 8.15527678e-01 -3.79850239e-01
7.50580788e-01 -4.53860760e-02 5.90883553e-01 8.29875052e-01
1.60103142e-01 7.37278983e-02 -9.46844459e-01 -5.07315516e-01
9.32517588e-01 2.74622858e-01 2.32429832e-01 -6.16521239e-01
-9.78199959e-01 1.15384901e+00 4.36450720e-01 4.81965899e-01
-5.01724124e-01 5.06986201e-01 1.10010616e-01 -3.63648593e-01
6.39707148e-02 1.00132748e-01 -6.45774961e-01 2.86577791e-02
-7.33357012e-01 1.88442037e-01 8.28993618e-01 7.98437834e-01
1.28732932e+00 9.60221067e-02 -2.86539793e-01 8.77003193e-01
6.72837257e-01 1.85810611e-01 7.43140340e-01 -9.85206664e-01
-1.10903971e-01 -1.48299299e-02 2.58494895e-02 -6.84480608e-01
-3.49712014e-01 -2.43705928e-01 -8.48494709e-01 -5.66411577e-02
-1.02938227e-02 1.51027784e-01 -1.12533808e+00 1.75674093e+00
-1.84036776e-01 1.31502658e-01 -1.36273757e-01 4.77858245e-01
7.19860733e-01 1.05034792e+00 4.94156659e-01 4.08905298e-02
1.28958929e+00 -7.55410969e-01 -4.59257811e-01 -3.00675035e-01
1.99381039e-01 1.14265919e-01 9.45333421e-01 4.77968395e-01
-1.03803313e+00 -8.98113132e-01 -8.78814697e-01 -2.68926293e-01
-4.74680871e-01 -1.37143582e-01 8.03608119e-01 4.74159211e-01
-1.13348949e+00 8.24344158e-01 -1.16187906e+00 -4.06580895e-01
7.25780964e-01 3.15008491e-01 -1.78207308e-01 -2.50871450e-01
-1.21210968e+00 8.29620719e-01 5.24797499e-01 -1.86244160e-01
-1.91943169e+00 -6.54292881e-01 -8.00205410e-01 3.37867349e-01
-1.79190010e-01 -8.39944482e-01 1.25479090e+00 -5.62062502e-01
-1.68204904e+00 9.04812813e-01 4.60254680e-03 -5.32064795e-01
-2.59284586e-01 -3.65556121e-01 7.85237625e-02 2.52488703e-01
-2.21967667e-01 1.13608801e+00 1.26222861e+00 -1.29696345e+00
-7.07333565e-01 -4.43899900e-01 -2.51714945e-01 -1.18735863e-03
-3.49497944e-01 -2.71830380e-01 -1.74779773e-01 -5.41855395e-01
8.80767033e-02 -5.82681239e-01 -1.45804256e-01 -9.74571481e-02
-9.56482515e-02 -6.07305884e-01 4.21753794e-01 -5.41728318e-01
7.17197776e-01 -2.23911619e+00 3.44074279e-01 1.44280374e-01
7.97359273e-02 -1.82632193e-01 -7.66298398e-02 1.04449831e-01
1.93377823e-01 -7.58196488e-02 -2.15453014e-01 -3.29874277e-01
3.72050166e-01 7.79888391e-01 -4.99232888e-01 2.33397216e-01
2.60935456e-01 7.47167528e-01 -8.89763534e-01 -1.06920131e-01
5.27615808e-02 3.95771533e-01 -7.93913066e-01 2.74335295e-01
-3.53143573e-01 1.85760111e-01 -1.28267124e-01 1.70295253e-01
2.75207341e-01 -2.17684761e-01 5.48654385e-02 1.93678010e-02
2.87080079e-01 2.85707414e-01 -7.91023731e-01 1.64304602e+00
-2.25048169e-01 5.08616865e-01 -3.32227916e-01 -1.15545738e+00
8.81945908e-01 1.32435456e-01 1.22455791e-01 -2.90762693e-01
-2.41640016e-01 -3.26783091e-01 -1.81789786e-01 -2.79174358e-01
1.39309108e-01 -6.43521845e-01 -2.40980655e-01 2.27358446e-01
9.37572539e-01 -3.21491718e-01 -3.11466843e-01 1.61252916e-01
1.11831415e+00 -7.12690875e-02 2.70768672e-01 -4.06053811e-01
1.04606397e-01 -4.67178822e-01 7.81776488e-01 1.29859686e+00
-1.08761534e-01 4.28031266e-01 5.47894180e-01 -8.13492596e-01
-9.33775961e-01 -1.70359254e+00 -1.74104720e-01 1.71108699e+00
-4.69452024e-01 -3.89081597e-01 -3.98311019e-01 -2.76128888e-01
1.38599738e-01 8.17324102e-01 -8.82415950e-01 -3.56292158e-01
-2.13968873e-01 -9.92003083e-01 3.61634970e-01 8.85555923e-01
4.72796589e-01 -1.12197638e+00 -6.57421410e-01 3.24453115e-01
3.96306574e-01 -9.23324645e-01 3.09124649e-01 7.31262922e-01
-1.16594529e+00 -5.83958983e-01 -6.28573120e-01 -9.02940691e-01
4.63160485e-01 -4.79238540e-01 1.17059600e+00 -4.68852133e-01
-6.64199233e-01 6.36307836e-01 -4.06020023e-02 -2.85498321e-01
-2.50155535e-02 -1.50383011e-01 2.10009769e-01 9.58722234e-02
4.70185548e-01 -6.74502254e-01 -3.15542281e-01 -1.10606678e-01
-8.89856994e-01 -9.89309698e-02 5.76041579e-01 1.09640145e+00
4.18173492e-01 9.79162604e-02 3.30216467e-01 -6.14587963e-01
2.61544406e-01 -5.04834294e-01 -4.92466271e-01 1.24484086e-02
-8.87818336e-02 2.90030092e-01 2.43070990e-01 -3.72561246e-01
-1.33424699e+00 -8.71744659e-03 -1.27534568e-01 -3.40040326e-01
-5.51573753e-01 4.92923141e-01 -1.99772343e-01 4.51228797e-01
6.79853618e-01 4.73441899e-01 -6.02915585e-01 -7.76010931e-01
5.93740582e-01 1.74369052e-01 6.35199428e-01 -1.18829918e+00
4.59092528e-01 6.74904764e-01 -4.82235774e-02 -9.83479679e-01
-1.05698121e+00 -3.00801486e-01 -1.12002873e+00 1.81882590e-01
1.34055102e+00 -1.08692586e+00 -3.91927660e-01 9.36189294e-01
-1.08120334e+00 -7.18532562e-01 -5.70802331e-01 6.64078355e-01
-9.33501840e-01 2.47588933e-01 -1.22615337e+00 -6.97504878e-01
2.45556250e-01 -6.60084307e-01 9.25271928e-01 2.06427306e-01
-2.24487215e-01 -1.15920293e+00 4.43194628e-01 3.18789296e-02
2.56418943e-01 -2.02311948e-01 1.68841934e+00 -8.14554691e-01
-6.50344253e-01 -4.21091057e-02 9.54673216e-02 6.17877781e-01
-1.26655564e-01 -6.23103008e-02 -1.18073654e+00 -3.27841282e-01
6.57148883e-02 -6.04632676e-01 1.47326624e+00 4.89342213e-01
1.50878751e+00 -3.53177369e-01 -3.39033931e-01 4.33663934e-01
1.16704595e+00 4.65508699e-02 5.54248750e-01 -5.39685600e-02
6.15873754e-01 4.21952695e-01 -1.64669119e-02 7.84353435e-01
4.63210970e-01 6.05321825e-01 1.77110285e-01 2.72790223e-01
2.53387034e-01 -2.88532138e-01 4.77388352e-01 8.39862049e-01
-7.72733390e-02 4.42080855e-01 -1.23816586e+00 7.45579123e-01
-1.80234098e+00 -1.00437391e+00 3.63370478e-01 1.97400963e+00
1.22582698e+00 2.87903488e-01 -4.22222093e-02 -4.09519345e-01
7.03338087e-01 1.50179505e-01 -4.87307638e-01 -3.92188191e-01
2.71586590e-02 6.54733658e-01 -9.18734744e-02 4.46728349e-01
-1.32650065e+00 9.46142256e-01 7.00983238e+00 6.47897184e-01
-5.28057516e-01 3.34602803e-01 3.35512936e-01 3.15231591e-04
-1.59527734e-01 -5.66246919e-02 -1.18172836e+00 2.26701289e-01
1.10454702e+00 1.45924523e-01 9.78962481e-02 1.10728693e+00
-4.72278357e-01 -1.18732624e-01 -1.62829936e+00 9.23809350e-01
1.57468736e-01 -1.34264541e+00 5.41126251e-01 -1.16831847e-02
7.92871594e-01 3.00323784e-01 2.65674204e-01 8.84214759e-01
1.09499788e+00 -9.85118389e-01 4.43822891e-01 9.88822520e-01
1.30690426e-01 -5.23924470e-01 3.86855900e-01 6.54244184e-01
-5.65514326e-01 -3.21157366e-01 -8.65069211e-01 -1.14789300e-01
-2.51773447e-01 6.34272695e-01 -5.76438308e-01 -1.09238088e-01
1.02219892e+00 8.17922473e-01 -4.10358101e-01 9.69962597e-01
-4.79577810e-01 6.83121860e-01 -2.07503751e-01 1.50197521e-01
4.83603865e-01 2.75072545e-01 1.21561378e-01 1.38464451e+00
1.01527013e-01 2.05745682e-01 1.77487239e-01 1.03542638e+00
-5.64800575e-02 -4.37418997e-01 -3.50731343e-01 2.33468145e-01
2.26449326e-01 1.05024409e+00 -3.91322643e-01 -7.48594522e-01
-2.65679747e-01 8.74664843e-01 5.48900604e-01 3.49055886e-01
-5.36681056e-01 -7.49919489e-02 6.36590838e-01 -1.16836496e-01
8.56412709e-01 -4.24274951e-01 8.01395550e-02 -1.23898304e+00
-6.67504191e-01 -3.63390207e-01 6.54307783e-01 -8.82737339e-01
-1.75547576e+00 1.64515868e-01 3.42800826e-01 -4.63181078e-01
-3.42022061e-01 -8.79975915e-01 -5.95835626e-01 7.04636931e-01
-1.29784000e+00 -9.14770663e-01 -6.77032396e-02 8.84328365e-01
4.14233923e-01 -3.45599532e-01 1.26145303e+00 -2.33862832e-01
-2.71366924e-01 9.92015079e-02 4.36633766e-01 3.59978229e-01
4.22851831e-01 -1.40528882e+00 3.70247573e-01 2.56267428e-01
4.22264189e-01 8.54134977e-01 3.34936351e-01 -3.71019125e-01
-7.87514985e-01 -9.84840989e-01 6.95114911e-01 -6.64578319e-01
5.52198470e-01 -6.17484093e-01 -1.19172394e+00 9.17864561e-01
3.28335240e-02 2.01502249e-01 1.13161516e+00 3.98154527e-01
-8.53696525e-01 -1.28065839e-01 -9.08632636e-01 9.13941190e-02
5.96434832e-01 -8.96026671e-01 -1.43750572e+00 2.38265291e-01
2.90588349e-01 2.81244725e-01 -9.58203852e-01 2.77009308e-01
5.71686685e-01 -9.25094545e-01 9.62989330e-01 -1.08922732e+00
1.64056510e-01 -1.71096791e-02 -5.15265584e-01 -1.23121512e+00
-9.82077122e-01 -8.88812914e-02 -4.60919797e-01 8.02213848e-01
5.71872331e-02 -1.32064834e-01 6.48858845e-01 6.43437028e-01
-8.52130726e-02 -5.24408460e-01 -1.03706157e+00 -8.47239316e-01
5.95925748e-01 -3.51144731e-01 1.13055460e-01 9.17253613e-01
3.73754501e-02 4.61750388e-01 -2.43303478e-01 1.81888953e-01
9.19909120e-01 -9.45659950e-02 4.86341774e-01 -1.65226173e+00
-7.11381257e-01 -2.12067395e-01 -7.85966814e-01 -1.04516649e+00
4.65984404e-01 -7.60764182e-01 4.00427669e-01 -1.50928140e+00
9.03573871e-01 -2.10204802e-04 -5.43421984e-01 6.72239244e-01
1.15383081e-01 -1.21947132e-01 7.22344443e-02 3.20329994e-01
-8.28752637e-01 8.63266885e-01 5.30904412e-01 -2.19978437e-01
2.14894503e-01 -1.57957956e-01 -3.99011336e-02 1.21606076e+00
4.56901252e-01 -6.31748021e-01 -1.07225724e-01 -5.10682285e-01
1.43031031e-01 -3.25248748e-01 7.44656801e-01 -1.19987345e+00
3.27878147e-01 -8.09139907e-02 1.01291299e+00 -6.97625637e-01
5.40892065e-01 -5.49822211e-01 -2.03254804e-01 4.53394920e-01
-5.63910007e-01 -6.47555113e-01 7.57710636e-02 8.44170451e-01
-1.86853662e-01 -2.98249990e-01 1.06446564e+00 -6.07048094e-01
-1.14025509e+00 3.66909832e-01 -8.84813309e-01 -1.21790737e-01
7.94136941e-01 1.21422168e-02 -1.65625215e-01 -2.81716555e-01
-1.55933213e+00 -2.59796549e-02 2.36220565e-03 3.93397033e-01
6.38595521e-01 -1.36542606e+00 -3.79930943e-01 2.76374012e-01
1.39003336e-01 9.30158049e-02 2.46446311e-01 2.68133402e-01
-7.25875199e-02 5.56778669e-01 -6.19216025e-01 -8.39475453e-01
-5.30111849e-01 4.42345768e-01 3.00205231e-01 -2.91360110e-01
-4.76667255e-01 1.26001346e+00 7.53178179e-01 -3.40010017e-01
4.98935759e-01 -5.49706280e-01 -6.83367923e-02 1.53275535e-01
4.10195529e-01 1.24111831e-01 -1.90599695e-01 -3.33598644e-01
-3.82452518e-01 3.23910296e-01 -3.74539703e-01 -3.52393895e-01
1.54249477e+00 5.12258895e-02 -1.67692319e-01 1.12286937e+00
1.10438454e+00 -9.34469402e-01 -1.69591951e+00 -5.83365321e-01
4.53023016e-01 -9.35274810e-02 9.95710790e-02 -5.24101079e-01
-4.89412934e-01 1.53320396e+00 5.15068293e-01 -3.12829643e-01
6.53185844e-01 4.11602944e-01 2.58197337e-01 9.78109479e-01
5.53195357e-01 -1.23799562e+00 6.88442290e-01 1.08419752e+00
6.83705330e-01 -8.12622666e-01 -1.30662620e-01 3.71352196e-01
-3.92400265e-01 1.15814984e+00 5.26968718e-01 -3.65174502e-01
1.12961543e+00 4.37105112e-02 -6.47949696e-01 -2.72605151e-01
-1.11557734e+00 -1.49738923e-01 1.89684317e-01 7.94767261e-01
9.19091180e-02 4.54359048e-04 4.64211524e-01 1.15084302e+00
1.33478731e-01 -3.43104079e-02 2.40200832e-01 8.98404002e-01
-1.03517580e+00 -6.62097335e-01 -7.94296563e-02 3.98767352e-01
-2.94874400e-01 -2.22790435e-01 -5.08659240e-03 4.83712375e-01
1.80433378e-01 1.27753034e-01 6.05289638e-01 -1.20976061e-01
-6.50551245e-02 7.03332961e-01 8.63561988e-01 -1.22501850e+00
-8.92334208e-02 -1.77595779e-01 -2.66521037e-01 -4.43526268e-01
-3.78264368e-01 -7.80078471e-01 -1.25787628e+00 1.60817757e-01
2.11721256e-01 1.35390162e-01 5.37098110e-01 1.14349580e+00
1.37148544e-01 2.69759089e-01 3.47351730e-01 -1.26194525e+00
-7.86815882e-01 -1.09073281e+00 -1.05203164e+00 3.91574085e-01
3.46509218e-01 -7.52260387e-01 -3.34142596e-01 5.01022577e-01]
|
[9.263550758361816, 2.9338130950927734]
|
9fddd881-a9fb-428c-9deb-9643a0ae435f
|
instance-variant-loss-with-gaussian-rbf
|
2305.04239
| null |
https://arxiv.org/abs/2305.04239v1
|
https://arxiv.org/pdf/2305.04239v1.pdf
|
Instance-Variant Loss with Gaussian RBF Kernel for 3D Cross-modal Retriveal
|
3D cross-modal retrieval is gaining attention in the multimedia community. Central to this topic is learning a joint embedding space to represent data from different modalities, such as images, 3D point clouds, and polygon meshes, to extract modality-invariant and discriminative features. Hence, the performance of cross-modal retrieval methods heavily depends on the representational capacity of this embedding space. Existing methods treat all instances equally, applying the same penalty strength to instances with varying degrees of difficulty, ignoring the differences between instances. This can result in ambiguous convergence or local optima, severely compromising the separability of the feature space. To address this limitation, we propose an Instance-Variant loss to assign different penalty strengths to different instances, improving the space separability. Specifically, we assign different penalty weights to instances positively related to their intra-class distance. Simultaneously, we reduce the cross-modal discrepancy between features by learning a shared weight vector for the same class data from different modalities. By leveraging the Gaussian RBF kernel to evaluate sample similarity, we further propose an Intra-Class loss function that minimizes the intra-class distance among same-class instances. Extensive experiments on three 3D cross-modal datasets show that our proposed method surpasses recent state-of-the-art approaches.
|
['Heng Tao Shen', 'Ning Xie', 'Zhenjiang Du', 'Guan Wang', 'Jiwei Wei', 'Zengyu Liu', 'Zhitao Liu']
|
2023-05-07
| null | null | null | null |
['cross-modal-retrieval']
|
['miscellaneous']
|
[-6.16576895e-02 -5.67202628e-01 -2.67610908e-01 -2.72976816e-01
-1.10893607e+00 -7.86469460e-01 5.44750631e-01 3.87418181e-01
-2.91709036e-01 1.76501602e-01 9.41297039e-02 1.81464225e-01
-5.86869895e-01 -7.36002386e-01 -3.45972478e-01 -8.94655108e-01
3.39998752e-02 1.63762555e-01 1.07938908e-01 2.82895211e-02
2.94874966e-01 5.64831793e-01 -1.54088736e+00 2.77697325e-01
9.11782682e-01 1.19432783e+00 -6.50231466e-02 -8.72637033e-02
-3.08859199e-01 -1.24125056e-01 -3.96503419e-01 -2.19040215e-01
4.29106355e-01 -2.97218412e-02 -4.36147064e-01 2.82260239e-01
6.02912784e-01 7.85632059e-02 -4.70517635e-01 1.08856225e+00
5.06014347e-01 2.21014887e-01 9.15987611e-01 -1.44857800e+00
-8.41230333e-01 -1.44993305e-01 -8.29799891e-01 5.87219093e-03
4.40855443e-01 -2.09606096e-01 1.15204716e+00 -1.20759320e+00
3.47757965e-01 1.24275064e+00 4.29347903e-01 2.75590569e-01
-1.43047404e+00 -5.90165913e-01 2.27989957e-01 1.82527244e-01
-1.86666381e+00 -2.04222262e-01 1.21399903e+00 -5.13509095e-01
4.03206736e-01 4.33919668e-01 5.20306349e-01 7.73039877e-01
-1.17202386e-01 9.04219449e-01 8.61410439e-01 -2.95211881e-01
7.70667121e-02 2.57081985e-01 -5.37653863e-02 4.89228427e-01
1.96274538e-02 -2.24068537e-01 -5.18882513e-01 -4.75900710e-01
6.39383733e-01 4.53983873e-01 -5.95279455e-01 -9.82064009e-01
-1.35882044e+00 8.06664586e-01 5.54401219e-01 3.69061768e-01
-2.21016198e-01 -2.53495574e-01 4.78209913e-01 2.81342924e-01
5.57379842e-01 8.79481360e-02 -2.44217008e-01 1.56782374e-01
-5.57212770e-01 1.62909657e-01 3.27550918e-01 8.40170205e-01
8.20686340e-01 -4.69888717e-01 -1.91394046e-01 1.29177916e+00
5.14897346e-01 5.03119946e-01 4.00499135e-01 -5.58404207e-01
7.32592285e-01 1.01652932e+00 -1.20311677e-02 -1.59578419e+00
-1.35140330e-01 -2.05335066e-01 -7.62825966e-01 6.38759579e-04
3.22540402e-01 3.71727616e-01 -6.78423345e-01 1.82705390e+00
6.54362082e-01 1.94330350e-01 -1.34926170e-01 1.20421600e+00
6.64087772e-01 7.09553897e-01 -3.70299220e-02 5.90209551e-02
1.22817719e+00 -6.05613351e-01 -3.25960219e-01 3.08639351e-02
4.37400192e-01 -9.16886628e-01 1.12118101e+00 -1.84194781e-02
-8.80413592e-01 -3.47699881e-01 -1.01174402e+00 1.74600128e-02
-5.32976568e-01 3.48163545e-02 2.92110682e-01 4.08377290e-01
-5.15617549e-01 3.09089571e-01 -6.25708103e-01 -3.13860811e-02
2.36533642e-01 1.28295004e-01 -6.94117188e-01 -4.77918923e-01
-1.24724960e+00 5.83492577e-01 4.07657713e-01 4.75999974e-02
-2.78809905e-01 -9.04883206e-01 -9.16243374e-01 -1.19539164e-02
3.21512371e-01 -6.23718441e-01 4.11555439e-01 -5.92702210e-01
-1.12014270e+00 9.99947011e-01 1.31015189e-03 3.39924574e-01
4.61786747e-01 -3.13638113e-02 -5.73184550e-01 2.84635961e-01
1.42845005e-01 3.98297846e-01 1.13837242e+00 -1.58783245e+00
-4.41772461e-01 -6.86787546e-01 1.70323983e-01 3.45952451e-01
-6.83380365e-01 -3.34423721e-01 -9.08550680e-01 -6.51799440e-01
6.34651482e-01 -1.03508806e+00 1.63147330e-01 5.00763893e-01
-2.02553585e-01 -4.67553765e-01 9.20303166e-01 -3.28544408e-01
1.11363447e+00 -2.64779997e+00 5.09263933e-01 4.93488431e-01
2.12755278e-01 -6.54565543e-02 -3.84548664e-01 3.57472181e-01
3.01658213e-02 -8.32249075e-02 -2.42812857e-01 -2.14437351e-01
1.60191610e-01 2.57776052e-01 -1.04614913e-01 6.82259142e-01
4.60704952e-01 5.53366721e-01 -1.00954878e+00 -5.90020895e-01
2.89508462e-01 8.63113582e-01 -3.42133582e-01 1.01075741e-02
2.21979275e-01 1.06835917e-01 -7.50556171e-01 7.61601806e-01
9.14773524e-01 -3.87581319e-01 -8.24487284e-02 -6.20265007e-01
1.57219216e-01 -5.41425385e-02 -1.46500027e+00 1.95390677e+00
-5.94553590e-01 3.52919757e-01 -8.91657099e-02 -1.28326213e+00
8.34383607e-01 1.88414693e-01 6.57504678e-01 -6.66022778e-01
-3.36822197e-02 3.93552542e-01 -5.35621881e-01 -5.10785043e-01
4.26918775e-01 -1.16599426e-01 -2.35259265e-01 2.54995704e-01
-7.70507082e-02 -2.80457400e-02 -6.23689182e-02 3.94615456e-02
6.06498361e-01 -2.20578596e-01 -1.72182247e-02 -4.90384847e-02
7.91679561e-01 -5.05615413e-01 5.23444235e-01 3.54124010e-01
-2.22214505e-01 9.32804823e-01 3.42924953e-01 -1.74856141e-01
-8.28063011e-01 -1.26612282e+00 -4.72437352e-01 7.11673498e-01
7.55471587e-01 -3.01924169e-01 -1.67411625e-01 -8.84905636e-01
5.17813981e-01 1.55359149e-01 -7.59361923e-01 -5.39982975e-01
-3.17965627e-01 -4.92180824e-01 1.91808879e-01 3.06231141e-01
1.97267786e-01 -4.58039373e-01 -2.09207147e-01 -5.57266399e-02
-2.29856819e-01 -1.00775337e+00 -7.41119146e-01 -1.85079053e-01
-7.67775297e-01 -1.03236401e+00 -1.01928592e+00 -8.71317685e-01
8.08296502e-01 7.96103179e-01 9.51823056e-01 -6.66160807e-02
-2.60694683e-01 8.75758290e-01 -3.74450982e-01 7.54564255e-02
1.37776732e-01 -7.02437013e-02 9.32812393e-02 3.83130133e-01
3.68228704e-01 -5.04956722e-01 -7.80539691e-01 4.11724865e-01
-1.33713627e+00 -3.30494493e-01 3.86213958e-01 9.31933105e-01
8.89881253e-01 -2.50004474e-02 4.04637128e-01 -2.36685827e-01
5.33555090e-01 -7.38055408e-01 -4.39088851e-01 5.56364536e-01
-3.59744817e-01 9.01423991e-02 3.16905946e-01 -7.56786525e-01
-4.29397345e-01 -3.44139248e-01 5.30401051e-01 -9.43331778e-01
1.40521407e-01 7.30058372e-01 -3.72861624e-01 -2.28519931e-01
2.23436207e-01 3.01576167e-01 1.77591324e-01 -3.10609877e-01
3.37331295e-01 4.78797406e-01 1.06957756e-01 -6.87191784e-01
9.93489802e-01 6.06559753e-01 1.20962471e-01 -7.63263285e-01
-6.55545890e-01 -7.40892172e-01 -3.04569125e-01 -1.49228767e-01
5.02545118e-01 -9.15368915e-01 -4.52597767e-01 3.15230906e-01
-9.05988693e-01 4.86780554e-01 -2.10705295e-01 6.14750981e-01
-2.79650033e-01 5.71897984e-01 -7.92380646e-02 -5.34121454e-01
-1.62552148e-01 -1.10901046e+00 1.39826560e+00 2.15714097e-01
1.12570986e-01 -1.01967669e+00 2.08203599e-01 2.88263053e-01
2.25851059e-01 2.72243202e-01 1.07488930e+00 -5.58542073e-01
-4.11751360e-01 -5.81352472e-01 -5.02863407e-01 3.37530643e-01
3.66039813e-01 -8.37322623e-02 -5.87225974e-01 -4.43814069e-01
-3.47052008e-01 -2.84073651e-01 7.97806263e-01 9.35579762e-02
1.32527316e+00 -7.80716762e-02 -4.24177229e-01 3.74706268e-01
1.46349823e+00 -1.68971285e-01 3.62266511e-01 3.37354481e-01
7.14601576e-01 6.13611221e-01 6.13583922e-01 5.60606480e-01
4.40294981e-01 8.39625061e-01 4.27792072e-01 1.34687722e-01
2.25968838e-01 -9.06116143e-02 2.45649517e-02 7.95213997e-01
2.07658887e-01 -1.03268154e-01 -7.02516079e-01 7.72174776e-01
-1.86834800e+00 -8.31627667e-01 3.34616691e-01 2.44773960e+00
7.28716075e-01 -1.49371818e-01 -2.57647736e-03 2.22856775e-01
8.50184798e-01 4.01807845e-01 -4.87762511e-01 2.37632155e-01
-1.79533243e-01 -2.01058656e-01 9.15391184e-03 2.39445642e-01
-1.18031585e+00 3.64875644e-01 4.47993994e+00 1.09308171e+00
-1.31298459e+00 -1.74868211e-01 2.17190772e-01 -2.61083633e-01
-6.81055665e-01 -2.34587900e-02 -4.30986792e-01 6.15650833e-01
1.96682170e-01 -3.01638573e-01 3.19646627e-01 6.88292384e-01
-1.82362989e-01 3.10157806e-01 -1.05855119e+00 1.25709641e+00
3.87340426e-01 -9.94816244e-01 3.11936200e-01 1.30477250e-01
6.45149291e-01 -2.12939963e-01 3.18520933e-01 2.78777242e-01
-5.36327124e-01 -6.38072550e-01 5.33767521e-01 5.47585726e-01
6.22203827e-01 -8.66046190e-01 4.37976062e-01 -1.17966952e-02
-1.26439703e+00 1.24840848e-01 -4.04155850e-01 5.83525240e-01
-2.40349714e-02 7.59898961e-01 -2.59336144e-01 8.04955602e-01
6.80538535e-01 8.46114516e-01 -5.54077506e-01 1.28932536e+00
2.54651040e-01 -9.54144374e-02 -4.71452951e-01 2.43403688e-01
2.50481188e-01 -4.56093580e-01 8.51651192e-01 8.79456162e-01
4.57441449e-01 -1.80913210e-02 3.96676451e-01 6.63146496e-01
-1.12480134e-01 2.98029095e-01 -5.80668330e-01 -1.26570873e-02
6.54263854e-01 1.13662159e+00 -2.27154464e-01 -5.93229495e-02
-6.54143810e-01 1.03509057e+00 2.73044348e-01 4.75591391e-01
-7.20306396e-01 -5.95027089e-01 8.90696824e-01 -1.08955674e-01
3.19618702e-01 -2.88427860e-01 -6.50649518e-02 -1.30272770e+00
5.29591560e-01 -5.91722667e-01 5.83719909e-01 -4.37737525e-01
-1.85965288e+00 3.10888708e-01 5.15878433e-04 -1.88776851e+00
1.82441667e-01 -4.38800007e-01 -2.86628902e-01 8.81581664e-01
-1.75362587e+00 -1.16365302e+00 -3.89837712e-01 8.55759323e-01
1.90162808e-01 -4.36504520e-02 6.93773031e-01 6.47398055e-01
-4.01499480e-01 9.42032337e-01 3.64722520e-01 2.07559601e-03
9.81748283e-01 -9.41217959e-01 -4.64059502e-01 2.43384242e-01
1.20607585e-01 7.85670817e-01 3.52761656e-01 -9.82883647e-02
-1.62245059e+00 -9.62116420e-01 7.58394361e-01 -1.62332028e-01
6.85666859e-01 -1.23020217e-01 -1.29596400e+00 1.20612904e-01
-2.57682055e-01 5.34116924e-01 1.03619993e+00 1.13133356e-01
-9.12064493e-01 -4.21123803e-01 -1.19587374e+00 6.32117093e-01
7.83661246e-01 -8.37373435e-01 -3.73413563e-01 2.79685318e-01
5.40327013e-01 -2.78302073e-01 -1.17723989e+00 6.48590922e-01
7.72420704e-01 -6.26930833e-01 1.31561899e+00 -7.22180903e-01
4.59517598e-01 -5.78724980e-01 -5.49676836e-01 -1.22615445e+00
-2.69033164e-01 5.14246449e-02 -2.75389105e-01 1.23318779e+00
2.03342155e-01 -5.80150902e-01 4.32926685e-01 6.45939767e-01
1.74184013e-02 -9.21152055e-01 -1.24667323e+00 -8.64437282e-01
2.64232367e-01 -2.56616384e-01 5.61818123e-01 1.26732206e+00
-1.07805186e-03 9.71302986e-02 -1.56199858e-01 5.34256339e-01
7.53172040e-01 5.76423287e-01 5.76738894e-01 -1.26727390e+00
-3.48757990e-02 -5.80202460e-01 -8.56164396e-01 -8.74042928e-01
3.37137252e-01 -9.79875207e-01 -2.49582425e-01 -1.21833086e+00
3.72020990e-01 -7.81230152e-01 -7.98060715e-01 3.98900628e-01
-3.51745307e-01 4.32013720e-01 3.49206150e-01 3.75653476e-01
-6.28805935e-01 1.03077030e+00 1.25450480e+00 -5.22909105e-01
-1.89752966e-01 -2.94786245e-01 -4.84971911e-01 3.95930380e-01
4.75876808e-01 -3.91008019e-01 -4.25920486e-01 -5.93045473e-01
2.72113889e-01 -5.01049869e-02 5.66903353e-01 -7.89939344e-01
1.77312717e-01 -3.21817487e-01 3.43089312e-01 -5.52655339e-01
5.17679632e-01 -1.28243220e+00 6.83955476e-03 1.34653994e-03
-2.61444628e-01 -2.01826438e-01 2.02038720e-01 9.16040659e-01
-5.55901647e-01 -5.97006045e-02 6.50442481e-01 2.41600007e-01
-4.85568523e-01 5.57920158e-01 2.08410800e-01 1.38628677e-01
1.04406118e+00 -3.53139490e-01 -5.74410893e-02 -3.51702385e-02
-5.39036870e-01 4.38345611e-01 6.48972869e-01 8.61945152e-01
8.04253459e-01 -1.97085488e+00 -6.58636451e-01 2.61672318e-01
8.66276622e-01 -4.37864363e-02 5.37477374e-01 7.92073846e-01
1.43883854e-01 2.57753253e-01 -3.76572609e-02 -9.39234078e-01
-1.13049555e+00 5.21637499e-01 1.63241237e-01 -1.09374642e-01
-3.81721258e-01 6.44108534e-01 1.71428010e-01 -5.78418493e-01
4.28734347e-02 1.19037703e-01 -1.45580932e-01 3.88372511e-01
3.85054827e-01 2.75250435e-01 8.70657787e-02 -7.14061797e-01
-6.69589043e-01 1.03318560e+00 -2.80475348e-01 1.01997159e-01
1.19913578e+00 -2.03229055e-01 1.97830126e-02 6.73770845e-01
1.87101471e+00 -1.52879450e-02 -1.09793103e+00 -6.34574234e-01
-3.43227386e-01 -9.70940471e-01 1.08140774e-01 -3.31040919e-01
-1.18197167e+00 6.85268462e-01 8.12148690e-01 2.72304982e-01
1.10669851e+00 1.83854640e-01 7.70057321e-01 4.71221283e-02
2.91201383e-01 -1.04917359e+00 2.56220162e-01 3.19882572e-01
9.02782142e-01 -1.56398749e+00 2.85551269e-02 -3.17322373e-01
-4.12375301e-01 1.03329384e+00 3.51749301e-01 -1.70267925e-01
8.60268176e-01 -5.58384359e-01 -8.84817839e-02 -3.13384533e-01
-3.50266606e-01 1.91539630e-01 8.66887569e-01 3.53581280e-01
3.04737926e-01 2.25252681e-03 -4.55803186e-01 2.50681967e-01
4.93391097e-01 -3.63587946e-01 -2.15552628e-01 9.85045433e-01
-4.41765934e-02 -1.06365538e+00 -3.81598622e-01 2.56421894e-01
-1.06068723e-01 1.49595708e-01 -1.83839202e-01 7.50731885e-01
-9.37893912e-02 6.61195636e-01 3.05006057e-02 -3.18073064e-01
5.05014062e-01 -1.33397859e-02 5.28401554e-01 -1.50104642e-01
3.55613744e-03 1.02961669e-02 -4.92330104e-01 -3.76378834e-01
-5.46960592e-01 -6.43105090e-01 -1.00426948e+00 -1.62578210e-01
-4.49530602e-01 1.20873600e-01 5.94172835e-01 7.31090546e-01
7.25496411e-01 7.01706707e-02 1.15263748e+00 -8.85035396e-01
-6.46953523e-01 -4.93024707e-01 -6.68820381e-01 7.88636923e-01
4.55781251e-01 -1.10116768e+00 -6.81266069e-01 -3.44942510e-01]
|
[11.232954025268555, 1.0055433511734009]
|
c2209e5a-d6a3-4b4f-aaeb-cfb652f506b1
|
an-efficient-membership-inference-attack-for
|
2305.18355
| null |
https://arxiv.org/abs/2305.18355v1
|
https://arxiv.org/pdf/2305.18355v1.pdf
|
An Efficient Membership Inference Attack for the Diffusion Model by Proximal Initialization
|
Recently, diffusion models have achieved remarkable success in generating tasks, including image and audio generation. However, like other generative models, diffusion models are prone to privacy issues. In this paper, we propose an efficient query-based membership inference attack (MIA), namely Proximal Initialization Attack (PIA), which utilizes groundtruth trajectory obtained by $\epsilon$ initialized in $t=0$ and predicted point to infer memberships. Experimental results indicate that the proposed method can achieve competitive performance with only two queries on both discrete-time and continuous-time diffusion models. Moreover, previous works on the privacy of diffusion models have focused on vision tasks without considering audio tasks. Therefore, we also explore the robustness of diffusion models to MIA in the text-to-speech (TTS) task, which is an audio generation task. To the best of our knowledge, this work is the first to study the robustness of diffusion models to MIA in the TTS task. Experimental results indicate that models with mel-spectrogram (image-like) output are vulnerable to MIA, while models with audio output are relatively robust to MIA. {Code is available at \url{https://github.com/kong13661/PIA}}.
|
['Kaidi Xu', 'Xiaoshuang Shi', 'Xiaofeng Zhu', 'HengTao Shen', 'RuiPeng Ma', 'Jinhao Duan', 'Fei Kong']
|
2023-05-26
| null | null | null | null |
['inference-attack', 'audio-generation', 'membership-inference-attack']
|
['adversarial', 'audio', 'computer-vision']
|
[ 4.46626283e-02 -2.34434884e-02 3.50417078e-01 -1.29494563e-01
-1.37535548e+00 -7.32330859e-01 7.41550267e-01 4.65553673e-03
-3.95592034e-01 5.53064764e-01 -1.12652242e-01 -5.37238002e-01
1.20025016e-02 -8.86253536e-01 -7.63660073e-01 -7.40398824e-01
-2.18237221e-01 1.57888770e-01 1.64171472e-01 9.80523378e-02
7.47152567e-02 1.63103864e-01 -1.04261434e+00 2.15011731e-01
8.04210722e-01 1.04234195e+00 -1.61000729e-01 8.73830497e-01
5.28093696e-01 6.57599568e-01 -1.11153615e+00 -9.86067712e-01
4.33551759e-01 -4.39538240e-01 -5.57973325e-01 -1.14422083e-01
2.16660827e-01 -4.83097076e-01 -6.28216863e-01 1.33843327e+00
9.69923735e-01 1.43215612e-01 6.06096506e-01 -1.61499286e+00
-7.72541583e-01 6.39871836e-01 -4.77606744e-01 9.49908271e-02
4.76624995e-01 3.23622763e-01 8.11098278e-01 -1.03879619e+00
5.26607871e-01 1.15485764e+00 5.47129691e-01 5.15464365e-01
-1.05307508e+00 -1.20365596e+00 -2.56772310e-01 2.37646103e-01
-2.05363321e+00 -5.37754714e-01 6.93526685e-01 -2.19253778e-01
6.62707925e-01 5.58073640e-01 2.51896501e-01 1.47412074e+00
2.83093043e-02 9.01843011e-01 1.29432070e+00 -4.90855798e-02
3.84701073e-01 2.91685015e-01 -2.50946641e-01 6.21017516e-01
-5.71401529e-02 3.01699996e-01 -7.62157321e-01 -7.27631211e-01
4.80596781e-01 -4.44528282e-01 -4.50959682e-01 2.44867751e-05
-9.71275508e-01 9.23959613e-01 -5.53571843e-02 -1.43734783e-01
-1.78721339e-01 3.56147498e-01 1.02141865e-01 5.03498316e-01
5.47200799e-01 -4.53431718e-03 7.84441605e-02 -2.40939096e-01
-1.02163923e+00 3.88280243e-01 9.47730422e-01 1.18970096e+00
4.42532539e-01 2.05148607e-01 -3.84863466e-01 5.44833243e-01
5.16455770e-01 7.58923411e-01 2.86970049e-01 -9.33478773e-01
5.73475480e-01 -3.22903514e-01 4.02270332e-02 -1.13321590e+00
1.05819307e-01 -4.80955541e-01 -1.11467469e+00 9.41986889e-02
5.05080223e-01 -5.00798821e-01 -7.20838308e-01 1.82522249e+00
3.89456570e-01 4.24299747e-01 3.12222481e-01 8.02671254e-01
6.37943208e-01 7.97085285e-01 -2.34577954e-01 -3.69357288e-01
1.14811850e+00 -5.41816473e-01 -6.47232771e-01 4.17062402e-01
3.40378463e-01 -1.03099883e+00 8.78743052e-01 6.23811781e-01
-1.28114367e+00 -4.93347049e-01 -7.97798812e-01 1.71444982e-01
-2.05022469e-01 7.17588663e-02 3.18864286e-01 1.11953962e+00
-1.31343567e+00 2.81406552e-01 -7.42981672e-01 -1.24492720e-01
5.31978190e-01 3.58567089e-01 -1.18042581e-01 -3.97032872e-02
-1.71642530e+00 3.64538729e-01 -1.72724035e-02 -3.33655663e-02
-1.40173221e+00 -4.21385944e-01 -7.48542249e-01 -1.88908592e-01
5.83001494e-01 -7.66705513e-01 1.25886571e+00 -3.75047952e-01
-1.42630994e+00 4.54267353e-01 -2.67606109e-01 -1.00615287e+00
1.01766062e+00 -7.75827840e-02 -4.01841819e-01 2.61343539e-01
-3.92934419e-02 6.37837768e-01 1.41166234e+00 -1.21390903e+00
-3.79117459e-01 -1.87255785e-01 -1.03985928e-01 -9.85459685e-02
-4.36138451e-01 7.98489433e-03 -6.06626630e-01 -1.00229347e+00
-2.31063008e-01 -1.11716294e+00 -2.28614673e-01 -1.08279243e-01
-8.57182562e-01 -3.06413442e-01 8.73941481e-01 -5.51403761e-01
1.55094671e+00 -2.26952434e+00 -3.53000402e-01 5.53780615e-01
9.78958830e-02 2.86882997e-01 7.70780966e-02 7.69320428e-01
3.43836904e-01 3.40261936e-01 -3.17368835e-01 -8.60334754e-01
2.34123930e-01 -7.36796334e-02 -8.45588684e-01 5.86055100e-01
-1.31021887e-01 8.68126512e-01 -6.69963360e-01 -5.14355063e-01
-5.63679114e-02 5.92236698e-01 -4.56664920e-01 1.25088602e-01
-1.05944961e-01 4.49229985e-01 -2.78984189e-01 7.02877700e-01
7.64161587e-01 3.73500516e-03 -2.63932049e-01 7.48712793e-02
1.79750875e-01 -2.08601668e-01 -1.13405824e+00 1.58781767e+00
-6.53319284e-02 5.99898875e-01 1.17576040e-01 -5.92457950e-01
9.19713974e-01 7.33453393e-01 2.45241970e-01 -2.87564456e-01
1.30044520e-01 1.67798817e-01 -8.19810405e-02 -2.90917218e-01
6.42898619e-01 7.78594241e-02 -1.93130925e-01 6.67231083e-01
-2.17531309e-01 -4.73483026e-01 -5.07969223e-02 5.73802054e-01
1.00791657e+00 -3.48260969e-01 -1.41105741e-01 2.26613849e-01
2.69402623e-01 -3.77557904e-01 3.34921598e-01 1.24230146e+00
-2.49796838e-01 9.05493259e-01 2.57093668e-01 1.82220951e-01
-7.53222108e-01 -1.24323785e+00 -3.81044969e-02 5.06718457e-01
1.70604989e-01 -8.12667668e-01 -1.05416465e+00 -5.80056667e-01
-1.25644758e-01 8.80769789e-01 -4.01737034e-01 -9.97864604e-02
-1.90460131e-01 -7.79644728e-01 1.21312034e+00 1.94071934e-01
7.44958520e-01 -8.17850113e-01 -2.88807541e-01 9.93863046e-02
-3.90854299e-01 -1.05349088e+00 -8.52788806e-01 -4.25702363e-01
-5.54598272e-01 -7.25498855e-01 -1.04029250e+00 -2.71852076e-01
4.87348706e-01 1.14439197e-01 5.51056862e-01 -3.51984650e-01
-2.36125559e-01 7.58784056e-01 -1.79560721e-01 -7.22477555e-01
-5.77467561e-01 -8.61703306e-02 1.34140983e-01 4.18319881e-01
1.94155782e-01 -6.57321751e-01 -7.57094800e-01 5.16548216e-01
-1.13481355e+00 -2.67161399e-01 3.03076208e-01 5.95047414e-01
6.52901292e-01 5.27850986e-01 5.67582369e-01 -7.87868917e-01
1.14800906e+00 -6.02160692e-01 -5.49401760e-01 9.33137536e-02
-7.65798867e-01 -2.88546920e-01 4.70017344e-01 -6.02865219e-01
-8.64390075e-01 -6.12728223e-02 -4.19562459e-01 -8.78374100e-01
-1.36305645e-01 4.82126862e-01 -1.06595330e-01 5.30865416e-02
7.18124688e-01 5.91671765e-01 6.39284998e-02 -2.04202697e-01
4.25288647e-01 8.17238748e-01 5.85619688e-01 -6.39562368e-01
9.82961357e-01 7.99835324e-01 -1.39005810e-01 -9.72049356e-01
-3.14776123e-01 -1.93689883e-01 1.71081070e-02 -2.03729942e-01
5.83444357e-01 -9.40550983e-01 -7.54006207e-01 8.96643996e-01
-1.05729055e+00 -2.52802640e-01 -2.42048442e-01 5.81776023e-01
-7.06403375e-01 7.98085749e-01 -7.37322688e-01 -1.21773767e+00
-5.74252963e-01 -1.05804932e+00 8.98283601e-01 -9.37359259e-02
-3.60325068e-01 -7.46692061e-01 -2.78427601e-01 5.69369912e-01
3.80962580e-01 1.99408516e-01 3.90719980e-01 -7.23197281e-01
-9.82633054e-01 -3.77520859e-01 2.22811908e-01 4.06155050e-01
-1.34697452e-01 6.23115338e-03 -1.15674782e+00 -6.12796664e-01
4.24515307e-01 -3.39457273e-01 8.83966506e-01 4.16650295e-01
1.29801059e+00 -6.88553810e-01 -1.51532337e-01 6.67032421e-01
8.98254514e-01 3.92896384e-01 8.45825195e-01 2.27253139e-02
3.68826210e-01 4.20120776e-01 8.04117858e-01 9.08157051e-01
3.68657380e-01 4.72651690e-01 4.05481905e-01 4.58361814e-03
9.16582271e-02 -4.07970577e-01 5.82779109e-01 6.22108996e-01
7.12456275e-03 -6.24733746e-01 -6.32875979e-01 5.27684212e-01
-1.78939152e+00 -1.12618613e+00 -1.70932204e-01 2.36264181e+00
8.86505365e-01 1.67099655e-01 1.85304835e-01 2.04550415e-01
5.87324500e-01 2.70280391e-01 -4.84372258e-01 -1.86914608e-01
-2.60289937e-01 2.47180983e-01 4.64409709e-01 4.52845007e-01
-1.11616492e+00 8.59596908e-01 5.72104740e+00 1.45677483e+00
-8.96568835e-01 3.63702387e-01 8.67447197e-01 -2.03181505e-01
-4.71921206e-01 3.34237106e-02 -7.16532946e-01 6.33046448e-01
9.10743475e-01 -4.01699543e-01 3.52059603e-01 6.35311842e-01
2.34835044e-01 3.04756127e-03 -8.99225175e-01 1.08319545e+00
9.72424969e-02 -1.02795780e+00 2.10037380e-02 3.11590523e-01
5.63489199e-01 -2.06315652e-01 9.17611539e-01 2.58718822e-02
4.90325630e-01 -1.02345884e+00 8.27503920e-01 3.76863331e-01
8.98169875e-01 -8.63630772e-01 4.18107182e-01 5.68231404e-01
-9.41425502e-01 1.06696852e-01 -3.01305473e-01 3.93484473e-01
3.20694834e-01 6.34560406e-01 -1.07704771e+00 5.78598797e-01
7.73810923e-01 1.88258082e-01 -3.10567588e-01 9.20291781e-01
-2.55859524e-01 1.00499177e+00 -4.72073853e-01 6.66405400e-03
1.35831758e-01 -1.28837377e-01 9.53860164e-01 1.11126268e+00
9.02013063e-01 1.25065997e-01 -8.91133249e-02 8.99638355e-01
-4.71983887e-02 2.65930612e-02 -8.55723619e-01 -3.35329324e-02
6.18218780e-01 7.82023668e-01 -4.18916613e-01 -1.11537732e-01
-2.66379770e-02 1.26944947e+00 -1.65071040e-01 6.70111537e-01
-1.19101715e+00 -4.32928473e-01 5.27637899e-01 1.37253448e-01
3.74012947e-01 -4.76977229e-01 8.74246731e-02 -9.23640788e-01
-8.21333006e-03 -9.90206063e-01 6.77998900e-01 -7.64670908e-01
-1.31235826e+00 6.25473917e-01 1.32167384e-01 -1.25991845e+00
-4.23842043e-01 1.38734698e-01 -4.79203552e-01 8.50424945e-01
-1.09392333e+00 -1.11395848e+00 1.85863480e-01 1.18035579e+00
3.70494336e-01 -2.36250594e-01 8.16249073e-01 1.76292479e-01
-3.03956479e-01 1.18517208e+00 1.44498020e-01 2.53547519e-01
7.43740141e-01 -9.75527823e-01 5.47164977e-01 9.60251868e-01
6.45403862e-01 7.99720109e-01 7.14830399e-01 -6.73404098e-01
-1.32043898e+00 -1.16349328e+00 8.42623293e-01 -4.68013734e-01
4.80632514e-01 -4.74059403e-01 -7.25916266e-01 4.66202378e-01
2.97462553e-01 -2.32676834e-01 8.20763826e-01 -4.33448255e-01
-3.21101546e-01 -1.98514257e-02 -1.24030864e+00 5.96502244e-01
7.51294255e-01 -1.00947404e+00 -1.04901753e-02 3.97974998e-01
7.70650566e-01 -6.15651846e-01 -9.25235689e-01 2.76765943e-01
2.99977750e-01 -1.02446449e+00 1.10185099e+00 3.21093500e-02
8.85764603e-03 -3.24395537e-01 -1.71204731e-01 -9.57294345e-01
1.39631256e-01 -1.49573302e+00 -4.60468620e-01 1.47945464e+00
4.74698693e-01 -7.64731526e-01 7.66952097e-01 6.46505058e-01
2.86643744e-01 -5.89309692e-01 -1.22916079e+00 -9.83982861e-01
-2.55375057e-02 -8.72725546e-01 5.95540464e-01 7.28136122e-01
-3.73205304e-01 4.48338687e-02 -8.61328483e-01 6.01265907e-01
8.55504930e-01 -1.86833199e-02 9.24700260e-01 -5.23534536e-01
-5.93953550e-01 -5.12643792e-02 -1.80743262e-01 -1.15194690e+00
-1.25842050e-01 -8.06756616e-01 -2.20463857e-01 -1.06188643e+00
-2.68686235e-01 -4.79557395e-01 -2.61544108e-01 1.11130893e-01
7.47414911e-03 3.42709064e-01 3.29904735e-01 3.45370710e-01
-4.69853997e-01 6.70479774e-01 1.03701353e+00 -2.54790336e-01
-6.99190423e-02 6.92378163e-01 -5.83554089e-01 4.84531283e-01
1.05639493e+00 -6.72984004e-01 -8.22621644e-01 -2.40284860e-01
5.80682121e-02 1.99421063e-01 4.33962613e-01 -9.78866637e-01
6.42834783e-01 1.18000701e-01 -1.18497960e-01 -7.81698763e-01
7.44584084e-01 -7.02264845e-01 5.46324134e-01 4.49656337e-01
-3.76349151e-01 2.91584488e-02 8.00830498e-02 1.04099619e+00
-3.28856707e-01 -9.19738412e-02 3.80971223e-01 -6.81836084e-02
-2.76308954e-01 6.67441428e-01 -5.72486043e-01 5.45399562e-02
1.15363479e+00 -2.09070116e-01 -2.79771686e-01 -1.20577359e+00
-7.20730662e-01 8.19089040e-02 2.38694459e-01 8.13059211e-02
1.01992404e+00 -1.16976476e+00 -9.05772150e-01 2.13440597e-01
-1.35184944e-01 1.36795640e-01 3.28790545e-01 8.79090369e-01
-1.12995319e-01 1.50137693e-01 5.45816362e-01 -5.30129433e-01
-1.57939386e+00 7.51310647e-01 -2.84515359e-02 -5.52338846e-02
-2.85060704e-01 1.25525212e+00 2.48471662e-01 -4.03574780e-02
6.22194588e-01 -7.29728341e-02 3.52942586e-01 1.56376278e-03
3.24624002e-01 3.75953764e-01 -1.34437427e-01 -5.54417312e-01
-1.48460224e-01 1.72266290e-01 -7.29768425e-02 -7.58096397e-01
7.29718328e-01 -1.45468473e-01 1.97592318e-01 1.00624055e-01
1.20374835e+00 3.86753082e-01 -1.14641404e+00 -2.65973896e-01
-3.65351766e-01 -7.31418788e-01 -2.91029990e-01 -5.56459308e-01
-1.09157991e+00 1.15451622e+00 5.07946372e-01 4.05215204e-01
1.17132103e+00 -1.58410534e-01 1.22752905e+00 1.32434532e-01
5.10259151e-01 -1.05184126e+00 2.33196229e-01 2.82097518e-01
9.46306527e-01 -8.40408623e-01 -2.37014517e-01 -3.65578294e-01
-9.32888091e-01 6.01074219e-01 3.00212771e-01 3.37323308e-01
9.08552051e-01 3.02277356e-01 1.02127001e-01 5.44080697e-02
-7.81807423e-01 4.07061130e-02 8.17480460e-02 7.99092054e-01
-1.82695940e-01 -6.51035830e-03 -6.80065826e-02 5.77472389e-01
-4.76987183e-01 6.11332292e-03 4.48080003e-01 9.56385136e-01
4.79228646e-02 -1.19500661e+00 -5.25125086e-01 1.64575532e-01
-7.52971292e-01 -3.62115622e-01 -5.78053176e-01 3.41820002e-01
-1.05005734e-01 1.36310136e+00 -1.38224185e-01 -6.82262182e-01
1.04423754e-01 -9.26526189e-02 1.00039497e-01 -2.80998737e-01
-6.85768008e-01 2.47514799e-01 6.84346184e-02 -3.93606275e-01
-1.96259841e-01 -7.45320559e-01 -1.04213917e+00 -6.41750872e-01
-4.07482982e-01 3.38706642e-01 5.37454605e-01 2.11914480e-01
5.80264449e-01 -3.45110781e-02 8.70263338e-01 -2.02986881e-01
-1.02438200e+00 -7.79620409e-01 -7.04465210e-01 3.19664836e-01
1.16734721e-01 -1.35700881e-01 -4.54438955e-01 -7.45469239e-03]
|
[5.767412185668945, 7.744274616241455]
|
05823270-d435-40d9-8dd9-beb3e547e8a9
|
sentitel-tabsa-for-twitter-reviews-on-uganda
| null | null |
https://aclanthology.org/2020.winlp-1.14
|
https://aclanthology.org/2020.winlp-1.14.pdf
|
SentiTel: TABSA for Twitter reviews on Uganda Telecoms
|
In this paper, we present a fine-grained opinion mining dataset called SentiTel. SentiTel is human annotated for targeted aspect-based sentiment analysis (TABSA). SentiTel contains Twitter reviews about three major Ugandan telecoms posted in the period between February 2019 and September 2019. The dataset contains reviews that are in English or have a codemix of English and Luganda. SentiTel contains 6320 reviews that are annotated with the target telecom, aspect and sentiment towards the aspect of the target telecom. The reviews contain at least one target telecom. We present two models on the TABSA task; random forest (RF) which is the baseline model and the BERT based model. The best result is presented by the BERT model with an AUC of 0.950 and 0.965 on the aspect category detection task and sentiment classification task respectively. The results show that a great performance can be obtained on a downstream task by fine-tuning the pre-trained BERT model. Finally, the results also confirm that fine-grained information can be extracted from the short and unstructured text from Twitter with limited cues.
|
['Joyce Nakatumba Nabende', 'David Kabiito']
|
2020-07-01
| null | null | null |
ws-2020-7
|
['aspect-category-detection']
|
['natural-language-processing']
|
[-6.45390823e-02 3.83715093e-01 -2.65523225e-01 -5.41893780e-01
-8.03656280e-01 -6.91763401e-01 9.62547958e-01 3.64355475e-01
-4.65328068e-01 6.52409673e-01 3.63949716e-01 -4.08350289e-01
2.17215374e-01 -6.95226610e-01 -1.98989376e-01 -5.76957285e-01
2.53073722e-01 5.53034604e-01 -2.32566092e-02 -6.96057320e-01
4.06878620e-01 -4.70705144e-02 -1.33951902e+00 6.54506266e-01
6.96160018e-01 1.36613309e+00 -6.68533370e-02 8.12013984e-01
-9.45338979e-02 7.48900831e-01 -8.63682926e-01 -9.85295534e-01
2.73110300e-01 -1.22745000e-01 -6.98936045e-01 1.62231445e-01
-2.11795971e-01 3.04052711e-01 6.29790783e-01 9.32288826e-01
3.71026784e-01 -2.03986302e-01 7.49379337e-01 -1.14407790e+00
-9.70706642e-02 7.25603878e-01 -8.06943774e-01 5.11678644e-02
4.81663942e-01 -2.95884103e-01 1.33855009e+00 -8.23165417e-01
8.06313813e-01 1.00685191e+00 7.63759851e-01 3.99171710e-02
-4.06443745e-01 -4.74037141e-01 3.16787213e-01 -1.87570378e-01
-9.60413694e-01 -2.58492529e-01 3.97264630e-01 -5.35845995e-01
1.26747537e+00 2.48446301e-01 6.88442945e-01 7.62429953e-01
6.82434320e-01 7.57692754e-01 1.47128427e+00 -4.39115167e-01
3.08775991e-01 8.32307100e-01 1.81458846e-01 3.32886368e-01
2.78962493e-01 -5.13230801e-01 -6.27405465e-01 -2.90022403e-01
-2.38570079e-01 -1.48752004e-01 2.48594344e-01 4.30645160e-02
-1.07982635e+00 1.09442675e+00 2.13220268e-01 3.78043503e-01
-7.60992825e-01 -5.92402458e-01 9.02565122e-01 4.89411980e-01
1.27870536e+00 4.98856187e-01 -1.21765208e+00 -3.25823575e-01
-7.97990143e-01 -1.28779516e-01 1.18868911e+00 9.43405986e-01
6.76076770e-01 -1.88849792e-01 2.70846933e-01 7.98602462e-01
4.04494733e-01 7.92322934e-01 7.10057437e-01 -1.75298586e-01
6.64669573e-01 8.82605612e-01 1.29690543e-01 -9.33631718e-01
-6.54159427e-01 -3.96915257e-01 -5.55598021e-01 -2.16869533e-01
-9.07128453e-02 -8.24570179e-01 -1.00710845e+00 9.72747147e-01
4.04560566e-01 -5.98584712e-01 1.93701312e-01 5.79200983e-01
1.12028074e+00 6.97359443e-01 1.06059715e-01 -2.78487295e-01
1.76501572e+00 -1.13236177e+00 -8.25892389e-01 -5.69752872e-01
7.42840052e-01 -1.01628411e+00 7.37391472e-01 6.15314126e-01
-6.34459794e-01 -1.98915154e-01 -9.49584424e-01 3.83000225e-01
-1.00062335e+00 9.80999097e-02 1.00409985e+00 1.01637161e+00
-8.06207418e-01 6.04619309e-02 -5.15395880e-01 -6.41481161e-01
3.08340758e-01 4.01675731e-01 -5.57025969e-01 1.46559745e-01
-1.12452519e+00 6.92671359e-01 -1.71029605e-02 2.25144960e-02
-2.78436512e-01 -1.64849952e-01 -9.30833757e-01 -3.58006060e-01
5.04324697e-02 -4.04158324e-01 1.32444334e+00 -1.44359481e+00
-1.27014577e+00 1.17207050e+00 -3.73066336e-01 -3.61870706e-01
2.79796988e-01 -1.33512020e-01 -7.92110026e-01 1.03663474e-01
7.32140064e-01 3.28398794e-01 8.50147486e-01 -9.53039765e-01
-1.26061332e+00 -5.13496935e-01 2.59576499e-01 1.77343860e-01
-1.53527439e-01 5.17170250e-01 -3.51246536e-01 -5.23702145e-01
-2.25737497e-01 -1.15419579e+00 -3.43765736e-01 -1.07371318e+00
-6.63086772e-01 -1.72160998e-01 5.55272996e-01 -4.45724219e-01
1.21647477e+00 -1.85047662e+00 -4.49185699e-01 1.92042664e-01
4.27956507e-02 2.14175418e-01 1.55962527e-01 5.58706582e-01
-9.36281905e-02 3.32026124e-01 1.62677432e-03 -3.50782484e-01
1.39508935e-04 -2.45657459e-01 -2.15251893e-01 2.49834031e-01
2.54728887e-02 1.05257368e+00 -9.56103563e-01 -3.11031222e-01
-1.02952413e-01 3.11597973e-01 -1.32336184e-01 -8.83454010e-02
6.57356530e-02 1.22408383e-01 -4.54379797e-01 8.76439989e-01
5.87380767e-01 -2.02831149e-01 2.25282252e-01 -1.99655116e-01
-1.95279643e-01 5.40338516e-01 -7.22316742e-01 9.50193882e-01
-7.08811522e-01 7.75159240e-01 1.81563139e-01 -3.90013069e-01
9.77959156e-01 3.26752096e-01 3.24772298e-01 -5.06563663e-01
4.83139038e-01 4.03731048e-01 -1.91797540e-01 -2.99751461e-01
9.18279469e-01 -4.00342345e-01 -6.62697911e-01 5.45002162e-01
-8.42947215e-02 -5.12650907e-01 3.97799551e-01 2.57596850e-01
9.32344139e-01 -2.72932261e-01 8.00580859e-01 -4.43647593e-01
6.09706759e-01 4.42341119e-01 3.50418031e-01 5.48810244e-01
-3.03883851e-01 5.03408849e-01 7.32578754e-01 -3.07739049e-01
-5.83040774e-01 -4.04206902e-01 -1.22969180e-01 1.49702013e+00
-2.86525786e-01 -6.08221710e-01 -6.45698249e-01 -1.29272270e+00
-3.19342315e-01 5.43406963e-01 -1.00706553e+00 3.08672607e-01
5.62330931e-02 -1.06307471e+00 6.37599677e-02 1.50104433e-01
6.40233517e-01 -1.19120848e+00 -2.25091308e-01 1.56537127e-02
-2.67453223e-01 -1.29240763e+00 -2.80021697e-01 7.24210382e-01
-5.73584199e-01 -9.30311084e-01 -5.98459482e-01 -8.92405570e-01
5.78239322e-01 2.35291690e-01 1.25827742e+00 -4.08850014e-01
4.17710185e-01 3.97971757e-02 -8.79463732e-01 -9.58569407e-01
-9.38345492e-02 5.57634711e-01 -1.06831461e-01 1.19963698e-01
1.10581756e+00 -8.46581981e-02 -4.59183484e-01 4.72193062e-01
-6.61748588e-01 -1.67217165e-01 5.32413185e-01 5.49044311e-01
3.28858256e-01 1.63774222e-01 8.41703892e-01 -1.74909377e+00
7.83536792e-01 -7.73381472e-01 -1.92068160e-01 -2.53837973e-01
-7.69098699e-01 -6.01134837e-01 4.83354390e-01 7.14707822e-02
-1.05263555e+00 7.56296962e-02 -3.21623534e-01 5.71900487e-01
-1.34123832e-01 9.32005346e-01 3.93906422e-02 4.14866447e-01
6.94799244e-01 -2.46857718e-01 -3.86441559e-01 -6.31392002e-02
1.21867932e-01 1.44611943e+00 -9.99930277e-02 1.30122021e-01
2.98497826e-01 6.23731315e-01 -4.69575852e-01 -7.74929821e-01
-1.37603879e+00 -9.34724808e-01 -5.62420189e-01 -2.06143305e-01
7.85119176e-01 -1.21480548e+00 -4.49973553e-01 7.54840910e-01
-7.90303528e-01 -1.76922809e-02 -2.24239677e-01 1.97248444e-01
-2.33446643e-01 2.31793011e-03 -6.76375091e-01 -1.03755403e+00
-8.10670674e-01 -9.93391156e-01 1.30220830e+00 1.04600623e-01
-6.54234231e-01 -1.15445590e+00 1.97089359e-01 5.69468141e-01
4.48580325e-01 2.14341074e-01 4.61337060e-01 -1.19234884e+00
3.56494159e-01 -8.46336305e-01 -1.39829159e-01 3.67132932e-01
4.35936809e-01 9.18513387e-02 -1.25470781e+00 -4.91785370e-02
2.10323021e-01 -2.21404210e-01 7.59413064e-01 4.66422141e-01
3.64482366e-02 -1.70930579e-01 -1.82083398e-01 -1.14129186e-01
1.14759696e+00 3.56693745e-01 3.83107305e-01 9.61567283e-01
5.43650925e-01 9.96676207e-01 1.17824435e+00 4.84574020e-01
7.19145417e-01 2.59181648e-01 2.72361219e-01 -6.07143268e-02
4.15362000e-01 -5.86719997e-02 6.41381085e-01 1.09512520e+00
1.24261402e-01 -2.93965071e-01 -9.86279845e-01 9.41346526e-01
-1.82137299e+00 -5.49399555e-01 -5.18616498e-01 1.69588125e+00
5.00949919e-01 6.45137548e-01 2.81791657e-01 3.80832285e-01
7.28356242e-01 3.63364577e-01 -1.37747779e-01 -1.00984979e+00
-3.25217873e-01 1.11983284e-01 5.05038500e-01 4.04409140e-01
-1.50365233e+00 1.05186725e+00 6.11064768e+00 8.09608757e-01
-9.87486720e-01 2.78204769e-01 9.80018675e-01 1.48067519e-01
-2.49683261e-01 -2.45506078e-01 -9.06602561e-01 3.97067904e-01
1.41309416e+00 -2.79402100e-02 -2.20152825e-01 1.05512381e+00
4.26127255e-01 -4.66481984e-01 -4.19886678e-01 5.35586774e-01
1.30667999e-01 -9.18910503e-01 -1.51603296e-01 1.43010527e-01
1.29918945e+00 4.80026245e-01 3.26758763e-03 5.20616770e-01
3.14547211e-01 -7.22963691e-01 6.37789369e-01 1.54973045e-01
6.09863758e-01 -9.52560365e-01 1.74679863e+00 2.62784839e-01
-1.12099898e+00 1.03149064e-01 -1.00690663e-01 -2.50591218e-01
3.22336823e-01 1.11955333e+00 -9.81223106e-01 5.12196124e-01
8.57595205e-01 1.15469456e+00 -6.16181850e-01 6.21589005e-01
-4.84223306e-01 6.89276516e-01 -9.14742276e-02 -5.67317069e-01
4.64450896e-01 -2.23130032e-01 3.82889599e-01 1.39629304e+00
5.33443466e-02 -2.87393749e-01 -1.20297089e-01 -3.35206807e-01
-3.26264054e-02 5.31973302e-01 -7.56488025e-01 -1.56629920e-01
1.46266806e-03 1.90205681e+00 -1.13551438e+00 -4.03110385e-01
-5.44250548e-01 7.79964983e-01 -1.39187738e-01 1.43841088e-01
-3.08081746e-01 -7.16099799e-01 3.28819096e-01 1.38991550e-01
6.35256410e-01 4.31645006e-01 -6.11491680e-01 -1.08802283e+00
-1.85993820e-01 -1.07129550e+00 3.07638049e-01 -8.38716209e-01
-1.33465624e+00 1.35984027e+00 -4.41900998e-01 -1.27183461e+00
-4.35168654e-01 -8.97069514e-01 -5.85208118e-01 6.41031563e-01
-1.42428541e+00 -1.27747154e+00 7.66402707e-02 2.70040333e-01
6.78205550e-01 -3.94005567e-01 7.65289187e-01 1.44728258e-01
-3.84139806e-01 2.18194067e-01 -1.43772379e-01 6.44824177e-04
6.91829443e-01 -1.42412674e+00 5.72228551e-01 4.54274476e-01
-2.01340213e-01 6.88790977e-01 8.32231939e-01 -7.89247334e-01
-8.19098771e-01 -1.13029170e+00 1.71733904e+00 -8.28944802e-01
1.04220307e+00 -3.98917228e-01 -9.84573811e-02 5.62031507e-01
4.37479794e-01 -4.67850596e-01 1.13405025e+00 3.93730193e-01
-1.81756407e-01 -4.86341715e-02 -1.37933719e+00 1.99321836e-01
2.07321361e-01 -6.38378263e-01 -5.54066360e-01 4.51566368e-01
5.45623124e-01 -1.21948101e-01 -8.23936522e-01 3.23950678e-01
6.16895378e-01 -1.01797533e+00 1.51622429e-01 -4.34488654e-01
6.54156983e-01 -2.01601341e-01 -3.35921526e-01 -1.52045608e+00
1.46887407e-01 -5.33219397e-01 2.79926926e-01 1.40795922e+00
1.16472697e+00 -7.76985884e-01 9.09404933e-01 4.63454843e-01
1.36163130e-01 -8.86573374e-01 -6.39182091e-01 -2.62640208e-01
-1.62920222e-01 -9.01082814e-01 5.56794941e-01 7.54974604e-01
3.79429013e-01 1.00616491e+00 -4.74676453e-02 -2.91876256e-01
-6.35514557e-02 1.27832249e-01 7.73455858e-01 -1.34091806e+00
2.77820397e-02 -2.65603215e-01 -3.49287480e-01 -8.24144125e-01
1.84935629e-02 -5.06839037e-01 1.03581384e-01 -1.60896587e+00
2.49507561e-01 -2.14359984e-01 -1.59954622e-01 2.58772612e-01
-2.12050110e-01 5.39504588e-01 -6.33226112e-02 2.09986135e-01
-9.87007320e-01 2.67699540e-01 9.69761550e-01 -1.61248431e-01
-1.71539024e-01 6.67341530e-01 -1.36408925e+00 1.01695347e+00
9.00574386e-01 -5.21801412e-01 -2.30596066e-01 1.36188030e-01
1.07459688e+00 -3.07598084e-01 -4.36050177e-01 -2.33343765e-01
8.31258204e-03 3.42930377e-01 6.64933026e-02 -9.32866454e-01
2.36652792e-01 -8.34293664e-01 -2.58186072e-01 2.05009192e-01
-2.30903216e-02 3.62489969e-01 1.59128815e-01 4.76825207e-01
-5.22721589e-01 -3.40428084e-01 5.09834766e-01 -1.76250756e-01
-4.28306013e-01 -8.67891535e-02 -9.00296450e-01 2.62567736e-02
8.45458746e-01 -9.79949981e-02 -3.51843834e-01 -7.82094479e-01
-6.35545850e-01 6.10288456e-02 2.21385762e-01 5.80706775e-01
1.21748829e-02 -9.03906345e-01 -5.28437614e-01 7.79613331e-02
4.54850882e-01 -3.23118180e-01 -2.04884648e-01 9.99041080e-01
-3.91483217e-01 7.46659815e-01 1.44278288e-01 -3.28066021e-01
-1.05544043e+00 2.52239406e-01 -1.22466022e-02 -6.79302394e-01
2.74398863e-01 8.14803481e-01 1.78481936e-02 -9.87208486e-01
-2.60275483e-01 -2.51329035e-01 -9.50653195e-01 8.74342620e-01
5.95598936e-01 2.76537806e-01 5.39714456e-01 -1.21025312e+00
-6.28059447e-01 5.00282645e-01 -3.33624154e-01 -2.94810832e-01
1.43188894e+00 -4.92427617e-01 -4.82382566e-01 7.91250825e-01
1.15388656e+00 5.68165481e-01 -5.41458189e-01 1.21750280e-01
1.35183260e-01 -1.04118340e-01 1.34866893e-01 -1.10974479e+00
-1.21726906e+00 3.95084828e-01 2.02137902e-01 9.33128178e-01
1.05335772e+00 3.78744118e-02 5.87327838e-01 1.93678379e-01
5.45521379e-01 -1.20999742e+00 -5.23775041e-01 9.01572764e-01
5.48282444e-01 -1.74165905e+00 5.46589158e-02 -5.75961173e-01
-1.35572290e+00 7.16490924e-01 2.12759525e-01 -7.11212680e-03
1.22211719e+00 6.06171414e-02 6.91819668e-01 -5.78829706e-01
-8.22668433e-01 -4.64439154e-01 3.02946538e-01 5.80929935e-01
6.38577282e-01 3.02514941e-01 -5.72679877e-01 9.99713719e-01
-7.75324821e-01 -3.27651858e-01 5.47754705e-01 8.62351596e-01
-4.84086245e-01 -8.59185815e-01 -1.33337572e-01 8.90662014e-01
-1.12049544e+00 -3.65009338e-01 -7.99417317e-01 7.52696633e-01
-8.50010738e-02 1.79762173e+00 -2.06840307e-01 -6.35292590e-01
4.33863580e-01 -2.36120299e-01 -5.02656162e-01 -8.34791660e-01
-1.44540989e+00 3.01175058e-01 8.39514792e-01 -2.62512177e-01
-7.30884254e-01 -8.23795319e-01 -8.64490867e-01 -2.47005001e-01
-6.02747798e-01 6.81744814e-01 9.13342834e-01 1.09779000e+00
3.47056806e-01 2.71404088e-01 1.01397812e+00 -7.67014265e-01
3.52230340e-01 -1.32400978e+00 -8.69550407e-01 1.46261573e-01
4.20938939e-01 -2.14863092e-01 -7.39101529e-01 -4.95209619e-02]
|
[11.204784393310547, 6.875819683074951]
|
2424c359-0699-4056-a442-0f894ee8a569
|
improving-the-quality-control-of-seismic-data
|
2201.06616
| null |
https://arxiv.org/abs/2201.06616v2
|
https://arxiv.org/pdf/2201.06616v2.pdf
|
Improving the quality control of seismic data through active learning
|
In image denoising problems, the increasing density of available images makes an exhaustive visual inspection impossible and therefore automated methods based on machine-learning must be deployed for this purpose. This is particulary the case in seismic signal processing. Engineers/geophysicists have to deal with millions of seismic time series. Finding the sub-surface properties useful for the oil industry may take up to a year and is very costly in terms of computing/human resources. In particular, the data must go through different steps of noise attenuation. Each denoise step is then ideally followed by a quality control (QC) stage performed by means of human expertise. To learn a quality control classifier in a supervised manner, labeled training data must be available, but collecting the labels from human experts is extremely time-consuming. We therefore propose a novel active learning methodology to sequentially select the most relevant data, which are then given back to a human expert for labeling. Beyond the application in geophysics, the technique we promote in this paper, based on estimates of the local error and its uncertainty, is generic. Its performance is supported by strong empirical evidence, as illustrated by the numerical experiments presented in this article, where it is compared to alternative active learning strategies both on synthetic and real seismic datasets.
|
['Stephan Clémençon', 'Emilie Chautru', 'Raphaël Butez', 'Mathieu Chambefort']
|
2022-01-17
| null | null | null | null |
['geophysics']
|
['miscellaneous']
|
[ 0.4167132 0.12970562 0.31097716 -0.32914177 -1.3010716 -0.36437005
0.3799286 0.68533367 -0.96763456 0.7691945 -0.22107905 -0.2624063
-0.46270123 -0.83638024 -0.4399854 -1.2657846 -0.3610929 0.5408014
0.34571868 -0.04727588 0.5007102 0.6865145 -1.3646693 -0.07114991
1.0852562 1.0404596 0.3738437 0.53279144 -0.07103734 0.7163254
-0.49926502 -0.1316758 0.21426941 -0.39233932 -0.50016433 0.4287022
-0.09905447 -0.13422813 0.49261928 1.0368856 0.5623438 0.5005661
0.78957486 -0.6397909 0.22730964 0.22923918 -0.43638772 0.23951444
0.15821418 0.22808248 0.7002086 -1.0991807 0.28138596 0.6469422
0.5657667 0.12586108 -1.3633219 -0.2421122 -0.15299219 0.45713905
-1.1162853 -0.6416983 1.0750781 -0.70894355 0.39223102 0.24194309
0.6756581 0.72675127 -0.07707267 0.3612439 1.2036184 -0.7758032
0.9274141 -0.0246265 -0.02292277 0.43862462 0.37336808 0.01337022
-0.48869708 -0.14288369 0.59784216 -0.2550862 -0.43639955 -0.3693786
-0.6724999 0.8711625 0.2641431 0.54930013 -0.8767102 -0.12353002
0.3729044 0.45188624 1.00955 0.5655049 -0.05819361 0.0073282
-1.4301722 -0.11521852 0.74029905 0.15137602 0.78777945 0.15300651
0.5449421 0.80284035 0.51155454 0.25324458 0.02284319 -0.93245476
0.4010204 0.6354913 0.2504099 -1.1957895 -0.19370599 -0.30346215
-1.0111872 0.99470407 0.6228393 -0.1210462 -0.69168615 0.8976994
0.3712271 0.07594703 -0.03387572 0.86742306 0.3315012 0.68706834
0.06952413 -0.79848796 1.0497148 -0.28297383 -0.6683552 -0.255358
0.17946364 -0.75636595 0.79554486 0.95470136 -1.0301341 -0.4228173
-0.97114676 0.4723069 -0.07715489 -0.07100058 0.03493924 0.56526524
-0.8101281 0.84593946 -1.2559423 -0.09760672 0.41426358 0.3039115
-0.29658556 -0.03140102 -1.0197327 0.8919515 0.37779534 0.61791134
-0.86354727 -0.34135702 -0.7768135 0.02020459 0.5307101 -0.16392112
1.0039936 -1.1370783 -1.3242341 0.5931826 0.17984039 -0.4641868
0.83439684 -0.10011159 -0.16847077 0.6550086 -0.12988026 -0.07427541
1.3442719 -1.6037723 -0.41756856 -0.31600165 -0.01747907 -0.00854853
-0.29890123 0.05824676 -0.19056262 -0.80491406 0.4944837 -0.55978817
-0.520917 0.16629674 0.0261453 -0.0741378 0.5269171 -0.99184686
0.9828648 -2.1953366 0.23446813 0.72971904 0.2064067 0.30471548
0.4935012 0.52816564 -0.19350879 -0.19788349 -0.9595983 -0.35733736
-0.3374219 0.10883004 -0.07074746 0.71365565 0.3472674 0.14177598
-0.9636324 -0.7810807 0.37213296 0.42587823 -0.21906158 0.3333415
0.07904129 0.652273 -0.41192088 0.47788426 0.5365644 -0.14987957
-0.02571911 -0.06890169 -0.34408635 -0.24434121 -1.4319154 1.3219713
-0.5646091 0.44286147 0.3488068 -1.6817465 1.1591452 0.73080313
0.60213906 -0.6314033 0.01111122 0.649746 -0.2135704 -0.83801454
0.05508117 -0.37383935 0.19345017 0.42499772 -0.19449052 -0.20190367
0.42357367 -0.17409337 1.0631275 0.04128943 0.4066113 -0.35946226
0.681728 0.23786744 0.6216224 0.5310175 -0.02714394 0.45730582
0.44099602 -0.2737611 -1.1929117 -0.64685506 -0.059423 0.5170301
-0.19028006 0.15355462 -0.7634326 -0.48346913 -0.46309495 0.5651164
-0.49285474 0.06617587 -0.7876975 -0.8038133 -0.04559572 0.25806683
0.30384034 -1.1943536 -1.1428772 0.597183 0.06816344 -0.5888661
0.2738995 0.5693396 -1.3039707 -1.0821064 -0.9192403 -0.43762136
0.78561 -0.17497508 0.9997908 0.25599718 -0.24563663 0.32781023
-0.73257643 -0.39160714 -0.56726557 -0.19662374 -0.27430427 0.2845804
-0.17482966 -0.6973172 -0.7047455 0.11440662 -1.1615826 -0.41039786
0.69696933 0.9249675 0.55681527 0.5705614 0.58470666 -0.920616
0.37388557 -0.36548445 -0.95156705 0.08039252 -0.39692312 -0.18990442
0.673789 -0.18046358 -1.2856803 0.3639579 -0.32187167 0.03531914
-0.23610148 0.9208198 -0.13790663 -0.09847053 0.88422006 -0.00750598
0.05352037 -0.545044 -0.2575048 0.5584395 0.40345263 -0.19853115
0.98543864 0.45037213 0.1761547 -1.2979083 -0.5683823 -0.68112636
-0.69592017 -0.74095887 0.7318708 -0.5134236 -0.3334865 0.5107741
-1.0551058 -0.07375145 -0.37491402 0.63437235 -0.5882962 0.64379436
-0.2716441 -1.2349787 -0.12547441 -1.1002188 0.64714426 -0.03100907
-0.09286657 -1.1285741 0.08839648 0.37834096 0.2636726 0.26431125
0.799609 -0.5120335 -0.5405602 -0.37164262 0.2574544 0.72664756
0.01820405 -0.11486539 -1.0058856 -0.25859636 0.6517103 -0.22944126
0.8757266 0.42010456 0.8889606 -0.09594678 0.09745876 0.05240101
1.643443 0.4552389 0.74586844 0.4667177 0.16376416 0.92615604
0.87604797 0.45417458 -0.31745893 0.46690544 0.58464205 -0.4048914
0.24158898 0.4925817 0.06993298 0.9357238 -0.5581413 -0.13393283
-1.1094714 0.5531324 -1.6525874 -1.0460733 -0.19461884 2.537644
0.7510308 0.4455278 -0.07399401 1.2008479 0.5724672 -0.07947511
-0.2303494 0.03175722 0.09460801 0.28716344 0.37221566 0.6133571
-1.2092533 0.06242688 4.589964 0.88404 -1.0609181 0.05036636
0.67759055 0.4111603 -0.06479514 0.05915602 -0.16019745 0.53455377
0.7204756 0.3626376 -0.07120055 0.5580205 0.6347441 -0.716568
-0.6837788 0.9291034 -0.12869482 -0.99269587 -0.45744148 -0.05644586
0.44972023 -0.4926989 -0.50729555 -0.2941707 -0.2469911 -0.6687256
0.62993 0.9661744 0.20068753 -0.94347423 0.876954 0.71013194
-0.8888489 -0.21655625 -0.22601633 0.12603417 0.546254 1.1478938
-0.6189067 0.7098712 0.73557234 0.37443155 -0.33101705 1.5751014
-0.33784738 0.9744834 -0.28634778 0.10660665 0.14655668 -0.44679764
0.67906433 0.9554079 0.48984495 0.24937606 0.15003386 0.48837563
0.5262153 0.37808707 -0.25805947 0.07595178 0.24771997 1.1805702
-1.4015937 -0.26762003 -0.25226417 0.7525812 -0.06832647 0.26695243
-0.2562819 -0.43335047 -0.14939773 0.41730934 0.261736 -0.3318137
-0.23487273 -0.77431595 0.17182155 -0.79734933 0.28767025 -0.38409805
-1.2860885 0.61851406 0.15243511 -1.5679473 -0.4329077 -0.4050995
-0.60784334 0.8468433 -1.3791085 -0.5561509 -0.24874665 0.29719228
0.6979734 -0.0544431 0.64837974 0.60032886 -0.3481095 -0.14048085
0.18270405 0.15419164 0.39639685 -1.2444422 0.06245852 1.0431088
0.0884022 0.14755234 1.1980915 -0.64826226 -1.0495574 -0.48184192
0.9882658 0.24635687 0.6583561 -0.04313876 -1.3400878 0.06664867
0.07764854 0.11011957 0.40923503 -0.25646544 0.33428347 -0.24910036
-1.0127888 0.08718283 0.28443328 -0.37563887 -0.74371517 0.20032285
-0.16350977 0.10962639 -0.7359996 0.4290314 0.01853901 -1.0474384
0.8578288 -0.12104386 0.20415452 -0.44882762 0.18947026 -1.2379754
0.0551797 -0.33241895 0.25721395 1.0793908 0.48628235 -0.5079792
0.87536204 0.3342309 -0.17488484 -0.6675707 -1.1744401 -0.6378968
-0.3955737 -0.63732606 -0.12016603 0.8912397 -0.2463368 0.01582972
-0.2365644 0.20142384 1.1433759 -0.14681374 0.35435203 -1.5236821
-0.45517597 -0.2311576 -0.4516595 -0.5041445 -0.19734256 -0.3577459
0.3305216 -1.5068707 -0.29815322 -0.54662263 -0.08309532 0.2297151
-0.10118014 0.45991012 -0.08257187 0.3298847 -0.14697532 0.42606637
0.89738923 -0.24654737 -0.06213856 0.44584826 -0.08234195 1.0524207
0.673382 -0.6739152 -0.37146765 -0.18715857 0.63598514 0.34646517
0.45177558 -1.1016684 0.49679238 0.03211033 0.27912885 -0.4354885
0.4869426 -1.3058635 0.22230583 0.55524755 -0.12405804 -0.19645905
-0.27705535 0.5185468 -0.6116354 -0.9929405 1.0422007 -0.36774993
-0.76800394 -0.16920218 -0.57913345 -0.2620396 0.9880306 -0.48094836
0.44493303 -0.42115155 -1.2916504 -0.1366395 0.18707989 -0.4428846
0.68107975 -0.76948315 -0.84285426 0.0096237 -0.00665485 0.09250118
0.50903344 1.0128444 -0.8462511 -0.24421309 0.14783254 -0.6731888
-1.2026043 0.29953018 0.17901926 -0.21487705 -0.6704867 0.7355645
-0.41694853 0.11724042 0.2584616 -0.07266276 -0.67637205 0.7542112
0.5171305 0.7621849 0.63355213 -0.47738877 -0.02089204 0.5862838
0.388555 -0.31299138 1.715448 -0.02123102 -0.2510762 0.58359855
0.7963595 -0.01716638 -1.3984469 -0.15914956 0.6258212 -0.49070013
0.30303124 -0.4042268 -1.0951418 0.9025434 0.81632257 0.6488291
1.5727963 -0.17558537 0.15457414 0.47555807 0.3811213 -1.3707485
0.11806428 0.02384038 0.96686774 -1.3630188 0.26678318 -0.38202086
-0.32915714 1.2299725 -0.14916381 -0.2753627 0.8265045 0.4009957
0.39798906 -0.29907912 -0.31230873 -0.01956399 0.23196475 0.17973886
0.33785176 -0.274185 -0.58491284 -0.01376841 0.30284005 -0.0827684
0.31140208 1.2745899 -0.81234586 -1.322674 -0.9255412 0.29712042
-0.46851814 0.12262118 0.01485496 0.7005705 0.13297792 1.096517
-0.04981938 0.24015184 0.40824285 0.01819113 0.32036516 -0.5432881
-0.5753465 0.48303252 0.08853412 -0.2108776 -0.9562331 -0.88381517
-1.149646 0.16821456 -0.40776503 0.5139427 0.9161177 0.98263747
-0.45823905 0.3532473 0.8571518 -1.0783508 -0.62282354 -0.9249694
-0.71305794 0.38589108 0.29837757 -0.6749607 -0.76090807 0.39862484]
|
[7.965648651123047, 2.200573444366455]
|
d7b1c497-6f52-46ae-b5a9-1ef644447860
|
enhancing-next-active-object-based-egocentric
|
2305.12953
| null |
https://arxiv.org/abs/2305.12953v2
|
https://arxiv.org/pdf/2305.12953v2.pdf
|
Enhancing Next Active Object-based Egocentric Action Anticipation with Guided Attention
|
Short-term action anticipation (STA) in first-person videos is a challenging task that involves understanding the next active object interactions and predicting future actions. Existing action anticipation methods have primarily focused on utilizing features extracted from video clips, but often overlooked the importance of objects and their interactions. To this end, we propose a novel approach that applies a guided attention mechanism between the objects, and the spatiotemporal features extracted from video clips, enhancing the motion and contextual information, and further decoding the object-centric and motion-centric information to address the problem of STA in egocentric videos. Our method, GANO (Guided Attention for Next active Objects) is a multi-modal, end-to-end, single transformer-based network. The experimental results performed on the largest egocentric dataset demonstrate that GANO outperforms the existing state-of-the-art methods for the prediction of the next active object label, its bounding box location, the corresponding future action, and the time to contact the object. The ablation study shows the positive contribution of the guided attention mechanism compared to other fusion methods. Moreover, it is possible to improve the next active object location and class label prediction results of GANO by just appending the learnable object tokens with the region of interest embeddings.
|
['Alessio Del Bue', 'Vittorio Murino', 'Pietro Morerio', 'Cigdem Beyan', 'Sanket Thakur']
|
2023-05-22
| null | null | null | null |
['short-term-object-interaction-anticipation', 'action-anticipation']
|
['computer-vision', 'computer-vision']
|
[ 2.86838174e-01 2.55783591e-02 -2.73231030e-01 -4.37492311e-01
-6.50387764e-01 -1.48264527e-01 6.37481451e-01 -2.49999419e-01
-3.52949679e-01 4.43028957e-01 8.79812837e-01 5.95899940e-01
-1.95147812e-01 -2.76731610e-01 -7.17606664e-01 -8.26427877e-01
-3.55032444e-01 4.34798658e-01 2.93810606e-01 1.56705469e-01
2.88596660e-01 2.89897323e-01 -1.72811103e+00 6.06513441e-01
5.05430460e-01 1.30293894e+00 4.81375754e-01 5.21909654e-01
2.90665716e-01 1.38545632e+00 -1.36238784e-01 -1.03369907e-01
1.90820903e-01 -3.38767946e-01 -8.42458725e-01 1.50758058e-01
5.81462979e-01 -8.26557159e-01 -5.94633937e-01 5.71303964e-01
4.11006778e-01 5.00937104e-01 4.37657535e-01 -1.56467605e+00
-3.62018287e-01 6.41686618e-01 -4.14638281e-01 3.73185754e-01
4.49150980e-01 3.74628693e-01 9.15773630e-01 -9.29699063e-01
8.37238014e-01 1.23217487e+00 3.20173800e-01 5.93918741e-01
-6.62028432e-01 -5.63212395e-01 7.73501098e-01 1.00243402e+00
-1.11099386e+00 -4.52830076e-01 1.02124119e+00 -6.56809330e-01
9.42976475e-01 -2.60959603e-02 1.02229595e+00 1.39372802e+00
1.60744876e-01 1.44459867e+00 5.38492084e-01 -1.66912943e-01
7.23920837e-02 -2.94256270e-01 -2.68162310e-01 6.81552351e-01
-5.08658350e-01 4.68614884e-03 -8.19139361e-01 1.56972602e-01
4.02968764e-01 4.41809475e-01 -1.72248691e-01 -6.02633357e-01
-1.67326725e+00 4.68191803e-01 4.69408303e-01 2.75388241e-01
-8.40266347e-01 6.40329719e-01 5.79210579e-01 -4.58912551e-01
7.03252435e-01 2.31078491e-01 -4.93103623e-01 -6.52310252e-01
-7.91639209e-01 3.58625740e-01 6.66688830e-02 9.81154442e-01
5.75688243e-01 -7.87568539e-02 -7.57177830e-01 3.14087451e-01
2.28562713e-01 2.55269110e-01 3.85981917e-01 -1.12169218e+00
5.61059594e-01 7.54389405e-01 3.39419901e-01 -9.58352506e-01
-3.93346936e-01 -1.77278072e-01 -2.56121993e-01 -1.37689590e-01
1.65985435e-01 -2.56793499e-02 -8.86455297e-01 1.93726981e+00
5.18173695e-01 7.32641399e-01 -1.84719786e-01 1.16354918e+00
6.30233407e-01 5.75023770e-01 5.13470531e-01 -7.36568719e-02
1.36531043e+00 -1.53123236e+00 -8.86100888e-01 -3.76867056e-01
8.11643243e-01 -4.94283527e-01 7.68943429e-01 1.53102189e-01
-1.15747583e+00 -8.14503014e-01 -5.61568975e-01 -1.58174008e-01
-1.86064422e-01 4.62586492e-01 8.54854107e-01 -8.29682946e-02
-6.82233810e-01 4.93360311e-01 -8.86513948e-01 -3.33026260e-01
7.26274252e-01 3.47963929e-01 -4.87432390e-01 -1.74122661e-01
-1.01551807e+00 7.57790148e-01 3.91782820e-01 3.23224425e-01
-1.37229013e+00 -7.93074727e-01 -8.54690909e-01 3.28060865e-01
6.06523037e-01 -7.67959714e-01 1.10328412e+00 -1.23852384e+00
-1.20998263e+00 4.86113608e-01 -3.37015688e-01 -6.31784916e-01
4.00804073e-01 -7.31300235e-01 -1.33501798e-01 4.45789218e-01
3.11880350e-01 1.13115561e+00 8.53882194e-01 -8.09431493e-01
-9.82474327e-01 -3.75632286e-01 2.42768690e-01 4.86428797e-01
-4.31511074e-01 -1.36205971e-01 -4.25578952e-01 -5.95251679e-01
-2.47668654e-01 -1.04034805e+00 -6.81107268e-02 -9.05051921e-03
-7.51242265e-02 -6.90472126e-01 1.14125800e+00 -6.77446663e-01
1.07957506e+00 -2.38142252e+00 4.52255994e-01 -4.72782731e-01
5.53461649e-02 1.12838104e-01 -3.33455563e-01 4.21252429e-01
-1.65839270e-01 -2.30895624e-01 2.10319906e-01 -5.22611618e-01
-2.80453805e-02 -1.00482270e-01 -4.04902458e-01 4.02342260e-01
1.88683063e-01 1.20716572e+00 -1.26715386e+00 -6.27521574e-01
5.56156874e-01 5.60622215e-01 -5.10451674e-01 4.58339870e-01
-3.16852689e-01 6.20716691e-01 -7.52908587e-01 7.08728552e-01
2.93089509e-01 -8.60637799e-02 -9.55794230e-02 -4.72140074e-01
-4.20589559e-02 9.07277614e-02 -6.94371700e-01 2.14445353e+00
-4.05807793e-01 7.32388496e-01 -3.84291947e-01 -9.24736083e-01
5.59296131e-01 4.56028640e-01 1.27906632e+00 -6.98499322e-01
1.01183854e-01 -2.47649878e-01 -2.24739760e-01 -1.06635356e+00
3.41865003e-01 2.89601535e-01 -6.36020675e-02 3.38397235e-01
1.05482556e-01 4.96293217e-01 2.99321800e-01 3.71823877e-01
1.08151400e+00 8.94333899e-01 1.45663500e-01 2.42586434e-01
4.92709398e-01 3.02365758e-02 7.15171337e-01 4.95086551e-01
-5.19970417e-01 5.78544974e-01 2.92874068e-01 -7.05444872e-01
-8.56130123e-01 -6.84623957e-01 4.57713574e-01 1.28091097e+00
3.20958287e-01 -4.56713498e-01 -7.23709166e-01 -9.67948914e-01
-1.51984289e-01 7.61127591e-01 -1.08596492e+00 -3.75586092e-01
-8.80599022e-01 -1.47354290e-01 -3.64923514e-02 1.00891006e+00
6.24402761e-01 -1.45804262e+00 -8.21997404e-01 3.14256966e-01
-7.89153576e-01 -1.38848555e+00 -7.93272495e-01 -3.87911081e-01
-6.86674356e-01 -1.26470721e+00 -6.89254642e-01 -4.91864324e-01
6.90635383e-01 3.24585497e-01 7.82616198e-01 -2.53433943e-01
-1.38005152e-01 8.26656759e-01 -5.41014254e-01 -1.92635909e-01
4.05111313e-01 -1.60867617e-01 -9.37017892e-03 6.62928820e-01
4.34085339e-01 -4.59170699e-01 -1.08250499e+00 3.27501118e-01
-4.73274022e-01 3.92953902e-01 5.84241390e-01 6.62022471e-01
4.65508878e-01 -3.10697824e-01 4.54525769e-01 -4.43748653e-01
-1.60940126e-01 -5.44727027e-01 -5.97769432e-02 2.98271179e-01
-2.20720649e-01 -2.36169487e-01 3.68350536e-01 -5.88364959e-01
-1.21482515e+00 4.68401253e-01 1.94570199e-01 -7.77723908e-01
-1.84964061e-01 2.37182140e-01 -3.26966643e-01 3.28656018e-01
6.85555562e-02 2.93413341e-01 -2.79753745e-01 -3.00114334e-01
3.34930480e-01 3.11227143e-01 4.26657766e-01 -3.19609702e-01
2.24472150e-01 8.20522904e-01 -5.14497831e-02 -3.34521443e-01
-1.08038449e+00 -7.14656234e-01 -8.86330068e-01 -8.44574690e-01
1.24834299e+00 -1.18823302e+00 -1.01624680e+00 4.30793107e-01
-1.19956446e+00 -3.97006750e-01 -4.60135818e-01 8.46490860e-01
-1.10208774e+00 3.04461241e-01 -2.79858351e-01 -7.48792946e-01
-2.27560699e-01 -1.04003763e+00 1.46086025e+00 1.26099378e-01
-2.07435012e-01 -7.25968361e-01 -7.47440159e-02 8.11807275e-01
1.81845456e-01 3.98838490e-01 5.75347304e-01 -3.84491324e-01
-1.09889805e+00 -3.53845417e-01 -1.52305514e-01 -4.76146638e-02
4.05990630e-02 -2.14701355e-01 -7.66448796e-01 -1.43087223e-01
-1.85601905e-01 -2.09731817e-01 1.01562190e+00 6.53492272e-01
1.40901351e+00 -2.96348274e-01 -6.27824903e-01 4.06300128e-01
9.59883749e-01 3.15630913e-01 7.43409812e-01 1.80896223e-01
9.55668330e-01 5.67555666e-01 1.42088330e+00 6.16316974e-01
4.61542606e-01 1.06859434e+00 8.45802188e-01 1.27009779e-01
-2.07454249e-01 -3.49817753e-01 6.62817121e-01 3.32349002e-01
-3.73165548e-01 -5.06664991e-01 -5.05936265e-01 9.44004655e-01
-2.40796447e+00 -1.38401973e+00 9.62511450e-02 1.78065896e+00
7.06847608e-02 -8.81112963e-02 1.33841887e-01 -2.42531717e-01
5.96429706e-01 3.72285664e-01 -6.47668540e-01 9.75030214e-02
1.73549742e-01 -3.99486721e-01 1.10415131e-01 2.07296193e-01
-1.31757593e+00 1.11853337e+00 5.31209898e+00 7.47966051e-01
-9.41163301e-01 3.56733352e-01 4.61820126e-01 -5.79236269e-01
1.26764879e-01 7.91209042e-02 -8.02464187e-01 5.58082759e-01
6.31641984e-01 -7.97573477e-03 1.93075180e-01 1.10636520e+00
5.87771773e-01 -3.23732048e-01 -1.38274002e+00 9.29481447e-01
4.29826140e-01 -1.18007982e+00 4.94379736e-02 -5.78929484e-02
6.71180308e-01 -1.54514864e-01 3.37383412e-02 4.17800874e-01
-1.83003038e-01 -5.87960243e-01 9.21761870e-01 1.09226823e+00
4.52221334e-01 -5.67827582e-01 7.29867637e-01 1.70122564e-01
-1.35702348e+00 -5.47532201e-01 -6.54000044e-02 -1.62974775e-01
6.07922196e-01 2.25754142e-01 -7.59647489e-01 3.25055689e-01
7.69122243e-01 1.38638413e+00 -5.93724191e-01 9.57708240e-01
-2.43315071e-01 5.23998737e-01 -4.12550867e-02 -9.27795842e-03
4.59817350e-01 6.44177869e-02 6.17528200e-01 8.76479805e-01
3.74388397e-01 2.47837067e-01 1.98970169e-01 4.95872378e-01
2.35936657e-01 -6.29340038e-02 -5.62444389e-01 -1.26900047e-01
-1.26536831e-01 1.23158741e+00 -5.89590311e-01 -5.90761304e-01
-3.36320966e-01 1.02218175e+00 3.56569886e-01 4.81583357e-01
-1.22852075e+00 3.30540091e-02 6.66219413e-01 2.23958224e-01
7.18357205e-01 -1.00680113e-01 2.34500676e-01 -1.07886159e+00
2.42804423e-01 -4.14435416e-01 4.51438904e-01 -1.22850549e+00
-7.02585936e-01 3.63597810e-01 1.97122917e-01 -1.58734846e+00
-5.46451926e-01 -2.48297691e-01 -6.49589539e-01 2.82528251e-01
-1.04108405e+00 -1.61100912e+00 -6.50468469e-01 5.70714474e-01
1.08015776e+00 -1.41283065e-01 4.74490225e-01 3.48440647e-01
-4.60059553e-01 2.41583899e-01 -2.21744090e-01 3.69935408e-02
6.35315299e-01 -8.70324790e-01 -4.06939723e-03 7.67643332e-01
5.68583161e-02 1.58761874e-01 4.95067388e-01 -8.13078880e-01
-1.22089577e+00 -1.24381304e+00 9.29280281e-01 -5.71509719e-01
3.95633519e-01 -4.84632462e-01 -4.30127025e-01 9.83136773e-01
9.88734215e-02 1.99949920e-01 3.89946043e-01 -4.06850874e-02
1.59323677e-01 -3.18785310e-01 -8.22067559e-01 7.14989662e-01
1.48956704e+00 -2.19947189e-01 -3.08486879e-01 6.17127895e-01
7.38291025e-01 -1.52957410e-01 -6.87315047e-01 5.77542067e-01
7.75236547e-01 -8.88980210e-01 9.83109832e-01 -8.55985045e-01
7.72826552e-01 -2.79331058e-01 -5.49889132e-02 -9.80150223e-01
-5.93254566e-01 -3.33359718e-01 -5.56498468e-01 1.18119717e+00
-1.67188242e-01 3.48043665e-02 9.44106340e-01 5.29565692e-01
-4.15380001e-01 -1.08751214e+00 -1.03007746e+00 -3.87028754e-01
-7.98073292e-01 -3.85046780e-01 5.37395477e-01 5.98992467e-01
-3.67390215e-02 6.98790550e-02 -9.53642190e-01 -8.23772401e-02
4.10631537e-01 2.70016998e-01 9.62832451e-01 -8.36808443e-01
1.07379250e-01 -7.59323463e-02 -7.25509882e-01 -1.18323088e+00
2.90247500e-01 -6.02400899e-01 1.39742196e-01 -1.60652709e+00
3.31672132e-01 -1.26108497e-01 -5.39355159e-01 5.12936354e-01
-1.16719045e-01 2.37606287e-01 3.53888690e-01 7.41931424e-02
-1.21968937e+00 1.04547453e+00 1.42263174e+00 -3.40619564e-01
-8.36482197e-02 -3.92770115e-03 -2.53711194e-01 7.61516809e-01
4.14065123e-01 -5.79423010e-01 -4.48722303e-01 -4.08850312e-01
-3.74734849e-02 1.46891803e-01 7.03006744e-01 -1.26226950e+00
3.22655231e-01 -3.98489177e-01 3.82071078e-01 -9.95935202e-01
8.22056174e-01 -9.98865902e-01 1.93115715e-02 4.19699192e-01
-5.14637887e-01 -1.86213642e-01 -9.64827240e-02 8.40986848e-01
-1.85081363e-01 3.18193771e-02 2.68936396e-01 -7.58971944e-02
-1.25223160e+00 6.61411941e-01 -4.07359958e-01 -1.84698939e-01
1.60729575e+00 -3.62117440e-01 -3.64435464e-02 -3.99022639e-01
-9.92775798e-01 4.13870573e-01 8.28338116e-02 8.85678351e-01
7.74213076e-01 -1.45693314e+00 -6.69869184e-01 8.80702585e-02
2.65765607e-01 -2.28954241e-01 8.23085189e-01 1.17625189e+00
5.74443974e-02 5.21429896e-01 -4.91599739e-01 -7.04814970e-01
-1.40272522e+00 9.01209593e-01 1.89608648e-01 -2.13821888e-01
-7.64765739e-01 7.50045419e-01 7.14226365e-01 2.48416573e-01
2.97263712e-01 -1.13639615e-01 -5.31685352e-01 2.14787334e-01
5.63394308e-01 5.28400302e-01 -4.13091689e-01 -7.78658330e-01
-4.78495598e-01 6.70038581e-01 -2.15410888e-01 1.15388058e-01
1.42848039e+00 -2.50873566e-01 1.54620618e-01 4.23486292e-01
1.04342651e+00 -3.74354601e-01 -1.82126892e+00 -6.08737171e-02
-4.20288920e-01 -7.97114968e-01 -7.03921616e-02 -6.17278576e-01
-1.27225471e+00 9.44589794e-01 7.99539030e-01 -3.80669981e-01
1.18535924e+00 9.75801349e-02 9.52227592e-01 1.39579117e-01
3.84602547e-01 -1.13963497e+00 6.69594407e-01 1.55447721e-01
1.00429285e+00 -1.15445542e+00 1.03629589e-01 -4.13011283e-01
-9.42349315e-01 8.86463881e-01 1.09458733e+00 -2.98454277e-02
5.80921054e-01 -3.59883487e-01 -2.29789615e-01 -2.22156495e-01
-9.17523742e-01 -2.45449126e-01 4.02798176e-01 5.02954006e-01
2.98065878e-02 -2.04724938e-01 -1.98835075e-01 4.49241638e-01
3.34867120e-01 2.91447341e-01 1.82157800e-01 7.92731047e-01
-2.36537814e-01 -6.43037677e-01 5.05641326e-02 3.88807446e-01
-2.39290833e-01 1.57388657e-01 -1.96284771e-01 6.13211751e-01
5.46587884e-01 6.20597303e-01 3.08761954e-01 -4.22520995e-01
1.70587718e-01 4.32056151e-02 4.56897080e-01 -3.64441574e-01
-4.23228234e-01 -5.06649315e-02 -1.11185424e-02 -1.03313673e+00
-8.69757175e-01 -1.00190878e+00 -1.09988546e+00 1.28077745e-01
-2.08680630e-01 -2.06508450e-02 4.96095896e-01 1.17928541e+00
7.02163935e-01 6.38033867e-01 5.84362984e-01 -1.30278063e+00
3.27090323e-02 -1.04979038e+00 -1.71452790e-01 6.95562184e-01
2.11535394e-01 -1.14550042e+00 -2.10271418e-01 2.13739842e-01]
|
[8.291463851928711, 0.52166348695755]
|
4e9ec6fd-4ed4-4791-a95a-426c274bd2d7
|
gaussian-processes-for-music-audio-modelling
|
1606.01039
| null |
http://arxiv.org/abs/1606.01039v2
|
http://arxiv.org/pdf/1606.01039v2.pdf
|
Gaussian Processes for Music Audio Modelling and Content Analysis
|
Real music signals are highly variable, yet they have strong statistical
structure. Prior information about the underlying physical mechanisms by which
sounds are generated and rules by which complex sound structure is constructed
(notes, chords, a complete musical score), can be naturally unified using
Bayesian modelling techniques. Typically algorithms for Automatic Music
Transcription independently carry out individual tasks such as multiple-F0
detection and beat tracking. The challenge remains to perform joint estimation
of all parameters. We present a Bayesian approach for modelling music audio,
and content analysis. The proposed methodology based on Gaussian processes
seeks joint estimation of multiple music concepts by incorporating into the
kernel prior information about non-stationary behaviour, dynamics, and rich
spectral content present in the modelled music signal. We illustrate the
benefits of this approach via two tasks: pitch estimation, and inferring
missing segments in a polyphonic audio recording.
|
['Dan Stowell', 'Pablo A. Alvarado']
|
2016-06-03
| null | null | null | null |
['music-transcription']
|
['music']
|
[ 4.22590554e-01 -2.70194739e-01 3.46588612e-01 2.10744977e-01
-9.46137607e-01 -8.72257054e-01 5.06116450e-01 -4.83024679e-02
-9.34278592e-02 4.85365212e-01 3.50002974e-01 1.91013440e-01
-6.66351795e-01 -2.43539289e-01 -3.85773718e-01 -9.17775989e-01
-2.52052337e-01 2.30025887e-01 2.14644179e-01 1.37931988e-01
3.99932593e-01 1.73292264e-01 -1.70466590e+00 -1.10370917e-02
3.61978680e-01 7.70373404e-01 3.64272863e-01 1.46267712e+00
1.77087843e-01 7.28838205e-01 -8.27767551e-01 -2.63236314e-01
-1.08751096e-01 -6.20932460e-01 -2.69334584e-01 -7.80176977e-03
-8.99547413e-02 -4.61884253e-02 1.61778092e-01 9.90205109e-01
5.94010830e-01 2.68930614e-01 1.03733599e+00 -7.07858860e-01
1.41442614e-02 9.30926383e-01 -1.86401144e-01 -1.45318925e-01
5.01359820e-01 -2.26348385e-01 1.19366741e+00 -7.91880846e-01
1.77645057e-01 9.93326485e-01 1.01830041e+00 4.02073422e-03
-1.57934678e+00 -5.34823179e-01 -6.51139081e-01 1.89788714e-02
-1.61779833e+00 -6.83628798e-01 1.05134737e+00 -7.43433177e-01
3.87007207e-01 4.94206876e-01 6.81310177e-01 1.09163904e+00
3.51088382e-02 8.30029309e-01 8.05736780e-01 -7.42533803e-01
2.17062280e-01 -1.09270163e-01 -2.22548153e-02 8.40611383e-02
-2.33789772e-01 2.91367948e-01 -1.02714241e+00 -6.61373734e-01
9.05796409e-01 -5.80902696e-01 -2.51156598e-01 -4.82067047e-03
-1.27926123e+00 4.48200524e-01 -5.89430332e-01 3.64281178e-01
-5.33803582e-01 5.08009613e-01 3.84403169e-01 4.77962866e-02
1.38445124e-02 4.43111748e-01 -4.11812395e-01 -6.48675680e-01
-1.36232984e+00 6.96432710e-01 1.03473151e+00 6.79239213e-01
1.26413360e-01 2.76262730e-01 -3.00259382e-01 8.79993618e-01
5.82780361e-01 6.36062026e-01 4.02623683e-01 -1.23110044e+00
-1.06765926e-01 -6.06628120e-01 4.82065529e-01 -7.62141645e-01
-1.40038788e-01 -5.99336684e-01 -5.51485002e-01 3.02473195e-02
6.46421313e-01 -9.10339281e-02 -4.36295539e-01 1.70708466e+00
1.60504743e-01 5.99897325e-01 -1.16143145e-01 5.18965065e-01
3.98743749e-01 6.16279662e-01 -1.12070419e-01 -5.49471974e-01
1.43610764e+00 -2.28677869e-01 -1.12552226e+00 3.28876704e-01
-1.60070881e-01 -1.40570831e+00 7.15551555e-01 1.17566705e+00
-1.37553787e+00 -9.06549096e-01 -9.84527409e-01 3.11975121e-01
5.26067615e-01 3.06053370e-01 3.51235062e-01 9.11581993e-01
-5.79462528e-01 1.03457904e+00 -7.36669958e-01 3.71390909e-01
-2.23390296e-01 2.12802157e-01 1.70182407e-01 5.93339682e-01
-9.89145219e-01 3.81504714e-01 5.28138459e-01 -1.11995381e-03
-1.07673740e+00 -8.72105837e-01 -4.53776658e-01 1.79906249e-01
4.91109155e-02 -5.03286481e-01 1.84152663e+00 -6.74058497e-01
-1.83401334e+00 3.42294663e-01 -1.34403467e-01 -3.22536141e-01
2.89284319e-01 -5.53526461e-01 -4.77008224e-01 1.44522741e-01
-2.72929311e-01 -2.42309235e-02 1.68607867e+00 -1.00593650e+00
-3.44178975e-01 4.04729135e-02 -6.72676802e-01 4.61487398e-02
2.04988986e-01 3.43560278e-01 -3.05343598e-01 -1.08631098e+00
3.39757234e-01 -8.88686359e-01 -4.74713966e-02 -5.33408999e-01
-5.25094092e-01 1.64055318e-01 2.33511269e-01 -9.33177471e-01
1.45237911e+00 -2.56207681e+00 3.08384895e-01 5.08616745e-01
-2.95631796e-01 -1.42814904e-01 1.86401501e-01 4.96394902e-01
-8.00984353e-03 -3.37946296e-01 -2.26100773e-01 -2.70884305e-01
3.71207893e-01 -9.88317356e-02 -8.00857961e-01 2.72785306e-01
-1.08466893e-01 3.27348977e-01 -7.79064775e-01 -2.86656976e-01
1.54603943e-01 8.28280807e-01 -4.51287150e-01 2.23360032e-01
-2.05762923e-01 5.27864218e-01 -1.49036393e-01 2.82570004e-01
3.07200372e-01 4.01077002e-01 1.80169821e-01 -2.60668695e-01
-2.28859574e-01 4.63110417e-01 -1.81487787e+00 1.82450247e+00
-1.58811092e-01 6.35776639e-01 4.87765044e-01 -5.96242368e-01
1.13308072e+00 9.73902822e-01 4.27959859e-01 2.08437651e-01
-9.03659221e-03 2.74397433e-01 3.59760791e-01 -3.22709233e-01
5.09376109e-01 -6.15352154e-01 -1.22319190e-02 4.71745849e-01
3.36718738e-01 -6.26011848e-01 4.84675243e-02 -1.74083173e-01
7.76202023e-01 4.25679535e-01 4.89218205e-01 -2.56653190e-01
3.70824754e-01 -5.36190569e-01 3.23123664e-01 7.85707176e-01
1.52466046e-02 6.59375310e-01 3.57083440e-01 1.63868934e-01
-9.88106728e-01 -1.43329453e+00 -3.88469547e-01 1.11385882e+00
-4.74055231e-01 -7.53040314e-01 -7.91488588e-01 4.60141540e-01
-1.37345761e-01 7.63939381e-01 -2.18721136e-01 -8.75401646e-02
-4.55084831e-01 -7.33299136e-01 8.29859316e-01 2.07850516e-01
-2.97051072e-01 -1.25103688e+00 -5.99659204e-01 8.49479973e-01
-3.83845896e-01 -8.18263710e-01 -4.28722769e-01 3.30954850e-01
-9.50069129e-01 -7.11119354e-01 -5.74226797e-01 -3.01984519e-01
-4.49898243e-01 -2.18185320e-01 9.97642457e-01 -4.61496174e-01
-6.92778289e-01 6.57117605e-01 -1.20926373e-01 -9.50661838e-01
-7.08158016e-01 -4.09290433e-01 2.45705098e-01 3.52353722e-01
-1.92487970e-01 -1.18807757e+00 -4.26858634e-01 2.66505599e-01
-8.02633703e-01 -1.38515934e-01 5.33479273e-01 4.99102265e-01
5.89263320e-01 4.65713501e-01 4.78201568e-01 -3.80236536e-01
9.38561141e-01 -1.03944331e-01 -2.30808944e-01 -8.84835124e-02
1.16180107e-01 -1.54249772e-01 1.05015539e-01 -7.26923168e-01
-1.12186837e+00 1.38993174e-01 -1.55076399e-01 -3.87678891e-01
-4.03396904e-01 4.36750263e-01 -1.60569042e-01 4.34697300e-01
7.75263369e-01 3.84385616e-01 -1.88384786e-01 -6.63721383e-01
4.31227475e-01 5.52396178e-01 1.14185953e+00 -9.13917422e-01
7.20502436e-01 3.31598908e-01 1.77269384e-01 -1.25600123e+00
-7.15431690e-01 -7.42491305e-01 -8.37217271e-01 -3.07246238e-01
8.39237392e-01 -9.11446393e-01 -7.76368141e-01 5.52834153e-01
-1.19905078e+00 -1.53749168e-01 -4.34930176e-01 1.04620481e+00
-1.24788666e+00 4.55967963e-01 -6.41802907e-01 -1.69053459e+00
-2.63778508e-01 -5.47169805e-01 9.96757746e-01 -1.58350006e-01
-9.49547827e-01 -8.98313463e-01 5.26234269e-01 1.01285189e-01
1.17348485e-01 2.02878803e-01 7.98093200e-01 -4.00560081e-01
-4.20008600e-01 -1.46957546e-01 4.30647194e-01 3.89613718e-01
1.22597277e-01 4.04664874e-01 -1.38472009e+00 5.49573228e-02
4.89280224e-01 4.12243381e-02 5.64290285e-01 9.29500639e-01
8.05380285e-01 1.58655159e-02 1.34263486e-01 3.17975998e-01
1.11265159e+00 9.52966511e-02 5.59880435e-01 -3.40952665e-01
3.15969378e-01 6.55607760e-01 5.82097232e-01 9.08482075e-01
-4.69523698e-01 7.96474695e-01 1.18394904e-01 5.53827405e-01
-1.03647277e-01 -1.18769094e-01 6.74159288e-01 1.42982817e+00
-4.69558328e-01 1.50700077e-01 -6.46757483e-01 5.23480475e-01
-1.70514059e+00 -1.43248522e+00 -4.79121417e-01 2.35128713e+00
9.62303758e-01 2.85822541e-01 3.62912774e-01 9.64313507e-01
7.97462344e-01 -2.69551456e-01 -3.21932614e-01 -1.52189314e-01
-1.56647135e-02 6.24011040e-01 1.46957815e-01 5.03683150e-01
-1.17305803e+00 4.26059574e-01 7.44550323e+00 1.19388926e+00
-4.78332341e-01 -1.02466129e-01 -4.15448636e-01 -1.65666088e-01
-1.17621459e-01 2.21784730e-02 -6.45148993e-01 1.56220168e-01
1.18768394e+00 -1.30475778e-02 5.85201502e-01 3.84933293e-01
5.27888477e-01 -9.84206721e-02 -1.00566399e+00 1.19380379e+00
-5.99797666e-02 -9.05770123e-01 -2.50872552e-01 -6.27941499e-03
4.42417115e-01 -3.27733576e-01 1.83738515e-01 1.68373123e-01
1.39300033e-01 -9.02788103e-01 1.33142006e+00 1.03615725e+00
5.26696265e-01 -7.88620591e-01 1.29703104e-01 5.40256858e-01
-1.25422919e+00 -3.09353508e-02 -9.20099244e-02 -1.19496979e-01
4.03155357e-01 8.77798140e-01 -8.17088723e-01 6.21759892e-01
3.56430799e-01 6.27025306e-01 -2.54105106e-02 1.52824771e+00
-3.34391743e-01 1.21831787e+00 -3.59577119e-01 4.28485930e-01
-2.19489962e-01 -3.27647299e-01 1.09527540e+00 1.34913719e+00
7.58573830e-01 -7.59913325e-02 -1.03345709e-02 9.15117204e-01
6.49933100e-01 1.70016944e-01 -1.19826958e-01 -1.66181877e-01
4.11545813e-01 1.21797168e+00 -8.05139363e-01 -1.68496132e-01
-6.17022440e-02 6.34214699e-01 -5.24368286e-01 2.66418874e-01
-6.97307706e-01 -3.64255041e-01 5.56414783e-01 -1.59634516e-01
5.50577462e-01 -4.59940404e-01 -1.22534178e-01 -7.85968781e-01
-2.72453606e-01 -9.68712628e-01 1.35563493e-01 -1.03169084e+00
-1.14430749e+00 -5.23189195e-02 -1.18818581e-01 -1.22055101e+00
-5.92986822e-01 -5.11397719e-01 -8.95742655e-01 1.07133567e+00
-8.40849578e-01 -7.22298682e-01 4.61388469e-01 7.54045546e-01
5.24493635e-01 1.34022748e-02 1.16610944e+00 1.35518923e-01
7.28632733e-02 -2.51620784e-02 3.02184224e-01 -2.08409354e-01
6.78555250e-01 -1.51679397e+00 2.59958357e-01 3.19994748e-01
8.85652661e-01 5.56218266e-01 1.25423253e+00 -5.96745670e-01
-1.07740474e+00 -5.18061280e-01 5.83559692e-01 -6.47992134e-01
9.25197840e-01 -3.05098802e-01 -8.85330141e-01 2.05666706e-01
-1.24700688e-01 -7.93190241e-01 1.20765531e+00 3.60872269e-01
-1.68357670e-01 1.81439742e-01 -4.84365135e-01 2.88739473e-01
4.71809119e-01 -9.35523152e-01 -1.03081441e+00 -2.09450405e-02
3.76935095e-01 1.79505814e-02 -8.66770864e-01 1.42229408e-01
1.08726203e+00 -1.06319678e+00 1.19244921e+00 -2.71616220e-01
-1.54746026e-01 -6.48880601e-01 -3.99667829e-01 -1.11814392e+00
-3.94309729e-01 -1.45981658e+00 -4.41981256e-01 1.43598640e+00
1.38011761e-02 2.71943033e-01 3.04000616e-01 5.62115274e-02
-9.00690854e-02 1.27308205e-01 -8.11247528e-01 -8.24053645e-01
-3.82826418e-01 -1.11276639e+00 1.89053625e-01 5.96564233e-01
1.49336532e-02 4.19414669e-01 -6.48850381e-01 3.36493820e-01
1.12042642e+00 -5.10098301e-02 7.21070468e-01 -1.75606227e+00
-1.07098055e+00 -5.27650774e-01 -2.44326726e-01 -6.77536190e-01
-1.72857389e-01 -6.02622807e-01 1.67155772e-01 -8.58111262e-01
4.52806503e-02 -5.88827133e-02 -4.47172910e-01 -3.36974263e-01
-5.51917590e-02 1.65581867e-01 1.96945861e-01 3.84215504e-01
-1.71410426e-01 4.41778928e-01 9.89331365e-01 2.08492950e-01
-3.65456790e-01 6.06220961e-01 -9.75994989e-02 1.16508853e+00
5.32176256e-01 -5.85358918e-01 -4.25410837e-01 2.04522833e-01
5.30221760e-01 5.48265576e-01 4.27520752e-01 -1.30522728e+00
1.96488187e-01 2.33386010e-01 2.49097139e-01 -7.94880450e-01
6.37991369e-01 -5.27996063e-01 7.89252400e-01 2.55572498e-01
-4.25725877e-01 -4.28856552e-01 1.70422599e-01 7.52826691e-01
-3.80955935e-01 -7.18983173e-01 5.67377031e-01 -9.46911145e-03
-6.42365366e-02 -2.31526911e-01 -7.32304931e-01 -1.56581238e-01
3.87999147e-01 -5.18581830e-02 8.03256214e-01 -5.49823463e-01
-1.53555453e+00 -7.42490292e-01 -2.31072187e-01 7.79840425e-02
2.73538947e-01 -1.29025137e+00 -8.35610449e-01 1.46674752e-01
-3.49091947e-01 -4.62658793e-01 4.68569487e-01 7.37256527e-01
-2.04469338e-01 2.06594661e-01 7.58296698e-02 -6.93088114e-01
-1.45292139e+00 3.99655700e-01 8.27257186e-02 -8.58992636e-02
-4.40739691e-01 8.21202338e-01 2.64000565e-01 4.85507846e-02
2.80656725e-01 -3.72444600e-01 -3.53936613e-01 3.33106488e-01
7.95394838e-01 5.35438597e-01 -1.09593801e-01 -6.23306870e-01
9.11799297e-02 7.26653099e-01 4.64258075e-01 -8.86531115e-01
9.13210392e-01 -3.28180403e-01 -9.62466225e-02 1.33084595e+00
4.78923917e-01 5.89864075e-01 -1.04955876e+00 -5.03798313e-02
2.59517193e-01 -2.22417191e-01 9.73302796e-02 -5.67351460e-01
-2.52007544e-02 9.42106843e-01 2.08864719e-01 4.10033911e-01
9.70350027e-01 -2.07791314e-01 3.83982152e-01 2.13255212e-01
2.69410521e-01 -1.04180324e+00 1.15371704e-01 4.50945854e-01
9.19307709e-01 -2.82558680e-01 -8.35993960e-02 -3.05560231e-01
-2.83082038e-01 1.36483490e+00 -3.22784543e-01 -2.11814731e-01
9.63070273e-01 3.72369051e-01 -5.31175965e-03 -2.71084402e-02
-6.64238572e-01 -1.26016796e-01 7.15928555e-01 5.50446272e-01
5.62636197e-01 2.86969692e-01 8.40154067e-02 1.08448124e+00
-8.55503976e-01 -3.15785468e-01 4.85135317e-01 4.86199588e-01
-6.41275704e-01 -1.10873449e+00 -1.07314801e+00 2.25401148e-02
-9.23644125e-01 -2.87532508e-01 -1.30506173e-01 2.35919371e-01
1.22004807e-01 9.80980515e-01 -8.04455653e-02 8.43430098e-05
2.73849517e-01 6.35918260e-01 8.64622593e-01 -5.09671152e-01
-4.55106229e-01 1.07847154e+00 9.94870588e-02 -1.02258682e-01
-5.12107074e-01 -1.16853786e+00 -8.56354892e-01 2.77639925e-01
-4.65424478e-01 4.20793086e-01 9.67254817e-01 7.93099046e-01
-2.55130470e-01 7.90969074e-01 3.69061917e-01 -1.12718213e+00
-7.64623284e-01 -1.18047535e+00 -1.35214674e+00 3.05414915e-01
3.35142374e-01 -3.14670354e-01 -4.64371532e-01 6.02795362e-01]
|
[15.729720115661621, 5.486980438232422]
|
b39d7f4b-05c1-4003-b1a2-5bea689d6b50
|
lanns-a-web-scale-approximate-nearest
|
2010.09426
| null |
https://arxiv.org/abs/2010.09426v1
|
https://arxiv.org/pdf/2010.09426v1.pdf
|
LANNS: A Web-Scale Approximate Nearest Neighbor Lookup System
|
Nearest neighbor search (NNS) has a wide range of applications in information retrieval, computer vision, machine learning, databases, and other areas. Existing state-of-the-art algorithm for nearest neighbor search, Hierarchical Navigable Small World Networks(HNSW), is unable to scale to large datasets of 100M records in high dimensions. In this paper, we propose LANNS, an end-to-end platform for Approximate Nearest Neighbor Search, which scales for web-scale datasets. Library for Large Scale Approximate Nearest Neighbor Search (LANNS) is deployed in multiple production systems for identifying topK ($100 \leq topK \leq 200$) approximate nearest neighbors with a latency of a few milliseconds per query, high throughput of 2.5k Queries Per Second (QPS) on a single node, on large ($\sim$180M data points) high dimensional (50-2048 dimensional) datasets.
|
['Niranjan Balasubramanian', 'Rushi Bhatt', 'Rajeev Kumar', 'Ashish Bhutani', 'Dhritiman Das', 'Ishita Doshi']
|
2020-10-19
| null | null | null | null |
['2048']
|
['playing-games']
|
[-5.54038346e-01 -5.70548475e-01 -3.68109465e-01 -5.89790106e-01
-9.67625558e-01 -6.66051090e-01 5.02015173e-01 5.45741618e-01
-6.31552517e-01 7.10173786e-01 2.98898786e-01 -4.65085596e-01
-9.54245329e-01 -1.29466283e+00 -4.84621406e-01 -6.88484758e-02
-5.93506336e-01 1.22714138e+00 9.34024632e-01 -1.08144999e-01
9.04672921e-01 1.12660229e+00 -1.51292360e+00 3.63757610e-01
4.25645202e-01 1.27098858e+00 -2.24572003e-01 5.78362286e-01
-3.88082027e-01 3.57296675e-01 -1.59739912e-01 -1.67302847e-01
5.31528950e-01 3.71654034e-01 -1.18287599e+00 -1.29844987e+00
7.49035060e-01 -8.04970920e-01 -8.38565171e-01 5.24583280e-01
8.74657869e-01 3.49710464e-01 5.12739718e-01 -1.55172205e+00
-2.91797847e-01 4.96828556e-01 -3.05134177e-01 3.52897882e-01
4.81187671e-01 -1.02106310e-01 6.37112200e-01 -1.04034996e+00
6.85704648e-01 1.29542601e+00 1.00711584e+00 -1.19364774e-02
-8.30315769e-01 -7.94455767e-01 -6.82280362e-01 6.20832801e-01
-1.97614491e+00 -4.23459768e-01 -7.39374831e-02 3.57405186e-01
1.43567145e+00 4.83541250e-01 5.79967499e-01 2.71639079e-01
4.31845039e-01 1.75628379e-01 6.28563225e-01 1.65441558e-02
6.78701341e-01 -2.88138181e-01 2.63056308e-01 2.14222267e-01
1.60557032e-01 2.20200464e-01 -8.98103714e-01 -9.34498787e-01
5.41801512e-01 4.11873668e-01 1.14138745e-01 -4.41990614e-01
-1.33798683e+00 8.78480256e-01 9.67073917e-01 2.90079266e-01
-4.21635062e-01 5.74429214e-01 5.21532238e-01 5.59309781e-01
2.30533734e-01 3.02503824e-01 -4.41798478e-01 -4.33147281e-01
-1.06838894e+00 5.14827490e-01 1.14157522e+00 1.53407955e+00
8.81562412e-01 -6.94935501e-01 -2.39045575e-01 7.36438930e-01
4.89816219e-02 5.41662693e-01 5.09215772e-01 -1.44500315e+00
4.93872732e-01 5.85003376e-01 2.91245878e-01 -1.13720155e+00
-6.06586993e-01 2.10773304e-01 -1.20080364e+00 2.11593118e-02
1.66722182e-02 5.46721041e-01 -3.79891276e-01 9.14541543e-01
8.18657398e-01 1.61363259e-01 5.14116585e-02 8.12720001e-01
5.75559735e-01 1.08943784e+00 -2.26550117e-01 1.39969112e-02
1.30416238e+00 -9.53252852e-01 7.76218846e-02 4.85164262e-02
8.60278606e-01 -9.58360016e-01 9.70418155e-01 7.31959045e-02
-1.01318729e+00 -1.86134920e-01 -7.45683253e-01 -5.25491059e-01
-9.15455818e-01 -1.00140047e+00 4.16108608e-01 -4.96962965e-02
-1.32294154e+00 7.75132179e-01 -6.07294858e-01 -8.99974167e-01
1.57279313e-01 7.50015557e-01 -5.52944243e-01 -7.64863193e-01
-9.85261619e-01 6.50420010e-01 5.56808233e-01 -2.94473320e-01
-6.38073757e-02 -9.65362847e-01 -4.17270422e-01 1.46064833e-01
1.21047445e-01 -8.34626734e-01 8.68782282e-01 4.82834935e-01
-6.58575475e-01 3.73495191e-01 -3.26267153e-01 -6.14774764e-01
2.28262112e-01 -1.20412230e-01 -5.63137949e-01 5.21034338e-02
3.77703309e-01 7.80757248e-01 -1.26494542e-01 -3.93216133e-01
-7.63151944e-01 -9.43901479e-01 -4.16616440e-01 2.93867081e-01
-5.40258825e-01 -2.46692952e-02 -3.77130061e-01 -3.57420206e-01
5.80894947e-01 -8.51422250e-01 -4.63867307e-01 6.84308052e-01
-7.21161589e-02 -6.92168415e-01 1.01758873e+00 1.18933938e-01
1.13550985e+00 -1.98035383e+00 -6.10526800e-01 8.42064559e-01
1.44448787e-01 2.43388608e-01 -3.39216322e-01 1.05202496e+00
5.80268443e-01 4.28112783e-02 1.62635997e-01 1.42500028e-01
2.24533394e-01 2.83278227e-01 -2.44821906e-01 4.49407905e-01
-1.17288363e+00 9.20005798e-01 -8.74392688e-01 -7.02953279e-01
2.99898058e-01 4.50457573e-01 -1.88405693e-01 2.99035892e-04
2.98003078e-01 -5.51503658e-01 -6.58800423e-01 6.24303341e-01
9.26156223e-01 -2.33291700e-01 -3.67952675e-01 -2.89618909e-01
-2.38225520e-01 6.74958825e-02 -1.39871502e+00 2.03110385e+00
-3.70486587e-01 3.84053618e-01 -2.92137057e-01 -3.67018312e-01
9.15357888e-01 8.14404562e-02 6.73562527e-01 -9.67404842e-01
-5.08898735e-01 9.62010324e-01 -7.22274005e-01 -4.64345478e-02
5.95033467e-01 6.91307962e-01 -1.54028654e-01 7.42023051e-01
-5.01277149e-01 -1.14231385e-01 2.02027082e-01 2.99618602e-01
1.69057524e+00 -8.20802391e-01 4.26258355e-01 -4.21406716e-01
2.58675158e-01 6.78118289e-01 1.06534606e-03 1.12029517e+00
8.91860127e-02 3.68717343e-01 -1.85876295e-01 -1.36254299e+00
-1.21269894e+00 -1.17738092e+00 -2.89604217e-01 1.04803574e+00
5.39516449e-01 -5.22309244e-01 -5.32867849e-01 -8.74226689e-02
5.60818851e-01 4.32697684e-01 1.93537429e-01 -6.97385669e-02
-6.92611396e-01 -6.78962842e-02 6.69010460e-01 5.30237615e-01
8.15023124e-01 -1.23058844e+00 -6.14911318e-01 4.84012127e-01
1.24988995e-01 -8.81710231e-01 -1.00437868e+00 -1.84371427e-01
-1.12408435e+00 -1.10120404e+00 -6.37252629e-01 -6.79887831e-01
3.45427722e-01 4.02039737e-01 1.31727779e+00 -1.64354339e-01
-5.01584470e-01 1.60631523e-01 -2.84881800e-01 2.16207981e-01
1.36712983e-01 4.68149751e-01 4.22515154e-01 -8.89428616e-01
1.00064862e+00 -9.33819234e-01 -1.23625851e+00 6.30875707e-01
-7.73274720e-01 -7.32841790e-01 7.32753813e-01 4.67720658e-01
1.11704385e+00 7.36688450e-02 2.87390500e-01 -6.87995911e-01
7.62972176e-01 -8.32413197e-01 -8.31424057e-01 1.80880189e-01
-1.36343515e+00 1.75653249e-02 6.95303500e-01 -2.82440215e-01
-1.05126565e-02 5.26217669e-02 -1.56975940e-01 -3.07274431e-01
-2.73034543e-01 1.16879456e-01 4.73080724e-01 -3.04436177e-01
8.80778134e-01 4.41348284e-01 -1.88302472e-01 -4.53008652e-01
6.10383451e-01 1.25433934e+00 7.28484929e-01 -3.00845057e-01
6.44581735e-01 4.80798870e-01 3.00155729e-01 -4.15039301e-01
-1.54975727e-01 -1.17108309e+00 -5.55159628e-01 5.71595371e-01
3.72697920e-01 -6.84618950e-01 -1.03293490e+00 2.76478916e-01
-1.24502647e+00 -1.59109309e-01 -2.57125586e-01 3.89552146e-01
-6.82291806e-01 7.20354691e-02 -5.79324722e-01 -3.34890515e-01
-1.12640631e+00 -5.39933145e-01 1.02717018e+00 7.36573637e-02
-4.04521972e-01 -5.00903070e-01 1.79312512e-01 2.26123258e-02
9.64963019e-01 4.11360003e-02 8.13355148e-01 -1.08042300e+00
-9.55368519e-01 -5.72432339e-01 -7.90452540e-01 -5.89006245e-01
-1.75468639e-01 -7.36202002e-01 -1.10313155e-01 -4.94729310e-01
-6.71849251e-01 -2.03378960e-01 2.98052222e-01 2.53287286e-01
1.20551717e+00 -8.68499279e-01 -8.84425342e-01 8.08680832e-01
1.77373374e+00 2.20980838e-01 5.97511828e-01 4.22636598e-01
3.15069973e-01 1.74807891e-01 8.50763798e-01 6.90710723e-01
4.35902447e-01 7.73554802e-01 5.68509102e-01 1.25263155e-01
7.90339336e-02 -2.38692105e-01 -4.13725585e-01 2.78420925e-01
6.64320230e-01 -5.50031483e-01 -1.19938362e+00 7.74336815e-01
-1.92742538e+00 -8.23078930e-01 -2.78104454e-01 2.44193673e+00
5.45723736e-01 2.26992816e-02 6.42796531e-02 -4.41618860e-02
5.47661841e-01 -6.75321296e-02 -1.06864297e+00 -5.38835347e-01
4.31890398e-01 8.02992657e-02 9.50111926e-01 2.57764280e-01
-5.72495162e-01 8.23143065e-01 5.89942503e+00 1.28107035e+00
-7.05327511e-01 -2.34505072e-01 4.75515842e-01 -2.52302498e-01
-1.46473348e-01 -9.06582475e-02 -8.68760884e-01 7.30504930e-01
1.34017861e+00 -2.22726643e-01 8.72397363e-01 1.28568876e+00
3.21969017e-02 -1.00149743e-01 -1.24060869e+00 1.39940584e+00
-4.01427031e-01 -1.84941733e+00 2.59407043e-01 3.16894144e-01
5.07099271e-01 6.96470082e-01 -2.42053911e-01 -1.42131045e-01
3.66544396e-01 -1.05234706e+00 4.68220599e-02 3.67613554e-01
1.15953028e+00 -1.28115845e+00 9.69947338e-01 5.38161337e-01
-1.48207319e+00 -1.69378117e-01 -7.20663488e-01 2.11191580e-01
1.99492589e-01 7.57547855e-01 -7.86701977e-01 -3.74549665e-02
1.52584875e+00 7.79160261e-02 -2.17316344e-01 1.32969570e+00
1.15159261e+00 -2.08715081e-01 -1.22870529e+00 -3.99457693e-01
3.76645952e-01 1.33104026e-01 2.48410821e-01 7.88250148e-01
8.72690439e-01 5.87868810e-01 4.27497067e-02 3.00647467e-01
-1.21261351e-01 2.09446728e-01 -9.35341656e-01 4.24753040e-01
1.64074659e+00 8.43162954e-01 -6.26291990e-01 -3.28507990e-01
8.14043060e-02 1.16774642e+00 2.12587357e-01 -5.89424372e-02
-2.75285691e-01 -1.38757288e+00 7.88295448e-01 6.52827203e-01
1.16591327e-01 -1.85106441e-01 -1.26532748e-01 -2.12250873e-01
2.97096848e-01 -6.33235991e-01 7.46368408e-01 -7.58469820e-01
-1.50909650e+00 6.57480776e-01 9.69518051e-02 -1.16648209e+00
-6.03574038e-01 -1.47946686e-01 -3.18531930e-01 8.27473402e-01
-1.19580495e+00 -8.73370469e-01 -6.43537402e-01 1.08063805e+00
2.40549847e-01 -8.71561170e-02 9.65922117e-01 5.02980053e-01
4.11827087e-01 7.05616474e-01 8.56420338e-01 -9.48787555e-02
7.34778583e-01 -9.73578155e-01 1.16905344e+00 -8.24197605e-02
1.51050717e-01 6.20980203e-01 3.19972783e-01 -4.39686626e-01
-1.54260027e+00 -1.02978635e+00 1.36468875e+00 -1.32780001e-01
4.55855042e-01 -6.32833764e-02 -5.70934534e-01 1.26044512e-01
-2.61255831e-01 9.52042520e-01 3.92515987e-01 -2.61930436e-01
-4.48601156e-01 -7.77353287e-01 -1.86079657e+00 4.58279997e-01
1.51809084e+00 -4.09340292e-01 3.42765413e-02 5.77784479e-01
8.50363791e-01 -4.73401815e-01 -1.51201117e+00 3.42061907e-01
8.32193851e-01 -1.32846868e+00 1.57747316e+00 -2.78611392e-01
-4.52443600e-01 -2.78364778e-01 -5.18945217e-01 -6.01372182e-01
-2.33927131e-01 -5.01509666e-01 -2.47582987e-01 1.02861154e+00
9.09945071e-02 -9.42364216e-01 1.13322294e+00 8.58066857e-01
3.00710291e-01 -1.10835862e+00 -1.54297674e+00 -8.88289273e-01
-3.09883449e-02 -1.40606239e-01 1.36180925e+00 5.23220241e-01
-6.42711073e-02 -3.73707190e-02 -4.55314666e-03 3.38551909e-01
8.03545237e-01 2.79611170e-01 7.73568690e-01 -1.38114989e+00
2.70373732e-01 -9.03456584e-02 -9.29406822e-01 -1.34348714e+00
-3.28302175e-01 -4.97572184e-01 -3.14648896e-01 -1.74436378e+00
6.29046932e-02 -1.15326941e+00 -1.74872681e-01 4.33608443e-01
6.75901175e-01 5.40262282e-01 -1.78219989e-01 7.03411579e-01
-8.55984986e-01 1.48254082e-01 4.20654207e-01 -3.84290889e-02
-7.82301053e-02 1.60406187e-01 -8.50215033e-02 2.93305486e-01
7.03519106e-01 -9.70739543e-01 -5.45765460e-01 -7.08869040e-01
6.60609230e-02 2.83945888e-01 2.03141794e-01 -1.37765443e+00
1.11944437e+00 -2.47602209e-01 2.78735191e-01 -1.09309304e+00
5.39935112e-01 -1.22518730e+00 3.78307134e-01 6.66992486e-01
-2.34873131e-01 7.27326095e-01 -1.93115547e-01 6.56021774e-01
-2.90115118e-01 -1.17606737e-01 5.02555192e-01 -2.45311141e-01
-8.81998718e-01 8.26169670e-01 1.31383717e-01 1.36915728e-01
1.16247928e+00 -4.59541708e-01 -4.24843460e-01 -2.93597907e-01
-1.01182543e-01 5.68751633e-01 6.04962528e-01 1.67729989e-01
8.97377431e-01 -1.70928228e+00 -6.54622316e-01 5.08805104e-02
2.59688556e-01 2.26705030e-01 2.05507398e-01 9.83614251e-02
-1.23755324e+00 7.09822476e-01 9.28323567e-02 -6.11725152e-01
-1.12142694e+00 7.32405901e-01 1.33662388e-01 -9.86784026e-02
-5.56710780e-01 7.07582295e-01 -9.44549024e-01 -7.20104396e-01
4.04412150e-01 -4.56362218e-02 4.78519469e-01 -4.40057278e-01
7.64564514e-01 1.26341915e+00 1.41955405e-01 -2.10701779e-01
-6.77563667e-01 9.52300787e-01 -1.10439457e-01 2.56965104e-02
1.22481787e+00 -3.76369804e-01 -4.73783255e-01 3.80808897e-02
1.79628122e+00 -4.69407231e-01 -4.53019410e-01 -3.13426048e-01
3.23799282e-01 -6.11524343e-01 -1.05586119e-01 -4.11350250e-01
-5.48325300e-01 3.18215877e-01 9.83267307e-01 1.80148199e-01
8.65543544e-01 1.92479610e-01 1.65911198e+00 1.14706933e+00
1.11143112e+00 -9.50819016e-01 -4.53088075e-01 3.93518239e-01
6.21892691e-01 -1.25741947e+00 1.85151622e-01 -8.50002840e-03
2.07117721e-01 1.26739168e+00 4.40151334e-01 -2.10170329e-01
1.12668204e+00 2.39213467e-01 -1.19690178e-02 -2.66488552e-01
-6.62569761e-01 4.60693270e-01 8.23563337e-02 6.78248703e-01
-1.71291903e-01 -3.18849891e-01 -1.14467941e-01 -1.30200133e-01
-3.91094565e-01 1.50947616e-01 -2.27616802e-01 7.96625793e-01
-7.31334388e-01 -1.14762294e+00 -2.92972326e-01 7.37416685e-01
-3.55924815e-02 -5.65711737e-01 3.99859250e-02 5.63341796e-01
-1.52926400e-01 8.65188181e-01 7.58484662e-01 -1.86007634e-01
3.12718600e-01 -3.29396248e-01 -2.95322686e-01 5.57602681e-02
-3.39552402e-01 -8.54775250e-01 -2.69810289e-01 -1.31625390e+00
2.73178250e-01 -1.25545159e-01 -1.55044746e+00 -1.37400091e+00
2.14311600e-01 6.36517823e-01 7.99751580e-01 3.07870626e-01
1.18767512e+00 -7.20165193e-01 5.67400873e-01 -4.74704593e-01
-7.71665990e-01 -5.63648462e-01 -7.38366902e-01 -4.04450670e-02
1.56877995e-01 1.23851188e-01 -2.79284805e-01 -6.52484119e-01]
|
[8.615116119384766, 3.5770280361175537]
|
640c4356-e733-43d4-be3c-83b109701810
|
viena2-a-driving-anticipation-dataset
|
1810.09044
| null |
http://arxiv.org/abs/1810.09044v2
|
http://arxiv.org/pdf/1810.09044v2.pdf
|
VIENA2: A Driving Anticipation Dataset
|
Action anticipation is critical in scenarios where one needs to react before
the action is finalized. This is, for instance, the case in automated driving,
where a car needs to, e.g., avoid hitting pedestrians and respect traffic
lights. While solutions have been proposed to tackle subsets of the driving
anticipation tasks, by making use of diverse, task-specific sensors, there is
no single dataset or framework that addresses them all in a consistent manner.
In this paper, we therefore introduce a new, large-scale dataset, called
VIENA2, covering 5 generic driving scenarios, with a total of 25 distinct
action classes. It contains more than 15K full HD, 5s long videos acquired in
various driving conditions, weathers, daytimes and environments, complemented
with a common and realistic set of sensor measurements. This amounts to more
than 2.25M frames, each annotated with an action label, corresponding to 600
samples per action class. We discuss our data acquisition strategy and the
statistics of our dataset, and benchmark state-of-the-art action anticipation
techniques, including a new multi-modal LSTM architecture with an effective
loss function for action anticipation in driving scenarios.
|
['Mohammad Sadegh Aliakbarian', 'Mathieu Salzmann', 'Lars Petersson', 'Lars Andersson', 'Fatemeh Sadat Saleh', 'Basura Fernando']
|
2018-10-22
| null | null | null | null |
['action-anticipation']
|
['computer-vision']
|
[ 4.79431331e-01 -1.15364596e-01 -5.53532243e-02 -6.58804059e-01
-6.19633377e-01 -3.39264601e-01 7.32783973e-01 -2.47775372e-02
-6.45740747e-01 6.79850757e-01 2.33807340e-01 -1.53003186e-01
-2.65717924e-01 -5.48803985e-01 -6.47707641e-01 -9.10223961e-01
-1.29654855e-01 3.43340248e-01 3.99275869e-01 -2.91219026e-01
1.10159092e-01 6.13223791e-01 -2.24450684e+00 3.14192265e-01
4.73502696e-01 1.26621747e+00 5.39516568e-01 9.67133939e-01
2.20926538e-01 1.07029772e+00 -3.58597189e-01 -1.71831742e-01
2.79532105e-01 -7.30854198e-02 -5.73546827e-01 3.29863608e-01
4.66134518e-01 -3.97686154e-01 -6.39157593e-01 6.01350129e-01
6.40254140e-01 3.89023453e-01 1.39050186e-01 -1.71126497e+00
1.33781120e-01 3.04769605e-01 -1.28796309e-01 2.64503211e-01
2.11431161e-01 7.08660126e-01 6.92568839e-01 -3.05801123e-01
4.07118976e-01 1.09478140e+00 3.83115798e-01 5.80917299e-01
-6.81362748e-01 -4.62868989e-01 1.50443733e-01 8.42226207e-01
-9.09796536e-01 -7.04292536e-01 7.12520003e-01 -4.42321271e-01
1.14117765e+00 1.63255855e-01 8.89228940e-01 1.31784511e+00
4.08399582e-01 1.07493925e+00 8.95959675e-01 3.78895923e-02
2.67365724e-01 -1.43476069e-01 5.46022318e-02 1.99683793e-02
-1.53585136e-01 2.21749917e-01 -5.87274194e-01 4.29449350e-01
-1.58582747e-01 5.19282930e-02 1.19013071e-01 -3.31611782e-01
-1.49018455e+00 4.94813532e-01 -1.19751999e-02 2.42269300e-02
-7.34235048e-01 3.09379727e-01 8.64637733e-01 2.57661283e-01
2.05565989e-01 -7.57551864e-02 -5.09844840e-01 -6.93735123e-01
-6.17994726e-01 5.16116142e-01 4.11845773e-01 9.74480391e-01
7.52515554e-01 2.95157850e-01 -4.91026253e-01 4.93783832e-01
1.19083218e-01 7.09440589e-01 3.15762401e-01 -1.25631177e+00
4.91106778e-01 2.75823295e-01 3.05352032e-01 -6.74151123e-01
-6.27948523e-01 1.29322568e-02 -7.02608168e-01 2.38473728e-01
2.93747604e-01 -2.68447936e-01 -7.67893970e-01 1.67323220e+00
2.46575698e-01 4.67357725e-01 2.63355196e-01 8.20037067e-01
6.48001075e-01 5.03133476e-01 2.15448320e-01 -1.69808954e-01
1.22486639e+00 -9.06754196e-01 -8.57761621e-01 -6.77651346e-01
5.64874232e-01 -6.44406617e-01 9.01977479e-01 5.65328062e-01
-9.18229401e-01 -8.52754951e-01 -1.02781153e+00 1.11315390e-02
-5.15723944e-01 -5.38996384e-02 4.74730819e-01 3.95289987e-01
-9.15024042e-01 3.19704145e-01 -6.97333395e-01 -4.81278419e-01
2.20019534e-01 1.48885638e-01 -3.87267500e-01 -2.47830674e-01
-1.46643913e+00 1.18496978e+00 5.15675902e-01 5.19610584e-01
-1.25992608e+00 -4.53435302e-01 -9.73464191e-01 -1.91250995e-01
5.31385005e-01 -2.91984439e-01 1.50869322e+00 -6.12143338e-01
-1.36043572e+00 6.51763022e-01 -2.33936265e-01 -1.03673220e+00
6.74489081e-01 -4.91409689e-01 -8.19208264e-01 -2.07934722e-01
-2.88519487e-02 1.07023501e+00 8.08954060e-01 -8.76178563e-01
-9.47503269e-01 -1.45042509e-01 4.21433628e-01 1.18392475e-01
-1.62972156e-02 1.16539262e-01 -2.76784271e-01 -5.02088554e-02
-4.68483388e-01 -1.16516519e+00 -4.31066245e-01 -2.57665753e-01
-2.60633320e-01 -9.49901268e-02 1.14971936e+00 -4.09915477e-01
1.09950435e+00 -2.25163555e+00 -4.62003751e-03 -4.06744331e-01
-4.55866829e-02 5.38406014e-01 -2.79485017e-01 4.85250175e-01
-1.07311960e-02 -4.42621887e-01 -3.06782544e-01 -5.59481561e-01
3.15511853e-01 6.87957168e-01 -4.87557799e-01 5.27970612e-01
2.07461044e-01 7.41667986e-01 -9.62807715e-01 -4.73475963e-01
8.55194986e-01 4.85620230e-01 -1.24587707e-01 1.72617525e-01
-2.14097500e-01 5.14659047e-01 -3.56618464e-01 4.43686932e-01
6.26195014e-01 4.58466470e-01 -2.27794752e-01 5.13007678e-02
-5.04324257e-01 3.14441502e-01 -1.21287608e+00 1.57092702e+00
-5.69892108e-01 1.20589983e+00 -2.53432065e-01 -1.02342403e+00
7.55319059e-01 2.75351346e-01 7.50828505e-01 -9.73674834e-01
2.87107173e-02 1.83287963e-01 1.06430791e-01 -9.40314174e-01
6.84428453e-01 1.99340627e-01 -2.64631778e-01 2.11874872e-01
-2.38291204e-01 -1.17640242e-01 7.57845163e-01 -1.11773252e-01
1.26471078e+00 2.17832640e-01 1.88076749e-01 1.10323511e-01
7.61067092e-01 3.82586345e-02 7.58653760e-01 6.59427643e-01
-8.29991460e-01 3.71692151e-01 3.25537682e-01 -7.85658896e-01
-9.33881462e-01 -8.16908956e-01 -6.42153695e-02 8.68311703e-01
1.38299525e-01 -2.62976050e-01 -6.30217612e-01 -3.36290509e-01
-9.99507234e-02 1.08851886e+00 -4.86546636e-01 -3.22872341e-01
-7.51485705e-01 -5.29328346e-01 4.71531093e-01 6.10229909e-01
1.02915537e+00 -1.52598834e+00 -1.44120157e+00 3.84783775e-01
-3.74427885e-01 -1.47626162e+00 -1.28177732e-01 1.08870164e-01
-4.07312542e-01 -1.10903776e+00 -1.15032390e-01 -1.24231033e-01
5.34476154e-02 5.17675340e-01 1.29277384e+00 -4.56710458e-01
-1.82586700e-01 4.78161186e-01 -3.88615102e-01 -7.23525465e-01
-3.68218690e-01 -2.43009701e-01 1.38371363e-01 2.27577627e-01
6.60462141e-01 -2.90059119e-01 -5.22494972e-01 3.60226959e-01
-8.19427609e-01 1.90789159e-02 6.71664059e-01 5.21961749e-01
5.88229537e-01 1.17453307e-01 4.68213409e-01 -1.61409855e-01
1.66000932e-01 -3.82972836e-01 -5.23628652e-01 1.86294299e-02
-1.79357249e-02 -2.22009823e-01 6.23949170e-01 -2.68545777e-01
-1.03457761e+00 3.97299469e-01 -3.42810243e-01 -3.89297128e-01
-6.43452466e-01 1.82608292e-01 -4.07032609e-01 1.45428017e-01
3.14512223e-01 2.81111717e-01 -1.13582715e-01 -7.00056627e-02
3.09960306e-01 7.08174109e-01 6.34496510e-01 -9.57814232e-02
4.83278543e-01 5.19999206e-01 1.45342633e-01 -9.59807575e-01
-8.84856880e-01 -5.93983591e-01 -7.67547846e-01 -8.34321976e-01
1.00035012e+00 -9.36518610e-01 -8.79732370e-01 1.00297296e+00
-1.10726798e+00 -6.73818707e-01 -5.17425478e-01 6.84747696e-01
-1.04041541e+00 1.19249932e-01 -2.59523034e-01 -8.96896482e-01
-5.01491688e-03 -1.36880159e+00 1.02516603e+00 1.19007239e-02
2.20337678e-02 -7.20349967e-01 1.46724815e-02 4.55369920e-01
5.52372396e-01 4.27277237e-01 2.10478753e-01 -3.87826800e-01
-5.51625550e-01 -3.33471030e-01 7.47081935e-02 5.35002887e-01
1.59709938e-02 9.78912041e-02 -1.17403364e+00 -1.56105846e-01
-6.32087979e-03 -4.04462188e-01 9.11729693e-01 5.96885502e-01
1.26379192e+00 7.01146647e-02 -1.33049279e-01 1.82213336e-01
1.03091061e+00 4.15541440e-01 1.01981270e+00 4.68351305e-01
5.40333331e-01 8.10927808e-01 1.28795385e+00 6.46894991e-01
6.55119538e-01 8.46435666e-01 8.72445941e-01 1.46081954e-01
-4.25072527e-03 1.78359345e-01 7.36455560e-01 4.57277507e-01
5.50526194e-02 -5.89904845e-01 -9.02003646e-01 7.88504541e-01
-1.97023892e+00 -1.32485330e+00 -3.73455822e-01 2.21512866e+00
2.33293369e-01 4.12402838e-01 1.73917696e-01 4.93553191e-01
5.00900865e-01 6.66322172e-01 -8.96910250e-01 -4.68621314e-01
-7.86350742e-02 -3.75306636e-01 7.81797767e-01 3.41070116e-01
-1.40213370e+00 7.81493306e-01 6.31527996e+00 6.33957982e-01
-1.21515250e+00 -5.98588772e-03 4.09996927e-01 -3.75387311e-01
9.43991467e-02 -5.22010326e-02 -9.61329401e-01 6.13982737e-01
1.46366084e+00 -7.94806853e-02 2.68232524e-01 8.24746549e-01
7.86759138e-01 -3.83833349e-01 -9.56207931e-01 9.90282834e-01
-3.59155089e-02 -1.00045538e+00 -4.76902366e-01 -1.72795266e-01
3.88936430e-01 4.26400214e-01 -5.04174903e-02 4.57032740e-01
2.05114305e-01 -7.99938560e-01 8.71336520e-01 6.14630401e-01
4.57020581e-01 -7.13925421e-01 6.59169376e-01 4.15756404e-01
-1.29235768e+00 -3.96769524e-01 -1.03633381e-01 -2.88310409e-01
6.21598840e-01 6.12705767e-01 -4.65996861e-01 4.52587515e-01
8.79670739e-01 1.05426514e+00 -5.53481817e-01 9.80838001e-01
-2.49711096e-01 3.91846269e-01 -2.43419647e-01 7.75734410e-02
6.03029132e-01 1.04535418e-02 5.98971307e-01 1.14583695e+00
4.24192458e-01 -1.24593168e-01 1.96208954e-01 1.34742811e-01
4.01536375e-01 -4.25495088e-01 -8.63538384e-01 6.90887719e-02
2.64941126e-01 1.35280228e+00 -2.16043666e-01 -4.34675783e-01
-5.77643633e-01 6.13753378e-01 -2.22254947e-01 2.64670342e-01
-1.24720836e+00 -3.20428461e-01 1.26346254e+00 5.96927889e-02
2.36853570e-01 -5.27661622e-01 1.11576803e-01 -6.46941245e-01
3.11175913e-01 -6.58511758e-01 1.83640450e-01 -9.66157138e-01
-7.27128446e-01 4.36600417e-01 2.65056729e-01 -1.64142764e+00
-6.38076246e-01 -5.14990866e-01 -6.62850440e-01 4.97675210e-01
-1.89201665e+00 -9.94143665e-01 -6.16819620e-01 4.88873035e-01
9.35029328e-01 -1.72173046e-02 3.79536957e-01 7.25710571e-01
-6.84016049e-01 4.47809808e-02 -1.41522139e-01 -3.61144334e-01
7.60721326e-01 -9.92567480e-01 6.62034333e-01 1.04038417e+00
-3.17396194e-01 -1.52648285e-01 1.19336474e+00 -2.00461254e-01
-1.63107359e+00 -1.46272051e+00 1.12727904e+00 -3.85172367e-01
6.79695010e-01 -3.29661638e-01 -6.76956177e-01 8.48545730e-01
3.51109475e-01 2.18341723e-01 2.43676752e-01 -4.57850754e-01
3.77107441e-01 -4.26230818e-01 -1.04973245e+00 5.07884860e-01
1.25212514e+00 -3.38584781e-01 -2.95776695e-01 4.68313277e-01
6.33166611e-01 -5.46614289e-01 -7.07688928e-01 5.54806352e-01
4.57370371e-01 -1.31156433e+00 9.29628491e-01 -4.55844432e-01
2.65255809e-01 -5.68537533e-01 -3.25219840e-01 -1.21993625e+00
-7.80225545e-02 -4.52397853e-01 -2.47669980e-01 9.15580928e-01
3.21066491e-02 -5.90285897e-01 5.38531005e-01 5.49794257e-01
-7.70003617e-01 -6.59488797e-01 -1.20432746e+00 -8.39428782e-01
-3.77278715e-01 -1.19628310e+00 7.57771909e-01 3.11469913e-01
-4.85705197e-01 -1.07765824e-01 -7.54966199e-01 5.33666126e-02
4.00766492e-01 -9.29520056e-02 9.95010793e-01 -9.51644897e-01
6.75417185e-02 -2.68398732e-01 -7.26815283e-01 -9.80441511e-01
3.94852489e-01 -2.48984501e-01 4.19023782e-01 -1.61244380e+00
-2.34662190e-01 -2.19942957e-01 -3.21066767e-01 5.88931680e-01
1.59724683e-01 6.09975271e-02 1.41803771e-01 -1.40348494e-01
-8.25928748e-01 6.32162273e-01 9.51682150e-01 -2.87830353e-01
-1.17793400e-02 3.92294340e-02 -8.70130435e-02 7.01010287e-01
8.62670243e-01 -5.77904209e-02 -6.84446692e-01 -4.63027388e-01
1.14393078e-01 -1.00778639e-02 6.32338226e-01 -1.57180858e+00
2.56127983e-01 -4.79025513e-01 -2.74877250e-02 -1.09523058e+00
7.18764305e-01 -7.95081377e-01 3.28177094e-01 5.78255773e-01
-3.75322223e-01 9.96720791e-02 4.71298486e-01 4.26262379e-01
-4.60537910e-01 1.15752947e-02 7.64032722e-01 5.97754680e-02
-1.65126991e+00 4.01950568e-01 -6.44147336e-01 2.82300133e-02
1.50517118e+00 -3.18037003e-01 -3.30431670e-01 -3.41869235e-01
-5.18236101e-01 6.75209939e-01 6.90474808e-02 8.99851024e-01
5.52738369e-01 -1.45716214e+00 -8.97684753e-01 2.21609190e-01
3.68940711e-01 -1.27439005e-02 8.84474158e-01 9.99163866e-01
-2.67586280e-02 6.61215067e-01 -4.78440821e-01 -7.23605275e-01
-1.16083026e+00 6.42186344e-01 2.95630783e-01 -1.30323470e-01
-6.64332092e-01 4.20222580e-01 -5.41686527e-02 -2.16759607e-01
6.27876669e-02 -3.75999719e-01 -4.05435443e-01 2.51352906e-01
7.41393268e-01 5.25258660e-01 2.32436493e-01 -1.00366175e+00
-4.21686053e-01 3.12124699e-01 2.12387785e-01 1.04449339e-01
1.10000539e+00 -2.88214415e-01 3.51270944e-01 6.33988678e-01
8.66677582e-01 -4.95924503e-01 -1.81464171e+00 -9.58581548e-03
-1.51889861e-01 -5.99683821e-01 -2.49905586e-02 -4.98798668e-01
-1.17453456e+00 9.60506380e-01 8.41482460e-01 1.90532237e-01
1.36603415e+00 -2.60365486e-01 1.07095969e+00 4.96629089e-01
4.40954715e-01 -1.35242128e+00 -1.78290918e-01 7.52480805e-01
8.72720957e-01 -1.51076078e+00 -2.92616040e-01 1.10986836e-01
-8.28828692e-01 7.33627379e-01 5.74632108e-01 2.64468461e-01
2.67954499e-01 2.12807849e-01 7.42809772e-02 2.12013423e-02
-1.07469475e+00 -5.53597391e-01 -1.00113742e-01 6.18083596e-01
8.04813951e-02 1.99222788e-01 -2.04430018e-02 -9.88961533e-02
-9.63222831e-02 1.12486087e-01 5.72668433e-01 9.15172815e-01
-4.51953083e-01 -8.77852738e-01 -1.38755038e-01 3.58082682e-01
4.68516424e-02 2.17497200e-01 3.78524698e-02 8.05454612e-01
4.10019279e-01 1.25123286e+00 2.50005096e-01 -5.59805036e-01
5.89774847e-01 -1.09187374e-02 1.45181268e-01 5.94677520e-04
-1.26428083e-01 -5.20287454e-01 4.54128742e-01 -9.29100454e-01
-8.10480118e-01 -1.27445734e+00 -1.19391334e+00 -4.59572107e-01
3.57710421e-01 -4.24906164e-01 7.30990171e-01 1.19663119e+00
4.75819200e-01 7.50081122e-01 6.92192197e-01 -1.24080122e+00
-1.84791386e-01 -8.93553734e-01 -3.83014947e-01 3.07487547e-01
5.74451625e-01 -9.42559004e-01 -3.74262422e-01 1.07147411e-01]
|
[7.242404460906982, 0.29085931181907654]
|
08842831-27bc-4e15-a2d0-507d758ac671
|
deep-neural-networks-for-hdr-imaging
|
1611.00591
| null |
http://arxiv.org/abs/1611.00591v1
|
http://arxiv.org/pdf/1611.00591v1.pdf
|
Deep Neural Networks for HDR imaging
|
We propose novel methods of solving two tasks using Convolutional Neural
Networks, firstly the task of generating HDR map of a static scene using
differently exposed LDR images of the scene captured using conventional cameras
and secondly the task of finding an optimal tone mapping operator that would
give a better score on the TMQI metric compared to the existing methods. We
quantitatively show the performance of our networks and illustrate the cases
where our networks performs good as well as bad.
|
['Kshiteej Sheth']
|
2016-09-04
| null | null | null | null |
['tone-mapping']
|
['computer-vision']
|
[ 4.78739798e-01 9.61100608e-02 3.90526146e-01 -3.04582387e-01
-6.39603615e-01 -3.34262669e-01 4.87734824e-01 -7.89960623e-01
-4.11660314e-01 8.65132391e-01 1.47511393e-01 -2.25907013e-01
1.02694407e-01 -7.46106863e-01 -7.80403137e-01 -5.33226073e-01
-3.08488924e-02 3.65893811e-01 2.16825679e-01 -4.95489597e-01
3.73313516e-01 5.70416331e-01 -1.63559330e+00 6.73724562e-02
5.36280930e-01 1.18694186e+00 2.95594990e-01 8.95214677e-01
4.19360191e-01 1.32060468e+00 -8.56812716e-01 -4.68457073e-01
7.02739358e-01 -5.57444572e-01 -1.23459613e+00 1.99401811e-01
6.35952652e-01 -6.72208130e-01 -8.17990303e-01 1.22118878e+00
5.39722443e-01 2.21341953e-01 4.18487728e-01 -9.07965064e-01
-1.10171270e+00 2.35422298e-01 -4.63166356e-01 5.10825157e-01
5.80391943e-01 1.33050650e-01 7.87828743e-01 -7.49153376e-01
6.97647929e-01 1.08967674e+00 4.00357306e-01 3.51791680e-01
-1.09570062e+00 -3.53477597e-01 -6.69548988e-01 4.22861576e-01
-1.54374707e+00 -5.51990569e-01 6.81264222e-01 -6.24249503e-02
1.06994355e+00 3.11508149e-01 1.44254878e-01 6.06253326e-01
2.10577622e-01 2.80929923e-01 1.25571668e+00 -4.77487087e-01
-1.37914255e-01 -1.33319885e-01 -3.84084493e-01 6.11561477e-01
-2.86349833e-01 1.58662587e-01 -4.32718813e-01 2.74509609e-01
1.06704497e+00 -5.71017861e-01 -4.36840594e-01 -1.04473773e-02
-1.44607794e+00 6.90220177e-01 4.62857693e-01 3.14648867e-01
-1.19326398e-01 2.74677068e-01 2.87182815e-02 8.07894886e-01
4.44381386e-01 7.62175083e-01 -1.09643303e-01 1.99230835e-01
-8.15798283e-01 -1.13583639e-01 3.47410202e-01 8.23805928e-01
8.90112638e-01 2.36296639e-01 -1.75463989e-01 9.14327502e-01
-1.83511391e-01 4.15760547e-01 1.34567067e-01 -1.57670522e+00
2.60776073e-01 -9.64063555e-02 2.74982572e-01 -7.47339368e-01
-1.94669306e-01 9.33148935e-02 -7.88022697e-01 5.02348900e-01
3.89064908e-01 -2.14214414e-01 -1.05127656e+00 1.61709082e+00
-2.23351032e-01 1.36279792e-01 3.14900428e-01 1.09440053e+00
7.19250143e-01 1.02426028e+00 -4.10919160e-01 -2.86083013e-01
9.19435740e-01 -8.05758476e-01 -9.04520571e-01 -1.21174015e-01
3.05176288e-01 -1.00293303e+00 9.82536316e-01 4.99844223e-01
-1.53365684e+00 -8.23876679e-01 -1.40921199e+00 -4.57660496e-01
-1.33683816e-01 2.60796756e-01 4.01391178e-01 6.01267397e-01
-1.86271262e+00 6.87842369e-01 7.10602254e-02 -2.54126012e-01
3.02188396e-01 4.47883427e-01 -4.61994171e-01 -2.16727674e-01
-1.51814437e+00 1.27408767e+00 4.09387380e-01 2.60423750e-01
-1.01172805e+00 -5.42495131e-01 -7.59869218e-01 6.74350336e-02
1.49327144e-01 -2.91262746e-01 1.19680142e+00 -1.17373574e+00
-1.88613367e+00 1.20770228e+00 1.78095877e-01 -3.69879663e-01
6.26991093e-01 -1.94114521e-01 -4.60322827e-01 3.12897652e-01
-7.46459663e-02 8.65316093e-01 6.03950083e-01 -1.04493082e+00
-6.34563088e-01 9.43119451e-02 4.76558506e-01 4.50550348e-01
7.89825544e-02 2.22945362e-01 -4.20843959e-01 -4.86206591e-01
-6.20421618e-02 -8.73547435e-01 -1.40776977e-01 2.57295296e-02
-6.09073818e-01 7.44142523e-03 8.87933373e-01 -7.80794382e-01
7.65239894e-01 -2.12484360e+00 1.25847206e-01 4.38064225e-02
1.43109232e-01 2.50000507e-01 -1.51748762e-01 1.15419284e-01
-4.67934757e-01 1.58981793e-02 -1.51122566e-02 1.78663254e-01
-1.64535865e-01 -1.76530808e-01 -4.45453316e-01 6.46969140e-01
2.09941223e-01 7.46479034e-01 -6.25746608e-01 -3.52957547e-01
5.45344234e-01 3.90312016e-01 -4.73934598e-02 5.91995656e-01
-1.35070533e-01 3.81186843e-01 2.37489656e-01 2.19372347e-01
6.87045872e-01 -2.40160331e-01 1.69360727e-01 -4.45324868e-01
-8.28896090e-02 5.60147241e-02 -1.09238005e+00 1.37955749e+00
-4.12479699e-01 1.44483745e+00 -4.06372458e-01 -7.81388342e-01
1.09944344e+00 2.76619732e-01 4.61749405e-01 -1.31182051e+00
5.53191192e-02 1.46876901e-01 -2.38245264e-01 -4.01900232e-01
8.20049644e-01 -3.33999366e-01 3.17684142e-03 6.77388012e-01
8.95765647e-02 -1.98912874e-01 -5.17145731e-02 -9.64856124e-04
1.16029000e+00 -3.55201177e-02 1.34335473e-01 -3.62326652e-01
5.41576385e-01 -2.45539814e-01 1.43104598e-01 7.19747007e-01
-3.11382920e-01 1.04721642e+00 4.66934294e-01 -9.44513440e-01
-1.48771632e+00 -1.18387055e+00 -3.89160439e-02 9.03742790e-01
5.78636646e-01 2.18659848e-01 -7.20291555e-01 -2.02857286e-01
-6.38756275e-01 6.35157049e-01 -6.94094539e-01 -5.05290292e-02
-8.20067346e-01 -8.11408937e-01 7.03691423e-01 3.50928783e-01
1.05816519e+00 -1.25907350e+00 -6.35632038e-01 -1.91134959e-01
-5.69390178e-01 -1.36838770e+00 -4.81796682e-01 2.60440588e-01
-3.21727455e-01 -8.64164889e-01 -9.55891609e-01 -9.26725864e-01
4.81539607e-01 5.49072027e-01 1.39309084e+00 1.80275515e-01
-2.59629190e-01 -1.40357882e-01 -1.38925701e-01 6.25329092e-02
-5.04171193e-01 -2.63412148e-01 -4.80032638e-02 -9.08690467e-02
1.51003614e-01 -2.92453438e-01 -5.45224130e-01 5.19874811e-01
-1.29169595e+00 2.85669446e-01 3.42054784e-01 4.07045454e-01
3.90973032e-01 4.98689055e-01 3.46289515e-01 -6.60869062e-01
4.68617082e-01 -6.39860332e-02 -7.39337325e-01 4.42336649e-01
-2.97151834e-01 9.82574150e-02 8.14688861e-01 -1.01627246e-01
-1.17108262e+00 1.45883858e-01 -1.33145437e-01 -4.87825871e-01
-1.45168036e-01 -2.17629328e-01 -1.31622523e-01 -3.73261601e-01
8.59599471e-01 2.08861262e-01 -3.69873673e-01 1.27132414e-02
4.31873232e-01 5.30190110e-01 8.69838476e-01 -3.32372159e-01
9.65009332e-01 5.60011208e-01 3.03002018e-02 -7.79416740e-01
-7.61676729e-01 -7.35421628e-02 -6.98899746e-01 -4.26629126e-01
1.35828269e+00 -6.97914660e-01 -3.66782635e-01 8.29207540e-01
-1.23111641e+00 -6.63594484e-01 -2.46917918e-01 1.28113821e-01
-1.09405458e+00 2.42728636e-01 -7.66155064e-01 -3.62173170e-01
-4.98655327e-02 -1.32475591e+00 1.09323287e+00 2.79191107e-01
2.20745072e-01 -8.38689685e-01 2.76194036e-01 2.35522211e-01
4.66861844e-01 1.48317337e-01 1.07635868e+00 -2.16730703e-02
-8.64851713e-01 3.58866334e-01 -7.00707138e-01 3.36144090e-01
1.27214819e-01 -9.70196798e-02 -1.30765212e+00 -2.16129810e-01
-5.85298650e-02 -6.83326483e-01 6.83548570e-01 5.65080583e-01
1.31644106e+00 -6.10452592e-02 1.38362169e-01 9.86963391e-01
1.96410584e+00 5.44310212e-01 1.65618825e+00 4.76310551e-01
5.83449125e-01 4.75898653e-01 4.65905845e-01 -1.19941798e-03
6.40777275e-02 1.08474684e+00 2.39009872e-01 -6.22193933e-01
-2.83223003e-01 1.80319443e-01 3.51336926e-01 4.09140348e-01
-5.25133573e-02 -5.79200983e-01 -6.96662843e-01 4.14021224e-01
-1.43804860e+00 -9.74829435e-01 1.34139210e-01 2.01580572e+00
5.77544630e-01 2.66961735e-02 5.15024923e-02 1.18429340e-01
9.27801251e-01 4.53570992e-01 -2.81991243e-01 -8.26604187e-01
-4.39803302e-01 3.91306400e-01 5.86729467e-01 4.13331270e-01
-1.20734918e+00 9.47363377e-01 8.58560085e+00 5.55584729e-01
-1.04622817e+00 -9.27779302e-02 1.09235001e+00 6.64937273e-02
-1.06478348e-01 -3.38412523e-01 -2.40443036e-01 2.87474185e-01
1.06626797e+00 -6.75615296e-02 8.21617663e-01 3.51582885e-01
1.04934916e-01 -2.12734759e-01 -1.17007971e+00 1.14851964e+00
3.41903001e-01 -1.31754053e+00 1.13846578e-01 4.71792556e-02
1.01391816e+00 8.68775770e-02 3.98920268e-01 -1.36768252e-01
6.23382449e-01 -1.33146060e+00 7.13974357e-01 6.23147666e-01
1.31052792e+00 -7.59532869e-01 6.92009330e-01 -3.25792938e-01
-7.98439443e-01 8.55644196e-02 -6.66425407e-01 2.57138878e-01
2.37258703e-01 4.60457355e-01 -5.46705246e-01 3.56160045e-01
8.48677158e-01 3.96987706e-01 -6.50255740e-01 8.79009783e-01
4.76949476e-02 9.29605812e-02 2.60925237e-02 6.20591283e-01
2.40151390e-01 -1.25534400e-01 2.92875171e-01 1.04559493e+00
4.87192661e-01 1.75458014e-01 -3.06674838e-01 8.82898450e-01
-3.92236531e-01 -3.18238735e-01 -1.05221641e+00 2.75676638e-01
8.60975012e-02 1.02887297e+00 -8.81850541e-01 -1.79404601e-01
-4.26086843e-01 1.39291370e+00 2.27549851e-01 5.22202015e-01
-8.82277191e-01 -7.18347847e-01 4.43889886e-01 -1.52537629e-01
4.27664310e-01 -8.44464004e-02 1.54782813e-02 -1.06421494e+00
-1.53056443e-01 -9.54955995e-01 1.99400052e-01 -1.64954460e+00
-1.07739449e+00 9.57082033e-01 -2.36384749e-01 -1.26234746e+00
-1.93988949e-01 -6.35834932e-01 -7.48145640e-01 8.56260240e-01
-1.63552809e+00 -6.36526942e-01 -4.96557623e-01 7.15618074e-01
5.85017622e-01 -2.64890075e-01 5.27102828e-01 4.27995563e-01
-3.64134908e-01 3.62429202e-01 9.68255326e-02 1.75146788e-01
7.18666077e-01 -1.28777742e+00 6.74378037e-01 1.05655837e+00
-4.07123640e-02 -8.24955553e-02 7.84258544e-01 -1.58643931e-01
-8.60850751e-01 -1.00797570e+00 6.13007545e-01 -3.36667508e-01
2.87559927e-01 -1.19468689e-01 -7.05044985e-01 6.04642987e-01
4.79470521e-01 -2.33807608e-01 3.29139829e-01 -4.52843726e-01
-1.90322414e-01 -2.31022224e-01 -1.23901939e+00 5.96421123e-01
9.92131889e-01 -8.15292358e-01 -3.24769288e-01 3.93203467e-01
8.75218451e-01 -4.68864173e-01 -8.18624735e-01 9.60754603e-02
3.41383338e-01 -1.40853429e+00 1.13087654e+00 -2.24916518e-01
9.40109432e-01 -4.51225281e-01 -4.57874030e-01 -1.45720863e+00
-4.81799155e-01 -6.08096004e-01 3.42333674e-01 6.56677008e-01
3.40246171e-01 -2.90961444e-01 6.41522646e-01 2.81461447e-01
4.63113822e-02 -3.59481305e-01 -9.52460170e-01 -6.01014555e-01
-4.04464118e-02 -1.37318715e-01 6.03416681e-01 9.46916699e-01
-5.75898111e-01 4.29855853e-01 -8.36165190e-01 -2.96642166e-02
5.90085506e-01 -1.00038372e-01 5.27525723e-01 -9.47762549e-01
-4.31433320e-02 -2.57994384e-01 -6.32703424e-01 -8.46006632e-01
2.26200372e-01 -4.99966890e-01 4.09146965e-01 -1.37555265e+00
3.19042504e-01 -2.80768514e-01 -1.48717865e-01 6.26405925e-02
-1.46227144e-02 9.56177771e-01 2.23481819e-01 2.93379277e-01
-8.88553858e-01 2.06664577e-01 1.31554258e+00 -6.50432780e-02
1.53504694e-02 -5.14337778e-01 -7.44288325e-01 3.27439100e-01
7.33288884e-01 -6.47067353e-02 -5.85315049e-01 -9.32511926e-01
4.85919684e-01 3.36392879e-01 3.69490296e-01 -1.36445248e+00
-2.34953826e-03 -1.10208742e-01 7.35792875e-01 -3.81620318e-01
3.70533764e-01 -7.92402804e-01 4.79741514e-01 1.36623055e-01
-5.74842393e-01 3.22215706e-01 6.64175227e-02 1.96169242e-01
-1.74285337e-01 -1.66797549e-01 1.26888895e+00 -1.60530299e-01
-9.93150592e-01 1.72439307e-01 -4.40837324e-01 -7.00090080e-02
9.58562553e-01 -3.57188672e-01 -6.77091956e-01 -8.24344218e-01
-5.64724922e-01 -2.11963609e-01 6.35082722e-01 4.29224312e-01
7.74454951e-01 -1.55188096e+00 -6.51234329e-01 2.80667335e-01
-8.90672654e-02 -3.57067436e-01 1.19242124e-01 3.40851575e-01
-1.25490117e+00 5.14580250e-01 -7.95770526e-01 -3.42321396e-01
-9.36774850e-01 5.29468894e-01 7.96007633e-01 -2.56301075e-01
-5.90520322e-01 5.27972460e-01 4.35556531e-01 -1.85600877e-01
7.39433467e-02 1.12326615e-01 -3.95785183e-01 -5.27817726e-01
6.30920470e-01 4.76598054e-01 1.60980761e-01 -1.01638222e+00
-6.53537214e-02 6.45391643e-01 1.88675076e-01 -3.06157440e-01
1.34782529e+00 -4.84071344e-01 -2.48212680e-01 2.54744649e-01
1.37789774e+00 -6.08706892e-01 -1.28893185e+00 -2.51936078e-01
-3.15605044e-01 -7.22077787e-01 3.80875468e-01 -8.40962052e-01
-1.41758204e+00 7.78596342e-01 1.15451181e+00 3.73935580e-01
1.51934206e+00 -9.31077451e-02 9.73772109e-01 5.47956109e-01
4.85243022e-01 -1.25949180e+00 4.44197327e-01 4.42255735e-01
6.66098416e-01 -1.20816433e+00 -3.67648825e-02 -1.68299317e-01
-7.64168799e-01 1.10570931e+00 6.45256698e-01 -3.99457544e-01
4.37698960e-01 1.54019624e-01 4.29352224e-01 -3.22559983e-01
-6.49566472e-01 -2.51114428e-01 2.04198137e-01 8.60010982e-01
2.27618381e-01 -3.10019642e-01 2.31504574e-01 -5.04852116e-01
-1.88406184e-01 1.49756268e-01 9.95065451e-01 4.80019927e-01
-4.66323167e-01 -5.98291576e-01 -3.32746178e-01 3.19992691e-01
-7.11777747e-01 -2.90649887e-02 -4.77752268e-01 6.13318443e-01
1.79369114e-02 7.94019938e-01 4.03751761e-01 -6.28820360e-01
2.88380742e-01 -2.51445293e-01 6.33202314e-01 -2.49612212e-01
-2.10677207e-01 -2.49380752e-01 1.41032383e-01 -6.55402303e-01
-7.12307394e-01 -1.07828133e-01 -7.43353605e-01 -7.20095813e-01
-7.77984485e-02 -3.73228610e-01 2.45631576e-01 9.80148494e-01
-1.17251515e-01 4.66314316e-01 1.16545415e+00 -8.77408922e-01
1.13616390e-02 -6.70006752e-01 -7.76253581e-01 4.93265361e-01
5.87286115e-01 -3.61427337e-01 -3.50617319e-01 3.81457061e-01]
|
[10.970787048339844, -2.287334680557251]
|
72e06319-3338-4f42-8e92-5fa4eceba345
|
ucphrase-unsupervised-context-aware-quality
|
2105.14078
| null |
https://arxiv.org/abs/2105.14078v1
|
https://arxiv.org/pdf/2105.14078v1.pdf
|
UCPhrase: Unsupervised Context-aware Quality Phrase Tagging
|
Identifying and understanding quality phrases from context is a fundamental task in text mining. The most challenging part of this task arguably lies in uncommon, emerging, and domain-specific phrases. The infrequent nature of these phrases significantly hurts the performance of phrase mining methods that rely on sufficient phrase occurrences in the input corpus. Context-aware tagging models, though not restricted by frequency, heavily rely on domain experts for either massive sentence-level gold labels or handcrafted gazetteers. In this work, we propose UCPhrase, a novel unsupervised context-aware quality phrase tagger. Specifically, we induce high-quality phrase spans as silver labels from consistently co-occurring word sequences within each document. Compared with typical context-agnostic distant supervision based on existing knowledge bases (KBs), our silver labels root deeply in the input domain and context, thus having unique advantages in preserving contextual completeness and capturing emerging, out-of-KB phrases. Training a conventional neural tagger based on silver labels usually faces the risk of overfitting phrase surface names. Alternatively, we observe that the contextualized attention maps generated from a transformer-based neural language model effectively reveal the connections between words in a surface-agnostic way. Therefore, we pair such attention maps with the silver labels to train a lightweight span prediction model, which can be applied to new input to recognize (unseen) quality phrases regardless of their surface names or frequency. Thorough experiments on various tasks and datasets, including corpus-level phrase ranking, document-level keyphrase extraction, and sentence-level phrase tagging, demonstrate the superiority of our design over state-of-the-art pre-trained, unsupervised, and distantly supervised methods.
|
['Jingbo Shang', 'Jiawei Han', 'Liyuan Liu', 'Yu Meng', 'Zhenyu Bi', 'Zihan Wang', 'Xiaotao Gu']
|
2021-05-28
| null | null | null | null |
['phrase-tagging', 'phrase-ranking']
|
['natural-language-processing', 'natural-language-processing']
|
[ 3.92062694e-01 -9.36311930e-02 -5.10315478e-01 -3.31784874e-01
-1.20392871e+00 -8.99838567e-01 3.03410888e-01 4.55704898e-01
-5.26479244e-01 7.92647779e-01 4.45050180e-01 -5.10306418e-01
-1.50308579e-01 -8.25505614e-01 -7.43266106e-01 -5.78463972e-01
1.20838419e-01 4.62607175e-01 9.11648497e-02 -3.10363531e-01
3.42184722e-01 -9.60899368e-02 -1.37444842e+00 4.20459390e-01
1.05504835e+00 1.00340378e+00 3.99987340e-01 3.62245023e-01
-6.44240916e-01 3.50026608e-01 -3.86681437e-01 -5.47455311e-01
1.35971867e-02 -1.43975690e-01 -9.23331141e-01 -1.10859752e-01
5.76536119e-01 2.32250214e-01 -8.43924284e-03 1.14580381e+00
1.11310340e-01 -5.77064566e-02 4.82085556e-01 -6.62552297e-01
-8.99045169e-01 1.18550456e+00 -5.50727189e-01 5.59325039e-01
2.97864884e-01 -2.60320455e-02 2.05636716e+00 -1.26179290e+00
3.93851846e-01 7.18596458e-01 7.17023313e-01 4.35533643e-01
-1.27207625e+00 -7.71207392e-01 3.82950187e-01 -7.88429901e-02
-1.35475075e+00 -2.28360370e-01 8.55733752e-01 -4.06542838e-01
1.25200927e+00 1.64342925e-01 5.12911797e-01 1.37069452e+00
1.35337878e-02 7.04546928e-01 8.60101044e-01 -8.02807987e-01
-1.34920618e-02 1.31471962e-01 4.65298265e-01 7.56291926e-01
3.55445981e-01 -1.15329608e-01 -8.37730587e-01 -2.35962972e-01
2.59310067e-01 5.88469356e-02 -3.97936791e-01 -6.65107071e-02
-1.16885674e+00 7.65905857e-01 2.25846156e-01 6.32773101e-01
-3.62121195e-01 -8.85583758e-02 3.63253206e-01 1.35688320e-01
5.96184909e-01 8.50051522e-01 -9.94106829e-01 -2.06378087e-01
-1.15370393e+00 5.27392440e-02 6.70187950e-01 1.04554594e+00
9.97901917e-01 -3.35499704e-01 -4.71699208e-01 7.78214812e-01
8.01411271e-02 5.16865432e-01 6.13719046e-01 -2.48391882e-01
7.24718690e-01 7.75478542e-01 -1.45542873e-02 -8.30490828e-01
-3.08639467e-01 -8.41007650e-01 -4.85206336e-01 -6.39782488e-01
2.19966158e-01 6.63305223e-02 -9.83415663e-01 1.97382903e+00
-1.55509831e-02 -1.98363466e-03 -8.54338855e-02 6.03353918e-01
5.12127459e-01 6.01221502e-01 2.93256313e-01 -2.96377718e-01
1.77387202e+00 -8.60589564e-01 -5.32322705e-01 -4.65228498e-01
6.75844610e-01 -5.11056066e-01 1.77295065e+00 1.92026690e-01
-7.20887601e-01 -3.76161397e-01 -8.58948469e-01 -4.68985029e-02
-4.35245335e-01 -4.82759215e-02 4.63160217e-01 4.95158345e-01
-6.36684477e-01 6.30145609e-01 -4.70229656e-01 -1.37626126e-01
3.22365373e-01 1.49378493e-01 -1.73111051e-01 8.35472643e-02
-1.53842914e+00 5.42870641e-01 5.90137064e-01 -2.12116152e-01
-5.32943904e-01 -9.50907528e-01 -8.04353952e-01 3.43813747e-01
7.50643194e-01 -6.74728215e-01 1.23029494e+00 -8.00265491e-01
-1.01284349e+00 9.60167170e-01 -4.18172568e-01 -4.38241988e-01
-3.00143063e-01 -5.10844409e-01 -5.64708710e-01 7.33458623e-02
5.33381224e-01 1.97283894e-01 9.29674029e-01 -8.63300145e-01
-9.64209616e-01 -1.80823490e-01 -1.14003131e-02 8.39060396e-02
-8.43210995e-01 -8.27604309e-02 -3.99044573e-01 -7.38551795e-01
-1.17650419e-01 -7.09699690e-01 -1.26650268e-02 -6.27177417e-01
-5.58253527e-01 -5.27010977e-01 4.24360812e-01 -4.42826688e-01
1.74787259e+00 -2.03808212e+00 -1.07127711e-01 2.39793599e-01
1.63828433e-01 6.55523092e-02 -1.78946540e-01 5.12011111e-01
9.60922018e-02 3.82724792e-01 -2.52108783e-01 -2.56495148e-01
1.66534886e-01 2.42157564e-01 -7.47591972e-01 -7.34414533e-02
5.32925069e-01 1.02461421e+00 -1.29242420e+00 -5.93964040e-01
-4.57937062e-01 -5.02971821e-02 -4.40971375e-01 1.19240627e-01
-5.82581818e-01 3.21834862e-01 -4.39002365e-01 7.44563818e-01
-1.39602192e-03 -5.91077685e-01 7.45848566e-02 -1.63502321e-01
6.29273206e-02 1.09645545e+00 -6.61750197e-01 1.91493046e+00
-8.57023358e-01 2.76612431e-01 -3.98856938e-01 -8.29024017e-01
8.41382563e-01 4.74868983e-01 2.17620984e-01 -7.11670518e-01
-1.36648118e-01 4.79613185e-01 -1.57886356e-01 -3.98923576e-01
7.33503580e-01 -4.56762344e-01 -5.04443586e-01 4.63147223e-01
3.82595986e-01 2.97146469e-01 4.00700629e-01 2.79012650e-01
1.51851773e+00 -4.36823741e-02 4.20543134e-01 -2.76628703e-01
4.87618744e-01 7.43913203e-02 8.52727294e-01 8.39097023e-01
8.87164101e-02 5.84485769e-01 2.07699925e-01 -2.61960477e-01
-1.07400846e+00 -9.84240890e-01 -3.31125915e-01 1.56382644e+00
-7.64130568e-03 -9.28056955e-01 -4.49381709e-01 -1.19667673e+00
-9.66635719e-03 5.64046860e-01 -4.70762879e-01 -1.14975244e-01
-6.46887779e-01 -5.10256588e-01 6.13194585e-01 6.66833937e-01
-4.93599474e-02 -1.31247485e+00 -1.32066041e-01 7.00281084e-01
-4.30097222e-01 -1.26886535e+00 -7.65357077e-01 7.04572260e-01
-6.79210722e-01 -8.48829925e-01 -3.93623173e-01 -7.86872089e-01
4.73499030e-01 2.21243843e-01 1.72609699e+00 4.55941893e-02
7.24809989e-02 9.88960266e-02 -6.80364192e-01 -2.95359224e-01
-4.19545770e-02 5.43891490e-01 2.54741937e-01 -1.75764412e-01
1.05215847e+00 -8.57046366e-01 -4.77511644e-01 3.48886102e-02
-7.79994905e-01 -2.40863577e-01 9.19939816e-01 1.24377084e+00
7.40992010e-01 2.23394573e-01 7.58607566e-01 -1.17751348e+00
7.01652169e-01 -3.82118583e-01 -4.04073834e-01 1.94486886e-01
-8.40602279e-01 4.30764139e-01 8.23527515e-01 -5.69831312e-01
-7.14627087e-01 -1.67334363e-01 -1.74212784e-01 -1.93802282e-01
-1.17322892e-01 7.46013463e-01 -3.04526627e-01 4.25515741e-01
8.07565033e-01 3.59889239e-01 -7.44085073e-01 -6.60211086e-01
4.55168784e-01 6.44667447e-01 5.96290648e-01 -9.58695590e-01
1.03885972e+00 1.55102924e-01 -4.86501396e-01 -4.59704518e-01
-1.74850857e+00 -9.55745101e-01 -5.74327409e-01 3.69820029e-01
6.47503972e-01 -8.53329122e-01 -2.50568360e-01 -1.36687294e-01
-1.08330524e+00 1.21553376e-01 -4.64156836e-01 9.28868502e-02
-2.30994686e-01 3.04862380e-01 -5.75066149e-01 -6.93709433e-01
-8.36234689e-01 -6.25620484e-01 1.27057540e+00 7.43829384e-02
-5.87943196e-01 -9.38185751e-01 3.17022920e-01 3.63496363e-01
1.89993098e-01 -3.33765775e-01 1.14994156e+00 -1.01432550e+00
-3.42550755e-01 -1.87006772e-01 -7.75338858e-02 2.89382666e-01
3.27166349e-01 -4.19864029e-01 -9.62026536e-01 -8.74629989e-02
-3.00011933e-01 -3.56793970e-01 1.06893170e+00 -1.14960745e-01
1.08177328e+00 -6.78408504e-01 -3.83373082e-01 3.60969454e-01
1.41267622e+00 -1.39560640e-01 1.45796593e-02 5.08697569e-01
8.54145408e-01 5.48907459e-01 6.49962783e-01 2.16352701e-01
1.17927894e-01 5.28713286e-01 2.98502739e-03 1.31334484e-01
2.25265548e-01 -8.17892671e-01 3.79117399e-01 1.12452257e+00
1.40577555e-01 -2.37818986e-01 -1.03571105e+00 9.84127998e-01
-1.65901291e+00 -1.01555347e+00 1.43465891e-01 1.90823829e+00
1.55339003e+00 7.30094135e-01 2.11484823e-02 3.32700729e-01
5.96033216e-01 9.20313671e-02 -2.38066480e-01 -2.63177037e-01
3.62773202e-02 6.57239258e-01 3.97145152e-01 1.12641238e-01
-1.18054450e+00 1.13545144e+00 4.87693977e+00 1.14204621e+00
-9.30761456e-01 2.46152744e-01 2.27983162e-01 -4.41557318e-02
-7.30386913e-01 1.55012026e-01 -1.34788656e+00 7.07193434e-01
1.03575623e+00 -1.04963236e-01 1.73452318e-01 1.02974403e+00
-4.58277725e-02 3.07124794e-01 -1.30401635e+00 8.03998530e-01
-3.90818380e-02 -1.24293053e+00 1.23088971e-01 9.45315287e-02
6.24098063e-01 5.17498441e-02 1.51911318e-01 6.09338820e-01
3.63547355e-01 -8.40105891e-01 8.05214882e-01 2.36344799e-01
8.96016061e-01 -7.03212321e-01 7.58235395e-01 4.20636207e-01
-1.27915740e+00 -5.49472980e-02 -4.06551540e-01 1.45187294e-02
1.84863552e-01 1.19115293e+00 -8.77133369e-01 4.76728529e-01
7.89352655e-01 6.37630522e-01 -3.09826702e-01 5.91237366e-01
-6.04365766e-01 1.15123069e+00 -2.60398209e-01 -1.45822227e-01
5.11074960e-01 1.80366233e-01 6.39559031e-01 1.64037931e+00
2.39730790e-01 -6.45218715e-02 1.94386125e-01 1.09899008e+00
-4.02229100e-01 2.73037851e-01 -5.93536139e-01 -4.08241183e-01
6.50149524e-01 1.42646873e+00 -4.82681632e-01 -3.11679095e-01
-6.54354692e-01 7.50739515e-01 6.92093849e-01 2.02016830e-01
-4.41874176e-01 -4.21193957e-01 7.28703022e-01 1.44446149e-01
6.60340846e-01 -1.61046937e-01 -4.66505349e-01 -1.33880603e+00
4.51028943e-01 -7.71877050e-01 4.51406777e-01 -2.17009738e-01
-1.87013209e+00 6.99784696e-01 -3.49457562e-01 -1.33955681e+00
-1.64013207e-01 -5.35932362e-01 -5.72721004e-01 9.36094761e-01
-1.84329438e+00 -1.13317287e+00 8.65316316e-02 3.42816055e-01
6.42071187e-01 -4.51215655e-02 9.66428459e-01 1.60051152e-01
-3.70566696e-01 8.30875754e-01 -8.44922885e-02 3.66303980e-01
7.47873068e-01 -1.63333976e+00 5.08838654e-01 8.60491276e-01
9.03025687e-01 1.02606034e+00 5.71097732e-01 -6.42453194e-01
-1.08597577e+00 -1.27494121e+00 1.46708453e+00 -8.28421831e-01
1.03854847e+00 -6.36750937e-01 -1.28325474e+00 5.03406525e-01
-1.21469565e-01 2.30110642e-02 1.02895343e+00 9.01715398e-01
-8.63315165e-01 -1.27765939e-01 -4.67404425e-01 4.75002289e-01
1.28862286e+00 -9.09435630e-01 -1.27768135e+00 2.15663284e-01
1.20612645e+00 -2.28989404e-02 -5.91260076e-01 4.68098760e-01
3.77252102e-01 -4.49842244e-01 7.79220939e-01 -6.61911070e-01
5.23043156e-01 -2.51458913e-01 -1.03431091e-01 -1.17928350e+00
-5.02398133e-01 -5.20688236e-01 -4.14407462e-01 1.52389443e+00
1.00590599e+00 -1.49795204e-01 8.80240679e-01 2.94479638e-01
-2.51722306e-01 -8.10633719e-01 -7.93370426e-01 -9.44189847e-01
1.32745594e-01 -6.41481817e-01 6.40572131e-01 8.93052936e-01
4.34271485e-01 9.33831513e-01 -2.82001704e-01 2.37872899e-01
3.52892160e-01 4.24715489e-01 2.73888439e-01 -1.29485869e+00
-6.04196906e-01 -3.03523719e-01 -2.07870483e-01 -1.15525699e+00
2.68598765e-01 -1.05218899e+00 3.55724961e-01 -1.11875665e+00
4.40083981e-01 -7.05950081e-01 -7.33076692e-01 8.31092417e-01
-6.94141746e-01 1.11748874e-01 -2.64000446e-01 3.67539555e-01
-8.64229262e-01 5.37397087e-01 9.70885098e-01 -3.44807446e-01
-2.03445345e-01 -1.36830360e-01 -1.01661813e+00 6.49443150e-01
6.68487728e-01 -8.42290759e-01 -2.16094479e-01 -2.66974449e-01
8.70283902e-01 -3.47408742e-01 1.84632502e-02 -7.62243569e-01
2.02964067e-01 -1.55109227e-01 -1.61876380e-02 -5.14688432e-01
-2.67891698e-02 -8.46746206e-01 -4.68191713e-01 -4.92670052e-02
-4.00604963e-01 -8.30148607e-02 3.10845822e-02 7.16893017e-01
-3.39344740e-01 -4.50966954e-01 3.64343733e-01 -3.44223827e-01
-7.03801811e-01 3.59767079e-01 -1.01998217e-01 4.58794981e-01
3.76753926e-01 -2.30471678e-02 -3.15627158e-01 7.49779269e-02
-4.08786237e-01 -8.23096838e-03 2.76646852e-01 4.67795879e-01
4.86553162e-01 -1.28269875e+00 -4.90205646e-01 1.25151217e-01
7.36380637e-01 2.41984174e-01 -4.11255807e-02 5.66247344e-01
2.82089919e-01 6.08408630e-01 3.19371611e-01 -5.41024208e-01
-9.82119739e-01 7.50170231e-01 -1.55445918e-01 -7.75542855e-01
-7.49903500e-01 9.73830223e-01 2.74529427e-01 -2.87970930e-01
1.95869163e-01 -5.03568649e-01 -1.46431863e-01 -3.80850956e-02
6.26357794e-01 -4.61041480e-01 4.43280280e-01 -2.86147982e-01
-2.86814034e-01 4.83521342e-01 -4.58463967e-01 -4.34753895e-02
1.28435254e+00 -1.32702380e-01 2.88159121e-02 5.46986759e-01
1.02980745e+00 3.46739084e-01 -8.62686872e-01 -8.52815866e-01
7.63290465e-01 -2.33267456e-01 2.55463421e-02 -8.59414577e-01
-5.44540167e-01 7.80358076e-01 -2.21858616e-03 2.52077222e-01
1.03092039e+00 3.23952138e-01 1.26556969e+00 4.69797969e-01
5.10745287e-01 -1.07504165e+00 3.88552994e-01 7.63315082e-01
5.23432553e-01 -1.23185074e+00 -4.23399448e-01 -3.32058609e-01
-3.42154562e-01 8.62282217e-01 6.03579760e-01 1.81378543e-01
3.98369193e-01 2.30090007e-01 -8.70119929e-02 -3.38760793e-01
-9.58361268e-01 -4.41749394e-01 5.70926726e-01 3.68220806e-01
5.13920009e-01 -7.19146654e-02 -2.98980594e-01 1.18148279e+00
-4.90633309e-01 -3.46452266e-01 3.42838168e-02 1.04115701e+00
-7.75497198e-01 -1.31668103e+00 -1.10964455e-01 6.40383124e-01
-8.45151305e-01 -8.58032048e-01 -2.53226429e-01 2.33565599e-01
2.53011823e-01 7.73973763e-01 -3.99478264e-02 -4.26029295e-01
2.88075238e-01 4.75414395e-01 1.52770191e-01 -1.24950898e+00
-5.98496199e-01 -1.33786112e-01 2.36111417e-01 -2.62052268e-01
-2.17631981e-01 -3.66196245e-01 -1.15649951e+00 6.43897504e-02
-6.92321837e-01 5.20941377e-01 3.30163687e-01 1.18042028e+00
3.06412548e-01 3.81063372e-01 5.73249757e-01 -3.49441081e-01
-5.78437030e-01 -1.11451173e+00 -6.37288928e-01 5.84619403e-01
3.54871035e-01 -6.60126686e-01 -3.58431816e-01 4.74619977e-02]
|
[10.8886137008667, 7.553292274475098]
|
d9b6e13e-9cde-4a4e-9b9c-9151de6eea44
|
geometry-aware-multi-task-learning-for
|
2111.10882
| null |
https://arxiv.org/abs/2111.10882v1
|
https://arxiv.org/pdf/2111.10882v1.pdf
|
Geometry-Aware Multi-Task Learning for Binaural Audio Generation from Video
|
Binaural audio provides human listeners with an immersive spatial sound experience, but most existing videos lack binaural audio recordings. We propose an audio spatialization method that draws on visual information in videos to convert their monaural (single-channel) audio to binaural audio. Whereas existing approaches leverage visual features extracted directly from video frames, our approach explicitly disentangles the geometric cues present in the visual stream to guide the learning process. In particular, we develop a multi-task framework that learns geometry-aware features for binaural audio generation by accounting for the underlying room impulse response, the visual stream's coherence with the sound source(s) positions, and the consistency in geometry of the sounding objects over time. Furthermore, we introduce a new large video dataset with realistic binaural audio simulated for real-world scanned environments. On two datasets, we demonstrate the efficacy of our method, which achieves state-of-the-art results.
|
['Kristen Grauman', 'Ruohan Gao', 'Rishabh Garg']
|
2021-11-21
| null | null | null | null |
['audio-generation', 'room-impulse-response']
|
['audio', 'audio']
|
[-5.23522981e-02 -6.36755645e-01 4.80689913e-01 1.28093185e-02
-1.31015456e+00 -7.24555671e-01 3.56473029e-01 -4.15839814e-02
1.25343859e-01 3.48067999e-01 8.46507430e-01 9.04876664e-02
-4.10924479e-02 -4.94992048e-01 -1.07825530e+00 -5.87716699e-01
-3.58228534e-01 -2.39131227e-01 3.08205068e-01 5.89924045e-02
2.14358404e-01 1.03632033e-01 -2.08230901e+00 6.81275547e-01
3.49281996e-01 1.28239703e+00 2.09473073e-01 1.31947649e+00
4.55464691e-01 8.26995671e-01 -5.38981497e-01 1.52534828e-01
3.59340400e-01 -4.17672276e-01 -3.71732533e-01 -4.26930040e-02
1.29624569e+00 -3.65926504e-01 -6.94565952e-01 6.67682707e-01
7.63429761e-01 3.99584830e-01 6.13439143e-01 -1.20516717e+00
-4.34687018e-01 3.33030224e-01 -3.06207627e-01 2.56871223e-01
9.33355451e-01 3.79618943e-01 1.09795606e+00 -9.05996144e-01
3.42347801e-01 1.11666512e+00 6.26165092e-01 4.35004607e-02
-1.02080107e+00 -8.18791389e-01 2.60299742e-02 5.21661222e-01
-1.31673706e+00 -7.03736901e-01 9.40785170e-01 -7.18647659e-01
5.35710633e-01 2.72353888e-01 9.63996053e-01 9.11063015e-01
3.59048657e-02 6.35357022e-01 8.87403548e-01 -3.56758595e-01
3.15351844e-01 -3.07750612e-01 -4.97625262e-01 2.62719005e-01
-4.64785784e-01 4.83713597e-01 -1.09459543e+00 8.65963697e-02
9.37491536e-01 -3.20537865e-01 -6.22391522e-01 -6.73732400e-01
-1.19060504e+00 3.40603054e-01 4.83402640e-01 1.08349621e-02
-7.72504732e-02 4.57883418e-01 1.67215139e-01 3.98529395e-02
1.82390809e-01 3.42992932e-01 -3.96828689e-02 -3.77077430e-01
-9.30395246e-01 3.90118390e-01 3.62812221e-01 8.97295773e-01
4.35922176e-01 3.81679922e-01 -7.69521147e-02 7.42676258e-01
4.27993089e-01 5.10231316e-01 2.07562476e-01 -1.30663264e+00
3.64825934e-01 -2.91577846e-01 6.62031099e-02 -9.44494009e-01
-8.71183202e-02 -4.79900390e-01 -2.33527869e-01 2.54377753e-01
4.36181188e-01 -4.79363576e-02 -6.79042459e-01 1.70971429e+00
3.76810312e-01 9.65617657e-01 -3.80348057e-01 1.28948677e+00
7.99255550e-01 6.74762428e-01 -1.99627668e-01 -5.57333902e-02
1.14125621e+00 -9.67967093e-01 -5.64890027e-01 -2.23043840e-03
-4.19056825e-02 -9.25505579e-01 1.15833950e+00 6.16147339e-01
-1.25897706e+00 -8.41377318e-01 -8.39706481e-01 8.58339667e-02
1.25516444e-01 -2.25947633e-01 3.91936243e-01 4.87211525e-01
-1.03456879e+00 1.24880798e-01 -6.40197694e-01 1.15709625e-01
1.20055780e-01 -2.91419446e-01 -2.41822749e-01 -1.72405556e-01
-9.83285904e-01 2.17087895e-01 -3.68096858e-01 -6.72025383e-02
-1.59477270e+00 -1.30996776e+00 -1.02342486e+00 9.05027837e-02
2.96969533e-01 -7.17314303e-01 1.70708823e+00 -9.51974869e-01
-1.36408782e+00 1.78278372e-01 -1.95184961e-01 -1.16002247e-01
3.59202534e-01 -4.69037980e-01 -4.60726112e-01 6.10829830e-01
4.74858396e-02 5.44067562e-01 1.08035421e+00 -1.64651632e+00
-5.78005373e-01 -7.77558982e-02 9.12992135e-02 4.16531950e-01
-1.07909337e-01 -4.37711179e-02 -2.90580064e-01 -8.02802980e-01
1.19260391e-02 -5.22439837e-01 1.16885193e-01 2.50853837e-01
-2.73354977e-01 3.03015888e-01 5.48382044e-01 -5.71724057e-01
1.11362290e+00 -2.55155492e+00 1.71563268e-01 -2.82917377e-02
3.25494558e-01 -3.09861124e-01 -2.04256967e-01 2.67527193e-01
-1.00955233e-01 -3.93067569e-01 3.12479317e-01 -4.49734628e-01
2.20254753e-02 -4.92798090e-01 -6.12247169e-01 3.25802743e-01
-4.16226774e-01 4.24868077e-01 -1.10588980e+00 -3.97687882e-01
3.35573256e-01 7.68728256e-01 -1.01690316e+00 4.11118090e-01
5.65987825e-02 5.43899000e-01 -5.73621988e-02 7.53254116e-01
5.06028533e-01 1.14841558e-01 -2.92401046e-01 -4.25720602e-01
-1.87050939e-01 2.70190299e-01 -1.34383416e+00 1.89424670e+00
-6.51953638e-01 7.97645867e-01 3.40451777e-01 -3.23584348e-01
4.42580491e-01 5.07968783e-01 4.85098988e-01 -7.32411861e-01
-1.57315925e-01 1.02584161e-01 -2.69343317e-01 -4.90072817e-01
4.67707098e-01 -3.93552072e-02 5.35528138e-02 3.65667552e-01
1.57959089e-01 -4.46450502e-01 -2.36826494e-01 1.48808926e-01
9.33988452e-01 1.58143252e-01 4.00972813e-02 2.14054152e-01
-1.12698898e-02 -6.09786153e-01 2.08810344e-01 7.04288304e-01
-2.24862456e-01 1.31933117e+00 3.64920162e-02 -1.63850516e-01
-9.38050568e-01 -1.80345428e+00 2.19499804e-02 1.16333354e+00
1.30039588e-01 -7.86577225e-01 -6.95802152e-01 -1.36514574e-01
-1.79140270e-01 5.02719045e-01 -4.33062494e-01 -3.51569392e-02
-3.38066995e-01 4.75055166e-02 4.21717167e-01 5.89401603e-01
1.61880568e-01 -7.11926639e-01 -5.85110545e-01 1.12497404e-01
-5.26865304e-01 -1.23840857e+00 -7.44752645e-01 -3.66919309e-01
-3.14815253e-01 -1.03924811e+00 -6.16598487e-01 -4.02087539e-01
-2.58236509e-02 5.44404566e-01 1.02407634e+00 -6.14577293e-01
-7.26156116e-01 1.07010114e+00 -3.75269771e-01 -3.05383384e-01
7.82660022e-02 -5.31161666e-01 9.05375853e-02 2.10030913e-01
-3.27368915e-01 -1.08291054e+00 -9.67988968e-01 2.26945624e-01
-7.27298856e-01 3.67242098e-02 1.71863943e-01 2.70977527e-01
5.74456692e-01 -1.63362436e-02 4.05648738e-01 1.37337178e-01
2.10288256e-01 -4.20927197e-01 -1.84234232e-01 -2.23437548e-01
1.84772208e-01 -5.73893011e-01 5.81474125e-01 -5.92775524e-01
-9.69919622e-01 7.82383084e-02 5.38233668e-02 -7.44883776e-01
-4.01804984e-01 6.08477965e-02 -2.22532526e-01 -1.40404895e-01
6.87579691e-01 1.84716374e-01 -4.86994386e-01 -4.40877348e-01
3.97862971e-01 5.61415374e-01 8.84844065e-01 -5.45352876e-01
5.83293200e-01 6.94069445e-01 -2.11222675e-02 -1.08494186e+00
-7.86759019e-01 -4.03895676e-01 -3.97770554e-01 -7.29862928e-01
8.83686900e-01 -1.31710911e+00 -7.42293060e-01 5.33914030e-01
-9.24665213e-01 -5.03015935e-01 -2.33458474e-01 9.14208770e-01
-8.80026639e-01 1.87750891e-01 -5.05082309e-01 -8.38313580e-01
3.76092404e-01 -1.02690995e+00 1.16783988e+00 7.25574493e-02
-1.50942728e-01 -6.81737602e-01 4.67744470e-01 4.26618785e-01
2.10097983e-01 2.86210347e-02 7.41450429e-01 2.33686902e-02
-1.13976181e+00 1.38074771e-01 1.03991941e-01 1.13649167e-01
3.58110145e-02 3.87813523e-02 -1.41500449e+00 -1.47111192e-01
-3.36681187e-01 -4.71493930e-01 6.49774849e-01 1.05904770e+00
1.34664357e+00 -2.64500141e-01 2.42338534e-02 9.15620923e-01
1.17876840e+00 1.04757637e-01 6.17467761e-01 9.98008400e-02
9.06170309e-01 5.76753318e-01 5.65017641e-01 7.02247083e-01
4.53720480e-01 1.02207565e+00 6.57498598e-01 1.03344508e-02
-5.84614754e-01 -6.34526193e-01 4.18632209e-01 7.10731864e-01
-8.69610161e-02 -2.05194071e-01 -5.95469952e-01 5.97762227e-01
-1.33562899e+00 -1.05984807e+00 1.47119567e-01 2.22163749e+00
6.98173821e-01 -2.11107627e-01 2.16624409e-01 1.88514307e-01
5.94610453e-01 3.01814675e-01 -1.95323750e-01 -1.32161573e-01
-6.06022179e-02 3.74792695e-01 -8.17488432e-02 6.76632106e-01
-1.09431422e+00 4.95353401e-01 6.99526024e+00 6.32746935e-01
-1.35129213e+00 -3.10799107e-02 7.91404918e-02 -8.70618641e-01
-4.75558847e-01 -1.84568614e-01 -3.66758138e-01 3.50821614e-01
7.05938458e-01 -6.55512437e-02 6.29080117e-01 6.79564416e-01
3.60249847e-01 -8.56458768e-02 -1.27721834e+00 1.17990077e+00
3.66534442e-01 -1.27376521e+00 -4.91045266e-02 3.40677751e-03
5.26881218e-01 -2.77409345e-01 5.87165236e-01 6.16535209e-02
1.32181048e-01 -1.14684570e+00 1.35090911e+00 5.98951876e-01
9.47036207e-01 -7.17147350e-01 -1.60593554e-01 -1.37319878e-01
-1.66684055e+00 -2.31391564e-01 1.72164500e-01 -2.54666984e-01
4.53925341e-01 3.17649156e-01 -7.40958154e-01 3.80395144e-01
1.17118776e+00 6.37606621e-01 -6.08976007e-01 1.65709627e+00
-1.66289359e-01 7.23340869e-01 -2.54646719e-01 4.21056896e-01
-6.94323480e-02 3.29074830e-01 9.40511584e-01 1.15241301e+00
6.85243249e-01 1.99917722e-02 2.22198531e-01 5.86254299e-01
2.04555720e-01 5.27679399e-02 -6.57903016e-01 2.10270643e-01
7.29612231e-01 1.01428092e+00 -2.42957070e-01 2.34518006e-01
-4.31827784e-01 5.81445038e-01 -2.61778086e-01 6.69107616e-01
-9.41194654e-01 -4.18053061e-01 9.22040761e-01 4.26280558e-01
5.74947536e-01 -3.79782945e-01 -1.30904717e-02 -8.44830155e-01
1.14135355e-01 -8.57520223e-01 1.91761151e-01 -1.51966488e+00
-1.00200641e+00 5.17229021e-01 4.69628572e-02 -1.62186301e+00
-5.10583460e-01 -4.05979127e-01 -6.49142921e-01 5.62305152e-01
-1.32378101e+00 -1.26501501e+00 -4.47552770e-01 8.42917383e-01
5.04756510e-01 7.17761964e-02 5.93477070e-01 5.40374160e-01
5.06987469e-03 6.97561145e-01 2.00369824e-02 5.80345318e-02
1.21598017e+00 -1.03209710e+00 4.38441522e-03 7.14048445e-01
4.37493294e-01 2.93161094e-01 1.01915073e+00 -1.92438334e-01
-1.23635221e+00 -8.87943685e-01 2.92722195e-01 -4.84542936e-01
6.41853869e-01 -5.87828994e-01 -6.79358721e-01 6.72439396e-01
7.41949975e-02 2.44808316e-01 1.03279483e+00 7.62789771e-02
-7.03091681e-01 -3.10774475e-01 -5.24707198e-01 4.94681686e-01
9.43692625e-01 -8.50994706e-01 -4.58352596e-01 -7.60000721e-02
8.01126778e-01 -5.64431727e-01 -4.68925565e-01 2.34450176e-01
8.95737290e-01 -1.18424344e+00 1.34985602e+00 -3.32715422e-01
7.48569548e-01 -5.05521059e-01 -5.62468529e-01 -1.56836760e+00
-4.05771047e-01 -6.46579623e-01 -3.37425739e-01 9.66629028e-01
-1.67442355e-02 -4.36615981e-02 2.49135181e-01 -4.39622663e-02
-4.90943253e-01 -3.12201142e-01 -9.68149841e-01 -5.57449043e-01
-2.06078514e-01 -8.66053879e-01 4.66921657e-01 7.47579873e-01
-7.43621867e-03 8.94135162e-02 -4.27832633e-01 3.94536763e-01
7.43507624e-01 2.35965922e-01 8.50091815e-01 -1.10115242e+00
-6.66361153e-01 -1.74591839e-01 -5.64381659e-01 -1.19575679e+00
1.44265518e-01 -4.08978075e-01 2.41985053e-01 -1.25687551e+00
-1.48107290e-01 -1.13916226e-01 -2.37463236e-01 2.62957830e-02
1.25677466e-01 5.52880168e-01 2.95573205e-01 -1.28520623e-01
-6.65558457e-01 6.44662082e-01 1.39250505e+00 1.17702655e-01
-3.04015309e-01 -2.93287560e-02 -6.31223440e-01 8.28660727e-01
1.89520001e-01 -1.24441668e-01 -5.63754320e-01 -6.25265718e-01
3.22886914e-01 4.45404798e-01 9.36794996e-01 -1.28602505e+00
4.49236393e-01 -2.36327752e-01 4.83925492e-01 -4.64206398e-01
7.55793393e-01 -5.04152834e-01 1.20832942e-01 -2.75903493e-01
-5.69316030e-01 -2.39322186e-01 3.33819598e-01 8.60633016e-01
-4.28550869e-01 2.77380466e-01 7.71934688e-01 5.23682684e-02
-4.90301371e-01 3.39428186e-01 -5.66455543e-01 2.96963662e-01
7.37082422e-01 -2.76917815e-01 -1.23389170e-01 -9.22725976e-01
-7.95936525e-01 -1.81208953e-01 3.43129545e-01 4.38524932e-01
6.86296880e-01 -1.55631530e+00 -5.43122649e-01 3.07145387e-01
1.05322808e-01 -4.64964025e-02 8.85368943e-01 7.02494323e-01
-2.40098163e-01 2.28397325e-01 -2.58244336e-01 -9.58827019e-01
-1.31167328e+00 3.48577440e-01 4.21802253e-01 6.63736105e-01
-5.47895849e-01 1.04485881e+00 8.01140189e-01 8.63604173e-02
4.78763580e-01 -3.85045558e-01 -6.19177222e-02 1.98431127e-02
6.77409470e-01 4.77922201e-01 -8.24764892e-02 -5.94762504e-01
-5.74405305e-02 6.87429726e-01 3.48215908e-01 -6.51737154e-01
1.22935939e+00 -4.44697142e-01 3.58087301e-01 8.52770030e-01
1.25921357e+00 6.97989702e-01 -1.87599075e+00 -2.06742376e-01
-8.27562571e-01 -1.03496850e+00 2.53042638e-01 -7.07998097e-01
-7.85169065e-01 1.25352335e+00 5.97031236e-01 1.35066314e-02
1.28886950e+00 1.63464665e-01 7.51666665e-01 -2.41210703e-02
2.21251786e-01 -8.70726943e-01 8.39350998e-01 3.27959031e-01
8.90036047e-01 -7.23246336e-01 -4.07272100e-01 -4.54473943e-01
-5.46613991e-01 1.06854248e+00 6.03758991e-01 7.45735466e-02
7.42739439e-01 3.65565509e-01 2.04604447e-01 5.47908880e-02
-7.64465690e-01 -9.87547934e-02 5.68786025e-01 9.07708824e-01
3.10527265e-01 -6.24590032e-02 8.45762789e-01 7.05421269e-01
-7.51310110e-01 -4.02370721e-01 4.84081805e-01 7.16570795e-01
-4.96524066e-01 -3.45429689e-01 -4.95141268e-01 -1.67252317e-01
-2.54629076e-01 -2.21841067e-01 -2.27496132e-01 3.17058563e-01
1.02884546e-01 8.82250726e-01 3.60230267e-01 -4.31676418e-01
4.88903165e-01 6.81748167e-02 9.96639550e-01 -4.95649785e-01
-2.50551969e-01 6.68792248e-01 -2.89575756e-01 -8.03460002e-01
-1.66476861e-01 -9.20076489e-01 -9.16924536e-01 4.87894826e-02
1.67182073e-01 3.85007448e-02 5.57814062e-01 6.78137422e-01
3.54554445e-01 8.51038456e-01 7.82815933e-01 -1.33593786e+00
-9.45655853e-02 -7.12575614e-01 -7.92421043e-01 2.83680350e-01
8.91264319e-01 -7.50910759e-01 -7.17613518e-01 3.77055168e-01]
|
[14.970193862915039, 5.095459461212158]
|
1713c657-9b84-4b4d-a789-920bc093c466
|
rt-track-robust-tricks-for-multi-pedestrian
|
2303.09668
| null |
https://arxiv.org/abs/2303.09668v1
|
https://arxiv.org/pdf/2303.09668v1.pdf
|
Rt-Track: Robust Tricks for Multi-Pedestrian Tracking
|
Object tracking is divided into single-object tracking (SOT) and multi-object tracking (MOT). MOT aims to maintain the identities of multiple objects across a series of continuous video sequences. In recent years, MOT has made rapid progress. However, modeling the motion and appearance models of objects in complex scenes still faces various challenging issues. In this paper, we design a novel direction consistency method for smooth trajectory prediction (STP-DC) to increase the modeling of motion information and overcome the lack of robustness in previous methods in complex scenes. Existing methods use pedestrian re-identification (Re-ID) to model appearance, however, they extract more background information which lacks discriminability in occlusion and crowded scenes. We propose a hyper-grain feature embedding network (HG-FEN) to enhance the modeling of appearance models, thus generating robust appearance descriptors. We also proposed other robustness techniques, including CF-ECM for storing robust appearance information and SK-AS for improving association accuracy. To achieve state-of-the-art performance in MOT, we propose a robust tracker named Rt-track, incorporating various tricks and techniques. It achieves 79.5 MOTA, 76.0 IDF1 and 62.1 HOTA on the test set of MOT17.Rt-track also achieves 77.9 MOTA, 78.4 IDF1 and 63.3 HOTA on MOT20, surpassing all published methods.
|
['Shan Zhao', 'Yang Yang', 'Limin Zhao', 'Mengzhen Li', 'Housheng Xie', 'Yunhua Jia', 'Yukuan Zhang']
|
2023-03-16
| null | null | null | null |
['trajectory-prediction']
|
['computer-vision']
|
[-4.11397159e-01 -8.26031148e-01 2.90353242e-02 -1.20493710e-01
-3.16526294e-01 -3.78208518e-01 4.31215584e-01 -9.77461562e-02
-3.01169485e-01 6.03871107e-01 -1.17337205e-01 1.66212663e-01
2.67182128e-03 -3.78897160e-01 -6.95512414e-01 -8.64835203e-01
-1.29272621e-02 2.49791458e-01 7.73067355e-01 1.12737916e-01
-3.27421799e-02 4.20999527e-01 -1.70414615e+00 -1.10800788e-01
6.84379697e-01 9.26738679e-01 1.36948869e-01 6.57802939e-01
8.18333402e-02 7.28777945e-01 -4.62084651e-01 -5.88800669e-01
2.88134366e-01 2.77916156e-02 -2.79080302e-01 -6.20168298e-02
9.85682189e-01 -3.99599522e-01 -5.84854007e-01 1.03973770e+00
5.63837290e-01 2.93669492e-01 6.63229048e-01 -1.74335933e+00
-7.66516328e-01 -1.80624835e-02 -9.49175298e-01 2.30678901e-01
1.56316355e-01 2.30726808e-01 4.83627737e-01 -8.64480793e-01
5.43803394e-01 1.53396940e+00 9.13971364e-01 5.98544419e-01
-9.13429022e-01 -9.45504665e-01 3.61277610e-01 6.08815551e-01
-1.50376034e+00 -3.97210002e-01 4.77652609e-01 -5.70375204e-01
4.88682717e-01 4.42521930e-01 8.08022976e-01 1.00089061e+00
2.00724944e-01 9.24403012e-01 9.08738732e-01 -3.14288288e-02
-2.76351482e-01 1.54885158e-01 4.32860553e-01 7.02920616e-01
6.09218240e-01 1.52282536e-01 -3.41614068e-01 -8.21769163e-02
7.84601331e-01 1.73385486e-01 -7.64528066e-02 -2.84876257e-01
-1.27751315e+00 4.17050093e-01 3.48086327e-01 5.44304028e-02
-2.64959782e-01 2.12819993e-01 4.64151204e-01 9.99328345e-02
1.47650152e-01 -2.35899016e-01 -1.36483416e-01 -1.56344716e-02
-6.66733742e-01 4.14968044e-01 2.16563225e-01 1.31412721e+00
4.99802470e-01 2.24961326e-01 -4.76527870e-01 8.22952807e-01
6.27958179e-01 9.90477026e-01 3.41836810e-01 -7.57609427e-01
1.77069411e-01 5.84440649e-01 3.61067474e-01 -1.34053993e+00
-3.43438923e-01 -4.15268153e-01 -1.04309475e+00 1.76741034e-01
4.11215961e-01 1.93939283e-01 -7.83244193e-01 1.72410572e+00
7.20322728e-01 5.64083636e-01 -2.19250754e-01 8.52122247e-01
9.70166445e-01 6.50263190e-01 1.95665807e-01 -1.92140460e-01
1.41044652e+00 -1.17960358e+00 -8.84285986e-01 9.56923813e-02
3.71679962e-01 -9.04750645e-01 3.28545660e-01 3.19061339e-01
-8.40196550e-01 -1.14664638e+00 -8.40973020e-01 2.22402737e-01
-2.94841290e-01 5.13793707e-01 3.25521588e-01 6.10565603e-01
-9.56751525e-01 3.14775467e-01 -8.13478947e-01 -4.56020683e-01
5.49077213e-01 6.16069734e-01 -4.77522135e-01 -2.04362467e-01
-7.92721629e-01 8.42836499e-01 1.86997294e-01 1.56428561e-01
-9.58270371e-01 -5.03599942e-01 -6.75984144e-01 -3.44741970e-01
1.40049040e-01 -7.49328554e-01 7.27668107e-01 -5.86649835e-01
-1.20819712e+00 6.73420250e-01 -4.81019735e-01 -1.94159389e-01
6.58571959e-01 -3.81673574e-01 -9.08961535e-01 -1.25559688e-01
1.51093259e-01 7.61179507e-01 9.94008958e-01 -1.18575442e+00
-8.55732858e-01 -3.08983862e-01 -2.85273731e-01 -1.02010161e-01
-4.58012134e-01 3.63211155e-01 -8.44214261e-01 -6.52052939e-01
-7.41120279e-02 -1.24548125e+00 -6.32081106e-02 5.14916241e-01
-1.92418799e-01 -4.39436913e-01 1.28366148e+00 -8.00957561e-01
1.24766612e+00 -2.14525843e+00 5.83635531e-02 -2.53487319e-01
3.50635856e-01 7.37257779e-01 -2.70918071e-01 6.66960608e-03
2.77055562e-01 -3.00460488e-01 3.33347768e-01 -5.83655596e-01
4.10852246e-02 8.16498473e-02 1.33438051e-01 7.84413278e-01
1.50682986e-01 8.71008277e-01 -7.74838567e-01 -8.76965761e-01
5.70622325e-01 5.78409970e-01 -2.06158474e-01 1.97732836e-01
2.53252536e-01 5.27441621e-01 -3.66119415e-01 8.77572417e-01
1.25597405e+00 -2.30175123e-01 -2.91347563e-01 -5.18505454e-01
-3.20218801e-01 -4.54002231e-01 -1.65162265e+00 1.24969649e+00
6.67443797e-02 6.03685319e-01 -1.03538122e-03 -5.76151013e-01
8.97081017e-01 1.35601431e-01 8.50472212e-01 -5.53552687e-01
2.04711348e-01 -7.33792856e-02 -5.45201413e-02 -5.84444702e-01
6.38619363e-01 4.27725971e-01 3.26023638e-01 -4.14311141e-02
-1.26745492e-01 7.94287860e-01 1.69474646e-01 7.55701810e-02
9.02979791e-01 2.94372022e-01 4.08827811e-02 -2.48151213e-01
7.73177445e-01 -1.58370331e-01 8.68759692e-01 6.89194798e-01
-7.55716980e-01 4.25321996e-01 -4.61204946e-01 -6.03033125e-01
-9.99453962e-01 -1.01681328e+00 -3.17647010e-01 7.34539986e-01
6.99496984e-01 -4.55174834e-01 -3.83931607e-01 -5.94916105e-01
2.62217015e-01 1.18324883e-01 -4.94380981e-01 -1.33368522e-01
-8.97278249e-01 -9.32566226e-01 4.84685212e-01 6.83908522e-01
7.20454693e-01 -7.06714571e-01 -3.41092139e-01 2.65362799e-01
-3.13887030e-01 -1.24996221e+00 -8.20611358e-01 -5.96167326e-01
-5.41974723e-01 -1.14247441e+00 -8.00744832e-01 -7.56480277e-01
6.88567996e-01 8.12406957e-01 7.82304108e-01 3.40103209e-01
-3.73849362e-01 3.01180840e-01 -3.72109652e-01 -3.01906705e-01
-1.34319216e-01 -4.81086522e-01 5.41975796e-01 3.92569125e-01
3.88098180e-01 -1.54699728e-01 -6.68306887e-01 8.63290429e-01
-5.06020367e-01 2.27948390e-02 5.27342200e-01 7.70738423e-01
5.62930107e-01 -3.74474749e-02 2.47300029e-01 2.55949236e-02
-1.63575321e-01 -4.02984440e-01 -7.29471743e-01 4.19804931e-01
-4.36721176e-01 -3.29832971e-01 5.46414495e-01 -1.00196826e+00
-8.21698546e-01 1.02961697e-01 4.75363247e-02 -7.94845819e-01
-1.05930388e-01 -3.16258699e-01 -2.45424658e-01 -5.52691638e-01
7.70460591e-02 3.80237699e-01 -1.19429305e-01 -5.55331171e-01
2.06805721e-01 6.15606189e-01 6.00572526e-01 -4.62928414e-01
1.12058902e+00 6.19582057e-01 1.36296958e-01 -8.19166541e-01
-4.81634527e-01 -7.47881234e-01 -7.06488431e-01 -5.27355671e-01
8.88773084e-01 -1.11076891e+00 -1.07495749e+00 8.24993849e-01
-1.17301452e+00 4.24309708e-02 2.16988295e-01 6.75917923e-01
-1.79860190e-01 6.48714840e-01 -3.93860906e-01 -1.01809227e+00
-3.57583135e-01 -1.07043433e+00 9.97783899e-01 4.29373652e-01
9.83292535e-02 -6.66667640e-01 1.26018282e-02 1.39327630e-01
4.22682136e-01 4.36567903e-01 1.28790677e-01 -2.19998360e-01
-9.38925087e-01 -1.68319106e-01 -6.03901803e-01 1.40811428e-01
2.54080951e-01 2.60689199e-01 -8.32284153e-01 -7.05321133e-01
-2.98354238e-01 5.68008423e-02 7.01949179e-01 5.42340100e-01
8.58750820e-01 -2.19146356e-01 -8.49813521e-01 7.35039532e-01
1.36159039e+00 4.13041919e-01 5.10270238e-01 4.96985167e-01
1.07447624e+00 2.46786714e-01 9.44167674e-01 3.84246349e-01
6.13152206e-01 1.25141084e+00 3.41937959e-01 1.67480320e-01
-4.82612461e-01 -4.08180431e-03 5.61776161e-01 9.54378009e-01
-3.05889219e-01 -2.16276288e-01 -4.90235537e-01 7.76601017e-01
-2.18583655e+00 -1.22019947e+00 -6.45662069e-01 2.08686209e+00
3.60343933e-01 4.48056608e-02 4.81276840e-01 -1.61717281e-01
1.01653838e+00 -6.51768222e-02 -4.53423232e-01 3.36165071e-01
-2.66264975e-01 -4.88830090e-01 5.91110229e-01 2.09487230e-01
-1.39219701e+00 7.88317561e-01 5.57347059e+00 8.43403220e-01
-7.95897126e-01 3.07530135e-01 8.70494843e-02 6.48079589e-02
4.03621018e-01 -1.60071701e-01 -1.56414032e+00 8.47780526e-01
6.55337512e-01 3.08852363e-02 1.06727503e-01 8.53798687e-01
1.06775485e-01 1.37798712e-01 -7.37924099e-01 1.33823574e+00
1.70472533e-01 -1.09213614e+00 7.47539252e-02 -6.46821335e-02
6.27458215e-01 -2.13604391e-01 6.38467297e-02 3.83986413e-01
2.69240230e-01 -6.94036007e-01 8.10444117e-01 6.58842623e-01
6.21499002e-01 -3.66980374e-01 7.42090166e-01 8.65271240e-02
-1.94173372e+00 -1.00141443e-01 -6.24533832e-01 2.33135387e-01
3.14210355e-01 2.62100756e-01 -3.50076437e-01 1.00274324e+00
9.94240284e-01 1.12869763e+00 -8.84504557e-01 1.62808180e+00
3.47600877e-01 2.54616857e-01 -5.17082572e-01 -1.18250668e-01
2.36212276e-02 -4.47205715e-02 7.85475373e-01 1.22753739e+00
3.38623315e-01 -6.47048429e-02 5.03253698e-01 4.44744140e-01
2.43104398e-01 4.73306403e-02 -2.90435046e-01 3.71457934e-01
5.45743823e-01 1.32782042e+00 -4.77373332e-01 -5.35268664e-01
-4.70357060e-01 8.86167109e-01 1.79155856e-01 8.42551067e-02
-1.36524558e+00 2.27107797e-02 9.32675719e-01 -8.09306726e-02
5.87807357e-01 -3.29346716e-01 3.39838117e-01 -1.22042966e+00
2.01132059e-01 -7.88606226e-01 4.58349615e-01 -5.41283846e-01
-1.31976521e+00 5.58166742e-01 8.56246650e-02 -1.57760596e+00
3.78644943e-01 -6.82146192e-01 -6.17740154e-01 5.35100400e-01
-1.39961839e+00 -1.69874179e+00 -7.60610282e-01 6.66036665e-01
6.22209668e-01 -1.97708741e-01 3.66671503e-01 8.62151802e-01
-1.06568801e+00 9.43763256e-01 4.32006419e-01 1.74317122e-01
9.62851286e-01 -8.86568487e-01 3.85430336e-01 1.00590718e+00
-2.06416845e-02 6.09287024e-01 6.21473134e-01 -9.03141439e-01
-1.63245237e+00 -1.57320750e+00 5.44841051e-01 -6.71014309e-01
3.63945007e-01 -2.13743344e-01 -9.32615399e-01 6.74328744e-01
-1.03693470e-01 3.61902267e-01 3.43125075e-01 -2.22772107e-01
-2.76743829e-01 -2.90152043e-01 -1.07924378e+00 4.99689668e-01
1.22374523e+00 1.43617913e-02 -1.63743466e-01 1.06424093e-01
5.98173022e-01 -4.51443285e-01 -1.03341091e+00 4.62997317e-01
8.82623315e-01 -6.25565827e-01 1.37149072e+00 -3.60981882e-01
-4.84227151e-01 -1.02436709e+00 -3.00536335e-01 -7.43928552e-01
-8.67230058e-01 -6.28565550e-01 -5.08694768e-01 1.51147163e+00
-4.19862509e-01 -4.35221404e-01 5.88043213e-01 3.87157917e-01
-1.26011789e-01 -4.82057780e-01 -9.27966535e-01 -1.21165764e+00
-2.67135173e-01 -3.11533827e-02 7.31390357e-01 8.93796086e-01
-7.28916347e-01 -9.27958265e-02 -8.78884375e-01 5.17450154e-01
1.16904366e+00 1.47903822e-02 1.15948546e+00 -1.40075469e+00
-4.06018794e-02 -2.79516190e-01 -7.51142502e-01 -1.13795006e+00
-4.92515415e-02 -6.21248126e-01 -2.04747647e-01 -1.36204255e+00
4.05486166e-01 -6.47302330e-01 -5.83934486e-01 3.53389174e-01
-4.00954545e-01 3.44100118e-01 3.70263934e-01 4.25539225e-01
-1.13702857e+00 7.66771376e-01 1.26234090e+00 -4.69902813e-01
4.81395312e-02 6.11948315e-03 -1.86620697e-01 5.39965332e-01
6.07839167e-01 -6.13316357e-01 -2.53517535e-02 -3.80583346e-01
-5.00447154e-01 -3.50901693e-01 8.56643021e-01 -1.42175603e+00
4.77943152e-01 -1.54266000e-01 7.63177931e-01 -1.17971468e+00
6.28784001e-01 -9.46861327e-01 6.77008867e-01 6.24236763e-01
3.27110857e-01 4.30388510e-01 5.42798042e-01 7.56076276e-01
6.36494905e-02 -1.48057807e-02 8.74718308e-01 1.92670807e-01
-1.08596814e+00 7.28076935e-01 -1.00492775e-01 -2.75020897e-01
1.26979899e+00 -3.98659319e-01 -4.98347670e-01 1.00632742e-01
-4.31518763e-01 5.04211426e-01 5.71564376e-01 7.26837039e-01
6.69879913e-01 -1.83822989e+00 -8.32005739e-01 1.05095506e-01
1.26991853e-01 -3.49915564e-01 5.32041967e-01 1.15403974e+00
-1.51318252e-01 3.51946086e-01 -4.84363645e-01 -8.70403767e-01
-1.81467247e+00 6.97358489e-01 1.60351500e-01 -1.00106947e-01
-8.30666244e-01 5.87057412e-01 2.46293962e-01 -4.88329902e-02
3.12106282e-01 3.79376784e-02 -3.09797049e-01 -1.54835910e-01
8.33893001e-01 7.62283206e-01 -3.62020761e-01 -1.20996082e+00
-6.51701152e-01 1.06072605e+00 -3.35009724e-01 3.94311845e-01
9.87168908e-01 -1.97738141e-01 -2.57814601e-02 1.37829185e-01
1.11328423e+00 -1.25329271e-01 -1.40999484e+00 -1.63133219e-01
-1.88385457e-01 -8.59200001e-01 -1.93868920e-01 -4.63639766e-01
-9.33801472e-01 6.27634704e-01 1.12064290e+00 -1.88870892e-01
8.62372458e-01 -3.46761256e-01 9.76969123e-01 2.16555521e-01
4.82993066e-01 -7.59590268e-01 2.76520759e-01 3.82759839e-01
6.07535720e-01 -1.37977731e+00 1.40980735e-01 -4.47310716e-01
-4.01920676e-01 9.45583820e-01 1.03202152e+00 5.05899824e-02
5.61599791e-01 -1.73230376e-02 -6.92025274e-02 -3.27387378e-02
-4.33170199e-01 -2.41216362e-01 4.21950370e-01 8.00943255e-01
-6.61398172e-02 -1.00602247e-01 -1.41705724e-03 3.50816697e-01
2.95942485e-01 -1.01782523e-01 1.20587647e-01 7.63387740e-01
-3.77916932e-01 -1.13056159e+00 -6.67534351e-01 2.83779860e-01
-2.16310352e-01 3.70499343e-01 -1.52092539e-02 7.75876999e-01
4.00318593e-01 9.93227959e-01 -1.52319461e-01 -6.92756116e-01
1.96555808e-01 -2.70932913e-01 5.05313396e-01 7.47480243e-02
-4.21731859e-01 1.57524735e-01 -3.17502134e-02 -3.41517568e-01
-7.32197940e-01 -9.70482647e-01 -7.96952426e-01 -4.70461726e-01
-7.01990485e-01 -7.13756233e-02 4.27010596e-01 5.95461011e-01
4.23182428e-01 4.95522648e-01 5.12482464e-01 -8.78937840e-01
-3.52241307e-01 -9.06246185e-01 -3.16007584e-01 6.56571031e-01
4.71731782e-01 -1.19556868e+00 -3.97486277e-02 1.12412289e-01]
|
[6.449484348297119, -2.022310495376587]
|
41fc4f2d-70d7-4a7b-af45-333b5757422a
|
atrial-fibrillation-detection-and-ecg
|
2011.06187
| null |
https://arxiv.org/abs/2011.06187v1
|
https://arxiv.org/pdf/2011.06187v1.pdf
|
Atrial Fibrillation Detection and ECG Classification based on CNN-BiLSTM
|
It is challenging to visually detect heart disease from the electrocardiographic (ECG) signals. Implementing an automated ECG signal detection system can help diagnosis arrhythmia in order to improve the accuracy of diagnosis. In this paper, we proposed, implemented, and compared an automated system using two different frameworks of the combination of convolutional neural network (CNN) and long-short term memory (LSTM) for classifying normal sinus signals, atrial fibrillation, and other noisy signals. The dataset we used is from the MIT-BIT Arrhythmia Physionet. Our approach demonstrated that the cascade of two deep learning network has higher performance than the concatenation of them, achieving a weighted f1 score of 0.82. The experimental results have successfully validated that the cascade of CNN and LSTM can achieve satisfactory performance on discriminating ECG signals.
|
['Weiheng Li', 'Jiacheng Wang']
|
2020-11-12
| null | null | null | null |
['ecg-classification', 'atrial-fibrillation-detection']
|
['medical', 'medical']
|
[ 8.58052894e-02 -4.84904289e-01 3.90491188e-01 -1.24055028e-01
-8.78810287e-01 -2.99242079e-01 -2.45468497e-01 -6.40880018e-02
-2.48011023e-01 8.30797315e-01 -1.32832274e-01 -6.09073937e-01
-1.53736696e-01 -3.28753501e-01 -1.24043301e-01 -6.91147327e-01
-6.35425568e-01 -8.39796811e-02 -4.39884961e-01 3.19190979e-01
-2.73096450e-02 7.12075651e-01 -9.88458276e-01 6.47835076e-01
5.05880713e-01 1.28083265e+00 -1.16606236e-01 1.30924761e+00
3.28301847e-01 7.42465317e-01 -1.09165609e+00 2.66290843e-01
1.90492999e-02 -8.96334052e-01 -7.28378415e-01 -4.51167405e-01
9.09390077e-02 -3.62816006e-01 -7.65889660e-02 6.09195590e-01
1.52935600e+00 -6.67536736e-01 4.11462128e-01 -7.23507762e-01
-2.37246007e-01 5.75126648e-01 -2.76615202e-01 9.02487576e-01
2.36608654e-01 2.07792476e-01 3.81745815e-01 -8.42798531e-01
2.51824170e-01 6.28530502e-01 1.43811035e+00 3.06447536e-01
-8.83494556e-01 -7.15646923e-01 -6.72303975e-01 3.76553237e-01
-1.53699076e+00 -8.79334733e-02 8.15366387e-01 -4.12078291e-01
1.03139269e+00 2.87315071e-01 9.80197549e-01 1.09779012e+00
5.97054362e-01 3.72533619e-01 9.90338683e-01 -6.40150428e-01
-1.71491176e-01 -4.76688370e-02 2.30103523e-01 6.37565315e-01
1.80707663e-01 1.69900268e-01 -2.75504738e-01 -5.38616359e-01
9.95895565e-01 1.39238061e-02 -4.49680984e-01 3.15846980e-01
-1.30455530e+00 3.11712354e-01 3.93497527e-01 9.14822221e-01
-5.66972256e-01 3.09021771e-01 6.82543576e-01 6.38963819e-01
2.32734844e-01 5.20078838e-01 -5.65392315e-01 -3.10714185e-01
-9.42712486e-01 -1.78186014e-01 5.60377955e-01 6.37838617e-02
-9.09264460e-02 6.33772075e-01 -5.02956092e-01 6.28193259e-01
1.82312086e-01 5.41460276e-01 6.76342487e-01 -6.15696132e-01
1.13878861e-01 3.09285104e-01 -9.79289860e-02 -1.04697168e+00
-7.98806489e-01 -1.04273140e+00 -1.27902222e+00 2.36665100e-01
3.74193847e-01 -7.37340391e-01 -9.75994289e-01 1.15366411e+00
-3.02028507e-01 4.48156357e-01 1.82492942e-01 8.69540393e-01
1.14581990e+00 3.62255245e-01 9.77698117e-02 -2.09983915e-01
1.20238841e+00 -2.23326206e-01 -9.49923337e-01 2.74655432e-01
8.33815753e-01 -5.86122394e-01 6.35783553e-01 5.33882201e-01
-1.10041869e+00 -1.00680518e+00 -1.30640996e+00 4.02558625e-01
-1.90410987e-02 7.19958901e-01 2.25618035e-01 9.51736271e-01
-1.21526933e+00 8.83265674e-01 -7.47592807e-01 -1.87056661e-01
6.24089718e-01 3.91054213e-01 -1.48888677e-03 4.72818285e-01
-1.31630957e+00 7.01767266e-01 1.85942277e-01 5.63239992e-01
-8.11132550e-01 -4.24390018e-01 -4.96777236e-01 1.29818156e-01
-5.09773612e-01 -8.84461582e-01 9.28896070e-01 -9.90284801e-01
-1.19236255e+00 8.29759002e-01 7.77213424e-02 -8.65892887e-01
5.73956966e-01 -3.28047663e-01 -6.15000963e-01 1.92257002e-01
-3.46631497e-01 2.48359993e-01 8.48883033e-01 -8.04951668e-01
-4.07479763e-01 -4.37854201e-01 -4.59700108e-01 -5.57812043e-02
-2.11142227e-01 1.43590599e-01 2.85027415e-01 -6.90904498e-01
1.62690595e-01 -5.69659114e-01 -3.78744975e-02 -2.97137588e-01
-4.17868853e-01 1.46722227e-01 9.34217334e-01 -9.36719596e-01
1.39175546e+00 -2.34573984e+00 -2.52030045e-01 4.12484169e-01
4.39813942e-01 6.91049933e-01 1.24989867e-01 1.71917126e-01
-2.94600576e-01 4.55622286e-01 -1.94481879e-01 1.34919733e-01
-6.16953552e-01 3.28026935e-02 1.83967222e-02 5.25475383e-01
7.04856887e-02 1.09711194e+00 -4.17478710e-01 -3.94288450e-01
1.83590755e-01 8.22632134e-01 4.80767816e-01 8.79215226e-02
5.31019092e-01 6.87345147e-01 -2.01731175e-01 5.54536939e-01
4.20123100e-01 -3.50776017e-01 4.21899408e-01 -3.23397875e-01
2.57615894e-01 -1.21601280e-02 -1.03131247e+00 1.70499837e+00
-1.60937816e-01 8.12749743e-01 -2.85857141e-01 -8.49440932e-01
1.01559854e+00 1.04820561e+00 5.44296622e-01 -4.59863096e-01
4.27557558e-01 3.26383412e-01 5.54781735e-01 -8.66810977e-01
-5.38710475e-01 -9.18132886e-02 3.70635688e-01 3.90731663e-01
3.78306434e-02 3.01784098e-01 -3.26937586e-01 -2.50131071e-01
1.11912489e+00 -1.69032365e-01 2.91954756e-01 -2.51695484e-01
5.62143564e-01 -3.67679238e-01 5.19165218e-01 9.43561196e-01
-3.68250549e-01 8.09400558e-01 3.93280804e-01 -1.07747054e+00
-6.67675197e-01 -9.12248254e-01 -2.61929274e-01 1.59198478e-01
-4.68849659e-01 -1.50014877e-01 -6.73592925e-01 -6.09287262e-01
-2.28076622e-01 -1.26048610e-01 -3.96168470e-01 -1.04627885e-01
-7.39223778e-01 -8.65054905e-01 1.47174287e+00 9.00096118e-01
5.78673184e-01 -1.37224960e+00 -1.14850688e+00 4.47056949e-01
-1.46314293e-01 -8.09166253e-01 2.28395499e-02 2.87717253e-01
-1.12371469e+00 -1.14052165e+00 -1.06532311e+00 -1.03703904e+00
1.57868698e-01 -5.01234055e-01 1.20735157e+00 3.75115335e-01
-8.63771260e-01 8.50172192e-02 -9.11951438e-02 -7.13845193e-01
-2.28370786e-01 -6.74401298e-02 -4.92946133e-02 9.98763815e-02
1.81654423e-01 -6.17371857e-01 -8.20526958e-01 -1.36401564e-01
-2.85407871e-01 -3.46353650e-01 6.05167925e-01 8.22142601e-01
3.78784269e-01 -1.64369315e-01 1.05171001e+00 -6.52441740e-01
1.02744341e+00 -4.82873581e-02 2.09363438e-02 2.37184539e-02
-5.88700235e-01 -4.97364521e-01 5.68153620e-01 -7.05643147e-02
-4.12089914e-01 1.89453602e-01 -4.28807199e-01 -4.61491436e-01
-3.75261456e-01 6.09428465e-01 2.74954081e-01 -1.67183727e-01
6.89312756e-01 1.38158247e-01 -4.46543992e-02 -4.00003821e-01
-3.81655931e-01 9.67610955e-01 6.13292098e-01 -7.60789076e-03
9.86840501e-02 2.10046932e-01 6.61415011e-02 -7.32635796e-01
-2.85535038e-01 -1.91111863e-01 -7.07332313e-01 -3.85527313e-01
1.04832304e+00 -9.15059090e-01 -5.87272763e-01 8.04391384e-01
-1.34778631e+00 -1.94311649e-01 -2.45879889e-01 6.85240030e-01
-1.00314662e-01 3.79795015e-01 -9.56139743e-01 -9.99632835e-01
-9.85407650e-01 -6.29120886e-01 5.97171426e-01 -7.56641477e-02
-4.39152718e-01 -1.05391276e+00 1.99053198e-01 -1.45397872e-01
6.53479934e-01 8.94200683e-01 8.21940243e-01 -5.28055489e-01
8.70273560e-02 -3.61028999e-01 -1.32072315e-01 6.24021709e-01
2.80184031e-01 8.99534747e-02 -1.34159756e+00 -3.80837530e-01
3.78261805e-01 -8.26071054e-02 8.14815879e-01 9.69930649e-01
1.25670505e+00 7.97414184e-02 -1.23629048e-01 6.64926648e-01
1.23684132e+00 6.02758288e-01 1.00922823e+00 -2.12316457e-02
6.65772259e-01 -4.70563173e-02 5.14225662e-03 3.78666222e-01
-1.08866438e-01 2.34902829e-01 5.93178980e-02 -9.30635273e-01
-2.53911614e-01 5.42282224e-01 2.10486457e-01 7.54984498e-01
-4.50718880e-01 1.23335995e-01 -1.24336493e+00 4.85047758e-01
-1.67926824e+00 -8.63343596e-01 -5.86573184e-01 1.94485545e+00
6.16137743e-01 8.77835155e-02 9.20556560e-02 8.81185353e-01
7.81332135e-01 -2.26352930e-01 -2.95682639e-01 -7.14003682e-01
-1.89896345e-01 7.31772661e-01 1.55726168e-02 7.96832293e-02
-1.27641368e+00 4.29476425e-02 7.13708496e+00 3.86360109e-01
-1.70793223e+00 1.67859435e-01 9.09715056e-01 6.35008365e-02
5.36415756e-01 -7.29147375e-01 -2.77999528e-02 3.23725224e-01
1.42456114e+00 1.16196491e-01 -2.36811131e-01 3.31787527e-01
3.87696743e-01 3.50809485e-01 -9.21908259e-01 1.39643824e+00
4.16399771e-03 -1.30401492e+00 -2.03062132e-01 -2.91079342e-01
2.64519304e-01 9.52858552e-02 6.42646849e-02 8.57889093e-03
-7.20654964e-01 -1.43061852e+00 2.97645628e-01 8.96435380e-01
1.17281175e+00 -6.59287095e-01 1.39867139e+00 1.71349257e-01
-1.23188424e+00 -2.37062365e-01 1.40641868e-01 -3.55031967e-01
-1.06249943e-01 7.34766126e-01 -9.70172882e-01 5.56258500e-01
1.05665970e+00 9.29427803e-01 -5.80927730e-01 1.28503454e+00
7.43996352e-03 1.00273478e+00 1.84498690e-02 2.79551208e-01
-8.19014311e-02 4.28217977e-01 5.47085166e-01 1.38161826e+00
3.89017910e-01 4.02423963e-02 2.45042890e-01 6.51066184e-01
1.80864722e-01 5.80532178e-02 -7.26873338e-01 7.12258816e-02
5.95807657e-03 1.28091502e+00 -7.95718968e-01 -6.63130939e-01
-9.82659459e-02 6.09832585e-01 -4.79588568e-01 3.63245159e-01
-8.02786052e-01 -8.28888595e-01 -3.34652141e-02 4.21083458e-02
-6.53016195e-02 1.14443839e-01 -1.00416827e+00 -7.02102125e-01
1.77659094e-01 -9.99804676e-01 4.72296417e-01 -7.04309940e-01
-1.09510362e+00 1.15072381e+00 -6.57419741e-01 -1.45137525e+00
-2.26208493e-01 -4.36468691e-01 -8.97118747e-01 1.20486259e+00
-1.21417725e+00 -7.87525177e-01 -4.04932112e-01 5.81182837e-01
1.45878047e-01 -2.23636389e-01 1.36577964e+00 6.52190089e-01
-4.70097065e-01 4.38014448e-01 -4.32617009e-01 5.20953178e-01
5.71969986e-01 -1.43866587e+00 2.23312229e-01 7.45750010e-01
2.10743994e-01 4.01947320e-01 2.91999340e-01 -4.47570562e-01
-6.81959510e-01 -1.06948829e+00 9.16420341e-01 -1.07800581e-01
-1.83328047e-01 2.02904552e-01 -9.21417832e-01 2.30308697e-01
4.02341068e-01 2.13816926e-01 8.44923675e-01 -1.16388574e-01
1.35068089e-01 -2.25057214e-01 -1.03895378e+00 6.35315627e-02
4.51582134e-01 -6.51479363e-01 -5.80543339e-01 3.39727178e-02
-1.81625172e-01 -3.60295027e-01 -1.15249407e+00 9.44571853e-01
9.05550122e-01 -1.06245613e+00 9.55654323e-01 -5.79614460e-01
6.42499328e-02 -2.36882836e-01 3.81884396e-01 -1.16229057e+00
-2.81798422e-01 -4.97265667e-01 -1.57834217e-01 6.71258152e-01
6.02958083e-01 -6.74161553e-01 5.13914227e-01 -2.19760656e-01
-1.74087018e-01 -1.08581543e+00 -6.57655299e-01 -5.22552311e-01
-2.11115688e-01 -3.58547717e-01 1.37835279e-01 8.08058143e-01
-1.90568820e-01 3.08346987e-01 -5.20904243e-01 1.08498380e-01
2.71768093e-01 -2.80561037e-02 2.04970449e-01 -1.50244868e+00
-2.12346479e-01 -2.59063035e-01 -7.84685612e-01 -9.66241732e-02
-4.42643166e-01 -8.18845272e-01 -2.92230457e-01 -1.68339694e+00
1.09585106e-01 -2.32381254e-01 -1.10946488e+00 7.10644782e-01
-1.37527943e-01 7.88576305e-01 -1.50327504e-01 5.76125383e-02
-1.92016691e-01 -2.28040308e-01 1.03884196e+00 -8.77122581e-02
-4.54773724e-01 1.67664096e-01 -3.49762440e-01 9.43635106e-01
1.16761971e+00 -5.54611504e-01 -1.33817643e-01 -2.70106465e-01
-7.71449804e-02 3.14013571e-01 5.84358573e-01 -1.62722397e+00
-7.26300701e-02 9.05665755e-01 1.17246985e+00 -5.72577178e-01
1.09152816e-01 -4.57487285e-01 3.20308834e-01 1.23081875e+00
-3.47118586e-01 3.58962923e-01 3.96021038e-01 1.85822636e-01
-3.97581995e-01 2.41608232e-01 6.15556419e-01 -2.60768592e-01
-1.71459049e-01 2.12222859e-02 -7.26472497e-01 -1.78509027e-01
7.38691986e-01 -4.03143823e-01 9.11141858e-02 -2.21049160e-01
-1.25383663e+00 -1.31133839e-01 -4.54437882e-01 1.23470567e-01
1.07114065e+00 -1.23376727e+00 -1.01408029e+00 4.01369870e-01
-2.48470351e-01 -5.71431994e-01 3.20789009e-01 1.44099760e+00
-1.01323748e+00 4.43430036e-01 -7.64132380e-01 -1.05572426e+00
-1.78672290e+00 -8.51181298e-02 9.19861495e-01 -3.14494908e-01
-9.24012065e-01 7.47271538e-01 -5.66045761e-01 1.41530946e-01
5.73003650e-01 -6.59986436e-01 -6.21065080e-01 -1.28870413e-01
5.85957646e-01 5.31242013e-01 4.60301101e-01 -1.77347213e-01
-5.43417633e-01 7.25326955e-01 3.44928175e-01 1.74294978e-01
1.07361484e+00 1.16508082e-01 -5.21439090e-02 6.07147634e-01
9.31359887e-01 -2.85531431e-01 -3.83340508e-01 2.46989101e-01
-7.95364473e-03 -4.76476178e-02 2.12139651e-01 -1.24996865e+00
-1.31101561e+00 1.33020759e+00 1.57341492e+00 4.77197558e-01
1.50362849e+00 -6.17101014e-01 1.03877223e+00 3.31061155e-01
6.96076900e-02 -7.06063390e-01 -1.08428262e-01 1.01563171e-01
5.69963098e-01 -8.43924701e-01 -3.41955483e-01 1.79087609e-01
-5.93028784e-01 1.62590063e+00 1.18982434e-01 -2.47244760e-01
9.57305908e-01 5.09618759e-01 8.20247293e-01 -3.64286005e-01
-5.54684460e-01 -2.02800781e-02 4.36074913e-01 5.84055245e-01
8.15469563e-01 7.22903162e-02 -4.18429762e-01 7.90550888e-01
2.77292401e-01 5.43436587e-01 3.21432292e-01 9.63156819e-01
-2.36058876e-01 -9.30797458e-01 -3.16538513e-01 6.85941517e-01
-1.26508200e+00 -8.59710351e-02 -5.18721402e-01 3.84096205e-01
5.06350160e-01 9.52726543e-01 -2.19974875e-01 -4.28425461e-01
1.07734837e-01 4.37014490e-01 4.71474558e-01 -5.18716455e-01
-1.26481092e+00 2.51294464e-01 8.75954628e-02 -3.98682892e-01
-3.39741707e-01 -1.85269639e-01 -1.19632566e+00 3.82427365e-01
-2.34962702e-01 4.06893417e-02 5.37277341e-01 8.38236213e-01
6.14597797e-01 1.28503299e+00 4.10474151e-01 -5.66388249e-01
-2.65019417e-01 -1.20652103e+00 -8.94148469e-01 1.31471902e-01
6.70543611e-01 -7.11538792e-02 -1.68962821e-01 3.12824339e-01]
|
[14.311387062072754, 3.2938904762268066]
|
a8b9297c-c97e-46d2-9031-3089ac4986a2
|
hydraplus-net-attentive-deep-features-for
|
1709.09930
| null |
http://arxiv.org/abs/1709.09930v1
|
http://arxiv.org/pdf/1709.09930v1.pdf
|
HydraPlus-Net: Attentive Deep Features for Pedestrian Analysis
|
Pedestrian analysis plays a vital role in intelligent video surveillance and
is a key component for security-centric computer vision systems. Despite that
the convolutional neural networks are remarkable in learning discriminative
features from images, the learning of comprehensive features of pedestrians for
fine-grained tasks remains an open problem. In this study, we propose a new
attention-based deep neural network, named as HydraPlus-Net (HP-net), that
multi-directionally feeds the multi-level attention maps to different feature
layers. The attentive deep features learned from the proposed HP-net bring
unique advantages: (1) the model is capable of capturing multiple attentions
from low-level to semantic-level, and (2) it explores the multi-scale
selectiveness of attentive features to enrich the final feature representations
for a pedestrian image. We demonstrate the effectiveness and generality of the
proposed HP-net for pedestrian analysis on two tasks, i.e. pedestrian attribute
recognition and person re-identification. Intensive experimental results have
been provided to prove that the HP-net outperforms the state-of-the-art methods
on various datasets.
|
['Xiaogang Wang', 'Haiyu Zhao', 'Lu Sheng', 'Xihui Liu', 'Jing Shao', 'Shuai Yi', 'Maoqing Tian', 'Junjie Yan']
|
2017-09-28
|
hydraplus-net-attentive-deep-features-for-1
|
http://openaccess.thecvf.com/content_iccv_2017/html/Liu_HydraPlus-Net_Attentive_Deep_ICCV_2017_paper.html
|
http://openaccess.thecvf.com/content_ICCV_2017/papers/Liu_HydraPlus-Net_Attentive_Deep_ICCV_2017_paper.pdf
|
iccv-2017-10
|
['pedestrian-attribute-recognition']
|
['computer-vision']
|
[-2.86089003e-01 -5.06071985e-01 1.08689994e-01 -5.75924456e-01
-2.55659282e-01 -9.67396982e-03 7.36939549e-01 3.43439984e-03
-6.64440572e-01 5.44245124e-01 4.57068264e-01 -2.98400316e-02
1.85725372e-02 -8.89763236e-01 -6.57548308e-01 -8.64896536e-01
-1.11514494e-01 -8.79366845e-02 4.97510165e-01 -3.18355173e-01
-7.22936839e-02 5.30583978e-01 -1.70805573e+00 2.92966753e-01
6.56298697e-01 1.27476680e+00 5.91727607e-02 5.98354101e-01
3.36982936e-01 7.15479374e-01 -3.68721873e-01 -8.02009463e-01
1.98972374e-01 -3.33376899e-02 -6.79186344e-01 2.09273160e-01
5.07789552e-01 -5.21874249e-01 -7.81657159e-01 1.22377992e+00
5.25078595e-01 2.93607831e-01 4.30278480e-01 -1.38325727e+00
-9.76926565e-01 7.47424960e-02 -5.97794712e-01 6.97104752e-01
2.15260357e-01 5.67140520e-01 7.64841974e-01 -8.70952129e-01
9.95199475e-03 1.65146649e+00 7.36489296e-01 5.81686616e-01
-6.27614260e-01 -5.28632045e-01 5.57208002e-01 8.85624051e-01
-1.40531313e+00 -3.17314148e-01 5.44571996e-01 -4.41085607e-01
6.70654058e-01 2.99979120e-01 8.94963741e-01 1.11556613e+00
1.54687345e-01 1.01443493e+00 7.94241905e-01 -3.42031494e-02
-2.36320257e-01 1.13290630e-01 5.56737423e-01 8.48473012e-01
4.13599521e-01 1.52896509e-01 -2.34127745e-01 1.26461536e-01
7.52474666e-01 5.03200769e-01 -1.78053066e-01 -1.05944768e-01
-1.12859571e+00 7.26014256e-01 1.01743186e+00 7.56695867e-02
-5.34961283e-01 -2.44665258e-02 7.06351697e-01 -3.50211114e-02
2.06921130e-01 1.27839953e-01 -2.53355652e-01 1.81783095e-01
-3.72962236e-01 2.00822011e-01 1.33791432e-01 9.19620991e-01
6.41706645e-01 6.46785498e-02 -6.28004670e-01 7.02050209e-01
4.23794657e-01 3.98900688e-01 6.34542882e-01 -5.01170397e-01
2.74621397e-01 7.52568007e-01 8.05812478e-02 -1.09537637e+00
-3.94878775e-01 -5.41350543e-01 -1.25247180e+00 1.26992658e-01
2.86945075e-01 -1.62397414e-01 -7.10757554e-01 1.73661673e+00
3.66150260e-01 1.87791035e-01 5.02588227e-03 1.21362627e+00
1.34749973e+00 6.59627557e-01 6.87933743e-01 2.50971198e-01
1.79479873e+00 -1.21008670e+00 -3.73507321e-01 -1.87490597e-01
-2.79021300e-02 -4.31408226e-01 8.02597821e-01 -1.31363094e-01
-6.67945623e-01 -1.21561420e+00 -9.20114338e-01 -1.19347624e-01
-6.02997363e-01 2.26649404e-01 5.93032837e-01 6.82490587e-01
-9.71367002e-01 2.40421966e-01 -3.92319113e-01 -7.95812249e-01
9.13129151e-01 3.85441303e-01 -5.21503568e-01 -1.49743900e-01
-1.38692212e+00 7.00121701e-01 3.35970551e-01 3.78997803e-01
-1.03484285e+00 -3.62398326e-01 -8.21085393e-01 3.99494290e-01
1.42278343e-01 -9.73230600e-01 8.61125767e-01 -1.04736090e+00
-1.17379749e+00 9.17581677e-01 -1.95203841e-01 -4.49927121e-01
3.91699493e-01 -3.84385943e-01 -4.04539734e-01 5.43463863e-02
3.15324545e-01 7.13063300e-01 7.78490365e-01 -9.66199815e-01
-1.16051018e+00 -4.06226724e-01 3.62554044e-01 7.40276501e-02
-7.55524695e-01 2.85501480e-01 -4.18063045e-01 -7.25135386e-01
-4.50907856e-01 -4.72017586e-01 -3.09645623e-01 1.05522476e-01
-3.66490990e-01 -5.18153608e-01 9.12152886e-01 -6.74731493e-01
8.75000596e-01 -2.04041529e+00 -7.71476142e-03 -1.60308808e-01
3.89126152e-01 8.15956056e-01 -1.75547108e-01 1.99356377e-02
-6.04964122e-02 -1.87173396e-01 1.04748040e-01 -1.93620294e-01
8.37257132e-03 -2.07650021e-01 3.31385713e-03 4.75843698e-01
4.05179620e-01 1.14666665e+00 -9.20425892e-01 -5.01887083e-01
6.12471879e-01 6.17852032e-01 -3.85322899e-01 3.96498293e-01
2.27491945e-01 4.68732387e-01 -6.64312661e-01 8.14768195e-01
7.31348932e-01 -1.94622532e-01 -3.41666430e-01 -5.12117147e-01
-2.81959504e-01 -4.45504248e-01 -7.58054495e-01 1.14311028e+00
1.51450597e-02 8.03623378e-01 -1.20654352e-01 -1.10221016e+00
7.64823675e-01 1.05017632e-01 1.50826037e-01 -7.88092911e-01
4.07498151e-01 -1.62802592e-01 2.47949939e-02 -6.58361375e-01
6.03113413e-01 2.49884158e-01 -2.24614322e-01 -6.08610138e-02
2.40676716e-01 9.48688149e-01 8.99186730e-02 6.05366752e-03
6.34132743e-01 -2.68510908e-01 5.14033973e-01 -6.50415540e-01
1.30132711e+00 -4.80149120e-01 7.42602587e-01 8.11585546e-01
-8.43842745e-01 1.86379373e-01 1.73945606e-01 -1.15911567e+00
-9.92306530e-01 -8.69606137e-01 -1.32328153e-01 1.53476965e+00
4.40797150e-01 -1.41275138e-01 -9.76870418e-01 -7.07591593e-01
2.40081660e-02 1.57331169e-01 -8.16540241e-01 -3.73491466e-01
-5.08806288e-01 -9.58387017e-01 4.68863249e-01 9.71886218e-01
1.24139607e+00 -1.17052209e+00 -6.12808943e-01 -2.54292525e-02
-1.98463485e-01 -1.27414560e+00 -5.91716588e-01 -3.03851366e-01
-1.44465670e-01 -1.31292164e+00 -9.46150243e-01 -1.08583307e+00
5.91316402e-01 6.48008645e-01 1.02230585e+00 2.19939381e-01
-3.65585983e-01 5.18214047e-01 -2.77302772e-01 -4.41155821e-01
1.84315100e-01 7.35173747e-02 1.76499471e-01 7.04969823e-01
8.63883734e-01 -1.52544335e-01 -9.04185057e-01 4.27191257e-01
-4.88125384e-01 -1.45338327e-01 7.28430986e-01 1.03781891e+00
3.66239786e-01 7.57998824e-02 7.98368633e-01 -4.88438934e-01
4.97862756e-01 -4.28842157e-01 -4.80520040e-01 2.50328451e-01
1.96810409e-01 -2.36721292e-01 8.31101537e-01 -1.66815713e-01
-1.13297582e+00 -4.12722230e-02 -4.15613085e-01 -1.17384210e-01
-6.01849377e-01 1.01475045e-02 -8.48630965e-01 -2.55009145e-01
2.80490100e-01 3.80307645e-01 -2.59145975e-01 -2.87266791e-01
2.48539791e-01 7.18959928e-01 8.25338781e-01 -5.26520014e-01
6.49008512e-01 3.94022822e-01 -4.60822880e-02 -1.06543541e+00
-9.32126522e-01 -5.07755101e-01 -8.24126124e-01 -3.87811631e-01
1.25427878e+00 -1.10746813e+00 -1.13392961e+00 9.93201971e-01
-1.18940330e+00 1.11825995e-01 -1.59408614e-01 2.61799574e-01
-2.26970702e-01 4.90679562e-01 -5.62941015e-01 -7.17117786e-01
-5.12703538e-01 -1.33373833e+00 1.02269554e+00 8.31917346e-01
2.36790538e-01 -8.90216589e-01 -4.62986022e-01 2.97434479e-01
3.72220218e-01 1.33768320e-01 5.63524365e-01 -5.31943679e-01
-7.30980158e-01 -2.60005891e-01 -8.93219471e-01 1.66435048e-01
8.22878852e-02 -8.93234611e-02 -1.20400059e+00 -4.76031601e-01
-3.58221114e-01 -2.77548730e-01 1.21660149e+00 5.95604181e-01
1.42975819e+00 -3.85812938e-01 -3.98226887e-01 7.81335592e-01
1.00620246e+00 -5.35378344e-02 7.42670476e-01 7.88977504e-01
8.76528800e-01 5.40659904e-01 4.62907761e-01 5.79773128e-01
6.58278108e-01 7.04783261e-01 6.28713787e-01 -4.09215361e-01
-1.74377605e-01 -6.35246634e-02 7.05343410e-02 2.58560985e-01
-4.04754966e-01 -8.04146826e-02 -5.67956269e-01 6.74528360e-01
-2.10720754e+00 -1.26629770e+00 -4.53611501e-02 2.00930357e+00
1.68145031e-01 -4.08853265e-03 5.81070960e-01 -8.58582780e-02
1.14730883e+00 1.40298054e-01 -5.08031428e-01 -1.09123833e-01
-1.46844625e-01 -2.85274833e-01 2.38701805e-01 1.28219023e-01
-1.87430573e+00 9.11898971e-01 5.10033560e+00 6.96188986e-01
-8.46179962e-01 1.07763605e-02 8.58726382e-01 3.03811014e-01
4.55499202e-01 -6.71169281e-01 -1.07037818e+00 8.59629869e-01
5.70182383e-01 -7.03562722e-02 -1.58871427e-01 1.02720761e+00
-7.16632977e-02 3.35685462e-01 -9.63604450e-01 1.04474008e+00
8.15352723e-02 -1.11064029e+00 5.05727112e-01 -1.23106793e-01
5.21858394e-01 -2.95710385e-01 3.79637212e-01 4.78400856e-01
2.50746548e-01 -9.77232933e-01 6.76561117e-01 7.26233840e-01
6.19455993e-01 -1.11948740e+00 1.22402394e+00 6.69113845e-02
-1.70935571e+00 -4.67432767e-01 -7.91685402e-01 -2.95618158e-02
6.62580132e-02 7.23727047e-02 -2.25082815e-01 5.86852431e-01
1.14624369e+00 9.96360719e-01 -9.58806038e-01 1.45264089e+00
-4.13399488e-02 1.71405494e-01 1.68338895e-01 -1.49371967e-01
4.97718543e-01 1.43667161e-01 4.65408385e-01 1.59355831e+00
1.29531026e-01 2.46941343e-01 2.84041822e-01 5.61701715e-01
-1.01583935e-01 -2.10417300e-01 -3.67362440e-01 4.29358661e-01
2.84344733e-01 1.48258185e+00 -3.92048359e-01 -4.05282736e-01
-6.63582802e-01 9.55393553e-01 3.28708380e-01 3.48128945e-01
-8.36782336e-01 -4.65745300e-01 1.21573675e+00 6.16712403e-03
6.97728395e-01 7.17925057e-02 1.59431577e-01 -1.08161807e+00
-1.39429778e-01 -7.35769570e-01 6.08563900e-01 -4.69185263e-01
-1.76169682e+00 9.53355730e-01 -2.44117677e-01 -1.09282434e+00
1.42628118e-01 -8.98397624e-01 -8.01346064e-01 9.76500511e-01
-1.70829022e+00 -1.70535839e+00 -9.47748661e-01 1.02870715e+00
7.13499486e-01 -6.54551566e-01 6.89851522e-01 5.52224874e-01
-9.45704877e-01 1.00343895e+00 -1.38858899e-01 6.92282677e-01
4.90742028e-01 -9.88922894e-01 6.25783324e-01 1.14333594e+00
-4.32112455e-01 6.46221399e-01 3.43387216e-01 -3.39265019e-01
-1.14244616e+00 -1.62942290e+00 5.45248628e-01 -3.32854599e-01
3.52178752e-01 -1.32868022e-01 -7.92266190e-01 5.96256733e-01
3.84336770e-01 4.11184490e-01 7.39223957e-01 1.35896998e-02
-3.70086461e-01 -4.30080503e-01 -1.10159492e+00 3.64848286e-01
9.37126935e-01 -3.38008761e-01 -4.55536813e-01 1.60479933e-01
5.87446094e-01 1.36905769e-02 -5.94264925e-01 2.41187528e-01
6.71834767e-01 -1.09514511e+00 1.45448947e+00 -9.98751938e-01
2.46491820e-01 -3.41060758e-01 -2.47222811e-01 -1.14883339e+00
-1.09658611e+00 -1.12496309e-01 4.25129011e-02 1.31028175e+00
-2.45895624e-01 -6.90680265e-01 4.60473210e-01 4.77428705e-01
-2.92850941e-01 -6.78771317e-01 -9.04231787e-01 -5.77612638e-01
-6.45625740e-02 1.23848446e-01 8.93102169e-01 6.72948599e-01
-4.73798573e-01 4.20569509e-01 -6.64006412e-01 4.57742244e-01
1.05097294e+00 -1.11597717e-01 8.22361588e-01 -1.47398579e+00
3.97014692e-02 -6.06744170e-01 -1.09071910e+00 -1.22548926e+00
1.15386561e-01 -4.31363404e-01 -1.93588793e-01 -1.36205983e+00
7.22365022e-01 -1.86752647e-01 -6.09321773e-01 2.24489227e-01
-6.86047137e-01 2.70896792e-01 2.65064180e-01 -5.20152226e-02
-1.18173540e+00 8.29292238e-01 1.00809085e+00 -4.77059215e-01
2.40298018e-01 1.70617819e-01 -9.86348450e-01 8.68915200e-01
5.53689241e-01 1.60012633e-01 -2.05262508e-02 -4.72668231e-01
-5.56263924e-01 -4.60716635e-01 9.33877587e-01 -1.36436820e+00
4.53795373e-01 6.01125211e-02 9.94785428e-01 -5.75179875e-01
3.03328723e-01 -7.72206306e-01 -4.01986569e-01 5.39996564e-01
-1.56678647e-01 9.94802266e-02 2.59821683e-01 6.78000987e-01
-2.54165679e-01 3.80990617e-02 9.87100542e-01 -2.42379069e-01
-1.48723650e+00 8.75205278e-01 -3.51611882e-01 -6.02834634e-02
1.16532242e+00 -2.19257295e-01 -4.58790153e-01 -1.53434351e-01
-4.36145186e-01 2.92866230e-01 2.01900288e-01 5.63831806e-01
8.14725935e-01 -1.69527996e+00 -9.05671477e-01 3.66851628e-01
2.46804535e-01 -2.69427866e-01 6.20117068e-01 4.68851328e-01
-1.92638159e-01 6.85298324e-01 -7.76345074e-01 -7.24052846e-01
-1.50308633e+00 8.63572717e-01 4.07884240e-01 -4.89287786e-02
-6.06854141e-01 1.00888050e+00 8.24641049e-01 -1.04060590e-01
3.30884844e-01 1.54178619e-01 -6.88346922e-01 -1.73770756e-01
9.67697859e-01 4.72580761e-01 -2.66200781e-01 -1.19108248e+00
-5.51094651e-01 5.64237893e-01 -3.00034910e-01 6.79617167e-01
1.33576190e+00 -3.03343117e-01 1.32299475e-02 -2.09153265e-01
1.10185468e+00 -6.63328588e-01 -1.61551642e+00 -4.15074706e-01
-3.45391601e-01 -6.63479030e-01 -1.38654292e-01 -3.98899287e-01
-1.16859770e+00 1.15988827e+00 8.85355711e-01 1.04798615e-01
1.09983492e+00 -1.49426877e-01 9.69027996e-01 4.53454226e-01
2.38861248e-01 -8.10727596e-01 2.66414344e-01 4.43948120e-01
7.25027204e-01 -1.59021187e+00 -4.58820164e-02 -2.16445610e-01
-6.38504565e-01 1.03263557e+00 1.07132316e+00 -5.52831180e-02
4.52270567e-01 -3.09879094e-01 -2.41831601e-01 -2.86237001e-02
-2.48865813e-01 -5.04428267e-01 4.36100334e-01 1.01717079e+00
2.93753505e-01 1.39210582e-01 3.02317917e-01 1.06524098e+00
1.19543709e-01 -2.52322882e-01 7.26905540e-02 5.29838145e-01
-6.99394882e-01 -6.67366028e-01 -5.32531440e-01 3.31728905e-01
-2.13527784e-01 3.77934948e-02 -2.78852224e-01 6.15401566e-01
4.16180730e-01 9.16191816e-01 9.44168195e-02 -5.89178205e-01
4.13880736e-01 -4.67320085e-01 3.05078030e-01 -9.74696279e-02
-6.10811591e-01 -5.01405478e-01 -1.06415480e-01 -4.64168727e-01
-5.03695011e-01 -6.75466120e-01 -5.62664509e-01 -5.81484199e-01
1.15416506e-02 5.44947013e-02 -3.74374464e-02 9.46963847e-01
3.65091205e-01 8.22913468e-01 6.26482606e-01 -1.08880341e+00
-2.79773057e-01 -8.55379224e-01 -2.86209047e-01 6.29749775e-01
5.09090483e-01 -9.94095922e-01 2.40022764e-01 6.70074970e-02]
|
[14.523003578186035, 0.9606629610061646]
|
240769e2-71ef-4b05-a243-c508c4a1a899
|
reciprocal-feature-learning-via-explicit-and
|
2105.06229
| null |
https://arxiv.org/abs/2105.06229v2
|
https://arxiv.org/pdf/2105.06229v2.pdf
|
Reciprocal Feature Learning via Explicit and Implicit Tasks in Scene Text Recognition
|
Text recognition is a popular topic for its broad applications. In this work, we excavate the implicit task, character counting within the traditional text recognition, without additional labor annotation cost. The implicit task plays as an auxiliary branch for complementing the sequential recognition. We design a two-branch reciprocal feature learning framework in order to adequately utilize the features from both the tasks. Through exploiting the complementary effect between explicit and implicit tasks, the feature is reliably enhanced. Extensive experiments on 7 benchmarks show the advantages of the proposed methods in both text recognition and the new-built character counting tasks. In addition, it is convenient yet effective to equip with variable networks and tasks. We offer abundant ablation studies, generalizing experiments with deeper understanding on the tasks. Code is available.
|
['Wenming Tan', 'Fei Wu', 'Wenqi Ren', 'Yi Niu', 'ShiLiang Pu', 'Zhanzhan Cheng', 'Yunlu Xu', 'Hui Jiang']
|
2021-05-13
| null | null | null | null |
['scene-text-recognition']
|
['computer-vision']
|
[ 4.76671100e-01 -3.57410789e-01 -3.44489068e-01 -5.30725837e-01
-4.18791234e-01 -4.45645809e-01 7.63501048e-01 -6.58077821e-02
-6.81972563e-01 8.24331820e-01 9.94786061e-03 -3.66928309e-01
1.00712273e-02 -5.64949989e-01 -2.30374604e-01 -8.65316570e-01
3.15460414e-01 2.60532290e-01 2.62017697e-01 1.35438532e-01
6.67650819e-01 2.91534066e-01 -1.24532557e+00 2.94787139e-01
9.60309386e-01 8.78170907e-01 2.45473102e-01 5.20868421e-01
-7.06837058e-01 1.02296698e+00 -3.66368830e-01 -6.55480206e-01
-1.27612380e-02 -6.45681620e-02 -6.19966209e-01 -2.90252250e-02
1.53591380e-01 -4.30607021e-01 -5.53931415e-01 8.94370198e-01
3.47682595e-01 1.51577527e-02 6.07902884e-01 -1.12941146e+00
-7.28062749e-01 7.93433189e-01 -8.24181080e-01 2.95130163e-01
2.74970204e-01 1.90149739e-01 1.16393769e+00 -1.15047503e+00
2.19511706e-02 9.73108053e-01 8.02545846e-01 5.87684393e-01
-9.33129013e-01 -8.15345287e-01 5.31898022e-01 1.31961569e-01
-1.19779813e+00 -7.32147038e-01 5.68144500e-01 -3.76775891e-01
9.67576861e-01 2.87774563e-01 2.70893514e-01 1.17531526e+00
-5.11739373e-01 1.36353135e+00 9.32117045e-01 -5.74700654e-01
-2.28377104e-01 2.11275861e-01 9.28222477e-01 9.05222774e-01
5.42585254e-01 -9.31058377e-02 -5.09125531e-01 -1.40161559e-01
6.64503634e-01 3.50385636e-01 -1.94895193e-01 -1.75219700e-02
-1.19785476e+00 6.56612933e-01 1.24697657e-02 4.07605439e-01
3.16808462e-01 3.99028987e-01 6.89876497e-01 1.66777134e-01
3.93511564e-01 2.52538949e-01 -5.45187891e-01 -4.26470637e-01
-8.93960893e-01 -2.08443388e-01 9.15404975e-01 1.28405225e+00
7.42944360e-01 1.97370246e-01 -5.58922172e-01 7.39486992e-01
2.49228403e-01 2.40862533e-01 6.25145495e-01 -3.85007054e-01
5.93034327e-01 9.00501847e-01 -9.44281220e-02 -8.67112815e-01
-1.92026719e-01 -5.08274257e-01 -1.04404807e+00 -2.86093742e-01
5.07855058e-01 3.03033297e-03 -9.42299247e-01 1.55277431e+00
-6.19752752e-03 3.60915750e-01 -5.62919140e-01 5.91484904e-01
6.49481177e-01 4.14155394e-01 1.24206133e-01 -7.58388266e-02
1.38598228e+00 -1.29419601e+00 -8.39792907e-01 -1.75580487e-01
8.38661671e-01 -5.24765134e-01 1.19502115e+00 2.61415958e-01
-5.63006699e-01 -3.22943956e-01 -1.08131480e+00 -2.65583366e-01
-4.78309691e-01 5.97187757e-01 1.13696432e+00 9.44297433e-01
-5.66196680e-01 4.94146198e-01 -8.90325785e-01 -2.36132160e-01
5.17646968e-01 5.30057192e-01 -1.86717197e-01 2.05823138e-01
-1.02459037e+00 5.70214331e-01 3.93165827e-01 4.48355943e-01
-2.90486395e-01 -5.80053777e-02 -7.74346888e-01 2.45015368e-01
6.10454857e-01 -4.52088952e-01 1.28553617e+00 -9.20783818e-01
-1.54568517e+00 7.21779525e-01 -3.81865412e-01 -2.45044231e-01
7.37456501e-01 -3.64476681e-01 -1.19236715e-01 -1.65176138e-01
-2.58804928e-03 5.50837219e-02 9.69985485e-01 -7.99272835e-01
-6.19048595e-01 -3.59874904e-01 -1.45815760e-01 1.29364803e-01
-8.98459852e-01 -3.95830683e-02 -8.57183337e-01 -8.05628777e-01
1.06276795e-01 -4.67471659e-01 1.42116100e-03 6.12423532e-02
-5.62246680e-01 -3.87444168e-01 8.97798419e-01 -5.69465816e-01
1.42710888e+00 -2.02622151e+00 -1.47942185e-01 5.32411002e-02
5.43610394e-01 3.99820924e-01 -4.86106016e-02 2.18384340e-01
1.64049059e-01 2.71743387e-01 -3.83956164e-01 -7.80436218e-01
2.40763351e-01 1.81815684e-01 -4.35435265e-01 4.68385190e-01
2.87875444e-01 1.05730069e+00 -6.93656027e-01 -6.66191459e-01
7.72047043e-03 -8.04680660e-02 -1.87070340e-01 6.64439099e-03
3.10300812e-02 7.96419457e-02 -9.20361578e-01 8.92728627e-01
6.17129385e-01 -5.00053883e-01 2.70059913e-01 -8.86564851e-02
-2.37270802e-01 3.48330796e-01 -1.05606937e+00 1.42319536e+00
-1.65135533e-01 4.95148301e-01 -4.59864996e-02 -1.35070193e+00
9.73694623e-01 2.41773315e-02 1.16454057e-01 -7.03319848e-01
2.91060805e-01 3.49723905e-01 -1.04869559e-01 -4.62131888e-01
5.56463897e-01 2.31497422e-01 -9.75536183e-02 5.91469705e-01
3.42040025e-02 3.57728809e-01 2.82826602e-01 2.21904933e-01
9.44410145e-01 2.22175404e-01 5.98675430e-01 -2.43548825e-01
8.39851320e-01 -3.59418094e-01 5.50152659e-01 1.08231664e+00
-3.35295558e-01 2.61158973e-01 5.56815743e-01 -4.30547565e-01
-8.35472643e-01 -5.02453148e-01 -2.75397539e-01 1.28635836e+00
8.90777335e-02 -4.31293249e-01 -4.65821296e-01 -9.18371081e-01
1.70177713e-01 3.62455785e-01 -9.24188137e-01 6.12919889e-02
-7.66268790e-01 -7.26392686e-01 9.76950228e-01 1.04693508e+00
7.44371474e-01 -9.21389103e-01 -1.02746382e-01 -8.13343748e-02
-3.05445716e-02 -9.43313122e-01 -7.66812384e-01 4.72850829e-01
-9.62292433e-01 -9.91693258e-01 -6.85908794e-01 -8.94645631e-01
4.55743074e-01 5.29837966e-01 9.03233945e-01 6.02598369e-01
-4.01122987e-01 2.30271503e-01 -5.05958617e-01 -5.33879101e-02
1.62139937e-01 5.30396879e-01 -7.52237737e-02 1.43266134e-02
7.17027009e-01 -3.92830461e-01 -2.22563654e-01 3.46506119e-01
-6.67960048e-01 2.15896711e-01 8.46305907e-01 1.36933827e+00
1.61976665e-01 -3.29466641e-01 5.34562528e-01 -1.17070556e+00
5.19236028e-01 -4.23814118e-01 -5.13041615e-01 3.75367612e-01
-7.41920173e-01 1.51071385e-01 5.25830865e-01 -5.46857953e-01
-1.12643576e+00 1.45705119e-01 2.46585496e-02 5.86370341e-02
4.31492627e-02 6.06501639e-01 -2.85150111e-01 -5.16650714e-02
1.77268282e-01 7.73252606e-01 -3.34239334e-01 -6.42610252e-01
1.46206930e-01 9.04646218e-01 3.12977016e-01 -1.04749072e+00
8.35315108e-01 2.71667331e-01 -3.00315946e-01 -7.76035368e-01
-8.02239478e-01 -5.69808304e-01 -7.49117792e-01 2.99022317e-01
6.15446806e-01 -6.74791157e-01 -9.28768158e-01 7.14533806e-01
-1.18101394e+00 -2.83831269e-01 2.62743048e-02 2.72577792e-01
-2.73099810e-01 9.79577184e-01 -7.32111633e-01 -1.09894252e+00
-5.43358624e-01 -1.02867281e+00 9.95782018e-01 2.39334285e-01
9.56511945e-02 -9.12279010e-01 -4.79003638e-02 4.09592539e-02
3.67988944e-01 -3.79770130e-01 9.08135295e-01 -1.08601022e+00
-6.34909809e-01 -2.98208237e-01 -7.59973288e-01 2.06440091e-01
1.64587289e-01 1.30868956e-01 -1.06536150e+00 -1.92243934e-01
-1.43963352e-01 -5.66120446e-01 1.28352010e+00 -1.04219951e-01
1.61452055e+00 -9.86536443e-02 -3.95776719e-01 8.90125036e-01
1.11169493e+00 7.66617805e-02 6.38980925e-01 4.07061785e-01
7.62818575e-01 5.22107482e-01 4.49213058e-01 5.24247169e-01
1.20584153e-01 5.96863866e-01 -1.33984789e-01 -3.09939161e-02
1.95422962e-01 -3.19428504e-01 2.09025443e-01 1.05933523e+00
-1.09365985e-01 -2.34493181e-01 -9.96126771e-01 6.46992326e-02
-2.17987490e+00 -9.45878983e-01 -3.57510030e-01 1.94950795e+00
1.05147815e+00 2.59642273e-01 -7.79416040e-02 2.90465951e-01
1.02738512e+00 1.43975258e-01 -5.50529003e-01 -1.88145712e-01
-8.21226686e-02 2.12375119e-01 2.80204684e-01 2.13066801e-01
-1.26716065e+00 9.88751054e-01 6.32524681e+00 1.16116917e+00
-7.91408181e-01 -1.36229366e-01 7.16843545e-01 2.60987580e-01
-6.53470755e-02 4.33027446e-02 -1.14004183e+00 5.45039892e-01
4.14835811e-01 -1.51419729e-01 3.04419875e-01 8.77227664e-01
-1.62458256e-01 -4.83702011e-02 -1.41420269e+00 1.14437926e+00
-2.68893205e-02 -1.21799529e+00 1.77047625e-01 -7.85423964e-02
3.15677196e-01 -3.25328588e-01 9.86944586e-02 7.00751007e-01
5.36748171e-02 -1.08821654e+00 7.46784806e-01 4.98843610e-01
1.22541475e+00 -4.21416640e-01 7.83451676e-01 5.05907834e-01
-1.57661259e+00 -1.40655115e-01 -3.43145490e-01 -3.71366769e-01
-1.53588444e-01 5.10545313e-01 -3.74258548e-01 4.95069116e-01
3.36952925e-01 8.87886226e-01 -7.47733653e-01 9.59019482e-01
-2.62146890e-01 6.99003458e-01 -2.01922998e-01 -3.77315998e-01
1.64964169e-01 -2.89293468e-01 2.28061482e-01 1.70882404e+00
2.27803066e-02 -5.49654709e-03 2.04516649e-01 9.32084203e-01
-2.29158312e-01 1.51306808e-01 -5.28054237e-01 -4.88876641e-01
5.36892295e-01 1.39998269e+00 -8.41121495e-01 -4.58255947e-01
-6.30295277e-01 9.88201857e-01 5.88509202e-01 3.06717366e-01
-9.36357379e-01 -7.52575219e-01 1.06824450e-01 -4.74457294e-01
2.33715534e-01 -4.76959735e-01 -6.65736079e-01 -1.55385017e+00
1.46212041e-01 -8.00459445e-01 2.65513182e-01 -2.57417768e-01
-1.59578705e+00 2.15282217e-01 -3.94532591e-01 -9.54518855e-01
1.90363020e-01 -9.05871630e-01 -8.13198686e-01 1.13395584e+00
-1.58363676e+00 -1.09153163e+00 -5.49432695e-01 6.42790735e-01
7.80995071e-01 -2.09794417e-01 7.65388548e-01 4.70700622e-01
-1.08070838e+00 1.22141814e+00 1.12363726e-01 6.45273447e-01
6.67985022e-01 -1.15017962e+00 5.39492369e-01 7.40904748e-01
4.34589572e-02 9.75537896e-01 6.97468743e-02 -6.27331674e-01
-1.50276685e+00 -8.49034309e-01 6.23805165e-01 -3.99258018e-01
8.65348041e-01 -6.80226624e-01 -9.34549272e-01 7.59876013e-01
-1.47582382e-01 -5.80291189e-02 7.97554851e-01 3.98497522e-01
-6.01237655e-01 9.41104293e-02 -7.60949850e-01 6.06496990e-01
1.18183565e+00 -6.85185075e-01 -6.74205601e-01 -2.33494386e-04
7.18575001e-01 -2.48348862e-02 -2.99403220e-01 2.12949321e-01
8.49071622e-01 -6.60522580e-01 6.82219565e-01 -6.36054456e-01
4.47744578e-01 -1.92693602e-02 -1.65389746e-01 -4.51385051e-01
-6.18856490e-01 -5.54287851e-01 -5.70406139e-01 1.56993926e+00
4.09142435e-01 -1.01675260e+00 1.04589999e+00 5.93428910e-01
-1.32888645e-01 -6.97470069e-01 -7.15360045e-01 -8.54253650e-01
6.04108311e-02 -2.83271253e-01 4.99309272e-01 1.21429658e+00
1.17105544e-01 4.83992875e-01 -4.53681409e-01 -2.61038423e-01
3.98955286e-01 7.68407509e-02 7.95934737e-01 -1.25880921e+00
-5.59551656e-01 -6.62166476e-01 -1.52495310e-01 -1.83645844e+00
2.43896335e-01 -1.03853524e+00 1.25640389e-02 -1.00663722e+00
8.65309596e-01 -6.83688700e-01 -4.44532216e-01 5.94557941e-01
-4.38921809e-01 3.21120955e-03 4.94543202e-02 4.34112847e-01
-7.69303441e-01 8.56110930e-01 1.04403102e+00 -3.24775547e-01
1.71601120e-02 9.10218582e-02 -8.22196662e-01 7.19675958e-01
9.60860312e-01 -4.30499762e-01 -2.53990203e-01 -7.32660651e-01
2.22520068e-01 -2.29480624e-01 1.43203944e-01 -6.72575891e-01
4.82648671e-01 -2.25327462e-01 4.71748501e-01 -6.55916750e-01
1.59333050e-01 -7.21517205e-01 -5.93757689e-01 3.97312582e-01
-4.00577217e-01 -6.32203743e-02 1.49461269e-01 9.34901714e-01
-1.00848623e-01 -6.66575909e-01 3.66669893e-01 -8.47153515e-02
-7.07270384e-01 5.23245215e-01 -2.40305632e-01 -2.79840399e-02
6.55031502e-01 -5.81121624e-01 -5.32303333e-01 4.22222987e-02
-5.13593614e-01 3.05486977e-01 2.90328324e-01 1.14235021e-01
5.93556046e-01 -1.08431947e+00 -4.68643874e-01 2.11985841e-01
1.04024909e-01 -2.28812918e-01 -7.51490146e-02 1.14912069e+00
-3.73589039e-01 6.52975917e-01 9.44023281e-02 -5.26163161e-01
-9.25860941e-01 4.67192262e-01 1.98101848e-01 -5.85676730e-01
-5.45861006e-01 7.11607873e-01 3.00797611e-01 -3.86604995e-01
5.44425786e-01 -1.97330400e-01 -2.85572052e-01 -3.32039855e-02
6.67512894e-01 4.92935836e-01 -1.33341864e-01 -2.83462666e-02
-2.60343313e-01 3.34815919e-01 -4.39841479e-01 8.18966329e-02
1.23048007e+00 -1.42681032e-01 -1.93222225e-01 6.01790905e-01
9.61274207e-01 1.46093383e-01 -1.02182734e+00 -4.56329644e-01
4.13812131e-01 -5.48826456e-01 -1.71451747e-01 -6.44511819e-01
-8.91760707e-01 1.13858736e+00 2.02260837e-01 1.78405419e-01
1.05979252e+00 -5.61357915e-01 5.29394686e-01 9.87616420e-01
4.07049894e-01 -9.78968203e-01 9.70638767e-02 9.90855873e-01
3.62601638e-01 -1.27178407e+00 2.16245964e-01 -5.55999637e-01
-3.18506628e-01 1.49585938e+00 9.31446135e-01 6.78959023e-03
4.13674951e-01 5.54212809e-01 -4.93940264e-01 -1.26928642e-01
-7.55290985e-01 -9.10020545e-02 -4.39888658e-03 4.24455941e-01
7.79659867e-01 -6.31186441e-02 -4.95193690e-01 1.11633217e+00
4.45436805e-01 -1.57892518e-02 2.47412786e-01 1.05777550e+00
-5.20349562e-01 -1.16819108e+00 -7.13737635e-03 7.85074651e-01
-2.53369540e-01 -5.21219432e-01 -4.72615898e-01 8.82289290e-01
-4.69954126e-02 5.67080438e-01 9.66931600e-03 -3.84222656e-01
-7.17731044e-02 2.44268984e-01 4.31760788e-01 -7.62545347e-01
-4.69904035e-01 -2.69593626e-01 1.40298247e-01 -3.35087776e-01
-1.33391678e-01 -5.71054876e-01 -1.14379299e+00 -4.27056193e-01
-8.99165034e-01 7.33528063e-02 4.67367321e-01 9.91753876e-01
1.25326350e-01 4.34949040e-01 5.11709750e-01 -6.47839546e-01
-1.05547202e+00 -1.08847308e+00 -6.72431767e-01 4.81582806e-02
6.36378750e-02 -7.84489930e-01 -4.35096085e-01 -7.74747431e-02]
|
[11.903766632080078, 2.1768527030944824]
|
2cad59ab-f23b-463d-8ab6-1c5ccf2972a1
|
rethinking-clustering-based-pseudo-labeling
|
2209.13635
| null |
https://arxiv.org/abs/2209.13635v1
|
https://arxiv.org/pdf/2209.13635v1.pdf
|
Rethinking Clustering-Based Pseudo-Labeling for Unsupervised Meta-Learning
|
The pioneering method for unsupervised meta-learning, CACTUs, is a clustering-based approach with pseudo-labeling. This approach is model-agnostic and can be combined with supervised algorithms to learn from unlabeled data. However, it often suffers from label inconsistency or limited diversity, which leads to poor performance. In this work, we prove that the core reason for this is lack of a clustering-friendly property in the embedding space. We address this by minimizing the inter- to intra-class similarity ratio to provide clustering-friendly embedding features, and validate our approach through comprehensive experiments. Note that, despite only utilizing a simple clustering algorithm (k-means) in our embedding space to obtain the pseudo-labels, we achieve significant improvement. Moreover, we adopt a progressive evaluation mechanism to obtain more diverse samples in order to further alleviate the limited diversity problem. Finally, our approach is also model-agnostic and can easily be integrated into existing supervised methods. To demonstrate its generalization ability, we integrate it into two representative algorithms: MAML and EP. The results on three main few-shot benchmarks clearly show that the proposed method achieves significant improvement compared to state-of-the-art models. Notably, our approach also outperforms the corresponding supervised method in two tasks.
|
['Ling Shao', 'Jianbing Shen', 'Xingping Dong']
|
2022-09-27
| null | null | null | null |
['unsupervised-few-shot-image-classification']
|
['computer-vision']
|
[-3.29695567e-02 -1.48017257e-01 -6.25641465e-01 -4.44318920e-01
-8.55990350e-01 -6.18075371e-01 5.82726836e-01 2.47528866e-01
-4.83990252e-01 4.87202764e-01 3.93082276e-02 -6.50658309e-02
-3.21125209e-01 -5.35457850e-01 -3.64192456e-01 -1.03017163e+00
1.35145351e-01 4.66649204e-01 1.11673594e-01 3.57913114e-02
1.99953139e-01 1.60747156e-01 -1.57994699e+00 1.44787103e-01
8.86230588e-01 8.09750617e-01 7.27157742e-02 3.65957096e-02
-2.93669641e-01 7.72185445e-01 -2.98431456e-01 -3.40611160e-01
1.76196665e-01 -4.30204988e-01 -8.69694054e-01 5.51727593e-01
1.24103807e-01 6.32763579e-02 -1.62853807e-01 1.06961715e+00
3.68478984e-01 1.12361208e-01 1.01809716e+00 -1.42283607e+00
-4.16012555e-01 6.93345308e-01 -7.12892711e-01 -2.23643646e-01
-1.49905205e-01 -6.92680553e-02 1.30425918e+00 -1.13138247e+00
5.15086472e-01 1.04703021e+00 6.36925101e-01 5.58642268e-01
-1.37022400e+00 -4.64456797e-01 2.29096234e-01 3.54363859e-01
-1.50415254e+00 -3.73778462e-01 1.09383833e+00 -3.70246947e-01
5.08453190e-01 2.27698326e-01 2.85219938e-01 1.07136488e+00
-3.42410386e-01 1.07629514e+00 1.29829407e+00 -6.83728099e-01
6.40713692e-01 5.32254338e-01 4.53098446e-01 7.69364357e-01
8.67591575e-02 1.20203979e-02 -2.27868617e-01 -1.99714333e-01
1.92577586e-01 2.47803822e-01 -1.89578265e-01 -8.46458077e-01
-1.27664685e+00 8.51702273e-01 5.51219285e-01 5.42190671e-01
9.51926261e-02 -5.85230514e-02 4.81711358e-01 2.19870552e-01
5.42070150e-01 5.22495270e-01 -3.90155911e-01 -7.14622736e-02
-1.00734663e+00 -2.73500025e-01 7.00528741e-01 1.01704121e+00
9.89762723e-01 -1.31937802e-01 -1.17014870e-01 1.05238962e+00
4.00696158e-01 -1.18374020e-01 8.39745343e-01 -1.10526264e+00
3.14859331e-01 8.77663612e-01 4.03640755e-02 -9.36313152e-01
-4.09011960e-01 -7.07694530e-01 -9.14863765e-01 -1.01601616e-01
2.77679086e-01 1.90558210e-01 -8.40008020e-01 1.65880466e+00
2.87506223e-01 3.67945850e-01 6.98775798e-02 6.98110163e-01
5.02996504e-01 3.96988034e-01 7.10762665e-02 -4.18886811e-01
1.18940926e+00 -1.65924597e+00 -5.95664024e-01 -1.27198011e-01
9.81907964e-01 -3.77735734e-01 1.18874860e+00 3.73004019e-01
-5.17505765e-01 -6.30038857e-01 -1.22472501e+00 4.11379725e-01
-6.10639691e-01 3.49332184e-01 6.50905848e-01 8.35998774e-01
-8.33169520e-01 7.65861034e-01 -8.29448938e-01 -4.87627357e-01
2.79559255e-01 1.11028984e-01 -3.51672471e-01 -4.99644950e-02
-9.86819744e-01 6.26121938e-01 7.50309408e-01 -2.66014397e-01
-6.38977826e-01 -5.35835385e-01 -7.62180090e-01 7.06176162e-02
6.55582249e-01 -3.88276041e-01 1.02405226e+00 -8.35622191e-01
-1.45911980e+00 7.29692996e-01 -2.08930522e-01 -2.96856940e-01
3.20501596e-01 5.08613102e-02 -4.96642470e-01 2.30239168e-01
-3.52060944e-02 6.73852205e-01 8.41746330e-01 -1.79740250e+00
-6.12475991e-01 -1.79350436e-01 -1.08119398e-02 5.90718314e-02
-1.13776064e+00 -2.89873183e-01 -7.19235599e-01 -7.81322956e-01
3.63764316e-01 -9.74030018e-01 -4.14040267e-01 -2.00346887e-01
-5.30275226e-01 -2.92574048e-01 8.23043525e-01 5.25722392e-02
1.47575462e+00 -2.28125906e+00 1.16013989e-01 1.97157860e-01
4.60528314e-01 2.11983755e-01 -2.04333246e-01 5.81181765e-01
-6.58638328e-02 3.37583691e-01 -6.13646805e-01 -7.43849397e-01
1.97548240e-01 3.38669151e-01 -1.91506684e-01 4.20827895e-01
1.79918647e-01 9.01189804e-01 -1.28892636e+00 -7.57814288e-01
3.20544690e-01 1.89237893e-01 -3.25052232e-01 9.04679075e-02
-1.45403072e-01 3.86940867e-01 -2.78805852e-01 8.38981271e-01
4.47067022e-01 -5.51466942e-01 3.92848492e-01 -3.46323609e-01
6.77054524e-02 1.86590217e-02 -1.25095415e+00 1.74702120e+00
-4.38764334e-01 3.12194705e-01 -2.86921471e-01 -1.56556189e+00
8.32344174e-01 1.67326957e-01 6.49142265e-01 -3.37926447e-01
2.00265467e-01 3.30361068e-01 -9.08840001e-02 -3.25211495e-01
3.47917557e-01 -2.70871390e-02 -7.34674409e-02 6.96304679e-01
2.05795586e-01 3.56067121e-01 3.90356064e-01 3.18826586e-01
9.16121483e-01 -6.26245793e-03 2.45360881e-01 -4.30873901e-01
5.92822313e-01 -9.68293101e-02 6.41193509e-01 8.09540510e-01
-2.62630731e-01 5.55787921e-01 2.43103862e-01 -1.29321694e-01
-8.07673573e-01 -1.07448113e+00 -2.10190177e-01 1.01452494e+00
3.80256057e-01 -8.39771628e-01 -6.58484817e-01 -1.16870999e+00
-1.69813991e-01 5.92738748e-01 -6.65612698e-01 -3.26698750e-01
-2.12653026e-01 -1.05086255e+00 4.25218970e-01 6.38169646e-01
4.51509774e-01 -6.95627213e-01 -1.89748883e-01 2.57031351e-01
-2.53263533e-01 -9.83631849e-01 -2.90523261e-01 3.83545876e-01
-1.05912578e+00 -1.01764357e+00 -5.15917420e-01 -9.22400713e-01
9.10518646e-01 4.54190135e-01 1.07852995e+00 1.72736123e-01
-4.78969254e-02 4.37942445e-01 -7.18026161e-01 -6.48757117e-03
-4.10812140e-01 4.18841898e-01 1.70862228e-01 2.22787291e-01
4.95053232e-01 -6.51389897e-01 -5.00430703e-01 6.11392975e-01
-9.45698023e-01 -9.47510004e-02 5.53375661e-01 1.20611441e+00
6.87731743e-01 4.53650475e-01 6.09475434e-01 -1.19150555e+00
3.10162246e-01 -5.22322237e-01 -2.74027228e-01 3.99010211e-01
-1.08084142e+00 2.00178221e-01 8.28944504e-01 -4.68867749e-01
-7.37568736e-01 1.64194018e-01 1.86213553e-01 -5.09672165e-01
-1.70260772e-01 6.47754252e-01 -3.24757874e-01 1.95607413e-02
5.04062355e-01 4.34753746e-01 -1.51931524e-01 -7.17136025e-01
6.13334179e-01 8.57965112e-01 3.65180790e-01 -6.34052873e-01
1.06167889e+00 7.15188503e-01 -1.58796579e-01 -5.70442915e-01
-1.08822298e+00 -8.55941892e-01 -7.93046236e-01 -5.73204905e-02
3.18336010e-01 -9.24833179e-01 -3.07155967e-01 2.07646027e-01
-6.58779144e-01 -1.85797930e-01 -3.11530739e-01 4.90072429e-01
-5.29564083e-01 6.81615233e-01 -5.81926465e-01 -7.68897772e-01
-1.13161534e-01 -1.14470387e+00 8.41733456e-01 -7.07534477e-02
-1.12126261e-01 -1.27239060e+00 9.11248028e-02 2.56200701e-01
2.36062825e-01 6.25785515e-02 9.97525573e-01 -9.58371520e-01
-2.21516952e-01 -1.53680503e-01 -1.03399053e-01 4.90405202e-01
4.49989676e-01 5.27853072e-02 -1.04967272e+00 -6.00787342e-01
2.14307327e-02 -5.58419108e-01 1.17130518e+00 -1.61905080e-01
1.44234157e+00 -1.26374573e-01 -5.41750193e-01 5.07912219e-01
1.54685807e+00 -7.44086653e-02 3.98359060e-01 5.51629722e-01
8.41143787e-01 6.72154903e-01 9.39702272e-01 3.90384406e-01
5.28670788e-01 7.78897762e-01 2.97560751e-01 -1.02890059e-01
-2.19973736e-02 -2.34894544e-01 3.55955869e-01 1.38693511e+00
1.52983293e-01 -1.04977667e-01 -7.42602289e-01 5.81999600e-01
-2.19450426e+00 -8.40002179e-01 -1.24210164e-01 2.10034227e+00
9.15346146e-01 1.51984155e-01 2.33549774e-01 6.21303797e-01
5.91144502e-01 1.70961782e-01 -4.82091129e-01 -2.43177898e-02
-7.31777623e-02 -2.02441085e-02 3.68119389e-01 2.30933607e-01
-1.32027006e+00 9.73119140e-01 6.03971386e+00 1.33939815e+00
-9.71051931e-01 2.51970917e-01 6.52462959e-01 1.85113519e-01
-2.50852108e-01 6.38582706e-02 -7.25109756e-01 6.36337996e-01
7.88970172e-01 -7.46155828e-02 2.69380331e-01 9.49415684e-01
-1.06090993e-01 1.44288093e-01 -1.37360573e+00 9.23488379e-01
1.01110466e-01 -1.10171700e+00 -6.93250820e-02 -4.54321876e-02
9.69128609e-01 -1.41608238e-01 6.18558154e-02 5.35173714e-01
2.06730515e-01 -7.47057080e-01 5.27227461e-01 1.68622941e-01
6.09946728e-01 -9.01644409e-01 6.62697196e-01 5.95179200e-01
-1.34513783e+00 -1.98589817e-01 -5.79065561e-01 1.36356056e-01
-1.14384793e-01 7.80781269e-01 -4.63136852e-01 8.11538279e-01
5.74831843e-01 9.67425406e-01 -8.54359567e-01 1.03141403e+00
-3.14594030e-01 8.55390370e-01 -1.52345747e-01 1.35352269e-01
3.85526419e-01 -1.86262012e-01 3.28782767e-01 1.24128222e+00
1.34399831e-01 -3.36016297e-01 4.25932109e-01 7.48352587e-01
-2.18694180e-01 2.75597572e-01 -3.92213911e-01 -7.48676136e-02
8.00443172e-01 1.54589939e+00 -9.71264005e-01 -6.09775186e-01
-3.48928034e-01 9.99853492e-01 5.52308798e-01 2.73720026e-01
-8.10647130e-01 -3.42784882e-01 2.18229890e-01 -1.39528051e-01
3.67455184e-01 -1.68456584e-01 -1.83711648e-01 -1.30496359e+00
-2.79171877e-02 -9.00923133e-01 4.39522117e-01 -2.52325594e-01
-1.64412904e+00 4.87144560e-01 -3.15707624e-02 -1.70909512e+00
-2.37569496e-01 -5.04249573e-01 -5.28611839e-01 2.63029814e-01
-1.66677010e+00 -1.07594121e+00 -2.83477724e-01 5.49189687e-01
5.74628651e-01 -2.85928786e-01 8.01909924e-01 4.78732646e-01
-8.14892530e-01 9.69149113e-01 5.74217677e-01 1.63475454e-01
9.03288126e-01 -1.48116636e+00 1.91295929e-02 6.63200617e-01
5.28788567e-01 6.51378632e-01 3.68879497e-01 -2.31757596e-01
-1.19913507e+00 -1.38890243e+00 7.23850548e-01 -5.89053929e-01
8.38070750e-01 -4.15601373e-01 -9.29367900e-01 4.65510517e-01
-6.15471974e-03 2.73979362e-02 1.12202168e+00 4.72972430e-02
-6.61035717e-01 -1.95844606e-01 -1.00862992e+00 6.43179655e-01
1.06571293e+00 -3.39015007e-01 -5.58500826e-01 3.43342513e-01
7.30695486e-01 1.29630879e-01 -9.05517161e-01 6.17023587e-01
2.89674968e-01 -1.04107869e+00 8.79929304e-01 -4.80308950e-01
4.42685157e-01 -5.16994715e-01 -1.69056118e-01 -1.37659168e+00
-4.54601109e-01 -2.94786304e-01 -4.36624885e-01 1.64572585e+00
4.33178574e-01 -6.07522964e-01 9.22593653e-01 2.74612933e-01
-4.17982861e-02 -1.00713861e+00 -6.98906958e-01 -1.30398560e+00
1.79459810e-01 -4.31816518e-01 5.47939658e-01 1.42868805e+00
2.32882455e-01 3.64447862e-01 -4.91671205e-01 3.18114646e-02
9.31756496e-01 3.32380444e-01 7.00105548e-01 -1.25800800e+00
-4.60590780e-01 -5.08702755e-01 -2.83742428e-01 -1.01538789e+00
3.44709635e-01 -1.04518771e+00 3.23416665e-02 -1.25279808e+00
4.25944269e-01 -7.79589295e-01 -9.23568010e-01 5.78581691e-01
-2.56721020e-01 4.82495308e-01 1.52488649e-01 6.17454588e-01
-9.95150387e-01 6.92284226e-01 8.21038485e-01 -3.85618836e-01
-2.60392725e-01 -2.70783365e-01 -7.66125619e-01 8.03174853e-01
9.45209861e-01 -7.55449891e-01 -5.69787025e-01 -1.47244245e-01
-7.64426291e-02 -5.36373675e-01 1.18474394e-01 -9.76371706e-01
3.15055251e-01 -3.79699003e-03 4.61774059e-02 -4.29550499e-01
2.63858616e-01 -1.00205612e+00 -5.48248030e-02 4.61031467e-01
-3.86688828e-01 -3.49414855e-01 -2.07517207e-01 7.66462982e-01
-4.03814435e-01 -4.51548725e-01 7.72677362e-01 1.81875676e-02
-8.09175134e-01 3.24214756e-01 -1.93019331e-01 -5.82404807e-02
1.04935169e+00 -1.68762848e-01 -1.77182227e-01 -1.47509664e-01
-5.71211040e-01 4.46964622e-01 7.56738305e-01 3.95029843e-01
3.90934110e-01 -1.65613532e+00 -4.72218812e-01 2.92756408e-03
6.81722999e-01 -2.39179164e-01 -5.09189330e-02 1.10158801e+00
-1.09762974e-01 2.19451085e-01 2.42173344e-01 -7.86034703e-01
-9.21772182e-01 9.74648595e-01 2.50551067e-02 -5.31349897e-01
-3.65692317e-01 4.66639519e-01 1.08503565e-01 -6.12074614e-01
4.47418898e-01 9.10867155e-02 -1.85526088e-01 1.85004413e-01
4.86020803e-01 4.57147449e-01 3.28453258e-04 -5.61237454e-01
-3.90603125e-01 6.65950060e-01 -1.66128784e-01 5.01930565e-02
1.21642053e+00 -2.61932224e-01 -3.37786525e-02 8.57379496e-01
1.38442326e+00 5.78097627e-03 -1.10955524e+00 -5.34088671e-01
4.14542079e-01 -3.56117070e-01 -4.93006259e-02 -5.79119325e-01
-1.24116528e+00 9.80680406e-01 4.70714390e-01 3.19478869e-01
1.07705855e+00 7.72928894e-02 5.83183825e-01 6.75748825e-01
4.46803749e-01 -1.28789616e+00 2.94929445e-01 2.19979227e-01
2.53894508e-01 -1.55072165e+00 5.15625477e-02 -5.29010773e-01
-5.79240441e-01 1.01197088e+00 6.27293587e-01 3.90254706e-02
6.80648685e-01 -3.97626460e-02 1.52240366e-01 -5.71903810e-02
-7.27964461e-01 -3.40675831e-01 2.91367114e-01 4.27119017e-01
3.34811836e-01 -5.45126759e-02 -3.94373953e-01 6.14727557e-01
1.06420174e-01 -2.71451771e-01 2.09327802e-01 9.41071153e-01
-4.63175565e-01 -1.47695291e+00 -4.14533913e-02 3.83590162e-01
-9.43383873e-02 -1.18727852e-02 -4.04991508e-01 7.88310528e-01
1.14876203e-01 1.11329460e+00 -1.31981388e-01 -6.52962565e-01
6.91625401e-02 1.84799343e-01 3.17160130e-01 -7.09220588e-01
-3.39255989e-01 1.61120251e-01 -2.34094888e-01 -4.09066737e-01
-8.43409538e-01 -1.95365757e-01 -1.06814742e+00 -6.30201548e-02
-5.52393794e-01 3.31469893e-01 3.82076740e-01 9.42311525e-01
3.59515280e-01 2.91556299e-01 1.14719498e+00 -6.53995037e-01
-7.21983731e-01 -9.27485347e-01 -7.09442139e-01 5.83525360e-01
-5.16217798e-02 -8.97116899e-01 -6.75567567e-01 1.90965645e-02]
|
[9.382707595825195, 3.151479959487915]
|
c3567c66-b2d2-4131-a151-954ff18ed73f
|
video-description-a-survey-of-methods
|
1806.00186
| null |
https://arxiv.org/abs/1806.00186v4
|
https://arxiv.org/pdf/1806.00186v4.pdf
|
Video Description: A Survey of Methods, Datasets and Evaluation Metrics
|
Video description is the automatic generation of natural language sentences that describe the contents of a given video. It has applications in human-robot interaction, helping the visually impaired and video subtitling. The past few years have seen a surge of research in this area due to the unprecedented success of deep learning in computer vision and natural language processing. Numerous methods, datasets and evaluation metrics have been proposed in the literature, calling the need for a comprehensive survey to focus research efforts in this flourishing new direction. This paper fills the gap by surveying the state of the art approaches with a focus on deep learning models; comparing benchmark datasets in terms of their domains, number of classes, and repository size; and identifying the pros and cons of various evaluation metrics like SPICE, CIDEr, ROUGE, BLEU, METEOR, and WMD. Classical video description approaches combined subject, object and verb detection with template based language models to generate sentences. However, the release of large datasets revealed that these methods can not cope with the diversity in unconstrained open domain videos. Classical approaches were followed by a very short era of statistical methods which were soon replaced with deep learning, the current state of the art in video description. Our survey shows that despite the fast-paced developments, video description research is still in its infancy due to the following reasons. Analysis of video description models is challenging because it is difficult to ascertain the contributions, towards accuracy or errors, of the visual features and the adopted language model in the final description. Existing datasets neither contain adequate visual diversity nor complexity of linguistic structures. Finally, current evaluation metrics ...
|
['Ajmal Mian', 'Nayyer Aafaq', 'Wei Liu', 'Syed Zulqarnain Gilani', 'Mubarak Shah']
|
2018-06-01
| null | null | null | null |
['video-description']
|
['computer-vision']
|
[ 1.30308628e-01 -2.72608370e-01 -3.16692561e-01 -2.78040677e-01
-7.23197639e-01 -5.61031222e-01 1.06549180e+00 1.19217537e-01
-4.40117657e-01 8.31133306e-01 5.94251454e-01 2.56001294e-01
-1.57082662e-01 -2.73987859e-01 -2.92771429e-01 -5.84342241e-01
-1.40692383e-01 4.51164514e-01 2.44346216e-01 -2.49354675e-01
4.52049285e-01 3.98073226e-01 -2.09040666e+00 6.36514843e-01
1.35459453e-01 8.37510884e-01 4.13404584e-01 6.60268545e-01
-4.32034910e-01 1.12812936e+00 -5.98657489e-01 -4.18995589e-01
-7.35457242e-02 -5.56332648e-01 -7.95385718e-01 4.91844118e-01
6.08697951e-01 -3.52557749e-01 -5.64060509e-01 8.36156309e-01
7.96137333e-01 5.41749224e-02 6.91273093e-01 -1.36525285e+00
-6.12539887e-01 2.73445904e-01 -1.99733272e-01 2.52095342e-01
9.40261424e-01 1.61249548e-01 9.45073068e-01 -7.02776611e-01
1.11893594e+00 1.13738000e+00 6.84754372e-01 7.41901040e-01
-7.87118673e-01 -3.61373156e-01 -1.16815679e-01 6.18663788e-01
-1.48734856e+00 -4.33627188e-01 5.50441086e-01 -7.54160106e-01
1.22014189e+00 6.65075481e-02 6.93545163e-01 1.42536330e+00
3.64734209e-03 7.88350821e-01 8.67555797e-01 -5.78323245e-01
1.51065230e-01 4.22846317e-01 -1.62924100e-02 6.11486375e-01
1.26867935e-01 -7.36714080e-02 -6.51490688e-01 9.97710004e-02
4.76218164e-01 -3.12391520e-01 -2.69741833e-01 -6.10299945e-01
-1.33060992e+00 8.75335038e-01 -3.33289616e-02 5.77506125e-01
-2.32481167e-01 -2.56988667e-02 1.11964154e+00 2.05849558e-01
2.47169241e-01 1.53094202e-01 -2.33923629e-01 -4.25649703e-01
-1.09077322e+00 6.98120236e-01 7.52605140e-01 1.31116772e+00
3.81740242e-01 -1.53960541e-01 -4.12963122e-01 8.31080616e-01
-3.47216018e-02 2.23131061e-01 6.77675724e-01 -8.32615197e-01
3.40111583e-01 3.85701925e-01 8.91736522e-02 -9.75269198e-01
-6.89440966e-01 1.32864326e-01 -6.29463434e-01 1.14052795e-01
3.94347191e-01 8.66397470e-03 -7.52628386e-01 1.45875156e+00
-1.99027523e-01 -3.14165175e-01 5.41143753e-02 9.10675347e-01
1.38200891e+00 5.48396647e-01 1.40435219e-01 -2.37592652e-01
1.36346722e+00 -7.74836600e-01 -9.01437700e-01 -8.44430625e-02
5.46378434e-01 -8.65621030e-01 8.23994517e-01 2.16444656e-01
-1.00663459e+00 -5.78521848e-01 -8.99316669e-01 -1.50164261e-01
-6.21426165e-01 1.16919160e-01 5.11369646e-01 4.96753216e-01
-1.23234880e+00 1.97365955e-01 -3.59244138e-01 -8.59925807e-01
4.23671514e-01 2.97051996e-01 -6.62159860e-01 -3.02047819e-01
-1.08663309e+00 1.07009399e+00 6.49273753e-01 -2.61400908e-01
-7.39057899e-01 -2.64600873e-01 -1.01231921e+00 -2.30542600e-01
4.18979347e-01 -6.64603174e-01 1.13575149e+00 -1.16103232e+00
-1.12271404e+00 1.30968249e+00 -3.90804186e-02 -6.48858666e-01
5.49341261e-01 1.48879737e-01 -2.96851486e-01 2.90861845e-01
1.83166146e-01 1.02285528e+00 6.18824184e-01 -1.05867171e+00
-9.31342006e-01 -1.19002156e-01 2.86630332e-01 1.91629499e-01
-1.37811005e-01 4.00995910e-01 -5.68715155e-01 -6.08434916e-01
-2.90019661e-01 -8.89469624e-01 1.74660340e-01 -3.36556286e-02
-5.65822013e-02 -4.91385311e-01 6.06885076e-01 -5.56276321e-01
1.24541080e+00 -2.16211534e+00 2.36211613e-01 -4.18141186e-01
2.08976746e-01 3.35822344e-01 -8.62807557e-02 8.59527111e-01
4.75114398e-02 2.40424290e-01 -1.85822561e-01 -3.12683195e-01
1.64675608e-01 1.00987576e-01 -1.43094271e-01 4.84102786e-01
2.61386540e-02 5.49072087e-01 -9.03900623e-01 -6.82078779e-01
4.96842027e-01 7.91435480e-01 -2.95703262e-01 2.09162131e-01
-1.90536454e-01 6.53665736e-02 -1.68681681e-01 6.64783299e-01
2.51572162e-01 8.40380937e-02 -2.62960047e-02 -1.38114884e-01
-3.24403167e-01 1.31314039e-01 -1.06081736e+00 1.69598556e+00
-2.43953839e-01 1.34912622e+00 -2.57073313e-01 -9.26696658e-01
7.62175798e-01 6.23247623e-01 5.97855806e-01 -7.80243933e-01
1.97791845e-01 4.09775198e-01 -2.07907453e-01 -1.03463888e+00
5.19499540e-01 -4.83460985e-02 -6.52545318e-02 5.51819801e-02
2.11374789e-01 -1.24558762e-01 6.78283036e-01 1.56090222e-02
1.02848220e+00 3.72936279e-01 7.94924855e-01 1.16908289e-01
7.58330405e-01 3.58463615e-01 1.31420881e-01 6.72145605e-01
-5.42256176e-01 1.00563383e+00 5.30003428e-01 -7.92561710e-01
-1.27248192e+00 -7.70858943e-01 -1.25539526e-01 8.76274407e-01
6.07222766e-02 -6.06438100e-01 -8.46836329e-01 -3.85754287e-01
-3.02880436e-01 4.32677507e-01 -4.41916078e-01 7.38835931e-02
-4.87072974e-01 -4.49413538e-01 5.87864041e-01 3.33508253e-01
6.65131271e-01 -1.59819603e+00 -9.69100416e-01 9.80465710e-02
-3.47541839e-01 -1.54638553e+00 -6.48967922e-02 -3.03711772e-01
-5.84520698e-01 -1.10695362e+00 -9.61386740e-01 -1.05931854e+00
2.70041078e-01 2.18522429e-01 1.27426982e+00 -3.04627083e-02
-3.30021083e-01 6.86868727e-01 -7.35964835e-01 -3.21035892e-01
-5.37986755e-01 -1.23397209e-01 3.44165079e-02 -2.90826052e-01
8.01844418e-01 -4.65712622e-02 -3.35789382e-01 -6.00545965e-02
-1.02817976e+00 5.62239774e-02 5.14913619e-01 7.86917567e-01
4.07926977e-01 -7.11996555e-02 2.78846532e-01 -3.12469453e-01
6.89726651e-01 -4.20079917e-01 -2.70018965e-01 1.55931920e-01
-2.94625252e-01 -4.91406471e-02 1.61681011e-01 -3.25724125e-01
-6.82905972e-01 5.73092066e-02 -7.93868601e-02 -1.72458440e-01
-5.55757105e-01 5.53032577e-01 1.05154686e-01 -2.73445770e-02
5.17979026e-01 4.51089382e-01 4.41599153e-02 -1.81347072e-01
8.29200372e-02 7.97234893e-01 5.42286873e-01 1.94444414e-02
4.24701869e-01 4.68871117e-01 -7.64148682e-02 -1.19467902e+00
-6.13392472e-01 -6.37068272e-01 -6.71654403e-01 -5.28630674e-01
9.65459287e-01 -8.65393460e-01 -3.55657816e-01 5.01213312e-01
-1.52654433e+00 9.59881842e-02 -4.79131564e-02 5.75552404e-01
-9.24537420e-01 4.71853226e-01 -2.16649592e-01 -7.33986735e-01
-1.71388239e-01 -1.34244144e+00 9.71563280e-01 4.20116447e-02
-4.27666306e-01 -9.32852685e-01 -3.00033074e-02 5.23417950e-01
4.03804570e-01 4.71231699e-01 7.40781248e-01 -6.62269592e-01
-4.36852932e-01 -4.36104447e-01 -3.98233086e-01 3.12265992e-01
3.50447930e-02 -4.63951416e-02 -8.67479146e-01 -5.94247803e-02
-2.03590974e-01 -3.47791702e-01 6.81275964e-01 6.00058377e-01
7.71983922e-01 -2.01963037e-01 -3.51279639e-02 3.95488083e-01
1.51015925e+00 3.03626001e-01 7.14961588e-01 8.24397802e-01
4.71545607e-01 8.44129145e-01 5.52777827e-01 5.56387544e-01
4.65739936e-01 1.02107644e+00 3.90876651e-01 2.32773930e-01
-5.26611149e-01 1.36730000e-01 5.70489645e-01 3.10304135e-01
-2.19162107e-01 -4.12446529e-01 -1.13664556e+00 7.06827462e-01
-1.69288743e+00 -1.36268485e+00 -4.91386615e-02 2.19180632e+00
4.02980119e-01 1.64706215e-01 2.66182870e-01 2.31367424e-01
7.59193420e-01 2.01742291e-01 -7.03313649e-02 -4.47738796e-01
-3.07091892e-01 -4.12121773e-01 3.25384766e-01 1.80524975e-01
-1.23001015e+00 8.52689862e-01 6.39854431e+00 7.59054840e-01
-1.11440635e+00 -8.51219445e-02 2.04503641e-01 -2.98676528e-02
4.48746011e-02 -3.05719554e-01 -8.26103687e-01 4.49498385e-01
8.83068085e-01 -1.20181613e-01 3.72895658e-01 7.02056408e-01
5.15536845e-01 -2.46045932e-01 -1.16407359e+00 1.46745503e+00
8.46169829e-01 -1.52712762e+00 2.66678244e-01 -1.81771711e-01
4.64430869e-01 2.52124041e-01 -1.18247002e-01 2.78223783e-01
-4.70150977e-01 -9.45096314e-01 9.55066204e-01 3.78680527e-01
8.21724176e-01 -4.94854450e-01 1.06560886e+00 9.41487029e-02
-9.70676720e-01 -1.79015785e-01 -2.69693077e-01 -2.60968357e-01
2.72605777e-01 -5.03188856e-02 -9.13517594e-01 2.69557625e-01
7.62708008e-01 7.43883550e-01 -5.81752539e-01 1.33369052e+00
1.17431313e-01 2.18046829e-01 1.76912874e-01 -2.84620613e-01
5.39881706e-01 3.13364044e-02 7.68486679e-01 1.61642313e+00
2.24355415e-01 -2.16330796e-01 1.29207343e-01 5.17461598e-01
1.50076360e-01 3.20156246e-01 -9.46132839e-01 -2.49443784e-01
2.36425117e-01 8.83096635e-01 -8.50167215e-01 -2.51913637e-01
-8.16377521e-01 7.83899665e-01 -1.66085325e-02 1.37877241e-01
-8.23415160e-01 -2.01264322e-01 5.32978475e-01 2.75018632e-01
2.29482979e-01 -3.76885742e-01 -1.01169869e-01 -8.62965167e-01
2.09997654e-01 -9.95548964e-01 3.67382228e-01 -1.02120304e+00
-1.03103662e+00 9.16802347e-01 4.75282073e-01 -1.53660762e+00
-5.82622111e-01 -7.36285329e-01 1.18833333e-02 4.79930311e-01
-1.40707076e+00 -1.14549637e+00 -5.10465920e-01 4.11098719e-01
1.17103302e+00 -7.35846162e-01 8.37858200e-01 3.73374164e-01
-3.17266077e-01 2.33623818e-01 6.17881529e-02 1.43279746e-01
7.72870839e-01 -9.80529010e-01 1.17811397e-01 5.63348353e-01
2.51756370e-01 2.14497298e-01 1.06914854e+00 -3.84253085e-01
-1.04231691e+00 -6.09420538e-01 1.32174134e+00 -4.50356126e-01
6.70071423e-01 -2.75165975e-01 -4.86672461e-01 3.22243214e-01
3.48540336e-01 -8.57096538e-02 4.73812252e-01 -3.85612398e-01
-2.61512727e-01 9.28224530e-03 -9.92315948e-01 6.21591806e-01
8.88928950e-01 -6.14645004e-01 -5.75981379e-01 5.61933756e-01
4.59092349e-01 -3.66226703e-01 -4.82221782e-01 2.41731480e-01
7.30897903e-01 -1.18475628e+00 7.80843437e-01 -5.25914609e-01
5.64035833e-01 -5.53778298e-02 -3.08371782e-01 -7.39088833e-01
-1.20644882e-01 -5.35697758e-01 1.67623773e-01 1.32831633e+00
1.93339229e-01 1.36898905e-01 6.29400849e-01 5.15005589e-01
-2.10381657e-01 -6.04714692e-01 -9.15074289e-01 -6.49643838e-01
-2.67951578e-01 -5.09337068e-01 1.64401561e-01 5.97460449e-01
-1.13245994e-01 2.77950555e-01 -5.49176991e-01 -4.22006935e-01
3.51051360e-01 -2.73656577e-01 6.74508035e-01 -1.08897412e+00
2.51819223e-01 -6.56647027e-01 -1.12338507e+00 -4.72904623e-01
3.10797036e-01 -7.22122073e-01 -7.45878369e-02 -1.98726559e+00
4.88009125e-01 2.31328502e-01 2.07551226e-01 2.64069647e-01
3.97873640e-01 3.50444019e-01 3.36277246e-01 1.74715266e-01
-7.44304478e-01 1.70854777e-01 8.62795889e-01 -3.57257754e-01
-1.45752937e-01 -2.20331281e-01 -3.04817140e-01 8.05147409e-01
7.33011186e-01 -2.51992553e-01 -4.03665960e-01 -4.97110307e-01
2.72833288e-01 -2.00642675e-01 4.40822393e-01 -1.29403448e+00
1.45484522e-01 7.41692260e-02 1.48690090e-01 -4.74167764e-01
2.47252807e-01 -8.35296452e-01 5.64205572e-02 2.83228755e-01
-5.05590796e-01 2.67964900e-01 2.01628789e-01 3.39504629e-01
-5.10317981e-01 -3.97497743e-01 6.33549929e-01 -4.20381248e-01
-1.64443946e+00 1.01263449e-01 -5.94971716e-01 1.12072766e-01
1.04650259e+00 -6.99612081e-01 -2.79365946e-02 -5.69848597e-01
-6.36120439e-01 -1.35085642e-01 4.44375038e-01 8.98949444e-01
7.46899724e-01 -1.15778303e+00 -9.50558126e-01 -1.71516865e-01
5.28026104e-01 -3.79139870e-01 2.37320647e-01 6.99198008e-01
-8.47127020e-01 7.24910855e-01 -5.94970584e-01 -5.83314121e-01
-1.57148969e+00 6.57411933e-01 1.84253857e-01 3.23696472e-02
-8.00756156e-01 4.60471332e-01 -9.34943929e-02 1.55294612e-01
6.94409668e-01 -2.54177935e-02 -7.01272964e-01 3.38433504e-01
7.16613948e-01 4.22284395e-01 -3.52072753e-02 -1.22250414e+00
-4.30173159e-01 6.71887040e-01 9.17409137e-02 -6.23157434e-02
1.31098902e+00 -4.12558317e-01 2.81733703e-02 5.13955057e-01
1.25666344e+00 -3.90007049e-01 -8.78541708e-01 -8.45015887e-03
2.02965796e-01 -1.25298753e-01 1.15305059e-01 -6.94990814e-01
-7.39978850e-01 6.88696325e-01 7.72784054e-01 2.51270324e-01
1.08969223e+00 6.60504028e-02 5.94888747e-01 3.78586799e-01
4.75366592e-01 -1.22927868e+00 -1.23277064e-02 6.51804745e-01
1.17124474e+00 -1.49848700e+00 1.42403230e-01 -1.38122529e-01
-9.19166744e-01 1.30394006e+00 2.88474768e-01 1.30904332e-01
3.50690335e-01 1.74785689e-01 3.65597196e-02 -2.47899562e-01
-5.05893528e-01 -5.54191291e-01 2.77408779e-01 8.25009108e-01
7.06055641e-01 -2.29232833e-01 -7.39821851e-01 3.33755881e-01
-9.37255844e-02 2.17972830e-01 7.51083672e-01 7.70013511e-01
-4.67501312e-01 -1.02701485e+00 -3.14601809e-01 2.87865460e-01
-5.29010415e-01 6.94420189e-02 -5.40083289e-01 1.19950294e+00
3.81352514e-01 9.09738779e-01 -9.89848003e-02 -3.71308416e-01
3.32574934e-01 -2.14326587e-02 5.37764311e-01 -6.71703279e-01
-3.76497418e-01 -2.14694783e-01 3.16963911e-01 -4.59459275e-01
-9.83992815e-01 -9.40698445e-01 -1.10079622e+00 -7.68103749e-02
1.54057845e-01 -1.22232221e-01 8.08817983e-01 1.10125840e+00
2.11700257e-02 2.89753616e-01 1.35793597e-01 -1.07617784e+00
-3.41728270e-01 -9.47452962e-01 -4.46227223e-01 5.59223652e-01
4.95154649e-01 -8.11764359e-01 -4.30653363e-01 4.41791058e-01]
|
[10.577402114868164, 0.703292191028595]
|
cd090cd2-8128-41cf-8275-70e561420d89
|
mask-cnn-localizing-parts-and-selecting
|
1605.06878
| null |
http://arxiv.org/abs/1605.06878v1
|
http://arxiv.org/pdf/1605.06878v1.pdf
|
Mask-CNN: Localizing Parts and Selecting Descriptors for Fine-Grained Image Recognition
|
Fine-grained image recognition is a challenging computer vision problem, due
to the small inter-class variations caused by highly similar subordinate
categories, and the large intra-class variations in poses, scales and
rotations. In this paper, we propose a novel end-to-end Mask-CNN model without
the fully connected layers for fine-grained recognition. Based on the part
annotations of fine-grained images, the proposed model consists of a fully
convolutional network to both locate the discriminative parts (e.g., head and
torso), and more importantly generate object/part masks for selecting useful
and meaningful convolutional descriptors. After that, a four-stream Mask-CNN
model is built for aggregating the selected object- and part-level descriptors
simultaneously. The proposed Mask-CNN model has the smallest number of
parameters, lowest feature dimensionality and highest recognition accuracy when
compared with state-of-the-arts fine-grained approaches.
|
['Chen-Wei Xie', 'Xiu-Shen Wei', 'Jianxin Wu']
|
2016-05-23
| null | null | null | null |
['fine-grained-image-recognition']
|
['computer-vision']
|
[ 0.02622079 -0.23480462 -0.14804405 -0.5628798 -0.6910321 -0.4506332
0.52689284 0.01915467 -0.21490067 0.43192837 0.27019483 0.58224154
-0.35978687 -0.61532265 -0.5261732 -0.6524363 0.20408857 0.3785735
0.30043623 0.25268146 0.22725707 1.0663501 -1.920466 0.3742102
0.58236545 1.7369967 0.06054978 0.20253496 -0.03532903 0.3931024
-0.5952593 -0.2010454 0.25042158 -0.10670919 -0.3890932 0.44809547
1.0453433 -0.24600653 -0.44810662 0.85950667 0.6405282 0.16257973
0.62841517 -1.1152092 -0.4868382 0.17541851 -0.32324767 0.01232674
-0.04463375 0.05748911 0.70256895 -1.23146 0.24993555 1.3159221
0.5928466 0.5199454 -1.0475754 -0.9209598 0.4092749 0.10341641
-2.0605931 -0.597741 0.6783584 -0.38912424 0.88665533 0.35981524
0.4870725 0.82042354 0.14932042 0.34578523 0.9470841 -0.04071832
0.1831203 -0.27246606 -0.09595122 1.0149087 0.3786602 -0.01126375
-0.5502959 -0.06891032 0.88545877 0.49639058 0.10176772 -0.5559247
-1.2780018 0.5961263 0.6130437 0.3094508 -0.6379972 0.15659891
0.17529804 -0.15610714 0.3819723 0.0987405 -0.6445894 0.04545145
-1.226051 0.22145271 0.53595716 0.8964512 1.0165898 0.01466674
-0.8419841 0.98178416 0.2096902 0.27966562 0.67153674 -0.52437854
0.34082043 0.92384416 -0.03935854 -1.1614076 -0.46066102 -0.88209504
-1.1184663 0.00826871 0.17994916 0.19130246 -1.2357025 1.4898936
0.40578458 0.13626163 -0.5177857 0.9714792 1.0853044 0.16491832
0.1474259 0.13280326 1.5027317 -1.216693 -0.20348735 -0.23737581
0.15124486 -0.70967454 0.64771235 0.11637687 -0.82078505 -1.1218752
-0.9005327 -0.2578335 -0.48360658 0.64034855 0.4755675 0.46816143
-0.80842376 0.6247291 -0.6292966 -0.2611995 0.71704054 0.62498194
-0.63367146 -0.19465436 -0.5202369 0.56394213 0.49256158 0.3506445
-1.0858214 -0.5500105 -0.9891723 0.51196605 0.06068173 -0.71851295
0.8328686 -0.71935064 -1.3154325 0.90573007 -0.06198452 -0.0478104
0.40010533 -0.13695909 -0.2378211 0.03012712 0.16594402 0.8147803
1.0701233 -0.9669323 -0.72022605 -0.72409326 -0.14055994 0.02590825
-0.22984087 0.05503202 -0.80775696 -0.93102235 0.18059663 -0.80221003
-0.29526284 0.1894451 -0.43699408 -0.32920292 0.71646696 -0.32698575
1.136634 -2.3131955 -0.0052273 0.15131442 0.34754482 0.2337813
-0.3014872 0.21151106 -0.11313253 -0.1199306 0.22781187 -0.32030755
0.28317666 0.05315081 0.10074345 0.6608537 0.4403229 1.0185854
-0.45802322 -0.5702548 0.54817194 0.566498 -0.29966578 0.37463376
0.05584326 0.2562946 -0.7598114 0.99108505 0.86189073 -0.26816574
-0.20290788 -0.7312212 -0.11065682 -0.21619494 -1.3843633 1.6393104
-0.4333643 0.16527434 0.07104141 -0.7027941 1.1601868 -0.08518431
0.28407604 -0.88540334 0.31852123 0.28648642 -0.19655825 -0.20994663
0.29028955 -0.0177683 -0.42774227 0.11457545 0.3128046 0.18889739
0.00878429 -0.34198022 0.8182311 -0.12089413 0.5170773 -0.21152656
0.5244043 -0.29358092 0.59029293 0.65260535 -0.26002076 0.81929016
-0.18363269 -0.82456106 -0.66149074 -0.78284866 -0.06644762 1.2970806
0.35499454 -0.34670457 -0.85016876 -0.59173983 0.30527756 -0.04509035
-0.9048095 -0.31335282 -0.6051279 -0.20723756 0.6098708 0.7221716
0.67852116 -1.1866698 -0.5585334 0.1492336 0.09619782 -1.1062158
-0.8012189 0.3978103 -0.68921465 -1.0519512 -0.82749397 -1.0054516
0.7765077 0.43920326 1.0063576 0.05012806 -0.62864196 -0.11321994
-0.26444355 -0.12211833 0.38484153 0.18881066 -0.11083309 0.637484
0.332141 -0.3488901 -0.95544976 0.6135364 -0.7062457 -0.13387173
1.0216237 1.0792437 1.1227949 -0.02091949 0.20099156 -0.3367559
0.28533888 -0.04126233 -0.5308992 0.292529 -0.266717 0.12819861
0.7128591 -0.3792815 -0.66632324 0.4286971 -0.02965571 -0.7290084
-0.69806796 0.08206264 -0.37573275 -0.49817657 0.4216416 0.29536295
-0.4431812 -0.90944654 0.29798818 0.76061386 0.47555092 -0.48061553
0.7934723 0.4458507 0.01593019 -0.65083927 -1.0517001 -0.6694471
-1.035261 0.13532458 0.9052283 -1.1364257 -0.6460075 0.7084647
-0.9562554 -0.09179173 -0.40807784 0.35459957 -0.34991288 0.01137455
-0.33716345 -0.33278954 -0.5875414 -1.2158867 1.8006827 0.5437079
0.05162226 -0.45052564 -0.23152141 0.2676327 0.50175565 0.37778708
0.4254818 -0.59402245 -0.6651915 -0.4264337 -0.7136305 0.16373487
0.19751875 -0.23787253 -0.9291801 -0.47653285 -0.49127248 -0.3994991
1.0049969 0.33100978 1.5785881 -0.33788526 -0.4416587 1.0724707
1.323907 -0.1662312 0.26110542 0.06775709 1.0703694 0.44976482
0.71088576 0.83324414 0.2483563 0.9521458 0.47814226 -0.2005922
-0.49425697 -0.30878374 -0.06904268 0.5234254 0.04739236 -0.01243572
-0.4119295 0.55547696 -1.6185213 -0.85755664 0.28514707 1.9633036
0.44595027 -0.08549053 0.10336205 -0.22693004 0.8879032 0.41557255
-0.6702132 -0.19532849 -0.06446765 0.68436676 0.54539484 -0.01667912
-1.4102206 1.1292324 5.1966186 1.2869343 -1.32561 -0.02171
0.5579189 -0.07947959 0.2866932 -0.44844237 -1.1092077 0.28243467
0.51986223 0.27061495 0.29373503 1.0371841 0.02153581 0.43269226
-1.1239613 1.3542768 0.3968238 -1.2982125 0.21899252 0.00694606
0.73091626 0.05720675 -0.23447502 0.17648955 -0.10725745 -1.4009184
1.1516156 0.49828422 1.1971163 -0.81202483 0.7132599 0.26326072
-1.6686271 -0.2395676 -0.6077995 0.07300089 -0.4156642 0.50390136
-0.39225554 0.39731959 0.6845061 0.4565865 -0.78254336 1.2630063
0.07905414 0.14425957 -0.15082492 0.02623005 0.26572752 0.28591114
-0.1530068 1.5820965 0.35957518 -0.10293744 0.5812883 0.59616643
-0.2565666 -0.03105168 -0.09268083 0.13641717 0.41568115 1.6546717
-0.86659527 -0.33421102 -0.2559048 1.1823235 0.4531176 0.08906818
-0.66627294 -0.63315576 0.9508054 -0.0334621 1.0540055 -0.06686912
-0.1599846 -1.2198572 0.16484855 -0.9380142 0.4744232 -0.3480049
-1.3828703 0.9029431 -0.36884272 -1.1973286 -0.09049701 -0.6943593
-0.36215666 1.0556542 -1.4476703 -1.6779538 -1.0357865 0.9949247
0.63328654 0.01218927 1.0129305 0.43914756 -0.60362726 1.0716356
-0.10082106 0.47743478 0.60634387 -0.88483196 0.32782352 0.5408836
0.05664974 0.58696485 0.1326792 -0.47209787 -1.3733368 -1.7232455
0.8366415 -0.1266744 0.16816783 -0.6681294 -0.4656749 0.23094374
-0.364236 0.83889204 0.48270264 -0.03306475 -0.7812798 -0.6234443
-1.2776972 0.22313489 1.1737767 -0.61909676 -0.35676667 0.176642
0.57104594 -0.4466746 -0.8945586 0.48398295 1.0737848 -0.8628394
1.0474423 -0.45904383 0.10159256 -0.47955883 -0.3214004 -1.194736
-0.997319 -0.20798284 -0.11919745 1.0940794 0.0054334 -0.34527668
0.91832733 0.1932338 -0.4362601 -0.8408068 -1.054252 -0.68027
-0.30450603 -0.14959098 0.8505542 0.519892 -0.67476827 0.18872909
-0.29855338 0.21783309 0.69802827 0.75549906 0.7746315 -1.422073
-0.23602249 -0.50556296 -0.86133975 -1.0875791 -0.02702516 -0.6176939
0.36059904 -1.4153782 0.390972 -0.3547089 -0.675412 0.63475114
-0.06946339 0.88269573 0.33092725 0.22721608 -0.9359934 0.5564714
1.4205416 -0.31117257 0.08127002 -0.09005789 -0.8395248 0.55561596
0.561782 -0.35195938 -0.03238498 -0.31182453 -0.5777598 -0.1100527
0.6207629 -1.1429719 0.23423749 -0.26795685 1.1413627 -0.7405782
0.4755838 -0.76030785 0.23516537 0.40758756 -0.22394222 -0.08077332
0.30668667 0.28230473 -0.60385495 0.16745389 0.91961795 -0.27850902
-0.7895548 0.9332276 0.16229644 -0.01030328 1.079276 -0.50216573
-0.21154082 0.14950393 -0.7183559 -0.230323 0.35301915 0.58003765
0.6867028 -1.5448904 -0.67953044 0.5388086 0.62099713 0.07880203
0.7069211 0.4975036 -0.2773345 0.7301888 -0.5446232 -0.41709793
-1.3884138 0.4686184 0.40137228 -0.3112409 -0.31454232 1.076883
0.7065843 -0.36495915 0.40927815 -0.49603036 -0.18602045 0.18288887
0.6330835 0.25144365 0.37769103 -1.1299659 -0.80678064 1.0144113
-0.16410957 0.5986989 1.4339439 -0.03409049 0.01451503 -0.14840041
1.4135453 0.02119314 -1.3780518 -0.14307135 -0.34583712 -0.68642
0.01611523 -0.88284534 -1.0753572 0.884284 0.86123943 -0.12369534
1.2600392 0.24284275 0.7339835 0.03157057 0.5271118 -0.8948169
-0.04728448 0.54427 1.0848241 -1.0702863 -0.08172823 -0.3458245
-0.16419613 1.1439713 0.78476685 -0.3606825 0.7206171 -0.0903404
0.04008171 -0.3417513 -0.45986542 -0.38982996 0.79298234 0.67887133
0.3777949 0.3520538 0.02005316 0.75691104 -0.15982206 0.09413499
-0.35175094 0.5106804 -0.45160118 -0.842333 -0.36170855 0.61928624
-0.43738344 -0.05290981 -0.49854627 0.5323988 0.8555806 0.7702837
0.28960136 -0.6379556 0.6363508 -0.28171167 0.53798205 -0.7593519
-0.93720156 0.08727335 0.03240442 -1.0524035 -0.13800897 -0.39921203
-0.815952 -0.1577903 -0.31919727 -0.10669886 0.57046413 0.9781814
0.8976999 0.566663 0.6582678 -1.3319919 -0.67416304 -1.1223022
-0.73111194 0.5057977 0.35242215 -0.8572297 -0.06088838 -0.23281099]
|
[9.60496711730957, 2.0020620822906494]
|
3b1c72cd-114c-4a4c-9c84-1538a9c2742f
|
benchmarks-for-corruption-invariant-person-re
|
2111.00880
| null |
https://arxiv.org/abs/2111.00880v2
|
https://arxiv.org/pdf/2111.00880v2.pdf
|
Benchmarks for Corruption Invariant Person Re-identification
|
When deploying person re-identification (ReID) model in safety-critical applications, it is pivotal to understanding the robustness of the model against a diverse array of image corruptions. However, current evaluations of person ReID only consider the performance on clean datasets and ignore images in various corrupted scenarios. In this work, we comprehensively establish six ReID benchmarks for learning corruption invariant representation. In the field of ReID, we are the first to conduct an exhaustive study on corruption invariant learning in single- and cross-modality datasets, including Market-1501, CUHK03, MSMT17, RegDB, SYSU-MM01. After reproducing and examining the robustness performance of 21 recent ReID methods, we have some observations: 1) transformer-based models are more robust towards corrupted images, compared with CNN-based models, 2) increasing the probability of random erasing (a commonly used augmentation method) hurts model corruption robustness, 3) cross-dataset generalization improves with corruption robustness increases. By analyzing the above observations, we propose a strong baseline on both single- and cross-modality ReID datasets which achieves improved robustness against diverse corruptions. Our codes are available on https://github.com/MinghuiChen43/CIL-ReID.
|
['Feng Zheng', 'Zhiqiang Wang', 'Minghui Chen']
|
2021-11-01
| null | null | null | null |
['generalizable-person-re-identification']
|
['computer-vision']
|
[-1.24761432e-01 -5.19219160e-01 2.33712085e-02 -2.50914663e-01
-7.26304591e-01 -6.34564400e-01 7.00531006e-01 -2.06707090e-01
-5.53940415e-01 6.94864094e-01 5.53524852e-01 -3.13836522e-02
2.17768196e-02 -5.23602366e-01 -8.81354153e-01 -5.79008698e-01
-7.96441808e-02 4.98463325e-02 -3.67807895e-01 -2.57431835e-01
4.57662418e-02 2.93875128e-01 -1.44641197e+00 6.85650706e-02
6.97796583e-01 5.64070106e-01 -6.05254769e-01 6.24704301e-01
5.19587040e-01 7.00049102e-01 -8.11773002e-01 -9.63783979e-01
5.31810641e-01 -1.12875178e-01 -9.24055457e-01 -2.66392112e-01
8.97376478e-01 -4.46300417e-01 -9.85034108e-01 1.02545750e+00
1.01334095e+00 6.08786196e-02 6.42602444e-01 -1.51833391e+00
-1.21098638e+00 5.72130799e-01 -5.96616626e-01 5.47605693e-01
5.95974803e-01 5.07017851e-01 3.72289866e-01 -7.17283964e-01
4.91576880e-01 1.58584595e+00 8.76937389e-01 8.61606479e-01
-9.42507446e-01 -1.07721889e+00 3.68429899e-01 3.46306086e-01
-1.57733428e+00 -6.71742320e-01 3.82412583e-01 -3.33972752e-01
6.73024237e-01 3.85536671e-01 -1.46409506e-02 1.93168151e+00
-2.70571895e-02 7.74281144e-01 1.29034567e+00 -4.04338874e-02
-2.33929321e-01 4.18636575e-02 5.70746183e-01 3.03015590e-01
5.44534981e-01 4.99789208e-01 -6.24700129e-01 -8.83064196e-02
2.18223929e-01 2.39174496e-02 -3.78572524e-01 6.30415827e-02
-1.22791636e+00 3.48581254e-01 6.17338836e-01 4.06153919e-03
-5.69121055e-02 2.04418749e-01 7.16846049e-01 4.77450192e-01
2.77380288e-01 6.58265054e-02 -2.75930822e-01 2.92359684e-02
-5.38273275e-01 5.69764793e-01 4.42918122e-01 1.28045189e+00
3.59196067e-01 1.48052141e-01 -5.15846670e-01 7.60671794e-01
-5.82186207e-02 8.86145234e-01 4.12278056e-01 -5.60730338e-01
7.03356862e-01 2.08313271e-01 1.74697742e-01 -8.64675701e-01
-3.22543383e-01 -5.84341586e-01 -1.42295003e+00 -2.00276867e-01
3.40024799e-01 -3.03759426e-02 -1.15616870e+00 1.90294671e+00
9.11772624e-02 3.68977875e-01 1.53378248e-01 7.71950006e-01
1.23668981e+00 1.27477348e-01 4.48972940e-01 2.53071219e-01
1.47258556e+00 -7.60267735e-01 -5.98026454e-01 -3.25874649e-02
4.43599433e-01 -5.69889963e-01 7.78254211e-01 1.68015406e-01
-7.88066983e-01 -8.29905212e-01 -8.93676519e-01 -1.22885391e-01
-7.73012459e-01 -1.42543688e-01 2.78029561e-01 8.74412775e-01
-1.02863669e+00 2.56817818e-01 -2.80920744e-01 -6.06805861e-01
6.15629554e-01 2.77649105e-01 -9.12429452e-01 -4.95976239e-01
-1.31489289e+00 1.06226492e+00 1.24486975e-01 2.57442921e-01
-1.00928092e+00 -8.57399762e-01 -9.13878918e-01 -3.31068993e-01
-1.26680791e-01 -1.13803244e+00 9.24465656e-01 -6.41040385e-01
-7.83041000e-01 1.06759858e+00 -4.03863490e-01 -5.01883149e-01
1.08608055e+00 -4.57641453e-01 -7.35316098e-01 -2.58615792e-01
1.61987871e-01 7.21199691e-01 7.92264462e-01 -1.60117316e+00
-3.23717982e-01 -3.99950176e-01 8.25694948e-02 2.00702809e-02
-1.76341102e-01 2.89389372e-01 -5.46228111e-01 -8.54292333e-01
-4.12715882e-01 -1.08950233e+00 1.38103530e-01 -5.00375211e-01
-8.63552392e-01 -5.54741696e-02 7.24528015e-01 -9.20847356e-01
9.50788915e-01 -2.18242645e+00 -7.32339621e-02 1.19610623e-01
2.13530615e-01 4.37823445e-01 -3.98627460e-01 3.01400304e-01
-5.57994068e-01 5.04511774e-01 -8.04466009e-02 -7.34486163e-01
1.09901525e-01 2.07896605e-02 -3.33013326e-01 9.00183976e-01
-3.08184028e-02 1.15119398e+00 -5.96658409e-01 -3.51456374e-01
1.86475232e-01 5.93246818e-01 -2.11014614e-01 1.60210043e-01
6.85688555e-01 8.80708098e-01 -1.88615546e-01 1.04454994e+00
1.17792702e+00 2.07968369e-01 -4.09689903e-01 -2.76922017e-01
1.64231628e-01 -1.30362600e-01 -1.06964576e+00 1.28782880e+00
-1.95184633e-01 4.95843768e-01 -1.78553075e-01 -5.96666455e-01
5.30005872e-01 2.50941187e-01 1.21778250e-01 -1.12900603e+00
2.54816115e-01 -2.01869562e-01 -2.94303924e-01 -4.56072778e-01
6.63436055e-01 3.73863727e-01 -2.89935201e-01 2.36184850e-01
-6.12822101e-02 6.24912798e-01 2.81481564e-01 3.51330191e-01
1.10691190e+00 -2.28030324e-01 -9.92030501e-02 -1.66328862e-01
4.69072372e-01 -3.91546965e-01 5.70483208e-01 1.23592150e+00
-5.87746620e-01 9.26299274e-01 1.28214866e-01 -6.34070814e-01
-1.03403449e+00 -1.28943753e+00 -3.38355899e-01 7.25504577e-01
3.82957548e-01 -1.93178996e-01 -7.06345022e-01 -7.32710958e-01
4.92136598e-01 3.39795798e-01 -9.56736743e-01 -3.38917196e-01
-7.68142581e-01 -1.12304544e+00 1.11619830e+00 5.74645698e-01
9.99428809e-01 -6.28360212e-01 1.93384454e-01 -3.44825625e-01
-4.74628091e-01 -1.29489994e+00 -6.71857238e-01 -5.13688326e-01
-4.70829099e-01 -1.31530106e+00 -8.84414911e-01 -5.70446730e-01
8.10751498e-01 6.65026665e-01 1.13448417e+00 4.57598299e-01
-4.16515917e-01 7.78031707e-01 -5.06863475e-01 -6.16347551e-01
-1.99988082e-01 1.37817105e-02 6.24719083e-01 -1.40918314e-01
3.51342320e-01 -2.97699094e-01 -6.46394193e-01 5.56370318e-01
-8.01070273e-01 -3.48390102e-01 4.14466798e-01 7.46239781e-01
2.06968173e-01 -1.30213827e-01 5.61766028e-01 -7.50526130e-01
6.48230970e-01 -5.88467419e-01 -1.47164240e-01 2.76597619e-01
-3.96213859e-01 -2.63117284e-01 3.16066355e-01 -5.21480680e-01
-9.35856044e-01 -4.29907233e-01 -3.07816733e-02 -3.38432789e-01
-3.93787831e-01 -1.99092273e-02 -2.87196040e-01 -1.61735728e-01
7.64036536e-01 2.71222562e-01 -3.72287065e-01 -5.22534668e-01
3.88467550e-01 6.84100747e-01 1.06799960e+00 -7.04351008e-01
1.32432902e+00 5.13097525e-01 -2.41664261e-01 -6.77705765e-01
-4.73586380e-01 -5.37044644e-01 -7.63385177e-01 -1.02582432e-01
6.31244838e-01 -1.48615003e+00 -8.96788597e-01 1.05004668e+00
-1.07438016e+00 -1.01297982e-01 1.87020645e-01 1.81861892e-01
-7.87609518e-02 6.92120135e-01 -5.85417688e-01 -5.93715131e-01
-5.89702308e-01 -9.64305937e-01 8.95277143e-01 3.32451761e-01
-1.36716023e-01 -6.41139448e-01 -2.08580848e-02 8.00682724e-01
5.64559698e-01 3.89574617e-01 4.69692200e-01 -5.14886677e-01
-4.51790661e-01 -5.38508117e-01 -3.49200755e-01 3.58612925e-01
8.21925998e-02 -2.08528072e-01 -1.27960849e+00 -7.50704408e-01
-5.93796074e-01 -2.46963039e-01 1.08547807e+00 1.14962630e-01
1.25937223e+00 -3.41679960e-01 -3.32901031e-01 9.22219217e-01
1.17890191e+00 -3.23837906e-01 1.05714881e+00 7.14283824e-01
9.22409773e-01 3.35278004e-01 2.32486010e-01 2.69019663e-01
8.38116407e-01 6.04407489e-01 3.02834451e-01 -2.44709700e-01
-3.90447229e-01 -2.80146629e-01 4.55570370e-01 3.33931923e-01
-3.64559472e-01 -5.18352449e-01 -8.45830500e-01 6.63229287e-01
-1.76385880e+00 -1.13648129e+00 -1.76809192e-01 2.25904727e+00
5.88653803e-01 -2.22515866e-01 4.31386203e-01 -5.12267638e-04
8.99742901e-01 -6.14393409e-03 -4.70601887e-01 -3.52724898e-03
-5.80005765e-01 -1.65374011e-01 8.99869204e-01 1.38989359e-01
-1.45158112e+00 8.87485862e-01 6.56326008e+00 3.38753581e-01
-6.98305666e-01 2.71669775e-01 6.18383527e-01 -1.60270885e-01
1.47150427e-01 -4.99851882e-01 -8.69873405e-01 7.63288677e-01
8.51995230e-01 -1.51244134e-01 3.20703834e-01 5.50777018e-01
3.26043665e-02 3.76503170e-02 -1.28466010e+00 1.36586010e+00
4.05214131e-01 -9.19955909e-01 1.11329459e-01 -1.58457477e-02
8.02932620e-01 5.14452271e-02 4.65524703e-01 5.02322614e-01
4.18986887e-01 -1.34812355e+00 7.59486079e-01 6.21545255e-01
7.20247626e-01 -7.21080363e-01 1.14349675e+00 -2.96792053e-02
-8.96681070e-01 -1.62769377e-01 -3.30899328e-01 2.60384202e-01
-3.20871249e-02 3.30632210e-01 -3.07698876e-01 9.31399167e-01
1.34104013e+00 6.43447399e-01 -1.08711481e+00 1.22758532e+00
-1.52554259e-01 3.92624259e-01 -7.23090097e-02 6.74124122e-01
-3.17385823e-01 2.52655268e-01 5.47570467e-01 1.60676491e+00
5.52104041e-02 6.02362677e-03 -1.63606167e-01 5.79282522e-01
-4.57505941e-01 -4.04919744e-01 -6.74080551e-01 5.35435140e-01
5.84724128e-01 8.15868974e-01 6.47756010e-02 -3.22450727e-01
-2.42189169e-01 1.07457232e+00 2.49205574e-01 6.85402989e-01
-1.03860164e+00 -7.87680782e-03 1.07495117e+00 -1.03506438e-01
-5.75993024e-02 -2.43786365e-01 -3.51765484e-01 -1.34834659e+00
2.17470780e-01 -1.18456006e+00 6.65923476e-01 -6.27271295e-01
-1.62897694e+00 4.49350744e-01 3.35852623e-01 -1.20486856e+00
1.23996779e-01 -3.90491217e-01 -6.10152721e-01 1.04011369e+00
-1.88589978e+00 -1.63157415e+00 -6.30699337e-01 8.76116037e-01
2.56094426e-01 -3.15002143e-01 6.98743641e-01 6.70515656e-01
-1.23827505e+00 1.44388211e+00 -2.07082536e-02 7.28828609e-01
1.18438029e+00 -9.50584829e-01 8.84502113e-01 1.37250447e+00
-3.03534091e-01 1.00048828e+00 7.77295411e-01 -6.71578407e-01
-1.64404535e+00 -1.46960807e+00 5.54431677e-01 -1.03600955e+00
1.96530655e-01 -3.53675246e-01 -8.86667728e-01 9.86448646e-01
2.27615565e-01 1.10168211e-01 5.83320439e-01 1.72462463e-01
-1.05446231e+00 -2.12457791e-01 -1.30396867e+00 5.42652845e-01
1.41304052e+00 -5.65309286e-01 -3.85358930e-01 2.97864765e-01
3.83288085e-01 -6.23415053e-01 -9.07939494e-01 4.68503654e-01
5.94082952e-01 -7.43863881e-01 1.46643496e+00 -8.03152442e-01
2.85821110e-02 -4.02922034e-01 -8.19753110e-02 -1.30267441e+00
-5.76085985e-01 -3.41481239e-01 -2.54489221e-02 1.59516454e+00
7.37445280e-02 -8.99363875e-01 3.74966323e-01 8.63359869e-01
5.33443093e-02 -1.87276732e-02 -1.01338685e+00 -1.17691588e+00
2.99907416e-01 -2.47452989e-01 8.66605997e-01 1.02327907e+00
-4.92440403e-01 -4.42550778e-01 -8.86917114e-01 7.08574891e-01
1.06608069e+00 -5.11675894e-01 1.34333825e+00 -8.36783230e-01
1.99909642e-01 -2.34327897e-01 -6.18908763e-01 -5.75724483e-01
3.08867425e-01 -7.18452513e-01 -2.15854436e-01 -1.22841060e+00
5.98757029e-01 -4.47692841e-01 -5.16058087e-01 6.62381887e-01
-5.56311548e-01 5.19836426e-01 3.47649336e-01 4.70521331e-01
-5.58019161e-01 3.02301437e-01 8.28552485e-01 -5.64345837e-01
2.12283641e-01 -1.24863721e-01 -1.07779288e+00 2.84767419e-01
9.15716350e-01 -4.01225626e-01 -4.24646481e-04 -7.97884464e-01
-2.20635738e-02 -6.81718349e-01 9.17653024e-01 -1.00573683e+00
3.52905571e-01 1.41556785e-01 7.60749638e-01 -3.74569058e-01
8.13468397e-02 -4.28726673e-01 2.82405108e-01 3.39684129e-01
-1.97105438e-01 4.63079751e-01 4.20394391e-01 5.52104175e-01
4.15924034e-04 2.62345195e-01 8.32315266e-01 -1.01318277e-01
-8.55572939e-01 5.75728297e-01 6.68882206e-02 -4.24419530e-03
7.30402350e-01 -2.90594995e-01 -9.40732598e-01 -3.16198468e-01
-2.80066490e-01 5.13599694e-01 5.99971235e-01 1.05167818e+00
4.49006379e-01 -1.32786524e+00 -1.15274596e+00 6.99084550e-02
4.84279066e-01 -2.26865530e-01 7.92251110e-01 6.88218772e-01
-2.15812251e-01 3.53521734e-01 -2.70772099e-01 -4.92350250e-01
-1.48835874e+00 8.12447846e-01 5.00225961e-01 9.21888426e-02
-5.79038084e-01 8.82397950e-01 2.42811874e-01 -5.10452628e-01
4.69229519e-01 8.88914913e-02 -4.80400175e-02 -2.86012769e-01
9.54693317e-01 7.47057438e-01 2.87096411e-01 -1.15191400e+00
-6.43600821e-01 5.78888535e-01 -3.00764769e-01 3.61407787e-01
1.03832734e+00 -1.79875925e-01 6.91301152e-02 -2.23729685e-01
1.23742485e+00 -2.25589007e-01 -8.80030036e-01 -2.30844989e-01
-3.25469166e-01 -6.93446219e-01 -3.02153379e-01 -9.50358152e-01
-1.04504561e+00 4.35055166e-01 9.75705504e-01 -2.20835954e-01
9.98566628e-01 -1.53146312e-01 7.32254744e-01 1.58448890e-01
4.33743209e-01 -8.27341974e-01 -1.02099046e-01 4.77638781e-01
1.10387874e+00 -1.54525995e+00 1.50766954e-01 -3.39502484e-01
-7.18915462e-01 5.27171671e-01 7.39811957e-01 6.77298605e-02
3.57485116e-01 3.10304388e-02 2.02526748e-01 4.29282300e-02
-3.26686740e-01 -1.16244473e-01 3.00585210e-01 1.04610169e+00
8.98627341e-02 1.24464996e-01 1.37810931e-01 4.35649484e-01
-4.22253877e-01 -2.27449670e-01 5.32822728e-01 7.31507063e-01
2.64130503e-01 -1.18316722e+00 -8.24877441e-01 1.38106152e-01
-4.62622553e-01 -2.02361904e-02 -4.59013581e-01 8.78833532e-01
3.17553729e-01 1.20100844e+00 -1.63464442e-01 -5.18579543e-01
7.73607254e-01 -1.49368852e-01 4.81942087e-01 -1.11128047e-01
-6.85180306e-01 -6.89380288e-01 1.71438858e-01 -5.08552730e-01
-4.51806068e-01 -7.53895700e-01 -5.23916781e-01 -1.00846767e+00
-1.16949668e-03 -3.49132568e-01 5.02226770e-01 8.65856946e-01
6.26449049e-01 3.49342525e-01 5.29387236e-01 -6.60519898e-01
-4.49078709e-01 -1.01107645e+00 -3.04176152e-01 8.96264553e-01
6.12596750e-01 -6.52595580e-01 -4.30579454e-01 9.16333050e-02]
|
[14.642303466796875, 0.9346874952316284]
|
d74231fe-86c4-466c-bc9a-ad962d65ccfd
|
occ3d-a-large-scale-3d-occupancy-prediction
|
2304.14365
| null |
https://arxiv.org/abs/2304.14365v2
|
https://arxiv.org/pdf/2304.14365v2.pdf
|
Occ3D: A Large-Scale 3D Occupancy Prediction Benchmark for Autonomous Driving
|
Robotic perception requires the modeling of both 3D geometry and semantics. Existing methods typically focus on estimating 3D bounding boxes, neglecting finer geometric details and struggling to handle general, out-of-vocabulary objects. 3D occupancy prediction, which estimates the detailed occupancy states and semantics of a scene, is an emerging task to overcome these limitations. To support 3D occupancy prediction, we develop a label generation pipeline that produces dense, visibility-aware labels for any given scene. This pipeline comprises three stages: voxel densification, occlusion reasoning, and image-guided voxel refinement. We establish two benchmarks, derived from the Waymo Open Dataset and the nuScenes Dataset, namely Occ3D-Waymo and Occ3D-nuScenes benchmarks. Furthermore, we provide an extensive analysis of the proposed dataset with various baseline models. Lastly, we propose a new model, dubbed Coarse-to-Fine Occupancy (CTF-Occ) network, which demonstrates superior performance on the Occ3D benchmarks. The code, data, and benchmarks are released at https://tsinghua-mars-lab.github.io/Occ3D/.
|
['Huitong Yang', 'Yucheng Mao', 'Hang Zhao', 'Yilun Wang', 'Yue Wang', 'Longfei Yun', 'Tao Jiang', 'Xiaoyu Tian']
|
2023-04-27
| null | null | null | null |
['occlusion-handling']
|
['computer-vision']
|
[-1.51359662e-01 -5.56463525e-02 -1.76888525e-01 -3.89622360e-01
-7.00150251e-01 -3.45513225e-01 8.56440008e-01 2.21533682e-02
-3.67447615e-01 6.41576350e-01 3.05007458e-01 -1.56032071e-01
8.08403641e-02 -8.94077599e-01 -8.48869264e-01 -5.22024751e-01
-1.92006767e-01 8.75193179e-01 6.02539480e-01 -3.37184779e-02
2.43424580e-01 4.77246940e-01 -1.96461630e+00 2.84698997e-02
6.46423995e-01 1.35018361e+00 4.77924675e-01 3.76585931e-01
-4.79524098e-02 6.38486981e-01 -9.38964635e-02 5.21220684e-01
4.37630057e-01 3.05591911e-01 -7.89775968e-01 1.21094063e-02
4.87697542e-01 -4.91804153e-01 -5.72714210e-01 8.23923290e-01
1.25687078e-01 5.63096523e-01 9.55812752e-01 -1.37295282e+00
-2.65298158e-01 2.86195904e-01 -6.48400664e-01 7.71133453e-02
1.72087520e-01 4.30882633e-01 8.35132718e-01 -1.03718805e+00
7.33055830e-01 1.43485141e+00 4.13611084e-01 2.10585743e-01
-1.07192183e+00 -5.18556893e-01 6.41272187e-01 1.03764065e-01
-1.75749207e+00 -2.26288661e-01 4.61842477e-01 -6.88441217e-01
1.21061099e+00 1.74191356e-01 8.90207708e-01 1.09373295e+00
-3.63416933e-02 7.45764375e-01 1.31122673e+00 3.92435864e-02
5.00547349e-01 -3.16259623e-01 1.81794629e-01 7.45296061e-01
2.34242499e-01 3.02952588e-01 -5.28823256e-01 2.60917358e-02
8.41002405e-01 -3.14522274e-02 -1.58185326e-02 -9.25891817e-01
-1.30130804e+00 6.53233349e-01 7.33904302e-01 -3.72773200e-01
-2.27758944e-01 4.88536894e-01 3.07664722e-01 -4.56175119e-01
5.18976808e-01 1.79743379e-01 -3.10989588e-01 1.32276848e-01
-5.00995219e-01 6.68808639e-01 4.98078823e-01 1.51067519e+00
1.07940698e+00 -3.32925558e-01 -1.72358021e-01 6.51732981e-01
5.62317729e-01 5.03599465e-01 -8.97656456e-02 -1.15357268e+00
3.30106974e-01 5.29713333e-01 4.09913659e-01 -4.89918888e-01
-7.22313643e-01 -1.96823478e-01 -6.89390719e-01 3.78532529e-01
3.38964574e-02 3.26629519e-01 -1.44592857e+00 1.57781780e+00
9.72315252e-01 1.98121086e-01 -1.22830458e-01 1.21994627e+00
1.02856314e+00 6.23108029e-01 4.37604785e-02 2.88986802e-01
1.35023022e+00 -1.49666739e+00 -2.83047050e-01 -4.54014152e-01
5.55952668e-01 -2.75673509e-01 1.15856171e+00 1.98729202e-01
-7.14669466e-01 -5.65763712e-01 -1.15235925e+00 -7.77714670e-01
-6.89463079e-01 -1.93643019e-01 7.86421537e-01 4.98472676e-02
-8.39552402e-01 6.10788465e-02 -9.66767192e-01 -3.87209445e-01
6.31514311e-01 1.19562425e-01 -1.47996172e-01 -2.86706209e-01
-8.64360929e-01 1.04968238e+00 7.48914957e-01 1.41216563e-02
-1.52376187e+00 -6.07474089e-01 -1.18093872e+00 -3.56131077e-01
5.89284062e-01 -8.28672111e-01 1.31326687e+00 2.27114424e-01
-9.27582741e-01 8.54589283e-01 -1.22078009e-01 -2.94799060e-01
5.90680838e-01 -2.44706333e-01 3.06537356e-02 -3.81110936e-01
2.92992324e-01 1.22567832e+00 4.59296137e-01 -1.57433057e+00
-8.77434552e-01 -5.27684748e-01 3.88261676e-01 4.55244720e-01
5.18242776e-01 -7.51321375e-01 -7.36833990e-01 -6.18122444e-02
4.92724299e-01 -8.90056133e-01 -5.33260643e-01 4.00171131e-02
-7.13431120e-01 -4.17971939e-01 6.82400703e-01 -3.08851540e-01
7.69805908e-01 -2.12702370e+00 1.82511508e-01 2.25952089e-01
3.57319206e-01 -4.57071185e-01 -1.27025349e-02 1.67572752e-01
3.51189047e-01 -1.36793122e-01 -2.70902753e-01 -6.27700746e-01
6.11002088e-01 3.31031084e-01 -2.94655234e-01 6.92781210e-01
3.81965041e-02 7.86011934e-01 -9.28675413e-01 -5.81650615e-01
8.04535925e-01 5.40994704e-01 -7.19239473e-01 7.39184394e-02
-8.45768452e-01 6.76990569e-01 -5.73720396e-01 8.35066140e-01
8.80347431e-01 -2.72192985e-01 -1.27875701e-01 -2.27812231e-01
-6.11026049e-01 5.43905675e-01 -1.22534597e+00 2.28039336e+00
-3.20708096e-01 2.17382595e-01 -2.04321042e-01 -3.87061328e-01
7.36993074e-01 -3.42400342e-01 4.51133370e-01 -8.20593596e-01
3.58514249e-01 2.98729569e-01 -5.90136647e-01 -1.96398854e-01
9.86063659e-01 3.78399640e-01 -6.01690233e-01 -1.88427698e-02
-1.35927066e-01 -6.55254185e-01 2.42850438e-01 6.34298921e-02
9.83685255e-01 6.46865427e-01 4.93239254e-01 -3.85560751e-01
7.66623691e-02 5.68968415e-01 5.13075233e-01 7.43243337e-01
-3.98788840e-01 5.14000654e-01 2.74746001e-01 -5.04073083e-01
-1.26311123e+00 -1.22265673e+00 -5.10440290e-01 8.09274971e-01
9.56555307e-01 -4.10053462e-01 -5.43873787e-01 -3.53731185e-01
2.34275877e-01 6.66025698e-01 -6.69529915e-01 1.70674622e-01
-2.43463770e-01 -4.36713487e-01 1.53822675e-01 4.89669800e-01
6.21560037e-01 -9.32125151e-01 -7.86881387e-01 -8.72107148e-02
-3.28892142e-01 -1.28885043e+00 -1.99591503e-01 2.99221754e-01
-4.55735832e-01 -1.08497787e+00 -1.87765360e-01 -7.41846025e-01
4.96824563e-01 3.81135613e-01 1.07615781e+00 -7.06937835e-02
-4.36643630e-01 8.61508995e-02 -3.18729281e-01 -4.66173261e-01
2.20685571e-01 3.57870519e-01 2.34735042e-01 -7.06101179e-01
3.38316023e-01 -5.68095684e-01 -7.51673102e-01 4.26019162e-01
-5.46932518e-01 6.36148572e-01 3.47177207e-01 3.75622153e-01
1.50299168e+00 1.94957014e-02 -3.24884094e-02 -6.17601693e-01
-2.13784263e-01 -6.72002137e-01 -9.90480363e-01 -1.56274959e-01
-2.28346109e-01 -5.82296029e-02 9.61335748e-02 -3.90453190e-02
-9.46122468e-01 3.58364999e-01 -1.40763074e-01 -4.78151292e-01
-4.99293536e-01 1.81820408e-01 -3.09103429e-01 1.63669482e-01
6.42583430e-01 -1.15812331e-01 -6.07714236e-01 -5.71538687e-01
6.87330186e-01 5.90456188e-01 8.17448854e-01 -7.76544273e-01
7.35373139e-01 9.33032513e-01 -2.01404244e-02 -6.08646572e-01
-1.24597657e+00 -7.21549511e-01 -8.46667826e-01 -2.37956807e-01
1.02739859e+00 -1.25437713e+00 -5.67959607e-01 4.21929270e-01
-1.03668654e+00 -8.43628168e-01 -4.48254228e-01 4.68007743e-01
-8.81697178e-01 -3.59163843e-02 -3.40703189e-01 -8.90580237e-01
2.82247011e-02 -1.33975732e+00 1.57414138e+00 6.25882074e-02
-6.02034070e-02 -5.03676534e-01 2.30981439e-01 4.55921590e-01
4.33375873e-03 6.02489710e-01 7.57784784e-01 -2.50692032e-02
-1.27024829e+00 1.75227597e-01 -4.74588841e-01 -1.58041909e-01
-1.79606497e-01 -3.83698344e-01 -1.10039532e+00 -9.60285068e-02
-4.62481469e-01 -6.49568439e-01 1.12289405e+00 4.25366044e-01
1.51131785e+00 1.86809942e-01 -5.27164578e-01 8.85147989e-01
1.29047358e+00 -1.77749753e-01 5.67288399e-01 4.46778238e-01
8.84052157e-01 4.84678179e-01 1.10827017e+00 6.14772022e-01
9.29614186e-01 7.40805626e-01 1.09135842e+00 1.89526733e-02
-3.44797581e-01 -5.72323799e-01 -1.47608936e-01 6.41869009e-01
2.21845061e-02 -2.61219233e-01 -1.02085674e+00 5.90667248e-01
-2.05771780e+00 -4.54327404e-01 -2.78125793e-01 1.81833196e+00
4.01532829e-01 2.56573379e-01 -3.32009345e-02 -3.19626123e-01
5.14985323e-01 5.08380592e-01 -1.14343083e+00 2.00715765e-01
-4.50197011e-02 -8.96766856e-02 6.54971004e-01 7.12768555e-01
-1.46406949e+00 1.38173664e+00 5.57520771e+00 8.03448558e-01
-5.16295850e-01 2.45211974e-01 4.66776133e-01 -4.17949229e-01
-3.00042123e-01 8.61863941e-02 -1.02997565e+00 1.40317976e-01
4.65618789e-01 3.95313501e-01 5.93684554e-01 1.15840995e+00
1.14383861e-01 -4.92426455e-01 -1.16521513e+00 9.38726544e-01
-7.23422691e-02 -1.30151367e+00 -1.14863597e-01 4.23491746e-01
8.98925066e-01 6.90939665e-01 -7.81975314e-02 6.29234493e-01
5.83460748e-01 -1.07736099e+00 1.26584780e+00 5.21580338e-01
9.25757468e-01 -5.86785972e-01 3.80520523e-01 5.02732575e-01
-1.63103580e+00 1.17015012e-01 -6.09799981e-01 -1.19265229e-01
3.67187887e-01 4.75429982e-01 -5.26677966e-01 3.46388876e-01
1.02389193e+00 6.57410741e-01 -2.59959847e-01 1.03940058e+00
-2.37616941e-01 -1.45969465e-01 -5.79218864e-01 1.47001535e-01
3.19488138e-01 -1.15117401e-01 3.26194644e-01 7.36693680e-01
2.64788061e-01 7.73409009e-02 5.81809998e-01 1.13233614e+00
2.81781685e-02 -2.43464977e-01 -6.00585222e-01 3.22956800e-01
7.67400205e-01 1.28528655e+00 -7.92648971e-01 -2.43097007e-01
-2.06352785e-01 6.17493093e-01 6.96313024e-01 1.06056578e-01
-1.17586231e+00 1.42316684e-01 1.07477689e+00 1.95507303e-01
3.54879349e-01 -6.44175649e-01 -4.27996039e-01 -9.51167285e-01
-1.66157767e-01 -3.56034189e-01 1.71062112e-01 -1.07936561e+00
-9.85051870e-01 3.81840855e-01 4.32675958e-01 -1.06735694e+00
2.22608134e-01 -7.32509673e-01 -4.75336164e-02 6.64090455e-01
-1.71813190e+00 -1.13237309e+00 -9.85369265e-01 3.82343918e-01
6.72944129e-01 3.96445572e-01 5.33214927e-01 2.05670312e-01
-5.24550676e-01 -1.47765920e-01 -2.08980262e-01 -2.72567809e-01
2.49211073e-01 -1.11982381e+00 7.22553313e-01 4.92010862e-01
-1.03435747e-01 1.17983982e-01 5.30721486e-01 -8.28212976e-01
-1.34192181e+00 -1.63731372e+00 3.91529918e-01 -7.16147602e-01
4.84373838e-01 -7.96511054e-01 -6.31382406e-01 8.00250709e-01
-2.83258855e-01 1.61906034e-01 8.02083760e-02 1.08795807e-01
-4.74992901e-01 3.33128124e-01 -1.04025149e+00 7.61544287e-01
1.98278809e+00 -3.99429590e-01 -3.38432342e-01 6.41855001e-01
1.19775808e+00 -1.09982038e+00 -8.79891276e-01 6.04674399e-01
5.03603816e-01 -9.45618451e-01 1.08258343e+00 -1.13762721e-01
2.69475847e-01 -7.79120207e-01 -8.26014638e-01 -1.04334009e+00
-4.59900290e-01 7.02640340e-02 -3.58782351e-01 6.98693037e-01
3.26444715e-01 -3.38344067e-01 7.81597197e-01 3.42226297e-01
-7.99801171e-01 -6.86335206e-01 -1.01529896e+00 -7.43585765e-01
-1.55258596e-01 -9.01252806e-01 1.04792798e+00 6.53060675e-01
-2.49032527e-01 2.30634332e-01 -2.39135221e-01 5.52448034e-01
8.36615026e-01 1.80478796e-01 1.21937299e+00 -1.09111381e+00
1.25473097e-01 -2.06705809e-01 -3.93568277e-01 -1.46393847e+00
1.28030464e-01 -9.16503727e-01 6.02560639e-01 -2.02018142e+00
2.20768169e-01 -7.85603046e-01 -1.58691511e-01 4.57712263e-01
3.03856730e-01 3.93809885e-01 -2.42940113e-01 2.99658626e-01
-1.04946387e+00 9.66654420e-01 1.45209670e+00 -2.94155240e-01
-1.42250866e-01 -5.88196099e-01 -3.19171131e-01 7.39643812e-01
8.64713728e-01 -2.22012848e-01 -4.75848317e-01 -5.34911335e-01
3.58981974e-02 -3.25137883e-01 6.73917174e-01 -1.30206895e+00
4.79840264e-02 -4.30631787e-01 2.72965968e-01 -1.53436446e+00
1.04759240e+00 -7.70892203e-01 2.82888174e-01 2.03840971e-01
-5.43616153e-02 -2.55031794e-01 1.50777787e-01 6.49923444e-01
2.44589105e-01 7.89009407e-02 6.79911613e-01 -2.56502777e-01
-1.30684638e+00 8.98576915e-01 -6.22486211e-02 5.32300510e-02
1.16310692e+00 -1.92024276e-01 -6.90389574e-01 1.63521767e-01
-5.81818700e-01 8.07602465e-01 8.58860672e-01 5.60505748e-01
5.79935789e-01 -1.45790768e+00 -3.64635855e-01 2.16922238e-01
7.45645702e-01 1.09894550e+00 6.48937464e-01 9.16693449e-01
-8.07040572e-01 5.37816644e-01 -4.93433513e-02 -9.39237297e-01
-7.18384326e-01 5.59423268e-01 3.48830134e-01 1.75804924e-02
-1.07354248e+00 8.26575637e-01 6.60163879e-01 -8.98177326e-01
5.11992455e-01 -8.96364510e-01 -2.44966689e-02 -3.83614302e-01
3.13005030e-01 3.22634220e-01 -1.52364701e-01 -8.53991508e-01
-3.91964972e-01 3.43893975e-01 1.95080355e-01 -2.43527573e-02
1.11601365e+00 -3.97580415e-01 -2.19947919e-02 6.83640599e-01
8.28889489e-01 -4.79988486e-01 -1.66225708e+00 -2.88288951e-01
-2.22499683e-01 -5.57220876e-01 2.63985842e-01 -5.39609849e-01
-5.87920487e-01 6.77814960e-01 5.96583724e-01 5.66753268e-04
5.22864699e-01 4.86962795e-01 6.56202614e-01 5.05051136e-01
1.06612134e+00 -1.18884802e+00 -6.44438490e-02 9.36217368e-01
9.67773318e-01 -1.31131160e+00 1.01885520e-01 -6.55509710e-01
-3.85610551e-01 3.26802194e-01 1.13265669e+00 -6.00912124e-02
7.79991686e-01 9.23873484e-02 -2.54476905e-01 -4.98457313e-01
-6.59140527e-01 -5.61615884e-01 2.01975405e-01 7.24662900e-01
-1.60129219e-01 3.75307947e-01 3.20838481e-01 3.07369143e-01
-5.72260141e-01 -2.64665753e-01 2.14495346e-01 9.60899472e-01
-8.46935332e-01 -4.92544979e-01 -4.28472936e-01 5.52026331e-01
5.25938094e-01 4.82625812e-02 -9.50847715e-02 9.78129268e-01
6.35730565e-01 7.05683351e-01 4.41824615e-01 -4.82148290e-01
3.87403339e-01 -3.01818520e-01 5.97108185e-01 -8.71945381e-01
3.28008562e-01 5.85635491e-02 2.59459734e-01 -9.31661248e-01
-3.79317760e-01 -7.13085234e-01 -1.61864972e+00 -3.59952807e-01
-1.21696442e-01 -1.72457352e-01 8.89645696e-01 8.68170261e-01
2.29525134e-01 8.37725997e-01 1.61788940e-01 -1.51188409e+00
-5.10221459e-02 -1.05967772e+00 -8.39695752e-01 9.70292836e-02
2.22942576e-01 -1.39738595e+00 -1.57021910e-01 -4.70684737e-01]
|
[8.177175521850586, -2.4868249893188477]
|
5b3b9696-b7e8-4944-b89c-a3e5af52e2c1
|
one-class-learning-towards-generalized-voice-1
|
2010.13995
| null |
https://arxiv.org/abs/2010.13995v1
|
https://arxiv.org/pdf/2010.13995v1.pdf
|
One-class learning towards generalized voice spoofing detection
|
Human voices can be used to authenticate the identity of the speaker, but the automatic speaker verification (ASV) systems are vulnerable to voice spoofing attacks, such as impersonation, replay, text-to-speech, and voice conversion. Recently, researchers developed anti-spoofing techniques to improve the reliability of ASV systems against spoofing attacks. However, most methods encounter difficulties in detecting unknown attacks in practical use, which often have different statistical distributions from known attacks. In this work, we propose an anti-spoofing system to detect unknown logical access attacks (i.e., synthetic speech) using one-class learning. The key idea is to compact the genuine speech representation and inject an angular margin to separate the spoofing attacks in the embedding space. Our system achieves an equal error rate of 2.19% on the evaluation set of ASVspoof 2019 Challenge, outperforming all existing single systems.
|
['Zhiyao Duan', 'Fei Jiang', 'You Zhang']
|
2020-10-27
|
one-class-learning-towards-generalized-voice
|
https://arxiv.org/abs/2010.13995
|
https://arxiv.org/pdf/2010.13995.pdf
| null |
['voice-anti-spoofing']
|
['audio']
|
[ 5.84549792e-02 -2.32783973e-01 -1.70640603e-01 -6.43471256e-02
-8.39406669e-01 -1.00930464e+00 6.05982244e-01 -1.21139780e-01
-9.18438062e-02 3.54698658e-01 3.52101654e-01 -8.47871602e-01
4.32844400e-01 -3.06659609e-01 -4.35371995e-01 -6.95240259e-01
5.82023337e-02 -1.63598247e-02 2.63502687e-01 -1.74353689e-01
1.94672905e-02 5.69048822e-01 -1.13835335e+00 1.71840027e-01
4.72026885e-01 8.24754119e-01 -1.99983388e-01 9.46872830e-01
-1.49298370e-01 4.21413273e-01 -1.11338270e+00 -2.68633395e-01
5.01996316e-02 -3.36869925e-01 -5.80462337e-01 -4.05461431e-01
1.84223056e-01 -2.76807159e-01 -7.62327075e-01 1.30362523e+00
8.96717966e-01 -3.93569976e-01 2.59231448e-01 -1.66738343e+00
-2.49351308e-01 9.34773922e-01 -1.88103154e-01 4.03169960e-01
6.91996098e-01 1.07074246e-01 5.68572342e-01 -8.90412033e-01
2.24805139e-02 1.75961220e+00 5.48662066e-01 9.36214745e-01
-1.00197160e+00 -1.39910972e+00 -4.29839224e-01 3.60564321e-01
-1.69596088e+00 -1.26649892e+00 8.35871994e-01 -1.80637985e-01
4.32886064e-01 4.99261588e-01 2.81750113e-02 1.34146965e+00
8.89930353e-02 7.36502171e-01 1.04777658e+00 -3.47398967e-01
4.55521271e-02 5.22461593e-01 1.39673948e-01 4.08665776e-01
2.22986892e-01 5.34158707e-01 -6.97698355e-01 -7.57715523e-01
-2.87453067e-02 -3.33870709e-01 -6.84187353e-01 1.63129866e-01
-1.13945603e+00 8.65322351e-01 -1.81264896e-02 1.13083303e-01
-7.43667111e-02 -2.00935200e-01 6.16115093e-01 4.39599246e-01
-9.15609375e-02 1.71657816e-01 -2.93746799e-01 1.07720539e-01
-7.37206221e-01 -7.17555452e-03 8.71964216e-01 7.01972604e-01
2.42122129e-01 5.10219216e-01 1.42299592e-01 7.83299923e-01
8.16143274e-01 1.24515522e+00 8.30919623e-01 -2.25884244e-01
5.00903606e-01 -2.66761392e-01 4.66003222e-03 -1.11933458e+00
2.13239696e-02 -2.51193225e-01 -7.53023744e-01 8.08081031e-03
1.73864231e-01 -2.67601430e-01 -7.10808396e-01 1.47294700e+00
3.92686903e-01 4.97012407e-01 3.73355418e-01 4.28905278e-01
5.62693715e-01 6.09799325e-01 -2.95286417e-01 -2.73501158e-01
1.39779925e+00 -7.79713809e-01 -1.18809128e+00 -3.04100126e-01
4.18362141e-01 -1.27252567e+00 8.35759223e-01 3.08985949e-01
-4.41509277e-01 -3.90078068e-01 -1.19670081e+00 6.47166848e-01
-2.29635388e-01 -2.30823889e-01 2.09209651e-01 1.78020680e+00
-8.59450102e-01 1.66857671e-02 -5.00872731e-01 1.01719186e-01
2.45755821e-01 4.19305503e-01 -5.31814873e-01 1.08831726e-01
-1.75356066e+00 4.60790396e-01 2.19784573e-01 -9.91736352e-02
-1.11390352e+00 -4.45596457e-01 -8.54126751e-01 -1.92082182e-01
-3.11162751e-02 8.41752440e-02 1.21155500e+00 -4.32000726e-01
-1.55429375e+00 6.19910359e-01 -4.17195886e-01 -7.69020140e-01
4.57593620e-01 6.28413260e-02 -1.53795505e+00 1.18647240e-01
-1.30003318e-01 -2.60566235e-01 1.60750449e+00 -1.13830817e+00
-3.54359210e-01 -3.58211488e-01 -7.10250139e-01 -5.40669918e-01
-4.41218287e-01 6.23067558e-01 1.72922492e-01 -7.00721025e-01
3.89095038e-01 -1.16984403e+00 3.91616672e-01 -4.32402939e-01
-7.70325661e-01 2.01485604e-02 1.55689132e+00 -1.03926754e+00
1.41633344e+00 -2.78036571e+00 -5.96423388e-01 3.12243134e-01
-1.79612130e-01 8.58196020e-01 1.80821285e-01 2.93978482e-01
-2.41701171e-01 4.63491887e-01 1.02278315e-01 -3.05105656e-01
-4.40389737e-02 -6.34923428e-02 -9.30915654e-01 8.23060751e-01
-2.54864514e-01 2.69685358e-01 -8.27967703e-01 -5.55515468e-01
1.61365896e-01 6.48848474e-01 -1.38890207e-01 1.54344782e-01
3.87893200e-01 1.03937924e-01 -2.46212557e-01 4.87227887e-01
1.02428365e+00 4.42665756e-01 5.91644794e-02 -1.53195986e-03
3.18018824e-01 8.83316875e-01 -1.54766309e+00 5.76005280e-01
-4.25427169e-01 7.94443309e-01 5.93707263e-01 -4.35367227e-01
1.00859976e+00 9.52120006e-01 -1.27206864e-02 -3.17714699e-02
3.45441312e-01 3.24824333e-01 1.47476465e-01 -2.64507979e-01
8.69540274e-02 -1.00988746e-02 9.64200497e-02 5.14353752e-01
6.04415685e-02 1.28336936e-01 -6.36676669e-01 1.05330676e-01
8.87596428e-01 -9.42218125e-01 1.65925756e-01 -1.20798964e-02
1.06966484e+00 -8.32753956e-01 5.59127510e-01 6.25422955e-01
-7.23073065e-01 2.79972106e-01 -5.10926656e-02 1.86715692e-01
-5.92195809e-01 -1.04740131e+00 -2.31215134e-01 5.15666842e-01
1.33578539e-01 -4.19533432e-01 -7.49777675e-01 -9.51407015e-01
2.18599722e-01 6.89765096e-01 4.25583534e-02 -4.49434251e-01
-5.00591040e-01 -2.88451999e-01 1.60605919e+00 1.19342841e-01
4.47309136e-01 -8.73459399e-01 5.25325894e-01 2.71252960e-01
-3.99452001e-01 -1.47410643e+00 -8.87388349e-01 -2.66374111e-01
-2.95685858e-01 -9.17897403e-01 -3.37676048e-01 -9.01168585e-01
1.24227643e-01 7.42983878e-01 3.02997947e-01 7.06579760e-02
5.51848635e-02 -1.87657308e-03 -6.41430989e-02 -6.53830111e-01
-1.33015525e+00 -8.96999463e-02 8.79925072e-01 4.99508113e-01
4.45762664e-01 -2.24584848e-01 -1.13513440e-01 8.95744503e-01
-6.22610092e-01 -7.24447250e-01 6.37288019e-02 7.91021407e-01
-1.60921618e-01 6.12719774e-01 7.77772009e-01 -3.56248707e-01
6.83608174e-01 -2.70460337e-01 -2.82678604e-01 1.50688410e-01
-6.69831097e-01 8.30729157e-02 5.15107274e-01 -9.13857162e-01
-7.13834643e-01 -3.56538743e-02 -4.86948699e-01 -3.32779258e-01
-1.88656837e-01 -3.69343013e-02 -7.73723722e-01 -4.80448633e-01
6.26890600e-01 6.47911727e-01 2.22835869e-01 -4.75530624e-01
1.89585701e-01 1.59245574e+00 5.29307604e-01 -5.20373210e-02
1.40188313e+00 2.38100588e-01 -5.44979215e-01 -1.42566633e+00
-1.11825719e-01 -6.68112040e-01 8.41242447e-03 5.87072410e-02
3.09172690e-01 -9.52776134e-01 -9.76184785e-01 1.19616282e+00
-1.29417658e+00 3.66906673e-01 5.27841330e-01 7.70372152e-01
4.75452356e-02 8.88010681e-01 -6.23939455e-01 -1.08039093e+00
-5.12274325e-01 -1.17302847e+00 7.56272376e-01 2.71476265e-02
-2.15597481e-01 -7.40471959e-01 -2.69486867e-02 5.97627401e-01
6.23844624e-01 -4.29051787e-01 4.28929955e-01 -1.13920033e+00
1.44178588e-02 -6.49942279e-01 1.78757772e-01 7.21209109e-01
5.45108438e-01 -8.48740861e-02 -1.39338696e+00 -6.00354731e-01
3.70191634e-01 7.31639378e-03 5.38521826e-01 -3.14176306e-02
7.42845893e-01 -8.21872413e-01 -4.62532461e-01 6.04566693e-01
7.16908634e-01 3.22313011e-01 6.88440382e-01 -9.14185718e-02
6.04999602e-01 4.47970420e-01 1.50683060e-01 3.40382040e-01
-3.61197405e-02 6.27992511e-01 1.67162806e-01 4.25406128e-01
-7.60888830e-02 -6.01630807e-01 8.30508351e-01 8.21037412e-01
6.80383921e-01 -4.46434289e-01 -8.77204955e-01 3.60130638e-01
-9.30784941e-01 -1.26529193e+00 -3.93360630e-02 2.41085577e+00
7.75302351e-01 4.32988793e-01 1.41540870e-01 8.85453701e-01
1.26477289e+00 6.60255790e-01 -2.14856744e-01 -3.05802375e-01
-2.93469310e-01 -1.01832598e-01 8.26700091e-01 9.15229857e-01
-1.23301256e+00 1.04177606e+00 6.16143942e+00 8.34342301e-01
-1.51730895e+00 2.24454790e-01 1.61328554e-01 4.50740635e-01
-1.33912144e-02 -1.85757205e-01 -1.19682264e+00 7.88052440e-01
1.18447685e+00 -1.79124564e-01 5.84128261e-01 8.07678223e-01
1.25773251e-01 7.83574581e-01 -5.07842302e-01 1.29308438e+00
4.21918511e-01 -8.75985026e-01 -1.72230557e-01 2.51560867e-01
1.57509223e-01 8.21270272e-02 4.46882145e-03 1.14338316e-01
2.86246419e-01 -9.40107822e-01 5.93429744e-01 -3.08528394e-01
7.75213003e-01 -6.70083702e-01 7.49873996e-01 3.02011222e-01
-1.19626641e+00 -1.29827052e-01 -5.92268147e-02 5.13834715e-01
3.24263185e-01 4.48914289e-01 -1.37479270e+00 6.58783838e-02
3.51086229e-01 3.49885374e-02 -2.29631007e-01 7.72648752e-01
-4.51559842e-01 1.20398617e+00 -5.05102754e-01 -2.18396410e-01
-2.34038353e-01 5.66406190e-01 9.16208684e-01 1.15161061e+00
2.45208174e-01 -4.10868943e-01 -5.39429337e-02 2.84017682e-01
-1.51573241e-01 -1.98087767e-02 -7.07592845e-01 -4.45887417e-01
1.11350536e+00 7.21000612e-01 -1.72314029e-02 3.54987010e-02
1.78857580e-01 1.05912900e+00 -4.35447961e-01 4.09055322e-01
-7.48693824e-01 -9.22546864e-01 9.84893084e-01 7.32161775e-02
1.53326094e-01 -4.09255981e-01 8.03526342e-02 -1.08714843e+00
-8.68007541e-02 -1.45177639e+00 1.69779673e-01 -1.87590539e-01
-1.00694215e+00 6.40079618e-01 -7.13155329e-01 -1.13440084e+00
-1.89222679e-01 -3.87611419e-01 -6.09223723e-01 1.04182327e+00
-1.31504560e+00 -8.87034476e-01 2.00496733e-01 7.85655081e-01
2.43363261e-01 -7.35397875e-01 8.96590948e-01 4.07972217e-01
-4.48726296e-01 1.13201010e+00 1.13953203e-01 6.91691399e-01
8.88432741e-01 -6.75228775e-01 9.81080353e-01 1.03662515e+00
2.42591307e-01 7.32893407e-01 7.59618819e-01 -5.25583446e-01
-1.58931100e+00 -8.00226688e-01 1.21589875e+00 -2.13716894e-01
7.55673766e-01 -5.25407970e-01 -9.44943130e-01 4.83145833e-01
-1.47802293e-01 2.26348639e-01 8.08144748e-01 -3.15644115e-01
-9.36723232e-01 -1.36803657e-01 -1.26749098e+00 2.22172648e-01
4.01083618e-01 -1.29812491e+00 -5.37470698e-01 1.63924202e-01
9.60904956e-01 5.11057042e-02 -2.77134180e-01 4.00436716e-03
6.89803064e-01 -5.59979975e-01 1.34299946e+00 -7.39361823e-01
-5.37337184e-01 -5.27689993e-01 -4.42795336e-01 -1.15188360e+00
3.71565633e-02 -1.42108464e+00 -3.07028204e-01 1.71865582e+00
3.80353332e-01 -1.02675724e+00 6.09458566e-01 -1.54432759e-01
4.04093355e-01 1.99303463e-01 -1.11698735e+00 -1.16469932e+00
-3.21491271e-01 -5.56203127e-01 1.05095327e+00 1.16786659e+00
1.16815418e-02 2.89976507e-01 -7.89439559e-01 1.07064271e+00
9.95382130e-01 -7.58853257e-01 9.56502855e-01 -9.53641057e-01
-1.99034750e-01 -2.00929999e-01 -6.07019067e-01 -1.05605996e+00
3.44737977e-01 -6.59503639e-01 6.61851764e-02 -6.45060480e-01
-6.05541050e-01 -3.37983578e-01 -4.31653291e-01 6.56311736e-02
-8.13096985e-02 -3.61931808e-02 3.52341868e-02 1.29045874e-01
1.54026583e-01 4.24201995e-01 5.42642176e-01 -5.38685977e-01
-1.76088467e-01 6.11577332e-01 -3.39854240e-01 7.03048646e-01
1.08252692e+00 -7.91463673e-01 -2.97875911e-01 4.27579544e-02
-3.85387570e-01 2.24774539e-01 1.33468673e-01 -1.06272078e+00
1.76479861e-01 -1.44148633e-01 -2.99296439e-01 -4.20678437e-01
1.14516191e-01 -8.20960641e-01 -1.61404029e-01 7.72518814e-01
-3.23457271e-01 -2.57650614e-01 3.35927233e-02 6.35831237e-01
-2.24691689e-01 -1.61070779e-01 1.07787597e+00 4.30694669e-01
-1.55253008e-01 2.00402811e-01 -4.04021174e-01 -2.01196894e-02
6.60229564e-01 3.00997108e-01 -5.72190523e-01 -6.50201797e-01
-3.69291484e-01 -1.11199290e-01 2.78117478e-01 6.17159903e-01
8.31772268e-01 -1.17300475e+00 -9.30871427e-01 8.03006053e-01
1.08108714e-01 -8.12394261e-01 1.11124434e-01 3.01077306e-01
-4.31054205e-01 4.60887432e-01 2.38606811e-01 -4.10045207e-01
-2.01309967e+00 7.88992703e-01 5.10375023e-01 2.57961750e-01
-2.83743441e-01 7.80531168e-01 -4.30787712e-01 -5.21388769e-01
5.35946548e-01 2.81788975e-01 5.28629981e-02 -1.77254900e-01
1.26605892e+00 4.64721262e-01 1.16558082e-01 -1.19663942e+00
-8.55019569e-01 1.21754482e-01 -1.10763714e-01 -3.87740344e-01
3.63359183e-01 -2.70868212e-01 8.12810063e-02 1.31668700e-02
1.65495980e+00 6.46599948e-01 -4.15902019e-01 -6.13859117e-01
-3.73893790e-02 -6.58309996e-01 2.37627625e-01 -3.91662240e-01
-7.52699375e-01 1.06274498e+00 9.03039217e-01 6.40410364e-01
4.27390903e-01 -1.65274933e-01 1.24758816e+00 2.96279490e-01
3.91243637e-01 -6.27475619e-01 -1.73813477e-01 4.11301613e-01
6.48649931e-01 -1.27489173e+00 -3.52542222e-01 -4.88057524e-01
-5.97814500e-01 8.32456529e-01 -2.07049269e-02 4.39937115e-01
1.09571493e+00 3.79870772e-01 4.85451221e-01 3.87710661e-01
-2.74025500e-01 5.27913332e-01 4.95420210e-02 8.51893663e-01
-1.04111269e-01 3.59646916e-01 1.47479281e-01 4.83914584e-01
-6.98489845e-01 -6.28845632e-01 5.59550643e-01 7.34741211e-01
-4.48843002e-01 -1.13476312e+00 -1.00594950e+00 -7.86083639e-02
-8.09570074e-01 -2.61713006e-02 -4.83397782e-01 -1.34338962e-03
-2.47625053e-01 1.63645399e+00 -3.91162336e-01 -1.00007153e+00
4.94408943e-02 2.98740357e-01 -2.58848935e-01 -1.80190489e-01
-1.73741385e-01 -7.60467723e-02 1.01951249e-01 -1.25280470e-01
-7.19960853e-02 -5.70316076e-01 -1.07029414e+00 -6.47189319e-01
-8.64348948e-01 4.98740107e-01 1.18921995e+00 8.00014794e-01
2.67851889e-01 1.21953443e-01 1.50981164e+00 -2.31899306e-01
-1.15605903e+00 -9.57132280e-01 -7.80888021e-01 4.02883530e-01
1.17787445e+00 -4.52187866e-01 -1.09799004e+00 -1.50538921e-01]
|
[14.07840633392334, 5.87157678604126]
|
ae40648d-bbc0-4486-96b0-0703cc0e5cd5
|
multitrack-music-transformer-learning-long
|
2207.06983
| null |
https://arxiv.org/abs/2207.06983v4
|
https://arxiv.org/pdf/2207.06983v4.pdf
|
Multitrack Music Transformer
|
Existing approaches for generating multitrack music with transformer models have been limited in terms of the number of instruments, the length of the music segments and slow inference. This is partly due to the memory requirements of the lengthy input sequences necessitated by existing representations. In this work, we propose a new multitrack music representation that allows a diverse set of instruments while keeping a short sequence length. Our proposed Multitrack Music Transformer (MMT) achieves comparable performance with state-of-the-art systems, landing in between two recently proposed models in a subjective listening test, while achieving substantial speedups and memory reductions over both, making the method attractive for real time improvisation or near real time creative applications. Further, we propose a new measure for analyzing musical self-attention and show that the trained model attends more to notes that form a consonant interval with the current note and to notes that are 4N beats away from the current step.
|
['Taylor Berg-Kirkpatrick', 'Julian McAuley', 'Shlomo Dubnov', 'Ke Chen', 'Hao-Wen Dong']
|
2022-07-14
| null | null | null | null |
['music-generation', 'music-generation']
|
['audio', 'music']
|
[ 5.43657541e-01 -3.21896374e-01 -4.26323153e-02 1.68351293e-01
-9.07508612e-01 -8.57466877e-01 3.24373543e-01 -8.78246799e-02
-1.91315427e-01 5.56424201e-01 4.39675331e-01 -4.88105156e-02
-4.19053406e-01 -6.99946165e-01 -4.83752429e-01 -5.52399457e-01
9.35697258e-02 6.58710778e-01 1.92109674e-01 -4.10487115e-01
5.05239725e-01 2.02360481e-01 -1.81190717e+00 5.94646513e-01
5.57138860e-01 8.10367405e-01 4.34586614e-01 1.09684694e+00
-2.58812040e-01 8.21130633e-01 -8.84063244e-01 -3.55124474e-01
2.72425443e-01 -9.78498995e-01 -6.97606862e-01 -3.05596590e-01
7.17314780e-01 -2.61438023e-02 -1.43083900e-01 7.39323676e-01
7.74864614e-01 5.13583720e-01 4.83585715e-01 -9.50259745e-01
-5.08851647e-01 1.20834255e+00 -2.60301709e-01 2.94725686e-01
4.52526569e-01 -1.19218223e-01 1.33679390e+00 -7.57598460e-01
4.65389729e-01 1.04234993e+00 8.78413558e-01 5.47395706e-01
-1.21649885e+00 -9.00786877e-01 9.14301798e-02 3.58350575e-01
-1.37507105e+00 -6.09000027e-01 8.14663053e-01 -2.70752966e-01
1.18179965e+00 6.45921171e-01 8.16958785e-01 8.87454093e-01
-2.44339794e-01 7.85299599e-01 7.98268199e-01 -5.38053513e-01
1.13807157e-01 -3.18329841e-01 -3.53118509e-01 4.30617630e-01
-3.66628915e-01 -6.88085658e-03 -1.00401831e+00 -1.17619090e-01
1.05956328e+00 -1.77944407e-01 -2.21439362e-01 5.36011942e-02
-1.31272459e+00 6.11224234e-01 7.09308088e-02 6.64285600e-01
-3.89380455e-01 3.38206887e-01 6.24552250e-01 3.58567238e-01
-9.14230049e-02 7.45107293e-01 -2.95341790e-01 -7.02117264e-01
-1.61107111e+00 4.58145797e-01 6.18970513e-01 8.80436599e-01
3.05593431e-01 3.67583692e-01 -3.99680316e-01 1.02246678e+00
-1.23574652e-01 2.47859031e-01 7.87018955e-01 -1.11098373e+00
4.23353732e-01 2.08149761e-01 -3.19565684e-02 -7.64824212e-01
-2.38694232e-02 -7.56754816e-01 -6.91275001e-01 1.71473980e-01
5.59156716e-01 2.95273632e-01 -4.78132784e-01 1.88604641e+00
-2.23895103e-01 4.64997411e-01 -8.96863267e-02 7.30700910e-01
3.01372498e-01 6.45340204e-01 -2.01679513e-01 -2.83623099e-01
1.34837580e+00 -1.05386722e+00 -5.79405963e-01 -1.04992233e-01
2.20183119e-01 -1.18872821e+00 1.38406682e+00 7.62498736e-01
-1.61161757e+00 -9.23921764e-01 -9.57574010e-01 -1.04825191e-01
1.00982897e-01 1.49853230e-01 5.86079180e-01 6.16958082e-01
-8.65863442e-01 1.14547300e+00 -5.32878757e-01 -5.98512590e-02
2.33755350e-01 2.96461642e-01 2.20816001e-01 2.98064798e-01
-8.36280346e-01 5.47061443e-01 2.60154605e-01 -8.89386013e-02
-6.75909400e-01 -7.76856244e-01 -3.56413037e-01 3.68575335e-01
1.33309782e-01 -6.88287675e-01 1.47764671e+00 -9.84274030e-01
-1.72885191e+00 6.00685835e-01 -2.04570085e-01 -5.78744769e-01
4.48731780e-01 -4.91866916e-01 -4.07043844e-01 5.07530607e-02
1.78401601e-02 4.62391853e-01 8.87937784e-01 -7.74210572e-01
-4.70193356e-01 -1.40146554e-01 -4.68769893e-02 3.02935719e-01
-3.31229955e-01 3.68982255e-02 -4.89814430e-01 -1.12268591e+00
5.55802733e-02 -1.24442172e+00 -3.35998530e-03 -2.73978055e-01
-4.96008575e-01 -1.16152667e-01 5.30033410e-01 -2.77872294e-01
1.54895329e+00 -2.20204878e+00 3.99213612e-01 -7.87130073e-02
-2.77714580e-01 3.30381721e-01 -2.18712077e-01 5.29951334e-01
-8.71864706e-02 5.53818000e-03 -9.66769177e-03 -4.62229759e-01
4.39645760e-02 1.59287885e-01 -6.51996076e-01 -8.05962905e-02
-2.58668125e-01 6.20575488e-01 -8.52640867e-01 -4.38336343e-01
-2.77690068e-02 4.02843833e-01 -7.00598955e-01 1.50129810e-01
-1.94655344e-01 4.12640303e-01 -1.53280407e-01 4.03764337e-01
2.64832944e-01 2.08478495e-02 6.58806935e-02 -2.36514360e-01
-2.81980634e-01 9.24963832e-01 -1.34685278e+00 2.23452282e+00
-5.79865813e-01 4.69645441e-01 -2.21370414e-01 -4.54525918e-01
9.94438410e-01 5.92766464e-01 2.98522532e-01 -6.09608471e-01
7.28717819e-02 2.63468683e-01 2.29403794e-01 -1.89140394e-01
8.35690379e-01 -3.96438509e-01 -1.98945738e-02 7.44333684e-01
-5.69289252e-02 -3.21591228e-01 4.52956766e-01 -1.21399507e-01
1.00295711e+00 3.82677674e-01 1.66634813e-01 -4.77904715e-02
1.87078580e-01 -2.78730333e-01 5.02350986e-01 9.18048203e-01
5.81696294e-02 7.56673098e-01 7.24718422e-02 -3.09073150e-01
-1.20197010e+00 -1.08801115e+00 3.08534235e-01 1.48423457e+00
-2.24780589e-01 -8.83962870e-01 -5.18448412e-01 -1.28486753e-01
-3.91506970e-01 7.69904852e-01 -3.24459046e-01 -1.83022022e-01
-9.52992916e-01 -2.15920374e-01 1.05526006e+00 8.14956844e-01
9.98017639e-02 -1.46350610e+00 -8.24451268e-01 6.09816790e-01
-4.97883439e-01 -5.89866102e-01 -8.98021579e-01 2.61242568e-01
-1.15644300e+00 -7.70094395e-01 -7.10228682e-01 -7.27230370e-01
-1.06275722e-01 2.15923954e-02 1.24655008e+00 -1.57853425e-01
-3.43197316e-01 -2.34806702e-01 -2.34243959e-01 -3.56991172e-01
-3.06186229e-01 1.76191494e-01 8.16606581e-02 -2.60602355e-01
2.52528898e-02 -1.00535846e+00 -4.55206901e-01 1.73499063e-01
-8.02548885e-01 3.07839155e-01 6.66951537e-01 7.59158969e-01
7.32087493e-01 7.23332241e-02 6.65239692e-01 -6.56547010e-01
6.91181540e-01 -1.33167624e-01 -8.35986659e-02 -4.76269834e-02
-3.24834079e-01 2.11485952e-01 7.46034980e-01 -8.51785362e-01
-7.16219783e-01 2.07619295e-02 -3.70910794e-01 -6.39697254e-01
7.23880604e-02 2.53257036e-01 1.64932340e-01 3.75254989e-01
7.12776005e-01 4.20968622e-01 -2.71904558e-01 -5.48370659e-01
4.06380266e-01 3.28504384e-01 7.18403399e-01 -6.47174239e-01
5.96372783e-01 9.70511064e-02 -1.11703508e-01 -6.23605371e-01
-8.71530175e-01 -3.03269237e-01 -5.43122709e-01 -9.39442515e-02
4.10505801e-01 -6.06847823e-01 -8.61337125e-01 2.22824782e-01
-1.00028777e+00 -2.79325813e-01 -7.33943284e-01 5.51788032e-01
-9.85158920e-01 2.03986838e-01 -8.38648856e-01 -9.23992097e-01
-6.24906421e-01 -8.43654573e-01 9.19516981e-01 7.62453750e-02
-8.39200497e-01 -5.63511789e-01 3.15992683e-01 1.46221235e-01
5.17601311e-01 -3.25940326e-02 1.09527636e+00 -2.46725872e-01
-5.08223772e-01 -5.20017222e-02 1.98331848e-01 1.71800718e-01
9.56673473e-02 -2.03240097e-01 -1.00332606e+00 -1.63804963e-01
-1.55297115e-01 -4.45280135e-01 8.32520723e-01 3.55302840e-01
1.18384075e+00 -2.67780781e-01 -2.23051067e-02 6.55177772e-01
1.26729083e+00 3.48475039e-01 7.55148172e-01 2.43822515e-01
5.03561795e-01 2.67660528e-01 4.94929999e-01 6.01035178e-01
-1.14404550e-02 9.23764825e-01 2.40720451e-01 1.70684770e-01
-3.71911585e-01 -4.95587647e-01 4.07004535e-01 1.29461491e+00
-4.61592108e-01 -2.16807067e-01 -3.81778389e-01 8.32626641e-01
-1.82300913e+00 -1.59516323e+00 -3.74399573e-02 2.43897796e+00
9.96502042e-01 2.22573131e-01 4.06406999e-01 7.83777237e-01
4.94196832e-01 1.35849535e-01 -4.16834205e-01 -6.65725350e-01
-2.25818872e-01 8.62557352e-01 7.82341212e-02 2.04867199e-01
-6.48386359e-01 9.36669469e-01 7.26486683e+00 9.56045508e-01
-1.06529248e+00 -1.89760216e-02 -7.20385090e-02 -6.74153864e-01
-2.63445199e-01 -4.13198210e-02 -6.18196428e-01 2.78319150e-01
1.02830005e+00 -2.02132151e-01 8.18614423e-01 6.39869452e-01
1.11552261e-01 4.18229282e-01 -1.21258295e+00 1.20494556e+00
1.95696816e-01 -1.38946402e+00 2.34293982e-01 -9.63502452e-02
6.30967736e-01 -1.64344952e-01 1.99707359e-01 3.35520625e-01
-4.54896241e-02 -1.14745128e+00 1.17391968e+00 5.21688938e-01
8.49226713e-01 -8.40906322e-01 1.08815603e-01 2.61929989e-01
-1.53310466e+00 -2.32990384e-01 -2.82318026e-01 -3.40948254e-01
1.69676021e-01 2.79445350e-01 -8.35172057e-01 2.94739246e-01
5.58640122e-01 4.69271779e-01 -3.42178583e-01 1.15418744e+00
-8.60725418e-02 8.41629207e-01 -1.64783210e-01 1.85364000e-02
6.49045780e-02 4.76950258e-02 6.91263616e-01 1.26496601e+00
8.75674307e-01 -1.80131853e-01 1.05007663e-01 9.85014677e-01
-1.02196885e-02 2.16429323e-01 -3.78333956e-01 -2.15160176e-01
7.30073392e-01 9.76540148e-01 -6.21215522e-01 -4.77807283e-01
5.92977740e-02 1.18799400e+00 2.66414642e-01 3.38651650e-02
-7.55019844e-01 -5.83467722e-01 6.52505040e-01 1.22247614e-01
5.59080184e-01 -3.06906521e-01 -2.60674387e-01 -8.39052320e-01
-2.00219285e-02 -1.01353359e+00 2.87160039e-01 -9.06691194e-01
-1.00624812e+00 8.99783194e-01 -4.64199781e-01 -1.40407670e+00
-6.06913626e-01 -2.57439733e-01 -6.37323499e-01 8.50634038e-01
-1.01972008e+00 -9.82595205e-01 -1.83273181e-02 6.13383174e-01
8.33312452e-01 -1.75672114e-01 1.37639797e+00 4.41129237e-01
-1.36538278e-02 7.53390431e-01 -1.99088320e-01 -1.24206863e-01
8.48296642e-01 -1.17222500e+00 4.69366252e-01 5.54501772e-01
8.92398477e-01 7.24181950e-01 7.84814000e-01 -3.32753420e-01
-1.22225726e+00 -7.36106396e-01 1.04195929e+00 -2.88130164e-01
4.47524101e-01 -2.34771028e-01 -8.06744039e-01 5.99406719e-01
2.26247683e-01 -5.95593512e-01 1.06216836e+00 4.71702576e-01
-6.04895175e-01 2.03449186e-02 -5.58999121e-01 6.66249871e-01
1.23104823e+00 -6.88750267e-01 -7.50953615e-01 -3.85934263e-02
3.95444214e-01 -3.45347166e-01 -7.20835984e-01 1.00454032e-01
9.50377762e-01 -1.10515726e+00 9.95971024e-01 -3.22944820e-01
4.27289814e-01 -4.11673576e-01 -1.29943922e-01 -1.25808334e+00
-6.79932415e-01 -8.31477761e-01 -1.56418756e-01 1.27472365e+00
2.76739329e-01 1.26694635e-01 7.68467844e-01 -2.76367515e-01
-4.00188088e-01 -4.77531284e-01 -9.17122185e-01 -8.43861163e-01
-2.05834478e-01 -7.45671809e-01 7.32506692e-01 7.72399485e-01
1.11165710e-01 5.88355720e-01 -7.25445092e-01 -2.34055743e-01
4.30613816e-01 5.73556304e-01 5.15030682e-01 -1.33984089e+00
-1.01829910e+00 -7.95454085e-01 -1.93436310e-01 -1.14706182e+00
-1.52034596e-01 -9.56497371e-01 2.57755462e-02 -1.19973171e+00
1.89028084e-01 -6.38204217e-01 -5.41370153e-01 4.78675485e-01
-1.90425552e-02 6.46909356e-01 4.70434636e-01 3.15800279e-01
-2.80545622e-01 4.22675699e-01 1.23176348e+00 -2.04149127e-01
-4.74949569e-01 1.74234971e-01 -6.36015713e-01 7.51664996e-01
6.62538946e-01 -5.40023327e-01 -5.96177995e-01 -5.00772774e-01
3.77912849e-01 3.12245280e-01 1.54791743e-01 -1.39974010e+00
3.04704368e-01 4.07490283e-02 3.03812087e-01 -7.25969732e-01
8.08809757e-01 -3.77615511e-01 4.16449934e-01 4.30490822e-01
-7.38810956e-01 2.85416335e-01 4.30382162e-01 3.04264516e-01
-2.23678917e-01 -4.64341849e-01 6.42775714e-01 -2.59009182e-01
-2.60903478e-01 -3.62210721e-02 -3.31608951e-01 1.20759740e-01
5.34160435e-01 -4.12277311e-01 7.27867261e-02 -3.49122107e-01
-8.27827454e-01 -5.35075724e-01 2.94890285e-01 6.59042120e-01
5.19833922e-01 -1.46344697e+00 -6.82856977e-01 1.63589746e-01
1.33686289e-01 -4.17433918e-01 3.39482188e-01 5.29129922e-01
-2.36123532e-01 2.40976095e-01 -4.46097761e-01 -3.64091903e-01
-1.59814608e+00 3.61007005e-01 7.71509409e-02 -2.84076035e-01
-7.85324037e-01 1.02141094e+00 -1.44648015e-01 -1.12373933e-01
2.71748334e-01 -3.94627750e-01 5.23075201e-02 3.86644565e-02
7.32019663e-01 4.77320105e-01 6.00717291e-02 -5.50942123e-01
-1.14623211e-01 8.14299881e-01 -1.93590596e-02 -3.74516129e-01
1.03983355e+00 2.66724169e-01 1.17732398e-01 9.31775212e-01
6.44436896e-01 4.43210065e-01 -1.00544202e+00 -5.64220175e-02
-1.66395456e-01 -6.04912996e-01 -1.43082142e-01 -9.05492604e-01
-6.93597794e-01 8.55738938e-01 5.17525852e-01 4.93171029e-02
1.19849038e+00 -1.20673321e-01 1.09140372e+00 3.10866505e-01
3.76954496e-01 -1.10981488e+00 3.74764293e-01 5.61401129e-01
8.49986672e-01 -2.73472905e-01 -3.31156164e-01 -1.23343214e-01
-6.83460832e-01 1.02211988e+00 2.41336867e-01 -2.50493169e-01
1.74849741e-02 4.64970469e-01 8.57501570e-03 6.61256909e-02
-8.54187548e-01 -1.13616444e-01 3.08681935e-01 4.14315403e-01
8.65286112e-01 8.13437775e-02 -5.36090694e-02 5.11894643e-01
-8.83108377e-01 7.44798928e-02 3.74136120e-01 6.62268817e-01
-5.11250317e-01 -1.33435762e+00 -3.80561173e-01 2.27667421e-01
-5.67811251e-01 -4.26017046e-01 -3.49905550e-01 3.16372156e-01
3.94588828e-01 8.98301601e-01 1.28151447e-01 -3.12309235e-01
4.05982852e-01 3.42882663e-01 8.72015715e-01 -6.25831008e-01
-1.00530553e+00 4.82947737e-01 -2.40056943e-02 -4.36204255e-01
-3.53873998e-01 -7.63558328e-01 -1.21530747e+00 -1.05253659e-01
-3.80162001e-01 2.45704755e-01 5.37604272e-01 6.58735633e-01
4.06637639e-01 9.55955446e-01 2.86365390e-01 -1.05013764e+00
-4.55897033e-01 -1.19308972e+00 -6.88497066e-01 3.93171221e-01
5.44020608e-02 -3.82466018e-01 -3.28281149e-02 1.49552613e-01]
|
[15.899904251098633, 5.4330573081970215]
|
fc67732a-5dbd-4da4-828d-f0a8ffaa2ff1
|
simplified-boardgames
|
1606.02645
| null |
http://arxiv.org/abs/1606.02645v2
|
http://arxiv.org/pdf/1606.02645v2.pdf
|
Simplified Boardgames
|
We formalize Simplified Boardgames language, which describes a subclass of
arbitrary board games. The language structure is based on the regular
expressions, which makes the rules easily machine-processable while keeping the
rules concise and fairly human-readable.
|
['Marek Szykuła', 'Jakub Kowalski', 'Jakub Sutowicz']
|
2016-06-08
| null | null | null | null |
['board-games']
|
['playing-games']
|
[-4.06683505e-01 6.63519144e-01 -2.99840808e-01 -5.73593900e-02
8.92224562e-05 -9.68062878e-01 5.86644590e-01 -4.32955753e-03
-2.19156638e-01 9.11720753e-01 8.43838677e-02 -9.27032292e-01
-1.54523939e-01 -1.45029759e+00 -3.96362752e-01 -1.05631448e-01
-4.67353433e-01 4.79568928e-01 9.75018919e-01 -9.67823505e-01
-1.85039669e-01 1.60513781e-02 -1.33928633e+00 1.99755579e-01
6.13938034e-01 6.77221954e-01 6.96435049e-02 1.06265593e+00
-8.94490406e-02 1.86570108e+00 -5.34410655e-01 -8.22101951e-01
5.37274063e-01 -2.61461437e-01 -1.22651494e+00 1.89802032e-02
-5.38945377e-01 -4.02068555e-01 -6.94990396e-01 1.21768498e+00
-4.73485231e-01 -3.18410695e-02 2.86162168e-01 -1.41097772e+00
-1.69100091e-01 5.26452959e-01 -5.31592518e-02 -2.94946045e-01
1.08641303e+00 3.02568320e-02 1.51332223e+00 2.65637308e-01
6.14621103e-01 9.51575160e-01 5.84680200e-01 4.64231789e-01
-6.44119799e-01 -2.98207790e-01 3.31380069e-01 2.96584517e-02
-1.51302910e+00 -5.72135299e-02 1.70285590e-02 -1.50746077e-01
1.38143635e+00 8.84716988e-01 1.32936549e+00 1.52137309e-01
7.50401139e-01 7.59025931e-01 8.75059247e-01 -6.69353127e-01
1.22394808e-01 -3.95970434e-01 4.51754004e-01 9.79441941e-01
7.79290617e-01 3.06997180e-01 7.67815411e-02 -4.23059791e-01
1.36778152e+00 -4.35028076e-02 6.83375299e-02 -3.02782893e-01
-6.93031549e-01 8.96498680e-01 -4.06194806e-01 1.55443788e-01
-2.05428042e-02 1.53362989e-01 6.11174822e-01 5.44977546e-01
-1.43969417e-01 4.07386690e-01 -2.91119814e-01 -4.83396500e-01
-5.74983954e-01 7.81392872e-01 1.46719611e+00 1.34543538e+00
6.41309559e-01 1.55730009e-01 7.14055542e-03 4.14673746e-01
5.83916306e-01 4.15528983e-01 3.22663277e-01 -1.06186664e+00
-8.44189338e-03 9.61777508e-01 3.95494848e-01 -7.66600490e-01
-3.06736767e-01 3.21804546e-02 -5.37120461e-01 5.19965410e-01
1.00976400e-01 3.86986434e-02 -3.79294813e-01 1.29516912e+00
7.68260732e-02 -1.06826968e-01 9.47092846e-02 3.58072370e-01
1.16318548e+00 9.62638855e-01 1.93726972e-01 -2.66277641e-01
1.60138476e+00 -1.16916180e+00 -5.98713994e-01 -1.47456244e-01
5.46180367e-01 -9.85467061e-03 1.13322198e+00 5.57015002e-01
-1.54000556e+00 -4.84051649e-03 -1.46828175e+00 -5.25009492e-03
-3.40994328e-01 -3.99443090e-01 1.32345188e+00 1.03756678e+00
-1.18926835e+00 1.77864864e-01 -9.12561357e-01 -4.02736701e-02
-1.48672536e-01 8.57968807e-01 -5.30775547e-01 1.26104370e-01
-1.16334641e+00 8.36238682e-01 9.85452116e-01 -2.51322478e-01
-8.90584230e-01 6.88664839e-02 -1.28503931e+00 -1.15904154e-03
7.46749640e-01 -6.89934433e-01 1.81343448e+00 -7.70035684e-01
-2.36578178e+00 1.28052413e+00 2.80504674e-01 -5.10000646e-01
4.39894438e-01 2.44993284e-01 -6.42959118e-01 -1.33998379e-01
2.43686885e-01 6.59598485e-02 2.16697857e-01 -4.47231680e-01
-8.32243383e-01 1.69527218e-01 1.47835648e+00 3.60971600e-01
3.20330381e-01 5.89874685e-01 -6.77402258e-01 -4.14278895e-01
-3.54205251e-01 -7.13373125e-01 -6.91952705e-01 -4.43883598e-01
-3.18587244e-01 -2.90002793e-01 -9.41261202e-02 -1.70621276e-01
1.89042723e+00 -2.00713825e+00 -1.50977701e-01 3.44482005e-01
6.89770997e-01 1.26628831e-01 2.72130203e-02 9.29903209e-01
7.91472048e-02 2.46355355e-01 2.56803602e-01 6.16622090e-01
5.12757123e-01 8.86860132e-01 -1.84687182e-01 2.34200194e-01
-2.43880406e-01 1.01564229e+00 -8.06799650e-01 -4.32945013e-01
1.96305737e-01 -7.72689223e-01 -9.43046272e-01 5.67673206e-01
-3.99004012e-01 -5.30148268e-01 -8.42413068e-01 4.47304368e-01
4.60440159e-01 -2.10886113e-02 4.45841223e-01 7.99099565e-01
-4.04251546e-01 6.75050497e-01 -1.66215158e+00 1.31338859e+00
-3.47793818e-01 1.95310321e-02 2.69934237e-01 -5.50961912e-01
6.98794067e-01 3.44926924e-01 -1.90587431e-01 -3.38253111e-01
1.62723899e-01 -1.25628456e-01 1.84177384e-01 -4.32522714e-01
6.61779284e-01 -4.64003533e-01 -7.37244308e-01 6.82275653e-01
-2.44001761e-01 -8.59691918e-01 1.00755596e+00 4.36235964e-01
1.26812112e+00 -9.18319391e-04 1.57604051e+00 -5.43720722e-01
6.13153517e-01 3.22942436e-01 6.62701607e-01 1.21779358e+00
3.44321877e-02 -1.18549084e-02 8.27105999e-01 -8.48587811e-01
-5.87260723e-01 -1.05363190e+00 5.81228793e-01 1.21877718e+00
4.63709772e-01 -1.59651613e+00 -5.17854214e-01 -1.09711722e-01
-3.35674524e-01 3.41541827e-01 -3.23359162e-01 -1.99989468e-01
-1.98902890e-01 -1.29102841e-01 9.54501748e-01 5.04268050e-01
8.14856052e-01 -9.00807559e-01 -5.89280725e-01 4.77732867e-01
8.67202580e-02 -9.77022588e-01 -1.51996195e-01 5.31084724e-02
-6.62388980e-01 -1.34379995e+00 1.40908539e-01 -7.58441389e-01
1.80788934e-01 2.51177698e-01 1.37414455e+00 6.68574035e-01
9.03908908e-02 3.14973503e-01 -5.05864859e-01 -6.38096213e-01
-5.56477249e-01 -4.72928807e-02 -2.08601862e-01 -7.39601135e-01
1.54486045e-01 -3.81347746e-01 1.18374608e-01 9.20405611e-02
-9.65239286e-01 4.03379202e-01 -1.16090268e-01 7.20433712e-01
3.84682268e-01 5.20723879e-01 -5.32422543e-01 -1.00041509e+00
1.05671465e+00 2.28965417e-01 -8.78948033e-01 3.54318380e-01
9.86568164e-03 2.62077805e-02 8.21209192e-01 -6.79942966e-02
-7.02458203e-01 -4.19556797e-01 -2.41237834e-01 8.65837574e-01
5.15063219e-02 7.29929507e-01 -3.50237906e-01 -2.82550424e-01
4.34833795e-01 1.03821307e-01 -2.32554600e-01 3.34262073e-01
2.87045658e-01 4.43692386e-01 4.47672755e-01 -1.04339314e+00
7.17743754e-01 2.41566420e-01 -1.00626387e-01 -7.30334699e-01
-3.49976897e-01 -1.98154286e-01 -1.58548817e-01 -2.11008966e-01
5.37218928e-01 -8.81225586e-01 -1.20496237e+00 3.40482235e-01
-9.51023102e-01 -7.90180564e-01 -6.29658401e-01 -8.50039646e-02
-9.06847775e-01 3.36902767e-01 -1.02271247e+00 -8.32829118e-01
-1.56068755e-02 -8.73186827e-01 5.78565776e-01 2.73987383e-01
-6.43289447e-01 -8.41434300e-01 5.78073144e-01 -1.34308070e-01
-3.92186455e-02 1.77154317e-01 9.50657666e-01 -5.17097771e-01
-3.93863440e-01 -5.65132678e-01 2.08960965e-01 1.88320324e-01
-1.28020734e-01 3.67520362e-01 -1.85788304e-01 1.17768943e-01
-1.94936082e-01 -1.03118932e-02 -2.37046897e-01 2.20322892e-01
7.17875481e-01 -3.48401040e-01 -2.61327475e-01 2.75039941e-01
1.13957095e+00 4.74691510e-01 1.03620613e+00 8.59274924e-01
-1.23614497e-01 -5.83936311e-02 4.16069567e-01 5.44756770e-01
6.97446287e-01 6.39070928e-01 1.94083124e-01 -4.49667349e-02
5.63268602e-01 -4.47273850e-01 3.39237899e-01 8.99440289e-01
-6.84261560e-01 9.81701091e-02 -9.48862791e-01 2.48744905e-01
-2.00325346e+00 -1.30153060e+00 -3.20376813e-01 1.57731211e+00
8.67848039e-01 4.07599121e-01 5.91933250e-01 2.76585132e-01
4.68937099e-01 4.38013405e-01 3.56178880e-01 -1.13828528e+00
-1.29301786e-01 6.95478559e-01 5.30858636e-01 9.00797963e-01
-9.32738006e-01 1.30889118e+00 8.95227337e+00 8.74111533e-01
-5.25134981e-01 2.48646718e-02 -1.52160019e-01 1.81992240e-02
-2.65369803e-01 3.49877447e-01 -3.55967641e-01 -6.50879666e-02
5.82162023e-01 -1.02413607e+00 6.26241326e-01 7.83075392e-01
1.71197593e-01 -1.95749089e-01 -8.00897062e-01 9.34111476e-01
-4.55742598e-01 -1.51859224e+00 7.95657411e-02 -1.47200733e-01
4.49168384e-01 -3.41951191e-01 -4.87602562e-01 8.72715533e-01
1.54863560e+00 -1.12150323e+00 1.25372791e+00 -9.85818505e-02
6.26314938e-01 -5.68864644e-01 6.53531671e-01 5.98991752e-01
-1.52102363e+00 -4.05763596e-01 -6.65019095e-01 -1.46172750e+00
1.62711963e-02 -9.86874700e-02 -1.58186823e-01 9.23683465e-01
5.38890302e-01 5.11767387e-01 -1.94454432e-01 1.20250189e+00
-9.39729750e-01 4.33664501e-01 -4.86815095e-01 -4.11380917e-01
6.78465009e-01 -5.33162832e-01 2.80004472e-01 9.56107616e-01
1.53089094e-03 9.19255137e-01 5.77143312e-01 2.96328068e-01
4.77772981e-01 6.62484244e-02 -8.61613572e-01 1.03831723e-01
2.05078319e-01 8.78114045e-01 -9.29137886e-01 -4.29815382e-01
-7.12743223e-01 6.16960704e-01 2.53244430e-01 1.71791211e-01
-8.64167809e-01 -4.52611357e-01 7.51219213e-01 3.63431603e-01
-7.92522042e-04 -4.74571973e-01 -7.70134106e-02 -1.22218156e+00
-8.88763443e-02 -1.24416864e+00 7.54419446e-01 -8.89113307e-01
-8.28026712e-01 8.27604175e-01 4.49249268e-01 -1.07173169e+00
-3.86454940e-01 -1.02652073e+00 -1.02537692e+00 2.67196923e-01
-8.24854732e-01 -8.83863628e-01 6.10125586e-02 8.17767501e-01
9.03802365e-03 -9.98285487e-02 1.29170859e+00 -2.82007903e-01
-5.84117591e-01 1.77828699e-01 -4.03811693e-01 3.88475619e-02
-1.91749871e-01 -1.19195879e+00 4.28747892e-01 9.13310051e-01
-1.23557284e-01 7.59563208e-01 8.20108473e-01 -3.49914879e-01
-1.38299787e+00 -7.16027558e-01 6.81562841e-01 -1.15774855e-01
1.07586884e+00 -3.52029622e-01 -8.23868364e-02 1.08057272e+00
4.03521061e-01 -3.53858352e-01 7.57859826e-01 2.43231252e-01
-3.49595010e-01 -2.07477584e-01 -7.50044882e-01 1.15745270e+00
1.29508138e+00 -7.22409964e-01 -1.11853898e+00 2.08199501e-01
2.77436078e-01 -6.44786835e-01 -4.13540483e-01 6.88098148e-02
5.62925935e-01 -9.40118611e-01 6.37631834e-01 -1.00303745e+00
-2.68749823e-03 -6.59011245e-01 -1.43822655e-01 -8.68306756e-01
-7.94300020e-01 -1.25660181e+00 5.04076891e-02 4.97295648e-01
5.26179135e-01 -8.51632953e-01 6.77166283e-01 9.19898748e-01
-7.45529383e-02 -3.88986856e-01 -9.84591603e-01 -1.18178988e+00
5.51439263e-03 -8.19757581e-01 1.03473938e+00 4.77065504e-01
1.41636384e+00 9.42564756e-02 -3.80392790e-01 -1.59808919e-01
-1.84361681e-01 1.80054829e-01 9.54700768e-01 -1.22495627e+00
-8.83666158e-01 -5.84701300e-01 -8.76476049e-01 -1.38914335e+00
3.38776231e-01 -8.80478680e-01 -1.13030955e-01 -1.69066274e+00
1.36292979e-01 -1.93131804e-01 3.79075289e-01 7.83275962e-01
2.91227669e-01 8.72718692e-02 7.04484135e-02 -9.85477492e-02
-1.45714879e+00 2.66737789e-02 1.24023890e+00 -2.83404380e-01
-1.46973759e-01 1.22776754e-01 -9.68195319e-01 9.32808101e-01
5.37235439e-01 -3.78354758e-01 -5.67220807e-01 -2.66279608e-01
9.96566951e-01 3.74746293e-01 4.37545478e-02 -8.17393780e-01
2.25140780e-01 -8.22737873e-01 -7.94889927e-01 3.26698154e-01
-1.33127600e-01 -6.03584588e-01 5.65152764e-01 7.65020847e-01
1.08695373e-01 3.16000164e-01 3.48170668e-01 3.36553715e-02
-4.86652374e-01 -5.35619318e-01 1.66544482e-01 -3.92243177e-01
-7.65625954e-01 2.85710484e-01 -1.37211728e+00 1.73868269e-01
1.46900630e+00 -6.67581618e-01 -9.57370549e-03 -9.99369323e-01
-6.57769978e-01 1.27701804e-01 7.81975389e-01 -1.84871361e-01
1.37543023e-01 -1.26859891e+00 -3.96627665e-01 1.32301584e-01
1.43018007e-01 -8.36096033e-02 4.43985239e-02 2.26797849e-01
-1.57063508e+00 3.00763309e-01 -5.93663931e-01 1.43790245e-01
-1.36338508e+00 4.57907915e-01 3.55506152e-01 -9.15209472e-01
-7.27047622e-01 5.43164372e-01 4.65294123e-01 -3.69622409e-01
-3.26216482e-02 -8.06393147e-01 -2.72769809e-01 -1.02123439e+00
1.06606090e+00 1.70767039e-01 -2.44211406e-01 -1.48214772e-01
-4.93066698e-01 3.94132882e-01 3.45528647e-02 -2.64940083e-01
1.25424314e+00 1.98089749e-01 -8.26266885e-01 2.13516966e-01
-4.79300134e-03 1.50165364e-01 -4.45821375e-01 2.33761907e-01
-9.93831307e-02 -6.08933493e-02 -3.13906580e-01 -3.96908641e-01
-3.98473054e-01 1.69281647e-01 -5.68518341e-01 9.01414514e-01
8.39610517e-01 -2.31323108e-01 4.17825580e-01 6.08093917e-01
1.20215940e+00 -1.02884161e+00 -6.45085931e-01 1.17796361e+00
6.41262054e-01 -2.43264645e-01 1.74740002e-01 -9.38681066e-01
-6.22691691e-01 1.09909546e+00 3.96260291e-01 -6.72949612e-01
6.54984295e-01 1.08627951e+00 -4.60483171e-02 -2.82906592e-01
-1.03124011e+00 -2.38761455e-01 -5.28406322e-01 8.41091812e-01
1.20163821e-01 7.04587817e-01 -7.58146048e-01 1.63286376e+00
-7.38315761e-01 2.45147750e-01 1.00948501e+00 1.25560653e+00
-5.14450729e-01 -1.47025669e+00 -5.13523936e-01 1.18896216e-01
-5.02035201e-01 -1.87012509e-01 -5.91522157e-01 1.39498711e+00
-4.81285676e-02 1.20481217e+00 -2.10023094e-02 -3.46716762e-01
5.83195329e-01 -2.90110648e-01 9.02422726e-01 -1.03063154e+00
-6.57052577e-01 -4.96708810e-01 7.26513863e-01 -5.26733041e-01
-2.16143325e-01 -2.07283914e-01 -1.27466059e+00 -1.16033328e+00
-8.59327316e-02 7.48907208e-01 -5.68382442e-01 1.04433250e+00
-2.55396694e-01 1.82661220e-01 -5.78748398e-02 -1.80686295e-01
-2.11322844e-01 -2.63258129e-01 -1.27305794e+00 -7.73721561e-02
-3.71177197e-01 -5.05533755e-01 4.65447187e-01 -1.76238403e-01]
|
[3.4427685737609863, 1.4646984338760376]
|
15833bf7-33a8-4833-a9f0-d210a7e83137
|
collecting-fluency-corrections-for-spoken
| null | null |
https://aclanthology.org/W17-5010
|
https://aclanthology.org/W17-5010.pdf
|
Collecting fluency corrections for spoken learner English
|
We present crowdsourced collection of error annotations for transcriptions of spoken learner English. Our emphasis in data collection is on fluency corrections, a more complete correction than has traditionally been aimed for in grammatical error correction research (GEC). Fluency corrections require improvements to the text, taking discourse and utterance level semantics into account: the result is a more naturalistic, holistic version of the original. We propose that this shifted emphasis be reflected in a new name for the task: {`}holistic error correction{'} (HEC). We analyse crowdworker behaviour in HEC and conclude that the method is useful with certain amendments for future work.
|
['Paula Buttery', 'Emma Flint', 'Andrew Caines']
|
2017-09-01
| null | null | null |
ws-2017-9
|
['grammatical-error-detection']
|
['natural-language-processing']
|
[ 1.33407339e-01 6.97757125e-01 4.64152575e-01 -6.52461588e-01
-9.86646533e-01 -3.81131887e-01 5.71393847e-01 6.85829520e-01
-1.04505825e+00 9.80444849e-01 1.17691648e+00 -2.87755877e-01
1.20457254e-01 -1.18609868e-01 -7.86087215e-01 1.76826894e-01
5.73236942e-01 5.37579656e-01 2.34903619e-01 -7.14131832e-01
5.81656992e-01 -1.39996767e-01 -1.54409003e+00 4.05672610e-01
1.15640330e+00 -9.84680429e-02 2.58919388e-01 9.60531414e-01
-3.22370768e-01 1.36628485e+00 -1.04254401e+00 -9.20168638e-01
-4.41773981e-01 -6.63784325e-01 -1.67494524e+00 -1.47273585e-01
5.53146482e-01 -8.26823264e-02 2.27726460e-01 1.10474384e+00
7.62341440e-01 6.50707603e-01 5.02901733e-01 -6.70468807e-01
-9.53915596e-01 9.99727249e-01 3.36258054e-01 3.52770060e-01
8.91210854e-01 -1.88135803e-01 6.48998201e-01 -1.43282664e+00
9.80949998e-01 1.07220280e+00 1.08569109e+00 1.02482378e+00
-7.83676088e-01 -5.56001291e-02 5.39528504e-02 2.72683889e-01
-1.32667887e+00 -9.33564425e-01 1.81619242e-01 -4.96131897e-01
1.65313280e+00 4.53211844e-01 5.54344952e-01 1.19615972e+00
-2.40207359e-01 6.39663339e-01 9.47001576e-01 -1.13545656e+00
1.47312880e-01 2.73396999e-01 1.50799945e-01 6.64607108e-01
2.81235486e-01 -2.42828712e-01 -8.68599713e-01 2.60346293e-01
-5.80722950e-02 -4.25803900e-01 -5.15636384e-01 5.13924003e-01
-9.20980334e-01 6.91424370e-01 -1.81297466e-01 5.69959641e-01
-1.56999871e-01 2.78540999e-01 7.47817218e-01 5.73195577e-01
8.93867016e-01 7.07604408e-01 -1.92967221e-01 -1.08082151e+00
-7.98045218e-01 7.50956535e-01 9.60011065e-01 1.43021679e+00
5.20560443e-01 1.19001560e-01 -3.98580581e-01 1.00877285e+00
4.10908937e-01 4.32855725e-01 8.00763071e-01 -1.18367136e+00
6.19328976e-01 5.12484729e-01 4.50726807e-01 -7.43425846e-01
-4.14972514e-01 -6.97671324e-02 -1.61811948e-01 -2.47665942e-01
4.50003296e-01 -8.21098983e-02 -2.71972090e-01 1.39687443e+00
6.06219023e-02 -4.45650369e-01 9.44961386e-04 7.46652842e-01
1.17314708e+00 1.84936710e-02 4.21887338e-01 -2.45987996e-01
9.52890038e-01 -9.95989203e-01 -1.30948532e+00 -1.53144911e-01
1.42371249e+00 -1.11267066e+00 1.38507700e+00 1.10763647e-01
-1.19858885e+00 -3.18000674e-01 -6.68699205e-01 -6.66926622e-01
-5.00028193e-01 -4.26289476e-02 1.76511794e-01 6.85499728e-01
-1.22392690e+00 6.78832889e-01 -4.86154139e-01 -6.94373190e-01
-2.78731603e-02 -1.90838635e-01 -2.27497697e-01 4.11118232e-02
-1.19143975e+00 1.38136601e+00 1.26349226e-01 -9.17464867e-02
-3.19692820e-01 -4.44318861e-01 -9.93879378e-01 -5.54277420e-01
1.28364623e-01 -4.04701047e-02 2.14409709e+00 -9.98636961e-01
-1.52894211e+00 1.24409580e+00 -6.08706117e-01 -2.11522058e-01
6.06296897e-01 -4.59485143e-01 -4.61661607e-01 -3.03801060e-01
4.02244687e-01 2.54315376e-01 1.95035681e-01 -1.16004169e+00
-9.42614079e-01 -2.13017926e-01 -1.52229980e-01 3.92561495e-01
3.19995955e-02 8.03425014e-01 1.49526387e-01 -1.02681673e+00
-2.51221150e-01 -6.97091877e-01 2.79317886e-01 -7.93674529e-01
-3.03215347e-02 -8.78179312e-01 8.66829976e-02 -1.26053596e+00
1.93845034e+00 -1.63878095e+00 3.37372839e-01 -2.68646866e-01
5.61086118e-01 5.52751362e-01 1.22504517e-01 8.62626612e-01
1.44549161e-01 5.02599955e-01 -2.68145502e-01 -9.80098903e-01
1.35807440e-01 2.38002002e-01 3.51588689e-02 3.80994171e-01
-2.96964403e-02 9.47413981e-01 -1.31323373e+00 -2.82479316e-01
-1.47702277e-01 6.07565977e-02 -6.60178661e-01 5.66815212e-02
-9.41319950e-03 4.06381518e-01 1.38940543e-01 4.38836217e-01
2.07901433e-01 4.63216037e-01 -2.35360950e-01 6.60972118e-01
-5.69110572e-01 6.34979844e-01 -7.80141115e-01 1.90632105e+00
-2.81374127e-01 8.22375476e-01 -1.59845322e-01 -4.09240156e-01
9.34848368e-01 5.50377548e-01 -1.19840354e-01 -5.64769626e-01
1.99230418e-01 7.76652157e-01 -1.16568714e-01 -9.64295566e-01
1.29646397e+00 -2.39661917e-01 -2.19172299e-01 6.48104191e-01
3.39223504e-01 -4.56434101e-01 1.57629579e-01 2.36201048e-01
1.00796354e+00 9.00592655e-02 4.07473117e-01 -4.71695334e-01
4.14628863e-01 3.91009182e-01 2.98243970e-01 1.04538739e+00
-7.86433339e-01 5.00879228e-01 -1.13015331e-01 -1.63867816e-01
-1.16226256e+00 -3.71281892e-01 1.06450088e-01 1.44023931e+00
-3.78274947e-01 -9.69348848e-01 -1.31438661e+00 -7.20935702e-01
-3.08950543e-01 1.10179961e+00 -5.40951490e-01 -7.73296654e-02
-6.44755542e-01 2.53534131e-02 1.14721453e+00 5.97467363e-01
2.13242784e-01 -1.25138855e+00 -6.00708306e-01 5.63559711e-01
-7.10733354e-01 -9.03515100e-01 -6.84663713e-01 7.95802101e-02
-4.56045806e-01 -1.03664541e+00 -5.30578256e-01 -9.22619283e-01
2.36443937e-01 5.26859388e-02 1.40514910e+00 6.86633348e-01
1.09157644e-01 9.87366974e-01 -1.34727347e+00 -1.05599451e+00
-9.04052675e-01 -1.30854202e-02 1.66168496e-01 -7.32927680e-01
9.96197402e-01 -6.69311881e-02 1.74923092e-01 -1.83881924e-01
-4.96689856e-01 -1.39807925e-01 -1.87179074e-01 7.68014252e-01
1.36455446e-01 -6.42940938e-01 6.24114752e-01 -1.31363487e+00
1.30627763e+00 -2.27784038e-01 3.37451398e-01 3.00324738e-01
-7.30856895e-01 -2.20956281e-01 3.86066824e-01 1.13344781e-01
-1.33300531e+00 -3.70021015e-01 -5.55380046e-01 1.06836990e-01
-2.35840574e-01 4.57122862e-01 2.27921039e-01 -2.70157252e-02
1.15749276e+00 -2.22231410e-02 -1.09616682e-01 -5.01375079e-01
3.15325141e-01 1.06763840e+00 7.22210467e-01 -7.82828510e-01
1.93662837e-01 -3.04433197e-01 -6.19573653e-01 -7.65358448e-01
-8.68336141e-01 -5.11210442e-01 -8.82521808e-01 -6.62757158e-01
8.44057679e-01 -8.68547261e-01 -3.63878280e-01 3.71415824e-01
-1.69023955e+00 -5.79751015e-01 -7.38046706e-01 4.90048647e-01
-6.14135385e-01 2.54444182e-01 -6.35210872e-01 -1.03383672e+00
-1.36474192e-01 -9.53942001e-01 9.98731911e-01 1.03654347e-01
-9.35463548e-01 -1.32218087e+00 4.02307659e-01 4.22375351e-01
4.98264700e-01 2.29484849e-02 6.83592856e-01 -1.05932713e+00
2.65065879e-01 -1.10743850e-01 2.40658149e-01 4.50104266e-01
-1.55923784e-01 5.88744879e-02 -9.55691576e-01 5.66193610e-02
-1.05899379e-01 -4.99629110e-01 4.76396829e-01 -3.19794357e-01
6.54934108e-01 -6.09070420e-01 3.69179010e-01 9.01618823e-02
1.17024958e+00 -6.07350469e-02 4.62361574e-01 3.82128686e-01
7.52133906e-01 9.45050836e-01 6.36270821e-01 5.95759869e-01
8.58624697e-01 6.56348944e-01 3.86661626e-02 4.94696140e-01
-3.95201147e-01 -6.49035990e-01 4.16332304e-01 1.62786841e+00
-4.26182032e-01 -4.75573868e-01 -1.21753180e+00 8.20702255e-01
-1.98635137e+00 -9.79228556e-01 -9.16856766e-01 1.86082053e+00
1.11205304e+00 -5.16416192e-01 4.65828218e-02 1.83563143e-01
7.23941326e-01 -8.77988413e-02 2.49555275e-01 -1.08086908e+00
-6.99041411e-02 4.42429245e-01 3.03851455e-01 1.25174761e+00
-7.64690220e-01 1.23119223e+00 6.87273073e+00 4.87429768e-01
-4.42732394e-01 7.18496680e-01 -1.85986161e-01 7.86238834e-02
-6.26132309e-01 -3.00172367e-03 -7.92630255e-01 5.81056297e-01
1.05753887e+00 -2.25402087e-01 7.21996605e-01 4.69515204e-01
4.29498345e-01 -4.31743771e-01 -1.04796159e+00 8.23293984e-01
5.63862681e-01 -1.05950928e+00 -2.65532374e-01 -3.50721687e-01
7.46497810e-01 -1.35843784e-01 -5.78027964e-01 7.99158752e-01
6.76238358e-01 -1.06982183e+00 1.35021615e+00 9.49409842e-01
6.89400434e-01 -6.86397731e-01 1.08829713e+00 4.60785151e-01
-5.10672390e-01 2.19696000e-01 -4.32904541e-01 -7.43655324e-01
1.14663839e-01 1.70968860e-01 -8.94530892e-01 2.21215621e-01
8.52809727e-01 6.63098514e-01 -1.01234913e+00 6.33215606e-01
-4.69660372e-01 6.88892722e-01 1.39621809e-01 -5.65529764e-01
-1.20484876e-02 8.91239494e-02 8.61946762e-01 1.77900219e+00
1.55720040e-01 2.88321197e-01 -1.75477955e-02 4.00148809e-01
-2.94848531e-01 4.90665734e-01 -6.99236989e-01 2.68093143e-02
1.02919865e+00 5.67714989e-01 -2.46633187e-01 -4.43866223e-01
-2.57236153e-01 1.34520328e+00 9.61339474e-01 1.31742865e-01
-4.24771488e-01 -5.65971732e-01 4.00120586e-01 2.45779846e-02
-2.51595467e-01 4.13233452e-02 -4.65036303e-01 -9.81507361e-01
1.96779534e-01 -1.07749224e+00 -4.31852713e-02 -7.21831679e-01
-1.14384508e+00 4.21432436e-01 -2.29014605e-01 -8.04613113e-01
-2.73125708e-01 -3.57395202e-01 -4.98804361e-01 1.03378618e+00
-1.15326917e+00 -6.46650732e-01 -4.99405831e-01 3.85961711e-01
9.03587341e-01 -1.00392327e-01 1.10354829e+00 3.11704099e-01
-5.54457963e-01 9.00264084e-01 1.76959679e-01 -2.25202311e-02
1.23367441e+00 -1.74662280e+00 3.78772110e-01 1.12726331e+00
-4.65859361e-02 6.06232285e-01 1.13829339e+00 -9.47291255e-01
-6.31178379e-01 -1.23591852e+00 2.26234722e+00 -1.21218693e+00
4.72712845e-01 -9.95958224e-02 -9.31329072e-01 9.07193899e-01
5.06246328e-01 -3.89398545e-01 8.27739060e-01 9.41898748e-02
4.24164273e-02 6.34714544e-01 -1.15253317e+00 4.43455368e-01
1.36515880e+00 -8.69895339e-01 -1.07929337e+00 5.57954133e-01
1.00830102e+00 -8.84962380e-01 -8.19236875e-01 -9.09247026e-02
5.12227789e-02 -8.30564559e-01 1.63408443e-01 -9.32647347e-01
7.32038498e-01 -1.73421487e-01 -2.87927032e-01 -1.80898774e+00
-3.06794912e-01 -6.95190191e-01 7.67180249e-02 1.72587276e+00
2.91419655e-01 -3.04368824e-01 1.08708836e-01 8.25138450e-01
-1.13323402e+00 -3.36326420e-01 -1.03728044e+00 -6.39283597e-01
3.60712469e-01 -7.89650559e-01 5.61077893e-01 1.13590658e+00
7.00327158e-01 2.42559180e-01 -3.25404316e-01 -3.31522822e-01
4.07733135e-02 -1.16858220e+00 6.08573020e-01 -1.20923150e+00
1.79840252e-01 -4.34571743e-01 -7.04625472e-02 -5.70000231e-01
3.31983984e-01 -1.09357667e+00 5.78546464e-01 -1.50951493e+00
-2.35160068e-01 -1.16409272e-01 6.53559923e-01 4.03786600e-01
-4.42808777e-01 2.60345221e-01 3.71404290e-01 2.75236726e-01
-9.31522548e-01 6.92528009e-01 9.74390745e-01 2.13318735e-01
-3.09787422e-01 -2.72874475e-01 -8.14228654e-01 7.55592883e-01
5.66525698e-01 -7.35517859e-01 -4.69116233e-02 -8.80836189e-01
4.73476529e-01 -3.18017006e-01 2.75556028e-01 -8.81232560e-01
4.82512295e-01 -6.94516897e-02 -2.10170895e-01 3.98136824e-02
-3.97729665e-01 -4.74765718e-01 -1.71184644e-01 2.26345025e-02
-7.67056942e-01 6.27258718e-01 2.68592760e-02 3.21728081e-01
-2.38837793e-01 -1.02596736e+00 4.35783058e-01 -4.55535263e-01
-7.84538209e-01 -4.76301104e-01 -1.01740658e+00 6.77450418e-01
6.84557140e-01 -2.96262026e-01 -3.81997913e-01 -4.48180676e-01
-8.77116442e-01 1.06460854e-01 5.58823705e-01 4.37249541e-01
4.03755665e-01 -1.36738837e+00 -9.68512237e-01 -1.57303885e-01
6.08668327e-01 1.25993580e-01 -1.48141934e-02 8.14384460e-01
-8.40962470e-01 4.70998615e-01 2.81028897e-01 -1.36366300e-02
-1.14721298e+00 3.72454152e-02 4.42717493e-01 5.94076514e-02
-6.44806623e-01 1.15165663e+00 -9.31619227e-01 -8.57190967e-01
5.08848608e-01 -2.91419029e-01 -5.13163328e-01 1.69478655e-01
7.31353581e-01 1.01678598e+00 6.23932540e-01 -1.01632619e+00
-2.74187177e-01 -1.47673100e-01 1.40892342e-02 -3.07713270e-01
1.09841835e+00 -7.93527424e-01 -1.91210330e-01 1.02483940e+00
6.43505037e-01 4.93817508e-01 -6.06561363e-01 -2.15538144e-01
5.61508536e-01 -3.89338762e-01 -3.03224236e-01 -1.24620211e+00
4.09708023e-02 6.40788376e-01 3.65113020e-01 -5.33899777e-02
3.89462650e-01 -1.63209885e-01 5.05991697e-01 5.42269170e-01
3.73510361e-01 -1.85452950e+00 -3.26500148e-01 1.39420509e+00
1.19214046e+00 -1.11777306e+00 -2.49128565e-01 -1.28513753e-01
-8.70202839e-01 1.03026199e+00 8.31474006e-01 1.28442794e-01
2.32592046e-01 -1.54018253e-01 1.40722871e-01 -3.59391868e-01
-4.78086799e-01 -2.25257710e-01 -1.63121317e-02 1.04229379e+00
1.19555032e+00 2.40342721e-01 -1.04227030e+00 1.02043331e+00
-6.07585967e-01 -1.09263569e-01 1.00114501e+00 1.05851328e+00
-6.70072079e-01 -9.80825484e-01 -3.49786401e-01 3.32672864e-01
-2.97655672e-01 -3.48264784e-01 -8.28592300e-01 7.62753069e-01
4.42649096e-01 1.36295605e+00 1.46602029e-02 -3.76442462e-01
8.70513916e-01 6.87101662e-01 4.51075435e-01 -1.17949617e+00
-1.32014763e+00 -6.85556889e-01 5.08210480e-01 -4.53828722e-01
-5.22578835e-01 -8.17480147e-01 -1.27914703e+00 -5.65781891e-01
-2.50333428e-01 3.98853838e-01 5.25736690e-01 1.33249712e+00
1.72744587e-01 5.74748337e-01 -7.70768092e-05 -7.20315516e-01
-5.29070377e-01 -1.48694003e+00 -4.37838674e-01 6.48953080e-01
1.77434176e-01 -4.78765696e-01 -4.01990861e-01 1.30163193e-01]
|
[11.063887596130371, 10.72111701965332]
|
caf44986-6540-45c8-801a-94d2dfd46c51
|
solar-irradiance-forecasting-with-transformer
| null | null |
https://www.mdpi.com/2076-3417/12/17/8852
|
https://www.mdpi.com/2076-3417/12/17/8852/pdf?version=1662438246
|
Solar Irradiance Forecasting with Transformer Model
|
Solar energy is one of the most popular sources of renewable energy today. It is therefore essential to be able to predict solar power generation and adapt energy needs to these predictions. This paper uses the Transformer deep neural network model, in which the attention mechanism is typically applied in NLP or vision problems. Here, it is extended by combining features based on their spatiotemporal properties in solar irradiance prediction. The results were predicted for arbitrary long-time horizons since the prediction is always 1 day ahead, which can be included at the end along the timestep axis of the input data and the first timestep representing the oldest timestep removed. A maximum worst-case mean absolute percentage error of 3.45% for the one-day-ahead prediction was obtained, which gave better results than the directly competing methods.
|
['Iveta Dirgová Luptáková', 'Martin Kubovčík', 'Jiří Pospíchal']
|
2022-09-02
| null | null | null |
mdpi-applied-sciences-2022-9
|
['solar-irradiance-forecasting']
|
['time-series']
|
[-2.79807989e-02 -6.84406757e-02 5.23344129e-02 -2.19799966e-01
-1.64423943e-01 -5.31191051e-01 9.81613040e-01 -8.15415010e-03
-2.05810189e-01 1.20292783e+00 2.80385196e-01 -2.69475102e-01
-3.11919421e-01 -1.13813841e+00 -7.36778140e-01 -9.99996483e-01
1.13442928e-01 4.35947925e-02 -1.52995393e-01 1.13582827e-01
1.32094532e-01 6.58589780e-01 -1.84218144e+00 -2.04396620e-01
1.19048560e+00 1.27824783e+00 5.03120780e-01 7.31349409e-01
-6.49121180e-02 2.42936984e-01 -4.90370572e-01 1.15963146e-01
4.23614323e-01 -4.71364230e-01 -8.51890072e-02 -4.07040507e-01
1.68071866e-01 -2.22338185e-01 -2.19829530e-01 7.25048065e-01
7.20957339e-01 2.88226634e-01 8.05244565e-01 -9.53609467e-01
-6.41560972e-01 1.05957173e-01 -4.24240194e-02 3.49805444e-01
1.46796545e-02 1.14828646e-01 7.20745623e-01 -8.20170701e-01
1.82930604e-01 4.30212796e-01 4.62187529e-01 1.21374212e-01
-7.67648757e-01 -2.64315128e-01 -1.88473463e-01 6.41873777e-01
-1.22954679e+00 -1.54840365e-01 7.83453405e-01 -5.43755174e-01
1.47835767e+00 2.42215037e-01 1.08279967e+00 1.00356340e+00
6.66923404e-01 1.60641566e-01 1.14397001e+00 -4.20004129e-01
5.29235065e-01 -8.57562870e-02 -2.56451547e-01 1.63228437e-01
-3.01668365e-02 7.68124700e-01 -3.57891858e-01 2.91437835e-01
3.19770217e-01 5.62497713e-02 -4.55386668e-01 8.30413215e-03
-9.27141368e-01 8.18416834e-01 8.31506789e-01 3.30943853e-01
-6.72867239e-01 -8.90410990e-02 -3.12634036e-02 7.30710998e-02
5.28614759e-01 1.64013669e-01 -6.84955657e-01 -1.64835095e-01
-1.00437021e+00 -1.14024356e-01 6.62820756e-01 4.60482955e-01
5.36763728e-01 4.71786320e-01 -2.34693184e-01 5.71552575e-01
9.65768471e-02 7.65618861e-01 7.48990715e-01 -7.24744022e-01
-2.37569623e-02 4.68773127e-01 4.72405702e-01 -3.46226811e-01
-6.22777939e-01 -6.60784066e-01 -1.18321836e+00 5.26424646e-01
2.31177118e-02 -6.61923528e-01 -1.24737132e+00 1.35318279e+00
1.35462448e-01 2.68598437e-01 2.44275510e-01 9.28729057e-01
6.02141976e-01 1.23723459e+00 -1.48028079e-02 -7.69592285e-01
8.56982112e-01 -9.08228040e-01 -6.82467699e-01 -2.33682133e-02
1.69626683e-01 -4.49389547e-01 5.76100767e-01 2.62412220e-01
-6.87719762e-01 -6.66641235e-01 -1.04254186e+00 1.70286037e-02
-1.08916664e+00 2.48293012e-01 2.47690096e-01 3.85720193e-01
-1.11286163e+00 9.56930816e-01 -6.00003839e-01 -5.64279735e-01
6.17205277e-02 1.28425449e-01 1.14562824e-01 5.41814864e-01
-1.19672382e+00 1.25048578e+00 6.26697183e-01 3.10263127e-01
-5.24714887e-01 -6.78719461e-01 -4.62916315e-01 5.76378644e-01
4.20372821e-02 -8.45929623e-01 1.09374559e+00 -1.06697536e+00
-1.91547668e+00 2.39202902e-02 -4.08874989e-01 -7.46119320e-01
4.10301447e-01 -2.01051161e-01 -6.10334814e-01 -4.94756699e-01
-2.95355648e-01 4.97360557e-01 9.11113083e-01 -6.77878380e-01
-8.04708064e-01 -2.41422668e-01 -3.54672581e-01 4.82467651e-01
-3.09881747e-01 -5.91475308e-01 1.23777203e-01 -3.19672525e-01
-4.87500668e-01 -9.00419056e-01 -1.11363484e-02 -9.19975266e-02
-3.78444046e-02 -5.47212481e-01 8.88371050e-01 -8.30539107e-01
8.96400571e-01 -1.87402225e+00 2.12614894e-01 -1.40081570e-01
-4.79909003e-01 3.62749517e-01 1.04211621e-01 4.52335060e-01
-2.69085526e-01 -1.11191437e-01 -2.34004244e-01 -4.28617597e-02
7.56829977e-02 3.01055193e-01 -4.63677973e-01 1.36819363e-01
-5.77068925e-02 1.01608539e+00 -6.47364140e-01 3.13825041e-01
6.92517400e-01 7.13497102e-01 3.63208175e-01 1.29013091e-01
-4.90419567e-01 3.59457701e-01 -1.22956201e-01 2.53500998e-01
4.15505350e-01 -1.75605565e-01 -1.88186049e-01 -1.59252644e-01
-7.93779373e-01 2.58758485e-01 -7.16237247e-01 1.46814430e+00
-6.96630657e-01 1.04766464e+00 -5.18496394e-01 -8.01677823e-01
9.18964267e-01 3.13930452e-01 5.52340031e-01 -1.18421030e+00
-2.82255411e-02 -3.44711216e-03 -2.27377594e-01 -3.05795133e-01
3.13667148e-01 -1.15247712e-01 4.29028720e-01 1.50062013e-02
-3.45146358e-01 -2.56314188e-01 1.66715086e-01 -5.86969137e-01
6.54575765e-01 2.96465039e-01 6.36153519e-01 -3.61991733e-01
4.44981396e-01 -2.42140681e-01 6.56127810e-01 2.99149603e-01
7.27710575e-02 5.19117475e-01 -1.11259231e-02 -7.31963992e-01
-1.21580350e+00 -8.95146728e-01 -3.46886575e-01 6.47325099e-01
-2.15516105e-01 2.90744123e-03 -3.53833199e-01 -5.48426926e-01
9.48444456e-02 1.37604678e+00 -5.88903606e-01 1.32100834e-02
-2.46020868e-01 -7.63081610e-01 -2.34593049e-01 7.13505328e-01
5.56465268e-01 -1.24889326e+00 -1.07631087e+00 2.21631452e-01
1.92658961e-01 -7.26302624e-01 1.14607438e-02 5.23509741e-01
-7.07469940e-01 -8.86059582e-01 -1.00600529e+00 -1.82655722e-01
1.75518528e-01 -1.59635112e-01 1.08135128e+00 -3.92210484e-01
-1.05966367e-01 9.07295272e-02 -2.34937772e-01 -8.42787862e-01
1.12896070e-01 1.27189130e-01 1.08818956e-01 -2.56223530e-01
3.08392286e-01 -6.53329670e-01 -7.95442581e-01 -2.70959467e-01
-4.74623591e-01 1.89334661e-01 5.17348230e-01 7.63031602e-01
5.84783375e-01 2.81492144e-01 7.23814309e-01 -1.30230978e-01
4.42784667e-01 -5.51190376e-01 -1.17587543e+00 3.93959492e-01
-9.52153742e-01 3.55316177e-02 1.13459253e+00 -8.26224461e-02
-1.01971436e+00 9.78254601e-02 -3.63998488e-02 -3.87619942e-01
-3.66254926e-01 5.13549447e-01 -4.36995476e-02 -3.59795205e-02
3.15480113e-01 6.89646065e-01 -6.96797550e-01 -3.95085394e-01
2.18014956e-01 4.81586993e-01 2.66510069e-01 1.60212025e-01
5.80709636e-01 -6.70149848e-02 3.50354463e-01 -1.13026094e+00
-5.92153668e-01 7.26771355e-02 -6.08416796e-01 -2.70005673e-01
7.69409120e-01 -8.13331008e-01 -6.76434398e-01 5.43773830e-01
-1.11414766e+00 -5.56282759e-01 -7.24667370e-01 5.01590312e-01
-3.88588905e-01 1.50776580e-01 1.95676342e-01 -9.83941615e-01
-6.97758198e-01 -4.83999789e-01 6.62801325e-01 9.27684546e-01
2.56380200e-01 -1.08989620e+00 2.84267515e-01 -2.93599725e-01
6.53634369e-01 3.02918822e-01 9.96888995e-01 -3.55220288e-01
-5.32145619e-01 2.25403875e-01 -2.13677526e-01 5.40571809e-01
2.55940676e-01 2.10044101e-01 -1.05018997e+00 -2.98082143e-01
-2.22110283e-02 1.09227575e-01 8.97789598e-01 7.55861342e-01
1.29518342e+00 -2.65966684e-01 -2.24363059e-01 6.52030289e-01
1.81414628e+00 6.29646361e-01 6.07398748e-01 1.02281265e-01
2.41896331e-01 1.62679181e-01 2.65526772e-01 4.53622013e-01
3.60244900e-01 3.99052352e-01 8.45023870e-01 -5.69192357e-02
2.42818985e-02 -4.33261581e-02 2.75881857e-01 5.20040870e-01
-1.94227308e-01 -7.48409569e-01 -6.80490911e-01 8.89146388e-01
-1.77623057e+00 -1.01877248e+00 4.31364886e-02 2.58678818e+00
1.55654639e-01 -1.31534457e-01 -1.88285708e-01 3.11387349e-02
1.86761305e-01 4.22747165e-01 -1.07488155e+00 -4.88253862e-01
-3.30422342e-01 2.64230669e-01 5.88452458e-01 6.36861980e-01
-9.30462837e-01 3.73679727e-01 6.76382971e+00 2.97023743e-01
-1.29965556e+00 -1.46926209e-01 3.63859624e-01 -3.74107778e-01
-8.02681148e-02 -1.90572485e-01 -8.06011200e-01 1.04703557e+00
1.37821174e+00 -5.04055142e-01 6.28189921e-01 5.67096412e-01
6.11482918e-01 -5.39560080e-01 -1.00092685e+00 7.09839225e-01
-1.68971699e-02 -1.21565974e+00 -2.55548835e-01 5.44968061e-02
9.81224120e-01 4.59131092e-01 -1.26289800e-01 1.95192888e-01
2.88395435e-02 -1.04252505e+00 1.54595792e-01 1.20751894e+00
5.83365619e-01 -7.21413255e-01 7.15036213e-01 8.01283300e-01
-1.29570723e+00 -4.87223089e-01 -4.60142523e-01 -3.64577472e-01
2.50991613e-01 9.50591207e-01 -7.76716888e-01 7.62590408e-01
9.47697997e-01 9.15743530e-01 -2.31718689e-01 1.35629070e+00
-2.85959601e-01 3.89776319e-01 -9.30130363e-01 -3.61609071e-01
1.96297616e-02 -5.55957139e-01 4.05195475e-01 9.35018480e-01
1.26919401e+00 1.29617527e-01 -2.44647861e-01 5.67729473e-01
-4.76828124e-03 -2.25731134e-01 -9.57683504e-01 1.07814968e-01
4.89986062e-01 1.17537689e+00 -1.70814201e-01 -3.17777872e-01
-6.03234589e-01 9.97664750e-01 1.87073201e-02 6.13894522e-01
-8.09566140e-01 -2.94326782e-01 6.34611845e-01 -1.74030289e-01
8.67219150e-01 -3.41722369e-02 -2.65642077e-01 -9.59968448e-01
1.07271940e-01 4.60540541e-02 2.24530280e-01 -1.43879354e+00
-1.27353442e+00 4.18332398e-01 -1.80540636e-01 -9.06842530e-01
-6.28336906e-01 -6.45694494e-01 -1.00132692e+00 1.31431055e+00
-1.93191457e+00 -9.07502830e-01 -4.97906089e-01 2.62892157e-01
7.88771629e-01 -1.04055710e-01 1.02731776e+00 -8.70041922e-02
-5.58572233e-01 -3.06337010e-02 9.24263716e-01 -4.23524231e-01
2.29925364e-01 -1.43836510e+00 2.71878958e-01 9.56962347e-01
1.52552977e-01 -1.59182861e-01 8.45449686e-01 -4.59144086e-01
-8.93162310e-01 -1.12073171e+00 1.11749196e+00 -1.18113041e-01
4.96741354e-01 1.48079753e-01 -9.12693977e-01 5.58517098e-01
8.65304291e-01 -5.95648997e-02 4.88790184e-01 -2.23981962e-01
2.51449853e-01 -4.38978642e-01 -1.05731344e+00 2.72460639e-01
6.62652612e-01 -3.31316173e-01 -4.85866815e-01 4.46700394e-01
2.65145272e-01 -2.87451118e-01 -1.12467384e+00 6.89510167e-01
4.49969471e-01 -1.14951658e+00 8.00116479e-01 -2.08153412e-01
1.94205478e-01 -4.23762292e-01 -3.07606310e-02 -1.78201914e+00
-4.96076494e-01 -3.36472988e-01 -7.40755796e-01 9.66436923e-01
3.22209001e-01 -7.17950404e-01 7.31493652e-01 4.80167985e-01
-1.26640111e-01 -9.45304513e-01 -1.26380789e+00 -6.05334103e-01
1.52502492e-01 -5.07709794e-02 6.78119838e-01 5.66304386e-01
-5.82664430e-01 2.93906182e-01 -3.00825566e-01 4.72327620e-01
4.46259409e-01 5.63264430e-01 2.89416283e-01 -1.41448379e+00
1.62711293e-01 -4.82873946e-01 8.79093334e-02 -9.20422375e-01
1.33972406e-01 -3.88022542e-01 8.86975527e-02 -1.96994376e+00
-1.72429234e-01 2.80434608e-01 -5.75921953e-01 4.59482431e-01
2.97226906e-02 -9.97357816e-02 2.01310292e-01 -2.11894438e-01
2.13179454e-01 1.19250393e+00 7.82584012e-01 -1.22203752e-01
-2.82941699e-01 3.39605600e-01 -5.14324382e-02 6.65465236e-01
1.15319610e+00 -6.02694266e-02 -5.55964053e-01 -4.24866438e-01
1.72693089e-01 6.49002753e-03 4.36565936e-01 -1.25253272e+00
4.89126563e-01 -4.25831199e-01 9.51976597e-01 -1.11020410e+00
6.25637352e-01 -1.12921977e+00 6.30560338e-01 2.82076925e-01
2.21104562e-01 -1.84398685e-02 2.33890906e-01 5.64849734e-01
7.76394084e-02 -7.80771598e-02 7.06934571e-01 9.61622968e-02
-1.01645720e+00 2.30963781e-01 -3.87189627e-01 -6.32489204e-01
1.30347204e+00 -3.03595930e-01 -4.65022504e-01 -2.48457968e-01
-7.19021857e-01 3.37290674e-01 4.12635863e-01 3.98414314e-01
4.18836415e-01 -1.16842711e+00 -5.37450731e-01 2.76749820e-01
-2.36924857e-01 -1.70864075e-01 2.71091491e-01 5.20019889e-01
-4.33050208e-02 7.24610627e-01 -2.29166180e-01 -4.83983576e-01
-9.98087347e-01 5.96877396e-01 8.28684390e-01 -1.89225018e-01
-7.17685580e-01 4.22777057e-01 -2.09889755e-01 -2.07547415e-02
6.40471056e-02 -3.83307606e-01 -4.25665468e-01 2.29151517e-01
2.63163537e-01 4.46133822e-01 1.65701807e-01 -3.90832961e-01
-1.83364138e-01 9.42539573e-01 5.38205147e-01 1.29617020e-01
1.45850956e+00 -2.06608176e-01 9.78689715e-02 8.01922798e-01
8.09683263e-01 -3.80949259e-01 -1.64025009e+00 1.00552507e-01
-2.42994040e-01 -1.46353006e-01 5.74407518e-01 -1.56089497e+00
-1.00088668e+00 1.04891515e+00 1.26285505e+00 5.31577349e-01
1.53155172e+00 -5.31058192e-01 6.51095629e-01 5.14186144e-01
2.30093718e-01 -1.20913506e+00 -7.83296764e-01 7.17290282e-01
9.70417976e-01 -1.22320056e+00 1.32210866e-01 3.97170037e-01
-2.47592434e-01 1.12331688e+00 4.59573120e-01 1.10768065e-01
7.59466588e-01 4.78664599e-02 -9.84040499e-02 3.48023206e-01
-9.48527813e-01 -2.60290831e-01 4.21148926e-01 5.10097802e-01
1.72463909e-01 9.82744023e-02 -4.69051391e-01 2.74526089e-01
-1.70685858e-01 2.75228173e-01 2.49142662e-01 5.03026962e-01
-5.49916685e-01 -7.09628940e-01 -1.82264373e-01 6.06918991e-01
-1.29715532e-01 -1.55793861e-01 -2.29697287e-01 5.03238440e-01
3.05838943e-01 7.27448463e-01 2.54988700e-01 2.29150970e-02
3.18744749e-01 4.36724365e-01 2.08310992e-01 -4.31820937e-02
-3.93485695e-01 -3.16803843e-01 -1.77039549e-01 -4.48902875e-01
-3.88690591e-01 -5.12585461e-01 -1.04795849e+00 -3.30299556e-01
-1.82588831e-01 1.64834723e-01 1.11477304e+00 1.05146849e+00
7.05745220e-01 6.50878131e-01 8.72926712e-01 -1.15873063e+00
-1.94883317e-01 -1.16133797e+00 -3.34789962e-01 -1.46031380e-01
5.47598600e-01 -4.65533048e-01 -5.34717560e-01 2.43989732e-02]
|
[6.253729820251465, 2.8142998218536377]
|
4ccc1666-1de8-486f-8c18-269ddef9d3d6
|
slot-order-matters-for-compositional-scene
|
2206.01370
| null |
https://arxiv.org/abs/2206.01370v2
|
https://arxiv.org/pdf/2206.01370v2.pdf
|
Towards Improving the Generation Quality of Autoregressive Slot VAEs
|
Unconditional scene inference and generation are challenging to learn jointly with a single compositional model. Despite encouraging progress on models that extract object-centric representations ("slots") from images, unconditional generation of scenes from slots has received less attention. This is primarily because learning the multi-object relations necessary to imagine coherent scenes is difficult. We hypothesize that most existing slot-based models have a limited ability to learn object correlations. We propose two improvements that strengthen slot correlation learning. The first is to condition the slots on a global, scene-level variable that captures higher-order correlations between slots. Second, we address the fundamental lack of a canonical order for objects by proposing to learn a consistent order to use for the autoregressive generation of scene objects. Specifically, we train an autoregressive slot prior to sequentially generate scene objects following the learned order. Slot inference entails estimating a randomly ordered set of slots using existing approaches for extracting slots from images, then aligning those slots to ordered slots generated autoregressively with the prior. Our experiments across three multi-object environments demonstrate clear gains in scene generation quality. Detailed ablation studies are also provided that validate the two proposed improvements.
|
['Anand Rangarajan', 'Sanjay Ranka', 'Pan He', 'Patrick Emami']
|
2022-06-03
| null | null | null | null |
['scene-generation']
|
['computer-vision']
|
[ 7.74629354e-01 3.50191772e-01 -8.86565149e-02 -7.46763825e-01
-9.13522065e-01 -4.00121570e-01 1.06380939e+00 -2.14818746e-01
5.47339581e-02 5.98883331e-01 4.26655680e-01 -1.53399110e-01
-2.38621652e-01 -7.64434934e-01 -1.00943160e+00 -5.52708745e-01
1.39461428e-01 8.19509566e-01 2.11229995e-01 5.89094497e-02
1.88080594e-01 2.63778955e-01 -1.83372319e+00 6.76360726e-01
6.88782454e-01 6.94576144e-01 7.04751134e-01 7.85542727e-01
-2.34123796e-01 1.05139387e+00 -5.37471533e-01 -2.13226229e-01
3.24786603e-01 -7.82765210e-01 -7.37134993e-01 6.85937822e-01
7.08542109e-01 -5.39920151e-01 -1.23708345e-01 6.81132555e-01
2.77512744e-02 2.99928933e-01 8.06296706e-01 -1.34167743e+00
-6.45880938e-01 7.48958111e-01 -4.22835380e-01 -1.58780158e-01
9.84130055e-02 1.46788314e-01 1.40340972e+00 -8.52561414e-01
7.91688323e-01 1.55678487e+00 3.22406769e-01 4.50743586e-01
-1.65581536e+00 -3.87902617e-01 4.21061009e-01 1.48911342e-01
-1.25062585e+00 -5.84014475e-01 7.74877727e-01 -5.14213800e-01
1.02222931e+00 3.07533205e-01 5.12502313e-01 1.13186467e+00
-8.46307874e-02 8.79875839e-01 1.22923291e+00 -5.34100294e-01
1.46997452e-01 2.37308115e-01 -5.22368588e-02 4.85133737e-01
2.92515755e-01 -2.29374003e-02 -8.82950246e-01 1.40691951e-01
1.14358675e+00 -4.11459357e-01 2.36924857e-01 -5.57204664e-01
-1.29623640e+00 7.24961340e-01 3.94423604e-01 6.79863524e-03
-4.85495389e-01 5.20670652e-01 -1.10090129e-01 -3.18351328e-01
4.60142910e-01 7.57126987e-01 -3.05122375e-01 2.61898190e-01
-8.95240128e-01 7.20846951e-01 5.19341946e-01 1.29321015e+00
1.06164169e+00 7.43164495e-02 -3.90200585e-01 7.74972856e-01
4.59538221e-01 3.65738600e-01 2.97652259e-02 -1.14007652e+00
2.48785481e-01 2.02537432e-01 1.47835702e-01 -9.32932436e-01
-2.48691902e-01 -4.64619279e-01 -4.69905555e-01 -2.94663087e-02
2.52501488e-01 1.12088583e-01 -1.04254425e+00 1.93974185e+00
2.56876707e-01 5.18283427e-01 -4.10064198e-02 7.97848344e-01
6.45944953e-01 6.44201159e-01 4.70868737e-01 6.85158744e-02
1.49042153e+00 -1.08999383e+00 -4.65106368e-01 -6.00332260e-01
2.62006640e-01 -9.35154676e-01 1.13635755e+00 1.16124734e-01
-1.14131844e+00 -7.47510552e-01 -8.38787854e-01 -3.09725642e-01
-9.62028205e-02 2.46051565e-01 1.17917621e+00 3.74987036e-01
-9.60695863e-01 1.98999345e-01 -8.08295310e-01 -3.05953115e-01
4.36179429e-01 1.01552457e-01 -2.03149438e-01 4.05818298e-02
-7.64125586e-01 8.24082851e-01 4.48049903e-01 -8.43752362e-03
-1.19496942e+00 -9.73788142e-01 -1.15217590e+00 -5.16468659e-02
4.35711503e-01 -1.26556754e+00 1.41943812e+00 -9.30369139e-01
-1.23272491e+00 9.05452013e-01 -6.31748259e-01 -5.38436532e-01
1.35035157e-01 -4.12407458e-01 1.26862943e-01 1.92786545e-01
5.66438437e-01 1.20197117e+00 8.16149890e-01 -1.81471670e+00
-6.26718462e-01 -5.74903712e-02 4.03069228e-01 5.82550049e-01
1.33570030e-01 -1.81786641e-01 -4.60362792e-01 -5.41426599e-01
4.90223140e-01 -1.00433278e+00 -4.43041623e-01 -7.26967528e-02
-6.53792918e-01 -1.71493635e-01 6.06795073e-01 -2.52649218e-01
6.82092190e-01 -2.05667686e+00 6.91369995e-02 -5.15543669e-02
5.28562488e-03 -4.65214640e-01 -4.01185125e-01 2.20559508e-01
-1.97804764e-01 -1.05999205e-02 -7.99439624e-02 -9.43009198e-01
1.63231656e-01 4.46419388e-01 -8.39153826e-01 2.63947956e-02
7.07044125e-01 9.37528253e-01 -8.68375599e-01 -5.36418796e-01
4.11096901e-01 3.10495019e-01 -1.05165315e+00 4.60946590e-01
-8.44030261e-01 4.64062333e-01 -2.88914591e-01 3.13499331e-01
6.66479826e-01 -5.57338476e-01 2.19641790e-01 -4.18559760e-01
5.95981814e-03 7.27500021e-01 -1.27998614e+00 1.90071130e+00
-4.13705498e-01 5.41191459e-01 -2.99825490e-01 -9.59564865e-01
6.13506854e-01 1.15208082e-01 4.31420535e-01 -4.70744431e-01
-2.60763556e-01 -1.38387352e-01 -2.70336896e-01 -3.49816680e-01
9.31663156e-01 -4.79085535e-01 -1.84389561e-01 4.25475836e-01
2.52763033e-01 -8.70785773e-01 2.99989372e-01 4.57747728e-01
7.30726361e-01 7.32511878e-01 1.01630934e-01 -1.07334994e-01
-1.55417137e-02 2.57116944e-01 5.48289478e-01 1.10169315e+00
2.23759741e-01 9.28483963e-01 4.38173890e-01 -1.73183903e-01
-1.25238144e+00 -1.61386669e+00 -3.72732170e-02 1.12655342e+00
3.51787746e-01 -5.30203819e-01 -4.68987882e-01 -3.21568727e-01
-2.74733335e-01 1.15852702e+00 -5.14501750e-01 1.43509954e-01
-3.96946937e-01 -7.03492343e-01 1.64461359e-01 6.40832305e-01
3.93587291e-01 -1.11519837e+00 -7.33365834e-01 9.08740982e-02
-5.92623591e-01 -1.44190395e+00 -1.09880872e-01 1.32107094e-01
-6.87075317e-01 -8.79775643e-01 -1.21803001e-01 -6.96157694e-01
8.74174595e-01 5.72940290e-01 1.53379285e+00 -7.49211460e-02
-4.68001455e-01 6.38932824e-01 -5.87371700e-02 -5.53453088e-01
-2.40574583e-01 -2.69297093e-01 -1.31747350e-01 -1.86267585e-01
1.81269437e-01 -5.24668813e-01 -3.44573885e-01 1.20753437e-01
-1.06882048e+00 8.36570024e-01 6.48409247e-01 8.82952750e-01
7.05947459e-01 2.77085185e-01 3.33487540e-01 -1.04413903e+00
2.14678720e-01 -3.79354715e-01 -6.67746067e-01 9.75947455e-02
-9.89976749e-02 2.80480444e-01 3.84649597e-02 -2.20322102e-01
-1.71635258e+00 3.61689597e-01 1.61242485e-01 -1.37785837e-01
-5.82088232e-01 4.26309317e-01 -1.68558583e-01 6.75818205e-01
6.31946623e-01 1.00296035e-01 -4.22537178e-01 -9.35352743e-02
7.63809979e-01 -1.36641562e-01 6.22527838e-01 -1.08354473e+00
8.91320944e-01 6.88933253e-01 1.75974786e-01 -8.68747115e-01
-1.21759427e+00 -4.55551684e-01 -6.12433970e-01 -1.46453260e-02
1.08219814e+00 -1.22862458e+00 -3.14005613e-01 2.91923523e-01
-1.44793677e+00 -6.37786984e-01 -5.11186719e-01 4.87041861e-01
-1.05379879e+00 1.36403203e-01 -3.52173597e-01 -9.37023938e-01
3.70813161e-01 -9.61707592e-01 1.58226573e+00 3.13214511e-02
-5.61078906e-01 -8.76531005e-01 -1.09279372e-01 4.23773348e-01
2.26205081e-01 8.77789333e-02 1.01046395e+00 -2.87978984e-02
-1.31805074e+00 2.92961240e-01 -3.44051570e-01 -6.45478368e-02
1.82296649e-01 -9.94260758e-02 -1.13792515e+00 1.87016219e-01
-2.45369337e-02 -4.62616533e-01 8.59712541e-01 6.59152627e-01
1.20969164e+00 -1.71630174e-01 -3.50971937e-01 4.16701734e-01
1.36327732e+00 2.90265642e-02 7.19578385e-01 1.94063619e-01
6.35383844e-01 8.84971261e-01 6.03484392e-01 4.60180283e-01
5.91972709e-01 6.88021064e-01 2.74658620e-01 -8.65936056e-02
-4.53469843e-01 -6.34058058e-01 1.40035108e-01 3.71671230e-01
1.61194533e-01 -1.38084382e-01 -6.19397938e-01 7.45484591e-01
-1.93442488e+00 -1.17324150e+00 -1.78546831e-02 1.88142371e+00
8.48295093e-01 2.20785871e-01 -1.52740911e-01 -2.91546762e-01
4.18317080e-01 2.62964785e-01 -2.67030567e-01 1.93909615e-01
-2.96842009e-01 3.18355024e-01 8.47167745e-02 6.77655995e-01
-1.16498411e+00 1.33934200e+00 6.95280409e+00 5.22709787e-01
-7.55062759e-01 -1.69196486e-01 7.42324293e-01 -5.90750873e-02
-7.55532086e-01 6.29067600e-01 -1.04892910e+00 -1.12502791e-01
5.35958111e-01 -2.84227002e-02 2.90623844e-01 1.07838416e+00
2.66403824e-01 -4.92264837e-01 -1.34204745e+00 8.33945453e-01
1.71122551e-01 -1.41680789e+00 4.07650381e-01 -3.30002862e-03
9.86176074e-01 -4.48282123e-01 2.88305849e-01 2.14141801e-01
6.91090167e-01 -1.04989290e+00 1.13667929e+00 5.40568292e-01
4.36333567e-01 -2.49509603e-01 5.87945320e-02 1.30711600e-01
-1.22225034e+00 1.95409805e-01 -4.53841388e-01 -3.25590402e-01
4.47463006e-01 6.24304593e-01 -1.30766213e+00 3.41884613e-01
3.93557519e-01 6.18647456e-01 -6.55976415e-01 7.88587391e-01
-4.55477029e-01 5.61899126e-01 -2.73473471e-01 3.26238036e-01
1.58816203e-01 -1.03939340e-01 4.19977903e-01 1.05326712e+00
2.88664073e-01 5.37440851e-02 1.95987612e-01 1.40785837e+00
3.08502585e-01 -2.90195256e-01 -6.93905771e-01 1.41921826e-02
3.18734199e-01 1.31242156e+00 -8.54365647e-01 -5.95735669e-01
-3.13759476e-01 9.44321036e-01 2.06040397e-01 4.46641386e-01
-8.27324152e-01 1.94886774e-01 7.02141166e-01 1.48579299e-01
3.68787527e-01 -6.32855237e-01 -7.14650273e-01 -1.10345662e+00
-9.14276987e-02 -7.05130696e-01 1.03123665e-01 -1.25366056e+00
-1.28270626e+00 3.59174788e-01 6.06257498e-01 -1.03857386e+00
-6.41975522e-01 -5.72751164e-01 -3.40148211e-01 8.82693172e-01
-1.39788735e+00 -1.49863350e+00 -3.33917350e-01 4.83587652e-01
9.45116103e-01 1.76926151e-01 8.91747296e-01 -9.55163836e-02
-1.34795338e-01 1.40511900e-01 -7.34988570e-01 -2.19139591e-01
6.04525626e-01 -1.37783611e+00 4.30245370e-01 1.05364025e+00
5.61578870e-01 9.50109124e-01 9.94098008e-01 -7.09850788e-01
-1.14823270e+00 -1.04365206e+00 9.24210489e-01 -8.48611355e-01
4.56364602e-01 -6.57035708e-01 -4.77270246e-01 1.09592557e+00
2.28771374e-01 -2.77546465e-01 6.22146666e-01 3.82463127e-01
-6.35766923e-01 8.03016350e-02 -5.37885427e-01 9.37955499e-01
1.18679404e+00 -6.93215847e-01 -5.39572835e-01 4.92158175e-01
7.48476744e-01 -4.50838029e-01 -3.03483576e-01 4.02806580e-01
4.57348913e-01 -1.10638690e+00 1.15810871e+00 -5.14501154e-01
8.31132054e-01 -5.64468145e-01 -3.29377115e-01 -1.14230955e+00
-4.31615084e-01 -2.67501175e-01 1.66026801e-01 1.20683753e+00
4.66487050e-01 -1.62171870e-01 9.27911401e-01 8.58056188e-01
-2.81089038e-01 -2.42450148e-01 -4.18302745e-01 -5.60707808e-01
-2.41254941e-01 -7.24014163e-01 5.31786382e-01 6.64419293e-01
-4.21250314e-01 7.82725215e-01 -5.68228364e-01 4.06380892e-01
7.59783924e-01 4.09486085e-01 1.26648319e+00 -8.79345059e-01
-7.93727219e-01 -2.30256110e-01 -2.15801284e-01 -1.47514570e+00
2.28082195e-01 -6.14204824e-01 6.53220475e-01 -1.81009138e+00
5.01062572e-01 -6.77225173e-01 7.37618981e-03 3.25399309e-01
-3.79681557e-01 1.91042051e-01 4.11676854e-01 1.26574367e-01
-7.81345427e-01 6.09116852e-01 1.18180215e+00 -1.89194139e-02
-2.53828019e-02 -2.11121604e-01 -9.87769186e-01 7.83744395e-01
4.49229062e-01 -2.68016756e-01 -9.58215654e-01 -6.71439052e-01
1.26021117e-01 -1.76622987e-01 7.21955478e-01 -1.03357852e+00
6.05749227e-02 -5.03313541e-01 5.30002892e-01 -8.49708736e-01
8.63004684e-01 -5.79631627e-01 4.45701063e-01 -1.54044002e-01
-5.39142251e-01 -2.94778287e-01 2.21672192e-01 4.91629034e-01
-6.33789375e-02 -1.41862378e-01 5.44323027e-01 -4.47364450e-01
-9.55594838e-01 1.00421589e-02 -2.99241155e-01 -2.41740584e-01
7.49571979e-01 -1.30294204e-01 -2.59258002e-01 -4.63506311e-01
-7.48041630e-01 -5.51153868e-02 2.59033173e-01 5.45271218e-01
5.51483750e-01 -1.38247037e+00 -5.94421387e-01 1.93593159e-01
2.22272187e-01 3.76483291e-01 3.47886145e-01 3.09676677e-01
-1.46478817e-01 5.59357524e-01 -8.48891288e-02 -9.40266013e-01
-1.03044009e+00 3.77304703e-01 4.44277152e-02 -2.59242415e-01
-3.10963541e-01 1.09229958e+00 9.84072387e-01 -3.38677227e-01
-8.15791115e-02 -2.96742558e-01 2.96354830e-01 -1.71723753e-01
3.34547967e-01 -1.45450845e-01 -2.64118910e-01 -7.30685413e-01
-1.54211605e-02 3.85498494e-01 -1.21582754e-01 -6.32220626e-01
1.27076590e+00 -1.99050248e-01 -7.71128535e-02 5.79028070e-01
7.93415606e-01 -7.35441819e-02 -1.61482465e+00 -2.14033470e-01
-4.64765690e-02 -7.19604969e-01 -3.55822057e-01 -5.04341900e-01
-3.83667052e-01 7.17556894e-01 9.72945243e-02 2.80391518e-02
9.44152594e-01 4.15115923e-01 2.97864377e-01 3.27643007e-01
3.94784540e-01 -9.18327510e-01 6.50908768e-01 5.36516130e-01
9.91035879e-01 -1.13319051e+00 1.01999931e-01 -8.55494797e-01
-6.80435002e-01 8.03579986e-01 1.05951536e+00 -1.05640724e-01
5.38001001e-01 1.61643729e-01 -2.58684363e-02 -2.85319030e-01
-9.19763923e-01 -4.37228709e-01 3.34696382e-01 7.11418986e-01
5.68904102e-01 1.52652726e-01 7.21119344e-02 2.63221681e-01
-4.83632326e-01 -2.33658537e-01 4.12935764e-01 7.78845668e-01
-4.25665557e-01 -1.10521698e+00 -3.34230810e-01 3.64326566e-01
-7.57743344e-02 -3.53139997e-01 -7.66821653e-02 4.93682981e-01
3.28785866e-01 9.96434987e-01 4.25204307e-01 2.40830462e-02
7.13902041e-02 3.81763116e-03 7.50464320e-01 -1.22123861e+00
2.17798457e-01 2.60351568e-01 1.90041438e-01 -5.66505551e-01
-8.40510070e-01 -7.80862033e-01 -1.09021199e+00 2.36095399e-01
-1.93019092e-01 -1.33977637e-01 6.80451751e-01 1.04477894e+00
2.98808128e-01 7.07459927e-01 2.91208655e-01 -1.16192639e+00
-1.60921261e-01 -7.28336334e-01 -4.45986390e-01 8.17582488e-01
1.14143305e-01 -8.88400555e-01 -1.65337294e-01 7.49327064e-01]
|
[10.31635570526123, 0.1875665932893753]
|
56370129-1198-4ad5-87f2-ecad7944ac1b
|
remote-atrial-fibrillation-burden-estimation
|
2008.02228
| null |
https://arxiv.org/abs/2008.02228v1
|
https://arxiv.org/pdf/2008.02228v1.pdf
|
Remote atrial fibrillation burden estimation using deep recurrent neural network
|
The atrial fibrillation burden (AFB) is defined as the percentage of time spend in atrial fibrillation (AF) over a long enough monitoring period. Recent research has demonstrated the added prognosis value that becomes available by using the AFB as compared with the binary diagnosis. We evaluate, for the first time, the ability to estimate the AFB over long-term continuous recordings, using a deep recurrent neutral network (DRNN) approach. Methods: The models were developed and evaluated on a large database of p=2,891 patients, totaling t=68,800 hours of continuous electrocardiography (ECG) recordings acquired at the University of Virginia heart station. Specifically, 24h beat-to-beat time series were obtained from a single portable ECG channel. The network, denoted ArNet, was benchmarked against a gradient boosting (XGB) model, trained on 21 features including the coefficient of sample entropy (CosEn) and AFEvidence. Data were divided into training and test sets, while patients were stratified by the presence and severity of AF. The generalizations of ArNet and XGB were also evaluated on the independent test PhysioNet LTAF database. Results: the absolute AF burden estimation error |E_AF|, median and interquartile, on the test set, was 1.2 (0.1-6.7) for ArNet and 3.1 (0.0-11.7) for XGB for AF individuals. Generalization results on LTAF were consistent with E_AF of 2.6 (1.1-14.7) for ArNet and 3.6 (1.0-16.7) for XGB. Conclusion: This research demonstrates the feasibility of AFB estimation from 24h beat-to-beat interval time series utilizing recent advances in DRNN. Significance: The novel data-driven approach enables robust remote diagnosis and phenotyping of AF.
|
['Joachim Behar', 'Yehoshua Y. Zeevi', 'Meyer Elbaz', 'Mandel Franck', 'Shany Biton', 'Julien Oster', 'Armand Chocron']
|
2020-08-05
| null | null | null | null |
['electrocardiography-ecg']
|
['methodology']
|
[ 1.54096082e-01 -2.33833909e-01 -2.31063843e-01 -4.91607100e-01
-9.03639615e-01 -7.19663322e-01 4.25832532e-02 2.04952389e-01
-3.65258098e-01 1.24601579e+00 1.94921196e-01 -7.99826443e-01
-6.70345783e-01 -6.54487729e-01 -2.48901442e-01 -6.57933295e-01
-8.45954835e-01 3.52524310e-01 -8.21729720e-01 3.23051304e-01
-1.81456044e-01 5.60151875e-01 -7.03947842e-01 1.46302238e-01
8.76451969e-01 1.29798532e+00 -4.62190509e-01 9.17749345e-01
6.19292736e-01 3.20342302e-01 -9.07274961e-01 1.02910087e-01
3.31195325e-01 -7.69508362e-01 -1.78766727e-01 -6.28553629e-01
1.91956103e-01 -3.02879959e-01 -5.33245504e-02 3.32180768e-01
1.35203719e+00 -2.55634964e-01 5.50165832e-01 -9.46824968e-01
-1.22445927e-03 7.73714066e-01 -1.74392641e-01 7.68501997e-01
3.82042043e-02 2.01411307e-01 6.73724353e-01 -6.57824636e-01
3.40720087e-01 4.90449905e-01 1.40630817e+00 3.08975786e-01
-1.38919890e+00 -8.19208801e-01 -3.48226398e-01 -1.34995058e-01
-1.57421958e+00 -3.15562844e-01 5.48322082e-01 -5.66118836e-01
8.78852367e-01 3.60295802e-01 1.46939361e+00 1.12914193e+00
7.23775446e-01 -1.40404582e-01 1.34560466e+00 -3.10718268e-01
1.58365533e-01 -2.01264888e-01 4.65778828e-01 3.52288306e-01
6.07561350e-01 6.13597035e-01 -3.01219702e-01 -8.52309525e-01
5.99501967e-01 6.89285398e-02 -6.32427812e-01 3.31114471e-01
-1.37220442e+00 6.65422916e-01 1.26179308e-01 2.08514094e-01
-7.16738760e-01 -6.79404512e-02 5.14365256e-01 5.35235286e-01
5.37849367e-01 6.38334811e-01 -7.59041429e-01 -5.92436492e-01
-1.21195567e+00 2.16814026e-01 9.57602024e-01 1.91958576e-01
9.20449197e-02 2.14973465e-01 -2.69976079e-01 6.91770196e-01
1.06461290e-02 8.86577725e-01 6.15435958e-01 -9.80569839e-01
1.74090654e-01 2.89673805e-01 6.57024831e-02 -8.64494324e-01
-8.35231125e-01 -1.29664516e+00 -1.40822327e+00 -1.19041584e-01
3.88096035e-01 -6.32204652e-01 -7.66098142e-01 1.79774857e+00
-2.90467262e-01 8.09870102e-03 1.59575313e-01 5.39248228e-01
7.35239744e-01 2.52327025e-01 5.40327653e-02 -8.38899136e-01
1.10968280e+00 -1.12083063e-01 -6.02522612e-01 4.67666471e-03
6.40240729e-01 -2.03443214e-01 7.52122402e-01 6.68559074e-01
-1.03674233e+00 -5.07435381e-01 -9.81595635e-01 6.90075636e-01
1.91418808e-02 5.34494296e-02 2.39973366e-01 1.01163852e+00
-1.06873691e+00 7.46017277e-01 -8.68455589e-01 -3.33195142e-02
7.00122774e-01 3.72978747e-01 -4.90974374e-02 1.94528878e-01
-1.48686326e+00 8.07106495e-01 1.26200035e-01 3.47500861e-01
-9.33427691e-01 -9.51561451e-01 -4.36848402e-01 -5.67431897e-02
-4.08226937e-01 -1.05591273e+00 6.13558471e-01 -7.42634714e-01
-8.62020016e-01 5.93867898e-01 -1.48558453e-01 -8.50481510e-01
6.19405448e-01 -3.00676107e-01 -5.58199942e-01 -9.43681374e-02
1.53253525e-01 2.22725309e-02 5.56314170e-01 -6.54871345e-01
-1.78586110e-01 -6.52699530e-01 -4.36991453e-01 6.39114454e-02
-7.67206699e-02 -2.52328634e-01 5.97118616e-01 -9.74110186e-01
9.82271973e-03 -9.31566417e-01 -2.15253174e-01 -3.54976803e-01
-2.92669326e-01 2.71690816e-01 3.80622715e-01 -1.09257805e+00
1.73414600e+00 -2.11945176e+00 -7.39153326e-02 6.01759493e-01
6.13354027e-01 1.83486134e-01 3.81119162e-01 1.91901043e-01
-3.53615135e-01 4.20721233e-01 -6.48198128e-01 1.81463033e-01
-6.02048039e-01 4.43412969e-03 -2.32648015e-01 6.47966504e-01
-2.82041371e-01 1.13650203e+00 -5.55632770e-01 -1.13410793e-01
-9.11697596e-02 5.88088691e-01 -2.42609248e-01 2.41641724e-03
4.22139257e-01 6.69360995e-01 -7.85335824e-02 5.41922867e-01
4.90849644e-01 -3.04523677e-01 3.18097442e-01 -1.18165560e-01
-9.46409255e-02 2.01027334e-01 -5.63781142e-01 1.54927790e+00
-2.74842352e-01 6.17540061e-01 -3.56837481e-01 -9.90951359e-01
1.17074966e+00 7.02002227e-01 7.33265817e-01 -2.18299806e-01
1.10451885e-01 3.74408185e-01 7.32273102e-01 -1.05064444e-01
-6.81881666e-01 -4.55226213e-01 9.49306265e-02 7.21320093e-01
-2.55119473e-01 1.37590945e-01 -1.54658675e-01 -2.07283556e-01
1.36335397e+00 -3.06377918e-01 6.38757944e-01 -5.62844574e-01
2.19946712e-01 -1.88473016e-01 7.84397185e-01 1.13696039e+00
-4.30660993e-01 6.09888196e-01 5.13121903e-01 -1.03743684e+00
-5.84005594e-01 -1.30159020e+00 -7.89212823e-01 1.12694971e-01
-4.31393534e-01 -6.35274589e-01 -3.70024621e-01 -6.97301686e-01
2.20459148e-01 4.24286216e-01 -5.16174376e-01 -3.38464856e-01
-6.56872749e-01 -1.35050714e+00 1.04361558e+00 7.67625153e-01
4.60591406e-01 -1.12166095e+00 -1.18520665e+00 5.05929887e-01
-3.07783693e-01 -3.21667880e-01 -9.91984904e-02 2.98057020e-01
-1.28386497e+00 -8.32735300e-01 -7.83481777e-01 -2.00546548e-01
1.52893633e-01 -7.15750515e-01 1.47790909e+00 5.97927868e-02
-5.87198436e-01 2.02797964e-01 -2.68559568e-02 -6.87372625e-01
-2.08770335e-01 5.66551536e-02 3.59206766e-01 -2.05634683e-01
2.43535399e-01 -1.03258908e+00 -1.10835648e+00 2.34239429e-01
-4.02630270e-02 -2.49008358e-01 5.73987544e-01 1.02498090e+00
7.48675108e-01 -3.41848910e-01 1.26061237e+00 -5.96247852e-01
7.16008067e-01 -4.70943928e-01 -2.71086872e-01 3.67046334e-02
-1.46506643e+00 -5.25937200e-01 4.46694046e-01 -3.34362388e-01
-3.55710089e-01 -2.83715516e-01 7.22632557e-02 -2.98156172e-01
-9.15397611e-03 7.74200916e-01 1.83684066e-01 2.55829006e-01
9.59617674e-01 1.05208091e-01 8.46243352e-02 -2.66764686e-02
-9.31080580e-02 8.21483433e-01 4.92942035e-01 -4.72851455e-01
2.14631766e-01 3.52350682e-01 1.83605164e-01 -5.35634041e-01
-6.62777066e-01 -1.62734821e-01 -4.42320824e-01 -6.39573410e-02
6.78391755e-01 -1.04813778e+00 -5.63212574e-01 3.57440472e-01
-6.28184855e-01 -4.40582842e-01 -6.25061572e-01 8.34076941e-01
-5.30817091e-01 1.41298592e-01 -3.11957866e-01 -9.46913064e-01
-1.26413441e+00 -5.38374722e-01 4.50027347e-01 -3.88652384e-01
-8.46817195e-01 -9.56170738e-01 3.11427087e-01 1.31852310e-02
6.09322786e-01 9.18280840e-01 8.87479901e-01 -7.18173802e-01
8.59415382e-02 -3.99874240e-01 1.38137072e-01 7.31360853e-01
5.40052772e-01 -2.86353767e-01 -8.30965161e-01 -5.99577308e-01
3.33367705e-01 1.55707344e-01 5.84662139e-01 1.02160740e+00
1.17472017e+00 -2.11652577e-01 -3.58710080e-01 8.83689404e-01
1.12045419e+00 7.37149656e-01 5.13146639e-01 1.04325838e-01
3.58855814e-01 -2.30135500e-01 2.76795775e-01 6.51655853e-01
1.93290357e-02 3.39654326e-01 -8.99404064e-02 -2.01771200e-01
-1.46579286e-02 1.84857622e-01 2.56892920e-01 5.98020613e-01
-6.08313560e-01 -4.81264526e-03 -1.29299235e+00 4.76887584e-01
-1.39964402e+00 -9.23768342e-01 -2.69003436e-02 2.45284104e+00
7.50747800e-01 2.52271324e-01 2.84111917e-01 5.39614081e-01
7.08384335e-01 -8.45563933e-02 -7.72400677e-01 -4.34517294e-01
-3.46143514e-01 4.63287979e-01 1.73858106e-01 3.18573892e-01
-7.97209740e-01 -1.01304509e-01 6.09031200e+00 -6.73333630e-02
-1.38186979e+00 8.36200193e-02 1.21793425e+00 -3.55042070e-01
1.49968430e-01 -8.00950527e-02 -3.76105875e-01 6.64821982e-01
1.64964890e+00 -5.41459262e-01 3.47467571e-01 3.64133090e-01
3.66600543e-01 2.29722977e-01 -1.14771199e+00 1.24379814e+00
-2.17873976e-02 -1.28212357e+00 -2.98750758e-01 1.30872086e-01
4.35508549e-01 3.71883392e-01 -2.60528643e-02 8.03190619e-02
-4.47000146e-01 -1.26121080e+00 4.13526028e-01 7.89218307e-01
1.70152104e+00 -6.12410843e-01 1.07178509e+00 2.11653277e-01
-9.09217477e-01 -4.41198140e-01 2.16371760e-01 -2.24695191e-01
2.61429161e-01 1.21678436e+00 -1.02167869e+00 4.89523113e-01
1.00256145e+00 1.08570457e+00 -2.15572774e-01 7.85319507e-01
1.49214223e-01 1.14671528e+00 -5.39721906e-01 3.71779621e-01
-4.39204842e-01 -7.39963949e-02 6.51568651e-01 9.59937453e-01
5.53368151e-01 2.13242292e-01 -9.06241983e-02 7.61319578e-01
5.65075725e-02 -1.02749489e-01 -6.41097963e-01 -5.86603284e-02
5.96827090e-01 9.45505440e-01 -4.61038977e-01 -5.71360290e-01
-1.56213388e-01 2.55247355e-01 -2.48900831e-01 3.22389483e-01
-6.50246024e-01 -6.00669026e-01 1.67653874e-01 4.59033042e-01
-2.58587062e-01 2.57469237e-01 -9.57354426e-01 -9.28989828e-01
2.12965086e-01 -1.07200861e+00 6.64929926e-01 -4.39049155e-01
-1.33450735e+00 9.57996428e-01 -1.37661725e-01 -1.26061320e+00
-5.64782679e-01 -2.15645507e-01 -5.92821538e-01 1.46510732e+00
-9.62340176e-01 -4.13011014e-01 -3.55676681e-01 4.23756748e-01
7.33231707e-03 -3.55748385e-01 1.45375896e+00 3.49081337e-01
-3.07690173e-01 4.98932034e-01 -1.90979868e-01 8.02439302e-02
6.22586846e-01 -1.18036151e+00 2.64334351e-01 4.97352213e-01
-9.36606154e-02 9.30890918e-01 1.98745370e-01 -8.15553308e-01
-7.94632971e-01 -1.11732733e+00 1.06256509e+00 -8.17863822e-01
4.38111462e-02 2.08837483e-02 -5.49127638e-01 5.25502264e-01
-1.10821329e-01 3.99541229e-01 9.43139493e-01 2.97434866e-01
1.68453809e-02 -4.71378922e-01 -1.31119609e+00 5.73380329e-02
9.18244839e-01 -5.12590170e-01 -6.05230451e-01 7.17568249e-02
-6.59163743e-02 -3.38586032e-01 -1.60923553e+00 1.15675056e+00
1.22857332e+00 -1.01881611e+00 1.00278199e+00 -5.19311786e-01
1.48464227e-02 -7.57324276e-03 4.93182093e-02 -1.27338433e+00
-3.38944256e-01 -9.10824299e-01 -2.26018071e-01 5.67079008e-01
6.06527150e-01 -1.14867151e+00 6.33359969e-01 2.19922408e-01
-1.75658867e-01 -1.25659359e+00 -1.16131735e+00 -5.70448518e-01
1.05654597e-01 -2.99723059e-01 4.60441768e-01 9.68146265e-01
-1.52216747e-01 2.55478233e-01 -1.47511184e-01 -5.08392006e-02
5.22355139e-01 3.05222690e-01 1.77995153e-02 -1.83584368e+00
-3.54112327e-01 -1.18172720e-01 -3.66773456e-01 4.72215228e-02
-1.78120360e-01 -1.25710595e+00 -5.20598412e-01 -1.42715073e+00
1.10205360e-01 -8.10627401e-01 -9.46788371e-01 5.32985806e-01
1.28298048e-02 1.78876847e-01 -1.41681403e-01 3.08386058e-01
4.04867440e-01 8.48283619e-03 6.96978092e-01 -1.67149287e-02
-5.25059402e-01 3.71779293e-01 -5.74879646e-01 5.11948109e-01
1.21154130e+00 -5.02203047e-01 -5.58564842e-01 5.57432603e-03
1.89489886e-01 5.65369248e-01 5.31595290e-01 -1.15673029e+00
-4.48095739e-01 5.35594046e-01 9.58261967e-01 -4.57759649e-01
2.31309086e-01 -5.02960503e-01 6.79553449e-01 9.76196110e-01
-1.94629714e-01 6.77111268e-01 2.81541258e-01 4.01457548e-01
-5.10921516e-02 4.65986252e-01 4.61433202e-01 -1.17270596e-01
3.77228528e-01 4.14424688e-01 -3.45694065e-01 4.11242932e-01
7.41629183e-01 -2.00144708e-01 -2.29902759e-01 -4.33291376e-01
-1.27296841e+00 -2.77913481e-01 -1.35025010e-01 3.89074609e-02
6.89311028e-01 -1.13270545e+00 -1.04655313e+00 5.45011401e-01
-2.36527044e-02 -1.18033454e-01 1.31035671e-01 1.33634567e+00
-4.60196942e-01 2.30551377e-01 -2.70459354e-01 -8.08476865e-01
-1.09982729e+00 4.37588803e-02 7.70179629e-01 -2.10303411e-01
-1.06801116e+00 5.92229605e-01 -2.04660207e-01 1.32021219e-01
1.26200527e-01 -3.45290482e-01 9.54717174e-02 7.31831118e-02
3.47880751e-01 6.13311946e-01 2.32825607e-01 -1.62633076e-01
-5.23694634e-01 6.19269371e-01 2.54136413e-01 -1.41461715e-01
1.14231277e+00 1.04212917e-01 -3.96028966e-01 6.83879972e-01
8.43393385e-01 -1.18494652e-01 -6.41291201e-01 5.10485768e-01
-2.64524400e-01 -5.82168326e-02 -2.50202254e-03 -1.49612153e+00
-9.90237355e-01 8.39758039e-01 1.44008791e+00 1.39520958e-01
1.28007877e+00 -3.16121668e-01 4.85419571e-01 3.14710408e-01
1.88501090e-01 -4.31166261e-01 -6.85945272e-01 -1.04397431e-01
9.05250311e-01 -6.26521409e-01 6.72629923e-02 2.24393472e-01
-2.27489874e-01 9.69198883e-01 -6.23830222e-02 8.45814869e-02
1.19760752e+00 2.41719052e-01 5.49429774e-01 -2.43007988e-01
-6.44609571e-01 8.03877175e-01 -8.81090611e-02 3.07613432e-01
7.02038169e-01 4.39440340e-01 -7.01595724e-01 9.58186507e-01
-3.94642591e-01 5.08785188e-01 3.07716280e-01 8.49814892e-01
1.92197859e-01 -1.02276361e+00 2.93936171e-02 1.25600517e+00
-7.35109806e-01 -3.32891583e-01 -4.14614417e-02 4.52465355e-01
2.87113845e-01 8.19252610e-01 -8.66023153e-02 -1.85677066e-01
1.71106279e-01 5.14725804e-01 2.55519092e-01 -3.04662734e-01
-8.14276636e-01 -9.66622233e-02 2.06098691e-01 -2.55036891e-01
-3.84078592e-01 -7.74077475e-01 -8.47053885e-01 2.72608101e-01
-2.54455894e-01 4.46200103e-01 5.34248352e-01 3.99228275e-01
7.78413177e-01 6.69565618e-01 5.91548502e-01 -2.64998853e-01
-4.43991810e-01 -1.17961073e+00 -8.51237595e-01 4.25906777e-02
8.42768788e-01 -3.89927030e-01 -8.57873619e-01 -1.30864335e-02]
|
[14.3283109664917, 3.2525110244750977]
|
65631162-5de7-42c9-a26a-ee2a80678089
|
online-clustering-of-contextual-cascading
|
1711.08594
| null |
http://arxiv.org/abs/1711.08594v2
|
http://arxiv.org/pdf/1711.08594v2.pdf
|
Online Clustering of Contextual Cascading Bandits
|
We consider a new setting of online clustering of contextual cascading
bandits, an online learning problem where the underlying cluster structure over
users is unknown and needs to be learned from a random prefix feedback. More
precisely, a learning agent recommends an ordered list of items to a user, who
checks the list and stops at the first satisfactory item, if any. We propose an
algorithm of CLUB-cascade for this setting and prove a $T$-step regret bound of
order $\tilde{O}(\sqrt{T})$. Previous work corresponds to the degenerate case
of only one cluster, and our general regret bound in this special case also
significantly improves theirs. We conduct experiments on both synthetic and
real data, and demonstrate the effectiveness of our algorithm and the advantage
of incorporating online clustering method.
|
['Shuai Li']
|
2017-11-23
| null | null | null | null |
['online-clustering']
|
['computer-vision']
|
[-2.10873902e-01 1.40713885e-01 -6.44624829e-01 -2.27983803e-01
-8.77948284e-01 -9.60661709e-01 -9.70270336e-02 2.45398343e-01
-3.90124738e-01 8.82502913e-01 -9.42539051e-03 -5.67818642e-01
-5.30506551e-01 -5.80294549e-01 -1.07928479e+00 -8.90237987e-01
-6.42077267e-01 7.49498844e-01 5.00595905e-02 2.10191488e-01
-1.70829237e-01 -4.09342721e-02 -1.05011189e+00 2.22909361e-01
9.19294536e-01 1.10461485e+00 9.30234939e-02 5.94769597e-01
-5.60014546e-02 5.07647991e-01 -2.99050152e-01 -4.63919163e-01
7.72715688e-01 -6.95531666e-01 -8.33176315e-01 6.70778036e-01
-3.26137245e-02 -9.53379869e-02 -1.62092969e-01 1.15804124e+00
1.71180695e-01 1.04047440e-01 3.17482978e-01 -1.08817494e+00
-6.08337581e-01 1.29714501e+00 -9.45912302e-01 2.91246086e-01
8.32178667e-02 -1.04701579e-01 1.59133232e+00 -2.45191962e-01
2.67056495e-01 8.41175616e-01 4.48014498e-01 2.94589847e-01
-1.32875311e+00 -7.10512221e-01 7.31473625e-01 7.22841546e-02
-1.14307559e+00 -9.86548811e-02 6.13032281e-01 -2.06150770e-01
3.23285848e-01 4.32578355e-01 8.15808475e-01 4.31665868e-01
-6.11589015e-01 1.15154612e+00 1.20042813e+00 -4.81537521e-01
8.02914917e-01 9.51980352e-02 3.16001564e-01 6.20688498e-01
4.58049119e-01 -6.80043623e-02 -4.18536007e-01 -4.40283120e-01
6.77835345e-01 3.78098458e-01 -2.55592644e-01 -5.11533618e-01
-8.46336961e-01 9.32112992e-01 4.53144789e-01 -4.37741987e-02
-3.52445036e-01 3.14946830e-01 -4.25438955e-02 7.62327731e-01
6.14096880e-01 4.26228270e-02 -5.32871723e-01 -5.21000996e-02
-9.36371207e-01 -2.78380252e-02 1.15684128e+00 1.18744743e+00
7.86193669e-01 -3.20879221e-01 3.63589860e-02 7.33496666e-01
-2.46373266e-02 4.98213857e-01 -7.48322234e-02 -1.22343218e+00
6.41631782e-01 4.76684302e-01 7.63888836e-01 -5.70843160e-01
-2.65852481e-01 -8.18816006e-01 -7.57111907e-01 -1.80288002e-01
5.61275959e-01 -6.85875654e-01 -4.92433488e-01 1.66964853e+00
4.72324580e-01 2.34093264e-01 -3.91109079e-01 1.22233772e+00
-1.99081540e-01 5.35738885e-01 -4.07266706e-01 -9.26205158e-01
6.56451166e-01 -1.14009356e+00 -4.28964823e-01 -1.26032382e-01
4.04519111e-01 -4.09776807e-01 8.05246234e-01 5.76334000e-01
-1.21013761e+00 2.32831046e-01 -7.18580663e-01 5.86787939e-01
-5.83743937e-02 -1.92168817e-01 1.09481418e+00 8.13167691e-01
-1.14109254e+00 5.91748953e-01 -7.55018413e-01 -1.95454955e-01
4.11063403e-01 4.67142195e-01 4.08919781e-01 -1.39217600e-01
-6.48629844e-01 -3.77541147e-02 2.30215386e-01 -7.77741745e-02
-7.79411376e-01 -5.43019116e-01 -3.36057357e-02 2.57719964e-01
1.02664459e+00 -7.44314194e-01 1.48776388e+00 -1.51491261e+00
-1.32974887e+00 3.72801244e-01 -2.56405681e-01 -6.58926547e-01
5.40688694e-01 -1.57517388e-01 -1.03928365e-01 5.37033677e-02
8.49532112e-02 -2.24479765e-01 6.01970494e-01 -1.15657830e+00
-1.04185557e+00 -4.60082352e-01 5.48483670e-01 2.24718332e-01
-4.18753415e-01 -1.09470196e-01 -5.28108835e-01 -3.11100900e-01
2.66131554e-02 -1.08072698e+00 -5.46664536e-01 -2.08724424e-01
-2.82891661e-01 -1.96996421e-01 2.90942043e-01 -1.66514650e-01
1.35756862e+00 -1.99991679e+00 -4.01258580e-02 6.77702785e-01
9.65806842e-02 -2.88957983e-01 1.29123777e-01 5.69329262e-01
2.30681673e-01 3.41400236e-01 -7.04341903e-02 -2.29217619e-01
1.26212403e-01 1.64820209e-01 -2.19675675e-01 5.47467053e-01
-7.92906880e-01 4.59116668e-01 -8.91272783e-01 -2.14389548e-01
-2.94455141e-01 -4.66252744e-01 -9.40127492e-01 -3.60111371e-02
-6.06874883e-01 1.24956213e-01 -6.32948875e-01 5.18763125e-01
5.41123331e-01 -9.60164368e-01 5.92538178e-01 5.11241734e-01
-7.56999850e-02 8.16739723e-02 -1.66059363e+00 1.25534654e+00
-4.96796489e-01 5.97507581e-02 6.71296537e-01 -1.06537294e+00
9.62332264e-02 3.30990314e-01 6.41067863e-01 -2.33266756e-01
-6.44990355e-02 1.22709803e-01 -1.07377231e-01 -1.96515128e-01
-7.01217055e-02 -1.16385050e-01 -2.73405947e-02 1.12636888e+00
-3.43826056e-01 6.83189690e-01 1.64148182e-01 5.95098615e-01
1.23324692e+00 -4.06170785e-01 1.56621709e-01 -3.03279459e-01
-1.11622177e-02 3.85660753e-02 5.42573929e-01 1.40285826e+00
-9.27532166e-02 1.50968164e-01 5.98160088e-01 -3.77966017e-01
-1.00971043e+00 -8.47488403e-01 5.25994539e-01 1.70461798e+00
2.65158623e-01 -1.17712446e-01 -7.09883690e-01 -6.87735677e-01
3.58261853e-01 4.44589704e-01 -6.95594549e-01 4.32241827e-01
-1.25520006e-01 -8.43984365e-01 -2.20873073e-01 1.76154807e-01
3.59971851e-01 -4.67872381e-01 -2.56334245e-01 4.60674554e-01
-2.60534942e-01 -6.57848716e-01 -1.06021726e+00 1.93118840e-01
-9.61110055e-01 -1.08178449e+00 -3.45831335e-01 -7.52435744e-01
6.75344646e-01 7.23966122e-01 9.00571704e-01 1.65593043e-01
1.26719788e-01 4.28935885e-01 -4.01305318e-01 -2.19409019e-01
1.74759403e-01 2.22015798e-01 9.93398502e-02 4.42744404e-01
1.46234870e-01 -6.93788767e-01 -1.09730351e+00 4.03642744e-01
-7.01452613e-01 -8.66452903e-02 4.73594427e-01 6.99480295e-01
5.77207267e-01 1.26516208e-01 7.49852479e-01 -1.33154571e+00
5.35790205e-01 -7.74438918e-01 -9.28181529e-01 4.12141651e-01
-6.96519494e-01 -5.11119738e-02 9.94777262e-01 -3.80116165e-01
-7.52029598e-01 3.36703598e-01 5.16734898e-01 -3.95520717e-01
1.48912191e-01 6.15976870e-01 1.62259564e-01 1.46654099e-01
4.06597614e-01 1.61305219e-01 -2.44117811e-01 -5.27489364e-01
5.75894713e-01 6.46951437e-01 2.52794862e-01 -7.94649780e-01
6.92537785e-01 6.99604630e-01 -2.88092643e-01 -4.06196445e-01
-1.20062768e+00 -7.72996068e-01 -2.11915284e-01 -1.91247240e-01
-4.04803306e-02 -1.01171446e+00 -1.38683748e+00 -1.55674577e-01
-6.50499642e-01 -5.75847626e-01 -1.77709371e-01 2.80817568e-01
-6.49755478e-01 2.97998041e-01 -7.58803189e-01 -1.33560276e+00
-2.74462044e-01 -4.98004645e-01 3.95022482e-01 1.13670900e-01
3.83071333e-01 -8.55425119e-01 -5.29775955e-02 4.51192558e-01
2.33396456e-01 -5.70799261e-02 7.40215302e-01 -6.52945042e-01
-9.48581040e-01 -8.71692225e-02 -1.26128113e-02 -7.21027702e-02
6.74490780e-02 -3.22244018e-01 -4.85591114e-01 -7.24468827e-01
-2.35759571e-01 -2.99908876e-01 7.87544370e-01 5.23047686e-01
1.32951641e+00 -9.97324347e-01 -5.06195247e-01 2.80519068e-01
1.66098046e+00 1.96612716e-01 -2.32290298e-01 5.71889952e-02
1.95383295e-01 7.95722529e-02 5.50350606e-01 8.89902592e-01
2.76109457e-01 3.94207567e-01 5.34547091e-01 1.73469838e-02
4.62091565e-01 -3.34783256e-01 1.79647028e-01 6.70110166e-01
-7.11256117e-02 -2.72619516e-01 -5.55651009e-01 9.70788002e-01
-2.33723950e+00 -9.12172973e-01 1.67071909e-01 2.55207038e+00
1.08596063e+00 2.24587187e-01 7.25483179e-01 -1.10708019e-02
8.55957508e-01 -1.39771596e-01 -9.18092549e-01 -1.90544158e-01
4.75431681e-02 -5.23440801e-02 9.67006624e-01 6.57080054e-01
-8.95438552e-01 8.18650246e-01 6.21005917e+00 8.51327896e-01
-8.13606858e-01 3.78446043e-01 9.19884980e-01 -8.14659894e-01
-2.40985796e-01 2.19347954e-01 -5.13329864e-01 6.41188741e-01
9.33354497e-01 -3.59622121e-01 1.26330757e+00 9.16533291e-01
5.80087066e-01 -1.82657167e-01 -1.13641870e+00 7.77720690e-01
-3.78525198e-01 -1.41583931e+00 -3.58544052e-01 2.21073225e-01
1.26600063e+00 1.75109014e-01 -3.00781592e-03 7.32722059e-02
1.25078368e+00 -5.74690223e-01 5.76048732e-01 4.81261052e-02
4.04183596e-01 -1.15449405e+00 1.98891729e-01 9.29726243e-01
-9.58153903e-01 -7.17330933e-01 -2.69973963e-01 -3.44668597e-01
2.39081820e-03 7.85948753e-01 -7.14745283e-01 4.97088194e-01
9.06675756e-01 3.90467405e-01 -6.67705908e-02 1.47735870e+00
9.55785960e-02 1.19978416e+00 -7.23057151e-01 -3.97449523e-01
4.19430315e-01 -6.23207033e-01 3.77154350e-01 1.14567733e+00
5.69179542e-02 6.90098405e-01 6.71923280e-01 4.67872947e-01
-3.80802870e-01 2.58771598e-01 -1.64892122e-01 1.29772678e-01
8.16158593e-01 1.11650836e+00 -7.89581478e-01 -4.91865665e-01
-4.56019402e-01 8.63906384e-01 6.71065748e-01 4.91713494e-01
-7.80466914e-01 -3.77707854e-02 5.04886806e-01 1.48507640e-01
9.14441824e-01 -1.11593142e-01 -4.85084385e-01 -1.01359499e+00
6.80496963e-03 -5.18633485e-01 8.40189159e-01 -2.22039431e-01
-1.44540167e+00 -8.13997909e-02 -3.96777123e-01 -1.00385034e+00
6.79542571e-02 -1.50010921e-02 -5.67374051e-01 2.79631436e-01
-1.15610957e+00 -7.11862206e-01 3.35629255e-01 8.23890746e-01
4.85485554e-01 2.50711888e-01 4.09210980e-01 1.82320029e-01
-5.01538277e-01 8.12209368e-01 1.11850727e+00 -1.37464553e-01
4.35387999e-01 -1.46509099e+00 -4.23231632e-01 7.02315867e-01
4.20290790e-02 6.40532613e-01 8.68744612e-01 -4.27978694e-01
-1.56973708e+00 -1.27429008e+00 5.10766983e-01 -2.82571409e-02
9.32357967e-01 -4.68028218e-01 -3.68059933e-01 9.14977372e-01
2.69353926e-01 3.26419598e-03 9.60358500e-01 4.62818205e-01
-2.68860161e-01 -4.55400735e-01 -1.11500359e+00 6.26384199e-01
1.41596603e+00 -7.44474009e-02 1.53425727e-02 7.04375863e-01
8.92454684e-01 -2.12279856e-01 -8.67621601e-01 -2.58171260e-01
5.00011206e-01 -8.57937515e-01 3.91355485e-01 -7.16735005e-01
-1.30940722e-02 -2.21785858e-01 -1.11851342e-01 -1.40868998e+00
-5.01843810e-01 -1.17977512e+00 -3.24333012e-01 8.74987304e-01
8.78409386e-01 -5.06194830e-01 1.09554398e+00 6.57598615e-01
1.68727502e-01 -6.81362867e-01 -9.56642687e-01 -6.75522864e-01
1.17343165e-01 -2.35604167e-01 5.40515006e-01 9.37894583e-01
6.15554690e-01 2.41747141e-01 -5.98837078e-01 3.56196493e-01
9.93388951e-01 9.22515750e-01 6.00033581e-01 -1.00506723e+00
-1.01888096e+00 -2.33090371e-01 6.51833415e-01 -1.51925480e+00
-2.57480562e-01 -7.56312311e-01 -1.53830260e-01 -1.23049057e+00
7.18273282e-01 -7.21711755e-01 -4.83621955e-01 3.76326621e-01
2.28945492e-03 -9.24786106e-02 3.03346425e-01 2.61853963e-01
-1.57117903e+00 8.72298330e-02 9.76202309e-01 1.68854967e-02
-3.28027666e-01 7.22940683e-01 -1.05362570e+00 4.78260756e-01
7.34106243e-01 -4.75871474e-01 -3.11830878e-01 -3.92188609e-01
4.52907294e-01 6.49133086e-01 -4.16445471e-02 -4.48891789e-01
7.06515729e-01 -5.76407850e-01 3.41728255e-02 -6.79910302e-01
-2.04458274e-02 -1.08866060e+00 1.95070043e-01 4.89239782e-01
-6.26263022e-01 -1.74239114e-01 -3.58647883e-01 1.29049373e+00
2.91673362e-01 8.51716623e-02 6.23741269e-01 -4.27441090e-01
8.65165517e-02 5.60381711e-01 -3.06824714e-01 5.19073643e-02
1.08776450e+00 -3.85740921e-02 -1.59877941e-01 -8.68385613e-01
-1.05485928e+00 7.43501365e-01 4.72159654e-01 -8.58818889e-02
5.99904209e-02 -1.19124103e+00 -5.06261587e-01 -2.47315615e-01
-1.53566882e-01 -1.10849835e-01 6.53377548e-02 8.56866062e-01
1.18069179e-01 4.56716031e-01 4.79405016e-01 -3.86760145e-01
-1.02637446e+00 9.53474462e-01 3.16544652e-01 -2.42854729e-01
-2.92890877e-01 6.64050281e-01 -1.27084300e-01 -8.52644816e-02
6.04518056e-01 -1.34994850e-01 3.63423139e-01 -1.84681386e-01
3.51127714e-01 5.34360647e-01 -5.94370887e-02 1.90886855e-01
-3.41245607e-02 -1.08285435e-02 -3.19115430e-01 -3.10741872e-01
1.44016910e+00 -4.80126292e-01 -1.48954242e-01 4.89085943e-01
8.60592544e-01 9.89326239e-02 -1.34082890e+00 -8.18837583e-01
1.20953090e-01 -4.18141216e-01 -3.49339664e-01 -9.29828346e-01
-1.25350142e+00 1.07821010e-01 4.55066949e-01 8.22357535e-01
1.01445532e+00 2.80665100e-01 5.53515375e-01 4.31928784e-01
8.58715892e-01 -1.39600205e+00 -3.27102616e-02 1.84727237e-01
2.85805255e-01 -1.06073153e+00 -1.44386679e-01 -2.46856064e-01
-4.64424729e-01 7.05747783e-01 3.13260615e-01 -3.33992362e-01
9.15424168e-01 -9.00829583e-02 -4.69610959e-01 2.09509358e-02
-1.24728835e+00 -3.02930802e-01 -3.91008168e-01 3.40336598e-02
2.09172174e-01 5.53735375e-01 -5.90403914e-01 8.79935920e-01
-8.44787657e-02 9.39862952e-02 5.59601307e-01 8.05176139e-01
-8.36989045e-01 -8.70661855e-01 -4.02630746e-01 6.97317958e-01
-9.46480513e-01 1.26980171e-02 -4.61136222e-01 3.09466541e-01
-4.77053449e-02 1.29907250e+00 -1.03129465e-02 -1.16973117e-01
-4.98676822e-02 -2.01334476e-01 1.62651271e-01 -5.22485912e-01
-5.92575729e-01 6.53902531e-01 -2.15000939e-02 -4.61436838e-01
-4.72375482e-01 -7.55066514e-01 -9.85894978e-01 -5.51760793e-01
-5.28487146e-01 7.46541381e-01 3.47091943e-01 8.59892666e-01
4.41152126e-01 -8.52238759e-02 1.42063797e+00 -2.83930033e-01
-7.34341145e-01 -7.82040894e-01 -9.60840702e-01 3.36756289e-01
4.81064349e-01 -3.31901908e-01 -4.54334527e-01 -6.16943575e-02]
|
[4.595046043395996, 3.383399724960327]
|
b43387e3-9dcd-4fb7-877c-85c3755b4f5d
|
domain-adaptation-in-multilingual-and-multi-1
|
2205.07283
| null |
https://arxiv.org/abs/2205.07283v1
|
https://arxiv.org/pdf/2205.07283v1.pdf
|
Domain Adaptation in Multilingual and Multi-Domain Monolingual Settings for Complex Word Identification
|
Complex word identification (CWI) is a cornerstone process towards proper text simplification. CWI is highly dependent on context, whereas its difficulty is augmented by the scarcity of available datasets which vary greatly in terms of domains and languages. As such, it becomes increasingly more difficult to develop a robust model that generalizes across a wide array of input examples. In this paper, we propose a novel training technique for the CWI task based on domain adaptation to improve the target character and context representations. This technique addresses the problem of working with multiple domains, inasmuch as it creates a way of smoothing the differences between the explored datasets. Moreover, we also propose a similar auxiliary task, namely text simplification, that can be used to complement lexical complexity prediction. Our model obtains a boost of up to 2.42% in terms of Pearson Correlation Coefficients in contrast to vanilla training techniques, when considering the CompLex from the Lexical Complexity Prediction 2021 dataset. At the same time, we obtain an increase of 3% in Pearson scores, while considering a cross-lingual setup relying on the Complex Word Identification 2018 dataset. In addition, our model yields state-of-the-art results in terms of Mean Absolute Error.
|
['Mihai Dascalu', 'Dumitru-Clementin Cercel', 'Răzvan-Alexandru Smădu', 'George-Eduard Zaharia']
|
2022-05-15
|
domain-adaptation-in-multilingual-and-multi
|
https://aclanthology.org/2022.acl-long.6
|
https://aclanthology.org/2022.acl-long.6.pdf
|
acl-2022-5
|
['lexical-complexity-prediction', 'complex-word-identification']
|
['natural-language-processing', 'natural-language-processing']
|
[ 4.17153329e-01 -7.56046399e-02 -3.09241749e-02 -3.77158597e-02
-8.09597492e-01 -5.90989888e-01 9.51997817e-01 6.26548052e-01
-9.67460930e-01 6.60320520e-01 1.63245350e-01 -2.86679536e-01
-1.68555111e-01 -6.06307089e-01 -5.28866291e-01 -4.11016285e-01
4.69404399e-01 5.34711957e-01 3.32743861e-02 -3.94246429e-01
1.58528119e-01 3.33004415e-01 -1.59181082e+00 -7.94812664e-02
1.37622941e+00 7.40838647e-01 2.46603176e-01 2.76800066e-01
-3.86035800e-01 1.40241105e-02 -7.23154068e-01 -7.89625704e-01
1.74951509e-01 -1.19582698e-01 -5.80426097e-01 -2.27210283e-01
5.85906863e-01 1.49927363e-01 -4.72350381e-02 1.03003895e+00
5.13408601e-01 2.07964033e-01 6.94342494e-01 -7.20194399e-01
-4.79567915e-01 7.41329134e-01 -3.30251127e-01 -1.43090382e-01
2.27402002e-01 1.64728060e-01 1.12045014e+00 -7.67923594e-01
7.12884784e-01 1.10091591e+00 6.62015736e-01 5.60392082e-01
-1.44539928e+00 -5.68379819e-01 3.82252991e-01 3.24834883e-01
-1.35442173e+00 -3.32900226e-01 6.70630038e-01 -3.81317139e-01
9.30810928e-01 2.84791648e-01 3.77272755e-01 1.27633226e+00
-2.11290970e-01 6.99734688e-01 1.02154016e+00 -7.24834919e-01
8.18001479e-02 2.41449758e-01 2.31506974e-01 8.87653604e-03
5.78694761e-01 -2.51532346e-01 -3.02977026e-01 2.01270744e-01
1.56362146e-01 -3.12552869e-01 -2.04193085e-01 -2.62608975e-01
-1.11848068e+00 7.40600228e-01 1.65471658e-02 5.07971823e-01
-1.56459868e-01 -4.00854051e-01 6.89398110e-01 1.60414964e-01
5.86370409e-01 8.60745311e-01 -4.91801441e-01 -2.81447262e-01
-1.06963134e+00 4.26521897e-01 7.30887949e-01 7.45981455e-01
4.97770607e-01 1.62075162e-02 -3.12083960e-01 1.18412483e+00
-6.47175610e-02 6.76884174e-01 5.31728446e-01 -3.52225751e-01
7.42018163e-01 7.04285741e-01 -7.36702532e-02 -7.61817694e-01
-5.72807074e-01 -6.83780849e-01 -9.43895042e-01 1.55011147e-01
6.16355538e-01 9.93542522e-02 -7.99810648e-01 1.95397067e+00
2.45485142e-01 -1.66684270e-01 -5.32401502e-02 5.85516691e-01
4.95523542e-01 4.27733541e-01 4.17976022e-01 -1.52115181e-01
1.45269036e+00 -8.43755066e-01 -6.88839197e-01 -3.91535223e-01
8.02069306e-01 -9.53252137e-01 1.55964303e+00 7.77291000e-01
-7.63280869e-01 -5.13828993e-01 -9.90969658e-01 -2.25125507e-01
-7.19251454e-01 3.21213275e-01 4.29579198e-01 7.39682436e-01
-6.62415147e-01 6.70788229e-01 -4.83842254e-01 -5.25973678e-01
1.69689357e-01 2.03729361e-01 -5.12401938e-01 -2.12984219e-01
-1.18143904e+00 1.25314760e+00 6.05526328e-01 -2.37943187e-01
-1.41407728e-01 -1.03022516e+00 -7.63897181e-01 1.76813051e-01
4.98863280e-01 -4.69952345e-01 8.15129757e-01 -5.83132923e-01
-1.52495456e+00 7.18587995e-01 -1.87605843e-01 -6.04466558e-01
8.29033315e-01 -5.60881615e-01 -5.14159083e-01 -3.33980709e-01
-1.16644725e-01 3.53811771e-01 8.26012373e-01 -8.64210010e-01
-6.07555091e-01 -2.75053203e-01 -2.63637379e-02 -1.33009963e-02
-7.30998516e-01 -2.64828056e-01 -7.27744639e-01 -1.01976192e+00
-4.80576485e-01 -9.94628489e-01 -3.20552886e-02 -3.03541362e-01
-3.57972056e-01 -4.31529224e-01 5.05184889e-01 -9.61595953e-01
1.76454771e+00 -2.07493830e+00 5.31642616e-01 1.00207880e-01
-4.61049203e-04 8.76062810e-01 -3.05334717e-01 3.41163337e-01
-1.13214992e-01 2.83582121e-01 -5.34614801e-01 -8.05207610e-01
1.69344679e-01 3.77799235e-02 -3.57535928e-02 1.78319573e-01
5.16457856e-01 6.91646397e-01 -7.37581909e-01 -3.04715842e-01
3.60763878e-01 4.85293627e-01 -7.69592285e-01 -3.32356133e-02
-2.39948615e-01 5.98412991e-01 9.40904990e-02 1.83559433e-01
8.40780675e-01 2.31289595e-01 3.91053975e-01 -1.81203112e-01
-2.46880129e-01 4.43678379e-01 -1.26695609e+00 1.89175153e+00
-9.91302013e-01 4.76976067e-01 -1.95501566e-01 -1.10727179e+00
9.08434570e-01 6.53308630e-02 1.26996905e-01 -8.46436679e-01
1.68623686e-01 3.37415904e-01 1.08945847e-01 -1.57583177e-01
6.59168661e-01 -1.65583361e-02 -3.21605086e-01 2.44071275e-01
1.32167190e-01 -1.42562807e-01 6.65039599e-01 -1.00886755e-01
8.40646505e-01 1.92195207e-01 3.53597343e-01 -3.44677478e-01
9.67831850e-01 -2.77005762e-01 4.51582789e-01 5.37788212e-01
8.84162337e-02 3.96122783e-01 3.95244151e-01 -2.25945756e-01
-1.21798778e+00 -7.36567736e-01 -3.15116018e-01 9.24353242e-01
-2.11419329e-01 -6.70103014e-01 -8.42432678e-01 -6.02637053e-01
1.25498906e-01 1.10008276e+00 -6.14552498e-01 -3.26776892e-01
-7.73909152e-01 -6.88638747e-01 4.91731495e-01 2.37932950e-01
2.18598455e-01 -9.18456554e-01 -2.83446342e-01 1.74240857e-01
-2.49730080e-01 -1.44656932e+00 -2.96626329e-01 1.90554962e-01
-7.02463686e-01 -8.76135647e-01 -7.19449162e-01 -4.35138851e-01
1.58949941e-01 -6.80966675e-03 1.20099103e+00 -1.62066165e-02
-2.12737232e-01 -1.33410946e-01 -5.45509517e-01 -4.48862553e-01
-6.16777539e-01 7.44774699e-01 7.22362399e-02 -7.73523524e-02
3.86459708e-01 -4.59033757e-01 -3.54679167e-01 -4.76019718e-02
-1.13153338e+00 -7.20600486e-02 8.13307881e-01 8.40755105e-01
4.68836159e-01 -1.17122069e-01 5.43509126e-01 -1.10986459e+00
6.28278255e-01 -4.30666029e-01 -6.72791600e-01 3.54365587e-01
-9.98710573e-01 3.21244389e-01 1.04056084e+00 -6.20650053e-01
-1.01908231e+00 -1.14774115e-01 -3.33994359e-01 1.26721337e-02
-2.58194327e-01 4.44000393e-01 -3.13315839e-01 1.77020624e-01
6.32729232e-01 -8.16230197e-03 -2.00529262e-01 -9.53033924e-01
5.50227225e-01 6.02651834e-01 4.01006132e-01 -5.14103115e-01
9.48703825e-01 3.43724526e-02 -2.67820638e-02 -1.04167366e+00
-7.68296719e-01 -3.82306308e-01 -8.73239756e-01 2.83531547e-01
5.89965463e-01 -7.37798870e-01 -5.10588348e-01 4.98054683e-01
-1.10885572e+00 -3.32467049e-01 -1.86819032e-01 3.61224174e-01
-8.73232111e-02 7.94947922e-01 -1.22564703e-01 -5.89237571e-01
-4.48391289e-01 -1.03181720e+00 9.48164940e-01 1.23149911e-02
-6.16042733e-01 -1.05962789e+00 -2.41234973e-02 2.98277020e-01
5.77149153e-01 1.90717295e-01 1.29080892e+00 -8.49664330e-01
-1.17433801e-01 -2.46652678e-01 -1.88074812e-01 6.53292239e-01
9.99186784e-02 -8.59464929e-02 -8.94874156e-01 -2.23212138e-01
-2.84690052e-01 8.69252309e-02 8.02672625e-01 4.70476784e-02
1.11068821e+00 2.54590996e-02 3.31233954e-03 6.21539414e-01
1.33158219e+00 -1.80925384e-01 6.09405637e-01 5.19054711e-01
8.10779452e-01 5.97134233e-01 7.40777075e-01 5.08415282e-01
3.31374764e-01 1.13438725e+00 2.42380351e-01 -4.56506722e-02
-3.88212860e-01 -1.64474204e-01 -8.42458475e-03 9.50877249e-01
9.15154889e-02 -1.52559206e-01 -1.06724083e+00 6.12649679e-01
-1.58699870e+00 -5.27826190e-01 -3.13657016e-01 2.60879803e+00
1.02584386e+00 3.29234809e-01 1.67191952e-01 5.49222708e-01
5.40167332e-01 -3.02621885e-03 -4.27438647e-01 -6.30768239e-01
-3.02850246e-01 5.86914182e-01 4.83001143e-01 5.60671806e-01
-1.10611141e+00 1.15707695e+00 5.08364820e+00 1.24823713e+00
-1.27630019e+00 7.53852203e-02 3.64168495e-01 -1.11170607e-02
-3.67002875e-01 -2.12248877e-01 -9.82140601e-01 6.35088742e-01
1.04321456e+00 -2.40748003e-01 6.30920887e-01 6.63646519e-01
9.21114683e-02 -1.13174506e-01 -1.14829099e+00 1.04692137e+00
1.20623969e-01 -8.87318134e-01 1.86591297e-01 3.73087567e-03
5.25265872e-01 -2.11087480e-01 6.70083389e-02 6.31006837e-01
-1.50791436e-01 -9.94216025e-01 7.88199306e-01 3.49854946e-01
9.57700670e-01 -9.74056482e-01 7.52318203e-01 4.47940081e-01
-9.65503931e-01 -2.88198981e-02 -2.66787887e-01 -2.03406904e-02
1.45733554e-03 9.02497828e-01 -7.16386020e-01 7.32184649e-01
4.65398043e-01 4.62959856e-01 -8.27532530e-01 1.05216348e+00
-3.57787281e-01 5.24988949e-01 -3.71138006e-01 5.72247058e-02
-7.18346387e-02 -3.13345551e-01 4.42036778e-01 1.57068586e+00
3.40060532e-01 -5.30464709e-01 -6.57046884e-02 5.90417743e-01
-1.73000932e-01 6.35598958e-01 -4.63547319e-01 -8.77008401e-03
6.09937012e-01 1.30634594e+00 -3.36026996e-01 -3.00826877e-01
-3.67411971e-01 1.00797379e+00 5.52810967e-01 7.66817108e-02
-8.70719612e-01 -5.82285941e-01 7.73459077e-01 -3.43823731e-02
4.56282794e-01 -3.20528656e-01 -6.41587198e-01 -1.25845134e+00
2.13629663e-01 -9.89349246e-01 1.96427822e-01 5.97767867e-02
-1.27603471e+00 5.88447869e-01 -2.78542191e-02 -1.18318248e+00
-1.59832746e-01 -7.51699209e-01 -3.93852800e-01 1.15246606e+00
-1.92634416e+00 -1.14667261e+00 -3.45639437e-01 3.92188430e-01
5.97887516e-01 5.12079038e-02 1.02727532e+00 7.52371669e-01
-6.99848115e-01 1.07336450e+00 2.52719790e-01 -1.60343483e-01
1.02211201e+00 -1.21864307e+00 7.39050925e-01 9.94361401e-01
4.11836840e-02 6.68112278e-01 6.90760672e-01 -4.92928177e-01
-1.16299236e+00 -1.16992044e+00 1.15704668e+00 -4.72974330e-01
7.47042775e-01 -6.17231905e-01 -1.14342153e+00 1.90442130e-01
-5.17581441e-02 -3.57288033e-01 6.59664750e-01 5.40773988e-01
-5.08964181e-01 -1.41050875e-01 -9.58585322e-01 9.23398018e-01
1.10607409e+00 -4.10247713e-01 -4.89898741e-01 1.92029059e-01
7.57193446e-01 -2.41539970e-01 -1.03064263e+00 3.30332279e-01
4.81526077e-01 -8.11649978e-01 9.02946115e-01 -3.66976857e-01
3.33453357e-01 -1.10390410e-01 -2.17476152e-02 -1.55192566e+00
-2.34371960e-01 -4.39505488e-01 2.67919432e-02 1.63862526e+00
5.75094581e-01 -6.78450406e-01 4.84506518e-01 4.66829509e-01
-1.09175697e-01 -6.14493132e-01 -1.10275543e+00 -1.03803825e+00
5.50156057e-01 -6.54255927e-01 5.96897781e-01 8.78651798e-01
-1.10559657e-01 4.72338706e-01 -4.17428732e-01 -2.39952728e-01
4.81414109e-01 -1.25775322e-01 9.05332088e-01 -1.54364514e+00
-2.48034611e-01 -7.37233162e-01 -1.79191902e-01 -7.98244596e-01
3.88013422e-01 -1.02775919e+00 -1.68483719e-01 -1.24169278e+00
-7.88081344e-03 -5.55828035e-01 -1.37045234e-01 3.11301291e-01
-4.67493027e-01 2.43606970e-01 6.18378878e-01 2.13738251e-02
-2.88664699e-01 6.79817200e-01 9.00156319e-01 -4.74902838e-02
-1.54561862e-01 -1.31448776e-01 -7.45800614e-01 6.95999622e-01
9.58731115e-01 -3.06737393e-01 -2.01224789e-01 -6.38690233e-01
1.54138088e-01 -5.36533833e-01 2.12979540e-02 -1.15642023e+00
-1.07736342e-01 1.84803933e-01 8.01515356e-02 -3.83199126e-01
3.01229507e-01 -7.82761157e-01 -1.81023590e-02 4.00464982e-01
-1.80904523e-01 -1.71350017e-02 5.11327565e-01 3.59148741e-01
-7.49568492e-02 -3.22778732e-01 8.01809907e-01 1.91715345e-01
-5.30107260e-01 -1.05584338e-01 -1.96847901e-01 3.47410113e-01
7.13980794e-01 -4.45940830e-02 -2.67392963e-01 -2.17049960e-02
-4.07190949e-01 7.44387433e-02 6.20902717e-01 6.40102684e-01
-3.43227270e-03 -1.10254776e+00 -7.32690573e-01 1.55932635e-01
3.62342596e-01 -1.99313704e-02 2.59320915e-01 9.65359926e-01
-2.54925281e-01 5.84703624e-01 -8.19339156e-02 -4.74600196e-01
-1.15875220e+00 6.12731516e-01 -2.22881138e-01 -8.36724877e-01
-5.75299799e-01 4.88060623e-01 4.60714959e-02 -5.50318897e-01
2.61884362e-01 -4.50556606e-01 -2.89995402e-01 3.45465750e-01
4.67228144e-01 4.91654307e-01 5.36357999e-01 -5.50854445e-01
-3.06746036e-01 6.56456411e-01 -2.03749895e-01 8.92264321e-02
1.41655588e+00 -1.41727418e-01 3.85808088e-02 2.28862479e-01
1.15187776e+00 4.53343272e-01 -9.14207339e-01 -4.84072000e-01
4.99796003e-01 -3.58993083e-01 -2.00105190e-01 -9.90924537e-01
-6.43098474e-01 9.92442131e-01 5.61961472e-01 -3.86894979e-02
1.15338552e+00 -2.56927043e-01 8.34700942e-01 2.40943730e-01
1.15091741e-01 -1.27026510e+00 -2.68412381e-01 9.10524309e-01
8.51342857e-01 -1.38387811e+00 -6.78944308e-03 -4.30733025e-01
-6.71946466e-01 1.05724871e+00 2.33314633e-01 -3.28868516e-02
2.90831596e-01 -2.55093351e-03 -1.39484689e-01 2.81632572e-01
-4.62777972e-01 -3.98633838e-01 3.76370698e-01 7.01564848e-01
5.99732697e-01 1.18510261e-01 -8.56393397e-01 7.02845633e-01
-4.08956200e-01 -2.46281400e-01 3.19805920e-01 3.04076135e-01
-8.20816606e-02 -1.63100660e+00 -2.21470818e-01 2.07685262e-01
-4.78629142e-01 -4.12746400e-01 -1.59451067e-01 1.03589141e+00
1.81191906e-01 8.53542984e-01 -2.23538309e-01 -2.80203551e-01
6.86046958e-01 3.05325508e-01 2.70239115e-01 -6.90633178e-01
-5.92044711e-01 -1.64197877e-01 1.91208765e-01 -3.21324676e-01
-4.73831631e-02 -6.94837570e-01 -8.36621642e-01 -3.39846551e-01
-1.88483298e-01 -8.88277814e-02 9.22326267e-01 9.69954133e-01
2.95961231e-01 6.15314901e-01 2.84556568e-01 -9.11316574e-01
-6.41042531e-01 -1.18159187e+00 -2.71354914e-01 8.96251321e-01
4.12012152e-02 -7.28196204e-01 -2.06388772e-01 -1.55082747e-01]
|
[10.356927871704102, 9.658768653869629]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.