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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
18f1e3f3-fe23-4c04-95a7-e18f3077393c
|
a-high-precision-self-supervised-monocular
|
2203.04812
| null |
https://arxiv.org/abs/2203.04812v1
|
https://arxiv.org/pdf/2203.04812v1.pdf
|
A high-precision self-supervised monocular visual odometry in foggy weather based on robust cycled generative adversarial networks and multi-task learning aided depth estimation
|
This paper proposes a high-precision self-supervised monocular VO, which is specifically designed for navigation in foggy weather. A cycled generative adversarial network is designed to obtain high-quality self-supervised loss via forcing the forward and backward half-cycle to output consistent estimation. Moreover, gradient-based loss and perceptual loss are introduced to eliminate the interference of complex photometric change on self-supervised loss in foggy weather. To solve the ill-posed problem of depth estimation, a self-supervised multi-task learning aided depth estimation module is designed based on the strong correlation between the depth estimation and transmission map calculation of hazy images in foggy weather. The experimental results on the synthetic foggy KITTI dataset show that the proposed self-supervised monocular VO performs better in depth and pose estimation than other state-of-the-art monocular VO in the literature, indicating the designed method is more suitable for foggy weather.
|
['Guowen An', 'Fengchao Li', 'Jiangang Yu', 'Xiuyuan Li']
|
2022-03-09
| null | null | null | null |
['monocular-visual-odometry']
|
['robots']
|
[-2.29594931e-01 -2.42867344e-03 4.82736409e-01 -5.00453115e-01
-3.48656774e-01 -2.69723386e-01 2.90089220e-01 -8.14824939e-01
-4.74217683e-01 1.14083636e+00 -4.71839607e-02 1.89501435e-01
1.37705237e-01 -9.86964703e-01 -6.61719382e-01 -1.02624857e+00
-5.05640507e-02 1.92237273e-01 1.51793748e-01 -4.47493315e-01
9.77521390e-03 1.77792549e-01 -1.52825844e+00 -3.30351859e-01
1.54835904e+00 1.00594473e+00 4.52723652e-01 7.14311421e-01
2.02115610e-01 8.61537516e-01 -7.57235706e-01 -3.63185614e-01
7.72159338e-01 -6.31767333e-01 -1.60751641e-01 -1.67421903e-02
8.26954365e-01 -7.24601686e-01 -5.02920270e-01 1.04557991e+00
7.94209599e-01 1.90984353e-01 7.72759378e-01 -1.01271129e+00
-1.29005283e-01 -1.53983757e-01 -3.69664431e-01 1.54164106e-01
7.75967017e-02 2.53189325e-01 5.92011392e-01 -8.28526378e-01
3.10894519e-01 9.46756721e-01 6.60778642e-01 4.92841840e-01
-4.63914871e-01 -7.70941138e-01 4.25773673e-02 1.53939709e-01
-1.30151069e+00 -1.83324471e-01 7.99419761e-01 -4.78724688e-01
5.49570858e-01 1.14937976e-01 9.02417541e-01 4.65946704e-01
6.21682286e-01 4.63102341e-01 1.26865697e+00 -7.45349526e-02
1.00657322e-01 1.71084896e-01 -6.34758055e-01 1.01444459e+00
3.54702562e-01 6.94330156e-01 -5.17772079e-01 2.29604825e-01
7.59259462e-01 -1.79185435e-01 -6.09640777e-01 -3.47722709e-01
-7.13854969e-01 8.76927376e-01 9.84742880e-01 -1.92785054e-01
-2.26300105e-01 2.09416658e-01 -1.05714865e-01 4.37129825e-01
8.66140246e-01 4.18684393e-01 -1.94578037e-01 6.80861995e-02
-1.17157578e+00 3.93270433e-01 6.23266637e-01 9.09835100e-01
9.95057285e-01 9.17265892e-01 2.08197668e-01 2.36058041e-01
6.00501597e-01 1.16499829e+00 3.74796480e-01 -8.22762907e-01
4.02654052e-01 1.37536958e-01 3.24163437e-01 -8.84318471e-01
-2.61359662e-01 -7.41672397e-01 -7.78957129e-01 8.09160948e-01
1.30261764e-01 -4.82160449e-01 -1.03322160e+00 1.46567464e+00
5.32726765e-01 9.97169465e-02 4.59524423e-01 1.41522253e+00
1.07905316e+00 7.40276694e-01 -4.28325206e-01 -1.60350725e-01
7.00061262e-01 -1.01454532e+00 -9.09639060e-01 -6.32915497e-01
5.96858971e-02 -6.60683751e-01 7.26836026e-01 2.96392977e-01
-8.76020491e-01 -6.00437999e-01 -1.36330175e+00 1.61499262e-01
-1.71394885e-01 -5.85212074e-02 7.04558611e-01 7.14009047e-01
-8.97766113e-01 2.60110557e-01 -5.07572412e-01 1.58868402e-01
1.47724107e-01 -1.54438466e-02 -1.01524211e-01 -9.92651209e-02
-1.62261951e+00 1.11044407e+00 3.33520263e-01 3.75853121e-01
-1.30828118e+00 -7.23576725e-01 -1.08419979e+00 -3.57065916e-01
-9.17608216e-02 -1.01556730e+00 8.00452769e-01 -1.08204806e+00
-1.68739522e+00 6.21806681e-01 1.34195834e-02 -5.86897671e-01
9.14186180e-01 -5.25362968e-01 -2.35971525e-01 4.38329399e-01
1.03007015e-02 7.22923577e-01 1.37567472e+00 -1.48930335e+00
-5.42626441e-01 -2.71872997e-01 9.04261228e-03 9.00255859e-01
9.52446312e-02 -7.60500550e-01 -6.08463101e-02 -5.98049164e-01
2.92631984e-01 -7.41116762e-01 -1.46637186e-01 1.84086502e-01
-5.14577888e-02 6.76995695e-01 9.60911214e-01 -7.55678654e-01
7.28314042e-01 -1.82464683e+00 1.35900646e-01 -1.29396304e-01
1.16346449e-01 2.08404317e-01 3.79602164e-01 1.52753651e-01
4.09664690e-01 -5.51273346e-01 -5.03808379e-01 -6.05067074e-01
-4.29029703e-01 2.42750794e-01 -3.36496800e-01 9.67886269e-01
3.81343849e-02 7.73370326e-01 -9.71588910e-01 -4.37481374e-01
7.17325270e-01 4.57497895e-01 -4.07002836e-01 6.63786709e-01
-2.10602462e-01 8.44941378e-01 -2.29951143e-01 6.51585519e-01
1.23959720e+00 4.27827179e-01 -5.58884561e-01 6.70502952e-04
-2.22344756e-01 -3.23993936e-02 -8.77849638e-01 1.62597179e+00
-7.64899790e-01 7.69865513e-01 2.88346738e-01 -3.28152299e-01
1.16994536e+00 -5.49178980e-02 -4.37390245e-03 -7.89534509e-01
1.68072760e-01 3.60726416e-01 -4.45493281e-01 -6.01111829e-01
6.74500763e-01 -6.22974992e-01 1.86023340e-01 -1.22408144e-01
-5.41008301e-02 -1.23850715e+00 -5.40145695e-01 7.29562938e-02
4.99434143e-01 4.14291650e-01 4.82217921e-03 -1.61753953e-01
6.73274994e-01 5.60091436e-02 8.44432592e-01 4.17313576e-01
-3.60818446e-01 1.00459206e+00 -3.74319941e-01 -3.03844035e-01
-8.76927316e-01 -1.03667808e+00 3.78402807e-02 2.17499033e-01
7.13853359e-01 3.69130254e-01 -4.25852180e-01 -4.68630522e-01
1.70838479e-02 3.81471366e-01 -5.19377172e-01 -1.77552104e-01
-3.25608104e-01 -8.25673819e-01 5.27724087e-01 1.22195259e-01
1.47895312e+00 -7.46349216e-01 -6.91858828e-01 1.10038318e-01
-4.06182170e-01 -1.11855948e+00 -2.65952349e-01 -5.09915985e-02
-9.06818151e-01 -9.68914390e-01 -1.03400469e+00 -7.56349802e-01
5.59627116e-01 4.69266593e-01 1.00117171e+00 -2.85215765e-01
3.78701324e-03 1.71789557e-01 -4.93853092e-01 -6.82449818e-01
-1.02321409e-01 -3.38537455e-01 -9.58065391e-02 2.07949251e-01
-3.70295137e-01 -6.18853629e-01 -9.46776152e-01 4.28824872e-01
-7.88913488e-01 -2.65910905e-02 4.66385573e-01 1.02345729e+00
3.29175979e-01 9.94961709e-02 2.69547701e-01 -6.14428639e-01
-5.28535880e-02 -3.63200814e-01 -1.05046344e+00 -2.61126131e-01
-7.52403557e-01 -2.02096388e-01 6.43125772e-01 -4.03968170e-02
-1.52529407e+00 2.61150878e-02 1.18727458e-03 -5.73615670e-01
3.09693307e-01 3.07186335e-01 -1.67753816e-01 -8.96647513e-01
7.09085047e-01 6.58036053e-01 8.58716518e-02 1.43413678e-01
2.74030268e-01 3.96242082e-01 7.34448075e-01 3.91868167e-02
1.49203956e+00 1.01683080e+00 3.44865203e-01 -9.92563725e-01
-7.61717021e-01 -5.78898430e-01 -8.14970359e-02 -7.30960548e-01
1.07058966e+00 -1.62918329e+00 -3.59665513e-01 1.16294611e+00
-9.50170219e-01 -4.77060676e-01 -6.20293058e-02 7.48391032e-01
-5.24585366e-01 4.77476865e-01 -3.65398496e-01 -1.09109247e+00
-7.47361422e-01 -8.33518565e-01 8.95287156e-01 5.17917991e-01
5.03616631e-01 -1.26019108e+00 1.78003326e-01 5.71133971e-01
5.30614078e-01 5.26020885e-01 -2.32508802e-03 5.81965148e-01
-1.08344805e+00 -2.97884438e-02 6.05835058e-02 5.37580550e-01
7.95054436e-02 -3.50379258e-01 -1.27545595e+00 -5.35356998e-01
3.75332564e-01 -5.74053943e-01 1.16345239e+00 4.26564515e-01
4.09884989e-01 -2.20212787e-01 1.30976558e-01 1.55990231e+00
1.88754678e+00 1.99150562e-01 8.98658633e-01 5.37836850e-01
8.22214484e-01 3.51802796e-01 1.21533966e+00 4.06955659e-01
5.31465650e-01 3.72425258e-01 1.09355676e+00 -3.78450871e-01
-6.84064701e-02 -2.80105352e-01 3.04684669e-01 4.00418997e-01
2.33521070e-02 -5.01771152e-01 -2.91706353e-01 5.69332600e-01
-1.50360298e+00 -7.67793894e-01 -1.03549786e-01 2.12968731e+00
5.70111811e-01 1.06367521e-01 -4.87936527e-01 -3.43766473e-02
3.00912261e-01 6.45635784e-01 -5.32662213e-01 -2.30614424e-01
-5.98462403e-01 1.87826544e-01 1.12351346e+00 9.31784809e-01
-1.16423762e+00 1.06967199e+00 5.81257153e+00 5.96855223e-01
-1.27209580e+00 1.52251184e-01 3.30117404e-01 8.72628391e-02
-6.28392756e-01 -1.44944817e-01 -7.84192920e-01 5.00559032e-01
4.46588993e-01 1.56969428e-01 8.46306086e-02 6.40636683e-01
4.43420172e-01 -6.11030638e-01 -9.40529108e-02 1.04718947e+00
1.91146538e-01 -1.07807815e+00 -1.04982719e-01 -1.54920578e-01
1.21960008e+00 9.84759703e-02 7.97702894e-02 -1.11507207e-01
3.23475510e-01 -7.86406517e-01 7.97775686e-01 6.44630015e-01
9.20727968e-01 -8.23049963e-01 9.95668888e-01 2.60123074e-01
-1.13097572e+00 1.42903313e-01 -5.62246025e-01 -3.26492339e-01
3.92462313e-01 9.93084073e-01 -6.77355647e-01 1.03286266e+00
7.56549835e-01 7.60246158e-01 -1.31607294e-01 1.27345943e+00
-8.44611943e-01 4.15583760e-01 -2.56928921e-01 6.98060840e-02
4.75370824e-01 -4.67057794e-01 8.37207258e-01 7.52223313e-01
3.45964402e-01 2.65852571e-01 -1.54831469e-01 7.19840109e-01
3.22744846e-01 -2.05144554e-01 -7.74128616e-01 6.42325759e-01
2.29065508e-01 1.05272603e+00 -3.70943956e-02 -9.90667269e-02
-1.11647539e-01 1.18870234e+00 -1.73898041e-01 2.40627185e-01
-9.02597785e-01 -5.11493206e-01 9.97886598e-01 -8.49228501e-02
3.41849267e-01 -4.32270318e-01 -4.30276245e-01 -1.09217048e+00
7.06158727e-02 -2.95218498e-01 4.28341515e-02 -1.14434874e+00
-8.53164613e-01 7.34157205e-01 -1.89091161e-01 -1.86943352e+00
-2.10266456e-01 -1.19159162e-01 -8.72758210e-01 1.18938899e+00
-2.23236370e+00 -1.19542110e+00 -1.11020923e+00 5.35471976e-01
2.35090941e-01 -2.08310828e-01 4.90796566e-01 2.24059045e-01
-6.44789562e-02 6.18373096e-01 1.70497477e-01 -2.97614813e-01
7.78805852e-01 -1.16279471e+00 4.29929674e-01 1.30607283e+00
-4.18725222e-01 -1.63854614e-01 9.79166269e-01 -7.54784524e-01
-1.15552545e+00 -1.41592968e+00 5.73003590e-01 -5.96834905e-02
7.80341178e-02 -1.18178301e-01 -6.36442482e-01 2.70615906e-01
2.73302030e-02 1.22033477e-01 -1.14698723e-01 -9.57228601e-01
-4.60186787e-02 -5.17297447e-01 -1.60701334e+00 2.22836494e-01
7.77527630e-01 -3.62244397e-01 -3.87648493e-01 3.39155465e-01
1.00775158e+00 -1.05663133e+00 -4.69216228e-01 6.41829073e-01
3.45906526e-01 -1.33342481e+00 8.80909204e-01 2.63991386e-01
5.21044314e-01 -6.29987299e-01 3.71181816e-02 -1.78289747e+00
7.40957186e-02 -8.36237907e-01 -5.12092151e-02 8.88507366e-01
2.50159860e-01 -1.05283868e+00 1.06169033e+00 -1.73733264e-01
-5.94227493e-01 -4.11673188e-01 -9.16097045e-01 -8.38417649e-01
-9.55906883e-02 1.21029783e-02 2.34191999e-01 6.83174968e-01
-6.80054605e-01 -1.24924153e-01 -8.80112886e-01 7.87984967e-01
1.10474682e+00 1.26766920e-01 9.62699533e-01 -7.94299364e-01
-2.04868212e-01 3.71180594e-01 -6.72260702e-01 -1.22509825e+00
-6.35358170e-02 -2.67847002e-01 5.12240708e-01 -1.46095288e+00
-5.53160727e-01 -3.99966866e-01 3.46175820e-01 -1.82860017e-01
-3.84442061e-01 5.84550440e-01 -4.32967544e-02 1.97065815e-01
-7.90911354e-03 1.09446275e+00 1.83936751e+00 -1.74085602e-01
-1.86960116e-01 1.82915926e-01 -8.71310830e-02 4.53165501e-01
7.42560983e-01 -3.45938474e-01 -6.99425399e-01 -6.64135277e-01
1.11836210e-01 2.36495093e-01 1.96803734e-01 -1.56964302e+00
2.51959682e-01 -1.83388636e-01 5.13676763e-01 -7.06374407e-01
6.71392977e-01 -6.64099514e-01 -8.07355940e-02 6.46271348e-01
3.98203671e-01 -1.92294896e-01 7.45095238e-02 5.44609129e-01
-7.06653774e-01 9.48311121e-04 1.26851428e+00 -2.92045981e-01
-1.10915172e+00 5.60780883e-01 -1.29087731e-01 3.02602112e-01
9.56034005e-01 -4.00163502e-01 -2.50111997e-01 -1.01464438e+00
-2.75851309e-01 3.81884068e-01 5.36975861e-01 2.87871202e-03
1.14904130e+00 -8.89981270e-01 -7.09452391e-01 5.15717924e-01
2.22027659e-01 4.75607902e-01 3.25710475e-01 4.32004452e-01
-1.22787058e+00 5.36171496e-02 -3.03875983e-01 -4.47129369e-01
-9.79877591e-01 -1.04553424e-01 8.57001841e-01 6.35763109e-02
-5.18323779e-01 1.31763482e+00 2.37468213e-01 -4.75450844e-01
8.16331152e-03 -8.00461248e-02 3.42376940e-02 -2.06504926e-01
3.48572373e-01 3.80043119e-01 9.38081294e-02 -4.65708345e-01
-1.60365105e-01 7.40530789e-01 4.73880738e-01 -3.06811273e-01
1.15649033e+00 -5.98247826e-01 8.67674798e-02 1.92392394e-01
9.57979560e-01 -1.10375866e-01 -1.93771958e+00 2.25902162e-03
-1.10038960e+00 -7.58218884e-01 4.38711286e-01 -7.02098131e-01
-1.44010472e+00 7.12802887e-01 7.15173364e-01 -6.14032924e-01
1.27262330e+00 -6.25117481e-01 1.03324449e+00 1.70987681e-01
7.00921357e-01 -6.88873291e-01 1.67913541e-01 8.06635261e-01
7.03711390e-01 -1.41403913e+00 2.17199281e-01 -5.26498258e-01
-6.93980694e-01 9.12052333e-01 9.27203298e-01 -4.18935537e-01
5.54102659e-01 -7.33665600e-02 5.41943073e-01 -1.42838866e-01
-1.34833127e-01 -2.91410953e-01 1.71964750e-01 6.93498611e-01
-2.87660658e-01 -6.82040229e-02 -3.07257682e-01 -9.75302383e-02
-5.60048819e-01 -2.32638747e-01 7.93732464e-01 8.62488747e-01
-5.88603556e-01 -3.80919933e-01 -5.85611939e-01 9.57273915e-02
-7.70526901e-02 -2.09099695e-01 -7.25496337e-02 7.44254470e-01
3.72698992e-01 9.97370839e-01 6.17644638e-02 -4.55655754e-01
6.64231181e-02 -5.19094884e-01 6.02183282e-01 -3.12600195e-01
-4.67138052e-01 -1.82536945e-01 6.30877167e-03 -5.27515769e-01
-5.35520434e-01 -2.26156622e-01 -1.07530975e+00 -2.64781922e-01
-2.76801497e-01 2.38586664e-01 7.53830492e-01 9.08617854e-01
-1.16149880e-01 3.36366981e-01 1.17081594e+00 -1.14999807e+00
-1.34223476e-01 -9.86393929e-01 -7.64519393e-01 -2.48245727e-02
8.82778645e-01 -7.28857875e-01 -1.14777708e+00 -3.24199140e-01]
|
[10.884488105773926, -3.2460615634918213]
|
f6d279ae-e0f6-423d-9bc6-0c02b28f36ea
|
uncertainty-estimation-with-normalized-logits
|
2302.07608
| null |
https://arxiv.org/abs/2302.07608v1
|
https://arxiv.org/pdf/2302.07608v1.pdf
|
Uncertainty-Estimation with Normalized Logits for Out-of-Distribution Detection
|
Out-of-distribution (OOD) detection is critical for preventing deep learning models from making incorrect predictions to ensure the safety of artificial intelligence systems. Especially in safety-critical applications such as medical diagnosis and autonomous driving, the cost of incorrect decisions is usually unbearable. However, neural networks often suffer from the overconfidence issue, making high confidence for OOD data which are never seen during training process and may be irrelevant to training data, namely in-distribution (ID) data. Determining the reliability of the prediction is still a difficult and challenging task. In this work, we propose Uncertainty-Estimation with Normalized Logits (UE-NL), a robust learning method for OOD detection, which has three main benefits. (1) Neural networks with UE-NL treat every ID sample equally by predicting the uncertainty score of input data and the uncertainty is added into softmax function to adjust the learning strength of easy and hard samples during training phase, making the model learn robustly and accurately. (2) UE-NL enforces a constant vector norm on the logits to decouple the effect of the increasing output norm from optimization process, which causes the overconfidence issue to some extent. (3) UE-NL provides a new metric, the magnitude of uncertainty score, to detect OOD data. Experiments demonstrate that UE-NL achieves top performance on common OOD benchmarks and is more robust to noisy ID data that may be misjudged as OOD data by other methods.
|
['Yu Qiao', 'Mouxiao Huang']
|
2023-02-15
| null | null | null | null |
['medical-diagnosis']
|
['medical']
|
[-4.92020734e-02 4.67492968e-01 -2.64728129e-01 -7.17184663e-01
-5.12460411e-01 -1.44157067e-01 4.10301596e-01 2.14248911e-01
-4.17727172e-01 8.83176208e-01 -1.21980153e-01 -3.93915474e-01
-1.07924536e-01 -6.78865254e-01 -8.24180782e-01 -7.02011228e-01
9.13218185e-02 3.51843745e-01 3.32944483e-01 2.49097899e-01
-1.29713574e-02 2.20395282e-01 -1.40961552e+00 5.17240949e-02
1.23854303e+00 1.66203856e+00 -3.90297323e-02 1.64599597e-01
-1.17437817e-01 8.29697907e-01 -9.96591091e-01 -1.88620761e-01
2.58849859e-01 -7.52620101e-02 -2.20782563e-01 -3.09174716e-01
2.13206172e-01 -5.73347509e-01 -2.66778737e-01 1.29607534e+00
5.36096454e-01 8.39210674e-02 9.12502110e-01 -1.64574897e+00
-5.17250121e-01 4.13501620e-01 -5.25006115e-01 1.35919496e-01
-2.43878812e-01 8.56689885e-02 5.40052831e-01 -6.78261042e-01
1.57528237e-01 1.23632789e+00 4.64612156e-01 5.80168664e-01
-8.80834103e-01 -1.02900290e+00 2.28481695e-01 1.83767766e-01
-1.28880334e+00 -2.50944644e-01 6.03012502e-01 -4.18557018e-01
7.51458526e-01 -2.40133400e-03 1.52921930e-01 1.23024917e+00
7.61610329e-01 8.11038315e-01 8.75958204e-01 -8.25689062e-02
4.52238292e-01 4.34695542e-01 2.34475330e-01 3.84102196e-01
5.18665731e-01 5.57994366e-01 -3.05037856e-01 -1.33979758e-02
5.75161576e-01 4.32223380e-02 -3.20951223e-01 -2.37156317e-01
-8.44102979e-01 6.89246237e-01 5.76750159e-01 -1.03701003e-01
-2.14011624e-01 -6.91702962e-02 4.13299471e-01 3.94194782e-01
2.50846922e-01 3.94260913e-01 -5.00772893e-01 -2.42782339e-01
-4.85698491e-01 2.35920861e-01 6.26729488e-01 9.06662405e-01
3.42411071e-01 2.05705062e-01 -4.88795429e-01 8.90175045e-01
4.24232692e-01 4.56729740e-01 7.42620766e-01 -7.08043396e-01
5.57583332e-01 5.64140618e-01 1.00119889e-01 -1.02292955e+00
-5.15106201e-01 -6.01200402e-01 -1.09923077e+00 5.71584821e-01
3.23315829e-01 -3.53138626e-01 -1.22527695e+00 1.81856942e+00
3.58877331e-01 1.18990257e-01 1.39529049e-01 9.05412793e-01
8.77603233e-01 5.60892284e-01 5.72262071e-02 -1.82502210e-01
1.13143599e+00 -6.62802875e-01 -8.91919672e-01 -4.62921351e-01
3.89793098e-01 -3.81761402e-01 8.52114856e-01 6.15082443e-01
-5.39440632e-01 -5.18734455e-01 -1.44910049e+00 1.87505424e-01
-2.74706274e-01 -1.22868039e-01 4.01711673e-01 4.30874228e-01
-3.49434823e-01 6.97713077e-01 -6.73058510e-01 4.04789448e-01
5.63198566e-01 3.48998159e-01 -2.78941929e-01 6.94663972e-02
-1.57724178e+00 9.11581755e-01 7.61701167e-01 2.12116972e-01
-7.55952358e-01 -6.65891469e-01 -9.41837311e-01 3.73636298e-02
5.20347118e-01 -2.76404828e-01 1.17065895e+00 -8.91218603e-01
-1.31494308e+00 2.96546757e-01 2.07866624e-01 -5.23879588e-01
8.90317619e-01 -2.40047634e-01 -6.93334460e-01 -5.87363183e-01
3.76540609e-02 6.82646930e-01 1.04728222e+00 -1.16280591e+00
-8.01693738e-01 -3.87311310e-01 -3.46419245e-01 3.80351543e-02
-1.96175486e-01 -5.15538692e-01 -3.41650456e-01 -6.26402199e-01
1.27497956e-01 -6.95839465e-01 8.09012577e-02 -4.49388660e-02
-6.61727726e-01 -5.20766079e-01 6.58935070e-01 -3.80514711e-01
1.49073088e+00 -2.33316612e+00 -5.20490050e-01 4.07572240e-01
4.31270450e-01 2.75690585e-01 1.49542317e-01 -3.81212592e-01
-7.02528059e-02 7.42667094e-02 -1.76402122e-01 7.93458968e-02
1.76509514e-01 4.04902160e-01 -2.01719761e-01 1.98929712e-01
5.71404338e-01 5.11088371e-01 -8.66606891e-01 -4.04722065e-01
1.41211286e-01 2.96271414e-01 -2.59030312e-01 4.12740916e-01
-1.24684833e-01 1.47318184e-01 -5.73136508e-01 6.27629101e-01
7.96185851e-01 -5.77727631e-02 -5.52329659e-01 -1.39635757e-01
2.01565981e-01 2.57277399e-01 -1.51326966e+00 8.24388981e-01
-1.46738604e-01 5.68525314e-01 -2.85207510e-01 -7.51362443e-01
1.00335312e+00 2.60671854e-01 1.08960755e-01 -8.62901270e-01
4.44306314e-01 3.07513624e-01 2.47337490e-01 -4.63890076e-01
2.44110510e-01 -9.45733264e-02 -1.66707188e-01 -1.53889172e-02
-1.57788813e-01 -3.86811397e-03 -1.31572053e-01 -2.00210616e-01
8.74061465e-01 -1.41171232e-01 2.36745134e-01 -2.03542367e-01
1.31465629e-01 -3.67424399e-01 1.21873736e+00 8.46676171e-01
-5.41657925e-01 4.73457038e-01 8.04359853e-01 -2.39203557e-01
-7.63129056e-01 -1.25236011e+00 -7.36381829e-01 4.91681337e-01
2.09973678e-01 1.37272105e-01 -5.83907604e-01 -9.52749372e-01
4.11359280e-01 1.06564450e+00 -5.77267885e-01 -7.62581646e-01
-9.65109617e-02 -8.61407220e-01 5.29894769e-01 8.33079278e-01
3.39431882e-01 -8.31732929e-01 -4.21486855e-01 2.68518031e-01
9.62719917e-02 -1.00462401e+00 -4.56099451e-01 7.71452248e-01
-6.31159842e-01 -9.57248688e-01 -5.63556194e-01 -5.23387432e-01
7.17476368e-01 -2.21505359e-01 9.20913696e-01 -2.44661778e-01
-1.19772799e-01 -3.70102763e-01 5.03861979e-02 -1.13695824e+00
-4.76704240e-01 -3.29038531e-01 3.71143669e-01 -1.90910921e-01
7.18705714e-01 -2.15111777e-01 -4.50747430e-01 6.30175829e-01
-7.96202064e-01 -3.39051187e-01 5.88921368e-01 1.19831729e+00
7.87332356e-01 5.21146536e-01 1.05684483e+00 -7.21515477e-01
9.75507259e-01 -6.28193140e-01 -7.10843503e-01 -4.40875962e-02
-1.04981649e+00 3.58160973e-01 6.15709126e-01 -6.83976352e-01
-9.07328427e-01 -3.33943754e-01 -1.40446499e-01 -7.03698397e-01
-9.65873972e-02 3.59643728e-01 -3.81696939e-01 3.31329435e-01
6.16412938e-01 -2.47569516e-01 5.58362640e-02 -2.64876813e-01
-2.97688097e-01 1.06041253e+00 5.45590878e-01 -3.59368712e-01
4.06668901e-01 -9.94768962e-02 -1.07653968e-01 -5.36545277e-01
-9.04109538e-01 -1.60425127e-01 1.04344618e-02 -1.05566919e-01
6.79134190e-01 -9.34695423e-01 -6.31979406e-01 6.22980237e-01
-8.26654911e-01 -1.85085669e-01 -9.63891670e-02 7.45065868e-01
-2.97988839e-02 1.77045390e-01 -3.22848558e-01 -9.36096191e-01
-2.92374223e-01 -1.20271277e+00 7.93439269e-01 5.66812098e-01
-2.88146228e-01 -6.72708988e-01 -4.11088049e-01 3.85397226e-02
1.67593181e-01 4.01105046e-01 9.91178989e-01 -9.37151670e-01
-9.50124189e-02 -5.20363986e-01 -2.67318994e-01 8.61930072e-01
1.09898843e-01 -4.84881829e-03 -1.11633742e+00 2.04163250e-02
1.64935067e-01 -4.91743594e-01 8.52363169e-01 4.71415430e-01
1.33116591e+00 -1.89768746e-01 -3.40021342e-01 3.60580057e-01
9.38797891e-01 4.53725010e-01 4.37956452e-01 3.51384073e-01
4.03640658e-01 4.99480903e-01 8.82468402e-01 3.35246265e-01
3.09436291e-01 3.78627270e-01 6.21469021e-01 -2.11492311e-02
2.48718157e-01 -2.82132059e-01 2.97021776e-01 4.17550832e-01
4.52412516e-01 -3.50247383e-01 -7.79061973e-01 3.66697311e-01
-1.87173343e+00 -4.40633535e-01 -2.73364317e-02 2.56790090e+00
1.02767086e+00 8.99768651e-01 -1.98360786e-01 3.39204341e-01
6.31321073e-01 -2.47098550e-01 -1.05103397e+00 -4.66069162e-01
1.32222399e-01 -2.76406080e-01 5.85629821e-01 2.75659204e-01
-1.19388616e+00 4.09459889e-01 5.46405458e+00 9.97861743e-01
-1.20570147e+00 -1.87571004e-01 1.04117966e+00 -1.28622249e-01
-1.63613439e-01 -6.57500803e-01 -1.00648844e+00 9.18733418e-01
6.83977842e-01 2.29955122e-01 1.94940884e-02 1.03887820e+00
2.26996750e-01 -2.60474056e-01 -1.36145175e+00 9.85308170e-01
-2.86771536e-01 -7.36610889e-01 -3.43043327e-01 -5.28471358e-02
6.93462849e-01 1.21039279e-01 1.05341211e-01 6.48821652e-01
2.68826485e-01 -1.21489894e+00 7.17279732e-01 4.90454614e-01
6.01900101e-01 -1.01244450e+00 1.28325498e+00 6.90823972e-01
-5.78550160e-01 -2.56121755e-01 -6.55144632e-01 2.10335851e-01
-5.75513355e-02 1.18543530e+00 -8.58852983e-01 1.81034461e-01
7.97518313e-01 3.16168904e-01 -1.84260890e-01 9.73884165e-01
-4.88228053e-01 5.62335193e-01 -6.60262465e-01 -1.39154062e-01
2.68708020e-01 1.32200062e-01 4.82009500e-01 8.25940311e-01
3.04070711e-01 -1.47625297e-01 4.03801724e-02 1.05951273e+00
-1.32547379e-01 -2.84492254e-01 -3.44047695e-01 5.41217327e-02
7.64260232e-01 7.16859341e-01 -3.12195241e-01 -7.32482076e-02
-1.25489831e-01 5.51260054e-01 2.20293805e-01 1.70599937e-01
-9.37330127e-01 -7.02945292e-01 5.96998930e-01 -8.01441073e-02
-4.67958264e-02 3.36130202e-01 -5.08626878e-01 -7.73316860e-01
4.04849917e-01 -7.40442097e-01 3.70801806e-01 -6.18333280e-01
-1.49421597e+00 7.58442283e-01 -1.61556259e-01 -1.44888401e+00
-3.42623353e-01 -8.59640718e-01 -5.13061702e-01 8.16175997e-01
-1.71383429e+00 -3.17472696e-01 -2.41800785e-01 2.41318822e-01
3.38070124e-01 -1.06745668e-01 5.33049226e-01 3.57814282e-01
-8.37474346e-01 1.01201475e+00 2.20964864e-01 1.29801869e-01
1.11328554e+00 -1.37690043e+00 9.43674296e-02 5.52935839e-01
-5.21412313e-01 4.03781831e-01 7.41106451e-01 -7.75006890e-01
-7.80572653e-01 -1.26262987e+00 6.54389083e-01 -3.90173197e-01
3.46776724e-01 -1.08589627e-01 -1.17573094e+00 2.55281210e-01
-4.37170923e-01 2.50538170e-01 4.57317144e-01 -6.09275214e-02
-2.38157719e-01 -3.88445914e-01 -1.37741399e+00 3.48110586e-01
6.03321135e-01 -2.12245762e-01 -8.14170659e-01 -8.54583271e-03
7.67659962e-01 -7.28348970e-01 -1.00197399e+00 8.71972620e-01
5.59763134e-01 -8.30086529e-01 7.32702851e-01 -3.24034035e-01
3.75636607e-01 -3.20727408e-01 7.82110542e-03 -1.46305454e+00
-1.42302543e-01 -9.09445807e-02 -4.79959697e-01 1.22655892e+00
6.26206875e-01 -9.07165706e-01 5.56108534e-01 8.94494832e-01
-2.50237286e-01 -1.12467229e+00 -1.09156370e+00 -9.09199536e-01
9.31356847e-03 -6.19661510e-01 6.96099460e-01 8.75503123e-01
-4.36532050e-02 2.33572200e-01 -4.29544657e-01 4.29362506e-01
5.50241649e-01 -4.88660574e-01 3.83410692e-01 -1.35875666e+00
-3.41809660e-01 -4.10473436e-01 -5.02075911e-01 -1.01159954e+00
-9.43830162e-02 -5.01824021e-01 4.14599180e-01 -1.28216815e+00
-2.36291438e-01 -5.50159335e-01 -6.65609658e-01 5.59451878e-01
-4.54445601e-01 -1.63501486e-01 -1.89239532e-01 1.04179561e-01
-3.43676597e-01 6.46817386e-01 1.24966812e+00 -2.08364889e-01
-2.81150877e-01 3.58642370e-01 -6.82887018e-01 8.80369961e-01
6.39597297e-01 -7.03855991e-01 -6.90192044e-01 -2.92799145e-01
1.35314271e-01 -5.03453650e-02 1.78945392e-01 -9.27446246e-01
7.16095492e-02 -2.03361288e-01 8.02527845e-01 -4.89656538e-01
3.47683392e-02 -1.08351827e+00 -2.02013046e-01 3.71417165e-01
-2.72721201e-01 -3.79589945e-01 2.92462766e-01 8.08498263e-01
-2.71201372e-01 -3.61886591e-01 7.95499325e-01 2.93643713e-01
-5.71698606e-01 2.80311733e-01 -2.88868964e-01 2.60193765e-01
1.04377127e+00 -3.02284747e-01 -2.40023032e-01 -2.06150264e-01
-4.39951032e-01 8.90127003e-01 3.96471024e-02 8.09055150e-01
5.76452672e-01 -1.39264679e+00 -5.62952638e-01 5.67479193e-01
4.42495584e-01 5.90250015e-01 1.86583325e-01 6.54149950e-01
-7.28480592e-02 3.04217100e-01 1.19282342e-01 -8.03320408e-01
-8.05920064e-01 3.32013398e-01 5.18309355e-01 -1.50701270e-01
-3.91421288e-01 9.09487069e-01 3.39646995e-01 -4.36469138e-01
9.76135850e-01 -6.29655361e-01 -1.75770551e-01 -1.51531711e-01
6.53903902e-01 3.89728099e-01 2.68712252e-01 -1.63678452e-01
-3.22610855e-01 -3.54085676e-02 -3.38048100e-01 4.99403059e-01
9.43499029e-01 -5.32004461e-02 3.90996546e-01 6.05265677e-01
9.31709468e-01 -4.04461831e-01 -1.61366391e+00 -1.15146689e-01
-4.28325310e-03 -2.51383990e-01 5.24266899e-01 -1.25333381e+00
-8.43171954e-01 8.79665792e-01 9.35303152e-01 6.03876859e-02
8.59920323e-01 -2.45258376e-01 7.50979364e-01 5.04936755e-01
1.49902388e-01 -1.57358789e+00 -8.10578186e-03 4.06370699e-01
7.23953426e-01 -1.69358063e+00 -3.00047576e-01 -4.07830998e-02
-8.82756472e-01 1.00601959e+00 9.94358838e-01 6.62361160e-02
7.90384173e-01 5.82556963e-01 1.93854511e-01 8.60916451e-02
-7.39541233e-01 4.99219686e-01 5.43324113e-01 5.72688997e-01
1.88332126e-01 1.20312713e-01 -1.04900323e-01 1.32152843e+00
-1.44358939e-02 6.55603856e-02 2.51360148e-01 7.49325335e-01
-5.16143024e-01 -5.51493168e-01 -3.63047600e-01 9.22649980e-01
-4.80983734e-01 1.15209930e-01 -8.61789808e-02 7.30089366e-01
4.13938135e-01 8.93281817e-01 2.47174561e-01 -4.79221255e-01
4.81099695e-01 -5.46490075e-03 -8.36730152e-02 -3.07900816e-01
-6.42330423e-02 -1.38500705e-02 1.06451176e-01 -4.76712883e-01
3.06860924e-01 -2.46951759e-01 -1.69815099e+00 -7.42662996e-02
-7.03417718e-01 4.60613742e-02 5.53801954e-01 1.03516424e+00
4.21368837e-01 1.05060542e+00 6.05769694e-01 -2.91185379e-01
-1.26563036e+00 -1.00491369e+00 -7.00006247e-01 2.37226129e-01
6.45172238e-01 -9.85423863e-01 -6.50461078e-01 -4.87045228e-01]
|
[7.78695821762085, 3.705505132675171]
|
5459497f-aeeb-445a-9c8e-168669b64e3c
|
egodistill-egocentric-head-motion
|
2301.02217
| null |
https://arxiv.org/abs/2301.02217v1
|
https://arxiv.org/pdf/2301.02217v1.pdf
|
EgoDistill: Egocentric Head Motion Distillation for Efficient Video Understanding
|
Recent advances in egocentric video understanding models are promising, but their heavy computational expense is a barrier for many real-world applications. To address this challenge, we propose EgoDistill, a distillation-based approach that learns to reconstruct heavy egocentric video clip features by combining the semantics from a sparse set of video frames with the head motion from lightweight IMU readings. We further devise a novel self-supervised training strategy for IMU feature learning. Our method leads to significant improvements in efficiency, requiring 200x fewer GFLOPs than equivalent video models. We demonstrate its effectiveness on the Ego4D and EPICKitchens datasets, where our method outperforms state-of-the-art efficient video understanding methods.
|
['Kristen Grauman', 'Tushar Nagarajan', 'Shuhan Tan']
|
2023-01-05
| null | null | null | null |
['video-understanding']
|
['computer-vision']
|
[-1.38597354e-01 -1.49506956e-01 -5.73933005e-01 -4.33112621e-01
-7.24623382e-01 -4.71108198e-01 5.73680878e-01 -4.27921265e-01
-2.55575329e-01 4.25734311e-01 7.01519608e-01 4.06769775e-02
7.44250044e-02 -4.00120974e-01 -9.88594115e-01 -4.57503796e-01
-1.28692284e-01 3.52595985e-01 1.56020328e-01 4.23438102e-02
4.56416547e-01 1.54186273e-03 -1.64927387e+00 3.12338263e-01
3.02977681e-01 8.24287236e-01 3.80480915e-01 1.10119462e+00
1.46618828e-01 1.31179690e+00 1.65544748e-01 -6.50357408e-03
3.74279171e-01 -3.44763368e-01 -1.00125563e+00 2.66816884e-01
7.57957399e-01 -1.03340113e+00 -1.10017514e+00 6.24986470e-01
4.47419107e-01 3.40785593e-01 4.10336822e-01 -1.27571630e+00
-2.71960825e-01 2.19791174e-01 -5.51608503e-01 5.04260242e-01
7.75902033e-01 -3.99549082e-02 1.00123346e+00 -9.22085702e-01
8.94601405e-01 1.16228640e+00 5.58125913e-01 5.36301315e-01
-9.52242494e-01 -5.31079948e-01 3.45332384e-01 7.68498242e-01
-1.19150543e+00 -8.36070180e-01 5.25594473e-01 -3.00377518e-01
1.06938434e+00 -1.85495540e-01 6.74474955e-01 1.13935089e+00
-5.39768487e-02 1.37031758e+00 4.90109801e-01 -1.75031558e-01
1.55126616e-01 -2.89390653e-01 2.91254278e-02 8.76685441e-01
-4.62261736e-02 -1.53270513e-01 -1.08381152e+00 -1.26583338e-01
1.00712121e+00 1.54997647e-01 -3.31665397e-01 -8.41150522e-01
-1.37378967e+00 7.70863533e-01 6.78387806e-02 -2.06973299e-01
-3.05320650e-01 8.10285091e-01 5.97527325e-01 4.32702899e-01
6.86865270e-01 -1.72131866e-01 -6.67538464e-01 -7.75942087e-01
-8.65101695e-01 4.73970830e-01 8.70801866e-01 1.32823169e+00
9.26980793e-01 -7.91718159e-03 5.52578270e-01 4.13300663e-01
1.96774304e-01 5.02681792e-01 4.22204196e-01 -1.63444066e+00
3.80623549e-01 1.43103451e-01 1.45929512e-02 -9.14177954e-01
-1.93175703e-01 5.30866943e-02 -2.25637063e-01 -2.80914783e-01
1.87802911e-01 -8.22476074e-02 -7.56260693e-01 1.73335361e+00
5.33289671e-01 1.05194831e+00 2.97038797e-02 9.10159588e-01
6.55992389e-01 7.08768785e-01 -7.25378841e-02 -3.45237516e-02
1.05762506e+00 -1.31769967e+00 -5.22549450e-01 -2.18648762e-01
9.52304840e-01 -4.48058516e-01 6.97524488e-01 3.28590393e-01
-1.19894850e+00 -3.80058914e-01 -8.29131007e-01 -2.98233956e-01
6.61482438e-02 -2.75144726e-01 1.03698444e+00 2.84036189e-01
-1.12313890e+00 6.17461741e-01 -1.13524318e+00 -6.61654651e-01
4.04211700e-01 2.98925012e-01 -7.32377112e-01 -3.44066709e-01
-6.86512411e-01 4.87301171e-01 4.46109891e-01 -4.58894998e-01
-1.28716481e+00 -9.37815368e-01 -1.27644777e+00 4.50021401e-02
5.96522748e-01 -1.05198073e+00 1.42213702e+00 -9.36159015e-01
-1.43916500e+00 7.88483262e-01 -5.94523966e-01 -5.98422885e-01
3.57322842e-01 -9.85230207e-01 -8.11442584e-02 6.92932308e-01
1.21134713e-01 8.25974882e-01 1.03114581e+00 -9.34934855e-01
-7.75230169e-01 -3.07248741e-01 2.71164328e-01 6.31836772e-01
-3.51695985e-01 -1.18676595e-01 -9.47236776e-01 -5.61472178e-01
3.22626650e-01 -9.97713149e-01 -8.42186138e-02 8.08265507e-02
4.38051596e-02 -7.51708597e-02 1.13042891e+00 -4.29487377e-01
8.45335662e-01 -1.98316574e+00 4.06631052e-01 -2.39501879e-01
4.01302457e-01 6.39547110e-02 -1.55814737e-01 2.15916052e-01
-1.45972539e-02 -2.67014503e-01 3.39061141e-01 -5.29275060e-01
-3.81774232e-02 3.40169072e-01 -4.44084257e-01 6.65488780e-01
-2.56036401e-01 8.80917311e-01 -1.30248797e+00 -4.81174707e-01
5.90215087e-01 5.26705086e-01 -1.15971696e+00 3.77542078e-01
-1.56478077e-01 3.87872308e-01 -5.98900914e-01 6.96020961e-01
4.32431042e-01 -3.77962321e-01 1.97317958e-01 -2.44925618e-01
1.74706161e-01 2.59509683e-01 -9.54269290e-01 2.45653296e+00
-3.78743500e-01 8.82879555e-01 -1.85145333e-01 -1.11030376e+00
2.43114367e-01 4.50467229e-01 8.24050784e-01 -4.26315606e-01
1.26143664e-01 -8.95407051e-02 -7.46113777e-01 -6.82048142e-01
5.47524035e-01 1.29685789e-01 2.06232886e-04 5.21724761e-01
4.21034127e-01 -5.57846278e-02 5.56547977e-02 6.36074066e-01
1.23762476e+00 8.41187894e-01 3.37843508e-01 -2.46177003e-01
3.05900306e-01 3.20186745e-03 6.35366857e-01 7.24711895e-01
-4.45957989e-01 9.14195240e-01 5.03488421e-01 -8.63162398e-01
-1.08123970e+00 -1.09870267e+00 3.09428364e-01 1.24789464e+00
4.16018575e-01 -9.59636450e-01 -1.01037478e+00 -7.53139734e-01
-1.42687038e-01 4.37129319e-01 -4.68269438e-01 1.40272062e-02
-7.47009039e-01 -2.96277374e-01 2.23439038e-01 8.17037642e-01
5.32927513e-01 -4.87450272e-01 -5.50186217e-01 4.99406457e-02
-5.95610499e-01 -1.69882095e+00 -6.25960171e-01 -2.96100646e-01
-1.15017736e+00 -1.22153580e+00 -5.39827943e-01 -5.98425746e-01
5.46380401e-01 1.13917077e+00 1.36266541e+00 1.57977454e-02
-1.43325210e-01 1.15677881e+00 -4.80922669e-01 -9.11737010e-02
1.99038178e-01 6.94656149e-02 5.26496291e-01 -1.50991887e-01
7.33747303e-01 -7.34075427e-01 -8.50373745e-01 2.57492572e-01
-6.30207837e-01 2.78930604e-01 -5.27826278e-03 6.94075108e-01
6.74966216e-01 -5.42938769e-01 2.69472837e-01 -8.06775510e-01
-1.87020004e-01 -8.42367768e-01 -3.18534106e-01 -1.42686397e-01
-2.75302500e-01 5.11414604e-03 4.47456270e-01 -1.91677853e-01
-9.75853741e-01 2.63657987e-01 1.55719578e-01 -9.78476405e-01
-5.65901659e-02 1.74672812e-01 -2.79797371e-02 -1.96508348e-01
3.48890901e-01 3.27388942e-01 5.14634065e-02 -4.15100724e-01
6.28658414e-01 3.50994200e-01 7.78687477e-01 -7.07788765e-01
7.85027564e-01 9.61096287e-01 -1.43406495e-01 -1.12161565e+00
-1.02227020e+00 -8.40313733e-01 -7.06125259e-01 -1.63848445e-01
8.22545290e-01 -1.66429818e+00 -7.54185796e-01 2.78647900e-01
-1.09074247e+00 -3.01362842e-01 -8.69811401e-02 7.62834311e-01
-1.41068101e+00 7.59575069e-01 -5.66386342e-01 -3.19676727e-01
-2.10102588e-01 -9.73155737e-01 1.22263658e+00 2.18731523e-01
-2.82659203e-01 -1.05908871e+00 2.38707349e-01 6.31124616e-01
6.71574995e-02 7.05936342e-04 3.57313454e-01 -4.03334498e-01
-1.12099779e+00 -1.20205909e-01 -3.05813164e-01 1.42844632e-01
-1.38942629e-01 -3.98946673e-01 -9.35073137e-01 -3.76339406e-01
-9.27878693e-02 -5.50217867e-01 9.18278813e-01 4.03221846e-01
1.28841162e+00 -4.14202549e-02 -3.52950275e-01 1.22433019e+00
1.33051050e+00 -3.15489143e-01 6.43416822e-01 4.15641487e-01
8.77052486e-01 1.73900753e-01 7.68303096e-01 7.23260283e-01
8.37096453e-01 6.22322440e-01 4.43462133e-01 2.73016065e-01
-1.52547076e-01 -4.76700127e-01 5.49097061e-01 9.19280171e-01
-4.18893158e-01 -2.26833135e-01 -5.31859398e-01 8.75181019e-01
-2.32486224e+00 -1.18205035e+00 2.94013798e-01 2.00708723e+00
1.27621651e-01 -1.09721959e-01 2.25867634e-03 -3.14873964e-01
3.93467456e-01 6.40022635e-01 -5.94028592e-01 -4.78369966e-02
1.04594320e-01 -1.38331875e-01 4.45050806e-01 4.62121457e-01
-1.31901109e+00 1.30113757e+00 7.32858849e+00 4.12328333e-01
-7.35092461e-01 2.38936320e-01 2.94678301e-01 -5.03882170e-01
-1.32228345e-01 2.30195448e-01 -5.94519556e-01 3.37598830e-01
9.66555536e-01 -2.99295664e-01 5.56495607e-01 1.28187323e+00
3.07200134e-01 -2.09855840e-01 -1.34041905e+00 1.47885692e+00
5.73019445e-01 -1.54557145e+00 9.51590240e-02 -6.11186437e-02
8.58655751e-01 5.64545393e-01 -1.89921021e-01 2.10368708e-01
1.67946979e-01 -6.29798412e-01 4.67830956e-01 3.49665076e-01
5.93029916e-01 -6.36708617e-01 4.16965574e-01 1.99686348e-01
-1.46379972e+00 2.68338397e-02 -6.23235822e-01 -2.28072003e-01
5.34819841e-01 1.55102819e-01 -5.61944246e-01 3.70357841e-01
8.98232818e-01 1.31239438e+00 -5.34780957e-02 6.58596456e-01
-7.23169744e-02 5.80027819e-01 -4.27402377e-01 4.70141619e-01
3.38957101e-01 -1.11718766e-01 7.05435634e-01 1.12474358e+00
3.68763298e-01 4.04766560e-01 1.48197204e-01 3.00856113e-01
-2.82693923e-01 -1.28411651e-01 -1.02196300e+00 -1.12856152e-02
2.48279616e-01 1.00808918e+00 -3.69899154e-01 -6.33132279e-01
-7.92639792e-01 1.42065918e+00 3.79509479e-01 3.81755739e-01
-9.70257103e-01 -8.34489837e-02 1.33366835e+00 4.67293262e-02
5.30803680e-01 -5.77904522e-01 2.24018067e-01 -1.90402472e+00
-2.04386204e-01 -7.15997756e-01 4.48315769e-01 -9.13974285e-01
-7.44431376e-01 8.17335099e-02 9.09020081e-02 -1.20315814e+00
-7.34862864e-01 -5.20392060e-01 -5.83611727e-01 1.96456984e-01
-1.63541961e+00 -1.20173502e+00 -6.73852623e-01 8.84153008e-01
1.06776094e+00 -2.24156365e-01 6.56359434e-01 3.35854560e-01
-3.24135542e-01 3.97729218e-01 1.89689726e-01 1.75267905e-02
7.91961491e-01 -1.09697485e+00 7.93966055e-01 6.93980932e-01
2.98379958e-01 4.71210510e-01 7.11249948e-01 -4.31163549e-01
-2.08694673e+00 -8.71747494e-01 8.13560665e-01 -8.04397583e-01
6.82404160e-01 -2.18292266e-01 -3.88610780e-01 1.36991835e+00
-4.85677784e-03 3.14927280e-01 6.47410631e-01 1.73267990e-01
-5.38604856e-01 4.62505482e-02 -7.16378272e-01 7.61962891e-01
1.58748281e+00 -7.83548713e-01 -7.38256216e-01 5.21303773e-01
7.83454657e-01 -5.10740519e-01 -7.87233770e-01 1.61396310e-01
7.28118896e-01 -1.05389118e+00 1.31504214e+00 -8.50153744e-01
5.97686589e-01 -2.18360141e-01 -4.56278682e-01 -1.02274919e+00
-2.21607238e-01 -8.92660260e-01 -9.19246316e-01 6.54901087e-01
-2.58289933e-01 -1.75579369e-01 1.30841112e+00 6.55179679e-01
-3.27521563e-02 -3.88742298e-01 -7.69555926e-01 -6.59597516e-01
-4.42913860e-01 -6.25829697e-01 1.86205417e-01 7.43489385e-01
3.11379582e-01 1.84912264e-01 -9.24216986e-01 -3.24360095e-02
9.69676554e-01 -4.18434013e-03 1.33616805e+00 -7.00819194e-01
-3.00318390e-01 2.44811177e-01 -1.01165271e+00 -1.80445254e+00
4.17620689e-01 -5.56001544e-01 -1.81903869e-01 -1.27638936e+00
5.58243036e-01 1.23457670e-01 -2.51321048e-01 1.97484359e-01
-1.19728453e-01 4.90399420e-01 3.21563154e-01 1.39075920e-01
-1.24119818e+00 6.15558028e-01 8.76971424e-01 1.40076831e-01
1.26712685e-02 -3.82205456e-01 -5.37377298e-01 1.13700998e+00
4.68365461e-01 -2.73350656e-01 -8.21981311e-01 -8.64241540e-01
4.49866522e-03 5.79268932e-02 2.94662893e-01 -1.09014010e+00
3.77213657e-01 -8.49282667e-02 2.72603869e-01 -7.61319399e-01
6.79559588e-01 -6.47521198e-01 -1.40929237e-01 1.91025808e-01
1.33322338e-02 1.66727945e-01 1.98857542e-02 9.28780794e-01
-1.76618591e-01 1.53009519e-01 3.97537619e-01 -3.15737933e-01
-1.28503156e+00 6.83695972e-01 -4.13040310e-01 3.12223673e-01
1.06875145e+00 -3.60263497e-01 -1.76533207e-01 -9.32916760e-01
-5.64254880e-01 3.30816060e-01 7.58495212e-01 5.63418627e-01
7.87407577e-01 -1.24072433e+00 -5.86891949e-01 7.34242648e-02
1.22803308e-01 -7.33846426e-02 6.19293094e-01 7.79597461e-01
-8.15131068e-01 4.72801924e-01 -5.70292957e-02 -8.58747184e-01
-1.30568898e+00 6.16933286e-01 -1.49646588e-02 1.66210428e-01
-1.10916889e+00 7.26655066e-01 8.29179287e-01 -1.16711244e-01
3.08497638e-01 1.29907593e-01 1.43917099e-01 -2.95177341e-01
8.87289464e-01 6.84014857e-01 -4.36341166e-01 -6.90109253e-01
-2.11685598e-01 7.67490566e-01 -2.92164773e-01 4.52022292e-02
1.46705639e+00 -6.04555905e-01 3.26386333e-01 2.38425955e-01
1.52778101e+00 -3.77577901e-01 -1.77993417e+00 -4.49895531e-01
-2.72987098e-01 -8.38882804e-01 3.01487535e-01 2.59537995e-02
-1.00275552e+00 6.24094903e-01 3.64132822e-01 -4.39229965e-01
1.00186348e+00 6.37419596e-02 1.28055727e+00 7.95368969e-01
7.21341789e-01 -1.33265948e+00 2.83473194e-01 7.19574571e-01
3.60609502e-01 -1.56440532e+00 3.48009288e-01 -4.18185174e-01
-6.13500416e-01 1.05707955e+00 4.81516629e-01 -4.32276577e-01
7.30736434e-01 6.10577203e-02 -2.90213544e-02 -2.39160866e-01
-9.07734275e-01 -1.20582759e-01 -1.25352442e-01 5.07087111e-01
1.48632869e-01 -2.30615318e-01 -2.59600300e-02 2.64788836e-01
1.37456834e-01 3.09813976e-01 5.56659400e-01 1.02425027e+00
-4.45717365e-01 -7.74803400e-01 -5.14029823e-02 1.19745634e-01
-4.15845066e-01 -2.61789933e-02 1.00062095e-01 8.42133224e-01
-3.68250102e-01 7.00536311e-01 2.27587491e-01 -4.42769259e-01
-4.90117669e-02 4.20962200e-02 7.06741869e-01 -3.58315289e-01
6.01081215e-02 7.94177428e-02 1.21225424e-01 -1.37770844e+00
-6.42645359e-01 -7.97001839e-01 -1.26105344e+00 -8.76325130e-01
-4.30447012e-02 -3.41330804e-02 6.45571053e-01 1.06613922e+00
7.29049385e-01 5.58504239e-02 7.29173958e-01 -1.31958437e+00
-2.56981760e-01 -4.82728094e-01 -3.25321108e-01 6.15205586e-01
3.24340403e-01 -7.80626059e-01 -4.61643815e-01 4.41485852e-01]
|
[8.565838813781738, 0.5670571327209473]
|
3d0539c0-d4e1-4842-9821-5aa2521fff79
|
semi-supervised-deep-regression-with
|
2302.07579
| null |
https://arxiv.org/abs/2302.07579v1
|
https://arxiv.org/pdf/2302.07579v1.pdf
|
Semi-Supervised Deep Regression with Uncertainty Consistency and Variational Model Ensembling via Bayesian Neural Networks
|
Deep regression is an important problem with numerous applications. These range from computer vision tasks such as age estimation from photographs, to medical tasks such as ejection fraction estimation from echocardiograms for disease tracking. Semi-supervised approaches for deep regression are notably under-explored compared to classification and segmentation tasks, however. Unlike classification tasks, which rely on thresholding functions for generating class pseudo-labels, regression tasks use real number target predictions directly as pseudo-labels, making them more sensitive to prediction quality. In this work, we propose a novel approach to semi-supervised regression, namely Uncertainty-Consistent Variational Model Ensembling (UCVME), which improves training by generating high-quality pseudo-labels and uncertainty estimates for heteroscedastic regression. Given that aleatoric uncertainty is only dependent on input data by definition and should be equal for the same inputs, we present a novel uncertainty consistency loss for co-trained models. Our consistency loss significantly improves uncertainty estimates and allows higher quality pseudo-labels to be assigned greater importance under heteroscedastic regression. Furthermore, we introduce a novel variational model ensembling approach to reduce prediction noise and generate more robust pseudo-labels. We analytically show our method generates higher quality targets for unlabeled data and further improves training. Experiments show that our method outperforms state-of-the-art alternatives on different tasks and can be competitive with supervised methods that use full labels. Our code is available at https://github.com/xmed-lab/UCVME.
|
['Kwang-Ting Cheng', 'Xiaomeng Li', 'Weihang Dai']
|
2023-02-15
| null | null | null | null |
['age-estimation', 'age-estimation']
|
['computer-vision', 'miscellaneous']
|
[ 1.87249258e-02 3.78792018e-01 -3.97644818e-01 -7.85720706e-01
-1.38242185e+00 -3.93755168e-01 2.67453313e-01 3.83065306e-02
-3.69494468e-01 1.19035769e+00 -1.13623239e-01 -1.40472800e-01
1.91578902e-02 -4.91234422e-01 -8.93839180e-01 -7.35765994e-01
2.36681715e-01 8.24770868e-01 5.90376034e-02 3.83703530e-01
-1.73937500e-01 -2.10784655e-02 -1.05446410e+00 -3.06757167e-02
1.23209107e+00 1.04594326e+00 -1.76016912e-01 6.14781976e-01
1.87937215e-01 6.00996673e-01 -5.11927664e-01 -5.12451351e-01
1.46392152e-01 -5.15914977e-01 -5.79094529e-01 -5.51771745e-02
4.79249567e-01 -2.39725620e-01 -8.18435289e-03 1.03017390e+00
5.14844120e-01 9.93294716e-02 1.20761383e+00 -1.20469642e+00
-5.04398286e-01 7.63397038e-01 -5.96139789e-01 -1.56964183e-01
-4.19930547e-01 -3.06849871e-02 9.94101107e-01 -8.40137661e-01
4.69277412e-01 1.04147589e+00 9.01528180e-01 8.20049644e-01
-1.56080973e+00 -8.16540718e-01 -5.95203182e-03 -2.64212102e-01
-1.29983878e+00 -2.71397144e-01 5.16892314e-01 -6.10116124e-01
3.50272089e-01 1.33264512e-01 1.84350118e-01 1.15430117e+00
2.33769596e-01 8.04529548e-01 1.27988064e+00 -2.77333558e-01
2.91337341e-01 4.27095175e-01 1.11811005e-01 6.96682334e-01
1.50031865e-01 1.74999595e-01 -2.22805396e-01 -2.67687261e-01
1.00436306e+00 8.46903324e-02 -2.89091229e-01 -6.31009042e-01
-1.24618840e+00 9.47777152e-01 3.10169101e-01 -5.10261022e-02
-1.93397805e-01 4.41699296e-01 2.97342896e-01 -1.07390292e-01
8.01509440e-01 4.52356875e-01 -6.27780318e-01 -7.72550479e-02
-1.31780052e+00 2.09820971e-01 5.48919141e-01 8.94370675e-01
4.99514043e-01 1.42246053e-01 -6.22017324e-01 1.21079373e+00
5.02896845e-01 7.86807656e-01 1.65180147e-01 -1.36754453e+00
2.99625188e-01 2.92526245e-01 5.64210524e-04 -4.28535730e-01
-4.65919882e-01 -7.15111256e-01 -1.16993093e+00 2.58476585e-01
7.00590193e-01 -3.44241768e-01 -1.19799697e+00 1.91360986e+00
3.37241858e-01 2.36460999e-01 -1.12397760e-01 9.38599646e-01
8.33829045e-01 5.16766369e-01 4.57099602e-02 -2.79861480e-01
1.15192211e+00 -8.72302055e-01 -5.80700219e-01 -7.93334190e-03
6.00383878e-01 -3.84757996e-01 8.21214736e-01 5.15344739e-01
-1.01014256e+00 -5.09928167e-01 -8.81518006e-01 4.92118904e-03
1.39252350e-01 3.16257149e-01 3.69788826e-01 6.44737303e-01
-7.35549390e-01 9.79940116e-01 -1.26159692e+00 2.28797525e-01
8.25492799e-01 3.39657098e-01 -1.41559273e-01 1.84611112e-01
-1.11020052e+00 7.78589666e-01 1.12281017e-01 6.78879917e-02
-8.77304912e-01 -1.07668293e+00 -1.07813334e+00 -1.74467623e-01
4.18539852e-01 -6.55473173e-01 1.30487192e+00 -8.42458725e-01
-1.52225566e+00 8.84694457e-01 -1.19974919e-01 -7.09037483e-01
9.76236820e-01 -1.47453442e-01 3.89487483e-02 1.02460673e-02
8.36324766e-02 9.55840945e-01 1.04188693e+00 -1.20571566e+00
-2.74091125e-01 -3.32629651e-01 -5.11276007e-01 -8.13943520e-02
2.09398828e-02 -1.75749168e-01 -4.11214024e-01 -7.53623486e-01
1.38494015e-01 -1.05871904e+00 -5.07184803e-01 2.29941487e-01
-6.87453210e-01 -1.24375582e-01 2.23663434e-01 -8.18333805e-01
1.05670595e+00 -1.83380675e+00 2.83204585e-01 2.75916129e-01
5.81667364e-01 2.78533772e-02 3.28452438e-01 -3.59246075e-01
1.28062917e-02 3.66476804e-01 -9.49289083e-01 -7.69154251e-01
4.60606255e-02 2.56562203e-01 -1.63191661e-01 4.58325267e-01
4.96325016e-01 1.00659978e+00 -8.28773558e-01 -8.43041539e-01
2.09318846e-01 5.79991162e-01 -4.09554869e-01 9.55809504e-02
-3.32392752e-01 1.06590521e+00 -3.18111598e-01 5.38557410e-01
6.31918311e-01 -4.94236588e-01 3.52246612e-02 8.48312527e-02
4.34829772e-01 -4.10604030e-02 -9.75069702e-01 1.49139321e+00
-5.35461307e-01 4.09530252e-01 -1.44239262e-01 -9.94984984e-01
1.03240836e+00 2.89813071e-01 4.84833062e-01 -1.37672812e-01
1.14628360e-01 3.42924893e-01 -1.02953270e-01 1.06897525e-01
1.78092659e-01 -2.96433538e-01 -9.45848078e-02 1.53489545e-01
1.53481081e-01 -3.38041395e-01 2.21175663e-02 2.22375304e-01
7.65222311e-01 6.52432442e-01 1.07046187e-01 -2.24179819e-01
2.64429837e-01 -5.98102324e-02 1.08403742e+00 7.75171340e-01
-2.27137744e-01 1.17327714e+00 7.06748843e-01 6.51943758e-02
-1.11071527e+00 -1.29859173e+00 -7.75856793e-01 5.81219614e-01
-1.72820494e-01 -1.44770265e-01 -7.25322664e-01 -1.02229023e+00
1.40218720e-01 6.84194803e-01 -6.26878202e-01 -4.20353711e-02
-3.36408496e-01 -9.84660208e-01 4.26622272e-01 8.57593715e-01
1.51575342e-01 -7.79388726e-01 -2.71562487e-01 5.53737581e-02
-8.91521871e-02 -9.76585329e-01 -5.39181292e-01 3.17315161e-01
-1.25142932e+00 -8.59863698e-01 -1.33833313e+00 -2.94484615e-01
8.21002424e-01 -6.92348421e-01 1.32653844e+00 -2.74791777e-01
-3.65596265e-01 2.15023249e-01 -6.54097721e-02 -4.88331437e-01
-6.81639433e-01 2.78141294e-02 6.93515763e-02 -1.25825971e-01
-1.78427305e-02 -3.98070693e-01 -6.33141816e-01 4.50643539e-01
-5.26528060e-01 2.16304939e-02 4.38566506e-01 1.19590437e+00
1.19547105e+00 -3.41535985e-01 6.51278973e-01 -1.47201002e+00
1.03886493e-01 -3.95630151e-01 -7.84056783e-01 1.79643720e-01
-9.50199544e-01 3.42070460e-01 5.04113853e-01 -5.27812302e-01
-1.11304951e+00 1.08190514e-01 -8.78953412e-02 -9.37877178e-01
1.07438728e-01 2.87845939e-01 2.49220684e-01 2.70742387e-01
9.12577152e-01 -1.85032964e-01 3.10383171e-01 -3.91935438e-01
2.77749151e-01 5.31295002e-01 4.32962149e-01 -7.46938884e-01
5.76226532e-01 3.63843739e-01 3.15177172e-01 -4.07906502e-01
-8.78428936e-01 -2.05766931e-01 -5.59090614e-01 -1.45858288e-01
8.16955805e-01 -9.68531013e-01 -3.95673543e-01 4.29426312e-01
-8.37075710e-01 -6.41105533e-01 -4.78568941e-01 7.06164479e-01
-6.79981411e-01 3.80567789e-01 -6.95793509e-01 -9.22511220e-01
-4.96397704e-01 -1.25141513e+00 1.18706226e+00 1.89197078e-01
-2.76513934e-01 -1.18205857e+00 6.91459104e-02 4.78590697e-01
2.12183729e-01 4.31658864e-01 6.58384323e-01 -7.38334596e-01
-3.82283449e-01 -7.71274976e-03 -2.08618224e-01 7.79482841e-01
-1.52138442e-01 1.51052684e-01 -9.79110539e-01 -2.53728777e-03
-3.51149291e-01 -5.40301681e-01 1.21239245e+00 1.08014607e+00
1.49231899e+00 1.55077279e-01 -3.79821360e-01 6.74363732e-01
1.06302965e+00 -2.86677778e-01 5.86359084e-01 -2.20475301e-01
8.19059610e-01 7.42568314e-01 7.77758300e-01 3.54692012e-01
2.95551449e-01 5.21861970e-01 1.81698114e-01 -2.96770662e-01
-8.06306079e-02 -1.37959048e-01 6.83978051e-02 6.37421548e-01
-2.12345980e-02 2.63240077e-02 -9.39421177e-01 3.61574292e-01
-2.10557103e+00 -4.70367938e-01 -4.82634127e-01 2.49364066e+00
1.24325716e+00 1.55415803e-01 4.55681570e-02 -2.42968336e-01
7.48815298e-01 -2.44162977e-01 -8.42065811e-01 -5.28872050e-02
1.30972758e-01 3.93128961e-01 6.44183517e-01 4.35064793e-01
-1.15636289e+00 7.53766716e-01 5.58332968e+00 9.72025037e-01
-7.56317198e-01 4.10786331e-01 1.36596274e+00 -6.50757924e-02
-3.67885739e-01 -1.18671276e-01 -8.29155624e-01 5.33497930e-01
9.53701258e-01 3.64853829e-01 -6.04552142e-02 1.01245546e+00
5.07915728e-02 -1.56665832e-01 -1.29358971e+00 8.71882319e-01
-2.26963088e-01 -1.19873619e+00 -3.66299182e-01 -2.66771950e-02
1.00413251e+00 -1.77976713e-02 1.44865826e-01 5.23411930e-01
3.88468862e-01 -1.13287079e+00 5.33434749e-01 7.45870173e-01
1.04292655e+00 -6.45904481e-01 8.05330992e-01 1.56918496e-01
-6.94510639e-01 4.25163448e-01 -4.72758710e-01 4.40431058e-01
1.82325974e-01 1.13953209e+00 -6.76550567e-01 4.39852834e-01
6.13949776e-01 8.46661091e-01 -2.84976214e-01 8.31885099e-01
-2.98288554e-01 8.33376586e-01 -2.99883813e-01 1.38698384e-01
-2.72882372e-01 -2.80896962e-01 4.08619434e-01 9.54347312e-01
2.61316210e-01 -1.41836673e-01 7.15309605e-02 1.42660296e+00
-1.90125152e-01 4.19612266e-02 -1.48334160e-01 8.57783854e-02
3.46480817e-01 1.26561248e+00 -7.05314875e-01 -3.46497953e-01
-1.02498762e-01 7.93537855e-01 2.41404176e-01 2.91298509e-01
-1.13680577e+00 -1.23227216e-01 3.04798424e-01 9.26640630e-02
9.15980414e-02 1.41422570e-01 -5.79685807e-01 -1.24468470e+00
-3.90110314e-02 -5.92613280e-01 4.29567605e-01 -7.92395890e-01
-1.45313179e+00 4.85256612e-01 1.65770769e-01 -1.31352234e+00
-4.77803469e-01 -6.48544133e-01 -3.62649560e-01 9.87070560e-01
-1.45754898e+00 -1.00391543e+00 -8.84881467e-02 1.59793109e-01
3.88573587e-01 1.63917523e-03 6.88369691e-01 1.58426404e-01
-6.04843497e-01 6.80733502e-01 3.60609442e-01 2.43267730e-01
1.14457226e+00 -1.65116560e+00 1.56720355e-01 7.40700424e-01
1.13254443e-01 4.60285366e-01 5.37853360e-01 -8.29871118e-01
-6.47639155e-01 -1.22527349e+00 5.08220911e-01 -7.78584421e-01
4.08081561e-01 -2.12736547e-01 -9.76191819e-01 5.88939846e-01
-3.72127235e-01 4.16652679e-01 6.43985510e-01 2.20420256e-01
-2.53422320e-01 -1.22218817e-01 -1.22147393e+00 3.22574705e-01
7.90085077e-01 -1.50879994e-01 -1.34888887e-01 4.26377118e-01
6.45357192e-01 -7.67682076e-01 -1.24948573e+00 8.76078129e-01
5.06199718e-01 -7.62318671e-01 7.81786442e-01 -4.33661073e-01
7.93025315e-01 -1.51776329e-01 6.98815063e-02 -1.28791881e+00
9.68285501e-02 -4.15694982e-01 -3.14095229e-01 1.32345533e+00
9.07598138e-01 -6.77167416e-01 9.75826204e-01 8.22659791e-01
-8.28702524e-02 -1.09109259e+00 -8.18933129e-01 -7.27698624e-01
4.67893273e-01 -5.82794726e-01 1.04472116e-01 7.79164612e-01
-3.92641604e-01 8.92000720e-02 -5.73672116e-01 9.27207023e-02
1.06260848e+00 5.48515934e-03 5.87682068e-01 -1.47140539e+00
-7.07791030e-01 -3.31839830e-01 -2.34777808e-01 -7.02739298e-01
3.13402355e-01 -9.56528544e-01 3.78778249e-01 -1.35358024e+00
3.28644216e-01 -9.06645596e-01 -2.30148524e-01 4.73890930e-01
-4.48536515e-01 4.45619702e-01 -1.25951350e-01 3.09613943e-01
-4.11296666e-01 6.36415243e-01 1.22109032e+00 -9.20277834e-02
-2.70477235e-01 5.12478173e-01 -4.80568022e-01 6.11213744e-01
7.74016619e-01 -7.05528438e-01 -3.59990031e-01 -3.55422460e-02
4.05268744e-02 3.32531691e-01 3.93054038e-01 -8.10065567e-01
-2.23724753e-01 1.12149410e-01 6.11255169e-01 -2.86847919e-01
3.19091827e-01 -5.34314036e-01 7.30954036e-02 3.79849225e-01
-4.86363292e-01 -5.37326932e-01 7.29304999e-02 6.11982882e-01
-3.62884738e-02 -3.94420981e-01 1.03976274e+00 -3.02714799e-02
-9.59641114e-02 4.37344670e-01 -2.26258524e-02 5.02314270e-01
8.62439811e-01 5.28653637e-02 -2.30030715e-02 -4.58436996e-01
-1.09479761e+00 6.12050533e-01 1.87449083e-01 5.07687740e-02
5.13131738e-01 -1.01955760e+00 -1.07118690e+00 -1.26731545e-01
5.04425727e-02 5.02014875e-01 3.16616625e-01 1.11972487e+00
-2.92498797e-01 1.50793403e-01 2.24841192e-01 -1.12934804e+00
-1.14706659e+00 2.17356741e-01 4.89626139e-01 -4.46703643e-01
-4.06650037e-01 1.01578999e+00 3.34889919e-01 -8.45027566e-01
3.78823578e-01 -5.74835479e-01 6.48148824e-03 -2.81017512e-01
1.75225332e-01 4.80946481e-01 -1.03153430e-01 -3.09121609e-01
-1.94366321e-01 5.38636684e-01 -3.63963023e-02 -1.93527207e-01
1.10335863e+00 9.96091440e-02 4.79122018e-03 7.38333642e-01
8.76082003e-01 -2.59021968e-01 -1.62949717e+00 -3.75067979e-01
-1.81307256e-01 -1.74575433e-01 2.16647044e-01 -9.88928318e-01
-1.13683009e+00 1.09387779e+00 6.39815509e-01 -2.91755348e-01
7.91297913e-01 1.32815763e-01 4.31077629e-01 2.79243067e-02
1.66233435e-01 -1.01100135e+00 -4.84422827e-03 3.23796682e-02
5.91360331e-01 -1.71741569e+00 1.46352723e-01 -6.27927661e-01
-1.12793076e+00 8.41634870e-01 6.75380170e-01 -3.01199369e-02
7.57239342e-01 3.35088432e-01 9.15079862e-02 1.73729822e-01
-5.11979342e-01 -4.28735167e-02 7.75011957e-01 6.07218266e-01
5.37678540e-01 2.51022160e-01 -1.96712643e-01 8.55361164e-01
8.60534906e-02 5.79922684e-02 3.19745630e-01 5.18830001e-01
-1.19615197e-01 -1.22514832e+00 -3.88220072e-01 9.83023942e-01
-6.78704083e-01 -1.08307086e-01 4.38864678e-02 5.03358543e-01
-1.44524306e-01 6.28814161e-01 7.57219940e-02 -7.40616247e-02
-1.37190536e-01 1.90694362e-01 5.36913991e-01 -8.12376738e-01
-2.62574822e-01 2.29323104e-01 4.11758497e-02 -3.74371439e-01
-3.82342786e-01 -7.18614340e-01 -1.55598950e+00 8.31303671e-02
-4.94182587e-01 1.36392355e-01 6.50344193e-01 8.93655658e-01
2.09114373e-01 7.34150648e-01 3.92107368e-01 -6.46010220e-01
-9.22633052e-01 -1.05504262e+00 -6.15947664e-01 2.37344891e-01
2.04601422e-01 -8.49963903e-01 -3.30096692e-01 7.02459887e-02]
|
[14.524795532226562, -2.0681979656219482]
|
a0e35e95-2772-421a-b7b0-21f15aeaf6d3
|
cascade-graph-neural-networks-for-rgb-d
|
2008.03087
| null |
https://arxiv.org/abs/2008.03087v1
|
https://arxiv.org/pdf/2008.03087v1.pdf
|
Cascade Graph Neural Networks for RGB-D Salient Object Detection
|
In this paper, we study the problem of salient object detection (SOD) for RGB-D images using both color and depth information.A major technical challenge in performing salient object detection fromRGB-D images is how to fully leverage the two complementary data sources. Current works either simply distill prior knowledge from the corresponding depth map for handling the RGB-image or blindly fuse color and geometric information to generate the coarse depth-aware representations, hindering the performance of RGB-D saliency detectors.In this work, we introduceCascade Graph Neural Networks(Cas-Gnn),a unified framework which is capable of comprehensively distilling and reasoning the mutual benefits between these two data sources through a set of cascade graphs, to learn powerful representations for RGB-D salient object detection. Cas-Gnn processes the two data sources individually and employs a novelCascade Graph Reasoning(CGR) module to learn powerful dense feature embeddings, from which the saliency map can be easily inferred. Contrast to the previous approaches, the explicitly modeling and reasoning of high-level relations between complementary data sources allows us to better overcome challenges such as occlusions and ambiguities. Extensive experiments demonstrate that Cas-Gnn achieves significantly better performance than all existing RGB-DSOD approaches on several widely-used benchmarks.
|
['Siwei Lyu', 'Hong Cheng', 'Fan Yang', 'Ao Luo', 'Zhicheng Jiao', 'Xin Li']
|
2020-08-07
| null |
https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/1571_ECCV_2020_paper.php
|
https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123570341.pdf
|
eccv-2020-8
|
['rgb-d-salient-object-detection']
|
['computer-vision']
|
[ 3.27148736e-02 3.15924942e-01 -1.93607569e-01 -2.76574880e-01
-5.28664351e-01 -3.52353036e-01 4.45579320e-01 2.20368475e-01
-6.80732876e-02 2.66849190e-01 2.32938617e-01 -2.52464086e-01
1.30931035e-01 -8.24943304e-01 -6.68371558e-01 -5.81666350e-01
2.03428790e-01 -1.93766896e-02 8.25618804e-01 -6.08803570e-01
2.14468077e-01 7.61662781e-01 -1.93298292e+00 1.50851846e-01
7.74055600e-01 1.35134268e+00 4.07629311e-01 5.65019071e-01
-2.56174356e-01 1.01539993e+00 -1.61326140e-01 -1.71839535e-01
4.44778085e-01 -1.39099911e-01 -6.18983448e-01 2.01237239e-02
5.64319670e-01 -6.11897230e-01 -5.42626083e-01 1.22905982e+00
4.02452230e-01 4.70391996e-02 2.34172478e-01 -1.47545826e+00
-1.00251150e+00 3.52486759e-01 -8.57269943e-01 4.84499663e-01
4.24249232e-01 1.33133546e-01 9.68588948e-01 -1.13626206e+00
4.72116560e-01 1.40177512e+00 4.97995585e-01 3.84936184e-01
-9.05845940e-01 -4.69476044e-01 4.99798745e-01 2.51523793e-01
-1.15855539e+00 -1.61238555e-02 1.36048567e+00 -1.05051249e-01
8.84222448e-01 8.12368095e-02 1.01295042e+00 8.27195227e-01
-1.64755911e-01 1.32256114e+00 9.73359585e-01 -2.73061275e-01
2.86618650e-01 -9.23762769e-02 1.10733323e-01 9.17074680e-01
4.30000454e-01 2.33747587e-01 -9.46762681e-01 2.13917330e-01
1.25521553e+00 4.90845144e-01 -2.23006442e-01 -9.72092450e-01
-1.16713071e+00 7.50262141e-01 1.48587239e+00 6.17014803e-02
-3.17297280e-01 2.22439364e-01 -1.12146251e-01 -9.11946371e-02
3.34346801e-01 3.70876282e-01 -1.09722003e-01 3.52325648e-01
-7.08456933e-01 9.96217355e-02 3.61717254e-01 1.19972813e+00
1.11152375e+00 9.21623781e-02 -1.65093452e-01 2.07317933e-01
4.73355830e-01 7.09133089e-01 2.41579607e-01 -6.38102949e-01
4.53167230e-01 1.22873187e+00 1.72829866e-01 -1.30035472e+00
-5.72689414e-01 -2.68023044e-01 -6.10214055e-01 5.18651903e-01
2.96103358e-01 3.15949947e-01 -1.26526344e+00 1.41699290e+00
5.15623868e-01 2.05620080e-01 5.74266687e-02 1.46219969e+00
1.43269908e+00 2.77082026e-01 -7.28646219e-02 5.93712866e-01
1.10059059e+00 -1.08743572e+00 -4.46454018e-01 -7.14291275e-01
1.64885879e-01 -5.02658904e-01 1.07851970e+00 -2.15521097e-01
-1.09640074e+00 -5.39313734e-01 -1.30985975e+00 -9.92701530e-01
-8.45966995e-01 6.31733611e-02 1.04757035e+00 2.12418005e-01
-1.40079105e+00 1.71074182e-01 -9.28026021e-01 -1.76671371e-01
6.24954998e-01 2.79887378e-01 -2.38856614e-01 -3.34689826e-01
-1.05196953e+00 8.85947227e-01 3.82122278e-01 3.84529591e-01
-9.34075654e-01 -6.65801466e-01 -1.28042459e+00 -5.57804964e-02
4.28230047e-01 -6.78544641e-01 1.00785589e+00 -6.53629303e-01
-1.11252427e+00 1.00884616e+00 -6.69107437e-02 -1.96174443e-01
3.19946289e-01 -2.87743390e-01 -4.57360335e-02 4.14885759e-01
2.26423338e-01 9.87294495e-01 9.57043409e-01 -1.38039756e+00
-7.72154808e-01 -5.36271214e-01 6.09575331e-01 3.81121546e-01
-1.49076596e-01 -3.92545342e-01 -7.02786803e-01 -4.86447185e-01
7.72472203e-01 -5.64428806e-01 -2.58605033e-01 5.47024131e-01
-5.87035835e-01 -1.30762279e-01 1.00474429e+00 -3.91082764e-01
6.01114869e-01 -2.02915883e+00 5.01163960e-01 -7.11030588e-02
7.82439530e-01 1.59429505e-01 -1.20785475e-01 6.04149923e-02
1.12874702e-01 -2.93003082e-01 -1.19205981e-01 -4.14390713e-01
-3.48623022e-02 3.61770540e-01 -4.76128519e-01 4.23835903e-01
6.93959713e-01 1.41624749e+00 -1.34992611e+00 -5.00421047e-01
6.42723441e-01 6.89162195e-01 -3.55938852e-01 2.91266710e-01
-2.21079886e-01 6.25811815e-02 -7.21416473e-01 1.17084324e+00
7.88232446e-01 -4.56581265e-01 -1.87793776e-01 -3.49663317e-01
-7.32610002e-02 3.48841697e-01 -1.05387342e+00 2.07847118e+00
-2.63939559e-01 6.75059617e-01 -1.07012719e-01 -8.47501755e-01
1.05518556e+00 -3.76532376e-01 1.87237591e-01 -8.84799957e-01
3.17371368e-01 1.63184941e-01 -6.08285010e-01 -2.22030804e-01
6.10306859e-01 1.79231256e-01 -1.06064439e-01 2.96639889e-01
1.47485241e-01 -5.40030539e-01 -2.66034067e-01 4.65475857e-01
8.58432233e-01 2.40477860e-01 2.02186242e-01 -1.46346182e-01
3.01928699e-01 1.03793547e-01 3.81354570e-01 8.06635320e-01
-4.02469695e-01 9.29697692e-01 4.91473049e-01 -6.71749890e-01
-6.05360508e-01 -1.49119890e+00 3.01990300e-01 1.04211438e+00
1.08902287e+00 -1.38722524e-01 -2.42945343e-01 -7.70683289e-01
3.49878013e-01 3.32471043e-01 -9.92001832e-01 -3.65167320e-01
-3.89151871e-01 -4.42539632e-01 -3.49268317e-03 9.07115757e-01
7.22579777e-01 -8.18613350e-01 -1.06018579e+00 -8.76156241e-02
1.93885609e-03 -1.21349478e+00 -6.48148209e-02 4.93427336e-01
-9.13958490e-01 -1.08302581e+00 -6.62819088e-01 -9.17022109e-01
7.64784992e-01 1.07009602e+00 1.15171254e+00 3.00243676e-01
-5.16151607e-01 3.52543563e-01 -5.14166057e-01 -5.59681475e-01
1.67145044e-01 5.58846258e-02 -3.25831592e-01 -2.40063563e-01
4.70289499e-01 -5.19370794e-01 -8.34261119e-01 1.75098311e-02
-9.69151378e-01 5.16388774e-01 7.26200223e-01 5.02598703e-01
7.06433594e-01 -4.31644917e-01 1.81676164e-01 -2.67321974e-01
9.82189365e-03 -3.85543257e-01 -6.47665262e-01 1.93374187e-01
-2.99993932e-01 1.32673785e-01 -5.11667021e-02 -6.87458515e-02
-7.96528578e-01 3.33132774e-01 6.14594519e-02 -7.32672036e-01
3.63244526e-02 1.58303127e-01 -1.41162038e-01 -3.39050531e-01
4.95627105e-01 1.34942681e-01 -3.52714658e-02 -4.44003969e-01
7.50101864e-01 2.27933154e-01 7.59646833e-01 -1.88880950e-01
1.13789546e+00 7.86034346e-01 7.27403462e-02 -5.27055383e-01
-1.32547069e+00 -5.50367296e-01 -8.62503111e-01 -1.27320126e-01
8.15580785e-01 -1.25724423e+00 -4.67040211e-01 5.14809728e-01
-1.07906568e+00 -4.30411279e-01 -3.09870571e-01 1.22657530e-01
-4.19567853e-01 7.24542364e-02 -4.70785469e-01 -5.40965140e-01
-1.17047578e-01 -1.08918762e+00 1.67771327e+00 7.43299663e-01
3.30540657e-01 -8.97265971e-01 -3.53560060e-01 3.45202871e-02
3.57123166e-01 4.86459851e-01 6.79258585e-01 -7.12691694e-02
-1.22054279e+00 -1.86074495e-01 -9.14086640e-01 -9.09596682e-03
2.29604334e-01 -4.02361840e-01 -1.16545486e+00 -2.52230912e-02
-1.15793265e-01 -3.48511279e-01 1.02506077e+00 3.13169241e-01
1.06200445e+00 9.99683514e-02 -3.71346831e-01 8.48714530e-01
1.57058513e+00 -4.46170807e-01 4.23431754e-01 4.25546348e-01
1.23014665e+00 3.75209242e-01 6.24834299e-01 2.90847123e-01
8.64997685e-01 4.22539830e-01 1.09444022e+00 -7.09581673e-01
-5.43963850e-01 -5.95747590e-01 1.23714142e-01 3.81964922e-01
-1.98320858e-02 9.82454494e-02 -9.28579807e-01 6.27172291e-01
-1.98675442e+00 -5.06245315e-01 -3.32651922e-04 1.59288347e+00
7.34712601e-01 9.38412771e-02 4.53773737e-02 1.39472991e-01
5.46973407e-01 4.02280211e-01 -9.03365314e-01 7.49663338e-02
-4.35375065e-01 -1.70415069e-03 5.64265490e-01 3.73116821e-01
-1.15014517e+00 1.11728477e+00 5.88921595e+00 2.99121469e-01
-1.22200966e+00 -2.60201879e-02 4.17915612e-01 -4.15479183e-01
-5.50609887e-01 -4.99453023e-02 -7.80968428e-01 -3.46715264e-02
9.14421398e-03 1.06503300e-01 2.81135350e-01 1.16270769e+00
-2.86423653e-01 -2.64899671e-01 -1.09350979e+00 1.21668577e+00
3.29387486e-01 -1.57633698e+00 1.46054953e-01 -2.34928533e-01
8.06925774e-01 3.47744614e-01 2.95383006e-01 7.43521526e-02
6.39161587e-01 -7.42396712e-01 1.02899098e+00 3.80864978e-01
4.16557193e-01 -5.10158539e-01 6.65646136e-01 -3.17257568e-02
-1.33863163e+00 -2.51439482e-01 -4.79977876e-01 -2.40125030e-01
-2.99052056e-03 6.63660765e-01 -6.43356264e-01 6.00729406e-01
1.09886467e+00 1.25637996e+00 -9.79509830e-01 8.78403604e-01
-7.50877678e-01 -2.13481426e-01 -3.77826750e-01 -1.61939114e-01
4.12045211e-01 1.74995571e-01 4.22563463e-01 7.13468730e-01
1.47456616e-01 1.16642818e-01 1.73507124e-01 1.25127959e+00
1.06810756e-01 -4.21883941e-01 -5.43206990e-01 2.25981817e-01
2.99325794e-01 1.29643989e+00 -9.92613792e-01 -1.93136796e-01
-5.00803530e-01 1.05153847e+00 6.53818429e-01 5.39387524e-01
-7.93437243e-01 -3.49121660e-01 7.59007633e-01 3.59633844e-03
6.38007700e-01 -3.42340648e-01 -4.59083289e-01 -1.26969790e+00
9.85839292e-02 -3.73343915e-01 3.53169203e-01 -1.31985855e+00
-1.06714499e+00 5.10446846e-01 -7.49721453e-02 -1.15045524e+00
3.35443430e-02 -8.38512242e-01 -4.24970835e-01 8.85985970e-01
-2.41980672e+00 -1.58956409e+00 -8.21329653e-01 8.44947040e-01
2.49952078e-01 3.00231308e-01 4.03166175e-01 -1.37352079e-01
-4.31240082e-01 2.00398043e-01 -4.11289215e-01 1.70248225e-01
1.59256011e-01 -1.47100949e+00 5.01017153e-01 1.11586154e+00
3.00941199e-01 5.14068544e-01 4.60949183e-01 -4.33041245e-01
-1.69717848e+00 -1.00596380e+00 4.48086590e-01 -7.07884073e-01
4.76914912e-01 -6.22668862e-01 -7.05321252e-01 5.68607032e-01
-1.64870664e-01 5.76587498e-01 1.38901919e-01 -1.15861885e-01
-6.67237163e-01 -4.29325588e-02 -8.72061729e-01 6.33349836e-01
1.36101246e+00 -8.67380083e-01 -8.84841442e-01 8.37824270e-02
1.10389256e+00 -7.97769845e-01 -4.72955346e-01 3.16061914e-01
2.14598596e-01 -1.34200037e+00 1.49173534e+00 -3.61010015e-01
6.15084648e-01 -7.00848162e-01 -2.81839073e-01 -1.07643557e+00
-1.34672318e-02 -2.57218152e-01 -3.88835073e-01 8.61952722e-01
1.30810076e-02 -3.46245140e-01 7.52900362e-01 5.55771887e-01
-2.78599173e-01 -7.91015506e-01 -7.78556526e-01 -3.06215107e-01
-3.75664294e-01 -3.94885421e-01 9.59912479e-01 6.55497193e-01
-1.39614940e-01 3.79761569e-02 -4.07424159e-02 5.80098331e-01
6.91023707e-01 8.67735744e-01 7.39288211e-01 -1.23455763e+00
9.92799848e-02 -5.29061258e-01 -8.17362010e-01 -1.33602870e+00
-2.00864837e-01 -9.31561649e-01 2.52654493e-01 -1.95394325e+00
-5.29251061e-02 -4.77449536e-01 -6.18077755e-01 7.65963614e-01
-4.71644074e-01 5.17713964e-01 2.26343691e-01 -7.61250407e-02
-8.22626948e-01 7.40145981e-01 1.39393878e+00 -2.69606769e-01
-3.35949600e-01 -4.76395637e-01 -1.00330949e+00 7.21086264e-01
5.28734148e-01 -2.13365242e-01 -5.31738400e-01 -7.63496399e-01
3.70370090e-01 -1.98142424e-01 1.04621589e+00 -1.02717757e+00
4.10559595e-01 -1.15945697e-01 7.46466637e-01 -9.66264546e-01
4.06997114e-01 -6.90227687e-01 -5.95616877e-01 1.54585719e-01
-7.04888254e-03 -5.53860776e-02 4.21138048e-01 7.34483838e-01
-2.69681782e-01 3.12556267e-01 4.76914197e-01 -1.38878495e-01
-1.32570124e+00 4.08748746e-01 1.55584142e-01 2.46520862e-02
9.20033514e-01 -4.62730885e-01 -4.77181792e-01 -2.47519925e-01
-4.23929453e-01 2.55456507e-01 7.30746090e-01 8.39513183e-01
1.05630529e+00 -1.40485597e+00 -1.49099782e-01 5.64098895e-01
4.41649109e-01 5.53910732e-01 3.26404572e-01 7.50032604e-01
-3.60216230e-01 1.75804868e-01 -4.95766640e-01 -9.03914332e-01
-6.75736547e-01 7.19362020e-01 2.30131701e-01 2.19506323e-01
-8.19353163e-01 1.17246878e+00 3.73415798e-01 -2.27883816e-01
3.29988390e-01 -8.58073890e-01 -1.10809207e-01 2.21693236e-02
5.29311061e-01 4.59307171e-02 3.18226703e-02 -6.35542154e-01
-5.81290603e-01 6.96418762e-01 1.43493637e-01 2.85575151e-01
1.50253689e+00 -4.33205992e-01 -1.00513846e-01 3.37735713e-01
1.11026382e+00 -3.75920445e-01 -1.71119225e+00 -3.87243807e-01
-1.44160539e-01 -6.35292888e-01 5.02189577e-01 -4.40905720e-01
-1.32269633e+00 1.13602996e+00 6.01217568e-01 8.14443231e-02
1.23077583e+00 3.58326435e-01 6.53500497e-01 2.83300430e-01
4.31987345e-01 -8.42682183e-01 5.83356261e-01 1.97017267e-01
9.27609861e-01 -1.59935546e+00 1.58326879e-01 -5.64597964e-01
-5.23914635e-01 1.01633704e+00 7.42183030e-01 -3.04413229e-01
6.31168842e-01 7.36467615e-02 1.47717029e-01 -5.43449283e-01
-2.09278986e-01 -8.38616371e-01 5.84414423e-01 8.89572561e-01
-2.08522752e-01 -8.33833143e-02 6.46010280e-01 5.29591620e-01
-6.11815080e-02 -2.30624840e-01 3.91011178e-01 1.12496281e+00
-5.13519406e-01 -6.06813848e-01 -2.00334191e-01 1.76131174e-01
2.06038207e-01 -1.23206630e-01 -5.33084154e-01 9.41488385e-01
4.85732816e-02 7.98489511e-01 1.08871400e-01 -3.63184661e-01
2.86699295e-01 -4.50558513e-01 3.64411741e-01 -6.08870506e-01
-1.01120070e-01 -1.96352825e-01 -4.87768710e-01 -1.02897310e+00
-7.09752202e-01 -3.38983566e-01 -1.40228689e+00 6.56258613e-02
-2.45735958e-01 -4.50205296e-01 5.95232487e-01 8.29675138e-01
4.63803947e-01 6.11502767e-01 3.87811810e-01 -1.40663314e+00
8.24972522e-03 -3.67919743e-01 -5.55551171e-01 3.86084437e-01
8.48484039e-01 -1.16317725e+00 -3.79981101e-01 -2.73243308e-01]
|
[9.714618682861328, -0.7133990526199341]
|
83f8156c-2c5a-44e9-911b-7279114b74f1
|
a-convolutional-attention-network-for-extreme
|
1602.03001
| null |
http://arxiv.org/abs/1602.03001v2
|
http://arxiv.org/pdf/1602.03001v2.pdf
|
A Convolutional Attention Network for Extreme Summarization of Source Code
|
Attention mechanisms in neural networks have proved useful for problems in
which the input and output do not have fixed dimension. Often there exist
features that are locally translation invariant and would be valuable for
directing the model's attention, but previous attentional architectures are not
constructed to learn such features specifically. We introduce an attentional
neural network that employs convolution on the input tokens to detect local
time-invariant and long-range topical attention features in a context-dependent
way. We apply this architecture to the problem of extreme summarization of
source code snippets into short, descriptive function name-like summaries.
Using those features, the model sequentially generates a summary by
marginalizing over two attention mechanisms: one that predicts the next summary
token based on the attention weights of the input tokens and another that is
able to copy a code token as-is directly into the summary. We demonstrate our
convolutional attention neural network's performance on 10 popular Java
projects showing that it achieves better performance compared to previous
attentional mechanisms.
|
['Miltiadis Allamanis', 'Charles Sutton', 'Hao Peng']
|
2016-02-09
| null | null | null | null |
['extreme-summarization']
|
['natural-language-processing']
|
[ 2.58192182e-01 1.26892567e-01 -1.68015182e-01 -3.32623720e-01
-6.71745002e-01 -3.70989412e-01 5.75398207e-01 3.38509500e-01
-1.37382716e-01 1.58819541e-01 5.65494537e-01 -4.32713896e-01
8.18577856e-02 -5.08839846e-01 -7.98537195e-01 -3.42236131e-01
-2.51900971e-01 3.17607448e-02 2.91200131e-01 -5.91905825e-02
8.25371742e-01 2.27325827e-01 -1.69639778e+00 7.49969184e-01
6.59463942e-01 5.10750890e-01 5.58747172e-01 1.06887507e+00
-4.83435005e-01 1.02245605e+00 -9.45014775e-01 3.94497775e-02
-1.47442907e-01 -3.81232679e-01 -1.14085317e+00 -4.17885393e-01
5.75282693e-01 -4.85608168e-02 -2.55851269e-01 9.03184831e-01
2.01931328e-01 1.57309070e-01 5.95649064e-01 -9.44451988e-01
-1.30084121e+00 9.50588167e-01 -5.50826848e-01 7.44469583e-01
2.38344625e-01 1.50944009e-01 1.24609184e+00 -9.50189292e-01
5.60299337e-01 1.11470866e+00 6.97254002e-01 5.33799171e-01
-1.19969833e+00 -1.19077750e-01 2.23608136e-01 9.32115689e-02
-8.34327817e-01 -3.42216313e-01 5.31495750e-01 -6.48712635e-01
1.81708646e+00 3.21673274e-01 3.37488562e-01 8.78122687e-01
7.74709642e-01 8.11919868e-01 1.65554136e-01 -4.94768262e-01
1.05897181e-01 -1.96986049e-01 4.46357697e-01 6.65549040e-01
7.13808276e-03 -4.17563826e-01 -5.46008706e-01 -3.09997320e-01
5.10763645e-01 3.19790155e-01 -1.44561633e-01 6.06577145e-03
-1.55072629e+00 7.04459727e-01 7.01181173e-01 6.28073096e-01
-5.49224794e-01 7.91396737e-01 4.68171507e-01 3.81669015e-01
3.22464675e-01 7.83562660e-01 -7.14620829e-01 -1.99437141e-01
-9.50029254e-01 -2.00694683e-03 5.17807961e-01 1.05336916e+00
9.12746668e-01 -2.79040597e-02 -7.23964453e-01 5.91058195e-01
7.09598064e-02 1.36116803e-01 8.56777728e-01 -8.41485798e-01
3.71201038e-01 8.58572721e-01 -1.42519316e-02 -7.22465217e-01
-3.07179153e-01 -3.65168989e-01 -3.17618757e-01 1.44507632e-01
7.35342354e-02 -2.75237020e-02 -1.03159904e+00 1.64828718e+00
-2.68833101e-01 -1.67331249e-01 -1.02144949e-01 5.54283917e-01
7.54760563e-01 7.40901053e-01 -1.43762995e-02 1.83461979e-01
1.17871523e+00 -1.03026271e+00 -4.52918500e-01 -4.93552506e-01
7.57876158e-01 -8.69266927e-01 1.25845218e+00 -1.38698563e-01
-1.25130892e+00 -6.15450263e-01 -9.16401923e-01 -3.84530246e-01
-4.67348695e-01 1.77436367e-01 5.96360147e-01 2.00084969e-01
-1.31659055e+00 1.02419329e+00 -8.73275995e-01 -6.17846727e-01
4.53767896e-01 4.03175384e-01 -1.33334443e-01 3.13423127e-01
-7.92030990e-01 7.36088037e-01 2.10475594e-01 8.33342504e-03
-8.00230920e-01 -6.55578077e-01 -8.68076980e-01 7.97327816e-01
-9.95943844e-02 -7.53452837e-01 1.44243348e+00 -1.27543783e+00
-1.17681968e+00 5.35323322e-01 -3.19139481e-01 -2.02895343e-01
1.19105829e-02 -1.34225115e-01 -1.29388282e-02 -6.61614686e-02
5.02898932e-01 6.94572508e-01 8.64981711e-01 -5.92007935e-01
-6.13290012e-01 -1.92757040e-01 1.99196517e-01 -1.29267156e-01
-5.74077964e-01 3.29722285e-01 -4.71905619e-01 -6.91774368e-01
-1.48700371e-01 -6.63853765e-01 -2.36590698e-01 -2.43972003e-01
-4.83552694e-01 -4.86393422e-01 7.20714152e-01 -4.30303752e-01
1.29517317e+00 -2.10487151e+00 1.47323787e-01 -1.16539516e-01
9.85593647e-02 4.99444194e-02 -4.35733080e-01 5.65015912e-01
-3.61225694e-01 3.88396293e-01 -2.39023894e-01 -1.34968787e-01
-3.89950536e-02 -1.60973549e-01 -5.43767631e-01 2.31731042e-01
6.88764095e-01 1.02421284e+00 -1.09758961e+00 -2.55799145e-01
-2.64672935e-01 2.90150106e-01 -9.48136508e-01 2.52196729e-01
-4.06945080e-01 -7.57629275e-02 -4.62038547e-01 4.18470323e-01
2.61876057e-03 -4.07898039e-01 2.10195817e-02 1.36748508e-01
-4.21938211e-01 6.31426811e-01 -4.24170375e-01 1.87704277e+00
-6.17600620e-01 1.24112308e+00 -3.81008774e-01 -8.32109034e-01
8.17866921e-01 1.73860401e-01 1.78715661e-01 -5.19100070e-01
8.16223491e-03 6.73195049e-02 1.67159647e-01 -7.67280936e-01
9.24869537e-01 2.64027625e-01 -3.93239290e-01 8.64884734e-01
1.22459993e-01 3.43422890e-01 1.99795336e-01 2.86273897e-01
1.67108083e+00 3.67022127e-01 -2.82810517e-02 -5.66039145e-01
4.01394367e-01 -7.19405711e-02 3.89546007e-01 9.82606530e-01
1.01796530e-01 7.18467355e-01 1.00828516e+00 -6.03395700e-01
-1.08645296e+00 -5.90967238e-01 2.51458347e-01 1.77327728e+00
-2.23262310e-01 -6.16423190e-01 -7.52037406e-01 -7.84227014e-01
-9.89625789e-03 7.18051016e-01 -1.18183756e+00 -3.19870025e-01
-7.38518238e-01 -1.91822007e-01 3.63642693e-01 9.45598006e-01
-1.81022044e-02 -1.63578212e+00 -1.04947639e+00 3.18774164e-01
-4.99556698e-02 -2.03216016e-01 -8.52908552e-01 5.48782706e-01
-7.76521325e-01 -8.93028438e-01 -7.21597672e-01 -8.70007992e-01
8.76072168e-01 1.19678244e-01 1.18897045e+00 3.01415443e-01
-4.48019743e-01 2.39368975e-01 -7.12574422e-02 -5.26034594e-01
-4.66424704e-01 6.04154050e-01 -3.63184750e-01 -3.13972771e-01
5.35738647e-01 -4.29903358e-01 -4.35278922e-01 -1.78827256e-01
-9.27107155e-01 -9.65872407e-02 7.05601752e-01 9.00965631e-01
5.08466214e-02 -5.40039301e-01 7.07634091e-01 -9.23223019e-01
7.58487940e-01 -7.10239649e-01 -3.10855657e-01 2.38161966e-01
-7.91203007e-02 7.09418297e-01 7.52661824e-01 -4.23915654e-01
-8.81099582e-01 -4.09368314e-02 7.55598769e-02 -3.58695596e-01
1.01671681e-01 7.17387736e-01 2.76703179e-01 4.47485805e-01
8.50273311e-01 2.80843467e-01 -2.66947746e-01 -4.85316545e-01
1.69877067e-01 6.37925029e-01 5.04222870e-01 -6.22832000e-01
3.68113220e-01 2.22805887e-01 -4.44766045e-01 -6.68533087e-01
-4.92113829e-01 -4.47535068e-01 -6.52577341e-01 9.75455046e-02
9.00513172e-01 -3.42659920e-01 -5.94936550e-01 8.88005570e-02
-1.54049170e+00 -3.74570191e-01 -3.45535308e-01 1.57613203e-01
-6.50842488e-01 1.28609136e-01 -5.53415596e-01 -4.39125985e-01
-4.12034094e-01 -1.16418111e+00 1.00885010e+00 2.39742190e-01
-5.22283852e-01 -8.52585852e-01 1.23178877e-01 -4.99955922e-01
8.12761009e-01 3.61558348e-02 1.23224545e+00 -8.38251591e-01
-6.94718897e-01 -3.62839811e-02 -2.04396337e-01 1.72634237e-02
6.11701719e-02 4.57975715e-01 -1.07085681e+00 -1.55313075e-01
-2.75753886e-01 4.90700677e-02 1.21990156e+00 4.01876301e-01
1.31243110e+00 -4.33899075e-01 -5.31196594e-01 4.26737607e-01
1.28868449e+00 9.58415791e-02 6.51051164e-01 3.53617311e-01
4.74190652e-01 3.78482848e-01 2.39380449e-01 3.38783532e-01
1.30750641e-01 2.00541228e-01 6.53024256e-01 4.75401841e-02
-8.39788932e-03 1.23835325e-01 6.18346453e-01 7.58302629e-01
-4.07045968e-02 -9.58515108e-02 -9.41592634e-01 1.26106524e+00
-1.80543971e+00 -1.15802622e+00 -6.62534013e-02 2.08674049e+00
8.03127408e-01 2.55949467e-01 -2.17314005e-01 -2.92121410e-01
8.07202101e-01 1.75727457e-01 -5.18218100e-01 -1.11544263e+00
3.81719947e-01 3.08478028e-01 1.66733816e-01 1.47872493e-01
-1.02473164e+00 7.29801834e-01 6.67378426e+00 1.78253487e-01
-1.19582880e+00 -1.90085024e-02 2.44581640e-01 -3.66502665e-02
-4.69420910e-01 1.29309505e-01 -5.20257413e-01 5.03140152e-01
1.22148979e+00 -3.45437139e-01 1.98624566e-01 1.07991338e+00
-9.22866985e-02 -9.78948362e-03 -1.58691442e+00 2.63212323e-01
-7.24207386e-02 -1.68082166e+00 -4.56526177e-03 -1.63127467e-01
8.26575756e-01 2.52074927e-01 1.02810487e-01 5.69148123e-01
2.40265548e-01 -9.11549211e-01 8.34221840e-01 7.08227038e-01
6.11327112e-01 -6.30368471e-01 6.54803455e-01 1.42840877e-01
-1.04044914e+00 -2.99896419e-01 -6.04309201e-01 -1.96539417e-01
-6.02570295e-01 1.56657666e-01 -9.92700577e-01 3.48496996e-02
7.67129064e-01 8.10913622e-01 -6.38192773e-01 1.07476294e+00
-6.95810020e-02 4.10257012e-01 1.59965575e-01 -3.35432649e-01
4.95375425e-01 4.28582877e-01 5.14768600e-01 1.70443368e+00
4.52095836e-01 -3.14707100e-01 -2.55442321e-01 1.23327112e+00
-1.88313946e-01 4.04963940e-02 -7.11541712e-01 -2.67834961e-01
3.46862763e-01 1.14226520e+00 -8.15317392e-01 -5.47063708e-01
-6.71181798e-01 1.08443296e+00 4.06563073e-01 3.61929268e-01
-8.17893028e-01 -1.19839585e+00 8.20349813e-01 -1.92403719e-02
9.35244203e-01 -5.91992140e-02 -1.96901545e-01 -1.10131335e+00
1.35146547e-02 -4.44720268e-01 3.61977935e-01 -9.87188220e-01
-8.92201126e-01 7.91227102e-01 -4.19560939e-01 -9.07004237e-01
-3.02942723e-01 -3.94214451e-01 -1.19612122e+00 1.21584964e+00
-1.29825425e+00 -1.00294900e+00 -1.90176025e-01 2.53069639e-01
8.71036291e-01 -1.81484401e-01 8.80295336e-01 -2.40854900e-02
-4.77067083e-01 5.31391501e-01 1.19020030e-01 1.83300272e-01
6.75522625e-01 -1.58518124e+00 8.74736130e-01 1.09065914e+00
-4.38680835e-02 1.38542271e+00 6.81963682e-01 -5.64140439e-01
-1.15883315e+00 -1.17645383e+00 1.36500621e+00 -6.46865308e-01
7.43772686e-01 -1.81500509e-01 -1.03542829e+00 1.19225621e+00
5.66583216e-01 3.34164873e-02 4.53202218e-01 1.26760975e-01
-5.64883530e-01 1.37342185e-01 -5.95268011e-01 5.24408162e-01
7.35295773e-01 -6.20951235e-01 -8.95669937e-01 1.77203566e-01
7.64834821e-01 -2.47649834e-01 -6.53966784e-01 -1.94876760e-01
5.66370070e-01 -8.96719933e-01 6.06360853e-01 -9.46050346e-01
9.48889077e-01 -2.76904732e-01 -1.71842352e-02 -1.16082728e+00
-7.09341466e-01 -7.27005959e-01 -3.43695818e-03 1.21292913e+00
5.78252554e-01 -3.69099468e-01 4.15139645e-01 4.86216247e-01
-6.58390880e-01 -5.59358418e-01 -4.94766921e-01 -5.07692516e-01
1.20013662e-01 -9.90153104e-02 6.69728875e-01 8.57156575e-01
3.57579410e-01 5.52184105e-01 7.78723732e-02 -1.51426643e-02
6.77805170e-02 5.21072030e-01 4.32763577e-01 -1.12409365e+00
-1.72428459e-01 -8.29338670e-01 -2.86964685e-01 -1.04794347e+00
1.94520861e-01 -1.17328417e+00 3.34360957e-01 -1.58880711e+00
5.16899467e-01 -2.05936775e-01 -4.08328265e-01 9.24122274e-01
-4.53702807e-01 -1.07437946e-01 -1.71956792e-02 2.26469547e-01
-7.16989875e-01 1.80738658e-01 6.92373931e-01 -3.40399563e-01
-1.89323038e-01 -5.07204756e-02 -8.53202641e-01 4.27269459e-01
6.09136224e-01 -6.38908684e-01 -1.28745615e-01 -8.21710110e-01
3.52940202e-01 -1.67312846e-03 1.60141006e-01 -8.53785634e-01
3.70472819e-01 -1.14235580e-01 4.51968431e-01 -5.89785993e-01
-2.06584156e-01 -4.55310285e-01 -1.85364231e-01 5.56184590e-01
-9.54741836e-01 3.70243281e-01 3.22578073e-01 4.23747867e-01
1.54678952e-02 -4.64241266e-01 5.07454157e-01 -3.93252999e-01
-6.45575523e-01 1.19707271e-01 -7.63688922e-01 -8.54038894e-02
5.94151556e-01 -1.45312726e-01 -4.62389678e-01 -3.07593904e-02
-6.26304030e-01 5.72636388e-02 6.00677431e-01 7.53061354e-01
4.27140057e-01 -1.28466165e+00 -6.45505667e-01 2.26064131e-01
2.43774995e-01 -2.09312841e-01 8.60846788e-02 6.61265254e-01
-4.88839239e-01 6.37006938e-01 -3.92482013e-01 -5.82014501e-01
-1.06386411e+00 9.32149112e-01 2.08257645e-01 -1.34154096e-01
-6.39659226e-01 8.20811152e-01 2.66589433e-01 -3.12447369e-01
7.91533515e-02 -9.53925371e-01 -1.52796075e-01 -3.76283042e-02
7.16488898e-01 1.66730583e-01 2.10315332e-01 -3.11690599e-01
-5.70555151e-01 5.04243672e-01 -2.56682485e-01 1.99506000e-01
1.65316808e+00 1.25577390e-01 -3.81145954e-01 6.17471814e-01
1.26444006e+00 -5.22132143e-02 -1.19235241e+00 -2.10108146e-01
3.39234859e-01 -2.69978732e-01 -1.63993329e-01 -5.02275407e-01
-9.63012874e-01 1.17016387e+00 6.00833036e-02 6.52208805e-01
8.43416631e-01 1.11723788e-01 4.54719037e-01 4.06831414e-01
-1.18666463e-01 -8.98228586e-01 4.63067949e-01 8.29207420e-01
1.06146932e+00 -8.97786796e-01 -2.17812568e-01 2.01704696e-01
-2.11505711e-01 1.66727448e+00 7.31988609e-01 -2.46507034e-01
2.64616460e-01 2.91385144e-01 -1.46633431e-01 -4.46608245e-01
-1.32688189e+00 -2.53737345e-02 2.11162701e-01 4.64585513e-01
9.08127666e-01 -3.12777668e-01 1.53545374e-02 4.90814358e-01
-2.80855643e-03 -1.47734001e-01 8.47421885e-01 1.18739355e+00
-6.36544645e-01 -7.84727633e-01 -1.89906254e-01 5.69219589e-01
-8.07091713e-01 -3.92081797e-01 -2.66219050e-01 5.58260083e-01
-8.00876543e-02 4.31456625e-01 5.31782448e-01 -1.76013082e-01
1.57848626e-01 1.95914894e-01 3.19776118e-01 -1.11602783e+00
-1.18623388e+00 -1.79674760e-01 -2.34506771e-01 -6.04117453e-01
-9.71288141e-03 -7.25647449e-01 -1.28164732e+00 -1.04859628e-01
-2.05893457e-01 5.50482050e-02 6.61558807e-01 6.51525438e-01
7.05718517e-01 1.09856606e+00 5.20327389e-01 -9.87395644e-01
-4.57832456e-01 -1.03795111e+00 4.08838429e-02 2.65718728e-01
7.90972710e-01 -1.65223613e-01 -1.69213310e-01 3.25995773e-01]
|
[7.638814449310303, 7.9031572341918945]
|
c9e7b881-45df-422c-9438-f1560884349a
|
modeling-human-cognition-with-a-hybrid-deep
|
2301.06216
| null |
https://arxiv.org/abs/2301.06216v2
|
https://arxiv.org/pdf/2301.06216v2.pdf
|
Modeling Human Cognition with a Hybrid Deep Reinforcement Learning Agent
|
Human cognition model could help us gain insights in how human cognition behaviors work under external stimuli, pave the way for synthetic data generation, and assist in adaptive intervention design for cognition regulation. When the external stimuli is highly dynamic, it becomes hard to model the effect that how the stimuli influences human cognition behaviors. Here we propose a novel hybrid deep reinforcement learning (HDRL) framework integrating drift-diffusion model to simulate the effect of dynamic time pressure on human cognition performance. We start with a N=50 user study to investigate how different factors may affect human performance, which help us gain prior knowledge in framework design. The evaluation demonstrates that this framework could improve human cognition modeling quantitatively and capture the general trend of human cognition behaviors qualitatively. Our framework could also be extended to explore and simulate how different external factors play a role in human behaviors.
|
['Xinyu Zhang', 'Songlin Xu']
|
2023-01-15
| null | null | null | null |
['synthetic-data-generation', 'synthetic-data-generation']
|
['medical', 'miscellaneous']
|
[-3.89159620e-01 -2.31018275e-01 5.52096367e-02 2.07751766e-02
1.82109386e-01 -4.64529574e-01 3.48322451e-01 2.84210026e-01
-4.88875717e-01 5.18721163e-01 3.85295421e-01 -3.77759248e-01
-2.22249985e-01 -9.03914869e-01 -5.19930959e-01 -3.55093628e-01
-9.82188582e-02 1.14278615e-01 2.12061614e-01 -4.86825138e-01
2.05278575e-01 1.41192623e-03 -1.70537424e+00 -3.30138542e-02
1.02830863e+00 2.99838424e-01 4.75887746e-01 6.33872628e-01
3.43353719e-01 7.20500827e-01 -8.72308731e-01 -6.00327663e-02
1.94020674e-01 -4.41052616e-01 -4.51778859e-01 1.17957946e-02
-2.71524876e-01 -4.91336972e-01 -4.10802603e-01 8.61712515e-01
7.50672162e-01 3.79264683e-01 6.00703537e-01 -9.75024283e-01
-1.29115391e+00 4.65706974e-01 -2.06871435e-01 3.13571423e-01
5.94831586e-01 6.32016063e-01 2.77330846e-01 -1.90604164e-03
1.80569783e-01 1.61582088e+00 3.86165708e-01 6.91888332e-01
-9.34157133e-01 -5.69308221e-01 4.45021123e-01 5.01217008e-01
-7.71055102e-01 2.76746213e-01 6.74053967e-01 -5.88600874e-01
3.90076399e-01 7.42658079e-02 1.18397164e+00 1.47230625e+00
4.56639469e-01 1.06084168e+00 1.44185519e+00 -4.23274815e-01
5.02715051e-01 1.33021653e-01 7.78653920e-01 3.86807889e-01
1.89334095e-01 5.49317241e-01 -4.02029902e-01 -2.16751546e-02
8.75926137e-01 2.18274504e-01 -1.23759046e-01 1.65381774e-01
-7.30740368e-01 5.41419148e-01 6.27522588e-01 1.28543451e-01
-6.03143454e-01 3.01923156e-01 1.75872207e-01 4.08729196e-01
6.33250326e-02 6.07927024e-01 -4.36931610e-01 -5.22313893e-01
-2.30853215e-01 7.75494874e-01 3.30248564e-01 6.14277184e-01
3.31665277e-01 1.17899373e-01 -8.65362287e-01 7.69219875e-01
9.91071761e-02 7.41759002e-01 8.52585495e-01 -1.16089392e+00
-1.37480870e-01 6.71324730e-01 6.12839043e-01 -8.91585588e-01
-8.09653342e-01 -4.27674919e-01 -3.04786772e-01 2.92723998e-02
3.99497896e-01 -5.62169313e-01 -8.08186769e-01 1.82984638e+00
7.33430386e-02 6.33779690e-02 -2.55884945e-01 1.14318407e+00
4.37864065e-01 2.98110992e-01 3.06349725e-01 -1.97365239e-01
1.47127807e+00 -5.27829051e-01 -8.33138943e-01 -1.39009148e-01
5.79149783e-01 -4.15434420e-01 1.77542734e+00 2.43981570e-01
-1.11728048e+00 -1.05554092e+00 -7.16197491e-01 4.04244572e-01
-3.83015454e-01 -1.21976942e-01 8.74311149e-01 1.25955021e+00
-9.29397762e-01 3.31897736e-01 -1.06804383e+00 -2.73736149e-01
4.01742756e-01 2.42498845e-01 7.76419342e-01 2.00248472e-02
-1.73527420e+00 7.35592604e-01 4.28079814e-02 -1.21269464e-01
-1.04127669e+00 -9.00895596e-01 -2.51056254e-01 1.89909577e-01
4.22782838e-01 -9.64394331e-01 1.40523148e+00 -1.01586890e+00
-1.51736295e+00 -1.51066650e-02 1.38285518e-01 -4.73692447e-01
5.11919439e-01 -4.26361889e-01 -3.58190000e-01 -4.47318405e-01
-2.40742728e-01 5.12755096e-01 3.92076135e-01 -1.13402152e+00
-1.94955900e-01 -4.91642267e-01 3.52788121e-01 3.01335812e-01
-5.47452390e-01 -2.68858105e-01 -1.83996707e-01 -6.34062648e-01
-6.36575818e-01 -1.22749460e+00 -2.95900822e-01 -5.35951495e-01
1.35594578e-02 -3.04418236e-01 5.21941364e-01 -4.70727146e-01
1.37068260e+00 -1.86569571e+00 -1.86585754e-01 2.09439829e-01
1.07048973e-01 3.84384334e-01 -9.81941447e-02 5.03105819e-01
4.22254384e-01 7.97214508e-02 3.10994923e-01 1.72111169e-01
1.77164584e-01 -1.55651849e-02 -1.05764821e-01 -1.09973580e-01
-2.30978832e-01 1.16241217e+00 -1.00587440e+00 1.97949231e-01
-1.25047937e-01 2.54207581e-01 -9.00832772e-01 3.26449424e-01
-3.82229209e-01 4.76715595e-01 -8.52952302e-01 3.46274465e-01
4.85354692e-01 -1.70313105e-01 -9.75109171e-03 4.41006988e-01
6.38989266e-03 -1.07098594e-01 -8.60001504e-01 1.13668168e+00
-3.36805135e-01 5.23540974e-01 -3.84984255e-01 -3.42797339e-01
1.02928996e+00 -6.95187598e-02 2.44547710e-01 -1.48018587e+00
2.91511506e-01 -5.51721215e-01 6.85754120e-01 -7.47565329e-01
5.36493123e-01 -5.64217046e-02 2.34022941e-02 6.41015887e-01
-6.04116917e-01 8.95422846e-02 3.06633990e-02 7.14417621e-02
1.34515429e+00 -4.40974310e-02 -2.39966691e-01 -4.24946547e-01
9.11075845e-02 -1.46004722e-01 4.61488008e-01 1.02376735e+00
-6.12982154e-01 -1.53132111e-01 5.33734441e-01 -3.91155481e-01
-7.99262464e-01 -9.98540759e-01 1.89621419e-01 1.19010675e+00
2.55930334e-01 -1.03181772e-01 -1.04105520e+00 -7.90470913e-02
1.02532111e-01 9.75665689e-01 -7.98200727e-01 -9.87366974e-01
-2.84158904e-02 -1.02355886e+00 3.99190396e-01 7.04807818e-01
8.35739493e-01 -1.37877715e+00 -1.08538020e+00 -4.34899982e-03
-1.91052958e-01 -5.78409851e-01 -3.52083325e-01 -2.37250328e-01
-6.72956586e-01 -8.85880947e-01 -7.25065351e-01 -1.75866738e-01
2.87518233e-01 3.15937132e-01 7.87627280e-01 2.98802167e-01
-1.75149590e-01 9.58834529e-01 -3.53816450e-01 -8.93125415e-01
-3.10555458e-01 -1.31130457e-01 3.94402772e-01 -4.01477635e-01
6.39945328e-01 -2.23477155e-01 -1.22644460e+00 4.61737812e-01
-1.02761173e+00 -9.54158679e-02 1.67506814e-01 5.02333581e-01
2.78791338e-02 3.16256046e-01 9.53810930e-01 -9.17144775e-01
1.81937122e+00 -5.53916097e-01 -1.54693589e-01 4.40756649e-01
-9.79797363e-01 -1.09321095e-01 6.47961617e-01 -7.60788739e-01
-1.37889278e+00 -6.02840543e-01 2.69956261e-01 -3.17136616e-01
-2.27171645e-01 3.39529991e-01 1.74588338e-01 5.37900090e-01
7.04172254e-01 8.61318633e-02 -1.17907047e-01 -1.79986849e-01
3.81039679e-01 6.50002182e-01 5.50618172e-02 -1.10354328e+00
2.09780246e-01 4.33127284e-02 -3.65688860e-01 -5.58437169e-01
-5.04311800e-01 -4.29341756e-02 -2.06845075e-01 -5.85999966e-01
8.24216723e-01 -9.49570954e-01 -1.38356841e+00 4.55268264e-01
-7.09170282e-01 -1.04482746e+00 5.98268956e-02 2.80874491e-01
-7.43853986e-01 -1.44515738e-01 -8.13823879e-01 -1.05796218e+00
-1.23291565e-02 -1.22037673e+00 6.16136551e-01 7.69664288e-01
-5.31930566e-01 -1.17328703e+00 1.80499479e-01 5.50460219e-01
5.76063454e-01 -2.70833701e-01 1.07447398e+00 -2.80745655e-01
-6.30331516e-01 1.39232814e-01 5.13059758e-02 4.23961915e-02
6.63238987e-02 -2.89490938e-01 -7.01176107e-01 -3.12995493e-01
-1.14570390e-02 -4.05840278e-01 4.04276699e-01 6.50522411e-01
1.30508840e+00 9.17316787e-03 -7.35319108e-02 -7.52624199e-02
9.40326035e-01 6.80113018e-01 8.29564035e-01 4.37078297e-01
5.05719721e-01 6.81560338e-01 7.66616642e-01 5.66334963e-01
7.66233563e-01 5.49990714e-01 -4.78540845e-02 -1.01393983e-02
1.20111771e-01 -3.84647280e-01 4.22923505e-01 5.30751944e-01
-3.57741028e-01 -4.32822168e-01 -1.04743457e+00 2.44827360e-01
-2.07616377e+00 -9.96379375e-01 -1.64066628e-01 2.10536218e+00
5.76427519e-01 2.04610705e-01 5.02530932e-01 -1.55862853e-01
4.45115268e-01 -3.85491312e-01 -8.47743452e-01 -5.10407150e-01
1.79415196e-01 -1.06590211e-01 3.18572998e-01 2.20631436e-01
-1.95171401e-01 9.57283378e-01 6.81517267e+00 4.52414691e-01
-9.29221749e-01 4.57268655e-02 8.58114898e-01 -1.87904269e-01
-4.40578908e-01 -2.47420266e-01 -6.45749331e-01 6.15957141e-01
1.30817688e+00 -4.13664103e-01 5.29542029e-01 3.96110803e-01
1.08370185e+00 -2.23878041e-01 -6.99205518e-01 6.11351132e-01
-3.92787397e-01 -6.22859597e-01 -9.91059467e-02 3.67019400e-02
7.61408567e-01 -5.01930535e-01 4.65536356e-01 9.06982958e-01
7.37244725e-01 -7.24757135e-01 4.97468263e-01 1.16877306e+00
7.04280883e-02 -7.94911981e-01 3.51944506e-01 6.46338701e-01
-6.84246540e-01 -5.86701512e-01 -5.24951875e-01 -7.12542474e-01
-7.44582713e-02 1.43559799e-01 -7.25668788e-01 -6.45134822e-02
5.30474186e-01 3.37590933e-01 -9.30720389e-01 7.32608914e-01
4.96562272e-02 9.22191799e-01 2.32552454e-01 -5.07316470e-01
5.30056171e-02 -4.89988253e-02 8.87241885e-02 8.79767656e-01
3.48728269e-01 4.73445982e-01 2.48696208e-01 1.23323405e+00
2.89831370e-01 4.29294109e-02 -4.64349121e-01 -2.09738776e-01
5.03408730e-01 6.64556026e-01 -1.01251495e+00 -3.60743880e-01
-3.25623453e-01 6.92570090e-01 1.34813398e-01 8.79569113e-01
-9.86732304e-01 1.98993281e-01 7.42814898e-01 5.44248879e-01
-1.03339195e-01 -2.85923421e-01 -3.79500180e-01 -8.03780437e-01
-3.99720132e-01 -1.04369688e+00 1.22350693e-01 -1.04821146e+00
-1.05351079e+00 3.76756266e-02 7.04790503e-02 -8.44949424e-01
-4.18526307e-03 -4.28356320e-01 -7.04527736e-01 7.22339690e-01
-9.84228611e-01 -4.95677173e-01 -4.44118857e-01 7.09807456e-01
6.49428606e-01 -1.40553236e-01 3.32205713e-01 9.61632729e-02
-8.30465436e-01 5.43027401e-01 -1.16074696e-01 -3.14015627e-01
6.83071017e-01 -9.96771395e-01 1.08462289e-01 4.38041151e-01
-4.53836620e-01 1.15617275e+00 7.75881648e-01 -1.03766143e+00
-1.25520074e+00 -8.34847748e-01 -2.20055282e-01 -6.32655025e-01
4.51659620e-01 -1.62354976e-01 -9.77315247e-01 3.36538374e-01
2.70224631e-01 -7.33764648e-01 7.91870654e-01 4.17115539e-01
2.00155079e-01 2.79993773e-01 -7.29130864e-01 1.19466805e+00
1.20904791e+00 -2.68269420e-01 -3.37493449e-01 -6.03833757e-02
1.07626939e+00 -2.69728810e-01 -6.63216114e-01 7.95495883e-02
5.00178993e-01 -1.01093602e+00 9.65223849e-01 -6.71263218e-01
6.18030012e-01 5.82214668e-02 1.06238067e-01 -1.66344428e+00
-6.63863242e-01 -4.71303701e-01 -5.28074093e-02 9.26606476e-01
1.23080574e-01 -5.38350284e-01 6.08026326e-01 1.36230934e+00
4.82284687e-02 -6.18032277e-01 -2.46634688e-02 -7.17499077e-01
3.05433810e-01 -5.36133111e-01 1.04988348e+00 7.16519117e-01
9.16947946e-02 1.78160936e-01 -3.83587420e-01 -2.56311614e-03
1.14259720e-01 -1.40444919e-01 7.81078160e-01 -1.05124462e+00
-5.16712308e-01 -5.49399853e-01 -9.50631872e-02 -1.09484589e+00
1.05812870e-01 -3.13965887e-01 -2.18310535e-01 -1.32245207e+00
4.13483024e-01 -3.88804197e-01 -4.75027889e-01 2.14523718e-01
-7.26291299e-01 -6.19025946e-01 3.31065267e-01 1.68196820e-02
-6.05814993e-01 9.33481336e-01 1.83456290e+00 1.67429194e-01
-6.27515137e-01 2.39698768e-01 -1.01164174e+00 3.89550060e-01
7.88137972e-01 -7.64647275e-02 -1.03595006e+00 -1.40597746e-01
2.97862589e-01 3.55714679e-01 4.57193196e-01 -1.03118360e+00
1.27744943e-01 -5.52773118e-01 4.40201163e-01 -3.81115153e-02
-1.82982996e-01 -7.86161900e-01 -5.65029383e-02 6.78545177e-01
-5.16523302e-01 3.45669121e-01 6.38781846e-01 8.50468755e-01
3.52434963e-01 1.03711262e-01 3.16127390e-01 -1.74021468e-01
-4.85498071e-01 1.05646223e-01 -8.27336550e-01 -1.12881072e-01
9.29125369e-01 -1.37862131e-01 -3.85778248e-01 -7.50861645e-01
-6.59642875e-01 5.80816150e-01 4.52983409e-01 8.33645403e-01
3.40074003e-01 -1.33380723e+00 -1.98321342e-01 8.91260132e-02
-4.24784794e-02 -6.39999270e-01 6.47600472e-01 4.03388798e-01
-5.56705631e-02 4.39755201e-01 -4.31174874e-01 -2.69879222e-01
-8.46217215e-01 6.69383883e-01 4.26607281e-01 -9.79896709e-02
-2.33317032e-01 4.61852252e-01 4.23243463e-01 -2.78527588e-01
2.22613886e-01 -3.36407036e-01 -3.06967378e-01 -1.55687198e-01
6.38068855e-01 7.98308969e-01 -2.14021236e-01 1.71235263e-01
1.08177774e-01 1.66851223e-01 -9.89046544e-02 -5.74266240e-02
9.47376966e-01 -3.62812728e-01 5.81298530e-01 7.86544979e-01
3.74953270e-01 -3.19529533e-01 -1.56871831e+00 -2.64380383e-03
-3.16239834e-01 -4.34029073e-01 8.59263726e-03 -1.12354410e+00
-7.02455223e-01 7.95351684e-01 1.22319055e+00 3.49361032e-01
1.33320451e+00 -3.45126212e-01 7.18734503e-01 3.78494203e-01
4.47397858e-01 -1.39271021e+00 7.69824266e-01 3.86723667e-01
6.68678284e-01 -1.18817437e+00 -4.09565866e-01 2.10590109e-01
-8.84779274e-01 7.74286807e-01 1.14306736e+00 -1.30126014e-01
8.56579125e-01 1.46471888e-01 2.04945356e-01 -3.46974373e-01
-1.03422701e+00 -3.61876696e-01 1.21589050e-01 7.22690880e-01
8.54926944e-01 5.62326908e-01 -5.26396215e-01 9.57884908e-01
-3.73752892e-01 4.27611679e-01 7.40402520e-01 6.16821408e-01
-4.02257204e-01 -1.28670716e+00 -4.69467252e-01 5.03083289e-01
-5.45002148e-02 2.05357581e-01 -4.56572801e-01 7.24223733e-01
2.06519529e-01 9.22887444e-01 1.09424338e-01 -4.02163446e-01
8.08715641e-01 1.84427947e-01 4.10592854e-01 -5.00868142e-01
-6.92593515e-01 1.35143489e-01 -3.03191751e-01 -4.55781668e-01
-1.24014899e-01 -7.78373539e-01 -1.42348647e+00 -6.38177216e-01
3.12477015e-02 -2.68934965e-01 1.67582333e-01 8.34989190e-01
4.46488321e-01 1.24800670e+00 3.18929762e-01 -2.74676263e-01
-2.94455945e-01 -1.00614798e+00 -5.79972506e-01 6.41296148e-01
-7.68526420e-02 -1.00927210e+00 9.54001863e-03 1.47827889e-03]
|
[4.1854753494262695, 2.135960340499878]
|
6a28d737-16a4-4060-a7ab-b9fbf415c414
|
contextualized-knowledge-aware-attentive
|
2104.05216
| null |
https://arxiv.org/abs/2104.05216v1
|
https://arxiv.org/pdf/2104.05216v1.pdf
|
Contextualized Knowledge-aware Attentive Neural Network: Enhancing Answer Selection with Knowledge
|
Answer selection, which is involved in many natural language processing applications such as dialog systems and question answering (QA), is an important yet challenging task in practice, since conventional methods typically suffer from the issues of ignoring diverse real-world background knowledge. In this paper, we extensively investigate approaches to enhancing the answer selection model with external knowledge from knowledge graph (KG). First, we present a context-knowledge interaction learning framework, Knowledge-aware Neural Network (KNN), which learns the QA sentence representations by considering a tight interaction with the external knowledge from KG and the textual information. Then, we develop two kinds of knowledge-aware attention mechanism to summarize both the context-based and knowledge-based interactions between questions and answers. To handle the diversity and complexity of KG information, we further propose a Contextualized Knowledge-aware Attentive Neural Network (CKANN), which improves the knowledge representation learning with structure information via a customized Graph Convolutional Network (GCN) and comprehensively learns context-based and knowledge-based sentence representation via the multi-view knowledge-aware attention mechanism. We evaluate our method on four widely-used benchmark QA datasets, including WikiQA, TREC QA, InsuranceQA and Yahoo QA. Results verify the benefits of incorporating external knowledge from KG, and show the robust superiority and extensive applicability of our method.
|
['Ying Shen', 'Wai Lam', 'Min Yang', 'Yaliang Li', 'Yuexiang Xie', 'Yang Deng']
|
2021-04-12
| null | null | null | null |
['answer-selection']
|
['natural-language-processing']
|
[-4.82007191e-02 7.05226138e-02 -3.31796333e-02 -5.99134147e-01
-7.92709529e-01 -5.75527072e-01 2.59117872e-01 1.29812881e-01
-4.88570720e-01 6.86084270e-01 7.66127586e-01 -3.71441454e-01
-3.95959646e-01 -1.09287059e+00 -6.52082980e-01 -3.14452648e-01
2.77829677e-01 6.25398874e-01 5.53517520e-01 -7.78465748e-01
3.11185777e-01 -1.05965082e-02 -1.23569012e+00 6.45998538e-01
1.47817481e+00 9.80735302e-01 1.59106120e-01 5.74990749e-01
-5.78933060e-01 1.35309732e+00 -6.72134519e-01 -8.52339268e-01
-3.45505089e-01 -5.92277288e-01 -1.68441498e+00 -2.34840721e-01
2.83698827e-01 -1.75152183e-01 -4.40489769e-01 8.97013485e-01
6.60776258e-01 4.81031358e-01 3.26535344e-01 -8.52722645e-01
-1.24057639e+00 7.05467761e-01 -1.79189608e-01 3.53879601e-01
6.67604685e-01 3.07358980e-01 1.18390524e+00 -7.28072643e-01
7.07825065e-01 1.46993351e+00 3.89110744e-01 4.29711580e-01
-5.48092961e-01 -8.96061733e-02 4.52944368e-01 1.08882415e+00
-1.11141717e+00 -6.02255277e-02 8.76508474e-01 -2.06734046e-01
1.24137568e+00 1.79976791e-01 3.19133997e-01 7.97378302e-01
3.51077802e-02 1.08560431e+00 7.50274539e-01 -3.27865005e-01
-5.13176210e-02 -7.05741942e-02 6.97829366e-01 7.17956066e-01
-2.32399136e-01 -3.54001909e-01 -3.85244131e-01 -3.07163805e-01
2.16420397e-01 -2.40659937e-01 -7.81257749e-01 -3.06895494e-01
-1.10726976e+00 9.96090829e-01 6.84693098e-01 8.78725722e-02
-4.07387346e-01 1.51547408e-02 6.19783759e-01 6.04454041e-01
2.30291039e-01 4.54334944e-01 -8.36651623e-01 1.41438827e-01
-6.38137758e-03 7.10248947e-02 8.78533304e-01 1.04247844e+00
8.58274937e-01 -1.59677491e-01 -7.19180644e-01 8.77000034e-01
2.82723278e-01 5.93324423e-01 5.62095404e-01 -6.52605534e-01
9.99427259e-01 1.29669678e+00 -1.31986037e-01 -1.06028771e+00
-5.47753930e-01 -5.29558837e-01 -7.05649078e-01 -5.84206223e-01
1.62682399e-01 -2.83161968e-01 -8.24318051e-01 1.63531828e+00
6.61275566e-01 -2.65101731e-01 4.65967864e-01 9.37187791e-01
1.64708841e+00 4.36875850e-01 2.01442465e-01 -1.29019588e-01
1.63820434e+00 -1.35834599e+00 -9.83630002e-01 -2.22714901e-01
8.07377338e-01 -3.22159588e-01 1.36728323e+00 -7.36280307e-02
-8.75050545e-01 -5.62373161e-01 -6.55760705e-01 -4.84347403e-01
-5.53651750e-01 -8.69200826e-02 4.29312915e-01 3.77346307e-01
-8.48134637e-01 1.00433379e-01 -1.68019459e-01 -2.64004529e-01
3.56725156e-01 4.99900252e-01 -2.56141663e-01 -5.16820014e-01
-2.00818825e+00 1.11846316e+00 6.51767731e-01 4.34332967e-01
-6.10698879e-01 -4.55448478e-01 -1.05167186e+00 3.83631289e-01
1.05221450e+00 -1.23206937e+00 1.16213274e+00 -9.42506909e-01
-1.50556099e+00 4.01744664e-01 -3.24779838e-01 -1.28154024e-01
-3.96069214e-02 -2.34584242e-01 -5.34844279e-01 3.90132248e-01
-2.58531119e-03 5.63134030e-02 5.49791455e-01 -1.08018649e+00
-1.84624836e-01 -5.54490387e-01 7.62160838e-01 7.13810742e-01
-2.54071414e-01 -4.05250490e-03 -8.07038486e-01 -2.14918777e-01
-1.54160276e-01 -3.45238149e-01 -1.95659101e-01 -7.89027393e-01
-4.23847884e-01 -5.38294375e-01 5.30410826e-01 -8.74270022e-01
1.56046689e+00 -1.69642389e+00 4.80281591e-01 -1.18703786e-02
2.63927519e-01 4.79990333e-01 -2.67418474e-01 7.16569364e-01
2.15746880e-01 -9.20515358e-02 -2.19914347e-01 1.97578609e-01
-8.25819895e-02 3.34746301e-01 -1.74215570e-01 -1.94276676e-01
4.41430092e-01 1.54945588e+00 -1.24836552e+00 -5.70858300e-01
-2.11019576e-01 1.26122624e-01 -5.46275675e-01 2.70753473e-01
-6.17198348e-01 3.64373922e-01 -8.61319244e-01 6.77283108e-01
4.92519766e-01 -6.65912867e-01 1.63014293e-01 -5.99128723e-01
5.71850538e-01 3.54288220e-01 -9.20178294e-01 1.69898903e+00
-5.09282172e-01 -9.05771181e-03 6.67141378e-02 -9.17066276e-01
6.87718630e-01 2.82315195e-01 -2.62770385e-01 -1.00374246e+00
1.58623576e-01 -5.61982058e-02 8.91650170e-02 -9.97586668e-01
5.76798975e-01 -6.52742982e-02 -7.99488276e-02 4.24328566e-01
4.84813631e-01 -6.02577720e-03 7.97376633e-02 6.96344972e-01
1.14454472e+00 -9.27913189e-02 4.58266288e-01 -1.10791124e-01
1.14183092e+00 2.09560677e-01 4.30075020e-01 6.64914429e-01
-2.40637109e-01 3.06938231e-01 5.81891000e-01 -3.38975102e-01
-9.72330719e-02 -6.75636172e-01 3.35553408e-01 1.37416553e+00
2.22104684e-01 -4.43346083e-01 -5.91566861e-01 -1.36607480e+00
-9.39888358e-02 7.64528930e-01 -7.49377131e-01 -5.90902328e-01
-6.34303868e-01 -6.04111016e-01 3.04093152e-01 6.66958392e-01
8.47953618e-01 -1.39936233e+00 -6.56049773e-02 1.76369876e-01
-5.81840992e-01 -9.93151248e-01 -5.98122180e-01 -1.06714107e-01
-6.12212360e-01 -1.37862396e+00 -2.82720685e-01 -7.22342849e-01
5.60498655e-01 3.02253067e-01 1.63000607e+00 4.52700526e-01
2.35450134e-01 1.05910838e+00 -7.60166585e-01 -2.11607799e-01
5.83889382e-03 2.84929216e-01 -3.37467760e-01 1.99834347e-01
5.75622082e-01 -1.51860103e-01 -6.22977078e-01 3.09375525e-01
-9.45740879e-01 -2.08527878e-01 4.52702850e-01 1.07731879e+00
3.55592787e-01 -7.24114776e-02 1.18364453e+00 -1.33060157e+00
1.11730301e+00 -7.44047165e-01 -2.02500433e-01 9.49581385e-01
-3.38586688e-01 1.25579968e-01 7.03933179e-01 -6.39856327e-04
-1.59767735e+00 -2.17447788e-01 -4.31114823e-01 -5.41253947e-02
4.60507274e-02 1.20832717e+00 -6.84386671e-01 -1.74811080e-01
6.06294394e-01 2.80762404e-01 -3.29930544e-01 -1.38432339e-01
8.37706983e-01 5.70750892e-01 5.01030385e-01 -7.59264648e-01
5.00954866e-01 2.42868334e-01 -5.04199445e-01 -3.73266339e-01
-1.31093597e+00 -6.06442392e-01 -5.65716445e-01 -6.19036593e-02
9.36874807e-01 -6.23082757e-01 -9.68481421e-01 3.73802602e-01
-1.27491999e+00 -1.15143441e-01 -1.83739483e-01 2.33397931e-01
-3.76911402e-01 6.82670653e-01 -5.91758668e-01 -5.97265482e-01
-6.92350984e-01 -1.06176841e+00 6.68550491e-01 4.33608204e-01
2.97269434e-01 -1.39073086e+00 8.28546658e-03 1.07517648e+00
5.63654542e-01 -1.30315423e-01 1.36821210e+00 -1.07018721e+00
-7.06296206e-01 -7.23455660e-03 -3.87659699e-01 3.21046621e-01
1.43385530e-01 -5.81810117e-01 -9.00129676e-01 1.74969006e-02
1.00769550e-01 -8.50516737e-01 1.05181456e+00 -5.66034280e-02
1.14176404e+00 -2.30944350e-01 -3.99226546e-02 1.91058472e-01
1.31245935e+00 1.32093400e-01 5.74524164e-01 2.32607871e-01
9.75430012e-01 8.85194659e-01 4.71088588e-01 2.55254470e-02
1.09465015e+00 3.77163410e-01 5.74944675e-01 1.54959485e-01
-7.33558983e-02 -1.86748683e-01 4.54637520e-02 1.35485947e+00
5.17410785e-02 -4.49273497e-01 -9.40221310e-01 7.97303140e-01
-2.14093018e+00 -8.29131603e-01 -2.26181224e-01 1.82197118e+00
9.98162448e-01 -1.55260444e-01 -2.79355764e-01 -3.62469494e-01
5.55874944e-01 -3.04432027e-03 -8.11376691e-01 -2.39004582e-01
-2.84804463e-01 6.06530905e-02 -3.49204615e-02 6.74349070e-01
-1.00740099e+00 1.04542816e+00 5.33312130e+00 9.98661578e-01
-5.18779874e-01 1.25035286e-01 4.33265865e-01 3.64206940e-01
-6.96804345e-01 -1.28137559e-01 -7.29534268e-01 2.03826874e-01
8.16895664e-01 -2.36922622e-01 1.49192899e-01 6.49619401e-01
-3.22461009e-01 -1.68299794e-01 -7.65254557e-01 5.08337438e-01
3.96621704e-01 -1.47395778e+00 4.57916260e-01 -4.75574940e-01
5.84118903e-01 -2.82221287e-02 -2.19520599e-01 9.51738000e-01
4.86339808e-01 -8.87263656e-01 -6.01619296e-03 7.00338244e-01
3.24665040e-01 -7.91499376e-01 1.31430483e+00 3.74367207e-01
-1.27534986e+00 -1.79676116e-01 -3.80665541e-01 6.19537830e-02
3.46638054e-01 2.59886801e-01 -4.77647662e-01 1.49077404e+00
6.97970033e-01 5.01330853e-01 -7.29476213e-01 8.31937313e-01
-6.96270466e-01 6.97829962e-01 1.16611965e-01 -1.16828337e-01
3.08485985e-01 -1.95460524e-02 6.50047287e-02 1.02942324e+00
-3.06813717e-01 6.93245351e-01 1.68452635e-01 6.46166623e-01
-3.52324307e-01 3.49273503e-01 -3.53027791e-01 7.75568038e-02
2.89971024e-01 1.14166355e+00 -3.92959416e-01 -5.53501248e-01
-8.12376499e-01 9.05588269e-01 8.98929596e-01 5.78924835e-01
-4.73531008e-01 -6.72599077e-01 1.13359451e-01 -4.90810156e-01
4.17900413e-01 9.10413265e-02 1.50509849e-01 -1.43982363e+00
1.93168774e-01 -1.00161278e+00 1.08913517e+00 -8.06816101e-01
-1.69826734e+00 8.27683032e-01 -9.91418436e-02 -7.68772125e-01
4.83053140e-02 -6.06713057e-01 -7.51155257e-01 1.03499413e+00
-2.01969242e+00 -1.22365212e+00 -3.50378990e-01 9.65638340e-01
3.73017222e-01 -5.12157679e-02 6.95839882e-01 3.35454106e-01
-5.30419528e-01 5.94056368e-01 -1.42193198e-01 3.41298670e-01
6.41984463e-01 -1.30406809e+00 1.63385168e-01 4.99729276e-01
-1.26885790e-02 9.06359136e-01 5.53430170e-02 -7.64659166e-01
-1.75650740e+00 -1.10801542e+00 1.04530180e+00 -7.81890392e-01
6.39126599e-01 -5.31549193e-03 -1.38686728e+00 7.75062323e-01
6.00649118e-01 -5.97969368e-02 9.74911273e-01 4.64915067e-01
-3.64376366e-01 8.59919190e-02 -9.28279757e-01 3.87540251e-01
8.91171098e-01 -7.20998168e-01 -1.15300143e+00 5.15872061e-01
1.41364384e+00 -5.15160263e-01 -9.09668863e-01 6.58523083e-01
-5.35909645e-02 -7.69486248e-01 1.05397105e+00 -1.15113401e+00
2.41448596e-01 -4.47873652e-01 -7.31477514e-02 -1.39048481e+00
-4.57107604e-01 -1.34320170e-01 -4.69357759e-01 1.33347404e+00
5.54488420e-01 -5.82482517e-01 5.94906151e-01 5.52217841e-01
-4.29473191e-01 -1.13182628e+00 -7.79475689e-01 -3.91376734e-01
1.71075612e-02 -2.35695109e-01 7.88776278e-01 1.19405949e+00
2.05878139e-01 1.07637417e+00 -2.92362154e-01 2.67524898e-01
-5.68633154e-02 4.60098028e-01 5.15598953e-01 -8.76879692e-01
-3.15648556e-01 -1.90242276e-01 -1.40830129e-01 -1.24559689e+00
1.84749722e-01 -9.79694605e-01 -2.71330595e-01 -2.10909152e+00
2.53285229e-01 3.14518064e-02 -3.67025673e-01 3.32326174e-01
-9.84597802e-01 -3.98249000e-01 4.00670469e-02 -2.27660522e-01
-1.38720095e+00 1.07682347e+00 1.67182410e+00 -3.32327664e-01
1.04470074e-03 -2.13537455e-01 -9.14640665e-01 5.77465475e-01
5.28457105e-01 -9.53652561e-02 -8.58173251e-01 -8.73751342e-01
7.82733142e-01 1.28813162e-01 2.08472982e-01 -3.99822354e-01
6.68480396e-01 -2.04590917e-01 8.85626674e-02 -7.18311787e-01
1.87193334e-01 -6.64934635e-01 -6.54180110e-01 2.28193417e-01
-2.49558166e-01 8.72846767e-02 1.91465840e-01 8.58894587e-01
-6.05065227e-01 -3.06244165e-01 1.96816534e-01 -5.07264197e-01
-1.09045315e+00 3.86342287e-01 -4.22752500e-02 7.20256627e-01
6.07526243e-01 1.96548566e-01 -9.49131668e-01 -5.82622409e-01
-6.89507961e-01 1.22738898e+00 -2.29872480e-01 3.93104315e-01
9.18008327e-01 -1.36342406e+00 -6.55642331e-01 -5.64361587e-02
4.82549846e-01 1.50507942e-01 9.30503190e-01 7.93825269e-01
-3.80226523e-01 4.59039241e-01 1.31645590e-01 -2.65718788e-01
-9.55778658e-01 8.98742199e-01 4.51251268e-01 -6.21137977e-01
-1.54163092e-01 1.13191378e+00 3.01686496e-01 -1.11923349e+00
1.79257229e-01 -2.47040182e-01 -9.05114353e-01 8.32795128e-02
5.56919217e-01 2.13996455e-01 2.93052524e-01 -4.64881957e-01
-3.48487020e-01 5.15058875e-01 -3.19591761e-01 4.49558198e-01
9.12177205e-01 -2.72485554e-01 -3.62276286e-01 1.13548741e-01
8.23942423e-01 -1.89010337e-01 -5.85639358e-01 -7.51849890e-01
1.90987363e-01 -9.45102572e-02 -2.79125839e-01 -1.27188087e+00
-1.02972841e+00 1.03141177e+00 2.66625732e-02 1.37945171e-02
1.17493749e+00 1.86845049e-01 8.60274374e-01 9.71475482e-01
2.63363272e-01 -1.07817078e+00 3.35393280e-01 1.00419772e+00
1.14093113e+00 -1.39012706e+00 -1.48194730e-01 -5.48323214e-01
-9.72467780e-01 9.45778370e-01 1.26668596e+00 2.97447145e-01
4.98672724e-01 -5.82302094e-01 2.75648683e-01 -7.44448066e-01
-9.95078921e-01 -6.10459268e-01 6.25003815e-01 5.53964794e-01
2.64931917e-01 -2.79693216e-01 -2.25936025e-01 1.20230436e+00
3.04149389e-02 -2.57005006e-01 2.30651245e-01 1.00948453e+00
-4.70395327e-01 -1.01031935e+00 -1.06663071e-01 5.52343130e-01
-1.73019409e-01 -4.14717019e-01 -5.84729612e-01 5.57543039e-01
2.00805236e-02 1.23000669e+00 -5.64493954e-01 -4.34105039e-01
6.54869914e-01 2.91782409e-01 1.81909248e-01 -7.69126832e-01
-1.12150860e+00 -5.76119483e-01 4.59417224e-01 -4.60831285e-01
-5.02846062e-01 -4.78498973e-02 -1.26466775e+00 3.39287892e-02
-6.76270664e-01 4.11774307e-01 -1.66924987e-02 1.29445744e+00
7.04555690e-01 7.82964826e-01 2.86914319e-01 4.41032238e-02
-6.33864224e-01 -9.06678736e-01 -2.02327535e-01 5.89285493e-01
2.17484966e-01 -6.53321922e-01 -1.30855337e-01 -2.31168523e-01]
|
[10.69620418548584, 7.923274517059326]
|
2881f046-527d-45dc-aa74-959c1dfcba3b
|
a-span-based-dynamic-local-attention-model
| null | null |
https://aclanthology.org/2021.acl-short.26
|
https://aclanthology.org/2021.acl-short.26.pdf
|
A Span-based Dynamic Local Attention Model for Sequential Sentence Classification
|
Sequential sentence classification aims to classify each sentence in the document based on the context in which sentences appear. Most existing work addresses this problem using a hierarchical sequence labeling network. However, they ignore considering the latent segment structure of the document, in which contiguous sentences often have coherent semantics. In this paper, we proposed a span-based dynamic local attention model that could explicitly capture the structural information by the proposed supervised dynamic local attention. We further introduce an auxiliary task called span-based classification to explore the span-level representations. Extensive experiments show that our model achieves better or competitive performance against state-of-the-art baselines on two benchmark datasets.
|
['Zipeng Chen', 'Jiangyue Yan', 'Zhenxi Lin', 'Qianli Ma', 'Xichen Shang']
|
2021-08-01
| null | null | null |
acl-2021-5
|
['sentence-classification']
|
['natural-language-processing']
|
[ 2.71594673e-01 -2.17528313e-01 -5.96875250e-01 -8.35120618e-01
-6.79452240e-01 -4.66357678e-01 4.23112780e-01 6.01629734e-01
-4.03089732e-01 6.38246775e-01 8.24633837e-01 -4.19186890e-01
2.66624570e-01 -5.55082977e-01 -5.91705322e-01 -5.58635533e-01
-5.84935918e-02 -1.02647841e-01 4.23020005e-01 -1.83085412e-01
6.33342922e-01 1.71270043e-01 -1.13485730e+00 8.58856440e-01
7.58546770e-01 9.10616100e-01 4.34973627e-01 6.12780988e-01
-5.78140497e-01 1.25609052e+00 -7.83795297e-01 -4.01533544e-02
-2.05274969e-01 -7.17299700e-01 -1.20960796e+00 1.95625618e-01
5.36491096e-01 -1.76200747e-01 -3.95910531e-01 9.94531989e-01
1.99908659e-01 4.69980121e-01 4.36493099e-01 -7.21660197e-01
-9.81320560e-01 9.74338830e-01 -6.65623069e-01 7.85281956e-01
5.27882457e-01 -5.86624146e-02 1.39786422e+00 -6.42138302e-01
4.47609812e-01 1.34984016e+00 4.55357313e-01 3.02794188e-01
-9.77615476e-01 -2.72147328e-01 1.04851854e+00 5.37620783e-01
-1.03570807e+00 -2.35169858e-01 1.06653082e+00 -3.37704688e-01
1.28357744e+00 1.76695928e-01 4.49610353e-01 1.07398772e+00
5.97598970e-01 1.10673749e+00 9.33718204e-01 -5.90889812e-01
5.31544946e-02 -3.74643207e-01 1.29445171e+00 5.51552296e-01
-2.22471520e-01 -4.98555779e-01 -4.36305672e-01 -5.46368808e-02
1.15754977e-01 2.38302663e-01 -2.58280993e-01 9.81414914e-02
-8.68351579e-01 8.58739793e-01 5.82880616e-01 7.22303808e-01
-2.99268752e-01 1.13386080e-01 1.02184987e+00 2.93494195e-01
8.62056315e-01 1.30852968e-01 -5.63837528e-01 -3.53808738e-02
-9.90027726e-01 3.49380560e-02 4.34796870e-01 8.36839020e-01
4.53010142e-01 -1.79027051e-01 -7.33278692e-01 7.13407934e-01
3.33517760e-01 -2.03487918e-01 8.86501968e-01 -4.71539676e-01
9.31056917e-01 6.04894519e-01 -1.74082533e-01 -1.05099118e+00
-4.40812111e-01 -7.07616627e-01 -7.96098053e-01 -5.74656725e-01
-1.52940243e-01 9.18314829e-02 -7.78777242e-01 1.66088521e+00
-1.07774464e-02 2.81541318e-01 1.38231739e-02 7.33178914e-01
8.80126953e-01 9.51720417e-01 2.49196589e-01 -6.24068201e-01
1.30010259e+00 -1.67900932e+00 -9.69164073e-01 -2.92559266e-01
6.90885365e-01 -4.74760115e-01 1.33547974e+00 -3.92747559e-02
-9.41428125e-01 -8.15419674e-01 -1.19445634e+00 -2.55639136e-01
-1.76980555e-01 -4.15783674e-02 4.97006595e-01 3.44593763e-01
-8.62528324e-01 4.34769392e-01 -8.39789450e-01 -3.60292554e-01
2.83627421e-01 6.49505015e-03 1.54215489e-02 8.03015903e-02
-1.47114611e+00 6.20583773e-01 5.40896893e-01 3.20362180e-01
-7.08920538e-01 -3.10522109e-01 -1.00374067e+00 4.08905327e-01
2.18980923e-01 -4.66439337e-01 1.37386572e+00 -9.81785774e-01
-1.35192192e+00 7.85468757e-01 -8.79813135e-01 -6.90698802e-01
-5.57826012e-02 -3.31269026e-01 -2.16771588e-01 2.65325487e-01
2.04696357e-01 2.72753298e-01 4.69365746e-01 -8.73179436e-01
-3.12933058e-01 -4.11260933e-01 3.76009852e-01 3.90262693e-01
-5.84480703e-01 3.10175747e-01 -3.76164168e-01 -8.74035835e-01
2.65482012e-02 -5.07596195e-01 -4.69115764e-01 -5.05981326e-01
-4.30600405e-01 -7.68726110e-01 8.32364261e-01 -5.83907723e-01
1.81044030e+00 -2.04510689e+00 8.90775025e-02 -4.87196028e-01
-9.66101885e-02 -5.59851248e-03 -2.23364636e-01 8.49267304e-01
-1.29840132e-02 3.51854891e-01 -1.78022176e-01 -5.49462140e-01
-1.57246292e-01 -3.52027863e-02 -6.81161642e-01 3.62185657e-01
-2.10967407e-01 1.09685755e+00 -8.79416764e-01 -6.69679761e-01
-1.42600343e-01 -4.48969379e-02 -1.99193060e-01 2.13035658e-01
-3.54424000e-01 3.12275440e-01 -5.06735086e-01 2.37494916e-01
4.32992786e-01 -4.77470249e-01 2.44252086e-01 -6.53237775e-02
4.15495783e-02 7.77510464e-01 -4.70823467e-01 2.06608534e+00
-3.79330158e-01 6.09739482e-01 -2.92548746e-01 -1.19203174e+00
6.94669187e-01 2.28027999e-01 1.80518940e-01 -4.90452379e-01
5.98887168e-02 -3.07732075e-01 -2.87772566e-02 -5.13570905e-01
6.75276220e-01 -8.73442665e-02 -3.91242206e-01 5.62757194e-01
-6.23341985e-02 5.06801903e-01 3.48634183e-01 4.44532365e-01
9.61619139e-01 -1.17474742e-01 4.95676935e-01 -4.59369451e-01
1.02761471e+00 -3.02614033e-01 5.51042616e-01 8.99974823e-01
-3.22349638e-01 4.62720126e-01 7.12099016e-01 -5.66250920e-01
-6.78552270e-01 -6.13381267e-01 -1.70956090e-01 1.42653763e+00
3.26140016e-01 -5.90444148e-01 -7.46032834e-01 -1.05256259e+00
-4.11534727e-01 7.16640234e-01 -8.11196744e-01 -1.64017335e-01
-6.42789543e-01 -5.68124235e-01 3.33687574e-01 8.08250606e-01
6.44441903e-01 -1.34295416e+00 -2.94858307e-01 3.58180821e-01
-3.24981034e-01 -9.95492935e-01 -1.06660366e+00 7.93026313e-02
-9.90883291e-01 -7.99361527e-01 -4.85999107e-01 -1.16813743e+00
5.90182304e-01 4.04349864e-01 1.00645697e+00 9.53645334e-02
2.11942017e-01 6.85278676e-04 -7.55560279e-01 1.32243723e-01
4.87975143e-02 5.70990741e-01 -2.43435517e-01 9.30242017e-02
4.26856011e-01 -3.01969230e-01 -4.91009533e-01 -6.02076873e-02
-8.22600365e-01 -8.64940416e-03 2.71448106e-01 9.93976951e-01
4.17881966e-01 -2.97434423e-02 8.45854461e-01 -1.03252006e+00
1.00573492e+00 -5.94380856e-01 -3.28616463e-02 5.14107168e-01
-2.97414869e-01 1.20816998e-01 8.75257313e-01 -2.04450950e-01
-1.04971170e+00 -1.53148994e-01 -3.70864987e-01 -1.38548285e-01
-2.29076594e-01 1.01204681e+00 -1.53594792e-01 5.41923106e-01
1.61213011e-01 7.46121526e-01 -3.89603764e-01 -5.88455498e-01
2.18890816e-01 9.35395241e-01 1.44530699e-01 -5.12175381e-01
8.02077726e-02 1.31573036e-01 -2.29236096e-01 -6.03438973e-01
-1.63056982e+00 -5.44579446e-01 -8.81097734e-01 7.59663284e-02
9.08842146e-01 -7.73949206e-01 -5.49340963e-01 4.10798639e-01
-1.47174859e+00 -2.08663508e-01 2.10791882e-02 1.38092101e-01
-2.48337775e-01 8.54306638e-01 -1.03419340e+00 -7.50343382e-01
-5.43225765e-01 -1.06247330e+00 1.07194972e+00 3.55545096e-02
-1.90345570e-01 -1.20643723e+00 1.28298700e-01 2.85830557e-01
1.90495431e-01 -6.88641146e-02 8.95535350e-01 -7.83458829e-01
-1.94968164e-01 -1.02935448e-01 -6.08093143e-02 2.97035158e-01
2.39764526e-01 -3.43321741e-01 -6.93551362e-01 -4.60274607e-01
3.24378729e-01 -4.05366451e-01 1.32493770e+00 3.32912058e-01
1.63454139e+00 -5.30782938e-01 -3.42658252e-01 2.16632798e-01
1.22118986e+00 3.37151080e-01 3.90639037e-01 2.64886677e-01
6.99986815e-01 6.62693620e-01 6.52507961e-01 1.23090528e-01
3.49063426e-01 6.78902328e-01 7.70545080e-02 3.62984210e-01
6.77906945e-02 -2.60635614e-01 4.92480725e-01 1.17754221e+00
5.63720942e-01 -6.44209266e-01 -7.73497045e-01 5.31558692e-01
-2.16810489e+00 -1.20212328e+00 -1.41188353e-01 1.59124708e+00
8.38014483e-01 6.11704051e-01 -8.02966952e-02 -6.69290721e-02
9.01424408e-01 8.42875779e-01 -3.88599902e-01 -7.61532664e-01
-1.33461505e-01 -1.73246071e-01 3.50399613e-02 7.43281841e-01
-1.40357828e+00 9.72641587e-01 6.88894367e+00 9.41534162e-01
-9.75259483e-01 2.08860561e-01 8.70011866e-01 -7.35786781e-02
-3.57654691e-01 -1.19919650e-01 -1.08249152e+00 8.49936604e-01
1.11979830e+00 -4.00041252e-01 -2.03149319e-01 8.24768007e-01
4.31891978e-01 1.00345746e-01 -9.51659083e-01 4.37962502e-01
5.14109373e-01 -1.40199935e+00 2.20563635e-01 -2.40747169e-01
7.87910163e-01 -1.31203920e-01 -1.40065834e-01 4.78510261e-01
-2.54712868e-02 -7.88253784e-01 6.39476955e-01 6.74702108e-01
4.45263535e-01 -7.52682209e-01 9.28768694e-01 5.84756970e-01
-1.54084766e+00 -1.18042812e-01 -3.19077462e-01 -5.22428811e-01
3.36921841e-01 4.13544446e-01 -2.69875735e-01 6.71723425e-01
4.97638971e-01 1.17388606e+00 -8.01695049e-01 6.56890154e-01
-3.66265446e-01 1.01680005e+00 1.98230833e-01 -4.88822341e-01
6.52849734e-01 9.43770781e-02 2.79668719e-01 1.51945949e+00
-2.18888730e-01 8.60813931e-02 6.94169581e-01 3.89849007e-01
-2.00380370e-01 3.45782101e-01 -5.09621561e-01 9.58971381e-02
3.28762829e-01 8.22900295e-01 -8.40675473e-01 -5.27754784e-01
-6.05384648e-01 1.19489598e+00 6.48522913e-01 3.34214061e-01
-7.02480316e-01 -5.36673725e-01 3.64037722e-01 -2.46681213e-01
5.47271371e-01 -3.44720751e-01 -3.23342055e-01 -1.40445018e+00
3.21748048e-01 -5.43573380e-01 4.73143518e-01 -4.82811272e-01
-1.63802660e+00 8.65746975e-01 -1.43474549e-01 -1.09652722e+00
-1.20006353e-01 -2.34452203e-01 -9.75406468e-01 8.24622452e-01
-1.32117164e+00 -1.18208945e+00 5.49163930e-02 2.19465196e-01
1.34041500e+00 2.36748345e-02 7.04963863e-01 -1.22185074e-01
-6.84384525e-01 6.54608846e-01 2.35188469e-01 3.00626963e-01
3.47204506e-01 -1.13778448e+00 5.01028061e-01 9.79832292e-01
1.61093310e-01 1.05524898e+00 3.93096328e-01 -5.81031740e-01
-7.64860988e-01 -1.00586784e+00 1.29766667e+00 -1.74654722e-01
7.90524781e-01 -4.71843064e-01 -1.01335180e+00 8.96695495e-01
8.05258930e-01 -3.70387509e-02 8.42845500e-01 3.34945858e-01
-3.22068661e-01 -1.66913196e-01 -6.70911908e-01 3.80014330e-01
1.12864304e+00 -7.11430788e-01 -1.04466295e+00 4.13469166e-01
1.40175116e+00 -1.96782753e-01 -4.36256021e-01 2.45525852e-01
2.58378923e-01 -6.31062150e-01 5.31175971e-01 -9.48790312e-01
7.65203595e-01 -1.34223759e-01 -7.53812566e-02 -1.10614967e+00
-6.78259373e-01 -2.32488409e-01 -4.31327522e-01 1.42434192e+00
2.76284486e-01 -3.35089356e-01 6.65195167e-01 1.70526281e-01
-4.73047793e-01 -1.25757301e+00 -8.65221322e-01 -8.17852199e-01
2.38626719e-01 -2.23893985e-01 4.40355152e-01 8.49757552e-01
2.49031186e-01 9.61477995e-01 -3.56607527e-01 -1.39127597e-02
2.42729947e-01 6.12952650e-01 1.32054865e-01 -7.83322215e-01
-3.24389219e-01 -6.28909528e-01 -4.55436885e-01 -1.75228333e+00
8.18509042e-01 -1.02998042e+00 1.72389328e-01 -1.75874197e+00
6.08231306e-01 2.06891730e-01 -7.19429791e-01 3.32041800e-01
-6.71304047e-01 -1.74410731e-01 8.86911899e-02 3.29595208e-01
-1.34631002e+00 9.21449602e-01 1.07479036e+00 -4.66823488e-01
-4.04468039e-03 -1.53129265e-01 -7.26433814e-01 4.40002143e-01
8.59058082e-01 -4.33488011e-01 -4.66749817e-01 -7.10500658e-01
-1.28390402e-01 4.50152233e-02 -1.25548527e-01 -8.60002816e-01
4.58858222e-01 -2.22266778e-01 1.05238333e-01 -1.01013434e+00
-2.23991442e-02 -5.28468966e-01 -5.94744742e-01 4.74430382e-01
-1.15746367e+00 1.41786799e-01 3.22320983e-02 8.39170635e-01
-5.72787225e-01 -5.71384966e-01 4.70302403e-01 -1.35099947e-01
-7.15178370e-01 2.29072630e-01 -6.43871427e-01 -2.65827999e-02
1.14899790e+00 1.92030117e-01 -2.64425993e-01 -2.21267149e-01
-7.96801388e-01 5.00493526e-01 1.00476019e-01 6.00348353e-01
5.88243425e-01 -1.35789013e+00 -6.10904515e-01 -1.42089546e-01
7.99717233e-02 -2.08136961e-01 4.86120492e-01 4.71037745e-01
-3.46376389e-01 8.41684163e-01 2.00512096e-01 -4.51470464e-01
-1.37431216e+00 9.90836680e-01 2.14167967e-01 -8.00463200e-01
-5.89172721e-01 9.09439206e-01 3.17996562e-01 -1.85926989e-01
4.40984011e-01 -3.73724967e-01 -6.14332020e-01 7.74208978e-02
6.08239412e-01 -2.21479946e-04 -1.69861034e-01 -5.86219966e-01
-4.59578186e-01 5.27657092e-01 -5.50935209e-01 2.67774668e-02
1.09388161e+00 -5.60733140e-01 -4.70625669e-01 1.05443263e+00
1.54936349e+00 -2.22755194e-01 -1.02799368e+00 -4.18942273e-01
3.06988657e-01 -3.38002026e-01 -6.84813112e-02 -3.97243500e-01
-7.31063545e-01 9.45245743e-01 4.75336164e-02 3.82729769e-01
1.12637115e+00 1.04327969e-01 9.63724017e-01 5.07511854e-01
2.98587263e-01 -7.24945068e-01 3.20831120e-01 1.03832197e+00
9.91473973e-01 -1.10908771e+00 6.87441463e-03 -3.85370672e-01
-6.14510894e-01 1.12074506e+00 8.56405497e-01 -3.03391516e-01
6.98040545e-01 -9.82869044e-02 -1.09570809e-01 -2.53562815e-02
-1.14328134e+00 1.04894983e-02 2.92130977e-01 -1.08464643e-01
8.55792701e-01 -3.44885319e-01 -9.22679782e-01 8.85848880e-01
8.40963125e-02 -2.79399097e-01 2.82874525e-01 1.12641346e+00
-6.83535635e-01 -1.16484451e+00 2.21403129e-02 3.71171653e-01
-6.09157860e-01 -3.83221209e-01 -3.40177298e-01 8.60221460e-02
-3.51413339e-02 1.06980062e+00 2.43927062e-01 -2.32174441e-01
1.12604737e-01 3.15493196e-01 2.43204474e-01 -1.03803980e+00
-8.15144658e-01 2.83655655e-02 -7.69068347e-03 -2.42747843e-01
-4.68945980e-01 -7.38313794e-01 -1.28801799e+00 1.23714752e-01
-2.25691855e-01 5.36304176e-01 1.31788149e-01 1.07282770e+00
2.82967418e-01 8.42709601e-01 9.84910548e-01 -6.23483121e-01
-6.25344753e-01 -1.40728712e+00 -4.11164790e-01 3.79753530e-01
5.47502935e-01 -2.09957913e-01 -2.97585368e-01 7.99130201e-02]
|
[11.105637550354004, 8.765676498413086]
|
5e5815a2-4af7-4431-a2cc-e5f15b4c0ed2
|
vsvc-backdoor-attack-against-keyword-spotting
|
2212.10103
| null |
https://arxiv.org/abs/2212.10103v1
|
https://arxiv.org/pdf/2212.10103v1.pdf
|
VSVC: Backdoor attack against Keyword Spotting based on Voiceprint Selection and Voice Conversion
|
Keyword spotting (KWS) based on deep neural networks (DNNs) has achieved massive success in voice control scenarios. However, training of such DNN-based KWS systems often requires significant data and hardware resources. Manufacturers often entrust this process to a third-party platform. This makes the training process uncontrollable, where attackers can implant backdoors in the model by manipulating third-party training data. An effective backdoor attack can force the model to make specified judgments under certain conditions, i.e., triggers. In this paper, we design a backdoor attack scheme based on Voiceprint Selection and Voice Conversion, abbreviated as VSVC. Experimental results demonstrated that VSVC is feasible to achieve an average attack success rate close to 97% in four victim models when poisoning less than 1% of the training data.
|
['Shunhui Ji', 'Yan Xiao', 'Hai Dong', 'Pengcheng Zhang', 'Hanbo Cai']
|
2022-12-20
| null | null | null | null |
['voice-conversion', 'voice-conversion', 'keyword-spotting']
|
['audio', 'speech', 'speech']
|
[-1.06760055e-01 -1.85859486e-01 -2.02043742e-01 1.23181229e-03
-2.99318761e-01 -1.08323658e+00 1.76721171e-01 -5.38306475e-01
-5.25284410e-01 3.07172865e-01 -4.30843771e-01 -1.15914702e+00
1.97252840e-01 -6.62896395e-01 -7.50526905e-01 -5.17467678e-01
2.19864577e-01 -2.16082871e-01 1.61046997e-01 3.69953662e-02
-8.81127864e-02 8.86266828e-01 -1.06560302e+00 -8.19016695e-02
4.07035619e-01 8.91917348e-01 2.24501371e-01 7.10900009e-01
-5.23862876e-02 2.83264279e-01 -1.48167336e+00 -3.76610518e-01
4.73819107e-01 1.06565699e-01 -3.06520253e-01 -6.84070587e-01
2.51482189e-01 -8.06874335e-01 -8.03907454e-01 1.25129092e+00
8.19389582e-01 -2.28542745e-01 -1.14546999e-01 -1.54244196e+00
-1.98141679e-01 9.63413835e-01 -5.14783198e-03 1.25813186e-01
-1.54918646e-02 5.12183905e-01 6.00345671e-01 -6.41798496e-01
1.00335322e-01 9.09843922e-01 5.01753807e-01 9.68633771e-01
-1.13967824e+00 -1.61154366e+00 -3.17582488e-01 -5.04838442e-03
-1.46921515e+00 -7.33146250e-01 8.60271990e-01 -1.08890466e-01
7.63970256e-01 6.60382867e-01 4.96562839e-01 1.64370275e+00
1.29597262e-01 5.12219310e-01 8.52379680e-01 -3.23923618e-01
3.33135039e-01 4.41725701e-01 2.07019448e-01 3.55322182e-01
6.38450384e-01 5.06077230e-01 -6.62621498e-01 -6.36965454e-01
8.49673569e-01 -3.20696473e-01 -5.53983212e-01 1.94055811e-01
-6.11317515e-01 7.57868946e-01 -4.99945730e-02 1.32858396e-01
-1.31940365e-01 2.03248128e-01 4.34457362e-01 3.36048186e-01
-1.83531478e-01 6.04044616e-01 -6.29586637e-01 -3.76719296e-01
-7.22400129e-01 -3.62739526e-02 9.11140382e-01 1.02249992e+00
1.89329997e-01 8.46777081e-01 1.74696460e-01 5.22268474e-01
3.02597135e-01 7.19362736e-01 5.93722343e-01 -3.14405262e-01
6.31135702e-01 -1.29904643e-01 -2.09417507e-01 -8.15530241e-01
-7.41260350e-02 -4.78042156e-01 -4.25226897e-01 2.17140049e-01
9.36603248e-02 -8.47923338e-01 -9.58159089e-01 1.70945013e+00
2.29547396e-01 4.58570898e-01 1.52356610e-01 7.33377814e-01
7.05407739e-01 6.11117661e-01 -7.71091878e-02 -9.34695676e-02
1.19392478e+00 -4.94762719e-01 -1.03940415e+00 -4.11687838e-03
3.23054075e-01 -7.61669099e-01 1.38404524e+00 5.23891807e-01
-4.14817423e-01 -2.70282865e-01 -1.41439342e+00 5.35156012e-01
-3.71141374e-01 1.46346629e-01 5.05273819e-01 1.49960661e+00
-6.92635059e-01 4.10036951e-01 -6.75877452e-01 2.39323914e-01
2.25341231e-01 7.93729901e-01 -2.08619177e-01 5.06277502e-01
-1.69543815e+00 4.07562166e-01 1.61285803e-01 2.76468009e-01
-1.36549830e+00 -8.18804026e-01 -5.69598854e-01 1.27142996e-01
3.95504475e-01 -1.37211934e-01 1.46519887e+00 -1.06182575e-01
-2.07083106e+00 1.26601264e-01 3.99325907e-01 -6.24437094e-01
3.48655909e-01 -4.00762230e-01 -1.08431196e+00 5.84166385e-02
-3.79871070e-01 -6.58637099e-03 1.47083402e+00 -9.70806956e-01
-3.40983927e-01 9.39752907e-02 1.78809054e-02 -3.81926507e-01
-9.38105583e-01 3.99024308e-01 -1.66402698e-01 -7.01876760e-01
-3.00895900e-01 -1.04062045e+00 -7.85516854e-03 -3.70590389e-01
-1.19648147e+00 9.06001925e-02 1.48348641e+00 -4.64754045e-01
1.50304258e+00 -2.44743824e+00 -6.46230817e-01 6.31649017e-01
1.81532189e-01 1.16642487e+00 8.27730671e-02 4.00163949e-01
-1.48759678e-01 5.99277318e-01 3.73773068e-01 -2.52963424e-01
1.24761507e-01 1.49899796e-01 -1.03408015e+00 3.73699993e-01
-2.13668466e-01 5.03741860e-01 -2.56489605e-01 1.18502952e-01
2.09346041e-01 3.81891578e-01 -6.02726579e-01 4.38857228e-01
9.64518115e-02 -1.95387863e-02 -3.79682630e-01 5.65851927e-01
5.33504844e-01 5.45977414e-01 1.15232632e-01 -1.46763906e-01
-9.60516483e-02 6.50251389e-01 -1.06127560e+00 6.50736392e-01
-5.20310998e-01 8.92219305e-01 2.85961211e-01 -3.76634270e-01
8.81340623e-01 8.18939924e-01 -2.20046669e-01 9.07466188e-02
4.52954412e-01 1.85812190e-01 2.44814575e-01 -1.91708341e-01
2.50446707e-01 1.81306362e-01 -2.34387398e-01 3.12103152e-01
1.56736583e-01 -4.22009379e-02 -6.77132607e-01 -3.45698185e-02
9.62656558e-01 -7.30554938e-01 -1.74312860e-01 -9.28695686e-03
2.82938272e-01 -4.66655403e-01 7.21721292e-01 1.00032067e+00
-2.35501453e-01 -4.30524200e-02 4.17873055e-01 -3.54328938e-02
-6.46609068e-01 -8.59169245e-01 -8.22909269e-03 4.83833522e-01
-1.52726829e-01 -7.07971931e-01 -1.02295363e+00 -5.94081700e-01
5.32328114e-02 9.35255826e-01 1.55723572e-01 -6.50305152e-01
-2.65523404e-01 1.00183286e-01 1.76520693e+00 4.41943228e-01
6.82986379e-01 -8.68361831e-01 -3.22256356e-01 1.45670369e-01
3.48848015e-01 -1.48109996e+00 -7.38053799e-01 4.28072095e-01
-4.58605826e-01 -7.88728416e-01 -2.03498989e-01 -4.92030680e-01
4.22578037e-01 2.66372114e-01 1.67038441e-01 -1.05675779e-01
-3.05358410e-01 9.21625420e-02 -8.02221149e-03 -6.74988031e-01
-6.57903671e-01 1.01835713e-01 1.01887321e+00 1.42502785e-02
3.95113409e-01 -5.98333299e-01 -1.58674985e-01 4.09488738e-01
-8.04211676e-01 -5.61380684e-01 2.33533382e-01 6.77708507e-01
3.33926640e-02 4.89933699e-01 5.74483633e-01 -5.76703310e-01
1.05127442e+00 4.24602143e-02 -1.07642472e+00 1.20315179e-01
-5.83109438e-01 -2.74962991e-01 9.24287021e-01 -1.03003466e+00
-4.56706852e-01 -1.43460691e-01 -1.99943811e-01 -1.10966468e+00
-1.09453753e-01 3.00687283e-01 -7.41832554e-01 -6.06573105e-01
4.05538231e-01 2.09493950e-01 -1.29942939e-01 -5.79249382e-01
1.99673891e-01 1.26045358e+00 5.00606298e-01 -3.28343004e-01
1.36091554e+00 -1.31866202e-01 -5.15617609e-01 -1.17411685e+00
-6.32896647e-03 5.71971424e-02 2.06914544e-01 -2.90510088e-01
4.64635253e-01 -7.54302084e-01 -1.15748703e+00 9.41927612e-01
-1.34469998e+00 -1.78857103e-01 3.13397437e-01 8.19956064e-01
1.58341199e-01 1.68491915e-01 -7.30625629e-01 -9.12682831e-01
-5.78105569e-01 -1.29534006e+00 3.16314816e-01 2.91017115e-01
-3.74649942e-01 -4.75577921e-01 -5.67597389e-01 3.65566283e-01
5.51788330e-01 -2.73062766e-01 9.63092446e-01 -1.13963103e+00
-4.49014932e-01 -6.14525080e-01 4.83983576e-01 7.93725610e-01
1.45599917e-01 2.63128787e-01 -1.30763638e+00 -2.76446223e-01
5.19305170e-01 -6.74355105e-02 1.19794346e-01 3.68465157e-03
1.41215158e+00 -6.74941063e-01 -2.62574971e-01 7.67781496e-01
8.58252466e-01 7.25063503e-01 4.76754427e-01 7.09854439e-02
7.60617852e-01 8.38724524e-02 2.33887643e-01 3.11138928e-01
-6.95111215e-01 6.47044301e-01 3.66889745e-01 1.17461421e-01
2.02676326e-01 -6.09587014e-01 6.05267227e-01 8.73450816e-01
5.66473544e-01 -4.11986470e-01 -7.42159009e-01 -4.55400208e-03
-1.02378964e+00 -6.57050550e-01 1.68437421e-01 2.33375812e+00
1.02292323e+00 6.36296511e-01 -3.91810685e-01 4.78914857e-01
1.07784915e+00 1.23855799e-01 -5.50016820e-01 -6.29074514e-01
3.02390426e-01 3.72173697e-01 9.99436140e-01 4.11373645e-01
-9.81616080e-01 1.31257522e+00 6.25656271e+00 1.06326747e+00
-1.57747090e+00 1.96135603e-02 1.86418593e-01 -2.60121197e-01
-2.44600445e-01 -2.20338747e-01 -1.28454733e+00 6.03031337e-01
1.07445693e+00 -1.91514701e-01 4.93700892e-01 1.05459070e+00
2.82022148e-01 5.60639739e-01 -9.07731950e-01 1.03087461e+00
-3.13317955e-01 -1.17340446e+00 1.03913203e-01 4.14897233e-01
2.77306996e-02 -2.62677193e-01 4.05636728e-01 3.07320207e-01
3.92691731e-01 -9.31507349e-01 6.64712191e-01 -2.26610348e-01
1.27891052e+00 -1.23125494e+00 3.89147729e-01 3.20296377e-01
-8.16638231e-01 -2.83159167e-01 -3.19466650e-01 2.94004738e-01
-5.37829474e-02 4.01916683e-01 -1.27840006e+00 6.61716461e-02
4.18249935e-01 -3.95002604e-01 8.45955610e-02 7.56815314e-01
-5.95138192e-01 1.43287981e+00 -7.30640113e-01 -3.43309671e-01
1.74092367e-01 1.76056966e-01 8.58259976e-01 6.91049278e-01
2.54828364e-01 3.51770669e-02 -2.07304269e-01 8.87133002e-01
-4.33227181e-01 -3.88981283e-01 -8.27593684e-01 -6.04589760e-01
1.32176566e+00 1.13992357e+00 -1.21089660e-01 1.75217196e-01
3.82077023e-02 7.99703240e-01 -2.02669173e-01 4.92319793e-01
-1.02326334e+00 -1.01484036e+00 1.23344004e+00 -2.00277343e-01
1.52973726e-01 -4.32850122e-01 -3.66763910e-04 -8.04804742e-01
-2.98099630e-02 -1.15328825e+00 -2.32178241e-01 -4.18246418e-01
-8.34826350e-01 7.43181169e-01 -4.05364513e-01 -1.15039718e+00
-7.68369138e-02 -6.46037221e-01 -6.74140155e-01 9.65473354e-01
-9.51382399e-01 -6.78805768e-01 3.68294567e-01 8.45785856e-01
1.87653825e-01 -7.20332980e-01 1.09834266e+00 4.79372233e-01
-1.09651315e+00 1.33455932e+00 -1.57357603e-01 6.37525856e-01
4.98309195e-01 -6.60188079e-01 6.34233177e-01 1.08478367e+00
4.24077719e-01 1.24343896e+00 6.54246628e-01 -6.14734530e-01
-1.72826588e+00 -9.85577881e-01 5.06017327e-01 1.76244155e-01
6.78074241e-01 -8.98492694e-01 -5.71789503e-01 5.03044724e-01
-1.88102648e-02 -3.71688511e-03 1.16208100e+00 -1.51849672e-01
-4.83881772e-01 -2.27156997e-01 -1.02763915e+00 1.07716095e+00
5.53320587e-01 -1.28359056e+00 -1.89994365e-01 1.69351906e-01
1.22728229e+00 -4.58062410e-01 -4.30481672e-01 2.01472968e-01
4.50732112e-01 -2.49170154e-01 8.09994400e-01 -8.53647828e-01
-3.54113728e-01 -2.14467198e-01 -1.87517405e-01 -1.30022025e+00
3.02494407e-01 -1.48593330e+00 -3.11149508e-01 1.53183246e+00
4.97862875e-01 -1.03069019e+00 7.20953405e-01 8.00943553e-01
-6.76194206e-02 -3.09453279e-01 -1.20733988e+00 -1.12130213e+00
-3.67602766e-01 -8.57463300e-01 9.76379216e-01 1.00997806e+00
1.91492494e-03 1.26649261e-01 -6.35677338e-01 1.01409733e+00
3.81143302e-01 -6.44357741e-01 7.82393277e-01 -6.99242055e-01
-3.13633502e-01 -1.30542859e-01 -3.95998210e-01 -9.50740397e-01
2.75962740e-01 -5.81880331e-01 -1.37378201e-01 -4.70698178e-01
-8.38948488e-01 -3.74718279e-01 -3.34028572e-01 6.24566734e-01
2.47159392e-01 -1.87821075e-01 3.35600048e-01 -5.76838739e-02
4.02997285e-01 3.72648388e-01 6.29521430e-01 -1.31913468e-01
-2.13534221e-01 5.02057850e-01 -4.32371736e-01 7.07240701e-01
1.17625129e+00 -6.89067483e-01 -7.69269288e-01 -2.01147705e-01
-3.41566175e-01 1.47285283e-01 3.20066541e-01 -1.08085644e+00
5.81870317e-01 -1.64012592e-02 -6.05798066e-02 -3.06070924e-01
3.57994139e-01 -1.19351304e+00 9.87877920e-02 6.18131280e-01
-3.01835567e-01 -2.58342117e-01 4.50938493e-01 5.71003616e-01
8.78114328e-02 -3.87217581e-01 3.14231992e-01 5.02203524e-01
-4.20918643e-01 3.33769828e-01 -8.14334691e-01 -4.72208679e-01
9.53929365e-01 -1.53376907e-02 -3.83999079e-01 -4.73042160e-01
-5.36213934e-01 -1.05241895e-01 -1.72688991e-01 6.01989567e-01
8.20144892e-01 -1.20764053e+00 1.98348641e-01 7.14961767e-01
-3.87430191e-01 -4.01914775e-01 -6.44667074e-02 4.15766165e-02
-3.37649792e-01 6.62400186e-01 2.55439728e-01 -2.47451082e-01
-1.83894658e+00 3.83595467e-01 5.69696665e-01 4.23503399e-01
-4.68890905e-01 9.11834896e-01 -3.22700888e-01 -5.23231566e-01
7.20660150e-01 -3.52770537e-01 5.31955883e-02 -2.89306462e-01
6.09003723e-01 1.14468947e-01 2.49345198e-01 4.66133468e-02
-4.75670666e-01 -1.12205464e-02 -1.42272934e-01 -4.05390829e-01
5.90510011e-01 4.33026344e-01 1.70211062e-01 9.12366137e-02
1.20320964e+00 4.59834814e-01 -9.61784482e-01 -9.70097557e-02
-2.44121924e-01 -5.19397080e-01 4.44286823e-01 -6.42046690e-01
-1.27934563e+00 9.39306021e-01 4.36748981e-01 3.15748304e-01
6.01091087e-01 -5.54656923e-01 1.29268336e+00 7.91053891e-01
5.81383049e-01 -9.78988588e-01 -2.43022323e-01 3.82099301e-01
4.90247875e-01 -4.70221996e-01 -5.32329917e-01 -4.00473535e-01
-4.69850957e-01 1.14953804e+00 6.68096662e-01 1.45608038e-01
9.96589601e-01 8.59330177e-01 3.81892204e-01 -4.03208360e-02
-6.35161281e-01 5.85390031e-01 -1.48486957e-01 6.39057577e-01
-3.73192489e-01 2.94015706e-01 1.00084931e-01 1.17364621e+00
-5.92604160e-01 -1.94911867e-01 6.30724609e-01 8.30101609e-01
-2.86198765e-01 -1.12004054e+00 -5.57513654e-01 3.33157808e-01
-8.04743350e-01 -3.39180261e-01 -5.11797667e-01 6.13127768e-01
-1.41616389e-01 1.23071551e+00 -1.08550407e-01 -1.31097174e+00
4.31719095e-01 1.87963098e-01 -2.93393463e-01 -5.14727354e-01
-7.61457503e-01 -2.97020655e-02 3.88087153e-01 -5.35879850e-01
5.04423797e-01 -2.29350477e-01 -1.18204367e+00 -6.57196105e-01
-8.41224790e-01 3.04203004e-01 9.46019351e-01 7.25962937e-01
4.06157136e-01 5.77531159e-01 1.10752857e+00 -3.51312041e-01
-1.23975313e+00 -6.98729694e-01 -7.97804356e-01 -5.19753754e-01
3.23965222e-01 -4.38027889e-01 -7.15736330e-01 -5.02169549e-01]
|
[13.956228256225586, 5.817488193511963]
|
18f64a60-4480-4304-9eb1-63aa4479e0e7
|
the-hanabi-challenge-a-new-frontier-for-ai
|
1902.00506
| null |
https://arxiv.org/abs/1902.00506v2
|
https://arxiv.org/pdf/1902.00506v2.pdf
|
The Hanabi Challenge: A New Frontier for AI Research
|
From the early days of computing, games have been important testbeds for studying how well machines can do sophisticated decision making. In recent years, machine learning has made dramatic advances with artificial agents reaching superhuman performance in challenge domains like Go, Atari, and some variants of poker. As with their predecessors of chess, checkers, and backgammon, these game domains have driven research by providing sophisticated yet well-defined challenges for artificial intelligence practitioners. We continue this tradition by proposing the game of Hanabi as a new challenge domain with novel problems that arise from its combination of purely cooperative gameplay with two to five players and imperfect information. In particular, we argue that Hanabi elevates reasoning about the beliefs and intentions of other agents to the foreground. We believe developing novel techniques for such theory of mind reasoning will not only be crucial for success in Hanabi, but also in broader collaborative efforts, especially those with human partners. To facilitate future research, we introduce the open-source Hanabi Learning Environment, propose an experimental framework for the research community to evaluate algorithmic advances, and assess the performance of current state-of-the-art techniques.
|
['Marc G. Bellemare', 'Shibl Mourad', 'Neil Burch', 'Subhodeep Moitra', 'Michael Bowling', 'Marc Lanctot', 'Hugo Larochelle', 'Jakob N. Foerster', 'Emilio Parisotto', 'Nolan Bard', 'Iain Dunning', 'H. Francis Song', 'Vincent Dumoulin', 'Sarath Chandar', 'Edward Hughes']
|
2019-02-01
| null | null | null | null |
['game-of-hanabi']
|
['playing-games']
|
[-2.88035542e-01 1.37978628e-01 1.14125438e-01 -1.81947742e-02
-1.97400808e-01 -4.88279879e-01 6.98488176e-01 3.33153233e-02
-6.73667252e-01 7.69175708e-01 1.46793678e-01 -5.57416022e-01
-1.05134599e-01 -7.88623631e-01 -1.11938186e-01 -4.61065620e-01
-3.28257471e-01 1.08356583e+00 5.44775963e-01 -9.68655348e-01
3.30251575e-01 -1.78230911e-01 -1.58104432e+00 1.88246995e-01
3.16251248e-01 6.36802256e-01 1.45942327e-02 8.66066337e-01
4.97192323e-01 1.92636693e+00 -4.85809743e-01 -8.71600628e-01
5.55173576e-01 -4.26313102e-01 -1.08007574e+00 -2.62057126e-01
-2.02335849e-01 -4.21538591e-01 -3.70804638e-01 1.03442037e+00
3.40030760e-01 1.64097697e-01 1.54969662e-01 -1.48838615e+00
-2.62087345e-01 1.06130016e+00 -5.58561504e-01 1.32795423e-01
3.24180126e-01 5.34135044e-01 1.40755308e+00 -9.15541034e-03
5.29548287e-01 1.24755132e+00 8.28167737e-01 6.64399743e-01
-8.76411259e-01 -5.82471371e-01 8.68600532e-02 8.30045104e-01
-1.14421999e+00 -2.39708692e-01 4.78507638e-01 -5.16586185e-01
1.02246344e+00 2.69314617e-01 9.00176525e-01 7.86842108e-01
1.54898167e-01 1.19615424e+00 1.16097462e+00 -3.49065512e-01
5.69234014e-01 -3.12395126e-01 -9.36884210e-02 8.67266417e-01
2.97544122e-01 3.60151321e-01 -4.05704647e-01 -2.41361320e-01
7.33008087e-01 -3.68340164e-01 1.25745341e-01 -5.29013455e-01
-1.12037313e+00 1.39561093e+00 1.76145211e-01 2.75860310e-01
-4.35155153e-01 3.18120688e-01 2.86591291e-01 4.84800637e-01
-6.05372377e-02 9.04006660e-01 -3.01820099e-01 -8.32996905e-01
-5.17781258e-01 8.79944086e-01 1.46384168e+00 4.56999719e-01
4.65423018e-01 -1.40586436e-01 4.98337358e-01 2.00192615e-01
1.80631518e-01 -1.35098770e-01 4.14672107e-01 -1.50302792e+00
9.55513269e-02 4.38524008e-01 2.64562607e-01 -1.03260028e+00
-6.49359405e-01 -3.35386962e-01 -5.00766635e-01 8.19305658e-01
7.12022126e-01 -5.04189968e-01 -2.48995900e-01 1.68720388e+00
2.82591254e-01 1.49044871e-01 2.43581519e-01 9.77942586e-01
3.33889544e-01 2.26155445e-01 -6.64919689e-02 1.53006673e-01
1.32507241e+00 -9.62227404e-01 -2.10735068e-01 -6.42175019e-01
8.12076509e-01 -5.08437634e-01 7.38668084e-01 8.68079782e-01
-1.20923781e+00 -7.36052841e-02 -1.23402870e+00 6.79451674e-02
3.41240503e-02 -8.02811086e-01 1.34236050e+00 8.10365736e-01
-1.10726511e+00 5.11623681e-01 -1.00349510e+00 -4.11537647e-01
3.48672301e-01 4.11570489e-01 -1.07302308e-01 4.44123223e-02
-1.15272570e+00 1.38554740e+00 5.33834815e-01 -2.82865882e-01
-9.68941689e-01 -2.13000283e-01 -6.83896959e-01 -5.89244906e-03
1.18407333e+00 -7.14523375e-01 1.78207624e+00 -9.63068187e-01
-1.48262191e+00 1.05115914e+00 5.32687664e-01 -9.05543685e-01
7.02309072e-01 3.94505374e-02 1.18669063e-01 -3.52408886e-01
1.33109614e-01 4.17047381e-01 3.27434033e-01 -9.81820881e-01
-1.20761991e+00 -4.64572936e-01 9.64923203e-01 5.46222448e-01
1.69112056e-01 2.75656790e-01 -4.29267483e-03 -1.06715076e-01
-1.91960409e-01 -1.05406749e+00 -5.97330928e-01 -4.41931725e-01
1.03623578e-02 -4.80946839e-01 2.69050688e-01 -2.75685728e-01
8.25318277e-01 -2.03634667e+00 4.02452469e-01 -7.04474524e-02
8.82542312e-01 1.98338434e-01 7.81522840e-02 5.59641242e-01
3.20296466e-01 1.76295824e-03 2.18254700e-01 -2.71995366e-02
4.72408324e-01 4.05189663e-01 2.39523873e-01 3.17186058e-01
-2.26362079e-01 9.87212718e-01 -1.12190044e+00 -3.42781454e-01
1.78648055e-01 -2.66244292e-01 -8.26014102e-01 1.88223943e-01
-3.75062257e-01 -1.53770568e-02 -5.01446605e-01 1.97370544e-01
1.70298845e-01 -3.48354936e-01 5.59741080e-01 6.42034411e-01
-2.02721804e-01 3.92354578e-01 -1.25608480e+00 1.25244129e+00
-2.18134327e-03 5.97035229e-01 5.01226366e-01 -1.04740596e+00
2.34499499e-01 1.44494921e-01 3.71732473e-01 -6.38581991e-01
4.82236475e-01 -1.48792744e-01 9.11373556e-01 -3.50609511e-01
5.27321994e-01 -5.32869756e-01 -2.46022418e-01 7.59872377e-01
-1.96463227e-01 -4.75225508e-01 5.37969530e-01 1.75605744e-01
1.47967446e+00 -1.68414451e-02 8.07327569e-01 -1.93364471e-01
3.47376853e-01 6.62949920e-01 5.62374949e-01 1.26240599e+00
-5.64666331e-01 1.48227319e-01 7.85090268e-01 -9.59718883e-01
-1.01551557e+00 -7.38592803e-01 4.75133687e-01 1.59547877e+00
1.92140058e-01 -6.47383988e-01 -7.98766196e-01 -3.52687418e-01
-1.70301661e-01 7.34030128e-01 -6.81999326e-01 -3.26706059e-02
-5.22577047e-01 -7.39272773e-01 8.13237488e-01 4.54707652e-01
7.47903407e-01 -1.04278874e+00 -1.11711156e+00 3.82986456e-01
-1.61091805e-01 -1.03994167e+00 2.69551128e-01 2.80925483e-01
-2.16841444e-01 -1.44036782e+00 -7.91945904e-02 -7.42856026e-01
-2.27169350e-01 2.35094205e-01 1.48411763e+00 6.04834378e-01
-1.12562299e-01 4.04047728e-01 -6.53880775e-01 -7.83323586e-01
-6.33911788e-01 4.08284813e-02 1.73133820e-01 -5.98761261e-01
5.66670775e-01 -6.04430318e-01 -2.37681225e-01 4.25039679e-01
-5.81809461e-01 4.21893805e-01 4.21574295e-01 7.53963351e-01
-3.78986686e-01 4.90375727e-01 -1.03046261e-02 -8.88516068e-01
8.39658320e-01 -6.15611255e-01 -7.36355901e-01 -1.53622761e-01
-2.89558977e-01 -1.62353098e-01 5.38715005e-01 -7.85695538e-02
-9.62424994e-01 -5.18254578e-01 -8.17111433e-02 3.06995928e-01
-1.80896640e-01 7.79243410e-01 2.46330619e-01 -1.95082482e-02
1.02993536e+00 3.85980979e-02 1.97675407e-01 -5.02061844e-03
2.93109983e-01 5.89239359e-01 3.86136383e-01 -9.05627847e-01
7.94808269e-01 3.87307703e-01 -2.44396955e-01 -7.75158048e-01
-7.26027250e-01 -3.30569178e-01 -1.75211444e-01 -1.96060196e-01
8.55867684e-01 -1.10874903e+00 -1.45372367e+00 6.61068082e-01
-8.54667604e-01 -7.71862149e-01 -1.28437355e-01 2.37980351e-01
-7.70974576e-01 2.11451694e-01 -8.29994500e-01 -8.65242243e-01
3.20994824e-01 -1.11197972e+00 3.38846922e-01 4.82591808e-01
-4.80973214e-01 -9.78604317e-01 3.51143539e-01 9.82485592e-01
4.54449147e-01 -1.56432301e-01 8.96857977e-01 -1.04796934e+00
-5.24399817e-01 -3.13663594e-02 4.37068194e-02 3.18362676e-02
-2.39064366e-01 -3.94108683e-01 -6.44393146e-01 -2.42549971e-01
1.68996334e-01 -8.71921480e-01 2.74602145e-01 2.37680208e-02
4.26968545e-01 4.53556329e-03 -3.07923667e-02 1.91928357e-01
1.03457403e+00 5.53172946e-01 6.15938902e-01 8.39022636e-01
2.81733215e-01 5.02705634e-01 4.29787248e-01 6.78995967e-01
1.03723145e+00 6.15490079e-01 6.56938791e-01 2.84274668e-01
3.33488435e-01 -4.88867946e-02 2.25258783e-01 4.25440103e-01
-7.29641795e-01 -1.55724794e-01 -1.17990696e+00 2.54573017e-01
-2.26311135e+00 -1.32892156e+00 1.73656702e-01 1.78535831e+00
6.92270637e-01 4.22422260e-01 5.72502255e-01 8.39219093e-02
4.41498727e-01 1.85732156e-01 -4.00440753e-01 -2.37477273e-01
5.02534257e-03 2.80272990e-01 9.21632424e-02 6.22204065e-01
-1.09985805e+00 1.37346148e+00 6.29721308e+00 7.12793946e-01
-6.32072985e-01 1.32805526e-01 4.18452382e-01 -2.52472097e-03
2.24823505e-01 4.56037112e-02 -4.61404145e-01 6.20750524e-02
5.16304910e-01 -4.81538653e-01 1.02978599e+00 1.05498350e+00
-2.45986376e-02 -4.68457162e-01 -1.20096886e+00 8.14500153e-01
2.63626277e-02 -1.38671553e+00 -7.47118473e-01 3.38744670e-01
5.30069947e-01 4.36156571e-01 -6.85758963e-02 8.40664685e-01
1.50147259e+00 -1.23031032e+00 8.67017984e-01 -2.33636573e-01
-1.93210244e-01 -6.36684895e-01 1.07756138e+00 9.02119994e-01
-7.80106843e-01 -4.17015702e-01 -4.50991035e-01 -1.33290207e+00
-2.47983590e-01 -2.84556717e-01 -9.39188182e-01 2.65148133e-01
6.37935519e-01 4.27018970e-01 -2.86387384e-01 9.43517864e-01
-3.33998084e-01 5.56809187e-01 -4.09486473e-01 -4.83795106e-01
5.98010898e-01 -1.22577697e-01 4.22121555e-01 5.88576555e-01
-3.61078203e-01 6.96905911e-01 4.71464694e-01 7.06355512e-01
1.44400775e-01 -2.10125402e-01 -3.74047697e-01 -2.85657525e-01
2.79376090e-01 1.17723238e+00 -9.28320110e-01 -1.23252481e-01
-4.10312146e-01 5.66851020e-01 4.90074128e-01 -1.88535407e-01
-7.76018620e-01 2.01805726e-01 1.00131619e+00 -1.39726494e-02
2.20917255e-01 -4.44389313e-01 -2.75601119e-01 -1.29492807e+00
-2.57614821e-01 -1.68617547e+00 4.50471193e-01 -5.40205717e-01
-1.22707760e+00 4.30064410e-01 -1.48588762e-01 -6.94120467e-01
-6.54137433e-01 -7.28691876e-01 -7.42134154e-01 2.29397133e-01
-9.55792487e-01 -1.04367375e+00 -1.14091039e-01 4.98115927e-01
4.84564394e-01 -4.02211428e-01 8.41960192e-01 -1.94952115e-01
-3.45088512e-01 5.52263074e-02 -1.67011634e-01 4.40621912e-01
2.78373510e-01 -1.20289958e+00 6.15904570e-01 6.37721956e-01
4.09280449e-01 3.52890849e-01 9.98527229e-01 -2.58527458e-01
-1.47447538e+00 -1.11086659e-01 2.78401315e-01 -6.90218925e-01
1.12023509e+00 -3.18268985e-01 -3.13903928e-01 1.04512334e+00
3.46422940e-01 -6.85812593e-01 5.93419790e-01 7.93948948e-01
-2.80192703e-01 3.57843250e-01 -7.71258891e-01 9.52483237e-01
9.99457061e-01 -2.52788037e-01 -1.10452378e+00 2.44115084e-01
2.78381228e-01 -5.69400370e-01 -3.36435467e-01 -6.66225180e-02
6.24930501e-01 -1.68414128e+00 6.51105285e-01 -1.00503862e+00
5.98295510e-01 -2.79859185e-01 -1.90294832e-01 -1.40182149e+00
-6.12630129e-01 -8.96868825e-01 3.82000268e-01 5.06435275e-01
2.55423814e-01 -4.30540293e-01 1.47643340e+00 7.63875484e-01
1.31103218e-01 -4.77459788e-01 -6.95508182e-01 -4.59471852e-01
3.28918904e-01 -8.10635030e-01 3.15367222e-01 8.53819489e-01
6.11502826e-01 5.05790889e-01 -4.60287422e-01 -9.69093293e-02
7.50036359e-01 1.16914988e-03 1.10424256e+00 -1.46658385e+00
-8.94276440e-01 -8.02239120e-01 -8.47422302e-01 -9.12408412e-01
7.26557076e-02 -5.69950938e-01 1.79457426e-01 -1.16268229e+00
3.31580251e-01 -4.30599064e-01 3.01929060e-02 5.65336704e-01
-1.58242702e-01 3.18931520e-01 7.07446396e-01 7.82704540e-03
-1.22679591e+00 3.20049971e-02 1.15018773e+00 -8.69558379e-02
9.37387645e-02 1.40507057e-01 -1.16687274e+00 1.34018373e+00
8.00042391e-01 -2.31555000e-01 -2.91844159e-01 -4.11648244e-01
8.48838449e-01 -1.11711044e-02 2.01079220e-01 -1.51848900e+00
6.80554986e-01 -6.81865394e-01 -9.66110528e-02 2.63055801e-01
4.68748838e-01 -5.01608670e-01 7.12898001e-02 7.13619947e-01
-2.19635263e-01 8.77718348e-03 1.01153702e-02 4.96482290e-02
-1.67882532e-01 -3.03816259e-01 6.65683687e-01 -7.42232442e-01
-1.17843664e+00 -4.88578454e-02 -9.95815039e-01 5.77197731e-01
1.35852444e+00 -8.17552432e-02 -3.05963546e-01 -1.01838279e+00
-6.53179228e-01 6.00511789e-01 7.05236673e-01 1.21604800e-01
2.06206068e-01 -5.84917843e-01 -9.81513500e-01 -8.32779557e-02
-1.46306664e-01 -2.66044885e-01 9.63540748e-02 5.94883204e-01
-9.18295622e-01 1.06348909e-01 -6.82543397e-01 -8.72206688e-02
-1.40049374e+00 3.45092773e-01 5.15449464e-01 -5.67530334e-01
-3.30327421e-01 1.15570402e+00 3.35464597e-01 -5.27155876e-01
1.18342303e-01 1.10136524e-01 -2.07513750e-01 -1.83212817e-01
6.95720851e-01 2.52672613e-01 -1.89418256e-01 -2.46302217e-01
-3.05898458e-01 -1.23042658e-01 -3.07680368e-01 -2.41943389e-01
1.53345978e+00 9.78249013e-02 -1.48979515e-01 2.20794827e-01
1.40351415e-01 -6.03211746e-02 -1.03115475e+00 -1.11554541e-01
1.49879232e-01 -1.95604235e-01 -6.46209642e-02 -9.36353326e-01
-5.56049287e-01 5.49223661e-01 -1.07882336e-01 7.17581868e-01
7.20783353e-01 8.39984789e-02 5.12217879e-01 8.19515109e-01
1.14927411e+00 -1.06264329e+00 1.56279802e-02 1.14129257e+00
2.72019714e-01 -1.36398411e+00 7.99597949e-02 -5.48381917e-03
-9.32973087e-01 8.86557162e-01 6.65516019e-01 -2.62953371e-01
2.56820112e-01 6.69522524e-01 2.39020377e-01 -3.18055928e-01
-1.21991432e+00 -5.89631796e-01 -4.16602671e-01 8.23389113e-01
2.00740352e-01 3.26923639e-01 -8.76859799e-02 8.43386650e-01
-7.97276795e-01 1.70878023e-01 9.06437397e-01 1.20201778e+00
-7.15181291e-01 -1.24472547e+00 -4.37356561e-01 2.24359483e-01
-3.73450875e-01 -1.49551854e-01 -6.90546215e-01 1.14948392e+00
3.05862188e-01 1.19649315e+00 -8.69976953e-02 -6.28726125e-01
1.29477575e-01 -1.45004377e-01 7.29962051e-01 -6.41448677e-01
-7.71952391e-01 -3.88313979e-01 4.51971054e-01 -7.44144320e-01
-1.90611854e-01 -7.35693514e-01 -1.03545666e+00 -1.13652110e+00
-3.36124264e-02 5.08551598e-01 2.43868500e-01 1.30670035e+00
-1.00973740e-01 1.18377261e-01 1.15225472e-01 -9.56650794e-01
-8.51231158e-01 -6.90977097e-01 -5.77627897e-01 1.05170265e-01
-1.33514732e-01 -7.36216664e-01 -3.27450842e-01 -3.69438261e-01]
|
[3.554410457611084, 1.5000791549682617]
|
a5603dba-80a3-4c88-ae5c-052028146fc8
|
visual-information-guided-zero-shot
|
2201.09107
| null |
https://arxiv.org/abs/2201.09107v2
|
https://arxiv.org/pdf/2201.09107v2.pdf
|
Visual Information Guided Zero-Shot Paraphrase Generation
|
Zero-shot paraphrase generation has drawn much attention as the large-scale high-quality paraphrase corpus is limited. Back-translation, also known as the pivot-based method, is typical to this end. Several works leverage different information as "pivot" such as language, semantic representation and so on. In this paper, we explore using visual information such as image as the "pivot" of back-translation. Different with the pipeline back-translation method, we propose visual information guided zero-shot paraphrase generation (ViPG) based only on paired image-caption data. It jointly trains an image captioning model and a paraphrasing model and leverage the image captioning model to guide the training of the paraphrasing model. Both automatic evaluation and human evaluation show our model can generate paraphrase with good relevancy, fluency and diversity, and image is a promising kind of pivot for zero-shot paraphrase generation.
|
['Xiaojun Wan', 'Zhe Lin']
|
2022-01-22
| null |
https://aclanthology.org/2022.coling-1.568
|
https://aclanthology.org/2022.coling-1.568.pdf
|
coling-2022-10
|
['paraphrase-generation', 'paraphrase-generation']
|
['computer-code', 'natural-language-processing']
|
[ 5.93856946e-02 -7.22596720e-02 -4.49808031e-01 -2.45686755e-01
-7.38174796e-01 -5.90282559e-01 9.38837647e-01 -1.81125596e-01
2.03366429e-02 6.19949937e-01 7.18738377e-01 -2.52790153e-01
4.95827198e-01 -7.57093012e-01 -1.10014927e+00 -2.60517299e-01
1.14277256e+00 3.70638818e-01 1.24948464e-01 -4.34100449e-01
4.99395251e-01 6.88582957e-02 -1.17648900e+00 7.34891057e-01
9.91171062e-01 5.43430984e-01 7.45483220e-01 3.14775825e-01
-4.93939102e-01 7.57984042e-01 -5.55978596e-01 -6.16443276e-01
1.94421828e-01 -1.04777622e+00 -9.36928630e-01 1.07235424e-01
4.62142140e-01 -2.24063918e-01 -3.92315179e-01 8.83856416e-01
5.36063015e-01 -1.15156196e-01 5.55860519e-01 -1.24244058e+00
-1.54298460e+00 5.21538079e-01 -6.78859234e-01 2.55839806e-02
8.67540240e-01 5.41068256e-01 9.24637616e-01 -1.31428909e+00
1.05987906e+00 1.20431304e+00 3.93440485e-01 6.36693656e-01
-1.21082151e+00 -4.59502190e-01 -1.27652109e-01 6.17549241e-01
-1.08737814e+00 -3.00417453e-01 9.50850904e-01 -4.61391419e-01
6.41811490e-01 1.31130308e-01 1.08281958e+00 1.60156643e+00
1.51030898e-01 1.07458651e+00 1.14492476e+00 -6.33980930e-01
1.15053877e-02 2.76821047e-01 -2.33840793e-01 5.80227911e-01
-5.50193600e-02 2.52089871e-04 -5.17188370e-01 2.44124457e-01
9.17010367e-01 3.36183757e-01 -3.67908239e-01 -5.60662031e-01
-1.44692314e+00 8.49113405e-01 8.93345237e-01 1.40343264e-01
-2.09591061e-01 6.99599385e-02 4.02907491e-01 4.79925692e-01
4.19133037e-01 7.29789674e-01 3.63161445e-01 6.16241200e-03
-1.24238944e+00 2.17827082e-01 4.04025167e-01 1.08199787e+00
6.93443477e-01 -8.26969892e-02 -9.16363180e-01 1.26331413e+00
2.30638348e-02 7.28686333e-01 7.77872086e-01 -7.52833307e-01
7.83599198e-01 7.65975177e-01 2.45398894e-01 -7.11417019e-01
2.29783818e-01 -2.73575366e-01 -6.70912325e-01 -1.07156960e-02
3.13310958e-02 1.88088104e-01 -1.11346257e+00 1.56308377e+00
2.79660914e-02 8.64969045e-02 7.83788681e-04 1.25971580e+00
1.08548975e+00 8.77109110e-01 -2.22980425e-01 2.45322324e-02
1.38193130e+00 -1.65222967e+00 -6.15037739e-01 -5.10217905e-01
3.80103081e-01 -1.05675471e+00 1.71652973e+00 -1.36274457e-01
-1.19634068e+00 -6.05174303e-01 -1.02372158e+00 -2.57822901e-01
-2.29358301e-01 7.20004290e-02 2.23046258e-01 1.50766954e-01
-9.38619494e-01 3.52058381e-01 -2.71640539e-01 -6.02705300e-01
4.48270589e-01 -4.36655551e-01 -3.31003010e-01 -5.56633115e-01
-1.26605797e+00 1.18866169e+00 1.97991177e-01 -3.73888940e-01
-7.83026993e-01 -6.15175962e-01 -8.77635241e-01 -4.21563797e-02
1.53185368e-01 -1.41505814e+00 1.16677964e+00 -1.12304926e+00
-1.56855226e+00 1.22348034e+00 -2.40613207e-01 -3.11990976e-01
7.05804408e-01 -4.27300334e-02 1.57538861e-01 3.72810602e-01
4.68358308e-01 9.70731616e-01 9.69179332e-01 -1.25538194e+00
-1.27222776e-01 -7.67504424e-02 6.65758029e-02 5.99671304e-01
-2.14778662e-01 4.96925488e-02 -4.72480208e-01 -7.71172762e-01
-3.69493999e-02 -8.85327637e-01 -6.28140047e-02 7.83349127e-02
-4.28905934e-01 6.95958659e-02 7.11877346e-01 -7.42589056e-01
1.04199541e+00 -1.92422163e+00 2.99608409e-01 -4.13165778e-01
-1.15688831e-01 2.36883447e-01 -4.46772963e-01 8.85172606e-01
-1.05307959e-01 -2.90235672e-02 -1.91898812e-02 -3.56693864e-01
-1.17615730e-01 -1.48735438e-02 -6.65060341e-01 3.53251048e-03
4.47430462e-02 1.77403295e+00 -1.25225103e+00 -6.85581625e-01
4.67284143e-01 1.37744322e-01 -3.44111592e-01 5.12454629e-01
-1.51232481e-01 3.63352597e-01 -1.63516954e-01 4.22385573e-01
5.57423115e-01 -4.66350138e-01 -3.86932522e-01 -2.68996179e-01
5.98513260e-02 2.19589267e-02 -1.76765740e-01 2.25971103e+00
-8.15732360e-01 7.30187595e-01 -4.32709754e-01 -6.38058305e-01
9.71009910e-01 8.52609947e-02 -2.00240803e-03 -1.03669679e+00
-1.71350539e-01 1.55814514e-01 -4.38647181e-01 -6.23394847e-01
6.09652042e-01 -4.04071152e-01 -1.60191566e-01 5.29849231e-01
-7.02478662e-02 -6.65928245e-01 1.05571039e-01 5.78086078e-01
8.22582662e-01 5.01968503e-01 3.56932223e-01 6.53340966e-02
3.12814265e-01 3.54257882e-01 -1.32429609e-02 6.94827080e-01
5.66496961e-02 1.44383228e+00 3.71048421e-01 -2.05846310e-01
-1.64188015e+00 -1.25349259e+00 3.37054074e-01 6.53593600e-01
6.63882554e-01 -3.30746740e-01 -9.38832819e-01 -4.64128405e-01
-3.28507274e-01 8.30061436e-01 -6.43343508e-01 -4.73317742e-01
-4.53084916e-01 -2.14875132e-01 2.87208021e-01 5.14300525e-01
7.36451149e-01 -1.36933768e+00 -3.55846167e-01 1.76094938e-02
-7.00332165e-01 -1.15618598e+00 -8.86224508e-01 -5.95794559e-01
-6.42190933e-01 -7.82193959e-01 -1.45228088e+00 -1.09132028e+00
7.74131536e-01 7.99999595e-01 1.27360547e+00 -1.35768980e-01
-2.00712487e-01 9.49314833e-02 -6.61769927e-01 -2.48258114e-01
-6.73080444e-01 -1.18504725e-01 -4.24098015e-01 -1.12283349e-01
2.33364347e-02 -6.42136097e-01 -1.15860343e+00 3.58975142e-01
-7.49611437e-01 1.00223577e+00 8.51696432e-01 1.01399934e+00
5.18680811e-01 -1.11322689e+00 5.92109501e-01 -6.10211492e-01
1.08272409e+00 -4.95761037e-01 -3.24801393e-02 6.09054446e-01
-4.65011150e-01 2.53578760e-02 1.03215492e+00 -4.14802372e-01
-8.99815202e-01 -1.72593564e-01 9.54867452e-02 -1.07482731e+00
-9.37062898e-04 1.26272723e-01 -8.43583234e-03 1.98613688e-01
7.87346840e-01 7.89288163e-01 9.36804935e-02 -2.48430535e-01
8.60064089e-01 7.82911122e-01 6.15876555e-01 -2.59715497e-01
7.79472291e-01 4.69789177e-01 -3.00361812e-01 -4.69130874e-01
-8.43013287e-01 -3.96842659e-01 -4.05087352e-01 -1.68261617e-01
8.62032115e-01 -1.01817727e+00 -1.62583306e-01 3.41744512e-01
-1.44733179e+00 -2.25189433e-01 -5.74969530e-01 1.03678279e-01
-8.72068822e-01 5.90857029e-01 -4.74402487e-01 -2.53698289e-01
-6.64627910e-01 -1.11808908e+00 1.38395333e+00 1.87911794e-01
-1.70657128e-01 -6.06848359e-01 2.52546132e-01 6.98823988e-01
3.45543623e-01 1.11380130e-01 7.34813511e-01 -2.29248986e-01
-8.19911301e-01 -1.39207602e-01 -6.33837283e-01 7.15791136e-02
-3.45126390e-02 -4.09236491e-01 -5.62619448e-01 -1.34425670e-01
-5.33383489e-02 -5.88878274e-01 9.17977691e-01 2.17264041e-01
8.87555420e-01 -4.59804714e-01 -2.14780331e-01 7.42706358e-01
1.33982337e+00 -1.45139858e-01 9.65907156e-01 3.43124688e-01
9.11238849e-01 4.38517004e-01 8.04396927e-01 2.12306425e-01
3.88244212e-01 9.80228901e-01 2.43348569e-01 -2.55958945e-01
-6.33876383e-01 -1.10564303e+00 4.12061393e-01 9.17283237e-01
2.99319357e-01 -2.70374894e-01 -6.00984693e-01 7.23388135e-01
-2.11374736e+00 -1.26539624e+00 4.49718125e-02 2.00078201e+00
8.10206115e-01 -2.75054336e-01 8.01859275e-02 -2.04656318e-01
9.01320934e-01 4.82347161e-01 -4.44791853e-01 -5.70246041e-01
-1.20337531e-01 -1.28553391e-01 7.12030083e-02 3.38179171e-01
-3.61497343e-01 1.26390946e+00 5.12247229e+00 1.13703203e+00
-9.75830615e-01 3.16406786e-01 4.93278742e-01 -1.03233501e-01
-5.74231684e-01 1.72508195e-01 -3.60319465e-01 8.67847085e-01
4.26405907e-01 -3.28952789e-01 3.92489225e-01 7.65086532e-01
4.76241738e-01 -6.50502443e-02 -1.01942158e+00 1.41281164e+00
6.73731565e-01 -1.65714288e+00 5.76646805e-01 -2.61656314e-01
9.21147406e-01 -1.85414299e-01 3.72470543e-03 3.04528803e-01
2.43660301e-01 -8.72664034e-01 8.43880653e-01 5.17506540e-01
1.17857277e+00 -5.23533762e-01 4.83389199e-01 3.80379945e-01
-1.04311359e+00 1.35256335e-01 -4.90520000e-01 4.01449949e-02
5.93059838e-01 2.72461802e-01 -8.52716148e-01 6.00242317e-01
1.30080372e-01 8.60234797e-01 -7.46841252e-01 1.09043109e+00
-8.20872068e-01 2.19830856e-01 2.34421954e-01 -2.51447737e-01
2.87859768e-01 -3.83070201e-01 5.68571508e-01 1.05705643e+00
5.26438415e-01 -1.56050384e-01 1.41336843e-01 1.11103523e+00
-1.34501263e-01 3.81589562e-01 -9.42913115e-01 3.93712297e-02
4.31734741e-01 1.06939554e+00 -5.61633468e-01 -5.68861783e-01
-4.02453065e-01 1.71821010e+00 2.78580606e-01 4.22263712e-01
-9.02873278e-01 -4.92216766e-01 1.66516483e-01 3.55651319e-01
1.37836509e-03 6.98214024e-02 -3.43503505e-01 -1.56418037e+00
4.06068489e-02 -6.38871729e-01 -1.30497187e-01 -1.80307615e+00
-1.26884174e+00 6.04083836e-01 5.82071245e-02 -1.66866446e+00
-3.56230646e-01 -2.68638283e-01 -1.00485134e+00 1.03672838e+00
-1.26421344e+00 -1.73615003e+00 -7.27899075e-01 4.37955230e-01
1.20950258e+00 -1.82494283e-01 4.71779555e-01 -1.36718035e-01
-2.76104122e-01 6.24236107e-01 -2.59689074e-02 9.94151644e-03
7.96952903e-01 -1.02024090e+00 8.22063208e-01 7.69185424e-01
3.43867451e-01 4.37529862e-01 7.79703677e-01 -7.15128362e-01
-1.11370862e+00 -1.12329292e+00 8.59152257e-01 -5.14961660e-01
5.70893347e-01 -4.06042159e-01 -6.66827917e-01 2.84511983e-01
6.00610137e-01 -3.87123287e-01 1.72901943e-01 -3.60611498e-01
-4.17532831e-01 7.10504279e-02 -8.36112022e-01 1.16246653e+00
1.41968656e+00 -6.34531975e-01 -1.02853167e+00 6.20990038e-01
1.08142614e+00 -2.84604341e-01 -2.56167710e-01 9.99005046e-03
4.57299113e-01 -1.05504858e+00 1.30287611e+00 -3.72485846e-01
1.35950255e+00 -9.67532024e-02 2.48310700e-01 -1.65628910e+00
-4.24567610e-01 -5.53782642e-01 1.44361943e-01 1.05002737e+00
2.61188537e-01 -2.47137487e-01 8.47830415e-01 1.39773697e-01
-4.00393039e-01 -7.81765461e-01 -6.06550813e-01 -1.02549100e+00
-9.09949988e-02 1.59486502e-01 6.37776673e-01 7.90100276e-01
3.30049336e-01 9.93402600e-01 -7.80064821e-01 -5.96785486e-01
5.07097065e-01 7.66108751e-01 1.13618016e+00 -5.09588063e-01
-3.98402482e-01 -4.25724477e-01 -2.13845998e-01 -1.29380345e+00
1.21765025e-01 -1.47495484e+00 -1.84773400e-01 -2.12938237e+00
7.91177571e-01 2.78520495e-01 1.37833789e-01 2.43671149e-01
-2.77448416e-01 4.83141243e-01 5.36539674e-01 6.82215631e-01
-5.49675226e-01 7.74496496e-01 1.82223296e+00 -1.48170099e-01
-1.57920539e-01 -3.82589519e-01 -6.83091879e-01 3.26515794e-01
6.29406214e-01 -2.61525601e-01 -6.18708253e-01 -5.04293442e-01
2.16327146e-01 4.84847546e-01 5.97406507e-01 -7.72870064e-01
9.37080011e-02 -4.24314946e-01 3.20429146e-01 -4.91220713e-01
5.28405905e-01 -3.39883059e-01 1.29277170e-01 4.40284729e-01
-5.31795084e-01 3.83828193e-01 -2.14611351e-01 6.40729070e-01
-3.79549563e-01 -4.44404930e-01 9.71267402e-01 -5.54261744e-01
-7.02802896e-01 1.69891924e-01 1.15917183e-01 2.69734025e-01
9.87372160e-01 -7.72998750e-01 -4.54198480e-01 -6.34142876e-01
-4.04491246e-01 3.35931182e-01 1.00441587e+00 7.14742661e-01
9.80094373e-01 -1.56812465e+00 -8.81575644e-01 1.45600706e-01
5.37928045e-01 -4.00916904e-01 3.33713651e-01 5.64608276e-01
-6.31032109e-01 4.28223640e-01 -5.04695535e-01 -5.42292476e-01
-1.15799212e+00 8.12318385e-01 2.71389028e-03 -3.10946643e-01
-7.97881484e-01 6.79836035e-01 6.33547425e-01 -3.79305072e-02
-3.58041525e-01 1.10153191e-01 -3.56531814e-02 -9.97108221e-02
4.25915807e-01 -6.41607791e-02 -3.99645895e-01 -5.27336419e-01
-1.09961722e-02 7.41312087e-01 -9.70964804e-02 -4.57040548e-01
1.02111661e+00 -3.49185973e-01 2.05097318e-01 3.62325013e-01
1.22969699e+00 -3.36613506e-01 -9.95804012e-01 -1.36194453e-01
-4.10899460e-01 -8.82475376e-01 -5.30607224e-01 -8.97937119e-01
-6.46904588e-01 1.24785995e+00 2.31288865e-01 -2.16417149e-01
8.12641799e-01 1.50319681e-01 1.18954599e+00 7.66307637e-02
5.93115449e-01 -1.01954234e+00 7.68066406e-01 3.20651352e-01
1.48190725e+00 -1.27977121e+00 -1.65845513e-01 -2.87484944e-01
-1.08583987e+00 6.23400152e-01 8.06976914e-01 -3.03755313e-01
-2.06502140e-01 -5.31357288e-01 3.44114676e-02 2.55742352e-02
-7.20629156e-01 -1.47436216e-01 2.96419293e-01 5.92545927e-01
1.67251021e-01 -9.35016498e-02 -6.04034066e-01 4.36590821e-01
-3.96511734e-01 2.67945737e-01 5.52695394e-01 5.80923975e-01
-4.93163317e-01 -1.08119321e+00 -3.00792873e-01 2.72497445e-01
-6.25052974e-02 -3.96658599e-01 -6.78985536e-01 3.23968023e-01
-1.88633800e-01 7.42833138e-01 -7.28230998e-02 -3.81793529e-01
4.44510043e-01 -2.20899060e-02 7.46065676e-01 -9.51639771e-01
-5.53364396e-01 -3.25817287e-01 -1.88209683e-01 -5.62294781e-01
-4.55443263e-02 -5.20188920e-02 -8.86593878e-01 -1.04725242e-01
-1.12499498e-01 1.06567815e-01 6.39293611e-01 8.36750925e-01
6.46545708e-01 1.31211445e-01 7.70542741e-01 -9.58154082e-01
-4.11020041e-01 -9.14182127e-01 -1.82977110e-01 1.16642773e+00
-8.91828835e-02 -3.44775558e-01 -2.22175926e-01 8.16770941e-02]
|
[11.222764015197754, 1.0104187726974487]
|
43bcdaa5-02dc-4f63-96b2-4d9513b9b827
|
recent-advances-in-artificial-intelligence
|
2301.05864
| null |
https://arxiv.org/abs/2301.05864v1
|
https://arxiv.org/pdf/2301.05864v1.pdf
|
Recent advances in artificial intelligence for retrosynthesis
|
Retrosynthesis is the cornerstone of organic chemistry, providing chemists in material and drug manufacturing access to poorly available and brand-new molecules. Conventional rule-based or expert-based computer-aided synthesis has obvious limitations, such as high labor costs and limited search space. In recent years, dramatic breakthroughs driven by artificial intelligence have revolutionized retrosynthesis. Here we aim to present a comprehensive review of recent advances in AI-based retrosynthesis. For single-step and multi-step retrosynthesis both, we first list their goal and provide a thorough taxonomy of existing methods. Afterwards, we analyze these methods in terms of their mechanism and performance, and introduce popular evaluation metrics for them, in which we also provide a detailed comparison among representative methods on several public datasets. In the next part we introduce popular databases and established platforms for retrosynthesis. Finally, this review concludes with a discussion about promising research directions in this field.
|
['Mingli Song', 'Tingjun Hou', 'Shaolun Yao', 'Lingxiang Jia', 'Tiantao Liu', 'Zunlei Feng', 'Jie Song', 'Zipeng Zhong']
|
2023-01-14
| null | null | null | null |
['retrosynthesis']
|
['medical']
|
[ 3.84341270e-01 -2.32281506e-01 -8.63019049e-01 6.97951987e-02
-3.67979825e-01 -1.14974630e+00 4.83904928e-01 8.22431207e-01
-3.66636395e-01 1.42241383e+00 -9.44785550e-02 -2.53124863e-01
-1.01797529e-01 -6.26879752e-01 -3.90823036e-01 -9.59831059e-01
-2.11403314e-02 4.97057050e-01 1.67596601e-02 -3.97130102e-01
5.81557214e-01 7.96342194e-01 -1.40569484e+00 1.06725946e-01
9.00837600e-01 9.16350007e-01 3.17182213e-01 2.22213984e-01
-1.61315784e-01 3.42673063e-01 -6.23870432e-01 -7.15025604e-01
1.79377303e-01 -5.03275514e-01 -5.86067319e-01 -4.37976599e-01
-7.72755593e-02 6.48032129e-02 -1.91129278e-02 8.77407789e-01
7.09643424e-01 3.36409032e-01 9.45264697e-01 -9.07407880e-01
-4.58272427e-01 8.40756118e-01 8.57259706e-02 -8.89815688e-02
5.88002026e-01 1.00974835e-01 1.13712251e+00 -1.29940438e+00
8.15123916e-01 9.25961852e-01 4.53158170e-01 5.61632633e-01
-9.80738163e-01 -7.72380471e-01 3.02679062e-01 2.09810004e-01
-1.47559607e+00 -5.69915473e-01 6.34541869e-01 -3.60256255e-01
1.25644863e+00 3.08584362e-01 7.60356486e-01 8.82813454e-01
4.86960739e-01 3.60570729e-01 7.52634943e-01 -1.02228217e-01
5.26317835e-01 3.18028815e-02 -4.84914362e-01 6.06393516e-01
4.80144650e-01 1.06852382e-01 -5.30608892e-01 -2.03724533e-01
3.42270970e-01 5.55473231e-02 1.39971465e-01 -3.37461591e-01
-1.29422879e+00 9.04687345e-01 3.41054201e-01 5.98719001e-01
-4.33360964e-01 -1.07012883e-01 6.25158966e-01 -2.07976520e-01
1.27360761e-01 1.30379879e+00 -6.50307477e-01 2.78170139e-01
-9.40165579e-01 3.25617164e-01 7.43974328e-01 1.03758228e+00
3.06331247e-01 1.58407897e-01 2.46979490e-01 5.65582097e-01
4.08445261e-02 -5.15532456e-02 4.09177005e-01 -4.65512842e-01
6.24688603e-02 1.18468717e-01 1.02603614e-01 -6.73193395e-01
-6.18197203e-01 -2.98861284e-02 -7.23964274e-01 -3.83481771e-01
2.41589412e-01 9.45239794e-03 -8.25177193e-01 1.18127072e+00
5.09018421e-01 -4.58299279e-01 -1.08936327e-02 5.76089859e-01
1.13339269e+00 9.60930645e-01 4.78116900e-01 -1.00569808e+00
1.33569217e+00 -1.11951959e+00 -9.72795308e-01 5.47991335e-01
1.42345488e-01 -1.14678657e+00 2.26219878e-01 9.48898554e-01
-1.32139266e+00 -1.74352676e-01 -1.20954907e+00 -1.86726123e-01
-1.17376697e+00 1.13077708e-01 1.11237192e+00 9.52514172e-01
-3.45981628e-01 9.76509690e-01 -5.06220520e-01 -2.92528659e-01
2.87897766e-01 6.46771908e-01 -2.79599249e-01 3.46919179e-01
-1.17029679e+00 9.17451739e-01 8.81882846e-01 7.81976581e-02
-1.22496355e+00 -1.07314348e+00 -5.47154307e-01 -1.90680608e-01
9.30191755e-01 -7.36116946e-01 1.47237349e+00 -1.61576554e-01
-1.89046633e+00 5.28319776e-01 1.32766455e-01 -3.83273810e-01
2.80134916e-01 8.70295018e-02 -5.34071267e-01 1.23141855e-01
1.13075636e-01 8.83218348e-01 4.59603667e-01 -8.80151570e-01
-5.81385136e-01 -4.06589545e-02 -1.52282923e-01 1.26382913e-02
1.12277560e-01 2.94559568e-01 -2.08058909e-01 -9.27156746e-01
-3.14218849e-02 -7.08610058e-01 -6.77183211e-01 -1.42156556e-01
-6.55618370e-01 -3.37845027e-01 1.11511812e-01 -8.15899670e-02
1.46912539e+00 -1.28485727e+00 3.58957559e-01 1.45503536e-01
1.38859883e-01 2.55477548e-01 1.94947019e-01 1.45832992e+00
-5.27428746e-01 4.56758052e-01 -1.56022370e-01 3.44727308e-01
-2.69269168e-01 -3.71941537e-01 -3.90795529e-01 3.60449493e-01
5.30951191e-03 7.01510012e-01 -1.32137620e+00 -2.67404437e-01
3.62757504e-01 1.38243735e-01 -4.61637288e-01 -3.17585081e-01
-7.75668442e-01 6.56542063e-01 -6.67660892e-01 1.38207245e+00
2.40358725e-01 -7.55585805e-02 6.60882533e-01 -3.73485535e-01
-8.77426744e-01 5.08106232e-01 -8.17148566e-01 1.89338982e+00
-1.61902159e-01 -2.11406082e-01 -4.02172506e-01 -8.61822605e-01
7.73488879e-01 6.52285218e-01 7.85900593e-01 -4.14957374e-01
2.15262666e-01 6.46955907e-01 1.55694097e-01 -6.99856551e-03
5.48340678e-01 -5.59344888e-01 -2.61792578e-02 -3.45624685e-02
1.25237303e-02 -4.87654388e-01 8.89348507e-01 1.60436016e-02
5.45623481e-01 2.50162214e-01 1.35360098e+00 -2.34830782e-01
1.07878017e+00 3.62790972e-01 5.06810486e-01 3.74946058e-01
-3.56503315e-02 4.32177670e-02 1.24589264e-01 -7.92412758e-01
-1.07923126e+00 -6.93894684e-01 -1.47808447e-01 8.62373650e-01
3.44627649e-02 -1.10335624e+00 -3.92206937e-01 -5.86652815e-01
-2.42875904e-01 3.82496119e-01 -3.81315559e-01 6.70967698e-02
-3.40071380e-01 -8.87641847e-01 5.05063057e-01 2.08575979e-01
2.28598818e-01 -7.64835119e-01 1.96475908e-03 7.14367151e-01
-3.60239744e-02 -8.03310513e-01 -2.78027624e-01 2.75416225e-01
-9.16052818e-01 -1.07717645e+00 -7.34070301e-01 -6.27897143e-01
2.81712085e-01 3.66720527e-01 7.23789811e-01 -1.35827869e-01
-7.09241629e-01 -2.53407151e-01 -2.12206587e-01 -9.26413119e-01
-4.94316250e-01 7.55171180e-02 4.04743165e-01 -6.66606486e-01
-3.69670130e-02 -5.88246405e-01 -9.62059379e-01 3.73064816e-01
-7.46597707e-01 -2.83852398e-01 7.18295515e-01 6.11137927e-01
1.01485109e+00 2.37659842e-01 8.77953649e-01 -5.68045855e-01
4.63972747e-01 -5.51985443e-01 -1.02316582e+00 3.10668141e-01
-1.01690793e+00 5.92420287e-02 8.88306916e-01 -2.20439360e-01
-9.75402832e-01 5.29134989e-01 -3.96275938e-01 2.09717289e-01
1.24983460e-01 6.94787025e-01 -2.85287887e-01 -1.30739138e-02
6.17805243e-01 1.29288584e-01 -2.11151525e-01 -5.73485732e-01
7.07589388e-01 3.54319997e-02 9.88245457e-02 -6.79468811e-01
5.21736741e-01 3.70366454e-01 5.05855739e-01 -9.59039450e-01
-5.01651227e-01 -5.93487203e-01 -3.53656858e-01 2.38497686e-02
8.76642942e-01 -6.75147414e-01 -1.17199969e+00 -6.38394058e-03
-9.32748795e-01 8.73609856e-02 5.61156236e-02 4.31587696e-01
-6.62256181e-01 4.01257604e-01 -5.39047182e-01 -5.14939487e-01
-5.92994452e-01 -1.56181216e+00 8.39443326e-01 1.34749711e-01
-2.56639004e-01 -7.79841959e-01 3.26648563e-01 3.19840729e-01
-1.05815686e-01 2.54231304e-01 1.03431833e+00 -8.22694480e-01
-6.82531118e-01 -1.06546301e-02 3.04492265e-01 -2.19793364e-01
5.93598068e-01 2.96608627e-01 -7.10152686e-01 1.24328390e-01
-4.22445446e-01 -2.96654612e-01 8.18900347e-01 5.30109704e-01
1.33020306e+00 -2.47517556e-01 -8.04511905e-01 2.54032165e-01
1.42888916e+00 1.01698029e+00 4.85201776e-01 2.77991623e-01
4.27612871e-01 6.50770843e-01 1.27721465e+00 5.63556671e-01
-3.05705518e-01 5.04369140e-01 4.37186331e-01 2.63717026e-01
2.02529296e-01 -3.47035736e-01 -3.29035781e-02 4.50769365e-01
-7.99083829e-01 -4.96407866e-01 -5.30194998e-01 -3.71815567e-03
-1.50685453e+00 -9.75291371e-01 9.87283215e-02 2.28980494e+00
1.20000350e+00 -3.05596143e-01 7.08436906e-01 2.49420777e-01
5.88456511e-01 -1.60860077e-01 -5.16556203e-01 -5.32591701e-01
-9.13711563e-02 6.86576247e-01 6.51417851e-01 2.77328730e-01
-1.24442053e+00 1.48327589e+00 7.65497780e+00 1.16596746e+00
-1.17848349e+00 -4.42882925e-01 4.02942151e-01 -1.15948029e-01
8.94250721e-02 -1.60335854e-03 -9.29555297e-01 3.31082255e-01
9.15354252e-01 -5.10971785e-01 4.06518877e-01 7.43373513e-01
4.03047115e-01 -5.26485592e-02 -1.35150027e+00 9.14829373e-01
-3.14073920e-01 -2.28568816e+00 3.54757458e-01 1.60424858e-02
8.22445333e-01 -5.22998035e-01 5.34169860e-02 -1.62468970e-01
-1.47212908e-01 -1.09235811e+00 7.05218315e-01 -8.66270158e-03
7.07971513e-01 -1.00979781e+00 8.71239305e-02 -1.09099038e-01
-1.44727802e+00 -1.16582930e-01 -2.07006887e-01 1.06536783e-01
1.42567396e-01 5.16208827e-01 -1.00003302e+00 1.23147881e+00
2.63075650e-01 9.15008366e-01 1.30243734e-01 1.08794272e+00
-5.02471328e-01 -3.82916592e-02 -1.21778317e-01 -4.66670483e-01
3.40591341e-01 -6.45605147e-01 3.15550566e-01 1.13203645e+00
1.36320263e-01 4.03785467e-01 3.54798675e-01 6.84184909e-01
-1.97514981e-01 4.68849152e-01 -5.33593178e-01 -8.02670419e-01
7.72774518e-01 1.13029778e+00 -1.14997566e+00 -6.24577522e-01
-3.35275531e-01 5.13005555e-01 -4.37766016e-01 1.69511810e-01
-8.46233368e-01 -5.80909014e-01 6.79749250e-01 -2.67257571e-01
2.61116952e-01 -1.89456537e-01 1.06514066e-01 -7.22743332e-01
-6.52631402e-01 -1.16845059e+00 6.18376434e-01 -3.32032174e-01
-8.87539923e-01 1.53341383e-01 2.07765833e-01 -1.09636307e+00
3.31214815e-01 -8.59934092e-01 -1.78645536e-01 2.80477971e-01
-1.19648099e+00 -6.08585775e-01 4.06911016e-01 3.18386592e-02
1.00237405e+00 -1.28486425e-01 8.54438186e-01 4.15570229e-01
-9.33328211e-01 1.01336822e-01 2.56238371e-01 -7.33827710e-01
1.08337259e+00 -9.11162615e-01 1.35374829e-01 2.86992580e-01
-2.53430486e-01 1.15690362e+00 8.74495804e-01 -8.84974599e-01
-1.70524848e+00 -1.02895880e+00 8.15107882e-01 -1.95372924e-01
8.55916619e-01 -2.88480699e-01 -1.51447833e-01 2.89446086e-01
3.44010703e-02 -6.36730731e-01 1.12568545e+00 -3.23151827e-01
-1.47330388e-01 -3.78993489e-02 -1.14477575e+00 9.25756752e-01
1.09984398e+00 2.73805019e-02 -2.87618011e-01 8.25435698e-01
6.40634656e-01 -3.06940824e-01 -1.29544365e+00 2.77519345e-01
9.11917925e-01 -6.65691495e-01 1.35811567e+00 -7.16172636e-01
4.98593003e-01 -5.10223627e-01 1.97085410e-01 -8.59553218e-01
-3.62865180e-01 -1.17531037e+00 2.05944166e-01 5.42442858e-01
5.91240108e-01 -6.10831439e-01 6.21710062e-01 1.80728897e-01
-4.78058755e-01 -8.31802666e-01 -4.52537566e-01 -1.13088417e+00
2.87919194e-01 6.63575083e-02 8.60785246e-01 9.03484821e-01
6.26198173e-01 6.36990786e-01 -3.35875154e-01 -4.23713058e-01
2.86080301e-01 2.13379011e-01 3.00830543e-01 -9.87064123e-01
8.41482878e-02 -7.72014499e-01 -6.74002469e-02 -8.03568959e-01
-2.25059658e-01 -9.27512288e-01 -1.04824804e-01 -1.29704475e+00
2.82266259e-01 -1.32242963e-01 -2.04787537e-01 1.70110092e-01
1.15885742e-01 1.66535974e-01 -3.08082819e-01 1.31316492e-02
-4.14756149e-01 4.51974720e-01 1.35480380e+00 -2.93092400e-01
-4.58483636e-01 -1.93610996e-01 -7.90922821e-01 6.13944829e-01
1.05673516e+00 -5.61884224e-01 -6.09637439e-01 4.38359678e-01
7.54330575e-01 -9.45876241e-02 -4.20801252e-01 -3.79861832e-01
9.46485922e-02 -6.67505324e-01 2.50220031e-01 -1.15126717e+00
4.39734757e-01 -7.02163756e-01 4.67937112e-01 9.34237421e-01
-1.80986941e-01 -2.16075271e-01 3.06227297e-01 7.17096329e-01
-1.10445330e-02 -2.22352579e-01 6.54116690e-01 -5.64171612e-01
-4.59710091e-01 7.32356608e-01 -9.97326791e-01 -6.90393448e-01
1.21297824e+00 -3.23316365e-01 -1.54983863e-01 1.65294647e-01
-8.01619232e-01 7.21420199e-02 2.52400786e-01 2.18540579e-01
4.73526329e-01 -1.01812375e+00 -8.82968679e-02 -7.00655699e-01
5.66500783e-01 -2.86012739e-01 2.58340943e-03 1.06841981e+00
-8.60171974e-01 1.28334928e+00 -1.61229745e-01 -9.20521766e-02
-1.42964983e+00 1.40692365e+00 1.74445540e-01 -6.36866540e-02
2.13068709e-01 6.95849717e-01 4.03846294e-01 -1.18425421e-01
2.23867834e-01 -5.97415388e-01 -2.18190029e-01 2.18163326e-01
5.56629956e-01 5.77728033e-01 3.67367923e-01 -1.66659027e-01
-7.53391027e-01 4.29253250e-01 -1.62230000e-01 2.07338467e-01
1.02366304e+00 2.14601606e-01 -1.78241506e-01 4.16346602e-02
6.90168083e-01 -2.93855881e-03 -3.64002496e-01 3.03597987e-01
-5.77885620e-02 -3.67728546e-02 -1.78637460e-01 -9.88421440e-01
-4.61466819e-01 5.87825775e-01 -1.00352466e-01 -2.14168489e-01
9.27883565e-01 2.39885431e-02 6.15310729e-01 9.53674495e-01
3.55147690e-01 -1.24685061e+00 2.13824168e-01 1.67785138e-01
1.02618456e+00 -9.91172969e-01 7.51085877e-01 -8.68591726e-01
-2.54444659e-01 1.42590296e+00 -3.22027989e-02 3.60487312e-01
3.88137728e-01 -3.75921309e-01 -6.82262659e-01 -3.04127157e-01
-7.35113978e-01 -1.76892877e-01 1.71056271e-01 4.09699231e-01
9.02031541e-01 -3.46075967e-02 -1.12235522e+00 4.51199710e-01
-2.13669658e-01 -1.94637641e-01 1.84332356e-01 1.35118628e+00
-5.54097235e-01 -1.90835428e+00 -3.21251392e-01 -1.67152688e-01
-7.25014269e-01 -5.37428737e-01 -1.28985155e+00 8.28110814e-01
2.01078117e-01 1.25918937e+00 -6.40923023e-01 -3.39092091e-02
2.06924364e-01 5.98379336e-02 1.09637022e+00 -6.49601996e-01
-6.88881099e-01 5.07079780e-01 6.15414739e-01 -5.04034519e-01
-5.80758095e-01 -5.12537122e-01 -1.35276103e+00 -3.09516698e-01
-6.24554217e-01 6.08272970e-01 1.16418207e+00 6.22630000e-01
2.29719013e-01 3.42202425e-01 6.45022690e-01 -7.22951531e-01
-2.04112828e-01 -3.52456659e-01 -2.65215248e-01 -6.47163451e-01
-1.26166567e-01 -7.36396670e-01 1.07559174e-01 3.59033376e-01]
|
[4.491466045379639, 6.109657287597656]
|
25c6baf6-d303-4545-86d4-86dae823cf3d
|
attention-based-spatial-temporal-graph-neural
|
2305.00985
| null |
https://arxiv.org/abs/2305.00985v1
|
https://arxiv.org/pdf/2305.00985v1.pdf
|
Attention-based Spatial-Temporal Graph Neural ODE for Traffic Prediction
|
Traffic forecasting is an important issue in intelligent traffic systems (ITS). Graph neural networks (GNNs) are effective deep learning models to capture the complex spatio-temporal dependency of traffic data, achieving ideal prediction performance. In this paper, we propose attention-based graph neural ODE (ASTGODE) that explicitly learns the dynamics of the traffic system, which makes the prediction of our machine learning model more explainable. Our model aggregates traffic patterns of different periods and has satisfactory performance on two real-world traffic data sets. The results show that our model achieves the highest accuracy of the root mean square error metric among all the existing GNN models in our experiments.
|
['Jane Macfarlane', 'Hadi Meidani', 'Weiheng Zhong']
|
2023-05-01
| null | null | null | null |
['traffic-prediction']
|
['time-series']
|
[-5.58997393e-01 -2.30502442e-01 -3.93597573e-01 -1.69742703e-01
3.12611997e-01 2.44320542e-01 4.47539270e-01 -5.37599623e-01
3.05321246e-01 6.09114468e-01 5.57933860e-02 -1.00794601e+00
-2.86386281e-01 -1.01056933e+00 -4.80301231e-01 -2.80413687e-01
-2.66105533e-01 7.09559619e-01 8.06619465e-01 -7.67227411e-01
-9.20997038e-02 7.82516539e-01 -1.14183021e+00 -6.70734271e-02
1.02881491e+00 1.26185179e+00 -1.03883475e-01 6.04784727e-01
-5.77621281e-01 1.48910964e+00 -5.54658711e-01 -6.47575378e-01
2.43140697e-01 -7.37405494e-02 -5.10682046e-01 -2.22045362e-01
2.54329175e-01 -1.36272684e-01 -1.66171885e+00 7.72247434e-01
1.01124734e-01 3.17941487e-01 6.12985611e-01 -1.85938740e+00
-1.07825422e+00 4.17692840e-01 -3.80148530e-01 1.00621355e+00
-5.34789741e-01 6.03267312e-01 7.18584299e-01 -1.23933025e-01
1.04786329e-01 1.39009857e+00 8.41295719e-01 3.98831993e-01
-8.80879521e-01 -8.47055256e-01 4.34241861e-01 9.06365216e-01
-1.21661985e+00 -1.79722235e-01 9.55805779e-01 -5.19609928e-01
1.10130382e+00 5.05327305e-04 6.58626497e-01 8.82917285e-01
7.95851052e-01 5.88796318e-01 3.54520947e-01 2.70979911e-01
-6.50352985e-02 -4.84105468e-01 4.63629305e-01 7.44278550e-01
2.93019980e-01 2.26894096e-01 3.75163406e-02 2.25090459e-01
8.99032712e-01 4.40512031e-01 2.23652676e-01 2.08493724e-01
-5.61849415e-01 7.27575719e-01 1.14135075e+00 1.26333982e-01
-6.93227172e-01 8.85728657e-01 4.47290391e-01 2.99571127e-01
8.21144521e-01 -1.15222715e-01 -1.04277126e-01 -2.05669776e-01
-3.77729505e-01 5.04166037e-02 4.83821481e-01 8.89787138e-01
5.60477912e-01 9.34137642e-01 -2.37981781e-01 6.44695580e-01
1.69318065e-01 7.19587743e-01 1.24306150e-01 -7.09543169e-01
6.79312885e-01 7.49113262e-01 -4.46691662e-01 -1.53210616e+00
-5.95192075e-01 -6.25186861e-01 -1.18450141e+00 -5.93311898e-02
2.39034772e-01 -2.24018380e-01 -1.05610228e+00 1.40884089e+00
2.70268950e-03 9.39959288e-01 -2.46828049e-01 8.15827250e-01
8.43281329e-01 1.01501989e+00 3.78197879e-01 2.16275886e-01
9.24925745e-01 -1.22267854e+00 -9.49580371e-01 -1.80829152e-01
4.44319129e-01 -8.01640376e-02 5.72531044e-01 -1.75817356e-01
-7.17741787e-01 -8.13524961e-01 -5.22421002e-01 5.55859096e-02
-6.32046998e-01 -3.78903478e-01 9.34242368e-01 3.39370608e-01
-9.94091988e-01 4.94284719e-01 -8.05529952e-01 -2.79961824e-01
8.22633624e-01 4.90951985e-01 2.41968557e-01 -2.16860417e-02
-1.52021563e+00 8.17936242e-01 3.77653569e-01 4.04420167e-01
-6.78033233e-01 -7.08118021e-01 -4.93747205e-01 3.40591937e-01
5.88411331e-01 -5.62501371e-01 1.01727211e+00 -5.57899654e-01
-1.30205750e+00 8.42926577e-02 -1.04425333e-01 -9.09329772e-01
2.78632373e-01 2.52684921e-01 -1.31804168e+00 -2.93181390e-01
-9.98974368e-02 2.09223926e-01 5.11097014e-01 -6.81621134e-01
-4.99885499e-01 1.03939466e-01 2.13277131e-01 -5.20531535e-01
-9.54414383e-02 -2.92188168e-01 -4.95000124e-01 -6.45504892e-01
-3.95433635e-01 -7.58886158e-01 -4.68254238e-01 -2.94415325e-01
-2.41219819e-01 -8.52032244e-01 1.14637315e+00 -7.35493898e-01
1.82148623e+00 -1.70379186e+00 -5.82775533e-01 2.46180803e-01
8.62106979e-01 9.50190425e-01 -2.92299062e-01 4.50191557e-01
-1.61622800e-02 9.52392519e-02 2.40828127e-01 3.02712113e-01
6.51002452e-02 6.07888281e-01 -4.06381041e-01 8.14594924e-02
2.25330532e-01 1.74108505e+00 -7.67144561e-01 -3.28447819e-01
3.86942804e-01 3.37249309e-01 -3.32003683e-01 1.20172128e-01
-5.38016438e-01 3.09433520e-01 -8.08040082e-01 1.16541244e-01
7.13987470e-01 -6.33000731e-01 -1.27174690e-01 -6.56181052e-02
2.26692095e-01 1.83046877e-01 -5.03392458e-01 7.78130829e-01
-4.63862121e-01 1.20336866e+00 -6.52090132e-01 -1.30411303e+00
1.08952630e+00 1.16934357e-02 5.55834293e-01 -1.51031673e+00
3.33241075e-01 -2.07595885e-01 4.58882570e-01 -6.27444923e-01
1.61078662e-01 1.99923739e-01 2.06275344e-01 4.19182539e-01
-1.61935657e-01 7.12422609e-01 3.10668141e-01 2.11566061e-01
1.23077488e+00 -6.87856436e-01 -1.92536771e-01 -1.47717744e-01
6.95769250e-01 -1.16533391e-01 5.63387036e-01 6.39229298e-01
-4.20437813e-01 -5.98048307e-02 8.12774181e-01 -1.22052896e+00
-1.14147449e+00 -8.86314929e-01 2.87206888e-01 9.55353677e-01
1.18577786e-01 -1.91337898e-01 -5.30434251e-01 -7.49421537e-01
2.78778195e-01 7.09588289e-01 -6.78206146e-01 -4.83574927e-01
-1.01037598e+00 -4.33567971e-01 5.44282019e-01 6.98415577e-01
7.28581011e-01 -1.19573021e+00 3.91705692e-01 4.76765394e-01
-1.43998235e-01 -1.34681857e+00 -4.84215409e-01 -4.90801811e-01
-6.12470627e-01 -1.24564254e+00 -2.98650831e-01 -5.67894340e-01
3.36724639e-01 5.46728969e-01 1.16762424e+00 5.79317868e-01
-8.27445555e-03 2.78140511e-02 4.00425568e-02 -3.75447154e-01
-3.47363561e-01 3.51590991e-01 1.02447504e-02 3.56398582e-01
7.80611277e-01 -7.77559280e-01 -5.95763683e-01 5.12187123e-01
-5.42053461e-01 -5.34163602e-02 3.98627490e-01 4.74988610e-01
2.37441093e-01 4.13860321e-01 8.51845920e-01 -7.30195463e-01
9.58245337e-01 -9.93655503e-01 -7.37597823e-01 1.63562134e-01
-8.45520377e-01 5.71141504e-02 1.14643037e+00 -3.28687519e-01
-6.98874056e-01 -8.30310404e-01 -1.26681343e-01 -8.18085968e-01
-3.39372754e-02 5.25582492e-01 1.01867437e-01 -6.81019947e-02
3.97978693e-01 2.98541218e-01 -6.79760706e-03 -3.59735638e-01
1.28678858e-01 5.53115368e-01 4.94131178e-01 -2.58109182e-01
8.06138456e-01 2.09143803e-01 4.62063342e-01 -6.69942796e-01
-7.62403905e-01 -1.14151597e-01 -4.82539684e-01 -7.49555349e-01
6.60346508e-01 -5.53259909e-01 -1.27178812e+00 4.92055535e-01
-1.13369405e+00 -5.50686598e-01 -2.95439344e-02 3.74699324e-01
-3.69107991e-01 1.96487233e-01 -7.38010705e-01 -6.97523475e-01
-2.14670017e-01 -8.81251514e-01 4.46397245e-01 2.50154674e-01
3.03459853e-01 -1.55087984e+00 6.14458695e-03 2.15979442e-01
9.32759404e-01 3.15778941e-01 1.15789390e+00 -6.24983490e-01
-9.03550327e-01 -2.39817291e-01 -8.38616192e-01 4.36367616e-02
6.26980588e-02 4.01883721e-02 -4.44424480e-01 1.54885471e-01
-5.39276004e-01 4.11737412e-01 9.87057149e-01 6.18344665e-01
1.72218609e+00 -5.41730940e-01 -4.53962892e-01 5.73917806e-01
1.12131989e+00 4.97631252e-01 9.92317975e-01 1.52341694e-01
1.20061827e+00 3.21185946e-01 7.94546008e-02 -9.66411009e-02
8.60998809e-01 6.40903056e-01 7.15022266e-01 -1.27254918e-01
-4.94937956e-01 -3.88511360e-01 1.20333090e-01 1.17568469e+00
-2.73523122e-01 -7.94591963e-01 -1.32879114e+00 4.10802633e-01
-2.34543753e+00 -1.37723553e+00 -8.96601796e-01 1.43428075e+00
-2.90060878e-01 5.41431963e-01 5.09231925e-01 -1.95851937e-01
1.00237870e+00 4.06995147e-01 -8.85345101e-01 -7.22096324e-01
1.00173227e-01 1.43845260e-01 7.12061405e-01 4.38097477e-01
-8.14187646e-01 1.24034619e+00 7.10047293e+00 1.02008688e+00
-1.13099825e+00 3.01650986e-02 6.55604362e-01 1.78721189e-01
-1.37197360e-01 -3.62813503e-01 -5.15891373e-01 1.02763700e+00
1.74435270e+00 -5.91198802e-01 6.86816871e-01 6.79067373e-01
5.15913785e-01 6.37477219e-01 -6.00221634e-01 9.40683603e-01
-3.11194122e-01 -1.73628962e+00 4.88433719e-01 4.05692458e-01
6.67033494e-01 5.10645628e-01 2.33209953e-01 7.11733222e-01
8.11401248e-01 -1.25955558e+00 9.43476260e-02 1.01339769e+00
6.21180415e-01 -9.66420889e-01 7.79715598e-01 2.88732946e-01
-1.51108086e+00 -3.87939006e-01 -5.22773504e-01 -3.43747497e-01
5.50111890e-01 4.73570824e-01 -5.84178627e-01 4.35341835e-01
2.88373500e-01 1.17951369e+00 -7.21428394e-01 1.12166202e+00
4.12681364e-02 1.11403239e+00 -3.15145850e-02 -2.83883631e-01
6.07987702e-01 -3.46051276e-01 4.31484461e-01 9.51018274e-01
1.69098496e-01 2.20409110e-01 2.76971370e-01 9.85017121e-01
-2.39230201e-01 -4.43869591e-01 -6.90519094e-01 -2.28783369e-01
3.33231509e-01 7.68836737e-01 -4.49873954e-01 -3.38495970e-01
-3.99688721e-01 2.82567203e-01 5.63308954e-01 6.78338468e-01
-1.44025612e+00 -5.08803189e-01 1.07398796e+00 4.77632314e-01
5.38311183e-01 -4.53643322e-01 -8.11227486e-02 -9.05431449e-01
-1.22040309e-01 -2.48899207e-01 4.65121448e-01 -8.75049293e-01
-1.61685634e+00 7.52653778e-01 -2.91529596e-01 -1.21191561e+00
-5.29501028e-02 -8.56142819e-01 -1.20823324e+00 7.60355175e-01
-1.68785071e+00 -1.17625129e+00 -3.51574481e-01 7.85856307e-01
4.49664444e-01 -5.02037823e-01 6.67616129e-02 6.51627123e-01
-1.09904885e+00 6.04199469e-01 7.92747214e-02 4.94618714e-01
-1.61170736e-01 -8.24409962e-01 1.34406340e+00 8.11023295e-01
-1.03546627e-01 3.04737210e-01 3.93843651e-01 -8.09246480e-01
-1.20651567e+00 -1.66886675e+00 8.34709883e-01 -4.44456667e-01
1.24728966e+00 -1.02459349e-01 -1.04608405e+00 9.41964805e-01
-3.67260836e-02 2.77461648e-01 2.92331547e-01 1.67134702e-02
-2.28868425e-01 -5.35834372e-01 -6.16109431e-01 6.30591452e-01
1.44984674e+00 -4.96986628e-01 -1.21962875e-01 4.04251486e-01
9.36635315e-01 -9.82860178e-02 -7.56430447e-01 1.05190963e-01
2.65316248e-01 -5.77522337e-01 8.34338844e-01 -1.36099243e+00
2.71712333e-01 -2.69763708e-01 2.58476943e-01 -1.40894949e+00
-1.01454782e+00 -5.91850162e-01 -9.27732110e-01 8.74302328e-01
2.49305844e-01 -1.14854908e+00 8.37436318e-01 6.29130244e-01
-3.03630352e-01 -6.56867862e-01 -9.22686815e-01 -1.04164648e+00
4.37250026e-02 -7.92506695e-01 1.32064152e+00 8.09522629e-01
-4.75179940e-01 4.79975224e-01 -5.75876594e-01 1.01074964e-01
5.50529718e-01 -1.36171997e-01 1.00285065e+00 -1.61057770e+00
3.35914373e-01 -9.60032463e-01 -1.11013615e+00 -1.29126048e+00
5.47524095e-01 -9.94565368e-01 -7.87714362e-01 -1.73842311e+00
-2.05052868e-01 -3.71255457e-01 -6.28467739e-01 3.01803440e-01
-1.42518222e-01 -6.57887831e-02 6.55225068e-02 6.35863990e-02
-1.01450765e+00 8.41172993e-01 1.34883142e+00 -4.15607899e-01
1.40162930e-01 3.13611090e-01 -6.26531243e-01 3.27180326e-01
1.00491953e+00 -4.02410597e-01 -4.74936992e-01 -7.83980429e-01
-5.16605303e-02 -7.58318827e-02 4.77093071e-01 -1.12514806e+00
5.96299529e-01 -3.85535061e-01 1.13385715e-01 -8.25946152e-01
-1.48559054e-02 -1.03730452e+00 2.24682882e-01 5.85217118e-01
-3.19917649e-02 3.20865601e-01 2.86208302e-01 8.99685323e-01
-8.92102718e-02 7.07798541e-01 2.54147410e-01 2.55423933e-01
-1.16067863e+00 1.37274301e+00 -6.50426030e-01 1.24492645e-01
9.33048785e-01 -3.28621149e-01 -7.11138010e-01 -7.32203424e-01
-5.89748204e-01 7.97821701e-01 -2.78246880e-01 8.80945265e-01
4.10946339e-01 -2.04773235e+00 -6.77013695e-01 1.04546472e-01
2.69409679e-02 -4.91872013e-01 4.87290800e-01 7.48747587e-01
-6.84144676e-01 9.65788901e-01 -2.51703620e-01 -4.53624606e-01
-5.52024066e-01 9.15405273e-01 7.68303156e-01 -3.74526829e-01
-8.17149341e-01 2.26389065e-01 7.55847692e-02 -4.61057991e-01
6.25529438e-02 -1.83463588e-01 -3.76312703e-01 -5.74370205e-01
4.15712714e-01 9.23944533e-01 -1.01022281e-01 -9.11014736e-01
-2.23084122e-01 4.28324401e-01 7.15690702e-02 7.69848824e-01
1.36284590e+00 -2.15667620e-01 1.67003460e-02 2.49729753e-01
1.24285185e+00 -7.46656954e-01 -1.28050780e+00 -5.45634270e-01
-9.78747755e-02 -3.39806765e-01 1.74035862e-01 -4.36995327e-01
-1.82610214e+00 1.19967628e+00 3.54431540e-01 7.66721368e-01
8.31731975e-01 -2.57207960e-01 1.53051388e+00 4.33419645e-01
1.61466226e-01 -8.82119238e-01 -8.74039382e-02 9.40085411e-01
5.61970055e-01 -1.10264838e+00 -2.89114624e-01 -2.73027718e-01
-5.45924067e-01 1.22552800e+00 9.96692002e-01 -4.63478357e-01
1.13919222e+00 -2.42821306e-01 -1.07445605e-01 -6.66029036e-01
-1.04384720e+00 -6.22631967e-01 5.59789777e-01 8.01190019e-01
-2.06241027e-01 2.10292399e-01 1.05965421e-01 5.43984830e-01
-1.97468236e-01 9.01567712e-02 1.82863280e-01 1.85011819e-01
-4.47333604e-01 -5.03247619e-01 1.80518985e-01 6.96328104e-01
-1.52716860e-01 1.50299087e-01 -3.17165852e-01 8.43777478e-01
-2.53400832e-01 1.11371517e+00 5.31839430e-01 -8.33311141e-01
4.72836256e-01 -2.09102944e-01 -1.71957806e-01 7.25260377e-02
-1.66636154e-01 -7.13267624e-01 4.24991101e-02 -7.49288321e-01
7.96601623e-02 7.02084750e-02 -1.08960307e+00 -1.47393703e+00
-1.63421154e-01 1.82951778e-01 2.28710115e-01 1.11528742e+00
8.48482966e-01 1.22359002e+00 9.55835044e-01 -4.68147814e-01
1.22377694e-01 -6.99000895e-01 -5.12975335e-01 4.28896368e-01
4.33163315e-01 -8.89994860e-01 -3.61494943e-02 -3.67980391e-01]
|
[6.457385063171387, 2.072253465652466]
|
0ac0098b-b1d3-4004-b64d-67957c464744
|
sequence-classification-with-human-attention
| null | null |
https://aclanthology.org/K18-1030
|
https://aclanthology.org/K18-1030.pdf
|
Sequence Classification with Human Attention
|
Learning attention functions requires large volumes of data, but many NLP tasks simulate human behavior, and in this paper, we show that human attention really does provide a good inductive bias on many attention functions in NLP. Specifically, we use estimated human attention derived from eye-tracking corpora to regularize attention functions in recurrent neural networks. We show substantial improvements across a range of tasks, including sentiment analysis, grammatical error detection, and detection of abusive language.
|
['Anders S{\\o}gaard', 'Joachim Bingel', 'Maria Barrett', 'Nora Hollenstein', 'Marek Rei']
|
2018-10-01
| null | null | null |
conll-2018-10
|
['grammatical-error-detection']
|
['natural-language-processing']
|
[-4.03708458e-01 3.88189167e-01 -1.71156302e-01 -5.37437916e-01
-4.53408599e-01 -4.55056787e-01 1.80638954e-01 1.70040801e-01
-6.24582291e-01 7.39852428e-01 4.54835713e-01 -5.04177332e-01
3.40824127e-01 -3.71392041e-01 -5.34444094e-01 -1.73068017e-01
3.29589874e-01 3.91930252e-01 -3.88526618e-01 -4.30729330e-01
4.66329992e-01 6.22910559e-01 -7.01711833e-01 -1.51226759e-01
8.70295882e-01 6.08808160e-01 -1.56787887e-01 7.60271966e-01
-1.92729175e-01 1.23225331e+00 -1.01722336e+00 -8.73326063e-01
-3.70416939e-01 -4.56956118e-01 -1.10031199e+00 -3.76738518e-01
2.33921751e-01 -1.29349753e-01 -3.23268533e-01 1.18224132e+00
5.20581007e-01 5.92498064e-01 5.71641505e-01 -9.68786418e-01
-1.71328938e+00 6.61717057e-01 -4.13078398e-01 8.23483944e-01
3.60770494e-01 4.50856537e-01 1.41862607e+00 -1.05072951e+00
4.37212378e-01 1.57601368e+00 4.90123749e-01 1.01143003e+00
-9.86568928e-01 -6.80201948e-01 1.36726171e-01 -9.15409848e-02
-9.58337784e-01 -3.37217122e-01 5.04484713e-01 -4.72693205e-01
1.38523912e+00 1.27130419e-01 3.47216278e-01 1.44207418e+00
1.82605788e-01 1.11111546e+00 5.08086383e-01 -3.29191178e-01
-1.83413595e-01 2.17821375e-01 9.35781181e-01 6.52318299e-01
1.52480111e-01 -1.31852165e-01 -4.31627572e-01 -1.74341187e-01
5.22386849e-01 -1.38676157e-02 -4.50667113e-01 6.90892696e-01
-7.51886368e-01 1.35940421e+00 6.87794387e-01 3.59904289e-01
-3.64081055e-01 2.80804932e-01 3.85859907e-01 2.38626957e-01
8.20643365e-01 1.05850041e+00 -5.81901133e-01 -1.89971641e-01
-3.80547315e-01 -1.86942704e-02 7.59123921e-01 7.83651531e-01
6.60564065e-01 3.42691571e-01 -4.82356995e-01 1.02541184e+00
1.81863233e-01 6.73164845e-01 7.92278826e-01 -9.74027991e-01
4.03283805e-01 4.54989880e-01 6.16118833e-02 -1.06890178e+00
-6.50665998e-01 -2.96912253e-01 -5.94463229e-01 -3.05146605e-01
3.00542653e-01 -4.89724845e-01 -8.69598031e-01 1.77250838e+00
-1.34508654e-01 -5.84858835e-01 -1.63681567e-01 7.30902910e-01
9.22549188e-01 7.35696614e-01 3.45367849e-01 -3.02623570e-01
1.26499009e+00 -1.05100143e+00 -1.12462413e+00 -6.45654261e-01
6.60048783e-01 -3.53093714e-01 1.71270978e+00 3.59227061e-01
-1.28809750e+00 -3.55126917e-01 -4.16127980e-01 -6.33000731e-01
-3.25165957e-01 -1.81166008e-01 6.71507657e-01 4.77614850e-01
-8.03533494e-01 5.23297608e-01 -5.42927265e-01 -4.47599024e-01
6.99434519e-01 2.27846950e-01 -3.48288901e-02 4.49080259e-01
-1.36308277e+00 1.17833829e+00 -1.02899134e-01 1.74876794e-01
-7.26846039e-01 -3.67476851e-01 -1.08134389e+00 3.59403461e-01
1.07674308e-01 -5.33363521e-01 1.48655427e+00 -1.17174482e+00
-1.39694428e+00 1.05997491e+00 -4.46485370e-01 -6.69853866e-01
4.21083197e-02 -8.56524050e-01 -1.91777050e-01 -9.88366231e-02
-3.01362015e-02 6.91611648e-01 7.71067977e-01 -5.02638221e-01
-8.42308179e-02 -2.79921532e-01 1.66348256e-02 -3.75632532e-02
-7.30589688e-01 6.99680209e-01 -2.24466473e-01 -6.84054077e-01
-8.20433319e-01 -7.03865290e-01 -2.09027976e-01 -2.12963551e-01
-5.49526036e-01 -7.91446805e-01 4.21383023e-01 -8.05519283e-01
1.41159892e+00 -2.20105290e+00 1.87076196e-01 2.92443559e-02
3.84731740e-01 3.80551010e-01 -9.17696878e-02 -1.96437299e-01
-6.16596341e-02 6.12699866e-01 -1.01045251e-01 -5.02160490e-01
1.47983938e-01 2.72055060e-01 -5.38012505e-01 3.92234802e-01
3.69667739e-01 1.39300323e+00 -1.03001368e+00 -3.87911797e-01
-1.21838808e-01 4.39081252e-01 -6.92789853e-01 3.70983988e-01
-7.99110681e-02 2.72139519e-01 -2.70664304e-01 6.59451067e-01
-1.25792921e-01 -5.36661744e-01 -3.78586918e-01 4.33561087e-01
2.77411669e-01 5.96432447e-01 1.15255511e-03 1.22364104e+00
-4.51784521e-01 1.14795089e+00 -1.00061581e-01 -7.23163426e-01
9.88911808e-01 2.12830037e-01 -2.47315973e-01 -6.99285686e-01
6.13849759e-01 -3.20672959e-01 3.06184798e-01 -7.27230132e-01
6.46247566e-01 -2.27771655e-01 -1.84712663e-01 8.11529934e-01
2.33530849e-01 -4.79980297e-02 3.59296836e-02 2.96382368e-01
9.05711174e-01 -5.16612411e-01 4.64366138e-01 -2.95054317e-01
4.57834005e-01 -2.29808301e-01 5.20342350e-01 8.12803447e-01
-5.87836206e-01 3.52763355e-01 1.04673982e+00 -4.01511788e-01
-7.66014755e-01 -5.00917375e-01 4.43981476e-02 1.88143063e+00
-4.96884733e-01 -1.84432998e-01 -9.18777525e-01 -8.96367967e-01
-4.26222235e-02 1.29548740e+00 -1.03157091e+00 -4.39021528e-01
-5.30349374e-01 -7.33601213e-01 5.72773993e-01 7.97548413e-01
-1.22285783e-01 -1.93378353e+00 -4.04678047e-01 -2.93771535e-01
-1.21734748e-02 -9.66400146e-01 -7.93503940e-01 4.23079610e-01
-6.17834687e-01 -1.04345500e+00 -6.66266382e-01 -6.98740304e-01
8.17669451e-01 -2.40895554e-01 1.40361953e+00 4.90450948e-01
-1.85183123e-01 1.39202401e-01 -2.89606273e-01 -7.18020856e-01
-2.73687869e-01 3.58737320e-01 4.69131246e-02 -3.66449386e-01
9.05391157e-01 -9.34353471e-02 -4.95327078e-03 1.63872838e-02
-2.63081223e-01 -6.67572260e-01 1.57809496e-01 8.49693716e-01
-1.82273388e-01 -7.01444328e-01 5.04620850e-01 -1.28998554e+00
1.24619341e+00 -6.07289255e-01 -3.99443924e-01 1.05902262e-01
-4.07151431e-01 4.69791517e-02 6.43621564e-01 -4.69285280e-01
-6.81696594e-01 -4.62223589e-01 -4.34346527e-01 -4.85498071e-01
-8.67708027e-02 3.93197596e-01 1.78860739e-01 2.19089821e-01
9.72011507e-01 -2.74513066e-01 -1.63179904e-01 -5.61463833e-01
2.44122475e-01 6.05856717e-01 3.42504233e-01 -2.86509514e-01
3.79908621e-01 -5.36063649e-02 -7.75199175e-01 -7.67044544e-01
-1.41690981e+00 -9.79557484e-02 -3.37071031e-01 8.69896412e-02
8.25380206e-01 -5.15440524e-01 -1.23205519e+00 3.48228477e-02
-1.50744069e+00 -5.70573270e-01 -3.72093767e-01 3.55854809e-01
-2.57383913e-01 -1.05721150e-02 -1.11077392e+00 -9.71551239e-01
-9.53086257e-01 -9.00469720e-01 7.02628493e-01 2.92448729e-01
-8.78983617e-01 -1.14066517e+00 1.11329399e-01 1.15645520e-01
3.87056649e-01 -2.61998385e-01 9.54159319e-01 -1.00991023e+00
7.85570070e-02 5.00232317e-02 -2.63820350e-01 5.91117859e-01
-1.64936036e-01 2.62210518e-01 -9.67668355e-01 -2.22157338e-03
-8.63856915e-03 -7.27366865e-01 9.14699435e-01 6.11354232e-01
1.46067238e+00 -4.17564631e-01 -1.41088143e-01 5.45034528e-01
5.25115371e-01 3.28224227e-02 6.47107244e-01 6.80119395e-02
8.15463006e-01 6.84962392e-01 2.61534810e-01 3.55776310e-01
4.02537994e-02 5.97087406e-02 1.28113180e-01 6.36258572e-02
1.98408455e-01 -1.73041359e-01 6.09873176e-01 5.89049041e-01
1.66937739e-01 -6.27774775e-01 -1.10945725e+00 7.84152269e-01
-1.61835837e+00 -7.30029345e-01 1.25954049e-02 1.75672245e+00
9.09445047e-01 3.42326462e-01 1.41859040e-01 -3.14476311e-01
9.47663903e-01 8.70677978e-02 -6.19469106e-01 -1.10919440e+00
2.80968323e-02 4.49382752e-01 2.15169489e-01 8.97144437e-01
-7.55720019e-01 1.35320318e+00 8.23733616e+00 2.70767301e-01
-1.14034057e+00 2.70174712e-01 8.09448898e-01 -3.48110229e-01
-3.33012909e-01 -5.44150949e-01 -9.34697568e-01 4.66878414e-01
1.21583569e+00 -3.86548102e-01 4.09496486e-01 1.00753427e+00
4.66894135e-02 3.10130060e-01 -1.01231396e+00 8.84974301e-01
3.52098197e-01 -8.03903818e-01 -4.65284660e-02 -7.95473456e-02
3.91535372e-01 2.71296442e-01 2.61494994e-01 7.08258450e-01
8.70173633e-01 -1.45057893e+00 4.49453771e-01 4.61112559e-01
5.16121805e-01 -9.47183609e-01 7.01363325e-01 4.11839068e-01
-9.42576826e-02 -1.29944727e-01 -6.06392860e-01 -3.78958642e-01
2.76030213e-01 3.48610938e-01 -6.43859267e-01 -6.92743778e-01
8.15687120e-01 8.44241560e-01 -5.59902549e-01 4.15792078e-01
-1.04836130e+00 8.38433862e-01 -1.05172865e-01 -6.37331188e-01
3.11593980e-01 2.64594465e-01 4.08214450e-01 1.34236789e+00
-9.83603969e-02 2.01777309e-01 -3.48827153e-01 1.11686838e+00
-7.41816044e-01 2.06543341e-01 -7.31534302e-01 -6.33976340e-01
1.55637488e-01 1.34223521e+00 -3.80752474e-01 -3.22648376e-01
-3.67098540e-01 9.05200660e-01 1.04024613e+00 4.58411515e-01
-1.08527493e+00 -6.36266232e-01 8.65384519e-01 -1.09200485e-01
2.48727366e-01 -7.41783977e-02 -3.66588473e-01 -1.22435272e+00
-5.38078308e-01 -6.89315319e-01 4.33942765e-01 -1.10801733e+00
-1.36135638e+00 6.19203448e-01 -5.23431182e-01 -2.36687198e-01
-3.83100450e-01 -7.98432946e-01 -8.64064455e-01 8.77547324e-01
-1.14266658e+00 -5.29624045e-01 -2.88944994e-03 5.89085639e-01
7.10480154e-01 -8.34628344e-02 7.01133370e-01 2.61464566e-01
-1.05882668e+00 9.24776554e-01 -5.11690378e-01 6.27111018e-01
7.09515154e-01 -1.32929730e+00 9.25380826e-01 6.00983977e-01
3.79912823e-01 1.04719484e+00 7.25759387e-01 -5.20160496e-01
-7.84702361e-01 -1.00134552e+00 1.36429870e+00 -8.64593267e-01
9.46407676e-01 -3.02899688e-01 -1.15332294e+00 1.36409819e+00
5.27313232e-01 3.37302033e-03 7.68603325e-01 8.64284635e-01
-3.88009280e-01 5.43687701e-01 -8.60444486e-01 4.54287142e-01
8.92410219e-01 -8.30550790e-01 -9.73386884e-01 6.08947635e-01
1.03889585e+00 -4.01213199e-01 -4.14291620e-01 -7.21872672e-02
1.00781269e-01 -3.42735320e-01 6.94777429e-01 -1.34741902e+00
7.44332075e-01 4.70526338e-01 3.14809442e-01 -1.44617736e+00
-7.04162419e-01 -9.14906800e-01 -3.46763730e-01 1.20625556e+00
7.28115797e-01 -6.96150005e-01 6.10679030e-01 1.11424589e+00
-1.10361919e-01 -5.38640976e-01 -3.18624467e-01 -4.07487452e-01
4.92832154e-01 -2.45702758e-01 1.19988471e-01 1.25044680e+00
4.80194837e-01 1.02995694e+00 -4.64665234e-01 -2.90762633e-01
1.29840568e-01 -3.44829738e-01 5.74720562e-01 -1.07243490e+00
-1.65770024e-01 -5.51744998e-01 1.37374371e-01 -1.06389904e+00
9.16188896e-01 -9.09808815e-01 4.91199270e-02 -1.11861253e+00
2.50530809e-01 2.34908089e-01 -2.73139954e-01 8.41617942e-01
-6.07686698e-01 2.94516832e-01 1.35414541e-01 -5.02237864e-02
-8.62389624e-01 7.28013396e-01 1.06332576e+00 -5.36115915e-02
-9.06031653e-02 -3.73324603e-01 -1.31717622e+00 9.92227852e-01
7.97551870e-01 -6.02162004e-01 3.29482555e-02 -7.70293355e-01
6.28538072e-01 -4.11719531e-01 1.71772212e-01 -2.84629166e-01
1.82788953e-01 -1.77512094e-01 3.17024231e-01 -2.11966261e-01
1.45050213e-01 -3.76768857e-01 -9.20222223e-01 3.52762699e-01
-7.06307769e-01 2.37587199e-01 3.06551397e-01 2.73963898e-01
-1.36938706e-01 -4.95174050e-01 7.89502144e-01 -1.84585124e-01
-3.27697933e-01 1.89431712e-01 -4.33093160e-01 7.36427248e-01
5.02786279e-01 3.59569907e-01 -5.32353282e-01 -6.46265090e-01
-9.27160800e-01 5.12790978e-01 1.98205680e-01 5.00045836e-01
5.08132219e-01 -1.07278478e+00 -5.64092636e-01 8.85526612e-02
-1.38605818e-01 -2.78822899e-01 -1.81420371e-01 7.65082598e-01
-3.43123049e-01 4.26105142e-01 5.74721508e-02 -2.05338016e-01
-1.17504990e+00 8.17247212e-01 2.91985005e-01 2.59098113e-02
-3.24625760e-01 1.44032502e+00 4.43635583e-01 -3.86751413e-01
3.77733797e-01 -4.00648534e-01 -4.77340788e-01 9.96462107e-02
8.48342717e-01 3.60514224e-02 -2.90308237e-01 -4.80175257e-01
-2.70345330e-01 2.82944083e-01 -5.01743615e-01 9.39418301e-02
1.25605166e+00 -3.16889994e-02 -1.91426165e-02 6.57185376e-01
1.21682906e+00 -8.77502114e-02 -9.32486117e-01 -1.16873667e-01
4.48798202e-03 -3.69819850e-01 1.08056016e-01 -7.92935431e-01
-1.07120109e+00 1.31925642e+00 -2.16101691e-01 2.87795395e-01
6.27208292e-01 2.39231870e-01 9.98603761e-01 7.30045378e-01
-1.91696227e-01 -1.24490499e+00 1.82570711e-01 1.10303593e+00
1.11396801e+00 -1.40162218e+00 -5.05277216e-02 -2.43671965e-02
-1.02852368e+00 7.80472517e-01 9.30644870e-01 -2.77641058e-01
5.60279071e-01 2.07892045e-01 -2.13089008e-02 -6.08750582e-01
-9.23420429e-01 -1.15268618e-01 1.40615806e-01 3.06557059e-01
9.11491811e-01 -2.21760213e-01 -1.62449703e-01 7.86192060e-01
-3.75314116e-01 -1.64509058e-01 3.80164087e-01 5.03939211e-01
-6.21016622e-01 -4.24956948e-01 -2.64622450e-01 3.61480296e-01
-9.23287928e-01 -5.76884747e-01 -8.02082837e-01 7.31829345e-01
-4.81314451e-01 5.34692585e-01 5.16252637e-01 -8.41146521e-03
2.72564113e-01 2.44656354e-01 2.92568117e-01 -8.87825429e-01
-1.25405490e+00 -3.76814395e-01 1.08537920e-01 -6.24070942e-01
1.21918648e-01 -4.60481703e-01 -1.25622356e+00 -5.80926955e-01
-3.49052101e-01 2.22510755e-01 2.73095608e-01 8.80029023e-01
2.50649631e-01 5.10700941e-01 2.78363556e-01 -4.94023383e-01
-4.05878872e-01 -1.38034010e+00 -4.21084732e-01 6.68045104e-01
5.12376189e-01 -5.00948906e-01 -8.75174403e-01 -1.79815903e-01]
|
[10.657845497131348, 8.53451919555664]
|
ff8333ac-6f96-414f-bf49-14fb5eb39f3e
|
nonnegative-dictionary-learning-in-the
| null | null |
http://papers.nips.cc/paper/4273-nonnegative-dictionary-learning-in-the-exponential-noise-model-for-adaptive-music-signal-representation
|
http://papers.nips.cc/paper/4273-nonnegative-dictionary-learning-in-the-exponential-noise-model-for-adaptive-music-signal-representation.pdf
|
Nonnegative dictionary learning in the exponential noise model for adaptive music signal representation
|
In this paper we describe a maximum likelihood likelihood approach for dictionary learning in the multiplicative exponential noise model. This model is prevalent in audio signal processing where it underlies a generative composite model of the power spectrogram. Maximum joint likelihood estimation of the dictionary and expansion coefficients leads to a nonnegative matrix factorization problem where the Itakura-Saito divergence is used. The optimality of this approach is in question because the number of parameters (which include the expansion coefficients) grows with the number of observations. In this paper we describe a variational procedure for optimization of the marginal likelihood, i.e., the likelihood of the dictionary where the activation coefficients have been integrated out (given a specific prior). We compare the output of both maximum joint likelihood estimation (i.e., standard Itakura-Saito NMF) and maximum marginal likelihood estimation (MMLE) on real and synthetical datasets. The MMLE approach is shown to embed automatic model order selection, akin to automatic relevance determination.
|
['Cédric Févotte', 'Onur Dikmen']
|
2011-12-01
| null | null | null |
neurips-2011-12
|
['audio-signal-processing']
|
['audio']
|
[ 3.93950582e-01 -7.58576617e-02 2.10210219e-01 -1.02878951e-01
-9.75297153e-01 -4.95177090e-01 4.00676757e-01 -1.48882084e-02
-5.98789990e-01 6.45367086e-01 2.34247580e-01 -1.30256861e-01
-3.83039623e-01 -4.04648989e-01 -3.68709058e-01 -1.01356030e+00
5.22745512e-02 4.34218496e-01 -5.42633124e-02 -1.64539680e-01
1.54260874e-01 1.50634393e-01 -1.48211074e+00 -1.59368783e-01
6.36187851e-01 7.17718303e-01 4.51306880e-01 1.03911328e+00
4.15410966e-01 1.70540318e-01 -5.10272563e-01 -1.54215023e-01
1.02153018e-01 -5.22475541e-01 -3.90719652e-01 2.42903352e-01
1.84009597e-01 1.07342210e-02 1.43538058e-01 1.12362599e+00
5.52635670e-01 3.14816117e-01 9.12365675e-01 -9.16497886e-01
-6.31164834e-02 4.65837479e-01 -3.27273816e-01 4.02925760e-01
3.44463021e-01 -2.47757941e-01 1.32646382e+00 -1.29892969e+00
4.68151152e-01 1.11444223e+00 6.75842881e-01 -6.63143545e-02
-1.59549332e+00 -3.07306319e-01 -2.42677182e-01 1.35945510e-02
-1.77480972e+00 -6.58547044e-01 7.67567337e-01 -6.87294424e-01
8.19575548e-01 2.36397624e-01 6.81286991e-01 8.27474356e-01
1.20868884e-01 7.08636403e-01 8.68544757e-01 -1.04696083e+00
4.74959344e-01 2.12181315e-01 -4.55404148e-02 4.34010565e-01
1.19388953e-01 2.05384821e-01 -8.10827613e-01 -7.04185426e-01
8.93477857e-01 -7.05207050e-01 -2.21646369e-01 -3.47543359e-01
-9.48752284e-01 9.99584079e-01 -5.64649165e-01 3.68891299e-01
-5.26895821e-01 3.41502339e-01 1.95776030e-01 3.08265269e-01
3.08925569e-01 2.41442084e-01 -3.07598293e-01 -2.51504838e-01
-1.22158778e+00 3.10119927e-01 9.30684268e-01 7.28454769e-01
6.49007022e-01 3.80197406e-01 8.19442496e-02 1.12430799e+00
8.46813440e-01 7.12702215e-01 1.82553291e-01 -1.14200318e+00
7.18775392e-02 -2.23762140e-01 1.14899457e-01 -9.19973731e-01
-2.36447275e-01 -5.36643267e-01 -6.09016061e-01 1.33760437e-01
5.11539996e-01 -1.67978436e-01 -6.07060790e-01 1.93588877e+00
2.86056459e-01 4.03538883e-01 -1.09465718e-01 7.63812006e-01
2.51736820e-01 7.40004301e-01 -1.01139657e-01 -8.47450137e-01
1.13590455e+00 -2.88439035e-01 -1.15804040e+00 -1.22565344e-01
2.61903793e-01 -1.16864252e+00 6.81591988e-01 9.21469867e-01
-1.15298927e+00 -5.86335957e-01 -1.07341945e+00 2.58793235e-01
2.20583111e-01 3.33427697e-01 4.20367479e-01 7.57615209e-01
-1.00259936e+00 4.56754386e-01 -9.01560664e-01 -6.85192198e-02
-3.56787801e-01 5.77373326e-01 -1.63159847e-01 5.22205293e-01
-1.25290000e+00 8.32834184e-01 5.19730806e-01 1.18563741e-01
-9.09641266e-01 -4.27608252e-01 -7.46302843e-01 -1.25962734e-01
1.91186190e-01 -6.37282193e-01 1.30085552e+00 -9.02573526e-01
-1.53654456e+00 6.15619361e-01 -4.26411122e-01 -4.10115659e-01
9.39472318e-02 -3.27098876e-01 -2.15546429e-01 2.06947625e-01
-1.74889505e-01 4.18409497e-01 1.48515201e+00 -1.14219344e+00
-3.22726309e-01 2.46922988e-02 -3.86258364e-01 2.01632753e-01
-1.41893685e-01 -1.23488223e-02 -3.06275249e-01 -1.11402130e+00
5.13952136e-01 -1.03615654e+00 -3.16886723e-01 -3.31775486e-01
-1.67381793e-01 -1.16282232e-01 3.53922367e-01 -9.18805957e-01
1.72096717e+00 -2.40069318e+00 5.25810182e-01 6.04026675e-01
-1.70436651e-01 -2.33013928e-01 9.34259966e-02 5.77595472e-01
-1.99822649e-01 -2.45176464e-01 -4.42698628e-01 -4.67949450e-01
5.04798181e-02 3.21961671e-01 -4.47688341e-01 5.83145499e-01
1.09312097e-02 2.75713980e-01 -7.16127515e-01 -5.46614766e-01
2.02301785e-01 7.19787657e-01 -9.02733922e-01 1.00412235e-01
4.33415314e-03 2.57575691e-01 -9.83977988e-02 2.88709491e-01
2.93398321e-01 5.38675450e-02 3.71742338e-01 -4.39066291e-01
-1.93780452e-01 2.85853088e-01 -1.92705667e+00 1.62780690e+00
-3.01471114e-01 6.91986322e-01 3.25844526e-01 -8.18442225e-01
8.66947472e-01 7.47502863e-01 5.72493970e-01 9.80660841e-02
1.88823417e-01 5.61242938e-01 9.43041146e-02 -2.78607160e-01
5.31703889e-01 -5.36357462e-01 9.28200968e-03 4.08069521e-01
5.88247418e-01 -3.55562121e-01 4.02023822e-01 3.32159460e-01
5.49408555e-01 1.05807364e-01 8.21193695e-01 -5.43040276e-01
5.40575147e-01 -4.40632313e-01 3.50368977e-01 7.46380985e-01
1.92133516e-01 7.21770883e-01 2.96813846e-01 1.44725129e-01
-1.04129744e+00 -1.13727474e+00 -4.07585263e-01 8.67992520e-01
-3.35447937e-01 -7.10195363e-01 -7.93673158e-01 1.47251248e-01
-3.40493113e-01 7.98448145e-01 -3.36491764e-01 -7.88833350e-02
-5.91253042e-01 -9.52591181e-01 3.28780532e-01 1.08196288e-01
-2.75164008e-01 -8.07433903e-01 -4.78391588e-01 4.84033227e-01
-3.80307257e-01 -8.76168549e-01 -4.30767268e-01 5.26371717e-01
-1.03318107e+00 -6.09381676e-01 -4.55360234e-01 -4.71895665e-01
4.08586413e-01 -2.68833339e-01 9.33802366e-01 -2.88694143e-01
-1.47915572e-01 7.27153182e-01 -2.51146942e-01 -3.44868660e-01
-7.13579297e-01 -2.92240858e-01 3.62073481e-01 1.77355602e-01
1.66629985e-01 -9.66388285e-01 -2.34585240e-01 4.54958156e-02
-9.83158350e-01 -2.75559574e-01 5.49171627e-01 9.81832683e-01
9.18458521e-01 1.62634254e-01 3.90608370e-01 -5.27300298e-01
9.07073140e-01 -1.96894586e-01 -6.93372071e-01 4.44305427e-02
-5.61004460e-01 1.61102071e-01 2.13820502e-01 -7.05251217e-01
-8.08018446e-01 2.49717727e-01 -2.28998989e-01 -4.80953693e-01
8.58519226e-02 8.06193948e-01 -6.74485043e-02 7.75167197e-02
7.45619893e-01 1.43387184e-01 -2.44669601e-01 -4.87450212e-01
3.25312525e-01 6.28872097e-01 5.39838850e-01 -5.76724589e-01
5.90521872e-01 2.53526121e-01 -2.28956230e-02 -1.11372352e+00
-6.14066660e-01 -6.74296498e-01 -7.33520508e-01 -4.00132865e-01
7.77564347e-01 -8.96163881e-01 -4.26033974e-01 2.20093042e-01
-1.23393154e+00 1.07078873e-01 -6.14765227e-01 9.49776590e-01
-8.53185654e-01 5.19884884e-01 -4.20161277e-01 -1.34521747e+00
-1.85371209e-02 -1.02710164e+00 1.12042582e+00 -1.90181985e-01
-7.06528902e-01 -1.19987893e+00 3.69178653e-01 -2.50730608e-02
-4.49782126e-02 -8.37296620e-02 8.27077687e-01 -5.07866859e-01
-3.05583030e-01 -1.01966448e-01 5.01080751e-01 5.29619873e-01
-1.54108480e-01 2.98746638e-02 -1.00772393e+00 -4.44574535e-01
5.32161713e-01 1.45926803e-01 7.17124820e-01 8.50295186e-01
3.25020254e-01 -2.21517235e-01 9.35655013e-02 2.62700081e-01
1.42362559e+00 2.64376789e-01 3.45864594e-01 -4.58382182e-02
2.91575223e-01 3.26559961e-01 7.08628774e-01 8.78651321e-01
-1.90019518e-01 8.84136915e-01 9.13685858e-02 3.77355367e-01
3.69822942e-02 -1.70723230e-01 6.02826834e-01 1.44960308e+00
2.13748161e-02 -2.54651248e-01 -7.29539394e-01 7.05573976e-01
-1.70468605e+00 -9.45764780e-01 -2.05646649e-01 2.45831704e+00
9.75126266e-01 2.26021871e-01 9.39286947e-02 4.13461268e-01
5.72426856e-01 2.09579654e-02 -1.87295869e-01 -3.51693690e-01
-1.79734051e-01 3.41932505e-01 2.53340751e-01 1.11054921e+00
-1.02462316e+00 4.62395132e-01 7.32383919e+00 1.14406443e+00
-5.59387684e-01 3.00072640e-01 1.98622029e-02 6.49129301e-02
-3.28096539e-01 3.00429225e-01 -8.91690433e-01 3.79969895e-01
1.20705390e+00 1.43339895e-02 5.92969835e-01 4.88133550e-01
4.55673218e-01 -5.00972748e-01 -8.54738474e-01 1.11562777e+00
1.25379547e-01 -7.84675121e-01 -1.30666465e-01 2.47378618e-01
5.97250044e-01 -2.74511158e-01 7.26466477e-02 -9.56455916e-02
-9.55094397e-02 -6.89640760e-01 9.77731347e-01 5.76946437e-01
5.01330197e-01 -6.95840955e-01 4.07908350e-01 5.61955929e-01
-1.02680230e+00 4.67231981e-02 -3.00503135e-01 7.21758530e-02
5.90617776e-01 1.04957926e+00 -9.14098263e-01 2.05118403e-01
2.40481213e-01 3.96421909e-01 -5.65408915e-02 1.00463235e+00
-3.40844274e-01 9.66374874e-01 -8.18949699e-01 3.02412093e-01
-1.33332968e-01 -7.04297245e-01 1.22693467e+00 1.29230177e+00
5.41073322e-01 1.31716356e-01 1.20620996e-01 8.27127457e-01
3.83724838e-01 2.65630722e-01 -1.89330205e-01 -1.07400030e-01
4.97720450e-01 1.12138760e+00 -7.65982926e-01 -5.06383032e-02
-5.86234853e-02 6.52874410e-01 -1.62067056e-01 3.55457962e-01
-5.25624752e-01 -7.86677077e-02 2.07553193e-01 2.71074846e-02
4.86990482e-01 -3.78331661e-01 -1.77633822e-01 -9.76648450e-01
-1.24268904e-01 -9.39468145e-01 3.21811348e-01 -6.49294734e-01
-9.78640258e-01 4.51010019e-01 4.57858264e-01 -1.28454983e+00
-7.74359882e-01 -4.88351852e-01 -2.86497980e-01 1.03801167e+00
-9.13538456e-01 -6.47594333e-01 5.73857486e-01 4.00543958e-01
5.08321226e-01 -1.08986661e-01 9.88799572e-01 4.54893351e-01
-1.85899466e-01 2.96424001e-01 2.76630014e-01 -3.31017911e-01
5.41677892e-01 -1.22757626e+00 -1.73554793e-02 9.36361909e-01
5.31856775e-01 8.95557940e-01 1.31535316e+00 -6.11368120e-01
-9.23221469e-01 -3.08523029e-01 1.13520694e+00 -2.89526492e-01
8.04589629e-01 -3.51720780e-01 -7.79511511e-01 5.00110745e-01
1.32849738e-01 -4.25513417e-01 1.02517247e+00 1.15535416e-01
1.12357754e-02 1.33190766e-01 -6.34137392e-01 3.06933075e-01
4.57565993e-01 -7.52359509e-01 -5.83396256e-01 2.69232839e-01
3.48393291e-01 -2.23272234e-01 -8.82718384e-01 2.25498483e-01
6.66458726e-01 -8.09071124e-01 8.73819232e-01 -1.79797009e-01
-1.21278428e-01 -5.67000508e-01 -5.64821243e-01 -1.04368579e+00
-3.74542683e-01 -1.08923376e+00 -3.83584559e-01 1.01682067e+00
5.18273711e-01 -1.85394704e-01 4.16647315e-01 1.28767684e-01
1.14609231e-03 -6.10308528e-01 -1.40006089e+00 -6.11958086e-01
-3.34545791e-01 -9.24058318e-01 -3.13638330e-01 7.63813853e-01
-9.32720453e-02 6.25871599e-01 -6.92739606e-01 2.94041544e-01
6.76692128e-01 -1.22927688e-01 3.51444334e-01 -1.25523913e+00
-1.02003407e+00 -7.40270838e-02 -2.31095627e-01 -1.25418043e+00
-1.23084940e-01 -7.40787029e-01 1.64222986e-01 -1.05529964e+00
-3.86283509e-02 -2.04492450e-01 -1.80291131e-01 -1.98739897e-02
3.21000954e-03 7.20926598e-02 2.46791989e-01 3.40867251e-01
-2.50374347e-01 4.60887164e-01 9.03864861e-01 1.30122170e-01
-4.61365342e-01 1.96726099e-01 -3.11965764e-01 9.40027773e-01
4.88372505e-01 -9.02942419e-01 -4.98828173e-01 -1.16347611e-01
7.04650104e-01 3.31559598e-01 3.10265690e-01 -8.38848650e-01
3.22655678e-01 1.53245181e-01 -4.68023401e-03 -5.25357962e-01
7.78034568e-01 -6.92667425e-01 6.53124154e-01 1.78854883e-01
-4.00869429e-01 -7.92950243e-02 1.90801304e-02 7.70658791e-01
-4.20043200e-01 -8.89876425e-01 7.89746046e-01 9.52468961e-02
-3.52661461e-01 -2.28736505e-01 -7.97954857e-01 -1.12583712e-01
4.40528214e-01 -2.93679655e-01 5.92162490e-01 -7.31167853e-01
-1.29411554e+00 -3.33183944e-01 2.56513394e-02 7.66355917e-03
6.63401306e-01 -1.34592116e+00 -8.04266930e-01 3.54472727e-01
-3.34174454e-01 -2.89343685e-01 1.26447663e-01 1.21582818e+00
-1.29559904e-01 2.03377873e-01 3.58127892e-01 -6.98325098e-01
-1.28248382e+00 9.57155526e-02 3.00803602e-01 -4.34947163e-01
-8.26336667e-02 1.03467751e+00 3.49907316e-02 -2.06609741e-01
1.35914728e-01 -7.08580390e-02 -2.18722209e-01 3.44731033e-01
3.89393806e-01 4.27513093e-01 7.50467703e-02 -8.85752797e-01
-1.98774725e-01 6.12119019e-01 2.72268206e-01 -9.30121899e-01
1.09383678e+00 -1.46600291e-01 -3.61289978e-01 8.86719048e-01
1.10093594e+00 4.29429322e-01 -8.73172224e-01 -6.48565590e-02
2.61140405e-03 -2.85548925e-01 3.92729878e-01 -4.61496413e-01
-4.71304893e-01 9.07419086e-01 7.21426129e-01 2.24762112e-01
1.13825524e+00 -2.73502648e-01 3.94044131e-01 2.26619080e-01
2.82355458e-01 -1.31025875e+00 -1.29547879e-01 4.48086232e-01
9.39487875e-01 -8.16186309e-01 4.25855964e-01 -4.87186790e-01
-3.54263306e-01 1.26650298e+00 -6.46176934e-02 -1.99155658e-01
1.05984592e+00 4.14450586e-01 -9.48116928e-02 -6.84874579e-02
-7.76198566e-01 -1.64497912e-01 5.25628030e-01 4.47176725e-01
4.60656911e-01 2.76831444e-02 -5.17107666e-01 4.75752801e-01
-3.84286821e-01 -3.69294971e-01 4.18647468e-01 6.35323346e-01
-5.84456027e-01 -1.38980436e+00 -6.26746237e-01 1.68659136e-01
-5.04993439e-01 -4.04279113e-01 -2.68118352e-01 4.34109747e-01
2.08329290e-01 1.14701664e+00 -1.37471080e-01 -1.98639795e-01
1.68407425e-01 2.18349397e-01 6.25304282e-01 -6.63686633e-01
-3.98709655e-01 1.12668967e+00 -3.83852609e-03 1.60347354e-02
-6.88175023e-01 -1.14629579e+00 -1.00631809e+00 3.34001899e-01
-9.26595509e-01 4.90895748e-01 8.18716228e-01 8.98962975e-01
-2.01207802e-01 3.05397391e-01 4.16775674e-01 -6.94573045e-01
-8.08964312e-01 -9.97612059e-01 -8.33230257e-01 2.33556982e-03
2.18136743e-01 -5.00213385e-01 -7.43987262e-01 4.49889481e-01]
|
[15.441733360290527, 5.584949493408203]
|
8a519d65-6d72-4980-91f6-2d03cd3f96b1
|
3d-multi-object-tracking-with-differentiable
|
2206.13785
| null |
https://arxiv.org/abs/2206.13785v1
|
https://arxiv.org/pdf/2206.13785v1.pdf
|
3D Multi-Object Tracking with Differentiable Pose Estimation
|
We propose a novel approach for joint 3D multi-object tracking and reconstruction from RGB-D sequences in indoor environments. To this end, we detect and reconstruct objects in each frame while predicting dense correspondences mappings into a normalized object space. We leverage those correspondences to inform a graph neural network to solve for the optimal, temporally-consistent 7-DoF pose trajectories of all objects. The novelty of our method is two-fold: first, we propose a new graph-based approach for differentiable pose estimation over time to learn optimal pose trajectories; second, we present a joint formulation of reconstruction and pose estimation along the time axis for robust and geometrically consistent multi-object tracking. In order to validate our approach, we introduce a new synthetic dataset comprising 2381 unique indoor sequences with a total of 60k rendered RGB-D images for multi-object tracking with moving objects and camera positions derived from the synthetic 3D-FRONT dataset. We demonstrate that our method improves the accumulated MOTA score for all test sequences by 24.8% over existing state-of-the-art methods. In several ablations on synthetic and real-world sequences, we show that our graph-based, fully end-to-end-learnable approach yields a significant boost in tracking performance.
|
['Matthias Nießner', 'Norman Müller', 'Zeju Qiu', 'Dominik Schmauser']
|
2022-06-28
| null | null | null | null |
['3d-multi-object-tracking']
|
['computer-vision']
|
[ 4.34858315e-02 -3.29106987e-01 1.48119375e-01 -2.43752360e-01
-9.61949348e-01 -9.02982354e-01 3.04531157e-01 -2.05700278e-01
-2.28638500e-01 4.05619711e-01 -6.32574782e-02 9.29336548e-02
-2.04541430e-01 -2.36531943e-01 -1.24181223e+00 -3.51163357e-01
-2.67402321e-01 6.49811327e-01 5.10106504e-01 1.96108297e-02
-1.73416853e-01 8.74064744e-01 -1.37189627e+00 -1.91690251e-01
4.10369098e-01 1.09312093e+00 2.69248337e-01 1.07286477e+00
5.91247022e-01 6.55755997e-01 -2.59564579e-01 -2.73247987e-01
7.13623881e-01 -1.60293967e-01 -4.04024601e-01 4.52808082e-01
1.21175301e+00 -5.81054330e-01 -6.72882557e-01 9.00303602e-01
5.49332440e-01 1.88704774e-01 3.58243465e-01 -1.35371566e+00
-3.05010110e-01 -5.49525581e-02 -6.09144866e-01 1.42870061e-02
5.53264737e-01 4.71478522e-01 7.32853889e-01 -8.77947152e-01
8.99415970e-01 1.30934846e+00 1.16135240e+00 5.97253382e-01
-1.15113711e+00 -6.45926535e-01 2.35326305e-01 -7.15515912e-02
-1.35325181e+00 -3.74646217e-01 6.93300247e-01 -5.44046044e-01
8.53120923e-01 2.43497685e-01 1.01224184e+00 1.09315205e+00
3.27954531e-01 7.18494236e-01 6.69974506e-01 -2.07238961e-02
-3.99972722e-02 -4.45573509e-01 -1.16591930e-01 1.07723868e+00
5.28594851e-01 4.20565665e-01 -4.52742308e-01 -1.01136915e-01
9.79125977e-01 3.19464862e-01 -9.09848511e-02 -1.11860502e+00
-1.69311810e+00 2.94838101e-01 6.30213618e-01 -1.87898755e-01
-3.30261230e-01 8.23036969e-01 -1.46523695e-02 -1.25998780e-01
3.54238957e-01 1.27651215e-01 -4.79458809e-01 -9.13037285e-02
-6.82437539e-01 7.19988644e-01 5.34465373e-01 1.37253582e+00
5.33820748e-01 6.62870333e-02 -2.41343707e-01 1.60975978e-01
7.26542294e-01 1.21363938e+00 -1.84180409e-01 -1.46715963e+00
4.98512357e-01 3.09723437e-01 4.76553053e-01 -1.05327106e+00
-6.15311861e-01 -5.01445651e-01 -3.58413309e-01 1.73309490e-01
5.51876426e-01 -4.57293876e-02 -8.98933291e-01 1.70863974e+00
7.21149504e-01 6.17705464e-01 -1.53878093e-01 1.18721795e+00
6.17106259e-01 3.29911351e-01 -1.54253140e-01 -1.47809079e-02
1.17019880e+00 -1.04518533e+00 -4.60759223e-01 -2.96987414e-01
2.19733357e-01 -6.41802251e-01 5.15565157e-01 5.20958146e-03
-1.06593859e+00 -6.70544207e-01 -8.81702363e-01 1.92125827e-01
1.86068133e-01 5.64527735e-02 5.23542941e-01 5.34040749e-01
-8.56088638e-01 7.04806626e-01 -1.25553298e+00 -4.47927862e-01
4.15359557e-01 4.52716410e-01 -3.21146101e-01 -1.28232583e-01
-5.00340164e-01 6.88757420e-01 3.47902894e-01 6.67736605e-02
-1.34647071e+00 -9.33256626e-01 -9.00408208e-01 -3.77129704e-01
5.40355504e-01 -1.17060137e+00 1.31162012e+00 -3.11834663e-01
-1.26349819e+00 7.87364542e-01 -8.31719786e-02 -5.00346959e-01
7.34133720e-01 -5.84996819e-01 -9.26475078e-02 5.86010776e-02
2.28101507e-01 6.99708343e-01 7.59760678e-01 -1.51956713e+00
-6.82817638e-01 -5.07356882e-01 -1.90129384e-01 1.95314378e-01
1.77233860e-01 -2.20156476e-01 -9.43638027e-01 -6.17737412e-01
3.72262985e-01 -1.47573352e+00 -3.80398989e-01 5.97609818e-01
-3.46228331e-01 1.38842851e-01 8.94893169e-01 -5.73841155e-01
4.46869612e-01 -1.82427740e+00 4.77842450e-01 -1.17756955e-01
3.58429492e-01 -9.20847654e-02 -7.51584843e-02 -4.56315912e-02
2.83913761e-01 -4.08764720e-01 -1.64433584e-01 -9.47384179e-01
9.79195088e-02 2.99001902e-01 -2.41679773e-01 9.44331944e-01
-4.22708541e-02 1.27911258e+00 -1.15803409e+00 -3.50124180e-01
5.53130805e-01 5.70415556e-01 -6.35707796e-01 2.35870481e-01
-5.13893545e-01 8.45260561e-01 -5.26691794e-01 9.14589167e-01
6.99818909e-01 -4.76520658e-01 9.51287299e-02 -4.44662660e-01
-3.72062176e-02 -2.06107244e-01 -1.20683730e+00 2.53168178e+00
1.92373842e-02 5.51789165e-01 -6.09669909e-02 -3.86279047e-01
5.14217377e-01 2.13938616e-02 9.99628723e-01 -4.25903559e-01
1.75263330e-01 9.23663899e-02 -5.22738874e-01 -3.08022141e-01
6.22878492e-01 7.93569833e-02 -1.66901410e-01 2.33617723e-01
9.24090371e-02 -2.64181942e-01 -6.58364370e-02 1.30504444e-01
1.34520578e+00 8.02054703e-01 -1.35120172e-02 1.98271610e-02
2.65915036e-01 1.87228397e-01 6.31298363e-01 7.47570574e-01
-3.16650569e-01 7.78857827e-01 -2.49896780e-01 -5.47704935e-01
-1.16441369e+00 -1.43733001e+00 2.58783430e-01 7.06601620e-01
5.68080783e-01 -3.26761097e-01 -3.65652651e-01 -7.35842884e-01
5.47670960e-01 3.28334987e-01 -4.38361645e-01 2.76388470e-02
-9.47097361e-01 -2.36203253e-01 3.94282490e-01 5.47567070e-01
4.39257443e-01 -5.13830483e-01 -9.28925395e-01 2.28172988e-01
-2.36089125e-01 -1.69987571e+00 -6.54666662e-01 -4.42906562e-03
-1.02235591e+00 -1.13245082e+00 -7.28088498e-01 -3.86836618e-01
6.44381225e-01 6.47586346e-01 1.12539864e+00 -1.80958748e-01
-3.20395052e-01 9.53318775e-01 -1.01450540e-01 -1.67452306e-01
-3.50338370e-01 -2.94117004e-01 3.22353661e-01 -1.87544823e-01
-2.14786351e-01 -4.07628566e-01 -6.23205900e-01 4.55090940e-01
-2.64128983e-01 1.23937950e-01 4.22568232e-01 2.66948611e-01
9.68901157e-01 -4.54144359e-01 -5.99361435e-02 -1.27568766e-01
-9.32472795e-02 -2.06777602e-01 -1.14526451e+00 1.60093769e-01
-2.60625094e-01 2.76100710e-02 2.15579808e-01 -4.82007205e-01
-5.20750284e-01 8.53593290e-01 1.05565988e-01 -1.23406768e+00
-4.40403745e-02 -1.61256775e-01 7.03325197e-02 -6.28683329e-01
4.83564556e-01 1.95701435e-01 -8.77163187e-03 -4.11849916e-01
5.47254145e-01 9.04709660e-03 9.88494515e-01 -7.21118391e-01
1.36936867e+00 7.18160152e-01 3.84129971e-01 -4.75898743e-01
-9.07662988e-01 -5.21215320e-01 -7.78511345e-01 -7.24307179e-01
1.10065460e+00 -1.35614347e+00 -1.28124774e+00 5.15024602e-01
-1.27069366e+00 -4.59126920e-01 -3.04621488e-01 7.00373888e-01
-9.46085572e-01 3.30989748e-01 -3.49292397e-01 -7.49222815e-01
-1.43900141e-01 -1.18382931e+00 1.70310736e+00 -2.29210197e-03
9.01028663e-02 -7.93124139e-01 2.83106953e-01 2.57830441e-01
3.23117897e-02 7.73599207e-01 -8.22378602e-03 -9.56590325e-02
-1.34886765e+00 -5.96855395e-02 -1.10857785e-01 -1.34953901e-01
1.40537918e-01 -2.86866069e-01 -7.13703334e-01 -7.25528955e-01
-1.63856789e-01 -2.19187558e-01 5.87464273e-01 4.98675734e-01
8.21231782e-01 1.06329853e-02 -6.09072745e-01 9.90272939e-01
1.52820504e+00 -9.55857486e-02 4.56506610e-02 2.88521558e-01
1.15740371e+00 1.13714291e-02 8.60736191e-01 4.15164918e-01
6.45648718e-01 1.08496428e+00 8.32451940e-01 2.52231687e-01
-4.05084580e-01 -4.11416024e-01 4.07484710e-01 6.10966682e-01
-1.61434278e-01 -1.57146245e-01 -8.63622904e-01 5.04689753e-01
-2.01216006e+00 -7.68358648e-01 -4.62509692e-02 2.08181381e+00
3.56253833e-01 1.31129116e-01 4.13956523e-01 -4.30309594e-01
5.58452249e-01 1.38807893e-01 -8.03819954e-01 8.71370971e-01
1.00953855e-01 -2.26315543e-01 9.43314970e-01 4.53698605e-01
-1.17489183e+00 8.43533993e-01 6.15289211e+00 2.12246537e-01
-8.51499081e-01 1.33714944e-01 -1.47407100e-01 -4.97033566e-01
-5.66918515e-02 -1.66812778e-01 -1.09455597e+00 1.36338681e-01
8.14495087e-01 -1.99956670e-02 4.02435392e-01 8.69618356e-01
7.97089934e-02 2.85786569e-01 -1.29725528e+00 1.21638155e+00
6.45982847e-02 -1.65421677e+00 -2.77439862e-01 1.46131903e-01
8.09794009e-01 4.45910752e-01 -1.08899131e-01 -1.79160703e-02
4.74764407e-01 -5.05190015e-01 1.36369824e+00 5.66265106e-01
6.01382673e-01 -4.64023650e-01 1.03821769e-01 3.56393844e-01
-1.68767929e+00 1.27551183e-01 -1.97029456e-01 1.53602567e-02
4.18064028e-01 1.43332466e-01 -7.16707826e-01 9.80866730e-01
6.92820013e-01 1.18117464e+00 -6.61088109e-01 1.31054568e+00
-1.56517625e-02 1.18897669e-01 -6.03405893e-01 6.05562590e-02
1.08346649e-01 6.96258843e-02 1.00535083e+00 9.65962827e-01
4.42144513e-01 8.12039971e-02 5.22267520e-01 9.77057874e-01
-3.50707918e-02 -6.85237050e-01 -6.51026726e-01 2.99812347e-01
4.57959801e-01 1.15034902e+00 -8.88759434e-01 -1.49099380e-01
-2.27272987e-01 9.84620512e-01 2.10473403e-01 2.80711174e-01
-1.27818859e+00 2.59602576e-01 9.84614015e-01 3.44895385e-02
7.31957078e-01 -8.12001169e-01 6.39218315e-02 -1.16746497e+00
2.05887035e-01 -5.89194000e-01 1.77820310e-01 -8.50594699e-01
-1.02632642e+00 2.87695825e-01 1.14683919e-01 -1.58752275e+00
-4.26350176e-01 -5.11563241e-01 1.15166366e-01 5.50541341e-01
-1.31132257e+00 -1.41412497e+00 -6.83026075e-01 6.72881842e-01
5.37208200e-01 1.49482340e-01 4.59795803e-01 4.94449466e-01
-1.62375867e-01 2.89490849e-01 -1.21491970e-02 2.14147598e-01
3.95466149e-01 -1.13030720e+00 9.27089214e-01 1.11469948e+00
3.84773701e-01 4.68573868e-01 7.69735396e-01 -8.90934408e-01
-2.25355148e+00 -1.54385495e+00 1.84530973e-01 -1.08769870e+00
4.77947325e-01 -6.13505781e-01 -4.52412784e-01 1.15078819e+00
-2.35179365e-01 6.92269623e-01 1.32803825e-05 -3.96581411e-01
-1.07392661e-01 -4.26860079e-02 -9.83107388e-01 4.69466627e-01
1.64802897e+00 -2.17750221e-01 -4.43788320e-01 5.36237836e-01
1.26694870e+00 -1.27294779e+00 -9.21968460e-01 4.63857859e-01
6.60778701e-01 -7.06305444e-01 1.58343756e+00 -3.91427130e-01
-1.38532996e-01 -8.97872269e-01 -4.66879696e-01 -9.29183483e-01
-3.50786030e-01 -8.12400162e-01 -5.73981047e-01 6.68733418e-01
-1.47831425e-01 -2.24575505e-01 1.04529738e+00 2.71320105e-01
-3.79659206e-01 -5.68523943e-01 -1.05858040e+00 -1.12101889e+00
-4.81419742e-01 -7.97773302e-01 4.08029258e-01 4.77457047e-01
-9.56507325e-01 4.02035452e-02 -6.70576155e-01 6.41926348e-01
1.45653725e+00 4.16367084e-01 1.37173784e+00 -1.11574674e+00
-4.00891691e-01 -1.27055300e-02 -6.70005918e-01 -1.73029637e+00
8.82430747e-02 -7.56025493e-01 3.61241251e-01 -1.53292871e+00
9.28449407e-02 -4.38932061e-01 8.10394064e-02 2.30284095e-01
-5.91221452e-02 3.43564957e-01 6.64949238e-01 2.08989084e-01
-1.22500181e+00 6.32636428e-01 1.34719789e+00 -1.15690269e-01
-7.07463846e-02 1.47451565e-01 -1.86966583e-01 5.69303989e-01
5.20018339e-02 -7.06069887e-01 -1.84683651e-01 -6.45935535e-01
-1.42392561e-01 3.75312686e-01 1.00095975e+00 -1.42171156e+00
4.46403474e-01 -1.31031632e-01 6.27891004e-01 -1.15811741e+00
7.97400951e-01 -1.23809838e+00 6.69989467e-01 7.69999802e-01
9.50413123e-02 2.56653666e-01 4.93625432e-01 9.64061677e-01
4.49395686e-01 4.38217402e-01 6.25044405e-01 -5.61870858e-02
-9.96926904e-01 6.93771780e-01 2.79065192e-01 5.89572676e-02
1.30902421e+00 -3.18888515e-01 -1.14060782e-01 -1.77143127e-01
-5.90369463e-01 3.65978956e-01 7.28940606e-01 6.72648251e-01
7.54599035e-01 -1.65492070e+00 -6.40336692e-01 1.31196678e-01
9.58058834e-02 2.88584679e-01 9.46442708e-02 8.12364459e-01
-5.93561530e-01 3.87700826e-01 -2.21888736e-01 -1.46201694e+00
-1.40763927e+00 5.08661270e-01 4.94343847e-01 -1.70488767e-02
-1.07725179e+00 8.67084801e-01 -1.76968295e-02 -6.32243931e-01
4.01901275e-01 -6.96503282e-01 5.69604754e-01 -5.90578556e-01
2.52626836e-01 3.64498913e-01 -1.13583714e-01 -8.54654610e-01
-7.82295644e-01 1.13620114e+00 3.43433052e-01 -1.31488949e-01
1.44878852e+00 -2.54268974e-01 3.85497659e-01 4.11368310e-01
1.16435087e+00 -8.01664740e-02 -2.02909875e+00 -1.87737256e-01
-2.45555863e-01 -8.59667122e-01 -3.25944006e-01 -4.88807410e-01
-1.08185363e+00 2.45215788e-01 8.10407579e-01 -2.44692102e-01
6.83009207e-01 2.46703893e-01 7.67880797e-01 5.07502913e-01
7.86880195e-01 -3.91857386e-01 3.59905303e-01 5.88270128e-01
7.25474060e-01 -1.21469653e+00 1.72426254e-01 -3.48408073e-01
-2.31185585e-01 1.01059556e+00 4.37676698e-01 -1.69992790e-01
2.45587185e-01 2.56830841e-01 -1.29739970e-01 -3.79757941e-01
-3.95045221e-01 -1.41725510e-01 4.45919752e-01 6.33410156e-01
-1.07625298e-01 -1.99223027e-01 6.73648834e-01 4.69358228e-02
-8.52688849e-02 5.11980020e-02 4.48062830e-02 1.11890686e+00
-3.32122266e-01 -6.05440080e-01 -6.79258645e-01 -4.20969799e-02
-7.22054914e-02 4.51580733e-01 -4.60642129e-02 9.37490642e-01
-6.67608604e-02 5.86611867e-01 -1.00793414e-01 -5.58587551e-01
6.86568856e-01 -3.22028905e-01 1.10873437e+00 -3.17440510e-01
-3.45960379e-01 7.56614208e-02 -9.86112133e-02 -1.12032485e+00
-6.37443304e-01 -1.00129056e+00 -1.35613966e+00 -3.20205331e-01
-1.85151145e-01 -3.35897624e-01 6.88330829e-01 9.03092325e-01
5.43228745e-01 8.41915667e-01 2.46669888e-01 -1.51579404e+00
-4.57633317e-01 -4.75664556e-01 -2.15169713e-01 4.55468893e-01
7.99138010e-01 -9.32284355e-01 -3.25635634e-02 1.46537811e-01]
|
[6.9193034172058105, -2.311037302017212]
|
bdbdca73-f43c-481c-9ac9-c8f486c0aebf
|
lightweight-image-inpainting-by-stripe-window
|
2301.00553
| null |
https://arxiv.org/abs/2301.00553v1
|
https://arxiv.org/pdf/2301.00553v1.pdf
|
Lightweight Image Inpainting by Stripe Window Transformer with Joint Attention to CNN
|
Image inpainting is an important task in computer vision. As admirable methods are presented, the inpainted image is getting closer to reality. However, the result is still not good enough in the reconstructed texture and structure based on human vision. Although more and more larger models have been proposed recently because of the advancement of computer hardware, we would like to build a suitable model for personal use or small-sized institution. Therefore, we propose a lightweight model that combines the special transformer and the traditional convolutional neural network (CNN). Furthermore, we noticed most researchers only consider three primary colors (RGB) in inpainted images, but we think this is not enough so we propose a new loss function to intensify color details. Extensive experiments on commonly seen datasets (Places2 and CelebA) validate the efficacy of our proposed model compared with other state-of-the-art methods. Index Terms - HSV color space, image inpainting, joint attention mechanism, stripe window, vision transformer
|
['Kuan-Hsien Liu', 'Po-Wei Chen', 'Tsung-Jung Liu']
|
2023-01-02
| null | null | null | null |
['image-inpainting']
|
['computer-vision']
|
[ 4.87296320e-02 -2.36921072e-01 3.50559428e-02 -2.19243601e-01
-2.45172337e-01 1.04190625e-01 1.32655859e-01 -3.44683588e-01
-3.70056689e-01 8.61110926e-01 -3.52074504e-02 -1.82213396e-01
3.52299064e-01 -7.45882630e-01 -9.63357151e-01 -5.98979354e-01
3.63265693e-01 -1.86870530e-01 3.96200657e-01 -3.96277130e-01
4.12265360e-01 4.10572320e-01 -1.52131367e+00 3.29173476e-01
1.10121727e+00 1.15266454e+00 3.87796760e-01 2.97287166e-01
-2.67137975e-01 1.02751553e+00 -4.86325920e-01 -5.80381632e-01
4.86543357e-01 -7.08813667e-01 -5.87393761e-01 2.03296661e-01
6.09994054e-01 -7.12751210e-01 -5.88797867e-01 1.25218165e+00
4.42229986e-01 -1.15184300e-02 1.80224508e-01 -1.15771663e+00
-1.36071813e+00 3.02008092e-01 -1.01230216e+00 2.04368532e-01
2.21888348e-01 3.83947529e-02 4.16215599e-01 -7.91526437e-01
4.61712688e-01 1.07718551e+00 7.82920539e-01 5.78787446e-01
-7.13914573e-01 -7.46159673e-01 1.83483541e-01 5.05231678e-01
-1.10244775e+00 -1.04535379e-01 1.15200114e+00 2.17528284e-01
5.99017203e-01 3.38472486e-01 9.15329874e-01 1.02765667e+00
2.85052955e-01 9.13050115e-01 1.55343664e+00 -5.77919781e-01
-9.70733762e-02 3.43647301e-01 -4.15219605e-01 8.13481688e-01
2.01705500e-01 9.51954499e-02 -4.30294156e-01 2.90402204e-01
1.14950979e+00 4.67454225e-01 -4.98206943e-01 -5.57549931e-02
-1.11353171e+00 5.46813250e-01 7.07754016e-01 3.02415401e-01
-3.46666813e-01 1.14371419e-01 1.94843903e-01 4.50511456e-01
6.22421265e-01 -2.38624141e-02 -1.47029102e-01 2.89115906e-02
-1.15594065e+00 2.31788188e-01 2.92750716e-01 1.00814760e+00
5.12446940e-01 1.71021029e-01 -7.63652101e-03 9.35711384e-01
8.26622620e-02 2.98171133e-01 5.04289091e-01 -8.87916267e-01
3.01905453e-01 6.43161893e-01 1.97466120e-01 -1.23025930e+00
2.48643644e-02 -1.39422044e-01 -1.21378529e+00 5.53870440e-01
1.16271012e-01 1.14054777e-01 -1.08398402e+00 1.40036511e+00
4.84345444e-02 2.45186985e-01 -3.35571378e-01 1.25571263e+00
8.11119437e-01 7.40260720e-01 -6.92932010e-02 -1.96472242e-01
1.14945292e+00 -1.35657692e+00 -9.52004015e-01 -1.58167854e-01
-2.50322819e-01 -1.13584459e+00 1.18776178e+00 7.38514364e-01
-1.34526575e+00 -1.02042925e+00 -1.13475955e+00 -4.80095208e-01
-4.00758803e-01 5.64990789e-02 9.19624567e-01 6.22117937e-01
-1.12837958e+00 7.74677694e-01 -5.66501737e-01 -4.63920206e-01
4.14220184e-01 -1.60747305e-01 -3.33222330e-01 -3.40092570e-01
-1.16819322e+00 1.08812010e+00 1.98904246e-01 5.13026059e-01
-6.27342999e-01 -2.87551969e-01 -3.78034294e-01 -7.44047388e-02
6.96168020e-02 -5.11664569e-01 9.14459407e-01 -1.30694246e+00
-1.46406567e+00 7.27185249e-01 7.76022747e-02 -2.71681726e-01
8.98658514e-01 -2.82464564e-01 -5.78061640e-01 2.30291918e-01
-1.84628934e-01 7.25146174e-01 1.11707890e+00 -1.45300007e+00
-7.40947306e-01 -1.94291130e-01 1.86395094e-01 2.36571267e-01
-3.59788239e-01 -7.04315607e-04 -6.24866784e-01 -8.07363689e-01
1.16277598e-01 -5.83620131e-01 -1.28361657e-01 6.62888229e-01
-1.55007169e-01 6.69338331e-02 1.14789569e+00 -1.17412102e+00
1.12688720e+00 -2.13886380e+00 -7.90406018e-02 -2.22053796e-01
1.27587721e-01 4.67694730e-01 4.95999083e-02 2.95059830e-01
-4.82441634e-02 -1.02291718e-01 -3.99709880e-01 -4.57613349e-01
-2.98887342e-01 1.65147200e-01 -3.03383589e-01 4.76127177e-01
1.84589818e-01 7.44459569e-01 -6.20467603e-01 -6.32707894e-01
3.63036275e-01 7.68023729e-01 -3.43861639e-01 2.65561640e-01
-7.33121438e-03 2.34129280e-01 -2.73701042e-01 9.99301136e-01
1.24918854e+00 5.22281677e-02 -3.49805653e-01 -5.19497633e-01
-2.51174659e-01 -4.06354994e-01 -1.07633960e+00 1.96588326e+00
-3.13991845e-01 6.75801337e-01 6.70917034e-02 -1.04269481e+00
9.32799518e-01 1.06023997e-01 3.31041783e-01 -1.07790017e+00
2.68074483e-01 2.15058625e-01 -2.38511816e-01 -8.62413168e-01
6.11964405e-01 -1.04933374e-01 5.11648357e-01 2.04399303e-01
-2.28478000e-01 -6.42125756e-02 4.84548621e-02 3.24400775e-02
6.04820013e-01 6.18715644e-01 -8.51344690e-02 6.28469214e-02
3.32625508e-01 -3.56601290e-02 4.57047403e-01 4.74752784e-01
-4.63942021e-01 1.14325917e+00 2.31752038e-01 -7.45244980e-01
-1.38917494e+00 -8.49831164e-01 3.52962175e-03 7.31363595e-01
5.66575706e-01 1.56148344e-01 -7.89237022e-01 -4.54838783e-01
-1.15309782e-01 3.28059673e-01 -8.35965991e-01 -1.55255213e-01
-4.71660346e-01 -5.00989437e-01 4.75333780e-01 5.34378648e-01
1.29117215e+00 -1.29126656e+00 -7.65107334e-01 8.22382048e-02
-3.74440253e-01 -9.34283555e-01 -6.79125667e-01 -1.51238471e-01
-9.81028795e-01 -9.38793361e-01 -1.18776131e+00 -1.02926481e+00
7.00574994e-01 6.65026844e-01 9.72643554e-01 3.92756253e-01
-4.63562906e-01 -2.04658136e-02 -5.66309988e-01 -3.96835506e-01
-5.64779826e-02 -4.91600007e-01 -3.06043267e-01 1.20396979e-01
2.18610108e-01 -5.95468044e-01 -1.02385020e+00 -1.33973390e-01
-1.26151073e+00 3.07315737e-01 9.13933039e-01 9.38924193e-01
3.97775024e-01 1.68160751e-01 1.09237045e-01 -8.23221803e-01
6.56140685e-01 -2.16485653e-02 -3.46632630e-01 5.37364304e-01
-6.52130902e-01 -1.43180370e-01 6.87631130e-01 -5.41799128e-01
-1.31520891e+00 -2.73013115e-01 -3.81862037e-02 -6.94187820e-01
-6.30536973e-02 2.18127161e-01 1.09293967e-01 -4.59374189e-01
3.19510579e-01 6.43040359e-01 1.92771003e-01 -5.88603199e-01
2.54874527e-01 7.86908448e-01 5.03639638e-01 -3.14886421e-01
6.45580411e-01 5.63341856e-01 -1.15351327e-01 -5.69153190e-01
-5.05712628e-01 3.58967334e-02 -1.75054148e-01 -2.63990521e-01
7.59804428e-01 -8.80510211e-01 -8.32413495e-01 7.52046168e-01
-1.20671487e+00 -3.82711142e-02 2.52320692e-02 3.15180928e-01
-4.24364299e-01 7.04465806e-01 -9.29071188e-01 -6.33744776e-01
-4.74722981e-01 -1.09620309e+00 8.01202595e-01 5.46018660e-01
5.70016742e-01 -7.32385099e-01 -2.41501719e-01 3.83556098e-01
9.35163319e-01 4.30845797e-01 6.63828433e-01 3.87107462e-01
-8.04176867e-01 3.65007333e-02 -6.68355942e-01 6.75475895e-01
2.57198393e-01 2.40321964e-01 -9.41100776e-01 -9.61381122e-02
2.54094213e-01 -4.00373548e-01 1.09933650e+00 2.76034921e-01
1.54283202e+00 -1.98388085e-01 -3.48003693e-02 7.89982438e-01
1.83697677e+00 3.09169978e-01 1.21324563e+00 5.58883488e-01
5.68636000e-01 3.67734015e-01 6.25421762e-01 2.55147129e-01
3.05191487e-01 2.61808097e-01 5.67909122e-01 -6.57421589e-01
-4.35664952e-01 -3.40687931e-01 2.31400594e-01 7.43162751e-01
-5.58614016e-01 2.14453470e-02 -2.46102154e-01 4.02147233e-01
-1.69704759e+00 -1.02205122e+00 -9.28108096e-02 1.99845719e+00
9.49406266e-01 5.82276247e-02 -9.09126326e-02 1.26414552e-01
6.05348885e-01 1.32336868e-02 -5.29313087e-01 -4.55081761e-01
-3.61923307e-01 3.04180294e-01 5.47468901e-01 1.22259691e-01
-8.84451866e-01 6.79355145e-01 6.01702213e+00 9.45126832e-01
-1.60415018e+00 1.76889077e-01 1.05449820e+00 2.46778354e-01
-1.35440528e-01 -5.93737811e-02 -2.04398662e-01 7.64606535e-01
2.03067839e-01 1.75595537e-01 6.85484946e-01 7.65579820e-01
9.68865901e-02 -3.11557353e-01 -6.32448256e-01 1.26096392e+00
4.38015968e-01 -1.23555899e+00 1.98315710e-01 -3.89515400e-01
7.19303370e-01 -3.20577234e-01 3.72286499e-01 1.18262924e-01
-2.79508382e-01 -9.41462278e-01 7.60671496e-01 7.82022476e-01
6.98940039e-01 -8.15006077e-01 7.16931343e-01 8.55645314e-02
-8.12397420e-01 -8.30661505e-02 -7.75904119e-01 -1.08690895e-02
-7.70008788e-02 4.83344883e-01 -1.39061466e-01 6.77439153e-01
1.20998466e+00 7.08130062e-01 -8.82715225e-01 1.30028772e+00
1.27774984e-01 1.99121103e-01 -9.93534103e-02 6.80516213e-02
3.17832291e-01 -3.43770385e-01 -1.43807143e-01 8.65086794e-01
5.65695703e-01 1.81009054e-01 -2.01176479e-01 9.65517402e-01
-3.17389071e-02 5.02280444e-02 -5.29556513e-01 8.31129923e-02
6.27071634e-02 1.29159820e+00 -8.51815462e-01 -4.02065694e-01
-7.56688058e-01 1.54677379e+00 1.59711331e-01 4.14052218e-01
-1.04340672e+00 -5.32572746e-01 3.67328793e-01 4.40083444e-02
3.00277531e-01 1.26987405e-04 -2.64887154e-01 -1.25864005e+00
2.18031853e-01 -8.93207729e-01 -1.13153486e-02 -1.20044446e+00
-1.33472002e+00 8.75347614e-01 -2.63709843e-01 -1.60382962e+00
4.39728618e-01 -5.59326947e-01 -5.78546464e-01 7.26752162e-01
-1.83937740e+00 -1.42346311e+00 -6.43694937e-01 8.56956422e-01
6.44833267e-01 7.34469071e-02 5.30923426e-01 6.32008195e-01
-3.09681505e-01 4.41101819e-01 2.25220770e-01 2.82605272e-03
1.01813638e+00 -8.80442262e-01 1.63342729e-01 9.22692180e-01
-1.95533112e-01 6.24334931e-01 8.35137248e-01 -4.99594957e-01
-1.36128247e+00 -9.17515218e-01 5.85467696e-01 1.05548754e-01
1.06847867e-01 -2.18906347e-02 -9.12568510e-01 4.72574353e-01
9.31711793e-01 2.39344001e-01 2.63154488e-02 -5.81312656e-01
-3.22080672e-01 -5.39820671e-01 -1.37838387e+00 6.15433097e-01
8.78935933e-01 -1.68193057e-01 -3.74756843e-01 2.28167847e-01
6.01397395e-01 -4.78804708e-01 -4.81704414e-01 2.61722386e-01
6.52866483e-01 -1.51465476e+00 8.70403111e-01 -1.63405642e-01
7.58492172e-01 -4.78809834e-01 -3.38468254e-02 -1.16132581e+00
-3.12389374e-01 -3.62660617e-01 8.50823075e-02 9.51843262e-01
-1.98054388e-02 -5.63425779e-01 7.57932007e-01 5.14860094e-01
-3.34681459e-02 -8.42286825e-01 -6.36206150e-01 -4.77816105e-01
-1.48754776e-01 9.12262574e-02 4.91686970e-01 9.74739134e-01
-3.05857986e-01 -8.17847997e-02 -1.02022028e+00 -9.20550302e-02
8.03201497e-01 3.62590075e-01 4.15110081e-01 -9.11482334e-01
-2.36788094e-01 -3.58595997e-01 -1.99209347e-01 -8.42562675e-01
-3.92719448e-01 -1.19598240e-01 -8.27922449e-02 -1.68015528e+00
3.34813327e-01 -3.82445842e-01 -4.19952661e-01 4.33668375e-01
-2.89866567e-01 7.63941765e-01 3.12593818e-01 4.86333817e-02
-4.21016544e-01 6.79339409e-01 1.84651899e+00 -3.28572214e-01
1.82155833e-01 -3.49487156e-01 -8.08857977e-01 4.16925400e-01
7.15854466e-01 -1.59716904e-01 -3.65425944e-01 -6.56209946e-01
4.95103784e-02 6.44222423e-02 4.79731768e-01 -1.08162189e+00
3.39021891e-01 -2.79539078e-01 9.14502144e-01 -5.68984389e-01
4.33304250e-01 -1.11942768e+00 2.59914517e-01 4.98959124e-01
-1.88261509e-01 2.70852864e-01 9.65003148e-02 3.40991318e-01
-5.60386717e-01 4.55348529e-02 1.00733280e+00 -5.86850941e-01
-9.06309247e-01 3.79847229e-01 1.06031321e-01 -3.70229900e-01
1.07955849e+00 -5.86290896e-01 -2.53780931e-01 -4.57769096e-01
-2.89714307e-01 -1.72568843e-01 4.81976300e-01 4.19509649e-01
1.11757088e+00 -1.28813291e+00 -6.45583928e-01 3.64431769e-01
-1.09422110e-01 -9.94966701e-02 5.00153065e-01 7.54090428e-01
-1.20947993e+00 1.17488801e-01 -9.78465915e-01 -2.21025899e-01
-1.04157901e+00 7.35306442e-01 5.39592803e-02 1.47782773e-01
-7.93208599e-01 9.14435506e-01 -4.07808460e-02 -3.63459662e-02
3.97761315e-01 -5.25743186e-01 -4.24238201e-03 -3.86239052e-01
5.89185894e-01 2.47806549e-01 -2.13335436e-02 -2.77407467e-01
-8.86735320e-02 5.40161133e-01 -2.36654893e-01 3.22665840e-01
1.44636250e+00 -3.85046184e-01 -4.31866169e-01 1.66649699e-01
8.99558067e-01 -1.58874810e-01 -1.46641195e+00 -2.08811834e-03
-6.05598450e-01 -8.65233004e-01 -3.31447683e-02 -7.34818995e-01
-1.56809616e+00 9.51323390e-01 1.01958704e+00 2.30549723e-01
1.67797863e+00 -5.69045842e-01 1.15885305e+00 1.57291487e-01
4.00080383e-01 -1.21885037e+00 1.95060015e-01 1.28074782e-02
9.31615174e-01 -1.41511309e+00 9.96913463e-02 -2.11146697e-01
-6.95450366e-01 1.27519524e+00 9.31808770e-01 -6.28430486e-01
5.71506262e-01 1.63458303e-01 3.40526611e-01 3.25321347e-01
-4.93872553e-01 -5.97501509e-02 -1.55939177e-01 7.18880832e-01
4.00327086e-01 -2.87942588e-01 -5.59915602e-01 1.87116429e-01
1.63101271e-01 3.34485412e-01 6.57716632e-01 8.98680270e-01
-4.46366370e-01 -1.10277581e+00 -5.92306376e-01 1.68460920e-01
-5.96518397e-01 -1.53185129e-01 -1.04516096e-01 9.33820546e-01
4.01075393e-01 7.41554618e-01 -1.58262387e-01 -3.35709214e-01
1.37720183e-01 -3.35500538e-01 8.66295218e-01 1.32935777e-01
-3.71062934e-01 -7.84280375e-02 -4.39884096e-01 -6.69028759e-01
-6.59111977e-01 4.24521081e-02 -7.56331801e-01 -5.80780625e-01
-1.14936255e-01 -7.14460611e-02 7.73697793e-01 5.37868023e-01
2.16948211e-01 5.00368655e-01 5.69302917e-01 -9.17537749e-01
-2.45856136e-01 -1.16814315e+00 -7.57163942e-01 5.48680067e-01
4.13480699e-01 -5.95924556e-01 -8.27199444e-02 2.88591176e-01]
|
[11.244551658630371, -1.6545051336288452]
|
858120dc-bbec-4c0d-9956-28fd1f5f5657
|
an-empirical-study-on-multi-task-learning-for
| null | null |
https://aclanthology.org/2020.coling-industry.6
|
https://aclanthology.org/2020.coling-industry.6.pdf
|
An Empirical Study on Multi-Task Learning for Text Style Transfer and Paraphrase Generation
|
The topic of this paper is neural multi-task training for text style transfer. We present an efficient method for neutral-to-style transformation using the transformer framework. We demonstrate how to prepare a robust model utilizing large paraphrases corpora together with a small parallel style transfer corpus. We study how much style transfer data is needed for a model on the example of two transformations: neutral-to-cute on internal corpus and modern-to-antique on publicly available Bible corpora. Additionally, we propose a synthetic measure for the automatic evaluation of style transfer models. We hope our research is a step towards replacing common but limited rule-based style transfer systems by more flexible machine learning models for both public and commercial usage.
|
['Katarzyna Beksa', 'Tymoteusz Krumholc', 'Jaroslaw Piersa', 'Katarzyna Witkowska', 'Hyungtak Choi', 'Kseniia Ryzhova', 'Pawel Bujnowski']
|
2020-12-01
| null | null | null |
coling-2020-8
|
['paraphrase-generation', 'paraphrase-generation']
|
['computer-code', 'natural-language-processing']
|
[ 5.59231997e-01 7.04387724e-02 -9.35779139e-02 -6.98003352e-01
-9.85000908e-01 -8.31489444e-01 9.56163049e-01 -4.82097805e-01
-6.12429559e-01 1.19592762e+00 3.81818384e-01 -3.86810720e-01
3.70566219e-01 -7.36121058e-01 -8.39295447e-01 -3.46357226e-01
7.91159153e-01 9.11722481e-01 3.78388762e-02 -8.35118949e-01
3.04241627e-01 3.49209547e-01 -9.01342511e-01 7.28242338e-01
7.45048046e-01 4.11169052e-01 4.84676734e-02 8.42752814e-01
-4.24173445e-01 5.21272421e-01 -8.34161997e-01 -9.44507837e-01
2.60963708e-01 -7.96297550e-01 -1.31246912e+00 -4.71454233e-01
7.59220243e-01 -2.23226950e-01 4.81796898e-02 8.62872601e-01
6.85635090e-01 5.58850281e-02 1.00929105e+00 -1.07859170e+00
-9.30413604e-01 7.77389228e-01 -2.43179634e-01 -2.81664915e-02
3.04865897e-01 -2.70799585e-02 8.81748974e-01 -7.63186216e-01
9.97385144e-01 1.52164590e+00 5.75034559e-01 1.06344187e+00
-1.43529034e+00 -6.74759865e-01 -1.32056743e-01 8.83660093e-03
-8.11102688e-01 -4.52083260e-01 1.02438962e+00 -2.18910456e-01
7.90862918e-01 3.77478898e-01 4.53342766e-01 1.76121950e+00
3.26287359e-01 8.57559979e-01 1.53875768e+00 -7.16403425e-01
-7.07782805e-02 4.75302815e-01 -2.87105478e-02 4.13461179e-01
-1.14255175e-01 -2.56477352e-02 -6.87133193e-01 -2.34695852e-01
8.63816559e-01 -6.66375875e-01 5.25314882e-02 -7.01021627e-02
-1.18584037e+00 7.83696175e-01 2.83142235e-02 4.72655952e-01
9.50989872e-02 -2.20261179e-02 8.03937197e-01 1.15135479e+00
8.48032415e-01 6.24475658e-01 -6.02274060e-01 -2.02687114e-01
-1.08112693e+00 4.81197804e-01 9.87613201e-01 1.43901408e+00
7.48825133e-01 1.80678919e-01 -3.75311464e-01 1.19627678e+00
-1.31456435e-01 6.84833765e-01 5.56652725e-01 -8.82424057e-01
5.97074330e-01 1.10454805e-01 8.34941715e-02 -3.80070806e-01
4.15387414e-02 -2.93683827e-01 -7.78134942e-01 2.22712353e-01
5.18420279e-01 -3.00944120e-01 -5.22509158e-01 1.97827208e+00
4.06804644e-02 -4.35656011e-01 4.45742346e-02 2.10513741e-01
5.01089811e-01 6.72401309e-01 1.09849826e-01 1.63269900e-02
1.21124327e+00 -1.08697510e+00 -6.06086373e-01 -2.00748846e-01
8.95039439e-01 -1.13485563e+00 1.62741029e+00 2.04366595e-01
-1.52986467e+00 -6.65448785e-01 -8.84622395e-01 -3.68566692e-01
-4.31517959e-01 4.46467251e-02 3.58106107e-01 6.22272909e-01
-1.21154714e+00 8.48837197e-01 -3.15134019e-01 -7.06841171e-01
2.52855062e-01 1.06525443e-01 -3.07668984e-01 3.13446254e-01
-1.21833026e+00 1.47143972e+00 3.69769990e-01 -4.55560386e-01
-6.52213395e-01 -8.60592961e-01 -6.22063398e-01 -2.14627460e-01
-2.23684430e-01 -1.13870740e+00 1.43816876e+00 -1.57836413e+00
-2.02074623e+00 1.45951855e+00 -1.09991811e-01 -2.80805320e-01
8.26917946e-01 -4.45566773e-01 -4.13479805e-01 -2.29525760e-01
8.14024508e-02 8.38776767e-01 8.16167116e-01 -1.17854106e+00
-5.48609257e-01 1.19286682e-02 -3.42866406e-02 3.40852737e-01
-5.42267382e-01 5.86467922e-01 5.00103906e-02 -1.13249075e+00
-7.64218569e-01 -1.01031125e+00 4.97833379e-02 6.54273108e-02
-1.71276435e-01 -3.12239408e-01 7.15719044e-01 -8.80768836e-01
8.69916201e-01 -1.66020679e+00 6.30408883e-01 -1.35428369e-01
-4.16180342e-01 2.70070761e-01 -5.15848577e-01 5.15631855e-01
-1.98445152e-02 1.16862267e-01 -3.16204280e-01 -4.81243521e-01
1.00626230e-01 1.29140809e-01 -5.50371349e-01 4.45611123e-03
1.66501001e-01 9.91905987e-01 -8.24451089e-01 -5.83106220e-01
-8.23082700e-02 2.25164965e-01 -7.64375865e-01 4.69865769e-01
-2.89877266e-01 5.76358736e-01 -2.54552543e-01 2.03489751e-01
4.39912230e-01 1.92547232e-01 -3.97541486e-02 -4.18655872e-02
3.60540636e-02 5.38103163e-01 -6.41579866e-01 2.27005601e+00
-1.01131046e+00 5.10643244e-01 -1.90123022e-02 -6.36518598e-01
1.07809114e+00 3.80687535e-01 -1.38745248e-01 -4.34602231e-01
1.60004184e-01 3.85956436e-01 -2.50134826e-01 3.72587629e-02
7.20422149e-01 -7.05790818e-01 -3.26574951e-01 8.70838583e-01
5.60474575e-01 -8.11742246e-01 3.22567612e-01 3.41551565e-02
6.30283892e-01 8.56001556e-01 2.86070079e-01 -8.97998512e-01
7.78916359e-01 -1.58358276e-01 3.12623173e-01 6.23375952e-01
-3.62368785e-02 5.30795634e-01 2.78283447e-01 -4.99471128e-01
-1.58871663e+00 -1.07491803e+00 1.36954710e-01 1.62321711e+00
-4.38572228e-01 -2.84189939e-01 -1.10596931e+00 -8.37505639e-01
-3.84729236e-01 1.31319511e+00 -6.75022900e-01 -2.33786970e-01
-1.07447469e+00 -4.37878519e-01 9.42923307e-01 5.36063671e-01
5.15428185e-01 -1.39781570e+00 4.04986441e-02 1.55345142e-01
-2.92237371e-01 -7.42605329e-01 -9.11450028e-01 4.92683388e-02
-9.73879874e-01 -4.05620545e-01 -1.04348814e+00 -1.02003586e+00
4.34555888e-01 -3.47749859e-01 1.55418265e+00 -1.67250052e-01
4.19516474e-01 1.72332212e-01 -2.78505832e-02 -6.30210042e-01
-1.16399109e+00 6.48220658e-01 -5.46124764e-02 -3.04470569e-01
2.74792731e-01 -7.25799859e-01 -1.60740972e-01 2.21061885e-01
-8.87745142e-01 4.93915260e-01 3.61032963e-01 9.70189393e-01
1.27165213e-01 -1.11099553e+00 7.63907492e-01 -1.17317641e+00
9.35434878e-01 -9.63140875e-02 -2.90244013e-01 5.99725306e-01
-4.49117452e-01 4.20899004e-01 9.09807265e-01 -4.35824633e-01
-1.77870524e+00 -2.27023035e-01 -4.27195787e-01 -1.35493889e-01
-2.97088683e-01 -1.45040005e-01 -2.18498722e-01 -3.15430909e-02
1.05215096e+00 2.37387151e-01 -1.55388713e-01 -5.67534149e-01
7.46619046e-01 6.37341797e-01 4.48727190e-01 -1.32906854e+00
1.04473805e+00 2.83787787e-01 -1.94530711e-01 -6.81641698e-01
-9.50183272e-01 1.24838643e-01 -9.01448369e-01 1.06450841e-01
7.12372899e-01 -8.14903438e-01 5.97087853e-02 5.18084049e-01
-1.49579060e+00 -7.66725481e-01 -5.54517269e-01 -3.89489233e-02
-9.26958382e-01 4.15982515e-01 -9.69939172e-01 -8.78159627e-02
-9.66601014e-01 -8.54919493e-01 1.00392592e+00 -6.50988743e-02
-8.43781412e-01 -1.56300569e+00 7.27776349e-01 1.81530088e-01
7.42537737e-01 -1.52483970e-01 1.06609988e+00 -7.60601461e-01
7.71122659e-03 3.58785242e-01 5.47110289e-02 6.38750553e-01
3.03199261e-01 -5.57569824e-02 -9.79519188e-01 -4.39374447e-01
1.41515672e-01 -7.16790140e-01 5.78887939e-01 -7.62878731e-02
9.17585909e-01 -3.11079234e-01 5.36739714e-02 8.25780988e-01
1.13522315e+00 1.16684780e-01 6.81057394e-01 4.62113976e-01
8.57130170e-01 5.38050354e-01 3.04201722e-01 -6.46180958e-02
2.34357744e-01 7.98035264e-01 -5.63770294e-01 -1.36004448e-01
-4.39592332e-01 -3.73958588e-01 5.72632790e-01 1.27101684e+00
-2.37116471e-01 -1.56127915e-01 -5.18341243e-01 5.43815434e-01
-1.38531482e+00 -9.76731181e-01 2.01105729e-01 2.00495672e+00
1.55288196e+00 2.13262588e-01 1.74306363e-01 -2.61241883e-01
7.72235572e-01 8.49935859e-02 -9.73943099e-02 -1.08767581e+00
-2.83927262e-01 7.39646435e-01 1.54507354e-01 7.34648228e-01
-9.37776923e-01 1.46514630e+00 7.08801699e+00 1.05238616e+00
-1.02766061e+00 2.71018714e-01 5.98356903e-01 -2.01075986e-01
-6.99933290e-01 -9.82415164e-04 -9.01630223e-01 3.38973761e-01
1.14559054e+00 -5.08208930e-01 5.45503676e-01 7.53858685e-01
-9.75864455e-02 5.91580331e-01 -1.56136250e+00 6.77254200e-01
2.68206984e-01 -1.28213322e+00 7.06579149e-01 -3.47952753e-01
9.68665183e-01 -1.15377598e-01 -1.03514425e-01 5.44531345e-01
5.90578556e-01 -6.62127256e-01 8.76649380e-01 3.50577831e-01
1.20837784e+00 -6.62046611e-01 3.23949754e-01 -1.19174421e-02
-7.64660716e-01 6.20845795e-01 -3.62780720e-01 -1.54285893e-01
2.92405456e-01 1.73348382e-01 -4.82115149e-01 6.07389510e-01
2.33668223e-01 6.21886313e-01 -7.19540656e-01 3.91463310e-01
-4.20969009e-01 6.19769216e-01 -6.62805587e-02 -1.38659000e-01
1.27297953e-01 -3.45115423e-01 5.44545710e-01 1.66279089e+00
4.19055969e-01 -3.77970994e-01 -2.27533638e-01 9.58691239e-01
-2.70553440e-01 4.11172062e-01 -9.32529986e-01 1.89258575e-01
1.52490675e-01 1.24734056e+00 -4.40056950e-01 -7.21978724e-01
-4.14715409e-01 1.50249565e+00 5.43430805e-01 3.51633787e-01
-6.82110310e-01 -5.26831627e-01 5.17489851e-01 -1.82757378e-01
-3.28007191e-02 -1.80301927e-02 -7.65337110e-01 -1.47332323e+00
-1.05575278e-01 -1.19939351e+00 2.81549275e-01 -8.09361041e-01
-1.57904589e+00 8.48527610e-01 1.36084527e-01 -9.40283716e-01
-5.15260637e-01 -6.34890378e-01 -9.84872520e-01 1.19418132e+00
-1.10721529e+00 -1.67814755e+00 1.09208815e-01 6.35429978e-01
8.32386672e-01 -6.06557846e-01 1.17739451e+00 1.99596792e-01
-8.68131518e-02 1.00098574e+00 1.44601867e-01 6.59029037e-02
1.44591737e+00 -1.46969318e+00 1.05684459e+00 6.51007295e-01
-2.41661295e-02 5.60837924e-01 8.63791049e-01 -5.34759402e-01
-7.37285197e-01 -1.00854743e+00 1.22823942e+00 -7.34061301e-01
7.51273692e-01 -5.51976979e-01 -9.56673324e-01 9.97655749e-01
1.01211357e+00 -7.09037125e-01 6.54605210e-01 2.78762817e-01
-3.90431255e-01 -8.06716755e-02 -1.17726660e+00 9.98046339e-01
1.23452294e+00 -7.59158790e-01 -1.15803564e+00 3.61368090e-01
6.98962510e-01 -1.88732967e-01 -9.69857752e-01 1.83472693e-01
5.55565059e-01 -5.63077867e-01 6.86061442e-01 -9.69814599e-01
8.65002096e-01 2.13077098e-01 1.04969867e-01 -1.93091238e+00
-4.63434339e-01 -9.37150002e-01 5.28386950e-01 1.57502210e+00
5.17924905e-01 -5.06041050e-01 6.21108294e-01 3.45713943e-01
-2.13788643e-01 -1.86353952e-01 -8.02421033e-01 -7.44040310e-01
1.18691373e+00 4.06050645e-02 4.20208693e-01 1.14249921e+00
-1.24133885e-01 9.00923669e-01 -5.54534078e-01 -7.78591931e-01
5.24721205e-01 1.94438745e-03 9.27479029e-01 -1.08475280e+00
-4.66124654e-01 -6.27304375e-01 3.04096431e-01 -8.43502343e-01
6.20391011e-01 -1.26000559e+00 -2.93572366e-01 -1.15034807e+00
1.70674384e-01 -1.96071982e-01 -1.39349818e-01 3.40520799e-01
-1.56710744e-01 2.02190071e-01 3.17156881e-01 1.50114633e-02
-1.89778462e-01 7.46521533e-01 1.48207080e+00 -2.36503452e-01
-6.44697249e-02 2.26147417e-02 -7.49873698e-01 6.66655540e-01
1.11829758e+00 -6.23187542e-01 -3.83259624e-01 -6.54398143e-01
2.30693057e-01 -1.93858787e-01 -6.26934916e-02 -7.06796527e-01
-1.96820855e-01 -2.12349683e-01 2.64932066e-01 -1.63190976e-01
1.40711382e-01 -4.09883261e-01 2.77017597e-02 2.93472499e-01
-8.58706176e-01 3.71981353e-01 3.84654492e-01 -1.48996338e-01
1.89486127e-02 -4.07481432e-01 1.08979356e+00 -3.64425153e-01
-2.80603856e-01 -7.01573491e-02 -3.26933920e-01 4.47831124e-01
5.96629739e-01 2.34721899e-01 -3.22208256e-01 -3.48707438e-01
-6.43906593e-01 -1.46528617e-01 7.16907620e-01 6.31978750e-01
1.11027323e-01 -1.71390021e+00 -1.17530227e+00 1.27532899e-01
1.13069028e-01 -5.23433864e-01 -2.40474284e-01 8.33025947e-02
-5.92532337e-01 2.84164786e-01 -9.38342452e-01 -2.34422535e-01
-1.23200285e+00 4.46846783e-01 3.66684794e-01 -4.10421371e-01
-3.94135416e-01 7.19502747e-01 3.93002182e-01 -8.39935958e-01
-8.21631104e-02 -1.27789497e-01 1.95232242e-01 -1.56638533e-01
4.21729952e-01 3.08731019e-01 3.88243198e-02 -3.38873714e-01
8.22697207e-02 5.18796742e-01 -4.18182611e-01 -6.81466460e-01
1.15709519e+00 -4.80117351e-02 -2.98875600e-01 7.45231867e-01
1.13137853e+00 -7.14715868e-02 -9.95326459e-01 -3.45586628e-01
-6.28628731e-02 -3.28047760e-02 -4.64725316e-01 -9.46264386e-01
-6.44239128e-01 9.72077668e-01 3.66926879e-01 -3.70845288e-01
9.62364256e-01 -2.97141135e-01 7.49666512e-01 8.22794795e-01
3.74925792e-01 -1.40494573e+00 -8.51250142e-02 7.66416490e-01
1.33987522e+00 -8.86934161e-01 -2.29285136e-01 -5.74377850e-02
-7.79266119e-01 9.77516413e-01 7.53002524e-01 -2.66431957e-01
4.31189865e-01 3.10075462e-01 2.30203301e-01 2.38198176e-01
-8.65051270e-01 3.58932018e-01 2.94601798e-01 5.44441521e-01
9.06910598e-01 -4.77176532e-02 -5.62871218e-01 2.70544767e-01
-6.75227940e-01 1.62292346e-01 4.53771174e-01 7.97828317e-01
-9.18231606e-02 -1.72020245e+00 -2.01393396e-01 2.22053885e-01
-5.18771768e-01 -4.66080636e-01 -7.75722384e-01 6.81854427e-01
-3.18293303e-01 5.56664884e-01 -3.94505747e-02 -1.53471768e-01
3.09662193e-01 6.80729449e-01 9.74864304e-01 -6.91449821e-01
-1.14578688e+00 -6.19649030e-02 4.99550641e-01 -1.81867301e-01
-2.79013544e-01 -4.65977579e-01 -4.61530834e-01 -5.22622347e-01
1.95673004e-01 6.89162165e-02 4.29375470e-01 1.04127133e+00
3.16339917e-02 4.03414607e-01 4.30678934e-01 -1.07127345e+00
-6.72119260e-01 -1.29940355e+00 -3.51338923e-01 7.73682594e-01
-2.16267049e-01 -2.28418618e-01 -1.91506177e-01 4.49477464e-01]
|
[11.553722381591797, 9.677913665771484]
|
0ff370f5-2972-4781-90ba-71ae1853ad3a
|
flickr1024-a-dataset-for-stereo-image-super
|
1903.06332
| null |
https://arxiv.org/abs/1903.06332v2
|
https://arxiv.org/pdf/1903.06332v2.pdf
|
Flickr1024: A Large-Scale Dataset for Stereo Image Super-Resolution
|
With the popularity of dual cameras in recently released smart phones, a growing number of super-resolution (SR) methods have been proposed to enhance the resolution of stereo image pairs. However, the lack of high-quality stereo datasets has limited the research in this area. To facilitate the training and evaluation of novel stereo SR algorithms, in this paper, we present a large-scale stereo dataset named Flickr1024, which contains 1024 pairs of high-quality images and covers diverse scenarios. We first introduce the data acquisition and processing pipeline, and then compare several popular stereo datasets. Finally, we conduct crossdataset experiments to investigate the potential benefits introduced by our dataset. Experimental results show that, as compared to the KITTI and Middlebury datasets, our Flickr1024 dataset can help to handle the over-fitting problem and significantly improves the performance of stereo SR methods. The Flickr1024 dataset is available online at: https://yingqianwang.github.io/Flickr1024.
|
['Longguang Wang', 'Wei An', 'Jungang Yang', 'Yulan Guo', 'Yingqian Wang']
|
2019-03-15
| null | null | null | null |
['stereo-image-super-resolution']
|
['computer-vision']
|
[ 4.06688720e-01 -5.83905935e-01 5.85866943e-02 -3.46167356e-01
-1.08325958e+00 -5.08689344e-01 5.77191114e-01 -4.97877896e-01
-1.69554636e-01 6.45724058e-01 5.62011182e-01 1.01151720e-01
1.84091628e-02 -5.49635470e-01 -4.36800510e-01 -5.25865674e-01
3.42301577e-01 -1.55590594e-01 5.45304716e-01 -7.51803070e-02
3.10296714e-01 2.25109980e-01 -1.92850554e+00 5.48869371e-01
9.90230024e-01 9.58166778e-01 6.49006903e-01 5.17397046e-01
5.88190198e-01 5.97134650e-01 1.23707801e-01 -1.26512289e-01
6.20951712e-01 -2.46315897e-01 -7.98990250e-01 1.63836062e-01
1.05957460e+00 -8.22603345e-01 -4.67542350e-01 1.11897838e+00
7.96444952e-01 9.66504663e-02 -9.86867994e-02 -9.89277959e-01
-5.45291364e-01 2.22763404e-01 -7.75687933e-01 3.81391227e-01
7.56607890e-01 1.52334228e-01 9.78412509e-01 -9.39702272e-01
7.30311751e-01 1.11114550e+00 6.44043803e-01 2.64363587e-01
-1.16934168e+00 -9.51991200e-01 -3.62524539e-01 2.41848022e-01
-1.55688357e+00 -7.52164543e-01 6.64489150e-01 -2.50580519e-01
5.47560036e-01 2.72421151e-01 5.93430221e-01 1.06458688e+00
-3.64820331e-01 7.18784332e-01 1.48658431e+00 -1.66751698e-01
-2.13881388e-01 -2.00230747e-01 -9.46391821e-02 1.67807743e-01
8.85882825e-02 3.53976220e-01 -7.85027564e-01 -1.72176331e-01
1.24050605e+00 -4.93584480e-03 -4.45602000e-01 -4.13640052e-01
-1.34184408e+00 5.00564933e-01 4.17334616e-01 2.49084264e-01
-9.95443240e-02 -1.73375130e-01 2.69204378e-01 1.63255364e-01
3.88879418e-01 1.65782064e-01 -2.89251655e-01 -3.05480272e-01
-8.76629889e-01 1.80868149e-01 -6.39684349e-02 1.15372789e+00
7.46999204e-01 -2.95736134e-01 2.11447984e-01 1.22300410e+00
-1.90087184e-02 7.06571639e-01 1.77845687e-01 -1.60727108e+00
6.60594940e-01 2.81247079e-01 2.14031771e-01 -1.20355594e+00
-1.03365347e-01 8.18052441e-02 -8.42691422e-01 -2.20628291e-01
3.20118725e-01 3.24608952e-01 -4.22333539e-01 1.15304494e+00
3.74409080e-01 3.07687908e-01 -2.11800247e-01 1.28353477e+00
1.04208291e+00 3.71264905e-01 -5.53658843e-01 -6.26369193e-02
1.03075814e+00 -9.13621843e-01 -4.23500121e-01 -1.46056801e-01
1.94303378e-01 -1.08293068e+00 1.12456334e+00 5.75136483e-01
-1.06317079e+00 -9.02319610e-01 -5.99932551e-01 -4.30836737e-01
2.75109053e-01 -2.74995193e-02 5.89230001e-01 2.36275852e-01
-1.20284438e+00 4.20592636e-01 -6.28756583e-01 -5.41105151e-01
3.80251586e-01 -1.76881645e-02 -4.92269039e-01 -7.29054332e-01
-1.15537190e+00 4.16417301e-01 2.68768370e-01 -7.13265091e-02
-5.19842505e-01 -6.71249092e-01 -8.93431306e-01 -4.09601331e-01
2.95632094e-01 -4.48347092e-01 1.15323174e+00 -8.98989081e-01
-1.12354302e+00 9.84754205e-01 -2.12959945e-01 -1.10105164e-01
6.34154618e-01 -4.60107744e-01 -5.93211591e-01 4.53024566e-01
2.52159178e-01 9.05347526e-01 6.77242994e-01 -1.28513598e+00
-9.22674417e-01 -4.97711539e-01 1.23203605e-01 3.04734588e-01
-1.49405703e-01 2.08770171e-01 -9.52481925e-01 -6.10928237e-01
2.36372694e-01 -9.28920329e-01 -1.70731321e-01 -2.23842725e-01
-2.96107501e-01 3.14602584e-01 6.72861159e-01 -7.46747077e-01
1.16568518e+00 -2.56478381e+00 -1.02878891e-01 -8.85280818e-02
1.38874859e-01 5.81825785e-02 -2.54251748e-01 2.92460710e-01
-7.15177804e-02 -2.28808463e-01 -9.33515728e-02 -2.19765589e-01
-5.02139926e-01 -4.25576344e-02 -2.85986960e-01 5.92690289e-01
-2.40632430e-01 5.75284004e-01 -1.01225233e+00 -6.23146653e-01
5.79665303e-01 7.08885431e-01 -6.40025496e-01 1.22524858e-01
3.12930167e-01 8.99034202e-01 -3.83404791e-01 7.05720901e-01
1.03795695e+00 -4.03851151e-01 -1.99236646e-02 -4.23855573e-01
-3.10238481e-01 1.48211271e-01 -1.35130858e+00 1.93952358e+00
-2.44265735e-01 8.06711197e-01 -2.81762853e-02 -2.16976196e-01
8.09890628e-01 -2.98288241e-02 6.19542480e-01 -1.22590911e+00
-1.70474753e-01 3.32638323e-01 -3.99612218e-01 -2.54850179e-01
1.00793254e+00 1.02895208e-01 1.60101205e-01 3.02491002e-02
-5.16735673e-01 -2.62600064e-01 2.82796592e-01 1.18281513e-01
8.97320092e-01 9.31302682e-02 2.86623925e-01 -1.28736898e-01
6.19370401e-01 -5.69120608e-02 6.97603464e-01 5.41024148e-01
-1.85054049e-01 1.34799433e+00 -2.88161367e-01 -3.52753371e-01
-1.31922591e+00 -1.05006790e+00 -4.78117019e-01 7.73469031e-01
6.54097497e-01 -5.29584646e-01 -4.54557836e-01 -1.50048271e-01
3.82822286e-03 2.47033745e-01 -2.63475001e-01 4.50402230e-01
-4.49096084e-01 -3.10485691e-01 2.88940281e-01 4.97245729e-01
1.17453051e+00 -8.02058578e-01 -4.00883973e-01 -2.37851769e-01
-7.38988340e-01 -1.59029853e+00 -7.11965919e-01 -7.48241663e-01
-1.00636494e+00 -1.31089187e+00 -7.49587953e-01 -4.18957382e-01
4.19891059e-01 1.19179165e+00 1.04463172e+00 -6.86691981e-03
-2.63611108e-01 5.50838590e-01 -5.56910872e-01 2.70702213e-01
-7.33121634e-02 -1.89158812e-01 1.10741340e-01 -8.38608891e-02
3.53132129e-01 -5.39073944e-01 -8.87135565e-01 6.55580819e-01
-1.03396237e+00 5.47313988e-01 2.77936786e-01 5.46238542e-01
8.81709456e-01 1.79386079e-01 -3.71059775e-02 -7.06741452e-01
1.85343325e-01 -1.09113425e-01 -8.21404576e-01 -1.64147839e-01
-4.22386020e-01 -3.94006461e-01 4.34989423e-01 -7.11932257e-02
-1.24113190e+00 6.98809549e-02 4.82883975e-02 -4.26185459e-01
-2.22233072e-01 1.35936081e-01 -6.87275752e-02 -1.43737823e-01
4.50574517e-01 3.10022652e-01 -7.17457235e-02 -7.66210318e-01
1.45383880e-01 9.12274420e-01 6.61749482e-01 -2.51900464e-01
7.82080591e-01 9.19824541e-01 -2.22753674e-01 -1.11033559e+00
-9.50794816e-01 -8.32699716e-01 -6.43470347e-01 -2.71121055e-01
7.00146317e-01 -1.48272526e+00 -1.92123383e-01 7.13805854e-01
-6.67016566e-01 -3.33046943e-01 -2.95376461e-02 5.00829339e-01
-7.40189910e-01 7.56112635e-01 -6.99700594e-01 -4.85741258e-01
-1.34928137e-01 -1.14947379e+00 1.51163983e+00 1.68744087e-01
1.30405985e-02 -5.92571139e-01 1.22367948e-01 1.05207562e+00
3.09552938e-01 -2.85709389e-02 -1.10139567e-02 2.50524014e-01
-9.01556075e-01 1.24351159e-01 -6.70377791e-01 4.42145169e-01
1.98879585e-01 -1.72996625e-01 -1.03440106e+00 -5.70108533e-01
-1.36587545e-01 -3.88700426e-01 7.19037294e-01 4.41276014e-01
1.31394517e+00 2.55475938e-01 -1.92667879e-02 8.67178380e-01
1.55221498e+00 -1.77004442e-01 9.93757308e-01 6.62426889e-01
8.34148467e-01 3.33468258e-01 1.24320400e+00 4.73566175e-01
6.62835777e-01 1.06881630e+00 2.58368343e-01 -1.39554694e-01
-2.24698216e-01 -3.51355433e-01 3.10699791e-01 9.15531158e-01
-3.44862580e-01 2.38632813e-01 -8.87738585e-01 3.78904045e-01
-1.63633204e+00 -1.06418848e+00 -3.78598511e-01 2.29017091e+00
7.76638746e-01 -2.86895454e-01 1.62514582e-01 7.00482354e-02
8.41344833e-01 4.67162192e-01 -4.11362052e-01 2.93811232e-01
-4.69006389e-01 -1.60115272e-01 5.28967321e-01 5.19783556e-01
-1.16994584e+00 8.87617111e-01 5.86213684e+00 8.71968567e-01
-1.06195903e+00 -5.33737196e-03 6.24037147e-01 -1.35291427e-01
-2.25675792e-01 -9.73018855e-02 -8.69633377e-01 5.49992800e-01
5.59803247e-01 -1.13341317e-01 6.96251214e-01 6.46257401e-01
5.30640006e-01 -4.70965981e-01 -8.55117083e-01 1.61455154e+00
1.14410006e-01 -1.20533228e+00 -1.72350153e-01 2.95916855e-01
1.04070425e+00 5.66013455e-01 1.46544933e-01 -1.50110826e-01
1.39350086e-01 -8.05940747e-01 4.66317058e-01 3.77494663e-01
1.19936097e+00 -6.44869268e-01 6.33467734e-01 1.50598064e-01
-1.37577176e+00 -4.71665673e-02 -4.74088728e-01 -3.37535106e-02
2.17673883e-01 6.66656852e-01 -1.70820251e-01 7.19829500e-01
1.25023472e+00 1.34782135e+00 -8.87163460e-01 1.11634791e+00
-8.37819725e-02 2.40074813e-01 -2.06685543e-01 7.40552902e-01
-2.09311977e-01 -4.03952420e-01 4.51395243e-01 8.27911496e-01
4.88890529e-01 3.20444107e-01 1.15143694e-01 4.94895875e-01
1.41890869e-01 -1.04191065e-01 -7.05004334e-01 1.77028507e-01
6.58584476e-01 1.18141866e+00 -3.87653768e-01 -1.04306705e-01
-6.76061928e-01 1.05211627e+00 -4.48343568e-02 3.11944604e-01
-7.82469213e-01 2.03110382e-01 7.97717690e-01 3.54849219e-01
2.93238252e-01 -3.35064679e-01 -1.09601334e-01 -1.56902242e+00
1.44802675e-01 -1.33357537e+00 4.78691250e-01 -1.20999503e+00
-1.27079403e+00 4.33987588e-01 1.35403564e-02 -1.71884799e+00
2.90240231e-03 -1.81026354e-01 -3.99870872e-02 6.89514816e-01
-1.64985955e+00 -1.00895381e+00 -8.65409613e-01 6.98806524e-01
6.20402277e-01 1.44068450e-01 3.77149940e-01 5.69718122e-01
-3.60478401e-01 2.37767398e-01 4.48832810e-01 -4.56224643e-02
9.70137060e-01 -7.76133478e-01 3.81739974e-01 9.33479369e-01
-1.01264484e-01 5.13858378e-01 6.06465459e-01 -5.67702234e-01
-1.32703412e+00 -9.50540960e-01 5.75488806e-01 -5.34605086e-01
4.62746173e-01 -1.20314561e-01 -7.79008865e-01 5.27849257e-01
-2.28830110e-02 3.06827843e-01 4.45944488e-01 -1.45880401e-01
-2.52716213e-01 -2.84996510e-01 -1.14304841e+00 5.69528878e-01
1.48748958e+00 -7.40993142e-01 -2.27564305e-01 -1.28820658e-01
4.75965291e-01 -5.52855551e-01 -9.87794280e-01 6.39244199e-01
6.63956583e-01 -1.63885093e+00 1.32531118e+00 1.86444297e-01
6.14475489e-01 -4.82783735e-01 -4.60436374e-01 -1.02456450e+00
-4.41208661e-01 -2.40424931e-01 1.05147973e-01 1.17190111e+00
-1.74834356e-01 -5.45927763e-01 6.19894743e-01 5.05833805e-01
1.89558446e-01 -3.03706199e-01 -6.81158066e-01 -8.40382695e-01
-3.95761341e-01 -4.94223803e-01 5.98114967e-01 9.79263902e-01
-3.40501308e-01 1.24633655e-01 -5.83718419e-01 2.44388551e-01
1.02633595e+00 5.38227975e-01 1.13651764e+00 -1.00128913e+00
-2.87216723e-01 1.77260656e-02 -3.74451935e-01 -1.40960121e+00
-2.61846125e-01 -4.89923120e-01 -1.84818089e-01 -1.31912732e+00
6.49706841e-01 -4.62946713e-01 -4.88662645e-02 -3.93093005e-02
-2.78722465e-01 7.95870841e-01 1.51826441e-01 5.38989186e-01
-8.70577931e-01 3.75155300e-01 1.49678183e+00 2.57665843e-01
-2.39631012e-02 -2.23008931e-01 -6.00339174e-01 6.91093028e-01
8.10652018e-01 1.09218448e-01 -3.14007819e-01 -6.74411476e-01
1.75120503e-01 1.85405478e-01 5.08632123e-01 -9.99335945e-01
3.07811908e-02 -1.99210316e-01 3.52746129e-01 -8.68433118e-01
4.67450291e-01 -9.51270401e-01 6.18876815e-01 9.45644006e-02
-1.43577680e-01 7.92108774e-02 1.79377664e-02 4.77314293e-01
-4.99737591e-01 3.52237046e-01 1.04734886e+00 -9.51796472e-02
-1.06241810e+00 3.55113685e-01 2.70821393e-01 2.05154881e-01
7.80129790e-01 -5.36052465e-01 -2.48150259e-01 -5.28580904e-01
-2.74598867e-01 4.50688869e-01 1.21129096e+00 5.89074194e-01
7.51888156e-01 -1.54163361e+00 -6.58324897e-01 3.30360532e-01
3.43669683e-01 2.51626194e-01 6.96370184e-01 8.85128438e-01
-5.91803014e-01 4.42946643e-01 -3.91971081e-01 -8.39836001e-01
-1.68664849e+00 3.67265642e-01 7.64638633e-02 5.57639860e-02
-8.23739588e-01 6.18838549e-01 3.27131480e-01 -3.13468784e-01
-3.57226506e-02 -1.87700987e-01 -1.91489205e-01 -1.86658621e-01
8.55810761e-01 7.74533331e-01 -1.93931654e-01 -9.77922022e-01
-2.70960361e-01 6.94239080e-01 4.32799608e-02 -8.19595605e-02
1.44164395e+00 -7.83101022e-01 -5.13257198e-02 3.53203982e-01
1.10357070e+00 2.29612976e-01 -1.39634907e+00 -4.77226675e-01
-2.41058484e-01 -1.39822650e+00 1.47521958e-01 -3.00254703e-01
-1.11390865e+00 5.85588217e-01 6.57711327e-01 -1.90552846e-02
1.46417952e+00 5.40778134e-03 1.04220676e+00 -1.01166911e-01
1.00370979e+00 -1.08289504e+00 1.77933052e-02 5.70550025e-01
7.96357274e-01 -1.61146426e+00 7.86948949e-02 -7.73665965e-01
-6.91725612e-01 9.04231012e-01 6.35479331e-01 3.31011452e-02
2.58891761e-01 5.36535159e-02 1.09825417e-01 4.03560661e-02
-3.78659278e-01 -3.03401679e-01 1.79029465e-01 5.47953367e-01
4.42581981e-01 -1.62983492e-01 -1.93019316e-01 4.37013470e-02
-3.34619731e-01 4.05443996e-01 5.73780894e-01 6.58182263e-01
-3.00209463e-01 -1.02372527e+00 -5.31828344e-01 3.36598188e-01
-2.75785178e-01 -2.96009094e-01 -1.87059775e-01 5.85425436e-01
-4.36034575e-02 1.03616881e+00 5.76308817e-02 -4.09131050e-01
2.60330349e-01 -6.58449769e-01 5.85579515e-01 -4.58546817e-01
-3.70497964e-02 6.88330606e-02 2.13761076e-01 -1.12882853e+00
-8.12660217e-01 -9.73338962e-01 -8.63802254e-01 -5.89587867e-01
-2.28379101e-01 -2.05635101e-01 1.12055391e-01 4.86069381e-01
5.13214409e-01 -8.97700489e-02 8.71133387e-01 -1.01492071e+00
-1.57322973e-01 -7.67458558e-01 -6.94761693e-01 6.50133729e-01
2.97604680e-01 -5.65505266e-01 -4.15543914e-01 1.61259025e-01]
|
[10.67517375946045, -2.2087149620056152]
|
642ac01b-7625-4c36-bc30-4432d4fb5d97
|
single-node-injection-label-specificity
|
2305.02901
| null |
https://arxiv.org/abs/2305.02901v1
|
https://arxiv.org/pdf/2305.02901v1.pdf
|
Single Node Injection Label Specificity Attack on Graph Neural Networks via Reinforcement Learning
|
Graph neural networks (GNNs) have achieved remarkable success in various real-world applications. However, recent studies highlight the vulnerability of GNNs to malicious perturbations. Previous adversaries primarily focus on graph modifications or node injections to existing graphs, yielding promising results but with notable limitations. Graph modification attack~(GMA) requires manipulation of the original graph, which is often impractical, while graph injection attack~(GIA) necessitates training a surrogate model in the black-box setting, leading to significant performance degradation due to divergence between the surrogate architecture and the actual victim model. Furthermore, most methods concentrate on a single attack goal and lack a generalizable adversary to develop distinct attack strategies for diverse goals, thus limiting precise control over victim model behavior in real-world scenarios. To address these issues, we present a gradient-free generalizable adversary that injects a single malicious node to manipulate the classification result of a target node in the black-box evasion setting. We propose Gradient-free Generalizable Single Node Injection Attack, namely G$^2$-SNIA, a reinforcement learning framework employing Proximal Policy Optimization. By directly querying the victim model, G$^2$-SNIA learns patterns from exploration to achieve diverse attack goals with extremely limited attack budgets. Through comprehensive experiments over three acknowledged benchmark datasets and four prominent GNNs in the most challenging and realistic scenario, we demonstrate the superior performance of our proposed G$^2$-SNIA over the existing state-of-the-art baselines. Moreover, by comparing G$^2$-SNIA with multiple white-box evasion baselines, we confirm its capacity to generate solutions comparable to those of the best adversaries.
|
['Qi Xuan', 'Zhen Wang', 'Shanqing Yu', 'Hongjie Ni', 'Jinhuan Wang', 'Yuqian Lv', 'Jian Zhang', 'Dayuan Chen']
|
2023-05-04
| null | null | null | null |
['specificity']
|
['natural-language-processing']
|
[ 2.98131227e-01 3.78176719e-01 -2.45363176e-01 2.63071179e-01
-5.46086431e-01 -9.93914068e-01 5.43210208e-01 -9.41632167e-02
-3.38574976e-01 6.39162540e-01 -3.87274683e-01 -7.23156929e-01
-7.51221925e-02 -1.09521520e+00 -1.04511344e+00 -7.17881322e-01
-6.42790973e-01 4.08218563e-01 1.86347678e-01 -6.65609479e-01
1.76435988e-02 5.81565022e-01 -5.52975357e-01 -4.83455956e-01
7.29052484e-01 5.59480429e-01 -3.57104957e-01 7.76499033e-01
3.69566411e-01 4.72031534e-01 -1.06654215e+00 -5.69199026e-01
8.53686035e-01 -4.89577711e-01 -5.11566043e-01 -2.60195345e-01
3.63532305e-01 -3.66762727e-01 -8.72223973e-01 1.51992691e+00
5.85659087e-01 9.62799415e-02 2.97760844e-01 -1.71163464e+00
-6.69601023e-01 9.46622074e-01 -7.38129139e-01 2.27830201e-01
1.60335630e-01 7.99818695e-01 9.75059688e-01 6.06680699e-02
5.17637134e-01 1.48338580e+00 5.53049684e-01 1.00272489e+00
-1.64484155e+00 -1.01095176e+00 5.81620872e-01 -3.12306613e-01
-1.14284015e+00 -1.52040422e-01 8.65279973e-01 1.92107186e-01
8.18021119e-01 3.57330859e-01 4.12840515e-01 1.73123097e+00
1.49560779e-01 6.59991026e-01 9.42407370e-01 1.06867656e-01
4.53256309e-01 -2.21466109e-01 -2.22809851e-01 7.94197977e-01
6.28963530e-01 6.29545450e-01 -1.72047634e-02 -6.00398242e-01
8.54507387e-01 -7.46790543e-02 -4.77165699e-01 -5.43369889e-01
-7.67927408e-01 9.88836586e-01 1.04823637e+00 -1.98675156e-01
-3.23982656e-01 9.52674866e-01 4.37300742e-01 5.92483222e-01
-1.62521247e-02 9.21668112e-01 -2.10960925e-01 7.52195045e-02
-2.97718972e-01 3.69027078e-01 9.83074725e-01 7.32853949e-01
6.92138433e-01 6.87856555e-01 -1.43159971e-01 9.47343484e-02
2.32169256e-01 7.90913403e-01 9.33888033e-02 -7.84265935e-01
6.07949793e-01 5.99204779e-01 -1.50707096e-01 -1.34924471e+00
-2.41727382e-01 -7.62664974e-01 -8.28510225e-01 6.34076536e-01
4.91240472e-01 -6.87996149e-01 -1.00252795e+00 2.32681870e+00
4.70565975e-01 4.69457835e-01 1.42502010e-01 6.54328465e-01
4.07235026e-01 3.48449320e-01 6.96399137e-02 1.42359719e-01
8.64158213e-01 -9.07352388e-01 -5.58793172e-02 -6.12548530e-01
5.80098629e-01 1.98393539e-01 1.20311344e+00 1.35814920e-01
-7.98619449e-01 1.54972404e-01 -1.05195427e+00 8.48206997e-01
-3.88819218e-01 -7.36116469e-01 6.65784597e-01 1.01045823e+00
-1.29123580e+00 7.48036325e-01 -9.57764387e-01 -3.06452006e-01
5.47425687e-01 7.44529188e-01 -3.09110999e-01 8.64557698e-02
-1.30364215e+00 5.79010606e-01 2.12732688e-01 1.32580493e-02
-1.84974957e+00 -6.61626160e-01 -8.58502924e-01 1.21023662e-01
1.05038548e+00 -6.35800004e-01 9.85222042e-01 -6.51994288e-01
-1.53805363e+00 3.54629397e-01 6.00213230e-01 -8.69110286e-01
6.45910680e-01 1.50475740e-01 -2.89081693e-01 1.18776798e-01
-1.05833560e-01 4.13757414e-01 1.27495027e+00 -1.45777094e+00
-1.02520078e-01 -2.36855701e-01 6.63721502e-01 2.07308754e-01
-3.38799179e-01 -2.22424358e-01 -2.04577297e-01 -8.08983922e-01
-3.96161765e-01 -1.08514392e+00 -7.17859507e-01 -2.05835879e-01
-8.24466944e-01 1.70961320e-01 1.17323232e+00 -1.86654285e-01
1.11662292e+00 -1.90782166e+00 3.30644041e-01 5.21734834e-01
7.33563006e-01 5.28672516e-01 -6.63243234e-01 5.03054917e-01
3.49642746e-02 6.16821885e-01 -2.96917349e-01 -6.37257174e-02
1.88534886e-01 3.09791774e-01 -5.51131666e-01 5.50110638e-01
2.44044587e-02 1.27339792e+00 -1.20468199e+00 8.48053247e-02
-2.33697310e-01 2.02096403e-01 -6.97046220e-01 2.07993880e-01
-5.14492452e-01 3.45821083e-01 -6.26270533e-01 7.22745478e-01
4.65789974e-01 -2.48683169e-01 3.49891186e-01 2.82427460e-01
8.36136997e-01 -1.21347636e-01 -8.98388624e-01 1.12944388e+00
-1.85470834e-01 1.67201936e-01 4.98720646e-01 -9.89739180e-01
7.43808687e-01 -4.82802950e-02 1.83927447e-01 -5.16109467e-01
3.29525501e-01 8.23052786e-03 3.09986949e-01 1.77018985e-01
8.47882591e-03 1.18199937e-01 -4.05755520e-01 6.72704101e-01
-2.23124087e-01 -2.90920705e-01 3.94661725e-02 5.45109332e-01
1.96819997e+00 -3.62751216e-01 1.70288503e-01 -1.31501764e-01
2.94006169e-01 -1.76940754e-01 5.42121232e-01 1.50977325e+00
-7.08215356e-01 -3.30045931e-02 1.07718658e+00 -3.45879167e-01
-6.61468327e-01 -1.26283598e+00 6.28668070e-01 1.07129991e+00
4.65409517e-01 -3.81956458e-01 -8.80175591e-01 -1.53364277e+00
3.34243447e-01 6.32488489e-01 -9.24687326e-01 -8.05145502e-01
-7.68565238e-01 -8.43269527e-01 1.25860798e+00 2.48911366e-01
7.10140646e-01 -1.20870650e+00 -2.27600098e-01 1.69024318e-01
4.30162042e-01 -7.77317584e-01 -5.85110426e-01 1.48811951e-01
-5.91955304e-01 -1.27837741e+00 -1.96610808e-01 -4.15403336e-01
8.92492592e-01 2.93027252e-01 1.12991643e+00 4.50136155e-01
-2.03563064e-01 5.42984903e-01 -2.11432993e-01 -2.38297090e-01
-7.86823511e-01 3.55621457e-01 2.21841767e-01 -2.10357770e-01
-9.88856032e-02 -8.69554400e-01 -6.37608290e-01 4.03804511e-01
-1.03521049e+00 -6.53795838e-01 5.94705820e-01 8.15377593e-01
4.64133620e-02 1.41082823e-01 5.87378561e-01 -1.12217844e+00
1.14754748e+00 -5.56993246e-01 -9.95518088e-01 1.64396703e-01
-6.47609055e-01 1.78524911e-01 1.20787036e+00 -7.65198290e-01
-3.69375050e-01 -4.15475458e-01 9.79858413e-02 -7.89491892e-01
1.50782168e-01 3.00670534e-01 -2.95452595e-01 -7.95567989e-01
1.28521800e+00 2.76025504e-01 2.39684388e-01 5.94981164e-02
4.86698717e-01 -2.46885836e-01 4.62033570e-01 -8.76656294e-01
1.81786537e+00 2.84814477e-01 2.93269843e-01 -4.09632295e-01
-3.79840314e-01 2.84217715e-01 1.59373432e-01 -6.94665313e-03
3.27356875e-01 -4.73425090e-01 -1.09616148e+00 5.65362990e-01
-5.96502244e-01 -8.97402406e-01 -2.66732156e-01 -1.83800757e-01
-3.38228434e-01 4.81805831e-01 -5.90845287e-01 -5.96999168e-01
-5.47957718e-01 -1.35227847e+00 6.15088105e-01 2.76219815e-01
1.39149189e-01 -1.20843315e+00 1.04758099e-01 5.18288836e-02
8.07582915e-01 7.94409752e-01 9.22821403e-01 -9.30096447e-01
-7.80526459e-01 -2.62750238e-01 8.26705154e-03 2.52739005e-02
7.06415251e-02 -2.25606471e-01 -4.11805004e-01 -1.10706377e+00
-2.04207644e-01 -6.95867598e-01 6.84572160e-01 2.97293309e-02
1.15352154e+00 -7.73296773e-01 -4.49187219e-01 9.40246165e-01
1.25147474e+00 1.83508858e-01 3.38377446e-01 5.04145324e-01
8.67920339e-01 -8.14097524e-02 2.58575231e-01 1.42863125e-01
-2.40181610e-02 2.81437397e-01 1.25781465e+00 -3.28285247e-02
2.14470878e-01 -5.30465126e-01 5.82226396e-01 -8.65768716e-02
3.28733981e-01 -5.52834988e-01 -8.43047857e-01 2.96678841e-01
-1.66001284e+00 -7.79968619e-01 5.49235225e-01 2.05299044e+00
6.70909584e-01 5.28090358e-01 3.53038907e-01 -2.05137029e-01
6.30254686e-01 6.44438684e-01 -1.22323418e+00 -2.67171592e-01
1.32745594e-01 1.66141316e-01 8.69311512e-01 5.29600501e-01
-1.13638079e+00 1.30030704e+00 6.16303921e+00 8.91604602e-01
-1.22281492e+00 -2.22693861e-01 5.33395469e-01 -1.41793966e-01
-3.97239417e-01 9.50896218e-02 -5.70518494e-01 2.06453234e-01
8.45867217e-01 -5.09757042e-01 1.17423296e+00 1.04082763e+00
-1.79197371e-01 4.69106287e-01 -9.74688888e-01 5.12508273e-01
-1.02939881e-01 -1.23532736e+00 9.76787135e-02 3.06057364e-01
5.92005730e-01 2.29778051e-01 5.00149190e-01 7.71653652e-01
1.33638382e+00 -1.23708725e+00 1.25524849e-01 -3.59991252e-01
7.34877765e-01 -9.28387105e-01 1.77784160e-01 3.85549545e-01
-1.03847396e+00 -2.34360546e-01 -2.81684250e-01 2.33301058e-01
-2.14044422e-01 -1.37183011e-01 -1.07905674e+00 4.36331511e-01
4.92722988e-01 2.18567073e-01 -7.04747200e-01 5.46413362e-01
-6.37523890e-01 8.34065974e-01 -4.26401854e-01 -1.78733364e-01
6.50286615e-01 -1.15664847e-01 1.11782265e+00 6.77291811e-01
-7.94823691e-02 -2.01251939e-01 5.04308403e-01 1.06938827e+00
-6.02424085e-01 -2.91509718e-01 -9.88341391e-01 -4.52402413e-01
5.80728114e-01 1.29180920e+00 -5.01393735e-01 7.27234185e-02
5.03922515e-02 8.42345655e-01 7.17107177e-01 5.82990229e-01
-1.22156084e+00 -6.38628662e-01 1.06637168e+00 -7.68214613e-02
9.65276286e-02 -1.04010403e-01 9.21875238e-02 -1.15758014e+00
-1.57058761e-01 -1.68904364e+00 6.15461230e-01 -6.77350210e-03
-1.42160892e+00 7.87772596e-01 4.73598670e-03 -7.63306260e-01
-3.34287614e-01 -4.91096228e-01 -1.09296894e+00 6.76754177e-01
-1.19667065e+00 -1.09936929e+00 -1.28927559e-01 9.96819317e-01
6.67536259e-03 -4.72565442e-01 7.31438100e-01 -1.89093262e-01
-1.02567339e+00 1.32413852e+00 -8.01383182e-02 4.27143902e-01
4.12315845e-01 -1.38083708e+00 1.10559165e+00 1.31977403e+00
5.19250259e-02 7.93888509e-01 8.63574922e-01 -9.14043367e-01
-1.82490790e+00 -1.31926167e+00 -5.22521973e-01 -3.68619949e-01
1.13641214e+00 -5.78042567e-01 -9.02677476e-01 8.87432516e-01
-6.24555387e-02 4.54078674e-01 1.63407832e-01 -9.42354649e-02
-6.45301223e-01 -1.18350148e-01 -1.51212859e+00 1.34820199e+00
1.28159559e+00 -4.34519112e-01 9.75429416e-02 3.67298603e-01
1.11892772e+00 -4.98059750e-01 -5.63544095e-01 4.89524037e-01
-9.29339509e-03 -6.08437717e-01 1.16488278e+00 -1.06302202e+00
-3.51408124e-02 -2.61292726e-01 9.38126519e-02 -1.62800884e+00
-1.56609938e-01 -1.32701075e+00 -5.76696515e-01 6.77785754e-01
4.73175853e-01 -1.17216599e+00 1.24355674e+00 4.43880796e-01
2.59108841e-01 -6.59129322e-01 -9.40379858e-01 -9.71921980e-01
3.46439034e-01 -2.57110354e-02 6.42060280e-01 7.87114143e-01
-1.49970070e-01 8.83280039e-02 -5.29042065e-01 6.05661452e-01
1.03546679e+00 -2.38443598e-01 1.34430659e+00 -6.54572070e-01
-8.70338261e-01 -6.43489778e-01 -3.80359918e-01 -8.01871002e-01
5.24827123e-01 -9.02266204e-01 -1.07755750e-01 -9.62333322e-01
-2.26711780e-01 -3.98476541e-01 -3.00187409e-01 7.13416278e-01
-5.58424771e-01 1.77395687e-01 3.88690323e-01 -2.88793538e-02
-5.95416248e-01 5.28623819e-01 1.21920407e+00 -4.05339181e-01
-2.84190834e-01 1.98678210e-01 -1.07916737e+00 5.49637735e-01
8.64332795e-01 -6.23438299e-01 -7.92315841e-01 -2.34801024e-01
1.51399121e-01 3.83333378e-02 4.24593419e-01 -9.03378129e-01
1.57349452e-01 -4.54936892e-01 2.40525305e-02 2.25917816e-01
9.16818157e-02 -6.67988062e-01 -2.83993669e-02 1.02039516e+00
-2.75636941e-01 2.41885245e-01 2.30563179e-01 1.16872525e+00
3.27477723e-01 8.33001221e-04 9.22974706e-01 -2.47112677e-01
-4.41810429e-01 8.11295807e-01 -1.41718537e-01 4.59727347e-01
1.32445419e+00 7.82708451e-02 -9.31859732e-01 -7.90908635e-01
-5.12668967e-01 5.68664670e-01 6.60321951e-01 2.81608671e-01
5.45517385e-01 -9.78446603e-01 -6.44518554e-01 3.01372409e-01
-7.68249333e-02 4.40465361e-02 -1.22749962e-01 3.90524298e-01
-6.50979280e-01 -2.22343355e-01 -1.19823135e-01 -2.65141040e-01
-9.16953146e-01 1.00024891e+00 7.78460503e-01 -7.84489632e-01
-5.61562300e-01 8.52727592e-01 3.64463508e-01 -7.29321122e-01
3.97335678e-01 1.78026870e-01 1.22912630e-01 -8.50918889e-01
1.78019091e-01 2.76063532e-01 -3.86063427e-01 -1.20422475e-01
-2.75234729e-01 -7.15823695e-02 -3.95025611e-01 -3.37616578e-02
1.03231037e+00 3.46207350e-01 6.26549423e-02 -5.05241811e-01
8.41950536e-01 6.49707839e-02 -1.40048790e+00 -2.73566902e-01
-2.19871148e-01 -5.20713270e-01 -2.22374380e-01 -7.77364612e-01
-1.33127058e+00 4.26635772e-01 1.07748128e-01 5.39009631e-01
1.03828573e+00 -3.23463112e-01 7.08594918e-01 8.04409266e-01
5.99236310e-01 -4.51991320e-01 3.82570475e-01 3.52817655e-01
8.24363470e-01 -1.18068838e+00 -6.40422776e-02 -2.74356097e-01
-5.32851696e-01 6.67910218e-01 1.26591301e+00 -6.27241075e-01
4.54348147e-01 1.87201560e-01 8.29465762e-02 -3.27990413e-01
-6.81509137e-01 2.00453743e-01 -1.41812816e-01 8.03244650e-01
-5.60172558e-01 -4.04070988e-02 2.63759255e-01 2.49117792e-01
-1.00528196e-01 -7.72428989e-01 6.47697031e-01 9.74394262e-01
-1.08352177e-01 -1.29202127e+00 -3.72866869e-01 3.17810357e-01
-5.32183051e-01 5.13265692e-02 -7.14126647e-01 1.12652695e+00
-6.69944763e-01 9.21477675e-01 -5.49492300e-01 -6.04886711e-01
3.19960892e-01 -4.74506557e-01 1.98519334e-01 -4.61648047e-01
-9.84731078e-01 -2.58078247e-01 -5.74761368e-02 -9.02518451e-01
3.65675211e-01 -9.78396237e-02 -9.97228920e-01 -7.24739611e-01
-1.68496579e-01 2.49548584e-01 2.83471316e-01 4.95612293e-01
4.79239047e-01 5.73319435e-01 9.43680346e-01 -8.59688044e-01
-1.42747772e+00 -6.19919062e-01 -5.11678934e-01 4.96101916e-01
3.34803134e-01 -4.90824014e-01 -7.87565470e-01 -9.46802199e-01]
|
[6.140402793884277, 7.349571704864502]
|
6070364d-b3e6-47f6-9198-5c79def1643e
|
position-guided-point-cloud-panoptic
|
2303.13509
| null |
https://arxiv.org/abs/2303.13509v1
|
https://arxiv.org/pdf/2303.13509v1.pdf
|
Position-Guided Point Cloud Panoptic Segmentation Transformer
|
DEtection TRansformer (DETR) started a trend that uses a group of learnable queries for unified visual perception. This work begins by applying this appealing paradigm to LiDAR-based point cloud segmentation and obtains a simple yet effective baseline. Although the naive adaptation obtains fair results, the instance segmentation performance is noticeably inferior to previous works. By diving into the details, we observe that instances in the sparse point clouds are relatively small to the whole scene and often have similar geometry but lack distinctive appearance for segmentation, which are rare in the image domain. Considering instances in 3D are more featured by their positional information, we emphasize their roles during the modeling and design a robust Mixed-parameterized Positional Embedding (MPE) to guide the segmentation process. It is embedded into backbone features and later guides the mask prediction and query update processes iteratively, leading to Position-Aware Segmentation (PA-Seg) and Masked Focal Attention (MFA). All these designs impel the queries to attend to specific regions and identify various instances. The method, named Position-guided Point cloud Panoptic segmentation transFormer (P3Former), outperforms previous state-of-the-art methods by 3.4% and 1.2% PQ on SemanticKITTI and nuScenes benchmark, respectively. The source code and models are available at https://github.com/SmartBot-PJLab/P3Former .
|
['Jiangmiao Pang', 'Dahua Lin', 'Chen Change Loy', 'Tai Wang', 'Wenwei Zhang', 'Zeqi Xiao']
|
2023-03-23
| null | null | null | null |
['panoptic-segmentation', 'point-cloud-segmentation']
|
['computer-vision', 'computer-vision']
|
[ 2.34198391e-01 -2.76953951e-02 -1.01226911e-01 -3.71806562e-01
-8.97369623e-01 -7.71364510e-01 5.57795763e-01 -1.44421101e-01
-1.34824872e-01 2.27164283e-01 -2.15765476e-01 -7.27546886e-02
-5.55644222e-02 -7.61431694e-01 -8.98274958e-01 -8.38997543e-01
1.39401928e-01 8.28919530e-01 8.04696679e-01 -1.74611613e-01
5.04693985e-01 7.16654062e-01 -1.64381838e+00 5.90272769e-02
1.07008421e+00 1.04955018e+00 7.73234904e-01 3.16976458e-01
-3.95632595e-01 7.99184218e-02 -2.54273534e-01 -3.02687317e-01
6.11180961e-01 1.94002017e-01 -8.19489062e-01 1.80391818e-01
8.80869389e-01 -1.78165182e-01 3.33477594e-02 1.15589631e+00
4.48716193e-01 2.43325785e-01 6.16266489e-01 -1.19572532e+00
-4.93486434e-01 2.60748088e-01 -1.02299237e+00 1.27653509e-01
-8.22614785e-03 3.03091764e-01 1.22687757e+00 -1.26527190e+00
6.02012336e-01 1.25810194e+00 5.74920118e-01 1.62087128e-01
-1.17568982e+00 -6.19319975e-01 6.38913274e-01 1.25591353e-01
-1.57658195e+00 -1.25107020e-01 8.69133532e-01 -6.80864453e-01
7.67593145e-01 4.79497671e-01 5.71803451e-01 6.85858428e-01
-1.58402577e-01 7.88905799e-01 8.55028570e-01 6.34544790e-02
-2.88958680e-02 1.46834552e-01 2.07268208e-01 6.47183836e-01
3.46099772e-02 7.51930326e-02 -3.66905004e-01 3.12081482e-02
8.33966672e-01 1.49005234e-01 -4.08897340e-01 -5.45365512e-01
-1.21414006e+00 8.35027575e-01 9.08971965e-01 6.09028116e-02
-4.89941835e-01 7.72258788e-02 5.58738448e-02 -1.90522045e-01
6.37665093e-01 4.44617659e-01 -6.43439412e-01 2.26316452e-01
-1.02866936e+00 4.86991733e-01 4.24856156e-01 1.14345896e+00
1.23318017e+00 -1.11935399e-01 -2.38728046e-01 9.22805130e-01
4.39842910e-01 4.93593782e-01 3.07200663e-02 -1.26601255e+00
3.46489906e-01 8.25276017e-01 1.73338950e-01 -9.87276316e-01
-2.42172301e-01 -6.68461025e-01 -3.54891628e-01 2.51908302e-01
1.75196722e-01 1.94586173e-01 -1.34506297e+00 1.22119391e+00
7.14278519e-01 3.68321061e-01 -4.07398999e-01 1.21168542e+00
1.04658520e+00 7.54855633e-01 -3.92843364e-03 2.17958882e-01
1.48436272e+00 -1.29549468e+00 -2.01893210e-01 -4.04769301e-01
1.40902326e-01 -9.16256964e-01 1.24887037e+00 2.15225220e-01
-9.61650252e-01 -8.06170285e-01 -7.43809044e-01 -2.95043558e-01
-5.16240895e-01 8.69072527e-02 6.26732469e-01 3.11592638e-01
-1.18901300e+00 4.65202004e-01 -6.75097048e-01 -4.86530095e-01
6.99775994e-01 3.02704990e-01 -4.18968387e-02 -3.60532664e-02
-7.13559449e-01 5.37672281e-01 2.39204302e-01 1.27084389e-01
-9.28576171e-01 -1.05129445e+00 -6.22559130e-01 -6.29965588e-02
7.14908898e-01 -8.74909163e-01 1.19346356e+00 -5.26094794e-01
-1.30763328e+00 1.07922947e+00 -3.83032888e-01 -3.76969010e-01
4.58263546e-01 -3.47034931e-01 1.82245642e-01 3.45563322e-01
4.82052237e-01 1.32843506e+00 1.02373886e+00 -1.83053160e+00
-9.76298571e-01 -3.74358416e-01 2.02753276e-01 4.73692983e-01
4.10320312e-01 -1.27098218e-01 -1.11286902e+00 -4.22159642e-01
6.04503930e-01 -8.65795851e-01 -4.94468480e-01 5.59004769e-02
-6.85692906e-01 -3.52798522e-01 1.04375231e+00 -4.72112238e-01
8.87265325e-01 -2.13197207e+00 1.73880652e-01 1.07179202e-01
3.32299888e-01 1.01660704e-02 -8.33243206e-02 3.16781670e-01
1.59479707e-01 2.53793538e-01 -6.24095201e-01 -6.34032488e-01
1.27486676e-01 2.87502319e-01 -5.51624238e-01 3.83391261e-01
2.31024131e-01 1.04354513e+00 -7.02471256e-01 -6.53005242e-01
6.49223566e-01 2.13366255e-01 -5.98526120e-01 1.13545403e-01
-6.04598761e-01 5.72092354e-01 -6.80322647e-01 1.19753408e+00
1.12650216e+00 -2.19844520e-01 -6.15209341e-01 -4.31156874e-01
-6.00619495e-01 1.37393743e-01 -1.00225496e+00 1.75796330e+00
-5.43957166e-02 3.64146411e-01 2.42182985e-01 -8.02855968e-01
7.58875549e-01 -1.79300547e-01 5.91218114e-01 -4.08119500e-01
1.98600516e-02 7.84714371e-02 -2.40654796e-01 -5.51718354e-01
6.74373150e-01 1.97518721e-01 1.74419910e-01 -2.06773773e-01
-4.07032706e-02 -6.06289268e-01 2.70553362e-02 2.09767774e-01
6.48154795e-01 2.82730222e-01 -1.40526146e-01 -2.07010508e-01
3.90241116e-01 3.49483699e-01 6.13466918e-01 8.62734795e-01
-2.51863629e-01 1.11337221e+00 1.32589474e-01 -2.01314583e-01
-5.27387023e-01 -1.32183921e+00 -4.81326938e-01 1.13003421e+00
7.59335458e-01 -3.00325632e-01 -4.79936451e-01 -5.57079792e-01
2.57774711e-01 6.66986108e-01 -5.34608305e-01 2.99974114e-01
-3.81590396e-01 -5.74349999e-01 4.61357348e-02 5.06030977e-01
6.16483569e-01 -1.07253516e+00 -5.72391808e-01 6.66298941e-02
-2.27105588e-01 -1.26347685e+00 -4.36942399e-01 6.57985657e-02
-8.20559382e-01 -9.13686812e-01 -5.74587405e-01 -7.11717010e-01
5.21704555e-01 5.74107230e-01 1.13798487e+00 -1.29159600e-01
-2.73744285e-01 4.95396316e-01 -4.14291531e-01 -5.17303467e-01
3.26658070e-01 2.23807663e-01 -3.29327703e-01 1.15330130e-01
2.90981561e-01 -6.42064691e-01 -8.87247682e-01 3.91863018e-01
-6.53400481e-01 8.65889266e-02 6.05373740e-01 4.06717002e-01
1.19841588e+00 -3.01194787e-01 -2.20183153e-02 -9.05629575e-01
7.17212185e-02 -5.67889571e-01 -8.16184223e-01 -8.52889642e-02
-4.49651837e-01 -4.35467035e-01 9.87187922e-02 -1.47324041e-01
-8.78595710e-01 2.73759723e-01 -2.40384981e-01 -9.02430236e-01
-3.94324571e-01 2.64074743e-01 -1.83381960e-01 -3.09037894e-01
3.43866199e-01 2.60261267e-01 -4.18635041e-01 -7.38702238e-01
5.08330166e-01 4.19951469e-01 4.58722234e-01 -5.58062911e-01
9.87952948e-01 8.67622077e-01 -3.10093045e-01 -9.82688606e-01
-8.81803811e-01 -8.82075012e-01 -7.38836944e-01 -2.07452700e-01
1.03545833e+00 -8.69953096e-01 -4.44577664e-01 3.74316543e-01
-1.18539298e+00 -2.49609232e-01 -3.63444865e-01 1.93901226e-01
-4.65961993e-01 2.98196822e-01 -3.23928207e-01 -7.61702657e-01
-1.90273479e-01 -1.36242676e+00 1.67302465e+00 2.74996310e-01
2.04561099e-01 -5.90801656e-01 -1.15961798e-01 6.03531182e-01
3.22692811e-01 2.90547073e-01 7.58354604e-01 -5.09477794e-01
-1.33221030e+00 1.97294906e-01 -5.79834759e-01 2.67923862e-01
-2.58650798e-02 2.10030988e-01 -1.21424162e+00 -1.36183128e-01
-1.50561705e-03 4.81294356e-02 1.15420759e+00 8.61202896e-01
1.33050013e+00 1.18927352e-01 -7.28832722e-01 1.00215614e+00
1.46867371e+00 1.14892058e-01 4.43811208e-01 1.32143587e-01
1.06028712e+00 7.25198030e-01 8.38542879e-01 1.27390102e-01
6.33659065e-01 8.00997972e-01 9.28257465e-01 -3.09470594e-01
-1.32868469e-01 -1.86378732e-01 6.01299182e-02 4.75184172e-01
-1.80399656e-01 -1.70169488e-01 -1.01613021e+00 7.66586185e-01
-1.84900737e+00 -5.89894712e-01 -3.31110954e-01 1.83862174e+00
5.20018160e-01 9.04066861e-02 -5.19862100e-02 -3.34155828e-01
6.31374538e-01 5.00645638e-01 -7.97968030e-01 -2.48093978e-02
-1.00082137e-01 2.32672676e-01 6.92293406e-01 6.51116788e-01
-1.25500321e+00 1.54748631e+00 5.39081430e+00 9.98241067e-01
-1.12044036e+00 3.08110565e-01 4.66493785e-01 1.00124525e-02
-3.06392610e-01 1.56825542e-01 -9.29208696e-01 3.27379972e-01
3.27692986e-01 2.05627143e-01 2.59348840e-01 8.38702977e-01
2.37127170e-01 -2.23454192e-01 -8.07456136e-01 9.69243228e-01
-1.93550915e-01 -1.27522361e+00 2.06589192e-01 -3.28415297e-02
5.88314235e-01 7.66402543e-01 1.67847320e-01 2.34146759e-01
1.34531260e-01 -9.13048029e-01 8.49266708e-01 4.78850722e-01
5.86928070e-01 -4.23164487e-01 4.01486099e-01 2.56898522e-01
-1.42860508e+00 -2.03040931e-02 -5.88245332e-01 2.11284295e-01
2.17712253e-01 4.71141130e-01 -6.38891935e-01 8.08829486e-01
1.13025939e+00 8.37576628e-01 -6.35328591e-01 1.30798984e+00
-9.86754075e-02 7.07494795e-01 -5.86340308e-01 3.81686509e-01
5.75961232e-01 -6.05236888e-01 9.51474905e-01 1.02800512e+00
2.10345194e-01 -4.19526286e-02 4.23016250e-01 1.34675288e+00
3.29527147e-02 5.06379046e-02 -4.20597821e-01 2.56206691e-01
5.49714148e-01 1.34200859e+00 -9.06759322e-01 -2.46713772e-01
-3.67756844e-01 9.49206054e-01 1.46002993e-01 6.37585521e-01
-9.66203272e-01 -7.27455169e-02 9.43831086e-01 3.02992165e-01
6.91975474e-01 -1.45310983e-01 -3.55075210e-01 -8.88625443e-01
1.26925424e-01 -4.95137930e-01 1.66610643e-01 -8.83096278e-01
-1.30204690e+00 6.13297582e-01 2.86852598e-01 -1.19553733e+00
3.12039912e-01 -4.14528549e-01 -6.59051895e-01 9.78215277e-01
-1.79398680e+00 -1.39386570e+00 -5.16817153e-01 5.06489336e-01
8.98791790e-01 3.11066657e-01 3.80870700e-01 3.42541486e-01
-5.11255503e-01 1.54513031e-01 -1.66128680e-01 -2.14481756e-01
3.80451798e-01 -1.33980942e+00 5.49915910e-01 8.19576800e-01
3.02234381e-01 4.58650559e-01 6.14539862e-01 -6.07326567e-01
-1.06847429e+00 -1.34375989e+00 6.10793769e-01 -7.23560631e-01
4.44010794e-01 -4.86442775e-01 -1.09391725e+00 5.55675030e-01
1.80003256e-01 4.63100187e-02 1.35557592e-01 -9.88142416e-02
-8.45453516e-02 -1.47242218e-01 -1.03179872e+00 6.35102987e-01
1.34692478e+00 -3.23838681e-01 -5.69326162e-01 5.02413273e-01
1.22395170e+00 -6.77387059e-01 -4.59400445e-01 6.84590638e-01
1.32124454e-01 -1.17348361e+00 1.23172140e+00 -7.21691996e-02
4.96168099e-02 -7.32280076e-01 -2.53139615e-01 -1.00242686e+00
-4.06253874e-01 -5.43461144e-01 2.41911635e-01 1.25392747e+00
3.77615035e-01 -8.07265937e-01 7.92421699e-01 1.82505399e-01
-6.40702009e-01 -9.54961300e-01 -9.24624562e-01 -5.44498801e-01
-8.85041058e-02 -5.94983995e-01 6.90240264e-01 8.79614949e-01
-8.05826843e-01 2.01282829e-01 1.31227791e-01 7.16761172e-01
6.31080329e-01 6.20279849e-01 6.89684451e-01 -1.34507692e+00
3.51043791e-02 -5.64236224e-01 -2.48285592e-01 -1.59473705e+00
-9.95429531e-02 -9.59696591e-01 8.05056319e-02 -1.75843227e+00
-9.48606431e-02 -6.69000566e-01 -1.84242725e-01 2.90980399e-01
-1.36221588e-01 3.82861882e-01 2.91382343e-01 4.41096365e-01
-4.78122771e-01 6.44889235e-01 1.43279302e+00 -1.71206385e-01
-3.10668021e-01 2.50945449e-01 -6.67409003e-01 7.84295380e-01
7.81420171e-01 -4.34971392e-01 -4.79801774e-01 -5.92648864e-01
-1.00500286e-01 -2.81590879e-01 6.75287902e-01 -1.03762746e+00
2.24434033e-01 -1.23394847e-01 5.71940243e-02 -1.31880879e+00
7.44397938e-01 -8.98798168e-01 1.93631984e-02 1.08920515e-01
1.08588926e-01 -1.22504488e-01 1.94759503e-01 5.73946357e-01
-2.51895487e-01 -1.52857661e-01 6.19495332e-01 -2.88164914e-01
-9.68694746e-01 7.62188852e-01 -3.59656215e-02 7.49865472e-02
9.88390326e-01 -5.89582562e-01 -1.69334099e-01 -1.29322940e-03
-6.67929292e-01 6.10566199e-01 5.42296052e-01 4.24245596e-01
7.11891651e-01 -9.32114422e-01 -6.00054681e-01 2.27854073e-01
2.23892808e-01 7.24697411e-01 3.71541083e-01 1.04655540e+00
-6.11271679e-01 3.74748230e-01 1.76501617e-01 -1.19009185e+00
-9.95580852e-01 2.56168306e-01 4.59587514e-01 1.79116160e-01
-7.03500986e-01 1.27106798e+00 6.28406107e-01 -5.95315218e-01
3.33660156e-01 -8.21777105e-01 -2.07791060e-01 2.21275941e-01
-6.30518049e-02 1.50692731e-01 -3.66519168e-02 -7.58642852e-01
-4.43840563e-01 1.11342132e+00 -7.88027793e-02 1.81057706e-01
1.21642923e+00 -2.17687935e-01 -1.89653859e-01 4.57606524e-01
9.77154613e-01 1.09375298e-01 -1.58434737e+00 -4.03755993e-01
-1.48647144e-01 -7.39513695e-01 9.59891677e-02 -7.26874232e-01
-1.22312975e+00 9.98127878e-01 6.56734765e-01 2.79462129e-01
9.14233863e-01 4.98930156e-01 8.28671515e-01 5.91264032e-02
3.85925949e-01 -7.09391654e-01 -2.11591884e-01 6.01728141e-01
1.06712663e+00 -1.36905062e+00 -1.17335550e-01 -9.83291745e-01
-6.72815561e-01 5.18921256e-01 8.36133957e-01 -2.60427445e-01
6.50399506e-01 1.23643577e-02 1.43771991e-01 -6.03117585e-01
-4.16616082e-01 -6.85497463e-01 5.06202221e-01 6.12242281e-01
-1.42658889e-01 8.45284015e-02 6.38819858e-02 5.08558691e-01
-3.39775890e-01 -4.55380291e-01 -7.36866146e-02 5.40293217e-01
-6.73449636e-01 -7.62517571e-01 -4.08991277e-01 5.37760556e-01
-1.70212202e-02 -2.03509808e-01 -5.01932085e-01 9.34344649e-01
5.56147933e-01 7.11522758e-01 2.96289355e-01 -2.94963747e-01
4.13315773e-01 -5.93286045e-02 2.72969961e-01 -8.24676394e-01
-5.78284323e-01 2.56194383e-01 -2.73353904e-01 -9.24490571e-01
-5.03426015e-01 -7.19013274e-01 -1.30232823e+00 8.55661482e-02
-4.03464764e-01 -4.01959084e-02 4.94124711e-01 6.08406901e-01
5.86927891e-01 4.43487614e-01 4.88575369e-01 -1.41014946e+00
-1.63702220e-01 -8.48357618e-01 -4.32750851e-01 1.61484197e-01
3.70700985e-01 -8.71212602e-01 -4.17027116e-01 -6.93692639e-02]
|
[7.984828948974609, -3.0314862728118896]
|
4b482d9c-e772-419c-a7ba-e8f937e29b10
|
double-deck-multi-agent-pickup-and-delivery
|
2304.14309
| null |
https://arxiv.org/abs/2304.14309v1
|
https://arxiv.org/pdf/2304.14309v1.pdf
|
Double-Deck Multi-Agent Pickup and Delivery: Multi-Robot Rearrangement in Large-Scale Warehouses
|
We introduce a new problem formulation, Double-Deck Multi-Agent Pickup and Delivery (DD-MAPD), which models the multi-robot shelf rearrangement problem in automated warehouses. DD-MAPD extends both Multi-Agent Pickup and Delivery (MAPD) and Multi-Agent Path Finding (MAPF) by allowing agents to move beneath shelves or lift and deliver a shelf to an arbitrary location, thereby changing the warehouse layout. We show that solving DD-MAPD is NP-hard. To tackle DD-MAPD, we propose MAPF-DECOMP, an algorithmic framework that decomposes a DD-MAPD instance into a MAPF instance for coordinating shelf trajectories and a subsequent MAPD instance with task dependencies for computing paths for agents. We also present an optimization technique to improve the performance of MAPF-DECOMP and demonstrate how to make MAPF-DECOMP complete for well-formed DD-MAPD instances, a realistic subclass of DD-MAPD instances. Our experimental results demonstrate the efficiency and effectiveness of MAPF-DECOMP, with the ability to compute high-quality solutions for large-scale instances with over one thousand shelves and hundreds of agents in just minutes of runtime.
|
['Hang Ma', 'Baiyu Li']
|
2023-04-27
| null | null | null | null |
['multi-agent-path-finding']
|
['playing-games']
|
[-4.93040323e-01 2.80128896e-01 1.01668254e-01 -2.29416624e-01
-7.35161006e-01 -1.32332134e+00 5.94705790e-02 6.86156809e-01
-9.50683355e-02 9.09297466e-01 -1.71798214e-01 -4.50031191e-01
-9.97144938e-01 -1.16803980e+00 -1.14624846e+00 -3.88731629e-01
-6.80298865e-01 1.61302853e+00 4.75381374e-01 -6.28875613e-01
-8.39690492e-02 6.34617090e-01 -1.11607325e+00 -1.06883030e-02
7.01777756e-01 5.61359942e-01 8.61813366e-01 6.22672856e-01
7.44641274e-02 1.55295849e-01 -5.25078177e-01 -2.60588259e-01
6.58213139e-01 1.58588111e-01 -9.83065248e-01 3.12205523e-01
-4.27528292e-01 -3.14031214e-01 -6.13864586e-02 5.32908857e-01
1.40656352e-01 3.19045663e-01 4.33732569e-01 -2.42001057e+00
-4.95516211e-01 7.72240520e-01 -8.23954940e-01 -6.61607906e-02
5.28868139e-01 -2.64299731e-03 7.12482274e-01 -4.02201056e-01
8.56682658e-01 1.29583287e+00 4.72636521e-01 -1.02682032e-01
-1.22197139e+00 8.40686634e-02 5.75497329e-01 1.04717180e-01
-1.16414213e+00 1.69349715e-01 1.87297419e-01 -1.06778368e-01
1.35755754e+00 3.44595551e-01 5.21965206e-01 1.94734395e-01
5.45833826e-01 7.77761817e-01 8.88596773e-01 -4.99141365e-02
4.30086493e-01 -1.71178073e-01 2.75343508e-01 5.89658201e-01
8.28798175e-01 -3.48170847e-01 -2.50218034e-01 -2.79130816e-01
5.21333039e-01 2.34726593e-02 -1.00819781e-01 -8.67563248e-01
-1.43987918e+00 9.85765457e-01 1.93065688e-01 -5.00205338e-01
-7.33830929e-01 5.51683418e-02 3.88089895e-01 4.83349979e-01
-3.36376019e-02 6.77955866e-01 -9.37143862e-01 -1.27108976e-01
-2.85215266e-02 9.41097379e-01 1.33275354e+00 1.73881233e+00
7.36300290e-01 -5.06786764e-01 3.88909698e-01 3.58317733e-01
-1.13373876e-01 6.83947325e-01 -2.93243289e-01 -1.42150593e+00
9.15124953e-01 4.51771200e-01 9.58820641e-01 -9.27586198e-01
-1.13821208e+00 1.35246679e-01 -4.04599100e-01 2.60806710e-01
3.68870497e-01 -1.89273521e-01 -5.47109127e-01 1.36417866e+00
7.41644681e-01 -3.93148214e-01 1.70839325e-01 9.39411938e-01
1.50512546e-01 1.01364934e+00 -5.68170607e-01 -6.03512406e-01
1.51758885e+00 -1.56069756e+00 -6.04981184e-01 -2.90585995e-01
8.29730392e-01 -4.93567824e-01 6.05125725e-01 5.56529582e-01
-1.56866646e+00 1.28371269e-01 -9.65049028e-01 1.06492847e-01
-6.74326420e-01 -5.60305595e-01 8.57063949e-01 3.29054259e-02
-1.15809143e+00 -1.71784665e-02 -8.44393969e-01 -4.66420442e-01
-2.92229921e-01 6.72016680e-01 -6.16064847e-01 -7.24247515e-01
-5.66923141e-01 1.15278447e+00 2.87042320e-01 1.27590224e-01
-9.96922433e-01 -7.04208553e-01 -1.07683420e+00 8.19391498e-05
1.19386494e+00 -6.70570612e-01 1.50087082e+00 2.52991289e-01
-1.32554579e+00 3.19841862e-01 -7.92754665e-02 -1.69116557e-01
2.88422793e-01 2.76345849e-01 -2.62919635e-01 -6.90856902e-03
6.45852625e-01 3.28172296e-01 1.77897327e-02 -1.58000159e+00
-1.11170471e+00 -3.94954652e-01 5.13963580e-01 3.84633750e-01
4.26889122e-01 -2.75499701e-01 -3.19868326e-01 7.89693370e-02
4.92936485e-02 -1.27222633e+00 -5.73718846e-01 -5.99160314e-01
-6.34179175e-01 -3.91788006e-01 5.23658812e-01 -3.79662551e-02
6.43185437e-01 -1.80972791e+00 4.83427346e-01 2.96917468e-01
1.41369984e-01 -4.28689599e-01 -7.11452425e-01 1.13971257e+00
3.72369766e-01 -3.08692664e-01 1.73641499e-02 -2.79231369e-01
6.83409870e-01 9.49341953e-01 -6.95191175e-02 4.26535100e-01
-1.20298304e-01 8.66564274e-01 -1.07712126e+00 9.64668915e-02
-8.52833614e-02 -4.24830168e-01 -5.19421697e-01 -1.19898647e-01
-4.44748342e-01 -3.66997331e-01 -3.83839309e-01 8.28302085e-01
1.18450058e+00 3.09117045e-02 4.44555074e-01 2.27339998e-01
-4.93785560e-01 1.07805304e-01 -1.52238452e+00 1.74599755e+00
-4.01572019e-01 -7.18692690e-03 7.87969887e-01 -5.72443783e-01
6.46681964e-01 -4.32741374e-01 5.98187447e-01 -7.39580572e-01
-1.95119560e-01 3.24254274e-01 -2.06401169e-01 -4.64247555e-01
1.09135473e+00 4.19536263e-01 -7.50584722e-01 8.50465715e-01
-4.64197069e-01 -4.52704817e-01 9.07758832e-01 2.50732154e-01
1.53030062e+00 -1.30864158e-01 -4.35920507e-02 -4.27783430e-01
-1.56406209e-01 7.32376158e-01 5.54286838e-01 7.51905560e-01
-1.57296568e-01 -9.54116881e-02 6.95683599e-01 -6.64825022e-01
-7.73399591e-01 -1.42420292e+00 4.85854983e-01 1.23587954e+00
9.18203950e-01 -2.40134120e-01 -5.98936856e-01 -5.03800273e-01
6.60628974e-01 9.66260791e-01 -3.47231269e-01 4.38459039e-01
-8.06024313e-01 -1.14425981e+00 -2.83314496e-01 2.76170731e-01
-3.17930616e-02 -5.63922524e-01 -7.77142525e-01 6.87097371e-01
-2.96530277e-01 -1.21335638e+00 -6.72928631e-01 8.16285074e-01
-2.05407292e-01 -1.39542007e+00 -3.06087404e-01 -1.18796504e+00
1.05392301e+00 1.02526069e+00 9.56473351e-01 -3.70266765e-01
-1.21053241e-01 4.00546551e-01 -4.63178605e-01 -4.26027924e-01
-1.48116335e-01 3.26056063e-01 1.33022904e-01 -6.51949465e-01
1.50426188e-02 -3.43971342e-01 -3.41420919e-01 8.96836042e-01
-6.40619338e-01 1.40455319e-02 4.80301648e-01 3.67199212e-01
8.71276498e-01 8.13500762e-01 7.54330158e-01 -4.82839495e-01
1.02627277e+00 -7.62040079e-01 -9.51731622e-01 3.66856992e-01
-5.28041124e-01 -2.84783304e-01 4.72375900e-01 -1.59108296e-01
-4.35554773e-01 1.17115386e-01 5.55169046e-01 6.11553118e-02
6.38883412e-02 7.22432435e-01 -9.00844261e-02 7.50127062e-02
-3.10977578e-01 4.77094948e-03 2.72466481e-01 -1.41591564e-01
5.47458708e-01 1.36547172e-02 6.88624740e-01 -7.87462592e-01
6.99527621e-01 3.08390021e-01 3.68499994e-01 -1.67177945e-01
-2.98260272e-01 -2.41433173e-01 -2.69551665e-01 1.02317393e-01
5.15119553e-01 -5.32538116e-01 -1.59093428e+00 3.36481541e-01
-1.16824889e+00 -9.73196626e-01 -1.90546572e-01 2.08786801e-01
-9.74434912e-01 -1.80762202e-01 -6.42215252e-01 -7.41747618e-01
2.28598982e-01 -1.26947093e+00 9.64969575e-01 -5.07445186e-02
1.33108854e-01 -7.40773797e-01 2.10616037e-01 2.50052601e-01
1.64928764e-01 4.97510761e-01 1.15430892e+00 -7.09810913e-01
-1.11276484e+00 4.00216021e-02 -8.00707415e-02 -8.29848886e-01
1.73392147e-01 -3.87578070e-01 3.37908894e-01 -5.64475596e-01
-5.47419310e-01 -1.37072891e-01 -1.23065621e-01 4.82869476e-01
3.03185880e-01 -5.46156406e-01 -8.27393293e-01 9.40581188e-02
1.63354445e+00 6.79349124e-01 1.56994566e-01 1.20023263e+00
-8.24007147e-04 6.31377995e-01 1.36018312e+00 5.76091945e-01
1.54454327e+00 7.78312981e-01 1.01551259e+00 1.15526572e-01
5.13519406e-01 4.31453995e-02 2.12872818e-01 7.23772585e-01
1.22566491e-01 -9.30408835e-01 -7.39603460e-01 9.40386176e-01
-2.40092397e+00 -6.03907347e-01 -4.44551826e-01 1.82294321e+00
3.39226931e-01 2.21480746e-02 5.86408138e-01 -1.67432185e-02
6.54287100e-01 -4.27479446e-01 -5.85378110e-01 -9.55358982e-01
-5.15205413e-02 -4.35186803e-01 9.40786958e-01 7.88020372e-01
-8.53932679e-01 6.24702990e-01 6.14711189e+00 7.78536350e-02
2.41548289e-03 1.76331282e-01 -5.88155314e-02 -3.38600427e-01
-2.80927509e-01 -1.76940680e-01 -8.74896586e-01 4.41286862e-01
7.81920552e-01 -3.54005396e-01 1.15042043e+00 8.00355494e-01
4.11401540e-01 -4.91398126e-01 -1.27380371e+00 5.62556148e-01
-7.30264187e-02 -1.23040521e+00 -3.27342778e-01 3.74627054e-01
7.82381833e-01 -7.47834519e-02 -2.18939751e-01 3.11422855e-01
1.18198681e+00 -4.31169331e-01 9.12847340e-01 1.22250512e-03
-7.86710978e-02 -1.24196815e+00 7.54757404e-01 4.84443903e-01
-1.41837561e+00 -4.29020226e-01 -4.36305583e-01 -1.56072587e-01
1.00832343e+00 3.00384104e-01 -1.19463205e+00 1.08123279e+00
8.20073247e-01 -6.45018220e-02 2.11378440e-01 1.23956215e+00
1.11229084e-01 -6.22556388e-01 -7.68029153e-01 -2.43973479e-01
6.35917604e-01 -5.23168802e-01 5.11629403e-01 8.20894718e-01
4.29267079e-01 3.39078307e-01 8.50886583e-01 4.16141003e-01
3.26513082e-01 -5.14497101e-01 -5.45750082e-01 3.94801319e-01
8.88820171e-01 1.33168495e+00 -1.09520233e+00 1.85893074e-01
-1.07074820e-01 8.54673922e-01 2.16724142e-01 3.20946932e-01
-1.04997885e+00 -6.85017884e-01 1.08351874e+00 9.92272124e-02
6.14346981e-01 -7.50018001e-01 -6.54912964e-02 -2.08032101e-01
9.11468267e-02 -5.22393644e-01 4.36056316e-01 -1.15342343e+00
-1.06821179e+00 5.01141608e-01 3.32753181e-01 -8.36589992e-01
-3.09234321e-01 -5.99640608e-01 -2.75845498e-01 3.80297989e-01
-1.96429479e+00 -1.14999151e+00 -2.97461569e-01 5.12194395e-01
6.34729087e-01 1.61089644e-01 7.09361553e-01 1.53824031e-01
-4.99464422e-01 6.43809736e-02 4.02041256e-01 -6.64698482e-01
3.03426415e-01 -1.45082450e+00 4.15466189e-01 6.25992596e-01
-9.33287621e-01 2.96864510e-01 8.92730355e-01 -6.33630276e-01
-2.46261048e+00 -1.21513319e+00 4.99841273e-01 -2.76588112e-01
8.08905184e-01 -4.60360557e-01 -1.78952485e-01 1.24666643e+00
3.51211250e-01 -3.40980202e-01 3.58113766e-01 -9.70391631e-02
2.47462898e-01 -3.69580567e-01 -1.32405305e+00 4.90107924e-01
1.23661292e+00 3.86940569e-01 -4.33333218e-01 7.45335817e-01
1.15868545e+00 -7.19410777e-01 -8.87769759e-01 -1.14467293e-02
4.26856540e-02 -4.41235304e-01 7.62748241e-01 -3.11397165e-01
-1.34084344e-01 -7.39835382e-01 -4.12365973e-01 -1.96174049e+00
-6.73254907e-01 -9.82814550e-01 1.23742081e-01 1.06208479e+00
7.20495701e-01 -1.11890912e+00 4.22497779e-01 8.45431089e-01
-8.01963210e-01 -7.39377499e-01 -9.32871222e-01 -1.06254840e+00
-2.61645406e-01 7.50126466e-02 1.25756025e+00 8.04450214e-01
2.84856200e-01 -5.06903566e-02 3.22121158e-02 9.91910994e-01
6.36569858e-01 6.41263545e-01 9.77566123e-01 -8.07479799e-01
-1.51893079e-01 -7.08800033e-02 1.55035093e-01 -1.05567992e+00
3.62266861e-02 -8.71592581e-01 3.04068416e-01 -2.41602683e+00
-1.75098907e-02 -9.10352051e-01 3.31071973e-01 7.88516343e-01
5.35563648e-01 -3.07198942e-01 5.79118729e-01 -9.56041459e-03
-1.07267463e+00 1.79799184e-01 1.48311019e+00 -2.25906983e-01
-5.21086156e-01 -7.96808973e-02 -6.69071853e-01 2.96912700e-01
6.42653286e-01 -4.57675159e-01 -6.15382910e-01 -9.32515502e-01
6.12169504e-01 4.96054649e-01 2.21970994e-02 -3.03948760e-01
6.14677668e-01 -7.37436831e-01 -2.50099987e-01 -1.04622626e+00
5.32879710e-01 -1.12406957e+00 4.49727267e-01 4.94179726e-01
2.46243924e-01 1.20186388e+00 3.86945128e-01 5.77385843e-01
2.15996623e-01 -2.22513989e-01 -9.26386490e-02 -1.97735295e-01
-7.81262159e-01 9.16877836e-02 -6.90967619e-01 -2.47586787e-01
2.00243354e+00 -1.37001514e-01 -1.22382164e+00 -1.41550109e-01
-6.95761204e-01 1.38391101e+00 4.20041382e-01 3.22532207e-01
6.70709789e-01 -9.80092764e-01 -5.43505311e-01 -1.02482708e-02
-1.87015414e-01 6.86039507e-01 2.81344861e-01 9.72387910e-01
-8.96537423e-01 4.87930208e-01 -3.21120709e-01 -2.78537929e-01
-8.10633361e-01 1.24833369e+00 -1.29466206e-01 -4.87437755e-01
-5.14554441e-01 5.33047795e-01 3.40905003e-02 -7.07166016e-01
1.12338932e-02 -7.03097165e-01 4.56417501e-01 4.12976593e-02
3.69267583e-01 7.72538662e-01 1.70523524e-01 6.02377579e-02
-6.68178797e-01 2.69994348e-01 -2.65812099e-01 -3.47105041e-02
1.86931908e+00 -5.73189139e-01 -5.12367189e-01 -1.65642574e-01
4.78691250e-01 -1.61235511e-01 -9.67559278e-01 4.23436761e-01
-3.22036296e-02 -1.67379573e-01 -4.41263944e-01 -1.18873096e+00
-8.34889472e-01 -2.85106570e-01 -2.44499967e-01 7.95216203e-01
1.10436249e+00 2.35008478e-01 9.84410346e-01 7.49394834e-01
1.38447785e+00 -1.15821338e+00 -1.66112363e-01 4.12915111e-01
9.53039825e-01 -6.02941811e-01 6.68603228e-04 -8.39143336e-01
-7.95997977e-01 9.07439053e-01 6.66989684e-01 -1.20282032e-01
3.91262472e-01 9.75518763e-01 -2.77010739e-01 -3.07470679e-01
-9.54979479e-01 -1.67509139e-01 -1.05108941e+00 8.68731380e-01
-9.54544842e-01 4.81587499e-01 -1.98489502e-01 7.42527485e-01
-4.16824162e-01 -7.72757158e-02 1.16518867e+00 1.55380309e+00
-5.12946963e-01 -9.62751746e-01 -5.92380404e-01 5.78080453e-02
4.30803180e-01 6.13774657e-01 -8.53489488e-02 1.31105256e+00
1.87471528e-02 1.28657866e+00 5.55645049e-01 -1.08332932e-01
8.37352097e-01 -4.66375381e-01 5.72237611e-01 -2.84694225e-01
-3.51413488e-01 -8.99571106e-02 6.48699343e-01 -6.01751745e-01
-1.24324478e-01 -5.73118746e-01 -1.69537020e+00 -8.95571291e-01
-1.62689269e-01 4.67180073e-01 7.84516990e-01 6.16705060e-01
7.11628616e-01 6.70934618e-01 7.72549391e-01 -1.05659854e+00
-4.24478084e-01 -2.94919670e-01 -8.72151136e-01 -3.21910888e-01
2.47627631e-01 -8.52196157e-01 1.50607213e-01 -4.36242342e-01]
|
[4.960483074188232, 1.7095260620117188]
|
2e7a0a74-553f-455c-bea7-bd22197cfc95
|
self-supervised-learning-based-depth
|
2304.06966
| null |
https://arxiv.org/abs/2304.06966v1
|
https://arxiv.org/pdf/2304.06966v1.pdf
|
Self-Supervised Learning based Depth Estimation from Monocular Images
|
Depth Estimation has wide reaching applications in the field of Computer vision such as target tracking, augmented reality, and self-driving cars. The goal of Monocular Depth Estimation is to predict the depth map, given a 2D monocular RGB image as input. The traditional depth estimation methods are based on depth cues and used concepts like epipolar geometry. With the evolution of Convolutional Neural Networks, depth estimation has undergone tremendous strides. In this project, our aim is to explore possible extensions to existing SoTA Deep Learning based Depth Estimation Models and to see whether performance metrics could be further improved. In a broader sense, we are looking at the possibility of implementing Pose Estimation, Efficient Sub-Pixel Convolution Interpolation, Semantic Segmentation Estimation techniques to further enhance our proposed architecture and to provide fine-grained and more globally coherent depth map predictions. We also plan to do away with camera intrinsic parameters during training and apply weather augmentations to further generalize our model.
|
['Haoyang Pei', 'Mohit Kewlani', 'Akash Mishra', 'Mayank Poddar']
|
2023-04-14
| null | null | null | null |
['self-driving-cars', 'monocular-depth-estimation']
|
['computer-vision', 'computer-vision']
|
[ 1.75205678e-01 8.43557492e-02 -4.04304788e-02 -7.58692026e-01
-1.89289227e-01 -4.42825556e-01 8.41999412e-01 -1.47272646e-01
-4.53071117e-01 4.83193964e-01 8.55920557e-03 -2.94318557e-01
4.09984589e-01 -1.04273474e+00 -5.17017424e-01 -2.57899642e-01
-1.67515278e-02 4.32642877e-01 4.76425886e-01 -1.94151029e-01
3.85632277e-01 8.41778159e-01 -1.76434684e+00 1.69692785e-02
3.32468987e-01 1.11643136e+00 2.72449225e-01 9.34287846e-01
-3.44085433e-02 7.72875965e-01 -2.20362976e-01 -1.37071967e-01
3.81792486e-01 -4.99929376e-02 -6.87767744e-01 3.25706273e-01
6.68287992e-01 -7.90831387e-01 -5.56353509e-01 8.28018486e-01
2.65887320e-01 1.37632072e-01 3.09675246e-01 -1.06424582e+00
-5.04412614e-02 -2.79954206e-02 -5.17322421e-01 1.70457363e-01
5.21531582e-01 2.31697887e-01 4.42044377e-01 -6.62184358e-01
7.03058958e-01 1.13706636e+00 5.62798500e-01 5.13586879e-01
-6.78314030e-01 -4.30931330e-01 2.56621659e-01 4.26946700e-01
-1.00363028e+00 -2.23670438e-01 7.87785769e-01 -2.63860494e-01
1.09828091e+00 -9.24858265e-03 7.07457423e-01 7.85969615e-01
1.08169436e-01 9.32558715e-01 1.11531568e+00 -4.16320890e-01
3.01996142e-01 1.54675037e-01 -2.24912822e-01 7.14037597e-01
5.69261052e-02 5.49054801e-01 -3.98322135e-01 3.68549794e-01
8.97396982e-01 1.38848359e-02 -2.21329823e-01 -7.31117666e-01
-1.31920195e+00 9.06242549e-01 8.31497490e-01 1.88368693e-01
-2.12484136e-01 3.96614552e-01 1.29382238e-01 1.28027707e-01
5.67833722e-01 3.01476955e-01 -6.43470943e-01 -2.71644264e-01
-1.09606874e+00 2.86666036e-01 4.54753280e-01 8.06094289e-01
1.28400314e+00 -4.97308653e-03 3.53998125e-01 2.54989624e-01
4.13128614e-01 2.49933586e-01 4.11302030e-01 -1.47450471e+00
4.39621180e-01 6.46677375e-01 1.48377150e-01 -6.67425275e-01
-6.56251848e-01 -2.47502998e-01 -2.59658396e-01 7.76098073e-01
5.05261719e-01 -2.55870819e-01 -1.00600612e+00 1.25878382e+00
4.73499507e-01 2.80000150e-01 -1.92058843e-03 1.19742024e+00
6.31055355e-01 3.53665590e-01 -2.21864983e-01 4.66066152e-01
9.29723978e-01 -7.85927713e-01 -2.17842981e-01 -7.35857427e-01
8.84786725e-01 -6.55566394e-01 5.74661076e-01 5.72339773e-01
-8.22111130e-01 -6.62304938e-01 -1.20956135e+00 -4.09166992e-01
-5.00248253e-01 1.25087172e-01 1.02563262e+00 8.87864470e-01
-1.14760149e+00 5.88469028e-01 -1.13194418e+00 -5.88664412e-01
2.29162544e-01 3.45162332e-01 -5.30785918e-01 -3.06207240e-01
-1.09772551e+00 1.15454054e+00 3.50396097e-01 1.57351509e-01
-7.96515882e-01 -5.74817002e-01 -1.37028074e+00 -3.81181568e-01
-5.22599407e-02 -9.23813820e-01 1.08247101e+00 -7.92249739e-01
-1.55643702e+00 1.16251683e+00 -2.94924170e-01 -6.81716919e-01
5.65931380e-01 -3.45064253e-01 1.88371554e-01 -4.58080545e-02
-7.00213313e-02 1.43068993e+00 2.50310808e-01 -1.06182182e+00
-9.39959288e-01 -8.65419507e-01 3.41394037e-01 4.96664375e-01
1.37814820e-01 -3.48041743e-01 -4.31706667e-01 -3.43668647e-02
5.66555858e-01 -1.11658740e+00 -4.90122348e-01 4.87258583e-01
-9.56172049e-02 2.80762434e-01 8.48705649e-01 -5.30132353e-01
4.00378734e-01 -1.78477299e+00 1.30168095e-01 -2.51074106e-01
6.42696097e-02 5.91989234e-02 2.05914304e-01 7.64221177e-02
1.51354969e-01 -5.45809448e-01 -8.81484449e-02 -5.51436603e-01
-1.77381247e-01 1.13052882e-01 -8.32943693e-02 6.53850853e-01
9.56192911e-02 9.43319380e-01 -6.94937050e-01 -6.44797683e-02
9.15619195e-01 7.47879028e-01 -4.45288926e-01 -1.04254279e-02
-3.88917387e-01 6.70140803e-01 -2.34785780e-01 5.56182444e-01
9.41416502e-01 2.33833641e-01 -2.21033022e-01 4.68009338e-02
-4.20171678e-01 3.29817533e-01 -1.15030217e+00 1.99616790e+00
-6.79985821e-01 1.29767001e+00 9.72655416e-03 -8.03372979e-01
1.04571450e+00 -1.94860458e-01 4.95175362e-01 -9.61700261e-01
2.33576357e-01 6.98851794e-02 -3.14287513e-01 -2.06134275e-01
9.69267726e-01 -3.54531221e-02 1.32536381e-01 -3.59122828e-02
-1.97898060e-01 -5.65350294e-01 -3.42664182e-01 -1.26739547e-01
8.52409601e-01 8.61252785e-01 1.37481123e-01 2.23760068e-01
6.04163945e-01 4.08471286e-01 1.10905699e-01 2.36689195e-01
-4.56525385e-01 8.36585283e-01 1.35256141e-01 -6.39279962e-01
-9.38533187e-01 -8.45948696e-01 -1.82623476e-01 4.91397083e-01
3.84123564e-01 1.83830902e-01 -6.19043529e-01 -2.53536254e-01
-4.90965247e-02 5.88837147e-01 -5.48107803e-01 2.13332504e-01
-5.50233305e-01 -4.88967627e-01 3.35653067e-01 8.17838252e-01
9.10000741e-01 -7.70437777e-01 -1.11995363e+00 2.50145108e-01
8.71399697e-03 -1.37790334e+00 2.25133881e-01 4.33773816e-01
-1.15314376e+00 -8.79696906e-01 -8.93797040e-01 -4.68240410e-01
2.06825942e-01 4.61470306e-01 8.93632412e-01 -4.53521580e-01
-2.67244130e-01 4.07773823e-01 -1.34792924e-01 -4.59987968e-01
7.75462389e-02 4.89866696e-02 -1.69308230e-01 -3.44194502e-01
8.17196786e-01 -3.54442447e-01 -9.45428967e-01 3.47727239e-01
-5.65280497e-01 1.68287590e-01 2.04644248e-01 2.09637657e-01
4.57199484e-01 -2.46229410e-01 -1.38637871e-01 -7.68703759e-01
-2.11210191e-01 -2.45061338e-01 -1.02261639e+00 -3.67314428e-01
-3.98670763e-01 -3.81003097e-02 1.10291854e-01 1.85036883e-02
-1.11168277e+00 4.63977367e-01 -3.83869380e-01 -4.49492931e-01
-6.89950109e-01 2.05183044e-01 -1.34712949e-01 -4.03872848e-01
7.88478792e-01 8.50396603e-02 -1.06753446e-01 -5.54158948e-02
4.71637368e-01 5.15132606e-01 4.35841650e-01 -4.81551662e-02
6.46907985e-01 9.64240491e-01 1.93399265e-01 -9.07360375e-01
-5.73751330e-01 -7.97840178e-01 -8.62566471e-01 -2.26809651e-01
1.04997194e+00 -1.40179956e+00 -6.35699213e-01 5.10886729e-01
-1.19131374e+00 -7.10896075e-01 -5.71002578e-03 7.40511596e-01
-7.59698331e-01 4.94591117e-01 -5.53756058e-01 -8.36402357e-01
1.72643989e-01 -1.07567465e+00 1.52680743e+00 4.29170460e-01
4.60604429e-02 -1.23543394e+00 3.14890951e-01 6.83945000e-01
3.10996652e-01 3.17806810e-01 1.32950574e-01 2.00925872e-01
-1.13955677e+00 -3.24370474e-01 -3.49604309e-01 1.21422239e-01
-8.01700056e-02 -2.74617761e-01 -1.37180507e+00 -4.38145883e-02
1.25754606e-02 -1.00009710e-01 1.03789854e+00 6.49903297e-01
9.59178507e-01 4.27008122e-01 -4.21258152e-01 9.82662022e-01
1.57470155e+00 2.78079957e-01 1.05203557e+00 7.85634279e-01
8.18587124e-01 9.84614134e-01 8.82922590e-01 3.08349282e-01
8.38341415e-01 8.86275351e-01 8.79846096e-01 -2.35063538e-01
-1.29928455e-01 -1.27612159e-01 1.59781143e-01 -6.49869293e-02
6.80462196e-02 -4.36861739e-02 -9.32735145e-01 6.08251393e-01
-1.59701169e+00 -8.10380459e-01 -3.18338305e-01 2.12008595e+00
1.19460173e-01 2.66065538e-01 5.36053739e-02 2.03751311e-01
2.34146506e-01 -7.38401785e-02 -6.42466724e-01 -4.76031423e-01
-2.03757938e-02 1.12588763e-01 8.60260010e-01 8.18225741e-01
-1.18088841e+00 1.10291278e+00 5.88433218e+00 1.01056337e-01
-1.44647694e+00 -1.78256199e-01 6.94557250e-01 6.99861795e-02
-2.95986444e-01 1.02809362e-01 -1.04987252e+00 4.33497541e-02
8.13646674e-01 5.05460024e-01 2.46986657e-01 9.60462570e-01
2.71481335e-01 -7.83576727e-01 -9.41429079e-01 9.62739348e-01
2.80829854e-02 -1.21790516e+00 -6.22278452e-01 2.75430650e-01
8.31594467e-01 5.87061524e-01 -1.28284581e-02 1.95328280e-01
1.77493051e-01 -9.44985807e-01 4.67591584e-01 2.75136322e-01
5.14064848e-01 -6.67344630e-01 7.05437183e-01 5.72831452e-01
-1.16405141e+00 5.06050624e-02 -5.69977224e-01 -6.53035998e-01
1.67569116e-01 3.99918795e-01 -1.08602023e+00 4.87226903e-01
6.28377318e-01 9.10301268e-01 -5.86863279e-01 1.16842473e+00
-1.55409396e-01 -3.53318490e-02 -4.20139462e-01 7.48515800e-02
3.67657870e-01 -5.28387167e-02 1.32450283e-01 9.80632901e-01
1.86797068e-01 -1.51216149e-01 -6.62069991e-02 7.98626363e-01
4.24110532e-01 -2.02514127e-01 -8.78780305e-01 4.80001658e-01
2.99625192e-02 1.11677086e+00 -7.66500294e-01 -1.26313940e-01
-6.74275100e-01 1.19514656e+00 1.54477313e-01 2.78014600e-01
-7.42277265e-01 -1.45605013e-01 9.83880520e-01 4.04264033e-01
3.95166665e-01 -6.81419432e-01 -7.35391617e-01 -1.23239362e+00
-2.98882246e-01 -1.44229993e-01 -7.27469474e-02 -1.13625860e+00
-5.87421238e-01 4.71151650e-01 -9.19678956e-02 -1.14629722e+00
-5.36568344e-01 -9.79905128e-01 -4.96183664e-01 1.04697371e+00
-2.04321647e+00 -1.16916895e+00 -8.67219925e-01 4.42820579e-01
5.60976505e-01 1.42996252e-01 5.77314436e-01 2.41396025e-01
2.97117587e-02 9.84541848e-02 -2.89853990e-01 -6.88869357e-02
4.35718805e-01 -1.26367664e+00 7.89879084e-01 8.86069715e-01
1.78160697e-01 1.23253301e-01 8.34202647e-01 -5.02898455e-01
-1.08132803e+00 -1.01405513e+00 6.94199979e-01 -7.39696324e-01
3.77532065e-01 -2.32557356e-01 -4.77043301e-01 8.09313893e-01
-1.38872291e-03 7.14693293e-02 1.06815681e-01 -1.59212053e-01
-4.51245308e-02 -3.63313816e-02 -1.18822253e+00 4.15784180e-01
8.64350080e-01 -6.44504905e-01 -6.07406721e-02 7.87298754e-02
5.59834123e-01 -8.64784002e-01 -5.66584468e-01 5.35480320e-01
4.03557152e-01 -1.48587298e+00 1.07575977e+00 9.58820730e-02
3.26153457e-01 -4.39108670e-01 -2.82343388e-01 -9.30289686e-01
1.31204903e-01 -1.28756046e-01 1.05561510e-01 6.85960472e-01
1.56237870e-01 -4.92954046e-01 1.78116500e+00 6.47507131e-01
-2.58974701e-01 -5.45263767e-01 -9.11058068e-01 -3.22476447e-01
1.33850485e-01 -8.45883191e-01 2.90416986e-01 6.75628901e-01
-2.88231045e-01 1.47261739e-01 -2.56676316e-01 2.86321849e-01
6.28398716e-01 2.33813897e-02 1.08366644e+00 -1.22254765e+00
-3.28231126e-01 -2.40015015e-01 -9.73395765e-01 -1.53148365e+00
8.87901802e-03 -4.20990378e-01 1.71751324e-02 -1.71088564e+00
-3.50716650e-01 -4.85853225e-01 3.49729836e-01 8.37967619e-02
7.54078627e-02 6.34701073e-01 -2.74023656e-02 -1.61477774e-01
-3.01998287e-01 4.89806890e-01 1.22634387e+00 1.57880723e-01
-1.54105619e-01 2.78945982e-01 -1.75055012e-01 5.93096614e-01
8.94615114e-01 -4.39539999e-02 -5.53411484e-01 -5.28913975e-01
1.84326500e-01 3.15676332e-01 5.84872007e-01 -1.30003738e+00
2.84861535e-01 -1.00538824e-02 7.47863948e-01 -8.75007570e-01
7.98254728e-01 -8.70662808e-01 4.64595109e-02 4.55442250e-01
1.55669287e-01 -1.21990405e-01 3.17843229e-01 3.29832345e-01
-3.39088053e-01 -1.38685465e-01 5.97717464e-01 -5.15937448e-01
-1.37925482e+00 2.86535114e-01 -3.52202177e-01 -4.17561889e-01
1.17320704e+00 -9.16573107e-01 2.42471844e-02 -4.53796118e-01
-6.89344883e-01 9.07110348e-02 8.88196945e-01 4.80054021e-01
6.98754251e-01 -9.64830399e-01 -3.06844890e-01 2.43419483e-01
1.81808114e-01 2.69402623e-01 1.67502597e-01 3.80901784e-01
-1.20990837e+00 9.04926360e-01 -3.13019723e-01 -9.24869418e-01
-9.41667318e-01 2.95486420e-01 7.65675902e-01 -3.36836353e-02
-6.36743665e-01 8.84069204e-01 3.87362659e-01 -6.29853666e-01
2.61876792e-01 -4.11353588e-01 -1.08431585e-01 -2.52491444e-01
4.58823830e-01 2.33910978e-01 2.97332615e-01 -4.94412631e-01
-4.49730843e-01 8.93743813e-01 3.40932757e-01 -3.70247215e-01
1.16476929e+00 -4.96961296e-01 4.07192916e-01 2.41133422e-01
1.02702391e+00 -3.41975152e-01 -1.91647756e+00 -5.69354109e-02
3.64971310e-02 -5.14105022e-01 4.59607184e-01 -5.89182019e-01
-1.14474511e+00 1.31202960e+00 9.95827794e-01 -3.28676194e-01
1.10215104e+00 -1.03413150e-01 5.70893526e-01 1.79427013e-01
7.39743829e-01 -7.43393898e-01 4.83279526e-02 5.99027097e-01
2.78919458e-01 -1.63510358e+00 -1.81132033e-02 -3.12957585e-01
-5.72159231e-01 1.34786344e+00 8.70226085e-01 -1.41645312e-01
4.48347628e-01 3.10977548e-01 2.98465073e-01 -1.18472680e-01
-4.30079818e-01 -7.21172988e-01 1.46073654e-01 1.06986821e+00
6.84359848e-01 -7.24168345e-02 3.21908891e-01 -3.81221801e-01
-1.91521540e-01 1.61258817e-01 7.06325650e-01 6.72287285e-01
-7.13046193e-01 -1.11627591e+00 -4.29058552e-01 -1.66937366e-01
-4.85026166e-02 7.31833130e-02 -2.90451616e-01 7.77872801e-01
2.97976255e-01 8.20630193e-01 2.66177982e-01 -3.00785244e-01
2.24498019e-01 -2.87931144e-01 8.00182879e-01 -6.40019357e-01
-1.92872152e-01 -2.45836720e-01 -1.83663461e-02 -7.19949007e-01
-5.02453029e-01 -8.15972030e-01 -1.14850032e+00 -4.76953685e-01
-1.41461357e-01 -5.61991453e-01 1.34430683e+00 9.31342661e-01
3.50532010e-02 3.13631237e-01 3.07228982e-01 -1.36883283e+00
1.56613946e-01 -7.91017056e-01 -1.16507664e-01 -2.24983498e-01
3.54734004e-01 -5.60300827e-01 -1.70289651e-01 -1.94069073e-01]
|
[8.585369110107422, -2.3394546508789062]
|
9196d56a-ee65-4e29-8cdd-d97041f1f8e0
|
spending-thinking-time-wisely-accelerating-1
|
2210.12628
| null |
https://arxiv.org/abs/2210.12628v1
|
https://arxiv.org/pdf/2210.12628v1.pdf
|
Spending Thinking Time Wisely: Accelerating MCTS with Virtual Expansions
|
One of the most important AI research questions is to trade off computation versus performance since ``perfect rationality" exists in theory but is impossible to achieve in practice. Recently, Monte-Carlo tree search (MCTS) has attracted considerable attention due to the significant performance improvement in various challenging domains. However, the expensive time cost during search severely restricts its scope for applications. This paper proposes the Virtual MCTS (V-MCTS), a variant of MCTS that spends more search time on harder states and less search time on simpler states adaptively. We give theoretical bounds of the proposed method and evaluate the performance and computations on $9 \times 9$ Go board games and Atari games. Experiments show that our method can achieve comparable performances to the original search algorithm while requiring less than $50\%$ search time on average. We believe that this approach is a viable alternative for tasks under limited time and resources. The code is available at \url{https://github.com/YeWR/V-MCTS.git}.
|
['Yang Gao', 'Pieter Abbeel', 'Weirui Ye']
|
2022-10-23
|
spending-thinking-time-wisely-accelerating
|
https://openreview.net/forum?id=33nhOe3cTd
|
https://openreview.net/pdf?id=33nhOe3cTd
| null |
['board-games', 'atari-games']
|
['playing-games', 'playing-games']
|
[ 7.30641559e-02 -1.64038077e-01 -2.91904986e-01 -9.12747085e-02
-1.04105699e+00 -5.80178678e-01 2.46168777e-01 -2.82898933e-01
-7.99087763e-01 9.85250473e-01 -4.62599695e-01 -5.85530400e-01
-2.19886646e-01 -8.40652764e-01 -3.62908959e-01 -7.08813488e-01
7.06971660e-02 6.89363718e-01 5.57597816e-01 -1.84490025e-01
4.56953257e-01 5.09069599e-02 -1.34331596e+00 -1.42980069e-01
1.09844649e+00 1.12327278e+00 4.53766018e-01 6.02693081e-01
3.40432197e-01 6.68504477e-01 -2.81523317e-01 -7.05578864e-01
6.39802277e-01 -7.29438782e-01 -1.01396883e+00 -3.91150922e-01
-9.30601060e-02 -3.15599382e-01 -4.56330180e-01 1.46447551e+00
6.61092222e-01 3.82335097e-01 1.05577961e-01 -1.20070422e+00
-8.79905820e-02 8.71913552e-01 -8.11676562e-01 5.89463592e-01
2.73107350e-01 3.87051076e-01 1.36035228e+00 -3.02497268e-01
4.19055223e-01 9.32575524e-01 2.58850932e-01 5.90791941e-01
-1.08305132e+00 -9.02772427e-01 2.61982549e-02 7.54074216e-01
-1.61491168e+00 -3.13983768e-01 5.60455978e-01 1.31406814e-01
7.81526208e-01 4.74231511e-01 8.43705356e-01 8.49058211e-01
2.04149351e-01 1.21426857e+00 1.39211643e+00 -2.85242200e-01
6.36190057e-01 -4.30406541e-01 -1.29827857e-01 8.25316250e-01
3.43155116e-01 2.31783763e-01 -6.84390008e-01 -2.11402833e-01
7.74500370e-01 -2.47386262e-01 -1.58712894e-01 -4.18224186e-01
-9.99293268e-01 9.64774728e-01 3.67436618e-01 2.30696931e-01
-3.14305484e-01 7.44560540e-01 1.43492982e-01 2.68704176e-01
2.01202005e-01 5.18712223e-01 -1.46727443e-01 -8.32870901e-01
-1.12594819e+00 6.17077947e-01 6.52444720e-01 7.46683896e-01
3.71805936e-01 3.54754701e-02 7.57158622e-02 4.74322200e-01
1.33841457e-02 5.14667273e-01 4.05312151e-01 -1.41738343e+00
5.94518423e-01 3.88445735e-01 4.74117905e-01 -7.66511619e-01
-3.88605118e-01 -5.77510118e-01 -7.79160202e-01 1.38642356e-01
5.75135291e-01 -1.94417700e-01 -5.83186805e-01 1.85140967e+00
5.67371905e-01 7.87640512e-02 -3.29424441e-01 1.01071906e+00
3.38796467e-01 5.63302696e-01 -2.53639847e-01 -2.90775448e-01
1.29445398e+00 -1.01938665e+00 -6.75479054e-01 -4.88076657e-01
8.03391695e-01 -6.78088844e-01 1.19311273e+00 4.81157362e-01
-1.47435451e+00 -6.38198927e-02 -9.77680922e-01 2.86510110e-01
8.63408446e-02 -4.72635441e-02 1.08766305e+00 1.03790116e+00
-8.77483785e-01 6.34378910e-01 -1.26079202e+00 -6.92274496e-02
4.99319583e-01 5.63321829e-01 1.83787659e-01 -1.46170348e-01
-1.18288100e+00 8.56445312e-01 3.30782861e-01 1.39238730e-01
-9.56691980e-01 2.32610255e-02 -6.27585888e-01 1.78101003e-01
1.17408454e+00 -4.24320161e-01 1.70977879e+00 -2.80974001e-01
-1.68547130e+00 5.33253372e-01 -2.08886132e-01 -6.55881166e-01
8.46662939e-01 -1.76306561e-01 2.03544214e-01 2.09800862e-02
5.77220730e-02 3.24525833e-01 3.27942520e-01 -6.63133621e-01
-6.60224378e-01 -4.43042040e-01 4.37726855e-01 5.91127455e-01
-3.80503573e-03 6.69822022e-02 -4.72221315e-01 -4.27713543e-01
3.21333855e-01 -1.35950637e+00 -6.57238483e-01 -4.05517519e-01
-2.56428838e-01 -4.66835797e-01 -6.10054284e-02 -8.14260766e-02
1.61372125e+00 -1.74738562e+00 1.02718532e-01 9.20236483e-02
1.29495889e-01 2.23091945e-01 -1.14359660e-02 5.60125411e-01
6.09565318e-01 7.23343715e-02 -1.36347607e-01 1.73274130e-02
3.39196086e-01 1.06405728e-01 2.58739274e-02 5.63670099e-01
-6.62837565e-01 1.14367461e+00 -8.93274188e-01 -5.46556234e-01
4.01830040e-02 -2.57257879e-01 -8.96793485e-01 -1.09388813e-01
-3.21332246e-01 2.78974205e-01 -8.43328774e-01 5.10997474e-01
5.45414984e-01 -5.99874973e-01 4.65222210e-01 4.95640010e-01
1.80966929e-02 4.49610889e-01 -1.58246243e+00 1.69038069e+00
-1.03438377e-01 2.64169276e-01 4.34627198e-02 -1.06262207e+00
4.02209312e-01 1.74507946e-02 1.34035617e-01 -8.75374854e-01
6.10614240e-01 4.61728245e-01 4.52055335e-01 -1.21165544e-01
5.60681701e-01 3.24437618e-02 -4.65040177e-01 6.63894892e-01
-5.09376764e-01 -4.26962674e-01 5.23667455e-01 6.23005964e-02
1.28349471e+00 1.81361228e-01 5.53226173e-01 -2.92572945e-01
2.99418390e-01 2.10445806e-01 7.15109527e-01 1.07218778e+00
-4.38228071e-01 -7.45973065e-02 3.50850314e-01 -5.26257575e-01
-6.54321432e-01 -8.13757479e-01 2.25921571e-01 1.33609211e+00
7.28633046e-01 -5.85050046e-01 -8.52180243e-01 -2.23109424e-01
-4.61019337e-01 7.15906382e-01 -5.73098004e-01 -8.13958272e-02
-6.49538040e-01 -7.42893517e-01 5.59017658e-01 3.99820596e-01
9.60287094e-01 -1.04814887e+00 -1.31938839e+00 3.38758640e-02
-5.34824193e-01 -8.62499118e-01 -3.60874683e-01 1.60069808e-01
-8.91981006e-01 -9.00480747e-01 -5.54865003e-01 -4.70980436e-01
3.19797486e-01 4.09854949e-01 8.85997474e-01 2.59300411e-01
-2.41052315e-01 -1.34588704e-01 -5.20221472e-01 -2.25677773e-01
-3.71268690e-02 3.86456102e-01 1.08721778e-01 -5.41571558e-01
2.92567641e-01 -4.89409566e-01 -9.55429077e-01 5.46237946e-01
-5.82565129e-01 2.83485353e-01 5.91536641e-01 9.54858720e-01
5.62682092e-01 4.30185735e-01 3.23810130e-02 -8.02453339e-01
6.87564611e-01 -8.95798653e-02 -1.11022818e+00 1.50980935e-01
-6.21320248e-01 1.46603316e-01 3.72260660e-01 -3.53654534e-01
-9.36415195e-01 -1.16290763e-01 -5.01805320e-02 -1.26660436e-01
2.27182195e-01 4.90312785e-01 1.45819753e-01 -1.28366679e-01
6.88012004e-01 5.38845181e-01 -4.05535311e-01 -3.03757221e-01
8.09816271e-02 3.15673947e-01 1.72729388e-01 -8.03639352e-01
7.94650733e-01 4.03350234e-01 3.15327123e-02 -2.76103616e-01
-9.02964771e-01 -1.63286790e-01 -3.60352686e-03 -1.37117296e-01
5.91259360e-01 -6.52208388e-01 -1.35388172e+00 3.29732537e-01
-6.94953382e-01 -7.26813138e-01 -1.82248175e-01 6.23991251e-01
-8.55176449e-01 4.88766402e-01 -6.47171140e-01 -1.16353083e+00
-3.01705331e-01 -1.33251810e+00 5.87562680e-01 4.06572461e-01
-1.90745220e-01 -4.49768424e-01 -9.83006880e-03 8.88117313e-01
2.35695556e-01 -1.09487474e-01 5.20556748e-01 -3.80229920e-01
-9.85169470e-01 -2.47494340e-01 -1.63954068e-02 -2.76898116e-01
-3.48677605e-01 -6.22215807e-01 -4.79269207e-01 -5.20600975e-01
-3.84092182e-02 -6.30557716e-01 6.91580474e-01 5.52007318e-01
1.23932350e+00 -2.56942421e-01 -2.69337803e-01 5.26285231e-01
1.41689527e+00 6.19379103e-01 6.20513678e-01 6.33969545e-01
1.31839186e-01 1.11535817e-01 9.92512047e-01 8.87977958e-01
3.82525682e-01 9.19545889e-01 5.51060200e-01 3.43381673e-01
3.97016495e-01 -2.41473660e-01 2.66425401e-01 6.71162903e-01
-5.61356664e-01 -4.15136963e-01 -9.99866545e-01 4.64787155e-01
-2.11340284e+00 -1.01298702e+00 1.37146413e-01 2.21454167e+00
1.04568648e+00 4.30530548e-01 1.72798842e-01 3.03030729e-01
5.86985111e-01 2.19579235e-01 -8.72170746e-01 -4.51244354e-01
2.63395667e-01 5.37609875e-01 6.07676029e-01 4.50502515e-01
-7.62944162e-01 1.44166398e+00 5.83208418e+00 1.58215773e+00
-6.08983219e-01 2.18024790e-01 5.47271729e-01 -4.93856221e-01
5.99857904e-02 1.68519869e-01 -6.39414907e-01 4.95494515e-01
6.47288978e-01 -5.70635736e-01 8.43071401e-01 1.03432870e+00
4.95825857e-02 -6.68867886e-01 -7.98363090e-01 1.18804705e+00
-3.14739823e-01 -1.21501553e+00 -3.19400519e-01 3.26591372e-01
7.86193728e-01 -9.94380116e-02 2.59763509e-01 2.84266859e-01
9.32615519e-01 -1.04742885e+00 8.21464837e-01 -2.48661637e-01
6.88756049e-01 -7.97670662e-01 8.83108377e-01 7.80120730e-01
-1.30090463e+00 -1.00533053e-01 -4.72783297e-01 -6.19044065e-01
1.97677150e-01 2.23155767e-02 -5.20330131e-01 4.04826343e-01
8.60262811e-01 -5.50161153e-02 -2.26290599e-01 1.17949200e+00
-4.41379339e-01 5.45583606e-01 -5.24812162e-01 -6.82008266e-01
5.57458699e-01 -3.44203949e-01 4.45924014e-01 5.57448328e-01
3.34454149e-01 6.03382647e-01 3.60937715e-01 6.92144394e-01
-1.39070153e-02 -9.17051956e-02 -1.73268974e-01 -1.30477503e-01
8.35938036e-01 8.44907224e-01 -1.10328948e+00 -1.64199337e-01
6.79653091e-03 9.01659906e-01 3.21376085e-01 -1.10841645e-02
-1.24786150e+00 -3.14429015e-01 3.51366848e-01 -1.15667582e-01
3.86710048e-01 -3.40210170e-01 -4.13560808e-01 -1.18867600e+00
1.62773579e-02 -1.06485891e+00 6.43648207e-01 -4.89065319e-01
-6.55725241e-01 6.35550201e-01 -2.85084713e-02 -1.08781445e+00
-3.32202643e-01 -2.80615538e-01 -2.26076692e-01 4.73188967e-01
-1.16547537e+00 -6.10508263e-01 -1.22666582e-01 6.25391066e-01
7.57309794e-01 9.20510888e-02 5.85000753e-01 -1.16127647e-01
-5.90374410e-01 8.91923249e-01 2.37804055e-01 -1.78723887e-01
3.87511253e-02 -9.48063552e-01 5.17386794e-01 9.23379064e-01
1.59835458e-01 4.71954197e-01 9.44546998e-01 -4.22607839e-01
-1.55713594e+00 -2.54057348e-01 6.24505639e-01 -2.04041675e-01
6.97613478e-01 -1.76511228e-01 -2.49816686e-01 4.48811859e-01
4.97241803e-02 -3.71290743e-01 5.84867179e-01 7.13083595e-02
3.59840989e-02 1.95216373e-01 -9.19826806e-01 1.02703083e+00
1.50599897e+00 -2.00347483e-01 -5.04997671e-01 2.37209812e-01
2.85553128e-01 -7.80643344e-01 -4.50058252e-01 2.91620255e-01
6.31047666e-01 -1.30032420e+00 7.87408352e-01 -1.49308667e-01
2.80200452e-01 -2.26300880e-01 -2.30631143e-01 -9.99356747e-01
-3.26506168e-01 -9.96898115e-01 -1.63880065e-01 3.38908643e-01
3.60959440e-01 -7.68600702e-01 1.31385779e+00 6.49028361e-01
2.94845223e-01 -1.13670588e+00 -1.36056805e+00 -1.08994436e+00
6.31722808e-02 -6.00904584e-01 5.48071384e-01 5.78182936e-01
1.45180464e-01 1.87985167e-01 -5.83745182e-01 -3.94375473e-02
8.07915151e-01 6.81889236e-01 7.61459589e-01 -9.24275458e-01
-6.33298993e-01 -5.29182374e-01 -1.41234323e-01 -1.44618928e+00
-1.97803617e-01 -6.67864978e-01 -3.01734544e-03 -1.46503925e+00
3.75453264e-01 -6.09527886e-01 -2.04895765e-01 4.31111664e-01
-1.04636990e-01 4.12421197e-01 5.17888606e-01 2.80537039e-01
-1.11209702e+00 8.02162707e-01 1.56185913e+00 1.71022490e-01
-4.00194600e-02 1.71026200e-01 -6.75772369e-01 8.64031553e-01
1.04392910e+00 -5.57914376e-01 -4.92744952e-01 -4.75943357e-01
6.17000282e-01 4.10117298e-01 -1.07597083e-01 -1.19929171e+00
4.25809026e-01 -6.38289988e-01 -1.17189176e-01 -5.45553267e-01
6.86469376e-01 -5.76230705e-01 1.91077217e-01 1.06104529e+00
-3.45448107e-01 1.88838392e-01 -8.58035088e-02 4.39501911e-01
1.04603946e-01 -4.06929910e-01 7.17145920e-01 -4.63041365e-01
-6.15852892e-01 3.03637654e-01 -4.44201738e-01 2.18560278e-01
1.14534605e+00 -5.39651632e-01 -2.28129163e-01 -5.39591730e-01
-5.48148453e-01 5.47310054e-01 4.70417380e-01 -7.52930492e-02
4.34964240e-01 -1.11851013e+00 -3.98233414e-01 -3.51038009e-01
-1.94077387e-01 -4.25654883e-03 4.79586720e-01 8.41422915e-01
-7.10965335e-01 6.27434313e-01 -1.29244298e-01 -2.45920286e-01
-1.47522759e+00 3.95738244e-01 1.84220761e-01 -7.98611581e-01
-4.01539803e-01 1.20898104e+00 2.01388806e-01 -7.58004859e-02
1.60794005e-01 -5.38701098e-03 1.23774722e-01 -3.61245483e-01
4.62720364e-01 6.44507170e-01 -3.20698708e-01 -1.72691882e-01
-4.54333067e-01 4.53695476e-01 -4.80293781e-02 -4.61997539e-01
1.29478538e+00 -1.96216419e-01 2.33364806e-01 1.24638505e-01
4.22849417e-01 -2.15774760e-01 -1.09961712e+00 -3.88142407e-01
-1.30659714e-02 -8.41126680e-01 1.77087530e-01 -7.26995587e-01
-9.60809946e-01 7.07105696e-01 4.00277466e-01 2.24707007e-01
1.10017800e+00 -5.99596314e-02 1.01399004e+00 7.36081779e-01
1.31660521e+00 -1.27976930e+00 5.26879877e-02 5.04592121e-01
3.85421485e-01 -9.72689569e-01 1.96006685e-01 -4.28975284e-01
-7.19793558e-01 4.96516109e-01 6.18549585e-01 -2.90408581e-01
2.32289225e-01 5.86661957e-02 -3.18683118e-01 -3.31317395e-01
-1.04123747e+00 -4.94563550e-01 -3.06876779e-01 1.42121434e-01
4.10738625e-02 2.55681157e-01 -8.10015917e-01 6.17893755e-01
-7.12419033e-01 1.35371611e-01 4.00286615e-01 1.15303779e+00
-4.38553005e-01 -1.18985748e+00 -2.43405208e-01 4.60164636e-01
-6.55000329e-01 -2.26919189e-01 -1.91684604e-01 6.24933720e-01
-7.44860545e-02 1.12934589e+00 -2.45470181e-01 -4.24811810e-01
-3.46960053e-02 -2.32976034e-01 8.94350529e-01 -3.69975984e-01
-4.30400610e-01 1.77018754e-02 1.44626722e-01 -9.16156650e-01
-3.34834695e-01 -6.10504806e-01 -1.09390712e+00 -9.06527936e-01
-6.22169912e-01 4.86651540e-01 2.94691026e-01 8.42179894e-01
2.13367581e-01 1.65875316e-01 4.34130341e-01 -7.24722624e-01
-1.00470912e+00 -6.72748089e-01 -6.44710064e-01 -6.25489652e-02
-3.75967741e-01 -1.01559544e+00 -2.23090664e-01 -4.44526613e-01]
|
[3.763401985168457, 1.730046272277832]
|
8acc39ee-58a7-4290-821e-0a626651e5fe
|
efficient-hdr-reconstruction-from-real-world
|
2306.10311
| null |
https://arxiv.org/abs/2306.10311v2
|
https://arxiv.org/pdf/2306.10311v2.pdf
|
Efficient HDR Reconstruction from Real-World Raw Images
|
High dynamic range (HDR) imaging is still a significant yet challenging problem due to the limited dynamic range of generic image sensors. Most existing learning-based HDR reconstruction methods take a set of bracketed-exposure sRGB images to extend the dynamic range, and thus are computational- and memory-inefficient by requiring the Image Signal Processor (ISP) to produce multiple sRGB images from the raw ones. In this paper, we propose to broaden the dynamic range from the raw inputs and perform only one ISP processing for the reconstructed HDR raw image. Our key insights are threefold: (1) we design a new computational raw HDR data formation pipeline and construct the first real-world raw HDR dataset, RealRaw-HDR; (2) we develop a lightweight-efficient HDR model, RepUNet, using the structural re-parameterization technique; (3) we propose a plug-and-play motion alignment loss to mitigate motion misalignment between short- and long-exposure images. Extensive experiments demonstrate that our approach achieves state-of-the-art performance in both visual quality and quantitative metrics.
|
['Jingyu Yang', 'Yihao Liu', 'Qirui Yang']
|
2023-06-17
| null | null | null | null |
['hdr-reconstruction']
|
['computer-vision']
|
[ 5.05524516e-01 -4.40556616e-01 1.52163312e-01 -3.84097427e-01
-8.58171761e-01 -4.08148199e-01 3.39774460e-01 -5.62383950e-01
-4.69144136e-01 3.18017483e-01 1.21953167e-01 -3.14764023e-01
-8.13119188e-02 -6.54598713e-01 -8.31046820e-01 -7.78075755e-01
-5.47577143e-02 -7.19964784e-03 6.69843912e-01 -1.53836235e-01
1.19355418e-01 5.61993837e-01 -1.49365258e+00 -5.53751215e-02
9.19343352e-01 9.86272931e-01 6.75796092e-01 1.00639701e+00
4.22804564e-01 9.90382254e-01 -2.12684497e-01 2.11168647e-01
7.06423402e-01 -5.98328590e-01 -6.08479500e-01 4.38419163e-01
5.58390737e-01 -9.36551452e-01 -8.27622592e-01 7.72084475e-01
7.52259195e-01 7.85565078e-02 -5.14738634e-02 -8.14505756e-01
-5.40701091e-01 1.80053338e-01 -8.63064408e-01 2.27459535e-01
2.89022267e-01 6.35042667e-01 4.37241346e-01 -7.65177906e-01
9.04886305e-01 8.84823978e-01 5.41690052e-01 3.47922593e-01
-1.37084639e+00 -6.61503077e-01 -1.99541584e-01 1.47606596e-01
-1.15374756e+00 -6.11225486e-01 7.38819063e-01 -2.11161017e-01
8.86388123e-01 3.02644968e-01 6.98681176e-01 6.87975883e-01
1.19695634e-01 1.57190949e-01 1.46681821e+00 -4.02245671e-01
1.80336013e-01 -7.07910061e-01 -6.18439801e-02 4.49966043e-01
-1.85430665e-02 3.35512400e-01 -6.38727844e-01 2.42515296e-01
1.29023027e+00 2.63215452e-02 -6.48725092e-01 -3.68570715e-01
-1.54780436e+00 4.76919085e-01 5.43041825e-01 2.79607661e-02
-2.10249066e-01 3.60647082e-01 4.04354408e-02 4.92040098e-01
2.62953818e-01 2.71238208e-01 -1.96726680e-01 1.76629331e-02
-9.91052210e-01 -9.32354294e-03 2.92235076e-01 7.42710948e-01
9.36993480e-01 -6.35801465e-04 8.52386877e-02 8.58024359e-01
2.50912517e-01 7.72444546e-01 2.70948470e-01 -1.35233092e+00
3.34168017e-01 6.06120154e-02 5.35122901e-02 -8.12563300e-01
-4.45080608e-01 -2.26070825e-02 -1.04619050e+00 4.04950142e-01
1.56862155e-01 1.93661854e-01 -1.03203928e+00 1.35687685e+00
3.73631895e-01 2.31310517e-01 4.47600819e-02 1.34787393e+00
7.40229726e-01 1.00300813e+00 -2.30126798e-01 -5.56150973e-01
1.07634628e+00 -9.18365479e-01 -6.62927210e-01 -3.44546974e-01
5.60354367e-02 -8.38776827e-01 1.33220482e+00 4.94891375e-01
-1.18975329e+00 -6.33030117e-01 -1.34089386e+00 -4.64568704e-01
3.10503274e-01 -6.53416067e-02 4.91581827e-01 2.41005227e-01
-1.11197507e+00 6.33615971e-01 -8.59457195e-01 -1.38766885e-01
1.34332538e-01 2.67725348e-01 -2.73733675e-01 -5.41639030e-01
-8.96140218e-01 6.10539079e-01 2.30418816e-01 1.29207432e-01
-8.96364629e-01 -1.02000856e+00 -6.58880889e-01 -4.80421305e-01
4.56185132e-01 -8.49780023e-01 9.29984748e-01 -6.63636744e-01
-1.83673978e+00 9.85555530e-01 1.60758913e-01 -2.83961922e-01
4.92522627e-01 -2.86200106e-01 -3.51282030e-01 3.98398340e-01
-2.88766444e-01 5.97169161e-01 1.02939570e+00 -1.47250807e+00
-4.82970953e-01 -3.30514461e-01 -4.26529437e-01 2.26690531e-01
4.29187082e-02 7.83926398e-02 -8.86625171e-01 -6.31515086e-01
4.78833407e-01 -1.06647003e+00 -3.35562408e-01 2.59952247e-01
-3.80652934e-01 5.88457286e-01 1.08627462e+00 -7.88431048e-01
1.25137663e+00 -2.11913705e+00 1.84249133e-01 -3.58622037e-02
4.91697997e-01 2.77275622e-01 -1.44547269e-01 6.59627095e-02
-6.26294538e-02 -4.30150032e-01 -3.28329831e-01 -2.41310060e-01
-4.25356179e-01 1.50213614e-01 -6.06500447e-01 7.01072276e-01
-2.86098421e-01 7.63293028e-01 -6.73580229e-01 -4.08629894e-01
7.61622429e-01 4.98135895e-01 -3.53992611e-01 6.32980406e-01
-9.81345698e-02 8.41130733e-01 -6.48170635e-02 5.45992076e-01
9.17991340e-01 -3.61259550e-01 2.12626368e-01 -7.91166723e-01
-4.74780381e-01 1.47300847e-02 -1.21486318e+00 1.89941740e+00
-4.48317230e-01 6.49080753e-01 -6.92826360e-02 -3.74508530e-01
8.98825228e-01 -2.66907066e-01 7.62507677e-01 -1.16856015e+00
-5.33427931e-02 3.35959285e-01 -2.23289430e-01 -5.02033710e-01
6.89858675e-01 -7.76291192e-02 -9.97965634e-02 5.08831918e-01
-3.06186080e-01 -5.02071619e-01 -7.29355589e-02 1.72710523e-01
1.20637238e+00 2.43292525e-01 3.02158028e-01 3.09151551e-03
2.49337912e-01 -2.62759000e-01 5.85190713e-01 5.72310448e-01
-1.00065753e-01 1.16044223e+00 1.67105466e-01 -4.30231065e-01
-1.68690145e+00 -1.38448441e+00 -4.38716896e-02 8.80727112e-01
5.00986040e-01 -9.27346125e-02 -3.22077185e-01 -4.66962792e-02
-2.63281792e-01 2.22454950e-01 -1.01814330e-01 7.19007775e-02
-1.02477419e+00 -8.42656314e-01 2.05121830e-01 3.48322868e-01
1.00093853e+00 -8.41277897e-01 -1.02883911e+00 7.22129568e-02
-1.75132900e-01 -1.33356476e+00 -7.10406303e-01 8.76016468e-02
-8.53302538e-01 -8.28256905e-01 -5.67548215e-01 -5.26065290e-01
3.05523306e-01 7.12519467e-01 1.15350080e+00 4.62529697e-02
-5.27597308e-01 1.91392213e-01 -3.94740194e-01 3.18492919e-01
-3.46041828e-01 -3.01661223e-01 -4.63283285e-02 -9.27029103e-02
-2.97470957e-01 -7.55486786e-01 -1.00729644e+00 4.46435928e-01
-1.16030431e+00 5.84475398e-01 6.85017884e-01 6.76065385e-01
1.08850420e+00 8.07436332e-02 2.09719598e-01 -5.14658451e-01
-5.64734451e-02 5.46194240e-02 -9.70878661e-01 1.01934914e-02
-6.78044498e-01 -3.48766036e-02 4.94933546e-01 -5.45804262e-01
-1.08964276e+00 3.09074163e-01 -5.39506078e-02 -7.08402216e-01
1.96821153e-01 -1.60389498e-01 -3.83523941e-01 -3.71994078e-01
4.62604314e-01 4.21843022e-01 1.69361904e-01 -3.02268177e-01
6.00489199e-01 5.72391748e-01 1.19807065e+00 -2.72494078e-01
1.06253505e+00 8.10549021e-01 6.38204813e-02 -1.03427374e+00
-6.83595836e-01 -4.36418861e-01 -4.50518757e-01 -3.99646640e-01
9.92580354e-01 -1.16849852e+00 -7.12182999e-01 9.29123819e-01
-7.10056603e-01 -9.03584182e-01 -2.52989262e-01 3.73062879e-01
-7.23238528e-01 5.33212006e-01 -9.00374234e-01 -2.81665444e-01
-6.19004846e-01 -1.18560755e+00 1.29009533e+00 3.24188173e-01
2.70589381e-01 -4.72629219e-01 2.22719207e-01 3.13146293e-01
5.89379370e-01 2.80379385e-01 6.59419537e-01 4.54460651e-01
-1.20537174e+00 4.65974689e-01 -6.04607582e-01 3.16826791e-01
-5.20301498e-02 -2.39179552e-01 -8.17322850e-01 -5.88787377e-01
5.32394648e-02 -3.25165451e-01 8.99716377e-01 4.98090982e-01
1.20592237e+00 -3.41949202e-02 -9.79721313e-04 1.22025144e+00
1.84687626e+00 4.73399274e-02 1.37504339e+00 5.77625275e-01
1.05618131e+00 2.69494802e-01 8.14289629e-01 4.66435462e-01
3.02632362e-01 1.17981195e+00 2.83253342e-01 -4.31601226e-01
-6.20707035e-01 -1.27767026e-01 5.82756221e-01 8.74499440e-01
6.44023493e-02 -1.25124976e-01 -6.58379257e-01 2.60495961e-01
-1.62208021e+00 -8.87897074e-01 -3.02632093e-01 2.25059962e+00
9.54118490e-01 -2.91174948e-01 7.51873478e-02 -7.62025639e-03
4.44036245e-01 6.12531483e-01 -7.59347558e-01 3.06594104e-01
-4.20889676e-01 9.10370424e-02 9.54458594e-01 4.73308444e-01
-9.10782576e-01 8.13404024e-01 6.18481874e+00 4.98179644e-01
-1.41080034e+00 7.70608857e-02 5.61262250e-01 -4.09434646e-01
-1.66674629e-01 5.63242994e-02 -6.28326654e-01 6.11797988e-01
8.92633319e-01 1.35420218e-01 7.11273193e-01 5.63414335e-01
4.22490269e-01 -3.12856972e-01 -9.65586901e-01 1.45294464e+00
1.16821401e-01 -1.22355735e+00 -2.80608356e-01 2.32660264e-01
6.68289661e-01 1.58855498e-01 1.75303131e-01 -2.89846331e-01
2.76215494e-01 -8.08839202e-01 7.12946653e-01 6.07612550e-01
1.23721635e+00 -6.47342324e-01 2.08095267e-01 -6.16223216e-02
-1.15330231e+00 -5.27994335e-02 -4.52504098e-01 4.04974073e-01
6.18801057e-01 9.10742223e-01 -2.79647380e-01 4.55015004e-01
1.01135695e+00 7.47092307e-01 -7.50462174e-01 7.02431321e-01
-6.78578168e-02 3.77738595e-01 -2.43854269e-01 7.35436559e-01
-3.84336621e-01 -3.07988554e-01 4.66031551e-01 8.88994157e-01
5.03674150e-01 4.42758888e-01 2.24853858e-01 6.27838135e-01
-6.09354079e-02 -4.96040165e-01 -3.99652630e-01 3.27758729e-01
5.43663800e-01 1.24714124e+00 -4.60030049e-01 -6.81126490e-02
-3.94984037e-01 1.46648788e+00 -7.95086548e-02 1.89676940e-01
-9.07196939e-01 -3.87327760e-01 5.89495420e-01 4.05424774e-01
4.21406269e-01 -4.31499630e-01 -2.04496354e-01 -1.24907649e+00
1.10498518e-01 -9.74502623e-01 2.39904299e-01 -1.15540826e+00
-1.12530339e+00 4.46853578e-01 -1.30165473e-01 -1.37052631e+00
-9.88202840e-02 -2.84310043e-01 -2.10110039e-01 5.73043406e-01
-1.80379570e+00 -1.17870533e+00 -9.15137410e-01 6.21552527e-01
4.47947025e-01 2.19883710e-01 2.30631277e-01 6.29956603e-01
-5.18166602e-01 1.63701952e-01 2.43805408e-01 -1.15267508e-01
9.61214006e-01 -9.17457700e-01 3.57839525e-01 1.25865054e+00
-2.10178003e-01 3.82884949e-01 8.57572019e-01 -5.85674763e-01
-2.12142658e+00 -1.18625939e+00 1.53689861e-01 -2.08031327e-01
4.42327052e-01 -3.07774305e-01 -1.09478664e+00 6.07304990e-01
7.81580893e-05 3.54063749e-01 2.51773804e-01 -5.79834342e-01
-3.73757988e-01 -5.54073930e-01 -1.10788774e+00 5.23614466e-01
1.30174220e+00 -5.41038930e-01 -1.58036783e-01 2.05241498e-02
1.08856773e+00 -6.86248004e-01 -1.11072505e+00 3.68848383e-01
5.76954126e-01 -1.25752091e+00 1.32774007e+00 3.47246766e-01
5.54493129e-01 -8.73544872e-01 -3.16430300e-01 -1.02853596e+00
-2.08221152e-01 -8.62914264e-01 -3.14842284e-01 1.16134179e+00
-1.54814601e-01 -4.67119604e-01 6.01443470e-01 4.05066222e-01
-2.30212420e-01 -4.46470231e-01 -7.57255673e-01 -7.16212630e-01
-3.76516581e-01 -4.01757717e-01 2.80723274e-01 6.86584055e-01
-6.00678086e-01 2.03274161e-01 -9.37995553e-01 3.29225123e-01
1.09237397e+00 3.72601777e-01 1.05289137e+00 -7.00044990e-01
-8.01112175e-01 2.23636180e-01 -2.15768874e-01 -1.46229053e+00
-3.65269482e-01 -1.95538476e-01 4.24556345e-01 -1.29228151e+00
3.69826347e-01 -6.78984284e-01 2.19626278e-01 1.52654484e-01
-7.33623728e-02 5.85857987e-01 4.20652390e-01 6.36468053e-01
-7.90338099e-01 3.92288417e-01 1.25287092e+00 3.30118537e-01
-3.76554042e-01 -5.53756118e-01 -3.20536315e-01 4.13084120e-01
3.71826828e-01 -1.70724824e-01 -3.82543504e-01 -6.49265051e-01
1.64087251e-01 4.38148171e-01 5.60849845e-01 -1.16767311e+00
1.93402469e-01 -1.51198179e-01 5.96984804e-01 -6.75030589e-01
1.60094038e-01 -7.46457696e-01 6.27698481e-01 3.34888101e-01
-2.96195038e-02 -3.63868079e-03 -1.25990465e-01 5.36062002e-01
1.34166896e-01 3.44941705e-01 1.47105336e+00 1.87291145e-01
-1.12524855e+00 4.87123579e-01 2.10226309e-02 -1.01529747e-01
1.02731848e+00 -1.12989537e-01 -6.41958177e-01 -2.44678095e-01
-1.66119918e-01 -7.00781345e-02 1.10195148e+00 2.32885003e-01
8.55749786e-01 -1.22051764e+00 -5.45770347e-01 2.27387741e-01
3.01522035e-02 2.38894850e-01 7.51382709e-01 7.41009712e-01
-1.02197134e+00 -1.93257499e-02 -2.93294609e-01 -7.95513809e-01
-1.36908054e+00 6.51311457e-01 3.99468571e-01 -3.28952968e-01
-1.30184257e+00 4.67475921e-01 1.87172256e-02 -1.01002514e-01
-1.79630905e-01 -1.21213868e-01 3.83641094e-01 -3.91759813e-01
8.19791198e-01 4.90084052e-01 -4.78028245e-02 -5.76956809e-01
-1.36466175e-01 9.58749592e-01 -1.05939554e-02 -3.10363114e-01
1.56941259e+00 -7.57046640e-01 -7.31492341e-02 2.09118471e-01
1.27800691e+00 -1.15873538e-01 -1.77973461e+00 -2.55665213e-01
-4.08554643e-01 -9.52659428e-01 3.54869068e-01 -4.66066241e-01
-1.22521174e+00 5.30849874e-01 1.16703761e+00 -2.34117404e-01
1.77771854e+00 1.44929543e-01 1.22748971e+00 5.78542165e-02
4.92298245e-01 -9.89761353e-01 4.14050102e-01 2.60452241e-01
6.53701186e-01 -1.14008713e+00 3.74420643e-01 -4.24724609e-01
-5.12523413e-01 1.06146145e+00 5.19158959e-01 -3.74054722e-02
2.25588843e-01 4.87310112e-01 1.10442810e-01 -1.27366260e-01
-5.87568521e-01 2.77782939e-02 -7.10375682e-02 7.31932402e-01
1.61200121e-01 -3.47287297e-01 -3.38647328e-02 -7.02811033e-02
-4.54263203e-03 2.01503664e-01 8.08659077e-01 9.48098719e-01
-4.64436769e-01 -9.82538760e-01 -4.25165266e-01 1.93104044e-01
-7.64488652e-02 6.33326247e-02 2.77929038e-01 5.58911324e-01
-1.56594649e-01 6.48248136e-01 1.80900991e-01 -4.57553059e-01
3.70197654e-01 -5.98938525e-01 6.82005763e-01 -2.07621589e-01
-9.04151201e-02 1.37996927e-01 -1.84307382e-01 -1.11415911e+00
-4.91703629e-01 -4.11204755e-01 -1.29845250e+00 -5.15122414e-01
3.86627652e-02 -6.17728889e-01 7.32912362e-01 6.26414061e-01
5.16703188e-01 6.28008842e-01 9.78761375e-01 -1.19956791e+00
-3.88760298e-01 -6.17729068e-01 -6.70585155e-01 4.07692045e-01
5.35878062e-01 -2.85479456e-01 -4.30643082e-01 3.78466278e-01]
|
[10.812997817993164, -2.227518081665039]
|
72b2b171-8d42-415c-85d3-2f1a98e02733
|
feature-mixing-for-writer-retrieval-and
|
2306.12939
| null |
https://arxiv.org/abs/2306.12939v1
|
https://arxiv.org/pdf/2306.12939v1.pdf
|
Feature Mixing for Writer Retrieval and Identification on Papyri Fragments
|
This paper proposes a deep-learning-based approach to writer retrieval and identification for papyri, with a focus on identifying fragments associated with a specific writer and those corresponding to the same image. We present a novel neural network architecture that combines a residual backbone with a feature mixing stage to improve retrieval performance, and the final descriptor is derived from a projection layer. The methodology is evaluated on two benchmarks: PapyRow, where we achieve a mAP of 26.6 % and 24.9 % on writer and page retrieval, and HisFragIR20, showing state-of-the-art performance (44.0 % and 29.3 % mAP). Furthermore, our network has an accuracy of 28.7 % for writer identification. Additionally, we conduct experiments on the influence of two binarization techniques on fragments and show that binarizing does not enhance performance. Our code and models are available to the community.
|
['Robert Sablatnig', 'Marco Peer']
|
2023-06-22
| null | null | null | null |
['retrieval']
|
['methodology']
|
[ 1.39140449e-02 -5.24804235e-01 -2.93067694e-01 -1.00801162e-01
-1.16594946e+00 -6.63889468e-01 1.04250884e+00 -2.23352194e-01
-3.08693975e-01 3.18731219e-01 1.02405161e-01 1.74908340e-01
-1.58203796e-01 -5.84046781e-01 -5.15298963e-01 -7.57392049e-01
6.53897300e-02 5.24270058e-01 -2.56974827e-02 -1.87279761e-01
6.13335252e-01 9.35438037e-01 -1.38299835e+00 5.71369588e-01
2.50054657e-01 1.21122158e+00 -2.30332334e-02 7.46837199e-01
-7.54817501e-02 5.93317986e-01 -1.03000915e+00 -7.44033396e-01
3.96352112e-01 -9.93503630e-02 -5.23068309e-01 -1.24678552e-01
1.03992212e+00 -7.13370979e-01 -8.08527052e-01 6.46777689e-01
7.07626164e-01 -7.27827102e-02 1.01484311e+00 -7.41599143e-01
-9.87809837e-01 6.16510928e-01 -9.37455833e-01 3.24319512e-01
4.74139675e-02 -2.84640104e-01 1.02963603e+00 -1.26978445e+00
5.35556257e-01 1.19526541e+00 6.82701111e-01 3.31656516e-01
-7.93590307e-01 -8.73035669e-01 -2.40961060e-01 4.68274623e-01
-2.01273727e+00 -7.51744688e-01 5.25222600e-01 -1.80101469e-01
8.73446226e-01 3.54587883e-01 1.27207771e-01 1.11004448e+00
3.51205945e-01 9.67496455e-01 8.13288927e-01 -4.66288149e-01
-2.36167237e-01 3.79060023e-02 4.24213886e-01 6.72986031e-01
1.58912420e-01 -1.24435453e-02 -1.08398831e+00 1.59291863e-01
7.96868205e-01 3.66170667e-02 1.76710114e-01 1.75411671e-01
-7.34812319e-01 7.86785066e-01 3.28486681e-01 2.18513921e-01
-1.57985836e-01 1.20310783e-01 3.04106057e-01 1.87201321e-01
1.72096938e-01 3.57663661e-01 2.62673855e-01 -7.68907592e-02
-1.67761004e+00 5.22775829e-01 8.56276393e-01 9.31337535e-01
4.57854897e-01 -8.31966102e-02 -6.57998323e-01 1.13882065e+00
1.00302801e-01 8.77270162e-01 4.33591634e-01 -8.96551073e-01
3.84237319e-01 3.11840057e-01 9.24122855e-02 -1.34664774e+00
-2.71237105e-01 -6.61322594e-01 -7.19669938e-01 -2.63893194e-02
4.00127172e-01 2.50111639e-01 -1.07744372e+00 1.07896388e+00
-5.95165610e-01 -2.98320323e-01 -2.37002105e-01 1.10606062e+00
1.05766070e+00 6.31007493e-01 -5.29949844e-01 1.94186196e-01
1.29220772e+00 -1.38918221e+00 -7.91563153e-01 -1.40159819e-02
-1.59239508e-02 -1.02812731e+00 7.28842080e-01 6.44595265e-01
-1.42058742e+00 -6.65036440e-01 -1.19564724e+00 -3.76285583e-01
-4.49459195e-01 7.02404499e-01 1.21161155e-01 7.75075078e-01
-1.21118224e+00 5.02654850e-01 -7.36343682e-01 -4.82506037e-01
2.64169395e-01 3.30420405e-01 -1.75784007e-01 1.49614617e-01
-8.20634782e-01 8.46913576e-01 6.50257692e-02 8.39883238e-02
-9.60985363e-01 -2.95159876e-01 -2.46846542e-01 6.11993074e-01
1.01985395e-01 -3.99106257e-02 1.16484118e+00 -5.52424669e-01
-1.38058555e+00 8.42701852e-01 -3.41276616e-01 -5.40173292e-01
5.76719105e-01 -4.39463675e-01 -7.13270962e-01 3.27466369e-01
4.33886163e-02 5.81740081e-01 1.10677767e+00 -1.07786822e+00
-6.94635808e-01 -4.73410875e-01 -2.87313312e-01 1.86482415e-01
-9.02033150e-01 4.35720801e-01 -1.29560769e+00 -6.35673642e-01
-2.80040912e-02 -8.54977727e-01 4.84166652e-01 -1.46911666e-01
-4.76135671e-01 -1.12287626e-01 7.22791791e-01 -1.22375464e+00
1.49164140e+00 -2.08629680e+00 -1.57885611e-01 5.29934585e-01
3.47940683e-01 3.05972129e-01 -4.17096406e-01 6.35960758e-01
2.59855032e-01 -3.14170085e-02 1.76055178e-01 -5.76323390e-01
2.79147625e-01 -3.41379762e-01 -3.94144863e-01 4.13227499e-01
1.65630773e-01 7.64921784e-01 -2.06025451e-01 -3.15424949e-01
-8.52363333e-02 5.84498584e-01 -1.23793803e-01 4.27587470e-03
4.79684263e-01 -4.12131727e-01 9.08331349e-02 1.02587736e+00
8.41197491e-01 -2.98348695e-01 2.39447474e-01 -1.38246387e-01
4.38957028e-02 1.13690488e-01 -1.14656198e+00 1.28934574e+00
-2.31448978e-01 1.11696589e+00 1.11699410e-01 -4.16313529e-01
1.22131300e+00 -7.68471658e-02 4.35584895e-02 -1.16098487e+00
7.11299554e-02 1.93316415e-01 -9.10241976e-02 -9.80206355e-02
1.25530958e+00 6.58846080e-01 -2.43931543e-02 7.37145185e-01
7.99252242e-02 6.11370981e-01 5.79752624e-01 3.83621186e-01
1.19901991e+00 -3.24837625e-01 -4.18131739e-01 -9.89101827e-02
5.83780885e-01 -4.12187219e-01 5.28079085e-02 1.40123391e+00
2.06762534e-02 9.59085524e-01 5.29813886e-01 -5.25464714e-01
-1.19759560e+00 -1.26535583e+00 -2.71024853e-01 1.25640059e+00
1.50862619e-01 -6.63860261e-01 -7.30781853e-01 -4.82995391e-01
1.18838124e-01 3.08260709e-01 -7.04859853e-01 -5.24510927e-02
-6.35141432e-01 -6.26962900e-01 9.57299829e-01 7.95030415e-01
5.48453987e-01 -9.95812297e-01 -1.96436420e-01 -2.34334052e-01
1.87935471e-03 -9.14588869e-01 -5.58486998e-01 5.56880012e-02
-2.57418513e-01 -8.56844366e-01 -1.10808480e+00 -7.78426170e-01
4.32432532e-01 2.54046530e-01 1.03935349e+00 4.73779626e-02
-2.83851504e-01 1.29014224e-01 -2.08356649e-01 -1.62305042e-01
-1.64505094e-01 5.77800751e-01 6.31552041e-02 -5.58345467e-02
5.37662745e-01 -3.86144556e-02 -5.80849826e-01 3.45482409e-01
-6.57320082e-01 -3.48618120e-01 7.68671453e-01 9.06742334e-01
3.16373020e-01 -8.53446424e-02 1.16608255e-01 -5.04703760e-01
9.61335599e-01 -5.14095016e-02 -4.66383427e-01 3.50646168e-01
-7.13997066e-01 -2.18628988e-01 4.16067004e-01 -3.40784371e-01
-9.67674494e-01 -1.14173561e-01 1.29123420e-01 -5.73887706e-01
1.56613410e-01 2.97234863e-01 8.86792615e-02 -2.45630756e-01
6.66097701e-01 5.31304717e-01 -6.65046051e-02 -7.66562998e-01
3.35304052e-01 1.15175152e+00 6.93920612e-01 -2.30841354e-01
6.75468922e-01 4.86005902e-01 -3.68701398e-01 -6.93735242e-01
-6.38346910e-01 -4.56788570e-01 -4.24620360e-01 -7.74630159e-02
3.27235520e-01 -8.84509444e-01 -8.13644707e-01 7.66872168e-01
-9.98855412e-01 8.00495669e-02 2.58374929e-01 1.22616604e-01
-2.63985336e-01 4.54808980e-01 -1.06393349e+00 -6.76085055e-01
-8.43508720e-01 -1.01537418e+00 1.21548474e+00 4.94864613e-01
-1.85238183e-01 -4.53090340e-01 -7.39161251e-03 4.25918430e-01
5.07856429e-01 -6.28181159e-01 4.80546653e-01 -9.25661922e-01
-4.28367823e-01 -3.80247891e-01 -9.48852956e-01 2.29132161e-01
-9.42878425e-02 -9.30491649e-03 -1.23097587e+00 -5.37526131e-01
-6.24698699e-01 -1.50935665e-01 1.29643714e+00 2.06813604e-01
1.37743425e+00 -8.48204419e-02 -2.52262235e-01 7.29823351e-01
1.27586043e+00 3.24941128e-01 5.88236272e-01 6.25676632e-01
8.15542281e-01 2.40471214e-01 2.08480105e-01 6.24820113e-01
8.96598622e-02 9.27331388e-01 9.23479646e-02 4.35500853e-02
-3.74960065e-01 1.32504806e-01 1.96953773e-01 5.89708447e-01
-9.51949582e-02 -3.78804803e-01 -1.08236444e+00 5.42087853e-01
-1.66463101e+00 -1.03459215e+00 1.53921753e-01 1.99612260e+00
7.32280791e-01 7.79877380e-02 3.37319195e-01 7.01481253e-02
7.45403349e-01 2.56911725e-01 -3.39159310e-01 -6.22061789e-01
-2.97488123e-01 4.13936764e-01 8.23608875e-01 3.04935336e-01
-1.31426156e+00 1.15807927e+00 7.06498575e+00 1.09720504e+00
-1.26215768e+00 -1.79994926e-01 5.99419296e-01 -5.28396726e-01
3.23052317e-01 -7.19523489e-01 -1.42942536e+00 6.48008466e-01
1.00019157e+00 -6.87866136e-02 7.87321687e-01 5.45704305e-01
-9.99757722e-02 1.44189864e-01 -9.32093859e-01 1.25564098e+00
7.78585553e-01 -1.54503930e+00 8.81831124e-02 1.71308801e-01
8.18231761e-01 -6.41794950e-02 5.51693857e-01 3.46659958e-01
-1.20104432e-01 -1.31081069e+00 7.59320855e-01 8.27797592e-01
7.88145363e-01 -1.18693578e+00 8.36903989e-01 -1.95229352e-02
-8.67353082e-01 -1.67305574e-01 -4.30281460e-01 2.31662869e-01
-3.85793477e-01 2.17878342e-01 -7.62924016e-01 3.14066797e-01
1.10187030e+00 6.49657249e-01 -1.05621397e+00 1.08786023e+00
2.11291667e-02 4.30883288e-01 -2.37210855e-01 -1.88528553e-01
1.88731626e-01 1.09089702e-01 2.45259449e-01 1.65555918e+00
2.81181216e-01 -6.53477967e-01 -1.28436238e-01 7.66029835e-01
-4.60094005e-01 2.53864024e-02 -1.95100740e-01 -1.86394617e-01
5.40241063e-01 1.47675502e+00 -7.41613209e-01 -3.85581553e-01
-2.43529320e-01 1.18793893e+00 3.28084022e-01 3.82141858e-01
-7.49067008e-01 -8.77917051e-01 4.07102138e-01 5.17641753e-03
7.99765408e-01 8.56870972e-03 -5.04392564e-01 -8.23739588e-01
2.06593961e-01 -1.05911899e+00 2.37906232e-01 -5.83367884e-01
-1.32899868e+00 4.75238264e-01 -3.99107277e-01 -8.28210890e-01
-3.59121352e-01 -8.45874369e-01 -3.98834556e-01 1.23410296e+00
-1.30926943e+00 -1.20485306e+00 -3.89415950e-01 3.31260711e-01
3.36037725e-01 -7.75540113e-01 9.07067239e-01 5.71644247e-01
-7.44574010e-01 1.35304523e+00 9.31097984e-01 5.21972597e-01
1.12206054e+00 -1.14715052e+00 6.47925198e-01 8.43544960e-01
2.48147845e-01 8.89524579e-01 2.46122643e-01 -4.16733533e-01
-1.53837538e+00 -9.31955516e-01 9.97581482e-01 -6.08826518e-01
5.97305536e-01 -5.24459779e-01 -8.49252582e-01 4.37423170e-01
4.09210056e-01 -3.95793676e-01 5.17558157e-01 4.35309708e-01
-5.66815674e-01 -3.58543962e-01 -7.84321487e-01 5.76218486e-01
6.87292695e-01 -7.25832999e-01 -1.28681377e-01 2.91506410e-01
-1.15218967e-01 -6.01991832e-01 -7.80962050e-01 -3.35025080e-02
1.17643011e+00 -8.80493999e-01 1.05636060e+00 -1.61144271e-01
7.65884876e-01 -3.15658376e-02 -5.36089391e-02 -7.50546157e-01
-7.43991613e-01 -2.06414267e-01 -5.86272478e-01 1.25582314e+00
3.57109189e-01 -7.60456175e-02 1.06133938e+00 3.18217069e-01
1.55106977e-01 -6.37462735e-01 -5.89411736e-01 -8.89055967e-01
-9.25541576e-03 -3.25083770e-02 5.20337701e-01 6.05678976e-01
-3.59661192e-01 2.70948678e-01 -6.69181585e-01 -6.37565032e-02
4.16055650e-01 2.08216846e-01 7.95187771e-01 -9.45442677e-01
-1.18486628e-01 -8.83545220e-01 -2.59374410e-01 -8.13086152e-01
1.15914673e-01 -7.90835142e-01 -7.43622258e-02 -1.21705890e+00
7.37299979e-01 1.70687512e-01 -7.22568333e-01 4.97428745e-01
2.42030825e-02 8.41855168e-01 4.65361089e-01 5.56162298e-01
-8.39912355e-01 2.00306058e-01 8.61958146e-01 -4.89578813e-01
-1.35386854e-01 -1.85937956e-01 -8.79324675e-01 3.87514800e-01
8.68743420e-01 -1.45487458e-01 8.83398652e-02 -5.77136755e-01
-7.09834620e-02 -3.00504804e-01 1.66347265e-01 -9.96704578e-01
3.33254397e-01 4.83887404e-01 1.18530619e+00 -1.10382199e+00
5.98812163e-01 -4.58791614e-01 -2.94294268e-01 2.20435143e-01
-5.45652688e-01 1.07609056e-01 3.22940081e-01 1.84272110e-01
-1.02848969e-01 -3.20945412e-01 5.45790255e-01 2.87365317e-01
-7.75206685e-01 2.00784188e-02 -3.44825834e-01 -4.31207687e-01
3.88600439e-01 -8.03159922e-02 -6.81151211e-01 -4.75745261e-01
-4.24726605e-01 6.68427795e-02 3.96475255e-01 6.43473744e-01
6.15261674e-01 -1.28482103e+00 -8.79698396e-01 3.48498374e-01
2.16650575e-01 -8.78828645e-01 -3.79015668e-03 4.03369099e-01
-8.19562197e-01 6.99492097e-01 -4.58955646e-01 -5.30952752e-01
-1.55491900e+00 1.42583996e-01 2.32209325e-01 -4.21673059e-01
-3.63300890e-01 8.37215602e-01 4.56035137e-04 -2.01392964e-01
6.02625906e-01 4.06895012e-01 -3.28118861e-01 1.65338367e-01
9.77363110e-01 7.50569582e-01 4.59514588e-01 -6.39897883e-01
-5.03249168e-01 4.44252014e-01 -8.93861353e-01 -3.00599754e-01
9.94991779e-01 8.68975893e-02 -2.08146721e-02 6.40322268e-02
1.24158204e+00 1.46051317e-01 -9.23876524e-01 -4.03592855e-01
-9.64617506e-02 -4.71258402e-01 2.40936324e-01 -1.25473738e+00
-1.09188080e+00 8.27361405e-01 9.87209558e-01 9.68999043e-02
1.06650805e+00 -1.95476651e-01 7.57836342e-01 7.92098105e-01
-2.13352874e-01 -1.40816128e+00 2.02393644e-02 8.16625237e-01
9.29162562e-01 -1.03911376e+00 2.48746470e-01 1.85035631e-01
-2.60866314e-01 1.15143728e+00 5.07234395e-01 -2.48439044e-01
2.52516448e-01 4.20338571e-01 2.04549715e-01 -1.16557039e-01
-4.77843195e-01 6.31686002e-02 6.14345253e-01 2.72891998e-01
6.12492561e-01 9.64700207e-02 -4.77261275e-01 7.09356129e-01
-3.49528134e-01 -3.77676755e-01 1.56961620e-01 7.46141315e-01
-1.95927039e-01 -9.65841115e-01 -6.43294632e-01 7.10414231e-01
-7.69347548e-01 -2.40301713e-01 -9.30472195e-01 7.73931921e-01
-3.39821339e-01 6.46737278e-01 4.75594312e-01 -5.09965181e-01
-7.00271130e-02 -8.87579620e-02 4.96090829e-01 -2.94744849e-01
-7.66804695e-01 4.59352940e-01 -1.86755788e-02 -5.21758318e-01
1.50249213e-01 -5.63108027e-01 -7.82191813e-01 -7.37266123e-01
-2.51993984e-01 -5.13331546e-03 8.69558215e-01 5.33760369e-01
4.64021981e-01 5.15129685e-01 5.97099602e-01 -7.93594539e-01
-5.14813185e-01 -1.13157964e+00 -9.92673695e-01 1.44796163e-01
1.72316492e-01 -3.42877805e-01 -9.24403220e-02 -1.95879359e-02]
|
[11.870752334594727, 2.4622952938079834]
|
0083077c-f3b9-413b-a921-daaf7f964b78
|
single-image-reflection-removal-through
|
1911.06634
| null |
https://arxiv.org/abs/1911.06634v2
|
https://arxiv.org/pdf/1911.06634v2.pdf
|
Single Image Reflection Removal through Cascaded Refinement
|
We address the problem of removing undesirable reflections from a single image captured through a glass surface, which is an ill-posed, challenging but practically important problem for photo enhancement. Inspired by iterative structure reduction for hidden community detection in social networks, we propose an Iterative Boost Convolutional LSTM Network (IBCLN) that enables cascaded prediction for reflection removal. IBCLN is a cascaded network that iteratively refines the estimates of transmission and reflection layers in a manner that they can boost the prediction quality to each other, and information across steps of the cascade is transferred using an LSTM. The intuition is that the transmission is the strong, dominant structure while the reflection is the weak, hidden structure. They are complementary to each other in a single image and thus a better estimate and reduction on one side from the original image leads to a more accurate estimate on the other side. To facilitate training over multiple cascade steps, we employ LSTM to address the vanishing gradient problem, and propose residual reconstruction loss as further training guidance. Besides, we create a dataset of real-world images with reflection and ground-truth transmission layers to mitigate the problem of insufficient data. Comprehensive experiments demonstrate that the proposed method can effectively remove reflections in real and synthetic images compared with state-of-the-art reflection removal methods.
|
['Yixiao Yang', 'Stephen Lin', 'Kun He', 'John E. Hopcroft', 'Chao Li']
|
2019-11-15
|
single-image-reflection-removal-through-1
|
http://openaccess.thecvf.com/content_CVPR_2020/html/Li_Single_Image_Reflection_Removal_Through_Cascaded_Refinement_CVPR_2020_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2020/papers/Li_Single_Image_Reflection_Removal_Through_Cascaded_Refinement_CVPR_2020_paper.pdf
|
cvpr-2020-6
|
['reflection-removal']
|
['computer-vision']
|
[ 8.40468884e-01 1.67216152e-01 4.86640781e-01 -2.01571826e-02
-4.74134892e-01 1.13142356e-01 2.55408347e-01 -4.13174987e-01
-5.25223389e-02 4.72486556e-01 5.77992737e-01 -8.82288143e-02
2.75428081e-03 -9.96677935e-01 -8.62058938e-01 -1.12314832e+00
2.76666045e-01 -2.10253328e-01 1.03658728e-01 -3.19519341e-01
-1.31146181e-02 2.18159288e-01 -1.16938841e+00 7.71259964e-01
8.44798684e-01 1.02695334e+00 2.97477990e-01 5.29533744e-01
1.40069798e-01 1.23779249e+00 -3.74651670e-01 -3.05903852e-01
3.09781700e-01 -5.09945691e-01 -6.48451149e-01 3.94839406e-01
5.07782459e-01 -4.59415615e-01 -5.48330724e-01 9.95200336e-01
3.67995918e-01 3.55609804e-02 4.67801571e-01 -5.77094316e-01
-8.43555689e-01 5.74365854e-01 -9.46480215e-01 -3.22131157e-01
7.37618133e-02 -4.77628633e-02 7.89363921e-01 -9.57759321e-01
3.88621420e-01 1.28942072e+00 9.71397698e-01 3.62686217e-01
-1.05436683e+00 -6.86847627e-01 3.31777245e-01 9.31107104e-02
-1.24330914e+00 -5.10461330e-01 9.71856296e-01 -1.51847020e-01
6.27885342e-01 2.54767537e-01 7.01633155e-01 8.72027218e-01
6.53370470e-02 6.96661949e-01 8.80228460e-01 -3.72083455e-01
-3.97667080e-01 9.31895748e-02 -2.66623318e-01 1.00925505e+00
1.84118167e-01 2.55209133e-02 -4.58708376e-01 6.55190051e-02
8.60559702e-01 4.00812149e-01 -5.79586446e-01 1.66319698e-01
-9.25406277e-01 4.61031556e-01 1.06199217e+00 1.99345261e-01
-5.25132596e-01 8.00991431e-02 -1.41290680e-01 3.85745347e-01
7.63344467e-01 7.32338130e-02 -3.12098209e-02 8.70276809e-01
-7.28653014e-01 -1.62798673e-01 4.53679115e-01 5.19517541e-01
1.04805171e+00 1.58238053e-01 -4.03060734e-01 9.83812094e-01
6.01783276e-01 5.65273345e-01 -1.89873487e-01 -1.04948056e+00
4.09974128e-01 7.79533982e-01 2.41565838e-01 -1.14127660e+00
-8.66471007e-02 -9.00592625e-01 -1.56474829e+00 2.54392892e-01
4.97249439e-02 -1.58942610e-01 -9.96872127e-01 1.47184134e+00
4.32237983e-01 4.43323821e-01 1.75433457e-02 1.10127175e+00
8.83082211e-01 7.75681317e-01 -4.02073771e-01 -4.23089445e-01
9.65297937e-01 -1.30001676e+00 -5.85628271e-01 -3.05594176e-01
2.67100841e-01 -9.81508195e-01 5.47775328e-01 5.58932126e-01
-1.12722707e+00 -5.83797336e-01 -9.96520042e-01 -2.22520381e-01
4.54496026e-01 3.40034634e-01 3.17447692e-01 1.75138861e-01
-1.10625446e+00 7.03291833e-01 -6.08103454e-01 8.39465633e-02
5.46245217e-01 1.25938669e-01 -1.94237335e-03 -5.78666866e-01
-9.58672881e-01 4.64122087e-01 -3.88152838e-01 9.30092990e-01
-9.60943401e-01 -5.48255563e-01 -5.83394170e-01 1.83644947e-02
2.64383167e-01 -7.74090648e-01 7.50706315e-01 -1.17554152e+00
-1.35595238e+00 5.76113939e-01 -2.93545365e-01 -2.38907889e-01
5.15899539e-01 -3.34170043e-01 -2.69091964e-01 1.73295647e-01
-1.77991495e-01 1.75207064e-01 1.40077531e+00 -1.67051375e+00
-3.72896373e-01 -2.10997060e-01 -3.97522859e-02 3.41136932e-01
-4.34920400e-01 -2.40491703e-01 -3.34804952e-01 -4.45884109e-01
4.30940866e-01 -7.06663489e-01 -3.39733481e-01 1.39371470e-01
-7.21565604e-01 2.63871104e-01 8.60314846e-01 -9.09538209e-01
9.77197647e-01 -2.02067280e+00 1.91508412e-01 1.90846652e-01
7.62440741e-01 5.01512170e-01 -4.92474556e-01 4.11700070e-01
-1.14214867e-02 4.40520607e-03 -3.59695822e-01 -6.65345252e-01
-4.97049630e-01 -6.86559975e-02 -2.77616441e-01 5.77686787e-01
1.64262936e-01 6.56987429e-01 -7.66947925e-01 -3.02149057e-01
1.69333503e-01 1.14205658e+00 -4.02206928e-01 2.53927737e-01
6.25473866e-03 7.71582603e-01 -3.15439612e-01 3.30940783e-01
1.03252959e+00 -5.12105167e-01 1.56381458e-01 -6.50656521e-01
-3.22857887e-01 1.53810278e-01 -1.01472652e+00 1.12052643e+00
-5.72915971e-01 7.91703939e-01 3.04860711e-01 -9.59185958e-01
1.07189500e+00 2.14997038e-01 4.29053366e-01 -8.42787564e-01
5.09155802e-02 -5.53895272e-02 -4.98027392e-02 -6.37691319e-01
2.35457376e-01 -2.31391519e-01 6.84543669e-01 5.99955738e-01
-3.95536125e-01 2.00070351e-01 -3.60797107e-01 1.10965878e-01
1.09150410e+00 -1.67141035e-02 -3.00143719e-01 3.30961138e-01
7.07208991e-01 -5.55417717e-01 7.06240475e-01 7.16082096e-01
3.15555096e-01 8.61712396e-01 -6.49097636e-02 -4.98977482e-01
-1.05094576e+00 -8.39319468e-01 3.80431354e-01 8.56815755e-01
3.18330079e-01 -4.72782105e-02 -7.78739214e-01 -2.86900848e-01
-4.49278563e-01 2.06737831e-01 -4.84274089e-01 -3.12333256e-01
-8.08201313e-01 -7.60011315e-01 1.05497636e-01 8.98436457e-02
9.79173958e-01 -1.09759331e+00 -4.11540568e-02 1.92571446e-01
-7.75891900e-01 -1.22889662e+00 -3.08983237e-01 -4.02683407e-01
-7.60275006e-01 -1.15344715e+00 -8.39629889e-01 -9.54677522e-01
9.11559939e-01 9.85314727e-01 1.11056066e+00 8.98638606e-01
-2.97392219e-01 9.85197201e-02 -2.29802594e-01 -2.71940589e-01
-3.68507951e-01 -2.97752529e-01 -4.61614043e-01 5.66939771e-01
-1.80177346e-01 -6.49884105e-01 -1.10961115e+00 3.68089199e-01
-7.19331682e-01 4.55791712e-01 7.94568837e-01 7.44129479e-01
5.26671052e-01 4.05939549e-01 1.08090125e-01 -7.97432482e-01
3.77544880e-01 -1.11752711e-01 -3.72508913e-01 3.01175028e-01
-3.59506965e-01 -1.54284224e-01 6.82699859e-01 -1.81751683e-01
-1.54610884e+00 -6.88099489e-02 -4.68656644e-02 -4.70901430e-01
2.40086257e-01 3.68779331e-01 6.90672770e-02 -4.01202798e-01
4.41562653e-01 4.07104373e-01 -4.58860770e-02 -5.34013689e-01
-8.90689436e-03 4.14357096e-01 2.39430264e-01 -3.77102755e-02
1.00165343e+00 9.73694980e-01 8.63951147e-02 -1.17810369e+00
-1.40199184e+00 -2.22609714e-01 -4.81155217e-01 -5.09886205e-01
6.05800092e-01 -1.06621075e+00 -9.66848552e-01 9.55983996e-01
-1.29252839e+00 -5.82541406e-01 -6.36981800e-02 2.94367999e-01
6.78533912e-02 4.95337427e-01 -8.21611226e-01 -1.08826768e+00
-8.18389595e-01 -7.69395530e-01 1.03027880e+00 1.52493909e-01
4.73842114e-01 -9.40738201e-01 -9.95532498e-02 6.79988503e-01
3.13431561e-01 5.17096967e-02 7.47996688e-01 4.26245809e-01
-9.93839204e-01 -2.23904965e-03 -6.84303284e-01 8.76586199e-01
1.56541437e-01 5.32259531e-02 -1.08768022e+00 -5.03237665e-01
4.51611161e-01 -3.15336995e-02 1.55114853e+00 5.48143923e-01
8.81304383e-01 -5.76690853e-01 -2.36186594e-01 7.58297384e-01
1.36979938e+00 -3.58293325e-01 1.12802815e+00 2.03648992e-02
1.25548613e+00 7.57883132e-01 3.27728182e-01 2.24391505e-01
4.71564740e-01 8.47229213e-02 7.61021495e-01 -8.78625810e-01
-6.35276198e-01 -1.30662978e-01 4.30658966e-01 1.03628755e+00
-4.09094095e-01 -4.46635187e-01 -3.83069336e-01 4.42361951e-01
-1.79189408e+00 -9.87794816e-01 -6.17475390e-01 2.27679300e+00
5.99866867e-01 -1.47239259e-02 -2.89837480e-01 1.87035859e-01
8.53115916e-01 4.02231932e-01 -4.65431571e-01 2.04108700e-01
-1.89103931e-01 -1.64732467e-02 3.67049456e-01 8.32084119e-01
-7.61277378e-01 7.63462365e-01 5.86489964e+00 5.27968824e-01
-1.25659227e+00 2.91351806e-02 7.74150968e-01 8.15056730e-03
-6.19928837e-01 -5.94769642e-02 -5.56142747e-01 1.51828289e-01
2.15091914e-01 4.62509215e-01 7.02700615e-01 -5.85343093e-02
5.89460373e-01 1.15613915e-01 -6.73778415e-01 7.44619071e-01
6.23845793e-02 -1.28133750e+00 2.06469804e-01 -3.96915153e-03
9.38454092e-01 2.86444306e-01 1.17787361e-01 -1.58791378e-01
3.60954493e-01 -9.85718966e-01 3.66027594e-01 9.32680666e-01
5.05458891e-01 -5.63820779e-01 5.76989174e-01 3.37167650e-01
-1.28716338e+00 -1.68903366e-01 -5.81165850e-01 -1.31004885e-01
1.97863728e-01 1.07577598e+00 -4.96508121e-01 5.82540810e-01
8.07935357e-01 1.17282534e+00 -1.15350671e-01 9.49375808e-01
-6.11827910e-01 6.57656014e-01 -9.87712964e-02 3.52334440e-01
1.35926113e-01 -6.54157817e-01 7.65402138e-01 9.20612097e-01
3.15679342e-01 2.20105097e-01 1.13813013e-01 9.72299874e-01
-3.03169042e-01 -4.06957328e-01 -3.35158497e-01 2.75598168e-01
8.16746727e-02 1.51726031e+00 -3.41220438e-01 -2.35372052e-01
-4.27887470e-01 9.67875123e-01 3.19956630e-01 8.31887543e-01
-5.90675414e-01 -8.42642039e-02 4.91551250e-01 4.98437762e-01
3.62922341e-01 2.95141730e-02 -1.10817418e-01 -1.08412075e+00
1.53752550e-01 -6.53705537e-01 8.49520788e-02 -1.00898266e+00
-1.28841794e+00 6.75965071e-01 -7.83613086e-01 -1.28991866e+00
4.58277941e-01 -3.62189531e-01 -8.71632636e-01 8.71418715e-01
-2.14380455e+00 -1.44098592e+00 -7.99424708e-01 6.13523901e-01
2.93201894e-01 2.51043439e-01 4.12916243e-01 5.50605953e-01
-6.15057349e-01 1.93400353e-01 8.35630447e-02 1.58673465e-01
6.37905657e-01 -6.67947948e-01 2.56743878e-01 1.01612997e+00
-1.75104722e-01 4.73748356e-01 5.42798042e-01 -6.82385564e-01
-1.19119763e+00 -1.45426643e+00 5.65975368e-01 1.69912905e-01
2.78230578e-01 -1.86822832e-01 -1.05345738e+00 5.90061963e-01
2.47543037e-01 -1.87677845e-01 3.49851042e-01 -1.57665133e-01
-3.35533828e-01 -3.63089472e-01 -9.31161463e-01 5.27055979e-01
1.23368907e+00 -4.64442700e-01 7.72394836e-02 4.59554046e-01
8.28546643e-01 -1.50969669e-01 -6.37036443e-01 3.83443207e-01
5.24056494e-01 -1.31932962e+00 1.17445791e+00 8.55042562e-02
7.94088960e-01 -2.76344299e-01 3.03991940e-02 -1.32415378e+00
-6.12630069e-01 -6.47515893e-01 2.79473178e-02 1.01915133e+00
3.12043875e-01 -6.34021580e-01 9.01767194e-01 1.28313929e-01
-4.15582597e-01 -5.66163957e-01 -3.58188033e-01 -2.40771547e-01
-3.86818498e-01 -1.17456771e-01 3.35380405e-01 9.54709351e-01
-7.21626699e-01 5.74628830e-01 -9.10602033e-01 4.45257097e-01
1.24574816e+00 3.27175498e-01 6.96927845e-01 -1.24932587e+00
-2.13075712e-01 -3.22096795e-02 1.80079371e-01 -1.31995034e+00
-9.45470408e-02 -5.59527040e-01 2.54765004e-01 -2.06634855e+00
5.30348718e-01 -2.74682283e-01 -3.29115063e-01 4.49659497e-01
-2.49637991e-01 6.61724091e-01 7.27519393e-02 2.64657676e-01
-6.24685407e-01 8.85381460e-01 1.88568890e+00 -4.80227709e-01
-2.57934600e-01 1.79430008e-01 -9.45463002e-01 9.52707350e-01
3.61292571e-01 -4.72361356e-01 -2.86595911e-01 -9.67048585e-01
5.76326907e-01 4.73132357e-02 7.06778109e-01 -8.76495838e-01
3.08972090e-01 1.89215615e-01 4.81995612e-01 -5.96377194e-01
5.64615309e-01 -7.83721030e-01 8.96778889e-03 4.92953539e-01
-2.82059222e-01 -7.55495667e-01 -2.13488668e-01 6.79747164e-01
-5.19944169e-02 1.30163848e-01 8.97247016e-01 -3.53815377e-01
-1.23599410e-01 5.16093135e-01 -2.30424732e-01 -2.70049334e-01
2.78275847e-01 -3.01289320e-01 -4.79820549e-01 -7.87487388e-01
-6.46588683e-01 1.60266161e-01 2.73311496e-01 7.72724375e-02
1.03475511e+00 -1.01976061e+00 -1.23785710e+00 1.44043609e-01
-2.57164001e-01 3.21517676e-01 6.73231184e-01 1.05964363e+00
-2.88728029e-01 -3.92665446e-01 8.20870548e-02 -5.60762584e-01
-1.57441473e+00 1.41928717e-01 6.54939950e-01 -3.14869195e-01
-8.52985740e-01 1.00963366e+00 6.78437829e-01 -3.57711136e-01
1.71816237e-02 -9.28213075e-02 -3.39450777e-01 -2.53480375e-01
7.36058831e-01 5.15968263e-01 1.11760143e-02 -5.84850371e-01
3.06910668e-02 7.69365907e-01 -1.57999530e-01 2.52561957e-01
1.85962939e+00 -6.25787020e-01 -6.07977688e-01 1.98374484e-02
1.20362520e+00 -1.66256428e-02 -1.49459207e+00 -7.99540043e-01
-7.39184499e-01 -4.82947618e-01 4.80243891e-01 -6.23798728e-01
-1.58929133e+00 9.60475028e-01 4.97921765e-01 2.08729938e-01
1.29787338e+00 -3.71541083e-01 1.11857355e+00 4.09244150e-01
-1.19394489e-01 -9.78668690e-01 5.75493693e-01 4.50637639e-01
1.05436385e+00 -1.12835038e+00 3.45904022e-01 -7.59700894e-01
-3.84355217e-01 9.85410273e-01 4.13710535e-01 -1.58035368e-01
5.12034357e-01 1.13699250e-01 6.71954884e-04 -4.04735863e-01
-5.84979653e-01 -1.01526938e-01 3.00738603e-01 5.37704706e-01
3.46465498e-01 -2.98365146e-01 3.72202307e-01 -5.33064567e-02
2.03156307e-01 -6.12822995e-02 5.28515697e-01 4.39538628e-01
-5.10388255e-01 -6.85614586e-01 -4.60884124e-01 4.10096258e-01
-2.98690796e-01 -3.79137576e-01 -5.26972592e-01 2.24405125e-01
1.03988871e-02 1.20935464e+00 -5.04621230e-02 -5.11792064e-01
1.75534248e-01 -7.60370791e-01 5.04137754e-01 -4.63837892e-01
-5.27193427e-01 4.34842467e-01 5.68559170e-02 -3.78600985e-01
-6.61173642e-01 -2.05943614e-01 -1.02545679e+00 -4.79012698e-01
-5.67728579e-01 -8.73311982e-02 3.22667837e-01 1.02060652e+00
2.97243059e-01 9.20577645e-01 1.18993676e+00 -1.06556404e+00
-1.07286245e-01 -9.48827624e-01 -5.67946494e-01 3.14398944e-01
8.53983462e-01 -2.00170532e-01 -5.01282573e-01 3.69823747e-03]
|
[10.669526100158691, -2.8090732097625732]
|
0c36c834-c69b-4cf7-803e-745ac33ca90b
|
can-shuffling-video-benefit-temporal-bias
|
2207.14698
| null |
https://arxiv.org/abs/2207.14698v2
|
https://arxiv.org/pdf/2207.14698v2.pdf
|
Can Shuffling Video Benefit Temporal Bias Problem: A Novel Training Framework for Temporal Grounding
|
Temporal grounding aims to locate a target video moment that semantically corresponds to the given sentence query in an untrimmed video. However, recent works find that existing methods suffer a severe temporal bias problem. These methods do not reason the target moment locations based on the visual-textual semantic alignment but over-rely on the temporal biases of queries in training sets. To this end, this paper proposes a novel training framework for grounding models to use shuffled videos to address temporal bias problem without losing grounding accuracy. Our framework introduces two auxiliary tasks, cross-modal matching and temporal order discrimination, to promote the grounding model training. The cross-modal matching task leverages the content consistency between shuffled and original videos to force the grounding model to mine visual contents to semantically match queries. The temporal order discrimination task leverages the difference in temporal order to strengthen the understanding of long-term temporal contexts. Extensive experiments on Charades-STA and ActivityNet Captions demonstrate the effectiveness of our method for mitigating the reliance on temporal biases and strengthening the model's generalization ability against the different temporal distributions. Code is available at https://github.com/haojc/ShufflingVideosForTSG.
|
['Jianxin Liao', 'Qi Qi', 'Jingyu Wang', 'Pengfei Ren', 'Haifeng Sun', 'Jiachang Hao']
|
2022-07-29
| null | null | null | null |
['language-based-temporal-localization']
|
['computer-vision']
|
[ 2.90082805e-02 -2.20145598e-01 -6.72565639e-01 -4.50388581e-01
-6.92939162e-01 -7.45961845e-01 6.19209051e-01 -4.48610261e-02
-2.14392081e-01 2.52078384e-01 4.92110342e-01 -1.52772903e-01
6.08474808e-03 -4.43956375e-01 -8.17366183e-01 -4.18859184e-01
-1.52256802e-01 -3.08558792e-01 5.34485221e-01 -1.85365617e-01
3.11036676e-01 4.79960218e-02 -1.51278281e+00 8.96016479e-01
6.44490540e-01 1.26209855e+00 2.90780604e-01 4.52519268e-01
8.28446820e-02 1.02230501e+00 -4.34427202e-01 -2.69581497e-01
3.45131934e-01 -5.95983565e-01 -7.44026721e-01 8.54578763e-02
8.42352569e-01 -4.23633844e-01 -8.63887787e-01 1.00488114e+00
2.38121808e-01 3.11609894e-01 1.68901801e-01 -1.79485297e+00
-7.71394134e-01 4.64217901e-01 -5.34351945e-01 7.13082135e-01
5.47345519e-01 3.56598169e-01 1.13223088e+00 -8.05091739e-01
7.87668407e-01 1.00262237e+00 6.36044919e-01 5.01040637e-01
-9.57961440e-01 -9.08120811e-01 5.88871717e-01 6.44696176e-01
-1.41427565e+00 -5.30034244e-01 9.88749444e-01 -6.54097974e-01
6.59089267e-01 2.82465726e-01 7.95244277e-01 1.63643909e+00
-2.50762478e-02 8.71171176e-01 1.10902166e+00 3.42839584e-02
3.74365635e-02 -1.30616099e-01 -9.05580632e-03 6.81129873e-01
-1.79352269e-01 2.92406708e-01 -1.26483059e+00 8.57288539e-02
6.41200900e-01 1.85345739e-01 -4.88552332e-01 -2.95507133e-01
-1.59325635e+00 6.68175817e-01 5.34155369e-01 5.02427459e-01
-2.97098666e-01 1.34017691e-01 4.88118500e-01 1.16652980e-01
3.85876834e-01 3.50696921e-01 -3.32393229e-01 -2.01110825e-01
-1.20550668e+00 1.96056589e-01 2.66318589e-01 1.13483727e+00
7.22358584e-01 -1.38846770e-01 -6.06701016e-01 4.65247303e-01
1.66571125e-01 4.10522699e-01 5.47073603e-01 -9.99117553e-01
8.40404689e-01 4.36799645e-01 7.54866898e-02 -1.55517948e+00
-3.71706933e-02 -2.64954120e-01 -4.36661392e-01 -3.15842181e-01
5.32941043e-01 2.39849642e-01 -8.99330556e-01 2.01692414e+00
3.27271461e-01 6.32244766e-01 -2.91990906e-01 1.15164673e+00
7.52251208e-01 5.74468195e-01 2.94516027e-01 3.70219746e-03
1.44213009e+00 -8.77156973e-01 -7.85976171e-01 -3.68042022e-01
6.46190763e-01 -6.49551630e-01 1.43997419e+00 -4.62229922e-02
-7.94159532e-01 -5.91393352e-01 -1.17458940e+00 -7.02751502e-02
-3.50334466e-01 3.38709876e-02 4.34987903e-01 1.49759933e-01
-7.94226170e-01 5.85707784e-01 -7.10570276e-01 -5.43731570e-01
3.45585793e-01 -1.34707659e-01 -3.71450871e-01 3.18760462e-02
-1.47723639e+00 5.81887364e-01 5.47306955e-01 2.06993356e-01
-9.57879245e-01 -7.84557223e-01 -9.20629680e-01 -3.41740340e-01
4.51950192e-01 -4.74438697e-01 1.30350327e+00 -1.24171472e+00
-8.61198902e-01 1.03264940e+00 -3.05690527e-01 -6.05044425e-01
5.22285700e-01 -2.18231514e-01 -5.94361424e-01 5.37847161e-01
5.48219085e-01 8.54780674e-01 8.35772753e-01 -1.11202490e+00
-8.00176561e-01 -1.58453032e-01 2.82104552e-01 1.79522693e-01
-4.14439470e-01 -2.31832877e-01 -7.05293953e-01 -1.05446291e+00
3.06236833e-01 -9.59126174e-01 2.64601976e-01 -3.15956585e-02
-3.89422685e-01 9.01017245e-03 9.86816049e-01 -8.62323999e-01
1.58444417e+00 -2.26467991e+00 -4.28420380e-02 1.57275833e-02
9.24274325e-02 -3.41392159e-01 -2.02552691e-01 4.42792237e-01
-1.13696836e-01 7.48615116e-02 1.01960458e-01 -1.03957608e-01
-1.05280273e-01 2.10467592e-01 -7.60386646e-01 5.40584505e-01
3.96754295e-02 9.32780743e-01 -1.20083916e+00 -8.23166251e-01
5.56856804e-02 2.37912670e-01 -5.16567469e-01 3.25502694e-01
-2.66286105e-01 5.73075056e-01 -2.75045097e-01 6.70607805e-01
2.48713493e-01 -4.06101525e-01 5.99161722e-02 -7.36819029e-01
2.35531060e-03 3.99205357e-01 -9.17376578e-01 2.13950586e+00
-1.45295620e-01 8.34394753e-01 -3.13078552e-01 -7.08747387e-01
5.57697475e-01 3.23084921e-01 8.06010187e-01 -1.14234138e+00
-6.11153170e-02 6.40111789e-02 -2.77579427e-01 -8.24427724e-01
6.65854812e-01 1.74470823e-02 -1.49712443e-01 3.05326164e-01
3.17533426e-02 2.97588110e-01 1.78565592e-01 5.32039404e-01
7.95410752e-01 5.37361681e-01 -1.13439798e-01 -3.77147645e-02
8.82876068e-02 2.42526382e-01 7.57838786e-01 6.31700993e-01
-5.04125774e-01 7.21068978e-01 3.21175724e-01 -3.97526383e-01
-9.05925512e-01 -1.01592910e+00 2.43245751e-01 1.39483535e+00
6.98302865e-01 -7.54629076e-01 -4.82519388e-01 -8.60768676e-01
-1.54775307e-01 6.16356492e-01 -9.73607600e-01 -2.67193586e-01
-6.56825483e-01 -2.60745853e-01 5.84538996e-01 6.45472169e-01
6.70708835e-01 -8.63884985e-01 -7.50748158e-01 -2.08372578e-01
-1.06498361e+00 -1.33988023e+00 -1.05064189e+00 -2.67310053e-01
-7.46287704e-01 -1.16406167e+00 -5.53228319e-01 -6.89936340e-01
4.36497897e-01 7.08673835e-01 1.08497465e+00 3.94410901e-02
4.76386845e-02 5.87133348e-01 -5.58632612e-01 -1.90945596e-01
4.21556979e-02 -8.33985284e-02 -3.05395126e-02 2.46236548e-01
5.20712912e-01 -4.23274189e-01 -9.91386533e-01 7.02064395e-01
-9.17610466e-01 3.90284240e-01 1.69001967e-01 6.06799304e-01
6.16635203e-01 -1.72387645e-01 1.94991276e-01 -3.84994298e-01
1.68639705e-01 -8.09212327e-01 -3.12969744e-01 1.69068277e-01
-4.70445216e-01 -4.62224893e-02 1.47894472e-01 -7.24279225e-01
-8.20789456e-01 -1.53739944e-01 3.47522050e-01 -9.72014964e-01
8.56732428e-02 4.45922762e-01 5.78376018e-02 3.13447356e-01
5.57362974e-01 2.91475981e-01 -2.20451385e-01 -2.29208946e-01
2.82863736e-01 2.47096092e-01 7.57005930e-01 -6.88714743e-01
7.98257768e-01 9.28599596e-01 -2.72394955e-01 -3.48875731e-01
-1.29560852e+00 -7.94759572e-01 -5.09937763e-01 -6.54187739e-01
1.17447150e+00 -1.09747326e+00 -4.63222146e-01 1.52928606e-01
-1.10304999e+00 -5.13418674e-01 -8.43767673e-02 4.45426762e-01
-6.66122079e-01 4.06381667e-01 -4.59761620e-01 -3.95262569e-01
-6.09089155e-03 -8.04151595e-01 1.10827601e+00 9.63543542e-03
-4.15144920e-01 -8.03348601e-01 2.31250357e-02 5.01344621e-01
-3.71641181e-02 3.90514702e-01 5.66615999e-01 -5.09532154e-01
-7.55102098e-01 4.65499200e-02 -1.74980089e-01 -6.11109361e-02
4.43099104e-02 -1.10312141e-01 -1.06519485e+00 -3.52543294e-01
3.10169384e-02 -1.91856131e-01 8.54535520e-01 2.06343904e-01
1.14315403e+00 -5.30073106e-01 -1.80612683e-01 7.22384393e-01
1.21606791e+00 1.21554613e-01 5.79988778e-01 7.12396204e-01
8.88596714e-01 8.34313154e-01 1.16882551e+00 3.68084580e-01
4.04788226e-01 8.93907130e-01 5.84082723e-01 -2.23397445e-02
-2.19953552e-01 -8.25255632e-01 4.58459288e-01 6.24570072e-01
1.05558865e-01 -1.29149139e-01 -1.02804780e+00 9.45789695e-01
-2.23151565e+00 -1.59622490e+00 2.08318561e-01 2.12725425e+00
8.24648619e-01 5.58135882e-02 2.76867956e-01 -3.79858096e-03
8.41055095e-01 5.36224365e-01 -3.55871439e-01 2.77297318e-01
-1.53515533e-01 -5.86359203e-01 4.50175524e-01 2.87640840e-01
-1.13689470e+00 7.76197016e-01 5.24777412e+00 8.01975429e-01
-1.24052751e+00 3.29276830e-01 5.23547411e-01 -2.82577932e-01
-3.00575584e-01 6.40655607e-02 -6.54136300e-01 7.74237871e-01
7.53956318e-01 -8.52935389e-02 2.63673574e-01 6.23314261e-01
6.05853796e-01 -3.65358926e-02 -1.33344221e+00 1.20912302e+00
2.20897958e-01 -1.42342222e+00 9.59030837e-02 -1.25928521e-01
6.76741362e-01 -9.20701623e-02 3.15562755e-01 2.67512470e-01
-8.56713429e-02 -8.91215265e-01 1.42310798e+00 4.53749776e-01
6.79813683e-01 -2.41570011e-01 3.77578348e-01 6.43145293e-02
-1.48150873e+00 -1.23826399e-01 1.30668283e-01 1.51777521e-01
1.94110543e-01 1.27332583e-01 -7.43012846e-01 4.49872583e-01
1.11055028e+00 1.05611765e+00 -7.52609611e-01 8.56223047e-01
-1.51421532e-01 5.77793181e-01 -2.22651027e-02 4.16029721e-01
4.67681378e-01 2.12175306e-02 6.35356545e-01 1.32236731e+00
3.66429389e-01 -7.46258199e-02 2.63226122e-01 6.02249980e-01
1.87836930e-01 2.01632138e-02 -7.59393036e-01 -1.15622886e-01
5.99476933e-01 8.70796740e-01 -7.83252001e-01 -3.22439820e-01
-4.92964089e-01 1.03629351e+00 6.11733459e-02 6.78884566e-01
-1.42960155e+00 -9.36535224e-02 5.86402833e-01 4.49834317e-01
3.26553226e-01 -3.60562831e-01 -2.54553378e-01 -1.38274157e+00
3.12937498e-01 -9.80501950e-01 8.41024578e-01 -1.18946779e+00
-1.22016597e+00 4.30910349e-01 3.14162314e-01 -1.70132923e+00
-1.29877431e-02 -2.64063507e-01 -4.71000463e-01 5.08857906e-01
-1.39010000e+00 -1.37180269e+00 -7.12954104e-01 1.02513707e+00
7.68247545e-01 1.85236931e-01 1.05338067e-01 4.49742585e-01
-4.00506049e-01 5.00186145e-01 -5.00842154e-01 3.75214964e-01
1.12109911e+00 -9.59634125e-01 -1.12751640e-01 1.23720455e+00
3.58931214e-01 6.79027140e-01 9.76786077e-01 -7.10641086e-01
-1.15127492e+00 -1.26363826e+00 8.21651697e-01 -5.98471999e-01
9.32842553e-01 -2.56442368e-01 -9.56968904e-01 7.70732939e-01
3.09565336e-01 1.54735809e-02 5.99059045e-01 -5.68389781e-02
-9.46677566e-01 -2.23049045e-01 -7.47987211e-01 7.76363134e-01
1.35357392e+00 -1.13416958e+00 -6.21584833e-01 2.78338462e-01
8.57255638e-01 -4.98536080e-01 -7.42121637e-01 3.96905065e-01
7.22820461e-01 -9.57543671e-01 9.94470537e-01 -6.73494220e-01
5.96211493e-01 -6.22294307e-01 -4.82357740e-01 -8.53742301e-01
-2.94817895e-01 -7.16245949e-01 -2.54071832e-01 1.31930709e+00
1.62556738e-01 -1.46299273e-01 7.03172505e-01 4.38567281e-01
-7.66919330e-02 -5.55913508e-01 -8.81251633e-01 -1.06397986e+00
-3.63864720e-01 -6.67944372e-01 6.01731479e-01 1.22519565e+00
1.10358946e-01 2.10448638e-01 -5.87092996e-01 4.21447128e-01
4.24710512e-01 3.36611718e-01 6.20349646e-01 -6.14759982e-01
-2.54740685e-01 -2.79521316e-01 -3.99096102e-01 -1.17194068e+00
9.91499498e-02 -8.52703393e-01 1.33324206e-01 -1.27921367e+00
2.23704979e-01 -1.63984910e-01 -6.07277155e-01 6.47649825e-01
-5.00388294e-02 5.02247632e-01 3.35061222e-01 5.21996975e-01
-1.00535142e+00 4.36678469e-01 1.22767913e+00 -2.85363406e-01
-1.18728116e-01 -3.90319079e-01 -4.91556168e-01 5.63452542e-01
7.74215162e-01 -4.80671793e-01 -7.99259424e-01 -5.47038674e-01
3.03802580e-01 -8.00467003e-03 8.37035537e-01 -9.80301261e-01
2.72801101e-01 -5.10582745e-01 2.19320446e-01 -7.09519029e-01
2.52720624e-01 -8.71433973e-01 1.59564078e-01 3.26775461e-01
-5.40953577e-01 5.09019494e-01 3.31175596e-01 8.43733013e-01
-4.02857631e-01 1.89116657e-01 6.49025559e-01 3.92466271e-03
-1.25971889e+00 4.47719723e-01 -1.13371402e-01 3.52168977e-01
1.14338291e+00 -5.48252821e-01 -4.90920335e-01 -5.08118272e-01
-7.08993614e-01 2.72526324e-01 5.08083045e-01 8.39257300e-01
6.08212769e-01 -1.59419560e+00 -3.22660804e-01 -8.97672772e-02
5.07334411e-01 -4.51148301e-01 5.25766253e-01 1.21808302e+00
-1.76186055e-01 4.12366420e-01 -2.36854240e-01 -9.10588443e-01
-1.34223151e+00 8.64831984e-01 3.30431730e-01 8.69806781e-02
-5.48216462e-01 7.67678201e-01 4.97706592e-01 3.64452600e-01
3.11836302e-01 -3.64180714e-01 9.19536781e-03 2.69150198e-01
3.85286331e-01 2.25762904e-01 -1.83172882e-01 -8.95271122e-01
-4.51145232e-01 5.24522483e-01 1.38092071e-01 -3.59795213e-01
9.09331679e-01 -4.60013360e-01 2.24769711e-01 6.46898806e-01
1.30620837e+00 9.27474233e-04 -1.52040553e+00 -3.31254780e-01
1.13058463e-01 -7.94544339e-01 -1.52229339e-01 -6.08990550e-01
-9.67698157e-01 7.53715098e-01 5.04212499e-01 9.23104137e-02
1.23699701e+00 3.81187052e-02 7.91456163e-01 2.43593939e-02
3.88998777e-01 -1.06969917e+00 5.67347765e-01 2.96288699e-01
9.85246241e-01 -1.34274936e+00 -1.63527682e-01 -3.25235486e-01
-8.41364980e-01 8.49037826e-01 8.87384117e-01 1.78256527e-01
3.71399909e-01 -2.11658880e-01 1.15298674e-01 -3.12759012e-01
-7.16844261e-01 -2.97011524e-01 5.90424001e-01 4.69467044e-01
2.22758025e-01 -2.52928376e-01 -9.88661721e-02 4.43342716e-01
-3.85860391e-02 -2.08316192e-01 1.61736637e-01 8.71243119e-01
-1.07029885e-01 -6.61659956e-01 -3.44611555e-01 2.66203959e-03
-3.62957358e-01 -8.35761130e-02 -2.94175476e-01 7.97559500e-01
3.62842321e-01 1.00500906e+00 1.42827496e-01 -6.42565668e-01
2.60579646e-01 -1.71524865e-04 3.41230094e-01 -4.28109199e-01
-4.25984532e-01 1.21733591e-01 6.95595741e-02 -1.08288777e+00
-7.28624165e-01 -5.60362637e-01 -9.83673453e-01 -2.00679317e-01
1.62509471e-01 2.22122073e-01 2.30966568e-01 6.91227078e-01
5.32374918e-01 3.41366917e-01 4.29334044e-01 -7.40005374e-01
-1.88569769e-01 -6.48438334e-01 -2.54099697e-01 1.06460404e+00
4.74120319e-01 -8.48917425e-01 -3.80172759e-01 5.14140487e-01]
|
[9.994281768798828, 0.7427493333816528]
|
13af72d2-a543-40e3-a41a-64488db3e94f
|
lazy-modeling-of-variants-of-token-swapping
|
1809.05959
| null |
http://arxiv.org/abs/1809.05959v1
|
http://arxiv.org/pdf/1809.05959v1.pdf
|
Lazy Modeling of Variants of Token Swapping Problem and Multi-agent Path Finding through Combination of Satisfiability Modulo Theories and Conflict-based Search
|
We address item relocation problems in graphs in this paper. We assume items
placed in vertices of an undirected graph with at most one item per vertex.
Items can be moved across edges while various constraints depending on the type
of relocation problem must be satisfied. We introduce a general problem
formulation that encompasses known types of item relocation problems such as
multi-agent path finding (MAPF) and token swapping (TSWAP). In this formulation
we express two new types of relocation problems derived from token swapping
that we call token rotation (TROT) and token permutation (TPERM). Our solving
approach for item relocation combines satisfiability modulo theory (SMT) with
conflict-based search (CBS). We interpret CBS in the SMT framework where we
start with the basic model and refine the model with a collision resolution
constraint whenever a collision between items occurs in the current solution.
The key difference between the standard CBS and our SMT-based modification of
CBS (SMT-CBS) is that the standard CBS branches the search to resolve the
collision while in SMT-CBS we iteratively add a single disjunctive collision
resolution constraint. Experimental evaluation on several benchmarks shows that
the SMT-CBS algorithm significantly outperforms the standard CBS. We also
compared SMT-CBS with a modification of the SAT-based MDD-SAT solver that uses
an eager modeling of item relocation in which all potential collisions are
eliminated by constrains in advance. Experiments show that lazy approach in
SMT-CBS produce fewer constraint than MDD-SAT and also achieves faster solving
run-times.
|
['Pavel Surynek']
|
2018-09-16
| null | null | null | null |
['multi-agent-path-finding']
|
['playing-games']
|
[ 2.49379098e-01 1.93335801e-01 -2.45862126e-01 -1.10329194e-02
-3.28462362e-01 -7.80409992e-01 1.76519454e-01 5.51958084e-01
-2.83620089e-01 1.38781726e+00 -2.04244927e-01 -4.63385433e-01
-7.50255883e-01 -1.05584681e+00 -5.64583957e-01 -4.40950990e-01
-3.93601745e-01 1.06436121e+00 9.01283979e-01 -6.08754396e-01
3.65322620e-01 4.66909707e-01 -9.64241743e-01 4.25718546e-01
5.95852613e-01 5.77738404e-01 3.05523843e-01 4.20298427e-01
-4.93692398e-01 4.23261851e-01 -5.90878367e-01 -6.10215217e-02
7.04911709e-01 -1.74792916e-01 -1.57538927e+00 7.47690648e-02
-1.12922542e-01 7.00493902e-02 4.10360783e-01 6.71012759e-01
-1.13321178e-01 2.71352768e-01 3.78917634e-01 -2.28815269e+00
-2.99564060e-02 1.06354630e+00 -1.19747186e+00 3.92973512e-01
9.14745569e-01 -3.41245770e-01 9.32429314e-01 -3.78558785e-01
1.00562763e+00 1.12458682e+00 5.17717242e-01 1.33935556e-01
-1.17644906e+00 -5.02150893e-01 6.38413966e-01 5.82037628e-01
-1.50419223e+00 1.17532909e-01 1.68244004e-01 2.03267783e-01
1.73131585e+00 8.59244585e-01 5.55854619e-01 1.43892497e-01
5.11350334e-01 4.22920734e-01 1.12572312e+00 -6.72453165e-01
4.24705833e-01 -3.18995072e-03 4.42975909e-01 2.88459748e-01
9.48972523e-01 -1.66120738e-01 -3.08784574e-01 -3.63968045e-01
6.74902618e-01 -1.14535481e-01 1.49560645e-01 -3.54565442e-01
-1.31570590e+00 7.87923038e-01 1.12373471e-01 2.88662702e-01
-4.38378125e-01 4.12303746e-01 5.67732334e-01 4.57396239e-01
-3.27330112e-01 3.04713339e-01 -6.81476951e-01 2.55291194e-01
-7.58284926e-01 6.87094629e-01 8.74037862e-01 1.43579757e+00
6.45097196e-01 -1.13552734e-01 -5.34610115e-02 3.00272733e-01
1.38889402e-01 3.11163008e-01 -1.07993864e-01 -6.49570048e-01
6.16292655e-01 3.19019973e-01 3.63930970e-01 -6.17310107e-01
-4.76849288e-01 -2.58398682e-01 -2.80986518e-01 2.13034019e-01
1.59414098e-01 -9.85105857e-02 -8.97069693e-01 1.76354504e+00
6.39678717e-01 4.99275148e-01 9.87918302e-02 7.47694850e-01
2.19602242e-01 8.58215451e-01 -1.17780611e-01 -7.35918343e-01
1.44050062e+00 -1.11146975e+00 -6.74163163e-01 8.25779513e-02
7.08496094e-01 -7.79144764e-01 -1.56877667e-01 7.53829837e-01
-1.54721940e+00 1.57694966e-01 -1.01712549e+00 4.15351927e-01
-5.99837184e-01 -1.00771558e+00 8.72529984e-01 6.63536608e-01
-1.24086952e+00 -6.85662962e-03 -6.65236652e-01 -4.29672718e-01
-3.23086530e-01 9.68348086e-01 -6.29294634e-01 -2.65272021e-01
-1.02430463e+00 9.60875332e-01 5.09086251e-01 -1.17340729e-01
-5.15061736e-01 -4.82139498e-01 -7.87675917e-01 1.70306548e-01
1.17787623e+00 -7.16211855e-01 1.35469353e+00 -7.00273573e-01
-1.09820306e+00 3.66550237e-01 -3.77743393e-01 -4.82843101e-01
-2.10678261e-02 6.79781437e-01 -5.39756358e-01 -1.14782155e-01
4.54621553e-01 3.93626899e-01 2.47086987e-01 -1.44829428e+00
-1.12337780e+00 6.61561266e-02 5.12154818e-01 6.24436326e-02
6.17983997e-01 2.00065821e-01 -2.97246277e-01 -2.03278869e-01
3.12791795e-01 -1.14410865e+00 -3.91955942e-01 -7.94226289e-01
-6.80139363e-01 -2.63045847e-01 7.08579123e-01 1.61714256e-01
1.37215066e+00 -1.45042241e+00 2.94864058e-01 7.31197536e-01
-6.33323491e-02 -1.01072848e-01 -3.29684764e-01 1.42398536e+00
-4.79000509e-01 1.64508611e-01 2.23674588e-02 -1.03330374e-01
9.65704620e-02 7.06950665e-01 -3.66503261e-02 3.13206881e-01
-9.87672955e-02 4.74983633e-01 -9.04704332e-01 -5.69673598e-01
6.81660650e-03 -3.66006017e-01 -7.73350716e-01 -5.62942863e-01
-4.60166126e-01 -2.63374805e-01 -3.41010869e-01 6.89867556e-01
1.28938138e+00 1.27707362e-01 6.27257347e-01 2.05479279e-01
-6.69600427e-01 3.04530919e-01 -2.16121364e+00 1.41490448e+00
-7.28968754e-02 -1.56102210e-01 1.85981825e-01 -9.15975451e-01
2.89904445e-01 2.24072754e-01 4.47592705e-01 -7.30773807e-01
9.29095671e-02 1.50944039e-01 3.67556103e-02 -2.82894164e-01
8.45982730e-01 -1.67545855e-01 -3.48143905e-01 6.05294645e-01
-3.68001610e-01 2.01743647e-01 7.89974451e-01 5.66035986e-01
1.29337156e+00 -2.52249926e-01 5.94995916e-01 -5.15027642e-01
6.52894795e-01 6.11096084e-01 9.77141201e-01 7.76324332e-01
1.58086181e-01 -9.70299318e-02 6.54391050e-01 -3.51232320e-01
-4.36145961e-01 -9.92724955e-01 4.41854261e-02 7.38503993e-01
5.94987929e-01 -6.15906835e-01 -3.61754417e-01 -6.23461723e-01
1.03342019e-01 7.42539406e-01 -4.94303554e-01 2.91005850e-01
-7.11013138e-01 -5.45549512e-01 7.46880695e-02 4.60902750e-01
1.06204301e-01 -8.21008921e-01 -6.81128919e-01 5.02758265e-01
-7.67513216e-02 -9.47409093e-01 -4.08386797e-01 5.15783489e-01
-4.31602329e-01 -1.19100654e+00 2.55812883e-01 -9.98837471e-01
8.21438432e-01 8.32405686e-01 8.65557492e-01 4.55792814e-01
-2.37771496e-01 7.52883479e-02 -7.53560305e-01 -4.36169833e-01
1.91203758e-01 -4.31134403e-02 -4.39885706e-02 -6.35710835e-01
8.07583183e-02 -3.78011227e-01 -2.10012332e-01 4.81839478e-01
-1.02304256e+00 -9.99670550e-02 2.88143337e-01 6.85889006e-01
7.24270999e-01 7.05536485e-01 6.98572755e-01 -1.13116252e+00
4.67454702e-01 -9.16553557e-01 -8.15553486e-01 2.93122232e-01
-5.59717596e-01 9.18987021e-02 5.14059246e-01 -2.42052630e-01
-7.05322564e-01 -1.76660269e-01 2.99268633e-01 2.82178894e-02
7.83785880e-02 9.85201836e-01 -1.09713942e-01 -8.93825740e-02
-1.57112211e-01 -3.30903620e-01 -4.94250387e-01 -3.89676285e-03
4.00460996e-02 3.64906117e-02 2.73568600e-01 -8.73078227e-01
7.28287876e-01 3.30039620e-01 7.06821442e-01 1.61892530e-02
7.09153861e-02 -5.65473199e-01 -1.92150846e-01 6.66938722e-02
3.97618204e-01 -3.35522443e-01 -9.91889119e-01 1.08904444e-01
-1.05229068e+00 -2.88098156e-01 -3.09805632e-01 2.11838722e-01
-2.40333870e-01 4.13706064e-01 -4.95728105e-01 -9.49232399e-01
-7.91133642e-02 -1.12742841e+00 6.27634525e-01 3.58836949e-02
-2.69149691e-01 -9.10233676e-01 3.59825045e-01 -2.99225915e-02
2.14950442e-01 5.43282628e-01 1.14975417e+00 -7.63873816e-01
-8.47787619e-01 2.32954100e-02 -7.83719309e-03 -6.80134773e-01
9.98163149e-02 -1.71020746e-01 1.23591088e-01 -3.96319926e-01
-3.94628435e-01 2.91693300e-01 1.16865195e-01 4.27077740e-01
4.52455789e-01 -2.92113572e-01 -8.77874911e-01 5.17668650e-02
2.19992661e+00 6.49314761e-01 7.24880099e-01 9.30817485e-01
5.78596676e-03 3.00511569e-01 1.08914340e+00 4.67711180e-01
5.85949898e-01 9.39901769e-01 7.34324872e-01 3.82764488e-02
2.35793352e-01 2.82184541e-01 7.74522424e-02 3.93022627e-01
-1.02406807e-01 -9.55313563e-01 -6.79520071e-01 8.64223242e-01
-2.23351717e+00 -1.06317008e+00 -1.03319383e+00 2.34762812e+00
6.75087869e-01 2.89107174e-01 4.19110894e-01 5.54805040e-01
7.31887639e-01 -5.07622838e-01 4.91751172e-02 -1.42436075e+00
2.65939713e-01 4.39159751e-01 1.11628413e+00 9.91087556e-01
-3.12189758e-01 9.39220905e-01 6.42497587e+00 5.04740417e-01
-7.19265461e-01 2.18167990e-01 -3.07723105e-01 -1.80426374e-01
-2.23723039e-01 4.93702203e-01 -9.23270822e-01 2.86461174e-01
7.04997540e-01 -3.75789702e-01 5.96452534e-01 3.45678329e-01
3.18008661e-01 -7.38834739e-01 -1.17133224e+00 3.96910876e-01
1.64326634e-02 -1.21565342e+00 -3.24806222e-03 2.49582604e-01
8.28632593e-01 -4.58399713e-01 -1.90822750e-01 1.19855113e-01
5.18170297e-01 -7.08290875e-01 8.81140471e-01 -8.10527802e-02
2.04083294e-01 -1.05588281e+00 8.89744520e-01 1.34855568e-01
-1.61377025e+00 -2.12526321e-03 -2.66982555e-01 -3.59714627e-01
6.38814390e-01 2.58073866e-01 -1.04535079e+00 1.31301033e+00
4.53462124e-01 -9.45177153e-02 1.21866927e-01 1.50827765e+00
2.39368826e-02 -3.53054926e-02 -6.59709454e-01 2.74801522e-01
5.02778769e-01 -2.53532588e-01 6.25336707e-01 1.05986977e+00
1.94545597e-01 3.05201977e-01 4.30279940e-01 4.87428308e-01
5.22653699e-01 -2.09617928e-01 -1.23044200e-01 7.74846852e-01
1.01323664e+00 1.18606579e+00 -1.26660872e+00 -2.35263586e-01
-4.98299927e-01 5.67863464e-01 -6.03978783e-02 1.44389600e-01
-1.08940375e+00 -3.04757684e-01 5.25798917e-01 3.21602494e-01
5.99069536e-01 -2.03482494e-01 -6.54569566e-02 -4.16262984e-01
-1.75561532e-01 -8.60144496e-01 6.58284724e-01 -7.24091709e-01
-9.73868787e-01 5.32437146e-01 9.20078278e-01 -8.61925423e-01
-2.06783101e-01 -3.93425465e-01 -8.73422563e-01 6.87133908e-01
-1.59194505e+00 -1.07096529e+00 5.68085536e-02 9.42043781e-01
6.51459813e-01 4.26668733e-01 6.35441601e-01 9.07312632e-02
-7.69384563e-01 4.38491911e-01 -2.82630801e-01 -6.52984738e-01
4.37930286e-01 -1.31976509e+00 2.64686439e-02 1.14329004e+00
-3.66382211e-01 8.48522604e-01 8.93430352e-01 -8.96918535e-01
-1.76436102e+00 -8.69326949e-01 9.60520029e-01 2.15597123e-01
4.89538014e-01 -1.32382125e-01 -5.09440899e-01 1.30767143e+00
6.78138614e-01 -3.27139705e-01 5.24127603e-01 1.10188341e-02
-7.64506459e-02 -5.87966293e-02 -1.51399064e+00 2.66615957e-01
9.66131628e-01 4.10267234e-01 -6.61664546e-01 5.53325117e-01
4.03006226e-01 -4.79516119e-01 -4.69590127e-01 9.65323746e-02
1.69229522e-01 -6.91853046e-01 7.86262691e-01 -5.45317292e-01
-3.16726118e-01 -8.83590996e-01 -5.57584167e-02 -1.08574200e+00
-8.60722542e-01 -6.23580337e-01 4.83001530e-01 1.14030325e+00
7.14534938e-01 -9.14063811e-01 4.85140890e-01 7.29303300e-01
-5.31075001e-01 -6.77787066e-01 -1.15967929e+00 -1.27941251e+00
-1.58715487e-01 -1.37371063e-01 1.14841974e+00 1.09938121e+00
7.93202400e-01 9.40330699e-02 -2.91339129e-01 5.51922977e-01
4.89974797e-01 2.14236438e-01 4.78538573e-01 -1.11790812e+00
-6.06041312e-01 -1.27716318e-01 -1.11279078e-01 -2.39731997e-01
-1.92951765e-02 -1.05659413e+00 -1.52854711e-01 -2.20934463e+00
3.27421904e-01 -7.55192697e-01 -2.03020930e-01 1.03230774e+00
4.33973551e-01 -1.41816661e-01 2.03484669e-01 -2.44266763e-01
-8.91013980e-01 -4.23506349e-01 9.07822788e-01 1.19373024e-01
-3.74620080e-01 -3.66553247e-01 -4.50154573e-01 2.07170993e-01
6.84001088e-01 -9.02912617e-01 -5.63315630e-01 -4.47401285e-01
8.57677937e-01 5.55540562e-01 1.42908478e-02 -5.90902925e-01
4.84727949e-01 -9.55420256e-01 -3.36871594e-01 -8.38113129e-01
1.63885832e-01 -1.10602641e+00 1.01844096e+00 7.31540859e-01
1.55775890e-01 7.71002233e-01 5.62376320e-01 3.48906726e-01
1.90655708e-01 -7.03088343e-01 2.60579973e-01 -9.22394171e-02
-7.92347848e-01 -1.01952709e-01 -8.56480837e-01 -4.73144710e-01
1.75042379e+00 -6.70295060e-01 -8.96853924e-01 3.47784325e-03
-6.72701597e-01 7.35897601e-01 4.54312533e-01 1.34796143e-01
4.83996183e-01 -1.08164310e+00 -2.98353821e-01 -8.94106403e-02
-1.32511765e-01 4.07291278e-02 1.28536016e-01 1.14016128e+00
-6.81956291e-01 6.97662830e-01 -4.77217406e-01 -7.13542849e-02
-1.48271406e+00 1.09718704e+00 -3.07903998e-02 -6.98011339e-01
-3.09926003e-01 6.82053268e-01 -1.71987683e-01 3.19510370e-01
-2.68733799e-01 -6.64457500e-01 3.96918841e-02 -2.76764959e-01
3.71995807e-01 6.24774694e-01 2.59436578e-01 -3.59643936e-01
-1.04590452e+00 3.90715033e-01 -4.52421427e-01 -1.86143473e-01
1.37394035e+00 -2.37760171e-01 -5.24531007e-01 -1.41699106e-01
3.66780221e-01 4.56460476e-01 -6.64243773e-02 1.36595145e-01
2.52350662e-02 -5.24608731e-01 2.71063554e-03 -9.48081851e-01
-1.10533309e+00 -4.00764942e-01 -1.04487546e-01 6.43251002e-01
1.23155034e+00 -1.95195198e-01 8.37536335e-01 1.61450118e-01
1.12976635e+00 -9.77013052e-01 -4.04432744e-01 4.46013659e-01
6.12074554e-01 -3.47677708e-01 3.07508171e-01 -1.24652398e+00
-5.19512773e-01 7.60272026e-01 8.22113633e-01 -3.11292112e-01
2.48471603e-01 8.73456299e-01 -7.06389189e-01 -2.27169648e-01
-1.05440986e+00 -2.52683222e-01 -6.53727651e-01 7.71354437e-01
-2.46540219e-01 8.99963006e-02 -1.03959608e+00 6.67581320e-01
-7.38182440e-02 3.82634439e-02 1.14322579e+00 1.65785098e+00
-4.26424772e-01 -1.99375010e+00 -8.20128143e-01 -8.94337427e-03
-2.65842915e-01 -6.47774786e-02 -3.01182628e-01 1.28369868e+00
5.70144713e-01 1.33579135e+00 1.08403333e-01 -1.26544073e-01
2.85442561e-01 -1.62224337e-01 8.47326994e-01 -8.67074788e-01
-1.01689267e+00 3.10772955e-01 6.56894267e-01 -4.33310896e-01
-5.18445790e-01 -9.48891103e-01 -1.82442498e+00 -7.35488772e-01
-8.25059414e-01 5.97539902e-01 4.15544689e-01 1.00154972e+00
1.62009314e-01 6.97371483e-01 2.82042950e-01 -2.86205858e-01
1.21216826e-01 -3.84816527e-01 -8.05765450e-01 -1.99806571e-01
4.33843583e-02 -9.27693069e-01 3.70411612e-02 -3.67875695e-01]
|
[4.984126091003418, 1.8358161449432373]
|
bc128722-46df-4f4a-ab72-c48d4b8dd645
|
segmenting-transparent-objects-in-the-wild
|
2003.13948
| null |
https://arxiv.org/abs/2003.13948v3
|
https://arxiv.org/pdf/2003.13948v3.pdf
|
Segmenting Transparent Objects in the Wild
|
Transparent objects such as windows and bottles made by glass widely exist in the real world. Segmenting transparent objects is challenging because these objects have diverse appearance inherited from the image background, making them had similar appearance with their surroundings. Besides the technical difficulty of this task, only a few previous datasets were specially designed and collected to explore this task and most of the existing datasets have major drawbacks. They either possess limited sample size such as merely a thousand of images without manual annotations, or they generate all images by using computer graphics method (i.e. not real image). To address this important problem, this work proposes a large-scale dataset for transparent object segmentation, named Trans10K, consisting of 10,428 images of real scenarios with carefully manual annotations, which are 10 times larger than the existing datasets. The transparent objects in Trans10K are extremely challenging due to high diversity in scale, viewpoint and occlusion as shown in Fig. 1. To evaluate the effectiveness of Trans10K, we propose a novel boundary-aware segmentation method, termed TransLab, which exploits boundary as the clue to improve segmentation of transparent objects. Extensive experiments and ablation studies demonstrate the effectiveness of Trans10K and validate the practicality of learning object boundary in TransLab. For example, TransLab significantly outperforms 20 recent object segmentation methods based on deep learning, showing that this task is largely unsolved. We believe that both Trans10K and TransLab have important contributions to both the academia and industry, facilitating future researches and applications.
|
['Wenhai Wang', 'Mingyu Ding', 'Chunhua Shen', 'Ping Luo', 'Wenjia Wang', 'Enze Xie']
|
2020-03-31
| null |
https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/2016_ECCV_2020_paper.php
|
https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123580681.pdf
|
eccv-2020-8
|
['transparent-objects']
|
['computer-vision']
|
[ 1.38939902e-01 3.26673277e-02 2.46473521e-01 -3.32583368e-01
-1.68443218e-01 -5.70153296e-01 1.56745791e-01 -5.08890212e-01
-1.52573764e-01 4.72267538e-01 -4.12546128e-01 -2.09706813e-01
2.47805968e-01 -7.10774899e-01 -7.57899404e-01 -7.28579640e-01
1.36868641e-01 2.86792219e-01 8.42187524e-01 -2.65636891e-02
1.47569895e-01 1.58376724e-01 -1.32691419e+00 4.17384893e-01
1.25783169e+00 1.19456279e+00 4.14417565e-01 4.24579501e-01
-3.29355538e-01 2.40105644e-01 -7.07275808e-01 -5.14264643e-01
7.49708593e-01 -1.53311998e-01 -6.97839677e-01 3.44346255e-01
7.13618219e-01 -5.78451812e-01 -1.73116592e-03 9.63666499e-01
3.25378418e-01 1.17814597e-02 6.45540237e-01 -1.14476204e+00
-8.24975312e-01 4.34538990e-01 -9.33249056e-01 -5.08487858e-02
6.29073456e-02 1.63243368e-01 6.34663045e-01 -9.04455483e-01
3.87349576e-01 1.07559979e+00 7.04995155e-01 5.41099131e-01
-9.25390005e-01 -8.70942235e-01 6.36907518e-01 -1.28906354e-01
-1.21145713e+00 -2.42507130e-01 7.02091098e-01 -5.26804328e-01
4.39693362e-01 6.07078254e-01 9.56445813e-01 1.00550985e+00
-1.50426880e-01 9.93660629e-01 1.28282166e+00 -1.78346485e-01
5.49043193e-02 5.02139628e-01 1.96244240e-01 6.89516068e-01
4.92257535e-01 -2.04074502e-01 -7.65455961e-02 1.27714068e-01
9.32615995e-01 8.14193580e-03 -5.27699590e-01 -4.51826960e-01
-1.27300262e+00 5.19834816e-01 6.13796473e-01 -2.22518351e-02
2.67584771e-01 5.28689055e-03 3.03558648e-01 4.29465668e-03
4.58020210e-01 2.03409657e-01 -5.69925487e-01 2.69728303e-01
-6.84044540e-01 7.39915594e-02 7.00115263e-01 1.41021085e+00
7.22098172e-01 4.41928772e-04 -7.94455558e-02 8.52324665e-01
4.93539780e-01 4.84990507e-01 5.16919434e-01 -7.31355309e-01
6.19480669e-01 8.80789518e-01 1.72800586e-01 -9.64534581e-01
-4.07448053e-01 -9.14079323e-02 -5.93386471e-01 2.33015761e-01
5.79535604e-01 -4.39911671e-02 -1.15572858e+00 1.10963690e+00
7.83441305e-01 1.55729100e-01 -9.46316347e-02 1.09117937e+00
1.40127611e+00 5.59642375e-01 -2.72110730e-01 3.24781947e-02
1.41364110e+00 -1.45062673e+00 -5.25013983e-01 -2.35036552e-01
4.50363606e-01 -1.03300691e+00 1.55081260e+00 5.49773991e-01
-8.31369817e-01 -5.37200749e-01 -7.60160148e-01 5.03440052e-02
-4.01896030e-01 2.94468313e-01 1.10575747e+00 1.01696181e+00
-7.24038005e-01 2.27507688e-02 -5.48070610e-01 -3.47701550e-01
6.73153818e-01 3.94693881e-01 -5.29703125e-02 3.94729562e-02
-7.44212091e-01 3.00050765e-01 3.49854708e-01 5.02008379e-01
-6.93474412e-01 -7.18878210e-01 -6.55686080e-01 -4.37622577e-01
6.96649075e-01 -5.66723347e-01 1.27124083e+00 -9.32754040e-01
-1.56525624e+00 8.02304149e-01 1.80383608e-01 -3.83554660e-02
1.05485785e+00 -4.51102287e-01 -2.21730739e-01 1.24188669e-01
-6.63291961e-02 5.01507580e-01 9.87901628e-01 -1.76964593e+00
-5.23936629e-01 -1.94264725e-01 2.02719763e-01 3.53635028e-02
-4.86846477e-01 -3.53730144e-03 -1.09615219e+00 -7.11689651e-01
2.69888341e-01 -1.05348039e+00 -3.17773700e-01 3.52792919e-01
-7.96263754e-01 -1.73247345e-02 1.28694904e+00 -4.70189273e-01
9.43421423e-01 -1.95477998e+00 -4.11054552e-01 6.71479292e-03
4.23688889e-01 6.27041101e-01 1.55569986e-02 4.99425530e-02
2.08736852e-01 3.05872560e-01 -5.15059650e-01 -3.69720846e-01
-1.91230565e-01 1.01830512e-01 -3.09294045e-01 3.69228780e-01
-8.76558870e-02 9.36785161e-01 -5.99809706e-01 -8.29130828e-01
3.85700762e-01 2.80881077e-01 -4.01751101e-01 7.87674338e-02
-3.66971523e-01 5.89341164e-01 -7.27109492e-01 8.18390965e-01
1.20545590e+00 -1.35641977e-01 -3.41657519e-01 -4.06599164e-01
-9.50765908e-02 -3.15862447e-01 -1.16533506e+00 1.33990335e+00
-3.03125143e-01 5.62136471e-01 1.86204873e-02 -6.76081359e-01
8.59561503e-01 4.84411679e-02 4.55458641e-01 -6.81376278e-01
2.64681250e-01 2.44471863e-01 -9.38431919e-02 -8.73679817e-01
5.39757073e-01 1.80988103e-01 1.93907559e-01 3.11116517e-01
-7.74091542e-01 -5.33591270e-01 1.92549750e-01 1.21437296e-01
4.87037331e-01 3.89639258e-01 -2.22338870e-01 -1.06423497e-01
1.41230568e-01 -8.82189125e-02 9.22545791e-01 7.34946966e-01
-3.80231559e-01 1.12061965e+00 1.54189065e-01 -5.88736892e-01
-7.22934604e-01 -9.61849630e-01 -4.74501669e-01 5.46895981e-01
1.05380094e+00 -2.05831960e-01 -1.02425802e+00 -8.37595046e-01
4.24848348e-02 2.00327963e-01 -6.77057266e-01 9.34482515e-02
-6.75316095e-01 -1.05174100e+00 1.03346087e-01 6.77371681e-01
9.88375306e-01 -1.13489652e+00 -7.03669012e-01 -4.75598238e-02
-2.37416849e-01 -1.77201712e+00 -6.53032124e-01 -2.96838015e-01
-1.05912459e+00 -1.22762060e+00 -9.09424424e-01 -6.73667610e-01
8.25545609e-01 7.47924387e-01 1.13069606e+00 1.26772076e-01
-6.24751151e-01 1.74704745e-01 -3.98896784e-01 -6.14422917e-01
-2.12500572e-01 -2.02261284e-01 -1.32804304e-01 1.84412450e-01
-2.61792094e-02 -2.99843222e-01 -1.08758652e+00 7.46102929e-01
-9.57605779e-01 3.19322079e-01 5.74833333e-01 2.62362182e-01
4.99200225e-01 2.05635607e-01 3.06465715e-01 -1.24061346e+00
2.51352668e-01 -9.88711268e-02 -7.62913764e-01 3.90594035e-01
-1.83010146e-01 -7.19086170e-01 4.12524879e-01 -6.06901467e-01
-1.28348804e+00 3.57474461e-02 2.76219398e-01 -2.42578819e-01
-8.25124159e-02 -6.20142147e-02 -3.59145850e-01 -3.92246932e-01
2.58839399e-01 3.01405992e-02 -3.83610815e-01 -4.98845726e-01
2.09536165e-01 6.97542727e-01 2.66335070e-01 -6.26888335e-01
9.41931069e-01 9.99710619e-01 -4.60846752e-01 -9.51262355e-01
-8.42512906e-01 -4.54485148e-01 -6.14485443e-01 -3.70583355e-01
9.17388439e-01 -7.58190691e-01 -7.00957060e-01 8.22267294e-01
-1.04514253e+00 -7.13079453e-01 -1.64552331e-01 4.69946176e-01
-2.04286307e-01 6.83569610e-01 -6.57658577e-01 -7.90533662e-01
-3.71002108e-01 -1.36282170e+00 1.17568076e+00 5.73107898e-01
8.86064991e-02 -7.95014799e-01 -4.74900752e-01 8.71141255e-01
1.40290737e-01 2.31595621e-01 7.26382375e-01 2.00275537e-02
-1.12797713e+00 2.96104997e-02 -6.75841808e-01 2.65651345e-01
4.72442716e-01 4.62214231e-01 -1.24879515e+00 -2.40782395e-01
-8.40835795e-02 -2.81313390e-01 9.48813736e-01 3.36964488e-01
1.54217994e+00 1.07470620e-02 -4.44642007e-01 7.61038661e-01
1.36909413e+00 3.52495044e-01 5.62567234e-01 4.48122710e-01
1.08064330e+00 7.57603884e-01 8.36731672e-01 3.37042093e-01
2.95721084e-01 7.36980975e-01 6.34270668e-01 -5.91311693e-01
-3.37124348e-01 2.16169268e-01 1.88766301e-01 8.18192184e-01
-3.01534891e-01 -3.25249463e-01 -6.65457904e-01 4.15883213e-01
-1.64963651e+00 -3.90626758e-01 -6.76135898e-01 2.13603473e+00
7.79268086e-01 2.19949648e-01 1.22970067e-01 -7.65502527e-02
5.85405767e-01 -2.23574534e-01 -7.80108750e-01 -1.53524891e-01
-2.05155343e-01 -3.39479536e-01 4.00872499e-01 3.77482474e-02
-1.19419980e+00 1.11421835e+00 6.20103168e+00 8.65905583e-01
-1.19820487e+00 -6.59860298e-02 8.32438409e-01 2.10841447e-01
-3.30766171e-01 -8.29208866e-02 -9.45610464e-01 6.25536203e-01
4.02729493e-03 3.57265592e-01 -2.04605199e-02 8.22153568e-01
3.28186929e-01 -3.70410293e-01 -8.32077086e-01 9.88245010e-01
1.39772072e-02 -8.92602324e-01 2.35877428e-02 -2.44638342e-02
1.11416757e+00 -1.11849271e-02 3.55115801e-01 1.10301096e-02
2.15343237e-01 -9.04974461e-01 9.92715180e-01 2.15534493e-01
5.43455303e-01 -1.16484985e-01 5.54263234e-01 9.37338248e-02
-1.33313167e+00 4.90666814e-02 -5.89561701e-01 2.91075408e-01
2.04846375e-02 8.42881501e-01 -6.33147478e-01 4.93647844e-01
1.30958807e+00 5.32888114e-01 -6.63845062e-01 1.48768032e+00
-8.29571560e-02 7.57535040e-01 -5.02303243e-01 1.64397553e-01
2.34016612e-01 -6.68392301e-01 4.05155510e-01 1.15300775e+00
6.08222894e-02 -1.90641239e-01 4.89349902e-01 1.14286435e+00
-3.73765714e-02 2.25356400e-01 -3.19364905e-01 1.73664048e-01
1.65228814e-01 1.40707016e+00 -1.47461033e+00 -3.45606774e-01
-6.59323394e-01 1.02914762e+00 -1.78150013e-02 6.72003269e-01
-1.14324057e+00 -2.81969219e-01 4.41722929e-01 7.65879303e-02
4.11562562e-01 -1.46004468e-01 -4.24109727e-01 -1.19611239e+00
6.00770950e-01 -7.82565773e-01 1.17168836e-01 -7.99436092e-01
-1.29363930e+00 6.78163469e-01 -2.25019101e-02 -1.42291629e+00
7.90921211e-01 -6.79151475e-01 -6.22037709e-01 5.35700142e-01
-1.66426778e+00 -1.38959539e+00 -9.30490792e-01 4.11991149e-01
8.06659818e-01 1.64317772e-01 3.47709268e-01 2.90881068e-01
-9.26276565e-01 5.73406160e-01 2.43770733e-01 1.39194384e-01
6.94563746e-01 -1.10964930e+00 5.13056040e-01 6.53303921e-01
8.15106090e-03 5.83813071e-01 3.66454184e-01 -5.47044575e-01
-1.16458654e+00 -1.20351183e+00 -1.54367521e-01 -5.93920469e-01
3.54992568e-01 -8.09576929e-01 -9.55743492e-01 5.29281855e-01
-6.47401661e-02 3.45915794e-01 4.40975100e-01 -3.16286176e-01
-2.15079233e-01 -1.47704855e-01 -1.07987416e+00 9.34523702e-01
1.23170733e+00 3.63597497e-02 -2.51388162e-01 5.05609214e-01
1.06107330e+00 -6.41353905e-01 -6.73322439e-01 6.16532922e-01
5.68649232e-01 -1.33330631e+00 1.04617846e+00 -7.28595778e-02
2.86645710e-01 -3.46406043e-01 1.09887235e-01 -8.08704793e-01
2.29697108e-01 -6.91722810e-01 8.33044481e-03 1.45669425e+00
4.03936028e-01 -8.40203762e-01 1.11152244e+00 9.21749115e-01
-3.25980514e-01 -1.02317154e+00 -5.94624102e-01 -8.93838644e-01
-7.99794029e-03 -4.00054395e-01 7.81663239e-01 8.15598190e-01
-7.92756021e-01 -7.56096616e-02 -2.34654233e-01 1.12093262e-01
5.80466866e-01 7.86114216e-01 9.55049634e-01 -1.13616979e+00
-8.80866274e-02 -3.46721649e-01 -1.93138242e-01 -1.37564898e+00
-2.52049893e-01 -4.03045475e-01 1.64615288e-01 -1.77865350e+00
4.20155823e-01 -1.20605803e+00 1.61704123e-01 3.10545087e-01
-4.42833364e-01 6.98739231e-01 8.01855847e-02 3.65441293e-01
-6.93726540e-01 7.22924232e-01 2.12831879e+00 -4.02880788e-01
-4.40206766e-01 3.58097255e-01 -6.46550238e-01 1.17749846e+00
7.36792564e-01 -2.26346046e-01 -5.85143626e-01 -7.15504527e-01
1.26555897e-02 -4.82067972e-01 2.92509109e-01 -8.03972304e-01
-2.02634111e-01 -3.10103029e-01 4.20697629e-01 -6.69977248e-01
4.79366720e-01 -1.04499054e+00 -2.09859878e-01 1.71067074e-01
2.07172751e-01 -4.20768648e-01 2.88467556e-01 7.16723800e-01
-3.83326486e-02 -2.72612005e-01 7.21989989e-01 -3.85722667e-01
-9.22861159e-01 6.44165516e-01 -6.03488684e-02 2.15570971e-01
1.43378770e+00 -9.37500477e-01 -3.24824750e-01 2.83001047e-02
-3.50991309e-01 1.97756529e-01 6.39976919e-01 3.86489660e-01
6.81534290e-01 -8.27503622e-01 -4.26005065e-01 1.00765251e-01
2.18538672e-01 8.29438746e-01 3.36824536e-01 8.57081771e-01
-8.00774753e-01 9.87631604e-02 1.09732017e-01 -6.82977319e-01
-1.44650984e+00 3.93082678e-01 2.96939403e-01 2.24189878e-01
-1.21315849e+00 1.16633666e+00 1.05415261e+00 -6.58545375e-01
1.53144717e-01 -8.29603255e-01 -1.05919160e-01 -1.95827290e-01
2.45342463e-01 1.30118221e-01 2.19832867e-01 -5.12952149e-01
-1.41045585e-01 1.05962920e+00 -3.25297058e-01 3.96519631e-01
1.07517302e+00 -1.75976798e-01 -7.10314512e-02 2.83576339e-01
9.32527661e-01 2.03600571e-01 -1.52770412e+00 -1.21858388e-01
-2.11034000e-01 -8.31423223e-01 -3.68861705e-01 -4.80971932e-01
-1.59407473e+00 8.56099904e-01 4.64127123e-01 7.76763558e-02
1.02663815e+00 -2.38522273e-02 1.22725117e+00 3.69107634e-01
4.37919050e-01 -1.11466539e+00 4.84050959e-01 1.14691138e-01
7.80372322e-01 -1.57282400e+00 1.56358644e-01 -1.40924788e+00
-6.47297502e-01 9.31840956e-01 1.17249393e+00 2.57539153e-01
3.82456779e-01 2.77166992e-01 5.78986526e-01 -1.65221423e-01
-9.14510116e-02 -1.27987303e-02 1.82322174e-01 7.85524011e-01
2.77198702e-01 -4.54127751e-02 -3.87773812e-02 5.42508721e-01
-3.10083956e-01 -2.95307934e-01 5.51788986e-01 8.47267330e-01
-2.45627761e-01 -8.27405632e-01 -5.80191910e-01 4.31728482e-01
-2.48580426e-01 3.36874127e-02 -4.79082048e-01 8.85713160e-01
2.68869370e-01 9.40349042e-01 -1.24745727e-01 -6.09439798e-02
1.92319334e-01 -6.41346693e-01 4.20213103e-01 -5.35822570e-01
-4.67600942e-01 2.06871573e-02 -1.23373173e-01 -5.26876092e-01
-4.74316299e-01 -3.78351569e-01 -1.33712637e+00 1.38176337e-01
-8.05872858e-01 6.53848192e-03 6.75104439e-01 6.22907162e-01
6.16859756e-02 7.44384348e-01 4.65823323e-01 -8.96841168e-01
-2.13026285e-01 -7.54728317e-01 -4.88275707e-01 5.03105938e-01
8.42570662e-02 -6.87246859e-01 -2.04866573e-01 2.72593468e-01]
|
[10.054588317871094, -0.7690457105636597]
|
80777f68-fb3d-4501-8d70-e1b2c20fa92c
|
learning-to-navigate-in-turbulent-flows-with
|
2306.04781
| null |
https://arxiv.org/abs/2306.04781v1
|
https://arxiv.org/pdf/2306.04781v1.pdf
|
Learning to Navigate in Turbulent Flows with Aerial Robot Swarms: A Cooperative Deep Reinforcement Learning Approach
|
Aerial operation in turbulent environments is a challenging problem due to the chaotic behavior of the flow. This problem is made even more complex when a team of aerial robots is trying to achieve coordinated motion in turbulent wind conditions. In this paper, we present a novel multi-robot controller to navigate in turbulent flows, decoupling the trajectory-tracking control from the turbulence compensation via a nested control architecture. Unlike previous works, our method does not learn to compensate for the air-flow at a specific time and space. Instead, our method learns to compensate for the flow based on its effect on the team. This is made possible via a deep reinforcement learning approach, implemented via a Graph Convolutional Neural Network (GCNN)-based architecture, which enables robots to achieve better wind compensation by processing the spatial-temporal correlation of wind flows across the team. Our approach scales well to large robot teams -- as each robot only uses information from its nearest neighbors -- , and generalizes well to robot teams larger than seen in training. Simulated experiments demonstrate how information sharing improves turbulence compensation in a team of aerial robots and demonstrate the flexibility of our method over different team configurations.
|
['David Saldaña', 'Kostas Daniilidis', 'Juan Calderon', 'Siddharth Mayya', 'Diego Patiño']
|
2023-06-07
| null | null | null | null |
['navigate']
|
['reasoning']
|
[-1.45944342e-01 -1.06074505e-01 4.17935848e-01 5.08672148e-02
3.17816764e-01 -7.66846836e-01 1.44718692e-01 1.45198360e-01
-3.21919918e-01 6.35045111e-01 6.10569827e-02 -6.13372214e-03
-6.53233469e-01 -9.08388376e-01 -5.84154069e-01 -7.26195157e-01
-5.25003016e-01 6.48747981e-01 3.24320704e-01 -9.96081948e-01
6.80437535e-02 4.26258206e-01 -1.45108032e+00 -1.03910692e-01
7.73738623e-01 5.49294591e-01 4.97722805e-01 1.00191212e+00
3.69146973e-01 9.50027049e-01 -6.29731834e-01 4.81455326e-01
7.52176583e-01 -3.17374885e-01 -5.71663618e-01 3.77819926e-01
5.20502031e-01 -2.01563030e-01 -6.09193370e-02 6.64957583e-01
4.49421108e-01 4.84860420e-01 6.27468765e-01 -9.30330098e-01
-2.79614866e-01 4.93961334e-01 -3.20300192e-01 -1.46195590e-01
-1.85719952e-01 4.47593480e-01 7.79387116e-01 -1.35024011e-01
5.34404635e-01 1.25298953e+00 9.54934180e-01 3.44936758e-01
-1.02672493e+00 -4.59180415e-01 2.96199501e-01 -2.32084990e-01
-1.03137517e+00 1.40289471e-01 1.51632458e-01 -6.75670862e-01
7.93372393e-01 -3.93731326e-01 9.25098062e-01 5.02970338e-01
5.94285429e-01 -2.06805393e-01 7.32582867e-01 4.30259993e-03
2.00242355e-01 -2.86784440e-01 -6.31860673e-01 8.24457765e-01
5.48007607e-01 8.38253200e-02 -3.28599393e-01 3.29601914e-01
1.09145427e+00 -3.04337684e-02 -5.93294501e-01 -5.85659385e-01
-1.34934247e+00 7.98549116e-01 1.06097066e+00 1.96118146e-01
-3.99005264e-01 6.66150451e-01 3.38694483e-01 6.78759873e-01
1.17967755e-01 1.21157598e+00 -6.63049579e-01 9.31989774e-02
-4.92673069e-01 4.75654274e-01 9.26488340e-01 6.27164006e-01
7.88862824e-01 4.42888290e-01 3.87745947e-01 5.22565126e-01
1.05771013e-01 6.93253517e-01 2.94750333e-01 -1.51295805e+00
4.31328297e-01 3.48002166e-01 5.52683473e-01 -1.25435829e+00
-7.93899953e-01 -4.07407969e-01 -9.38129842e-01 9.18731034e-01
4.81213659e-01 -9.73595440e-01 -4.81711835e-01 1.63791060e+00
4.10752684e-01 -2.93321721e-02 3.37589473e-01 1.36154425e+00
-2.65133400e-02 5.00401855e-01 -5.76958179e-01 -1.31804287e-01
7.99766183e-01 -1.27815962e+00 -6.57368660e-01 -5.39093435e-01
8.64979684e-01 -5.25919497e-01 5.82060754e-01 4.69859898e-01
-7.99197912e-01 -5.57495236e-01 -9.61155236e-01 5.50933778e-01
-3.20564866e-01 6.16643950e-02 4.81836468e-01 1.82592705e-01
-1.47752547e+00 1.02731621e+00 -8.44773114e-01 -5.77127635e-01
4.48284857e-02 6.74178004e-01 -4.35011834e-01 1.46311641e-01
-9.50308740e-01 1.01188195e+00 -2.29778513e-02 3.29865843e-01
-8.59226942e-01 -5.39174616e-01 -7.26125896e-01 -6.47083670e-02
3.66489053e-01 -1.15369630e+00 1.26007926e+00 -1.08973718e+00
-1.84502435e+00 -5.71787395e-02 2.68515974e-01 -5.80659688e-01
4.62215930e-01 -4.23218340e-01 4.30789381e-01 3.13678384e-01
2.36227721e-01 6.31199718e-01 8.67654979e-01 -1.43392730e+00
-6.40558124e-01 -3.49861234e-02 2.96564966e-01 7.37546623e-01
-4.71446663e-01 -5.63200474e-01 2.76093602e-01 -5.05117536e-01
-1.78052038e-01 -1.27841270e+00 -9.93719757e-01 8.19414631e-02
2.45345309e-01 3.35712075e-01 7.85830915e-01 1.85693592e-01
5.65621972e-01 -1.77325082e+00 7.49348402e-01 -6.55502900e-02
1.63720310e-01 2.57296085e-01 -1.80133000e-01 8.84411037e-01
2.25852340e-01 3.41768079e-02 -8.23092386e-02 -3.52719128e-01
-4.48872715e-01 5.51265359e-01 -3.48648429e-01 4.04647887e-01
1.98045209e-01 1.78287178e-01 -1.25551438e+00 1.81835636e-01
1.68775618e-01 3.33623946e-01 -7.40313113e-01 7.61109069e-02
-2.07807690e-01 7.64461398e-01 -4.57221657e-01 6.80400357e-02
6.20540917e-01 -4.20364700e-02 5.24094284e-01 2.52824336e-01
-5.94688356e-01 -3.10437351e-01 -1.24061453e+00 1.49981129e+00
-8.35104048e-01 6.24539912e-01 1.00025833e+00 -8.45609426e-01
1.05563068e+00 1.98512465e-01 5.71395457e-01 -3.51929516e-02
3.07008773e-01 2.87195444e-01 2.08230451e-01 -6.41295135e-01
7.28724122e-01 -1.91036925e-01 3.74719910e-02 1.95500746e-01
-2.39553973e-01 -1.09481668e+00 1.36784375e-01 4.18881997e-02
1.42262661e+00 -1.40038222e-01 -1.78001955e-01 -4.69212621e-01
6.98001608e-02 4.88448769e-01 5.11804879e-01 8.37568223e-01
-1.92208469e-01 3.28113735e-01 2.42948711e-01 -6.19651139e-01
-8.59628201e-01 -5.77299654e-01 3.45083088e-01 7.81783462e-01
7.15878606e-01 -3.61331254e-01 -4.04851049e-01 -2.46260628e-01
3.63080293e-01 -7.46527128e-03 -7.24656880e-01 -1.41004473e-01
-7.16553271e-01 -4.72016454e-01 1.68578982e-01 3.77492189e-01
7.09811389e-01 -7.21430242e-01 -1.29043233e+00 3.81245166e-01
-5.55644336e-04 -1.23376811e+00 -2.65171975e-01 5.95307350e-01
-8.02229464e-01 -1.01248705e+00 -3.14750880e-01 -7.84622073e-01
5.63858092e-01 9.15376663e-01 4.69466031e-01 2.18413845e-01
-3.69183540e-01 6.05747640e-01 -6.49290919e-01 -3.83913517e-01
-2.03253150e-01 -5.27919270e-02 5.32985806e-01 -2.40953550e-01
-6.33861601e-01 -7.13596106e-01 -5.35497129e-01 5.73762596e-01
-5.99963188e-01 -1.51958898e-01 3.97717118e-01 9.52589929e-01
9.89073142e-02 5.60904086e-01 1.04491338e-01 -4.87862259e-01
5.42041898e-01 -3.64770323e-01 -7.34517813e-01 -3.34207326e-01
-2.56909756e-03 -9.59843546e-02 1.18719292e+00 -4.04589921e-01
-7.32075810e-01 3.71734291e-01 6.12706542e-01 -6.85683250e-01
-7.52396137e-02 2.91380435e-01 4.98645157e-01 -4.60760087e-01
7.24182010e-01 -4.04405534e-01 3.06161642e-01 9.80362967e-02
2.79560089e-01 3.76302540e-01 1.86959341e-01 -1.77796975e-01
9.67847466e-01 8.54236126e-01 6.82009339e-01 -9.92997825e-01
-7.31784463e-01 -4.21980888e-01 -9.32525992e-01 -5.51372290e-01
1.09363008e+00 -1.04608917e+00 -1.02988183e+00 5.14197946e-01
-1.16337335e+00 -1.17505419e+00 -1.02071963e-01 8.16035151e-01
-5.09320736e-01 -2.43962742e-02 -6.03621244e-01 -9.15435553e-01
3.07531599e-02 -1.18335116e+00 8.13416958e-01 2.31125608e-01
1.65136039e-01 -1.24171543e+00 3.42408508e-01 -1.98290199e-02
7.19535887e-01 4.43841904e-01 1.81603670e-01 8.15048292e-02
-5.58264315e-01 6.97717965e-02 6.29119277e-02 1.26057073e-01
3.22714686e-01 1.65113628e-01 -4.41143066e-01 -4.51803982e-01
-7.79295713e-02 -5.53461313e-01 9.98801351e-01 6.11021519e-01
4.41742927e-01 -2.13039935e-01 -3.70952189e-01 6.37905777e-01
1.52334118e+00 -1.11841984e-01 -2.37527370e-01 4.84247833e-01
9.03621793e-01 9.86022294e-01 7.65655398e-01 6.50664210e-01
4.82041925e-01 5.07957578e-01 1.26581502e+00 -2.22936515e-02
-6.29809499e-02 1.98629722e-01 3.93094093e-01 7.60081887e-01
-4.39002752e-01 -4.37996536e-01 -9.22712386e-01 5.81126690e-01
-2.26046610e+00 -8.30624580e-01 -5.59775829e-01 1.88780594e+00
1.09342113e-01 -2.20464244e-01 -2.95158088e-01 -3.75064284e-01
6.00071132e-01 3.73513848e-02 -3.30951065e-01 -5.88942826e-01
-7.67056830e-04 -1.62893012e-01 1.20936894e+00 7.86210835e-01
-1.31213486e+00 1.08371854e+00 6.40411091e+00 7.43431747e-02
-1.26927805e+00 -2.81035423e-01 -8.38724971e-02 -1.48728848e-01
3.37132663e-01 -2.46544071e-02 -5.98683417e-01 -4.76960763e-02
7.61389852e-01 3.18800956e-01 7.55086362e-01 7.47406423e-01
5.88597775e-01 -3.61808568e-01 -4.70815182e-01 4.77810949e-01
-1.05846331e-01 -1.30929697e+00 -1.25643834e-01 1.40370265e-01
8.31217110e-01 3.36828738e-01 3.35076228e-02 -8.97988528e-02
9.67375040e-01 -8.10969412e-01 8.22242081e-01 4.05180603e-01
3.14062238e-01 -5.68345189e-01 8.50494981e-01 4.76435184e-01
-1.52522480e+00 -7.07088053e-01 -6.93878710e-01 -8.78423512e-01
7.79785812e-02 3.58906776e-01 -1.27315152e+00 7.90711403e-01
9.45149541e-01 9.51259732e-01 -2.15249494e-01 9.19208109e-01
-1.64968535e-01 1.55337155e-01 -3.87419581e-01 -2.43989170e-01
4.96046841e-01 -4.41324115e-01 7.31054842e-01 8.26600969e-01
5.31049252e-01 -1.40683856e-02 7.35293448e-01 1.78918272e-01
2.97031909e-01 -2.19562352e-01 -1.09860539e+00 2.81312525e-01
2.69153118e-01 1.51603270e+00 -6.91523433e-01 1.04764802e-02
1.91493612e-02 7.58748591e-01 6.28073514e-01 2.28358388e-01
-4.62765217e-01 -6.19402945e-01 1.01042461e+00 -7.03301653e-03
6.01915061e-01 -9.12440002e-01 -1.97427034e-01 -7.07939506e-01
-4.14899468e-01 -3.69490921e-01 -1.83132529e-01 -8.82997334e-01
-1.20141697e+00 6.78414643e-01 -3.06250453e-01 -1.62628281e+00
-1.01974040e-01 -8.45020354e-01 -7.29446352e-01 6.35987997e-01
-1.60473740e+00 -1.00480473e+00 -6.43081069e-01 2.49657467e-01
4.68513876e-01 -7.77752697e-02 7.71822333e-01 -3.10814023e-01
-3.40156287e-01 -2.86620051e-01 1.15721293e-01 -1.91178098e-01
1.18578136e+00 -1.56425476e+00 6.20242395e-03 6.80098951e-01
-5.88006079e-01 3.44447404e-01 1.09031641e+00 -8.06275249e-01
-1.55978549e+00 -1.43134320e+00 1.86917201e-01 -3.30262989e-01
1.14943242e+00 -1.64777443e-01 -5.51633835e-01 4.81391191e-01
4.81331259e-01 9.67885405e-02 3.33496153e-01 -5.26232608e-02
-1.58510134e-02 -2.88751036e-01 -7.00276077e-01 6.46544635e-01
1.03380501e+00 2.22913131e-01 -1.64053872e-01 6.46958947e-01
9.38123107e-01 -4.95999277e-01 -7.04098880e-01 3.91458303e-01
2.72128135e-01 -1.07553411e+00 5.11110961e-01 -1.65558174e-01
4.26207423e-01 -6.84303880e-01 5.35934530e-02 -2.31420732e+00
-4.34092879e-01 -7.95320928e-01 7.46645272e-01 5.38123965e-01
2.53057748e-01 -7.47190773e-01 4.43424255e-01 -3.06072056e-01
-6.62365317e-01 -4.09128547e-01 -6.16646290e-01 -7.05624521e-01
1.36398450e-01 8.82088020e-02 1.05186082e-01 9.44830596e-01
1.27106458e-01 3.12316507e-01 -5.23138702e-01 8.84640515e-01
3.91019344e-01 6.87637404e-02 1.07467198e+00 -1.19778669e+00
-5.13730764e-01 -2.38041833e-01 -1.59594819e-01 -8.56820703e-01
9.01542976e-02 -5.65970421e-01 6.27573967e-01 -1.71034932e+00
-8.00471246e-01 -5.80882072e-01 6.44051731e-01 4.86037344e-01
1.20199537e-02 8.58884081e-02 4.43823189e-01 3.34830403e-01
-5.92837989e-01 6.63353682e-01 1.80863047e+00 8.11164528e-02
-4.20600235e-01 -9.47952121e-02 -1.89906284e-01 7.29255199e-01
9.97620165e-01 -1.35338306e-01 -4.38731730e-01 -1.18221712e+00
3.21478993e-01 2.09568620e-01 2.30683982e-01 -1.55393314e+00
6.75055504e-01 -2.52387881e-01 2.62215167e-01 8.02682638e-02
5.33324659e-01 -9.34918702e-01 -1.58928409e-01 8.95991027e-01
-1.17368832e-01 3.73991817e-01 3.08228999e-01 9.37512994e-01
-1.91588059e-01 -2.89748088e-02 7.67970979e-01 -5.71659625e-01
-3.49141061e-01 1.06145427e-01 -1.03388095e+00 -2.02379867e-01
1.11452150e+00 8.62998888e-03 -6.00894094e-01 -4.23185676e-01
-5.27236938e-01 9.32801068e-01 6.86985850e-01 1.75782844e-01
4.80297297e-01 -6.26183987e-01 -4.45049107e-01 6.52344674e-02
-2.40424871e-01 3.93209338e-01 3.37975651e-01 8.98554802e-01
-1.12238395e+00 -3.77694070e-02 -4.78552610e-01 -5.42051256e-01
-1.00277138e+00 1.98696613e-01 7.45724618e-01 -2.26469189e-01
-4.80635047e-01 1.06771576e+00 1.73459828e-01 -6.77412271e-01
-1.08769648e-01 -4.72474694e-01 -3.23761344e-01 -4.65956405e-02
2.79516578e-01 2.10143477e-01 -1.31412283e-01 -3.96596491e-01
-2.33258363e-02 1.29015052e+00 4.09542888e-01 -1.81110546e-01
1.47976017e+00 -4.06612009e-01 -1.40944287e-01 4.06001925e-01
4.77092147e-01 2.17735786e-02 -1.72213638e+00 4.24402416e-01
-5.97273469e-01 -4.44887310e-01 1.61309883e-01 -4.23426658e-01
-1.04133725e+00 8.53001952e-01 8.56952146e-02 5.26531696e-01
9.24571157e-01 -6.07372761e-01 5.44585466e-01 7.60932088e-01
5.61729789e-01 -1.01630425e+00 5.93017161e-01 1.21390748e+00
7.72242606e-01 -1.10684299e+00 -1.93786353e-01 -4.50893223e-01
-9.47055101e-01 1.54179358e+00 7.41334081e-01 -7.54753172e-01
6.24527395e-01 4.62203860e-01 4.99398679e-01 -3.20534743e-02
-1.11825013e+00 -3.78822118e-01 -6.99829102e-01 8.69240224e-01
-1.53859973e-01 2.93697175e-02 3.93625617e-01 -1.49080336e-01
-4.58650440e-01 -4.33153570e-01 1.25420201e+00 9.97554123e-01
-1.01642501e+00 -9.27848935e-01 -5.35045266e-01 1.88677654e-01
1.56949908e-01 3.57769877e-01 -2.91520864e-01 6.60161793e-01
4.82381880e-01 1.19202363e+00 2.67262608e-01 -5.60855746e-01
4.37106013e-01 -6.53039932e-01 3.01836848e-01 -8.37018788e-01
-1.00365281e+00 -3.24771516e-02 -4.35918868e-02 -5.97649634e-01
-5.89880526e-01 -6.39622271e-01 -1.38354588e+00 -1.30629510e-01
-1.45323575e-01 3.62939715e-01 5.53047955e-01 6.86963618e-01
3.39319438e-01 9.99973834e-01 8.98150444e-01 -1.50234938e+00
-3.91910791e-01 -7.69960940e-01 -7.13082552e-01 -6.70882687e-02
7.58584321e-01 -1.02695715e+00 -7.70343602e-01 -5.55183664e-02]
|
[4.536502838134766, 1.4710510969161987]
|
5fb977fe-52ba-488d-9e97-15792669d126
|
convergence-of-gradient-descent-with-linearly
|
2302.01463
| null |
https://arxiv.org/abs/2302.01463v2
|
https://arxiv.org/pdf/2302.01463v2.pdf
|
Gradient Descent with Linearly Correlated Noise: Theory and Applications to Differential Privacy
|
We study gradient descent under linearly correlated noise. Our work is motivated by recent practical methods for optimization with differential privacy (DP), such as DP-FTRL, which achieve strong performance in settings where privacy amplification techniques are infeasible (such as in federated learning). These methods inject privacy noise through a matrix factorization mechanism, making the noise linearly correlated over iterations. We propose a simplified setting that distills key facets of these methods and isolates the impact of linearly correlated noise. We analyze the behavior of gradient descent in this setting, for both convex and non-convex functions. Our analysis is demonstrably tighter than prior work and recovers multiple important special cases exactly (including anticorrelated perturbed gradient descent). We use our results to develop new, effective matrix factorizations for differentially private optimization, and highlight the benefits of these factorizations theoretically and empirically.
|
['Brendan Mcmahan', 'Keith Rush', 'Zachary Charles', 'Ryan McKenna', 'Anastasia Koloskova']
|
2023-02-02
| null | null | null | null |
['stochastic-optimization']
|
['methodology']
|
[-1.83801521e-02 -7.02637881e-02 -2.68635958e-01 -1.82809904e-01
-8.81005526e-01 -1.31135929e+00 1.79310262e-01 2.62253154e-02
-4.26179886e-01 7.34979153e-01 6.48625851e-01 -7.72022009e-01
-2.02875242e-01 -6.05384886e-01 -9.16713715e-01 -8.28554690e-01
-4.46520030e-01 -5.85976504e-02 -6.15670681e-01 -3.28455389e-01
-1.00503102e-01 4.82215762e-01 -8.03574383e-01 2.27396190e-01
7.88569331e-01 5.84005237e-01 -7.51033485e-01 8.51932466e-01
2.67605633e-01 6.79775178e-01 -3.42518389e-01 -9.86993670e-01
1.18951881e+00 -3.48327041e-01 -7.04623282e-01 -1.70373932e-01
5.60561240e-01 -5.98055124e-01 -7.56318808e-01 1.18210638e+00
5.83669543e-01 1.21702947e-01 1.77415144e-02 -1.33541501e+00
-5.34829080e-01 1.02880716e+00 -6.66559696e-01 9.14021879e-02
1.96585909e-01 2.03744069e-01 1.21595061e+00 -5.16427279e-01
5.99279463e-01 1.11005533e+00 1.06732500e+00 6.37695909e-01
-1.40719080e+00 -6.66502059e-01 1.06095016e-01 -4.10800308e-01
-1.08056331e+00 -6.58205807e-01 2.17965588e-01 -2.33080834e-01
4.72647697e-01 8.63400161e-01 3.22303593e-01 1.03156900e+00
-2.77731031e-01 1.08413744e+00 1.07900488e+00 -3.08788270e-01
1.88168928e-01 2.62808502e-01 2.33197331e-01 6.22043014e-01
7.11734116e-01 3.39224815e-01 -6.38012886e-01 -1.17740691e+00
2.27426365e-01 3.43945712e-01 -8.97483408e-01 -5.30306041e-01
-8.97122920e-01 8.29624534e-01 1.82768121e-01 8.24812949e-02
-1.12389572e-01 2.97096372e-01 5.34440875e-01 7.61250675e-01
4.44067121e-01 3.72696698e-01 -6.23289526e-01 -4.45736982e-02
-7.51985312e-01 4.44524974e-01 1.43004608e+00 1.05500734e+00
9.04909670e-01 -3.58841002e-01 -5.18502474e-01 3.40040147e-01
-4.70902734e-02 2.73490906e-01 8.75579640e-02 -1.06969821e+00
7.35992074e-01 5.37276343e-02 4.73241240e-01 -1.10105300e+00
-3.80217731e-02 -5.75921714e-01 -7.20090151e-01 -1.22095766e-02
5.94591200e-01 -8.74843419e-01 -2.85374284e-01 1.96210384e+00
5.03290892e-01 1.19747207e-01 1.81111574e-01 8.28647733e-01
-4.80399169e-02 2.37611741e-01 -2.90802300e-01 -2.45324641e-01
9.41029251e-01 -9.15908396e-01 -4.77885127e-01 4.67484295e-02
9.93807912e-01 -2.45152935e-01 9.62303817e-01 2.92564601e-01
-8.23233187e-01 5.58044672e-01 -8.50168824e-01 -2.20046893e-01
-1.55828409e-02 -3.13930780e-01 1.20396054e+00 1.32048202e+00
-1.28569734e+00 7.82774806e-01 -8.45389485e-01 -1.07440069e-01
7.07500219e-01 4.55215812e-01 -6.06601894e-01 -1.28119722e-01
-9.65819895e-01 -4.88527026e-03 -9.61153060e-02 -4.57745157e-02
-7.27209747e-01 -1.12526727e+00 -5.83566904e-01 2.15849280e-01
3.95312160e-01 -1.02750480e+00 1.18937635e+00 -8.89260232e-01
-1.28720081e+00 8.22769940e-01 -1.52628943e-01 -8.55118215e-01
1.18604374e+00 -3.30456495e-01 1.16681628e-01 -9.14538279e-02
-1.80647656e-01 -3.90505552e-01 6.16234183e-01 -1.03513408e+00
-6.63939416e-01 -8.11031342e-01 4.51907516e-01 1.74398571e-01
-9.30023015e-01 7.34947324e-02 -2.66865581e-01 -4.96943712e-01
-3.54181379e-01 -9.71451998e-01 -6.96748316e-01 2.34554484e-01
-5.05637228e-01 4.29449946e-01 7.42760599e-01 -5.09459555e-01
1.16947842e+00 -2.24412417e+00 -7.18607232e-02 5.17673790e-01
8.58307600e-01 1.39287323e-01 2.47099064e-02 6.56550348e-01
2.54370898e-01 4.76778567e-01 -2.67910272e-01 -7.38967896e-01
3.39783877e-01 1.51237369e-01 -4.96920079e-01 1.10334694e+00
-5.65512359e-01 9.73915577e-01 -9.91331279e-01 1.61025941e-01
-3.63498062e-01 1.30424038e-01 -1.01043737e+00 7.65637979e-02
-8.04689601e-02 4.37103450e-01 -5.81917822e-01 5.58468997e-01
1.02398443e+00 -1.58509001e-01 4.29934084e-01 -4.78987880e-02
6.37753084e-02 4.93393764e-02 -1.02575922e+00 1.59229839e+00
-3.70630831e-01 3.78562152e-01 9.83934700e-01 -6.85525358e-01
1.34558409e-01 2.02892259e-01 6.27063215e-01 2.38779802e-02
1.76189646e-01 6.96372613e-02 -4.86993492e-01 -4.32480633e-01
5.19829512e-01 -1.38591468e-01 -7.74547318e-03 8.77324581e-01
-1.78203285e-01 4.45536077e-01 -2.43876696e-01 6.31535947e-01
1.49622142e+00 -4.55701351e-01 3.67989689e-02 -4.95251477e-01
1.06430471e-01 -4.17839050e-01 8.29715490e-01 1.33903253e+00
-3.47200274e-01 4.32414085e-01 6.80236280e-01 -2.75364161e-01
-8.27934206e-01 -6.74680829e-01 6.06843345e-02 1.29243708e+00
4.47519161e-02 -9.15075660e-01 -6.74301744e-01 -1.01538086e+00
7.72737503e-01 2.31364623e-01 -7.05457330e-01 -4.14989516e-02
-8.02129656e-02 -1.36503386e+00 9.16833818e-01 5.91351762e-02
2.39142850e-01 2.19420210e-01 7.39910901e-02 -1.26611367e-01
1.02002881e-01 -8.98362339e-01 -9.80465710e-01 2.56272048e-01
-8.27494740e-01 -1.05634129e+00 -3.52887183e-01 -1.32314831e-01
6.47340238e-01 4.99499172e-01 8.40652704e-01 1.17259085e-01
-1.19202033e-01 9.15104091e-01 -8.81192312e-02 -1.34694621e-01
-3.71670693e-01 2.46076565e-02 1.94634229e-01 4.75157738e-01
2.65162885e-01 -8.79609287e-01 -8.24186206e-01 8.94604027e-02
-1.12072265e+00 -5.32155156e-01 1.47004440e-01 1.02690327e+00
4.27592158e-01 -2.34467253e-01 -4.94990200e-02 -1.56292355e+00
1.07698023e+00 -9.32123005e-01 -7.52100646e-01 3.00827026e-01
-6.61543727e-01 2.49238506e-01 8.53291452e-01 -2.41923943e-01
-8.32375824e-01 -5.72518399e-03 8.67514536e-02 -5.70268095e-01
4.83680338e-01 1.58289745e-01 -1.47416383e-01 -9.45859492e-01
1.05461609e+00 3.92058454e-02 7.97208250e-02 -5.54836154e-01
7.20087647e-01 6.13314927e-01 4.59003299e-01 -1.17061424e+00
8.54816973e-01 1.02323925e+00 1.00363053e-01 -3.91923398e-01
-4.98845637e-01 -4.10806358e-01 1.06723651e-01 5.41654825e-01
-6.06221110e-02 -1.00117421e+00 -1.20841467e+00 3.41466248e-01
-7.05590010e-01 -3.57856780e-01 -5.90323567e-01 3.27029735e-01
-3.42772096e-01 9.17820334e-01 -8.71754229e-01 -9.37808990e-01
-5.36566615e-01 -8.93929124e-01 6.62255526e-01 -2.81545036e-02
2.59657770e-01 -1.20396972e+00 3.41269761e-01 2.10511610e-01
5.66442609e-01 4.01828408e-01 4.70865309e-01 -8.06539714e-01
-6.69396043e-01 -2.15919271e-01 4.61471118e-02 4.52900797e-01
-5.58483563e-02 -2.10931793e-01 -1.00870609e+00 -7.80880451e-01
3.31586480e-01 -3.44661653e-01 1.04141760e+00 -9.86940116e-02
1.35386395e+00 -1.13137758e+00 -3.84814888e-01 1.51156700e+00
1.46833456e+00 -7.53199279e-01 3.05090457e-01 1.82871167e-02
5.90549350e-01 1.54743671e-01 3.95016521e-02 1.08075976e+00
2.16690570e-01 2.20364496e-01 3.63146245e-01 -3.34387794e-02
4.39157844e-01 -3.78765792e-01 3.17814708e-01 3.12173545e-01
2.27467626e-01 -3.25641602e-01 -4.21394169e-01 4.69013572e-01
-2.00504875e+00 -8.66847098e-01 1.17689827e-02 2.41765976e+00
1.06649148e+00 -5.25335252e-01 2.43443847e-01 -3.35583597e-01
5.61995924e-01 1.72659770e-01 -7.20346510e-01 -3.88161719e-01
-5.40296853e-01 2.62759268e-01 1.40620005e+00 5.56965828e-01
-1.09181786e+00 5.84071815e-01 7.25183725e+00 4.89396274e-01
-7.12893963e-01 4.19425994e-01 8.80571008e-01 -6.17904127e-01
-8.77818227e-01 2.37000600e-01 -6.21428251e-01 3.44652712e-01
7.75090694e-01 -6.70736015e-01 1.01746666e+00 1.05528927e+00
-1.68199763e-01 3.53084683e-01 -1.27868187e+00 1.16806483e+00
-3.16812515e-01 -1.38297641e+00 -2.94640243e-01 3.78702164e-01
1.16112649e+00 3.46973002e-01 4.44401234e-01 4.45993766e-02
1.14073598e+00 -8.57955337e-01 3.39257121e-01 1.37546629e-01
6.54624522e-01 -7.79217601e-01 2.30998531e-01 1.43952176e-01
-4.69199747e-01 -5.95617056e-01 -4.46777344e-01 1.48863737e-02
-1.89666271e-01 9.46520567e-01 -4.32229757e-01 6.46572948e-01
6.51318967e-01 4.52225327e-01 -1.80795401e-01 9.87226725e-01
1.63246077e-02 8.18633497e-01 -8.66935492e-01 2.17522368e-01
1.14574924e-01 -4.49113905e-01 9.43041265e-01 1.32444942e+00
2.04640314e-01 5.13536893e-02 -7.82654993e-03 7.21252739e-01
-8.06602359e-01 3.28600377e-01 -6.93350673e-01 -1.46191269e-01
5.90753853e-01 1.21612918e+00 1.64486811e-01 1.53239099e-02
-4.65481669e-01 1.33929706e+00 5.42689323e-01 5.90952396e-01
-4.80480433e-01 -3.19343895e-01 1.57698107e+00 -1.34730116e-01
2.99291402e-01 -3.09382051e-01 -3.61022592e-01 -1.86375284e+00
3.54978442e-01 -1.10608089e+00 7.24463701e-01 4.17506278e-01
-1.67238653e+00 5.76414652e-02 -4.96665955e-01 -6.85193419e-01
-1.67589281e-02 -3.62981915e-01 -4.56053734e-01 8.68974090e-01
-1.34486985e+00 -7.90911853e-01 1.91308290e-01 9.91457164e-01
-6.10222340e-01 5.11876903e-02 7.67095327e-01 4.35234457e-01
-5.25432110e-01 1.49902725e+00 1.06347871e+00 2.15181112e-02
6.42531157e-01 -1.28607357e+00 4.45935667e-01 1.19857836e+00
2.31598482e-01 1.17379487e+00 6.16443336e-01 -4.01998281e-01
-2.29671049e+00 -1.07183301e+00 3.06207716e-01 -5.47720492e-01
8.08165431e-01 -7.31552064e-01 -4.41205353e-01 1.01617706e+00
-1.75636753e-01 4.88964230e-01 1.16792619e+00 2.90972918e-01
-8.12363267e-01 -3.21107358e-01 -1.78067970e+00 6.63264811e-01
1.43758559e+00 -7.97379136e-01 3.34784865e-01 7.90770411e-01
9.62696195e-01 -6.35637224e-01 -6.83983088e-01 -3.08787245e-02
5.50519049e-01 -1.12776017e+00 7.66016066e-01 -1.14198697e+00
-3.43316376e-01 -7.75694707e-03 -3.51290643e-01 -1.11715555e+00
-1.14264540e-01 -1.91650379e+00 -5.76419950e-01 9.29687858e-01
3.04057121e-01 -1.12382972e+00 1.10158587e+00 1.18808103e+00
4.31885362e-01 -5.99500835e-01 -8.14076304e-01 -7.61514783e-01
2.14396358e-01 -1.56849697e-01 8.38292778e-01 1.27602923e+00
1.19665399e-01 -4.49123263e-01 -6.90225422e-01 3.62380505e-01
8.12336445e-01 -3.91354710e-02 1.05620539e+00 -6.36538982e-01
-9.28911746e-01 -1.34119764e-01 -3.69670242e-01 -1.16738093e+00
1.60973147e-01 -1.13850522e+00 -4.96130973e-01 -6.21692300e-01
1.81595415e-01 -6.72579706e-01 -2.75677443e-01 6.32704735e-01
-2.98778832e-01 5.23730703e-02 9.00338069e-02 1.39114320e-01
-4.64103490e-01 4.32615280e-01 7.85692811e-01 3.36353071e-02
-1.99150175e-01 2.71165013e-01 -1.48734772e+00 1.32405311e-01
6.14558220e-01 -3.84341687e-01 -3.55910242e-01 -4.95387912e-01
6.08792484e-01 -1.92370698e-01 2.55848706e-01 -5.30241013e-01
3.76856774e-01 -1.49928927e-01 -5.04468977e-02 2.77988732e-01
-1.12694547e-01 -8.82996202e-01 2.81771183e-01 3.09976280e-01
-4.74349618e-01 3.51390019e-02 -7.73842484e-02 9.82042253e-01
2.71643460e-01 8.47597793e-02 5.33643663e-01 -1.96871161e-01
1.32213896e-02 9.24418688e-01 -6.85590412e-03 5.59273660e-01
9.51039433e-01 2.36626327e-01 -4.31063920e-01 -7.49614775e-01
-6.10332966e-01 5.03402472e-01 8.43970597e-01 -2.33869344e-01
1.56621099e-01 -9.92883921e-01 -7.69723713e-01 1.92540482e-01
-1.00922674e-01 -2.25214541e-01 1.41090780e-01 9.79885101e-01
-3.99438471e-01 -8.14072415e-02 4.11814213e-01 -4.45216708e-02
-1.12573791e+00 6.80350661e-01 6.09340847e-01 -2.16323584e-01
-5.69242477e-01 1.13314438e+00 1.22914061e-01 -4.90891099e-01
4.62964654e-01 -9.28573683e-02 9.58988309e-01 -4.53558713e-01
7.01909423e-01 4.15788174e-01 1.02995865e-01 -9.82238799e-02
-8.96667764e-02 -2.16892257e-01 -3.71175528e-01 -1.21245958e-01
1.23214257e+00 -2.75651455e-01 -5.24990141e-01 -2.99314290e-01
1.73192084e+00 8.18599582e-01 -1.26513028e+00 -5.00108063e-01
-2.49709472e-01 -1.01320255e+00 -1.28264874e-01 -4.89004463e-01
-1.22856379e+00 4.12133873e-01 2.71930695e-01 3.69930528e-02
1.09385407e+00 -3.55286807e-01 9.13257182e-01 6.50358856e-01
5.66730022e-01 -6.81420982e-01 -7.24752128e-01 2.42517024e-01
2.40519688e-01 -9.13405478e-01 -1.55673530e-02 -2.13269562e-01
-2.61157781e-01 8.72199714e-01 5.81309386e-02 -4.21405993e-02
7.30370939e-01 6.91686034e-01 -1.66142538e-01 1.22060940e-01
-7.20030546e-01 2.93182671e-01 -4.46555495e-01 5.20153463e-01
2.85335141e-03 1.74329430e-01 -5.15280247e-01 8.57766926e-01
-1.64105803e-01 -9.01926830e-02 5.73069215e-01 1.12192988e+00
1.11631230e-01 -1.39850056e+00 -2.30265856e-01 4.76934910e-01
-1.11319077e+00 -2.65141487e-01 -5.78540146e-01 2.41022557e-01
-2.64424652e-01 8.21000338e-01 -4.23879653e-01 -4.73105520e-01
-1.48386389e-01 -1.73541144e-01 3.55917960e-01 -2.40616545e-01
-1.19197357e+00 -5.44430077e-01 4.02715094e-02 -9.65794921e-01
6.77831620e-02 -5.85282505e-01 -6.99754953e-01 -9.64472115e-01
-3.47986072e-01 5.55154443e-01 5.97474575e-01 4.29522753e-01
1.02088070e+00 -3.74223500e-01 1.14826834e+00 -2.19254494e-01
-1.40003693e+00 -1.89793855e-01 -8.15474153e-01 5.16253829e-01
8.03076863e-01 2.16458693e-01 -8.67154896e-01 -4.86203343e-01]
|
[5.848475933074951, 6.556402683258057]
|
8c6548f2-ccd4-40b2-8913-cf71b972fde7
|
jigsaw-vit-learning-jigsaw-puzzles-in-vision
|
2207.11971
| null |
https://arxiv.org/abs/2207.11971v2
|
https://arxiv.org/pdf/2207.11971v2.pdf
|
Jigsaw-ViT: Learning Jigsaw Puzzles in Vision Transformer
|
The success of Vision Transformer (ViT) in various computer vision tasks has promoted the ever-increasing prevalence of this convolution-free network. The fact that ViT works on image patches makes it potentially relevant to the problem of jigsaw puzzle solving, which is a classical self-supervised task aiming at reordering shuffled sequential image patches back to their natural form. Despite its simplicity, solving jigsaw puzzle has been demonstrated to be helpful for diverse tasks using Convolutional Neural Networks (CNNs), such as self-supervised feature representation learning, domain generalization, and fine-grained classification. In this paper, we explore solving jigsaw puzzle as a self-supervised auxiliary loss in ViT for image classification, named Jigsaw-ViT. We show two modifications that can make Jigsaw-ViT superior to standard ViT: discarding positional embeddings and masking patches randomly. Yet simple, we find that Jigsaw-ViT is able to improve both in generalization and robustness over the standard ViT, which is usually rather a trade-off. Experimentally, we show that adding the jigsaw puzzle branch provides better generalization than ViT on large-scale image classification on ImageNet. Moreover, the auxiliary task also improves robustness to noisy labels on Animal-10N, Food-101N, and Clothing1M as well as adversarial examples. Our implementation is available at https://yingyichen-cyy.github.io/Jigsaw-ViT/.
|
['Johan A. K. Suykens', 'Qinghua Tao', 'Yahui Liu', 'Xi Shen', 'Yingyi Chen']
|
2022-07-25
| null | null | null | null |
['learning-with-noisy-labels', 'learning-with-noisy-labels']
|
['computer-vision', 'natural-language-processing']
|
[ 2.41644233e-01 -2.94679636e-03 1.94932632e-02 -1.84481010e-01
-4.07520592e-01 -7.60349214e-01 3.82025540e-01 -1.96443662e-01
-4.96476352e-01 6.01431310e-01 -9.85659808e-02 -3.68325889e-01
-2.09395781e-01 -8.32704246e-01 -9.85122144e-01 -7.49412358e-01
-1.41953276e-02 1.43001303e-01 3.01728815e-01 -4.00647074e-01
1.44790426e-01 4.25887674e-01 -1.16062176e+00 4.14531708e-01
6.06423855e-01 1.06132340e+00 4.07878280e-01 6.26413703e-01
2.32407883e-01 7.57530749e-01 -5.84786177e-01 -7.56286919e-01
5.23064673e-01 -3.09266508e-01 -1.11281478e+00 6.72424166e-03
5.27915478e-01 8.11987370e-02 -4.33214009e-01 1.13824224e+00
3.18574786e-01 7.59737985e-03 4.41805929e-01 -1.45229948e+00
-1.22220516e+00 4.25567836e-01 -4.08318281e-01 1.62453398e-01
-7.19861407e-03 5.36091387e-01 1.13440669e+00 -6.21670723e-01
6.92313194e-01 1.02252650e+00 1.05021083e+00 6.23589396e-01
-1.30821586e+00 -6.59241378e-01 -6.80323243e-02 6.07374430e-01
-1.16566873e+00 -1.44149363e-02 8.13671470e-01 -2.93707937e-01
9.01771665e-01 2.48594150e-01 3.15349787e-01 1.39296162e+00
1.33139089e-01 7.41410613e-01 1.18227804e+00 -2.63470203e-01
1.90256804e-01 -2.03555673e-01 9.33064818e-02 8.07715833e-01
4.89035770e-02 2.40888730e-01 -3.55980217e-01 6.62165508e-02
7.80753195e-01 -1.51986852e-01 -3.03560078e-01 -4.67910200e-01
-1.20952988e+00 9.41502810e-01 9.96542454e-01 3.30031335e-01
-1.06382564e-01 3.93531352e-01 5.40534973e-01 6.82639420e-01
2.34161988e-01 8.83053601e-01 -5.90988457e-01 9.29968283e-02
-5.42349041e-01 3.00185621e-01 5.75999260e-01 8.56843472e-01
8.36300373e-01 3.34964599e-03 -4.77475263e-02 9.58630681e-01
-2.46144861e-01 3.24139804e-01 5.46809554e-01 -1.15127218e+00
3.34755599e-01 4.24069911e-01 -3.44237626e-01 -1.05300045e+00
-4.35723990e-01 -3.67496133e-01 -1.24401593e+00 4.96659040e-01
7.16851115e-01 1.49928108e-01 -1.06345844e+00 1.85140336e+00
9.09226574e-03 1.18928947e-01 5.07037081e-02 9.85806465e-01
9.69345391e-01 5.05667746e-01 -1.39797047e-01 4.16628778e-01
1.36640811e+00 -1.35712337e+00 -1.62161380e-01 -3.46218944e-01
4.73185182e-01 -5.87038755e-01 1.45656109e+00 4.85061795e-01
-7.61849046e-01 -5.24322510e-01 -9.96864021e-01 -2.81184256e-01
-5.45763254e-01 -1.71223059e-02 8.35548162e-01 5.56710660e-01
-1.04717779e+00 1.01164854e+00 -5.34835994e-01 -5.11671484e-01
9.69581485e-01 4.19055492e-01 -9.14750993e-01 -2.32584178e-01
-9.90525484e-01 9.10672903e-01 2.78344274e-01 -8.98034871e-03
-8.41978848e-01 -7.21238136e-01 -9.79956567e-01 -1.45082492e-02
3.81947339e-01 -5.98481894e-01 1.08483171e+00 -1.08683395e+00
-1.37765384e+00 1.00699615e+00 1.76723301e-01 -6.49286151e-01
4.17068332e-01 2.13479742e-01 -2.76358407e-02 1.08205922e-01
1.42136186e-01 8.42224717e-01 1.14279258e+00 -1.05761921e+00
-3.05776507e-01 -3.35214168e-01 3.70606691e-01 -1.60823673e-01
-1.56373233e-01 -2.18336940e-01 -2.06512764e-01 -9.19064641e-01
-7.63053223e-02 -9.20176506e-01 -2.75350899e-01 2.17374623e-01
-4.09905136e-01 -3.14963281e-01 7.49110639e-01 -5.40905774e-01
5.41895926e-01 -2.31198168e+00 1.88592955e-01 2.67412253e-02
2.69437879e-01 5.38213670e-01 -7.38284111e-01 2.95549154e-01
-3.37645382e-01 1.30800530e-01 -5.91569245e-01 -3.01046699e-01
5.33108190e-02 5.37951648e-01 -2.62610257e-01 4.57449585e-01
5.56579709e-01 1.38102448e+00 -8.32480490e-01 -6.72090799e-02
6.99418262e-02 2.60184944e-01 -6.05638385e-01 -1.07315481e-02
-3.44984651e-01 4.33090419e-01 2.46127788e-02 7.41914451e-01
8.09381962e-01 -3.88712138e-01 -1.08361408e-01 -2.70275444e-01
2.31547296e-01 -2.11532824e-02 -6.81685567e-01 1.81595659e+00
-5.14879882e-01 8.14159572e-01 -1.80189312e-02 -1.47122920e+00
8.10896635e-01 -1.20392837e-01 2.98510864e-02 -9.02695715e-01
9.47766975e-02 1.74973384e-01 -7.29039460e-02 -4.52019602e-01
3.80497426e-01 -1.71737626e-01 -1.56728670e-01 3.27107161e-01
4.52373207e-01 -1.98085129e-01 1.60040885e-01 3.15781981e-02
1.34580934e+00 1.15160674e-01 1.41626596e-01 -2.68102109e-01
2.75949001e-01 1.22119829e-01 5.55215597e-01 7.88387835e-01
-1.60702720e-01 9.17277038e-01 5.76852798e-01 -6.17102027e-01
-1.17802787e+00 -1.11889565e+00 2.07132343e-02 8.88934314e-01
3.30476791e-01 -2.22717538e-01 -7.01963425e-01 -9.89857376e-01
2.01546341e-01 3.00409436e-01 -9.38594997e-01 -4.06742662e-01
-6.04208708e-01 -6.23963475e-01 8.97036135e-01 4.76486981e-01
9.02428389e-01 -1.44623256e+00 -3.72948945e-01 2.11819634e-02
-1.01883478e-01 -1.12099981e+00 -4.92281795e-01 4.37701017e-01
-5.28764486e-01 -1.40012884e+00 -8.73119891e-01 -1.04049444e+00
6.11165464e-01 3.69003087e-01 1.10327804e+00 2.42194921e-01
-5.34745097e-01 3.20645124e-01 -5.24384260e-01 -8.06498900e-02
-3.85088831e-01 3.16075087e-01 -1.47364840e-01 -7.44688883e-02
2.11830914e-01 -8.68916512e-01 -4.73528534e-01 4.84542549e-01
-1.14736140e+00 -1.29355490e-01 5.01287758e-01 1.24357808e+00
3.09621900e-01 4.81998436e-02 5.40362179e-01 -8.45058799e-01
5.26065171e-01 -3.30466360e-01 -5.38584888e-01 2.08031863e-01
-3.79620790e-01 9.32932645e-02 1.08921504e+00 -6.64816141e-01
-5.07553935e-01 4.96113375e-02 -5.30884862e-01 -5.65796733e-01
-3.32290202e-01 1.61770195e-01 -1.61606163e-01 -5.15656412e-01
7.63331473e-01 2.86796659e-01 1.18822984e-01 -5.45671821e-01
4.83999878e-01 3.96232307e-01 8.24764431e-01 -4.68019009e-01
9.60408032e-01 4.56461400e-01 -6.93286359e-02 -8.56979966e-01
-6.80810332e-01 -2.12077826e-01 -3.72406095e-01 2.40019277e-01
8.34340155e-01 -4.82722968e-01 -1.03259182e+00 7.70789385e-01
-1.20985651e+00 -8.06831956e-01 -5.90420842e-01 3.21374722e-02
-6.94268107e-01 5.45621037e-01 -6.71214104e-01 -1.68165147e-01
-1.79770172e-01 -1.15610898e+00 7.36166000e-01 1.45290524e-01
-6.61227033e-02 -9.17463601e-01 -6.12885170e-02 4.81980592e-01
4.30897832e-01 1.65705562e-01 1.07133973e+00 -4.04606432e-01
-4.91096079e-01 1.27764106e-01 -5.42075098e-01 7.32259572e-01
1.13876179e-01 -4.10534680e-01 -1.03549600e+00 -3.13986838e-01
-6.92703873e-02 -6.27231121e-01 1.25575292e+00 1.15190744e-01
1.38060248e+00 -3.66674513e-01 5.66168725e-02 1.15690351e+00
1.44949448e+00 8.94921366e-03 8.93436611e-01 6.45616531e-01
6.76079929e-01 3.81888419e-01 2.61234939e-01 7.51464441e-02
2.90751696e-01 6.15946114e-01 8.31370115e-01 -2.42327675e-01
-4.23120826e-01 -1.30445763e-01 2.29871109e-01 3.55664164e-01
3.25777344e-02 -3.44315991e-02 -6.39553428e-01 6.23866796e-01
-1.73333335e+00 -9.80544329e-01 1.38045009e-02 1.81841779e+00
7.43490696e-01 -1.12668253e-01 1.59825869e-02 2.98213750e-01
4.50852156e-01 2.61239737e-01 -5.89324892e-01 -5.74654341e-01
-3.71935427e-01 5.79385698e-01 7.50331223e-01 3.26949388e-01
-1.27811015e+00 1.08597529e+00 5.58319569e+00 1.06877255e+00
-1.16475284e+00 3.16250980e-01 6.05288088e-01 3.05212587e-01
-9.74617675e-02 -2.44300619e-01 -3.70318055e-01 4.72431332e-01
2.83851445e-01 2.10162640e-01 7.63121963e-01 8.22605073e-01
-3.35882515e-01 1.03289285e-03 -1.02768576e+00 1.13992000e+00
8.35104138e-02 -1.67850280e+00 -2.23129429e-02 -1.72293827e-01
6.47775888e-01 6.41964227e-02 2.14165390e-01 2.94958860e-01
4.09153908e-01 -1.26033962e+00 6.39391899e-01 -7.53387958e-02
8.84331942e-01 -5.78886211e-01 7.05656767e-01 2.84970969e-01
-8.31013203e-01 -2.47761860e-01 -6.77500367e-01 -7.46998563e-02
-1.81500942e-01 4.51273590e-01 -5.32980978e-01 5.34698486e-01
9.19602573e-01 6.69772625e-01 -6.27877176e-01 1.06563604e+00
-3.31288993e-01 5.18391967e-01 -2.13362381e-01 8.35519135e-02
5.07179260e-01 -1.49218850e-02 4.31484520e-01 1.04910195e+00
9.03033763e-02 -6.16590418e-02 -1.52126834e-01 8.98672462e-01
-2.22565457e-01 -2.69059956e-01 -7.21317053e-01 2.71391924e-02
-3.79699096e-02 1.15415525e+00 -7.69489646e-01 6.92085922e-02
-1.93383381e-01 1.41957963e+00 5.50295889e-01 3.40055346e-01
-8.10635567e-01 -4.99111742e-01 9.82718110e-01 -9.09382179e-02
7.62627959e-01 -3.26921731e-01 -3.65875304e-01 -1.09170771e+00
5.99111430e-02 -1.06764102e+00 2.24194139e-01 -6.95965648e-01
-1.53838408e+00 8.95616353e-01 -2.73063213e-01 -1.06453574e+00
1.23138033e-01 -1.01870275e+00 -5.38725495e-01 5.36164105e-01
-1.60800695e+00 -1.36456430e+00 -3.92356932e-01 8.62399399e-01
4.87336218e-01 -2.63127804e-01 9.08345520e-01 3.12859714e-01
-3.56442630e-01 9.29615259e-01 9.68067199e-02 3.32760990e-01
5.22625625e-01 -1.14649189e+00 7.70790637e-01 7.25847602e-01
4.29891318e-01 2.30819553e-01 5.24348319e-01 -2.48231128e-01
-1.26412535e+00 -1.24623370e+00 6.17854893e-01 -2.63683885e-01
6.90780222e-01 -5.17552972e-01 -8.98525596e-01 5.81189513e-01
1.80739298e-01 2.57033139e-01 2.24559128e-01 -1.72059700e-01
-9.25572395e-01 -2.01555297e-01 -1.27987587e+00 6.58849895e-01
1.31481802e+00 -5.27226985e-01 -5.17119110e-01 3.87887180e-01
7.54806876e-01 -3.17128003e-01 -5.71954846e-01 3.19163293e-01
3.73382628e-01 -9.89288151e-01 1.17254460e+00 -7.06298411e-01
6.99271262e-01 -1.76235974e-01 -2.28291824e-01 -1.63272965e+00
-6.36638939e-01 -6.57245159e-01 3.82443577e-01 8.78524780e-01
2.82605588e-01 -8.93318713e-01 1.02469742e+00 3.07270885e-02
-1.23453803e-01 -6.98558629e-01 -1.22460818e+00 -1.18863404e+00
3.83407652e-01 -5.09592772e-01 5.56685388e-01 9.90802228e-01
-2.68216580e-01 1.83357939e-01 -4.47249651e-01 1.71895802e-01
5.72417617e-01 7.64827207e-02 8.31918597e-01 -9.16344404e-01
-4.31037635e-01 -5.34841835e-01 -8.02768946e-01 -1.06871259e+00
1.83296278e-01 -1.13730085e+00 3.10192555e-02 -1.37817037e+00
-6.74447194e-02 -5.83890796e-01 -1.26740858e-01 9.24782515e-01
7.82257468e-02 9.13748264e-01 5.00812769e-01 -6.75936788e-02
-3.73033047e-01 4.14472967e-01 1.43407798e+00 -5.62060058e-01
1.43219039e-01 4.85845059e-02 -7.39180803e-01 5.79188764e-01
1.06983054e+00 -5.09235203e-01 -2.53570348e-01 -5.47340989e-01
1.98964223e-01 -3.66469741e-01 8.78716826e-01 -9.48503613e-01
1.43083468e-01 2.42384877e-02 5.70242740e-02 -1.51451558e-01
4.46247250e-01 -7.91748047e-01 7.27200285e-02 4.71387237e-01
-1.59421489e-01 -7.58425100e-03 2.75187075e-01 3.35981458e-01
-1.57377034e-01 -4.96915281e-01 8.93173099e-01 -2.42554009e-01
-1.00205815e+00 3.07954758e-01 -1.99437380e-01 1.44851282e-01
8.96509588e-01 -3.78356487e-01 -5.57884216e-01 -2.72915244e-01
-7.06037879e-01 4.70182858e-02 4.87327784e-01 5.00537515e-01
7.13882446e-01 -1.27120852e+00 -6.53836668e-01 4.43420410e-01
1.84018940e-01 -1.99366100e-02 2.76890516e-01 7.54266798e-01
-7.41856694e-01 1.93409473e-01 -5.12698829e-01 -5.94340742e-01
-1.35665023e+00 7.62324750e-01 3.06144953e-01 -3.11619014e-01
-7.86701441e-01 1.23762810e+00 4.33118194e-01 -7.54898965e-01
3.35328132e-01 -5.18657625e-01 1.35894567e-01 -2.70732403e-01
3.72410417e-01 1.65123388e-01 1.44570127e-01 -3.70348454e-01
-3.14689457e-01 6.13101065e-01 -4.35096025e-02 3.47772419e-01
1.56178558e+00 1.75863415e-01 -2.32272699e-01 1.90757886e-02
1.36687291e+00 -2.48297378e-01 -1.28021157e+00 -2.35788882e-01
-1.70368552e-02 -4.25201952e-01 -4.03146982e-01 -8.88354480e-01
-1.33574259e+00 9.59825873e-01 4.09464896e-01 2.39572659e-01
1.20819700e+00 1.60878733e-01 9.56265032e-01 5.60025930e-01
6.41674876e-01 -5.24471700e-01 3.15832615e-01 7.10782290e-01
1.10813642e+00 -1.30384111e+00 -3.01143199e-01 -4.39966440e-01
-6.10061526e-01 9.15286660e-01 5.57315826e-01 -6.06785595e-01
5.47936678e-01 2.47425050e-01 2.12210789e-01 -1.00299321e-01
-4.78324801e-01 -2.99532205e-01 -1.47277256e-02 9.84972417e-01
-2.84528404e-01 -6.56914189e-02 -3.78413014e-02 7.50401914e-01
-3.53535891e-01 8.35448429e-02 4.48731184e-01 5.58793724e-01
-3.74794379e-02 -1.21385920e+00 -3.91018152e-01 3.70344758e-01
-3.70382041e-01 -1.87901840e-01 -3.74410719e-01 8.54938030e-01
4.69080597e-01 6.95274055e-01 -1.18782111e-01 -5.50580561e-01
3.84543359e-01 -2.61906236e-01 7.02454150e-01 -4.30179536e-01
-8.98937583e-01 -5.62383235e-01 -1.01936452e-01 -6.65889621e-01
-2.36432016e-01 -2.65659720e-01 -8.74148846e-01 -4.16296601e-01
-1.52760863e-01 8.03670753e-03 4.77174163e-01 8.44805539e-01
2.91702777e-01 4.36334342e-01 6.26892686e-01 -8.84394646e-01
-6.73365891e-01 -7.44442225e-01 -5.06033838e-01 5.97465217e-01
3.82751882e-01 -5.64664245e-01 -4.32344347e-01 -6.77540749e-02]
|
[9.556662559509277, 1.959618091583252]
|
71b40280-114f-4875-88ea-0bca004ff483
|
graph-guided-deformation-for-point-cloud
|
2112.01840
| null |
https://arxiv.org/abs/2112.01840v1
|
https://arxiv.org/pdf/2112.01840v1.pdf
|
Graph-Guided Deformation for Point Cloud Completion
|
For a long time, the point cloud completion task has been regarded as a pure generation task. After obtaining the global shape code through the encoder, a complete point cloud is generated using the shape priorly learnt by the networks. However, such models are undesirably biased towards prior average objects and inherently limited to fit geometry details. In this paper, we propose a Graph-Guided Deformation Network, which respectively regards the input data and intermediate generation as controlling and supporting points, and models the optimization guided by a graph convolutional network(GCN) for the point cloud completion task. Our key insight is to simulate the least square Laplacian deformation process via mesh deformation methods, which brings adaptivity for modeling variation in geometry details. By this means, we also reduce the gap between the completion task and the mesh deformation algorithms. As far as we know, we are the first to refine the point cloud completion task by mimicing traditional graphics algorithms with GCN-guided deformation. We have conducted extensive experiments on both the simulated indoor dataset ShapeNet, outdoor dataset KITTI, and our self-collected autonomous driving dataset Pandar40. The results show that our method outperforms the existing state-of-the-art algorithms in the 3D point cloud completion task.
|
['Shaojie Shen', 'Liang Heng', 'Lingyun Xu', 'Jieqi Shi']
|
2021-11-11
| null | null | null | null |
['point-cloud-completion']
|
['computer-vision']
|
[-2.22768933e-02 2.74466634e-01 4.01739597e-01 -2.68824250e-01
-4.92580414e-01 -6.32613063e-01 5.67153096e-01 -1.89341187e-01
-1.09835327e-01 3.50614637e-01 -2.82896101e-01 -2.47614831e-01
1.08541034e-01 -1.11128747e+00 -1.30247808e+00 -4.49816763e-01
1.50521062e-02 7.81142890e-01 5.78735247e-02 -3.30525339e-01
3.73874575e-01 8.31982195e-01 -1.40259767e+00 -3.06714684e-01
1.17495799e+00 1.01621497e+00 2.77715772e-01 3.69085401e-01
-2.34671831e-01 9.25741792e-02 -1.13827676e-01 -5.15146792e-01
5.62696815e-01 1.14155091e-01 -4.00583357e-01 2.70441413e-01
6.52153373e-01 -2.51762092e-01 -4.59402174e-01 1.05181026e+00
4.33546215e-01 1.90604001e-01 6.38848543e-01 -1.28751624e+00
-8.84101629e-01 1.96654454e-01 -7.40632415e-01 -6.96371436e-01
1.23776242e-01 3.03465664e-01 5.00434995e-01 -1.31202960e+00
7.85365164e-01 1.42044079e+00 7.80476868e-01 6.24516070e-01
-1.14688671e+00 -7.93228030e-01 2.98050314e-01 -2.64066964e-01
-1.55256939e+00 -1.92455456e-01 1.20482421e+00 -6.64031029e-01
6.53421938e-01 -2.46014521e-02 8.24414670e-01 6.94067240e-01
3.35181534e-01 4.30507243e-01 4.84294593e-01 -1.06977606e-02
3.45408738e-01 -3.27770144e-01 -4.00433302e-01 7.88227916e-01
2.39949554e-01 2.97454298e-01 -1.22045718e-01 -2.60051191e-01
1.29018652e+00 2.00158730e-01 -2.16284350e-01 -8.80535066e-01
-1.28509665e+00 6.42099619e-01 8.86585474e-01 -2.54633814e-01
-3.30149621e-01 5.95833600e-01 -5.91722131e-02 6.90069571e-02
7.02766716e-01 1.32064447e-01 -2.35125452e-01 1.17412239e-01
-7.17452526e-01 7.29818583e-01 6.81472838e-01 1.58287919e+00
1.11873102e+00 1.19290546e-01 4.21272777e-02 4.49583173e-01
6.55652225e-01 7.69331217e-01 -2.34549403e-01 -1.24883544e+00
6.74369812e-01 7.96775341e-01 1.32691309e-01 -1.30103219e+00
-1.37084469e-01 -5.87524891e-01 -1.14012849e+00 6.11064017e-01
2.09246472e-01 -2.03242198e-01 -9.64633286e-01 1.54804480e+00
4.02681679e-01 4.62522447e-01 -1.52747318e-01 9.32550132e-01
7.03064799e-01 6.23151898e-01 -1.47965580e-01 2.76460856e-01
9.65953052e-01 -8.32582593e-01 -3.84845823e-01 -1.21384658e-01
4.88883793e-01 -6.45462632e-01 9.61628377e-01 1.69367060e-01
-1.22900510e+00 -8.00446987e-01 -1.02195811e+00 -3.62535149e-01
-1.90077610e-02 2.26224903e-02 6.06621087e-01 1.07497387e-01
-1.10975647e+00 7.29775190e-01 -1.08365691e+00 -1.48989946e-01
6.20633304e-01 2.22795144e-01 -2.78014362e-01 -1.81825489e-01
-6.16114616e-01 5.48827469e-01 -9.24256817e-03 3.34917992e-01
-7.47474790e-01 -1.03833580e+00 -8.11198771e-01 -5.24403453e-02
3.27798754e-01 -1.29557264e+00 9.21580970e-01 -5.34538329e-01
-1.51629484e+00 8.17571044e-01 -1.49713457e-01 -2.86511689e-01
7.95659363e-01 -7.23972470e-02 2.60145038e-01 -3.10324639e-01
3.83912362e-02 9.27458167e-01 8.33767653e-01 -1.68168819e+00
-3.00331175e-01 -5.45231044e-01 -6.38524711e-04 3.42409819e-01
4.17933822e-01 -6.71941519e-01 -6.97980821e-01 -6.76684439e-01
5.22411287e-01 -1.11907423e+00 -4.80377257e-01 5.11327684e-01
-3.03381979e-01 -1.44505054e-01 9.69053566e-01 -4.34999257e-01
6.13735318e-01 -2.28120494e+00 3.18503797e-01 3.64884019e-01
3.60158652e-01 -6.71362132e-02 -1.00705251e-01 3.13386530e-01
1.23969339e-01 2.83732533e-01 -6.07529938e-01 -8.61566067e-01
2.64022887e-01 4.30192798e-01 -5.73488593e-01 5.19402206e-01
2.31271937e-01 1.24320650e+00 -9.45260406e-01 -2.53928155e-01
3.37163597e-01 7.66895115e-01 -8.57236266e-01 1.93766788e-01
-4.40748006e-01 7.82262146e-01 -6.76382184e-01 5.75543940e-01
1.21091962e+00 -1.32070035e-01 -5.03979325e-01 -2.23972678e-01
-2.27905303e-01 -1.93213195e-01 -1.22311628e+00 2.53884220e+00
-4.02076721e-01 1.64115116e-01 3.31788689e-01 -5.90191305e-01
1.21237719e+00 -5.70987687e-02 4.97856349e-01 -2.50926733e-01
3.08952164e-02 3.15975726e-01 -2.16617748e-01 -1.71629846e-01
5.07658243e-01 1.50465250e-01 8.38445798e-02 9.42888409e-02
-3.63621384e-01 -9.53280807e-01 -3.53905380e-01 2.06121877e-01
7.94848621e-01 7.11183310e-01 -3.20560694e-01 -2.17969805e-01
3.70605707e-01 2.14494988e-01 5.49139380e-01 3.46075743e-01
2.44341075e-01 1.13240147e+00 1.86757848e-01 -4.32145953e-01
-1.28686678e+00 -1.22167778e+00 -1.52422830e-01 3.23822141e-01
5.60817301e-01 -2.75478601e-01 -7.27243721e-01 -4.19690400e-01
3.48910451e-01 5.89140534e-01 -3.51232082e-01 -6.44449964e-02
-8.15440595e-01 -1.64891347e-01 2.39515483e-01 4.89816368e-01
6.83078170e-01 -8.66391718e-01 -2.88490832e-01 1.58652708e-01
1.53530702e-01 -1.13341165e+00 -7.18060553e-01 -2.98218399e-01
-1.08602226e+00 -9.56774950e-01 -6.51047468e-01 -8.59307647e-01
9.88402486e-01 3.38099629e-01 1.14483893e+00 4.06868964e-01
7.64159719e-03 2.66682029e-01 -3.46823670e-02 -4.77532744e-01
-2.40613088e-01 -7.63703894e-04 -1.60812974e-01 5.70277870e-02
-9.92194340e-02 -1.01984882e+00 -8.51714611e-01 2.18571022e-01
-8.29050124e-01 4.56420243e-01 5.59341967e-01 4.05745685e-01
1.13263416e+00 -1.62985817e-01 7.53068030e-02 -7.26900756e-01
3.58388424e-01 -3.14060032e-01 -8.35544407e-01 -1.75072327e-01
-2.84488261e-01 -1.56712309e-02 5.18705785e-01 -2.06671208e-01
-8.44432294e-01 4.91480261e-01 -3.11753482e-01 -9.36937749e-01
-1.56386212e-01 2.04193845e-01 -3.62559617e-01 -4.88748401e-01
4.58712816e-01 2.08123192e-01 1.03257589e-01 -5.82103431e-01
5.66747129e-01 2.39793643e-01 6.09978437e-01 -8.63357484e-01
1.38543797e+00 7.54292428e-01 4.03788298e-01 -6.65444791e-01
-2.84279704e-01 -3.63795161e-02 -8.74367118e-01 -2.44682729e-01
9.42189634e-01 -9.94751811e-01 -7.54819691e-01 6.25410557e-01
-1.58043921e+00 -4.01830465e-01 -4.32741076e-01 1.07945517e-01
-7.90101945e-01 5.12551665e-01 -2.46557146e-01 -5.63310087e-01
-2.90594757e-01 -1.20946932e+00 1.56113613e+00 -4.96404581e-02
2.91353911e-01 -8.58003974e-01 -4.88859192e-02 -5.82346395e-02
3.68669182e-01 8.17362547e-01 6.80110276e-01 1.89178139e-01
-1.17910278e+00 -1.01099946e-01 -2.81346738e-01 5.00182994e-02
1.32073194e-01 9.79326218e-02 -8.37793410e-01 -3.25303614e-01
1.06013410e-01 7.61793703e-02 6.44141912e-01 3.07980001e-01
1.59477627e+00 -5.51273637e-02 -4.39072639e-01 1.30613434e+00
1.47161114e+00 -3.35535780e-02 6.31280661e-01 -3.57089452e-02
1.25703335e+00 3.87231022e-01 4.51907694e-01 2.81476110e-01
7.33667612e-01 6.63762152e-01 9.19827163e-01 -2.12961435e-01
-2.77715296e-01 -6.68929338e-01 -8.37223604e-02 8.22755992e-01
-4.06380236e-01 -5.60133047e-02 -1.03473318e+00 2.64134794e-01
-1.97405279e+00 -4.48976040e-01 -5.21408916e-01 2.05896378e+00
5.24253249e-01 -1.31670274e-02 -3.53949726e-01 -2.48457596e-01
6.26971424e-01 1.22830302e-01 -7.87357688e-01 -3.57422000e-03
1.27910346e-01 1.57051548e-01 5.38912237e-01 7.31615007e-01
-7.96575248e-01 1.03960335e+00 5.03334951e+00 7.90701807e-01
-9.88014340e-01 -7.47481361e-02 4.69571412e-01 2.23059893e-01
-5.93535066e-01 2.44494140e-01 -4.84905958e-01 3.61730427e-01
2.81031102e-01 -2.53608555e-01 5.72846472e-01 9.39139247e-01
3.35634172e-01 3.32404047e-01 -1.18432641e+00 1.25911677e+00
-1.56968504e-01 -1.36994910e+00 3.22937995e-01 2.85177797e-01
9.33773100e-01 1.71402127e-01 6.70304149e-02 5.39873764e-02
2.77902782e-01 -9.05415058e-01 9.32331979e-01 9.45089757e-01
9.40694392e-01 -7.42445230e-01 2.74311185e-01 6.73434317e-01
-1.36384654e+00 4.12184417e-01 -7.86477983e-01 -1.05675682e-01
2.56807357e-01 7.63535500e-01 -3.84236008e-01 8.24273229e-01
3.53703231e-01 9.19200659e-01 -3.06152940e-01 1.09833288e+00
-1.68397784e-01 2.12054804e-01 -3.89898568e-01 3.45557451e-01
1.40623495e-01 -6.46336019e-01 6.47473752e-01 6.44752920e-01
5.91774464e-01 1.81199446e-01 3.28401536e-01 1.52726126e+00
-2.04621822e-01 -2.50728969e-02 -8.75977337e-01 4.15132761e-01
3.80473793e-01 1.18206787e+00 -6.04899466e-01 -2.00287357e-01
-1.96159393e-01 8.31039667e-01 3.25194210e-01 5.14991701e-01
-8.05821478e-01 -2.48447478e-01 7.91878283e-01 3.54796350e-01
8.75294879e-02 -8.40683937e-01 -7.56089926e-01 -1.24774992e+00
1.90966219e-01 -2.21652031e-01 -4.46176082e-01 -9.30922091e-01
-1.29507780e+00 5.22503078e-01 -1.00060888e-01 -1.51165748e+00
9.74949002e-02 -3.98679972e-01 -7.23352849e-01 1.16823375e+00
-1.62116349e+00 -1.32762682e+00 -7.04859078e-01 5.44452369e-01
4.88074720e-01 2.11770251e-01 4.19723243e-01 1.96818545e-01
-1.16113983e-01 1.79211333e-01 -1.02855042e-01 3.76600735e-02
3.63659590e-01 -1.09895420e+00 1.10022306e+00 7.47991145e-01
-2.06429884e-01 5.51528931e-01 4.24167484e-01 -9.51604307e-01
-1.81474888e+00 -1.52403820e+00 4.74252135e-01 -5.82761943e-01
2.54266024e-01 -5.88086069e-01 -1.09337783e+00 6.35956764e-01
-1.90959439e-01 2.76275903e-01 -2.22841322e-01 -5.50740004e-01
2.73563229e-02 -3.51647735e-02 -1.11691022e+00 6.01413488e-01
1.75506425e+00 -2.76954710e-01 -3.92623365e-01 3.55401933e-01
1.05778289e+00 -1.11344731e+00 -7.91434169e-01 5.64629257e-01
2.04560921e-01 -6.09113932e-01 1.08806264e+00 -1.74347684e-01
5.62814891e-01 -5.83127558e-01 -8.55150521e-02 -1.43710637e+00
-4.03740406e-01 -6.72951996e-01 2.60984134e-02 1.15357637e+00
1.36480302e-01 -5.96954823e-01 9.90627587e-01 7.36424625e-01
-7.47406602e-01 -7.41507888e-01 -9.10921395e-01 -6.12100422e-01
3.97412777e-01 -6.56114638e-01 1.19832754e+00 7.71895707e-01
-6.65689588e-01 -7.77506009e-02 -1.65715218e-02 4.09803420e-01
9.74396229e-01 2.57366538e-01 1.23428929e+00 -1.31786644e+00
-4.32822220e-02 -3.52216929e-01 -3.12322706e-01 -1.58845878e+00
1.85358569e-01 -1.04819226e+00 1.33562386e-01 -1.60909212e+00
-3.39095294e-01 -9.20445561e-01 4.54527587e-01 2.40871727e-01
-6.53765425e-02 1.21398322e-01 2.03117594e-01 3.47081482e-01
-3.65686789e-02 8.73545587e-01 1.94223726e+00 -1.28593177e-01
-3.01346898e-01 7.65363127e-02 -5.51809192e-01 7.40113854e-01
6.18267477e-01 -3.14060450e-01 -5.98510146e-01 -9.91392732e-01
4.31303948e-01 1.08565196e-01 6.39143586e-01 -1.07467926e+00
4.23986584e-01 -2.71757931e-01 1.98468432e-01 -8.73631537e-01
3.46536666e-01 -1.02371871e+00 4.81591910e-01 3.03562522e-01
1.51766408e-02 1.30091593e-01 1.52650505e-01 6.41534388e-01
6.41663373e-02 1.48922682e-01 5.20524263e-01 -2.16022730e-01
-4.54973578e-01 1.12869883e+00 3.00414056e-01 -4.05459963e-02
9.04396772e-01 -3.49695653e-01 -7.67346323e-02 -1.33339897e-01
-6.27222657e-01 4.13776100e-01 1.00809348e+00 4.13662702e-01
8.66024375e-01 -1.62508416e+00 -7.97066212e-01 5.89466989e-01
3.76529023e-02 1.19848013e+00 1.84549719e-01 4.83769864e-01
-8.41441810e-01 -1.26417689e-02 8.49503577e-02 -9.73319888e-01
-6.22810245e-01 5.82241118e-01 4.22907561e-01 2.18881860e-01
-9.92118359e-01 6.37257397e-01 4.31139112e-01 -7.98485398e-01
1.57677546e-01 -7.74278820e-01 1.80652589e-01 -5.67551374e-01
-7.34196454e-02 3.05665255e-01 1.19925827e-01 -7.28006542e-01
-8.36896449e-02 1.24532235e+00 4.48056728e-01 1.04275763e-01
1.29637241e+00 -1.28955290e-01 -1.11554839e-01 1.33684739e-01
1.06543326e+00 -5.63159818e-04 -1.64737177e+00 -1.22952461e-01
-5.37282526e-01 -6.32240355e-01 -7.84741417e-02 -2.32837632e-01
-1.30702913e+00 9.98369515e-01 3.21213067e-01 -6.69438671e-03
6.78538918e-01 -1.52953386e-01 9.04088557e-01 3.25141728e-01
8.60716820e-01 -6.16255403e-01 -2.30693579e-01 6.10726297e-01
1.32847285e+00 -1.07901943e+00 -1.65285721e-01 -8.33250463e-01
-1.21998467e-01 1.07930398e+00 7.52063751e-01 -7.48868585e-01
8.02155375e-01 2.00758025e-01 -1.83682203e-01 -3.88447851e-01
-4.18499291e-01 8.03421587e-02 3.16894233e-01 6.94702387e-01
2.95080487e-02 6.29679188e-02 -9.27087665e-02 3.58238220e-01
-5.84885478e-01 1.49510399e-01 2.81171173e-01 7.51669466e-01
-3.47022474e-01 -1.10733175e+00 -4.64969814e-01 1.45940661e-01
1.31694555e-01 1.23909272e-01 -2.71006107e-01 7.46018469e-01
2.82515496e-01 5.10719240e-01 2.42784679e-01 -3.71998847e-01
7.29732931e-01 -2.58578271e-01 3.77587050e-01 -6.55475080e-01
-1.97410345e-01 -9.54896361e-02 -4.03105110e-01 -7.31928825e-01
-1.19072497e-01 -4.75512803e-01 -1.47415495e+00 -5.40336132e-01
2.81668957e-02 -7.02427030e-02 8.05402994e-01 6.06809199e-01
7.41847575e-01 5.56565881e-01 6.45966887e-01 -1.55729151e+00
-3.62512082e-01 -6.12864435e-01 -4.37670618e-01 5.68535447e-01
1.69881240e-01 -6.72647774e-01 -3.17903489e-01 1.87115893e-02]
|
[8.457903861999512, -3.5563912391662598]
|
49831644-9d5e-46c6-a7c8-9813b909a49d
|
on-the-development-of-a-bayesian-optimisation
|
2207.09154
| null |
https://arxiv.org/abs/2207.09154v2
|
https://arxiv.org/pdf/2207.09154v2.pdf
|
Investigating Bayesian optimization for expensive-to-evaluate black box functions: Application in fluid dynamics
|
Bayesian optimization provides an effective method to optimize expensive-to-evaluate black box functions. It has been widely applied to problems in many fields, including notably in computer science, e.g. in machine learning to optimize hyperparameters of neural networks, and in engineering, e.g. in fluid dynamics to optimize control strategies that maximize drag reduction. This paper empirically studies and compares the performance and the robustness of common Bayesian optimization algorithms on a range of synthetic test functions to provide general guidance on the design of Bayesian optimization algorithms for specific problems. It investigates the choice of acquisition function, the effect of different numbers of training samples, the exact and Monte Carlo based calculation of acquisition functions, and both single-point and multi-point optimization. The test functions considered cover a wide selection of challenges and therefore serve as an ideal test bed to understand the performance of Bayesian optimization to specific challenges, and in general. To illustrate how these findings can be used to inform a Bayesian optimization setup tailored to a specific problem, two simulations in the area of computational fluid dynamics are optimized, giving evidence that suitable solutions can be found in a small number of evaluations of the objective function for complex, real problems. The results of our investigation can similarly be applied to other areas, such as machine learning and physical experiments, where objective functions are expensive to evaluate and their mathematical expressions are unknown.
|
['Kevin Wilson', 'Sylvain Laizet', 'Andrew Wynn', "Joseph O'Connor", 'Richard D. Whalley', 'Yu Guan', 'Mike Diessner']
|
2022-07-19
| null | null | null | null |
['bayesian-optimisation']
|
['methodology']
|
[ 4.88465205e-02 -3.27693522e-01 1.84193641e-01 -2.04287902e-01
-4.54396427e-01 -3.46264660e-01 4.09831434e-01 3.02184105e-01
-6.34933114e-01 1.03042829e+00 -3.44962507e-01 -4.54482615e-01
-6.39681101e-01 -6.24398291e-01 -5.13484478e-01 -1.10908663e+00
-2.28202850e-01 8.12892020e-01 2.82708794e-01 -1.02377981e-01
5.84413052e-01 7.60830104e-01 -1.50465357e+00 -6.18766546e-01
6.06898546e-01 8.08273852e-01 8.98028389e-02 7.57212162e-01
2.90275723e-01 -1.04999602e-01 -6.57513976e-01 -4.61001880e-02
-4.83337790e-02 -2.72269368e-01 -4.62784439e-01 -1.75784424e-01
-4.77280989e-02 -5.21188378e-02 -2.58029830e-02 7.65677869e-01
9.65674222e-01 8.95800412e-01 8.55918407e-01 -5.82503676e-01
1.61248803e-01 9.91966054e-02 -1.05167687e-01 4.56336737e-01
-4.76250844e-03 5.00177383e-01 5.64750195e-01 -4.35983688e-01
1.94800377e-01 1.17072105e+00 4.62678671e-01 2.60574669e-01
-1.30781150e+00 -4.13647383e-01 -1.83562472e-01 5.58969006e-03
-1.12312412e+00 -4.04390991e-01 6.17747903e-01 -7.44950354e-01
7.55207837e-01 8.37745965e-02 7.55948424e-01 6.53367221e-01
5.67965865e-01 3.15574743e-02 1.13769448e+00 -4.88544732e-01
5.80883265e-01 1.99031979e-01 1.17502354e-01 5.30651093e-01
6.90945923e-01 7.65812814e-01 -4.92088646e-01 -2.89907068e-01
7.31876493e-01 -5.39658129e-01 -2.78338104e-01 -4.39083427e-01
-8.52599442e-01 1.09401977e+00 3.01728398e-02 8.39143991e-02
-2.16980845e-01 2.32436717e-01 1.51448250e-01 -1.60465747e-01
3.08202028e-01 1.05324566e+00 -5.89616299e-01 -5.09983718e-01
-7.25262463e-01 8.49920154e-01 1.18222177e+00 2.87016422e-01
4.88542289e-01 3.75684649e-01 -1.64260063e-02 8.55792582e-01
6.87874436e-01 7.14131534e-01 1.34757996e-01 -1.05924714e+00
3.24883848e-01 -1.64416179e-01 4.65377361e-01 -7.60138631e-01
-5.72567523e-01 -4.70907718e-01 -3.43566209e-01 5.45376837e-01
7.91851521e-01 -5.88449955e-01 -6.10578656e-01 1.27783298e+00
6.08496904e-01 -1.72898099e-02 -1.37791082e-01 1.15986562e+00
5.89556575e-01 6.91306174e-01 -1.38539061e-01 -1.98810458e-01
1.25758791e+00 -2.52341896e-01 -3.58383864e-01 -2.90965438e-01
4.27716434e-01 -1.07671690e+00 8.17267716e-01 5.70359647e-01
-1.04035902e+00 -1.64524466e-01 -1.04088688e+00 5.15459239e-01
-2.82209158e-01 -1.49850100e-01 6.20605648e-01 9.40328538e-01
-3.52489412e-01 1.17271197e+00 -1.09474277e+00 -1.71855718e-01
1.00223579e-01 4.65477675e-01 2.59652942e-01 2.19056889e-01
-9.79401529e-01 1.25893688e+00 4.13960874e-01 1.92536846e-01
-7.54007280e-01 -7.76729107e-01 -6.97339416e-01 2.03378927e-02
3.97982657e-01 -7.58478642e-01 1.23339283e+00 -2.35355377e-01
-1.96322525e+00 2.42089301e-01 -1.72095180e-01 -2.25021794e-01
3.09049636e-01 -2.37627625e-01 -7.07259178e-02 -1.93533272e-01
-3.04799914e-01 2.26759151e-01 7.19417751e-01 -9.29580688e-01
-1.18882924e-01 -1.11115381e-01 -2.26556540e-01 1.84080422e-01
2.67621845e-01 1.87428743e-02 3.12127043e-02 -5.37049174e-01
-1.43160120e-01 -1.02634466e+00 -4.48490769e-01 -3.51437569e-01
-1.56873822e-01 2.10776940e-01 5.77191055e-01 -5.03602982e-01
1.07132804e+00 -1.50603199e+00 3.12633991e-01 5.48482358e-01
-2.73571193e-01 1.75726026e-01 3.60749483e-01 4.47019339e-01
1.10478878e-01 4.47092764e-02 -5.42228699e-01 1.31841019e-01
-2.43025664e-02 6.27818406e-02 -1.35175675e-01 6.95922375e-01
2.79428035e-01 5.34773529e-01 -7.61575878e-01 -2.09655225e-01
5.84541023e-01 5.30730426e-01 -6.47801578e-01 2.16408372e-01
-1.61702603e-01 7.57661998e-01 -5.91598034e-01 1.95247203e-01
1.48831010e-01 -3.45054716e-02 -9.55625176e-02 4.67837835e-03
-3.33082676e-01 2.28670225e-01 -1.45606709e+00 8.66418481e-01
-5.22572100e-01 7.88732052e-01 2.91725397e-01 -1.19152069e+00
9.20379937e-01 5.92237860e-02 4.85027760e-01 -2.04851314e-01
4.10139889e-01 2.17260286e-01 5.00493348e-01 -5.21117508e-01
4.91473138e-01 -6.32163823e-01 3.75517160e-02 3.47112417e-01
-6.55765161e-02 -8.99687111e-01 3.90069723e-01 -3.62158060e-01
6.85025990e-01 1.60289958e-01 3.04187655e-01 -6.86337054e-01
4.75373536e-01 -9.80617553e-02 2.26329103e-01 7.40677655e-01
5.41684777e-02 4.11406457e-01 3.26034248e-01 -3.06126416e-01
-1.01593566e+00 -7.09333420e-01 -6.29674375e-01 6.49348974e-01
2.66473055e-01 -1.56169012e-01 -5.10372639e-01 1.84972167e-01
2.71308184e-01 7.36436725e-01 -2.50710815e-01 -2.97353327e-01
-7.19683707e-01 -1.55985129e+00 4.15428542e-02 2.17252493e-01
1.35387219e-02 -7.25405991e-01 -1.12021077e+00 3.34816188e-01
4.79157388e-01 -8.94421995e-01 1.09846868e-01 3.47654104e-01
-1.09810567e+00 -1.14165342e+00 -5.30328453e-01 1.30368592e-02
2.45360285e-01 -2.96859622e-01 1.06707835e+00 1.56853437e-01
-4.22557771e-01 6.01306021e-01 -1.03272840e-01 -6.96357191e-01
-6.14256561e-01 7.85078928e-02 1.65490523e-01 -4.11341906e-01
8.27962682e-02 -3.88253897e-01 -4.41128284e-01 7.52996564e-01
-7.74263024e-01 -4.96597409e-01 2.88026035e-01 9.34770048e-01
3.39433759e-01 2.36983657e-01 3.83394957e-01 -6.28359616e-01
9.10747409e-01 -3.80105406e-01 -1.30313623e+00 1.33226439e-01
-6.65513992e-01 3.61230284e-01 4.76211160e-01 -4.77278650e-01
-8.90052915e-01 -2.29818016e-01 -2.15844020e-01 -1.07922137e-01
-1.43935248e-01 5.76670110e-01 9.19764638e-02 -4.70251530e-01
7.08599627e-01 -2.40139484e-01 2.13163450e-01 -5.02582073e-01
-1.25345632e-01 3.12781632e-01 8.12693611e-02 -1.23016119e+00
6.71733558e-01 6.20679446e-02 7.57454395e-01 -1.27605784e+00
-4.42879021e-01 -2.98993319e-01 -3.39283824e-01 -4.78024721e-01
6.24665260e-01 -2.95623660e-01 -9.78446722e-01 3.75811309e-01
-7.72987425e-01 -6.25364482e-01 -2.00357556e-01 9.45144176e-01
-7.12317646e-01 2.13413909e-01 -2.39432022e-01 -1.07838154e+00
3.05713341e-02 -1.55386662e+00 9.91241574e-01 6.44876719e-01
-4.11571831e-01 -1.53614283e+00 1.03893034e-01 1.63309798e-01
5.63874900e-01 1.31221756e-01 9.11236823e-01 -4.90285486e-01
-6.38117313e-01 -1.15703344e-01 3.88790756e-01 1.44041404e-01
-1.23560779e-01 4.96818781e-01 -7.26619959e-01 -3.14132750e-01
9.33277607e-02 -1.02307819e-01 8.03612471e-01 8.48816097e-01
9.16275918e-01 4.00393195e-02 -3.18801045e-01 4.90787566e-01
1.22650838e+00 3.11688602e-01 3.44220579e-01 2.12231442e-01
3.35085243e-01 7.57271945e-01 6.16330802e-01 6.84288502e-01
-2.37060666e-01 1.01871514e+00 2.27416396e-01 2.46900111e-01
2.84762114e-01 2.87195712e-01 1.26499742e-01 5.74445188e-01
-1.86664149e-01 -1.37628585e-01 -1.03042221e+00 6.52673766e-02
-1.57842660e+00 -7.67449677e-01 -2.71127969e-02 2.56777763e+00
5.49425364e-01 2.05140218e-01 1.15199029e-01 1.00559413e-01
7.20819831e-01 -4.24242914e-02 -5.11291206e-01 -5.63941300e-01
2.55573511e-01 5.54332197e-01 6.96167469e-01 7.74961710e-01
-1.01443768e+00 4.95820165e-01 7.16349983e+00 8.49944651e-01
-1.26182926e+00 -1.10491313e-01 5.30631900e-01 -2.58560061e-01
3.55547108e-02 1.61292672e-01 -1.08185446e+00 6.52990341e-01
1.10776818e+00 -1.46073669e-01 7.40460813e-01 4.89841580e-01
6.09546781e-01 -7.05941677e-01 -8.38969529e-01 7.45411396e-01
-4.20810640e-01 -1.25520539e+00 -4.98185217e-01 6.97712526e-02
6.12100840e-01 -1.29123166e-01 -1.20295666e-01 1.43373832e-01
-6.66850992e-03 -1.16838837e+00 5.42726517e-01 7.17848718e-01
1.81490839e-01 -6.15527034e-01 8.44790518e-01 2.96990544e-01
-6.87558293e-01 1.35349676e-01 -2.96326697e-01 -2.34252572e-01
6.05529547e-01 8.72174442e-01 -8.39075983e-01 3.02431613e-01
6.31219625e-01 2.08387122e-01 -6.85426593e-02 1.63358998e+00
-5.70810251e-02 8.49353671e-01 -9.63869750e-01 -7.37106383e-01
9.34663489e-02 -6.05245233e-01 9.88133609e-01 1.02064359e+00
3.73516470e-01 2.02383194e-02 -9.57335532e-02 1.06691766e+00
6.52160287e-01 5.66326566e-02 -3.25528592e-01 -1.11544192e-01
4.84672666e-01 1.02116394e+00 -9.21808422e-01 8.71754363e-02
1.53010771e-01 -3.00306678e-01 -1.09755449e-01 6.52707100e-01
-7.42122531e-01 -5.23955405e-01 7.65667200e-01 3.10061067e-01
3.82633716e-01 -6.36859775e-01 -4.50830847e-01 -7.06144631e-01
-2.65432030e-01 -4.78202492e-01 1.82348415e-01 -4.41364169e-01
-9.57005680e-01 -8.54992867e-02 1.00282860e+00 -6.36901677e-01
-5.77635467e-01 -1.12800705e+00 -8.45005929e-01 1.00014520e+00
-1.18873394e+00 -1.32093430e-01 -2.47742739e-02 -1.29017696e-01
1.07250080e-01 -3.09479415e-01 3.53723586e-01 1.10601194e-01
-8.19471002e-01 1.02639534e-01 5.18766642e-01 -2.34642372e-01
3.31184000e-01 -8.53825331e-01 1.39431152e-02 7.04922557e-01
-2.41783425e-01 6.97383285e-01 1.42786074e+00 -6.39017761e-01
-1.42993605e+00 -5.70037842e-01 -7.65979663e-03 -2.90382892e-01
4.90045279e-01 -2.87457630e-02 -9.58915055e-01 -5.31308465e-02
-3.13371420e-01 -2.22388893e-01 4.22797829e-01 2.49064788e-01
6.12725437e-01 -1.34681314e-02 -1.10459149e+00 5.38247824e-01
3.52950156e-01 3.22373137e-02 -3.02760959e-01 2.58259118e-01
8.85006562e-02 -8.11514616e-01 -1.19689989e+00 6.89205050e-01
6.65124893e-01 -8.63168061e-01 9.55737531e-01 -5.81516445e-01
1.77225679e-01 -2.36294046e-01 -3.21354344e-02 -1.56347430e+00
7.16794580e-02 -8.33310962e-01 -1.70684457e-01 8.96522284e-01
3.47119957e-01 -1.12467289e+00 6.56709015e-01 8.36710632e-01
-1.48720533e-01 -1.23572636e+00 -1.28834200e+00 -7.33129025e-01
3.20187896e-01 -4.82353926e-01 3.78370643e-01 2.26197079e-01
-4.16238517e-01 2.72814542e-01 3.66941988e-02 9.01920125e-02
6.33872688e-01 -3.51814483e-03 8.46662045e-01 -1.23815334e+00
-6.17189944e-01 -7.72607386e-01 -2.35246792e-01 -8.52291465e-01
4.86751646e-02 -3.50192338e-01 2.32540801e-01 -9.27620769e-01
-3.02587897e-01 -7.66197562e-01 3.22655290e-01 -2.36004159e-01
-3.78504127e-01 -1.66549787e-01 1.10686652e-01 -1.96069747e-01
1.42151445e-01 7.33959854e-01 1.14752090e+00 2.40851626e-01
-3.10340941e-01 3.88054341e-01 -3.90944719e-01 5.82777500e-01
7.63336420e-01 -6.06748819e-01 -2.08621562e-01 1.45950347e-01
2.85888612e-01 -1.62032945e-03 4.26183134e-01 -1.05547309e+00
-1.46324877e-02 -4.56838369e-01 2.49393970e-01 -4.80910614e-02
5.11216938e-01 -6.61132336e-01 1.43082172e-01 2.32501015e-01
1.85513552e-02 -7.30925333e-03 4.97037053e-01 4.31516021e-01
-7.39861233e-03 -8.90978038e-01 1.09947300e+00 -8.71890709e-02
-3.91582340e-01 -1.14192087e-02 -4.89634126e-01 3.34106833e-01
9.34291840e-01 -2.89793581e-01 -1.22096650e-01 -4.91864830e-01
-5.22818148e-01 2.97876924e-01 4.52603787e-01 -1.31412089e-01
4.22341973e-01 -7.92527020e-01 -5.05087316e-01 1.83453918e-01
-4.28377151e-01 3.00197303e-01 4.86310422e-02 9.39219952e-01
-8.65166187e-01 5.19978046e-01 7.09082484e-02 -8.80590677e-01
-9.93124783e-01 1.69912931e-02 7.68772840e-01 -2.36098826e-01
-2.62443304e-01 7.13331044e-01 -5.02570570e-02 -2.34955102e-01
-7.03740343e-02 -3.32812995e-01 3.93325761e-02 -1.91422373e-01
8.31424370e-02 6.78606570e-01 1.34035304e-01 -4.53680903e-01
-1.99577287e-01 9.05410409e-01 4.73233163e-01 -4.20360297e-01
1.21768034e+00 1.26053408e-01 1.47901088e-01 5.49763739e-01
7.67348051e-01 -1.37902692e-01 -1.40505493e+00 2.85810590e-01
-1.16157711e-01 -4.48286682e-01 2.26795554e-01 -5.19818068e-01
-8.29797029e-01 9.08278883e-01 4.48685944e-01 4.25565206e-02
7.04183698e-01 -1.76720664e-01 2.39142194e-01 4.37304646e-01
2.76640683e-01 -1.12038887e+00 -6.10665567e-02 7.02501953e-01
7.50891209e-01 -9.56508279e-01 5.79148591e-01 -2.24531174e-01
-3.65087003e-01 1.36547875e+00 3.44531327e-01 -1.78890392e-01
9.67355371e-01 3.76426786e-01 -2.88067311e-01 -7.79441074e-02
-5.75957417e-01 8.66339877e-02 4.64542747e-01 3.20882559e-01
4.03475523e-01 -7.28947669e-02 -3.39265883e-01 1.77211061e-01
-2.87869096e-01 -1.37732714e-01 4.69856799e-01 8.95668149e-01
-5.67800939e-01 -1.17658532e+00 -8.35158944e-01 7.40765095e-01
-3.77490520e-01 1.35425851e-01 1.34205729e-01 9.15545821e-01
-2.00773641e-01 8.02926064e-01 -6.76632896e-02 4.26604971e-02
2.68797040e-01 2.21580684e-01 6.99499607e-01 -4.59459066e-01
-4.99197692e-01 1.09782834e-02 4.03396726e-01 -1.83367193e-01
-4.40927774e-01 -9.51870739e-01 -1.02631760e+00 -2.59099394e-01
-6.71926260e-01 4.46850032e-01 8.94897401e-01 1.24965692e+00
-7.61900749e-03 5.87733746e-01 7.70731858e-05 -1.45723701e+00
-6.88577890e-01 -8.99653435e-01 -5.60827792e-01 -6.52200058e-02
2.32780144e-01 -1.40465474e+00 -5.45749664e-01 -4.49246734e-01]
|
[6.238903999328613, 3.679579257965088]
|
b362a91c-1caf-4661-9389-bc87fb330373
|
quantifying-the-scanner-induced-domain-gap-in
|
2103.16515
| null |
https://arxiv.org/abs/2103.16515v1
|
https://arxiv.org/pdf/2103.16515v1.pdf
|
Quantifying the Scanner-Induced Domain Gap in Mitosis Detection
|
Automated detection of mitotic figures in histopathology images has seen vast improvements, thanks to modern deep learning-based pipelines. Application of these methods, however, is in practice limited by strong variability of images between labs. This results in a domain shift of the images, which causes a performance drop of the models. Hypothesizing that the scanner device plays a decisive role in this effect, we evaluated the susceptibility of a standard mitosis detection approach to the domain shift introduced by using a different whole slide scanner. Our work is based on the MICCAI-MIDOG challenge 2021 data set, which includes 200 tumor cases of human breast cancer and four scanners. Our work indicates that the domain shift induced not by biochemical variability but purely by the choice of acquisition device is underestimated so far. Models trained on images of the same scanner yielded an average F1 score of 0.683, while models trained on a single other scanner only yielded an average F1 score of 0.325. Training on another multi-domain mitosis dataset led to mean F1 scores of 0.52. We found this not to be reflected by domain-shifts measured as proxy A distance-derived metric.
|
['Andreas Maier', 'Francesco Ciompi', 'Natalie ter Hoeve', 'Katharina Breininger', 'Nikolas Stathonikos', 'Robert Klopfleisch', 'Mitko Veta', 'Christof Bertram', 'Marc Aubreville']
|
2021-03-30
| null | null | null | null |
['mitosis-detection']
|
['medical']
|
[ 2.59685993e-01 -1.74015183e-02 1.61126480e-01 -2.24750102e-01
-1.20289469e+00 -8.28101695e-01 6.90720856e-01 4.68724042e-01
-7.48323739e-01 8.00795496e-01 -1.32041857e-01 -3.49451244e-01
2.65845517e-03 -4.62192595e-01 -8.66789877e-01 -1.27012980e+00
2.87507147e-01 7.53491938e-01 5.19518733e-01 2.64258802e-01
3.90268266e-01 5.65089405e-01 -9.69365358e-01 4.38500315e-01
1.74512938e-01 4.51679140e-01 1.17094859e-01 1.10848236e+00
-7.34307170e-02 3.65886956e-01 -8.96372199e-01 -2.61520952e-01
8.05660486e-02 -3.39807302e-01 -6.81077123e-01 -4.82077040e-02
4.28974479e-01 -2.50622064e-01 -1.40983388e-01 8.05210412e-01
7.68012285e-01 -7.70491421e-01 8.98437858e-01 -9.46989179e-01
-7.70222582e-03 5.08956671e-01 -7.53696799e-01 5.91206968e-01
-7.07571488e-03 4.71246123e-01 4.47868258e-01 -3.53901386e-01
1.16034317e+00 6.32631302e-01 8.17431867e-01 3.12268257e-01
-1.67618132e+00 -4.97441471e-01 -5.70370913e-01 -7.19081312e-02
-1.42660487e+00 -3.36661220e-01 2.06387937e-01 -7.65175581e-01
8.11881423e-01 -5.98899312e-02 6.66732192e-01 9.70153451e-01
7.19662368e-01 1.34944186e-01 1.33604491e+00 -3.46305937e-01
3.05080682e-01 3.20258677e-01 -1.03066131e-01 4.43494588e-01
6.15275323e-01 -2.30532754e-02 -3.39756459e-01 -2.19423473e-02
6.86846495e-01 -3.73514354e-01 -3.30277056e-01 -3.11711818e-01
-1.36222899e+00 6.16424561e-01 1.45987108e-01 8.94066811e-01
1.03262812e-01 4.39057425e-02 6.04581594e-01 4.12286937e-01
-1.45541698e-01 6.64871812e-01 -5.06739318e-01 4.35236283e-02
-1.16607022e+00 -1.12988465e-01 4.80618507e-01 5.20418763e-01
5.39268553e-01 -6.38176322e-01 -4.14505340e-02 1.77635863e-01
8.74365047e-02 1.23237878e-01 8.97226691e-01 -6.43811703e-01
-2.55586177e-01 6.26954138e-01 -9.45526138e-02 -7.43934989e-01
-9.90595102e-01 -5.10731816e-01 -6.04883492e-01 3.06589723e-01
1.49101675e+00 1.81880966e-01 -9.90799785e-01 1.51343453e+00
1.75794810e-01 -1.83947347e-02 -9.50045288e-02 7.02516854e-01
6.06471002e-01 -8.83773342e-02 9.60839018e-02 2.48787813e-02
1.37571645e+00 -4.61005300e-01 -4.06155884e-01 8.71955454e-02
1.39448202e+00 -1.00347686e+00 9.94087040e-01 5.36462307e-01
-6.81504905e-01 -1.34552300e-01 -1.32577169e+00 1.60191245e-02
-7.27509022e-01 -1.45255569e-02 2.14221999e-01 8.47049475e-01
-1.26559544e+00 6.06981993e-01 -7.62317657e-01 -9.34978783e-01
4.79002684e-01 5.56306362e-01 -5.06822586e-01 1.29010547e-02
-6.88908160e-01 9.19829190e-01 2.89220005e-01 -3.42613757e-01
-6.76705062e-01 -1.17553198e+00 -2.87810296e-01 -3.29838365e-01
-1.74535170e-01 -3.69222939e-01 1.10873199e+00 -6.29080057e-01
-1.15575814e+00 1.58519280e+00 2.53457632e-02 -4.53888059e-01
9.09239590e-01 6.03553295e-01 -2.21483067e-01 2.04965875e-01
1.66627496e-01 6.57951951e-01 3.83655906e-01 -9.49549019e-01
-6.74086452e-01 -4.79027033e-01 -5.32111287e-01 -3.64999205e-01
-3.55348103e-02 -1.79504231e-01 -2.94042826e-01 -5.01713865e-02
-1.78382874e-01 -9.66752052e-01 -1.29619002e-01 1.68326534e-02
-7.22258091e-02 2.58711040e-01 5.81540525e-01 -5.58299363e-01
6.84963763e-01 -2.07965684e+00 -4.72505927e-01 1.77646190e-01
3.58255595e-01 9.72376764e-02 -1.14639811e-01 1.59765765e-01
-1.34878173e-01 3.69438201e-01 3.37782502e-02 -1.49865951e-02
-1.10630438e-01 -2.05084667e-01 3.95389467e-01 1.14373398e+00
1.43279344e-01 7.72758901e-01 -7.60501981e-01 -6.58067882e-01
9.29886252e-02 5.71711659e-01 -7.87203312e-02 -1.67920545e-01
7.34547153e-02 5.30626357e-01 2.33574972e-01 6.15059853e-01
9.06658888e-01 -4.76238400e-01 3.59082550e-01 -1.12577036e-01
1.70331057e-02 8.41207616e-03 -6.39507294e-01 1.75307405e+00
1.30769936e-02 9.04198825e-01 -1.17317021e-01 -5.63981712e-01
7.05230057e-01 1.78167090e-01 3.60886186e-01 -8.04305673e-01
3.33183825e-01 5.31014264e-01 5.77777565e-01 -2.38821685e-01
3.51308696e-02 -3.03845555e-01 1.30103052e-01 4.07079518e-01
1.37330696e-01 -2.74682432e-01 2.60262102e-01 8.63560885e-02
1.66275287e+00 -3.95147741e-01 1.91632256e-01 -6.19034052e-01
3.12600940e-01 2.78158247e-01 2.36037984e-01 6.40390277e-01
-8.44831645e-01 1.00369155e+00 1.05116999e+00 -3.64108145e-01
-1.41376758e+00 -9.05137956e-01 -4.43826407e-01 4.85043406e-01
-1.58146441e-01 3.64790112e-02 -7.18567073e-01 -7.62408793e-01
7.97466468e-03 1.67593181e-01 -1.02112341e+00 -1.87067002e-01
-2.54284352e-01 -1.11310148e+00 1.05013299e+00 2.25539953e-01
8.50044936e-02 -4.28146720e-01 -9.82656360e-01 1.03561074e-01
1.23971045e-01 -1.11583030e+00 -3.00275266e-01 5.51376939e-01
-7.59751916e-01 -1.24074328e+00 -9.33520854e-01 -6.23714924e-01
5.77644467e-01 -2.71034036e-02 1.25216639e+00 1.53506428e-01
-6.45933449e-01 -6.81832731e-02 -1.28480569e-01 -5.20694137e-01
-6.98266566e-01 4.05558258e-01 -3.64258587e-01 -3.18072617e-01
8.31721127e-01 -1.25867635e-01 -7.28496432e-01 3.53570342e-01
-1.01762843e+00 -3.91120493e-01 9.97800589e-01 8.39225113e-01
7.67789185e-01 1.53034449e-01 3.39877814e-01 -1.10452616e+00
4.91931848e-02 -3.02657902e-01 -6.91083193e-01 1.52502432e-01
-7.13108301e-01 7.55978035e-05 3.68309975e-01 -4.87429857e-01
-7.18673527e-01 3.53352726e-01 9.59219486e-02 -9.86141562e-02
-4.37457412e-01 3.55703264e-01 -6.33741543e-02 -3.18185419e-01
9.00263011e-01 -4.12972532e-02 3.14601660e-01 1.00253798e-01
-5.37423253e-01 4.62605298e-01 4.66175526e-01 7.32598156e-02
4.78189856e-01 7.61005461e-01 3.86863440e-01 -8.43709230e-01
-3.00748557e-01 -7.25925386e-01 -5.52139759e-01 -6.39498159e-02
7.50224054e-01 -7.53209293e-01 -4.96057123e-01 7.78453350e-01
-8.26887190e-01 -7.75563180e-01 4.90031801e-02 4.77012724e-01
-3.14564198e-01 1.11595713e-01 -6.60179317e-01 -2.09772363e-01
-4.69012409e-02 -1.32817090e+00 1.15563500e+00 2.75737792e-01
-5.33791363e-01 -1.09783340e+00 3.25208962e-01 4.59516317e-01
5.42770326e-01 5.50034523e-01 7.82919586e-01 -8.89293730e-01
-3.17500740e-01 -4.23068911e-01 -2.73229897e-01 -3.73524457e-01
8.66576061e-02 4.96665329e-01 -1.28229296e+00 -4.00071681e-01
-6.48192018e-02 -1.74670413e-01 8.88184845e-01 5.25090218e-01
8.12850773e-01 5.44418216e-01 -6.53235972e-01 5.42121589e-01
1.68844211e+00 1.67360976e-01 8.87456059e-01 6.88287258e-01
2.22991720e-01 5.72765768e-01 2.61269689e-01 -1.21837214e-01
-7.91025013e-02 4.59308982e-01 3.08240235e-01 -3.67297471e-01
-1.95102677e-01 6.61257580e-02 5.71184121e-02 1.63913742e-02
4.74790573e-01 -8.17498937e-02 -1.37842882e+00 8.38747978e-01
-1.27096546e+00 -5.95681131e-01 -4.78740662e-01 2.20864940e+00
7.33358502e-01 4.09689635e-01 2.20480785e-01 3.04444730e-01
6.75543427e-01 -4.48489517e-01 -5.79081357e-01 -2.45663762e-01
-3.88990134e-01 7.10416734e-02 9.56540406e-01 3.11645508e-01
-7.36448467e-01 3.85396838e-01 6.92683506e+00 7.79337108e-01
-1.48956728e+00 -1.62667409e-01 1.00899720e+00 -2.51431584e-01
1.11440994e-01 -1.59072533e-01 -6.76460743e-01 6.83359325e-01
1.35667932e+00 -7.63331726e-02 -2.06205428e-01 3.77477646e-01
1.31513029e-01 -5.48104465e-01 -1.30321753e+00 8.73049259e-01
-1.57774791e-01 -1.32806909e+00 -3.89176697e-01 7.92627156e-01
5.45601487e-01 1.14724092e-01 1.97936788e-01 8.21173098e-03
6.77520260e-02 -1.29161918e+00 1.14651352e-01 4.33024853e-01
8.10537398e-01 -6.74831450e-01 1.42588913e+00 9.10351649e-02
-5.23663282e-01 5.34035444e-01 -3.45032185e-01 2.02868327e-01
-3.41215432e-01 9.58882034e-01 -1.66418040e+00 -7.01352628e-03
7.11927712e-01 8.42413679e-02 -9.68505383e-01 1.10593081e+00
3.09017450e-01 5.65359116e-01 -3.99011612e-01 4.67530638e-02
-1.02442563e-01 3.89024496e-01 5.08504845e-02 1.62391627e+00
2.72686034e-01 -4.85572636e-01 -6.05168581e-01 5.07594585e-01
4.08845767e-02 -6.68251663e-02 -4.70892549e-01 -3.12263101e-01
3.82028550e-01 1.65723276e+00 -1.31908071e+00 6.04097918e-03
-4.63821083e-01 6.73999131e-01 5.30163422e-02 -7.98734576e-02
-9.05028582e-01 -4.07008052e-01 2.56764203e-01 5.24731755e-01
3.65172625e-01 2.71550506e-01 -4.93409425e-01 -6.05126619e-01
-3.98775458e-01 -8.01141024e-01 4.40752923e-01 -4.54040617e-01
-1.14070046e+00 2.51038164e-01 -4.13834095e-01 -8.70340645e-01
1.51653007e-01 -8.41594279e-01 -4.47946608e-01 9.64725912e-01
-1.45904541e+00 -8.29485416e-01 -3.32124352e-01 2.01292560e-01
2.16092411e-02 -1.74388289e-02 7.12215722e-01 3.01984161e-01
-3.07374835e-01 8.97366285e-01 1.40027434e-01 4.42200527e-02
1.40552378e+00 -1.49724829e+00 9.23539624e-02 4.85288829e-01
-2.33541697e-01 4.60885167e-01 9.73644495e-01 -2.43742868e-01
-1.01657724e+00 -9.26093519e-01 1.02301276e+00 -8.47665370e-01
8.10176313e-01 -1.55979851e-02 -9.20990765e-01 4.46698695e-01
2.13544726e-01 -1.51666468e-02 1.19978964e+00 -3.53737801e-01
-1.93635806e-01 8.88593495e-02 -1.55138361e+00 2.55719095e-01
5.12924016e-01 -3.62971693e-01 6.52099997e-02 2.16339737e-01
2.33939260e-01 -4.73497868e-01 -1.11779368e+00 1.42506048e-01
6.55988634e-01 -1.20350826e+00 5.38293660e-01 -2.39053428e-01
5.00920892e-01 -4.19321150e-01 5.98086268e-02 -1.13508594e+00
-4.33403045e-01 8.04798119e-03 4.38974112e-01 1.11868858e+00
7.04136789e-01 -5.91102540e-01 1.16103601e+00 4.11510527e-01
7.86843151e-02 -6.27040803e-01 -9.71552014e-01 -7.34650791e-01
5.98643899e-01 1.57494947e-01 5.70408165e-01 7.70002723e-01
1.11652650e-01 8.32408294e-02 5.11083663e-01 1.60449767e-03
4.81437266e-01 -4.04169947e-01 8.38536263e-01 -1.15674305e+00
-3.95208567e-01 -7.72734821e-01 -9.48178589e-01 -1.82350323e-01
-2.03045234e-01 -5.45834661e-01 1.00886412e-01 -1.06283176e+00
4.79027152e-01 -1.93558708e-01 -3.86937827e-01 1.00509785e-01
1.41507819e-01 5.40838301e-01 -2.42931679e-01 2.67985314e-01
-4.28275436e-01 -5.51928282e-01 1.13313437e+00 -1.67259127e-01
7.62140676e-02 -4.22096074e-01 -7.98346937e-01 5.94091535e-01
8.78680468e-01 -6.56655073e-01 -1.00922771e-01 -2.23313570e-01
3.48336577e-01 -2.88793236e-01 2.44060174e-01 -1.29678440e+00
4.31766540e-01 1.38298199e-01 9.37004924e-01 -4.68285859e-01
7.30741099e-02 -7.57667124e-01 5.09634733e-01 8.54234517e-01
-2.05992401e-01 -2.39760831e-01 4.17878091e-01 4.15372908e-01
8.32198933e-02 -1.42529458e-01 1.01545942e+00 -2.41522565e-01
-2.56377041e-01 -1.40245348e-01 -7.61796474e-01 -8.99282843e-02
1.08381283e+00 -6.23018324e-01 -7.84185648e-01 6.12846725e-02
-6.14877045e-01 6.48211222e-03 1.15905488e+00 -2.22520098e-01
-1.69480905e-01 -8.62897575e-01 -6.31277025e-01 4.17825729e-02
3.80618989e-01 3.19174156e-02 3.18894118e-01 1.29947686e+00
-9.25561547e-01 5.63917100e-01 -3.14028680e-01 -1.02503479e+00
-1.47444403e+00 4.16311532e-01 5.87118566e-01 -7.45414913e-01
-3.58058675e-03 1.00239909e+00 7.36617073e-02 -2.68025130e-01
3.39716747e-02 -3.60171169e-01 7.33629391e-02 2.19759703e-01
4.96810377e-01 2.38922313e-01 6.62023664e-01 -4.22240406e-01
-6.39178872e-01 5.59590101e-01 -6.25025690e-01 -8.91079307e-02
9.88967538e-01 1.18875206e-01 1.87812418e-01 4.69191730e-01
1.23905361e+00 -1.06960133e-01 -1.10724020e+00 2.39549085e-01
1.12218320e-01 -2.83590376e-01 2.63140984e-02 -1.02427948e+00
-9.57369506e-01 7.44289160e-01 9.81757820e-01 9.19214413e-02
9.61963117e-01 -3.27728363e-03 4.59748864e-01 1.51537852e-02
3.97277713e-01 -1.10839343e+00 1.02203861e-02 1.97070003e-01
3.27661097e-01 -1.45281053e+00 9.09056067e-02 -3.25123444e-02
-2.98056066e-01 1.11633313e+00 5.11927068e-01 1.32547811e-01
3.34157586e-01 6.87892616e-01 4.73149657e-01 -2.76936799e-01
-7.33143568e-01 1.50877625e-01 -5.18445551e-01 8.95644128e-01
7.10567951e-01 -9.62583944e-02 -3.68124276e-01 3.90703738e-01
-2.46684954e-01 5.11668265e-01 9.54328477e-01 1.03198218e+00
-2.49360248e-01 -1.01955080e+00 -5.13898313e-01 4.78847355e-01
-6.08295679e-01 4.63233411e-01 -7.94273138e-01 1.13122129e+00
1.74726978e-01 6.11402154e-01 2.83775866e-01 -2.46949524e-01
3.46081480e-02 9.99716744e-02 6.60958230e-01 -4.38893884e-01
-5.18060029e-01 3.40790525e-02 -3.90858144e-01 -2.06407040e-01
-4.01095271e-01 -9.68906820e-01 -1.39191222e+00 -5.56908846e-01
-2.88558513e-01 -1.62436917e-01 6.35144651e-01 6.52370036e-01
2.80807197e-01 6.40532672e-01 8.54575485e-02 -6.01223946e-01
-2.86676735e-01 -8.99143636e-01 -8.73077095e-01 2.27216274e-01
4.15980965e-01 -4.62452263e-01 -7.90813446e-01 3.92790139e-01]
|
[15.134425163269043, -3.151902914047241]
|
4aa43c8e-ecd6-4d57-916b-7305ba5a7d9c
|
provably-efficient-safe-exploration-via
|
2003.00534
| null |
https://arxiv.org/abs/2003.00534v2
|
https://arxiv.org/pdf/2003.00534v2.pdf
|
Provably Efficient Safe Exploration via Primal-Dual Policy Optimization
|
We study the Safe Reinforcement Learning (SRL) problem using the Constrained Markov Decision Process (CMDP) formulation in which an agent aims to maximize the expected total reward subject to a safety constraint on the expected total value of a utility function. We focus on an episodic setting with the function approximation where the Markov transition kernels have a linear structure but do not impose any additional assumptions on the sampling model. Designing SRL algorithms with provable computational and statistical efficiency is particularly challenging under this setting because of the need to incorporate both the safety constraint and the function approximation into the fundamental exploitation/exploration tradeoff. To this end, we present an \underline{O}ptimistic \underline{P}rimal-\underline{D}ual Proximal Policy \underline{OP}timization (OPDOP) algorithm where the value function is estimated by combining the least-squares policy evaluation and an additional bonus term for safe exploration. We prove that the proposed algorithm achieves an $\tilde{O}(d H^{2.5}\sqrt{T})$ regret and an $\tilde{O}(d H^{2.5}\sqrt{T})$ constraint violation, where $d$ is the dimension of the feature mapping, $H$ is the horizon of each episode, and $T$ is the total number of steps. These bounds hold when the reward/utility functions are fixed but the feedback after each episode is bandit. Our bounds depend on the capacity of the state-action space only through the dimension of the feature mapping and thus our results hold even when the number of states goes to infinity. To the best of our knowledge, we provide the first provably efficient online policy optimization algorithm for CMDP with safe exploration in the function approximation setting.
|
['Mihailo R. Jovanović', 'Xiaohan Wei', 'Zhaoran Wang', 'Zhuoran Yang', 'Dongsheng Ding']
|
2020-03-01
| null | null | null | null |
['safe-exploration']
|
['robots']
|
[-1.14321634e-02 5.50981462e-01 -4.29400116e-01 1.24926761e-01
-8.19369137e-01 -6.06250525e-01 1.60155505e-01 2.65059143e-01
-9.14863050e-01 1.01784348e+00 -2.93277293e-01 -5.88123679e-01
-7.39715874e-01 -7.50610590e-01 -8.70075762e-01 -1.00409544e+00
-5.39416254e-01 4.58373040e-01 -1.34992257e-01 3.97510082e-02
1.18477717e-01 2.90170699e-01 -1.10792053e+00 -6.43597484e-01
9.48944330e-01 1.57080042e+00 2.09448799e-01 6.28437459e-01
3.39237124e-01 6.41715467e-01 -3.07289630e-01 -9.37747583e-02
6.59152865e-01 -4.84415323e-01 -5.18905103e-01 1.57011136e-01
-3.74015331e-01 -6.61927879e-01 -3.69426489e-01 1.23073816e+00
4.39727545e-01 5.44312000e-01 2.98686832e-01 -1.37031054e+00
9.78058800e-02 5.92950046e-01 -5.98274946e-01 1.62060201e-01
-4.54835854e-02 3.32456529e-01 9.99038637e-01 -1.23003229e-01
4.86069351e-01 9.92355645e-01 -7.77965486e-02 5.07260919e-01
-1.12630141e+00 -4.90159690e-01 5.44319272e-01 -5.11665046e-02
-1.01792502e+00 -2.87073348e-02 2.34980643e-01 -3.52179319e-01
7.23280907e-01 1.57035485e-01 7.31094658e-01 5.36436439e-01
1.26164377e-01 9.11036611e-01 1.10223901e+00 -3.33989620e-01
8.59093070e-01 6.28455654e-02 -4.58583571e-02 6.41956985e-01
2.58048743e-01 5.59883595e-01 -2.09846154e-01 -1.10408835e-01
8.44587445e-01 1.41920730e-01 -1.31190360e-01 -4.26056445e-01
-6.47600591e-01 1.08920586e+00 2.59948242e-02 -2.90240854e-01
-6.58124804e-01 4.78330463e-01 1.78907707e-01 5.75358510e-01
9.03664678e-02 5.20893574e-01 -3.22203815e-01 -5.03878415e-01
-7.71873176e-01 4.12605137e-01 7.34391630e-01 1.00967562e+00
4.91057158e-01 2.23818123e-01 -4.15443897e-01 3.36671025e-01
-9.93118882e-02 7.82693028e-01 -7.93339610e-02 -1.41871071e+00
8.01729321e-01 1.38101056e-01 1.00391340e+00 -2.79456019e-01
-2.66416281e-01 -6.51422739e-01 -2.55579919e-01 3.76856267e-01
7.11163044e-01 -8.88670087e-01 -5.68829000e-01 2.09521985e+00
3.39633584e-01 -3.27217668e-01 8.40849523e-03 7.00865746e-01
-4.98636276e-01 6.30801499e-01 -1.40536174e-01 -8.40258658e-01
1.03387320e+00 -4.72012609e-01 -5.68284631e-01 -2.43338019e-01
4.89977926e-01 -2.01210082e-01 1.06098402e+00 4.46716428e-01
-1.39231920e+00 1.96422547e-01 -9.76825595e-01 6.42553568e-01
3.42101395e-01 -1.68524861e-01 4.39146221e-01 6.87173545e-01
-5.13109267e-01 6.57565057e-01 -9.85890746e-01 2.10989818e-01
3.66664350e-01 4.94412124e-01 1.66640654e-01 2.70521361e-02
-9.55882967e-01 7.70160496e-01 4.99546111e-01 1.66583098e-02
-1.39054024e+00 -3.97462457e-01 -4.74981844e-01 3.78709406e-01
1.24994993e+00 -1.59556329e-01 1.38755584e+00 -6.56706870e-01
-1.49508190e+00 6.89506158e-02 3.38145941e-01 -8.62538517e-01
8.12993288e-01 -1.10799171e-01 2.20965177e-01 1.98115245e-01
1.24014067e-02 1.49503157e-01 6.90684617e-01 -7.32737005e-01
-8.81366611e-01 -5.45928359e-01 5.11399686e-01 5.02870977e-01
-3.10601085e-01 -2.15118036e-01 2.80085560e-02 -3.89992684e-01
-3.80377769e-01 -1.26459980e+00 -4.42676038e-01 -2.03669116e-01
-1.47619933e-01 7.97574595e-02 3.85987788e-01 -5.42900980e-01
1.42030644e+00 -1.95314658e+00 2.36208752e-01 2.95583814e-01
-3.01176786e-01 -1.11022353e-01 6.42551407e-02 4.75504816e-01
4.83784854e-01 -9.97566357e-02 -3.27634931e-01 -2.08329961e-01
3.36386025e-01 3.78424525e-01 -4.68748212e-01 6.00580275e-01
-5.79970539e-01 4.17588383e-01 -6.96888864e-01 -3.13909724e-02
-9.05406401e-02 -1.49719998e-01 -5.80372691e-01 1.66694567e-01
-5.98943710e-01 2.05426499e-01 -8.81247520e-01 2.57273525e-01
3.26329648e-01 6.73340186e-02 2.44025260e-01 6.89179063e-01
-3.23589772e-01 2.92779412e-02 -1.51566708e+00 1.17350590e+00
-4.82285559e-01 2.97044702e-02 3.25326174e-01 -9.57925200e-01
5.29014170e-01 1.25183597e-01 7.73247659e-01 -6.85276330e-01
2.17057660e-01 1.33573472e-01 -6.08775765e-02 4.86039370e-03
2.21096337e-01 -3.99082303e-01 -2.75919229e-01 7.31824040e-01
-1.97206944e-01 1.35443788e-02 2.45342135e-01 3.11301742e-03
1.03378487e+00 1.05937846e-01 3.02201867e-01 -3.37058425e-01
6.30158260e-02 -2.01061666e-01 9.16426003e-01 8.79397213e-01
-2.51019359e-01 -4.32732642e-01 1.33603418e+00 -2.43000612e-02
-1.03914070e+00 -7.19591260e-01 -8.01109970e-02 9.00705457e-01
1.71685189e-01 6.34715185e-02 -7.26465225e-01 -6.96497202e-01
1.80464581e-01 1.17265141e+00 -8.66978526e-01 -8.03690255e-02
-3.28477085e-01 -6.37386918e-01 2.12147489e-01 4.38824207e-01
5.36085129e-01 -8.59301627e-01 -1.23829293e+00 2.77442843e-01
2.15774611e-01 -8.27580094e-01 -6.84447289e-01 4.26975787e-01
-6.54937983e-01 -6.64898992e-01 -5.39075494e-01 -1.17432782e-02
6.53052032e-01 -2.81376421e-01 3.06396604e-01 -6.83432400e-01
4.02671210e-02 5.60026765e-01 -2.42511276e-02 -6.44554496e-01
6.16182312e-02 -1.40471905e-01 1.22546278e-01 -3.26140113e-02
-1.00604929e-01 -3.16021472e-01 -8.58638465e-01 1.54148310e-01
-7.36981332e-01 -5.74376322e-02 2.98898190e-01 8.61190259e-01
8.33483815e-01 3.53594571e-01 5.33003151e-01 -5.29786050e-01
6.26083195e-01 -3.06970865e-01 -1.44602966e+00 2.46345356e-01
-7.55133152e-01 5.32006979e-01 6.64830089e-01 -5.22985935e-01
-9.82097149e-01 -2.23004483e-02 3.59608829e-01 -3.86591852e-01
4.70426917e-01 3.06323290e-01 -2.15467989e-01 3.02084684e-01
3.47441077e-01 3.41923416e-01 3.23485732e-02 -3.52465928e-01
1.38219535e-01 2.94541866e-01 1.37768134e-01 -9.92818117e-01
5.36227405e-01 2.90610522e-01 4.77954090e-01 -4.67399359e-01
-8.13815773e-01 1.55166864e-01 5.25682978e-02 -1.48707196e-01
5.04568875e-01 -6.96331143e-01 -1.61348641e+00 -1.34172887e-01
-3.45079035e-01 -8.37244630e-01 -8.32747936e-01 6.56163752e-01
-1.34127641e+00 2.24277869e-01 -2.26489425e-01 -1.66652405e+00
-2.08873704e-01 -1.06812716e+00 3.78395647e-01 2.00145736e-01
2.75357276e-01 -5.02911031e-01 -1.61055818e-01 1.74264252e-01
1.80769980e-01 4.06772435e-01 7.11701512e-01 -2.33617753e-01
-7.34847844e-01 -1.05291158e-01 1.39753312e-01 3.88054907e-01
-2.51681775e-01 -6.53366864e-01 -2.69754648e-01 -7.08783269e-01
3.25978816e-01 -3.70365351e-01 6.21692538e-01 4.25605595e-01
1.12643600e+00 -1.08701754e+00 -4.22927691e-03 3.64154279e-01
1.46655929e+00 8.58626187e-01 2.59523869e-01 4.50356901e-01
5.52361943e-02 3.77662301e-01 1.07494903e+00 1.25941551e+00
2.31953591e-01 7.00591981e-01 8.34992051e-01 6.18454754e-01
7.73682952e-01 -3.20161700e-01 6.51492119e-01 -3.05899084e-01
-8.30508098e-02 -1.20057970e-01 -6.98205769e-01 5.62495530e-01
-1.98190498e+00 -1.01594865e+00 6.72275305e-01 3.05497885e+00
9.22936320e-01 3.66614908e-01 4.30530220e-01 -8.77237618e-02
4.27104145e-01 -1.95257604e-01 -1.16282248e+00 -7.55487859e-01
2.15787306e-01 1.55135274e-01 1.10889983e+00 7.82251120e-01
-7.15857267e-01 5.92420757e-01 4.26966238e+00 9.65340376e-01
-8.70973945e-01 1.01178614e-02 5.92146039e-01 -7.99578965e-01
-6.68350309e-02 2.04070777e-01 -9.48871374e-01 8.30895305e-01
1.13571489e+00 -3.65751505e-01 9.78700697e-01 1.04408014e+00
4.11562264e-01 -5.73218405e-01 -1.02196181e+00 5.37837327e-01
-5.99356055e-01 -8.80505085e-01 -7.83004642e-01 5.34442067e-01
6.37472570e-01 -3.00797909e-01 2.71143526e-01 3.63098055e-01
6.15235507e-01 -8.25649083e-01 8.84995520e-01 3.51171404e-01
8.18732023e-01 -1.40425861e+00 3.49817187e-01 7.35806942e-01
-9.90789592e-01 -9.37904000e-01 -1.27202883e-01 -8.63668919e-02
2.66931683e-01 2.66407460e-01 -5.11318445e-01 2.40280643e-01
4.15470392e-01 -2.06264347e-01 3.40682745e-01 8.98589253e-01
-2.78031021e-01 3.93849462e-01 -5.94961524e-01 -2.33034372e-01
6.15797520e-01 -4.10007924e-01 6.50211275e-01 6.26521468e-01
3.94341350e-01 2.82757789e-01 5.92313409e-01 6.69616222e-01
1.80464312e-01 -6.06949218e-02 -2.25016579e-01 -3.22799176e-01
5.78176141e-01 6.70710981e-01 -5.55952609e-01 -5.75487688e-02
4.24871631e-02 6.67628467e-01 1.87876403e-01 3.30658197e-01
-9.15504456e-01 -4.07141566e-01 7.40932286e-01 -4.70663905e-02
5.11291564e-01 -3.38212430e-01 -2.77957588e-01 -8.60277116e-01
1.89128503e-01 -4.48278546e-01 6.34439349e-01 -1.09334085e-02
-6.92667305e-01 4.27183397e-02 2.19865695e-01 -8.56246650e-01
-5.71056187e-01 -1.39373958e-01 -2.02888474e-01 7.90228069e-01
-1.23461437e+00 -3.45893502e-01 4.33547676e-01 5.47158420e-01
3.36699307e-01 -6.71608374e-02 5.02706766e-01 -2.21509919e-01
-7.81616807e-01 4.89251971e-01 5.82257986e-01 -3.23235691e-01
-6.97954893e-02 -1.28978050e+00 -3.15687269e-01 7.13960648e-01
-6.51491523e-01 3.07001561e-01 9.85208869e-01 -5.45082986e-01
-1.50093842e+00 -7.85379529e-01 2.31133789e-01 1.33333102e-01
6.82510555e-01 -3.11131150e-01 -4.13240641e-01 7.14797258e-01
-2.34545380e-01 -1.19562196e-02 3.48043859e-01 -1.90816924e-01
1.01456925e-01 -2.52116680e-01 -1.37275195e+00 7.16887355e-01
8.10383737e-01 -1.35523006e-01 -9.25720111e-03 2.29622066e-01
6.45634651e-01 -4.14110363e-01 -1.12035489e+00 2.69779742e-01
4.97259349e-01 -3.80656064e-01 4.81214821e-01 -5.99171221e-01
-1.29443973e-01 -5.62183186e-02 -3.64780307e-01 -9.85305667e-01
6.37846962e-02 -1.24632084e+00 -5.75473070e-01 6.30081236e-01
3.91428351e-01 -6.70204103e-01 7.89025366e-01 1.05901349e+00
1.63039342e-01 -1.08215690e+00 -1.40334094e+00 -1.10334611e+00
3.24295580e-01 -1.81047380e-01 4.26768422e-01 4.09586072e-01
2.81120896e-01 -1.83601752e-01 -6.02480769e-01 4.20035198e-02
9.99990046e-01 2.54948199e-01 3.31165999e-01 -6.40791237e-01
-1.02982438e+00 -2.57846892e-01 4.16247904e-01 -9.56172824e-01
-1.78702418e-02 -1.81364268e-01 -3.49892490e-03 -1.17532587e+00
1.97102353e-01 -7.75162876e-01 -3.60592097e-01 5.52618265e-01
9.80035439e-02 -9.47743952e-01 5.73828757e-01 -7.85026923e-02
-6.77167237e-01 8.83574188e-01 1.26084507e+00 1.35121182e-01
-5.27555227e-01 5.04206300e-01 -6.29963875e-01 4.24983293e-01
8.12346160e-01 -2.98401117e-01 -8.57586026e-01 -1.26638442e-01
3.01677644e-01 1.07204998e+00 2.01265886e-02 -3.93240362e-01
-8.45947489e-02 -8.98521125e-01 -2.18101397e-01 -3.56841803e-01
5.02058148e-01 -7.39176631e-01 1.05678424e-01 8.60910177e-01
-6.09208047e-01 -1.03002317e-01 -7.49932900e-02 8.45674336e-01
3.26060206e-01 -3.29840422e-01 1.01035321e+00 -2.01399699e-01
-1.94978938e-01 5.44168532e-01 -3.34762633e-01 -1.05948490e-03
1.36289763e+00 1.82860959e-02 -7.11864159e-02 -7.36570776e-01
-8.87262523e-01 7.78129339e-01 2.63388127e-01 6.09240402e-03
2.44270384e-01 -9.26649749e-01 -2.96474189e-01 -1.43485755e-01
-4.12304670e-01 -5.13650216e-02 4.53378171e-01 8.61581743e-01
5.69092445e-02 5.87295175e-01 -1.09449312e-01 3.81418876e-03
-7.46187747e-01 7.60491908e-01 5.76706111e-01 -6.36638105e-01
-4.10254925e-01 5.69108844e-01 7.32728541e-02 2.93779850e-01
6.36818409e-01 -1.06064856e-01 2.99946278e-01 4.03175391e-02
3.73875260e-01 8.39946210e-01 -3.47027481e-01 4.37502526e-02
2.00158823e-02 -1.12931013e-01 -1.22552171e-01 -7.18289614e-01
1.32976019e+00 -2.19086751e-01 2.98975736e-01 1.77004442e-01
7.44333863e-01 -4.51979786e-01 -2.02215505e+00 -2.98055947e-01
-5.81769869e-02 -3.66161972e-01 9.70132351e-02 -8.47638965e-01
-8.15652490e-01 7.06043303e-01 6.78208113e-01 8.78203213e-02
1.03187275e+00 -3.42580348e-01 4.93782192e-01 3.74453306e-01
7.92970717e-01 -1.61955285e+00 -4.99405079e-02 4.58957821e-01
6.56461060e-01 -5.82175910e-01 -6.36076927e-02 2.07380593e-01
-9.36998963e-01 7.02838004e-01 4.81299937e-01 -1.05960570e-01
3.67741227e-01 1.74140498e-01 -8.79163802e-01 2.73084611e-01
-8.39626133e-01 -3.13935310e-01 -3.77404332e-01 6.86145574e-02
-2.38081291e-01 5.49221575e-01 -6.86274648e-01 6.05669022e-01
-1.62253494e-03 -6.00994797e-03 5.48481166e-01 1.22116148e+00
-8.19028974e-01 -1.14872718e+00 -3.17681640e-01 3.42364788e-01
-6.31710768e-01 2.28648737e-01 2.65260726e-01 6.66451693e-01
-1.28195897e-01 7.65286267e-01 -1.08969018e-01 3.62302184e-01
2.61637539e-01 3.55109759e-02 6.29606605e-01 -2.41184980e-01
-1.45725071e-01 3.28413308e-01 1.34401605e-01 -5.91056466e-01
2.67333835e-01 -8.31006885e-01 -1.42540061e+00 -3.16414446e-01
-1.21275656e-01 4.08756584e-01 5.87967336e-01 9.51370180e-01
2.04318821e-01 1.20421760e-01 9.90428090e-01 -2.22636834e-01
-1.34448612e+00 -6.19689763e-01 -9.58658159e-01 -1.67719409e-01
2.94354886e-01 -6.93728864e-01 -2.78610975e-01 -7.68089831e-01]
|
[4.334485054016113, 2.866194248199463]
|
a20402ba-05ec-4dff-975d-336134848fff
|
learning-to-drive-using-sparse-imitation
|
2205.12128
| null |
https://arxiv.org/abs/2205.12128v1
|
https://arxiv.org/pdf/2205.12128v1.pdf
|
Learning to Drive Using Sparse Imitation Reinforcement Learning
|
In this paper, we propose Sparse Imitation Reinforcement Learning (SIRL), a hybrid end-to-end control policy that combines the sparse expert driving knowledge with reinforcement learning (RL) policy for autonomous driving (AD) task in CARLA simulation environment. The sparse expert is designed based on hand-crafted rules which is suboptimal but provides a risk-averse strategy by enforcing experience for critical scenarios such as pedestrian and vehicle avoidance, and traffic light detection. As it has been demonstrated, training a RL agent from scratch is data-inefficient and time consuming particularly for the urban driving task, due to the complexity of situations stemming from the vast size of state space. Our SIRL strategy provides a solution to solve these problems by fusing the output distribution of the sparse expert policy and the RL policy to generate a composite driving policy. With the guidance of the sparse expert during the early training stage, SIRL strategy accelerates the training process and keeps the RL exploration from causing a catastrophe outcome, and ensures safe exploration. To some extent, the SIRL agent is imitating the driving expert's behavior. At the same time, it continuously gains knowledge during training therefore it keeps making improvement beyond the sparse expert, and can surpass both the sparse expert and a traditional RL agent. We experimentally validate the efficacy of proposed SIRL approach in a complex urban scenario within the CARLA simulator. Besides, we compare the SIRL agent's performance for risk-averse exploration and high learning efficiency with the traditional RL approach. We additionally demonstrate the SIRL agent's generalization ability to transfer the driving skill to unseen environment.
|
['Alper Yilmaz', 'Yuci Han']
|
2022-05-24
| null | null | null | null |
['safe-exploration']
|
['robots']
|
[-3.37048203e-01 3.76869917e-01 -2.70700250e-02 2.37176940e-01
-4.44215238e-01 -4.56385165e-01 5.16700029e-01 -7.04848394e-02
-6.77365363e-01 1.01591575e+00 -2.58139998e-01 -5.59928060e-01
-1.84917465e-01 -7.89858878e-01 -9.24145639e-01 -8.58165205e-01
-3.17227840e-01 3.37908596e-01 3.51696610e-01 -5.64528644e-01
7.18622059e-02 5.01961708e-01 -1.68704081e+00 -5.39895296e-01
1.36309838e+00 6.51072681e-01 6.91335738e-01 5.75901628e-01
2.65539318e-01 9.79527473e-01 -3.45409185e-01 2.48907432e-01
5.08378148e-01 -1.90285504e-01 -1.88664719e-01 -1.59348607e-01
-2.28341073e-01 -4.79970455e-01 -4.80639488e-01 7.59045243e-01
5.34837902e-01 5.61459899e-01 4.41591382e-01 -1.44686508e+00
7.30109662e-02 1.91726461e-01 -4.96043503e-01 2.45828673e-01
5.51853813e-02 8.79620612e-01 3.54279310e-01 -3.98401082e-01
3.74075681e-01 1.14907753e+00 3.25688839e-01 5.30258775e-01
-7.87711918e-01 -7.36283660e-01 4.79628265e-01 4.37641963e-02
-1.32086217e+00 -1.58440500e-01 6.73931301e-01 -3.80804151e-01
9.63454068e-01 -1.84817776e-01 8.74686539e-01 9.83169258e-01
4.57899392e-01 8.30332518e-01 1.31789327e+00 2.21372284e-02
6.90527141e-01 3.36260438e-01 -2.75311977e-01 7.77674377e-01
2.49477074e-01 1.09496605e+00 2.78583961e-03 1.06011681e-01
6.30954504e-01 -1.99379608e-01 2.35077348e-02 -5.09421468e-01
-7.99709082e-01 6.43466473e-01 4.82175857e-01 -2.01395109e-01
-7.74596632e-01 2.58370638e-01 3.31619084e-01 6.37361705e-01
-1.95411310e-01 4.26346898e-01 -2.86069036e-01 -3.12485397e-01
-5.57493746e-01 5.22673249e-01 6.84226871e-01 9.10059631e-01
8.41138482e-01 6.89355135e-01 -1.09025791e-01 4.63288724e-01
2.67335027e-01 7.58063853e-01 3.70711237e-01 -1.03897536e+00
3.61109704e-01 5.19711316e-01 5.72027445e-01 -7.36543715e-01
-3.30662787e-01 -7.52701938e-01 -4.84824091e-01 1.11573863e+00
1.08875632e-01 -6.87107682e-01 -7.22713530e-01 1.82342625e+00
5.55823088e-01 3.69119763e-01 5.29538095e-01 9.07008410e-01
1.51095033e-01 7.73318589e-01 2.33144984e-01 -2.74298370e-01
8.79585862e-01 -7.81017363e-01 -4.86179322e-01 -4.31702167e-01
5.68735957e-01 -1.29076436e-01 9.53732610e-01 4.24960136e-01
-9.87075269e-01 -7.19368160e-01 -1.17946029e+00 6.88461185e-01
-3.58010381e-01 2.32680459e-02 3.29212576e-01 3.60463232e-01
-9.19761658e-01 4.80683148e-01 -5.99199951e-01 -3.00832331e-01
2.78323501e-01 5.27836025e-01 -3.54655609e-02 9.25117135e-02
-1.38990521e+00 1.16974401e+00 4.47777867e-01 -3.00393179e-02
-1.65335941e+00 -6.01897538e-01 -6.88449442e-01 -4.72205058e-02
7.51640737e-01 -4.81980711e-01 1.20951211e+00 -7.49429286e-01
-1.99544501e+00 9.70468521e-02 2.87720174e-01 -7.92957664e-01
7.68481672e-01 -2.10794270e-01 -2.03271061e-01 -4.83239666e-02
1.47222623e-01 7.27128863e-01 8.71958852e-01 -1.41754591e+00
-8.82685721e-01 1.23269409e-01 2.60467678e-01 6.23805881e-01
1.13996476e-01 -6.41527057e-01 1.69769466e-01 -3.85451138e-01
-6.19412065e-01 -1.11726940e+00 -5.18084049e-01 -2.38882869e-01
1.94700167e-01 -2.06548452e-01 1.08130658e+00 -2.31775016e-01
1.11200202e+00 -2.28904748e+00 3.19289640e-02 3.87377053e-01
-1.55651063e-01 5.30099392e-01 -1.35745525e-01 6.00973785e-01
2.72120863e-01 -3.67132813e-01 -1.66855872e-01 1.31088689e-01
-9.02600586e-02 5.46912253e-01 -3.60993594e-01 2.82662570e-01
1.46809131e-01 7.43333161e-01 -1.21586621e+00 -4.24370527e-01
3.69707704e-01 1.31914496e-01 -5.12608349e-01 5.21497488e-01
-3.85737807e-01 6.91578865e-01 -8.16327691e-01 3.88059586e-01
5.34010708e-01 2.47525379e-01 -6.02753162e-02 3.82594258e-01
-5.25716007e-01 -2.47579157e-01 -1.24726725e+00 1.19015169e+00
-8.22301328e-01 2.84598053e-01 4.21729416e-01 -9.82800066e-01
1.08518863e+00 2.16704696e-01 3.23462367e-01 -9.83046055e-01
2.95665145e-01 2.28662401e-01 7.68066719e-02 -6.86183870e-01
2.56585449e-01 -4.67367563e-03 -1.85501069e-01 4.24457550e-01
-1.96931750e-01 -2.73182571e-01 4.64065671e-02 1.59167796e-01
1.02717519e+00 4.23393697e-01 2.46236295e-01 -2.81740725e-01
7.70539224e-01 2.98652112e-01 7.08496630e-01 9.01165843e-01
-5.30398011e-01 -5.06859720e-01 2.23118395e-01 -2.71086454e-01
-8.41412067e-01 -1.08200276e+00 2.57140607e-01 9.18997407e-01
4.51255947e-01 1.97789431e-01 -4.47082967e-01 -7.04054058e-01
3.00264806e-01 1.08804011e+00 -5.16953468e-01 -5.51126361e-01
-7.97095418e-01 -1.94332063e-01 3.74636918e-01 2.83794194e-01
9.30823386e-01 -1.06445181e+00 -1.14769316e+00 4.19247448e-01
4.05688882e-01 -8.85960698e-01 -2.58913904e-01 2.93673843e-01
-5.71538270e-01 -8.96750629e-01 -3.81310523e-01 -6.86889112e-01
6.08929396e-01 2.83521205e-01 6.10824049e-01 6.15767613e-02
1.14361972e-01 3.58708233e-01 -1.28151879e-01 -5.10217309e-01
-6.90175712e-01 -1.62283763e-01 2.32262000e-01 -2.26199925e-01
-2.53328308e-03 -6.01390362e-01 -7.19948232e-01 4.41436827e-01
-5.66071272e-01 -3.34855579e-02 8.30992162e-01 8.26023042e-01
3.28050971e-01 4.96781975e-01 1.13390696e+00 -3.40882808e-01
7.95035303e-01 -7.15294898e-01 -1.04645801e+00 -1.00483403e-01
-8.13531041e-01 2.30518058e-01 1.06087565e+00 -6.60882771e-01
-1.29734242e+00 2.32896969e-01 -5.10739312e-02 -4.78648424e-01
-2.42135957e-01 2.14034945e-01 1.18136138e-01 -2.12267846e-01
5.82002878e-01 4.36900109e-01 4.58811581e-01 -5.23442868e-03
2.39680260e-01 6.20000541e-01 4.25030202e-01 -7.47370422e-01
1.00109291e+00 2.38510415e-01 9.84529555e-02 -6.05688512e-01
-3.47527951e-01 -9.77788046e-02 -1.36675546e-02 -7.54085302e-01
6.03245854e-01 -1.03656912e+00 -1.07797909e+00 3.16936582e-01
-6.58801973e-01 -6.99588716e-01 -4.90404218e-01 5.07144272e-01
-8.46227944e-01 1.16923377e-01 -1.16292022e-01 -1.22811127e+00
-1.03465848e-01 -1.20808756e+00 5.47980726e-01 5.92589974e-01
1.39167562e-01 -8.87551129e-01 2.10540399e-01 -9.38970298e-02
5.07423401e-01 4.24074382e-01 6.83130920e-01 -1.29235283e-01
-7.28848994e-01 -6.35639504e-02 2.53953636e-01 2.13885739e-01
-3.16377401e-01 -2.98097074e-01 -6.53660715e-01 -6.81646168e-01
-6.93185255e-02 -5.12682080e-01 5.65076590e-01 1.96656972e-01
5.02144277e-01 -3.07422340e-01 -3.59684974e-01 1.75175548e-01
1.52632165e+00 7.03544021e-01 3.76152307e-01 6.05067968e-01
2.82037079e-01 6.10147536e-01 1.12135601e+00 4.14744526e-01
4.66879427e-01 3.71705174e-01 6.85499370e-01 -8.05085003e-02
1.06746525e-01 -5.33102036e-01 7.26235926e-01 3.74044746e-01
1.07023209e-01 1.13972411e-01 -6.60183012e-01 5.20177543e-01
-2.06313872e+00 -9.54177737e-01 3.28476906e-01 2.30866385e+00
5.77044725e-01 4.34231907e-01 3.87738466e-01 -5.39030656e-02
3.79828334e-01 -1.73817977e-01 -9.80249405e-01 -7.32740104e-01
7.85857290e-02 -2.26435125e-01 7.79736459e-01 6.03422940e-01
-6.51539385e-01 1.06683433e+00 6.04466534e+00 8.46693039e-01
-1.15465665e+00 -7.16031343e-02 2.24028751e-01 -1.38467196e-02
-6.53614774e-02 4.08373140e-02 -6.06849968e-01 5.22191465e-01
1.00231934e+00 -3.62879992e-01 8.08436930e-01 1.03236306e+00
8.01874518e-01 -5.97063065e-01 -6.95752084e-01 4.83891517e-01
-5.67351520e-01 -9.54048276e-01 -3.94889802e-01 -1.79931242e-02
6.69578731e-01 1.13989167e-01 1.34133011e-01 1.00873375e+00
8.62010598e-01 -8.96096945e-01 7.44760275e-01 3.53000641e-01
3.89669448e-01 -1.11593807e+00 6.22380078e-01 9.20462549e-01
-1.08244133e+00 -7.99610317e-01 -1.19634263e-01 -2.55915523e-01
1.09914035e-01 3.76567282e-02 -7.91414917e-01 4.19370800e-01
3.71192932e-01 3.72366369e-01 -2.56366611e-01 8.27891827e-01
-2.51030535e-01 4.90393549e-01 -3.34680200e-01 -3.55231434e-01
9.02591646e-01 -3.46637815e-01 9.42395627e-01 7.80894279e-01
2.21716821e-01 1.63785607e-01 5.14851868e-01 7.76776731e-01
5.81461549e-01 -9.20184925e-02 -1.05813742e+00 2.24276245e-01
5.27242601e-01 1.01087689e+00 -2.77078897e-01 -3.67560774e-01
-9.97782275e-02 2.90610492e-01 3.83605659e-01 5.13475657e-01
-9.57118273e-01 -3.08436036e-01 6.62066340e-01 7.65192462e-03
3.49419862e-01 -2.14173838e-01 9.99773201e-03 -5.41627645e-01
-2.37868622e-01 -8.60257328e-01 6.79885447e-02 -6.26420379e-01
-7.37281799e-01 5.36690652e-01 2.02362016e-01 -1.32340777e+00
-5.74673831e-01 -2.74594516e-01 -8.12299848e-01 7.59730399e-01
-1.89294434e+00 -5.42658567e-01 -1.72882527e-01 6.61194324e-01
6.51952803e-01 -5.26754737e-01 2.71908611e-01 8.10617283e-02
-7.47220159e-01 3.14831734e-01 4.72602993e-02 -5.15656412e-01
2.63542205e-01 -1.05872095e+00 -1.20640583e-01 6.67248487e-01
-8.31381321e-01 2.19464719e-01 1.11902094e+00 -7.54507124e-01
-1.57976890e+00 -1.15794003e+00 4.43259552e-02 1.56391844e-01
6.46983802e-01 -5.59250303e-02 -7.50609636e-01 2.37285286e-01
3.18524688e-01 -2.77929634e-01 -1.21297389e-01 -4.49974209e-01
2.82616824e-01 -3.31276238e-01 -1.33425570e+00 8.88081551e-01
6.45464122e-01 -1.01131499e-01 -4.53572154e-01 4.54399288e-02
5.18049479e-01 -3.29170942e-01 -4.86991674e-01 3.59130830e-01
3.87569845e-01 -7.58459926e-01 6.69699609e-01 -3.16095442e-01
-9.85349566e-02 -5.97145855e-01 2.47491628e-01 -1.63364804e+00
-2.16385260e-01 -8.80355954e-01 -1.23175792e-01 7.49367714e-01
1.81438506e-01 -8.43052447e-01 6.17072344e-01 1.70963362e-01
-2.24858567e-01 -9.55793321e-01 -1.07627010e+00 -1.14275122e+00
2.53908545e-01 -1.86116979e-01 4.29052711e-01 3.26107621e-01
-9.41788554e-02 1.02490105e-01 -5.20074666e-01 4.66711640e-01
7.03272223e-01 -1.03564262e-01 9.56570566e-01 -7.85043955e-01
-3.95312816e-01 -2.71578968e-01 -1.76538937e-02 -9.30793226e-01
3.36758792e-01 -5.61692715e-01 3.59499067e-01 -1.20369601e+00
-3.16540897e-01 -7.27978826e-01 -2.76205689e-01 3.56464475e-01
-8.47758353e-02 -6.20477617e-01 1.66513398e-01 -1.16036308e-03
-4.90037680e-01 7.05544651e-01 1.53283548e+00 -7.17614964e-02
-5.27288020e-01 1.21049851e-01 -4.23599571e-01 4.87449080e-01
1.03427517e+00 -3.99893731e-01 -1.00403619e+00 -7.32403919e-02
2.88579259e-02 4.63104844e-01 4.20330971e-01 -1.13912654e+00
2.15384543e-01 -5.33416927e-01 -1.11406192e-01 -5.79470515e-01
1.02234833e-01 -1.05963516e+00 9.78199393e-02 9.84986246e-01
-8.00015107e-02 1.68629929e-01 4.74512488e-01 8.72061193e-01
-3.51824798e-02 4.70986683e-03 1.02673507e+00 -2.15017542e-01
-1.00897121e+00 1.26790896e-01 -9.14064229e-01 4.42100987e-02
1.64424539e+00 -3.96319747e-01 -1.05428688e-01 -3.74134213e-01
-5.13714671e-01 1.12614715e+00 2.42513835e-01 3.69671583e-01
6.19673550e-01 -1.05396056e+00 -5.25855124e-01 3.72028142e-01
-2.46281207e-01 -1.02075882e-01 2.90471166e-01 8.04790020e-01
-3.13871294e-01 2.85942882e-01 -5.36716700e-01 -4.52600390e-01
-7.04798281e-01 8.15260768e-01 5.99565566e-01 -3.05967242e-01
-8.23232710e-01 1.62858918e-01 2.53630310e-01 -4.19309497e-01
2.80218780e-01 -6.90560564e-02 -1.88665345e-01 -3.62838298e-01
2.10487753e-01 5.15093267e-01 -3.42087030e-01 -3.34489554e-01
-2.49583423e-01 4.36209500e-01 1.02246426e-01 -2.17498273e-01
1.05275869e+00 -3.34639370e-01 6.80023313e-01 2.24894136e-01
6.75218344e-01 -2.55437225e-01 -1.88117409e+00 8.22404027e-02
-3.33152652e-01 -1.94915906e-01 3.51985186e-01 -9.37913060e-01
-7.95510948e-01 5.39681673e-01 7.45032787e-01 -6.75352514e-02
9.51311648e-01 -5.52309453e-01 7.67925382e-01 5.22016227e-01
5.93291104e-01 -1.58683515e+00 1.65035143e-01 5.41018546e-01
8.41325581e-01 -1.32283044e+00 -2.79734075e-01 1.45523608e-01
-1.01169920e+00 7.22748518e-01 1.04027879e+00 -5.75619102e-01
4.94235396e-01 4.26727951e-01 7.25447461e-02 -1.34770889e-02
-1.07091057e+00 -3.43683809e-01 -3.36256474e-01 6.40340209e-01
-6.09736979e-01 8.81346762e-02 -2.06301317e-01 8.23419243e-02
1.94017887e-02 1.91296309e-01 5.29626787e-01 1.10779071e+00
-8.54079485e-01 -6.99774504e-01 -2.97041982e-01 1.22102492e-01
1.89943865e-01 4.17761117e-01 2.11463168e-01 1.14944875e+00
3.29261214e-01 9.01586056e-01 -1.13312423e-01 -2.80382097e-01
4.81865823e-01 -2.46708870e-01 1.44320026e-01 -2.34862566e-01
-5.42948186e-01 -5.90054020e-02 9.00464281e-02 -5.78134060e-01
1.17998645e-01 -6.67304277e-01 -1.51509058e+00 -2.04363823e-01
3.28177549e-02 3.32370847e-01 5.22949576e-01 9.79800045e-01
3.84316772e-01 6.02965593e-01 1.13165510e+00 -5.86021423e-01
-1.02085006e+00 -5.34754932e-01 -3.39386135e-01 3.26317316e-03
6.22035921e-01 -1.05452263e+00 -4.34660405e-01 -6.87764585e-01]
|
[4.972261428833008, 1.4598803520202637]
|
8edc6ec1-7cf0-467e-9247-3af41b428aa7
|
counterfactual-diagnosis
|
1910.06772
| null |
https://arxiv.org/abs/1910.06772v3
|
https://arxiv.org/pdf/1910.06772v3.pdf
|
Counterfactual diagnosis
|
Machine learning promises to revolutionize clinical decision making and diagnosis. In medical diagnosis a doctor aims to explain a patient's symptoms by determining the diseases \emph{causing} them. However, existing diagnostic algorithms are purely associative, identifying diseases that are strongly correlated with a patients symptoms and medical history. We show that this inability to disentangle correlation from causation can result in sub-optimal or dangerous diagnoses. To overcome this, we reformulate diagnosis as a counterfactual inference task and derive new counterfactual diagnostic algorithms. We show that this approach is closer to the diagnostic reasoning of clinicians and significantly improves the accuracy and safety of the resulting diagnoses. We compare our counterfactual algorithm to the standard Bayesian diagnostic algorithm and a cohort of 44 doctors using a test set of clinical vignettes. While the Bayesian algorithm achieves an accuracy comparable to the average doctor, placing in the top 48% of doctors in our cohort, our counterfactual algorithm places in the top 25% of doctors, achieving expert clinical accuracy. This improvement is achieved simply by changing how we query our model, without requiring any additional model improvements. Our results show that counterfactual reasoning is a vital missing ingredient for applying machine learning to medical diagnosis.
|
['Saurabh Johri', 'Jonathan G. Richens', 'Ciaran M. Lee']
|
2019-10-15
| null | null | null | null |
['counterfactual-inference']
|
['miscellaneous']
|
[ 4.36711282e-01 1.01741278e+00 -5.44996619e-01 -5.04606724e-01
-7.38726795e-01 -3.77111554e-01 6.38215125e-01 4.15911108e-01
-2.66228735e-01 1.13282967e+00 3.94267827e-01 -7.45646596e-01
-7.78868079e-01 -6.23492539e-01 -4.34685588e-01 -6.92241967e-01
-1.77515209e-01 1.31879604e+00 -3.74793172e-01 3.42436880e-01
-3.27146165e-02 2.46488959e-01 -1.10570765e+00 4.71837521e-01
1.03710461e+00 3.55027080e-01 -2.79580235e-01 6.74710572e-01
3.41363400e-01 1.02177298e+00 -4.20380920e-01 -8.29206347e-01
7.09041581e-02 -8.23216140e-01 -9.66278970e-01 -8.41106325e-02
-2.77328752e-02 -4.82980102e-01 -1.67848453e-01 8.26912761e-01
4.91878718e-01 -3.40579718e-01 9.86152470e-01 -1.11872315e+00
-5.31388760e-01 1.12904060e+00 -3.52221191e-01 -3.79398698e-03
5.43538868e-01 2.01452613e-01 1.02240050e+00 9.04386267e-02
6.33162498e-01 1.45690060e+00 6.79870546e-01 8.98288012e-01
-1.55569708e+00 -5.79276443e-01 5.35432622e-02 8.00567344e-02
-1.00448656e+00 -3.28641742e-01 4.10242558e-01 -6.39038026e-01
5.39751470e-01 6.26637220e-01 8.92389059e-01 1.10035253e+00
3.92545044e-01 5.79139829e-01 1.31398225e+00 -3.67259383e-01
5.60446382e-01 1.96359321e-01 -2.39180475e-02 5.33705771e-01
6.66603148e-01 5.90377688e-01 -1.30646214e-01 -8.60461712e-01
5.95735133e-01 2.92906970e-01 -4.54604864e-01 -3.23522091e-01
-1.50532949e+00 9.30573881e-01 2.50676841e-01 1.51240220e-02
-6.01915956e-01 1.64626762e-01 3.69025320e-02 1.71026979e-02
2.34596863e-01 7.86961794e-01 -7.75460780e-01 1.94939628e-01
-9.90379035e-01 8.02353442e-01 1.05913389e+00 1.79382816e-01
-1.61939651e-01 -4.74109828e-01 -4.04635191e-01 4.72529143e-01
1.24099895e-01 7.29808569e-01 5.47376513e-01 -1.25757217e+00
-9.95419845e-02 4.55263346e-01 5.26478767e-01 -6.61179841e-01
-4.70271707e-01 -5.67079484e-01 -7.80559182e-01 1.43635944e-02
6.14023507e-01 -3.58180881e-01 -8.35064709e-01 1.73667693e+00
2.94401616e-01 2.15051055e-01 1.56670153e-01 6.75978959e-01
1.32708132e-01 1.38182998e-01 3.21395904e-01 -7.33900309e-01
1.39408517e+00 -2.63988934e-02 -8.04319799e-01 -5.63266799e-02
1.01755130e+00 -5.55896997e-01 4.16172385e-01 6.13105237e-01
-1.12117267e+00 1.79710776e-01 -6.81747437e-01 4.90887761e-01
2.17221066e-01 -3.84851873e-01 1.00658584e+00 6.40431166e-01
-8.28155875e-01 9.27503288e-01 -6.12118840e-01 -1.62474558e-01
6.60681009e-01 4.24620152e-01 -2.20545735e-02 -2.21319467e-01
-1.32353997e+00 1.04280460e+00 3.69974762e-01 -4.82370049e-01
-7.79207349e-01 -1.25590348e+00 -5.79867899e-01 3.09755057e-02
5.11600614e-01 -1.53216159e+00 1.53196394e+00 -7.06709146e-01
-7.66715765e-01 1.08635533e+00 -2.50559479e-01 -8.05816412e-01
9.29033041e-01 1.22803666e-01 -3.73595029e-01 2.35899054e-02
2.39695057e-01 6.97012544e-01 3.27683747e-01 -1.23180199e+00
-9.40352440e-01 -5.39923906e-01 -4.81667183e-02 4.74723764e-02
3.97473693e-01 -3.09321642e-01 3.36982727e-01 -4.95866060e-01
5.56380413e-02 -9.68367457e-01 -6.77438021e-01 -2.24529967e-01
-6.42995179e-01 -2.19267085e-01 1.64142132e-01 -2.93173343e-01
1.10161412e+00 -1.61498260e+00 -2.62218863e-01 9.88290459e-02
6.22698426e-01 -1.31487340e-01 5.08612812e-01 6.33428842e-02
-5.66267371e-01 2.06237882e-01 -4.52611566e-01 1.20767631e-01
-6.68927794e-03 3.78356308e-01 -4.09324795e-01 4.65733230e-01
1.07999854e-01 8.97584379e-01 -1.17583179e+00 -5.41859210e-01
1.55396417e-01 1.21563226e-01 -8.64602983e-01 -3.51617336e-02
-1.99944228e-01 2.92739570e-01 -3.70300412e-01 3.42407197e-01
4.44401771e-01 -5.73463559e-01 7.49652386e-01 2.96315610e-01
4.40382302e-01 6.55472875e-01 -9.50487137e-01 1.17782044e+00
-2.01337382e-01 -5.50427139e-02 -2.64059812e-01 -1.07471097e+00
2.93980181e-01 8.18489611e-01 7.11587608e-01 -1.14545099e-01
-1.09636700e-02 3.71910125e-01 6.66280806e-01 -7.44376779e-01
-4.97572958e-01 -9.87994194e-01 3.22388043e-03 7.49565542e-01
-4.96644944e-01 -2.44483709e-01 -2.01016173e-01 3.19996744e-01
1.20818448e+00 -2.99831241e-01 7.23945260e-01 -3.52147579e-01
-1.38559118e-01 4.17024463e-01 7.88232327e-01 1.01022363e+00
-1.79162428e-01 2.95205504e-01 9.00875688e-01 -6.22004569e-01
-7.94134974e-01 -1.45961213e+00 -5.09199142e-01 2.52307117e-01
-2.94977367e-01 1.43953085e-01 -6.23716891e-01 -9.43092525e-01
3.81773978e-01 1.28527582e+00 -9.43509459e-01 -5.06960869e-01
-1.58848107e-01 -1.43829012e+00 4.09249932e-01 3.77925277e-01
-1.11554533e-01 -6.13445520e-01 -7.60462523e-01 1.49071246e-01
-3.55891615e-01 -4.92399067e-01 -1.01201333e-01 -1.46479189e-01
-9.84199822e-01 -1.46052599e+00 -5.90938866e-01 -6.45714207e-03
6.23305976e-01 -3.58494461e-01 1.35662866e+00 -1.27533749e-01
-4.54764366e-01 -9.07633081e-02 1.02726325e-01 -6.35442495e-01
-7.96674430e-01 -6.41414106e-01 2.80793220e-01 -3.77169520e-01
6.71968043e-01 -6.14806354e-01 -9.80429113e-01 -1.83313832e-01
-6.76801801e-01 2.23361865e-01 6.70986354e-01 1.09232497e+00
1.24394581e-01 1.44032408e-02 6.33756161e-01 -1.62027419e+00
5.27393579e-01 -6.36479974e-01 -2.74228305e-01 1.51040345e-01
-1.21219456e+00 3.43858629e-01 2.55065739e-01 -4.25238162e-01
-1.14620376e+00 1.10499389e-01 5.80997989e-02 -1.83115397e-02
-3.49768072e-01 4.27016377e-01 1.15255155e-01 8.79424512e-01
8.57315838e-01 -2.47432873e-01 -1.05044849e-01 -3.15720528e-01
3.09943259e-01 5.48664570e-01 4.71035928e-01 -3.27446878e-01
2.72205502e-01 6.43806219e-01 2.07274348e-01 1.82045862e-01
-1.22343659e+00 -4.78212945e-02 -2.54485130e-01 2.19750866e-01
8.00252378e-01 -6.80188715e-01 -1.17327154e+00 -3.58196169e-01
-9.62415576e-01 -1.49089947e-01 -5.17728031e-01 9.02855098e-01
-6.36322439e-01 -1.28224403e-01 -3.13636243e-01 -9.58889723e-01
3.47382203e-02 -1.06222785e+00 9.05073285e-01 -3.25682610e-01
-1.02133071e+00 -1.35890830e+00 2.42388755e-01 5.23185253e-01
-1.82929114e-01 5.93599141e-01 1.39139748e+00 -8.42673898e-01
-2.63104558e-01 -3.67762357e-01 -1.30549565e-01 -3.70019853e-01
4.09717977e-01 -2.18335584e-01 -1.00953197e+00 9.28671509e-02
2.21984878e-01 8.63829255e-02 9.27504659e-01 9.92184162e-01
9.80447054e-01 -7.14524746e-01 -9.80288863e-01 2.25744352e-01
1.26628280e+00 6.24193311e-01 3.73806208e-01 -1.71870738e-01
4.08435494e-01 1.08092618e+00 3.81501377e-01 4.61200416e-01
4.81014758e-01 5.42547226e-01 2.63431907e-01 -1.00653052e-01
1.32403627e-01 -2.19907671e-01 -2.13718608e-01 -5.03971986e-03
-2.35207807e-02 -2.50446610e-02 -1.21645737e+00 8.97899926e-01
-2.01368618e+00 -1.07352293e+00 -2.27203488e-01 2.28493285e+00
1.29339898e+00 1.93718672e-01 1.25557274e-01 3.03181827e-01
4.78197366e-01 -6.00917995e-01 -5.82962573e-01 -4.71349597e-01
1.91950813e-01 1.00496799e-01 6.10321522e-01 6.98625684e-01
-7.97200918e-01 2.42163152e-01 7.72615623e+00 5.98156452e-01
-7.60351300e-01 8.46809670e-02 1.10764408e+00 -5.19008398e-01
-7.62777627e-01 2.50297226e-02 -2.75163561e-01 4.70783204e-01
1.24996722e+00 -4.32866305e-01 5.59378937e-02 6.70683861e-01
6.75721228e-01 -2.51788169e-01 -1.81075943e+00 8.07849348e-01
-2.81193942e-01 -1.50538087e+00 1.83453202e-01 2.72178322e-01
9.97581363e-01 -5.52475512e-01 1.41918749e-01 -1.85446978e-01
1.27986085e+00 -1.51921487e+00 3.65005910e-01 5.95595777e-01
8.12207580e-01 -6.48216903e-01 8.97703230e-01 4.06966984e-01
5.86393178e-02 -2.14279324e-01 1.37672782e-01 -3.56557012e-01
1.49720237e-01 1.15419185e+00 -1.58273649e+00 3.82824212e-01
3.26357126e-01 3.95797104e-01 3.84470113e-02 8.03823650e-01
-2.27533892e-01 7.46792078e-01 -2.82799229e-02 2.40570486e-01
-1.38471931e-01 2.99973786e-01 4.13227439e-01 1.07170355e+00
1.10224657e-01 3.22705686e-01 -2.05004230e-01 1.05905700e+00
3.09645627e-02 -3.08779299e-01 -8.21135044e-01 1.01444356e-01
3.47805768e-01 5.98298013e-01 -5.13813436e-01 -8.63985896e-01
1.38753042e-01 6.82631493e-01 -2.00620055e-01 1.27951622e-01
-6.83166742e-01 1.38983190e-01 7.30738938e-01 2.62544245e-01
-2.88617939e-01 8.84013236e-01 -8.21426928e-01 -1.03932071e+00
-3.75419766e-01 -8.92389655e-01 6.92919314e-01 -6.63886726e-01
-1.26000881e+00 -2.45992895e-02 2.81927049e-01 -7.82109976e-01
-9.39679801e-01 -5.20265579e-01 -2.33831689e-01 7.59722590e-01
-1.04498780e+00 -5.92096925e-01 4.43242371e-01 1.88596115e-01
1.64939374e-01 3.56766939e-01 1.00562441e+00 3.22141387e-02
-2.53863275e-01 3.36263925e-01 1.42021598e-02 -1.17725074e-01
8.43213737e-01 -1.69198179e+00 1.88509095e-02 3.91535729e-01
-2.53857702e-01 7.51135707e-01 1.11338520e+00 -8.06764841e-01
-7.40781248e-01 -1.01258385e+00 1.43068898e+00 -8.01835656e-01
4.04434264e-01 2.74244845e-01 -6.79213047e-01 7.67593563e-01
-1.11760274e-01 -4.01178658e-01 9.32740986e-01 3.08196396e-01
-2.56924480e-01 2.84002423e-02 -1.56435061e+00 8.52280080e-01
1.10385990e+00 -1.98540106e-01 -1.11006927e+00 7.02467620e-01
6.35593116e-01 -3.72188240e-02 -7.82884717e-01 6.18340015e-01
6.87035024e-01 -1.02319646e+00 1.03593802e+00 -1.19219649e+00
9.83469069e-01 2.97928485e-03 1.67111292e-01 -1.51065147e+00
-3.28990996e-01 -4.67978477e-01 2.05227248e-02 5.12843370e-01
7.07469642e-01 -7.79667437e-01 6.81085467e-01 1.01399779e+00
3.42354596e-01 -9.35928285e-01 -7.60939419e-01 -3.64854842e-01
1.64297178e-01 -4.45688248e-01 6.43439591e-01 1.44854164e+00
2.97313035e-01 4.62680534e-02 -1.47624567e-01 1.56802431e-01
9.11021054e-01 4.68998790e-01 7.94170126e-02 -1.37997842e+00
-8.12302709e-01 -4.72496897e-01 -2.12347861e-02 -3.62887323e-01
-1.92048535e-01 -7.94243753e-01 -1.63262859e-01 -1.65796375e+00
8.44712317e-01 -4.77481753e-01 -2.70698160e-01 4.43451554e-01
-5.89690983e-01 -1.75074991e-02 -3.47429395e-01 7.21580163e-02
-8.89490843e-02 -2.23954305e-01 1.24555385e+00 -7.56069645e-02
-1.50490925e-02 2.37758681e-01 -1.17031574e+00 1.13764477e+00
6.37838185e-01 -8.62289667e-01 -5.24189651e-01 7.21245408e-02
3.80195647e-01 5.16959906e-01 7.46776521e-01 -2.77536780e-01
-1.31359994e-01 -4.91332442e-01 4.88132894e-01 -2.25591645e-01
1.06986113e-01 -5.18346369e-01 6.26182616e-01 1.48645234e+00
-7.29428530e-01 -1.99465781e-01 -1.07305020e-01 6.36279285e-01
2.58601636e-01 -1.34866834e-01 6.87626600e-01 -4.75925356e-01
9.21891108e-02 -2.86774486e-01 -4.53057438e-01 2.02757362e-02
1.02883554e+00 1.46277279e-01 -2.81557620e-01 -3.27094287e-01
-1.12922275e+00 1.62102476e-01 2.81190544e-01 -2.40281180e-01
4.62299824e-01 -8.85915220e-01 -9.93005633e-01 -4.12800133e-01
-7.00322986e-02 -1.51155621e-01 3.38879913e-01 1.10401464e+00
-4.40376312e-01 8.02311122e-01 2.79064327e-01 -3.77238125e-01
-1.19003975e+00 7.81826019e-01 5.30197740e-01 -6.83735192e-01
-3.12456995e-01 7.60802448e-01 5.70568740e-01 -2.24430442e-01
-9.71945748e-02 -3.44501704e-01 1.46200344e-01 -1.03413790e-01
5.74918389e-01 4.17521894e-01 -2.11006120e-01 1.70002937e-01
-4.44123298e-01 4.12395149e-02 -2.30611995e-01 -3.47200543e-01
1.23647511e+00 1.39294118e-02 -1.20146036e-01 3.59836102e-01
4.55400079e-01 2.07394622e-02 -8.11027408e-01 1.16625339e-01
4.11468595e-02 -4.15927917e-01 -1.55534903e-02 -1.48251808e+00
-5.51158190e-01 5.44350445e-01 4.13738310e-01 3.20699513e-01
9.92667496e-01 2.78611600e-01 3.75200659e-01 1.67964607e-01
7.05930442e-02 -5.22275448e-01 -5.52883029e-01 -4.98562068e-01
6.33594751e-01 -1.39756310e+00 2.91230381e-01 -4.71900105e-01
-6.09376848e-01 4.24718559e-01 -6.79266602e-02 -3.59503143e-02
6.31235182e-01 2.42468730e-01 2.22046569e-01 -3.62641096e-01
-1.34392715e+00 4.47955094e-02 1.46531135e-01 6.46950901e-01
5.31507373e-01 7.70488083e-01 -5.97779632e-01 6.78623974e-01
-4.24882591e-01 3.28775883e-01 5.98892510e-01 4.93859559e-01
4.05297391e-02 -1.12894046e+00 -5.90541065e-01 9.41254795e-01
-9.30426538e-01 -1.96394295e-01 -3.52679312e-01 7.36033678e-01
5.00514865e-01 9.06805933e-01 6.91637844e-02 -1.18223600e-01
1.47346154e-01 2.71194458e-01 5.98603785e-01 -7.05448449e-01
-2.08469272e-01 4.18671407e-02 2.00111732e-01 -4.25097913e-01
-5.09438455e-01 -1.01719046e+00 -1.19685268e+00 -4.09064770e-01
-7.48228561e-03 2.90439785e-01 2.24836290e-01 1.08241034e+00
3.23400199e-01 7.35516846e-01 2.22672105e-01 5.45114987e-02
-9.18838918e-01 -6.30205035e-01 -5.18903911e-01 5.50934613e-01
5.78875601e-01 -5.91156483e-01 -3.60864252e-01 3.36391330e-01]
|
[8.38121223449707, 5.543274402618408]
|
8068bc49-9897-4483-a83d-cef72dd7f569
|
going-the-extra-mile-in-face-image-quality
|
2207.04904
| null |
https://arxiv.org/abs/2207.04904v1
|
https://arxiv.org/pdf/2207.04904v1.pdf
|
Going the Extra Mile in Face Image Quality Assessment: A Novel Database and Model
|
Computer vision models for image quality assessment (IQA) predict the subjective effect of generic image degradation, such as artefacts, blurs, bad exposure, or colors. The scarcity of face images in existing IQA datasets (below 10\%) is limiting the precision of IQA required for accurately filtering low-quality face images or guiding CV models for face image processing, such as super-resolution, image enhancement, and generation. In this paper, we first introduce the largest annotated IQA database to date that contains 20,000 human faces (an order of magnitude larger than all existing rated datasets of faces), of diverse individuals, in highly varied circumstances, quality levels, and distortion types. Based on the database, we further propose a novel deep learning model, which re-purposes generative prior features for predicting subjective face quality. By exploiting rich statistics encoded in well-trained generative models, we obtain generative prior information of the images and serve them as latent references to facilitate the blind IQA task. Experimental results demonstrate the superior prediction accuracy of the proposed model on the face IQA task.
|
['Dietmar Saupe', 'Yanning Zhang', 'Hantao Liu', 'Yu Zhu', 'Jinqiu Sun', 'Oliver Wiedemann', 'Vlad Hosu', 'Hanhe Lin', 'Shaolin Su']
|
2022-07-11
| null | null | null | null |
['face-image-quality', 'face-image-quality-assessment']
|
['computer-vision', 'computer-vision']
|
[ 3.46852005e-01 -1.86459839e-01 2.36667201e-01 -5.83170533e-01
-7.25775719e-01 -2.88811922e-01 4.56608027e-01 -7.03119516e-01
1.26213029e-01 5.40302396e-01 5.24349034e-01 1.89081714e-01
-2.44374365e-01 -6.30369067e-01 -4.70648825e-01 -6.57853484e-01
1.73321635e-01 -3.10091749e-02 -5.31027615e-01 7.85273612e-02
2.92182416e-01 3.97199601e-01 -2.10626054e+00 3.98801297e-01
1.14735687e+00 1.35416210e+00 5.61818890e-02 4.88356113e-01
2.86429584e-01 5.66616714e-01 -7.92618692e-01 -9.21257854e-01
2.28221431e-01 -5.27089357e-01 -4.01697844e-01 4.07925934e-01
9.14657772e-01 -7.58437693e-01 -5.37004769e-01 1.30148280e+00
9.03011799e-01 -1.66885942e-01 8.21699500e-01 -1.04558730e+00
-1.55191886e+00 6.39539585e-02 -4.21057135e-01 3.04290354e-01
4.08911020e-01 6.42987430e-01 7.51276076e-01 -1.14893639e+00
3.02028388e-01 1.66171527e+00 5.06308258e-01 8.80397856e-01
-1.01441085e+00 -7.73721218e-01 -2.11393252e-01 4.44803298e-01
-1.26114857e+00 -1.21289718e+00 6.53133452e-01 -4.50766474e-01
5.86105049e-01 4.57451046e-02 5.68512022e-01 1.09793568e+00
1.40451774e-01 3.41999084e-01 1.38359237e+00 -1.07218802e-01
1.51549354e-01 -1.42182574e-01 -2.59733349e-01 6.76016450e-01
1.25438750e-01 3.62927288e-01 -7.06007361e-01 -5.80586568e-02
7.67613828e-01 -3.14954937e-01 -4.54568774e-01 4.43562567e-02
-6.86017811e-01 4.11627263e-01 4.02014345e-01 -8.76472145e-02
-3.92722249e-01 -6.99394494e-02 3.81603232e-03 1.61394119e-01
6.52903497e-01 1.79668039e-01 -7.11891577e-02 1.90540567e-01
-9.28348184e-01 1.61253601e-01 2.04800650e-01 7.84380019e-01
5.13970554e-01 4.15010959e-01 -7.38212466e-01 1.23612177e+00
5.27670681e-01 9.00133908e-01 2.54087925e-01 -1.20692134e+00
1.27772465e-01 3.85430872e-01 2.91452587e-01 -1.03884840e+00
-1.01812035e-01 -5.20403087e-01 -9.72085595e-01 4.13188368e-01
2.36121416e-01 2.83888996e-01 -1.14271438e+00 1.68212187e+00
4.60858569e-02 2.84055144e-01 -1.48701146e-01 9.58395898e-01
1.01554883e+00 4.23777848e-01 -2.28956360e-02 -6.61503673e-01
1.50924289e+00 -4.51758057e-01 -9.92958009e-01 -2.21433356e-01
-4.78561968e-01 -8.59199882e-01 1.08229554e+00 6.39557600e-01
-1.35112226e+00 -9.17814851e-01 -9.56368089e-01 -4.15840596e-02
6.97499812e-02 4.23020035e-01 5.55937469e-01 1.08237338e+00
-1.59084642e+00 5.38889647e-01 -2.66053170e-01 1.09907679e-01
1.18081307e+00 1.70172289e-01 -2.56499529e-01 -5.58774948e-01
-9.34310615e-01 7.58082926e-01 -3.75167042e-01 4.16582614e-01
-1.31321943e+00 -8.92456591e-01 -8.32494438e-01 -1.60882156e-02
-7.71697164e-02 -8.10154736e-01 8.88759792e-01 -9.61438835e-01
-1.57613957e+00 9.70676541e-01 -5.44936121e-01 -8.48244056e-02
2.34863847e-01 -3.23739827e-01 -7.75792301e-01 6.48434460e-02
-6.96491599e-02 4.66287106e-01 1.56284750e+00 -1.39286816e+00
-1.28014326e-01 -8.09560061e-01 -1.19154230e-01 1.16615176e-01
-3.29249978e-01 3.72816116e-01 -5.14326692e-01 -5.73551595e-01
-1.68357179e-01 -3.68162930e-01 3.11532497e-01 1.59948274e-01
-1.31106645e-01 -1.70469180e-01 5.39903939e-01 -1.06868505e+00
1.09601092e+00 -2.17611361e+00 2.04453096e-02 -9.18839201e-02
3.40215892e-01 4.80090201e-01 -5.43745100e-01 -2.69395888e-01
-7.28889108e-02 3.51391360e-02 -1.68591723e-01 -4.57213283e-01
1.31064683e-01 -1.86137766e-01 -1.85676754e-01 5.67723751e-01
5.55528581e-01 9.11705077e-01 -6.55824482e-01 -3.21238697e-01
1.34218976e-01 8.96916389e-01 -5.43141782e-01 5.63950419e-01
-6.13809936e-03 4.62837845e-01 2.01348420e-02 9.88024652e-01
1.17612100e+00 -9.15841386e-02 -1.58970192e-01 -6.66489780e-01
3.51018041e-01 8.89050737e-02 -9.31073666e-01 1.45248401e+00
-5.05309999e-01 5.71675539e-01 1.59696996e-01 -5.16796112e-01
9.78530526e-01 3.20335895e-01 2.31596798e-01 -1.02643931e+00
2.32954890e-01 -1.90001018e-02 -7.98585564e-02 -4.77848977e-01
7.12012798e-02 2.62805447e-02 4.63653862e-01 2.72738725e-01
4.26847160e-01 -6.54373690e-02 1.69393659e-01 -1.21827297e-01
6.93020582e-01 -1.31415293e-01 -1.63524169e-02 -9.03355926e-02
6.81776702e-01 -1.03700590e+00 5.26392639e-01 4.11623836e-01
-5.75151384e-01 7.73036420e-01 2.98940629e-01 -2.91457683e-01
-1.16347408e+00 -1.20726788e+00 -4.73204941e-01 9.71888542e-01
-2.45757047e-02 -2.18744770e-01 -9.67359424e-01 -2.76427984e-01
-9.48824808e-02 2.63591319e-01 -5.72153211e-01 -3.89527261e-01
-1.52714118e-01 -1.16712201e+00 3.18128556e-01 3.71944785e-01
8.47801030e-01 -1.04946446e+00 6.97243363e-02 -2.48825416e-01
-2.39757881e-01 -1.01395154e+00 -5.85171163e-01 -7.94846714e-01
-5.12995303e-01 -1.13611650e+00 -6.81100070e-01 -5.81751108e-01
7.08789468e-01 1.31982237e-01 1.44627953e+00 1.53675735e-01
-6.52759075e-01 4.70119536e-01 -4.43886556e-02 -1.88160151e-01
-4.19490367e-01 -9.77202535e-01 2.58887470e-01 4.21952754e-01
1.64957151e-01 -6.11636639e-01 -1.08170462e+00 2.53881961e-01
-5.76115906e-01 -3.04432929e-01 7.06671834e-01 9.64582682e-01
4.06078398e-01 3.53818417e-01 8.42809081e-01 -4.63703513e-01
9.01244581e-01 -1.71739891e-01 -5.42282403e-01 2.38880903e-01
-8.45170617e-01 -2.23858193e-01 2.72300154e-01 -3.22292715e-01
-1.59560728e+00 -3.66533309e-01 -1.83049977e-01 -4.03050244e-01
-1.92873821e-01 -1.84806914e-03 -7.90360034e-01 -3.25219810e-01
7.33843565e-01 3.19350421e-01 -5.58988936e-02 -3.70880127e-01
4.34078783e-01 7.41347373e-01 9.05842304e-01 -2.27502123e-01
7.52462327e-01 2.74049252e-01 -2.29905788e-02 -5.26170135e-01
-7.39599049e-01 -1.55951222e-02 -3.45091313e-01 -6.04306877e-01
5.59404731e-01 -1.24949324e+00 -8.57978284e-01 9.38207030e-01
-9.18708086e-01 5.52353822e-03 1.30686164e-01 1.22408800e-01
-5.78794599e-01 3.54809195e-01 -6.88300729e-01 -1.08000851e+00
-5.75673342e-01 -1.22574210e+00 1.12034750e+00 4.11494106e-01
3.89262587e-01 -4.35081184e-01 -2.24867210e-01 8.76242220e-01
6.87065840e-01 -2.73029637e-02 8.21705759e-01 4.03076887e-01
-6.20627642e-01 5.68201877e-02 -5.50820589e-01 7.51215696e-01
3.79390091e-01 -4.99901511e-02 -1.46905351e+00 -4.31270331e-01
9.15215090e-02 -4.70998198e-01 9.32868063e-01 6.24200761e-01
1.46963584e+00 -5.13074219e-01 1.91308528e-01 8.04928601e-01
1.19646525e+00 1.14876397e-01 1.10586941e+00 -3.22032094e-01
4.50225770e-01 5.08057296e-01 4.02218044e-01 6.67923987e-01
1.98834181e-01 5.37643075e-01 5.60532987e-01 -4.36805822e-02
-7.27321744e-01 -4.74914536e-02 5.30800819e-01 5.22923231e-01
-3.21599245e-01 -2.07809269e-01 -5.01553774e-01 4.84455526e-01
-1.00367248e+00 -1.09574425e+00 3.32096308e-01 2.03277540e+00
8.93811285e-01 -3.11809540e-01 -1.16465241e-01 1.59863114e-01
8.39599133e-01 1.36402145e-01 -8.02330434e-01 4.75408770e-02
-3.02723616e-01 3.49881619e-01 -9.65368077e-02 2.51173198e-01
-1.11901653e+00 6.05452061e-01 7.04811239e+00 6.91652894e-01
-7.91570961e-01 5.63862436e-02 1.11565542e+00 -1.48604199e-01
-3.51933450e-01 -5.69283068e-01 -5.29645205e-01 6.97266936e-01
9.38400686e-01 -1.18445307e-01 9.79027092e-01 6.41627669e-01
3.31177205e-01 2.41551861e-01 -9.35487688e-01 1.52634501e+00
5.86540818e-01 -9.25943673e-01 1.96251050e-01 3.63243856e-02
8.63244236e-01 -4.52016801e-01 9.83966887e-01 4.77051027e-02
1.42933965e-01 -1.45590687e+00 6.29554749e-01 8.95996451e-01
1.28675699e+00 -7.83630788e-01 6.54368579e-01 -2.91714579e-01
-8.44136119e-01 -3.74869257e-01 -6.80192292e-01 2.92052329e-01
-1.68542862e-01 7.79953718e-01 -4.63402420e-01 3.47037166e-01
9.64622378e-01 7.25971580e-01 -8.56609285e-01 9.41714406e-01
-1.24959745e-01 5.31539500e-01 3.18968892e-01 6.59645796e-01
-7.01917470e-01 -1.98310018e-01 3.92597884e-01 7.51022816e-01
5.26377439e-01 2.31280759e-01 -5.32292306e-01 1.14180684e+00
-5.22128344e-01 -1.04824871e-01 -2.84444749e-01 -9.53753069e-02
5.23985386e-01 1.25833476e+00 -6.88356087e-02 -3.60378325e-02
-4.55227077e-01 1.01429045e+00 1.32964417e-01 5.66599071e-01
-7.20945597e-01 8.27616677e-02 1.09994483e+00 5.52574508e-02
1.15257300e-01 1.20862164e-01 -1.68464482e-01 -1.07112145e+00
2.45420963e-01 -1.11323345e+00 1.33473381e-01 -1.11646628e+00
-1.70785856e+00 8.38000059e-01 -4.30148631e-01 -1.01718068e+00
-1.07791901e-01 -8.04989457e-01 -5.11095583e-01 1.20466113e+00
-1.71780205e+00 -1.29540622e+00 -6.49033606e-01 8.55864584e-01
4.09978300e-01 -4.33307618e-01 5.71188986e-01 4.61020589e-01
-5.24218202e-01 8.34123254e-01 -1.09608294e-02 6.22061379e-02
8.92478108e-01 -9.47365522e-01 3.23699743e-01 1.02649403e+00
-1.44601628e-01 5.39138317e-01 5.00521600e-01 -5.00506878e-01
-1.47566974e+00 -1.23527253e+00 3.83978277e-01 -5.12908936e-01
9.99306142e-02 -3.43746424e-01 -8.86032045e-01 1.42073169e-01
6.97131902e-02 2.67364651e-01 6.60070539e-01 4.67054322e-02
-6.19947910e-01 -5.15370429e-01 -1.43762147e+00 2.09431291e-01
1.31161773e+00 -8.70660722e-01 -1.70972824e-01 7.94629380e-02
4.62354213e-01 -9.19661969e-02 -9.60540593e-01 7.83651233e-01
6.08305693e-01 -1.25648832e+00 1.27455997e+00 -3.16112459e-01
4.50765371e-01 -2.68658191e-01 -8.05559009e-02 -1.38154447e+00
-6.71539843e-01 -3.77189338e-01 -2.97454596e-01 1.48223603e+00
6.76520541e-02 -3.53164613e-01 4.21506435e-01 5.40182054e-01
-1.03896886e-01 -3.55038583e-01 -8.19364488e-01 -5.35500765e-01
-2.93718845e-01 -2.25090340e-01 1.01291800e+00 5.71766794e-01
-4.54676360e-01 -9.95797068e-02 -6.37919486e-01 3.30500752e-01
1.05931461e+00 7.04417303e-02 4.86640245e-01 -1.25000942e+00
-8.89351666e-02 -5.39516568e-01 -6.01044118e-01 -5.69518268e-01
2.33842820e-01 -5.82980156e-01 6.33079335e-02 -1.41376305e+00
4.79876846e-01 8.07367824e-03 -2.65539438e-01 1.00713484e-01
-6.52575850e-01 6.44803405e-01 -1.07943222e-01 1.42893493e-01
-4.17256266e-01 8.11871350e-01 1.61497772e+00 -3.34273398e-01
1.87029228e-01 -1.51799500e-01 -9.92131710e-01 5.19467115e-01
4.65067506e-01 1.89366043e-01 -5.29291213e-01 -5.41202962e-01
1.62655219e-01 9.46795046e-02 5.85906208e-01 -1.03163064e+00
-1.89760134e-01 -1.03110142e-01 1.09758496e+00 -1.73022062e-01
4.48380828e-01 -3.80940199e-01 8.37197900e-02 2.02903990e-02
-1.48043096e-01 -4.23045963e-01 9.98137295e-02 6.27117217e-01
-2.83718705e-01 1.72199070e-01 1.15237069e+00 1.59609262e-02
-5.82325757e-01 7.23266482e-01 8.74591321e-02 -1.12587713e-01
4.85357940e-01 -4.11403216e-02 -5.83123207e-01 -5.67314744e-01
-7.32951760e-01 -2.79090524e-01 3.11189204e-01 5.05926609e-01
1.06168187e+00 -1.55361187e+00 -1.08258510e+00 5.43830514e-01
1.62205890e-01 -3.99484724e-01 8.99386585e-01 3.06874812e-01
8.68585519e-03 1.00813761e-01 -5.87919891e-01 -4.16877955e-01
-1.22682369e+00 7.43807971e-01 4.35088813e-01 3.44821930e-01
-1.95308298e-01 1.00371158e+00 4.80465949e-01 1.61240220e-01
7.27228001e-02 8.59822333e-02 -5.49948633e-01 -1.45105794e-01
1.11664653e+00 4.25523579e-01 3.24826807e-01 -8.62601101e-01
-1.41435519e-01 4.98091817e-01 1.60757899e-01 1.48548201e-01
1.26562619e+00 -4.21695620e-01 -2.54084408e-01 -1.05253518e-01
8.36628437e-01 -2.22683057e-01 -1.51490784e+00 -2.34937534e-01
-4.96280760e-01 -1.08337438e+00 2.79701918e-01 -1.05105472e+00
-1.43519831e+00 9.66017127e-01 1.22874022e+00 -2.49878168e-01
1.84225094e+00 -3.63148982e-03 3.52590770e-01 -2.18457639e-01
3.79020303e-01 -8.93447876e-01 5.48567533e-01 -4.15093787e-02
1.41986394e+00 -1.28373551e+00 -9.53797176e-02 -4.11711872e-01
-4.20385063e-01 7.69353807e-01 6.62687540e-01 1.58635259e-01
5.12116730e-01 4.38284017e-02 2.82968469e-02 -9.77039710e-02
-7.55729795e-01 -1.93398789e-01 6.58294082e-01 1.17339063e+00
3.57380182e-01 1.41486730e-02 1.46857828e-01 8.57401073e-01
-3.65400016e-01 1.76290628e-02 3.57490271e-01 1.76898137e-01
-4.05095816e-01 -6.30383790e-01 -4.75158304e-01 6.48724794e-01
-4.62634355e-01 -2.44426817e-01 -1.64335966e-01 -8.02658945e-02
3.54645461e-01 1.26219010e+00 4.06167395e-02 -2.53903121e-01
2.79976815e-01 6.35883734e-02 8.63568962e-01 -2.34877437e-01
4.65009660e-02 -6.91021159e-02 -1.98317394e-01 -7.77130961e-01
-5.82846820e-01 -6.49629056e-01 -3.78068626e-01 -3.38656276e-01
-1.90320060e-01 -4.90003198e-01 5.08324862e-01 6.02329791e-01
4.28381056e-01 5.27972758e-01 8.71037841e-01 -7.92888165e-01
-3.49598110e-01 -1.32310569e+00 -7.27381229e-01 6.42384827e-01
4.20590132e-01 -9.43468571e-01 -4.66677785e-01 4.87835348e-01]
|
[12.852072715759277, 0.028658581897616386]
|
644bb7b7-ce52-422e-8533-0105135afb00
|
referring-expression-generation-in-time
| null | null |
https://aclanthology.org/L18-1476
|
https://aclanthology.org/L18-1476.pdf
|
Referring Expression Generation in time-constrained communication
| null |
["r{\\'e}", "Andr{\\'e} Mariotti", 'Iv Paraboni']
|
2018-05-01
|
referring-expression-generation-in-time-1
|
https://aclanthology.org/L18-1476
|
https://aclanthology.org/L18-1476.pdf
|
lrec-2018-5
|
['referring-expression-generation']
|
['computer-vision']
|
[-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]
|
[-7.206428527832031, 3.620995283126831]
|
a3979f27-8db2-44e3-93c3-74a2ce4b94e0
|
bigvgan-a-universal-neural-vocoder-with-large
|
2206.04658
| null |
https://arxiv.org/abs/2206.04658v2
|
https://arxiv.org/pdf/2206.04658v2.pdf
|
BigVGAN: A Universal Neural Vocoder with Large-Scale Training
|
Despite recent progress in generative adversarial network (GAN)-based vocoders, where the model generates raw waveform conditioned on acoustic features, it is challenging to synthesize high-fidelity audio for numerous speakers across various recording environments. In this work, we present BigVGAN, a universal vocoder that generalizes well for various out-of-distribution scenarios without fine-tuning. We introduce periodic activation function and anti-aliased representation into the GAN generator, which brings the desired inductive bias for audio synthesis and significantly improves audio quality. In addition, we train our GAN vocoder at the largest scale up to 112M parameters, which is unprecedented in the literature. We identify and address the failure modes in large-scale GAN training for audio, while maintaining high-fidelity output without over-regularization. Our BigVGAN, trained only on clean speech (LibriTTS), achieves the state-of-the-art performance for various zero-shot (out-of-distribution) conditions, including unseen speakers, languages, recording environments, singing voices, music, and instrumental audio. We release our code and model at: https://github.com/NVIDIA/BigVGAN
|
['Sungroh Yoon', 'Bryan Catanzaro', 'Boris Ginsburg', 'Wei Ping', 'Sang-gil Lee']
|
2022-06-09
| null | null | null | null |
['audio-generation', 'music-generation', 'music-generation']
|
['audio', 'audio', 'music']
|
[ 3.07397600e-02 7.13907108e-02 3.60628366e-01 -1.40194759e-01
-1.59870386e+00 -7.82853067e-01 3.22489887e-01 -7.78506279e-01
2.78535575e-01 7.03134418e-01 5.16346574e-01 -1.96150333e-01
3.91027272e-01 -6.16290569e-01 -8.75482678e-01 -7.63339877e-01
1.14117644e-03 2.82714665e-01 -2.10836932e-01 -2.54828066e-01
-6.54357910e-01 2.48267755e-01 -1.31771302e+00 2.98266381e-01
5.84250808e-01 9.24928725e-01 -1.23795316e-01 1.26318240e+00
4.12347734e-01 7.95437932e-01 -1.21630979e+00 -6.11592472e-01
3.27325046e-01 -9.54473317e-01 -4.68818396e-01 -2.26885796e-01
6.97581649e-01 -4.54901725e-01 -6.16637588e-01 7.74660468e-01
1.21194458e+00 1.17183104e-01 6.73976243e-01 -1.21369028e+00
-8.61595809e-01 8.38364720e-01 -1.69080749e-01 1.30876184e-01
2.66540647e-01 5.55276930e-01 1.10303175e+00 -7.98140585e-01
2.50442207e-01 1.22103524e+00 7.90636122e-01 9.59837675e-01
-1.26114094e+00 -1.00450659e+00 -2.62515694e-01 -1.00376882e-01
-1.32888353e+00 -8.49562347e-01 9.62208152e-01 -1.70396045e-01
8.74319375e-01 3.62407476e-01 5.60738623e-01 1.89230251e+00
2.32077893e-02 5.66986144e-01 7.78088987e-01 -3.13213646e-01
2.41892084e-01 -2.40077808e-01 -6.74670041e-01 1.04868703e-01
-3.92213523e-01 3.45702261e-01 -7.64269352e-01 -2.20914915e-01
9.44449067e-01 -4.68817830e-01 -2.97283500e-01 3.42232287e-01
-9.32985961e-01 8.17496181e-01 2.59702623e-01 7.19221905e-02
-3.30063671e-01 6.31386220e-01 4.14991856e-01 5.43133557e-01
2.50669688e-01 4.63498294e-01 -1.74132764e-01 -5.94509065e-01
-1.00573790e+00 2.66321301e-01 8.19323361e-01 1.05920136e+00
1.69615000e-01 1.34994555e+00 -3.15077454e-01 1.10441244e+00
-6.35436699e-02 8.62377703e-01 7.04825521e-01 -1.03310680e+00
4.69471246e-01 -5.39801002e-01 -3.29383135e-01 -4.58039939e-01
3.08232401e-02 -7.65111625e-01 -1.16582811e+00 1.78817749e-01
1.37258261e-01 -7.54791260e-01 -1.16907978e+00 2.02797556e+00
9.90991518e-02 4.68752861e-01 2.46481553e-01 8.84107590e-01
1.11198759e+00 9.61759031e-01 -2.64255673e-01 2.68430300e-02
1.00949776e+00 -1.17192304e+00 -8.02670479e-01 -1.44463584e-01
-1.29826203e-01 -9.57085192e-01 1.38399410e+00 6.20023310e-01
-1.31558442e+00 -8.86205137e-01 -9.91725147e-01 -2.19205618e-02
1.32045433e-01 -1.12132818e-01 4.28526908e-01 8.02063644e-01
-1.32777119e+00 2.91747600e-01 -8.06295991e-01 2.67752826e-01
2.11322829e-01 1.43457517e-01 -9.65147093e-02 2.62846231e-01
-1.26924503e+00 2.55332053e-01 -3.17354053e-02 -9.01311822e-03
-1.67180324e+00 -9.14433777e-01 -7.31463134e-01 5.34876101e-02
-1.96167771e-02 -8.71023893e-01 1.55893707e+00 -1.05484772e+00
-2.04067516e+00 2.20809773e-01 -2.42848434e-02 -6.58709884e-01
4.48892862e-01 -3.51194859e-01 -9.25839722e-01 2.34170929e-02
-2.53792882e-01 6.81520939e-01 1.38669920e+00 -1.17288256e+00
-9.62587893e-02 1.43298149e-01 -2.27493823e-01 7.80746788e-02
-1.85467243e-01 -8.76935869e-02 -2.23296002e-01 -1.26797593e+00
-4.37381476e-01 -8.76417577e-01 -5.01904190e-02 -6.23828769e-01
-5.41116655e-01 2.03487724e-01 6.73564434e-01 -8.40460777e-01
1.06590474e+00 -2.36735535e+00 9.17074531e-02 -5.83117679e-02
-2.16088971e-04 2.02296510e-01 -3.58020514e-01 6.41343653e-01
-6.31547198e-02 1.12562448e-01 -1.99365780e-01 -5.33881009e-01
1.79447159e-01 8.59476030e-02 -7.92899191e-01 3.02242320e-02
2.35128000e-01 9.30307209e-01 -5.40462792e-01 -1.13527007e-01
7.91580305e-02 9.35714066e-01 -9.89027441e-01 7.20091403e-01
-5.30461520e-02 7.72158980e-01 -6.42261580e-02 5.94192922e-01
4.37186182e-01 2.09633961e-01 -1.51946858e-01 -5.28704971e-02
3.34393322e-01 6.13511205e-01 -9.60764527e-01 1.92599130e+00
-7.95843363e-01 7.23234713e-01 4.42557633e-01 -6.45427465e-01
1.02582169e+00 8.13505948e-01 1.45938978e-01 -3.93021196e-01
3.64178196e-02 3.02899420e-01 1.34180397e-01 -1.80917561e-01
2.62192219e-01 -4.30173248e-01 -1.61842778e-01 2.11846769e-01
7.54180014e-01 -6.96235836e-01 -2.09333792e-01 7.23257437e-02
1.08281314e+00 -1.19454920e-01 3.38492617e-02 1.06231263e-02
4.45697382e-02 -6.16168797e-01 3.89804453e-01 7.88489044e-01
1.36316285e-01 1.10891783e+00 2.25921020e-01 9.60820913e-02
-1.13205135e+00 -1.53551292e+00 6.28641769e-02 1.20233226e+00
-4.81500506e-01 -4.35801327e-01 -9.38504100e-01 -3.92749794e-02
-2.72977561e-01 7.37871408e-01 -3.38195056e-01 -2.97801167e-01
-6.48995936e-01 -4.66095686e-01 1.24599862e+00 6.34177268e-01
2.29138300e-01 -1.34508753e+00 8.74254480e-03 4.17132884e-01
-6.19513243e-02 -1.05093515e+00 -7.97769785e-01 2.46168897e-01
-5.82926869e-01 -3.66004974e-01 -8.18748534e-01 -8.31453025e-01
1.07686430e-01 -3.90247673e-01 1.47373903e+00 -5.42908847e-01
-2.05740213e-01 3.65298659e-01 -2.67634094e-01 -6.07519627e-01
-8.23109031e-01 1.21565863e-01 2.12505236e-01 -7.59822279e-02
-3.50122809e-01 -1.21108007e+00 -5.80694437e-01 9.24890786e-02
-7.92990744e-01 -1.50375307e-01 2.63163745e-01 1.09050918e+00
7.44585931e-01 5.85794486e-02 1.22782362e+00 -6.55904353e-01
7.35550404e-01 -4.35262799e-01 -3.12789857e-01 -2.00354517e-01
2.53042881e-03 -2.55190194e-01 1.11264002e+00 -6.82470202e-01
-8.85727167e-01 -3.80036294e-01 -1.06908941e+00 -7.29347706e-01
-1.77009180e-01 -1.52555481e-02 -3.56103390e-01 2.33685091e-01
8.08283687e-01 3.61457288e-01 -2.53282279e-01 -5.47180951e-01
5.53489983e-01 8.86536837e-01 1.09062421e+00 -7.14127898e-01
9.85645413e-01 4.22382280e-02 -3.38809520e-01 -9.73132074e-01
-6.61119103e-01 7.70148039e-02 -4.22273986e-02 9.21125636e-02
6.44082844e-01 -1.50652146e+00 -5.69892764e-01 6.15731955e-01
-8.82296681e-01 -8.94201696e-01 -7.22786129e-01 5.63025415e-01
-7.93452144e-01 -1.92863747e-01 -9.75758314e-01 -8.74887168e-01
-9.54009235e-01 -1.04830110e+00 1.09282410e+00 2.52029076e-02
-2.95023322e-01 -8.12865615e-01 2.48308495e-01 2.27574781e-01
7.69671261e-01 4.26423430e-01 5.28591454e-01 -4.24079895e-01
-3.36046606e-01 1.47912294e-01 5.47835350e-01 8.71398032e-01
1.73013836e-01 -7.41120568e-03 -1.42812848e+00 -5.00276148e-01
1.74968466e-01 -6.61516845e-01 6.20332062e-01 5.37865579e-01
1.37585402e+00 -5.73487461e-01 4.64245945e-01 1.19752598e+00
1.01766503e+00 2.71060616e-01 7.28231490e-01 -4.35179859e-01
7.92164981e-01 -4.27044928e-03 -2.58774031e-02 4.29906040e-01
-7.33215958e-02 6.13285303e-01 3.73924077e-01 -2.33647898e-01
-6.55897915e-01 -6.12173915e-01 6.07328534e-01 1.47388268e+00
-2.73810290e-02 -6.41343117e-01 -5.15138686e-01 5.58028340e-01
-1.01838768e+00 -9.95964468e-01 3.69941294e-01 1.89138806e+00
1.14782333e+00 7.44569600e-02 4.00204152e-01 2.30605707e-01
4.69875216e-01 2.94705480e-01 -7.33249843e-01 -6.30328715e-01
-2.28708014e-01 9.35214043e-01 9.15180221e-02 4.85484719e-01
-8.94282877e-01 8.68195593e-01 6.99332047e+00 1.11217260e+00
-1.45928943e+00 5.87303340e-01 5.49299717e-01 -5.21101594e-01
-5.40471017e-01 -5.14417768e-01 -6.53138697e-01 5.45489728e-01
1.41145158e+00 -8.84097144e-02 9.86581862e-01 9.08940494e-01
-4.16622311e-02 8.27854633e-01 -1.01500571e+00 1.08243382e+00
1.67829946e-01 -1.14727581e+00 1.51534993e-02 -1.74466997e-01
9.20567334e-01 3.06625545e-01 4.55686331e-01 6.07782364e-01
5.30482829e-01 -1.44679034e+00 9.88348663e-01 1.60888135e-02
1.55805039e+00 -9.03728724e-01 4.30568635e-01 4.85466979e-02
-1.15707922e+00 1.83059692e-01 -1.13082699e-01 1.07616879e-01
4.06067759e-01 4.29193228e-01 -8.91826749e-01 3.72766286e-01
6.63780391e-01 4.15561378e-01 -1.53659955e-01 5.78172028e-01
-3.49221826e-01 1.39669073e+00 -3.47769737e-01 2.70204067e-01
1.13619052e-01 1.17171071e-01 6.48990095e-01 1.24264038e+00
6.57853425e-01 -5.01129627e-02 -1.27552584e-01 7.91481614e-01
-6.33949697e-01 -2.63599306e-01 -5.34259200e-01 -1.44509763e-01
6.55733228e-01 9.73229647e-01 6.79112971e-02 -1.23518467e-01
-7.18260780e-02 9.69565809e-01 7.62490556e-02 6.45769238e-01
-1.11206067e+00 -5.33777058e-01 9.11281288e-01 -9.05004051e-03
4.03178900e-01 -1.79663837e-01 -6.63771108e-02 -1.13164127e+00
4.87168431e-02 -1.41908610e+00 1.48647010e-01 -8.45356345e-01
-1.42236865e+00 1.12467539e+00 -4.42689568e-01 -1.11532056e+00
-8.26868236e-01 -1.62784621e-01 -9.26748693e-01 9.49720502e-01
-1.02179384e+00 -1.12736487e+00 -1.60768196e-01 7.38824010e-01
7.36769080e-01 -4.82100010e-01 1.07261288e+00 5.45946002e-01
-3.45851243e-01 1.11096990e+00 2.06154928e-01 1.38595790e-01
7.46771514e-01 -1.18911111e+00 9.30296421e-01 8.49267185e-01
4.09643531e-01 3.98059130e-01 8.07990730e-01 -2.11618483e-01
-1.43681765e+00 -1.33247089e+00 1.51905611e-01 -3.55187446e-01
6.48250818e-01 -8.56874943e-01 -8.35359693e-01 7.88717866e-01
6.48242354e-01 8.87317434e-02 8.26726794e-01 -6.99709775e-03
-5.14602304e-01 -3.41366410e-01 -9.94767368e-01 5.37040889e-01
1.04270852e+00 -8.42573464e-01 -3.67554456e-01 9.45383087e-02
1.20115900e+00 -7.14011669e-01 -1.09362996e+00 2.73680091e-01
5.32881260e-01 -8.75792146e-01 1.14361918e+00 -5.09887815e-01
4.88839835e-01 -8.02974477e-02 -3.97354275e-01 -1.76057851e+00
-1.23912953e-01 -1.37874508e+00 -3.91139776e-01 1.64103365e+00
4.70348597e-01 -5.55623233e-01 4.78368968e-01 -1.03958711e-01
-6.49359763e-01 -6.11917078e-01 -1.07194746e+00 -9.29772675e-01
2.43775502e-01 -6.13960743e-01 8.70590150e-01 5.40745854e-01
-3.69764537e-01 4.73671824e-01 -9.68277156e-01 1.77928388e-01
5.21791160e-01 -8.92232731e-02 1.03162134e+00 -4.89668578e-01
-8.58803511e-01 -2.03823626e-01 -2.54568011e-01 -1.04555178e+00
1.18611783e-01 -7.72961557e-01 1.51752427e-01 -1.21236062e+00
-3.79020661e-01 -3.23459476e-01 -1.51829615e-01 3.47400963e-01
1.96941867e-02 6.55321181e-01 2.42952541e-01 -1.99755147e-01
-9.37075913e-02 9.14655328e-01 1.21316147e+00 -1.60118386e-01
-1.42516598e-01 6.91362321e-02 -8.28017771e-01 5.63727200e-01
9.72588599e-01 -4.17498291e-01 -5.61709702e-01 -6.20995820e-01
-2.43202552e-01 2.81899631e-01 3.13563198e-01 -1.46217537e+00
-1.33143023e-01 1.05509259e-01 3.58186424e-01 -2.27841377e-01
9.07417893e-01 -4.86904711e-01 5.88377535e-01 1.78059414e-01
-4.22302783e-01 -1.52521774e-01 2.66472936e-01 3.02913010e-01
-5.19091606e-01 1.51918724e-01 8.72434616e-01 1.63001701e-01
-4.18339409e-02 4.17725056e-01 -1.56892553e-01 8.01293552e-01
3.36038858e-01 3.14811856e-01 -2.78365523e-01 -1.01821125e+00
-6.96713805e-01 -3.26295704e-01 9.13244411e-02 4.44806844e-01
4.82548982e-01 -1.60227144e+00 -1.09622777e+00 5.54462969e-01
-3.24994981e-01 1.13636397e-01 6.01709902e-01 2.74923176e-01
-3.70786101e-01 1.85516462e-01 -9.40958112e-02 -4.49044108e-01
-8.61194134e-01 3.38326842e-01 3.67586285e-01 4.04376425e-02
-6.75508320e-01 1.03638732e+00 4.25524354e-01 -3.22464287e-01
3.64852756e-01 -1.32175639e-01 3.65502805e-01 -3.37638259e-01
4.10640419e-01 1.98614731e-01 1.74756140e-01 -4.47880536e-01
-5.96500337e-02 3.07240456e-01 4.82904553e-01 -4.33748782e-01
1.15157640e+00 2.07099825e-01 4.10228491e-01 5.72935402e-01
1.22836137e+00 5.48417807e-01 -1.37298560e+00 1.03233308e-01
-1.02516890e+00 -2.80880004e-01 -1.14658453e-01 -7.01416492e-01
-1.37884092e+00 1.18068504e+00 4.12132055e-01 1.53619081e-01
1.29741967e+00 1.29195713e-02 1.20226455e+00 3.37628201e-02
3.35837901e-01 -6.68507576e-01 2.75566727e-01 4.50929850e-01
1.26762283e+00 -8.09580505e-01 -4.54512417e-01 7.14996830e-02
-8.90137553e-01 6.97763145e-01 5.42695999e-01 -3.10059845e-01
6.27155185e-01 8.53857815e-01 4.00247872e-01 2.46548578e-01
-9.09226120e-01 1.56851575e-01 2.65835553e-01 8.44071388e-01
5.46283007e-01 3.14364523e-01 3.85310709e-01 9.17402387e-01
-1.16763139e+00 -4.17702228e-01 3.55692536e-01 2.91206211e-01
4.55291197e-02 -9.44673836e-01 -3.92598867e-01 3.79701316e-01
-9.61428881e-01 -4.37275976e-01 -1.61957368e-01 5.27151704e-01
-1.33538261e-01 1.05424607e+00 1.00009121e-01 -5.65324605e-01
2.98565179e-01 7.96513259e-02 3.26732725e-01 -4.09858048e-01
-8.79711032e-01 4.61326748e-01 2.68602729e-01 -3.10481846e-01
6.29008412e-02 -3.41316819e-01 -1.02053368e+00 -2.93838203e-01
-1.46102130e-01 2.89044529e-01 5.35468340e-01 3.25599402e-01
5.39515078e-01 1.05141318e+00 7.75935709e-01 -9.08396900e-01
-7.29419172e-01 -1.25599778e+00 -7.11617708e-01 3.99046600e-01
6.18873835e-01 -1.53162643e-01 -6.50037229e-01 2.91619629e-01]
|
[15.41445255279541, 6.152064800262451]
|
b72295d1-ac76-41e4-8c8c-d73cab83be62
|
mathematical-imaging-methods-for-mitosis
|
1609.04649
| null |
http://arxiv.org/abs/1609.04649v3
|
http://arxiv.org/pdf/1609.04649v3.pdf
|
Mathematical Imaging Methods for Mitosis Analysis in Live-Cell Phase Contrast Microscopy
|
In this paper we propose a workflow to detect and track mitotic cells in
time-lapse microscopy image sequences. In order to avoid the requirement for
cell lines expressing fluorescent markers and the associated phototoxicity,
phase contrast microscopy is often preferred over fluorescence microscopy in
live-cell imaging. However, common specific image characteristics complicate
image processing and impede use of standard methods. Nevertheless, automated
analysis is desirable due to manual analysis being subjective, biased and
extremely time-consuming for large data sets. Here, we present the following
workflow based on mathematical imaging methods. In the first step, mitosis
detection is performed by means of the circular Hough transform. The obtained
circular contour subsequently serves as an initialisation for the tracking
algorithm based on variational methods. It is sub-divided into two parts: in
order to determine the beginning of the whole mitosis cycle, a backwards
tracking procedure is performed. After that, the cell is tracked forwards in
time until the end of mitosis. As a result, the average of mitosis duration and
ratios of different cell fates (cell death, no division, division into two or
more daughter cells) can be measured and statistics on cell morphologies can be
obtained. All of the tools are featured in the user-friendly
MATLAB$^{\circledR}$ Graphical User Interface MitosisAnalyser.
|
[]
|
2017-02-10
| null | null | null | null |
['mitosis-detection']
|
['medical']
|
[ 4.59599227e-01 -4.40645039e-01 4.04953241e-01 8.43788981e-02
-3.99269253e-01 -8.30580115e-01 3.69652987e-01 6.93396986e-01
-8.87440979e-01 9.53361928e-01 -7.18842506e-01 -2.25016266e-01
3.94663699e-02 -6.60838842e-01 -1.00663371e-01 -1.37617636e+00
2.08701655e-01 5.92080653e-01 3.70037973e-01 4.87121820e-01
5.55324316e-01 8.31494033e-01 -1.46801531e+00 -5.37648737e-01
5.93072772e-01 5.12577295e-01 3.74125749e-01 8.42323482e-01
-1.70239061e-01 2.93300390e-01 -4.37212735e-01 -5.29801287e-03
-2.03847110e-01 -7.21158504e-01 -4.77159262e-01 2.67219454e-01
-4.06648546e-01 -7.21665025e-02 3.54806095e-01 9.75669503e-01
5.14201522e-01 1.63623571e-01 8.88938427e-01 -9.32340026e-01
2.82029808e-01 -4.46767733e-02 -4.84786600e-01 4.67758954e-01
2.97188550e-01 3.36133868e-01 2.25171492e-01 -9.02077794e-01
8.85929704e-01 5.58010280e-01 2.79248178e-01 3.23182404e-01
-1.52278030e+00 -1.71050563e-01 -4.16394264e-01 -1.43003128e-02
-1.40659034e+00 -3.71083170e-01 5.09420633e-01 -7.29670048e-01
4.23059791e-01 2.85020471e-01 8.32367539e-01 5.41296542e-01
3.68859619e-01 2.05500856e-01 1.37700963e+00 -4.66658801e-01
4.81667519e-01 -5.71456850e-02 -6.55450821e-02 4.66508210e-01
3.83280575e-01 1.23085506e-01 -1.43895671e-01 9.41437483e-02
7.06849396e-01 -3.40126194e-02 -3.91059905e-01 -3.42863500e-01
-1.22715664e+00 4.00300503e-01 -1.00493081e-01 8.13571572e-01
-3.92813355e-01 -8.06488395e-02 2.26181656e-01 -2.86734998e-01
2.12596998e-01 5.50852306e-02 2.84614582e-02 -3.68872762e-01
-1.24535632e+00 2.64685094e-01 4.98818040e-01 1.73552096e-01
6.80795133e-01 -2.71453857e-01 6.88850209e-02 2.70864815e-01
2.26677805e-01 3.12947601e-01 3.68119597e-01 -9.47477102e-01
-5.05138576e-01 6.58333302e-01 2.31555209e-01 -7.16630638e-01
-4.63538855e-01 -1.92825839e-01 -9.20654178e-01 5.86496234e-01
1.17444777e+00 -5.01401946e-02 -5.42449832e-01 1.08118546e+00
5.87360620e-01 -1.08100064e-01 -8.61494690e-02 7.58778930e-01
4.39015597e-01 5.64810634e-01 7.82925263e-02 -8.89702857e-01
1.58473253e+00 -1.61435366e-01 -8.72068465e-01 2.34844729e-01
5.33303976e-01 -8.63640130e-01 6.56421781e-01 3.65897357e-01
-1.11386502e+00 -1.36615068e-01 -8.43512356e-01 8.25798661e-02
-6.17425680e-01 1.76834881e-01 2.44193375e-01 4.72369999e-01
-7.42393374e-01 6.81333721e-01 -1.30308568e+00 -5.79049766e-01
3.94285589e-01 2.94104934e-01 -3.27978581e-01 1.37444168e-01
-3.82998079e-01 6.67473853e-01 4.39297974e-01 1.16103545e-01
-6.02361739e-01 -5.41936040e-01 -8.23875964e-01 -5.02606630e-02
5.98517284e-02 -5.74512839e-01 8.56673062e-01 -3.57189626e-01
-1.69102693e+00 1.34913027e+00 -5.17001510e-01 -7.67672509e-02
6.62704229e-01 5.38714468e-01 2.66688675e-01 2.23031998e-01
-5.41459732e-02 4.81588542e-01 3.80077094e-01 -1.29663086e+00
-7.46422768e-01 -4.74805176e-01 -3.41345906e-01 7.44562894e-02
1.66005909e-01 1.40321165e-01 -6.38908684e-01 -2.75747538e-01
7.51753226e-02 -8.03297997e-01 2.17503570e-02 2.75406968e-02
-1.61963597e-01 1.10827744e-01 9.43877280e-01 -5.59584379e-01
1.11240160e+00 -2.19106174e+00 2.42866889e-01 -5.26656397e-02
1.07603878e-01 2.00850457e-01 5.72344661e-01 3.93563420e-01
8.92855301e-02 -1.05847113e-01 -4.83991116e-01 -3.82876396e-01
-2.61357576e-01 -1.35555163e-01 3.45760435e-01 9.72091496e-01
-6.96145520e-02 5.50589561e-01 -9.57696855e-01 -8.42975616e-01
5.55589795e-01 7.35891342e-01 1.95485666e-01 1.07342809e-01
-1.02986753e-01 1.16849780e+00 -6.40029907e-02 7.56632268e-01
7.87373900e-01 -2.41785079e-01 1.78399339e-01 -7.23315775e-02
-7.38630712e-01 -2.25782260e-01 -1.02784443e+00 1.17448461e+00
3.63262929e-02 4.20097709e-01 3.18192780e-01 -8.00080001e-01
8.24281156e-01 4.34245825e-01 5.76781631e-01 -2.36781105e-01
5.43096006e-01 4.66192991e-01 -1.53440356e-01 -3.48042458e-01
1.40781507e-01 -3.40625077e-01 3.62546891e-01 5.22937536e-01
-2.13140026e-01 -3.29285592e-01 8.80226135e-01 -2.31873780e-01
7.18325615e-01 3.98452491e-01 5.10219812e-01 -4.07171279e-01
8.84643197e-01 5.66710271e-02 5.11088014e-01 6.63602874e-02
-3.24883670e-01 6.30860388e-01 5.92057347e-01 -3.18211079e-01
-9.58796680e-01 -7.54788041e-01 -2.60197967e-01 3.68220478e-01
1.56726941e-01 1.47170618e-01 -9.89959598e-01 -1.32542163e-01
-3.69923413e-01 4.69521850e-01 -4.89558309e-01 3.19824010e-01
-3.85854334e-01 -9.90642667e-01 1.75844669e-01 -2.37411559e-02
2.80674517e-01 -1.13047540e+00 -1.28525865e+00 1.95118591e-01
-1.58633262e-01 -7.30246902e-01 -4.23412062e-02 3.92376572e-01
-9.38931465e-01 -1.16883457e+00 -1.25219738e+00 -6.72311604e-01
1.02958238e+00 6.46648332e-02 5.57027519e-01 2.65607506e-01
-4.73918557e-01 9.29085389e-02 -6.41589910e-02 -1.76339597e-01
-4.27650571e-01 -9.32687819e-02 -2.57666707e-01 -9.97808725e-02
3.46005738e-01 -5.28731942e-01 -8.00271392e-01 3.12274307e-01
-9.69121933e-01 -1.00350179e-01 1.90829277e-01 5.32678485e-01
1.13584936e+00 3.00429493e-01 -2.35109832e-02 -5.76351583e-01
1.48141891e-01 2.58357096e-02 -1.21006620e+00 -5.47544798e-04
-2.85741508e-01 -3.86802703e-01 6.92181587e-01 -2.94420928e-01
-9.02835071e-01 4.15608108e-01 2.46259924e-02 -8.83747265e-02
-5.44971347e-01 5.68982065e-01 -3.05303726e-02 2.12143771e-02
1.55434519e-01 6.40691996e-01 4.22482222e-01 -4.99111079e-02
-3.25525999e-01 3.64437670e-01 5.27007818e-01 -4.77183685e-02
6.46927416e-01 8.58004272e-01 4.85285223e-01 -8.98909330e-01
-6.91946149e-02 -4.46137398e-01 -7.39311635e-01 -5.66357255e-01
9.16018009e-01 -2.57907599e-01 -1.02863288e+00 8.61422122e-01
-9.14856315e-01 -4.48189050e-01 -1.65159076e-01 6.10026300e-01
-6.56436324e-01 6.02989674e-01 -5.93933821e-01 -9.52400744e-01
-2.16844797e-01 -1.25365520e+00 6.32169127e-01 7.04874098e-01
-2.23145649e-01 -1.16709197e+00 2.43123025e-01 1.01763323e-01
1.93619132e-01 4.63619322e-01 8.02106738e-01 -2.04770029e-01
-6.85036123e-01 -4.60199773e-01 1.62566990e-01 -2.62651145e-01
2.48243928e-01 6.24100387e-01 -8.11304927e-01 -2.34342545e-01
-3.18480171e-02 1.86378554e-01 5.99097371e-01 8.31333637e-01
7.52540588e-01 4.06947345e-01 -7.50830591e-01 5.90783715e-01
1.54654288e+00 6.35797262e-01 7.39415765e-01 4.96864587e-01
4.33468558e-02 8.20273757e-01 8.56647193e-01 4.39037859e-01
-1.07823707e-01 5.33614695e-01 4.04703140e-01 -6.91157579e-02
1.86552763e-01 3.83725107e-01 1.90868452e-01 2.34467849e-01
-3.55647743e-01 -3.39182019e-01 -7.54010618e-01 6.34287834e-01
-1.29363441e+00 -8.66907835e-01 -5.24818420e-01 2.56595588e+00
7.17611611e-01 -5.08685247e-04 3.14397484e-01 6.93227112e-01
7.93135881e-01 -3.53536397e-01 -1.83416262e-01 -1.39091924e-01
-1.56533420e-01 -2.08058879e-02 3.67357910e-01 7.67609417e-01
-9.77471650e-01 6.18618906e-01 5.20440960e+00 6.02476537e-01
-1.35943592e+00 -2.76161462e-01 6.88483477e-01 8.74001235e-02
-6.65738583e-02 3.09553981e-01 -7.62466490e-01 7.00412571e-01
5.20077050e-01 -2.42554337e-01 2.04377875e-01 8.94145202e-03
5.98040581e-01 -9.28791881e-01 -7.99240410e-01 8.33469033e-01
-4.14450765e-01 -1.12361956e+00 -4.63567287e-01 3.93022925e-01
9.52320844e-02 -5.24179578e-01 -1.86361089e-01 -4.20439512e-01
-2.70111889e-01 -6.72964692e-01 4.74894375e-01 5.40226638e-01
9.40022111e-01 -6.53493702e-01 8.63772810e-01 4.64604616e-01
-1.08691478e+00 2.75510937e-01 -1.35418952e-01 1.46206617e-01
6.39737308e-01 8.92007828e-01 -7.75056958e-01 4.28620458e-01
3.57902348e-01 2.66660988e-01 -2.49832466e-01 1.35097253e+00
6.42942414e-02 3.05990607e-01 -4.70857829e-01 -5.10796718e-02
-1.84223980e-01 -5.72869122e-01 6.48585737e-01 1.09154344e+00
5.78889608e-01 1.04583114e-01 -3.17363948e-01 1.04413366e+00
3.94319236e-01 5.25280237e-02 -4.50737149e-01 -1.83525175e-01
3.17230791e-01 1.62278676e+00 -1.84710467e+00 -2.01014265e-01
-1.26747817e-01 9.21824396e-01 -1.59393743e-01 2.19650090e-01
-4.84138131e-01 -4.20283437e-01 3.58774997e-02 3.70244294e-01
3.49604875e-01 -2.13563830e-01 -1.74857929e-01 -6.59344971e-01
-3.62866968e-01 -3.72914344e-01 2.95003384e-01 -5.40832639e-01
-4.73865509e-01 1.71878085e-01 -4.83863056e-02 -9.61236477e-01
-2.30354518e-01 -4.76887852e-01 -6.99915946e-01 8.50549817e-01
-1.22128701e+00 -9.10789490e-01 -3.50883454e-01 1.52184919e-01
2.84426183e-01 3.61368388e-01 9.83895421e-01 1.74387529e-01
-7.07631171e-01 -7.20124245e-02 2.58090079e-01 -7.18325526e-02
4.02455479e-01 -1.34171796e+00 -3.81142795e-01 9.50434864e-01
-3.42359632e-01 4.77865160e-01 9.21970904e-01 -5.82350492e-01
-8.23113024e-01 -4.26588982e-01 1.15547001e+00 -7.81753100e-03
3.61552775e-01 -7.63437375e-02 -7.94418514e-01 3.38521183e-01
3.01270932e-02 1.02308795e-01 7.23244071e-01 -6.94792569e-01
6.49885535e-01 5.16466238e-02 -1.21338415e+00 5.33662975e-01
1.01191744e-01 -2.33621690e-02 1.46814182e-01 -8.67433250e-02
-2.14609057e-01 -5.11887789e-01 -9.71721292e-01 1.41491979e-01
5.72390616e-01 -1.08855426e+00 5.42370141e-01 3.10618669e-01
4.54216450e-02 -8.80163848e-01 4.28502530e-01 -6.52733624e-01
-1.04403175e-01 -6.57312155e-01 3.22683960e-01 1.26967442e+00
9.26568881e-02 -5.12286127e-01 8.44442904e-01 2.05080286e-01
1.82572782e-01 -5.10804713e-01 -1.15760255e+00 -4.59854990e-01
-1.87782139e-01 2.27420732e-01 9.26528424e-02 5.59513271e-01
1.88532934e-01 4.27956842e-02 2.96062320e-01 -8.17919746e-02
8.48518491e-01 2.21225187e-01 5.75529516e-01 -1.29945469e+00
9.24509615e-02 -5.71750760e-01 -4.44061995e-01 -4.90266383e-01
-2.96077579e-01 -3.43043417e-01 3.68211325e-03 -1.27773774e+00
3.40109080e-01 -1.14487819e-01 -3.76293026e-02 -3.84533890e-02
-1.57273620e-01 3.59833181e-01 -5.74158803e-02 1.82683647e-01
-2.83709198e-01 -3.04269902e-02 1.16131520e+00 1.28695831e-01
-2.15052485e-01 1.77002400e-01 -1.27681671e-02 6.90339506e-01
6.33983254e-01 -4.88903612e-01 -1.18456952e-01 2.75802165e-01
2.33182192e-01 1.94986761e-01 4.80414122e-01 -1.15126884e+00
3.62042874e-01 7.45612895e-03 2.93798387e-01 -9.16950107e-01
2.83362985e-01 -9.01430249e-01 5.95848322e-01 8.40545177e-01
1.09971680e-01 -1.20349258e-01 4.12217677e-02 2.87009925e-01
-1.21729679e-01 -7.09223270e-01 1.13161469e+00 -2.99344361e-01
-1.40841827e-01 1.12884324e-02 -1.19172513e+00 -3.71087492e-01
1.44313622e+00 -8.32752705e-01 -4.27677371e-02 -1.95932910e-01
-9.69316602e-01 -6.37720851e-03 1.14110029e+00 -6.48046017e-01
4.09852058e-01 -8.05033028e-01 -3.80188793e-01 -6.01108707e-02
-2.96941698e-02 3.36413503e-01 5.00572503e-01 1.58456111e+00
-1.10395563e+00 4.09526378e-01 -2.26930425e-01 -7.81745017e-01
-1.68832660e+00 6.35581136e-01 4.49998111e-01 -4.50735778e-01
-3.11114013e-01 5.36724091e-01 -6.41889423e-02 2.36159414e-01
-1.00197181e-01 -1.31645605e-01 -6.11345470e-01 2.25767836e-01
5.35090625e-01 7.07450211e-01 -2.91385176e-03 -7.55988777e-01
-4.59750116e-01 8.77604604e-01 2.35842302e-01 -1.60948589e-01
1.21860397e+00 -5.59739947e-01 -4.39005584e-01 6.92651451e-01
9.16082740e-01 2.50862297e-02 -1.46112359e+00 4.91752267e-01
-7.09080920e-02 -3.00225019e-01 -3.84618938e-02 -4.80514407e-01
-8.97385299e-01 9.04046893e-01 6.86831415e-01 3.20193559e-01
1.19235909e+00 -1.44683242e-01 4.28413510e-01 -2.39010677e-01
2.06035480e-01 -8.92554760e-01 -4.32321280e-01 1.41050100e-01
2.32615739e-01 -8.76305521e-01 1.13055855e-01 -5.48863888e-01
-1.66308761e-01 1.30270112e+00 2.55758688e-02 2.41262600e-01
4.44401652e-01 6.03144765e-01 2.96073526e-01 -2.53488362e-01
-4.47325289e-01 -2.70101726e-01 -4.40455467e-01 6.66527927e-01
6.07491255e-01 -3.93486440e-01 -9.15293634e-01 1.88791305e-01
1.57183945e-01 5.70640087e-01 6.39285386e-01 1.00842106e+00
-3.72880787e-01 -1.08810055e+00 -6.78553522e-01 -5.66428974e-02
-7.65195191e-01 3.91823858e-01 -2.15692818e-01 7.27924943e-01
1.88383684e-01 6.99812651e-01 3.36731464e-01 3.57732236e-01
-1.32872969e-01 1.74150676e-01 6.48910046e-01 -2.68328160e-01
-2.50208795e-01 4.94797051e-01 -4.27698195e-01 -1.06555007e-01
-7.49740660e-01 -1.02827072e+00 -1.60759056e+00 -2.81304359e-01
-3.35308701e-01 1.79368958e-01 6.84682608e-01 8.72317076e-01
-8.71930458e-03 3.06806624e-01 1.86483994e-01 -1.06216395e+00
2.43103564e-01 -7.34814048e-01 -7.99954712e-01 1.86322495e-01
1.85238108e-01 -6.29575551e-01 -6.17856145e-01 6.42447412e-01]
|
[14.35417366027832, -3.0614616870880127]
|
49c9b665-234f-43cc-b302-b8d5df5164ab
|
lvm-med-learning-large-scale-self-supervised
|
2306.11925
| null |
https://arxiv.org/abs/2306.11925v2
|
https://arxiv.org/pdf/2306.11925v2.pdf
|
LVM-Med: Learning Large-Scale Self-Supervised Vision Models for Medical Imaging via Second-order Graph Matching
|
Obtaining large pre-trained models that can be fine-tuned to new tasks with limited annotated samples has remained an open challenge for medical imaging data. While pre-trained deep networks on ImageNet and vision-language foundation models trained on web-scale data are prevailing approaches, their effectiveness on medical tasks is limited due to the significant domain shift between natural and medical images. To bridge this gap, we introduce LVM-Med, the first family of deep networks trained on large-scale medical datasets. We have collected approximately 1.3 million medical images from 55 publicly available datasets, covering a large number of organs and modalities such as CT, MRI, X-ray, and Ultrasound. We benchmark several state-of-the-art self-supervised algorithms on this dataset and propose a novel self-supervised contrastive learning algorithm using a graph-matching formulation. The proposed approach makes three contributions: (i) it integrates prior pair-wise image similarity metrics based on local and global information; (ii) it captures the structural constraints of feature embeddings through a loss function constructed via a combinatorial graph-matching objective; and (iii) it can be trained efficiently end-to-end using modern gradient-estimation techniques for black-box solvers. We thoroughly evaluate the proposed LVM-Med on 15 downstream medical tasks ranging from segmentation and classification to object detection, and both for the in and out-of-distribution settings. LVM-Med empirically outperforms a number of state-of-the-art supervised, self-supervised, and foundation models. For challenging tasks such as Brain Tumor Classification or Diabetic Retinopathy Grading, LVM-Med improves previous vision-language models trained on 1 billion masks by 6-7% while using only a ResNet-50.
|
['Mathias Niepert', 'Daniel Sonntag', 'Pengtao Xie', 'Shadi Albarqouni', 'Nhat Ho', 'Paul Swoboda', 'Binh T. Nguyen', 'Tri Cao', 'Tan N. Pham', 'Nghiem T. Diep', 'Hoang Nguyen', 'Duy M. H. Nguyen']
|
2023-06-20
| null | null | null | null |
['contrastive-learning', 'self-supervised-learning', 'graph-matching', 'medical-image-segmentation', 'lesion-segmentation', 'diabetic-retinopathy-grading', 'contrastive-learning']
|
['computer-vision', 'computer-vision', 'graphs', 'medical', 'medical', 'medical', 'methodology']
|
[ 4.61020976e-01 3.63948315e-01 -3.93413931e-01 -5.12160063e-01
-9.70758438e-01 -5.98702133e-02 3.99520844e-01 4.87973452e-01
-7.20852375e-01 4.50679630e-01 1.59824550e-01 -3.35362017e-01
-1.86680570e-01 -5.04268289e-01 -7.10874379e-01 -5.26027739e-01
-2.58042663e-01 8.18626344e-01 2.77136594e-01 -2.02849694e-02
-2.34528154e-01 3.48918319e-01 -9.90950346e-01 1.99504733e-01
9.19736326e-01 1.24636018e+00 5.20990968e-01 6.10683143e-01
-3.57554816e-02 9.26680207e-01 -1.15803361e-01 -3.47483128e-01
2.93512434e-01 -3.95298034e-01 -9.50812221e-01 3.17900658e-01
9.22817588e-01 -1.97463274e-01 -6.07333362e-01 1.23094475e+00
8.96589577e-01 -1.79582357e-01 6.47047877e-01 -1.01234186e+00
-9.82560396e-01 4.72009331e-01 -5.42138159e-01 4.99057502e-01
-1.68948188e-01 5.01941860e-01 8.51074755e-01 -7.64287591e-01
8.67726207e-01 1.01211286e+00 7.77632177e-01 7.86092758e-01
-1.26211810e+00 -3.61063272e-01 -8.59800056e-02 1.19154736e-01
-1.13357115e+00 -6.08673580e-02 5.15305042e-01 -6.57444417e-01
9.32136357e-01 -4.60161641e-02 7.19521999e-01 9.56837237e-01
4.33768809e-01 7.29920030e-01 1.19056737e+00 -1.89958960e-01
9.80402902e-02 -4.99697365e-02 1.67121217e-01 1.40722203e+00
2.22209200e-01 -3.40562388e-02 -3.83724123e-01 -1.64364710e-01
8.08366716e-01 3.95242088e-02 -3.48084360e-01 -5.98040164e-01
-1.40969384e+00 1.02210212e+00 7.65750885e-01 3.24658811e-01
-2.82532036e-01 1.49701729e-01 6.15428090e-01 1.79552987e-01
6.51838362e-01 4.56429511e-01 -4.66031164e-01 3.53866220e-01
-9.60440755e-01 -9.25523117e-02 6.04337990e-01 7.00718582e-01
6.94871008e-01 -2.67249253e-02 -3.53859961e-01 9.58526015e-01
3.74549806e-01 3.74983579e-01 7.73430407e-01 -4.14961249e-01
3.94990832e-01 7.66327143e-01 -6.42103314e-01 -7.90540636e-01
-8.81993771e-01 -9.15092766e-01 -1.30894876e+00 1.21052124e-01
3.77378464e-01 -2.45221960e-03 -1.47164166e+00 1.69554579e+00
3.44015837e-01 3.12406480e-01 -5.16775716e-03 8.44940782e-01
1.50722361e+00 1.19136214e-01 2.29941146e-03 -4.12343927e-02
1.46099389e+00 -1.32546568e+00 -4.80185539e-01 -5.28906047e-01
8.26588571e-01 -5.09062409e-01 1.04708469e+00 4.46770042e-02
-1.09047282e+00 -5.02716243e-01 -9.10639346e-01 -1.69277549e-01
-3.59677762e-01 1.16637938e-01 7.09338486e-01 3.96563083e-01
-1.38586569e+00 4.66364950e-01 -1.03158641e+00 -4.33960080e-01
1.14589965e+00 4.04169142e-01 -4.02160048e-01 -5.01300395e-01
-7.93271840e-01 1.08678281e+00 6.08270727e-02 -4.53090407e-02
-1.26614320e+00 -1.28380728e+00 -1.03092325e+00 -1.92789033e-01
2.57616788e-01 -1.30082226e+00 8.10277164e-01 -7.82453835e-01
-1.11807609e+00 1.57287872e+00 2.34750226e-01 -7.72879183e-01
7.20423579e-01 1.83864653e-01 -2.94642776e-01 4.59305674e-01
3.78792226e-01 9.21596110e-01 7.61393964e-01 -8.46374750e-01
-2.41895616e-01 -4.67450887e-01 -2.03604251e-01 1.25538409e-01
-4.21743065e-01 -1.19885497e-01 -5.87300241e-01 -6.86872721e-01
-1.53427973e-01 -9.42263782e-01 -7.66970515e-01 5.07068753e-01
-5.13263345e-01 -3.92242931e-02 5.24565279e-01 -5.33705413e-01
7.19698310e-01 -1.91825199e+00 1.65898740e-01 4.05259021e-02
7.83728778e-01 3.40526462e-01 -4.60009187e-01 -1.16286725e-01
-1.09951690e-01 -1.65930077e-01 -6.29112720e-01 -4.57335770e-01
-2.38572896e-01 2.13735312e-01 2.83236206e-01 7.08795249e-01
2.07989708e-01 1.35452688e+00 -1.00833762e+00 -7.73232579e-01
3.32729697e-01 4.56248373e-01 -5.02763271e-01 2.00063199e-01
-4.05602232e-02 5.44979513e-01 -2.00002655e-01 7.84460247e-01
5.69202185e-01 -9.13625062e-01 -9.83368903e-02 -4.88354981e-01
3.30978125e-01 -5.57808718e-03 -6.53527975e-01 2.24759674e+00
-5.01003385e-01 4.93648291e-01 1.90337613e-01 -1.42330217e+00
5.41944921e-01 7.50811100e-02 9.31958437e-01 -7.05963731e-01
2.25848258e-01 2.33481124e-01 1.16935968e-01 -6.61008477e-01
-1.84873521e-01 -1.17232822e-01 2.27581903e-01 1.10363431e-01
6.23532236e-01 -2.56328434e-01 3.94001126e-01 2.23959565e-01
1.50119197e+00 -3.40631038e-01 2.81232327e-01 -4.82756913e-01
4.12977010e-01 7.70391226e-02 3.98850322e-01 8.99586976e-01
-3.66643548e-01 7.40394652e-01 2.48686165e-01 -6.34674609e-01
-6.25049770e-01 -1.11692989e+00 -4.86938477e-01 1.00997531e+00
4.81850840e-02 -1.99986830e-01 -6.68996751e-01 -9.27581251e-01
6.10762052e-02 1.16802558e-01 -8.90218675e-01 -2.15674624e-01
-2.92626739e-01 -1.20394981e+00 4.81258005e-01 5.50663650e-01
5.74266195e-01 -8.48885953e-01 -4.70072895e-01 1.78152263e-01
1.97167806e-02 -1.39246845e+00 -7.04225779e-01 2.75950074e-01
-9.50355828e-01 -1.16232407e+00 -1.05319965e+00 -1.11633408e+00
9.98285770e-01 6.41181841e-02 1.47153509e+00 5.10243736e-02
-1.08097684e+00 6.05163991e-01 -3.71285714e-02 -3.35148394e-01
-4.88018185e-01 1.18601881e-01 -2.87487477e-01 3.00699975e-02
1.23314574e-01 -4.64337349e-01 -8.09607506e-01 1.53231218e-01
-8.83401036e-01 2.45681331e-01 8.95302415e-01 1.17182851e+00
9.10964489e-01 -5.00333011e-01 4.41840202e-01 -1.12212360e+00
4.32055652e-01 -3.82808357e-01 -4.84990627e-01 4.30957228e-01
-7.80619144e-01 -8.04530177e-03 3.86720121e-01 -3.82878810e-01
-5.30856788e-01 2.46518016e-01 -1.81611106e-01 -5.23775995e-01
6.06424809e-02 7.02608645e-01 2.79224634e-01 -5.56849301e-01
8.49067032e-01 2.95204699e-01 2.43636742e-01 -2.03958273e-01
4.72442359e-01 2.04439670e-01 7.70461321e-01 -3.00582230e-01
6.58534825e-01 6.66304350e-01 3.39790642e-01 -7.83686996e-01
-1.16211772e+00 -6.71225846e-01 -6.24163568e-01 -1.49503842e-01
1.25294375e+00 -1.05524468e+00 -3.96449327e-01 5.19194067e-01
-8.71535003e-01 -5.41573524e-01 -4.23180699e-01 4.86266226e-01
-4.64707673e-01 4.98980910e-01 -7.03304708e-01 -1.89180654e-02
-7.40967393e-01 -1.41170919e+00 1.15208387e+00 2.99124103e-02
7.62536675e-02 -1.41964006e+00 2.76216306e-02 6.21167302e-01
5.61725557e-01 3.76075804e-01 1.01861680e+00 -5.31353474e-01
-3.62172335e-01 -4.03935201e-02 -5.59726000e-01 5.65868258e-01
3.19994241e-01 -5.44550061e-01 -7.34874249e-01 -6.04000747e-01
-2.68037677e-01 -8.01529765e-01 1.33074844e+00 7.61476934e-01
1.39912081e+00 -7.25206807e-02 -4.22792077e-01 1.10260487e+00
1.41648114e+00 -3.91764730e-01 3.23421568e-01 9.04978812e-02
1.03881121e+00 2.65230626e-01 8.89037475e-02 2.43331149e-01
5.77712595e-01 3.92759770e-01 6.51952267e-01 -6.89074099e-01
-6.80787683e-01 -4.61604167e-03 -1.01748540e-03 8.54935527e-01
2.08239034e-01 1.52437435e-02 -1.09192050e+00 8.11066210e-01
-1.80229414e+00 -3.91471028e-01 6.60504773e-02 1.88907349e+00
1.03424001e+00 1.03548370e-01 -6.88532516e-02 -4.54199672e-01
5.15800416e-01 1.46487355e-01 -9.59955931e-01 1.81672066e-01
-2.31845364e-01 4.86597985e-01 8.27911019e-01 2.07273364e-01
-1.38959432e+00 7.54740238e-01 5.66275454e+00 8.05215359e-01
-1.33324242e+00 5.13946772e-01 9.12947178e-01 4.85637691e-03
9.86193214e-03 -4.66707259e-01 -5.41336000e-01 2.07775816e-01
6.33078218e-01 6.74412996e-02 1.79969877e-01 8.64770472e-01
-9.37590674e-02 7.71303698e-02 -1.29886186e+00 1.39154482e+00
4.13126171e-01 -1.84360492e+00 1.69148773e-01 -4.50593885e-03
9.94867146e-01 7.75506318e-01 2.52638698e-01 9.90704149e-02
2.94012457e-01 -1.27037764e+00 2.22717822e-01 3.47196937e-01
1.05788696e+00 -1.76645726e-01 7.22585380e-01 6.82354048e-02
-9.69429970e-01 1.52773768e-01 -4.51768011e-01 5.22838891e-01
4.86859865e-02 8.86906683e-01 -8.50842595e-01 5.82563818e-01
7.06744909e-01 1.04348683e+00 -7.93418825e-01 1.12014246e+00
7.26508573e-02 6.75046861e-01 -1.56721711e-01 3.21678817e-01
4.65984136e-01 1.51795475e-02 3.35044503e-01 1.38611674e+00
-5.94927557e-03 -2.66707808e-01 5.72720706e-01 9.26028132e-01
-3.76970679e-01 7.57172629e-02 -4.52679008e-01 8.76270011e-02
-1.95133850e-01 1.59492993e+00 -7.77684391e-01 -3.89259487e-01
-5.70482969e-01 8.92931521e-01 4.01097029e-01 1.13048993e-01
-6.96540713e-01 -8.11213776e-02 4.95807499e-01 1.76671639e-01
1.77510500e-01 6.81532323e-02 -3.00856799e-01 -1.19534397e+00
-1.72696888e-01 -7.72428155e-01 6.31675005e-01 -5.46158910e-01
-1.80657971e+00 8.32185805e-01 -2.49248177e-01 -1.06831813e+00
-7.09540695e-02 -9.12026405e-01 -3.59342963e-01 6.51281595e-01
-2.00377750e+00 -1.37279928e+00 -5.87833047e-01 8.37610364e-01
4.06992286e-01 -4.88854676e-01 8.78305554e-01 5.31469524e-01
-4.62023765e-01 6.68729067e-01 4.62194330e-05 4.62074518e-01
7.06833541e-01 -1.29184675e+00 3.21567535e-01 6.14497006e-01
1.01237744e-01 2.49713868e-01 2.81163990e-01 -4.03333545e-01
-1.32649088e+00 -1.55938375e+00 5.32213748e-01 -2.01287910e-01
7.29446709e-01 -2.11128682e-01 -8.29278469e-01 4.80829656e-01
1.64703652e-01 1.07237720e+00 6.60703123e-01 -2.69749552e-01
-3.37939590e-01 -2.34568909e-01 -1.26844871e+00 3.03647876e-01
1.25616884e+00 -4.85315979e-01 -3.54964077e-01 1.02357006e+00
6.46016002e-01 -7.47424424e-01 -1.03980696e+00 5.37306190e-01
7.54412562e-02 -7.35274434e-01 1.07788861e+00 -8.39589417e-01
6.48437917e-01 9.45473239e-02 -2.67176367e-02 -1.36100078e+00
-3.15892458e-01 -6.21118665e-01 7.68739879e-02 5.94670355e-01
5.06724834e-01 -7.09120810e-01 7.09421456e-01 2.71264344e-01
-4.98136789e-01 -1.37983632e+00 -1.11026025e+00 -7.70984828e-01
1.45025373e-01 -3.57749969e-01 1.17356759e-02 1.01641512e+00
-2.94364005e-01 3.59177768e-01 -1.99506417e-01 -5.31146005e-02
8.90876293e-01 8.76608957e-03 4.52911556e-01 -1.15325749e+00
-4.66901690e-01 -5.62493682e-01 -7.45817125e-01 -9.80960906e-01
1.81386113e-01 -1.52119768e+00 -1.60066411e-01 -1.94480622e+00
5.21044195e-01 -3.46621335e-01 -5.04543424e-01 6.50030255e-01
-2.96731330e-02 4.33422267e-01 -7.36744925e-02 -4.31958288e-02
-7.33182847e-01 4.19549793e-01 1.53105009e+00 -6.78226471e-01
1.66105658e-01 -1.96428850e-01 -6.09666705e-01 8.86617064e-01
4.83335227e-01 -5.01076043e-01 -4.18999165e-01 -6.32432520e-01
-6.12330996e-02 1.15342708e-02 6.90041363e-01 -1.08840322e+00
3.14322680e-01 6.94309026e-02 2.71665752e-01 -2.73776084e-01
1.60585225e-01 -5.21444678e-01 -5.46283901e-01 7.67390370e-01
-5.23692906e-01 -1.06315501e-03 1.54390499e-01 6.00644529e-01
-2.18269333e-01 2.64628194e-02 1.19323814e+00 -3.45708609e-01
-6.01253152e-01 8.62308145e-01 -3.25141400e-02 6.59159660e-01
8.69647563e-01 6.20933063e-02 -3.80245924e-01 -8.16498697e-02
-7.85718799e-01 3.11779290e-01 4.03425172e-02 3.01436007e-01
8.17751884e-01 -1.17881906e+00 -1.02234769e+00 7.77603267e-03
4.37803715e-01 2.10652277e-01 4.38995838e-01 1.19769740e+00
-6.25934660e-01 3.57211739e-01 -2.29829311e-01 -1.00203669e+00
-1.20873642e+00 4.37142998e-01 5.97168565e-01 -7.72584021e-01
-8.05150628e-01 1.04687917e+00 5.20068228e-01 -5.58417022e-01
3.80074918e-01 -6.86115503e-01 -6.75583025e-04 -3.00771683e-01
3.96634102e-01 7.00654611e-02 2.68617034e-01 -5.04488051e-01
-5.69715738e-01 6.61504209e-01 -2.12998092e-01 3.37613851e-01
1.56451154e+00 1.78938538e-01 -1.59889162e-01 8.78362432e-02
1.56146157e+00 -6.26549542e-01 -1.08403349e+00 -7.04670310e-01
-2.84329385e-01 -1.38740838e-01 4.01532143e-01 -9.51399922e-01
-1.57591188e+00 9.56412077e-01 9.93516862e-01 -2.31466755e-01
1.12920594e+00 3.41848314e-01 1.01730728e+00 3.27838510e-01
1.10195540e-01 -1.03816736e+00 4.64136928e-01 2.76296109e-01
8.41087461e-01 -1.65210116e+00 1.24983832e-01 -3.99088144e-01
-6.05856895e-01 9.34944570e-01 6.05884612e-01 -5.03509566e-02
8.46155763e-01 2.54653484e-01 1.08995624e-01 -4.79570687e-01
-5.59887886e-01 -3.78186494e-01 7.36576438e-01 7.31026173e-01
3.78152698e-01 1.88542604e-02 -1.28033042e-01 4.27412719e-01
5.88051453e-02 1.12556607e-01 3.19577962e-01 6.51177227e-01
-1.22986533e-01 -7.53144085e-01 1.37799323e-01 8.61702144e-01
-5.10425806e-01 -3.05832237e-01 -1.11992702e-01 6.08901262e-01
2.44867086e-01 6.37392819e-01 -1.27498180e-01 -9.61314216e-02
1.18499413e-01 -4.69525874e-01 6.56692863e-01 -8.69950652e-01
-4.99823421e-01 -1.01788528e-01 -6.03327565e-02 -6.19060338e-01
-5.31296790e-01 -4.36181426e-01 -1.25493300e+00 3.02407205e-01
-1.55167887e-02 -4.49083656e-01 5.60608327e-01 9.53343570e-01
4.02469486e-01 8.69545221e-01 2.65310317e-01 -7.22635388e-01
-7.86331892e-01 -8.62168372e-01 -4.17053401e-01 6.45611584e-01
3.34716946e-01 -5.33729374e-01 -1.08357504e-01 3.16525623e-02]
|
[14.73281478881836, -2.3029839992523193]
|
430751e4-fea3-474c-832b-6886f2101751
|
improving-lexical-embeddings-with-semantic
| null | null |
https://aclanthology.org/P14-2089
|
https://aclanthology.org/P14-2089.pdf
|
Improving Lexical Embeddings with Semantic Knowledge
| null |
['Mark Dredze', 'Mo Yu']
|
2014-06-01
| null | null | null |
acl-2014-6
|
['learning-word-embeddings']
|
['methodology']
|
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
|
[-7.188699722290039, 3.784574508666992]
|
5ad46faf-695c-45ce-88af-4f9e3ccc1da8
|
tracking-by-associating-clips
|
2212.10149
| null |
https://arxiv.org/abs/2212.10149v1
|
https://arxiv.org/pdf/2212.10149v1.pdf
|
Tracking by Associating Clips
|
The tracking-by-detection paradigm today has become the dominant method for multi-object tracking and works by detecting objects in each frame and then performing data association across frames. However, its sequential frame-wise matching property fundamentally suffers from the intermediate interruptions in a video, such as object occlusions, fast camera movements, and abrupt light changes. Moreover, it typically overlooks temporal information beyond the two frames for matching. In this paper, we investigate an alternative by treating object association as clip-wise matching. Our new perspective views a single long video sequence as multiple short clips, and then the tracking is performed both within and between the clips. The benefits of this new approach are two folds. First, our method is robust to tracking error accumulation or propagation, as the video chunking allows bypassing the interrupted frames, and the short clip tracking avoids the conventional error-prone long-term track memory management. Second, the multiple frame information is aggregated during the clip-wise matching, resulting in a more accurate long-range track association than the current frame-wise matching. Given the state-of-the-art tracking-by-detection tracker, QDTrack, we showcase how the tracking performance improves with our new tracking formulation. We evaluate our proposals on two tracking benchmarks, TAO and MOT17 that have complementary characteristics and challenges each other.
|
['Joon-Young Lee', 'In So Kweon', 'Seoung Wug Oh', 'KwanYong Park', 'Sanghyun Woo']
|
2022-12-20
| null | null | null | null |
['chunking']
|
['natural-language-processing']
|
[-2.85327546e-02 -6.96724832e-01 -3.68146062e-01 1.32370263e-01
-7.09885180e-01 -6.74784720e-01 4.44356203e-01 2.82042414e-01
-4.86670882e-01 6.06809080e-01 -1.27733409e-01 8.74126926e-02
-3.42888795e-02 -4.51652080e-01 -8.46317828e-01 -7.52438307e-01
-1.93333119e-01 3.36465597e-01 1.23911238e+00 1.50150761e-01
-8.64719134e-03 6.19476080e-01 -1.87955570e+00 7.06730690e-03
5.38342416e-01 1.28518772e+00 3.31056356e-01 7.08542347e-01
-1.64530084e-01 8.64532351e-01 -6.23786449e-01 -3.11495721e-01
4.04320747e-01 -1.46607012e-01 -2.13226065e-01 1.52671203e-01
9.14715946e-01 -5.28936505e-01 -4.51653451e-01 9.84796286e-01
3.81422490e-01 -1.92195568e-02 -1.20085083e-01 -1.49337029e+00
-1.51416585e-01 3.37059081e-01 -9.15279746e-01 4.36327308e-01
5.06759405e-01 1.61905318e-01 6.25898123e-01 -8.76748383e-01
8.17636490e-01 1.05173826e+00 9.57857490e-01 4.11444604e-01
-1.00222707e+00 -9.02341068e-01 3.21906477e-01 3.39953095e-01
-1.58027399e+00 -5.52704334e-01 2.86466748e-01 -6.11648798e-01
3.88935417e-01 3.49494696e-01 8.18827689e-01 4.35108423e-01
1.97483093e-01 7.43424594e-01 5.79531193e-01 -2.33057320e-01
-8.04884732e-02 -2.52297163e-01 2.41231918e-01 5.27933359e-01
5.39935470e-01 3.42413932e-01 -7.23807514e-01 -2.01910555e-01
7.21780956e-01 3.67311418e-01 -4.19880867e-01 -4.21410799e-01
-1.64243281e+00 3.33503544e-01 1.10957779e-01 2.30623364e-01
-3.00963283e-01 3.51174384e-01 6.69290125e-01 1.86203480e-01
1.83417082e-01 -2.64626503e-01 -1.26200899e-01 -1.93531647e-01
-1.38171518e+00 3.55023146e-01 4.38155204e-01 1.36152792e+00
6.77270114e-01 -2.50237025e-02 -6.05939209e-01 2.13324130e-01
2.77338207e-01 6.18804276e-01 2.02489197e-01 -9.03624117e-01
3.17544729e-01 2.16228947e-01 3.77674133e-01 -1.08825397e+00
-3.16361547e-01 -3.63123626e-01 -5.72373211e-01 2.82342494e-01
7.13050544e-01 -8.80018398e-02 -4.93775249e-01 1.77908885e+00
6.34709597e-01 6.12663388e-01 -2.99890637e-01 9.57126617e-01
7.23697662e-01 4.47734207e-01 4.43980694e-02 -7.36137867e-01
1.62543988e+00 -1.06582463e+00 -1.07951677e+00 1.24466702e-01
4.15499002e-01 -1.01452935e+00 1.75847873e-01 1.96774438e-01
-1.25221837e+00 -9.43976104e-01 -9.98843253e-01 2.97710925e-01
-7.66108707e-02 -1.75443277e-01 3.37417960e-01 5.03035247e-01
-9.27112520e-01 3.12834591e-01 -8.94744754e-01 -3.62139702e-01
2.60496229e-01 4.05248553e-01 -1.37931541e-01 -4.52157818e-02
-9.49102283e-01 6.50740445e-01 5.23741901e-01 -7.37844333e-02
-6.10649467e-01 -1.00027108e+00 -4.58311707e-01 9.93468165e-02
6.58130467e-01 -5.66836417e-01 1.24020481e+00 -7.56293595e-01
-1.14177966e+00 5.49047053e-01 -4.43971992e-01 -4.73229140e-01
7.32033908e-01 -5.48659027e-01 -6.79811060e-01 1.63547397e-01
1.03676498e-01 4.74713832e-01 8.75479996e-01 -1.04187012e+00
-1.10847270e+00 -9.26870480e-02 3.96315195e-02 1.63293127e-02
-3.29879910e-01 4.08081740e-01 -1.14779913e+00 -7.29433000e-01
2.91749835e-04 -9.18962121e-01 3.63817513e-02 4.98969257e-01
-1.77032799e-01 -3.55160125e-02 1.25508809e+00 -7.51785710e-02
1.67382467e+00 -2.34596682e+00 -1.72694772e-01 -1.03463307e-01
4.10712242e-01 4.14382339e-01 9.39195752e-02 2.68736631e-01
1.83675349e-01 -4.36203182e-01 4.67495978e-01 -4.29103613e-01
-9.82087106e-02 -1.78719796e-02 -3.58895451e-01 6.93303704e-01
-2.62738734e-01 6.83340549e-01 -1.01968038e+00 -8.74188423e-01
3.51432085e-01 5.08881271e-01 -2.82498777e-01 7.17908219e-02
-2.60105968e-01 4.67569351e-01 -3.19613397e-01 8.23294818e-01
9.79101300e-01 -2.68894911e-01 1.34493709e-01 -5.36772192e-01
-6.90114200e-01 -2.74229825e-01 -1.61467612e+00 1.71431720e+00
1.68402612e-01 6.67469919e-01 1.21963471e-01 -4.46547627e-01
5.79217076e-01 4.16289657e-01 9.45170879e-01 -6.00576103e-01
-7.15672001e-02 1.60751149e-01 -1.86086923e-01 -2.39638895e-01
7.83166647e-01 3.09661150e-01 1.17130138e-01 3.34516972e-01
-3.13861012e-01 6.32298887e-01 6.70508325e-01 2.56433547e-01
1.14058745e+00 2.52525270e-01 2.89121091e-01 -1.42571956e-01
5.21515369e-01 7.48243043e-03 1.02524900e+00 9.27125394e-01
-4.33165133e-01 4.76238638e-01 -6.58777878e-02 -6.04696155e-01
-7.31680036e-01 -1.01269066e+00 -1.62875786e-01 9.87628043e-01
7.39963531e-01 -7.38441050e-01 -3.79561752e-01 -4.18285161e-01
1.75668597e-01 -1.16869591e-01 -2.84199774e-01 1.29120126e-01
-9.63646412e-01 -3.51940781e-01 5.34881949e-01 5.59060395e-01
3.90142113e-01 -7.01122224e-01 -1.13969290e+00 5.57360351e-01
-1.09814778e-01 -1.45400620e+00 -1.01535022e+00 -3.14502716e-01
-7.75465786e-01 -1.13362527e+00 -6.91099405e-01 -5.24998724e-01
3.83071065e-01 7.79736161e-01 1.16948402e+00 5.15479028e-01
-2.27377638e-01 2.69506335e-01 -3.49429697e-01 -2.03551650e-01
-5.34654334e-02 -2.97166079e-01 2.86963224e-01 7.83077553e-02
2.85027295e-01 -1.44627094e-01 -6.72143221e-01 6.86048031e-01
-7.09246933e-01 8.46882015e-02 3.17502528e-01 5.73381066e-01
8.47685874e-01 -1.03928596e-01 2.71202803e-01 -4.96517301e-01
-2.33725578e-01 -2.98211753e-01 -1.04528761e+00 4.15452749e-01
-2.43694648e-01 -4.78181720e-01 4.31992024e-01 -7.20733583e-01
-8.35663080e-01 3.14961821e-01 2.95433670e-01 -8.82876039e-01
9.45132747e-02 -3.85303795e-02 -5.26569523e-02 -3.20956588e-01
1.42735258e-01 2.19599411e-01 -1.23552844e-01 -4.20934230e-01
2.14951098e-01 1.15017228e-01 8.32112491e-01 -4.24996376e-01
1.06214440e+00 7.56941974e-01 3.47338356e-02 -6.05522692e-01
-6.51952565e-01 -7.54794538e-01 -6.56530499e-01 -7.50102520e-01
8.43750060e-01 -1.04257023e+00 -1.21972370e+00 5.22727668e-01
-1.30907047e+00 1.05201177e-01 -2.11076334e-01 6.11601770e-01
-2.89653331e-01 6.60633862e-01 -6.35995388e-01 -6.67268276e-01
-2.60817885e-01 -1.21808040e+00 1.01879680e+00 4.07090634e-01
-4.08844613e-02 -6.63817883e-01 1.58311561e-01 -1.49830818e-01
3.99667174e-01 3.74120682e-01 4.97946404e-02 -2.34073833e-01
-1.28895473e+00 -2.80043513e-01 -3.01074207e-01 -3.95596623e-01
-1.87644660e-02 2.12853178e-01 -7.13101149e-01 -6.59484386e-01
-1.52303860e-01 1.18905209e-01 6.30175531e-01 4.89020407e-01
9.57191706e-01 -3.00128502e-03 -8.00031185e-01 6.57644749e-01
1.64962125e+00 3.49170119e-01 4.80116963e-01 5.15565872e-01
7.82997549e-01 2.00957507e-02 1.20703304e+00 4.76542652e-01
2.60437310e-01 1.39782488e+00 4.19852018e-01 -6.73700497e-02
-3.03061247e-01 1.44765124e-01 5.15515924e-01 7.18532324e-01
-1.24963075e-01 -3.13937634e-01 -5.32356918e-01 4.28859711e-01
-2.18405342e+00 -1.44879711e+00 -5.65455377e-01 2.47881866e+00
5.59416354e-01 2.07249075e-01 4.91320878e-01 -5.99365011e-02
1.15490949e+00 1.79638252e-01 -3.89076948e-01 4.20262337e-01
-2.59403020e-01 -2.36649290e-01 7.39857674e-01 2.37550899e-01
-1.21269178e+00 6.81760192e-01 6.33002090e+00 7.82066464e-01
-9.33237910e-01 3.21616679e-01 -2.77919948e-01 -4.97059703e-01
3.59915823e-01 4.16817442e-02 -1.44644535e+00 8.22265267e-01
6.42684281e-01 -6.55936599e-02 -9.18092281e-02 5.27513027e-01
1.62375495e-01 -2.37965450e-01 -1.22841573e+00 1.12025678e+00
-1.02015282e-03 -1.62357605e+00 -1.24662228e-01 7.41417855e-02
5.19264698e-01 -2.10609064e-02 -2.67739266e-01 -2.24560555e-02
-5.79950847e-02 -2.27047175e-01 1.33052719e+00 5.55458605e-01
7.02572405e-01 -4.41158384e-01 4.36321378e-01 1.44438908e-01
-2.06577516e+00 6.89742416e-02 -2.12756246e-01 7.13917613e-02
6.46700025e-01 6.72459304e-01 -2.34099496e-02 7.37999022e-01
7.94066906e-01 7.47135937e-01 -5.17256498e-01 1.61827004e+00
4.50724840e-01 2.47952297e-01 -5.13066769e-01 3.83870274e-01
-1.07486024e-01 7.28451908e-02 7.67552793e-01 1.37267947e+00
4.91373956e-01 6.75683320e-02 7.66055167e-01 3.64306450e-01
1.25466570e-01 -2.54462212e-01 -2.67569214e-01 4.71699148e-01
8.95322323e-01 1.13910222e+00 -8.49596739e-01 -6.68929517e-01
-9.57756281e-01 6.48702860e-01 8.90467037e-03 1.25178650e-01
-1.40821457e+00 -1.68248296e-01 7.86624789e-01 1.90748751e-01
7.58848250e-01 -2.45044202e-01 1.37458786e-01 -1.19802535e+00
2.62396187e-01 -5.73063254e-01 6.50983810e-01 -4.02873188e-01
-8.84017825e-01 4.56140727e-01 -4.69721891e-02 -1.87851679e+00
1.14503257e-01 -1.00682616e-01 -5.17277598e-01 3.59833509e-01
-1.59774172e+00 -9.90661383e-01 -5.58691800e-01 7.64807343e-01
6.21502280e-01 1.90113187e-01 3.60127687e-01 1.13819218e+00
-6.19592369e-01 7.60497153e-01 1.73374176e-01 1.31772354e-01
1.02012479e+00 -7.94196069e-01 1.54154330e-01 1.15253651e+00
-4.02452648e-02 3.66232246e-01 7.51111984e-01 -7.99521625e-01
-1.61646354e+00 -1.07258701e+00 6.55591726e-01 -3.89492393e-01
6.72883213e-01 -2.75357872e-01 -1.01853025e+00 7.43201435e-01
5.28661944e-02 5.16236722e-01 4.22434121e-01 -2.59360254e-01
-2.04639867e-01 -3.24078679e-01 -7.58494794e-01 2.42009729e-01
1.14060497e+00 -1.98187158e-01 -3.27428043e-01 2.98333436e-01
5.41641593e-01 -8.14050794e-01 -1.04076123e+00 3.10912281e-01
8.73778343e-01 -1.13421631e+00 1.16275358e+00 -1.17620984e-02
-3.56075287e-01 -1.08769119e+00 -8.62153713e-03 -5.17602265e-01
-4.87948805e-01 -8.37431192e-01 -6.85807049e-01 1.51625931e+00
-2.46926755e-01 -3.06629181e-01 7.81958938e-01 2.69284427e-01
-1.84598953e-01 -3.51568341e-01 -1.01277697e+00 -1.23751938e+00
-6.53437018e-01 -2.16383502e-01 6.42285943e-01 9.07457173e-01
-2.62078732e-01 -1.93825930e-01 -6.88135087e-01 4.04401004e-01
1.03526676e+00 4.78245109e-01 1.08881164e+00 -1.36518180e+00
-3.67636621e-01 -3.04739058e-01 -5.10266662e-01 -1.40522766e+00
-3.22427601e-01 -3.78465772e-01 3.03309351e-01 -9.26148713e-01
4.00604516e-01 -5.40107071e-01 -2.66119182e-01 2.45291293e-01
-3.22485477e-01 4.34068561e-01 8.04945886e-01 6.81606770e-01
-1.29256916e+00 1.32308841e-01 1.03120792e+00 3.35989073e-02
-1.98143115e-03 -7.31464475e-02 4.78675663e-02 7.19688773e-01
2.26968065e-01 -8.12845767e-01 8.20099339e-02 -5.12941539e-01
-9.72270668e-02 2.60254174e-01 4.24320966e-01 -1.43937624e+00
8.81919622e-01 1.21486615e-02 3.19769502e-01 -1.10279846e+00
3.44614863e-01 -1.16471004e+00 8.65466714e-01 7.51427770e-01
1.85111940e-01 4.64697778e-01 4.28478032e-01 7.29270458e-01
-2.74411500e-01 7.28841126e-02 8.97388041e-01 1.94940478e-01
-1.04775929e+00 5.48966706e-01 -2.54036576e-01 -4.66880994e-03
1.43768942e+00 -6.57952845e-01 -4.78247523e-01 4.22501191e-02
-6.29799485e-01 4.28148031e-01 8.34159315e-01 4.55023259e-01
1.89295173e-01 -1.68393993e+00 -5.98498821e-01 1.43439556e-02
1.12819709e-01 -1.74160361e-01 3.47170413e-01 1.36598730e+00
-2.66526431e-01 2.74748147e-01 -1.97656929e-01 -1.09570098e+00
-1.74993873e+00 8.62288952e-01 4.03267480e-02 -2.57140607e-01
-9.73821640e-01 5.06697118e-01 2.91080356e-01 6.04098797e-01
5.03849089e-01 -8.65475833e-02 8.18056688e-02 2.12503210e-01
9.83426690e-01 5.79716444e-01 -1.05204180e-01 -8.73302996e-01
-5.47945559e-01 1.07471359e+00 -2.00286075e-01 2.70990431e-01
7.67406285e-01 -3.96017522e-01 1.54573157e-01 3.08306128e-01
7.27729738e-01 1.90560430e-01 -1.50363004e+00 -4.10755634e-01
1.14764273e-03 -9.12963927e-01 -2.51396507e-01 -2.03003585e-01
-1.21840131e+00 3.40719163e-01 6.66189194e-01 3.42249751e-01
1.09039664e+00 -1.47902519e-01 1.11782765e+00 2.64563709e-02
6.06012225e-01 -8.22940767e-01 -2.25913212e-01 3.97985458e-01
2.87504911e-01 -1.05529344e+00 2.50570983e-01 -5.13297141e-01
-9.12295505e-02 1.07625103e+00 7.07668781e-01 1.43575594e-01
4.10943270e-01 6.61590099e-01 1.54180843e-02 -1.99058890e-01
-6.96235120e-01 -2.63641298e-01 1.17878690e-01 4.35132176e-01
2.97943801e-01 -3.10382485e-01 -2.01298907e-01 3.86880152e-02
4.71770912e-01 2.52754599e-01 2.59941846e-01 1.07127559e+00
-4.92040813e-01 -1.01704299e+00 -9.04561996e-01 1.52128935e-01
-4.90095079e-01 2.22816274e-01 2.20705688e-01 9.80840027e-01
3.02189291e-01 7.39785790e-01 2.76447266e-01 -1.80429906e-01
4.27035749e-01 -2.83137083e-01 4.36670631e-01 -2.88439542e-01
-7.29049802e-01 3.72794330e-01 -2.24923223e-01 -9.75437522e-01
-8.84084582e-01 -9.26182926e-01 -1.34122264e+00 -6.27194881e-01
-6.84916317e-01 1.46998450e-01 1.66156515e-01 7.48723984e-01
6.02231801e-01 6.61730587e-01 3.19788337e-01 -1.10239744e+00
-1.40533328e-01 -5.10173321e-01 -3.49249721e-01 4.98108923e-01
7.69801378e-01 -1.07685971e+00 -1.24368824e-01 1.79427758e-01]
|
[6.441262245178223, -2.0286006927490234]
|
da38c0ad-1fe4-49fc-a235-90639a03b616
|
color-constancy-with-derivative-colors
|
1611.08389
| null |
http://arxiv.org/abs/1611.08389v1
|
http://arxiv.org/pdf/1611.08389v1.pdf
|
Color Constancy with Derivative Colors
|
Information about the illuminant color is well contained in both achromatic
regions and the specular components of highlight regions. In this paper, we
propose a novel way to achieve color constancy by exploiting such clues. The
key to our approach lies in the use of suitably extracted derivative colors,
which are able to compute the illuminant color robustly with kernel density
estimation. While extracting derivative colors from achromatic regions to
approximate the illuminant color well is basically straightforward, the success
of our extraction in highlight regions is attributed to the different rates of
variation of the diffuse and specular magnitudes in the dichromatic reflection
model. The proposed approach requires no training phase and is simple to
implement. More significantly, it performs quite satisfactorily under
inter-database parameter settings. Our experiments on three standard databases
demonstrate its effectiveness and fine performance in comparison to
state-of-the-art methods.
|
['Long Quan', 'Huan Lei', 'Guang Jiang']
|
2016-11-25
| null | null | null | null |
['color-constancy']
|
['computer-vision']
|
[-4.53624576e-02 -6.03589594e-01 1.64703816e-01 -1.72753319e-01
-4.02621478e-01 -6.87223613e-01 5.77745497e-01 -1.24518089e-01
-3.49134922e-01 6.63204193e-01 -2.46937335e-01 6.10144623e-02
8.20402503e-02 -7.08140075e-01 -4.75195110e-01 -9.63558972e-01
1.63662001e-01 1.25466943e-01 3.37827593e-01 -7.05465004e-02
4.08594936e-01 6.93442345e-01 -1.72461140e+00 2.78160665e-02
1.11507201e+00 1.14523053e+00 -4.76203524e-02 8.54703069e-01
-1.62998408e-01 6.63563013e-01 -6.30258203e-01 -4.93990868e-01
4.01565969e-01 -4.58044708e-01 -2.24511936e-01 1.79346532e-01
4.85008270e-01 -4.00444776e-01 -8.75281319e-02 1.29324949e+00
2.09101483e-01 -3.39550362e-03 8.98417652e-01 -6.88297093e-01
-5.52054346e-01 -1.72002017e-01 -7.76693285e-01 -2.46933475e-01
2.13021919e-01 -6.56775832e-02 9.25846756e-01 -9.37008440e-01
3.94756198e-01 7.28667080e-01 5.59053540e-01 1.35769829e-01
-1.17153418e+00 -2.67370522e-01 -2.10268483e-01 2.77862310e-01
-1.64762521e+00 -4.52388346e-01 9.66419101e-01 -6.08850755e-02
3.83094788e-01 5.17315090e-01 6.11308992e-01 5.48912287e-01
1.14279442e-01 6.33827448e-01 1.61157453e+00 -8.70884776e-01
3.20237517e-01 6.19892895e-01 -1.29932776e-01 6.71496809e-01
4.12418008e-01 1.21707700e-01 -3.97760421e-01 -1.49761125e-01
9.39042270e-01 -8.30098465e-02 -6.65383101e-01 -5.95982671e-01
-8.21688056e-01 3.56500119e-01 4.09661949e-01 2.06084877e-01
-2.22804591e-01 2.57822368e-02 -6.71996996e-02 1.78190023e-01
6.85689867e-01 3.70681077e-01 -2.93392628e-01 -7.49799684e-02
-8.59076560e-01 -7.74363428e-02 8.79405558e-01 8.32275391e-01
9.71142590e-01 2.63217073e-02 2.21714735e-01 1.05161238e+00
1.45910233e-01 9.42460418e-01 -4.58864607e-02 -7.98707068e-01
-4.36127186e-02 2.70722479e-01 6.79274023e-01 -8.69567692e-01
-2.80442894e-01 -2.08100438e-01 -5.81356347e-01 5.88651836e-01
8.34410608e-01 -5.64048886e-02 -9.30394650e-01 1.23123527e+00
3.29438686e-01 3.95465046e-02 2.33637895e-02 1.05314732e+00
2.81673700e-01 5.28202236e-01 -5.38715601e-01 -1.28505379e-01
1.22363758e+00 -4.92294401e-01 -5.77727854e-01 1.70113578e-01
-1.22861601e-01 -1.25635266e+00 1.03981352e+00 7.37529159e-01
-9.53222752e-01 -5.48412442e-01 -9.93497372e-01 -3.06165963e-02
-2.70795912e-01 3.69938403e-01 6.39859140e-01 1.05715406e+00
-1.00085449e+00 4.58497375e-01 -5.10204434e-01 -1.40476199e-02
-7.20147192e-02 -5.83553538e-02 -1.45407826e-01 -2.19799846e-01
-7.01016545e-01 8.08876336e-01 1.64668635e-02 2.19625458e-01
-1.58949390e-01 -6.20501995e-01 -5.49286246e-01 -6.31834120e-02
3.30613464e-01 -1.93446577e-01 1.01642084e+00 -1.33563054e+00
-1.92140520e+00 7.53697991e-01 -3.32462668e-01 2.45654024e-02
6.20233655e-01 -3.35319251e-01 -4.49722290e-01 4.70686257e-01
-6.37092173e-01 9.27476212e-02 1.08164299e+00 -1.69036222e+00
-3.93704385e-01 -2.38776103e-01 -1.11832440e-01 -2.29061469e-02
-2.09160715e-01 -1.83388978e-01 -9.09289360e-01 -3.67900789e-01
1.88835293e-01 -7.63379276e-01 1.19463764e-01 3.31136853e-01
-4.41330642e-01 1.88076213e-01 4.38693017e-01 -4.53461111e-01
7.97861159e-01 -2.29014802e+00 -4.07437116e-01 5.73576689e-01
1.21793024e-01 2.69738138e-01 1.88461540e-03 4.53184128e-01
1.36305824e-01 -5.47915459e-01 -3.88493165e-02 -8.94054994e-02
-2.47910768e-02 -8.86003673e-02 -3.48038465e-01 8.51723552e-01
2.84518659e-01 4.51802403e-01 -6.48028433e-01 -2.00719416e-01
4.31932092e-01 8.67695630e-01 -3.66877884e-01 3.17452043e-01
-1.94146991e-01 8.73726159e-02 -1.76844746e-01 7.44725466e-01
1.30473840e+00 6.68832734e-02 1.08001903e-01 -4.89607006e-01
-4.33637530e-01 1.35067448e-01 -1.31694448e+00 1.24792886e+00
-5.12009263e-01 7.36338139e-01 1.31503463e-01 -4.27827984e-01
1.22071505e+00 -9.88929421e-02 1.99100167e-01 -1.01614606e+00
-1.18480632e-02 5.02205074e-01 -2.59613395e-01 -1.90470159e-01
7.83399105e-01 -2.66309887e-01 4.24676359e-01 3.33973199e-01
-2.64189214e-01 -3.33889514e-01 -4.21716198e-02 -2.28511214e-01
4.18355435e-01 3.30867887e-01 3.62919599e-01 -3.47656578e-01
4.90893126e-01 -3.20549101e-01 2.85007983e-01 6.23180270e-01
-7.06453919e-02 7.56884217e-01 4.44531679e-01 -3.44376266e-01
-1.03740048e+00 -1.34831357e+00 -3.77945900e-01 5.89461207e-01
4.07929033e-01 -1.28287539e-01 -6.91003203e-01 -3.09550732e-01
4.59637083e-02 7.31530130e-01 -4.99572933e-01 9.63434577e-02
-3.61992061e-01 -6.96463227e-01 1.50003850e-01 1.62891969e-01
6.56955361e-01 -5.35971284e-01 -7.10016370e-01 -1.30213037e-01
1.70556888e-01 -1.09674394e+00 -2.05872491e-01 1.18869036e-01
-6.48985982e-01 -1.21699977e+00 -1.09466827e+00 -1.96552023e-01
6.93450451e-01 5.87464273e-01 1.29235816e+00 -1.26740307e-01
-7.01155007e-01 5.72591722e-01 -4.67467129e-01 -4.62951571e-01
-2.58896172e-01 -4.06112283e-01 -3.60557139e-01 3.96555752e-01
4.32233214e-01 -3.49240273e-01 -8.97650659e-01 4.00671929e-01
-8.13935041e-01 -5.96462190e-02 5.63103974e-01 6.65210903e-01
5.54935336e-01 1.83666468e-01 -1.09327324e-01 -1.00921583e+00
1.79803908e-01 -4.56270017e-02 -1.22625089e+00 2.02209324e-01
-6.48268163e-01 1.05258383e-01 8.96377146e-01 -1.35174051e-01
-1.48822713e+00 -1.14587352e-01 2.10561812e-01 -1.94454581e-01
-2.73971349e-01 -1.72214061e-02 2.62220427e-02 -3.89307708e-01
7.07063556e-01 2.50185400e-01 -1.25095576e-01 -5.81262708e-01
5.02241075e-01 4.62788552e-01 5.52127242e-01 -6.61777794e-01
9.86855030e-01 1.00208127e+00 2.15035439e-01 -1.14698005e+00
-5.54821312e-01 -6.86235070e-01 -4.42907423e-01 -3.06091905e-01
3.43132794e-01 -7.10010529e-01 -8.94566238e-01 6.81754827e-01
-9.54397142e-01 -2.60208160e-01 -1.59540191e-01 3.94579411e-01
-3.97166550e-01 5.15080690e-01 -4.06329662e-01 -1.28500485e+00
-3.01655214e-02 -7.93384790e-01 9.08615053e-01 4.76866037e-01
3.13723624e-01 -1.06946564e+00 1.34453073e-01 -1.58416599e-01
5.43956995e-01 1.92692727e-01 8.26236963e-01 1.88499019e-01
-8.72722030e-01 -2.77474105e-01 -7.52982676e-01 5.03083289e-01
4.73147333e-01 3.99353027e-01 -1.41639602e+00 -5.64533696e-02
-7.98271876e-03 -8.76382887e-02 9.80427623e-01 3.82232279e-01
9.75765407e-01 2.31660958e-02 1.10158384e-01 8.06135118e-01
1.86001050e+00 -6.40751347e-02 9.40130174e-01 2.31761649e-01
6.07577205e-01 6.86945438e-01 7.25717604e-01 6.40153587e-01
7.38256052e-02 8.21939230e-01 3.89435589e-01 -5.69729567e-01
-3.04760098e-01 2.23124728e-01 2.05636799e-01 4.15547848e-01
-4.11209196e-01 -1.25228465e-01 -3.96171182e-01 3.01025122e-01
-1.33832300e+00 -1.05527997e+00 -3.56354266e-01 2.63060760e+00
9.04627681e-01 -1.12816438e-01 1.20484576e-01 2.14885473e-01
2.81626910e-01 1.57158852e-01 -3.98380727e-01 -3.91817689e-01
-4.75284427e-01 4.21778530e-01 7.22447097e-01 6.16583169e-01
-8.62301171e-01 6.28788173e-01 6.57942772e+00 6.24610305e-01
-1.34280288e+00 -4.01582569e-01 6.26423359e-01 9.11176652e-02
-4.86716837e-01 -2.07505822e-01 -5.97361565e-01 2.88587540e-01
6.63497925e-01 2.21647620e-01 6.51340544e-01 6.19701207e-01
7.65328258e-02 -7.54132748e-01 -7.77582705e-01 1.11799932e+00
1.68707415e-01 -1.00479925e+00 -2.08369493e-01 -3.07966799e-01
7.16723859e-01 -1.48188472e-01 4.29495424e-01 -3.80872607e-01
-2.75677182e-02 -7.61023939e-01 7.42347598e-01 9.44025099e-01
9.76849198e-01 -9.20809448e-01 4.55037922e-01 -9.21274796e-02
-1.04169774e+00 3.12102526e-01 -7.44494855e-01 1.10219419e-01
-3.26567054e-01 1.06099880e+00 -8.66011083e-01 6.63207769e-01
5.75952888e-01 4.30829078e-01 -4.58000779e-01 1.37470651e+00
-4.32069123e-01 3.94353151e-01 -3.11098963e-01 -1.62588641e-01
-4.33860198e-02 -6.69059634e-01 3.00145358e-01 1.49124134e+00
3.87778848e-01 -1.64055794e-01 -2.28028849e-01 1.09319365e+00
1.18605480e-01 1.27777010e-01 -2.98350632e-01 1.48760170e-01
2.00584859e-01 1.36922705e+00 -6.47030592e-01 -4.76119965e-02
-5.57965398e-01 1.32771397e+00 1.50280535e-01 9.16758657e-01
-6.14713252e-01 -6.13852262e-01 7.52983987e-01 4.11165692e-03
4.71490771e-01 -3.78526628e-01 -3.26315016e-01 -1.19146848e+00
3.17128062e-01 -7.05089092e-01 2.13872269e-02 -9.39373553e-01
-1.27377498e+00 5.35639226e-01 -1.41899049e-01 -1.21962333e+00
-1.96098816e-02 -1.14173710e+00 -4.45052683e-01 1.25320351e+00
-2.23093796e+00 -9.97338593e-01 -5.86916447e-01 9.62522864e-01
-3.92682627e-02 2.34296545e-01 9.12725568e-01 1.33969769e-01
-2.01345265e-01 4.83148992e-01 6.10084116e-01 -1.21052995e-01
1.07475913e+00 -1.58626997e+00 1.21601619e-01 9.43608880e-01
2.71987647e-01 6.90057814e-01 9.28386807e-01 -5.48100919e-02
-1.57377458e+00 -5.64773619e-01 4.79838729e-01 -1.07424162e-01
4.67289865e-01 -4.54768986e-01 -7.56456256e-01 -8.87280032e-02
1.31530866e-01 3.60010825e-02 6.75878584e-01 1.55595720e-01
-1.02065623e+00 -5.39312661e-01 -9.82318759e-01 5.91345549e-01
2.75939375e-01 -7.21914709e-01 -1.04090996e-01 1.68037429e-01
-4.29323390e-02 -3.67936730e-01 -3.77048045e-01 2.00984348e-02
7.00577021e-01 -1.71864879e+00 9.66890991e-01 1.23124354e-01
1.52399823e-01 -4.65079099e-01 -2.06782103e-01 -1.29278529e+00
-9.49171260e-02 -7.23370075e-01 9.74042639e-02 1.13239861e+00
2.24808693e-01 -9.13214207e-01 5.83626509e-01 5.67462146e-01
9.99432579e-02 -2.74576128e-01 -6.32797182e-01 -6.60168886e-01
-3.08023155e-01 -3.19290131e-01 3.85230511e-01 6.75900936e-01
-3.47561091e-01 -3.47057372e-01 -4.35502946e-01 2.76273698e-01
9.14252400e-01 6.11732662e-01 8.59062910e-01 -1.30190754e+00
-5.39385974e-01 -3.36160213e-01 -1.52408943e-01 -9.22163606e-01
-2.14302838e-01 -2.96798795e-01 1.59748122e-01 -1.09139597e+00
2.20151350e-01 -7.08455741e-01 -4.50130790e-01 2.46006399e-02
-3.39405000e-01 6.31812334e-01 1.43192694e-01 1.36078283e-01
-3.16205084e-01 3.66001189e-01 1.23321116e+00 1.97164237e-01
-3.62204731e-01 1.29552200e-01 -5.54537773e-01 7.77782142e-01
6.62013531e-01 -4.35632020e-02 -2.94464380e-01 -2.40676239e-01
3.77240598e-01 -5.00040233e-01 5.10495722e-01 -9.51288521e-01
-6.89923689e-02 -1.46655694e-01 7.01218724e-01 -4.06832129e-01
6.18517578e-01 -9.09690380e-01 -1.18245095e-01 1.14042982e-01
2.01913312e-01 -3.26346099e-01 2.44788721e-01 3.97094905e-01
-9.89493802e-02 -2.35082760e-01 1.08026016e+00 3.18759009e-02
-7.34801054e-01 -2.19458193e-02 -6.68365955e-02 -1.03095122e-01
5.77684999e-01 -3.06307226e-01 -4.95009959e-01 -5.10848999e-01
9.31830779e-02 -6.59503162e-01 9.80858862e-01 -1.07118577e-01
4.49245721e-01 -9.25062120e-01 -4.72249061e-01 3.90693843e-01
2.10900366e-01 -6.10131800e-01 2.17032164e-01 7.93658733e-01
-8.82644832e-01 2.74356574e-01 -1.12883829e-01 -5.26134074e-01
-1.15430927e+00 3.40250105e-01 3.94016773e-01 1.21374227e-01
-4.91588175e-01 7.97153771e-01 2.98534960e-01 1.59430102e-01
1.01901978e-01 -3.86019289e-01 1.70214921e-01 -2.75652856e-01
8.47561419e-01 4.23809767e-01 1.21637970e-01 -3.55444551e-01
-2.36805588e-01 8.60197186e-01 -1.07197734e-02 -6.99472800e-02
1.10079134e+00 -2.18510270e-01 -2.28551537e-01 6.38698697e-01
1.15587842e+00 7.91900814e-01 -1.43741393e+00 -8.20504725e-02
-2.87497014e-01 -9.39673483e-01 1.28816694e-01 -8.47638428e-01
-9.59885955e-01 8.43242049e-01 5.66194117e-01 3.39651465e-01
1.50222278e+00 -3.09694827e-01 3.61271352e-01 2.38140211e-01
5.28205633e-01 -1.09974909e+00 2.03977786e-02 2.47800305e-01
5.56336164e-01 -1.15066063e+00 1.54864132e-01 -7.27050781e-01
-4.22594070e-01 1.57926190e+00 1.43830940e-01 -1.83748841e-01
4.68795955e-01 3.21426362e-01 5.90622783e-01 5.49809672e-02
-2.13314727e-01 -4.14509922e-01 6.18054211e-01 7.37398744e-01
6.61228955e-01 -6.33319200e-04 -9.23323110e-02 -1.95756778e-01
5.01169004e-02 -3.03269714e-01 5.91348827e-01 4.30596173e-01
-5.27807832e-01 -1.01808143e+00 -6.36708617e-01 -8.84050950e-02
-4.18420315e-01 -1.68780208e-01 -4.29346710e-01 9.77672756e-01
-1.21141873e-01 9.69612420e-01 9.04179811e-02 1.48727626e-01
2.20024109e-01 -1.54710442e-01 7.86996245e-01 -6.26105368e-02
-1.84107766e-01 6.12673700e-01 2.79461052e-02 -7.37213969e-01
-5.05874693e-01 -4.72027570e-01 -9.63663220e-01 -3.01242411e-01
-2.75425225e-01 8.37207120e-03 8.19299579e-01 5.16918480e-01
6.62073120e-02 1.21454142e-01 9.45125699e-01 -7.16040730e-01
-3.75694931e-01 -5.30243158e-01 -1.18909299e+00 4.76255089e-01
4.72167790e-01 -5.16111374e-01 -5.60880840e-01 -7.59113356e-02]
|
[10.374496459960938, -2.6914613246917725]
|
5aa3d686-681a-4dcd-835c-81345d677d0e
|
mixdehazenet-mix-structure-block-for-image
|
2305.17654
| null |
https://arxiv.org/abs/2305.17654v1
|
https://arxiv.org/pdf/2305.17654v1.pdf
|
MixDehazeNet : Mix Structure Block For Image Dehazing Network
|
Image dehazing is a typical task in the low-level vision field. Previous studies verified the effectiveness of the large convolutional kernel and attention mechanism in dehazing. However, there are two drawbacks: the multi-scale properties of an image are readily ignored when a large convolutional kernel is introduced, and the standard series connection of an attention module does not sufficiently consider an uneven hazy distribution. In this paper, we propose a novel framework named Mix Structure Image Dehazing Network (MixDehazeNet), which solves two issues mentioned above. Specifically, it mainly consists of two parts: the multi-scale parallel large convolution kernel module and the enhanced parallel attention module. Compared with a single large kernel, parallel large kernels with multi-scale are more capable of taking partial texture into account during the dehazing phase. In addition, an enhanced parallel attention module is developed, in which parallel connections of attention perform better at dehazing uneven hazy distribution. Extensive experiments on three benchmarks demonstrate the effectiveness of our proposed methods. For example, compared with the previous state-of-the-art methods, MixDehazeNet achieves a significant improvement (42.62dB PSNR) on the SOTS indoor dataset. The code is released in https://github.com/AmeryXiong/MixDehazeNet.
|
['Bingrong Xu', 'DuanFeng Chu', 'Qian Xiong', 'LiPing Lu']
|
2023-05-28
| null | null | null | null |
['image-dehazing']
|
['computer-vision']
|
[ 2.68519316e-02 -2.81784952e-01 3.38754982e-01 -2.57922895e-02
-2.42559165e-01 1.51508600e-01 3.14910620e-01 -1.85616881e-01
-5.38481951e-01 3.14264417e-01 1.76313624e-01 -9.06446278e-02
-9.61840674e-02 -1.01070511e+00 -8.35802317e-01 -1.13727427e+00
1.44253850e-01 -4.18759555e-01 8.41433764e-01 -4.56979901e-01
2.87017256e-01 2.61121362e-01 -1.66901684e+00 3.50607067e-01
1.20275319e+00 1.07688379e+00 4.45169926e-01 5.73635340e-01
7.98980743e-02 8.90631020e-01 -6.74591899e-01 -2.46023446e-01
2.78823555e-01 -3.66432071e-01 -4.64380920e-01 7.94118866e-02
6.35392368e-01 -6.72410429e-01 -5.92622101e-01 1.40274000e+00
6.02027714e-01 1.83733776e-01 5.14213920e-01 -1.08863485e+00
-1.15476239e+00 3.40649277e-01 -8.23906422e-01 5.96786618e-01
-4.46439415e-01 1.71270683e-01 6.46508992e-01 -9.98672843e-01
-1.93283945e-01 1.16325927e+00 5.45476377e-01 2.50274867e-01
-7.87885904e-01 -9.34426963e-01 2.17926919e-01 5.68727136e-01
-1.66035604e+00 -1.03100032e-01 7.45443404e-01 -3.39857519e-01
7.46088922e-01 1.60717681e-01 4.42767501e-01 5.31428099e-01
4.52178329e-01 7.41049826e-01 1.15563023e+00 -1.11063354e-01
-7.62678757e-02 1.04963802e-01 2.46688038e-01 6.06182694e-01
4.17564243e-01 6.59908131e-02 -3.50778475e-02 2.30584338e-01
8.33868384e-01 3.92976254e-01 -5.69428921e-01 1.07845612e-01
-1.14977074e+00 6.27428591e-01 9.31271970e-01 4.23770010e-01
-2.96317339e-01 2.67206132e-01 1.89459085e-01 1.27487585e-01
6.20176136e-01 2.36981407e-01 2.76527181e-03 3.97736907e-01
-7.00986445e-01 1.01010367e-01 1.86852172e-01 9.19832110e-01
9.83958364e-01 2.67055243e-01 -2.06968531e-01 8.55960608e-01
3.74943321e-03 3.92946392e-01 6.08018398e-01 -6.47097826e-01
3.54199708e-01 4.69420165e-01 -1.11077890e-01 -1.08595395e+00
-1.33133680e-01 -4.06070560e-01 -1.34821439e+00 5.35276115e-01
6.17593378e-02 -5.31414337e-02 -1.11043930e+00 1.10000908e+00
1.33355752e-01 5.63906372e-01 6.13225102e-02 1.08096111e+00
1.05349457e+00 1.19764650e+00 -7.22296238e-02 6.87258318e-02
1.33553302e+00 -1.38490975e+00 -9.76302207e-01 -1.63844988e-01
9.94269550e-02 -9.55647707e-01 1.00926054e+00 4.46695149e-01
-1.09766114e+00 -9.59351718e-01 -1.33668351e+00 -2.72245675e-01
-6.09033167e-01 8.05750415e-02 3.34541947e-01 5.00749886e-01
-1.19966471e+00 4.88069862e-01 -4.34467942e-01 -2.89072488e-02
5.19285083e-01 2.13769704e-01 1.81318708e-02 -2.40759745e-01
-1.31687319e+00 7.05641925e-01 6.67302430e-01 3.74997199e-01
-8.62063527e-01 -7.61231363e-01 -6.92488968e-01 1.96318775e-01
3.90826464e-01 -3.77003938e-01 1.02213120e+00 -1.11479914e+00
-1.33023691e+00 4.07178164e-01 1.73382670e-01 -3.21239114e-01
3.17941278e-01 -4.52011883e-01 -5.29963255e-01 2.25001127e-01
-2.34808981e-01 4.55554038e-01 1.21079075e+00 -1.20508242e+00
-7.20986366e-01 -1.70380801e-01 2.69400477e-01 3.07328492e-01
-5.36723316e-01 1.60973743e-02 -6.05602622e-01 -1.10900450e+00
-2.57987529e-01 -4.63916630e-01 -6.00302890e-02 9.64811370e-02
-2.27079347e-01 -9.12953690e-02 1.02947462e+00 -8.18331301e-01
1.50045598e+00 -2.54907489e+00 5.95739894e-02 -1.75523072e-01
4.80835527e-01 7.53811240e-01 1.51706021e-02 1.76706374e-01
-2.54068643e-01 2.39003643e-01 -3.78699720e-01 -9.34588462e-02
-2.65695721e-01 -6.27836958e-02 -2.15932772e-01 4.52037126e-01
3.98364723e-01 6.22678518e-01 -6.75868213e-01 -4.64314967e-01
4.28130895e-01 7.72157490e-01 -3.91687095e-01 3.10904056e-01
1.19759455e-01 5.77533655e-02 -3.02407175e-01 6.21057689e-01
1.13286352e+00 -6.60591573e-02 -6.93142056e-01 -3.65360916e-01
-3.66845191e-01 -1.68091685e-01 -1.10355258e+00 1.20995224e+00
-4.19951946e-01 5.48355222e-01 1.04067223e-02 -8.28844190e-01
7.62876213e-01 2.49431834e-01 9.61321965e-02 -6.40529275e-01
4.09992695e-01 2.70496994e-01 4.28143665e-02 -6.67858005e-01
5.74396133e-01 -1.06911764e-01 5.10976493e-01 2.24091858e-02
9.69097205e-03 -1.02326408e-01 -7.63643607e-02 6.16496988e-02
7.47557998e-01 -2.56747156e-01 1.65430427e-01 -5.51973462e-01
7.43391037e-01 -2.88584620e-01 4.68586832e-01 4.37399358e-01
-3.49696606e-01 9.01740730e-01 2.11797923e-01 -4.67951864e-01
-1.03203487e+00 -7.70534813e-01 -2.08533168e-01 7.59724200e-01
6.64640009e-01 -3.40903103e-01 -8.86359811e-01 -4.64584619e-01
-2.54788429e-01 4.15874273e-01 -7.04257071e-01 -5.38381636e-01
-5.06269038e-01 -9.97975707e-01 4.31999743e-01 4.26335156e-01
1.39070070e+00 -1.03611302e+00 -4.02376801e-01 4.21702899e-02
-1.98159516e-01 -1.04598069e+00 -8.14251304e-01 -2.08530203e-01
-5.63628197e-01 -8.86966884e-01 -1.12035787e+00 -1.06633937e+00
5.78749299e-01 9.77794766e-01 6.70052171e-01 5.06403029e-01
-2.90422529e-01 -1.26089513e-01 -5.36464810e-01 -8.24017406e-01
-6.73340401e-03 -4.80155982e-02 -3.87729317e-01 4.61776614e-01
2.30623141e-01 -6.49592280e-01 -9.43883538e-01 3.93634528e-01
-1.47103453e+00 1.85245469e-01 9.22120452e-01 8.59713018e-01
3.87323707e-01 7.57262111e-01 2.41074190e-01 -7.00201869e-01
4.58678454e-01 -5.17608881e-01 -5.28941691e-01 -1.83620583e-02
-5.08753955e-01 -2.54925400e-01 6.67937338e-01 -5.45629621e-01
-1.17869830e+00 -4.55218226e-01 -2.95184851e-01 -6.28744960e-01
-1.92225218e-01 1.68278754e-01 -3.51797789e-01 -4.64197606e-01
3.97903204e-01 5.00925958e-01 5.49399070e-02 -4.69338745e-01
-9.03285071e-02 7.88011432e-01 4.42564279e-01 -3.37804735e-01
1.18558717e+00 4.93868738e-01 -3.33836377e-01 -1.01827085e+00
-5.50141275e-01 -4.24431860e-01 -2.94712931e-01 -3.70604806e-02
1.18924487e+00 -1.10722101e+00 -6.02254272e-01 1.13977039e+00
-1.07559705e+00 -5.15878141e-01 -7.51161352e-02 5.28386712e-01
-8.90504569e-02 4.62931216e-01 -6.93396449e-01 -6.33284152e-01
-4.72048908e-01 -1.20425141e+00 8.66227090e-01 5.11621058e-01
7.23463714e-01 -8.10077369e-01 -3.91589046e-01 1.73460469e-01
8.09123039e-01 6.19424284e-02 8.08498859e-01 -1.28271401e-01
-8.79505873e-01 9.63633955e-02 -6.48310602e-01 8.35931540e-01
1.25183254e-01 4.49012369e-02 -1.08130383e+00 -3.24272603e-01
1.51502296e-01 -2.87537742e-02 1.15479827e+00 2.51273483e-01
1.59304607e+00 -2.37793073e-01 1.52879506e-01 1.05873525e+00
1.70950878e+00 2.88475811e-01 1.15106785e+00 7.15130806e-01
8.55144203e-01 4.52719301e-01 5.79091072e-01 1.23471737e-01
2.24440292e-01 5.50986886e-01 8.01478148e-01 -5.84994256e-01
-4.00447488e-01 7.88469017e-02 3.65510225e-01 8.08880270e-01
-4.27830935e-01 -2.99726069e-01 -6.53334558e-01 6.67073071e-01
-1.55172861e+00 -7.12382197e-01 -1.59122050e-01 2.01503015e+00
5.89849234e-01 1.10319622e-01 -8.38111937e-02 1.82829440e-01
8.21747959e-01 4.66617465e-01 -3.50849926e-01 -2.72300273e-01
-2.23093390e-01 2.18404695e-01 7.61589468e-01 3.96354169e-01
-1.25254059e+00 7.94937551e-01 5.06764460e+00 1.23288810e+00
-1.01188362e+00 1.38780579e-01 6.24707520e-01 1.27022177e-01
2.46644709e-02 -3.17352116e-01 -7.90956736e-01 8.30320954e-01
5.24151564e-01 -1.32917598e-01 4.18309242e-01 5.69342613e-01
-2.83389166e-02 -9.21591837e-03 -3.82085145e-01 9.70889151e-01
2.66380668e-01 -1.13608682e+00 1.38847664e-01 8.04583430e-02
7.73433089e-01 -3.37610655e-02 3.00908267e-01 2.22954914e-01
-5.96498773e-02 -8.61372590e-01 7.39349961e-01 3.16112280e-01
6.68444216e-01 -9.41463590e-01 1.18338084e+00 1.48213908e-01
-1.36426950e+00 -1.59819916e-01 -7.80742586e-01 4.44569066e-02
-2.24212095e-01 7.04587996e-01 -1.83390900e-01 8.46238971e-01
1.32919621e+00 7.20410347e-01 -7.36583292e-01 1.29580700e+00
-3.68801534e-01 6.59170866e-01 -2.95047201e-02 2.60290384e-01
4.24615413e-01 -1.81845948e-01 3.47800851e-01 1.22385490e+00
3.64812523e-01 3.36897999e-01 -4.94420603e-02 6.89267874e-01
7.19019175e-02 5.96177317e-02 -3.14789116e-01 2.75640815e-01
5.97757436e-02 1.18935180e+00 -4.45403159e-01 -5.20572066e-01
-6.72019780e-01 1.13669288e+00 5.23837283e-02 4.57932264e-01
-1.09825885e+00 -8.64993215e-01 8.04676473e-01 1.17992982e-01
6.48752689e-01 -1.43394902e-01 -1.15377679e-02 -1.12570393e+00
6.13623997e-03 -9.18668151e-01 2.51504481e-01 -1.05428219e+00
-1.23805094e+00 7.83133626e-01 -2.07120515e-02 -1.19297945e+00
7.72939622e-01 -7.10276484e-01 -8.98796022e-01 1.06267655e+00
-2.10931039e+00 -1.15323138e+00 -9.17565763e-01 8.31377327e-01
8.29631925e-01 1.50164172e-01 4.04966444e-01 6.92119062e-01
-8.28376949e-01 5.84226847e-01 6.37940913e-02 1.64250866e-01
7.93521106e-01 -1.17094123e+00 3.16886842e-01 1.16730547e+00
-4.20346379e-01 4.81669456e-01 5.02157927e-01 -5.00701249e-01
-9.40741301e-01 -1.36703813e+00 3.26101959e-01 5.27282394e-02
6.74122572e-01 -1.35826215e-01 -1.38002145e+00 5.59452236e-01
6.10582411e-01 2.41976723e-01 2.07179800e-01 -5.55419803e-01
-4.40761596e-01 -3.93380672e-01 -8.21934938e-01 5.79512656e-01
7.47507572e-01 -2.50293493e-01 -4.89576787e-01 1.30777210e-01
1.06163442e+00 -4.30341721e-01 -8.22435141e-01 4.74743128e-01
1.79936320e-01 -1.17117512e+00 1.03177083e+00 6.99238330e-02
5.66556692e-01 -7.00760901e-01 1.34958522e-02 -1.29488802e+00
-6.35989785e-01 -6.01013243e-01 -3.68369929e-02 1.11886954e+00
-5.31890951e-02 -8.78831863e-01 4.33749616e-01 2.72513423e-02
-4.20625716e-01 -8.40456188e-01 -5.28198719e-01 -8.60523641e-01
1.11936167e-01 1.37670472e-01 7.23576069e-01 8.31693530e-01
-6.56503201e-01 8.16413835e-02 -6.98704898e-01 6.37723923e-01
6.48724794e-01 -1.74393445e-01 5.73381424e-01 -7.87585497e-01
-1.97089300e-01 -4.14235085e-01 -6.18999183e-01 -9.68901277e-01
-3.00576210e-01 -4.08040643e-01 1.13595255e-01 -1.50760961e+00
1.43829882e-01 -1.70133442e-01 -5.91335833e-01 3.78823787e-01
-6.18767619e-01 3.98719221e-01 2.43001506e-01 2.93118119e-01
-2.88593978e-01 7.32213616e-01 1.49901521e+00 -3.45160097e-01
1.06918707e-01 -5.05916029e-02 -8.60116541e-01 7.58562505e-01
1.17545950e+00 -2.46111900e-01 -4.71024185e-01 -7.61217117e-01
-3.68561417e-01 -6.26827359e-01 5.12935698e-01 -1.28884625e+00
3.48252237e-01 -4.41253670e-02 3.05491030e-01 -4.40675616e-01
2.46998191e-01 -9.78964031e-01 -4.27489616e-02 4.83779579e-01
1.11376852e-01 9.80226845e-02 3.56274039e-01 5.74763834e-01
-5.27905643e-01 -8.39271247e-02 1.04886866e+00 -1.17184587e-01
-1.11561334e+00 5.60797036e-01 -4.00430232e-01 -2.47910321e-01
1.10844505e+00 -2.72319824e-01 -7.00043976e-01 -1.82291523e-01
-2.13417172e-01 9.99432877e-02 3.82043660e-01 3.96439373e-01
9.21454549e-01 -1.27239943e+00 -7.91083217e-01 3.87529612e-01
1.08976923e-01 3.60220492e-01 7.40770221e-01 1.02454126e+00
-9.65904534e-01 6.21925704e-02 -3.05985898e-01 -2.39074811e-01
-1.21470916e+00 8.24827969e-01 5.48669159e-01 7.56805837e-02
-9.00039256e-01 9.32368159e-01 9.54999268e-01 1.45120412e-01
1.84059501e-01 -3.48845184e-01 -2.50634402e-01 -3.14418256e-01
1.00397170e+00 3.48442256e-01 9.57771465e-02 -6.05062127e-01
3.25880572e-02 7.59818196e-01 -2.94086635e-01 4.20644701e-01
1.29581845e+00 -1.77985668e-01 -3.61715674e-01 1.74529299e-01
1.12307036e+00 2.89798453e-02 -1.46177769e+00 -3.42605829e-01
-5.11583209e-01 -8.36856067e-01 3.47211301e-01 -4.40165967e-01
-1.43540382e+00 1.24609971e+00 7.80057430e-01 2.51949877e-01
1.57883573e+00 -3.67895365e-01 9.91268635e-01 1.33325428e-01
7.11641088e-02 -7.40847290e-01 3.19800109e-01 3.97465765e-01
1.00432503e+00 -1.09696436e+00 6.57914877e-02 -6.14912808e-01
-6.06461883e-01 8.82517099e-01 1.09646916e+00 -4.57463443e-01
8.79208207e-01 1.93059698e-01 1.71656728e-01 -1.88580260e-01
-3.17444414e-01 -3.26747537e-01 3.54924530e-01 4.85785782e-01
2.02791810e-01 -2.81678796e-01 -1.37452841e-01 3.45606446e-01
1.31703883e-01 -3.46499681e-01 6.56736076e-01 7.93095887e-01
-7.44114697e-01 -5.90515494e-01 -5.47692537e-01 9.86842513e-02
-6.61234260e-01 -3.64809364e-01 -1.84210353e-02 9.12999034e-01
5.57012320e-01 8.82695556e-01 6.25074422e-03 -6.28481567e-01
4.54248905e-01 -4.60063338e-01 1.13048621e-01 -4.90363955e-01
-5.79315603e-01 3.22254956e-01 -4.13074702e-01 -5.10191321e-01
-4.97533083e-01 -6.11965433e-02 -9.08862233e-01 -4.33016032e-01
-4.53350693e-01 1.68922171e-01 3.83188844e-01 5.98289132e-01
2.71084130e-01 1.18391025e+00 6.12794697e-01 -9.12257254e-01
-1.58384353e-01 -1.05957210e+00 -8.73376548e-01 2.62777686e-01
6.22507513e-01 -6.41575158e-01 -4.93874550e-01 5.33539727e-02]
|
[10.954547882080078, -3.018432378768921]
|
638649c8-f448-497c-9275-775e78b237b6
|
dynamic-anchor-learning-for-arbitrary
|
2012.04150
| null |
https://arxiv.org/abs/2012.04150v2
|
https://arxiv.org/pdf/2012.04150v2.pdf
|
Dynamic Anchor Learning for Arbitrary-Oriented Object Detection
|
Arbitrary-oriented objects widely appear in natural scenes, aerial photographs, remote sensing images, etc., thus arbitrary-oriented object detection has received considerable attention. Many current rotation detectors use plenty of anchors with different orientations to achieve spatial alignment with ground truth boxes, then Intersection-over-Union (IoU) is applied to sample the positive and negative candidates for training. However, we observe that the selected positive anchors cannot always ensure accurate detections after regression, while some negative samples can achieve accurate localization. It indicates that the quality assessment of anchors through IoU is not appropriate, and this further lead to inconsistency between classification confidence and localization accuracy. In this paper, we propose a dynamic anchor learning (DAL) method, which utilizes the newly defined matching degree to comprehensively evaluate the localization potential of the anchors and carry out a more efficient label assignment process. In this way, the detector can dynamically select high-quality anchors to achieve accurate object detection, and the divergence between classification and regression will be alleviated. With the newly introduced DAL, we achieve superior detection performance for arbitrary-oriented objects with only a few horizontal preset anchors. Experimental results on three remote sensing datasets HRSC2016, DOTA, UCAS-AOD as well as a scene text dataset ICDAR 2015 show that our method achieves substantial improvement compared with the baseline model. Besides, our approach is also universal for object detection using horizontal bound box. The code and models are available at https://github.com/ming71/DAL.
|
['Linhao Li', 'Hongwei Zhang', 'Lingjuan Miao', 'Zhiqiang Zhou', 'Qi Ming']
|
2020-12-08
| null | null | null | null |
['multi-oriented-scene-text-detection', 'object-detection-in-aerial-images']
|
['computer-vision', 'computer-vision']
|
[-3.49318199e-02 -4.44829464e-01 -5.14778435e-01 -3.12502623e-01
-7.93817103e-01 -5.47493398e-01 4.44334000e-01 2.34362945e-01
-4.11551505e-01 4.65703309e-01 -2.15285867e-01 -2.34076232e-01
-1.21016093e-01 -9.78573382e-01 -4.07129288e-01 -1.00215495e+00
-7.87094235e-02 2.71119416e-01 6.46758914e-01 6.38671517e-02
3.36806327e-02 6.64600730e-01 -1.21853459e+00 -2.07481191e-01
8.41157734e-01 1.08760369e+00 3.50045532e-01 2.07838967e-01
1.17090166e-01 1.91901311e-01 -5.40185750e-01 1.31128415e-01
3.84688586e-01 -1.28838299e-02 -2.99388170e-01 7.58385882e-02
2.01269239e-01 -4.43526238e-01 -4.84102555e-02 1.21258843e+00
6.00581467e-01 -5.95425516e-02 7.68370748e-01 -1.22195888e+00
-4.98988509e-01 4.29776669e-01 -1.10984790e+00 3.69926125e-01
8.41369759e-03 -2.93089077e-02 1.08455515e+00 -1.29092097e+00
2.37689123e-01 1.14004326e+00 6.78732514e-01 1.30672634e-01
-9.97640729e-01 -1.12914360e+00 5.01723766e-01 2.07018346e-01
-1.95943201e+00 -2.42220551e-01 5.67691505e-01 -3.74034375e-01
9.58677307e-02 4.26849753e-01 5.32478750e-01 5.30657113e-01
-1.03225954e-01 7.24725544e-01 9.56723750e-01 -2.73336530e-01
4.88871485e-02 1.78178653e-01 1.12209991e-01 5.58522820e-01
5.96568167e-01 1.72250867e-02 -3.45121659e-02 -1.44050851e-01
6.28012896e-01 4.33456421e-01 -4.20426786e-01 -3.85238826e-01
-1.33354139e+00 6.60163224e-01 1.08037782e+00 9.09041613e-02
-1.89557135e-01 -3.19717467e-01 1.42691135e-01 -7.13446289e-02
4.61805880e-01 4.38712128e-02 -3.42665374e-01 7.60087490e-01
-7.29502439e-01 -1.20055582e-02 5.15900888e-02 9.86864984e-01
9.02300775e-01 -1.75680131e-01 -3.38865936e-01 8.45441103e-01
5.81854582e-01 1.02178860e+00 2.21279562e-01 -4.62728798e-01
5.11595011e-01 8.38755310e-01 4.28607345e-01 -1.42694247e+00
-5.58106899e-01 -6.65675581e-01 -1.08893001e+00 2.48823371e-02
2.36222431e-01 -2.78641284e-02 -9.00039256e-01 1.31020927e+00
7.88910210e-01 3.21649104e-01 -4.07530963e-02 1.25012207e+00
7.18148649e-01 8.65580559e-01 5.94150387e-02 -2.81055570e-01
1.52336752e+00 -6.03924096e-01 -5.37078738e-01 -2.80003011e-01
5.95721245e-01 -6.24695301e-01 9.46518540e-01 1.45983696e-01
-2.74720043e-01 -6.71520054e-01 -1.36244607e+00 3.54188710e-01
-2.03404784e-01 8.61330748e-01 5.28448284e-01 3.93809944e-01
-3.01722735e-01 1.39753437e-02 -7.64507413e-01 -2.58715570e-01
5.46143472e-01 2.51004517e-01 -1.78911880e-01 -1.28217787e-01
-1.04869950e+00 5.72559655e-01 5.58513105e-01 4.77581441e-01
-6.12143815e-01 -2.21839473e-01 -6.64110780e-01 -2.65120625e-01
5.84838212e-01 -8.76017287e-02 9.35221553e-01 -6.06502771e-01
-8.16743314e-01 6.53815806e-01 -1.15546964e-01 -2.81172276e-01
4.14747685e-01 -7.63425305e-02 -5.78763068e-01 9.88611765e-03
4.26055610e-01 6.99876904e-01 6.52357817e-01 -1.32840896e+00
-1.15369391e+00 -5.33306062e-01 2.32175384e-02 2.42169648e-01
-5.01035571e-01 -1.16849355e-02 -5.00864923e-01 -5.38211465e-01
9.12581444e-01 -9.42414224e-01 -3.27520609e-01 2.24592373e-01
-3.81204754e-01 -4.36540842e-01 9.76597667e-01 -3.16833436e-01
1.38566971e+00 -2.27923179e+00 -2.52697110e-01 4.78860259e-01
5.86626083e-02 2.88419515e-01 -2.82067470e-02 -5.26221097e-02
6.42903894e-02 1.33462846e-01 -1.40207395e-01 9.00967717e-02
-3.72183442e-01 1.38015628e-01 -4.08744216e-01 8.32776904e-01
2.14148462e-01 4.53317672e-01 -9.86726582e-01 -7.43395329e-01
2.72341639e-01 2.40193993e-01 -2.08179519e-01 -9.44987964e-03
1.49961963e-01 4.00294989e-01 -8.31380665e-01 1.16386652e+00
1.06863523e+00 -3.81710649e-01 1.25394855e-02 -4.29115236e-01
-2.01106787e-01 -9.02391523e-02 -1.77342200e+00 1.02715814e+00
-1.44763321e-01 3.07978988e-01 -5.84894232e-02 -8.89323950e-01
1.31395006e+00 1.45822853e-01 3.16670507e-01 -5.53767920e-01
1.94227695e-03 4.25102830e-01 -1.08974993e-01 -3.65176469e-01
4.15703654e-01 1.36959732e-01 3.20107415e-02 -1.55797869e-01
-5.51506460e-01 1.01939306e-01 8.50497335e-02 -3.82788144e-02
5.25415242e-01 -4.41217460e-02 5.51207602e-01 -2.00423434e-01
7.29684651e-01 1.03396446e-01 8.80076885e-01 7.91358709e-01
-2.96025127e-01 6.70310676e-01 3.63435112e-02 -4.88117635e-01
-7.38515794e-01 -8.90080571e-01 -6.08923554e-01 8.96389306e-01
7.36801147e-01 -3.74502242e-02 -3.55436593e-01 -6.77270114e-01
-4.68301997e-02 2.43039176e-01 -5.30518234e-01 -6.64043799e-02
-5.00238538e-01 -1.22203755e+00 2.72979587e-01 5.72331548e-01
7.60588050e-01 -7.72154093e-01 -3.70971054e-01 9.72928703e-02
-3.03568661e-01 -9.31025028e-01 -2.38528624e-01 -1.02813140e-01
-8.18681061e-01 -1.17411149e+00 -6.94928288e-01 -6.66433394e-01
9.44644690e-01 1.06071353e+00 6.85360372e-01 3.90531212e-01
-1.10980988e-01 -2.18070596e-01 -5.51917076e-01 -4.90101814e-01
8.20231959e-02 2.21483320e-01 1.32718652e-01 1.15490764e-01
4.29144800e-01 -1.79361045e-01 -8.56657207e-01 8.50885451e-01
-7.26032853e-01 -5.09592183e-02 6.83762491e-01 7.67956853e-01
8.16421509e-01 2.80041657e-02 5.96812725e-01 -3.76694024e-01
-1.23721734e-01 -6.61706090e-01 -1.13916266e+00 2.70942897e-01
-5.81507385e-01 -3.83380532e-01 3.90192717e-01 -5.77448249e-01
-6.03651106e-01 1.81001097e-01 -3.33375856e-02 -2.54922092e-01
-5.93885630e-02 4.11966026e-01 -2.28522718e-01 -8.38577449e-02
7.68405318e-01 5.77665679e-02 -4.24396873e-01 -3.40078235e-01
-1.02161542e-01 9.97879207e-01 2.40034595e-01 -3.40043366e-01
1.16088080e+00 7.85964668e-01 9.52179208e-02 -6.31778955e-01
-1.01015460e+00 -9.01269794e-01 -5.06737173e-01 -1.93016469e-01
5.36704600e-01 -1.28788960e+00 -3.44335407e-01 3.20148200e-01
-1.01316893e+00 1.17439747e-01 6.67770579e-02 7.35751987e-01
2.80437857e-01 2.86225438e-01 -1.48832291e-01 -1.02752876e+00
-3.54804009e-01 -1.10768759e+00 1.32085514e+00 5.51841378e-01
3.66983831e-01 -4.15606588e-01 -2.80742854e-01 2.10894588e-02
-5.49330702e-03 2.20998004e-01 3.05983484e-01 -5.51722527e-01
-8.91416728e-01 -5.70608079e-01 -6.71715677e-01 1.92020118e-01
1.47663906e-01 1.81806311e-01 -7.65246570e-01 -4.81860995e-01
-4.23916847e-01 -2.03072429e-01 8.93094778e-01 3.19463104e-01
1.20884848e+00 -1.42552495e-01 -8.00584316e-01 5.98213792e-01
1.36112583e+00 1.92265108e-01 4.57418889e-01 5.27688801e-01
6.67611301e-01 3.61356080e-01 1.43436909e+00 4.43511873e-01
4.19155300e-01 8.29765201e-01 6.62632167e-01 -3.16567302e-01
2.08540067e-01 -9.13060233e-02 2.49867335e-01 2.86446214e-01
-1.67744920e-01 -1.71445176e-01 -9.31210935e-01 4.99292165e-01
-1.77747560e+00 -8.64436626e-01 -5.69967747e-01 2.44431376e+00
6.21046007e-01 1.82496652e-01 -3.10008705e-04 3.37847471e-01
9.28515375e-01 2.09974065e-01 -4.62571204e-01 7.69832134e-01
-1.94728747e-01 -3.39083433e-01 9.12590444e-01 3.89639378e-01
-1.51709747e+00 7.00009465e-01 5.09508801e+00 1.18210351e+00
-1.22126949e+00 1.60437882e-01 5.98018825e-01 3.34928602e-01
1.87097460e-01 1.21583201e-01 -1.43722558e+00 4.39784646e-01
1.49638310e-01 2.38573790e-01 -1.05494283e-01 1.15777862e+00
3.61156762e-01 -9.16979387e-02 -4.11782771e-01 7.71372974e-01
-1.92245334e-01 -9.68207896e-01 -7.88861066e-02 -5.83184417e-03
5.86541355e-01 1.75893992e-01 -1.67380329e-02 2.34712884e-01
7.29494914e-02 -6.12995863e-01 7.30691433e-01 1.30448312e-01
7.93894947e-01 -5.04419029e-01 9.40192342e-01 5.21008790e-01
-1.65796793e+00 -3.16538960e-01 -8.80646169e-01 1.95704833e-01
-1.12661347e-01 7.08583832e-01 -8.12000990e-01 7.64246345e-01
7.66842782e-01 7.93399870e-01 -8.26822877e-01 1.31643116e+00
-4.15399432e-01 5.51027894e-01 -5.99072874e-01 -1.45611376e-01
1.64549321e-01 -2.02824637e-01 5.75508833e-01 1.01602638e+00
6.03686392e-01 3.06613714e-01 6.47241294e-01 4.62871104e-01
1.71672240e-01 4.24432546e-01 -3.51178169e-01 6.91178262e-01
8.87399733e-01 1.51690233e+00 -1.03126884e+00 -3.63374323e-01
-2.47554660e-01 4.49953020e-01 9.15648639e-02 3.94812047e-01
-1.14071107e+00 -3.21941197e-01 2.39943966e-01 2.08593190e-01
2.17588440e-01 -2.68250376e-01 -1.33002087e-01 -1.19868791e+00
1.31639585e-01 -6.98274255e-01 5.26637375e-01 -6.90971494e-01
-9.77208316e-01 2.94826865e-01 1.07456453e-01 -1.82925200e+00
4.39957201e-01 -6.58300519e-01 -5.94690561e-01 6.02925777e-01
-1.72249019e+00 -1.18089008e+00 -8.34387720e-01 3.08306664e-01
3.00575465e-01 8.84244367e-02 4.25616264e-01 5.55859506e-01
-8.08136582e-01 3.79940778e-01 1.46495208e-01 4.56604779e-01
7.60599613e-01 -8.56089592e-01 -8.90145972e-02 1.04077053e+00
1.69587195e-01 4.64273065e-01 5.39939404e-01 -4.68079984e-01
-8.71378541e-01 -1.33750284e+00 4.61566418e-01 -2.86876291e-01
4.05267805e-01 -2.23509058e-01 -9.17497694e-01 4.85886395e-01
-4.65100050e-01 4.85316873e-01 3.31524849e-01 -1.17457435e-01
-2.78084219e-01 -7.17840970e-01 -9.67747509e-01 5.19708157e-01
9.47241724e-01 -6.43634330e-03 -2.26618215e-01 7.23324358e-01
5.88699996e-01 -4.34271932e-01 -5.19138634e-01 9.94032919e-01
4.91662979e-01 -7.49684989e-01 1.26410949e+00 5.02598397e-02
-4.73190993e-02 -1.08810663e+00 -2.76284635e-01 -7.72517204e-01
-4.38594252e-01 2.10392192e-01 1.86573833e-01 1.34131014e+00
2.93279558e-01 -7.60714412e-01 4.98110205e-01 -1.83292732e-01
-7.82093965e-03 -6.23751998e-01 -8.97610009e-01 -8.50077450e-01
-3.49090368e-01 -2.32356399e-01 7.99256623e-01 9.79965687e-01
-5.79054713e-01 3.21102619e-01 -3.02366138e-01 9.59740698e-01
5.67954779e-01 2.89143831e-01 9.69224274e-01 -1.43701231e+00
6.45453185e-02 -2.27085873e-01 -4.09706801e-01 -1.36838639e+00
-2.54825145e-01 -7.78170824e-01 9.82056111e-02 -1.45940876e+00
1.34792954e-01 -1.18185091e+00 -3.60013396e-01 6.10897243e-01
-4.35507506e-01 6.54560864e-01 -9.46086124e-02 7.57812083e-01
-7.11873949e-01 4.87572849e-01 1.11102867e+00 -1.75860658e-01
-2.04073519e-01 3.23073715e-01 -4.54428017e-01 8.60963643e-01
9.36245084e-01 -7.07517803e-01 -9.93375331e-02 -2.29813546e-01
2.07505509e-01 -2.65632510e-01 6.13218963e-01 -1.14863801e+00
1.34200662e-01 -3.62275541e-01 6.41803801e-01 -9.61445332e-01
2.77615860e-02 -1.06111753e+00 -5.90760894e-02 6.85875654e-01
1.47945255e-01 -2.42390186e-01 -2.07951441e-02 6.65914237e-01
-9.57587659e-02 -2.78058469e-01 9.80017304e-01 1.82235092e-01
-7.44878769e-01 5.14736772e-01 9.24693942e-02 -2.70531148e-01
1.22616756e+00 -1.91357687e-01 -3.32156628e-01 4.77859899e-02
-4.66782868e-01 6.00312948e-01 3.50720376e-01 2.92509079e-01
4.72804457e-01 -1.43679178e+00 -8.31738532e-01 1.75053641e-01
5.44566810e-01 4.52873528e-01 -2.39693541e-02 9.97594059e-01
-4.25738245e-01 9.91088375e-02 2.71693617e-01 -1.11561358e+00
-1.28349125e+00 3.81465167e-01 3.55316222e-01 -1.56568438e-01
-2.62131661e-01 7.08727479e-01 1.78011492e-01 -4.76158082e-01
1.34663343e-01 -4.21969861e-01 -4.78984416e-01 2.85930485e-01
5.75520098e-01 2.58914232e-01 -1.54721946e-01 -7.29384601e-01
-5.35277367e-01 8.60725343e-01 7.19252750e-02 2.66453117e-01
9.37429607e-01 -2.74982244e-01 -7.13001490e-02 2.30911285e-01
7.55469620e-01 2.74792910e-01 -1.15225685e+00 -3.61706823e-01
-5.52832969e-02 -6.99644446e-01 -5.46945110e-02 -3.11622113e-01
-1.03576171e+00 8.06928873e-01 9.07688975e-01 2.98302859e-01
1.15864098e+00 3.58225591e-02 3.50049853e-01 5.04566073e-01
4.76356357e-01 -8.39667857e-01 1.81323159e-02 1.92340702e-01
6.80321574e-01 -1.60352266e+00 4.09013778e-01 -5.84277570e-01
-2.98556000e-01 9.19400573e-01 9.12976384e-01 -8.55876058e-02
4.77356642e-01 -1.12054691e-01 5.97648509e-02 5.63820936e-02
-3.72560509e-02 -3.37598145e-01 2.94076264e-01 4.90101218e-01
5.65119684e-02 1.98736101e-01 -4.43583071e-01 3.45588148e-01
2.61794239e-01 -2.85353929e-01 1.89059317e-01 7.07704067e-01
-8.07422936e-01 -8.12695146e-01 -9.11709130e-01 3.71254414e-01
-2.87418634e-01 4.67272177e-02 1.14689775e-01 9.01989281e-01
3.64869297e-01 9.42787290e-01 2.94566695e-02 -3.09090644e-01
3.14682662e-01 -3.97727579e-01 -7.72828236e-02 -5.52178323e-01
9.52751264e-02 2.40434468e-01 -2.57683337e-01 -1.85805380e-01
-7.03708827e-01 -5.92523396e-01 -1.49345267e+00 9.11168754e-02
-1.10591662e+00 1.69422179e-01 3.34131867e-01 6.43526912e-01
1.28448337e-01 2.42185563e-01 9.48526204e-01 -7.36071646e-01
-6.56422377e-01 -1.04078650e+00 -5.25761187e-01 -7.79690817e-02
3.40752363e-01 -9.08343852e-01 -5.34401834e-01 -2.46399879e-01]
|
[8.805208206176758, -0.7994343042373657]
|
92878183-45e1-435e-b2be-ddf404fe73a2
|
unsupervised-salience-learning-for-person-re
| null | null |
http://openaccess.thecvf.com/content_cvpr_2013/html/Zhao_Unsupervised_Salience_Learning_2013_CVPR_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2013/papers/Zhao_Unsupervised_Salience_Learning_2013_CVPR_paper.pdf
|
Unsupervised Salience Learning for Person Re-identification
|
Human eyes can recognize person identities based on some small salient regions. However, such valuable salient information is often hidden when computing similarities of images with existing approaches. Moreover, many existing approaches learn discriminative features and handle drastic viewpoint change in a supervised way and require labeling new training data for a different pair of camera views. In this paper, we propose a novel perspective for person re-identification based on unsupervised salience learning. Distinctive features are extracted without requiring identity labels in the training procedure. First, we apply adjacency constrained patch matching to build dense correspondence between image pairs, which shows effectiveness in handling misalignment caused by large viewpoint and pose variations. Second, we learn human salience in an unsupervised manner. To improve the performance of person re-identification, human salience is incorporated in patch matching to find reliable and discriminative matched patches. The effectiveness of our approach is validated on the widely used VIPeR dataset and ETHZ dataset.
|
['Rui Zhao', 'Wanli Ouyang', 'Xiaogang Wang']
|
2013-06-01
| null | null | null |
cvpr-2013-6
|
['patch-matching']
|
['computer-vision']
|
[ 2.93261886e-01 -3.82755637e-01 3.11441887e-02 -4.07029003e-01
-2.85083562e-01 -5.80927253e-01 5.02400935e-01 3.47779661e-01
-5.76219738e-01 5.48305690e-01 4.16353643e-01 7.80055285e-01
-3.61056104e-02 -5.99700272e-01 -4.53463823e-01 -6.54147506e-01
2.00922772e-01 2.61933595e-01 1.64502189e-01 -6.31072521e-02
5.45457661e-01 4.28199649e-01 -1.85643518e+00 6.51079118e-02
6.49854124e-01 5.59808135e-01 2.30294362e-01 1.63388737e-02
3.10661405e-01 4.02600467e-01 -3.02389413e-01 -5.16640604e-01
7.22445190e-01 -3.08487892e-01 -7.32235432e-01 4.83451545e-01
9.65488434e-01 -2.01586470e-01 -2.26789430e-01 1.31533861e+00
5.29594123e-01 4.00733024e-01 4.86308545e-01 -1.20308268e+00
-4.22647685e-01 2.41959497e-01 -9.12516356e-01 3.93995613e-01
7.82191396e-01 4.68120873e-02 9.54591870e-01 -8.51615548e-01
6.81865036e-01 1.12327647e+00 8.40860844e-01 3.57013464e-01
-1.31626046e+00 -5.74251890e-01 2.36295611e-01 5.77716589e-01
-1.77023065e+00 -6.32939398e-01 1.21791744e+00 -4.75293517e-01
4.79258209e-01 2.51480669e-01 8.63176167e-01 8.26842666e-01
-4.05794501e-01 5.24307191e-01 1.18992531e+00 -3.20664167e-01
-1.77170672e-02 3.94497722e-01 5.04668802e-02 5.43915868e-01
3.99152964e-01 3.60496938e-01 -6.78572357e-01 -1.02000125e-01
7.67202914e-01 4.67348576e-01 -2.66370475e-01 -6.19423687e-01
-1.37491190e+00 6.47176743e-01 6.43939137e-01 2.91057318e-01
-2.98567712e-01 -2.13251755e-01 2.49309212e-01 1.21838532e-01
2.07959279e-01 4.96550977e-01 1.18912213e-01 2.42300212e-01
-9.78306770e-01 2.51502275e-01 2.44064167e-01 9.11532402e-01
1.03532350e+00 -2.16622010e-01 -1.87238663e-01 8.56214821e-01
2.00125024e-01 4.89957631e-01 4.02802765e-01 -6.93618000e-01
1.13473631e-01 7.72712350e-01 6.20448850e-02 -1.61580038e+00
-3.94724041e-01 -3.20121825e-01 -8.50694656e-01 -1.24971904e-02
4.62312579e-01 2.51882195e-01 -5.01388788e-01 1.58628964e+00
5.44246793e-01 4.07670975e-01 -3.45347747e-02 1.24188423e+00
7.75199950e-01 1.16082281e-01 5.22929132e-02 -1.36686042e-01
1.47211039e+00 -8.38179052e-01 -3.76798362e-01 -2.63766348e-01
2.85781980e-01 -9.30827677e-01 6.64379656e-01 1.02183908e-01
-7.92655051e-01 -8.35192084e-01 -7.73738742e-01 -7.81663954e-02
-2.99217522e-01 2.54681259e-01 4.33385491e-01 5.93524933e-01
-7.59673715e-01 5.30040026e-01 -2.52308577e-01 -6.45915687e-01
3.11777860e-01 4.84132499e-01 -7.54256129e-01 -1.87487721e-01
-9.48977590e-01 7.45859087e-01 3.39137375e-01 2.22205505e-01
-5.61395764e-01 -4.88069743e-01 -1.06096375e+00 -1.02264673e-01
2.67670989e-01 -6.39257312e-01 5.66392362e-01 -1.19495487e+00
-1.16234434e+00 1.13251567e+00 -3.11533451e-01 -2.64469922e-01
4.24069136e-01 -7.60319680e-02 -3.36640626e-01 3.53860319e-01
3.50092918e-01 8.29865992e-01 1.10087478e+00 -1.44370592e+00
-5.79759419e-01 -6.96345210e-01 5.83669320e-02 5.10489941e-01
-3.21814924e-01 8.06507915e-02 -3.68209183e-01 -8.63029301e-01
2.54747391e-01 -9.66561079e-01 -2.29933694e-01 -1.82287414e-02
-2.43470490e-01 -8.68548527e-02 4.93023694e-01 -7.76061952e-01
8.51094127e-01 -2.19056630e+00 2.53556371e-01 5.38541794e-01
2.25506872e-01 -2.06915066e-02 -8.86867642e-02 2.47438937e-01
-1.13003217e-02 -4.36270714e-01 -1.39450267e-01 -3.41398418e-01
-1.12721972e-01 -8.12117830e-02 -4.10387442e-02 7.95349598e-01
8.14644620e-02 7.06834733e-01 -9.02476668e-01 -8.21124315e-01
4.01396364e-01 3.46925110e-01 -5.78804493e-01 2.84227163e-01
4.47866648e-01 7.12624311e-01 -1.64365262e-01 7.90929556e-01
7.05214679e-01 -5.78228533e-02 1.17561005e-01 -6.26972437e-01
-1.57844335e-01 -2.64596701e-01 -1.47188318e+00 1.81437457e+00
-7.22700283e-02 4.71150547e-01 -2.91893840e-01 -1.03501904e+00
9.51917410e-01 1.44848078e-01 5.57059765e-01 -5.40476382e-01
2.56762803e-01 -9.45794135e-02 -3.22734475e-01 -3.29256117e-01
6.79227233e-01 6.86381683e-02 2.23273765e-02 3.39696616e-01
-7.27203637e-02 3.59998137e-01 5.29034100e-02 -8.87062773e-03
4.05283451e-01 9.80095565e-02 5.35388052e-01 -3.18063468e-01
8.77320826e-01 2.43164925e-03 7.68551171e-01 4.78864819e-01
-4.30696160e-01 8.10948133e-01 -1.54403389e-01 -7.16558635e-01
-1.08192337e+00 -8.44831705e-01 -1.52506068e-01 9.32572186e-01
8.78923953e-01 -4.55175608e-01 -7.02298701e-01 -5.92710733e-01
1.00574605e-01 1.67232007e-01 -6.47508442e-01 -1.15851022e-01
-4.12435770e-01 -5.84586203e-01 2.03057006e-01 4.65587437e-01
8.03903639e-01 -9.65980887e-01 -6.37152672e-01 -1.53789878e-01
-2.63960660e-01 -1.15123296e+00 -8.46916378e-01 -4.43363219e-01
-5.27902484e-01 -1.27077591e+00 -8.29994082e-01 -1.12723994e+00
1.16206419e+00 8.53062272e-01 7.85319507e-01 2.11315483e-01
-5.87030947e-01 6.00480855e-01 -2.80393869e-01 9.61411968e-02
3.17205817e-01 -9.66657028e-02 4.53244954e-01 4.40520167e-01
6.85433865e-01 -5.66950917e-01 -8.11342657e-01 5.21655142e-01
-4.20553327e-01 -5.00685982e-02 4.84416306e-01 9.07628655e-01
6.08094692e-01 1.36538684e-01 3.37391466e-01 -6.72331572e-01
2.11711928e-01 -1.06473267e-01 -5.51925838e-01 3.44584823e-01
-3.27312022e-01 -1.41185716e-01 3.84223402e-01 -5.34288228e-01
-1.07292116e+00 5.19958854e-01 3.53323042e-01 -3.63958776e-01
-4.27798837e-01 1.09224662e-01 -2.45903909e-01 -4.52819496e-01
5.67305684e-01 2.86224276e-01 -5.12522571e-02 -3.57807130e-01
2.29300991e-01 6.16566777e-01 6.58024251e-01 -5.18968880e-01
1.21882319e+00 6.77169919e-01 -1.77601799e-01 -7.54153907e-01
-7.00577855e-01 -8.78596425e-01 -1.22900844e+00 -2.50627935e-01
7.89643347e-01 -1.24131656e+00 -8.51728439e-01 3.53633225e-01
-8.89775932e-01 3.61220568e-01 -1.79022402e-01 6.00419581e-01
-2.41585359e-01 8.68220031e-01 -1.88720718e-01 -6.14827633e-01
-3.22953194e-01 -8.89259040e-01 1.07806587e+00 5.56838572e-01
-2.20252350e-01 -8.28508556e-01 1.70941755e-01 5.22800863e-01
1.03647307e-01 8.21089000e-02 2.09545866e-01 -4.37696874e-01
-5.07073700e-01 -2.80954927e-01 -3.57743561e-01 -5.41890450e-02
3.69027436e-01 -3.90660644e-01 -1.01425505e+00 -6.23686552e-01
-2.24032983e-01 -5.77167757e-02 7.21240461e-01 1.08867288e-01
7.19722033e-01 -1.70195878e-01 -5.00880063e-01 6.57198310e-01
1.31342554e+00 -1.36832863e-01 3.57942015e-01 4.05049980e-01
1.03042805e+00 8.81703019e-01 6.40201807e-01 4.75661606e-01
4.66055274e-01 8.98631513e-01 -1.31503597e-01 -1.87400356e-01
7.99558312e-02 -2.95047045e-01 2.45611578e-01 4.79371786e-01
-3.02816331e-01 5.91467023e-01 -6.16139174e-01 6.33424580e-01
-1.91224468e+00 -1.38209558e+00 2.23192722e-01 2.45793915e+00
7.59708285e-01 -2.77600437e-01 4.41335738e-01 6.07446767e-02
1.13937235e+00 2.93066557e-02 -3.17815930e-01 3.19305599e-01
-1.88441172e-01 -1.99020132e-01 4.38549936e-01 3.56344581e-01
-1.40366173e+00 1.00190425e+00 5.73541021e+00 4.38249677e-01
-8.86021376e-01 -6.87000081e-02 2.00353697e-01 -4.76340353e-02
-6.79584146e-02 1.06885105e-01 -8.45588863e-01 4.69250172e-01
-5.37155867e-02 -2.04931334e-01 4.06186074e-01 7.74471521e-01
-4.85198125e-02 -2.39788353e-01 -1.02952456e+00 1.65528965e+00
6.49761200e-01 -9.45210516e-01 -6.23009019e-02 -2.96339132e-02
7.68592954e-01 -6.85781777e-01 4.52104248e-02 -1.04153953e-01
-9.53998268e-02 -8.41910541e-01 3.87247354e-01 6.16535366e-01
3.86955589e-01 -8.35063756e-01 8.27497482e-01 4.82859574e-02
-1.38328767e+00 -4.39668335e-02 -6.95378959e-01 -1.57926127e-01
6.17300682e-02 3.39408070e-01 -6.78667247e-01 5.14070749e-01
8.29862833e-01 1.14428449e+00 -9.76021290e-01 1.32652521e+00
2.60599609e-02 -1.11340716e-01 -2.04506665e-01 3.26776475e-01
-2.65384704e-01 -3.32526237e-01 6.87173665e-01 9.31603432e-01
7.77994171e-02 1.71013027e-02 6.22653246e-01 8.18569839e-01
2.10226804e-01 3.27604473e-01 -7.64388263e-01 4.02512252e-01
4.92110610e-01 1.39031935e+00 -7.28859425e-01 -1.86791673e-01
-4.16670203e-01 1.36870515e+00 3.01577628e-01 1.55498058e-01
-7.15648472e-01 -6.79227412e-02 5.60849130e-01 1.20605886e-01
2.27412865e-01 -7.80795366e-02 2.94558215e-03 -1.47597277e+00
4.68703099e-02 -8.49351406e-01 4.94391739e-01 -4.62934226e-01
-1.58456039e+00 5.57253540e-01 6.08399883e-02 -1.60936165e+00
-3.78542393e-02 -2.09837303e-01 -4.36159730e-01 6.17746115e-01
-1.49122679e+00 -1.43662918e+00 -7.69531071e-01 1.00036573e+00
3.12793642e-01 -4.69447821e-01 7.25608468e-01 3.61652076e-01
-6.39667153e-01 8.19557905e-01 -2.05785304e-01 3.40262532e-01
1.11115932e+00 -1.11220360e+00 1.16397396e-01 9.41221297e-01
4.12493467e-01 9.42801237e-01 5.43312430e-01 -7.23953605e-01
-1.03139079e+00 -9.97910380e-01 7.73804367e-01 -3.29048365e-01
1.14897318e-01 -1.98767841e-01 -7.29727983e-01 5.78860879e-01
1.51573330e-01 6.51606470e-02 9.94110584e-01 2.73206115e-01
-5.15309632e-01 -2.30279893e-01 -1.27949154e+00 6.61458373e-01
1.15094495e+00 -6.48211956e-01 -8.78136277e-01 2.31654301e-01
8.12518001e-02 -1.91029370e-01 -8.67801607e-01 2.41859660e-01
6.33273661e-01 -8.84793282e-01 1.24988544e+00 -2.88884878e-01
-1.14205822e-01 -7.34288633e-01 -1.17052779e-01 -1.03784943e+00
-5.83324313e-01 -4.78445619e-01 4.25076097e-01 1.28830969e+00
-2.00116023e-01 -4.94384170e-01 7.73966730e-01 5.78471839e-01
3.97050709e-01 1.39359077e-02 -6.94553733e-01 -5.84286690e-01
-5.99993706e-01 3.40946883e-01 5.37931144e-01 1.18668234e+00
3.13416064e-01 3.26798499e-01 -7.27439165e-01 3.73840034e-01
1.17131817e+00 5.35869837e-01 9.62643087e-01 -1.50705540e+00
-1.61951438e-01 -1.21379100e-01 -9.47725415e-01 -7.37469137e-01
1.53656051e-01 -8.06082547e-01 -4.76174466e-02 -9.08116937e-01
7.06649542e-01 -2.90070087e-01 -3.52176487e-01 4.16995525e-01
-4.92218345e-01 6.18663192e-01 4.20743138e-01 5.23473859e-01
-7.63148248e-01 4.99841064e-01 8.37090969e-01 -2.79527664e-01
-2.24080712e-01 -1.62051558e-01 -6.61875904e-01 6.74431860e-01
7.73765445e-01 -2.41169751e-01 -2.33684421e-01 -2.14303404e-01
-1.51828766e-01 -4.05254334e-01 8.49576771e-01 -1.24985397e+00
5.16718268e-01 -7.32789710e-02 7.95752347e-01 -6.16725981e-01
2.74865001e-01 -1.05628741e+00 1.68025121e-01 1.94604367e-01
-2.12624446e-01 1.87128946e-01 -2.86214333e-02 5.90161085e-01
-3.11567277e-01 -2.58935481e-01 9.05715048e-01 -3.09044391e-01
-1.14457154e+00 3.15785140e-01 1.95666745e-01 -8.01864043e-02
1.18342710e+00 -4.96858597e-01 3.06903180e-02 -1.41315356e-01
-6.12263858e-01 9.94514897e-02 9.59225833e-01 3.83268088e-01
7.90929973e-01 -1.54051733e+00 -5.56447506e-01 5.20301700e-01
4.37662601e-01 -2.81606585e-01 5.31501710e-01 8.38158250e-01
-2.25904077e-01 1.17817581e-01 -6.97169304e-01 -6.84504330e-01
-1.70192742e+00 8.55846465e-01 2.62682378e-01 2.22079486e-01
-7.66902626e-01 6.74432576e-01 2.58337945e-01 -2.77338713e-01
2.10017264e-01 2.47852266e-01 -6.40932739e-01 2.51657695e-01
6.18900418e-01 2.07785293e-01 -4.12190050e-01 -1.22154140e+00
-5.26508093e-01 1.12015414e+00 -2.56027400e-01 1.44264519e-01
1.04454613e+00 -4.19100642e-01 -1.23586476e-01 1.36278138e-01
1.17441809e+00 2.93637812e-01 -1.14692366e+00 -6.03631198e-01
-4.84254071e-03 -9.95249927e-01 -1.55009538e-01 -2.97038764e-01
-1.10853338e+00 6.52307034e-01 8.80914509e-01 -3.42752934e-01
1.03583193e+00 -2.79636115e-01 6.51311636e-01 3.15311074e-01
5.13094544e-01 -1.27736151e+00 8.76648799e-02 -2.93581504e-02
6.91093326e-01 -1.62206900e+00 3.51725131e-01 -5.96234381e-01
-9.01512742e-01 9.28674221e-01 7.34879017e-01 -2.17548519e-01
4.18576062e-01 -2.87734538e-01 -7.06500933e-03 -1.45628929e-01
-3.18054929e-02 -7.06605673e-01 4.64854419e-01 7.97771335e-01
1.37591854e-01 -8.44881833e-02 -1.42920703e-01 3.66295606e-01
-2.28913069e-01 -2.86176801e-01 1.62947521e-01 7.64388442e-01
-3.89582813e-01 -9.99014378e-01 -5.95728576e-01 -6.62313402e-02
-2.84897182e-02 -1.34106919e-01 -4.96831208e-01 5.70589066e-01
2.55162477e-01 8.68550777e-01 1.98892057e-01 -4.65371132e-01
3.12395334e-01 -1.84594914e-01 7.03427494e-01 -5.79410374e-01
-6.91274405e-01 1.31104708e-01 -2.51920134e-01 -4.94244277e-01
-8.80083442e-01 -9.61196661e-01 -8.75436485e-01 -2.85736412e-01
-2.01771006e-01 8.10580403e-02 -3.20069455e-02 8.05768609e-01
2.92669833e-01 -3.00008245e-02 7.49288619e-01 -1.04791629e+00
-3.40856820e-01 -6.78567171e-01 -6.58233285e-01 1.10726464e+00
1.92299157e-01 -1.05417550e+00 -2.56338626e-01 4.45732087e-01]
|
[14.756058692932129, 1.0083731412887573]
|
1ced3f32-6c60-45f5-b056-95cd1f691ebe
|
text-diae-degradation-invariant-autoencoders
|
2203.04814
| null |
https://arxiv.org/abs/2203.04814v4
|
https://arxiv.org/pdf/2203.04814v4.pdf
|
Text-DIAE: A Self-Supervised Degradation Invariant Autoencoders for Text Recognition and Document Enhancement
|
In this paper, we propose a Text-Degradation Invariant Auto Encoder (Text-DIAE), a self-supervised model designed to tackle two tasks, text recognition (handwritten or scene-text) and document image enhancement. We start by employing a transformer-based architecture that incorporates three pretext tasks as learning objectives to be optimized during pre-training without the usage of labeled data. Each of the pretext objectives is specifically tailored for the final downstream tasks. We conduct several ablation experiments that confirm the design choice of the selected pretext tasks. Importantly, the proposed model does not exhibit limitations of previous state-of-the-art methods based on contrastive losses, while at the same time requiring substantially fewer data samples to converge. Finally, we demonstrate that our method surpasses the state-of-the-art in existing supervised and self-supervised settings in handwritten and scene text recognition and document image enhancement. Our code and trained models will be made publicly available at~\url{ http://Upon_Acceptance}.
|
['Dimosthenis Karatzas', 'Lluis Gomez', 'Josep Lladós', 'Yousri Kessentini', 'Alicia Fornés', 'Ali Furkan Biten', 'Andres Mafla', 'Sanket Biswas', 'Mohamed Ali Souibgui']
|
2022-03-09
| null | null | null | null |
['document-enhancement', 'scene-text-recognition']
|
['computer-vision', 'computer-vision']
|
[ 9.00468230e-01 -8.64551365e-02 1.00058913e-01 -5.71986496e-01
-9.83003914e-01 -4.58723903e-01 8.80331814e-01 -2.28707656e-01
-4.39400434e-01 4.05524671e-01 5.95333613e-02 -2.55672693e-01
-1.69651378e-02 -1.81606904e-01 -6.39329493e-01 -7.89155841e-01
3.25103551e-01 2.55725116e-01 -1.01601988e-01 -1.49111301e-01
4.10671860e-01 3.65497947e-01 -1.46703601e+00 6.08439684e-01
9.09742892e-01 9.76090610e-01 2.66571373e-01 1.00726688e+00
1.41919881e-01 1.02931988e+00 -4.54102576e-01 -6.88326001e-01
1.36420861e-01 -6.06522202e-01 -6.37993336e-01 6.76788330e-01
6.69499099e-01 -5.26001215e-01 -5.85700929e-01 9.97446299e-01
7.69464016e-01 7.31077716e-02 7.49836743e-01 -8.94394934e-01
-9.97289836e-01 5.53131223e-01 -6.35841966e-01 5.00698900e-03
-1.80532038e-01 1.17043026e-01 8.12404573e-01 -1.29405999e+00
4.13886607e-01 8.79065514e-01 7.77983606e-01 5.72897255e-01
-1.21142042e+00 -3.83159995e-01 1.11808293e-01 -8.35048482e-02
-1.11583388e+00 -9.07058597e-01 8.32434773e-01 -2.72716552e-01
8.20434153e-01 1.83936462e-01 -1.23063764e-02 1.43797076e+00
8.35617259e-02 1.37211823e+00 1.22743487e+00 -7.92658746e-01
3.14860716e-02 3.15259248e-01 -1.28621301e-02 8.19353938e-01
-1.44759759e-01 2.35367678e-02 -8.27381194e-01 3.24784189e-01
6.32310987e-01 -2.37088785e-01 -2.60291636e-01 -3.43396991e-01
-1.05378807e+00 6.43658936e-01 -1.85793862e-01 3.12618792e-01
-8.34669694e-02 1.17242418e-01 5.79488695e-01 4.99466151e-01
6.65540457e-01 1.16467826e-01 -2.66672194e-01 -2.61758417e-01
-1.39347374e+00 -2.22168252e-01 5.77314436e-01 1.07175887e+00
4.76298004e-01 3.84018868e-01 -5.18031120e-01 9.76101995e-01
2.20655620e-01 5.39697945e-01 5.59674442e-01 -6.18714809e-01
6.60316586e-01 3.49636436e-01 -8.49440023e-02 -4.47363973e-01
9.12753195e-02 -5.39419293e-01 -9.54071343e-01 3.98417354e-01
1.78625718e-01 -2.18789130e-01 -1.16709936e+00 1.37343597e+00
-1.30796373e-01 -2.79328674e-01 1.29005581e-01 7.62232304e-01
4.06929314e-01 4.85298723e-01 -1.42559230e-01 -6.92554712e-02
1.13507140e+00 -1.39673769e+00 -9.83222544e-01 -2.93797493e-01
3.39699835e-01 -1.12482023e+00 1.40195107e+00 6.34772837e-01
-1.26590955e+00 -6.00794137e-01 -1.26203167e+00 -2.65470147e-01
-2.85866112e-01 9.92039204e-01 3.67555976e-01 8.85830879e-01
-1.13635683e+00 7.80856311e-01 -9.33856308e-01 -3.02420199e-01
4.33078617e-01 2.91315287e-01 -6.77991733e-02 5.32548912e-02
-7.22923160e-01 8.06708574e-01 8.49249214e-02 1.02887027e-01
-1.00955164e+00 -4.42737997e-01 -7.09431946e-01 1.54214785e-01
2.23574758e-01 -3.15500259e-01 1.54468560e+00 -1.22811866e+00
-1.94848585e+00 1.09726095e+00 -1.06116429e-01 -2.78261125e-01
8.15652251e-01 -5.29576063e-01 -5.06987989e-01 2.19298080e-01
-2.31037989e-01 4.75412250e-01 1.60186410e+00 -1.22802448e+00
-4.54801321e-01 -2.57707983e-01 -4.07397032e-01 4.19025421e-01
-8.40033531e-01 3.04833859e-01 -7.45090246e-01 -1.17071652e+00
-2.12180674e-01 -5.69804430e-01 1.71783119e-01 1.86558455e-01
-5.24440289e-01 2.15697125e-01 1.24811041e+00 -7.18384206e-01
1.04057467e+00 -2.29800463e+00 1.01644069e-01 -1.30366758e-01
3.15447859e-02 4.37431216e-01 -2.83113420e-01 5.51104546e-01
-4.82889563e-02 -9.48232338e-02 -5.22366524e-01 -1.00005496e+00
2.50913471e-01 -2.42578790e-01 -5.33467352e-01 4.69171584e-01
4.43033546e-01 7.86386609e-01 -4.39799339e-01 -5.30534804e-01
4.84014183e-01 7.01558411e-01 -1.98430762e-01 1.54812500e-01
-1.07176401e-01 2.32779860e-01 -3.19534093e-01 7.03238726e-01
7.06081748e-01 -3.44619185e-01 1.38671845e-01 -4.22853768e-01
-2.06469595e-01 1.18820369e-01 -9.97646689e-01 1.65723121e+00
-4.29543465e-01 1.01446581e+00 1.75223991e-01 -1.05839503e+00
8.89608979e-01 2.59244055e-01 3.51236224e-01 -6.23640478e-01
3.32087249e-01 2.22638056e-01 -4.15509373e-01 -4.07130867e-01
7.79639602e-01 2.31633931e-01 2.70737797e-01 8.10427129e-01
4.62138832e-01 -1.25872061e-01 2.89989144e-01 1.76497951e-01
8.53984773e-01 3.56385499e-01 -5.60712107e-02 -2.54606664e-01
5.99869847e-01 -2.39172265e-01 -2.27498561e-02 8.36605668e-01
-7.31746554e-02 7.46232688e-01 2.64998019e-01 6.99238405e-02
-1.47541463e+00 -8.27091694e-01 -2.85858035e-01 1.32066607e+00
-1.27073646e-01 -2.08804235e-01 -7.86170840e-01 -7.20035911e-01
-2.22836897e-01 6.04781091e-01 -6.41289413e-01 3.89899313e-02
-3.97194296e-01 -6.46129370e-01 9.63162303e-01 6.57154918e-01
7.85179734e-01 -9.56514657e-01 -3.85753661e-01 -3.47006023e-02
2.71401647e-03 -1.20355105e+00 -9.15883183e-01 5.96901715e-01
-8.72282922e-01 -8.09533894e-01 -1.03648555e+00 -1.02714407e+00
9.25624490e-01 5.26897497e-02 6.88885212e-01 -2.04286262e-01
-3.32687378e-01 7.56564736e-01 -4.96952534e-01 -2.84965396e-01
-6.38928175e-01 6.66390806e-02 -1.63042173e-01 3.18245173e-01
2.02651337e-01 -3.35142106e-01 -5.24095833e-01 3.37104946e-01
-1.18478823e+00 1.60761118e-01 8.31082165e-01 1.28119230e+00
5.66768348e-01 9.17812958e-02 2.03660786e-01 -7.59185910e-01
7.58335531e-01 2.47595429e-01 -4.75063294e-01 6.11631453e-01
-7.84736753e-01 1.86566010e-01 7.06756711e-01 -5.42259037e-01
-1.43700135e+00 2.81452894e-01 -1.45313591e-02 -3.69922221e-01
-1.42165944e-01 3.15026343e-01 -5.00480197e-02 -2.69786775e-01
6.09322846e-01 6.45771682e-01 -3.12467590e-02 -5.03169000e-01
2.82336950e-01 1.03159952e+00 8.03304970e-01 -4.82481658e-01
8.18134606e-01 4.78648961e-01 -2.22128272e-01 -7.95631528e-01
-7.12247014e-01 -2.61536986e-01 -7.06606150e-01 -2.19217300e-01
6.30885601e-01 -8.20156038e-01 -4.22428638e-01 9.56340432e-01
-7.83772647e-01 -6.62994087e-01 -3.66039485e-01 2.62617141e-01
-7.79420972e-01 8.04228783e-01 -8.62903476e-01 -9.97109056e-01
-6.67705119e-01 -1.04798877e+00 1.31863976e+00 5.96552715e-02
2.40123332e-01 -1.05510521e+00 -1.44647419e-01 3.03462058e-01
5.62916279e-01 -1.29260883e-01 4.71555412e-01 -4.94142205e-01
-2.80377060e-01 -2.47180536e-01 -2.91417748e-01 7.41443574e-01
1.95402756e-01 1.55254565e-02 -1.30220199e+00 -6.60522580e-01
1.85882468e-02 -8.98158073e-01 1.22708273e+00 3.44408929e-01
1.26010823e+00 -1.69822052e-01 8.12042728e-02 7.33187139e-01
1.28192949e+00 4.44776081e-02 7.17069387e-01 4.00726527e-01
4.29119051e-01 4.34619039e-01 5.76868236e-01 6.35001063e-01
-3.30148563e-02 5.93299866e-01 3.99914384e-02 -4.75420713e-01
-3.95798653e-01 -3.40313017e-01 5.86360753e-01 7.01873362e-01
1.26105517e-01 -5.48655152e-01 -7.43640959e-01 4.03975755e-01
-1.79230499e+00 -8.04130435e-01 7.35471398e-02 2.04369068e+00
1.07471883e+00 3.18045914e-02 -1.69898152e-01 3.93825144e-01
8.21167290e-01 2.40314484e-01 -7.76249528e-01 -3.43895376e-01
-4.35544074e-01 3.13750356e-01 5.63125849e-01 3.51554811e-01
-1.55063832e+00 1.06111085e+00 6.25241280e+00 7.82362878e-01
-1.09473777e+00 -3.67765576e-02 7.27298379e-01 1.21145211e-01
8.98981616e-02 -1.36878684e-01 -7.77974427e-01 1.39816821e-01
6.61665857e-01 1.08384741e-02 5.80708027e-01 6.70964360e-01
9.21297669e-02 1.47591606e-01 -1.23074222e+00 8.94129932e-01
4.67826843e-01 -1.04641509e+00 4.01831493e-02 -2.80374408e-01
8.14628184e-01 -1.77362695e-01 6.19139850e-01 9.45995525e-02
2.82133818e-01 -8.90075207e-01 8.40967715e-01 3.94768327e-01
1.34276152e+00 -3.82886738e-01 3.23095292e-01 -4.63311002e-02
-9.57209527e-01 -1.00559942e-01 -1.04167186e-01 4.50453192e-01
7.57484809e-02 5.94309032e-01 -3.11616868e-01 4.75435108e-01
5.59856355e-01 8.24830770e-01 -5.97602725e-01 7.13063896e-01
-2.90170163e-01 8.55919719e-01 -1.95270255e-02 8.75956342e-02
2.91873187e-01 1.53007105e-01 4.16349888e-01 1.57290936e+00
2.21835688e-01 -1.83903158e-01 -8.29199702e-02 8.30911398e-01
-2.50067115e-01 3.29381414e-02 -2.68649459e-01 -3.47358167e-01
8.58944207e-02 1.11787021e+00 -5.27472496e-01 -3.88986409e-01
-3.51273656e-01 1.62404788e+00 3.90900522e-02 5.29725194e-01
-7.86991000e-01 -9.50454295e-01 -1.55404592e-02 -2.52194732e-01
6.49969399e-01 -1.24120288e-01 -3.84553730e-01 -1.43044114e+00
2.43496299e-01 -1.13461649e+00 1.46825269e-01 -8.49102259e-01
-1.26883090e+00 5.72423637e-01 -4.12081599e-01 -1.22502065e+00
-1.83282450e-01 -9.44799006e-01 -4.78865743e-01 7.47280240e-01
-1.64126670e+00 -1.42472982e+00 -2.91761339e-01 7.55652964e-01
8.82225156e-01 -4.39493209e-01 7.08068430e-01 3.53858799e-01
-6.96676493e-01 1.07296801e+00 6.53343320e-01 2.27620214e-01
1.13075328e+00 -1.27597034e+00 4.10511553e-01 1.14631033e+00
-1.97593849e-02 3.54559571e-01 5.85522771e-01 -4.89767134e-01
-1.41268659e+00 -1.04807854e+00 7.67467916e-01 -2.01440990e-01
6.31265879e-01 -5.67503750e-01 -6.79794073e-01 7.83053994e-01
4.11795139e-01 -3.03320289e-01 4.59954888e-01 -2.26695508e-01
-2.85209030e-01 -6.58177258e-03 -1.04356670e+00 6.24324024e-01
8.20783496e-01 -6.35628581e-01 -2.99222350e-01 3.50299180e-01
3.25313777e-01 -5.06772816e-01 -6.92773938e-01 2.66837150e-01
5.16127765e-01 -7.60765910e-01 7.33682394e-01 -4.23120975e-01
8.67708147e-01 6.17446890e-03 -1.60197929e-01 -9.73326802e-01
-2.44471893e-01 -9.03760850e-01 -3.08572054e-01 1.56431353e+00
4.30468291e-01 -3.56395155e-01 7.92900383e-01 3.90900046e-01
-2.90278763e-01 -4.18231606e-01 -6.65434659e-01 -7.53512681e-01
2.04678625e-02 -2.33726859e-01 1.99784160e-01 1.05809391e+00
2.24618986e-02 2.41272613e-01 -9.01207685e-01 -3.07371709e-02
8.23218286e-01 4.63094302e-02 6.20623171e-01 -7.09548652e-01
-5.02202809e-01 -4.71825659e-01 -3.20473500e-02 -1.33879220e+00
-1.13916092e-01 -7.78040349e-01 2.41520107e-01 -1.09657967e+00
3.58522207e-01 -1.22667305e-01 -1.92576095e-01 7.28927016e-01
-1.12157948e-01 3.34318191e-01 1.78364992e-01 1.92165971e-01
-7.10630655e-01 8.94222021e-01 1.10055268e+00 -4.19565827e-01
-1.83178782e-01 -9.73936841e-02 -6.33617878e-01 3.32807422e-01
1.00400460e+00 -1.12357803e-01 -3.85026872e-01 -7.30447173e-01
-2.13272899e-01 -1.77616298e-01 2.78422475e-01 -8.61267209e-01
3.71819109e-01 2.19855115e-01 5.76217175e-01 -6.59027338e-01
2.76002854e-01 -6.31575286e-01 -4.49761152e-01 1.85596719e-01
-8.65064919e-01 -1.02229513e-01 4.26954895e-01 4.24425542e-01
-2.31152415e-01 -4.75368708e-01 9.57913041e-01 3.74957800e-01
-4.38596249e-01 1.65325522e-01 -6.11792684e-01 -9.33361575e-02
5.79964936e-01 -2.49433115e-01 -3.88949722e-01 -4.72632706e-01
-5.38714945e-01 6.15695305e-03 4.22466457e-01 2.76866794e-01
8.14209461e-01 -1.00714648e+00 -8.85639489e-01 2.05294698e-01
1.39939412e-01 -4.66256052e-01 1.89046621e-01 7.93345451e-01
-3.11756670e-01 5.15859425e-01 -1.36840045e-01 -4.68657881e-01
-1.40036929e+00 4.27274883e-01 3.75581175e-01 -5.32890916e-01
-7.27862239e-01 6.03156030e-01 2.91287005e-02 -4.40589905e-01
7.14453280e-01 -1.14603089e-02 1.38162300e-01 -5.00706792e-01
4.68371719e-01 3.26917291e-01 2.06454515e-01 -2.60902047e-01
6.66552186e-02 5.70755482e-01 -5.12817323e-01 -3.34784657e-01
1.40176070e+00 -2.69859463e-01 2.61728644e-01 9.64289606e-02
1.05081773e+00 -4.50829975e-02 -1.68210661e+00 -3.74887288e-01
-2.10537076e-01 -3.42535824e-01 4.06483859e-01 -1.16458964e+00
-1.03838384e+00 9.40319300e-01 8.01166296e-01 -3.19909364e-01
1.65111876e+00 -3.77807319e-01 5.01963973e-01 7.59558916e-01
-2.26698980e-01 -1.49964929e+00 4.82309014e-01 4.17406708e-01
9.58133817e-01 -1.26495528e+00 2.06296235e-01 -2.12162852e-01
-9.79645193e-01 1.25238955e+00 5.24394870e-01 1.48149058e-01
3.49800289e-01 6.07893586e-01 7.47784376e-02 -9.26872939e-02
-7.14673936e-01 -1.22503683e-01 3.34917814e-01 5.52897036e-01
6.98665380e-01 -4.78212684e-01 -1.82354569e-01 3.79678696e-01
1.31091565e-01 9.99581292e-02 4.12709296e-01 1.20043647e+00
-7.87328705e-02 -1.34577990e+00 -2.68488914e-01 4.58908737e-01
-6.10240698e-01 -3.47323269e-01 -6.62560344e-01 5.11451423e-01
-3.78699511e-01 7.82398582e-01 -1.02726147e-01 -3.15083265e-01
2.34268278e-01 1.27348915e-01 6.14544392e-01 -1.97803319e-01
-5.49234509e-01 1.39546722e-01 -9.13278684e-02 -9.33338553e-02
-4.86326754e-01 -6.61073565e-01 -7.15270281e-01 -2.66531467e-01
-7.13984847e-01 -3.11316878e-01 7.04052925e-01 5.07442415e-01
3.42129260e-01 6.07395947e-01 8.91215920e-01 -8.72940063e-01
-9.23766434e-01 -9.59362268e-01 -6.58411324e-01 3.41959566e-01
3.85006458e-01 -3.13776910e-01 -2.82564819e-01 6.55423701e-01]
|
[11.846346855163574, 2.2009079456329346]
|
fb1f8925-6271-41c5-be5c-110c34e34546
|
antman-sparse-low-rank-compression-to-1
|
1910.01740
| null |
https://arxiv.org/abs/1910.01740v1
|
https://arxiv.org/pdf/1910.01740v1.pdf
|
AntMan: Sparse Low-Rank Compression to Accelerate RNN inference
|
Wide adoption of complex RNN based models is hindered by their inference performance, cost and memory requirements. To address this issue, we develop AntMan, combining structured sparsity with low-rank decomposition synergistically, to reduce model computation, size and execution time of RNNs while attaining desired accuracy. AntMan extends knowledge distillation based training to learn the compressed models efficiently. Our evaluation shows that AntMan offers up to 100x computation reduction with less than 1pt accuracy drop for language and machine reading comprehension models. Our evaluation also shows that for a given accuracy target, AntMan produces 5x smaller models than the state-of-art. Lastly, we show that AntMan offers super-linear speed gains compared to theoretical speedup, demonstrating its practical value on commodity hardware.
|
['Samyam Rajbhandari', 'Harsh Shrivastava', 'Yuxiong He']
|
2019-10-02
|
antman-sparse-low-rank-compression-to
|
https://openreview.net/forum?id=BJgsN3R9Km
|
https://openreview.net/pdf?id=BJgsN3R9Km
|
iclr-2019-5
|
['low-rank-compression']
|
['computer-code']
|
[ 2.23371565e-01 1.84886858e-01 -5.77338576e-01 -3.59824359e-01
-9.62558925e-01 -4.40393627e-01 2.33633712e-01 -7.88220465e-02
-5.41326165e-01 5.40174663e-01 4.58558887e-01 -6.55614316e-01
-2.66787201e-01 -5.16588211e-01 -8.58275950e-01 -1.09760672e-01
3.00779164e-01 5.95403492e-01 -3.19458008e-01 1.25505462e-01
1.49905346e-02 2.33992845e-01 -1.50101197e+00 4.29258257e-01
9.40952063e-01 7.61964262e-01 4.08865631e-01 9.79032099e-01
8.25111270e-02 1.45726895e+00 -2.77073115e-01 -2.75327623e-01
2.69586295e-01 9.81317610e-02 -1.08483088e+00 -3.43321979e-01
8.64547074e-01 -6.61906123e-01 -7.12988138e-01 6.85081661e-01
2.75173753e-01 3.43136013e-01 2.39629179e-01 -7.59507000e-01
-6.20389879e-01 9.34228182e-01 -4.30642694e-01 4.56966609e-01
7.06302747e-02 1.14828557e-01 1.35757697e+00 -8.61206830e-01
2.38895833e-01 1.19247675e+00 5.88635385e-01 6.15149796e-01
-1.37866640e+00 -7.72362947e-01 2.63341486e-01 3.01757634e-01
-1.26316547e+00 -8.42867494e-01 6.99465722e-02 9.70169380e-02
1.65811253e+00 4.57428306e-01 8.32618296e-01 9.22237337e-01
-1.79147065e-01 1.24414420e+00 9.44918096e-01 -4.39129382e-01
-5.36612794e-03 -2.21740037e-01 7.29632020e-01 8.30923676e-01
3.68847489e-01 -1.27832085e-01 -9.56815898e-01 -1.14913531e-01
5.39164186e-01 -1.03409834e-01 -2.90372610e-01 2.08591193e-01
-8.31429064e-01 5.97644091e-01 2.48436660e-01 1.17577255e-01
-3.62065732e-01 7.09197402e-01 4.00410086e-01 2.43275449e-01
3.40992481e-01 6.22745574e-01 -9.04605865e-01 -8.58233035e-01
-1.29820514e+00 2.45401278e-01 1.09078336e+00 1.16251695e+00
3.75459075e-01 5.11846185e-01 -5.48266694e-02 8.82738173e-01
1.57283053e-01 7.20757663e-01 5.43618739e-01 -1.22048938e+00
6.71387494e-01 4.92196977e-01 -2.47959018e-01 -8.48281980e-01
-5.91211677e-01 -9.44943190e-01 -9.43993509e-01 -5.66441894e-01
2.27647558e-01 -4.56699394e-02 -9.30522859e-01 1.65646386e+00
6.77096024e-02 1.14512511e-01 3.12574118e-01 5.87218940e-01
7.74019063e-01 7.86401033e-01 -1.08993128e-02 -1.54475346e-01
1.27608883e+00 -1.36993229e+00 -5.09509027e-01 -6.24660015e-01
1.07184184e+00 -4.70974773e-01 1.07334173e+00 7.19039500e-01
-1.44119716e+00 -2.99828827e-01 -7.99463212e-01 -5.79150438e-01
1.95580140e-01 3.90649885e-01 1.06241548e+00 4.89054382e-01
-1.05742192e+00 4.66322482e-01 -1.35127437e+00 -6.31172583e-02
5.96304357e-01 6.32053137e-01 -4.64112870e-02 -4.29637462e-01
-7.54507720e-01 1.03911984e+00 3.58574092e-01 5.48062474e-02
-9.61167812e-01 -1.24984658e+00 -6.43327832e-01 3.97601038e-01
2.68001705e-01 -7.52065957e-01 1.59208512e+00 -4.14401799e-01
-1.41084254e+00 3.68879855e-01 -4.18828100e-01 -1.09293258e+00
3.67427506e-02 -7.35976577e-01 -2.71070302e-01 -6.97951168e-02
-4.33799744e-01 6.49477184e-01 4.96721745e-01 -6.35209501e-01
-5.52815259e-01 -2.73430347e-01 1.98893502e-01 3.59020025e-01
-7.42529392e-01 -1.50254920e-01 -6.69119596e-01 -3.60998422e-01
3.27117831e-01 -1.08800566e+00 -3.06956112e-01 -4.83541489e-01
-3.28380108e-01 -2.01782495e-01 5.64829528e-01 -9.99481797e-01
1.37057626e+00 -1.78652644e+00 8.33943784e-02 1.94906995e-01
6.73432767e-01 4.24606562e-01 -3.96511346e-01 6.17727898e-02
3.31728071e-01 -5.80006912e-02 1.10866398e-01 -3.22900325e-01
-1.13720633e-01 4.87267524e-01 -3.84871513e-01 4.32318300e-02
-2.56592631e-01 1.09154296e+00 -6.61840737e-01 -1.37962461e-01
-9.04121548e-02 6.16559565e-01 -9.22696531e-01 -4.66062874e-02
-3.26295584e-01 -1.62680030e-01 -2.81548232e-01 6.07041180e-01
4.85909045e-01 -6.50339961e-01 4.01999891e-01 -1.78345963e-01
2.87287533e-01 8.62603128e-01 -7.93630242e-01 1.69595397e+00
-8.59401464e-01 9.30183172e-01 1.35060072e-01 -8.21664691e-01
5.59054971e-01 6.19118027e-02 1.10760078e-01 -6.32148683e-01
-7.52647519e-02 9.19350088e-02 1.36201531e-01 -1.32380292e-01
7.04726577e-01 3.30114096e-01 4.95552212e-01 7.12171793e-01
1.15543246e-01 1.18778400e-01 2.52137005e-01 4.56209511e-01
1.43391061e+00 -7.24787265e-02 5.55667281e-02 -2.45006606e-01
-2.00662650e-02 1.14132524e-01 3.47470522e-01 8.79540324e-01
2.70012677e-01 2.44053937e-02 2.89245218e-01 -4.94421780e-01
-9.19643819e-01 -7.67535269e-01 2.24314965e-02 1.65091002e+00
-5.65665960e-01 -6.60531461e-01 -7.73340404e-01 -2.84736484e-01
-6.74421638e-02 8.77326846e-01 -1.19628832e-01 -5.27330376e-02
-7.48528719e-01 -1.00531125e+00 8.35245728e-01 9.08350468e-01
7.01565325e-01 -5.12328207e-01 -6.33979082e-01 7.40154609e-02
-2.42106929e-01 -1.43854713e+00 -2.61347055e-01 2.88978428e-01
-1.37571371e+00 -7.92053103e-01 -2.27452412e-01 -7.50371933e-01
6.64987564e-01 4.65470046e-01 1.44858158e+00 1.89818978e-01
-1.57199174e-01 3.70160311e-01 -8.60643834e-02 -3.15645367e-01
-1.34068653e-01 6.74275935e-01 2.64261037e-01 -8.42688501e-01
5.52990377e-01 -7.15493441e-01 -4.98610228e-01 -2.17710748e-01
-6.47277176e-01 5.34769297e-01 8.21248949e-01 8.54564786e-01
6.00161195e-01 3.00413445e-02 3.58358622e-01 -9.41417336e-01
6.46310031e-01 -2.11047351e-01 -7.31897831e-01 4.32053596e-01
-1.07226717e+00 3.27888936e-01 6.39601529e-01 -5.69546521e-01
-8.58817995e-01 1.87554985e-01 -6.48002401e-02 -4.50470567e-01
3.83322358e-01 8.22618425e-01 2.73043334e-01 3.19852494e-02
8.04735541e-01 5.20426869e-01 -5.36623076e-02 -5.25047064e-01
5.47093391e-01 5.31516671e-01 5.18179119e-01 -5.62311113e-01
5.84881604e-01 9.82704312e-02 -1.90602273e-01 -8.18546653e-01
-1.43068027e+00 -3.53140861e-01 -2.64034867e-01 3.46823394e-01
3.93079489e-01 -1.52753186e+00 -8.15623701e-01 -4.50229906e-02
-1.06868529e+00 -5.84366024e-01 -1.89635932e-01 7.00955391e-01
-4.15050477e-01 -6.68270215e-02 -1.07442963e+00 -7.25077093e-01
-1.06269050e+00 -8.55838001e-01 6.62085950e-01 -1.37685305e-02
-3.09160143e-01 -8.50241482e-01 -1.38646141e-01 1.09671962e+00
5.37522256e-01 -6.30286276e-01 1.06169200e+00 -7.56482065e-01
-8.01970541e-01 -1.27637610e-01 -4.73307610e-01 3.64974499e-01
-4.65699881e-01 -4.04765099e-01 -1.00599718e+00 -4.81138200e-01
-4.66538928e-02 -6.82631493e-01 9.51219022e-01 3.56707960e-01
1.34768379e+00 -7.06973314e-01 -1.68766648e-01 7.69805789e-01
1.30041051e+00 -1.83681220e-01 4.38828379e-01 -5.98206185e-02
1.13899934e+00 1.42581791e-01 1.62342861e-01 2.90567517e-01
5.32250166e-01 3.83637875e-01 1.13944411e-01 1.42312795e-01
-1.72268018e-01 -4.37477648e-01 4.35012341e-01 1.67105019e+00
-1.85551316e-01 -1.77132159e-01 -1.22953761e+00 4.81708944e-01
-1.78808713e+00 -7.31548071e-01 -1.11154668e-01 1.96005929e+00
1.04284286e+00 1.79772839e-01 -9.87860858e-02 3.04150730e-02
2.72696428e-02 1.35319084e-01 -8.94638181e-01 -5.40677309e-01
4.74040359e-02 3.94847602e-01 7.45295823e-01 6.77441418e-01
-5.19641638e-01 1.29404140e+00 7.70836067e+00 1.17262864e+00
-7.78599977e-01 3.86384755e-01 7.45277345e-01 -7.03836501e-01
-3.70558560e-01 -1.25126764e-01 -1.13161087e+00 -1.60542265e-01
1.51487315e+00 -1.97561055e-01 1.00699222e+00 1.09059310e+00
5.04910275e-02 9.57827345e-02 -1.20198989e+00 1.27369237e+00
4.85558882e-02 -1.61334860e+00 2.53984392e-01 1.12106405e-01
8.68504405e-01 6.25023305e-01 9.16703939e-02 6.21430457e-01
8.05164158e-01 -1.48892152e+00 1.53371781e-01 3.84466171e-01
5.30340374e-01 -8.89031768e-01 4.65008289e-01 6.37243569e-01
-1.05341160e+00 -1.85888588e-01 -8.14037263e-01 -4.82655853e-01
-1.70151055e-01 6.05350852e-01 -1.02156520e+00 4.13978193e-03
4.32436854e-01 6.47779405e-01 -5.06410956e-01 4.36882317e-01
-1.99500278e-01 1.29163802e+00 -5.25709331e-01 -5.37037142e-02
2.01383576e-01 1.16838805e-01 1.35125920e-01 1.30574059e+00
2.20072623e-02 3.55028093e-01 1.02972277e-02 7.05273867e-01
-5.41850567e-01 3.43287811e-02 -4.15819019e-01 -1.45826429e-01
7.46853054e-01 1.09872603e+00 -2.62492627e-01 -4.89941090e-01
-2.76191324e-01 7.65525103e-01 8.88024569e-01 2.99157172e-01
-7.54854679e-01 8.20577983e-03 5.97720206e-01 6.28960729e-02
2.03400359e-01 -5.18094838e-01 -6.18172944e-01 -1.23543227e+00
3.73204947e-02 -1.28116393e+00 2.07307383e-01 -7.97297955e-01
-7.92112947e-01 5.16843140e-01 -1.63609281e-01 -5.16653776e-01
-2.40660936e-01 -4.96332586e-01 -3.27495746e-02 4.76777524e-01
-1.20574427e+00 -1.07810163e+00 -1.66969597e-01 2.76945502e-01
6.67759001e-01 -3.06979299e-01 1.01965463e+00 1.17773287e-01
-8.44709635e-01 1.09172547e+00 5.01474679e-01 1.15481943e-01
-1.49614492e-03 -9.59891438e-01 5.36314130e-01 8.57669055e-01
4.93446290e-01 7.42154360e-01 4.96721447e-01 -4.48465526e-01
-2.02773142e+00 -1.16704333e+00 1.08534658e+00 -6.26742303e-01
6.85581684e-01 -2.93765157e-01 -7.42521465e-01 1.01232862e+00
4.20407392e-02 -1.45693734e-01 8.96524012e-01 7.29317844e-01
-7.10460603e-01 -2.61626214e-01 -9.16805208e-01 7.56918967e-01
1.28815722e+00 -7.75347114e-01 -3.86789262e-01 6.88562453e-01
9.52423096e-01 -7.20727980e-01 -1.22086525e+00 2.82616138e-01
6.41084135e-01 -4.88344043e-01 9.99987006e-01 -8.41905713e-01
7.08688915e-01 2.90081888e-01 -3.97029579e-01 -1.01563084e+00
-4.60837990e-01 -5.51455379e-01 -9.35424805e-01 7.40900099e-01
7.18632400e-01 -3.67445141e-01 1.21776474e+00 1.11261928e+00
8.31437930e-02 -1.18057251e+00 -8.65598917e-01 -7.16967762e-01
6.90055415e-02 -9.65958595e-01 5.13396263e-01 7.45070457e-01
7.92189166e-02 7.50217617e-01 -4.82165188e-01 7.94910640e-02
5.57246923e-01 -7.10361451e-02 7.08908856e-01 -9.59861100e-01
-6.90092325e-01 -2.56099641e-01 -1.66455597e-01 -1.73212719e+00
3.56023401e-01 -1.07198703e+00 -3.47062737e-01 -1.68778110e+00
6.03283167e-01 -4.62999225e-01 -2.43160248e-01 9.22886252e-01
-3.46682370e-02 1.16646312e-01 4.13767815e-01 3.30547094e-01
-8.75926435e-01 3.18382800e-01 1.18172729e+00 -1.21583343e-01
-1.93571851e-01 -3.29228759e-01 -9.06734467e-01 7.37121403e-01
8.15245569e-01 -3.77787411e-01 -8.67336392e-01 -1.10716653e+00
4.24902946e-01 9.18909311e-02 -6.63414598e-02 -1.22740376e+00
4.30260956e-01 1.30261749e-01 4.03132290e-01 -6.28847957e-01
5.49881697e-01 -6.04152858e-01 -1.98198214e-01 6.27402604e-01
-8.66886973e-01 2.17500135e-01 5.15415192e-01 4.94010299e-01
8.79966319e-02 -1.61649421e-01 5.19533813e-01 -1.76146328e-02
-2.75579244e-01 1.53887197e-01 -4.65933830e-01 3.85107249e-01
4.43578541e-01 9.46154669e-02 -3.95436525e-01 -4.79998976e-01
-5.39327562e-01 3.45499128e-01 -1.60608962e-01 1.61650524e-01
7.06994891e-01 -9.72117066e-01 -7.78282404e-01 6.00985736e-02
-3.15418631e-01 2.96402961e-01 2.14507237e-01 7.60916650e-01
-5.86440384e-01 8.95715773e-01 3.56275588e-01 -5.09438753e-01
-1.52023375e+00 2.40021750e-01 2.18930721e-01 -7.71115422e-01
-5.96031427e-01 1.08065736e+00 -1.44535929e-01 -5.20590603e-01
5.07508993e-01 -6.49492264e-01 1.13311328e-01 -3.95728588e-01
7.97879457e-01 8.95485282e-01 1.58295020e-01 -6.03815205e-02
-1.07779518e-01 3.66234072e-02 -6.04566157e-01 7.79815987e-02
1.36402845e+00 2.10036963e-01 -1.91109166e-01 1.40560567e-01
1.09803450e+00 -1.15820542e-01 -1.10201395e+00 -4.49168414e-01
-4.21120094e-05 -1.11386247e-01 5.31532943e-01 -9.70623791e-01
-1.13369656e+00 5.98888040e-01 4.31990802e-01 -3.54800791e-01
1.13710964e+00 -2.19805896e-01 1.14721715e+00 1.24480367e+00
3.26898158e-01 -1.04248500e+00 -2.33461738e-01 9.31965709e-01
5.52434087e-01 -1.17175579e+00 4.86814469e-01 -1.86825231e-01
-4.86224860e-01 7.65353084e-01 8.34058821e-01 5.43999150e-02
5.31613469e-01 4.21016812e-01 -2.22249821e-01 -1.06855236e-01
-1.43896091e+00 2.62368768e-01 3.68134290e-01 3.11422348e-01
4.98152822e-01 2.57689059e-01 1.76892038e-02 6.68381512e-01
-8.69244993e-01 1.33379340e-01 3.02034944e-01 6.70809448e-01
-4.99288678e-01 -7.28159726e-01 -1.65657237e-01 1.13005352e+00
-3.97566438e-01 -7.14389026e-01 5.01548462e-02 6.17486298e-01
-3.95321220e-01 1.01818728e+00 1.53570741e-01 -7.22406387e-01
1.79296985e-01 2.02386081e-02 6.44522190e-01 -5.86656332e-01
-8.06516051e-01 -2.31255695e-01 3.66908491e-01 -7.15393722e-01
-8.35645050e-02 -2.75205582e-01 -1.37680042e+00 -8.44445467e-01
-4.25360501e-01 -7.41010997e-04 5.86016536e-01 1.08049572e+00
6.73701167e-01 4.05174166e-01 5.30888028e-02 -3.24161023e-01
-9.36296999e-01 -1.08053994e+00 -9.89989415e-02 -2.83071399e-01
1.69018671e-01 6.94247782e-02 -2.03253195e-01 -1.38719767e-01]
|
[8.77525806427002, 3.6551153659820557]
|
3c938875-5d9c-49b6-a5a1-cd3589a55570
|
a-zero-few-shot-anomaly-classification-and
|
2305.17382
| null |
https://arxiv.org/abs/2305.17382v2
|
https://arxiv.org/pdf/2305.17382v2.pdf
|
A Zero-/Few-Shot Anomaly Classification and Segmentation Method for CVPR 2023 VAND Workshop Challenge Tracks 1&2: 1st Place on Zero-shot AD and 4th Place on Few-shot AD
|
In this technical report, we briefly introduce our solution for the Zero/Few-shot Track of the Visual Anomaly and Novelty Detection (VAND) 2023 Challenge. For industrial visual inspection, building a single model that can be rapidly adapted to numerous categories without or with only a few normal reference images is a promising research direction. This is primarily because of the vast variety of the product types. For the zero-shot track, we propose a solution based on the CLIP model by adding extra linear layers. These layers are used to map the image features to the joint embedding space, so that they can compare with the text features to generate the anomaly maps. Besides, when the reference images are available, we utilize multiple memory banks to store their features and compare them with the features of the test images during the testing phase. In this challenge, our method achieved first place in the zero-shot track, especially excelling in segmentation with an impressive F1 score improvement of 0.0489 over the second-ranked participant. Furthermore, in the few-shot track, we secured the fourth position overall, with our classification F1 score of 0.8687 ranking first among all participating teams.
|
['Jiangning Zhang', 'Yue Han', 'Xuhai Chen']
|
2023-05-27
| null | null | null | null |
['anomaly-classification']
|
['computer-vision']
|
[ 2.91334484e-02 -9.50814709e-02 1.34601876e-01 -1.63345784e-01
-6.75702035e-01 -3.36509705e-01 3.94690692e-01 2.39619061e-01
-2.57805824e-01 -1.20307561e-02 -4.58739460e-01 -2.88752392e-02
-2.69622896e-02 -3.80032331e-01 -7.58116007e-01 -4.92735922e-01
3.92090045e-02 5.52442968e-02 4.74803239e-01 -3.02833647e-01
4.08099145e-01 2.40366355e-01 -1.94638371e+00 4.61148530e-01
9.44153190e-01 1.49826348e+00 1.16506040e-01 6.15970254e-01
-6.80641532e-02 5.16454160e-01 -7.71715581e-01 -7.13166833e-01
4.58383769e-01 -2.93758214e-01 -4.24125791e-01 2.17622489e-01
9.07767177e-01 -3.24039936e-01 -1.52233541e-01 1.29604697e+00
3.73902470e-01 1.12441614e-01 5.79589963e-01 -1.60434544e+00
-6.61149621e-01 2.13409156e-01 -8.71612430e-01 2.33535588e-01
4.43346709e-01 5.17823935e-01 8.80060375e-01 -1.44641137e+00
6.69331014e-01 8.37904572e-01 5.98570228e-01 4.13834900e-01
-8.98958802e-01 -5.51768839e-01 3.43923450e-01 7.79513896e-01
-1.35010946e+00 -1.40412822e-01 7.72955835e-01 -7.70637155e-01
8.11974108e-01 1.77832216e-01 6.02352679e-01 1.28305876e+00
2.42700011e-01 8.39483142e-01 6.36300802e-01 -2.50035733e-01
2.04097450e-01 4.45698977e-01 2.73866713e-01 6.36128128e-01
2.93930113e-01 -1.54562697e-01 -6.52488112e-01 1.93708181e-01
2.37492695e-01 1.11327358e-01 -1.00378133e-01 -6.08433068e-01
-1.12702966e+00 6.10417783e-01 5.59451818e-01 2.11694792e-01
-4.64072585e-01 -3.45555067e-01 4.86842185e-01 4.77124155e-01
2.09916055e-01 5.99363923e-01 -1.89823642e-01 5.88569464e-03
-9.89201069e-01 2.20496431e-01 4.58327979e-01 9.21686769e-01
3.91722739e-01 7.46627003e-02 -4.65988934e-01 7.74478495e-01
1.62210345e-01 2.58237690e-01 6.58465862e-01 -6.79893315e-01
5.76117873e-01 7.17119992e-01 1.81439504e-01 -1.20469534e+00
-1.53107271e-01 -6.29200995e-01 -9.10008967e-01 5.83886981e-01
4.37962830e-01 2.24948660e-01 -1.07971764e+00 1.25080073e+00
2.31313780e-01 1.84462354e-01 -1.96675688e-01 8.10449064e-01
5.69794059e-01 6.33120894e-01 -9.01081488e-02 -5.00353687e-02
1.40894115e+00 -1.22623241e+00 -5.93395054e-01 -3.55110407e-01
5.81730664e-01 -8.30273807e-01 1.40700996e+00 8.39170873e-01
-8.84351850e-01 -9.66756880e-01 -1.58064127e+00 2.81396806e-01
-6.46532953e-01 2.50294894e-01 1.67046309e-01 4.04780805e-01
-8.95225704e-01 7.08636761e-01 -6.98664427e-01 -3.29185486e-01
4.59169179e-01 -7.28212995e-03 -4.19538736e-01 -3.43438298e-01
-9.15666521e-01 8.27343583e-01 2.79673785e-01 1.13373637e-01
-8.97126138e-01 -7.26678967e-01 -8.37710857e-01 -2.86448635e-02
3.70268434e-01 -1.62566692e-01 1.05060112e+00 -5.63736439e-01
-1.04278970e+00 6.38987303e-01 1.71367675e-01 -3.40040654e-01
7.24442959e-01 -3.19132835e-01 -6.19966209e-01 2.27906425e-02
1.92472801e-01 5.76352060e-01 1.12500310e+00 -9.57860351e-01
-9.40210104e-01 -3.36803079e-01 -1.61318228e-01 -2.96911765e-02
-6.49178386e-01 -1.26518205e-01 -5.57300627e-01 -4.37442362e-01
2.34648094e-01 -6.32155240e-01 -1.39626250e-01 1.43461749e-01
-3.63818318e-01 -2.96193153e-01 9.22733724e-01 -7.61845291e-01
1.14926922e+00 -2.71067953e+00 -8.61031860e-02 2.49779567e-01
2.79715329e-01 3.30350339e-01 -1.42910123e-01 6.07423894e-02
-3.29158247e-01 -3.42678949e-02 -4.64525633e-02 -2.23413199e-01
1.39688507e-01 -3.87732208e-01 -1.75794095e-01 3.33820105e-01
5.94632566e-01 7.69720137e-01 -8.16877604e-01 -3.31862748e-01
3.22620958e-01 -3.07606440e-03 -3.10001284e-01 2.50069469e-01
3.25357020e-02 1.12841517e-01 -1.84710950e-01 6.88312829e-01
6.44704640e-01 -1.61016271e-01 -2.87136734e-01 -4.64569032e-01
-5.78399561e-02 -2.93009132e-01 -1.34048152e+00 1.59974658e+00
-2.09709615e-01 5.38733661e-01 -1.99048728e-01 -9.00904655e-01
1.04556477e+00 3.00104469e-01 4.84772772e-01 -8.49455357e-01
-4.14977521e-02 3.67052406e-01 3.61602455e-02 -5.95819652e-01
5.21976769e-01 3.78928125e-01 -1.37029916e-01 -1.17302081e-02
2.12729484e-01 1.01294078e-01 5.94360530e-01 1.93365365e-01
1.16135681e+00 -1.61815807e-02 1.59762248e-01 1.38780892e-01
4.68630880e-01 -7.09585547e-02 4.65254754e-01 7.57927001e-01
-5.65114021e-01 9.50520217e-01 5.63857019e-01 -6.00850999e-01
-1.09995174e+00 -1.16494751e+00 -8.40953458e-03 7.66413748e-01
2.93957859e-01 -3.87948900e-01 -6.09671056e-01 -9.14024293e-01
-1.39045209e-01 7.14274645e-01 -6.37468219e-01 -5.67571461e-01
-1.95936725e-01 -4.89295244e-01 1.77437082e-01 5.26600480e-01
6.11301601e-01 -1.06880856e+00 -5.86247444e-01 2.13235486e-02
-1.08362094e-01 -9.49826062e-01 -3.85472029e-01 1.78779885e-01
-4.67593104e-01 -1.10842204e+00 -8.62112343e-01 -8.61428738e-01
5.65157056e-01 1.57738745e-01 8.03030193e-01 -1.73024923e-01
-5.85470259e-01 1.65379465e-01 -4.48499084e-01 -5.48109531e-01
-1.95567921e-01 -6.96886107e-02 3.44468071e-03 2.77627319e-01
4.17083412e-01 -1.42278537e-01 -5.26434362e-01 4.49867636e-01
-8.31321001e-01 -2.47638434e-01 6.65327191e-01 7.94479251e-01
5.84226072e-01 -8.48474819e-03 3.55927825e-01 -5.34854949e-01
4.47704017e-01 -4.69679326e-01 -5.62042117e-01 1.92664519e-01
-7.52480328e-01 -4.53212053e-01 5.81304967e-01 -5.39036572e-01
-7.05269754e-01 2.10752785e-01 -1.41120821e-01 -6.35355353e-01
-1.16187721e-01 1.82630390e-01 -1.87116206e-01 6.08810298e-02
7.56901801e-01 -1.38486875e-02 2.88562588e-02 -4.66556191e-01
2.97010541e-01 6.87350094e-01 7.88071454e-01 -3.51975635e-02
1.01589012e+00 2.29263622e-02 -3.60583276e-01 -6.89079106e-01
-9.82787073e-01 -6.03246331e-01 -6.32343471e-01 -3.37287128e-01
9.44088936e-01 -8.71469378e-01 -2.38464132e-01 7.40578413e-01
-1.12815499e+00 1.13643467e-01 -6.48625374e-01 5.32709897e-01
-3.10980022e-01 4.07975852e-01 -3.31762820e-01 -6.72517657e-01
-9.90580246e-02 -1.19790232e+00 8.84099841e-01 2.93287814e-01
-2.88164437e-01 -4.49414551e-01 -2.83284992e-01 1.88961923e-01
5.50749362e-01 3.42973143e-01 7.95825720e-01 -9.31702793e-01
-4.82192278e-01 -6.89960420e-01 -1.79635406e-01 7.33607054e-01
-4.64074016e-02 7.92313018e-04 -1.04910636e+00 -5.17311454e-01
2.88816512e-01 -2.08582297e-01 8.79452109e-01 2.74634480e-01
1.35456383e+00 5.05077131e-02 -2.71534383e-01 4.15316015e-01
1.03494620e+00 2.60280281e-01 6.96149349e-01 5.74709773e-01
7.99937010e-01 4.44272518e-01 1.08419931e+00 2.77177572e-01
6.57784417e-02 8.63940239e-01 6.83333158e-01 -2.18092471e-01
-1.08483724e-01 -8.81666169e-02 6.33876204e-01 8.43903124e-01
2.28578001e-01 -8.89227837e-02 -7.50871420e-01 7.23990321e-01
-1.77352798e+00 -8.82626593e-01 -2.33349204e-01 2.31649137e+00
2.52110362e-01 5.89284539e-01 1.83566585e-01 2.65778720e-01
8.26792359e-01 8.72043818e-02 -5.98305821e-01 -4.22005355e-01
-4.17643320e-03 -4.36905846e-02 1.18015945e-01 -4.93472926e-02
-1.32666349e+00 5.47251701e-01 5.99291992e+00 8.79740179e-01
-9.22170818e-01 1.71369221e-02 6.15350544e-01 -1.39703169e-01
2.76152343e-01 -3.58461261e-01 -6.87061131e-01 7.01642573e-01
9.37618554e-01 -2.64798641e-01 2.62238807e-03 1.10367179e+00
-2.16031149e-01 1.30218521e-01 -1.11454654e+00 1.08107352e+00
3.63334477e-01 -9.69294786e-01 8.64097327e-02 1.64696854e-02
7.99215853e-01 -1.11259207e-01 4.91826028e-01 5.50866008e-01
-1.71957135e-01 -9.48503673e-01 7.69804060e-01 5.30175865e-01
6.35065556e-01 -6.23839676e-01 1.13080680e+00 1.62977874e-01
-1.08200657e+00 -3.82479876e-01 -4.53074694e-01 2.66810000e-01
1.38737273e-03 6.66471362e-01 -6.41751647e-01 8.05144250e-01
1.05647719e+00 6.62791252e-01 -9.50101793e-01 1.50908959e+00
-4.41398770e-02 3.98402035e-01 -1.52480632e-01 2.97173023e-01
1.42854393e-01 1.75023958e-01 6.25537694e-01 9.68435168e-01
7.25951493e-01 -6.28462851e-01 3.42653126e-01 6.82579815e-01
2.32586786e-02 1.60371155e-01 -6.95075274e-01 6.84657767e-02
1.37930557e-01 1.46453059e+00 -6.07683241e-01 -2.08110332e-01
-4.80332464e-01 1.15054679e+00 -1.91239432e-01 1.71725035e-01
-9.85039473e-01 -9.35578704e-01 4.81241167e-01 1.40993387e-01
5.95618546e-01 7.91747570e-02 -1.09455734e-01 -8.84455860e-01
4.70696747e-01 -1.01033306e+00 3.56002420e-01 -6.44378901e-01
-1.57901621e+00 7.90397346e-01 -1.67911708e-01 -1.67694747e+00
-2.34387249e-01 -9.12384033e-01 -7.00062156e-01 7.43029952e-01
-1.10599339e+00 -9.52512860e-01 -5.97635329e-01 4.08098668e-01
6.69160068e-01 -5.65096915e-01 6.40451491e-01 5.33577621e-01
-6.51023328e-01 9.11822140e-01 1.65684670e-02 1.40133962e-01
7.49718785e-01 -1.31005394e+00 6.95847213e-01 1.10731363e+00
3.86255920e-01 3.34087670e-01 6.91908121e-01 -4.42119002e-01
-9.13730145e-01 -1.13500106e+00 5.78695834e-01 -5.36259174e-01
6.41249180e-01 -2.34239191e-01 -1.25817847e+00 4.03479129e-01
1.66009385e-02 2.92117149e-01 4.17010844e-01 -9.97864082e-02
-3.30376327e-01 -2.71826327e-01 -1.09852636e+00 4.11768705e-01
7.98004150e-01 -3.89679074e-01 -4.84358460e-01 1.62802503e-01
7.48754919e-01 -4.64370221e-01 -7.25331426e-01 3.74051750e-01
3.86396348e-01 -9.43652987e-01 7.16845036e-01 -4.18221980e-01
5.14375627e-01 -6.30238771e-01 -1.90805450e-01 -1.18255126e+00
-4.00183439e-01 -2.70835161e-01 -2.47685567e-01 1.39462304e+00
5.65284789e-01 -3.02964926e-01 7.68800735e-01 2.29550809e-01
-2.57523924e-01 -7.56009758e-01 -8.51657569e-01 -1.03685188e+00
-3.76339257e-01 -4.83690172e-01 4.24489200e-01 6.80777431e-01
-2.41217792e-01 1.41431808e-01 -3.76167566e-01 1.37614921e-01
5.33867717e-01 -1.05973989e-01 7.41949439e-01 -1.30682576e+00
-3.10321897e-01 -2.60967672e-01 -9.27359939e-01 -7.09529281e-01
-4.04483914e-01 -8.96068692e-01 2.09726945e-01 -1.33243465e+00
2.62586504e-01 1.43630251e-01 -9.06810105e-01 3.75093490e-01
-3.07038039e-01 4.98834163e-01 3.71464074e-01 6.57048449e-02
-8.70252609e-01 3.35874170e-01 9.84421611e-01 -5.11920691e-01
-1.31079629e-01 2.04892516e-01 -7.77632058e-01 6.92933261e-01
7.43730664e-01 -3.17638993e-01 -2.80401140e-01 -2.06127465e-01
-7.26241106e-03 -6.10777199e-01 5.24359584e-01 -1.48724496e+00
2.01457754e-01 1.78209752e-01 6.65445209e-01 -7.89308190e-01
2.88304508e-01 -7.39215732e-01 -4.78278436e-02 4.22411323e-01
-1.67024821e-01 2.97403276e-01 2.48571664e-01 6.12824559e-01
-3.58136773e-01 -4.26136464e-01 6.32622778e-01 9.24167633e-02
-1.09741545e+00 2.49428615e-01 -1.08710080e-01 -6.16254583e-02
1.30752254e+00 -4.51783031e-01 -3.19534123e-01 -1.37941107e-01
-8.33864748e-01 3.72509241e-01 3.77587348e-01 8.32162499e-01
7.36910641e-01 -1.38108993e+00 -6.24291420e-01 4.53110576e-01
5.82530677e-01 -1.47204861e-01 5.10515392e-01 8.57396245e-01
-2.35921174e-01 -1.65505130e-02 -5.64146936e-01 -9.18062210e-01
-1.04768586e+00 6.48540795e-01 1.58620760e-01 -3.72213423e-01
-7.43417799e-01 8.52430582e-01 1.31972730e-01 -1.70600504e-01
4.28575993e-01 -7.68320337e-02 -3.31225604e-01 1.49396017e-01
8.91931415e-01 4.57274288e-01 3.83832812e-01 -3.91058207e-01
-3.86322439e-01 5.32749295e-01 -4.55515534e-01 9.74645466e-02
1.12439847e+00 1.16726458e-01 2.49625221e-01 6.99411750e-01
1.12949431e+00 -1.07133307e-01 -1.39691961e+00 -4.24026549e-02
2.80299395e-01 -4.83241320e-01 -7.02019706e-02 -7.94056833e-01
-1.15739024e+00 9.45342898e-01 1.19664979e+00 3.97248328e-01
9.31847394e-01 -1.24092571e-01 6.78715348e-01 3.02395731e-01
2.15633377e-01 -1.18341923e+00 4.98766899e-01 3.49852383e-01
8.63254845e-01 -1.36235917e+00 -1.82785451e-01 -2.90051252e-01
-9.29132760e-01 9.09977734e-01 9.69275832e-01 -3.03412527e-02
3.75415385e-01 -4.19443212e-02 1.24024160e-01 -3.54257643e-01
-6.29505813e-01 2.50810335e-05 6.18440390e-01 8.74218106e-01
1.44304395e-01 -2.19740734e-01 3.70491110e-03 6.86685503e-01
-6.73185661e-02 -2.09218055e-01 3.83336008e-01 5.59594989e-01
-5.45157313e-01 -8.90730262e-01 -1.40361115e-01 6.22727215e-01
-2.94745237e-01 2.72750914e-01 -2.55753249e-01 7.43024528e-01
3.20197493e-01 9.21678722e-01 1.37111261e-01 -7.88844049e-01
9.30631816e-01 2.79384702e-01 9.13884714e-02 -5.86594939e-01
-5.00354230e-01 -3.31683427e-01 -3.49314898e-01 -9.54087973e-01
2.59818882e-01 -6.35872483e-01 -7.62539566e-01 7.67656043e-02
-2.72937506e-01 3.68217267e-02 7.86332846e-01 6.28941655e-01
3.22678030e-01 8.93432498e-01 7.92287886e-01 -8.54679346e-01
-8.44830871e-01 -9.22460377e-01 -6.89599931e-01 6.22342646e-01
2.53957868e-01 -7.33536422e-01 -4.81515199e-01 1.24660861e-02]
|
[7.684839725494385, 1.9455512762069702]
|
cc306e11-e9dc-4163-8d56-80bd1ed889f8
|
combo-a-complete-benchmark-for-open-kg
|
2302.03905
| null |
https://arxiv.org/abs/2302.03905v1
|
https://arxiv.org/pdf/2302.03905v1.pdf
|
COMBO: A Complete Benchmark for Open KG Canonicalization
|
Open knowledge graph (KG) consists of (subject, relation, object) triples extracted from millions of raw text. The subject and object noun phrases and the relation in open KG have severe redundancy and ambiguity and need to be canonicalized. Existing datasets for open KG canonicalization only provide gold entity-level canonicalization for noun phrases. In this paper, we present COMBO, a Complete Benchmark for Open KG canonicalization. Compared with existing datasets, we additionally provide gold canonicalization for relation phrases, gold ontology-level canonicalization for noun phrases, as well as source sentences from which triples are extracted. We also propose metrics for evaluating each type of canonicalization. On the COMBO dataset, we empirically compare previously proposed canonicalization methods as well as a few simple baseline methods based on pretrained language models. We find that properly encoding the phrases in a triple using pretrained language models results in better relation canonicalization and ontology-level canonicalization of the noun phrase. We release our dataset, baselines, and evaluation scripts at https://github.com/jeffchy/COMBO/tree/main.
|
['Kewei Tu', 'Pengjun Xie', 'Yuting Zheng', 'Weiqi Wu', 'Yong Jiang', 'Chengyue Jiang']
|
2023-02-08
| null | null | null | null |
['open-knowledge-graph-canonicalization']
|
['knowledge-base']
|
[-3.41479331e-01 4.98648047e-01 -5.24128020e-01 -3.90473247e-01
-8.93830419e-01 -8.90093625e-01 5.25321186e-01 5.64654827e-01
-2.78082758e-01 8.37448359e-01 6.53361678e-01 -2.73776025e-01
-1.42803162e-01 -1.07084537e+00 -9.83074129e-01 -1.38428450e-01
-1.64119586e-01 8.52913916e-01 8.19206089e-02 -2.99676806e-01
-3.89624238e-01 -2.79549241e-01 -1.15387690e+00 4.02103156e-01
1.07671320e+00 8.80466521e-01 -2.83598125e-01 1.71785921e-01
-2.94348389e-01 2.31704071e-01 -1.27850562e-01 -1.05757391e+00
4.20958638e-01 1.09156445e-01 -1.23107541e+00 -4.30819154e-01
7.65097439e-01 2.25723833e-01 -4.46230978e-01 1.04969335e+00
2.90332615e-01 -1.65918604e-01 5.95177948e-01 -1.36374760e+00
-1.06043935e+00 1.51229167e+00 -2.04275534e-01 -1.50452867e-01
8.02013040e-01 -9.83541906e-02 1.98526347e+00 -1.10393286e+00
1.21314383e+00 1.27485394e+00 6.88847184e-01 3.24768573e-01
-1.36350095e+00 -6.00115836e-01 -1.04352213e-01 2.41558999e-01
-1.76039886e+00 -2.24570587e-01 1.57032683e-01 -1.74924687e-01
1.33247375e+00 1.98474780e-01 7.05987751e-01 9.81701612e-01
1.79787248e-01 6.43143654e-01 7.32874572e-01 -4.60943609e-01
-1.63113371e-01 -1.53514758e-01 7.84323990e-01 7.40089953e-01
9.92222965e-01 -2.89541990e-01 -7.17579424e-01 -3.46848667e-01
-3.56065040e-03 -5.13619125e-01 -3.55390906e-01 -4.99535263e-01
-1.36576843e+00 4.14591730e-01 6.05477154e-01 -1.08913481e-01
-2.72272497e-01 3.27724457e-01 4.49209720e-01 2.36388966e-01
2.44586959e-01 5.56809604e-01 -9.15981472e-01 -4.32632072e-03
-2.06156239e-01 5.33409476e-01 1.35317457e+00 1.69705117e+00
1.16634476e+00 -6.19335115e-01 -2.50961930e-01 8.50247025e-01
3.51171672e-01 5.09991884e-01 1.02589987e-01 -7.80800223e-01
8.23449373e-01 9.74455059e-01 -2.05198646e-01 -7.30135858e-01
-3.92929703e-01 -1.95637003e-01 -3.10876608e-01 -9.99437153e-01
1.28330946e-01 1.66944936e-02 -9.98245656e-01 1.73181653e+00
4.51982319e-01 -3.36955413e-02 5.61079741e-01 4.42309797e-01
1.55602908e+00 4.36469644e-01 3.20462346e-01 -9.67095792e-03
1.81561518e+00 -8.25118542e-01 -8.30229521e-01 -2.44563043e-01
1.07836246e+00 -6.55699730e-01 1.19179451e+00 -6.66407794e-02
-6.05455637e-01 9.91601720e-02 -1.07548606e+00 -7.24018872e-01
-9.04263973e-01 -2.89896190e-01 8.33007157e-01 5.72066844e-01
-7.91711986e-01 4.72957253e-01 -7.31528282e-01 -6.85742438e-01
1.67570814e-01 3.22922677e-01 -7.94362247e-01 -1.93507656e-01
-1.65610504e+00 9.02694583e-01 1.17205966e+00 -3.24115545e-01
-3.01501721e-01 -1.14308453e+00 -1.22258437e+00 5.18956780e-02
7.24737227e-01 -1.20254230e+00 1.12341118e+00 1.43041775e-01
-7.00879097e-01 9.75808501e-01 -2.50165135e-01 -5.12219787e-01
-1.51932716e-01 -4.95316356e-01 -6.50440335e-01 -1.34727582e-01
6.43086910e-01 7.48380482e-01 1.21552266e-01 -1.31172049e+00
-6.47433758e-01 -3.51197571e-01 4.45756108e-01 1.93087205e-01
-3.70400012e-01 -1.24021791e-01 -9.23400223e-01 -5.13153553e-01
5.09113252e-01 -1.00114238e+00 5.63529842e-02 -6.91978991e-01
-1.04191184e+00 -7.16407657e-01 4.70132470e-01 -6.14664257e-01
1.68820059e+00 -1.87517858e+00 1.14464395e-01 4.20415372e-01
5.58219314e-01 -3.67860973e-01 5.32856723e-03 6.70620859e-01
-2.06859857e-01 6.80823088e-01 -8.66855383e-02 -1.27724186e-01
4.83282983e-01 6.50709093e-01 -3.46215755e-01 -3.47846821e-02
9.52492096e-03 1.34310377e+00 -1.01240969e+00 -6.98249102e-01
-4.89336729e-01 1.01022288e-01 -6.87043428e-01 -2.55971432e-01
-6.43974304e-01 -2.27811277e-01 -2.69651175e-01 1.05854595e+00
5.83398104e-01 -3.79930109e-01 7.90764809e-01 -8.12959373e-01
4.80496317e-01 8.18676710e-01 -1.07506990e+00 1.55003285e+00
-2.85457224e-01 7.96440020e-02 -5.32907903e-01 -1.95722282e-01
5.83916485e-01 3.10658216e-01 4.98481005e-01 -2.95844227e-01
-4.29125652e-02 5.40419042e-01 -1.68241754e-01 -3.60637695e-01
1.03062356e+00 1.14796236e-01 -6.22821331e-01 2.84752965e-01
5.65818667e-01 -2.00885952e-01 9.28667307e-01 8.27496648e-01
1.24554527e+00 1.07610583e-01 6.29106283e-01 -3.13008428e-01
2.15127245e-01 1.85610250e-01 8.68228436e-01 4.12582219e-01
3.57193738e-01 3.31724137e-01 7.85135686e-01 -3.71028185e-01
-9.34241593e-01 -1.34301674e+00 -3.42598498e-01 8.67537439e-01
2.36628711e-01 -1.61250067e+00 -4.09748942e-01 -8.29327762e-01
4.46171671e-01 7.80744612e-01 -3.92134309e-01 -1.71478942e-01
-5.55415928e-01 -9.08682585e-01 9.51233685e-01 5.79508960e-01
1.36440292e-01 -4.71238494e-01 3.12986195e-01 4.46889624e-02
-6.54088736e-01 -1.72856975e+00 -5.69454134e-01 3.46339583e-01
-6.12663925e-01 -1.40026534e+00 3.34375203e-01 -6.55323744e-01
6.20107710e-01 -8.13279152e-02 1.71434987e+00 -6.87983334e-02
1.20574631e-01 3.53446692e-01 -6.01874650e-01 -3.00840557e-01
-2.37225115e-01 3.01875830e-01 3.59479725e-01 -4.18272048e-01
8.43903303e-01 -5.79491138e-01 -7.64428601e-02 5.64155206e-02
-8.83691907e-01 -5.46627454e-02 3.09187531e-01 6.82923555e-01
7.12695479e-01 -2.23424092e-01 1.40068308e-01 -1.34410238e+00
7.32349396e-01 -6.64781272e-01 -4.37160611e-01 6.02147162e-01
-8.29860568e-01 4.20609325e-01 1.90860420e-01 1.71926126e-01
-6.17545426e-01 -1.14846252e-01 2.57074349e-02 -5.01608811e-02
3.16033155e-01 1.18247795e+00 -4.68526453e-01 1.21798590e-01
7.01394320e-01 -2.48110592e-01 -6.04705930e-01 -5.23337245e-01
8.80547762e-01 4.03985977e-01 6.31902337e-01 -1.23552954e+00
1.06046975e+00 2.81432271e-01 -1.05929010e-01 -4.82577175e-01
-1.07632267e+00 -4.73646998e-01 -8.37830782e-01 3.27525288e-01
7.39819765e-01 -1.20765841e+00 -4.49633688e-01 -1.07956499e-01
-1.27601016e+00 1.79343894e-01 -4.34524029e-01 2.47040391e-01
-2.06678793e-01 5.73730290e-01 -5.57800651e-01 3.54911052e-02
-4.34030503e-01 -7.99614608e-01 1.08507717e+00 -7.30701238e-02
-4.01336223e-01 -9.14617479e-01 1.63881555e-01 6.57198966e-01
-3.07486206e-01 3.80827822e-02 1.24615073e+00 -8.14717114e-01
-7.22112298e-01 -1.94040403e-01 -2.97705144e-01 1.38492405e-01
9.94603336e-02 5.42819947e-02 -4.54378486e-01 -3.76484022e-02
-7.96558022e-01 -4.86385703e-01 9.19613421e-01 -1.96554601e-01
5.15424490e-01 -5.22814929e-01 -6.40287697e-01 8.30745339e-01
1.44349778e+00 -3.45699638e-01 5.02913237e-01 2.93286264e-01
1.01700616e+00 1.78775385e-01 5.36962271e-01 1.03594191e-01
1.21477258e+00 4.86657798e-01 1.17434382e-01 4.45971161e-01
-1.48549631e-01 -7.44365752e-01 4.67939198e-01 1.10623682e+00
-8.44284967e-02 -3.15422624e-01 -1.16490579e+00 7.55389035e-01
-1.74573863e+00 -5.58711946e-01 -4.32157010e-01 1.85602486e+00
1.32530010e+00 -3.67040164e-03 -2.63710797e-01 -2.64821023e-01
4.63727862e-01 2.85647083e-02 -1.62405595e-02 -9.02197435e-02
-6.03852928e-01 4.17108268e-01 7.94390738e-01 6.71814382e-01
-1.20013297e+00 1.33296096e+00 5.90089560e+00 8.02590549e-01
-5.37875950e-01 2.93005228e-01 -1.63835019e-01 1.85692981e-02
-8.55251193e-01 6.75641179e-01 -1.42380989e+00 1.45133704e-01
8.84885490e-01 -6.21636927e-01 2.21936673e-01 6.45350814e-01
-3.73957425e-01 4.80973301e-03 -1.54332078e+00 7.80589581e-01
-5.42144477e-02 -1.45138335e+00 4.39080626e-01 4.63259444e-02
8.72587383e-01 2.62820125e-01 -4.36898947e-01 6.31838977e-01
7.12687731e-01 -7.91600466e-01 7.09283352e-01 1.28641695e-01
7.23413885e-01 -2.66685426e-01 6.79537475e-01 -2.91612655e-01
-1.56137133e+00 1.49847031e-01 -4.23635393e-01 1.68920979e-01
2.20061824e-01 7.65588462e-01 -7.99268126e-01 1.31484318e+00
6.99757159e-01 9.03035402e-01 -8.62230778e-01 5.54604173e-01
-7.34369695e-01 4.57743675e-01 -7.46198535e-01 1.44134328e-01
1.54495507e-01 -1.81577012e-01 6.08648717e-01 1.07092512e+00
6.53165355e-02 1.08253293e-01 4.27792966e-01 7.81068504e-01
-6.42403066e-01 4.45700407e-01 -7.68971443e-01 -4.01964366e-01
8.42240036e-01 1.15330327e+00 -3.76154780e-01 -4.49795932e-01
-6.92927897e-01 4.48075861e-01 6.43133521e-01 4.25047159e-01
-6.74225509e-01 -4.21112388e-01 9.03274715e-01 2.97078677e-02
2.30181277e-01 -3.00002605e-01 -2.89389789e-01 -1.73366940e+00
4.24220920e-01 -6.95849478e-01 1.01546621e+00 -5.59145093e-01
-1.37937140e+00 3.61784548e-01 4.18230623e-01 -8.17086339e-01
-6.09957613e-02 -5.18892050e-01 -1.37251601e-01 5.82075715e-01
-1.19390893e+00 -1.46531570e+00 -8.19618478e-02 3.12869966e-01
-3.62153612e-02 2.76620891e-02 9.77673471e-01 6.20872080e-01
-5.84981024e-01 7.45078623e-01 -1.34642050e-01 4.52178776e-01
7.80803561e-01 -1.42035151e+00 6.97832048e-01 8.32519948e-01
3.91775936e-01 1.28143299e+00 7.59270489e-01 -1.25942945e+00
-1.69152713e+00 -1.18695724e+00 1.53684890e+00 -8.85918200e-01
1.11215186e+00 -5.38083673e-01 -7.30382204e-01 1.50274611e+00
2.97407329e-01 2.09962875e-01 7.55718112e-01 9.09478724e-01
-8.38000238e-01 -1.61816150e-01 -6.79824948e-01 6.85332477e-01
1.44043326e+00 -5.82248688e-01 -9.26719129e-01 5.95869780e-01
1.28766406e+00 -6.12834334e-01 -1.46453083e+00 6.48832083e-01
5.08300781e-01 -3.56357604e-01 9.16681230e-01 -9.64665353e-01
4.02227372e-01 -4.21268374e-01 -4.34581429e-01 -1.23971045e+00
-1.44559488e-01 -4.60493386e-01 -4.62217093e-01 1.29304588e+00
1.15207779e+00 -7.02513456e-01 6.81221664e-01 7.05852926e-01
-2.90212393e-01 -8.42418253e-01 -9.16858912e-01 -1.07365882e+00
9.07335430e-02 -6.30431175e-01 8.05027962e-01 1.05774534e+00
5.36117613e-01 7.79911220e-01 -2.76710708e-02 4.01132971e-01
7.28588283e-01 2.29341388e-01 8.37155640e-01 -1.06437349e+00
1.81815520e-01 1.02183543e-01 -5.03713429e-01 -8.09518099e-01
4.27483529e-01 -1.58153415e+00 -3.80788743e-01 -1.74694788e+00
4.16134536e-01 -5.05565226e-01 9.12017152e-02 1.00159574e+00
-2.87772864e-01 2.08501682e-01 1.53027251e-01 9.30855572e-02
-6.78708792e-01 6.95706427e-01 8.98194015e-01 -3.40420783e-01
5.87108210e-02 -7.13364005e-01 -1.01206875e+00 4.63771075e-01
4.94470000e-01 -5.68249166e-01 -2.99435407e-01 -5.80959439e-01
8.91108036e-01 -3.52819532e-01 3.06334525e-01 -7.28421152e-01
3.46299648e-01 -1.02992752e-03 -2.34572157e-01 -5.08308709e-01
2.45875418e-01 -5.69102108e-01 3.91623884e-01 2.03477472e-01
-1.49139222e-02 5.82975000e-02 5.44639900e-02 5.09521842e-01
-3.11120272e-01 -1.36146992e-01 2.29980707e-01 -2.05088124e-01
-9.77623403e-01 4.06705528e-01 3.22698891e-01 6.25059187e-01
8.28056991e-01 7.04504848e-02 -6.98491335e-01 -8.91677961e-02
-9.35747802e-01 7.78454185e-01 4.82688010e-01 6.93783700e-01
4.75991905e-01 -1.51992333e+00 -6.86415374e-01 -1.66469574e-01
6.31421983e-01 7.81086087e-02 -2.25094751e-01 8.00155759e-01
-5.97994149e-01 7.08324194e-01 1.74056441e-01 -1.38366893e-01
-1.12312245e+00 4.56724107e-01 5.14124446e-02 -4.47554260e-01
-5.74233651e-01 7.37153411e-01 -9.00537893e-02 -1.00856674e+00
-2.41555134e-03 -8.64345431e-01 5.68618923e-02 1.71185374e-01
-5.25149852e-02 8.66444409e-02 2.24046245e-01 -6.69409096e-01
-4.94774610e-01 3.79077911e-01 -1.35324240e-01 1.98622987e-01
1.18921816e+00 -9.52578187e-02 -8.79424393e-01 2.50565827e-01
1.14350259e+00 1.90858647e-01 3.87592092e-02 -6.01381958e-01
5.71168303e-01 -3.01307291e-01 -3.90643030e-01 -6.06474817e-01
-7.41961241e-01 2.58331411e-02 -2.28720486e-01 5.61075695e-02
5.71443200e-01 5.39491117e-01 9.25946474e-01 8.40504348e-01
6.83344126e-01 -8.93861473e-01 -4.40687150e-01 1.06189954e+00
8.08891356e-01 -1.00089228e+00 3.68562102e-01 -1.30978191e+00
-4.62663203e-01 6.93631530e-01 7.88704634e-01 1.96499676e-01
9.25048411e-01 2.06264392e-01 -1.57649182e-02 -5.89708030e-01
-1.01907492e+00 -4.44901884e-01 4.18448895e-01 4.16366965e-01
5.56646824e-01 3.18171024e-01 -7.37998843e-01 1.04496181e+00
-8.01277399e-01 -2.90772140e-01 6.25133276e-01 8.61768782e-01
-9.14231017e-02 -1.49882913e+00 -1.58758610e-01 7.61726141e-01
-3.23683083e-01 -6.54755950e-01 -5.07194877e-01 7.03309953e-01
1.80417433e-01 5.58452070e-01 -2.20387444e-01 -5.44555068e-01
3.92892331e-01 3.55642259e-01 4.90222067e-01 -1.05388451e+00
-3.47128481e-01 -3.92704457e-01 8.92958581e-01 -6.62991822e-01
-1.79338738e-01 -6.49153888e-01 -1.58742833e+00 -4.99421269e-01
-3.66025150e-01 4.68519807e-01 1.61223695e-01 8.65662158e-01
5.02434254e-01 2.44648680e-01 -1.37209430e-01 -1.03072569e-01
-3.61573547e-01 -1.02070570e+00 -6.17836237e-01 5.04735291e-01
-3.27511370e-01 -8.35636258e-01 -5.77858500e-02 1.72365859e-01]
|
[9.153640747070312, 8.287605285644531]
|
e84dfb0c-999b-4235-98de-601e58e48c1c
|
prepositional-phrase-attachment-problem
| null | null |
https://aclanthology.org/W15-0102
|
https://aclanthology.org/W15-0102.pdf
|
Prepositional Phrase Attachment Problem Revisited: how Verbnet can Help
| null |
['Yuliya Lierler', 'Daniel Bailey', 'Benjamin Susman']
|
2015-04-01
| null | null | null |
ws-2015-4
|
['prepositional-phrase-attachment']
|
['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]
|
[-7.403817176818848, 3.673025369644165]
|
b4442f0c-e7d0-44ad-a208-f0126b380db3
|
point-cloud-registration-for-lidar-and
|
2302.07184
| null |
https://arxiv.org/abs/2302.07184v1
|
https://arxiv.org/pdf/2302.07184v1.pdf
|
Point Cloud Registration for LiDAR and Photogrammetric Data: a Critical Synthesis and Performance Analysis on Classic and Deep Learning Algorithms
|
Recent advances in computer vision and deep learning have shown promising performance in estimating rigid/similarity transformation between unregistered point clouds of complex objects and scenes. However, their performances are mostly evaluated using a limited number of datasets from a single sensor (e.g. Kinect or RealSense cameras), lacking a comprehensive overview of their applicability in photogrammetric 3D mapping scenarios. In this work, we provide a comprehensive review of the state-of-the-art (SOTA) point cloud registration methods, where we analyze and evaluate these methods using a diverse set of point cloud data from indoor to satellite sources. The quantitative analysis allows for exploring the strengths, applicability, challenges, and future trends of these methods. In contrast to existing analysis works that introduce point cloud registration as a holistic process, our experimental analysis is based on its inherent two-step process to better comprehend these approaches including feature/keypoint-based initial coarse registration and dense fine registration through cloud-to-cloud (C2C) optimization. More than ten methods, including classic hand-crafted, deep-learning-based feature correspondence, and robust C2C methods were tested. We observed that the success rate of most of the algorithms are fewer than 40% over the datasets we tested and there are still are large margin of improvement upon existing algorithms concerning 3D sparse corresopondence search, and the ability to register point clouds with complex geometry and occlusions. With the evaluated statistics on three datasets, we conclude the best-performing methods for each step and provide our recommendations, and outlook future efforts.
|
['Shuang Song', 'Rongjun Qin', 'Ningli Xu']
|
2023-02-14
| null | null | null | null |
['point-cloud-registration']
|
['computer-vision']
|
[-1.48712136e-02 -4.57579851e-01 5.69383539e-02 -2.61364788e-01
-9.06868100e-01 -5.85031033e-01 9.65133429e-01 1.38661772e-01
-4.73924965e-01 2.94014245e-01 -2.30876908e-01 -1.06474645e-01
-3.85896146e-01 -7.61510551e-01 -7.38663733e-01 -6.79326117e-01
-1.75319642e-01 1.01106703e+00 1.00651711e-01 -4.73890573e-01
4.06404436e-01 1.35286677e+00 -1.85992777e+00 -2.75496244e-01
6.77961409e-01 1.08425438e+00 1.96358994e-01 2.16633067e-01
-1.65018871e-01 -2.01205611e-01 -2.24758893e-01 -1.57187402e-01
6.97606325e-01 3.78652662e-01 -6.01439953e-01 -5.51849529e-02
8.41369927e-01 -1.35711446e-01 -9.03118327e-02 8.05729985e-01
6.61391795e-01 1.00762628e-01 5.79024196e-01 -1.15611184e+00
-4.55588341e-01 -2.04674050e-01 -5.61301529e-01 -7.29655772e-02
5.86857617e-01 8.60151276e-02 6.83936238e-01 -1.44130874e+00
4.32402432e-01 9.71247077e-01 1.13715029e+00 8.01814348e-03
-9.92518842e-01 -6.87884927e-01 -1.27387673e-01 9.01386067e-02
-1.79763544e+00 -3.44456345e-01 8.17926943e-01 -7.32640564e-01
1.29276907e+00 3.58613133e-01 8.65415752e-01 5.20385444e-01
-4.91718948e-02 -1.55920964e-02 1.01274514e+00 -5.10788500e-01
1.55871406e-01 -2.28012741e-01 -1.20513171e-01 2.92156696e-01
3.05149198e-01 5.07229149e-01 -4.27086234e-01 -3.47077072e-01
9.95722651e-01 1.82116061e-01 -1.90902457e-01 -7.85952449e-01
-1.44297898e+00 7.45969296e-01 7.17930019e-01 3.57354403e-01
-4.63284761e-01 -3.58027183e-02 6.20571189e-02 2.39799857e-01
5.79993188e-01 2.49513224e-01 -5.10129094e-01 1.86607167e-02
-1.14273798e+00 4.12192196e-01 6.61708772e-01 1.21423268e+00
1.20711076e+00 4.97524142e-02 4.49559182e-01 5.19423068e-01
5.07289886e-01 1.02015579e+00 1.42267391e-01 -7.93121457e-01
5.33941984e-01 5.26322186e-01 1.96760491e-01 -1.28767443e+00
-5.81222117e-01 -2.48676479e-01 -8.59501123e-01 4.74089265e-01
1.46842487e-02 3.09941202e-01 -8.22156787e-01 7.95651555e-01
4.21975136e-01 3.22642237e-01 -1.44835308e-01 1.00231278e+00
1.02581704e+00 2.69551843e-01 -3.22172672e-01 7.04374611e-02
1.03636181e+00 -4.42863166e-01 -2.73145646e-01 -1.58421129e-01
3.40246230e-01 -1.14496183e+00 8.56056809e-01 -3.11675556e-02
-9.00362253e-01 -6.06105387e-01 -1.04566360e+00 -5.21159880e-02
-3.91478539e-01 4.35848348e-02 7.34487653e-01 4.69777822e-01
-1.21791518e+00 7.81404674e-01 -1.10663021e+00 -7.88859844e-01
4.17749941e-01 7.52060711e-01 -6.02069557e-01 -2.62779910e-02
-6.60313070e-01 1.12415576e+00 1.23216905e-01 3.03512871e-01
-4.64055777e-01 -8.24083447e-01 -8.26282978e-01 -2.45297983e-01
-8.01928341e-02 -8.23969245e-01 8.06404352e-01 -4.91613299e-01
-1.47027719e+00 1.24754620e+00 -2.36753434e-01 -2.11044550e-01
4.62700188e-01 -2.88007468e-01 -1.46746993e-01 -1.23361543e-01
9.07294676e-02 5.19985974e-01 4.47082341e-01 -1.55322194e+00
-3.97420168e-01 -5.92896581e-01 -2.28064895e-01 3.04785579e-01
2.16243893e-01 2.07948923e-01 -4.76017803e-01 -2.98665345e-01
8.92903984e-01 -1.18683112e+00 -3.16482961e-01 2.37670869e-01
4.93644476e-02 -8.88964685e-04 7.56103158e-01 -1.88748851e-01
4.50967133e-01 -1.91949952e+00 -5.40720597e-02 4.19548154e-01
5.37408050e-04 2.14314163e-01 -3.74024883e-02 6.16328359e-01
-2.11560890e-01 2.16898490e-02 -3.63381535e-01 -6.91520870e-01
-1.04536287e-01 2.37155125e-01 -3.24824482e-01 1.06548524e+00
-2.93744355e-02 8.86550307e-01 -6.54534698e-01 -2.92177200e-01
8.92921865e-01 7.43284166e-01 -1.63811669e-01 1.50521351e-02
2.41103321e-01 6.52573586e-01 -3.38240951e-01 1.08521128e+00
1.13841546e+00 -5.42356595e-02 -4.02837783e-01 -4.54568952e-01
-4.61269498e-01 1.37675524e-01 -1.39966202e+00 1.81096113e+00
-4.92623627e-01 5.69969654e-01 3.45752984e-02 -8.05530012e-01
1.27438343e+00 1.99279383e-01 9.18327987e-01 -5.17585278e-01
-6.48271339e-03 5.13386905e-01 -3.96531343e-01 -2.13828504e-01
5.61883628e-01 -1.63676143e-01 2.13972151e-01 5.70786372e-02
-5.85785951e-04 -7.54032671e-01 -4.54869896e-01 -3.97789836e-01
6.43873394e-01 3.22827637e-01 4.18309301e-01 -2.46681139e-01
5.95512390e-01 4.46292639e-01 2.17291668e-01 6.29455030e-01
-4.32416908e-02 1.28099072e+00 -4.61473018e-01 -7.20233381e-01
-1.10408151e+00 -8.99732828e-01 -4.01099980e-01 4.93693739e-01
3.59932899e-01 -2.53557980e-01 -9.53541100e-02 -5.95580824e-02
2.56596953e-01 1.94075093e-01 -1.62587196e-01 3.46668124e-01
-7.16902018e-01 -7.38289654e-01 2.85541475e-01 4.11426097e-01
4.87311572e-01 -7.69533038e-01 -4.34199244e-01 -2.57074280e-04
2.50820927e-02 -1.27728653e+00 1.73462927e-01 -7.46208876e-02
-1.36682558e+00 -1.14651120e+00 -5.11734724e-01 -6.15495920e-01
5.83785951e-01 6.85349345e-01 1.37359214e+00 2.28660434e-01
1.10804848e-01 6.20826006e-01 -3.60235929e-01 -3.76102358e-01
-1.75884087e-02 7.67936260e-02 4.14626122e-01 -4.11855489e-01
6.02721453e-01 -9.40242410e-01 -3.83058727e-01 4.76633072e-01
-5.33630550e-01 -3.72319400e-01 5.19438267e-01 4.52580690e-01
9.05769825e-01 -3.87456894e-01 -2.88926154e-01 -1.82000548e-01
2.20315605e-01 -2.32689396e-01 -1.00914037e+00 -1.57630015e-02
-7.25743651e-01 -3.95912021e-01 1.04241692e-01 -6.90162033e-02
-6.00659251e-01 5.62395871e-01 -3.31364125e-01 -6.91029608e-01
-4.22048807e-01 5.14890134e-01 -2.15927847e-02 -9.89772141e-01
7.72541583e-01 3.75811875e-01 5.98648898e-02 -5.98958671e-01
3.00968140e-01 5.39005935e-01 6.24043286e-01 -7.00671434e-01
1.38724124e+00 9.86548007e-01 2.02180639e-01 -1.03748226e+00
-3.42451245e-01 -8.08412135e-01 -1.36782444e+00 -1.89347625e-01
6.14514887e-01 -1.13264143e+00 -7.13670492e-01 5.47429323e-01
-1.44234610e+00 2.80901957e-02 -3.34888011e-01 6.48253441e-01
-6.37206256e-01 5.37154794e-01 -1.24412421e-02 -6.64091289e-01
-4.73549187e-01 -1.30301523e+00 1.67440915e+00 2.61550825e-02
3.03711351e-02 -9.91797388e-01 3.73734266e-01 3.62654507e-01
5.16353309e-01 5.24610937e-01 2.59950370e-01 -4.49802071e-01
-9.12468612e-01 -4.11000669e-01 -6.77498206e-02 4.94456999e-02
2.69756734e-01 1.06997885e-01 -1.04886007e+00 -4.76539820e-01
9.92540494e-02 9.42637622e-02 3.28095466e-01 2.80085742e-01
7.58536518e-01 1.58222258e-01 -3.74736726e-01 1.21100640e+00
1.82041562e+00 -1.32596046e-01 7.99590409e-01 8.53734136e-01
9.15763915e-01 2.85424680e-01 6.76789939e-01 3.45357537e-01
4.57502425e-01 1.03636992e+00 8.94344091e-01 -2.19468176e-01
6.17706683e-03 1.11630112e-01 -3.90939452e-02 8.66220593e-01
-8.85377467e-01 3.74281824e-01 -1.38094401e+00 4.65712875e-01
-1.61071193e+00 -7.58666277e-01 -4.40636635e-01 2.47464633e+00
3.40672404e-01 -1.95513919e-01 -3.17601919e-01 -5.26897386e-02
6.24675751e-01 2.08563745e-01 -2.89917380e-01 1.57716423e-01
-2.99872667e-01 4.30851787e-01 8.53733957e-01 5.01258492e-01
-1.19264591e+00 1.07048249e+00 6.69745541e+00 4.76301253e-01
-1.38459027e+00 8.77026543e-02 -1.39627904e-01 1.38919041e-01
-1.62509501e-01 2.10359603e-01 -8.16133380e-01 7.39813894e-02
5.86842656e-01 1.34263970e-02 4.32518214e-01 9.19168413e-01
2.09434226e-01 9.15012695e-03 -1.00234914e+00 1.46360755e+00
1.07161812e-01 -1.64652348e+00 -1.39605597e-01 2.10275143e-01
7.51015604e-01 1.04596591e+00 -2.03108460e-01 -5.77050187e-02
6.68780059e-02 -1.02351308e+00 6.54270887e-01 5.82311988e-01
8.23061466e-01 -3.43005061e-01 8.69964719e-01 2.40253344e-01
-1.37705672e+00 3.46036762e-01 -5.97825110e-01 -2.44991332e-01
9.70398039e-02 6.19570017e-01 -5.60139537e-01 1.09249258e+00
1.10405016e+00 7.91528225e-01 -5.17831028e-01 1.31807327e+00
2.09093932e-02 1.10817596e-01 -8.87279809e-01 4.19272989e-01
7.02296421e-02 -5.82406878e-01 6.15432143e-01 9.78077769e-01
6.85725629e-01 2.14036524e-01 1.66411683e-01 8.61940205e-01
2.91618735e-01 4.58091870e-03 -7.97488749e-01 5.01184046e-01
6.84989512e-01 1.30286622e+00 -6.33964300e-01 8.45289417e-03
-6.22327447e-01 5.15807807e-01 8.87295008e-02 2.04456985e-01
-5.37437737e-01 6.39287606e-02 9.00831103e-01 3.32115531e-01
1.27497107e-01 -8.60597491e-01 -6.70505702e-01 -1.13435209e+00
1.40886828e-01 -5.96356690e-01 -7.28283823e-02 -9.77796435e-01
-1.28407049e+00 6.15125239e-01 3.00366998e-01 -1.77778947e+00
-2.96807680e-02 -5.90002120e-01 -6.52653635e-01 1.15521204e+00
-1.76876330e+00 -1.38334751e+00 -8.91513705e-01 6.98893666e-01
2.60603458e-01 -3.19911182e-01 8.73166203e-01 3.19467515e-01
3.62805054e-02 7.49229565e-02 3.45355123e-01 -6.52569011e-02
4.94521350e-01 -8.54064584e-01 7.78667808e-01 7.32531667e-01
1.92653984e-01 6.64411306e-01 5.65306962e-01 -6.41557336e-01
-1.77156830e+00 -9.27822351e-01 7.49921799e-01 -7.36526668e-01
4.45086986e-01 -2.75754601e-01 -9.06015396e-01 7.63074100e-01
-1.62076607e-01 4.30681169e-01 3.47491354e-01 1.22403197e-01
6.52918592e-02 -1.70290083e-01 -1.18510258e+00 2.84150213e-01
1.05531335e+00 -4.13459510e-01 -7.44286895e-01 5.05300522e-01
3.96325111e-01 -9.98588920e-01 -1.14773321e+00 7.40536988e-01
4.11399662e-01 -1.13261461e+00 1.48257518e+00 -3.71504138e-04
-6.28453791e-02 -5.06921232e-01 -4.60218310e-01 -1.06693804e+00
-4.07285601e-01 -4.05656129e-01 1.85505971e-01 9.94279683e-01
-6.34266585e-02 -7.38099515e-01 9.50735271e-01 5.07172823e-01
-4.76294547e-01 -4.40383852e-01 -1.29100525e+00 -8.87199640e-01
4.47634188e-03 -7.12162673e-01 1.02399850e+00 1.16115093e+00
-6.52099252e-01 -1.13223061e-01 5.02652377e-02 6.91203713e-01
6.62431657e-01 3.61242801e-01 1.23446035e+00 -1.67816865e+00
4.24677789e-01 -2.44044915e-01 -7.96829104e-01 -8.37113023e-01
6.54940978e-02 -8.56072009e-01 -1.41559005e-01 -1.59159601e+00
-2.56623924e-01 -8.94410789e-01 1.49027213e-01 2.83184171e-01
2.39535168e-01 3.41987103e-01 2.43776262e-01 9.71535683e-01
-1.00582100e-01 6.65863514e-01 8.95685136e-01 -7.98046961e-02
-1.95006728e-01 1.99860916e-01 -2.17447236e-01 8.48284662e-01
6.15570903e-01 -4.62978959e-01 1.37551889e-01 -8.12708616e-01
1.70011297e-01 -2.40493402e-01 6.64141536e-01 -1.29949617e+00
4.81931597e-01 -2.06527859e-01 3.08302552e-01 -1.14806449e+00
5.50372779e-01 -1.30109096e+00 5.70643663e-01 1.33400396e-01
4.77277637e-01 4.26107764e-01 2.79714525e-01 2.61180639e-01
-3.33248913e-01 -3.78343835e-02 6.72431350e-01 -3.47463042e-01
-8.61912310e-01 5.78866303e-01 2.49334991e-01 -3.77500534e-01
8.05831373e-01 -7.61010408e-01 5.13001308e-02 -2.13610783e-01
-4.99744475e-01 -3.73742990e-02 9.66954470e-01 3.80054593e-01
7.66565859e-01 -1.52147722e+00 -8.15686703e-01 4.16043818e-01
3.08188379e-01 5.90359688e-01 -5.73100634e-02 7.98139274e-01
-9.42521334e-01 5.50116122e-01 -2.96590507e-01 -1.31834531e+00
-1.21249139e+00 2.82899141e-01 6.00264728e-01 1.63424388e-01
-6.42247677e-01 5.48305988e-01 -3.05847913e-01 -9.82840419e-01
-4.84802425e-02 -4.16797280e-01 -1.72712341e-01 -1.69000566e-01
-3.85083407e-02 4.35868919e-01 7.63061225e-01 -1.29326022e+00
-8.03828955e-01 1.48533535e+00 4.78828043e-01 7.89209902e-02
1.71922803e+00 -1.23605197e-02 -3.71482730e-01 2.01721311e-01
1.09067953e+00 -7.09330589e-02 -1.13787305e+00 -2.80275226e-01
-6.12430274e-02 -7.66075790e-01 1.27689734e-01 -2.23364443e-01
-1.23149204e+00 7.19304681e-01 9.92445350e-01 -9.56540480e-02
9.03782308e-01 1.82852820e-01 3.77202153e-01 5.55449069e-01
8.35292757e-01 -5.95951319e-01 -3.90503109e-01 7.95945466e-01
1.17996669e+00 -1.60128164e+00 7.02121556e-01 -5.63498080e-01
-2.12532923e-01 1.26296675e+00 3.18883836e-01 -3.53816867e-01
7.97207117e-01 1.51203051e-01 1.11211061e-01 -6.22355998e-01
1.05497994e-01 -2.99687922e-01 3.81962299e-01 1.05500233e+00
2.62126476e-01 -1.41468182e-01 9.07679275e-03 -1.94051284e-02
-6.02670968e-01 -5.69315515e-02 2.09745854e-01 9.78697419e-01
-2.87352234e-01 -1.05856574e+00 -9.51578736e-01 1.46406740e-01
7.45864585e-02 -6.58019185e-02 -2.37948492e-01 1.11921906e+00
3.59123871e-02 7.14757800e-01 2.68915087e-01 -3.94486040e-01
6.69969201e-01 -3.00235868e-01 5.23852408e-01 -5.00961185e-01
-6.35083079e-01 -7.60764107e-02 -2.83986717e-01 -8.28073978e-01
-9.11427021e-01 -9.36038911e-01 -9.31941092e-01 -3.93740773e-01
-4.74785924e-01 -1.64424300e-01 1.14068985e+00 1.02395487e+00
4.67923105e-01 -1.38884544e-01 5.70029080e-01 -1.83111537e+00
-3.25154483e-01 -7.18765676e-01 -4.14693266e-01 2.87584007e-01
3.27649981e-01 -8.73427749e-01 -3.70944232e-01 -8.88648704e-02]
|
[7.747100830078125, -2.835925340652466]
|
b4cb3425-f0f8-4ba1-97bb-bb6c8bb5aedd
|
lighttag-text-annotation-platform
|
2109.02320
| null |
https://arxiv.org/abs/2109.02320v1
|
https://arxiv.org/pdf/2109.02320v1.pdf
|
LightTag: Text Annotation Platform
|
Text annotation tools assume that their user's goal is to create a labeled corpus. However, users view annotation as a necessary evil on the way to deliver business value through NLP. Thus an annotation tool should optimize for the throughput of the global NLP process, not only the productivity of individual annotators. LightTag is a text annotation tool designed and built on that principle. This paper shares our design rationale, data modeling choices, and user interface decisions then illustrates how those choices serve the full NLP lifecycle.
|
['Tal Perry']
|
2021-09-06
| null |
https://aclanthology.org/2021.emnlp-demo.3
|
https://aclanthology.org/2021.emnlp-demo.3.pdf
|
emnlp-acl-2021-11
|
['text-annotation']
|
['natural-language-processing']
|
[ 7.78104439e-02 9.13326561e-01 -5.30554056e-01 -3.52604300e-01
-6.56309903e-01 -1.15180254e+00 5.16849160e-01 5.39618850e-01
-4.51918274e-01 5.53386807e-01 6.25777364e-01 -4.52524602e-01
6.64032623e-02 -4.54439551e-01 1.42100938e-02 -1.02319822e-01
7.13560700e-01 9.88099873e-01 1.47004068e-01 -6.49317056e-02
3.54933172e-01 3.42331141e-01 -8.41207743e-01 3.29876661e-01
4.57863897e-01 7.31193423e-01 1.04536280e-01 5.39729059e-01
-1.03564453e+00 1.14855266e+00 -5.35452843e-01 -4.99878615e-01
2.84044832e-01 -2.30129853e-01 -1.35199404e+00 -1.88815266e-01
-2.82597512e-01 9.36671272e-02 2.91814178e-01 9.73194480e-01
3.50440919e-01 -4.48422804e-02 4.23752040e-01 -1.29215801e+00
-4.57396656e-01 1.08503520e+00 -1.54176608e-01 -1.92317635e-01
3.69383723e-01 -1.23107815e-02 1.26185346e+00 -5.64755619e-01
1.07624638e+00 7.23680317e-01 6.40752375e-01 3.19129735e-01
-8.84378195e-01 -4.00472254e-01 -3.55003208e-01 -4.27730978e-01
-1.14028454e+00 -6.86882615e-01 1.97303161e-01 -7.75661111e-01
1.17036402e+00 4.17802930e-01 6.08419597e-01 3.45422119e-01
1.22763449e-02 4.86434519e-01 5.92117608e-01 -8.08716536e-01
6.89346120e-02 4.22940373e-01 3.23894560e-01 5.72200894e-01
3.30014199e-01 -7.77590454e-01 -7.63032436e-01 -2.40333989e-01
3.75270039e-01 -4.03685361e-01 -5.13279065e-02 -5.65956309e-02
-1.00541294e+00 4.70094323e-01 -4.36525375e-01 6.74361885e-01
-2.61501938e-01 2.95270622e-01 8.56362045e-01 2.88418457e-02
6.40966952e-01 7.34241486e-01 -8.73641372e-01 -6.72487378e-01
-7.18589067e-01 -7.69890919e-02 1.47454548e+00 1.34460151e+00
5.93008816e-01 -4.37805891e-01 -2.40329131e-01 7.22497702e-01
5.48821092e-01 1.00304857e-01 3.83000702e-01 -1.14766896e+00
4.91141260e-01 1.14206767e+00 7.40866244e-01 -5.12219131e-01
-5.24784029e-01 -1.48122549e-01 8.87142047e-02 -2.97737002e-01
6.58727705e-01 -3.30213606e-01 -4.87260520e-01 9.64331865e-01
3.04495245e-01 -1.07929766e+00 6.79352134e-02 4.66884315e-01
7.62957454e-01 4.37520862e-01 7.93868780e-01 -3.89112681e-01
1.59370959e+00 -8.84960949e-01 -1.29645383e+00 -3.33375305e-01
1.48723114e+00 -1.03133953e+00 1.18471885e+00 -2.07710499e-03
-1.17452371e+00 -5.30671813e-02 -5.07985950e-01 -6.22729301e-01
-6.11472964e-01 2.68084139e-01 8.00699413e-01 7.88789272e-01
-6.79961860e-01 3.54387909e-01 -6.73538089e-01 -5.37471175e-01
3.03746849e-01 1.35013685e-01 -2.44152561e-01 3.36538702e-01
-7.88862169e-01 9.48628247e-01 7.94185042e-01 -2.64913887e-01
-2.30360869e-02 -5.62062800e-01 -2.34774321e-01 4.84179795e-01
6.80859387e-01 -4.17976946e-01 1.80165362e+00 -6.97655499e-01
-1.14104211e+00 9.93729293e-01 -9.84933227e-02 -1.94741234e-01
5.81815481e-01 -2.54414231e-01 -1.57468036e-01 -3.24081063e-01
3.24944615e-01 4.33590174e-01 -1.52959034e-01 -8.38025928e-01
-1.09255779e+00 -9.92523283e-02 -2.86472619e-01 4.15564269e-01
-2.66545147e-01 6.61493182e-01 -5.44633448e-01 -1.20564595e-01
-7.81144947e-02 -7.04871416e-01 -1.54002070e-01 -2.50345230e-01
-1.23160899e-01 -5.79047084e-01 7.27171004e-01 -7.66453922e-01
1.40858781e+00 -2.04467988e+00 -4.85611796e-01 1.43555105e-01
3.02338898e-01 -2.40313280e-02 4.49406773e-01 9.29568946e-01
2.99954891e-01 9.55891728e-01 3.65398556e-01 -1.84853822e-01
5.23995042e-01 1.99372888e-01 -1.98679402e-01 2.28438407e-01
-4.14390266e-01 7.39590764e-01 -8.93499255e-01 -7.74785578e-01
-1.40444398e-01 -3.68731879e-02 -1.58553287e-01 -1.89039379e-01
-6.01843119e-01 2.25313291e-01 -6.77315593e-01 6.29685700e-01
-7.66429529e-02 -3.69376063e-01 7.05769002e-01 3.16252187e-02
-7.65227139e-01 4.67786670e-01 -6.55533433e-01 1.54530895e+00
-2.95918673e-01 6.68190122e-01 2.57951021e-01 -4.72536951e-01
1.03337955e+00 7.56720662e-01 8.21789265e-01 -3.50376070e-01
1.63924038e-01 4.38180268e-01 -3.37279998e-02 -7.60415554e-01
8.16506743e-01 -8.54729563e-02 -2.92550832e-01 7.94815063e-01
1.80535726e-02 -6.33396357e-02 3.72559756e-01 2.32434332e-01
1.01745069e+00 4.36400145e-01 7.43448079e-01 -3.08812201e-01
-4.05580327e-02 7.64372468e-01 6.33046150e-01 5.50146937e-01
-1.04865037e-01 -1.63456038e-01 9.49365437e-01 -6.07356548e-01
-1.48702514e+00 -2.04171300e-01 -4.49037179e-02 1.25944293e+00
-3.06848675e-01 -6.32179558e-01 -7.22863734e-01 -8.70324910e-01
-3.81876916e-01 1.05058682e+00 -1.49550200e-01 5.75412631e-01
-2.02986404e-01 -2.84273267e-01 6.15003347e-01 3.43991518e-01
4.00493413e-01 -9.62251484e-01 -8.65610957e-01 2.13798210e-01
-4.33858305e-01 -1.23123980e+00 -4.20388550e-01 3.01866740e-01
-1.82777956e-01 -9.99767363e-01 1.28362373e-01 -6.74134612e-01
4.51848745e-01 -1.97992310e-01 9.01272297e-01 -2.45690774e-02
-6.54204190e-02 4.45943683e-01 -6.42234504e-01 -9.07145202e-01
-6.90941453e-01 3.36858630e-01 -3.89475256e-01 -6.25672221e-01
8.49951208e-01 -1.50975212e-01 -1.58155456e-01 3.83625388e-01
-6.20220780e-01 5.46751440e-01 2.17746556e-01 2.33933270e-01
6.19321704e-01 2.44370475e-01 5.29371619e-01 -1.27122271e+00
7.93228567e-01 -3.29511315e-01 -4.83666390e-01 6.56822562e-01
-7.98588574e-01 -1.24614075e-01 2.31060386e-01 1.02574214e-01
-1.35474932e+00 3.72574925e-01 -1.89157635e-01 3.47025990e-01
2.27518585e-02 6.56255484e-01 -3.79147947e-01 2.70879596e-01
6.46115839e-01 -3.47944438e-01 -2.69247711e-01 -6.38946652e-01
5.14457941e-01 6.96025729e-01 2.27443010e-01 -4.51321214e-01
2.77175069e-01 2.82042652e-01 -1.63475826e-01 -4.62042332e-01
-8.44631851e-01 -8.65141809e-01 -7.97821760e-01 -4.78763074e-01
9.22070026e-01 -5.71583927e-01 -8.46330464e-01 -4.02765155e-01
-1.11023450e+00 -6.89163506e-01 -7.62729287e-01 1.89909145e-01
-6.49915457e-01 5.85938357e-02 -2.75469571e-01 -9.86092269e-01
-6.10533714e-01 -7.17789829e-01 6.15554214e-01 3.87008488e-02
-9.37901914e-01 -1.01449454e+00 -4.83975820e-02 7.26249218e-01
6.02852441e-02 2.70778328e-01 7.28712976e-01 -1.47229576e+00
-1.67246297e-01 -4.73132014e-01 -3.31576586e-01 -8.60022940e-03
5.79038337e-02 1.04912862e-01 -7.17846870e-01 3.96851063e-01
-1.58969343e-01 -1.21719263e-01 -2.46760353e-01 -4.69910316e-02
7.80066192e-01 -8.46500337e-01 -5.62868774e-01 -4.50986885e-02
1.52124941e+00 3.87427926e-01 4.28678274e-01 3.83396149e-01
6.47719502e-01 1.11488700e+00 8.60792279e-01 3.44549656e-01
1.15112446e-01 5.48674703e-01 -1.30021349e-01 2.88085997e-01
1.71474025e-01 -4.06545579e-01 -1.92261890e-01 6.43790841e-01
-8.02016556e-02 -6.32454574e-01 -1.66539228e+00 4.40423906e-01
-2.23698115e+00 -8.34404826e-01 -4.57531780e-01 1.60612607e+00
8.42188537e-01 1.79004848e-01 -6.18297346e-02 -7.15520605e-02
2.88683861e-01 -2.68535644e-01 -1.87774584e-01 -6.00465834e-01
2.83922881e-01 -2.83133298e-01 1.00706172e+00 2.56120682e-01
-4.48217779e-01 9.98795509e-01 6.56848955e+00 6.18047655e-01
-5.22340178e-01 5.47234178e-01 4.39736158e-01 -8.81350338e-02
-2.77366668e-01 3.54269058e-01 -1.10498929e+00 3.68815869e-01
1.02890623e+00 -4.82085526e-01 2.00657770e-01 1.07848144e+00
6.44223392e-01 -1.61220536e-01 -1.06842387e+00 4.82012004e-01
-3.06673616e-01 -1.61946332e+00 -4.19061124e-01 3.57477933e-01
4.16011959e-01 1.00925066e-01 -4.96759295e-01 8.77376571e-02
9.27718997e-01 -7.85466015e-01 1.03487718e+00 5.32700837e-01
8.20102751e-01 -3.97865295e-01 8.68092418e-01 5.60940921e-01
-8.34553778e-01 -1.73117351e-02 5.25551550e-02 -9.51944962e-02
6.74988091e-01 5.45372128e-01 -1.30296862e+00 5.91925718e-02
6.05655730e-01 -1.08024850e-01 -2.08224535e-01 5.52245259e-01
-1.53356373e-01 5.21397948e-01 -2.04223782e-01 -1.20784849e-01
1.91199064e-01 -1.69077769e-01 3.11170191e-01 1.58506966e+00
8.01105797e-02 3.04190576e-01 4.37698245e-01 3.94376069e-01
-3.68125528e-01 7.91095078e-01 -4.42730010e-01 -8.23739886e-01
9.54104304e-01 1.42624915e+00 -1.19854856e+00 -4.87447858e-01
-5.10176241e-01 3.40091646e-01 2.84466930e-02 7.02944724e-03
-4.97524559e-01 -3.57008815e-01 2.70756364e-01 3.91764194e-01
-2.27608576e-01 -2.10501403e-01 -8.05443764e-01 -5.22480786e-01
-1.49059638e-01 -6.10050559e-01 3.05194736e-01 -8.72923493e-01
-5.67675531e-01 -1.02557018e-01 -1.63236797e-01 -6.05325818e-01
-1.83177844e-01 -4.34591979e-01 -1.50369704e-01 8.68890524e-01
-7.99811482e-01 -1.41527355e+00 -2.69176334e-01 -2.25337043e-01
4.30153608e-01 2.80656666e-01 8.78509521e-01 1.60128042e-01
-2.50290841e-01 -6.06398396e-02 -5.11053484e-03 3.45685124e-01
6.00114048e-01 -1.13942015e+00 3.20740938e-01 4.34355766e-01
8.98391902e-02 6.45941973e-01 6.23356521e-01 -9.67541218e-01
-1.06409919e+00 -8.85057211e-01 1.65168476e+00 -7.10537672e-01
9.53757584e-01 -2.25688562e-01 -5.24792731e-01 1.10650134e+00
4.30111110e-01 -5.09198368e-01 1.20117748e+00 2.50889868e-01
7.93943740e-03 1.69465065e-01 -1.15407228e+00 4.33419496e-01
1.02226496e+00 -4.55351621e-01 -4.28633779e-01 7.89854169e-01
8.57028306e-01 -5.28372884e-01 -1.19057357e+00 -1.81089938e-01
8.26759934e-01 -2.89186805e-01 3.23303580e-01 -5.09312928e-01
1.06110908e-01 -2.74899781e-01 -7.55329952e-02 -6.95455492e-01
-2.12941661e-01 -9.60530162e-01 2.13362888e-01 1.67879641e+00
8.31552923e-01 -4.20156360e-01 6.80737615e-01 1.52527833e+00
-3.44192713e-01 -3.01764756e-01 -4.77740943e-01 -4.95728612e-01
-3.47125351e-01 -5.84867358e-01 7.52118587e-01 1.32046103e+00
7.32747138e-01 6.69540405e-01 5.42072244e-02 -2.26388231e-01
1.01949587e-01 -5.31727195e-01 8.07835042e-01 -1.43220687e+00
3.31774019e-02 -4.25263166e-01 2.21216008e-01 -5.08084297e-01
-1.84518352e-01 -7.31040895e-01 3.07406578e-03 -2.00328016e+00
2.47989595e-01 -7.29326725e-01 2.96655595e-01 9.32854414e-01
4.82049286e-01 -1.68468326e-01 2.04622418e-01 7.45070696e-01
-6.35125041e-01 -5.67764401e-01 9.91754234e-01 2.21301362e-01
-5.12404978e-01 -2.57564932e-01 -9.88714278e-01 8.63491535e-01
1.07624412e+00 -5.95260024e-01 -3.90817761e-01 -4.60407823e-01
1.10124302e+00 -1.41612619e-01 -2.45612681e-01 -5.43988526e-01
6.79039657e-01 -5.58962345e-01 4.06214148e-02 -4.35028613e-01
-1.73897505e-01 -1.04157591e+00 6.87456071e-01 1.12323597e-01
-6.72148645e-01 -1.05130516e-01 -7.90167600e-02 1.06376633e-01
9.86919552e-02 -8.21083367e-01 3.17326218e-01 -3.71469736e-01
-4.65753973e-01 1.31483739e-02 -4.75417376e-01 1.26987725e-01
8.99855196e-01 -2.07620010e-01 -3.38868469e-01 -8.79950523e-02
-8.92175496e-01 3.23442996e-01 7.13789165e-01 -8.31666961e-02
-4.33745176e-01 -7.07728624e-01 -2.79720694e-01 -5.01846075e-01
6.34218231e-02 2.60764569e-01 -3.11866611e-01 6.63664639e-01
-8.06777000e-01 8.61228704e-01 -2.52039079e-02 6.09670486e-03
-1.37959731e+00 3.92950803e-01 7.90980682e-02 -5.46092331e-01
-6.43982232e-01 5.44449568e-01 -1.94371030e-01 -1.68468058e-01
1.43097222e-01 -1.42404303e-01 -3.28513443e-01 3.85273188e-01
4.84436095e-01 4.09722000e-01 -1.40115721e-02 -3.28948349e-01
-2.44659651e-02 -4.66479175e-02 2.37995252e-01 -5.75130761e-01
1.22077274e+00 -5.05767703e-01 -3.51906478e-01 6.07285380e-01
6.44738913e-01 2.95162529e-01 -6.93328142e-01 -1.42034411e-01
6.31284475e-01 -2.98972100e-01 4.00055684e-02 -1.17508042e+00
-5.28820813e-01 2.47136354e-01 -8.64426717e-02 6.03977561e-01
5.95497787e-01 2.02327579e-01 4.76884902e-01 4.12326962e-01
1.82675272e-01 -1.75661886e+00 -2.87918746e-01 2.22088873e-01
5.44934928e-01 -7.67469704e-01 1.53966218e-01 -6.18875384e-01
-7.74760604e-01 1.15745986e+00 3.11225176e-01 8.59762490e-01
4.57501531e-01 4.75347847e-01 8.36269092e-03 -6.66522324e-01
-6.07236207e-01 1.36135565e-02 -2.91802615e-01 5.08489728e-01
8.18872571e-01 2.02808678e-02 -8.67758989e-01 8.27315569e-01
-2.37496898e-01 4.17991221e-01 2.91647702e-01 1.04765272e+00
-7.00464785e-01 -1.17827976e+00 -1.87298376e-02 5.02254546e-01
-8.85216475e-01 -7.85424002e-03 -7.95754850e-01 8.96156728e-01
4.26520914e-01 7.64495134e-01 1.06555000e-01 1.50164187e-01
2.41926938e-01 6.35195911e-01 -6.22403622e-02 -1.03342116e+00
-8.41526687e-01 4.85616177e-02 1.10881019e+00 -1.91377655e-01
-5.13692617e-01 -6.96272671e-01 -1.57477093e+00 -4.16653365e-01
-5.57917297e-01 8.10804963e-01 1.17877519e+00 1.01675057e+00
2.96766996e-01 2.17922136e-01 -4.03020345e-02 -1.53385336e-02
-3.49279167e-03 -1.05299187e+00 -5.25186419e-01 5.15350606e-03
-4.40937966e-01 -1.97715685e-02 6.39960840e-02 6.49864912e-01]
|
[9.411933898925781, 8.786335945129395]
|
b5551a82-5839-4ba5-b763-cfb432b0c2eb
|
a-tangled-web-the-faint-signals-of-deception
| null | null |
https://aclanthology.org/L16-1558
|
https://aclanthology.org/L16-1558.pdf
|
A Tangled Web: The Faint Signals of Deception in Text - Boulder Lies and Truth Corpus (BLT-C)
|
We present an approach to creating corpora for use in detecting deception in text, including a discussion of the challenges peculiar to this task. Our approach is based on soliciting several types of reviews from writers and was implemented using Amazon Mechanical Turk. We describe the multi-dimensional corpus of reviews built using this approach, available free of charge from LDC as the Boulder Lies and Truth Corpus (BLT-C). Challenges for both corpus creation and the deception detection include the fact that human performance on the task is typically at chance, that the signal is faint, that paid writers such as turkers are sometimes deceptive, and that deception is a complex human behavior; manifestations of deception depend on details of domain, intrinsic properties of the deceiver (such as education, linguistic competence, and the nature of the intention), and specifics of the deceptive act (e.g., lying vs. fabricating.) To overcome the inherent lack of ground truth, we have developed a set of semi-automatic techniques to ensure corpus validity. We present some preliminary results on the task of deception detection which suggest that the BLT-C is an improvement in the quality of resources available for this task.
|
['Franco Salvetti', 'James H. Martin', 'John B. Lowe']
|
2016-05-01
|
a-tangled-web-the-faint-signals-of-deception-1
|
https://aclanthology.org/L16-1558
|
https://aclanthology.org/L16-1558.pdf
|
lrec-2016-5
|
['deception-detection']
|
['miscellaneous']
|
[-4.18866239e-02 3.94005366e-02 1.53258577e-01 -5.92261136e-01
-7.39332020e-01 -8.22132289e-01 9.22505021e-01 -2.16911500e-03
-4.66624260e-01 6.33502543e-01 3.36076170e-01 -2.88117796e-01
1.98772848e-01 -1.01569884e-01 -3.77617776e-01 -2.71612078e-01
5.20902753e-01 4.86419171e-01 -2.33260319e-01 -4.35150295e-01
1.00011432e+00 3.41900319e-01 -1.06705117e+00 2.52472937e-01
8.62453878e-01 6.94921672e-01 -3.53206903e-01 8.30786169e-01
1.36645421e-01 1.04223943e+00 -1.21071446e+00 -9.29569006e-01
3.18641998e-02 -7.15445459e-01 -9.65871572e-01 2.14918330e-01
5.68940043e-01 -6.08193755e-01 6.11770973e-02 1.00990760e+00
2.19145834e-01 -4.25885106e-03 8.04123938e-01 -9.87507939e-01
-9.41984415e-01 3.23251992e-01 -2.84418672e-01 3.97257596e-01
7.16390669e-01 8.34940970e-02 6.89703882e-01 -9.50302601e-01
5.59973419e-01 1.18470621e+00 3.92872453e-01 8.17520797e-01
-8.19059908e-01 -6.04501009e-01 -4.11528766e-01 2.63384283e-01
-1.13686490e+00 -1.08766854e+00 7.73813248e-01 -7.40333498e-01
5.43282747e-01 1.68273613e-01 3.62305105e-01 1.70249724e+00
2.30268940e-01 6.47584498e-01 1.27256143e+00 -6.71044767e-01
4.19915795e-01 6.95032597e-01 4.01885182e-01 5.39501250e-01
2.28022128e-01 -9.41298231e-02 -8.67231965e-01 -5.74903309e-01
1.73754781e-01 -7.73858011e-01 -1.71786651e-01 -1.79007620e-01
-8.31027627e-01 9.32279050e-01 -2.09013477e-01 5.42545557e-01
3.61825451e-02 -2.95576453e-01 8.23260844e-01 3.15413147e-01
4.95447993e-01 5.74629068e-01 -5.39419539e-02 -7.57538438e-01
-1.04374588e+00 2.08482414e-01 1.17757308e+00 6.73663914e-01
5.59545029e-03 1.28095523e-02 2.51150757e-01 7.16235995e-01
1.10525019e-01 2.80682594e-01 8.46232116e-01 -7.46292531e-01
4.75419551e-01 4.95020419e-01 5.47553003e-01 -1.26580691e+00
3.79990004e-02 1.32358417e-01 -4.35099863e-02 1.12155929e-01
7.69036353e-01 -2.37118483e-01 -5.17224252e-01 1.12737250e+00
8.74953344e-02 -4.17630732e-01 -8.90890434e-02 1.15483022e+00
6.40706599e-01 2.56415993e-01 -2.08069548e-01 -2.20514506e-01
1.31028366e+00 -4.29779202e-01 -8.45786929e-01 -3.63602340e-01
7.48994827e-01 -7.44362533e-01 1.06881702e+00 5.63138783e-01
-1.06780779e+00 -1.97563007e-01 -1.20191848e+00 -2.42266431e-01
-4.38071072e-01 1.87483117e-01 1.64953545e-01 1.43027401e+00
-6.83350861e-01 3.73065442e-01 -3.40841919e-01 -2.65426010e-01
2.70331949e-01 -6.35547861e-02 -4.97959942e-01 -1.53320342e-01
-1.05905855e+00 1.44111133e+00 -2.89197415e-02 4.20984745e-01
-5.68968117e-01 6.00832589e-02 -9.78467882e-01 -2.64532477e-01
1.73962489e-01 8.90872478e-02 1.10106409e+00 -1.42430270e+00
-1.55586040e+00 1.32833123e+00 -1.30572811e-01 -1.64107293e-01
8.40838790e-01 -1.99649483e-01 -7.45394289e-01 1.37306988e-01
2.83120513e-01 -1.07448578e-01 1.20697045e+00 -1.06391954e+00
-1.80246681e-01 -7.26077557e-01 -2.97888458e-01 1.86564550e-01
-1.73122823e-01 6.87793136e-01 3.26553583e-01 -1.42248064e-01
-2.63446301e-01 -7.24440575e-01 6.09014511e-01 -4.76043314e-01
-3.17848772e-01 -1.60619453e-01 9.40628350e-01 -9.08920228e-01
8.76255214e-01 -2.30462122e+00 -4.43908513e-01 6.70748278e-02
3.42656106e-01 6.59128904e-01 1.32844165e-01 5.98793387e-01
2.06241548e-01 5.35531640e-01 1.65661685e-02 -3.28624457e-01
1.13112740e-01 1.29551530e-01 -9.94372666e-02 9.61968005e-01
-3.74044105e-02 8.07476401e-01 -8.96389127e-01 -4.66699809e-01
-1.20303310e-01 1.24049425e-01 2.83925265e-01 -2.09991485e-02
3.29468757e-01 7.19003826e-02 -3.02215964e-01 4.20396119e-01
4.55850601e-01 3.60924959e-01 9.51792896e-02 5.87768912e-01
-3.53078544e-01 7.33422220e-01 -6.31290078e-01 9.59205449e-01
-1.36785880e-01 1.16341293e+00 5.46930909e-01 -6.34312332e-01
1.07540894e+00 4.53872561e-01 -4.86115187e-01 -3.94704849e-01
2.69311011e-01 5.98462403e-01 2.52751529e-01 -6.77645922e-01
9.44574952e-01 -2.68646419e-01 -1.87607020e-01 8.41466367e-01
-1.38136223e-01 -4.70669121e-01 1.88430116e-01 3.67920220e-01
8.61484110e-01 9.07723661e-05 2.89611042e-01 -4.84039247e-01
4.93770272e-01 1.43886983e-01 3.98631871e-01 5.40489614e-01
-8.53171825e-01 5.21059692e-01 9.12002921e-01 -4.51384515e-01
-1.05565810e+00 -3.88764143e-01 -1.85674459e-01 7.09446847e-01
-1.51224747e-01 -9.95206535e-02 -8.25526357e-01 -9.66188133e-01
-1.32371098e-01 1.13758564e+00 -7.43235171e-01 -2.76786953e-01
-3.45736533e-01 -3.85858953e-01 8.34048569e-01 -3.10894661e-02
4.29287404e-01 -8.23547781e-01 -7.33207345e-01 -2.52053048e-03
-1.58815712e-01 -1.25811696e+00 -6.08087361e-01 -3.01086724e-01
-3.30103368e-01 -1.12316620e+00 -1.35879487e-01 -4.18565989e-01
5.62313795e-01 6.92847222e-02 9.62534666e-01 2.28192985e-01
-7.71377236e-02 2.99665391e-01 -5.12097120e-01 -5.35491943e-01
-8.44286799e-01 -3.90056968e-01 -6.13147803e-02 6.99813291e-02
7.15956807e-01 -1.08316459e-01 -2.00562358e-01 3.08563858e-01
-6.54987633e-01 -5.27513802e-01 2.18196094e-01 7.36056924e-01
-5.59273362e-01 -8.37691948e-02 5.33319831e-01 -1.05702126e+00
1.45361865e+00 -4.26857650e-01 -2.69036472e-01 -2.59553082e-02
-3.62197071e-01 -3.72294217e-01 4.72273350e-01 -2.43052587e-01
-9.79053676e-01 -3.08086902e-01 1.91716209e-01 1.00388236e-01
-4.09983814e-01 1.52098551e-01 1.07594661e-01 -1.46830559e-01
8.92247200e-01 2.45708138e-01 4.32293564e-01 -2.01880977e-01
-5.54192103e-02 1.14125764e+00 2.29875311e-01 -4.28791255e-01
4.47084635e-01 -2.55773719e-02 -3.76250774e-01 -9.83396828e-01
-7.76862919e-01 -4.93958384e-01 -6.87434137e-01 -3.75645459e-01
4.56178337e-01 -5.07398129e-01 -7.58008957e-01 7.02149749e-01
-1.09481704e+00 -1.61026105e-01 1.25548482e-01 3.59524906e-01
-1.49848014e-01 6.06906116e-01 -6.57028258e-01 -1.25331891e+00
-3.23437214e-01 -8.44673753e-01 6.85687959e-01 -2.18326271e-01
-6.19686604e-01 -9.64706719e-01 6.21536374e-02 1.11799586e+00
2.48483509e-01 3.32728982e-01 8.04874241e-01 -1.18299019e+00
2.65173912e-01 -6.87113404e-01 -9.72362906e-02 8.11614633e-01
3.67239094e-03 3.50249201e-01 -7.60336995e-01 1.28345788e-01
5.80502391e-01 -1.01996565e+00 1.33466244e-01 -2.07289517e-01
2.49893412e-01 -5.15534878e-01 2.38652453e-01 -4.32753593e-01
9.60807443e-01 1.03115924e-01 6.20310962e-01 8.97276551e-02
4.43698019e-01 8.36071312e-01 4.88628000e-01 3.35444003e-01
5.30833781e-01 5.72945237e-01 1.23578928e-01 6.07214153e-01
3.91765863e-01 2.93458183e-03 7.32521653e-01 7.06409216e-01
1.92045286e-01 -2.18191400e-01 -1.01143861e+00 6.25033021e-01
-1.52344120e+00 -9.28080857e-01 -6.75741494e-01 1.97995007e+00
7.54315376e-01 1.21141940e-01 3.05783987e-01 3.26113760e-01
7.68064022e-01 6.90007061e-02 -1.75829872e-01 -1.51328111e+00
1.03405699e-01 -2.55528182e-01 -2.42568869e-02 7.14616239e-01
-5.51437914e-01 7.81673133e-01 6.79602861e+00 4.51528519e-01
-8.10074210e-01 3.08056101e-02 5.70457637e-01 -5.98459169e-02
3.97026949e-02 -3.63160640e-01 -4.63827938e-01 7.60853946e-01
1.09189975e+00 -9.33650658e-02 2.85046369e-01 6.68930709e-01
4.62825865e-01 -5.70428431e-01 -1.12554610e+00 5.79748988e-01
8.41640592e-01 -5.78288257e-01 -4.02141899e-01 1.28032461e-01
3.37650687e-01 -2.97898591e-01 -3.80042121e-02 2.16076672e-01
2.41854593e-01 -9.95933592e-01 1.11748397e+00 1.31918877e-01
4.65802819e-01 -6.63429499e-01 1.10598361e+00 7.44646430e-01
6.68364242e-02 2.18012616e-01 -2.19200283e-01 -4.32539433e-01
-1.28809586e-01 6.14636421e-01 -1.00949061e+00 1.42145693e-01
1.35168731e-01 2.50433683e-01 -6.61091506e-01 3.14982712e-01
-6.02797747e-01 7.65538573e-01 -2.54559834e-02 -5.98960757e-01
1.49597898e-01 -2.32478410e-01 6.69306278e-01 1.10165143e+00
-1.19873263e-01 1.54204071e-01 -2.76306480e-01 9.79396284e-01
-1.98029563e-01 8.61412212e-02 -5.59263170e-01 -3.88092220e-01
2.93469191e-01 1.38204253e+00 -3.09317857e-01 -1.19167224e-01
-1.36901096e-01 1.08409536e+00 3.53895187e-01 5.45649156e-02
-3.68812531e-01 -3.56221884e-01 3.37042272e-01 1.36359468e-01
-1.26704993e-02 -3.21411282e-01 -5.16508877e-01 -1.19064760e+00
2.90425301e-01 -1.22998583e+00 -3.81620228e-02 -8.21912289e-01
-1.19525957e+00 3.63575876e-01 -2.94751108e-01 -4.51134980e-01
-5.16091943e-01 -6.94396615e-01 -6.11352384e-01 9.93182719e-01
-8.01597893e-01 -9.69460428e-01 -6.32790551e-02 4.45409000e-01
3.60602021e-01 -9.80702639e-02 6.92769110e-01 -1.51404023e-01
-5.66511631e-01 4.10484225e-01 -1.30538732e-01 3.24740738e-01
6.88674033e-01 -1.17173946e+00 2.78447151e-01 8.25804174e-01
5.16566075e-02 6.62849784e-01 8.10313046e-01 -5.54536521e-01
-1.17920172e+00 -1.55405000e-01 1.37591112e+00 -1.04778481e+00
9.43282664e-01 -8.16339076e-01 -7.11388528e-01 5.09827435e-01
1.03554122e-01 -3.28853637e-01 8.85052800e-01 2.23302558e-01
-5.00455320e-01 3.90874714e-01 -1.52965569e+00 3.53889287e-01
4.61092740e-01 -8.81014049e-01 -9.74771142e-01 3.54728729e-01
-2.35911265e-01 -2.96482325e-01 -2.60398567e-01 -4.35197026e-01
6.66607082e-01 -1.19076478e+00 1.69853307e-02 -4.74398643e-01
7.85012245e-01 1.92860827e-01 2.54830390e-01 -1.33085775e+00
-6.39904737e-02 -6.37682676e-01 2.30841711e-01 1.35825479e+00
2.87709296e-01 -5.27851820e-01 6.57700479e-01 1.38531733e+00
7.44034478e-04 -2.40021929e-01 -1.02254999e+00 -4.96126890e-01
2.85224468e-01 -2.90342510e-01 3.34328040e-02 1.11573553e+00
6.48612022e-01 5.54463387e-01 -4.98964071e-01 -2.97563225e-01
3.06192547e-01 -2.47160912e-01 7.57169127e-01 -1.12830424e+00
1.65232867e-01 -1.08419530e-01 -5.04719138e-01 -5.61750174e-01
4.19022977e-01 -2.29402915e-01 1.92850694e-01 -8.43596697e-01
2.29607020e-02 1.76255554e-01 4.85666811e-01 1.11708224e-01
-1.49313137e-01 6.66490123e-02 9.64380577e-02 3.04375798e-01
-4.01977599e-01 2.45647028e-01 1.14610767e+00 2.48324990e-01
-1.30827472e-01 -8.13418720e-03 -9.20554221e-01 6.46972179e-01
8.53117406e-01 -4.70194072e-01 -1.32458627e-01 -2.23108262e-01
2.98164517e-01 -5.78147452e-03 3.78092080e-01 -4.20940995e-01
2.07933366e-01 5.06191663e-02 3.35164994e-01 -1.29136875e-01
4.15732145e-01 -5.22915542e-01 -4.31361735e-01 1.77403793e-01
-3.55048001e-01 -2.80039907e-02 -7.09089711e-02 3.31342250e-01
-1.60372779e-01 -9.28583980e-01 7.31488824e-01 -3.97235394e-01
-2.87482202e-01 -6.55377924e-01 -8.41806829e-01 3.00825983e-01
8.59438002e-01 -4.08809751e-01 -5.65546155e-01 -7.63563514e-01
-2.38572463e-01 -1.13052450e-01 7.17271745e-01 2.91137427e-01
4.50473547e-01 -6.37604892e-01 -8.84881914e-01 6.57968745e-02
3.20024118e-02 -7.24342167e-01 -2.20837235e-01 6.93762362e-01
-4.90853757e-01 3.34656805e-01 -2.71147698e-01 1.79804847e-01
-1.40177774e+00 2.27263257e-01 2.38579050e-01 1.37863442e-01
-3.39557439e-01 5.84023952e-01 -3.35767090e-01 -1.37366861e-01
-2.12823048e-01 3.76405388e-01 -2.00829983e-01 2.24831551e-01
6.84573114e-01 7.23131299e-01 3.05916101e-01 -1.11534369e+00
-4.27736044e-01 -2.80290306e-01 -4.40045089e-01 -3.99953783e-01
1.00917506e+00 -2.10591212e-01 -3.15663159e-01 7.37024188e-01
1.09773195e+00 4.68385458e-01 -5.75429499e-01 2.61756271e-01
2.56954450e-02 -8.09586287e-01 8.24092701e-02 -1.04301620e+00
-5.00592172e-01 6.24881625e-01 -1.85320601e-01 6.71223581e-01
4.08301234e-01 -3.87315303e-01 5.70899248e-01 2.06328154e-01
3.40611368e-01 -1.52985477e+00 9.96522680e-02 4.70456302e-01
1.12239432e+00 -1.26111007e+00 2.15838581e-01 -1.90053642e-01
-9.76546228e-01 1.15179586e+00 4.48919326e-01 -6.85403869e-02
1.31600335e-01 -7.93143213e-02 3.88440967e-01 -4.55637485e-01
-5.17370641e-01 2.68363059e-01 1.64215669e-01 6.26470208e-01
5.29978037e-01 6.27398416e-02 -8.56816769e-01 5.10146320e-01
-5.51273227e-01 -2.48430938e-01 1.14489186e+00 8.78590882e-01
-1.40337467e-01 -8.79545391e-01 -4.51365173e-01 3.99665505e-01
-8.66090119e-01 -4.76323674e-03 -1.71360838e+00 8.05283010e-01
-7.01700449e-02 1.44447422e+00 -2.32719705e-01 -2.00114876e-01
3.13923150e-01 3.17201734e-01 3.15404385e-01 -7.14399040e-01
-1.01931322e+00 -5.62998354e-01 8.07943344e-01 -2.60402620e-01
-1.13272361e-01 -9.43371892e-01 -4.88828123e-01 -6.90542221e-01
-5.73754907e-01 3.33710760e-01 9.97666001e-01 1.24545729e+00
1.58967689e-01 -4.29927647e-01 4.48463112e-01 -6.35769129e-01
-8.41515362e-01 -1.14154887e+00 -9.64823425e-01 5.95481694e-01
3.44466746e-01 -2.72437632e-01 -9.47101533e-01 -1.13913372e-01]
|
[8.236237525939941, 10.409440040588379]
|
bfe0adae-a570-4629-b125-040a8a039aae
|
a-multi-source-graph-representation-of-the
| null | null |
https://aclanthology.org/2022.lrec-1.138
|
https://aclanthology.org/2022.lrec-1.138.pdf
|
A Multi-source Graph Representation of the Movie Domain for Recommendation Dialogues Analysis
|
In dialogue analysis, characterising named entities in the domain of interest is relevant in order to understand how people are making use of them for argumentation purposes. The movie recommendation domain is a frequently considered case study for many applications and by linguistic studies and, since many different resources have been collected throughout the years to describe it, a single database combining all these data sources is a valuable asset for cross-disciplinary investigations. We propose an integrated graph-based structure of multiple resources, enriched with the results of the application of graph analytics approaches to provide an encompassing view of the domain and of the way people talk about it during the recommendation task. While we cannot distribute the final resource because of licensing issues, we share the code to assemble and process it once the reference data have been obtained from the original sources.
|
['Sabrina Mennella', 'Maria Di Maro', 'Martina Di Bratto', 'Antonio Origlia']
| null | null | null | null |
lrec-2022-6
|
['movie-recommendation']
|
['miscellaneous']
|
[-2.38182545e-01 4.16000605e-01 -2.66778022e-01 -2.00470120e-01
-1.46618575e-01 -8.14336538e-01 1.04084742e+00 1.05215228e+00
-4.39210594e-01 7.48474240e-01 7.02063918e-01 -3.78877938e-01
-5.56571543e-01 -7.90445745e-01 -6.58216849e-02 -2.00586826e-01
1.11744307e-01 6.74533129e-01 5.06619275e-01 -6.93507016e-01
5.86734951e-01 5.00512004e-01 -1.27749002e+00 4.57383871e-01
8.60479772e-01 6.15791380e-01 8.92340466e-02 2.04779282e-01
-7.67943859e-01 6.75513685e-01 -6.90590501e-01 -1.17905545e+00
-1.62698194e-01 -4.79259372e-01 -1.22003388e+00 7.10406899e-02
-1.52841389e-01 2.21622258e-01 -9.04891565e-02 1.07073271e+00
2.79603273e-01 1.38587654e-01 7.22483635e-01 -8.13245952e-01
-2.51653921e-02 9.76236463e-01 2.10106000e-02 1.78286314e-01
7.40635872e-01 -3.88515174e-01 1.13013184e+00 -5.27290940e-01
1.14445317e+00 1.06796801e+00 3.07462215e-01 6.36030361e-02
-9.21153009e-01 3.24117303e-01 8.60591233e-02 1.25542983e-01
-7.65196443e-01 -1.94021419e-01 6.58198118e-01 -6.57908022e-01
6.37764692e-01 3.61138731e-01 7.86166251e-01 1.09373641e+00
-3.42850834e-01 2.51201004e-01 8.03231776e-01 -6.93968773e-01
2.13637296e-02 6.26197100e-01 5.81446588e-01 4.94757801e-01
5.88311076e-01 -6.96083844e-01 -5.27489126e-01 -4.14764494e-01
3.89503151e-01 -3.68745565e-01 -2.94687539e-01 -6.45751476e-01
-9.68553424e-01 9.38924074e-01 2.33982116e-01 9.30530906e-01
-5.04968047e-01 -5.26213765e-01 7.02156603e-01 3.80662888e-01
4.77242142e-01 4.56539571e-01 -2.88565576e-01 -4.94250655e-01
-6.27976000e-01 4.05658752e-01 1.50520730e+00 6.85756862e-01
4.47891146e-01 -6.21277690e-01 1.07421979e-01 8.99790883e-01
4.43241864e-01 -1.28243864e-01 2.46577382e-01 -6.77658498e-01
5.60661435e-01 1.23745012e+00 3.43963921e-01 -1.18549645e+00
-3.25111598e-01 -2.77610987e-01 -3.81367236e-01 -1.74467757e-01
1.00060034e+00 -2.23917395e-01 2.40651295e-01 1.10538709e+00
6.83009923e-01 -6.70259953e-01 8.67644772e-02 8.41138005e-01
1.21321428e+00 3.11159968e-01 -7.49802887e-02 -1.74944565e-01
1.67048085e+00 -3.99555475e-01 -9.27355111e-01 1.55335605e-01
8.11267078e-01 -8.45062256e-01 7.39418924e-01 5.34536242e-01
-9.37093198e-01 -1.31840050e-01 -9.99125123e-01 -1.68161392e-01
-9.61205721e-01 1.09548628e-01 5.48340023e-01 6.94879293e-01
-6.23496175e-01 6.79672241e-01 -3.43503386e-01 -9.13868964e-01
1.19089589e-01 -2.59697944e-01 -4.04158205e-01 1.64319813e-01
-1.35914719e+00 1.20236456e+00 3.25724125e-01 8.05642232e-02
9.67437923e-02 -2.77931631e-01 -4.84267145e-01 -1.23589560e-01
7.01217055e-01 -4.35707659e-01 9.04253483e-01 -7.43118763e-01
-1.30813158e+00 9.83842254e-01 3.03133070e-01 -3.88111472e-01
7.51356244e-01 -1.06702343e-01 -3.64443630e-01 4.08794805e-02
-2.43278015e-02 -4.09731597e-01 3.98017168e-01 -8.54026854e-01
-5.11039317e-01 -4.79360342e-01 5.38021326e-01 1.90891698e-01
-2.83829391e-01 5.40714622e-01 -5.19587040e-01 -4.67832774e-01
-2.46326134e-01 -6.05746627e-01 8.32358152e-02 -4.26144481e-01
-3.34155440e-01 -2.81316519e-01 4.02167171e-01 -8.20488334e-01
1.59268022e+00 -1.85104203e+00 5.67126870e-01 3.47796321e-01
4.74814087e-01 3.77787799e-01 5.76318800e-01 1.33708787e+00
3.32511932e-01 3.60695451e-01 -1.29480094e-01 -3.62388268e-02
2.96698958e-01 -5.64004760e-03 -1.12399712e-01 4.99558151e-01
-3.22115630e-01 6.49211645e-01 -8.22829366e-01 -4.56853062e-01
2.07348555e-01 4.69989270e-01 -1.44131646e-01 7.64022255e-03
-4.69598770e-01 3.77394944e-01 -8.83827448e-01 -5.31925373e-02
9.16757435e-02 -2.07285136e-01 7.06841409e-01 -2.32939020e-01
-1.52891010e-01 6.16351604e-01 -1.30795264e+00 1.63254774e+00
-4.31447566e-01 5.85661650e-01 1.78585157e-01 -7.90002406e-01
8.36742878e-01 4.22984004e-01 2.73601711e-01 -4.46369529e-01
2.24687085e-01 2.37665087e-01 1.36040568e-01 -5.89207828e-01
7.20683694e-01 2.42327511e-01 -1.03664123e-01 7.35941052e-01
5.92619032e-02 -2.12817136e-02 6.77482665e-01 5.91180623e-01
8.07939529e-01 1.47677451e-01 6.34733021e-01 -3.66924793e-01
8.65591884e-01 3.20462346e-01 1.06782615e-01 3.27835470e-01
2.88344681e-01 3.82054858e-02 1.13012254e+00 -4.52926815e-01
-1.06625986e+00 -4.65294868e-01 -2.27697313e-01 9.80684102e-01
-1.90230265e-01 -8.79697263e-01 -7.88783073e-01 -7.06160843e-01
-2.12896407e-01 7.17615128e-01 -5.18618584e-01 4.63738620e-01
-2.85490900e-01 -2.43647203e-01 3.36653948e-01 -2.06850111e-01
2.55104005e-01 -1.05813205e+00 -7.65528977e-01 1.57950923e-01
-2.39601016e-01 -1.02685738e+00 7.94784278e-02 -1.93622380e-01
-5.42763650e-01 -1.49810112e+00 -4.68697965e-01 -1.40416831e-01
4.06599522e-01 -1.46844104e-01 1.47994268e+00 5.18933952e-01
-1.23009034e-01 5.96232176e-01 -9.34643090e-01 -3.75202686e-01
-8.99882317e-01 2.83614129e-01 -2.98191577e-01 1.28937289e-01
1.90564618e-01 -3.76574904e-01 -8.70554373e-02 1.06879048e-01
-1.09197974e+00 -2.45372653e-01 -6.09912798e-02 3.77653867e-01
-1.44370019e-01 -1.52759105e-01 4.70343828e-01 -1.50385821e+00
1.35131955e+00 -6.83749437e-01 -5.71108818e-01 4.87936467e-01
-3.94930810e-01 -1.47756431e-02 3.45374197e-01 6.50825277e-02
-9.40642953e-01 -7.75824070e-01 6.40224814e-02 3.05851161e-01
-4.05944347e-01 1.02912831e+00 -2.65193015e-01 1.57515123e-01
5.87684751e-01 -3.80268663e-01 1.37957513e-01 -7.61659980e-01
7.10121989e-01 7.52218425e-01 -9.98947993e-02 -6.83709085e-01
3.35963696e-01 6.20545931e-02 -1.03048310e-01 -1.04446161e+00
-7.36105323e-01 -5.57184219e-01 -8.35150599e-01 -4.99696583e-01
8.70129883e-01 -5.15471458e-01 -8.37375402e-01 -4.24552560e-02
-1.27496970e+00 2.27487602e-04 -3.50626886e-01 3.07279080e-01
-2.31617585e-01 5.76924086e-01 -3.20864171e-01 -9.01290536e-01
-1.74127758e-01 -9.35541928e-01 3.54884863e-01 5.89304268e-02
-6.96373940e-01 -1.51157641e+00 3.63054499e-02 4.62882698e-01
4.44946140e-01 3.75318408e-01 1.14515948e+00 -1.21686852e+00
-6.85344413e-02 -3.46605539e-01 8.09589252e-02 -9.43297893e-02
1.49310917e-01 2.06046507e-01 -5.87929666e-01 9.51265693e-02
-5.58496304e-02 -1.16433188e-01 2.27543712e-01 -1.70577034e-01
4.39474076e-01 -3.50291342e-01 -1.50022462e-01 -4.08956826e-01
1.12436140e+00 -1.93505526e-01 4.95205432e-01 6.25036716e-01
4.87266064e-01 1.41039777e+00 4.82529163e-01 6.68839753e-01
4.20040637e-01 1.14348650e+00 3.03928494e-01 2.72739828e-01
4.64457981e-02 -1.94577109e-02 1.91778794e-01 6.50432587e-01
-3.14015865e-01 -3.13925147e-01 -9.24954236e-01 3.30837935e-01
-2.05285621e+00 -8.63663495e-01 -6.46945477e-01 2.32156658e+00
5.64028323e-01 1.88025475e-01 6.58309817e-01 7.91312605e-02
7.53646255e-01 2.23431647e-01 8.74367431e-02 -6.04388773e-01
6.70683617e-03 -7.66427442e-02 1.18322410e-01 6.04598582e-01
-5.26462793e-01 7.15252638e-01 5.60350084e+00 7.82534540e-01
-5.85725665e-01 -2.43921187e-02 9.19108912e-02 2.72414356e-01
-3.69359583e-01 2.59187132e-01 -4.64905530e-01 2.32198581e-01
1.03230560e+00 -4.26040560e-01 3.98041636e-01 4.71537441e-01
4.05192494e-01 -3.31656843e-01 -8.89830589e-01 5.98710477e-01
-9.96142775e-02 -1.45727468e+00 -3.65866944e-02 3.92476201e-01
2.77331448e-03 -1.12141460e-01 -5.20879924e-01 -2.48216882e-01
4.43815678e-01 -6.21189833e-01 7.16699660e-01 6.04602695e-01
3.18206310e-01 -4.65077072e-01 8.00862551e-01 5.25872409e-01
-8.27637732e-01 1.74916238e-01 -1.72328025e-01 -1.83970258e-02
3.86746854e-01 8.12089622e-01 -7.05167949e-01 1.29460692e+00
2.98254132e-01 6.60372794e-01 -6.43340468e-01 8.54743361e-01
-4.00352716e-01 5.35140872e-01 -9.53026637e-02 -4.88736540e-01
1.73098464e-02 -8.47428203e-01 8.75353873e-01 1.10547876e+00
8.59069079e-02 -9.78173763e-02 6.14813082e-02 8.20652843e-01
1.23502932e-01 8.04608822e-01 -6.97692275e-01 -4.14075911e-01
2.52219737e-01 1.63222432e+00 -8.61734688e-01 -1.88842878e-01
-5.73318303e-01 5.18700898e-01 4.38321501e-01 9.46597308e-02
-3.84475291e-01 -2.58428961e-01 3.09015542e-01 3.63871485e-01
2.11670637e-01 -3.52145821e-01 9.30763483e-02 -1.04984677e+00
-3.48005518e-02 -9.31304097e-01 5.02264917e-01 -6.58861995e-01
-1.29012084e+00 8.30027699e-01 2.49765813e-01 -9.84240234e-01
-4.29055721e-01 -4.26968306e-01 -3.89102310e-01 1.00311828e+00
-8.76085937e-01 -1.01471770e+00 -5.02281077e-02 4.80082035e-01
8.26323032e-02 -2.40954787e-01 9.27505493e-01 1.73869729e-01
-5.29509664e-01 -1.24378666e-01 6.31650686e-02 6.36338070e-02
5.93396544e-01 -1.20327485e+00 1.60287008e-01 6.70470893e-01
7.03119755e-01 9.36236441e-01 8.65113616e-01 -7.34616518e-01
-1.27835548e+00 -2.07841963e-01 1.21306384e+00 -6.86890662e-01
1.18483281e+00 -2.70682216e-01 -9.39807415e-01 3.85640919e-01
5.69914639e-01 -4.74636197e-01 8.60599577e-01 3.97934258e-01
-4.33375388e-02 2.40472585e-01 -9.50013936e-01 3.56675208e-01
8.67600918e-01 -4.66739476e-01 -1.06347215e+00 3.06923628e-01
2.07163930e-01 -2.57597744e-01 -1.30918694e+00 -2.60252982e-01
1.92490697e-01 -1.26286268e+00 4.22939360e-01 -8.11677933e-01
2.55271494e-01 -1.92669749e-01 -3.76389883e-02 -1.20253432e+00
8.22760090e-02 -7.01056421e-01 1.86561830e-02 1.69258690e+00
5.66399276e-01 -8.34407270e-01 5.27769208e-01 7.17068493e-01
9.23143327e-02 -4.83692914e-01 -7.45886862e-01 -7.76624233e-02
-1.63566709e-01 -4.60631669e-01 4.14232135e-01 9.09171820e-01
5.55552542e-01 5.65188825e-01 -3.49086449e-02 -3.46355230e-01
2.91288286e-01 2.06093192e-01 9.06801105e-01 -1.71467149e+00
-8.09757411e-02 -5.67937672e-01 -1.78849757e-01 -6.74895287e-01
5.85700683e-02 -9.94093895e-01 -7.01253116e-01 -2.05234170e+00
-9.69684646e-02 -3.65665406e-01 2.55064577e-01 2.76411907e-03
1.86317638e-01 -1.86222732e-01 2.64220804e-01 4.70996588e-01
-5.17480254e-01 1.68275982e-01 8.91807973e-01 4.08379227e-01
-1.93070740e-01 1.02047153e-01 -8.11630547e-01 7.53078043e-01
3.96360964e-01 -3.70570958e-01 -3.03388506e-01 1.23261608e-01
8.12095225e-01 3.37784626e-02 1.52226180e-01 -4.85347003e-01
1.49609491e-01 -8.04753155e-02 -2.23963961e-01 -2.70286053e-01
1.30391985e-01 -1.01349628e+00 4.89099145e-01 2.75282919e-01
-4.89991337e-01 -5.41234165e-02 -1.74091190e-01 5.09152830e-01
-3.50772232e-01 -6.76669002e-01 2.68476069e-01 -3.53846997e-01
-5.52954793e-01 -5.59983589e-02 -4.29583728e-01 2.63479799e-01
8.79712105e-01 -9.02580842e-02 -3.30493510e-01 -5.81126869e-01
-9.41696167e-01 -2.32670959e-02 6.23452187e-01 3.75445724e-01
1.02159373e-01 -7.63050079e-01 -6.86805487e-01 -4.01794732e-01
1.28885061e-01 -3.51277739e-01 3.42630898e-03 7.85709620e-01
-6.52732551e-01 4.34025407e-01 -2.13608786e-01 5.98383360e-02
-1.21299469e+00 4.90598172e-01 1.88615769e-01 -6.36765003e-01
-6.41610563e-01 1.44126207e-01 -4.05784190e-01 -2.21277744e-01
4.25647199e-02 2.61608083e-02 -1.08731568e+00 7.93119550e-01
4.21934038e-01 5.53684950e-01 2.56326497e-01 -8.62667322e-01
-2.06284180e-01 2.34940991e-01 1.89406797e-01 -3.61206800e-01
1.62078440e+00 -5.20957410e-01 -5.01265347e-01 8.89298081e-01
5.54571390e-01 5.42350233e-01 -4.71280366e-01 -2.59337276e-01
5.39859951e-01 -2.09082752e-01 -2.39542261e-01 -7.05668032e-01
-7.22330153e-01 6.82981730e-01 -1.11573003e-01 1.23221505e+00
5.48169076e-01 2.15501726e-01 -1.14739668e-02 2.75962591e-01
3.93403083e-01 -1.03081787e+00 -3.89218926e-01 4.93513584e-01
1.09641182e+00 -7.80178845e-01 5.22402287e-01 -9.11476851e-01
-9.38650548e-01 1.43876314e+00 -1.07372165e-01 2.93206275e-01
7.49071956e-01 -1.55759469e-01 -1.18509680e-01 -6.94722593e-01
-4.44006026e-01 -3.83553743e-01 3.96785200e-01 3.76137823e-01
7.76844978e-01 -1.93862304e-01 -1.16735911e+00 7.80068278e-01
-1.51378393e-01 -7.96091333e-02 6.53339148e-01 6.64383411e-01
-1.97129976e-02 -1.75387502e+00 -1.36872724e-01 4.05277073e-01
-8.53066742e-01 2.96300929e-02 -1.03180468e+00 8.69230926e-01
-2.06926376e-01 1.09154880e+00 -3.81606221e-01 -4.09220345e-02
6.37329161e-01 1.22639887e-01 4.33541626e-01 -7.27323890e-01
-1.25625598e+00 -1.35808945e-01 1.01735699e+00 -1.54197007e-01
-9.84837592e-01 -5.62473178e-01 -1.01182902e+00 -3.08253020e-01
-1.61583707e-01 7.01843023e-01 9.08410490e-01 1.14599037e+00
3.14547539e-01 3.20131779e-01 7.75671452e-02 -6.46298528e-01
-1.64127246e-01 -9.91113424e-01 -7.90503383e-01 6.15603924e-01
-3.70323509e-01 -7.44959414e-01 -8.69803056e-02 -1.05321020e-01]
|
[9.647112846374512, 9.417040824890137]
|
b480780d-2352-4581-822e-b51428bc0362
|
1st-place-solution-for-eccv-2022-ood-cv
|
2301.04796
| null |
https://arxiv.org/abs/2301.04796v1
|
https://arxiv.org/pdf/2301.04796v1.pdf
|
1st Place Solution for ECCV 2022 OOD-CV Challenge Object Detection Track
|
OOD-CV challenge is an out-of-distribution generalization task. To solve this problem in object detection track, we propose a simple yet effective Generalize-then-Adapt (G&A) framework, which is composed of a two-stage domain generalization part and a one-stage domain adaptation part. The domain generalization part is implemented by a Supervised Model Pretraining stage using source data for model warm-up and a Weakly Semi-Supervised Model Pretraining stage using both source data with box-level label and auxiliary data (ImageNet-1K) with image-level label for performance boosting. The domain adaptation part is implemented as a Source-Free Domain Adaptation paradigm, which only uses the pre-trained model and the unlabeled target data to further optimize in a self-supervised training manner. The proposed G&A framework help us achieve the first place on the object detection leaderboard of the OOD-CV challenge. Code will be released in https://github.com/hikvision-research/OOD-CV.
|
['Yueting Zhuang', 'ShiLiang Pu', 'Di Xie', 'Shicai Yang', 'WeiJie Chen', 'Binbin Chen', 'Wei Zhao']
|
2023-01-12
| null | null | null | null |
['source-free-domain-adaptation']
|
['computer-vision']
|
[ 8.11449364e-02 1.26003191e-01 -4.14435565e-01 -7.34881520e-01
-8.26044679e-01 -5.23822129e-01 7.02246606e-01 5.89090995e-02
-3.79918754e-01 3.10992897e-01 -2.34197155e-01 -2.61348695e-01
7.50787914e-01 -5.61843872e-01 -7.07729220e-01 -5.87196469e-01
1.31425887e-01 8.21705759e-01 5.52440524e-01 -3.29329111e-02
-2.46206403e-01 1.79947391e-01 -1.41524613e+00 5.46840549e-01
7.05424070e-01 1.32943714e+00 4.77472186e-01 7.54384756e-01
-1.41934693e-01 5.29704809e-01 -3.47035199e-01 -1.58410147e-01
6.01052701e-01 -2.85396308e-01 -7.21508145e-01 1.79278612e-01
4.43902284e-01 -4.84953940e-01 -8.71917307e-02 1.06119585e+00
6.55897975e-01 1.21507913e-01 8.61766279e-01 -1.29812777e+00
-7.50826061e-01 9.79634076e-02 -8.18676770e-01 3.48465621e-01
-2.55972385e-01 5.31473637e-01 6.70232713e-01 -1.13632953e+00
6.65257692e-01 1.36998880e+00 4.76874471e-01 9.46369588e-01
-1.28395069e+00 -8.87948871e-01 3.36295605e-01 8.02363381e-02
-1.20729876e+00 -1.27283081e-01 7.56598592e-01 -6.46189809e-01
8.48818600e-01 -2.85527974e-01 5.00234604e-01 1.28065765e+00
-2.08643690e-01 1.17931330e+00 1.07416904e+00 -3.87677819e-01
4.33299989e-01 6.03506207e-01 5.18813133e-01 4.66601640e-01
6.56643435e-02 3.64344329e-01 -1.74049646e-01 -1.54942885e-01
4.75572109e-01 -6.86675608e-02 1.47057384e-01 -8.02068770e-01
-6.83178902e-01 7.70159721e-01 8.99576187e-01 6.32800208e-03
-3.29117060e-01 -2.92058617e-01 5.69215357e-01 3.31751794e-01
8.64938557e-01 8.20616931e-02 -5.84620416e-01 4.63402301e-01
-9.27748084e-01 4.52038199e-01 5.59759080e-01 1.21873105e+00
9.95816648e-01 -1.86597168e-01 -3.74927908e-01 1.05103207e+00
5.40139675e-01 5.33313632e-01 5.50950110e-01 -5.28564692e-01
4.12848294e-01 7.34933615e-01 7.76013508e-02 -1.78456515e-01
-3.21989030e-01 -8.12426150e-01 -6.58735693e-01 5.08608103e-01
2.31687695e-01 -1.64072677e-01 -1.39462423e+00 1.78458869e+00
8.98407578e-01 4.17166233e-01 2.96705193e-03 9.83656704e-01
1.06897855e+00 6.56437218e-01 5.98397255e-01 2.47845188e-01
1.50020254e+00 -1.49365115e+00 -2.82167375e-01 -5.72216511e-01
8.23476255e-01 -5.69280624e-01 1.22569132e+00 2.01946929e-01
-7.92345345e-01 -1.18510342e+00 -1.17407012e+00 -3.32953513e-01
-6.03735685e-01 4.94552374e-01 2.32670829e-01 3.98399502e-01
-9.19522583e-01 -5.96401356e-02 -7.09102750e-01 -5.14836133e-01
8.66567552e-01 -5.22372387e-02 -3.30819577e-01 -4.93227959e-01
-9.37895298e-01 9.95043874e-01 7.49696016e-01 -1.65233120e-01
-1.59235036e+00 -9.16589618e-01 -8.71703088e-01 -3.07429105e-01
3.16747636e-01 -7.56512642e-01 1.56631076e+00 -1.08220243e+00
-1.35445058e+00 1.61841691e+00 2.68176533e-02 -5.69430530e-01
6.41342044e-01 -3.27343196e-01 -2.62238503e-01 -2.13613316e-01
2.89021641e-01 1.05697763e+00 1.16585708e+00 -1.37780964e+00
-9.12839293e-01 -5.00124872e-01 -2.23531485e-01 2.48563036e-01
-1.72930602e-02 -1.60485744e-01 -7.01057732e-01 -5.75488389e-01
-3.46433043e-01 -9.56716120e-01 -1.52759090e-01 1.26368016e-01
-3.34043413e-01 -6.64162338e-01 1.01698339e+00 -6.98756993e-01
9.63118374e-01 -2.34048200e+00 -1.61016673e-01 -5.76507486e-02
2.09892333e-01 6.85248494e-01 -4.92220223e-01 -7.40692914e-02
-3.27709734e-01 -3.54280442e-01 -3.54792446e-01 -7.07934499e-01
-7.70171136e-02 -1.36132553e-01 -2.44306266e-01 3.35535824e-01
7.11597979e-01 8.35996568e-01 -1.04890275e+00 -3.82202178e-01
1.49540693e-01 2.32053727e-01 -4.01312649e-01 7.45968103e-01
-6.09711409e-01 4.34767276e-01 -4.87673670e-01 6.96562409e-01
1.13918364e+00 -2.75501609e-01 -1.29485130e-01 -1.17318213e-01
-5.54326065e-02 2.93725848e-01 -1.06969821e+00 1.85341191e+00
-2.29535967e-01 2.73421347e-01 2.70400465e-01 -1.26194489e+00
1.05560291e+00 -4.25686091e-02 2.30337754e-02 -6.03658497e-01
1.81084499e-01 5.53730540e-02 -1.50821045e-01 -3.59985709e-01
1.69979423e-01 -9.32494104e-02 5.58573827e-02 2.29100630e-01
6.08037353e-01 -1.76081941e-01 1.17530003e-01 3.58412564e-01
1.00173223e+00 5.84386468e-01 3.02108049e-01 -3.08778614e-01
5.02459705e-01 3.46445531e-01 6.83474243e-01 8.53524745e-01
-5.28244495e-01 6.02816582e-01 3.03322256e-01 -4.52405989e-01
-1.08797574e+00 -1.14918447e+00 -3.01341772e-01 1.45795321e+00
1.40659362e-01 -2.25437745e-01 -8.82856131e-01 -1.21971107e+00
2.40674496e-01 7.46853709e-01 -9.02892232e-01 -3.99067461e-01
-2.97395229e-01 -6.80416048e-01 2.19050780e-01 7.13769078e-01
7.06344306e-01 -9.92768884e-01 -1.11409269e-01 1.32345706e-01
1.99373260e-01 -1.01166582e+00 -3.74475271e-01 6.64032102e-01
-8.06117833e-01 -8.77124846e-01 -1.03045428e+00 -9.51755583e-01
6.79916322e-01 2.24839985e-01 1.20602250e+00 -1.81044862e-01
-4.04929638e-01 2.14622706e-01 -3.52149189e-01 -8.61008465e-01
-5.83933055e-01 2.28806466e-01 -8.63695741e-02 -4.05894257e-02
7.85023928e-01 -2.22005904e-01 -6.99083388e-01 4.04393494e-01
-6.96271360e-01 1.15022928e-01 8.63872290e-01 8.71466637e-01
8.32942188e-01 -4.30731595e-01 6.20421231e-01 -9.82665598e-01
3.18922192e-01 -7.30398417e-01 -7.86849976e-01 1.92966983e-01
-5.09987831e-01 -1.87172651e-01 2.95510530e-01 -6.60370469e-01
-1.34121776e+00 4.20175999e-01 -2.31984183e-01 -6.83075070e-01
-4.24565047e-01 -1.45086106e-02 -4.12611425e-01 1.86571196e-01
1.06842220e+00 2.66958982e-01 2.51759570e-02 -8.79760385e-01
6.10079050e-01 8.87799561e-01 7.22660184e-01 -4.64466840e-01
1.05943632e+00 3.04544747e-01 -5.02303421e-01 -6.76464558e-01
-1.12006199e+00 -8.48254204e-01 -8.07999671e-01 -1.15374289e-01
1.02561319e+00 -1.46341109e+00 3.74909081e-02 7.56567538e-01
-9.85008299e-01 -8.82643104e-01 -4.62769985e-01 3.93353164e-01
-2.45657906e-01 9.95407626e-02 -4.82619673e-01 -5.82192183e-01
-3.32934350e-01 -9.21694458e-01 1.13417947e+00 4.69911128e-01
3.05794626e-01 -6.97446585e-01 3.48263204e-01 4.00398076e-01
2.82143205e-01 -3.37913074e-02 5.08395195e-01 -1.13710940e+00
-3.67220610e-01 -2.84269571e-01 -5.23235261e-01 1.04762161e+00
-1.38539121e-01 -4.19209063e-01 -1.20433307e+00 -4.07797426e-01
-1.74991041e-01 -9.12655115e-01 1.17463601e+00 3.79515737e-01
1.11621606e+00 1.47466362e-01 -7.43751168e-01 7.09064007e-01
1.10204124e+00 -2.05337986e-01 3.00055593e-01 3.09365064e-01
5.94876111e-01 4.83085424e-01 9.02638257e-01 1.39422491e-01
5.90924740e-01 6.61478996e-01 3.86018991e-01 -4.38984513e-01
-6.14168108e-01 -3.50883782e-01 1.78947344e-01 1.90421164e-01
2.71709293e-01 -8.98281187e-02 -1.04240179e+00 6.15445554e-01
-1.79582965e+00 -6.49663627e-01 -1.09265424e-01 2.10575199e+00
9.04353142e-01 3.34119201e-01 5.92942178e-01 -5.03507674e-01
9.02243078e-01 -1.44726530e-01 -8.98979425e-01 -1.31872982e-01
1.24800786e-01 4.75981273e-02 2.45429665e-01 3.70303720e-01
-1.50071144e+00 1.18713367e+00 5.12675476e+00 1.06665540e+00
-9.99340653e-01 4.56994802e-01 7.01484263e-01 9.86631438e-02
2.12599829e-01 -1.09966606e-01 -1.26609933e+00 5.41808963e-01
6.60814941e-01 1.80214271e-01 -6.83350936e-02 1.65472746e+00
-1.50542289e-01 -7.91845694e-02 -1.11304271e+00 6.98066950e-01
-6.77555799e-02 -1.03288651e+00 -4.50861119e-02 -1.26822531e-01
7.07457364e-01 5.54401040e-01 -2.75923498e-02 1.08080518e+00
3.59672248e-01 -4.46532071e-01 6.05169415e-01 1.14662386e-01
6.66126370e-01 -3.05077016e-01 5.69608212e-01 6.79465413e-01
-1.23501611e+00 -1.10585064e-01 -6.63290739e-01 1.74013704e-01
-4.31500785e-02 8.15764904e-01 -1.05299616e+00 4.21653897e-01
1.06481409e+00 6.69572830e-01 -5.92409492e-01 1.16211450e+00
-4.24011737e-01 8.59340012e-01 -4.29776132e-01 2.56726056e-01
2.50813514e-01 1.03817366e-01 4.92078036e-01 1.58203387e+00
-1.39946833e-01 2.66633295e-02 4.23617214e-01 8.85550857e-01
-2.45838597e-01 -1.14276923e-01 -4.05728102e-01 3.38186949e-01
4.08530325e-01 1.45213485e+00 -4.85227972e-01 -7.56060600e-01
-4.62612838e-01 1.04097867e+00 5.28782904e-01 3.34328473e-01
-7.34588563e-01 -1.30349204e-01 5.55336773e-01 1.52688995e-01
4.19744343e-01 8.48818868e-02 -1.28258198e-01 -1.15274370e+00
-2.02941656e-01 -8.47236395e-01 6.30301952e-01 -7.82783270e-01
-1.62618017e+00 5.84988713e-01 1.55364603e-01 -1.31609619e+00
-6.80768713e-02 -7.42617726e-01 -7.43435323e-01 1.02121985e+00
-1.80834615e+00 -1.60929680e+00 -6.20350361e-01 6.46181166e-01
6.35495186e-01 -3.67033780e-01 6.11176550e-01 2.38109335e-01
-6.16282880e-01 6.91478252e-01 9.74545851e-02 2.57892162e-01
1.07795012e+00 -1.42148829e+00 6.23475254e-01 6.25180185e-01
-2.94833601e-01 2.26095289e-01 4.91120905e-01 -7.30983853e-01
-7.22393870e-01 -1.48716140e+00 5.95490754e-01 -5.45555532e-01
4.02994305e-01 -8.98628891e-01 -1.12849760e+00 8.17346334e-01
1.18662111e-01 5.48182964e-01 5.56492209e-01 1.04926616e-01
-5.18249989e-01 -2.53005564e-01 -1.25417924e+00 5.21447025e-02
9.78013456e-01 -3.21858048e-01 -9.48517203e-01 4.83458191e-01
7.56949425e-01 -5.32154739e-01 -5.35511494e-01 4.98492151e-01
2.32172161e-01 -6.17872596e-01 1.00212991e+00 -6.13597572e-01
2.16765627e-01 -5.00877321e-01 1.61161106e-02 -1.20158410e+00
-3.92054439e-01 -1.79859444e-01 -1.74500972e-01 1.42943823e+00
3.84141088e-01 -4.03387159e-01 7.68774986e-01 1.30343154e-01
-3.26038361e-01 -6.42975152e-01 -7.31304705e-01 -9.63888705e-01
2.49918178e-01 -4.30944413e-01 2.94236153e-01 8.97076607e-01
-4.65960205e-01 6.30501449e-01 -1.86331853e-01 2.28740677e-01
1.04430413e+00 -3.53609696e-02 1.10091925e+00 -1.11912036e+00
-3.79934609e-01 -1.91537112e-01 -3.00346971e-01 -1.43624282e+00
7.19516678e-03 -1.05549765e+00 2.55424201e-01 -1.17576969e+00
4.66692805e-01 -4.91901636e-01 -4.23386663e-01 5.80799520e-01
-4.16568518e-01 2.14631453e-01 4.87500317e-02 3.85259748e-01
-7.86255479e-01 6.72950506e-01 1.13040709e+00 -2.50786066e-01
-3.17712516e-01 3.29757839e-01 -5.13043106e-01 5.54094136e-01
6.75423026e-01 -7.12875366e-01 -4.71635252e-01 -2.49990031e-01
-6.02112114e-01 -4.89473850e-01 5.67649961e-01 -1.09288251e+00
1.15479417e-02 1.98357895e-01 6.61856055e-01 -7.70916879e-01
1.97539002e-01 -5.32825768e-01 -4.20401454e-01 5.11041820e-01
-3.30412835e-01 -4.11140651e-01 3.44890177e-01 7.01677382e-01
-1.12805657e-01 -9.45751145e-02 1.22209942e+00 -8.67246836e-02
-1.08983195e+00 5.31593323e-01 -1.40251182e-02 9.15152133e-02
1.26407504e+00 -4.11512936e-03 -4.74309325e-01 1.22433923e-01
-9.77687776e-01 7.25447774e-01 2.70021141e-01 6.43509924e-01
1.42354771e-01 -1.32149482e+00 -9.74863350e-01 2.93562680e-01
7.28105426e-01 3.44589442e-01 4.31162953e-01 4.40152317e-01
-1.30815044e-01 1.07001990e-01 -1.71232134e-01 -9.74793613e-01
-9.59376276e-01 8.53919625e-01 4.70785260e-01 -5.54629982e-01
-5.88907003e-01 1.26921952e+00 7.52228081e-01 -9.07432616e-01
4.72714722e-01 -2.01636881e-01 3.89630757e-02 -1.40703619e-01
7.81514883e-01 1.08120866e-01 1.22406352e-02 -2.73620725e-01
-5.61690688e-01 2.26347521e-01 -4.41914737e-01 5.99888638e-02
1.18659401e+00 1.94115788e-02 4.41822261e-01 2.00584888e-01
1.18232453e+00 -7.15969145e-01 -1.67405319e+00 -6.04278684e-01
9.53525957e-03 -7.07069710e-02 1.45326227e-01 -1.29712737e+00
-6.56837761e-01 1.08332217e+00 1.14161193e+00 -1.29626870e-01
1.17739081e+00 3.65514964e-01 5.44780433e-01 2.50929713e-01
5.30154817e-02 -1.27086318e+00 2.92860478e-01 4.82342601e-01
9.32028949e-01 -1.66648912e+00 -2.18194634e-01 -1.51045874e-01
-7.73059428e-01 5.93027771e-01 1.22355294e+00 -3.82142693e-01
9.35779989e-01 1.48450464e-01 2.90341407e-01 9.71395746e-02
-7.67970741e-01 -5.42349756e-01 4.90900069e-01 1.11832225e+00
2.14089051e-01 -8.40892047e-02 -4.91665006e-02 9.21330750e-01
3.28661054e-01 2.05169588e-01 -3.47452283e-01 8.23067904e-01
-6.47865057e-01 -1.16985822e+00 -1.58312753e-01 3.57927531e-01
8.10164139e-02 5.71005829e-02 -3.46603453e-01 8.86682510e-01
4.47439611e-01 6.16150618e-01 -5.28988661e-03 -4.09825832e-01
5.21150947e-01 2.42793798e-01 2.48393029e-01 -1.05537939e+00
-5.62011242e-01 2.65500750e-02 -8.75578299e-02 -6.06631577e-01
-1.09482221e-01 -5.18798232e-01 -9.32574570e-01 5.17468415e-02
-2.19320670e-01 5.90320081e-02 7.25527406e-01 8.31544161e-01
4.55476910e-01 3.52383882e-01 4.59952444e-01 -9.92113352e-01
-7.46072173e-01 -1.29428637e+00 -4.86638367e-01 4.29806650e-01
4.20399249e-01 -6.84995472e-01 -1.35420606e-01 1.71584174e-01]
|
[9.40744400024414, 1.4724887609481812]
|
635efa7b-702a-4081-82c5-d7e0cb14a9e5
|
stvgformer-spatio-temporal-video-grounding
|
2207.02756
| null |
https://arxiv.org/abs/2207.02756v1
|
https://arxiv.org/pdf/2207.02756v1.pdf
|
STVGFormer: Spatio-Temporal Video Grounding with Static-Dynamic Cross-Modal Understanding
|
In this technical report, we introduce our solution to human-centric spatio-temporal video grounding task. We propose a concise and effective framework named STVGFormer, which models spatiotemporal visual-linguistic dependencies with a static branch and a dynamic branch. The static branch performs cross-modal understanding in a single frame and learns to localize the target object spatially according to intra-frame visual cues like object appearances. The dynamic branch performs cross-modal understanding across multiple frames. It learns to predict the starting and ending time of the target moment according to dynamic visual cues like motions. Both the static and dynamic branches are designed as cross-modal transformers. We further design a novel static-dynamic interaction block to enable the static and dynamic branches to transfer useful and complementary information from each other, which is shown to be effective to improve the prediction on hard cases. Our proposed method achieved 39.6% vIoU and won the first place in the HC-STVG track of the 4th Person in Context Challenge.
|
['Wei-Shi Zheng', 'Tiancai Ye', 'Zhi Jin', 'Jian-Fang Hu', 'Chaolei Tan', 'Zihang Lin']
|
2022-07-06
| null | null | null | null |
['video-grounding', 'spatio-temporal-video-grounding']
|
['computer-vision', 'computer-vision']
|
[-1.31147400e-01 2.04313453e-02 -3.85475427e-01 -5.37064731e-01
-5.76855719e-01 -4.67187881e-01 5.75342059e-01 -8.62289220e-02
-4.58293110e-01 4.58441556e-01 5.59308827e-01 -6.18004352e-02
1.73365325e-01 -2.78614312e-01 -8.77203345e-01 -4.50333595e-01
-3.89042735e-01 2.76352257e-01 7.69437492e-01 -2.63073504e-01
-4.78688739e-02 8.42772797e-02 -1.51216948e+00 1.08274257e+00
3.01793784e-01 9.26896513e-01 5.50119102e-01 8.57827187e-01
2.40388885e-01 1.07141364e+00 -2.89999396e-01 -1.11552186e-01
9.93339866e-02 -3.74220520e-01 -9.18778956e-01 -6.34149741e-03
6.59456015e-01 -3.16496819e-01 -5.64698339e-01 4.56600726e-01
5.31959236e-01 4.28137332e-01 2.72443652e-01 -1.41190791e+00
-4.58274603e-01 4.21759695e-01 -6.61196232e-01 6.49451435e-01
7.28388846e-01 2.14596644e-01 1.05888128e+00 -6.94699228e-01
9.08048272e-01 1.46513093e+00 3.10827106e-01 7.57888436e-01
-8.19700658e-01 -3.69166791e-01 9.00519431e-01 7.67134488e-01
-1.10774314e+00 -4.82527703e-01 6.64362848e-01 -6.24474764e-01
1.18025827e+00 1.85046047e-01 9.07460809e-01 1.18976009e+00
1.43778801e-01 1.23065865e+00 7.94437110e-01 -2.77046353e-01
1.57765634e-02 -2.08472490e-01 8.40293467e-02 8.54420006e-01
-4.84180361e-01 4.25941885e-01 -1.06786764e+00 1.38271675e-01
6.08513355e-01 -1.75869763e-01 -3.32904130e-01 -3.09192657e-01
-1.51709199e+00 5.45680821e-01 6.67936563e-01 4.76771623e-01
-1.61590651e-01 4.30275738e-01 4.06833887e-01 -2.06731372e-02
1.98814884e-01 -2.01474518e-01 -4.20300305e-01 -2.20754996e-01
-7.94020653e-01 1.83691576e-01 3.35149229e-01 9.50236678e-01
3.04873586e-01 -2.50705838e-01 -5.13839960e-01 4.33632523e-01
3.33586663e-01 3.19239646e-01 9.74252373e-02 -1.04302955e+00
7.31523275e-01 2.72544742e-01 3.68044406e-01 -1.10426724e+00
-5.94228029e-01 -1.51533097e-01 -3.08972031e-01 -1.31911308e-01
5.36510646e-01 3.41135748e-02 -9.72669601e-01 2.04742193e+00
3.93200427e-01 6.58705235e-01 -2.30869010e-01 1.35226309e+00
7.64818609e-01 6.97507501e-01 5.64079821e-01 -2.20529497e-01
1.45396507e+00 -1.23944557e+00 -7.08769083e-01 -3.81052941e-01
5.75285316e-01 -4.30470049e-01 1.05772769e+00 2.60440618e-01
-1.13121057e+00 -8.49877596e-01 -8.80264819e-01 -2.20611364e-01
-1.48831934e-01 2.47408524e-01 5.84213495e-01 1.30239785e-01
-1.34847462e+00 1.60327271e-01 -1.17064238e+00 -4.36656982e-01
2.11972579e-01 2.74741501e-01 -5.53940117e-01 -3.82641070e-02
-1.24635458e+00 6.72547936e-01 5.53437054e-01 1.16713144e-01
-1.03114617e+00 -5.65713584e-01 -1.00020897e+00 -1.90732941e-01
3.14995259e-01 -7.68890083e-01 1.13264477e+00 -1.15060604e+00
-1.10976899e+00 1.00414419e+00 -7.35189199e-01 -6.73793256e-01
5.19765258e-01 -5.32244623e-01 -5.10812938e-01 5.53700030e-01
1.94709867e-01 1.08567047e+00 8.28340650e-01 -1.28212416e+00
-9.91908252e-01 -3.66610348e-01 2.60718405e-01 3.76530588e-01
-1.15522519e-01 -3.34607661e-02 -1.03855324e+00 -7.99017966e-01
1.31546091e-02 -8.01569164e-01 7.17005655e-02 -2.90601850e-02
-2.12774530e-01 -3.88267964e-01 1.12422359e+00 -9.48438346e-01
1.32722378e+00 -2.02497387e+00 5.15965641e-01 -8.35744515e-02
1.31661639e-01 1.21802442e-01 -3.66430655e-02 1.95295289e-01
-2.08016038e-01 -1.46475658e-01 4.00911778e-01 -4.40861344e-01
-1.99914917e-01 1.35391325e-01 -2.30349392e-01 3.26370150e-01
9.46502239e-02 1.12567604e+00 -1.11133432e+00 -6.10226274e-01
4.08198833e-01 6.19470060e-01 -5.17835438e-01 2.46171817e-01
-3.88419867e-01 7.16051161e-01 -3.78660411e-01 5.64718425e-01
1.99371606e-01 -3.76571774e-01 1.56647831e-01 -5.30172288e-01
1.50395473e-02 9.19435471e-02 -9.28093493e-01 1.95310318e+00
-2.47586444e-01 8.29157352e-01 -2.13283971e-01 -8.54629517e-01
4.12806630e-01 4.83474642e-01 4.49152410e-01 -9.86585975e-01
-1.19495600e-01 -3.20904970e-01 -1.99582174e-01 -7.35530138e-01
3.34498584e-01 -6.35929313e-03 1.84994303e-02 1.67616636e-01
2.35698503e-02 7.33489990e-01 2.64959663e-01 4.49826062e-01
8.72132301e-01 5.91830313e-01 2.29311973e-01 -4.77921851e-02
6.05842113e-01 -1.06274523e-01 4.74356592e-01 6.55677080e-01
-4.63710040e-01 6.20745182e-01 3.73220503e-01 -8.09213877e-01
-6.31234288e-01 -1.20149672e+00 5.93199790e-01 1.51745248e+00
6.12590492e-01 -6.35951459e-01 -5.71498752e-01 -7.83859551e-01
-2.82415062e-01 5.78677475e-01 -7.44625449e-01 -1.82443559e-01
-8.58554959e-01 -9.81580764e-02 2.28549719e-01 9.70440805e-01
6.01589382e-01 -1.01618993e+00 -9.94669676e-01 7.68103376e-02
-8.94306302e-01 -1.48989701e+00 -8.39686215e-01 -1.86606541e-01
-4.76878226e-01 -1.04139435e+00 -5.94335496e-01 -8.21505189e-01
4.33849454e-01 3.07424963e-01 1.12552679e+00 1.02418751e-01
-2.44459555e-01 7.14701414e-01 -4.62305754e-01 4.45225760e-02
1.68423712e-01 -2.53689617e-01 3.15446444e-02 3.46468240e-01
3.23279649e-01 -2.54617602e-01 -9.08579230e-01 6.11519575e-01
-3.73788983e-01 4.61043119e-01 1.26289502e-01 7.34569013e-01
6.83906376e-01 -3.18326592e-01 1.86346278e-01 -3.86814326e-01
-1.05269916e-01 -1.88348517e-01 -3.66635591e-01 5.53572834e-01
2.02704087e-01 -5.95564060e-02 2.35878974e-01 -4.85044122e-01
-1.24039698e+00 2.99592972e-01 4.02043052e-02 -6.59884691e-01
-1.69832200e-01 3.49426568e-01 -2.09395230e-01 1.78210646e-01
4.99234647e-01 8.83693770e-02 -5.30499339e-01 -2.67926902e-01
4.58595425e-01 2.09286407e-01 8.89508426e-01 -6.19964600e-01
4.21678901e-01 7.71263242e-01 -1.59684137e-01 -6.72666728e-01
-1.01243412e+00 -7.74150968e-01 -9.07775283e-01 -7.49550223e-01
1.44810212e+00 -1.16556573e+00 -1.05075955e+00 2.33735740e-01
-1.33323872e+00 -5.67199051e-01 7.45041370e-02 4.52120185e-01
-8.12590957e-01 2.13209406e-01 -3.99361372e-01 -6.99321866e-01
7.12023750e-02 -9.89111841e-01 1.41339099e+00 1.91820517e-01
-2.42013678e-01 -1.09673023e+00 -1.99569210e-01 6.00020945e-01
-9.76438075e-02 4.58690822e-01 6.07740402e-01 -3.42279822e-01
-8.03660870e-01 8.91546458e-02 -9.76722538e-02 -1.21618621e-01
-6.17361292e-02 -2.75555968e-01 -7.76036441e-01 -5.01949668e-01
-3.13507617e-01 -1.40727609e-01 9.52220619e-01 6.54974878e-01
9.65454280e-01 9.27880853e-02 -5.86279154e-01 5.84507287e-01
1.03008413e+00 3.08623284e-01 5.96136332e-01 1.27296373e-01
8.05171728e-01 5.90476215e-01 1.05093622e+00 2.40713105e-01
6.58872187e-01 1.21758962e+00 3.52474034e-01 -7.22765848e-02
-2.71539330e-01 -4.09097195e-01 7.13335097e-01 4.40526813e-01
-5.20873070e-01 -3.40755075e-01 -1.02736676e+00 6.45425737e-01
-2.26115394e+00 -1.38532603e+00 2.54047532e-02 1.98832810e+00
3.57292622e-01 2.54211813e-01 4.53678966e-01 -2.56389856e-01
7.23787367e-01 2.86775440e-01 -3.96826446e-01 1.20321311e-01
-1.16967149e-01 -2.49178112e-01 1.19298548e-01 6.46316826e-01
-1.38422596e+00 1.20110953e+00 5.63521481e+00 5.99957168e-01
-1.03244257e+00 2.92760611e-01 5.47848046e-01 -3.42280030e-01
7.22293407e-02 -5.14023975e-02 -8.51887345e-01 3.03739130e-01
7.71385431e-01 5.18511236e-02 1.35815546e-01 5.36594033e-01
5.04873931e-01 -3.95416439e-01 -1.37577426e+00 1.17915785e+00
2.15127617e-01 -1.40317225e+00 -5.69698103e-02 -2.26623088e-01
5.00435472e-01 -1.17779039e-01 -8.56512263e-02 8.15841034e-02
-2.48041391e-01 -1.02521276e+00 1.01355338e+00 7.88725734e-01
6.27639115e-01 -6.80677593e-01 4.26888287e-01 3.97311330e-01
-1.83306921e+00 -1.28854498e-01 1.31431624e-01 -2.92830560e-02
6.69689476e-01 -4.57534827e-02 -3.82907838e-01 5.90519369e-01
9.07317817e-01 1.15474737e+00 -4.34441775e-01 8.00455093e-01
-1.83322385e-01 4.88680571e-01 -1.46573693e-01 4.46390867e-01
1.89905658e-01 1.75477013e-01 6.71850204e-01 1.25715339e+00
-3.52107771e-02 3.45756441e-01 5.02038836e-01 3.75028551e-01
4.51929331e-01 -3.28042537e-01 -3.73157114e-01 1.32212192e-01
9.30218175e-02 7.97651947e-01 -7.90946007e-01 -3.98479819e-01
-5.18499374e-01 1.16399610e+00 3.81776392e-01 7.05234170e-01
-1.19578815e+00 1.04782239e-01 4.41529721e-01 1.87037170e-01
4.86054510e-01 -4.22768891e-01 5.86518645e-03 -1.32571661e+00
1.85825214e-01 -5.97654998e-01 9.37710464e-01 -9.45286214e-01
-8.94079804e-01 6.86817050e-01 3.45470190e-01 -1.24759901e+00
-3.10184360e-01 -6.55042827e-01 -3.81070018e-01 5.40002584e-01
-1.29616725e+00 -1.49414992e+00 -4.74758506e-01 1.03864086e+00
7.92056680e-01 -1.02772087e-01 4.58239257e-01 2.62170583e-01
-1.30395770e-01 5.22463739e-01 -6.33369207e-01 1.13881245e-01
7.99008965e-01 -1.10531402e+00 2.41078168e-01 9.31915820e-01
3.81952435e-01 5.57702959e-01 7.77180135e-01 -7.06924915e-01
-1.12891495e+00 -9.22093809e-01 9.62250829e-01 -6.52279556e-01
3.63689780e-01 -4.49589431e-01 -6.70219839e-01 9.36404228e-01
3.47976200e-02 1.23756468e-01 4.09436494e-01 2.62556970e-02
-5.48263550e-01 1.52740898e-02 -7.06213236e-01 6.47322118e-01
1.28750229e+00 -7.73988008e-01 -6.81826234e-01 4.31338489e-01
8.32267940e-01 -6.46931231e-01 -5.89147270e-01 3.90208125e-01
6.86137557e-01 -1.15760887e+00 1.28249097e+00 -9.31637406e-01
4.13627595e-01 -5.21197975e-01 -2.09930047e-01 -8.03665519e-01
-2.25456879e-01 -6.33259833e-01 -4.37739342e-01 8.12141001e-01
1.20029077e-01 8.77110288e-03 6.82694912e-01 4.55526382e-01
-1.32276163e-01 -6.66474521e-01 -1.01697278e+00 -6.84173286e-01
-4.01122361e-01 -6.20411098e-01 2.35219067e-03 6.21272504e-01
2.07376644e-01 2.24069715e-01 -8.43345582e-01 2.75136858e-01
4.24074084e-01 1.80316269e-01 6.06539965e-01 -7.94801712e-01
-5.09384453e-01 -2.64044888e-02 -6.51047111e-01 -1.48993921e+00
1.45591721e-01 -6.09088004e-01 -6.67097047e-03 -1.45935011e+00
2.93178737e-01 2.32908547e-01 -3.65597516e-01 5.54258049e-01
-1.93068862e-01 1.58151031e-01 6.38390601e-01 8.51487834e-03
-1.14954221e+00 2.75429517e-01 1.14549184e+00 -1.57739595e-01
-3.21087629e-01 5.15913265e-03 -1.18874952e-01 7.23934829e-01
4.08842027e-01 -5.34674227e-02 -6.54946744e-01 -5.62586069e-01
-4.51215766e-02 4.60099906e-01 7.47119486e-01 -9.85266745e-01
4.53210622e-01 -2.53704339e-01 3.78013283e-01 -8.79977524e-01
6.25115812e-01 -7.35713780e-01 9.87389460e-02 4.28535908e-01
-4.55889553e-01 1.69194445e-01 2.36265525e-01 8.79503012e-01
-3.40735078e-01 6.06320560e-01 6.08239830e-01 -2.59872302e-02
-1.48312902e+00 3.90372276e-01 3.05857323e-03 7.78675452e-02
1.35951769e+00 -2.28199720e-01 -3.18280160e-01 -7.56622136e-01
-1.48939586e+00 5.52601755e-01 5.65804057e-02 8.57981443e-01
8.05300176e-01 -1.34872282e+00 -4.15422231e-01 -5.74586429e-02
2.18218684e-01 -4.77342576e-01 6.99246585e-01 1.05427659e+00
-1.87727749e-01 6.33685529e-01 -2.25455686e-01 -1.01050544e+00
-1.52490222e+00 7.25363731e-01 5.22289097e-01 -1.39491126e-01
-7.70163834e-01 1.15981662e+00 6.41278267e-01 3.99985254e-01
5.03466904e-01 -1.41362965e-01 -3.32556576e-01 3.40592042e-02
8.84283483e-01 1.99282318e-01 -2.72986054e-01 -1.30964482e+00
-6.06385231e-01 7.33059525e-01 6.11575656e-02 -2.80695915e-01
9.13227618e-01 -5.38068712e-01 1.13216087e-01 7.39886940e-01
1.12212098e+00 -3.99171114e-01 -1.62259889e+00 -3.25031787e-01
1.05978251e-01 -5.63634038e-01 -9.84126106e-02 -9.95856524e-01
-1.11657608e+00 1.02055180e+00 7.21632957e-01 -2.23244771e-01
1.24458551e+00 3.78297329e-01 7.17409313e-01 7.18741789e-02
5.96465409e-01 -9.66850102e-01 4.53746110e-01 5.61278045e-01
9.61204886e-01 -1.22558165e+00 -3.14418375e-01 -4.95865017e-01
-1.01284432e+00 1.00770462e+00 8.85126114e-01 2.55764753e-01
5.20636380e-01 1.17184907e-01 -5.00946268e-02 -2.15103820e-01
-1.10285997e+00 -3.94056410e-01 9.57056582e-01 7.58169115e-01
3.40017796e-01 3.02051902e-02 -4.61209118e-02 5.88025033e-01
1.81222305e-01 6.12157062e-02 -2.07787082e-01 7.37653255e-01
-2.65328884e-01 -8.47287059e-01 -1.62368894e-01 -1.27245724e-01
-1.86956137e-01 1.58365816e-01 -1.17175646e-01 9.60647583e-01
3.70497167e-01 1.01858044e+00 2.48470172e-01 -5.19859493e-01
3.94897997e-01 7.95121565e-02 4.73780364e-01 -3.46641451e-01
-4.44952786e-01 4.12450016e-01 1.77733943e-01 -1.15479624e+00
-8.22285891e-01 -7.84199059e-01 -1.47476876e+00 5.26878461e-02
3.07250947e-01 -7.42753521e-02 1.16200045e-01 1.23493588e+00
3.75282258e-01 4.78319436e-01 1.12427831e-01 -1.12175548e+00
2.14220658e-01 -4.86503065e-01 -1.33413941e-01 5.87540925e-01
6.55113637e-01 -9.13002491e-01 -4.80895676e-02 6.84460402e-01]
|
[9.455328941345215, 0.6582515239715576]
|
2d02c72a-eccf-440e-b77e-e7741d2c7ef0
|
improving-speaker-independent-lipreading-with
|
1708.01565
| null |
http://arxiv.org/abs/1708.01565v1
|
http://arxiv.org/pdf/1708.01565v1.pdf
|
Improving Speaker-Independent Lipreading with Domain-Adversarial Training
|
We present a Lipreading system, i.e. a speech recognition system using only
visual features, which uses domain-adversarial training for speaker
independence. Domain-adversarial training is integrated into the optimization
of a lipreader based on a stack of feedforward and LSTM (Long Short-Term
Memory) recurrent neural networks, yielding an end-to-end trainable system
which only requires a very small number of frames of untranscribed target data
to substantially improve the recognition accuracy on the target speaker. On
pairs of different source and target speakers, we achieve a relative accuracy
improvement of around 40% with only 15 to 20 seconds of untranscribed target
speech data. On multi-speaker training setups, the accuracy improvements are
smaller but still substantial.
|
['Michael Wand', 'Juergen Schmidhuber']
|
2017-08-04
| null | null | null | null |
['lipreading']
|
['computer-vision']
|
[ 3.54628235e-01 4.62752312e-01 -3.31804156e-01 -3.59580040e-01
-1.37817824e+00 -5.05799949e-01 6.61467612e-01 -5.51570535e-01
-4.76522267e-01 5.91031134e-01 2.78779715e-01 -4.66141433e-01
7.09521890e-01 1.06441274e-01 -6.93885386e-01 -4.95745659e-01
3.06398809e-01 1.37333691e-01 2.12613940e-02 8.90011266e-02
-5.43544218e-02 4.47591007e-01 -1.45471454e+00 2.09309474e-01
2.65932441e-01 1.16320407e+00 -4.15173685e-03 1.43958151e+00
-4.78806272e-02 6.41956747e-01 -7.99252152e-01 -4.08217400e-01
2.14007884e-01 -2.04235107e-01 -6.55121148e-01 8.81266966e-02
8.00260782e-01 -5.49662113e-01 -7.27704227e-01 7.50108421e-01
9.37927723e-01 1.61908984e-01 4.65990186e-01 -1.28674412e+00
-8.12994123e-01 1.36355788e-01 -1.54165283e-01 -6.06103055e-02
4.28917229e-01 4.11081254e-01 5.82233727e-01 -1.14716864e+00
1.88968629e-01 1.28779268e+00 5.85454583e-01 1.21504259e+00
-1.20600355e+00 -7.34789193e-01 -1.36982828e-01 1.11041948e-01
-1.33839440e+00 -1.77992988e+00 7.92540133e-01 -1.36970282e-01
1.10379469e+00 4.99837324e-02 1.85384080e-01 1.42388356e+00
-1.53353557e-01 7.01835930e-01 7.88462222e-01 -5.39917886e-01
4.89375629e-02 3.41466129e-01 -2.02638730e-01 4.13916051e-01
-4.90964413e-01 4.24521267e-01 -8.71962130e-01 9.67476591e-02
6.11141741e-01 -4.49200481e-01 -3.60674977e-01 -2.89947748e-01
-1.10246992e+00 7.59514570e-01 1.67047724e-01 -1.42271742e-01
6.86043426e-02 -2.23265067e-02 7.04466701e-01 3.55498910e-01
4.63875800e-01 -2.11864874e-01 -2.75418431e-01 -2.14317426e-01
-1.12830615e+00 -3.75629812e-01 8.12745154e-01 9.49489653e-01
3.79976749e-01 6.94414318e-01 2.08775094e-03 1.29799545e+00
4.84877497e-01 9.71907854e-01 8.05389822e-01 -8.95364821e-01
5.22016764e-01 -1.36840671e-01 7.02222958e-02 -3.72172952e-01
6.78248517e-03 9.40537676e-02 -6.28256738e-01 7.95209408e-01
4.28292811e-01 -4.56346154e-01 -1.35821939e+00 1.74035907e+00
5.09003624e-02 2.42473677e-01 6.00691855e-01 6.18356228e-01
9.77673292e-01 6.69409573e-01 2.85484791e-02 -2.41327509e-01
9.99575317e-01 -1.24993491e+00 -7.68832743e-01 -4.77651119e-01
-1.11942686e-01 -1.03868723e+00 1.18767691e+00 6.94289729e-02
-1.48481011e+00 -5.72673142e-01 -1.16169095e+00 -2.95184106e-01
-4.80256230e-01 2.24467978e-01 2.34860647e-02 9.05751705e-01
-1.41168654e+00 2.39112139e-01 -4.90737766e-01 -1.63312346e-01
3.45710635e-01 6.05755746e-01 -5.36781788e-01 1.07821040e-01
-1.10516059e+00 9.10037935e-01 -5.00701480e-02 -1.58747241e-01
-1.17197418e+00 -7.05535829e-01 -1.23493063e+00 -1.56390622e-01
5.94546348e-02 -3.94266218e-01 1.58062840e+00 -1.29839754e+00
-2.46999788e+00 1.12764347e+00 -6.28679276e-01 -6.87994361e-01
5.43403983e-01 4.20141481e-02 -7.58393645e-01 3.56842428e-01
-3.18891883e-01 9.76693869e-01 1.61488020e+00 -9.60666001e-01
-2.93027997e-01 -5.34570366e-02 -2.69146204e-01 2.90334195e-01
-1.76354349e-01 5.11141062e-01 -2.94858217e-01 -7.49030888e-01
-2.93375641e-01 -8.64845097e-01 5.69577992e-01 4.10525769e-01
-3.28472137e-01 -1.97525546e-01 1.12877429e+00 -9.00011420e-01
3.34413439e-01 -2.28878117e+00 -1.92682564e-01 -4.81723040e-01
-5.92476875e-02 7.17374206e-01 -5.01362085e-01 -2.78092418e-02
-2.42460921e-01 1.60429597e-01 -3.32989804e-02 -8.82747114e-01
-5.26490770e-02 -2.86739562e-02 -5.36170840e-01 5.22534490e-01
2.18210936e-01 9.85022187e-01 -4.76331651e-01 -3.42406064e-01
4.91878271e-01 8.60190034e-01 -1.10940613e-01 6.76185429e-01
-5.57055660e-02 2.25174099e-01 3.98004949e-01 6.71949387e-01
7.27521539e-01 2.36782447e-01 -4.27217364e-01 1.01289093e-01
1.48102358e-01 6.20117605e-01 -7.06427693e-01 1.68831980e+00
-7.01114774e-01 1.26016355e+00 5.26658118e-01 -6.62219346e-01
8.44352841e-01 7.57275283e-01 2.91897133e-02 -4.87167150e-01
-1.14645563e-01 7.85560384e-02 -3.12205106e-01 -3.24763358e-01
2.09894732e-01 -4.65167165e-01 2.50575632e-01 4.16142553e-01
2.14858189e-01 -3.05934101e-02 -4.96318072e-01 -8.09506103e-02
5.67379653e-01 -1.28393024e-01 1.09600546e-02 7.09878281e-02
6.15306616e-01 -6.19194567e-01 2.89387435e-01 3.37262362e-01
-6.46853864e-01 8.82930398e-01 3.66552137e-02 -2.82275490e-02
-1.19987094e+00 -1.39974225e+00 -5.53181954e-02 1.41846120e+00
-1.93350166e-01 8.53624791e-02 -7.99872458e-01 -5.01697898e-01
-1.63176000e-01 6.45918727e-01 -4.03509349e-01 -9.66746062e-02
-4.10294533e-01 1.32666782e-01 1.36719537e+00 4.68359709e-01
4.93275434e-01 -1.17878151e+00 -8.33565295e-02 1.92901061e-03
-1.65549785e-01 -1.25548160e+00 -1.06261420e+00 -5.13890907e-02
-2.86206007e-01 -6.68664932e-01 -1.37408841e+00 -1.07572305e+00
3.34625602e-01 1.81603536e-01 9.23401892e-01 -5.23765683e-01
3.34672481e-02 3.56555045e-01 1.18325047e-01 -5.65683544e-01
-9.15568769e-01 -3.60013485e-01 5.10114014e-01 3.76212820e-02
1.25234365e-01 -3.17166209e-01 -4.54528660e-01 5.07148743e-01
-3.03985238e-01 -3.90610322e-02 3.01925719e-01 9.61819112e-01
2.03249663e-01 -6.93666339e-01 7.80544221e-01 1.62453681e-01
4.25156891e-01 1.08191624e-01 -4.55806196e-01 2.08052099e-01
-3.41222584e-01 -7.04277903e-02 5.36661148e-01 -9.65692878e-01
-1.07639503e+00 -1.58321291e-01 -2.42117986e-01 -1.04283822e+00
-3.21653455e-01 -2.84851730e-01 -3.48620415e-01 -3.28090459e-01
5.78303158e-01 5.31516850e-01 6.70458317e-01 -5.12586892e-01
4.36084241e-01 1.17704463e+00 1.00853539e+00 6.05216585e-02
5.56633413e-01 1.23797357e-01 -5.41241109e-01 -1.25271261e+00
-8.14971849e-02 -2.84048468e-01 -4.65348780e-01 -2.66560446e-02
6.88887954e-01 -1.15576971e+00 -9.59971070e-01 9.27784979e-01
-1.21717358e+00 -7.43387043e-01 -2.07411602e-01 3.85935634e-01
-8.25459242e-01 2.71347404e-01 -4.86246735e-01 -9.02962029e-01
-5.39901733e-01 -1.09543478e+00 1.04982829e+00 1.58677936e-01
-7.34306574e-02 -9.56587851e-01 -1.54625289e-02 4.73452985e-01
4.75857913e-01 -2.96500087e-01 5.43052077e-01 -7.46403992e-01
-3.50908071e-01 -1.73926368e-01 -2.30664790e-01 9.05002713e-01
3.82323027e-01 2.98700016e-02 -1.60523033e+00 -5.33976793e-01
-1.27801448e-01 -6.68066859e-01 8.07067156e-01 5.36475718e-01
7.20899880e-01 -7.09387124e-01 -2.40367785e-01 5.71179748e-01
8.74885857e-01 1.36493415e-01 6.51498556e-01 -3.03046197e-01
5.70131183e-01 5.27818620e-01 1.49637446e-01 4.82169017e-02
1.92713261e-01 8.48812759e-01 5.63272387e-02 -3.93776923e-01
-8.36988986e-01 -5.06105721e-01 9.92047071e-01 4.48123097e-01
5.17999530e-01 -2.93124825e-01 -7.29622304e-01 7.67482340e-01
-1.16597235e+00 -1.15133691e+00 7.10275054e-01 2.45317650e+00
9.25239146e-01 -4.05909009e-02 4.86922026e-01 1.70994893e-01
1.12504184e+00 3.76571149e-01 -8.79996181e-01 -8.13432395e-01
5.39466739e-02 1.48361146e-01 5.11512518e-01 1.09452283e+00
-1.05976427e+00 1.13887107e+00 7.75072289e+00 6.09067321e-01
-1.74180031e+00 1.96077973e-02 4.93884742e-01 -4.19273615e-01
6.84423447e-02 -5.91636837e-01 -7.09841847e-01 4.75775093e-01
1.55906630e+00 -4.10524011e-01 6.51247978e-01 8.11790287e-01
4.22974169e-01 2.00771138e-01 -1.15029669e+00 1.31273401e+00
4.24661219e-01 -1.21093404e+00 -1.38477728e-01 8.16498399e-02
2.06615895e-01 4.29664910e-01 4.89042640e-01 2.33738020e-01
2.59101212e-01 -1.37930167e+00 6.76683605e-01 2.33902231e-01
1.65192986e+00 -7.61742830e-01 1.85413316e-01 1.87209100e-01
-9.48081195e-01 1.20538726e-01 -3.74311715e-01 3.10237527e-01
9.74506065e-02 -4.50117067e-02 -1.25384688e+00 -2.18304470e-01
3.78312409e-01 3.84571701e-01 -1.69465870e-01 5.76285839e-01
-1.22941479e-01 6.00403905e-01 -3.80856425e-01 1.09971479e-01
-8.68175626e-02 6.04275882e-01 7.19459057e-01 1.15197277e+00
-1.04384847e-01 -2.10984468e-01 -2.50594586e-01 5.22451282e-01
-4.41871047e-01 -1.68853134e-01 -8.58581066e-01 9.65993032e-02
4.27785069e-01 8.51385355e-01 2.63805836e-01 -2.94941932e-01
-2.93209046e-01 1.17234242e+00 1.28259316e-01 6.80709600e-01
-7.35190868e-01 -5.94800949e-01 1.14586353e+00 -4.02360633e-02
4.30215418e-01 -1.62829295e-01 -3.27972949e-01 -1.10654056e+00
1.04636319e-01 -1.07808948e+00 -1.42525226e-01 -8.79269063e-01
-1.02908957e+00 6.59086287e-01 -6.09722853e-01 -1.02373934e+00
-9.39854681e-01 -6.12148941e-01 -7.13789821e-01 1.28838384e+00
-1.66475475e+00 -1.33326030e+00 1.53679233e-02 9.77177382e-01
8.73742938e-01 -6.05202377e-01 1.26232648e+00 -1.93586305e-01
-5.00411928e-01 1.44338119e+00 1.82555705e-01 4.43798423e-01
1.09203362e+00 -9.66205537e-01 7.94380307e-01 7.55605280e-01
7.92970881e-02 3.49279135e-01 6.49464726e-01 -1.48349896e-01
-1.03687298e+00 -1.08816588e+00 1.07360816e+00 -3.27462703e-01
4.04551059e-01 -6.78359270e-01 -8.26615691e-01 6.94446623e-01
6.00128114e-01 2.51447111e-01 9.02932525e-01 -7.94633925e-02
-7.37674475e-01 -1.60533249e-01 -1.40912855e+00 6.18443370e-01
7.09986925e-01 -1.24819505e+00 -7.93644547e-01 2.17266276e-01
9.20569599e-01 -4.06043112e-01 -7.07228303e-01 1.93729773e-01
8.02001417e-01 -6.46604300e-01 1.20223320e+00 -5.95672846e-01
-4.46648598e-02 3.41544896e-02 -1.25653297e-01 -1.17544496e+00
2.55186796e-01 -1.13480759e+00 -1.34616271e-01 1.34737480e+00
5.72888613e-01 -8.50696325e-01 7.20288873e-01 8.04268360e-01
1.20650850e-01 -4.40728426e-01 -1.41863084e+00 -9.40370262e-01
4.02081192e-01 -3.95321757e-01 1.31168216e-01 3.88628840e-01
2.39444137e-01 4.03907031e-01 -4.92610574e-01 1.56179011e-01
6.12210333e-01 -2.34722540e-01 8.46512735e-01 -7.20841765e-01
-6.26678094e-02 -2.25296885e-01 -3.64729553e-01 -1.31806624e+00
7.33229816e-01 -5.97918153e-01 3.69933724e-01 -9.86362338e-01
-1.73876569e-01 -8.93042162e-02 -2.96764523e-01 7.61970341e-01
6.37709722e-02 4.84331071e-01 1.78830683e-01 1.63963571e-01
-9.16406736e-02 6.15252614e-01 6.40616000e-01 -5.35805225e-01
-2.44320661e-01 2.99553901e-01 -3.36697489e-01 7.32556045e-01
7.96520591e-01 -1.77763686e-01 -4.08174336e-01 -3.99611324e-01
-9.10691619e-01 2.72501022e-01 5.68455160e-01 -9.15991426e-01
2.94569433e-01 3.81995626e-02 4.74389821e-01 -3.02787840e-01
9.55612361e-01 -3.27248037e-01 -5.37450492e-01 7.71292299e-02
-6.93721414e-01 -3.78103137e-01 7.73148656e-01 4.46269542e-01
-1.26363263e-01 4.97417115e-02 1.34957898e+00 2.95665652e-01
-5.36499798e-01 2.46561646e-01 -4.22109932e-01 1.56574219e-01
8.39918852e-01 -3.05202335e-01 -2.17867017e-01 -9.50262189e-01
-8.85218441e-01 -3.10915798e-01 5.85114360e-01 6.80902779e-01
8.55484128e-01 -1.18896735e+00 -7.95589328e-01 6.38675451e-01
-7.45039508e-02 -3.93361449e-01 8.64001215e-02 2.42280692e-01
9.43872929e-02 7.61095643e-01 -1.55802652e-01 -6.64112568e-01
-1.89780807e+00 6.49333179e-01 7.41644025e-01 2.76824027e-01
-5.05063653e-01 1.14955270e+00 9.47091207e-02 -2.35725507e-01
5.93346059e-01 7.89998770e-02 1.50170371e-01 -2.04184204e-01
8.48189890e-01 2.19956860e-01 5.36959106e-03 -9.32671368e-01
-5.53010345e-01 4.52930808e-01 -1.26209527e-01 -7.20415413e-01
7.24461973e-01 -2.33189970e-01 6.35457397e-01 4.14934814e-01
1.66949189e+00 2.12585896e-01 -1.72298324e+00 -1.50809035e-01
-6.16370440e-01 -2.85345763e-01 2.49744043e-01 -1.01018214e+00
-7.42211640e-01 1.34279549e+00 8.19495678e-01 -1.23521604e-01
9.18657780e-01 4.42227162e-03 7.81880558e-01 2.81995982e-01
5.74846333e-03 -8.56326997e-01 3.44589241e-02 3.62904012e-01
1.24979126e+00 -1.53534842e+00 -4.91254330e-01 -5.34118488e-02
-7.83335924e-01 1.00169110e+00 1.35515332e-01 2.48055890e-01
5.42931914e-01 4.82844174e-01 5.90311050e-01 6.52029932e-01
-7.12801456e-01 -2.18302101e-01 4.36666310e-01 1.06446517e+00
9.33565944e-02 5.22354916e-02 7.52280712e-01 1.85014352e-01
-2.73251981e-01 5.40485643e-02 1.74130768e-01 3.02463472e-01
-2.79503703e-01 -9.49989796e-01 -3.21858019e-01 -1.91420808e-01
-5.80999672e-01 -4.03554469e-01 -4.51837599e-01 4.81705606e-01
-4.39844102e-01 1.40917981e+00 2.32924625e-01 -2.76414603e-01
8.79150704e-02 5.70055485e-01 1.97972804e-01 -3.45829844e-01
-1.78823739e-01 4.92442660e-02 1.41804934e-01 -5.49673796e-01
-6.64579198e-02 -5.03465116e-01 -1.07762885e+00 -5.15501261e-01
-1.81069732e-01 -2.88068265e-01 9.26726401e-01 7.35823989e-01
6.39099300e-01 1.04153842e-01 9.05135930e-01 -9.77532148e-01
-7.35113859e-01 -1.12745214e+00 -2.80013204e-01 -6.09180480e-02
1.19091761e+00 -3.93422365e-01 -6.99076116e-01 3.66178125e-01]
|
[14.343911170959473, 5.018189430236816]
|
4f1f9457-4524-469b-a8f3-84a1892a3a8d
|
adversarial-attack-and-defense-for-dehazing
|
2303.17255
| null |
https://arxiv.org/abs/2303.17255v1
|
https://arxiv.org/pdf/2303.17255v1.pdf
|
Adversarial Attack and Defense for Dehazing Networks
|
The research on single image dehazing task has been widely explored. However, as far as we know, no comprehensive study has been conducted on the robustness of the well-trained dehazing models. Therefore, there is no evidence that the dehazing networks can resist malicious attacks. In this paper, we focus on designing a group of attack methods based on first order gradient to verify the robustness of the existing dehazing algorithms. By analyzing the general goal of image dehazing task, five attack methods are proposed, which are prediction, noise, mask, ground-truth and input attack. The corresponding experiments are conducted on six datasets with different scales. Further, the defense strategy based on adversarial training is adopted for reducing the negative effects caused by malicious attacks. In summary, this paper defines a new challenging problem for image dehazing area, which can be called as adversarial attack on dehazing networks (AADN). Code is available at https://github.com/guijiejie/AADN.
|
['James Tin-Yau Kwok', 'Yuan Yan Tang', 'Chengwei Peng', 'Xiaofeng Cong', 'Jie Gui']
|
2023-03-30
| null | null | null | null |
['image-dehazing', 'single-image-dehazing']
|
['computer-vision', 'computer-vision']
|
[ 3.13554019e-01 -3.02059114e-01 1.22337915e-01 2.83209290e-02
-2.57822156e-01 -5.31604886e-01 6.62795961e-01 -2.05700159e-01
-1.93196610e-01 4.29776102e-01 1.06002996e-02 -3.56090516e-01
1.22147761e-02 -9.35853839e-01 -5.51322281e-01 -1.18103039e+00
5.56991063e-02 -4.99783814e-01 4.66204852e-01 -3.92921448e-01
4.62814689e-01 4.56984252e-01 -1.25091529e+00 2.32551679e-01
8.22446227e-01 9.39360738e-01 -2.05686092e-01 8.71879101e-01
4.59871441e-01 8.11014771e-01 -1.04053307e+00 -5.55561125e-01
5.41763365e-01 -5.62572718e-01 -4.06318724e-01 -2.44163368e-02
3.15153539e-01 -7.62827992e-01 -7.72161663e-01 1.64905417e+00
7.42808759e-01 -4.95680645e-02 6.04755461e-01 -1.58187521e+00
-1.09384310e+00 3.48990828e-01 -5.37951410e-01 4.99761552e-01
-4.63331342e-02 3.87956291e-01 4.09444392e-01 -7.03790605e-01
4.32071686e-02 1.21165168e+00 5.09771287e-01 8.30312729e-01
-6.17179096e-01 -1.21572793e+00 1.23946309e-01 4.21651185e-01
-1.52706647e+00 -3.39975923e-01 8.85967195e-01 -2.47539088e-01
3.56909126e-01 3.72489750e-01 1.91834554e-01 1.20910645e+00
5.93463123e-01 5.80878556e-01 1.30228949e+00 -2.89141893e-01
1.32204637e-01 3.66333425e-01 -8.28370079e-02 5.01664877e-01
4.19483364e-01 4.54240680e-01 -1.96232468e-01 -1.00130916e-01
4.70954895e-01 8.41349512e-02 -4.59409416e-01 1.18401699e-01
-5.76828957e-01 9.50677276e-01 3.77749711e-01 2.88174897e-01
-1.19671948e-01 -4.10133339e-02 5.53987443e-01 4.51631337e-01
5.81595540e-01 2.37923563e-01 -1.46820307e-01 5.25710285e-01
-5.92508495e-01 1.50689065e-01 5.52172482e-01 6.16532803e-01
3.74006152e-01 4.40847158e-01 2.19762102e-02 5.51011741e-01
4.09383029e-01 5.79435766e-01 4.97958481e-01 -4.11450893e-01
2.33044058e-01 1.97571620e-01 -1.38796866e-01 -1.75382292e+00
1.08336568e-01 -2.14576051e-01 -1.08704793e+00 4.65451270e-01
-6.18691929e-02 -5.38511336e-01 -9.82293963e-01 1.42539477e+00
3.10218573e-01 7.31164753e-01 3.28822672e-01 9.13226545e-01
9.24430490e-01 1.05412579e+00 2.24332884e-01 -1.15032256e-01
1.18384635e+00 -9.12121952e-01 -8.33270192e-01 -1.13207221e-01
2.76260853e-01 -9.35782433e-01 8.68653536e-01 7.21340835e-01
-7.66154945e-01 -5.69309533e-01 -1.57178223e+00 3.33325773e-01
-7.92877853e-01 -2.86973178e-01 3.68961871e-01 1.24912798e+00
-8.76507223e-01 3.67256045e-01 -5.73015153e-01 -8.43941942e-02
5.31208992e-01 2.86324352e-01 -2.52945811e-01 -1.69188991e-01
-1.67302012e+00 1.01209080e+00 5.52279651e-01 2.80352056e-01
-1.49451709e+00 -6.16010606e-01 -5.47215998e-01 -2.29781747e-01
4.23799574e-01 -1.31122246e-01 9.70883787e-01 -1.06270146e+00
-1.37955284e+00 7.81472683e-01 6.01426303e-01 -6.85132384e-01
5.24369776e-01 -1.87116906e-01 -8.30367565e-01 2.25506008e-01
-3.44978541e-01 1.98313475e-01 1.28800023e+00 -1.41292107e+00
-4.94661480e-01 -4.10156667e-01 1.15405753e-01 5.62689863e-02
-7.23724365e-01 3.52506876e-01 -4.07856032e-02 -1.21286356e+00
-4.24423575e-01 -6.82278514e-01 -1.83228612e-01 -6.03653640e-02
-5.40653348e-01 5.69924414e-02 1.36206138e+00 -9.10620153e-01
1.64458585e+00 -2.47545028e+00 -1.40562177e-01 2.07931906e-01
2.38330483e-01 9.15264904e-01 -9.50785503e-02 5.91181576e-01
-2.63964564e-01 6.27855539e-01 -4.32328194e-01 2.95466632e-01
-1.37933895e-01 2.45413873e-02 -6.68344438e-01 6.99151874e-01
1.91679329e-01 4.82020497e-01 -5.16075134e-01 -3.36860955e-01
2.67430276e-01 6.42380059e-01 -1.90079838e-01 4.18164909e-01
2.45623998e-02 8.33247676e-02 -4.85965610e-01 6.27738595e-01
1.20126259e+00 2.75874943e-01 -4.05633807e-01 -2.57021964e-01
4.08776179e-02 -4.45676953e-01 -1.24117219e+00 8.59117568e-01
-1.72706217e-01 5.09136796e-01 1.84446216e-01 -9.69936430e-01
8.44238937e-01 4.67439085e-01 1.52857065e-01 -2.83355236e-01
5.37281334e-01 1.61587343e-01 1.85140774e-01 -6.06270909e-01
1.11300848e-01 -6.26910031e-02 1.38980493e-01 3.79812002e-01
-5.04163839e-02 -1.03461675e-01 -3.23027223e-02 2.24690095e-01
9.49956000e-01 -2.97529846e-01 1.32642478e-01 -2.29683921e-01
8.75667989e-01 -7.05592930e-02 3.96381944e-01 6.79782093e-01
-5.74684978e-01 3.80810678e-01 1.65380970e-01 -3.43877375e-01
-8.73689055e-01 -8.29258382e-01 -2.77744625e-02 8.78426433e-01
5.42688608e-01 -3.41817915e-01 -1.20422077e+00 -8.00437272e-01
-2.19871044e-01 4.81895328e-01 -7.06241190e-01 -7.03147590e-01
-4.60506260e-01 -9.68315065e-01 9.20729399e-01 -4.75884527e-02
1.27454531e+00 -9.09702957e-01 -1.80254608e-01 -1.94552928e-01
1.53252259e-01 -9.20667291e-01 -5.93671143e-01 -3.66171867e-01
-4.00175184e-01 -1.19214392e+00 -6.96533084e-01 -8.68433297e-01
5.05411208e-01 6.67630434e-01 4.80682254e-01 5.74468195e-01
-3.28193158e-01 1.45709455e-01 -5.54098308e-01 -9.22918737e-01
-6.60000503e-01 -2.48245165e-01 7.05670938e-02 2.17157274e-01
4.82854605e-01 -5.32258391e-01 -7.26334035e-01 6.70405746e-01
-1.46935499e+00 -3.86385113e-01 7.94847667e-01 5.57215512e-01
2.24696651e-01 9.17381704e-01 4.99255270e-01 -9.13936794e-01
8.09072495e-01 -6.94102049e-01 -4.43367720e-01 1.53967515e-01
-6.44316792e-01 -3.65610361e-01 7.89810419e-01 -9.07298982e-01
-1.07948101e+00 -4.20132399e-01 -3.72196436e-01 -5.69266856e-01
-3.00198525e-01 3.10725600e-01 -4.14962769e-01 -6.15238011e-01
7.85913646e-01 2.83167273e-01 -4.18248251e-02 -2.39749372e-01
2.11679831e-01 8.68897319e-01 4.54725683e-01 -1.17627434e-01
1.42464983e+00 5.41858912e-01 -2.09842592e-01 -9.80805576e-01
-6.77987218e-01 -2.88344678e-02 -5.32832108e-02 -3.35063905e-01
9.70576942e-01 -6.72283411e-01 -4.98593420e-01 1.14180791e+00
-9.00642872e-01 -1.16410084e-01 3.81121159e-01 2.61910617e-01
-3.42458934e-02 7.28071809e-01 -6.71839535e-01 -6.69260085e-01
-5.39526403e-01 -1.00731277e+00 3.49658459e-01 3.56177390e-01
4.66820896e-01 -9.07364249e-01 -2.45591123e-02 2.28663370e-01
6.27456248e-01 4.95800048e-01 6.52025163e-01 -8.19182634e-01
-5.67780316e-01 -4.01601166e-01 -7.72261918e-02 9.51813877e-01
1.24160416e-01 3.00977916e-01 -8.87339592e-01 -4.72815871e-01
5.86574972e-01 -1.25599056e-01 6.63725078e-01 1.99458867e-01
1.29895473e+00 -6.78758562e-01 -8.72854963e-02 7.39629984e-01
1.47422838e+00 5.94056726e-01 1.07084036e+00 5.35424113e-01
6.60691440e-01 5.11517346e-01 6.28139615e-01 2.89540321e-01
-2.46713221e-01 2.54148811e-01 8.30313146e-01 -1.30969584e-01
2.03152582e-01 -1.64608374e-01 3.62319261e-01 5.52348852e-01
3.97820733e-02 -7.46866941e-01 -8.37880373e-01 1.88562408e-01
-1.43775892e+00 -1.00875211e+00 1.11030051e-02 1.79571307e+00
4.83340889e-01 2.93237567e-01 -2.00737584e-02 4.13334310e-01
1.12020564e+00 6.15277708e-01 -5.29349089e-01 -4.02808279e-01
-7.21267015e-02 -2.62348223e-02 7.79490650e-01 3.24297726e-01
-1.54733956e+00 1.02144647e+00 5.63586664e+00 1.19917345e+00
-1.11000288e+00 6.78122938e-02 7.47318447e-01 2.34844744e-01
5.06159328e-02 -1.52711958e-01 -7.29297161e-01 7.30144560e-01
8.06709230e-01 -3.26706469e-01 3.75615209e-01 8.25191557e-01
2.22933814e-01 3.30993146e-01 -3.98805499e-01 9.34928894e-01
5.29218495e-01 -1.02827990e+00 3.23409617e-01 -2.03204118e-02
7.10407019e-01 -4.11823809e-01 4.59969223e-01 1.19460553e-01
3.79129261e-01 -1.19275939e+00 3.19113523e-01 1.21147372e-01
3.89662445e-01 -1.02199447e+00 9.32106495e-01 3.20636690e-01
-8.76457989e-01 -1.45814270e-01 -4.41758335e-01 2.27526531e-01
-1.34101123e-01 1.35297403e-01 -3.19426209e-01 6.46240115e-01
8.33762884e-01 5.90529740e-01 -5.36817491e-01 8.91318917e-01
-4.49985802e-01 9.33490515e-01 6.69369698e-02 9.35592130e-02
4.16409135e-01 -4.12703678e-02 5.14256358e-01 1.11050034e+00
1.85610875e-01 4.31728691e-01 9.87827778e-02 3.02601963e-01
-3.87297384e-03 1.60560995e-01 -7.75153637e-01 -9.51129869e-02
5.23942173e-01 1.07963228e+00 -6.08092308e-01 -2.75875647e-02
-2.45976314e-01 8.21409404e-01 -3.26643616e-01 2.98874885e-01
-1.10659349e+00 -7.71801829e-01 5.70024550e-01 -1.09449930e-01
5.43212965e-02 -6.76538646e-02 5.78539521e-02 -9.42298055e-01
-1.90911815e-01 -1.44158816e+00 5.56407630e-01 -6.79998994e-01
-1.53700924e+00 5.55631638e-01 1.62798986e-02 -1.21448314e+00
4.06283021e-01 -7.69752026e-01 -9.62225139e-01 7.83274233e-01
-1.40791297e+00 -9.60150480e-01 -4.33641106e-01 8.24299395e-01
6.10388935e-01 -4.80277866e-01 5.19785166e-01 5.23455262e-01
-8.85183752e-01 7.10185945e-01 -4.10241187e-02 3.80523354e-01
6.98274612e-01 -6.77071631e-01 1.69805050e-01 1.46096230e+00
-2.78293073e-01 4.49715436e-01 9.30964947e-01 -6.37851477e-01
-1.13359082e+00 -1.06405246e+00 1.44496948e-01 -2.42280468e-01
8.00828755e-01 -1.00771219e-01 -1.18945551e+00 5.30375063e-01
6.11275077e-01 -1.24458317e-02 6.17736101e-01 -9.18238819e-01
-2.84406871e-01 -1.07814692e-01 -1.33944035e+00 8.04656863e-01
7.17065573e-01 -2.86438167e-01 -2.80835301e-01 3.68674874e-01
8.54577065e-01 -2.95441210e-01 -7.03952312e-01 5.44816077e-01
2.30784178e-01 -8.94990087e-01 1.22125256e+00 -6.33769095e-01
5.82784593e-01 -4.39352632e-01 -1.31023228e-01 -1.20058239e+00
-2.20646784e-01 -7.72503138e-01 -1.31929591e-01 1.24912417e+00
-9.30732023e-03 -8.30603182e-01 7.22357631e-01 1.02352411e-01
3.29982527e-02 -7.57676005e-01 -4.92192268e-01 -7.30715871e-01
1.31317303e-01 -1.25742942e-01 6.96382880e-01 1.20311284e+00
-6.08739316e-01 -3.14459354e-02 -8.21030140e-01 9.02506053e-01
6.34425938e-01 -7.12568104e-01 7.24646449e-01 -6.26573205e-01
-9.22207609e-02 -3.27212363e-01 -6.67259693e-01 -6.54793978e-01
3.65311690e-02 -4.44413930e-01 9.27328877e-03 -1.10777819e+00
-1.68973684e-01 -8.72309133e-02 -3.94943953e-01 1.86473817e-01
-5.22198617e-01 4.47681606e-01 1.32985577e-01 2.86823332e-01
-1.92439966e-02 3.92429650e-01 1.22703683e+00 -3.70576769e-01
1.58734694e-01 4.67600152e-02 -9.16621983e-01 8.62100482e-01
1.55413246e+00 -6.36024356e-01 -7.24925518e-01 -4.17846531e-01
-2.26462305e-01 -3.21460485e-01 3.48595023e-01 -9.00519133e-01
1.98639378e-01 -4.15796429e-01 5.59838787e-02 -4.02437001e-01
4.66190316e-02 -8.83334160e-01 7.07518170e-03 6.49564683e-01
-2.79271036e-01 3.07069093e-01 9.19866115e-02 6.63746715e-01
-3.86930853e-01 -3.95653695e-01 1.16798306e+00 -2.20310420e-01
-8.31772983e-01 6.91980958e-01 -5.33906996e-01 7.71150962e-02
1.52023339e+00 -2.61721730e-01 -7.07302034e-01 -3.99410725e-01
-3.15769881e-01 7.10813403e-02 1.13475032e-01 3.57609749e-01
8.81443083e-01 -1.24124146e+00 -8.90628874e-01 1.88751653e-01
-1.09264694e-01 -3.93932313e-01 6.01484060e-01 3.01838100e-01
-8.52293789e-01 -2.39970535e-02 -3.89191628e-01 -5.98231284e-03
-1.67904401e+00 1.16218793e+00 5.61738908e-01 -1.09159097e-01
-3.60380501e-01 1.00397265e+00 2.91931868e-01 5.58565259e-02
3.53074253e-01 4.87260044e-01 -4.80393142e-01 -2.66560435e-01
8.99537146e-01 4.17576313e-01 -1.49543583e-01 -6.91849589e-01
-1.75611377e-01 3.99367601e-01 -3.57479781e-01 8.98356140e-02
1.05386555e+00 3.73237543e-02 -3.17068189e-01 -1.72786906e-01
1.08504689e+00 1.74927469e-02 -1.08018064e+00 -5.77237234e-02
-2.66228050e-01 -9.22718644e-01 1.58789501e-01 -5.32820225e-01
-1.27419603e+00 9.26664114e-01 8.98777127e-01 7.23436654e-01
1.39733720e+00 -6.01238668e-01 7.84585238e-01 9.38197672e-02
-1.19972244e-01 -8.40092301e-01 4.39112186e-01 2.07671359e-01
9.75430906e-01 -1.09761190e+00 -2.20148321e-02 -5.42301953e-01
-6.79844856e-01 7.04110563e-01 9.02446687e-01 -6.44192874e-01
1.05108035e+00 2.52607465e-01 3.59150141e-01 4.75611091e-02
-4.14725333e-01 2.84278303e-01 7.99173564e-02 7.45253682e-01
-1.62731633e-01 -3.10183436e-01 -3.61769676e-01 2.47271672e-01
-1.93473995e-01 -3.88420582e-01 6.04222536e-01 9.58568275e-01
-5.05983353e-01 -9.43238437e-01 -6.38344586e-01 2.47163326e-01
-9.50257897e-01 -1.38470605e-01 -4.17962044e-01 8.59084487e-01
2.17353344e-01 1.22643554e+00 -4.91163939e-01 -8.08807671e-01
2.74604917e-01 -4.30392325e-01 2.43355501e-02 -4.42700028e-01
-5.58438122e-01 -2.77621269e-01 -1.90489739e-01 -1.80868417e-01
-4.47629511e-01 -4.43887636e-02 -5.70564032e-01 -7.80724227e-01
-4.58217978e-01 2.00705990e-01 4.61909413e-01 6.44682527e-01
1.07582696e-01 5.07870138e-01 1.02701235e+00 -5.91402411e-01
-8.05205464e-01 -8.03919196e-01 -5.46977341e-01 4.31063443e-01
3.64360750e-01 -5.50356269e-01 -8.43707800e-01 3.06168497e-02]
|
[5.513589382171631, 7.944633960723877]
|
aec0c9f7-85bd-4286-ba06-c5905b879793
|
towards-adversarial-retinal-image-synthesis
|
1701.08974
| null |
http://arxiv.org/abs/1701.08974v1
|
http://arxiv.org/pdf/1701.08974v1.pdf
|
Towards Adversarial Retinal Image Synthesis
|
Synthesizing images of the eye fundus is a challenging task that has been
previously approached by formulating complex models of the anatomy of the eye.
New images can then be generated by sampling a suitable parameter space. In
this work, we propose a method that learns to synthesize eye fundus images
directly from data. For that, we pair true eye fundus images with their
respective vessel trees, by means of a vessel segmentation technique. These
pairs are then used to learn a mapping from a binary vessel tree to a new
retinal image. For this purpose, we use a recent image-to-image translation
technique, based on the idea of adversarial learning. Experimental results show
that the original and the generated images are visually different in terms of
their global appearance, in spite of sharing the same vessel tree.
Additionally, a quantitative quality analysis of the synthetic retinal images
confirms that the produced images retain a high proportion of the true image
set quality.
|
['Aurélio Campilho', 'Ana Maria Mendonça', 'Michael David Abràmoff', 'Maria Inês Meyer', 'Adrian Galdran', 'Pedro Costa', 'Meindert Niemeijer']
|
2017-01-31
| null | null | null | null |
['medical-image-generation']
|
['medical']
|
[ 4.48950052e-01 5.82857966e-01 1.71292707e-01 -2.27603495e-01
-2.31431514e-01 -7.91052878e-01 5.19555569e-01 -4.45399135e-01
-1.99251711e-01 9.06504631e-01 -1.20735802e-01 -3.33386034e-01
2.96188533e-01 -7.97298491e-01 -9.50709462e-01 -7.75826156e-01
2.90177315e-01 -4.37030382e-02 2.76678145e-01 -4.01093848e-02
8.34384188e-02 6.24002993e-01 -1.48882389e+00 1.27359971e-01
9.38887954e-01 7.97956407e-01 -2.91286558e-01 7.81776249e-01
-6.37184083e-02 6.27866685e-01 -5.24830699e-01 -5.92040539e-01
8.03916037e-01 -1.09345043e+00 -6.85637116e-01 5.43088913e-01
7.38389313e-01 -4.87230599e-01 -1.45453751e-01 1.24412489e+00
1.96183071e-01 -2.75862038e-01 5.82411468e-01 -9.82530475e-01
-3.90049428e-01 1.94297835e-01 -4.10664469e-01 -7.82657936e-02
1.85612217e-01 3.58017147e-01 5.81246674e-01 -2.31184915e-01
8.77653718e-01 9.09074545e-01 2.76567191e-01 6.38187766e-01
-1.80247450e+00 -4.93982464e-01 -3.56694043e-01 -2.66742617e-01
-1.24886727e+00 -5.56660891e-01 6.67980373e-01 -7.14593112e-01
-7.86812454e-02 2.15777531e-01 9.74417567e-01 7.19660580e-01
3.07779878e-01 1.22038789e-01 1.85168588e+00 -6.17160201e-01
2.26926953e-01 4.40094978e-01 -4.87083256e-01 9.23161149e-01
3.54230583e-01 3.47824186e-01 -8.95413458e-02 6.44904971e-02
1.15203249e+00 -3.50944012e-01 -5.04565001e-01 -6.30226612e-01
-1.02988183e+00 6.37869418e-01 5.28481364e-01 2.56335914e-01
-2.60549039e-01 1.82077229e-01 -1.14782721e-01 3.79381031e-01
1.59816489e-01 6.06166661e-01 1.62071481e-01 4.84073818e-01
-8.37997377e-01 2.00883031e-01 7.57064581e-01 5.96902251e-01
7.83644795e-01 -4.47439291e-02 -1.52352214e-01 2.89093316e-01
2.89207280e-01 4.05340046e-01 4.57626104e-01 -1.17653298e+00
-7.93899521e-02 6.71972036e-01 3.30368578e-01 -7.48268485e-01
-5.84467389e-02 -3.88574183e-01 -6.32828653e-01 1.04785156e+00
8.56207073e-01 -2.41752237e-01 -1.00007594e+00 1.59944820e+00
4.62027609e-01 3.29305500e-01 1.53036937e-01 8.84639263e-01
4.76567179e-01 1.99644074e-01 -6.36057481e-02 -1.62598029e-01
1.18829393e+00 -6.78598702e-01 -6.27014756e-01 -1.22100115e-03
1.48528725e-01 -8.40168297e-01 8.82200181e-01 2.35214323e-01
-1.42339122e+00 -6.18149936e-01 -1.16366088e+00 4.05504815e-02
-2.88817901e-02 2.59719938e-01 3.84545922e-01 8.11437488e-01
-1.27189672e+00 4.66585964e-01 -6.61678255e-01 -1.09438725e-01
5.86685479e-01 1.66518837e-01 -4.38464999e-01 2.33204495e-02
-7.27950573e-01 8.03955078e-01 3.43232304e-01 -2.60650814e-01
-6.03108704e-01 -6.67985797e-01 -7.03315973e-01 -1.99535251e-01
5.33798449e-02 -1.17268836e+00 1.05394363e+00 -1.34229207e+00
-1.84921169e+00 1.29108429e+00 -1.50592163e-01 -6.39055014e-01
8.68343055e-01 4.16959167e-01 -2.66819507e-01 5.82047343e-01
-2.32288644e-01 7.39235997e-01 1.39975190e+00 -1.38944280e+00
-6.03192806e-01 -2.92347133e-01 1.05671525e-01 -3.07552367e-01
2.22827420e-01 -5.65592945e-02 -1.72920108e-01 -5.04483104e-01
-1.22465931e-01 -9.25011098e-01 -2.10141331e-01 6.72905743e-01
-5.99642754e-01 4.87352073e-01 1.84625939e-01 -5.51391363e-01
8.01690519e-01 -2.03918695e+00 2.01725423e-01 4.29155111e-01
5.59814692e-01 3.91522229e-01 -2.99345523e-01 -3.46210971e-02
-2.12615222e-01 2.09470645e-01 -3.35608155e-01 -2.63706625e-01
-5.03840566e-01 8.92845243e-02 -3.23884219e-01 5.65534472e-01
3.19996566e-01 9.57835555e-01 -8.23624015e-01 -6.03874803e-01
1.47258669e-01 4.74046558e-01 -4.75376427e-01 3.11318725e-01
-1.79983318e-01 1.04602051e+00 -4.08758342e-01 2.71581799e-01
5.47223568e-01 -7.26084858e-02 1.35230005e-01 -3.29175204e-01
-7.09312484e-02 -2.86640406e-01 -7.23741174e-01 1.39567137e+00
-4.08392459e-01 6.81516230e-01 -2.04467446e-01 -6.89906120e-01
9.00156915e-01 3.59244615e-01 3.13831419e-01 -5.53305745e-01
5.06380975e-01 4.64510709e-01 3.85966897e-01 -4.25396055e-01
-4.87932898e-02 -4.84748542e-01 4.67727900e-01 4.71569926e-01
-1.14612877e-01 -6.33258224e-01 3.29683781e-01 -2.08402947e-01
6.57892764e-01 1.23281583e-01 2.76034296e-01 1.68615550e-01
6.84357405e-01 -1.82385594e-01 8.33219439e-02 3.47201347e-01
-9.88170058e-02 8.98437440e-01 8.33354115e-01 -4.12451923e-01
-1.61648226e+00 -1.07675540e+00 -4.33828056e-01 -3.44869137e-01
5.48161566e-03 2.24560887e-01 -1.31264949e+00 -5.73739648e-01
-1.48064271e-01 3.74146581e-01 -7.15912044e-01 -1.99694917e-01
-3.83416623e-01 -2.54696220e-01 4.65070695e-01 -1.04188353e-01
4.85584229e-01 -8.30232620e-01 -9.28106844e-01 1.46168128e-01
5.81955388e-02 -1.32761431e+00 -2.91556478e-01 -6.89594090e-01
-8.22045743e-01 -1.29096997e+00 -1.11396313e+00 -7.34501600e-01
1.09333503e+00 -3.66805457e-02 9.26340997e-01 1.42910615e-01
-6.33768857e-01 1.44532062e-02 -1.57800227e-01 -2.24895984e-01
-1.06483197e+00 -3.57289135e-01 -8.17574039e-02 7.83241987e-01
-1.37875080e-01 -7.06214011e-01 -8.05503249e-01 2.45305255e-01
-1.16008067e+00 1.96916267e-01 7.36035883e-01 7.54614532e-01
8.02598059e-01 1.04369156e-01 2.65433252e-01 -8.79359365e-01
4.86692578e-01 -6.93781078e-02 -1.02795482e+00 2.31452242e-01
-4.02059913e-01 9.95706767e-02 7.22075999e-01 -5.21233857e-01
-7.94416666e-01 2.82917976e-01 3.05862278e-01 -4.84328568e-01
-1.94981724e-01 1.80571061e-02 9.72255170e-02 -5.96226394e-01
8.79066050e-01 2.91716248e-01 6.59591377e-01 -2.73510516e-01
6.84757054e-01 4.78581786e-01 7.71751881e-01 -1.47101566e-01
1.12802088e+00 9.41798329e-01 2.32811511e-01 -4.98954535e-01
-6.44001842e-01 1.59211800e-01 -5.94046950e-01 -1.60258308e-01
9.48135138e-01 -5.86249650e-01 -6.43110156e-01 6.49721980e-01
-1.05928731e+00 -4.37230349e-01 -7.00159132e-01 5.23966908e-01
-7.40034282e-01 3.05067122e-01 -2.65435487e-01 -4.44299281e-01
1.32963777e-01 -1.21875727e+00 8.87601376e-01 5.29463589e-01
9.03726891e-02 -9.29222524e-01 5.67326248e-02 2.49416143e-01
3.76651734e-01 6.69440389e-01 1.06695020e+00 -1.49730574e-02
-9.75803018e-01 -2.31191695e-01 -2.49045089e-01 8.19323778e-01
3.12434644e-01 2.42297351e-01 -8.05448472e-01 -2.13695690e-01
5.65747954e-02 3.25497799e-02 5.40463567e-01 5.20888865e-01
8.36948693e-01 -3.58115137e-01 -7.31381774e-02 8.41657639e-01
1.46050763e+00 1.09434448e-01 1.06241035e+00 1.13450423e-01
3.26100916e-01 8.53127182e-01 1.37269959e-01 2.14958992e-02
-9.17898938e-02 7.07302451e-01 2.55457669e-01 -6.29150569e-01
-7.89145708e-01 -2.12480336e-01 1.48421928e-01 -5.20748459e-02
-1.65785313e-01 9.93121974e-03 -5.49595892e-01 4.29266810e-01
-1.23105764e+00 -7.16180265e-01 -1.39974967e-01 2.37431407e+00
1.11748958e+00 7.12394267e-02 1.63426310e-01 -1.08619206e-01
6.67245328e-01 -2.31422961e-01 -5.38119972e-01 -1.84008121e-01
-7.88298398e-02 5.27676404e-01 5.54929733e-01 5.96713901e-01
-8.17609549e-01 8.93564522e-01 6.56092882e+00 3.18626493e-01
-1.47655249e+00 -1.81251958e-01 7.59819090e-01 1.67956613e-02
-3.23068589e-01 2.17650339e-01 -3.28634202e-01 5.16089201e-01
7.71005392e-01 -3.37689221e-01 4.59799886e-01 2.64144480e-01
2.85653204e-01 -9.27497000e-02 -9.21752989e-01 7.58739829e-01
2.25309245e-02 -1.35113001e+00 2.83434927e-01 2.06057191e-01
9.10767615e-01 -5.89253783e-01 2.98233837e-01 -6.29442334e-01
6.32784218e-02 -1.22419345e+00 4.89955783e-01 1.06880581e+00
1.15175140e+00 -3.95973563e-01 5.96057296e-01 -1.06362678e-01
-4.68509316e-01 2.84397811e-01 -1.23262279e-01 3.85109037e-01
2.09899202e-01 4.62322026e-01 -8.76108170e-01 2.57807910e-01
3.28515768e-01 4.24592882e-01 -7.36864984e-01 1.37007570e+00
-4.14660096e-01 3.83651406e-01 2.53484324e-02 4.48414177e-01
-1.69788584e-01 -5.66052914e-01 7.47073293e-01 2.73194134e-01
3.31531763e-01 -3.25308412e-01 -3.26491863e-01 1.52757835e+00
-1.10593297e-01 1.43650174e-01 -7.89622962e-01 -1.06585935e-01
1.21674038e-01 1.18949139e+00 -5.34868300e-01 -2.93444604e-01
-2.93063492e-01 9.42500532e-01 -1.75650239e-01 5.23336351e-01
-6.00888968e-01 -4.32490975e-01 5.46706140e-01 4.25940633e-01
1.99759066e-01 8.42114612e-02 -1.40904635e-01 -1.04259765e+00
1.85537592e-01 -8.37097168e-01 -4.72150743e-01 -1.00938439e+00
-9.07312870e-01 9.06412125e-01 -2.88889945e-01 -1.60064840e+00
-3.12693655e-01 -3.84192199e-01 -5.39621413e-01 1.10087764e+00
-1.88672018e+00 -1.13192761e+00 -4.30808038e-01 5.37631869e-01
-1.27823263e-01 -4.09151316e-01 6.90443635e-01 4.14644070e-02
-4.02006716e-01 6.46147549e-01 -4.97069098e-02 6.36123642e-02
7.16564059e-01 -1.15785432e+00 4.06164408e-01 1.00304294e+00
2.23554391e-02 5.04300714e-01 7.65962481e-01 -3.70446563e-01
-7.68370688e-01 -1.11439621e+00 5.72582841e-01 -3.22608113e-01
5.05306304e-01 1.20914057e-01 -7.62088716e-01 5.28538346e-01
3.28612864e-01 3.35023075e-01 4.23845321e-01 -8.43961000e-01
-2.64322042e-01 -1.71877056e-01 -1.50064099e+00 8.08207214e-01
5.60106516e-01 -3.38730037e-01 -4.24775124e-01 2.18171269e-01
5.16119480e-01 -3.77771527e-01 -9.43492770e-01 2.55417395e-02
6.26123846e-01 -1.16836321e+00 9.29798961e-01 -7.25697935e-01
6.02658033e-01 -5.84463239e-01 3.05921882e-01 -1.42099345e+00
2.63480723e-01 -9.89992917e-01 2.96925902e-01 9.44232166e-01
3.83885115e-01 -8.94340754e-01 6.84471846e-01 4.68381792e-01
2.00014949e-01 -4.55023974e-01 -8.08237135e-01 -7.94788301e-01
8.55551586e-02 4.07754391e-01 5.67360520e-01 5.87338269e-01
-4.64552492e-01 8.15751404e-02 -2.24729195e-01 4.25324850e-02
6.90756202e-01 2.30830505e-01 1.05648279e+00 -1.15805507e+00
-4.37094748e-01 -4.48817998e-01 -7.99884200e-01 -6.37701571e-01
6.63630962e-02 -7.71090448e-01 -1.07369103e-01 -1.11483383e+00
-1.12157635e-01 -4.46560591e-01 1.06835686e-01 2.29077652e-01
2.65400801e-02 7.50356019e-01 1.08192660e-01 3.20048124e-01
2.96795756e-01 2.02848002e-01 1.94617009e+00 4.34389934e-02
-3.43199283e-01 1.52573898e-01 -7.22266912e-01 5.85167050e-01
8.78747344e-01 -3.77376854e-01 -5.02482653e-01 -8.79862159e-02
9.10619721e-02 1.39066055e-01 6.77136719e-01 -1.05054498e+00
-1.59681022e-01 1.19431719e-01 1.47159323e-01 3.14355284e-01
1.59671634e-01 -7.16502845e-01 2.47012720e-01 6.80919409e-01
-3.09225470e-01 -4.57167894e-01 4.42294627e-02 4.49379295e-01
-4.51337457e-01 -2.84815907e-01 1.29424882e+00 -1.14081964e-01
-1.55044466e-01 1.15899280e-01 -2.00057045e-01 -7.64516070e-02
1.30962110e+00 -4.19179618e-01 -2.56528676e-01 -3.38220954e-01
-8.85665894e-01 -4.11683530e-01 9.09153044e-01 8.91811028e-02
4.87550318e-01 -9.34612989e-01 -9.11997557e-01 5.23873687e-01
1.14914231e-01 -1.03655957e-01 9.94746909e-02 1.07466972e+00
-9.26780105e-01 1.15878761e-01 -5.97077370e-01 -4.68314677e-01
-1.16332018e+00 5.29706120e-01 1.00186157e+00 1.54896721e-01
-6.12493277e-01 5.07353067e-01 2.38662601e-01 3.81897837e-02
-1.00740083e-01 -4.98470962e-01 -1.36455789e-01 -3.18870366e-01
5.81391275e-01 -1.56836540e-01 -1.18142918e-01 -6.00711882e-01
3.26693118e-01 9.94199455e-01 1.86633065e-01 1.87819693e-02
9.45858359e-01 -1.66242316e-01 -2.58587897e-01 -8.92573297e-02
1.13129663e+00 4.92194384e-01 -1.34624577e+00 -2.18407586e-01
-4.66633290e-01 -8.60753059e-01 4.08955254e-02 -8.11103225e-01
-1.28501976e+00 7.95511365e-01 9.86249268e-01 2.70759553e-01
1.21354163e+00 -9.55556557e-02 7.93808460e-01 -2.33231917e-01
2.63953120e-01 -3.44963759e-01 -3.24861892e-02 -3.09470385e-01
8.54876041e-01 -1.04534853e+00 -2.63562888e-01 -7.09675550e-01
-3.33992779e-01 1.13456190e+00 1.33913845e-01 -3.04064929e-01
4.24249500e-01 -1.29406184e-01 4.89368677e-01 -2.68678404e-02
-1.36331007e-01 -4.28592652e-01 4.87238824e-01 7.82521307e-01
5.68432137e-02 -6.73215538e-02 -5.35286725e-01 -1.99811868e-02
-3.78614157e-01 3.75551194e-01 1.05277932e+00 3.65993828e-01
-1.04849949e-01 -1.51268613e+00 -2.38209382e-01 1.13113590e-01
-5.27434051e-01 -1.82171669e-02 -4.43686604e-01 8.15312028e-01
1.94927171e-01 7.04939306e-01 -1.64513979e-02 2.57706307e-02
2.64248163e-01 -3.50265354e-02 8.52952361e-01 -6.13688409e-01
-2.90764630e-01 2.96467985e-03 -1.87220275e-01 -5.74329197e-01
-5.00715911e-01 -4.95542139e-01 -9.97209609e-01 1.07638808e-02
4.88305464e-02 -1.53549373e-01 8.00030351e-01 6.02895916e-01
2.27743134e-01 6.07856810e-01 7.97241688e-01 -5.35534620e-01
-4.90824163e-01 -5.15680194e-01 -7.55526781e-01 5.33152819e-01
7.09242940e-01 -5.14204025e-01 -4.18707281e-01 6.02856934e-01]
|
[15.590458869934082, -3.7630789279937744]
|
32a38869-fd90-4f09-8c76-445a927ac35d
|
a-continual-learning-framework-for-adaptive
|
2203.08796
| null |
https://arxiv.org/abs/2203.08796v1
|
https://arxiv.org/pdf/2203.08796v1.pdf
|
A Continual Learning Framework for Adaptive Defect Classification and Inspection
|
Machine-vision-based defect classification techniques have been widely adopted for automatic quality inspection in manufacturing processes. This article describes a general framework for classifying defects from high volume data batches with efficient inspection of unlabelled samples. The concept is to construct a detector to identify new defect types, send them to the inspection station for labelling, and dynamically update the classifier in an efficient manner that reduces both storage and computational needs imposed by data samples of previously observed batches. Both a simulation study on image classification and a case study on surface defect detection via 3D point clouds are performed to demonstrate the effectiveness of the proposed method.
|
['Tzyy-Shuh Chang', 'Judy Jin', 'Raed Al Kontar', 'Wenbo Sun']
|
2022-03-16
| null | null | null | null |
['defect-detection']
|
['computer-vision']
|
[ 2.17714369e-01 2.56237257e-02 3.87610972e-01 -5.10502636e-01
-2.04257354e-01 -2.00792938e-01 3.88673171e-02 8.56644273e-01
-1.58890989e-02 9.04542729e-02 -8.15557957e-01 -1.41582534e-01
-2.28718847e-01 -9.99944448e-01 -1.72096878e-01 -7.51274109e-01
7.83272758e-02 9.70549405e-01 4.60642457e-01 -9.64999869e-02
7.52429664e-01 1.02857554e+00 -1.90253139e+00 2.79788852e-01
8.08770955e-01 1.12641990e+00 6.75701976e-01 6.55909598e-01
-7.50012025e-02 1.66015923e-01 -9.29484367e-01 1.38874963e-01
3.13495874e-01 -2.21554458e-01 -8.13230395e-01 1.32872081e+00
-1.84678197e-01 -2.32351840e-01 2.19850317e-01 9.74681854e-01
1.39085680e-01 1.82137806e-02 8.77220690e-01 -9.09922838e-01
-2.98261851e-01 -4.42549527e-01 -4.62222308e-01 2.62059897e-01
2.26171628e-01 4.66021150e-02 4.38629210e-01 -9.60876465e-01
4.39211994e-01 1.09173381e+00 4.97235358e-01 6.02908917e-02
-7.77097046e-01 1.13865480e-01 -2.31973067e-01 1.56171039e-01
-9.53783989e-01 -4.56778519e-02 8.68398070e-01 -6.62890196e-01
9.08093452e-01 2.25133672e-01 9.16782558e-01 -2.83806920e-01
4.20008183e-01 6.73030257e-01 9.46307600e-01 -8.71260107e-01
5.34043372e-01 1.61333412e-01 4.86273408e-01 1.05189455e+00
5.50410926e-01 7.38390088e-02 -3.60569805e-02 -8.78140107e-02
1.02918601e+00 2.44928882e-01 1.43302456e-01 -6.47451043e-01
-4.35445130e-01 1.07352114e+00 4.12182737e-04 2.53966987e-01
-6.42430305e-01 -6.04983449e-01 7.21855104e-01 7.03056872e-01
7.47402251e-01 5.19436777e-01 -7.16274261e-01 2.14268044e-01
-4.31126803e-01 -1.57830670e-01 7.00315237e-01 9.09986258e-01
7.08335161e-01 -2.87600737e-02 3.39377671e-01 9.73727882e-01
5.08328438e-01 2.92245209e-01 9.02081933e-03 -6.93269551e-01
-1.24762043e-01 9.40834463e-01 9.59466994e-02 -8.44521940e-01
-3.39121759e-01 -3.12737077e-02 -5.58078885e-01 7.22133398e-01
-8.44829157e-02 1.14715517e-01 -1.20317698e+00 -4.06818688e-02
5.42888641e-01 -5.40929675e-01 2.92829927e-02 6.52995884e-01
4.65584010e-01 5.35541832e-01 -4.18914407e-01 -3.45841885e-01
1.28565550e+00 -8.27315211e-01 -6.59426928e-01 2.89450474e-02
7.64106750e-01 -8.46688449e-01 6.20017111e-01 1.00429702e+00
-9.48887348e-01 -9.48241353e-01 -9.06665087e-01 3.75755072e-01
-1.71030015e-01 5.70871532e-01 7.10075974e-01 5.64243495e-01
-8.43987107e-01 5.88247955e-01 -1.00241220e+00 -4.89729106e-01
3.73040318e-01 4.57119495e-01 -3.21836352e-01 -3.29135954e-01
-3.73959720e-01 8.29832792e-01 3.84044826e-01 5.47205031e-01
-9.94863629e-01 -1.41795516e-01 -9.88799989e-01 -3.07957262e-01
4.34530387e-03 -1.13706410e-01 1.26681149e+00 -5.18381953e-01
-1.25919330e+00 9.64104056e-01 -5.99623509e-02 -5.10745682e-02
-2.58087851e-02 1.11833811e-02 -3.09398890e-01 4.29979742e-01
1.79824546e-01 -7.10283741e-02 8.78610551e-01 -1.48828149e+00
-1.23651445e+00 -5.63172162e-01 -1.73547924e-01 -1.45777434e-01
6.24387600e-02 3.22545022e-02 -3.99430007e-01 -5.36686331e-02
8.42237294e-01 -5.26028633e-01 -3.58905762e-01 -1.17265821e-01
-1.05732247e-01 -5.47415078e-01 1.15056574e+00 -7.07676947e-01
4.26886499e-01 -2.08137131e+00 -2.14851424e-01 4.31444675e-01
-1.67178825e-01 4.58778083e-01 2.27639154e-01 4.79584485e-01
-3.78955738e-03 -5.07046938e-01 -2.93546617e-01 -2.65920252e-01
-4.79298443e-01 5.45087516e-01 4.12821352e-01 6.11400425e-01
4.62464184e-01 1.12319648e-01 -7.20559061e-01 -3.90591145e-01
7.22593844e-01 -2.36474991e-01 -2.79951900e-01 2.50070214e-01
1.36769876e-01 5.41027904e-01 -5.09801626e-01 1.16565847e+00
8.35876465e-01 1.06614344e-01 6.27497286e-02 -2.04683900e-01
-1.96407184e-01 -1.77130774e-01 -1.24208367e+00 1.30979657e+00
-5.52187800e-01 1.62755281e-01 6.29984677e-01 -1.63652909e+00
1.67167759e+00 4.48842347e-01 6.66008234e-01 -3.81444216e-01
9.77139995e-02 1.85272992e-01 -3.63950014e-01 -9.81776237e-01
4.05544221e-01 -2.25402787e-01 6.82133660e-02 -3.71639058e-02
5.54368682e-02 -7.36659467e-01 4.33136672e-01 -3.67746830e-01
1.11238265e+00 -2.33859360e-01 5.24656549e-02 -3.24262053e-01
5.69137216e-01 6.21031821e-01 4.55953658e-01 5.15001535e-01
-3.83808501e-02 2.32841954e-01 9.97103602e-02 -7.34858036e-01
-1.03267455e+00 -8.49893451e-01 -2.67956823e-01 4.12173092e-01
3.89394850e-01 3.89241576e-01 -4.27009702e-01 -7.89426446e-01
3.23859364e-01 2.70080924e-01 -2.50026852e-01 -8.12571272e-02
-1.96708888e-01 -5.48976481e-01 -4.20759350e-01 3.16168725e-01
2.81441927e-01 -1.08543122e+00 -8.04481626e-01 4.31405544e-01
4.89843100e-01 -6.65685177e-01 3.71204138e-01 2.29118526e-01
-1.48406136e+00 -1.55613482e+00 -2.40886196e-01 -1.48883164e+00
1.45576179e+00 4.48143005e-01 1.01936769e+00 6.14067793e-01
-8.95416141e-01 5.15923917e-01 -8.17854345e-01 -7.54872203e-01
-9.45937514e-01 -5.45768917e-01 -1.19043468e-02 -7.32984990e-02
4.72229004e-01 1.12970695e-01 -3.47739458e-01 4.32970017e-01
-8.07787478e-01 -5.30028701e-01 5.82067907e-01 9.73992586e-01
6.86757624e-01 9.55382943e-01 5.09572208e-01 -8.34684312e-01
4.54989702e-01 -1.17495805e-01 -9.65504050e-01 2.08006427e-01
-6.70921326e-01 -5.19889295e-01 2.11965814e-01 2.65416056e-01
-1.11977744e+00 2.17961669e-01 -2.79406428e-01 -2.04128578e-01
-6.87795758e-01 4.73136991e-01 1.84612516e-02 -1.73936218e-01
3.18857670e-01 3.19232233e-02 4.66550052e-01 -1.01562965e+00
-2.81169534e-01 1.12874401e+00 4.16422874e-01 -2.40125239e-01
6.05222762e-01 5.62360644e-01 9.58536416e-02 -1.12287247e+00
-4.92436469e-01 -1.15685189e+00 -1.12592936e+00 -3.12685370e-01
7.17827737e-01 -6.59816325e-01 -5.22268176e-01 7.80942976e-01
-1.20652401e+00 2.45291665e-01 -5.53534865e-01 5.63008547e-01
-3.42821896e-01 6.40351653e-01 -8.91719103e-01 -1.07674396e+00
-4.24460918e-01 -1.02157116e+00 1.18355596e+00 4.21076734e-03
3.50713819e-01 -1.12679398e+00 -2.02947587e-01 3.40183020e-01
-1.79678187e-01 1.11223876e-01 9.95518446e-01 -5.54377198e-01
-3.10851187e-01 -8.25280190e-01 5.07441983e-02 8.87220144e-01
8.16522837e-01 2.38563225e-01 -5.30477166e-01 -5.85178316e-01
6.08360887e-01 -7.10375374e-03 5.98390043e-01 3.69110644e-01
8.26198936e-01 3.38272452e-01 -5.05783498e-01 -1.43353477e-01
1.54116797e+00 8.90345991e-01 3.86538446e-01 1.96498409e-01
3.67237955e-01 8.64708722e-01 1.59648156e+00 6.64303780e-01
-1.44225314e-01 2.46055603e-01 7.07189739e-01 -4.78247970e-01
4.72091138e-01 4.72867936e-01 -1.94064915e-01 1.13110626e+00
-8.94003212e-02 -1.14044234e-01 -8.88778269e-01 8.30135167e-01
-1.50685775e+00 -4.83846188e-01 -5.98134398e-01 1.91209805e+00
3.29632103e-01 2.18323797e-01 -1.98035568e-01 8.92984331e-01
7.76925743e-01 -7.28346884e-01 -6.27951249e-02 -8.64438951e-01
5.50117254e-01 4.39345390e-01 3.18675876e-01 3.51283044e-01
-1.24200118e+00 4.02470917e-01 6.64370632e+00 3.52901280e-01
-5.64831734e-01 9.01146233e-02 2.66791075e-01 5.20188212e-01
3.19107622e-01 -1.15713775e-01 -6.16621256e-01 7.51623213e-02
3.50678921e-01 4.83919233e-01 -3.16349715e-01 1.00961721e+00
2.63638854e-01 -7.33842194e-01 -1.04734528e+00 8.33940566e-01
-5.02038635e-02 -1.25891888e+00 -9.50918943e-02 -3.85012142e-02
6.15116179e-01 -2.73519367e-01 -5.28453112e-01 -9.78765041e-02
1.82346627e-01 -2.10497782e-01 3.71674776e-01 5.46418548e-01
3.28766018e-01 -7.18943477e-01 1.26219523e+00 2.24336326e-01
-9.51247990e-01 -3.86332035e-01 -7.93430150e-01 -2.19785467e-01
3.13555062e-01 1.09421372e+00 -1.17551541e+00 8.68605375e-01
7.99654961e-01 5.43581188e-01 -3.58986467e-01 1.43463600e+00
-3.73398773e-02 4.01767731e-01 -1.16720349e-01 1.66170388e-01
1.81923315e-01 -5.41795850e-01 2.70994544e-01 9.07196045e-01
3.88704181e-01 1.48343081e-02 6.17094338e-01 2.93578476e-01
7.64652550e-01 -1.83067292e-01 -7.04642415e-01 3.29104096e-01
2.01528743e-01 1.12735736e+00 -1.12560034e+00 -2.82824725e-01
-4.54566360e-01 9.91107047e-01 -2.28691787e-01 -2.14834154e-01
-2.08446652e-01 -6.49861336e-01 4.00066555e-01 2.13587642e-01
7.02158093e-01 -4.33578193e-01 -1.78575292e-01 -2.34749034e-01
-4.18868549e-02 -3.61617923e-01 2.99772650e-01 -5.14975905e-01
-1.40352964e+00 3.03102404e-01 -1.62772660e-03 -1.36332750e+00
7.08819330e-02 -9.48123872e-01 -5.97116709e-01 6.57819510e-01
-1.26092708e+00 -7.58438170e-01 -3.36040944e-01 3.31953019e-01
9.84099329e-01 -3.55913311e-01 7.49347329e-01 2.22832918e-01
-5.35173297e-01 -1.29857942e-01 3.73677224e-01 9.29740965e-02
-1.10993408e-01 -1.32235205e+00 2.46139213e-01 7.62461245e-01
-2.89420605e-01 -3.96127738e-02 7.43474245e-01 -9.12519991e-01
-1.47151434e+00 -9.99744773e-01 4.94375020e-01 -1.12168454e-01
2.23354310e-01 2.13576462e-02 -1.01004171e+00 2.16851443e-01
-1.21294729e-01 1.08070724e-01 2.07224458e-01 -2.72472411e-01
6.38055623e-01 -2.46998414e-01 -1.67172587e+00 -2.92161405e-01
5.45866311e-01 -8.99475068e-02 -5.72797418e-01 7.83249140e-01
1.36245772e-01 -2.97947794e-01 -1.26051331e+00 5.66420794e-01
-1.61485180e-01 -7.99582243e-01 4.43888485e-01 -7.94409141e-02
-2.43929774e-02 -4.29643154e-01 3.61931294e-01 -1.35793459e+00
-3.69659662e-01 -1.14151686e-01 4.65303659e-01 1.03135717e+00
2.86372174e-02 -6.11573040e-01 9.40746605e-01 6.08225353e-02
-7.93134212e-01 -6.56654477e-01 -1.00136733e+00 -8.32753897e-01
-4.99757141e-01 -1.26027107e-01 1.93312198e-01 5.45112967e-01
-1.56679884e-01 3.31977233e-02 3.38493943e-01 6.82309628e-01
6.21244490e-01 6.68985248e-02 3.63782644e-01 -1.75186884e+00
3.95109020e-02 3.66810650e-01 -1.04411066e+00 -6.71304405e-01
-3.20699185e-01 -4.92502421e-01 5.01277506e-01 -2.11302590e+00
-3.21596235e-01 -8.20419908e-01 -3.18517797e-02 2.42113739e-01
2.73895919e-01 -3.58712114e-02 -2.89532840e-01 2.66926378e-01
-3.01203400e-01 2.61420071e-01 1.40220213e+00 -2.52103537e-01
-1.49875015e-01 6.31508470e-01 8.16596523e-02 6.10819340e-01
7.29848206e-01 -3.87339175e-01 -3.83056194e-01 -4.32551771e-01
-4.42235976e-01 2.39097521e-01 2.11349830e-01 -1.12185943e+00
-1.41731072e-02 1.34723917e-01 2.55406916e-01 -1.06480086e+00
2.31495112e-01 -1.30978668e+00 -5.17259426e-02 1.02415764e+00
3.36806238e-01 4.61654067e-02 9.01959240e-02 7.67330706e-01
-6.56894565e-01 -9.89423156e-01 9.04461741e-01 -3.33126634e-01
-9.11152422e-01 1.05942942e-01 -6.47610188e-01 -9.37948525e-01
1.49050975e+00 -6.51179373e-01 1.41533300e-01 3.72653574e-01
-1.07563806e+00 1.21079177e-01 4.97815043e-01 2.42606685e-01
9.48711693e-01 -8.70017767e-01 -5.40471077e-01 7.85187066e-01
2.65277982e-01 3.39398682e-01 1.84183657e-01 5.44563174e-01
-1.08642268e+00 4.56284404e-01 -2.80166060e-01 -1.06803918e+00
-1.48055220e+00 7.52542436e-01 2.40829021e-01 -1.49592131e-01
-6.91819429e-01 9.85599756e-01 -2.96687782e-01 -3.94764185e-01
-1.61396459e-01 -6.69264257e-01 -2.90350974e-01 -1.77174155e-02
1.71316713e-01 6.19101346e-01 9.21235621e-01 -3.22316021e-01
-2.29681209e-01 6.14936233e-01 -2.56360825e-02 3.56672317e-01
1.30294967e+00 -9.74869207e-02 -4.16313231e-01 4.00565118e-01
9.38993335e-01 -5.26491284e-01 -1.19987404e+00 -9.61380452e-02
3.43669653e-01 -7.65072107e-01 1.62350476e-01 -4.16934788e-01
-1.35744083e+00 6.74729645e-01 1.04408443e+00 8.79556417e-01
1.25641990e+00 2.95738935e-01 5.10716558e-01 3.58053952e-01
7.87345767e-01 -1.34627998e+00 1.42203480e-01 3.33961368e-01
9.27631199e-01 -1.29351628e+00 -1.61730591e-02 -9.06703115e-01
-6.44099772e-01 1.29553127e+00 5.59421599e-01 -2.26948038e-01
9.65431452e-01 2.10327566e-01 1.50079787e-01 -8.25812519e-01
-5.09733498e-01 -1.14953369e-01 -2.26750821e-01 1.12619269e+00
-4.81953546e-02 -1.53006867e-01 -2.58414954e-01 -2.48719990e-01
5.40050328e-01 -5.32239527e-02 5.83577633e-01 1.80766177e+00
-1.01287758e+00 -1.08347583e+00 -7.65868962e-01 9.31965828e-01
-2.89108813e-01 6.85201049e-01 1.44717425e-01 7.23000467e-01
4.32682157e-01 1.45200157e+00 7.75477648e-01 -2.20341787e-01
8.21252704e-01 -1.79715753e-01 5.21078289e-01 -1.26958823e+00
-5.23933351e-01 2.18949839e-02 1.88159972e-01 -6.01425469e-02
-4.98004526e-01 -8.28708589e-01 -1.36099946e+00 3.33667278e-01
-8.29286337e-01 5.72494209e-01 1.15262890e+00 8.76228571e-01
1.63726777e-01 5.57482541e-01 1.31154275e+00 -8.97860646e-01
-4.98904556e-01 -1.13051093e+00 -1.31131387e+00 2.76588559e-01
1.48701876e-01 -1.01196265e+00 -2.69280761e-01 2.18880668e-01]
|
[7.375646591186523, 1.8807308673858643]
|
f521d68f-0474-4f27-b7eb-46c603029cba
|
rbsr-efficient-and-flexible-recurrent-network
|
2306.17595
| null |
https://arxiv.org/abs/2306.17595v1
|
https://arxiv.org/pdf/2306.17595v1.pdf
|
RBSR: Efficient and Flexible Recurrent Network for Burst Super-Resolution
|
Burst super-resolution (BurstSR) aims at reconstructing a high-resolution (HR) image from a sequence of low-resolution (LR) and noisy images, which is conducive to enhancing the imaging effects of smartphones with limited sensors. The main challenge of BurstSR is to effectively combine the complementary information from input frames, while existing methods still struggle with it. In this paper, we suggest fusing cues frame-by-frame with an efficient and flexible recurrent network. In particular, we emphasize the role of the base-frame and utilize it as a key prompt to guide the knowledge acquisition from other frames in every recurrence. Moreover, we introduce an implicit weighting loss to improve the model's flexibility in facing input frames with variable numbers. Extensive experiments on both synthetic and real-world datasets demonstrate that our method achieves better results than state-of-the-art ones. Codes and pre-trained models are available at https://github.com/ZcsrenlongZ/RBSR.
|
['WangMeng Zuo', 'Hongzhi Zhang', 'Shuohao Zhang', 'Zhilu Zhang', 'Renlong Wu']
|
2023-06-30
| null | null | null | null |
['super-resolution']
|
['computer-vision']
|
[ 4.19031799e-01 -2.95655906e-01 -2.13229284e-01 -1.40750960e-01
-9.48676348e-01 -6.47569969e-02 1.56346574e-01 -5.80097854e-01
-1.51005834e-01 7.98254728e-01 3.27085823e-01 2.33149789e-02
-5.30572757e-02 -5.08040309e-01 -6.41221881e-01 -8.28414202e-01
1.97872773e-01 -3.92980784e-01 3.59218597e-01 -2.27499664e-01
1.49351750e-02 3.26375753e-01 -1.57929885e+00 4.59598541e-01
7.81566918e-01 1.05414701e+00 9.54274237e-01 5.31329572e-01
3.78857911e-01 1.22138011e+00 -1.71429455e-01 -7.33130006e-03
1.29631773e-01 -4.00447488e-01 -5.49575508e-01 2.27861702e-01
9.17544216e-02 -7.18157053e-01 -7.16645896e-01 9.73305941e-01
6.11408949e-01 2.54844069e-01 -2.60269325e-02 -5.68675995e-01
-5.01203775e-01 4.12462384e-01 -7.14443088e-01 6.59443617e-01
5.69179177e-01 1.52330041e-01 5.34137249e-01 -1.11754429e+00
6.92702889e-01 1.01328874e+00 4.92864609e-01 5.38893819e-01
-9.38649237e-01 -7.86295831e-01 1.84599221e-01 4.22478884e-01
-1.36501193e+00 -9.23314691e-01 8.24736059e-01 -5.07226586e-02
4.36045587e-01 2.95302689e-01 4.47767109e-01 1.34175801e+00
-1.61125764e-01 6.91337287e-01 9.99806345e-01 -1.62749901e-01
-2.23410353e-01 -2.12876171e-01 -4.45997082e-02 4.28364456e-01
-6.12637997e-02 1.87672928e-01 -8.62492621e-01 1.32111147e-01
1.37300134e+00 3.38724196e-01 -7.29806900e-01 3.43041390e-01
-1.26312804e+00 3.94761831e-01 2.42687404e-01 4.52833354e-01
-5.64139366e-01 4.51638103e-02 7.51751959e-02 4.18918468e-02
6.03923380e-01 7.46177286e-02 -1.25194356e-01 -1.13917366e-01
-8.22671533e-01 -2.61538313e-03 9.27402154e-02 8.34049940e-01
5.52723646e-01 2.18005076e-01 -3.50090206e-01 1.11961603e+00
-2.56965775e-02 3.26659858e-01 3.82250339e-01 -1.44570756e+00
3.95045310e-01 -6.92711994e-02 4.31445867e-01 -9.22926962e-01
-1.10468045e-01 -4.94471490e-01 -1.22639751e+00 -2.92772114e-01
5.27389795e-02 -4.61917333e-02 -6.91512644e-01 1.52677298e+00
3.65095198e-01 6.80742621e-01 -8.01796094e-02 1.21174121e+00
9.86159027e-01 9.82854962e-01 -3.80037785e-01 -7.44588435e-01
1.13108170e+00 -9.98664141e-01 -1.08168161e+00 -1.26784906e-01
-2.41263971e-01 -8.14501703e-01 8.39245260e-01 5.80114603e-01
-1.35674310e+00 -8.87940645e-01 -8.74157965e-01 -3.34347069e-01
2.81060278e-01 3.28919083e-01 4.41960990e-01 -1.01462556e-02
-1.08468509e+00 5.98937988e-01 -9.32599306e-01 1.58952445e-01
4.16043401e-01 -6.61138864e-03 -1.54391274e-01 -4.49242651e-01
-1.16873813e+00 5.95401883e-01 1.39411464e-01 3.02110076e-01
-8.87885153e-01 -5.76142609e-01 -7.30900049e-01 -2.61176646e-01
7.92009652e-01 -4.69766021e-01 1.30230272e+00 -6.09971941e-01
-1.76429200e+00 4.92492884e-01 -6.84613764e-01 -4.12732184e-01
3.70670825e-01 -3.85590971e-01 -5.46534598e-01 5.94996393e-01
-1.46224070e-02 3.76533329e-01 1.13677037e+00 -1.40360141e+00
-7.89576173e-01 -1.13103777e-01 1.33264393e-01 2.84116626e-01
-1.62204042e-01 9.33070332e-02 -8.40014398e-01 -7.70641565e-01
2.56680787e-01 -6.78548276e-01 -3.65506500e-01 -2.15274572e-01
-2.45415360e-01 7.50306845e-02 7.31577337e-01 -9.16057885e-01
1.38845181e+00 -2.24534750e+00 1.43594891e-01 -3.16445976e-01
5.00810623e-01 3.80472481e-01 -6.75087497e-02 1.47043779e-01
-1.37693107e-01 -1.99514285e-01 1.13543477e-02 -5.30456424e-01
-6.66958809e-01 1.53224155e-01 -5.62875092e-01 5.05115092e-01
9.81905609e-02 7.17145860e-01 -9.47569191e-01 -4.63000923e-01
3.97567630e-01 9.86974716e-01 -1.92948490e-01 3.98592651e-01
5.70952818e-02 1.00806570e+00 -5.57030201e-01 5.73264956e-01
6.36939049e-01 -7.32710600e-01 1.41899347e-01 -4.97020453e-01
-3.73431265e-01 3.48375469e-01 -1.15793490e+00 1.80160844e+00
-4.74474400e-01 5.27035475e-01 9.77863595e-02 -7.47840583e-01
8.00796390e-01 2.48618275e-01 5.10912538e-01 -1.05908561e+00
-8.81182775e-02 9.52251852e-02 -2.93831468e-01 -5.29157698e-01
7.39953876e-01 1.94906682e-01 4.59696829e-01 2.18421534e-01
-2.08812222e-01 2.94130862e-01 2.40478247e-01 1.19452849e-01
7.94839144e-01 1.66709244e-01 3.21198046e-01 3.51832092e-01
5.29771447e-01 -6.72094047e-01 9.43666816e-01 8.06890547e-01
8.86840001e-03 1.13688350e+00 3.31042558e-02 -3.60306382e-01
-1.07692420e+00 -8.47737968e-01 -1.41064167e-01 8.50569129e-01
4.16023523e-01 -3.94795924e-01 -4.55358386e-01 -5.32283969e-02
-6.02119148e-01 4.84353364e-01 -3.84503424e-01 1.54182449e-01
-8.00357223e-01 -6.15542531e-01 2.50739437e-02 3.34276140e-01
6.13473713e-01 -8.42510223e-01 -7.78701067e-01 3.06066573e-01
-9.26910758e-01 -1.54533362e+00 -5.00952125e-01 -1.90703019e-01
-7.74359167e-01 -8.86998713e-01 -8.67577851e-01 -3.49574059e-01
5.81901968e-01 9.69411731e-01 1.06125093e+00 1.94776669e-01
-1.66449040e-01 7.50133321e-02 -6.81976974e-01 1.06023185e-01
-1.14814833e-01 -1.55366167e-01 -3.40066217e-02 2.42657736e-01
-2.41878033e-01 -8.02811563e-01 -8.02212775e-01 5.03252566e-01
-1.01474202e+00 5.28775096e-01 5.32916844e-01 9.26079392e-01
1.08378863e+00 2.00866431e-01 5.14141321e-01 -6.82137132e-01
2.09344044e-01 -4.31446433e-01 -5.42467773e-01 6.17998047e-03
-3.08849335e-01 -3.52996379e-01 6.95988715e-01 -5.24160504e-01
-1.36533380e+00 -9.60897058e-02 -1.30473047e-01 -7.38696337e-01
1.30401030e-01 1.55287266e-01 8.66569579e-02 3.18439887e-03
2.47693062e-01 5.54384291e-01 -1.33716077e-01 -5.55907547e-01
2.28832170e-01 5.89814484e-01 7.17306972e-01 -4.78533775e-01
5.78410268e-01 8.41494501e-01 -1.77413672e-01 -8.64788890e-01
-1.22643065e+00 -5.09525239e-01 -2.17729867e-01 -3.99628758e-01
6.02285385e-01 -1.36762357e+00 -6.06641710e-01 5.42245746e-01
-1.13043392e+00 -4.00377572e-01 -2.65330642e-01 3.98133278e-01
-5.33896863e-01 3.41040790e-01 -9.48416054e-01 -8.91331553e-01
-2.30119780e-01 -1.15004301e+00 1.16287434e+00 5.89555383e-01
3.20862174e-01 -4.16491181e-01 -2.43969768e-01 5.21009684e-01
4.82724488e-01 5.27393408e-02 4.31911014e-02 8.54176208e-02
-1.00183249e+00 2.01817304e-01 -3.76807541e-01 3.77134115e-01
2.38579050e-01 -2.11590707e-01 -9.46824014e-01 -3.07533473e-01
4.13995296e-01 -2.57902831e-01 1.04676628e+00 6.09167159e-01
1.55284166e+00 -2.49642849e-01 -1.17473550e-01 8.95005763e-01
1.39149201e+00 3.18743825e-01 9.84159887e-01 1.37996212e-01
7.67314196e-01 3.19261253e-01 8.57464790e-01 5.70493877e-01
5.21383405e-01 8.82291913e-01 2.97003180e-01 -1.34987697e-01
-4.08161998e-01 -2.10189208e-01 3.48073304e-01 9.53992724e-01
-4.90860999e-01 -1.79585665e-01 -4.74201351e-01 3.71159047e-01
-2.01904082e+00 -1.21660459e+00 3.85459997e-02 2.07197070e+00
1.01824296e+00 -1.50472462e-01 -7.59622380e-02 -1.43397171e-02
8.65135193e-01 6.49580717e-01 -6.71679735e-01 4.93010461e-01
-4.68313754e-01 -7.81779289e-02 3.05921018e-01 5.80043018e-01
-8.62374008e-01 7.94761598e-01 5.58525658e+00 1.20131040e+00
-1.19288146e+00 3.54004353e-01 9.93490100e-01 -4.25124913e-01
-1.39824837e-01 -2.37351313e-01 -9.02083278e-01 7.54527450e-01
7.74070203e-01 -9.47386865e-03 8.69887412e-01 3.43775600e-01
5.79636991e-01 -3.29863667e-01 -6.94812357e-01 1.35988593e+00
2.57491004e-02 -1.55336583e+00 -2.63413101e-01 -1.50775492e-01
7.26620376e-01 2.27745902e-02 2.77033150e-01 -8.41420442e-02
-3.31327086e-03 -1.01055968e+00 6.15437746e-01 9.71964300e-01
1.11611164e+00 -6.89337969e-01 5.69731116e-01 2.63164341e-01
-1.30758142e+00 -1.07580297e-01 -3.76276016e-01 2.97310539e-02
3.95956397e-01 8.80440712e-01 -2.48136118e-01 9.68976319e-01
9.95610535e-01 1.13819492e+00 -3.17388594e-01 6.27895474e-01
-3.21392208e-01 4.30419654e-01 -2.19903767e-01 6.83535337e-01
-1.46145478e-01 -3.58760178e-01 6.52237058e-01 1.00182796e+00
4.43503976e-01 7.25047529e-01 1.15601465e-01 7.56332517e-01
1.12815192e-02 -4.57330376e-01 -5.18225312e-01 1.69604942e-01
7.23599136e-01 1.33887076e+00 -4.65950012e-01 -2.51412392e-01
-4.23303306e-01 9.47087049e-01 9.72449258e-02 5.80661416e-01
-8.08964968e-01 2.70845704e-02 4.51876700e-01 1.99264899e-01
4.69273210e-01 -1.90674484e-01 1.83441360e-02 -1.40866268e+00
3.15698475e-01 -1.03150439e+00 1.14201829e-01 -1.23853540e+00
-1.10308504e+00 6.82517290e-01 -2.12175697e-01 -1.52679515e+00
-3.33089203e-01 3.43607813e-02 -1.54736176e-01 8.40292573e-01
-1.91444981e+00 -1.02421868e+00 -5.12179792e-01 7.34203637e-01
8.31456602e-01 1.51118904e-01 3.56624544e-01 5.07984519e-01
-6.55982971e-01 2.71110475e-01 6.05684705e-02 -2.25778483e-02
6.36290967e-01 -5.84149659e-01 8.84660929e-02 1.25777841e+00
-1.38066441e-01 5.16669631e-01 7.24168241e-01 -5.46897829e-01
-1.45272553e+00 -1.01000214e+00 3.81318241e-01 -2.15258852e-01
3.85974944e-01 -3.59955616e-02 -1.04325569e+00 5.08942425e-01
5.99425985e-03 3.85805339e-01 2.57802665e-01 -2.69706488e-01
-1.91089556e-01 -3.23687851e-01 -8.12807500e-01 5.34575343e-01
1.22187936e+00 -6.49400115e-01 -1.38893157e-01 1.62321091e-01
1.14151776e+00 -9.02384996e-01 -7.10156798e-01 5.18838346e-01
5.01691401e-01 -1.48137712e+00 1.28775954e+00 1.47103280e-01
7.47116387e-01 -5.62623560e-01 -3.48053664e-01 -1.00101113e+00
-2.77933091e-01 -8.84403586e-01 -5.78035712e-01 1.02482712e+00
-2.22937182e-01 -5.61376870e-01 3.94315541e-01 1.69655219e-01
-4.33765650e-02 -9.55526292e-01 -8.20228279e-01 -5.88766932e-01
-7.22365260e-01 -4.53393996e-01 3.83804113e-01 8.33311915e-01
-4.54689711e-01 2.55733490e-01 -9.58526313e-01 4.11536723e-01
8.88973653e-01 1.26907185e-01 5.05526960e-01 -6.82584107e-01
-4.17799890e-01 3.97176780e-02 8.17658901e-02 -1.35391736e+00
-1.26807034e-01 -1.36065006e-01 -4.93424535e-02 -1.32071316e+00
4.00946766e-01 -3.10558319e-01 -2.46277153e-01 2.45422423e-01
-3.16861957e-01 3.76794010e-01 4.10864294e-01 4.78742182e-01
-9.24218893e-01 5.72674572e-01 1.49390018e+00 3.50801915e-01
-1.66059569e-01 -3.98712307e-02 -8.44689012e-01 7.30116487e-01
6.53738737e-01 -2.73641646e-01 -4.75559056e-01 -5.99433303e-01
8.75694901e-02 7.21689939e-01 5.43594718e-01 -7.82819688e-01
4.15654004e-01 -8.53703171e-02 4.26704079e-01 -8.17727029e-01
7.15727568e-01 -6.29973173e-01 5.18752038e-01 5.51982373e-02
-3.19775909e-01 -8.60934108e-02 -1.15496837e-01 5.64465702e-01
-3.49016458e-01 4.77615595e-02 1.00973749e+00 -1.95448771e-01
-6.20114088e-01 5.07073224e-01 1.55014113e-01 -1.08219609e-01
6.35444999e-01 -1.00132719e-01 -4.85110283e-01 -6.76125228e-01
-4.99674976e-01 2.79151440e-01 4.53321338e-01 3.07154447e-01
1.01257050e+00 -1.23258829e+00 -6.94022775e-01 1.18258424e-01
-2.81479567e-01 4.06268567e-01 8.83316338e-01 1.00155210e+00
-1.60518423e-01 2.58230865e-01 -2.55332310e-02 -5.57496428e-01
-1.12036312e+00 3.69105309e-01 2.78228223e-01 -3.86216938e-01
-1.01664352e+00 7.37404168e-01 2.82334149e-01 2.20524788e-01
1.70936644e-01 -9.43199694e-02 -4.24095571e-01 -2.15987340e-02
1.18728244e+00 5.31211734e-01 -2.18393728e-01 -6.02745175e-01
-4.48034629e-02 5.19865990e-01 -3.07597995e-01 -8.99363309e-02
1.64584994e+00 -7.42287517e-01 -2.92734578e-02 5.89194894e-01
8.81890535e-01 1.99978575e-02 -1.74958909e+00 -6.18397653e-01
-6.17883742e-01 -7.42163420e-01 2.29376540e-01 -5.47737062e-01
-1.24335015e+00 5.72140813e-01 5.53517878e-01 9.80760306e-02
1.63585615e+00 -1.11401923e-01 1.04335022e+00 1.25166476e-01
5.87873578e-01 -9.08226013e-01 2.02983782e-01 3.99082035e-01
9.07146513e-01 -1.27173936e+00 1.19888872e-01 -5.08123279e-01
-5.49409509e-01 1.03557003e+00 3.27879548e-01 1.51122594e-02
3.04975033e-01 2.54379869e-01 -4.63225842e-02 4.86882739e-02
-8.62174690e-01 -3.17312241e-01 -4.42107841e-02 5.18353283e-01
1.77467823e-01 -1.89873248e-01 -6.84944317e-02 6.20107293e-01
1.15095094e-01 4.67407256e-01 7.82203138e-01 5.83858430e-01
-2.29561731e-01 -7.83089399e-01 -4.50781763e-01 2.95522183e-01
-6.85610235e-01 -2.49744922e-01 5.10854959e-01 4.55830604e-01
-1.58696268e-02 1.14664459e+00 -1.66529551e-01 -3.81565243e-01
8.91089439e-02 -5.81592381e-01 4.23428893e-01 -3.89186382e-01
2.87043862e-02 5.33022881e-01 4.87694796e-03 -1.06835616e+00
-8.16841543e-01 -6.85968995e-01 -9.73761022e-01 -2.14058176e-01
-4.32863012e-02 -1.73509210e-01 2.44929865e-01 7.20353067e-01
4.70541000e-01 8.38529706e-01 9.58609641e-01 -1.28309381e+00
-2.45488167e-01 -8.39430571e-01 -5.98049104e-01 1.58491865e-01
8.05628240e-01 -4.73783046e-01 -3.91740173e-01 3.69757861e-01]
|
[10.994829177856445, -1.901167869567871]
|
63f24cd7-95ba-4aa0-ace1-d708d895a38f
|
safety-guided-deep-reinforcement-learning-via
|
1903.02526
| null |
http://arxiv.org/abs/1903.02526v2
|
http://arxiv.org/pdf/1903.02526v2.pdf
|
Safety-Guided Deep Reinforcement Learning via Online Gaussian Process Estimation
|
An important facet of reinforcement learning (RL) has to do with how the
agent goes about exploring the environment. Traditional exploration strategies
typically focus on efficiency and ignore safety. However, for practical
applications, ensuring safety of the agent during exploration is crucial since
performing an unsafe action or reaching an unsafe state could result in
irreversible damage to the agent. The main challenge of safe exploration is
that characterizing the unsafe states and actions is difficult for large
continuous state or action spaces and unknown environments. In this paper, we
propose a novel approach to incorporate estimations of safety to guide
exploration and policy search in deep reinforcement learning. By using a cost
function to capture trajectory-based safety, our key idea is to formulate the
state-action value function of this safety cost as a candidate Lyapunov
function and extend control-theoretic results to approximate its derivative
using online Gaussian Process (GP) estimation. We show how to use these
statistical models to guide the agent in unknown environments to obtain
high-performance control policies with provable stability certificates.
|
['Jiameng Fan', 'Wenchao Li']
|
2019-03-06
| null | null | null | null |
['safe-exploration']
|
['robots']
|
[-0.0080493 0.366557 -0.187415 0.27551457 -0.70412034 -0.7719135
0.47170454 0.35689232 -0.6709566 1.1638695 -0.22613555 -0.6040616
-0.47777772 -0.67906135 -0.95873255 -1.0472056 -0.5401658 0.23564652
0.12012979 -0.03522148 0.20802002 0.5458145 -1.0576073 -0.7739802
0.9914283 0.911608 0.05547358 0.74333304 0.4594128 0.76331604
-0.50168896 0.258488 0.5217345 -0.54372895 -0.6718825 -0.09787848
-0.27236333 -0.60324085 -0.32734886 1.5134361 0.3615386 0.56213284
0.48546544 -1.4725536 0.08329851 0.44595474 -0.37044215 -0.06112958
-0.07640023 0.5058298 0.5029991 0.1393793 0.27124527 1.1861914
0.25823295 0.79123443 -1.2491953 -0.5003755 0.44244224 0.02573365
-1.0375578 -0.07465114 0.43594286 -0.30320528 0.47289765 0.06915092
0.8977422 0.9835176 0.684565 0.7119633 1.22458 -0.09449193
0.90250415 -0.07384504 -0.3882155 0.6306763 0.48566344 0.75599545
0.08567626 -0.20906197 0.7137513 0.04554679 -0.28077942 -0.67299193
-0.93920577 0.90699077 0.18510216 -0.41807625 -0.5674438 0.7109603
0.3464917 0.38713634 -0.04453602 0.6495688 -0.2622741 -0.60322756
-0.39887965 0.6595181 0.98018825 0.78099704 0.43129033 0.34383425
-0.15678689 0.04351369 0.26004997 0.6608888 -0.13375512 -1.3547349
0.3506272 0.09985684 0.8801719 -0.59554064 -0.19899683 -0.46517012
-0.3636372 0.9178797 0.61763334 -0.68501496 -0.7801302 1.9875559
0.42346096 0.15499976 0.16961561 0.75353605 -0.5827284 0.7371994
0.16724597 -0.56637347 0.78828335 -0.4593685 -0.75100696 -0.23547736
0.57396364 -0.04357457 0.9732875 0.51276195 -1.2548999 0.09942114
-1.1259586 0.663488 0.02822581 -0.38287866 0.00797193 0.3650111
-0.73931843 0.8938261 -1.1894753 -0.14291196 0.3719784 0.44740072
0.12319329 0.4137159 -1.1378542 1.1485659 0.63235015 0.01184411
-1.859373 -0.650021 -0.72203976 0.07173327 1.108009 -0.31175345
1.5308104 -0.43426073 -1.9134415 -0.23138472 0.30167297 -0.79297125
0.97335386 -0.56294584 0.15176475 0.21163177 -0.02414706 0.2057085
1.0048835 -1.218799 -0.7449049 -0.18599671 0.26430368 0.39242092
-0.24655344 -0.47867736 0.14535132 -0.26092407 -0.43517956 -1.2523444
-0.6852103 0.11426836 -0.22130308 -0.02511234 0.87442136 -0.4259105
0.99782795 -1.8352854 0.17040113 0.28192616 -0.03732479 0.20035653
0.01571057 0.5998911 0.5763479 -0.0643305 -0.3922418 0.09506768
0.15939902 0.3855927 -0.8634589 0.8539205 0.12140504 0.685703
-1.2696819 -0.15610507 0.23471172 0.2688102 -0.47284234 0.33954313
-0.39655566 0.6370103 -1.0741575 0.14110513 0.37552404 0.26351964
0.08504501 0.6099032 -0.36681455 0.04672568 -1.1243793 1.0308946
-0.56224054 0.19423623 0.5243211 -0.92998946 0.6545162 0.06106822
0.43406174 -0.41736692 0.3873235 0.11821563 -0.15803978 -0.27394536
0.3944721 -0.38178986 -0.19872269 0.53820866 -0.39069977 -0.40516925
-0.20254639 -0.04857852 1.2221644 0.4056726 0.24320306 -0.54226685
0.44726303 0.09797674 0.5930822 0.6920442 -0.5477596 -0.42687696
0.91558546 -0.06108397 -0.9269128 -1.0132453 0.1897869 0.597188
0.46352896 -0.02098026 -0.6791785 -0.69943017 0.05224073 1.0588979
-0.75410116 -0.58889365 -0.6595026 -0.25969365 0.23838869 0.40283653
0.39252165 -0.8103981 -1.4346343 0.40331903 0.24313271 -0.68549794
-0.51376164 0.46498263 -0.6847614 -1.0612322 -0.5311221 -0.16663179
0.75560796 -0.16348536 0.3724307 -0.32326534 0.04365636 0.522799
-0.02236563 -0.56931484 -0.5320574 -0.27383548 0.34934208 -0.24622953
-0.29191983 -0.26431602 -0.5507668 0.2840084 -0.8138649 -0.2765674
0.18995619 0.8851341 0.75379163 0.67373097 0.6347657 -0.21561664
0.9380726 -0.43006065 -1.5181026 0.26796246 -0.6303187 0.5362823
0.93876255 -0.70037043 -0.77363116 0.13015209 0.17662999 -0.51753944
0.18650995 0.13819253 -0.238112 -0.05257871 0.25641328 0.32762197
0.3413081 -0.06403966 0.27971444 0.09814722 0.29525635 -1.0668592
0.9801003 0.5444685 0.48805904 -0.5604239 -0.5883991 -0.07197484
-0.11655955 -0.37768146 0.5050761 -0.57413167 -1.4968866 0.05673233
-0.6793555 -0.7996474 -0.63275146 0.42958254 -1.2229198 0.2713326
-0.11315109 -1.5712126 -0.11149624 -1.2630807 0.7023412 0.35794222
0.08748894 -0.90947616 0.26838547 -0.4265525 0.42531598 0.46536985
0.51041865 -0.25165868 -0.61321294 -0.04963473 0.31895608 0.1465126
-0.08437759 -0.27362293 -0.35819098 -0.60153604 0.45542297 -0.48032567
0.55132544 0.40744546 1.0907756 -0.87384003 -0.28934303 0.49551272
1.4640914 0.6795044 0.3346933 0.5786816 0.25905907 0.73344004
1.1035665 0.88006634 0.19734664 0.3708895 0.8717619 0.6311998
0.8247461 -0.52923054 0.85453963 -0.0451727 0.12443593 -0.05621572
-0.74765366 0.57367325 -2.0185013 -1.0330231 0.44454378 2.7770026
0.9928809 0.23097682 0.2242886 -0.09075873 0.4162297 -0.2920046
-1.0562525 -0.560209 0.34591374 -0.2510651 1.0715556 0.6920981
-0.9221832 0.77861 5.918864 0.8227165 -1.1549755 -0.22956575
0.52713346 -0.20465696 -0.18488644 0.13458054 -0.81355774 0.4620721
1.1482214 -0.55149066 0.828374 1.1029625 0.6929507 -0.41213283
-0.89149165 0.24753238 -0.6527499 -0.7983158 -0.64975756 0.35566893
0.61053354 -0.22566353 0.15655032 0.45316756 0.8443839 -0.9423386
0.86484796 0.46972215 0.41468436 -1.3959707 0.29671612 0.6775486
-0.9824804 -0.4574899 -0.2509112 -0.12827605 0.41062596 0.06347962
-0.7491666 0.10878723 0.27451658 0.30728975 0.07446892 1.0741464
-0.5682191 0.51688087 -0.5915528 -0.47761363 0.6480328 -0.4467895
0.91051656 0.59704715 0.48354498 -0.0690651 0.45047188 0.96766955
0.46547195 -0.33582154 -0.7340705 -0.3604417 0.42505863 0.96999097
-0.75687397 -0.14903155 0.07351435 0.618141 0.2083668 0.41258514
-1.0256884 -0.42178303 0.93805295 -0.1063608 0.26144138 -0.6497628
-0.23588443 -0.69001323 -0.11955169 -0.740122 0.15252753 -0.0333153
-0.7923037 0.30141252 0.12119445 -1.2988497 -0.50086725 -0.24776448
-0.7049486 0.73537594 -1.3936414 -0.52193636 0.2826727 0.46993038
0.35244438 0.09536769 0.40686613 -0.32452503 -0.6314441 0.23315035
0.37565967 -0.34685406 0.19724874 -1.4081357 0.06470881 0.97037494
-0.533656 0.45147824 1.2917094 -0.8757367 -1.8637851 -1.2308921
-0.01604116 -0.29234186 1.1551453 -0.0658206 -0.82610685 0.46857923
-0.05509477 0.00791864 -0.20181797 -0.42683896 0.25511605 0.06734643
-1.0685362 0.97268677 0.6298642 -0.2638255 -0.24349938 -0.06787019
0.76625305 -0.31617194 -0.58097816 0.370374 0.35781813 -0.31131324
0.67553586 -0.82840174 -0.04790813 -0.41149965 -0.04436294 -1.610435
0.19722405 -1.3185768 -0.63164353 0.5291156 0.01598419 -0.7002705
0.7909288 0.7051006 -0.02688413 -0.96041477 -1.1960756 -1.4295075
0.49699867 -0.30706236 0.32376078 0.33648998 0.38823423 -0.30498898
-0.538104 0.49429342 1.0774238 -0.03675589 0.5209069 -0.5981161
-0.25778478 -0.44394615 0.11503745 -0.85288274 0.4604763 -0.21932983
0.7107775 -1.3993423 -0.15619873 -0.53395724 -0.26507464 0.2567719
-0.12721328 -0.6798693 0.25138497 -0.05300963 -0.6711828 1.0749866
1.4112684 0.04496928 -0.5128082 0.38753328 -0.42895973 0.58013374
1.0751642 -0.43192187 -0.80507404 0.06524157 0.25965104 0.411656
0.38408452 -0.90074533 0.16780668 -0.9176483 -0.34161857 -0.49734932
0.31729245 -1.0736492 0.05871013 1.2602036 -0.6838001 -0.13894123
0.20989475 1.1197838 0.21739002 -0.2744117 0.9939129 0.16996303
-0.40888053 0.42752478 -0.7233903 0.22461703 1.4676913 0.23295632
-0.1820552 -0.7010594 -0.5946935 0.8265123 0.47127125 0.1667411
0.6586543 -1.0735054 -0.33376643 0.05642648 -0.2719492 -0.11880054
0.08137182 0.76758456 -0.3003621 0.40678427 -0.17824548 -0.1444133
-0.78176355 0.67663646 0.5687469 -0.241108 -0.5438973 0.54199153
0.09269087 0.02755962 0.5305734 -0.40480873 0.12576427 -0.21711186
0.62175614 0.5939195 -0.36798033 -0.14480959 -0.15844476 0.19568804
0.12694357 -0.6172253 1.1038677 -0.2108668 0.27679184 0.27063668
0.83410966 -0.35265294 -2.2344222 -0.00610159 0.08014815 -0.36751273
0.22888711 -0.58665437 -0.6328728 0.85405004 0.42434022 0.28289634
0.8441209 -0.31245753 0.6480038 0.5627234 0.82873946 -1.5252674
0.04015944 0.70324737 0.85809517 -0.87210834 -0.23514661 0.23253284
-0.90243536 1.0206866 0.66505116 -0.15806456 0.5974661 0.41385537
-0.36424235 0.19290924 -0.8698258 -0.19929379 -0.08482765 0.44615364
-0.33538964 0.23230053 -0.24587005 0.22208358 0.12899719 -0.06040788
0.75738114 1.2742938 -0.8651869 -0.9914405 -0.45994636 0.20680207
-0.44168666 0.3499982 0.1017192 0.66677326 -0.28829437 0.8092749
-0.17910556 -0.02159432 0.08142126 -0.16966537 0.4266438 -0.48224849
-0.01342976 0.19585748 -0.01570174 -0.834108 0.07871076 -0.7000451
-1.5251653 -0.27300745 -0.23656762 0.35474035 0.7050523 1.0038496
0.02649904 0.4237841 0.8480639 -0.48433873 -1.6039829 -0.4598205
-0.66175634 -0.2367173 0.80443424 -0.8876218 -0.4230798 -0.47494557]
|
[4.57480525970459, 2.258777379989624]
|
43cdf866-9bc6-44a6-9859-3e9be65900e4
|
shilling-black-box-review-based-recommender
|
2306.16526
| null |
https://arxiv.org/abs/2306.16526v1
|
https://arxiv.org/pdf/2306.16526v1.pdf
|
Shilling Black-box Review-based Recommender Systems through Fake Review Generation
|
Review-Based Recommender Systems (RBRS) have attracted increasing research interest due to their ability to alleviate well-known cold-start problems. RBRS utilizes reviews to construct the user and items representations. However, in this paper, we argue that such a reliance on reviews may instead expose systems to the risk of being shilled. To explore this possibility, in this paper, we propose the first generation-based model for shilling attacks against RBRSs. Specifically, we learn a fake review generator through reinforcement learning, which maliciously promotes items by forcing prediction shifts after adding generated reviews to the system. By introducing the auxiliary rewards to increase text fluency and diversity with the aid of pre-trained language models and aspect predictors, the generated reviews can be effective for shilling with high fidelity. Experimental results demonstrate that the proposed framework can successfully attack three different kinds of RBRSs on the Amazon corpus with three domains and Yelp corpus. Furthermore, human studies also show that the generated reviews are fluent and informative. Finally, equipped with Attack Review Generators (ARGs), RBRSs with adversarial training are much more robust to malicious reviews.
|
['Jason S. Chang', 'Hong-Han Shuai', 'Yun-Zhu Song', 'Yi-Syuan Chen', 'Hung-Yun Chiang']
|
2023-06-27
| null | null | null | null |
['review-generation']
|
['natural-language-processing']
|
[-5.38024902e-02 2.71287829e-01 -1.82449698e-01 -3.16681355e-01
-6.61512315e-01 -7.27791727e-01 8.18232656e-01 -2.61054367e-01
-1.89989358e-01 7.05930293e-01 2.12492675e-01 -4.17018741e-01
4.46691602e-01 -1.02324128e+00 -6.50833726e-01 -5.09167194e-01
3.73563558e-01 1.96622357e-01 -1.20431222e-01 -1.05584574e+00
4.31711972e-01 2.51176238e-01 -1.12261009e+00 3.01991761e-01
1.24659991e+00 5.29940844e-01 -4.26667333e-02 6.16786897e-01
1.39415994e-01 1.14648855e+00 -1.15930200e+00 -1.14114726e+00
4.72513944e-01 -4.29461777e-01 -4.86635447e-01 -3.53170872e-01
-4.34037223e-02 -6.46982610e-01 -5.10298371e-01 9.61831331e-01
5.41868150e-01 3.28389823e-01 5.77902377e-01 -1.13937497e+00
-1.59267902e+00 9.79753792e-01 -3.81335407e-01 -5.76107576e-02
4.83472317e-01 3.46442074e-01 1.33689380e+00 -9.56532300e-01
4.37850535e-01 1.22755098e+00 2.97302127e-01 1.20289946e+00
-7.80884266e-01 -7.89559901e-01 3.09778869e-01 -2.93375161e-02
-9.10567045e-01 -2.70194858e-01 7.26471364e-01 -1.55751273e-01
7.23513484e-01 4.13834661e-01 4.59399670e-01 1.73544621e+00
1.16144784e-01 1.10577726e+00 1.09410429e+00 -2.32819438e-01
3.36366594e-01 7.38719761e-01 2.16557547e-01 4.05265719e-01
4.07408446e-01 4.33778107e-01 -4.50978339e-01 -5.03990114e-01
2.28062674e-01 4.15692776e-01 -1.99040815e-01 -8.17917380e-03
-6.25868261e-01 1.45947731e+00 4.65293556e-01 8.75212103e-02
-3.15030515e-01 -1.05706602e-01 3.43851924e-01 4.42467332e-01
7.01036096e-01 1.11623728e+00 -4.22887653e-01 2.81113356e-01
-4.45171416e-01 3.11508596e-01 8.87933016e-01 1.01893818e+00
4.83665258e-01 4.18817997e-01 -3.40808362e-01 8.72483790e-01
3.81578773e-01 1.05916178e+00 9.12506163e-01 -2.30836973e-01
3.59336585e-01 3.93671334e-01 4.14101809e-01 -1.06655407e+00
4.59894203e-02 -4.56294179e-01 -6.62111223e-01 -7.22821876e-02
-4.91804583e-03 -3.25258911e-01 -7.51532972e-01 1.55723262e+00
1.73198685e-01 5.92037383e-03 3.87793213e-01 1.00810850e+00
7.00369120e-01 7.76244462e-01 -7.77410492e-02 -9.39839557e-02
1.04601336e+00 -1.28596807e+00 -5.99880815e-01 -2.95201510e-01
6.73284233e-01 -5.59981465e-01 1.63478065e+00 6.49212182e-01
-9.03064549e-01 -4.10412937e-01 -1.21421707e+00 4.36188787e-01
-4.23655212e-01 -3.33174877e-02 6.30407095e-01 1.20993650e+00
-7.29074240e-01 4.97197866e-01 -3.67814332e-01 3.22278365e-02
1.25815958e-01 1.45224690e-01 2.93873786e-03 -8.59595165e-02
-1.86914623e+00 1.00073957e+00 -4.50119525e-01 7.11721852e-02
-1.13433206e+00 -3.25903028e-01 -6.24071419e-01 -1.19812332e-01
4.61301506e-01 -3.81313056e-01 1.21594083e+00 -1.32255805e+00
-2.02325082e+00 3.78198922e-01 3.70343477e-01 -6.22260571e-01
3.34551454e-01 -7.22802222e-01 -6.43926442e-01 -2.21884735e-02
-1.20164305e-01 1.69239402e-01 1.33360004e+00 -1.17481399e+00
-6.43237308e-02 -3.26001257e-01 4.33345795e-01 2.41347358e-01
-5.70823908e-01 1.23826966e-01 1.15040153e-01 -8.69111061e-01
-6.58892274e-01 -1.17389464e+00 -4.88393068e-01 -8.78514409e-01
-6.40469491e-01 -2.03502163e-01 3.11443329e-01 -5.10097086e-01
1.19500327e+00 -1.84159088e+00 -6.31065294e-02 4.09625202e-01
8.43499303e-02 8.35049391e-01 -2.84628421e-01 4.56655592e-01
1.77092746e-01 5.50929189e-01 9.01767537e-02 -1.53246537e-01
9.68476851e-03 -1.00788385e-01 -1.01379287e+00 3.11559379e-01
1.05760485e-01 9.86415327e-01 -1.03433311e+00 3.38807166e-01
-8.27350914e-02 3.63573879e-01 -7.55989492e-01 6.20931506e-01
-2.51184702e-01 2.70113170e-01 -6.03334844e-01 3.47089380e-01
4.60525870e-01 -7.80480728e-02 4.29736115e-02 3.12092572e-01
4.67513055e-01 5.34779310e-01 -6.01889133e-01 1.15617728e+00
-7.47312188e-01 1.36606321e-01 -3.56953174e-01 -6.61048174e-01
1.34051025e+00 1.29884407e-01 -7.06863776e-02 -7.31628120e-01
9.92435664e-02 2.98184790e-02 -4.61147027e-03 -2.10990831e-01
9.62150574e-01 -8.30893219e-02 -3.02296907e-01 9.74155903e-01
-2.94762373e-01 -1.53759807e-01 -3.01514775e-01 7.47000873e-01
1.19705939e+00 -1.01996891e-01 9.91443247e-02 2.41951019e-01
7.56514728e-01 -3.53432149e-01 2.24977374e-01 1.08054090e+00
-6.59976527e-02 2.78361291e-01 1.62870973e-01 -3.69775236e-01
-8.66016567e-01 -7.98192799e-01 3.55285734e-01 1.34197366e+00
2.45084047e-01 -5.06486356e-01 -6.18736267e-01 -1.28362679e+00
1.99471861e-02 1.16975689e+00 -6.42796576e-01 -7.18319893e-01
-3.74187469e-01 -6.76444113e-01 6.62362337e-01 2.42719054e-01
5.79509325e-02 -1.39693272e+00 -2.13040873e-01 1.02878930e-02
-9.27033462e-03 -6.15037799e-01 -5.71496248e-01 -3.51270884e-02
-4.39462185e-01 -9.09850299e-01 -6.22886658e-01 -1.95894599e-01
8.08437049e-01 6.25943482e-01 1.00960147e+00 4.70108360e-01
8.90942570e-03 2.96713412e-01 -9.75678265e-01 -3.06557447e-01
-9.38323259e-01 1.66661978e-01 3.08957845e-01 -1.58605158e-01
4.10433054e-01 -2.14547783e-01 -6.75689280e-01 4.72550869e-01
-1.01502240e+00 -4.12665427e-01 5.02777696e-01 9.74323213e-01
1.67981745e-03 -3.21165055e-01 1.31083477e+00 -1.60480416e+00
1.30984604e+00 -6.72521412e-01 -3.84179026e-01 2.81028152e-01
-9.75540698e-01 7.33990073e-02 1.41138005e+00 -8.27370524e-01
-1.11031592e+00 -4.21513975e-01 -2.70152032e-01 -3.56758386e-01
1.56330138e-01 1.57161877e-01 -9.88554582e-02 1.28686175e-01
1.24997127e+00 3.44245642e-01 -1.48184642e-01 -1.29422382e-01
1.00032437e+00 8.81058991e-01 -1.05057374e-01 -3.92239720e-01
1.06058526e+00 1.62389144e-01 -8.12429905e-01 -3.62543643e-01
-1.06469011e+00 -2.79933870e-01 1.04973547e-01 -5.80423437e-02
3.13432544e-01 -9.22260940e-01 -5.85522115e-01 2.64460862e-01
-9.15017664e-01 -1.36475876e-01 -1.61188349e-01 2.11210668e-01
-3.75030130e-01 3.87119979e-01 -8.66402388e-01 -1.01464033e+00
-9.36615646e-01 -1.14223063e+00 4.89014447e-01 1.36372522e-01
-1.00632749e-01 -5.94250739e-01 3.48437876e-01 5.64233303e-01
6.04176402e-01 -3.87149036e-01 5.63667595e-01 -1.42516661e+00
-2.82144964e-01 -4.66846913e-01 3.54236484e-01 8.94607008e-01
1.75942257e-02 -1.73408855e-02 -8.76796961e-01 -3.72022212e-01
2.56334394e-01 -7.14108050e-01 6.66101992e-01 -1.40169904e-01
7.80089140e-01 -8.14466298e-01 1.38401449e-01 2.74254888e-01
1.01862454e+00 1.92781746e-01 7.55893946e-01 1.18641667e-01
6.94386184e-01 5.31340837e-01 9.62916851e-01 5.91579318e-01
1.16038978e-01 4.34625864e-01 4.61787194e-01 2.01240048e-01
3.31669450e-01 -5.28840661e-01 9.63494956e-01 9.17715430e-01
2.19810620e-01 -5.20478964e-01 -2.82026201e-01 9.63915437e-02
-1.62462044e+00 -1.03949225e+00 7.49686435e-02 2.26473403e+00
9.19952154e-01 1.47440940e-01 1.51759565e-01 -2.23367527e-01
7.52656460e-01 2.53400773e-01 -7.52576947e-01 -7.86428928e-01
4.34870385e-02 3.02024186e-01 4.63358074e-01 4.00286973e-01
-7.68563211e-01 1.37623501e+00 5.72295713e+00 6.40329063e-01
-9.39145327e-01 1.15489237e-01 7.26243556e-01 -2.52692103e-01
-8.05133402e-01 -8.67897421e-02 -9.59940970e-01 6.04064882e-01
1.07120645e+00 -4.25790846e-02 8.51222277e-01 1.15535462e+00
-1.12032322e-02 3.94655138e-01 -7.08520591e-01 4.75553662e-01
5.01528621e-01 -8.96958232e-01 1.51642412e-01 -1.03166558e-01
9.13557351e-01 -7.83798099e-02 4.82171237e-01 8.45777452e-01
1.06254578e+00 -1.00879359e+00 5.32294393e-01 3.70648980e-01
4.23928410e-01 -1.01994956e+00 7.00675249e-01 5.62674582e-01
-2.25188106e-01 -1.75298035e-01 -8.78334582e-01 7.08052441e-02
-8.29724148e-02 5.38346291e-01 -1.15409493e+00 3.19914609e-01
1.33710444e-01 5.01526177e-01 -6.09629095e-01 3.34015161e-01
-8.67648900e-01 9.02407110e-01 1.76747054e-01 -5.66778123e-01
1.08560529e-02 -1.45692408e-01 3.78520459e-01 8.01422358e-01
1.79119468e-01 1.56405136e-01 1.08319961e-01 7.92053998e-01
-2.46941105e-01 5.06310105e-01 -9.88823950e-01 -1.46362066e-01
5.57458162e-01 1.53978491e+00 -1.55707911e-01 -2.86814660e-01
-2.45473519e-01 1.10564101e+00 5.11444569e-01 3.50371659e-01
-1.01401412e+00 -3.74843180e-01 4.45093364e-01 -6.39921278e-02
6.75300732e-02 2.84982324e-01 -4.86376546e-02 -1.46407795e+00
-3.15426290e-01 -1.59319472e+00 -3.42876837e-02 -7.35389292e-01
-1.63208580e+00 9.62778926e-01 -6.28588676e-01 -1.02294600e+00
-3.46654534e-01 -1.72421917e-01 -5.00933528e-01 8.26293290e-01
-1.46428716e+00 -9.97916639e-01 2.56758094e-01 7.44156361e-01
6.36761189e-01 -5.93942285e-01 8.35683107e-01 -1.19224600e-02
-6.51436746e-01 9.80426908e-01 -3.78812477e-02 -7.98190292e-03
1.11946583e+00 -1.18703830e+00 7.74526358e-01 7.69742608e-01
3.72589946e-01 1.20057130e+00 7.94686913e-01 -8.92199874e-01
-1.44363666e+00 -1.02942562e+00 4.16174084e-01 -6.97295606e-01
6.73374116e-01 -4.36290830e-01 -7.97115922e-01 4.18085843e-01
1.59880713e-01 -1.87650129e-01 9.72484767e-01 1.71049252e-01
-8.68130267e-01 -5.45860827e-02 -1.18584800e+00 7.61033118e-01
6.84134424e-01 -6.44610405e-01 -5.77508330e-01 4.86371189e-01
1.08022964e+00 -5.21033481e-02 -4.07046676e-01 1.11474767e-02
3.80840898e-01 -9.35343981e-01 8.45937312e-01 -9.69373286e-01
6.77910388e-01 2.97014434e-02 -5.03973849e-03 -1.78031588e+00
-2.80619651e-01 -8.39713633e-01 -1.55094668e-01 1.05745387e+00
6.52024388e-01 -5.11048317e-01 7.13172615e-01 6.89216912e-01
3.07569299e-02 -6.63619757e-01 -1.66740000e-01 -6.48936033e-01
1.65028587e-01 -2.19724923e-01 7.79448688e-01 7.93076217e-01
3.12491655e-01 7.73653984e-01 -1.12020564e+00 1.02603555e-01
4.20594327e-02 -8.92343447e-02 1.17498028e+00 -7.99278677e-01
-7.38671899e-01 -3.10256444e-02 3.28904092e-01 -1.19535410e+00
4.12414640e-01 -8.62848699e-01 2.51944423e-01 -8.60604584e-01
-4.06576954e-02 -5.75438380e-01 -3.32904220e-01 3.19608748e-01
-6.05515838e-01 4.47238572e-02 2.39679560e-01 2.96220005e-01
-7.15424478e-01 6.43054247e-01 1.28934300e+00 -5.84764928e-02
-2.87784576e-01 5.95916212e-01 -1.36008179e+00 4.94016290e-01
8.90447795e-01 -5.39796770e-01 -7.58574486e-01 1.84488688e-02
7.97487080e-01 6.24780022e-02 -1.85903713e-01 -2.43787050e-01
-1.39230087e-01 -1.55427679e-01 -5.67887016e-02 -2.18026012e-01
1.15604430e-01 -5.58908224e-01 -4.82107401e-01 3.95226240e-01
-9.42137778e-01 1.94157347e-01 -4.22998786e-01 9.43382382e-01
2.07908407e-01 -4.73414868e-01 5.84059656e-01 -2.10479930e-01
4.36142349e-04 2.57555366e-01 -6.32320225e-01 1.02264568e-01
8.82474065e-01 3.93048406e-01 -5.89550316e-01 -8.13976705e-01
-3.63906175e-01 -5.46343476e-02 3.64852518e-01 6.95748150e-01
8.76205325e-01 -1.10979450e+00 -6.47648692e-01 3.43401551e-01
2.19632685e-01 -5.42516530e-01 1.39349401e-01 1.10497132e-01
-5.64194061e-02 2.14655757e-01 9.21451151e-02 1.29067317e-01
-9.76901293e-01 1.07772982e+00 1.57896671e-02 -6.18428230e-01
-3.13660055e-01 8.95657659e-01 1.90691829e-01 -8.03595901e-01
1.11559734e-01 1.76804811e-01 -4.06209916e-01 -2.32442632e-01
7.03417122e-01 2.41835967e-01 9.02741868e-03 -3.57727498e-01
-9.83317122e-02 -3.86948556e-01 -7.54706681e-01 -1.12905003e-01
1.04004169e+00 -5.76029681e-02 2.55145699e-01 -3.54208127e-02
7.03538299e-01 7.09267974e-01 -9.10018325e-01 -1.56000242e-01
-1.93675756e-01 -5.30463457e-01 -3.33168864e-01 -1.09821784e+00
-1.17239201e+00 7.69089341e-01 8.14849138e-02 5.25496900e-01
7.41487026e-01 -3.91298532e-01 1.01693308e+00 6.56153917e-01
4.03706342e-01 -9.36760604e-01 5.88944554e-01 4.68520880e-01
7.77872741e-01 -1.16068518e+00 -1.30959809e-01 -1.97187006e-01
-1.42951393e+00 7.04318523e-01 8.51466298e-01 -6.00367785e-01
4.49526995e-01 -1.28127575e-01 1.73289508e-01 6.68966249e-02
-1.06655705e+00 2.27815688e-01 3.56345326e-02 7.27488875e-01
3.34905207e-01 3.33412170e-01 -3.69266838e-01 1.10213280e+00
-4.11889553e-01 -5.12284935e-01 9.66798961e-01 4.33242589e-01
-6.11401200e-01 -1.39309514e+00 -2.20376670e-01 5.28695524e-01
-7.13434398e-01 -5.67882478e-01 -6.62111819e-01 2.70597160e-01
-4.32826936e-01 1.27959168e+00 -5.97393453e-01 -8.64516020e-01
3.12267691e-01 -1.55758545e-01 7.59905577e-02 -9.51846361e-01
-1.35939658e+00 -2.44656116e-01 1.53038278e-01 -4.29842085e-01
1.16122216e-01 -2.16202766e-01 -9.29331899e-01 -3.48730117e-01
-9.64636505e-01 4.65394199e-01 5.34058630e-01 8.17942739e-01
3.93373132e-01 1.74621329e-01 1.46166706e+00 -1.03451230e-01
-1.51759934e+00 -9.68111634e-01 -6.81143522e-01 8.06542993e-01
-7.80565962e-02 -2.78269827e-01 -6.73572481e-01 -4.90906656e-01]
|
[6.165739059448242, 8.18825912475586]
|
bea2e3bd-a49e-4f6b-a017-7ea6ee7d01fc
|
efficient-teacher-semi-supervised-object
|
2302.07577
| null |
https://arxiv.org/abs/2302.07577v3
|
https://arxiv.org/pdf/2302.07577v3.pdf
|
Efficient Teacher: Semi-Supervised Object Detection for YOLOv5
|
Semi-Supervised Object Detection (SSOD) has been successful in improving the performance of both R-CNN series and anchor-free detectors. However, one-stage anchor-based detectors lack the structure to generate high-quality or flexible pseudo labels, leading to serious inconsistency problems in SSOD. In this paper, we propose the Efficient Teacher framework for scalable and effective one-stage anchor-based SSOD training, consisting of Dense Detector, Pseudo Label Assigner, and Epoch Adaptor. Dense Detector is a baseline model that extends RetinaNet with dense sampling techniques inspired by YOLOv5. The Efficient Teacher framework introduces a novel pseudo label assignment mechanism, named Pseudo Label Assigner, which makes more refined use of pseudo labels from Dense Detector. Epoch Adaptor is a method that enables a stable and efficient end-to-end semi-supervised training schedule for Dense Detector. The Pseudo Label Assigner prevents the occurrence of bias caused by a large number of low-quality pseudo labels that may interfere with the Dense Detector during the student-teacher mutual learning mechanism, and the Epoch Adaptor utilizes domain and distribution adaptation to allow Dense Detector to learn globally distributed consistent features, making the training independent of the proportion of labeled data. Our experiments show that the Efficient Teacher framework achieves state-of-the-art results on VOC, COCO-standard, and COCO-additional using fewer FLOPs than previous methods. To the best of our knowledge, this is the first attempt to apply Semi-Supervised Object Detection to YOLOv5.Code is available: https://github.com/AlibabaResearch/efficientteacher
|
['Lulu Hu', 'Wenlong Guan', 'Mingtao Chen', 'Bowen Xu']
|
2023-02-15
| null | null | null | null |
['semi-supervised-object-detection']
|
['computer-vision']
|
[-3.20420384e-01 -1.15291951e-02 -2.89161384e-01 -3.73985171e-01
-8.42524886e-01 -4.58279938e-01 3.77279311e-01 -1.76273957e-01
-6.82830930e-01 4.36025888e-01 -3.31925571e-01 8.05597473e-03
3.11475873e-01 -4.16824192e-01 -7.43930638e-01 -7.87992418e-01
3.26055855e-01 3.67472142e-01 8.23235095e-01 1.14278510e-01
-1.56430021e-01 5.01287699e-01 -1.75329757e+00 1.19757496e-01
7.32526183e-01 9.44020212e-01 5.77358186e-01 4.33344662e-01
-6.85310438e-02 7.63204098e-01 -5.72619081e-01 -8.92831981e-02
3.67724597e-01 -2.80901045e-01 -5.68596780e-01 2.60478705e-02
8.46809804e-01 -4.62957144e-01 -2.94790864e-01 1.05220842e+00
9.11510646e-01 8.22801795e-03 6.12717032e-01 -1.35061121e+00
-6.06812954e-01 5.44356823e-01 -8.16152215e-01 3.38429451e-01
-1.10693961e-01 3.51378590e-01 8.97766650e-01 -1.29260719e+00
4.70596164e-01 1.13829124e+00 7.92528033e-01 8.29449356e-01
-1.37156820e+00 -1.08126652e+00 1.72879666e-01 9.80152264e-02
-1.64746618e+00 -3.73340398e-01 6.46129608e-01 -4.23355937e-01
6.88944638e-01 -2.32217968e-01 5.68637967e-01 1.02123308e+00
-3.41510653e-01 1.08628166e+00 1.02181315e+00 -5.04028797e-01
1.42104715e-01 4.67571795e-01 2.98359394e-01 9.90921140e-01
2.49142975e-01 2.90568173e-01 -5.26082397e-01 -6.60016760e-02
7.87204564e-01 1.08628690e-01 -1.53223976e-01 -6.25113189e-01
-1.01772034e+00 7.97183096e-01 9.82447505e-01 -9.14171711e-03
-3.39641757e-02 3.07590008e-01 1.89563543e-01 1.51796281e-01
2.77404308e-01 2.75933385e-01 -5.53002596e-01 2.15760306e-01
-1.12122047e+00 -1.45285338e-01 5.57885170e-01 1.28246963e+00
9.52619195e-01 8.43270570e-02 -2.12516174e-01 8.51108849e-01
6.61642015e-01 8.13909829e-01 4.95686620e-01 -9.97294784e-01
5.05958498e-02 7.51591861e-01 -1.06737666e-01 -3.37002069e-01
-3.27402443e-01 -6.56314015e-01 -5.54458916e-01 4.82232124e-01
5.72230279e-01 -7.10374340e-02 -1.18932211e+00 1.75000298e+00
6.71555758e-01 4.21699136e-01 -2.11252227e-01 1.00273788e+00
1.23352408e+00 3.94216657e-01 1.26615688e-01 1.28340349e-01
1.31019747e+00 -1.40482581e+00 -3.66233736e-01 -1.57574922e-01
7.49896765e-01 -8.45097482e-01 1.06074917e+00 1.50032446e-01
-8.13834071e-01 -8.41820657e-01 -1.11842132e+00 -2.15676576e-01
-1.83874130e-01 5.63489914e-01 5.52741468e-01 5.30110300e-01
-1.08377349e+00 2.48362109e-01 -8.79881203e-01 -2.80730486e-01
7.92417586e-01 4.28428501e-01 -3.25891137e-01 -9.90195349e-02
-7.94829786e-01 6.96554005e-01 2.46574968e-01 7.17141898e-04
-1.23790812e+00 -7.02113152e-01 -7.37085700e-01 -2.44379237e-01
4.35745627e-01 -6.19294226e-01 1.46118271e+00 -1.15709805e+00
-1.43997395e+00 1.11817610e+00 -1.17789052e-01 -4.68827337e-01
3.63873303e-01 -8.89432132e-02 -1.37621805e-01 1.94846004e-01
2.45020628e-01 1.29118323e+00 1.08417487e+00 -1.06925881e+00
-7.59961188e-01 -1.74911126e-01 -1.77244484e-01 1.19586125e-01
-2.60158122e-01 -8.29613358e-02 -6.09949946e-01 -4.24244523e-01
-2.68766694e-02 -1.01582205e+00 -1.90451846e-01 4.79347378e-01
-2.68315941e-01 -5.44976950e-01 8.03673744e-01 8.64396840e-02
9.53388631e-01 -2.49324918e+00 -2.60894150e-01 -8.01278651e-02
4.98423278e-01 7.17141449e-01 -4.29117858e-01 -1.05814613e-01
-2.22873352e-02 -2.31178939e-01 -1.65442809e-01 -7.04990566e-01
-1.20208614e-01 1.06380150e-01 -1.13997608e-01 6.66575432e-01
2.25816295e-01 7.42985427e-01 -1.06232870e+00 -7.27284193e-01
1.36990577e-01 5.88935733e-01 -3.34988177e-01 3.13052595e-01
-2.05663711e-01 1.76011190e-01 -3.76824349e-01 8.51398945e-01
7.70757914e-01 -5.18023074e-01 -3.65641624e-01 -1.02760576e-01
-1.44345552e-01 2.39196524e-01 -1.34118915e+00 1.79648709e+00
-2.33851701e-01 7.00999558e-01 -1.38130309e-02 -4.15752023e-01
9.68690634e-01 1.73970804e-01 2.07901523e-01 -5.27214587e-01
2.77703077e-01 3.26598793e-01 -2.17937544e-01 -6.93495870e-02
3.13415408e-01 2.64275253e-01 3.30021054e-01 3.82858098e-01
4.75554198e-01 -3.45797390e-02 2.00328827e-01 4.28138465e-01
1.03237259e+00 2.76226401e-01 9.49529409e-02 -8.75034705e-02
3.91513020e-01 5.03992550e-02 7.62071848e-01 7.70033717e-01
-5.70621490e-01 8.13858509e-01 9.72135141e-02 -2.66337961e-01
-9.04955924e-01 -1.29003751e+00 -4.89102989e-01 1.13236821e+00
2.58148670e-01 -2.51148373e-01 -5.97633660e-01 -9.04369116e-01
1.91298157e-01 3.14066887e-01 -6.14595652e-01 -1.16203003e-01
-1.79190576e-01 -4.51655626e-01 6.84012949e-01 6.51730955e-01
6.24858797e-01 -1.10328269e+00 -4.80096936e-01 1.42032668e-01
1.41953543e-01 -9.67361927e-01 -8.42177689e-01 5.74613214e-01
-7.50235021e-01 -1.15069115e+00 -8.23262572e-01 -9.92826104e-01
9.38571215e-01 7.54647553e-01 1.01368773e+00 7.53116980e-02
-3.96508515e-01 2.56333977e-01 -3.25037062e-01 -5.35690486e-01
-1.69552267e-01 8.70004967e-02 1.03348173e-01 -1.28742486e-01
6.64283812e-01 -2.53768861e-01 -8.58059108e-01 6.97128057e-01
-7.61830986e-01 -7.14232102e-02 6.37394190e-01 9.21637595e-01
9.41986442e-01 -5.36315084e-01 4.60597724e-01 -9.11305785e-01
-7.38183036e-02 -3.45221221e-01 -9.40733731e-01 -3.20937932e-02
-8.45261931e-01 6.37278035e-02 3.77112508e-01 -7.90374756e-01
-8.35329413e-01 5.81691504e-01 -1.06596358e-01 -7.85725296e-01
-1.49124593e-01 -2.33755916e-01 5.03416499e-03 -3.13299596e-01
1.03236282e+00 -4.77695838e-03 -1.87839270e-02 -6.20315850e-01
4.41027403e-01 7.39015758e-01 5.26977837e-01 -2.32585505e-01
8.65803897e-01 6.07690156e-01 -2.42948845e-01 -4.93327409e-01
-1.16498661e+00 -1.05942488e+00 -6.07400060e-01 -3.07225306e-02
7.51696229e-01 -1.44933259e+00 -3.46629798e-01 7.40503311e-01
-9.38813925e-01 -6.53708100e-01 -6.71575248e-01 5.29107094e-01
-3.14875603e-01 1.34082958e-01 -5.64441681e-01 -4.00674939e-01
-4.59346294e-01 -1.16232049e+00 1.12605309e+00 5.38037419e-01
1.07046209e-01 -6.48201287e-01 1.33334011e-01 2.04378486e-01
2.47168675e-01 -9.13558677e-02 -1.00560918e-01 -5.24081409e-01
-6.97343290e-01 -7.57763013e-02 -5.16144872e-01 6.36709869e-01
1.01609379e-02 2.91686505e-02 -1.30877531e+00 -5.43354392e-01
-3.60628426e-01 -7.61400998e-01 1.18252409e+00 3.52020741e-01
7.93637872e-01 2.18920097e-01 -3.89838099e-01 9.70426798e-01
1.33029759e+00 -3.21376204e-01 3.05633128e-01 3.41153860e-01
9.14407849e-01 1.46073073e-01 7.66130805e-01 2.95217395e-01
4.52356249e-01 6.09199762e-01 4.41176057e-01 -4.80953276e-01
-7.53827333e-01 -3.70903343e-01 5.37697911e-01 4.48184490e-01
3.60543638e-01 1.14429682e-01 -6.96099579e-01 6.42472744e-01
-1.70614040e+00 -7.10894227e-01 -4.22362238e-01 2.10315871e+00
1.09318221e+00 6.34382963e-02 3.63607287e-01 -1.76744193e-01
7.73770869e-01 -1.19946390e-01 -7.32835710e-01 1.26509145e-01
-1.05684012e-01 9.30965319e-02 7.39976645e-01 2.02681020e-01
-1.16454709e+00 1.22119999e+00 5.34630775e+00 1.06511068e+00
-1.01525986e+00 4.89926487e-01 3.27165246e-01 -2.03561395e-01
1.80607900e-01 3.50426533e-03 -1.51975429e+00 4.27270621e-01
6.60782993e-01 3.12498868e-01 1.46502957e-01 1.41303325e+00
-6.94298223e-02 -2.39248753e-01 -1.12958646e+00 1.01821482e+00
2.33731940e-02 -1.10774112e+00 -3.43667001e-01 -1.31640047e-01
9.00339484e-01 7.09171593e-01 1.00825898e-01 2.05810547e-01
5.62944651e-01 -5.45490384e-01 9.49264824e-01 2.16478318e-01
1.01497340e+00 -4.76163059e-01 6.14897490e-01 2.80735284e-01
-1.16246617e+00 6.75803795e-02 -6.31372750e-01 3.45867932e-01
-1.52213350e-01 6.37139261e-01 -9.91951048e-01 -4.93980944e-02
8.37728977e-01 7.40019083e-01 -8.38038921e-01 1.44431520e+00
-7.66379654e-01 8.02684307e-01 -5.91472507e-01 -6.06952496e-02
3.14338535e-01 3.02542984e-01 4.76607710e-01 1.29389107e+00
-2.84536108e-02 -2.48256758e-01 2.60124803e-01 7.96052992e-01
-2.50747561e-01 -1.10326402e-01 -3.55241597e-01 4.20293003e-01
7.59668052e-01 1.50381398e+00 -7.22314537e-01 -2.57127672e-01
-4.41857427e-01 8.00146282e-01 5.35871804e-01 3.02624226e-01
-8.33555460e-01 -2.39078730e-01 2.83851385e-01 1.76909670e-01
6.11062467e-01 -1.31553784e-01 -1.18830055e-01 -1.06895983e+00
-2.83143997e-01 -6.66161358e-01 5.69560647e-01 -7.36533642e-01
-1.29763162e+00 5.86096764e-01 -1.68010473e-01 -1.38488054e+00
1.28202796e-01 -4.71380264e-01 -4.61224228e-01 7.11919188e-01
-1.89389491e+00 -1.25775123e+00 -5.44637024e-01 8.35701227e-01
6.10485971e-01 -2.83242404e-01 6.17952526e-01 3.75418186e-01
-7.09728003e-01 9.25280869e-01 1.80071726e-01 2.92342424e-01
1.13178265e+00 -1.46997893e+00 1.69324145e-01 8.62257123e-01
4.71368104e-01 4.27102208e-01 3.12038362e-01 -3.34296465e-01
-9.63993847e-01 -1.33285201e+00 5.25980890e-01 -4.50198084e-01
5.79579532e-01 -4.53638941e-01 -7.93154597e-01 6.42634451e-01
-4.64148335e-02 6.85424924e-01 6.49728000e-01 -2.64364667e-02
-6.55712545e-01 -4.45752144e-01 -1.11155856e+00 2.68179089e-01
9.68656600e-01 -3.92250597e-01 -4.96396393e-01 4.43478078e-01
7.64422178e-01 -4.66574967e-01 -4.24300492e-01 2.13711470e-01
3.86057794e-01 -8.25577259e-01 8.23005378e-01 1.44031972e-01
-4.52776775e-02 -8.73374820e-01 1.73879817e-01 -1.01333821e+00
-4.19873387e-01 -4.68103409e-01 -2.38378182e-01 1.46294069e+00
5.08244693e-01 -6.22602463e-01 8.79402757e-01 1.90902010e-01
-2.94618160e-01 -6.19918942e-01 -9.25048947e-01 -9.12560403e-01
-2.84822017e-01 -2.03390419e-01 3.34615201e-01 7.80073941e-01
-6.26013279e-01 2.41456315e-01 -5.99437319e-02 2.26998284e-01
8.56182456e-01 5.43687493e-02 9.92676914e-01 -1.07566333e+00
-4.99198407e-01 -2.67146587e-01 -3.42139602e-01 -1.31242239e+00
4.99425307e-02 -9.71741855e-01 2.99423873e-01 -1.22822070e+00
2.60613620e-01 -9.63722527e-01 -5.00623524e-01 7.95022011e-01
-1.89652547e-01 7.20695317e-01 1.50705963e-01 5.21371126e-01
-1.11229968e+00 6.87256753e-01 1.27259219e+00 -9.03912783e-02
-3.58134300e-01 1.94777668e-01 -4.97922450e-01 8.17794025e-01
6.83965981e-01 -1.05374217e+00 -3.72865289e-01 -4.74173099e-01
-1.86999455e-01 -5.77131629e-01 4.31136996e-01 -1.01077855e+00
5.59832215e-01 2.24223137e-01 4.46946323e-01 -6.44225955e-01
1.24196500e-01 -7.78999150e-01 -2.63323873e-01 4.96506006e-01
-1.73660487e-01 -1.94812253e-01 8.87623429e-02 5.95676005e-01
-1.24900445e-01 -3.43415856e-01 1.29442859e+00 2.38118060e-02
-9.89253700e-01 5.13958037e-01 -7.37192705e-02 1.92915127e-01
1.11071408e+00 -2.25142732e-01 -3.57415736e-01 1.11019097e-01
-4.49782014e-01 5.65578043e-01 4.87105846e-01 2.40290537e-01
6.27544641e-01 -1.27062917e+00 -5.74851036e-01 1.28453493e-01
3.13147902e-01 5.88758945e-01 2.68051606e-02 8.36989164e-01
-5.10072470e-01 1.66092999e-02 -7.01598590e-04 -9.75581825e-01
-1.24280035e+00 5.01558483e-01 4.16299731e-01 -1.19024687e-01
-6.28913164e-01 1.40141320e+00 2.11745232e-01 -4.35718924e-01
6.41665697e-01 -1.48535684e-01 -3.86046842e-02 3.18032429e-02
6.29478753e-01 3.04096073e-01 -2.06401289e-01 -5.07678568e-01
-4.22497332e-01 5.47718346e-01 -3.68345082e-01 8.77886117e-02
1.09852409e+00 -6.53130114e-02 1.84820637e-01 4.44214612e-01
1.06604397e+00 -1.17912434e-01 -1.71746957e+00 -5.79001427e-01
-2.06819922e-01 -3.62100095e-01 3.51096272e-01 -7.41006136e-01
-1.06902993e+00 8.32651734e-01 9.99058008e-01 -3.14685404e-01
1.10089517e+00 2.47006476e-01 8.08418036e-01 2.64005125e-01
3.28165144e-01 -1.01842809e+00 3.57527912e-01 3.88115674e-01
4.30760682e-01 -1.47341752e+00 -5.22619253e-03 -3.62349480e-01
-5.61632633e-01 7.65829802e-01 1.14874506e+00 -3.03966254e-01
6.13292277e-01 3.07061136e-01 3.71035546e-01 -8.70618522e-02
-6.35383070e-01 -6.05889142e-01 3.38997722e-01 6.17615700e-01
1.28912434e-01 -6.28243908e-02 -4.58999649e-02 1.82081476e-01
2.37020344e-01 -6.96993759e-03 1.96876407e-01 7.64998198e-01
-5.80636203e-01 -1.12144983e+00 -2.84080863e-01 2.53313601e-01
-2.72466063e-01 -2.57798851e-01 -1.86470404e-01 7.46924102e-01
4.18753862e-01 8.49503100e-01 1.06395572e-01 -1.51450753e-01
2.57314622e-01 -2.07897186e-01 4.92334187e-01 -9.44927752e-01
-6.31055772e-01 2.36215651e-01 -3.16053092e-01 -5.29816151e-01
-5.18208861e-01 -5.91670036e-01 -1.56090820e+00 -6.48894161e-02
-9.37960029e-01 4.72440161e-02 6.49705768e-01 6.38208449e-01
4.03325081e-01 3.47767442e-01 7.32965410e-01 -7.90551841e-01
-6.50751889e-01 -1.02723300e+00 -7.33578146e-01 2.37485036e-01
5.01083076e-01 -8.77334297e-01 -4.28529233e-01 -6.28337413e-02]
|
[9.193265914916992, 1.2882637977600098]
|
db5bbd5a-9ef5-40f4-a20b-706c83752905
|
learning-unseen-modality-interaction
|
2306.12795
| null |
https://arxiv.org/abs/2306.12795v1
|
https://arxiv.org/pdf/2306.12795v1.pdf
|
Learning Unseen Modality Interaction
|
Multimodal learning assumes all modality combinations of interest are available during training to learn cross-modal correspondences. In this paper, we challenge this modality-complete assumption for multimodal learning and instead strive for generalization to unseen modality combinations during inference. We pose the problem of unseen modality interaction and introduce a first solution. It exploits a feature projection module to project the multidimensional features of different modalities into a common space with rich information reserved. This allows the information to be accumulated with a simple summation operation across available modalities. To reduce overfitting to unreliable modality combinations during training, we further improve the model learning with pseudo-supervision indicating the reliability of a modality's prediction. We demonstrate that our approach is effective for diverse tasks and modalities by evaluating it for multimodal video classification, robot state regression, and multimedia retrieval.
|
['Cees G. M. Snoek', 'Hazel Doughty', 'Yunhua Zhang']
|
2023-06-22
| null | null | null | null |
['video-classification', 'retrieval']
|
['computer-vision', 'methodology']
|
[ 5.73255181e-01 2.48861611e-02 -3.34907115e-01 -4.86120582e-01
-1.09175670e+00 -7.27847040e-01 9.13800359e-01 -3.53668779e-02
-5.94743907e-01 8.65484893e-01 3.04663219e-02 -6.96533024e-02
-8.37178305e-02 -1.58421189e-01 -9.33100879e-01 -6.37828290e-01
1.45142794e-01 3.82054299e-01 -4.45416644e-02 -7.94318765e-02
1.59861352e-02 4.11540180e-01 -1.64161181e+00 6.16612554e-01
3.82736623e-01 8.45868826e-01 1.93553064e-02 8.01951885e-01
9.17213261e-02 7.47389853e-01 -1.98112652e-01 -3.00963432e-01
-1.27140433e-02 -3.41160238e-01 -9.78304923e-01 3.24226707e-01
4.56648290e-01 -2.26922914e-01 -3.24163795e-01 8.44188690e-01
2.19692692e-01 2.21196115e-01 8.45445514e-01 -1.61187255e+00
-5.78812100e-02 3.06694210e-01 -3.53090763e-01 -2.06854656e-01
8.95279109e-01 -1.16062127e-01 9.98049021e-01 -1.02996433e+00
8.51883292e-01 1.16708267e+00 5.43225884e-01 7.42781639e-01
-1.43587184e+00 -1.55669928e-01 1.16572231e-01 2.95600176e-01
-1.25036418e+00 -7.59355366e-01 7.02852547e-01 -3.91744286e-01
8.78569067e-01 3.98273170e-01 3.20089519e-01 1.35644329e+00
-1.79037228e-01 1.01721346e+00 9.01359022e-01 -7.60067105e-01
7.86171630e-02 3.15573692e-01 1.00453876e-01 8.35914671e-01
-2.79091686e-01 -1.11615071e-02 -9.45997298e-01 -1.45729661e-01
4.84021425e-01 -1.33981571e-01 -2.74757832e-01 -7.30158031e-01
-1.57419181e+00 5.31686544e-01 1.57115608e-01 1.10959642e-01
-2.36002073e-01 -4.39197607e-02 3.90071750e-01 5.48852265e-01
-7.59714236e-03 2.80581802e-01 -6.05173647e-01 -1.36180460e-01
-4.82267231e-01 -2.63635200e-02 8.89919698e-01 9.12979782e-01
1.01692975e+00 -3.53533238e-01 2.72566944e-01 9.19129908e-01
4.84397382e-01 6.31888151e-01 3.91824186e-01 -1.23552549e+00
6.40622556e-01 4.33275938e-01 3.17648016e-02 -5.96104920e-01
-5.27708530e-01 2.96559423e-01 -6.07100487e-01 7.38826022e-02
5.49694657e-01 -2.73631901e-01 -8.79721642e-01 2.01474285e+00
2.30914041e-01 1.48222208e-01 4.42819059e-01 7.96448231e-01
6.81311369e-01 6.02895439e-01 2.43708298e-01 -2.34371185e-01
1.01523745e+00 -6.60384834e-01 -5.22986352e-01 -1.70829311e-01
6.15986943e-01 -4.77769345e-01 7.58443296e-01 3.48094016e-01
-9.82444525e-01 -4.38404530e-01 -8.86555314e-01 -2.30895672e-02
-4.42364126e-01 4.45677377e-02 6.00681782e-01 3.96972805e-01
-8.30160201e-01 5.35631180e-01 -9.82780039e-01 -4.87443238e-01
-3.25830765e-02 7.01239586e-01 -9.78361189e-01 -2.55978048e-01
-1.06311989e+00 1.13915384e+00 6.76246107e-01 1.05209783e-01
-8.72068465e-01 -3.88467014e-01 -1.23773611e+00 -1.86863303e-01
3.57125223e-01 -8.34562123e-01 1.06902099e+00 -1.16508603e+00
-1.50072062e+00 7.00385213e-01 -3.84224534e-01 2.61998940e-02
1.51368216e-01 -4.97163050e-02 -3.33596140e-01 2.85481155e-01
-3.81806105e-01 1.07277644e+00 1.07354534e+00 -1.57488298e+00
-6.14272416e-01 -3.01361799e-01 5.46494164e-02 6.56107306e-01
-3.62411439e-01 -1.42636895e-01 -5.44800580e-01 -6.51145950e-02
4.26128983e-01 -1.24521816e+00 1.99009478e-01 -2.44797856e-01
-3.99334311e-01 -2.33548842e-02 6.48795068e-01 -3.54877919e-01
6.50157750e-01 -2.10463333e+00 9.28314686e-01 6.03757679e-01
-7.65204355e-02 -3.08115542e-01 -4.95965302e-01 4.98114169e-01
-6.97283074e-02 -1.92190707e-01 -1.67175010e-01 -8.04271877e-01
1.12741411e-01 6.62478685e-01 -1.56722665e-01 4.46077794e-01
4.59711939e-01 6.91030562e-01 -7.37131059e-01 -6.88067138e-01
4.32355642e-01 6.32659256e-01 -4.20074522e-01 3.49422216e-01
-1.59913570e-01 7.13008106e-01 -2.08053514e-01 7.74743915e-01
3.48083496e-01 -1.76299199e-01 4.67944145e-01 -4.58331794e-01
2.22543955e-01 -4.69176620e-02 -1.28428340e+00 2.06606174e+00
-3.66295606e-01 5.67354798e-01 -1.10242635e-01 -1.08215356e+00
3.84026080e-01 4.77150112e-01 5.95910072e-01 -3.76022995e-01
8.09103474e-02 6.57922635e-03 -3.68904740e-01 -8.06464612e-01
5.12993455e-01 -2.41440520e-01 -2.51612574e-01 2.76493043e-01
7.18519628e-01 -7.19125271e-02 3.60218808e-02 2.19359085e-01
7.94568002e-01 4.53600556e-01 -2.66638976e-02 3.29585612e-01
6.09319568e-01 -1.83277681e-01 1.49268419e-01 6.82593763e-01
9.23178941e-02 5.84866285e-01 4.55635637e-01 6.41612932e-02
-8.61952245e-01 -1.26217234e+00 -1.18643962e-01 1.44311070e+00
1.66900307e-01 -2.61691332e-01 -3.79200757e-01 -9.06509042e-01
-1.97287858e-01 5.67088544e-01 -5.52169681e-01 -8.51559937e-02
-2.59510905e-01 -5.29907107e-01 4.17691469e-01 4.01222229e-01
6.96929395e-02 -8.25373948e-01 -2.34907418e-01 -2.71476060e-01
-4.38175261e-01 -1.14951050e+00 -3.78601849e-02 4.04604405e-01
-8.60573232e-01 -9.83893394e-01 -5.39129555e-01 -7.43307829e-01
8.18152130e-01 -2.40107663e-02 8.34630847e-01 -2.02811956e-01
5.88780753e-02 1.10607100e+00 -2.41465792e-01 8.75933468e-02
-5.90719938e-01 4.63521704e-02 2.89634347e-01 1.16297305e-02
1.42322674e-01 -3.58240515e-01 3.43933627e-02 1.91687912e-01
-9.61097777e-01 -2.40460336e-02 5.43646455e-01 1.14684427e+00
4.46322858e-01 -2.06484884e-01 3.02980632e-01 -5.41767597e-01
2.86293030e-01 -5.33215404e-01 -3.04784864e-01 5.78956902e-01
-1.34627923e-01 3.97263408e-01 1.67536184e-01 -6.74191177e-01
-1.15209508e+00 5.33806384e-01 1.21082835e-01 -6.67486668e-01
-4.50271785e-01 8.14180732e-01 -2.65201986e-01 -2.15965480e-01
4.34080660e-01 1.62628368e-01 1.60425752e-01 -3.82991970e-01
5.08397043e-01 3.84448707e-01 6.47845209e-01 -7.31459439e-01
6.51465833e-01 2.49162436e-01 2.60075927e-01 -8.29115033e-01
-6.41036808e-01 -4.58303541e-01 -7.87442625e-01 -4.48302180e-01
6.45288646e-01 -1.02443254e+00 -9.29860890e-01 3.00291926e-01
-1.05239975e+00 -5.68112619e-02 -4.68640365e-02 8.90468836e-01
-7.26972163e-01 6.25213683e-01 -6.12577081e-01 -7.78743505e-01
3.71572703e-01 -9.89974856e-01 1.19168198e+00 1.50744319e-01
-2.24602565e-01 -1.27738500e+00 2.47480527e-01 3.24100882e-01
-1.18042216e-01 -1.02410547e-01 7.39120483e-01 -7.42579758e-01
-4.19051468e-01 -2.66100526e-01 -7.48092160e-02 3.34419817e-01
-2.82869246e-02 -2.36654673e-02 -1.20993888e+00 -3.22804898e-01
-3.23180974e-01 -8.00967097e-01 8.73633742e-01 3.59236263e-02
7.92747915e-01 5.38010262e-02 -3.28281522e-01 3.33789140e-01
1.16172159e+00 -1.94849148e-01 3.77942592e-01 1.88515991e-01
7.58054554e-01 7.23868966e-01 4.96453047e-01 2.82871157e-01
4.75910962e-01 5.04625261e-01 3.59129131e-01 1.59332797e-01
2.61259288e-01 -3.06777388e-01 5.04453838e-01 7.26306558e-01
-1.55937551e-02 -3.01059753e-01 -7.80960858e-01 1.79841965e-01
-1.93726671e+00 -1.04306781e+00 3.01012397e-01 2.27621174e+00
7.19598830e-01 -6.41304404e-02 1.33275807e-01 -2.54205544e-03
4.82949466e-01 -2.83893824e-01 -4.24066007e-01 -1.26220301e-01
-3.15179557e-01 -2.39258230e-01 1.94143817e-01 7.07165778e-01
-1.26095343e+00 6.43707097e-01 7.26797056e+00 3.32637697e-01
-9.21479940e-01 -6.98130578e-02 1.59564406e-01 4.69955690e-02
-4.66484219e-01 7.83422403e-03 -5.61249733e-01 6.58388585e-02
8.20291162e-01 4.30021495e-01 6.08443439e-01 4.34919447e-01
-3.83392155e-01 -4.20371443e-01 -1.67842412e+00 1.10995924e+00
3.33589286e-01 -8.88404131e-01 -3.40658315e-02 -1.70935363e-01
5.79100013e-01 3.04864403e-02 2.33413041e-01 3.74977946e-01
-5.98408766e-02 -7.72832215e-01 4.63313133e-01 9.44609463e-01
6.24967277e-01 -4.62141812e-01 5.88714123e-01 4.59418267e-01
-7.67950952e-01 -1.83648169e-01 -1.39453216e-03 1.00610718e-01
2.37803683e-01 -4.82718535e-02 -7.90357530e-01 6.68164551e-01
1.53335929e-01 7.08797455e-01 -6.19268239e-01 7.05084324e-01
7.38308430e-02 1.29482850e-01 -7.11944282e-01 2.95216203e-01
-1.09266505e-01 -9.92584899e-02 5.58112502e-01 1.14239991e+00
3.30376089e-01 -9.04905051e-02 2.45001301e-01 3.49371552e-01
-7.33073847e-03 -1.35963500e-01 -6.75211668e-01 -8.62382427e-02
3.61558467e-01 1.17938864e+00 -2.42215261e-01 -3.60338986e-01
-6.43626750e-01 1.13577843e+00 3.74502271e-01 5.15240073e-01
-6.87848091e-01 8.89804214e-02 2.93689787e-01 -6.72664225e-01
6.33738786e-02 -3.10787648e-01 -7.66045302e-02 -1.50681376e+00
-4.81421221e-03 -7.35306084e-01 7.33701766e-01 -8.69416475e-01
-1.31356370e+00 3.39228660e-01 3.21431130e-01 -1.33763170e+00
-8.19708526e-01 -7.39193201e-01 -1.82645991e-01 6.89557314e-01
-1.40301538e+00 -1.43334985e+00 -3.29175629e-02 1.00642991e+00
1.38693109e-01 -2.50190943e-01 1.05906701e+00 2.25822911e-01
-4.26549703e-01 5.94796479e-01 -6.81325644e-02 -1.76963344e-01
8.96767080e-01 -1.16331995e+00 -4.80875790e-01 4.35552210e-01
2.18248099e-01 6.46553457e-01 7.27036655e-01 -4.48510021e-01
-1.92813659e+00 -4.79000896e-01 8.06069374e-01 -7.15384126e-01
7.37757206e-01 -9.32105631e-02 -7.57389963e-01 8.50016892e-01
1.77790597e-01 -1.52965978e-01 8.96227241e-01 5.08192599e-01
-6.73376441e-01 6.29227608e-02 -9.06130552e-01 4.56244111e-01
7.23529160e-01 -9.65818048e-01 -7.18794346e-01 1.67569816e-01
2.94618189e-01 -3.81201446e-01 -1.09930098e+00 4.71378863e-01
7.96326458e-01 -5.85735857e-01 8.05033505e-01 -9.09898520e-01
2.24572405e-01 -2.67380029e-01 -6.11908078e-01 -1.17136419e+00
2.07750246e-01 -3.71020406e-01 -2.15829119e-01 1.12646842e+00
7.25999892e-01 -2.88274050e-01 6.16158783e-01 1.13119221e+00
1.17759835e-02 -1.03355430e-01 -1.07307482e+00 -4.74181861e-01
-2.19133019e-01 -5.30331969e-01 7.37590566e-02 1.09703636e+00
7.27192521e-01 5.35775125e-01 -5.83441377e-01 3.56222212e-01
6.73031926e-01 -2.60764752e-02 6.66119695e-01 -1.04147494e+00
-4.42936718e-01 -4.61160317e-02 -4.37071353e-01 -1.04066777e+00
6.34391010e-01 -7.52934396e-01 1.37303993e-01 -9.48074460e-01
5.61768532e-01 -1.43492699e-01 -4.05770093e-01 7.27753937e-01
3.41004096e-02 2.93918669e-01 3.71074229e-01 2.92996973e-01
-9.62118328e-01 3.22414070e-01 9.77954984e-01 -1.81554452e-01
-2.53918350e-01 -8.48318487e-02 -1.76782995e-01 6.03979588e-01
5.43541908e-01 -1.24566801e-01 -2.98304498e-01 -4.62672412e-01
3.25696975e-01 4.24494267e-01 5.96120954e-01 -8.43119025e-01
3.31618130e-01 -1.76628441e-01 6.24027908e-01 -4.21306878e-01
9.87651825e-01 -1.16919446e+00 2.57061034e-01 -9.29008201e-02
-7.56080210e-01 -2.43997797e-01 2.14164868e-01 5.93738258e-01
-2.69636929e-01 -2.46533170e-01 3.32974523e-01 2.47116446e-01
-8.94919097e-01 -9.14637223e-02 -3.96801531e-01 -4.23426151e-01
7.07326055e-01 -6.54086098e-02 -1.76847592e-01 -5.38129330e-01
-1.40841997e+00 2.56521672e-01 5.29466212e-01 4.95358139e-01
7.05162287e-01 -1.39848268e+00 -3.75894457e-01 3.28306556e-01
3.12313765e-01 -4.99310464e-01 3.06409061e-01 9.36815679e-01
1.87730074e-01 3.06028396e-01 -3.26376081e-01 -9.81300592e-01
-1.47437465e+00 6.03593111e-01 3.14661950e-01 1.64254680e-01
-5.32111488e-02 5.37544549e-01 -1.65804103e-01 -6.24676347e-01
4.26732302e-01 1.45232037e-01 -2.03371063e-01 1.06863819e-01
2.80118555e-01 1.73140228e-01 -1.17504217e-01 -9.60440159e-01
-3.85849148e-01 5.00110328e-01 3.32371518e-02 -6.42054200e-01
1.01854944e+00 -5.24637580e-01 -8.14550519e-02 1.00536299e+00
1.46654189e+00 -3.44438106e-01 -1.28464115e+00 -4.35827285e-01
8.45128894e-02 -1.69018850e-01 -1.72360122e-01 -8.06560278e-01
-6.33294106e-01 8.08745205e-01 6.68692648e-01 -2.09895354e-02
1.10616457e+00 3.86799306e-01 1.87554747e-01 1.00715542e+00
1.52327865e-01 -1.09162426e+00 -8.06600600e-03 7.26240575e-01
6.04181707e-01 -1.70630562e+00 -5.39981499e-02 -2.13695943e-01
-9.06692386e-01 1.28700066e+00 4.33039844e-01 4.63006318e-01
5.84842503e-01 7.00629056e-02 -4.63081850e-03 -1.54987611e-02
-8.36937010e-01 -2.80734539e-01 5.57455361e-01 6.51418030e-01
3.01263452e-01 -1.49057746e-01 2.67354399e-01 1.24519333e-01
3.67834300e-01 -3.88931371e-02 2.29485869e-01 1.17368603e+00
-3.12200338e-01 -1.23460639e+00 -4.01726484e-01 1.51177019e-01
-8.57805833e-02 1.09557875e-01 -3.51293296e-01 6.95571125e-01
-1.86715394e-01 9.14473653e-01 -3.09568830e-03 -5.54448903e-01
1.14374571e-01 5.80772758e-01 9.31552827e-01 -2.79534400e-01
-6.80306181e-02 2.28262916e-01 3.16068798e-01 -5.17147839e-01
-8.92726600e-01 -9.48616087e-01 -1.06990135e+00 -1.77484229e-02
-3.14912766e-01 3.53202340e-03 9.34671104e-01 1.27337718e+00
1.67433619e-01 1.16710894e-01 6.77492678e-01 -1.22524917e+00
-5.92193604e-01 -7.27890372e-01 -3.07642788e-01 6.61535621e-01
5.84837019e-01 -7.16770887e-01 -3.98941457e-01 4.52883750e-01]
|
[10.731208801269531, 1.688222885131836]
|
9660986e-b601-41cc-907d-833f742eed7d
|
warppinn-cine-mr-image-registration-with
|
2211.12549
| null |
https://arxiv.org/abs/2211.12549v1
|
https://arxiv.org/pdf/2211.12549v1.pdf
|
WarpPINN: Cine-MR image registration with physics-informed neural networks
|
Heart failure is typically diagnosed with a global function assessment, such as ejection fraction. However, these metrics have low discriminate power, failing to distinguish different types of this disease. Quantifying local deformations in the form of cardiac strain can provide helpful information, but it remains a challenge. In this work, we introduce WarpPINN, a physics-informed neural network to perform image registration to obtain local metrics of the heart deformation. We apply this method to cine magnetic resonance images to estimate the motion during the cardiac cycle. We inform our neural network of near-incompressibility of cardiac tissue by penalizing the jacobian of the deformation field. The loss function has two components: an intensity-based similarity term between the reference and the warped template images, and a regularizer that represents the hyperelastic behavior of the tissue. The architecture of the neural network allows us to easily compute the strain via automatic differentiation to assess cardiac activity. We use Fourier feature mappings to overcome the spectral bias of neural networks, allowing us to capture discontinuities in the strain field. We test our algorithm on a synthetic example and on a cine-MRI benchmark of 15 healthy volunteers. We outperform current methodologies both landmark tracking and strain estimation. We expect that WarpPINN will enable more precise diagnostics of heart failure based on local deformation information. Source code is available at https://github.com/fsahli/WarpPINN.
|
['Francisco Sahli Costabal', 'Daniel E. Hurtado', 'Sergio Uribe', 'Hernán Mella', 'Pablo Arratia López']
|
2022-11-22
| null | null | null | null |
['landmark-tracking']
|
['computer-vision']
|
[ 1.76170945e-01 -4.49217707e-02 -2.06701038e-03 -3.73970062e-01
-5.59266567e-01 -5.41721940e-01 1.75244153e-01 4.69917729e-02
-4.29712921e-01 6.39105916e-01 1.88079894e-01 -2.62749624e-02
-6.73090816e-02 -6.82298958e-01 -4.55268145e-01 -8.17153513e-01
-5.77670932e-01 5.32511592e-01 8.77363384e-02 -4.00401681e-04
-1.79984402e-02 6.59672916e-01 -5.29266298e-01 -2.73425411e-02
7.14473963e-01 8.49152625e-01 5.67878261e-02 7.15762496e-01
4.00147349e-01 5.39046466e-01 -3.45880032e-01 1.93124592e-01
2.16683671e-01 -6.80013359e-01 -9.66511846e-01 -2.82830983e-01
5.60157180e-01 -6.14444256e-01 -2.30787843e-01 8.49422514e-01
9.09799755e-01 1.10734478e-01 5.05610704e-01 -6.16102755e-01
-3.04382861e-01 5.09358823e-01 -2.16245234e-01 6.82964087e-01
3.69828902e-02 2.10106775e-01 6.73812151e-01 -7.55364537e-01
8.37398648e-01 6.73605561e-01 1.24518991e+00 5.26435316e-01
-1.42338288e+00 -2.64151216e-01 -7.05575705e-01 5.01117436e-03
-1.05971444e+00 -2.21631676e-01 1.13372791e+00 -8.31541240e-01
4.99303043e-01 3.22118312e-01 8.18045139e-01 6.56977177e-01
6.49292290e-01 2.06456557e-01 1.11587393e+00 -1.84203118e-01
-7.60131255e-02 -3.80038053e-01 1.19278006e-01 7.76479363e-01
3.34953815e-02 3.36205304e-01 -6.00064807e-02 -3.09531450e-01
1.26649117e+00 -6.29194733e-03 -8.46505344e-01 -4.58339632e-01
-1.47235453e+00 7.03460515e-01 5.99106669e-01 6.46391690e-01
-4.84525919e-01 4.65904206e-01 4.67104435e-01 7.73175135e-02
5.99972129e-01 5.66025913e-01 -3.28013480e-01 -1.32666111e-01
-1.04331362e+00 2.38965467e-01 6.24996960e-01 -2.33175978e-01
4.83432949e-01 2.89099842e-01 -2.61465430e-01 8.42298150e-01
3.23421508e-01 4.00442451e-01 6.52162910e-01 -1.62019479e+00
3.83651704e-02 2.49074847e-01 -1.99495301e-01 -1.20758247e+00
-7.67583311e-01 -4.76492286e-01 -1.16134655e+00 4.29248273e-01
8.22231233e-01 -3.16220790e-01 -6.61059499e-01 1.67721760e+00
4.28516984e-01 2.96769261e-01 -4.15874541e-01 1.41875148e+00
5.46990812e-01 1.50124088e-01 -4.19974923e-02 -2.55096167e-01
1.13903558e+00 -4.84520316e-01 -5.37628591e-01 -1.22653812e-01
7.61975169e-01 -4.90349382e-01 7.54276454e-01 -1.01510786e-01
-1.51704848e+00 -4.11241651e-01 -9.79465663e-01 1.44863963e-01
1.30969375e-01 -1.10294841e-01 2.37350032e-01 3.71183246e-01
-1.12616897e+00 1.43377423e+00 -1.45205915e+00 2.98021007e-02
3.38127106e-01 2.77154475e-01 -2.61957169e-01 2.06103355e-01
-1.37217617e+00 9.61193085e-01 2.00402066e-01 3.40302497e-01
-5.64351559e-01 -1.21756089e+00 -7.73950994e-01 -4.22836393e-02
-2.20717430e-01 -9.29327309e-01 8.00269485e-01 -9.20826614e-01
-1.33433342e+00 9.67466772e-01 2.40983620e-01 -2.95865476e-01
7.63046980e-01 1.82617441e-01 3.23800817e-02 4.74100441e-01
-1.64008457e-02 2.93511897e-01 6.43563449e-01 -9.95556712e-01
3.63874823e-01 -3.29203248e-01 -4.29599345e-01 -1.15274712e-02
4.24524508e-02 1.85939800e-02 -2.11325586e-02 -9.56367850e-01
3.52151811e-01 -1.04022300e+00 -2.57203341e-01 3.67909044e-01
-1.68677822e-01 4.58408862e-01 5.26370108e-01 -1.24106121e+00
9.41176891e-01 -2.09019756e+00 1.93597794e-01 3.35154951e-01
6.63663745e-01 1.73259422e-01 -6.47296151e-03 1.97563153e-02
-3.44804406e-01 2.01558277e-01 -8.55776846e-01 1.26153827e-01
-3.93624634e-01 -2.36809465e-05 1.32975847e-01 7.70394981e-01
2.58476198e-01 1.19233108e+00 -8.09044480e-01 -4.38098878e-01
-1.22461542e-01 8.73484075e-01 -5.02379954e-01 1.25617683e-01
3.00299972e-01 8.35077167e-01 -2.55774468e-01 2.74247855e-01
6.02234006e-01 -3.00355196e-01 4.08429891e-01 -4.33145165e-01
1.25773743e-01 3.38291414e-02 -7.80644834e-01 1.78222132e+00
-1.08954668e-01 6.00914776e-01 3.14469367e-01 -1.24032688e+00
7.34889448e-01 6.63238406e-01 1.01780784e+00 -4.90771890e-01
3.20471823e-01 4.40465510e-01 4.95339423e-01 -4.60562974e-01
-2.45267078e-01 -4.40126359e-01 2.65528262e-01 5.30136645e-01
4.77344915e-02 -1.33621573e-01 5.19370399e-02 -1.65000424e-01
1.21128869e+00 3.42998445e-01 1.93112195e-02 -5.86645544e-01
3.89427513e-01 -2.12391824e-01 6.25931084e-01 5.35243571e-01
-5.85486948e-01 1.06733751e+00 4.86317456e-01 -8.06041718e-01
-1.19677627e+00 -1.17966723e+00 -4.98364031e-01 2.98275828e-01
-1.63621485e-01 7.05324858e-02 -7.79748499e-01 -4.21115309e-01
1.28377050e-01 8.95591229e-02 -5.82944155e-01 -1.54714555e-01
-1.21231675e+00 -9.25193965e-01 6.39410675e-01 6.61469996e-01
2.51174778e-01 -1.13963151e+00 -9.37122762e-01 3.70820642e-01
-3.35932285e-01 -9.00131822e-01 -7.20645308e-01 -6.09700717e-02
-1.33193743e+00 -1.08059502e+00 -1.03787792e+00 -6.61054552e-01
6.90740168e-01 -5.03729045e-01 1.25900602e+00 4.32368457e-01
-6.82155669e-01 2.30244502e-01 2.55623013e-01 2.14874163e-01
-7.36646652e-01 -2.72519559e-01 -8.17019194e-02 -2.19328627e-01
-5.02216637e-01 -8.85439813e-01 -1.07913315e+00 2.39490464e-01
-7.24970281e-01 -2.25949109e-01 1.31531298e-01 8.48403633e-01
6.88442707e-01 -5.32213271e-01 3.63869995e-01 -7.34474301e-01
5.54738104e-01 -3.34705919e-01 -3.37382406e-01 -1.50632665e-01
-6.31576359e-01 2.28709932e-02 3.61125410e-01 -4.61351603e-01
-6.30642831e-01 -7.07174912e-02 -2.01428846e-01 -5.13770461e-01
1.23163447e-01 7.70834327e-01 5.49563468e-01 -3.30035359e-01
8.02248895e-01 1.67574003e-01 4.42764431e-01 -2.98784494e-01
-3.70560773e-02 -1.17281586e-01 6.36448324e-01 -6.39417589e-01
5.91008306e-01 4.13932174e-01 4.17794913e-01 -5.26229084e-01
-3.57478470e-01 -1.78229079e-01 -6.95213675e-01 -5.13378799e-01
9.22622859e-01 -3.38997573e-01 -7.45386183e-01 5.14087200e-01
-1.07371151e+00 -7.19887257e-01 -6.85946107e-01 6.55508161e-01
-6.57864392e-01 6.32669389e-01 -1.02117515e+00 -2.33751386e-01
-7.69536912e-01 -1.07136309e+00 6.60919487e-01 -1.26672342e-01
-2.51264215e-01 -1.57194006e+00 5.55260301e-01 1.92575634e-01
7.96044469e-01 1.06099701e+00 8.84770989e-01 -2.85665333e-01
-3.59867752e-01 -9.52331871e-02 -4.99486029e-02 6.34854078e-01
1.12404749e-01 -1.93676814e-01 -6.84692264e-01 -3.67923468e-01
3.37197453e-01 -1.36005715e-01 8.12448323e-01 9.07040238e-01
9.81480896e-01 -1.47057816e-01 -3.96813229e-02 8.78375888e-01
1.45988870e+00 1.42468676e-01 4.62552667e-01 -6.32863119e-02
9.12512898e-01 4.89554197e-01 -9.38331634e-02 2.05209017e-01
9.53663811e-02 7.05550671e-01 1.79837406e-01 -3.61069679e-01
-5.20888984e-01 2.92106241e-01 8.16303343e-02 1.12284851e+00
-6.71966255e-01 3.82204443e-01 -1.31535959e+00 4.45954353e-01
-1.54333150e+00 -8.80451620e-01 -2.14627787e-01 2.23248601e+00
1.18702662e+00 9.07944664e-02 -7.48030320e-02 -3.58217545e-02
5.24958491e-01 1.58169150e-01 -4.93259817e-01 -1.80187926e-01
1.17145590e-01 3.96635562e-01 3.39220762e-01 8.64988148e-01
-1.02043939e+00 5.58929294e-02 6.64818525e+00 5.48773035e-02
-1.54479933e+00 3.48104119e-01 8.43498409e-01 1.80453092e-01
-1.60000801e-01 -1.66717067e-01 5.83895072e-02 4.51269746e-01
9.77519453e-01 5.93547560e-02 4.39406335e-01 2.50978678e-01
3.20715904e-01 1.99903503e-01 -9.40710306e-01 5.81119299e-01
-3.67038436e-02 -1.47647810e+00 -6.41002536e-01 -9.52119231e-02
3.41731608e-01 3.89258295e-01 -2.70827621e-01 -2.53574312e-01
-3.53502154e-01 -8.97467852e-01 3.11171502e-01 1.02617013e+00
9.96753156e-01 -2.26627260e-01 7.28912771e-01 3.45196016e-02
-9.77521479e-01 3.74845862e-01 1.07965060e-01 2.47633830e-01
3.02662194e-01 7.41983056e-01 -7.95387506e-01 3.30344588e-01
5.61200023e-01 7.32680917e-01 -3.06440979e-01 9.04944003e-01
1.79717377e-01 6.76531017e-01 -3.18082511e-01 6.60498679e-01
-3.25611949e-01 -4.93590027e-01 7.97684848e-01 9.48456049e-01
3.55526388e-01 4.09111977e-02 2.83879787e-01 1.32929718e+00
-5.25858393e-03 -2.17527896e-02 -3.64044577e-01 1.15369923e-01
-1.13223568e-01 1.35302567e+00 -8.79446149e-01 -1.86443344e-01
-2.20576059e-02 7.26281226e-01 9.75923613e-02 4.05043215e-01
-5.22487223e-01 -2.13058591e-01 3.49155754e-01 5.40425956e-01
-1.48853242e-01 -3.48697752e-01 -2.95917571e-01 -1.17147017e+00
1.35809571e-01 -6.51305854e-01 1.65041402e-01 -7.53552437e-01
-1.06343794e+00 5.67458510e-01 -6.66391253e-02 -8.12928617e-01
-4.28556949e-01 -4.98112381e-01 -7.77569473e-01 1.23834431e+00
-1.32733822e+00 -7.55088806e-01 -4.28137690e-01 2.25912169e-01
-5.41497841e-02 3.16683173e-01 7.81301618e-01 5.20603359e-01
-2.30430469e-01 2.27310985e-01 -1.68460727e-01 5.23969889e-01
7.52189279e-01 -1.44939864e+00 2.38585353e-01 6.16123080e-01
-3.64188164e-01 6.57209933e-01 6.05840623e-01 -7.67093122e-01
-1.02200568e+00 -7.84775257e-01 6.87188566e-01 -3.99682879e-01
6.80759668e-01 1.01529814e-01 -1.25149608e+00 5.81114650e-01
1.15801636e-02 8.20979655e-01 4.15524364e-01 -4.90370631e-01
6.51262105e-02 -8.99026170e-02 -1.12142622e+00 5.04132695e-02
6.64767325e-01 -4.29344535e-01 -2.17555180e-01 2.65885919e-01
3.72962922e-01 -6.95587933e-01 -1.59000421e+00 6.30633235e-01
7.40688264e-01 -7.78170764e-01 1.11725211e+00 -3.41331452e-01
6.12515867e-01 -9.63578001e-02 3.43139350e-01 -1.23266089e+00
-3.95351976e-01 -5.25316715e-01 -2.05276832e-01 7.64077246e-01
9.34874937e-02 -8.98091793e-01 7.21344292e-01 5.48743367e-01
-2.01779097e-01 -1.00954771e+00 -1.01691377e+00 -5.44425726e-01
4.74353969e-01 -3.56706828e-02 -7.70863891e-02 1.36487210e+00
-1.83686227e-01 -2.14882493e-01 -1.89379491e-02 -1.75370187e-01
7.31298327e-01 -1.35076553e-01 -1.38196260e-01 -1.33629632e+00
-3.60033512e-01 -5.53920031e-01 -2.99560130e-01 -3.02447081e-01
6.98150396e-02 -1.23510337e+00 8.73403531e-03 -1.38892567e+00
1.59074500e-01 -5.71386755e-01 -3.93140495e-01 5.10686755e-01
-1.29694983e-01 6.00401044e-01 9.23257321e-02 4.90702271e-01
4.17680711e-01 1.26673102e-01 1.69255280e+00 -4.66361754e-02
-2.25073531e-01 -2.36825839e-01 -7.02728033e-02 7.45835066e-01
1.03985858e+00 -6.01045907e-01 -8.19742829e-02 -2.25573853e-01
9.31594744e-02 5.27902007e-01 7.79983342e-01 -1.10713279e+00
-1.89154357e-01 1.23190738e-01 4.73120928e-01 7.76007473e-02
1.53913766e-01 -6.75413728e-01 4.05201733e-01 9.86781478e-01
-3.26167136e-01 1.90094367e-01 -6.98015280e-03 -7.82094747e-02
-2.04999566e-01 -2.35513747e-01 1.00576985e+00 -1.99265644e-01
1.23479083e-01 4.80628341e-01 -2.05463111e-01 4.52571899e-01
4.51261312e-01 -6.76849037e-02 -1.67571247e-01 -9.50646177e-02
-9.84496832e-01 -1.98548600e-01 3.63129973e-01 -5.89953102e-02
5.98700762e-01 -1.43188941e+00 -9.49049771e-01 3.16536784e-01
-5.16843200e-01 -1.99175209e-01 3.76219422e-01 1.53843510e+00
-1.16968834e+00 -6.88799992e-02 -4.51633453e-01 -8.11940789e-01
-8.89316201e-01 2.27625057e-01 1.14658499e+00 -5.33515751e-01
-7.52634108e-01 4.63152677e-01 -7.10561946e-02 -4.01832819e-01
-2.38483608e-01 -5.23834646e-01 -2.13251680e-01 -1.57103926e-01
1.33590132e-01 4.16664273e-01 4.42169383e-02 -6.95231676e-01
-4.79552239e-01 9.32946920e-01 4.79845345e-01 5.42870499e-02
1.20213437e+00 9.70269516e-02 -4.34687436e-01 4.19820487e-01
1.30330229e+00 -1.55402064e-01 -1.35068214e+00 -5.90657331e-02
-1.67381987e-01 -1.67039484e-01 3.98350090e-01 -8.88480306e-01
-1.56929183e+00 8.94080281e-01 1.14118075e+00 3.87242138e-01
1.16788578e+00 -2.55976617e-01 9.59535420e-01 -1.99484155e-01
-1.87076643e-01 -6.25171244e-01 -4.94155996e-02 1.64434180e-01
1.02319980e+00 -1.11862493e+00 1.31751910e-01 -2.13369802e-01
-2.43662909e-01 1.32384753e+00 1.17670380e-01 -4.48339999e-01
8.34632158e-01 5.01993418e-01 5.36716282e-01 -2.22407013e-01
-1.64099768e-01 4.05707389e-01 4.94190216e-01 3.33004057e-01
7.87613750e-01 -4.96513732e-02 -6.46572828e-01 8.77895877e-02
1.50288388e-01 2.13347301e-01 4.29042459e-01 9.33392823e-01
-2.12543055e-01 -1.10997820e+00 -3.02867651e-01 3.16180706e-01
-8.81994128e-01 -1.24472737e-01 4.32563052e-02 5.54073155e-01
3.97902802e-02 6.39755726e-02 1.56383042e-03 1.45328239e-01
1.61967337e-01 3.85484904e-01 5.84368169e-01 -2.97503084e-01
-7.67112672e-01 1.63151354e-01 -1.30322069e-01 -6.45992339e-01
-6.41441345e-01 -7.66724348e-01 -1.34213078e+00 1.70637108e-02
9.00173113e-02 -2.04728901e-01 5.99687099e-01 6.89033985e-01
2.33620197e-01 7.69060194e-01 4.99476016e-01 -1.15350962e+00
-4.66555595e-01 -7.69766212e-01 -6.29084587e-01 7.27418065e-01
6.46605134e-01 -4.63294744e-01 -3.82880539e-01 2.85966307e-01]
|
[13.995209693908691, -2.4768311977386475]
|
5006dfa2-bb58-4c0a-b347-d030b34f60f7
|
improving-open-information-extraction-via
|
1905.13413
| null |
https://arxiv.org/abs/1905.13413v1
|
https://arxiv.org/pdf/1905.13413v1.pdf
|
Improving Open Information Extraction via Iterative Rank-Aware Learning
|
Open information extraction (IE) is the task of extracting open-domain assertions from natural language sentences. A key step in open IE is confidence modeling, ranking the extractions based on their estimated quality to adjust precision and recall of extracted assertions. We found that the extraction likelihood, a confidence measure used by current supervised open IE systems, is not well calibrated when comparing the quality of assertions extracted from different sentences. We propose an additional binary classification loss to calibrate the likelihood to make it more globally comparable, and an iterative learning process, where extractions generated by the open IE model are incrementally included as training samples to help the model learn from trial and error. Experiments on OIE2016 demonstrate the effectiveness of our method. Code and data are available at https://github.com/jzbjyb/oie_rank.
|
['Pengcheng Yin', 'Zhengbao Jiang', 'Graham Neubig']
|
2019-05-31
|
improving-open-information-extraction-via-1
|
https://aclanthology.org/P19-1523
|
https://aclanthology.org/P19-1523.pdf
|
acl-2019-7
|
['open-information-extraction']
|
['natural-language-processing']
|
[-8.37460831e-02 7.78083920e-01 -6.65131867e-01 -4.55267727e-01
-1.22560620e+00 -8.93911123e-01 5.76246977e-01 5.31872094e-01
-4.05901074e-01 1.13566875e+00 4.45440561e-01 -3.72772932e-01
1.98054351e-02 -6.82759821e-01 -8.22100341e-01 3.19009960e-01
1.12983063e-01 6.00810826e-01 2.01666951e-01 1.67197824e-01
3.91878545e-01 -1.33451805e-01 -1.09318757e+00 3.99357826e-01
1.32832050e+00 9.45987165e-01 -1.89264700e-01 6.09173536e-01
-2.95682222e-01 9.56574619e-01 -4.78619516e-01 -9.15118515e-01
1.81155905e-01 4.15611006e-02 -1.29201341e+00 -4.33712900e-01
1.44722402e-01 -8.10375884e-02 8.09920281e-02 1.20989156e+00
8.55287388e-02 -3.38337243e-01 9.80832934e-01 -1.45454228e+00
-7.00493276e-01 9.95606959e-01 -2.48103440e-01 2.64150321e-01
6.02234662e-01 -1.31899536e-01 1.47461057e+00 -1.26787639e+00
8.57557058e-01 8.35651577e-01 6.72946632e-01 2.91925669e-01
-1.19198072e+00 -8.44063580e-01 -1.11040600e-01 2.58383732e-02
-1.43173301e+00 -6.02280259e-01 3.54054034e-01 -5.85085213e-01
9.24800813e-01 1.21345744e-01 3.91090512e-01 7.94999301e-01
4.09985125e-01 6.47786736e-01 1.00604856e+00 -5.21045923e-01
2.01381132e-01 6.30190969e-01 4.46743220e-01 7.85287917e-01
8.69949102e-01 -3.73470336e-01 -5.92661560e-01 -6.04118764e-01
1.20026030e-01 -2.84439564e-01 -3.49558771e-01 -1.58114880e-01
-9.70055580e-01 7.37644076e-01 3.45048636e-01 -3.10068950e-03
-3.29973549e-01 -1.38848543e-01 4.23038363e-01 3.80069852e-01
9.23862517e-01 8.64325106e-01 -1.00303447e+00 -2.10221335e-01
-9.31861579e-01 2.75817066e-01 1.35305548e+00 1.20209873e+00
9.07726049e-01 -8.44203353e-01 -1.99340954e-01 9.83094692e-01
5.17475486e-01 3.43043476e-01 2.23443910e-01 -1.04840183e+00
7.68669724e-01 9.05913889e-01 3.63284260e-01 -8.48318994e-01
-1.74429432e-01 -1.14476472e-01 -1.45946428e-01 -2.10437089e-01
3.56491387e-01 -4.25440013e-01 -6.71480536e-01 1.49192500e+00
1.89203516e-01 -1.17471829e-01 3.55887800e-01 4.06585604e-01
1.05993938e+00 4.40675855e-01 2.02289909e-01 -1.05547532e-01
1.37852943e+00 -6.69029176e-01 -8.61315787e-01 -2.77389437e-01
7.13413775e-01 -7.45063663e-01 7.88221419e-01 2.79091716e-01
-9.73031521e-01 9.34623554e-02 -1.38630486e+00 -2.95031846e-01
-6.09762728e-01 4.43658978e-02 3.87676388e-01 2.25000590e-01
-4.61090744e-01 4.12445039e-01 -6.10249758e-01 -4.31984151e-03
4.99258369e-01 6.84534758e-02 -4.48313028e-01 7.63119534e-02
-1.62916028e+00 1.00144565e+00 5.72230041e-01 -1.03101589e-01
-5.50317407e-01 -9.02325869e-01 -1.10964501e+00 1.45435318e-01
8.03917408e-01 -5.41184425e-01 1.47756875e+00 -4.75041389e-01
-8.84643197e-01 7.77405381e-01 -2.89755702e-01 -6.45677149e-01
3.34570199e-01 -4.29766804e-01 -4.04331625e-01 1.81927681e-01
6.68622732e-01 6.03274643e-01 4.14915890e-01 -1.27255213e+00
-7.18552530e-01 -1.34331629e-01 1.39312834e-01 -1.15428604e-01
-1.66400030e-01 1.51475251e-01 -3.33586723e-01 -3.84215772e-01
6.68568760e-02 -7.92018652e-01 -6.57954952e-03 -2.34926254e-01
-4.70862597e-01 -5.49507201e-01 3.81617635e-01 -9.12644565e-01
1.58895826e+00 -1.67341125e+00 -2.50594825e-01 3.76571029e-01
6.05443597e-01 -2.10363194e-02 2.35640988e-01 3.11250269e-01
1.36530280e-01 7.99784660e-01 -3.39484096e-01 3.10034920e-02
-1.12333991e-01 -1.16931014e-01 -3.19829822e-01 4.09161188e-02
5.64482689e-01 7.31899381e-01 -1.06119871e+00 -8.89698625e-01
-5.61688900e-01 2.43309308e-02 -5.26403904e-01 4.86993529e-02
-5.51104307e-01 -1.50170466e-02 -4.83915031e-01 5.31317234e-01
4.03673500e-01 -5.97151458e-01 1.12540789e-01 -2.21764177e-01
7.12610707e-02 9.27680016e-01 -1.13455594e+00 1.22194636e+00
-4.72441733e-01 4.90682036e-01 -1.55825824e-01 -3.12943637e-01
8.80030096e-01 5.63986063e-01 2.73521811e-01 -1.28713921e-01
1.28356516e-01 3.89920801e-01 -2.13174418e-01 -4.07550693e-01
6.40534878e-01 1.64952844e-01 -3.10033083e-01 6.50823474e-01
5.35345197e-01 -3.83860826e-01 5.28383315e-01 7.20288157e-01
1.19172072e+00 7.61417905e-03 6.03316486e-01 -2.38260120e-01
3.83197457e-01 1.98034123e-01 9.48490739e-01 6.87709033e-01
-1.22636721e-01 3.80999446e-01 8.47625732e-01 -1.36703178e-01
-1.14220870e+00 -9.38408673e-01 -5.34888923e-01 6.19687736e-01
-1.61923438e-01 -9.70526099e-01 -6.48278534e-01 -1.21886611e+00
1.90794021e-01 9.22070980e-01 -5.38597703e-01 -3.01946979e-02
-7.95618165e-03 -3.29114228e-01 6.43427849e-01 4.49958801e-01
2.64366090e-01 -7.47961164e-01 -3.07018042e-01 2.78193410e-02
-7.11830437e-01 -1.13498366e+00 -4.03626114e-01 3.60539824e-01
-5.74368894e-01 -1.32659853e+00 -1.34803727e-01 -4.98331249e-01
7.68912435e-01 -4.86299157e-01 1.35262513e+00 1.83888018e-01
1.60659805e-01 1.13510244e-01 -4.99329239e-01 -6.64021194e-01
-7.71245897e-01 3.23499531e-01 1.87860847e-01 -4.99102563e-01
7.53286898e-01 -1.12249069e-01 -2.09172890e-01 1.41336888e-01
-7.34805703e-01 8.50546956e-02 3.93537045e-01 1.02949321e+00
4.49698836e-01 -1.30846471e-01 7.73989201e-01 -1.40115345e+00
9.10023391e-01 -7.66491294e-01 -5.66606402e-01 4.85031128e-01
-1.22384250e+00 2.96105653e-01 2.06656918e-01 -1.01260744e-01
-1.16901791e+00 -3.20374995e-01 -1.30398333e-01 -7.52225593e-02
1.24722362e-01 1.08711183e+00 8.44154134e-02 2.67096817e-01
6.89175189e-01 -4.47117001e-01 -1.37003824e-01 -2.77490288e-01
3.19995463e-01 1.19868326e+00 2.93519437e-01 -7.35373199e-01
6.48262620e-01 -1.22131230e-02 -5.37762702e-01 -3.19046676e-01
-1.34309006e+00 -2.97246248e-01 -6.19182408e-01 9.97608062e-03
4.91872966e-01 -1.13684237e+00 -3.04411083e-01 -1.50627598e-01
-1.10692811e+00 -1.66308254e-01 -2.18729958e-01 3.18465233e-01
-1.96686715e-01 1.25666007e-01 -6.86183214e-01 -6.57459259e-01
-6.79290593e-01 -8.16108048e-01 7.78466880e-01 3.19549680e-01
-6.44136906e-01 -8.81938875e-01 1.60792544e-01 5.90810239e-01
-1.44847974e-01 -1.31585047e-01 7.99175382e-01 -1.20519185e+00
-3.17406744e-01 -5.08320808e-01 -2.74459511e-01 4.19761479e-01
1.33722454e-01 3.66789877e-01 -8.24307859e-01 9.39818695e-02
-3.26431930e-01 -6.27325118e-01 7.20822990e-01 -3.24573368e-02
9.31191862e-01 -7.38324940e-01 -4.63699520e-01 1.70066357e-01
1.32544637e+00 -3.49203080e-01 4.92686898e-01 3.66286874e-01
3.40319365e-01 4.09850776e-01 9.89287972e-01 4.82972741e-01
6.21995389e-01 8.20671394e-02 -1.61212757e-01 3.92012656e-01
9.86161456e-02 -5.99624395e-01 1.16674855e-01 7.73595393e-01
2.36294478e-01 -6.29393458e-02 -1.19878757e+00 6.17788970e-01
-1.61741662e+00 -9.74669278e-01 1.78350508e-01 2.03661370e+00
1.52233839e+00 7.84863412e-01 -2.83163756e-01 -2.80024353e-02
5.25539756e-01 -1.28449813e-01 -2.66349316e-01 -3.51357222e-01
1.33649513e-01 1.57789201e-01 5.75432420e-01 7.27433205e-01
-8.10349286e-01 8.42748940e-01 6.41427755e+00 5.01900494e-01
-6.48854256e-01 1.50365070e-01 6.17970169e-01 1.44771233e-01
-6.91882908e-01 5.13939977e-01 -1.15927374e+00 5.47814965e-01
1.16159201e+00 -5.99891603e-01 -7.63798282e-02 8.20416510e-01
-4.84584868e-02 -3.53025347e-01 -1.14121521e+00 2.43483901e-01
-7.44803250e-02 -1.44736958e+00 -1.55855060e-01 -6.39537498e-02
7.40233064e-01 3.01058829e-01 -5.12150943e-01 5.51910996e-01
7.43208051e-01 -5.63026488e-01 7.83285856e-01 6.86929226e-01
7.51567960e-01 -5.41432679e-01 9.19024706e-01 4.81622100e-01
-9.22360361e-01 -1.01258889e-01 -2.02441767e-01 7.37517253e-02
-1.24963433e-01 8.60529542e-01 -1.31320918e+00 3.79939049e-01
7.74773419e-01 6.71683311e-01 -8.79136384e-01 7.66423404e-01
-5.84206283e-01 7.84230769e-01 -2.37596363e-01 -2.43918285e-01
-2.58095890e-01 8.50353949e-03 5.43400943e-01 1.27231967e+00
-2.11497933e-01 -2.24764887e-02 1.52436867e-01 1.24657810e+00
-5.45480013e-01 6.82075601e-03 -6.84583962e-01 -3.91387761e-01
9.64004517e-01 1.22454178e+00 -4.46018279e-01 -5.01661360e-01
-6.01261675e-01 4.09122705e-01 7.69006073e-01 2.34843463e-01
-7.31022954e-01 -7.82275617e-01 3.76120389e-01 -3.43000330e-02
2.95832187e-01 -3.20078917e-02 -5.16056478e-01 -1.39435172e+00
2.34393612e-01 -7.69667864e-01 6.98173463e-01 -9.06719029e-01
-1.47558296e+00 4.36212122e-01 1.78564057e-01 -1.16245210e+00
-4.44947273e-01 -4.09031212e-01 -2.79398769e-01 8.21292937e-01
-1.34989905e+00 -7.51904249e-01 1.14448266e-02 -9.37611237e-02
3.69862735e-01 1.45093247e-01 7.45620131e-01 9.70642120e-02
-4.72681224e-01 6.64151847e-01 -1.67752311e-01 6.65518105e-01
9.75283742e-01 -1.39004302e+00 4.73318994e-01 8.97126913e-01
2.98727155e-02 1.10047281e+00 8.83692026e-01 -1.22882605e+00
-8.57521176e-01 -8.74569654e-01 1.26948988e+00 -9.19274151e-01
8.97850275e-01 -4.00834113e-01 -1.00905609e+00 9.24368978e-01
2.22465649e-01 1.62811726e-01 9.30048347e-01 5.20668864e-01
-6.04963779e-01 9.78070274e-02 -1.25623560e+00 4.59163070e-01
6.72603250e-01 -7.08645940e-01 -1.13371122e+00 1.75408855e-01
9.93151009e-01 -4.16300684e-01 -1.39963758e+00 5.81788182e-01
5.67773879e-01 -3.44098330e-01 5.56890786e-01 -7.23133922e-01
6.77847028e-01 -3.13605636e-01 -4.13915850e-02 -1.31560898e+00
-9.79273114e-03 -1.94580123e-01 -4.75349158e-01 1.38513517e+00
1.37067580e+00 -7.31828570e-01 3.38654011e-01 1.03457201e+00
2.82871544e-01 -1.00343525e+00 -6.79342389e-01 -5.17424941e-01
6.77973926e-02 -4.19060141e-01 4.91313905e-01 9.19880569e-01
6.91287696e-01 5.86259007e-01 1.21228524e-01 3.40007961e-01
5.07282078e-01 -6.41915435e-03 5.94967067e-01 -1.48211420e+00
-7.93656148e-03 1.29455831e-02 -1.00895092e-01 -6.66766465e-01
3.58499497e-01 -9.22372341e-01 2.15758920e-01 -1.64273584e+00
5.10578752e-01 -7.34889567e-01 -2.22457901e-01 7.90347934e-01
-4.79657352e-01 3.81914675e-02 -2.96098366e-02 2.79740989e-01
-7.54233479e-01 4.35132414e-01 8.73879135e-01 -1.81279898e-01
-1.31309152e-01 -6.30182996e-02 -1.11419594e+00 8.87450397e-01
7.16314316e-01 -8.80422056e-01 -1.39035374e-01 -1.60860047e-01
7.14145362e-01 5.74640930e-02 1.62449151e-01 -8.90982330e-01
3.29209328e-01 -6.72942847e-02 4.54593599e-01 -5.56699216e-01
-1.01042211e-01 -6.13745809e-01 -2.60842860e-01 8.32106173e-02
-7.27285862e-01 -3.81494910e-02 1.57883897e-01 4.14231420e-01
-1.30452126e-01 -7.16685832e-01 2.93279618e-01 -8.77358019e-02
-3.96266043e-01 1.51358351e-01 -1.88022628e-01 7.73606837e-01
5.37015438e-01 1.88009620e-01 -5.51586807e-01 -2.69064575e-01
-5.72125733e-01 5.69614232e-01 3.78841609e-01 4.77486402e-01
5.35412669e-01 -9.59227204e-01 -7.84239829e-01 -9.74010527e-02
5.35552979e-01 7.84899816e-02 -4.14827019e-01 6.89555824e-01
-2.50901043e-01 3.98214787e-01 2.06308886e-01 -2.28512004e-01
-1.03792191e+00 3.30977708e-01 1.89763278e-01 -5.23508787e-01
-4.14029568e-01 7.17810154e-01 -5.53654671e-01 -7.65644729e-01
2.51297474e-01 -3.57900351e-01 -1.88118353e-01 -1.06311113e-01
5.69021046e-01 1.32920235e-01 2.15975672e-01 -2.20858276e-01
-3.59719723e-01 -1.78778276e-01 -4.60418254e-01 -3.72459322e-01
1.24001563e+00 -1.33803114e-01 -3.05059791e-01 5.65488458e-01
9.99559164e-01 4.10442114e-01 -8.29424918e-01 -4.16621923e-01
4.51022983e-01 -4.65170622e-01 5.79282194e-02 -1.03006017e+00
-3.21012169e-01 1.66600659e-01 1.75920296e-02 2.05229476e-01
5.41278541e-01 2.84888476e-01 4.69041944e-01 6.26737535e-01
3.69789600e-01 -1.29527164e+00 -2.82437772e-01 6.82964742e-01
1.00025284e+00 -1.60790837e+00 4.12847430e-01 -5.67138195e-01
-6.93813086e-01 9.10755694e-01 8.01302195e-01 2.64951140e-01
9.00336266e-01 6.21283591e-01 2.76801921e-02 -2.94888705e-01
-1.11737418e+00 2.10716903e-01 2.93904781e-01 1.22583821e-01
7.90185273e-01 -6.36461899e-02 -6.42057896e-01 9.84554291e-01
-1.74736425e-01 2.73655981e-01 7.12424815e-01 1.06256747e+00
-4.18033749e-01 -9.47012842e-01 -1.14036761e-01 1.10724437e+00
-5.06553769e-01 -3.20472270e-01 -4.21587825e-01 3.28097433e-01
-4.43220846e-02 1.05630302e+00 -1.12907201e-01 -4.08049554e-01
9.79535133e-02 4.69268978e-01 1.19275190e-01 -9.30186450e-01
-3.63403827e-01 -5.55988729e-01 7.15325236e-01 -2.88362056e-01
-2.26775423e-01 -6.74810410e-01 -1.46268129e+00 -1.18283585e-01
-8.02290380e-01 7.50225782e-01 5.84063113e-01 1.02606833e+00
5.35816610e-01 2.94012278e-01 4.76488441e-01 -9.58848745e-02
-7.30904698e-01 -1.19860590e+00 -1.98133573e-01 1.99528515e-01
3.22655916e-01 -7.18435824e-01 -5.33880472e-01 8.39275420e-02]
|
[9.454195976257324, 8.58975887298584]
|
b944e1a2-caab-418a-93d9-c8e2e01d2377
|
chemical-reaction-aware-molecule
|
2109.09888
| null |
https://arxiv.org/abs/2109.09888v2
|
https://arxiv.org/pdf/2109.09888v2.pdf
|
Chemical-Reaction-Aware Molecule Representation Learning
|
Molecule representation learning (MRL) methods aim to embed molecules into a real vector space. However, existing SMILES-based (Simplified Molecular-Input Line-Entry System) or GNN-based (Graph Neural Networks) MRL methods either take SMILES strings as input that have difficulty in encoding molecule structure information, or over-emphasize the importance of GNN architectures but neglect their generalization ability. Here we propose using chemical reactions to assist learning molecule representation. The key idea of our approach is to preserve the equivalence of molecules with respect to chemical reactions in the embedding space, i.e., forcing the sum of reactant embeddings and the sum of product embeddings to be equal for each chemical equation. This constraint is proven effective to 1) keep the embedding space well-organized and 2) improve the generalization ability of molecule embeddings. Moreover, our model can use any GNN as the molecule encoder and is thus agnostic to GNN architectures. Experimental results demonstrate that our method achieves state-of-the-art performance in a variety of downstream tasks, e.g., 17.4% absolute Hit@1 gain in chemical reaction prediction, 2.3% absolute AUC gain in molecule property prediction, and 18.5% relative RMSE gain in graph-edit-distance prediction, respectively, over the best baseline method. The code is available at https://github.com/hwwang55/MolR.
|
['Martin D. Burke', 'Jiawei Han', 'Heng Ji', 'Kyunghyun Cho', 'Xiaomeng Jin', 'Weijiang Li', 'Hongwei Wang']
|
2021-09-21
|
chemical-reaction-aware-molecule-1
|
https://openreview.net/forum?id=6sh3pIzKS-
|
https://openreview.net/pdf?id=6sh3pIzKS-
|
iclr-2022-4
|
['chemical-reaction-prediction']
|
['medical']
|
[ 4.57780123e-01 7.11935014e-02 -4.41704392e-01 -8.39473307e-02
-3.74068201e-01 -8.50471318e-01 6.65424228e-01 7.42164612e-01
-2.53646731e-01 9.55837309e-01 5.65654878e-03 -6.35530174e-01
6.38081506e-02 -1.13557208e+00 -1.06162381e+00 -8.11504126e-01
-6.70393556e-02 1.52348682e-01 -1.48783699e-01 -2.97999471e-01
3.17876875e-01 8.42184663e-01 -8.70580196e-01 1.16185188e-01
9.65083063e-01 7.97509909e-01 4.37660217e-02 5.72752953e-01
-1.08804077e-01 8.14982295e-01 -3.53329837e-01 -6.19040489e-01
2.11394027e-01 -5.03399730e-01 -6.63996220e-01 -7.86762178e-01
3.08207929e-01 1.71659991e-01 -7.28931129e-01 9.72919464e-01
6.20791018e-01 3.14489365e-01 9.56842482e-01 -8.13746214e-01
-1.00043023e+00 5.51522613e-01 -8.77022222e-02 -1.58138499e-01
4.15688574e-01 3.90121251e-01 1.24181771e+00 -8.67023408e-01
8.25390637e-01 9.81887639e-01 5.18730104e-01 6.54340506e-01
-1.49708283e+00 -6.29599631e-01 -3.92559059e-02 4.43376042e-02
-1.38134527e+00 -4.12499547e-01 5.59132397e-01 -5.64481139e-01
1.51544058e+00 3.87265295e-01 4.20809925e-01 1.05127001e+00
6.16146743e-01 1.30501956e-01 4.99982536e-01 -1.97429985e-01
2.98424423e-01 6.23693690e-02 2.07068883e-02 7.39809632e-01
4.74432826e-01 2.09758848e-01 -5.15254378e-01 -5.85321710e-02
5.57988405e-01 3.98000121e-01 -3.77685428e-01 -6.03439987e-01
-1.22392535e+00 1.10130847e+00 1.00344205e+00 1.22932591e-01
-3.01304042e-01 4.11379963e-01 4.69070703e-01 3.26038480e-01
2.54561961e-01 1.05027997e+00 -3.96610141e-01 6.75663128e-02
-3.81386548e-01 6.55886084e-02 7.62665093e-01 7.67130852e-01
6.61226690e-01 -2.53505297e-02 -4.06309962e-02 4.37985718e-01
1.74043521e-01 1.03763133e-01 4.67011660e-01 -2.57383794e-01
6.20100737e-01 6.38107061e-01 -1.45408019e-01 -9.46551263e-01
-3.41574013e-01 -4.58072931e-01 -1.00349891e+00 -2.97111664e-02
1.95634916e-01 1.17365889e-01 -9.35016930e-01 1.69811177e+00
2.26902470e-01 -7.43838996e-02 3.32345873e-01 4.55291599e-01
8.17505836e-01 9.23272908e-01 3.01432878e-01 -5.86841218e-02
9.69835103e-01 -1.06103313e+00 -3.97564232e-01 1.35960981e-01
1.19923341e+00 -4.22061950e-01 8.61720741e-01 1.13801792e-01
-9.08275902e-01 -4.91275221e-01 -1.31802440e+00 -3.08380842e-01
-1.01872933e+00 -1.13912880e-01 9.45682466e-01 6.64404571e-01
-5.03564000e-01 1.32419944e+00 -7.09341586e-01 2.45375317e-02
4.62259710e-01 6.20793998e-01 -7.02148855e-01 -2.86294341e-01
-1.46044397e+00 9.70989406e-01 7.56184697e-01 -9.06590745e-02
-6.73343599e-01 -1.15232217e+00 -9.64606762e-01 1.73476875e-01
2.06171215e-01 -7.34860897e-01 6.66848958e-01 -5.87314844e-01
-1.54006779e+00 2.51757681e-01 -1.34373516e-01 -5.23476720e-01
2.46700853e-01 1.38683781e-01 -3.55236679e-01 7.97415804e-03
-3.03994983e-01 6.81855381e-01 1.96850315e-01 -7.65787244e-01
-2.41143461e-02 -1.35345802e-01 2.70490259e-01 9.89955887e-02
-3.17025870e-01 -5.54846883e-01 4.59361076e-02 -6.29632711e-01
-1.51755214e-01 -7.87024140e-01 -5.44219732e-01 5.47863767e-02
-4.46247786e-01 -6.20844327e-02 2.05737919e-01 -4.78804588e-01
1.29530191e+00 -1.81291699e+00 4.75578398e-01 4.36879575e-01
4.48082983e-01 3.30208004e-01 -4.09268111e-01 1.04361033e+00
-7.39370942e-01 5.15297830e-01 -3.10245663e-01 -1.28178401e-02
-2.46412791e-02 -1.12149358e-01 -2.08091646e-01 4.98845160e-01
3.31063956e-01 1.22497141e+00 -1.18947148e+00 1.24038406e-01
2.82886446e-01 7.68511176e-01 -7.35325277e-01 5.25371470e-02
-3.45707387e-01 3.17161024e-01 -2.72596538e-01 6.04560137e-01
5.06611526e-01 -2.16496006e-01 4.23389256e-01 -4.63300675e-01
-1.65185377e-01 7.84668207e-01 -8.75206470e-01 1.74093413e+00
-4.85863924e-01 2.16082916e-01 -9.03183043e-01 -7.75976360e-01
7.71541834e-01 2.11520448e-01 3.75517815e-01 -6.44374967e-01
-1.19164959e-01 2.90630490e-01 3.06770295e-01 9.67303216e-02
2.61570036e-01 -2.87412077e-01 2.60041982e-01 3.74455154e-02
9.02566537e-02 9.36955214e-02 1.86017230e-01 2.39534780e-01
8.85977507e-01 -9.27733332e-02 6.01846218e-01 -7.28401244e-02
6.31935656e-01 -3.14319611e-01 3.48609239e-01 4.08438087e-01
3.84309232e-01 2.68105924e-01 7.26684690e-01 -4.49517936e-01
-1.10252666e+00 -8.56496871e-01 5.01732044e-02 7.18582928e-01
-9.92542058e-02 -7.83848166e-01 -3.55553597e-01 -7.45605826e-01
2.29699880e-01 8.16463411e-01 -7.34696090e-01 -5.72642505e-01
-4.78352338e-01 -7.45877028e-01 5.88838518e-01 4.88226563e-01
-6.37776479e-02 -7.38551199e-01 1.56022713e-01 3.64204079e-01
3.79762679e-01 -4.51815575e-01 -6.33649766e-01 5.43589294e-01
-6.95765734e-01 -1.07392991e+00 -3.77817065e-01 -4.89477932e-01
7.38171577e-01 1.52447775e-01 6.40196919e-01 -1.35009691e-01
-4.01565552e-01 -3.00092965e-01 -1.70079976e-01 -1.44002691e-01
-4.52351928e-01 2.80693591e-01 2.43887845e-02 -1.42664760e-01
1.05610192e-01 -6.05001569e-01 -7.76309907e-01 -4.91424650e-02
-9.17886972e-01 -6.80268928e-02 5.14958978e-01 9.61541772e-01
7.04527080e-01 -2.33621836e-01 5.91418803e-01 -1.01059091e+00
5.06071627e-01 -5.47492385e-01 -5.53192437e-01 2.92813897e-01
-9.08509791e-01 3.86867285e-01 9.84795511e-01 -4.64899212e-01
-3.04684788e-01 2.18088150e-01 -2.23636165e-01 -3.06384891e-01
2.52907574e-01 7.52021611e-01 -4.85234886e-01 -3.96983504e-01
7.07066774e-01 3.73286664e-01 -9.50350836e-02 -3.12812150e-01
6.88185930e-01 2.50344455e-01 5.66147454e-03 -5.11968911e-01
7.05619693e-01 6.35379180e-02 5.24883986e-01 -5.61717749e-01
-2.81711638e-01 -4.02854472e-01 -7.33248413e-01 4.26659405e-01
8.16838264e-01 -8.59709322e-01 -1.07132697e+00 -9.44610313e-02
-1.14362431e+00 -2.32747868e-01 -1.25955909e-01 4.41653222e-01
-3.76887232e-01 6.61711693e-01 -6.71789348e-01 -4.74024713e-01
-6.27533138e-01 -1.23818147e+00 5.10210514e-01 -9.24076065e-02
-3.62322479e-01 -1.14783430e+00 9.76423025e-02 -6.74676299e-02
2.68961251e-01 6.47944450e-01 1.34619761e+00 -8.14518034e-01
-6.09750211e-01 -2.99697459e-01 -8.51013064e-02 3.00642788e-01
5.64973295e-01 -2.18254924e-02 -8.26607823e-01 -4.10391152e-01
-6.55339301e-01 -7.91761652e-02 1.15361893e+00 1.20886564e-02
1.13999689e+00 -5.28666615e-01 -5.59329093e-01 6.99818432e-01
1.62447381e+00 4.69231874e-01 7.84642041e-01 1.02875523e-01
1.15727079e+00 4.23744082e-01 1.46776780e-01 1.49619669e-01
-8.17720219e-02 7.07381725e-01 5.21011114e-01 -6.64061308e-02
-7.14670941e-02 -8.56825352e-01 4.89473850e-01 6.88083529e-01
-4.76836562e-02 -6.28804922e-01 -6.99196100e-01 1.99014440e-01
-1.67704225e+00 -8.85387599e-01 -9.26355049e-02 2.45434189e+00
9.94191468e-01 2.30380654e-01 5.46802767e-02 4.66256589e-02
2.65986860e-01 9.28516984e-02 -7.85964191e-01 -6.65787160e-01
-6.25390336e-02 5.57340980e-01 8.86354029e-01 7.23415434e-01
-9.32715774e-01 9.91968632e-01 5.23672676e+00 9.61927295e-01
-1.25516224e+00 -9.79447514e-02 5.58472455e-01 -4.27121967e-02
-4.70331460e-01 -3.84013914e-02 -7.16995895e-01 5.19359887e-01
1.22491336e+00 -2.25376382e-01 5.10564089e-01 5.91014087e-01
-1.53418649e-02 5.70838213e-01 -1.61489141e+00 9.16400194e-01
-1.99651316e-01 -1.95269549e+00 6.30867541e-01 2.12417871e-01
4.75744516e-01 -1.12526178e-01 9.70913097e-02 1.33007884e-01
-4.41656262e-02 -1.60031343e+00 4.79566216e-01 4.55716938e-01
1.16243696e+00 -1.01049948e+00 4.85085070e-01 -5.28196879e-02
-1.22313321e+00 2.16704145e-01 -5.44013798e-01 2.31862273e-02
-5.52914292e-02 5.40837944e-01 -1.01249540e+00 9.47237074e-01
-1.88004464e-01 1.01898503e+00 -3.70047897e-01 8.78388643e-01
-1.89614132e-01 3.12983990e-01 -1.01480275e-01 -3.29187453e-01
4.22284156e-01 -4.60349262e-01 2.46636823e-01 1.38142967e+00
1.58702448e-01 4.46994379e-02 -6.64122216e-03 8.27084184e-01
-4.51202184e-01 3.49015445e-01 -8.65923882e-01 -6.28621817e-01
3.93482596e-01 8.44300985e-01 -2.28043675e-01 -2.06550702e-01
-2.47478321e-01 1.06432974e+00 2.70714909e-01 3.25451374e-01
-8.62186909e-01 -8.82132292e-01 1.19080210e+00 -5.73513769e-02
3.83816421e-01 -2.37313390e-01 -9.01496708e-02 -9.77283895e-01
-1.23763278e-01 -8.18294168e-01 1.58261701e-01 -3.54579270e-01
-1.14824963e+00 4.65219021e-01 -5.24725795e-01 -1.00089014e+00
5.16962893e-02 -1.05209565e+00 -5.29035866e-01 1.06926334e+00
-1.66432655e+00 -9.77228105e-01 2.91627377e-01 1.24275669e-01
3.04708332e-01 -1.02377068e-02 1.14480519e+00 4.46877807e-01
-7.77308345e-01 9.36404526e-01 4.20872688e-01 -1.32286981e-01
7.00697541e-01 -1.21110094e+00 7.21296906e-01 3.66597265e-01
6.58095181e-02 1.29789805e+00 3.93894523e-01 -6.16507232e-01
-1.77157331e+00 -1.45786536e+00 1.01310670e+00 -6.44179642e-01
6.89975202e-01 -7.98578262e-01 -8.40438068e-01 3.81916195e-01
-2.26905331e-01 -1.51526824e-01 1.06540501e+00 -8.65674019e-02
-9.63438451e-01 -2.91556790e-02 -1.02588463e+00 7.90894806e-01
1.15478218e+00 -6.20142758e-01 2.17260551e-02 7.30398715e-01
1.15167928e+00 -3.16878080e-01 -1.33695352e+00 2.91127354e-01
5.92202961e-01 -2.89810807e-01 1.35525358e+00 -1.25326574e+00
5.96337676e-01 -4.37533110e-01 -2.52948195e-01 -1.25081384e+00
-5.77671945e-01 -6.10142052e-01 -3.18433613e-01 7.51821280e-01
9.35999274e-01 -1.01510537e+00 5.83338082e-01 3.09410959e-01
-6.32866025e-02 -1.14835048e+00 -7.05955327e-01 -1.00894809e+00
3.26724082e-01 8.19872245e-02 6.09323561e-01 1.16911316e+00
2.68924505e-01 6.89369380e-01 -2.70320565e-01 1.70204401e-01
2.32990906e-01 3.56389064e-04 4.85401064e-01 -1.05841815e+00
-3.65354925e-01 -6.66381180e-01 -5.20759642e-01 -1.08766210e+00
8.45394358e-02 -1.54326308e+00 -5.95390737e-01 -1.51057541e+00
6.03681095e-02 -2.79796153e-01 -7.52658486e-01 6.82779312e-01
9.15039517e-03 -1.05610609e-01 1.44979686e-01 -3.43315937e-02
-3.02467018e-01 7.09388673e-01 1.00517905e+00 -4.77384806e-01
-1.12670928e-01 -2.81197667e-01 -8.40735734e-01 -1.05096646e-01
9.89271343e-01 -4.40944076e-01 -5.20303011e-01 -5.08405007e-02
4.75636899e-01 -2.30469540e-01 8.20676684e-02 -6.16189122e-01
1.59702629e-01 -2.91978359e-01 4.69349444e-01 -1.33037239e-01
4.86974031e-01 -5.06814659e-01 4.39210862e-01 9.52990770e-01
-4.21899647e-01 -5.77129088e-02 2.81335115e-01 8.49641562e-01
3.62330712e-02 -1.99766070e-01 6.36135459e-01 -4.59902100e-02
-1.56852335e-01 6.05019689e-01 -1.96365900e-02 -5.28052509e-01
8.22212577e-01 -2.42714956e-01 -4.00660396e-01 2.97658853e-02
-2.97042817e-01 -4.15330753e-02 6.34834468e-01 2.49204710e-01
5.80592394e-01 -1.23119640e+00 -4.12067592e-01 1.49508389e-02
3.59386325e-01 -3.12253654e-01 -6.92284182e-02 5.68354845e-01
-6.95112824e-01 8.78047168e-01 -3.80593129e-02 7.67865106e-02
-1.15638721e+00 8.93821299e-01 4.49793011e-01 -3.22218865e-01
-2.89718121e-01 9.28614557e-01 4.40197617e-01 -4.99272466e-01
8.17427784e-02 -4.68508273e-01 1.41258404e-01 8.21893942e-03
4.28914100e-01 2.08505109e-01 2.55583227e-01 -1.85015023e-01
-5.26667356e-01 5.17210007e-01 -3.68897170e-01 5.61476648e-01
1.32119894e+00 7.17079103e-01 -1.55855596e-01 3.04738849e-01
1.69195175e+00 6.62444010e-02 -8.28279972e-01 4.32182364e-02
-1.37713686e-01 -1.24647215e-01 -1.65827647e-01 -9.13439035e-01
-6.26456618e-01 1.04889965e+00 3.88752431e-01 -2.06352174e-01
5.40353477e-01 -2.78871506e-01 7.33658552e-01 8.02388191e-01
1.16600413e-02 -4.79661971e-01 -8.19385648e-02 3.86403888e-01
8.39397967e-01 -1.03095078e+00 1.85031846e-01 -5.76900780e-01
-3.45638007e-01 1.29935265e+00 2.27431044e-01 1.03789434e-01
3.36254358e-01 -2.92731941e-01 -5.91085672e-01 -2.55874157e-01
-7.97052264e-01 7.38462731e-02 2.57686406e-01 4.29789186e-01
7.87917435e-01 2.01863796e-01 -4.17803943e-01 3.43254536e-01
-8.78497139e-02 -3.90726000e-01 3.14968199e-01 7.58924127e-01
-2.64964521e-01 -1.37844145e+00 3.33973765e-01 2.65091926e-01
-2.10399345e-01 -6.32217050e-01 -7.36210406e-01 5.76379299e-01
3.01185232e-02 6.01070940e-01 -3.67472708e-01 -4.47358221e-01
5.56422412e-01 1.96916670e-01 5.86674631e-01 -6.95284307e-01
-5.73160231e-01 -4.54860717e-01 8.17472041e-02 -4.20577735e-01
1.18516535e-01 -1.47869870e-01 -1.34130669e+00 -3.88805687e-01
-4.14346397e-01 3.25174689e-01 7.17598617e-01 2.78183520e-01
7.63855636e-01 5.92995346e-01 6.32729352e-01 -5.03836870e-01
-5.84004939e-01 -7.19177127e-01 -2.70858288e-01 1.70726970e-01
2.95098037e-01 -4.94028926e-01 -2.61440933e-01 -1.64009437e-01]
|
[4.996126651763916, 5.905185222625732]
|
bc7d91dc-72aa-4d7a-b8db-945f64c2577f
|
transliterating-kurdish-texts-in-latin-into
|
2110.12374
| null |
https://arxiv.org/abs/2110.12374v1
|
https://arxiv.org/pdf/2110.12374v1.pdf
|
Transliterating Kurdish texts in Latin into Persian-Arabic script
|
Kurdish is written in different scripts. The two most popular scripts are Latin and Persian-Arabic. However, not all Kurdish readers are familiar with both mentioned scripts that could be resolved by automatic transliterators. So far, the developed tools mostly transliterate Persian-Arabic scripts into Latin. We present a transliterator to transliterate Kurdish texts in Latin into Persian-Arabic script. We also discuss the issues that should be considered in the transliteration process. The tool is a part of Kurdish BLARK, and it is publicly available for non-commercial use
|
['Hossein Hassani']
|
2021-10-24
| null | null | null | null |
['transliteration']
|
['natural-language-processing']
|
[-2.91692674e-01 -5.23865260e-02 3.43357660e-02 -3.34051728e-01
-3.29070151e-01 -1.23433232e+00 5.97182691e-01 -4.67849299e-02
-4.64732468e-01 1.04985189e+00 8.69832933e-02 -8.52553487e-01
-1.87151767e-02 -5.68136811e-01 9.66547336e-03 -3.13462973e-01
6.04170442e-01 9.99504209e-01 -9.60639566e-02 -6.98152959e-01
4.91171539e-01 7.75575340e-01 -4.59921747e-01 -1.55189678e-01
1.30980253e+00 -3.25981677e-01 3.70410420e-02 7.72165358e-01
-8.50809216e-02 7.47159600e-01 -7.24049628e-01 -6.39205575e-01
2.15486601e-01 -8.84988666e-01 -1.29307401e+00 7.44538158e-02
1.79402679e-01 -6.09449029e-01 -5.94426766e-02 9.09475982e-01
3.35111201e-01 -3.91930729e-01 7.57004380e-01 -6.70179904e-01
-9.34034407e-01 1.24247134e+00 -4.17826384e-01 2.77089085e-02
5.40077627e-01 -1.89669758e-01 4.68553960e-01 -1.09229612e+00
8.28975201e-01 1.34894681e+00 3.66717309e-01 1.38299644e-01
-6.05729043e-01 -5.90351939e-01 -4.95775849e-01 4.42558415e-02
-1.71029794e+00 -1.70712605e-01 3.49219859e-01 -4.49039727e-01
7.44752347e-01 5.09877026e-01 6.82305336e-01 5.83228350e-01
2.21523300e-01 3.33816350e-01 1.77624047e+00 -7.59123802e-01
-2.07498476e-01 3.93421113e-01 1.66027263e-01 4.31840152e-01
6.38312995e-01 -5.84028721e-01 -3.45318496e-01 3.99744585e-02
7.32161403e-01 -4.43916053e-01 8.97705629e-02 8.37259829e-01
-1.05637872e+00 7.10095167e-01 -1.85428128e-01 8.36647868e-01
-2.32964978e-02 -4.25714582e-01 3.47724348e-01 2.96777070e-01
2.60548383e-01 4.45344895e-01 -2.23861367e-01 -6.56443179e-01
-1.10803866e+00 1.98090866e-01 8.59673858e-01 9.64839041e-01
6.97948933e-01 3.23029637e-01 3.38408381e-01 9.24735188e-01
6.00997567e-01 1.16343105e+00 4.30401444e-01 -6.49442732e-01
4.66579646e-01 4.54865575e-01 2.96490908e-01 -8.64134729e-01
-1.10247314e-01 8.48694369e-02 -2.29001656e-01 5.33839352e-02
6.59107864e-01 -4.79009688e-01 -7.86358893e-01 5.67733943e-01
2.48285666e-01 -1.00071347e+00 1.28551736e-01 1.04553199e+00
7.62166440e-01 7.81320274e-01 -6.94569424e-02 -3.41569006e-01
1.41678894e+00 -7.95884967e-01 -1.04578507e+00 -3.45705539e-01
2.91704625e-01 -1.68535316e+00 1.05890441e+00 5.09926736e-01
-1.18204057e+00 -8.26158449e-02 -1.01229167e+00 4.99884449e-02
-3.89745086e-01 5.55145919e-01 2.31111888e-03 1.26490986e+00
-7.82588542e-01 3.39087367e-01 -8.00523221e-01 -1.08146513e+00
-2.60788172e-01 1.59842610e-01 -3.37821811e-01 1.05253190e-01
-1.16684389e+00 1.37227023e+00 6.90380991e-01 3.23371708e-01
-2.28563949e-01 3.23769629e-01 -6.45963132e-01 -7.85743654e-01
8.47011656e-02 2.27780044e-01 1.42390621e+00 -1.28779078e+00
-1.78172994e+00 1.29235268e+00 -4.55632396e-02 1.52307332e-01
8.35987985e-01 -4.51787382e-01 -6.05800986e-01 2.94271767e-01
2.37055421e-01 5.85594997e-02 6.24765635e-01 -1.09752810e+00
-5.29791057e-01 -1.60584778e-01 -1.21204376e-01 1.93801254e-01
-6.25309274e-02 1.10117579e+00 -4.77611154e-01 -1.05011761e+00
9.39389542e-02 -9.80898738e-01 2.62361407e-01 -4.90977436e-01
-3.11756611e-01 -1.51623696e-01 7.64276803e-01 -1.59453237e+00
1.51088274e+00 -1.73416710e+00 -3.20449948e-01 2.54969448e-01
-1.94852665e-01 6.26036644e-01 1.14945821e-01 1.07840908e+00
-6.24109544e-02 3.44341129e-01 -2.11835012e-01 3.35001707e-01
1.55243412e-01 4.00406241e-01 -1.56101584e-01 7.37679422e-01
3.06242913e-01 5.65034389e-01 -1.18244243e+00 -5.07159114e-01
4.36979420e-02 3.91874731e-01 4.03459489e-01 -3.46182883e-02
1.33625001e-01 5.30043364e-01 -3.66150200e-01 9.20318723e-01
1.05385458e+00 4.42835182e-01 6.02131724e-01 4.04162854e-01
-6.08982265e-01 4.99436229e-01 -1.03429508e+00 9.54362333e-01
-6.21122830e-02 6.60978079e-01 -2.23307684e-01 -4.00656313e-01
1.17391777e+00 3.79885525e-01 -1.95270956e-01 -2.94609755e-01
3.55649620e-01 1.08490539e+00 1.42382860e-01 -4.61620688e-01
1.08996892e+00 -1.79802537e-01 -3.05540841e-02 9.29816544e-01
-2.83763856e-01 -5.29923677e-01 9.74317849e-01 -1.55673120e-02
1.47782877e-01 5.10715008e-01 8.88125539e-01 -4.02042359e-01
6.56570554e-01 4.01354432e-01 3.90164226e-01 3.12875122e-01
-6.08148947e-02 5.21622598e-01 4.57115710e-01 -2.45599002e-01
-1.20824957e+00 -9.64654803e-01 -2.76848257e-01 9.15701151e-01
-4.83062327e-01 -7.21218765e-01 -1.12557161e+00 -5.50524294e-01
-6.58883393e-01 8.31998229e-01 -2.38438711e-01 5.01527250e-01
-8.24203610e-01 -5.13635516e-01 9.77944672e-01 6.94526285e-02
6.95000231e-01 -1.00665402e+00 -6.91312432e-01 2.64208853e-01
-2.35941634e-01 -9.61197197e-01 -3.98425609e-01 -1.96073696e-01
-5.23015440e-01 -9.64403391e-01 -1.27607143e+00 -9.20422554e-01
8.10025930e-01 7.94463754e-02 6.66279674e-01 1.21027334e-02
8.65775645e-02 3.15454006e-01 -8.86614799e-01 -3.83424610e-01
-1.10642290e+00 1.20288953e-01 -6.02430515e-02 -5.36177158e-01
4.95154589e-01 -2.57185381e-02 4.06551212e-02 1.85986713e-01
-8.95002842e-01 3.29132751e-02 4.57824230e-01 9.73777100e-02
4.20325547e-02 -1.53288200e-01 7.17045516e-02 -1.26315904e+00
5.00662923e-01 -3.73746663e-01 -3.86777133e-01 3.03765476e-01
-5.38723111e-01 -2.86610693e-01 8.84622991e-01 -2.70265073e-01
-1.03332806e+00 -2.14418933e-01 -2.81508029e-01 6.26204669e-01
-9.20641869e-02 7.42571712e-01 -7.10262102e-05 -9.35866833e-02
7.70160496e-01 2.41412118e-01 -1.87392816e-01 -3.63376737e-01
3.30712736e-01 1.31366885e+00 2.81250238e-01 -4.41830784e-01
1.04498935e+00 7.30502233e-02 -3.27717513e-01 -1.13123608e+00
-1.07184149e-01 -1.03453724e-02 -8.57022047e-01 -4.30176437e-01
6.15600705e-01 -7.31530309e-01 -1.12330385e-01 8.39993119e-01
-1.22846830e+00 -4.56654131e-01 1.30613789e-01 3.19705337e-01
-2.48932600e-01 4.48658049e-01 -7.22604334e-01 -8.16104293e-01
-4.52334195e-01 -8.90193880e-01 4.15871680e-01 6.82415903e-01
-8.92418325e-01 -1.00072074e+00 2.78602540e-01 2.57508099e-01
2.61978626e-01 1.53047875e-01 8.06060433e-01 -6.50214374e-01
1.71162277e-01 -1.28606788e-03 -2.79970676e-01 1.79683506e-01
6.63596272e-01 7.62197435e-01 -4.17268693e-01 -1.33145034e-01
-1.88462198e-01 -2.17970029e-01 -1.03932850e-01 -1.76755637e-01
-1.77025720e-01 -8.43449056e-01 3.28364581e-01 -2.52944194e-02
1.30475819e+00 4.48554337e-01 6.06337249e-01 7.89864600e-01
3.66360188e-01 6.11354351e-01 1.00545728e+00 5.41737854e-01
5.15398502e-01 2.64206767e-01 -3.34903419e-01 1.98627710e-01
-9.84184816e-02 -9.37802494e-02 1.05405247e+00 1.31263483e+00
-3.32582176e-01 1.62314042e-01 -1.64179957e+00 5.45992315e-01
-1.55432737e+00 -5.38323581e-01 -7.69864082e-01 1.88479662e+00
1.23891020e+00 -1.99049532e-01 4.82220709e-01 2.53223419e-01
5.58816612e-01 -1.44953519e-01 2.31380999e-01 -1.06612372e+00
-4.71922457e-01 4.80313599e-01 4.56965774e-01 9.58516955e-01
-9.76731241e-01 1.41791749e+00 7.01731586e+00 5.61284602e-01
-1.50120771e+00 5.22143440e-03 1.20104164e-01 4.98781055e-01
-3.87802839e-01 4.24312472e-01 -7.67527461e-01 2.23575413e-01
1.04191494e+00 -6.35352433e-01 4.86136854e-01 4.05944973e-01
7.08261549e-01 -2.75548249e-01 -4.32845175e-01 6.50614142e-01
4.12117064e-01 -9.65340316e-01 1.20278731e-01 -3.39451820e-01
7.65544295e-01 -1.65099412e-01 -1.04603700e-01 -9.15465802e-02
4.55912292e-01 -1.04843140e+00 1.13681579e+00 2.58784682e-01
8.44688714e-01 -7.28813708e-01 7.53750920e-01 -8.11214969e-02
-7.25249290e-01 3.42653096e-01 -4.55331445e-01 -3.00943077e-01
6.68431818e-02 3.26476216e-01 -1.18780625e+00 6.58836007e-01
5.07496655e-01 5.95032215e-01 -1.06478906e+00 7.91514993e-01
-8.59549165e-01 9.60943580e-01 -1.73660725e-01 -3.37872356e-01
5.07031381e-01 -9.23742115e-01 4.50718671e-01 1.48622835e+00
6.27104998e-01 -2.87442617e-02 -1.08287901e-01 5.01798809e-01
6.82838440e-01 8.11912417e-01 -4.52226907e-01 -8.24178219e-01
3.40209246e-01 1.03102648e+00 -1.15208662e+00 -2.63104111e-01
-3.28348607e-01 1.37957776e+00 -1.52082205e-01 5.05786479e-01
-6.50469005e-01 -7.89790452e-01 6.20292015e-02 4.65854317e-01
-1.05616398e-01 -6.32987678e-01 -3.01804185e-01 -7.26953268e-01
-1.97427437e-01 -1.50904667e+00 2.62144685e-01 -8.88684034e-01
-8.08704913e-01 9.07430589e-01 1.90304428e-01 -1.22572339e+00
-4.19512957e-01 -8.61748397e-01 -2.70932019e-01 1.12988639e+00
-8.92949164e-01 -1.60324883e+00 1.03249244e-01 3.79839502e-02
3.68857771e-01 -2.16130316e-01 9.71302092e-01 2.68300653e-01
-5.94255924e-01 3.57763320e-01 3.59329671e-01 2.75337636e-01
1.21916687e+00 -1.19523609e+00 2.81933010e-01 1.22086346e+00
-1.36484668e-01 9.71826255e-01 1.04117143e+00 -8.84105325e-01
-8.59657824e-01 -8.64754379e-01 1.72134209e+00 -2.17586070e-01
9.67442572e-01 6.78774640e-02 -4.59287673e-01 9.96462941e-01
9.39258635e-01 -8.56344998e-01 9.15571570e-01 -5.81306577e-01
-1.49713710e-01 1.48521826e-01 -9.17398512e-01 1.07328832e+00
1.60573691e-01 -5.04905164e-01 -8.24637949e-01 4.55502450e-01
1.29748777e-01 -4.07124817e-01 -7.08580792e-01 -3.46930474e-01
5.57916522e-01 -5.91291726e-01 1.96516141e-01 -3.03857684e-01
3.89334977e-01 -7.92609990e-01 2.61069983e-01 -1.17159319e+00
-4.75174561e-02 -1.36864662e+00 5.84178269e-01 1.50002122e+00
3.67089123e-01 -9.14321661e-01 1.58737212e-01 3.50384384e-01
2.21631378e-01 2.22135499e-01 -6.64335132e-01 -6.56173050e-01
5.99036038e-01 1.74326953e-02 2.08839774e-01 1.09592009e+00
3.40478063e-01 3.73955250e-01 -4.16274130e-01 -1.62691608e-01
1.09451272e-01 -1.43478721e-01 5.97006381e-01 -6.91849470e-01
-4.60182838e-02 -3.79543602e-01 -7.09683374e-02 -5.52419722e-01
-1.36496976e-01 -7.33166933e-01 -1.27165288e-01 -1.74794912e+00
-2.92885378e-02 -1.41071290e-01 6.91165984e-01 7.66946733e-01
-1.63290992e-01 6.55925810e-01 4.90237862e-01 2.59858787e-01
-1.18289879e-02 -4.87572290e-02 1.07402027e+00 2.29701772e-02
-4.32709545e-01 -2.80588418e-01 -7.18416929e-01 7.89164126e-01
1.27616179e+00 -6.53247356e-01 -1.06034622e-01 -4.13641572e-01
8.11338544e-01 -4.82398987e-01 -3.57368469e-01 -6.45583451e-01
-1.37201622e-01 -7.21397340e-01 5.39646521e-02 -6.41004920e-01
-3.98764670e-01 -5.60948431e-01 4.74016219e-01 5.56310773e-01
4.06813532e-01 6.89368665e-01 3.66472721e-01 -7.25326180e-01
-4.02772687e-02 -7.99209356e-01 8.47305775e-01 -1.17329754e-01
-4.38106209e-01 -3.41870338e-01 -1.45639932e+00 -1.23549448e-02
1.10960662e+00 -4.58331078e-01 -3.03628266e-01 -4.71763074e-01
-3.30475152e-01 -1.10038035e-01 8.93774569e-01 6.70659728e-03
3.71791095e-01 -1.29980826e+00 -1.13553047e+00 -1.15138628e-01
-1.58084482e-01 -4.92428988e-01 -1.99043244e-01 9.99309361e-01
-2.12221408e+00 5.55420518e-01 -6.94421589e-01 1.92409456e-01
-1.25772500e+00 2.70808656e-02 5.66714592e-02 2.58854800e-03
-2.11912408e-01 2.84546077e-01 -6.37842953e-01 -7.43599474e-01
-3.16797167e-01 1.56502351e-01 -4.41353559e-01 1.92793444e-01
6.25536501e-01 6.62162244e-01 -1.38146088e-01 -1.29480875e+00
-6.14165962e-01 8.08871508e-01 -5.43546975e-02 -7.13491917e-01
9.73864257e-01 -4.76398855e-01 -9.73746002e-01 8.20103109e-01
4.23714429e-01 8.61750305e-01 -2.59074420e-01 1.88037068e-01
-8.51384830e-03 -4.02905583e-01 -7.39127040e-01 -9.39087272e-01
-2.23846719e-01 5.84324837e-01 9.95706394e-02 5.99227175e-02
1.00689709e+00 -4.56644982e-01 7.06464231e-01 4.51192141e-01
1.07777171e-01 -1.65197229e+00 -3.56010407e-01 1.22596550e+00
1.06313872e+00 -6.72603786e-01 2.88429618e-01 -4.74423766e-01
-9.02742863e-01 1.79134262e+00 3.08709323e-01 -9.36761573e-02
4.42911953e-01 4.68462817e-02 9.99067485e-01 2.65572786e-01
1.01536483e-01 -3.39686610e-02 6.44043237e-02 6.60490155e-01
1.11027873e+00 9.49082598e-02 -1.61714900e+00 6.34359896e-01
-8.20451856e-01 -1.01260282e-01 1.30579972e+00 1.10340798e+00
-2.30771407e-01 -1.60919714e+00 -1.20018685e+00 -1.89005479e-01
-5.77278376e-01 -1.29002765e-01 -1.09130716e+00 9.74718928e-01
6.08835518e-02 1.02772582e+00 -1.77502245e-01 -1.22503608e-01
-1.37117818e-01 1.90485157e-02 5.92089057e-01 -5.35820365e-01
-9.15933132e-01 4.92214024e-01 5.20229101e-01 2.63221622e-01
-3.55078548e-01 -7.33822942e-01 -1.38505185e+00 -1.06378829e+00
2.83657521e-01 4.22265500e-01 3.69916826e-01 1.00159621e+00
-5.06704271e-01 -1.66295424e-01 3.13915163e-01 -2.92496443e-01
-2.94394910e-01 -1.25657022e+00 -7.40787804e-01 -1.62948236e-01
-1.27823517e-01 1.22588560e-01 -3.39934193e-02 2.83938199e-01]
|
[10.52642822265625, 10.521719932556152]
|
853fb49d-3e6f-417e-8126-627f78706c9f
|
nms-threshold-matters-for-ego4d-moment
|
2307.02025
| null |
https://arxiv.org/abs/2307.02025v1
|
https://arxiv.org/pdf/2307.02025v1.pdf
|
NMS Threshold matters for Ego4D Moment Queries -- 2nd place solution to the Ego4D Moment Queries Challenge 2023
|
This report describes our submission to the Ego4D Moment Queries Challenge 2023. Our submission extends ActionFormer, a latest method for temporal action localization. Our extension combines an improved ground-truth assignment strategy during training and a refined version of SoftNMS at inference time. Our solution is ranked 2nd on the public leaderboard with 26.62% average mAP and 45.69% Recall@1x at tIoU=0.5 on the test set, significantly outperforming the strong baseline from 2023 challenge. Our code is available at https://github.com/happyharrycn/actionformer_release.
|
['Yin Li', 'Fangzhou Mu', 'Lin Sui']
|
2023-07-05
| null | null | null | null |
['action-localization', 'action-recognition']
|
['computer-vision', 'computer-vision']
|
[-2.25943699e-01 2.19834372e-01 -7.24344015e-01 -2.70907074e-01
-1.01144385e+00 -7.16716111e-01 8.26496065e-01 -5.21170557e-01
-8.13640654e-01 1.01411700e+00 1.03362775e+00 3.13722372e-01
1.80654034e-01 -2.85913825e-01 -7.87350476e-01 -3.92632157e-01
-4.03652698e-01 4.45889771e-01 3.52176666e-01 -1.04119420e-01
-1.48664378e-02 1.57100677e-01 -1.00060225e+00 5.85406363e-01
1.98966682e-01 1.02871943e+00 1.77349910e-01 7.74067163e-01
5.63014984e-01 1.24163151e+00 -4.49944586e-01 -1.05345733e-01
4.85804796e-01 -5.28427586e-02 -1.13329649e+00 -3.06538761e-01
7.60936618e-01 -6.48108006e-01 -1.15812838e+00 9.56149280e-01
7.78653681e-01 5.46713173e-01 4.44437824e-02 -1.27208471e+00
-4.43524837e-01 7.67662823e-01 -4.85388130e-01 6.71317816e-01
5.92610955e-01 4.16861624e-01 1.01991844e+00 -8.99763942e-01
9.48496163e-01 1.20954859e+00 6.09409153e-01 8.09963703e-01
-8.53309214e-01 -6.01955533e-01 5.28799415e-01 7.00804651e-01
-1.54265010e+00 -8.58339489e-01 1.04188338e-01 -3.40931654e-01
1.46724761e+00 -1.14782348e-01 6.96146429e-01 1.76845098e+00
1.73882172e-01 1.45221901e+00 8.11028183e-01 4.54065561e-01
9.43084583e-02 -5.44301808e-01 -1.39808595e-01 5.40742338e-01
-2.24351913e-01 -1.29491789e-02 -1.22726512e+00 1.14200339e-01
7.27482736e-01 -3.15624267e-01 -2.67166495e-01 1.25446811e-01
-1.75126731e+00 2.69029737e-01 6.53855801e-01 1.31100342e-01
-7.21402168e-01 1.16071796e+00 4.53701794e-01 -5.97946234e-02
6.21852756e-01 1.23825669e-01 -5.59951663e-01 -8.81118536e-01
-7.09186137e-01 5.18995166e-01 3.11572015e-01 1.10183883e+00
2.82336295e-01 -2.61379685e-02 -6.32916629e-01 2.99551457e-01
1.45350337e-01 6.44908547e-01 3.01007092e-01 -1.67231727e+00
7.46244252e-01 1.56017542e-01 6.00622654e-01 -7.25974381e-01
-5.88791847e-01 -5.87391198e-01 -4.77767050e-01 -1.74441367e-01
3.86282593e-01 -3.40678036e-01 -9.23473179e-01 2.00298595e+00
2.22274974e-01 5.68536282e-01 -1.62566844e-02 1.08296192e+00
9.41529930e-01 3.78292650e-01 2.87261866e-02 3.27938408e-01
9.28749263e-01 -1.52932227e+00 -9.36951280e-01 -4.22516763e-01
7.74808705e-01 -3.38361830e-01 7.21838117e-01 3.18274587e-01
-1.08897769e+00 -4.07007188e-01 -8.24920058e-01 -2.01554969e-01
-3.07153791e-01 4.02270317e-01 8.42250168e-01 -7.80408159e-02
-1.54908347e+00 5.82068145e-01 -1.26654172e+00 -5.74312866e-01
6.84611440e-01 2.07721278e-01 -6.17406309e-01 2.43100468e-02
-1.35319781e+00 1.10601807e+00 3.84836376e-01 3.11995029e-01
-1.37232649e+00 -4.53991026e-01 -6.37067080e-01 -4.50545102e-01
4.67545152e-01 -5.74157238e-01 1.97267485e+00 -4.50073421e-01
-1.56985760e+00 7.43511856e-01 -2.87828803e-01 -1.01613259e+00
1.05681884e+00 -7.97930181e-01 -3.07146370e-01 1.29622743e-01
5.77673614e-01 1.16513598e+00 6.03580847e-03 -3.94217163e-01
-8.66952777e-01 -1.17006756e-01 3.53436112e-01 4.06866401e-01
2.20225677e-01 -2.04594731e-01 -8.37413371e-01 -2.77767122e-01
2.11732537e-01 -1.20386875e+00 -3.67465019e-01 -1.41203150e-01
-4.38297898e-01 -4.07825351e-01 2.97817290e-01 -6.76645398e-01
8.63848984e-01 -1.83898997e+00 2.98153549e-01 -3.28398496e-01
1.87387839e-01 -1.45750195e-01 -4.03916508e-01 6.17343485e-01
1.01591617e-01 -1.95369810e-01 2.97034234e-01 -8.06482017e-01
3.82486373e-01 -8.51746500e-02 -1.76711947e-01 9.49762881e-01
-3.02141756e-01 1.39732468e+00 -1.39022624e+00 -1.42601803e-01
3.24553281e-01 3.28650147e-01 -2.94387609e-01 -2.29358613e-01
-1.99584410e-01 6.75200045e-01 -4.97064829e-01 8.31204355e-01
2.38581628e-01 -3.61721307e-01 -2.40700059e-02 -1.89781874e-01
-1.99952930e-01 4.72150564e-01 -1.01910460e+00 2.63182616e+00
-3.50538194e-02 1.01866496e+00 -9.10836011e-02 -3.10152233e-01
3.39666367e-01 3.37260008e-01 8.28729033e-01 -8.58770370e-01
1.02740593e-01 -1.70066074e-01 -2.40712404e-01 -3.50411266e-01
6.26932740e-01 5.20863175e-01 -3.19280714e-01 2.35833868e-01
3.25167656e-01 3.22043896e-01 2.85668015e-01 5.70393384e-01
1.72155440e+00 8.96752357e-01 -9.42732859e-03 -2.45427430e-01
1.36719123e-01 -2.15553436e-02 6.96617782e-01 1.02647781e+00
-8.02947819e-01 5.78281999e-01 4.38258171e-01 -6.76272988e-01
-3.66659105e-01 -1.17968798e+00 2.65209407e-01 1.18245983e+00
1.63834065e-01 -8.11720610e-01 -5.28515399e-01 -8.78123045e-01
-1.25515774e-01 8.31127822e-01 -9.28109229e-01 1.20941989e-01
-4.97844666e-01 -4.36273217e-01 1.00947440e+00 8.12570453e-01
8.78873110e-01 -1.05131865e+00 -3.78583491e-01 3.40153575e-02
-7.24531472e-01 -1.50378788e+00 -5.50739288e-01 1.97211117e-03
-6.81885183e-01 -1.22505474e+00 -6.18785620e-01 -1.30640224e-01
3.11904907e-01 2.53664672e-01 1.02578425e+00 -5.08518398e-01
2.65715152e-01 5.57085812e-01 -4.08269614e-01 -2.28843912e-01
3.92023921e-01 4.19397652e-01 4.76743221e-01 -2.00654596e-01
5.01182199e-01 -6.34190202e-01 -8.89680326e-01 2.38812074e-01
-2.49997869e-01 1.77214175e-01 3.80959213e-01 2.23458841e-01
6.52198434e-01 -5.32750547e-01 4.10657078e-01 -1.14749528e-01
3.05363566e-01 -6.58204556e-01 -5.62753797e-01 -1.46780089e-01
-1.44850805e-01 -2.54702270e-01 -1.75400913e-01 -1.51592135e-01
-8.80918205e-01 4.06098157e-01 -1.07118577e-01 -4.94786322e-01
-2.45268494e-01 1.97680265e-01 1.75950099e-02 1.91297278e-01
7.63429344e-01 1.97603300e-01 -3.32406491e-01 -5.26324928e-01
5.65937996e-01 1.33084327e-01 7.85533428e-01 -3.39674652e-01
3.80905747e-01 8.13259482e-01 -2.38798246e-01 -3.81921858e-01
-1.21712816e+00 -6.23630345e-01 -8.39589238e-01 -6.70561075e-01
1.11725581e+00 -1.46629632e+00 -1.00136375e+00 6.38944030e-01
-1.22119904e+00 -1.06963253e+00 -1.90487280e-01 8.07536125e-01
-7.71295726e-01 -3.37267704e-02 -6.32971406e-01 -3.68283123e-01
-2.64147401e-01 -8.99782181e-01 1.19067311e+00 1.25020057e-01
-3.92458797e-01 -7.12891817e-01 4.46670890e-01 4.82056469e-01
4.23471302e-01 4.62575823e-01 -6.34574115e-01 -4.01514888e-01
-7.33700752e-01 -3.81286711e-01 -4.29235324e-02 1.11508161e-01
1.55064166e-02 -4.96471137e-01 -1.01450753e+00 -2.34682813e-01
-6.65635228e-01 -6.16763830e-01 1.24596250e+00 7.16802716e-01
9.39842999e-01 -8.58139992e-02 -5.25710523e-01 5.77471554e-01
9.02673602e-01 -1.92208558e-01 6.65999711e-01 6.92594826e-01
4.92106080e-01 4.83242683e-02 8.68120551e-01 5.81572056e-01
7.40466297e-01 9.72365379e-01 7.57554054e-01 3.18396300e-01
-3.72928649e-01 -4.52965528e-01 6.86104119e-01 2.16979712e-01
-4.87399489e-01 -4.79445249e-01 -1.01531839e+00 7.44708002e-01
-2.39892054e+00 -1.33562684e+00 -1.03369139e-01 1.83584857e+00
5.61264157e-01 1.73485905e-01 2.30996490e-01 -8.86491001e-01
4.30651069e-01 7.93492317e-01 -6.50751472e-01 4.03234810e-01
-2.87424088e-01 -3.30066025e-01 9.41868663e-01 8.65123570e-01
-1.49748337e+00 1.46351540e+00 6.32201433e+00 6.30329728e-01
-6.31100237e-01 5.92758596e-01 2.10506752e-01 -9.57047760e-01
3.36755991e-01 -1.03280105e-01 -9.37986851e-01 4.40606862e-01
9.40103292e-01 -2.16935560e-01 4.77907181e-01 8.48833025e-01
6.91796243e-01 -4.56080556e-01 -1.09065294e+00 1.12962449e+00
-9.64229405e-02 -1.49894059e+00 -5.99264264e-01 1.24398850e-01
1.02698064e+00 1.24226856e+00 -7.26665705e-02 4.71239060e-01
6.73023522e-01 -9.36774313e-01 9.43667829e-01 9.96924698e-01
5.45657694e-01 -4.34164494e-01 8.51528347e-01 2.04728201e-01
-1.19363964e+00 1.05998561e-01 -1.21096991e-01 -3.84086460e-01
7.24054575e-01 3.63135338e-02 -7.97055542e-01 2.66166508e-01
9.04326141e-01 1.28611648e+00 -5.77452242e-01 1.15799844e+00
-8.19877565e-01 4.68381047e-01 -4.20682073e-01 4.68813926e-02
7.35137641e-01 2.17407778e-01 8.07757795e-01 1.28986943e+00
1.24390379e-01 2.42116913e-01 3.30227047e-01 3.50950181e-01
-2.85851389e-01 -5.40036023e-01 -6.72571063e-01 2.81215981e-02
5.82367480e-01 1.12241387e+00 -3.95758539e-01 -4.34442848e-01
1.92118213e-02 1.31320024e+00 4.09026146e-01 5.44535041e-01
-1.24052620e+00 4.66081761e-02 9.87716675e-01 -1.74625546e-01
1.16786160e-01 -6.03471279e-01 1.81221649e-01 -1.13021374e+00
1.99720580e-02 -4.62207228e-01 3.50334048e-01 -1.04446411e+00
-6.45131707e-01 4.88103449e-01 1.85245052e-01 -1.22391999e+00
-3.28025073e-01 -1.97287664e-01 -4.02190417e-01 5.17666638e-01
-1.02830243e+00 -1.16161942e+00 -3.75097841e-01 5.10288596e-01
6.37232423e-01 -3.05347070e-02 6.65948987e-01 4.56294507e-01
-5.11104167e-01 4.56770927e-01 -7.70580173e-02 1.97204173e-01
7.09299564e-01 -1.29746354e+00 1.02140796e+00 9.07758296e-01
8.27388912e-02 2.23207220e-01 7.57012427e-01 -7.40484834e-01
-1.34085953e+00 -1.25626791e+00 1.05866241e+00 -1.12775922e+00
1.00659764e+00 -4.03890997e-01 -9.93572250e-02 1.37570179e+00
5.33865571e-01 7.38668814e-02 2.21150190e-01 2.13041097e-01
9.66473520e-02 8.57704431e-02 -6.83294415e-01 7.36697376e-01
1.69487417e+00 -4.28898811e-01 -3.46407622e-01 1.03168821e+00
6.70645356e-01 -8.32616329e-01 -7.38995612e-01 3.14528137e-01
6.76991761e-01 -6.54778659e-01 7.97685862e-01 -6.94257617e-01
7.55305365e-02 -5.35437584e-01 -4.94190931e-01 -9.87277210e-01
-5.62355995e-01 -7.31438696e-01 -6.11008644e-01 5.37746966e-01
4.87551481e-01 -3.26646805e-01 1.20959771e+00 1.93096682e-01
-4.08945441e-01 -7.61826813e-01 -1.31790411e+00 -1.03532910e+00
-4.83473986e-01 -7.69385338e-01 8.83715972e-02 6.26265943e-01
-1.70504242e-01 1.94558516e-01 -7.87301898e-01 2.38600612e-01
6.47471726e-01 -4.23620850e-01 1.07840085e+00 -4.14812684e-01
-3.02135348e-01 -3.13546747e-01 -5.12936056e-01 -1.64370215e+00
2.09077939e-01 -8.48172903e-01 1.64592862e-01 -2.10080957e+00
7.67975748e-02 2.83892959e-01 -4.85025167e-01 1.03462040e+00
4.01551545e-01 6.03606701e-01 2.25740612e-01 -4.89343377e-03
-1.51763773e+00 9.59084570e-01 1.06153631e+00 -2.31169492e-01
-4.70179990e-02 3.36032957e-02 -3.23072076e-01 8.78449082e-01
1.23017180e+00 -5.32991588e-01 -3.57942998e-01 -9.23285127e-01
1.56501412e-01 -1.94295809e-01 6.31412089e-01 -1.40661871e+00
4.49595571e-01 -2.58862317e-01 4.25855905e-01 -1.08327806e+00
8.46019804e-01 -1.05602823e-01 1.98616222e-01 5.31407833e-01
-4.04466063e-01 2.63973147e-01 3.30426216e-01 6.47787035e-01
1.18714400e-01 4.30858761e-01 1.77641973e-01 -3.15806866e-01
-1.39697003e+00 5.17515600e-01 -3.09347361e-01 1.24079376e-01
8.09983432e-01 1.92695390e-02 -7.09759295e-01 -7.36999094e-01
-1.13851166e+00 7.15803444e-01 1.50179356e-01 8.89504731e-01
3.77855211e-01 -1.69560111e+00 -8.49385142e-01 -7.49417543e-01
1.21758217e-02 -3.40834945e-01 4.62107629e-01 1.46721435e+00
-2.63745844e-01 8.14540029e-01 -1.42360777e-01 -5.65188944e-01
-9.08652961e-01 -1.46916807e-01 7.03736842e-01 1.09288236e-02
-6.45190775e-01 1.32622802e+00 -1.05786391e-01 -3.89456540e-01
5.75701892e-01 -6.65678158e-02 4.70944867e-02 3.72908683e-03
7.93456912e-01 7.17685282e-01 -2.42093965e-01 -7.09284425e-01
-9.16071892e-01 1.14097089e-01 9.76193175e-02 -5.47621608e-01
1.24842238e+00 -1.30291581e-01 1.45990908e-01 4.55096126e-01
9.44079816e-01 -1.84035480e-01 -1.61889184e+00 -2.37386540e-01
-4.33020405e-02 -3.83173347e-01 1.96392030e-01 -1.16793728e+00
-8.78051758e-01 2.76665390e-01 5.91966927e-01 -4.30612862e-01
6.71744823e-01 3.46817046e-01 5.59710324e-01 9.68271136e-01
6.84873044e-01 -1.42078662e+00 8.35011899e-02 9.04079974e-01
1.27927744e+00 -1.27057099e+00 2.00225800e-01 2.80660808e-01
-6.04012430e-01 5.14782190e-01 9.17836666e-01 -2.54466623e-01
3.34668368e-01 -6.02265820e-03 8.94740894e-02 -4.65847850e-01
-1.11327589e+00 -4.42747802e-01 2.21947894e-01 3.69052559e-01
1.66098937e-01 1.40033573e-01 -2.62453347e-01 3.71502310e-01
-2.35383332e-01 9.03669596e-02 3.30781192e-01 9.36905563e-01
-4.13780183e-01 -5.58614612e-01 2.32259974e-01 1.08811915e-01
-3.88177782e-01 -7.98157081e-02 -7.03889310e-01 8.81414175e-01
1.43966928e-01 9.32487071e-01 -3.44435126e-02 -8.02372098e-01
4.28201854e-01 -5.15162647e-02 6.06649399e-01 -4.76229489e-01
-2.38742530e-01 -1.16858639e-01 6.19654536e-01 -1.70414865e+00
-6.27413929e-01 -1.02410042e+00 -1.32586908e+00 -4.26835150e-01
1.71148166e-01 -1.05356693e-01 5.20073950e-01 7.46015310e-01
7.83256352e-01 5.31406343e-01 1.17524430e-01 -1.29983628e+00
-2.82493591e-01 -1.33357978e+00 -2.36667603e-01 -2.30659977e-01
2.79217750e-01 -8.00910473e-01 -2.18711689e-01 -1.84576124e-01]
|
[8.354452133178711, 0.41934242844581604]
|
988fdff7-9129-4b05-b4f5-96271be8c4c8
|
disentangling-confidence-score-distribution
|
2210.08830
| null |
https://arxiv.org/abs/2210.08830v1
|
https://arxiv.org/pdf/2210.08830v1.pdf
|
Disentangling Confidence Score Distribution for Out-of-Domain Intent Detection with Energy-Based Learning
|
Detecting Out-of-Domain (OOD) or unknown intents from user queries is essential in a task-oriented dialog system. Traditional softmax-based confidence scores are susceptible to the overconfidence issue. In this paper, we propose a simple but strong energy-based score function to detect OOD where the energy scores of OOD samples are higher than IND samples. Further, given a small set of labeled OOD samples, we introduce an energy-based margin objective for supervised OOD detection to explicitly distinguish OOD samples from INDs. Comprehensive experiments and analysis prove our method helps disentangle confidence score distributions of IND and OOD data.\footnote{Our code is available at \url{https://github.com/pris-nlp/EMNLP2022-energy_for_OOD/}.}
|
['Weiran Xu', 'Yuanmeng Yan', 'Pei Wang', 'Yutao Mou', 'Keqing He', 'Zhiyuan Zeng', 'Yanan Wu']
|
2022-10-17
| null | null | null | null |
['intent-detection']
|
['natural-language-processing']
|
[-3.07829171e-01 5.68233848e-01 -6.05334222e-01 -1.03710794e+00
-9.57673311e-01 -5.53659320e-01 6.07668221e-01 2.36173496e-01
-3.40917617e-01 9.07416165e-01 2.59604782e-01 -4.02872205e-01
1.97080255e-01 -3.01830262e-01 -1.61085278e-01 -2.44719416e-01
2.16287121e-01 6.76446021e-01 2.18701679e-02 1.16174258e-01
2.07287222e-01 -8.51627514e-02 -1.13745534e+00 1.33253083e-01
1.19668925e+00 9.05258000e-01 2.97612399e-01 7.60871828e-01
-2.58231848e-01 5.36078811e-01 -9.19627905e-01 -4.67227906e-01
3.07753105e-02 -3.36766899e-01 -7.87748694e-01 -2.04562470e-01
3.19806218e-01 -3.41681868e-01 -3.37886214e-01 1.29177928e+00
5.77034533e-01 3.49457055e-01 9.76581037e-01 -1.55829954e+00
-5.86088777e-01 5.30196726e-01 -4.55307007e-01 2.97235996e-01
5.71541071e-01 -7.57712871e-02 1.23992777e+00 -1.02580881e+00
3.75986844e-01 1.37783015e+00 2.42217034e-01 7.17397928e-01
-1.14068460e+00 -9.79357421e-01 1.31878793e-01 6.35588821e-03
-1.19625437e+00 -4.61015880e-01 8.37265730e-01 -3.03047806e-01
9.58656132e-01 3.75797749e-01 1.38228446e-01 1.48938537e+00
1.46851137e-01 1.19499409e+00 1.17393661e+00 -4.47005451e-01
4.66001093e-01 7.37031221e-01 7.36367583e-01 7.82860160e-01
3.85967374e-01 1.25835896e-01 -7.63590097e-01 -6.96626842e-01
2.32980236e-01 -5.85890636e-02 -2.07172185e-01 3.73128876e-02
-8.11437905e-01 1.20988822e+00 1.03836983e-01 2.76162058e-01
2.00149417e-02 -3.00513417e-01 -7.14583471e-02 3.52227062e-01
7.41413832e-01 3.72961670e-01 -8.20455313e-01 -2.36558422e-01
-6.63542569e-01 1.85491323e-01 1.14453316e+00 1.07099366e+00
6.41721666e-01 -2.89474040e-01 -2.79798806e-01 1.15436971e+00
8.50612521e-01 3.47644389e-01 5.07187605e-01 -1.08870995e+00
3.93137783e-01 5.93064487e-01 2.63273388e-01 -7.26866484e-01
-3.69641930e-01 3.38809527e-02 -5.51651537e-01 2.54773170e-01
6.09296620e-01 -5.82131982e-01 -6.66905880e-01 1.94521034e+00
5.33915937e-01 -3.42012495e-01 5.26714604e-03 8.46892953e-01
9.71369445e-01 4.62511122e-01 1.71393245e-01 -3.02729577e-01
1.58771312e+00 -5.36212385e-01 -1.09702742e+00 -5.59299648e-01
6.17592931e-01 -6.61749184e-01 1.10969067e+00 2.52575904e-01
-9.07447278e-01 -4.67215538e-01 -1.16316664e+00 -1.41669765e-01
-3.08447540e-01 -2.07991013e-03 5.87410033e-01 6.86580241e-01
-5.94560206e-01 2.80085742e-01 -5.17234266e-01 -1.80956185e-01
2.62520075e-01 1.88106537e-01 2.73359045e-02 1.05356872e-01
-1.55750072e+00 8.84108126e-01 5.33933699e-01 -2.96599299e-01
-4.40508395e-01 -3.51383120e-01 -8.99571776e-01 -2.68183015e-02
5.82991362e-01 -2.28340283e-01 1.64811540e+00 -5.10980844e-01
-1.29061294e+00 8.99531007e-01 -5.19707799e-01 -2.06686512e-01
5.76184928e-01 -2.54959375e-01 -4.57775235e-01 -1.82326153e-01
1.28550008e-01 7.63003230e-01 8.21363628e-01 -1.10361493e+00
-5.94142854e-01 -4.23777938e-01 -2.11947113e-01 2.53629506e-01
-3.03269416e-01 4.77472730e-02 -2.28026897e-01 -6.06548786e-01
1.03195384e-01 -8.12289476e-01 1.72914028e-01 -1.01996176e-01
-8.79397452e-01 -1.04107153e+00 5.46187997e-01 -5.07860005e-01
1.27788365e+00 -2.06491947e+00 -4.61658895e-01 6.04911000e-02
5.54780364e-01 3.89317088e-02 2.96669006e-01 1.71661586e-01
1.53585672e-01 2.83255190e-01 -1.45088986e-01 -1.95006207e-01
7.17056036e-01 4.13650237e-02 -4.27901521e-02 9.27750021e-02
3.93245488e-01 4.87192899e-01 -7.59442091e-01 -7.86659658e-01
1.06124856e-01 3.43345940e-01 -3.58052015e-01 6.54073536e-01
-5.09726048e-01 1.50193140e-01 -5.66701591e-01 6.39662743e-01
7.86943316e-01 -4.18822050e-01 3.67897302e-01 -1.03366785e-01
2.79163778e-01 6.89601958e-01 -1.23161542e+00 1.33583784e+00
-1.85366377e-01 7.05895483e-01 7.37708583e-02 -5.99531353e-01
8.86921942e-01 4.30117488e-01 1.06396060e-02 -3.34056079e-01
4.32237118e-01 2.05321103e-01 -5.83507977e-02 -3.59817684e-01
5.74910045e-01 1.00745171e-01 -4.03091401e-01 5.01091003e-01
4.15895581e-01 -7.54335672e-02 -3.91383208e-02 2.80255795e-01
8.33621740e-01 -2.49464214e-01 4.02469426e-01 -2.53642917e-01
9.43906978e-02 -2.11014315e-01 6.47063375e-01 8.98270667e-01
-5.73210120e-01 1.90716818e-01 6.84954584e-01 1.78498477e-01
-5.13617754e-01 -1.22688007e+00 -5.76613069e-01 1.13051343e+00
1.49854213e-01 -2.46044591e-01 -6.64480984e-01 -1.13579023e+00
1.34568229e-01 1.41957974e+00 -3.27038884e-01 -1.06187522e-01
-1.64062247e-01 -8.51703763e-01 4.47785467e-01 2.53608286e-01
4.14196581e-01 -7.56333232e-01 -1.48110136e-01 -5.35800494e-02
-2.72092998e-01 -9.45879459e-01 -8.32324445e-01 7.63907015e-01
-5.73534489e-01 -1.02082241e+00 -5.26466966e-01 -8.00581157e-01
5.72522938e-01 -1.18055686e-01 1.06817472e+00 -1.97878063e-01
5.56069985e-02 2.01538369e-01 -8.16417038e-02 -7.86466718e-01
-7.06898451e-01 -2.21450135e-01 3.07692587e-01 -4.56961930e-01
1.30934072e+00 -1.95956454e-01 -4.57584769e-01 5.06080985e-01
-4.91913795e-01 -3.10672551e-01 2.73611724e-01 7.90619671e-01
2.36240864e-01 -9.90292430e-02 8.91703069e-01 -8.45721364e-01
1.25226641e+00 -6.85781598e-01 -5.77524483e-01 2.15837121e-01
-1.04233563e+00 3.12581211e-01 5.24095967e-02 -7.40353763e-01
-1.32310998e+00 -1.48058146e-01 -1.84706524e-01 -2.20190883e-01
-6.20204866e-01 2.80784488e-01 -2.66253918e-01 7.06727266e-01
5.44629574e-01 -1.99356556e-01 -1.94283769e-01 -5.53786218e-01
1.16750821e-01 1.32991230e+00 1.66175067e-01 -4.83556241e-01
4.23550814e-01 -2.28952821e-02 -8.38745475e-01 -1.01621962e+00
-9.84830201e-01 -4.86410320e-01 -2.56772876e-01 -1.03938244e-01
1.03658009e+00 -8.49458992e-01 -9.27619159e-01 2.47268498e-01
-1.08583057e+00 -4.58382368e-01 -1.28721252e-01 8.73876572e-01
-1.05684526e-01 4.20055091e-01 -6.48685515e-01 -1.29204404e+00
-1.94696829e-01 -7.70312369e-01 8.16810250e-01 6.52807832e-01
-9.19734240e-01 -1.10700703e+00 1.23349555e-01 6.99568391e-01
-3.16413902e-02 -1.65720120e-01 6.75058186e-01 -1.63644958e+00
-1.05380304e-01 -4.45755303e-01 -1.59748960e-02 3.59895557e-01
1.87430784e-01 -2.51644045e-01 -1.21560431e+00 -8.17365944e-02
4.31080073e-01 -7.89506555e-01 7.57817626e-01 3.91402692e-01
9.15061712e-01 -3.78134310e-01 -3.89678687e-01 -2.20725194e-01
7.49783456e-01 3.04841965e-01 -2.52261888e-02 -1.99148819e-01
1.12858847e-01 6.17128730e-01 9.53333318e-01 6.83326304e-01
3.91281486e-01 6.78609848e-01 4.56251577e-02 2.47701794e-01
2.55525619e-01 -3.11052173e-01 4.99822110e-01 5.08437335e-01
4.64749068e-01 -8.95368278e-01 -5.92713356e-01 2.59258449e-01
-1.61728048e+00 -7.50708103e-01 -1.73560008e-01 2.13975739e+00
1.25771236e+00 5.48557341e-01 1.06518883e-02 -2.22819805e-01
9.64996338e-01 2.91241109e-01 -8.17472458e-01 -3.76258850e-01
9.41459760e-02 -3.96594293e-02 1.19100869e-01 9.74387527e-01
-1.02886367e+00 5.80378354e-01 5.62773037e+00 9.45931852e-01
-3.91131192e-01 2.72950411e-01 6.35595620e-01 -3.05929314e-02
-4.77932453e-01 -2.18980536e-01 -1.23625672e+00 6.82463586e-01
1.03168380e+00 -1.18325189e-01 -2.03998275e-02 8.94589961e-01
1.75641552e-01 -4.20695364e-01 -1.09241617e+00 6.19676888e-01
-8.37591961e-02 -6.50610864e-01 -7.05906391e-01 1.75037220e-01
3.63213778e-01 2.65683681e-01 -1.66010082e-01 5.90397239e-01
9.41026390e-01 -7.08440006e-01 3.17987412e-01 2.19481349e-01
5.89243412e-01 -3.58500838e-01 6.95710182e-01 6.13543272e-01
-7.39449501e-01 3.33620250e-01 -3.85704637e-01 3.85346860e-01
6.23050728e-04 1.01780808e+00 -1.09653890e+00 -3.09088593e-03
6.82083130e-01 9.59064737e-02 -2.99758255e-01 3.63310397e-01
-4.96183366e-01 7.52555788e-01 -4.19823050e-01 -6.41370773e-01
-1.72903463e-01 -6.87360298e-03 7.23833501e-01 1.13434160e+00
2.71043368e-02 2.01083019e-01 1.23352610e-01 1.38281846e+00
-3.79656076e-01 -2.97889143e-01 -4.12468225e-01 -2.32128099e-01
8.52502227e-01 1.19801486e+00 -4.99558866e-01 -3.89479876e-01
-3.71036023e-01 1.00198054e+00 3.53533715e-01 2.61099428e-01
-9.82261419e-01 -6.45590901e-01 6.30250454e-01 -3.84350032e-01
-5.22459447e-02 -8.05058554e-02 3.49667505e-03 -1.32471001e+00
-1.12895109e-02 -6.59420371e-01 6.14587426e-01 -7.31643677e-01
-1.61241698e+00 3.76448393e-01 5.63935414e-02 -9.10476625e-01
-4.27840412e-01 -6.30275965e-01 -7.23781288e-01 8.68069410e-01
-1.19433832e+00 -3.78811538e-01 -2.94175237e-01 2.36352190e-01
9.26158607e-01 -4.91664931e-02 8.21323931e-01 1.18528798e-01
-4.03152823e-01 8.13810289e-01 1.04837738e-01 3.15671474e-01
1.15440750e+00 -1.52828264e+00 1.75081298e-01 3.57670754e-01
2.32987832e-02 6.19357169e-01 8.61921072e-01 -9.53988433e-01
-8.48421574e-01 -6.45126879e-01 1.33055890e+00 -7.48699546e-01
4.44710463e-01 -6.05560005e-01 -9.99259710e-01 5.53350925e-01
4.92925584e-01 -4.70368505e-01 1.00620115e+00 2.23654807e-01
-2.45540336e-01 5.30078232e-01 -1.50991201e+00 4.27558303e-01
8.12923133e-01 -5.02456844e-01 -1.09934378e+00 6.36309862e-01
8.89827251e-01 -2.87759632e-01 -9.35490310e-01 2.04541489e-01
5.51643431e-01 -1.02519619e+00 8.39162409e-01 -4.27507222e-01
4.27974621e-03 7.62774721e-02 -3.11807007e-01 -1.10185254e+00
1.66710913e-02 -4.40274060e-01 -5.68390489e-01 1.41918325e+00
6.69439137e-01 -8.65422726e-01 8.27706397e-01 1.00423789e+00
5.43275960e-02 -5.18606424e-01 -8.81575108e-01 -7.67078876e-01
-5.56157678e-02 -5.71699142e-01 -6.04649400e-03 1.23905051e+00
4.42366600e-01 5.44800520e-01 -2.40479589e-01 2.50159860e-01
8.11089516e-01 -4.50679809e-02 3.33875507e-01 -1.42303598e+00
-3.89710724e-01 -1.57020226e-01 3.00133884e-01 -1.27419293e+00
2.75145084e-01 -7.89600432e-01 1.93861216e-01 -1.20086336e+00
7.50382841e-02 -2.93127149e-01 -2.25445151e-01 4.19073761e-01
-3.99263829e-01 -1.15323707e-01 -2.44980678e-01 1.51602834e-01
-6.09664202e-01 6.61181509e-01 6.41869903e-01 -1.98639393e-01
-4.16552931e-01 2.67579198e-01 -7.18698978e-01 9.14602041e-01
1.09241176e+00 -9.66088772e-01 -3.65865350e-01 1.11272320e-01
-1.56786352e-01 1.61561862e-01 1.28340289e-01 -4.94994849e-01
9.05271992e-03 -2.97186166e-01 4.68585938e-01 -6.99962914e-01
5.91362834e-01 -5.12196600e-01 -3.86445045e-01 6.04643583e-01
-7.89254189e-01 -4.25624579e-01 -6.32993951e-02 8.09225857e-01
-5.54827452e-02 -6.32049620e-01 7.10167706e-01 -1.10805944e-01
-3.30555648e-01 -1.22593269e-01 -7.92956054e-01 6.90435231e-01
6.62368238e-01 7.05920532e-02 -4.15359408e-01 -6.59236670e-01
-6.43069863e-01 5.85749447e-01 2.66850535e-02 6.16269588e-01
5.22343695e-01 -1.06862748e+00 -5.82605183e-01 1.63883299e-01
1.38015762e-01 -1.70116603e-01 1.11207269e-01 7.67894447e-01
1.90230161e-01 4.60724473e-01 3.54578286e-01 -4.64771390e-01
-1.53190541e+00 1.82150960e-01 2.29434073e-01 -2.00847536e-01
8.03872272e-02 1.01266563e+00 7.54960626e-02 -9.81464744e-01
7.31943905e-01 -3.07852626e-01 -3.80088836e-02 1.52753636e-01
2.44361356e-01 3.46078902e-01 -2.92127132e-01 -8.97188485e-02
-5.10299385e-01 -2.25524202e-01 -1.88384667e-01 -2.28593379e-01
9.95940924e-01 -1.91685796e-01 4.81503993e-01 6.09800935e-01
1.16528201e+00 1.55485928e-01 -1.05289710e+00 -3.80277574e-01
2.41895780e-01 -3.41694534e-01 -5.00841206e-03 -9.95811403e-01
-3.26939106e-01 7.65446365e-01 7.82297790e-01 6.05032623e-01
4.46085900e-01 5.05451560e-01 5.82435846e-01 7.24679768e-01
-1.05229542e-01 -1.46943045e+00 3.48060310e-01 3.11110526e-01
6.30192578e-01 -1.96280730e+00 -1.47990212e-01 -3.23738426e-01
-8.27307940e-01 8.28564167e-01 8.66524756e-01 2.75491923e-01
7.89168298e-01 2.02328891e-01 4.47613560e-02 -3.25657547e-01
-8.56848478e-01 1.10430643e-01 3.65953773e-01 4.70884144e-01
6.48504019e-01 -3.96539532e-02 -3.64367962e-01 8.75233054e-01
-1.10809743e-01 -1.44535035e-01 3.18398565e-01 9.12293673e-01
-7.12688208e-01 -1.01030922e+00 -2.39982516e-01 7.31620550e-01
-4.73390996e-01 -1.24738716e-01 -8.78424227e-01 5.96639156e-01
-3.10115933e-01 1.36747313e+00 1.64521616e-02 -4.98483539e-01
7.62155503e-02 6.97842002e-01 7.72163447e-04 -4.68050957e-01
-2.45079085e-01 1.90042943e-01 5.40510654e-01 -1.59385964e-01
-4.21422631e-01 -4.29523438e-01 -1.43497121e+00 -2.55997062e-01
-8.19542706e-01 4.92855638e-01 5.84872067e-01 6.95555329e-01
4.01690573e-01 2.02033207e-01 6.60829604e-01 -3.92104059e-01
-9.16269481e-01 -1.28025162e+00 -6.80432200e-01 1.70071200e-01
2.52578676e-01 -6.21085167e-01 -9.49494958e-01 -3.92037123e-01]
|
[12.51302433013916, 7.703769683837891]
|
1fec431b-4521-4675-9c0a-4a166d80cb29
|
convolutional-neural-networks-for-facial
|
1704.06756
| null |
http://arxiv.org/abs/1704.06756v1
|
http://arxiv.org/pdf/1704.06756v1.pdf
|
Convolutional Neural Networks for Facial Expression Recognition
|
We have developed convolutional neural networks (CNN) for a facial expression
recognition task. The goal is to classify each facial image into one of the
seven facial emotion categories considered in this study. We trained CNN models
with different depth using gray-scale images. We developed our models in Torch
and exploited Graphics Processing Unit (GPU) computation in order to expedite
the training process. In addition to the networks performing based on raw pixel
data, we employed a hybrid feature strategy by which we trained a novel CNN
model with the combination of raw pixel data and Histogram of Oriented
Gradients (HOG) features. To reduce the overfitting of the models, we utilized
different techniques including dropout and batch normalization in addition to
L2 regularization. We applied cross validation to determine the optimal
hyper-parameters and evaluated the performance of the developed models by
looking at their training histories. We also present the visualization of
different layers of a network to show what features of a face can be learned by
CNN models.
|
['Shima Alizadeh', 'Azar Fazel']
|
2017-04-22
| null | null | null | null |
['l2-regularization']
|
['methodology']
|
[ 1.03351168e-01 1.16319619e-01 5.07783234e-01 -9.25910771e-01
2.03739569e-01 -1.17567860e-01 4.25856829e-01 -5.73020615e-02
-7.82950282e-01 6.67309582e-01 -2.57319599e-01 -4.13269550e-02
1.81878656e-01 -8.62439513e-01 -5.93468666e-01 -7.62445152e-01
-2.87134051e-01 1.47725359e-01 6.31620511e-02 -1.68150946e-01
2.55102128e-01 1.24438357e+00 -2.00556564e+00 5.63769519e-01
3.10380638e-01 1.24136007e+00 -2.82685071e-01 7.32181787e-01
-1.61681816e-01 7.34597564e-01 -5.32849073e-01 -2.77672946e-01
1.87994108e-01 -4.00682092e-01 -6.01298273e-01 2.47794658e-01
4.31467980e-01 -2.79666513e-01 8.11342224e-02 9.60057318e-01
4.29631710e-01 1.87378246e-02 5.33756971e-01 -1.34111631e+00
-2.48313826e-02 -1.62594438e-01 -3.48273218e-01 5.81969060e-02
5.03749400e-02 1.63436502e-01 2.96042383e-01 -9.72011566e-01
8.17662597e-01 1.14859295e+00 6.17152750e-01 6.79363847e-01
-1.26092446e+00 -6.99857891e-01 -3.60878348e-01 2.03832120e-01
-1.31780970e+00 -4.53593433e-01 7.43388116e-01 -4.01625216e-01
1.07249796e+00 4.59611863e-02 9.60880816e-01 8.81526172e-01
1.52349204e-01 2.95649946e-01 1.43025303e+00 -6.30500376e-01
3.18593919e-01 5.10286391e-01 -4.30863257e-03 1.03343320e+00
-9.39228460e-02 -4.50569764e-03 -4.63394910e-01 -1.06044717e-01
7.71313250e-01 -2.09446892e-01 1.64833501e-01 -1.12770401e-01
-3.95990640e-01 9.75279987e-01 3.72169703e-01 5.09500027e-01
-4.25363243e-01 2.09409893e-01 5.87064683e-01 2.04288095e-01
3.60773891e-01 1.52974829e-01 -4.00602460e-01 -8.38704407e-03
-8.40447426e-01 -5.17672412e-02 7.55322993e-01 3.45976233e-01
1.13924360e+00 2.27757394e-01 1.24100119e-01 7.49732494e-01
1.25606477e-01 2.34647971e-02 5.48857689e-01 -9.24343109e-01
-2.44792238e-01 8.89419258e-01 -2.57644862e-01 -1.08444762e+00
-5.63122988e-01 -6.05943948e-02 -8.77864897e-01 1.04325306e+00
5.36139369e-01 -4.06449765e-01 -1.09250700e+00 1.45615828e+00
2.08226353e-01 -3.43414582e-02 9.09363478e-02 9.22306836e-01
8.09547424e-01 6.16102099e-01 2.82766581e-01 -2.87561566e-02
1.18288839e+00 -5.86927891e-01 -5.94140887e-01 2.02942669e-01
9.36357439e-01 -6.24207139e-01 8.37692320e-01 5.33866882e-01
-9.90840197e-01 -7.72207677e-01 -1.04038155e+00 8.17391649e-02
-8.68907511e-01 3.79579961e-01 5.51326513e-01 8.68938148e-01
-1.41970170e+00 9.92138326e-01 -7.43487895e-01 -4.95778292e-01
6.62656665e-01 6.74696684e-01 -8.05826008e-01 3.32823575e-01
-9.88229573e-01 7.33139277e-01 3.43711764e-01 4.20661539e-01
-6.23961627e-01 -2.47787267e-01 -7.39958107e-01 2.69718230e-01
-2.66744018e-01 -2.17838049e-01 7.46238530e-01 -1.85911345e+00
-1.76254952e+00 1.26053095e+00 -9.67019890e-03 -3.43821675e-01
3.09968978e-01 1.39253736e-01 -2.32085362e-01 3.37735772e-01
-7.24309504e-01 9.99985993e-01 8.71155083e-01 -1.04563808e+00
-3.79804194e-01 -4.22742039e-01 -1.48888960e-01 -1.59038246e-01
-5.69367588e-01 2.88912743e-01 -1.70575470e-01 -1.24317169e-01
-4.20862027e-02 -8.65515351e-01 -2.21503805e-03 2.55368441e-01
5.32958806e-02 -1.25184413e-02 1.04108059e+00 -4.99630690e-01
8.11165154e-01 -2.22105336e+00 -6.64342493e-02 5.40964425e-01
4.91968282e-02 3.60593706e-01 -1.00870557e-01 6.28503188e-02
-2.98114568e-01 5.79713061e-02 -6.28471896e-02 -4.55329180e-01
-2.60440528e-01 3.93122911e-01 2.03377679e-01 5.32081902e-01
4.31013644e-01 4.59964991e-01 -3.00020307e-01 -6.22282147e-01
2.19547227e-01 1.00537682e+00 -6.36625707e-01 3.30011219e-01
-6.48652315e-02 3.24479669e-01 -1.25521362e-01 4.27118003e-01
8.47607672e-01 1.90573260e-01 1.19216740e-01 -2.44526505e-01
-1.62192896e-01 -3.51736814e-01 -9.41823304e-01 1.33925796e+00
-5.41495502e-01 8.81134033e-01 1.58497065e-01 -8.72054577e-01
1.33180237e+00 3.27071905e-01 4.06146497e-01 -7.04460442e-01
5.87630987e-01 1.79779634e-01 7.36738294e-02 -7.39330113e-01
1.40465662e-01 -2.27153376e-01 5.74981391e-01 2.13135839e-01
4.52625185e-01 7.75771737e-02 6.25741705e-02 -3.54296535e-01
6.79144561e-01 2.01209173e-01 -3.21232192e-02 -3.47069412e-01
6.08895063e-01 -1.77612156e-01 2.46785298e-01 3.01168472e-01
-1.52239501e-01 4.49070066e-01 9.46550250e-01 -8.90176117e-01
-1.03026164e+00 -3.75720888e-01 -1.85067773e-01 1.08583069e+00
-6.30995214e-01 -1.28686056e-01 -9.55326617e-01 -4.49011117e-01
-3.47127706e-01 2.64482737e-01 -1.05295730e+00 9.78258159e-03
-4.73219812e-01 -9.18790221e-01 6.02618456e-01 2.19691530e-01
5.58919847e-01 -1.47999775e+00 -1.28484285e+00 -7.13103861e-02
5.79035461e-01 -8.55354905e-01 3.42607230e-01 6.15680516e-01
-1.01358247e+00 -1.00834084e+00 -5.91801405e-01 -8.76652718e-01
8.88139844e-01 -5.38837731e-01 8.63373637e-01 2.28532970e-01
-6.81113005e-01 1.71820194e-01 -3.71498585e-01 -4.56457615e-01
-3.33915800e-01 -9.18838754e-02 -2.61577457e-01 3.47472042e-01
5.27982950e-01 -5.51107705e-01 -5.23904622e-01 -7.27005675e-02
-1.13561082e+00 -4.07096036e-02 5.12229860e-01 7.57183015e-01
2.39258781e-01 -1.35839075e-01 4.73884568e-02 -7.71257699e-01
5.24817467e-01 -1.53254300e-01 -7.34909773e-01 -2.10970137e-02
-2.53920078e-01 2.33945139e-02 6.61305189e-01 -4.14385855e-01
-8.29630494e-01 4.24035668e-01 -4.44089085e-01 -5.65291762e-01
-4.94765639e-01 2.14802533e-01 5.80133758e-02 -5.89641571e-01
5.51461279e-01 -4.80505452e-02 3.92558664e-01 -2.89535403e-01
-1.17022395e-01 5.66222608e-01 1.51099078e-02 -2.66133577e-01
1.48504615e-01 4.44535464e-01 3.22849661e-01 -1.06717062e+00
-2.88865000e-01 -4.96477075e-02 -8.12001526e-01 -6.40280366e-01
1.05832231e+00 -4.14114863e-01 -1.18284595e+00 7.83146858e-01
-1.29438663e+00 -4.03367102e-01 9.19574350e-02 2.62599975e-01
-2.84425378e-01 -7.48355910e-02 -6.24379277e-01 -9.16430712e-01
-2.24463418e-01 -1.02911639e+00 7.35893488e-01 6.46448612e-01
4.00770679e-02 -1.01302338e+00 4.45018001e-02 -3.18018347e-01
6.79630458e-01 5.42703331e-01 8.54097962e-01 -5.18919110e-01
3.62996245e-03 -2.83616424e-01 -3.35146546e-01 7.44748235e-01
-1.66127101e-01 5.02559960e-01 -1.41403639e+00 -6.00302182e-02
1.19517319e-01 -5.54322481e-01 6.40503526e-01 2.56845087e-01
1.48623252e+00 -2.51093417e-01 4.20556702e-02 8.15084577e-01
1.56808555e+00 2.87559390e-01 1.03424025e+00 5.74349821e-01
3.18709344e-01 9.70575929e-01 1.85627565e-01 5.74906349e-01
-3.71233493e-01 6.65514708e-01 5.69807529e-01 -5.60899436e-01
1.56905621e-01 1.58661962e-01 9.84132141e-02 -3.11905313e-02
-4.09985244e-01 1.71220154e-01 -8.41199934e-01 1.85883164e-01
-1.42519414e+00 -7.80201793e-01 -1.11196069e-02 1.96563673e+00
5.02955616e-01 1.71744406e-01 2.01381296e-01 1.01728790e-01
4.01807129e-01 -1.39355198e-01 -4.68921661e-02 -1.07103765e+00
-2.42159694e-01 7.52462268e-01 2.98551023e-01 4.29993451e-01
-1.03788853e+00 9.71734524e-01 6.20078564e+00 2.96291560e-01
-1.91758823e+00 -1.39406130e-01 1.00903058e+00 -6.64947182e-02
4.40278023e-01 -3.01869482e-01 -6.23142421e-01 2.25296602e-01
1.09665120e+00 4.25855041e-01 2.12896779e-01 1.11142433e+00
2.33004197e-01 -3.54929835e-01 -7.28223205e-01 9.55589116e-01
1.02515973e-01 -1.11908174e+00 -6.13255985e-02 8.79332274e-02
4.93933380e-01 -1.83149219e-01 -1.37978550e-02 7.82784671e-02
-2.57040381e-01 -1.40014613e+00 3.50226969e-01 6.04513764e-01
6.03308201e-01 -7.49403119e-01 9.72200215e-01 -1.50366992e-01
-6.19645000e-01 -1.57492142e-02 -5.45094073e-01 -1.72579318e-01
-3.01638842e-01 2.70285428e-01 -9.16052997e-01 -2.75535528e-02
8.70210826e-01 2.91808695e-01 -8.15409005e-01 7.62561858e-01
1.66017637e-01 3.96816015e-01 -5.05204558e-01 -2.14633822e-01
5.07948995e-01 -3.02080393e-01 -5.71599267e-02 1.47665918e+00
3.58681977e-01 -8.50707740e-02 -4.01307702e-01 7.59737968e-01
1.13333955e-01 4.01225179e-01 -5.80055237e-01 1.18328869e-01
-1.06792428e-01 1.65314722e+00 -8.52357268e-01 -2.79209018e-01
-2.15164647e-01 9.75388169e-01 3.42479408e-01 3.16871881e-01
-6.94038868e-01 -5.47421694e-01 5.65391898e-01 3.28055352e-01
3.58902872e-01 -1.95559606e-01 -1.94481671e-01 -7.35495925e-01
-2.27925479e-01 -5.51024973e-01 1.60607889e-01 -9.20398653e-01
-6.14388525e-01 1.05482471e+00 -6.40362501e-02 -7.78906226e-01
-2.58944809e-01 -1.10210419e+00 -6.72132969e-01 1.01322496e+00
-1.42284000e+00 -9.27721143e-01 -6.32039189e-01 6.51720941e-01
7.07123801e-02 -2.84231484e-01 1.13704431e+00 2.42551342e-01
-5.80840111e-01 3.57610673e-01 -2.62423038e-01 2.85073489e-01
4.29982632e-01 -9.09062743e-01 -3.03486884e-01 3.78963232e-01
-1.22006901e-01 3.73636663e-01 7.93784380e-01 -1.37729064e-01
-8.42887282e-01 -7.14508057e-01 9.49166834e-01 2.88361281e-01
4.30154532e-01 -4.49463189e-01 -1.06552958e+00 5.07471502e-01
3.13082457e-01 2.87048846e-01 7.01731920e-01 -1.34881869e-01
-2.16985971e-01 -2.69104332e-01 -1.25923252e+00 2.28374213e-01
4.10068989e-01 -2.41301134e-01 1.08002173e-02 1.94215961e-02
-1.19905509e-01 -1.92285895e-01 -6.89600945e-01 2.64995575e-01
6.70339465e-01 -1.61454093e+00 4.66902077e-01 -7.89506555e-01
5.71344018e-01 -5.76628000e-02 1.67117760e-01 -1.16132414e+00
-2.64622290e-02 -1.75308853e-01 3.37243617e-01 9.61779058e-01
3.44613731e-01 -5.62052131e-01 1.07603204e+00 5.28876603e-01
2.39205211e-01 -9.11676466e-01 -8.77093732e-01 -2.52718091e-01
-9.00141671e-02 -8.19687769e-02 2.48580575e-01 6.02516115e-01
-1.73046961e-01 -4.32685353e-02 -1.87906116e-01 -4.07790840e-02
3.22458863e-01 -2.39121497e-01 6.24592841e-01 -1.07704246e+00
4.60297801e-02 -5.36222458e-01 -9.88475859e-01 -5.18842973e-02
3.54195356e-01 -6.11975551e-01 -2.10636064e-01 -8.76282334e-01
-9.16505232e-02 -2.59819329e-01 -1.89025223e-01 7.92810738e-01
4.44770515e-01 7.55186260e-01 1.42805353e-01 -2.54746020e-01
2.62445081e-02 3.80105823e-01 9.01578605e-01 2.24681213e-01
-2.64397949e-01 -2.29995340e-01 -1.07202254e-01 7.10380554e-01
1.00439620e+00 -4.95375812e-01 -8.24429244e-02 -2.23192602e-01
1.57046378e-01 -1.84548840e-01 6.11514926e-01 -1.22346640e+00
7.71300271e-02 1.66803285e-01 9.19947743e-01 -3.24109532e-02
5.51849067e-01 -9.82840300e-01 1.66947424e-01 6.60402417e-01
-3.21215987e-01 -1.12020336e-02 5.88774920e-01 -1.95547312e-01
-4.47028965e-01 -3.86246651e-01 1.23395622e+00 -1.06324799e-01
-7.72781968e-01 -3.91849019e-02 -4.67384845e-01 -7.46153474e-01
1.08247328e+00 -3.87834013e-01 7.34875798e-02 -2.83852875e-01
-1.17262316e+00 -3.16714168e-01 4.84595329e-01 7.36286193e-02
6.91405714e-01 -1.04301155e+00 -5.14122605e-01 6.22218549e-01
-2.59210970e-02 -4.43117261e-01 2.66912580e-01 8.11864793e-01
-1.20814240e+00 1.26309842e-01 -1.12608433e+00 -5.50851882e-01
-1.79944003e+00 2.89581716e-01 8.65032136e-01 5.42480610e-02
-2.44002402e-01 7.10309267e-01 -1.37541324e-01 -1.63122579e-01
3.37677777e-01 -9.49671641e-02 -6.18021131e-01 -4.39048372e-03
5.78892231e-01 1.05916180e-01 2.53481179e-01 -6.07541203e-01
-3.71168405e-01 6.66959882e-01 6.29700869e-02 -8.78330171e-02
1.54943740e+00 2.80965686e-01 -3.65789711e-01 2.16128647e-01
1.62527418e+00 -3.65553498e-01 -1.29455030e+00 3.74681562e-01
-2.13112142e-02 -2.92589217e-01 2.24642262e-01 -6.83228970e-01
-1.31699586e+00 1.16785133e+00 1.06061411e+00 1.12605132e-01
1.56746590e+00 -4.02083486e-01 1.37004405e-01 2.82032043e-01
1.98261626e-02 -1.12979412e+00 8.74557998e-03 5.93472540e-01
6.99750781e-01 -1.13748968e+00 -2.04845995e-01 -2.09518746e-01
-5.60338795e-01 1.90308082e+00 6.15902662e-01 -4.77972627e-01
8.32099617e-01 4.35779184e-01 3.80497575e-01 -3.95054966e-01
-7.12969482e-01 -2.86411136e-01 7.92861953e-02 4.54535455e-01
7.21186876e-01 -3.19174439e-01 -3.71079355e-01 1.56294256e-01
-1.32767096e-01 4.89701360e-01 4.33547974e-01 9.42242265e-01
-4.15392578e-01 -9.30286646e-01 -2.97809631e-01 2.10678801e-01
-5.23912728e-01 2.81022519e-01 -4.86981064e-01 1.02749729e+00
4.16907400e-01 5.25615394e-01 3.51848692e-01 -4.06304955e-01
2.50821769e-01 4.11457539e-01 6.79908514e-01 -2.78149307e-01
-9.08068299e-01 -8.20112154e-02 -7.19908327e-02 -6.26338184e-01
-4.25966382e-01 -3.84240985e-01 -1.10255277e+00 -3.41255963e-01
2.88686156e-01 -1.63191706e-02 1.20426011e+00 7.52494633e-01
2.36739025e-01 4.05188024e-01 6.05267644e-01 -1.18950832e+00
-1.32637369e-02 -1.18576252e+00 -6.68930531e-01 5.24311244e-01
1.64479107e-01 -4.59795654e-01 -5.02813041e-01 2.99685508e-01]
|
[13.522807121276855, 1.8095557689666748]
|
2a0a880b-a8cb-4175-b415-4b2a4ab436aa
|
a-photo-click-thiol-ene-collagen-based
|
2304.01942
| null |
https://arxiv.org/abs/2304.01942v2
|
https://arxiv.org/pdf/2304.01942v2.pdf
|
A photo-click thiol-ene collagen-based hydrogel platform for skeletal muscle tissue engineering
|
UV-cured collagen-based hydrogels hold promise in skeletal muscle regeneration due to their soft elastic properties and porous architecture. However, the complex triple helix conformation of collagen and environmental conditions, i.e. molecular oxygen, pose risks to reaction controllability, wet-state integrity and reproducibility. To address this challenge, a photo-click hydrogel platform is presented through an oxygen-insensitive thiol-ene reaction between 2-iminothiolane (2IT)-functionalized type I collagen and multi-arm, non-homopolymerizable norbornene-terminated polyethylene glycol (PEG). UV-induced network formation is demonstrated by oscillatory time sweeps on the reacting thiol-ene mixture, so that significantly increased storage moduli are measured and adjusted depending on the photo-initiator concentration. Variations in PEG functionality (4-arm and 8-arm) and PEG content generate hydrogels with skeletal muscle native stiffness (Ec= 1.3-11.5 kPa), diffusion-controlled swelling behavior and erosion-driven degradability. In vitro, no cytotoxic effect is detected on C2C12 murine myoblasts, while myogenic differentiation is successfully accomplished on hydrogel-seeded cells in low serum culture medium. In vivo, 7-day subcutaneous implantation of selected thiol-ene hydrogel in rats reveal higher cell infiltration, blood vessel formation, and denser tissue interface compared to a clinical gold standard collagen matrix (Mucograft, Geistlich). These results therefore support the applicability and further development of this hydrogel platform for skeletal muscle regeneration.
|
['Giuseppe Tronci', 'David J. Wood', 'Xuebin B. Yang', 'Roisin Holmes']
|
2023-03-29
| null | null | null | null |
['culture']
|
['speech']
|
[ 3.48569870e-01 -3.96323204e-02 -4.14923638e-01 4.51376885e-01
-4.35810536e-01 -8.15350235e-01 8.97131711e-02 5.34664631e-01
-6.56399488e-01 9.69033957e-01 5.42810798e-01 1.98155984e-01
1.34874567e-01 -6.08621478e-01 -4.98626560e-01 -1.21231794e+00
-4.21075016e-01 8.74917284e-02 4.14374352e-01 -2.92079300e-01
1.35397300e-01 6.30708873e-01 -8.63621294e-01 1.51245803e-01
8.87747705e-01 7.85564601e-01 3.63351852e-01 3.46103907e-01
2.17937499e-01 3.08257073e-01 -1.25085667e-01 1.28026977e-01
-9.08765644e-02 -3.31824645e-02 -1.46727845e-01 1.59140006e-01
-2.30790034e-01 -4.89212781e-01 8.34647864e-02 4.94384617e-01
4.03741896e-01 -2.98778921e-01 7.11240530e-01 -6.93857074e-01
-2.74529725e-01 3.66971493e-01 -2.07243845e-01 -3.74829501e-01
5.85125566e-01 2.17848912e-01 4.61717576e-01 -9.58960474e-01
1.30251944e+00 8.07095110e-01 4.30893749e-01 6.90791547e-01
-1.54439521e+00 -9.54149142e-02 -4.19388801e-01 -3.84507567e-01
-5.26673555e-01 -6.87997267e-02 2.23147631e-01 -9.15908396e-01
5.00841379e-01 6.77267253e-01 1.18635976e+00 1.13254476e+00
1.29037511e+00 2.03939244e-01 1.52561653e+00 -4.02597263e-02
5.53925157e-01 1.00184523e-01 -5.27738869e-01 1.64723918e-01
6.49388552e-01 1.80085033e-01 -1.60724133e-01 -1.75380006e-01
7.71512926e-01 8.62848014e-02 -6.25973105e-01 -3.20368290e-01
-1.07067215e+00 6.33102179e-01 9.52188298e-02 4.85137552e-01
-5.71142673e-01 4.99131940e-02 6.92615211e-01 8.96794945e-02
4.35220778e-01 7.85957798e-02 1.03902845e-02 -3.06335151e-01
-5.42691350e-01 1.12681746e-01 9.31750894e-01 4.26572978e-01
-8.22684020e-02 -1.91581007e-02 1.06875531e-01 5.74041009e-01
4.56639975e-01 5.14316976e-01 3.50677729e-01 -8.99462640e-01
1.87212288e-01 3.41122806e-01 2.03067303e-01 -6.81421757e-01
-3.39601249e-01 -2.16322571e-01 -7.21026242e-01 6.03198946e-01
5.02686739e-01 -2.05274731e-01 -4.81469482e-01 1.55072129e+00
6.80084527e-01 -8.96697879e-01 2.20661417e-01 1.39396000e+00
4.64930803e-01 5.00452340e-01 5.65341294e-01 -6.35758162e-01
1.37379038e+00 -4.68570054e-01 -9.03051615e-01 7.37720057e-02
3.96762848e-01 -6.17132068e-01 8.14972579e-01 2.43112788e-01
-1.30909622e+00 9.78976786e-02 -1.13625789e+00 3.03893209e-01
-1.19600393e-01 -1.46462843e-01 4.15492624e-01 4.96817797e-01
-6.17793262e-01 9.97845054e-01 -9.87371683e-01 -5.34949243e-01
3.69126871e-02 3.44344854e-01 -6.87181592e-01 -2.13952601e-01
-9.45499003e-01 4.45901901e-01 1.42315224e-01 1.96414545e-01
-6.70435369e-01 -7.23801970e-01 -7.56816924e-01 -1.35288566e-01
-5.35435975e-01 -9.47357595e-01 2.32156679e-01 -3.19544494e-01
-2.20617127e+00 5.78359485e-01 4.84340578e-01 -3.76879930e-01
9.03402030e-01 -1.37123466e-01 -1.55145377e-01 7.64909446e-01
1.60353512e-01 5.40265799e-01 6.83254674e-02 -1.31040907e+00
6.15817346e-02 -9.83722582e-02 -2.50887990e-01 -1.00323997e-01
-3.56823206e-01 2.26451650e-01 4.67737168e-01 -7.77997613e-01
4.19080317e-01 -1.32509172e+00 -3.29898149e-01 5.02296150e-01
-3.85601133e-01 2.65503138e-01 4.98811990e-01 -6.04829550e-01
8.22915018e-01 -2.01959229e+00 2.85868496e-01 4.13756162e-01
6.22918457e-02 -9.01411250e-02 -5.83278984e-02 1.53476095e+00
-6.86384514e-02 4.46444333e-01 3.93875092e-02 4.38127398e-01
-6.26871064e-02 3.34706120e-02 2.75766701e-01 8.55724394e-01
-1.29193619e-01 4.52236533e-01 -1.02044296e+00 -5.40542603e-01
-3.91145386e-02 7.19510317e-01 -3.59925717e-01 -2.74586946e-01
-2.95292735e-01 7.16294885e-01 -2.51958430e-01 1.08923018e+00
6.71004832e-01 2.24090531e-01 6.81250513e-01 -6.18176162e-01
-5.73838890e-01 -2.25983456e-01 -5.52276433e-01 1.80355561e+00
2.99252775e-02 3.12162608e-01 3.43848735e-01 1.94965631e-01
9.07393396e-01 6.17041528e-01 9.47163343e-01 -8.24182987e-01
2.05052614e-01 6.57874048e-01 -1.57292798e-01 -9.71257210e-01
5.22411503e-02 -8.45495462e-01 3.91933739e-01 -1.73435993e-02
-5.61805129e-01 4.35347706e-01 6.50495648e-01 1.25447109e-01
9.61456478e-01 1.60504669e-01 -4.92471904e-01 -7.47377753e-01
2.39861056e-01 4.77293432e-02 4.02739584e-01 -2.87070051e-02
6.30039498e-02 5.75375259e-01 6.25801086e-01 -8.27767402e-02
-1.32795322e+00 -9.97131646e-01 -3.05907488e-01 6.19767904e-01
1.95433453e-01 -3.92590553e-01 -5.56575418e-01 3.61785650e-01
4.19228852e-01 -2.22616848e-02 -5.62294781e-01 2.08514228e-01
-4.84923929e-01 -3.74578126e-02 2.53873646e-01 2.72889882e-01
3.11046481e-01 -4.73976344e-01 -5.02905309e-01 9.00428653e-01
1.17208630e-01 -9.05455351e-01 -3.14205259e-01 -4.34677362e-01
-1.30531347e+00 -8.02019179e-01 -1.02917480e+00 -5.37969887e-01
5.29581010e-01 -2.86066920e-01 2.81603783e-01 -4.55125347e-02
-3.32009196e-01 3.36431682e-01 -3.31941873e-01 -2.30165217e-02
-6.98956192e-01 -2.48853505e-01 1.51524007e-01 -1.84130996e-01
-4.68942463e-01 -1.00001144e+00 -1.38965869e+00 4.43253815e-01
-1.00525022e+00 -1.07089601e-01 7.76536465e-01 5.85899234e-01
7.25001276e-01 -6.70468092e-01 6.83473587e-01 -8.03360701e-01
6.76705480e-01 -5.12074828e-01 -1.00457231e-02 -1.24614991e-01
-5.08573592e-01 -2.48456180e-01 3.64184454e-02 -9.18060780e-01
-8.45972419e-01 -2.74999976e-01 3.22981691e-03 1.31716624e-01
3.27970654e-01 1.36477208e+00 1.79855563e-02 -1.37094110e-01
7.73287654e-01 -9.00373459e-02 9.26524460e-01 -1.29875883e-01
-1.56415448e-01 5.46091080e-01 2.95414478e-01 -7.57440746e-01
6.04564309e-01 6.12870276e-01 1.62652701e-01 -8.00699115e-01
-5.63982548e-03 -1.54956341e-01 1.59469575e-01 -8.02279711e-01
9.48196232e-01 -1.11497581e+00 -8.04505110e-01 5.22689641e-01
-7.02870667e-01 -6.15303338e-01 -2.96273977e-02 8.64114761e-01
-3.39001834e-01 4.62597996e-01 -1.50532722e+00 -5.89885175e-01
-5.53720057e-01 -8.46826732e-01 3.40511382e-01 -8.02663267e-02
-3.62994641e-01 -1.02814603e+00 4.00252581e-01 5.50802052e-01
7.74416745e-01 1.60259628e+00 9.50805664e-01 5.25215149e-01
-4.82987940e-01 -5.78225911e-01 3.40798825e-01 2.91625619e-01
1.39979735e-01 2.82512963e-01 -3.10447991e-01 -8.21032465e-01
-2.50819504e-01 -2.33751401e-01 5.32224774e-01 4.20018584e-01
6.90349117e-02 -6.99044526e-01 -6.46597743e-01 -8.81506652e-02
1.94731319e+00 1.21961594e-01 1.21002984e+00 5.78470469e-01
1.44141778e-01 4.21490550e-01 7.95314491e-01 6.05692983e-01
-2.44267002e-01 1.56157389e-01 4.60282922e-01 -5.13986647e-01
-1.97840869e-01 -2.53770649e-01 7.54597068e-01 7.95393169e-01
-6.23118222e-01 -3.54470968e-01 -3.06090534e-01 5.04037380e-01
-1.00863981e+00 -6.87029779e-01 -4.12914127e-01 2.40494657e+00
1.42516160e+00 3.58034313e-01 4.50372338e-01 -1.47861585e-01
5.84399343e-01 -5.01127481e-01 -2.92189479e-01 -2.04701990e-01
-2.16421247e-01 1.36984855e-01 6.72890127e-01 4.80896503e-01
-4.05583531e-01 1.91580296e-01 5.35841846e+00 1.86766610e-01
-1.53531015e+00 -8.98140967e-02 -3.80585074e-01 -3.02851588e-01
-7.85773635e-01 4.39689547e-01 -6.33785427e-02 5.96982658e-01
5.03055990e-01 -5.65736890e-02 -3.33501756e-01 2.58138299e-01
6.22191727e-01 -5.88370383e-01 -7.87572980e-01 2.00474724e-01
-6.09152973e-01 -1.63915026e+00 -3.77370507e-01 6.09604955e-01
4.61618304e-01 -2.73551017e-01 -1.75594002e-01 -5.29514849e-01
-6.90206170e-01 -3.38453889e-01 9.56152499e-01 7.22137272e-01
1.20145941e+00 -3.82055730e-01 6.34139240e-01 -2.04878166e-01
-7.29896843e-01 -1.16949141e-01 5.33613414e-02 3.22706206e-03
5.44920385e-01 9.69818294e-01 -6.99094296e-01 1.83718339e-01
4.52733278e-01 4.40748423e-01 1.16546310e-01 1.10006964e+00
1.17399067e-01 4.00363028e-01 -3.39281291e-01 -1.91043422e-01
-8.37800652e-02 -5.17108560e-01 8.29669893e-01 8.94302964e-01
-1.40557275e-03 1.76250815e-01 -2.84997020e-02 6.56533897e-01
3.25091779e-01 2.42392406e-01 -3.13518465e-01 -3.43562722e-01
4.02374864e-01 1.08266497e+00 -9.00466502e-01 5.57777956e-02
-1.78298354e-02 7.10020483e-01 -3.09211195e-01 6.93657637e-01
-4.98260260e-01 -3.94187242e-01 6.92295253e-01 8.55632842e-01
6.63025379e-02 -7.50116527e-01 1.84664160e-01 -5.84926963e-01
1.84985712e-01 -5.18645644e-01 8.09797347e-02 -2.96777219e-01
-1.00658393e+00 -6.08073287e-02 -2.77687877e-01 -1.29314113e+00
6.51140690e-01 -3.38496298e-01 -4.02680397e-01 4.55472767e-01
-1.21305227e+00 -1.08534670e+00 -1.34376675e-01 -3.51757258e-01
6.78923503e-02 5.92479050e-01 9.25122619e-01 2.77109444e-01
-5.12166917e-01 1.23825707e-01 3.65454018e-01 -3.55332315e-01
8.33160877e-01 -7.61045039e-01 -9.02479172e-01 2.44474337e-01
-1.30503035e+00 6.85777962e-01 1.24017477e+00 -1.26950288e+00
-1.56425774e+00 -6.58958852e-01 5.79168022e-01 4.25409049e-01
5.54830253e-01 -3.95253450e-01 -5.49784303e-01 2.96301633e-01
3.79094362e-01 -5.54521680e-01 1.29964054e+00 -6.07291758e-01
-5.06047942e-02 7.50474036e-02 -1.48151374e+00 7.39507437e-01
5.99830806e-01 -3.12850513e-02 3.61625522e-01 9.63123739e-01
3.92109305e-01 -8.36749613e-01 -2.16775346e+00 4.03851956e-01
1.13605011e+00 -5.77247322e-01 5.23202777e-01 -6.06004298e-01
7.68354833e-01 -2.97610223e-01 8.36500451e-02 -5.46669006e-01
-3.48290563e-01 -7.77712703e-01 4.65848655e-01 9.85051334e-01
4.48479682e-01 -8.77903521e-01 7.00295329e-01 1.01897323e+00
-3.82708549e-01 -8.48439217e-01 -9.56353128e-01 -9.33048904e-01
2.84897506e-01 8.30569804e-01 -3.87788057e-01 6.98910713e-01
7.24788070e-01 -2.20552668e-01 -1.16623286e-02 -7.95350745e-02
3.74165595e-01 -1.40810281e-01 3.89118850e-01 -1.21868610e+00
1.35592446e-01 2.26900131e-01 -5.10130525e-01 -3.08062702e-01
-2.49599338e-01 -8.06690276e-01 -3.66327077e-01 -1.70202720e+00
-8.70269723e-03 -7.01748610e-01 2.68394917e-01 2.13117778e-01
5.54582238e-01 2.49912823e-03 5.12275063e-02 4.72871989e-01
1.28963232e-01 5.25754511e-01 1.40903056e+00 -2.74014831e-01
-3.05931836e-01 -3.74765188e-01 4.11186293e-02 -3.16987664e-01
9.30700541e-01 -5.29403090e-01 -3.02071869e-01 -2.48855576e-02
5.18237650e-01 6.37776852e-01 2.54262149e-01 -1.22631860e+00
8.75827000e-02 -2.31045514e-01 -4.93979454e-02 8.33549500e-02
1.59230992e-01 -7.78821588e-01 1.19923139e+00 1.11031663e+00
-1.91151708e-01 -3.48173916e-01 -2.12643459e-01 6.73221946e-01
8.36928040e-02 -6.29418641e-02 5.25831997e-01 -1.62827089e-01
2.78717846e-01 -1.68817848e-01 -1.38627362e+00 -5.24824560e-01
1.20183647e+00 -1.02644455e+00 -7.06383348e-01 6.95245415e-02
-1.21481943e+00 8.59502237e-03 1.03998911e+00 -2.70204514e-01
5.87507010e-01 -1.39305770e+00 -8.40939403e-01 -2.01463595e-01
-8.35892409e-02 -2.28892833e-01 1.02142215e+00 1.56522012e+00
-1.14858747e+00 -1.78451046e-01 -7.10183561e-01 -9.27212894e-01
-1.00277472e+00 2.10952863e-01 2.51962394e-01 3.67382728e-02
-5.52686334e-01 3.67117912e-01 -4.33319092e-01 4.45323199e-01
-2.20230103e-01 -2.95712352e-01 2.15356603e-01 1.19919600e-02
4.67351861e-02 6.13942087e-01 8.37841630e-03 3.34466577e-01
-1.81451306e-01 4.16987032e-01 -8.39097723e-02 -1.18145414e-01
1.44966722e+00 -9.77092013e-02 -4.31877434e-01 3.98878068e-01
9.55156088e-01 2.63012409e-01 -1.48671520e+00 3.76636326e-01
-2.93468922e-01 -2.21475378e-01 -2.79334396e-01 -6.19256556e-01
-1.18597651e+00 -5.69051579e-02 7.10875869e-01 2.30782300e-01
7.19582319e-01 1.49279768e-02 9.38839614e-01 -3.83391142e-01
1.99120119e-01 -1.44918406e+00 3.76818210e-01 -3.74733955e-01
1.47893083e+00 -4.63097215e-01 2.89627105e-01 -1.16326582e+00
-1.53031871e-01 1.35925639e+00 1.68909192e-01 -2.35672101e-01
4.66096401e-01 2.34644055e-01 1.86551005e-01 -2.97771156e-01
-7.39351630e-01 3.87805432e-01 -5.71165621e-01 5.55794477e-01
7.36640036e-01 1.17906697e-01 -1.56088996e+00 1.76734537e-01
1.31804302e-01 3.02056760e-01 8.17892492e-01 1.37590730e+00
-4.70373094e-01 -1.22676170e+00 1.23088852e-01 -2.17227964e-03
-2.29212582e-01 3.64676595e-01 5.58800250e-02 1.01358175e+00
-3.23884904e-01 6.21213138e-01 -4.61834013e-01 9.60892532e-03
4.04144973e-01 -2.87537673e-03 5.43945491e-01 -1.45366818e-01
-5.95581234e-01 7.61326313e-01 6.01084590e-01 -4.84409392e-01
-7.89848268e-01 -8.64320457e-01 -1.35053730e+00 -2.04521880e-01
-4.00549561e-01 1.23273626e-01 1.20146561e+00 6.15250349e-01
4.83286083e-01 1.51433617e-01 7.32593596e-01 -6.98499620e-01
-2.43375853e-01 -7.37298667e-01 -1.05924022e+00 3.59826684e-01
2.29085997e-01 -7.48820305e-01 -6.53626502e-01 1.62931547e-01]
|
[13.633379936218262, -3.0355491638183594]
|
3f5816de-0d53-4fbd-9882-7a21d9bbb4ef
|
automatic-design-method-of-building-pipeline
|
2305.10760
| null |
https://arxiv.org/abs/2305.10760v1
|
https://arxiv.org/pdf/2305.10760v1.pdf
|
Automatic Design Method of Building Pipeline Layout Based on Deep Reinforcement Learning
|
The layout design of pipelines is a critical task in the construction industry. Currently, pipeline layout is designed manually by engineers, which is time-consuming and laborious. Automating and streamlining this process can reduce the burden on engineers and save time. In this paper, we propose a method for generating three-dimensional layout of pipelines based on deep reinforcement learning (DRL). Firstly, we abstract the geometric features of space to establish a training environment and define reward functions based on three constraints: pipeline length, elbow, and installation distance. Next, we collect data through interactions between the agent and the environment and train the DRL model. Finally, we use the well-trained DRL model to automatically design a single pipeline. Our results demonstrate that DRL models can complete the pipeline layout task in space in a much shorter time than traditional algorithms while ensuring high-quality layout outcomes.
|
['Jia-Rui Lin', 'Zhe Zheng', 'Chen Yang']
|
2023-05-18
| null | null | null | null |
['layout-design']
|
['computer-vision']
|
[-1.34929761e-01 7.86702707e-02 3.59493077e-01 -2.56296605e-01
-4.69260931e-01 -1.03522384e+00 5.32775768e-04 1.48671255e-01
-2.41987184e-01 4.04756218e-01 -9.35116410e-02 -7.07817554e-01
-1.93808615e-01 -1.08296406e+00 -8.32544088e-01 -3.42739284e-01
-1.39497042e-01 4.09707695e-01 1.88483059e-01 2.69063637e-02
5.26217282e-01 9.97613788e-01 -1.16038620e+00 -1.21969037e-01
8.46808970e-01 5.18031180e-01 6.09235883e-01 8.90915751e-01
1.02567472e-01 7.44478405e-01 -8.85650456e-01 8.91477019e-02
1.69059947e-01 -1.87059969e-01 -9.24597502e-01 3.09163034e-01
-3.24730456e-01 -6.97934806e-01 -4.31015193e-01 4.49285477e-01
4.85511988e-01 -4.66217194e-03 6.30206287e-01 -1.25111747e+00
-4.59365517e-01 6.99755073e-01 -6.93967640e-01 -4.93624091e-01
1.40349895e-01 4.83509690e-01 7.67630458e-01 -5.50476015e-01
3.24880749e-01 1.07291174e+00 4.69554454e-01 3.47691953e-01
-1.05947423e+00 -3.83627594e-01 2.11499766e-01 -1.80091769e-01
-1.09314299e+00 2.36585476e-02 8.07483554e-01 -6.51080966e-01
7.57324398e-01 -9.02806595e-02 9.27089572e-01 5.81538081e-01
1.42010614e-01 8.70987117e-01 4.50397372e-01 -4.82851714e-01
7.01990128e-01 -3.98010343e-01 -5.13137758e-01 8.56411934e-01
2.34942928e-01 -6.84179068e-02 -2.25771829e-01 2.79636264e-01
1.25832045e+00 -8.89164209e-02 1.89754263e-01 -6.60148263e-01
-8.13042700e-01 7.63430238e-01 6.10882938e-01 1.09382838e-01
-1.57149062e-01 6.93801880e-01 1.93282306e-01 -1.65359616e-01
-1.90096349e-01 1.12120223e+00 -3.79731238e-01 -2.57882267e-01
-7.36218631e-01 4.51094657e-01 7.65174627e-01 1.31363547e+00
6.45947576e-01 -5.43459989e-02 -9.03531164e-02 4.95425731e-01
4.27192867e-01 3.47447664e-01 -2.05352738e-01 -1.33052981e+00
5.34770608e-01 6.41954780e-01 5.39496720e-01 -7.12481499e-01
-6.20745420e-01 -1.40157998e-01 -4.52153146e-01 4.47453052e-01
2.97498703e-01 -7.37953603e-01 -7.17099726e-01 1.24300611e+00
1.80052668e-01 -2.74776369e-01 -1.27103791e-01 6.71155095e-01
1.74822107e-01 6.55195475e-01 8.20425376e-02 2.45219350e-01
9.76834953e-01 -1.21930349e+00 -5.66171825e-01 -3.89003038e-01
9.54338253e-01 -5.65724850e-01 1.22194445e+00 4.47879672e-01
-1.32121289e+00 -4.60310221e-01 -1.30168331e+00 1.28787428e-01
-7.99094960e-02 6.63580358e-01 9.15644407e-01 6.08326614e-01
-9.92077529e-01 9.84568000e-01 -1.14301288e+00 1.83007941e-01
6.43504322e-01 4.86124516e-01 8.41930807e-02 -8.90604779e-02
-6.81598306e-01 7.39821732e-01 1.05763815e-01 5.52668333e-01
-1.23546720e+00 -7.17841029e-01 -1.11263764e+00 2.66101629e-01
5.26525915e-01 -5.02712607e-01 1.90073097e+00 -7.97381848e-02
-1.86328566e+00 1.19500421e-01 3.55170041e-01 4.92366739e-02
4.43300843e-01 -5.04546046e-01 2.96807140e-01 3.56026068e-02
-1.57906506e-02 6.46373034e-01 3.35926503e-01 -1.39011800e+00
-6.39585078e-01 2.66644228e-02 3.74346256e-01 3.19605246e-02
-1.83776960e-01 -3.80558819e-01 -5.66597402e-01 -1.50885358e-01
-9.49888229e-02 -7.61254728e-01 -7.38882542e-01 6.02731295e-02
-5.96455812e-01 -2.87094235e-01 6.02330029e-01 -7.20484972e-01
1.35509026e+00 -2.08864260e+00 1.52330309e-01 2.21104562e-01
1.64119769e-02 5.50160147e-02 -8.07855427e-02 6.76456988e-01
3.26651335e-01 2.24412441e-01 -1.35145292e-01 -1.98073894e-01
2.61772990e-01 2.87383139e-01 7.73351490e-02 1.55355990e-01
5.75658023e-01 9.60071206e-01 -1.27664673e+00 -4.74941373e-01
3.37719917e-01 8.11233744e-02 -8.63968134e-01 7.30048954e-01
-4.89168286e-01 3.45396668e-01 -8.02960932e-01 6.34461343e-01
4.04095381e-01 -4.24176157e-01 4.82107490e-01 -1.66544855e-01
-4.46631879e-01 8.97266865e-02 -1.08422768e+00 2.03828788e+00
-9.72186506e-01 5.12390137e-01 -3.15494165e-02 -7.15613246e-01
1.11995137e+00 -2.08233614e-02 3.78344029e-01 -4.60719138e-01
1.80575982e-01 1.03191227e-01 -9.06015784e-02 -8.00621212e-01
5.10053098e-01 3.93497586e-01 -4.60217476e-01 5.48212349e-01
-3.66181999e-01 -7.78738320e-01 4.85837162e-01 -4.05825488e-02
1.47370088e+00 7.45007873e-01 -2.69246370e-01 3.87296575e-04
7.15587735e-02 1.01159252e-01 5.18537939e-01 5.00804961e-01
2.79293835e-01 2.74414986e-01 8.47833633e-01 -3.91014993e-01
-1.33245277e+00 -9.03568625e-01 6.27980232e-01 6.52113795e-01
1.99762195e-01 -2.98688203e-01 -1.18257844e+00 -6.78440154e-01
-3.20650339e-02 8.59782279e-01 -2.82375127e-01 -2.13458255e-01
-9.03716028e-01 -7.89407790e-02 4.01026249e-01 9.69923556e-01
2.64620185e-01 -1.12001240e+00 -1.24530697e+00 3.44754010e-01
2.86043286e-01 -8.35483193e-01 -7.37791359e-01 1.79248765e-01
-6.79037809e-01 -1.00160038e+00 -5.58755040e-01 -1.04302561e+00
1.24650407e+00 7.57828057e-02 1.05003154e+00 3.13122571e-01
-6.81402206e-01 -3.15547548e-02 -2.97697306e-01 -2.94925660e-01
-3.57921600e-01 2.97040284e-01 -5.02376020e-01 -7.08551288e-01
-2.84131527e-01 -4.33520257e-01 -8.31470370e-01 2.89464444e-01
-6.39882505e-01 5.21026850e-01 9.45859790e-01 5.03715932e-01
2.36487404e-01 6.09312892e-01 5.80173373e-01 -5.68686306e-01
6.68901563e-01 -5.68699054e-02 -1.16772985e+00 2.43546322e-01
-2.89099842e-01 3.25034708e-01 7.17993617e-01 -1.59276873e-01
-1.15993142e+00 7.65582085e-01 6.96254941e-03 -2.14611202e-01
-1.68731600e-01 4.24785823e-01 -5.51244438e-01 2.01401547e-01
1.85364291e-01 -3.82796824e-01 -2.55073100e-01 -5.43661773e-01
4.05580670e-01 4.99350488e-01 2.70242333e-01 -9.66245890e-01
1.02484083e+00 4.97492738e-02 8.78956988e-02 -3.32807213e-01
-6.78605974e-01 8.84470642e-02 -9.37535882e-01 -3.20778430e-01
8.30319166e-01 -4.70353067e-01 -1.13140464e+00 2.93865085e-01
-1.34315825e+00 -1.08816862e+00 -2.78167814e-01 9.38610882e-02
-5.90521455e-01 7.51335025e-02 -4.99875247e-01 -9.85719144e-01
-3.50984856e-02 -1.39462519e+00 1.19924355e+00 5.14685810e-01
-2.57893771e-01 -6.82601571e-01 -1.69390172e-01 -5.88589988e-04
2.25556001e-01 4.10509676e-01 9.62692499e-01 1.70964181e-01
-1.06278086e+00 -3.60765291e-04 -1.74138889e-01 5.69815673e-02
3.35848749e-01 3.90434176e-01 -5.10078311e-01 -9.02136937e-02
-7.73118138e-01 -2.34757647e-01 -6.39201002e-03 4.41645205e-01
1.71239030e+00 -1.15018949e-01 -4.16368753e-01 4.17269200e-01
1.21924245e+00 6.33150637e-01 3.97779793e-01 5.52075922e-01
6.52275145e-01 9.34578538e-01 1.02661896e+00 6.30184829e-01
3.74288499e-01 2.74008304e-01 4.53925371e-01 -3.21317613e-01
2.94735909e-01 -7.08412647e-01 1.01529267e-02 4.73348171e-01
1.35405779e-01 -3.27869833e-01 -9.97898459e-01 6.91896021e-01
-1.99019647e+00 -5.09648979e-01 2.38153145e-01 1.83239591e+00
8.51559818e-01 3.24199706e-01 -2.22502174e-04 2.15096906e-01
4.50822592e-01 -3.25235963e-01 -5.56414902e-01 -5.26587248e-01
6.56565785e-01 2.34861597e-01 7.53534198e-01 3.64143759e-01
-9.53128934e-01 1.00610387e+00 6.52921629e+00 3.18939656e-01
-7.44143963e-01 -5.12060463e-01 5.81869245e-01 -5.02168052e-02
-4.32113558e-01 1.97593242e-01 -5.70168018e-01 2.85100371e-01
8.24358165e-01 4.13689315e-02 6.06109202e-01 9.92073476e-01
5.94435275e-01 -1.61879629e-01 -1.45044219e+00 6.96711838e-01
-5.01874745e-01 -1.56712425e+00 -4.57999170e-01 6.34268150e-02
6.01245821e-01 -7.80159891e-01 -2.66843736e-01 1.98374480e-01
9.54494178e-01 -1.10710549e+00 9.63073254e-01 3.73508751e-01
6.10343516e-01 -1.11226034e+00 4.72334057e-01 3.15009713e-01
-1.20191979e+00 -2.77607679e-01 -8.35716128e-02 -8.11948255e-02
6.12894356e-01 5.06922305e-01 -1.06559622e+00 4.24798489e-01
6.76980376e-01 4.26413774e-01 -4.18360859e-01 1.18148434e+00
-7.01797485e-01 2.83985019e-01 -6.32604957e-02 -4.67103988e-01
2.64877349e-01 -1.18435159e-01 -2.93271452e-01 9.75246310e-01
4.78160322e-01 -2.29229987e-01 3.50838453e-01 1.24865735e+00
8.49861950e-02 -4.41029817e-01 -4.39739764e-01 -3.27521503e-01
8.29358459e-01 1.41772413e+00 -9.02971208e-01 1.96840242e-01
-6.30153567e-02 8.42043281e-01 3.57675999e-01 2.53558934e-01
-1.05496144e+00 -9.01862502e-01 3.70819211e-01 6.93677440e-02
4.04673278e-01 -7.87843406e-01 -5.00307083e-01 -2.05618382e-01
3.65336053e-02 -4.02467340e-01 -1.96677402e-01 -1.02134311e+00
-8.13722372e-01 2.94210494e-01 -1.86894145e-02 -1.00266337e+00
-1.80922240e-01 -7.97355413e-01 -6.04089499e-01 7.27143228e-01
-1.53037977e+00 -1.01959753e+00 -4.53520894e-01 -2.79553354e-01
7.24485934e-01 1.52960801e-02 5.93073189e-01 3.46023321e-01
-8.66721928e-01 3.72226626e-01 -3.30426514e-01 2.99758196e-01
3.40062290e-01 -1.51999140e+00 8.41886342e-01 5.97436249e-01
-3.91634524e-01 6.31663322e-01 6.72824323e-01 -6.67104483e-01
-1.77350962e+00 -1.04297519e+00 3.60932410e-01 -1.89450994e-01
5.42114198e-01 -3.91932428e-01 -4.09348905e-01 4.34937119e-01
1.62210003e-01 -3.42296660e-01 5.24220765e-01 -1.56765327e-01
1.70593768e-01 3.84446494e-02 -8.53809237e-01 8.26380968e-01
9.71075714e-01 -6.46130294e-02 -9.05110464e-02 3.61731529e-01
9.68050420e-01 -6.73824966e-01 -9.14964139e-01 6.27109259e-02
5.09387493e-01 -4.81090218e-01 8.38415444e-01 -3.18584770e-01
7.63582528e-01 -6.67245328e-01 3.39591146e-01 -1.46929359e+00
-3.83181721e-01 -9.02394533e-01 -3.29829976e-02 1.21030438e+00
6.11399531e-01 1.13053858e-01 1.08405268e+00 8.93010378e-01
-5.14378965e-01 -9.35008883e-01 -2.28650391e-01 -5.67557514e-01
-9.56580937e-02 -1.53769821e-01 1.10335469e+00 2.42410302e-01
-1.17378965e-01 2.78089792e-01 -7.66109154e-02 4.95824754e-01
5.15167296e-01 8.45057964e-02 8.10333729e-01 -9.21580911e-01
-6.08099997e-01 -3.12247723e-01 2.91395873e-01 -1.18830132e+00
9.92524251e-02 -2.66497314e-01 7.05621064e-01 -2.11614084e+00
-2.86105663e-01 -9.68550444e-01 2.96823978e-01 5.57647645e-01
1.98344052e-01 -5.91976643e-01 -2.99701169e-02 -2.89636612e-01
-5.48592329e-01 5.05221248e-01 1.65511847e+00 -1.32953554e-01
-5.28121471e-01 5.97620867e-02 -7.60386527e-01 7.12374330e-01
9.35453475e-01 -1.11325167e-01 -6.07029438e-01 -1.01940906e+00
5.01441181e-01 2.14997053e-01 -8.63015279e-03 -8.18451166e-01
3.20259839e-01 -4.80081588e-01 5.91737866e-01 -8.67314756e-01
4.10574749e-02 -9.56486404e-01 -1.72987580e-01 7.21076071e-01
-3.27041388e-01 2.84640670e-01 2.99207449e-01 2.88089424e-01
2.82748342e-01 -6.51209652e-01 3.95007074e-01 -1.50114179e-01
-5.71493089e-01 1.20835245e-01 -5.68683982e-01 -3.13334912e-01
1.19955909e+00 -3.42452340e-02 -1.77717194e-01 -4.41468833e-03
-4.40626472e-01 7.39951849e-01 4.34618533e-01 3.19851846e-01
5.93558669e-01 -1.10673261e+00 -2.09124222e-01 1.30680397e-01
-2.28174895e-01 9.11288440e-01 1.76604703e-01 -9.01873969e-03
-1.22359943e+00 5.55102862e-02 -1.68186352e-01 -4.21013832e-01
-8.51152182e-01 5.68261385e-01 2.25885168e-01 -3.82217854e-01
-7.18328357e-01 8.30427229e-01 2.56782584e-03 -4.77784842e-01
3.70989740e-01 -4.49409306e-01 6.97705448e-02 -4.29600358e-01
4.29878056e-01 4.12690848e-01 -1.71519041e-01 3.43264520e-01
-2.29052708e-01 6.03950083e-01 1.09925166e-01 -2.10160598e-01
1.64828193e+00 1.88100129e-01 1.10207506e-01 3.72928083e-02
7.14769661e-01 -2.66564697e-01 -1.92311430e+00 4.26383287e-01
3.93138647e-01 -4.72593486e-01 -2.40231082e-02 -6.87356830e-01
-9.75916743e-01 1.01054525e+00 1.77180469e-01 1.02514893e-01
9.18688774e-01 -1.71706125e-01 8.14038873e-01 3.76856327e-01
4.10047799e-01 -1.57470429e+00 6.71300173e-01 1.69097751e-01
8.93278122e-01 -5.68857908e-01 2.26954110e-02 -4.94108945e-01
-5.33767521e-01 1.15907192e+00 1.06208968e+00 -1.49137557e-01
3.33200663e-01 7.69876003e-01 -2.78681159e-01 -1.64954409e-01
-4.85056788e-01 2.35184267e-01 -2.45435774e-01 7.18225062e-01
4.28016424e-01 5.92604317e-02 2.59828836e-01 6.52858078e-01
-2.30003476e-01 -1.81190282e-01 7.06055045e-01 1.50590420e+00
-5.02116799e-01 -1.51241994e+00 -2.09906146e-01 5.89383207e-02
1.24743976e-01 4.21595901e-01 -1.21726342e-01 5.89514792e-01
3.67258266e-02 9.57842290e-01 1.11314043e-01 -3.19013983e-01
7.98280478e-01 -5.29907942e-01 5.97879827e-01 -7.92524636e-01
-4.20482457e-01 5.17104799e-03 3.19607913e-01 -4.39323515e-01
1.46305695e-01 -4.12564844e-01 -1.71869278e+00 -3.30064222e-02
-3.10593963e-01 1.76458165e-01 8.68782938e-01 7.69878626e-01
3.50631922e-01 1.29656446e+00 9.04149830e-01 -1.20574951e+00
-6.31329000e-01 -5.26938021e-01 -4.81217235e-01 1.41956490e-02
2.30610166e-02 -8.10560286e-01 2.07245767e-01 1.65094376e-01]
|
[5.495396137237549, 2.9416074752807617]
|
f0d0ab7e-d556-414e-95c1-3b102a0841c4
|
harnessing-the-power-of-adversarial-prompting
|
2306.11648
| null |
https://arxiv.org/abs/2306.11648v1
|
https://arxiv.org/pdf/2306.11648v1.pdf
|
Harnessing the Power of Adversarial Prompting and Large Language Models for Robust Hypothesis Generation in Astronomy
|
This study investigates the application of Large Language Models (LLMs), specifically GPT-4, within Astronomy. We employ in-context prompting, supplying the model with up to 1000 papers from the NASA Astrophysics Data System, to explore the extent to which performance can be improved by immersing the model in domain-specific literature. Our findings point towards a substantial boost in hypothesis generation when using in-context prompting, a benefit that is further accentuated by adversarial prompting. We illustrate how adversarial prompting empowers GPT-4 to extract essential details from a vast knowledge base to produce meaningful hypotheses, signaling an innovative step towards employing LLMs for scientific research in Astronomy.
|
['Kartheik Iyer', 'Sandor Kruk', 'Yuan-Sen Ting', 'Ioana Ciucă']
|
2023-06-20
| null | null | null | null |
['astronomy']
|
['miscellaneous']
|
[ 8.54748487e-02 5.33844948e-01 -2.49835864e-01 1.22236304e-01
-8.68087590e-01 -1.07947445e+00 9.71830964e-01 7.20779672e-02
-3.56182814e-01 9.05623078e-01 4.81980771e-01 -7.86303937e-01
-2.66155392e-01 -6.01451755e-01 -9.89960790e-01 -9.72337574e-02
3.60885598e-02 1.92170218e-01 -3.17714512e-01 -1.01228736e-01
6.61274016e-01 7.80767202e-01 -9.33295667e-01 2.13471919e-01
9.24517989e-01 2.66780466e-01 1.99197367e-01 8.14070523e-01
-5.26822746e-01 9.70863223e-01 -1.07418311e+00 -5.74754655e-01
2.29501247e-01 1.11936264e-01 -8.49849045e-01 -4.36946481e-01
5.35306931e-01 1.70755342e-01 -3.26460570e-01 7.91305542e-01
5.62047064e-01 1.13011980e-02 4.96761978e-01 -1.10241246e+00
-8.82353902e-01 7.13802338e-01 4.72230166e-02 7.41859674e-01
6.67888820e-01 5.44368386e-01 9.21291173e-01 -9.24157143e-01
9.09768999e-01 1.49495947e+00 5.38368881e-01 4.18169707e-01
-1.01250517e+00 -8.86502922e-01 6.51776642e-02 -1.63530502e-02
-1.26966000e+00 -4.13901329e-01 5.37848055e-01 -5.70240319e-01
1.35857499e+00 5.33130825e-01 4.62524533e-01 1.52334011e+00
4.51666117e-01 4.33498055e-01 1.26308036e+00 -5.46271205e-01
3.93256724e-01 2.63071775e-01 -3.56478021e-02 4.87873137e-01
4.15502399e-01 6.22674167e-01 -7.75135219e-01 -5.53420961e-01
8.05747211e-01 -6.93391263e-01 -8.33853334e-02 6.25961840e-01
-1.55245697e+00 6.73088491e-01 4.33042377e-01 1.85524613e-01
-4.13364470e-01 -1.24384486e-03 8.25716108e-02 3.58558595e-01
4.32806551e-01 1.60877967e+00 -5.78550935e-01 1.02679273e-02
-1.03575075e+00 7.32332349e-01 9.96062398e-01 1.14840460e+00
4.44205664e-02 1.91450909e-01 -4.78182197e-01 3.31956089e-01
2.73361474e-01 6.08804405e-01 4.66015548e-01 -1.16025472e+00
3.99474680e-01 4.36145782e-01 1.90030321e-01 -7.97143638e-01
-2.00112119e-01 -8.55744660e-01 -8.54941383e-02 5.57302088e-02
1.27173632e-01 -4.20719326e-01 -7.82147944e-01 1.71386552e+00
1.16688311e-01 6.39855623e-01 4.38106477e-01 5.13060927e-01
1.22663474e+00 5.39612293e-01 5.94203234e-01 4.89280820e-02
1.39387918e+00 -8.14455688e-01 -4.40680593e-01 -5.45688570e-01
7.05345154e-01 -9.00408149e-01 1.11309600e+00 5.05321205e-01
-9.66021895e-01 -4.60437357e-01 -9.73748088e-01 -1.40379950e-01
-6.13836825e-01 -4.02438372e-01 7.28589118e-01 5.78705072e-01
-8.74768436e-01 3.50775987e-01 -3.92101079e-01 -2.27402061e-01
4.63843882e-01 1.05300449e-01 -2.49060884e-01 -7.14598894e-02
-1.38173187e+00 1.00400341e+00 6.23396754e-01 -4.26170319e-01
-8.58952999e-01 -1.34790885e+00 -6.53337061e-01 -1.00090243e-01
5.82713068e-01 -9.60853279e-01 1.29200149e+00 -1.74873486e-01
-1.14261782e+00 4.95122343e-01 -1.42609775e-01 -5.67766011e-01
2.38618657e-01 -8.49797800e-02 -7.60175586e-01 2.49397442e-01
1.46871775e-01 9.01250482e-01 6.21510029e-01 -1.17600179e+00
-3.11085343e-01 1.67912364e-01 2.23670617e-01 3.27180438e-02
-2.52793491e-01 4.48349178e-01 -3.51722389e-01 -1.10876572e+00
-4.87506926e-01 -8.25235188e-01 -5.70551455e-01 -3.66969913e-01
-6.72639549e-01 -2.60595173e-01 6.47125781e-01 -8.76573265e-01
1.17516255e+00 -1.62423301e+00 1.23191094e-02 1.36826202e-01
3.23885441e-01 3.36827546e-01 -4.24064904e-01 3.58091623e-01
6.65840283e-02 7.87824094e-01 1.45585313e-01 9.95676145e-02
-1.48565531e-01 3.25184941e-01 -7.06642449e-01 -2.60096431e-01
5.47715306e-01 1.42366064e+00 -1.08416259e+00 -4.95186597e-01
-1.03902847e-01 1.09165825e-01 -7.05437481e-01 1.63402990e-01
-7.72954226e-01 7.56045818e-01 -8.25692594e-01 8.67025375e-01
2.17582628e-01 -2.60661632e-01 -1.67121276e-01 3.41785967e-01
-2.92139471e-01 4.58018512e-01 -7.14581370e-01 1.63359177e+00
-4.88967568e-01 6.07717931e-01 -1.69880778e-01 -6.53500319e-01
7.73631036e-01 4.26299363e-01 9.50977858e-03 -5.38959742e-01
-1.91079184e-01 2.14516986e-02 6.58523440e-02 -7.14005709e-01
4.17397141e-01 -2.44378701e-01 -2.53946990e-01 3.89584810e-01
1.86484292e-01 -4.25683916e-01 7.51905069e-02 5.14406085e-01
1.26665187e+00 -3.06372810e-02 1.20492838e-01 -2.81409472e-01
2.56569594e-01 1.99250206e-01 3.79576206e-01 1.27929556e+00
1.21922128e-01 1.41696021e-01 3.38366807e-01 -3.86131704e-02
-9.08191741e-01 -8.69283140e-01 -4.55470644e-02 6.19665802e-01
-6.20320201e-01 -6.48737729e-01 -3.89961481e-01 -9.90810454e-01
1.89505786e-01 1.26088607e+00 -4.62835014e-01 -2.33724669e-01
-3.04068238e-01 -4.47879940e-01 1.16557801e+00 2.77691573e-01
3.42781872e-01 -1.37568676e+00 -3.54947925e-01 4.94308993e-02
-1.44299418e-01 -1.36584675e+00 -1.39757484e-01 -9.43517014e-02
-8.15303624e-01 -7.68402278e-01 -5.53464651e-01 -4.45556015e-01
4.36507434e-01 -2.05565810e-01 1.35645735e+00 -1.38781473e-01
-5.51100731e-01 6.53189778e-01 -1.71111166e-01 -9.64385867e-01
-9.16549921e-01 5.93880862e-02 9.80094541e-03 -8.95086288e-01
2.73334891e-01 -3.96574318e-01 -9.49068815e-02 -8.56272876e-02
-1.05128193e+00 -3.93164679e-02 8.66906941e-01 4.60098535e-01
2.63065755e-01 -4.92780246e-02 1.00081992e+00 -9.36262667e-01
8.91148210e-01 -8.64691079e-01 -4.99326885e-01 2.58531511e-01
-7.15427041e-01 1.58376411e-01 5.98992407e-01 -5.17277181e-01
-1.13808930e+00 -5.52748263e-01 -5.57828285e-02 -3.03410500e-01
-2.21990332e-01 1.12908328e+00 -1.03007086e-01 -5.44983327e-01
9.15607154e-01 -6.36831522e-02 -2.67286658e-01 -4.02591676e-01
7.27092683e-01 5.05909443e-01 6.70132518e-01 -8.82082522e-01
1.07596767e+00 -1.17085487e-01 1.57746732e-01 -7.30784416e-01
-8.54502738e-01 -9.32230875e-02 -1.83455259e-01 -8.98820814e-03
6.03675306e-01 -9.39418614e-01 -4.21818316e-01 -2.67627597e-01
-9.44927812e-01 -1.69734493e-01 -3.62795740e-01 4.77529287e-01
-2.16945380e-01 3.42777371e-01 -2.84038544e-01 -5.24700701e-01
-3.51939410e-01 -8.44523251e-01 7.75812089e-01 5.22003472e-01
-5.44703484e-01 -1.33870244e+00 -5.78869432e-02 5.87643802e-01
3.59951407e-01 2.35492945e-01 1.17260516e+00 -1.16778946e+00
-6.63290024e-01 -2.46127248e-01 5.77822551e-02 1.54531717e-01
-2.18531996e-01 -1.95968986e-01 -9.50522304e-01 -5.16219288e-02
1.59019619e-01 -4.13402826e-01 4.29507941e-01 4.26739380e-02
1.35064209e+00 -6.67420387e-01 -3.91552210e-01 4.76644814e-01
1.07074964e+00 2.62841843e-02 4.75594431e-01 4.69926745e-01
6.55184388e-01 5.75194478e-01 7.30558932e-01 8.43570158e-02
5.95430173e-02 3.55820835e-01 3.68089341e-02 3.48528847e-02
-1.27013385e-01 -5.48825145e-01 2.02371523e-01 6.09855115e-01
2.71259040e-01 -3.38019758e-01 -1.06317842e+00 4.75761831e-01
-1.27395308e+00 -8.02799046e-01 -5.85750379e-02 1.95631218e+00
1.15824115e+00 4.41506654e-01 -4.27208632e-01 -4.88667756e-01
1.10874437e-01 5.33760563e-02 -5.87233007e-01 -4.62093145e-01
-3.54636759e-01 6.46165311e-01 6.31050766e-01 5.29700756e-01
-5.64009488e-01 1.13068771e+00 8.29613018e+00 1.08319438e+00
-9.55663681e-01 -7.25924969e-02 3.67759436e-01 -3.58390421e-01
-8.03860843e-01 3.25637937e-01 -7.03472435e-01 4.73387271e-01
1.54744995e+00 -7.37675607e-01 3.22339922e-01 5.13144791e-01
3.60977292e-01 1.47374973e-01 -9.61577058e-01 3.79936904e-01
-1.21574298e-01 -1.91270149e+00 4.53970999e-01 2.77019203e-01
7.04187512e-01 6.34944737e-02 1.36371136e-01 7.34615386e-01
5.17681956e-01 -1.43009889e+00 3.43585879e-01 6.75773919e-01
5.71525455e-01 -5.67845762e-01 3.12132925e-01 5.96398354e-01
-4.80353653e-01 -1.32331416e-01 -4.01855707e-01 -1.75372168e-01
-8.82774740e-02 5.58288336e-01 -1.67389178e+00 9.52782094e-01
2.94421375e-01 3.80601913e-01 -9.88583744e-01 9.90073502e-01
-3.38537395e-01 1.27610743e+00 -1.79856911e-01 1.41974673e-01
1.75318539e-01 2.88613856e-01 1.08482158e+00 1.62253928e+00
2.65811920e-01 1.52954116e-01 1.94272652e-01 1.21241176e+00
-4.49855864e-01 -3.58011760e-02 -8.46003354e-01 -8.46490383e-01
8.79980922e-01 1.25337064e+00 -3.10219198e-01 -6.11073613e-01
-4.25988168e-01 4.21990663e-01 1.01430543e-01 4.42130297e-01
-4.85958040e-01 -2.00609863e-01 3.31558019e-01 -1.06869929e-01
2.72757746e-02 -3.12160224e-01 -7.47657359e-01 -8.20646286e-01
-2.63942987e-01 -1.25662732e+00 2.80304641e-01 -1.27778542e+00
-1.32703328e+00 2.45792195e-01 1.10338792e-01 -7.40406692e-01
-1.69332847e-01 -4.88774210e-01 -6.94911480e-01 1.35595000e+00
-1.30420160e+00 -1.33872509e+00 1.11433662e-01 1.25205323e-01
5.76688647e-01 -5.53940892e-01 1.00593674e+00 -2.03090012e-02
-4.04723585e-01 6.78860009e-01 -3.36274028e-01 -2.25227118e-01
8.24651778e-01 -1.31501389e+00 9.60383892e-01 9.50433731e-01
2.78623998e-01 1.13269866e+00 1.06015921e+00 -1.30079353e+00
-1.43233156e+00 -1.25114894e+00 1.24694848e+00 -1.24796844e+00
8.85633647e-01 -2.57153660e-01 -1.06765974e+00 7.52627611e-01
2.60482043e-01 -3.48894536e-01 1.05048966e+00 -4.27352674e-02
-3.67868870e-01 7.69296467e-01 -1.28031540e+00 8.85219395e-01
9.70538259e-01 -6.02089942e-01 -1.12106705e+00 5.02827764e-01
1.30443633e+00 -4.86650288e-01 -1.30151296e+00 3.39594215e-01
1.69455841e-01 -6.30820245e-02 1.33814180e+00 -1.19301689e+00
7.19031453e-01 -8.53899047e-02 1.17973112e-01 -1.26350522e+00
-3.01310033e-01 -9.28246498e-01 -3.19547117e-01 1.28008199e+00
4.86625016e-01 -5.92011154e-01 4.61152077e-01 6.75562978e-01
-2.78044164e-01 -5.45318902e-01 -7.30617523e-01 -8.13535035e-01
7.03436196e-01 -7.08009005e-01 5.91708064e-01 1.24204242e+00
-2.04944819e-01 1.61791489e-01 -1.15948573e-01 5.31881154e-01
4.27715093e-01 -2.80505389e-01 6.59529150e-01 -1.22303236e+00
-4.62544829e-01 -3.18913102e-01 -3.62421162e-02 -6.97129607e-01
3.79099369e-01 -1.25329292e+00 -3.05170149e-01 -1.38262999e+00
-9.50792059e-02 -3.69416833e-01 -4.05963629e-01 5.43881059e-01
-5.39309204e-01 6.09856322e-02 1.74540102e-01 1.66933998e-01
-2.98926830e-01 1.40589699e-01 1.26740599e+00 9.94650871e-02
6.08546771e-02 -2.89717227e-01 -1.51848340e+00 5.86958528e-01
6.70166135e-01 -6.18052483e-01 -3.21409464e-01 2.66716088e-04
2.24886298e-01 -1.32709607e-01 6.24814987e-01 -8.13567042e-01
2.59459496e-01 -4.29906279e-01 6.58248782e-01 -1.65314302e-01
-3.76505926e-02 -2.35445634e-01 -1.77240949e-02 3.15171510e-01
-8.30627024e-01 -1.64157636e-02 1.06538224e+00 6.10380232e-01
3.23927924e-02 -4.24377292e-01 5.65457828e-02 -4.42291707e-01
-7.13050544e-01 -1.38940886e-01 -5.80222905e-01 2.93308198e-01
6.65774584e-01 1.91368401e-01 -5.15528262e-01 -1.99972719e-01
-6.68130934e-01 3.54925096e-01 2.57481635e-01 7.10897923e-01
3.47775728e-01 -1.06519496e+00 -8.08694363e-01 1.77110970e-01
8.29510763e-02 -1.39670670e-01 1.14204124e-01 3.77112597e-01
-4.74479795e-02 9.28416908e-01 9.04994644e-03 -1.66613087e-01
-1.01896822e+00 7.59037614e-01 -2.83464417e-02 -3.37954670e-01
-4.64804649e-01 1.05923009e+00 5.18213511e-02 -5.64323366e-01
1.46985471e-01 -2.80210644e-01 -6.24495037e-02 -4.45268840e-01
4.59575176e-01 1.89984813e-01 2.13108491e-02 6.39222041e-02
-2.95040816e-01 2.46384889e-02 -2.21392855e-01 -5.06604612e-01
1.00366569e+00 1.69301048e-01 9.14873853e-02 2.28920683e-01
6.35946572e-01 7.79043317e-01 -9.10538495e-01 -2.31584683e-01
3.15095156e-01 -2.89028317e-01 3.36360395e-01 -1.75649428e+00
-3.75702053e-01 4.10994351e-01 9.19305980e-02 4.16645780e-02
7.72749007e-01 5.24556637e-02 4.57887203e-01 6.99042439e-01
1.29990697e-01 -7.05373347e-01 1.17756747e-01 5.11981606e-01
1.26142466e+00 -1.04963386e+00 6.21178187e-02 -2.47968435e-01
-6.98101223e-01 8.80962312e-01 6.17677093e-01 3.26005742e-02
3.54337394e-01 4.00006086e-01 9.79346558e-02 -5.22758365e-01
-9.51890945e-01 2.79259324e-01 5.69857121e-01 5.16866684e-01
3.29957813e-01 -1.94721386e-01 -2.12324500e-01 8.04690838e-01
-5.91452599e-01 2.33158946e-01 5.45945704e-01 1.20142198e+00
-1.98313564e-01 -1.34329331e+00 -7.22389102e-01 4.37132329e-01
-7.49315917e-01 -4.64926451e-01 -8.52970421e-01 7.62592971e-01
1.10193707e-01 1.08799851e+00 -3.30588490e-01 -2.71196514e-01
2.00917363e-01 4.99361217e-01 2.14429453e-01 -9.53957558e-01
-8.67719650e-01 -6.39135092e-02 4.88902926e-01 -3.29208642e-01
4.37379964e-02 -4.34891135e-01 -1.16203034e+00 -9.74925235e-02
2.81187780e-02 3.34795088e-01 7.82668829e-01 1.17063427e+00
9.01308239e-01 9.89056230e-01 6.19383268e-02 -4.48402017e-01
-4.72818404e-01 -1.10290468e+00 1.59889579e-01 -7.72892982e-02
1.44417286e-01 -4.12971973e-01 -3.69769305e-01 1.08151250e-02]
|
[10.861407279968262, 8.287883758544922]
|
d769feae-1f31-4c11-a671-a4830810a56d
|
joint-debiased-representation-and-image
|
2209.06941
| null |
https://arxiv.org/abs/2209.06941v1
|
https://arxiv.org/pdf/2209.06941v1.pdf
|
Joint Debiased Representation and Image Clustering Learning with Self-Supervision
|
Contrastive learning is among the most successful methods for visual representation learning, and its performance can be further improved by jointly performing clustering on the learned representations. However, existing methods for joint clustering and contrastive learning do not perform well on long-tailed data distributions, as majority classes overwhelm and distort the loss of minority classes, thus preventing meaningful representations to be learned. Motivated by this, we develop a novel joint clustering and contrastive learning framework by adapting the debiased contrastive loss to avoid under-clustering minority classes of imbalanced datasets. We show that our proposed modified debiased contrastive loss and divergence clustering loss improves the performance across multiple datasets and learning tasks. The source code is available at https://anonymous.4open.science/r/SSL-debiased-clustering
|
['Mina Rezaei', 'Shekoofeh Azizi', 'Bernd Bischl', 'Emilio Dorigatti', 'JaeEun Nam', 'Shunjie-Fabian Zheng']
|
2022-09-14
| null | null | null | null |
['image-clustering']
|
['computer-vision']
|
[-1.24918170e-01 -3.09547007e-01 -4.68043178e-01 -6.31387115e-01
-8.63704264e-01 -6.81434333e-01 5.21570742e-01 5.02281785e-01
-3.56302083e-01 5.85395932e-01 2.02463180e-01 -1.45932257e-01
-1.26917481e-01 -5.10601223e-01 -8.17646742e-01 -9.24525499e-01
4.70759571e-02 5.84699094e-01 -2.78824344e-02 3.89565855e-01
2.87875623e-01 2.43928522e-01 -1.54963720e+00 4.65605229e-01
1.03309667e+00 6.34819567e-01 5.76492064e-02 4.64379907e-01
-1.40359163e-01 6.06983244e-01 -7.61920035e-01 -2.02747002e-01
1.75587162e-01 -5.76873600e-01 -5.12235701e-01 -1.89390257e-01
6.07754469e-01 4.52847183e-02 -3.35420161e-01 1.09016597e+00
5.89834094e-01 2.22568095e-01 1.12246919e+00 -1.70291615e+00
-9.68839169e-01 5.68349004e-01 -1.24928868e+00 3.89629424e-01
-2.02220067e-01 2.71550566e-02 1.04293287e+00 -9.48184490e-01
3.00405711e-01 1.48717117e+00 3.85229826e-01 5.09904146e-01
-1.46378720e+00 -1.13078070e+00 5.07701576e-01 3.89917672e-01
-1.45580757e+00 -2.95408040e-01 9.28744376e-01 -6.70250297e-01
4.91237879e-01 1.54429913e-01 4.13298965e-01 8.91982853e-01
-2.48487979e-01 1.02371764e+00 1.09951961e+00 -2.52232969e-01
3.49888623e-01 -1.64889004e-02 2.53449827e-01 5.51114202e-01
5.00711441e-01 4.05274658e-03 -3.36339116e-01 -1.68417513e-01
5.03598511e-01 4.80323136e-01 -1.04283549e-01 -9.00303483e-01
-7.95675457e-01 9.96327400e-01 8.87408912e-01 1.38238996e-01
5.20352880e-03 3.01605016e-01 4.92910087e-01 2.06324294e-01
7.82945752e-01 1.26561865e-01 -1.98434651e-01 1.19179644e-01
-1.13614976e+00 1.79515898e-01 1.87089443e-01 7.02481806e-01
8.65052402e-01 -1.05001815e-01 -2.78298527e-01 1.09533024e+00
3.11275661e-01 4.59590971e-01 3.14556599e-01 -1.01004183e+00
2.35198200e-01 6.61896110e-01 -8.72713402e-02 -9.51356709e-01
-3.61513168e-01 -5.27621806e-01 -1.10831320e+00 4.94902521e-01
6.05830073e-01 -1.15865178e-01 -1.03785467e+00 1.98208392e+00
2.22089335e-01 -5.76584600e-03 -2.80123919e-01 7.89064705e-01
7.07894206e-01 7.17134237e-01 2.76178777e-01 -1.89530700e-01
9.47123349e-01 -7.98412621e-01 -6.62084281e-01 -1.98328316e-01
4.09146041e-01 -5.04462898e-01 1.30050075e+00 2.97419846e-01
-1.07736063e+00 -4.76207405e-01 -1.06579053e+00 -1.52527004e-01
-2.80588299e-01 -1.09511040e-01 4.79162753e-01 3.41734767e-01
-7.10330844e-01 3.16932857e-01 -9.98081625e-01 4.21773680e-02
1.14421904e+00 5.79036735e-02 -1.65549129e-01 -4.45167094e-01
-6.62280321e-01 5.81143320e-01 2.98210502e-01 -2.17179120e-01
-9.14164007e-01 -8.64293098e-01 -7.38861740e-01 6.67425320e-02
1.20065920e-01 -4.47355717e-01 7.57847309e-01 -1.09283280e+00
-7.49002278e-01 1.12763262e+00 -3.55427824e-02 -3.45942348e-01
6.77637398e-01 -3.71355087e-01 1.00001737e-01 8.99138302e-02
4.01732661e-02 8.87723446e-01 9.23670232e-01 -1.63073432e+00
-3.25251907e-01 -7.10099280e-01 -3.93137932e-01 3.99233729e-01
-3.63200456e-01 -3.40412289e-01 -3.23933542e-01 -8.09484839e-01
-6.25429526e-02 -6.20434105e-01 -1.63169488e-01 4.72810298e-01
-3.55997115e-01 -1.84018284e-01 9.42250431e-01 -5.19974828e-01
9.75931168e-01 -2.48665214e+00 1.88054651e-01 6.67035654e-02
4.76074785e-01 2.81305671e-01 -2.63569891e-01 1.11217521e-01
-2.71754175e-01 1.00452252e-01 -5.07467210e-01 -4.24583882e-01
3.64516228e-02 -5.35241468e-03 -2.36502364e-01 7.69595623e-01
4.83030230e-02 7.97649503e-01 -8.98692489e-01 -4.20621246e-01
2.76813090e-01 6.19770467e-01 -5.65567851e-01 3.98357034e-01
-1.86047778e-01 5.15722990e-01 1.09755836e-01 4.64837402e-01
1.07869637e+00 -3.36682409e-01 3.04288745e-01 9.87054966e-03
2.43456990e-01 2.15657111e-02 -1.10424471e+00 1.38994455e+00
-1.35192364e-01 7.12392151e-01 1.43498912e-01 -1.41310656e+00
9.45115268e-01 -2.25547835e-01 3.35253656e-01 -7.88123429e-01
-1.62601426e-01 -8.82602930e-02 -2.29274973e-01 -1.24776751e-01
1.27108037e-01 -4.46321040e-01 -2.62691323e-02 5.27899027e-01
-1.51390851e-01 -5.76250367e-02 -5.47938757e-02 3.58350903e-01
7.13672578e-01 5.39505742e-02 2.70203888e-01 -2.34110475e-01
-1.16449185e-01 -5.24255931e-02 7.66784847e-01 6.84932709e-01
-4.04322147e-01 7.91215181e-01 7.77751982e-01 -1.93088785e-01
-1.04066873e+00 -1.54697323e+00 -2.56637007e-01 1.42035651e+00
1.63051829e-01 -7.35365748e-02 -5.07148862e-01 -9.50601041e-01
4.11922038e-01 6.60094917e-01 -7.51455545e-01 -4.41469938e-01
-2.72884369e-01 -1.10165501e+00 4.89632487e-01 6.44458950e-01
1.79787040e-01 -7.63235688e-01 -1.30120188e-01 -2.16239467e-01
-9.89855900e-02 -2.59144574e-01 -3.99240434e-01 4.62412000e-01
-8.68945658e-01 -1.32699239e+00 -8.88305128e-01 -7.73866475e-01
7.69678593e-01 5.43923795e-01 1.10290420e+00 1.69682339e-01
-4.63912040e-01 2.79172778e-01 -2.04664856e-01 -4.03505176e-01
-3.64582628e-01 -1.18775934e-01 -1.54256269e-01 -1.93889275e-01
5.00469446e-01 -5.66633940e-01 -8.19863737e-01 2.05054864e-01
-9.46086824e-01 -1.81735769e-01 2.84493536e-01 8.81570399e-01
7.04423547e-01 -9.58682597e-02 7.79940486e-01 -9.93144572e-01
3.17596197e-01 -7.31111109e-01 -5.90858459e-01 1.09268397e-01
-7.49487877e-01 2.42164992e-02 5.07335961e-01 -4.79382843e-01
-8.71278346e-01 -1.74873024e-01 1.61497742e-01 -7.89363980e-01
-2.21551940e-01 9.05221030e-02 -9.38084647e-02 3.84289235e-01
7.18768835e-01 -6.39061853e-02 2.33022720e-01 -5.88673711e-01
5.81911981e-01 6.50033653e-01 4.65635419e-01 -4.32281762e-01
6.96064472e-01 7.14598477e-01 -2.86254704e-01 -4.82031673e-01
-9.87376928e-01 -5.53673089e-01 -5.20450473e-01 -6.00776225e-02
6.64610267e-01 -1.18500233e+00 -4.81395483e-01 5.33196211e-01
-4.97958481e-01 -5.35136461e-01 -4.23940539e-01 3.69553596e-01
-5.06141484e-01 5.56588829e-01 -4.80820745e-01 -7.27488697e-01
-1.68912083e-01 -7.04880834e-01 7.35855997e-01 1.77679598e-01
-1.23697639e-01 -9.66760933e-01 2.64368445e-01 4.37912166e-01
3.08567077e-01 3.55383575e-01 1.33117771e+00 -6.30999029e-01
-1.97717279e-01 5.26497401e-02 -4.75632995e-01 6.01581693e-01
1.75519466e-01 2.37282127e-01 -9.96940613e-01 -7.27740884e-01
-3.60137194e-01 -6.60257161e-01 1.53963590e+00 6.61701858e-01
1.59401166e+00 -2.08099872e-01 -3.11067820e-01 6.79115176e-01
1.29670691e+00 -1.03317901e-01 5.60529351e-01 1.26666129e-01
8.75976920e-01 6.88622415e-01 5.13994515e-01 6.93143785e-01
3.74975085e-01 4.30034399e-01 6.03589058e-01 -3.25156897e-01
-2.99109966e-01 -2.92939276e-01 1.29518345e-01 5.63594222e-01
2.46081203e-01 -1.83130428e-01 -9.87274885e-01 7.56234288e-01
-1.87726271e+00 -1.08070147e+00 6.87599927e-02 2.43193555e+00
9.85251367e-01 3.64450067e-02 4.52079535e-01 1.93365440e-01
1.12362456e+00 3.87040287e-01 -9.00145948e-01 -6.62364587e-02
-1.43533424e-01 4.71790880e-02 1.63122669e-01 4.46909487e-01
-1.28444052e+00 7.27184892e-01 6.09640265e+00 9.11308408e-01
-1.10786605e+00 1.10124446e-01 9.32089746e-01 -5.48645318e-01
-5.95845401e-01 -1.13688782e-01 -4.84866947e-01 6.33332610e-01
6.19012058e-01 -5.54808937e-02 2.01206714e-01 7.59852409e-01
5.69606461e-02 4.00514789e-02 -9.20052230e-01 1.10477638e+00
-1.83464866e-02 -1.21444154e+00 7.17641190e-02 4.45404761e-02
8.43430758e-01 1.21593222e-01 5.38978457e-01 3.81193370e-01
6.18849158e-01 -1.12570477e+00 6.09121859e-01 3.41134191e-01
8.76095593e-01 -1.08129096e+00 3.54470253e-01 2.32437506e-01
-8.86561930e-01 -1.95625395e-01 -6.33080125e-01 4.91943471e-02
-3.36748511e-01 1.00118399e+00 -5.31335354e-01 1.44488290e-02
8.56957853e-01 9.07282889e-01 -8.53206694e-01 1.22307444e+00
-3.85823101e-02 7.78166592e-01 -4.36474569e-02 1.97058201e-01
-1.11957714e-01 -1.40346810e-01 3.28852415e-01 1.31648397e+00
1.01208597e-01 -4.85704869e-01 2.06671938e-01 1.00127292e+00
-3.25992078e-01 9.01066046e-03 -5.55913270e-01 8.87506641e-03
7.62503326e-01 1.05675042e+00 -7.14134097e-01 -3.45540047e-01
-2.98310459e-01 7.70485282e-01 7.87683606e-01 5.17488956e-01
-7.96641886e-01 -3.69026512e-01 7.69002318e-01 2.00634331e-01
4.50654358e-01 -1.41404375e-01 -4.41149622e-01 -1.09839571e+00
-1.61667943e-01 -8.20816398e-01 1.06296587e+00 -5.13027072e-01
-1.79706275e+00 -1.41070522e-02 1.06095463e-01 -1.25694251e+00
6.22631200e-02 -3.70406240e-01 -6.11241698e-01 4.01551664e-01
-1.53156376e+00 -9.59845483e-01 -2.86555171e-01 5.69523215e-01
3.27298880e-01 8.53775516e-02 4.01712716e-01 2.04326212e-01
-5.81457138e-01 7.35289216e-01 6.67777658e-01 5.28528541e-02
1.11453915e+00 -1.52062869e+00 -8.68505314e-02 5.87510884e-01
1.78459451e-01 4.27355677e-01 5.25002301e-01 -4.59608138e-01
-8.03807735e-01 -1.26748633e+00 2.52650350e-01 -3.15304250e-01
2.45906025e-01 -6.02367043e-01 -1.36076939e+00 7.27694154e-01
1.54957250e-01 2.55140543e-01 9.43404853e-01 2.74423838e-01
-9.57037508e-01 -1.73157915e-01 -1.25118732e+00 3.45105588e-01
8.45173001e-01 -5.44981599e-01 -4.85267729e-01 3.72392923e-01
3.94010425e-01 3.63212936e-02 -7.44489372e-01 3.17005992e-01
3.29806745e-01 -1.03905606e+00 1.01851916e+00 -5.36730647e-01
2.24234894e-01 -3.55535388e-01 -2.63595372e-01 -1.35024738e+00
-4.23219204e-01 -1.46509901e-01 -2.20711991e-01 1.30971253e+00
1.56386748e-01 -5.14279604e-01 8.54979634e-01 3.99026312e-02
-8.39184131e-03 -4.37604696e-01 -8.92548501e-01 -7.99808443e-01
8.06250691e-01 -1.70743808e-01 2.60108680e-01 1.19424582e+00
-9.21832994e-02 2.38155454e-01 -2.91049719e-01 2.61764806e-02
1.04481149e+00 4.24008101e-01 6.71242595e-01 -1.31627738e+00
-1.96638837e-01 -5.86981535e-01 -2.70419151e-01 -9.21626627e-01
1.73479244e-01 -1.11175382e+00 3.06368787e-02 -1.39650190e+00
8.15928638e-01 -5.57109714e-01 -4.85910445e-01 5.74614048e-01
-5.83719015e-01 4.49991435e-01 2.29346082e-01 4.49766546e-01
-9.60487187e-01 8.40498567e-01 1.04621184e+00 -3.44747990e-01
-7.41622075e-02 -9.49593708e-02 -8.81682336e-01 5.72047710e-01
7.95681715e-01 -7.54586458e-01 -4.49829102e-01 -1.76494643e-01
6.98512932e-03 -3.96269470e-01 4.19773430e-01 -8.20321023e-01
-1.36323553e-02 -1.50944978e-01 8.66615117e-01 -5.97509563e-01
1.94021851e-01 -3.70029122e-01 -2.09871173e-01 4.92698669e-01
-6.45589352e-01 -9.83653441e-02 1.78360611e-01 8.06002140e-01
-2.25403413e-01 5.43857887e-02 1.37573612e+00 -2.29661143e-03
-4.21096623e-01 2.74236888e-01 -2.76152194e-01 6.01404309e-01
1.05997789e+00 5.20708784e-03 -7.55023539e-01 -3.91925305e-01
-6.49027705e-01 3.38298142e-01 7.31450558e-01 3.65069926e-01
7.23851323e-01 -1.37000299e+00 -8.37646663e-01 2.20020548e-01
2.25798368e-01 2.10360095e-01 4.64482218e-01 6.31424785e-01
-2.89625794e-01 -2.47828752e-01 -4.42339331e-01 -7.78771102e-01
-1.38137019e+00 8.54808748e-01 2.67720461e-01 1.31054558e-02
-4.98238355e-01 8.68564129e-01 6.24128401e-01 -6.35235012e-01
5.30348837e-01 8.96307677e-02 -1.64899155e-01 3.44647884e-01
6.11417174e-01 6.63326561e-01 -1.83333144e-01 -4.04474735e-01
-5.07383347e-01 4.56944972e-01 -3.58341366e-01 8.03969130e-02
1.47345912e+00 -1.67584941e-01 1.42530026e-02 6.48218274e-01
1.42386997e+00 -1.24647833e-01 -1.58501410e+00 -1.82665750e-01
-1.47528529e-01 -5.00090897e-01 -7.27703944e-02 -6.37868285e-01
-1.16309416e+00 1.26243854e+00 7.97356546e-01 7.98308328e-02
1.09061778e+00 3.00371736e-01 3.41508090e-01 1.05967782e-01
-1.66450217e-01 -1.08375371e+00 6.69384480e-01 1.47769257e-01
8.28788102e-01 -1.37499619e+00 1.22094348e-01 -1.05296448e-01
-8.17347229e-01 6.02798164e-01 7.40435600e-01 -2.81930268e-01
6.74366891e-01 2.79044390e-01 2.69552588e-01 -3.64170462e-01
-8.13857317e-01 -1.31635323e-01 2.43404210e-01 9.90555644e-01
5.23489833e-01 2.18401507e-01 -7.04810396e-02 3.60181749e-01
2.08307520e-01 -5.07694244e-01 3.15193862e-01 8.88901651e-01
-5.69907665e-01 -7.59026587e-01 -4.06397432e-01 6.01592541e-01
-4.14988637e-01 1.10814884e-01 -4.54739988e-01 6.06138647e-01
-6.66809082e-03 8.19936156e-01 5.95164299e-01 -2.11472943e-01
9.37743764e-03 1.10283561e-01 4.32070225e-01 -5.21001458e-01
-2.31699064e-01 1.19484983e-01 -4.79892045e-01 -3.29868108e-01
-4.86082435e-01 -6.85331225e-01 -1.36286151e+00 -3.24858338e-01
-3.82262357e-02 1.25429735e-01 3.99419576e-01 4.82922375e-01
4.07403976e-01 2.91996032e-01 7.82646477e-01 -7.41188586e-01
-5.52615345e-01 -9.43827987e-01 -7.45271742e-01 8.45152497e-01
5.65618098e-01 -8.31911027e-01 -8.23071659e-01 -1.41671836e-01]
|
[9.517868995666504, 2.9468367099761963]
|
9778ce4c-aea5-40c7-a912-4179205e557b
|
msdoctr-lite-a-lite-transformer-for-full-page
|
2303.13931
| null |
https://arxiv.org/abs/2303.13931v1
|
https://arxiv.org/pdf/2303.13931v1.pdf
|
MSdocTr-Lite: A Lite Transformer for Full Page Multi-script Handwriting Recognition
|
The Transformer has quickly become the dominant architecture for various pattern recognition tasks due to its capacity for long-range representation. However, transformers are data-hungry models and need large datasets for training. In Handwritten Text Recognition (HTR), collecting a massive amount of labeled data is a complicated and expensive task. In this paper, we propose a lite transformer architecture for full-page multi-script handwriting recognition. The proposed model comes with three advantages: First, to solve the common problem of data scarcity, we propose a lite transformer model that can be trained on a reasonable amount of data, which is the case of most HTR public datasets, without the need for external data. Second, it can learn the reading order at page-level thanks to a curriculum learning strategy, allowing it to avoid line segmentation errors, exploit a larger context and reduce the need for costly segmentation annotations. Third, it can be easily adapted to other scripts by applying a simple transfer-learning process using only page-level labeled images. Extensive experiments on different datasets with different scripts (French, English, Spanish, and Arabic) show the effectiveness of the proposed model.
|
['Sinda Ben Salem', 'Yousri Kessentini', 'Ahmed Cheikh Rouhou', 'Marwa Dhiaf']
|
2023-03-24
| null | null | null | null |
['handwriting-recognition']
|
['computer-vision']
|
[ 4.33554828e-01 -3.69944900e-01 -2.15820596e-01 -4.51966316e-01
-6.98315978e-01 -6.93156719e-01 3.20735514e-01 -2.61579216e-01
-4.25119430e-01 4.99929130e-01 -2.05136314e-01 -5.83552301e-01
6.04757331e-02 -7.51802683e-01 -7.60961950e-01 -5.87542832e-01
5.84770918e-01 6.82108283e-01 6.21410310e-01 -2.14917243e-01
2.69387603e-01 6.12003922e-01 -1.14616799e+00 4.45868909e-01
1.08133543e+00 1.09435403e+00 5.48207760e-01 3.96097988e-01
-2.41146088e-01 9.95563209e-01 -5.83858848e-01 -6.38008058e-01
1.05339304e-01 -3.89589280e-01 -7.88120449e-01 6.17415190e-01
4.11792189e-01 -4.46964949e-01 -4.17807281e-01 7.15575397e-01
4.04824972e-01 -9.04340670e-02 5.15533686e-01 -7.93630719e-01
-6.00951493e-01 4.92337704e-01 -8.34863782e-01 -1.15917817e-01
2.27695648e-02 -9.66136158e-02 9.88600314e-01 -9.26738262e-01
5.11901736e-01 9.52618718e-01 5.38951457e-01 4.83964115e-01
-8.49723935e-01 -3.03565949e-01 1.56396359e-01 4.28736299e-01
-1.19418502e+00 -9.44362655e-02 8.06856751e-01 -2.48535872e-01
7.45018125e-01 2.87629604e-01 4.50519204e-01 1.15717280e+00
-2.21364468e-01 1.27442777e+00 1.35826111e+00 -5.29366612e-01
-4.30684835e-02 5.86028956e-02 2.37800494e-01 8.31059754e-01
-1.18360974e-01 -5.86691558e-01 -5.34239411e-01 3.05212468e-01
7.98360050e-01 1.76333934e-02 -4.00671721e-01 -2.51925528e-01
-1.00152314e+00 5.40561616e-01 2.44182169e-01 3.63271326e-01
6.71369443e-03 -2.95327902e-01 4.60686356e-01 1.78264678e-01
6.90532774e-02 1.88462093e-01 -3.24857086e-01 -3.46057177e-01
-8.79467547e-01 -1.44699812e-01 7.22733498e-01 1.04171538e+00
7.26157069e-01 -5.99519210e-03 -6.83640912e-02 1.31992507e+00
5.95671348e-02 5.20865440e-01 7.45405674e-01 -2.48470411e-01
1.00719213e+00 7.32942879e-01 -1.07789062e-01 -8.59356880e-01
-9.48783010e-02 -2.25079060e-01 -1.05911076e+00 -2.45267466e-01
6.48459971e-01 1.63856789e-01 -1.07226253e+00 1.08098269e+00
-7.39733055e-02 -2.57289648e-01 -4.94253114e-02 8.90702724e-01
3.72263581e-01 7.53753424e-01 -3.02432775e-01 -1.89062729e-02
1.29338527e+00 -1.25077760e+00 -6.11863613e-01 -4.75263536e-01
4.38784093e-01 -8.19603801e-01 1.49489474e+00 6.88511252e-01
-6.02842152e-01 -4.40589696e-01 -1.09352696e+00 -3.12165141e-01
-3.19243044e-01 6.19928598e-01 5.53557575e-01 6.05989516e-01
-6.77234590e-01 5.16255617e-01 -7.39029348e-01 -3.83543551e-01
5.32139659e-01 2.01279670e-01 -2.15153202e-01 -4.42373872e-01
-7.96367764e-01 6.73486173e-01 4.76751804e-01 5.10217726e-01
-5.75613022e-01 -2.48033538e-01 -6.24426246e-01 1.18290216e-01
6.18064106e-01 1.49386019e-01 1.13412082e+00 -1.04123449e+00
-1.84477735e+00 6.07671320e-01 1.11184148e-02 -6.69972301e-02
8.32964778e-01 -3.43774021e-01 -1.89767852e-01 2.35014603e-01
-2.65724838e-01 1.97532624e-01 9.10627961e-01 -9.55552995e-01
-3.49975556e-01 -3.87066960e-01 -2.26253688e-01 2.22816601e-01
-7.58592546e-01 -6.52008355e-02 -8.31109941e-01 -8.60062778e-01
1.34701595e-01 -8.91180992e-01 9.17432643e-03 -1.62392676e-01
-2.76467860e-01 -2.64979124e-01 1.05905390e+00 -9.22328711e-01
1.02744460e+00 -2.17356753e+00 1.76791400e-01 2.47583702e-01
-1.17958657e-01 5.69403887e-01 -1.75374404e-01 3.72052759e-01
3.77135426e-01 -6.36284277e-02 -3.90256584e-01 -3.59226507e-03
-9.28326771e-02 2.65608102e-01 -4.29383755e-01 7.70386830e-02
3.06051403e-01 8.55009317e-01 -5.46836913e-01 -5.65039814e-01
1.13572724e-01 2.48889610e-01 -1.55759409e-01 2.68541843e-01
-4.70303327e-01 2.49274194e-01 -4.65193659e-01 7.44417489e-01
4.66294110e-01 -5.01794457e-01 4.80205923e-01 -1.48116916e-01
-5.16652800e-02 1.02126032e-01 -1.12714505e+00 1.48324370e+00
-5.53565979e-01 6.31248534e-01 -3.17671359e-01 -1.00190842e+00
1.32235456e+00 1.23165667e-01 3.71706001e-02 -9.06825423e-01
9.34639946e-02 5.53028464e-01 -1.08884558e-01 -4.29659605e-01
5.06956220e-01 2.35587418e-01 -9.36149061e-02 6.41618073e-01
-7.51630962e-02 -4.80480604e-02 3.59462172e-01 -1.56089142e-01
8.57895434e-01 2.66993195e-01 -1.25038192e-01 1.00320272e-01
5.85991979e-01 -3.85270678e-02 6.62465096e-01 4.86280292e-01
9.61214900e-02 6.17768884e-01 4.47031587e-01 -5.62407196e-01
-1.15372014e+00 -6.12660050e-01 3.43066826e-02 1.01065087e+00
5.40002361e-02 -1.91198811e-01 -7.66518116e-01 -7.03386486e-01
-2.62225896e-01 2.07363233e-01 -1.71113700e-01 2.43945152e-01
-9.05415475e-01 -8.01796436e-01 6.34932697e-01 8.16324890e-01
1.03078651e+00 -9.74425375e-01 -5.47162175e-01 1.22406200e-01
-3.91170532e-01 -1.40994549e+00 -6.26953781e-01 1.05627559e-01
-9.84389842e-01 -9.79847074e-01 -1.08313012e+00 -1.02669072e+00
8.16783428e-01 1.80558100e-01 5.95400155e-01 -1.04803918e-02
-3.45758498e-01 6.12690486e-02 -5.43771386e-01 3.43776457e-02
-1.75968438e-01 3.85474503e-01 -2.93198228e-01 3.43237460e-01
2.71603823e-01 -2.23361805e-01 -1.98596433e-01 6.88467979e-01
-8.77781451e-01 2.48061374e-01 9.62741613e-01 1.08991396e+00
5.87797225e-01 1.00310266e-01 3.85081828e-01 -1.05005741e+00
5.04949689e-01 2.37192392e-01 -7.83333123e-01 7.89306641e-01
-3.74337673e-01 -3.49848494e-02 9.64117467e-01 -5.96096098e-01
-1.24312651e+00 1.07802428e-01 4.96583693e-02 -3.34623247e-01
4.07679640e-02 5.44036031e-01 -3.80787641e-01 -1.49911925e-01
2.29792506e-01 7.37010837e-01 -2.27253348e-01 -8.05225074e-01
1.47314698e-01 1.07853472e+00 4.35248494e-01 -7.01828957e-01
5.88828504e-01 6.12002388e-02 -1.76801324e-01 -1.02853918e+00
-7.71586597e-01 -2.17638344e-01 -9.65937257e-01 -6.12574928e-02
6.30190194e-01 -6.35795057e-01 -6.58696055e-01 1.15920651e+00
-8.76351535e-01 -7.48809338e-01 6.75548688e-02 2.33418137e-01
-3.84459525e-01 7.59499729e-01 -9.62961197e-01 -4.21628058e-01
-2.80821830e-01 -1.19818175e+00 8.20635021e-01 3.33883405e-01
2.30013460e-01 -8.14604938e-01 -3.18375170e-01 6.93921804e-01
2.77676731e-01 -1.57130778e-01 1.10534096e+00 -4.76757795e-01
-8.97334814e-01 -1.95312992e-01 -3.97326857e-01 4.99193668e-01
2.72537529e-01 2.79298685e-02 -8.44362020e-01 -3.70322019e-01
-1.45832092e-01 -8.12147081e-01 8.45532715e-01 -2.79197007e-01
1.57552946e+00 -2.82365173e-01 -5.82851060e-02 4.58251864e-01
1.35090077e+00 1.07850850e-01 7.11431623e-01 4.48660523e-01
9.99164462e-01 4.01194066e-01 7.33275473e-01 3.49503040e-01
3.59102100e-01 6.70533001e-01 -1.13361396e-01 -1.35284349e-01
-1.29468232e-01 -3.78160179e-01 3.20468605e-01 1.24470997e+00
-3.45120244e-02 -2.99476177e-01 -1.13660538e+00 4.01150882e-01
-1.78520143e+00 -5.84633052e-01 5.77766746e-02 2.06399608e+00
1.02360821e+00 1.06140807e-01 8.44649747e-02 3.29164177e-01
6.11114085e-01 1.14858106e-01 -6.12222791e-01 -3.50120485e-01
-2.40744054e-01 1.29980132e-01 5.52545607e-01 4.82289828e-02
-1.01738286e+00 1.13165200e+00 5.58831549e+00 1.11493838e+00
-1.42449832e+00 -1.63239405e-01 6.46460176e-01 3.71471494e-01
3.37463133e-02 -3.05611312e-01 -8.91930103e-01 3.80530655e-01
6.51234329e-01 3.00272226e-01 3.85401696e-01 7.82336771e-01
-1.02897026e-01 -4.11789007e-02 -9.90490854e-01 1.04822099e+00
2.38386348e-01 -1.09062600e+00 2.60906190e-01 3.36704440e-02
4.97681707e-01 -2.79177487e-01 4.34323661e-02 3.57563458e-02
5.45483232e-02 -9.04955685e-01 6.42615736e-01 1.64535865e-01
9.65147853e-01 -6.35002971e-01 7.05023229e-01 3.66499603e-01
-1.04315829e+00 -9.64355990e-02 -5.34139693e-01 2.94154793e-01
-1.20126069e-01 4.03950572e-01 -7.20138669e-01 5.33233881e-01
6.13136411e-01 7.48557210e-01 -8.70669544e-01 8.45449567e-01
-4.34260577e-01 6.29788339e-01 -1.39928833e-01 -3.18722457e-01
3.32483053e-01 -3.53361279e-01 -1.64430082e-01 1.14059067e+00
2.92709410e-01 -3.88078317e-02 2.75314748e-01 4.01200116e-01
-2.22760767e-01 3.56986672e-01 -3.62800807e-01 -2.99765706e-01
2.40454465e-01 1.27365017e+00 -9.36334550e-01 -1.90241456e-01
-5.31152070e-01 1.29078043e+00 5.80061972e-01 2.55987912e-01
-6.51807785e-01 -6.97062373e-01 -2.38932014e-01 -7.94679970e-02
7.20218420e-01 -3.59000504e-01 -2.09755778e-01 -1.22331429e+00
4.62850600e-01 -1.24211991e+00 2.17058688e-01 -5.03437817e-01
-1.12935770e+00 6.90447807e-01 -4.95734245e-01 -1.17862666e+00
3.92290577e-02 -1.11118567e+00 -3.06805998e-01 6.17723405e-01
-1.58128238e+00 -1.36466753e+00 -3.34657341e-01 7.33279765e-01
7.50601947e-01 -3.42711717e-01 6.86213970e-01 4.78928268e-01
-8.15226197e-01 8.63546908e-01 3.25115293e-01 6.91466630e-01
7.58348048e-01 -1.23002565e+00 1.96464375e-01 8.09012890e-01
2.62263566e-01 3.57544720e-01 1.97116323e-02 -3.89871299e-01
-1.76178443e+00 -1.04923201e+00 6.50214553e-01 -5.92161901e-02
6.40171289e-01 -6.67087257e-01 -1.06864476e+00 6.79274917e-01
1.10057943e-01 -1.64266959e-01 8.37563217e-01 1.18975028e-01
-6.16824508e-01 -4.00820881e-01 -8.38323474e-01 4.90211874e-01
6.60056591e-01 -5.55667043e-01 -4.71715748e-01 3.40907305e-01
2.20674202e-01 -5.58776617e-01 -9.41775918e-01 1.36769697e-01
4.64403212e-01 -6.50809109e-01 3.96176189e-01 -1.22061230e-01
5.20426214e-01 -1.95534855e-01 -1.93490498e-02 -1.02823842e+00
-1.04492627e-01 -4.48347718e-01 5.27936518e-02 1.42212045e+00
4.59606260e-01 -5.55899382e-01 7.30949521e-01 5.44879377e-01
-1.82347223e-02 -7.07613289e-01 -7.43253946e-01 -8.75677526e-01
-5.58951236e-02 -9.37222466e-02 6.28561676e-01 8.29386771e-01
-9.20605883e-02 5.12503982e-01 -6.12348795e-01 7.22033950e-03
4.82561022e-01 4.97308522e-01 6.26132548e-01 -1.00415492e+00
-4.92753029e-01 -2.12783471e-01 -1.70071036e-01 -1.44477570e+00
1.21944211e-02 -7.61426985e-01 1.92922994e-01 -1.39132321e+00
3.21728498e-01 -6.97099447e-01 1.07377311e-02 8.78870666e-01
-2.58765463e-02 2.77091891e-01 3.88448983e-01 4.46354002e-01
-6.40418231e-01 5.55308878e-01 1.43328261e+00 -3.21758240e-01
7.13569224e-02 -5.81489056e-02 -2.53411502e-01 5.60473859e-01
7.33136952e-01 -1.26556426e-01 -3.65005285e-01 -8.53153586e-01
3.63666900e-02 1.88414361e-02 -2.01647207e-02 -8.22659314e-01
2.38799065e-01 -9.79807973e-02 4.98769134e-01 -6.77549422e-01
1.65557891e-01 -7.68019557e-01 -3.07471573e-01 2.71961868e-01
-2.38611266e-01 -3.71614769e-02 2.22902477e-01 4.30184484e-01
-3.28759760e-01 -4.98992801e-01 8.36098373e-01 -1.87205479e-01
-8.92167687e-01 2.72727460e-01 -4.00293827e-01 2.51661241e-02
9.33042526e-01 -2.48402268e-01 -3.41647148e-01 1.02836035e-01
-2.69504458e-01 1.71385556e-01 5.12673318e-01 5.85722744e-01
6.78602576e-01 -1.02526724e+00 -4.46385235e-01 2.29438946e-01
1.47424132e-01 1.19014740e-01 9.18169916e-02 4.89015490e-01
-7.93303728e-01 5.13503373e-01 -2.87099123e-01 -6.68396473e-01
-1.20180535e+00 3.75323504e-01 1.19906433e-01 -5.20519197e-01
-7.30588555e-01 5.19503653e-01 -9.74434689e-02 -5.35294890e-01
2.73539752e-01 -1.47554189e-01 -4.30163406e-02 -2.16359030e-02
5.43151200e-01 2.83599317e-01 2.53192186e-01 -3.49986076e-01
-1.23427667e-01 8.26703429e-01 -5.16410291e-01 2.90163849e-02
1.35110354e+00 9.25914422e-02 -2.36679018e-01 4.58780915e-01
8.14665973e-01 -2.40762010e-01 -1.27940595e+00 -2.75605232e-01
2.55444825e-01 -5.61304271e-01 -2.11153880e-01 -8.46816599e-01
-1.17183352e+00 1.25931621e+00 4.26141649e-01 -1.77960709e-01
1.16567099e+00 -3.04495245e-01 8.98209810e-01 7.98906088e-01
5.28031051e-01 -1.34149325e+00 3.54396075e-01 6.02779865e-01
7.72908986e-01 -1.09549940e+00 -2.45091274e-01 -4.13085371e-01
-8.33797991e-01 1.44118130e+00 7.90492117e-01 2.73678392e-01
2.00540662e-01 2.35230029e-01 2.08757028e-01 1.70571059e-01
-4.52920705e-01 1.29008085e-01 4.53928858e-02 4.47713047e-01
4.70742702e-01 -2.69856527e-02 -2.07738534e-01 4.79198813e-01
5.97314760e-02 1.13288566e-01 7.14646518e-01 9.84693527e-01
-2.84303546e-01 -1.42561042e+00 -1.81184202e-01 4.67159182e-01
-3.38498622e-01 -6.84294943e-03 -4.86159146e-01 6.16487861e-01
-3.86921495e-01 6.73235655e-01 -7.26431608e-02 -1.35497823e-01
3.88365269e-01 7.18534067e-02 6.76206589e-01 -5.01425922e-01
-3.44343156e-01 1.35213032e-01 -1.95255354e-01 -2.09741354e-01
-2.37961888e-01 -5.44875622e-01 -9.65753794e-01 -1.35597959e-01
-3.41611952e-01 1.62046694e-03 4.61770862e-01 9.49473560e-01
1.99840322e-01 4.77589726e-01 6.66125178e-01 -3.84961516e-01
-8.37897778e-01 -9.55906391e-01 -6.74203575e-01 2.80316353e-01
-7.97981769e-02 -3.24283183e-01 1.84548944e-01 2.00089216e-01]
|
[11.859535217285156, 2.3261313438415527]
|
50abc5cd-cc96-4539-a18b-f4c097abf446
|
re-matching-a-fine-grained-semantic-matching
|
2306.04954
| null |
https://arxiv.org/abs/2306.04954v1
|
https://arxiv.org/pdf/2306.04954v1.pdf
|
RE-Matching: A Fine-Grained Semantic Matching Method for Zero-Shot Relation Extraction
|
Semantic matching is a mainstream paradigm of zero-shot relation extraction, which matches a given input with a corresponding label description. The entities in the input should exactly match their hypernyms in the description, while the irrelevant contexts should be ignored when matching. However, general matching methods lack explicit modeling of the above matching pattern. In this work, we propose a fine-grained semantic matching method tailored for zero-shot relation extraction. Following the above matching pattern, we decompose the sentence-level similarity score into entity and context matching scores. Due to the lack of explicit annotations of the redundant components, we design a feature distillation module to adaptively identify the relation-irrelevant features and reduce their negative impact on context matching. Experimental results show that our method achieves higher matching $F_1$ score and has an inference speed 10 times faster, when compared with the state-of-the-art methods.
|
['Mingming Sun', 'Minlong Peng', 'Junzhe Wang', 'Zhongyu Wei', 'Tao Gui', 'Qi Zhang', 'Xin Zhao', 'WenYu Zhan', 'Jun Zhao']
|
2023-06-08
| null | null | null | null |
['relation-extraction']
|
['natural-language-processing']
|
[ 3.39026183e-01 2.42442936e-01 -4.93644685e-01 -3.84687752e-01
-4.93649811e-01 -2.75911480e-01 4.00215447e-01 6.61194324e-01
-4.00222182e-01 5.30219913e-01 3.25856358e-01 -9.55019444e-02
-2.14984119e-01 -1.21022880e+00 -2.23368332e-01 -3.14313531e-01
2.25393862e-01 3.72467965e-01 5.35697699e-01 -3.57219100e-01
1.22365333e-01 -3.09821405e-02 -1.95189083e+00 3.96185189e-01
1.00717020e+00 1.01478231e+00 2.70736396e-01 8.32731649e-02
-7.34620869e-01 7.11708009e-01 -3.08755666e-01 -7.42718697e-01
-4.46975231e-02 -4.67895508e-01 -1.07372260e+00 -4.36780363e-01
2.98339605e-01 -1.19878734e-02 -5.95492721e-01 1.38214350e+00
4.52903211e-01 4.18620974e-01 3.69578719e-01 -1.25835598e+00
-5.28067291e-01 7.16780186e-01 -1.72621980e-01 2.77603775e-01
8.32091272e-01 -2.43604735e-01 1.72977436e+00 -1.04420793e+00
8.73900056e-01 1.01843929e+00 4.07249749e-01 5.13414621e-01
-9.93954778e-01 -7.37660170e-01 1.39823034e-01 6.84595764e-01
-1.56047285e+00 -3.84897798e-01 5.41696548e-01 -1.34271219e-01
1.39014113e+00 3.65556151e-01 3.22236866e-01 6.72899008e-01
-3.14129025e-01 5.98564088e-01 4.98367101e-01 -7.48526573e-01
1.27674624e-01 -1.90668441e-02 6.02088153e-01 6.78512633e-01
4.04829085e-01 -1.65065154e-01 -5.93183160e-01 -3.57589424e-01
1.21564843e-01 1.18063815e-01 -1.00424543e-01 -1.01212196e-01
-9.70942914e-01 7.38212287e-01 2.68050522e-01 5.34237862e-01
-1.11572534e-01 -1.48454204e-01 5.82645655e-01 2.26337656e-01
2.48457104e-01 5.59736133e-01 -3.83337677e-01 4.06150073e-02
-6.57805502e-01 3.56924206e-01 8.51728439e-01 1.54321039e+00
1.07912827e+00 -7.89778054e-01 -7.94853687e-01 9.37252164e-01
-7.01429248e-02 1.90234914e-01 4.35800612e-01 -6.81578338e-01
7.26700068e-01 1.14785588e+00 -9.66458488e-03 -1.15837252e+00
-3.59163880e-01 -6.36321725e-03 -5.98135352e-01 -5.39349198e-01
5.01887091e-02 2.27772653e-01 -6.71547234e-01 1.68509400e+00
5.81081867e-01 3.25287730e-01 1.64073676e-01 8.58413637e-01
1.38215292e+00 3.91384780e-01 3.23629856e-01 -3.22960287e-01
1.89867127e+00 -8.20664287e-01 -1.12320948e+00 -2.80437380e-01
8.47942472e-01 -7.04013526e-01 1.13743234e+00 -4.42368776e-01
-7.14319468e-01 -3.13071758e-01 -1.14437878e+00 -3.37279588e-01
-5.55330455e-01 -2.16154546e-01 8.00693512e-01 3.44049811e-01
-1.18975662e-01 9.63776410e-01 -3.29698384e-01 -4.91663188e-01
1.54228255e-01 1.67361900e-01 -3.97261500e-01 -2.22796857e-01
-1.93667722e+00 1.02742612e+00 6.65593624e-01 -3.57282490e-01
6.79332316e-02 -7.58117437e-01 -1.22837186e+00 4.02010649e-01
8.85048926e-01 -7.52235949e-01 1.21312881e+00 -3.06680113e-01
-8.77001166e-01 9.38220263e-01 -7.48935401e-01 -1.48243397e-01
-1.63788363e-01 -6.57053888e-02 -8.59888792e-01 2.34700367e-02
3.89685303e-01 1.47613227e-01 2.55761832e-01 -7.48686910e-01
-9.37005043e-01 -2.48222157e-01 4.22304153e-01 2.23056063e-01
-4.99171436e-01 3.56602371e-01 -7.15482295e-01 -4.89791453e-01
3.58654261e-01 -5.16724467e-01 -2.54314631e-01 -3.33170146e-01
-4.10272896e-01 -6.03185534e-01 5.03777862e-01 -2.65817761e-01
1.88712299e+00 -1.99721968e+00 -1.48894742e-01 2.26376995e-01
3.02389205e-01 2.96907812e-01 -1.30869970e-01 7.46775806e-01
1.03651676e-02 1.08177729e-01 -1.97970286e-01 -7.97477141e-02
1.70035586e-01 2.59939998e-01 -2.03105032e-01 -1.06515519e-01
2.04259127e-01 9.91997540e-01 -1.36591458e+00 -8.79318297e-01
-2.40236032e-03 -1.68368295e-02 -3.92398775e-01 3.92832279e-01
-1.51276141e-01 -2.52178699e-01 -5.00902951e-01 6.61166310e-01
5.29307485e-01 -3.15156519e-01 5.30736804e-01 -5.60458064e-01
1.71806216e-01 7.98912406e-01 -1.28515089e+00 1.76719236e+00
-3.65399748e-01 1.56251922e-01 -4.14127529e-01 -7.52358496e-01
9.45229888e-01 4.48112279e-01 3.92436713e-01 -7.62188792e-01
1.21570714e-01 2.21121296e-01 -2.48875506e-02 -7.11990952e-01
6.90973699e-01 -2.89782792e-01 -2.65991807e-01 2.76552498e-01
2.60798186e-01 2.02443928e-01 5.16581595e-01 2.84591228e-01
1.39477789e+00 -1.19177938e-01 8.29204977e-01 -8.37947726e-02
6.45914316e-01 9.64013189e-02 1.06261027e+00 5.85482717e-01
-1.08547941e-01 3.03856105e-01 3.34685802e-01 -4.28110927e-01
-7.69323945e-01 -6.40120327e-01 -4.29792739e-02 1.10041690e+00
6.62702858e-01 -1.14100564e+00 -6.78448915e-01 -8.26490402e-01
1.34251997e-01 6.92965567e-01 -5.06088197e-01 -3.84924144e-01
-5.24071932e-01 -4.80934948e-01 6.60621285e-01 4.34075952e-01
3.21074009e-01 -1.09292054e+00 -4.68108058e-01 3.46912771e-01
-5.73644817e-01 -1.35295463e+00 -4.81015235e-01 1.88058108e-01
-5.16515493e-01 -1.26976931e+00 -6.87679872e-02 -8.91218424e-01
5.16731441e-01 4.00088280e-01 1.22331917e+00 2.48516440e-01
-3.22148263e-01 -3.44790369e-01 -6.55179381e-01 -5.92713393e-02
-1.24139920e-01 1.72197431e-01 -9.80812833e-02 -2.77058035e-01
1.23564351e+00 -5.91658652e-01 -4.07732159e-01 1.45386070e-01
-7.86205530e-01 5.41662574e-02 3.18414629e-01 9.39073503e-01
6.77196741e-01 1.02812136e-02 5.49751341e-01 -1.26495898e+00
6.08108878e-01 -4.69784170e-01 -2.96368718e-01 6.41113937e-01
-9.69923317e-01 3.68754447e-01 6.51421607e-01 -4.08042461e-01
-9.68543053e-01 1.56823248e-02 -4.18245094e-03 -3.46256167e-01
-8.57983381e-02 7.04771698e-01 -6.79296732e-01 2.28630573e-01
4.49112952e-01 -1.38983458e-01 -5.18807888e-01 -5.51867843e-01
5.36099017e-01 6.09747410e-01 6.26079202e-01 -6.25761747e-01
8.29437613e-01 1.57984421e-01 5.57102822e-02 -2.21998990e-01
-1.31099820e+00 -9.09932315e-01 -6.54124916e-01 2.41417259e-01
7.01094508e-01 -7.20591068e-01 -7.16764927e-01 -2.78028846e-01
-1.28065729e+00 5.25820196e-01 -4.50132608e-01 2.83732563e-01
-2.76983619e-01 5.35063505e-01 -5.82786500e-01 -6.03137672e-01
-6.36797607e-01 -7.11392760e-01 8.37017894e-01 3.67044181e-01
-5.35550833e-01 -5.68312526e-01 2.00338066e-01 2.31275916e-01
1.28351197e-01 -1.82746090e-02 1.28142524e+00 -9.89696562e-01
-3.58187616e-01 -3.24139327e-01 -5.03239095e-01 -2.91032732e-01
3.11331123e-01 -3.48924667e-01 -1.00088918e+00 1.66402221e-01
-3.02105248e-01 1.46717086e-01 7.96381950e-01 -3.46668899e-01
8.86162758e-01 -3.94894749e-01 -6.15471661e-01 4.57538635e-01
1.35740924e+00 1.21026129e-01 5.67465067e-01 2.59815961e-01
7.07435489e-01 7.19267607e-01 1.06266809e+00 4.70777243e-01
4.36272204e-01 8.77185166e-01 1.41262740e-01 1.71278194e-01
-1.20389476e-01 -7.02637076e-01 -1.44552618e-01 7.41471231e-01
3.19636315e-02 -7.07765110e-03 -6.10917091e-01 6.46220744e-01
-2.19388366e+00 -1.28700280e+00 -1.52309045e-01 2.24568081e+00
1.21487427e+00 1.21619008e-01 -1.80465952e-01 2.14139551e-01
9.41215217e-01 6.48417026e-02 -2.99190283e-01 -2.57190704e-01
-9.88146961e-02 4.77774352e-01 3.85515273e-01 4.19784367e-01
-1.12555707e+00 1.27823400e+00 5.80470085e+00 1.15851855e+00
-3.36173862e-01 2.31659934e-01 -1.38453439e-01 6.34992272e-02
-6.27714992e-01 5.25476098e-01 -1.09413910e+00 4.06663269e-01
7.56649673e-01 -6.09357595e-01 2.57529944e-01 7.47984111e-01
-2.29223654e-01 1.33975014e-01 -1.33531845e+00 8.57960999e-01
-5.10720573e-02 -1.35685134e+00 8.20971280e-02 -2.15816602e-01
3.83289278e-01 -4.46342915e-01 -6.89687371e-01 5.27216673e-01
1.08502351e-01 -8.96130562e-01 3.73842835e-01 3.75477016e-01
8.46694350e-01 -8.67303252e-01 8.91411185e-01 2.90987611e-01
-1.72943449e+00 -4.46381792e-02 -5.40962934e-01 -2.22932249e-01
1.89071149e-01 7.62243390e-01 -5.49156129e-01 9.63875294e-01
4.50144082e-01 6.20276093e-01 -2.42717788e-01 7.14428782e-01
-5.26088655e-01 2.02706650e-01 -6.68741986e-02 -2.47913674e-01
-1.01321466e-01 -5.89115061e-02 5.10896921e-01 1.46162391e+00
2.02232063e-01 5.39461553e-01 2.35195771e-01 9.23089623e-01
-2.00245962e-01 4.80434984e-01 -5.80573440e-01 8.09248611e-02
1.10991776e+00 1.34307063e+00 -3.94845605e-01 -6.76581025e-01
-7.65662670e-01 1.01947987e+00 6.02592885e-01 3.66801433e-02
-6.85651958e-01 -1.01562333e+00 1.02649295e+00 -1.31749496e-01
3.36146832e-01 4.13070351e-01 -2.85953045e-01 -1.35429394e+00
2.73640811e-01 -4.96803105e-01 8.17892373e-01 -3.59296113e-01
-1.48444009e+00 4.65324432e-01 6.85496852e-02 -1.40550375e+00
-2.11985797e-01 -1.94383159e-01 -6.58061266e-01 9.26192462e-01
-1.45843256e+00 -9.90788519e-01 -2.91583687e-01 3.43513399e-01
2.33406961e-01 8.78344551e-02 1.29466736e+00 6.14165366e-01
-6.04772747e-01 9.39132929e-01 -4.80548263e-01 2.96953112e-01
6.36209607e-01 -9.90319014e-01 5.65279841e-01 8.82902622e-01
1.72822937e-01 1.03881431e+00 6.71706140e-01 -7.98250377e-01
-1.21991730e+00 -9.86362040e-01 1.85838521e+00 -1.92337647e-01
6.90884471e-01 -1.87038571e-01 -1.11521673e+00 2.82799244e-01
-6.70297891e-02 1.83730200e-01 9.42632616e-01 4.56751913e-01
-9.40717876e-01 -6.42782301e-02 -1.16815901e+00 7.05268323e-01
1.55697310e+00 -8.46494257e-01 -1.10419476e+00 1.26176879e-01
1.19347501e+00 -1.72764122e-01 -9.91889179e-01 7.38709688e-01
5.54508328e-01 -4.92191762e-01 9.39813673e-01 -8.59990954e-01
3.12401682e-01 -5.20440876e-01 -3.80156130e-01 -8.38444293e-01
-6.18780077e-01 -5.18740118e-01 -6.56644225e-01 1.54773045e+00
5.61353266e-01 -2.11827472e-01 5.89375317e-01 8.81253541e-01
3.34005281e-02 -6.93102956e-01 -9.94346559e-01 -9.87331927e-01
-4.23391700e-01 -2.31289178e-01 1.03332925e+00 1.23068225e+00
8.63537312e-01 8.10784698e-01 -2.01411590e-01 7.83132240e-02
4.98409003e-01 7.02620625e-01 3.38066339e-01 -1.38281083e+00
-2.65070230e-01 -3.69279712e-01 -4.55869794e-01 -7.62253702e-01
3.71403933e-01 -1.21392322e+00 3.10565457e-02 -1.53382587e+00
6.61880910e-01 -3.70495915e-01 -5.44716954e-01 7.63508439e-01
-7.15915382e-01 -1.59941584e-01 3.28738876e-02 1.58932239e-01
-6.91834390e-01 4.49663013e-01 9.66442168e-01 -1.09386362e-01
-6.55033886e-02 -2.69252986e-01 -7.69574583e-01 4.88410205e-01
5.08120537e-01 -6.60678208e-01 -3.50418091e-01 -6.42127693e-02
1.88293874e-01 -1.39870301e-01 -1.16534874e-01 -6.92088842e-01
4.97665823e-01 -5.14246941e-01 -2.16048896e-01 -4.23907429e-01
2.16982812e-01 -8.70385110e-01 1.27217025e-01 1.40952736e-01
-4.89787877e-01 -2.27585062e-01 -2.06376627e-01 4.84680772e-01
-3.66380215e-01 -5.62877536e-01 4.25861448e-01 -1.04168877e-01
-1.00635362e+00 3.86412084e-01 1.07613780e-01 3.55987251e-01
7.98686683e-01 -1.04346136e-02 -4.14868951e-01 2.98365895e-02
-4.63939160e-01 2.40383923e-01 4.48926419e-01 5.99420846e-01
6.00945055e-01 -1.60621488e+00 -4.06934649e-01 1.12180682e-02
8.79444003e-01 -1.85125753e-01 6.49975091e-02 4.46148783e-01
1.60272464e-01 3.31747621e-01 2.07520828e-01 1.70013621e-01
-1.45239854e+00 9.47976351e-01 7.22316355e-02 -5.19074857e-01
-7.73410320e-01 6.03923380e-01 -7.82856569e-02 -4.14877981e-01
3.09097916e-01 -1.93703845e-01 -4.60093021e-01 1.61607966e-01
8.23878944e-01 1.61112696e-01 2.65570670e-01 -5.55795729e-01
-6.84146166e-01 4.20024842e-01 -2.01138258e-02 1.32910594e-01
9.72036421e-01 -4.30468991e-02 -3.59982818e-01 2.94108629e-01
1.13649774e+00 -5.52030578e-02 -3.93115848e-01 -7.58141279e-01
6.53638840e-01 -6.60759628e-01 -2.40984574e-01 -4.14080173e-01
-7.70852089e-01 4.96507525e-01 6.82272166e-02 1.19258881e-01
9.12716329e-01 2.41296574e-01 1.24117708e+00 5.49941182e-01
3.95167381e-01 -1.13301766e+00 -4.27092165e-01 7.57071435e-01
2.07971871e-01 -1.07217574e+00 1.16379723e-01 -1.08929181e+00
-4.83605027e-01 8.47123265e-01 7.56389022e-01 1.20774500e-01
4.47657108e-01 2.55535871e-01 -2.40688473e-01 -3.87861073e-01
-9.34242427e-01 -8.47617626e-01 5.96999347e-01 4.30209249e-01
6.23381019e-01 4.22856361e-02 -8.33629310e-01 1.17628765e+00
-2.33072732e-02 -1.95975736e-01 -1.09040618e-01 1.01844001e+00
-7.17127502e-01 -1.43479443e+00 2.40611717e-01 4.23064679e-01
-3.72116327e-01 -5.76871514e-01 -3.58447880e-01 3.77149224e-01
3.16189915e-01 1.12633133e+00 -1.25863463e-01 -7.11501598e-01
6.81831419e-01 3.93235952e-01 2.90127009e-01 -9.52850521e-01
-7.52436578e-01 -4.39759731e-01 5.58730960e-01 -6.66049778e-01
-3.32124442e-01 -3.73842508e-01 -1.66653097e+00 -2.60278702e-01
-6.39445782e-01 4.20540303e-01 2.16750726e-01 1.21498668e+00
5.14184594e-01 3.29707235e-01 4.99744505e-01 -6.56449199e-02
-4.41706002e-01 -9.13218260e-01 -3.97622794e-01 9.60268259e-01
-1.29214868e-01 -8.35775912e-01 -1.77695274e-01 -1.88071102e-01]
|
[9.35133171081543, 8.562419891357422]
|
12409ccb-35b8-404a-932e-30b7cb491baf
|
towards-head-motion-compensation-using-multi
|
1807.03651
| null |
http://arxiv.org/abs/1807.03651v1
|
http://arxiv.org/pdf/1807.03651v1.pdf
|
Towards Head Motion Compensation Using Multi-Scale Convolutional Neural Networks
|
Head pose estimation and tracking is useful in variety of medical
applications. With the advent of RGBD cameras like Kinect, it has become
feasible to do markerless tracking by estimating the head pose directly from
the point clouds. One specific medical application is robot assisted
transcranial magnetic stimulation (TMS) where any patient motion is compensated
with the help of a robot. For increased patient comfort, it is important to
track the head without markers. In this regard, we address the head pose
estimation problem using two different approaches. In the first approach, we
build upon the more traditional approach of model based head tracking, where a
head model is morphed according to the particular head to be tracked and the
morphed model is used to track the head in the point cloud streams. In the
second approach, we propose a new multi-scale convolutional neural network
architecture for more accurate pose regression. Additionally, we outline a
systematic data set acquisition strategy using a head phantom mounted on the
robot and ground-truth labels generated using a highly accurate tracking
system.
|
['Alexander Schlaefer', 'Lars Matthäus', 'Omer Rajput', 'Nils Gessert', 'Martin Gromniak']
|
2018-07-10
| null | null | null | null |
['head-pose-estimation']
|
['computer-vision']
|
[-8.60052779e-02 3.31277221e-01 1.30870938e-01 -4.08854336e-01
-6.81940496e-01 -7.87819698e-02 2.02988163e-01 -8.77395943e-02
-7.33148575e-01 5.34248710e-01 1.06303461e-01 1.02376893e-01
1.33164614e-01 -2.49516234e-01 -5.59854269e-01 -7.09498405e-01
1.11770406e-01 8.32457960e-01 1.92999750e-01 -2.90903360e-01
1.40588358e-01 8.60712647e-01 -1.34845531e+00 -4.94568288e-01
5.14321804e-01 9.30464208e-01 4.39248085e-01 3.83124858e-01
1.55347005e-01 3.35276246e-01 -3.78432423e-01 5.57633080e-02
1.30834594e-01 -2.19603091e-01 -5.97285151e-01 2.19116867e-01
1.58821270e-01 -4.10395920e-01 2.89533008e-03 9.29434359e-01
9.61991727e-01 -4.21041921e-02 3.41541737e-01 -1.25210309e+00
2.85962433e-01 1.93595141e-01 -7.12797284e-01 -3.41355979e-01
6.95934832e-01 -1.95161849e-01 1.31184191e-01 -6.91047251e-01
6.82176471e-01 8.89495254e-01 7.68189967e-01 9.17362630e-01
-6.89890623e-01 -6.18920445e-01 -8.32295790e-02 1.44414932e-01
-1.44803476e+00 -5.29622018e-01 8.42288077e-01 -4.91285324e-01
4.97399688e-01 1.44266963e-01 8.22998881e-01 7.01603234e-01
2.83508003e-01 6.40491426e-01 9.69125450e-01 -6.55823946e-01
4.98049587e-01 2.38308311e-02 -1.34632387e-03 5.00501394e-01
2.53694266e-01 -7.46822357e-02 -4.66305673e-01 -1.27273127e-01
9.12727356e-01 8.02887157e-02 -6.35819197e-01 -9.03275788e-01
-1.14725852e+00 5.57981730e-01 7.89462090e-01 6.07951820e-01
-7.14064062e-01 1.67213738e-01 2.18008146e-01 -4.85748947e-01
2.66682655e-01 1.27083570e-01 -3.64514142e-01 -2.18149111e-01
-1.14479268e+00 2.75408309e-02 6.92473948e-01 9.20811772e-01
2.79028058e-01 -1.72138110e-01 1.01536021e-01 2.84829080e-01
7.54304290e-01 2.38993868e-01 8.86653185e-01 -7.10686982e-01
1.95805445e-01 5.16522408e-01 3.52589965e-01 -7.10650444e-01
-9.66820955e-01 -1.79040030e-01 -5.41274965e-01 3.81654590e-01
4.08375919e-01 -1.99774966e-01 -1.15179741e+00 1.42704833e+00
8.40570331e-01 2.13904083e-01 -4.05071855e-01 1.27246869e+00
6.18351102e-01 -1.46902427e-01 -2.01241411e-02 -2.14400515e-01
1.52227330e+00 -5.51724315e-01 -8.78122270e-01 -2.37959012e-01
8.15059364e-01 -4.48418975e-01 6.15570784e-01 4.53089207e-01
-8.68419588e-01 -1.11599728e-01 -9.53030884e-01 1.22037567e-01
-2.71279782e-01 1.14995152e-01 5.14213324e-01 7.42157340e-01
-1.22307348e+00 4.32640523e-01 -1.43270123e+00 -6.17455542e-01
2.03439161e-01 1.02271664e+00 -7.27627456e-01 9.94462669e-02
-7.90021181e-01 1.42393470e+00 2.36572370e-01 5.49812317e-01
-2.20701396e-01 -2.08263174e-01 -9.41015482e-01 -4.36642140e-01
-1.79415450e-01 -6.72850966e-01 1.35031044e+00 -6.13488317e-01
-1.99757516e+00 1.02567232e+00 -1.53298721e-01 -2.33692974e-01
5.65532804e-01 -4.17879879e-01 -5.44117652e-02 -3.54667986e-03
-8.93294886e-02 6.63471222e-01 9.18254733e-01 -1.16564453e+00
-2.67090499e-01 -9.74574089e-01 -3.79389822e-01 3.14853460e-01
-1.17515273e-01 2.74572819e-01 -6.12523258e-01 -7.46165961e-02
6.12875700e-01 -1.31850874e+00 -2.75981724e-01 2.11358175e-01
-3.79137665e-01 7.44800493e-02 7.31421888e-01 -8.65239382e-01
5.94420195e-01 -1.85699821e+00 2.82989621e-01 8.03846568e-02
3.67826641e-01 1.32516563e-01 4.29069012e-01 -2.38557518e-01
-2.33012572e-01 -7.09698737e-01 -1.72716379e-01 -7.09426761e-01
-2.30837971e-01 -1.51106697e-02 2.39181072e-01 9.67423737e-01
-4.59918648e-01 7.64380038e-01 -7.30294883e-01 -5.51635325e-01
4.53830063e-01 9.07785833e-01 -2.66678154e-01 2.37593651e-01
2.38582060e-01 1.04379225e+00 -4.22435313e-01 6.85492396e-01
7.39552855e-01 1.72641993e-01 5.56157203e-03 -2.69288749e-01
-1.38431802e-01 6.71641380e-02 -1.14020169e+00 1.97644544e+00
-4.61796969e-01 2.82905698e-01 6.01481378e-01 -5.68529248e-01
6.97779775e-01 7.20071554e-01 8.91130328e-01 -4.67711002e-01
6.14583015e-01 3.94609153e-01 -2.16981575e-01 -4.69257981e-01
4.58748966e-01 -3.42483789e-01 9.88915339e-02 3.62674057e-01
-1.99724495e-01 -1.68929592e-01 -6.93717957e-01 -2.33296320e-01
7.28234947e-01 5.46357036e-01 4.33689624e-01 7.31924623e-02
3.92578274e-01 5.08101359e-02 3.17176312e-01 1.32452354e-01
-4.14329410e-01 9.15875673e-01 -2.45331511e-01 -3.14700395e-01
-6.46879613e-01 -5.15368879e-01 -2.48200521e-01 6.33352578e-01
1.03857173e-02 1.03932194e-01 -1.12127757e+00 -2.73674637e-01
-5.45859411e-02 3.87433946e-01 -4.72541451e-01 -9.30435117e-03
-8.82068336e-01 -5.60151041e-01 1.81205124e-01 6.05232477e-01
2.97236830e-01 -9.49653506e-01 -1.16014588e+00 2.68148810e-01
-1.93376631e-01 -9.47507203e-01 -3.00889283e-01 2.38253742e-01
-1.09759724e+00 -8.13745499e-01 -1.12140298e+00 -7.04675376e-01
7.52958715e-01 -7.39599019e-02 5.65344453e-01 -2.23253638e-01
-1.46406457e-01 5.38386881e-01 -2.97652125e-01 -5.45956254e-01
-6.55368343e-02 1.32879853e-01 4.28981930e-01 -1.21259674e-01
2.69969970e-01 -5.09094000e-01 -6.49360240e-01 1.50907919e-01
-4.09777492e-01 -5.83440885e-02 4.04872626e-01 3.05957884e-01
4.56460178e-01 -4.91854042e-01 1.57863602e-01 -4.26790386e-01
4.89029378e-01 -3.02020878e-01 -6.19869530e-01 7.65511841e-02
-2.84490764e-01 -1.21241771e-01 1.75779685e-01 -5.16033232e-01
-6.68405831e-01 9.43032503e-01 -3.75865638e-01 -5.69033086e-01
-3.23490798e-01 3.06142360e-01 -4.78645772e-01 -3.79149884e-01
5.02832174e-01 3.41956690e-02 2.05152243e-01 -4.99588907e-01
3.44463706e-01 1.00306904e+00 8.35326910e-01 -6.59742281e-02
5.09144783e-01 5.87481737e-01 1.10036902e-01 -7.28382826e-01
-2.74219334e-01 -5.91546893e-01 -1.32071865e+00 -4.07530785e-01
9.23611403e-01 -6.60396039e-01 -1.05347848e+00 5.99002957e-01
-1.43259704e+00 -3.23897004e-01 4.43726592e-02 9.02363539e-01
-7.66160965e-01 2.93151915e-01 -2.18760967e-01 -9.47045982e-01
-6.53919637e-01 -1.33230281e+00 1.52399683e+00 1.52314574e-01
-3.76771927e-01 -8.57234061e-01 1.26912460e-01 6.30452335e-02
3.34496051e-01 5.55870533e-01 1.88173875e-01 -3.63741726e-01
-2.47872964e-01 -9.11959887e-01 2.71783561e-01 -4.06180799e-01
2.96030164e-01 -5.51566482e-01 -1.11803389e+00 -4.49281365e-01
5.98261654e-01 5.59121594e-02 -8.99322107e-02 7.82689989e-01
6.51747763e-01 8.68604332e-02 -7.27849960e-01 7.40947008e-01
1.13775229e+00 2.26528436e-01 6.29503846e-01 7.86126673e-01
9.01361763e-01 5.82721293e-01 5.90707242e-01 3.37483466e-01
5.28106391e-01 1.19563746e+00 4.71776158e-01 9.54436436e-02
6.52367249e-02 2.91663129e-02 1.34113207e-01 8.31216693e-01
-1.02439255e-01 2.24861011e-01 -9.49439287e-01 2.80121326e-01
-1.73332858e+00 -2.33540118e-01 -1.54066086e-01 2.55365229e+00
5.80760956e-01 -2.12843522e-01 3.29193950e-01 3.28821182e-01
7.23269880e-01 -4.89534229e-01 -5.28374493e-01 -5.60669694e-03
5.44427216e-01 1.17670625e-01 5.54270804e-01 5.21447897e-01
-8.91186357e-01 7.78486967e-01 5.68672991e+00 1.49680376e-02
-1.71246660e+00 3.95281196e-01 -3.19664106e-02 1.16355089e-03
4.12844002e-01 -2.26373702e-01 -8.55490983e-01 4.56789136e-01
9.43817556e-01 2.13807106e-01 1.34032980e-01 8.88373733e-01
3.93468052e-01 -5.44098377e-01 -1.13692451e+00 1.27166629e+00
1.76637471e-01 -6.57543302e-01 -7.08990693e-01 2.15317473e-01
-2.42017638e-02 6.97060823e-02 -3.02566350e-01 -1.12049475e-01
-2.77809501e-01 -8.02804649e-01 8.53826761e-01 6.85899079e-01
8.26038480e-01 -3.97180736e-01 5.79676747e-01 6.58404589e-01
-1.13378298e+00 3.28927666e-01 -9.51002389e-02 4.08146605e-02
5.83577275e-01 2.85088748e-01 -1.18379772e+00 4.09048349e-01
6.26535296e-01 2.00553864e-01 -3.28746736e-01 1.43166614e+00
-1.85101673e-01 1.03404388e-01 -5.99567711e-01 1.65933773e-01
-2.87872016e-01 7.39891157e-02 4.92550999e-01 6.53428495e-01
3.57852578e-01 1.05195746e-01 -2.09295481e-01 4.69851971e-01
2.56064594e-01 -5.02264872e-02 -5.86813509e-01 7.23157823e-01
2.88685292e-01 1.44003761e+00 -8.70538056e-01 1.16447374e-01
-2.09626377e-01 1.16680717e+00 2.23384157e-01 -1.52935892e-01
-6.17382586e-01 -2.96227247e-01 1.44945472e-01 3.34290057e-01
-8.76290724e-02 -4.16759640e-01 -3.14528823e-01 -1.08000183e+00
1.07290596e-01 -4.77060586e-01 -5.68704270e-02 -1.02580094e+00
-5.28452516e-01 8.52976263e-01 4.12856005e-02 -1.41876960e+00
-6.80515528e-01 -5.69462359e-01 -3.99581611e-01 1.00864255e+00
-1.18616486e+00 -1.08097923e+00 -6.18099272e-01 6.73007369e-01
9.67060700e-02 3.82020175e-01 9.22414422e-01 2.26184189e-01
-5.19029975e-01 4.96241361e-01 -2.02254951e-01 6.14429526e-02
5.34774363e-01 -9.93140578e-01 2.52211362e-01 5.09527266e-01
-3.30608875e-01 6.81466103e-01 8.34333837e-01 -8.74372602e-01
-1.59591889e+00 -8.55443895e-01 7.87777543e-01 -7.35935807e-01
3.05521518e-01 -3.36539119e-01 -7.90375650e-01 8.62779737e-01
-2.25891024e-01 2.74696141e-01 1.63811371e-01 -4.44296151e-01
3.91205430e-01 8.27825591e-02 -1.50922894e+00 2.36163944e-01
6.15318060e-01 -4.26339626e-01 -5.32984555e-01 4.26085979e-01
2.94501752e-01 -1.07978523e+00 -7.43030608e-01 2.42866442e-01
6.82976604e-01 -4.58196759e-01 6.40749693e-01 -5.59671372e-02
-3.68076682e-01 -3.94005477e-01 8.95619020e-02 -1.29899108e+00
5.43315448e-02 -6.34413958e-01 -2.08094746e-01 8.49779487e-01
-2.00897396e-01 -6.54113531e-01 1.34485543e+00 1.25869572e+00
-1.09883294e-01 -5.14845192e-01 -1.31855023e+00 -5.62652469e-01
-2.13934094e-01 -3.24048936e-01 6.43222511e-01 6.27004862e-01
3.19683403e-01 1.95658132e-01 -1.99058279e-01 4.07410920e-01
6.19123399e-01 -2.93141335e-01 6.55615687e-01 -1.31594968e+00
2.16236696e-01 2.01559104e-02 -7.97906816e-01 -9.03796434e-01
9.79970619e-02 -6.35674775e-01 4.92982030e-01 -1.61570764e+00
-1.01089187e-01 -4.80609387e-01 1.34675726e-01 3.48647296e-01
1.71608746e-01 1.75216183e-01 8.49365070e-02 3.30106944e-01
-2.76287049e-01 4.66550142e-01 1.02818596e+00 3.69729191e-01
-4.26529825e-01 4.33998257e-01 -1.63036525e-01 9.50255752e-01
6.51030540e-01 -5.98454118e-01 -1.59762800e-02 -5.30763865e-01
-2.09205016e-01 4.54465449e-01 3.37265611e-01 -1.26092279e+00
7.99255490e-01 4.20909733e-01 4.34743553e-01 -6.92268133e-01
7.53184319e-01 -1.20651042e+00 2.50159204e-01 5.95629752e-01
2.49868870e-01 2.01009646e-01 1.06700942e-01 1.53065979e-01
7.05404282e-02 -2.97629476e-01 8.45814645e-01 -1.47528827e-01
-3.17409664e-01 2.55454063e-01 -2.65572667e-01 -6.24983847e-01
1.03154695e+00 -5.89027047e-01 2.77225941e-01 -3.35705459e-01
-9.79274094e-01 -1.39537156e-01 6.38925552e-01 2.82165408e-01
6.98382616e-01 -1.28362679e+00 -1.20201275e-01 1.95534006e-01
-4.84646633e-02 3.74226272e-01 -7.43752271e-02 1.33055496e+00
-5.52986503e-01 6.76802516e-01 -1.46737605e-01 -7.44832039e-01
-1.10480440e+00 3.41975182e-01 7.89439082e-01 2.25305483e-01
-6.50248468e-01 6.37123644e-01 -2.07984447e-01 -6.86156929e-01
4.21040893e-01 -4.63812202e-01 -3.93891484e-01 -1.36506051e-01
4.68260825e-01 2.68418759e-01 6.07894599e-01 -1.25469911e+00
-7.05252051e-01 7.60942161e-01 2.85609514e-01 -3.04177254e-01
1.46470404e+00 -2.13643566e-01 -1.72486484e-01 3.53578597e-01
1.02857471e+00 -1.02821298e-01 -1.10259795e+00 2.22530179e-02
2.55773187e-01 -1.99568570e-01 3.47214073e-01 -5.42403460e-01
-1.18272030e+00 7.92336166e-01 1.19640851e+00 -2.86291122e-01
1.02077484e+00 -8.95253047e-02 7.16543436e-01 4.49198894e-02
9.83724892e-01 -7.61250257e-01 -4.26111072e-01 1.89042106e-01
8.98905396e-01 -1.23174644e+00 8.14801380e-02 -2.68961847e-01
-5.37085474e-01 1.01340461e+00 5.15071750e-01 1.55665025e-01
6.73910439e-01 4.86749291e-01 1.90984890e-01 -4.41843033e-01
1.56463921e-01 -9.45006311e-02 2.03564003e-01 7.78530836e-01
5.55275559e-01 1.90328866e-01 -9.61438566e-02 4.89454389e-01
-3.52331012e-01 6.06178164e-01 3.20652008e-01 1.35265458e+00
-4.19800103e-01 -9.94063914e-01 -9.48886335e-01 2.01190725e-01
-2.12248102e-01 1.80730313e-01 -1.38486922e-01 7.88483083e-01
-2.52913442e-02 7.30868578e-01 -4.95086536e-02 -3.17693681e-01
5.20696402e-01 1.68078005e-01 7.74495244e-01 -6.80883825e-01
-4.73983765e-01 2.84025520e-01 -4.11717623e-01 -5.85054934e-01
-3.76662612e-01 -8.07797730e-01 -1.46941173e+00 1.40245095e-01
-7.90201485e-01 -1.85937844e-02 1.42427182e+00 9.89899218e-01
1.61964640e-01 2.10510433e-01 4.44226265e-01 -1.73684239e+00
-2.53261924e-01 -1.04991317e+00 -7.42978930e-01 -6.39825910e-02
4.01771754e-01 -9.52864408e-01 -6.83117658e-02 -1.20230749e-01]
|
[13.646628379821777, 0.22659942507743835]
|
79056510-8fc8-4b4a-9856-58eea9b7ceb6
|
securing-behavior-based-opinion-spam
|
1811.03739
| null |
http://arxiv.org/abs/1811.03739v1
|
http://arxiv.org/pdf/1811.03739v1.pdf
|
Securing Behavior-based Opinion Spam Detection
|
Reviews spams are prevalent in e-commerce to manipulate product ranking and
customers decisions maliciously. While spams generated based on simple spamming
strategy can be detected effectively, hardened spammers can evade regular
detectors via more advanced spamming strategies. Previous work gave more
attention to evasion against text and graph-based detectors, but evasions
against behavior-based detectors are largely ignored, leading to
vulnerabilities in spam detection systems. Since real evasion data are scarce,
we first propose EMERAL (Evasion via Maximum Entropy and Rating sAmpLing) to
generate evasive spams to certain existing detectors. EMERAL can simulate
spammers with different goals and levels of knowledge about the detectors,
targeting at different stages of the life cycle of target products. We show
that in the evasion-defense dynamic, only a few evasion types are meaningful to
the spammers, and any spammer will not be able to evade too many detection
signals at the same time. We reveal that some evasions are quite insidious and
can fail all detection signals. We then propose DETER (Defense via Evasion
generaTion using EmeRal), based on model re-training on diverse evasive samples
generated by EMERAL. Experiments confirm that DETER is more accurate in
detecting both suspicious time window and individual spamming reviews. In terms
of security, DETER is versatile enough to be vaccinated against diverse and
unexpected evasions, is agnostic about evasion strategy and can be released
without privacy concern.
|
['Philip S. Yu', 'Sihong Xie', 'Shuaijun Ge', 'Guixiang Ma']
|
2018-11-09
| null | null | null | null |
['spam-detection']
|
['natural-language-processing']
|
[ 2.81999528e-01 9.94329304e-02 -6.00607879e-02 -1.11087039e-01
3.72878951e-03 -1.10291159e+00 9.00956750e-01 4.66584004e-02
-7.12628588e-02 3.89299870e-01 -4.99917090e-01 -7.29767144e-01
3.00164986e-02 -1.17750227e+00 -5.03773689e-01 -4.20267224e-01
-3.17803733e-02 5.78846514e-01 4.98838812e-01 -8.52152407e-01
2.02397466e-01 4.24189150e-01 -1.11125708e+00 3.24674547e-01
9.81078923e-01 4.86005157e-01 -3.57024997e-01 7.87583590e-01
-2.80766897e-02 4.78819728e-01 -1.08263350e+00 -7.94862866e-01
7.48150229e-01 -4.83133346e-01 -1.61000833e-01 1.75846711e-01
6.67558834e-02 -3.55093867e-01 -1.89922810e-01 1.42736113e+00
1.15373760e-01 -3.81694853e-01 6.15808129e-01 -1.56576598e+00
-1.10070968e+00 5.22929609e-01 -6.96415246e-01 -1.03062391e-01
1.61111727e-01 6.97115481e-01 8.47420275e-01 -3.25726509e-01
4.20176446e-01 1.60471809e+00 3.83687884e-01 1.07758272e+00
-1.61254072e+00 -7.87485301e-01 3.24269414e-01 -4.38433826e-01
-6.43044710e-01 -1.07597224e-01 5.56346655e-01 -4.81137522e-02
4.31465656e-01 6.40890896e-01 5.51095486e-01 1.52263093e+00
4.50369149e-01 9.65018690e-01 1.28979254e+00 6.63461611e-02
6.36812091e-01 8.05410981e-01 5.74768364e-01 4.85445857e-01
9.20261025e-01 6.57125771e-01 -1.35229334e-01 -9.84459221e-01
3.55682284e-01 1.71863124e-01 6.05252460e-02 -4.72457975e-01
-2.75842935e-01 1.45794368e+00 2.77931005e-01 -1.05930060e-01
-2.76591390e-01 7.18762428e-02 5.39377093e-01 1.06697023e+00
3.02085698e-01 8.29618812e-01 -7.64317036e-01 3.11046273e-01
-3.51591669e-02 3.54612440e-01 1.28396344e+00 6.92516863e-01
5.02409458e-01 4.16215122e-01 -4.43875603e-02 5.13739824e-01
3.04130316e-01 1.01137233e+00 6.46737754e-01 -3.89220923e-01
-1.09316319e-01 9.74635720e-01 1.41882554e-01 -1.17540002e+00
7.55134076e-02 -4.33654994e-01 -4.05862868e-01 6.46196127e-01
3.73965651e-01 -3.03319484e-01 -9.33646321e-01 1.53015029e+00
2.12124050e-01 -2.67445773e-01 -2.17814326e-01 7.86806047e-01
1.36690095e-01 3.29690784e-01 2.40429312e-01 -1.74519002e-01
1.44441044e+00 -6.12379789e-01 -3.87585282e-01 -6.57800496e-01
6.78944647e-01 -4.29984957e-01 1.09474421e+00 5.75535178e-01
-8.06618214e-01 -4.11509573e-02 -1.16500890e+00 8.53280783e-01
-8.14164817e-01 -5.38081288e-01 9.34345067e-01 1.51582134e+00
-8.35340679e-01 4.39166456e-01 -4.98332560e-01 -2.11713165e-01
6.80350065e-01 6.28847778e-01 2.93655604e-01 1.25419110e-01
-1.53852797e+00 9.02343452e-01 -3.31869841e-01 -4.34300274e-01
-1.10388851e+00 -4.21724051e-01 -4.85334992e-01 -2.91123129e-02
5.00244498e-01 -7.35043764e-01 1.24325192e+00 -1.67287338e+00
-1.13232005e+00 6.12726271e-01 2.20439002e-01 -1.02102935e+00
7.79668689e-01 2.12590590e-01 -7.38252401e-01 -5.46299443e-02
1.51057482e-01 2.53983766e-01 1.53983700e+00 -1.31660891e+00
-3.62199873e-01 -6.88166022e-01 -5.43593653e-02 -1.51736826e-01
-5.21973848e-01 7.06997365e-02 4.49974626e-01 -4.57307428e-01
-3.88008535e-01 -7.86486924e-01 -5.74828088e-01 -3.16940904e-01
-4.44007874e-01 7.94066302e-03 1.51960421e+00 -8.76806751e-02
9.69004035e-01 -1.71548986e+00 -7.80065298e-01 5.84123254e-01
4.82898355e-01 9.24254835e-01 -5.12736380e-01 2.38054499e-01
1.40360236e-01 5.37911952e-01 5.83558790e-02 3.91473442e-01
9.14167985e-02 -4.34144437e-02 -6.42000079e-01 4.67418730e-01
1.93176568e-01 1.30251789e+00 -1.06703854e+00 2.55664676e-01
1.90195352e-01 1.73209101e-01 -2.44696051e-01 -2.44965866e-01
-4.73517358e-01 -2.80987561e-01 -8.38483751e-01 8.68074358e-01
8.22802603e-01 -2.31527761e-01 3.10594052e-01 5.67366183e-01
5.55875778e-01 1.92383379e-01 -6.27643347e-01 7.09566325e-02
-2.76274621e-01 9.39130783e-02 5.56299567e-01 -8.62445891e-01
9.33499753e-01 -1.21783704e-01 1.21249795e-01 -6.08552456e-01
4.94021803e-01 4.06816483e-01 1.98064074e-01 1.28374383e-01
3.24123979e-01 8.27162713e-02 -2.29231030e-01 9.32145596e-01
-3.91837209e-01 -8.26745927e-02 1.43210396e-01 6.13438547e-01
1.62693524e+00 -7.64610887e-01 3.02984506e-01 -3.23399544e-01
4.52809513e-01 4.03133184e-01 1.87928230e-01 1.48582602e+00
-6.93374574e-01 -2.62983769e-01 7.83936322e-01 -2.95786142e-01
-7.63575852e-01 -1.37223709e+00 1.27792999e-01 1.21744549e+00
4.44959819e-01 -2.10244387e-01 -7.72158206e-01 -1.60997295e+00
7.50197470e-01 9.18702006e-01 -5.88958204e-01 -7.56380200e-01
-2.61042506e-01 -1.22126114e+00 4.93859291e-01 -9.19508338e-02
3.64047855e-01 -9.25018728e-01 -1.51387855e-01 3.41029435e-01
6.72371745e-01 -4.09330457e-01 -5.98450124e-01 7.71375746e-02
-8.33033323e-01 -1.32243323e+00 -1.56375587e-01 -4.63374138e-01
9.90676999e-01 8.27578008e-01 1.06346440e+00 3.92217785e-01
-4.76393998e-01 4.76168662e-01 -2.42948249e-01 -8.66172254e-01
-1.30907142e+00 -1.31806016e-01 3.04158181e-01 5.99906780e-02
1.30802333e+00 -3.57632488e-01 -4.81725574e-01 9.38036919e-01
-8.58723164e-01 -7.48231828e-01 6.26827955e-01 9.07312930e-01
-3.66870254e-01 1.10206857e-01 1.35270667e+00 -1.55673063e+00
1.47147357e+00 -4.63881493e-01 -8.41025293e-01 1.19030431e-01
-1.30198431e+00 -2.44869709e-01 1.03378570e+00 -1.09349585e+00
-9.64380920e-01 1.20426178e-01 2.44179681e-01 4.43953648e-02
-1.32075533e-01 -3.72844875e-01 -5.08526415e-02 -3.33060503e-01
1.44183719e+00 3.44636470e-01 5.92328012e-01 -1.45515323e-01
6.17084742e-01 8.42620194e-01 -1.97788015e-01 -2.47176975e-01
1.23115754e+00 6.08818650e-01 -4.01387066e-01 -7.89899111e-01
-3.40326667e-01 -2.41008759e-01 1.79703936e-01 -1.19466476e-01
8.54933634e-02 -4.97495830e-01 -1.05800855e+00 5.28023183e-01
-7.71595895e-01 1.46676227e-01 -3.56023520e-01 -2.92561024e-01
-5.32847978e-02 5.14105618e-01 -9.75503862e-01 -1.09615898e+00
-6.34973347e-01 -9.31255996e-01 7.24047959e-01 1.60288200e-01
-4.30612534e-01 -9.81529772e-01 -2.02651218e-01 3.21656436e-01
8.78029585e-01 -2.73524106e-01 7.95189559e-01 -1.54738915e+00
-7.46361732e-01 -8.30512881e-01 -6.34706244e-02 4.74847108e-01
2.72050172e-01 -1.61247507e-01 -7.28884101e-01 -3.66091341e-01
4.77038175e-01 -3.33829105e-01 8.80095422e-01 3.06570828e-01
5.80260694e-01 -7.84891129e-01 -5.70990384e-01 -1.30439132e-01
9.79743421e-01 4.40417528e-01 4.01446372e-01 2.61376023e-01
9.19879302e-02 8.10511172e-01 5.16451836e-01 1.58944577e-01
-3.49984199e-01 1.62766501e-01 6.47050798e-01 -5.09289727e-02
5.58747590e-01 -4.46380109e-01 9.14234817e-01 -1.22464195e-01
7.49001503e-01 -2.02859223e-01 -1.61714330e-01 2.48041376e-01
-1.80426753e+00 -9.85634685e-01 -5.97198486e-01 2.21450520e+00
6.81378305e-01 5.84244430e-01 3.79579455e-01 -2.38136292e-01
9.31287050e-01 2.52143312e-02 -9.95217919e-01 -8.81893516e-01
-2.01854602e-01 1.20020330e-01 1.03259110e+00 5.57154119e-01
-9.85606074e-01 1.26695430e+00 6.83648348e+00 9.80838418e-01
-6.64163530e-01 1.20577902e-01 7.89242387e-01 -9.00958776e-02
-5.36912441e-01 2.76132256e-01 -9.18140173e-01 5.80500364e-01
9.21041489e-01 -6.08964503e-01 4.97176558e-01 1.42559004e+00
4.64519650e-01 -3.25958640e-03 -8.31681192e-01 2.80200005e-01
3.15246917e-02 -1.31117773e+00 1.95931390e-01 2.41265208e-01
6.12245083e-01 -1.37405440e-01 2.83691257e-01 6.25582755e-01
1.37563980e+00 -8.49174321e-01 5.06564928e-03 -2.05669016e-01
2.53013987e-02 -6.98292494e-01 6.28775001e-01 5.21072090e-01
-5.19655108e-01 -4.45089728e-01 -5.71204364e-01 -2.72939913e-02
2.00079963e-01 7.96253383e-01 -1.05489850e+00 -2.23302588e-01
9.28639472e-02 1.79953158e-01 -8.11732709e-01 2.69450635e-01
-3.47012877e-01 7.64823616e-01 -2.94398338e-01 -8.04262817e-01
8.28357115e-02 -3.46821845e-01 8.65914106e-01 8.03918540e-01
-3.14596564e-01 -2.76096582e-01 1.62286848e-01 1.23007834e+00
-3.07961684e-02 -1.26363739e-01 -1.17134202e+00 -3.51981908e-01
3.64936054e-01 1.45536399e+00 -5.48384964e-01 -3.13915163e-01
-1.60126984e-01 9.13373590e-01 -3.34264994e-01 3.57897550e-01
-5.89577913e-01 -2.21365452e-01 8.86650383e-01 4.39107448e-01
-1.84766069e-01 2.58229375e-01 -4.98588413e-01 -9.31112051e-01
-4.73453671e-01 -1.49209130e+00 3.19668025e-01 -3.13545585e-01
-2.01583838e+00 2.72744745e-01 -5.81874073e-01 -9.29474294e-01
-2.43523404e-01 -9.05245960e-01 -8.35618436e-01 6.87246680e-01
-8.29759419e-01 -8.81026030e-01 4.74236757e-01 4.40440178e-01
6.29510224e-01 -6.31187558e-01 5.15061259e-01 -3.36932659e-01
-1.92388922e-01 4.83978689e-01 -7.60594457e-02 1.23496883e-01
4.45958763e-01 -1.04015648e+00 8.06785643e-01 7.55722642e-01
-5.62732182e-02 1.04875910e+00 8.86665344e-01 -1.19450200e+00
-1.56061602e+00 -1.22741115e+00 4.43458855e-01 -8.39541733e-01
1.12892306e+00 -7.26888120e-01 -7.45900273e-01 4.20422554e-01
-2.19984770e-01 -4.37380433e-01 2.85232335e-01 5.79738989e-02
-4.81404305e-01 2.30241362e-02 -1.53656328e+00 9.41609681e-01
8.85560513e-01 -3.31914812e-01 -3.06898832e-01 6.77410662e-01
6.74249470e-01 4.25515741e-01 1.04092546e-01 5.85540757e-02
2.91236013e-01 -9.89352643e-01 9.03824687e-01 -1.07703781e+00
8.45536031e-03 -5.01236580e-02 3.60920221e-01 -1.28906476e+00
-8.53980035e-02 -8.50718200e-01 -3.41504812e-01 8.82036030e-01
5.71329832e-01 -1.22118068e+00 1.07742345e+00 5.18464684e-01
7.13160396e-01 -5.00702083e-01 -4.08625692e-01 -1.24636781e+00
7.00896308e-02 -3.49826902e-01 4.40252393e-01 7.40088820e-01
1.34711280e-01 7.05433428e-01 -5.33292651e-01 -2.99912803e-02
9.14944232e-01 -1.98369846e-02 1.03525710e+00 -1.26293957e+00
-4.86485392e-01 -5.70506394e-01 -9.05885622e-02 -9.38139856e-01
4.97639142e-02 -8.41687143e-01 -2.56790727e-01 -6.90944672e-01
7.47982264e-02 -6.08121306e-02 9.95048955e-02 2.33858109e-01
-6.42136484e-02 4.02405448e-02 -3.74152698e-02 8.22204947e-02
-4.03280228e-01 1.93659991e-01 1.05273151e+00 -4.63049024e-01
-6.22973919e-01 7.17782080e-01 -1.43205154e+00 7.79987693e-01
9.41438675e-01 -6.39251351e-01 -6.17262185e-01 4.94793534e-01
2.29725137e-01 -3.19845736e-01 4.65576977e-01 -3.95647250e-02
-1.71301648e-01 -1.81498006e-01 1.24526545e-01 -1.92029625e-01
-1.63404077e-01 -6.53539717e-01 -2.14310259e-01 1.04898119e+00
-3.69587004e-01 -1.19848199e-01 -2.26875708e-01 1.07807219e+00
4.19876903e-01 -4.00445521e-01 9.63324606e-01 -4.96606380e-01
-2.47032121e-01 3.91030580e-01 -9.10060823e-01 -3.31473611e-02
1.21257472e+00 -2.11842924e-01 -9.98624265e-01 -6.89337134e-01
-4.67954904e-01 5.65252304e-01 6.42252088e-01 6.56112492e-01
4.35244948e-01 -8.56676638e-01 -6.29871547e-01 5.03539503e-01
5.92155429e-03 -1.12616169e+00 -1.67158961e-01 4.43933815e-01
-1.34725764e-01 1.37403950e-01 1.20624088e-01 -4.21499223e-01
-1.32475543e+00 9.35870349e-01 4.95158195e-01 -4.13301378e-01
-1.95732415e-01 5.30181706e-01 4.27147418e-01 -4.98859346e-01
-1.00930236e-01 6.26268804e-01 5.52666970e-02 -2.13886872e-01
6.62429094e-01 3.76910418e-01 -1.02465607e-01 -1.73271462e-01
-3.16749662e-01 -3.28199476e-01 -7.12386668e-01 1.29653960e-01
7.64773428e-01 5.25305681e-02 -2.17210464e-02 -3.97151649e-01
8.41130555e-01 3.06247592e-01 -8.42467725e-01 7.74189159e-02
1.63512886e-01 -8.32056403e-01 -4.18750554e-01 -1.18674362e+00
-7.79583514e-01 4.41303402e-01 4.62025493e-01 1.23743474e+00
7.20981836e-01 -2.12536514e-01 8.96584451e-01 4.25714940e-01
4.94236261e-01 -1.28592491e+00 3.01655948e-01 2.54607886e-01
4.58394468e-01 -1.28122663e+00 -1.45385817e-01 -6.63299263e-01
-8.18428576e-01 5.55180311e-01 9.50926304e-01 -4.76137996e-01
5.96504927e-01 3.24551851e-01 1.79036811e-01 -2.60252982e-01
-1.06047523e+00 1.24435030e-01 -1.95470855e-01 1.11024714e+00
-2.59476215e-01 2.05869704e-01 -7.72927701e-01 6.82149053e-01
2.07451686e-01 -5.07010639e-01 9.34832335e-01 7.20675468e-01
-8.40497971e-01 -1.45098341e+00 -4.17878389e-01 1.03257096e+00
-4.65703338e-01 3.11923884e-02 -1.21134353e+00 7.17520356e-01
-3.63295913e-01 1.37123489e+00 -4.21675056e-01 -5.68176866e-01
3.36211711e-01 -7.80116692e-02 -1.49474248e-01 -6.87695801e-01
-1.09229851e+00 -6.98641017e-02 1.92624122e-01 -4.41512972e-01
3.73569548e-01 -5.00961423e-01 -8.18437874e-01 -7.35936761e-01
-7.57092416e-01 1.73072413e-01 4.24275637e-01 4.79240000e-01
4.61202383e-01 -2.81197131e-02 1.19001949e+00 -4.93552350e-02
-1.71370792e+00 -5.67085743e-01 -9.54458594e-01 7.87792563e-01
-3.31532918e-02 -2.59847969e-01 -1.08730257e+00 -5.52052736e-01]
|
[7.757177829742432, 9.970415115356445]
|
7575ea95-9da6-4a63-9faa-39fdd162b021
|
crossget-cross-guided-ensemble-of-tokens-for
|
2305.17455
| null |
https://arxiv.org/abs/2305.17455v1
|
https://arxiv.org/pdf/2305.17455v1.pdf
|
CrossGET: Cross-Guided Ensemble of Tokens for Accelerating Vision-Language Transformers
|
Vision-language models have achieved tremendous progress far beyond what we ever expected. However, their computational costs and latency are also dramatically growing with rapid development, making model acceleration exceedingly critical for researchers with limited resources and consumers with low-end devices. Although extensively studied for unimodal models, the acceleration for multimodal models, especially the vision-language Transformers, is still relatively under-explored. Accordingly, this paper proposes \textbf{Cross}-\textbf{G}uided \textbf{E}nsemble of \textbf{T}okens (\textbf{\emph{CrossGET}}) as a universal vison-language Transformer acceleration framework, which adaptively reduces token numbers during inference via cross-modal guidance on-the-fly, leading to significant model acceleration while keeping high performance. Specifically, the proposed \textit{CrossGET} has two key designs:1) \textit{Cross-Guided Matching and Ensemble}. \textit{CrossGET} incorporates cross-modal guided token matching and ensemble to merge tokens effectively, only introducing cross-modal tokens with negligible extra parameters. 2) \textit{Complete-Graph Soft Matching}. In contrast to the previous bipartite soft matching approach, \textit{CrossGET} introduces an efficient and effective complete-graph soft matching policy to achieve more reliable token-matching results. Extensive experiments on various vision-language tasks, datasets, and model architectures demonstrate the effectiveness and versatility of the proposed \textit{CrossGET} framework. The code will be at https://github.com/sdc17/CrossGET.
|
['Jiaqi Wang', 'Chun Yuan', 'Zhendong Yang', 'Anyi Rao', 'Chaofan Tao', 'Dachuan Shi']
|
2023-05-27
| null | null | null | null |
['visual-reasoning', 'image-captioning', 'visual-reasoning']
|
['computer-vision', 'computer-vision', 'reasoning']
|
[ 3.60238492e-01 8.63616616e-02 -3.93550128e-01 -3.50534827e-01
-9.72885966e-01 -5.86326480e-01 6.51887834e-01 6.35176525e-02
-4.87974435e-01 2.46748596e-01 -8.03254247e-02 -5.45122981e-01
-2.55306643e-02 -7.22403526e-01 -7.32792497e-01 -6.92672968e-01
6.07587159e-01 5.33904910e-01 2.76406407e-01 -8.96977782e-02
-1.29488632e-01 -7.56052360e-02 -1.56270814e+00 1.50317878e-01
1.20735347e+00 1.08376825e+00 7.13681281e-01 5.54672599e-01
-3.05430859e-01 4.65706378e-01 -1.08495481e-01 -1.04372621e+00
2.22415701e-01 -2.22630594e-02 -5.84041595e-01 -9.29978639e-02
7.48843431e-01 -3.01864505e-01 -7.05249906e-01 1.45642805e+00
6.36352360e-01 -1.16165187e-02 3.92083436e-01 -1.40629470e+00
-8.34839046e-01 8.64052951e-01 -9.40549195e-01 7.79568553e-02
1.82978645e-01 3.70105535e-01 1.33216941e+00 -1.03074348e+00
3.63413900e-01 1.26643455e+00 5.47383666e-01 5.49197674e-01
-1.13869011e+00 -8.72611105e-01 4.45427775e-01 3.81902784e-01
-1.37278283e+00 -4.73887920e-01 6.00339174e-01 -2.11973846e-01
1.02589059e+00 4.62837994e-01 6.26625359e-01 9.79612410e-01
-1.56860724e-01 1.18761730e+00 9.28460777e-01 -3.60943735e-01
-2.07057223e-01 -3.19180153e-02 5.11237323e-01 1.18433237e+00
2.98703581e-01 5.84794804e-02 -5.71459174e-01 -7.87333399e-02
5.33633411e-01 1.14847399e-01 -2.31368884e-01 -7.41950888e-03
-1.25827563e+00 6.12718940e-01 3.12455833e-01 1.88703477e-01
-2.06839278e-01 3.52297038e-01 4.85456258e-01 -1.34086497e-02
2.17832506e-01 -3.81130785e-01 -1.19457021e-02 -1.72184110e-01
-1.05994308e+00 8.80304873e-02 3.84556055e-01 1.38580430e+00
9.16952610e-01 2.92660713e-01 -3.78298193e-01 1.18392622e+00
3.83171082e-01 9.12344813e-01 9.40685421e-02 -7.64157534e-01
8.19784403e-01 6.35428786e-01 -3.82149577e-01 -9.04259622e-01
-3.65222096e-01 -3.64073962e-01 -1.06406641e+00 -3.12357455e-01
2.52317011e-01 6.86387643e-02 -1.18920493e+00 1.76318598e+00
2.54246950e-01 1.57238171e-01 -2.69093007e-01 7.24670053e-01
1.11398315e+00 6.60211027e-01 2.44473755e-01 5.37335165e-02
1.66406906e+00 -1.07309973e+00 -4.43538904e-01 -4.34209317e-01
5.71941733e-01 -1.02312827e+00 1.31646836e+00 2.77403712e-01
-1.13878763e+00 -6.39689445e-01 -7.17532694e-01 -3.37477446e-01
-2.43037194e-01 3.81164789e-01 8.59763443e-01 6.57242894e-01
-1.17629731e+00 -7.11896494e-02 -5.84572494e-01 -2.53665358e-01
3.77851874e-01 6.06846154e-01 -2.41712108e-01 -3.12873840e-01
-1.08497322e+00 5.85717499e-01 5.01759827e-01 4.66664374e-01
-6.61195695e-01 -5.65987110e-01 -1.02945399e+00 1.47693217e-01
6.31446719e-01 -9.64650214e-01 1.20476937e+00 -6.88270628e-01
-1.16918993e+00 9.50427115e-01 -3.35347027e-01 -1.70392036e-01
3.67285162e-01 2.22241253e-01 -3.42529207e-01 8.12720321e-03
-1.84793901e-02 9.93582010e-01 8.49923670e-01 -1.20134735e+00
-6.53278947e-01 -4.88101214e-01 1.09498292e-01 1.45082504e-01
-6.61292851e-01 -1.48743726e-02 -1.20852494e+00 -7.62404740e-01
1.27230644e-01 -1.01742983e+00 7.39224404e-02 -2.10135996e-01
-7.48862803e-01 -3.29866827e-01 6.18439496e-01 -7.63680577e-01
1.39459300e+00 -1.94395769e+00 1.94914844e-02 2.32404441e-01
5.40900886e-01 5.31087697e-01 -2.95649678e-01 3.42318743e-01
1.87857822e-01 -1.15954176e-01 -1.86282009e-01 -6.74997509e-01
4.10985798e-01 2.77852118e-01 -1.64727971e-01 9.78099927e-02
-1.31352395e-01 1.22749054e+00 -7.27805197e-01 -7.75171518e-01
3.67331177e-01 4.40120935e-01 -5.12774169e-01 -2.73730874e-01
-2.66857982e-01 -1.28741562e-01 -5.32412589e-01 9.85395789e-01
8.42913151e-01 -3.96105200e-01 1.45769328e-01 -5.64121306e-01
1.07117496e-01 -1.27428636e-01 -1.10559809e+00 1.52120459e+00
-3.27122003e-01 4.29299623e-01 2.61458725e-01 -9.86417055e-01
7.99061120e-01 -5.53535931e-02 3.06923062e-01 -1.04447806e+00
2.75508046e-01 1.18684359e-01 -2.03677788e-01 -3.71659160e-01
9.47237134e-01 1.50010452e-01 -1.35089830e-01 7.58490041e-02
5.88884577e-02 1.10765442e-01 3.08005184e-01 4.88486469e-01
7.81380177e-01 5.24082184e-02 -2.33597219e-01 1.51293003e-03
5.81838429e-01 -1.10941090e-01 5.94415605e-01 8.31975996e-01
-2.10634008e-01 3.91369879e-01 2.78374583e-01 1.93622321e-01
-7.66945481e-01 -1.22053790e+00 1.78026691e-01 1.15260959e+00
4.13375765e-01 -6.70632303e-01 -8.28053534e-01 -4.70430285e-01
-5.65242544e-02 6.22823954e-01 -1.29422635e-01 -2.25963026e-01
-4.33924049e-01 -6.89471126e-01 9.42242622e-01 6.11228645e-01
7.65495002e-01 -7.68419087e-01 -1.84063788e-03 -5.60681559e-02
-5.38400590e-01 -1.28481758e+00 -9.90103066e-01 -1.72890708e-01
-5.45991004e-01 -6.93234742e-01 -7.89513528e-01 -9.47141886e-01
7.27788627e-01 4.83986318e-01 9.72492039e-01 1.51277214e-01
-3.67547840e-01 6.53105676e-01 -9.12817940e-02 -3.72187048e-01
-1.15399525e-01 -1.06397539e-01 -1.38435647e-01 -9.74437408e-03
4.76255685e-01 -3.15463960e-01 -7.14744389e-01 3.15205693e-01
-9.23655510e-01 4.05138224e-01 7.16240346e-01 1.01289892e+00
9.36248124e-01 -1.56142592e-01 2.40314350e-01 -5.95302105e-01
5.21259248e-01 -1.12513624e-01 -7.46416211e-01 5.78400552e-01
-7.15951264e-01 -8.73271674e-02 6.28701925e-01 -4.92681086e-01
-1.09446108e+00 -7.06148371e-02 -2.01202437e-01 -6.26743972e-01
1.95636377e-01 4.67668623e-01 -2.04234511e-01 -4.90211230e-03
6.05579130e-02 6.39433324e-01 -1.93503469e-01 -2.79879153e-01
4.27118450e-01 6.75452948e-01 7.63778210e-01 -6.93415761e-01
7.62860715e-01 5.10073423e-01 -2.44669497e-01 -9.35647905e-01
-4.63642776e-01 -5.13214409e-01 -1.06667459e-01 -4.32183862e-01
8.58163774e-01 -9.48627591e-01 -1.20567036e+00 7.59147823e-01
-9.06356096e-01 -2.50710964e-01 7.76807144e-02 5.18040478e-01
-3.27986747e-01 8.10196996e-01 -4.89150256e-01 -8.47249269e-01
-6.92588925e-01 -1.36779571e+00 1.30170286e+00 4.77429688e-01
2.10598171e-01 -1.00189936e+00 -4.26958829e-01 1.17517030e+00
1.24204695e-01 -3.15952659e-01 9.63000894e-01 -4.38937843e-01
-8.87759447e-01 -2.26358265e-01 -7.24870920e-01 3.17929029e-01
-2.20690086e-01 4.89994548e-02 -9.41520274e-01 -4.15475309e-01
-6.01575792e-01 -2.74856418e-01 1.05599177e+00 4.11765546e-01
1.21430063e+00 -1.45836109e-02 -4.73091185e-01 5.81091583e-01
1.17907023e+00 6.53459355e-02 4.43546414e-01 -6.58154860e-02
1.06555223e+00 1.32651895e-01 5.54252148e-01 3.40096444e-01
1.14381731e+00 7.02104688e-01 4.46024448e-01 -3.37224960e-01
-2.20198765e-01 -3.33945781e-01 5.47277868e-01 1.05919015e+00
-1.03262529e-01 -4.45609093e-01 -8.81824613e-01 5.39365828e-01
-1.92023075e+00 -1.05715406e+00 -2.02150315e-01 2.15168929e+00
6.16456389e-01 2.42537692e-01 -4.05335836e-02 -2.20613942e-01
9.33893979e-01 3.04732621e-01 -6.31183147e-01 -2.51964897e-01
-1.40131369e-01 1.38417453e-01 5.98778427e-01 5.01436234e-01
-8.68299544e-01 1.13247609e+00 3.85256910e+00 1.48917782e+00
-1.00471747e+00 9.69345421e-02 4.09484267e-01 -2.92624775e-02
-7.32782125e-01 2.44015694e-01 -1.20275033e+00 5.16727328e-01
2.93310374e-01 -6.73777908e-02 5.23598850e-01 5.52669346e-01
1.25843897e-01 -3.41298282e-01 -8.15098226e-01 1.30136633e+00
1.09162189e-01 -1.35659242e+00 3.44179600e-01 1.78072199e-01
4.05139297e-01 9.73718613e-02 1.77534804e-01 5.81091762e-01
2.72096425e-01 -5.45406878e-01 9.09026563e-01 6.45055115e-01
9.19208169e-01 -5.89092970e-01 3.48902911e-01 3.55172396e-01
-1.65401852e+00 9.66451988e-02 -3.50728214e-01 4.62808162e-01
2.52156824e-01 6.89478695e-01 -5.47157109e-01 9.24223781e-01
7.84056962e-01 4.79395062e-01 -4.68891352e-01 8.12540770e-01
7.89606944e-02 6.38952434e-01 -3.39651018e-01 -9.28205475e-02
4.09446478e-01 -3.65245283e-01 6.48958743e-01 1.36271405e+00
2.57169932e-01 -4.09158021e-02 4.48819011e-01 7.27587581e-01
-1.89583510e-01 -6.68658176e-03 -3.60225827e-01 -1.14170313e-01
6.18708849e-01 1.42303765e+00 -7.63880730e-01 -4.17553663e-01
-5.82367837e-01 8.74931216e-01 3.15186948e-01 3.41437548e-01
-1.19242263e+00 -4.54591036e-01 5.04412651e-01 -7.96788037e-02
4.15378541e-01 -1.79229334e-01 -1.88550234e-01 -1.20154893e+00
3.01310837e-01 -7.55693555e-01 6.43435001e-01 -7.91089654e-01
-1.19284642e+00 4.52444941e-01 1.62591234e-01 -9.95255351e-01
1.40344396e-01 -5.15276611e-01 -3.31840515e-01 8.20393205e-01
-1.60914898e+00 -1.61302435e+00 -4.95062411e-01 1.04378343e+00
5.12327194e-01 -9.50049087e-02 4.65108186e-01 6.32830262e-01
-7.42039025e-01 1.26781428e+00 6.36856481e-02 1.62969291e-01
4.92901742e-01 -9.71608877e-01 3.76008451e-01 9.55314755e-01
2.47252092e-01 7.73885965e-01 2.95120537e-01 -7.01914072e-01
-1.69334185e+00 -1.05915570e+00 7.60648072e-01 7.55664101e-03
8.01779032e-01 -5.84228575e-01 -7.49279559e-01 6.53147876e-01
2.53602684e-01 -2.06905738e-01 3.77892345e-01 -8.41648206e-02
-5.15732944e-01 -3.54634762e-01 -9.39480007e-01 9.50125694e-01
1.34125757e+00 -7.92663574e-01 -3.26622784e-01 3.77292961e-01
5.28805971e-01 -4.92220342e-01 -7.16956198e-01 3.88019353e-01
4.39256042e-01 -8.90219688e-01 9.91661549e-01 3.89973423e-03
-3.62599967e-03 -3.50045860e-01 -1.12495445e-01 -7.03968585e-01
-2.95877159e-01 -8.62539768e-01 -2.00208589e-01 1.48328006e+00
4.50573206e-01 -8.74897420e-01 6.66696310e-01 5.25227904e-01
-3.74799609e-01 -6.45630002e-01 -1.06047320e+00 -7.50835299e-01
-3.71972382e-01 -6.72208250e-01 3.54300529e-01 6.98726475e-01
-1.84475958e-01 1.64522886e-01 -4.46956784e-01 1.24734662e-01
8.59592259e-01 2.00765461e-01 6.26780868e-01 -8.14880192e-01
-4.78295147e-01 -6.50774002e-01 -2.31037661e-01 -1.42528415e+00
1.51921257e-01 -1.34154725e+00 -1.25303373e-01 -1.79918575e+00
6.19951844e-01 -5.12653351e-01 -2.59147525e-01 8.55648935e-01
-3.70436698e-01 2.97302693e-01 4.80463177e-01 8.11400786e-02
-6.28858209e-01 5.49368203e-01 1.34790933e+00 -4.93256539e-01
1.72423705e-01 -5.90992756e-02 -7.19419241e-01 6.60511255e-01
4.62920606e-01 -2.40367159e-01 -6.00933731e-01 -6.70313716e-01
2.70943731e-01 9.92081016e-02 5.98149717e-01 -6.55268967e-01
6.43740892e-01 6.78126886e-02 8.39394554e-02 -9.18497860e-01
6.95804000e-01 -6.09061539e-01 6.60036653e-02 2.04011291e-01
-3.27772796e-02 3.58104706e-01 3.44513446e-01 6.10077560e-01
-1.84405163e-01 -1.32200360e-01 5.62643468e-01 5.15364744e-02
-9.47283089e-01 5.49208760e-01 -7.71728903e-02 1.91158891e-01
9.46373403e-01 -6.16965890e-01 -6.24481261e-01 -3.18032712e-01
-5.56478381e-01 5.94384491e-01 2.18417361e-01 5.27550399e-01
7.13859856e-01 -1.14787591e+00 -5.44751942e-01 7.97359124e-02
2.25916609e-01 2.20966190e-02 8.12540114e-01 1.11261642e+00
-6.54224455e-02 3.83592129e-01 1.46504059e-01 -8.90769005e-01
-1.59852576e+00 4.46859121e-01 9.17052999e-02 -3.05125445e-01
-4.74437773e-01 9.97871637e-01 3.42342347e-01 -5.00244975e-01
3.58514220e-01 -3.28653842e-01 2.47534707e-01 1.82303086e-01
2.82182336e-01 4.98440027e-01 2.57406924e-02 -7.20769107e-01
-3.52966517e-01 6.34010613e-01 -4.78940278e-01 1.04033895e-01
9.60732162e-01 -1.60861745e-01 -2.11382061e-01 7.45891826e-04
9.51908708e-01 -1.33713156e-01 -9.12429333e-01 -4.30721343e-01
-2.06379876e-01 -3.81395966e-01 1.36336267e-01 -7.90487707e-01
-1.35004652e+00 7.53025234e-01 4.94831115e-01 -1.15276212e-02
1.26619196e+00 8.82997513e-02 1.18191659e+00 3.93366784e-01
4.59718168e-01 -1.30336165e+00 -4.30039801e-02 4.40994531e-01
5.41509449e-01 -1.06756318e+00 -7.32001439e-02 -4.85727787e-01
-7.65535533e-01 8.12647820e-01 6.59770012e-01 3.63957345e-01
4.63013560e-01 2.62648225e-01 -1.39341041e-01 -2.12238431e-01
-5.30526698e-01 -4.61287498e-01 4.16662902e-01 4.88501459e-01
6.50770664e-02 1.87786624e-01 -2.39344895e-01 4.81455773e-01
-8.31324086e-02 -7.83887804e-02 6.10874891e-02 7.43776321e-01
-2.88630545e-01 -1.14902520e+00 -2.77968615e-01 6.50787652e-01
-1.46155372e-01 -5.31946301e-01 -1.98402405e-01 6.14716589e-01
1.82470828e-01 1.05504954e+00 -1.00514159e-01 -4.81946379e-01
3.79420906e-01 -7.91863054e-02 5.75479865e-01 -2.30047718e-01
-7.72685230e-01 2.30900750e-01 7.48964325e-02 -4.81047750e-01
-2.06720799e-01 -6.60046160e-01 -1.57589233e+00 -4.84711140e-01
-5.58976054e-01 -2.84410089e-01 4.74523723e-01 8.21896315e-01
5.29833257e-01 4.93647993e-01 3.06487173e-01 -6.49743557e-01
-5.25805295e-01 -6.13315403e-01 -3.31536055e-01 1.38340250e-01
-5.89475073e-02 -6.56993449e-01 -2.43161824e-02 2.93754991e-02]
|
[10.25671672821045, 0.9679552912712097]
|
d6d91c79-27c5-423f-9fc6-514c2242ddba
|
loa-logical-optimal-actions-for-text-based-1
|
2110.10973
| null |
https://arxiv.org/abs/2110.10973v1
|
https://arxiv.org/pdf/2110.10973v1.pdf
|
LOA: Logical Optimal Actions for Text-based Interaction Games
|
We present Logical Optimal Actions (LOA), an action decision architecture of reinforcement learning applications with a neuro-symbolic framework which is a combination of neural network and symbolic knowledge acquisition approach for natural language interaction games. The demonstration for LOA experiments consists of a web-based interactive platform for text-based games and visualization for acquired knowledge for improving interpretability for trained rules. This demonstration also provides a comparison module with other neuro-symbolic approaches as well as non-symbolic state-of-the-art agent models on the same text-based games. Our LOA also provides open-sourced implementation in Python for the reinforcement learning environment to facilitate an experiment for studying neuro-symbolic agents. Code: https://github.com/ibm/loa
|
['Alexander Gray', 'Ryosuke Kohita', 'Akifumi Wachi', 'Asim Munawar', 'Don Joven Agravante', 'Michiaki Tatsubori', 'Masaki Ono', 'Subhajit Chaudhury', 'Daiki Kimura']
|
2021-10-21
|
loa-logical-optimal-actions-for-text-based
|
https://aclanthology.org/2021.acl-demo.27
|
https://aclanthology.org/2021.acl-demo.27.pdf
|
acl-2021-5
|
['text-based-games']
|
['playing-games']
|
[-3.89117420e-01 6.34858251e-01 1.46223426e-01 -1.49997398e-01
1.53905571e-01 -3.80073637e-01 6.29667819e-01 -1.96286961e-01
-2.13100240e-01 9.18700397e-01 -1.12657353e-01 -8.22264433e-01
-5.51802456e-01 -1.30513036e+00 -7.12524116e-01 -1.92954779e-01
-2.22264364e-01 9.13767695e-01 5.01779556e-01 -8.62363935e-01
2.19195783e-01 3.48506600e-01 -1.76129937e+00 6.90420926e-01
5.65573156e-01 3.79553139e-01 1.19934745e-01 9.03864026e-01
-1.40980050e-01 1.52802467e+00 -3.47292483e-01 -3.55126001e-02
2.95273900e-01 -7.28693247e-01 -1.32952857e+00 -7.77074873e-01
-2.43072346e-01 -5.51840246e-01 -4.92092907e-01 1.13168669e+00
-1.53102688e-02 2.54208982e-01 4.21611547e-01 -1.71987247e+00
-4.20122534e-01 1.26539946e+00 2.37430498e-01 7.06821308e-02
7.20375061e-01 9.08089340e-01 6.03464127e-01 -1.82661593e-01
7.25736201e-01 1.57086897e+00 4.96234506e-01 9.84779537e-01
-8.76681626e-01 -7.66605377e-01 -7.46218637e-02 6.95079803e-01
-9.72936988e-01 6.74787015e-02 4.33499753e-01 -3.60911012e-01
1.68576813e+00 6.93800747e-02 1.22922683e+00 1.18511605e+00
1.11708216e-01 6.63902581e-01 1.19189000e+00 -6.80045426e-01
4.47746962e-01 -1.30735710e-01 2.51874626e-01 1.30972469e+00
-7.29757994e-02 9.69728708e-01 -6.52694941e-01 -6.93551078e-02
1.41209769e+00 -3.11029643e-01 5.45519888e-01 -1.51748508e-01
-1.01189148e+00 8.97465229e-01 2.85091192e-01 4.53182429e-01
-5.42727172e-01 6.98031843e-01 5.62320292e-01 5.34693778e-01
-3.93833756e-01 7.48844624e-01 -5.59029043e-01 -7.04692245e-01
-3.60082299e-01 8.64775956e-01 1.03896201e+00 7.95678079e-01
4.67732161e-01 6.31067693e-01 -2.81864792e-01 3.66440713e-01
7.81354189e-01 3.91560644e-01 6.51925147e-01 -1.57954311e+00
-2.98437446e-01 8.48448813e-01 4.35416438e-02 -7.08526015e-01
-6.21587813e-01 2.06741676e-01 -1.28823906e-01 1.30450630e+00
7.80831337e-01 -3.11784416e-01 -6.91252291e-01 1.73192143e+00
1.78602219e-01 5.80758035e-01 3.85565758e-01 7.24098802e-01
9.54055488e-01 6.35375082e-01 3.08803856e-01 2.45588616e-01
1.32723439e+00 -1.00489259e+00 -3.49034876e-01 1.70835555e-01
7.81305075e-01 9.66054127e-02 1.30907941e+00 7.04548836e-01
-1.29370522e+00 -4.41036135e-01 -9.93828893e-01 2.01830208e-01
-9.76510465e-01 -2.16959029e-01 1.07386458e+00 5.06468952e-01
-1.27503443e+00 8.20902884e-01 -1.08595550e+00 -4.13794011e-01
2.91094154e-01 6.57169700e-01 -1.31409168e-02 4.61443037e-01
-1.55583012e+00 1.34858060e+00 1.15498531e+00 -3.40702444e-01
-1.27791464e+00 -5.55411041e-01 -1.15173483e+00 2.57157031e-02
7.89826691e-01 -7.53945768e-01 1.86980224e+00 -9.67885137e-01
-2.42652583e+00 6.59697533e-01 4.87188011e-01 -8.75097334e-01
3.19718689e-01 -9.83966812e-02 1.63366236e-02 -3.22125517e-02
-1.85395360e-01 7.40937948e-01 2.22667739e-01 -9.87019062e-01
-5.47757447e-01 -1.66352078e-01 8.51386428e-01 1.51029885e-01
4.68712986e-01 1.37302592e-01 1.20889492e-01 -2.56097347e-01
-6.60506010e-01 -6.91211998e-01 -2.17365295e-01 -1.73943415e-01
1.32644605e-02 -5.23513317e-01 6.12521470e-01 -4.62828010e-01
1.04132020e+00 -1.56872964e+00 5.21883667e-02 4.01819468e-01
5.31437397e-02 4.47135389e-01 -1.79198250e-01 6.29602849e-01
-2.75237590e-01 8.72741267e-02 2.05340624e-01 4.72458363e-01
4.77014244e-01 4.13569629e-01 -8.64880010e-02 -4.32934493e-01
-8.35045800e-03 1.04399264e+00 -1.39243424e+00 -5.11413336e-01
8.84000123e-01 -9.10919160e-02 -5.84859908e-01 1.88491151e-01
-8.85113716e-01 3.70426297e-01 -6.63216114e-01 4.37313378e-01
2.51305997e-02 8.21417570e-03 4.33725208e-01 5.43818355e-01
-1.04886323e-01 3.77847224e-01 -1.24060369e+00 1.86792755e+00
-4.19281602e-01 3.14073861e-01 -2.63443470e-01 -9.94160354e-01
5.93584776e-01 4.43123162e-01 8.66649970e-02 -4.96081084e-01
5.74747205e-01 1.23480875e-02 3.68801683e-01 -8.99075627e-01
1.76153690e-01 7.64040351e-02 -4.18259166e-02 5.89390278e-01
3.97648633e-01 -3.79157722e-01 5.99810839e-01 2.39835545e-01
1.29227018e+00 1.07131779e+00 7.62295783e-01 -1.44057587e-01
4.31677699e-01 5.56436121e-01 2.40231201e-01 1.13431609e+00
-1.02560028e-01 -3.09067696e-01 5.90002477e-01 -6.92812800e-01
-1.09372818e+00 -7.58372843e-01 3.57617408e-01 1.38269007e+00
-1.32066563e-01 -5.33753812e-01 -1.12055075e+00 -3.51492077e-01
-6.40968531e-02 1.31371522e+00 -6.77698493e-01 -2.07002684e-01
-4.49892938e-01 -4.62740660e-04 1.08613670e+00 6.18337512e-01
6.59356236e-01 -2.14402890e+00 -1.04508448e+00 8.71919021e-02
3.23284209e-01 -6.34699345e-01 4.80536431e-01 1.66168720e-01
-5.87336779e-01 -1.26810670e+00 3.40416700e-01 -6.83007836e-01
1.41245082e-01 -5.60955524e-01 1.12098813e+00 5.84180415e-01
-1.73398063e-01 4.96221006e-01 -5.13807714e-01 -7.80197799e-01
-9.46889222e-01 -2.14991838e-01 -7.37086833e-02 -1.02091765e+00
4.58739102e-01 -9.85340416e-01 -9.09079835e-02 -5.10121956e-02
-6.08148098e-01 7.34828651e-01 8.21835101e-02 8.23275328e-01
-4.21416573e-02 -9.35665425e-03 4.51431185e-01 -6.45046294e-01
1.22526026e+00 -3.09866548e-01 -1.07828498e+00 2.50667959e-01
-4.41260397e-01 4.84035432e-01 6.40884936e-01 -2.98262119e-01
-8.17912936e-01 -6.62764013e-02 -1.98796734e-01 -2.35205084e-01
-7.87485182e-01 8.53546143e-01 2.54623264e-01 5.55228591e-02
1.04514468e+00 2.09068194e-01 4.16157454e-01 1.18538663e-01
5.05610466e-01 5.14675319e-01 5.37820816e-01 -1.15261936e+00
4.56453085e-01 -1.84180722e-01 5.98072410e-02 -2.47626469e-01
-1.87285528e-01 1.96186796e-01 -4.50016648e-01 -6.43701792e-01
8.94956529e-01 -3.16104054e-01 -1.59817374e+00 6.25584900e-01
-1.10824454e+00 -1.42592835e+00 -6.44195080e-01 3.51018131e-01
-1.20288002e+00 -2.10144579e-01 -6.14318967e-01 -9.08979177e-01
-2.18012199e-01 -1.13539064e+00 3.57722163e-01 6.37546837e-01
-6.58632755e-01 -9.18719769e-01 3.46298575e-01 2.09281549e-01
2.08312407e-01 3.63236159e-01 1.16280520e+00 -1.10240948e+00
-3.46439391e-01 2.27412447e-01 1.92727059e-01 5.20037338e-02
-2.55486161e-01 1.20925196e-01 -6.23677075e-01 4.39233094e-01
-5.06279767e-01 -9.55027640e-01 6.88660145e-03 3.74317259e-01
9.06776011e-01 -2.47826084e-01 -1.20612733e-01 4.25510183e-02
9.92773890e-01 8.44529748e-01 8.38935256e-01 9.31118131e-01
1.47575840e-01 2.90700823e-01 7.45932698e-01 5.27831435e-01
3.23502511e-01 6.46280408e-01 6.25248969e-01 4.57385242e-01
1.84546053e-01 -2.31116131e-01 4.47106779e-01 2.14167103e-01
-6.76218212e-01 2.65177667e-01 -1.46873653e+00 2.20117643e-01
-2.56756592e+00 -1.43307936e+00 2.24242490e-02 1.66918981e+00
1.24884856e+00 1.47410795e-01 4.60271090e-01 1.00700974e-01
4.02165294e-01 -3.09707254e-01 -5.89475095e-01 -9.94469881e-01
5.26156127e-01 6.74974203e-01 -1.84887573e-01 1.03388906e+00
-8.05188477e-01 1.56744897e+00 6.62951660e+00 1.00207186e+00
-8.64551723e-01 2.92926747e-02 5.28966449e-02 1.09855421e-01
2.61301786e-01 -9.66240391e-02 -3.97547394e-01 -3.26538924e-03
1.31239533e+00 -5.46536446e-01 1.29764855e+00 1.31678951e+00
5.54334641e-01 -3.44241142e-01 -1.08421528e+00 6.10090911e-01
-4.01778251e-01 -1.77031517e+00 1.27559096e-01 -1.91070139e-01
2.94452995e-01 2.51214020e-03 -3.61332327e-01 8.92068505e-01
1.24222982e+00 -1.23880529e+00 8.55610967e-01 6.60872102e-01
3.93845469e-01 -6.87629521e-01 5.84092975e-01 4.67065603e-01
-7.02411354e-01 -2.94608623e-01 -8.14115107e-02 -7.40382671e-01
-3.27938169e-01 -5.37094712e-01 -8.69509757e-01 4.16185319e-01
7.02671230e-01 5.89771390e-01 -4.72006470e-01 7.85619855e-01
-6.85642123e-01 5.98754108e-01 -3.96592170e-01 -5.23313284e-01
3.69144917e-01 -1.97196513e-01 4.61115628e-01 1.09049594e+00
-2.33554184e-01 6.03807509e-01 3.55199613e-02 1.32568002e+00
4.89215553e-01 -2.39031747e-01 -8.66752207e-01 -1.75584555e-01
4.03164893e-01 9.57029462e-01 -8.51321697e-01 -5.37369430e-01
-1.50641561e-01 4.52257305e-01 3.97767991e-01 1.35888562e-01
-1.09286463e+00 -4.70168173e-01 5.77146530e-01 -1.19655937e-01
7.38841072e-02 -3.21503997e-01 -3.83004844e-01 -6.83836341e-01
-7.21463144e-01 -1.44869030e+00 3.28011632e-01 -1.38948488e+00
-6.80528760e-01 4.66214389e-01 6.46833003e-01 -7.40329325e-01
-1.14516389e+00 -1.11968803e+00 -8.96934032e-01 5.92079759e-01
-6.56875730e-01 -1.18190467e+00 -4.87167269e-01 9.72074151e-01
4.05805916e-01 -7.35820174e-01 1.41289246e+00 -3.56987029e-01
-3.03721517e-01 1.75337479e-01 -2.47271463e-01 3.14124912e-01
-1.84442952e-01 -1.39497042e+00 3.84197801e-01 4.54708397e-01
-1.31005999e-02 4.88210738e-01 7.24854410e-01 -7.14374125e-01
-1.11315835e+00 -5.22654057e-01 5.51965125e-02 -4.26630259e-01
9.15694714e-01 -6.53982535e-02 -6.39737844e-01 9.78488922e-01
6.70689702e-01 -4.81274247e-01 4.56419230e-01 1.05331175e-01
-2.03597918e-03 2.72145331e-01 -1.10028481e+00 1.06187308e+00
1.09085691e+00 -4.12434638e-01 -9.92726624e-01 2.29174048e-01
5.64421952e-01 -6.71228528e-01 -6.76270545e-01 1.26620140e-02
4.87938613e-01 -1.14264870e+00 8.12486112e-01 -1.01357937e+00
7.43086278e-01 -6.41476870e-01 3.56899023e-01 -1.19534433e+00
-2.99507022e-01 -7.01829255e-01 -1.78230330e-01 6.33498490e-01
3.77840936e-01 -6.99512899e-01 7.49407232e-01 7.04407990e-01
-8.32828134e-02 -7.33505011e-01 -7.65547633e-01 -7.93389559e-01
4.01344933e-02 -9.77342248e-01 6.35615408e-01 7.88091898e-01
9.46294546e-01 5.96098043e-02 4.17922474e-02 1.75418227e-03
4.46593985e-02 -1.95347443e-01 7.63336837e-01 -9.40905452e-01
-6.91477180e-01 -8.03415179e-01 -5.63646853e-01 -2.79088974e-01
3.39381218e-01 -1.13097656e+00 -5.97092174e-02 -1.56908846e+00
-7.69021586e-02 5.81015795e-02 -1.33299783e-01 1.37413538e+00
6.36527061e-01 -2.80507654e-01 2.93989956e-01 -2.30270505e-01
-6.42436206e-01 2.10329980e-01 1.15095544e+00 3.03237922e-02
-4.58013296e-01 -3.08104157e-01 -3.64578843e-01 1.12698650e+00
1.20299482e+00 -6.19888306e-01 -6.67142868e-01 5.65186515e-02
4.62497801e-01 1.84000105e-01 8.32263827e-01 -1.29209960e+00
2.67661989e-01 -7.42320120e-01 -5.43217920e-02 -8.98448005e-02
2.08604693e-01 -6.86595321e-01 1.25805616e-01 9.78738070e-01
-5.28301060e-01 -4.80391569e-02 7.16259956e-01 -8.80776569e-02
4.90791500e-02 -6.26366377e-01 5.36779284e-01 -5.87714732e-01
-7.36829579e-01 -3.96585166e-01 -1.15369189e+00 -5.91002777e-02
1.33746648e+00 -4.40191060e-01 -4.09120768e-01 -6.76727951e-01
-1.12732613e+00 2.65903682e-01 5.56758121e-02 2.40751609e-01
5.23238659e-01 -1.05832446e+00 -3.42453659e-01 -5.87139614e-02
-2.92909145e-01 -1.90944523e-01 -1.78736538e-01 4.15984035e-01
-1.20714784e+00 4.09382015e-01 -1.01552355e+00 8.65708739e-02
-1.38053930e+00 4.82169569e-01 9.99904335e-01 -6.25942945e-01
-4.95831460e-01 4.50242519e-01 -4.06597197e-01 -8.05379152e-01
2.30775595e-01 -9.29964602e-01 -4.08500671e-01 -7.99206734e-01
7.04843879e-01 3.04544508e-01 -3.66431504e-01 1.98642071e-02
-2.66279012e-01 -9.50118750e-02 2.49578327e-01 -5.61017334e-01
1.57316124e+00 5.67401290e-01 -1.22965813e-01 6.17473781e-01
-8.08809698e-02 -7.33022749e-01 -8.87499869e-01 4.76682670e-02
-3.95935513e-02 1.97637022e-01 -1.63952619e-01 -1.36373818e+00
-5.91944337e-01 5.58944523e-01 4.30950552e-01 3.72410148e-01
6.75566435e-01 -9.76840034e-02 -1.12396374e-01 1.13823569e+00
7.17460394e-01 -1.16734505e+00 1.03151716e-01 1.15929329e+00
1.35823476e+00 -8.64235044e-01 -7.38170221e-02 -2.17155069e-02
-7.20474482e-01 1.44940925e+00 1.09003997e+00 -5.49175620e-01
7.31452346e-01 5.68866253e-01 1.03284568e-02 -4.68464375e-01
-9.17451739e-01 -2.39057034e-01 -3.22930813e-01 1.09182346e+00
1.02171451e-01 -2.22248156e-02 -1.82467699e-01 1.01742423e+00
-6.95924282e-01 8.09871018e-01 4.80232716e-01 1.15622687e+00
-2.42151409e-01 -1.17897248e+00 -4.59126800e-01 1.37779981e-01
-1.58600435e-01 -1.85061604e-01 -4.01555359e-01 1.17582417e+00
1.34109110e-01 7.87006378e-01 -2.08667994e-01 -6.36314094e-01
2.26496965e-01 4.61504519e-01 8.37532341e-01 -6.80086732e-01
-1.26915348e+00 -6.19457424e-01 4.47937012e-01 -9.19619024e-01
-3.55367184e-01 -5.38824737e-01 -2.05704594e+00 -8.04102421e-01
1.99325010e-01 1.90182365e-02 4.10329193e-01 1.09122229e+00
1.82193905e-01 6.69984579e-01 -4.01090741e-01 -1.03480566e+00
-1.09973006e-01 -1.08085239e+00 -4.37404335e-01 1.68283507e-01
-5.33247948e-01 -6.36365354e-01 1.72897503e-01 5.86977750e-02]
|
[3.855628490447998, 1.3490887880325317]
|
e3265c13-847a-4009-a9d6-1dcbae665dc5
|
schema-first-learn-versatile-knowledge-graph
|
2306.03659
| null |
https://arxiv.org/abs/2306.03659v1
|
https://arxiv.org/pdf/2306.03659v1.pdf
|
Schema First! Learn Versatile Knowledge Graph Embeddings by Capturing Semantics with MASCHInE
|
Knowledge graph embedding models (KGEMs) have gained considerable traction in recent years. These models learn a vector representation of knowledge graph entities and relations, a.k.a. knowledge graph embeddings (KGEs). Learning versatile KGEs is desirable as it makes them useful for a broad range of tasks. However, KGEMs are usually trained for a specific task, which makes their embeddings task-dependent. In parallel, the widespread assumption that KGEMs actually create a semantic representation of the underlying entities and relations (e.g., project similar entities closer than dissimilar ones) has been challenged. In this work, we design heuristics for generating protographs -- small, modified versions of a KG that leverage schema-based information. The learnt protograph-based embeddings are meant to encapsulate the semantics of a KG, and can be leveraged in learning KGEs that, in turn, also better capture semantics. Extensive experiments on various evaluation benchmarks demonstrate the soundness of this approach, which we call Modular and Agnostic SCHema-based Integration of protograph Embeddings (MASCHInE). In particular, MASCHInE helps produce more versatile KGEs that yield substantially better performance for entity clustering and node classification tasks. For link prediction, using MASCHInE has little impact on rank-based performance but increases the number of semantically valid predictions.
|
['Davy Monticolo', 'Armelle Brun', 'Pierre Monnin', 'Heiko Paulheim', 'Nicolas Hubert']
|
2023-06-06
| null | null | null | null |
['graph-embedding', 'link-prediction', 'knowledge-graph-embedding', 'knowledge-graph-embeddings', 'knowledge-graph-embeddings']
|
['graphs', 'graphs', 'graphs', 'graphs', 'methodology']
|
[-3.29166383e-01 4.44781274e-01 -5.42514086e-01 -2.67461807e-01
5.84326684e-02 -6.44733846e-01 4.76944208e-01 6.77640498e-01
-5.76668754e-02 3.73673320e-01 5.37325859e-01 -1.85005322e-01
-4.30677891e-01 -1.29447055e+00 -5.92863023e-01 -4.99159753e-01
-3.27057511e-01 4.62180793e-01 3.45829129e-01 -4.46282953e-01
-1.11827411e-01 4.39200878e-01 -1.31200409e+00 1.30444810e-01
7.85190284e-01 7.73774743e-01 -8.68002847e-02 3.85795087e-01
-3.91812503e-01 6.34847224e-01 -3.25499535e-01 -1.07223976e+00
1.25123739e-01 -3.45646739e-02 -8.90180945e-01 -2.82827675e-01
1.79245517e-01 7.80382603e-02 -7.16135621e-01 8.59352767e-01
2.52773672e-01 1.91880018e-01 9.01892126e-01 -1.59400761e+00
-1.13684380e+00 9.24853444e-01 -3.20491314e-01 9.26434919e-02
2.68279135e-01 -2.62789637e-01 1.83051026e+00 -7.89791405e-01
8.95136535e-01 1.22954202e+00 8.03306460e-01 2.82381922e-01
-1.37751734e+00 -3.34571302e-01 1.60783976e-01 4.45188910e-01
-1.55748034e+00 -2.72257254e-03 7.20915854e-01 -4.73131925e-01
1.06200588e+00 2.58072406e-01 7.43897855e-01 8.20253491e-01
6.33076876e-02 7.40241349e-01 4.85552788e-01 -3.45308483e-01
7.27476031e-02 4.41973120e-01 3.17334920e-01 7.15543628e-01
9.31834877e-01 -1.97010681e-01 -4.45922464e-01 -2.64800519e-01
5.75778723e-01 1.20164119e-01 -4.29566205e-01 -1.20713639e+00
-1.00201726e+00 1.12092257e+00 9.30732429e-01 3.10772687e-01
-3.31215322e-01 2.36833468e-01 4.69912380e-01 3.21071953e-01
1.72558770e-01 7.17327297e-01 -4.16495055e-01 1.56017289e-01
-4.43462312e-01 2.05204442e-01 9.38913763e-01 1.06090903e+00
8.88184011e-01 -1.19465113e-01 3.27487737e-02 8.83472502e-01
2.98824519e-01 1.84375606e-02 2.93287903e-01 -4.31687742e-01
3.82164985e-01 1.11773610e+00 -1.99641883e-01 -1.44690275e+00
-4.44142550e-01 -3.50511998e-01 -5.62990844e-01 -3.00028056e-01
2.18051439e-03 1.83880180e-01 -7.34661341e-01 1.81652141e+00
5.14377058e-01 1.92184925e-01 3.02282631e-01 6.64543450e-01
9.13582981e-01 5.28724790e-01 1.96581632e-01 4.06237274e-01
1.31999910e+00 -8.12971592e-01 -4.47822869e-01 -2.26445608e-02
1.04610133e+00 -2.40515739e-01 7.82453716e-01 -1.46000519e-01
-5.84061503e-01 -2.40523949e-01 -9.70812380e-01 -3.12802345e-02
-9.36278224e-01 -3.23348999e-01 1.10795307e+00 7.19604969e-01
-1.00361633e+00 5.97834647e-01 -6.74151957e-01 -6.06301486e-01
4.46664155e-01 2.14785039e-01 -8.05717826e-01 -1.88520864e-01
-1.44253635e+00 9.62258339e-01 1.08068097e+00 -2.76982635e-01
-2.27147490e-01 -8.25169384e-01 -1.32541442e+00 4.36388135e-01
5.38142681e-01 -8.49285185e-01 5.34704328e-01 -5.61636388e-01
-7.90225029e-01 7.77806044e-01 1.52803704e-01 -5.37699223e-01
-4.73039076e-02 7.12234154e-02 -8.74509454e-01 1.18478738e-01
-4.09724331e-03 6.13717675e-01 6.01772189e-01 -1.31999624e+00
-5.74317932e-01 -2.46542737e-01 4.10886973e-01 1.20803691e-01
-9.00372386e-01 -3.37128967e-01 -6.05139852e-01 -7.04776406e-01
-4.68740799e-02 -8.67285371e-01 2.30446812e-02 -4.64568622e-02
-5.11393189e-01 -5.12614608e-01 6.96300864e-01 -4.87559259e-01
1.62542117e+00 -2.12058473e+00 3.03764939e-01 5.86565733e-01
7.13671029e-01 3.04126859e-01 -1.91174775e-01 1.04931736e+00
-1.93147942e-01 4.34365809e-01 -6.76114857e-02 1.66555762e-01
2.51334876e-01 5.45686603e-01 -1.87032834e-01 1.17357984e-01
3.90445262e-01 1.39393938e+00 -1.06070507e+00 -4.54054356e-01
6.43048733e-02 2.85765588e-01 -7.18292832e-01 4.00694087e-03
-2.42105737e-01 -3.68269920e-01 -4.78211522e-01 4.64881301e-01
3.84920835e-01 -5.34603775e-01 5.82360744e-01 -4.99623090e-01
5.04887700e-01 9.14262012e-02 -1.14600933e+00 1.41761088e+00
-4.73693728e-01 5.48829913e-01 -4.46717501e-01 -1.20612574e+00
8.63769472e-01 1.47416651e-01 4.49495345e-01 -2.76282609e-01
-4.56882827e-02 4.22956161e-02 6.69094622e-02 -3.59584570e-01
6.94072187e-01 -6.77782148e-02 -1.02068335e-01 3.51547420e-01
2.39779487e-01 2.45131463e-01 3.55307400e-01 8.82468343e-01
1.30772555e+00 -8.16931129e-02 6.81669652e-01 -1.20607823e-01
1.70060962e-01 9.44545120e-02 7.23830104e-01 4.19660538e-01
9.25714299e-02 1.15910627e-01 7.10461497e-01 -2.73284346e-01
-9.94041741e-01 -1.40171099e+00 -6.49881363e-02 9.62131262e-01
4.08077151e-01 -1.00161910e+00 -2.18959391e-01 -1.08625996e+00
6.72779858e-01 6.58214808e-01 -7.34673738e-01 -5.38071334e-01
-2.68636018e-01 -4.38358992e-01 5.01121938e-01 9.25577700e-01
-9.19969082e-02 -7.84161448e-01 -1.50837794e-01 2.46450111e-01
1.50601164e-01 -1.21963930e+00 -2.81138271e-01 2.57012229e-02
-6.70681119e-01 -1.43926203e+00 -3.75338227e-01 -8.19334567e-01
6.46069169e-01 3.60146046e-01 1.36617887e+00 2.19242632e-01
-2.52124786e-01 6.91691399e-01 -8.06187570e-01 -2.10839212e-01
-2.12483168e-01 1.57206818e-01 1.72253042e-01 1.38334716e-02
5.86833179e-01 -7.71221459e-01 -2.78093457e-01 2.77470976e-01
-1.05121148e+00 -7.74845704e-02 6.16468906e-01 9.21967685e-01
4.15233165e-01 3.05845231e-01 7.12202787e-01 -1.30902302e+00
6.90481067e-01 -8.43897045e-01 -2.68567532e-01 5.50793946e-01
-8.62634659e-01 3.80274981e-01 6.68104768e-01 -2.98863739e-01
-6.06451392e-01 -4.73376393e-01 6.47904677e-03 -6.43678546e-01
2.55302519e-01 9.57739234e-01 -1.62066445e-01 -1.17043585e-01
6.68323696e-01 2.13392712e-02 -2.93742776e-01 -3.95610809e-01
8.78950536e-01 5.01239181e-01 4.77431238e-01 -6.46439433e-01
1.14390790e+00 3.18655133e-01 4.13698889e-02 -6.28212333e-01
-5.54469049e-01 -6.75296485e-01 -4.45098549e-01 1.54053256e-01
7.62743771e-01 -8.79337072e-01 -4.54066545e-01 -2.85116255e-01
-8.61098409e-01 -4.28973325e-02 -3.41917783e-01 3.10074449e-01
-1.23264141e-01 3.40466410e-01 -4.04427111e-01 -2.98873156e-01
-1.90932840e-01 -6.84959114e-01 6.28470600e-01 1.39365435e-01
-3.91231179e-01 -1.46193874e+00 3.97571325e-02 6.49085268e-02
1.92255631e-01 4.31466967e-01 1.45220673e+00 -1.04544532e+00
-5.80643535e-01 -2.28139311e-01 -4.01224941e-01 2.12532297e-01
2.31863812e-01 9.12643597e-03 -6.87642753e-01 -2.66852111e-01
-9.91158128e-01 -1.82886124e-01 8.41785967e-01 -4.05077152e-02
9.10111129e-01 -2.62891322e-01 -7.83836007e-01 6.25463605e-01
1.60459328e+00 -2.30704576e-01 5.20407915e-01 2.40329027e-01
1.08247912e+00 5.44008136e-01 3.81396860e-01 2.88564205e-01
8.38918626e-01 8.46878409e-01 4.11138684e-01 1.96366280e-01
-1.93749264e-01 -6.89518452e-01 2.19925180e-01 9.57927942e-01
1.38485357e-01 -3.20867002e-01 -9.95018423e-01 8.52716446e-01
-1.82487106e+00 -8.72624040e-01 -1.47835001e-01 1.80425239e+00
8.60414982e-01 -4.77692075e-02 3.86279337e-02 1.31879598e-01
6.96654916e-01 2.39573598e-01 -4.57710862e-01 -4.76806819e-01
-5.48320040e-02 3.62964302e-01 4.23494190e-01 1.08326308e-01
-8.39904904e-01 1.08281195e+00 4.81182194e+00 7.73131967e-01
-7.89971650e-01 1.28652558e-01 -1.21247321e-01 3.03779900e-01
-8.45142424e-01 2.25225970e-01 -6.80060923e-01 4.08029705e-01
7.47820735e-01 -6.25479162e-01 2.77755499e-01 1.07566357e+00
-4.39091653e-01 2.79154539e-01 -1.28281307e+00 1.05348647e+00
-3.48851979e-02 -1.66632402e+00 2.69883722e-01 1.34338021e-01
8.30798924e-01 -8.93041715e-02 -2.13264152e-01 6.44045174e-01
6.92352295e-01 -9.08885837e-01 4.76628214e-01 2.36079738e-01
7.46819854e-01 -8.10196340e-01 7.24767268e-01 -7.89234713e-02
-1.52725410e+00 -8.48459750e-02 -6.39164686e-01 3.33657324e-01
1.05832256e-01 7.47723937e-01 -9.28726375e-01 1.21495152e+00
6.17279708e-01 8.62830341e-01 -6.65538609e-01 9.04475331e-01
-6.01858616e-01 4.82334763e-01 -1.16374157e-01 9.46417265e-03
2.51321197e-01 -1.45150051e-01 4.18229550e-01 1.22107267e+00
6.52159452e-02 -1.16081811e-01 6.19233139e-02 7.74283350e-01
-5.90559900e-01 1.80547908e-01 -8.97642493e-01 -5.29626429e-01
9.17810738e-01 1.42503870e+00 -5.60972452e-01 -2.50200838e-01
-5.25516212e-01 8.73053312e-01 7.37944722e-01 2.76736200e-01
-7.30001450e-01 -5.86581171e-01 1.08306980e+00 6.90795332e-02
6.83356464e-01 -1.22827113e-01 -5.85814118e-02 -1.23033655e+00
-1.49017805e-02 -5.67081690e-01 8.16049278e-01 -5.60354352e-01
-1.59364438e+00 3.53276223e-01 -3.44066881e-02 -9.29031789e-01
-5.86626753e-02 -6.81505382e-01 -6.66449010e-01 6.09568119e-01
-1.49617612e+00 -1.23360908e+00 -3.31482679e-01 4.06788379e-01
-1.04357742e-01 -4.92335744e-02 9.61332560e-01 2.22847253e-01
-5.78705430e-01 8.69102955e-01 1.43478557e-01 4.69966918e-01
5.51200747e-01 -1.47797775e+00 4.44182634e-01 6.62853718e-01
6.58989549e-01 1.06492460e+00 4.01063353e-01 -6.85590625e-01
-1.51551259e+00 -1.45204949e+00 9.05064881e-01 -6.21334970e-01
9.29242134e-01 -3.00577879e-01 -9.89713669e-01 9.02173221e-01
-1.55054301e-01 2.51836061e-01 1.08291686e+00 7.12028980e-01
-9.47211444e-01 -4.66398560e-02 -9.41247404e-01 7.58687496e-01
1.30963790e+00 -6.61350846e-01 -7.14549959e-01 5.55653460e-02
8.89016569e-01 -9.03765019e-03 -1.49315453e+00 3.69583219e-01
5.55997849e-01 -7.51487970e-01 1.15483189e+00 -1.04321504e+00
3.28179628e-01 -3.78326237e-01 -1.92782700e-01 -1.64776540e+00
-5.13905108e-01 -1.07035510e-01 -7.36086845e-01 1.29662776e+00
4.76949543e-01 -7.87112296e-01 7.59665191e-01 4.17674035e-01
-7.32444972e-02 -1.06773460e+00 -5.77985942e-01 -1.22285438e+00
-8.29790086e-02 -2.69700378e-01 8.75298977e-01 1.57847142e+00
2.28672251e-01 4.72316831e-01 -8.20145607e-02 4.51445520e-01
4.65083450e-01 2.86278248e-01 8.54419470e-01 -1.52717376e+00
-1.67455226e-01 -4.65730548e-01 -1.28506446e+00 -6.68827653e-01
3.01465660e-01 -1.57378924e+00 -5.71260333e-01 -1.89033270e+00
1.44379795e-01 -8.23543310e-01 -5.03165781e-01 7.28916049e-01
-4.35503781e-01 -3.78020890e-02 1.86762899e-01 6.22412637e-02
-5.95530510e-01 7.25810945e-01 6.61994100e-01 -2.32421160e-01
-8.93513337e-02 -4.46048230e-01 -1.04462349e+00 4.55381453e-01
7.45747030e-01 -5.13863742e-01 -7.25906074e-01 -3.16948563e-01
4.81690854e-01 -3.87354761e-01 3.47438693e-01 -7.19986498e-01
2.18406290e-01 -1.20565994e-03 9.57061276e-02 -5.65067716e-02
2.21016154e-01 -9.25926208e-01 3.60843033e-01 1.04876645e-01
-1.13185160e-01 7.16527039e-03 2.07322836e-02 1.10272777e+00
-3.56976390e-01 -2.23337799e-01 4.13180202e-01 2.27229565e-01
-1.26382482e+00 4.70449060e-01 3.66243631e-01 3.05536985e-01
1.11601162e+00 -3.86928022e-01 -2.78280199e-01 -2.62178361e-01
-6.80360913e-01 3.62502754e-01 6.26973569e-01 7.41649628e-01
6.00487530e-01 -1.62212813e+00 -3.96528780e-01 -4.64013927e-02
9.25483048e-01 -1.01702414e-01 4.01095152e-02 7.05330849e-01
-3.65752161e-01 2.02497140e-01 1.37578761e-02 -2.70189494e-01
-1.11341560e+00 7.17739642e-01 -1.09166810e-02 -3.82978886e-01
-8.39300096e-01 9.04701710e-01 2.47288942e-01 -4.92251128e-01
-9.88563709e-03 -2.23987311e-01 -2.08389699e-01 2.07501039e-01
2.06400931e-01 2.56768078e-01 6.18411005e-02 -4.34828937e-01
-3.68188024e-01 2.32404366e-01 -3.33336860e-01 4.17050481e-01
1.50386655e+00 2.23578125e-01 -7.89824799e-02 1.68679520e-01
1.14963233e+00 1.93242162e-01 -7.31432915e-01 -5.68760395e-01
5.47122955e-01 -6.27120137e-01 -3.28272358e-02 -5.53757489e-01
-1.09824681e+00 5.65334976e-01 -1.97800063e-03 4.24350917e-01
7.47174501e-01 3.20873708e-01 8.07504952e-01 4.45678055e-01
5.57903230e-01 -9.00977314e-01 -2.62436736e-02 2.07834676e-01
5.78287065e-01 -9.98077929e-01 8.63522738e-02 -6.72494948e-01
-7.77118146e-01 9.93598759e-01 4.93553817e-01 1.36228889e-01
5.60679615e-01 -1.59655154e-01 -3.81008744e-01 -5.85127413e-01
-6.52814031e-01 -5.08856416e-01 3.55188012e-01 8.95580649e-01
1.91494226e-01 3.32495898e-01 -1.68344423e-01 8.26943815e-01
-1.24638885e-01 -3.49180013e-01 3.46765935e-01 8.50242972e-01
-3.76899809e-01 -1.27889538e+00 1.10424280e-01 8.28251243e-01
5.14328256e-02 -2.26537168e-01 -4.93268400e-01 8.90784204e-01
9.03861523e-02 7.53506780e-01 -2.50911206e-01 -6.68591917e-01
3.71240497e-01 1.88293204e-01 3.36717606e-01 -8.87520730e-01
-2.49510393e-01 -7.68221557e-01 3.31532717e-01 -4.50370580e-01
-7.50957057e-02 -2.23273247e-01 -1.32237756e+00 -6.58924580e-01
-5.32061577e-01 3.74426693e-01 3.40301633e-01 5.90621352e-01
7.09847271e-01 5.01616776e-01 4.90230769e-01 -1.68356478e-01
-3.74668717e-01 -6.23492718e-01 -8.82240236e-01 7.90103137e-01
-2.88921803e-01 -1.09125066e+00 -7.40093738e-02 -1.34796098e-01]
|
[8.70962905883789, 7.875404357910156]
|
41c3a014-a631-412c-975f-822a7d1e0216
|
design-and-analysis-of-robust-deep-learning
|
2106.09664
| null |
https://arxiv.org/abs/2106.09664v1
|
https://arxiv.org/pdf/2106.09664v1.pdf
|
Design and Analysis of Robust Deep Learning Models for Stock Price Prediction
|
Building predictive models for robust and accurate prediction of stock prices and stock price movement is a challenging research problem to solve. The well-known efficient market hypothesis believes in the impossibility of accurate prediction of future stock prices in an efficient stock market as the stock prices are assumed to be purely stochastic. However, numerous works proposed by researchers have demonstrated that it is possible to predict future stock prices with a high level of precision using sophisticated algorithms, model architectures, and the selection of appropriate variables in the models. This chapter proposes a collection of predictive regression models built on deep learning architecture for robust and precise prediction of the future prices of a stock listed in the diversified sectors in the National Stock Exchange (NSE) of India. The Metastock tool is used to download the historical stock prices over a period of two years (2013- 2014) at 5 minutes intervals. While the records for the first year are used to train the models, the testing is carried out using the remaining records. The design approaches of all the models and their performance results are presented in detail. The models are also compared based on their execution time and accuracy of prediction.
|
['Sidra Mehtab', 'Jaydip Sen']
|
2021-06-17
| null | null | null | null |
['stock-price-prediction']
|
['time-series']
|
[-9.21772063e-01 -6.15453839e-01 -2.11733595e-01 -3.16450566e-01
-4.16917235e-01 -6.47860765e-01 4.23545241e-01 -1.75720915e-01
-2.01198295e-01 9.74207044e-01 -1.00031346e-02 -8.11839998e-01
-2.64174581e-01 -1.15479076e+00 -4.84670043e-01 -6.06317818e-01
-4.24314469e-01 5.95604897e-01 5.57404533e-02 -3.74653786e-01
6.55778587e-01 6.70867145e-01 -1.38571811e+00 -1.13822706e-01
2.37951443e-01 1.52208245e+00 4.39242907e-02 4.44592446e-01
-3.38902056e-01 1.26051116e+00 -6.02059722e-01 -3.33592772e-01
1.10970390e+00 6.11006841e-02 -3.34805667e-01 -4.29838389e-01
-1.26595959e-01 -8.43752682e-01 -2.25147575e-01 7.38963008e-01
1.33285120e-01 -1.61469907e-01 3.56391698e-01 -1.34889412e+00
-6.52599931e-01 7.70501792e-01 -2.69410968e-01 7.37055302e-01
-3.97281587e-01 -5.46151726e-03 1.23429692e+00 -8.27815831e-01
4.02303226e-02 6.08033538e-01 7.15273142e-01 -7.13635683e-02
-8.44433367e-01 -1.16431403e+00 -1.29724219e-01 1.55906782e-01
-1.17828953e+00 -8.02324563e-02 6.61179125e-01 -6.24289334e-01
1.29133749e+00 2.69643635e-01 1.03726447e+00 2.78274745e-01
6.67291820e-01 3.52216482e-01 1.17821646e+00 -6.70816749e-02
2.75268495e-01 2.67824769e-01 2.22232223e-01 -1.97620586e-01
6.26942813e-01 4.30764556e-01 -2.13597879e-01 -1.68728068e-01
9.22020972e-01 3.02738458e-01 2.16406241e-01 2.22667262e-01
-9.53494251e-01 1.09370625e+00 2.87266642e-01 3.88467491e-01
-1.06172907e+00 -6.84947073e-02 4.60735649e-01 8.01768243e-01
5.79254746e-01 3.91483307e-01 -1.16937900e+00 -2.65021771e-01
-1.56550753e+00 7.25267351e-01 1.00878727e+00 7.05073953e-01
2.64509767e-01 8.38866770e-01 2.31152982e-01 1.24534316e-01
2.91618824e-01 3.31970215e-01 8.53902876e-01 -5.97868860e-01
3.48780453e-01 5.80475092e-01 6.07695997e-01 -9.96679902e-01
-3.69283646e-01 -6.72248840e-01 -5.51174402e-01 4.93069977e-01
2.20048279e-01 -4.20297056e-01 -6.89796031e-01 8.94698739e-01
-2.23216161e-01 4.69254345e-01 2.96067685e-01 4.33358520e-01
3.63727003e-01 1.21242511e+00 3.77713926e-02 -5.07825077e-01
8.04145813e-01 -6.20646775e-01 -6.05466187e-01 7.06395954e-02
2.40642980e-01 -5.62039435e-01 1.57339528e-01 2.65553057e-01
-1.06415451e+00 -5.09661317e-01 -9.67048883e-01 3.43976229e-01
-6.98978782e-01 -4.03917618e-02 5.59171557e-01 3.00233722e-01
-9.08747852e-01 1.03478932e+00 -7.84673095e-01 6.03675961e-01
8.99194926e-02 6.22679055e-01 8.65265056e-02 9.08609629e-01
-1.31855154e+00 1.26997221e+00 6.08916044e-01 2.49293476e-01
-3.88468802e-01 -8.81299019e-01 -3.13776344e-01 2.81179130e-01
-2.98322052e-01 -2.00046953e-02 1.31761265e+00 -1.03906012e+00
-1.30751932e+00 3.61301750e-01 3.54272127e-01 -1.19989121e+00
5.69159806e-01 -7.29147643e-02 -6.13916874e-01 -4.76340234e-01
-1.13429219e-01 1.61176044e-02 3.14869404e-01 -4.47217673e-01
-1.09411705e+00 -2.17170089e-01 -1.62896976e-01 -5.93574978e-02
1.68236837e-01 1.23513468e-01 2.77431875e-01 -8.82469594e-01
6.42893091e-02 -7.73512244e-01 -1.95443645e-01 -8.08780909e-01
9.86070633e-02 -1.71841264e-01 6.78763270e-01 -1.16475821e+00
1.43124747e+00 -1.72594190e+00 -5.51299989e-01 4.54689324e-01
-3.41433883e-01 1.55506641e-01 4.80859041e-01 6.16428733e-01
-5.69473624e-01 1.67762861e-01 1.37301967e-01 2.27654397e-01
1.99809447e-01 1.34399936e-01 -1.04020274e+00 4.26276952e-01
4.17547226e-02 9.56729829e-01 -1.47669032e-01 6.85718432e-02
3.94912064e-01 2.09961474e-01 3.43758911e-02 9.05213952e-02
-2.66496301e-01 -2.06601005e-02 -3.82708222e-01 5.23131371e-01
7.99628854e-01 -1.93388388e-01 -1.22703612e-01 1.52181461e-01
-7.55157053e-01 5.68846524e-01 -1.36182523e+00 3.99375170e-01
-1.36769369e-01 5.32813132e-01 -4.91005749e-01 -8.72412860e-01
1.29511762e+00 5.28137267e-01 7.25400209e-01 -8.13594699e-01
7.09949955e-02 7.59486139e-01 1.38436735e-01 -5.96456043e-02
6.69234574e-01 -5.43806672e-01 1.44914806e-01 5.80121934e-01
-3.31087142e-01 1.37246788e-01 3.20643038e-01 -4.49830830e-01
4.58212227e-01 1.24753214e-01 3.35993052e-01 -3.00016135e-01
3.44360083e-01 1.49692580e-01 6.77730203e-01 2.76024848e-01
-2.12140419e-02 3.23127918e-02 2.97634989e-01 -1.20787406e+00
-1.36835527e+00 -6.07194185e-01 -3.67609322e-01 7.20915139e-01
-4.64175791e-01 2.50200838e-01 -1.04792625e-01 -2.85869595e-02
4.38269973e-01 1.03731167e+00 -5.28582811e-01 4.89114255e-01
-2.72324294e-01 -1.03071952e+00 -8.12268406e-02 6.03006840e-01
6.69362605e-01 -1.29927051e+00 -9.59943712e-01 5.63857079e-01
4.10913914e-01 -6.72886431e-01 2.12823436e-01 3.04322660e-01
-1.14146638e+00 -8.12881887e-01 -7.48958647e-01 -5.01198769e-01
8.69554132e-02 -1.78062886e-01 1.25419772e+00 7.30857998e-02
4.14237648e-01 -3.00987035e-01 -2.19788938e-03 -1.19582546e+00
-1.26514971e-01 6.96021393e-02 1.11749326e-03 -1.63172349e-01
7.78980672e-01 -3.89432192e-01 -5.72350085e-01 -8.09846371e-02
-6.06215477e-01 -3.45305577e-02 4.86203045e-01 4.71520662e-01
5.47147036e-01 5.92822492e-01 9.75671470e-01 -4.10057724e-01
5.57820380e-01 -8.24045002e-01 -1.54867744e+00 8.38570669e-02
-1.10861135e+00 -9.87850875e-02 4.54825312e-01 -2.94367820e-02
-6.78587139e-01 -1.61959808e-02 -1.29830554e-01 -3.36490273e-01
3.33351463e-01 9.94217336e-01 6.38988972e-01 1.36699080e-01
-1.76749617e-01 6.94249332e-01 -2.15367302e-02 -5.81568539e-01
-4.11461473e-01 5.67480803e-01 2.94918388e-01 1.06593333e-01
9.32566583e-01 1.29399404e-01 -1.18747659e-01 -3.82129937e-01
-4.91210550e-01 -3.50998193e-01 -8.13647509e-01 -1.53738394e-01
2.97127634e-01 -1.09136963e+00 -5.35971344e-01 7.55927920e-01
-6.21825099e-01 -2.56090034e-02 -2.90776759e-01 6.31131232e-01
-3.37019950e-01 -3.78684610e-01 -7.81001627e-01 -1.25020301e+00
-8.16637516e-01 -8.77390862e-01 3.87586713e-01 2.29666814e-01
-2.60268331e-01 -1.18518054e+00 2.05298781e-01 -1.36903599e-01
7.80638099e-01 3.94805163e-01 7.31169939e-01 -1.37072277e+00
-6.65179968e-01 -8.47712994e-01 -3.58114168e-02 6.03394508e-01
1.38673559e-01 2.88422376e-01 -6.17834806e-01 -1.58741534e-01
2.77407587e-01 1.84994072e-01 3.57486427e-01 8.32677901e-01
4.95735139e-01 -6.59621537e-01 2.53922760e-01 4.40342426e-01
1.84482884e+00 8.48996997e-01 6.08517349e-01 1.10750675e+00
-3.36580314e-02 3.55897605e-01 5.01764894e-01 6.96799874e-01
2.36939982e-01 9.38742459e-02 2.18979433e-01 2.35012412e-01
9.74623024e-01 -1.58335026e-02 3.46384943e-01 8.85325074e-01
-2.43634373e-01 3.02096277e-01 -8.87581289e-01 5.74647605e-01
-1.48693395e+00 -1.37531567e+00 -1.27081513e-01 2.19825339e+00
5.75242937e-01 5.13974607e-01 4.31519896e-01 5.14803052e-01
3.04193288e-01 2.24437729e-01 -5.82211673e-01 -3.73949438e-01
-8.10768008e-02 2.20885560e-01 1.23599100e+00 1.76183373e-01
-1.15104306e+00 6.41450405e-01 7.37811852e+00 1.45613164e-01
-1.44372404e+00 -4.45642471e-01 9.41830337e-01 -2.11247444e-01
-1.24404274e-01 -7.87968859e-02 -1.00481725e+00 8.85160565e-01
1.59831870e+00 -8.48142862e-01 7.49392211e-02 1.11733937e+00
5.19741535e-01 6.60921261e-02 -4.59253967e-01 6.57321870e-01
-6.16804302e-01 -1.92145443e+00 4.99184020e-02 2.17763662e-01
7.33440042e-01 3.03017586e-01 1.86295152e-01 4.89299506e-01
3.09844166e-01 -9.41515744e-01 1.01480663e+00 9.71923172e-01
9.85739306e-02 -1.17221546e+00 1.26051652e+00 5.14546633e-01
-1.17530346e+00 -5.69206178e-01 -4.59946334e-01 -6.11116230e-01
-7.92801753e-03 3.62214863e-01 -4.89452094e-01 3.41717303e-01
8.86019647e-01 6.70052588e-01 -1.43610105e-01 1.06677532e+00
4.12211239e-01 6.29891753e-01 -4.44439441e-01 -4.47133817e-02
6.58380747e-01 -5.60372412e-01 -1.05719030e-01 7.40784645e-01
7.49021292e-01 2.38665625e-01 -2.68098474e-01 7.44261622e-01
2.41805047e-01 2.02955410e-01 -4.95923221e-01 -8.64070281e-02
4.93841708e-01 6.59354389e-01 -4.20897961e-01 -3.75689298e-01
-8.52762938e-01 1.35381296e-01 -3.19649220e-01 7.85610452e-02
-6.14501953e-01 -1.08979426e-01 5.36135674e-01 3.68225694e-01
5.29693127e-01 -3.27748984e-01 -6.24571145e-01 -8.02430272e-01
-3.18018831e-02 -6.47302747e-01 2.70528883e-01 -6.64639771e-01
-1.17585754e+00 4.43781734e-01 4.16733474e-02 -1.35972416e+00
-6.77837312e-01 -8.29388678e-01 -8.67638528e-01 1.36881721e+00
-1.82542312e+00 -7.06544280e-01 3.67836893e-01 4.43029672e-01
6.30403101e-01 -8.74895871e-01 6.76358402e-01 2.03273669e-02
-5.40381610e-01 -2.53636569e-01 7.37428784e-01 5.38203478e-01
-2.64346778e-01 -1.26426005e+00 8.98890138e-01 7.68009186e-01
1.07768388e-03 3.74891311e-01 8.03884029e-01 -1.00803339e+00
-9.95820582e-01 -1.06703520e+00 1.22565985e+00 -1.12947129e-01
1.18640387e+00 3.36996168e-01 -1.07564986e+00 1.15836406e+00
1.95093393e-01 -1.91574499e-01 5.73467612e-01 -4.78500873e-01
7.94132501e-02 -4.18450326e-01 -9.51048076e-01 5.22329919e-02
-3.69930148e-01 -1.00267775e-01 -8.15421939e-01 4.09070998e-02
2.13080227e-01 -2.68168062e-01 -1.08541822e+00 1.54797226e-01
8.00476849e-01 -1.10267389e+00 7.85005927e-01 -4.53238159e-01
-1.00292107e-02 -9.42809787e-03 -1.36032462e-01 -1.08622348e+00
-4.04605538e-01 -4.83168662e-01 -8.10665563e-02 9.20606732e-01
6.61038458e-01 -9.33688521e-01 7.28061080e-01 1.27235007e+00
3.45384687e-01 -6.72465861e-01 -1.04202199e+00 -7.53996491e-01
3.96696180e-01 -4.53435868e-01 1.34363139e+00 7.86151409e-01
-3.62068802e-01 -2.72426575e-01 -5.12766719e-01 2.58129597e-01
3.36481869e-01 4.75195497e-01 4.79890794e-01 -1.52699053e+00
1.38927668e-01 -5.37753582e-01 -2.81241298e-01 -4.36863810e-01
5.14990166e-02 -2.46226132e-01 -5.59908032e-01 -1.40560126e+00
-2.68070757e-01 -4.92994070e-01 -8.22959900e-01 2.12814763e-01
3.75349075e-01 -2.48521701e-01 3.11668724e-01 7.96727359e-01
3.47559482e-01 9.02948603e-02 5.92591941e-01 7.49762878e-02
-1.84933513e-01 6.62414551e-01 -3.31768036e-01 6.04728580e-01
1.13194084e+00 -1.40708327e-01 -2.32071757e-01 -3.11954767e-02
4.87742841e-01 2.02661812e-01 1.87822402e-01 -8.79174590e-01
2.01911077e-01 -4.40136552e-01 7.38868773e-01 -1.36825335e+00
1.17004581e-01 -1.04504669e+00 8.43858123e-01 6.37467027e-01
-1.61645174e-01 9.50131178e-01 3.79046410e-01 1.07784122e-01
-5.86555660e-01 -3.64226222e-01 6.89289451e-01 -4.86864179e-01
-1.00388622e+00 4.96038556e-01 -2.77183712e-01 -5.47751904e-01
1.39292669e+00 -3.63992929e-01 1.51247308e-01 -5.79320848e-01
-5.52210689e-01 2.02606454e-01 -5.83018325e-02 3.60566109e-01
4.62733746e-01 -1.27394235e+00 -8.88836563e-01 3.25768322e-01
-4.67457324e-01 -4.02024448e-01 -9.34338644e-02 4.15184289e-01
-1.06546617e+00 9.67969596e-01 -4.47187513e-01 1.16870493e-01
-6.53540075e-01 5.50840974e-01 7.97167301e-01 -4.58893955e-01
-5.94607770e-01 5.21604300e-01 -4.51671124e-01 2.30979189e-01
6.40906766e-02 -5.31331122e-01 -4.88909751e-01 5.24420261e-01
8.08753073e-01 3.76268357e-01 2.28668585e-01 -8.40482056e-01
-8.27247798e-02 4.67651069e-01 2.29949597e-02 1.93752293e-02
2.10149050e+00 -8.02478641e-02 2.92067528e-02 9.02556181e-01
9.16663110e-01 -4.87915784e-01 -1.31625259e+00 -6.55528903e-02
6.60040736e-01 -2.14348808e-01 3.96401793e-01 -8.21570218e-01
-1.44102597e+00 5.06740630e-01 5.78097105e-01 6.16064847e-01
9.76655662e-01 -7.11126029e-01 9.19797242e-01 8.46815631e-02
2.78840244e-01 -1.32577169e+00 -9.99907672e-01 5.49111068e-01
9.50327516e-01 -1.21748435e+00 1.20908104e-01 5.32346010e-01
-7.60719061e-01 1.37859786e+00 -1.67106111e-02 -5.67375243e-01
1.28318012e+00 4.41959143e-01 2.27041870e-01 -6.28110543e-02
-1.05407643e+00 3.04422557e-01 7.65325949e-02 1.46978512e-01
4.25703138e-01 2.01596409e-01 -6.44677952e-02 8.03721666e-01
-7.44554996e-01 4.12913769e-01 5.61748326e-01 9.29013193e-01
-4.38133568e-01 -5.59911609e-01 -4.36747134e-01 8.62454355e-01
-1.06217909e+00 -2.06543356e-01 3.13244313e-02 1.21875024e+00
-3.12650442e-01 5.25236309e-01 6.83328569e-01 -1.44766703e-01
3.30123037e-01 3.73722732e-01 -3.75915319e-01 -1.43351868e-01
-8.46794248e-01 1.61713377e-01 -2.03864858e-01 -4.10844311e-02
-2.50388891e-01 -8.55521262e-01 -1.21568346e+00 -8.39141309e-01
-2.89413389e-02 3.19551706e-01 6.79533601e-01 9.39175010e-01
6.95534647e-02 1.33864164e-01 1.07095110e+00 -7.89593875e-01
-1.05838823e+00 -1.04616416e+00 -1.33665442e+00 6.83845058e-02
5.82434297e-01 -3.76646310e-01 -5.82683206e-01 8.40266794e-03]
|
[4.449654579162598, 4.24291467666626]
|
43439e2d-93a7-4597-89f4-bd79e3823fff
|
nsp-bert-a-prompt-based-few-shot-learner
| null | null |
https://aclanthology.org/2022.coling-1.286
|
https://aclanthology.org/2022.coling-1.286.pdf
|
NSP-BERT: A Prompt-based Few-Shot Learner through an Original Pre-training Task —— Next Sentence Prediction
|
Using prompts to utilize language models to perform various downstream tasks, also known as prompt-based learning or prompt-learning, has lately gained significant success in comparison to the pre-train and fine-tune paradigm. Nonetheless, virtually most prompt-based methods are token-level such as PET based on mask language model (MLM). In this paper, we attempt to accomplish several NLP tasks in the zero-shot and few-shot scenarios using a BERT original pre-training task abandoned by RoBERTa and other models——Next Sentence Prediction (NSP). Unlike token-level techniques, our sentence-level prompt-based method NSP-BERT does not need to fix the length of the prompt or the position to be predicted, allowing it to handle tasks such as entity linking with ease. NSP-BERT can be applied to a variety of tasks based on its properties. We present an NSP-tuning approach with binary cross-entropy loss for single-sentence classification tasks that is competitive compared to PET and EFL. By continuing to train BERT on RoBERTa’s corpus, the model’s performance improved significantly, which indicates that the pre-training corpus is another important determinant of few-shot besides model size and prompt method.
|
['Hangping Qiu', 'Chao Hao', 'Yu Zheng', 'Yi Sun']
| null | null | null | null |
coling-2022-10
|
['sentence-classification']
|
['natural-language-processing']
|
[ 1.36961147e-01 3.03491712e-01 -3.62872392e-01 -5.20321369e-01
-1.24423647e+00 -5.24962068e-01 7.59076118e-01 6.31585121e-01
-9.29448247e-01 9.21304822e-01 2.63257951e-01 -5.40792525e-01
-8.76451880e-02 -5.12483418e-01 -5.08302450e-01 -2.87384778e-01
-6.27153218e-02 5.55946767e-01 4.14283603e-01 -4.08551514e-01
3.34615409e-01 2.19478473e-01 -1.09858608e+00 4.08660054e-01
7.70451427e-01 5.01074970e-01 5.27189076e-01 7.36523867e-01
-4.80855376e-01 8.00772846e-01 -6.07457519e-01 -5.60318649e-01
6.34225132e-03 -3.47641110e-01 -1.01398075e+00 -6.48423135e-01
1.73334152e-01 -5.93334697e-02 -1.54864222e-01 5.28427064e-01
8.09625745e-01 5.03971338e-01 5.70412576e-01 -8.56168449e-01
-2.16493100e-01 9.09724653e-01 -3.42459410e-01 5.60679376e-01
4.07784581e-01 1.87374860e-01 1.27353919e+00 -8.65084589e-01
5.89574754e-01 1.26991534e+00 8.95551860e-01 6.50980294e-01
-1.31402326e+00 -4.17877376e-01 2.69232154e-01 2.29425162e-01
-9.50403512e-01 -4.77697551e-01 5.52734971e-01 -4.01078194e-01
1.54756141e+00 1.13701649e-01 2.33041108e-01 1.03315568e+00
2.98533291e-01 9.91073191e-01 1.14889026e+00 -9.62691367e-01
1.36810720e-01 3.35850954e-01 3.26109290e-01 5.15407920e-01
-2.20416203e-01 1.83736235e-01 -6.82608426e-01 -1.18392304e-01
1.69062167e-01 -4.72466558e-01 5.61652891e-02 1.65906534e-01
-1.13511300e+00 1.03812289e+00 1.41548187e-01 5.76312542e-01
-2.89706290e-01 -4.76071564e-03 7.38774300e-01 4.89981353e-01
6.75510645e-01 9.80396926e-01 -9.42355573e-01 -5.04584372e-01
-1.34435081e+00 1.54611602e-01 1.00613999e+00 7.65584826e-01
4.36059833e-01 -1.28232881e-01 -8.21788669e-01 8.01718771e-01
1.25868201e-01 -4.50320914e-02 4.94882137e-01 -5.63218832e-01
7.21264482e-01 1.56578898e-01 6.54771626e-02 -4.19874638e-01
-6.29107714e-01 -3.45370919e-01 -4.34407830e-01 -2.16147348e-01
5.56151986e-01 -4.53549147e-01 -8.03870499e-01 1.91033888e+00
1.74697727e-01 1.45914018e-01 1.37636643e-02 3.39285284e-01
7.20596731e-01 8.88489962e-01 6.46649182e-01 -4.46129441e-01
1.29527092e+00 -1.04522383e+00 -5.97146451e-01 -4.32191998e-01
1.22732687e+00 -8.48701596e-01 1.22295368e+00 1.44392341e-01
-1.07516944e+00 -6.07015789e-01 -7.24278152e-01 -1.43951014e-01
-3.01723123e-01 -1.79861575e-01 5.34226358e-01 4.24573570e-01
-1.16267133e+00 7.74941146e-01 -6.12247109e-01 -5.18553972e-01
1.97099730e-01 3.90718579e-01 -1.54273465e-01 9.80957523e-02
-1.65460336e+00 1.33470607e+00 6.95152402e-01 -4.77692306e-01
-4.17160690e-01 -8.34699392e-01 -7.75350630e-01 3.96593332e-01
2.83776790e-01 -6.85529888e-01 1.59021139e+00 -5.51215708e-01
-1.50893497e+00 8.19670379e-01 -2.43716538e-01 -7.43269444e-01
2.55148917e-01 -2.17480481e-01 -1.59027189e-01 1.18608763e-02
2.16424957e-01 8.92834544e-01 5.93302071e-01 -7.93699324e-01
-6.96288705e-01 1.89418584e-01 9.31149051e-02 2.70912260e-01
-3.63075048e-01 3.92773271e-01 1.07314169e-01 -5.06475985e-01
-4.72745240e-01 -5.03795028e-01 -3.85566682e-01 -3.81756485e-01
-3.27632606e-01 -6.76347494e-01 4.36786234e-01 -4.77378398e-01
1.41907597e+00 -2.07643843e+00 -3.50921988e-01 -2.38403499e-01
-8.46627206e-02 7.36619174e-01 -5.21543205e-01 8.66863132e-01
-1.46732181e-01 2.17517868e-01 -7.23517537e-02 -5.43082416e-01
-4.77899574e-02 -1.84831768e-02 -2.59225935e-01 4.13874090e-02
4.37956154e-01 1.08137906e+00 -1.14070237e+00 -7.57949352e-01
1.43950984e-01 1.96851760e-01 -4.98970449e-01 4.14552778e-01
-3.35774571e-01 3.53895247e-01 -2.90332139e-01 1.18721031e-01
1.43458813e-01 -1.86710894e-01 -6.59312773e-03 1.73892140e-01
-2.66651958e-01 9.06317949e-01 -7.11623490e-01 1.59786630e+00
-6.59485340e-01 6.29387021e-01 -4.83729644e-03 -1.12071562e+00
1.05604601e+00 6.35215461e-01 3.92797470e-01 -7.80697525e-01
-1.07762873e-01 1.60916224e-01 1.03658862e-01 -6.73857570e-01
6.06090069e-01 -5.17171443e-01 -2.24200606e-01 4.18633074e-01
2.36327186e-01 1.13071306e-02 4.32984799e-01 3.76407802e-01
1.39789176e+00 1.60045370e-01 6.24574065e-01 -1.80154145e-01
4.00535703e-01 1.29230186e-01 4.90791172e-01 1.08351040e+00
-3.21608514e-01 2.99393207e-01 4.28255290e-01 4.86252457e-02
-9.86182213e-01 -8.16567123e-01 -2.04382896e-01 1.42020094e+00
-2.99739212e-01 -4.44550097e-01 -6.18368447e-01 -9.33887005e-01
-6.84837922e-02 1.31146133e+00 -1.80408150e-01 -1.56128779e-01
-6.50812566e-01 -5.17402232e-01 4.74727303e-01 4.91730273e-01
1.75686359e-01 -1.43952191e+00 -3.94504011e-01 7.38912404e-01
-2.39864185e-01 -9.54217672e-01 -6.15447760e-01 7.97554910e-01
-7.57513702e-01 -6.51445448e-01 -5.99453032e-01 -8.68477285e-01
3.89663190e-01 -8.40148479e-02 1.29767942e+00 -2.81677544e-01
-9.37933475e-02 3.24582517e-01 -5.99209607e-01 -3.41531336e-01
-5.40099680e-01 4.29411381e-01 -7.73467347e-02 -3.40509266e-01
6.24433458e-01 -4.19720829e-01 -2.44372740e-01 -1.21241815e-01
-5.34135818e-01 -8.03653970e-02 7.39284337e-01 1.11185527e+00
2.33232573e-01 -1.84290975e-01 1.18402088e+00 -1.17847657e+00
9.35542464e-01 -4.45512086e-01 -2.43794352e-01 4.36085701e-01
-6.74023092e-01 1.32462621e-01 7.31488466e-01 -4.73226100e-01
-1.11643445e+00 -6.97612762e-02 -4.96228218e-01 1.64504871e-02
-3.17324430e-01 8.17901254e-01 6.86605573e-02 2.41164207e-01
6.29583240e-01 7.86557272e-02 -2.62000203e-01 -5.51411390e-01
4.25658017e-01 6.61631823e-01 2.52801865e-01 -5.40104568e-01
6.39114201e-01 -3.30184281e-01 -1.77026927e-01 -5.94934106e-01
-1.29698169e+00 -7.57075191e-01 -8.35030019e-01 -4.05027457e-02
8.34030092e-01 -8.11423123e-01 -5.75908303e-01 6.26198202e-02
-1.47207236e+00 -6.47838891e-01 -5.76309860e-01 4.07293737e-01
-5.47898829e-01 2.24312410e-01 -8.00934553e-01 -1.02755976e+00
-5.78576922e-01 -5.88190556e-01 8.68736506e-01 1.74376905e-01
-5.06185591e-01 -1.38292515e+00 8.50885659e-02 1.77630842e-01
5.72598398e-01 -1.39581636e-01 1.09795713e+00 -1.30656254e+00
-1.03459969e-01 -1.84870377e-01 -4.43558171e-02 3.40686291e-01
-1.17741890e-01 -4.09140289e-01 -8.76009762e-01 -2.89277971e-01
1.20812178e-01 -4.86880571e-01 9.36644077e-01 3.83115083e-01
6.65171206e-01 -3.43492538e-01 -3.55057567e-01 1.72305435e-01
1.30947649e+00 1.94391921e-01 2.78965682e-01 5.04497588e-01
3.75632048e-01 7.93829083e-01 9.44855988e-01 2.62899339e-01
4.60410833e-01 6.75774693e-01 -1.83024853e-01 -9.35140997e-02
-1.18007615e-01 -4.39753503e-01 5.55610299e-01 1.03110409e+00
4.49693769e-01 -3.80710393e-01 -9.79327381e-01 4.26046550e-01
-1.87096751e+00 -1.21027076e+00 -9.13973823e-02 1.99253654e+00
1.26972306e+00 5.59479237e-01 6.01298474e-02 3.41484547e-02
6.34103894e-01 1.96775973e-01 -2.93951243e-01 -8.67120147e-01
-1.62890349e-02 2.75931209e-01 2.81092614e-01 6.40726626e-01
-1.04691958e+00 1.13423753e+00 6.36590862e+00 1.06894445e+00
-9.82212663e-01 3.95266712e-01 6.18677378e-01 -2.76210774e-02
-1.89165264e-01 1.11652553e-01 -1.31311357e+00 5.18051803e-01
1.50769818e+00 -4.56788212e-01 4.59020548e-02 6.90723538e-01
5.35170734e-01 -2.71451503e-01 -1.37709904e+00 6.30645514e-01
-7.09322616e-02 -1.21149313e+00 -2.89110392e-01 -4.09564316e-01
5.67712069e-01 6.01531006e-02 -3.05314749e-01 1.02744389e+00
2.89502829e-01 -8.02716851e-01 6.08765125e-01 3.99004132e-01
8.49124730e-01 -5.74058115e-01 7.65596092e-01 8.27705801e-01
-1.03298008e+00 -1.40206859e-01 -2.41001353e-01 -2.11779535e-01
5.96914947e-01 7.34271526e-01 -1.29434359e+00 4.41071481e-01
1.92544311e-01 4.17687416e-01 -4.01591122e-01 1.23294866e+00
-3.07028294e-01 1.12136436e+00 -2.39932552e-01 -3.95035386e-01
3.56732070e-01 2.49461174e-01 6.17147207e-01 1.71121454e+00
6.69044405e-02 1.77621067e-01 4.38993871e-01 4.97115642e-01
-2.32470017e-02 3.22398692e-01 -5.36325872e-01 -1.05093159e-01
8.62610221e-01 1.16130149e+00 -5.29661477e-01 -3.84910077e-01
-5.68373382e-01 7.43281126e-01 6.37848854e-01 1.77648187e-01
-6.53398633e-01 -6.54565752e-01 2.85358410e-02 2.25904331e-01
3.46378654e-01 -2.08310768e-01 -3.81193072e-01 -8.25275898e-01
-1.46734983e-01 -7.17896581e-01 3.68932843e-01 -4.84148443e-01
-1.46184325e+00 6.15183890e-01 2.05473915e-01 -9.74968553e-01
-4.60863709e-01 -3.60352784e-01 -9.67551589e-01 1.00326777e+00
-1.69428587e+00 -9.08790767e-01 5.08219957e-01 2.55911052e-01
8.43166411e-01 -7.67258108e-02 9.66214776e-01 3.56265247e-01
-4.15902913e-01 8.05107772e-01 6.95632920e-02 1.14615850e-01
1.05955708e+00 -1.38892889e+00 4.63456422e-01 7.50440836e-01
1.45605430e-01 5.68004370e-01 6.58157766e-01 -5.37855148e-01
-7.12529302e-01 -1.07389379e+00 1.85219812e+00 -4.17882621e-01
6.59600496e-01 -4.70803976e-01 -9.02929366e-01 5.41760683e-01
3.32391560e-01 -2.98982829e-01 6.68891668e-01 5.64274192e-01
-2.09527649e-02 -1.77600816e-01 -9.49682653e-01 4.00563270e-01
7.29751408e-01 -5.44724643e-01 -9.97421801e-01 6.30260825e-01
9.39866841e-01 -2.23415360e-01 -6.67629540e-01 2.08036259e-01
-7.73335993e-02 -6.52126372e-01 6.94725275e-01 -8.28656733e-01
2.80517697e-01 3.34970832e-01 2.55127430e-01 -1.40827441e+00
-5.65448105e-01 -8.52652550e-01 -1.46643341e-01 1.69375777e+00
8.13541591e-01 -6.06379449e-01 5.62188983e-01 7.29158700e-01
-3.31306845e-01 -1.01229858e+00 -1.00885272e+00 -1.07642090e+00
1.98424131e-01 -2.74123162e-01 2.22522259e-01 8.26693892e-01
2.01367378e-01 9.85648811e-01 -2.08571345e-01 -3.82131130e-01
2.74826795e-01 -2.28277922e-01 4.18588340e-01 -1.28229618e+00
-4.32243794e-01 -3.88082683e-01 9.79433283e-02 -1.22760487e+00
3.48869890e-01 -1.23193550e+00 4.65751886e-01 -1.59024000e+00
1.89892381e-01 -5.79803467e-01 -4.44160670e-01 7.48623669e-01
-5.21501422e-01 -3.24897081e-01 3.18790644e-01 -7.11111072e-03
-7.42993653e-01 6.20083094e-01 1.04719460e+00 1.62624083e-02
-5.01340806e-01 1.15645878e-01 -5.42572081e-01 4.02608871e-01
7.25153089e-01 -8.24945271e-01 -2.07985684e-01 -2.44831949e-01
1.47284847e-02 3.11544955e-01 -6.93481714e-02 -6.75794840e-01
5.31261921e-01 -4.77579385e-02 -1.71664897e-02 -5.64345121e-01
2.50648171e-01 -3.23217809e-01 -4.77215677e-01 4.71152514e-01
-8.79788935e-01 4.12065722e-02 1.55330211e-01 4.77066755e-01
-1.75882816e-01 -7.41600633e-01 8.11319649e-01 -2.24555150e-01
-8.08063030e-01 7.90784135e-02 -4.56693113e-01 4.83461797e-01
9.26880121e-01 -4.23245095e-02 -2.49865428e-01 -2.05908567e-01
-7.53577411e-01 3.70163918e-01 -2.46092856e-01 2.63000548e-01
3.15635234e-01 -9.35088038e-01 -9.52745020e-01 -1.79767162e-01
-4.88346219e-02 -6.28521442e-02 -6.32563513e-03 1.08121800e+00
-4.54830155e-02 8.47036064e-01 2.46821925e-01 -4.51805443e-01
-1.11119580e+00 6.65006578e-01 -7.39591718e-02 -9.97876048e-01
-6.77054703e-01 1.10949850e+00 -1.20244592e-01 -4.69143420e-01
2.93606579e-01 2.02137157e-01 -2.25887656e-01 2.29428619e-01
3.40968341e-01 9.03014168e-02 4.45395969e-02 -1.49994984e-01
-3.25410306e-01 -4.48740534e-02 -4.32931751e-01 -2.59250164e-01
1.36324871e+00 -2.20345005e-01 1.21099375e-01 6.95486963e-01
1.06631577e+00 -1.71321020e-01 -1.17809582e+00 -4.19242859e-01
6.52399600e-01 -1.40919685e-01 7.15396879e-03 -1.12796223e+00
-3.15399170e-01 9.89228249e-01 1.75659239e-01 1.97429821e-01
8.62371445e-01 3.08147445e-02 8.93584251e-01 6.32418573e-01
3.77231807e-01 -1.28987277e+00 1.84669346e-01 9.29156423e-01
5.36282361e-01 -1.27836549e+00 -2.68199444e-01 -1.97904468e-01
-6.68664336e-01 9.95358050e-01 6.87526703e-01 1.56922787e-01
5.95100284e-01 4.47602153e-01 1.02842867e-01 7.40100369e-02
-1.16927111e+00 -2.26522774e-01 2.14169070e-01 5.51412165e-01
8.15567076e-01 -1.51409224e-01 -6.98305190e-01 6.43067062e-01
-2.24301249e-01 -9.05294940e-02 1.80876479e-01 8.73550355e-01
-8.03116620e-01 -1.44740212e+00 -2.52583567e-02 7.14286864e-01
-4.89493817e-01 -5.60776412e-01 -1.39335871e-01 6.88890338e-01
-1.09896399e-01 1.13981915e+00 -9.67634618e-02 -2.49487802e-01
3.53209406e-01 5.84506452e-01 2.81178445e-01 -1.12824547e+00
-1.08318114e+00 -1.01672731e-01 5.66642821e-01 -1.78196236e-01
-2.32172370e-01 -7.84234643e-01 -1.22282398e+00 -1.88471571e-01
-6.98612273e-01 5.43136895e-01 4.80343968e-01 1.28169119e+00
1.94846109e-01 5.30748248e-01 6.82119429e-01 -7.73619175e-01
-1.02495968e+00 -1.19255459e+00 -3.61647010e-01 2.19169125e-01
2.15221956e-01 -3.67232561e-01 -2.29349360e-01 -9.04622898e-02]
|
[10.6441650390625, 8.709274291992188]
|
d1433860-dfe3-4f24-8b05-ac00d6c98aeb
|
core-text-improving-scene-text-detection-with
|
2112.07513
| null |
https://arxiv.org/abs/2112.07513v1
|
https://arxiv.org/pdf/2112.07513v1.pdf
|
CORE-Text: Improving Scene Text Detection with Contrastive Relational Reasoning
|
Localizing text instances in natural scenes is regarded as a fundamental challenge in computer vision. Nevertheless, owing to the extremely varied aspect ratios and scales of text instances in real scenes, most conventional text detectors suffer from the sub-text problem that only localizes the fragments of text instance (i.e., sub-texts). In this work, we quantitatively analyze the sub-text problem and present a simple yet effective design, COntrastive RElation (CORE) module, to mitigate that issue. CORE first leverages a vanilla relation block to model the relations among all text proposals (sub-texts of multiple text instances) and further enhances relational reasoning via instance-level sub-text discrimination in a contrastive manner. Such way naturally learns instance-aware representations of text proposals and thus facilitates scene text detection. We integrate the CORE module into a two-stage text detector of Mask R-CNN and devise our text detector CORE-Text. Extensive experiments on four benchmarks demonstrate the superiority of CORE-Text. Code is available: \url{https://github.com/jylins/CORE-Text}.
|
['Ting Yao', 'Hongyang Chao', 'Xuehang Yang', 'Rongfeng Lai', 'Yingwei Pan', 'Jingyang Lin']
|
2021-12-14
| null | null | null | null |
['scene-text-detection', 'relational-reasoning']
|
['computer-vision', 'natural-language-processing']
|
[ 2.88151115e-01 5.81309721e-02 -1.37790591e-01 -3.58068198e-01
-7.83234715e-01 -3.89095902e-01 1.00976646e+00 1.86797399e-02
-1.57810852e-01 5.35671823e-02 1.86847135e-01 -2.57945329e-01
1.52577505e-01 -6.92564726e-01 -6.44689679e-01 -6.33734345e-01
5.82812726e-01 5.72610676e-01 5.72476327e-01 -8.76148194e-02
2.48785064e-01 2.84363210e-01 -1.35947347e+00 6.26503348e-01
7.56121635e-01 9.17395949e-01 3.66352767e-01 4.61455584e-01
-3.77538502e-01 8.79639864e-01 -6.08936667e-01 -4.95987922e-01
4.63324696e-01 -2.49908432e-01 -5.19274235e-01 2.20928982e-01
8.24760914e-01 -4.62043166e-01 -7.90036023e-01 1.13459003e+00
3.37587893e-01 1.99080929e-01 6.37742162e-01 -9.52366948e-01
-7.75788128e-01 8.24685991e-01 -1.02013445e+00 3.78299892e-01
1.63742244e-01 1.41031116e-01 1.46671927e+00 -1.41716146e+00
4.72465247e-01 1.37646472e+00 4.38493639e-01 2.30295494e-01
-1.14566731e+00 -7.12546587e-01 5.04483819e-01 5.72258234e-02
-1.63445103e+00 -5.32678843e-01 8.18298876e-01 -2.35563427e-01
1.04821062e+00 3.49633247e-01 3.73515993e-01 1.02029812e+00
2.66473532e-01 1.24291766e+00 8.17232192e-01 -2.52237827e-01
-1.83790773e-02 3.60683948e-02 9.97730568e-02 8.79597127e-01
3.64241153e-01 -4.60705787e-01 -7.25094378e-01 2.16947883e-01
8.13556612e-01 3.73035401e-01 -2.65663266e-01 -2.73234278e-01
-1.32347119e+00 6.17529690e-01 5.75275540e-01 2.25993946e-01
-1.57490864e-01 5.68476133e-02 4.73525822e-01 5.50792133e-03
5.37985742e-01 -1.48247238e-02 -2.81284094e-01 3.88958186e-01
-9.54216719e-01 1.10344827e-01 5.57752252e-01 1.17081499e+00
7.15557516e-01 -1.98874369e-01 -4.56253052e-01 1.04039752e+00
2.93943822e-01 4.99723613e-01 4.08679992e-01 -1.78670406e-01
9.99133170e-01 1.09096813e+00 -4.64718044e-01 -1.00802612e+00
-2.87259072e-01 -4.93554085e-01 -8.21084738e-01 -1.47772327e-01
1.43975794e-01 2.86723733e-01 -8.63983154e-01 1.08013213e+00
6.46441936e-01 1.75964721e-02 -3.12473804e-01 1.00832748e+00
1.23304021e+00 6.09965801e-01 6.29186630e-02 7.70327523e-02
1.63510084e+00 -1.30065262e+00 -4.60675925e-01 -6.05255544e-01
6.10957384e-01 -9.40512538e-01 1.17301905e+00 1.90270811e-01
-1.02597499e+00 -3.96344304e-01 -8.37455928e-01 -6.12928867e-01
-4.58983749e-01 5.32109380e-01 4.23714966e-01 2.50645787e-01
-8.83095264e-01 1.07573867e-01 -6.03978693e-01 -4.65933949e-01
8.11785996e-01 4.23571497e-01 -1.97480574e-01 -2.43034009e-02
-7.46889710e-01 5.30844629e-01 3.30248266e-01 3.11766118e-01
-6.38441145e-01 -3.47071558e-01 -8.55693460e-01 8.51004571e-02
8.70714962e-01 -7.03699410e-01 1.12691104e+00 -9.12759840e-01
-1.09663701e+00 1.13887882e+00 -4.45502967e-01 -2.21068054e-01
6.46386206e-01 -2.97455370e-01 -4.73661311e-02 3.13140333e-01
3.77318293e-01 6.28733516e-01 1.29659283e+00 -1.16273963e+00
-6.97316706e-01 -5.64599395e-01 1.22745901e-01 5.68160832e-01
-2.66480118e-01 5.30045740e-02 -8.60312045e-01 -8.75358760e-01
5.08491457e-01 -6.09560907e-01 4.30120230e-02 2.29450628e-01
-7.84082472e-01 -5.60716450e-01 9.79834318e-01 -2.95224160e-01
1.17309701e+00 -2.08728552e+00 -3.57506126e-02 -2.57347614e-01
7.20228791e-01 9.77857113e-02 2.90365014e-02 3.97137523e-01
-6.61214963e-02 -6.90651238e-02 2.13853177e-02 -7.33664155e-01
1.19793139e-01 3.21627781e-02 -7.17137337e-01 5.93093336e-01
4.02792364e-01 1.19858980e+00 -5.11469960e-01 -9.29524183e-01
6.99320018e-01 3.35467368e-01 -1.99212119e-01 -5.04856035e-02
-6.14778578e-01 4.32711393e-02 -8.97801876e-01 9.39140916e-01
7.92899966e-01 -6.01307213e-01 -3.69135737e-02 -2.68264294e-01
-6.85197338e-02 4.11737502e-01 -1.07897520e+00 1.52369821e+00
-7.51756355e-02 8.34724307e-01 6.15755841e-02 -1.12006807e+00
8.29734862e-01 -6.98608458e-02 2.53516793e-01 -9.41568315e-01
4.19100910e-01 -1.16939418e-01 -2.31153443e-01 -6.36423409e-01
7.80358434e-01 7.85126388e-02 1.16402455e-01 3.67354691e-01
-1.35715887e-01 -5.14888167e-02 -4.93190587e-02 4.59562749e-01
1.01478648e+00 3.42248172e-01 4.59556103e-01 -1.39848098e-01
5.37028372e-01 -1.23583898e-01 1.73578113e-01 9.68177855e-01
-3.46847147e-01 7.36905992e-01 6.39945567e-01 -4.22921598e-01
-8.51348162e-01 -9.06555831e-01 -3.66391361e-01 1.46676528e+00
4.88543391e-01 -6.97116494e-01 -3.77685428e-01 -8.69129062e-01
7.16889575e-02 4.51729476e-01 -8.30784142e-01 2.29439020e-01
-6.13995552e-01 -6.42044902e-01 4.69071656e-01 6.08685195e-01
6.53765738e-01 -1.01886475e+00 -5.21021247e-01 -2.57323921e-01
-1.52895942e-01 -1.43262887e+00 -5.35309553e-01 4.13973600e-01
-7.29587615e-01 -7.62941897e-01 -3.10970485e-01 -8.74749422e-01
6.69335783e-01 8.17110240e-01 9.30558324e-01 2.22608477e-01
-6.43577933e-01 3.40382248e-01 -3.78946483e-01 -3.28178048e-01
6.42087385e-02 -9.57007706e-02 -4.43477631e-01 1.38902903e-01
6.25373423e-01 -2.42258847e-01 -7.71993220e-01 3.90096843e-01
-8.69829535e-01 3.66625518e-01 7.78944790e-01 8.28125060e-01
6.80781245e-01 2.56118566e-01 1.90793797e-01 -8.69035542e-01
2.86429107e-01 -2.34431833e-01 -6.77605033e-01 2.50614256e-01
-2.29986668e-01 -1.83599606e-01 7.35641420e-01 -3.93161744e-01
-1.03315508e+00 1.26688644e-01 1.78980306e-01 -4.85643715e-01
-2.02646121e-01 1.82336554e-01 -1.48441717e-01 2.43887931e-01
5.12594223e-01 6.42651498e-01 -4.34584320e-01 -2.16539968e-02
3.57453734e-01 6.26358569e-01 2.45495379e-01 -6.39723897e-01
9.43968952e-01 8.93530965e-01 -1.74608584e-02 -1.07448065e+00
-1.10993326e+00 -8.17998528e-01 -6.49217844e-01 -1.77556008e-01
7.43064523e-01 -1.21157062e+00 -6.34001315e-01 2.91621864e-01
-1.00187755e+00 -5.12826443e-01 -1.68675691e-01 5.74231185e-02
-3.10428381e-01 4.35254484e-01 -4.77430105e-01 -6.92329526e-01
-4.54344124e-01 -1.04213691e+00 1.80146861e+00 1.23405159e-01
2.62642980e-01 -8.11617792e-01 -4.07639593e-01 6.66023195e-01
-1.97360143e-02 -1.42162651e-01 6.88496470e-01 -6.75615668e-01
-9.86406565e-01 -1.57517761e-01 -7.38264441e-01 -1.62438527e-01
-9.93850008e-02 -8.59297216e-02 -1.29652715e+00 -6.81271479e-02
-7.02725649e-02 -3.45648885e-01 1.24766099e+00 3.93714041e-01
1.19691563e+00 -6.02650009e-02 -4.68386799e-01 5.38427711e-01
1.37319076e+00 -1.94981188e-01 4.92791057e-01 2.61307269e-01
9.88685787e-01 6.08030915e-01 7.46137142e-01 4.52835292e-01
4.83413905e-01 5.91513336e-01 4.02800590e-01 -2.15107843e-01
-2.49807477e-01 -2.36541837e-01 1.84778333e-01 4.79207456e-01
2.62682378e-01 -3.72097820e-01 -8.62583578e-01 3.41887027e-01
-2.10126758e+00 -8.33604872e-01 -3.81829500e-01 1.78364313e+00
6.51391447e-01 3.21271747e-01 -9.06076096e-03 -7.81102702e-02
8.78831804e-01 3.63150507e-01 -7.31546402e-01 1.79591537e-01
-1.62741974e-01 1.00261427e-01 3.01954925e-01 1.63917080e-01
-1.40594912e+00 1.46357214e+00 4.82097292e+00 1.00549281e+00
-1.02649915e+00 -4.93319705e-02 6.46239638e-01 -2.15319857e-01
1.41035631e-01 1.36772590e-02 -1.07265770e+00 -4.19390714e-03
1.38264373e-01 -2.71235555e-02 3.34296316e-01 8.88231456e-01
1.40455127e-01 -2.34546602e-01 -1.18422484e+00 1.09094894e+00
3.65712166e-01 -1.21566463e+00 3.55602324e-01 -1.14314288e-01
5.55400014e-01 9.11941454e-02 1.73949063e-01 4.40507829e-01
-3.09764128e-02 -8.83995652e-01 9.09153163e-01 1.28612205e-01
7.81245351e-01 -3.40221673e-01 3.87091964e-01 3.16897929e-01
-1.65550613e+00 -1.28227443e-01 -7.17112839e-01 9.11182389e-02
-3.30960602e-01 5.55018127e-01 -8.23161960e-01 5.14472783e-01
8.52887154e-01 1.08272231e+00 -8.32012951e-01 6.07888520e-01
-2.29865253e-01 4.58367586e-01 -2.69081384e-01 -2.01418743e-01
1.89631209e-01 -6.64824694e-02 5.42690575e-01 1.32254076e+00
-2.17761099e-01 1.84780836e-01 2.42609873e-01 1.17580152e+00
-3.14315766e-01 3.20751846e-01 -5.92332780e-01 4.03861664e-02
4.06390369e-01 1.60841680e+00 -1.13846564e+00 -3.90067786e-01
-7.62629390e-01 9.85798120e-01 5.89623094e-01 3.25050265e-01
-9.12612855e-01 -2.99309105e-01 1.88709930e-01 6.99063689e-02
6.33069515e-01 -1.79282073e-02 -5.42652667e-01 -1.37996078e+00
4.97809023e-01 -9.19597208e-01 3.63433927e-01 -8.69510829e-01
-1.22013927e+00 3.86811525e-01 -2.95311630e-01 -1.25704539e+00
5.87393165e-01 -8.49648356e-01 -5.65447867e-01 7.30044365e-01
-1.45856452e+00 -1.61797321e+00 -5.70777953e-01 9.34994102e-01
1.21203566e+00 4.23502550e-02 1.94313109e-01 4.83687259e-02
-1.05248821e+00 6.46178067e-01 -1.82738975e-01 3.81893426e-01
7.10290611e-01 -1.19080901e+00 5.71034968e-01 8.03473234e-01
2.88665891e-01 8.72735977e-01 3.85157496e-01 -7.73198724e-01
-1.70942605e+00 -1.30456829e+00 6.22233391e-01 -6.22488916e-01
1.01376724e+00 -8.94461215e-01 -8.99031281e-01 9.68679070e-01
2.60410318e-03 1.28726304e-01 1.50972545e-01 7.88865983e-02
-6.41551554e-01 -1.18766360e-01 -6.36695266e-01 1.04795790e+00
1.10805786e+00 -7.20962346e-01 -7.16896236e-01 5.51855326e-01
7.12329924e-01 -5.14632702e-01 -3.86304110e-01 2.10156962e-01
3.91786486e-01 -1.18263519e+00 9.31230068e-01 -1.03201292e-01
6.70485795e-01 -3.62358809e-01 -1.15275458e-01 -3.75488698e-01
-1.29492298e-01 -4.66166168e-01 -9.67877507e-02 1.22376478e+00
1.77637488e-01 -7.64781654e-01 7.64655948e-01 2.13060036e-01
-1.92908451e-01 -8.02446902e-01 -9.44879055e-01 -4.56225932e-01
8.57187156e-03 -4.91510332e-01 2.46883452e-01 1.05036223e+00
-5.01442887e-03 7.40778208e-01 -6.69793040e-02 3.12544227e-01
5.86054504e-01 3.41228426e-01 9.42593098e-01 -9.84378636e-01
-2.50407517e-01 -7.88100779e-01 -3.48915756e-01 -1.70505631e+00
1.44284889e-01 -9.70515072e-01 2.49857128e-01 -1.56646979e+00
5.76767921e-01 -3.81644547e-01 -8.28074943e-03 4.87598777e-01
-4.85038757e-01 1.58539250e-01 1.00809582e-01 4.36435014e-01
-1.06827855e+00 6.67323112e-01 1.27481878e+00 -2.48731151e-01
-1.72428694e-02 -2.45073408e-01 -6.63279176e-01 7.43952096e-01
6.45577729e-01 -4.44451094e-01 -3.61065835e-01 -5.31255245e-01
3.68079931e-01 -1.02468424e-01 6.27114415e-01 -7.14893579e-01
4.99977916e-01 -2.95003261e-02 6.71296597e-01 -1.15151584e+00
3.21305990e-01 -7.91632831e-01 -5.87435067e-01 2.09164452e-02
-4.79497850e-01 -2.55915940e-01 1.82445124e-01 7.19720304e-01
3.82244773e-02 -9.32209119e-02 6.00800335e-01 -7.83675760e-02
-4.88939911e-01 2.79419661e-01 -8.77956972e-02 1.65364042e-01
9.62359726e-01 -2.90918618e-01 -7.90474474e-01 -3.92630510e-02
-1.84858322e-01 3.26573879e-01 3.38210821e-01 4.13421452e-01
7.87589073e-01 -7.90760577e-01 -6.44438624e-01 1.06593788e-01
4.83626395e-01 5.13276875e-01 2.68021494e-01 1.07217085e+00
-3.63684833e-01 5.00380039e-01 4.45315629e-01 -9.85661089e-01
-1.38326597e+00 6.81668758e-01 4.25018966e-01 -1.90618932e-01
-1.24678707e+00 9.06926453e-01 1.00944674e+00 -2.06291035e-01
3.88972342e-01 -4.68116254e-01 -3.08236368e-02 -1.40619680e-01
5.40772140e-01 -1.60227753e-02 -2.88090166e-02 -7.40133286e-01
-2.69814879e-01 7.40247905e-01 -4.93053466e-01 2.43875355e-01
9.94619012e-01 -3.21775973e-01 -2.58723497e-01 4.12236035e-01
8.36478412e-01 4.56338823e-02 -1.26496267e+00 -6.06250703e-01
-1.28101766e-01 -4.21030968e-01 1.37556121e-01 -4.35040683e-01
-9.75384533e-01 8.17884922e-01 1.60079196e-01 1.23934805e-01
1.18885767e+00 4.09760684e-01 5.29942393e-01 7.19719172e-01
-4.48994786e-02 -9.27522063e-01 4.05511141e-01 4.07402813e-01
8.35235715e-01 -1.40906978e+00 3.43683720e-01 -6.64417326e-01
-5.99213719e-01 1.20592070e+00 7.79516160e-01 -1.70521632e-01
4.79145259e-01 3.38893384e-01 -1.00266501e-01 -5.26273549e-01
-8.98062468e-01 -3.74355286e-01 3.81795973e-01 1.32492408e-01
3.91652912e-01 2.70879939e-02 2.86800433e-02 2.71374553e-01
2.20772270e-02 -4.80768412e-01 2.53433049e-01 9.02140021e-01
-4.88020986e-01 -5.94327867e-01 -4.66552764e-01 6.88315213e-01
-3.23410243e-01 -4.77115095e-01 -7.17706442e-01 8.61061752e-01
-7.15523167e-03 8.70179832e-01 6.20548427e-03 -2.67188609e-01
2.57220358e-01 -1.28575802e-01 3.96676391e-01 -6.85888529e-01
-7.74777353e-01 4.80648339e-01 -1.80989772e-01 -4.13900554e-01
-2.21195161e-01 -6.68186605e-01 -1.44652891e+00 -1.43754274e-01
-7.34287322e-01 -4.92593586e-01 4.25036222e-01 9.59698081e-01
2.12139994e-01 6.52334213e-01 5.30137956e-01 -7.05555618e-01
-4.55695778e-01 -8.75220597e-01 -4.96564597e-01 3.25754195e-01
3.89175802e-01 -5.70438921e-01 -3.57641667e-01 1.32938951e-01]
|
[12.031734466552734, 2.2603437900543213]
|
9eaedd2e-7add-4225-a27e-feff4b3dc6a4
|
effective-slot-filling-via-weakly-supervised
| null | null |
https://ojs.aaai.org/index.php/AAAI/article/view/17643
|
https://ojs.aaai.org/index.php/AAAI/article/view/17643/17450
|
Effective Slot Filling via Weakly-Supervised Dual-Model Learning
|
Slot filling is a challenging task in Spoken Language Understanding (SLU). Supervised methods usually require large amounts of annotation to maintain desirable performance. A solution to relieve the heavy dependency on labeled data is to employ bootstrapping, which leverages unlabeled data. However, bootstrapping is known to suffer from semantic drift. We argue that semantic drift can be tackled by exploiting the correlation between slot values (phrases) and their respective types. By using some particular weakly-labeled data, namely the plain phrases included in sentences, we propose a weaklysupervised slot filling approach. Our approach trains two models, namely a classifier and a tagger, which can effectively learn from each other on the weakly-labeled data. The experimental results demonstrate that our approach achieves better results than standard baselines on multiple datasets, especially in the low-resource setting.
|
['Gang Chen', 'Sai Wu', 'Lidan Shou', 'Ke Chen', 'Jue Wang']
|
2021-05-18
| null | null | null |
aaai-2021-5
|
['slot-filling']
|
['natural-language-processing']
|
[ 1.62689120e-01 4.51505840e-01 -5.61566710e-01 -6.81853533e-01
-1.11323977e+00 -5.31892896e-01 3.38578612e-01 8.55811685e-02
-6.23352349e-01 1.09054995e+00 1.88661039e-01 -2.04458520e-01
4.05696541e-01 -5.67509890e-01 -6.91760838e-01 -7.17512608e-01
2.85058320e-01 7.05831468e-01 3.18394452e-01 -1.61049008e-01
9.88391787e-02 -4.77400243e-01 -1.49406374e+00 1.86419517e-01
1.20592237e+00 8.23596120e-01 2.39366129e-01 1.40602455e-01
-8.77112150e-01 5.43396294e-01 -4.37370658e-01 -1.70253232e-01
2.68186498e-02 -5.06680071e-01 -1.04712796e+00 3.01450461e-01
-1.41438529e-01 -3.64116691e-02 1.89800382e-01 1.16506243e+00
2.91648179e-01 -4.15988034e-03 2.20951557e-01 -1.17002964e+00
-8.54799375e-02 9.94692266e-01 -4.95177686e-01 -5.43977283e-02
2.22644508e-01 -3.63324493e-01 1.18953478e+00 -1.01223350e+00
6.02610052e-01 1.24821270e+00 6.61802769e-01 8.89154673e-01
-1.09175587e+00 -7.54745007e-01 5.37688375e-01 -1.74165964e-01
-1.08421826e+00 -5.92614830e-01 6.77510321e-01 -1.88663602e-01
6.75586343e-01 -4.25411249e-03 1.77801445e-01 1.13430429e+00
-7.56530464e-01 1.41133237e+00 1.18074012e+00 -7.67339766e-01
6.88632011e-01 5.00016868e-01 5.93265593e-01 4.40592736e-01
1.31327346e-01 -2.66840756e-01 -9.66698110e-01 -3.17425847e-01
1.71535134e-01 -1.63402528e-01 -3.31315339e-01 -5.76428950e-01
-8.77289176e-01 8.52684438e-01 -1.16955049e-01 1.97779045e-01
-2.07073569e-01 -1.77104548e-01 5.14261127e-01 2.87119567e-01
1.05281770e+00 1.78735271e-01 -1.13251877e+00 -4.71948028e-01
-8.63281071e-01 1.38544500e-01 8.69576991e-01 1.04204488e+00
8.74762714e-01 -4.95284140e-01 -1.66754961e-01 1.19655311e+00
3.60671818e-01 4.62663829e-01 7.44157434e-01 -4.65215832e-01
7.58347034e-01 6.07312560e-01 4.36182916e-01 -8.26516151e-02
-3.20802867e-01 -1.99666172e-02 -2.89803237e-01 -5.28101683e-01
5.51061332e-01 -3.78015578e-01 -1.26740861e+00 2.10614872e+00
5.07075727e-01 3.72420222e-01 3.12579513e-01 7.64826298e-01
6.27021670e-01 4.18428332e-01 3.97772759e-01 -4.40552473e-01
1.40661061e+00 -1.12965822e+00 -9.88247097e-01 -6.61844552e-01
9.72817063e-01 -3.24280053e-01 1.39345157e+00 3.94600704e-02
-6.68352902e-01 -4.89116460e-02 -8.77083659e-01 1.18171513e-01
-1.26088172e-01 -1.08851746e-01 6.24921381e-01 7.11659014e-01
-7.43842602e-01 4.59572971e-01 -1.08505607e+00 -4.60262686e-01
4.38824683e-01 2.22532287e-01 -1.93355903e-01 -1.40329868e-01
-1.32891715e+00 4.84460145e-01 4.72161382e-01 -1.00380465e-01
-5.26473641e-01 -4.46897686e-01 -9.38333929e-01 -3.26334219e-03
8.02667916e-01 -2.60392487e-01 1.71617782e+00 -7.43833840e-01
-1.45435655e+00 1.05961406e+00 -6.85670555e-01 -6.01388395e-01
5.01501381e-01 -5.59411943e-01 5.72738200e-02 -9.60306600e-02
3.62433493e-01 5.52958131e-01 8.10171664e-01 -1.25313282e+00
-8.54128122e-01 -4.39675331e-01 -3.35545242e-01 3.91650170e-01
-5.43942988e-01 -1.71821013e-01 -4.87726688e-01 -3.59577477e-01
5.45263588e-01 -8.75173986e-01 -2.39787892e-01 -4.26748872e-01
-4.23288643e-01 -6.61804795e-01 7.66481996e-01 -5.58895528e-01
1.17362916e+00 -2.23185945e+00 -7.90859908e-02 -6.15038276e-02
-2.09184006e-01 4.14735973e-01 1.12521136e-02 1.25438914e-01
3.09701592e-01 7.12210387e-02 -5.52354872e-01 -6.87713087e-01
-4.86630648e-02 5.41228235e-01 -5.87039113e-01 -1.92135237e-02
1.53476804e-01 8.72193933e-01 -1.26734829e+00 -3.70235473e-01
-2.52070457e-01 5.81688853e-03 -4.72855389e-01 2.77615666e-01
-6.48575306e-01 5.68475127e-01 -6.40061975e-01 7.34832406e-01
5.87861538e-01 -3.52459878e-01 4.76656675e-01 5.53560376e-01
2.39562452e-01 7.63857663e-01 -1.06218147e+00 2.13240647e+00
-3.20820361e-01 1.28307477e-01 8.28573331e-02 -1.28237045e+00
1.04511261e+00 5.71422100e-01 2.53778636e-01 -5.54496348e-01
-9.22245532e-02 5.97173810e-01 -4.31885034e-01 -4.06216145e-01
3.34173471e-01 -3.83118033e-01 -3.01502615e-01 7.19944179e-01
7.54755661e-02 -1.26328290e-01 1.48870751e-01 2.26599514e-01
1.02519584e+00 2.65937477e-01 8.50164816e-02 -2.66541004e-01
3.21713120e-01 -1.25988349e-02 1.01077247e+00 8.97118449e-01
-4.66648847e-01 3.28424156e-01 5.38136840e-01 -2.50614971e-01
-9.04872119e-01 -7.78037429e-01 -2.47614592e-01 1.23068893e+00
2.01208964e-01 -3.75790238e-01 -8.88505757e-01 -1.14761329e+00
1.61994703e-03 5.88954389e-01 -3.40932280e-01 -2.14980856e-01
-3.13837618e-01 -9.83252525e-01 2.65552580e-01 5.26621819e-01
4.82833236e-01 -1.04270291e+00 -3.50712210e-01 4.26379502e-01
-5.23318708e-01 -1.33510089e+00 -1.68600380e-01 6.20611250e-01
-1.08709490e+00 -8.33987653e-01 -5.57437181e-01 -8.18584740e-01
6.92275286e-01 4.17022854e-01 1.19499075e+00 7.57898241e-02
2.56677836e-01 1.59230843e-01 -6.44526660e-01 -5.72702110e-01
-2.85328418e-01 5.01014352e-01 2.67716736e-01 -7.10826963e-02
9.67622101e-01 -3.90426457e-01 -3.38694692e-01 3.37597966e-01
-7.11400211e-01 3.09187006e-02 2.45924816e-01 1.38391399e+00
3.74934196e-01 -1.20426662e-01 8.74163210e-01 -1.31868303e+00
4.72223133e-01 -4.07633662e-01 -3.97765130e-01 4.61356044e-01
-8.44556570e-01 3.37583750e-01 2.10722342e-01 -2.38020226e-01
-1.30993474e+00 1.57662153e-01 2.42067147e-02 7.83404186e-02
-2.00993896e-01 4.92078394e-01 -2.99057752e-01 4.66368288e-01
4.62169260e-01 1.96166128e-01 -1.28759220e-01 -7.54364491e-01
3.93418789e-01 9.93698120e-01 3.29645455e-01 -7.39526033e-01
5.51997900e-01 4.73699123e-01 -8.52111459e-01 -7.35961556e-01
-1.61047256e+00 -7.70558536e-01 -5.87202728e-01 2.59479642e-01
3.20751935e-01 -1.23801196e+00 -1.53522879e-01 5.69227159e-01
-8.67197156e-01 -4.84499693e-01 -3.39146256e-01 4.34669048e-01
-3.10413122e-01 3.44358593e-01 -6.79430783e-01 -1.15028226e+00
-3.98388475e-01 -7.09162533e-01 1.20378053e+00 4.40780640e-01
-2.52084464e-01 -8.84469390e-01 1.43285915e-01 4.98725265e-01
1.32017285e-01 -4.87561435e-01 6.63311064e-01 -1.20426667e+00
-3.73897612e-01 -2.92852998e-01 -9.60420892e-02 1.82966292e-01
2.57854044e-01 -5.88720262e-01 -1.37354124e+00 -3.24369580e-01
7.58225620e-02 -8.89044404e-01 1.08298421e+00 2.79766053e-01
8.84644628e-01 -4.52009924e-02 -5.67463636e-01 1.51666760e-01
8.75702739e-01 1.85650274e-01 1.61397904e-01 4.12151366e-01
4.84294742e-01 7.32676029e-01 1.07493234e+00 4.98160154e-01
5.15578628e-01 5.48694193e-01 -8.60227924e-03 1.66563597e-02
3.27401221e-01 -7.04224944e-01 1.76519841e-01 7.68035352e-01
4.86441076e-01 -7.97336176e-02 -9.00817871e-01 7.59546340e-01
-2.30083108e+00 -4.23775315e-01 2.01887667e-01 2.25140882e+00
1.50068247e+00 4.30181414e-01 5.64701706e-02 2.92293906e-01
7.11088002e-01 1.23116806e-01 -7.40101516e-01 1.05668515e-01
-1.16341762e-01 2.35364541e-01 3.67941409e-01 5.11283100e-01
-1.25978148e+00 1.49128985e+00 6.18415117e+00 5.75706422e-01
-8.84750485e-01 2.89396822e-01 6.46641791e-01 -4.71333647e-03
-3.74207050e-01 1.72195002e-01 -1.10856688e+00 6.71424568e-01
9.61281240e-01 -1.53487956e-04 6.17245883e-02 1.01009190e+00
1.30148843e-01 -3.76056522e-01 -1.02728128e+00 6.18305981e-01
-1.30403265e-01 -8.81881595e-01 -3.15341294e-01 -3.20926428e-01
7.00601041e-01 4.05072644e-02 -2.17041418e-01 5.70782721e-01
7.45704234e-01 -5.82061112e-01 5.04998744e-01 -8.43524262e-02
5.39521754e-01 -3.43536079e-01 8.49228263e-01 5.95019937e-01
-8.21464241e-01 3.05869374e-02 -3.05567563e-01 -1.78432256e-01
2.94134378e-01 7.85861969e-01 -9.54560041e-01 2.15206265e-01
6.76811695e-01 6.74987137e-01 -1.62689909e-01 7.64485359e-01
-6.78170562e-01 1.02476346e+00 -3.83852988e-01 3.77300829e-02
1.83361277e-01 -9.76143554e-02 1.78961039e-01 8.35097075e-01
1.15353532e-01 -2.56741028e-02 3.69988829e-01 5.66664696e-01
-1.68932647e-01 2.40712628e-01 -4.35681134e-01 -2.04367504e-01
8.39990079e-01 1.02962053e+00 -8.40755045e-01 -5.89952111e-01
-4.94734168e-01 1.00499022e+00 4.98276919e-01 3.84198785e-01
-3.59352380e-01 3.28083485e-02 6.86700821e-01 -2.84238577e-01
3.94370973e-01 -1.65995359e-02 -4.85523462e-01 -1.56753409e+00
3.56263906e-01 -5.63841045e-01 5.60774505e-01 -4.28917885e-01
-1.12193620e+00 4.54934210e-01 -2.02491984e-01 -9.34200406e-01
-4.60168093e-01 -2.18455315e-01 -2.91936219e-01 6.43396020e-01
-1.91125989e+00 -8.81981492e-01 -6.97105899e-02 3.66139710e-01
7.61292040e-01 4.46877070e-02 9.14607584e-01 2.11201385e-01
-6.84043050e-01 6.00195169e-01 7.05028996e-02 1.88700184e-01
8.24390471e-01 -1.31645405e+00 6.13673329e-01 7.29074776e-01
5.06207824e-01 4.25098032e-01 7.72536695e-01 -6.65303946e-01
-8.38560283e-01 -7.35360205e-01 1.26968622e+00 -4.57898468e-01
5.84070802e-01 -6.48452282e-01 -1.30127370e+00 6.76218987e-01
-2.30967417e-01 2.21424270e-02 8.14081311e-01 6.35022461e-01
-5.16068757e-01 2.14773819e-01 -8.81409049e-01 2.57357806e-01
1.10405791e+00 -7.44609416e-01 -8.97936583e-01 4.15408254e-01
9.82154250e-01 -5.50826728e-01 -1.97937489e-01 3.07656318e-01
8.39132294e-02 -6.39701962e-01 5.61996758e-01 -9.05698538e-01
8.44180062e-02 -1.22387365e-01 9.23680067e-02 -1.46508360e+00
4.19831127e-01 -5.43890536e-01 -5.20078689e-02 1.38407099e+00
6.72109544e-01 -5.21133721e-01 1.19445419e+00 1.16679084e+00
1.90719694e-01 -3.95917177e-01 -1.07711220e+00 -8.09664011e-01
-4.17565517e-02 -5.10932803e-01 5.71072817e-01 1.02248573e+00
4.75730836e-01 6.29521549e-01 -4.70991105e-01 2.13575503e-03
4.47199881e-01 2.11008191e-01 6.65415704e-01 -1.41520154e+00
1.52694266e-02 1.96846411e-01 6.58548772e-02 -1.12767124e+00
5.20674109e-01 -6.92778826e-01 5.37140191e-01 -1.09952140e+00
2.67848372e-01 -9.35492158e-01 -4.36555624e-01 8.34916532e-01
-6.71448290e-01 -3.52640450e-02 -1.35190815e-01 3.82023245e-01
-8.78146172e-01 7.44882464e-01 6.90396488e-01 -3.36020775e-02
-5.28211415e-01 1.31169140e-01 -6.92300975e-01 7.76059866e-01
7.47251153e-01 -7.44679272e-01 -4.26210225e-01 -3.29002678e-01
6.59090951e-02 2.19922885e-02 -2.08508283e-01 -5.25472462e-01
9.48830545e-02 -4.63950709e-02 -3.21225584e-01 -4.11402047e-01
1.07715063e-01 -5.30980468e-01 -5.37727952e-01 3.16982687e-01
-5.14802337e-01 -5.55310607e-01 7.38033578e-02 7.02272117e-01
-4.41815794e-01 -4.28369820e-01 4.96695429e-01 -3.17714870e-01
-8.08268368e-01 2.12660134e-01 3.68614830e-02 4.05432761e-01
7.54333079e-01 2.13042378e-01 -1.77765161e-01 -3.21685135e-01
-7.74934113e-01 6.12370074e-01 3.84128422e-01 6.29030108e-01
1.37333289e-01 -1.18914175e+00 -4.08358514e-01 3.03434938e-01
4.50328857e-01 4.97531295e-01 3.63924615e-02 5.21901846e-01
2.89613336e-01 4.33703214e-01 3.58348697e-01 -7.88779020e-01
-9.75184619e-01 3.92759323e-01 6.31228536e-02 -3.82224500e-01
-5.48044801e-01 1.15180433e+00 2.47864574e-02 -7.33755350e-01
5.92151523e-01 -2.41244435e-01 -7.54437670e-02 2.86597759e-01
6.01673424e-01 -2.23582387e-01 1.38710871e-01 -1.44380644e-01
-4.19283450e-01 1.59170359e-01 -3.47945750e-01 -2.80760556e-01
1.44877350e+00 -5.33722401e-01 1.99938938e-03 7.29583979e-01
8.27970326e-01 -2.61662126e-01 -1.56194162e+00 -1.00055468e+00
9.29100692e-01 -3.78882825e-01 -1.61086217e-01 -6.47081733e-01
-7.80700803e-01 7.69989491e-01 3.37490886e-01 2.01721370e-01
7.32835114e-01 1.04935706e-01 1.02468491e+00 4.97010946e-01
7.16583967e-01 -1.57024062e+00 -5.19449413e-02 5.76146364e-01
1.56803533e-01 -1.80391467e+00 -4.74708140e-01 -5.08363485e-01
-8.27487469e-01 5.97137511e-01 6.88686907e-01 4.64836657e-01
5.19260824e-01 2.76034981e-01 5.61194003e-01 3.99027765e-02
-9.29609299e-01 -3.85759979e-01 -1.50602579e-01 4.23517227e-01
4.25024003e-01 1.13286756e-01 -5.10277271e-01 8.43519926e-01
2.32880954e-02 2.07183674e-01 2.80860215e-01 1.21929121e+00
-5.96975446e-01 -1.62431943e+00 -2.17975080e-01 3.08064163e-01
-4.44950044e-01 -2.22795457e-01 -2.68248469e-01 3.46877873e-01
-2.44583055e-01 1.02796221e+00 3.22738327e-02 -3.49367969e-02
1.90732062e-01 8.08080554e-01 -2.68585924e-02 -9.76228595e-01
-7.12864473e-02 7.15093911e-02 2.30433926e-01 -4.03337479e-01
-6.77346885e-01 -8.05046558e-01 -1.37100279e+00 2.74998635e-01
-4.87463504e-01 7.67410278e-01 5.98990798e-01 1.30976188e+00
2.37030521e-01 1.11150019e-01 6.57108188e-01 -3.20205718e-01
-8.93128335e-01 -1.07840729e+00 -8.73192430e-01 6.28774047e-01
2.58085728e-01 -8.62348020e-01 -4.61780071e-01 -7.71367997e-02]
|
[12.45687198638916, 7.289758205413818]
|
dd9d50f5-2789-49d8-84e8-b4609a2a593e
|
lanit-language-driven-image-to-image
|
2208.14889
| null |
https://arxiv.org/abs/2208.14889v4
|
https://arxiv.org/pdf/2208.14889v4.pdf
|
LANIT: Language-Driven Image-to-Image Translation for Unlabeled Data
|
Existing techniques for image-to-image translation commonly have suffered from two critical problems: heavy reliance on per-sample domain annotation and/or inability of handling multiple attributes per image. Recent truly-unsupervised methods adopt clustering approaches to easily provide per-sample one-hot domain labels. However, they cannot account for the real-world setting: one sample may have multiple attributes. In addition, the semantics of the clusters are not easily coupled to the human understanding. To overcome these, we present a LANguage-driven Image-to-image Translation model, dubbed LANIT. We leverage easy-to-obtain candidate attributes given in texts for a dataset: the similarity between images and attributes indicates per-sample domain labels. This formulation naturally enables multi-hot label so that users can specify the target domain with a set of attributes in language. To account for the case that the initial prompts are inaccurate, we also present prompt learning. We further present domain regularization loss that enforces translated images be mapped to the corresponding domain. Experiments on several standard benchmarks demonstrate that LANIT achieves comparable or superior performance to existing models.
|
['Seungryong Kim', 'Seokju Cho', 'Sunwoo Kim', 'Youngjung Uh', 'Jaejun Yoo', 'Soohyun Kim', 'JiHye Park']
|
2022-08-31
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Park_LANIT_Language-Driven_Image-to-Image_Translation_for_Unlabeled_Data_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Park_LANIT_Language-Driven_Image-to-Image_Translation_for_Unlabeled_Data_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['unsupervised-image-to-image-translation']
|
['computer-vision']
|
[ 5.01426637e-01 -3.91437523e-02 -6.38242066e-01 -9.46217060e-01
-1.36812687e+00 -8.81155312e-01 8.13478708e-01 4.36473750e-02
-4.24615890e-01 5.75962603e-01 1.09962024e-01 -3.91029492e-02
1.75384298e-01 -3.18910599e-01 -7.97312260e-01 -5.86020529e-01
6.84745371e-01 1.03487396e+00 -5.41203991e-02 3.38713437e-01
-1.15998566e-01 -4.99905739e-03 -1.27239835e+00 6.10976815e-01
8.65078270e-01 8.75655770e-01 3.13585192e-01 3.06034952e-01
-3.08025718e-01 6.08923733e-01 -4.96948034e-01 -5.01120090e-01
3.01401258e-01 -5.95381916e-01 -8.20659935e-01 7.61130512e-01
5.84072113e-01 -3.60846758e-01 1.13896906e-01 1.20929122e+00
2.20381811e-01 -4.59222049e-02 1.12519097e+00 -1.58218014e+00
-9.33091521e-01 4.04902756e-01 -5.57830036e-01 -3.27369541e-01
4.94580209e-01 3.77025381e-02 1.01169729e+00 -1.12381983e+00
8.57839227e-01 1.20430911e+00 3.21728706e-01 4.71078366e-01
-1.66572189e+00 -6.52816415e-01 2.40180671e-01 1.95190132e-01
-1.54860544e+00 -5.32250464e-01 7.26648092e-01 -6.24818683e-01
3.90712887e-01 6.16643764e-02 6.87614456e-02 1.34070086e+00
-3.23170543e-01 8.61373961e-01 1.40297174e+00 -6.10468686e-01
1.38989016e-01 6.40166402e-01 -1.58055753e-01 3.05256605e-01
-1.13150991e-01 -2.27645472e-01 -5.96073806e-01 -1.93575978e-01
5.61976671e-01 -1.94140241e-01 -5.01758717e-02 -6.40980005e-01
-1.48762679e+00 8.47020805e-01 1.44887149e-01 -9.39357355e-02
-1.97503939e-01 -2.67722189e-01 2.67160028e-01 4.47218657e-01
4.77019250e-01 3.88226986e-01 -4.98851389e-01 1.18121915e-01
-9.84771073e-01 6.67764992e-02 5.99884391e-01 1.52515280e+00
1.12672043e+00 -3.27656090e-01 -1.36029109e-01 8.18968654e-01
1.26498505e-01 5.87689340e-01 4.00233269e-01 -1.21328115e+00
4.08503026e-01 5.08722961e-01 3.10178250e-01 -7.23229527e-01
1.05341868e-02 1.27360057e-02 -6.76534891e-01 -7.46168420e-02
5.01743972e-01 1.23934224e-01 -1.00159395e+00 1.84387672e+00
2.69028991e-01 9.97631773e-02 1.26545265e-01 1.06955159e+00
5.91341674e-01 5.28212726e-01 3.37427586e-01 -2.51519233e-01
1.45532119e+00 -8.50666523e-01 -6.51658952e-01 -5.00867188e-01
5.72599828e-01 -1.02333105e+00 1.63275313e+00 1.63168907e-01
-7.84767687e-01 -5.61923087e-01 -6.21735573e-01 -3.46294455e-02
-3.09410155e-01 4.84688520e-01 2.15531111e-01 5.04379451e-01
-9.94585097e-01 5.35144880e-02 -5.06986201e-01 -6.66644514e-01
2.84037411e-01 2.74927378e-01 -4.48339492e-01 -3.18762004e-01
-9.74736333e-01 6.93383396e-01 3.95940244e-01 -6.46288872e-01
-7.66037524e-01 -5.71161389e-01 -7.63135433e-01 -3.59292477e-01
4.76265937e-01 -7.63889849e-01 1.46597552e+00 -1.66527939e+00
-1.11544394e+00 1.43287301e+00 -4.71345931e-01 -2.03107208e-01
5.43743193e-01 -1.79134890e-01 -4.16036159e-01 3.58051836e-01
7.05419779e-01 1.12916040e+00 1.13491654e+00 -1.51890814e+00
-7.10004628e-01 -1.54565290e-01 -8.89879763e-02 5.38673639e-01
-5.51670074e-01 9.92211923e-02 -7.58764803e-01 -6.68534517e-01
4.56264541e-02 -1.11328018e+00 3.41071151e-02 2.81378448e-01
-5.09534955e-01 -2.25594416e-01 6.68039978e-01 -3.26277524e-01
7.63318300e-01 -2.33755755e+00 -4.77837808e-02 5.48346415e-02
1.04346864e-01 -7.76430294e-02 -2.73971707e-01 1.36980385e-01
-1.57033324e-01 9.93048921e-02 -2.46610418e-01 -4.82994139e-01
2.38581002e-02 4.27565098e-01 -4.81067270e-01 3.65748495e-01
4.27515179e-01 7.18356252e-01 -9.17060435e-01 -1.02359521e+00
2.78960038e-02 2.56142139e-01 -4.66014296e-01 3.24908078e-01
-4.58382905e-01 7.68443584e-01 -4.86622274e-01 5.48115015e-01
5.82745314e-01 -5.46271980e-01 2.16846719e-01 -3.19958925e-01
3.04562926e-01 2.18433425e-01 -1.07469201e+00 1.72676194e+00
-2.15486005e-01 5.08173108e-01 -3.56372520e-02 -8.38770628e-01
7.65543580e-01 4.33931112e-01 5.84717631e-01 -5.73053300e-01
-2.12416619e-01 2.78034568e-01 -3.59025002e-01 -2.88770527e-01
4.80008006e-01 -3.43641222e-01 -2.65157312e-01 6.80226088e-01
6.96660578e-02 -2.27247521e-01 1.56994924e-01 4.75392342e-01
5.77816367e-01 3.93988580e-01 1.32072091e-01 -1.46508545e-01
2.96067864e-01 3.74571919e-01 7.72400737e-01 8.06830585e-01
-2.58840948e-01 1.01452065e+00 4.14559931e-01 -1.58011079e-01
-1.37287152e+00 -1.13025641e+00 -3.09623554e-02 1.41676545e+00
2.17314079e-01 -2.34360591e-01 -8.16084206e-01 -7.04856336e-01
-1.83848336e-01 7.46315539e-01 -3.39147687e-01 -7.45876655e-02
-1.65144637e-01 -4.55860585e-01 4.22056347e-01 4.51154172e-01
2.22006232e-01 -8.90037656e-01 -2.90230870e-01 1.63765863e-01
-7.93399096e-01 -1.78163517e+00 -8.04131269e-01 1.12865448e-01
-5.06716549e-01 -7.93693721e-01 -6.78593874e-01 -1.08806980e+00
1.08925223e+00 2.81216800e-01 1.45002520e+00 -2.05049306e-01
1.19995371e-01 6.82278454e-01 -4.82976437e-01 -1.62308425e-01
-7.01016426e-01 1.63168117e-01 2.45173618e-01 2.89491296e-01
8.24086785e-01 -2.67718822e-01 -4.01867211e-01 6.18725777e-01
-9.21227098e-01 3.48618299e-01 5.81874073e-01 7.21404612e-01
9.56167877e-01 -1.57939032e-01 5.45676112e-01 -1.22837102e+00
4.14957345e-01 -3.97773564e-01 -2.90473938e-01 4.63389099e-01
-6.90701067e-01 9.22059268e-02 5.90334833e-01 -8.20145965e-01
-1.01297998e+00 6.27469361e-01 4.20292556e-01 -6.33323014e-01
-5.41271269e-01 3.94851685e-01 -3.29724967e-01 3.00401896e-01
7.37409413e-01 3.29758614e-01 -3.80138457e-02 -2.55033016e-01
6.51845157e-01 8.61789882e-01 7.80926466e-01 -1.09571779e+00
9.07706439e-01 4.52116013e-01 -6.10968828e-01 -5.40531158e-01
-1.08502555e+00 -6.13501906e-01 -8.91817212e-01 -1.18651003e-01
8.94026279e-01 -1.37750530e+00 -8.67051333e-02 1.78143248e-01
-1.15454793e+00 -3.41329008e-01 -1.70266956e-01 3.61220777e-01
-9.13779914e-01 3.69861871e-01 -5.04737258e-01 -5.16693473e-01
4.11814004e-02 -1.28790593e+00 1.21559036e+00 -9.57728550e-02
-5.09219944e-01 -8.67868304e-01 -5.01010656e-01 3.81315589e-01
1.49460450e-01 1.83416605e-02 1.07554650e+00 -7.22765207e-01
-5.17940104e-01 -7.48384595e-02 -4.60705698e-01 1.33590311e-01
2.21917629e-01 -1.56130120e-01 -9.57827151e-01 -3.06981087e-01
-3.21758911e-02 -7.75300264e-01 4.43171889e-01 1.39439061e-01
1.03102612e+00 -3.35387707e-01 -2.92875290e-01 5.44447184e-01
1.37082815e+00 -1.57475859e-01 3.81193906e-01 2.63716459e-01
7.23897219e-01 7.58380413e-01 9.42187250e-01 4.20404255e-01
6.40016079e-01 7.82122910e-01 -3.09168752e-02 -3.39352459e-01
-6.82928041e-02 -3.92771423e-01 3.61184090e-01 6.43757582e-01
4.64114785e-01 -1.31486624e-01 -9.46857214e-01 7.51176596e-01
-1.74456620e+00 -6.63310468e-01 -8.39120373e-02 2.24096537e+00
1.30213618e+00 1.34230837e-01 2.26218566e-01 -4.34370518e-01
8.89547944e-01 -2.62495428e-01 -6.51466727e-01 2.29020212e-02
-3.23390782e-01 -3.29667360e-01 5.38075268e-01 4.03247058e-01
-1.30324519e+00 1.18585563e+00 6.09716320e+00 8.13834727e-01
-1.01520109e+00 2.65166223e-01 8.08420122e-01 7.52745643e-02
-4.17441577e-01 1.54343829e-01 -8.23699117e-01 4.66822028e-01
7.33853102e-01 -2.73268849e-01 3.86472702e-01 8.41955721e-01
3.56241018e-02 9.11818966e-02 -1.54124737e+00 1.05085242e+00
1.98598266e-01 -8.39180350e-01 3.16256702e-01 2.90871616e-02
7.82098711e-01 -8.65323022e-02 2.69541681e-01 1.26876205e-01
5.33372402e-01 -8.33715498e-01 9.31921363e-01 2.47692972e-01
1.56262195e+00 -4.23996955e-01 2.85190314e-01 4.99141991e-01
-9.34559345e-01 1.50111362e-01 -4.44583267e-01 1.72220126e-01
-3.17832008e-02 4.01629299e-01 -1.11986697e+00 2.66610473e-01
5.32452583e-01 7.67359197e-01 -6.40999079e-01 5.40864885e-01
-2.40652323e-01 5.28956711e-01 -3.51979554e-01 3.34393382e-01
3.16674680e-01 -2.34514594e-01 2.20396012e-01 1.26534581e+00
2.66239643e-01 -2.90925391e-02 7.36744940e-01 9.52644169e-01
-1.81084082e-01 1.23428598e-01 -6.55186474e-01 -6.37714714e-02
7.95943439e-01 1.17980981e+00 -7.44209528e-01 -6.07591271e-01
-6.35380208e-01 1.31282461e+00 2.00778916e-01 6.17600143e-01
-7.27867305e-01 9.40357447e-02 6.54458702e-01 3.39499295e-01
7.63033032e-02 -1.40378773e-01 -5.01240134e-01 -1.33607817e+00
8.99900422e-02 -1.24057722e+00 4.70929980e-01 -9.92031515e-01
-1.59124959e+00 4.91713405e-01 1.56865001e-01 -1.54857683e+00
-3.42218131e-01 -4.44163144e-01 -1.23632858e-02 7.45730400e-01
-1.59237266e+00 -1.45872188e+00 -1.83658063e-01 8.94514441e-01
7.59000897e-01 -1.62036762e-01 1.01890731e+00 4.33389217e-01
-2.12234914e-01 7.64596343e-01 4.93905954e-02 3.52880657e-01
1.60319149e+00 -1.34371197e+00 2.86932439e-01 8.10775936e-01
2.94717342e-01 5.67199111e-01 8.04314494e-01 -5.97298205e-01
-9.81825590e-01 -1.26396978e+00 1.10827661e+00 -9.13654327e-01
6.10369265e-01 -4.53383803e-01 -9.63328719e-01 8.99260283e-01
2.34299108e-01 -5.96056283e-02 7.80660212e-01 -9.21149328e-02
-6.18141353e-01 1.24025680e-02 -9.97903645e-01 6.34027660e-01
7.29085684e-01 -8.27366114e-01 -4.09078479e-01 6.22215986e-01
7.11103678e-01 -3.79828751e-01 -6.91084385e-01 4.53299917e-02
1.08607531e-01 -5.03542066e-01 9.11020100e-01 -4.43765193e-01
6.19957924e-01 -4.45898563e-01 -3.51475090e-01 -1.09170640e+00
-3.32791328e-01 -4.78244096e-01 3.47562939e-01 1.42626297e+00
5.62474847e-01 -2.10784182e-01 6.51371241e-01 1.11706781e+00
5.66050000e-02 -1.98538229e-01 -7.20669448e-01 -7.80884445e-01
1.95303127e-01 -3.71970564e-01 5.69562316e-01 1.39740026e+00
-2.10549414e-01 6.71076059e-01 -5.38661897e-01 1.86510652e-01
8.01775396e-01 1.85636714e-01 7.89112031e-01 -1.17510605e+00
-1.52084276e-01 -1.60158828e-01 -1.63126901e-01 -1.20438123e+00
4.15089339e-01 -8.69093001e-01 2.74452299e-01 -1.32479501e+00
5.45204282e-01 -7.29356289e-01 -1.20211318e-01 6.69277906e-01
-2.52390176e-01 4.65966225e-01 5.58360368e-02 7.02412546e-01
-9.00071323e-01 2.01870456e-01 1.01969767e+00 -2.11330891e-01
-1.93632711e-02 -2.26392642e-01 -8.84063601e-01 6.92184627e-01
6.89511597e-01 -7.78726697e-01 -5.74790537e-01 -7.22429633e-01
1.77326217e-01 -7.27639198e-02 3.46569091e-01 -7.15829015e-01
3.99385810e-01 -4.00568843e-01 3.66957098e-01 -3.64533514e-01
2.31910378e-01 -1.08051801e+00 1.19432181e-01 -2.38719791e-01
-7.55681038e-01 -1.16855223e-02 -2.01668888e-01 6.34837627e-01
-3.21072668e-01 -7.20049068e-02 9.09434676e-01 -1.21714391e-01
-9.59110439e-01 2.40231916e-01 -2.18378231e-01 2.77837992e-01
1.04134107e+00 -3.65588188e-01 1.43752797e-02 -6.06009364e-01
-7.27068007e-01 2.80079931e-01 1.03371048e+00 4.88682926e-01
4.94607955e-01 -1.51227033e+00 -8.28264773e-01 2.02440187e-01
6.75277710e-01 -7.02944994e-02 -1.76588506e-01 6.70263052e-01
-7.35667273e-02 1.01682730e-02 5.50736766e-03 -9.60520029e-01
-1.22729886e+00 6.73186600e-01 1.04125246e-01 6.90748543e-02
-5.35250127e-01 5.81185818e-01 5.53074419e-01 -6.52381480e-01
2.21074119e-01 1.00934118e-01 2.74285406e-01 1.29807577e-01
2.76689291e-01 -4.16201234e-01 -1.87771946e-01 -9.33382571e-01
-3.46824616e-01 5.21730781e-01 -3.03840131e-01 -4.37842011e-01
9.97117519e-01 -5.34329712e-01 -2.81508807e-02 6.31705344e-01
1.09737551e+00 -3.58315587e-01 -1.50323820e+00 -8.60194862e-01
2.43335366e-01 -4.85912353e-01 -3.39440137e-01 -9.25894320e-01
-7.18222141e-01 8.08800697e-01 5.49676418e-01 -1.78200066e-01
1.20947945e+00 2.20545784e-01 6.78348243e-01 4.26943153e-01
2.98457652e-01 -1.41427600e+00 2.89287418e-01 3.58830363e-01
4.29913968e-01 -1.62404704e+00 -4.37461659e-02 -5.23838520e-01
-1.26287782e+00 8.23374271e-01 7.26631522e-01 2.38086447e-01
2.73413748e-01 1.18913531e-01 6.49173856e-01 1.30056411e-01
-6.08341277e-01 -1.45994872e-01 2.35057101e-01 8.87103736e-01
5.63964486e-01 2.38835335e-01 1.82530463e-01 4.82775837e-01
-4.07631658e-02 -9.27233547e-02 5.27407110e-01 6.66252017e-01
-2.96142310e-01 -1.36903942e+00 -5.02278268e-01 3.28485996e-01
-3.98752332e-01 -1.15647204e-01 -5.83238184e-01 4.89767581e-01
5.62597718e-03 9.34941292e-01 1.79596961e-01 -1.86087877e-01
1.36922449e-01 3.16594452e-01 2.40857258e-01 -8.83409023e-01
-1.19631134e-01 4.07011241e-01 -7.90441409e-02 -2.63504505e-01
-7.12446034e-01 -7.70315468e-01 -1.11490083e+00 -3.56433056e-02
-1.73730373e-01 2.27167774e-02 4.21180964e-01 9.71933484e-01
4.84836221e-01 2.43229333e-05 5.63451946e-01 -5.07416129e-01
-5.64373255e-01 -6.92262292e-01 -3.99966240e-01 1.05350363e+00
2.23973721e-01 -4.94592428e-01 -9.93995592e-02 9.46647286e-01]
|
[10.255189895629883, 1.2709052562713623]
|
82424fca-5070-4459-b419-a651ec311f10
|
3d-face-reconstruction-for-forensic
|
2303.11164
| null |
https://arxiv.org/abs/2303.11164v1
|
https://arxiv.org/pdf/2303.11164v1.pdf
|
3D Face Reconstruction for Forensic Recognition -- A Survey
|
3D face reconstruction algorithms from images and videos are applied to many fields, from plastic surgery to the entertainment sector, thanks to their advantageous features. However, when looking at forensic applications, 3D face reconstruction must observe strict requirements that still make unclear its possible role in bringing evidence to a lawsuit. Shedding some light on this matter is the goal of the present survey, where we start by clarifying the relation between forensic applications and biometrics. To our knowledge, no previous work adopted this relation to make the point on the state of the art. Therefore, we analyzed the achievements of 3D face reconstruction algorithms from surveillance videos and mugshot images and discussed the current obstacles that separate 3D face reconstruction from an active role in forensic applications.
|
['Gian Luca Marcialis', 'Martin Drahansky', 'Tomáš Goldmann', 'Giulia Orrù', 'Simone Maurizio La Cava']
|
2023-02-03
| null | null | null | null |
['3d-face-reconstruction', 'face-reconstruction']
|
['computer-vision', 'computer-vision']
|
[ 2.02567086e-01 1.53489083e-01 -1.42556682e-01 -3.30358028e-01
-6.83231130e-02 -3.13685626e-01 4.86038625e-01 -2.06940755e-01
-3.83865714e-01 7.07718611e-01 -3.75848450e-02 -4.98914540e-01
-3.73340994e-01 -6.03291750e-01 -3.82890075e-01 -8.00636053e-01
-4.72906269e-02 1.96041211e-01 1.47159649e-02 2.32334808e-03
5.52328467e-01 1.25865865e+00 -1.72862470e+00 2.16938496e-01
3.96923095e-01 8.12245727e-01 -2.26088315e-01 2.37346098e-01
-2.93742806e-01 3.33674103e-01 -6.61277890e-01 -1.11178541e+00
3.33329052e-01 -4.47132736e-01 -6.55176878e-01 4.06364173e-01
3.74695629e-01 -9.28224862e-01 -6.38952851e-02 1.01382160e+00
5.41806281e-01 -2.90575117e-01 5.71545660e-01 -1.08408833e+00
-4.10633028e-01 7.44169727e-02 -5.45142770e-01 3.31986040e-01
9.16431963e-01 -2.31456999e-02 2.24118546e-01 -9.48930562e-01
8.45652759e-01 1.24226415e+00 4.70318913e-01 7.38935769e-01
-8.09542954e-01 -5.64075232e-01 -5.95575050e-02 4.87391531e-01
-1.23593080e+00 -8.56259167e-01 1.04177094e+00 -5.24690807e-01
5.54662824e-01 2.53311068e-01 7.69362569e-01 1.17490149e+00
4.28229310e-02 4.63822126e-01 1.44504201e+00 -5.18583655e-01
6.44015148e-02 2.87993222e-01 2.37905234e-01 6.59092307e-01
7.03478813e-01 4.46879165e-03 -6.20615900e-01 -1.39062434e-01
8.13299596e-01 -2.23311573e-01 -2.27928922e-01 -9.50473025e-02
-3.12306225e-01 6.16851211e-01 -3.80744576e-01 5.91263115e-01
-2.23220095e-01 -3.45748335e-01 3.53804529e-01 2.46879533e-01
2.87472844e-01 -2.40245447e-01 4.31176201e-02 -1.01310834e-01
-9.08376932e-01 -1.56465992e-01 7.11305916e-01 4.78687018e-01
4.30492312e-01 -9.73140355e-03 5.98750889e-01 4.70516086e-01
4.41128194e-01 5.21359980e-01 3.98230292e-02 -9.27995622e-01
-8.70428532e-02 4.70511466e-01 -3.74852657e-01 -1.03698957e+00
2.55672671e-02 3.47642377e-02 -3.58785778e-01 5.63284874e-01
6.15308762e-01 2.37760350e-01 -2.59032488e-01 1.06210983e+00
6.53241932e-01 1.75918788e-01 -2.84938544e-01 8.78288746e-01
1.25378478e+00 7.36749917e-02 -2.45877773e-01 -6.05884731e-01
1.44829631e+00 6.40834197e-02 -9.37924802e-01 3.72392982e-01
-5.09126037e-02 -9.99163270e-01 4.33790594e-01 7.47402966e-01
-1.08404255e+00 -3.76514018e-01 -6.71429753e-01 8.45227316e-02
-1.63238630e-01 2.47909542e-04 5.94411075e-01 1.37002087e+00
-7.81696856e-01 4.61041778e-01 -7.68236518e-01 -8.50425243e-01
4.63270724e-01 3.85383964e-01 -8.77831161e-01 -9.23526660e-02
-7.49640167e-01 1.05696762e+00 -5.02076745e-02 4.62053925e-01
-3.67641360e-01 -3.26644003e-01 -5.87694585e-01 -3.71432185e-01
5.63161254e-01 -2.70348907e-01 7.29306161e-01 -6.12062275e-01
-1.66409063e+00 1.70622945e+00 -2.98858374e-01 -2.17413738e-01
5.98101854e-01 -8.01951960e-02 -6.39829636e-01 6.58076286e-01
-1.42978132e-01 -4.34333272e-03 9.82824385e-01 -1.23145425e+00
-1.62406296e-01 -8.81933451e-01 1.44249544e-01 -5.16736925e-01
-1.22541629e-01 4.83158529e-01 -3.43068153e-01 -2.58716434e-01
1.44392297e-01 -5.80978274e-01 3.29317778e-01 4.25255388e-01
-1.07624285e-01 -1.16725653e-01 1.05953050e+00 -7.46581733e-01
9.29590881e-01 -2.30985856e+00 -1.86143041e-01 1.07435480e-01
9.63493362e-02 5.57739556e-01 2.32750043e-01 6.82592034e-01
-5.43890446e-02 1.23742171e-01 -1.55594096e-01 -3.44044417e-01
-1.80207416e-01 2.09942594e-01 -3.44082624e-01 1.06278157e+00
3.53953950e-02 5.33425152e-01 -6.32078767e-01 -8.64527404e-01
4.96650338e-01 7.45934308e-01 -3.73419255e-01 8.05560872e-02
5.45225859e-01 4.94532555e-01 -6.20768249e-01 1.12096608e+00
9.75959122e-01 4.62133646e-01 3.36881280e-01 -1.61890462e-01
-3.02667022e-01 6.19970728e-03 -1.09893560e+00 1.28669715e+00
8.91984999e-02 5.52263856e-01 5.26166618e-01 -1.18182075e+00
9.42542493e-01 5.26822984e-01 7.23772287e-01 -5.82825720e-01
4.26388830e-01 3.96039695e-01 -3.28937210e-02 -9.81458426e-01
3.02222669e-01 -5.53590775e-01 4.96559799e-01 4.87650543e-01
9.73198339e-02 -2.96764560e-02 -1.67385582e-02 -2.25061953e-01
5.01888454e-01 5.04336417e-01 5.97184956e-01 -1.25471741e-01
8.91489983e-01 -3.07145923e-01 3.37337911e-01 1.99381143e-01
-5.08115768e-01 5.45846999e-01 6.95261180e-01 -5.31894624e-01
-4.90072787e-01 -1.03343344e+00 -6.53305054e-01 1.61644086e-01
-7.66727999e-02 -1.89161718e-01 -1.02775931e+00 -7.86813200e-01
-2.14607548e-02 2.95308650e-01 -6.21361434e-01 1.20322585e-01
-5.60184181e-01 -6.32389128e-01 5.49910367e-01 -1.78144574e-01
3.66636217e-01 -1.01858962e+00 -8.26119483e-01 -1.75114870e-01
-1.06077855e-02 -1.20076621e+00 1.17349483e-01 -4.39291984e-01
-1.26433146e+00 -1.61373496e+00 -5.10434806e-01 -2.02111825e-01
7.36626446e-01 3.23664725e-01 8.63534391e-01 5.66689193e-01
-3.34738314e-01 8.65520358e-01 -5.09115040e-01 -5.59709311e-01
-7.37539649e-01 -3.39886844e-01 3.75530005e-01 2.10678622e-01
6.03516936e-01 -8.37011576e-01 -2.80083060e-01 4.15319324e-01
-1.05892909e+00 -7.67032564e-01 2.67281860e-01 3.35626096e-01
2.36579999e-01 -1.73687354e-01 3.75958025e-01 -9.53407526e-01
3.39055628e-01 -2.95252174e-01 -5.68594098e-01 1.24966629e-01
-3.21404278e-01 -4.30319875e-01 3.17804456e-01 -6.55107275e-02
-1.10104191e+00 -8.63320455e-02 -5.16095400e-01 -4.95802224e-01
-4.82303292e-01 -6.60063103e-02 -3.98510456e-01 -3.07059735e-01
5.49558401e-01 3.00507635e-01 5.29168427e-01 -7.00723588e-01
-1.08010940e-01 7.02784061e-01 5.63042939e-01 -5.08825958e-01
9.41214561e-01 1.05337203e+00 3.59183818e-01 -1.49668789e+00
-4.52969283e-01 -4.25088882e-01 -9.44141746e-01 -6.91877127e-01
6.79597080e-01 -2.85120845e-01 -9.57476556e-01 3.72969121e-01
-1.31772637e+00 5.70320010e-01 -2.11008519e-01 5.60667932e-01
-5.12812614e-01 1.04568040e+00 -3.42510790e-01 -1.33074188e+00
-1.77520499e-01 -1.14965487e+00 9.36430037e-01 2.25730129e-02
-4.49222997e-02 -9.07799900e-01 -5.51627092e-02 7.59994507e-01
5.05373292e-02 4.12044734e-01 6.48653090e-01 -1.75603583e-01
-4.41414028e-01 -3.23484600e-01 1.91577926e-01 3.25646192e-01
2.81722248e-01 3.51834804e-01 -1.55243647e+00 -1.78017840e-02
6.44067168e-01 8.71103927e-02 4.53250259e-01 3.44215691e-01
8.76576424e-01 3.33011076e-02 -1.45334229e-01 6.15078330e-01
1.26864660e+00 4.36923414e-01 1.19740987e+00 1.74197942e-01
1.16429389e-01 1.27341175e+00 6.13535583e-01 4.59364682e-01
-2.69490868e-01 7.78684199e-01 7.07778513e-01 2.01668620e-01
-4.63224709e-01 -1.51952520e-01 5.00945091e-01 3.95878255e-01
-7.13123024e-01 1.24672972e-01 -3.75955611e-01 1.18895210e-01
-1.17989826e+00 -1.41970098e+00 -4.46334988e-01 2.43494105e+00
5.63843139e-02 4.04408462e-02 3.02008122e-01 6.27466738e-01
8.75457585e-01 3.29854600e-02 -1.11727111e-01 -3.35049719e-01
-1.49932057e-01 1.66044846e-01 -1.21896148e-01 3.67596716e-01
-7.25985706e-01 5.19194901e-01 6.61757612e+00 7.19525754e-01
-1.34616077e+00 7.55289495e-02 3.50415230e-01 4.81525064e-02
-1.73224241e-01 8.43802840e-02 -7.70304620e-01 4.55711812e-01
5.66444039e-01 2.77600884e-01 1.67908579e-01 6.58948302e-01
3.28173757e-01 -4.23212379e-01 -8.79041851e-01 1.20673239e+00
6.42990053e-01 -9.28813577e-01 6.66471422e-02 7.58502066e-01
-1.68808661e-02 -6.89422846e-01 9.21782181e-02 -3.10092181e-01
-7.42150009e-01 -8.81775498e-01 7.60284066e-01 4.90989745e-01
8.30852568e-01 -6.06854379e-01 6.55077696e-01 2.18049943e-01
-7.05568135e-01 2.72463739e-01 -3.82394552e-01 -1.92910030e-01
6.13048017e-01 6.87105775e-01 -1.09273386e+00 6.80094719e-01
4.97975528e-01 3.53266388e-01 -1.17622577e-01 9.35337424e-01
-2.22887561e-01 4.70540941e-01 -1.25496864e-01 2.11448282e-01
-1.48358345e-01 -5.81452250e-01 7.15490699e-01 9.40893292e-01
4.84305173e-01 4.01239872e-01 -5.24551153e-01 7.60685265e-01
3.01008135e-01 2.72310466e-01 -1.09241748e+00 3.85046899e-02
1.28180444e-01 1.13358653e+00 -1.05883098e+00 1.25809267e-01
-6.72306895e-01 9.23146009e-01 -2.56334633e-01 -1.49108022e-01
-6.06847525e-01 1.27432257e-01 7.37632096e-01 7.82001853e-01
8.12365189e-02 -3.18572760e-01 -1.60754740e-01 -1.17783761e+00
1.21976413e-01 -6.68146610e-01 3.27063709e-01 -4.73258913e-01
-1.06900930e+00 4.82307792e-01 3.33891362e-01 -1.40517163e+00
-2.01548070e-01 -8.54565740e-01 -5.38796186e-01 4.89855140e-01
-1.39019144e+00 -1.22360098e+00 3.44677903e-02 5.03655434e-01
4.32364613e-01 -2.58235693e-01 6.82554126e-01 2.64446944e-01
-4.52747583e-01 3.54294688e-01 -1.90758035e-01 -2.72253454e-02
5.87871134e-01 -5.73642254e-01 -6.40780316e-04 8.64053607e-01
1.90742627e-01 8.15771520e-01 7.66413510e-01 -6.47730470e-01
-1.78790522e+00 -9.66279656e-02 1.12750673e+00 -7.17949390e-01
2.24885017e-01 -1.79044768e-01 -7.02248096e-01 2.73774922e-01
1.01117074e-01 -1.56656541e-02 9.15970147e-01 -8.59797746e-02
-2.28120089e-01 -2.22443044e-02 -1.68576348e+00 3.27269435e-01
1.14129794e+00 -5.87055922e-01 -7.67299652e-01 2.41243243e-02
-2.25939110e-01 -1.10257581e-01 -8.69407833e-01 5.62782288e-02
1.01515400e+00 -1.67928791e+00 1.10497022e+00 -3.65910232e-01
1.70259565e-01 -5.50117418e-02 -6.35516495e-02 -3.95692825e-01
3.02514464e-01 -4.25401509e-01 1.20540023e-01 1.31108189e+00
-2.78195769e-01 -6.36379898e-01 9.03689742e-01 4.63438272e-01
-4.79159597e-03 -5.42978764e-01 -1.33993042e+00 -7.77836382e-01
-3.14174294e-02 -6.57455742e-01 5.31773448e-01 8.99753988e-01
-1.98783986e-02 -2.28840888e-01 -5.31839311e-01 3.67792696e-02
9.40033436e-01 1.41782716e-01 8.72837484e-01 -1.45218217e+00
5.92809310e-03 -2.99907357e-01 -7.58657217e-01 -4.84169453e-01
2.87653238e-01 -5.61914027e-01 -7.03079760e-01 -1.08846617e+00
1.71267405e-01 7.95421153e-02 3.69164616e-01 -6.14337809e-02
3.59355509e-01 5.17240107e-01 3.22529584e-01 1.56901330e-01
-8.27157754e-04 1.16563939e-01 1.14959347e+00 2.81386644e-01
2.06814840e-01 1.42926797e-01 -7.23876774e-01 8.03778946e-01
3.78764123e-01 -3.24041247e-01 -2.85365850e-01 -2.69460171e-01
-7.72080198e-02 1.82946086e-01 2.88129956e-01 -6.40952408e-01
3.17802615e-02 -1.39234096e-01 3.07011366e-01 -4.98268425e-01
6.23469353e-01 -1.21565092e+00 4.39314574e-01 6.66692376e-01
5.27195930e-01 -2.20224068e-01 1.90630406e-02 1.81511626e-01
-1.62851587e-01 -7.85465300e-01 8.26627910e-01 -3.16789120e-01
-4.86354589e-01 8.47457200e-02 -5.73365211e-01 -5.05132616e-01
1.17008901e+00 -1.04853821e+00 -8.34539067e-03 -4.36768800e-01
-8.55619311e-01 -6.28439307e-01 7.17841148e-01 3.18418927e-02
8.73910844e-01 -9.67308104e-01 -5.38194478e-01 4.79356974e-01
-2.92450547e-01 -5.97623765e-01 5.77974916e-01 9.57902193e-01
-6.66871428e-01 4.18679088e-01 -5.75651467e-01 -4.90467906e-01
-1.82964337e+00 7.32170165e-01 2.35953480e-01 3.18550527e-01
-5.48419893e-01 2.75898874e-01 -7.40502030e-02 -9.52529833e-02
1.53787494e-01 1.33386090e-01 -5.49856663e-01 2.38454610e-01
8.53192151e-01 7.77226031e-01 3.15366387e-01 -1.13604939e+00
-5.46052217e-01 1.00598693e+00 3.11663657e-01 -7.41841495e-02
1.29167688e+00 -1.07034318e-01 -4.08773810e-01 9.82391313e-02
1.03630435e+00 5.00856996e-01 -7.10242152e-01 2.45669812e-01
-1.16283163e-01 -1.09797704e+00 -2.81082898e-01 -2.76235729e-01
-1.31538069e+00 1.08559811e+00 4.39543962e-01 3.32186252e-01
1.27419543e+00 1.35111138e-01 4.20980066e-01 2.25937039e-01
7.93447554e-01 -7.82961786e-01 -1.72307953e-01 4.26700152e-02
9.19540584e-01 -1.04400146e+00 2.51658380e-01 -9.13599908e-01
-2.50627935e-01 1.47643900e+00 1.81358263e-01 -6.09062612e-03
7.72310495e-01 3.52498680e-01 -4.04494256e-02 -3.07687163e-01
-5.84595166e-02 -2.61108339e-01 6.08926862e-02 9.67075169e-01
5.44019401e-01 -3.30758095e-01 -9.89589274e-01 3.44546854e-01
-4.77073991e-05 3.10991406e-01 6.56282723e-01 9.89421606e-01
-3.61763090e-01 -1.59634566e+00 -1.09471774e+00 2.27452263e-01
-8.89609635e-01 3.75918984e-01 -6.67724311e-01 1.09314549e+00
3.70218903e-01 8.68146896e-01 -1.89810216e-01 -1.57924041e-01
4.23612982e-01 1.29987583e-01 9.22681868e-01 -2.33438954e-01
-4.64926511e-01 1.20725028e-01 1.50024682e-01 -4.93888617e-01
-9.17679667e-01 -1.19183958e+00 -7.04010367e-01 -5.87696493e-01
-2.30914161e-01 2.85447568e-01 9.35179830e-01 9.18909729e-01
-6.94627762e-02 -2.67444164e-01 4.13957119e-01 -7.43000925e-01
-3.50344479e-01 -6.05359614e-01 -1.01377237e+00 2.53247887e-01
1.68801188e-01 -8.41792941e-01 -3.63117337e-01 1.43883124e-01]
|
[12.810683250427246, 0.7478553056716919]
|
8cf545f8-4efe-4d1d-a282-d3243316195d
|
neural-automated-essay-scoring-and-coherence
|
1804.06898
| null |
https://arxiv.org/abs/1804.06898v3
|
https://arxiv.org/pdf/1804.06898v3.pdf
|
Neural Automated Essay Scoring and Coherence Modeling for Adversarially Crafted Input
|
We demonstrate that current state-of-the-art approaches to Automated Essay Scoring (AES) are not well-suited to capturing adversarially crafted input of grammatical but incoherent sequences of sentences. We develop a neural model of local coherence that can effectively learn connectedness features between sentences, and propose a framework for integrating and jointly training the local coherence model with a state-of-the-art AES model. We evaluate our approach against a number of baselines and experimentally demonstrate its effectiveness on both the AES task and the task of flagging adversarial input, further contributing to the development of an approach that strengthens the validity of neural essay scoring models.
|
['Ted Briscoe', 'Youmna Farag', 'Helen Yannakoudakis']
|
2018-04-18
|
neural-automated-essay-scoring-and-coherence-1
|
https://aclanthology.org/N18-1024
|
https://aclanthology.org/N18-1024.pdf
|
naacl-2018-6
|
['automated-essay-scoring']
|
['natural-language-processing']
|
[ 3.88020724e-01 2.27777570e-01 1.96807981e-01 -5.66374600e-01
-1.09493637e+00 -9.18912530e-01 7.19022632e-01 6.25916570e-02
-1.78704605e-01 6.23284280e-01 6.28591776e-01 -5.42047381e-01
-1.84822813e-01 -7.81992853e-01 -4.28970516e-01 -1.66191667e-01
1.03308052e-01 4.24731255e-01 -9.34043229e-02 -7.22676635e-01
5.66028535e-01 1.78347975e-01 -9.50022757e-01 6.31681144e-01
8.56190920e-01 6.25931203e-01 -5.96878767e-01 1.21415555e+00
2.53316276e-02 1.49633837e+00 -1.31255925e+00 -1.20611060e+00
4.65956628e-02 -8.84624004e-01 -1.16624582e+00 -4.47059929e-01
1.17533076e+00 -4.39932555e-01 -7.30698586e-01 1.12503374e+00
3.39273006e-01 7.18234405e-02 8.85681033e-01 -8.72671545e-01
-9.61063623e-01 8.00059915e-01 1.56293780e-01 5.91474295e-01
5.63947439e-01 5.28587461e-01 1.52468455e+00 -4.03769255e-01
5.77626944e-01 9.34498429e-01 1.08044159e+00 8.92393708e-01
-1.11817098e+00 -5.86515784e-01 -2.88891867e-02 2.85576254e-01
-6.33631885e-01 -5.59919894e-01 8.70107651e-01 -2.76999652e-01
9.63236749e-01 5.85539162e-01 4.52271879e-01 1.56339979e+00
3.86975527e-01 1.06838500e+00 1.31864572e+00 -3.29635501e-01
-1.10908262e-01 -3.14652413e-01 6.98046744e-01 9.88433838e-01
-2.99711972e-02 3.71516556e-01 -9.32074189e-01 -3.39177638e-01
1.87540606e-01 -6.49191976e-01 -1.55832842e-01 2.93814778e-01
-8.98321509e-01 1.15841854e+00 1.04887687e-01 5.05957246e-01
3.41099650e-02 3.16267252e-01 6.68747842e-01 8.89631927e-01
5.80283642e-01 1.15325952e+00 -9.88509655e-02 -3.74880880e-01
-1.42865551e+00 5.68291366e-01 1.14860034e+00 4.15188849e-01
1.54645756e-01 2.57498950e-01 -7.38282561e-01 2.82751143e-01
-1.40234232e-01 3.34334046e-01 4.51947957e-01 -1.07871807e+00
6.32710159e-01 4.39200997e-01 -1.63947836e-01 -1.08734357e+00
-3.99711221e-01 -5.62172413e-01 -4.83153164e-01 2.35904410e-01
5.33488750e-01 -2.82886624e-03 -3.84752691e-01 1.91884482e+00
-5.07573545e-01 3.23040664e-01 8.54130909e-02 6.69946373e-01
8.44505727e-01 2.82131433e-01 -1.20190255e-01 8.96255672e-02
9.51909006e-01 -1.06004775e+00 -6.18842483e-01 -3.18005919e-01
7.93826163e-01 -6.20747864e-01 1.40144598e+00 5.60370743e-01
-1.68616843e+00 -4.60549533e-01 -1.34319937e+00 -8.12686831e-02
-9.37597528e-02 -1.87133268e-01 4.81263101e-01 9.25897062e-01
-1.15437448e+00 1.01719964e+00 -4.50408548e-01 3.26283276e-01
4.98940647e-01 1.88920468e-01 -2.99168408e-01 1.94760948e-01
-1.56222165e+00 1.58588886e+00 -5.30665368e-02 -2.14233041e-01
-8.27841878e-01 -7.44276643e-01 -7.64820874e-01 2.11574435e-01
-9.98376235e-02 -7.52938867e-01 1.59947181e+00 -9.45391953e-01
-1.64795792e+00 9.50102866e-01 8.90101045e-02 -6.14727020e-01
4.23589140e-01 -1.81918174e-01 -3.62930238e-01 2.09050760e-01
-1.34060934e-01 2.50453986e-02 7.25502491e-01 -7.01157391e-01
1.02764785e-01 -2.38154441e-01 2.16007486e-01 1.67232320e-01
-9.23268855e-01 1.69925660e-01 3.72656018e-01 -8.36266637e-01
-6.49936318e-01 -7.24650145e-01 7.77228475e-02 -5.23111522e-01
-4.82607007e-01 -2.90778995e-01 5.03456056e-01 -8.69038105e-01
1.67713165e+00 -1.62152171e+00 2.59407163e-01 2.02805907e-01
3.24389726e-01 3.81602645e-01 -5.99454880e-01 3.12443167e-01
5.92505513e-03 1.74787909e-01 -2.76472539e-01 -6.62648678e-01
3.98031682e-01 -1.61517769e-01 -7.19266653e-01 3.25002998e-01
3.50465775e-01 1.40481925e+00 -1.02670777e+00 -2.27836415e-01
-1.15951255e-01 -1.29500955e-01 -3.49997282e-01 5.05250514e-01
-1.55255064e-01 3.42935264e-01 -1.60843030e-01 2.67695695e-01
1.75699905e-01 -1.34489372e-01 -2.17280649e-02 3.33273947e-01
3.88477504e-01 1.03507102e+00 -5.96546650e-01 1.81909430e+00
-4.06192541e-01 1.02210760e+00 -2.81844914e-01 -7.09718227e-01
7.64438391e-01 4.33319360e-01 -1.59473300e-01 -6.77301824e-01
1.87629342e-01 1.60585374e-01 7.15222210e-02 -5.26330471e-01
7.69769907e-01 -2.21177444e-01 -5.95488071e-01 1.29560304e+00
3.50992978e-01 -6.11492276e-01 3.29471529e-01 7.38756180e-01
1.77744341e+00 -2.03552619e-01 2.44696841e-01 -2.31442168e-01
8.61507833e-01 3.28283943e-02 -1.68132171e-01 1.23117912e+00
-3.66103798e-01 5.82704008e-01 7.00066745e-01 -6.16715133e-01
-1.17751586e+00 -1.07167101e+00 1.67439729e-01 1.27571964e+00
-3.06546450e-01 -5.89865983e-01 -1.06272113e+00 -1.12653446e+00
-3.65094058e-02 1.05045223e+00 -9.07415748e-01 -5.90293884e-01
-7.59221256e-01 -4.31885839e-01 1.57807088e+00 4.96050209e-01
1.50156498e-01 -1.20473337e+00 -3.75630736e-01 1.49781346e-01
-4.13981408e-01 -8.03219736e-01 -6.87138021e-01 9.37994197e-02
-6.56405807e-01 -1.15358388e+00 -3.36123183e-02 -3.66498291e-01
2.62959242e-01 -1.01254821e-01 1.95805895e+00 5.12493491e-01
-1.36888236e-01 5.36355257e-01 -3.74076553e-02 -3.65723997e-01
-1.17534196e+00 2.98666745e-01 -1.71952527e-02 -5.25793314e-01
5.56742668e-01 -5.27280807e-01 -3.82560462e-01 -1.59654677e-01
-1.01864719e+00 -4.17925596e-01 3.54617029e-01 1.07345450e+00
-4.36816663e-01 -5.86600900e-01 6.85103595e-01 -1.19943106e+00
1.70996189e+00 -3.00174773e-01 -1.32067665e-01 3.00445437e-01
-5.20390451e-01 2.98218727e-01 9.70515311e-01 -3.44317555e-01
-7.38052011e-01 -6.20774567e-01 -2.55571216e-01 -1.54271081e-01
-2.33377405e-02 5.50678194e-01 2.36870125e-01 -3.99337947e-01
1.14509583e+00 1.59718350e-01 3.46887857e-02 9.86059085e-02
5.83115816e-01 4.51645523e-01 1.07749867e+00 -6.13194764e-01
9.40475762e-01 4.36063530e-03 6.21693842e-02 2.87413504e-02
-1.20669711e+00 -1.91925362e-01 -7.32532859e-01 -2.69499153e-01
3.53483707e-01 -5.03253818e-01 -5.81488252e-01 3.42905700e-01
-1.40442121e+00 -3.39595854e-01 -3.24832380e-01 -1.88229457e-01
-7.54474521e-01 7.31778622e-01 -1.09534931e+00 -5.35578251e-01
-6.42318010e-01 -8.10530841e-01 8.42602134e-01 -6.48457110e-02
-1.00626016e+00 -1.36580276e+00 7.92253852e-01 5.68569779e-01
5.90182304e-01 1.39073238e-01 9.59284842e-01 -1.01962793e+00
-2.21263349e-01 -5.10676861e-01 2.71263748e-01 5.58014512e-01
-4.57790047e-01 -9.21982601e-02 -1.18041432e+00 -4.39806283e-02
3.31823915e-01 -1.10871255e+00 1.04583085e+00 -1.45899564e-01
1.06460500e+00 -6.24901712e-01 4.14116144e-01 4.19235975e-01
8.87377858e-01 -5.87713122e-01 7.13986576e-01 3.37249696e-01
2.71411419e-01 5.19302487e-01 1.61287770e-01 1.53726161e-01
1.85498044e-01 4.63617921e-01 4.00348723e-01 2.64191747e-01
-2.16724262e-01 -1.48742929e-01 6.20399415e-01 1.19650114e+00
1.61644265e-01 -3.71778607e-01 -6.73437893e-01 5.01169324e-01
-1.66014886e+00 -1.51133347e+00 -1.62131339e-01 1.62276018e+00
1.36729646e+00 2.74994195e-01 -3.01281046e-02 3.50627601e-01
2.06797153e-01 5.99109113e-01 -2.20129117e-01 -1.25640213e+00
-2.48263702e-01 1.03694582e+00 -3.75662558e-02 7.75534749e-01
-1.31952918e+00 8.03679407e-01 7.97399187e+00 7.06277788e-01
-6.38858140e-01 1.25824854e-01 5.65170884e-01 -2.51155704e-01
-5.64225435e-01 -5.12472689e-01 -3.77896696e-01 5.00855505e-01
1.52476680e+00 -1.81510538e-01 5.59433460e-01 5.36956906e-01
-2.10512027e-01 4.12929952e-01 -1.29353893e+00 2.33169913e-01
5.41324377e-01 -1.43214548e+00 8.29408783e-03 -3.54243726e-01
1.04214001e+00 -3.13892514e-01 5.59321105e-01 4.88505810e-01
9.51699436e-01 -1.70547867e+00 7.26564884e-01 6.12142503e-01
5.90640903e-01 -5.70888758e-01 8.86979699e-01 4.13010210e-01
-4.45657164e-01 -2.03146681e-01 -2.89150864e-01 -4.96230751e-01
-1.60195589e-01 2.06958055e-01 -5.90041339e-01 3.87348026e-01
1.20547868e-01 5.75487316e-01 -9.43713665e-01 5.02648234e-01
-8.36406946e-01 9.79361713e-01 1.82705730e-01 -2.90053338e-01
4.14233088e-01 1.73548341e-01 6.48959756e-01 1.49975312e+00
-1.01306058e-01 5.99434786e-03 -2.28870124e-01 1.12889278e+00
-5.02767205e-01 -2.09980890e-01 -6.20313764e-01 -1.83368564e-01
4.67659950e-01 1.25956523e+00 1.28528431e-01 -3.40430975e-01
-3.30652803e-01 1.07536519e+00 8.00025821e-01 -6.79466641e-03
-6.14541531e-01 -4.46119249e-01 5.11785626e-01 -4.08081174e-01
-7.77483061e-02 -1.50358677e-01 -1.15800226e+00 -1.25442278e+00
7.14694634e-02 -1.49215388e+00 5.40214002e-01 -6.09361887e-01
-1.77007902e+00 5.88857293e-01 -4.89524722e-01 -8.57815266e-01
-7.45486438e-01 -6.10765040e-01 -1.38238049e+00 1.13450015e+00
-1.35561144e+00 -1.07527256e+00 -4.49840218e-01 5.47142804e-01
3.65536094e-01 -6.21760070e-01 1.47153544e+00 -2.57108420e-01
-8.91229510e-02 1.11400604e+00 8.99429061e-03 3.29836041e-01
9.43126678e-01 -1.52585828e+00 9.47605610e-01 1.05051577e+00
5.51111460e-01 7.91060925e-01 6.99056983e-01 -4.22896504e-01
-9.39963818e-01 -7.91962385e-01 1.22665143e+00 -1.34451735e+00
1.11753488e+00 -3.72597098e-01 -9.03732538e-01 6.31844938e-01
4.98251408e-01 -3.76922995e-01 9.29611802e-01 2.51569957e-01
-8.78001809e-01 3.38685364e-01 -1.00829411e+00 7.10726440e-01
9.05860424e-01 -1.18958569e+00 -1.20216215e+00 3.30212831e-01
3.66903365e-01 -4.49925274e-01 -7.63389230e-01 1.51392832e-01
4.44717646e-01 -1.07431698e+00 9.11363602e-01 -1.21742320e+00
1.17800128e+00 3.73089075e-01 2.59156913e-01 -1.58143210e+00
-5.90231597e-01 -1.00457811e+00 -5.71872592e-01 7.74761438e-01
3.24365944e-01 -1.76999986e-01 8.97639096e-01 5.72167218e-01
-3.68725777e-01 -6.81103230e-01 -1.10146248e+00 -4.97403979e-01
8.75659466e-01 -3.02139610e-01 4.79066730e-01 9.28557813e-01
6.08059585e-01 3.83004606e-01 -2.93430388e-01 -2.25050792e-01
4.83984500e-01 -9.53705385e-02 6.83037221e-01 -1.21512222e+00
-3.94725114e-01 -1.03542256e+00 -5.21368742e-01 -7.09720194e-01
8.24149072e-01 -1.08382797e+00 -9.14508626e-02 -1.04416227e+00
3.33554059e-01 1.26536086e-01 -4.28386062e-01 2.96321988e-01
-7.70032346e-01 7.46595144e-01 2.46940449e-01 1.83131531e-01
-8.23458850e-01 3.07244927e-01 9.62727606e-01 -4.81373727e-01
4.02684331e-01 -8.64260718e-02 -1.12354290e+00 3.24346364e-01
8.09303761e-01 -6.02990627e-01 -2.54005820e-01 -4.62742478e-01
5.04311025e-01 -1.34688750e-01 4.03593630e-01 -8.97788644e-01
4.43482071e-01 6.26064390e-02 3.14553469e-01 -6.18127435e-02
1.85216188e-01 -8.46064687e-02 -6.44893169e-01 5.11301041e-01
-1.11007619e+00 2.40823463e-01 1.46746576e-01 4.25486803e-01
-2.37104058e-01 -7.59483457e-01 6.66079998e-01 -3.45236421e-01
-8.21369290e-02 -8.74038637e-02 -3.95165116e-01 4.76267368e-01
5.48203588e-01 1.01787269e-01 -9.31831002e-01 -7.52247930e-01
-3.76331866e-01 -1.24898039e-01 2.92740554e-01 3.94325525e-01
5.88638783e-01 -1.18607819e+00 -1.19053233e+00 1.97530016e-01
-1.42749473e-01 -8.38469923e-01 -3.89655158e-02 5.06552100e-01
-4.66088474e-01 7.65211880e-01 -2.55996317e-01 3.12714186e-03
-1.49542284e+00 6.71979114e-02 4.92401809e-01 -1.06062090e+00
-2.39375144e-01 1.07247615e+00 -2.49229461e-01 -7.07186639e-01
1.93834275e-01 2.03150630e-01 -3.17831308e-01 -4.01003599e-01
6.54064596e-01 3.36176991e-01 4.53604132e-01 -2.93546945e-01
-2.71609277e-02 -1.86529234e-02 -2.49284133e-01 -3.04678142e-01
1.12623107e+00 3.43516856e-01 -2.45880857e-01 3.04318190e-01
8.70140195e-01 2.52635270e-01 -7.37107098e-01 -1.45987734e-01
1.50986537e-01 -3.19840491e-01 -1.40527442e-01 -1.27779365e+00
-5.00261128e-01 1.13219357e+00 -4.43951506e-03 1.55046314e-01
7.17415333e-01 -4.15538192e-01 9.78076816e-01 5.94336271e-01
2.48099882e-02 -1.07412148e+00 9.11743820e-01 1.11319470e+00
8.16152692e-01 -7.58751929e-01 5.81285730e-02 1.03952803e-01
-7.78408945e-01 1.54158604e+00 4.41512108e-01 -5.25213599e-01
1.42459311e-02 4.36034203e-01 2.44574040e-01 -4.31334436e-01
-1.04981411e+00 3.85036290e-01 7.36857712e-01 4.15123105e-01
6.50962293e-01 2.06559803e-02 -2.29940444e-01 1.03792775e+00
-7.54122138e-01 -2.35332772e-01 8.69292378e-01 7.72933483e-01
-2.72601455e-01 -1.04925942e+00 -2.42270693e-01 5.35667956e-01
-7.96262622e-01 -6.64855897e-01 -1.10421252e+00 3.79335701e-01
-3.17656964e-01 1.06929290e+00 -1.13497175e-01 -6.63835764e-01
1.59251019e-01 6.57842398e-01 8.24667633e-01 -5.67784071e-01
-1.66293716e+00 -9.59381580e-01 4.97211009e-01 -4.74870801e-01
-7.13004395e-02 -6.97795987e-01 -6.15423679e-01 -7.44691312e-01
-2.52817333e-01 1.04921818e-01 3.89537811e-01 1.25611091e+00
1.07704118e-01 6.05930686e-01 5.89946270e-01 -3.42955291e-01
-1.46468735e+00 -1.21938694e+00 -2.51527905e-01 8.09700370e-01
2.47077242e-01 -6.08214140e-02 -4.71261203e-01 -1.29077733e-01]
|
[11.312138557434082, 9.335247993469238]
|
88720a66-9502-47a0-88e1-d7936fb38a03
|
monai-an-open-source-framework-for-deep
|
2211.02701
| null |
https://arxiv.org/abs/2211.02701v1
|
https://arxiv.org/pdf/2211.02701v1.pdf
|
MONAI: An open-source framework for deep learning in healthcare
|
Artificial Intelligence (AI) is having a tremendous impact across most areas of science. Applications of AI in healthcare have the potential to improve our ability to detect, diagnose, prognose, and intervene on human disease. For AI models to be used clinically, they need to be made safe, reproducible and robust, and the underlying software framework must be aware of the particularities (e.g. geometry, physiology, physics) of medical data being processed. This work introduces MONAI, a freely available, community-supported, and consortium-led PyTorch-based framework for deep learning in healthcare. MONAI extends PyTorch to support medical data, with a particular focus on imaging, and provide purpose-specific AI model architectures, transformations and utilities that streamline the development and deployment of medical AI models. MONAI follows best practices for software-development, providing an easy-to-use, robust, well-documented, and well-tested software framework. MONAI preserves the simple, additive, and compositional approach of its underlying PyTorch libraries. MONAI is being used by and receiving contributions from research, clinical and industrial teams from around the world, who are pursuing applications spanning nearly every aspect of healthcare.
|
['Andrew Feng', 'Sebastien Ourselin', 'Prerna Dogra', 'Stephen Aylward', 'Klaus H. Maier-Hein', 'Keyvan Farahani', 'Haris Shuaib', 'S. Kevin Zhou', 'Ralf Floca', 'David Bericat', 'Daguang Xu', 'Holger R. Roth', 'Lee A. D. Cooper', 'Justin Kirby', 'Mona Flores', 'Jayashree Kalpathy-Cramer', 'Michael Baumgartner', 'Paul F. Jaeger', 'Lena Maier-Hein', 'Andre Dourson', 'Andres Diaz-Pinto', 'Vikash Gupta', 'Barbaros S. Erdal', 'Brad Genereaux', 'Hans Johnson', 'Benjamin Gorman', 'Yunguan Fu', 'Yipeng Hu', 'Yiwen Li', 'Guotai Wang', 'Tom Vercauteren', 'Marc Modat', 'Charlie Budd', 'Mohammad Zalbagi Darestani', 'Sachidanand Alle', 'Behrooz Hashemian', 'Michael Zephyr', 'Isaac Yang', 'Yucheng Tang', 'Mingxin Zheng', 'Yun Liu', 'Wentao Zhu', 'Ali Hatamizadeh', 'Ziyue Xu', 'Yufan He', 'Vishwesh Nath', 'Dong Yang', 'Can Zhao', 'Andriy Myronenko', 'Benjamin Murrey', 'Yiheng Wang', 'Eric Kerfoot', 'Nic Ma', 'Richard Brown', 'Wenqi Li', 'M. Jorge Cardoso']
|
2022-11-04
| null | null | null | null |
['medical-image-detection', 'medical-image-registration']
|
['computer-vision', 'medical']
|
[-3.54718976e-02 1.69099346e-01 8.19674227e-03 -1.70799032e-01
-6.29109383e-01 -2.80406386e-01 2.77006596e-01 4.43240792e-01
-4.01845761e-02 2.99487442e-01 3.82281661e-01 -5.50506175e-01
-1.62910521e-01 -7.48415291e-01 -4.12318170e-01 -6.18515134e-01
-1.67356417e-01 8.71997178e-01 -8.74644052e-03 -6.18454143e-02
-2.90855855e-01 7.50781894e-01 -1.04722095e+00 4.83932853e-01
6.03101730e-01 7.56799519e-01 -2.35661805e-01 9.44474638e-01
2.11393222e-01 9.86656547e-01 -2.22208709e-01 -7.26671070e-02
1.68403015e-01 -4.62400973e-01 -7.87207603e-01 -1.87531441e-01
1.08480126e-01 -1.53622761e-01 -5.86671904e-02 3.93295914e-01
6.32073402e-01 -4.29828078e-01 5.56287587e-01 -9.57529664e-01
-3.06096852e-01 2.60308951e-01 -4.66568433e-02 -3.82785611e-02
1.68881997e-01 6.26287103e-01 4.47024196e-01 -3.63947511e-01
7.23496556e-01 8.27931702e-01 8.18840265e-01 5.56930184e-01
-9.89479959e-01 -4.25920159e-01 -3.46321970e-01 1.12139238e-02
-1.28215611e+00 -3.74241650e-01 3.47569406e-01 -6.71013892e-01
9.63755429e-01 6.75580502e-01 9.91071105e-01 8.47344637e-01
7.47188687e-01 7.28547394e-01 8.42391908e-01 -2.91115999e-01
5.01638532e-01 6.76811785e-02 3.30316164e-02 8.16679120e-01
1.38310105e-01 -2.44196206e-01 -1.71122491e-01 -4.46028531e-01
8.15924525e-01 1.56455845e-01 -5.82140163e-02 -5.62970757e-01
-1.53188801e+00 5.59150934e-01 4.34395492e-01 5.01146674e-01
-8.10810983e-01 2.57335484e-01 5.23045063e-01 -5.05292835e-03
2.09392875e-01 6.06289327e-01 -6.53491616e-01 -2.87684023e-01
-6.68046653e-01 4.35672641e-01 7.79468596e-01 6.29808247e-01
6.25625402e-02 -6.75953850e-02 -3.96380760e-02 5.45001090e-01
2.62328386e-01 4.06763911e-01 5.53313017e-01 -1.20255744e+00
-3.44649464e-01 9.07792568e-01 -2.30647132e-01 -1.06776190e+00
-8.75168860e-01 -6.88877165e-01 -9.31335866e-01 2.44766891e-01
8.14892203e-02 6.59698322e-02 -7.64504373e-01 1.13678908e+00
4.78152215e-01 1.55471936e-01 1.26405269e-01 9.59345222e-01
1.14298415e+00 2.90565401e-01 4.26089287e-01 1.52052164e-01
1.35331559e+00 -6.76526129e-01 -2.85519242e-01 -2.07891092e-01
1.03249478e+00 -5.43846250e-01 9.22750771e-01 7.81189203e-01
-1.37954128e+00 -1.42930374e-01 -7.47640312e-01 -2.42183030e-01
-3.76824558e-01 -2.84278035e-01 6.70575798e-01 4.79879141e-01
-9.77247179e-01 4.05745089e-01 -1.24219477e+00 -4.88737285e-01
6.28840506e-01 4.09235358e-01 -4.44968224e-01 -2.61910170e-01
-8.54845345e-01 1.10115457e+00 2.15797678e-01 -1.37363508e-01
-8.35988879e-01 -1.14336348e+00 -7.40380168e-01 -1.47804663e-01
-3.62932011e-02 -1.36999917e+00 1.35367286e+00 -8.89217734e-01
-1.03301334e+00 9.72968280e-01 2.62797982e-01 -5.98809004e-01
5.86052001e-01 2.18198802e-02 -3.14782321e-01 7.84869790e-02
-1.37477264e-01 6.15236998e-01 3.11843120e-02 -8.11415195e-01
-3.60829353e-01 -4.83569324e-01 -3.54388326e-01 2.24813536e-01
-1.44751459e-01 1.35602072e-01 -4.46454197e-01 -3.25072259e-01
-2.42639810e-01 -9.64354575e-01 -5.53208530e-01 4.04223680e-01
-3.21127236e-01 3.18912625e-01 5.44366360e-01 -8.68638933e-01
9.95008826e-01 -2.10149717e+00 8.93825442e-02 3.14422965e-01
6.27598047e-01 4.15967435e-01 2.67659903e-01 4.16513711e-01
5.22788465e-02 -6.15042001e-02 -4.04230088e-01 1.14802673e-01
-1.71694160e-01 1.89486235e-01 2.41915122e-01 5.01031935e-01
-4.92479056e-02 9.00335789e-01 -8.23563933e-01 -5.38533032e-01
5.82189023e-01 7.92050004e-01 -4.92401749e-01 1.48419924e-02
-3.41717005e-01 7.13791668e-01 -5.22792578e-01 5.90922356e-01
1.70930833e-01 -5.53292394e-01 1.37511551e-01 -1.77666798e-01
-3.54598388e-02 1.53066158e-01 -8.54894340e-01 1.54919910e+00
-6.02401614e-01 2.93862581e-01 1.62847638e-01 -7.57783473e-01
6.00290418e-01 6.15847945e-01 1.00632715e+00 -4.55633432e-01
3.26935172e-01 3.91088068e-01 3.39756757e-01 -6.66772485e-01
-1.69408187e-01 -5.28083220e-02 9.16632339e-02 5.80385625e-01
-3.67279112e-01 -2.55518764e-01 9.41495299e-02 1.75198555e-01
1.33260739e+00 -3.87404971e-02 7.45250642e-01 -2.78762341e-01
4.36549366e-01 4.14456993e-01 3.26726586e-01 2.34927997e-01
-2.31793672e-01 6.10551953e-01 2.27533162e-01 -8.54809701e-01
-1.20547724e+00 -9.39545810e-01 -4.87245888e-01 7.14585304e-01
-3.96472633e-01 -1.77505314e-01 -7.58601606e-01 -4.18466270e-01
6.43844381e-02 6.37388468e-01 -5.55979729e-01 -1.58860892e-01
-4.39238340e-01 -6.70331061e-01 4.79783982e-01 6.82179213e-01
2.33048171e-01 -1.24115944e+00 -1.12642956e+00 5.27763188e-01
9.84181091e-02 -9.46651757e-01 -1.95750237e-01 -1.97727606e-01
-8.88144314e-01 -1.12567532e+00 -8.15038443e-01 -6.72653019e-01
4.20500308e-01 -3.65106255e-01 1.12903726e+00 1.13848127e-01
-9.51210320e-01 7.53974438e-01 -7.60077871e-03 -8.41617703e-01
-7.72427976e-01 -2.78528109e-02 -2.71755129e-01 -3.53908777e-01
2.75943816e-01 -2.95545399e-01 -7.63841033e-01 4.75205859e-04
-9.24023449e-01 4.64350641e-01 5.30717432e-01 5.11564434e-01
7.39327967e-01 -2.67856628e-01 4.77218330e-01 -9.42725778e-01
5.03237426e-01 -6.14827156e-01 -6.23914823e-02 1.61398709e-01
-6.49734676e-01 -2.50863522e-01 5.17660618e-01 2.75422679e-03
-4.54662800e-01 2.09286407e-01 -4.70541894e-01 -1.29197329e-01
-5.18006444e-01 7.75035322e-01 -8.21049064e-02 -4.22025733e-02
8.22906733e-01 -4.26937714e-02 4.39326227e-01 -1.61257327e-01
2.69541949e-01 6.90324366e-01 5.76675832e-01 -2.69265890e-01
-2.10562330e-02 5.38925707e-01 1.62205637e-01 -8.50301385e-01
-2.83322603e-01 -3.80520880e-01 -2.58188963e-01 -1.65200695e-01
9.09305692e-01 -5.91168761e-01 -5.57319522e-01 5.13723314e-01
-8.67869496e-01 -4.92503107e-01 -2.67396420e-01 3.65424901e-01
-4.26282436e-01 -1.40130790e-02 -4.80405301e-01 -3.91976982e-01
-9.65932369e-01 -1.31140769e+00 8.82330418e-01 7.05389902e-02
-7.32448399e-01 -1.24621701e+00 1.59318298e-01 3.70546937e-01
7.61408091e-01 6.53561056e-01 1.12117374e+00 -6.82816863e-01
-4.97061104e-01 -4.16949511e-01 -9.01746675e-02 1.59811571e-01
1.40350368e-02 2.97068536e-01 -7.98729837e-01 1.00585043e-01
-1.68910131e-01 -4.24270481e-02 1.82879657e-01 6.80827379e-01
1.24614191e+00 -3.55648696e-01 -5.99179149e-01 4.98406589e-01
1.02184248e+00 3.35065126e-01 5.36890805e-01 3.04479957e-01
5.71367085e-01 4.41219062e-01 5.33567853e-02 4.13629502e-01
6.85609758e-01 5.15149653e-01 3.02136719e-01 -4.69928622e-01
-1.50512576e-01 5.26266873e-01 -4.94330935e-02 7.67594993e-01
1.40696943e-01 2.34651640e-01 -1.50448799e+00 6.09757423e-01
-1.85022557e+00 -6.01580799e-01 -4.03081089e-01 2.06269145e+00
9.94968176e-01 8.65637884e-02 1.43900797e-01 -1.25853010e-02
-2.38678381e-02 -5.44341683e-01 -6.57287598e-01 -7.63909698e-01
2.47148827e-01 2.01881737e-01 3.48541200e-01 3.32389921e-01
-9.72169340e-01 5.00640988e-01 6.46117735e+00 2.51465797e-01
-1.46344125e+00 9.89457741e-02 8.53201032e-01 -3.79615054e-02
-2.06704110e-01 -5.81813812e-01 -6.34544566e-02 2.86316156e-01
1.30902076e+00 -4.84561712e-01 3.19821417e-01 9.45927620e-01
4.96627092e-01 4.28461172e-02 -1.16516972e+00 4.77430195e-01
-7.10788369e-02 -1.78899062e+00 -4.44288254e-01 1.33631872e-02
4.34966475e-01 3.38820606e-01 -5.91221601e-02 -1.16083577e-01
4.39708650e-01 -1.23573887e+00 3.47018540e-01 8.05565774e-01
8.07339370e-01 -5.77475846e-01 8.68074238e-01 3.22307914e-01
-7.62296796e-01 5.16134650e-02 2.16610521e-01 1.05620056e-01
-4.71384600e-02 7.20837831e-01 -1.14755166e+00 4.79142338e-01
8.52300406e-01 6.45381331e-01 -4.52598482e-01 1.27423489e+00
2.02429846e-01 7.01693594e-01 -2.82031864e-01 1.48472145e-01
7.03741470e-03 1.75804093e-01 2.99981505e-01 1.41197777e+00
9.01329052e-03 9.11822319e-02 3.53020549e-01 6.36913002e-01
3.06109726e-01 3.09416711e-01 -5.23800433e-01 -1.64288700e-01
1.62776813e-01 1.31138754e+00 -7.36808360e-01 -3.41430575e-01
-3.46342355e-01 4.91789162e-01 -4.15748209e-01 -1.10376380e-01
-6.61222816e-01 -1.04420267e-01 6.82448566e-01 4.70744014e-01
-2.66619205e-01 -5.16246678e-03 -8.99665236e-01 -5.46376526e-01
-2.79203445e-01 -1.40701532e+00 4.86652017e-01 -7.76517630e-01
-1.14357424e+00 6.26996934e-01 1.33560151e-01 -9.03560996e-01
-5.18065035e-01 -7.03491271e-01 -5.04988551e-01 8.43609929e-01
-1.00686026e+00 -1.38637030e+00 -4.87764359e-01 4.57376331e-01
9.22655910e-02 -1.76354647e-01 1.37787044e+00 4.14205313e-01
-5.29057920e-01 2.00090781e-01 -8.87844414e-02 1.58566594e-01
5.36188304e-01 -1.11758244e+00 4.68610942e-01 2.22802669e-01
-2.64786601e-01 7.44996727e-01 7.04147637e-01 -5.86880565e-01
-1.64625788e+00 -1.12312865e+00 7.45697320e-01 -4.20963734e-01
6.37356281e-01 6.88633174e-02 -8.13963473e-01 7.21170902e-01
-7.55064338e-02 5.51733114e-02 9.35805976e-01 -8.47185403e-02
-1.65458827e-03 -3.48579019e-01 -1.35805643e+00 5.53674698e-01
4.23161447e-01 -2.16549829e-01 -4.43884790e-01 6.32347226e-01
3.29539657e-01 -7.18780756e-01 -1.18881071e+00 4.16254193e-01
7.56842196e-01 -7.86526859e-01 1.03022611e+00 -4.92636681e-01
5.07902920e-01 -1.24371119e-01 3.59064817e-01 -1.06720746e+00
-2.74250329e-01 -3.76110792e-01 -2.73933798e-01 5.49569666e-01
4.79050547e-01 -8.04405749e-01 7.81065524e-01 9.65247571e-01
-5.07065773e-01 -1.20775259e+00 -5.73231339e-01 -2.30323762e-01
3.15463781e-01 -5.55728257e-01 6.62100852e-01 9.57153976e-01
2.28511736e-01 -6.12257197e-02 1.21214725e-01 5.70472591e-02
2.46261179e-01 -1.42773569e-01 8.21022272e-01 -1.23396051e+00
-3.85239184e-01 -7.07518637e-01 -7.01785505e-01 -3.71268019e-02
-5.77426076e-01 -9.44432974e-01 -1.29382119e-01 -2.16144609e+00
1.15020618e-01 -5.34353256e-01 -1.27824754e-01 8.64344895e-01
8.94855708e-02 4.01208282e-01 -8.37774053e-02 2.94276625e-01
-1.41315371e-01 -2.22151443e-01 1.21819925e+00 -9.66781825e-02
-8.11969712e-02 -1.38098449e-01 -9.24727499e-01 7.34206378e-01
8.74441028e-01 -2.37731174e-01 -1.69233814e-01 -2.75473297e-01
8.40653926e-02 -2.79827744e-01 5.64484000e-01 -1.32110000e+00
1.37768298e-01 -1.54409260e-01 3.91528994e-01 -1.25842050e-01
9.02299434e-02 -1.00089562e+00 5.63514888e-01 8.53079975e-01
-3.63506407e-01 3.71765997e-03 4.00607944e-01 -3.21359724e-01
5.75861521e-02 1.10155441e-01 7.77055979e-01 -3.75561088e-01
-4.16761577e-01 3.34898114e-01 -2.84155220e-01 6.59731217e-04
1.43010068e+00 -9.58406180e-02 -1.76028207e-01 -1.88963652e-01
-6.33567631e-01 3.44734102e-01 5.75177729e-01 1.68180183e-01
4.11901385e-01 -8.08312416e-01 -1.01284122e+00 3.59488517e-01
1.73253804e-01 2.31760964e-01 3.50780338e-01 1.00146186e+00
-1.18446147e+00 4.14508641e-01 -2.86706150e-01 -6.45548344e-01
-1.22663426e+00 5.73579490e-01 6.23468816e-01 -2.94837594e-01
-1.11960125e+00 4.62162584e-01 7.21186772e-02 -5.31826735e-01
2.02399001e-01 -3.74801815e-01 1.99829251e-01 -4.48801845e-01
8.01502943e-01 3.76375556e-01 5.16448438e-01 -3.82995605e-01
-6.02903903e-01 3.11137855e-01 -1.23992659e-01 9.72412676e-02
1.56323421e+00 3.39084327e-01 -1.14165105e-01 4.43241626e-01
7.07358003e-01 -2.75659800e-01 -8.57044697e-01 3.18908989e-01
-1.76170200e-01 2.03123465e-01 3.39802116e-01 -1.36639333e+00
-1.09170270e+00 9.16144729e-01 5.25475740e-01 2.09775403e-01
1.10049844e+00 7.97808841e-02 9.22678530e-01 -3.37121449e-02
2.31083646e-01 -7.95906246e-01 -3.79423141e-01 7.49818757e-02
9.52307343e-01 -7.65359938e-01 3.71106595e-01 -1.69072568e-01
-9.74425495e-01 1.06171298e+00 3.13674569e-01 1.18550681e-01
7.78721035e-01 6.46214068e-01 5.55031121e-01 -4.09749836e-01
-7.53081024e-01 1.37759402e-01 4.73803133e-01 6.89290166e-01
9.57232535e-01 2.54444599e-01 -1.99845478e-01 3.76245469e-01
-1.54788122e-01 6.00665033e-01 2.69013941e-01 1.04318225e+00
-1.78985909e-01 -1.01572430e+00 -4.97892648e-01 5.84625185e-01
-3.86726916e-01 -1.21921055e-01 -3.04788768e-01 7.54806578e-01
2.36858904e-01 4.60881233e-01 -1.33545965e-01 1.54960090e-02
2.69534916e-01 1.02721445e-01 3.42038542e-01 -5.88143528e-01
-9.30483162e-01 1.28375128e-01 1.10543154e-01 -6.16604447e-01
-5.61331622e-02 -7.22547054e-01 -1.71641290e+00 -2.59206116e-01
4.50093836e-01 -1.12062700e-01 9.83967781e-01 9.31149721e-01
8.57074559e-01 7.89370656e-01 3.46001680e-03 -5.10475397e-01
-3.90684009e-02 -6.65521622e-01 -1.72443777e-01 1.48016214e-01
1.61561400e-01 -1.38769582e-01 2.53461152e-01 6.03530034e-02]
|
[14.7726411819458, -2.3516926765441895]
|
4eebddca-05d6-4877-a848-9e8e57bd4010
|
squeezeseg-convolutional-neural-nets-with
|
1710.07368
| null |
http://arxiv.org/abs/1710.07368v1
|
http://arxiv.org/pdf/1710.07368v1.pdf
|
SqueezeSeg: Convolutional Neural Nets with Recurrent CRF for Real-Time Road-Object Segmentation from 3D LiDAR Point Cloud
|
In this paper, we address semantic segmentation of road-objects from 3D LiDAR
point clouds. In particular, we wish to detect and categorize instances of
interest, such as cars, pedestrians and cyclists. We formulate this problem as
a point- wise classification problem, and propose an end-to-end pipeline called
SqueezeSeg based on convolutional neural networks (CNN): the CNN takes a
transformed LiDAR point cloud as input and directly outputs a point-wise label
map, which is then refined by a conditional random field (CRF) implemented as a
recurrent layer. Instance-level labels are then obtained by conventional
clustering algorithms. Our CNN model is trained on LiDAR point clouds from the
KITTI dataset, and our point-wise segmentation labels are derived from 3D
bounding boxes from KITTI. To obtain extra training data, we built a LiDAR
simulator into Grand Theft Auto V (GTA-V), a popular video game, to synthesize
large amounts of realistic training data. Our experiments show that SqueezeSeg
achieves high accuracy with astonishingly fast and stable runtime (8.7 ms per
frame), highly desirable for autonomous driving applications. Furthermore,
additionally training on synthesized data boosts validation accuracy on
real-world data. Our source code and synthesized data will be open-sourced.
|
['Xiangyu Yue', 'Bichen Wu', 'Kurt Keutzer', 'Alvin Wan']
|
2017-10-19
| null | null | null | null |
['robust-3d-semantic-segmentation']
|
['computer-vision']
|
[ 1.93771627e-02 -4.36000936e-02 -1.26647592e-01 -8.43289733e-01
-9.91731346e-01 -4.42157269e-01 4.85467315e-01 -9.70484242e-02
-5.15291572e-01 4.18621480e-01 -6.41307414e-01 -4.73240972e-01
3.81613135e-01 -1.20248532e+00 -1.14909112e+00 -1.84481949e-01
-1.15269810e-01 1.02471161e+00 7.16283619e-01 -7.57472068e-02
1.30902156e-01 8.27310026e-01 -2.09866309e+00 6.61204085e-02
1.01084244e+00 1.16970098e+00 2.92860657e-01 7.69489944e-01
-2.60724813e-01 4.75455850e-01 -3.23793501e-01 -3.78825754e-01
4.44019347e-01 3.83929431e-01 -6.73698545e-01 8.70911703e-02
9.09465909e-01 -1.69983014e-01 -7.81078786e-02 7.88043022e-01
1.08929984e-01 1.63434684e-01 5.76634824e-01 -1.53050125e+00
-7.55636916e-02 1.94036618e-01 -4.42573905e-01 -6.45415112e-02
-4.17983085e-01 4.09402907e-01 8.54683459e-01 -1.03071404e+00
4.71898288e-01 1.40469134e+00 7.23433733e-01 3.21101189e-01
-1.15052164e+00 -1.05264461e+00 1.19884163e-01 2.12729111e-01
-1.73193610e+00 -2.72586524e-01 6.60260439e-01 -5.71442544e-01
9.42937374e-01 4.94037429e-03 6.73372626e-01 7.93327689e-01
7.93385431e-02 7.87342250e-01 7.60959268e-01 1.48520648e-01
3.57240379e-01 -6.64856061e-02 1.35313019e-01 7.27523029e-01
-6.69610575e-02 3.04796696e-01 -1.91128463e-01 2.50995576e-01
5.45795083e-01 1.67145452e-03 4.40977693e-01 -5.47293186e-01
-1.08748662e+00 9.01461482e-01 8.78781736e-01 -4.93062973e-01
-2.99123883e-01 5.11770010e-01 2.62242019e-01 -1.78200662e-01
4.87942398e-01 -2.70847052e-01 -4.99399871e-01 -3.38190980e-02
-1.06024861e+00 5.62488317e-01 4.72736895e-01 1.30799961e+00
1.33051527e+00 -1.67619124e-01 1.42358363e-01 6.83257043e-01
3.21942687e-01 7.79102206e-01 -1.39041916e-01 -1.31187129e+00
5.65959752e-01 6.01790249e-01 -2.69883196e-03 -7.77585924e-01
-2.95378745e-01 -3.07812214e-01 -5.04143536e-01 4.67677116e-01
1.93425521e-01 3.54605876e-02 -1.28583443e+00 1.43282497e+00
4.43158597e-01 6.02505684e-01 -1.66714013e-01 9.97942567e-01
9.31360781e-01 6.51624501e-01 3.53447527e-01 5.45365393e-01
1.26523089e+00 -9.53357875e-01 -8.52378905e-02 -5.52340388e-01
6.30690277e-01 -4.56308186e-01 8.74474347e-01 1.58864647e-01
-8.65828574e-01 -1.17494619e+00 -8.34784091e-01 -3.22036624e-01
-5.53529799e-01 4.00124520e-01 4.84839410e-01 4.95881736e-01
-1.05216217e+00 5.26470184e-01 -1.12529647e+00 -1.98952869e-01
8.01172912e-01 4.17027444e-01 -1.48582488e-01 -1.33560896e-01
-8.32047462e-01 8.11134160e-01 4.36123490e-01 1.90757498e-01
-9.66344476e-01 -7.04437852e-01 -1.18572295e+00 -1.35457695e-01
4.74021763e-01 -5.48908710e-01 1.35800397e+00 -3.90823007e-01
-1.35994101e+00 1.16801488e+00 -2.06642866e-01 -7.47189999e-01
5.52207172e-01 -1.68488577e-01 -2.26999030e-01 -2.84563564e-02
6.03862226e-01 1.55154121e+00 7.33566940e-01 -1.28188646e+00
-1.19634080e+00 -4.01964128e-01 -2.16001168e-01 -1.25939503e-01
4.92506802e-01 -2.54839867e-01 -7.15665221e-01 8.81258994e-02
1.76771671e-01 -1.14716077e+00 -5.04263163e-01 -2.07439624e-02
-4.82992291e-01 -4.69813943e-01 1.10834956e+00 -2.47942924e-01
4.17184412e-01 -2.08559656e+00 -3.77902210e-01 3.20161968e-01
2.19902292e-01 2.73101121e-01 -2.64849942e-02 -1.10237196e-01
-1.13132717e-02 2.71228757e-02 -5.69567621e-01 -6.50252700e-01
7.94663802e-02 4.91290450e-01 -3.75450194e-01 3.82332027e-01
6.58058345e-01 1.25388956e+00 -8.13616455e-01 -6.36601746e-01
7.06491470e-01 5.43390334e-01 -5.61595559e-01 7.67366961e-02
-5.78680038e-01 4.18080091e-01 -3.36833596e-01 6.89071417e-01
1.02936089e+00 -8.84229913e-02 -3.76524627e-01 8.58423766e-03
-3.16537201e-01 3.65111381e-01 -1.02326798e+00 1.65900433e+00
-5.55020332e-01 7.07897902e-01 -1.49767622e-01 -8.65114510e-01
1.39136565e+00 -4.25476432e-01 3.60468924e-01 -6.39034986e-01
1.72751293e-01 1.44814670e-01 -4.14233059e-01 -1.48191795e-01
8.46251726e-01 2.07740664e-01 -3.75605792e-01 2.79198196e-02
1.40814855e-05 -5.91527581e-01 1.04614772e-01 1.47831053e-01
8.44769180e-01 3.98836166e-01 -3.17710489e-01 -1.35133386e-01
3.87842864e-01 5.69142163e-01 5.59886634e-01 6.06503606e-01
-1.59311160e-01 7.43353248e-01 2.78657645e-01 -6.35744691e-01
-1.12525904e+00 -1.14863777e+00 -3.07968676e-01 8.66395175e-01
3.35383832e-01 -4.59142983e-01 -8.07874799e-01 -6.36098206e-01
2.53128201e-01 8.20079148e-01 -2.98898578e-01 6.17157072e-02
-7.74107993e-01 -1.60524264e-01 4.44299728e-01 7.34979153e-01
6.66900992e-01 -1.02019346e+00 -9.07769263e-01 3.23404640e-01
-5.01487665e-02 -1.64911592e+00 -7.86913335e-02 2.92996466e-01
-7.77113736e-01 -1.05447435e+00 -1.30536720e-01 -7.87535846e-01
3.46932769e-01 4.01752025e-01 1.33139837e+00 -2.53657829e-02
-3.42955828e-01 5.27040437e-02 -1.90581325e-02 -4.49969292e-01
-1.72649175e-01 4.18777227e-01 -2.92745948e-01 -4.89165783e-02
7.59131014e-01 -5.23206413e-01 -5.54655254e-01 5.17157972e-01
-3.75105351e-01 1.73362702e-01 4.32621688e-01 3.13582748e-01
9.50744927e-01 -7.91235417e-02 1.29345626e-01 -7.71633685e-01
-8.68744403e-02 -2.74308413e-01 -1.15815282e+00 -4.28890228e-01
-1.26888648e-01 -2.86929548e-01 3.10525537e-01 4.45705876e-02
-7.53074586e-01 6.10714018e-01 -5.05546689e-01 -7.97673821e-01
-8.22832942e-01 1.66781947e-01 -1.31493017e-01 9.30580720e-02
6.19717479e-01 1.46443900e-02 -5.70954755e-02 -3.19239348e-01
7.72406220e-01 8.18696916e-01 1.15964091e+00 -6.98207319e-01
9.60845947e-01 6.75794542e-01 -4.26408276e-02 -7.21153438e-01
-7.79869795e-01 -4.97373432e-01 -1.02473903e+00 -3.05191666e-01
1.11770570e+00 -1.28902578e+00 -9.50692892e-01 2.98203558e-01
-1.25674295e+00 -7.24182904e-01 -2.36829311e-01 3.67989719e-01
-7.41900206e-01 -8.19398090e-02 -3.82364005e-01 -6.52846694e-01
-8.80493745e-02 -1.34363782e+00 1.63405383e+00 2.21315920e-01
6.05834797e-02 -5.19448400e-01 -1.77498266e-01 5.53657234e-01
4.65269387e-02 4.01313037e-01 4.89823312e-01 -3.57844889e-01
-1.30814362e+00 -2.93869406e-01 -5.00426233e-01 2.69615352e-01
-3.89503717e-01 3.62397820e-01 -1.13983560e+00 8.10855329e-02
-5.57599962e-01 -3.20309043e-01 1.06453753e+00 4.06089485e-01
1.49640858e+00 3.68452162e-01 -5.91205776e-01 7.67972410e-01
1.18447602e+00 -3.95892002e-02 5.95092654e-01 1.30600914e-01
9.27663267e-01 5.91526270e-01 9.04693067e-01 4.13820706e-02
9.20854926e-01 7.01251388e-01 7.84577906e-01 -1.51204556e-01
-1.59411058e-01 -4.93206054e-01 5.47985435e-02 4.63250011e-01
1.20485552e-01 2.86604147e-02 -1.20849836e+00 6.98498011e-01
-1.80656731e+00 -7.09418595e-01 -5.75933397e-01 2.01706362e+00
3.58876795e-01 5.05747259e-01 2.86450505e-01 -4.61254455e-03
8.71482193e-01 -9.24825519e-02 -5.43018639e-01 -2.68525094e-01
2.55488634e-01 5.22842348e-01 8.49703133e-01 6.28018618e-01
-1.34474719e+00 1.67494941e+00 5.20513153e+00 9.08214152e-01
-1.22553480e+00 1.05484761e-01 7.82692790e-01 1.88237414e-01
1.09457880e-01 6.60051405e-02 -1.16478384e+00 4.15283263e-01
1.16682148e+00 2.22280428e-01 9.02421214e-03 1.24192929e+00
2.66995192e-01 -1.24833800e-01 -9.39938962e-01 8.63815010e-01
-4.39924896e-01 -1.43072903e+00 -3.17818195e-01 2.76407003e-01
5.01977444e-01 7.73303390e-01 -8.41202810e-02 6.34284198e-01
8.26424420e-01 -1.05163503e+00 9.20358717e-01 1.92905918e-01
9.60127056e-01 -1.12013698e+00 5.16888380e-01 6.41118348e-01
-1.50095832e+00 1.52386054e-01 -6.20306969e-01 -9.81380045e-02
2.86884248e-01 7.13027477e-01 -1.10988557e+00 3.62723112e-01
8.41618121e-01 9.33469594e-01 -6.31849289e-01 9.45598364e-01
-3.58802229e-01 6.42245710e-01 -6.18707716e-01 2.14347273e-01
4.89159614e-01 -2.71882206e-01 1.48212627e-01 1.31107712e+00
1.65286213e-01 1.13044098e-01 4.92878616e-01 1.23096395e+00
-6.25272691e-02 -2.80031323e-01 -7.32869446e-01 4.85984057e-01
5.01371861e-01 1.53362751e+00 -1.17983675e+00 -5.04894555e-01
-1.26614660e-01 5.50662339e-01 3.78674895e-01 1.30616441e-01
-1.12574029e+00 -3.21251810e-01 9.28098619e-01 2.22578645e-01
5.94491780e-01 -5.56885898e-01 -5.23393035e-01 -7.84916341e-01
-1.04849651e-01 -2.15096250e-01 -9.92692038e-02 -9.23267543e-01
-1.10935092e+00 6.10890269e-01 6.55355444e-03 -1.13277876e+00
-1.77794501e-01 -5.77944398e-01 -5.83425343e-01 8.44613075e-01
-1.74479377e+00 -1.28852224e+00 -7.72792041e-01 5.11110544e-01
6.26669645e-01 2.16734454e-01 4.20506239e-01 3.38602901e-01
-4.51040208e-01 1.56625047e-01 -5.23269773e-01 4.12975520e-01
2.85421044e-01 -1.16477644e+00 1.35382628e+00 6.16134584e-01
7.98545703e-02 1.26325101e-01 2.42424563e-01 -8.67615521e-01
-9.20736432e-01 -1.96524715e+00 9.49905694e-01 -6.93517864e-01
4.05528992e-01 -7.15977967e-01 -8.72167587e-01 8.11659276e-01
-3.27355415e-01 4.39158559e-01 1.41718864e-01 -1.36374563e-01
-2.29131490e-01 -1.83677480e-01 -1.05499709e+00 3.64645779e-01
1.39615583e+00 -5.37550926e-01 -3.05227458e-01 4.30767208e-01
9.63505208e-01 -7.24063694e-01 -5.89431643e-01 5.64694166e-01
4.38990220e-02 -9.26721454e-01 1.13948917e+00 -3.49295318e-01
4.63895798e-01 -6.55507028e-01 -2.05661058e-01 -1.03595734e+00
-9.42685753e-02 -1.77053839e-01 3.17357510e-01 9.79897380e-01
3.43172818e-01 -3.65432352e-01 1.25872588e+00 5.60338616e-01
-6.31955147e-01 -4.71722662e-01 -1.01710939e+00 -6.82014048e-01
1.67728826e-01 -1.23844397e+00 8.21530521e-01 5.65715790e-01
-6.87801957e-01 3.76923114e-01 1.09870337e-01 4.20000404e-01
8.56034696e-01 2.61146247e-01 1.33928967e+00 -1.51526725e+00
3.76904756e-01 -2.14737967e-01 -6.68978810e-01 -1.35203481e+00
5.37821710e-01 -1.10847437e+00 4.23232585e-01 -1.41067195e+00
-3.67141724e-01 -1.02548504e+00 2.84849554e-01 5.78822553e-01
7.69271553e-02 6.36881173e-01 1.84942663e-01 1.49235368e-01
-7.13518202e-01 6.27134442e-01 9.24372792e-01 -3.35293978e-01
-1.63805053e-01 4.33188528e-01 -1.89465806e-01 7.28542447e-01
8.08180511e-01 -5.87616086e-01 -2.32059628e-01 -3.72926593e-01
-1.57964855e-01 -8.11325479e-03 9.32378352e-01 -1.32187653e+00
3.06477934e-01 -1.23499162e-01 2.42214024e-01 -1.53766751e+00
6.60918891e-01 -8.10399413e-01 3.92131601e-03 2.08181649e-01
-6.06430583e-02 -4.88362573e-02 2.58659840e-01 3.89723033e-01
-8.82341042e-02 1.57100871e-01 7.66153455e-01 -6.52391836e-02
-9.61360931e-01 6.74760997e-01 -2.88821399e-01 -1.33230850e-01
1.10474873e+00 -4.15200412e-01 4.83330116e-02 -9.70930979e-02
-6.40369952e-01 5.89964092e-01 3.94856334e-01 5.46701729e-01
6.65308535e-01 -1.17754424e+00 -7.08698392e-01 4.38647717e-01
2.18464181e-01 9.52073991e-01 1.75786614e-01 3.35723311e-01
-7.23605037e-01 4.40847397e-01 -9.47916284e-02 -1.52362883e+00
-9.25511718e-01 3.77897739e-01 3.71452719e-01 1.87106431e-01
-9.01375890e-01 7.62632310e-01 -2.96184327e-02 -9.60011542e-01
1.34982377e-01 -7.37864554e-01 -4.41633491e-03 -1.53666794e-01
1.82834163e-01 1.34676889e-01 3.00719708e-01 -8.75893354e-01
-4.64062810e-01 7.48072088e-01 1.24122836e-01 -4.72107790e-02
1.18880463e+00 1.13088921e-01 2.82174945e-01 1.41396314e-01
1.20290589e+00 -4.37678903e-01 -1.67737663e+00 -2.50680387e-01
1.01066865e-01 -4.78902131e-01 6.39076307e-02 -3.93667340e-01
-1.24078143e+00 1.11410379e+00 5.76342702e-01 -1.56126291e-01
5.07518113e-01 1.66919053e-01 9.73296344e-01 4.87582892e-01
6.73334956e-01 -8.98265362e-01 -2.71534622e-01 7.10209668e-01
3.02962929e-01 -1.28307211e+00 -3.93927783e-01 -6.93655014e-01
-5.44350743e-01 8.87035310e-01 8.13515007e-01 -4.79090154e-01
6.69672251e-01 3.57910722e-01 2.83175945e-01 -2.23382935e-01
-5.58572650e-01 -6.81703329e-01 -2.70124665e-03 8.87007117e-01
-1.97889894e-01 3.33473265e-01 4.52069491e-01 2.71758407e-01
-8.09080541e-01 1.64094791e-01 2.15900227e-01 7.26611137e-01
-6.86015248e-01 -8.74633431e-01 -3.92490417e-01 2.91909486e-01
1.83951974e-01 1.25548154e-01 -1.49857640e-01 8.83851945e-01
6.86081648e-01 9.08589423e-01 7.69270718e-01 -5.67142546e-01
4.36855346e-01 -9.79966521e-02 -4.10033204e-02 -7.40402281e-01
-3.28211963e-01 -1.60085842e-01 -5.22682369e-02 -6.97664261e-01
-2.37260371e-01 -6.54666424e-01 -1.61303210e+00 -4.63499129e-01
-1.10801034e-01 1.27033368e-01 1.16240096e+00 8.81113470e-01
5.38418472e-01 4.95104522e-01 5.57841241e-01 -1.40710580e+00
1.09452128e-01 -6.49242759e-01 -1.53683096e-01 1.57313213e-01
1.34495094e-01 -7.61860788e-01 7.79905245e-02 7.28311017e-02]
|
[8.103917121887207, -2.5702691078186035]
|
8399cfb9-61ea-42da-87d2-a2008e73999f
|
deformable-registration-using-average
|
1907.09670
| null |
https://arxiv.org/abs/1907.09670v1
|
https://arxiv.org/pdf/1907.09670v1.pdf
|
Deformable Registration Using Average Geometric Transformations for Brain MR Images
|
Accurate registration of medical images is vital for doctor's diagnosis and quantitative analysis. In this paper, we propose a new deformable medical image registration method based on average geometric transformations and VoxelMorph CNN architecture. We compute the differential geometric information including Jacobian determinant(JD) and the curl vector(CV) of diffeomorphic registration field and use them as multi-channel of VoxelMorph CNN for second train. In addition, we use the average transformation to construct a standard brain MRI atlas which can be used as fixed image. We verify our method on two datasets including ADNI dataset and MRBrainS18 Challenge dataset, and obtain excellent improvement on MR image registration with average Dice scores and non-negative Jacobian locations compared with MIT's original method. The experimental results show the method can achieve better performance in brain MRI diagnosis.
|
['Zicong Zhou', 'Yongpei Zhu', 'Guojun Liao', 'Kehong Yuan']
|
2019-07-23
| null | null | null | null |
['deformable-medical-image-registration']
|
['medical']
|
[-1.35895044e-01 -3.04038879e-02 1.99716672e-01 -6.05258167e-01
-7.32799768e-01 -3.23083639e-01 2.99808741e-01 -8.06154460e-02
-7.09719539e-01 5.47872782e-01 2.42081195e-01 5.68004809e-02
3.80505100e-02 -6.26507699e-01 -4.27780330e-01 -7.50534534e-01
-3.28307629e-01 6.45379245e-01 4.86460388e-01 -3.16122860e-01
1.39820904e-01 4.14236367e-01 -4.38272446e-01 -1.98845804e-01
9.46264684e-01 8.35584283e-01 1.59503877e-01 1.06899813e-01
2.65382975e-01 3.13463598e-01 -2.61661679e-01 -3.48521054e-01
5.48864961e-01 -2.24022985e-01 -1.12519264e+00 -3.42663884e-01
3.99822950e-01 -5.56328237e-01 -4.07118201e-01 1.41203809e+00
8.23122561e-01 2.09440216e-01 6.60226405e-01 -8.39719653e-01
-9.84969795e-01 5.60272932e-01 -7.35512853e-01 7.77611077e-01
-1.36313498e-01 2.24049121e-01 3.91297311e-01 -6.64754152e-01
7.73336530e-01 9.69231904e-01 9.55224752e-01 4.38944638e-01
-8.83024871e-01 -8.11670184e-01 -5.92544436e-01 2.28118137e-01
-1.45799470e+00 -4.68596257e-02 6.07316375e-01 -5.10104358e-01
7.19913244e-01 1.92194641e-01 7.01758027e-01 3.72145921e-01
1.08216250e+00 1.06279716e-01 1.29869092e+00 2.33926609e-01
-8.28651115e-02 -7.71680117e-01 7.48430938e-02 8.24240863e-01
1.42450735e-01 -4.43831272e-02 4.27172899e-01 -1.32896736e-01
1.26168561e+00 1.48154646e-01 -3.41648340e-01 2.18617078e-02
-1.67975068e+00 6.48579061e-01 9.76003528e-01 6.77964926e-01
-5.21070480e-01 1.46821469e-01 2.30020240e-01 2.26299584e-01
5.74714601e-01 1.66380763e-01 1.19628338e-02 1.41565517e-01
-7.29672253e-01 1.84258670e-01 1.80237129e-01 6.11282647e-01
4.24707890e-01 -8.11268762e-02 -3.50921065e-01 8.83913577e-01
3.37313473e-01 5.12539625e-01 1.11683774e+00 -7.67173350e-01
2.86087722e-01 5.81860125e-01 -5.24712563e-01 -1.13730967e+00
-7.97306240e-01 -4.55760300e-01 -1.36111546e+00 7.75049627e-02
2.62603521e-01 3.74455657e-03 -1.05327141e+00 1.43574739e+00
4.32830364e-01 4.92714256e-01 -3.34910035e-01 1.12821162e+00
1.00387943e+00 7.70291463e-02 4.94117104e-02 -1.09290220e-01
1.34546947e+00 -8.44160259e-01 -7.63426065e-01 2.61292338e-01
7.80980468e-01 -7.11816132e-01 7.87577987e-01 -1.61283433e-01
-1.15985477e+00 -2.21038863e-01 -9.96679246e-01 -2.65252143e-01
-6.08145334e-02 -8.31123590e-02 6.10572636e-01 3.09895247e-01
-1.29657197e+00 8.80050778e-01 -1.40684855e+00 8.76385719e-02
7.70132303e-01 7.90873647e-01 -7.27295876e-01 1.44693956e-01
-1.05413413e+00 1.05477726e+00 7.90194348e-02 1.65635929e-01
-7.66227841e-01 -1.20732665e+00 -7.15547621e-01 -4.05000776e-01
-4.03363079e-01 -5.73962331e-01 8.99950981e-01 -3.21593463e-01
-1.23081112e+00 1.10789156e+00 2.83577800e-01 -2.82129586e-01
6.23273075e-01 3.18821669e-01 -3.63413066e-01 2.51810819e-01
2.13984072e-01 6.36178732e-01 2.83225477e-01 -7.24869549e-01
1.30940139e-01 -8.38780999e-01 -2.19490007e-01 1.99588165e-01
3.01985890e-02 2.30787501e-01 -8.44078213e-02 -7.86849141e-01
6.09361768e-01 -1.11890376e+00 -4.86915827e-01 1.74108088e-01
-3.71381670e-01 1.05896831e-01 6.95233226e-01 -1.31763268e+00
6.51761055e-01 -1.89892280e+00 3.82730365e-02 3.14687997e-01
8.10394287e-01 1.38039365e-01 -1.53752863e-01 -5.49820244e-01
-2.86203682e-01 7.61565715e-02 -6.24067366e-01 -2.97643188e-02
-5.54498672e-01 -4.66287397e-02 3.24786812e-01 9.92996037e-01
-1.00751020e-01 1.39814365e+00 -7.65975296e-01 -6.34878039e-01
4.37576547e-02 8.08757484e-01 -4.59937721e-01 1.11539803e-01
6.80287361e-01 1.12686121e+00 -5.28255045e-01 3.65676105e-01
1.02691829e+00 -1.39492959e-01 -1.90131620e-01 -6.76906645e-01
1.90481514e-01 -6.62165508e-02 -7.25750566e-01 1.87232959e+00
-1.40843466e-01 3.34987432e-01 -1.01876572e-01 -8.63264799e-01
7.57412374e-01 3.45556259e-01 9.45441067e-01 -7.99867928e-01
5.88950396e-01 2.25645334e-01 4.22800481e-01 -3.64975840e-01
-1.66851699e-01 -1.50425091e-01 3.82855445e-01 6.08666182e-01
1.06257774e-01 -2.58106291e-01 -1.89513892e-01 -5.42556681e-02
1.09414947e+00 -1.95192620e-01 7.32588023e-02 -8.46447229e-01
6.54755890e-01 -2.90070891e-01 4.00061518e-01 1.15716718e-01
-4.97634649e-01 9.54873085e-01 1.81023672e-01 -6.76511228e-01
-1.18968594e+00 -1.26885772e+00 -5.00753045e-01 3.11180741e-01
1.34698361e-01 1.33635581e-01 -1.17799056e+00 -6.47357345e-01
-2.83344328e-01 -5.89559525e-02 -6.97190642e-01 -2.68889785e-01
-1.03380597e+00 -1.30575681e+00 6.68695271e-01 5.05533338e-01
9.11729097e-01 -9.04243946e-01 -2.84659922e-01 5.27977385e-02
-4.12973106e-01 -9.89453554e-01 -1.16374564e+00 -4.60986644e-01
-1.19329381e+00 -1.05444002e+00 -1.03702021e+00 -1.08418608e+00
1.00871849e+00 1.27988622e-01 6.56787097e-01 3.91788512e-01
-6.43678308e-01 -2.75665149e-02 2.15473194e-02 5.51661663e-02
-3.08647394e-01 2.71363221e-02 1.53096035e-01 -2.32872099e-01
-1.98939934e-01 -9.42150116e-01 -1.18378234e+00 4.59820598e-01
-9.46396708e-01 -1.14705205e-01 4.65149850e-01 6.62998974e-01
9.11462665e-01 -2.62815356e-01 3.16036373e-01 -5.91293991e-01
8.74414206e-01 -2.91192979e-01 -4.55236971e-01 2.66635358e-01
-7.34408379e-01 1.61570590e-02 2.65188128e-01 -4.61116672e-01
-6.27986073e-01 -1.32661119e-01 -3.38594735e-01 -2.66908646e-01
3.08311641e-01 3.05485159e-01 3.11907142e-01 -8.35424721e-01
6.04672849e-01 2.11618185e-01 3.20516706e-01 -3.51821333e-01
6.00372255e-02 3.57995152e-01 8.31339419e-01 -3.61161202e-01
8.73434126e-01 6.37584448e-01 1.93660945e-01 -3.16948950e-01
-2.36449331e-01 -1.57150343e-01 -1.14450955e+00 -1.08302444e-01
1.28205073e+00 -7.00546265e-01 -7.36851513e-01 5.98327875e-01
-1.08640802e+00 -3.32646102e-01 -1.11877080e-02 8.45575690e-01
-3.34492922e-01 5.76946139e-01 -8.16957772e-01 2.59919733e-01
-1.09631443e+00 -1.69673491e+00 8.95447075e-01 9.84637886e-02
1.14761956e-01 -1.27974701e+00 3.27124536e-01 2.12531865e-01
8.92827868e-01 6.10651553e-01 8.40539277e-01 -5.26200533e-01
-2.60108888e-01 -1.57898188e-01 -3.68177503e-01 2.48884737e-01
4.05535400e-01 -6.17069840e-01 -4.67740178e-01 -3.92369837e-01
2.61485606e-01 8.46302211e-02 5.59028029e-01 5.68243682e-01
1.37410676e+00 -1.49168894e-01 -1.51828378e-01 1.06580770e+00
1.42297935e+00 2.80855924e-01 8.79589140e-01 1.04010127e-01
1.12291765e+00 7.92689398e-02 7.48573467e-02 -3.68144512e-02
7.23198235e-01 7.96877205e-01 2.68425435e-01 -3.49995881e-01
-4.97127205e-01 3.58143628e-01 3.92956547e-02 1.52226937e+00
-7.00907290e-01 6.15563333e-01 -1.15805864e+00 4.19337481e-01
-1.41794062e+00 -7.07369208e-01 -3.91117126e-01 2.09554577e+00
1.08905733e+00 -3.89592350e-01 -2.10223854e-01 -3.90693069e-01
7.16224015e-01 -2.11057305e-01 -4.76283222e-01 2.41653621e-01
4.89585996e-02 6.03529215e-01 7.28816330e-01 5.88993251e-01
-1.06831181e+00 7.01744020e-01 6.76471663e+00 5.69276094e-01
-1.40905654e+00 8.54897141e-01 7.31028318e-01 2.57758558e-01
-2.80653000e-01 -3.28678370e-01 -2.21996680e-01 5.07914007e-01
7.10567892e-01 -2.24057481e-01 5.11736870e-01 3.13426495e-01
2.96474565e-02 1.54324099e-01 -7.72478819e-01 1.06372678e+00
-3.18215936e-02 -1.26408446e+00 -6.05277345e-02 2.55305707e-01
6.71197653e-01 4.67709571e-01 1.09912030e-01 -2.04083994e-01
2.66086966e-01 -1.30492198e+00 1.12368084e-01 6.39088988e-01
1.06841040e+00 -5.41003525e-01 9.34503973e-01 -2.48445883e-01
-1.13288355e+00 6.34254813e-01 -4.76119071e-01 4.09298569e-01
7.92280883e-02 4.56006587e-01 -7.88060009e-01 5.40668964e-01
7.23416448e-01 6.97281539e-01 -6.34484410e-01 1.18766224e+00
1.90065354e-01 3.19562525e-01 -9.22207311e-02 5.47021627e-01
7.61900246e-02 -6.65743232e-01 3.17590237e-01 8.82675827e-01
4.05194312e-01 4.35658514e-01 1.54965937e-01 9.37073052e-01
-2.08660677e-01 4.39244896e-01 -3.36042017e-01 3.99172157e-01
-2.07330491e-02 1.67608821e+00 -1.01461279e+00 -1.60844818e-01
-1.37246281e-01 9.73983109e-01 9.97979864e-02 5.82683086e-03
-9.03181076e-01 -2.87124962e-01 4.04027611e-01 1.64078385e-01
-4.15315539e-01 -4.33946669e-01 -3.08558255e-01 -1.29536653e+00
1.11690154e-02 -5.16361475e-01 7.50474334e-02 -6.81339562e-01
-1.22698987e+00 8.83216560e-01 8.28223526e-02 -1.03135693e+00
1.96988449e-01 -2.89263278e-01 -8.61453652e-01 8.84673715e-01
-1.35491407e+00 -1.17396832e+00 -5.37230730e-01 9.19202864e-01
-6.92783371e-02 -1.91514164e-01 7.63523757e-01 8.13160896e-01
-2.75180966e-01 6.45289123e-01 4.47706021e-02 6.61189377e-01
7.64114141e-01 -1.11303949e+00 5.27957082e-01 5.24393737e-01
-2.90961385e-01 7.95584142e-01 1.68976799e-01 -9.07938540e-01
-1.25477815e+00 -1.26502931e+00 5.06122231e-01 -3.89308631e-01
7.27605700e-01 1.06229760e-01 -9.60593700e-01 9.57901478e-01
1.55251577e-01 5.71498275e-01 5.06675959e-01 -6.83701038e-01
-6.05334304e-02 -4.50318903e-02 -1.76967013e+00 3.84148329e-01
1.04526997e+00 -2.25327954e-01 -6.04303300e-01 7.95588672e-01
8.95747125e-01 -8.54309797e-01 -1.67043972e+00 5.95945895e-01
5.40610135e-01 -4.02675867e-01 1.19622314e+00 -4.05598164e-01
3.72059494e-01 -1.75049111e-01 1.34507224e-01 -1.48238575e+00
-4.49554086e-01 -3.07423860e-01 6.62513494e-01 7.63373733e-01
1.29616693e-01 -9.00968552e-01 1.99071437e-01 6.23512805e-01
-4.32898611e-01 -9.26292717e-01 -1.35478735e+00 -6.98217988e-01
5.49843013e-01 7.89467916e-02 8.48043680e-01 1.32279158e+00
-3.58393043e-01 -1.91527188e-01 -6.58799484e-02 4.37396159e-03
8.34609509e-01 -4.16088492e-01 1.53202951e-01 -1.28401637e+00
2.43357807e-01 -3.98570180e-01 -8.06573212e-01 -3.22660446e-01
1.32429063e-01 -1.60331368e+00 -1.81466013e-01 -1.41599846e+00
5.15359282e-01 -5.73405981e-01 -5.21658897e-01 6.21379256e-01
1.29827499e-01 9.01165426e-01 -1.56854898e-01 4.44647431e-01
-2.58183151e-01 4.48457181e-01 1.99701858e+00 -3.32060546e-01
-1.06510781e-01 -4.80667263e-01 -3.57068002e-01 6.12417638e-01
8.57725859e-01 -5.06982863e-01 -1.85251266e-01 -5.84468007e-01
-2.75461584e-01 -6.41582459e-02 5.96127927e-01 -1.15038848e+00
1.01850748e-01 5.83347566e-02 5.71576595e-01 -1.24861337e-01
-1.46848723e-01 -6.81416571e-01 2.73447812e-01 7.54792154e-01
-2.25848883e-01 6.28168643e-01 -6.14082217e-02 -1.14664502e-01
4.06482331e-02 2.37979248e-01 1.07805908e+00 -1.22134566e-01
-4.39872080e-03 1.07637727e+00 1.48678780e-01 8.36851820e-02
9.58880842e-01 1.18836708e-01 -3.49677682e-01 -3.30896652e-03
-8.79279792e-01 -3.17370333e-02 3.46632987e-01 3.95323873e-01
7.98204601e-01 -1.85589802e+00 -8.77521634e-01 5.32354042e-02
-3.35644215e-01 -6.57071471e-02 3.21289241e-01 1.72773528e+00
-1.09301376e+00 2.16189042e-01 -8.23278725e-01 -5.85743070e-01
-1.11193824e+00 3.04134507e-02 8.26499462e-01 -4.03220803e-01
-9.39718544e-01 5.61798751e-01 1.31300226e-01 -6.24710202e-01
-4.03299481e-01 -6.34658515e-01 -2.98674524e-01 -4.93665367e-01
5.66597939e-01 2.01608762e-01 3.63640755e-01 -1.16573715e+00
-5.84067047e-01 9.89111423e-01 -1.89317629e-01 -1.21697392e-02
1.58958316e+00 -5.66505268e-02 -8.44557822e-01 -1.56523257e-01
1.56793547e+00 -1.98519170e-01 -8.64116251e-01 -2.76583195e-01
-3.88125569e-01 -2.25768477e-01 3.18871528e-01 -6.40935361e-01
-1.77931643e+00 8.50699246e-01 1.49414980e+00 -2.05192789e-01
8.24913502e-01 1.95171069e-02 1.32516646e+00 2.81605497e-02
4.85652596e-01 -5.79347253e-01 -1.83987975e-01 3.06122869e-01
1.08340168e+00 -1.23348939e+00 1.08700983e-01 -3.27067435e-01
-5.58260918e-01 9.67696249e-01 5.26149333e-01 -6.50243342e-01
1.04485393e+00 2.84934014e-01 2.87321806e-01 -4.56572384e-01
1.83675617e-01 3.14825982e-01 7.32731819e-01 6.25755906e-01
5.24212241e-01 1.86527580e-01 -7.13135064e-01 5.24283111e-01
-4.63400185e-01 -5.65283187e-02 2.46872410e-01 4.90830302e-01
-1.87575802e-01 -9.75911558e-01 -8.59880075e-02 6.07023478e-01
-7.39253223e-01 -2.02420145e-01 -5.85083254e-02 5.49591124e-01
8.20726380e-02 2.39567533e-01 1.51983067e-01 -3.66170466e-01
5.84465079e-02 -2.49009132e-01 8.57232392e-01 -3.78656596e-01
-7.89070249e-01 -1.17343605e-01 -6.41894162e-01 -7.55334735e-01
-4.80015367e-01 -4.39764708e-01 -1.79313707e+00 -2.25221515e-01
2.79897675e-02 -2.00585827e-01 7.81519294e-01 1.20261610e+00
2.66905218e-01 5.07025123e-01 5.22208691e-01 -8.16072404e-01
-4.19275939e-01 -9.95224118e-01 -4.53840643e-01 6.39797390e-01
1.75455406e-01 -7.24197209e-01 1.02632113e-01 6.27926365e-02]
|
[14.008770942687988, -2.5443954467773438]
|
b4a1f214-cf8f-4375-b983-367f178efe4c
|
secure-and-privacy-preserving-automated-end
|
2211.07643
| null |
https://arxiv.org/abs/2211.07643v1
|
https://arxiv.org/pdf/2211.07643v1.pdf
|
Secure and Privacy-Preserving Automated End-to-End Integrated IoT-Edge-Artificial Intelligence-Blockchain Monitoring System for Diabetes Mellitus Prediction
|
Diabetes Mellitus, one of the leading causes of death worldwide, has no cure till date and can lead to severe health complications, such as retinopathy, limb amputation, cardiovascular diseases, and neuronal disease, if left untreated. Consequently, it becomes crucial to take precautionary measures to avoid/predict the occurrence of diabetes. Machine learning approaches have been proposed and evaluated in the literature for diabetes prediction. This paper proposes an IoT-edge-Artificial Intelligence (AI)-blockchain system for diabetes prediction based on risk factors. The proposed system is underpinned by the blockchain to obtain a cohesive view of the risk factors data from patients across different hospitals and to ensure security and privacy of the user data. Furthermore, we provide a comparative analysis of different medical sensors, devices, and methods to measure and collect the risk factors values in the system. Numerical experiments and comparative analysis were carried out between our proposed system, using the most accurate random forest (RF) model, and the two most used state-of-the-art machine learning approaches, Logistic Regression (LR) and Support Vector Machine (SVM), using three real-life diabetes datasets. The results show that the proposed system using RF predicts diabetes with 4.57% more accuracy on average compared to LR and SVM, with 2.87 times more execution time. Data balancing without feature selection does not show significant improvement. The performance is improved by 1.14% and 0.02% after feature selection for PIMA Indian and Sylhet datasets respectively, while it reduces by 0.89% for MIMIC III.
|
['Rajiv Janardhanan', 'Priya Ranjan', 'Juma Al Kaabi', 'Huned Materwala', 'Alain Hennebelle', 'Leila Ismail']
|
2022-11-13
| null | null | null | null |
['diabetes-prediction']
|
['medical']
|
[ 3.62690955e-01 -1.29236430e-01 -6.29018962e-01 -3.77451450e-01
-2.28428274e-01 -1.27974689e-01 4.17880386e-01 7.07449496e-01
-1.97509438e-01 1.13609183e+00 1.17021389e-01 -7.13387191e-01
-3.80584270e-01 -8.27589810e-01 -3.86357576e-01 -9.20159757e-01
-1.41328976e-01 5.05555451e-01 -2.83905774e-01 1.02106854e-01
3.49456459e-01 1.71127290e-01 -1.21834612e+00 2.91135967e-01
1.01817191e+00 1.41288149e+00 -5.75190961e-01 4.98523623e-01
2.03496307e-01 9.63313520e-01 -3.84391844e-01 -2.80229360e-01
4.19891447e-01 -3.98668081e-01 -1.02730483e-01 -4.95857179e-01
-1.21603735e-01 -3.47952694e-01 1.75301671e-01 7.36204982e-01
5.46728492e-01 -8.73242319e-01 8.55115771e-01 -1.55170918e+00
-3.68875235e-01 5.04420400e-01 -5.68093121e-01 -1.88153461e-01
1.53344765e-01 2.38567237e-02 2.73488283e-01 5.54621127e-03
4.55632478e-01 9.16453898e-01 7.18667507e-01 3.48524839e-01
-1.19085038e+00 -9.92052853e-01 -5.10974348e-01 4.24000412e-01
-9.78245616e-01 -1.74711362e-01 5.48932552e-01 -6.31332040e-01
8.02473247e-01 5.25003493e-01 9.09861505e-01 8.33393514e-01
8.87103140e-01 2.40453795e-01 1.73181343e+00 -5.33378780e-01
3.49006414e-01 3.74047637e-01 3.78724486e-01 5.80803156e-01
9.17842805e-01 3.73262823e-01 -2.42242649e-01 -5.90459347e-01
4.96075600e-01 3.24038297e-01 2.82102406e-01 -3.10176402e-01
-1.26890588e+00 8.04125488e-01 1.08124986e-01 -7.68512394e-03
-5.64996183e-01 -1.57238901e-01 5.41052878e-01 4.41235423e-01
7.07441047e-02 -1.50707215e-01 -9.71238852e-01 8.54375437e-02
-4.96952504e-01 1.74857274e-01 9.41526413e-01 6.82811081e-01
1.37355924e-01 -1.41890302e-01 1.83116525e-01 2.34649852e-01
6.70326531e-01 7.02647507e-01 3.17144424e-01 -6.09723210e-01
1.78633720e-01 1.10783792e+00 6.01382032e-02 -9.30508196e-01
-5.90092897e-01 -4.78999645e-01 -1.40068209e+00 3.91138971e-01
4.54769701e-01 -3.02214950e-01 -9.87678230e-01 1.05942535e+00
3.88498485e-01 3.98162045e-02 4.20208812e-01 5.59438705e-01
5.14585197e-01 4.42016512e-01 4.29065555e-01 -6.72404468e-01
1.42903018e+00 -4.59147304e-01 -6.32983267e-01 2.97620118e-01
4.89993513e-01 -6.90526605e-01 1.62693650e-01 9.44845557e-01
-5.94154656e-01 -3.92210446e-02 -1.00965798e+00 3.28711718e-01
-2.95303166e-01 1.84951015e-02 7.85533667e-01 9.70821500e-01
-3.14026564e-01 3.51118416e-01 -7.27437556e-01 -6.73224509e-01
6.59687400e-01 6.71081126e-01 -7.93209821e-02 -1.83447301e-01
-9.88575816e-01 9.51345146e-01 9.45230350e-02 -3.13008845e-01
-3.87584478e-01 -7.53223300e-01 -1.77073434e-01 -5.56771040e-01
-2.06139579e-01 -9.59088266e-01 4.72960263e-01 -8.99333119e-01
-1.16647005e+00 6.13124311e-01 1.41008571e-01 -8.52084458e-01
8.11576247e-01 -1.99143857e-01 -7.40520477e-01 -1.78747132e-01
-2.54915804e-01 1.42351463e-01 4.33592409e-01 -8.03561389e-01
-8.95917594e-01 -8.19922745e-01 -4.59785312e-01 -3.52949053e-01
2.57468551e-01 1.37806296e-01 5.16626596e-01 -5.13804972e-01
-6.93593249e-02 -1.08044565e+00 -3.09771150e-01 -1.16874417e-02
-3.82046372e-01 1.15062922e-01 5.49224436e-01 -9.70801234e-01
1.21827424e+00 -1.62201428e+00 -1.83514461e-01 5.21138012e-01
-5.50287105e-02 2.49974027e-01 6.07948124e-01 3.72978806e-01
9.46606100e-02 4.06034989e-03 -5.19352183e-02 6.17818594e-01
-4.01042283e-01 2.90688008e-01 6.93090186e-02 6.49246216e-01
-1.17509454e-01 4.63199854e-01 -4.24647063e-01 -5.21462739e-01
4.78017151e-01 5.88247538e-01 -3.11130613e-01 -1.33329749e-01
-5.63177988e-02 6.97193205e-01 -5.72458088e-01 1.17528307e+00
5.65132260e-01 1.43284453e-02 5.57258904e-01 -4.57162797e-01
-1.48628086e-01 -1.31292433e-01 -1.07217836e+00 1.07425368e+00
3.34947854e-02 1.42876590e-02 -2.89327204e-01 -1.25797653e+00
1.13569009e+00 4.91789609e-01 8.64484787e-01 -9.24187183e-01
5.34186661e-01 3.88179630e-01 2.64258116e-01 -8.77235115e-01
-6.17158055e-01 -1.08722255e-01 -9.51935723e-02 1.64199337e-01
-4.76720303e-01 8.73874545e-01 -1.24936119e-01 -2.50545830e-01
1.20047069e+00 4.32266146e-02 9.18953240e-01 -2.59954512e-01
6.43131375e-01 4.59673107e-01 8.56255472e-01 3.13900888e-01
-5.01761496e-01 -3.93477678e-02 4.95566219e-01 -1.01038277e+00
-9.52213168e-01 -8.08331668e-01 -4.20793861e-01 4.05866176e-01
1.83159038e-02 5.76422550e-02 -3.63107502e-01 -6.78041101e-01
6.51884198e-01 5.91854334e-01 -5.44482768e-01 1.01177007e-01
-4.58341569e-01 -1.31011856e+00 3.22866857e-01 1.39348373e-01
6.38068140e-01 -5.54620743e-01 -9.00696874e-01 3.56818795e-01
2.03970805e-01 -5.49714208e-01 7.56465554e-01 2.12685362e-01
-1.19905496e+00 -1.23133397e+00 -1.98310181e-01 -4.78654653e-01
5.16071558e-01 -4.87711012e-01 8.36430967e-01 4.14857455e-02
-3.90502900e-01 -4.56645727e-01 -3.30297619e-01 -7.34763026e-01
-6.49300039e-01 -2.29039639e-01 1.42803386e-01 -2.27412451e-02
7.27072537e-01 -6.45238936e-01 -1.05471575e+00 3.38816911e-01
-2.68580168e-01 1.43391490e-01 1.05834067e+00 5.19233465e-01
4.01488453e-01 -9.98012163e-03 9.70050752e-01 -1.10424995e+00
4.39575128e-02 -8.56845498e-01 -5.54767966e-01 2.24063963e-01
-1.53390992e+00 2.62079053e-02 3.66133213e-01 -2.02861175e-01
-4.83242482e-01 2.74003088e-01 3.05644542e-01 1.62587225e-01
-3.53578985e-01 3.33152413e-01 -2.77084950e-02 3.42412777e-02
4.67400849e-01 -2.89249280e-03 3.45371813e-01 -5.65104783e-01
-4.62464616e-02 1.28221202e+00 7.89055526e-02 -4.78967167e-02
1.28280729e-01 4.13488746e-01 4.37478900e-01 -2.84246296e-01
-1.89050570e-01 -2.10154548e-01 -2.38827959e-01 -1.48106784e-01
8.26973081e-01 -9.42759633e-01 -1.05505323e+00 5.34227490e-01
-6.24927819e-01 2.23226190e-01 3.39945883e-01 6.77480578e-01
-3.44338804e-01 -2.05328837e-01 -3.35825145e-01 -9.77532029e-01
-8.17978203e-01 -8.40794444e-01 3.33573252e-01 4.16848511e-02
-4.86914515e-01 -6.23099983e-01 6.12534292e-04 7.84861386e-01
5.63648880e-01 1.00761914e+00 1.46318412e+00 -7.20598578e-01
-4.86102194e-01 -3.59039396e-01 -2.36644715e-01 1.32624924e-01
3.56317788e-01 1.00076012e-01 -4.68717933e-01 -1.68385163e-01
-1.74497172e-01 1.31917551e-01 3.50439280e-01 4.63007033e-01
3.90245765e-01 -7.77065516e-01 -6.86712503e-01 2.64591128e-01
1.82991004e+00 8.21636856e-01 6.41529202e-01 4.62800324e-01
3.63925010e-01 3.29420418e-01 6.52606070e-01 6.53160870e-01
4.12846923e-01 4.89355654e-01 5.43822348e-01 3.44279297e-02
3.93126272e-02 2.03708217e-01 1.74031720e-01 5.51321566e-01
-3.49799722e-01 8.38986412e-02 -1.02403998e+00 2.86100209e-01
-1.91856718e+00 -6.53053999e-01 -6.46352828e-01 2.22843838e+00
8.05613577e-01 1.71154253e-02 5.23776054e-01 6.21475875e-01
5.21988690e-01 -6.50626481e-01 -6.54867113e-01 -6.49458945e-01
3.03291559e-01 2.18590632e-01 9.89944160e-01 5.05496077e-02
-9.74919736e-01 1.03928797e-01 5.52779961e+00 5.39450571e-02
-1.40033388e+00 7.83261433e-02 7.96032429e-01 -8.29413906e-02
2.58980811e-01 -1.04081491e-02 -4.02854562e-01 8.62097621e-01
1.12976921e+00 -4.85762618e-02 2.97709197e-01 6.58739746e-01
4.41527396e-01 -3.13708544e-01 -9.08146560e-01 7.98222721e-01
-2.55179346e-01 -1.12580705e+00 -7.54557103e-02 9.89205390e-02
6.40960217e-01 -9.06136259e-03 -3.22795033e-01 -1.01811737e-01
3.82774711e-01 -9.47016835e-01 2.65235513e-01 7.92428076e-01
5.96298814e-01 -8.16447556e-01 1.07154667e+00 2.56695956e-01
-4.43767160e-01 -5.39995432e-01 1.76063627e-01 -3.78199130e-01
-1.11631211e-02 8.97087276e-01 -8.61082435e-01 5.58937848e-01
7.26103306e-01 5.11573732e-01 -2.16041535e-01 9.20336008e-01
1.58120930e-01 9.00580287e-01 -3.81643802e-01 -1.86743170e-01
-3.47033679e-01 -2.72937268e-01 1.91379547e-01 8.11652064e-01
4.32498693e-01 1.22085042e-01 -1.48910806e-01 6.84692264e-02
4.55709130e-01 4.27683055e-01 -3.87596458e-01 2.86576629e-01
3.18727791e-01 8.56622458e-01 -3.00396591e-01 -4.25326854e-01
-5.40404499e-01 2.38412425e-01 -4.48794484e-01 -7.51728117e-02
-8.73323381e-01 -6.97233081e-02 6.15391314e-01 5.11085629e-01
3.46806124e-02 1.23201855e-01 -1.03216803e+00 -7.57930338e-01
-2.80103445e-01 -1.05235457e+00 6.33268476e-01 -2.81121850e-01
-1.17830431e+00 1.73623875e-01 -2.64983177e-01 -1.29372680e+00
-2.50164390e-01 -4.02945220e-01 -9.41912383e-02 6.50157273e-01
-1.51540327e+00 -1.24533331e+00 -2.57888079e-01 5.59988976e-01
-7.06643462e-02 -3.90082538e-01 1.00205994e+00 5.59654593e-01
-6.61387682e-01 3.28569740e-01 3.63048494e-01 -1.22470565e-01
6.94103956e-01 -7.50878394e-01 -4.25140381e-01 2.27334037e-01
-6.31123185e-01 3.86930913e-01 7.78640926e-01 -1.01598990e+00
-1.66051495e+00 -1.17210078e+00 1.02106476e+00 -1.20759696e-01
4.16187823e-01 -2.87945904e-02 -2.24737942e-01 4.68944997e-01
2.66359836e-01 8.86305720e-02 9.61462259e-01 -1.30298480e-01
-1.09931827e-01 -8.23745906e-01 -1.87652206e+00 1.69409722e-01
4.59811538e-01 4.20385569e-01 -3.35319906e-01 2.04601139e-01
7.27894083e-02 7.52265975e-02 -1.29650199e+00 7.31985569e-01
1.37731862e+00 -1.00995588e+00 8.58850598e-01 -7.18132496e-01
3.77187133e-01 -3.46759021e-01 -3.71215433e-01 -7.22903967e-01
-1.86826110e-01 -5.13578594e-01 -2.43020311e-01 1.28884828e+00
4.04379189e-01 -1.02900720e+00 7.98775613e-01 6.95614576e-01
6.54328525e-01 -9.93667603e-01 -8.99251342e-01 -4.78960782e-01
-3.77766378e-02 -1.62515879e-01 5.92317402e-01 1.07078111e+00
5.25989607e-02 2.35947192e-01 -4.52834010e-01 1.13625132e-01
1.08805418e+00 2.61609077e-01 7.48346567e-01 -1.49492610e+00
5.39994240e-02 8.89445171e-02 -9.59718704e-01 2.24149153e-01
-7.79475927e-01 -6.92622423e-01 -8.16048920e-01 -1.59780657e+00
2.85665840e-01 -9.07931626e-01 -8.34885001e-01 6.48268998e-01
2.79066414e-01 2.30350584e-01 -3.84287387e-02 2.81626701e-01
1.38923049e-01 -3.22399825e-01 6.65656388e-01 -3.07093635e-02
-1.69439152e-01 2.17286021e-01 -6.92972720e-01 5.86535871e-01
1.16601419e+00 -7.65794992e-01 -1.67602301e-01 9.96490046e-02
3.34459126e-01 2.92197108e-01 4.14033473e-01 -1.03053355e+00
4.12496412e-03 -5.48080266e-01 7.19646454e-01 -5.51157117e-01
-3.17628503e-01 -1.40804648e+00 1.00748706e+00 1.31568813e+00
-6.74774945e-02 -4.80540693e-02 -2.76463419e-01 4.51341033e-01
3.16042066e-01 3.27954441e-01 7.25582242e-01 1.26658469e-01
-9.54637229e-02 -2.49041349e-01 -3.15711290e-01 -5.65747976e-01
1.60888553e+00 -1.94443360e-01 -6.09472930e-01 1.33942828e-01
-7.34615326e-01 1.23690836e-01 4.09164518e-01 2.03377903e-01
2.22301006e-01 -1.20965779e+00 -9.36717153e-01 4.06872511e-01
6.56182244e-02 -4.55949813e-01 -6.24410845e-02 1.23499739e+00
-7.57461071e-01 5.61128974e-01 -6.76929712e-01 -5.39908886e-01
-1.61640310e+00 7.66711354e-01 -1.24520123e-01 -2.33170062e-01
-3.38172466e-01 -5.00658415e-02 -6.84113503e-01 -4.37862352e-02
1.91611856e-01 -6.54733121e-01 -2.43374541e-01 1.13433294e-01
2.75833189e-01 1.01801455e+00 1.16710991e-01 -2.99520344e-01
-7.53933966e-01 7.75338888e-01 -3.09994891e-02 3.55915695e-01
1.48100293e+00 -4.75864783e-02 -2.32668787e-01 1.97344631e-01
7.89546072e-01 -6.44346178e-02 -6.01668715e-01 1.27004847e-01
3.39020900e-02 -2.62379915e-01 4.67373990e-02 -1.79398620e+00
-1.06272721e+00 4.40777838e-01 1.31252611e+00 3.13368052e-01
1.26626766e+00 -3.07240218e-01 8.36865127e-01 -1.03351451e-01
6.02030456e-01 -6.24000490e-01 -9.10527468e-01 -3.41457754e-01
5.04810333e-01 -1.19500804e+00 4.61842895e-01 -4.38009620e-01
-5.25609732e-01 9.78394330e-01 4.47097197e-02 -1.42156169e-01
8.19829643e-01 4.61984843e-01 4.12373900e-01 2.13859022e-01
-8.88527095e-01 2.84424603e-01 -2.97608197e-01 7.09292352e-01
2.05407739e-01 6.00052774e-01 -8.67400527e-01 6.97656095e-01
5.40190339e-02 1.01603365e+00 1.91067293e-01 1.13382435e+00
-2.45344743e-01 -1.32775748e+00 -4.61722761e-01 9.75618601e-01
-8.19411278e-01 1.31833673e-01 -2.11379528e-01 7.65332580e-01
7.03066170e-01 9.53440964e-01 -2.26909831e-01 -3.71468157e-01
1.88577533e-01 3.20675820e-01 4.20864046e-01 1.69165224e-01
-7.07205296e-01 -1.16569325e-01 3.02992374e-01 -3.87259096e-01
-6.35917544e-01 -8.41570675e-01 -1.19310117e+00 -4.36739564e-01
-7.87530560e-03 -2.32560098e-01 1.13808107e+00 7.30768323e-01
6.78263426e-01 3.10435414e-01 6.98583007e-01 1.16470449e-01
-4.29648638e-01 -8.38649809e-01 -3.83304119e-01 1.46444663e-01
3.81812751e-01 -5.52786291e-01 -1.57124996e-01 3.65619600e-01]
|
[8.407461166381836, 4.971554756164551]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.