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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
5930e1af-7116-49fd-b5cd-76a214a2ecc1 | identifiable-causal-inference-with-noisy | 2306.10614 | null | https://arxiv.org/abs/2306.10614v1 | https://arxiv.org/pdf/2306.10614v1.pdf | Identifiable causal inference with noisy treatment and no side information | In some causal inference scenarios, the treatment (i.e. cause) variable is measured inaccurately, for instance in epidemiology or econometrics. Failure to correct for the effect of this measurement error can lead to biased causal effect estimates. Previous research has not studied methods that address this issue from a causal viewpoint while allowing for complex nonlinear dependencies and without assuming access to side information. For such as scenario, this paper proposes a model that assumes a continuous treatment variable which is inaccurately measured. Building on existing results for measurement error models, we prove that our model's causal effect estimates are identifiable, even without knowledge of the measurement error variance or other side information. Our method relies on a deep latent variable model where Gaussian conditionals are parameterized by neural networks, and we develop an amortized importance-weighted variational objective for training the model. Empirical results demonstrate the method's good performance with unknown measurement error. More broadly, our work extends the range of applications where reliable causal inference can be conducted. | ['Pekka Marttinen', 'Antti Pöllänen'] | 2023-06-18 | null | null | null | null | ['causal-inference', 'epidemiology', 'econometrics', 'causal-inference'] | ['knowledge-base', 'medical', 'miscellaneous', 'miscellaneous'] | [ 5.46208978e-01 4.28083718e-01 -7.39158452e-01 -5.60808063e-01
-9.07618165e-01 -2.57201463e-01 6.56448901e-01 4.75747138e-03
-2.96909958e-01 1.07965040e+00 5.49425721e-01 -5.86446166e-01
-3.39576572e-01 -9.28559244e-01 -1.15353096e+00 -6.63338244e-01
5.15145846e-02 5.23755074e-01 -3.89410198e-01 4.64523017e-01
2.33925208e-01 -1.69246599e-01 -7.70256519e-01 -3.03766429e-01
1.02718329e+00 2.53627360e-01 -7.73773789e-02 4.89213496e-01
3.00996095e-01 7.69628704e-01 -3.13923299e-01 -4.32345927e-01
-1.17419176e-01 -3.29647928e-01 -6.61960781e-01 -1.96217939e-01
4.15684283e-01 -7.88741171e-01 -1.59546167e-01 1.11876953e+00
2.95938849e-01 -1.88517392e-01 1.13522685e+00 -1.30621612e+00
-1.01922083e+00 1.12119401e+00 -6.42732501e-01 -2.51892433e-02
6.43773526e-02 1.71458587e-01 1.11376607e+00 -5.25749028e-01
4.39585537e-01 1.63786054e+00 8.23559880e-01 3.71073067e-01
-1.66344380e+00 -8.03338051e-01 3.06227833e-01 -1.11416675e-01
-1.08863211e+00 -2.67702758e-01 4.63133246e-01 -7.56251812e-01
4.59808797e-01 -9.29299146e-02 1.75879702e-01 1.69427156e+00
6.14671111e-01 4.81958985e-01 9.98808920e-01 -1.85889855e-01
4.03985858e-01 -1.83714464e-01 3.38540673e-01 3.17500889e-01
6.16919041e-01 5.98498225e-01 -1.01866409e-01 -5.22643089e-01
9.37522113e-01 3.33268017e-01 -2.53254116e-01 -2.22791240e-01
-1.02719605e+00 1.08966708e+00 2.70225763e-01 -6.85216710e-02
-7.05594897e-01 8.21435213e-01 1.55141447e-02 3.31825353e-02
6.51250958e-01 3.57562542e-01 -4.73350674e-01 1.13787064e-02
-9.14221227e-01 3.20171475e-01 4.61232245e-01 7.73633599e-01
3.49827796e-01 -5.79547845e-02 -5.04829347e-01 3.89467597e-01
2.94082046e-01 9.73943412e-01 -5.00108078e-02 -1.09420955e+00
3.41733426e-01 1.40842944e-01 6.85235381e-01 -8.74603808e-01
-5.94079554e-01 -2.60266699e-02 -9.80437934e-01 -2.77896166e-01
7.78398275e-01 -5.01570225e-01 -1.11421502e+00 2.39106822e+00
4.35365081e-01 7.58116484e-01 -3.33455533e-01 8.74692380e-01
3.73826593e-01 2.75406897e-01 5.08658051e-01 -4.31371391e-01
1.33535504e+00 -2.87923366e-01 -1.09007406e+00 -1.69171430e-02
5.37311733e-01 -4.48449910e-01 8.44960451e-01 1.30706593e-01
-1.19055283e+00 3.85771645e-03 -4.29002136e-01 -1.47417113e-01
-3.34316939e-02 -2.22245604e-01 8.47433805e-01 6.34714127e-01
-7.00228870e-01 5.14112234e-01 -1.18748605e+00 -1.29179284e-01
4.62528110e-01 3.02464694e-01 2.81228428e-03 -1.40073732e-01
-1.47180593e+00 7.89527357e-01 -2.31382132e-01 1.21704869e-01
-1.38158667e+00 -1.50932753e+00 -6.64805770e-01 3.48178446e-01
6.06406510e-01 -1.22040141e+00 1.34309292e+00 -8.14586461e-01
-1.06768966e+00 2.77234137e-01 -3.10045838e-01 -3.95752490e-01
5.94849825e-01 -1.45218045e-01 -9.80921462e-02 -2.48922184e-01
2.99689233e-01 3.98776680e-02 9.24292207e-01 -1.05909204e+00
-4.88686502e-01 -4.92922336e-01 1.38622135e-01 -1.62816569e-01
-1.55041143e-02 -5.24549093e-03 1.87010884e-01 -5.07978022e-01
-9.71514061e-02 -8.06739271e-01 -2.65387684e-01 -1.49709255e-01
-7.42326021e-01 -1.91898361e-01 2.36126930e-01 -5.43204904e-01
1.27980590e+00 -1.74000168e+00 1.85387492e-01 1.11020416e-01
5.15394270e-01 -3.21244121e-01 7.27223977e-02 4.10534441e-01
-2.18966305e-01 4.95362461e-01 -5.96974909e-01 -4.38406795e-01
1.45536572e-01 2.08943486e-01 -3.90028328e-01 6.98176444e-01
9.57838967e-02 9.60493624e-01 -1.05806088e+00 -1.82491004e-01
-2.25185547e-02 7.25887895e-01 -8.11833978e-01 5.29079065e-02
-1.27430752e-01 6.78576648e-01 -5.31227469e-01 2.74613857e-01
6.96098566e-01 -6.32501841e-01 4.16443914e-01 6.97376728e-02
7.02448562e-02 4.49465394e-01 -1.15379548e+00 1.15952086e+00
-6.21186376e-01 2.24627733e-01 7.32565373e-02 -1.27751935e+00
1.88147519e-02 5.66823006e-01 4.68250304e-01 -1.69516400e-01
-1.74102355e-02 1.76944658e-02 2.79355142e-02 -6.34361207e-01
1.21347360e-01 -6.30357623e-01 -1.06464989e-01 3.77266824e-01
-1.88292980e-01 2.99478292e-01 -4.08840507e-01 2.38639683e-01
1.23440874e+00 -5.53834140e-02 1.86075911e-01 -1.82489768e-01
-2.36016512e-01 -1.91071123e-01 8.23919475e-01 1.32092071e+00
1.55685008e-01 4.53525424e-01 8.24549019e-01 -7.43897073e-03
-1.02487910e+00 -1.31078780e+00 -6.45006597e-01 6.51791096e-01
-2.14651406e-01 1.70450762e-01 -6.35601699e-01 -5.12079239e-01
3.59745055e-01 1.06943262e+00 -1.12527180e+00 -1.99476644e-01
-2.44549707e-01 -1.24230289e+00 3.99402738e-01 6.51750743e-01
-7.43270442e-02 -4.90310669e-01 -2.05686286e-01 2.36017853e-01
-2.29287416e-01 -5.13615131e-01 -3.30087423e-01 -3.63163024e-01
-7.96356976e-01 -1.32699859e+00 -7.08748102e-01 5.99016771e-02
6.82966650e-01 -1.94794778e-02 1.20534110e+00 -1.37014091e-01
1.90378223e-02 6.80545151e-01 1.08275935e-01 -5.78126669e-01
-3.51535559e-01 -3.53343517e-01 9.74533036e-02 1.20917466e-02
4.58084941e-01 -6.62954807e-01 -9.19250965e-01 -1.75654367e-01
-8.29238951e-01 -2.09974140e-01 3.94377828e-01 1.08901429e+00
3.34672183e-01 -1.19668528e-01 8.26059639e-01 -1.55647051e+00
4.13715780e-01 -1.10996950e+00 -8.90219331e-01 1.60155892e-01
-9.06684279e-01 9.11002010e-02 2.06639096e-01 -7.31250703e-01
-1.29704809e+00 -5.76869130e-01 2.93841094e-01 -2.04210356e-01
-6.92422092e-02 8.19437921e-01 -4.84648645e-02 6.01608038e-01
3.71393263e-01 -3.80932391e-01 -1.40727133e-01 -4.38293457e-01
2.41499081e-01 4.03776377e-01 1.87676698e-01 -6.70530260e-01
4.69378889e-01 7.00990200e-01 2.89904445e-01 -3.36632043e-01
-1.04972315e+00 -3.00356485e-02 -2.51139343e-01 2.72905022e-01
1.05782437e+00 -9.74560976e-01 -1.00765300e+00 1.35278687e-01
-1.10289574e+00 -6.06126070e-01 -4.27040122e-02 1.16608131e+00
-5.15593410e-01 -1.31686255e-01 -5.98756373e-01 -1.25510311e+00
2.32698008e-01 -1.13586462e+00 1.22521758e+00 -2.61993408e-01
-2.42149428e-01 -1.61150825e+00 1.12065010e-01 1.98961407e-01
3.50522190e-01 2.71475643e-01 1.04398584e+00 -2.41330937e-01
-6.86854303e-01 -1.82453454e-01 -2.48736128e-01 -1.65143177e-01
2.10953444e-01 1.63086310e-01 -8.06666017e-01 -4.46503647e-02
-1.09893098e-01 8.24191049e-03 9.84194517e-01 1.49918902e+00
1.22778463e+00 -6.46890163e-01 -6.77158415e-01 3.84304136e-01
1.26780593e+00 8.37423280e-03 4.01110500e-01 -3.50705862e-01
8.25075567e-01 4.95984674e-01 1.87191099e-01 5.53552628e-01
8.15463364e-01 6.39427423e-01 5.14521539e-01 -1.91840842e-01
2.89429307e-01 -5.85648417e-01 4.99763526e-02 1.04771681e-01
-1.02134449e-02 -3.69778663e-01 -7.20531523e-01 7.87105918e-01
-1.91607451e+00 -1.27662790e+00 -5.87542832e-01 2.53125191e+00
1.07313979e+00 -1.37934059e-01 1.03560276e-01 -4.53504682e-01
7.16010571e-01 -2.56290048e-01 -7.30734944e-01 -1.65508822e-01
1.64158031e-01 2.62454618e-02 9.82132435e-01 9.31996882e-01
-8.78206432e-01 4.60596144e-01 7.18557215e+00 3.18393618e-01
-8.53187382e-01 5.24016023e-01 6.32576108e-01 -1.93416864e-01
-8.35056722e-01 3.02267194e-01 -5.97365737e-01 7.00984061e-01
1.14887977e+00 -1.51574895e-01 2.10767746e-01 2.74001062e-01
8.08532536e-01 -6.29740134e-02 -1.37174726e+00 2.49191120e-01
-4.42441553e-01 -1.10139787e+00 -4.01785344e-01 5.25450885e-01
9.44494545e-01 -2.12615356e-01 3.36402982e-01 2.76136458e-01
1.20223534e+00 -1.18215704e+00 5.19503891e-01 9.36011791e-01
8.04283082e-01 -5.55615425e-01 7.35397995e-01 2.60928005e-01
-3.45860630e-01 -1.53517410e-01 -4.22648966e-01 -3.82328153e-01
3.67206156e-01 1.10193098e+00 -6.93836331e-01 1.82092100e-01
3.18253011e-01 6.27424240e-01 1.61753029e-01 7.51152277e-01
-5.70208311e-01 1.49960363e+00 -3.29396904e-01 2.85170823e-01
-1.81162301e-02 -7.73347244e-02 5.05709767e-01 8.94101381e-01
3.76413286e-01 2.00330228e-01 -2.70285849e-02 1.36267447e+00
-1.95972100e-01 -2.38120690e-01 -7.19381809e-01 -3.21361981e-02
6.59299135e-01 7.55620837e-01 -1.48771942e-01 -3.97359341e-01
-5.81430674e-01 4.44183052e-01 2.08357200e-01 7.61687338e-01
-1.00198090e+00 4.27896112e-01 5.64695954e-01 1.80562258e-01
1.04816288e-01 1.68931410e-01 -5.34414828e-01 -1.27731359e+00
-2.52251118e-01 -4.40354764e-01 4.85919803e-01 -6.21931612e-01
-1.56760311e+00 -5.73149323e-01 3.49685282e-01 -4.47632730e-01
-3.91753703e-01 -2.42445081e-01 -5.22343338e-01 1.25042200e+00
-1.27955902e+00 -1.18949866e+00 5.11837065e-01 2.50303626e-01
7.14416653e-02 4.95539367e-01 6.50240898e-01 2.17614830e-01
-8.20198298e-01 4.59816068e-01 2.07772270e-01 -1.27789825e-01
8.98629725e-01 -1.62830555e+00 5.34396507e-02 6.30904913e-01
-3.69624376e-01 9.80239093e-01 1.07578540e+00 -1.03208733e+00
-1.33157051e+00 -9.67234373e-01 9.61443782e-01 -9.47692394e-01
9.32365358e-01 -3.43896627e-01 -9.21430290e-01 1.22482204e+00
-4.63642254e-02 -1.88692987e-01 7.35190153e-01 8.91328335e-01
-5.60217440e-01 1.71054795e-01 -1.24327433e+00 5.20476401e-01
9.51754272e-01 -3.12458485e-01 -4.11014646e-01 5.36943793e-01
8.97250175e-01 -2.59117723e-01 -1.06068242e+00 4.05809373e-01
6.87221408e-01 -5.16730845e-01 9.55706775e-01 -1.19393635e+00
9.75637734e-01 7.84914494e-02 -3.14937532e-02 -1.42832208e+00
-4.17752534e-01 -2.61303157e-01 -1.57748207e-01 1.16806221e+00
5.94462991e-01 -6.95479453e-01 3.93869072e-01 1.26353216e+00
2.18478426e-01 -2.70928472e-01 -8.81248713e-01 -4.66596931e-01
6.80790663e-01 -6.51107013e-01 7.56523728e-01 1.31725955e+00
-2.29579985e-01 3.38688433e-01 -7.86335170e-01 8.33665252e-01
1.04654253e+00 1.18730828e-01 5.79413116e-01 -1.40347910e+00
-5.06798148e-01 -2.95227885e-01 1.18918136e-01 -7.04814315e-01
3.23596984e-01 -4.78183508e-01 5.17362431e-02 -1.40251601e+00
8.33907306e-01 -4.86473113e-01 -2.47110918e-01 3.31186265e-01
-8.71822536e-01 -1.97685689e-01 -2.18445376e-01 -1.29447028e-01
-9.64172184e-02 4.27502334e-01 1.17439854e+00 -1.13097519e-01
5.07614762e-02 2.36816108e-01 -9.35681105e-01 7.22442448e-01
5.32733500e-01 -8.89060736e-01 -5.92191458e-01 -4.80069518e-01
4.80967522e-01 5.95912755e-01 9.31593060e-01 -4.10070457e-02
-3.47357914e-02 -8.10024440e-01 2.25537896e-01 -7.80671015e-02
1.28515676e-01 -8.07289660e-01 3.14748853e-01 2.60177195e-01
-7.56601155e-01 -2.17290446e-01 -1.66215539e-01 9.30182517e-01
3.12842906e-01 -1.42607927e-01 3.57778370e-01 7.14673549e-02
8.46281797e-02 4.37229455e-01 -2.55402774e-01 1.79743022e-01
4.48593438e-01 4.67599779e-01 -4.41718221e-01 -7.34676838e-01
-8.27774823e-01 3.59848291e-01 2.01066107e-01 7.44452700e-02
2.80455649e-01 -1.23412740e+00 -9.75853026e-01 -2.65722901e-01
-1.70944557e-01 -3.01133811e-01 6.21469498e-01 1.22930956e+00
5.62671065e-01 3.03096741e-01 6.27294958e-01 -4.13692951e-01
-7.85664022e-01 9.63798463e-01 2.33235195e-01 -2.29707003e-01
-4.42311704e-01 5.17901123e-01 8.69856179e-01 -3.72018129e-01
-5.99508658e-02 -3.90505940e-01 1.40741855e-01 -1.25617191e-01
6.46796823e-01 5.31378329e-01 -5.46997368e-01 -2.24854663e-01
-1.72805622e-01 1.74061388e-01 8.92753378e-02 -2.75986016e-01
1.31840849e+00 -4.16978300e-01 -1.64900601e-01 8.32125843e-01
1.01193333e+00 6.87162280e-02 -1.37240875e+00 -2.18621284e-01
-1.60598040e-01 -4.74954069e-01 3.67919385e-01 -9.32197809e-01
-8.67106676e-01 9.84458685e-01 4.12761956e-01 3.25516343e-01
7.73318589e-01 -1.49070755e-01 1.20187268e-01 -2.24032015e-01
2.48378679e-01 -8.16755295e-01 -4.36868876e-01 6.28496855e-02
6.97086990e-01 -1.53311002e+00 -1.96283287e-03 -2.42807880e-01
-1.57350034e-01 4.86590773e-01 1.57303378e-01 -1.38818830e-01
9.41247463e-01 2.14775100e-01 -2.58259177e-01 -2.17412472e-01
-9.32378709e-01 5.34439832e-02 3.01256508e-01 6.88779950e-01
6.75737739e-01 5.69127619e-01 -6.39994562e-01 5.96471012e-01
2.18095779e-02 1.72409639e-01 7.81964302e-01 3.71221006e-01
-7.72760017e-03 -7.84388304e-01 -5.60769796e-01 7.77358115e-01
-9.91166055e-01 -4.14734185e-01 1.25732794e-01 7.07465589e-01
-2.11415738e-01 1.00836062e+00 3.04045528e-01 3.43414307e-01
1.31207451e-01 4.46247533e-02 2.67668694e-01 -6.56460464e-01
1.68881621e-02 1.48460373e-01 -1.36400312e-01 -6.62375867e-01
-6.83672011e-01 -8.96422446e-01 -9.22689140e-01 -6.24420345e-01
-3.13187033e-01 -2.13163942e-02 5.12841046e-01 1.13635206e+00
2.63038754e-01 8.22920561e-01 4.73108262e-01 -3.62490356e-01
-8.93026650e-01 -1.01238823e+00 -6.10979915e-01 3.72174621e-01
8.09410334e-01 -1.04093945e+00 -5.45633376e-01 -9.24037546e-02] | [8.04196548461914, 5.312739849090576] |
9772a007-01c6-4b1c-8c5b-fa7aaa66211c | piece-wise-matching-layer-in-representation | 2010.06510 | null | https://arxiv.org/abs/2010.06510v1 | https://arxiv.org/pdf/2010.06510v1.pdf | Piece-wise Matching Layer in Representation Learning for ECG Classification | This paper proposes piece-wise matching layer as a novel layer in representation learning methods for electrocardiogram (ECG) classification. Despite the remarkable performance of representation learning methods in the analysis of time series, there are still several challenges associated with these methods ranging from the complex structures of methods, the lack of generality of solutions, the need for expert knowledge, and large-scale training datasets. We introduce the piece-wise matching layer that works based on two levels to address some of the aforementioned challenges. At the first level, a set of morphological, statistical, and frequency features and comparative forms of them are computed based on each periodic part and its neighbors. At the second level, these features are modified by predefined transformation functions based on a receptive field scenario. Several scenarios of offline processing, incremental processing, fixed sliding receptive field, and event-based triggering receptive field can be implemented based on the choice of length and mechanism of indicating the receptive field. We propose dynamic time wrapping as a mechanism that indicates a receptive field based on event triggering tactics. To evaluate the performance of this method in time series analysis, we applied the proposed layer in two publicly available datasets of PhysioNet competitions in 2015 and 2017 where the input data is ECG signal. We compared the performance of our method against a variety of known tuned methods from expert knowledge, machine learning, deep learning methods, and the combination of them. The proposed approach improves the state of the art in two known completions 2015 and 2017 around 4% and 7% correspondingly while it does not rely on in advance knowledge of the classes or the possible places of arrhythmia. | ['Sixian Zhang', 'Fatemeh Afghah', 'Behzad Ghazanfari'] | 2020-09-26 | null | null | null | null | ['ecg-classification'] | ['medical'] | [ 3.70328546e-01 -3.13866884e-01 9.57377180e-02 -2.70822674e-01
-5.70552111e-01 -5.81510007e-01 4.25886065e-01 5.66857994e-01
-4.88033801e-01 6.01186216e-01 -4.02215235e-02 -3.59647214e-01
-7.93548465e-01 -6.30502820e-01 -2.88893998e-01 -6.55380070e-01
-5.93884766e-01 6.43417761e-02 1.61024585e-01 -2.54123658e-01
4.50146645e-01 7.41505623e-01 -1.34547698e+00 7.11592376e-01
6.70106888e-01 1.27622604e+00 -5.01343533e-02 6.01369262e-01
1.08627602e-01 4.66390133e-01 -7.34341383e-01 2.98855156e-01
2.89090186e-01 -5.03205478e-01 -4.73147541e-01 -2.35321015e-01
-8.46384391e-02 2.87998497e-01 -2.01405063e-02 4.27502036e-01
1.13921380e+00 6.01899102e-02 7.85625160e-01 -6.99224770e-01
-1.01276860e-01 6.67203248e-01 -4.43498164e-01 7.81637251e-01
3.82464558e-01 -5.18014021e-02 3.56218219e-01 -7.06137776e-01
5.93132079e-01 6.54001594e-01 1.08693206e+00 2.87615150e-01
-1.01781404e+00 -4.92047161e-01 -1.09520115e-01 4.91480917e-01
-1.41199315e+00 -8.21918547e-02 1.02843690e+00 -5.77433765e-01
9.21569467e-01 3.10121685e-01 6.52408540e-01 8.45388055e-01
3.02109659e-01 1.20859906e-01 1.20544815e+00 -6.37769520e-01
3.33405674e-01 1.41612351e-01 3.96454573e-01 2.84112900e-01
-6.57346249e-02 1.28422946e-01 -2.98034728e-01 -4.00567979e-01
6.31210864e-01 7.95088634e-02 -2.90927082e-01 -8.95404592e-02
-1.26364613e+00 4.04978573e-01 1.32587403e-01 6.68528795e-01
-6.43938780e-01 -2.14714304e-01 8.43027949e-01 4.28852379e-01
6.03188314e-02 4.95035172e-01 -7.49892533e-01 -5.58823794e-02
-1.13123238e+00 1.37400717e-01 5.71971297e-01 3.05034161e-01
4.57090169e-01 1.65247381e-01 -5.18799305e-01 6.39934301e-01
-2.47386307e-01 1.25196457e-01 9.72270668e-01 -3.16143304e-01
4.50289130e-01 6.70607209e-01 -1.17943592e-01 -1.00808513e+00
-9.35532331e-01 -6.80848718e-01 -1.04431093e+00 9.87807587e-02
2.39735439e-01 -3.27295929e-01 -9.82129574e-01 1.45991349e+00
1.28597975e-01 5.57494044e-01 1.25968717e-02 5.92619121e-01
8.32314610e-01 6.14358902e-01 2.47295611e-02 -5.95401704e-01
1.46404123e+00 -3.55164945e-01 -6.85419798e-01 3.54097575e-01
6.03486598e-01 -6.08513296e-01 5.22487581e-01 5.72351038e-01
-9.76931155e-01 -9.75500882e-01 -1.16499722e+00 5.43914735e-01
-5.04830539e-01 4.78151411e-01 5.20983636e-01 7.01396763e-01
-9.35133398e-01 1.10959721e+00 -7.84679949e-01 -4.55108225e-01
1.21384598e-01 4.90873456e-01 -1.97822899e-01 4.92339224e-01
-1.44119608e+00 9.04087722e-01 3.26817989e-01 2.83987582e-01
-5.97667277e-01 -8.51206303e-01 -5.82881510e-01 1.80743411e-01
-1.89345375e-01 -5.41837454e-01 5.03234863e-01 -8.05788100e-01
-1.29642057e+00 6.70190394e-01 1.66260004e-01 -8.67277801e-01
5.59009731e-01 4.94254418e-02 -7.55142272e-01 2.78056324e-01
-3.08416128e-01 8.61798674e-02 1.01573682e+00 -6.23198450e-01
-4.69077289e-01 -2.96476841e-01 -1.52680978e-01 -9.24345180e-02
-1.88616499e-01 -3.98354530e-02 -7.60695757e-03 -9.01823938e-01
7.37292245e-02 -7.74703681e-01 -2.98892379e-01 -4.24704075e-01
-4.01433557e-02 -2.42253587e-01 5.43390870e-01 -5.27679682e-01
1.59282410e+00 -2.41935039e+00 9.35928747e-02 5.40914357e-01
-3.56870592e-02 4.20431495e-01 4.22267802e-02 6.49158239e-01
-6.05129778e-01 7.04507530e-02 -1.30596384e-01 1.95467904e-01
-4.06777859e-01 -8.97117332e-02 -5.27447760e-01 5.12140632e-01
1.43427685e-01 5.26194572e-01 -6.12877250e-01 -4.57519829e-01
5.04446685e-01 6.31183088e-01 -2.78882116e-01 6.99706972e-02
4.15646344e-01 6.55147851e-01 -3.65382940e-01 3.31703812e-01
5.25027514e-01 -5.77759407e-02 3.87887180e-01 -6.47192895e-01
-2.44755939e-01 3.19586456e-01 -1.48529375e+00 1.68032014e+00
-3.19076627e-01 1.46745577e-01 -6.19888723e-01 -1.43592560e+00
1.15736926e+00 6.60582185e-01 8.62852275e-01 -4.28629696e-01
1.42334148e-01 2.57961273e-01 4.11890090e-01 -5.44852078e-01
-1.25588283e-01 9.26388726e-02 3.23237181e-02 4.99795079e-01
1.15045577e-01 3.46043646e-01 2.53804505e-01 -3.64212841e-01
1.24304700e+00 2.37072662e-01 5.65388799e-01 -4.32125896e-01
7.85353363e-01 -4.14743841e-01 6.13731086e-01 9.33898568e-01
-1.14321917e-01 8.00151825e-01 3.17416072e-01 -1.01580966e+00
-5.51909447e-01 -9.40518975e-01 -5.34278989e-01 7.71290660e-01
-2.08860278e-01 -5.60504556e-01 -4.51915771e-01 -4.02583361e-01
-2.79638380e-01 6.90711066e-02 -8.76464248e-01 -2.72690415e-01
-9.49789822e-01 -9.30551469e-01 7.97019005e-01 5.45234799e-01
2.43582800e-01 -1.34181213e+00 -1.18528616e+00 4.79425251e-01
-1.52078988e-02 -8.82058144e-01 3.00668217e-02 5.02024174e-01
-1.27262211e+00 -9.07200098e-01 -6.38602793e-01 -5.13253868e-01
4.21423942e-01 -2.45346144e-01 9.68855441e-01 8.96624699e-02
-5.65420866e-01 1.86015993e-01 -4.71313655e-01 -5.79571843e-01
-1.46636650e-01 2.40752086e-01 -3.24521847e-02 2.83416837e-01
2.97714677e-02 -1.00424123e+00 -9.06234145e-01 1.92411587e-01
-7.12045848e-01 -3.74019623e-01 5.51685333e-01 7.34227121e-01
5.65630376e-01 -2.47020479e-02 1.05510449e+00 -9.23271060e-01
7.35993922e-01 -5.65257072e-01 -2.57019460e-01 2.75555909e-01
-7.69388795e-01 1.97718218e-02 7.97584951e-01 -5.93434632e-01
-4.78532165e-01 2.45596051e-01 -5.27782273e-03 -3.84294242e-01
-9.14689079e-02 6.46867752e-01 2.80668706e-01 -2.84073390e-02
9.39145863e-01 3.30831200e-01 -1.89869195e-01 -3.40246141e-01
-5.56506589e-02 4.23096746e-01 4.18755352e-01 -6.42823040e-01
5.62058270e-01 4.59458202e-01 1.58958077e-01 -3.50858837e-01
-3.11631531e-01 -2.98499942e-01 -8.05957735e-01 -1.43874317e-01
7.28419065e-01 -6.04205847e-01 -5.72116315e-01 2.64094025e-01
-1.02668858e+00 2.28863075e-01 -3.58281434e-01 5.69674730e-01
-3.84436458e-01 5.32289743e-01 -3.42582732e-01 -7.92507529e-01
-8.47215235e-01 -7.24151254e-01 6.57434523e-01 1.11601194e-02
-3.52160573e-01 -8.97148788e-01 3.02338470e-02 -2.23328769e-01
5.26006758e-01 7.85491288e-01 1.11797118e+00 -8.49157572e-01
-8.07315931e-02 -3.69338244e-01 3.53509367e-01 3.22725952e-01
3.81917059e-01 -1.53837398e-01 -9.95271742e-01 -2.87526757e-01
2.87267268e-01 6.27546087e-02 8.48210752e-01 4.76780713e-01
1.33836615e+00 1.18143320e-01 -3.74206692e-01 6.75579131e-01
1.30133975e+00 5.38462520e-01 8.41036856e-01 2.53314614e-01
1.87674403e-01 2.51793504e-01 3.93311262e-01 8.70203078e-01
-1.36873005e-02 5.74664354e-01 1.15365885e-01 -2.17922747e-01
3.06095406e-02 2.82678276e-01 1.38911530e-01 7.92205691e-01
-7.25185156e-01 2.29485884e-01 -8.80635619e-01 4.31818545e-01
-1.75836158e+00 -1.20011389e+00 2.80115992e-01 2.53016162e+00
6.70868933e-01 3.91321123e-01 2.27301732e-01 9.27520454e-01
7.64504313e-01 8.13027173e-02 -5.18820286e-01 -6.15772724e-01
-1.16329759e-01 7.43498266e-01 1.69541523e-01 -8.64512920e-02
-1.08669496e+00 1.66718364e-01 6.00528479e+00 6.62285745e-01
-1.73210895e+00 -1.16839096e-01 3.05352569e-01 1.01359576e-01
2.02677488e-01 -2.36214608e-01 -5.58734417e-01 4.29534346e-01
1.01175451e+00 -5.38333505e-02 3.22023988e-01 4.33030605e-01
3.38572025e-01 3.77395004e-01 -1.24761677e+00 1.26920331e+00
1.20613195e-01 -1.54995489e+00 -1.40098510e-02 -3.32497448e-01
4.02488321e-01 -1.33452728e-01 -8.62632468e-02 3.26718956e-01
-8.23033392e-01 -1.11078930e+00 5.03078401e-01 7.26713121e-01
7.15943336e-01 -4.97709870e-01 7.04528749e-01 1.64567128e-01
-1.44781530e+00 -3.72595996e-01 -3.15744281e-02 -2.43310466e-01
-8.49187840e-03 4.08105701e-01 -6.90124214e-01 9.81512189e-01
7.31343746e-01 8.03070068e-01 -5.64250588e-01 1.09131217e+00
1.47176802e-01 8.67187500e-01 -3.53036374e-01 3.06196600e-01
-1.03174195e-01 2.18760341e-01 4.56435084e-01 1.38126707e+00
3.69349122e-01 3.05072274e-02 1.20785147e-01 4.96407956e-01
4.06245083e-01 2.98596442e-01 -5.14596641e-01 3.14432561e-01
5.36804378e-01 1.39021099e+00 -8.47298145e-01 -2.80407071e-01
-1.59523353e-01 3.84517521e-01 -2.54480578e-02 2.88459480e-01
-9.58651543e-01 -6.76570296e-01 -1.28127620e-01 4.31615353e-01
3.47997218e-01 6.16065226e-02 -4.96595830e-01 -9.48211789e-01
2.75113851e-01 -9.55910385e-01 8.25950205e-01 -3.27298760e-01
-1.09833586e+00 1.11118793e+00 4.41843271e-02 -1.79162717e+00
-4.69805837e-01 -4.19045925e-01 -8.55014265e-01 8.85115027e-01
-1.24890888e+00 -8.42539966e-01 -1.45615116e-01 8.15678537e-01
3.38825226e-01 -3.06336343e-01 1.11218011e+00 6.84934258e-01
-1.85188234e-01 6.50252700e-01 -3.22821259e-01 9.67542082e-02
6.75068676e-01 -9.57572341e-01 4.37538438e-02 6.41562939e-01
1.82086468e-01 7.41437197e-01 4.47750509e-01 -1.61256045e-01
-1.14124608e+00 -9.59755063e-01 8.12651336e-01 -2.34376326e-01
2.43695617e-01 -1.72771826e-01 -1.00499237e+00 2.29084164e-01
6.16093632e-03 4.39688087e-01 6.25872552e-01 1.78486466e-01
-1.97337642e-01 -6.27376258e-01 -1.00865865e+00 8.64913538e-02
6.84510529e-01 -3.58830303e-01 -8.85076106e-01 1.55017465e-01
-1.21859357e-01 -3.48102003e-01 -1.07340467e+00 1.01773489e+00
7.45996237e-01 -8.44048679e-01 1.11803162e+00 -6.15892053e-01
-2.72718817e-02 -4.57289279e-01 1.17290825e-01 -1.01590824e+00
-4.15515542e-01 -8.42605054e-01 3.27594765e-02 1.03160596e+00
4.92837489e-01 -6.34095967e-01 3.51940274e-01 1.19818682e-02
-2.89986193e-01 -1.20777726e+00 -9.00817692e-01 -5.97018480e-01
-1.03386030e-01 -3.74741673e-01 4.66650903e-01 9.75765526e-01
1.06772311e-01 2.37754285e-01 -4.40216154e-01 1.90180279e-02
2.18454137e-01 3.94620776e-01 2.93415189e-01 -1.40001190e+00
-4.27698553e-01 -2.21155226e-01 -7.74319589e-01 -3.47881258e-01
-2.77847856e-01 -9.87520099e-01 -5.21627963e-01 -1.39112222e+00
-1.12489372e-01 -6.57967031e-01 -1.17607439e+00 5.33700407e-01
9.82560813e-02 1.40036747e-01 1.09837562e-01 2.16392487e-01
-1.86137199e-01 -1.56760037e-01 8.56169820e-01 -1.81229394e-02
-5.64242125e-01 2.80991912e-01 -2.55512774e-01 6.98485076e-01
8.90186608e-01 -5.12393534e-01 -5.66331148e-01 -8.32836181e-02
2.36030549e-01 3.21925193e-01 1.76704958e-01 -1.47581053e+00
2.17440531e-01 2.91934013e-01 6.00958884e-01 -6.75681710e-01
2.05731258e-01 -5.88585138e-01 3.72468323e-01 7.09955454e-01
-4.14399147e-01 4.58513230e-01 2.86136627e-01 3.00816059e-01
-2.33085603e-01 -5.65557629e-02 7.68067122e-01 -1.13769956e-01
-4.96879548e-01 1.03882566e-01 -3.20423543e-01 5.09147346e-02
9.09451246e-01 -4.12337303e-01 7.44808540e-02 8.08996186e-02
-1.04885507e+00 -3.01354617e-01 -4.01688129e-01 5.22312820e-01
5.62552869e-01 -1.09753263e+00 -7.86989033e-01 3.27500165e-01
5.52094169e-02 -4.60690051e-01 4.11364108e-01 1.14108646e+00
-3.41384947e-01 2.92230129e-01 -7.23457694e-01 -7.40013242e-01
-1.06923985e+00 6.35014236e-01 6.02880120e-01 -5.30275702e-01
-8.63561034e-01 2.26637229e-01 -2.26422213e-02 1.30516842e-01
3.07815045e-01 -6.69877827e-01 -7.42587388e-01 1.56803787e-01
4.80314463e-01 4.10912961e-01 5.30162096e-01 -2.71827102e-01
-7.70984173e-01 9.80490685e-01 1.05001062e-01 1.95264220e-01
1.27742195e+00 2.16124684e-01 -1.33821089e-02 6.27422810e-01
7.46308863e-01 -6.84669688e-02 -5.70688248e-01 3.89466286e-02
5.40354289e-02 8.66997540e-02 -2.91095674e-01 -1.00580263e+00
-1.08099902e+00 1.07146108e+00 1.16058969e+00 1.70845121e-01
1.59786463e+00 -4.99597967e-01 5.32769084e-01 2.43735984e-01
3.15690160e-01 -9.26553667e-01 -2.62928456e-01 2.12592512e-01
7.88218737e-01 -6.65552855e-01 1.70185074e-01 -1.25996292e-01
-3.27479780e-01 1.45303094e+00 1.74035132e-01 -5.00863969e-01
1.13011038e+00 2.95099854e-01 2.01275989e-01 -8.93107876e-02
-7.76553214e-01 4.28342484e-02 4.52616960e-01 5.84905863e-01
5.96330941e-01 -1.10401966e-01 -9.52404082e-01 9.39254820e-01
1.50652096e-01 3.24311197e-01 2.36068979e-01 9.28363144e-01
-5.14845848e-02 -1.24504638e+00 -1.63279116e-01 4.23514575e-01
-8.00384343e-01 -6.88068867e-02 2.79672831e-01 6.37708187e-01
6.88519120e-01 7.19031990e-01 -2.98020720e-01 -4.92621064e-01
3.96569401e-01 8.96913633e-02 6.00795209e-01 -5.57733834e-01
-1.22208273e+00 7.25120530e-02 -1.55222401e-01 -2.88959622e-01
-4.56687987e-01 -6.66614175e-01 -1.34664416e+00 5.56859851e-01
-1.39558166e-01 1.84453338e-01 6.83659613e-01 9.92143393e-01
6.41468823e-01 6.32405579e-01 6.45555735e-01 -6.51434660e-01
-7.66792297e-01 -1.09749830e+00 -4.66257304e-01 5.36743104e-01
2.88022310e-01 -5.54114640e-01 -2.27483392e-01 3.30503851e-01] | [14.288241386413574, 3.2783215045928955] |
a0ef53c1-719c-4ee9-89b9-0098669f21c6 | aware-of-the-history-trajectory-forecasting | 2207.09646 | null | https://arxiv.org/abs/2207.09646v1 | https://arxiv.org/pdf/2207.09646v1.pdf | Aware of the History: Trajectory Forecasting with the Local Behavior Data | The historical trajectories previously passing through a location may help infer the future trajectory of an agent currently at this location. Despite great improvements in trajectory forecasting with the guidance of high-definition maps, only a few works have explored such local historical information. In this work, we re-introduce this information as a new type of input data for trajectory forecasting systems: the local behavior data, which we conceptualize as a collection of location-specific historical trajectories. Local behavior data helps the systems emphasize the prediction locality and better understand the impact of static map objects on moving agents. We propose a novel local-behavior-aware (LBA) prediction framework that improves forecasting accuracy by fusing information from observed trajectories, HD maps, and local behavior data. Also, where such historical data is insufficient or unavailable, we employ a local-behavior-free (LBF) prediction framework, which adopts a knowledge-distillation-based architecture to infer the impact of missing data. Extensive experiments demonstrate that upgrading existing methods with these two frameworks significantly improves their performances. Especially, the LBA framework boosts the SOTA methods' performance on the nuScenes dataset by at least 14% for the K=1 metrics. | ['Ulrich Neumann', 'Siheng Chen', 'Zhenyang Ni', 'Yiqi Zhong'] | 2022-07-20 | null | null | null | null | ['trajectory-forecasting'] | ['computer-vision'] | [-4.86954600e-01 -3.22613388e-01 -5.30009568e-01 -5.52634358e-01
-5.27423024e-01 -5.49399495e-01 1.02824616e+00 4.21863705e-01
-2.89064646e-01 7.96681583e-01 7.35945940e-01 -4.41073120e-01
-2.50456452e-01 -1.30520618e+00 -8.12767506e-01 -7.48675048e-01
-3.65152985e-01 4.12795335e-01 7.87906170e-01 -3.60758156e-01
1.41837642e-01 6.69715047e-01 -1.63292122e+00 3.21613133e-01
9.32952046e-01 7.16901004e-01 2.33477682e-01 5.25921047e-01
2.11898237e-02 9.92255926e-01 -4.02067095e-01 -6.71517700e-02
8.94570649e-02 5.35030775e-02 -5.62699676e-01 -5.05026460e-01
5.22755310e-02 -8.58450770e-01 -7.98060775e-01 4.86411303e-01
1.55076627e-02 4.11659956e-01 3.13347489e-01 -1.59441400e+00
-4.57049549e-01 7.25237668e-01 -8.30633193e-02 6.90658569e-01
3.29424679e-01 3.57015222e-01 4.74154174e-01 -5.59307337e-01
6.90101802e-01 9.80184436e-01 9.26481068e-01 8.12806934e-02
-7.55962253e-01 -3.34261984e-01 7.77336955e-01 7.58211911e-01
-1.52138972e+00 -3.23175192e-01 5.65892100e-01 -5.20194232e-01
1.15445542e+00 4.73190069e-01 5.37201583e-01 9.33226168e-01
8.91575068e-02 9.00243700e-01 4.99951839e-01 2.44778380e-01
2.22968742e-01 1.09652027e-01 2.56759167e-01 2.15702578e-01
-2.14798376e-01 3.73445988e-01 -3.47490430e-01 -2.20222384e-01
2.94661969e-01 5.74887991e-01 -1.54214337e-01 1.27847329e-01
-1.31044078e+00 3.61356318e-01 7.76859879e-01 1.71495855e-01
-6.58307433e-01 -2.48313360e-02 8.10613558e-02 -1.76290739e-02
7.60922432e-01 -7.94406310e-02 -3.99786621e-01 -4.47777718e-01
-9.80262041e-01 7.87495613e-01 6.78978503e-01 1.09370625e+00
1.08909643e+00 -1.33071631e-01 -3.96972030e-01 2.63208359e-01
-8.93583074e-02 7.91909158e-01 8.60028714e-02 -7.70611405e-01
7.30549216e-01 1.00806165e+00 6.85293496e-01 -1.21432078e+00
-7.79161513e-01 -1.73948914e-01 -6.06749654e-01 -2.74466246e-01
3.73447865e-01 -3.45768541e-01 -5.75306833e-01 1.75084221e+00
5.62104881e-01 8.79828155e-01 2.58494392e-02 1.09366071e+00
4.56446648e-01 1.15058970e+00 8.11945274e-02 -6.04268201e-02
7.33958304e-01 -1.15165377e+00 -6.68018878e-01 1.71573177e-01
8.85847569e-01 -2.11866215e-01 6.20840192e-01 -1.70953304e-01
-6.25511408e-01 -5.19038916e-01 -7.09083974e-01 1.88579574e-01
-1.00230479e+00 1.14818603e-01 5.86660504e-01 3.70278746e-01
-1.04535484e+00 5.01187086e-01 -1.11211193e+00 -4.47609663e-01
1.48096025e-01 5.76738715e-02 -1.11026213e-01 -3.47068533e-02
-1.25083268e+00 1.01506257e+00 3.97589475e-01 8.41982663e-02
-9.82992530e-01 -1.25676084e+00 -4.79039580e-01 4.87941615e-02
4.71421033e-01 -2.97744989e-01 8.60054910e-01 -2.18164667e-01
-1.13088799e+00 -1.51905671e-01 -6.11137211e-01 -6.34598613e-01
5.93691111e-01 -5.23336232e-03 -1.03412652e+00 -2.07344338e-01
1.15693398e-01 3.32191855e-01 2.45256990e-01 -1.06787419e+00
-1.35593426e+00 -3.95148784e-01 3.15759867e-01 1.46124542e-01
-2.56009758e-01 -2.30623901e-01 -4.95346159e-01 -4.68882114e-01
-2.11772367e-01 -1.10331821e+00 -2.60538369e-01 -3.99489403e-01
-1.46076187e-01 -4.86186415e-01 1.05664241e+00 -7.71485567e-01
1.77495933e+00 -1.96924305e+00 -1.43434256e-01 3.29844087e-01
8.72926638e-02 2.59522527e-01 -1.36670554e-02 1.00547612e+00
5.68290293e-01 -1.86888933e-01 -1.15636706e-01 -2.37597287e-01
1.13624923e-01 3.39643478e-01 -8.39456439e-01 3.30821544e-01
-1.66547447e-01 9.38628614e-01 -1.08953762e+00 -6.82857782e-02
5.70761859e-01 4.46810156e-01 -3.23518842e-01 1.77428573e-01
-2.71842539e-01 5.99922597e-01 -6.67032003e-01 4.52322811e-01
8.70480239e-01 -9.34723914e-02 1.06467716e-01 2.16107711e-01
-8.12333465e-01 2.83885837e-01 -8.76118481e-01 1.53396511e+00
-1.66384339e-01 4.71673906e-01 -4.35115963e-01 -5.23434043e-01
7.57920921e-01 1.20527700e-01 7.94639051e-01 -8.14877927e-01
-5.37969410e-01 3.50819295e-03 -4.32288170e-01 -4.27534997e-01
8.11024487e-01 5.76800466e-01 -1.13034725e-01 4.37355399e-01
-5.75708508e-01 9.17978048e-01 2.80369043e-01 2.25938708e-01
1.34663892e+00 2.17377678e-01 -1.58827826e-02 -1.12570338e-01
6.17328048e-01 6.30666494e-01 7.78428614e-01 7.82394767e-01
-3.10252428e-01 2.45981812e-01 1.40859812e-01 -1.02678299e+00
-1.01916206e+00 -9.55628812e-01 -1.29854679e-01 1.33981788e+00
5.80745578e-01 -7.34597683e-01 -5.37535608e-01 -6.79989278e-01
2.30614185e-01 9.61640716e-01 -6.47886455e-01 1.05398655e-01
-9.49589193e-01 -1.02357161e+00 5.43739021e-01 7.32779026e-01
5.60331881e-01 -6.29900515e-01 -5.41705966e-01 3.39951694e-01
-3.28123480e-01 -9.88450766e-01 -2.40633935e-01 -6.02057576e-01
-6.28675163e-01 -8.93717825e-01 -4.37505335e-01 -4.76742871e-02
4.87036437e-01 7.12888122e-01 5.70826232e-01 6.57932088e-02
5.40789723e-01 2.18599305e-01 -5.71654916e-01 -2.15689540e-01
-1.20875768e-01 3.50442946e-01 1.63280472e-01 7.55015314e-02
5.58822453e-01 -7.32433677e-01 -7.76990831e-01 6.25192523e-01
-4.80249405e-01 2.60599047e-01 2.38540515e-01 7.40243420e-02
3.44666123e-01 2.08893374e-01 7.95679688e-01 -6.59168720e-01
2.69458294e-01 -1.31738293e+00 -5.47838390e-01 3.60479385e-01
-6.74568355e-01 -1.12365626e-01 8.39338243e-01 -1.19621582e-01
-1.40106630e+00 -1.03294715e-01 -1.25707552e-01 -2.87106127e-01
-4.27711070e-01 7.18233705e-01 4.33116555e-02 2.44061783e-01
4.91428822e-01 5.13515174e-01 -3.47982973e-01 -6.75274074e-01
4.60339308e-01 5.45364857e-01 5.86066365e-01 -2.91799724e-01
6.00906432e-01 8.45109105e-01 -2.34146044e-01 -4.78116125e-01
-4.51964080e-01 -7.42711127e-01 -7.30522215e-01 -4.67335552e-01
4.58129942e-01 -9.94733274e-01 -1.01226723e+00 4.56843793e-01
-9.88113582e-01 -5.13857484e-01 9.62972194e-02 4.94441211e-01
-3.32433671e-01 2.08209045e-02 -3.63079995e-01 -7.36829579e-01
2.30669379e-01 -1.05852604e+00 8.23806226e-01 1.63215473e-01
3.73769365e-02 -1.14335454e+00 4.24961805e-01 -4.00689542e-02
6.99722052e-01 3.40555876e-01 7.63656735e-01 -5.83309591e-01
-1.06670940e+00 -2.18878731e-01 -2.60774136e-01 -5.78804135e-01
1.30460292e-01 -2.80079067e-01 -8.56516182e-01 -2.59488583e-01
-5.61542094e-01 4.46035296e-01 7.32338786e-01 2.06530601e-01
1.03361392e+00 -6.11781538e-01 -9.89136279e-01 5.44831455e-01
1.21284318e+00 3.92678887e-01 5.01448572e-01 3.84895056e-01
7.46544182e-01 6.32871509e-01 7.99556077e-01 6.53666914e-01
1.32753515e+00 1.08409786e+00 4.72410828e-01 2.62590289e-01
-1.57612696e-01 -5.82159460e-01 3.66661787e-01 6.25979364e-01
-4.22454029e-01 -4.04703438e-01 -1.20991898e+00 9.29286540e-01
-2.69486594e+00 -1.38727832e+00 -4.10760880e-01 1.98929143e+00
2.14228705e-01 -3.54018927e-01 4.94197637e-01 -2.56809711e-01
5.15088439e-01 4.46423620e-01 -8.80224526e-01 2.47851804e-01
3.11975423e-02 -9.18194950e-01 7.99354911e-01 5.99728942e-01
-1.18801928e+00 9.58312154e-01 6.23012924e+00 8.01658332e-01
-1.07109463e+00 2.85528779e-01 2.82010227e-01 -2.06210420e-01
-4.28017467e-01 -4.34985682e-02 -1.09792948e+00 9.82417166e-01
1.35546231e+00 -3.10118049e-01 5.16924918e-01 1.02069950e+00
5.92599630e-01 -1.98563054e-01 -9.91771698e-01 6.11783385e-01
-2.44938642e-01 -1.89350724e+00 -1.00345183e-02 1.81311071e-01
8.51195753e-01 4.69379723e-01 -4.50782776e-02 5.07607460e-01
3.48557055e-01 -6.77859008e-01 7.80039072e-01 1.03616154e+00
2.45608792e-01 -8.91823113e-01 6.96598947e-01 8.83716285e-01
-1.62017345e+00 -3.29996318e-01 -2.86839187e-01 -3.89756620e-01
5.11622429e-01 3.17485034e-01 -1.00005782e+00 9.48053420e-01
7.72621334e-01 1.24510515e+00 -5.48913181e-01 9.49103594e-01
1.12088174e-01 7.67706573e-01 -5.14499009e-01 1.95446908e-01
5.66938758e-01 -2.49511540e-01 7.64668047e-01 1.18500304e+00
4.11859095e-01 4.53431189e-01 4.79154587e-01 7.50122726e-01
3.71891648e-01 -2.84259260e-01 -7.62781143e-01 5.10112405e-01
7.41886497e-01 9.50192213e-01 -3.79010260e-01 -5.34036875e-01
-4.77376223e-01 4.94500339e-01 4.46420431e-01 5.92476308e-01
-1.06755090e+00 1.13204964e-01 1.00968182e+00 3.37882966e-01
2.49495924e-01 -4.26307440e-01 -6.46423250e-02 -1.09369493e+00
-3.14271450e-02 -1.18021838e-01 3.55315298e-01 -5.59753418e-01
-1.00941718e+00 5.62475741e-01 3.79970014e-01 -1.34285951e+00
-6.10141456e-01 -9.75465626e-02 -8.20923984e-01 7.83689678e-01
-1.74803174e+00 -1.42452538e+00 -4.30826455e-01 5.60436964e-01
4.00573760e-01 -1.74230635e-01 6.13842010e-01 5.12541175e-01
-6.34889066e-01 2.24698514e-01 5.14142990e-01 -1.08281583e-01
3.39133233e-01 -1.00471902e+00 8.16601753e-01 1.11100113e+00
-6.63418695e-02 4.84512299e-01 5.07075429e-01 -9.95680511e-01
-1.49412560e+00 -1.62618852e+00 1.04586148e+00 -8.58767152e-01
6.07568800e-01 -2.49687443e-03 -1.06731629e+00 1.01763105e+00
-2.21578628e-01 -9.85784307e-02 4.28374380e-01 2.91336119e-01
-4.39690948e-02 -3.43132973e-01 -8.73581290e-01 5.98878801e-01
1.26459920e+00 -3.94886911e-01 -2.62968302e-01 2.86324263e-01
6.94813609e-01 -2.20789984e-01 -1.02735138e+00 4.02315378e-01
5.26602030e-01 -1.01255631e+00 1.09881794e+00 -5.66586256e-01
-4.15666066e-02 -6.42584801e-01 -2.38233238e-01 -1.40923524e+00
-5.47912478e-01 -2.85826325e-01 -3.26475918e-01 1.07726669e+00
3.73644441e-01 -6.20490551e-01 7.67328978e-01 8.85482669e-01
-4.14933890e-01 -5.52402139e-01 -8.14657450e-01 -7.99494267e-01
-1.70904517e-01 -6.19624734e-01 1.68879783e+00 9.35202241e-01
3.06221843e-01 -3.48790795e-01 -6.90576851e-01 7.16704071e-01
2.76995182e-01 4.43910122e-01 1.08293450e+00 -8.66658151e-01
4.08036828e-01 -2.82947361e-01 -3.82941514e-01 -1.27852452e+00
1.30956993e-01 -8.33356440e-01 -1.89115763e-01 -1.55236399e+00
-4.15282964e-04 -8.12126756e-01 -3.00672352e-01 4.66245770e-01
-1.02234423e-01 -2.38233492e-01 1.93975821e-01 6.82442427e-01
-8.35364342e-01 5.85828006e-01 6.42806947e-01 1.05858352e-02
-3.78102481e-01 2.90049165e-01 1.59835326e-03 6.50567174e-01
7.00306058e-01 -4.22702342e-01 -6.57775342e-01 -5.79760373e-01
1.33498341e-01 3.15745294e-01 5.92863977e-01 -1.19963574e+00
8.37491691e-01 -5.67206562e-01 1.30980298e-01 -1.34687817e+00
4.00271326e-01 -9.32229042e-01 6.00463033e-01 1.25893921e-01
-1.68902293e-01 1.98445171e-01 1.01449020e-01 1.00068891e+00
-2.00980231e-01 4.00490284e-01 -1.12921707e-01 6.30804375e-02
-1.43581760e+00 6.30483627e-01 -5.20091832e-01 -3.18669826e-01
1.29600382e+00 -7.40277693e-02 -8.39609265e-01 -3.45217019e-01
-5.41047454e-01 7.43475497e-01 5.45556426e-01 5.66420317e-01
4.38586265e-01 -1.54506934e+00 -5.42100430e-01 2.02705130e-01
2.60180116e-01 -2.21895128e-01 7.17486203e-01 1.01666379e+00
-2.19794929e-01 8.40307772e-01 -3.11774701e-01 -5.18675208e-01
-7.82476366e-01 6.82108104e-01 5.40964343e-02 -1.81182146e-01
-9.00651157e-01 2.70497084e-01 7.79744014e-02 -5.96219182e-01
4.80769910e-02 -4.51297522e-01 -4.33206081e-01 -2.76235025e-02
1.01007974e+00 1.22488451e+00 2.57800091e-02 -8.61682713e-01
-4.63086277e-01 1.04315460e-01 4.42408510e-02 -3.54268998e-02
1.43683362e+00 -5.12958825e-01 1.12042017e-01 5.99912584e-01
8.43892872e-01 -1.78892657e-01 -1.76130283e+00 -3.74366760e-01
1.53524116e-01 -6.38721883e-01 1.69334877e-02 -8.92558932e-01
-7.49147832e-01 5.42573571e-01 1.49480745e-01 4.24659371e-01
9.06798959e-01 -1.42605036e-01 1.12419057e+00 4.35468018e-01
8.68856490e-01 -1.05106223e+00 -7.65815437e-01 5.30155182e-01
7.72627950e-01 -1.18986022e+00 -1.34633973e-01 -2.15799227e-01
-8.17436695e-01 7.54067481e-01 5.51640868e-01 2.43303832e-02
8.34454834e-01 -3.56249474e-02 -2.02010021e-01 -2.98165213e-02
-1.03519249e+00 -4.38494831e-01 3.05921018e-01 6.68627203e-01
-4.03046608e-01 5.47351003e-01 2.67156959e-01 6.41149759e-01
-1.80309400e-01 6.50246739e-02 2.54545927e-01 6.92344129e-01
-5.65733254e-01 -8.14020097e-01 -1.19599015e-01 3.14118564e-01
2.84922600e-01 3.02513152e-01 4.49568639e-03 6.78507745e-01
2.54042327e-01 1.02707696e+00 3.54542643e-01 -6.47876382e-01
2.96816111e-01 -1.41426310e-01 -1.89457640e-01 -1.50900409e-02
-5.54921985e-01 -4.87815410e-01 2.04837903e-01 -9.72693324e-01
-5.08011997e-01 -8.54405880e-01 -1.23165405e+00 -1.18391228e+00
3.16524267e-01 9.74221230e-02 4.86268789e-01 1.08651221e+00
8.68873775e-01 3.20364773e-01 6.39459074e-01 -9.13490713e-01
-7.98712671e-02 -8.21376085e-01 -2.53950089e-01 2.86582351e-01
5.10270834e-01 -7.65960395e-01 1.04379065e-01 -4.90809307e-02] | [5.99837589263916, 0.9942309260368347] |
e3269aa8-cfd3-4f2d-a193-5e2d7304ac6f | a-fast-and-accurate-system-for-face-detection | 1809.07586 | null | http://arxiv.org/abs/1809.07586v1 | http://arxiv.org/pdf/1809.07586v1.pdf | A Fast and Accurate System for Face Detection, Identification, and Verification | The availability of large annotated datasets and affordable computation power
have led to impressive improvements in the performance of CNNs on various
object detection and recognition benchmarks. These, along with a better
understanding of deep learning methods, have also led to improved capabilities
of machine understanding of faces. CNNs are able to detect faces, locate facial
landmarks, estimate pose, and recognize faces in unconstrained images and
videos. In this paper, we describe the details of a deep learning pipeline for
unconstrained face identification and verification which achieves
state-of-the-art performance on several benchmark datasets. We propose a novel
face detector, Deep Pyramid Single Shot Face Detector (DPSSD), which is fast
and capable of detecting faces with large scale variations (especially tiny
faces). We give design details of the various modules involved in automatic
face recognition: face detection, landmark localization and alignment, and face
identification/verification. We provide evaluation results of the proposed face
detector on challenging unconstrained face detection datasets. Then, we present
experimental results for IARPA Janus Benchmarks A, B and C (IJB-A, IJB-B,
IJB-C), and the Janus Challenge Set 5 (CS5). | ['Jun-Cheng Chen', 'Rajeev Ranjan', 'Joshua Gleason', 'Jingxiao Zheng', 'Ankan Bansal', 'Anirudh Nanduri', 'Rama Chellappa', 'Hongyu Xu', 'Carlos D. Castillo', 'Boyu Lu'] | 2018-09-20 | null | null | null | null | ['robust-face-recognition'] | ['computer-vision'] | [-2.07179338e-01 -4.31714833e-01 3.41781527e-02 -6.75339699e-01
-6.43258274e-01 -5.64302862e-01 5.75397789e-01 -6.95128202e-01
-1.99751899e-01 1.93110451e-01 -2.16681838e-01 1.93182215e-01
2.72694111e-01 -4.09936011e-01 -7.15730548e-01 -6.30169392e-01
-3.59904766e-01 3.93830329e-01 -8.83292258e-02 1.81270346e-01
7.55010545e-02 1.22892261e+00 -1.76271737e+00 4.18296665e-01
-2.01974064e-01 1.36066115e+00 -1.96295649e-01 5.99637985e-01
2.59823471e-01 4.24977154e-01 -4.86276358e-01 -7.98636138e-01
4.75142062e-01 -5.07347621e-02 -6.96994901e-01 4.51993346e-02
1.27455854e+00 -7.04348922e-01 -3.59848589e-01 9.81805027e-01
8.76327455e-01 -2.40470126e-01 5.82793355e-01 -1.51597953e+00
-6.72699749e-01 1.47080585e-01 -9.83939528e-01 3.00732315e-01
5.58254242e-01 1.38988286e-01 4.78296489e-01 -1.67975247e+00
6.80595815e-01 1.97717822e+00 1.08958876e+00 1.03845334e+00
-9.91908610e-01 -1.10515606e+00 -2.07189530e-01 1.87936127e-01
-1.95641744e+00 -1.35378098e+00 3.28237832e-01 -4.82483745e-01
1.05567169e+00 -1.09802470e-01 2.88254857e-01 1.09621060e+00
-1.06419161e-01 6.37750387e-01 6.42912924e-01 -3.69534582e-01
-1.12698264e-02 -2.08361462e-01 -1.10606715e-01 1.29735959e+00
2.45761156e-01 2.48643860e-01 -6.98429108e-01 2.35712770e-02
7.09881186e-01 1.03336230e-01 -5.37532941e-02 -5.60644306e-02
-5.89501500e-01 6.03609681e-01 3.66234690e-01 -7.05988184e-02
-1.48065284e-01 1.65866837e-01 5.33249855e-01 1.84667856e-01
2.45943978e-01 -2.92247504e-01 -4.56163883e-01 3.99024278e-01
-1.01348341e+00 8.17350298e-02 6.90593064e-01 1.05483162e+00
5.34477353e-01 2.51243442e-01 -3.53037864e-01 1.06599736e+00
6.32919371e-01 8.75162482e-01 1.62047416e-01 -7.19038844e-01
7.05276430e-02 5.65717697e-01 2.94717290e-02 -1.14411104e+00
-4.98277992e-01 2.02172250e-01 -6.34483993e-01 3.39202672e-01
3.61935496e-01 -3.79896760e-02 -1.12417209e+00 1.55820680e+00
5.72591186e-01 3.55879545e-01 -1.29090339e-01 7.62143970e-01
1.48540103e+00 2.70949930e-01 7.28104822e-03 1.03879251e-01
1.74983728e+00 -6.15750074e-01 -4.39629853e-01 -1.71379715e-01
1.55462310e-01 -1.05991971e+00 4.97093529e-01 -3.42776887e-02
-7.61812687e-01 -9.83916998e-01 -8.97128046e-01 -1.99798867e-01
-4.46192235e-01 7.78389812e-01 4.08363432e-01 1.17369008e+00
-1.47933090e+00 2.63498396e-01 -7.85599411e-01 -7.40364015e-01
1.23022187e+00 9.49209511e-01 -8.65972996e-01 -3.11595827e-01
-5.19441545e-01 6.14211440e-01 9.82988253e-02 2.70338565e-01
-1.43800652e+00 -6.34047508e-01 -1.06668556e+00 7.80026987e-02
1.35224596e-01 -1.11755341e-01 1.14797103e+00 -9.20785069e-01
-1.21881354e+00 1.44603837e+00 -4.17206794e-01 -3.39067638e-01
6.39535859e-02 -9.34775621e-02 -6.08688176e-01 1.28139421e-01
1.38153687e-01 1.23637772e+00 1.10506022e+00 -8.83788109e-01
-4.75505084e-01 -8.89376581e-01 -3.32682580e-01 -3.09094906e-01
-2.93747514e-01 9.22982812e-01 -6.83931470e-01 -2.03184724e-01
-2.51478255e-01 -7.77662575e-01 4.17169780e-01 5.94188631e-01
-3.47666562e-01 -5.20080805e-01 1.40193677e+00 -6.88393295e-01
4.22962666e-01 -2.34200144e+00 -4.02977347e-01 -2.29619965e-02
-4.74194996e-02 6.63998306e-01 -6.12464130e-01 -2.95589529e-02
-2.76879519e-01 -1.55669749e-01 1.28285199e-01 -8.98637474e-01
9.77116916e-03 1.62330065e-02 -2.79463798e-01 8.88619602e-01
4.91239429e-01 9.10649836e-01 -3.58396351e-01 -4.51546043e-01
1.46786466e-01 8.49479079e-01 -4.82444406e-01 2.24129692e-01
1.55513927e-01 2.30188489e-01 2.44986229e-02 1.58867466e+00
1.18143547e+00 8.15570578e-02 -7.56663159e-02 -4.37448412e-01
-4.04008664e-02 -3.37944835e-01 -1.28704846e+00 1.11828983e+00
-2.24694550e-01 8.38945448e-01 6.01930618e-01 -5.90311527e-01
8.93862545e-01 4.82149988e-01 2.14050144e-01 -5.01353085e-01
2.52176255e-01 1.26986325e-01 -7.84140602e-02 -3.47928226e-01
-1.54106691e-01 3.39943647e-01 5.54895282e-01 3.26970845e-01
6.51419461e-01 2.71299303e-01 3.65324169e-01 -1.70466766e-01
7.19539464e-01 -1.58244669e-02 4.24900800e-01 -4.40764755e-01
9.49468970e-01 -7.90103972e-01 6.21110320e-01 3.82458687e-01
-6.22682929e-01 7.16011286e-01 3.81947875e-01 -1.06578302e+00
-6.62066936e-01 -7.92449713e-01 -4.26245540e-01 1.32422113e+00
-3.04800749e-01 -3.74705881e-01 -8.96142662e-01 -7.12025344e-01
4.74566698e-01 -2.62612820e-01 -8.96930993e-01 1.50605172e-01
-5.49918652e-01 -7.49852538e-01 9.46624517e-01 7.38120079e-01
7.39718735e-01 -1.22608995e+00 -3.80002022e-01 -3.24443460e-01
4.56930637e-01 -1.39333928e+00 -6.18234515e-01 -3.67829382e-01
-2.76872635e-01 -1.59927630e+00 -6.06906176e-01 -1.28385198e+00
8.60590279e-01 1.23438336e-01 1.05471957e+00 3.34406763e-01
-1.14642894e+00 6.35990024e-01 1.57895058e-01 -6.77805781e-01
1.05195036e-02 -4.03782099e-01 5.68287730e-01 3.07391226e-01
7.40549922e-01 1.03608742e-02 -6.80818558e-01 5.52308142e-01
-3.66656423e-01 -8.16545427e-01 3.33032072e-01 6.56612158e-01
3.75295788e-01 -5.03206372e-01 5.20348907e-01 -4.38163191e-01
7.49733523e-02 -1.54792875e-01 -1.15345728e+00 4.51261669e-01
-1.17589846e-01 -3.81265551e-01 3.51010501e-01 -9.22978520e-02
-9.56969857e-01 7.25793242e-01 -5.65420330e-01 -4.80668247e-01
-2.97600895e-01 -3.94477814e-01 -3.72665524e-01 -9.39484596e-01
6.10297680e-01 1.03740320e-01 1.15269929e-01 -5.76823533e-01
1.14493668e-01 7.63104022e-01 6.15553498e-01 -3.33935738e-01
6.50176764e-01 7.11443365e-01 7.34345019e-02 -1.14272606e+00
-7.03997076e-01 -4.27009523e-01 -1.00606191e+00 -3.07800382e-01
7.19487548e-01 -1.09138179e+00 -1.25319564e+00 9.02423382e-01
-1.41805351e+00 1.14918342e-02 2.33962342e-01 9.18640196e-03
-1.10052183e-01 1.28057688e-01 -6.23379290e-01 -8.60591531e-01
-6.72123492e-01 -1.44013464e+00 1.76635385e+00 4.28188413e-01
2.32532144e-01 -6.73137009e-01 -1.53087080e-01 9.94052589e-02
3.84937942e-01 1.86087549e-01 1.68733910e-01 -5.95583200e-01
-4.35747981e-01 -3.81398469e-01 -5.43979228e-01 3.73940706e-01
6.33454025e-02 5.13916492e-01 -1.48532939e+00 -7.18200326e-01
-3.12278152e-01 -6.48203313e-01 8.54910493e-01 4.93068874e-01
1.26800489e+00 -2.67622828e-01 -5.89266956e-01 1.10791540e+00
1.21194422e+00 1.91674292e-01 5.29525697e-01 -2.62732387e-01
6.11694574e-01 5.25148273e-01 1.28845587e-01 5.00085711e-01
-2.83556022e-02 8.07588696e-01 3.98414493e-01 -4.43441980e-02
-4.30229664e-01 6.78311139e-02 4.24641401e-01 -3.81740332e-02
-6.53361678e-02 1.58684820e-01 -8.87425542e-01 4.27711725e-01
-1.13563526e+00 -9.34746861e-01 2.45603025e-01 1.75816500e+00
4.48405564e-01 -5.70189238e-01 2.62036830e-01 -1.45411894e-01
9.86057758e-01 3.76610868e-02 -4.35450763e-01 -1.71714127e-01
-6.35543317e-02 5.34909487e-01 3.31628561e-01 2.69312680e-01
-1.56788778e+00 1.21209919e+00 6.87386036e+00 5.56589901e-01
-1.23208296e+00 2.24775940e-01 1.01010108e+00 -9.35015678e-02
9.64885652e-01 -6.67651772e-01 -1.61379731e+00 3.39073151e-01
6.84787095e-01 2.56986916e-01 5.12357414e-01 1.29938293e+00
-3.19513679e-01 1.33606017e-01 -1.60724866e+00 1.54648376e+00
5.18011153e-01 -1.38472641e+00 -1.08775325e-01 -1.46510333e-01
7.57550836e-01 1.10607050e-01 3.63703489e-01 3.35325211e-01
-2.75311340e-02 -1.47058713e+00 4.48718339e-01 1.12832665e-01
1.21922946e+00 -8.03975821e-01 8.23025405e-01 -2.69496381e-01
-1.56460512e+00 -3.29202294e-01 -7.95793712e-01 2.56582290e-01
-3.68329585e-01 7.72165731e-02 -8.98405194e-01 -8.84785950e-02
1.07650876e+00 7.59951651e-01 -7.77124166e-01 8.55020106e-01
-7.75821880e-03 1.56606808e-01 -3.18287462e-01 2.75004774e-01
-1.79464534e-01 2.45784178e-01 9.91980806e-02 1.30502868e+00
2.30941072e-01 1.75144851e-01 1.17471479e-01 7.70875037e-01
-8.01592767e-01 -4.62914556e-02 -5.95225215e-01 5.47202006e-02
6.62340462e-01 1.71668136e+00 -8.01189125e-01 -6.59230500e-02
-5.60466588e-01 7.25993276e-01 3.26208174e-01 5.11823408e-02
-7.01006234e-01 -8.32179859e-02 1.17240715e+00 -4.11507785e-02
4.89091098e-01 1.97394826e-02 2.91858852e-01 -6.19094551e-01
5.78586012e-03 -9.76486802e-01 4.62794185e-01 -2.23419666e-01
-1.25076866e+00 7.32883811e-01 -1.77633643e-01 -5.45949519e-01
9.65099260e-02 -1.26702690e+00 -7.39809990e-01 6.90827906e-01
-1.36483073e+00 -1.37352979e+00 -5.42269528e-01 8.66836488e-01
6.36684716e-01 -8.74308109e-01 9.10528660e-01 7.03915298e-01
-9.95250404e-01 1.06521034e+00 -7.38009065e-02 8.94545615e-01
8.19098592e-01 -6.65984154e-01 8.01001430e-01 7.93876290e-01
3.69641751e-01 6.90965712e-01 3.06943953e-02 -3.35875809e-01
-1.79587364e+00 -1.33209217e+00 4.49238688e-01 -5.54653525e-01
8.73490423e-02 -7.38698602e-01 -4.80496168e-01 9.03249145e-01
-4.09297347e-02 8.18825603e-01 4.33775961e-01 -8.21082518e-02
-6.35355473e-01 -5.84644556e-01 -1.60557473e+00 2.17389375e-01
1.18377149e+00 -6.75501347e-01 -4.54810932e-02 7.53181338e-01
3.44836414e-02 -3.22405487e-01 -5.73947191e-01 5.79092443e-01
8.23342741e-01 -1.00394964e+00 1.37725234e+00 -5.02620578e-01
-9.69277695e-02 -1.27833948e-01 -8.63894373e-02 -6.71987712e-01
-1.92472026e-01 -4.76434022e-01 -1.27962515e-01 1.30940330e+00
-7.63430726e-03 -4.90779161e-01 7.55464256e-01 3.99961889e-01
2.05007717e-01 -6.14233792e-01 -1.27208686e+00 -7.60113239e-01
-4.72457170e-01 -6.50201142e-02 7.49005497e-01 6.49959624e-01
-5.75775921e-01 -6.19664341e-02 -2.20530704e-01 4.01427925e-01
7.59216964e-01 3.19290869e-02 8.02116454e-01 -1.26408672e+00
3.57634872e-01 -3.92923772e-01 -9.50584769e-01 -6.07117176e-01
5.64773142e-01 -7.79919028e-01 -9.87172648e-02 -9.88057613e-01
3.38790208e-01 6.81812689e-02 1.15235291e-01 8.41442585e-01
1.88109785e-01 1.03413677e+00 3.33483443e-02 2.42810510e-02
-6.41988695e-01 3.15582663e-01 6.70156538e-01 -1.82584703e-01
3.79105687e-01 -1.09508105e-01 -2.51640528e-01 8.74788523e-01
3.83085072e-01 -2.44491607e-01 1.46470219e-01 -4.81807351e-01
-5.45221508e-01 -5.02395749e-01 5.82016349e-01 -1.25786686e+00
4.00077015e-01 4.28235441e-01 1.23869765e+00 -5.16974986e-01
6.29481018e-01 -6.52325869e-01 -1.90346614e-01 6.85950100e-01
7.79064000e-02 1.78456038e-01 6.09731734e-01 2.27705792e-01
-1.61656275e-01 -4.83824126e-02 1.48936164e+00 -9.30208527e-03
-1.07084548e+00 7.83260167e-01 1.97544202e-01 -2.42907569e-01
1.26681185e+00 -1.44936234e-01 -2.24863172e-01 1.04270466e-01
-5.74010849e-01 1.32488430e-01 2.26778507e-01 6.96713150e-01
8.94224405e-01 -1.37518322e+00 -9.92779493e-01 8.62130344e-01
3.72554958e-02 -4.18060809e-01 1.23959087e-01 5.94309151e-01
-8.45350742e-01 6.60051644e-01 -5.37529528e-01 -7.94440389e-01
-2.04841542e+00 5.01559258e-01 6.95610940e-01 5.60330451e-01
-2.59925872e-01 1.60123372e+00 3.47310483e-01 -4.58052725e-01
6.83423102e-01 1.01081230e-01 -2.08362505e-01 1.19319744e-01
1.17008901e+00 3.31314981e-01 2.41455495e-01 -1.29141462e+00
-1.05490088e+00 7.74870753e-01 -2.78046221e-01 6.00921392e-01
1.26678932e+00 3.04925382e-01 -4.32260245e-01 -5.19368827e-01
1.41803074e+00 -2.92060047e-01 -1.19845998e+00 -4.71118577e-02
1.50650352e-01 -7.27653742e-01 -1.66893542e-01 -6.35466933e-01
-1.64405155e+00 9.57812190e-01 1.24629748e+00 -4.09013391e-01
9.92589414e-01 1.26557395e-01 5.12204766e-01 5.06808877e-01
4.42617655e-01 -8.03428054e-01 1.90890566e-01 5.56356192e-01
1.22928607e+00 -1.54080939e+00 -6.08717427e-02 -4.62583721e-01
-2.03838959e-01 1.39474881e+00 9.86208498e-01 9.19377506e-02
8.31945181e-01 6.12181783e-01 -2.47016139e-02 -3.37092370e-01
-3.89586568e-01 -1.81180120e-01 4.35285717e-01 6.78790987e-01
5.45062304e-01 -1.46045864e-01 4.72878903e-01 2.35205486e-01
1.90745637e-01 -2.06396624e-01 -2.63113558e-01 6.33280218e-01
-4.19797897e-01 -8.57204437e-01 -8.15774739e-01 2.09072128e-01
-6.32456481e-01 -3.74247925e-03 -7.90629923e-01 9.08156514e-01
3.15374404e-01 8.30511451e-01 2.81333208e-01 -3.38684134e-02
1.42622352e-01 1.82806969e-01 6.93644941e-01 -6.51389956e-01
-2.94531971e-01 -1.99110836e-01 -2.84172207e-01 -8.64792585e-01
-2.78080672e-01 -6.77597225e-01 -9.17748094e-01 -4.37773854e-01
-1.59761071e-01 -3.57566357e-01 8.31924736e-01 7.44902492e-01
5.51298082e-01 -7.10602626e-02 5.53946733e-01 -1.23575985e+00
-5.34203947e-01 -1.04000092e+00 -5.78259110e-01 1.79761484e-01
3.24558496e-01 -8.92881036e-01 -5.70455566e-02 1.25369832e-01] | [13.335994720458984, 0.7265997529029846] |
4fe189b6-126c-4e6e-a30e-6b06c7a6f24a | human-body-model-fitting-by-learned-gradient | 2008.08474 | null | https://arxiv.org/abs/2008.08474v1 | https://arxiv.org/pdf/2008.08474v1.pdf | Human Body Model Fitting by Learned Gradient Descent | We propose a novel algorithm for the fitting of 3D human shape to images. Combining the accuracy and refinement capabilities of iterative gradient-based optimization techniques with the robustness of deep neural networks, we propose a gradient descent algorithm that leverages a neural network to predict the parameter update rule for each iteration. This per-parameter and state-aware update guides the optimizer towards a good solution in very few steps, converging in typically few steps. During training our approach only requires MoCap data of human poses, parametrized via SMPL. From this data the network learns a subspace of valid poses and shapes in which optimization is performed much more efficiently. The approach does not require any hard to acquire image-to-3D correspondences. At test time we only optimize the 2D joint re-projection error without the need for any further priors or regularization terms. We show empirically that this algorithm is fast (avg. 120ms convergence), robust to initialization and dataset, and achieves state-of-the-art results on public evaluation datasets including the challenging 3DPW in-the-wild benchmark (improvement over SMPLify 45%) and also approaches using image-to-3D correspondences | ['Xu Chen', 'Otmar Hilliges', 'Jie Song'] | 2020-08-19 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/3610_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123650732.pdf | eccv-2020-8 | ['image-to-3d'] | ['computer-vision'] | [-2.57768720e-01 1.11263327e-01 3.76669765e-02 -4.81608659e-01
-7.40831137e-01 -5.21052718e-01 5.50499141e-01 -1.71751246e-01
-9.51684415e-01 3.65961075e-01 2.69755185e-01 1.49129197e-01
9.55729559e-02 -1.89268529e-01 -1.01909089e+00 -3.31419468e-01
7.75252730e-02 1.25529432e+00 1.46661803e-01 -2.03976005e-01
-2.71902420e-02 8.38201523e-01 -1.12223172e+00 -1.92730248e-01
1.44566059e-01 1.00954378e+00 -1.15544081e-01 8.93573582e-01
2.09616020e-01 2.88271785e-01 -2.36106455e-01 -4.50634837e-01
8.98126960e-01 6.16551302e-02 -8.04045379e-01 4.51851070e-01
9.28705215e-01 -5.54102957e-01 -4.75301296e-01 7.52615988e-01
9.90312397e-01 3.19123864e-01 4.05838341e-01 -9.54991758e-01
-2.97805578e-01 -6.75453171e-02 -6.65245771e-01 -8.36108401e-02
3.84362221e-01 6.09000444e-01 8.65417302e-01 -1.28984630e+00
1.06139982e+00 1.29425502e+00 1.04647183e+00 5.58603168e-01
-1.43206429e+00 -2.13999927e-01 2.15197704e-03 -1.73191071e-01
-1.37514246e+00 -6.52847171e-01 5.90846479e-01 -4.65311974e-01
1.30304110e+00 -9.12808552e-02 9.99495447e-01 1.03196752e+00
-6.98696449e-02 8.83125484e-01 7.16575265e-01 -3.18250686e-01
-6.91982219e-03 -1.51142344e-01 -8.88783559e-02 1.13198781e+00
1.62907355e-02 2.15351060e-01 -5.74238956e-01 -6.77243024e-02
1.01531434e+00 -1.53099984e-01 -6.74383491e-02 -1.03937006e+00
-1.20573533e+00 5.29887259e-01 5.86670518e-01 -1.73804536e-01
-4.84442145e-01 4.79880303e-01 2.68293768e-01 5.52827455e-02
3.17142427e-01 3.59221995e-01 -7.28630722e-01 -1.88804179e-01
-1.00144529e+00 6.06137514e-01 7.40100205e-01 7.51839697e-01
7.31602490e-01 -1.25655472e-01 -7.48278350e-02 7.92514443e-01
3.37363631e-01 5.77062249e-01 1.66200981e-01 -1.31189394e+00
4.64681357e-01 4.35089588e-01 4.45628673e-01 -9.47831571e-01
-9.27538574e-01 -6.09288394e-01 -5.84985614e-01 2.55474091e-01
6.11576974e-01 -3.19326311e-01 -1.13126361e+00 1.62403417e+00
7.74602532e-01 1.69709697e-01 -3.50820929e-01 1.21152413e+00
5.76608479e-01 2.99573451e-01 -2.90674001e-01 1.49366125e-01
1.02107596e+00 -9.99849677e-01 -1.23626091e-01 -4.60257560e-01
4.24970776e-01 -6.49168074e-01 1.06692767e+00 3.10304612e-01
-1.50686514e+00 -6.12783432e-01 -1.01158893e+00 -1.87144622e-01
1.17423885e-01 3.10196549e-01 3.14533919e-01 4.18546379e-01
-1.26630759e+00 8.56893718e-01 -1.07743561e+00 -4.82532173e-01
5.07978201e-01 6.81745291e-01 -6.97433472e-01 2.29604468e-01
-4.50419366e-01 1.18175578e+00 3.11416566e-01 2.24132031e-01
-8.15795362e-01 -9.43611026e-01 -8.24020505e-01 -3.42391968e-01
5.58542490e-01 -1.19870806e+00 1.32986200e+00 -6.34221494e-01
-1.86008024e+00 1.21894693e+00 -1.32284135e-01 -6.27198279e-01
9.23183203e-01 -7.56145477e-01 3.00423324e-01 9.21032727e-02
-1.61375597e-01 1.10498083e+00 9.88082647e-01 -1.18623376e+00
-3.85062218e-01 -4.59427416e-01 -6.87821358e-02 4.74475741e-01
-1.34908944e-01 -9.01208743e-02 -1.14106476e+00 -3.57971042e-01
2.85620183e-01 -1.23541296e+00 -4.17171985e-01 3.93337131e-01
-3.83318007e-01 -7.32224435e-02 4.03821558e-01 -5.98147810e-01
6.33309543e-01 -1.63096166e+00 4.31672931e-01 4.23996866e-01
3.04097742e-01 1.41495332e-01 -3.53633761e-01 3.37629430e-02
8.37030075e-03 -3.03775668e-01 -1.84390709e-01 -1.09615171e+00
2.74412721e-01 2.30256557e-01 7.34921917e-02 9.82003868e-01
1.64092869e-01 1.27986526e+00 -5.53675950e-01 -3.34371746e-01
3.94525588e-01 5.34215331e-01 -9.88591373e-01 4.06749457e-01
-2.04117805e-01 6.28087640e-01 -8.21455792e-02 3.32039863e-01
5.00213087e-01 -4.48683769e-01 -6.08417131e-02 -5.13164520e-01
5.37777580e-02 1.17607169e-01 -1.48004687e+00 2.24487257e+00
-3.44867200e-01 4.14025635e-01 4.43379395e-02 -9.53178465e-01
7.59870529e-01 9.40898135e-02 7.05713391e-01 -2.72042572e-01
3.66720140e-01 1.76538020e-01 -2.32399210e-01 -3.18339318e-01
3.32106262e-01 4.30506282e-02 1.96347326e-01 5.73783100e-01
4.64541942e-01 -2.12671995e-01 1.18909217e-02 -1.53055936e-01
8.85034621e-01 7.37183988e-01 2.20865771e-01 -1.80456400e-01
3.61482531e-01 -9.80914608e-02 5.21631837e-01 5.43066084e-01
-8.28055516e-02 8.96125138e-01 9.63815376e-02 -9.90087688e-01
-1.53419948e+00 -9.77133214e-01 3.96822318e-02 1.01989830e+00
-1.20805964e-01 -2.80636281e-01 -7.68409550e-01 -6.78411424e-01
1.60430208e-01 2.84442425e-01 -6.05482876e-01 1.22763626e-01
-9.67616916e-01 -5.26367247e-01 4.95300233e-01 5.72192073e-01
4.36039865e-01 -8.79660428e-01 -7.82433927e-01 2.59773552e-01
-2.43290998e-02 -1.37748289e+00 -8.53889167e-01 -1.19078634e-02
-9.54862177e-01 -1.06908500e+00 -7.84307480e-01 -6.16587818e-01
9.57628965e-01 -2.57622123e-01 1.32213163e+00 8.83052200e-02
-4.11837935e-01 7.22081602e-01 1.30758047e-01 -2.13921741e-01
-1.46341845e-01 -2.70715933e-02 3.18249255e-01 -4.91507016e-02
1.42149642e-01 -5.59999824e-01 -7.72684753e-01 2.69360811e-01
-2.48054326e-01 -7.11051421e-03 4.70259786e-01 7.20064104e-01
8.71416450e-01 -5.48231840e-01 -1.49075791e-01 -5.29342532e-01
3.76205117e-01 1.41211197e-01 -8.70393693e-01 4.95235510e-02
-3.64301622e-01 4.23811853e-01 3.48975629e-01 -4.98034984e-01
-6.64604843e-01 7.02772379e-01 -3.70327115e-01 -7.55449295e-01
-2.71687448e-01 1.33131281e-01 1.90824673e-01 -3.82210582e-01
1.00852525e+00 -1.27723580e-02 2.45681897e-01 -7.01386750e-01
5.97868145e-01 2.70532984e-02 9.42979813e-01 -7.47624457e-01
1.02611208e+00 6.04744732e-01 2.00737745e-01 -6.41496897e-01
-9.48635340e-01 -5.28259575e-01 -1.08936155e+00 -2.64434278e-01
8.46810520e-01 -9.25733209e-01 -9.14525270e-01 4.79944497e-01
-1.26089859e+00 -5.61598599e-01 -4.18686032e-01 4.80443388e-01
-9.45826471e-01 4.85571861e-01 -5.66144586e-01 -4.94369388e-01
-6.02643788e-01 -1.11384773e+00 1.14192986e+00 -7.23937899e-02
-3.54025215e-01 -8.85736227e-01 2.62325138e-01 3.63852382e-01
9.65872481e-02 3.91680241e-01 3.44397873e-01 -5.24192929e-01
-5.19454062e-01 -3.47591043e-01 3.03013995e-02 2.15434641e-01
-2.67420977e-01 -2.93617874e-01 -8.26596677e-01 -6.14182293e-01
-1.73958391e-01 -5.89525223e-01 5.52876890e-01 5.70909500e-01
1.06073630e+00 -3.05411428e-01 -6.55458868e-02 1.12652254e+00
1.17609835e+00 -6.86896265e-01 3.11977655e-01 3.99775922e-01
8.79188538e-01 3.66814077e-01 4.50617462e-01 5.48244357e-01
5.40860355e-01 9.42318082e-01 4.80797589e-01 -1.96133465e-01
-1.82714209e-01 -2.69394755e-01 1.44557625e-01 4.21779305e-01
-3.79459232e-01 2.29982212e-01 -9.24077034e-01 4.90274251e-01
-1.85547984e+00 -6.12292528e-01 2.87500650e-01 2.42121124e+00
9.94845331e-01 2.44985968e-01 5.49885571e-01 -3.53377283e-01
2.89830476e-01 2.70538498e-02 -6.72718287e-01 4.83558811e-02
1.93274111e-01 3.97409111e-01 7.10027456e-01 7.79708982e-01
-1.24652863e+00 1.13858962e+00 6.20950174e+00 3.40480745e-01
-8.95948529e-01 -1.25545841e-02 4.28239882e-01 -4.86847043e-01
1.46651283e-01 -2.27483898e-01 -9.61853564e-01 -6.15334734e-02
6.45511746e-01 2.84675330e-01 6.78885877e-01 7.68243849e-01
2.54332960e-01 1.56382233e-01 -1.36221969e+00 1.34946311e+00
1.00574642e-02 -1.40829599e+00 -2.53420174e-01 -1.47568043e-02
7.74760187e-01 4.66676891e-01 3.09298243e-02 -2.29556803e-02
3.92619669e-01 -9.81458127e-01 9.98232543e-01 6.19853854e-01
6.22570813e-01 -7.51210749e-01 4.22063231e-01 6.41652465e-01
-9.21293914e-01 7.97245353e-02 -5.21758080e-01 1.75011784e-01
2.74380863e-01 3.67243171e-01 -1.00158429e+00 1.52281031e-01
7.66452134e-01 5.81773996e-01 -4.40093398e-01 1.18772149e+00
-3.08719039e-01 1.68239683e-01 -7.92518616e-01 1.23250328e-01
1.86283022e-01 -1.21762395e-01 8.23441684e-01 1.24063790e+00
2.49295741e-01 1.23077016e-02 5.25497675e-01 8.02508414e-01
-2.00336084e-01 2.17463952e-02 -3.19818377e-01 2.76694655e-01
2.31015012e-01 1.24031055e+00 -5.91931105e-01 -1.70986518e-01
-1.03729293e-01 1.20424426e+00 6.05111241e-01 2.62104899e-01
-6.17738366e-01 1.25698388e-01 5.94037414e-01 2.70448804e-01
5.55809677e-01 -7.22366929e-01 -2.55251110e-01 -1.06728518e+00
1.94834933e-01 -1.04355478e+00 2.40162328e-01 -5.35831451e-01
-1.16413450e+00 4.15240437e-01 7.12528378e-02 -8.65072608e-01
-5.05427241e-01 -7.64826775e-01 -2.25249425e-01 8.47535133e-01
-1.21233392e+00 -1.25672221e+00 -2.45707303e-01 5.58095694e-01
4.19647217e-01 -1.53811187e-01 6.37376845e-01 4.06049797e-03
-3.69432151e-01 6.98410153e-01 -3.05619866e-01 2.85286278e-01
7.95184553e-01 -1.40133691e+00 9.34045434e-01 7.16731369e-01
4.49729562e-01 6.37259185e-01 8.30746710e-01 -4.90023434e-01
-1.53274882e+00 -7.58470893e-01 7.56343901e-01 -8.19227755e-01
3.62577736e-01 -3.68058294e-01 -5.27322650e-01 7.03812778e-01
1.85198355e-02 4.23085064e-01 2.44890034e-01 2.50869900e-01
-1.74713984e-01 2.02258937e-02 -9.86656547e-01 6.96570218e-01
1.27430475e+00 -3.77101451e-01 -5.43397009e-01 5.10034740e-01
5.28730273e-01 -1.22432148e+00 -7.75589526e-01 2.26864204e-01
6.49222970e-01 -8.50600958e-01 1.44168782e+00 -8.55746150e-01
-2.85976734e-02 -3.83067548e-01 -5.17061800e-02 -1.14483976e+00
-4.03623551e-01 -9.65530038e-01 -3.63200992e-01 4.38646764e-01
3.40517044e-01 -2.97207654e-01 1.16683269e+00 8.62147868e-01
3.14890109e-02 -9.70866382e-01 -8.42308164e-01 -6.85512662e-01
-1.68582369e-02 -5.41188955e-01 3.95230561e-01 3.92177075e-01
-5.58304608e-01 2.39412054e-01 -5.58901906e-01 2.46362686e-01
9.57704425e-01 -1.43430993e-01 1.42686689e+00 -1.08295262e+00
-6.35600924e-01 -3.28890920e-01 -3.83080810e-01 -1.58314693e+00
1.66273758e-01 -8.06053400e-01 1.41028091e-01 -1.30603266e+00
1.22511119e-01 -1.85884297e-01 1.07943080e-01 6.80867612e-01
-1.26371965e-01 4.25884426e-01 4.42139000e-01 1.73369333e-01
-6.69245541e-01 4.65453327e-01 1.12612307e+00 7.93191567e-02
-3.41192245e-01 9.96514186e-02 -1.50329426e-01 9.74880993e-01
4.87242430e-01 -4.51757491e-01 -1.70410156e-01 -6.25677228e-01
3.20785791e-01 -2.04886630e-01 6.78371429e-01 -1.02532113e+00
3.13804835e-01 6.52885661e-02 6.77738070e-01 -7.60123670e-01
6.39666617e-01 -7.50542581e-01 1.02838174e-01 3.90809000e-01
-2.97376752e-01 2.17678025e-01 2.42139593e-01 4.44675565e-01
4.78521466e-01 -1.99206352e-01 9.63341355e-01 -2.16887757e-01
-6.29340351e-01 7.31505990e-01 2.98188031e-01 2.84612626e-01
5.31310022e-01 -2.67227679e-01 4.38718230e-01 -5.46632707e-01
-1.01431680e+00 2.15609238e-01 6.69910610e-01 3.54487449e-01
6.10038996e-01 -1.40360618e+00 -8.25344920e-01 2.02299938e-01
-2.09677175e-01 2.75355279e-01 -9.98385772e-02 9.13961709e-01
-7.61485696e-01 2.14259371e-01 -5.38899563e-02 -9.84802902e-01
-1.20552146e+00 3.05616438e-01 7.76229560e-01 -3.07735175e-01
-9.81535971e-01 1.16221988e+00 -1.70335695e-01 -8.66112471e-01
4.74620372e-01 -9.79754999e-02 1.86357051e-01 -3.17615181e-01
2.43939400e-01 3.08402479e-01 1.43855646e-01 -8.18843067e-01
-3.57579768e-01 9.32462811e-01 4.77429554e-02 -3.75959486e-01
1.54286218e+00 1.39801186e-02 1.36123344e-01 1.90119907e-01
1.51110411e+00 -1.40720621e-01 -1.83841634e+00 -4.73311305e-01
-5.48641533e-02 -5.35064757e-01 3.49370241e-02 -7.32698083e-01
-1.10110855e+00 7.50032306e-01 6.93049490e-01 -5.10150135e-01
6.40103996e-01 -1.21775120e-01 8.58152747e-01 8.60413790e-01
3.37797105e-01 -1.41258037e+00 4.34877932e-01 7.06457019e-01
1.12265027e+00 -1.21908939e+00 3.54737461e-01 9.77457687e-02
-5.13397336e-01 1.17433298e+00 4.91074294e-01 -5.19111097e-01
5.30194700e-01 2.00372219e-01 1.46668345e-01 -2.21254244e-01
-4.80033070e-01 -1.96997300e-01 7.70421565e-01 4.91313100e-01
2.40014032e-01 -3.01493198e-01 1.55401334e-01 5.56843206e-02
-4.36452866e-01 -3.73997800e-02 -5.82487471e-02 7.29994953e-01
-4.17440087e-01 -1.18559444e+00 -3.71076494e-01 2.33039632e-01
-2.68616080e-01 1.42318219e-01 -2.27477729e-01 7.40345240e-01
-3.93908843e-02 3.86981815e-01 9.30650160e-03 -1.66918725e-01
6.52375519e-01 7.74406716e-02 9.25641239e-01 -3.90771031e-01
-6.13893509e-01 2.60628045e-01 1.36727571e-01 -1.14072180e+00
-4.10410762e-01 -8.54685068e-01 -1.38690901e+00 -3.46430242e-01
-2.17820227e-01 -3.90754640e-01 6.57088935e-01 1.05438387e+00
5.10897100e-01 3.14306356e-02 2.68534333e-01 -1.40396905e+00
-8.71203423e-01 -6.84106827e-01 -2.38102302e-02 5.38734376e-01
2.72578806e-01 -7.37718999e-01 3.07079758e-02 -2.42851153e-02] | [7.048025131225586, -0.9460397958755493] |
7bb85593-1c84-4d42-bf69-3c49a1b6fdee | integrating-knowledge-supported-search-into | null | null | https://aclanthology.org/D18-2022 | https://aclanthology.org/D18-2022.pdf | Integrating Knowledge-Supported Search into the INCEpTION Annotation Platform | Annotating entity mentions and linking them to a knowledge resource are essential tasks in many domains. It disambiguates mentions, introduces cross-document coreferences, and the resources contribute extra information, e.g. taxonomic relations. Such tasks benefit from text annotation tools that integrate a search which covers the text, the annotations, as well as the knowledge resource. However, to the best of our knowledge, no current tools integrate knowledge-supported search as well as entity linking support. We address this gap by introducing knowledge-supported search functionality into the INCEpTION text annotation platform. In our approach, cross-document references are created by linking entity mentions to a knowledge base in the form of a structured hierarchical vocabulary. The resulting annotations are then indexed to enable fast and yet complex queries taking into account the text, the annotations, and the vocabulary structure. | ['Jan-Christoph Klie', 'Iryna Gurevych', 'Beto Boullosa', 'Naveen Kumar', 'Richard Eckart de Castilho'] | 2018-11-01 | null | null | null | emnlp-2018-11 | ['text-annotation'] | ['natural-language-processing'] | [-1.80607036e-01 6.45402789e-01 -5.59403539e-01 -8.15200359e-02
-6.33443594e-01 -1.03492725e+00 6.03145778e-01 1.06676042e+00
-5.99743366e-01 9.31805193e-01 5.13838828e-01 -1.17641836e-01
-3.89358521e-01 -8.31930339e-01 -1.53516039e-01 9.38982219e-02
1.70061544e-01 8.81472528e-01 8.49725008e-01 -2.67270476e-01
2.35634029e-01 1.30863383e-01 -1.46324587e+00 3.08399826e-01
9.31891799e-01 8.17592561e-01 1.30829498e-01 -1.30811393e-01
-1.01521420e+00 4.69447136e-01 -4.99695539e-01 -5.91567874e-01
-3.71477783e-01 1.62374839e-01 -1.29546130e+00 -4.96154875e-01
2.24774569e-01 2.67651021e-01 1.92826927e-01 1.02459288e+00
1.10428542e-01 1.21568412e-01 4.71315563e-01 -8.98228824e-01
-1.82914540e-01 1.11841750e+00 -1.50092751e-01 -4.71867174e-02
6.06180012e-01 -6.88528776e-01 1.20435262e+00 -8.06945860e-01
1.18783760e+00 8.49861801e-01 5.86984992e-01 2.31110364e-01
-8.64847720e-01 -3.05916995e-01 5.01456894e-02 -7.69900717e-03
-1.18513477e+00 -4.32859331e-01 1.89674929e-01 -7.31771708e-01
1.35273254e+00 3.78778964e-01 2.86431909e-01 4.94531959e-01
-4.51651156e-01 3.57216783e-02 3.98856372e-01 -6.43942416e-01
4.92783375e-02 5.45818865e-01 4.81454581e-01 6.44062757e-01
6.63308442e-01 -5.54550231e-01 -5.18699050e-01 -2.84815043e-01
2.98634380e-01 -5.90605438e-01 -2.90157616e-01 -4.63322550e-01
-9.09595847e-01 3.46674800e-01 3.46793056e-01 7.38878429e-01
-4.50318605e-01 -2.93669552e-01 7.83418536e-01 1.36194918e-02
4.54486996e-01 8.74944270e-01 -9.00133073e-01 3.22514743e-01
-8.38354588e-01 -2.93517094e-02 1.34285808e+00 1.28227222e+00
1.07461631e+00 -7.89917529e-01 4.80853729e-02 7.33003974e-01
4.12315905e-01 -1.40713781e-01 4.05045867e-01 -8.85147691e-01
5.98172069e-01 1.16730058e+00 5.59139967e-01 -6.70797110e-01
-6.12093270e-01 -3.95499706e-01 -9.85304341e-02 -3.96952599e-01
3.58097136e-01 -2.16172189e-02 -4.59139258e-01 1.47526503e+00
7.57370949e-01 -1.68596968e-01 1.91479936e-01 3.86373669e-01
1.58504355e+00 3.48235406e-02 4.78227824e-01 -5.61350226e-01
1.73692405e+00 -4.46625054e-01 -1.02048159e+00 7.60201588e-02
1.16277504e+00 -8.92681241e-01 5.31249106e-01 -3.16303223e-01
-9.82062519e-01 -2.67004818e-01 -7.04437017e-01 -3.36536855e-01
-1.12788343e+00 6.20141067e-03 5.23654819e-01 3.34922522e-01
-7.11851418e-01 4.80154037e-01 -6.75254583e-01 -8.47512722e-01
-2.02460691e-01 2.61364281e-01 -4.97772336e-01 2.36812040e-01
-1.65107560e+00 1.18087244e+00 1.16638911e+00 -4.37711775e-01
1.38922229e-01 -1.00330329e+00 -1.01501906e+00 1.93962112e-01
1.05910230e+00 -8.23567033e-01 1.17436445e+00 -5.90584040e-01
-8.39594841e-01 1.06640375e+00 -7.89526477e-02 -4.12775904e-01
4.87420782e-02 -8.10591057e-02 -5.13982773e-01 5.61940297e-02
6.30984664e-01 3.78437638e-01 -2.44063318e-01 -8.79314065e-01
-8.08149397e-01 -4.72903788e-01 3.82367522e-01 2.60755330e-01
-2.80650198e-01 4.33026254e-01 -8.78859162e-01 -4.42438692e-01
8.21524188e-02 -4.19024199e-01 -7.01695634e-03 -4.20060784e-01
-3.94857645e-01 -4.02049899e-01 5.60537815e-01 -8.48290265e-01
1.74585354e+00 -1.90430486e+00 6.09704927e-02 3.74049842e-01
1.68878347e-01 -1.52126709e-02 3.08645517e-01 9.71191943e-01
1.35959452e-02 3.40241879e-01 1.30111054e-01 7.77795687e-02
3.14021647e-01 3.25960606e-01 -3.09672147e-01 -2.47159570e-01
-2.17677668e-01 7.83897698e-01 -1.00413537e+00 -8.45595360e-01
-8.86478499e-02 2.67102987e-01 -3.48393023e-01 -2.65889645e-01
-6.86594546e-01 1.63480476e-01 -6.12568736e-01 3.67914528e-01
-9.69733521e-02 -2.04758614e-01 7.38763034e-01 -6.32526577e-01
-5.93417525e-01 8.72936249e-01 -1.21861649e+00 1.54632354e+00
-2.89689124e-01 1.58215329e-01 5.94805852e-02 -6.55977488e-01
7.54800975e-01 6.06023014e-01 5.60191989e-01 -1.30174071e-01
-9.10988748e-02 6.39577806e-01 -4.46820706e-01 -5.25159121e-01
9.09108162e-01 2.72561938e-01 -2.06606567e-01 1.14266962e-01
3.14883411e-01 2.18696192e-01 7.54700840e-01 5.67541957e-01
8.64380479e-01 4.59187448e-01 9.89460230e-01 -1.94335580e-01
6.84001267e-01 4.72450048e-01 4.15686011e-01 4.00485784e-01
6.30019188e-01 -4.18831885e-01 4.25758898e-01 -2.19774365e-01
-7.40508616e-01 -3.72151732e-01 -3.12508851e-01 1.16318142e+00
3.19269821e-02 -1.06440210e+00 -5.89251637e-01 -9.18611050e-01
8.79171211e-03 6.83855832e-01 -5.06129861e-01 3.92817110e-01
-2.32596487e-01 -1.83552831e-01 3.41143072e-01 4.26226854e-01
1.64041281e-01 -8.07342887e-01 -4.67134595e-01 3.80531967e-01
-3.94707799e-01 -1.26366186e+00 -2.72170365e-01 2.80406654e-01
-4.00950432e-01 -1.33160806e+00 -2.23474979e-01 -7.56838143e-01
4.01624352e-01 -5.29923201e-01 1.29130101e+00 1.52604073e-01
-5.53359985e-02 3.66310000e-01 -4.87813771e-01 -3.58096391e-01
-4.64082748e-01 4.83228475e-01 -3.92972678e-01 -6.95432007e-01
5.99620283e-01 -2.91938663e-01 2.38677561e-02 3.86133015e-01
-8.07810366e-01 -1.52761713e-01 1.02086611e-01 4.36143458e-01
6.06607974e-01 3.53149697e-02 6.38508856e-01 -1.11012077e+00
4.12698835e-01 -6.11368001e-01 -9.88732159e-01 7.89728403e-01
-7.34529734e-01 2.90031046e-01 -1.11404294e-02 -9.16604623e-02
-1.16938472e+00 4.25460003e-02 -5.29794358e-02 2.96290040e-01
-2.71725774e-01 1.48063564e+00 -4.19336915e-01 2.42568731e-01
6.88695133e-01 -3.82615030e-01 -5.25699615e-01 -8.04809928e-01
6.59486175e-01 6.37895703e-01 5.94298780e-01 -7.33529031e-01
4.33303297e-01 -1.54600307e-01 7.84273073e-02 -4.84791130e-01
-1.06148195e+00 -9.67325747e-01 -1.03145874e+00 1.03707708e-01
8.36467922e-01 -8.49393010e-01 -4.72491235e-01 -5.07318676e-01
-1.17421019e+00 4.86756042e-02 -4.91837174e-01 4.46726680e-01
-1.17102861e-01 5.24781942e-01 -2.86928892e-01 -5.05528271e-01
-3.01025778e-01 -6.35350466e-01 7.36038148e-01 4.05649915e-02
-7.29413688e-01 -1.09847152e+00 1.61848933e-01 4.08675015e-01
5.60216904e-02 1.28108442e-01 1.28388739e+00 -1.38382959e+00
-4.29296643e-01 -2.56464839e-01 -2.50177711e-01 -5.69934309e-01
1.04381174e-01 5.08592203e-02 -4.34488565e-01 2.60170788e-01
-8.05434406e-01 2.45179329e-02 5.89520752e-01 -1.89812750e-01
4.46729362e-01 -3.26383680e-01 -8.29045832e-01 2.27578923e-01
1.32151163e+00 8.68757144e-02 3.72494787e-01 6.95647061e-01
3.62265259e-01 1.08887613e+00 6.29756272e-01 2.77599156e-01
5.61701775e-01 1.05181289e+00 2.20680520e-01 2.84258783e-01
-6.55404702e-02 -1.51630715e-01 -3.86220515e-01 5.20751595e-01
-5.90679981e-02 -2.02711731e-01 -1.10968447e+00 5.80760419e-01
-1.87653756e+00 -1.03349936e+00 -4.41759676e-01 2.10766029e+00
1.41133463e+00 2.01834273e-02 3.78883556e-02 8.36719293e-03
5.64220726e-01 -4.04084742e-01 -2.10943192e-01 5.89854084e-02
9.00060833e-02 7.05062151e-02 5.35270751e-01 7.85017788e-01
-1.06085169e+00 1.20267749e+00 5.71498871e+00 5.82467675e-01
-5.28452456e-01 1.29529089e-01 -3.04678589e-01 3.17382097e-01
-3.44834805e-01 4.74269241e-01 -1.30611706e+00 1.82645291e-01
6.89678907e-01 -4.60217208e-01 -9.32774171e-02 8.83510351e-01
-1.67961031e-01 -6.11045808e-02 -9.88038361e-01 2.99957365e-01
-1.74076527e-01 -1.56758165e+00 -1.03837131e-02 7.94119388e-02
3.55492443e-01 -1.26451060e-01 -5.65873682e-01 3.39890510e-01
6.22768879e-01 -5.71889281e-01 5.64097404e-01 4.49833333e-01
9.79629099e-01 -3.82865906e-01 1.02141595e+00 1.24932937e-01
-1.41834378e+00 2.85475135e-01 2.87106857e-02 3.56775522e-01
2.34477311e-01 6.07997835e-01 -8.12419653e-01 1.12301266e+00
5.65674901e-01 5.41625202e-01 -3.65983188e-01 1.01183391e+00
-4.56903458e-01 1.44361809e-01 -2.23966256e-01 5.24058044e-02
-1.25765940e-02 1.11128658e-01 4.47147876e-01 1.34181392e+00
2.86074460e-01 1.56075999e-01 5.20473182e-01 5.98417699e-01
-2.10601449e-01 7.51107097e-01 -2.37280920e-01 -1.92487910e-01
8.24239910e-01 1.13465369e+00 -7.17690766e-01 -6.84779108e-01
-6.04285717e-01 4.64533120e-01 2.35410705e-01 3.24222535e-01
-2.94407785e-01 -7.89035499e-01 4.32880998e-01 2.50939071e-01
2.84421712e-01 -5.46063147e-02 1.62344888e-01 -1.09811413e+00
-8.40780362e-02 -5.90066969e-01 8.36740375e-01 -8.33877146e-01
-9.99343455e-01 5.15771747e-01 4.10033077e-01 -7.15213954e-01
-4.69373316e-01 -2.95169711e-01 5.20780645e-02 9.16335881e-01
-1.35862672e+00 -1.15799904e+00 -2.51472861e-01 3.33924264e-01
3.33919823e-02 2.06600964e-01 1.09727156e+00 5.36343038e-01
-2.95485079e-01 1.40474036e-01 -2.40018040e-01 4.59290177e-01
9.32399869e-01 -1.19961500e+00 5.93618676e-02 4.40019637e-01
9.21236351e-02 1.07583499e+00 6.22082233e-01 -1.20419466e+00
-4.45437312e-01 -8.33991885e-01 1.65717435e+00 -4.85627621e-01
1.06440353e+00 6.90918714e-02 -1.23331475e+00 8.32965732e-01
4.17200089e-01 -2.63280362e-01 1.21753633e+00 5.95454216e-01
-7.37578988e-01 3.84196699e-01 -7.85142422e-01 2.38795951e-01
1.05125487e+00 -4.25362021e-01 -1.05373967e+00 3.22918922e-01
8.42665136e-01 -6.34038389e-01 -1.33533311e+00 2.58924216e-01
5.32601476e-01 -8.52093026e-02 8.63583982e-01 -6.46411538e-01
-9.19111297e-02 -4.54301953e-01 -1.90912455e-01 -9.08969164e-01
-4.14127409e-02 -4.27654237e-01 -1.73185825e-01 1.71010458e+00
9.06917572e-01 -5.39070547e-01 2.76750892e-01 8.28899145e-01
-2.70992756e-01 -7.62651265e-02 -6.03130162e-01 -7.35181987e-01
-3.60678881e-01 -2.67765019e-02 8.69806170e-01 1.37340617e+00
8.34113002e-01 5.97984254e-01 2.76189476e-01 3.91041398e-01
2.70431608e-01 1.25490442e-01 4.27200854e-01 -2.03658652e+00
6.20045960e-02 -3.25301677e-01 -2.16668636e-01 -5.01315773e-01
3.08856368e-01 -1.14618313e+00 -2.85553426e-01 -2.04450941e+00
-7.05134571e-02 -6.75189555e-01 -8.20070226e-03 8.28372657e-01
3.90596092e-02 -1.62216023e-01 -2.91394651e-01 3.80358607e-01
-7.86446631e-01 -1.26099139e-01 3.53726476e-01 2.28801355e-01
-3.51309627e-01 -4.45537329e-01 -7.63116181e-01 8.88564527e-01
6.12156093e-01 -6.28318787e-01 -1.34532616e-01 -4.15670455e-01
6.94633186e-01 -3.68677564e-02 -8.12656283e-02 -6.46010399e-01
6.31200194e-01 -2.60779858e-01 -6.95099533e-02 -5.42789936e-01
7.69331083e-02 -8.44267726e-01 4.75276172e-01 1.06641971e-01
-5.88168085e-01 -3.66474599e-01 1.95675299e-01 2.79008597e-01
-3.42313439e-01 -7.59872079e-01 2.44322091e-01 -5.37660658e-01
-7.72560954e-01 -1.06763661e-01 -1.54002324e-01 3.64195108e-01
7.33648181e-01 -9.86470878e-02 -6.49964333e-01 1.55772492e-01
-1.05979705e+00 3.33458334e-01 5.80921173e-01 2.83904642e-01
-1.67272687e-01 -9.36552644e-01 -1.38214052e-01 -3.12358737e-01
4.31188941e-01 -3.61615531e-02 -3.04384261e-01 6.82200193e-01
-6.88519478e-02 8.40561867e-01 -2.80082431e-02 2.85390131e-02
-1.51628900e+00 6.99678481e-01 1.36288524e-01 -5.27584016e-01
-1.31966218e-01 4.14121211e-01 -5.24317250e-02 -3.74863386e-01
3.48967761e-01 -2.19839200e-01 -9.65650082e-01 7.85182536e-01
5.00624180e-01 1.98096350e-01 3.78720671e-01 -7.00946033e-01
-6.08953834e-01 6.39285386e-01 1.28404811e-01 -2.06669867e-01
1.23109758e+00 -3.60951662e-01 -6.30993962e-01 2.64994353e-01
5.18545806e-01 4.55680102e-01 -2.02830181e-01 -6.44001603e-01
9.15225804e-01 1.24955140e-02 -1.46602452e-01 -9.93258238e-01
-4.53695297e-01 2.99807429e-01 -2.60487407e-01 4.02380258e-01
6.55237794e-01 5.70358038e-01 1.86114579e-01 6.88181698e-01
3.14560741e-01 -1.04808879e+00 -2.60123193e-01 6.80569768e-01
6.74589157e-01 -8.02312613e-01 9.99697000e-02 -1.08584356e+00
-3.62767726e-01 1.08227789e+00 5.70672870e-01 8.30522120e-01
5.30488610e-01 3.55649263e-01 2.02133097e-02 -5.91156423e-01
-7.71113336e-01 -8.62480223e-01 5.92392921e-01 6.02953255e-01
8.15863132e-01 -4.98526156e-01 -7.69933999e-01 7.64577866e-01
-7.90363699e-02 1.49375051e-01 1.92313179e-01 8.58816504e-01
-7.75527358e-01 -1.44602394e+00 -9.60498229e-02 3.15472990e-01
-8.16303849e-01 -4.96155411e-01 -6.86411440e-01 8.39258790e-01
4.79391426e-01 8.03666413e-01 1.37476549e-01 1.85272664e-01
4.19930220e-01 5.14351785e-01 1.07753247e-01 -1.28444803e+00
-9.23338234e-01 1.97859153e-01 7.36628354e-01 -1.26889527e-01
-7.70401955e-01 -5.61663628e-01 -1.46763670e+00 2.07092598e-01
-7.88000286e-01 8.08608472e-01 6.85256481e-01 1.05752707e+00
4.57199901e-01 4.05393690e-01 -4.76551205e-01 -5.54479212e-02
1.16089620e-02 -7.99109638e-01 -2.27220923e-01 1.51651502e-01
-2.87168056e-01 -6.01622999e-01 9.00737126e-04 2.29760408e-01] | [9.333934783935547, 8.68156909942627] |
7bbf0a80-0fa2-40df-b9ef-d99f35fc9b14 | perceptual-image-restoration-with-high | 2103.03010 | null | https://arxiv.org/abs/2103.03010v1 | https://arxiv.org/pdf/2103.03010v1.pdf | Perceptual Image Restoration with High-Quality Priori and Degradation Learning | Perceptual image restoration seeks for high-fidelity images that most likely degrade to given images. For better visual quality, previous work proposed to search for solutions within the natural image manifold, by exploiting the latent space of a generative model. However, the quality of generated images are only guaranteed when latent embedding lies close to the prior distribution. In this work, we propose to restrict the feasible region within the prior manifold. This is accomplished with a non-parametric metric for two distributions: the Maximum Mean Discrepancy (MMD). Moreover, we model the degradation process directly as a conditional distribution. We show that our model performs well in measuring the similarity between restored and degraded images. Instead of optimizing the long criticized pixel-wise distance over degraded images, we rely on such model to find visual pleasing images with high probability. Our simultaneous restoration and enhancement framework generalizes well to real-world complicated degradation types. The experimental results on perceptual quality and no-reference image quality assessment (NR-IQA) demonstrate the superior performance of our method. | ['Jianhua Lu', 'Xiaoming Tao', 'Yiping Duan', 'Chaoyi Han'] | 2021-03-04 | null | null | null | null | ['no-reference-image-quality-assessment'] | ['computer-vision'] | [ 4.38694566e-01 -4.63411398e-02 1.00156851e-01 -2.11003542e-01
-1.01573002e+00 -2.59668291e-01 3.05729717e-01 -4.08776969e-01
-1.15893222e-01 7.35600233e-01 4.00579214e-01 1.26257213e-02
-3.37853819e-01 -6.30085528e-01 -6.86639547e-01 -1.07808304e+00
1.03787921e-01 -2.67434835e-01 -2.17970580e-01 3.88356261e-02
3.15275997e-01 3.58195633e-01 -1.47916341e+00 -5.52355312e-02
1.29753721e+00 8.02666485e-01 4.79145467e-01 6.57144070e-01
5.37596285e-01 4.34413671e-01 -6.23969555e-01 -1.98725343e-01
4.02431875e-01 -6.27931416e-01 -5.34618199e-01 7.25005805e-01
2.65300870e-01 -3.98206294e-01 -2.64907330e-01 1.55345845e+00
5.11979580e-01 1.97464570e-01 6.96608841e-01 -1.18156099e+00
-1.06034267e+00 -2.85950806e-02 -5.60486853e-01 1.45400003e-01
3.68007302e-01 3.78160179e-01 7.51542211e-01 -9.30425406e-01
3.77774984e-01 1.35922718e+00 2.43478358e-01 3.53029311e-01
-1.43436885e+00 -1.40011132e-01 1.49105899e-02 2.71739244e-01
-1.44775856e+00 -6.43650055e-01 8.00297797e-01 -3.89751941e-01
4.59207892e-01 2.60903537e-01 2.57455915e-01 6.77537978e-01
3.66484940e-01 6.02007270e-01 1.27738786e+00 -5.22857130e-01
2.99275488e-01 1.05420217e-01 -2.94339240e-01 4.03664976e-01
8.55345279e-02 2.72894055e-01 -3.00490648e-01 -5.40311262e-02
9.56876159e-01 -2.10305184e-01 -8.88681054e-01 -5.65139651e-01
-1.13598645e+00 6.40880406e-01 3.51957053e-01 2.69088030e-01
-5.64341962e-01 -8.88701752e-02 -1.97853729e-01 2.76594013e-01
3.88862789e-01 4.11265105e-01 6.44399449e-02 -5.25248703e-04
-9.80543673e-01 -1.27879098e-01 3.82161200e-01 6.64625347e-01
5.93432188e-01 9.52300057e-02 -4.38830882e-01 9.66385961e-01
5.03561854e-01 5.27741730e-01 2.11152509e-01 -1.45710480e+00
2.85994232e-01 1.28047273e-01 5.81314027e-01 -1.12446058e+00
2.66434997e-01 -7.60279000e-01 -1.05118334e+00 6.19196892e-01
3.89756650e-01 2.18914896e-01 -7.83412516e-01 1.80923617e+00
1.10503711e-01 2.32939925e-02 1.66025683e-01 1.33145535e+00
-1.08299954e-02 8.47437382e-01 -1.83605418e-01 -7.82548428e-01
8.32984149e-01 -8.97934496e-01 -9.56724763e-01 -9.46490914e-02
-1.00090191e-01 -9.19176042e-01 1.60005724e+00 5.64409435e-01
-1.45095026e+00 -8.53654325e-01 -1.18531132e+00 8.10123980e-02
4.52237278e-01 3.01292449e-01 -9.55111310e-02 6.87710583e-01
-1.26161945e+00 7.14127958e-01 -6.91548049e-01 -1.34634957e-01
2.43714124e-01 2.84441467e-02 -1.43203363e-01 -4.97937143e-01
-9.01586652e-01 8.42584193e-01 8.17323178e-02 2.73337901e-01
-1.26485598e+00 -4.26364303e-01 -6.68121040e-01 4.26180363e-02
1.58116981e-01 -8.34121943e-01 6.10502899e-01 -1.03777337e+00
-1.75381911e+00 7.38803566e-01 -1.98101282e-01 -1.27001435e-01
6.02536380e-01 -2.95032680e-01 -4.37972844e-01 3.63588333e-01
1.30672306e-01 4.87589687e-01 1.27192461e+00 -1.73831570e+00
-1.75084352e-01 -1.86294541e-01 7.71749094e-02 4.03286695e-01
-4.00089473e-01 -6.69303685e-02 -3.97724032e-01 -6.20852947e-01
3.07065725e-01 -6.55700922e-01 -2.09860831e-01 2.88161308e-01
-4.34387535e-01 6.53352290e-02 5.10307431e-01 -8.84097040e-01
1.06624901e+00 -2.45940208e+00 3.88641804e-01 1.15901254e-01
1.37942256e-02 6.64986018e-03 -2.33939618e-01 1.69946209e-01
-5.91586065e-03 -5.31248143e-03 -6.87693596e-01 -3.37628514e-01
1.02184974e-01 2.85082400e-01 -3.65976423e-01 9.81114864e-01
1.28270790e-01 4.82287258e-01 -8.19862604e-01 -4.53844398e-01
1.67940646e-01 6.43222153e-01 -5.20441949e-01 4.21363682e-01
1.55703992e-01 5.55214643e-01 -1.55101851e-01 4.58410889e-01
9.54422534e-01 -2.93072462e-01 -3.43266800e-02 -3.85418773e-01
1.11957878e-01 -2.18242258e-01 -1.21288812e+00 1.79108596e+00
-5.21977365e-01 3.61960381e-01 2.43752405e-01 -8.65256846e-01
9.38373208e-01 2.43111238e-01 3.04374695e-01 -4.55635875e-01
-1.24094896e-01 7.69823194e-02 -1.41917273e-01 -5.76555848e-01
2.21839547e-01 -3.45234126e-01 4.00302052e-01 2.33520105e-01
-1.63494334e-01 -4.96566184e-02 3.74335386e-02 3.93730588e-04
6.52800918e-01 1.79545209e-01 1.16507486e-01 -3.41776431e-01
5.93917072e-01 -5.87855279e-01 6.27696991e-01 6.21279657e-01
-4.41787779e-01 9.40256059e-01 3.40634644e-01 2.44603261e-01
-1.20793259e+00 -1.52191138e+00 -2.82780319e-01 4.43070292e-01
5.16136885e-01 1.77740231e-02 -7.04739571e-01 -2.86572218e-01
-4.24041033e-01 7.98701763e-01 -2.55091518e-01 -3.24655026e-01
-3.83319169e-01 -8.58872831e-01 4.86016311e-02 8.85713920e-02
6.26392901e-01 -7.24839509e-01 -4.05889273e-01 1.28028169e-01
-5.49686849e-01 -9.60195422e-01 -6.12737179e-01 -2.77408630e-01
-8.83335650e-01 -8.11360240e-01 -1.07428288e+00 -7.64309704e-01
8.42206001e-01 2.80217469e-01 1.01975441e+00 -4.18048017e-02
-2.32713506e-01 3.82519037e-01 -1.56672522e-01 5.60813725e-01
-5.67740202e-01 -7.50323713e-01 1.78035349e-01 4.15846258e-01
-3.73526335e-01 -7.35673010e-01 -9.46959734e-01 6.50288582e-01
-9.99049783e-01 -3.24701965e-02 6.37529433e-01 9.35086727e-01
7.56939232e-01 7.27126837e-01 4.70181674e-01 3.88016924e-02
6.93794668e-01 -3.53401661e-01 -4.37509269e-01 3.78140450e-01
-9.16322351e-01 -2.11332273e-02 4.78092611e-01 -5.15608013e-01
-1.18192160e+00 -2.05219030e-01 1.49959192e-01 -6.27090335e-01
-2.51259953e-01 2.33980000e-01 -6.61376536e-01 9.77633819e-02
5.68602562e-01 4.28418159e-01 5.45183681e-02 -4.77607429e-01
5.30698895e-01 5.45961022e-01 8.52751255e-01 -6.54732168e-01
9.76123929e-01 5.43448806e-01 9.45224687e-02 -6.71314597e-01
-5.90879381e-01 -3.28776896e-01 -3.97932440e-01 -4.55076039e-01
8.08259308e-01 -8.06877673e-01 -5.97497284e-01 3.92369062e-01
-9.05258179e-01 -1.64499596e-01 -3.91004711e-01 6.76587641e-01
-8.61275911e-01 9.56402838e-01 -5.32421350e-01 -1.15450132e+00
-7.12117255e-02 -1.30297506e+00 8.81036222e-01 5.79282679e-02
2.91241139e-01 -8.66714537e-01 -7.20059648e-02 3.65305662e-01
4.08137202e-01 2.43302375e-01 8.23939145e-01 4.77800936e-01
-6.37098074e-01 1.40536711e-01 -4.00688380e-01 8.49080145e-01
3.71335506e-01 -1.78917512e-01 -8.37184370e-01 -6.35747313e-01
6.44308090e-01 -1.69026792e-01 8.14440548e-01 7.18756258e-01
1.10394847e+00 -4.85936940e-01 8.75736922e-02 7.42562175e-01
1.66994703e+00 7.93094113e-02 1.12204123e+00 3.24343145e-01
4.06386256e-01 7.09552407e-01 7.23253489e-01 4.48240727e-01
3.41911539e-02 6.90133989e-01 6.32852972e-01 -1.43389359e-01
-4.06268239e-01 -1.66969046e-01 6.28303647e-01 7.50119507e-01
1.27647519e-01 -5.15687048e-01 -5.81157207e-01 6.69494689e-01
-1.59296668e+00 -8.92579138e-01 3.49510908e-02 2.31597924e+00
9.16448593e-01 5.89752942e-02 -2.56179512e-01 2.85547763e-01
9.46336806e-01 2.28705183e-02 -6.10007465e-01 1.54435128e-01
-5.28251052e-01 -3.13457072e-01 3.19258690e-01 7.97352314e-01
-8.65482867e-01 4.05433536e-01 6.54926252e+00 1.01938426e+00
-7.54127443e-01 7.50334039e-02 9.01107132e-01 1.78846866e-02
-4.65618223e-01 1.68293878e-01 -3.38980228e-01 5.04756689e-01
6.25557542e-01 -1.05294354e-01 7.03056455e-01 4.02044356e-01
6.49411261e-01 -2.26957217e-01 -8.27041745e-01 1.12801945e+00
1.32700890e-01 -7.98316121e-01 6.85597491e-03 2.45376810e-01
9.12841380e-01 -6.80397868e-01 5.22130549e-01 -2.88448572e-01
7.01704174e-02 -1.05582345e+00 7.20958173e-01 8.92446995e-01
8.01693201e-01 -6.63933635e-01 3.69908065e-01 3.42687845e-01
-6.53716922e-01 -8.25732499e-02 -6.22274160e-01 1.80772156e-01
4.51316863e-01 9.00825322e-01 -3.24738562e-01 5.07829845e-01
6.58066571e-01 7.24665880e-01 -4.25925225e-01 1.20065320e+00
-3.72695982e-01 4.83809173e-01 1.67725813e-02 6.78764701e-01
-2.03887358e-01 -5.69356441e-01 8.76815498e-01 7.51952052e-01
5.98256409e-01 2.52890922e-02 2.20128998e-01 1.09056950e+00
1.94451123e-01 8.05595070e-02 -2.55416751e-01 1.84471101e-01
8.33419785e-02 1.02102160e+00 -5.03997445e-01 -1.42596841e-01
-1.28717814e-02 1.40770888e+00 -8.32705051e-02 6.59750044e-01
-8.46888244e-01 -1.06359050e-01 5.71612954e-01 5.96595369e-02
1.07516110e-01 -3.08129936e-01 -2.89993107e-01 -1.24029827e+00
3.22190166e-01 -9.11402345e-01 7.30583593e-02 -1.12898684e+00
-1.49395251e+00 6.58290386e-01 -6.02797568e-02 -1.55787480e+00
2.85110413e-03 -1.86382189e-01 -4.68693346e-01 1.08987665e+00
-1.71128953e+00 -8.25149655e-01 -2.41382733e-01 8.47228408e-01
4.36497360e-01 7.12210536e-02 4.21607137e-01 3.76227945e-01
-4.77762789e-01 5.01756608e-01 3.39398474e-01 -3.07010382e-01
8.52259815e-01 -1.22329080e+00 -2.15506926e-01 1.28812242e+00
-2.40388721e-01 4.51516628e-01 1.08288682e+00 -5.64956844e-01
-1.14092970e+00 -8.99561226e-01 4.55760032e-01 6.74718171e-02
3.64288002e-01 1.65856451e-01 -9.99285161e-01 9.30158198e-02
3.81778032e-01 -3.25431600e-02 3.81582826e-01 -3.74376982e-01
-1.71000808e-01 -1.85421452e-01 -1.32581913e+00 5.51348150e-01
1.02234781e+00 -5.55039823e-01 -1.91746309e-01 3.84388208e-01
6.73833847e-01 1.42323570e-02 -1.09922707e+00 3.29751879e-01
1.37454137e-01 -9.24061537e-01 1.16083586e+00 -1.42583191e-01
5.80756128e-01 -7.76456118e-01 -6.14627361e-01 -1.39831483e+00
-4.60831761e-01 -6.87068522e-01 -9.08030942e-02 1.30137384e+00
6.22991435e-02 -3.95112038e-01 3.49889576e-01 2.97463179e-01
-1.49185807e-01 -4.15030509e-01 -9.12037849e-01 -1.09915078e+00
-1.08561851e-01 -1.41224027e-01 3.94679993e-01 6.87713623e-01
-3.38493407e-01 -8.56470838e-02 -6.78606808e-01 7.65721500e-01
1.29437256e+00 2.09287107e-01 4.45726961e-01 -7.28884876e-01
-7.35874176e-01 -3.07462484e-01 -3.50850105e-01 -1.36996770e+00
3.19828838e-02 -6.22613847e-01 3.63790423e-01 -1.58386147e+00
3.41362774e-01 -2.86856174e-01 -4.46847320e-01 -3.56363617e-02
-3.60029668e-01 3.32143962e-01 1.16965368e-01 6.23243570e-01
-4.23603296e-01 9.97718930e-01 1.74450123e+00 -2.68743932e-01
-6.80437237e-02 -1.13928504e-01 -5.71551323e-01 4.11867350e-01
7.51295030e-01 -3.31291497e-01 -5.81685305e-01 -5.03463447e-01
6.85804188e-02 4.37581062e-01 5.76410115e-01 -9.97976780e-01
-3.17974463e-02 -3.60212564e-01 4.28617001e-01 -3.31934214e-01
4.82245475e-01 -7.38664269e-01 1.75271615e-01 2.84637719e-01
-4.10012931e-01 -2.97526538e-01 -3.35008889e-01 7.98129857e-01
-3.63834918e-01 -2.12582961e-01 1.26252472e+00 1.79627597e-01
-3.57442915e-01 2.18637034e-01 -9.60529074e-02 -1.80694103e-01
8.61020923e-01 -3.35316151e-01 -2.31479347e-01 -7.17983127e-01
-1.01307583e+00 -7.52875805e-02 7.84430027e-01 2.79252350e-01
9.63646531e-01 -1.41023719e+00 -9.00970101e-01 8.09510797e-02
-4.09465060e-02 -4.86557364e-01 4.80333596e-01 7.74546504e-01
-2.10339308e-01 -8.61185044e-02 -2.14073747e-01 -8.09060991e-01
-9.92384553e-01 8.62906575e-01 5.59618950e-01 -9.58999991e-02
-5.58426976e-01 5.84478080e-01 2.94736922e-01 5.06792963e-02
1.86978042e-01 -2.72263139e-02 -7.86463693e-02 -5.02635896e-01
4.33870882e-01 5.26729584e-01 -2.67039329e-01 -8.05667877e-01
-7.52821639e-02 5.68747044e-01 3.56362104e-01 -4.39355910e-01
1.20617008e+00 -8.10072184e-01 -1.66132659e-01 2.04042971e-01
1.36404359e+00 9.03552584e-03 -1.82441139e+00 -1.17319621e-01
-2.73655325e-01 -1.14523816e+00 4.10661906e-01 -7.53537118e-01
-1.15761507e+00 8.31006765e-01 1.17586482e+00 2.10075051e-01
1.60277927e+00 -1.46374777e-01 4.88844812e-01 -3.17810684e-01
1.46188214e-01 -8.35429013e-01 5.06461322e-01 -1.25372142e-01
1.23805904e+00 -1.18666613e+00 5.65818995e-02 -3.50260019e-01
-4.98269320e-01 9.35638845e-01 4.45410579e-01 -1.36877060e-01
5.53227901e-01 -1.09292783e-01 8.70842114e-02 2.92580575e-02
-3.86085987e-01 -3.03192176e-02 3.46896291e-01 7.79278934e-01
1.51626274e-01 1.22323036e-02 -2.99885869e-01 1.49371522e-02
-1.91753041e-02 -1.06459789e-01 6.85117602e-01 6.28782272e-01
-4.91053998e-01 -9.82989430e-01 -6.82633638e-01 -5.19939438e-02
-4.48302895e-01 -6.33590743e-02 3.82529199e-01 9.66079682e-02
-1.09472591e-02 1.54378843e+00 -2.21797675e-01 1.69616938e-02
2.10294142e-01 -3.97441357e-01 6.14073336e-01 -2.06466839e-01
1.95555657e-01 5.01590490e-01 -3.11492443e-01 -5.96140087e-01
-5.49755991e-01 -5.87461948e-01 -7.63832033e-01 -1.08248122e-01
-3.46155167e-01 7.71271884e-02 6.36226773e-01 6.23084188e-01
1.44833818e-01 4.25205350e-01 1.12453783e+00 -6.51159167e-01
-8.15505862e-01 -7.56854355e-01 -8.33333492e-01 4.82550234e-01
5.78299880e-01 -4.79845554e-01 -7.31026888e-01 3.97196949e-01] | [11.44950008392334, -2.1646368503570557] |
87f7b56c-d382-42f2-84e1-208b26f28e12 | multi-channel-speech-separation-using | 2304.12023 | null | https://arxiv.org/abs/2304.12023v1 | https://arxiv.org/pdf/2304.12023v1.pdf | Multi-channel Speech Separation Using Spatially Selective Deep Non-linear Filters | In a multi-channel separation task with multiple speakers, we aim to recover all individual speech signals from the mixture. In contrast to single-channel approaches, which rely on the different spectro-temporal characteristics of the speech signals, multi-channel approaches should additionally utilize the different spatial locations of the sources for a more powerful separation especially when the number of sources increases. To enhance the spatial processing in a multi-channel source separation scenario, in this work, we propose a deep neural network (DNN) based spatially selective filter (SSF) that can be spatially steered to extract the speaker of interest by initializing a recurrent neural network layer with the target direction. We compare the proposed SSF with a common end-to-end direct separation (DS) approach trained using utterance-wise permutation invariant training (PIT), which only implicitly learns to perform spatial filtering. We show that the SSF has a clear advantage over a DS approach with the same underlying network architecture when there are more than two speakers in the mixture, which can be attributed to a better use of the spatial information. Furthermore, we find that the SSF generalizes much better to additional noise sources that were not seen during training. | ['Timo Gerkmann', 'Kristina Tesch'] | 2023-04-24 | null | null | null | null | ['speech-separation'] | ['speech'] | [ 3.78979295e-01 -1.51575908e-01 2.12480739e-01 -1.64488867e-01
-9.85218883e-01 -5.75318635e-01 5.24547160e-01 -3.75877529e-01
-3.69712472e-01 5.94126344e-01 6.24589860e-01 -1.74246624e-01
-5.30052781e-01 -2.34482422e-01 -6.40806377e-01 -1.02475238e+00
-8.29208940e-02 1.23761021e-01 2.16471329e-01 -1.13548629e-01
-5.32211885e-02 7.07615077e-01 -1.54382575e+00 3.25539142e-01
8.25842202e-01 7.60125935e-01 4.80324954e-01 7.76431739e-01
-3.02910190e-02 3.92884672e-01 -8.74224663e-01 3.15745890e-01
4.22911048e-01 -7.96120524e-01 -3.96090090e-01 3.00937351e-02
4.50399458e-01 1.09030589e-01 -3.96728069e-01 1.07536638e+00
7.30999351e-01 4.79904294e-01 5.72478592e-01 -8.24284077e-01
-1.28538221e-01 7.83542573e-01 -6.06339812e-01 3.83130610e-01
3.55023406e-02 -1.65725693e-01 6.95443094e-01 -9.82317626e-01
1.87492177e-01 1.25424504e+00 5.25656581e-01 4.47412014e-01
-1.22179210e+00 -7.30671346e-01 3.33975077e-01 1.42040461e-01
-1.38767779e+00 -1.00986540e+00 9.58332777e-01 -1.27677679e-01
6.33512855e-01 3.09021384e-01 2.36889020e-01 1.15484059e+00
-3.09033871e-01 8.98741245e-01 1.04627740e+00 -6.23945296e-01
2.46886373e-01 -4.77544777e-02 1.02007546e-01 6.20447472e-02
-4.14014533e-02 2.20534593e-01 -9.51621473e-01 1.53512806e-01
8.00874829e-01 -8.68327767e-02 -6.36839390e-01 -3.47706020e-01
-1.35880911e+00 5.45122385e-01 4.11225736e-01 8.29594553e-01
-5.07118821e-01 -1.37191996e-01 2.49317259e-01 2.96905577e-01
3.67492169e-01 5.65964043e-01 -4.44398850e-01 -3.19218822e-02
-1.52015889e+00 -7.95741677e-02 7.99747288e-01 5.29686630e-01
5.46265006e-01 4.84944612e-01 -2.00228989e-01 9.83040512e-01
2.32465610e-01 5.59280813e-01 6.96489513e-01 -7.97955930e-01
4.79916126e-01 -5.27523495e-02 -3.87524185e-03 -5.83672285e-01
-4.13895607e-01 -1.06267357e+00 -1.03151107e+00 3.21285665e-01
5.31724215e-01 -4.19091672e-01 -1.06212282e+00 1.93162978e+00
8.60065594e-02 6.04619741e-01 4.41565126e-01 1.04220116e+00
4.90584046e-01 6.41379058e-01 -4.30935264e-01 -2.38823548e-01
1.00941813e+00 -1.13158917e+00 -8.22904766e-01 -4.23484892e-01
7.38939121e-02 -7.33398199e-01 5.27934194e-01 5.21407068e-01
-9.88021672e-01 -7.26391315e-01 -1.10574961e+00 3.76395524e-01
-3.90751958e-01 2.85832644e-01 1.96133956e-01 7.51519620e-01
-1.15138853e+00 4.67422962e-01 -6.98136210e-01 -1.59663036e-01
1.96687952e-01 3.23816121e-01 -3.79155517e-01 8.73697028e-02
-1.20227826e+00 6.27957344e-01 2.09972277e-01 3.96646768e-01
-9.39043999e-01 -6.18443251e-01 -7.54816771e-01 3.55189025e-01
3.90709788e-01 -3.64394128e-01 9.47185695e-01 -1.29314005e+00
-1.83095241e+00 7.43454844e-02 -6.42270148e-01 -3.72117996e-01
7.89644718e-02 -2.76658326e-01 -6.31156087e-01 3.42472166e-01
7.00464100e-02 4.50983316e-01 1.34722352e+00 -1.30338037e+00
-6.23108804e-01 -3.92066151e-01 -3.38705808e-01 3.04145396e-01
-2.37235963e-01 6.64897114e-02 -2.54425436e-01 -9.63736534e-01
2.85692096e-01 -8.28821361e-01 -1.72290012e-01 -4.14465874e-01
-5.31349361e-01 2.59328604e-01 8.23525488e-01 -7.97106862e-01
1.01250851e+00 -2.45827699e+00 7.00883627e-01 2.87280113e-01
-7.71516711e-02 6.38555586e-01 -4.79299277e-01 3.61633629e-01
-3.82332802e-01 -3.26778561e-01 -3.22561145e-01 -7.23425686e-01
-1.12666391e-01 -1.22161523e-01 -2.90788651e-01 4.82774407e-01
1.75970532e-02 4.51937497e-01 -6.18268788e-01 8.87063053e-03
1.96092829e-01 6.82327032e-01 -3.04552048e-01 1.21406816e-01
2.85765588e-01 8.29424798e-01 9.05793980e-02 1.04379006e-01
9.59547818e-01 1.11647822e-01 1.77781150e-01 -6.78962618e-02
-1.97704807e-01 4.63838607e-01 -1.69099915e+00 1.96163774e+00
-6.25401974e-01 7.88321018e-01 9.02723312e-01 -1.11466777e+00
7.65963316e-01 6.45564973e-01 3.13605309e-01 -7.10900187e-01
-2.16086656e-01 4.05484349e-01 3.24218333e-01 -1.19011573e-01
2.41988540e-01 -3.64469886e-01 1.05066516e-01 4.40761119e-01
4.15093064e-01 2.57136524e-01 9.87400208e-03 5.04911058e-02
9.02886987e-01 -1.34806037e-01 5.27237244e-02 -3.19844574e-01
5.82213283e-01 -6.89483404e-01 4.90005910e-01 8.14423084e-01
-5.44887176e-03 7.88724124e-01 1.20750152e-01 2.20251828e-01
-5.18408656e-01 -1.21902609e+00 5.26051549e-03 1.12896633e+00
9.13265720e-02 -5.96323982e-02 -6.69602931e-01 -5.10216773e-01
-2.95614421e-01 7.14283884e-01 -2.90906966e-01 -1.00392997e-01
-8.18443418e-01 -5.19980788e-01 7.40447998e-01 4.89641815e-01
4.50027287e-01 -7.35631049e-01 -4.11854893e-01 2.66287267e-01
-2.28981569e-01 -1.01745784e+00 -5.51658869e-01 6.91181839e-01
-6.03494167e-01 -5.62608004e-01 -1.13611031e+00 -6.88252866e-01
3.39493573e-01 6.61437333e-01 4.43189174e-01 -5.20214081e-01
2.60962248e-01 2.58746952e-01 -2.99102247e-01 -3.36023062e-01
-2.19802096e-01 2.53858030e-01 1.98116511e-01 4.92060930e-01
1.90492243e-01 -8.96807551e-01 -4.57965672e-01 3.30566853e-01
-9.99372959e-01 -1.09465688e-01 8.11401248e-01 7.79550433e-01
8.60651582e-02 4.87344950e-01 1.01875842e+00 -3.49886477e-01
5.47790051e-01 -4.35614437e-01 -1.74131945e-01 4.70086597e-02
4.01181541e-02 2.44537100e-01 7.66077995e-01 -5.63794613e-01
-1.35745347e+00 1.73148349e-01 -3.21186900e-01 -5.34198701e-01
-5.81305921e-01 3.76874745e-01 -4.85910684e-01 9.79336426e-02
5.20726383e-01 4.22135562e-01 4.73866053e-02 -8.87132227e-01
4.14238006e-01 7.16189504e-01 3.31229508e-01 -1.34957731e-01
6.96863115e-01 4.77751076e-01 -5.05447537e-02 -1.07584882e+00
-5.48797965e-01 -8.11937034e-01 -8.89033020e-01 8.02209973e-02
6.93061054e-01 -9.30425584e-01 -3.30979913e-01 6.72861814e-01
-1.22547305e+00 -3.04685563e-01 -2.20731810e-01 9.68721092e-01
-2.04568058e-01 2.60468930e-01 -3.68612915e-01 -1.01130474e+00
1.51496917e-01 -1.09448004e+00 1.10067511e+00 1.17957726e-01
-3.40279490e-02 -1.02888703e+00 1.91857778e-02 6.15646876e-03
7.57499933e-01 -4.40787286e-01 4.97930884e-01 -8.92480612e-01
-3.68531108e-01 1.11589924e-01 9.30194333e-02 5.44213414e-01
4.59508896e-01 -5.77790916e-01 -1.37367988e+00 -3.94395053e-01
4.28971082e-01 2.20008343e-01 1.28654468e+00 7.35575080e-01
6.86432242e-01 -1.45695686e-01 -4.72808093e-01 6.01355910e-01
1.16814744e+00 3.74854088e-01 5.77549458e-01 1.88747924e-02
7.33856320e-01 7.87725866e-01 9.48962867e-02 8.97509232e-02
-1.48068622e-01 9.06196773e-01 2.89568305e-01 -4.85368907e-01
-5.49465239e-01 -4.58159931e-02 5.40412724e-01 7.28448033e-01
8.83585066e-02 -3.73886049e-01 -4.38503087e-01 6.38099372e-01
-1.67678404e+00 -1.07175064e+00 2.08640546e-01 2.45677233e+00
6.16835654e-01 -1.19463868e-01 2.82483369e-01 4.44926113e-01
8.12416971e-01 3.74210000e-01 -2.87046015e-01 -2.02832252e-01
-4.15608585e-01 4.12395835e-01 5.00811577e-01 8.09280932e-01
-1.08395803e+00 5.30860186e-01 6.23442507e+00 1.13367152e+00
-1.58820760e+00 1.80053070e-01 1.34011984e-01 -4.03827608e-01
-1.73922092e-01 -4.27739263e-01 -6.64034128e-01 4.52950597e-01
1.15146279e+00 4.35480386e-01 6.89044833e-01 1.54651821e-01
3.18686396e-01 -7.69019425e-02 -1.10405004e+00 8.39278698e-01
2.04542547e-01 -9.34924364e-01 -2.82477409e-01 -1.04003148e-02
4.46489692e-01 -8.25226009e-02 2.33960465e-01 1.33407876e-01
-7.51851797e-02 -1.19338119e+00 8.38261127e-01 4.91968542e-01
4.59349573e-01 -7.40662932e-01 4.25657451e-01 6.42757952e-01
-1.25189817e+00 -2.39015818e-01 -1.07173100e-01 9.17890854e-03
3.52059215e-01 8.98981869e-01 -7.81502068e-01 7.98446178e-01
5.51129580e-01 4.22572970e-01 -3.38970542e-01 1.06281435e+00
-2.37261385e-01 6.91694736e-01 -3.56311083e-01 3.67806822e-01
2.33229190e-01 -1.58380449e-01 1.09817469e+00 1.41787875e+00
5.30852497e-01 -3.12219530e-01 -1.98762834e-01 7.13699758e-01
2.04181150e-01 -1.68844491e-01 -6.00934565e-01 2.76329011e-01
3.92741621e-01 1.12224209e+00 -7.26521790e-01 -1.07972324e-01
-4.33183342e-01 1.14661515e+00 6.37393519e-02 8.44540596e-01
-6.02183580e-01 -5.81926167e-01 6.68784320e-01 9.42725316e-03
9.26779687e-01 -3.92704904e-01 -1.54954717e-01 -1.02191269e+00
-1.11859843e-01 -9.36286986e-01 -7.50038698e-02 -4.96139348e-01
-1.12741494e+00 7.97855318e-01 -1.17868990e-01 -1.26864791e+00
-3.37443918e-01 -5.04998386e-01 -6.22115135e-01 1.35363126e+00
-1.62086868e+00 -1.07982147e+00 3.26025009e-01 6.99419916e-01
6.33870780e-01 -2.60389686e-01 7.87291050e-01 4.83470291e-01
-4.60490465e-01 5.62074780e-01 3.60507518e-01 4.80021462e-02
9.07204986e-01 -1.12171912e+00 -1.49631882e-02 1.33887327e+00
4.58185911e-01 8.37016642e-01 6.18982494e-01 -3.38724226e-01
-1.05125165e+00 -1.02302134e+00 6.52139544e-01 2.15389431e-02
3.14275473e-01 -5.41636288e-01 -9.38325405e-01 6.01943552e-01
5.50053477e-01 -1.17780738e-01 7.84113228e-01 2.67613530e-01
-2.62407809e-01 -3.12249333e-01 -8.11088026e-01 4.21922654e-01
8.51898551e-01 -5.52354395e-01 -7.41084397e-01 -1.24751464e-01
4.82163429e-01 -2.00424373e-01 -2.36242890e-01 1.44430831e-01
3.75161707e-01 -1.10289872e+00 1.13571811e+00 -2.25452557e-01
-2.93763075e-02 -6.04000866e-01 -2.71536142e-01 -1.99678791e+00
-4.55561250e-01 -6.13371909e-01 1.53517768e-01 1.27375245e+00
6.39876962e-01 -8.39688301e-01 3.80120724e-01 -5.73486201e-02
-3.96384418e-01 -2.16008872e-01 -1.33416998e+00 -9.74361897e-01
1.45512177e-02 -3.72391164e-01 5.59607744e-01 6.62545502e-01
-4.76871058e-02 4.85031843e-01 -5.97671926e-01 6.01084888e-01
3.74952734e-01 1.37400880e-01 3.59155983e-01 -1.13455236e+00
-7.17384636e-01 -4.75774437e-01 -8.21818039e-03 -1.52103281e+00
3.41737151e-01 -7.11421967e-01 4.37855154e-01 -1.40523863e+00
-3.39204282e-01 -3.22138786e-01 -5.65862834e-01 5.38901575e-02
-1.79921657e-01 3.84837538e-02 2.76861250e-01 6.62704706e-02
-4.49783891e-01 5.36222875e-01 1.20259416e+00 -5.30526266e-02
-4.74699110e-01 4.82554466e-01 -8.87429297e-01 6.36141062e-01
5.44214845e-01 -3.29051673e-01 -4.66899037e-01 -5.58655441e-01
-2.58558542e-01 2.37296671e-01 3.98874223e-01 -1.22521448e+00
5.47806680e-01 1.94396436e-01 4.53367859e-01 -5.21629155e-01
7.60916889e-01 -1.03650832e+00 1.49084002e-01 1.61805734e-01
-4.38761622e-01 -5.33948660e-01 4.41985220e-01 5.99586308e-01
-5.79517484e-01 -3.81995887e-01 6.66067898e-01 -7.67095089e-02
-4.71996933e-01 -2.07328871e-01 -6.82687461e-01 -4.29402530e-01
6.81392252e-01 -2.42066115e-01 -3.83732542e-02 -5.48599422e-01
-8.83279324e-01 -2.65612185e-01 -8.22123140e-02 2.43350089e-01
4.88815993e-01 -1.26546228e+00 -6.86795890e-01 5.85965514e-01
-4.48855311e-01 -4.25750613e-02 5.31924546e-01 1.06844330e+00
2.58471072e-01 6.23885930e-01 -8.26445818e-02 -5.32744646e-01
-1.09861398e+00 6.43781006e-01 5.60799658e-01 -1.31847143e-01
-5.74286044e-01 1.05065787e+00 6.02020621e-01 -2.67895818e-01
3.65494460e-01 -2.21740529e-01 -3.06202948e-01 3.22490811e-01
6.95764840e-01 5.53648651e-01 2.66804636e-01 -8.69021535e-01
-4.53536570e-01 5.41932702e-01 2.42393032e-01 -6.30423725e-01
1.38180220e+00 -4.30858463e-01 -1.00482814e-02 5.29714763e-01
1.10169291e+00 7.87535667e-01 -1.35479641e+00 -4.29553241e-01
-2.61999309e-01 -4.02482569e-01 1.36348099e-01 -8.77731204e-01
-1.20887685e+00 1.13225567e+00 5.36516607e-01 2.97082901e-01
1.36733019e+00 -8.46172720e-02 4.84666288e-01 -7.35064223e-02
8.56013373e-02 -7.98600197e-01 7.05608502e-02 5.98638952e-01
9.16574597e-01 -8.00932467e-01 -5.30024052e-01 -4.29801315e-01
-5.82865357e-01 1.21532285e+00 1.52403936e-01 6.39439598e-02
8.36522102e-01 5.25003314e-01 1.49659246e-01 6.42974824e-02
-2.68827349e-01 -4.08814073e-01 4.07000095e-01 8.38764369e-01
2.45673418e-01 -3.26968022e-02 4.20266747e-01 7.47416317e-01
-5.69840185e-02 -3.98212582e-01 2.38406330e-01 6.11661971e-01
-4.82415587e-01 -1.16016233e+00 -8.45647454e-01 1.35510460e-01
-3.64540935e-01 -2.48672977e-01 -1.43794492e-01 3.77362043e-01
1.62326574e-01 1.22749436e+00 6.69965222e-02 -1.59664392e-01
2.67313123e-01 3.54751706e-01 3.17986012e-01 -7.35708237e-01
-4.04782712e-01 6.24755025e-01 -4.26187404e-02 -2.96707094e-01
-7.34726012e-01 -6.79347873e-01 -1.08422697e+00 2.07964793e-01
-3.80020201e-01 2.48850405e-01 6.69279099e-01 1.05327582e+00
4.90872443e-01 1.06698549e+00 5.83218932e-01 -1.26361215e+00
-3.01742226e-01 -1.23054028e+00 -9.70950246e-01 1.62633628e-01
9.41447973e-01 -6.49033904e-01 -6.68236732e-01 -7.85848722e-02] | [15.085515022277832, 5.770382404327393] |
b677dbdf-ec85-483d-b992-cf9a37bdaca7 | between-homomorphic-signal-processing-and | 1706.08231 | null | http://arxiv.org/abs/1706.08231v1 | http://arxiv.org/pdf/1706.08231v1.pdf | Between Homomorphic Signal Processing and Deep Neural Networks: Constructing Deep Algorithms for Polyphonic Music Transcription | This paper presents a new approach in understanding how deep neural networks
(DNNs) work by applying homomorphic signal processing techniques. Focusing on
the task of multi-pitch estimation (MPE), this paper demonstrates the
equivalence relation between a generalized cepstrum and a DNN in terms of their
structures and functionality. Such an equivalence relation, together with pitch
perception theories and the recently established
rectified-correlations-on-a-sphere (RECOS) filter analysis, provide an
alternative way in explaining the role of the nonlinear activation function and
the multi-layer structure, both of which exist in a cepstrum and a DNN. To
validate the efficacy of this new approach, a new feature designed in the same
fashion is proposed for pitch salience function. The new feature outperforms
the one-layer spectrum in the MPE task and, as predicted, it addresses the
issue of the missing fundamental effect and also achieves better robustness to
noise. | ['Li Su'] | 2017-06-26 | null | null | null | null | ['music-transcription'] | ['music'] | [ 4.47523594e-02 1.08349502e-01 2.75915086e-01 5.63403107e-02
-1.30843684e-01 -4.62975472e-01 6.02142572e-01 2.33903736e-01
-6.42869174e-01 5.17766416e-01 4.08286840e-01 -2.73938160e-02
-4.30149019e-01 -5.37261486e-01 -4.83019412e-01 -9.35897768e-01
-1.63483173e-01 -2.71819592e-01 2.47401014e-01 -4.47264433e-01
4.40078937e-02 7.32250810e-01 -1.84602571e+00 1.00023977e-01
4.54608738e-01 9.28490579e-01 3.05763483e-01 6.41214490e-01
2.52601415e-01 3.60482544e-01 -9.18246925e-01 -6.11701667e-01
2.63483137e-01 -4.18854803e-01 -5.03312349e-01 -5.25709331e-01
5.08343220e-01 -8.55653360e-02 -2.29245365e-01 1.20292091e+00
7.06877530e-01 2.75045514e-01 5.47945917e-01 -8.33403647e-01
-2.69720733e-01 7.59087861e-01 -1.89733222e-01 5.25788367e-01
1.84499666e-01 -3.55602354e-01 1.21384013e+00 -7.89049149e-01
4.27843511e-01 1.18888009e+00 9.50493217e-01 2.12447107e-01
-1.23694372e+00 -4.06562537e-01 -3.09072435e-01 1.35832995e-01
-1.40635324e+00 -3.05060387e-01 9.64042485e-01 -2.53412962e-01
1.00999522e+00 7.94638991e-02 7.97320724e-01 1.13787556e+00
5.13350070e-01 3.21041822e-01 9.29773152e-01 -7.31777191e-01
7.72266611e-02 1.82699472e-01 -4.66634296e-02 3.14797431e-01
1.44552812e-01 6.17642343e-01 -8.56636584e-01 8.90650302e-02
8.65292549e-01 -5.51409483e-01 -4.18493778e-01 -3.19053233e-02
-1.14555740e+00 6.61391318e-01 3.97101611e-01 8.56988549e-01
-4.11206186e-01 5.51824607e-02 5.09888947e-01 3.05890679e-01
2.54751801e-01 7.09050775e-01 -3.59096706e-01 -1.88212469e-01
-1.03130484e+00 1.18958421e-01 7.81916380e-01 2.82308757e-01
5.04131675e-01 5.56362450e-01 1.26465678e-01 5.52410901e-01
1.37403578e-01 2.73965418e-01 7.64154375e-01 -7.36309052e-01
4.32941988e-02 -1.61999226e-01 -3.94632161e-01 -1.08296204e+00
-7.89887726e-01 -7.93013096e-01 -6.77712262e-01 1.73257664e-01
4.86324519e-01 -1.54831186e-01 -2.48328969e-01 1.99015272e+00
7.86026865e-02 3.64363760e-01 3.53971541e-01 7.13815033e-01
6.95761502e-01 5.07489920e-01 -8.60874131e-02 -3.04379761e-01
1.57081926e+00 -4.13656592e-01 -1.00260043e+00 2.18591094e-01
1.60408318e-01 -7.38476038e-01 9.27390516e-01 5.97381592e-01
-1.11131871e+00 -1.00254071e+00 -1.40020835e+00 4.66875173e-02
-4.99006182e-01 1.37146294e-01 4.61457491e-01 9.91830051e-01
-1.02489436e+00 1.07029331e+00 -4.54133511e-01 -1.46664441e-01
-3.14322293e-01 2.93378174e-01 -2.60367602e-01 8.67485821e-01
-1.53449762e+00 1.06493020e+00 7.81179070e-01 -4.04227786e-02
-2.26743847e-01 -8.76214385e-01 -6.55967236e-01 5.25313199e-01
-2.30916575e-01 -3.22540045e-01 1.15158117e+00 -8.96570265e-01
-1.71682465e+00 4.91544545e-01 1.28959432e-01 -7.76989341e-01
3.11961602e-02 -3.32837611e-01 -8.52726698e-01 5.48549652e-01
-4.77802694e-01 5.85162401e-01 1.22371089e+00 -8.25619578e-01
-4.92230594e-01 -3.67844760e-01 -3.00095201e-01 1.09725334e-01
-3.73180538e-01 -1.55189753e-01 4.50737596e-01 -9.92564976e-01
2.15720445e-01 -5.04618526e-01 3.28013748e-01 -5.58457434e-01
-6.05018772e-02 -5.85964173e-02 5.37406802e-01 -6.42342687e-01
1.31165063e+00 -2.62486458e+00 1.02054449e-02 3.71059924e-01
2.07030755e-02 2.26084560e-01 -1.85165510e-01 4.88327354e-01
-6.22672677e-01 -9.77102369e-02 4.21722755e-02 -2.20015377e-01
9.50894356e-02 -1.00107543e-01 -6.15089059e-01 5.38400054e-01
3.86726588e-01 5.91103077e-01 -6.40281618e-01 7.69475773e-02
2.44488180e-01 9.61435258e-01 -3.71308088e-01 -3.54294211e-01
1.91753924e-01 1.16473496e-01 4.09424156e-01 -2.04223897e-02
6.60951197e-01 5.28284311e-01 2.68483430e-01 -6.46817386e-01
-4.26395446e-01 3.34710896e-01 -1.30809426e+00 1.46833682e+00
-2.69358784e-01 1.01456845e+00 -9.67413466e-03 -8.18067431e-01
1.26547003e+00 7.43335009e-01 3.32169652e-01 -8.84012938e-01
2.22565129e-01 3.85385960e-01 6.06515110e-01 -8.29144195e-02
7.20058322e-01 -4.49914902e-01 2.37765521e-01 5.40726557e-02
8.06319296e-01 6.67774528e-02 7.55853727e-02 -2.78827369e-01
7.11188376e-01 -2.37410627e-02 5.43946922e-01 -6.99203432e-01
5.29823065e-01 -7.84248888e-01 3.69955540e-01 4.98123735e-01
-2.81389922e-01 5.50029457e-01 5.05138159e-01 -2.58458406e-01
-9.77218330e-01 -1.02102959e+00 -3.57388347e-01 9.65119839e-01
-1.82429641e-01 -4.12094742e-01 -7.07227886e-01 1.78082407e-01
-1.64288320e-02 5.96329749e-01 -4.61049914e-01 -3.50813299e-01
-5.36985159e-01 -4.36737150e-01 9.89348233e-01 3.78862947e-01
3.23004246e-01 -1.06963563e+00 -1.09744418e+00 3.18166584e-01
1.84148015e-03 -1.25977349e+00 -1.37832627e-01 5.59865713e-01
-8.04250360e-01 -7.66288161e-01 -6.15765095e-01 -5.74943125e-01
-1.38485402e-01 -6.47461340e-02 8.88100505e-01 -3.38161439e-01
-1.69893876e-01 3.43600124e-01 -3.70306402e-01 -5.64571559e-01
-4.88418251e-01 6.96248468e-03 4.70107406e-01 1.22205503e-01
3.61468077e-01 -1.10480917e+00 -2.15750247e-01 -4.01268415e-02
-9.30408239e-01 -4.47743267e-01 6.16361201e-01 5.70063174e-01
2.16115624e-01 4.63057756e-01 8.26141357e-01 -1.50617674e-01
1.04763663e+00 -3.55752036e-02 -5.86834550e-01 -6.87589720e-02
-3.65296721e-01 2.05211580e-01 6.92845702e-01 -5.43407261e-01
-9.40079689e-01 -1.21994361e-01 -5.41558325e-01 -4.12876070e-01
-3.07249963e-01 4.10988241e-01 -1.28322378e-01 -1.98856935e-01
7.08485842e-01 1.83334664e-01 -1.47225469e-01 -6.15264893e-01
3.67533505e-01 2.57759273e-01 8.47656548e-01 -1.48404121e-01
7.32666254e-01 2.51389503e-01 2.71187037e-01 -1.31345069e+00
-4.09939110e-01 -4.33948547e-01 -7.38528013e-01 -9.15373266e-02
8.17810714e-01 -7.75563836e-01 -9.66757655e-01 5.34431040e-01
-1.47196436e+00 2.66732931e-01 -6.10910237e-01 1.00003254e+00
-6.91076577e-01 3.93837810e-01 -6.20342493e-01 -1.00544024e+00
-3.35364223e-01 -8.79889369e-01 6.44405961e-01 3.90260547e-01
-9.18430910e-02 -1.06116164e+00 2.35977009e-01 -3.75282049e-01
3.74944329e-01 1.05587862e-01 1.06711149e+00 -9.17965531e-01
5.41621819e-02 1.01264477e-01 1.82255834e-01 5.25936961e-01
-2.50520885e-01 1.46727294e-01 -1.45839286e+00 -1.21759154e-01
6.44655645e-01 1.34647250e-01 8.89010787e-01 5.75902224e-01
5.15448928e-01 -4.69100624e-02 3.65693808e-01 6.05006754e-01
1.53389370e+00 3.09472412e-01 7.30978191e-01 1.89489901e-01
5.90118133e-02 6.06110811e-01 7.32920915e-02 4.89918560e-01
-9.02708471e-02 6.39591575e-01 4.47887212e-01 4.22329828e-02
-2.29515821e-01 -2.62802929e-01 4.38489735e-01 1.10686278e+00
-2.50003785e-01 1.72702208e-01 -6.43260777e-01 3.27099770e-01
-1.27370000e+00 -1.23267972e+00 -8.19958597e-02 2.25221610e+00
5.65971553e-01 3.28180254e-01 3.22200835e-01 6.73233569e-01
6.67254925e-01 2.63331056e-01 -1.54102623e-01 -8.34655285e-01
-6.59452856e-01 8.05570126e-01 3.19143653e-01 4.85787243e-01
-1.08918297e+00 6.81898892e-01 6.75692749e+00 9.24571097e-01
-1.32179654e+00 -1.00119308e-01 -1.96329385e-01 2.97464997e-01
6.96863756e-02 -3.06955159e-01 -7.40369856e-01 2.01610550e-01
1.38786209e+00 -1.72908008e-01 5.13159811e-01 5.71356535e-01
5.48565760e-02 -2.04411998e-01 -1.08077133e+00 1.07473767e+00
9.33326110e-02 -1.25801909e+00 1.71190258e-02 1.05668932e-01
1.46890670e-01 -1.04560912e-01 2.71734059e-01 2.10550670e-02
-6.63805068e-01 -8.16198409e-01 1.07187331e+00 4.75788146e-01
4.45839196e-01 -1.04901934e+00 6.96125448e-01 2.22360522e-01
-1.52037084e+00 -3.07747602e-01 -3.51847917e-01 -1.91930994e-01
-3.85998562e-02 6.60338879e-01 -6.63502693e-01 7.47117579e-01
5.24343550e-01 3.80490214e-01 -3.16852480e-01 1.18762338e+00
-1.83019355e-01 3.35648268e-01 -2.15065107e-01 1.99588135e-01
2.95739155e-02 -2.05761969e-01 8.40562105e-01 1.33906806e+00
3.58258069e-01 -1.70250580e-01 -5.93649328e-01 1.00131929e+00
3.40230852e-01 7.22300783e-02 -4.35610116e-01 -1.41599402e-01
5.55176079e-01 1.24100876e+00 -8.35647404e-01 1.65839314e-01
-3.72627109e-01 6.51805937e-01 -1.15390271e-01 3.02596450e-01
-5.64256728e-01 -7.09728003e-01 7.32655525e-01 -2.42192447e-01
5.35552204e-01 -3.15994501e-01 -9.89030078e-02 -7.77786732e-01
1.49306015e-03 -7.67703891e-01 -1.96556449e-01 -6.39194965e-01
-1.01197362e+00 7.22513139e-01 -7.18982071e-02 -1.16623211e+00
-3.75883013e-01 -8.76538336e-01 -5.93945026e-01 1.05501473e+00
-1.47220731e+00 -6.82677269e-01 2.06153616e-01 6.67236090e-01
1.11958571e-01 -1.85910642e-01 1.01063371e+00 3.76794696e-01
-1.60516024e-01 5.53292811e-01 4.45809327e-02 -4.34676372e-02
4.24052447e-01 -1.32907259e+00 4.29327518e-01 8.07111800e-01
4.47322339e-01 6.56932950e-01 8.53615224e-01 -2.35629141e-01
-8.51415515e-01 -4.51975226e-01 8.74676704e-01 5.86687960e-02
7.00111091e-01 -4.95468050e-01 -9.64661181e-01 8.85262340e-03
3.67343366e-01 -4.06800151e-01 8.84963751e-01 9.17700231e-02
-3.88839930e-01 -1.44197822e-01 -9.46542144e-01 2.89763957e-01
5.64410388e-01 -9.38271344e-01 -1.08987772e+00 -1.56689212e-01
8.83506894e-01 -1.59418702e-01 -9.28865790e-01 3.14729631e-01
7.19276607e-01 -1.56569433e+00 9.92262781e-01 -2.64177501e-01
7.20893443e-02 -9.70324874e-02 -3.66173297e-01 -1.39987874e+00
-3.18210155e-01 -8.29625309e-01 -3.33125368e-02 9.71606672e-01
1.48205534e-01 -7.66325295e-01 3.01645637e-01 -2.06946209e-01
-2.98317343e-01 -4.75051314e-01 -1.27040839e+00 -8.90894055e-01
1.66117754e-02 -4.09351438e-01 3.64871204e-01 7.48431802e-01
1.32099897e-01 3.30718368e-01 -1.42658904e-01 2.31431127e-01
2.49759331e-01 -4.66214538e-01 7.05945939e-02 -1.84537780e+00
-3.51347834e-01 -5.64314485e-01 -6.54669940e-01 -5.68517089e-01
1.79981828e-01 -6.58240616e-01 -2.69805461e-01 -6.00928962e-01
-5.78373790e-01 2.55210131e-01 -4.46826726e-01 -1.64478093e-01
3.03731471e-01 4.18525748e-02 5.86779714e-01 -3.91667113e-02
1.65209740e-01 4.76811767e-01 7.88650274e-01 5.06176054e-01
-2.69685060e-01 4.49339226e-02 -3.04627180e-01 9.91229236e-01
6.71421766e-01 -2.15349928e-01 -4.35988992e-01 1.05392925e-01
2.83964038e-01 1.21824533e-01 4.84469682e-01 -1.52226949e+00
3.18306774e-01 6.04897857e-01 3.73932928e-01 -5.83237112e-01
7.41392314e-01 -8.27304244e-01 1.42545104e-01 5.76658666e-01
-9.43612680e-02 3.04241151e-01 5.02154589e-01 3.08601797e-01
-5.73726475e-01 -4.82810408e-01 9.96388078e-01 1.54659140e-03
-6.35706544e-01 -3.28801602e-01 -4.61134255e-01 -1.91519380e-01
6.46101475e-01 -5.28548539e-01 -4.35768992e-01 -1.66764468e-01
-9.19348061e-01 -6.90623820e-01 -1.02582000e-01 1.53733835e-01
5.23005962e-01 -1.22673869e+00 -3.82102579e-01 6.71157777e-01
-4.72937286e-01 -6.65976644e-01 2.75545776e-01 8.51585805e-01
-2.70530075e-01 7.29615510e-01 -7.41724491e-01 -2.36432716e-01
-1.17135954e+00 3.46338332e-01 7.83843994e-01 -1.55302390e-01
-5.61796904e-01 8.19666862e-01 1.00780070e-01 1.21036552e-01
4.37719405e-01 -6.27561986e-01 -3.81222546e-01 6.23191714e-01
5.23866892e-01 3.24230164e-01 3.99878532e-01 -8.26443195e-01
-2.71568298e-01 5.34097910e-01 2.91944385e-01 -4.40882891e-01
1.21036196e+00 7.54423216e-02 -9.37882662e-02 7.44144082e-01
1.12482107e+00 2.18156472e-01 -8.93612027e-01 -7.12506175e-02
1.32890731e-01 -9.46922824e-02 6.99655265e-02 -5.15953898e-01
-7.44978726e-01 1.18447495e+00 7.86324143e-01 6.59197986e-01
1.24731314e+00 -3.33680123e-01 2.89181948e-01 3.55037600e-01
3.14612202e-02 -1.26825976e+00 2.13376969e-01 7.01143444e-01
9.32826519e-01 -5.04492223e-01 -3.30573171e-01 -1.63976148e-01
-4.00949180e-01 1.64372742e+00 8.33102241e-02 -2.87671208e-01
7.95451880e-01 2.19644353e-01 -1.84075162e-01 1.75868154e-01
-4.66090173e-01 -3.91404122e-01 3.20607424e-01 8.68893623e-01
4.78047639e-01 -7.19809607e-02 -3.46138209e-01 6.81165993e-01
-8.05177510e-01 -5.11769533e-01 6.51373565e-01 3.33549380e-01
-4.59729254e-01 -8.60344529e-01 -5.11592031e-01 -1.35112107e-01
-5.59157670e-01 -2.71290898e-01 -1.91270679e-01 1.08958077e+00
2.66972452e-01 6.53414249e-01 2.64673710e-01 -5.20613968e-01
4.76412058e-01 3.59625697e-01 5.07787824e-01 -2.19928965e-01
-9.72012877e-01 3.54872793e-01 -2.06466079e-01 -3.53898883e-01
-5.85698009e-01 -5.32616556e-01 -1.06221163e+00 -8.73700436e-03
-4.05328631e-01 1.27796948e-01 9.60123181e-01 8.57666194e-01
1.48313269e-01 7.08912075e-01 3.72390509e-01 -9.42599475e-01
-7.59475470e-01 -8.94536376e-01 -1.20612097e+00 2.24899516e-01
7.83128321e-01 -5.98410785e-01 -5.13883412e-01 -2.11445436e-01] | [15.222196578979492, 5.518536567687988] |
ec27750f-7431-4625-9f6d-63100c8256fe | vitag-online-wifi-fine-time-measurements | null | null | https://www.winlab.rutgers.edu/~hansiiii/papers/ViTag_SECON2022_camera_ready_v6.pdf | https://www.winlab.rutgers.edu/~hansiiii/papers/ViTag_SECON2022_camera_ready_v6.pdf | ViTag: Online WiFi Fine Time Measurements Aided Vision-Motion Identity Association in Multi-person Environments | In this paper, we present ViTag to associate user identities across multimodal data, particularly those obtained from cameras and smartphones. ViTag associates a sequence of vision tracker generated bounding boxes with Inertial Measurement Unit (IMU) data and Wi-Fi Fine Time Measurements (FTM) from smartphones. We formulate the problem as association by sequence to sequence (seq2seq) translation. In this two-step process, our system first performs cross-modal translation using a multimodal LSTM encoder-decoder network (X-Translator) that translates one modality to another, e.g. reconstructing IMU and FTM readings purely from camera bounding boxes. Second, an association module finds identity matches between camera and phone domains, where the translated modality is then matched with the observed data from the same modality. In contrast to existing works, our proposed approach can associate identities in multi-person scenarios where all users may be performing the same activity. Extensive experiments in real-world indoor and outdoor environments demonstrate that online association on camera and phone data (IMU and FTM) achieves an average Identity Precision Accuracy (IDP) of 88.39% on a 1 to 3 seconds window, outperforming the state-of-the-art Vi-Fi (82.93%). Further study on modalities within the phone domain shows the FTM can improve association performance by 12.56% on average. Finally, results from our sensitivity experiments demonstrate the robustness of ViTag under different noise and environment variations. | ['Shubham Jain', 'Ashwin Ashok', 'Kristin Dana', 'Marco Gruteser', 'Nicholas Meegan', 'Hansi Liu', 'Abrar Alali', 'Bryan Bo Cao'] | 2022-09-20 | null | null | null | ieee-international-conference-on-sensing | ['multimodal-association'] | ['time-series'] | [ 5.00546753e-01 -2.84933299e-01 -1.53588196e-02 -2.70149767e-01
-9.86567318e-01 -7.30383694e-01 8.03484797e-01 -3.72057527e-01
-4.87909526e-01 8.92498791e-01 9.63507220e-02 5.85717149e-02
1.82940468e-01 -3.63729924e-01 -1.08275068e+00 -3.52238625e-01
1.51182443e-01 4.60635632e-01 -3.53407800e-01 2.32841194e-01
-1.65364906e-01 2.28419527e-02 -1.66015649e+00 1.84953645e-01
5.87379456e-01 1.18783224e+00 1.17224365e-01 1.04758835e+00
5.06604016e-02 5.28628051e-01 -7.29423642e-01 -7.71087348e-01
3.55106652e-01 -2.82149553e-01 -4.91757482e-01 1.63214337e-02
8.21445167e-01 -4.79000300e-01 -2.91958272e-01 8.33306551e-01
6.82607114e-01 1.13405325e-01 3.18381011e-01 -1.62629545e+00
-7.19359815e-01 1.25096649e-01 -4.33792859e-01 -2.85613418e-01
1.12656832e+00 -1.51504755e-01 5.29528141e-01 -1.03607798e+00
5.50947785e-01 9.91786480e-01 1.02867806e+00 5.60871303e-01
-9.87851620e-01 -6.82220042e-01 -1.82531804e-01 2.73124129e-01
-1.89737940e+00 -5.79547703e-01 1.60529509e-01 -4.01610404e-01
1.14054215e+00 4.56465662e-01 3.20953608e-01 1.54698884e+00
3.94958211e-03 5.65799713e-01 8.50377738e-01 -2.92149752e-01
-2.61612739e-02 2.50756621e-01 -2.08240911e-01 4.30194348e-01
8.66156220e-02 -6.17870167e-02 -1.04098809e+00 -2.67822117e-01
7.95164168e-01 -1.29407775e-02 -1.85593098e-01 1.76618062e-02
-1.85762846e+00 1.01905800e-01 -1.26810119e-01 7.63288215e-02
-4.77737218e-01 4.28842157e-01 3.61008346e-01 3.12935919e-01
1.75887957e-01 7.32352864e-03 -4.13854778e-01 -7.54144371e-01
-7.64782071e-01 -1.65236399e-01 7.82729626e-01 1.54112649e+00
6.28001273e-01 -1.43760204e-01 -9.37435776e-02 7.02161610e-01
3.50173980e-01 1.18683100e+00 4.57244635e-01 -9.12170231e-01
8.05549443e-01 6.47257864e-02 5.50072491e-01 -7.61961758e-01
-1.32560626e-01 -5.02450950e-02 -7.52193630e-01 -3.14764410e-01
5.81559718e-01 -5.25293112e-01 -5.40970266e-01 1.83957636e+00
3.90473604e-01 1.02724779e+00 1.77476723e-02 1.02918065e+00
6.70322597e-01 3.76329869e-01 -1.16421223e-01 -1.08233854e-01
1.55402029e+00 -5.80792069e-01 -8.45727742e-01 -1.65140435e-01
5.03088474e-01 -1.01173210e+00 7.69176006e-01 2.01743022e-01
-1.01167059e+00 -1.01052797e+00 -9.20462966e-01 3.00604254e-02
-4.14191514e-01 3.85959744e-01 1.77562565e-01 8.66831303e-01
-1.10385227e+00 1.52741268e-01 -7.06419528e-01 -8.69454622e-01
-1.60250142e-01 8.44017148e-01 -6.06564522e-01 9.80575159e-02
-1.29698360e+00 7.56528676e-01 1.16074599e-01 9.65358466e-02
-4.61134851e-01 -5.84662676e-01 -9.94630098e-01 -2.51668215e-01
1.65468544e-01 -1.05209851e+00 1.10686719e+00 -9.22276974e-01
-1.57226086e+00 7.34055698e-01 -8.20739269e-01 -4.65546697e-01
6.97378933e-01 -5.59174001e-01 -8.52819324e-01 -1.99750990e-01
2.65453815e-01 5.06776631e-01 9.73251045e-01 -1.12028360e+00
-1.02942526e+00 -1.75825015e-01 -3.04785501e-02 3.18968028e-01
-4.44893330e-01 -3.93831171e-03 -8.50567281e-01 -3.34117442e-01
-1.78938359e-02 -1.18271005e+00 5.43375015e-01 -3.42760533e-01
-4.09605891e-01 2.46567026e-01 5.80586493e-01 -9.45578754e-01
1.02407515e+00 -2.06815362e+00 8.74323696e-02 2.63676226e-01
1.42177893e-02 2.09677771e-01 3.14050587e-04 3.13369244e-01
1.35173500e-01 -2.02843338e-01 1.82549179e-01 -1.22795784e+00
3.67809325e-01 2.58072466e-01 -3.08440447e-01 6.13520801e-01
-2.96336740e-01 1.03111529e+00 -8.94037843e-01 -4.09327000e-01
5.53398311e-01 7.87703395e-01 -3.99422981e-02 3.00441176e-01
3.87082249e-01 7.33595073e-01 1.96409032e-01 8.23163688e-01
6.48472190e-01 -1.94758251e-01 2.79145330e-01 -3.00280720e-01
-3.10646109e-02 6.40829578e-02 -1.47665823e+00 1.96277821e+00
-7.56317198e-01 9.02297795e-01 -9.56316963e-02 -3.37172717e-01
6.90746665e-01 7.33867526e-01 4.26740855e-01 -6.56481445e-01
1.15099065e-01 1.87163204e-01 -4.96497303e-01 -6.45833492e-01
8.23119879e-01 3.76264006e-01 -3.45356345e-01 3.26647550e-01
6.83152303e-02 6.95065260e-01 -1.22514248e-01 -1.50233433e-01
8.45032156e-01 4.01114494e-01 1.48473144e-01 5.55284798e-01
7.23141193e-01 -4.44948792e-01 3.91913980e-01 1.09657943e+00
-1.98805451e-01 7.36300886e-01 -2.21390426e-01 -1.09872505e-01
-1.16769779e+00 -1.28988147e+00 8.39740708e-02 9.80134249e-01
2.40596846e-01 -3.61678809e-01 -6.98488116e-01 -3.98497373e-01
4.72698564e-04 4.39533830e-01 -6.63991749e-01 1.69578254e-01
-5.75455427e-01 -4.39657062e-01 9.64666784e-01 6.31512582e-01
8.54290187e-01 -3.86731386e-01 -2.06061110e-01 1.71772495e-01
-8.27772617e-01 -1.80799186e+00 -7.86149621e-01 -5.97361982e-01
-2.58281261e-01 -8.13543200e-01 -8.45354676e-01 -5.63665748e-01
5.38433671e-01 3.59059751e-01 8.08480620e-01 -3.92239243e-01
1.68280125e-01 8.65495563e-01 -2.63221949e-01 -9.36498493e-02
-9.04051885e-02 1.35238171e-01 7.27014363e-01 7.35948503e-01
7.62279034e-01 -2.81799495e-01 -3.51352990e-01 5.89446008e-01
-3.00233901e-01 5.82986884e-02 3.52369666e-01 7.26536572e-01
5.31336963e-01 -5.36322117e-01 3.77960771e-01 -2.50813186e-01
3.30315799e-01 -6.08061671e-01 -4.45125967e-01 4.53082830e-01
-3.90154123e-01 -4.66613829e-01 4.86032188e-01 -7.74576008e-01
-9.93816972e-01 2.58884102e-01 1.41631559e-01 -6.04920685e-01
-3.44214976e-01 7.89166987e-02 -3.80909890e-01 -8.49324539e-02
3.58449221e-01 5.69554329e-01 -1.76429316e-01 -1.93788171e-01
4.12409991e-01 1.20960033e+00 1.11396575e+00 -4.22568262e-01
7.52784550e-01 5.86667359e-01 -1.71636581e-01 -9.10968781e-01
-3.23423862e-01 -6.11556530e-01 -7.62735665e-01 -4.86464649e-01
9.25786495e-01 -1.20534575e+00 -1.35904872e+00 6.00761831e-01
-1.29790092e+00 -9.23116505e-02 2.01989070e-01 8.16501439e-01
-5.27004421e-01 5.54744124e-01 -4.85644549e-01 -1.06060350e+00
-2.77145028e-01 -9.30660546e-01 1.61487150e+00 2.91298538e-01
-5.48670292e-01 -9.41327870e-01 -1.17564604e-01 6.71881855e-01
1.74069121e-01 2.30141312e-01 -2.28081524e-01 -3.76065671e-01
-5.59515178e-01 -4.66897815e-01 -2.46206477e-01 -7.96096772e-03
3.16461772e-01 -4.41504836e-01 -1.28025293e+00 -3.17092508e-01
-2.07974076e-01 1.87382475e-01 1.85030490e-01 2.72749484e-01
4.39186692e-01 -2.07196668e-01 -4.71172631e-01 9.10683811e-01
1.19689727e+00 1.28310949e-01 6.73037887e-01 4.78490323e-01
1.09681547e+00 9.54973698e-02 8.07584405e-01 6.28526807e-01
8.07047784e-01 1.10232985e+00 1.86983332e-01 -9.60962623e-02
-3.64003219e-02 -2.27255970e-01 6.75528467e-01 5.44377983e-01
-1.78528354e-01 -3.14500570e-01 -6.80519283e-01 4.53465551e-01
-2.29932404e+00 -9.94876385e-01 -2.91540086e-01 2.74093270e+00
3.40311289e-01 -2.58817375e-01 3.05651844e-01 -8.12351704e-03
1.01911449e+00 -4.01692390e-01 -3.98807466e-01 -1.08904615e-01
-2.52969652e-01 -2.01998487e-01 9.01514053e-01 7.70331204e-01
-1.18311214e+00 7.87108779e-01 5.48829889e+00 4.02967095e-01
-9.16285455e-01 4.07386869e-01 1.77856863e-01 -3.74813110e-01
2.59457141e-01 -4.00128275e-01 -9.54959452e-01 8.52262616e-01
1.21827197e+00 2.22585097e-01 7.86888838e-01 3.49575192e-01
2.95109659e-01 -1.23461485e-01 -1.39408076e+00 1.73098779e+00
4.05158341e-01 -1.03377151e+00 -4.92390931e-01 2.32504651e-01
6.82202160e-01 1.27680182e-01 1.67168811e-01 1.92245781e-01
-7.63396695e-02 -7.85279453e-01 7.75982380e-01 9.20718133e-01
1.27002752e+00 -5.21379650e-01 1.04424489e+00 8.75538811e-02
-1.58847582e+00 9.43581834e-02 -1.54765785e-01 -8.23417678e-02
5.20979106e-01 2.97274977e-01 -1.24017465e+00 7.92848110e-01
6.87921464e-01 6.57231867e-01 -3.26902628e-01 7.20374644e-01
5.91879524e-02 3.44400406e-01 -4.61972982e-01 1.51016369e-01
-1.68328092e-01 -2.31134996e-01 5.30977726e-01 1.30968499e+00
8.21195781e-01 -2.16812491e-01 -3.37754488e-02 5.37291467e-01
-3.80734503e-02 -3.02233011e-01 -6.86528146e-01 3.58381450e-01
9.29706097e-01 1.06089115e+00 -9.31349546e-02 -6.85197771e-01
-6.68684006e-01 1.71637106e+00 -7.43367150e-02 6.42206907e-01
-1.49703288e+00 -1.85941249e-01 1.15292263e+00 -1.78342834e-01
1.68672040e-01 -4.69273001e-01 -3.21171939e-01 -1.30672550e+00
2.25652739e-01 -5.64280868e-01 2.60616302e-01 -1.06366861e+00
-1.05505896e+00 4.36562538e-01 -1.50228456e-01 -1.69542778e+00
-6.64003134e-01 -3.22532564e-01 -1.44351527e-01 1.13936055e+00
-1.02216351e+00 -1.58639717e+00 -5.33457637e-01 8.00203741e-01
1.37341529e-01 -3.01357657e-01 9.32243347e-01 8.19157779e-01
-6.66175008e-01 1.22269464e+00 4.39160556e-01 3.97183269e-01
1.01831174e+00 -1.20519519e+00 8.44567776e-01 9.38101709e-01
3.49714875e-01 8.70303333e-01 5.82668185e-01 -7.26989329e-01
-1.76177824e+00 -1.24295521e+00 1.28227389e+00 -8.95146430e-01
4.47843462e-01 -7.52044380e-01 -4.72670108e-01 9.31739509e-01
1.29806966e-01 -1.70230374e-01 9.37748075e-01 6.89289793e-02
-2.96109796e-01 -1.26997024e-01 -1.16522682e+00 6.61820471e-01
1.17925668e+00 -9.59368050e-01 -2.90051848e-01 1.59239933e-01
3.65788251e-01 -7.90266454e-01 -1.25015700e+00 -1.37415377e-03
1.09976602e+00 -6.87503040e-01 1.27936709e+00 -1.92103207e-01
-2.69970596e-01 -6.71083629e-01 -5.28866947e-01 -9.02015269e-01
-2.49083377e-02 -9.11521316e-01 -6.16714954e-01 1.49891520e+00
1.25847831e-01 -8.55886936e-01 5.86137593e-01 8.11290622e-01
1.98845074e-01 -4.92439419e-02 -1.14051282e+00 -1.02097368e+00
-7.33949602e-01 -8.68410289e-01 1.01782525e+00 8.79939437e-01
-1.16436603e-02 2.81368792e-01 -9.89520609e-01 6.82756901e-01
7.19857514e-01 -2.31426388e-01 1.31735814e+00 -7.22479224e-01
-3.91871065e-01 5.99393733e-02 -5.04371583e-01 -1.22224426e+00
6.71812743e-02 -5.54696739e-01 -1.38282627e-01 -1.29958081e+00
-1.42884076e-01 -2.15923823e-02 -1.00031354e-01 3.59333277e-01
-1.96871161e-01 7.85025656e-01 2.79968232e-01 2.80469328e-01
-7.48044491e-01 1.95935160e-01 4.85588878e-01 -1.28987044e-01
-1.70571670e-01 -1.29772663e-01 -2.00142890e-01 4.66740251e-01
6.38871253e-01 -9.08481851e-02 -2.26588100e-01 -6.62469387e-01
8.93425047e-02 9.33080241e-02 5.66645801e-01 -1.31105709e+00
4.31998014e-01 2.28505030e-01 7.17361748e-01 -4.18016762e-01
8.98489118e-01 -9.16825652e-01 6.20461285e-01 2.84373611e-01
1.06674217e-01 2.61050224e-01 2.41058275e-01 4.43815142e-01
2.15059564e-01 2.43568018e-01 1.05479017e-01 3.13488871e-01
-8.20253551e-01 1.14344746e-01 -3.94954056e-01 -3.73699009e-01
8.57249796e-01 -6.28557801e-01 -2.80791044e-01 -7.71587729e-01
-7.03534007e-01 1.33169338e-01 4.71088499e-01 6.74661040e-01
5.11680603e-01 -1.69199932e+00 -6.58773899e-01 2.61571318e-01
2.32608289e-01 -6.30856454e-01 1.51853964e-01 1.01657224e+00
-6.96163923e-02 5.21441400e-01 9.03154630e-03 -1.02332962e+00
-1.60369098e+00 2.94265240e-01 3.14448088e-01 2.51951158e-01
-1.39029771e-01 7.22479641e-01 -2.10445002e-01 -5.02666950e-01
2.70610064e-01 -2.08926246e-01 1.97926030e-01 2.87102652e-03
7.74765670e-01 6.45148098e-01 -2.85216197e-02 -1.20547569e+00
-6.25653625e-01 8.46046448e-01 2.93095291e-01 -4.90562886e-01
5.28597474e-01 -8.43801975e-01 1.70008436e-01 5.15616357e-01
1.31670249e+00 -1.00576781e-01 -1.14029610e+00 -3.52369905e-01
-2.96011716e-01 -5.89082301e-01 -5.73349774e-01 -7.32335150e-01
-5.63304543e-01 4.60877210e-01 9.89039779e-01 -1.08541138e-01
9.36040163e-01 -2.18493059e-01 1.09914839e+00 2.70047992e-01
6.43465817e-01 -1.15822017e+00 -3.73737276e-01 5.28809130e-01
3.70176673e-01 -1.35013020e+00 -3.32745343e-01 -2.18557581e-01
-5.94807088e-01 6.29656315e-01 5.53743958e-01 4.39006358e-01
1.78380251e-01 1.47010535e-01 2.21657932e-01 4.03437704e-01
-2.69628316e-01 -3.79702806e-01 3.72522920e-01 9.92749393e-01
3.32259178e-01 2.40466714e-01 1.12318024e-01 4.94643599e-01
-2.93935925e-01 6.69143200e-02 3.70621532e-01 6.36081457e-01
1.88674554e-01 -1.09839725e+00 -1.03933680e+00 2.22796619e-01
-2.46208742e-01 -1.83404628e-02 -2.97969460e-01 5.92721641e-01
5.24672329e-01 1.35797274e+00 3.11531961e-01 -9.96155500e-01
2.15343982e-01 1.93238273e-01 5.37726402e-01 -9.74305645e-02
-6.95016146e-01 7.16181323e-02 2.55754590e-01 -6.55639112e-01
-4.19530302e-01 -1.09596336e+00 -1.10867786e+00 -6.72718048e-01
-6.75401837e-02 -1.57022849e-01 8.16823542e-01 1.16623354e+00
6.78206265e-01 4.26135212e-01 4.31580782e-01 -9.30428684e-01
-2.23796815e-02 -1.01213002e+00 -1.85338140e-01 5.02803206e-01
4.75861937e-01 -5.71497738e-01 1.16403878e-01 4.36987132e-01] | [14.53492259979248, 1.0206551551818848] |
f5b948a5-213c-4974-a9ef-cba40599a621 | specializing-joint-representations-for-the | 1706.07625 | null | http://arxiv.org/abs/1706.07625v2 | http://arxiv.org/pdf/1706.07625v2.pdf | Specializing Joint Representations for the task of Product Recommendation | We propose a unified product embedded representation that is optimized for
the task of retrieval-based product recommendation. To this end, we introduce a
new way to fuse modality-specific product embeddings into a joint product
embedding, in order to leverage both product content information, such as
textual descriptions and images, and product collaborative filtering signal. By
introducing the fusion step at the very end of our architecture, we are able to
train each modality separately, allowing us to keep a modular architecture that
is preferable in real-world recommendation deployments. We analyze our
performance on normal and hard recommendation setups such as cold-start and
cross-category recommendations and achieve good performance on a large product
shopping dataset. | ['Vasile Flavian', 'Smirnova Elena', 'Nedelec Thomas'] | 2017-07-18 | null | null | null | null | ['product-recommendation'] | ['miscellaneous'] | [-3.88430990e-02 -2.62155503e-01 -2.57419765e-01 -5.74247181e-01
-5.60291767e-01 -8.59074712e-01 5.64603388e-01 1.89635128e-01
-2.49736100e-01 -9.18689072e-02 4.02300328e-01 -2.53158063e-01
-1.99893340e-01 -6.92034483e-01 -4.98644948e-01 -4.25646514e-01
2.05155507e-01 4.19044763e-01 1.06971823e-01 -5.09181321e-01
-7.46438187e-03 2.37311020e-01 -1.58600783e+00 5.25755286e-01
5.15998065e-01 1.18479061e+00 2.67198652e-01 5.05379915e-01
-5.61224632e-02 4.04684216e-01 1.15227193e-01 -1.01325417e+00
5.38629591e-01 9.61219706e-03 -4.34884787e-01 2.82262057e-01
6.00166142e-01 -4.78710175e-01 -4.31957573e-01 8.45643699e-01
3.91673952e-01 4.80628401e-01 7.91952729e-01 -8.72101724e-01
-1.23453033e+00 7.26503849e-01 -4.76618439e-01 -1.50290653e-01
3.27893585e-01 2.10961580e-01 1.57579088e+00 -1.09158123e+00
6.86812580e-01 1.18722141e+00 3.48711908e-01 3.60025764e-01
-1.46528828e+00 -1.93563908e-01 5.75600028e-01 6.36688694e-02
-1.19595563e+00 -3.63507777e-01 6.91154122e-01 -4.42290902e-01
8.75904799e-01 -2.30273549e-02 3.94796342e-01 1.15938687e+00
3.87785956e-02 1.00013471e+00 6.04135811e-01 -7.68894330e-02
4.33100984e-02 5.52156031e-01 3.87945145e-01 3.66676241e-01
1.13799371e-01 8.11417177e-02 -3.73714447e-01 3.16815227e-02
5.67805290e-01 6.70465589e-01 -3.02377671e-01 -8.04045677e-01
-1.16713476e+00 1.28188825e+00 5.34539819e-01 2.54814357e-01
-3.67336988e-01 5.95693849e-02 3.01001728e-01 5.44019282e-01
5.92468500e-01 4.75974321e-01 -5.99516809e-01 3.21497291e-01
-7.17421293e-01 4.54688460e-01 9.51717317e-01 8.56697738e-01
5.92654645e-01 -3.17555666e-01 -2.26524517e-01 9.73673046e-01
7.78152704e-01 4.51833069e-01 3.28667194e-01 -5.25250077e-01
2.75745571e-01 2.33512357e-01 2.50470400e-01 -8.36497843e-01
-3.08483332e-01 -8.03603649e-01 -4.19524968e-01 -1.30917095e-02
2.00798094e-01 4.08637486e-02 -9.66047227e-01 1.57038951e+00
1.45182341e-01 1.05678156e-01 7.55123869e-02 1.21977746e+00
8.26619446e-01 6.24390304e-01 -8.96618739e-02 7.16900006e-02
1.64768898e+00 -1.48262036e+00 -6.51122630e-01 -2.04919763e-02
7.59854496e-01 -8.77319694e-01 1.14208198e+00 5.09688318e-01
-9.78495479e-01 -7.25649297e-01 -1.33356357e+00 -2.83050060e-01
-7.74045289e-01 1.03472151e-01 8.86437118e-01 5.38433790e-01
-7.67404556e-01 5.83362639e-01 -5.49043715e-01 -4.24035937e-01
1.44809019e-02 1.76489845e-01 -4.96110767e-01 -6.53204441e-01
-1.05777681e+00 8.71676266e-01 2.31361687e-01 1.89653307e-01
-7.68256187e-01 -7.19101667e-01 -1.03377962e+00 3.08930993e-01
4.02094930e-01 -1.04518509e+00 1.16138780e+00 -7.37699687e-01
-1.63117301e+00 3.99377197e-01 2.46116742e-01 -2.44641513e-01
-3.49771716e-02 -3.49619806e-01 -7.18164325e-01 -1.50035948e-01
-3.11664313e-01 5.41742384e-01 1.04514241e+00 -1.22237194e+00
-5.56260705e-01 -4.76968855e-01 4.11720037e-01 1.70687497e-01
-5.72641909e-01 -6.22164495e-02 -8.48228931e-01 -7.56825149e-01
-6.66471869e-02 -1.14100289e+00 -2.84789413e-01 -1.15927331e-01
1.18703200e-02 -1.47286117e-01 4.62061524e-01 -5.34656763e-01
8.63410354e-01 -2.63176703e+00 3.60591501e-01 3.58326405e-01
9.90685225e-02 2.38626730e-02 -7.42620170e-01 5.29878020e-01
3.09929326e-02 -1.16417192e-01 2.93131620e-01 -8.99074256e-01
4.98019844e-01 2.69631922e-01 -4.09883469e-01 3.77104729e-01
1.14377260e-01 1.11475563e+00 -7.75952101e-01 4.01321705e-03
3.04006189e-01 6.42111063e-01 -8.25207472e-01 1.72879517e-01
-4.46970344e-01 2.06330910e-01 -4.28954810e-01 5.25647223e-01
7.68668115e-01 -2.86486536e-01 4.28705923e-02 -5.02938092e-01
3.23196918e-01 4.27800626e-01 -1.37443781e+00 2.21365833e+00
-8.52661908e-01 4.57632951e-02 1.07056715e-01 -6.54455423e-01
7.74974644e-01 2.48611555e-01 3.96919638e-01 -7.72411823e-01
1.72076896e-01 8.10335204e-03 -1.45868853e-01 -1.69257909e-01
1.03441226e+00 -2.73089996e-03 -6.28355220e-02 7.34954059e-01
6.61310852e-01 6.85333312e-02 7.20851421e-02 3.55070740e-01
9.20215368e-01 1.50760457e-01 -3.77335966e-01 1.03511708e-02
3.45802516e-01 -3.86833400e-01 -8.14749580e-03 7.35585988e-01
3.36677283e-01 8.15149009e-01 9.67922658e-02 -3.90298218e-01
-8.32518518e-01 -1.22725773e+00 -1.35186706e-02 1.70044994e+00
2.30973616e-01 -8.68236482e-01 2.40031779e-02 -1.03643668e+00
2.12973610e-01 5.49798965e-01 -6.59932673e-01 -3.63836467e-01
-1.50028318e-01 -3.76001269e-01 -4.15892988e-01 5.79403460e-01
-1.34167522e-01 -5.77685714e-01 2.26967484e-01 1.49539709e-01
2.58920938e-01 -1.03205252e+00 -7.86730707e-01 3.35184544e-01
-7.65741050e-01 -5.46651304e-01 -8.96955431e-01 -6.37681961e-01
3.30156922e-01 5.27192593e-01 1.21564579e+00 -1.47988141e-01
1.08568467e-01 7.00015426e-01 -7.83568263e-01 2.04495285e-02
4.40064594e-02 2.09393293e-01 -1.06168166e-01 3.87317002e-01
3.33471805e-01 -4.27843958e-01 -7.92938888e-01 2.99796045e-01
-1.07897186e+00 -3.24082732e-01 6.76913857e-01 1.01937139e+00
4.11836118e-01 -3.56942452e-02 2.30545476e-01 -8.25170815e-01
7.58023679e-01 -4.53147143e-01 -5.94860911e-01 3.49346191e-01
-7.34276056e-01 1.23988897e-01 4.79232907e-01 -5.69410026e-01
-9.53931808e-01 1.27826825e-01 -4.30188298e-01 -5.91197491e-01
-8.46415162e-02 7.58650064e-01 -3.31684388e-02 1.34931475e-01
4.72542703e-01 -6.53310195e-02 -1.27519518e-01 -1.13371897e+00
1.23645854e+00 6.29143536e-01 3.84457976e-01 -1.69914886e-01
6.56083703e-01 1.84559494e-01 -3.73181522e-01 -1.95167437e-01
-8.63442600e-01 -8.47069919e-01 -4.38637376e-01 2.28848800e-01
7.16498196e-01 -1.07666755e+00 -6.62594974e-01 -1.35980085e-01
-8.89674246e-01 6.99331164e-02 -4.30493832e-01 8.15608263e-01
-2.68831283e-01 3.44305962e-01 -9.12620604e-01 -4.41442907e-01
-4.22375709e-01 -1.22297871e+00 1.23581207e+00 6.87773600e-02
1.10122047e-01 -1.03788960e+00 2.89308637e-01 2.75655001e-01
5.64086139e-01 -4.96732622e-01 7.07133651e-01 -1.02599871e+00
-3.60993475e-01 -1.97986379e-01 -2.50146508e-01 4.68105376e-01
-7.99255222e-02 -2.33589917e-01 -9.46089149e-01 -5.02969146e-01
-2.05683216e-01 -3.11713338e-01 1.44479740e+00 1.49380267e-02
7.11597443e-01 7.77039751e-02 -2.64346361e-01 5.50566435e-01
1.40763521e+00 -2.67820299e-01 5.63852787e-01 -1.83803998e-02
7.37585783e-01 4.71137404e-01 7.85279334e-01 3.44954491e-01
5.09677529e-01 1.07855892e+00 4.08326149e-01 -1.35420099e-01
-1.23232178e-01 -4.07900929e-01 4.57611650e-01 6.50504887e-01
3.07568014e-01 -3.98492783e-01 -1.65788755e-01 4.65083301e-01
-2.12240911e+00 -7.09131598e-01 2.84684330e-01 2.25376821e+00
3.06222767e-01 2.67689861e-02 2.39891723e-01 -3.88492316e-01
1.67506710e-01 1.35114178e-01 -3.89180273e-01 -5.27804375e-01
1.12954170e-01 4.28943068e-01 4.83611017e-01 3.37440550e-01
-1.26311660e+00 9.19730783e-01 6.65501881e+00 6.92534566e-01
-1.08610260e+00 4.57828134e-01 2.46520117e-02 -2.33422592e-01
-8.40529680e-01 -1.77400410e-01 -6.78983033e-01 3.53652716e-01
9.21620369e-01 4.38069731e-01 4.48757917e-01 9.29512024e-01
-2.70812273e-01 2.35010445e-01 -1.38259053e+00 1.10093427e+00
2.59744525e-01 -1.12524796e+00 -9.00143944e-03 1.92380652e-01
5.91450572e-01 2.23461300e-01 5.39963067e-01 6.41313612e-01
5.47865689e-01 -8.00281167e-01 5.41686296e-01 2.90569365e-01
2.45939001e-01 -4.92569238e-01 9.23948228e-01 -7.77411088e-02
-1.04876089e+00 -1.53947100e-01 -4.83272374e-01 2.19550997e-01
4.97576505e-01 5.96379280e-01 -3.39303553e-01 8.01827610e-01
5.47167122e-01 1.26553118e+00 -5.30610681e-01 1.05764925e+00
-9.49218124e-03 1.78224653e-01 -3.12956005e-01 2.95685649e-01
1.52717888e-01 -4.30194706e-01 9.28190649e-02 1.13929617e+00
3.03392768e-01 -2.29743123e-01 2.90321797e-01 7.74413824e-01
-1.14202783e-01 3.33972514e-01 -5.12551486e-01 -3.53719145e-01
-1.57872930e-01 1.70922136e+00 -4.78760332e-01 -1.11630894e-01
-9.22666490e-01 1.24780631e+00 3.02825809e-01 4.26565260e-01
-8.70756984e-01 -8.58240202e-02 9.32495773e-01 -3.06566685e-01
1.06659663e+00 -3.92031282e-01 6.76687956e-02 -1.45983267e+00
-4.24043201e-02 -6.03092849e-01 4.83827561e-01 -5.98352134e-01
-1.68089223e+00 3.94759864e-01 -2.60106146e-01 -1.33194566e+00
-1.78421736e-01 -7.94121444e-01 -2.43364915e-01 8.96839321e-01
-1.64957678e+00 -1.45955408e+00 1.25999928e-01 8.43566477e-01
3.35096657e-01 -5.11700027e-02 7.87416935e-01 7.47960687e-01
-4.39916581e-01 9.08705473e-01 2.03693092e-01 -2.20082164e-01
1.01563334e+00 -1.14054143e+00 2.72310764e-01 4.77603316e-01
6.89432859e-01 1.03321505e+00 6.10952020e-01 -1.38426244e-01
-2.00404906e+00 -8.65985453e-01 4.87287551e-01 -4.77007985e-01
7.17084646e-01 -6.51490152e-01 -6.12317979e-01 7.54034817e-01
2.31280088e-01 4.93773967e-02 1.05738914e+00 7.70815134e-01
-8.93356323e-01 -2.49377757e-01 -1.02802086e+00 5.12488604e-01
8.41922402e-01 -6.70018196e-01 -5.38946331e-01 4.45671111e-01
9.77573335e-01 -5.41271083e-02 -1.36863840e+00 3.60851288e-01
7.88747489e-01 -5.41994870e-01 1.11267662e+00 -5.75697184e-01
1.84369683e-01 -5.03842175e-01 -3.51608783e-01 -1.44834208e+00
-7.89511502e-01 -2.65912056e-01 -2.76816487e-01 1.15832365e+00
8.66489232e-01 -5.63231528e-01 4.50193763e-01 5.37151814e-01
-1.72903165e-01 -5.62645733e-01 -3.14603478e-01 -5.53488493e-01
-1.94928795e-01 -4.24939543e-01 6.96028352e-01 6.75787091e-01
1.61488265e-01 7.15731800e-01 -6.44878507e-01 1.72549620e-01
1.82687074e-01 5.42035639e-01 6.47094727e-01 -1.08039284e+00
-8.74842048e-01 -3.48548710e-01 -3.55795056e-01 -1.54721761e+00
4.46874350e-02 -1.00413573e+00 -4.00467925e-02 -1.56639898e+00
2.35420197e-01 -3.43035877e-01 -8.98633897e-01 3.92789185e-01
1.15783826e-01 7.30629206e-01 4.92107779e-01 1.02593161e-01
-8.02669287e-01 6.03841662e-01 1.20125544e+00 -3.84065658e-01
-2.61427701e-01 -1.89901873e-01 -9.72151399e-01 1.32548317e-01
3.29315603e-01 -1.57569498e-01 -5.70230782e-01 -5.75079143e-01
4.84554619e-01 -3.17458123e-01 -1.10046323e-02 -5.03704786e-01
-1.34185085e-03 1.88973621e-01 3.12345028e-01 -5.33904374e-01
6.73690081e-01 -1.15433848e+00 1.80773497e-01 -1.72196016e-01
-4.79204088e-01 -2.34398440e-01 -9.10011753e-02 6.78027391e-01
-2.14310467e-01 -2.80636996e-01 2.88976729e-01 4.51921038e-02
-7.86777973e-01 3.52907807e-01 -4.50715795e-02 -7.13616848e-01
7.78998137e-01 1.44515634e-01 -2.73283005e-01 -4.79500860e-01
-1.16741252e+00 3.19408506e-01 5.75067282e-01 1.02700603e+00
5.69512427e-01 -1.40838385e+00 -4.46807206e-01 2.17185140e-01
7.92699397e-01 -7.49675393e-01 3.71936351e-01 9.96759832e-01
7.80700147e-02 4.56461042e-01 2.45755408e-02 -4.22887117e-01
-8.22281301e-01 1.29958415e+00 6.02590516e-02 -3.36833090e-01
-5.45824349e-01 7.91943967e-01 2.49889538e-01 -3.98694307e-01
2.64924586e-01 -2.93497562e-01 -4.27381366e-01 1.40055582e-01
7.22913086e-01 -2.81228423e-01 2.92558521e-01 -5.96298695e-01
-6.14164881e-02 5.22915244e-01 -5.53564548e-01 -1.87042400e-01
1.13894272e+00 -4.68051642e-01 2.66089946e-01 3.11212152e-01
1.41122723e+00 -3.26213688e-02 -8.90621006e-01 -4.69444603e-01
-2.81134039e-01 -5.92089295e-01 4.86667126e-01 -8.81102741e-01
-1.43835783e+00 5.21437466e-01 9.39278662e-01 1.84180543e-01
9.42429125e-01 3.50853354e-01 9.38822627e-01 3.90228659e-01
3.32471877e-01 -8.96284878e-01 -3.61405127e-02 2.80364156e-01
6.93421185e-01 -1.41305768e+00 8.63682199e-03 -3.57584804e-01
-7.59883523e-01 9.14014161e-01 2.11067289e-01 -1.84858918e-01
1.05985165e+00 -9.62818116e-02 -8.95226151e-02 -2.90877640e-01
-8.38267386e-01 -6.64635003e-01 7.04327703e-01 4.00053084e-01
5.38431704e-01 8.47925097e-02 -3.63084674e-01 8.54048252e-01
3.06689799e-01 -1.36821819e-02 2.05397769e-03 7.20404863e-01
-1.55472279e-01 -1.49589980e+00 5.94903976e-02 4.61671084e-01
-2.70612150e-01 -1.39743328e-01 -9.80811343e-02 2.83089966e-01
1.35257587e-01 1.10615766e+00 1.31484061e-01 -6.61092281e-01
4.33614999e-01 6.93007465e-03 5.79070687e-01 -8.00972998e-01
-7.61554599e-01 2.88208246e-01 1.98568344e-01 -8.05403769e-01
-2.33717054e-01 -3.28082860e-01 -4.73718166e-01 -1.44523516e-01
-6.11813784e-01 6.04930967e-02 9.14251864e-01 7.88019657e-01
6.52077138e-01 3.01813275e-01 5.80630600e-01 -1.03135109e+00
-5.00653148e-01 -9.28411484e-01 -9.49630737e-01 7.37391293e-01
2.64080882e-01 -8.95196915e-01 -1.80318907e-01 -1.62021220e-01] | [10.114974975585938, 5.715418815612793] |
f8832991-014d-4481-9b7a-86204c4b5863 | comae-a-multi-factor-hierarchical-framework | 2105.08316 | null | https://arxiv.org/abs/2105.08316v3 | https://arxiv.org/pdf/2105.08316v3.pdf | CoMAE: A Multi-factor Hierarchical Framework for Empathetic Response Generation | The capacity of empathy is crucial to the success of open-domain dialog systems. Due to its nature of multi-dimensionality, there are various factors that relate to empathy expression, such as communication mechanism, dialog act and emotion. However, existing methods for empathetic response generation usually either consider only one empathy factor or ignore the hierarchical relationships between different factors, leading to a weak ability of empathy modeling. In this paper, we propose a multi-factor hierarchical framework, CoMAE, for empathetic response generation, which models the above three key factors of empathy expression in a hierarchical way. We show experimentally that our CoMAE-based model can generate more empathetic responses than previous methods. We also highlight the importance of hierarchical modeling of different factors through both the empirical analysis on a real-life corpus and the extensive experiments. Our codes and used data are available at https://github.com/chujiezheng/CoMAE. | ['Minlie Huang', 'Yongcai Leng', 'Wei Chen', 'Yong liu', 'Chujie Zheng'] | 2021-05-18 | null | https://aclanthology.org/2021.findings-acl.72 | https://aclanthology.org/2021.findings-acl.72.pdf | findings-acl-2021-8 | ['empathetic-response-generation', 'open-domain-dialog'] | ['natural-language-processing', 'natural-language-processing'] | [-6.78010225e-01 -5.37738483e-03 -7.92870224e-02 -4.21500325e-01
-1.54015750e-01 -3.43372375e-01 4.17943865e-01 -2.33857259e-01
-2.00455070e-01 8.12787294e-01 7.93670952e-01 5.77837527e-02
-2.25805834e-01 -5.67766726e-01 2.92667061e-01 -3.80551815e-01
4.01917815e-01 4.39768195e-01 -2.22350955e-01 -6.91129029e-01
3.92854184e-01 2.62805939e-01 -1.03938210e+00 3.14964205e-01
9.59331751e-01 4.11077619e-01 -1.76356256e-01 6.30883873e-01
-2.49444976e-01 1.38147175e+00 -7.93722987e-01 -8.51568401e-01
-6.30869195e-02 -8.10641885e-01 -1.25485826e+00 -2.51171470e-01
-6.37402773e-01 -7.36247420e-01 -3.25190634e-01 7.34258831e-01
5.36951363e-01 1.73374549e-01 7.72823215e-01 -1.63611388e+00
-8.92200589e-01 8.43774140e-01 -3.48512352e-01 -2.51171023e-01
7.55283356e-01 3.28180864e-02 1.00403702e+00 -6.44470572e-01
3.01757246e-01 1.42007732e+00 4.59943652e-01 1.05591643e+00
-7.73094296e-01 -7.04434991e-01 -1.74610242e-01 3.00146490e-01
-8.49448979e-01 -2.56900758e-01 7.92562604e-01 -5.40407121e-01
5.45915186e-01 2.32822120e-01 6.12179041e-01 1.33685756e+00
4.88562025e-02 8.97294760e-01 1.40511978e+00 -2.71636635e-01
1.47708967e-01 2.79192954e-01 4.52831089e-01 2.11730912e-01
-2.57053047e-01 8.13243315e-02 -6.18010819e-01 -5.74861705e-01
8.77534568e-01 -1.00275226e-01 -1.36848345e-01 1.91293001e-01
-9.56162035e-01 1.40886557e+00 4.07963693e-01 3.88347536e-01
-5.97728133e-01 -1.02985471e-01 2.74339169e-01 4.54658240e-01
2.43578449e-01 7.30836332e-01 -1.54490516e-01 -5.97689867e-01
-2.63749540e-01 3.36891055e-01 1.28310466e+00 7.76581287e-01
4.04483467e-01 -2.65174866e-01 -2.13494822e-01 1.18906629e+00
2.51551151e-01 2.97284946e-02 4.98062223e-01 -1.32221770e+00
-1.32371741e-03 7.62793720e-01 3.46170008e-01 -9.82789695e-01
-5.70777535e-01 1.77705079e-01 -6.04014993e-01 -4.90300655e-02
4.45970386e-01 -6.27499819e-01 2.25909308e-01 1.91397059e+00
4.74275202e-01 -4.22172904e-01 3.52695554e-01 1.19239116e+00
1.15368509e+00 3.51635158e-01 3.36246341e-01 -1.06336147e-01
1.35836267e+00 -1.23728561e+00 -8.08319867e-01 -5.06658293e-02
7.97633946e-01 -9.34757888e-01 1.15407431e+00 3.43861252e-01
-1.11142492e+00 -1.35269180e-01 -3.60181034e-01 -1.13754824e-01
7.28661940e-02 1.01472229e-01 1.08154869e+00 4.17918295e-01
-7.31323421e-01 3.64452779e-01 -7.45187476e-02 -5.48478246e-01
-1.17049195e-01 3.08138669e-01 -2.69595712e-01 2.04809517e-01
-1.72477877e+00 1.13573360e+00 -1.08342864e-01 -1.95992738e-01
-2.34222841e-02 -3.55518043e-01 -2.81948209e-01 1.30167499e-01
9.60185938e-03 -8.65158975e-01 1.59409869e+00 -9.91066396e-01
-2.01736069e+00 5.84108293e-01 7.83535168e-02 -1.47739332e-02
5.60021758e-01 -4.46334034e-01 -7.44982436e-02 3.52241755e-01
-2.42281526e-01 7.93967187e-01 4.29454774e-01 -1.12255156e+00
-1.80875212e-01 -3.50911953e-02 4.43628877e-01 3.06997299e-01
-6.74893200e-01 7.38331497e-01 7.75929466e-02 -4.66863424e-01
-4.43967640e-01 -1.04626131e+00 -3.84776086e-01 -1.64559647e-01
-1.18817732e-01 -6.38432860e-01 3.44378620e-01 -3.99246603e-01
1.35638678e+00 -1.98446882e+00 2.77098298e-01 -2.66890436e-01
5.72484732e-01 1.24957688e-01 -1.89012080e-01 1.04502010e+00
-8.28559138e-03 -2.92987730e-02 1.59090191e-01 -2.70951062e-01
3.85836869e-01 4.89201434e-02 -3.38962287e-01 3.30339409e-02
-4.62495796e-02 9.49362874e-01 -7.76119053e-01 -7.07319260e-01
6.81896210e-02 3.20958734e-01 -6.65945709e-01 6.13840401e-01
1.78903669e-01 7.15720534e-01 -8.17219973e-01 2.89787889e-01
5.41998208e-01 -3.05185527e-01 -3.26702604e-03 2.84572691e-01
-2.48798858e-02 8.65788758e-02 -6.06453478e-01 1.23213124e+00
-3.52822095e-01 2.66034901e-01 1.13666475e-01 -2.84031272e-01
1.16009581e+00 5.66334367e-01 6.42632425e-01 -4.23339933e-01
6.35784864e-01 2.20670402e-02 3.12070966e-01 -4.68342394e-01
7.22642124e-01 -6.09290421e-01 -5.33447802e-01 1.07157421e+00
-1.65422171e-01 -2.47495443e-01 1.70043483e-02 5.66235185e-01
9.96694446e-01 -1.49201080e-01 3.65423232e-01 -1.10275358e-01
5.12949109e-01 -2.55741142e-02 4.73536015e-01 3.62965256e-01
-7.90467083e-01 1.93255395e-01 1.03844333e+00 -1.80549338e-01
-7.85616815e-01 -5.60567319e-01 -2.74520498e-02 1.25084114e+00
2.06990421e-01 -4.59599286e-01 -9.18225408e-01 -1.82159409e-01
-1.48964748e-01 8.48053277e-01 -5.89827240e-01 -3.04798782e-01
-2.82428950e-01 -5.46861351e-01 8.53007436e-01 3.30456704e-01
2.15930566e-01 -1.33511066e+00 -6.72553897e-01 7.90947080e-02
-6.32951021e-01 -1.15503085e+00 -1.96596935e-01 -3.42490554e-01
-6.02740228e-01 -8.85203719e-01 -6.28184795e-01 -3.45858335e-01
2.94351697e-01 3.11662495e-01 9.23712194e-01 4.45213675e-01
7.13489056e-02 4.14037406e-01 -7.91099668e-01 -2.32529685e-01
-6.53700292e-01 -2.00465515e-01 5.81958443e-02 -2.66157538e-01
7.78720379e-01 -5.54133296e-01 -6.46168947e-01 6.43371999e-01
-6.20846331e-01 3.17808717e-01 3.87362421e-01 9.21577334e-01
-4.54089046e-01 -4.65391040e-01 1.01230621e+00 -7.53993154e-01
1.54976153e+00 -7.80657709e-01 9.58453640e-02 -5.11522032e-02
-2.38625765e-01 -2.59031564e-01 5.81322670e-01 -6.67649806e-01
-1.24785972e+00 -2.53790587e-01 -2.26843551e-01 -2.73280293e-01
-2.62819171e-01 3.47934276e-01 1.46029323e-01 -7.71647319e-02
6.16192698e-01 -3.91874492e-01 1.14356436e-01 -2.41388887e-01
4.88179862e-01 9.85989571e-01 1.68549642e-01 -1.09913063e+00
4.01197582e-01 1.33105129e-01 -3.04210812e-01 -5.88849008e-01
-7.78680861e-01 -3.40583354e-01 -5.32611370e-01 -5.69228828e-01
6.91499412e-01 -8.78714025e-01 -1.23323309e+00 4.61659968e-01
-1.41666627e+00 -3.24390352e-01 1.02670856e-01 6.72495723e-01
-6.96713030e-01 5.96588671e-01 -1.33592474e+00 -9.86615717e-01
-4.53821003e-01 -8.34492624e-01 6.24305308e-01 4.95729536e-01
-1.15437245e+00 -9.51758504e-01 2.32728183e-01 9.09008265e-01
4.66353863e-01 -1.04536809e-01 9.47531164e-01 -6.07524216e-01
1.39243126e-01 -1.75007939e-01 -1.75813019e-01 1.08578116e-01
-2.12020829e-01 9.82180685e-02 -7.68191814e-01 2.75959820e-01
4.27257359e-01 -8.48974764e-01 9.65613350e-02 -9.05855224e-02
5.66940486e-01 -3.32640111e-01 1.28970310e-01 1.14854530e-01
8.90138209e-01 -1.02178585e-02 7.20065117e-01 1.38624728e-01
4.77789581e-01 1.31780756e+00 9.24568474e-01 1.10142076e+00
6.22421682e-01 5.35882175e-01 1.80818215e-01 2.01317053e-02
5.37783623e-01 -1.73405945e-01 4.04408962e-01 1.17472100e+00
-1.35823429e-01 -1.94810942e-01 -7.03872979e-01 3.25723082e-01
-2.08185077e+00 -1.03982699e+00 -6.45529330e-01 1.66061103e+00
8.28659713e-01 -4.94510829e-01 4.62890774e-01 -2.32582852e-01
6.97582006e-01 -1.12441041e-01 -3.26868862e-01 -1.03471243e+00
-1.08659975e-01 -1.68816913e-02 -2.14951739e-01 6.12523317e-01
-5.30915439e-01 1.26906717e+00 6.36334276e+00 4.80533391e-01
-7.30810702e-01 1.12296127e-01 2.39613160e-01 -2.15261012e-01
-3.91281337e-01 2.02543482e-01 -3.87833685e-01 4.39536572e-01
7.85705566e-01 -5.22704422e-01 4.04873341e-01 7.88346946e-01
3.80112171e-01 -1.81820720e-01 -9.52950537e-01 9.29971099e-01
-1.17703281e-01 -5.27655959e-01 -2.15718001e-01 7.49952719e-02
4.51564014e-01 -5.12003899e-01 -2.65944421e-01 4.77120042e-01
8.26574624e-01 -1.04395044e+00 3.90288442e-01 4.67646003e-01
1.32077098e-01 -6.93765700e-01 7.49427676e-01 7.06806421e-01
-4.94884461e-01 -2.02579305e-01 -3.43255401e-01 -6.35635138e-01
3.97333503e-01 1.04960598e-01 -3.52998823e-01 3.05768669e-01
2.91065782e-01 4.72457796e-01 -7.03973994e-02 6.13322079e-01
-6.56953514e-01 3.77586931e-01 1.26639835e-03 -4.67162848e-01
4.07002494e-02 -5.18025577e-01 4.10074532e-01 8.49467516e-01
2.03646198e-01 7.95719862e-01 -2.47961029e-01 1.00196660e+00
2.15348408e-01 4.87321794e-01 -3.29103947e-01 -2.54947752e-01
6.91033721e-01 1.50500917e+00 -6.71104044e-02 -1.25112519e-01
-3.42459679e-01 1.03939390e+00 5.62910795e-01 1.43731982e-01
-1.16867256e+00 -6.35782480e-02 7.83620536e-01 -3.62427771e-01
-1.96969241e-01 -1.00191586e-01 -3.74260783e-01 -1.19091618e+00
-3.09147239e-01 -1.09923947e+00 3.22943091e-01 -9.90615308e-01
-1.79013121e+00 5.94884634e-01 -9.28498209e-02 -1.12898076e+00
-4.24256414e-01 -4.68264431e-01 -8.82338107e-01 9.66061890e-01
-9.14224029e-01 -1.08297873e+00 -4.60994810e-01 7.09237635e-01
2.08075017e-01 2.87246108e-02 1.18444085e+00 1.79662660e-01
-6.86754048e-01 4.92402315e-01 -2.66184062e-01 -1.67871583e-02
1.12651515e+00 -8.35420370e-01 -6.51105046e-02 1.90229729e-01
-4.99262154e-01 9.40534472e-01 7.49666393e-01 -3.40502679e-01
-1.01189709e+00 -2.46649951e-01 1.19457519e+00 -5.96364260e-01
1.08577359e+00 -1.19770087e-01 -8.97042394e-01 3.93926650e-01
5.13297319e-01 -8.18904877e-01 1.39204180e+00 3.17831546e-01
-2.38206252e-01 6.11512840e-01 -1.06500220e+00 9.29277658e-01
7.95069218e-01 -3.89153928e-01 -8.34936917e-01 2.22938702e-01
6.15018129e-01 -1.03577383e-01 -1.23608863e+00 -2.23898161e-02
6.58976078e-01 -1.39628541e+00 6.18023753e-01 -8.83029342e-01
1.11685193e+00 2.81469792e-01 6.04554825e-02 -1.32891393e+00
-3.94877017e-01 -8.31222117e-01 -4.83899973e-02 1.29477119e+00
-7.86794052e-02 -7.70171463e-01 6.76466882e-01 1.25866032e+00
2.92649925e-01 -8.46423566e-01 -6.11632288e-01 -3.32275182e-01
6.50462806e-01 -1.07231878e-01 8.25290918e-01 1.24380481e+00
8.92420411e-01 9.33496118e-01 -9.04509842e-01 -4.15685773e-01
2.55064726e-01 2.47149616e-01 1.25300825e+00 -1.32432282e+00
-4.90410566e-01 -7.54662931e-01 -1.95778981e-02 -1.21715498e+00
4.79116619e-01 -2.81781971e-01 -1.59102127e-01 -1.46958983e+00
4.90743577e-01 -4.50708836e-01 9.29972753e-02 2.89694101e-01
-4.21321779e-01 -2.64832377e-01 4.08556312e-01 5.58577478e-01
-4.16106552e-01 9.40721989e-01 1.47105956e+00 6.27882481e-01
-9.55566466e-02 -1.14211276e-01 -1.18539011e+00 7.80542195e-01
1.12468719e+00 -5.73691785e-01 -2.76632726e-01 -2.38441810e-01
3.28185894e-02 6.69391274e-01 3.22630167e-01 -4.07815695e-01
3.08756053e-01 -6.44976258e-01 -1.00381084e-01 -4.86057550e-02
7.47060955e-01 -5.00618160e-01 1.84700917e-02 3.59856606e-01
-4.95266825e-01 4.56747636e-02 -1.40856162e-01 5.65479919e-02
-3.51923317e-01 -4.81005281e-01 7.28910446e-01 -3.40976715e-02
-3.48465204e-01 -5.98662198e-02 -5.27968109e-01 -1.54867359e-02
1.02688360e+00 -1.05800286e-01 -6.10148966e-01 -1.11758125e+00
-4.28081483e-01 4.11934406e-01 6.15589023e-01 6.46268427e-01
4.21162158e-01 -1.31107724e+00 -7.31141150e-01 -4.08129394e-01
3.69632691e-02 -6.62128448e-01 6.30334079e-01 1.04848301e+00
-3.25722188e-01 2.89963514e-01 -6.47688925e-01 7.47845545e-02
-1.42116368e+00 4.21399534e-01 1.90336481e-01 -2.67372727e-01
-2.46521682e-01 8.07696581e-01 5.08344233e-01 -6.97984278e-01
-2.05622660e-03 4.37872529e-01 -6.10490143e-01 3.31223011e-02
3.71905416e-01 4.81171608e-01 -7.97542334e-01 -9.74589586e-01
-9.61961225e-02 3.84704828e-01 2.73141712e-02 -2.78537542e-01
1.03246522e+00 -1.72691345e-01 -4.81298625e-01 3.82642388e-01
7.81969488e-01 1.48677200e-01 -6.94296658e-01 -1.04214571e-01
-1.69443503e-01 -6.61853671e-01 -4.87837315e-01 -5.35390556e-01
-5.89532256e-01 1.02380717e+00 -2.21864074e-01 2.10143566e-01
9.35558617e-01 -6.18164204e-02 1.03337908e+00 2.29129717e-01
4.12852108e-01 -1.19196808e+00 4.68828857e-01 7.66365647e-01
9.72652733e-01 -1.07183933e+00 -4.45979200e-02 -5.25485873e-01
-1.43319690e+00 1.10718024e+00 1.08777082e+00 -4.63227481e-02
3.34176272e-01 1.36739075e-01 6.11938894e-01 -2.21398592e-01
-1.09963810e+00 1.63558140e-01 -2.55752087e-01 2.90722817e-01
8.74852657e-01 3.01404834e-01 -8.94547284e-01 1.26217127e+00
-6.43888652e-01 1.17930524e-01 7.20834196e-01 7.13878214e-01
-2.27002650e-01 -1.47421741e+00 -3.70317459e-01 8.89127478e-02
-3.71820182e-01 4.74439226e-02 -1.29098177e+00 7.10639536e-01
-4.16510224e-01 1.43683410e+00 -3.14120293e-01 -6.03700876e-01
2.57992804e-01 -4.18837275e-03 3.17246109e-01 -3.37951303e-01
-1.14076507e+00 -1.76962242e-01 3.29671264e-01 -3.58803004e-01
-3.58519465e-01 -5.83837688e-01 -1.46517062e+00 -1.10600221e+00
-3.49212885e-01 5.29470384e-01 2.53555924e-01 8.99284899e-01
3.05071771e-01 1.13734603e-01 8.43798280e-01 -3.74686033e-01
-8.56946528e-01 -1.19154298e+00 -6.25235617e-01 7.63585567e-01
-4.10539001e-01 -7.48098254e-01 -5.59425175e-01 -4.53740746e-01] | [13.198698043823242, 7.582991123199463] |
e58600bf-94cc-40e5-9fcc-8ce35e74dc29 | comparing-a-composite-model-versus-chained | 2306.01551 | null | https://arxiv.org/abs/2306.01551v1 | https://arxiv.org/pdf/2306.01551v1.pdf | Comparing a composite model versus chained models to locate a nearest visual object | Extracting information from geographic images and text is crucial for autonomous vehicles to determine in advance the best cell stations to connect to along their future path. Multiple artificial neural network models can address this challenge; however, there is no definitive guidance on the selection of an appropriate model for such use cases. Therefore, we experimented two architectures to solve such a task: a first architecture with chained models where each model in the chain addresses a sub-task of the task; and a second architecture with a single model that addresses the whole task. Our results showed that these two architectures achieved the same level performance with a root mean square error (RMSE) of 0.055 and 0.056; The findings further revealed that when the task can be decomposed into sub-tasks, the chain architecture exhibits a twelve-fold increase in training speed compared to the composite model. Nevertheless, the composite model significantly alleviates the burden of data labeling. | ['Tayeb Lemlouma', 'Fanny Parzysz', 'Xavier Marjou', 'Antoine Le Borgne'] | 2023-06-02 | null | null | null | null | ['autonomous-vehicles'] | ['computer-vision'] | [-9.46299806e-02 2.13156849e-01 -3.31532687e-01 -4.42601830e-01
-6.21134818e-01 -4.27970469e-01 5.58696568e-01 -7.71459788e-02
-4.74873364e-01 8.29303145e-01 4.22927067e-02 -9.81348753e-01
-1.70910150e-01 -6.20811880e-01 -4.89272863e-01 -6.81479752e-01
6.19394258e-02 4.25446719e-01 -1.87662542e-02 1.75325554e-02
2.73793578e-01 5.68566203e-01 -1.25404847e+00 -6.41413266e-03
9.63530540e-01 1.06097639e+00 2.35287517e-01 3.73050541e-01
-2.18774512e-01 9.69746828e-01 -6.56344950e-01 -2.69080698e-01
3.05088520e-01 4.02860381e-02 -6.18106186e-01 1.30070761e-01
2.44324788e-01 -5.20784616e-01 -2.45022625e-01 5.39893746e-01
2.71100700e-01 -8.05925205e-02 6.37042999e-01 -1.41887784e+00
-2.32484013e-01 3.71590197e-01 -5.02088130e-01 2.74754703e-01
-2.43388727e-01 -2.13161167e-02 8.58221591e-01 -7.63140917e-01
2.62214184e-01 8.26799929e-01 7.27808356e-01 2.48250633e-01
-1.19079208e+00 -8.91299546e-01 4.52524871e-01 2.10395344e-02
-1.41525102e+00 -7.29566395e-01 4.05214429e-01 -4.81953472e-01
1.01261270e+00 1.11082494e-01 3.86608124e-01 5.60699165e-01
2.22914472e-01 3.68855268e-01 9.42008972e-01 -1.90408975e-01
2.31160745e-01 5.05589724e-01 1.29814938e-01 4.60612208e-01
5.38188696e-01 -2.10312337e-01 -8.28176588e-02 4.58624661e-02
5.80764890e-01 -2.41194710e-01 1.48844957e-01 -8.94587301e-03
-9.20401037e-01 7.42777884e-01 4.16316181e-01 5.74120879e-01
-4.81636703e-01 1.56168222e-01 3.06338128e-02 3.12302321e-01
2.75345057e-01 5.75212538e-01 -4.81618196e-01 1.50522232e-01
-1.00539410e+00 1.32676601e-01 7.94515312e-01 8.75940382e-01
7.41978884e-01 3.24833214e-01 9.68580022e-02 6.02181137e-01
5.55587113e-01 3.00159872e-01 -5.70188928e-03 -1.13220263e+00
6.79636717e-01 5.55061281e-01 3.33341837e-01 -8.03221822e-01
-9.54135180e-01 -8.58554900e-01 -6.19947851e-01 7.65035003e-02
7.35681832e-01 -7.09468901e-01 -9.61173177e-01 1.83345711e+00
1.04352079e-01 4.47551347e-02 -1.02002891e-02 7.25820184e-01
6.62829518e-01 7.16814756e-01 4.26672757e-01 1.81340352e-02
1.27087843e+00 -1.04452384e+00 -7.15363622e-01 -7.74099529e-01
8.12580407e-01 -5.54280818e-01 6.17916167e-01 -8.81463382e-03
-1.05847847e+00 -4.42329973e-01 -1.38510621e+00 7.20061734e-02
-5.03967941e-01 3.65461379e-01 6.85787261e-01 7.69743502e-01
-1.37290323e+00 3.92501801e-02 -4.56437379e-01 -4.39958870e-01
2.08103508e-01 4.29327488e-01 -1.74037606e-01 7.31666833e-02
-1.05388010e+00 1.28245950e+00 3.25741172e-01 2.96950936e-01
-6.67251527e-01 -4.74513978e-01 -6.30293608e-01 2.61292279e-01
1.33506566e-01 -5.21800876e-01 1.10778165e+00 -1.02178991e+00
-1.07956994e+00 5.66265166e-01 -2.55494267e-01 -4.84838694e-01
2.30103612e-01 2.60643661e-01 -5.79430223e-01 -1.76138073e-01
2.05724940e-01 8.71446013e-01 5.80833435e-01 -1.36601162e+00
-1.15266538e+00 -2.68223464e-01 1.05580002e-01 2.78887600e-01
-2.90136814e-01 4.05212268e-02 -7.58501768e-01 -3.86889547e-01
1.75828129e-01 -1.02207494e+00 -3.16561490e-01 -9.21467692e-02
-2.06996143e-01 -1.49983481e-01 6.41444564e-01 -7.13228285e-01
1.41515303e+00 -2.16479468e+00 -3.62956226e-01 3.52203459e-01
3.38546336e-01 2.02108994e-01 -2.58545905e-01 3.75970811e-01
1.60259996e-02 4.16610181e-01 1.60839766e-01 -2.59622902e-01
-5.21880500e-02 7.21664056e-02 6.23244643e-02 2.42523089e-01
7.82772005e-02 5.15661955e-01 -4.86533374e-01 -3.03466141e-01
5.64203598e-04 2.54697859e-01 -2.03861967e-01 -8.09279531e-02
1.10290997e-01 2.61637717e-01 -4.77495879e-01 5.91315627e-01
6.82961345e-01 -5.06432474e-01 4.87453789e-01 3.31347249e-02
-1.97219178e-01 3.77008408e-01 -1.01494741e+00 1.06743431e+00
-3.70799333e-01 1.03697455e+00 4.28704679e-01 -9.26924050e-01
8.51284921e-01 3.24970931e-01 6.67061985e-01 -9.41689432e-01
1.40437379e-01 2.11194307e-01 2.78909624e-01 -3.51424843e-01
5.90963006e-01 2.20509581e-02 1.50011731e-02 4.71609652e-01
-3.31725776e-01 4.18535024e-01 2.85722733e-01 3.06861345e-02
1.01296782e+00 -3.22658539e-01 -1.40866917e-02 -2.49322116e-01
3.00668269e-01 2.02689737e-01 5.99043906e-01 8.68298650e-01
-4.16137934e-01 1.74761936e-01 5.33772469e-01 -5.82703292e-01
-8.27160478e-01 -7.03856111e-01 -1.47597715e-01 9.06554878e-01
3.22170854e-02 -1.37484089e-01 -7.95231342e-01 -6.87094510e-01
3.50432396e-02 1.05061936e+00 -4.08027798e-01 2.50682503e-01
-4.03798044e-01 -8.03429127e-01 4.98189390e-01 5.64209461e-01
6.01218343e-01 -7.07800448e-01 -5.19101381e-01 1.09655641e-01
-3.74843299e-01 -1.32568896e+00 -1.13053262e-01 1.84510186e-01
-7.43091285e-01 -8.53609204e-01 -5.50161898e-01 -8.21945369e-01
7.94541776e-01 5.71360707e-01 9.56450701e-01 3.27991903e-01
5.67665100e-01 -6.51673377e-02 1.84806716e-02 -5.32239854e-01
-2.26053193e-01 3.82102191e-01 -1.80269763e-01 -6.83915988e-02
4.78271008e-01 -2.11505160e-01 -4.33077186e-01 5.70006073e-01
-5.99479377e-01 3.65255803e-01 9.14731383e-01 3.92416596e-01
3.19509298e-01 1.99488088e-01 1.00676501e+00 -6.18690193e-01
5.90053737e-01 -8.13466966e-01 -6.08182728e-01 1.59051314e-01
-8.70698631e-01 -1.61971420e-01 3.62903923e-01 -2.32746974e-02
-1.04170048e+00 1.16525151e-01 -2.01423436e-01 8.38566497e-02
-3.11158270e-01 8.74421120e-01 1.01557754e-01 1.50897149e-02
2.39642933e-01 1.97125487e-02 1.38552472e-01 -2.65835077e-01
1.09285703e-02 7.76638925e-01 2.22325996e-01 -2.62843639e-01
6.61973000e-01 3.63856226e-01 -7.87618607e-02 -8.58127773e-01
-6.40205801e-01 -2.06299573e-01 -4.75895792e-01 -4.02003109e-01
9.07262266e-01 -1.14454639e+00 -4.73499835e-01 2.80413538e-01
-9.66144323e-01 -5.62304258e-01 3.40266079e-01 6.34180486e-01
-1.91493943e-01 -8.30599631e-04 -3.79782319e-01 -5.64469934e-01
-9.34131816e-02 -1.22735739e+00 5.24686098e-01 3.66439402e-01
-2.44579136e-01 -9.36648726e-01 -3.39419454e-01 6.24292195e-01
7.47971237e-01 -2.18822751e-02 1.21635568e+00 -8.00525188e-01
-6.50167942e-01 -3.71093452e-01 -4.98003572e-01 -4.18681093e-02
3.60830836e-02 -6.48453012e-02 -7.87050307e-01 -2.81527728e-01
-7.84625672e-03 5.38385883e-02 5.38930476e-01 6.59014046e-01
8.65395546e-01 -2.67052293e-01 -4.61969584e-01 3.72655571e-01
1.39006042e+00 5.99697351e-01 6.95044994e-01 6.50065482e-01
3.90145689e-01 8.53807986e-01 4.13930804e-01 -8.31206366e-02
9.88063335e-01 7.12141335e-01 4.09949392e-01 -3.58148694e-01
-1.06177377e-02 3.93564105e-02 5.40090539e-02 4.41803634e-01
1.98597670e-01 -4.15282071e-01 -1.25640988e+00 5.51345170e-01
-1.72201836e+00 -8.18739653e-01 -2.57862598e-01 2.09618211e+00
1.22332714e-01 4.47172612e-01 1.84290826e-01 2.05824785e-02
4.84700233e-01 1.04165435e-01 -5.91084421e-01 -3.12006474e-01
2.47167319e-01 -6.89927995e-01 7.10171878e-01 2.84875780e-01
-9.27298844e-01 8.05041850e-01 7.64021730e+00 6.92815363e-01
-1.15330124e+00 -2.79664304e-02 9.70928907e-01 -7.80255646e-02
-8.55551809e-02 1.73304025e-02 -9.29724157e-01 6.49404347e-01
1.40439105e+00 1.66701928e-01 3.05611074e-01 5.81338704e-01
6.06739044e-01 -3.64744067e-01 -8.51956129e-01 6.65375471e-01
-1.05036028e-01 -1.31587720e+00 -1.93271264e-02 3.54715049e-01
6.78138852e-01 3.21876079e-01 -3.34269032e-02 4.01009351e-01
2.67716706e-01 -1.15050316e+00 7.93960631e-01 3.90966445e-01
4.86093223e-01 -6.08860433e-01 9.16806817e-01 5.12361109e-01
-1.16562486e+00 -3.67421448e-01 -7.38421036e-03 -3.17634284e-01
2.59049207e-01 1.38103977e-01 -7.38644540e-01 2.73064643e-01
5.83751500e-01 2.00159028e-01 -6.97397947e-01 1.17884076e+00
1.02587916e-01 6.63523495e-01 -1.50161505e-01 1.53401777e-01
6.00129902e-01 -3.23082268e-01 2.40580857e-01 1.00570560e+00
4.63027954e-01 -7.43185431e-02 1.27637133e-01 4.95966882e-01
7.85487890e-02 -2.45179191e-01 -6.03808641e-01 8.95525562e-04
9.50762808e-01 1.32550192e+00 -8.93105447e-01 -3.46863121e-01
-8.14535737e-01 1.59172297e-01 2.96863645e-01 7.56080627e-01
-1.00657296e+00 -3.55480283e-01 6.17440403e-01 1.86642185e-01
2.00250953e-01 -3.41783643e-01 -9.30651963e-01 -4.83266056e-01
-6.93664178e-02 -8.03186953e-01 2.11458817e-01 -9.55024540e-01
-7.36040175e-01 6.94916904e-01 -1.51695877e-01 -1.11352766e+00
-1.62028581e-01 -5.21079540e-01 -5.83565176e-01 9.39304769e-01
-1.61971498e+00 -1.13460016e+00 -4.69693452e-01 2.24418685e-01
4.99737591e-01 -1.40404090e-01 6.31290853e-01 5.72777510e-01
-1.06962669e+00 5.25707722e-01 1.57080948e-01 1.96451679e-01
3.91949952e-01 -9.09813821e-01 1.87031597e-01 8.99667323e-01
-2.66615927e-01 5.48691332e-01 4.52549547e-01 -4.71611917e-01
-1.26191711e+00 -1.17543805e+00 1.35405540e+00 -1.96117565e-01
3.43008399e-01 8.43142942e-02 -5.99210203e-01 8.87350202e-01
3.05274367e-01 -5.37708700e-01 8.35164845e-01 1.82082504e-01
3.94906178e-02 -3.65933567e-01 -1.02375126e+00 6.39798403e-01
5.18204391e-01 -2.86028773e-01 -7.18638077e-02 5.29606454e-02
5.00168979e-01 -2.77880371e-01 -8.70600283e-01 1.90912798e-01
6.97658956e-01 -9.24823105e-01 7.19755948e-01 -2.79105157e-01
1.66407600e-01 -4.41754848e-01 -8.89108926e-02 -1.01397502e+00
-5.63199341e-01 -1.91418260e-01 1.23397158e-02 1.12534416e+00
1.09196877e+00 -9.25274789e-01 8.76604021e-01 1.06354082e+00
-3.82367551e-01 -8.01196814e-01 -6.88264251e-01 -6.72484159e-01
9.81597602e-02 -5.44178963e-01 7.03247070e-01 8.28419805e-01
-3.60268503e-01 4.63342994e-01 -1.55750215e-01 3.97672623e-01
4.08532411e-01 -2.28144750e-01 7.90004015e-01 -1.32126510e+00
3.48125935e-01 -7.20521390e-01 -1.18539231e-02 -1.20723104e+00
1.05393797e-01 -6.13330543e-01 -1.49798155e-01 -1.97191501e+00
-6.71585323e-03 -9.33523238e-01 -3.06999058e-01 5.12399554e-01
4.82672974e-02 3.68849300e-02 2.20703423e-01 4.05547291e-01
-3.40374976e-01 4.00706753e-02 7.06168711e-01 -8.43174681e-02
-2.93856323e-01 2.20770597e-01 -1.38874638e+00 8.01491439e-01
1.13397872e+00 -4.49605256e-01 -6.20508194e-01 -9.68226969e-01
1.08742349e-01 2.35561818e-01 1.35556683e-01 -1.19986761e+00
6.38025224e-01 -3.23350161e-01 4.60259527e-01 -8.72862399e-01
3.65120053e-01 -1.02245986e+00 3.85956675e-01 2.70277828e-01
-1.67827472e-01 2.21520498e-01 5.35385847e-01 4.47768420e-01
-1.35211840e-01 -2.03079790e-01 5.27298033e-01 2.77295709e-01
-7.51648784e-01 -5.15312329e-02 -9.84527826e-01 -4.13698435e-01
1.14214993e+00 -5.91865301e-01 -3.70553672e-01 -5.21458745e-01
-5.33291578e-01 6.94696248e-01 2.93875724e-01 2.67620414e-01
2.47975558e-01 -1.21170390e+00 -5.86469889e-01 1.24697134e-01
-2.48277456e-01 -2.48007104e-01 8.71287808e-02 1.13834143e+00
-5.04943609e-01 9.46509421e-01 -1.44223541e-01 -3.18244487e-01
-1.04812431e+00 2.93722749e-01 4.72160012e-01 -4.10389185e-01
-1.12456791e-01 4.38131630e-01 7.49011040e-02 -3.16165775e-01
5.05648375e-01 -2.30771363e-01 -6.09190881e-01 1.93904594e-01
4.48563606e-01 5.49794734e-01 1.89898312e-01 -9.13696945e-01
-2.77576476e-01 3.37340027e-01 -7.24409968e-02 -8.02808329e-02
1.06158543e+00 -5.94504833e-01 4.08264026e-02 1.58944845e-01
8.59615743e-01 -3.03153574e-01 -1.37142718e+00 -1.21492103e-01
1.88056737e-01 -5.72358556e-02 4.32608545e-01 -1.02099013e+00
-1.16865051e+00 4.20670807e-01 3.27701539e-01 4.40176547e-01
1.13024724e+00 -3.69337767e-01 4.33024615e-01 5.37208617e-01
3.17063957e-01 -1.19526517e+00 -4.70839322e-01 6.15303040e-01
6.42618239e-01 -1.14666760e+00 6.36033574e-03 -2.30480865e-01
-7.50810325e-01 8.69312048e-01 7.37013936e-01 3.64334583e-01
7.37006903e-01 1.23288013e-01 1.91083446e-01 -3.56048822e-01
-8.36916685e-01 -3.01612228e-01 1.84373409e-01 5.34984708e-01
2.87518352e-01 3.58473063e-02 -3.06713432e-01 5.10657847e-01
-3.24351378e-02 -8.32898319e-02 3.52325439e-01 7.99604058e-01
-5.84527135e-01 -7.08681881e-01 -3.79589289e-01 7.88413048e-01
-2.96620786e-01 4.68778796e-02 -3.66644919e-01 1.05951321e+00
6.53016046e-02 1.58281291e+00 3.93947810e-01 -4.93902236e-01
1.69051111e-01 1.53723508e-01 -1.57271057e-01 -3.21245849e-01
-5.04889607e-01 -1.65085774e-02 5.16908467e-01 -4.16363418e-01
-2.85849214e-01 -4.06488508e-01 -1.20878005e+00 -5.90954900e-01
-3.48667711e-01 2.12051734e-01 9.67888355e-01 1.05102289e+00
6.91749811e-01 6.21072054e-01 6.11956060e-01 -8.60018313e-01
-3.05313408e-01 -9.12588000e-01 -4.00240690e-01 -2.24672332e-01
2.81665802e-01 -5.24261892e-01 -9.88549963e-02 -1.27361521e-01] | [6.5575971603393555, 1.9241260290145874] |
e5113f1b-c83d-433f-8255-434ea1f58dd6 | an-empirical-study-on-neural-keyphrase | 2009.10229 | null | https://arxiv.org/abs/2009.10229v3 | https://arxiv.org/pdf/2009.10229v3.pdf | An Empirical Study on Neural Keyphrase Generation | Recent years have seen a flourishing of neural keyphrase generation (KPG) works, including the release of several large-scale datasets and a host of new models to tackle them. Model performance on KPG tasks has increased significantly with evolving deep learning research. However, there lacks a comprehensive comparison among different model designs, and a thorough investigation on related factors that may affect a KPG system's generalization performance. In this empirical study, we aim to fill this gap by providing extensive experimental results and analyzing the most crucial factors impacting the generalizability of KPG models. We hope this study can help clarify some of the uncertainties surrounding the KPG task and facilitate future research on this topic. | ['Adam Trischler', 'Tong Wang', 'Xingdi Yuan', 'Sanqiang Zhao', 'Rui Meng', 'Daqing He'] | 2020-09-22 | null | https://aclanthology.org/2021.naacl-main.396 | https://aclanthology.org/2021.naacl-main.396.pdf | naacl-2021-4 | ['keyphrase-generation'] | ['natural-language-processing'] | [-1.75923601e-01 -2.20394596e-01 -4.60435778e-01 -6.55751005e-02
-5.35383403e-01 -3.64254981e-01 9.02144551e-01 5.87393828e-02
-5.22784472e-01 7.84806609e-01 4.82427329e-01 -6.53830290e-01
-3.14513743e-01 -8.09701443e-01 -7.32562304e-01 -5.28051972e-01
-1.19351700e-01 -1.67748276e-02 7.84050301e-03 -2.32320771e-01
4.76296216e-01 3.07453543e-01 -1.28022206e+00 3.60107213e-01
5.86977363e-01 1.00461900e+00 3.16908546e-02 5.19894004e-01
6.57108873e-02 9.49719191e-01 -9.79239166e-01 -5.77426374e-01
2.16682941e-01 -9.02583972e-02 -8.03218722e-01 -6.13301218e-01
5.23810863e-01 -6.05353177e-01 -5.03096938e-01 7.51297653e-01
7.49672592e-01 6.45661727e-02 4.67194021e-01 -1.63740182e+00
-1.04992998e+00 1.13335884e+00 -2.40555584e-01 6.22072279e-01
-1.79712981e-01 1.28365740e-01 1.34028184e+00 -7.43458331e-01
4.69814003e-01 1.08178473e+00 7.10857928e-01 4.61115807e-01
-8.61387432e-01 -1.19573498e+00 2.67786205e-01 4.57175404e-01
-1.46228981e+00 -2.65386581e-01 8.56285214e-01 -1.81836590e-01
1.19524336e+00 1.58494920e-01 8.29829276e-01 1.50101256e+00
3.48023623e-01 9.35300052e-01 1.21393847e+00 -4.58673537e-01
3.09083853e-02 -1.90173890e-02 3.73927444e-01 3.71537387e-01
7.10004449e-01 3.20794582e-01 -9.02630389e-01 -2.08436593e-01
7.75229275e-01 -3.46734703e-01 -3.65138888e-01 3.93635854e-02
-1.16661954e+00 9.82574701e-01 4.89443690e-01 4.50487465e-01
-4.30203259e-01 4.95317042e-01 3.30078512e-01 4.70537961e-01
5.63454568e-01 1.08764410e+00 -6.05293751e-01 -4.31064725e-01
-1.03557968e+00 7.08627999e-01 9.04131591e-01 7.04231501e-01
4.63512927e-01 1.95851699e-01 -6.43233359e-02 5.48599184e-01
3.66737917e-02 1.07597731e-01 6.33276641e-01 -6.81671262e-01
5.80799103e-01 3.98349911e-01 5.95208665e-04 -1.22896850e+00
-3.89003277e-01 -5.90663910e-01 -6.78167284e-01 -4.31114674e-01
3.67174953e-01 -3.36220860e-01 -7.66377032e-01 1.76674759e+00
-3.98617715e-01 3.40376586e-01 -8.34320337e-02 6.84663117e-01
1.11664474e+00 6.75029218e-01 2.37334087e-01 2.12693006e-01
1.15190136e+00 -9.60661054e-01 -5.03885984e-01 -4.31480497e-01
6.28531516e-01 -7.28693068e-01 1.22796535e+00 4.21702981e-01
-8.18868995e-01 -4.51909840e-01 -1.16757011e+00 9.71328169e-02
-5.32575369e-01 1.03130728e-01 1.21649361e+00 6.73585594e-01
-1.07650244e+00 4.50355321e-01 -6.96383178e-01 -6.24653757e-01
3.78594190e-01 1.49010256e-01 -1.87720478e-01 -1.47544518e-02
-1.83456910e+00 1.12564909e+00 7.36385345e-01 8.24195519e-02
-9.09959614e-01 -9.34138060e-01 -3.73470724e-01 1.30555719e-01
4.69083458e-01 -7.44565308e-01 1.45645738e+00 -4.88665909e-01
-1.00805974e+00 5.46695352e-01 2.47686371e-01 -8.82221997e-01
7.93459266e-02 -6.74843609e-01 -4.17835057e-01 -1.49425104e-01
-2.17135668e-01 7.87075758e-01 6.09492779e-01 -1.01063979e+00
-6.08097732e-01 -1.49804994e-01 3.63487512e-01 3.34367067e-01
-6.63392723e-01 4.07440774e-02 -4.55209643e-01 -9.90748286e-01
-2.62018681e-01 -8.29685152e-01 -1.12679556e-01 -6.32419407e-01
-2.99248695e-01 -4.74127829e-01 3.30863804e-01 -3.46620947e-01
1.63818967e+00 -1.92572641e+00 -1.55982018e-01 -1.95729196e-01
4.06296998e-01 3.81371588e-01 -3.40926230e-01 9.36056077e-01
7.96408653e-02 5.53154409e-01 4.26882029e-01 -1.51622072e-01
4.28364202e-02 2.70729139e-02 -8.50769699e-01 -3.52674420e-03
1.76556364e-01 1.39290488e+00 -8.15121949e-01 -1.66090816e-01
-1.24900781e-01 3.86858463e-01 -4.59330946e-01 -9.44069237e-04
-2.73166418e-01 -9.92072076e-02 -4.92765516e-01 5.48561871e-01
3.31609398e-01 -4.40926731e-01 -2.00448737e-01 -2.59331286e-01
4.55554351e-02 6.83935761e-01 -7.60100543e-01 1.23849571e+00
-2.69498855e-01 1.03366745e+00 -2.93056726e-01 -9.46565449e-01
8.03569019e-01 2.27195576e-01 5.01559198e-01 -6.14318192e-01
1.99745893e-01 1.45209819e-01 3.06030005e-01 4.29455899e-02
9.93207574e-01 -2.09322665e-02 1.36673585e-01 6.76197410e-01
8.84035304e-02 -1.34791940e-01 7.46290237e-02 3.41688871e-01
9.12826002e-01 -3.83616090e-01 1.91017732e-01 -3.85391921e-01
-1.27460897e-01 6.61463067e-02 2.76502162e-01 1.23821056e+00
-1.77003548e-01 3.54378670e-01 4.30890530e-01 -5.04499078e-01
-6.95564747e-01 -7.48164117e-01 -4.92864801e-03 1.09427738e+00
-7.21015930e-02 -9.25313115e-01 -5.84527910e-01 -3.29784006e-01
1.61880001e-01 8.05345237e-01 -8.09350848e-01 -5.12912214e-01
-4.06370729e-01 -1.29414833e+00 1.12720323e+00 7.93801606e-01
5.79251885e-01 -1.17050505e+00 -5.18725932e-01 1.50200799e-01
-3.30850005e-01 -1.04869223e+00 -4.86533977e-02 2.30007842e-01
-1.10569966e+00 -9.59075630e-01 -6.14375293e-01 -5.64285398e-01
2.54715651e-01 6.02630079e-01 1.30220985e+00 2.43392244e-01
1.93921834e-01 3.57233316e-01 -6.63419008e-01 -9.92155254e-01
-3.54778200e-01 8.30394864e-01 1.78307518e-01 -4.56094176e-01
8.05904984e-01 -4.41547841e-01 -6.03182256e-01 1.58398729e-02
-9.83312964e-01 1.53873786e-01 8.60209644e-01 6.78940415e-01
2.67430616e-04 2.62392908e-01 9.07181084e-01 -8.33687067e-01
1.55583179e+00 -5.80342889e-01 -2.87726730e-01 5.01858220e-02
-1.31888843e+00 -8.60716179e-02 3.31196219e-01 -6.24631882e-01
-7.05926478e-01 -9.20830011e-01 7.17941523e-02 -2.76340395e-01
2.43944123e-01 1.13404965e+00 3.43518823e-01 -2.44585648e-02
8.51728082e-01 3.64223838e-01 -2.06635535e-01 -5.34555495e-01
4.70029593e-01 5.51319242e-01 3.33958656e-01 -8.91309440e-01
9.34921503e-01 6.56829998e-02 -4.47963476e-01 -7.91522801e-01
-7.91063428e-01 -1.17038697e-01 -2.23210484e-01 -9.62564629e-03
3.63675863e-01 -1.08363163e+00 -1.52351737e-01 8.47948432e-01
-8.91849399e-01 -6.26220822e-01 7.61174783e-02 5.27164459e-01
-1.54411018e-01 1.54757649e-01 -9.96707499e-01 -3.34648401e-01
-6.49210513e-01 -1.20652926e+00 5.38891971e-01 4.76738840e-01
-4.59841639e-01 -9.66833055e-01 1.33578107e-02 4.25203174e-01
8.60921025e-01 -3.84997308e-01 1.02840114e+00 -7.55070627e-01
-6.85063303e-01 -3.37331176e-01 -3.17970425e-01 4.74246889e-01
5.56973070e-02 6.44379705e-02 -1.02822590e+00 -5.09114325e-01
-1.14988126e-01 -5.46645939e-01 1.14718270e+00 6.05735958e-01
1.14560699e+00 -2.86670893e-01 -3.97350132e-01 5.46791315e-01
1.10504305e+00 1.66822791e-01 5.72613060e-01 7.57503688e-01
6.60657287e-01 2.56991178e-01 5.68581104e-01 1.08337373e-01
5.92164755e-01 4.24455643e-01 2.88474560e-01 3.40533033e-02
-1.08623683e-01 -5.09579659e-01 2.32712328e-01 8.31091940e-01
-2.57783681e-01 -4.27763373e-01 -1.09415233e+00 5.02851009e-01
-1.48574054e+00 -7.51827121e-01 3.14688504e-01 2.00187159e+00
8.59950006e-01 2.73785084e-01 -5.43606840e-02 1.93996862e-01
3.06042999e-01 5.70088863e-01 -3.86519909e-01 -1.04161210e-01
-1.63069263e-01 3.01603884e-01 5.88355660e-01 2.55504072e-01
-9.84627664e-01 1.30933154e+00 7.66718149e+00 6.22538626e-01
-1.34760273e+00 -4.08699542e-01 5.71119010e-01 -5.06147556e-02
-3.20501089e-01 -4.61884104e-02 -9.08594847e-01 3.16338301e-01
9.87674534e-01 -7.45701849e-01 3.81841928e-01 7.30897486e-01
4.26627286e-02 3.60394865e-02 -1.05704558e+00 8.68286908e-01
1.33551985e-01 -1.58201182e+00 2.62300819e-01 1.41183734e-01
6.89019501e-01 4.79391724e-01 2.96318382e-01 6.66821837e-01
5.15322089e-01 -1.06949794e+00 5.06351411e-01 9.94476676e-02
3.74896049e-01 -5.51314473e-01 6.00301623e-01 2.23448157e-01
-7.56566107e-01 -2.22800329e-01 -3.87679935e-01 -4.28979933e-01
-2.10069880e-01 3.94558042e-01 -1.10945606e+00 5.22609055e-01
7.98248768e-01 4.97190952e-01 -9.85392332e-01 1.00267863e+00
-3.77721071e-01 1.30839157e+00 -8.82170126e-02 -2.87037134e-01
4.93068993e-01 6.03717007e-02 4.74093944e-01 1.03548908e+00
1.07178628e-01 -1.43255904e-01 -1.56907305e-01 8.14243078e-01
-2.38369018e-01 3.89589369e-02 -5.74323297e-01 -7.43165135e-01
8.21896195e-01 1.01705766e+00 -6.67154074e-01 -3.35379004e-01
-6.33399487e-01 5.99482954e-01 2.60447562e-01 6.61889553e-01
-7.78128564e-01 -1.68046504e-01 7.17768550e-01 1.04817033e-01
-8.74751061e-02 -4.09460336e-01 -4.29833204e-01 -1.30231738e+00
-4.33001965e-02 -1.22048891e+00 4.63833779e-01 -7.82613397e-01
-1.27238691e+00 4.94912326e-01 3.98718536e-01 -6.52853668e-01
-1.09805249e-01 -5.27426362e-01 -4.86277431e-01 7.95837164e-01
-1.51408947e+00 -1.15353775e+00 -2.41494715e-01 2.99243361e-01
3.56185853e-01 -3.53353500e-01 8.10052097e-01 2.28021741e-01
-6.77974761e-01 8.54032099e-01 -1.07484959e-01 2.94988096e-01
9.83441949e-01 -1.07536674e+00 9.73962307e-01 8.58786106e-01
3.30962151e-01 1.12705886e+00 6.28044486e-01 -7.02389240e-01
-1.29175794e+00 -7.49576867e-01 7.82469928e-01 -8.18098724e-01
7.90592194e-01 -6.72788173e-02 -8.23409259e-01 9.34240580e-01
2.21713170e-01 -5.25197506e-01 6.80457175e-01 6.76757574e-01
-5.48019826e-01 -4.67933267e-02 -4.66722071e-01 8.23636234e-01
8.74716699e-01 -6.07351422e-01 -7.55111277e-01 -1.60766542e-01
8.34152997e-01 -3.75359803e-01 -9.41744089e-01 6.99584484e-01
5.80695331e-01 -7.40105271e-01 1.02305818e+00 -7.62393713e-01
5.19760251e-01 2.20573425e-01 -8.98167565e-02 -1.52665508e+00
-4.91601527e-01 -4.63402480e-01 -3.96079957e-01 9.40152705e-01
4.16749001e-01 -8.50740612e-01 1.07273388e+00 5.71459174e-01
-5.40332124e-02 -1.14836645e+00 -4.17121708e-01 -6.06241405e-01
5.78513622e-01 -7.25547731e-01 8.17047298e-01 1.20448446e+00
-1.97867259e-01 6.08169556e-01 -5.25115311e-01 1.11100540e-01
1.82978079e-01 -5.75462766e-02 1.03909326e+00 -1.20574796e+00
-1.64134637e-01 -7.75136888e-01 -2.49840260e-01 -1.06427968e+00
-4.96334061e-02 -8.18049312e-01 -5.93603790e-01 -1.55181623e+00
2.44803563e-01 -3.15973550e-01 -7.59829462e-01 6.20009422e-01
-5.85413992e-01 1.74229756e-01 3.53609979e-01 3.14227700e-01
-3.13050181e-01 4.04948145e-01 1.21436334e+00 -1.53217793e-01
-2.16186419e-01 2.40628999e-02 -1.40727675e+00 3.68738979e-01
1.06937432e+00 -2.47411072e-01 -9.36876178e-01 -6.07182205e-01
5.43341875e-01 -4.03520584e-01 1.82637274e-01 -9.29472387e-01
2.46700153e-01 -3.73008996e-01 3.36239755e-01 -6.06809258e-01
1.85563818e-01 -4.68061000e-01 -8.13774988e-02 4.70028013e-01
-3.93301606e-01 4.04240072e-01 4.99630064e-01 3.84134322e-01
-2.18830764e-01 2.35555787e-02 3.23950678e-01 -1.09480910e-01
-9.93320286e-01 3.29634726e-01 -4.30273652e-01 1.00696512e-01
6.69597685e-01 -2.20088646e-01 -4.92584229e-01 -6.53877139e-01
-2.25883648e-02 2.34841332e-01 3.77233803e-01 1.10048938e+00
4.43941593e-01 -1.30615103e+00 -6.80584788e-01 1.02696694e-01
3.04917663e-01 -4.27403413e-02 3.62277515e-02 5.91875613e-01
-3.43386948e-01 9.76326466e-01 -2.15111718e-01 4.02489398e-03
-8.59017849e-01 2.61675626e-01 3.90764028e-01 -4.88685131e-01
-5.75812519e-01 1.02060604e+00 -2.18548607e-02 -2.20509797e-01
4.67712075e-01 -5.88459551e-01 -2.28619739e-01 2.42416173e-01
5.42638481e-01 3.91579062e-01 1.94311842e-01 -1.79108456e-01
-1.40308112e-01 -9.27746743e-02 -6.88308358e-01 -4.73787785e-02
1.13337529e+00 3.50603282e-01 7.86797106e-02 4.88227367e-01
8.22520137e-01 -3.07567984e-01 -9.36418951e-01 -3.40900600e-01
-1.11164659e-01 -2.85017371e-01 3.02896321e-01 -1.07506573e+00
-1.02228916e+00 6.91434503e-01 2.68691540e-01 2.73367167e-01
1.05524468e+00 -4.11842801e-02 9.46650445e-01 5.77949226e-01
5.50499737e-01 -1.03117824e+00 2.82463044e-01 7.28482962e-01
9.01523709e-01 -1.02779341e+00 8.52392688e-02 1.01977982e-01
-5.87671638e-01 8.89998198e-01 7.88801193e-01 5.54235689e-02
6.12615108e-01 1.98456347e-02 2.50228435e-01 -5.67005813e-01
-1.13502347e+00 2.39900276e-01 3.85012448e-01 4.84605223e-01
7.06313550e-01 1.14375159e-01 -5.69936037e-01 8.14167976e-01
-7.86714613e-01 3.03425342e-01 4.26433474e-01 8.18742752e-01
-2.40494847e-01 -1.17583263e+00 -2.35985279e-01 9.05793071e-01
-6.58697963e-01 -5.10028303e-01 -6.85719609e-01 1.05555499e+00
-4.20872062e-01 9.21483397e-01 -1.03078790e-01 -8.64468157e-01
2.06493735e-01 2.35245466e-01 5.15218377e-01 -6.01963043e-01
-4.34066921e-01 -3.72577965e-01 7.71734491e-02 -3.44163537e-01
-1.09782331e-01 -2.63558060e-01 -6.19009376e-01 -6.74504399e-01
-4.79862839e-01 2.75855154e-01 3.90333951e-01 7.18708575e-01
5.06398678e-01 3.98256004e-01 2.17492819e-01 -5.35136282e-01
-8.67962837e-01 -1.37709081e+00 -4.78587747e-01 2.51367658e-01
-4.13613133e-02 -6.68673158e-01 -3.50433856e-01 -4.16904896e-01] | [10.756481170654297, 8.282617568969727] |
1293a324-f525-46e7-8816-9274b057b14a | knowledge-prompted-estimator-a-novel-approach | 2306.07486 | null | https://arxiv.org/abs/2306.07486v1 | https://arxiv.org/pdf/2306.07486v1.pdf | Knowledge-Prompted Estimator: A Novel Approach to Explainable Machine Translation Assessment | Cross-lingual Machine Translation (MT) quality estimation plays a crucial role in evaluating translation performance. GEMBA, the first MT quality assessment metric based on Large Language Models (LLMs), employs one-step prompting to achieve state-of-the-art (SOTA) in system-level MT quality estimation; however, it lacks segment-level analysis. In contrast, Chain-of-Thought (CoT) prompting outperforms one-step prompting by offering improved reasoning and explainability. In this paper, we introduce Knowledge-Prompted Estimator (KPE), a CoT prompting method that combines three one-step prompting techniques, including perplexity, token-level similarity, and sentence-level similarity. This method attains enhanced performance for segment-level estimation compared with previous deep learning models and one-step prompting approaches. Furthermore, supplementary experiments on word-level visualized alignment demonstrate that our KPE method significantly improves token alignment compared with earlier models and provides better interpretability for MT quality estimation. Code will be released upon publication. | ['Yanfei Jiang', 'Daimeng Wei', 'Minghan Wang', 'Shimin Tao', 'Min Zhang', 'Hao Yang'] | 2023-06-13 | null | null | null | null | ['machine-translation'] | ['natural-language-processing'] | [ 1.59980040e-02 -1.02405213e-01 -6.37439013e-01 -3.60079199e-01
-1.81037366e+00 -6.11607492e-01 8.46759677e-01 3.04918498e-01
-3.24987769e-01 7.28657842e-01 5.70201039e-01 -7.56189287e-01
5.73835671e-02 -3.51079106e-01 -7.80089974e-01 -1.18554980e-01
4.69084769e-01 7.54744768e-01 -2.08241329e-01 -3.95350784e-01
4.25545096e-01 3.11560556e-02 -8.17274868e-01 6.46765172e-01
1.70929098e+00 6.11044288e-01 3.35570216e-01 6.88086808e-01
-3.52903783e-01 9.39913452e-01 -6.45233929e-01 -8.79556179e-01
-1.26188397e-01 -4.48100090e-01 -1.30199599e+00 -4.99060094e-01
7.46767521e-01 -2.94007361e-01 2.51668375e-02 9.68484640e-01
6.35074496e-01 -3.15153182e-01 5.86024225e-01 -1.13612950e+00
-1.16125154e+00 1.14409435e+00 -2.52016664e-01 6.02778137e-01
4.59366053e-01 4.66168195e-01 1.42526007e+00 -1.30029964e+00
5.14639437e-01 1.24377239e+00 7.29248405e-01 2.45477930e-01
-1.26740885e+00 -3.40940535e-01 -7.30517805e-02 6.59519255e-01
-9.92733479e-01 -5.27422369e-01 4.48668450e-01 -3.89179587e-01
1.67746174e+00 7.66452774e-02 3.57203662e-01 1.28712952e+00
6.24351025e-01 1.23217118e+00 1.27290404e+00 -7.63449073e-01
-4.13615890e-02 -1.42784998e-01 1.68221891e-01 6.34427667e-01
-4.28137630e-02 4.54911739e-02 -9.23925459e-01 6.21480159e-02
4.24323142e-01 -6.22452855e-01 5.12035340e-02 4.24448371e-01
-1.88290477e+00 5.84293246e-01 1.61540300e-01 4.80624884e-01
-5.82685173e-01 9.30928141e-02 3.28367352e-01 7.10271180e-01
7.26025105e-01 6.98290944e-01 -6.48235321e-01 -9.93215919e-01
-1.32342434e+00 2.15755269e-01 7.91502357e-01 1.08722198e+00
5.00775993e-01 4.71070446e-02 -8.51963043e-01 9.23453867e-01
1.65481925e-01 8.66805255e-01 7.36029506e-01 -8.32645953e-01
1.11379266e+00 5.55882454e-01 -1.17979953e-02 -6.13871098e-01
-1.16321005e-01 -7.56618857e-01 -6.38121486e-01 -4.40274388e-01
3.72098774e-01 1.26768708e-01 -7.13608325e-01 1.67432618e+00
-8.92945454e-02 -4.49142784e-01 6.31538928e-02 7.00719714e-01
7.17123389e-01 7.25708842e-01 4.75150757e-02 -4.20928121e-01
1.34154534e+00 -1.47411549e+00 -9.78632390e-01 -3.43811095e-01
1.02725661e+00 -1.26251972e+00 1.61070812e+00 3.41149449e-01
-1.32198727e+00 -8.10252011e-01 -8.44370365e-01 -3.99978995e-01
-1.88162804e-01 2.92696595e-01 2.16737092e-01 5.12203097e-01
-1.37470019e+00 7.10568905e-01 -8.44715655e-01 -4.43372726e-01
-1.05500156e-02 -3.62734906e-02 -1.34481102e-01 -3.86606157e-02
-1.51376927e+00 1.61942220e+00 -6.83668954e-03 -8.42248350e-02
-6.63458347e-01 -8.26044321e-01 -8.45175683e-01 6.72078729e-02
1.30846919e-02 -8.38066816e-01 1.95254278e+00 -6.46543741e-01
-1.74516654e+00 5.90186238e-01 -6.34023666e-01 -3.98997992e-01
4.90534663e-01 -6.31700754e-01 -4.09520656e-01 -2.06963703e-01
4.14880633e-01 6.79826736e-01 5.30587733e-01 -7.07469463e-01
-5.07277310e-01 1.55272223e-02 -1.20603271e-01 4.53655243e-01
-2.88931161e-01 3.59772921e-01 -2.24596888e-01 -5.26506007e-01
-5.32588325e-02 -5.35943449e-01 2.18744427e-02 -5.69016695e-01
-3.94912004e-01 -4.98100489e-01 6.10320605e-02 -1.18470871e+00
1.75398922e+00 -1.47396374e+00 2.97256112e-01 -3.04478079e-01
1.89891353e-01 3.63694161e-01 -5.50868452e-01 8.71799350e-01
3.30337107e-01 3.79913598e-01 -4.94246408e-02 -3.77246439e-01
4.49747503e-01 -1.30378827e-01 -6.10734895e-02 9.79012065e-03
5.74443400e-01 1.53695738e+00 -1.13571525e+00 -7.92647243e-01
6.42174184e-02 -2.58752201e-02 -3.82853836e-01 2.72996843e-01
-2.94088215e-01 2.73226291e-01 -1.22330256e-03 7.63169527e-01
3.83735806e-01 -2.99912900e-01 7.99324811e-02 -1.04992799e-01
-1.51580229e-01 1.08062279e+00 -3.29612881e-01 1.95041633e+00
-7.63006687e-01 8.19712460e-01 -6.07734561e-01 -3.83220851e-01
8.63377154e-01 4.75516856e-01 2.25513771e-01 -1.31794333e+00
-9.80388969e-02 5.23567259e-01 2.98542380e-01 -5.02995908e-01
8.68584514e-01 1.16069160e-01 -1.09514564e-01 6.60776913e-01
1.25340804e-01 -2.17487127e-01 2.97947615e-01 3.49988937e-01
1.10757661e+00 2.50624001e-01 4.13370132e-01 -2.92396963e-01
3.77535611e-01 1.72288641e-01 6.23092949e-01 6.28835917e-01
-5.24685383e-01 2.93500870e-01 2.05408469e-01 -2.69087940e-01
-1.36884701e+00 -1.03905535e+00 5.42202443e-02 9.92960215e-01
-2.02969417e-01 -7.32926309e-01 -1.15095711e+00 -6.18875802e-01
-1.68280110e-01 9.84622836e-01 -1.92233831e-01 -9.02779922e-02
-7.16138184e-01 -5.03659010e-01 8.19521606e-01 6.52263343e-01
3.39885980e-01 -9.81133401e-01 -3.66399661e-02 5.34389496e-01
-1.18569922e+00 -1.21647060e+00 -7.71597981e-01 -9.38608572e-02
-9.89982367e-01 -4.50280517e-01 -6.30270004e-01 -6.74243629e-01
1.55310348e-01 1.20466501e-01 1.70071805e+00 -1.20897330e-01
3.70244354e-01 1.66280288e-02 -5.34935892e-01 -1.32977039e-01
-8.63527060e-01 5.45605361e-01 3.22119117e-01 -6.85524523e-01
6.73592865e-01 -3.56588155e-01 -4.23827857e-01 2.29899660e-01
-5.05752981e-01 3.77194077e-01 1.11908972e+00 9.58926499e-01
4.55136836e-01 -5.58215201e-01 6.37814224e-01 -4.21931654e-01
1.14646113e+00 -1.79369822e-01 -6.11588508e-02 6.77144349e-01
-1.11332381e+00 2.58417934e-01 6.32546723e-01 -2.86046922e-01
-9.92797375e-01 -6.94248736e-01 -2.77278572e-01 6.71210960e-02
8.68202746e-02 9.13632512e-01 -6.75500780e-02 3.04230154e-01
8.26443970e-01 4.71663862e-01 -2.19274402e-01 -5.68849325e-01
7.10281491e-01 9.97270226e-01 5.23591399e-01 -9.44433987e-01
6.24304712e-01 -4.77627635e-01 -5.27652264e-01 -2.29596391e-01
-7.50034630e-01 -3.22111070e-01 -9.18257475e-01 -1.94361433e-01
5.41030347e-01 -1.00341845e+00 -4.02365834e-01 4.74707454e-01
-1.47686338e+00 -4.50840920e-01 5.17212860e-02 5.72778225e-01
-7.96218574e-01 5.19487679e-01 -1.11128378e+00 -4.60217923e-01
-9.37548935e-01 -1.20134938e+00 1.28505433e+00 -1.54911876e-01
-5.87901533e-01 -9.96641576e-01 2.79495031e-01 8.27813447e-01
6.64329410e-01 -3.59656751e-01 1.20278251e+00 -5.39412439e-01
-4.66059715e-01 1.90679416e-01 -1.94896623e-01 3.72336119e-01
1.07284166e-01 -4.11946699e-03 -6.41060174e-01 -1.59027755e-01
-1.24633931e-01 -3.26164812e-01 4.15812939e-01 3.92800927e-01
4.57002878e-01 -6.09917343e-01 8.97143185e-02 2.32613459e-01
1.08450401e+00 -1.60033777e-01 5.54809868e-01 6.29187286e-01
7.28144884e-01 3.93349409e-01 8.77207041e-01 6.29699305e-02
9.98365164e-01 8.99797320e-01 -3.12165618e-02 -1.63377538e-01
-3.49712223e-01 -4.72895294e-01 9.54512715e-01 1.85067821e+00
9.91008207e-02 -3.19336355e-01 -1.16389644e+00 5.09536505e-01
-1.98550177e+00 -8.76104116e-01 -4.50079888e-01 1.81353533e+00
1.23999631e+00 1.65937722e-01 8.77416953e-02 -5.38865365e-02
5.01163840e-01 -3.84008326e-02 -4.10459727e-01 -9.62794840e-01
-1.80700123e-01 5.55463023e-02 2.20557377e-01 7.28263140e-01
-4.54590678e-01 1.48125792e+00 6.57017326e+00 1.14585769e+00
-8.26451480e-01 4.88538742e-01 4.57751304e-01 1.19774111e-01
-7.17841089e-01 8.69606063e-02 -8.00236166e-01 4.71249074e-01
1.49608505e+00 -4.18225169e-01 5.09495437e-01 6.91751003e-01
5.70441663e-01 1.03293126e-02 -1.48140049e+00 7.57804036e-01
-1.35747660e-02 -1.15943539e+00 2.24907577e-01 6.92150667e-02
8.04195106e-01 3.58807176e-01 -7.16840252e-02 5.86858690e-01
4.21670645e-01 -7.54127383e-01 1.00824201e+00 4.33671683e-01
9.20131564e-01 -6.20578885e-01 9.18966532e-01 6.34541810e-01
-1.00935185e+00 1.36450037e-01 -3.67082089e-01 -2.78990388e-01
3.99147451e-01 7.89960623e-01 -9.99537289e-01 9.14223075e-01
3.44736934e-01 6.16439044e-01 -6.41787291e-01 6.67566836e-01
-4.13032115e-01 1.02130842e+00 2.10515976e-01 -1.72578901e-01
3.93019348e-01 2.78504323e-02 5.06728292e-01 1.60468733e+00
3.86200011e-01 -4.18340892e-01 3.32824029e-02 9.08032477e-01
-2.24784598e-01 4.02608544e-01 -2.24030048e-01 -3.34531218e-01
7.25980401e-01 1.04326451e+00 -8.85136649e-02 -6.17789507e-01
-3.53662372e-01 1.22158194e+00 7.55804956e-01 1.19916908e-01
-7.53916323e-01 -2.44290173e-01 7.12989271e-01 -2.91805387e-01
-8.80715698e-02 -4.07526225e-01 -6.04041576e-01 -1.36682713e+00
3.74089032e-01 -1.44144809e+00 4.00241315e-02 -6.63938761e-01
-1.40311027e+00 7.87916183e-01 -2.69548625e-01 -1.41015482e+00
-5.88099599e-01 -4.85950738e-01 -3.69177490e-01 1.24128652e+00
-1.63736415e+00 -1.40109956e+00 1.21536210e-01 1.70078352e-01
1.06094122e+00 -1.39155045e-01 8.18738937e-01 2.99318522e-01
-4.78843659e-01 9.97777641e-01 2.64759183e-01 6.65551275e-02
9.40766394e-01 -1.45350134e+00 1.08928931e+00 1.14843476e+00
3.75758559e-01 7.65119851e-01 5.55212379e-01 -7.31465280e-01
-1.49941170e+00 -8.66103411e-01 2.19216204e+00 -1.06067336e+00
7.72765696e-01 -2.53526121e-01 -7.22240031e-01 5.09808600e-01
6.79185390e-01 -6.79538190e-01 5.82785487e-01 4.87091571e-01
-4.15114373e-01 -1.51065156e-01 -7.47699797e-01 7.37882674e-01
1.02808762e+00 -9.08431470e-01 -9.98868525e-01 4.46812212e-01
1.18897474e+00 -4.38969076e-01 -1.19614625e+00 4.23812807e-01
5.56579888e-01 -6.61708534e-01 5.38918316e-01 -6.10400498e-01
8.51958692e-01 -1.57280177e-01 -1.45157933e-01 -1.75937974e+00
-8.39480639e-01 -8.44229221e-01 -3.25233996e-01 1.37216473e+00
8.35187674e-01 -3.38750899e-01 2.75107026e-01 3.16222370e-01
-6.20364308e-01 -9.10792470e-01 -8.83432269e-01 -1.22036302e+00
6.32179379e-01 -4.92753625e-01 8.99335206e-01 9.73171651e-01
5.61584890e-01 6.34584665e-01 -4.10127103e-01 -8.58210102e-02
3.96537960e-01 -1.70932505e-02 6.33959293e-01 -6.55274808e-01
-2.86552906e-01 -8.05462778e-01 1.77097339e-02 -1.21706474e+00
2.98889056e-02 -1.11805797e+00 2.38173798e-01 -2.01049089e+00
4.68851656e-01 4.67837963e-04 -2.39458293e-01 4.33716089e-01
-8.06847453e-01 -7.03953505e-02 8.30738097e-02 4.33550864e-01
-6.64314508e-01 6.91685498e-01 1.40445566e+00 -2.42584482e-01
7.21102878e-02 -3.16658616e-01 -6.79263055e-01 2.20069550e-02
7.54216850e-01 -4.44164127e-01 -1.71435356e-01 -1.07973123e+00
4.32523996e-01 6.69080839e-02 -1.36551782e-01 -7.99568474e-01
3.15737963e-01 -2.19474614e-01 2.04259843e-01 -7.96373725e-01
-1.08239599e-01 -1.69115633e-01 -1.00945868e-01 5.15998006e-01
-7.01901674e-01 8.41863513e-01 1.23888314e-01 6.57760501e-02
-3.90628308e-01 9.17898118e-02 2.90416867e-01 -6.94971979e-02
-4.72698450e-01 1.63601384e-01 -5.92218339e-01 2.14476719e-01
1.49156198e-01 -1.01363458e-01 -6.50844574e-01 -3.43659550e-01
-1.15507424e-01 4.08420920e-01 3.25510472e-01 6.36825860e-01
4.05945569e-01 -1.64121199e+00 -1.33024585e+00 -2.88149387e-01
5.74172676e-01 -5.68811476e-01 -3.72711606e-02 1.19032288e+00
-3.52618605e-01 8.91595483e-01 -7.98346698e-02 -7.31290162e-01
-1.02509141e+00 1.78027838e-01 8.29206547e-04 -8.14632535e-01
-2.70006061e-01 7.09508538e-01 -4.46911663e-01 -9.44369972e-01
-1.23791120e-04 -5.93924344e-01 2.39585564e-01 -2.72759378e-01
4.66293544e-01 5.28387547e-01 5.29780924e-01 -5.57135165e-01
-3.04179758e-01 3.22250754e-01 -3.51972550e-01 -5.77852249e-01
8.77189159e-01 -4.02580202e-01 -2.46875003e-01 5.13582170e-01
8.85879636e-01 -2.31768295e-01 -7.47097015e-01 -7.39770949e-01
5.70179105e-01 -4.47439700e-01 -7.84294754e-02 -1.50215292e+00
-2.47112378e-01 1.15818763e+00 2.93922246e-01 -3.40166807e-01
8.12654197e-01 -1.75913915e-01 1.19100273e+00 5.82253754e-01
4.62135732e-01 -1.45391023e+00 2.34329119e-01 9.16709363e-01
7.87370980e-01 -1.45639205e+00 -3.65356892e-01 -5.99617064e-02
-6.67400479e-01 1.01039779e+00 7.83041239e-01 7.07191885e-01
-1.13529913e-01 6.56586662e-02 5.18923342e-01 1.51128411e-01
-1.22795320e+00 -4.31890637e-02 6.79069102e-01 4.59144622e-01
1.00290513e+00 3.96555275e-01 -6.71537161e-01 5.05014479e-01
-6.57624006e-01 -6.45405725e-02 9.51646343e-02 6.43166602e-01
-5.64022422e-01 -1.39918721e+00 -5.84422275e-02 2.90140450e-01
-3.37251097e-01 -9.12903249e-01 -5.94008684e-01 4.36810255e-01
-2.01055214e-01 1.35213184e+00 -2.52658188e-01 -6.94403410e-01
2.13921741e-01 2.35362247e-01 6.58055902e-01 -5.59366465e-01
-9.76838291e-01 -1.47327989e-01 4.05615211e-01 -6.99209273e-01
1.01082344e-02 -6.04877472e-01 -8.05096209e-01 -7.84355879e-01
-4.49220598e-01 2.72590071e-01 8.65869462e-01 1.32753241e+00
5.40928364e-01 3.46664369e-01 6.21579230e-01 -3.40462506e-01
-9.07527804e-01 -1.64590394e+00 9.28397849e-02 1.79160476e-01
2.76051044e-01 -1.52182505e-01 4.10403730e-03 4.16441597e-02] | [11.666573524475098, 10.285221099853516] |
28017177-05b1-45e5-8303-c6c6a87b5339 | sparse-message-passing-network-with-feature | 2212.02992 | null | https://arxiv.org/abs/2212.02992v1 | https://arxiv.org/pdf/2212.02992v1.pdf | Sparse Message Passing Network with Feature Integration for Online Multiple Object Tracking | Existing Multiple Object Tracking (MOT) methods design complex architectures for better tracking performance. However, without a proper organization of input information, they still fail to perform tracking robustly and suffer from frequent identity switches. In this paper, we propose two novel methods together with a simple online Message Passing Network (MPN) to address these limitations. First, we explore different integration methods for the graph node and edge embeddings and put forward a new IoU (Intersection over Union) guided function, which improves long term tracking and handles identity switches. Second, we introduce a hierarchical sampling strategy to construct sparser graphs which allows to focus the training on more difficult samples. Experimental results demonstrate that a simple online MPN with these two contributions can perform better than many state-of-the-art methods. In addition, our association method generalizes well and can also improve the results of private detection based methods. | ['Guo Cao', 'Horst Bischof', 'Horst Possegger', 'Bisheng Wang'] | 2022-12-06 | null | null | null | null | ['multiple-object-tracking'] | ['computer-vision'] | [-1.71216547e-01 -2.92224914e-01 -3.59606683e-01 1.42417746e-02
-3.56809318e-01 -5.72716773e-01 6.43275619e-01 1.26726687e-01
-4.31293577e-01 6.82345688e-01 -1.85467660e-01 -8.45109895e-02
-1.11039191e-01 -7.44793594e-01 -8.20079446e-01 -5.98830402e-01
-4.65313047e-01 6.09808564e-01 9.06174064e-01 4.63432372e-02
-1.09544732e-01 5.93183577e-01 -1.59360230e+00 -2.13532805e-01
6.98576450e-01 9.57190633e-01 1.12198360e-01 5.46287179e-01
-1.46029666e-01 6.18909955e-01 -5.06899893e-01 -4.31766421e-01
5.26257575e-01 -2.09782898e-01 -3.95329326e-01 -2.36335069e-01
9.92663383e-01 -4.65734005e-01 -5.65624595e-01 1.18634760e+00
5.30810893e-01 2.43706748e-01 2.57738590e-01 -1.58440411e+00
-7.29722142e-01 6.36370003e-01 -5.62855601e-01 1.58035263e-01
8.99856165e-02 2.93490857e-01 9.35863912e-01 -6.58743024e-01
7.55108118e-01 1.31133306e+00 1.20755851e+00 7.50836670e-01
-1.28443885e+00 -9.58150208e-01 4.81909931e-01 1.99931651e-01
-1.53607750e+00 -4.21129107e-01 4.78432089e-01 -1.63507506e-01
5.62446356e-01 1.86381474e-01 8.15374076e-01 1.03355539e+00
4.39925455e-02 7.27597237e-01 7.42770791e-01 -1.17298573e-01
-5.46424985e-02 5.76163009e-02 1.03618875e-01 1.05383086e+00
9.08429980e-01 2.52957702e-01 -5.14601231e-01 -3.29184026e-01
6.64791942e-01 3.21971297e-01 -2.24148095e-01 -8.62712026e-01
-1.28344083e+00 7.32417464e-01 7.46454716e-01 3.71428013e-01
-6.01375997e-02 6.80370510e-01 2.19135374e-01 1.83727473e-01
1.86194435e-01 9.16996226e-02 -6.88885599e-02 1.52218759e-01
-1.10446322e+00 2.99413085e-01 8.62614930e-01 1.14941680e+00
8.65627229e-01 -4.96191308e-02 -5.06833911e-01 3.35471392e-01
5.34503698e-01 6.79074526e-01 -1.07897401e-01 -8.48527193e-01
1.14600457e-01 6.52363181e-01 6.68153912e-02 -1.09733260e+00
-4.39756870e-01 -5.80083132e-01 -6.83490157e-01 1.62890419e-01
6.40892982e-01 -5.08530103e-02 -9.02793944e-01 1.82267833e+00
6.69871330e-01 6.42315984e-01 -3.11043561e-01 8.17134619e-01
7.32577682e-01 2.09431753e-01 1.96065772e-02 3.02767623e-02
1.45952797e+00 -1.13252747e+00 -7.90345311e-01 -1.11356176e-01
7.04120576e-01 -4.53740865e-01 4.77562726e-01 -3.75025608e-02
-7.83999324e-01 -4.60453361e-01 -1.04563773e+00 8.51442739e-02
-5.50716639e-01 8.37865621e-02 9.01691377e-01 7.77855754e-01
-1.43563759e+00 6.49295986e-01 -9.71173882e-01 -6.05870843e-01
7.21822798e-01 6.18042231e-01 -2.01449290e-01 -1.12972230e-01
-9.36174452e-01 7.12790191e-01 2.98917741e-01 1.63815990e-02
-7.80788183e-01 -6.58851087e-01 -7.85276949e-01 4.77804281e-02
7.28151858e-01 -1.03046632e+00 9.84268725e-01 -3.51305395e-01
-1.39500642e+00 6.10542834e-01 -1.88988075e-01 -7.71527767e-01
6.13359630e-01 -1.73740789e-01 -2.12302700e-01 3.81331481e-02
5.65159470e-02 8.73418510e-01 7.93403387e-01 -1.18299496e+00
-7.84130752e-01 -2.35419810e-01 1.36852473e-01 -2.11262137e-01
-7.11078048e-01 -5.62367328e-02 -6.96564555e-01 -5.62296927e-01
-1.87328547e-01 -1.08214939e+00 -2.96389908e-01 7.11679399e-01
-3.35723251e-01 -3.40206265e-01 1.11054063e+00 -1.48991898e-01
1.37862706e+00 -2.16025567e+00 3.20371240e-02 1.32458523e-01
6.61725342e-01 5.28413117e-01 -2.88711071e-01 2.84965515e-01
6.03067935e-01 -4.55067158e-02 1.59854829e-01 -7.95366228e-01
2.94322789e-01 2.26186112e-01 -7.59690180e-02 7.37011373e-01
1.17633924e-01 9.88772094e-01 -1.00790191e+00 -6.86522067e-01
1.46592334e-01 5.59782982e-01 -6.67594492e-01 -1.81060448e-01
-3.42204422e-01 4.01574552e-01 -3.87620956e-01 8.09097826e-01
8.44155312e-01 -7.94303298e-01 2.98814774e-01 -2.18579397e-01
-2.08188593e-01 -5.13546616e-02 -1.38399696e+00 1.55787873e+00
9.16979164e-02 5.44998527e-01 2.92445958e-01 -6.54895127e-01
6.88616514e-01 -9.50377062e-02 6.61387205e-01 -5.07286668e-01
2.03297719e-01 1.32358775e-01 -1.67323858e-01 6.59032613e-02
6.76300645e-01 2.41707399e-01 2.30937958e-01 1.48008391e-01
-5.30137867e-02 6.91702902e-01 2.56753355e-01 3.59286577e-01
1.29222119e+00 6.74205273e-02 3.86948735e-02 -1.79874286e-01
5.15766740e-01 -9.98934917e-03 8.48450541e-01 1.14663208e+00
-6.01037323e-01 2.98712224e-01 1.23830728e-01 -4.12674278e-01
-5.83947182e-01 -9.74368811e-01 1.32241592e-01 1.18458128e+00
6.64436281e-01 -6.41699791e-01 -3.23463619e-01 -8.22325826e-01
5.68842828e-01 1.63641825e-01 -4.23929840e-01 -7.89740384e-02
-8.32422495e-01 -5.73365450e-01 7.10382521e-01 6.83753729e-01
3.34435940e-01 -5.81635714e-01 -3.09303999e-01 2.82111585e-01
4.54632491e-02 -1.29331338e+00 -6.78984582e-01 -2.54111290e-01
-7.98800230e-01 -1.12394524e+00 -5.99394977e-01 -6.75866604e-01
5.53808689e-01 6.93121910e-01 7.94284403e-01 5.43876052e-01
-1.62436873e-01 4.77702826e-01 -3.00383598e-01 -2.57685542e-01
-1.35089859e-01 3.06147337e-01 3.08101267e-01 1.09501109e-01
2.78409421e-01 -4.32998925e-01 -5.46786547e-01 4.17874575e-01
-6.10873938e-01 -3.16312522e-01 5.74784577e-01 6.53306007e-01
5.22649407e-01 -3.36254269e-01 4.05378640e-01 -5.90043783e-01
1.59843668e-01 -3.02704275e-01 -1.04390633e+00 2.81410933e-01
-7.17640698e-01 3.70388255e-02 5.07433832e-01 -7.88860381e-01
-5.85782886e-01 1.89685583e-01 2.05778077e-01 -8.31029534e-01
1.96111083e-01 -1.16734184e-01 2.07684606e-01 -8.32193851e-01
3.83368015e-01 7.60278702e-02 2.11808190e-01 -6.39088690e-01
4.08994257e-01 1.17537983e-01 4.14097756e-01 -3.44595194e-01
1.29111791e+00 9.59948897e-01 2.22358167e-01 -5.89644313e-01
-7.61740625e-01 -6.42954826e-01 -3.80547047e-01 -3.18953544e-01
6.71258569e-01 -1.09661138e+00 -1.18130898e+00 3.09546292e-01
-1.01446569e+00 -2.99971640e-01 -1.86306655e-01 6.18093431e-01
-1.54688925e-01 6.15828276e-01 -6.50382757e-01 -9.68511045e-01
-2.40720615e-01 -8.89778495e-01 1.30314505e+00 4.19528037e-01
1.45123154e-01 -9.55744922e-01 1.26327634e-01 -5.04327230e-02
7.53884315e-01 2.39738703e-01 1.14156179e-01 -7.56027043e-01
-1.41285253e+00 -1.39475256e-01 -4.92696762e-01 -2.06743687e-01
-4.35016025e-03 -3.49948257e-02 -7.12127864e-01 -9.01658475e-01
-5.08337855e-01 -3.93177941e-02 1.19937861e+00 3.73583734e-01
8.39374304e-01 -3.32649201e-01 -1.14337766e+00 9.54138279e-01
1.41052866e+00 -2.86448479e-01 3.31292570e-01 4.09070373e-01
1.01493227e+00 1.84144765e-01 5.44610262e-01 3.79363090e-01
5.92324495e-01 9.15792346e-01 5.81605911e-01 6.99845552e-02
-5.20203888e-01 -1.89386860e-01 6.32963717e-01 6.67575002e-01
2.21686233e-02 -2.90848970e-01 -5.49458504e-01 5.84661305e-01
-2.16771078e+00 -1.03253138e+00 -2.86502004e-01 2.20181155e+00
4.98387516e-01 1.17043465e-01 4.21482503e-01 -5.34468055e-01
9.59657371e-01 3.72652948e-01 -5.35122573e-01 5.29801667e-01
-1.68310151e-01 -3.21826674e-02 9.32162821e-01 3.28383893e-01
-1.23812711e+00 1.01655030e+00 6.54705524e+00 9.24488902e-01
-8.89972746e-01 3.21484059e-01 -2.57487833e-01 -2.28101537e-01
-5.51724397e-02 1.60493553e-01 -1.56199038e+00 4.74655777e-01
8.93989027e-01 1.76411923e-02 4.48872924e-01 7.89559960e-01
-3.19917023e-01 2.69827873e-01 -1.00890458e+00 9.31409538e-01
-1.39376719e-03 -1.61030948e+00 6.21181540e-03 2.48664811e-01
5.83007514e-01 3.22016239e-01 -1.26930401e-01 3.68080139e-01
6.36487305e-01 -5.48827291e-01 6.95104480e-01 5.86255074e-01
5.49419701e-01 -2.96107620e-01 4.75728005e-01 1.90877974e-01
-1.89347124e+00 -1.52910396e-01 -3.59424174e-01 1.91515014e-01
2.59372294e-01 3.82239372e-01 -5.62401056e-01 8.80005300e-01
7.29374349e-01 8.28018725e-01 -8.55835378e-01 1.58887374e+00
2.36585811e-01 2.87405550e-01 -7.75319815e-01 -3.77593994e-01
1.85424417e-01 1.75429821e-01 9.94477510e-01 1.13677180e+00
2.98670441e-01 -4.90978152e-01 6.73846662e-01 8.93551648e-01
-2.40472630e-01 -2.24846572e-01 -7.90957034e-01 -4.53544669e-02
8.09340417e-01 1.38488209e+00 -8.24099243e-01 -4.72274780e-01
-5.01709163e-01 6.67990744e-01 5.33717632e-01 1.89024329e-01
-1.08716989e+00 -1.49055615e-01 7.95265019e-01 1.13191947e-01
8.48886847e-01 -2.88481027e-01 3.18572283e-01 -1.31327689e+00
1.22195795e-01 -6.03117228e-01 6.18529201e-01 -8.08494389e-02
-1.50137341e+00 3.53793085e-01 -1.93527788e-01 -1.35660791e+00
1.15270063e-01 -5.33854663e-01 -3.41734588e-01 2.74424821e-01
-1.84526372e+00 -1.38772261e+00 -3.13361585e-01 6.08877540e-01
-4.83385846e-03 3.84636596e-03 4.29619670e-01 8.47571492e-01
-6.29611433e-01 9.60065305e-01 1.74584001e-01 2.95901775e-01
9.42204595e-01 -1.02362800e+00 3.93944681e-01 9.35137272e-01
2.01643959e-01 7.47335374e-01 4.99169320e-01 -8.71267557e-01
-1.80866718e+00 -1.43342960e+00 5.92778146e-01 -4.68350947e-01
8.69842708e-01 -5.07613540e-01 -8.33718121e-01 9.72757995e-01
1.05011873e-02 4.37335879e-01 3.47610176e-01 1.34073108e-01
-3.63656282e-01 -2.69067675e-01 -1.07188344e+00 5.42010427e-01
1.46533370e+00 -1.93873718e-01 -1.32573783e-01 3.93398464e-01
9.12086487e-01 -3.74367088e-01 -8.22888672e-01 3.87356609e-01
5.53159356e-01 -8.03616643e-01 1.12988830e+00 -4.36269045e-01
-7.23684192e-01 -8.77296269e-01 4.17177938e-02 -6.98000133e-01
-5.75907588e-01 -9.68760014e-01 -7.76961684e-01 1.33768976e+00
4.12216596e-02 -1.11317372e+00 9.53195870e-01 1.40845865e-01
-4.15051579e-02 -5.09966731e-01 -1.01729846e+00 -1.35164952e+00
-3.76233131e-01 4.55052182e-02 6.79958403e-01 9.64615405e-01
-4.77524996e-01 8.80665332e-02 -5.12948930e-01 5.29652417e-01
1.24827003e+00 2.95148790e-01 1.11218369e+00 -1.62064409e+00
-2.39764228e-01 -5.10728955e-01 -6.17824137e-01 -1.45126545e+00
-9.76556167e-02 -9.36995625e-01 -4.27139737e-02 -1.32784140e+00
2.22052380e-01 -6.80875123e-01 -3.21571589e-01 5.24312437e-01
-2.22508743e-01 2.69496620e-01 4.97426003e-01 3.39599371e-01
-1.41727757e+00 6.98461890e-01 1.06613231e+00 -1.12259910e-01
-1.20232850e-02 -5.19588664e-02 -5.02716362e-01 4.69230771e-01
4.83970344e-01 -8.38594913e-01 1.21167704e-01 -3.79120499e-01
-4.85985018e-02 -3.02951157e-01 7.90443778e-01 -1.29677594e+00
8.87000024e-01 1.29881456e-01 1.54332846e-01 -8.37036967e-01
2.85180539e-01 -9.52312469e-01 3.95948112e-01 7.90337443e-01
1.81326628e-01 1.20738007e-01 1.99862376e-01 1.18714631e+00
1.42945483e-01 9.07385200e-02 7.38363326e-01 9.77257863e-02
-7.17692792e-01 6.79141104e-01 5.65311685e-02 -1.63822100e-01
1.08156955e+00 -3.63801122e-01 -5.67665458e-01 -6.32671863e-02
-4.74571049e-01 7.59638965e-01 7.48539031e-01 4.34132457e-01
2.34188974e-01 -1.77315569e+00 -4.64210033e-01 5.18261008e-02
8.06278735e-02 -3.14139903e-01 2.89927647e-02 1.23991668e+00
-2.95283794e-01 3.05382937e-01 4.32419032e-02 -8.32036853e-01
-1.36709535e+00 6.51307762e-01 2.28990003e-01 -4.90671962e-01
-1.00242901e+00 7.20939577e-01 -4.32386026e-02 -4.44852799e-01
6.58681214e-01 -1.96557581e-01 -5.06964512e-02 -1.70008605e-03
6.08149946e-01 4.33214903e-01 -3.74432623e-01 -6.05909646e-01
-6.74840271e-01 6.27642334e-01 -2.10500613e-01 3.86720777e-01
1.15454352e+00 -1.75477311e-01 2.63993740e-02 2.83264294e-02
8.59652102e-01 5.41794524e-02 -1.40158772e+00 -4.19244349e-01
1.24454051e-01 -7.06692219e-01 -1.96768999e-01 -2.03192905e-01
-1.23048532e+00 4.05561477e-01 7.63679147e-01 5.51501274e-01
6.83016062e-01 5.48448041e-03 1.13786674e+00 6.14323199e-01
5.64846098e-01 -8.68183136e-01 -8.88044834e-02 4.97405410e-01
1.33762911e-01 -1.19798696e+00 9.63872150e-02 -4.24113721e-01
3.79474945e-02 9.08557653e-01 7.87889421e-01 -9.02549773e-02
5.45538306e-01 3.35237980e-01 -2.64306396e-01 -5.01857817e-01
-6.07084334e-01 -6.62097871e-01 1.96619228e-01 6.56764865e-01
-3.92025299e-02 -3.11884224e-01 -1.61925748e-01 1.34594098e-01
3.70949596e-01 3.46248783e-02 1.06638402e-01 9.73845422e-01
-6.09931111e-01 -1.16743577e+00 -5.06547570e-01 3.91222805e-01
-3.93609613e-01 1.62706688e-01 -3.22905242e-01 1.04387236e+00
1.02685571e-01 5.35589755e-01 7.24529568e-03 -4.07914132e-01
2.89179176e-01 -2.76700795e-01 5.55600464e-01 -4.35863495e-01
-6.41241312e-01 -1.19787484e-01 2.27978546e-02 -8.25863242e-01
-6.26994610e-01 -5.72908521e-01 -1.19824958e+00 -6.55342340e-01
-7.95834780e-01 -7.18958303e-02 3.21632177e-01 6.98176205e-01
7.55384147e-01 6.02764249e-01 1.79858312e-01 -9.70012844e-01
-6.35725379e-01 -6.24517560e-01 -4.03780520e-01 3.40665698e-01
5.91585219e-01 -1.09294343e+00 -3.48312914e-01 -4.20369208e-01] | [6.310306549072266, -2.079286813735962] |
7bc66d44-fa3a-4487-9ca0-f7b6e284a9fb | constructing-topological-maps-using-markov | null | null | http://papers.nips.cc/paper/3861-constructing-topological-maps-using-markov-random-fields-and-loop-closure-detection | http://papers.nips.cc/paper/3861-constructing-topological-maps-using-markov-random-fields-and-loop-closure-detection.pdf | Constructing Topological Maps using Markov Random Fields and Loop-Closure Detection | We present a system which constructs a topological map of an environment given a sequence of images. This system includes a novel image similarity score which uses dynamic programming to match images using both the appearance and relative positions of local features simultaneously. Additionally an MRF is constructed to model the probability of loop-closures. A locally optimal labeling is found using Loopy-BP. Finally we outline a method to generate a topological map from loop closure data. Results are presented on four urban sequences and one indoor sequence. | ['Roy Anati', 'Kostas Daniilidis'] | 2009-12-01 | null | null | null | neurips-2009-12 | ['loop-closure-detection'] | ['computer-vision'] | [ 3.98044974e-01 -1.20682217e-01 3.13259698e-02 -8.11816931e-01
-7.25176513e-01 -5.00761271e-01 7.85593390e-01 4.75622296e-01
-2.47127026e-01 3.23352277e-01 1.35372654e-02 -2.28388920e-01
-4.30532247e-02 -8.06366444e-01 -9.21049833e-01 -2.43720114e-01
-4.75092769e-01 4.97019082e-01 9.13460553e-01 -2.72676855e-01
8.40339184e-01 6.63365364e-01 -1.71057832e+00 2.71914095e-01
5.97376347e-01 6.98611975e-01 9.80863214e-01 1.08903575e+00
-3.41061577e-02 5.46173036e-01 -1.06303081e-01 1.08705543e-01
6.34871960e-01 -3.64367843e-01 -8.83096874e-01 2.41685256e-01
1.04250264e+00 3.74147855e-03 -3.79599839e-01 1.14166081e+00
2.31961325e-01 5.17306268e-01 6.86707437e-01 -1.12068176e+00
-1.81231588e-01 -1.51533708e-01 -5.29521346e-01 3.97769481e-01
1.01720524e+00 -6.22908361e-02 8.49644303e-01 -8.28931570e-01
1.04087484e+00 1.28596628e+00 8.64412785e-01 -2.91060358e-01
-1.29805887e+00 -1.64449401e-02 -2.02565044e-01 5.67717552e-01
-1.47349954e+00 -2.95092702e-01 6.75116420e-01 -5.72267711e-01
1.18972838e+00 4.35638815e-01 7.70549178e-01 2.75515079e-01
5.02744734e-01 4.80951846e-01 1.20894253e+00 -8.07673216e-01
3.79468314e-02 -3.20821881e-01 1.86571653e-03 1.34675324e+00
-5.42179167e-01 2.73965985e-01 -2.87600994e-01 -2.87224382e-01
8.89522493e-01 -1.66440234e-01 1.49317607e-01 -8.25561881e-01
-1.23819804e+00 4.54810053e-01 7.30446041e-01 2.96128511e-01
-2.37389877e-01 2.74916530e-01 2.18524739e-01 3.42215061e-01
3.20747565e-03 3.47121954e-01 -3.92024145e-02 7.71825109e-03
-8.80375087e-01 1.82538226e-01 5.11607289e-01 9.74277794e-01
1.42669439e+00 -5.87201118e-01 1.24196500e-01 7.75918603e-01
2.17759967e-01 5.64481199e-01 1.32605121e-01 -1.29885340e+00
2.74565607e-01 4.69579428e-01 -7.63410563e-03 -1.82331967e+00
-3.26331198e-01 3.37591797e-01 -2.20030591e-01 3.57059062e-01
2.22074650e-02 5.87571084e-01 -9.75654364e-01 1.25265050e+00
3.32446754e-01 3.22493434e-01 -2.55497158e-01 5.67339182e-01
4.17191535e-01 8.02698910e-01 -4.06905830e-01 2.02422943e-02
9.86486852e-01 -1.24715436e+00 -4.00638103e-01 -2.91990787e-01
2.95046240e-01 -9.14252937e-01 6.59647942e-01 -1.99200734e-02
-8.63380373e-01 -8.66490781e-01 -1.05608904e+00 6.25967374e-03
-4.47337419e-01 -2.82873772e-02 7.60520920e-02 3.50394428e-01
-1.52810133e+00 8.08525026e-01 -6.61571026e-01 -8.93819153e-01
-2.49926940e-01 3.10779423e-01 -4.88694668e-01 4.16342057e-02
-6.48782074e-01 1.22979045e+00 6.81922078e-01 -2.86780357e-01
-5.39453924e-01 -1.07531205e-01 -1.09890568e+00 -3.59539598e-01
-2.16815680e-01 -4.68515426e-01 1.00275123e+00 -9.09888268e-01
-1.21370578e+00 1.14865398e+00 -4.61244345e-01 -4.40298259e-01
6.01242959e-01 3.29555988e-01 -3.21915835e-01 4.53383416e-01
4.74481910e-01 1.12204444e+00 6.66726828e-01 -1.49429679e+00
-8.45812380e-01 4.99966927e-02 -2.13989049e-01 5.68251014e-01
4.21460122e-01 8.94001052e-02 -5.75204492e-01 -4.08817828e-01
5.83481610e-01 -1.05271244e+00 -5.89196980e-01 2.29702413e-01
-1.77334979e-01 8.13723430e-02 9.82383490e-01 -9.47283745e-01
1.02269590e+00 -2.00411296e+00 1.54248970e-02 8.98315132e-01
-1.68890744e-01 -1.73480973e-01 -4.02777731e-01 5.20902693e-01
-2.56465018e-01 -1.09552100e-01 -3.69140595e-01 -1.14944629e-01
-3.22668046e-01 3.90356094e-01 -8.83980021e-02 6.12669706e-01
-2.60222435e-01 7.76923776e-01 -1.20562947e+00 -8.82901311e-01
7.60843635e-01 -2.25205332e-01 -3.69653076e-01 1.41310170e-01
-4.95115668e-02 2.12549001e-01 -7.76368156e-02 2.72088081e-01
7.27041006e-01 1.55472173e-03 1.47296950e-01 -5.12576811e-02
-5.52687228e-01 -6.19707145e-02 -1.00424528e+00 1.82863009e+00
-2.66088784e-01 9.81623650e-01 -3.18409592e-01 -8.90208244e-01
1.40641010e+00 -1.60422638e-01 5.15944660e-01 -7.95278966e-01
-2.23032981e-02 6.43087998e-02 -4.70834702e-01 -6.97228849e-01
8.34791005e-01 2.19338909e-01 -7.36083165e-02 4.58316654e-01
-2.34711230e-01 -4.64895099e-01 2.33321905e-01 1.42097592e-01
1.29005980e+00 4.05823886e-01 3.60648811e-01 -4.47005540e-01
6.13380730e-01 3.37896913e-01 1.47797942e-01 8.13437819e-01
-3.66634667e-01 8.84721220e-01 -1.10538550e-01 -1.00258207e+00
-1.42887044e+00 -1.27728724e+00 1.69371124e-02 8.20760548e-01
7.83660650e-01 -4.35223699e-01 -6.90824151e-01 -4.65080112e-01
-2.08228648e-01 3.08041900e-01 -6.72531188e-01 2.24219307e-01
-7.32087612e-01 -1.45281285e-01 2.22779214e-01 1.45075589e-01
6.65368617e-01 -1.05952787e+00 -9.83601272e-01 7.71432444e-02
-7.92246521e-01 -1.00414705e+00 -7.18523681e-01 1.07856058e-01
-7.26787925e-01 -1.07226837e+00 -3.14529449e-01 -1.41352415e+00
1.03597832e+00 4.76852268e-01 9.50232863e-01 1.97368205e-01
-5.00905275e-01 4.84454423e-01 -2.32378140e-01 1.60696745e-01
-7.71543384e-01 -2.58211911e-01 8.67530257e-02 -3.27846527e-01
3.05843726e-02 -4.56444412e-01 -4.36301738e-01 6.47310674e-01
-3.62897396e-01 1.75323933e-01 2.01724887e-01 5.48962295e-01
9.01762426e-01 3.97897065e-02 -2.97615439e-01 1.20671941e-02
3.30566376e-01 -9.75310579e-02 -7.50743926e-01 7.10536778e-01
-1.82625890e-01 2.73203045e-01 1.46765709e-01 -1.64490908e-01
-9.32900667e-01 6.15315914e-01 1.01882860e-01 -4.12814796e-01
-1.30616903e-01 1.07575640e-01 4.96080726e-01 -6.06038868e-01
9.67507064e-01 3.48998427e-01 1.32555559e-01 -2.33036712e-01
4.87631708e-01 5.16980231e-01 9.79125619e-01 -5.47443867e-01
6.67841613e-01 5.23811638e-01 1.57823209e-02 -1.02805221e+00
-2.46642694e-01 -9.29014146e-01 -1.13296783e+00 -7.03189611e-01
7.81790316e-01 -4.34569389e-01 -3.24506283e-01 3.35451007e-01
-1.28263378e+00 -1.92271441e-01 -6.39925599e-02 2.89714575e-01
-1.09445739e+00 8.42131674e-01 -2.79008090e-01 -4.22588319e-01
4.02804762e-02 -1.02103889e+00 9.20389116e-01 8.24437067e-02
-2.76526213e-01 -1.01923907e+00 8.04890215e-01 9.73185990e-03
-5.76726533e-02 4.55306470e-01 4.65739459e-01 1.62322968e-02
-8.20666134e-01 4.18635458e-02 -3.15023094e-01 -2.72492263e-02
1.14260770e-01 3.73428255e-01 -5.73134959e-01 -1.87867969e-01
-3.91051620e-01 2.66671516e-02 5.64331055e-01 5.65330446e-01
7.13031828e-01 -1.64137527e-01 -8.24217677e-01 5.00038624e-01
1.62974858e+00 4.71046597e-01 7.88801134e-01 7.11490810e-01
4.78408128e-01 7.62799799e-01 8.59513700e-01 1.06021844e-01
3.41301084e-01 9.01843190e-01 2.29524389e-01 -1.51545005e-02
-1.59388229e-01 -5.81153035e-01 1.70213655e-01 8.65466237e-01
5.58478087e-02 1.64576590e-01 -1.29049397e+00 6.53248787e-01
-2.00858092e+00 -1.25103235e+00 -2.93773979e-01 1.95951426e+00
3.94253820e-01 -9.09467414e-02 3.85429114e-02 -8.09310749e-02
1.22884595e+00 1.20212160e-01 -8.14160928e-02 -6.06033564e-01
-8.12844634e-02 -2.62562037e-01 8.13771129e-01 9.63375330e-01
-1.34518027e+00 8.96652997e-01 8.42389297e+00 6.40868664e-01
-7.38477468e-01 -1.58873618e-01 4.45565253e-01 6.77843153e-01
-1.97438627e-01 1.98635668e-01 -2.44966418e-01 3.25475693e-01
5.28463662e-01 -4.22813930e-02 4.26596522e-01 7.58078933e-01
1.37655824e-01 -6.44472182e-01 -6.83193445e-01 1.10172510e+00
1.70803815e-01 -1.45557368e+00 -1.56851083e-01 -5.06172015e-04
9.32464898e-01 4.82532948e-01 -1.99275106e-01 -4.57115918e-01
4.75942135e-01 -5.41092277e-01 9.95313287e-01 7.86046326e-01
6.93838656e-01 -5.75311184e-01 1.20622672e-01 3.36426467e-01
-1.64477575e+00 5.11111915e-02 -3.83304358e-01 -7.02579096e-02
2.55817890e-01 1.03608303e-01 -1.14465296e+00 5.35987556e-01
8.90132546e-01 7.59751856e-01 -9.43269193e-01 1.54293275e+00
3.02954167e-02 -3.15028399e-01 -5.25461674e-01 7.75327766e-03
3.13943893e-01 -5.34395576e-01 6.74995899e-01 1.46259916e+00
5.17385840e-01 -1.24107338e-01 7.48027146e-01 4.47856009e-01
6.92993164e-01 2.07723349e-01 -1.12466776e+00 7.52966404e-01
4.46047455e-01 1.10975373e+00 -1.26133907e+00 -3.70048225e-01
-4.52966727e-02 1.25232852e+00 3.13670188e-01 1.76538415e-02
-5.88620842e-01 -4.65554595e-01 3.25600386e-01 3.58665250e-02
2.70780593e-01 -4.95770127e-01 -8.44093710e-02 -6.52178466e-01
-4.91015129e-02 -3.81554693e-01 2.07423508e-01 -1.11951220e+00
-8.22680652e-01 5.44852436e-01 3.63688976e-01 -1.37544119e+00
-3.73965353e-01 -2.92282820e-01 -6.49425566e-01 4.87624735e-01
-1.06954837e+00 -8.33621442e-01 -3.86460781e-01 3.82818639e-01
4.50815737e-01 1.30038522e-02 6.65997148e-01 5.75277209e-02
9.82377492e-03 1.94539111e-02 2.50203878e-01 -9.39173717e-03
5.10968685e-01 -1.21017575e+00 8.79790425e-01 9.55496490e-01
1.29338577e-01 3.84145856e-01 9.72695172e-01 -7.84952939e-01
-5.71700573e-01 -9.71608579e-01 1.03738678e+00 -5.42424023e-01
4.94950175e-01 -1.52159065e-01 -7.17624366e-01 6.55404866e-01
2.10976288e-01 7.92141929e-02 8.59327242e-02 -5.66688895e-01
-2.58462399e-01 1.91943690e-01 -1.43454754e+00 5.41731298e-01
1.36717343e+00 -7.54689634e-01 -7.48905540e-01 2.35302180e-01
3.67837518e-01 -6.41233802e-01 -5.64753354e-01 2.33295709e-01
6.44013941e-01 -1.19393373e+00 1.13265347e+00 1.99806362e-01
5.35878055e-02 -8.40036571e-01 -3.29588771e-01 -1.21984303e+00
-6.49049938e-01 -5.89908004e-01 5.35671651e-01 5.89097440e-01
1.13031678e-01 -2.73973227e-01 7.44752169e-01 4.25408259e-02
-7.54224360e-02 -3.02708417e-01 -1.04721403e+00 -1.04116011e+00
-5.34098744e-01 -1.92521751e-01 3.29735339e-01 7.67092109e-01
2.70166099e-01 3.01579703e-02 -1.72565445e-01 2.30357036e-01
7.33746827e-01 1.18271217e-01 6.80660009e-01 -9.73460615e-01
6.33928478e-02 -2.89664060e-01 -1.01384032e+00 -1.11523211e+00
1.38572186e-01 -9.70923841e-01 6.20089054e-01 -1.59838998e+00
2.72417545e-01 -6.82821214e-01 6.13419451e-02 2.22826496e-01
2.23255277e-01 3.76717329e-01 1.28177851e-01 2.55390346e-01
-8.56613457e-01 2.56975859e-01 9.48598921e-01 -8.17391947e-02
-1.86425507e-01 -4.66447711e-01 3.12137932e-01 7.70957172e-01
1.02514553e+00 -4.40757215e-01 -3.26524645e-01 -1.13518976e-01
7.91690797e-02 1.46055445e-01 3.21893662e-01 -1.26348519e+00
3.73278737e-01 -3.37192714e-01 2.50331670e-01 -1.09488618e+00
3.45569193e-01 -9.07948792e-01 6.18233800e-01 7.85498083e-01
-2.50538409e-01 8.67147207e-01 2.45342683e-02 8.06317866e-01
-1.23418979e-01 -4.54258412e-01 9.83992636e-01 -2.94638366e-01
-1.43761671e+00 -3.49331484e-03 -5.36162734e-01 -5.33675015e-01
1.44731748e+00 -7.46523559e-01 1.18560297e-02 -3.27389687e-01
-6.84071124e-01 3.62394899e-01 1.06696808e+00 5.95158875e-01
9.97817695e-01 -1.59949625e+00 -4.70137477e-01 4.59512889e-01
4.40029830e-01 -4.70542520e-01 7.96192512e-02 4.87393886e-01
-1.38510740e+00 1.91431269e-01 -5.92675388e-01 -9.42265987e-01
-1.86886775e+00 6.29738510e-01 5.79164684e-01 3.56978877e-03
-7.37345815e-01 4.99052495e-01 -5.14837168e-02 -8.21933210e-01
5.78761883e-02 -2.68771261e-01 -1.10452967e-02 -3.48633051e-01
4.22815025e-01 6.50701582e-01 -1.18131317e-01 -1.14630139e+00
-5.50340593e-01 1.07932436e+00 3.44985574e-01 -4.90450591e-01
8.34765553e-01 -6.73447013e-01 -3.33391070e-01 3.37785810e-01
1.53166795e+00 -3.72722864e-01 -1.28446126e+00 -2.47881934e-01
2.04004705e-01 -8.93681407e-01 -3.30680698e-01 -3.70204180e-01
-4.78292346e-01 3.89463007e-01 1.05445957e+00 5.79326004e-02
9.12745297e-01 -1.08688092e-02 4.41852897e-01 6.64021492e-01
6.57278180e-01 -1.17704809e+00 3.82234275e-01 7.90442348e-01
7.98204005e-01 -1.07776451e+00 -3.10280025e-02 -3.87496501e-01
-3.15405756e-01 1.19686270e+00 1.83895350e-01 -2.92541593e-01
6.66714370e-01 1.77927092e-01 5.01551405e-02 -2.39158928e-01
-1.93544984e-01 -1.73644423e-01 1.02133043e-01 9.25485492e-01
-1.46049008e-01 9.51061845e-02 -9.22025144e-02 -1.00001979e+00
-2.95956194e-01 -2.78965414e-01 5.18842041e-01 1.14756250e+00
-1.01756096e+00 -1.05927527e+00 -7.97565460e-01 1.35944009e-01
8.45940709e-02 1.81604177e-01 -2.52778322e-01 3.25077921e-01
1.51358768e-01 8.44577730e-01 3.96475017e-01 -6.72925115e-01
1.38604924e-01 -2.50278294e-01 8.62670422e-01 -4.34928745e-01
-1.82419732e-01 -3.27259928e-01 3.97428535e-02 -7.01853514e-01
-6.57049060e-01 -8.70287597e-01 -1.29513204e+00 -3.31149161e-01
-2.08661139e-01 1.14611080e-02 7.46387541e-01 5.11774778e-01
2.36770779e-01 -1.88115001e-01 9.34040725e-01 -1.21525908e+00
1.90114230e-01 -3.50746155e-01 -3.67866367e-01 2.68784732e-01
3.87407482e-01 -4.85171586e-01 3.51996347e-02 4.33485806e-01] | [7.556637287139893, -2.052532911300659] |
05c06649-0d35-446f-a3c0-f35284d650dd | detection-and-tracking-of-fingertips-for | null | null | https://ieeexplore.ieee.org/document/9035256 | https://ieeexplore.ieee.org/document/9035256 | Detection and tracking of fingertips for geometric transformation of objects in virtual environment | This paper presents an approach of two-stage convolutional neural network (CNN) for detection of fingertips so that an interaction of the fingertips with a 3D object in the virtual environment (VR) can be established. The first-stage CNN is assigned to detect and locate the hand. Subsequently, the detected hand is cropped, resized, and fed to the second stage CNN for predicting the coordinates of fingertips. Next, a tracker is employed to track the hand continuously so that the system becomes reliable in real-time performance. The VR environments are designed to demonstrate the performance of the fingertip-based interaction system. The proposed method focuses on the geometric transformation of a virtual 3D object by using the gesture of the thumb and index finger. In particular, the distance of the thumb and index fingertips is employed to scale a 3D object in virtual environment. To realize the system, a dataset of 1000 images, named, Thumb Index 1000 (TI1K) dataset, is developed including those variations that are commonly-seen in real-life thumb and index fingers. The system is evaluated with the aid of a number of participants and virtual objects that are distinctive in nature. The proposed approach attains the desired goal and performs in real-time seamlessly to facilitate the human-computer interaction in the VR environment. | ['S. M. Mahbubur Rahman', 'Mohammad Mahmudul Alam'] | 2020-03-16 | null | null | null | null | ['hand-detection', 'fingertip-detection'] | ['computer-vision', 'computer-vision'] | [-7.27476925e-02 -5.11153638e-01 1.97114497e-01 6.48712516e-02
7.27698207e-02 -7.38960147e-01 3.85744214e-01 -4.85963702e-01
-5.01980543e-01 -9.36711952e-02 -3.60108197e-01 -1.72448635e-01
-3.00641265e-03 -5.11369944e-01 -4.30032104e-01 -3.59283894e-01
3.44233550e-02 -4.59905155e-02 4.42183167e-01 -7.32447580e-02
6.12297297e-01 1.15924549e+00 -1.63793695e+00 7.08260983e-02
1.73726916e-01 1.14740324e+00 6.26854599e-01 8.39113891e-01
1.31842732e-01 -5.59217073e-02 -6.57675922e-01 1.06223918e-01
6.66326821e-01 2.99848169e-01 -1.93755671e-01 -1.88332617e-01
4.74140286e-01 -6.99007213e-01 -5.93343496e-01 8.80772293e-01
8.57534170e-01 2.90722966e-01 4.74468619e-01 -1.15187931e+00
-6.02486193e-01 1.26436353e-01 -8.54217410e-01 2.50932649e-02
4.27933693e-01 2.63207197e-01 4.40021902e-01 -1.11004639e+00
5.43664992e-01 1.45717692e+00 3.68371516e-01 4.28118080e-01
-3.61037076e-01 -7.94053793e-01 7.32076541e-02 -1.04190886e-01
-1.65492630e+00 -6.26814663e-02 1.12012780e+00 -5.62765837e-01
8.17681253e-01 3.31093639e-01 7.13412642e-01 9.64298725e-01
4.75521922e-01 8.50113153e-01 4.40811843e-01 -3.87750000e-01
-9.76066738e-02 1.01224519e-01 2.36686870e-01 6.09944880e-01
7.26372972e-02 3.05637747e-01 -1.07835874e-01 8.67741704e-02
1.72311330e+00 3.75805110e-01 -3.41376960e-02 -1.65901095e-01
-1.39262688e+00 8.32064152e-02 8.18607330e-01 3.68601412e-01
-6.38846636e-01 2.18675286e-02 3.46785218e-01 -1.12294853e-01
-5.40875606e-02 1.33783177e-01 -2.67695010e-01 -1.87732697e-01
-6.47763550e-01 2.67983109e-01 5.01548827e-01 1.12989068e+00
-1.41115814e-01 -7.48336688e-02 -4.17910784e-01 4.68184620e-01
6.13404214e-01 5.22827387e-01 1.10493623e-01 -7.40869462e-01
6.68263316e-01 7.09504902e-01 6.08898699e-01 -1.24677229e+00
-3.79304647e-01 -2.09604397e-01 -7.53318369e-01 7.67925143e-01
5.14557779e-01 -1.23847658e-02 -1.03306329e+00 1.22993827e+00
4.41969037e-01 5.92615679e-02 -3.86647940e-01 1.31899905e+00
9.51776624e-01 3.18549454e-01 -1.43796861e-01 3.57229739e-01
1.48491240e+00 -5.79257488e-01 -6.42786503e-01 4.05052043e-02
-1.72303468e-01 -8.74459326e-01 1.29566157e+00 4.14296865e-01
-9.33952808e-01 -1.31393051e+00 -9.91626263e-01 1.14466712e-01
-4.45559144e-01 7.62438774e-01 4.58392411e-01 4.29074496e-01
-6.35177433e-01 3.70316058e-01 -7.48448849e-01 -3.08861107e-01
1.85936466e-01 5.60588300e-01 -2.69423813e-01 4.22137886e-01
-9.72384751e-01 6.91689014e-01 2.53534496e-01 7.58161068e-01
-4.12984133e-01 -2.14442000e-01 -4.97437805e-01 -2.63099223e-02
3.96467596e-02 -3.76134366e-01 1.00157559e+00 -6.57711565e-01
-1.48236310e+00 7.61586905e-01 -5.00173448e-03 3.22293013e-01
9.68662500e-01 -3.49994987e-01 -4.33230162e-01 -1.55291826e-01
-1.32792830e-01 2.75848120e-01 1.14854574e+00 -1.16288388e+00
-7.50637949e-01 -7.81901538e-01 6.37463108e-02 2.24433079e-01
-1.30073130e-01 2.57815123e-01 -8.74926507e-01 -5.76242685e-01
2.20857456e-01 -8.85372639e-01 2.63082325e-01 2.69616514e-01
-8.32737446e-01 -5.32809079e-01 1.02484834e+00 -6.63099170e-01
9.92324829e-01 -2.26427746e+00 -4.12961803e-02 4.50632364e-01
3.88212025e-01 6.78097308e-01 -1.48338243e-01 8.18629656e-03
-3.54707986e-02 -2.11861148e-01 5.31771481e-01 -1.72044188e-01
-1.83075160e-01 -5.24562001e-01 -6.84515312e-02 2.53831565e-01
8.68960563e-03 8.83182764e-01 -7.00534046e-01 -3.72194350e-01
7.10124314e-01 7.87489116e-01 1.96461171e-01 3.83661985e-01
1.47980839e-01 4.39202309e-01 -6.41244173e-01 8.98949623e-01
7.96076298e-01 -3.69260646e-02 -1.86787888e-01 -5.07735729e-01
-4.07024562e-01 -3.60264510e-01 -1.61001813e+00 1.33316803e+00
-3.28366339e-01 6.28772855e-01 6.95393682e-02 -1.61351904e-01
1.16627181e+00 2.35258937e-01 2.04494983e-01 -5.16988456e-01
5.24419308e-01 2.56021786e-02 3.07227299e-02 -7.24353731e-01
5.42622745e-01 1.01942801e+00 2.08242789e-01 4.34941590e-01
-5.63066840e-01 2.99509376e-01 -2.78009146e-01 -2.16313213e-01
4.98190761e-01 2.92776823e-01 7.60674775e-02 2.50964433e-01
4.30434614e-01 -3.69762301e-01 -5.98948635e-02 8.11569631e-01
-2.97845989e-01 6.00795507e-01 -8.22892785e-02 -5.40713489e-01
-1.09762585e+00 -1.19633937e+00 1.17108170e-02 7.80362964e-01
3.90082300e-01 2.98890382e-01 -6.55021906e-01 -3.60065699e-01
3.31896752e-01 3.50728370e-02 -6.64708614e-01 1.32617891e-01
-7.33845294e-01 1.35685325e-01 3.24226201e-01 8.44756603e-01
7.44634092e-01 -1.50162661e+00 -1.13739300e+00 2.39670426e-02
4.09258962e-01 -1.00632191e+00 -6.33705497e-01 -4.36891317e-01
-5.89077234e-01 -1.26542628e+00 -1.09914458e+00 -1.06785345e+00
6.62202299e-01 4.20049727e-01 2.55950958e-01 -8.30100551e-02
-4.34585780e-01 1.95694968e-01 -1.14330687e-01 -4.65890497e-01
1.97971195e-01 1.47431999e-01 2.98032075e-01 1.42673582e-01
3.89060050e-01 -1.97543457e-01 -9.35484827e-01 3.56593758e-01
-4.11705285e-01 -1.54456005e-01 4.90100801e-01 2.51382679e-01
4.27483052e-01 -6.90456554e-02 2.01927543e-01 4.36200984e-02
8.93001139e-01 5.51361516e-02 -5.69333553e-01 2.66361505e-01
6.33171648e-02 -4.23839152e-01 5.17495751e-01 -9.96703088e-01
-8.11202466e-01 4.65016633e-01 7.18352124e-02 -6.76207840e-01
-2.90092587e-01 1.95138417e-02 -2.06818849e-01 -3.27150822e-01
5.22852182e-01 7.22293556e-02 -1.36748150e-01 -6.66307628e-01
4.60806847e-01 1.37455058e+00 8.91704917e-01 -3.49308103e-02
7.63866961e-01 2.59081215e-01 -1.70925960e-01 -8.83621454e-01
4.60411888e-03 -3.32735777e-01 -1.16512978e+00 -6.41631603e-01
7.60602832e-01 -5.23979962e-01 -1.33734965e+00 9.60488021e-01
-1.61238873e+00 -4.62048629e-04 2.16921464e-01 6.00446045e-01
-1.52355954e-01 1.95122376e-01 -5.37646651e-01 -1.16508102e+00
-7.29910612e-01 -1.05892086e+00 1.15190947e+00 6.72154427e-01
-1.34632766e-01 -4.66960073e-01 -3.57808679e-01 -3.71992350e-01
1.51589528e-01 3.53828490e-01 5.53911805e-01 -3.40684593e-01
-4.74152625e-01 -9.29982126e-01 -5.10491431e-01 -4.69948836e-02
5.82515597e-01 5.76672375e-01 -1.00548506e+00 -2.82681912e-01
-3.93587053e-01 2.38123298e-01 1.99934885e-01 5.70280671e-01
1.24585557e+00 3.21806669e-01 -4.96319056e-01 4.43742216e-01
1.04919612e+00 7.17310369e-01 4.38756317e-01 1.99871927e-01
1.07534122e+00 4.81131732e-01 8.65304589e-01 4.85024959e-01
1.58783376e-01 7.54077077e-01 4.53239948e-01 -1.98909685e-01
6.03168234e-02 -1.77802980e-01 -8.99347439e-02 4.59245354e-01
-6.98933482e-01 -2.88950596e-02 -5.75672925e-01 1.65186346e-01
-1.43074751e+00 -5.80727160e-01 -1.54949307e-01 2.45507789e+00
2.64387071e-01 2.06575304e-01 2.31118232e-01 2.63960719e-01
1.08561838e+00 -1.29495978e-01 -9.61955786e-01 -1.14207774e-01
3.56307149e-01 2.43029937e-01 2.67342836e-01 2.31924266e-01
-1.06947613e+00 1.09373784e+00 5.75913286e+00 1.85567737e-01
-1.56618810e+00 -5.61903238e-01 3.82827371e-02 -1.48309926e-02
3.81809264e-01 -5.90582430e-01 -8.25082779e-01 5.11216879e-01
-3.04678604e-02 2.95410991e-01 6.37903094e-01 8.29390824e-01
4.70213711e-01 2.09417224e-01 -1.07575428e+00 1.32571256e+00
-3.16202730e-01 -7.41592586e-01 -1.46827981e-01 -1.17578596e-01
9.76191387e-02 -2.32530847e-01 4.20643300e-01 -1.54643536e-01
-3.21699262e-01 -8.98175538e-01 7.99477279e-01 8.98165822e-01
1.21941674e+00 -7.41381407e-01 4.45365191e-01 3.48096520e-01
-1.69745958e+00 -2.67630070e-01 -2.98358202e-01 -7.43524358e-02
-1.04021609e-01 7.11243749e-02 -9.09366906e-01 1.44304335e-01
8.21826994e-01 3.22578311e-01 -4.53029543e-01 1.12871861e+00
-8.14468861e-02 -1.76209256e-01 -2.82811433e-01 -4.41141814e-01
-2.69286521e-02 2.36036032e-02 6.01141870e-01 9.80408609e-01
3.91158462e-01 5.77458441e-02 -7.64312670e-02 1.07959187e+00
1.08522817e-01 2.01366115e-02 -5.82480133e-01 -1.02996342e-01
8.67636144e-01 1.38644660e+00 -8.24823260e-01 -2.54510820e-01
-6.36856556e-02 1.20788157e+00 1.79533996e-02 5.63014030e-01
-6.00798070e-01 -1.17385459e+00 4.74915504e-01 -5.92023991e-02
2.66223967e-01 -5.50505638e-01 -2.27852151e-01 -8.24804962e-01
5.41065693e-01 -4.90728945e-01 -1.08781070e-01 -1.09999597e+00
-8.48660171e-01 7.34579027e-01 -4.81977642e-01 -1.25683069e+00
-2.48767629e-01 -9.02475953e-01 -7.57454455e-01 1.34741688e+00
-8.71283770e-01 -1.22271621e+00 -1.09530330e+00 7.52836168e-01
4.20466453e-01 -2.96048880e-01 6.13117337e-01 1.16407961e-01
-5.83470464e-01 7.99255848e-01 -4.57057171e-02 5.31649411e-01
7.08060384e-01 -9.04285431e-01 1.09055018e+00 5.49656928e-01
-1.64869949e-01 9.92784142e-01 1.72730669e-01 -8.79053652e-01
-1.62620986e+00 -7.70614505e-01 4.91559803e-01 -7.38245189e-01
1.11421764e-01 -6.13170624e-01 -6.40335441e-01 6.96442127e-01
-5.22538245e-01 4.26414199e-02 -4.75738421e-02 -2.68933564e-01
-2.54118860e-01 -7.73170441e-02 -1.22432077e+00 8.32538068e-01
1.11423361e+00 -6.75274312e-01 -4.90891725e-01 -3.87541056e-02
5.77314794e-01 -7.80238211e-01 -8.19603622e-01 1.40376851e-01
1.44819820e+00 -5.92472374e-01 1.27941453e+00 -4.78623152e-01
2.62984708e-02 -3.75753403e-01 7.13754520e-02 -8.74151528e-01
-4.95314032e-01 -4.07613903e-01 -3.71068299e-01 8.99153173e-01
-2.21543953e-01 -4.29312438e-01 7.68261194e-01 7.25138783e-01
3.44984174e-01 -5.29749036e-01 -6.99847400e-01 -4.32245910e-01
-4.15508002e-01 -3.91196609e-01 7.73888648e-01 6.21197283e-01
-2.17353269e-01 -1.58142388e-01 -1.23626150e-01 3.92075717e-01
5.48302710e-01 -2.34238636e-02 9.32542264e-01 -1.40472531e+00
1.71161488e-01 -4.86759633e-01 -5.28622150e-01 -1.25831747e+00
-4.05665040e-01 -4.79911625e-01 -3.05648744e-01 -1.23587322e+00
-1.59741059e-01 -4.17517692e-01 -5.23156524e-01 2.73962259e-01
-1.67416260e-01 1.49572670e-01 3.12209636e-01 3.20367485e-01
-9.44171194e-03 -7.88778216e-02 1.56846464e+00 -1.98100656e-02
-9.08575535e-01 5.71609020e-01 -1.86849371e-01 5.13873935e-01
6.57059491e-01 3.34372848e-01 -1.24242850e-01 -2.92461962e-01
-3.10675144e-01 2.60678113e-01 8.58000696e-01 -7.98362017e-01
3.61311853e-01 5.63791990e-02 1.03929341e+00 -1.03196001e+00
3.20412815e-01 -8.18190098e-01 -8.46348628e-02 6.25942767e-01
-3.32989305e-01 1.58265457e-01 8.74736756e-02 3.33962068e-02
3.84156555e-01 2.36709509e-03 5.00864923e-01 -2.31681332e-01
-7.61832595e-01 5.80454051e-01 -6.66907802e-02 -6.30871952e-01
1.03821421e+00 -5.99493086e-01 9.56157222e-02 -1.33465454e-01
-7.16134548e-01 -8.52832273e-02 1.83760583e-01 9.15494025e-01
1.08395720e+00 -1.44271123e+00 -3.79124820e-01 5.84376931e-01
4.30498905e-02 3.12778875e-02 2.99527586e-01 3.68763924e-01
-5.34953296e-01 5.09825885e-01 -5.07653713e-01 -6.13545716e-01
-1.52143693e+00 6.23918295e-01 5.81844449e-01 5.41650951e-01
-8.10879529e-01 7.17595935e-01 -2.06965387e-01 -4.07751143e-01
8.45269561e-01 -5.87746680e-01 -5.29100060e-01 -2.11638987e-01
7.45399594e-01 4.67638791e-01 -3.04158360e-01 -4.56597924e-01
-3.86366904e-01 9.94491041e-01 7.97351729e-03 -1.54453274e-02
8.63675296e-01 1.28542185e-01 1.27063885e-01 2.98145026e-01
1.11019623e+00 -7.61834159e-02 -1.52812040e+00 -2.67907649e-01
-3.54184479e-01 -7.86788285e-01 -1.81876302e-01 -8.30044150e-01
-1.11136627e+00 1.07928133e+00 1.28683829e+00 1.55166788e-02
6.97846889e-01 -2.84378380e-01 7.70528257e-01 4.62796450e-01
4.25087541e-01 -9.64039624e-01 5.02639636e-02 3.53086144e-01
1.09274554e+00 -1.06276631e+00 -4.04907912e-01 -1.22082166e-01
-3.25071335e-01 1.43522596e+00 8.97898734e-01 -2.50976205e-01
6.03862166e-01 2.07876742e-01 1.18848704e-01 -2.31842235e-01
2.46618807e-01 8.44980851e-02 6.59743071e-01 7.62935102e-01
1.63625956e-01 2.25469559e-01 2.66302824e-01 6.02920651e-01
-2.36564577e-01 2.32383981e-01 -8.52871686e-02 9.66771424e-01
-3.47576320e-01 -4.63589400e-01 -4.96153772e-01 2.80555338e-01
-1.41645819e-01 1.24325179e-01 -5.23418844e-01 8.66320133e-01
1.22137263e-01 7.08053470e-01 3.27844560e-01 -9.35212314e-01
6.49528861e-01 -2.80240834e-01 7.64960468e-01 -1.40907258e-01
-5.83575308e-01 -5.79060949e-02 -6.44567907e-01 -5.83020389e-01
6.70262724e-02 -3.00620407e-01 -1.34875059e+00 -1.43538058e-01
-1.84327632e-01 -4.34203148e-01 1.04082382e+00 7.16160595e-01
3.34876120e-01 5.44357181e-01 9.31448758e-01 -1.60903633e+00
-6.18289888e-01 -9.72824872e-01 -7.64851749e-01 8.08781236e-02
3.07675272e-01 -7.96603978e-01 -4.45119515e-02 -4.14737195e-01] | [6.476024627685547, -0.40534862875938416] |
f7bb8058-721b-49ac-b9c4-f63709270769 | improving-robustness-and-accuracy-via | 2107.13994 | null | https://arxiv.org/abs/2107.13994v1 | https://arxiv.org/pdf/2107.13994v1.pdf | Improving Robustness and Accuracy via Relative Information Encoding in 3D Human Pose Estimation | Most of the existing 3D human pose estimation approaches mainly focus on predicting 3D positional relationships between the root joint and other human joints (local motion) instead of the overall trajectory of the human body (global motion). Despite the great progress achieved by these approaches, they are not robust to global motion, and lack the ability to accurately predict local motion with a small movement range. To alleviate these two problems, we propose a relative information encoding method that yields positional and temporal enhanced representations. Firstly, we encode positional information by utilizing relative coordinates of 2D poses to enhance the consistency between the input and output distribution. The same posture with different absolute 2D positions can be mapped to a common representation. It is beneficial to resist the interference of global motion on the prediction results. Second, we encode temporal information by establishing the connection between the current pose and other poses of the same person within a period of time. More attention will be paid to the movement changes before and after the current pose, resulting in better prediction performance on local motion with a small movement range. The ablation studies validate the effectiveness of the proposed relative information encoding method. Besides, we introduce a multi-stage optimization method to the whole framework to further exploit the positional and temporal enhanced representations. Our method outperforms state-of-the-art methods on two public datasets. Code is available at https://github.com/paTRICK-swk/Pose3D-RIE. | ['Wen Gao', 'Xinfeng Zhang', 'Shanshe Wang', 'Haopeng Lu', 'Wenkang Shan'] | 2021-07-29 | null | null | null | null | ['monocular-3d-human-pose-estimation'] | ['computer-vision'] | [-9.63815376e-02 -2.72963941e-01 -3.74867648e-01 -1.90284416e-01
-3.80946904e-01 -2.27122039e-01 3.75070602e-01 -1.87159926e-01
-3.22147489e-01 5.08708775e-01 6.06637955e-01 4.45968568e-01
-1.38090760e-01 -6.77935481e-01 -5.01796842e-01 -5.49202323e-01
-1.79600000e-01 4.00748253e-01 3.71898234e-01 -3.86858702e-01
9.12649790e-04 3.84566575e-01 -1.37481964e+00 5.71091212e-02
5.16664207e-01 7.98769116e-01 2.51626790e-01 3.09548229e-01
4.26352650e-01 2.38010615e-01 -6.78102672e-01 -1.35342240e-01
3.68713498e-01 -5.43637753e-01 -5.89945257e-01 -8.14942736e-03
3.15611780e-01 -3.41679394e-01 -6.17914617e-01 6.65698826e-01
8.83610785e-01 4.53402638e-01 3.95565629e-01 -1.13737953e+00
-2.84136564e-01 5.60896359e-02 -7.03645945e-01 1.10682607e-01
8.32361341e-01 2.07835257e-01 7.00324774e-01 -7.55964935e-01
8.08835983e-01 1.22422647e+00 7.60811627e-01 4.70953912e-01
-9.53778744e-01 -6.42465115e-01 3.84688020e-01 3.57463896e-01
-1.55033314e+00 -1.12272002e-01 1.04954720e+00 -4.25671101e-01
7.51928329e-01 3.46761197e-01 1.02109885e+00 1.17577493e+00
5.14944196e-01 8.78962994e-01 5.45173764e-01 -1.40207678e-01
-1.95427045e-01 -4.66861784e-01 -6.67767748e-02 6.15932703e-01
6.50707409e-02 1.67530160e-02 -7.13283062e-01 3.73684689e-02
8.26768637e-01 2.33440742e-01 -4.61774439e-01 -4.81751382e-01
-1.53178310e+00 5.29328227e-01 6.57633245e-01 1.94739670e-01
-5.02678752e-01 2.78611332e-01 3.53460819e-01 -1.33530691e-01
3.33368808e-01 1.84457123e-01 -5.57260096e-01 -4.10004139e-01
-7.66167819e-01 5.91268599e-01 4.61327851e-01 8.39096725e-01
4.76647735e-01 -2.85669893e-01 -3.21583360e-01 6.97442651e-01
4.16087359e-01 3.79708171e-01 4.99545425e-01 -9.55545068e-01
8.06138992e-01 5.98900080e-01 1.52468622e-01 -1.54546797e+00
-8.21957886e-01 -5.18779933e-01 -8.32859695e-01 -1.70957744e-01
4.91596162e-01 -1.07277125e-01 -7.66015708e-01 1.88046205e+00
4.86104935e-01 -3.93068127e-04 -4.03147399e-01 1.22537780e+00
4.68268037e-01 6.39132500e-01 -9.83327627e-02 -3.67147103e-02
1.31952655e+00 -9.36913788e-01 -7.36307800e-01 -5.41300237e-01
5.02957225e-01 -6.88813210e-01 6.86079800e-01 1.16355166e-01
-9.16796327e-01 -8.52679133e-01 -1.04419219e+00 -1.53352479e-02
8.03381298e-03 3.65316927e-01 4.66312349e-01 3.94230574e-01
-4.92772132e-01 7.65161097e-01 -1.14658999e+00 -2.91637987e-01
-1.24898806e-01 4.48461294e-01 -5.07238805e-01 1.92201845e-02
-1.49201548e+00 9.56369281e-01 1.19383536e-01 6.06136203e-01
-2.05482960e-01 -4.19381976e-01 -1.00059414e+00 -4.55109060e-01
3.23850185e-01 -8.22819412e-01 9.29041386e-01 -4.11579281e-01
-1.44395113e+00 4.34780806e-01 -3.47132802e-01 -3.89097370e-02
8.34114790e-01 -8.65665495e-01 -1.14999115e-01 5.38667962e-02
1.37357146e-01 6.28124595e-01 5.92869043e-01 -1.03904521e+00
-2.99589187e-01 -5.86051941e-01 -2.61134416e-01 5.80202043e-01
-1.13192469e-01 -3.69862229e-01 -1.10164297e+00 -1.07809794e+00
5.84141135e-01 -1.25809312e+00 -3.43434095e-01 1.50399119e-01
-4.63700861e-01 -7.80648515e-02 5.95562577e-01 -9.27915394e-01
1.46613717e+00 -2.01867080e+00 7.03291416e-01 2.54440635e-01
-1.69068441e-01 9.63596329e-02 2.35762708e-02 4.41097200e-01
7.34801590e-02 -1.81530431e-01 1.10777107e-03 -3.79026324e-01
-1.02364026e-01 2.86650211e-01 -7.19433278e-03 6.52875125e-01
-3.86439487e-02 8.28257620e-01 -8.34664941e-01 -4.53271061e-01
3.09743822e-01 6.93729639e-01 -5.10434210e-01 1.82098046e-01
1.48127377e-01 8.50151598e-01 -6.56878054e-01 4.77686435e-01
4.63953823e-01 -7.27728903e-02 2.33602345e-01 -5.74514091e-01
2.18448162e-01 3.51014853e-01 -1.52942896e+00 2.08388066e+00
-1.01452149e-01 1.73301861e-01 -3.44381124e-01 -6.69728637e-01
9.99619782e-01 3.41239035e-01 9.23216701e-01 -6.16207719e-01
1.77235588e-01 7.36747086e-02 -1.13165267e-01 -3.26498240e-01
4.21744645e-01 2.14947611e-01 -2.29068384e-01 2.98283219e-01
-2.70194650e-01 2.16586053e-01 3.01656127e-02 -1.85295627e-01
7.57473052e-01 6.82474315e-01 4.33846056e-01 2.08978951e-01
3.53843659e-01 -2.44451240e-01 9.25869048e-01 3.48912954e-01
-4.44142163e-01 9.38552976e-01 1.43077701e-01 -4.64420676e-01
-8.82453918e-01 -1.05239737e+00 1.43891037e-01 8.82866859e-01
6.24493599e-01 -8.02336395e-01 -4.68541801e-01 -4.93604064e-01
1.02955334e-01 1.14748418e-01 -6.65447414e-01 -4.00307626e-01
-1.05648971e+00 -7.03297496e-01 5.15575647e-01 8.60806346e-01
6.29894316e-01 -7.49538958e-01 -8.51442814e-01 2.31779784e-01
-6.12354875e-01 -8.87656450e-01 -7.04597354e-01 -2.45403185e-01
-9.97255206e-01 -9.40137684e-01 -1.03541934e+00 -5.61791182e-01
5.08221626e-01 4.38275486e-02 7.31141746e-01 2.91627981e-02
-1.03672512e-01 2.69362390e-01 -4.45312828e-01 2.25562394e-01
1.58322290e-01 1.31813049e-01 2.81046629e-01 -2.05779627e-01
1.09826401e-02 -4.48943049e-01 -9.58828032e-01 8.53193581e-01
-3.77451867e-01 3.75426084e-01 4.72474098e-01 7.89447904e-01
6.73404098e-01 -7.15032443e-02 1.76410913e-01 -2.63419271e-01
4.71838862e-01 -3.55587393e-01 2.41333842e-01 9.97383669e-02
-2.24572271e-01 -1.31828822e-02 1.64518729e-01 -6.08570755e-01
-9.05959368e-01 4.38152313e-01 -3.35728824e-01 -3.08739334e-01
-3.89398113e-02 5.06545782e-01 -2.62226433e-01 2.74030805e-01
3.95529598e-01 2.12572515e-01 2.33194456e-01 -6.56864166e-01
2.17491165e-01 1.98611081e-01 5.68958700e-01 -5.43388247e-01
6.70841455e-01 3.08424443e-01 1.14394270e-01 -3.99383545e-01
-5.56512415e-01 -4.64536905e-01 -9.97541249e-01 -4.05748904e-01
7.15596855e-01 -8.68742645e-01 -5.94014883e-01 6.47489130e-01
-1.11013353e+00 -1.14303559e-01 7.85092860e-02 6.92722142e-01
-5.85605502e-01 6.70759737e-01 -6.45188749e-01 -6.89716101e-01
-3.24850559e-01 -1.11209369e+00 1.22308815e+00 1.94217153e-02
-8.01549435e-01 -5.53806901e-01 1.80689335e-01 2.77185410e-01
2.89984141e-02 4.93525058e-01 4.24240500e-01 -1.12897143e-01
-2.03674734e-01 -4.83370334e-01 2.47288093e-01 -3.08758337e-02
3.84232402e-01 -1.39821082e-01 -2.74923891e-01 -4.90283817e-01
-2.08480805e-01 -6.46855906e-02 7.22717822e-01 3.72459948e-01
8.11537743e-01 -2.74789929e-01 -5.64376950e-01 5.19534290e-01
8.40503454e-01 8.35826769e-02 6.65493965e-01 5.23792088e-01
9.35431719e-01 8.26411843e-01 1.04327607e+00 4.94497478e-01
4.85178590e-01 1.31118107e+00 2.89382607e-01 2.31269285e-01
-2.45047972e-01 -5.10591388e-01 4.87467557e-01 8.10724556e-01
-6.23726189e-01 -2.95544881e-02 -8.48260403e-01 3.86684507e-01
-2.17621112e+00 -8.74678195e-01 5.55240288e-02 2.34452391e+00
8.30397964e-01 1.56148702e-01 2.44486064e-01 2.36786410e-01
7.02666938e-01 4.52615768e-01 -4.97043669e-01 2.00629964e-01
2.12574497e-01 -1.32429540e-01 2.27276683e-01 4.06693757e-01
-1.12589753e+00 6.80711925e-01 5.68632507e+00 7.21578002e-01
-1.08946276e+00 -1.22570381e-01 2.28683501e-01 -2.79236883e-01
7.28509724e-02 -2.09926262e-01 -9.00713980e-01 7.33538806e-01
4.02271420e-01 1.85269088e-01 1.78417172e-02 6.90310538e-01
2.24072859e-01 -6.28149733e-02 -9.07450438e-01 9.11143243e-01
-2.81917565e-02 -8.24266195e-01 4.45949659e-02 -2.54192622e-03
5.37699938e-01 -4.02316779e-01 -1.14721293e-02 2.53032688e-02
-4.62461501e-01 -8.79029989e-01 8.67432296e-01 7.35308766e-01
4.90078956e-01 -6.95538044e-01 8.09065759e-01 5.56406438e-01
-1.66676044e+00 2.57600565e-02 -5.39951511e-02 -2.09320769e-01
4.67433363e-01 3.30991298e-01 -4.37329739e-01 8.30333352e-01
7.90407658e-01 9.11652863e-01 -6.03610039e-01 8.85443568e-01
-4.82746750e-01 1.82344958e-01 -5.83914459e-01 1.04660451e-01
-1.61429107e-01 -4.11894508e-02 7.16360629e-01 9.27951157e-01
3.86497080e-01 -2.92679649e-02 3.75690013e-01 4.41759497e-01
4.43372607e-01 -3.80195049e-03 -3.88990968e-01 3.43034923e-01
4.34228569e-01 7.75392115e-01 -4.85196769e-01 5.65588288e-03
-1.56795710e-01 1.31242096e+00 9.33074430e-02 3.65824252e-01
-8.69169950e-01 -1.09103605e-01 7.22547472e-01 2.45375067e-01
1.12769529e-01 -6.58408284e-01 -2.85424262e-01 -1.07855737e+00
4.19441789e-01 -7.60258734e-01 5.02591550e-01 -5.47082305e-01
-1.03326917e+00 3.83825511e-01 2.21424371e-01 -1.64535856e+00
-6.51812851e-01 -4.01068747e-01 -4.14016157e-01 7.71225691e-01
-7.18631387e-01 -1.10620344e+00 -3.94632012e-01 6.61597967e-01
4.00958002e-01 2.51165122e-01 7.87292898e-01 3.85891914e-01
-4.77144271e-01 7.27530777e-01 -3.82740080e-01 2.58978188e-01
9.03615892e-01 -8.98147464e-01 4.39955145e-01 7.94400990e-01
-2.06425842e-02 9.62204814e-01 8.04363847e-01 -9.27901566e-01
-1.31251395e+00 -7.86765814e-01 8.62021446e-01 -5.64439416e-01
1.33654058e-01 -1.71595573e-01 -8.00385058e-01 7.41740108e-01
-4.58051145e-01 -1.26430253e-02 4.81121331e-01 1.65958330e-01
-6.27906173e-02 -9.57597271e-02 -7.54895329e-01 7.20356345e-01
1.30422664e+00 -2.62971997e-01 -7.77191401e-01 1.65702049e-02
4.35796618e-01 -8.05803299e-01 -9.72105265e-01 8.84035289e-01
1.07255304e+00 -6.52961969e-01 1.29114127e+00 -3.45419943e-01
4.06419665e-01 -6.00356042e-01 -8.13014507e-02 -1.22621930e+00
-4.27536368e-01 -4.83673513e-01 -4.67079431e-01 8.67215812e-01
1.20091559e-02 -3.55109394e-01 9.50597703e-01 4.72899497e-01
1.64035901e-01 -1.14570475e+00 -1.09510529e+00 -6.73470497e-01
-1.98666930e-01 -3.71443361e-01 4.55203176e-01 5.74053109e-01
1.21142166e-02 1.25801578e-01 -9.88378882e-01 1.27313554e-01
3.52311164e-01 3.44606578e-01 1.11583436e+00 -9.80087280e-01
-3.62171948e-01 -1.97740585e-01 -8.36293340e-01 -1.61481512e+00
-2.84920037e-01 -4.54299808e-01 1.90209478e-01 -1.51739454e+00
1.35035768e-01 -1.70471847e-01 -3.63377154e-01 5.20051837e-01
-4.67267960e-01 4.86648232e-01 4.16014254e-01 5.44303060e-01
-4.39211845e-01 7.59041488e-01 1.59255421e+00 3.55573557e-02
-3.56434405e-01 3.41961294e-01 -2.62932926e-01 7.19476759e-01
7.46005774e-01 -4.16008770e-01 -3.78648400e-01 -2.90804535e-01
1.94612756e-01 1.06927052e-01 5.09443164e-01 -1.12540865e+00
2.71126896e-01 -4.12122272e-02 9.26616728e-01 -8.70882630e-01
6.83082223e-01 -6.33973539e-01 6.10032678e-01 7.82856345e-01
-9.32513624e-02 1.79696992e-01 -3.90116274e-02 7.20950425e-01
-2.20911786e-01 1.89244911e-01 4.12960976e-01 -4.03901860e-02
-8.22596788e-01 4.09605145e-01 -1.93538070e-01 -2.85504431e-01
9.73472297e-01 -5.38369298e-01 -1.52873294e-02 -4.91871357e-01
-9.76246238e-01 2.97422022e-01 4.65517133e-01 8.01885188e-01
6.29836679e-01 -1.53052545e+00 -3.52549940e-01 8.37496668e-02
6.00823313e-02 3.30276676e-02 3.81488353e-01 1.02210033e+00
-5.04599690e-01 3.07968229e-01 -5.36739588e-01 -7.11704314e-01
-1.30787027e+00 1.12372503e-01 2.77636111e-01 -3.65325630e-01
-7.95363963e-01 7.34377921e-01 -6.22300170e-02 -4.91307467e-01
2.68892527e-01 -1.38102010e-01 -2.07438573e-01 -2.22260915e-02
3.40453357e-01 5.46722651e-01 -1.42010510e-01 -9.43477035e-01
-7.38592446e-01 1.01404500e+00 1.48484036e-01 -2.39402223e-02
1.11474323e+00 -3.02057266e-01 2.29055896e-01 4.58297938e-01
1.22853947e+00 1.04267098e-01 -1.58550882e+00 -1.06094725e-01
-3.40018600e-01 -7.85937428e-01 -3.80999148e-01 -6.35000288e-01
-1.15399599e+00 7.66025841e-01 6.62173331e-01 -3.80043834e-01
1.14434648e+00 -9.05582607e-02 1.01209211e+00 1.13509580e-01
6.31466985e-01 -1.00696123e+00 2.03501955e-01 4.28769857e-01
1.12791169e+00 -9.51301455e-01 4.57770467e-01 -7.15752184e-01
-7.32844949e-01 1.10467315e+00 7.35340178e-01 -1.38899237e-01
5.20641148e-01 -6.55581951e-02 -2.43938994e-02 1.54571403e-02
-3.83798301e-01 -7.88270012e-02 8.57962966e-01 4.90058362e-01
5.79680860e-01 4.06947955e-02 -6.14255607e-01 7.72862554e-01
-2.92096019e-01 -1.56082988e-01 -1.50356814e-01 1.19094539e+00
-2.24587634e-01 -1.20340002e+00 -4.47087288e-01 -1.81459561e-02
-3.11753839e-01 4.38734174e-01 -3.81274104e-01 1.00572836e+00
1.79555997e-01 7.01039970e-01 -2.06298218e-03 -8.32641363e-01
5.58206737e-01 -2.87056230e-02 6.24138534e-01 -4.37054902e-01
-3.44675183e-01 2.88376480e-01 6.77697212e-02 -1.01332641e+00
-4.12023842e-01 -7.53823459e-01 -1.39668691e+00 -2.90016919e-01
-1.37368843e-01 -1.82760075e-01 2.10112065e-01 8.03511918e-01
4.81740236e-01 4.64149803e-01 3.59181941e-01 -1.35100853e+00
-4.44955587e-01 -9.70457852e-01 -4.58289117e-01 6.69769645e-01
1.45970985e-01 -1.13402069e+00 -3.55744027e-02 -1.04055919e-01] | [7.135378837585449, -0.6709673404693604] |
f4ea8df5-5798-4443-8072-e8deeba4c6eb | data-augmentation-for-opcode-sequence-based | 2106.11821 | null | https://arxiv.org/abs/2106.11821v2 | https://arxiv.org/pdf/2106.11821v2.pdf | Data Augmentation for Opcode Sequence Based Malware Detection | In this paper we study data augmentation for opcode sequence based Android malware detection. Data augmentation has been successfully used in many areas of deep-learning to significantly improve model performance. Typically, data augmentation simulates realistic variations in data to increase the apparent diversity of the training-set. However, for opcode-based malware analysis it is not immediately clear how to apply data augmentation. Hence we first study the use of fixed transformations, then progress to adaptive methods. We propose a novel data augmentation method -- Self-Embedding Language Model Augmentation -- that uses a malware detection network's own opcode embedding layer to measure opcode similarity for adaptive augmentation. To the best of our knowledge this is the first paper to carry out a systematic study of different augmentation methods for opcode sequence based Android malware classification. | ['Jesus Martinez del Rincon', 'Niall McLaughlin'] | 2021-06-22 | null | null | null | null | ['android-malware-detection'] | ['miscellaneous'] | [ 7.71313608e-01 1.29472762e-01 -4.20043945e-01 -3.01660925e-01
-1.54539898e-01 -3.37021589e-01 7.17329681e-01 2.15708539e-01
-4.14730906e-01 3.42129588e-01 -7.39101917e-02 -8.86810958e-01
4.27093983e-01 -6.20103240e-01 -5.93755066e-01 -4.77445632e-01
-3.93842727e-01 5.05958915e-01 1.72561333e-01 -4.64438498e-01
3.67873341e-01 5.36106884e-01 -1.57622123e+00 3.78504515e-01
6.81082547e-01 4.96744514e-01 -1.19540885e-01 1.17275310e+00
-3.88540328e-01 6.32929683e-01 -9.42816079e-01 -4.48501676e-01
3.26171160e-01 -5.30222237e-01 -6.60519898e-01 -3.04281443e-01
2.38430589e-01 -2.03266278e-01 -1.40060350e-01 1.05856812e+00
2.67085880e-01 -1.61847875e-01 6.15562856e-01 -1.49394381e+00
-5.16900122e-01 3.23968768e-01 -3.70286882e-01 3.89122695e-01
4.83854920e-01 1.93461567e-01 5.13167441e-01 -2.67446578e-01
4.76915628e-01 1.00770378e+00 1.04420030e+00 9.37020600e-01
-1.21281695e+00 -6.06176138e-01 -1.83772504e-01 1.46360770e-01
-8.86605322e-01 -1.94586605e-01 9.72244263e-01 -5.09662390e-01
1.08988917e+00 3.52766097e-01 8.92594695e-01 1.58513916e+00
2.71489888e-01 6.79140151e-01 1.32123065e+00 -4.30143386e-01
2.07126752e-01 1.87049255e-01 3.50110024e-01 8.18200827e-01
3.22029680e-01 9.76224840e-02 1.17970265e-01 -4.79360849e-01
2.17110500e-01 1.05501808e-01 1.69764489e-01 -3.12548667e-01
-6.96514666e-01 1.12425137e+00 2.78133303e-01 4.41221446e-01
-1.73919871e-01 2.38128215e-01 8.73135388e-01 3.71439606e-01
5.66161990e-01 7.94204175e-01 -6.64613128e-01 -6.08884335e-01
-8.79963517e-01 2.06373006e-01 8.44561160e-01 3.20268005e-01
5.74160278e-01 3.74456912e-01 4.23967719e-01 7.83664465e-01
2.99371541e-01 2.97306895e-01 1.11766696e+00 -7.39534318e-01
5.29704504e-02 6.50740087e-01 -4.58703399e-01 -9.23175573e-01
-2.60298491e-01 -1.45269543e-01 -5.40155649e-01 3.84045273e-01
2.92979807e-01 -8.25761035e-02 -1.08798599e+00 1.65861678e+00
2.42522672e-01 5.56467533e-01 2.23001763e-02 -1.38451727e-02
4.06055033e-01 5.51136196e-01 6.45604357e-02 -6.42430708e-02
1.34498787e+00 -8.11314523e-01 -6.61713898e-01 -9.41729322e-02
1.20502627e+00 -6.51824534e-01 1.46578562e+00 2.30107576e-01
-5.27756214e-01 -4.07280922e-01 -1.24583268e+00 1.18034765e-01
-1.02228165e+00 -2.41177589e-01 6.92821741e-01 1.46591520e+00
-9.06271219e-01 4.90316182e-01 -9.98897791e-01 -3.67865175e-01
7.69919991e-01 8.92294824e-01 -3.29461753e-01 1.06815971e-01
-8.70678544e-01 5.77157557e-01 5.37217557e-01 -5.90971112e-01
-9.29138958e-01 -5.37589908e-01 -9.73240793e-01 -2.32655138e-01
3.43145490e-01 -3.12342376e-01 1.36543608e+00 -1.17224598e+00
-1.67892945e+00 7.55958557e-01 -3.00926983e-01 -9.66861844e-01
2.94009726e-02 -9.28087905e-02 -5.46941996e-01 -3.00650805e-01
-2.13750020e-01 8.18803549e-01 1.22831798e+00 -1.32744777e+00
-1.30125403e-01 -4.21764523e-01 2.50196427e-01 -3.92404377e-01
-7.97741354e-01 -5.36452644e-02 7.83718005e-02 -8.17189515e-01
-5.95393419e-01 -1.35232604e+00 -2.58161664e-01 -5.06343603e-01
-1.74079701e-01 3.55538018e-02 1.54887152e+00 -6.80226147e-01
1.19838965e+00 -2.14854670e+00 -1.76556498e-01 8.86298791e-02
4.21927661e-01 1.00367045e+00 -3.56334865e-01 2.44646803e-01
-3.05327952e-01 5.44920743e-01 -8.30944657e-01 -6.03228986e-01
-2.45715037e-01 7.82972097e-01 -3.52678567e-01 2.17251435e-01
5.13127744e-01 1.05202436e+00 -8.75380337e-01 -3.04413110e-01
1.59331739e-01 8.80250812e-01 -8.93786788e-01 -7.97419064e-03
-3.92007768e-01 2.54432023e-01 -1.71527863e-01 5.10508478e-01
6.06728137e-01 1.34450182e-01 -1.12244911e-01 1.79942444e-01
8.02150741e-02 4.18774962e-01 -4.23919857e-01 1.18685758e+00
-7.38011658e-01 1.07428598e+00 -2.30994329e-01 -9.65609610e-01
7.35338688e-01 -1.02143355e-01 6.13435090e-01 -2.16012031e-01
3.17588538e-01 2.92551249e-01 8.02233458e-01 -5.30174434e-01
6.59957409e-01 2.83439428e-01 2.03845292e-01 4.32983071e-01
4.43632947e-03 3.88536826e-02 9.68594775e-02 4.76709902e-02
1.29262245e+00 -3.47122736e-02 4.18538600e-01 -9.50418413e-02
9.65994537e-01 6.74263062e-03 2.28164401e-02 6.85751438e-01
-2.88992018e-01 1.70365334e-01 7.26165950e-01 -4.37940329e-01
-1.20037603e+00 -6.69080675e-01 -3.16155097e-03 1.12080860e+00
-4.73461002e-01 -9.02684808e-01 -1.30138028e+00 -1.38695467e+00
-2.34506816e-01 4.24200356e-01 -1.00708628e+00 -3.18935782e-01
-9.02226090e-01 -9.67469275e-01 1.04585338e+00 3.37955207e-01
6.27315640e-01 -1.06477606e+00 -3.70380312e-01 -1.33465901e-01
3.25051546e-01 -8.44368756e-01 -3.50443333e-01 2.47498691e-01
-1.04906034e+00 -1.21597707e+00 -3.75814617e-01 -7.66870797e-01
7.13307977e-01 -1.61396474e-01 6.74469292e-01 4.39132035e-01
-2.31819913e-01 5.55087030e-01 -5.08209348e-01 -6.94436610e-01
-1.17544079e+00 4.84885216e-01 2.53496557e-01 -1.49538606e-01
6.54363871e-01 -5.17409503e-01 -1.26805350e-01 -4.99072559e-02
-1.05873299e+00 -6.32096469e-01 6.24329746e-01 8.72495294e-01
2.34460577e-01 -9.67872422e-03 1.72428608e-01 -1.14337790e+00
1.06949639e+00 -5.22758424e-01 -2.45825231e-01 -2.39680380e-01
-6.44650638e-01 7.89954066e-02 8.16801727e-01 -9.64664221e-01
-4.39701706e-01 2.01861367e-01 -8.91258240e-01 -2.72942990e-01
-2.87449539e-01 2.13970318e-01 -1.02052703e-01 -5.98277271e-01
7.44770944e-01 4.23918635e-01 5.43396115e-01 -3.12009990e-01
3.15001935e-01 8.68473470e-01 2.90853977e-01 -2.46358663e-01
8.41268837e-01 3.72294247e-01 2.38363311e-01 -1.52449107e+00
-1.28514841e-01 -2.08161548e-01 -5.41925251e-01 2.08211780e-01
8.76636446e-01 -2.49958798e-01 -6.91487908e-01 5.30782223e-01
-1.02212942e+00 -5.93289673e-01 -4.52610165e-01 6.99324533e-02
-3.77270579e-01 7.26440012e-01 -5.10024250e-01 -7.13507652e-01
-1.51230723e-01 -1.53353846e+00 1.00764751e+00 -1.56766132e-01
-2.72811323e-01 -1.32543457e+00 4.41444159e-01 1.24088027e-01
4.77532893e-01 9.11374539e-02 6.29465938e-01 -1.29448199e+00
2.72364374e-02 -3.66469502e-01 1.33764863e-01 6.50862098e-01
4.38671440e-01 1.23349488e-01 -1.21673238e+00 -2.16922835e-01
3.52650821e-01 5.18583544e-02 7.62023807e-01 1.29998058e-01
1.68430710e+00 -7.01215863e-01 -4.22545910e-01 6.34429812e-01
9.71783757e-01 3.42511564e-01 1.01178670e+00 3.89644593e-01
8.55872989e-01 4.35889632e-01 3.24682921e-01 3.81886517e-03
1.80220664e-01 7.65858471e-01 6.26057267e-01 -7.66388432e-04
-1.21055700e-01 -1.47025257e-01 5.98460793e-01 5.89175940e-01
1.14418395e-01 -2.57735372e-01 -1.05410254e+00 2.32169777e-01
-1.15850210e+00 -8.76310468e-01 -2.93623328e-01 2.07839370e+00
5.76039553e-01 2.68883377e-01 5.70576787e-01 4.86660242e-01
3.73488009e-01 2.52162308e-01 -2.50990868e-01 -1.00615156e+00
3.12543869e-01 5.59702516e-01 4.17173207e-01 7.57164299e-01
-1.22551024e+00 7.55084336e-01 6.33941746e+00 8.99464667e-01
-1.12359464e+00 5.15241146e-01 4.56040472e-01 2.51586437e-01
-2.21778266e-02 -1.57209963e-01 -7.22970843e-01 9.51396585e-01
1.35954797e+00 4.23351467e-01 4.62321609e-01 1.07457399e+00
-1.85049430e-01 2.52725303e-01 -7.89831281e-01 8.94962251e-01
3.55568081e-01 -1.22750330e+00 4.42311317e-02 7.45260239e-01
6.33519471e-01 2.11712644e-01 3.83336991e-01 5.01498461e-01
2.91449111e-02 -9.72409308e-01 -2.49419555e-01 1.63319074e-02
5.01931667e-01 -7.21664488e-01 8.71426642e-01 4.71631996e-02
-1.02685750e+00 -4.78980601e-01 1.76100284e-02 -1.19294515e-02
-4.35540751e-02 1.09854944e-01 -1.24237978e+00 -3.22714075e-02
3.11789185e-01 6.74970865e-01 -1.11430192e+00 6.83962762e-01
1.02868065e-01 1.00379145e+00 -2.19556317e-01 -1.61527291e-01
3.02846968e-01 1.00094594e-01 6.54631734e-01 1.35699785e+00
1.67938754e-01 -4.74081159e-01 1.01035461e-01 2.87471026e-01
-1.36193633e-01 9.48934108e-02 -1.11260545e+00 -6.33643568e-01
8.16376209e-02 1.09616184e+00 -6.59269392e-01 -5.42596698e-01
-2.29401290e-01 1.05240786e+00 -1.62612274e-02 -1.81655958e-01
-7.75570095e-01 -4.16117162e-01 1.17341304e+00 3.57583374e-01
5.52163906e-02 -6.47281826e-01 -3.02452713e-01 -8.18379641e-01
-3.91948074e-01 -1.23373461e+00 2.15227813e-01 -3.09761524e-01
-9.59279776e-01 6.29428983e-01 1.07180513e-01 -9.80417609e-01
-6.40890360e-01 -1.06281972e+00 -7.16705799e-01 3.26363802e-01
-1.08526337e+00 -1.08537138e+00 -5.57484925e-02 3.17567617e-01
7.02948511e-01 -7.31546640e-01 8.53577137e-01 2.85306662e-01
-5.44432819e-01 8.44876409e-01 5.20771295e-02 7.12961406e-02
1.07050881e-01 -1.31752300e+00 7.76788354e-01 7.68713892e-01
2.64467418e-01 1.01890230e+00 7.99731195e-01 -8.99622619e-01
-1.16207492e+00 -1.04789507e+00 5.60866714e-01 -8.43246996e-01
8.41808081e-01 -6.31785750e-01 -1.23702002e+00 6.75075889e-01
3.36268395e-01 -1.32283196e-01 1.02472246e+00 -2.01120004e-01
-4.65036005e-01 1.94464754e-02 -1.31939793e+00 7.13219047e-01
8.25853705e-01 -5.97280383e-01 -3.15092266e-01 3.43882114e-01
9.13838029e-01 -1.26661658e-01 -8.76333356e-01 5.35527110e-01
4.70740944e-01 -7.26523459e-01 1.19957662e+00 -7.01335549e-01
2.40875199e-01 -1.13979444e-01 -1.34809017e-02 -1.05374360e+00
2.37746432e-01 -6.66741550e-01 -7.58118033e-01 1.05701935e+00
1.32409260e-01 -1.01219916e+00 1.03497124e+00 -2.74795622e-01
-1.69857204e-01 -9.09291029e-01 -7.45964408e-01 -9.55681324e-01
8.16156715e-02 -5.89838505e-01 7.42763102e-01 1.03091586e+00
-2.53147393e-01 6.87640458e-02 -2.01722115e-01 5.28955571e-02
2.92743742e-01 -8.02013636e-01 1.02140963e+00 -1.11191761e+00
-4.10524815e-01 -7.26881087e-01 -8.74651670e-01 -7.18016148e-01
5.11894047e-01 -8.01868200e-01 -2.36926422e-01 -8.08651507e-01
-4.35789675e-02 -2.76827306e-01 7.08321435e-03 5.85357189e-01
-8.11949149e-02 7.85994470e-01 5.53251617e-03 -3.37930620e-02
-6.78119063e-02 3.86328101e-01 7.21355677e-01 -1.34909719e-01
-5.30449748e-01 1.04071416e-01 -4.31501985e-01 7.18673229e-01
1.54412413e+00 -5.06159544e-01 -6.93419755e-01 3.03629115e-02
7.03819096e-02 -9.41979885e-01 3.15746009e-01 -1.00009537e+00
-4.85459119e-01 2.47341037e-01 1.13210820e-01 -1.90422863e-01
5.05869627e-01 -8.49543512e-01 -3.50536555e-01 1.21646225e+00
-2.96780586e-01 5.03022492e-01 6.62901819e-01 5.18666744e-01
1.80449501e-01 -5.47831297e-01 6.37116015e-01 3.66422199e-02
-5.48488379e-01 1.87622547e-01 -9.98035550e-01 -1.37403488e-01
9.47601318e-01 -3.85602683e-01 -2.70868510e-01 -2.36107767e-01
-4.71023291e-01 -5.39773047e-01 9.31514800e-01 6.95338190e-01
6.53796017e-01 -1.09232795e+00 -2.86091805e-01 6.27757967e-01
6.21099137e-02 -8.51032853e-01 -2.54976898e-01 8.08715463e-01
-6.42535806e-01 1.73862547e-01 -3.18024516e-01 -6.61740124e-01
-1.92261791e+00 9.00355577e-01 3.11573654e-01 -4.09882337e-01
-2.02948391e-01 4.35428113e-01 -2.33019188e-01 -7.87150085e-01
-1.65532216e-01 -2.57612944e-01 -4.13598627e-01 -2.30982557e-01
5.95934987e-01 4.12522197e-01 -4.43035783e-03 -8.94051850e-01
-3.30717325e-01 4.04768586e-01 -2.82423884e-01 3.77739444e-02
9.73678768e-01 3.77414733e-01 -3.60235602e-01 3.63310218e-01
1.56751680e+00 2.84506619e-01 -7.48755097e-01 5.79536855e-01
4.67290543e-02 -5.90193927e-01 -3.03527266e-01 -4.74405646e-01
-6.58868492e-01 1.05966294e+00 1.16244841e+00 7.04837739e-01
1.03656673e+00 -2.88956523e-01 7.28149295e-01 6.39749706e-01
-8.71855542e-02 -6.06456339e-01 4.05028880e-01 6.51639283e-01
5.78432620e-01 -1.37974358e+00 -9.61694121e-02 -2.60440409e-01
-3.15115184e-01 9.96575356e-01 7.29143918e-01 -1.50365993e-01
6.69002295e-01 4.89376962e-01 -1.83306530e-01 -1.60026431e-01
-2.63456523e-01 -2.12429807e-01 2.36103192e-01 1.13494349e+00
4.76871163e-01 -1.44020736e-01 -4.30961609e-01 -2.51668096e-01
-3.80021036e-01 -2.28179902e-01 6.10887289e-01 9.58496928e-01
-1.91722199e-01 -1.95373046e+00 -2.95618773e-01 8.24550271e-01
-6.17558956e-01 -1.98322624e-01 -9.41646159e-01 1.00635636e+00
3.97233039e-01 8.06891143e-01 -5.03475182e-02 -9.03375328e-01
-2.25899398e-01 3.57288688e-01 5.61102390e-01 -7.48849571e-01
-8.15654695e-01 -5.37724555e-01 -4.55555283e-02 -2.70232975e-01
-2.99540997e-01 -6.51708305e-01 -1.10700727e+00 -2.63955116e-01
-8.43714699e-02 1.95527244e-02 1.23090720e+00 7.98357844e-01
5.04099905e-01 6.36875749e-01 5.27167797e-01 -9.67967629e-01
-1.97248295e-01 -1.05442965e+00 1.61826685e-01 2.62868673e-01
6.73759818e-01 -5.36151648e-01 -2.50385851e-01 -2.16319308e-01] | [14.423800468444824, 9.681509017944336] |
c3a6c01e-5312-4e84-8381-29fecb666ac8 | le-trading-algorithmique | 0810.4000 | null | https://arxiv.org/abs/0810.4000v2 | https://arxiv.org/pdf/0810.4000v2.pdf | Le trading algorithmique | The algorithmic trading comes from digitalisation of the processing of trading assets on financial markets. Since 1980 the computerization of the stock market offers real time processing of financial information. This technological revolution has offered processes and mathematic methods to identify best return on transactions. Current research relates to autonomous transaction systems programmed in certain periods and some algorithms. This offers return opportunities where traders can not intervene. There are about thirty algorithms to assist the traders, the best known are the VWAP, the TWAP, TVOL. The algorithms offer the latest strategies and decision-making are the subject of much research. These advances in modeling decision-making autonomous agent can envisage a rich future for these technologies, the players already in use for more than 30% of their trading. | ['Victor Lebreton'] | 2008-10-22 | null | null | null | null | ['algorithmic-trading'] | ['time-series'] | [-5.40418863e-01 2.16584504e-01 2.92495620e-02 -8.57871696e-02
4.21808571e-01 -9.66482341e-01 9.21952963e-01 -1.36829719e-01
-6.99014485e-01 7.47455001e-01 -3.02030712e-01 -5.12192786e-01
-1.52119160e-01 -1.37894392e+00 -8.57786164e-02 -5.27300477e-01
-6.11650586e-01 1.23882723e+00 5.25638461e-01 -7.38119364e-01
9.65463698e-01 5.38314104e-01 -1.55080318e+00 3.14167857e-01
2.69271821e-01 1.47419381e+00 -2.33814612e-01 9.34196293e-01
-6.67379677e-01 7.84554601e-01 -3.99871975e-01 -6.03863299e-01
1.28828883e+00 -2.39186108e-01 -2.45296776e-01 -5.03060222e-01
-8.97172332e-01 -6.62746310e-01 2.67887592e-01 7.93555200e-01
6.43128902e-02 -2.95008481e-01 5.69031358e-01 -1.28402007e+00
-3.03824931e-01 1.01759124e+00 -3.08139682e-01 4.58091289e-01
1.36743456e-01 2.76064068e-01 9.19416726e-01 -7.42021859e-01
4.51280475e-01 1.13559413e+00 3.98474365e-01 1.38522014e-01
-5.80407679e-01 -6.04223073e-01 -4.61504608e-01 3.60163674e-02
-5.92301190e-01 -4.07180339e-02 3.86498749e-01 -1.93510339e-01
1.27093852e+00 4.82712597e-01 1.31414187e+00 -1.01025437e-03
8.69729817e-01 6.78158820e-01 1.25389206e+00 -4.77492481e-01
4.12595779e-01 3.26680988e-01 2.34309554e-01 1.25984192e-01
7.16875553e-01 7.11365938e-01 -5.81998944e-01 -3.82983714e-01
1.20872843e+00 2.84854114e-01 4.27289814e-01 7.28609189e-02
-1.16093731e+00 1.09416807e+00 -3.72894585e-01 5.34255505e-01
-8.46454680e-01 5.90057299e-02 5.27026176e-01 1.16306770e+00
2.53777504e-01 7.52641141e-01 -6.76549554e-01 -5.53622961e-01
-6.08260214e-01 6.95580959e-01 1.67785788e+00 4.40521330e-01
4.69624549e-01 1.76306531e-01 3.66707832e-01 -6.00984842e-02
6.35424674e-01 6.02937043e-01 8.49036336e-01 -1.14394748e+00
2.58985192e-01 6.25558019e-01 5.91799080e-01 -7.79665887e-01
-3.75372082e-01 9.18987468e-02 -4.72621351e-01 9.96850729e-01
7.61320829e-01 -4.67550188e-01 -3.19046974e-01 4.72571224e-01
1.66177079e-01 -2.95402497e-01 8.86515751e-02 4.21731740e-01
-4.50009763e-01 8.49104941e-01 -3.43919158e-01 -4.16281849e-01
1.29322624e+00 -5.58193147e-01 -8.65094125e-01 6.19479716e-01
8.33230689e-02 -1.07341194e+00 -1.07605226e-01 1.19550145e+00
-1.33089447e+00 -1.78221241e-01 -1.01779914e+00 8.31071258e-01
-7.77131200e-01 -6.94426417e-01 1.02568102e+00 9.47961569e-01
-1.18156302e+00 9.81598616e-01 -8.84715736e-01 2.49946296e-01
7.38795614e-03 6.60633147e-01 5.04157364e-01 1.05478883e+00
-1.42043233e+00 1.15859246e+00 5.63390076e-01 2.95047730e-01
-4.27707940e-01 -6.31710351e-01 6.46244213e-02 -7.06286728e-02
3.80084217e-01 -4.62954193e-01 1.42539489e+00 -1.33037174e+00
-2.04063749e+00 8.80110979e-01 6.82472229e-01 -1.50515318e+00
1.33205914e+00 -1.85835689e-01 -3.94215554e-01 9.31155160e-02
-2.41175577e-01 1.83441609e-01 7.44254112e-01 -3.45620304e-01
-1.25733352e+00 -1.29875362e-01 -2.46195868e-01 1.34706467e-01
1.29081130e-01 3.64265054e-01 6.58268273e-01 -8.12742174e-01
6.65480793e-02 -6.05157614e-01 -4.44203794e-01 -2.08340690e-01
3.25298190e-01 -3.26787353e-01 6.13231659e-01 -4.64016378e-01
8.46265316e-01 -1.41510689e+00 -6.41335070e-01 6.46414638e-01
-5.25856763e-02 1.30993918e-01 6.59437239e-01 1.18267405e+00
1.55374527e-01 3.94785076e-01 1.20681562e-01 5.08394599e-01
4.76941586e-01 4.67047170e-02 -8.41765940e-01 1.71644121e-01
-1.38843656e-01 8.17491412e-01 -3.04258913e-01 -2.19352767e-01
2.53887683e-01 -2.53465354e-01 -4.56976220e-02 8.40782821e-02
-5.28971791e-01 -3.32793534e-01 -9.18819427e-01 6.57666326e-01
5.49827754e-01 3.45761031e-01 -5.50358221e-02 7.04700410e-01
-8.34302902e-01 7.54686967e-02 -1.63239515e+00 6.73239231e-01
3.10092092e-01 3.84585023e-01 3.21588218e-01 -6.75346613e-01
1.11918616e+00 6.53465688e-01 8.01842511e-01 -6.31450355e-01
1.61372826e-01 7.86733866e-01 4.13169622e-01 -2.28213042e-01
5.37474096e-01 -6.08478248e-01 2.45841101e-01 1.55287278e+00
-6.33244216e-01 -1.34530991e-01 3.97186518e-01 -2.13503569e-01
1.06035793e+00 1.67274952e-01 6.44451439e-01 -4.94304478e-01
3.98678362e-01 3.23714972e-01 5.22333860e-01 6.81643188e-01
-4.11945909e-01 -1.42604440e-01 5.32206297e-01 -1.10678422e+00
-1.11390233e+00 -8.84458065e-01 -2.22321507e-02 9.08078253e-01
-6.93631098e-02 4.19461280e-01 -5.97054183e-01 5.68098426e-02
4.95810807e-01 4.56380159e-01 -2.43290871e-01 4.95834947e-01
-6.30339921e-01 -7.93346703e-01 2.25711018e-01 4.87919338e-02
8.04529309e-01 -1.72423983e+00 -1.46961141e+00 8.60268891e-01
1.04021823e+00 -2.69264817e-01 8.21956098e-02 7.70588517e-02
-1.23133600e+00 -9.33439791e-01 -6.72910988e-01 -2.48529479e-01
2.00677752e-01 -9.63351652e-02 9.51921463e-01 2.87856102e-01
-5.36925122e-02 3.12498689e-01 -2.86318213e-01 -1.23845267e+00
-7.20296502e-01 -1.58456087e-01 5.95264360e-02 1.17353626e-01
4.31835711e-01 -5.25613844e-01 -7.07924187e-01 1.93387702e-01
-6.14241064e-01 -4.41647414e-03 7.06525743e-01 3.38401556e-01
1.65706739e-01 3.94130409e-01 6.74730659e-01 -1.12145102e+00
8.48969877e-01 -3.37738276e-01 -1.13194025e+00 1.67809278e-01
-1.09202588e+00 2.93431163e-01 3.86515588e-01 -1.16583286e-02
-1.18260252e+00 -3.41363877e-01 5.92273772e-01 1.71483025e-01
2.93865561e-01 2.14849487e-01 3.14044774e-01 -8.78047794e-02
8.68308172e-03 1.88751414e-01 4.20253664e-01 -1.52391419e-01
1.48732051e-01 4.10734266e-01 1.85875505e-01 -3.57321709e-01
7.41699994e-01 5.35348415e-01 -2.57957149e-02 -5.28169572e-01
4.45490718e-01 -2.34859586e-01 -4.83992189e-01 -5.35269558e-01
4.70732450e-01 -1.34124637e-01 -1.08454621e+00 9.68377590e-01
-1.15699673e+00 -7.40871578e-02 -4.23657835e-01 4.59441245e-01
-6.23096049e-01 -1.46734640e-01 -9.32692170e-01 -1.66037750e+00
-6.94670737e-01 -3.94477338e-01 1.96420714e-01 5.04641116e-01
-3.22097510e-01 -1.09302342e+00 5.81453145e-01 -1.56853609e-02
7.21405506e-01 2.41851926e-01 4.30081040e-01 -1.55181062e+00
-1.13096547e+00 -4.80075419e-01 3.75810117e-01 1.12146787e-01
-6.26101065e-03 3.59880775e-01 -5.78712642e-01 1.35582194e-01
7.44672000e-01 3.07835311e-01 4.60399091e-01 3.38838279e-01
-6.38573617e-02 -3.13043922e-01 -2.06227154e-01 2.48941686e-02
1.28812778e+00 1.33227849e+00 6.07576668e-01 8.93704295e-01
-2.89846092e-01 8.66157293e-01 1.01359200e+00 7.53759861e-01
-1.02133468e-01 4.02433537e-02 2.38064229e-01 4.45229679e-01
1.00564122e+00 1.35358199e-01 4.95431542e-01 7.53230691e-01
-7.11634040e-01 3.05493832e-01 -1.08878410e+00 -4.79080714e-02
-1.86477125e+00 -1.43340611e+00 -7.35751465e-02 2.15746832e+00
6.35292888e-01 9.15102005e-01 3.61400574e-01 2.51736432e-01
7.15304494e-01 -1.55450061e-01 -5.25887966e-01 -8.14451933e-01
6.60456270e-02 4.03097987e-01 1.00525320e+00 4.14943308e-01
-9.16805983e-01 5.45896947e-01 6.80727911e+00 4.01608527e-01
-8.18052590e-01 -3.66705865e-01 7.40829885e-01 1.22608547e-03
-3.38361293e-01 2.90047675e-01 -9.70347106e-01 7.12086201e-01
1.22615790e+00 -8.45918536e-01 5.85191011e-01 6.27225280e-01
6.12466276e-01 -4.55085218e-01 -8.36011410e-01 6.54786944e-01
-5.21502018e-01 -1.55878377e+00 1.00042010e-02 4.65958238e-01
4.32224691e-01 -2.57717192e-01 -6.57365322e-02 1.38917323e-02
7.00078487e-01 -6.89012110e-01 1.01531446e+00 1.25634718e+00
-5.18307507e-01 -1.00401855e+00 9.71582532e-01 3.49849075e-01
-1.10739517e+00 -9.28705260e-02 -3.23884636e-01 -5.75038731e-01
2.18488231e-01 2.74843514e-01 -7.39378691e-01 3.77396524e-01
4.41877574e-01 4.90114689e-01 -1.32971823e-01 9.32426929e-01
1.43441051e-01 4.01100963e-01 -7.02888131e-01 -7.20421493e-01
3.85968477e-01 -1.17208529e+00 5.04548907e-01 5.78979075e-01
4.33660179e-01 4.67319161e-01 -1.81088984e-01 8.16352487e-01
5.36340237e-01 1.11002229e-01 -5.60981214e-01 -2.14921713e-01
4.28558648e-01 9.63911891e-01 -1.43670011e+00 -4.51385587e-01
-4.25869338e-02 5.45008421e-01 -6.90370619e-01 -1.21063061e-01
-4.42318499e-01 -4.43540543e-01 4.47396666e-01 3.37494940e-01
1.14984721e-01 -3.02566767e-01 -7.74574459e-01 -6.59367204e-01
-1.11415043e-01 -1.04258823e+00 4.36408907e-01 -2.43389234e-01
-1.14011562e+00 6.25591502e-02 -1.59495860e-01 -1.23884308e+00
-7.69533515e-01 -9.75511611e-01 -1.00918782e+00 7.42463946e-01
-9.62807059e-01 -2.56891578e-01 7.66327381e-01 1.95685700e-01
5.42727232e-01 -1.04905951e+00 4.68587190e-01 -3.51510555e-01
5.04419915e-02 -4.75098848e-01 5.12205720e-01 1.60704270e-01
2.56611824e-01 -1.54733276e+00 5.34877300e-01 4.05595332e-01
2.46560991e-01 3.97068918e-01 8.17898452e-01 -7.32679486e-01
-1.61405611e+00 -5.57547361e-02 8.23889732e-01 -3.59231651e-01
1.20673621e+00 3.84324491e-02 -7.88314521e-01 3.58358324e-01
6.60360396e-01 -1.01906633e+00 3.51869464e-01 -5.37790000e-01
3.41361940e-01 -4.96012181e-01 -1.10991812e+00 4.60434198e-01
2.35065237e-01 1.21921338e-01 -1.07412016e+00 -1.39112055e-01
3.16513151e-01 3.45968977e-02 -4.88143831e-01 -3.61290723e-01
9.62034941e-01 -1.52631664e+00 6.97198987e-01 -1.94698304e-01
-2.37062514e-01 9.18818489e-02 4.51402247e-01 -7.58395851e-01
3.64607662e-01 -1.53794467e+00 2.02744380e-01 1.06092203e+00
6.80539310e-01 -1.75719655e+00 1.01692867e+00 1.34253323e+00
4.43212777e-01 -2.61614799e-01 -1.07787943e+00 -6.84472561e-01
4.96081822e-02 -1.41215399e-01 7.57100344e-01 6.83585405e-01
1.73094526e-01 -5.25068820e-01 6.46318495e-02 -5.86608708e-01
1.00956500e+00 4.98307824e-01 4.36807334e-01 -1.41340590e+00
-5.26841223e-01 -9.53891039e-01 -3.53583276e-01 -4.10266787e-01
-5.87060273e-01 -4.26112145e-01 -4.37530667e-01 -8.20177734e-01
-4.30676639e-01 -2.40387112e-01 -4.45454210e-01 1.37577800e-03
5.59952378e-01 -5.34773171e-01 5.28923810e-01 6.70388818e-01
-1.27162695e-01 -2.50425115e-02 1.11020720e+00 1.10069484e-01
-4.28182960e-01 4.25738752e-01 -2.52353549e-01 8.22571456e-01
1.22212160e+00 -5.28954923e-01 -2.06456050e-01 3.79080683e-01
9.26045835e-01 2.11991996e-01 -8.22858070e-04 -6.87451124e-01
5.37379503e-01 -4.13596362e-01 1.96128845e-01 -7.94683754e-01
-2.63747573e-01 -9.60232615e-01 7.24952221e-01 1.28298676e+00
-1.19726680e-01 5.13552368e-01 -3.89833152e-01 5.28738499e-01
-4.17766392e-01 -6.56248033e-01 4.20492947e-01 -6.13170862e-01
-5.81952751e-01 1.51527282e-02 -9.26955104e-01 -1.37862235e-01
1.57213616e+00 -5.25637090e-01 -7.91049302e-02 -3.67563188e-01
-6.13961399e-01 4.93582636e-01 2.30075136e-01 1.79982837e-02
2.59943783e-01 -8.55052054e-01 -7.07329452e-01 1.46640211e-01
-1.07842982e+00 -3.24279130e-01 -5.70143759e-01 4.52414811e-01
-1.77181268e+00 6.00775599e-01 -9.15652990e-01 1.42469674e-01
-8.07404697e-01 3.45738560e-01 7.12200642e-01 -6.84501171e-01
-4.51232016e-01 4.22305822e-01 -3.99646521e-01 2.79588014e-01
-2.27675214e-01 -4.20743793e-01 -3.23191613e-01 5.39750874e-01
7.77454019e-01 8.81403923e-01 -3.86602253e-01 1.95202991e-01
-2.98231412e-02 4.24186319e-01 -1.69736654e-01 -1.00895226e+00
1.68925440e+00 1.71827361e-01 -6.30951345e-01 7.65029430e-01
1.17833793e-01 -3.17491412e-01 -1.10821378e+00 1.98192865e-01
8.80090833e-01 -3.11802804e-01 -4.33416516e-01 -5.16703546e-01
-8.70478332e-01 5.89621365e-01 3.56558979e-01 1.34239912e+00
7.62943745e-01 -7.87190855e-01 7.27558970e-01 6.36473358e-01
1.03106785e+00 -1.74968028e+00 -4.12724614e-01 4.72663820e-01
9.64858234e-01 -8.95410657e-01 9.95251238e-02 1.06538460e-01
-6.19352460e-01 1.66590393e+00 -1.48158759e-01 -9.15903926e-01
1.35198307e+00 8.52030516e-01 3.39451373e-01 -2.54941285e-01
-1.12170899e+00 2.74452478e-01 -4.96001989e-01 3.75117123e-01
2.92483598e-01 3.51576954e-01 -9.42547739e-01 8.31818819e-01
-5.21177590e-01 1.10540107e-01 6.70519650e-01 1.24682343e+00
-9.50931907e-01 -1.38968432e+00 -8.40856016e-01 5.36815464e-01
-7.29690731e-01 2.81486083e-02 -7.70403385e-01 1.17708158e+00
-1.96181551e-01 5.80471218e-01 2.40769356e-01 2.21514165e-01
4.15838063e-01 3.96196663e-01 1.48819191e-02 -2.13171348e-01
-1.44898069e+00 2.48775601e-01 -4.55038920e-02 -4.28335696e-01
-3.34680855e-01 -1.18699861e+00 -1.78275311e+00 -7.09298015e-01
1.19828179e-01 4.81761664e-01 5.20102203e-01 5.03822148e-01
-1.20710887e-01 -1.56307116e-01 8.27106535e-01 -8.60072732e-01
-1.12670600e+00 -7.56366611e-01 -1.43842900e+00 -3.78344417e-01
-1.71173021e-01 -3.45178485e-01 -4.73293930e-01 2.10333481e-01] | [4.570631980895996, 4.062162399291992] |
ae80feac-fbae-4250-87ea-ef22f80cf83e | android-malware-detection-using-deep-learning | 1712.08996 | null | http://arxiv.org/abs/1712.08996v1 | http://arxiv.org/pdf/1712.08996v1.pdf | Android Malware Detection using Deep Learning on API Method Sequences | Android OS experiences a blazing popularity since the last few years. This
predominant platform has established itself not only in the mobile world but
also in the Internet of Things (IoT) devices. This popularity, however, comes
at the expense of security, as it has become a tempting target of malicious
apps. Hence, there is an increasing need for sophisticated, automatic, and
portable malware detection solutions. In this paper, we propose MalDozer, an
automatic Android malware detection and family attribution framework that
relies on sequences classification using deep learning techniques. Starting
from the raw sequence of the app's API method calls, MalDozer automatically
extracts and learns the malicious and the benign patterns from the actual
samples to detect Android malware. MalDozer can serve as a ubiquitous malware
detection system that is not only deployed on servers, but also on mobile and
even IoT devices. We evaluate MalDozer on multiple Android malware datasets
ranging from 1K to 33K malware apps, and 38K benign apps. The results show that
MalDozer can correctly detect malware and attribute them to their actual
families with an F1-Score of 96%-99% and a false positive rate of 0.06%-2%,
under all tested datasets and settings. | ['Mourad Debbabi', 'ElMouatez Billah Karbab', 'Abdelouahid Derhab', 'Djedjiga Mouheb'] | 2017-12-25 | null | null | null | null | ['android-malware-detection'] | ['miscellaneous'] | [ 7.37318397e-02 -3.98748726e-01 -5.92585683e-01 8.37198347e-02
-3.86849195e-01 -7.99799681e-01 5.15797257e-01 -1.45474896e-01
8.81229714e-02 3.80220503e-01 -3.19177628e-01 -6.48182929e-01
2.69629300e-01 -6.16149724e-01 -4.59447622e-01 -5.92082381e-01
-1.11510769e-01 1.74449176e-01 5.47361493e-01 9.25152898e-02
1.84847951e-01 2.39344478e-01 -1.55275559e+00 3.88260424e-01
7.56733537e-01 1.36401200e+00 4.48298175e-04 7.90611982e-01
4.08820584e-02 7.25187123e-01 -7.92281330e-01 -7.59998977e-01
5.84451519e-02 -9.80969444e-02 -3.74131531e-01 -3.30627561e-01
2.84990650e-02 -5.77467918e-01 -2.57319838e-01 1.35599589e+00
1.49031594e-01 -6.35753155e-01 4.14152354e-01 -1.53293097e+00
-3.53613734e-01 4.75151956e-01 -6.04457557e-01 2.64080584e-01
2.43988499e-01 3.54524672e-01 6.49765313e-01 -3.03729892e-01
1.35628819e-01 9.32908177e-01 8.34669113e-01 5.32088995e-01
-8.03733468e-01 -9.26433980e-01 -4.08136427e-01 2.09724501e-01
-1.36248517e+00 -3.89799565e-01 6.86825454e-01 -8.05834293e-01
7.92649209e-01 2.88142830e-01 6.43579245e-01 1.67202842e+00
7.21519947e-01 3.55988085e-01 9.11323726e-01 1.28418997e-01
2.52498865e-01 2.71681070e-01 3.19545716e-01 7.57042408e-01
4.73288834e-01 -1.05189927e-01 -7.50489347e-03 -6.50290072e-01
2.72232518e-02 5.88042974e-01 -1.08669795e-01 5.58933318e-02
-7.62155116e-01 8.19752634e-01 -5.28328866e-02 2.19600514e-01
-1.57131210e-01 2.71194316e-02 7.99313724e-01 -1.59095675e-01
3.60063463e-01 1.92316085e-01 -6.83789134e-01 -7.57519662e-01
-5.21523297e-01 -2.35088974e-01 7.77242661e-01 5.26687443e-01
5.81064343e-01 5.46108298e-02 4.65681821e-01 6.46425843e-01
5.22542596e-01 7.15129137e-01 1.08988678e+00 -6.50315762e-01
1.73700035e-01 1.02305448e+00 -2.62309164e-01 -1.42513609e+00
-1.08443685e-02 -1.77954227e-01 -6.26321554e-01 -2.36891001e-01
9.76286381e-02 -4.19130512e-02 -5.06071091e-01 1.39104903e+00
3.68887275e-01 4.63984728e-01 -2.95653164e-01 3.27513777e-02
3.27702790e-01 6.90540791e-01 -2.95685846e-02 -3.15653950e-01
1.44631898e+00 -5.89805305e-01 -4.94419962e-01 -2.07521230e-01
7.63046384e-01 -5.40428996e-01 1.52004576e+00 4.14616615e-01
-3.18374008e-01 -3.37097347e-01 -1.06820607e+00 3.38260978e-01
-5.94084561e-01 3.11616987e-01 4.21543688e-01 1.21433699e+00
-7.07978725e-01 2.03336015e-01 -1.13284457e+00 -1.77410990e-01
7.53480256e-01 5.10048509e-01 -2.16850445e-01 1.95278570e-01
-5.38988352e-01 2.11508259e-01 3.19843203e-01 -4.24859673e-01
-1.02679050e+00 -3.48999441e-01 -4.61485296e-01 1.33135110e-01
5.09920537e-01 -4.34005121e-03 1.27261293e+00 -8.97587180e-01
-1.46342838e+00 6.00109935e-01 2.63312086e-02 -4.91209656e-01
1.13716222e-01 -3.97929460e-01 -8.44426274e-01 -1.50928438e-01
-9.18563902e-02 -8.56476575e-02 1.27307999e+00 -9.02358115e-01
-4.67233479e-01 -5.41185796e-01 5.94368726e-02 -6.89481616e-01
-8.19354534e-01 4.05941233e-02 -1.48358986e-01 -3.13580483e-01
-6.31362259e-01 -1.42123389e+00 4.53428954e-01 -5.93229890e-01
-6.54274166e-01 -1.94990456e-01 1.63899338e+00 -8.75396371e-01
1.47575128e+00 -2.51163602e+00 -2.46878818e-01 -8.09035357e-03
5.42480230e-01 8.63903224e-01 3.27969939e-01 1.35076419e-01
1.84586018e-01 3.86702061e-01 -1.90825894e-01 -1.61492020e-01
-1.67587891e-01 1.05877317e-01 -5.71142375e-01 3.09381336e-01
-1.94113776e-01 7.47971714e-01 -8.14479470e-01 -1.22631840e-01
8.92257243e-02 6.13715529e-01 -4.96857285e-01 1.14468187e-01
-3.42434704e-01 6.17942549e-02 -3.40597391e-01 8.95181417e-01
4.75029320e-01 -5.74149787e-01 3.40345889e-01 1.49621680e-01
2.65984476e-01 4.76501703e-01 -3.88335705e-01 6.93854332e-01
-5.59155703e-01 7.58504152e-01 -1.35234848e-01 -5.03066719e-01
6.00264966e-01 6.00435473e-02 3.61022294e-01 -2.52837181e-01
6.01450801e-01 3.56093436e-01 3.50919873e-01 -5.78340232e-01
3.47813547e-01 5.14332056e-01 -5.48842289e-02 4.92636055e-01
-2.51380652e-01 5.90202391e-01 -1.22354783e-01 1.29115984e-01
1.45883083e+00 -3.81648213e-01 5.42742848e-01 -6.59953663e-03
8.64147305e-01 -2.89788038e-01 3.02553356e-01 2.26012111e-01
-3.16384315e-01 -1.51863948e-01 6.40890598e-01 -3.53925198e-01
-5.99428475e-01 -9.22233820e-01 1.38379633e-01 1.26176953e+00
2.96221767e-02 -9.02918875e-01 -1.11339486e+00 -1.35310066e+00
-1.22471035e-01 2.88195670e-01 -4.67000932e-01 -6.09675884e-01
-5.26143968e-01 -5.98861992e-01 7.91352928e-01 1.91208526e-01
7.96858847e-01 -1.00391889e+00 -4.59491074e-01 -1.44914851e-01
-1.45597924e-02 -1.31500936e+00 -5.71103573e-01 -1.52057931e-01
-5.23656547e-01 -1.38015521e+00 -3.93198095e-02 -6.26299202e-01
3.30033332e-01 3.55032563e-01 7.36101091e-01 1.77738115e-01
-1.34723559e-01 2.80722439e-01 -3.95149082e-01 -2.84065932e-01
-8.82918060e-01 4.56767499e-01 3.89845699e-01 2.20334738e-01
6.48320377e-01 -4.28709835e-01 -2.70014167e-01 4.74870414e-01
-7.56477058e-01 -5.01676440e-01 4.62584347e-01 3.30883443e-01
3.56400818e-01 4.41046923e-01 4.90375131e-01 -8.27691674e-01
4.90363181e-01 -9.28065062e-01 -7.28536963e-01 -3.75188477e-02
-5.69965363e-01 -4.40646946e-01 1.13226533e+00 -1.07280886e+00
-5.30605912e-01 -1.83714256e-02 -4.52237248e-01 -3.99284393e-01
-7.81736299e-02 -1.66999623e-02 -6.62321210e-01 -4.43452150e-02
6.45745039e-01 4.61746871e-01 8.67207646e-02 -2.86824882e-01
-3.98548581e-02 1.32358670e+00 2.69244701e-01 7.00345710e-02
7.02088416e-01 2.92147338e-01 -3.91042143e-01 -1.32618868e+00
-4.71656412e-01 -1.40036300e-01 4.54080440e-02 -1.39075727e-03
7.90465772e-01 -6.19138122e-01 -1.00233400e+00 8.83893609e-01
-9.57611322e-01 -3.41107309e-01 2.10471764e-01 1.11899450e-01
-5.01740985e-02 5.53300738e-01 -6.98523521e-01 -5.72393417e-01
-4.27525938e-01 -1.63658845e+00 1.09004855e+00 1.59388557e-02
-3.08255851e-01 -7.86881983e-01 -1.08725481e-01 3.67489308e-01
4.42262143e-01 -7.27749616e-02 9.46168184e-01 -9.47133720e-01
-5.09063125e-01 -5.18384039e-01 -1.66884124e-01 5.87595940e-01
6.96166873e-01 3.58019233e-01 -9.56072807e-01 -1.74553066e-01
3.29959810e-01 -1.16283614e-02 3.47973585e-01 4.98178788e-02
1.59548724e+00 -8.68466079e-01 -6.10107660e-01 6.10041499e-01
1.03897858e+00 6.78280652e-01 5.50861359e-01 3.66592482e-02
1.02353954e+00 2.17393160e-01 2.48087674e-01 2.45174333e-01
2.74507076e-01 6.60028934e-01 8.48550737e-01 6.71564758e-01
1.59951776e-01 -3.51477832e-01 8.41348886e-01 8.42191577e-01
4.03715730e-01 -2.73124933e-01 -8.40848088e-01 -3.21531855e-03
-1.31341767e+00 -8.23894501e-01 -1.47228926e-01 2.29168582e+00
6.00388348e-01 3.88742000e-01 5.77926397e-01 4.20416415e-01
7.79856205e-01 1.33664086e-01 -6.52567565e-01 -5.90865552e-01
4.23496604e-01 1.09733909e-03 4.12073672e-01 1.99737385e-01
-1.08293605e+00 7.91757286e-01 5.40051842e+00 1.00208509e+00
-1.70332718e+00 4.82424378e-01 7.70016849e-01 2.83849537e-01
1.99444473e-01 -3.25617135e-01 -8.51620138e-01 1.44528031e+00
1.32451916e+00 -1.67097757e-03 6.95793271e-01 1.47456181e+00
-9.01821107e-02 2.38750726e-01 -7.11308062e-01 1.24405646e+00
8.25112760e-02 -1.06252265e+00 -1.93289906e-01 7.64867544e-01
4.35969025e-01 1.43695906e-01 3.79962534e-01 4.01096523e-01
7.53190322e-03 -8.92378509e-01 3.40560526e-01 -1.08332336e-01
8.63783419e-01 -7.95381784e-01 8.06847215e-01 5.33168197e-01
-1.09529376e+00 -5.47852695e-01 -2.33719684e-02 -7.09409341e-02
-2.14541376e-01 6.00857496e-01 -1.10171223e+00 -2.29691043e-01
7.99654841e-01 8.15641522e-01 -7.48713434e-01 5.20723581e-01
-1.27651811e-01 1.08461368e+00 -4.80191447e-02 -3.43934864e-01
-2.69199073e-01 6.22943975e-02 2.97597259e-01 8.99994493e-01
4.88156438e-01 -4.37552929e-01 1.16580084e-01 3.07228833e-01
-4.53437597e-01 5.11961356e-02 -9.54446137e-01 -6.75696492e-01
6.13515437e-01 1.44192660e+00 -7.36464083e-01 -4.75765586e-01
-1.52390078e-01 8.84131491e-01 -6.34809881e-02 -1.70867190e-01
-1.23930812e+00 -3.45313072e-01 1.07043314e+00 4.33681369e-01
1.32494569e-01 -3.44936669e-01 -3.22259404e-02 -1.04672837e+00
1.12424847e-02 -1.25994360e+00 -1.29644843e-02 -3.14067513e-01
-1.17049778e+00 6.96398973e-01 -3.13107669e-01 -1.12149858e+00
-3.89070958e-01 -8.22066545e-01 -5.63494444e-01 1.30277216e-01
-7.49290526e-01 -8.66416633e-01 -4.20551419e-01 3.01637560e-01
4.57951099e-01 -4.33004737e-01 6.34515703e-01 4.51251596e-01
-7.82570183e-01 7.32820272e-01 2.59359300e-01 1.23936802e-01
2.53137082e-01 -8.53605568e-01 4.70191240e-01 6.97685719e-01
-5.54127432e-02 8.87672842e-01 3.12600523e-01 -8.44960988e-01
-1.51493144e+00 -1.31851351e+00 3.86025786e-01 -6.43564939e-01
9.37497735e-01 -6.23433888e-01 -8.91332984e-01 7.43618548e-01
-5.93919642e-02 -3.35303210e-02 8.23200464e-01 -3.75359505e-01
-6.35604739e-01 -4.07322675e-01 -1.31575263e+00 4.78773654e-01
8.37494016e-01 -6.87123597e-01 1.76324874e-01 4.09989923e-01
9.69791651e-01 -7.10911453e-02 -6.00947738e-01 2.77083486e-01
7.95054376e-01 -1.17760062e+00 8.35936606e-01 -2.62913018e-01
2.83710569e-01 -1.72481894e-01 -3.46185833e-01 -7.48436391e-01
7.90823549e-02 -6.71200693e-01 -9.62311387e-01 1.23957980e+00
1.71164036e-01 -1.00128961e+00 7.73612678e-01 -1.48464188e-01
1.15758009e-01 -9.23882604e-01 -8.43778074e-01 -8.56895268e-01
-4.26555723e-01 -6.67036772e-01 8.55750799e-01 8.90710294e-01
-3.76453698e-01 3.87155563e-01 -4.36383724e-01 1.20336480e-01
3.34116131e-01 -4.71177369e-01 9.34647620e-01 -1.25230718e+00
-5.72021842e-01 -4.85138923e-01 -6.14081621e-01 -7.23650575e-01
3.19831878e-01 -5.61127305e-01 -2.92088598e-01 -5.15919447e-01
2.22580314e-01 -2.11447626e-01 1.04496874e-01 4.74028170e-01
3.34407538e-02 5.55586159e-01 -2.67129600e-01 3.63314509e-01
-4.99038935e-01 1.90920070e-01 4.64976251e-01 -2.77581334e-01
-3.76943797e-01 3.83309126e-01 -7.15238273e-01 1.01096725e+00
1.07730651e+00 -3.16586733e-01 -5.99879026e-01 -3.38194966e-02
2.56311774e-01 -5.42142451e-01 2.84995049e-01 -1.15735245e+00
-5.01245499e-01 2.79003046e-02 -1.06519656e-02 -2.78769493e-01
4.59857136e-01 -7.87437379e-01 1.90013438e-01 9.34287667e-01
1.51536196e-01 2.99066484e-01 4.40760814e-02 5.98833323e-01
3.43467683e-01 5.69065548e-02 7.55334616e-01 1.48049325e-01
-4.29979682e-01 2.52490938e-01 -5.17502666e-01 -2.10268632e-01
1.34975302e+00 -1.90823615e-01 -7.44214892e-01 -8.24913085e-02
-6.29045665e-02 -4.90268588e-01 8.60609472e-01 6.18997991e-01
5.24083376e-01 -9.38540757e-01 9.67692435e-02 3.95821542e-01
2.73296267e-01 -6.34041846e-01 2.80874278e-02 7.84892678e-01
-4.99142736e-01 2.74873227e-01 -1.26038888e-03 -8.89212668e-01
-1.78673291e+00 8.85379255e-01 8.59802738e-02 -1.30292088e-01
-3.01255733e-01 3.40082735e-01 -1.28323108e-01 -2.25293532e-01
1.40724644e-01 -3.22756469e-01 -2.50438243e-01 -2.50159979e-01
9.00535047e-01 6.70433640e-01 6.07945360e-02 -9.64005172e-01
-6.50385380e-01 3.11443806e-01 -1.73175186e-01 5.64308226e-01
7.63126433e-01 2.33760640e-01 -3.10580552e-01 3.65523696e-01
1.51479435e+00 5.86788476e-01 -7.15235412e-01 4.73698556e-01
5.42655215e-02 -5.76877773e-01 -3.94566000e-01 -6.21568143e-01
-9.79490876e-01 8.53789628e-01 9.50085938e-01 9.59204793e-01
9.50380266e-01 1.75314918e-01 1.34531045e+00 2.01721802e-01
5.54976940e-01 -2.48040020e-01 2.87376612e-01 4.52031761e-01
1.53146967e-01 -1.13659525e+00 -3.32599968e-01 -4.04153585e-01
-2.93933332e-01 6.10422194e-01 6.16878450e-01 -2.50553712e-02
7.05439568e-01 2.18860388e-01 -2.12069333e-01 -9.71832965e-03
-3.80717009e-01 3.64157140e-01 -1.38959708e-03 7.37390459e-01
2.26473898e-01 4.14840341e-01 -1.06551409e-01 6.91770971e-01
-2.63217509e-01 -2.18793094e-01 5.22367299e-01 7.52433121e-01
-5.19604445e-01 -1.23975348e+00 -1.63068950e-01 8.49435985e-01
-9.41491723e-01 7.55131394e-02 -7.77364612e-01 5.44398248e-01
4.21693951e-01 1.06677544e+00 -1.40960023e-01 -1.26786888e+00
-3.80756289e-01 -4.48608138e-02 -2.56626736e-02 -5.33792019e-01
-4.63102043e-01 -2.47819811e-01 -1.70472354e-01 -5.19341111e-01
1.76846683e-01 -3.03070545e-01 -1.02525783e+00 -5.27231872e-01
-2.81643689e-01 2.56200973e-02 9.75023687e-01 9.17732894e-01
9.13409412e-01 3.49954724e-01 9.34222460e-01 -6.78131938e-01
-5.04983723e-01 -9.71659482e-01 -1.54479221e-01 6.80450648e-02
4.42616075e-01 -7.80874133e-01 -3.98566455e-01 -1.24530770e-01] | [14.42257022857666, 9.681090354919434] |
0b9fccb0-c16b-4671-ab25-9689325e9739 | distortion-adaptive-grape-bunch-counting-for | 2008.12511 | null | https://arxiv.org/abs/2008.12511v1 | https://arxiv.org/pdf/2008.12511v1.pdf | Distortion-Adaptive Grape Bunch Counting for Omnidirectional Images | This paper proposes the first object counting method for omnidirectional images. Because conventional object counting methods cannot handle the distortion of omnidirectional images, we propose to process them using stereographic projection, which enables conventional methods to obtain a good approximation of the density function. However, the images obtained by stereographic projection are still distorted. Hence, to manage this distortion, we propose two methods. One is a new data augmentation method designed for the stereographic projection of omnidirectional images. The other is a distortion-adaptive Gaussian kernel that generates a density map ground truth while taking into account the distortion of stereographic projection. Using the counting of grape bunches as a case study, we constructed an original grape-bunch image dataset consisting of omnidirectional images and conducted experiments to evaluate the proposed method. The results show that the proposed method performs better than a direct application of the conventional method, improving mean absolute error by 14.7% and mean squared error by 10.5%. | ['Yuzuko Utsumi', 'Ryota Akai', 'Yuka Miwa', 'Koichi Kise', 'Masakazu Iwamura'] | 2020-08-28 | null | null | null | null | ['object-counting'] | ['computer-vision'] | [ 1.71748862e-01 -1.42126098e-01 3.19949090e-01 -4.22578186e-01
-1.12931654e-01 -4.55990136e-01 8.93510342e-01 -1.80886641e-01
-7.60602832e-01 5.86631179e-01 1.72512993e-01 -3.28639567e-01
1.64490685e-01 -1.00015414e+00 -5.59813619e-01 -5.16495585e-01
3.65503311e-01 5.33708632e-01 2.47227296e-01 4.11167383e-01
4.84070212e-01 4.62336332e-01 -1.64380848e+00 -1.03395402e-01
9.56760645e-01 9.38437641e-01 6.70288146e-01 6.99439406e-01
-1.20961867e-01 7.16316700e-01 -7.62681663e-01 -4.60939884e-01
4.13771212e-01 -2.85278051e-03 -3.82420987e-01 5.16794920e-01
5.00790596e-01 -7.80773878e-01 -2.34947979e-01 1.43566132e+00
2.61583447e-01 8.18940178e-02 8.38369250e-01 -9.40696478e-01
-9.58910704e-01 3.01429965e-02 -7.66927600e-01 1.11350650e-02
2.94917524e-01 -4.45947796e-01 2.71164328e-01 -1.06961584e+00
4.39364105e-01 1.19406700e+00 4.98943985e-01 3.83065671e-01
-1.07582152e+00 -3.89233887e-01 -6.81398958e-02 2.36983806e-01
-1.70605433e+00 -1.33179575e-02 3.93423110e-01 -4.76390958e-01
5.21586537e-01 3.39098781e-01 7.12106407e-01 3.53557676e-01
-5.15215546e-02 7.24193633e-01 1.30113125e+00 -4.85567838e-01
4.99052197e-01 5.05625248e-01 -5.41607402e-02 4.71536368e-01
5.45388281e-01 -1.25803631e-02 7.48508796e-02 -1.61730513e-01
9.23183143e-01 1.13824502e-01 -2.54869103e-01 -6.33143187e-01
-8.44874561e-01 7.47545719e-01 2.35917538e-01 -1.73058920e-02
-2.84857064e-01 -3.83277416e-01 7.56643340e-02 -3.66979718e-01
5.87509096e-01 -1.64250568e-01 1.44760072e-01 -1.16678394e-01
-9.14628148e-01 1.96898758e-01 7.03588128e-01 1.12448525e+00
3.62303674e-01 3.14219035e-02 4.56678085e-02 1.07671833e+00
3.56239855e-01 8.78615558e-01 2.06383690e-01 -8.39306891e-01
3.50919336e-01 6.21578157e-01 4.38807666e-01 -1.27060127e+00
-2.16668934e-01 2.86894087e-02 -1.03712678e+00 2.83466488e-01
5.53856492e-01 2.27978185e-01 -7.90286005e-01 1.19767094e+00
4.16235983e-01 5.70119247e-02 -3.22300568e-02 1.19515371e+00
4.46284115e-01 6.39751494e-01 -2.14147449e-01 -5.82829677e-02
1.17243159e+00 -6.62455201e-01 -6.46428645e-01 -2.67211813e-02
1.68427780e-01 -8.97653103e-01 1.17260635e+00 6.92569494e-01
-9.02571201e-01 -4.38865930e-01 -1.04901981e+00 8.77645835e-02
-1.76716939e-01 5.26567638e-01 5.51316559e-01 9.58010018e-01
-8.92845154e-01 1.86383739e-01 -6.82596564e-01 -3.21723312e-01
7.75702000e-02 -2.68019317e-03 -3.73719662e-01 -3.98323685e-01
-4.93672431e-01 8.43194783e-01 2.51072645e-01 -1.36854723e-01
-4.68339890e-01 -5.63670516e-01 -8.77574205e-01 9.50553492e-02
-1.69743486e-02 -2.01800719e-01 1.05793905e+00 -4.31786746e-01
-1.51003659e+00 7.96168089e-01 -2.14259982e-01 -2.10390031e-01
7.13591576e-01 -1.09738722e-01 -1.82541788e-01 -3.86391692e-02
1.93886250e-01 3.58243972e-01 4.54332501e-01 -1.50763249e+00
-9.14537370e-01 -8.03104460e-01 2.37889192e-03 3.75645489e-01
-5.09651124e-01 -1.88449621e-01 -6.19027197e-01 -3.23580801e-01
1.58846840e-01 -7.22653866e-01 2.69476585e-02 -6.15296327e-02
-3.09633881e-01 2.40837246e-01 1.04513323e+00 -7.86149681e-01
9.70901966e-01 -2.01011443e+00 -3.47716272e-01 6.55820295e-02
-7.65740350e-02 3.78994703e-01 1.34944782e-01 -4.63470444e-02
2.60978520e-01 -2.25263670e-01 -1.33477569e-01 -6.59105599e-01
-3.09714973e-01 3.06039095e-01 -4.61229384e-02 6.75633013e-01
-2.12943271e-01 2.95767248e-01 -9.29865837e-01 -3.46424460e-01
7.86614180e-01 7.97310412e-01 -4.88955915e-01 2.25772727e-02
3.24829787e-01 4.75738496e-01 1.11708038e-01 5.55580735e-01
1.59710789e+00 5.16293161e-02 1.96434066e-01 -1.59359217e-01
-3.55359524e-01 -1.64890021e-01 -1.39517331e+00 1.11890900e+00
-5.60542464e-01 3.57441872e-01 -5.10460325e-03 -7.70282209e-01
1.35553265e+00 -1.17345788e-01 4.30034727e-01 -8.16887617e-01
3.41621079e-02 3.81993987e-02 -4.37824488e-01 -5.95559813e-02
9.75153565e-01 -1.38807148e-01 1.07237466e-01 1.75001714e-02
-2.08524048e-01 -4.78277415e-01 2.67920166e-01 -2.18821988e-02
6.44639492e-01 -6.69726506e-02 5.62709332e-01 -2.66186595e-01
6.94239259e-01 -1.40411006e-02 5.18285692e-01 6.32310152e-01
1.94731131e-02 1.01930285e+00 1.88074857e-02 -7.19550312e-01
-1.47516012e+00 -1.19772577e+00 -5.52680969e-01 2.14810386e-01
4.98943865e-01 -1.92705929e-01 -8.51250112e-01 -4.04152066e-01
-3.92903853e-03 1.10708690e+00 -3.75052482e-01 6.91035241e-02
-3.09636533e-01 -1.12367463e+00 1.15334257e-01 6.81244850e-01
1.00254786e+00 -5.61901808e-01 -5.64069629e-01 -2.83447094e-03
-3.86694461e-01 -1.45265114e+00 -5.34036994e-01 -2.61092901e-01
-8.88282001e-01 -1.01728189e+00 -1.00172150e+00 -4.01212841e-01
8.88014972e-01 8.01015735e-01 9.60648060e-01 -2.03359619e-01
-1.95983738e-01 3.00445050e-01 -1.97131217e-01 -6.24107182e-01
-2.50802696e-01 -3.34020823e-01 1.50942847e-01 2.67344397e-02
8.58855367e-01 -4.50193197e-01 -4.73905861e-01 5.67912877e-01
-8.48019719e-01 1.00665569e-01 1.99069872e-01 7.90943205e-01
6.51133418e-01 1.69557139e-01 -4.90907521e-04 -5.97748518e-01
6.41475976e-01 -3.43981951e-01 -1.42021847e+00 -5.37439995e-02
-5.15483618e-01 -3.90244931e-01 8.21755290e-01 -2.42302641e-01
-1.30187726e+00 3.08876067e-01 5.03568016e-02 -3.17402571e-01
-2.28082448e-01 2.08839148e-01 -1.55018568e-01 -3.70655023e-02
4.73679006e-01 5.57679296e-01 -9.86028984e-02 -6.16973281e-01
8.88337865e-02 9.63846326e-01 8.29113841e-01 -2.21458942e-01
5.81627965e-01 9.41601634e-01 -1.23563394e-01 -1.23646891e+00
-4.17163700e-01 -7.63927042e-01 -6.81305110e-01 -3.75759602e-01
7.98791230e-01 -9.21115816e-01 -8.99628758e-01 6.68592334e-01
-1.26403999e+00 1.38408974e-01 -2.15296537e-01 8.64587188e-01
-7.04380274e-01 5.65074861e-01 -3.92824978e-01 -1.27556920e+00
-3.69200706e-02 -9.26951945e-01 1.23345780e+00 3.63097608e-01
2.79888213e-01 -7.60714531e-01 1.49631485e-01 3.02155465e-01
1.52206793e-01 -1.83932319e-01 6.98810160e-01 -1.32649541e-01
-5.19441187e-01 -3.64725858e-01 -6.71107173e-01 6.41363740e-01
3.28102589e-01 -2.55518824e-01 -8.72075081e-01 -1.41364858e-01
2.06605271e-01 1.43990800e-01 3.86236072e-01 5.62170506e-01
1.28693545e+00 -1.41573101e-01 -8.50198492e-02 7.39162445e-01
1.63177121e+00 4.22956198e-01 1.10126936e+00 5.09063065e-01
5.84672987e-01 1.28557414e-01 7.80597627e-01 8.01620305e-01
6.71155632e-01 8.45973730e-01 6.92496061e-01 -4.14145663e-02
1.21005177e-01 -5.49408793e-01 -1.22628175e-01 9.05534685e-01
-5.05445004e-01 -1.76698416e-01 -7.60258257e-01 7.02498138e-01
-1.66875136e+00 -9.21164393e-01 -2.05617949e-01 2.67072535e+00
2.67960012e-01 -3.15267950e-01 1.34423316e-01 3.97469282e-01
7.67216563e-01 -3.51751357e-01 -2.31892094e-01 -4.75448459e-01
3.53950821e-02 -2.82957822e-01 9.76278186e-01 6.64073944e-01
-1.09676957e+00 6.07777476e-01 6.52520943e+00 5.08795679e-01
-9.45724785e-01 -7.50492364e-02 4.08779621e-01 2.77207673e-01
-1.00703202e-01 -2.68970191e-01 -9.91914690e-01 6.39766097e-01
4.70245779e-01 -2.17617586e-01 5.63957274e-01 1.24977028e+00
2.81218946e-01 -6.47457182e-01 -5.17050207e-01 1.18616235e+00
3.02784353e-01 -9.92419481e-01 1.08446762e-01 4.89786029e-01
8.88300598e-01 -4.13160235e-01 -9.25602615e-02 3.89471389e-02
4.45570856e-01 -8.70913446e-01 7.73218095e-01 6.57744110e-01
7.42947400e-01 -8.45200539e-01 1.01994920e+00 8.65438282e-01
-1.09791195e+00 2.27057058e-02 -1.08817708e+00 -3.24457258e-01
4.12986666e-01 9.65786576e-01 -1.02118838e+00 5.17566681e-01
7.42518365e-01 2.33432963e-01 -4.14639831e-01 1.37180793e+00
4.09846194e-02 2.17049971e-01 -5.72863340e-01 -5.36130629e-02
-6.00344688e-02 -7.57796645e-01 4.94682461e-01 1.08503163e+00
7.48789907e-01 -7.68039189e-03 -4.73237969e-03 8.70630205e-01
2.21638426e-01 8.45754519e-02 -1.01357329e+00 3.74468386e-01
3.87882084e-01 1.13505888e+00 -7.29759037e-01 -3.57907265e-01
-6.43182874e-01 1.09905708e+00 1.32943820e-02 -1.47602626e-03
-9.08154368e-01 -8.00141916e-02 3.14865738e-01 2.54633695e-01
2.47140869e-01 -4.45159674e-01 -4.21479285e-01 -1.28584814e+00
2.02848941e-01 -5.35847664e-01 -4.75571975e-02 -9.25005078e-01
-1.11474133e+00 4.54733402e-01 2.90033251e-01 -1.22712898e+00
-1.33841112e-01 -7.02342272e-01 -2.94676930e-01 9.19449031e-01
-1.23863816e+00 -9.31864917e-01 -8.69546831e-01 4.99933392e-01
4.67290848e-01 -1.09076172e-01 9.19145703e-01 5.80847919e-01
1.50701061e-01 2.91431785e-01 5.77033579e-01 -2.88909934e-02
1.28203422e-01 -1.31452823e+00 4.33783501e-01 8.96290123e-01
1.16254494e-01 3.47746402e-01 5.52836657e-01 -6.30787075e-01
-1.00873470e+00 -1.10731030e+00 8.95228624e-01 -2.89254814e-01
6.61062226e-02 -4.91964221e-01 -6.13291383e-01 5.51139534e-01
-3.04600000e-01 2.10600972e-01 2.03176692e-01 -3.53955954e-01
6.88337674e-03 2.08034422e-02 -1.52639246e+00 5.87107778e-01
9.26405370e-01 -3.26054007e-01 -2.98700303e-01 1.07714944e-01
-7.04296678e-02 -2.89254367e-01 -5.99563181e-01 5.19628227e-01
6.78324342e-01 -1.24063158e+00 1.06394553e+00 2.26403728e-01
2.91814417e-01 -6.21469200e-01 -7.07354724e-01 -1.33041191e+00
-3.17586362e-01 2.44285464e-01 2.07645148e-01 1.19581413e+00
-9.96567160e-02 -4.67666924e-01 9.94583726e-01 3.47171545e-01
1.49302348e-01 -2.03663334e-01 -8.85425627e-01 -1.02468371e+00
-5.22008836e-02 -3.26580435e-01 4.29760635e-01 6.07808173e-01
8.90666619e-02 8.19345564e-02 -5.87122619e-01 2.11355522e-01
8.29360664e-01 -2.23214611e-01 1.03920269e+00 -1.11824262e+00
4.77694012e-02 2.17532143e-01 -8.85324121e-01 -1.50653243e+00
-2.20324725e-01 -5.72058022e-01 1.12949692e-01 -1.53093863e+00
5.50096989e-01 -3.79464447e-01 4.24960941e-01 -3.34579825e-01
-1.08105671e-02 5.58276772e-01 4.24641192e-01 1.79914180e-02
-8.73164907e-02 6.02334261e-01 1.18037915e+00 -1.35813326e-01
-3.44921574e-02 3.43929380e-01 -5.23591697e-01 1.10083330e+00
8.35354030e-01 -3.04798037e-01 -6.34765506e-01 -4.78822976e-01
-1.40921250e-01 -3.05791898e-03 5.43857992e-01 -1.16704011e+00
1.14311032e-01 -5.20907342e-02 5.17884552e-01 -9.21573579e-01
4.89412099e-01 -1.04098773e+00 1.51749134e-01 1.97540432e-01
3.47768068e-01 1.00672491e-01 1.65515631e-01 5.21105468e-01
-2.40369543e-01 -4.23901081e-01 1.09816110e+00 -2.28560969e-01
-5.20688891e-01 -4.99374121e-02 -5.12510121e-01 -4.22930747e-01
8.44472170e-01 -5.86066961e-01 -2.40653887e-01 -6.57109618e-01
-4.96906191e-01 -3.54920149e-01 7.47360706e-01 2.78213799e-01
7.12150991e-01 -1.56898558e+00 -4.62394625e-01 3.41106296e-01
1.58616789e-02 4.69045341e-02 5.75889163e-02 5.91952860e-01
-7.78291821e-01 6.39899850e-01 -3.25133353e-01 -8.58414292e-01
-1.33912754e+00 6.43294275e-01 1.81867018e-01 -2.42762700e-01
-7.21211076e-01 3.51853669e-01 4.77960676e-01 -8.58614087e-01
3.81758325e-02 -4.09828335e-01 -3.96780521e-01 -4.91456628e-01
8.19203913e-01 5.73356152e-01 2.69709349e-01 -7.68914998e-01
-3.22639614e-01 6.83258832e-01 1.31603479e-01 -3.33508193e-01
1.06274581e+00 -3.80960375e-01 2.87400663e-01 2.59928256e-01
9.14240181e-01 1.62871450e-01 -1.16465318e+00 8.51342175e-03
-2.11159334e-01 -1.30282807e+00 7.28656352e-02 -5.76876402e-01
-9.91206408e-01 8.32480133e-01 7.60375261e-01 1.65871501e-01
1.03034401e+00 -5.71052670e-01 3.41598034e-01 6.78738877e-02
9.93345141e-01 -1.02843034e+00 -3.19757491e-01 6.48633897e-01
7.10854053e-01 -1.08753324e+00 -7.00695440e-02 -7.28910923e-01
-5.43908298e-01 1.00149691e+00 6.47354305e-01 -1.42393917e-01
6.31941795e-01 4.68947172e-01 -5.05918302e-02 2.20843449e-01
-7.19481632e-02 5.40385880e-02 -4.44399640e-02 1.06325722e+00
1.39010325e-01 1.37327701e-01 -5.03398061e-01 2.95984685e-01
-3.07410926e-01 4.91361022e-01 8.17651629e-01 7.70406246e-01
-3.35474133e-01 -6.52426481e-01 -8.51852953e-01 3.41714978e-01
-3.20372671e-01 9.28313956e-02 -3.55971232e-02 7.52618432e-01
-5.06047392e-04 1.04739547e+00 5.50140262e-01 -3.31271827e-01
6.23511791e-01 -3.40039432e-01 5.81232190e-01 -3.63139302e-01
3.19739699e-01 -1.64889321e-01 -4.51765843e-02 -2.46399000e-01
-3.59982193e-01 -6.11729681e-01 -8.21514189e-01 -7.77438700e-01
-6.37836397e-01 -8.46806616e-02 1.06109822e+00 6.54491365e-01
-1.37503862e-01 1.77918613e-01 5.22653341e-01 -1.02270067e+00
-6.10904276e-01 -9.24819648e-01 -9.95609224e-01 4.26036984e-01
-2.29774833e-01 -8.31362128e-01 -3.95339757e-01 2.05645353e-01] | [8.895583152770996, -2.49094295501709] |
63011ae4-7a66-46f0-897a-66a86dcfa4a8 | informative-visual-storytelling-with-cross | 1907.03240 | null | https://arxiv.org/abs/1907.03240v2 | https://arxiv.org/pdf/1907.03240v2.pdf | Informative Visual Storytelling with Cross-modal Rules | Existing methods in the Visual Storytelling field often suffer from the problem of generating general descriptions, while the image contains a lot of meaningful contents remaining unnoticed. The failure of informative story generation can be concluded to the model's incompetence of capturing enough meaningful concepts. The categories of these concepts include entities, attributes, actions, and events, which are in some cases crucial to grounded storytelling. To solve this problem, we propose a method to mine the cross-modal rules to help the model infer these informative concepts given certain visual input. We first build the multimodal transactions by concatenating the CNN activations and the word indices. Then we use the association rule mining algorithm to mine the cross-modal rules, which will be used for the concept inference. With the help of the cross-modal rules, the generated stories are more grounded and informative. Besides, our proposed method holds the advantages of interpretation, expandability, and transferability, indicating potential for wider application. Finally, we leverage these concepts in our encoder-decoder framework with the attention mechanism. We conduct several experiments on the VIsual StoryTelling~(VIST) dataset, the results of which demonstrate the effectiveness of our approach in terms of both automatic metrics and human evaluation. Additional experiments are also conducted showing that our mined cross-modal rules as additional knowledge helps the model gain better performance when trained on a small dataset. | ['Siliang Tang', 'Jiacheng Li', 'Haizhou Shi', 'Yueting Zhuang', 'Fei Wu'] | 2019-07-07 | null | null | null | null | ['visual-storytelling'] | ['natural-language-processing'] | [ 2.68071234e-01 1.83704510e-01 -1.84967905e-01 -3.18925172e-01
-6.46986485e-01 -3.61190081e-01 8.43418539e-01 1.42377377e-01
-1.58801824e-01 8.40457559e-01 7.49002814e-01 -1.38433486e-01
6.93465322e-02 -8.20930123e-01 -8.60658646e-01 -5.67862570e-01
-2.26509944e-02 3.06866527e-01 1.26781538e-01 -3.57422471e-01
9.84038934e-02 -2.44757444e-01 -1.82092202e+00 7.91248620e-01
8.35027993e-01 9.74817514e-01 5.06784916e-01 2.20535636e-01
-3.30639452e-01 1.24185848e+00 -5.86548209e-01 -6.21700525e-01
-1.28422260e-01 -6.32345498e-01 -7.92730153e-01 2.30889067e-01
8.93386006e-02 -4.31773007e-01 -2.88842827e-01 7.78615773e-01
3.62895459e-01 2.58417636e-01 6.40354753e-01 -1.24542975e+00
-6.67625725e-01 1.01072085e+00 -5.50526738e-01 -1.51135206e-01
6.02547824e-01 1.75408229e-01 1.44488108e+00 -1.11214793e+00
7.81394720e-01 1.37540317e+00 2.32098475e-01 4.57202613e-01
-9.32929039e-01 -6.81746244e-01 3.24408084e-01 4.98576343e-01
-1.37448943e+00 -3.37277412e-01 7.80636847e-01 -3.73668760e-01
9.24088538e-01 1.85706884e-01 8.90641987e-01 1.39347768e+00
-1.64311364e-01 1.22296083e+00 7.10772336e-01 -3.76197100e-01
6.93149045e-02 2.53405899e-01 -2.30822995e-01 6.09355092e-01
1.46044895e-01 -1.01494953e-01 -8.58036876e-01 3.15357178e-01
7.87585020e-01 -1.35757327e-01 -3.76496822e-01 -2.74545372e-01
-1.61887109e+00 9.53965664e-01 5.62789559e-01 3.81082118e-01
-4.23930228e-01 1.75197259e-01 5.12743652e-01 -7.44235739e-02
2.89063334e-01 3.95001262e-01 2.08040729e-01 -5.98365478e-02
-7.25942314e-01 2.22005546e-01 3.84399623e-01 1.04627419e+00
4.43749696e-01 -5.11443019e-02 -4.65884477e-01 9.37116742e-01
3.49498749e-01 2.96120375e-01 3.92684817e-01 -6.82459593e-01
7.67960370e-01 8.02273929e-01 -2.04250608e-02 -1.09877884e+00
-6.81843311e-02 -3.37417990e-01 -7.60464489e-01 -1.27201155e-01
7.49717951e-02 -3.05368714e-02 -7.67589331e-01 2.02012372e+00
1.02945708e-01 5.23292981e-02 2.03400016e-01 9.14667249e-01
1.17073083e+00 1.03105342e+00 2.41387442e-01 -1.29127666e-01
1.42964828e+00 -8.31938863e-01 -9.88677204e-01 -4.39822525e-01
3.72427702e-01 -4.98077929e-01 1.30648637e+00 1.80741653e-01
-9.71019745e-01 -5.23559272e-01 -1.18153107e+00 -1.94710150e-01
-3.15975159e-01 1.03406511e-01 6.61033750e-01 6.28164038e-03
-7.30698168e-01 2.31519744e-01 -4.69991237e-01 -4.88540262e-01
6.12130642e-01 -2.77343988e-02 -3.47549319e-01 -1.69408306e-01
-1.33140576e+00 8.42285335e-01 9.79848504e-01 -1.07798450e-01
-8.81367505e-01 -3.50835353e-01 -1.05095565e+00 1.46895006e-01
5.05920887e-01 -9.51117516e-01 1.06754243e+00 -9.66717958e-01
-7.91172385e-01 6.30073905e-01 -2.05862194e-01 -3.74962002e-01
3.70812714e-01 -1.55725479e-01 -3.31494093e-01 3.71324033e-01
3.63206446e-01 1.13925886e+00 6.23667896e-01 -1.48576248e+00
-8.76478016e-01 -3.54959518e-02 2.94615358e-01 4.72715825e-01
-6.18621171e-01 -2.95777142e-01 -7.14058459e-01 -7.72156060e-01
-1.00017168e-01 -5.82639635e-01 9.56983939e-02 -1.94648117e-01
-6.82547092e-01 -1.83032706e-01 6.91269517e-01 -6.79938614e-01
1.38952529e+00 -2.17415094e+00 2.98056304e-01 8.67474526e-02
1.89681321e-01 -1.21575370e-01 -1.00426607e-01 5.41300237e-01
-9.02862661e-03 8.54872465e-02 -3.11863273e-01 -3.13656986e-01
-3.79521586e-02 2.01389387e-01 -3.98783535e-01 -2.13472754e-01
4.46948439e-01 1.08557808e+00 -9.50611770e-01 -7.68682361e-01
1.79520488e-01 4.50027078e-01 -2.79307485e-01 2.18071237e-01
-4.81165588e-01 2.43406102e-01 -4.92855340e-01 5.46157598e-01
-2.14941590e-03 -2.27885887e-01 1.70815662e-01 -2.76661038e-01
1.85810193e-01 4.15123999e-01 -9.78332281e-01 1.68475139e+00
-4.43898946e-01 8.56407583e-01 -6.06445789e-01 -7.50371754e-01
7.50037551e-01 3.94665450e-01 2.40834534e-01 -6.61492527e-01
1.86106905e-01 -9.89206955e-02 -1.17425762e-01 -8.58302414e-01
6.21356666e-01 -3.69439811e-01 -2.12357491e-01 3.86984974e-01
9.48949456e-02 1.28710359e-01 3.67879927e-01 6.38318777e-01
7.16497660e-01 2.00321794e-01 3.55049551e-01 2.24280670e-01
2.01726362e-01 2.29966983e-01 2.07509890e-01 5.47889650e-01
3.83937150e-01 5.57304025e-01 6.48811221e-01 -2.20890358e-01
-1.09428871e+00 -8.32475722e-01 2.05109015e-01 9.18614686e-01
3.06823760e-01 -6.92276895e-01 -5.36510468e-01 -6.35600150e-01
-3.39603901e-01 1.27752137e+00 -7.47409999e-01 -1.93760693e-01
-3.24862003e-01 -5.88486135e-01 4.41986650e-01 7.23244488e-01
6.79168999e-01 -1.37483609e+00 -8.40113342e-01 2.08845019e-01
-9.01340306e-01 -1.21092343e+00 -1.73582137e-01 -1.38536081e-01
-4.99750078e-01 -8.78017008e-01 -5.29869914e-01 -8.01103950e-01
6.23592496e-01 4.14636195e-01 1.05629683e+00 1.26444325e-01
6.92213923e-02 2.51447827e-01 -6.04825735e-01 -4.41224784e-01
-3.23412776e-01 -8.66031423e-02 -3.51911575e-01 1.43196940e-01
1.45785391e-01 -4.26148146e-01 -3.68958086e-01 3.92497145e-02
-1.12102532e+00 8.56050611e-01 6.12167180e-01 8.63881171e-01
5.63180685e-01 2.29027629e-01 6.31475329e-01 -7.23908305e-01
7.36821055e-01 -6.78216517e-01 1.59794036e-02 3.28675300e-01
-1.70003012e-01 1.73442066e-01 3.93623888e-01 -5.31378746e-01
-1.23067021e+00 -1.05101727e-01 -3.60872410e-02 -4.11401033e-01
-6.86714128e-02 8.13164771e-01 -5.55926323e-01 5.89471221e-01
5.74254215e-01 3.24464947e-01 -2.17000380e-01 -7.36942515e-02
5.33563077e-01 4.02566910e-01 4.81540471e-01 -5.98453462e-01
6.55124426e-01 3.75855416e-01 -2.12828457e-01 -6.22717321e-01
-8.49782586e-01 -2.47907773e-01 -1.34637967e-01 -5.28956234e-01
9.35538054e-01 -9.31889713e-01 -5.18664360e-01 -2.38725562e-02
-1.24525487e+00 -8.95959958e-02 -2.51675218e-01 3.42055112e-01
-6.15412295e-01 3.80520135e-01 -4.37086284e-01 -9.12457705e-01
-2.01259166e-01 -1.01015460e+00 9.98960376e-01 7.99269378e-02
-4.29807365e-01 -8.70946527e-01 -3.49687785e-01 4.79713261e-01
-1.92769431e-02 4.63067412e-01 1.35282314e+00 -6.34716690e-01
-7.08422363e-01 -1.36053190e-01 -2.16387078e-01 -2.18197498e-02
-2.17254370e-01 -1.49197444e-01 -9.90182102e-01 -7.98708107e-03
-3.57424468e-01 -5.75405002e-01 9.74189878e-01 1.35052562e-01
1.07870173e+00 -4.69813585e-01 -2.38896623e-01 1.53088361e-01
1.46563637e+00 3.30219597e-01 8.51905286e-01 2.57995993e-01
8.07550788e-01 8.62593532e-01 9.34422255e-01 4.75334585e-01
7.01505244e-01 6.29752815e-01 7.25926936e-01 -1.65500537e-01
-1.54551208e-01 -7.64426053e-01 4.78142202e-01 6.18722677e-01
-3.17055315e-01 -4.90694284e-01 -8.56211901e-01 8.84178817e-01
-2.07934093e+00 -1.40642428e+00 -1.83094554e-02 1.96644342e+00
7.97881305e-01 1.76969543e-01 1.15183584e-01 1.99844003e-01
6.80145800e-01 1.51435688e-01 -3.31679076e-01 -1.45350888e-01
-3.08914572e-01 -1.96241885e-01 -1.13718040e-01 2.41004780e-01
-8.36775899e-01 1.04958653e+00 5.44319868e+00 1.02917659e+00
-8.25282633e-01 -8.00545700e-03 6.87706947e-01 -2.01054871e-01
-6.93905056e-01 -2.70876847e-02 -5.39541185e-01 2.97333032e-01
3.99244398e-01 -4.23455350e-02 2.91509032e-01 5.55472434e-01
2.16465801e-01 -3.15523714e-01 -1.15095973e+00 1.06079853e+00
3.13473552e-01 -1.45541036e+00 5.87645590e-01 -1.20671224e-02
5.57648778e-01 -5.56302786e-01 -6.42154962e-02 3.92671168e-01
1.96524516e-01 -1.02324319e+00 1.17092824e+00 4.06906843e-01
8.13926995e-01 -9.30264950e-01 7.35423267e-01 4.77189243e-01
-1.16316068e+00 -1.30771130e-01 -1.83530375e-01 5.28889068e-04
2.52853066e-01 3.39607388e-01 -1.16089880e+00 6.78530991e-01
4.38816696e-01 8.90278518e-01 -4.43675011e-01 8.03718150e-01
-5.92668295e-01 4.95031774e-01 5.45365401e-02 -2.60203272e-01
2.74951369e-01 6.70770407e-02 5.43435276e-01 1.18432117e+00
5.78798234e-01 2.35824481e-01 -3.80442329e-02 1.09746170e+00
-1.34524822e-01 2.38230303e-01 -8.58220518e-01 -2.69990623e-01
3.78728628e-01 1.06552243e+00 -7.48016059e-01 -5.52106917e-01
-4.94915962e-01 7.90095329e-01 3.03510755e-01 4.77171838e-01
-9.68842506e-01 -2.69389331e-01 3.74299407e-01 1.46963984e-01
3.65952194e-01 2.89412867e-02 -4.76086885e-01 -1.22911251e+00
7.72621110e-02 -8.97789657e-01 4.83555079e-01 -1.26978695e+00
-1.02453411e+00 6.38606608e-01 3.50325406e-01 -1.29211080e+00
-4.87637579e-01 -2.52536207e-01 -4.94481236e-01 4.21185762e-01
-1.23451114e+00 -1.39713717e+00 -3.09456289e-01 6.23127162e-01
8.92306685e-01 -1.05963603e-01 5.67460775e-01 1.19081736e-01
-4.54272926e-01 4.31166798e-01 -5.19428670e-01 1.64365217e-01
4.65281457e-01 -1.01181531e+00 9.50326174e-02 1.02594578e+00
4.04429853e-01 4.83413577e-01 8.14362168e-01 -7.79443860e-01
-1.04867005e+00 -1.04442430e+00 1.05331540e+00 -2.83369213e-01
4.91218597e-01 -3.99752527e-01 -8.11390460e-01 6.85182095e-01
5.03092706e-01 -6.91192031e-01 7.91178107e-01 5.76041192e-02
-3.81122172e-01 1.35467976e-01 -7.93215454e-01 9.18977022e-01
1.15654504e+00 -3.31749678e-01 -8.16840768e-01 1.99842975e-01
8.00214767e-01 -1.51839852e-01 -5.46791494e-01 1.79810628e-01
5.08806169e-01 -8.45703542e-01 9.59017813e-01 -5.20138860e-01
1.22301352e+00 -3.00370246e-01 -2.76958108e-01 -1.29361498e+00
-2.83558995e-01 -2.03789085e-01 -3.58491428e-02 1.50525987e+00
6.99229717e-01 -6.80311322e-02 2.73862958e-01 3.90470505e-01
-1.96865559e-01 -7.24912584e-01 -6.64856911e-01 -4.14435506e-01
-3.42925310e-01 -7.95780361e-01 6.68614447e-01 9.89883125e-01
4.23224360e-01 8.12292576e-01 -6.73415184e-01 -1.20211586e-01
3.81510973e-01 3.46804023e-01 6.33554816e-01 -8.31663549e-01
-2.07191750e-01 -2.64508694e-01 -2.82611579e-01 -9.26084995e-01
8.80905762e-02 -8.20490241e-01 2.84476370e-01 -1.91559577e+00
6.04911387e-01 -1.47582948e-01 -8.55637118e-02 8.03908408e-01
-3.34047407e-01 1.45680562e-01 4.22145396e-01 1.24847591e-01
-7.75216162e-01 6.92584872e-01 1.50203776e+00 -3.70460391e-01
-6.49658293e-02 -4.05709654e-01 -8.97435546e-01 5.36265254e-01
6.13918841e-01 -2.20326945e-01 -7.52528727e-01 -5.15945673e-01
5.10339677e-01 2.01498270e-01 5.61486065e-01 -9.12984788e-01
-1.13052186e-02 -1.65434882e-01 4.13404256e-01 -6.14232898e-01
5.86836874e-01 -8.87466490e-01 2.57976532e-01 1.50259599e-01
-4.86016691e-01 -1.33973897e-01 1.51090503e-01 4.41650808e-01
-4.50238675e-01 -7.50310346e-02 4.08345968e-01 -9.99720395e-02
-9.35549021e-01 5.08752614e-02 -3.43668967e-01 1.73110869e-02
1.05205989e+00 -4.64132130e-01 -2.15778559e-01 -8.53301704e-01
-5.34257114e-01 3.20205271e-01 3.00251991e-01 5.47216058e-01
1.02560234e+00 -1.70467889e+00 -7.90658891e-01 -2.73411013e-02
5.15160978e-01 -1.53572569e-02 1.83518231e-01 6.00567043e-01
-1.88842639e-01 3.68076384e-01 -3.59033257e-01 -5.06011844e-01
-1.09477162e+00 6.80735528e-01 -8.18167180e-02 -5.20695113e-02
-6.91918492e-01 6.38958395e-01 5.20352960e-01 3.39466929e-01
2.88155705e-01 -3.05745959e-01 -5.60803711e-01 4.09396797e-01
5.55167615e-01 -5.67377321e-02 -2.72065550e-01 -8.17881227e-01
-8.21536854e-02 2.36712620e-01 9.56000462e-02 -4.04288560e-01
1.38599861e+00 -1.80756763e-01 1.94239959e-01 5.75729132e-01
7.70474732e-01 -2.50969023e-01 -1.14838815e+00 -2.86426932e-01
-1.36808738e-01 -4.19125348e-01 -2.14333043e-01 -8.42406511e-01
-9.30530787e-01 1.02477419e+00 8.66817012e-02 1.12253547e-01
1.27554905e+00 3.25340003e-01 7.95644164e-01 1.48853377e-01
3.32218438e-01 -8.30620885e-01 3.33145887e-01 2.80372173e-01
1.15169215e+00 -1.12766480e+00 -2.48392001e-01 -4.70198512e-01
-1.26330113e+00 9.89731133e-01 5.88793337e-01 3.27236474e-01
-1.35047942e-01 -5.88826928e-03 -1.73419699e-01 -2.72715479e-01
-9.79783118e-01 -4.15185124e-01 3.36506456e-01 4.71955091e-01
2.81020045e-01 5.49359508e-02 -1.97011888e-01 8.03732514e-01
-2.55874515e-01 -1.55529574e-01 3.74974012e-01 6.59544885e-01
-6.02566898e-01 -6.61724150e-01 -2.82422811e-01 2.90991902e-01
-8.61005336e-02 -1.63593411e-01 -5.96235812e-01 9.50394809e-01
3.61327589e-01 1.03315508e+00 -5.18068746e-02 -5.12146771e-01
3.48820239e-01 1.54508457e-01 3.65127027e-01 -5.47043145e-01
-3.71133864e-01 1.44395292e-01 4.49497521e-01 -3.89350057e-01
-5.51908910e-01 -5.43705940e-01 -1.32376182e+00 -2.21724391e-01
-2.03875676e-01 -3.39488499e-02 3.44487458e-01 1.11325300e+00
8.44708234e-02 7.09518790e-01 3.06674153e-01 -5.03425300e-01
1.22240722e-01 -9.27198231e-01 -2.41064385e-01 7.67072260e-01
3.90562266e-02 -7.48602390e-01 -6.72390172e-03 3.05875480e-01] | [11.090639114379883, 0.7280737161636353] |
857716f6-12d5-455a-af37-b9c338da47e4 | statistical-spatial-analysis-for-cryo | 2107.06738 | null | https://arxiv.org/abs/2107.06738v1 | https://arxiv.org/pdf/2107.06738v1.pdf | Statistical spatial analysis for cryo-electron tomography | Cryo-electron tomography (cryo-ET) is uniquely suited to precisely localize macromolecular complexes in situ, that is in a close-to-native state within their cellular compartments, in three-dimensions at high resolution. Point pattern analysis (PPA) allows quantitative characterization of the spatial organization of particles. However, current implementations of PPA functions are not suitable for applications to cryo-ET data because they do not consider the real, typically irregular 3D shape of cellular compartments and molecular complexes. Here, we designed and implemented first and the second-order, uni- and bivariate PPA functions in a Python package for statistical spatial analysis of particles located in three dimensional regions of arbitrary shape, such as those encountered in cellular cryo-ET imaging (PyOrg). To validate the implemented functions, we applied them to specially designed synthetic datasets. This allowed us to find the algorithmic solutions that provide the best accuracy and computational performance, and to evaluate the precision of the implemented functions. Applications to experimental data showed that despite the higher computational demand, the use of the second-order functions is advantageous to the first-order ones, because they allow characterization of the particle organization and statistical inference over a range of distance scales, as well as the comparative analysis between experimental groups comprising multiple tomograms. Altogether, PyOrg is a versatile, precise, and efficient open-source software for reliable quantitative characterization of macromolecular organization within cellular compartments imaged in situ by cryo-ET, as well as to other 3D imaging systems where real-size particles are located within regions possessing complex geometry. | ['Vladan Lučić', 'Wolfgang Baumeister', 'Antonio Martinez-Sanchez'] | 2021-07-14 | null | null | null | null | ['electron-tomography'] | ['medical'] | [-1.54563963e-01 -5.95302343e-01 6.12153471e-01 8.97763595e-02
-2.95375437e-01 -6.57396793e-01 5.57093561e-01 5.84061861e-01
-7.46324718e-01 9.33795452e-01 -4.65680212e-01 -3.26908857e-01
-3.62486869e-01 -9.18972552e-01 -5.05613863e-01 -1.13151455e+00
-2.18786463e-01 1.18987215e+00 4.66316760e-01 1.35507479e-01
2.88503140e-01 1.20006347e+00 -1.56033158e+00 -5.41411713e-02
4.68321979e-01 7.31094301e-01 7.64184415e-01 7.13151872e-01
-1.23272397e-01 3.44201699e-02 -2.42860094e-01 9.17252973e-02
-1.25954494e-01 -1.37997568e-01 -4.26343143e-01 3.42215784e-02
-2.92715095e-02 2.62387067e-01 4.46088254e-01 6.10191882e-01
5.30812502e-01 -8.70384574e-02 8.80734742e-01 -5.11642098e-01
-1.87633589e-01 -2.40923136e-01 4.59134877e-02 3.76515359e-01
4.84582871e-01 9.97518227e-02 3.49429101e-01 -9.84129250e-01
1.08187747e+00 1.02903605e+00 7.60995507e-01 1.46707878e-01
-1.87347722e+00 -2.80091781e-02 -4.69409496e-01 -8.32176283e-02
-1.22480845e+00 -2.06791043e-01 2.76368409e-01 -7.75290847e-01
7.97110736e-01 4.07067627e-01 6.87919736e-01 7.29008436e-01
6.94254041e-01 -1.82351455e-01 1.71488440e+00 -2.93790638e-01
5.48088670e-01 6.69600675e-03 1.01819605e-01 4.38951313e-01
3.51670265e-01 -1.60251260e-01 -3.57790254e-02 -5.20479262e-01
8.16627145e-01 8.54209960e-02 -2.16360390e-01 -6.86151206e-01
-1.46060395e+00 3.76446575e-01 -1.25841200e-01 9.11872447e-01
-5.00651777e-01 -3.00299734e-01 4.57294673e-01 -1.96425036e-01
2.62626827e-01 6.46216393e-01 -5.88635623e-01 -3.10826510e-01
-8.28623235e-01 4.62949067e-01 6.82105303e-01 3.27498704e-01
8.44085336e-01 -4.54599828e-01 2.83551306e-01 4.67530072e-01
1.70855835e-01 4.80924547e-01 4.86825049e-01 -8.89351308e-01
-1.53953629e-03 5.73191226e-01 3.00053567e-01 -9.60876942e-01
-6.67295516e-01 -1.20242193e-01 -8.26332688e-01 4.23526555e-01
9.30222809e-01 2.43749514e-01 -3.60808700e-01 1.30106688e+00
5.60919642e-01 -4.15918469e-01 -6.81298599e-02 8.57277155e-01
4.61798340e-01 5.15755296e-01 -4.84106923e-03 -6.66183710e-01
1.62305534e+00 3.79070863e-02 -4.60644811e-01 4.71131146e-01
5.41049719e-01 -7.68791080e-01 1.01051033e+00 1.73329085e-01
-9.97527122e-01 -3.28351736e-01 -6.17648900e-01 1.93902835e-01
-5.37538350e-01 1.26410857e-01 3.97367001e-01 5.10339260e-01
-8.35840762e-01 1.00706387e+00 -1.07236910e+00 -7.30019331e-01
1.89100951e-01 1.82714745e-01 -8.56985629e-01 2.72331387e-01
-5.61851084e-01 9.13646936e-01 1.25902683e-01 1.02260448e-01
-5.43732703e-01 -7.33107150e-01 -5.65135658e-01 1.51129022e-01
-1.23115502e-01 -6.56032205e-01 4.89268661e-01 -9.19811055e-02
-1.52932966e+00 1.12910545e+00 -3.68613422e-01 -1.37611777e-01
4.29597229e-01 4.29480493e-01 6.99586049e-02 2.50608861e-01
8.77910927e-02 3.42609249e-02 4.68376540e-02 -1.31420112e+00
1.64636046e-01 -7.27678001e-01 -4.80404973e-01 -1.44625366e-01
1.65798113e-01 4.71964628e-02 1.85627669e-01 -3.90155837e-02
5.18378317e-01 -6.23242319e-01 -2.01986775e-01 -5.27312793e-02
-1.88268438e-01 1.33921146e-01 8.64391863e-01 -3.89644295e-01
6.01762712e-01 -2.03281498e+00 3.67474914e-01 2.70238012e-01
2.51922548e-01 2.15796635e-01 5.56810021e-01 9.17023361e-01
-1.06489591e-01 3.48441824e-02 -2.53145039e-01 -4.21125114e-01
-9.44747180e-02 1.23249739e-01 1.44494653e-01 9.65353429e-01
-2.33253315e-01 5.89795649e-01 -5.96950412e-01 -5.85810363e-01
5.19188166e-01 5.52698612e-01 -1.91657037e-01 1.01339519e-01
-1.53669551e-01 8.57759118e-01 -4.03429776e-01 4.42629367e-01
8.57965648e-01 -3.35469186e-01 5.97051978e-01 -1.22531384e-01
-8.84738564e-01 -1.23289367e-02 -9.37945366e-01 1.11623740e+00
-1.50093868e-01 3.66423696e-01 4.77803826e-01 -7.53261328e-01
1.08858740e+00 2.80310363e-01 6.45259380e-01 -6.49086237e-01
1.22286968e-01 3.49483639e-01 -1.42093077e-01 -3.07552844e-01
1.44133583e-01 -5.93645692e-01 3.53927404e-01 3.17082196e-01
-1.74393624e-01 3.00446548e-03 4.57185090e-01 -1.65843189e-01
9.88259375e-01 -1.11458860e-02 3.63481253e-01 -9.81992126e-01
7.12367058e-01 -1.34104341e-01 2.23739713e-01 2.76657641e-01
3.86083126e-02 7.69932687e-01 7.39763379e-01 -6.40812695e-01
-1.50617528e+00 -1.11981523e+00 -8.00681472e-01 4.55455095e-01
1.32419199e-01 -2.77880877e-01 -7.46279955e-01 1.53229237e-01
4.93025705e-02 -5.26851378e-02 -6.06219053e-01 6.27840579e-01
-3.77572179e-01 -1.03503668e+00 2.21428394e-01 -1.71478674e-01
3.21572199e-02 -1.07751143e+00 -8.45283091e-01 3.34156752e-01
1.26186997e-01 -1.14257002e+00 2.89771408e-01 3.32275242e-01
-9.88182127e-01 -1.45639896e+00 -8.61491442e-01 -2.79658914e-01
7.46884942e-01 3.02013624e-02 8.51103306e-01 1.60922915e-01
-4.66147184e-01 2.57532358e-01 -1.88332185e-01 2.10269183e-01
-5.04126191e-01 -1.95836976e-01 3.29670310e-01 -1.56779498e-01
-4.91168536e-02 -9.76845980e-01 -5.11758983e-01 7.21640110e-01
-8.94864619e-01 -3.27043653e-01 1.79647148e-01 7.07347095e-01
1.06286931e+00 5.12343347e-02 1.53455883e-01 -7.67653883e-01
5.34260929e-01 -2.79419333e-01 -9.09909010e-01 4.23229523e-02
-3.50271255e-01 -9.53044146e-02 1.12136793e+00 -1.24857053e-01
-6.71754181e-01 -6.57315031e-02 -2.87844151e-01 -3.09455022e-02
-5.99898100e-01 4.31878239e-01 -2.24898711e-01 -1.97122917e-01
5.30663908e-01 5.48372746e-01 4.48295653e-01 -6.22491241e-01
-4.54414070e-01 2.88282424e-01 1.82230994e-01 -9.74727809e-01
3.00999641e-01 8.74897718e-01 6.20689452e-01 -1.18550944e+00
1.99163314e-02 -5.12882411e-01 -8.13501120e-01 -1.36202723e-01
8.66500199e-01 -4.00697142e-01 -1.34819388e+00 4.24216509e-01
-9.01135147e-01 -3.80485564e-01 -1.35472491e-01 5.29017925e-01
-8.25645506e-01 8.17759216e-01 -5.76657832e-01 -8.37131321e-01
-1.19641706e-01 -1.42126644e+00 1.11320865e+00 1.36111518e-02
-7.37944543e-02 -1.09914589e+00 4.30484653e-01 2.01772153e-03
4.02709782e-01 4.81042832e-01 1.16095579e+00 -1.61306709e-01
-5.00514567e-01 4.31942055e-03 6.75068647e-02 -2.15991184e-01
-4.41358313e-02 4.96962637e-01 -6.86278403e-01 -3.66381943e-01
4.62264419e-02 7.51721635e-02 4.18151677e-01 5.43633640e-01
9.19320345e-01 1.85654730e-01 -5.42473972e-01 4.92644876e-01
1.49973798e+00 7.74645507e-02 9.42027748e-01 4.38721746e-01
2.64014546e-02 7.99045384e-01 5.56034803e-01 6.48169339e-01
-5.90091981e-02 9.97414470e-01 4.17924255e-01 7.11087063e-02
3.38509530e-01 1.68778151e-01 3.47104692e-03 6.05149686e-01
-4.49800789e-01 1.07784634e-02 -8.56666028e-01 2.49080181e-01
-1.58672118e+00 -1.13952065e+00 -6.47090614e-01 2.48331618e+00
4.74967241e-01 -8.58732760e-02 3.22836310e-01 1.57032281e-01
6.62584484e-01 -1.98948100e-01 -7.58230537e-02 -3.14020932e-01
-4.73054409e-01 1.82907566e-01 2.28381991e-01 3.60694975e-01
-7.02280581e-01 2.66668469e-01 6.72258091e+00 7.45576084e-01
-1.22588098e+00 6.66188523e-02 3.62827957e-01 2.64340907e-01
-7.17131272e-02 2.51250118e-01 -8.97766709e-01 9.01351571e-01
1.02056980e+00 1.07459784e-01 3.09912235e-01 5.88534594e-01
5.73223352e-01 -5.93925834e-01 -8.49224746e-01 7.00031996e-01
-6.18799567e-01 -1.75635457e+00 -2.77188987e-01 5.49296319e-01
2.34756470e-01 -1.08677195e-02 -3.68489176e-01 -4.54696715e-01
-4.28538114e-01 -8.33511055e-01 4.30675328e-01 8.27892780e-01
7.38410771e-01 -5.12357414e-01 8.97351027e-01 6.32632554e-01
-1.08038056e+00 3.53841245e-01 -7.54572272e-01 -7.61911795e-02
3.60110998e-01 1.09173131e+00 -8.03805590e-01 4.44085270e-01
7.03498006e-01 1.50819287e-01 -3.78344506e-01 8.74333620e-01
4.40423548e-01 1.46440715e-01 -6.12264097e-01 -2.14409545e-01
-1.64948255e-01 -8.68514836e-01 4.78560120e-01 1.21745372e+00
3.61171007e-01 -1.36020128e-03 -3.24975461e-01 1.08029079e+00
4.77673411e-01 1.57721221e-01 -5.89400172e-01 -1.38698652e-01
3.32252800e-01 1.45239353e+00 -1.31014109e+00 4.01889943e-02
5.23856245e-02 3.90263021e-01 4.73526090e-01 1.61878169e-01
-4.40735430e-01 -7.60901794e-02 7.83785105e-01 8.69483888e-01
3.38366419e-01 -7.79172778e-01 -1.30666122e-01 -8.45055163e-01
-1.05171148e-02 -3.94107163e-01 -6.59193918e-02 -8.62953603e-01
-1.19419432e+00 4.16840136e-01 6.16093576e-02 -7.56830275e-01
1.60756540e-02 -1.08126092e+00 -4.98276621e-01 8.49845648e-01
-9.33135808e-01 -7.95722723e-01 -1.07530870e-01 2.33759224e-01
-2.03582913e-01 1.62962049e-01 1.19200373e+00 2.23969847e-01
-3.17089647e-01 -3.11129600e-01 7.30856717e-01 -4.83429044e-01
3.49015057e-01 -1.33666432e+00 -1.25817150e-01 3.48110259e-01
-2.79428363e-01 1.00530887e+00 1.07512486e+00 -6.51677549e-01
-1.36520290e+00 -4.71336663e-01 7.98451483e-01 -4.68859494e-01
4.87968773e-01 -6.17537320e-01 -9.95664001e-01 2.42126063e-01
-1.73775852e-01 2.09693283e-01 7.58460402e-01 -3.32565010e-02
2.21502587e-01 2.09015086e-01 -1.38633990e+00 2.46537149e-01
6.04389191e-01 -3.95871729e-01 -2.26825550e-01 4.93240207e-01
-2.73029394e-02 -2.38137305e-01 -1.58907568e+00 3.76096010e-01
7.02462673e-01 -1.57019556e+00 1.01609719e+00 -2.60478377e-01
1.77315012e-01 -6.59420967e-01 -8.18926990e-02 -8.82778525e-01
-3.46129686e-01 -3.00534844e-01 3.91225666e-01 8.65813434e-01
2.17020288e-01 -8.40292096e-01 8.03052008e-01 3.01398933e-01
-1.40814960e-01 -9.70291257e-01 -1.29556525e+00 -6.39160097e-01
2.05039103e-02 1.12533882e-01 3.24271917e-01 7.36295402e-01
-1.26427934e-02 -1.66211143e-01 2.46599063e-01 1.76539034e-01
7.08645880e-01 3.81461799e-01 8.80535662e-01 -1.46466970e+00
-2.66098350e-01 -1.49477527e-01 -7.99599707e-01 -5.35186052e-01
1.65310010e-01 -3.94697070e-01 -1.51662707e-01 -1.26783431e+00
2.92652637e-01 -6.50696158e-01 3.56408864e-01 -1.97953954e-01
3.62904876e-01 2.36388460e-01 -1.10336594e-01 4.76169080e-01
-3.64352137e-01 4.05620635e-01 1.41144311e+00 5.04445791e-01
1.55298471e-01 -2.41710022e-02 1.11149959e-01 5.07671297e-01
4.37033117e-01 -3.72447908e-01 1.65033370e-01 2.50078559e-01
2.93233246e-01 3.99634391e-02 5.95933139e-01 -1.18140888e+00
2.65445858e-01 -1.27298966e-01 5.21606326e-01 -8.41964424e-01
5.48565447e-01 -9.46053684e-01 8.04957390e-01 3.62719983e-01
3.68744463e-01 1.37360230e-01 5.50332442e-02 3.40159744e-01
-1.25272736e-01 -3.77716005e-01 1.14148796e+00 -4.61099863e-01
-1.35849074e-01 -1.23495115e-02 -8.72575283e-01 -4.07156408e-01
1.13612533e+00 -6.30731642e-01 -5.38925946e-01 1.32368967e-01
-1.08904588e+00 -3.16068977e-01 1.34515750e+00 -6.97908282e-01
4.97308701e-01 -8.68375659e-01 -2.07011595e-01 2.53286839e-01
1.26616592e-02 6.76860064e-02 6.32979989e-01 1.22467864e+00
-1.19526589e+00 6.22475326e-01 -6.10513568e-01 -9.60790277e-01
-1.29911578e+00 5.38026869e-01 5.75731933e-01 -4.50235426e-01
-5.66674113e-01 5.23933806e-02 1.99770033e-01 -6.60813272e-01
-4.88429219e-01 -3.49622667e-01 -2.09647343e-01 -1.46703660e-01
3.73182297e-01 4.12678212e-01 3.46792400e-01 -7.16739118e-01
-3.99110168e-01 9.35684681e-01 3.54062051e-01 1.92794845e-01
1.46681702e+00 -3.06142837e-01 -7.13515759e-01 6.33323252e-01
9.79617596e-01 1.63708180e-01 -1.16143727e+00 2.76242971e-01
-1.11862183e-01 -3.97076845e-01 -3.60547572e-01 -3.24088812e-01
-4.44028646e-01 7.89267659e-01 3.00549775e-01 6.34843230e-01
6.86554551e-01 2.21694946e-01 1.78817213e-01 3.00032526e-01
7.76083350e-01 -6.58378243e-01 -3.69049013e-01 4.78501230e-01
6.18070900e-01 -5.79121172e-01 3.22504491e-01 -6.17420256e-01
1.44531680e-02 1.25406134e+00 1.78523183e-01 -1.20020717e-01
4.36040491e-01 5.28297126e-01 -2.64211744e-01 -4.98239875e-01
-7.35133648e-01 -7.51352161e-02 -4.03763592e-01 7.61199117e-01
4.78052735e-01 -6.36946335e-02 -5.70114970e-01 1.37879238e-01
-7.52508128e-03 -1.70044452e-01 5.87394655e-01 8.35528493e-01
-4.68114704e-01 -1.27044368e+00 -7.24179447e-01 2.34406576e-01
-4.99908447e-01 3.05330008e-01 -2.89154887e-01 9.94052708e-01
4.42174375e-02 3.67247373e-01 2.31460929e-01 1.14289507e-01
2.88922042e-01 7.98142031e-02 6.52107596e-01 -3.85289013e-01
-4.48006392e-01 6.82862289e-03 -8.57000500e-02 -5.12561917e-01
-6.28895164e-01 -8.33711088e-01 -1.33002818e+00 -5.32602966e-01
-2.58253366e-01 6.51599705e-01 1.05555868e+00 9.97404277e-01
4.94159818e-01 1.41478479e-01 4.14649665e-01 -1.33085120e+00
8.46287329e-03 -9.62092936e-01 -1.23345041e+00 3.03371310e-01
7.62440711e-02 -9.20617759e-01 -4.84660685e-01 -1.12724110e-01] | [13.416158676147461, -3.05718731880188] |
a60765b2-846b-4227-9f67-41162584ba84 | starvqa-co-training-space-time-attention-for | 2306.12298 | null | https://arxiv.org/abs/2306.12298v1 | https://arxiv.org/pdf/2306.12298v1.pdf | StarVQA+: Co-training Space-Time Attention for Video Quality Assessment | Self-attention based Transformer has achieved great success in many computer vision tasks. However, its application to video quality assessment (VQA) has not been satisfactory so far. Evaluating the quality of in-the-wild videos is challenging due to the unknown of pristine reference and shooting distortion. This paper presents a co-trained Space-Time Attention network for the VQA problem, termed StarVQA+. Specifically, we first build StarVQA+ by alternately concatenating the divided space-time attention. Then, to facilitate the training of StarVQA+, we design a vectorized regression loss by encoding the mean opinion score (MOS) to the probability vector and embedding a special token as the learnable variable of MOS, leading to better fitting of human's rating process. Finally, to solve the data hungry problem with Transformer, we propose to co-train the spatial and temporal attention weights using both images and videos. Various experiments are conducted on the de-facto in-the-wild video datasets, including LIVE-Qualcomm, LIVE-VQC, KoNViD-1k, YouTube-UGC, LSVQ, LSVQ-1080p, and DVL2021. Experimental results demonstrate the superiority of the proposed StarVQA+ over the state-of-the-art. | ['Sam Kwong', 'Guopu Zhu', 'Weixuan Tang', 'Yuan-Gen Wang', 'Fengchuang Xing'] | 2023-06-21 | null | null | null | null | ['video-quality-assessment', 'video-quality-assessment'] | ['computer-vision', 'time-series'] | [-3.24333876e-01 -5.77918589e-01 1.80278942e-01 -4.21484828e-01
-9.87496018e-01 -8.10351744e-02 9.93511230e-02 -1.73069745e-01
-3.21723491e-01 4.56153661e-01 4.52893823e-01 -7.35342205e-02
-4.67322730e-02 -5.02878487e-01 -6.76685035e-01 -5.99166095e-01
-5.73594794e-02 -7.50276670e-02 6.76056370e-02 -2.21828625e-01
4.69300933e-02 4.85976934e-02 -1.25400054e+00 2.00822458e-01
8.36039186e-01 1.58283341e+00 1.77997977e-01 7.63618410e-01
2.20790282e-01 1.06466544e+00 -5.24641931e-01 -9.77241039e-01
3.23893517e-01 -6.02517188e-01 -5.71423113e-01 2.54147470e-01
5.25918067e-01 -6.24397933e-01 -7.93011129e-01 1.02600300e+00
7.73355722e-01 1.43117145e-01 3.30611140e-01 -1.61659324e+00
-1.00142264e+00 1.10121153e-01 -5.26650071e-01 7.16821551e-01
3.16208631e-01 4.29767191e-01 1.37508154e+00 -9.72855151e-01
3.53291094e-01 1.40558720e+00 4.66361880e-01 4.29194927e-01
-7.37001777e-01 -6.83024585e-01 -1.26756402e-02 1.01950622e+00
-1.48202872e+00 -3.83484781e-01 8.30722988e-01 -3.03791463e-01
6.82339370e-01 2.40436327e-02 6.83137357e-01 9.97244716e-01
3.11733454e-01 8.97478819e-01 7.67089486e-01 6.39665276e-02
1.28949508e-01 -6.51683360e-02 -1.60889059e-01 6.94639981e-01
-4.17518646e-01 5.73991856e-04 -5.42419016e-01 1.64450064e-01
8.35212052e-01 6.39034063e-02 -5.14022052e-01 -2.77734190e-01
-1.10403740e+00 6.92995846e-01 5.93033552e-01 -7.73724467e-02
-5.45978546e-01 2.80605167e-01 6.95498228e-01 6.22045457e-01
4.38254684e-01 1.34109125e-01 -4.42563176e-01 -4.24173057e-01
-8.92191291e-01 1.24979869e-01 8.72220472e-02 9.62333083e-01
2.44225994e-01 3.91621917e-01 -5.79247653e-01 7.90182412e-01
4.31698054e-01 5.77506781e-01 5.13887644e-01 -1.30509651e+00
6.35909617e-01 1.22972533e-01 3.31111550e-01 -1.12259948e+00
1.22761510e-01 -4.47647214e-01 -8.91610146e-01 1.87429711e-01
7.74273574e-02 1.47904968e-02 -7.24045217e-01 1.53697431e+00
-2.65252125e-02 4.43445086e-01 4.25309427e-02 1.33867025e+00
8.77748013e-01 1.18799114e+00 -5.71031831e-02 -5.01841366e-01
1.12940860e+00 -1.24948692e+00 -1.14419639e+00 3.09504271e-01
1.67771261e-02 -4.98083174e-01 1.27626920e+00 6.61117136e-01
-1.37519407e+00 -1.14565456e+00 -1.14126384e+00 -2.20371112e-01
1.35233089e-01 -3.39309834e-02 1.79408237e-01 2.84900695e-01
-1.08762729e+00 6.34309769e-01 -5.86606205e-01 -7.23090023e-03
7.05243528e-01 2.03617826e-01 -1.80176958e-01 -2.30853662e-01
-1.37654769e+00 5.40555477e-01 -6.13156483e-02 6.74753264e-02
-1.46313560e+00 -6.86616302e-01 -6.92738593e-01 3.25340271e-01
4.52654660e-01 -4.80148643e-01 1.11501849e+00 -1.23322630e+00
-1.57192588e+00 5.08485138e-01 1.27614047e-02 -4.65542912e-01
3.54842514e-01 -4.38520432e-01 -7.15292692e-01 2.40765542e-01
2.24415027e-02 5.47548950e-01 9.74573493e-01 -1.05894291e+00
-7.38050222e-01 -2.26972371e-01 3.99322093e-01 3.80600035e-01
-3.86188805e-01 1.60395488e-01 -9.10628021e-01 -7.19146609e-01
-4.16080892e-01 -4.72916037e-01 8.69575441e-02 3.95547062e-01
1.17330607e-02 -3.64973843e-01 8.04719269e-01 -9.91713226e-01
1.42739606e+00 -2.42903137e+00 3.63708764e-01 -2.20119491e-01
3.52594972e-01 4.93681312e-01 -5.20674646e-01 5.40708080e-02
3.49771976e-02 -8.55897367e-02 1.16871923e-01 -3.13805729e-01
-1.01787575e-01 1.96682483e-01 -1.11827977e-01 4.95457888e-01
3.39333206e-01 9.79076087e-01 -8.99654329e-01 -6.14873827e-01
2.33585000e-01 5.85481763e-01 -7.19327152e-01 6.52073026e-01
9.97966975e-02 2.21769020e-01 -2.91688085e-01 6.92113280e-01
5.39108455e-01 -2.96344787e-01 -3.91506255e-01 -6.44232988e-01
-2.87351552e-02 -5.01946434e-02 -9.74410653e-01 1.88126349e+00
-4.30188864e-01 7.77858973e-01 3.68106808e-03 -9.07356799e-01
7.09819138e-01 5.75495124e-01 5.41737914e-01 -1.15065193e+00
4.34758544e-01 3.45692374e-02 -6.70767725e-02 -8.65171313e-01
3.45870525e-01 4.74707969e-02 3.10301244e-01 -9.31262597e-03
5.89411795e-01 1.66602358e-01 2.05072194e-01 2.96623737e-01
9.68357742e-01 1.11888580e-01 7.57721141e-02 1.30851880e-01
7.05736995e-01 -5.92504740e-01 9.01807368e-01 1.88401461e-01
-7.00535476e-01 8.52971733e-01 4.70634282e-01 -1.99091703e-01
-1.22318435e+00 -1.25950789e+00 1.49282485e-01 9.67645109e-01
3.43872845e-01 -4.32747066e-01 -7.27017939e-01 -7.25019872e-01
-3.32639068e-01 3.52186233e-01 -6.14907742e-01 -2.01834068e-01
-2.35369712e-01 -2.80477583e-01 1.01566523e-01 5.33033788e-01
8.44488680e-01 -1.13667285e+00 -3.87597740e-01 3.67705673e-01
-5.36856592e-01 -1.08264494e+00 -9.19181645e-01 -4.51738983e-01
-5.83239675e-01 -9.52881753e-01 -1.06974423e+00 -5.59276700e-01
8.15148428e-02 4.24264550e-01 1.23840916e+00 9.13257077e-02
-8.94393679e-03 4.29748207e-01 -6.40229881e-01 -1.58896238e-01
1.30640209e-01 -3.88102889e-01 1.97982714e-02 4.73675191e-01
1.63299918e-01 -4.77141112e-01 -8.09900463e-01 3.86281580e-01
-6.91510022e-01 -3.14713359e-01 3.71995956e-01 9.30210710e-01
7.15085745e-01 8.10069367e-02 5.84135234e-01 -2.19923869e-01
4.58279014e-01 -5.27744889e-01 -4.03821319e-01 1.45692229e-01
-5.55925786e-01 -3.06246638e-01 7.05074668e-01 -3.47456276e-01
-8.76354158e-01 -4.62570757e-01 -4.14193064e-01 -1.14909005e+00
2.77343959e-01 3.61498296e-01 -6.55986369e-01 1.58596545e-01
2.19538942e-01 2.22636491e-01 -4.80044410e-02 -3.80021423e-01
3.20274949e-01 7.43247807e-01 7.75382280e-01 -1.25276700e-01
7.81591356e-01 2.28036538e-01 -3.45829546e-01 -5.31477749e-01
-8.17895114e-01 -4.44382548e-01 -4.22033593e-02 -4.06867266e-01
1.23366785e+00 -1.20865631e+00 -7.16447234e-01 5.38549483e-01
-1.17199934e+00 -1.44796580e-01 -4.23827231e-01 4.99211341e-01
-6.79275870e-01 4.94036645e-01 -7.62246430e-01 -6.14253461e-01
-5.37089288e-01 -1.38667858e+00 8.22503626e-01 2.96295106e-01
4.22543555e-01 -6.41959965e-01 1.51436292e-02 5.59592366e-01
4.19262469e-01 -6.38201535e-02 6.01314485e-01 -1.47069618e-01
-6.45242274e-01 6.33175969e-02 -5.34703016e-01 8.99208486e-01
-5.42919114e-02 -5.30047789e-02 -8.29799652e-01 -4.24151868e-01
-1.48372147e-02 -5.71164370e-01 6.20040238e-01 5.30823946e-01
1.42800784e+00 -3.15466166e-01 4.56672609e-01 9.30145800e-01
1.46343708e+00 4.81771827e-01 9.67964590e-01 1.83504373e-01
7.72646666e-01 1.18496969e-01 8.53975594e-01 5.59931874e-01
5.99907219e-01 8.35816801e-01 8.11843634e-01 -1.37051478e-01
-2.63755739e-01 -7.32328966e-02 5.77819288e-01 1.22619355e+00
-2.84835041e-01 -6.25147581e-01 -2.82114565e-01 7.22637057e-01
-1.66889989e+00 -1.05560219e+00 -1.05436422e-01 2.02496505e+00
4.76723880e-01 1.00792512e-01 9.40049291e-02 4.31042463e-01
3.96821558e-01 3.42170030e-01 -5.27500868e-01 -4.12432939e-01
-1.12187840e-01 9.69858989e-02 1.24593817e-01 2.66788155e-01
-1.04024673e+00 4.93187338e-01 5.09875202e+00 1.00240457e+00
-1.12547469e+00 5.12146711e-01 7.29684114e-01 -2.80711889e-01
-1.45979077e-01 -4.03695703e-01 -1.25043973e-01 8.42115879e-01
9.57547903e-01 -1.55958891e-01 6.43655896e-01 6.89259350e-01
4.99418050e-01 3.79707783e-01 -9.94553030e-01 1.54834378e+00
2.89484620e-01 -1.05480134e+00 1.00260423e-02 -1.33422017e-01
6.26187384e-01 2.33031288e-02 3.13457996e-01 5.41575730e-01
-1.35625124e-01 -8.74316037e-01 9.01997328e-01 5.28639317e-01
1.09536123e+00 -8.05485189e-01 1.06753147e+00 -2.20848456e-01
-1.29462492e+00 -3.10559809e-01 -5.88807762e-01 1.77132174e-01
4.36703324e-01 3.23346108e-01 1.65818155e-01 6.14341617e-01
1.21435905e+00 1.15540016e+00 -5.41315317e-01 1.12277317e+00
-3.59174535e-02 7.36590326e-01 2.17449442e-01 3.23855698e-01
3.28851730e-01 -1.57756567e-01 4.85344559e-01 7.89972901e-01
6.13601685e-01 4.70157146e-01 -8.38023499e-02 5.67306757e-01
-3.87175441e-01 1.72831297e-01 -7.25434273e-02 2.61163525e-02
1.05920821e-01 1.13208747e+00 6.28654286e-02 -4.18174922e-01
-8.07325721e-01 1.32511663e+00 5.10436930e-02 4.82614905e-01
-1.30609798e+00 -4.16968882e-01 9.05603111e-01 4.60472219e-02
6.39510095e-01 1.09080173e-01 3.32905412e-01 -1.28995001e+00
9.33932513e-02 -1.08113468e+00 1.90208972e-01 -1.31487644e+00
-1.40943718e+00 8.02819669e-01 -3.30044985e-01 -1.49890065e+00
2.19816893e-01 -3.84292096e-01 -7.14105189e-01 7.42544115e-01
-1.68283951e+00 -8.64677250e-01 -5.24895847e-01 1.02150381e+00
9.47984338e-01 -3.99992615e-01 3.79212677e-01 9.47998464e-01
-5.91251016e-01 8.49417627e-01 1.77288309e-01 1.54794872e-01
8.71603668e-01 -1.12357986e+00 1.94702864e-01 7.81138182e-01
6.37407741e-03 -4.97076884e-02 7.25956678e-01 -1.03969119e-01
-1.40283465e+00 -1.24093485e+00 5.06710947e-01 -1.27166644e-01
5.44049025e-01 1.04791485e-02 -1.01666176e+00 3.41711462e-01
5.49041927e-01 4.92707402e-01 3.70229840e-01 -4.10052121e-01
-3.07782650e-01 -6.35280252e-01 -1.10448706e+00 2.94722944e-01
9.74976122e-01 -5.62353969e-01 -3.69387001e-01 1.57351121e-02
1.08976436e+00 -1.72239944e-01 -1.07736802e+00 2.03112170e-01
4.91335243e-01 -1.04235685e+00 9.90595818e-01 -5.76030731e-01
8.24226677e-01 -3.31157595e-01 -5.53874195e-01 -1.51007557e+00
-6.17732584e-01 -3.57644320e-01 -2.85882443e-01 1.47687745e+00
-6.51752725e-02 -8.55455846e-02 5.89910686e-01 2.10079893e-01
-3.13478798e-01 -8.10280144e-01 -1.03345191e+00 -6.11947477e-01
-2.23425776e-01 -4.20620173e-01 6.07920051e-01 7.50617146e-01
-5.04326761e-01 4.12904859e-01 -8.97340000e-01 9.47194099e-02
6.90220356e-01 -3.29949826e-01 6.07213795e-01 -8.82376671e-01
-4.94367391e-01 -2.15496987e-01 -8.08737397e-01 -1.22564280e+00
-2.19435975e-01 -3.68606776e-01 2.23165005e-03 -1.41970313e+00
2.80951351e-01 -4.67311628e-02 -7.59137809e-01 4.18013409e-02
-3.65392685e-01 2.85413891e-01 6.02433801e-01 3.86018828e-02
-1.15601182e+00 1.10745704e+00 1.45310140e+00 -4.57066417e-01
1.78096101e-01 -3.44593585e-01 -4.95740116e-01 3.19095314e-01
4.58827734e-01 -2.46560872e-01 -6.12734079e-01 -7.54706740e-01
1.31130621e-01 4.80282307e-01 2.84809321e-01 -1.21088600e+00
-8.76089483e-02 -8.99222717e-02 1.56520650e-01 -5.62065542e-01
5.20389915e-01 -9.31289256e-01 -8.79469588e-02 1.13997869e-01
-2.42099628e-01 3.29436988e-01 -9.69341695e-02 6.17911041e-01
-6.15687311e-01 8.77917632e-02 8.13187838e-01 6.17812574e-02
-9.34528649e-01 9.34428871e-01 -5.91864325e-02 3.74345869e-01
8.29022586e-01 -1.15319090e-02 -1.45736396e-01 -6.89937294e-01
-4.60460901e-01 5.32875061e-01 2.79368669e-01 5.04137814e-01
1.13433015e+00 -1.76000500e+00 -1.06030583e+00 3.35149206e-02
2.48436943e-01 -3.29265952e-01 8.21160495e-01 6.65130079e-01
-4.83428895e-01 6.02748208e-02 -5.47784865e-01 -5.49405575e-01
-1.26977384e+00 9.20144081e-01 2.58837789e-01 -3.06194186e-01
-3.33987951e-01 9.52937245e-01 2.07603276e-01 1.43083647e-01
4.13728714e-01 -1.43306673e-01 -4.03808028e-01 -1.33674771e-01
7.36190975e-01 4.95061934e-01 -8.34603533e-02 -8.52648795e-01
-1.82470083e-01 5.24731576e-01 1.30616844e-01 -1.80181246e-02
1.29108262e+00 -6.10629559e-01 2.74828881e-01 3.81741703e-01
1.57922888e+00 -3.24359000e-01 -1.60058701e+00 -3.93789172e-01
-6.84526563e-01 -8.26283991e-01 2.80789286e-01 -6.51970446e-01
-1.72897446e+00 1.28513682e+00 1.18705344e+00 -1.58884991e-02
1.47605646e+00 -2.59246618e-01 1.15201318e+00 -8.32581371e-02
3.33584815e-01 -9.21877980e-01 4.65182155e-01 2.13795096e-01
1.12578368e+00 -1.53382599e+00 -1.82558164e-01 7.28783086e-02
-9.16119695e-01 6.68224990e-01 6.96595073e-01 -1.25707582e-01
5.71429372e-01 -3.44098210e-01 2.26491615e-01 7.78236007e-03
-9.10829782e-01 1.23740463e-02 4.22575891e-01 6.62029147e-01
2.45037392e-01 -9.08023939e-02 -1.53805479e-01 7.58774579e-01
1.66298270e-01 2.44585574e-01 6.67052269e-01 4.35882956e-01
-1.90301821e-01 -6.65734172e-01 -2.33790994e-01 4.22688276e-01
-8.33230078e-01 -1.05507888e-01 3.28259289e-01 3.33156466e-01
3.92206967e-01 1.14239681e+00 1.26960903e-01 -6.79056406e-01
5.67659080e-01 -4.65088725e-01 2.47802958e-01 -1.09099038e-01
-4.58355337e-01 1.09369598e-01 -1.47336766e-01 -9.46747243e-01
-4.57803100e-01 -4.62630033e-01 -8.76231134e-01 -3.46110165e-01
-1.53057829e-01 2.81730413e-01 3.91935408e-01 7.37816989e-01
2.25240618e-01 8.95569563e-01 1.04194272e+00 -8.26936543e-01
-4.54190671e-01 -8.85689139e-01 -7.21232474e-01 7.39026427e-01
5.17017365e-01 -7.10136235e-01 -2.07430825e-01 2.24770531e-01] | [11.662569999694824, -1.7746471166610718] |
ff2594be-e9b1-48f3-8370-6dcfc2412787 | positive-and-unlabeled-learning-through | 1805.07331 | null | http://arxiv.org/abs/1805.07331v2 | http://arxiv.org/pdf/1805.07331v2.pdf | Positive and Unlabeled Learning through Negative Selection and Imbalance-aware Classification | Motivated by applications in protein function prediction, we consider a
challenging supervised classification setting in which positive labels are
scarce and there are no explicit negative labels. The learning algorithm must
thus select which unlabeled examples to use as negative training points,
possibly ending up with an unbalanced learning problem. We address these issues
by proposing an algorithm that combines active learning (for selecting negative
examples) with imbalance-aware learning (for mitigating the label imbalance).
In our experiments we observe that these two techniques operate
synergistically, outperforming state-of-the-art methods on standard protein
function prediction benchmarks. | ['Nicolò Cesa-Bianchi', 'Marco Frasca'] | 2018-05-18 | null | null | null | null | ['protein-function-prediction'] | ['medical'] | [ 7.34079778e-01 4.41512942e-01 -8.25423300e-01 -5.23745298e-01
-9.77209032e-01 -6.38344824e-01 1.70327917e-01 7.80457556e-01
-5.40590465e-01 1.21833444e+00 -2.53988028e-01 -3.65445048e-01
-2.02363580e-02 -5.39624453e-01 -6.65290594e-01 -9.69344020e-01
-1.29676670e-01 9.61068988e-01 2.76094079e-01 1.04442909e-01
4.27634746e-01 3.29997420e-01 -1.44799495e+00 4.63598102e-01
1.01787841e+00 6.68863595e-01 -3.01499903e-01 3.88302118e-01
-1.82514876e-01 1.07826948e+00 -6.02621675e-01 -3.30720872e-01
2.54418224e-01 -2.26003453e-01 -9.97017026e-01 2.58796781e-01
3.04958135e-01 1.23500533e-01 4.97470140e-01 7.96424627e-01
5.28948486e-01 6.92737028e-02 6.43158078e-01 -1.38857365e+00
-3.23694915e-01 3.45702469e-01 -7.78690934e-01 2.20472604e-01
1.01995938e-01 1.80495963e-01 1.39582491e+00 -1.08511591e+00
7.82847464e-01 8.40899169e-01 6.01234913e-01 5.14276326e-01
-1.71046853e+00 -4.28978473e-01 3.86727959e-01 2.66049683e-01
-9.68718648e-01 -7.28881776e-01 8.05167854e-01 -5.40686846e-01
1.13793528e+00 3.56884271e-01 5.69846928e-01 6.93540275e-01
-2.77972132e-01 1.06043875e+00 7.84972370e-01 -7.33516753e-01
5.91199636e-01 -4.36535291e-02 5.83124340e-01 2.72525936e-01
2.70852506e-01 1.05128177e-01 -7.07320929e-01 -8.04118335e-01
-7.95813091e-03 5.32812662e-02 -1.97532669e-01 -1.01155818e+00
-9.95848835e-01 8.70436966e-01 1.01751886e-01 -1.96695656e-01
-5.26455879e-01 -2.14629769e-01 5.45671761e-01 4.77912575e-01
8.90616059e-01 8.03153217e-01 -1.12388301e+00 1.33720905e-01
-7.53487647e-01 3.09467822e-01 7.21871614e-01 5.28214276e-01
9.50321794e-01 -4.42120016e-01 -8.06396678e-02 1.03399599e+00
2.45831802e-01 -7.32062533e-02 2.17965648e-01 -7.80693471e-01
3.46936762e-01 9.70444798e-01 3.70982081e-01 -2.66344517e-01
-4.09805030e-01 -2.32055724e-01 -3.12191933e-01 5.37064195e-01
7.06901431e-01 -1.68970674e-01 -9.83588398e-01 1.70066273e+00
5.74243963e-01 1.05229676e-01 1.17563635e-01 6.40722752e-01
3.55344802e-01 3.27700913e-01 5.33789396e-01 -8.97702634e-01
6.98735774e-01 -1.01017737e+00 -6.75734162e-01 -2.58806318e-01
1.11215365e+00 -7.08969533e-01 9.48067307e-01 6.86426044e-01
-9.89823163e-01 -1.58563942e-01 -1.06877768e+00 4.22243066e-02
-3.10974151e-01 -3.81922759e-02 8.92076612e-01 4.33197081e-01
-6.78093255e-01 6.01218343e-01 -7.70641804e-01 -1.34979367e-01
7.21970081e-01 8.01786482e-01 -4.55622077e-01 7.94748962e-02
-7.09087133e-01 8.44128430e-01 5.39247036e-01 -1.41253084e-01
-4.35243726e-01 -9.66012239e-01 -4.19085860e-01 -1.55994162e-01
6.25302196e-01 -2.70844728e-01 1.25558841e+00 -1.39212704e+00
-1.03926134e+00 1.09432721e+00 -2.42283836e-01 -4.74920392e-01
4.86722976e-01 -2.72451162e-01 -1.45123124e-01 -1.08912759e-01
-1.22538723e-01 5.89558721e-01 4.84111577e-01 -1.26802373e+00
-6.23744726e-01 -6.85895622e-01 8.17040280e-02 4.21845764e-01
-2.95010507e-01 6.71616057e-04 6.70176744e-02 -4.56971079e-01
2.42468983e-01 -8.49931896e-01 -5.92821538e-01 1.87448531e-01
-4.65535820e-01 -5.05533755e-01 1.05091715e+00 -2.35611908e-02
1.04164815e+00 -1.76068139e+00 1.30162776e-01 3.28661442e-01
5.46883821e-01 4.70450699e-01 -1.37618287e-02 2.17268988e-01
-5.17696023e-01 7.84710199e-02 -4.10188824e-01 -2.36755565e-01
-2.70782650e-01 1.68624461e-01 -2.67446816e-01 6.31568134e-01
5.14662862e-01 7.56165564e-01 -1.03541827e+00 -5.21769404e-01
9.27928686e-02 -3.69665250e-02 -6.20238245e-01 4.09582108e-01
-5.10740101e-01 4.52910662e-01 -2.96630740e-01 8.98373067e-01
6.05098546e-01 -6.92413747e-01 6.88771367e-01 2.14044884e-01
1.64670169e-01 4.40951526e-01 -9.09662008e-01 1.07305014e+00
1.06808305e-01 1.56568393e-01 -1.23954885e-01 -1.43297851e+00
8.81955624e-01 3.32128227e-01 8.31713438e-01 -3.88970315e-01
-1.45513147e-01 3.15485358e-01 1.40486494e-01 -4.19014871e-01
5.94140068e-02 -2.44703919e-01 3.63915294e-01 4.88735974e-01
1.89077601e-01 3.74135822e-01 3.64093393e-01 7.15821013e-02
1.27998614e+00 1.77541375e-01 7.44906664e-01 -3.47667247e-01
4.18262750e-01 2.51098156e-01 9.60589588e-01 5.92646897e-01
-6.66115582e-01 2.99643457e-01 8.92529368e-01 -4.92961764e-01
-1.00079966e+00 -7.41808355e-01 -1.03946656e-01 1.53484440e+00
-1.75495353e-02 -4.57069486e-01 -3.87740046e-01 -1.33370888e+00
4.37303334e-02 1.67668268e-01 -5.59483528e-01 -5.35315014e-02
-6.93685710e-01 -1.14959216e+00 2.66857669e-02 3.98673177e-01
-4.72052157e-01 -1.11322439e+00 -2.96020538e-01 2.23916322e-01
9.79351774e-02 -3.52952540e-01 1.06752150e-01 1.13222456e+00
-9.41466570e-01 -1.46398652e+00 -4.77430314e-01 -8.58221650e-01
8.52227032e-01 4.99250852e-02 1.64892268e+00 4.10066187e-01
-3.50292064e-02 -2.39795849e-01 -4.34639364e-01 -6.25516057e-01
-1.97568402e-01 2.90012032e-01 -1.32021084e-01 -3.75468373e-01
7.55442500e-01 -5.01248002e-01 -5.34670830e-01 2.27200955e-01
-7.63364494e-01 -1.81895956e-01 2.75028706e-01 1.31455553e+00
9.00268674e-01 -3.38740677e-01 1.23835397e+00 -1.69085562e+00
2.53362626e-01 -7.00655639e-01 -5.44785976e-01 5.24331987e-01
-8.11351657e-01 -1.00093730e-01 7.19168723e-01 -6.32905781e-01
-6.76876545e-01 6.46951973e-01 -2.41373628e-01 5.78751229e-02
-1.12028085e-01 2.66539901e-01 -2.67841458e-01 -2.69483209e-01
9.84993458e-01 -3.47508311e-01 -7.62343826e-03 -4.08852547e-01
1.13371372e-01 5.84120452e-01 4.47404347e-02 -3.99557531e-01
3.59262705e-01 2.45119616e-01 -2.30722323e-01 -3.97484034e-01
-1.16382837e+00 -7.87746549e-01 -8.54567707e-01 2.40944475e-02
1.69792980e-01 -8.71205986e-01 -6.10779345e-01 2.12071285e-01
-7.31158257e-01 -3.24766040e-01 -4.62467462e-01 3.78067404e-01
-7.75429606e-01 3.61441404e-01 -4.56671596e-01 -8.49194765e-01
-3.06859106e-01 -1.15234435e+00 9.13987637e-01 2.76041334e-03
-5.15882194e-01 -8.91882241e-01 5.01767516e-01 4.82769608e-01
-4.32384871e-02 1.64242033e-02 1.13647103e+00 -1.40468609e+00
-2.68985331e-01 -3.42823043e-02 1.97940215e-01 1.04951017e-01
2.67614145e-02 -4.30928245e-02 -1.13026237e+00 -3.69719535e-01
-2.37674966e-01 -9.41288650e-01 9.93707120e-01 2.14177743e-01
1.16471124e+00 -1.02550298e-01 -4.56534177e-01 2.67543048e-01
1.17080426e+00 1.57355964e-01 5.52750409e-01 1.03142589e-01
5.17445385e-01 7.70008028e-01 9.84175265e-01 3.66253823e-01
-1.10941462e-01 6.95166171e-01 4.14359003e-01 -5.85175157e-01
3.77944350e-01 8.20132568e-02 -1.12027243e-01 3.56118679e-01
2.73329198e-01 -3.87379766e-01 -1.01026797e+00 5.15758455e-01
-2.21804023e+00 -7.59369791e-01 -2.66746998e-01 2.40711594e+00
1.40038776e+00 2.92857587e-01 3.00048888e-01 6.66204572e-01
6.81273997e-01 2.58249380e-02 -1.06530952e+00 -9.55845788e-02
-2.24506825e-01 3.43804300e-01 4.72470492e-01 3.18634063e-01
-1.43876004e+00 7.02932060e-01 7.44253588e+00 8.39156270e-01
-1.05463326e+00 -7.65270814e-02 1.25368333e+00 -8.53411481e-02
-1.78840265e-01 1.81731880e-01 -6.78297520e-01 2.80055940e-01
8.70675683e-01 1.15440145e-01 -4.51854058e-02 9.02417421e-01
6.58531487e-02 -4.69569005e-02 -1.22091186e+00 6.66574001e-01
-2.49775872e-01 -1.34494829e+00 -3.23680013e-01 -2.46240243e-01
7.85491645e-01 1.11363426e-01 -2.37288535e-01 1.76420480e-01
5.53002179e-01 -9.08261299e-01 3.67373258e-01 1.39515147e-01
5.11796534e-01 -7.17850983e-01 5.93202770e-01 4.59070653e-01
-6.28345668e-01 -7.59145319e-02 -3.17185432e-01 -1.62487343e-01
-2.87090000e-02 9.36073422e-01 -9.45158124e-01 1.78676546e-01
3.33531350e-01 8.74390662e-01 -3.84787917e-01 1.14712489e+00
-9.67247561e-02 8.86429787e-01 -2.93671399e-01 2.50976086e-01
-2.91003078e-01 6.58642442e-04 2.23630160e-01 8.15892339e-01
-3.51580590e-01 7.62568191e-02 5.44007182e-01 5.51927447e-01
-3.11499178e-01 5.79478145e-01 -4.91041601e-01 -1.91727981e-01
5.08099794e-01 1.10499966e+00 -1.03636873e+00 -4.79308039e-01
-4.22247201e-01 6.36444211e-01 8.73906255e-01 2.91151404e-01
-3.26018453e-01 -6.70536887e-03 4.86278802e-01 7.93989748e-02
1.14597969e-01 2.81253219e-01 -5.82615912e-01 -1.19243944e+00
-2.85736714e-02 -8.45679402e-01 5.89743555e-01 -1.83422029e-01
-1.73962092e+00 1.02244169e-02 -5.49504697e-01 -1.19692087e+00
-1.46348268e-01 -6.52995110e-01 -3.22749704e-01 6.02624953e-01
-1.58804011e+00 -7.09808111e-01 2.65072614e-01 2.16231853e-01
3.69649827e-01 -2.68375594e-02 9.08978999e-01 3.66621047e-01
-5.61905563e-01 5.14304161e-01 2.26336613e-01 -2.23291814e-01
9.88563716e-01 -1.47598839e+00 1.84447438e-01 4.84144628e-01
9.96378213e-02 5.91945410e-01 5.76483011e-01 -6.88810170e-01
-1.19083643e+00 -8.97358775e-01 1.15469933e+00 -4.16912585e-01
4.54607725e-01 -4.39698070e-01 -1.17375302e+00 6.98881507e-01
-2.30773956e-01 5.03257990e-01 1.38978422e+00 4.33048636e-01
-4.67343837e-01 1.60961151e-01 -1.39755058e+00 1.62218347e-01
9.80254233e-01 -2.19940513e-01 -3.06432039e-01 8.44205976e-01
5.67809820e-01 -8.29640627e-02 -6.90206110e-01 7.43223131e-01
3.77325356e-01 -9.70493615e-01 1.00106692e+00 -1.19818032e+00
2.76058853e-01 -2.17743531e-01 3.82289514e-02 -1.04533529e+00
-2.76453555e-01 -4.48959947e-01 -4.34931785e-01 9.10638094e-01
9.25432742e-01 -5.09833574e-01 1.27446163e+00 5.72151423e-01
2.28870437e-02 -1.23502982e+00 -6.47086501e-01 -3.59607756e-01
1.90344706e-01 7.98707381e-02 2.69605577e-01 1.28672326e+00
3.90391469e-01 6.39540255e-01 -3.59096140e-01 -4.02138114e-01
4.20624942e-01 1.03611887e-01 5.25507569e-01 -1.58087909e+00
-4.28917676e-01 -1.43470034e-01 -2.67272323e-01 -8.01634848e-01
2.95686632e-01 -7.00413227e-01 1.83264345e-01 -1.04587007e+00
4.96147215e-01 -7.83462286e-01 -5.65127134e-01 9.52788889e-01
-7.03236222e-01 6.14163876e-01 -2.11208090e-01 4.19555157e-01
-1.09447515e+00 2.46537611e-01 6.35351241e-01 2.24308930e-02
-2.34423444e-01 1.57139733e-01 -8.32762420e-01 6.49508953e-01
6.78325891e-01 -5.54049551e-01 -5.16613245e-01 2.09300965e-01
4.14187402e-01 -1.10113546e-01 -2.08166927e-01 -4.46510434e-01
-1.78675484e-02 -5.32069862e-01 5.22895277e-01 -6.06631637e-01
1.03820190e-01 -8.06434751e-01 -1.18814163e-01 4.04692143e-01
-9.33931828e-01 -2.39642218e-01 -2.02291220e-01 5.66733003e-01
-2.20501404e-02 -2.08377317e-01 1.07578611e+00 -5.48890196e-02
-5.04502237e-01 3.44480038e-01 -2.39576831e-01 2.03274906e-01
1.11664772e+00 -1.42490000e-01 -5.83290637e-01 1.39479265e-01
-9.66383874e-01 2.26548120e-01 6.82394326e-01 2.33895946e-02
1.99526072e-01 -1.02583456e+00 -5.36302805e-01 2.66658276e-01
5.60774982e-01 -4.77040969e-02 -2.05874816e-01 9.17329073e-01
-2.29828119e-01 2.17366949e-01 8.81777108e-02 -5.49745619e-01
-1.57888734e+00 8.05904388e-01 1.43761009e-01 -6.05305374e-01
-1.42010808e-01 8.73892009e-01 1.90694839e-01 -8.37657452e-01
3.85982573e-01 3.30170780e-01 -3.78327519e-01 3.31199646e-01
3.96443814e-01 3.24748963e-01 7.13719130e-01 -3.51100028e-01
-3.55588377e-01 -1.11447729e-01 -3.80185276e-01 4.35913473e-01
1.49527884e+00 1.02816336e-01 -2.30989397e-01 5.76500177e-01
1.08162296e+00 2.11675633e-02 -1.10035372e+00 -4.76655930e-01
6.44915760e-01 -3.46418291e-01 -3.53970081e-01 -1.13968337e+00
-8.88696551e-01 4.58683372e-01 6.88957751e-01 2.59350300e-01
1.09251738e+00 -1.30651504e-01 3.98999482e-01 6.97217286e-01
2.09043950e-01 -1.16714990e+00 1.06095888e-01 4.63680595e-01
1.65771931e-01 -1.64037359e+00 1.71303973e-01 -7.25089610e-01
-5.05145073e-01 9.79483664e-01 6.32739604e-01 -9.84999016e-02
4.56167847e-01 5.60774565e-01 2.69191116e-01 -1.45187244e-01
-1.47399747e+00 -2.09445953e-01 1.13105454e-01 5.84415615e-01
1.02798653e+00 7.96174705e-02 -7.19397306e-01 1.63210899e-01
3.19741130e-01 6.03751093e-02 1.91588253e-01 1.41831613e+00
-6.19328499e-01 -1.65139771e+00 -1.53080732e-01 9.85811412e-01
-6.43113852e-01 -1.16096914e-01 -8.83103669e-01 3.13354254e-01
1.87021643e-01 9.38642204e-01 9.15093720e-03 -7.31567964e-02
-3.04600131e-02 3.51757705e-01 4.80202377e-01 -8.41216147e-01
-7.87565172e-01 1.18963756e-01 2.91115582e-01 -3.80391628e-01
-6.95852041e-01 -5.46753705e-01 -1.38375306e+00 -2.24139635e-03
-8.87857735e-01 4.16255295e-01 1.36322275e-01 9.38517034e-01
4.08051670e-01 1.70067713e-01 7.03170240e-01 -5.11590540e-01
-7.09908545e-01 -8.71023297e-01 -4.70355302e-01 5.64243257e-01
4.46918994e-01 -7.31076896e-01 -3.98254126e-01 1.22615151e-01] | [9.504339218139648, 4.015408992767334] |
2dbb871d-8143-4eed-b679-5b7ae321ea1a | eventbert-a-pre-trained-model-for-event | 2110.06533 | null | https://arxiv.org/abs/2110.06533v1 | https://arxiv.org/pdf/2110.06533v1.pdf | EventBERT: A Pre-Trained Model for Event Correlation Reasoning | Event correlation reasoning infers whether a natural language paragraph containing multiple events conforms to human common sense. For example, "Andrew was very drowsy, so he took a long nap, and now he is very alert" is sound and reasonable. In contrast, "Andrew was very drowsy, so he stayed up a long time, now he is very alert" does not comply with human common sense. Such reasoning capability is essential for many downstream tasks, such as script reasoning, abductive reasoning, narrative incoherence, story cloze test, etc. However, conducting event correlation reasoning is challenging due to a lack of large amounts of diverse event-based knowledge and difficulty in capturing correlation among multiple events. In this paper, we propose EventBERT, a pre-trained model to encapsulate eventuality knowledge from unlabeled text. Specifically, we collect a large volume of training examples by identifying natural language paragraphs that describe multiple correlated events and further extracting event spans in an unsupervised manner. We then propose three novel event- and correlation-based learning objectives to pre-train an event correlation model on our created training corpus. Empirical results show EventBERT outperforms strong baselines on four downstream tasks, and achieves SoTA results on most of them. Besides, it outperforms existing pre-trained models by a large margin, e.g., 6.5~23%, in zero-shot learning of these tasks. | ['Daxin Jiang', 'Guodong Long', 'Tao Shen', 'Xiubo Geng', 'Yucheng Zhou'] | 2021-10-13 | null | null | null | null | ['cloze-test'] | ['natural-language-processing'] | [ 1.39928460e-01 -9.22688842e-02 -2.57450968e-01 -2.88390428e-01
-8.36350739e-01 -6.43916845e-01 8.56407106e-01 5.24388611e-01
-3.60788316e-01 9.50641155e-01 7.32030571e-01 -3.56807381e-01
-2.62424886e-01 -9.02108371e-01 -6.10888541e-01 -2.25563258e-01
1.36673242e-01 5.70124209e-01 4.35140580e-01 -3.09383452e-01
1.16883844e-01 8.23071674e-02 -1.25908387e+00 5.83315730e-01
5.40850818e-01 4.85005945e-01 1.47991925e-02 7.04721928e-01
4.48815618e-03 1.75105762e+00 -6.76770151e-01 -6.73180580e-01
-3.94915164e-01 -8.00712228e-01 -1.26512527e+00 -1.95176508e-02
-1.43287048e-01 -4.57494885e-01 -5.05954027e-01 7.36434698e-01
2.90603608e-01 7.35415161e-01 7.71727204e-01 -1.24840903e+00
-4.43156272e-01 1.10291433e+00 -3.43977243e-01 7.14508474e-01
7.75081277e-01 2.28935525e-01 1.28789651e+00 -5.19833028e-01
8.44014227e-01 9.90623176e-01 6.12532794e-01 2.46419594e-01
-7.08838999e-01 -5.70574045e-01 2.17384622e-01 5.92169762e-01
-1.25261128e+00 -1.53434634e-01 7.41727769e-01 -3.38061363e-01
1.33716774e+00 3.14951718e-01 4.19265747e-01 1.60794652e+00
1.19836852e-01 9.51712012e-01 5.83480537e-01 -3.22597831e-01
2.98683971e-01 -1.62239611e-01 1.62658587e-01 4.49877352e-01
1.32574951e-02 -3.61620039e-01 -8.38043153e-01 -1.16739079e-01
2.61506170e-01 3.46595407e-01 -1.68461487e-01 4.87943441e-01
-1.49393559e+00 7.55607843e-01 -1.53792808e-02 3.55574220e-01
-3.98998648e-01 1.00671630e-02 6.30846202e-01 1.75555378e-01
1.78095311e-01 4.31502074e-01 -2.52679169e-01 -7.06559837e-01
-8.23814869e-01 5.37141860e-01 1.01892304e+00 1.05265117e+00
1.89984232e-01 -2.32231647e-01 -5.39033830e-01 6.68024719e-01
2.30327006e-02 2.77720600e-01 5.93631744e-01 -8.09380054e-01
8.68077815e-01 7.03291893e-01 2.44514063e-01 -9.86647844e-01
-5.35121202e-01 -1.03621855e-01 -7.42954612e-01 -3.81091595e-01
3.03705633e-01 -3.19145292e-01 -4.91937876e-01 1.70378911e+00
1.94477841e-01 3.96676481e-01 3.88436258e-01 7.43644416e-01
1.02357948e+00 9.20830727e-01 2.95630157e-01 -5.38267434e-01
1.66466057e+00 -8.20489943e-01 -9.66024339e-01 -4.03127640e-01
5.61967194e-01 -7.43008733e-01 1.37250543e+00 3.05538356e-01
-1.11741221e+00 -2.02769503e-01 -1.15506244e+00 -3.15323621e-02
-2.13761881e-01 -2.03286141e-01 6.42389834e-01 1.69285070e-02
-1.35349398e-02 5.93902588e-01 -9.13303375e-01 -5.61617076e-01
3.01769853e-01 -3.91012222e-01 -2.06836104e-01 -2.63531059e-01
-1.60807455e+00 9.08626974e-01 8.38779151e-01 -3.92334819e-01
-9.41756487e-01 -6.28156781e-01 -1.01277053e+00 2.75880903e-01
8.04672718e-01 -5.71924090e-01 1.62206531e+00 -8.72305259e-02
-9.68753934e-01 6.79695725e-01 -3.20992589e-01 -5.72310269e-01
7.93798804e-01 -4.22246546e-01 -8.50249290e-01 2.42978260e-01
5.35070956e-01 -2.29885112e-02 1.91717401e-01 -6.96554959e-01
-7.51822054e-01 -3.95467915e-02 2.99636722e-01 2.55031675e-01
-1.26609564e-01 4.71035749e-01 -4.72241879e-01 -6.64921105e-01
1.10835545e-02 -6.81658387e-01 1.47397788e-02 -5.43656766e-01
-6.47954106e-01 -3.24519545e-01 6.77481651e-01 -3.24532062e-01
1.56626332e+00 -2.00680089e+00 -3.24955553e-01 -2.33294159e-01
8.24075341e-02 -5.25060557e-02 2.88519412e-01 8.94810319e-01
-9.52756554e-02 7.85726681e-02 -2.97945261e-01 6.61998838e-02
9.33808684e-02 3.42813313e-01 -6.79188907e-01 3.59684490e-02
2.40217790e-01 6.79611921e-01 -1.37922120e+00 -6.89949512e-01
5.00112735e-02 5.57067245e-02 -2.80140907e-01 3.48319203e-01
-3.53286088e-01 1.76271304e-01 -3.90990257e-01 5.30493796e-01
1.02710865e-01 -5.60057759e-01 1.27563462e-01 4.46355492e-02
6.02301098e-02 5.35053849e-01 -1.19186068e+00 1.51699281e+00
-3.23613524e-01 7.99834073e-01 -1.03343272e+00 -7.90381074e-01
6.07867181e-01 7.75890231e-01 3.47072691e-01 -5.10984540e-01
1.92752689e-01 -2.13073745e-01 -1.17133938e-01 -9.06400561e-01
4.92073029e-01 -3.70078623e-01 -5.29535115e-01 8.62579942e-01
-3.64915915e-02 -1.95173696e-01 7.86209583e-01 6.82901978e-01
1.58152306e+00 -3.66781689e-02 7.80370593e-01 3.80531669e-01
4.41868246e-01 2.99458712e-01 6.74224019e-01 7.59898484e-01
-1.77559137e-01 7.32630551e-01 8.43209386e-01 -3.30936939e-01
-8.56100917e-01 -1.21128571e+00 2.33095109e-01 1.01701677e+00
3.02161902e-01 -8.93923342e-01 -3.08594227e-01 -6.65897787e-01
-4.05169129e-01 1.46051848e+00 -4.84350622e-01 -3.42541754e-01
-4.57986683e-01 -6.35866225e-01 7.98357606e-01 8.95936430e-01
7.78908014e-01 -1.16670418e+00 -7.04439342e-01 4.88463819e-01
-7.74936974e-01 -1.41248155e+00 -2.38201201e-01 2.30059117e-01
-3.58227462e-01 -1.29694366e+00 -1.02705918e-01 -3.55033994e-01
2.42670238e-01 2.86709309e-01 1.21895182e+00 -6.61971271e-02
-2.05464870e-01 -4.45699878e-03 -7.00845063e-01 -6.01364374e-01
-2.10334390e-01 -2.73495495e-01 -4.51156357e-03 -3.62518877e-01
7.93662727e-01 -5.70506692e-01 -2.00791597e-01 2.34627694e-01
-9.12443280e-01 8.19592550e-02 2.14869708e-01 7.92016566e-01
3.75079155e-01 6.98877275e-01 4.15969044e-01 -1.06441641e+00
6.77950978e-01 -9.10931468e-01 6.22537211e-02 2.77508020e-01
-1.79263994e-01 -1.42037913e-01 8.56996715e-01 -5.61651766e-01
-1.45566297e+00 -4.92068499e-01 1.08831814e-02 -1.89775869e-01
-1.60328656e-01 6.86695993e-01 -1.74431354e-01 1.02029812e+00
9.72941160e-01 1.82823554e-01 -7.32581615e-01 7.89070204e-02
1.95610821e-01 5.18509984e-01 9.73027945e-01 -6.83322370e-01
8.13406825e-01 6.43766403e-01 -3.47223341e-01 -3.98315966e-01
-1.61349738e+00 -6.89922810e-01 -2.91176528e-01 -2.32358098e-01
8.78863811e-01 -1.00973713e+00 -5.68773985e-01 1.40882790e-01
-1.34242105e+00 -3.75282168e-01 -1.46043435e-01 8.38176668e-01
-7.22877026e-01 1.77149668e-01 -7.05659986e-01 -7.61987627e-01
-2.37289801e-01 -6.47386789e-01 5.91932893e-01 2.89460957e-01
-9.89875972e-01 -8.86699021e-01 6.91145584e-02 5.50830126e-01
-3.06852877e-01 4.94982988e-01 9.13462698e-01 -1.06518304e+00
-2.58706808e-01 -2.95554459e-01 -1.28065452e-01 -2.02598646e-01
2.16786742e-01 4.62860689e-02 -6.19946241e-01 2.49137223e-01
-5.05673811e-02 -5.05667508e-01 5.42483032e-01 -4.19382006e-02
9.03538465e-01 -3.84548634e-01 -3.37286860e-01 1.48500293e-01
1.19490802e+00 4.67934430e-01 5.80890238e-01 4.53773469e-01
4.87160921e-01 4.01635110e-01 9.44278896e-01 9.69865739e-01
6.97429955e-01 2.36340001e-01 5.28735062e-03 3.61527413e-01
5.07178204e-03 -5.51507890e-01 1.75199047e-01 4.99038756e-01
-1.64274484e-01 -6.30207539e-01 -1.15383995e+00 8.16460550e-01
-2.17683840e+00 -1.65448213e+00 -2.85899609e-01 1.68857753e+00
1.06915259e+00 4.99161661e-01 -5.62416390e-02 2.97547817e-01
6.39860451e-01 2.46652722e-01 -4.54004139e-01 -3.04369301e-01
-1.07744440e-01 -1.07579291e-01 -3.95732522e-02 1.56686142e-01
-1.12735963e+00 9.63824868e-01 5.47600222e+00 6.42712891e-01
-5.43747723e-01 1.83572412e-01 2.63064861e-01 -3.56044173e-01
-3.26655149e-01 1.92195699e-01 -7.96767950e-01 6.71259105e-01
8.94882262e-01 -6.55950129e-01 6.52990937e-02 8.87975335e-01
3.27706784e-01 -2.64318496e-01 -1.41903913e+00 8.70572090e-01
3.19520950e-01 -1.50310874e+00 -7.82276243e-02 -5.27689099e-01
7.95254290e-01 -2.10399330e-01 -5.65478146e-01 6.57290339e-01
8.07956636e-01 -9.01837707e-01 7.78673351e-01 2.64733911e-01
4.00139093e-01 -8.48831654e-01 7.60181606e-01 7.00293422e-01
-1.15005147e+00 4.32752110e-02 -6.53278902e-02 -3.97969902e-01
5.19281209e-01 5.34321845e-01 -9.34517205e-01 5.00674844e-01
7.50945330e-01 6.02528095e-01 -9.44006816e-02 8.33762169e-01
-8.41455400e-01 8.07032466e-01 -1.46144882e-01 -2.67751485e-01
4.36093360e-01 4.00896162e-01 5.93331754e-01 1.34123182e+00
1.56742737e-01 8.59367847e-01 1.75791100e-01 8.08028817e-01
-2.21582457e-01 -1.75371706e-01 -6.54523671e-01 -1.37209833e-01
6.63020670e-01 1.00731361e+00 -8.15034926e-01 -7.68578529e-01
-4.78542030e-01 9.21249628e-01 2.96914756e-01 2.78684974e-01
-1.26243162e+00 -5.86125016e-01 3.11434567e-01 -8.67559463e-02
7.74049759e-02 -2.35893801e-02 -1.79180764e-02 -1.28001070e+00
1.83917284e-01 -5.91000140e-01 9.31859970e-01 -1.07687235e+00
-1.21150815e+00 4.62211996e-01 2.12868646e-01 -1.42195129e+00
-6.15937531e-01 5.87592125e-02 -1.26880300e+00 3.81518096e-01
-1.23077989e+00 -8.95933390e-01 -4.85645324e-01 6.71545088e-01
9.25326347e-01 -1.36093661e-01 6.34332359e-01 1.23564065e-01
-6.26213074e-01 1.96757913e-01 -3.96269709e-01 4.38006967e-01
7.85774350e-01 -1.21752763e+00 3.14623386e-01 1.15618205e+00
1.10465899e-01 6.29688740e-01 9.68200386e-01 -8.43828559e-01
-1.04285157e+00 -1.32751203e+00 1.21929216e+00 -5.54654956e-01
9.48262691e-01 6.66684136e-02 -1.09281731e+00 1.06105900e+00
9.68339145e-02 -2.38498390e-01 1.02178884e+00 2.20017284e-01
-5.72183967e-01 2.24813864e-01 -7.66755760e-01 9.28712964e-01
1.06673419e+00 -5.73469877e-01 -1.46687424e+00 8.48383605e-01
7.81770408e-01 -4.82198954e-01 -6.66041791e-01 1.55369371e-01
1.34978190e-01 -7.62959123e-01 8.45417142e-01 -1.07896805e+00
9.90529537e-01 -4.42677081e-01 -1.93790242e-01 -1.14216185e+00
-2.77548164e-01 -7.50699818e-01 -3.57824355e-01 1.45306313e+00
5.06186664e-01 -1.15048468e-01 5.78504324e-01 7.32517719e-01
-8.71344283e-02 -4.11532700e-01 -4.86899406e-01 -9.20133054e-01
-2.40459830e-01 -9.24769819e-01 5.81501245e-01 1.21547031e+00
6.84834957e-01 7.33072698e-01 -4.32556540e-01 2.55406231e-01
2.55202383e-01 2.45653197e-01 5.85381508e-01 -8.89905751e-01
-2.64108151e-01 -3.19303483e-01 -1.48869917e-01 -5.20356238e-01
1.84074849e-01 -7.26180673e-01 1.61326870e-01 -1.81026769e+00
5.71370721e-01 1.31330594e-01 -1.85257927e-01 6.01225197e-01
-4.99946952e-01 -1.89263076e-01 4.91031492e-03 1.61387518e-01
-1.09000850e+00 2.02656925e-01 9.04897511e-01 -2.80697197e-01
-2.89430052e-01 3.95983318e-03 -7.02118456e-01 1.04531860e+00
6.48802996e-01 -4.34134632e-01 -7.29476213e-01 -2.43171975e-01
5.67749560e-01 5.27521372e-01 3.04526508e-01 -1.01083934e+00
6.41352952e-01 -5.48641086e-01 2.96563655e-01 -7.47963369e-01
1.87055245e-01 -5.24573505e-01 1.47871897e-01 2.38787696e-01
-5.90156972e-01 -6.45517632e-02 1.20036686e-02 6.33818507e-01
-4.90366966e-01 -3.57181340e-01 4.09872979e-01 -3.10881197e-01
-1.03505242e+00 2.04952791e-01 -4.84653831e-01 6.27459168e-01
1.25777566e+00 8.31189901e-02 -5.21789253e-01 -4.61097360e-01
-5.79122126e-01 3.33337367e-01 -3.04851010e-02 4.01328057e-01
6.22848094e-01 -1.33101141e+00 -8.87007475e-01 -4.66833413e-01
5.43542325e-01 1.74874514e-01 5.05054116e-01 7.15940118e-01
-4.13131386e-01 4.21967544e-02 1.66148674e-02 -2.97682583e-01
-1.09887016e+00 6.12041950e-01 -3.03649157e-01 -4.05548275e-01
-1.04571474e+00 8.18154156e-01 -1.93870515e-01 -2.20839400e-02
1.94776222e-01 -3.04158777e-01 -2.79223502e-01 2.34132499e-01
9.43798244e-01 2.94667870e-01 -1.43784666e-02 -1.35530069e-01
-5.16779363e-01 1.66033491e-01 -2.25748613e-01 -1.13267496e-01
1.17055953e+00 -3.18015665e-02 1.84468985e-01 6.68560326e-01
9.81475830e-01 -3.01824347e-03 -9.77000117e-01 -4.43722308e-01
1.14151187e-01 -3.49343181e-01 -3.30711812e-01 -8.68819952e-01
-3.23348969e-01 7.56157875e-01 -4.35532779e-01 1.45927355e-01
1.01316094e+00 2.12617815e-01 9.15880382e-01 8.22173655e-01
1.84151217e-01 -1.24063289e+00 3.84005725e-01 7.89049327e-01
7.29018033e-01 -1.35016644e+00 2.16008574e-01 -2.43236497e-01
-1.29228222e+00 1.02417135e+00 7.06351638e-01 1.75921634e-01
2.62798786e-01 3.56529802e-01 -2.69293010e-01 -3.65668446e-01
-1.19739568e+00 -2.57280767e-01 2.41535343e-02 1.79339483e-01
4.76969630e-01 5.09423465e-02 -2.30161622e-01 9.28987622e-01
-4.70335335e-01 -5.24048060e-02 7.86308408e-01 1.00877023e+00
-4.55240220e-01 -3.88108462e-01 -2.15894699e-01 3.41292799e-01
-4.00712103e-01 -8.19457620e-02 -1.97485745e-01 8.97903621e-01
3.67942043e-02 1.20676255e+00 7.56490007e-02 -2.11888105e-01
4.54454154e-01 1.15307458e-01 7.91263357e-02 -7.53032982e-01
-3.57653558e-01 -2.30353206e-01 5.69031000e-01 -4.15104836e-01
-3.02897483e-01 -8.60754967e-01 -1.70903742e+00 -5.36394656e-01
8.49783644e-02 2.68159434e-02 -1.59719004e-03 1.46592402e+00
-2.30614945e-01 7.67491758e-01 4.16716814e-01 3.28422897e-02
-2.23708630e-01 -9.03024137e-01 -3.27350229e-01 6.82420254e-01
-9.90662798e-02 -6.45299971e-01 -2.39460170e-01 4.29015785e-01] | [11.100076675415039, 8.878588676452637] |
a4efd1be-5f4e-4238-a4f2-fd43fedbd321 | hat-hierarchical-aggregation-transformers-for | 2107.05946 | null | https://arxiv.org/abs/2107.05946v2 | https://arxiv.org/pdf/2107.05946v2.pdf | HAT: Hierarchical Aggregation Transformers for Person Re-identification | Recently, with the advance of deep Convolutional Neural Networks (CNNs), person Re-Identification (Re-ID) has witnessed great success in various applications. However, with limited receptive fields of CNNs, it is still challenging to extract discriminative representations in a global view for persons under non-overlapped cameras. Meanwhile, Transformers demonstrate strong abilities of modeling long-range dependencies for spatial and sequential data. In this work, we take advantages of both CNNs and Transformers, and propose a novel learning framework named Hierarchical Aggregation Transformer (HAT) for image-based person Re-ID with high performance. To achieve this goal, we first propose a Deeply Supervised Aggregation (DSA) to recurrently aggregate hierarchical features from CNN backbones. With multi-granularity supervisions, the DSA can enhance multi-scale features for person retrieval, which is very different from previous methods. Then, we introduce a Transformer-based Feature Calibration (TFC) to integrate low-level detail information as the global prior for high-level semantic information. The proposed TFC is inserted to each level of hierarchical features, resulting in great performance improvements. To our best knowledge, this work is the first to take advantages of both CNNs and Transformers for image-based person Re-ID. Comprehensive experiments on four large-scale Re-ID benchmarks demonstrate that our method shows better results than several state-of-the-art methods. The code is released at https://github.com/AI-Zhpp/HAT. | ['Huchuan Lu', 'Jinqing Qi', 'Pingping Zhang', 'Guowen Zhang'] | 2021-07-13 | null | null | null | null | ['person-retrieval'] | ['computer-vision'] | [-2.10889757e-01 -5.94884753e-01 -5.03647700e-03 -5.38969278e-01
-4.62772340e-01 -2.23360047e-01 6.16066754e-01 -7.22076232e-03
-4.32174295e-01 5.19054472e-01 4.73186821e-01 3.72620940e-01
-1.67003527e-01 -8.61141145e-01 -4.67177540e-01 -5.89130282e-01
2.80473202e-01 2.95341402e-01 7.52333626e-02 -1.76230967e-01
-1.80391997e-01 4.66366380e-01 -1.61741424e+00 1.78950578e-01
7.71883726e-01 9.28824723e-01 6.42635599e-02 1.33497804e-01
1.65951818e-01 5.28557003e-01 -3.86988133e-01 -7.25117803e-01
2.53111601e-01 -2.07945526e-01 -7.23671377e-01 8.78177360e-02
6.80562079e-01 -5.99705398e-01 -8.41244161e-01 1.08020961e+00
7.75741160e-01 2.54875511e-01 4.03593570e-01 -1.22372711e+00
-8.73360217e-01 1.99486002e-01 -5.14822245e-01 2.55593896e-01
3.90289843e-01 1.25874132e-01 8.62724960e-01 -9.49187338e-01
2.84598798e-01 1.35938537e+00 8.75436902e-01 7.01793075e-01
-9.07017708e-01 -9.19966578e-01 3.08682919e-01 4.82398927e-01
-1.62464428e+00 -3.56280506e-01 6.79935753e-01 -3.29746664e-01
8.85739386e-01 9.50290784e-02 8.35753977e-01 1.00909281e+00
-3.79539728e-01 1.01617241e+00 9.43429708e-01 -1.38511255e-01
-3.08741301e-01 -2.45212279e-02 3.28784108e-01 7.55874455e-01
3.32715452e-01 -3.58674414e-02 -5.04065454e-01 4.22478765e-02
9.71994817e-01 7.46787250e-01 -2.47260943e-01 -2.27532908e-03
-1.26109111e+00 5.01155376e-01 8.77287805e-01 4.26412344e-01
-3.63116533e-01 1.30076021e-01 3.45997036e-01 1.02015749e-01
3.64760548e-01 -2.02940609e-02 -1.02106057e-01 1.46180838e-01
-8.51345658e-01 3.93304080e-01 4.26054925e-01 8.88034880e-01
7.58376479e-01 -2.67677695e-01 -5.20128131e-01 1.04981744e+00
9.69269797e-02 6.13445401e-01 4.83178943e-01 -6.25557125e-01
5.27516186e-01 7.91466236e-01 2.88185570e-02 -1.10559416e+00
-3.90021771e-01 -7.80345082e-01 -1.26348579e+00 -3.39797467e-01
3.96451503e-01 1.95919678e-01 -9.26823437e-01 1.72697783e+00
1.50101855e-01 4.43040043e-01 -1.06400236e-01 1.12821388e+00
9.63607669e-01 5.20009637e-01 6.34085611e-02 1.52961254e-01
1.57942414e+00 -1.23985720e+00 -5.25458217e-01 2.02384032e-03
3.27025682e-01 -4.30571824e-01 8.31251740e-01 2.19965473e-01
-8.74007106e-01 -9.79922235e-01 -8.50694656e-01 -2.45413557e-01
-2.99328148e-01 3.97262335e-01 5.24252176e-01 5.88078678e-01
-1.02952659e+00 3.53202969e-01 -7.10473776e-01 -6.21126771e-01
6.11015320e-01 5.24601281e-01 -5.96617579e-01 -3.89639050e-01
-1.24030578e+00 3.63661885e-01 1.65051714e-01 3.68689686e-01
-7.80621946e-01 -5.80954015e-01 -7.09335148e-01 2.03187212e-01
2.45592833e-01 -9.93710160e-01 9.94478405e-01 -8.87405097e-01
-1.31666005e+00 8.00488591e-01 -4.47823703e-01 -3.12556475e-01
3.69132191e-01 -4.70161140e-01 -4.43587661e-01 1.90993845e-01
3.52877259e-01 6.16687953e-01 6.61713481e-01 -1.17654812e+00
-6.71591699e-01 -6.75061882e-01 1.50206253e-01 1.79765910e-01
-1.02434576e+00 1.71177864e-01 -9.81775522e-01 -8.27910662e-01
-1.37792215e-01 -9.09066617e-01 -1.43717945e-01 -1.61911115e-01
-5.41351140e-01 -5.38732767e-01 4.99520987e-01 -7.90837288e-01
1.14410818e+00 -2.07603335e+00 1.92278191e-01 1.29340678e-01
5.79798579e-01 4.95409310e-01 -1.77171290e-01 2.62853384e-01
1.34485485e-02 -9.88315791e-02 3.53136547e-02 -8.45099151e-01
9.40728933e-02 -7.20763728e-02 -4.78325374e-02 3.11600387e-01
-7.52746016e-02 1.24699140e+00 -7.37108052e-01 -4.73004550e-01
3.19656521e-01 8.08807373e-01 -4.58839178e-01 2.49966085e-01
3.98457855e-01 5.05628407e-01 -5.60442865e-01 8.37761760e-01
6.97827995e-01 -5.37804484e-01 -3.76948826e-02 -6.34365201e-01
-1.28342196e-01 -1.12380810e-01 -8.82616222e-01 1.71695101e+00
-1.70887232e-01 2.75229603e-01 -2.66687483e-01 -1.17485619e+00
9.29745317e-01 1.81224018e-01 5.41678905e-01 -9.39804375e-01
6.42368058e-03 1.44917190e-01 -4.50573295e-01 -1.65804461e-01
4.14454281e-01 2.48416126e-01 -7.82253891e-02 2.30723917e-01
1.21604063e-01 7.89443910e-01 6.55260310e-02 2.42997095e-01
8.47218215e-01 -1.06150009e-01 5.08326069e-02 -1.32677540e-01
9.11956787e-01 -4.15565491e-01 8.58276844e-01 7.63347507e-01
-3.78787160e-01 8.26717019e-01 2.69690901e-02 -7.49976635e-01
-1.02120733e+00 -9.71265614e-01 8.44295248e-02 9.06761944e-01
4.27985311e-01 -6.72320485e-01 -7.92795897e-01 -5.67007720e-01
9.31613892e-02 -3.29185203e-02 -4.53153819e-01 -7.77564049e-02
-6.56381369e-01 -8.22883606e-01 6.54368699e-01 8.23112965e-01
1.22404492e+00 -8.16718221e-01 3.28900702e-02 1.24173261e-01
-5.26048541e-01 -1.32532191e+00 -7.64760315e-01 -5.04376590e-01
-5.73873103e-01 -9.03701067e-01 -1.37346041e+00 -9.15567935e-01
6.90859139e-01 7.72178769e-01 9.08856988e-01 3.36397350e-01
-2.55853653e-01 6.03604078e-01 -4.71649379e-01 -1.18038490e-01
5.23312151e-01 3.00389647e-01 2.53299803e-01 3.01106215e-01
6.87793136e-01 -6.01510167e-01 -1.03366244e+00 4.53679204e-01
-6.51775658e-01 9.22910869e-02 5.89231968e-01 9.21262443e-01
5.64557135e-01 1.90308973e-01 3.85326296e-01 -4.12575006e-01
4.62720782e-01 -2.64960587e-01 -2.78760105e-01 4.25939411e-01
-4.33324873e-01 -2.83867985e-01 5.92316270e-01 -2.92978883e-01
-1.08420169e+00 -1.00039266e-01 -1.26955003e-01 -4.68277663e-01
-3.32441926e-01 3.12734812e-01 -4.13502365e-01 -1.66388357e-03
2.00792566e-01 3.79252255e-01 -2.09785283e-01 -7.60809183e-01
3.49752493e-02 7.02671468e-01 6.91063762e-01 -6.89531803e-01
8.43439460e-01 6.22296751e-01 -2.39803538e-01 -5.92757046e-01
-1.04796994e+00 -6.04930043e-01 -6.58685505e-01 -1.30846828e-01
9.47209775e-01 -1.35605478e+00 -9.11501288e-01 1.07523763e+00
-1.09248078e+00 -1.12134844e-01 1.20175434e-02 4.60387796e-01
-7.80932456e-02 5.76363981e-01 -7.56308079e-01 -4.93214637e-01
-5.78586280e-01 -1.08059454e+00 1.12265778e+00 6.38267994e-01
2.36131027e-01 -7.22514212e-01 -1.45042717e-01 7.39528060e-01
4.06348467e-01 -3.80171239e-02 2.75954902e-01 -5.13657451e-01
-7.55452275e-01 -2.24992588e-01 -6.75267458e-01 3.34940851e-01
2.67998546e-01 -6.38148963e-01 -8.93585205e-01 -6.94038153e-01
-3.90895724e-01 -2.57743359e-01 1.23562002e+00 1.95950925e-01
1.38145220e+00 -1.77333504e-01 -5.86942613e-01 9.23729718e-01
1.26645970e+00 -1.89433143e-01 7.08745062e-01 5.87981761e-01
1.17144358e+00 4.89953697e-01 3.84401768e-01 5.89353979e-01
9.18794334e-01 1.00767267e+00 1.03153676e-01 -2.84240186e-01
-2.03264356e-01 -3.59034330e-01 2.70224184e-01 6.49916828e-01
-5.66851616e-01 -2.48395160e-01 -8.87188554e-01 6.69767916e-01
-2.05708575e+00 -1.01853442e+00 3.46308462e-02 2.10220838e+00
4.53200758e-01 -2.73396283e-01 3.82712305e-01 -5.40395118e-02
1.01285970e+00 1.34243086e-01 -5.05505025e-01 5.05410731e-01
-1.47136688e-01 -1.65270902e-02 3.74316633e-01 2.24455148e-01
-1.39048922e+00 9.75681067e-01 4.72302723e+00 9.31270719e-01
-7.84751713e-01 2.61166185e-01 5.75954378e-01 1.11876372e-02
-1.02809705e-01 -2.06163362e-01 -1.24565947e+00 6.26184762e-01
4.18685406e-01 2.43815463e-02 3.46265972e-01 5.96731126e-01
-6.94445968e-02 4.41405952e-01 -9.51899886e-01 1.45789027e+00
3.54336321e-01 -1.08462989e+00 2.69931167e-01 1.02076538e-01
7.15342999e-01 -2.05145448e-01 1.37542531e-01 1.77741751e-01
1.67348310e-01 -1.10519505e+00 4.54744339e-01 7.08965123e-01
7.22713768e-01 -9.04104173e-01 1.05530632e+00 -1.64121017e-02
-1.68403232e+00 -2.82773733e-01 -4.64242280e-01 1.02577910e-01
1.52926385e-01 6.55261219e-01 -1.77686974e-01 8.83458138e-01
1.18180931e+00 1.30844796e+00 -7.36100018e-01 9.99626517e-01
-9.97661799e-02 3.87270808e-01 -1.70245126e-01 2.70978808e-01
-5.26587851e-02 -4.65281941e-02 3.24604154e-01 1.30515301e+00
4.06325191e-01 1.97465956e-01 3.44768405e-01 6.84312522e-01
-1.16633244e-01 3.87911648e-02 -2.14367285e-01 2.00790286e-01
4.34176385e-01 1.21100247e+00 -3.35831881e-01 -4.33636993e-01
-6.81068897e-01 1.35098362e+00 5.21088421e-01 5.29268563e-01
-6.88883841e-01 -3.13369095e-01 7.81932950e-01 1.84222862e-01
2.91832119e-01 -3.00689518e-01 1.70495033e-01 -1.59259737e+00
2.24108979e-01 -7.80289710e-01 6.92880929e-01 -5.42257786e-01
-1.81472480e+00 6.07590914e-01 -5.03399596e-02 -1.22223806e+00
1.08257495e-01 -5.47565937e-01 -3.79016638e-01 9.02769506e-01
-1.71702421e+00 -1.78487623e+00 -8.71767819e-01 1.12415326e+00
3.71675193e-01 -5.42355120e-01 6.56550884e-01 7.07657635e-01
-8.41068923e-01 1.06007636e+00 5.85854091e-02 6.89509511e-01
8.49454820e-01 -1.03472245e+00 3.83327454e-01 9.18135822e-01
-8.00650567e-02 8.88624787e-01 6.40403554e-02 -6.51133716e-01
-1.16600287e+00 -1.30603647e+00 8.98825645e-01 -2.33773232e-01
2.53330469e-01 -3.61615956e-01 -8.15194130e-01 7.65950799e-01
3.89624909e-02 1.14684470e-01 7.56393373e-01 2.24990353e-01
-6.57831848e-01 -6.45076156e-01 -8.80932570e-01 3.59626859e-01
1.40999031e+00 -6.95673466e-01 -4.00768161e-01 2.89546967e-01
6.31956339e-01 -9.08379406e-02 -9.98331904e-01 4.99449492e-01
7.63669372e-01 -1.06149507e+00 1.49266565e+00 -2.74480760e-01
1.91509217e-01 -4.35941517e-01 -1.33821085e-01 -9.93935704e-01
-8.06489944e-01 -3.25236917e-01 1.84277352e-02 1.56865776e+00
-2.40299523e-01 -8.67007971e-01 7.30662107e-01 5.24906516e-01
-2.17918642e-02 -4.56718683e-01 -7.41975665e-01 -8.20152342e-01
-6.50896654e-02 -7.99308531e-03 8.80434752e-01 8.32808375e-01
-4.07190025e-01 1.03650637e-01 -7.25115001e-01 3.44325185e-01
9.82108712e-01 1.57103866e-01 9.62150633e-01 -1.32179868e+00
-1.66523442e-01 -3.30225557e-01 -7.04204559e-01 -1.39258838e+00
1.64902896e-01 -9.45713639e-01 -3.25271398e-01 -1.57411730e+00
8.34333837e-01 -5.04256010e-01 -7.11242139e-01 6.29155993e-01
-4.63247597e-01 4.08871830e-01 4.43433285e-01 6.43080294e-01
-9.21407461e-01 8.21198165e-01 1.12705767e+00 -3.30458432e-01
9.45115276e-03 -8.21504816e-02 -7.83227682e-01 5.42946100e-01
1.00311124e+00 -1.64109766e-01 -2.49899849e-01 -7.21038878e-01
-2.92031337e-02 -3.93115222e-01 9.98002291e-01 -1.34009111e+00
6.49053752e-01 1.86026871e-01 7.59917736e-01 -7.65795827e-01
4.21010733e-01 -6.05881631e-01 1.31361693e-01 2.87508160e-01
-1.89505190e-01 7.44991452e-02 -8.83230194e-02 5.46537042e-01
-4.94697064e-01 1.42111257e-01 4.95703399e-01 -2.99963385e-01
-8.90762806e-01 8.93705606e-01 1.23240903e-01 -2.11205050e-01
7.00489640e-01 -8.88379067e-02 -3.08331221e-01 -3.58747840e-01
-5.15324295e-01 4.06762868e-01 4.95236069e-01 4.77193296e-01
7.73215413e-01 -1.62044263e+00 -7.24493802e-01 1.86237887e-01
2.24493891e-01 1.96129121e-02 6.88519597e-01 6.98666930e-01
-1.40955344e-01 5.44892430e-01 -3.31113368e-01 -5.70075750e-01
-1.24320185e+00 5.19766450e-01 3.46164882e-01 -4.75711107e-01
-7.25589216e-01 8.60883415e-01 5.71630120e-01 -4.30570513e-01
2.27623761e-01 2.57666081e-01 -3.80049527e-01 -1.63366333e-01
9.18259978e-01 3.57106090e-01 -2.11544037e-01 -8.43275487e-01
-4.01719183e-01 8.63817692e-01 -4.72330332e-01 8.58710259e-02
1.33919466e+00 -2.83500552e-01 -2.33836651e-01 -9.88022052e-03
1.19594514e+00 -2.28588179e-01 -1.20107818e+00 -6.44761741e-01
-3.31929356e-01 -6.29619598e-01 -1.24417894e-01 -5.38547158e-01
-1.28415394e+00 9.47963357e-01 7.95970261e-01 -2.39830941e-01
1.28475499e+00 -1.50216362e-02 1.35141647e+00 4.33908015e-01
4.57713991e-01 -1.06031072e+00 4.61668223e-01 3.40465426e-01
7.38377392e-01 -1.26262486e+00 -3.73493098e-02 -5.04304826e-01
-4.94584829e-01 9.66800988e-01 8.90516877e-01 -6.29987195e-02
6.15729213e-01 -1.49880037e-01 -2.21764416e-01 -1.43383950e-01
-1.71235994e-01 -5.47717631e-01 3.76077354e-01 6.80670679e-01
2.14296177e-01 1.59409165e-01 6.91706985e-02 1.01025403e+00
8.53744522e-03 2.69861251e-01 -2.00989187e-01 6.16624892e-01
-2.15849727e-01 -1.22621584e+00 -2.89084375e-01 3.55419666e-01
-3.15454185e-01 -2.02008709e-01 -1.75856546e-01 5.77641487e-01
3.70635867e-01 9.11044240e-01 3.38827632e-02 -7.09652066e-01
3.31956208e-01 -2.84598708e-01 3.91750991e-01 -3.72353494e-01
-5.08625507e-01 -1.15677767e-01 -1.96253434e-01 -6.31370425e-01
-7.41587222e-01 -6.72811627e-01 -8.94451141e-01 -5.23800194e-01
-8.99967179e-03 2.78066657e-03 1.15219764e-01 9.04290318e-01
5.35046637e-01 2.88881838e-01 4.77257133e-01 -7.82437205e-01
-2.30931357e-01 -8.12347710e-01 -4.74221736e-01 7.05534518e-01
2.88160920e-01 -7.84282207e-01 -6.95931464e-02 -3.37806414e-05] | [14.714652061462402, 0.9342683553695679] |
f974b709-f7cf-4197-aa1e-44a65a989fdb | flexible-and-scalable-state-tracking | 1811.12891 | null | http://arxiv.org/abs/1811.12891v1 | http://arxiv.org/pdf/1811.12891v1.pdf | Flexible and Scalable State Tracking Framework for Goal-Oriented Dialogue Systems | Goal-oriented dialogue systems typically rely on components specifically
developed for a single task or domain. This limits such systems in two
different ways: If there is an update in the task domain, the dialogue system
usually needs to be updated or completely re-trained. It is also harder to
extend such dialogue systems to different and multiple domains. The dialogue
state tracker in conventional dialogue systems is one such component - it is
usually designed to fit a well-defined application domain. For example, it is
common for a state variable to be a categorical distribution over a
manually-predefined set of entities (Henderson et al., 2013), resulting in an
inflexible and hard-to-extend dialogue system. In this paper, we propose a new
approach for dialogue state tracking that can generalize well over multiple
domains without incorporating any domain-specific knowledge. Under this
framework, discrete dialogue state variables are learned independently and the
information of a predefined set of possible values for dialogue state variables
is not required. Furthermore, it enables adding arbitrary dialogue context as
features and allows for multiple values to be associated with a single state
variable. These characteristics make it much easier to expand the dialogue
state space. We evaluate our framework using the widely used dialogue state
tracking challenge data set (DSTC2) and show that our framework yields
competitive results with other state-of-the-art results despite incorporating
little domain knowledge. We also show that this framework can benefit from
widely available external resources such as pre-trained word embeddings. | ['Dilek Hakkani-Tur', 'Tagyoung Chung', 'Shachi Paul', 'Jeremie Lecomte', 'Arindam Mandal', 'Rahul Goel'] | 2018-11-30 | null | null | null | null | ['goal-oriented-dialogue-systems'] | ['natural-language-processing'] | [-5.55890650e-02 3.04124862e-01 -3.04278851e-01 -4.89340305e-01
-3.94319266e-01 -1.01992369e+00 9.98858213e-01 1.81732073e-01
-5.62055588e-01 9.21375632e-01 2.35625833e-01 -3.75735015e-01
2.33835146e-01 -7.57822812e-01 -1.08883873e-01 -2.54524380e-01
8.22378173e-02 8.68224919e-01 6.74534082e-01 -8.78316641e-01
-2.66557466e-02 2.52020478e-01 -1.19296372e+00 -8.31135958e-02
8.19876909e-01 6.14560902e-01 2.25848943e-01 7.82226384e-01
-3.92107904e-01 4.74244088e-01 -9.81974185e-01 -2.83432662e-01
9.51546505e-02 -5.73840678e-01 -1.18227375e+00 2.11391345e-01
3.80009934e-02 -5.21128953e-01 -3.15104127e-01 8.41966271e-01
4.14351970e-01 4.94587600e-01 5.62017143e-01 -1.19277108e+00
-2.50359744e-01 5.75111926e-01 5.07689714e-02 -2.21723840e-01
5.07404327e-01 3.45807135e-01 1.05095780e+00 -3.17343235e-01
6.70119643e-01 1.36038375e+00 4.18908715e-01 9.18543875e-01
-1.36105299e+00 -2.05078319e-01 3.90326709e-01 -3.04608978e-02
-7.77617216e-01 -4.97772217e-01 6.90435588e-01 -2.56042778e-01
1.05402482e+00 2.45441586e-01 5.51212847e-01 1.20747018e+00
-1.66022077e-01 6.63757086e-01 1.04452479e+00 -5.46748638e-01
3.79613161e-01 4.40731317e-01 3.30930382e-01 4.62960362e-01
-1.98631674e-01 -3.00752014e-01 -2.23879293e-01 -3.29181820e-01
6.06373310e-01 -2.73678601e-01 -2.37168118e-01 -6.61381602e-01
-9.34601426e-01 1.06537092e+00 -5.63469306e-02 5.01298308e-01
-4.98172902e-02 -3.10702473e-01 8.17757487e-01 5.75989664e-01
2.07133070e-01 7.38624454e-01 -7.73589551e-01 -7.39082634e-01
-5.15339553e-01 3.55904639e-01 1.54086089e+00 7.48534262e-01
8.75817239e-01 -2.31400043e-01 -3.37257564e-01 1.18522584e+00
8.38112235e-02 7.41757825e-02 5.57281554e-01 -8.87135327e-01
4.01942432e-01 7.62518942e-01 3.93414021e-01 -5.67385674e-01
-5.02055466e-01 1.18975371e-01 -4.62868959e-01 1.29921734e-01
8.89370322e-01 -6.45431638e-01 -6.84743047e-01 2.11953497e+00
5.45944095e-01 -1.93487853e-01 2.82689303e-01 7.47103930e-01
7.20694363e-01 5.87310791e-01 -1.16608283e-02 -1.17822014e-01
1.39440918e+00 -9.78332698e-01 -8.49259436e-01 -4.97656316e-01
7.50098765e-01 -3.82487565e-01 1.29085994e+00 2.45935977e-01
-7.83374608e-01 -2.40819395e-01 -8.89324427e-01 -2.58394871e-02
-5.08783162e-01 -2.37598926e-01 6.07214272e-01 7.11460114e-01
-9.25028801e-01 4.95873302e-01 -6.93880439e-01 -6.97841346e-01
-3.22302222e-01 3.08546722e-01 -3.79899710e-01 1.65935919e-01
-1.73565590e+00 1.45913064e+00 4.15609747e-01 -2.19525099e-01
-6.60574019e-01 -2.93955743e-01 -1.05090714e+00 1.65165856e-01
6.75473511e-01 -3.74545813e-01 1.66417706e+00 -6.73341751e-01
-2.24993086e+00 4.72079068e-01 -1.60419568e-02 -4.01047230e-01
5.18479228e-01 -1.25591755e-01 -3.46963108e-01 -7.70215243e-02
-2.25445971e-01 3.65078062e-01 5.75055182e-01 -9.29301322e-01
-4.51698303e-01 -1.74233258e-01 7.93402910e-01 4.85409647e-01
-6.45344019e-01 8.63708481e-02 -6.64678693e-01 -1.93316177e-01
-4.77333069e-01 -1.15801728e+00 -3.13745230e-01 -1.34026438e-01
-3.29771876e-01 -4.02546227e-01 8.86139333e-01 -4.83061105e-01
1.41507769e+00 -1.82853699e+00 3.11016381e-01 -1.69420779e-01
2.73738038e-02 5.92825949e-01 -1.40827581e-01 6.84205174e-01
2.40623042e-01 1.34736598e-02 -2.92178690e-01 -3.42416495e-01
3.59994441e-01 5.16740024e-01 8.96400064e-02 1.89050719e-01
1.90681174e-01 5.34013510e-01 -1.09966433e+00 -2.92164803e-01
3.92137319e-01 2.29035795e-01 -5.63439131e-01 4.49441075e-01
-5.32749832e-01 4.75027472e-01 -3.48862290e-01 1.08495299e-02
4.19909120e-01 -1.57947898e-01 6.01682723e-01 1.24737166e-01
-1.60490140e-01 7.97730029e-01 -1.30530310e+00 1.81780279e+00
-7.06790030e-01 4.20608789e-01 2.20996559e-01 -1.01250434e+00
9.28485930e-01 5.51254094e-01 2.61078745e-01 -1.99974149e-01
9.21808407e-02 -1.58576414e-01 1.72701642e-01 -3.52342367e-01
7.20431626e-01 -1.86716333e-01 -5.95907509e-01 7.31783986e-01
4.09017891e-01 -3.13868821e-01 4.24719125e-01 2.23710984e-01
1.18467951e+00 -1.23907596e-01 5.70699632e-01 -7.84357265e-03
6.69939160e-01 1.22123305e-02 7.19484568e-01 6.43583715e-01
-4.14866745e-01 2.02899024e-01 8.35953236e-01 -5.10071702e-02
-7.56651700e-01 -7.21074164e-01 -1.98559776e-01 1.48912394e+00
2.42271777e-02 -5.67971885e-01 -6.69667482e-01 -1.11848164e+00
-6.70686513e-02 7.57269561e-01 -4.86025989e-01 -1.83173701e-01
-5.04552841e-01 -3.25047404e-01 5.56628942e-01 2.19907686e-01
5.95070064e-01 -8.97173464e-01 -4.68021542e-01 4.68238533e-01
-6.28408566e-02 -1.02783632e+00 -4.61706787e-01 3.57590377e-01
-5.40413022e-01 -9.87132549e-01 -6.32316172e-01 -5.57926416e-01
2.84032226e-01 -1.32214397e-01 1.06606913e+00 -1.77600592e-01
1.31021589e-01 4.64936584e-01 -4.00384128e-01 -7.07917735e-02
-9.88375127e-01 4.25518543e-01 9.75146815e-02 -2.21741229e-01
3.93563092e-01 -9.36857760e-02 -3.21115971e-01 4.24180537e-01
-7.65864372e-01 -2.21597090e-01 1.37936659e-02 1.20489824e+00
-2.79969156e-01 -1.55075893e-01 8.47613215e-01 -1.13632429e+00
1.09742367e+00 -5.15141964e-01 -4.79671568e-01 3.07348639e-01
-5.43256819e-01 2.95721382e-01 7.96515644e-01 -7.42501676e-01
-1.21031535e+00 3.28252427e-02 -1.49759963e-01 3.53516340e-02
-4.51070011e-01 6.39777303e-01 -3.27352464e-01 2.44247869e-01
6.82862103e-01 8.42953697e-02 3.22348505e-01 -5.24908960e-01
6.71165884e-01 8.86894763e-01 2.79544443e-01 -8.09100747e-01
6.32324338e-01 -1.15267031e-01 -4.76893097e-01 -9.18134749e-01
-4.22284991e-01 -5.61186790e-01 -7.47065902e-01 -1.07072778e-01
5.96461296e-01 -6.65566385e-01 -5.94632745e-01 4.35464233e-01
-1.12167168e+00 -6.85922682e-01 -1.67919844e-01 2.21323520e-01
-4.05282587e-01 7.56954849e-01 -6.31375968e-01 -8.65650415e-01
-1.77235827e-01 -1.17664826e+00 6.20706499e-01 4.09278423e-01
-7.21234024e-01 -1.42136014e+00 3.08016360e-01 6.15669750e-02
6.34864748e-01 -1.07102431e-01 8.14498007e-01 -1.27601087e+00
7.51253515e-02 -2.09174186e-01 1.66662067e-01 4.95012045e-01
5.90254307e-01 -5.08957729e-02 -8.93136799e-01 -4.15410191e-01
2.03986820e-02 -6.03261232e-01 4.90628064e-01 -2.31795043e-01
5.65162122e-01 -5.02164483e-01 -1.35772914e-01 -1.53384373e-01
8.01199198e-01 1.99797034e-01 2.80459106e-01 4.07062471e-01
3.36236238e-01 7.94268668e-01 7.42688417e-01 6.07828379e-01
7.51983643e-01 1.03862655e+00 5.98094501e-02 3.64084020e-02
1.73872218e-01 -5.34180440e-02 5.17923594e-01 6.82817757e-01
3.94399405e-01 -2.34572634e-01 -8.79276752e-01 7.06802547e-01
-1.98468530e+00 -8.71287763e-01 2.70026028e-01 2.30358267e+00
1.60869873e+00 9.21332557e-03 4.20939595e-01 -1.23392969e-01
7.35546470e-01 3.37691426e-01 -6.56917572e-01 -6.89747036e-01
2.55103111e-01 2.13149875e-01 1.54357493e-01 7.06176281e-01
-1.13084900e+00 1.20697188e+00 5.86020660e+00 4.98633116e-01
-1.24415910e+00 7.34440982e-02 5.85326776e-02 5.84678724e-02
-9.58177671e-02 1.71583831e-01 -8.96461785e-01 4.45973366e-01
1.00490820e+00 -3.17282706e-01 3.54649514e-01 7.07244873e-01
-4.12471630e-02 -2.44621843e-01 -1.22733939e+00 4.82987374e-01
-2.31468707e-01 -8.62774253e-01 -4.05999810e-01 -3.89047377e-02
3.31418395e-01 -1.28422618e-01 -2.11946368e-01 8.62344682e-01
8.80669177e-01 -7.62743294e-01 2.17903495e-01 4.29104567e-02
6.71286345e-01 -5.32020926e-01 5.68009138e-01 5.61384499e-01
-8.73681843e-01 2.56910524e-03 -2.58967996e-01 -1.44068480e-01
2.41252214e-01 1.20758742e-01 -1.30447674e+00 3.80305678e-01
1.99210018e-01 3.96761477e-01 -4.01976496e-01 8.42424393e-01
-3.42540026e-01 5.52044511e-01 -5.05483210e-01 -3.04264307e-01
2.73515165e-01 -9.77884009e-02 6.12404704e-01 1.27149987e+00
-8.37465450e-02 -9.64302942e-02 5.47384918e-01 5.71813762e-01
9.19982046e-02 6.97524771e-02 -6.73783004e-01 -3.19667198e-02
6.60451293e-01 1.38160586e+00 -3.03111941e-01 -3.19600612e-01
-6.22714758e-01 1.03448606e+00 3.56112957e-01 2.98918486e-01
-6.49382532e-01 -6.31835580e-01 9.76297319e-01 -2.86003023e-01
3.74865532e-01 -3.24147314e-01 5.82386330e-02 -1.44840622e+00
-2.58971035e-01 -1.02070951e+00 4.16996300e-01 -1.19502954e-01
-1.30150270e+00 5.72662771e-01 6.70053139e-02 -9.53992724e-01
-7.89867640e-01 -4.58358169e-01 -6.24984562e-01 1.11003196e+00
-1.39360476e+00 -9.09879148e-01 2.15712637e-02 6.43091917e-01
5.70547104e-01 -2.01583549e-01 1.29292607e+00 2.85979714e-02
-5.99817276e-01 7.29477167e-01 1.46849044e-02 3.36224377e-01
1.22937572e+00 -1.55276000e+00 4.16971803e-01 5.23414373e-01
-1.53993815e-01 8.19003284e-01 9.47437108e-01 -5.25289416e-01
-1.28406429e+00 -8.50783706e-01 8.38920653e-01 -5.88419676e-01
8.42237115e-01 -6.95176482e-01 -1.23921549e+00 5.55152953e-01
4.45159197e-01 -2.17209831e-01 6.63804054e-01 6.28488719e-01
-2.40398943e-01 1.32771686e-01 -1.28826702e+00 7.69560337e-01
5.47971070e-01 -6.44233286e-01 -8.63572776e-01 2.33804569e-01
5.67315042e-01 -6.11782670e-01 -1.04561889e+00 1.24189138e-01
2.94246256e-01 -7.41673887e-01 6.10519230e-01 -8.03014398e-01
3.33038457e-02 -1.60702810e-01 7.17045143e-02 -1.71113157e+00
-3.18566337e-02 -9.60287392e-01 -3.89280438e-01 1.46396482e+00
5.71290314e-01 -8.26749146e-01 5.72759211e-01 1.31019008e+00
9.65387821e-02 -4.70929265e-01 -8.93375039e-01 -8.64095032e-01
3.64264280e-01 -9.04563740e-02 6.66803241e-01 1.14650548e+00
7.22864628e-01 8.94716620e-01 -2.98866749e-01 8.13108906e-02
7.37557095e-03 1.74775235e-02 1.14746463e+00 -1.32684505e+00
-5.09187639e-01 -3.71510029e-01 -5.74216470e-02 -1.43729508e+00
4.09094334e-01 -5.90565979e-01 2.52798647e-01 -1.45570505e+00
-2.01666117e-01 -7.08196998e-01 -6.75100833e-02 7.18154490e-01
-3.64332914e-01 -5.54560125e-01 3.14517885e-01 -1.43158674e-01
-6.69780076e-01 7.51716554e-01 1.09586334e+00 -4.98789139e-02
-5.27135313e-01 1.89908609e-01 -6.16389453e-01 4.50313717e-01
9.20416474e-01 -1.83027044e-01 -6.02898717e-01 -3.33748423e-02
-2.20291749e-01 3.30229998e-01 -3.42684612e-02 -6.00304067e-01
2.84665614e-01 -3.01259547e-01 -1.58264294e-01 -9.77047086e-02
6.03156447e-01 -6.49685562e-01 -3.25092256e-01 1.28961086e-01
-4.61930186e-01 -2.75514662e-01 3.38219494e-01 4.91755307e-01
-1.36831507e-01 -5.74795544e-01 8.28643084e-01 -1.94957033e-01
-7.11568177e-01 -5.47409318e-02 -6.49929285e-01 2.76551515e-01
1.06109858e+00 -9.83889699e-02 -3.99533957e-01 -6.89151704e-01
-6.55593693e-01 4.89234716e-01 6.92262173e-01 6.21209145e-01
3.65658253e-02 -1.01314235e+00 -3.09026062e-01 -3.55071276e-02
2.41417542e-01 1.87805183e-02 1.12100998e-02 4.46442455e-01
5.82974367e-02 4.29532975e-01 -2.00530604e-01 -5.59599876e-01
-1.25378621e+00 2.98179150e-01 3.29692751e-01 -5.78473628e-01
-4.90474045e-01 5.06450832e-01 4.61745150e-02 -1.03650999e+00
2.31105551e-01 -4.12206590e-01 -4.58140552e-01 3.31168622e-01
3.49168658e-01 -4.19537611e-02 -6.39430732e-02 -4.98431295e-01
-1.28683522e-01 3.18183494e-03 -3.92159015e-01 -3.83246362e-01
1.25734639e+00 -2.87731618e-01 1.49596691e-01 7.00005949e-01
9.18033838e-01 -7.20242038e-02 -1.47622657e+00 -6.40155435e-01
2.56354287e-02 -2.01933131e-01 -9.51895416e-02 -1.10259402e+00
-6.41342342e-01 7.44274914e-01 2.78321773e-01 6.81882322e-01
6.81361020e-01 -6.09532297e-02 6.72755957e-01 5.69691718e-01
4.67730612e-01 -1.32246518e+00 2.67133135e-02 1.19409430e+00
7.24449813e-01 -1.34500015e+00 -2.18045041e-01 -2.52750292e-02
-1.19590664e+00 1.14191294e+00 8.33607435e-01 2.75063068e-01
4.15088147e-01 1.42974243e-01 2.35698134e-01 1.89352836e-02
-9.12300527e-01 -2.97795713e-01 9.19758342e-03 6.60474837e-01
6.09872699e-01 -4.87596653e-02 -3.16094905e-01 7.10800588e-01
-5.39380126e-02 -1.28032163e-01 6.97662532e-01 1.05954695e+00
-4.00820225e-01 -1.77546632e+00 -7.57682696e-02 3.36233526e-01
-2.49883607e-01 7.71510303e-02 -5.77245355e-01 8.37424695e-01
-5.34206033e-01 1.16214621e+00 -1.22107238e-01 -1.79429099e-01
4.77415353e-01 5.69709420e-01 3.71285945e-01 -1.15025640e+00
-9.33761597e-01 -1.07698783e-01 6.67383909e-01 -2.39340797e-01
-1.22278161e-01 -6.68931961e-01 -1.26248920e+00 -2.28146672e-01
-5.09150803e-01 2.88933456e-01 4.73319709e-01 8.88226092e-01
4.25566226e-01 2.96027184e-01 5.42046130e-01 -5.45474350e-01
-8.80579650e-01 -1.24627221e+00 -3.90465975e-01 3.90272975e-01
3.99433076e-01 -8.59510899e-01 -1.83976904e-01 -2.10756779e-01] | [12.881855010986328, 7.875612735748291] |
e637d8a0-c4da-49c4-8867-829037fb308b | dcdetector-dual-attention-contrastive | 2306.10347 | null | https://arxiv.org/abs/2306.10347v1 | https://arxiv.org/pdf/2306.10347v1.pdf | DCdetector: Dual Attention Contrastive Representation Learning for Time Series Anomaly Detection | Time series anomaly detection is critical for a wide range of applications. It aims to identify deviant samples from the normal sample distribution in time series. The most fundamental challenge for this task is to learn a representation map that enables effective discrimination of anomalies. Reconstruction-based methods still dominate, but the representation learning with anomalies might hurt the performance with its large abnormal loss. On the other hand, contrastive learning aims to find a representation that can clearly distinguish any instance from the others, which can bring a more natural and promising representation for time series anomaly detection. In this paper, we propose DCdetector, a multi-scale dual attention contrastive representation learning model. DCdetector utilizes a novel dual attention asymmetric design to create the permutated environment and pure contrastive loss to guide the learning process, thus learning a permutation invariant representation with superior discrimination abilities. Extensive experiments show that DCdetector achieves state-of-the-art results on multiple time series anomaly detection benchmark datasets. Code is publicly available at https://github.com/DAMO-DI-ML/KDD2023-DCdetector. | ['Liang Sun', 'Qingsong Wen', 'Tian Zhou', 'Chaoli Zhang', 'Yiyuan Yang'] | 2023-06-17 | null | null | null | null | ['contrastive-learning', 'contrastive-learning', 'anomaly-detection', 'time-series-anomaly-detection'] | ['computer-vision', 'methodology', 'methodology', 'time-series'] | [ 8.45155269e-02 -7.39427567e-01 1.82928324e-01 -2.56309420e-01
-7.89424241e-01 -4.62176800e-01 5.20602584e-01 3.35468739e-01
-7.46761113e-02 2.26090342e-01 6.58573806e-02 -3.52236778e-01
-2.74091065e-01 -7.31642663e-01 -5.56922317e-01 -9.51840758e-01
-4.81602758e-01 3.06345284e-01 -8.98137987e-02 -3.34549189e-01
3.78127933e-01 5.50945163e-01 -1.53005886e+00 1.82078212e-01
1.11866879e+00 1.34076273e+00 -1.96369499e-01 3.59535664e-01
-9.42164212e-02 6.24782503e-01 -6.12867713e-01 -1.87847316e-02
4.96801555e-01 -4.93654549e-01 -3.32051247e-01 -5.16328096e-01
4.61454093e-01 -1.49498180e-01 -4.76402462e-01 1.27113545e+00
5.09848654e-01 3.34714204e-01 8.32893491e-01 -1.76928520e+00
-8.09754372e-01 3.68655384e-01 -9.98323560e-01 1.15684438e+00
1.01566680e-01 2.76722044e-01 1.11431634e+00 -6.37521982e-01
-1.30141452e-01 1.14044368e+00 3.40667158e-01 3.69549781e-01
-1.21764195e+00 -1.01791430e+00 5.12300432e-01 6.90731585e-01
-1.25813365e+00 -1.80357486e-01 1.11446762e+00 -4.29539174e-01
7.84273505e-01 6.15066469e-01 4.66870576e-01 1.25151598e+00
2.13022292e-01 1.00056696e+00 7.55954146e-01 -1.39504420e-02
1.57808065e-01 -6.14424407e-01 3.46341938e-01 5.28502941e-01
3.29911351e-01 1.83622651e-02 -2.32874870e-01 -3.61403555e-01
4.68950242e-01 7.58814871e-01 -4.04688597e-01 -2.31465772e-01
-1.05623007e+00 7.12162316e-01 5.79185605e-01 4.02917027e-01
-5.75881898e-01 1.18822465e-02 9.15177524e-01 7.56194890e-01
6.52330756e-01 5.62343359e-01 -4.52857405e-01 -2.38363877e-01
-4.03039038e-01 2.41757289e-01 1.99062005e-01 1.54400721e-01
2.89137781e-01 5.60152769e-01 -5.37022233e-01 8.48908484e-01
1.75002888e-01 6.00485742e-01 9.41968918e-01 -2.70045847e-01
5.72178662e-01 6.48031414e-01 -4.25291330e-01 -1.18663859e+00
-1.84032530e-01 -7.22772598e-01 -1.06576240e+00 2.67307967e-01
3.62413824e-01 1.43424748e-02 -9.39688623e-01 1.80871320e+00
1.03455288e-02 7.51432419e-01 -1.49939120e-01 8.65267575e-01
4.26743209e-01 9.74830031e-01 -2.97839530e-02 -1.25755183e-02
1.09296453e+00 -7.50341713e-01 -5.53638816e-01 -3.12941819e-01
4.90358174e-01 -4.04117346e-01 1.19274879e+00 4.03928638e-01
-6.04720712e-01 -2.92465687e-01 -9.97022271e-01 3.13078910e-01
-3.59297603e-01 -2.22351581e-01 5.66140115e-01 1.51988834e-01
-4.21747476e-01 5.24702907e-01 -1.05735576e+00 1.77566409e-02
6.99515820e-01 7.45342597e-02 -3.09311450e-01 -1.06584907e-01
-1.13010025e+00 4.86571014e-01 3.34770322e-01 7.69892186e-02
-8.46677482e-01 -7.47373104e-01 -9.52819169e-01 2.59265989e-01
2.24224478e-01 -1.46437883e-01 1.04233181e+00 -9.85444725e-01
-9.26123738e-01 6.84640348e-01 -5.87378442e-03 -6.91105604e-01
4.50976461e-01 -4.95000988e-01 -8.85683000e-01 -9.16082412e-02
2.44007781e-01 -4.26811501e-02 1.10481751e+00 -6.32631660e-01
-5.61266840e-01 -7.07761943e-01 -5.12123525e-01 -2.52093598e-02
-4.10050333e-01 -7.14257956e-02 -1.26826447e-02 -1.18295372e+00
2.41168171e-01 -6.54640019e-01 -1.72960132e-01 -8.99083838e-02
-3.55445623e-01 -4.54101831e-01 1.20227468e+00 -7.46065497e-01
1.40050340e+00 -2.40293837e+00 -8.57477076e-03 4.43270862e-01
2.39755526e-01 4.14415121e-01 -4.13273245e-01 3.49275410e-01
-6.94329381e-01 -2.01073632e-01 -6.27990782e-01 6.51185140e-02
-2.80831102e-02 -2.64797695e-02 -9.42193806e-01 7.28623390e-01
4.98087049e-01 7.27897882e-01 -8.83254528e-01 1.95516288e-01
1.00423522e-01 2.15878680e-01 -6.10551476e-01 4.05638635e-01
-9.59874317e-02 6.11068666e-01 -6.22360051e-01 6.53304338e-01
7.44893789e-01 -8.40538591e-02 -4.17601734e-01 2.36805186e-01
1.18443295e-01 1.72838449e-01 -9.76287425e-01 1.43951619e+00
-2.04078779e-01 5.77425539e-01 -5.06986618e-01 -1.55165529e+00
1.12454712e+00 -3.17159221e-02 6.20211780e-01 -1.16044307e+00
4.84607322e-03 4.64629620e-01 3.46573472e-01 -4.87996876e-01
9.53365117e-02 2.31814057e-01 -1.76175982e-01 5.49256921e-01
-3.48630160e-01 3.50110769e-01 2.27764864e-02 -1.80977583e-02
1.27960777e+00 -1.39773339e-01 2.79018581e-01 -2.69263536e-01
6.38234437e-01 -4.60906446e-01 9.63163257e-01 5.32484651e-01
-2.86226958e-01 5.47824919e-01 7.91875660e-01 -6.86372757e-01
-5.85392952e-01 -1.21356606e+00 -1.59216151e-01 1.09446561e+00
-1.56862482e-01 -1.23512052e-01 -2.43173659e-01 -9.01445150e-01
2.76330441e-01 8.93573761e-01 -6.75723612e-01 -7.47407198e-01
-6.76740646e-01 -9.42429185e-01 4.61792469e-01 5.81890285e-01
4.29970145e-01 -1.16526318e+00 -4.94823068e-01 5.20645194e-02
-1.25002056e-01 -4.38092470e-01 -6.10011160e-01 1.54906124e-01
-7.89670467e-01 -1.18804038e+00 -6.13288462e-01 -5.75216770e-01
7.29918718e-01 2.55562842e-01 9.23281074e-01 -1.65999651e-01
-4.81194317e-01 2.56151348e-01 -4.94879425e-01 -7.04878688e-01
4.86626513e-02 -1.25892475e-01 1.87487900e-01 4.37978417e-01
7.44480848e-01 -8.16024542e-01 -6.96237504e-01 1.69978365e-01
-1.09600639e+00 -7.12362111e-01 2.68259555e-01 9.35559750e-01
5.93932986e-01 1.22693799e-01 8.77112627e-01 -5.73076010e-01
7.47502208e-01 -1.01081419e+00 -6.60478771e-01 -1.31404713e-01
-4.71864998e-01 4.59499098e-02 9.36273277e-01 -4.00205821e-01
-4.73024517e-01 -3.35431755e-01 -2.57001996e-01 -8.11644971e-01
-1.35490611e-01 4.55805928e-01 -5.05671985e-02 3.75304282e-01
6.83175683e-01 6.07464373e-01 2.15770707e-01 -5.95876217e-01
-5.86316139e-02 3.21246415e-01 5.39033830e-01 -5.63729167e-01
7.51262009e-01 3.46165687e-01 -1.19927704e-01 -7.37352908e-01
-7.23253667e-01 -4.68958288e-01 -2.48922721e-01 1.00065887e-01
4.68510598e-01 -7.92885482e-01 -3.99349302e-01 8.04569304e-01
-6.89231813e-01 -2.09727094e-01 -4.76024210e-01 4.04510677e-01
-3.26740921e-01 4.08984780e-01 -2.39631489e-01 -7.58661747e-01
-4.86304551e-01 -9.11963820e-01 8.51602376e-01 2.57527083e-02
-5.50307222e-02 -8.26574445e-01 3.78158748e-01 -1.82088822e-01
5.87340891e-01 5.50000012e-01 8.90471935e-01 -1.37492239e+00
-3.34096581e-01 -4.26943630e-01 -1.37942135e-01 5.68543196e-01
3.42350394e-01 -1.10377334e-01 -1.02501953e+00 -5.72299123e-01
5.51787280e-02 -1.92719139e-02 1.18767774e+00 2.95223653e-01
2.00689459e+00 -4.48031366e-01 -6.44195229e-02 7.61043847e-01
1.15946305e+00 4.91089880e-01 7.36571491e-01 3.45876068e-01
8.56725276e-01 2.68296450e-01 5.44671178e-01 5.19169509e-01
6.81951120e-02 5.80368280e-01 6.45489275e-01 1.28269166e-01
3.93427074e-01 -6.73529953e-02 5.33553839e-01 7.31012940e-01
1.37095645e-01 -2.37976044e-01 -1.19000089e+00 6.40271068e-01
-1.91239715e+00 -1.31985855e+00 1.12839326e-01 2.35228372e+00
3.99173886e-01 1.19544089e-01 2.27735505e-01 5.05959570e-01
7.58490801e-01 4.83194351e-01 -9.32635605e-01 -2.89733499e-01
-1.27007440e-01 1.66827574e-01 -4.24938798e-02 -1.07170343e-01
-1.44499135e+00 4.42655802e-01 4.87841988e+00 9.20058012e-01
-1.47033310e+00 -1.45614102e-01 7.37744033e-01 -1.22439280e-01
-3.08389723e-01 -5.64997673e-01 -2.31654674e-01 9.36380208e-01
9.04128253e-01 -5.19539595e-01 1.88396692e-01 7.14626670e-01
1.84975713e-02 5.80652773e-01 -1.00717211e+00 1.24111307e+00
1.10991143e-01 -8.84132504e-01 3.14283788e-01 4.18824406e-04
5.38798153e-01 2.74851948e-01 4.84822810e-01 6.72312081e-01
-8.14540312e-02 -9.37558115e-01 3.86928171e-01 5.82990468e-01
4.42734420e-01 -9.32305872e-01 7.05066800e-01 3.07355598e-02
-1.24851811e+00 -5.81980944e-01 -1.84919462e-01 1.11653544e-02
-2.08307594e-01 6.55240417e-01 -2.95896232e-01 5.73061764e-01
8.60515475e-01 1.15985537e+00 -5.39026976e-01 1.38763344e+00
3.98601964e-02 7.93968618e-01 -9.73267183e-02 3.03234607e-01
2.30699629e-01 -4.01924431e-01 1.12383258e+00 9.59872663e-01
6.45540118e-01 -2.00627625e-01 2.53258705e-01 6.27488196e-01
-2.41282396e-02 2.67851949e-01 -9.72936928e-01 -2.24259138e-01
3.08495700e-01 9.18241382e-01 -4.67805266e-01 -1.43027762e-02
-3.02308202e-01 9.51669872e-01 3.09018761e-01 2.64853984e-01
-8.17266047e-01 -5.91934919e-01 1.18464005e+00 8.29007030e-02
2.73684829e-01 2.75324285e-02 6.89606965e-02 -1.36931098e+00
3.07947427e-01 -1.14710021e+00 1.04582953e+00 -2.71852583e-01
-1.98488939e+00 6.85886681e-01 -1.43428490e-01 -1.82213402e+00
-3.09408903e-01 -5.98863602e-01 -1.21279597e+00 8.16597283e-01
-1.44230604e+00 -7.07065642e-01 -3.39774638e-01 6.78694367e-01
6.36563540e-01 -5.64629376e-01 6.74973011e-01 6.45760596e-01
-9.53987479e-01 8.33590448e-01 3.19291025e-01 4.02264625e-01
7.27275074e-01 -1.33694208e+00 7.23909795e-01 1.21525919e+00
1.18893236e-01 3.49762410e-01 4.20132041e-01 -4.53158855e-01
-1.12867808e+00 -1.43290079e+00 1.53616682e-01 -2.82544971e-01
7.51616418e-01 -8.03775117e-02 -1.64560044e+00 6.44542456e-01
-1.57374635e-01 4.34707642e-01 7.17624247e-01 9.18094162e-03
-7.48026192e-01 -4.13852781e-01 -1.03720403e+00 6.04278624e-01
9.94045258e-01 -5.63169658e-01 -6.59141064e-01 8.82889107e-02
5.38945675e-01 -2.39103198e-01 -4.32401061e-01 5.02753735e-01
2.33294919e-01 -7.64842212e-01 9.53577697e-01 -8.53407443e-01
3.28693718e-01 -4.28358525e-01 -1.93097174e-01 -1.62642562e+00
-4.32160795e-01 -5.00952363e-01 -5.10145247e-01 1.11895669e+00
1.88748494e-01 -1.40989971e+00 3.46904814e-01 1.86807476e-02
-2.79439509e-01 -7.43228436e-01 -9.17211950e-01 -1.12496459e+00
1.17910922e-01 -4.35400218e-01 8.79118085e-01 1.30957031e+00
-2.94515163e-01 4.40566875e-02 -2.15452701e-01 2.76607484e-01
6.06127858e-01 2.42107078e-01 4.86910224e-01 -1.59137344e+00
-7.12929443e-02 -8.89557898e-01 -8.61422658e-01 -7.82331169e-01
2.67452329e-01 -1.20542657e+00 -1.51883394e-01 -1.04678750e+00
-1.13789730e-01 -3.05177718e-01 -1.09714270e+00 5.33036947e-01
-4.20206696e-01 5.03728949e-02 -2.07671985e-01 1.50390610e-01
-4.89020169e-01 9.99442399e-01 8.25989306e-01 -3.40325445e-01
-2.34208584e-01 3.12784106e-01 -9.32407856e-01 5.56983054e-01
1.38754618e+00 -4.88880932e-01 -4.68491524e-01 -3.59060526e-01
-1.14941895e-01 -2.87973523e-01 4.55639094e-01 -1.03857791e+00
-6.77768439e-02 2.23312341e-02 4.07960325e-01 -5.52994847e-01
1.13987602e-01 -7.08639801e-01 -2.55112886e-01 7.56583512e-01
-3.46698999e-01 7.94391274e-01 4.04822677e-01 8.98262918e-01
-4.73177433e-01 1.24176294e-01 7.11097002e-01 2.35310122e-01
-9.25944507e-01 7.75977373e-01 -1.60952061e-01 4.71646786e-01
1.01738346e+00 1.41713992e-01 -4.69572186e-01 -2.70334512e-01
-2.80575097e-01 4.04617399e-01 -2.82709040e-02 8.90399337e-01
7.57347047e-01 -1.69084430e+00 -1.08295536e+00 7.05531597e-01
4.82419133e-01 -1.25262529e-01 4.55307394e-01 7.49836087e-01
-4.41437900e-01 2.49555372e-02 -7.05889285e-01 -7.18772829e-01
-1.02847242e+00 6.14408374e-01 3.82026851e-01 -1.32139251e-01
-9.80965734e-01 6.40727162e-01 3.67144585e-01 -3.37112486e-01
1.82998359e-01 -1.95233747e-01 -3.83598775e-01 9.71555263e-02
9.53153133e-01 6.44954920e-01 1.02124333e-01 -4.66239750e-01
-4.45875674e-01 4.38165247e-01 -3.66574049e-01 3.16143394e-01
1.49076426e+00 2.65836477e-01 -1.78548560e-01 5.05428612e-01
1.32823777e+00 -1.35284722e-01 -1.12479246e+00 -4.01283234e-01
9.85709205e-02 -8.12847912e-01 -2.89296787e-02 -4.72339541e-01
-1.40110385e+00 8.61295342e-01 9.35878456e-01 5.55323303e-01
1.52834475e+00 -3.24010372e-01 8.76300931e-01 3.65323454e-01
-3.50349909e-03 -5.57032406e-01 3.42102885e-01 5.83341241e-01
1.28117192e+00 -1.33881474e+00 -3.30789685e-01 9.31570679e-02
-6.71735048e-01 8.80339563e-01 9.02075946e-01 -4.17015374e-01
6.83075130e-01 -9.08112079e-02 1.84609424e-02 -3.52615833e-01
-6.09463930e-01 -9.77065265e-02 5.83537400e-01 4.78242636e-01
3.44554842e-01 4.42699753e-02 1.04394313e-02 5.98169029e-01
6.63967803e-02 -8.96664679e-01 1.72255039e-01 7.31380522e-01
-1.95259228e-01 -8.50714386e-01 -3.30462545e-01 9.19289052e-01
-7.78650224e-01 7.44281039e-02 -2.71299668e-02 3.51517022e-01
-3.21555406e-01 6.65991187e-01 6.01337016e-01 -2.84990579e-01
5.80565095e-01 1.81747407e-01 -1.46295258e-03 -3.53594244e-01
-4.10495013e-01 -2.23405167e-01 -5.22179365e-01 -6.23406649e-01
2.01074928e-01 -9.22267377e-01 -1.06452072e+00 -1.19102396e-01
1.51748791e-01 2.72164911e-01 1.86169326e-01 5.54363430e-01
6.45412922e-01 8.90958250e-01 1.01553714e+00 -6.11716628e-01
-5.80362439e-01 -7.36764729e-01 -5.47716558e-01 6.38164222e-01
8.68423462e-01 -7.11927593e-01 -5.56184351e-01 -5.36869407e-01] | [7.562896728515625, 2.37819242477417] |
3865edd1-ec5d-4c51-9771-757661f24039 | multi-omic-data-integration-and-feature | 2206.10699 | null | https://arxiv.org/abs/2206.10699v2 | https://arxiv.org/pdf/2206.10699v2.pdf | Multi-Omic Data Integration and Feature Selection for Survival-based Patient Stratification via Supervised Concrete Autoencoders | Cancer is a complex disease with significant social and economic impact. Advancements in high-throughput molecular assays and the reduced cost for performing high-quality multi-omics measurements have fuelled insights through machine learning . Previous studies have shown promise on using multiple omic layers to predict survival and stratify cancer patients. In this paper, we developed a Supervised Autoencoder (SAE) model for survival-based multi-omic integration which improves upon previous work, and report a Concrete Supervised Autoencoder model (CSAE), which uses feature selection to jointly reconstruct the input features as well as predict survival. Our experiments show that our models outperform or are on par with some of the most commonly used baselines, while either providing a better survival separation (SAE) or being more interpretable (CSAE). We also perform a feature selection stability analysis on our models and notice that there is a power-law relationship with features which are commonly associated with survival. The code for this project is available at: https://github.com/phcavelar/coxae | ['Sophia Tsoka', 'Min Wu', 'Sophia Karagiannis', 'Roman Laddach', 'Pedro Henrique da Costa Avelar'] | 2022-06-21 | null | null | null | null | ['data-integration'] | ['knowledge-base'] | [-1.12078495e-01 -2.44539678e-01 -2.95068383e-01 -3.43683690e-01
-4.79717195e-01 -1.81501403e-01 5.99227428e-01 7.17651367e-01
-3.63688260e-01 7.46593893e-01 5.91266453e-01 -3.52033198e-01
-4.64777648e-01 -7.81369567e-01 -3.80787432e-01 -9.16778207e-01
-2.68321067e-01 5.56695938e-01 -3.94712240e-01 -8.35160539e-02
-2.94809401e-01 5.81020415e-01 -1.42012656e+00 4.90056515e-01
6.25482321e-01 7.67461240e-01 -1.31254390e-01 1.01801205e+00
9.95831043e-02 7.36925721e-01 -4.21727240e-01 -3.41525316e-01
-1.62719905e-01 -4.37402159e-01 -7.09497094e-01 -3.83694828e-01
-1.00107752e-01 -6.80871904e-02 -4.43160117e-01 5.78454673e-01
5.20481944e-01 -2.98797518e-01 7.86219060e-01 -1.15968394e+00
-6.74202621e-01 3.06179136e-01 1.44081458e-01 -4.98832725e-02
1.39567256e-01 2.61690915e-02 9.92690921e-01 -6.73446238e-01
6.59337640e-01 1.03644443e+00 9.42211747e-01 5.25621653e-01
-1.08267188e+00 -3.14595699e-01 -3.77587229e-01 2.26881623e-01
-1.22717905e+00 -4.79288161e-01 2.97533453e-01 -5.83327770e-01
1.09206557e+00 6.24755323e-01 9.13273990e-01 1.34057975e+00
8.79312515e-01 8.17004025e-01 1.02383459e+00 -3.72354776e-01
3.06800544e-01 2.34726995e-01 2.25660056e-01 7.75405705e-01
4.21287715e-01 5.02039135e-01 -4.24774855e-01 -4.56932455e-01
3.10615778e-01 8.33670199e-01 -1.88490406e-01 -4.18541022e-03
-1.26594865e+00 1.07086492e+00 4.37142134e-01 6.55827045e-01
-6.35627449e-01 1.13506280e-01 2.43428186e-01 3.33498031e-01
5.64810574e-01 5.78863502e-01 -6.71181083e-01 1.67124242e-01
-9.61208701e-01 2.63720956e-02 8.84714305e-01 3.19984794e-01
4.79489505e-01 -9.89519879e-02 -5.35600865e-03 5.94490588e-01
2.38607422e-01 1.59930989e-01 8.39545071e-01 -7.52076805e-01
-6.00844562e-01 8.45696509e-01 -1.66427031e-01 -7.67609894e-01
-9.52291071e-01 -6.20588601e-01 -1.03032148e+00 6.45126551e-02
2.68576175e-01 -6.68982118e-02 -8.28605771e-01 1.43381572e+00
2.25630954e-01 1.27595887e-01 4.31311518e-01 6.60172820e-01
1.09217381e+00 3.08642000e-01 3.29733729e-01 2.97584627e-02
1.58609855e+00 -6.77285314e-01 -6.36711955e-01 9.53902826e-02
9.26373124e-01 -3.67043346e-01 6.00960433e-01 3.63364786e-01
-4.80721027e-01 -6.96188360e-02 -1.04953694e+00 -4.27381769e-02
-8.27546299e-01 1.32078633e-01 1.23280239e+00 6.27563596e-01
-9.53477263e-01 1.05427682e+00 -1.21720064e+00 -7.54591584e-01
5.09571493e-01 4.94338363e-01 -6.92867935e-01 -1.59557052e-02
-1.06559265e+00 8.83665144e-01 4.36632007e-01 -2.66698599e-01
-9.16920245e-01 -8.79497826e-01 -7.14364767e-01 7.76576251e-02
-2.13748276e-01 -1.28749299e+00 8.34215522e-01 -9.87222016e-01
-1.43251574e+00 6.49194479e-01 -3.90937962e-02 -6.10540330e-01
1.76581547e-01 -2.44251475e-01 -6.26028061e-01 1.46818876e-01
-2.96925873e-01 5.96867502e-01 2.23150060e-01 -7.44302094e-01
-4.31719810e-01 -5.33936203e-01 -2.49927700e-01 -1.04990877e-01
-7.08821416e-01 1.01823211e-01 2.06376880e-01 -5.02274215e-01
-2.20611811e-01 -1.00573385e+00 -4.22142535e-01 -1.58732384e-01
-3.35875839e-01 -2.08450660e-01 4.60141212e-01 -8.59288990e-01
1.07330847e+00 -1.84571993e+00 4.93130028e-01 -3.00976217e-01
4.77310449e-01 -4.18718979e-02 2.36649327e-02 8.19202423e-01
-2.03669131e-01 3.09623271e-01 -2.38932148e-01 -5.81621289e-01
-1.84303418e-01 2.82040864e-01 1.79623410e-01 7.88667023e-01
4.36332196e-01 9.06292319e-01 -8.00836205e-01 -1.45522997e-01
2.53254980e-01 9.51123118e-01 -3.64156216e-01 2.10961014e-01
-1.51719779e-01 3.36226046e-01 -2.05322593e-01 1.21322036e+00
2.43528500e-01 -6.39072001e-01 1.81099147e-01 1.16573581e-02
4.53662947e-02 -6.35836124e-02 -6.50290728e-01 1.53723943e+00
-2.13068902e-01 5.44494212e-01 -2.22105443e-01 -8.40014040e-01
4.56825852e-01 3.94555688e-01 7.95947611e-01 -2.69620657e-01
5.31494439e-01 1.86889932e-01 1.32644206e-01 -4.19278353e-01
1.01183742e-01 -4.84026879e-01 7.71183595e-02 1.92867294e-01
4.28661585e-01 3.99609566e-01 -3.38654220e-02 1.24531686e-01
1.32822204e+00 -1.62234381e-01 6.65982485e-01 -4.35501367e-01
5.54419637e-01 1.74098954e-01 6.16273224e-01 3.74657810e-01
-1.62326306e-01 4.29411083e-01 3.67361486e-01 -6.43411577e-01
-8.30543220e-01 -6.74762428e-01 -5.29268384e-01 8.09035778e-01
-3.74796033e-01 -4.81144428e-01 -3.89446437e-01 -4.53291059e-01
3.69282514e-01 5.98798513e-01 -9.93582904e-01 -1.94515079e-01
2.97674593e-02 -1.41265142e+00 7.46505201e-01 5.78811407e-01
-2.86207646e-01 -7.14752197e-01 -4.97018605e-01 7.68733546e-02
2.34761387e-01 -5.46332121e-01 4.34916168e-01 5.68658352e-01
-9.76352870e-01 -1.22782433e+00 -7.97905803e-01 -2.93942422e-01
4.32653725e-01 -1.53081223e-01 9.60188985e-01 2.47208938e-01
-5.47159612e-01 1.84765011e-01 -4.34520334e-01 -9.80727136e-01
-6.46934271e-01 1.21300362e-01 4.14264113e-01 -1.72146901e-01
7.07203150e-01 -4.39135253e-01 -7.76005626e-01 -8.88128728e-02
-1.03209198e+00 7.15626776e-02 7.89110124e-01 1.13779473e+00
6.17248535e-01 -6.40214309e-02 8.30635011e-01 -8.02526295e-01
4.46720958e-01 -1.02740443e+00 -1.19246833e-01 -3.06489486e-02
-1.02240801e+00 5.63554019e-02 6.83401406e-01 -8.31372887e-02
-4.06571805e-01 9.43500828e-03 -4.77775812e-01 -4.24074203e-01
-5.92597902e-01 8.13634932e-01 1.60920039e-01 5.48229441e-02
5.95891178e-01 1.07165624e-03 3.78382862e-01 -5.17565906e-01
-1.15371257e-01 8.32980037e-01 -5.80311054e-03 3.09515912e-02
3.02516341e-01 7.42178679e-01 2.49377698e-01 -7.41850555e-01
-6.24530613e-01 -6.09090567e-01 -5.39426386e-01 1.32444575e-01
9.39365029e-01 -1.04224885e+00 -7.93147087e-01 4.07959819e-01
-5.73282242e-01 -1.35611027e-01 -5.35498410e-02 7.64168501e-01
-3.32797170e-01 1.39772281e-01 -8.28592539e-01 -5.60243666e-01
-6.46340668e-01 -9.15172756e-01 1.06117642e+00 3.40824604e-01
-3.71559292e-01 -1.37364447e+00 4.83805269e-01 1.76920339e-01
5.37571132e-01 6.61945462e-01 9.94315326e-01 -1.11038184e+00
-2.53931098e-02 -6.08828843e-01 -3.42632644e-02 9.92318243e-02
2.13901058e-01 2.21564714e-02 -1.17460299e+00 -3.87812674e-01
-9.08361599e-02 -1.19355999e-01 1.15085185e+00 4.81583208e-01
1.01750994e+00 -1.26476586e-01 -6.42637491e-01 9.21306133e-01
1.65910757e+00 6.84883967e-02 3.95033211e-01 3.07259470e-01
4.42868024e-01 5.46202481e-01 1.43049106e-01 4.64481950e-01
3.76662523e-01 2.24398956e-01 3.83152574e-01 -2.78882831e-01
3.66699956e-02 8.92192796e-02 2.23573849e-01 7.07130373e-01
-6.16672151e-02 -4.59335268e-01 -1.07011449e+00 5.40608406e-01
-1.77487469e+00 -6.92132950e-01 -1.74131468e-01 1.97607934e+00
6.11307383e-01 -4.16391581e-01 3.38277698e-01 1.01769768e-01
5.24312025e-03 -2.58316636e-01 -5.13879836e-01 -4.25286651e-01
-3.85624021e-01 3.07116538e-01 4.75654989e-01 3.98113281e-01
-1.17150438e+00 4.70034212e-01 7.05000687e+00 2.92291462e-01
-1.18422830e+00 1.24773614e-01 6.95656598e-01 -2.83660740e-01
-2.05131456e-01 -1.18602477e-01 -5.16664684e-01 2.46819302e-01
1.64154279e+00 -1.80668578e-01 2.21987560e-01 6.32357836e-01
-1.86755061e-02 3.10853086e-02 -1.22641170e+00 6.44617140e-01
-1.45605486e-03 -1.56635499e+00 -1.50486544e-01 3.44504923e-01
3.59052986e-01 4.56853956e-01 -1.25551939e-01 1.68584630e-01
4.26306814e-01 -1.29490530e+00 1.34476677e-01 8.13942909e-01
6.22123063e-01 -6.97911680e-01 1.11464429e+00 2.11955220e-01
-6.77987814e-01 -3.91649902e-01 -1.99262246e-01 -1.49388492e-01
-2.58943170e-01 8.82019877e-01 -9.69493866e-01 8.66679907e-01
7.42919981e-01 1.03514302e+00 -8.28968167e-01 9.16210413e-01
1.14705354e-01 6.55496299e-01 -2.00424120e-01 -1.88672230e-01
-1.59029856e-01 5.41822612e-02 4.19654667e-01 1.45875514e+00
4.71905798e-01 -3.97072220e-03 -3.48545834e-02 6.49374306e-01
1.28002957e-01 2.66108125e-01 -4.27489877e-01 -7.50319958e-01
6.46495372e-02 1.33809781e+00 -6.32287741e-01 -4.15519714e-01
-5.39106250e-01 9.83708858e-01 7.44229779e-02 -1.55055337e-02
-5.55130720e-01 -2.83260196e-01 9.88894939e-01 -1.41048759e-01
-3.27983648e-02 -2.50769835e-02 -2.49948293e-01 -1.35006988e+00
-6.16693556e-01 -8.41548324e-01 8.73586893e-01 -3.43486607e-01
-1.46306622e+00 4.93335694e-01 -4.44430232e-01 -1.04332328e+00
-1.57422870e-01 -8.81399393e-01 -4.36200976e-01 7.28961289e-01
-1.53519475e+00 -1.25894654e+00 -2.74510980e-01 4.65816528e-01
1.97857752e-01 -2.87606627e-01 1.59669018e+00 3.01537633e-01
-9.79419649e-01 3.73653769e-01 5.14786065e-01 -4.49153371e-02
6.61785781e-01 -1.28488481e+00 -1.44151613e-01 4.43339497e-01
-3.68811153e-02 7.59310484e-01 6.64563060e-01 -7.05154955e-01
-1.78443348e+00 -1.08217835e+00 8.55182409e-01 -6.23253047e-01
6.77139819e-01 -1.69736847e-01 -6.06644154e-01 7.40348279e-01
2.32382357e-01 -1.40981838e-01 1.70133948e+00 5.00923932e-01
-1.32737175e-01 9.31018963e-02 -1.07655585e+00 3.30973953e-01
6.40088379e-01 -3.58058184e-01 -2.58790106e-01 4.68533069e-01
5.40297329e-01 -2.18501892e-02 -1.45608115e+00 4.14109856e-01
8.41120243e-01 -9.06450152e-01 9.09742773e-01 -9.96497095e-01
7.84848452e-01 -2.30266109e-01 -2.74763316e-01 -1.38754237e+00
-7.04639375e-01 3.16112600e-02 -4.77539062e-01 6.53789401e-01
3.42624158e-01 -7.89089859e-01 7.54000604e-01 3.14229310e-01
-1.19889714e-01 -1.06644034e+00 -7.93119550e-01 -6.78077579e-01
3.55756998e-01 -8.99659619e-02 6.02782249e-01 1.13855016e+00
2.09928140e-01 8.89957771e-02 -3.08637321e-01 2.39316851e-01
4.15998101e-01 3.56894955e-02 4.78114903e-01 -1.62042606e+00
-5.98311186e-01 -4.67157334e-01 -7.66060650e-01 7.34539777e-02
-4.42727767e-02 -1.24892163e+00 -5.30998528e-01 -1.54239213e+00
5.17072141e-01 -6.27645552e-02 -9.16316628e-01 6.22611761e-01
-8.14912021e-02 1.02986746e-01 -8.36526752e-02 2.54484385e-01
-3.17576826e-01 3.18867773e-01 6.21355772e-01 -5.23605384e-02
-1.83148712e-01 -2.35759035e-01 -1.15762949e+00 5.98240077e-01
8.92044246e-01 -5.41082263e-01 1.50893867e-01 -1.67153254e-01
2.67515600e-01 1.45459026e-01 4.16435063e-01 -1.11563945e+00
1.42455533e-01 -1.95056185e-01 1.00830829e+00 -2.62255222e-01
3.55003655e-01 -6.67389333e-01 6.56163692e-01 1.05993700e+00
-1.12539470e-01 -1.11382462e-01 3.29746276e-01 6.31878495e-01
-9.95798409e-02 4.98960055e-02 5.35030127e-01 -5.87914288e-02
-4.26021665e-01 3.41564894e-01 -5.67940831e-01 -8.84867132e-01
9.34749424e-01 -6.50907517e-04 -4.42504823e-01 -7.76730329e-02
-8.79006088e-01 1.32446870e-01 6.26687169e-01 3.65242839e-01
4.96202916e-01 -1.17324543e+00 -9.35188413e-01 1.57547787e-01
3.31085712e-01 -4.79256183e-01 2.42801011e-01 1.11156380e+00
-6.41674221e-01 6.75320804e-01 -2.90741652e-01 -5.05828142e-01
-1.29834127e+00 6.64946795e-01 5.30665517e-01 -3.21653545e-01
-5.34862101e-01 7.76555538e-01 -7.49445632e-02 -4.79844004e-01
-1.14765558e-02 -5.95641844e-02 -3.43373567e-01 2.54488856e-01
7.00402439e-01 4.50682789e-01 5.95468022e-02 -3.39035422e-01
-5.95320761e-01 5.58412634e-02 -1.00871436e-01 3.85523051e-01
1.93915761e+00 2.04199463e-01 -2.38793224e-01 6.27627969e-01
1.22598505e+00 -1.41917780e-01 -5.69902897e-01 7.89902955e-02
4.98243645e-02 -1.21686667e-01 4.22050446e-01 -1.13016486e+00
-7.75604665e-01 5.69915831e-01 8.55329812e-01 1.82512596e-01
1.15944219e+00 1.30741715e-01 6.63357437e-01 4.12632346e-01
-7.44409487e-02 -6.54122055e-01 -2.72526681e-01 1.78946108e-01
6.48206592e-01 -1.33425891e+00 3.04406315e-01 -8.90885480e-03
-3.20373923e-01 1.23396659e+00 1.18999697e-01 -6.93375543e-02
8.50236297e-01 4.00443166e-01 1.06165320e-01 -4.79316622e-01
-1.04644036e+00 -1.57329544e-01 1.55463889e-01 2.98418552e-01
7.36925066e-01 4.88736659e-01 -3.07151854e-01 9.61322606e-01
-2.24324450e-01 3.67294699e-01 4.07692194e-01 7.77340949e-01
-3.34329575e-01 -1.21047676e+00 -2.87873328e-01 7.65751481e-01
-7.70012081e-01 -1.28293529e-01 -6.07636988e-01 5.75009763e-01
3.16443108e-02 8.70517373e-01 6.73511019e-03 -6.91671431e-01
-1.56497717e-01 4.07555670e-01 1.03568554e-01 -4.32979822e-01
-7.53646791e-01 -5.33933342e-02 1.86432287e-01 -5.94334960e-01
-3.48065197e-01 -8.08676600e-01 -1.09018409e+00 -4.74628210e-01
-4.29942071e-01 -1.17117856e-02 9.24979925e-01 7.29943573e-01
8.60476375e-01 8.97592306e-01 4.10992295e-01 -5.98095238e-01
-1.97116062e-01 -1.04945350e+00 -5.20225465e-01 2.84605801e-01
5.27464688e-01 -4.58122432e-01 -4.37785506e-01 -1.27556443e-01] | [6.07305383682251, 5.680384635925293] |
c5063656-d65f-4545-b73e-938e70aeea54 | hyperspectral-image-reconstruction-from | 2209.07891 | null | https://arxiv.org/abs/2209.07891v1 | https://arxiv.org/pdf/2209.07891v1.pdf | Hyperspectral Image Reconstruction from Multispectral Images Using Non-Local Filtering | Using light spectra is an essential element in many applications, for example, in material classification. Often this information is acquired by using a hyperspectral camera. Unfortunately, these cameras have some major disadvantages like not being able to record videos. Therefore, multispectral cameras with wide-band filters are used, which are much cheaper and are often able to capture videos. However, using multispectral cameras requires an additional reconstruction step to yield spectral information. Usually, this reconstruction step has to be done in the presence of imaging noise, which degrades the reconstructed spectra severely. Typically, same or similar pixels are found across the image with the advantage of having independent noise. In contrast to state-of-the-art spectral reconstruction methods which only exploit neighboring pixels by block-based processing, this paper introduces non-local filtering in spectral reconstruction. First, a block-matching procedure finds similar non-local multispectral blocks. Thereafter, the hyperspectral pixels are reconstructed by filtering the matched multispectral pixels collaboratively using a reconstruction Wiener filter. The proposed novel procedure even works under very strong noise. The method is able to lower the spectral angle up to 18% and increase the peak signal-to-noise-ratio up to 1.1dB in noisy scenarios compared to state-of-the-art methods. Moreover, the visual results are much more appealing. | ['André Kaup', 'Jürgen Seiler', 'Frank Sippel'] | 2022-09-16 | null | null | null | null | ['spectral-reconstruction', 'material-classification'] | ['computer-vision', 'computer-vision'] | [ 8.60411525e-01 -6.70581579e-01 1.63995996e-01 4.09064591e-02
-6.76004827e-01 -4.99711454e-01 3.58010054e-01 1.95263565e-01
-5.32538295e-01 8.04821014e-01 -3.11159819e-01 4.56607901e-02
-4.49933171e-01 -1.04770029e+00 -5.70527732e-01 -1.23148465e+00
4.23771054e-01 -2.25944072e-01 2.43130714e-01 1.73220485e-02
9.25430208e-02 5.16206920e-01 -1.96813917e+00 2.18159705e-01
1.19318283e+00 1.01380682e+00 8.80921960e-01 3.13495278e-01
-1.50396466e-01 3.56049538e-01 -2.01870680e-01 -7.70890564e-02
4.19643760e-01 -5.45508027e-01 -1.88650176e-01 5.23808599e-01
5.06168425e-01 -4.22238678e-01 -5.57774790e-02 1.74377596e+00
1.71266347e-01 2.61550725e-01 3.03886443e-01 -4.04769212e-01
-9.03568417e-02 3.30791354e-01 -9.58706915e-01 -2.30659142e-01
3.15382510e-01 1.91516876e-01 7.92305231e-01 -7.95059979e-01
3.98267180e-01 7.13440180e-01 5.33703148e-01 -6.09262623e-02
-1.49031758e+00 -5.35815656e-01 -3.20624739e-01 4.01143223e-01
-1.26299953e+00 -3.36557895e-01 1.01779401e+00 -2.99820691e-01
2.56034642e-01 4.21032101e-01 8.68167698e-01 6.14923179e-01
-4.40474749e-02 2.26929158e-01 1.49683022e+00 -4.75686312e-01
1.36600006e-02 -1.44512402e-02 -5.95181920e-02 2.74065137e-01
5.35724699e-01 6.38804436e-02 -3.68665218e-01 9.55096334e-02
5.98569751e-01 4.01758045e-01 -8.99285913e-01 -2.30119154e-01
-1.15866876e+00 4.74980265e-01 4.18563217e-01 6.28164232e-01
-7.43106842e-01 -2.09437251e-01 7.70895109e-02 1.08263291e-01
3.13000947e-01 2.10240930e-01 9.68954265e-02 2.39844322e-01
-1.07675493e+00 -9.04993042e-02 3.06803316e-01 1.72883034e-01
9.52031791e-01 -5.20055331e-02 3.69589925e-01 9.83401477e-01
2.85237938e-01 7.26803780e-01 2.25491330e-01 -7.95688748e-01
2.25482836e-01 2.23760024e-01 2.28974089e-01 -1.02368248e+00
-2.56361902e-01 -6.30172133e-01 -1.15151870e+00 3.76872718e-01
5.22097647e-01 1.84030876e-01 -5.48381984e-01 1.19507265e+00
2.73768574e-01 1.48264885e-01 1.83663651e-01 1.06806350e+00
4.44152862e-01 9.47259605e-01 -7.94303790e-02 -8.71285915e-01
1.25818431e+00 -6.00894809e-01 -8.85688066e-01 -1.45794958e-01
-9.31634307e-02 -1.16566098e+00 8.39866817e-01 7.90672243e-01
-9.43752229e-01 -6.80625319e-01 -9.62165892e-01 4.39910263e-01
-8.31519812e-02 3.45183343e-01 2.21992195e-01 7.29536772e-01
-5.44350088e-01 6.75474823e-01 -5.44583738e-01 -3.43624264e-01
1.31840378e-01 -1.12485863e-01 -4.24418479e-01 -3.97906840e-01
-8.14652026e-01 6.75322473e-01 5.58349073e-01 2.99727559e-01
-3.11917186e-01 -6.64306998e-01 -5.49694419e-01 1.00015894e-01
7.30507433e-01 -2.65047699e-01 6.84692025e-01 -1.11263657e+00
-1.64581370e+00 5.82615614e-01 -1.64333701e-01 -1.25092298e-01
4.06698167e-01 -1.35957360e-01 -7.22407460e-01 6.23985708e-01
-1.10174119e-01 -1.67366415e-01 9.70041692e-01 -1.27435100e+00
-6.11953974e-01 -3.58526707e-01 -3.86833884e-02 1.84934720e-01
-3.66670191e-01 -1.76891729e-01 -2.87846416e-01 -5.50622880e-01
6.19029284e-01 -7.51268029e-01 -2.25520390e-03 1.20220423e-01
-5.23747616e-02 4.64772463e-01 8.72095168e-01 -7.64080942e-01
8.83492053e-01 -2.38088512e+00 7.81733468e-02 2.62290090e-01
-1.38563603e-01 5.28201401e-01 -3.77364904e-02 4.86493886e-01
-1.50452614e-01 -3.29836994e-01 -5.95197797e-01 1.38356760e-01
-5.85365355e-01 -2.87260711e-01 2.29943711e-02 8.17481995e-01
-2.83587664e-01 6.24203011e-02 -8.06153357e-01 -1.77344456e-01
6.29015744e-01 5.47478557e-01 -8.91133845e-02 -8.57459307e-02
4.59311251e-03 6.52418017e-01 -1.98143259e-01 6.21230602e-01
1.01188517e+00 3.53012653e-03 3.35846901e-01 -5.88746011e-01
-5.78394711e-01 -3.52004319e-01 -1.61381328e+00 1.68740392e+00
-5.70723176e-01 4.11482573e-01 6.93895817e-01 -1.12677252e+00
8.74676526e-01 4.42671984e-01 6.00337505e-01 -3.79752159e-01
-1.47927478e-01 6.75297737e-01 -2.03217998e-01 -5.66930175e-01
2.81840801e-01 -4.12695646e-01 6.13134444e-01 8.21032971e-02
-4.17008698e-01 -3.42198342e-01 3.44274014e-01 -3.95281643e-01
5.21182835e-01 1.36983082e-01 4.57631886e-01 -9.27343294e-02
1.03155684e+00 -1.20983660e-01 4.61306244e-01 4.01084125e-01
1.57122940e-01 3.94296288e-01 -1.28920496e-01 -3.44524607e-02
-9.45238709e-01 -8.13878715e-01 -4.60200340e-01 3.32944810e-01
4.39999163e-01 -9.13716331e-02 -4.82108742e-01 -3.19053940e-02
-8.43660831e-02 6.35145247e-01 -3.02381553e-02 1.80470452e-01
-1.18727110e-01 -8.80819619e-01 -7.94828683e-02 -1.96115356e-02
1.07577538e+00 -6.34182036e-01 -7.69987702e-01 4.14962202e-01
-3.35379660e-01 -1.05849290e+00 4.37649973e-02 -1.15473449e-01
-1.02974272e+00 -1.21475685e+00 -8.19069803e-01 -2.73952007e-01
5.03151417e-01 9.54042673e-01 3.70372295e-01 -1.68417931e-01
-2.64954865e-01 1.71155199e-01 -4.80538368e-01 6.10352904e-02
-3.58301550e-01 -3.88388008e-01 -2.09088787e-01 6.96350396e-01
8.57701749e-02 -6.26829267e-01 -6.37973130e-01 3.22772503e-01
-1.24905694e+00 1.95061266e-01 8.70191813e-01 9.36977267e-01
7.31043696e-01 9.45392668e-01 3.35008591e-01 -7.66815841e-01
1.13648497e-01 -5.76102696e-02 -1.01528478e+00 1.09248407e-01
-3.30841124e-01 -4.04273748e-01 9.23929751e-01 -2.97588706e-01
-1.41295481e+00 3.32634091e-01 5.38209230e-02 -3.08443695e-01
-3.92389119e-01 7.00219750e-01 -2.37116411e-01 -2.83803165e-01
5.82568705e-01 4.66395885e-01 1.69001743e-01 -6.39943540e-01
1.08275577e-01 7.08306015e-01 4.72380191e-01 -7.09286630e-02
1.06493437e+00 7.04405248e-01 4.34950411e-01 -1.47301626e+00
-5.56583881e-01 -8.39414835e-01 -4.72987294e-01 -5.83399475e-01
7.75991142e-01 -9.26176369e-01 -5.87923408e-01 7.48223186e-01
-8.01760793e-01 1.54791832e-01 -1.69831689e-03 8.53378713e-01
-2.78010786e-01 8.80478501e-01 -4.04194593e-01 -1.02865243e+00
-9.80105475e-02 -1.24800873e+00 5.77888072e-01 3.22351575e-01
3.80034626e-01 -5.83324313e-01 -5.10770440e-01 5.64327300e-01
4.17713702e-01 1.75524473e-01 6.32280409e-01 2.62653768e-01
-7.39829957e-01 -2.10356340e-01 -3.44135195e-01 4.46239322e-01
3.95889819e-01 5.93012152e-03 -9.72577214e-01 -3.33756566e-01
2.34083191e-01 7.45813027e-02 8.90419304e-01 6.04286730e-01
1.14760911e+00 1.54817894e-01 -1.21466979e-01 5.80564976e-01
2.00793123e+00 3.07222843e-01 6.89947605e-01 4.97770816e-01
4.60773736e-01 8.64333570e-01 8.32265437e-01 3.13405395e-01
-4.91357207e-01 6.15509510e-01 6.92488253e-01 -1.24177851e-01
-1.40967201e-02 2.83319712e-01 1.87774375e-01 3.97120386e-01
-3.93128753e-01 -1.94293082e-01 -5.18807888e-01 4.67555225e-01
-1.72044611e+00 -1.35646272e+00 -7.65891314e-01 2.55806112e+00
6.45887315e-01 -2.27538407e-01 -1.68033749e-01 6.20171785e-01
9.64241087e-01 4.73894328e-01 -2.69771099e-01 3.82445484e-01
-4.01527584e-01 2.69578665e-01 8.10569644e-01 3.50953668e-01
-1.04512167e+00 3.86903942e-01 4.33186674e+00 8.56900990e-01
-1.23013675e+00 8.31694156e-02 3.18306545e-03 -2.93673500e-02
-1.58812493e-01 6.82571754e-02 -2.69128084e-01 4.54626530e-01
5.17603517e-01 1.53666288e-01 7.02907979e-01 3.61859262e-01
6.18829668e-01 -7.88891912e-01 -4.19481575e-01 1.33474898e+00
4.08264473e-02 -9.34467256e-01 -1.86836466e-01 9.05305818e-02
6.42595172e-01 -4.89819050e-01 -1.19056731e-01 -5.79816103e-01
-4.97155070e-01 -5.09138525e-01 5.31625926e-01 7.28982568e-01
7.93839157e-01 -9.12232697e-01 7.40263999e-01 4.86520737e-01
-1.17047524e+00 -2.57860720e-01 -5.20506680e-01 9.27193090e-02
3.81886154e-01 1.27868605e+00 -4.91954796e-02 1.04488277e+00
7.81667650e-01 7.14260995e-01 -1.46623105e-01 1.39284277e+00
-3.30828249e-01 4.30642992e-01 -2.62817889e-01 2.76546746e-01
1.88661441e-01 -9.57104802e-01 7.32975602e-01 7.80436039e-01
8.99292529e-01 3.44423085e-01 1.11179374e-01 7.50541687e-01
1.58052698e-01 3.23390335e-01 -5.05166829e-01 -1.00925773e-01
7.88464174e-02 1.31728601e+00 -6.69279158e-01 -2.64984578e-01
-7.43176162e-01 1.08315361e+00 -5.39742708e-01 3.17043155e-01
-5.37853718e-01 -4.12686497e-01 4.51146781e-01 2.68416971e-01
2.90578097e-01 -2.30383098e-01 1.33070024e-02 -1.17150354e+00
6.18710853e-02 -1.01874793e+00 8.71995538e-02 -1.02114260e+00
-1.04799700e+00 1.90328777e-01 -1.33415431e-01 -1.58459365e+00
1.95653647e-01 -5.79088449e-01 -1.47493094e-01 1.07120764e+00
-1.73009384e+00 -9.59476590e-01 -7.60011554e-01 5.95344245e-01
5.66061318e-01 1.92892998e-01 6.91718340e-01 4.69997674e-01
-3.76722991e-01 -3.20637733e-01 7.27904081e-01 -3.30677629e-01
7.59513378e-01 -7.92536318e-01 -6.03693545e-01 1.21016014e+00
-2.93245148e-02 3.03030044e-01 7.10064113e-01 -6.39080524e-01
-1.36087561e+00 -8.37215483e-01 3.51532668e-01 6.69417322e-01
5.86378992e-01 1.84003681e-01 -1.02488840e+00 -2.07375716e-02
3.10468704e-01 1.44872684e-02 7.07982481e-01 -4.36896414e-01
4.70155589e-02 -5.51434040e-01 -1.09548438e+00 3.15063715e-01
6.27587259e-01 -5.89052558e-01 -2.38224506e-01 2.92148113e-01
1.01345554e-02 -3.24785453e-03 -7.98044503e-01 5.19137561e-01
6.86560512e-01 -1.32727623e+00 9.45121706e-01 3.89170229e-01
2.66212225e-01 -7.84448504e-01 -2.40864247e-01 -1.41856527e+00
-3.47856998e-01 -2.86658227e-01 6.15831077e-01 1.16415632e+00
2.10086271e-01 -8.66442859e-01 3.66527259e-01 1.78424306e-02
7.43278116e-02 2.31578220e-02 -5.76728225e-01 -8.45207512e-01
-7.44276822e-01 -1.49312004e-01 2.78674662e-01 9.82210875e-01
-2.99343646e-01 8.78984928e-02 -4.26473528e-01 5.53508282e-01
1.16373360e+00 5.01153648e-01 4.95467961e-01 -1.62919450e+00
-3.89812291e-01 -4.06200409e-01 -1.40511945e-01 -5.64302802e-01
-5.80282742e-03 -5.28249562e-01 -6.64323494e-02 -1.55131245e+00
1.95609465e-01 -1.00756675e-01 -2.25950945e-02 -5.51315304e-03
-3.98526043e-02 5.72423160e-01 2.35566825e-01 5.25883317e-01
1.97272986e-01 2.75864452e-01 1.29496455e+00 -2.88438499e-01
-2.99862117e-01 1.66921750e-01 -1.46815747e-01 7.92042494e-01
6.88009679e-01 -1.72008350e-01 -2.82452554e-01 -2.17429549e-01
1.29818499e-01 1.68872386e-01 4.65934306e-01 -1.35175312e+00
3.37351769e-01 -2.31526613e-01 3.69786024e-01 -6.84127808e-01
6.13590479e-01 -1.37597060e+00 9.24955368e-01 5.26007175e-01
1.59868300e-01 -6.26798391e-01 5.70754101e-03 7.64819801e-01
-4.99901325e-01 -8.39518547e-01 1.18579459e+00 -4.62680012e-01
-8.33594322e-01 -1.38736159e-01 -5.00815511e-01 -8.83103907e-01
1.19725299e+00 -4.61997718e-01 -2.89449513e-01 -3.69161427e-01
-6.32140636e-01 -4.05970603e-01 6.87586904e-01 -3.63051772e-01
3.93948197e-01 -8.76962304e-01 -6.26747608e-01 2.38158613e-01
1.34419218e-01 -1.64649680e-01 7.60366917e-01 1.13338077e+00
-7.84516752e-01 4.25217934e-02 -2.95450687e-01 -7.31808126e-01
-1.46973586e+00 4.18474972e-01 3.16743135e-01 4.96304110e-02
-5.82511663e-01 3.02684814e-01 -5.70427217e-02 1.49226144e-01
-2.40254208e-01 -4.95077251e-03 -3.26449633e-01 5.52409768e-01
5.58958352e-01 5.40247023e-01 2.46032313e-01 -7.10760057e-01
-7.83603638e-02 9.68516350e-01 4.81091768e-01 -1.19232304e-01
1.49155557e+00 -2.53634334e-01 -4.88149732e-01 3.98171932e-01
8.42150927e-01 5.57784915e-01 -1.00144064e+00 -2.90194303e-01
-4.59229380e-01 -1.01329589e+00 5.34761608e-01 -5.39460719e-01
-1.08439374e+00 9.21759248e-01 7.01482356e-01 4.48882252e-01
1.73887908e+00 -5.74939430e-01 4.61271197e-01 1.95257053e-01
4.39476579e-01 -1.14167523e+00 -2.83248574e-01 -1.11155575e-02
6.06140137e-01 -1.15686381e+00 3.19901556e-01 -8.36714447e-01
-1.80461794e-01 1.41958296e+00 4.65035476e-02 3.67151737e-01
3.53941143e-01 -6.02725260e-02 2.66568661e-02 1.18551053e-01
7.54227489e-02 -5.70723891e-01 6.68025911e-02 3.05761039e-01
3.12209696e-01 1.57396309e-02 -5.54559112e-01 -1.74413085e-01
3.40335071e-01 -8.47641081e-02 5.85325480e-01 5.03716350e-01
-6.52129650e-01 -1.05482876e+00 -9.99051213e-01 3.24637741e-01
-4.50105041e-01 6.48584729e-03 9.06622559e-02 6.01135731e-01
2.56065447e-02 1.33633518e+00 -2.69111753e-01 -7.54569396e-02
4.15928751e-01 1.32914528e-01 4.33175534e-01 -3.19070995e-01
-3.19161534e-01 6.67588532e-01 -1.17502794e-01 -5.58941066e-01
-1.13635719e+00 -8.24050188e-01 -7.83200204e-01 -2.62036771e-01
-7.35055983e-01 1.83034372e-02 9.14715886e-01 5.87144196e-01
-2.17832863e-01 4.32477981e-01 6.95439935e-01 -8.90873194e-01
-2.81739682e-01 -8.98842931e-01 -1.24050760e+00 3.95304710e-01
2.17794716e-01 -5.46633005e-01 -6.04922712e-01 2.13717848e-01] | [10.182194709777832, -2.2293944358825684] |
1931b48a-0ce0-45bf-8049-19ca54d3b039 | dsamnet-a-deeply-supervised-attention-metric | null | null | https://ieeexplore.ieee.org/document/9555146 | https://ieeexplore.ieee.org/document/9555146 | DSAMNet: A Deeply Supervised Attention Metric Based Network for Change Detection of High-Resolution Images | In view of the insufficient of current change detection, we propose a deeply-supervised attention metric-based network (DSAMNet) for bi-temporal image change detection. The DSAMNet contains a CBAM integrated change decision module to learn a change map directly from features from feature extractor, and an auxiliary deep supervision module to generate intermediate change results to help the training of hidden layers. We also provide a new benchmark-SYSU-CD-with totally 20000 image pairs for the training and testing of deep learning based CD methods. Comparative experiments on the SYSU-CD dataset have proved the effectiveness of the proposed method. | ['Qian Shi', 'Mengxi Liu'] | 2021-10-12 | null | null | null | ieee-international-geoscience-and-remote | ['change-detection', 'change-detection-for-remote-sensing-images'] | ['computer-vision', 'miscellaneous'] | [ 1.26891583e-01 -5.43264985e-01 1.41552195e-01 -5.21236658e-01
-4.36526656e-01 -4.32092696e-03 6.30014956e-01 -3.77900660e-01
-4.74581122e-01 5.31005323e-01 1.13952383e-01 -4.97060567e-02
2.65019476e-01 -9.00308013e-01 -6.65280581e-01 -6.91018999e-01
-2.56904006e-01 -7.95290992e-02 4.60447252e-01 -3.07400465e-01
4.06094939e-02 3.26902121e-01 -1.32745504e+00 1.97094887e-01
7.35688686e-01 1.47344923e+00 3.06059808e-01 7.93656588e-01
2.05422759e-01 1.04161119e+00 -3.93683553e-01 -1.02622341e-03
4.65974510e-01 -6.03680849e-01 -8.02705765e-01 6.09029606e-02
5.00732601e-01 -6.65463567e-01 -8.04276824e-01 1.01997018e+00
9.23272371e-01 2.05102533e-01 4.49120879e-01 -1.06894350e+00
-9.88294840e-01 3.72422814e-01 -5.77337027e-01 1.21578300e+00
2.97624797e-01 5.93054414e-01 8.25576663e-01 -1.01059020e+00
1.02288711e+00 1.28103256e+00 6.83651209e-01 3.69291127e-01
-7.90872097e-01 -5.68237901e-01 5.87956548e-01 9.51341689e-01
-1.26821399e+00 -3.07858407e-01 9.49666440e-01 -4.96259153e-01
1.23799276e+00 5.28138131e-03 7.63862848e-01 9.82702613e-01
3.86518538e-01 9.03687119e-01 8.60731244e-01 -8.33914950e-02
3.47248185e-03 -3.53772044e-01 2.11186260e-01 7.96775579e-01
8.60097259e-02 2.23475099e-01 -7.57482648e-02 4.97414231e-01
7.12768614e-01 1.37771159e-01 -6.37656212e-01 -1.70349166e-01
-1.42029619e+00 5.64176917e-01 1.15692544e+00 4.35508907e-01
-3.54111731e-01 3.06351393e-01 6.73460364e-01 6.91974998e-01
4.92728859e-01 1.94288850e-01 -4.75852013e-01 -1.09485060e-01
-4.31942225e-01 -1.51265055e-01 7.24871010e-02 9.68108416e-01
6.97692990e-01 1.21804319e-01 -4.86791253e-01 6.71295166e-01
2.20457632e-02 4.51596409e-01 1.07361460e+00 -5.16220391e-01
4.56043422e-01 6.64940476e-01 1.29122749e-01 -1.26517749e+00
-2.84648716e-01 -4.29349005e-01 -1.16632533e+00 6.61344156e-02
-3.52994204e-01 -1.07602671e-01 -1.06747973e+00 1.55199516e+00
3.86289865e-01 3.82936835e-01 -1.36584133e-01 1.05966055e+00
7.40548372e-01 8.07085216e-01 -4.17386979e-01 -1.97879493e-01
8.15860152e-01 -1.30956948e+00 -8.01721692e-01 1.33521669e-02
5.94033003e-01 -1.61751091e-01 1.30623519e+00 9.13562477e-02
-8.02559376e-01 -9.60526586e-01 -1.53746033e+00 -2.09961325e-01
-4.60265040e-01 1.48008019e-01 2.66180873e-01 -1.04317799e-01
-1.24434400e+00 4.26016092e-01 -1.07511806e+00 -3.92442703e-01
7.63611197e-01 3.61512691e-01 -2.17613533e-01 -2.72178382e-01
-1.35638022e+00 7.60080457e-01 5.01100242e-01 5.68980217e-01
-1.13519526e+00 -3.46398175e-01 -9.72490311e-01 -2.51844041e-02
1.02025568e-01 -5.13227701e-01 1.19434428e+00 -1.30341721e+00
-1.54853320e+00 8.72023344e-01 1.55265421e-01 -5.88496804e-01
6.38080895e-01 -4.43776026e-02 -7.78785408e-01 1.99336603e-01
1.67181179e-01 8.38093758e-01 8.30481291e-01 -9.04946566e-01
-9.27519739e-01 -1.78969409e-02 3.02733015e-02 2.60041922e-01
-1.84792757e-01 -2.66299069e-01 -6.53302908e-01 -7.90560961e-01
4.46347706e-02 -7.12581396e-01 -2.10646242e-01 -2.28100643e-02
-3.63746524e-01 -1.39716834e-01 1.22914600e+00 -8.26248527e-01
1.49371111e+00 -1.97822189e+00 -2.05418356e-02 -6.83434634e-03
1.93960086e-01 1.76989675e-01 -5.07528186e-01 -1.35603696e-01
-3.20225149e-01 -2.76961803e-01 -4.57690358e-01 -8.41488317e-02
-2.22058088e-01 -2.78632585e-02 -1.67002112e-01 5.63804269e-01
4.42592800e-01 1.07936597e+00 -9.81270969e-01 -3.03288281e-01
3.00640464e-01 -8.37653577e-02 -5.29777586e-01 3.52827311e-01
-2.91447341e-01 2.75971651e-01 -1.65305778e-01 5.48582673e-01
8.34930658e-01 -3.07581186e-01 -1.98698163e-01 -2.22210214e-01
-1.51912533e-02 1.97200239e-01 -9.98309255e-01 1.85090280e+00
-3.71136665e-01 1.00416267e+00 -5.01091242e-01 -9.38533306e-01
7.31395304e-01 -2.73922205e-01 4.74707097e-01 -1.18066597e+00
2.53419459e-01 -1.43926531e-01 1.95359379e-01 -7.35422730e-01
3.28240335e-01 3.90160769e-01 -5.23787253e-02 5.10695577e-01
9.60296951e-04 1.57491460e-01 1.65307447e-01 8.64969864e-02
1.38879263e+00 1.93902493e-01 2.56464601e-01 -2.54906297e-01
8.39000881e-01 -1.35901093e-01 9.71136451e-01 3.43244046e-01
-6.87880039e-01 4.92151409e-01 1.21501632e-01 -9.25428391e-01
-7.90974617e-01 -8.37972403e-01 -1.06329821e-01 7.51554370e-01
6.36314869e-01 -1.11137122e-01 -6.33664787e-01 -1.05176926e+00
-1.89216614e-01 3.17185313e-01 -8.25188756e-01 -5.41639149e-01
-7.79404223e-01 -8.46858501e-01 1.35073453e-01 5.53246498e-01
1.22116280e+00 -1.29970026e+00 -4.90165353e-01 2.78349161e-01
-1.85674548e-01 -1.04239774e+00 -9.27056253e-01 -1.26826465e-01
-5.80690145e-01 -1.21931469e+00 -4.18165624e-01 -1.25341654e+00
7.00252652e-01 3.67474020e-01 7.98332691e-01 -7.65584111e-02
-3.81858349e-01 1.72121912e-01 -2.57129282e-01 -1.84101000e-01
-2.31453434e-01 1.03388451e-01 -4.32212539e-02 1.49937734e-01
4.32714313e-01 -5.67952037e-01 -9.86920416e-01 3.87241423e-01
-8.69035125e-01 6.19723424e-02 5.07949889e-01 8.80933583e-01
5.99306822e-01 9.93552133e-02 7.38733649e-01 -4.46641028e-01
3.37364763e-01 -5.18213987e-01 -7.22425520e-01 1.50916934e-01
-9.52120423e-01 -1.02937222e-01 4.18393880e-01 -3.54846776e-01
-1.14469171e+00 -2.03754872e-01 -8.58685598e-02 -4.29899454e-01
1.63895831e-01 5.15125275e-01 -4.39602405e-01 2.13975549e-01
5.78536630e-01 4.26666647e-01 -3.02125007e-01 -4.69711661e-01
3.32102567e-01 8.36069047e-01 1.04504192e+00 1.44536391e-01
6.58948541e-01 3.77051353e-01 -5.41027665e-01 -2.09256411e-01
-5.80254018e-01 -1.83309793e-01 -9.26094592e-01 -2.75973231e-01
8.36758852e-01 -1.26124394e+00 -9.64531004e-02 1.27855802e+00
-1.17153907e+00 -7.22066760e-01 -1.83331117e-01 3.14355552e-01
-4.99736518e-01 2.30595857e-01 -9.12098885e-01 1.73460603e-01
-6.45705879e-01 -1.11027527e+00 8.02090526e-01 2.35928416e-01
4.55216289e-01 -1.12989795e+00 4.22355980e-01 6.60214946e-03
4.60934639e-01 4.68099415e-01 8.14316213e-01 -2.49455333e-01
-8.36552024e-01 -1.82988778e-01 -2.97363520e-01 4.68790054e-01
6.39108777e-01 -9.49193090e-02 -8.44092131e-01 -6.92791939e-01
-4.20065969e-02 -2.17708558e-01 1.03282022e+00 3.19594204e-01
1.34535313e+00 -3.94871593e-01 -4.52460170e-01 7.96564043e-01
1.72244930e+00 5.14564872e-01 7.79605150e-01 6.51044846e-01
8.85260344e-01 -2.81221956e-01 8.34682167e-01 4.14136589e-01
5.85244596e-01 7.57839561e-01 4.52702165e-01 -2.88556278e-01
-5.88968873e-01 2.86468547e-02 5.99481106e-01 1.25944042e+00
3.24650258e-01 -2.35121533e-01 -7.25344419e-01 6.84502542e-01
-1.74996185e+00 -1.17194998e+00 2.09880117e-02 1.76454425e+00
1.04708636e+00 3.85521621e-01 -1.32971555e-01 3.80578917e-03
9.91640270e-01 4.70529526e-01 -1.08575988e+00 -1.29817009e-01
-1.47939354e-01 -9.46045220e-02 1.77494004e-01 3.88472736e-01
-1.37162685e+00 9.51812148e-01 6.19806576e+00 6.03268266e-01
-1.61163938e+00 3.82231027e-01 7.08555698e-01 -1.85517505e-01
-3.37329246e-02 -5.00047982e-01 -4.97368574e-01 8.31077099e-01
5.85356593e-01 -1.31919399e-01 3.67517859e-01 6.62300944e-01
4.09135997e-01 -1.05143547e-01 -1.09665251e+00 1.15842676e+00
3.08053851e-01 -1.21320939e+00 1.65338188e-01 -4.59468752e-01
1.25971699e+00 5.43221414e-01 -3.74850221e-02 5.58326364e-01
2.54902840e-01 -4.85269934e-01 4.90660578e-01 6.82783961e-01
1.05472052e+00 -5.29663920e-01 9.26561177e-01 -1.45382077e-01
-1.33378386e+00 -2.93806702e-01 -4.46609676e-01 -6.27298728e-02
-1.91443443e-01 6.34863973e-01 -7.04235435e-01 7.15326965e-01
8.27395320e-01 1.66912520e+00 -9.54221129e-01 1.00967956e+00
-8.72228593e-02 5.15266180e-01 -9.15150940e-02 3.37492317e-01
4.03108597e-01 -1.42795742e-01 4.76992697e-01 1.29941511e+00
2.79978752e-01 -1.95136648e-02 -5.93918040e-02 7.29101419e-01
-3.36042762e-01 -1.96642771e-01 -4.13365692e-01 3.34382921e-01
4.34623778e-01 1.19538009e+00 -5.47354579e-01 -4.60625947e-01
-4.35597241e-01 1.63377261e+00 4.46591675e-01 3.58289629e-01
-9.91177201e-01 -8.74193788e-01 8.99251580e-01 -2.33543292e-01
8.47484827e-01 2.46601924e-03 5.99363148e-02 -1.44791555e+00
2.94857800e-01 -8.48183215e-01 5.71409941e-01 -9.54157233e-01
-1.13886738e+00 8.14360976e-01 -3.31588805e-01 -1.67017484e+00
-4.19569202e-02 -2.61772126e-01 -1.14565027e+00 6.53209686e-01
-1.72203469e+00 -9.04228330e-01 -8.98731351e-01 7.46868014e-01
7.77637362e-01 -2.52271980e-01 1.88294262e-01 4.94511902e-01
-1.28513181e+00 6.57602429e-01 3.63986313e-01 4.35917199e-01
7.05513477e-01 -1.22793114e+00 8.57535124e-01 1.44165456e+00
-1.78254217e-01 5.25801070e-02 2.06174135e-01 -5.23615360e-01
-1.02912915e+00 -1.81273293e+00 3.51095706e-01 -2.86444366e-01
5.13752043e-01 -1.35407269e-01 -8.10257137e-01 7.84716845e-01
3.98990124e-01 3.93722832e-01 1.22192033e-01 -4.80338395e-01
-5.09456694e-02 -5.71749210e-01 -9.60330129e-01 3.96955431e-01
1.48921311e+00 -6.07707858e-01 -6.59715354e-01 6.94905296e-02
1.13345480e+00 -5.71302176e-01 -8.72451782e-01 4.30048257e-01
5.32307550e-02 -7.03888059e-01 5.56905925e-01 -4.15045619e-01
4.74975884e-01 -6.01110935e-01 -4.54858206e-02 -1.54902387e+00
-7.20927179e-01 -3.15955460e-01 -8.10397118e-02 1.23892605e+00
-2.95798965e-02 -5.94601393e-01 2.94801116e-01 8.14571381e-02
-5.00722766e-01 -9.64661300e-01 -9.24542964e-01 -6.63836002e-01
-2.07065970e-01 -1.74914241e-01 9.21476126e-01 1.20385730e+00
-3.56208235e-01 5.00077844e-01 -8.59731585e-02 2.12661251e-01
-3.08253802e-03 2.49340832e-01 5.79176486e-01 -7.34596550e-01
-3.47599238e-02 -4.99637008e-01 -9.98139143e-01 -1.17932773e+00
-1.12284720e-03 -8.96124005e-01 1.95727199e-01 -1.55555916e+00
2.72379994e-01 -3.62635732e-01 -8.35208893e-01 4.69754398e-01
-6.03301883e-01 1.14613689e-01 -1.45576209e-01 2.83856392e-01
-8.28269184e-01 1.08789492e+00 1.49615192e+00 -5.77492237e-01
-2.99539983e-01 -1.68311432e-01 -3.42081636e-01 2.78261960e-01
4.89861280e-01 -2.91197002e-01 -5.03547788e-01 -5.49263239e-01
-2.05161422e-01 -3.88626635e-01 1.98697031e-01 -1.27403331e+00
2.76705138e-02 -1.61244199e-02 4.25257683e-01 -8.17368448e-01
-1.89529449e-01 -6.35805368e-01 -1.94005400e-01 8.10321867e-01
-4.09617782e-01 3.66844118e-01 6.44871294e-02 6.40213847e-01
-3.98265392e-01 2.60699928e-01 8.58513594e-01 2.68518984e-01
-1.33200479e+00 8.79370809e-01 -2.23603785e-01 3.64651754e-02
1.20569038e+00 -7.45347217e-02 -4.00131941e-01 -1.87190503e-01
-4.85060483e-01 5.52825809e-01 3.70925367e-01 6.40135646e-01
6.22834086e-01 -2.02862740e+00 -5.82298756e-01 2.76614785e-01
3.46127748e-01 1.87838078e-01 3.57614607e-01 7.54388690e-01
-7.40625739e-01 2.87089676e-01 -5.60194254e-01 -6.10787153e-01
-8.72933626e-01 8.46653342e-01 9.91833270e-01 -1.16734624e-01
-7.60270894e-01 6.91097856e-01 2.10014254e-01 -3.80469233e-01
5.05059473e-02 -8.48091960e-01 -1.79705083e-01 4.76999879e-02
6.27296329e-01 4.67412025e-01 2.19319597e-01 -3.36386621e-01
-3.82809103e-01 3.33761692e-01 -3.42889369e-01 9.71207470e-02
1.33170342e+00 -5.44577718e-01 -2.82398127e-02 3.38259250e-01
1.54946673e+00 -7.31229782e-01 -1.74569273e+00 -6.38671875e-01
-1.36253610e-01 -3.31083834e-01 1.86049446e-01 -9.59398508e-01
-1.35518241e+00 7.42309451e-01 1.32046962e+00 -2.39831001e-01
1.55371451e+00 -4.44940090e-01 1.10158968e+00 7.65411854e-01
1.88922167e-01 -1.22374463e+00 4.19048548e-01 4.38084006e-01
1.21674514e+00 -1.54263759e+00 -2.41771221e-01 2.21721709e-01
-4.59444702e-01 9.17089581e-01 1.18169391e+00 -2.18710124e-01
7.76384473e-01 -5.84305041e-02 2.03798920e-01 -1.27988368e-01
-8.98586333e-01 -2.13907152e-01 1.52704701e-01 5.52723765e-01
-1.54108971e-01 -2.52735078e-01 -9.76778567e-02 4.67585564e-01
1.87815979e-01 1.74556017e-01 6.13332987e-01 9.76936042e-01
-4.10202265e-01 -6.19283140e-01 1.28657028e-01 6.31116569e-01
1.08583704e-01 7.48930648e-02 -3.32671463e-01 7.04648614e-01
3.10288072e-01 5.90914071e-01 5.23832798e-01 -8.10604393e-01
2.97101825e-01 -4.00590718e-01 2.73608983e-01 -3.64454150e-01
-4.14862096e-01 -3.21829796e-01 -1.61950693e-01 -7.76319623e-01
-4.80304867e-01 -8.56678009e-01 -1.22284770e+00 -1.88510820e-01
-1.09887555e-01 -2.52005219e-01 4.77921367e-02 9.34792101e-01
6.14428759e-01 7.93490171e-01 1.46807384e+00 -7.89721966e-01
-3.29783559e-01 -1.37953043e+00 -3.48858416e-01 5.13158202e-01
6.29766345e-01 -5.23687243e-01 -8.10564756e-02 2.90718913e-01] | [9.64319133758545, -1.1186565160751343] |
b2a3cc6a-8328-458b-94e3-7a9fe38085f6 | is-someone-speaking-exploring-long-term | 2107.06592 | null | https://arxiv.org/abs/2107.06592v2 | https://arxiv.org/pdf/2107.06592v2.pdf | Is Someone Speaking? Exploring Long-term Temporal Features for Audio-visual Active Speaker Detection | Active speaker detection (ASD) seeks to detect who is speaking in a visual scene of one or more speakers. The successful ASD depends on accurate interpretation of short-term and long-term audio and visual information, as well as audio-visual interaction. Unlike the prior work where systems make decision instantaneously using short-term features, we propose a novel framework, named TalkNet, that makes decision by taking both short-term and long-term features into consideration. TalkNet consists of audio and visual temporal encoders for feature representation, audio-visual cross-attention mechanism for inter-modality interaction, and a self-attention mechanism to capture long-term speaking evidence. The experiments demonstrate that TalkNet achieves 3.5% and 2.2% improvement over the state-of-the-art systems on the AVA-ActiveSpeaker dataset and Columbia ASD dataset, respectively. Code has been made available at: https://github.com/TaoRuijie/TalkNet_ASD. | ['Haizhou Li', 'Mike Zheng Shou', 'Xinyuan Qian', 'Rohan Kumar Das', 'Zexu Pan', 'Ruijie Tao'] | 2021-07-14 | null | null | null | null | ['audio-visual-active-speaker-detection'] | ['computer-vision'] | [-2.04140827e-01 -9.51256081e-02 1.05305739e-01 -6.75043046e-01
-1.23605275e+00 -5.94616055e-01 7.06108391e-01 -1.32557884e-01
-1.89571828e-01 2.46576980e-01 6.65418625e-01 -2.10639253e-01
2.61887193e-01 -2.27078304e-01 -5.07397234e-01 -6.46951795e-01
-1.14339896e-01 5.57515472e-02 2.58217782e-01 2.84614153e-02
2.47355819e-01 2.70660728e-01 -1.71007121e+00 5.59831440e-01
4.23580736e-01 1.03054070e+00 2.51937151e-01 1.12251639e+00
-2.99983799e-01 1.02781296e+00 -5.97880185e-01 -2.32346207e-01
-2.17708617e-01 -3.91103715e-01 -7.43102133e-01 9.99781713e-02
5.03576875e-01 -4.02608991e-01 -4.53074843e-01 7.36623108e-01
1.06365097e+00 1.46751046e-01 1.45941094e-01 -1.49359012e+00
-8.17233622e-01 6.90763772e-01 -4.61086839e-01 6.18805945e-01
7.02660263e-01 6.06498420e-01 1.09154868e+00 -1.39927804e+00
1.10826693e-01 1.54701185e+00 2.42654249e-01 5.55664182e-01
-7.97563732e-01 -9.54744041e-01 4.08715069e-01 7.27266192e-01
-1.39745283e+00 -1.45189834e+00 8.38095188e-01 -5.30473053e-01
9.96768832e-01 3.46625477e-01 6.92010522e-01 1.26412761e+00
-9.99649018e-02 1.18187225e+00 7.19143927e-01 -5.73101997e-01
-4.54806350e-02 1.23673394e-01 4.08295780e-01 4.70875353e-01
-5.90574324e-01 1.85584158e-01 -1.21485758e+00 -6.16000630e-02
4.02371854e-01 -2.97873944e-01 -3.53238791e-01 8.52216780e-02
-1.13743722e+00 5.70207775e-01 3.01984042e-01 3.26109618e-01
-3.92484039e-01 9.06243399e-02 3.71023148e-01 2.19293579e-01
5.66604197e-01 -4.53106575e-02 5.15456200e-02 -5.00940025e-01
-8.51849675e-01 -3.27065080e-01 6.09115124e-01 6.03639543e-01
3.28691751e-01 2.06197053e-01 -4.71823096e-01 9.34616029e-01
7.15976655e-01 5.39885104e-01 4.13687259e-01 -8.82568479e-01
3.04297119e-01 2.80562639e-01 -2.15689875e-02 -5.52932084e-01
3.85945067e-02 -2.11400002e-01 -2.28611961e-01 3.66773129e-01
3.43774743e-02 3.94396596e-02 -8.44412029e-01 1.67994130e+00
4.58807498e-01 3.97020131e-01 -3.52497003e-03 9.08443093e-01
1.39995980e+00 7.38002419e-01 -4.29819077e-02 -3.87641698e-01
1.36781740e+00 -1.24671555e+00 -1.14655209e+00 -2.46399611e-01
-2.83143997e-01 -1.05667686e+00 1.10818255e+00 2.37057447e-01
-1.37985575e+00 -8.36831450e-01 -8.78643513e-01 -3.16559136e-01
-4.53806520e-02 7.80302286e-02 3.30379367e-01 2.41598636e-01
-1.38827837e+00 -3.46211255e-01 -8.46392453e-01 -2.24966183e-01
2.73555130e-01 2.16053262e-01 -2.61319488e-01 3.20225775e-01
-1.19659102e+00 4.25680429e-01 -3.80267680e-01 4.86841559e-01
-1.59258103e+00 -3.86091948e-01 -8.10559154e-01 8.78079608e-02
2.97142833e-01 -5.16975701e-01 1.89451134e+00 -1.13530958e+00
-1.76741338e+00 8.35954666e-01 -8.74416709e-01 -1.55282810e-01
4.47147429e-01 -2.73011982e-01 -7.11048782e-01 3.11524451e-01
-3.49101834e-02 5.37313044e-01 1.03924787e+00 -1.05250108e+00
-4.88826126e-01 -4.60956722e-01 2.24422678e-01 5.23826361e-01
-2.41844952e-01 5.43869376e-01 -8.71303141e-01 -3.99723262e-01
7.99682736e-02 -5.95668972e-01 5.15956581e-01 2.69917428e-01
-5.45833766e-01 -4.54879403e-01 1.05230343e+00 -7.90859938e-01
1.08965087e+00 -2.69946933e+00 -9.82890055e-02 -3.25317740e-01
3.73414248e-01 2.65711635e-01 -1.37725160e-01 3.45274717e-01
-1.53457463e-01 -1.96439460e-01 2.86696106e-01 -8.13824654e-01
-2.18318179e-02 -2.56985575e-01 -2.71984160e-01 4.18365628e-01
3.60803083e-02 6.96776986e-01 -7.47040689e-01 -5.82645595e-01
2.31306404e-01 8.55288208e-01 -3.38653997e-02 5.24884760e-01
2.57626951e-01 5.15007198e-01 -2.31793776e-01 7.42756605e-01
5.76168776e-01 -1.69885099e-01 -2.84830511e-01 6.70762360e-02
-4.35648948e-01 6.34815097e-01 -1.00940752e+00 1.66302812e+00
-4.87561494e-01 1.21614921e+00 5.72308362e-01 -5.82166076e-01
7.06801772e-01 9.59613025e-01 1.53169796e-01 -8.17976415e-01
1.57842655e-02 -5.86693734e-03 -3.24477367e-02 -7.64091432e-01
1.74187377e-01 1.24156341e-01 2.80640095e-01 2.87905127e-01
1.37774602e-01 4.57824856e-01 -1.10463358e-01 4.85883057e-01
8.80869746e-01 -3.54034275e-01 4.80239354e-02 1.15123428e-01
8.47225308e-01 -5.93912482e-01 4.61197942e-01 5.70561469e-01
-7.16400087e-01 6.27220511e-01 1.90079093e-01 -1.44843496e-02
-4.05335575e-01 -1.29721510e+00 1.02162257e-01 1.58141518e+00
1.58623219e-01 -4.80686396e-01 -4.97456968e-01 -3.65746200e-01
-1.29065529e-01 6.73824966e-01 -6.70537710e-01 -1.37307286e-01
-3.33445430e-01 7.77549669e-02 3.96392345e-01 6.41700387e-01
5.63845098e-01 -1.33112121e+00 -1.53493479e-01 9.68528464e-02
-4.50342119e-01 -9.67299283e-01 -9.45282578e-01 -2.11865947e-01
-1.86509877e-01 -7.98025787e-01 -7.78276742e-01 -7.75077879e-01
2.86378354e-01 4.83779311e-01 1.02066636e+00 -2.48529062e-01
-3.08642805e-01 6.84075654e-01 -2.08517596e-01 -5.49459994e-01
-2.74784952e-01 -4.74101692e-01 8.88006855e-03 3.73457700e-01
3.99025112e-01 -4.83339101e-01 -8.71344686e-01 3.78426969e-01
-1.02914132e-01 5.83029017e-02 4.22895551e-01 7.27093577e-01
3.04521292e-01 -5.01439810e-01 4.81939584e-01 -2.43717968e-01
4.23525751e-01 -1.42010257e-01 -3.48736823e-01 2.83895195e-01
-1.71537086e-01 -4.47322071e-01 -1.50007643e-02 -5.31830668e-01
-1.36377704e+00 6.44217283e-02 -2.74943084e-01 -7.08643794e-01
-9.08041671e-02 1.70008868e-01 -4.07164246e-01 2.66971916e-01
4.85763699e-01 3.76361549e-01 4.75189909e-02 -5.01012325e-01
2.46167734e-01 1.24345589e+00 6.83874786e-01 -8.06374028e-02
2.80609280e-01 5.52443266e-01 -8.00959766e-01 -9.41156805e-01
-7.66840219e-01 -9.22336459e-01 -4.79111314e-01 -6.62689209e-01
9.35722709e-01 -1.32254255e+00 -1.00898600e+00 6.89576149e-01
-1.17343056e+00 -2.77558595e-01 -8.84562731e-02 6.46195948e-01
-4.15709943e-01 1.69761360e-01 -5.76585412e-01 -1.30598593e+00
-5.35817981e-01 -1.32056618e+00 1.09472728e+00 1.82158023e-01
-2.16947421e-01 -7.04241335e-01 1.04478911e-01 6.63992524e-01
3.94180447e-01 -3.40160817e-01 1.86053347e-02 -4.39460754e-01
-6.06423795e-01 9.77902114e-02 3.36317047e-02 3.45045060e-01
1.83802083e-01 3.56098682e-01 -1.78526711e+00 -4.04626459e-01
-1.14038840e-01 -1.90982565e-01 9.12968695e-01 7.19879448e-01
1.02340209e+00 -3.06350678e-01 -2.08029106e-01 3.96320522e-01
7.00997233e-01 5.63649237e-01 3.60663414e-01 -2.25780815e-01
5.96634626e-01 4.97655034e-01 4.08287555e-01 4.59459543e-01
5.37906587e-01 9.47048306e-01 4.82977599e-01 -2.41620660e-01
-6.18193626e-01 -1.48933500e-01 7.84317017e-01 9.11498785e-01
1.53977245e-01 -4.02855784e-01 -1.06061804e+00 7.89475203e-01
-1.82404625e+00 -1.15683889e+00 -4.45849784e-02 2.13808012e+00
6.14289284e-01 1.15119696e-01 3.36688787e-01 2.83220321e-01
1.13475430e+00 1.93706840e-01 -7.28576601e-01 -3.70697349e-01
-9.11134407e-02 -3.63399982e-01 -3.19807708e-01 9.39245820e-01
-1.00267708e+00 6.45034432e-01 6.07195234e+00 5.58344960e-01
-1.55021131e+00 3.96976739e-01 5.33307612e-01 -4.86601263e-01
-7.98993334e-02 -2.11650923e-01 -6.77693784e-01 5.01612008e-01
1.11102676e+00 -2.78125197e-01 3.02579612e-01 7.05685139e-01
5.41606903e-01 4.43603061e-02 -1.23828030e+00 1.35974622e+00
3.58379185e-01 -9.95723546e-01 -3.60914499e-01 -3.03492136e-02
6.44992143e-02 1.65864959e-01 2.66019255e-01 2.34062836e-01
1.21086568e-01 -8.94193351e-01 1.28304136e+00 5.66597342e-01
8.42385471e-01 -5.63769162e-01 3.36568207e-01 1.15263619e-01
-1.55982304e+00 -1.68355122e-01 3.49820286e-01 1.00208431e-01
2.45815098e-01 2.69496977e-01 -9.91870642e-01 1.36112243e-01
1.08323860e+00 7.09421933e-01 -3.86983246e-01 1.09038460e+00
-3.81746680e-01 9.94248033e-01 -1.15511052e-01 2.75432110e-01
-1.25943780e-01 5.15710235e-01 9.78949785e-01 1.12732768e+00
1.91779315e-01 -6.76820800e-02 -1.64574676e-03 5.84110498e-01
3.67773995e-02 -6.81584254e-02 -4.03681695e-01 -1.18388228e-01
8.39777946e-01 1.13913429e+00 -1.85282812e-01 -3.64255875e-01
-5.49325109e-01 1.04892850e+00 1.07768565e-01 5.31084597e-01
-8.32965136e-01 -3.71774256e-01 6.70769572e-01 5.34853525e-02
3.32558036e-01 -5.52663766e-02 -2.73709986e-02 -9.20872509e-01
2.34671548e-01 -6.39034688e-01 6.07605696e-01 -1.17062819e+00
-1.09376347e+00 7.23962903e-01 -3.32360029e-01 -1.27096784e+00
-3.91694665e-01 -2.51438975e-01 -1.00526655e+00 1.01066673e+00
-1.32036734e+00 -1.27591288e+00 -5.32073855e-01 7.92218983e-01
1.01755071e+00 -2.64954418e-01 7.92186677e-01 3.23330253e-01
-6.41256213e-01 7.91746855e-01 -3.55038606e-02 3.39284062e-01
9.96140957e-01 -1.11751747e+00 4.12279189e-01 7.70393431e-01
2.69405395e-01 4.63769168e-01 7.08640993e-01 -7.86751434e-02
-1.35939777e+00 -6.53277099e-01 1.15280437e+00 -3.67750347e-01
5.64603686e-01 -7.80385971e-01 -8.08258653e-01 8.77981067e-01
7.48630881e-01 2.46425077e-01 8.22562516e-01 3.20220381e-01
-4.63538677e-01 -3.24053854e-01 -8.84537697e-01 2.37596840e-01
9.08309042e-01 -1.05583251e+00 -4.94034231e-01 3.13014872e-02
6.51383042e-01 -3.38361472e-01 -3.37401688e-01 3.91312093e-02
7.86244869e-01 -1.07475138e+00 9.93017614e-01 -2.12677330e-01
8.21301565e-02 -2.65100896e-01 -9.14704427e-02 -1.09098196e+00
-4.48531538e-01 -7.87552238e-01 -3.72387439e-01 1.65151751e+00
4.10653442e-01 -5.05250335e-01 1.92294389e-01 1.54951200e-01
-4.48228866e-01 -4.43128943e-01 -1.20489788e+00 -5.85392356e-01
-4.26665366e-01 -6.28416359e-01 4.44222391e-01 8.83019209e-01
1.83913037e-01 6.88073635e-01 -3.31638932e-01 3.16040844e-01
4.08577621e-01 1.17007524e-01 6.64829314e-01 -9.92965996e-01
-3.95921052e-01 -4.47731048e-01 -4.93693709e-01 -1.14661276e+00
2.18838722e-01 -4.21543986e-01 2.71314949e-01 -1.61973739e+00
3.33452404e-01 1.81877688e-01 -6.02335691e-01 6.57378316e-01
-1.17501326e-01 2.21779257e-01 4.05698009e-02 2.67910182e-01
-8.43757212e-01 7.41074562e-01 1.23938429e+00 -3.42323929e-01
-2.72996813e-01 2.36446559e-02 -6.22455955e-01 5.54661155e-01
5.98173141e-01 -7.68266525e-03 -5.26614785e-01 -5.76733112e-01
-2.69068301e-01 5.26580513e-01 5.39655983e-01 -8.08376431e-01
6.15569532e-01 1.99750680e-02 3.21466535e-01 -8.09784770e-01
9.81629431e-01 -3.47905725e-01 -1.56695917e-01 3.58424038e-02
-6.34631455e-01 3.15090716e-02 3.00003767e-01 4.51587558e-01
-6.46757901e-01 3.57426107e-01 6.91467166e-01 2.60553546e-02
-8.38571072e-01 2.87044138e-01 -5.47163904e-01 -1.60834491e-01
8.99194956e-01 -1.22509465e-01 -5.02273321e-01 -8.93956363e-01
-1.20097351e+00 4.00586277e-01 -2.01227248e-01 7.46235549e-01
7.69173563e-01 -1.34737742e+00 -8.89183581e-01 2.95967162e-01
2.85983086e-01 -3.38845402e-01 7.72074878e-01 9.35079455e-01
-3.14208642e-02 4.38707709e-01 4.34404686e-02 -1.01072955e+00
-1.99667656e+00 2.68645048e-01 4.29985702e-01 6.05257213e-01
-5.62951028e-01 1.54454184e+00 6.29503906e-01 1.19349562e-01
7.32561648e-01 4.58903052e-02 -3.13641101e-01 1.92714006e-01
9.44220781e-01 2.21599579e-01 3.23824547e-02 -1.07719696e+00
-7.16705382e-01 2.58499444e-01 -1.61681801e-01 -5.77698469e-01
1.02682137e+00 -6.77862227e-01 2.26562902e-01 1.02183187e+00
1.12619460e+00 2.65711337e-01 -1.36554039e+00 -5.15419602e-01
-5.41082680e-01 -5.54800212e-01 5.07787287e-01 -1.03318059e+00
-1.09220862e+00 1.41421890e+00 9.45687294e-01 2.26286680e-01
1.15485632e+00 4.87141520e-01 3.21255147e-01 -4.41417508e-02
-1.76321879e-01 -7.48916745e-01 4.59565908e-01 3.37883741e-01
1.41390193e+00 -1.43152153e+00 -4.48914319e-01 -3.67062867e-01
-8.50360811e-01 9.14037228e-01 6.42665088e-01 5.25213242e-01
7.89803386e-01 2.58267254e-01 7.21545935e-01 -1.39325231e-01
-1.25815535e+00 -3.28805834e-01 4.64809984e-01 3.99742782e-01
7.38038898e-01 6.86230063e-02 4.72869128e-01 5.76330483e-01
-2.68805295e-01 -2.72491425e-01 1.18047057e-03 8.41945767e-01
-2.98899949e-01 -6.83299065e-01 -4.50699657e-01 -2.30464563e-01
-3.42401922e-01 -1.15974464e-01 -6.37553215e-01 3.49304199e-01
-5.90960830e-02 1.54791057e+00 5.06484151e-01 -3.72212023e-01
2.01730743e-01 2.49521196e-01 2.54921615e-01 -5.34499109e-01
-4.65972751e-01 5.21514773e-01 2.16216415e-01 -5.76038778e-01
-4.24455166e-01 -8.50389719e-01 -1.13311934e+00 -1.62363991e-01
-3.42268646e-01 2.21755430e-01 6.98020875e-01 6.22020841e-01
5.34621179e-01 7.66697705e-01 9.24270868e-01 -6.09937966e-01
-2.45121464e-01 -1.28334284e+00 -4.24498767e-01 2.32454926e-01
9.77322102e-01 -6.80582225e-01 -6.27697110e-01 1.86588258e-01] | [14.432462692260742, 5.13247013092041] |
2888ea73-2955-4264-ac26-30df8a6a8b1c | iov-scenario-implementation-of-a-bandwidth | 2202.03488 | null | https://arxiv.org/abs/2202.03488v1 | https://arxiv.org/pdf/2202.03488v1.pdf | IoV Scenario: Implementation of a Bandwidth Aware Algorithm in Wireless Network Communication Mode | The wireless network communication mode represented by the Internet of vehicles (IoV) has been widely used. However, due to the limitations of traditional network architecture, resource scheduling in wireless network environment is still facing great challenges. This paper focuses on the allocation of bandwidth resources in the virtual network environment. This paper proposes a bandwidth aware multi domain virtual network embedding algorithm (BA-VNE). The algorithm is mainly aimed at the problem that users need a lot of bandwidth in wireless communication mode, and solves the problem of bandwidth resource allocation from the perspective of virtual network embedding (VNE). In order to improve the performance of the algorithm, we introduce particle swarm optimization (PSO) algorithm to optimize the performance of the algorithm. In order to verify the effectiveness of the algorithm, we have carried out simulation experiments from link bandwidth, mapping cost and virtual network request (VNR) acceptance rate. The final results show that the proposed algorithm is better than other representative algorithms in the above indicators. | ['Mohsen Guizani', 'Neeraj Kumar', 'Gagangeet Singh Aujla', 'Chao Wang', 'Peiying Zhang'] | 2022-02-03 | null | null | null | null | ['network-embedding'] | ['methodology'] | [-2.81937480e-01 -2.27221712e-01 -4.85360801e-01 1.56211793e-01
5.24561703e-01 -9.20273960e-02 1.22464746e-01 -3.35860193e-01
-4.85864580e-01 1.06631839e+00 -3.09392303e-01 -5.32105029e-01
-6.77910089e-01 -1.21934462e+00 4.96810935e-02 -5.85935354e-01
-1.96734145e-01 4.89930600e-01 4.22754616e-01 -3.28012228e-01
4.06100273e-01 6.35064006e-01 -1.33380222e+00 -6.63937509e-01
7.13829517e-01 8.83361578e-01 6.56298697e-01 2.27348179e-01
-6.62525237e-01 1.56199574e-01 -7.44719386e-01 -3.22205275e-02
2.03173846e-01 -7.91591033e-02 -7.47093439e-01 4.43186983e-02
-8.21670890e-01 -3.53474289e-01 -3.14195216e-01 9.17548001e-01
6.74048305e-01 4.12213951e-01 3.70705426e-01 -1.99030805e+00
1.73374228e-02 4.26829666e-01 -4.78973210e-01 6.69565976e-01
-4.86752540e-02 -3.36084157e-01 7.33153701e-01 -6.15731359e-01
8.02563727e-01 1.16379297e+00 7.14136362e-02 4.39990997e-01
-7.28931785e-01 -8.12294364e-01 1.24268651e-01 7.43754506e-01
-1.40443897e+00 -2.17182577e-01 8.52267087e-01 -3.64477426e-04
8.65845084e-01 3.41147602e-01 6.78604841e-01 3.62904549e-01
7.81849250e-02 2.49116361e-01 3.57699662e-01 -3.61942828e-01
2.75753796e-01 6.06319487e-01 -2.15224758e-01 2.58129776e-01
5.24158955e-01 1.87485758e-02 4.22128029e-02 -1.66654781e-01
7.53107548e-01 -2.23428741e-01 -3.62915814e-01 -5.64978957e-01
-7.02646554e-01 8.89123380e-01 2.97111630e-01 3.01214457e-01
-4.67681110e-01 1.41663134e-01 6.21500969e-01 3.16364020e-01
2.78645724e-01 -1.21801928e-01 -3.69535267e-01 -3.45240593e-01
-5.07196903e-01 -2.49934897e-01 7.75998771e-01 8.84448886e-01
4.15838599e-01 1.16660900e-01 4.71007913e-01 9.39249039e-01
5.63148320e-01 4.37839389e-01 2.24396363e-01 -8.44257414e-01
3.57706398e-01 3.15533489e-01 4.18906696e-02 -1.24457228e+00
-5.31916440e-01 -3.17259103e-01 -7.23203123e-01 1.75294787e-01
-2.61396706e-01 -5.12128294e-01 -1.47160277e-01 1.50866711e+00
6.37691081e-01 3.57504189e-01 1.70197532e-01 7.95574129e-01
7.84570158e-01 1.08376336e+00 1.00706138e-01 -8.64850104e-01
8.26667249e-01 -1.02529287e+00 -1.19870317e+00 3.11309069e-01
4.20385331e-01 -8.91595840e-01 3.73339593e-01 -8.04045647e-02
-8.73237610e-01 -3.72269362e-01 -9.62745845e-01 7.69602299e-01
-4.11163181e-01 -3.98083538e-01 6.43559337e-01 9.73186672e-01
-1.01541555e+00 3.91183756e-02 -2.22177878e-01 -7.11825132e-01
1.21846661e-01 5.14386356e-01 -1.81820795e-01 2.96007004e-02
-1.38594627e+00 7.36723483e-01 6.96324348e-01 -1.38953611e-01
-3.46539944e-01 -6.03136241e-01 -4.90967631e-01 2.81617492e-01
5.34662724e-01 -4.47876751e-01 6.12040460e-01 -8.72132063e-01
-1.51625788e+00 -1.18139692e-01 4.23018634e-03 -8.40982795e-02
4.57404315e-01 5.23922980e-01 -9.17329490e-01 1.69861883e-01
-1.69613034e-01 2.00162351e-01 1.47054657e-01 -1.16478431e+00
-8.98982584e-01 1.66068360e-01 2.47105628e-01 1.96909085e-01
-6.76372170e-01 2.38654822e-01 -6.08747602e-01 -1.99257597e-01
-1.15642667e-01 -7.15343773e-01 -1.48929790e-01 2.34548282e-02
1.20944772e-02 -4.46524739e-01 1.21741307e+00 -1.69107184e-01
1.61102724e+00 -2.07223940e+00 2.10040390e-01 7.51205802e-01
5.08001596e-02 4.59492683e-01 -2.33008996e-01 5.24626195e-01
1.68881625e-01 2.99842864e-01 4.65543151e-01 2.59955317e-01
-3.39369513e-02 4.28842306e-01 3.30068797e-01 2.19207585e-01
-5.62498689e-01 2.23479390e-01 -6.21982574e-01 -7.20696628e-01
4.10109550e-01 4.38460559e-01 -4.84350890e-01 9.04617738e-03
8.76879841e-02 -1.37228388e-02 -6.66831970e-01 4.28949445e-01
1.20124221e+00 -2.01856792e-01 3.97422403e-01 -3.00346464e-01
-3.46432745e-01 -5.73969185e-01 -1.60717332e+00 9.69898582e-01
-5.86252630e-01 5.88744760e-01 3.49237233e-01 -1.28478682e+00
7.23697841e-01 4.73544776e-01 9.26734209e-01 -8.80716324e-01
4.20654535e-01 2.66301185e-01 2.53962576e-02 -8.42514753e-01
3.34122658e-01 2.23901376e-01 5.34400344e-01 2.95572042e-01
-2.12362111e-01 5.35023332e-01 3.05715919e-01 2.02342138e-01
6.51593089e-01 -4.09324616e-01 2.80644625e-01 -1.11359529e-01
8.04850399e-01 -3.69726941e-02 7.05129445e-01 1.55577481e-01
-6.50429130e-01 -5.45949817e-01 4.73947853e-01 -2.06211895e-01
-1.13508332e+00 -7.16473639e-01 -1.81272745e-01 7.38864303e-01
1.02069771e+00 -4.78464626e-02 -3.88307989e-01 -2.63377488e-01
6.16953475e-03 6.18418217e-01 -6.63969740e-02 3.86622548e-02
-2.54077077e-01 -6.58361256e-01 1.49163818e-02 -1.99711546e-01
6.44040704e-01 -8.60578895e-01 -3.73331696e-01 6.60713136e-01
-1.78430930e-01 -1.20164597e+00 -3.09837669e-01 -5.64446568e-01
-5.84020019e-01 -9.84487057e-01 -4.78697628e-01 -8.81444275e-01
4.21937257e-01 7.50376225e-01 5.21673083e-01 5.70539713e-01
-3.39898378e-01 3.52731466e-01 -5.07532001e-01 -2.43679628e-01
-2.06154913e-01 1.60100967e-01 1.50099531e-01 -1.08995482e-01
1.32152736e-01 -5.72660148e-01 -4.83611256e-01 7.02920318e-01
-6.65494382e-01 5.24041913e-02 3.51200610e-01 6.01689219e-01
3.68659675e-01 9.35469866e-01 1.08970737e+00 -5.20779669e-01
7.86317050e-01 -9.19465780e-01 -8.72048736e-01 3.15707117e-01
-7.42572844e-01 -3.26204002e-01 4.17920113e-01 -4.21485603e-01
-8.67106199e-01 -7.37802684e-01 -7.17666596e-02 -1.46710882e-02
2.87792116e-01 5.30602336e-01 -4.87404108e-01 -5.13366759e-01
-4.33331653e-02 6.25218153e-02 3.50268543e-01 -2.50483871e-01
-6.93081766e-02 9.87799287e-01 -3.14895123e-01 -1.09755427e-01
6.99078619e-01 2.89350986e-01 3.55442524e-01 -9.11935806e-01
1.16143391e-01 -2.78895408e-01 1.32812396e-01 -7.50433564e-01
5.53395391e-01 -5.20547926e-01 -1.25645959e+00 1.05103822e-02
-1.02760601e+00 -3.50464322e-02 4.70640689e-01 7.65452743e-01
-2.82070130e-01 5.66165864e-01 -1.25516817e-01 -9.06016290e-01
-2.72994041e-01 -1.18141663e+00 -1.79464623e-01 5.90302229e-01
4.50974673e-01 -1.20400643e+00 -3.08030784e-01 1.85615525e-01
9.60430086e-01 2.47478962e-01 8.86690915e-01 -2.63630122e-01
-7.10191071e-01 6.73642233e-02 -6.63324237e-01 2.30307821e-02
2.41700038e-02 3.38309318e-01 -2.58041397e-02 -4.65855569e-01
-3.28863204e-01 3.00910771e-01 5.09931110e-02 4.17662948e-01
1.22013927e+00 -1.09104641e-01 -6.29451573e-01 6.19902194e-01
2.13055229e+00 6.15660906e-01 6.35364413e-01 8.08937371e-01
2.02796519e-01 6.07436657e-01 1.05380428e+00 8.59405160e-01
3.72170806e-01 8.10875416e-01 1.07394135e+00 -7.15229213e-02
1.29596086e-03 2.13507622e-01 -2.22658962e-02 8.03679466e-01
-2.96353605e-02 -8.99336994e-01 -5.76931298e-01 4.30014789e-01
-1.78418696e+00 -1.09132457e+00 -4.37568501e-02 1.81752682e+00
-1.43313199e-01 2.63099402e-01 8.01620707e-02 2.90401936e-01
9.96567786e-01 -4.55625616e-02 -3.42937827e-01 -6.87406898e-01
-4.51938296e-03 -2.81008780e-01 8.28939557e-01 4.74110484e-01
-4.82489079e-01 7.00980127e-01 5.68922377e+00 1.25075102e+00
-1.08631754e+00 1.84103459e-01 8.97649582e-03 -3.75328636e-05
-4.10322785e-01 -9.34892148e-02 -4.32525814e-01 9.03282642e-01
8.42980444e-01 -6.13265455e-01 8.84297669e-01 7.36043394e-01
7.64388978e-01 -2.21869886e-01 -2.01470852e-01 1.17650950e+00
-2.45484427e-01 -1.48408210e+00 1.71521142e-01 3.94186080e-01
3.74071330e-01 -4.95432280e-02 -1.78771302e-01 -3.86072062e-02
-3.14071566e-01 -6.98455334e-01 1.73277080e-01 2.51091570e-01
7.74553716e-01 -1.22754061e+00 1.03619838e+00 2.66972274e-01
-1.34662664e+00 -3.56200427e-01 -6.29659116e-01 1.97638929e-01
6.20054841e-01 3.16152275e-01 -4.16441560e-01 8.37765038e-01
3.04013222e-01 1.50171235e-01 2.42030367e-01 1.46491253e+00
4.65439528e-01 5.27496152e-02 -3.13199431e-01 -4.21347946e-01
9.44755599e-02 -3.56003493e-01 5.38846850e-01 6.31157041e-01
4.47736323e-01 3.16970438e-01 2.09216818e-01 3.79442066e-01
-2.36932444e-03 6.83171928e-01 -3.06153089e-01 1.49162695e-01
1.20771217e+00 1.16447854e+00 -6.92370534e-01 -1.40680104e-01
-7.05374539e-01 5.03934622e-01 -6.81973696e-02 5.12008309e-01
-1.22393548e+00 -9.41936195e-01 9.53210235e-01 -9.80353877e-02
2.57092953e-01 -2.14539111e-01 8.10432807e-02 -4.99667853e-01
-2.73797512e-01 -2.50284255e-01 7.46367648e-02 -5.60330927e-01
-6.32365346e-01 4.75639075e-01 -1.11066855e-01 -1.14829767e+00
4.30027902e-01 -4.09793109e-01 -6.27076864e-01 7.60962844e-01
-1.89700317e+00 -4.68874663e-01 -4.91300106e-01 5.97783625e-01
5.59994996e-01 -6.37090445e-01 6.22178555e-01 9.31008399e-01
-8.91832232e-01 6.76856637e-01 4.36613828e-01 -3.63335729e-01
3.01723182e-01 -2.52105594e-01 -4.03223217e-01 6.57440484e-01
-6.01641774e-01 8.49686265e-02 8.23701203e-01 -3.77164900e-01
-1.34777892e+00 -7.09174097e-01 6.74176753e-01 6.51205420e-01
4.58776146e-01 1.88491449e-01 -3.72519165e-01 7.74595365e-02
1.51338845e-01 -1.11094769e-02 7.78014004e-01 -4.04585034e-01
5.85897326e-01 -2.83768743e-01 -1.62620819e+00 4.62979913e-01
8.49928796e-01 2.17582047e-01 3.15241814e-01 3.28893960e-01
8.40763152e-01 -7.79762045e-02 -8.54251146e-01 2.47944325e-01
4.53728795e-01 -5.15869617e-01 1.05421126e+00 -3.70952249e-01
-2.90405154e-01 -4.12289679e-01 -2.76608855e-01 -1.18484461e+00
-3.88898700e-01 -2.77068287e-01 1.20288335e-01 1.22375834e+00
2.63684541e-01 -1.03927338e+00 6.83062017e-01 2.05249086e-01
3.39608461e-01 -6.47606909e-01 -1.13825476e+00 -7.75976241e-01
-5.59712827e-01 -1.29053771e-01 9.96208429e-01 7.95477509e-01
-4.85208165e-03 2.12810889e-01 -3.49865615e-01 3.06954592e-01
5.98425150e-01 -2.45940655e-01 7.68825054e-01 -1.34065628e+00
1.11036740e-01 -5.09054184e-01 -5.40430963e-01 -7.19074905e-01
3.28466713e-01 -5.80500424e-01 -6.20762289e-01 -1.88882065e+00
-7.42969150e-03 -9.89732504e-01 -3.68825287e-01 -2.80050993e-01
2.68038124e-01 -1.37714833e-01 1.96339071e-01 9.94673520e-02
-4.93645132e-01 4.97237891e-01 1.27994549e+00 1.87790155e-01
-1.82999760e-01 6.54407963e-02 -2.03914955e-01 2.55533546e-01
1.29806924e+00 -3.11987936e-01 -1.08275259e+00 -4.53663617e-01
-4.39980961e-02 6.73033416e-01 -3.26353461e-02 -6.83505595e-01
3.27966094e-01 -6.50597930e-01 -2.11571783e-01 -4.34248924e-01
2.77197361e-01 -1.41787100e+00 5.63998222e-01 6.87568426e-01
1.62782848e-01 1.84610844e-01 4.41962667e-02 6.89734936e-01
-6.14977367e-02 -2.67645806e-01 9.04296041e-01 3.49938244e-01
-1.19882739e+00 5.46130955e-01 -5.71801603e-01 -1.65169999e-01
1.48374152e+00 -6.08384311e-01 -3.77854019e-01 -1.94593906e-01
-5.68067133e-01 5.83852232e-01 2.91570634e-01 3.00392836e-01
7.09858000e-01 -1.32014215e+00 -3.17257345e-01 -3.06178629e-02
-8.77573043e-02 -7.51102328e-01 5.90119481e-01 7.00693309e-01
-9.60863352e-01 4.13153112e-01 -6.04098320e-01 -1.25326082e-01
-1.62680984e+00 6.78976953e-01 2.78642833e-01 8.85589272e-02
-2.32605591e-01 6.28453970e-01 -4.95735198e-01 -1.09870657e-01
4.16737288e-01 6.57979190e-01 -7.04469442e-01 -8.25836882e-02
2.72552222e-01 1.01858187e+00 -4.32856172e-01 -6.67586684e-01
-4.78383452e-01 5.43387890e-01 2.49528140e-01 -6.02597333e-02
1.17828178e+00 -7.24960804e-01 -5.08470953e-01 -4.07788277e-01
1.21992326e+00 -4.24019396e-02 -3.95399421e-01 -1.43105820e-01
-3.01473856e-01 -8.66760790e-01 4.52613294e-01 -3.70102316e-01
-1.30946302e+00 2.83165365e-01 8.75534058e-01 3.81369889e-01
1.03773606e+00 -5.30598104e-01 9.54497039e-01 -1.26308024e-01
6.89111471e-01 -1.46174788e+00 -7.01829568e-02 2.97938615e-01
2.01509535e-01 -1.08092010e+00 -7.04980791e-02 -9.48970973e-01
-8.47159475e-02 1.31675470e+00 9.50765133e-01 1.42707795e-01
1.15721810e+00 -1.66460462e-02 -1.67949498e-01 4.31541391e-02
-6.36992991e-01 -1.85727626e-01 -5.42386532e-01 5.69170952e-01
-5.40140644e-03 2.01121435e-01 -9.82201993e-01 5.37357144e-02
2.29698047e-01 -8.25266317e-02 9.36929464e-01 5.83913863e-01
-9.26613927e-01 -1.40089417e+00 -1.74241543e-01 3.08371603e-01
-3.57635796e-01 2.90272564e-01 6.48268878e-01 8.85316968e-01
1.59871802e-01 1.31758595e+00 3.77163887e-02 -3.49548161e-01
1.84188575e-01 -5.63297868e-01 -1.22964606e-01 -5.36460392e-02
2.63216764e-01 -1.10831760e-01 2.96759158e-01 -3.48186225e-01
-5.21658838e-01 -6.42106235e-02 -1.44867015e+00 -1.10081208e+00
-6.16894305e-01 7.68836737e-01 1.20916402e+00 5.55916905e-01
2.47026622e-01 9.40149009e-01 1.05585229e+00 -2.77694672e-01
-5.02025634e-02 -3.96605492e-01 -8.45784485e-01 -2.72894781e-02
-2.23916322e-02 -1.01911736e+00 -4.60614979e-01 -7.95240939e-01] | [5.8687872886657715, 1.7084227800369263] |
ad786363-5292-4684-b5ac-d32a686fedaf | human-skeletons-and-change-detection-for | null | null | https://www.sciencedirect.com/science/article/pii/S1077314223001194 | https://www.sciencedirect.com/science/article/pii/S1077314223001194 | Human skeletons and change detection for efficient violence detection in surveillance videos | In our constantly monitored world, surveillance cameras play a crucial role in curbing crime and violence in public spaces by serving as a deterrent. To enhance their effectiveness, there is a growing need for automated tools that can detect crimes in real time. In this paper, we propose a novel deep learning architecture that accurately and efficiently detects violent crimes in surveillance videos. We rely on what we believe are the most essential pieces of information to detect violence, namely: human bodies and their interaction. To this end, we employ human pose extractors and change detectors as the input of our proposal. Subsequently, we combine them using a novel method, which relies on additions instead of multiplications to guarantee the transmission of information even when one of the inputs provides a zero-valued signal; outperforming other combination alternatives of the literature. Finally, to account for both spatial and temporal information, we use a convolutional alternative of the standard LSTM, the ConvLSTM. The experiments performed on several benchmark datasets demonstrate the efficacy and efficiency of our proposal, achieving state-of-the-art results with much fewer trainable parameters. We release the code to replicate the proposed architecture at https://github.com/atmguille/Violence-Detection-With-Human-Skeletons | ['Juan C. San Miguel', 'Guillermo Garcia-Cobo'] | 2023-05-20 | null | null | null | computer-vision-and-image-understanding-2023 | ['change-detection'] | ['computer-vision'] | [ 2.12965161e-01 -2.94641376e-01 1.02650076e-01 -1.09617390e-01
-4.10349399e-01 -5.13216496e-01 6.18469238e-01 9.56927985e-03
-8.06366861e-01 4.69989061e-01 -5.11710858e-03 -1.49186850e-01
1.10785484e-01 -9.20228124e-01 -7.53467739e-01 -7.62804210e-01
1.21785803e-02 -1.67634189e-01 5.59138417e-01 -1.76751286e-01
3.10582016e-02 7.06720710e-01 -1.30571973e+00 2.55980641e-01
5.05764604e-01 1.03367972e+00 -1.22600406e-01 7.09721088e-01
4.95727718e-01 9.06278610e-01 -3.38618487e-01 -8.42254400e-01
3.16222876e-01 -1.31201178e-01 -5.04896998e-01 -1.42113775e-01
4.78048116e-01 -9.42139745e-01 -6.54223740e-01 1.04283345e+00
4.24364746e-01 6.07655905e-02 4.66447353e-01 -9.36839938e-01
-1.35498807e-01 1.79249361e-01 -8.78960490e-01 3.98084939e-01
3.95800829e-01 2.79702663e-01 6.00694358e-01 -6.05591416e-01
3.27113360e-01 1.06382430e+00 7.69328296e-01 6.02057815e-01
-7.86383092e-01 -6.97411299e-01 -1.60210282e-01 4.67022508e-01
-1.26687455e+00 -6.39838099e-01 8.57289970e-01 -4.29521918e-01
7.95432866e-01 2.24406108e-01 7.02145815e-01 1.63219070e+00
1.66259259e-01 6.91283345e-01 4.88729984e-01 -1.87874019e-01
-2.40665954e-02 -7.12459683e-02 -6.38333187e-02 8.76267850e-01
4.18139368e-01 6.90489933e-02 -3.18130612e-01 -1.43898189e-01
8.60915840e-01 4.74384695e-01 -3.01078171e-01 -1.17452152e-01
-8.59974980e-01 7.95772910e-01 3.58515173e-01 5.25924623e-01
-4.89705920e-01 3.96850705e-01 5.96849144e-01 -8.66001248e-02
4.84066457e-01 4.01249342e-02 -3.41811962e-02 -3.38042676e-01
-8.89056742e-01 1.44655988e-01 3.18334490e-01 3.78485650e-01
3.25303227e-01 -9.89525393e-02 -1.42318889e-01 5.93291938e-01
1.23088717e-01 4.93552476e-01 2.23967597e-01 -7.51735032e-01
6.19725645e-01 5.72801054e-01 9.23239440e-02 -1.53111970e+00
-6.34584904e-01 -2.93053031e-01 -1.07329178e+00 5.44250943e-02
4.17542040e-01 -2.82693774e-01 -5.43260396e-01 1.68263686e+00
5.30394435e-01 4.62385535e-01 -2.96395808e-01 8.67135108e-01
6.52937949e-01 5.77542424e-01 1.71674993e-02 -1.24395750e-01
1.30861926e+00 -5.41992009e-01 -6.14841640e-01 -2.93904319e-02
4.77990568e-01 -4.37477440e-01 5.77138007e-01 4.77492660e-01
-9.36040521e-01 -4.19310629e-01 -8.34589243e-01 -3.61037105e-02
-3.05202037e-01 3.81435305e-01 3.75175089e-01 7.07489729e-01
-7.80888855e-01 6.61676049e-01 -1.20299721e+00 -3.92554700e-01
5.48731089e-01 2.61138469e-01 -4.95689124e-01 2.33859763e-01
-1.18001866e+00 8.17413628e-01 2.87120372e-01 5.12755632e-01
-7.45960951e-01 -1.73398823e-01 -9.04648066e-01 -2.55366713e-02
4.71192539e-01 -3.26674342e-01 1.03886712e+00 -7.22279012e-01
-1.27432978e+00 6.81671381e-01 3.29702944e-02 -5.70017755e-01
8.83497953e-01 -5.28133988e-01 -1.86124533e-01 5.74041009e-01
-1.51574895e-01 3.69551063e-01 1.01305056e+00 -9.21594024e-01
-7.59251177e-01 -4.91775811e-01 1.76908180e-01 -2.31422007e-01
-7.90761650e-01 2.45829090e-01 -4.45776731e-01 -5.87998927e-01
-3.09920788e-01 -9.17643011e-01 -1.77435912e-02 1.53649136e-01
-5.22414207e-01 -1.09187767e-01 1.00222051e+00 -1.03621972e+00
1.30283177e+00 -2.17411304e+00 2.17057634e-02 8.34508762e-02
4.65391576e-01 8.58380020e-01 1.35291934e-01 4.06589031e-01
1.16664357e-01 -2.11863860e-01 -4.11424220e-01 -5.94048381e-01
-2.33818084e-01 -2.19109710e-02 -1.82765469e-01 9.52996373e-01
3.00229907e-01 7.74052441e-01 -8.42290401e-01 -4.63826835e-01
4.77900594e-01 7.93893099e-01 -3.17168057e-01 8.29444751e-02
2.92886138e-01 4.62284058e-01 -4.55260456e-01 5.91218829e-01
6.00298107e-01 1.49024099e-01 -9.08640996e-02 -9.45825949e-02
-2.06142053e-01 -3.14917713e-02 -1.14418685e+00 1.35616100e+00
-1.76556334e-01 6.96693599e-01 1.85488120e-01 -1.14633095e+00
7.24939287e-01 5.61177790e-01 4.58227277e-01 -7.62214839e-01
4.18034583e-01 1.85869157e-01 -2.26801604e-01 -7.88218319e-01
2.33269066e-01 2.03645498e-01 2.91000009e-02 2.06218824e-01
-2.64034010e-02 3.88072312e-01 2.49674693e-01 -7.56356213e-03
1.39249313e+00 -1.79692730e-02 3.29651326e-01 2.81493008e-01
6.12592280e-01 -4.77453411e-01 4.93966460e-01 6.72872245e-01
-3.89391989e-01 4.53784466e-01 5.43830395e-01 -6.95396125e-01
-9.09047961e-01 -7.43568361e-01 2.49560326e-02 8.01286995e-01
-1.31822139e-01 -1.22508623e-01 -9.65395629e-01 -6.98158801e-01
-4.59133923e-01 3.85620534e-01 -7.25012958e-01 -1.48626655e-01
-9.03195262e-01 -6.15382075e-01 9.12025452e-01 6.08762026e-01
7.49139667e-01 -1.08409441e+00 -1.46159554e+00 1.84251517e-01
-2.16165945e-01 -1.45785463e+00 -1.66022331e-01 -2.32846305e-01
-6.06804609e-01 -1.34681535e+00 -8.06134164e-01 -3.89320880e-01
4.02628630e-01 3.28267902e-01 4.79922652e-01 2.17340663e-01
-3.75116616e-01 3.21878046e-01 -4.94108051e-01 -2.62916952e-01
-1.19243145e-01 3.97993438e-02 -3.06216944e-02 3.91559303e-01
1.59513190e-01 -6.28881931e-01 -6.68001473e-01 -1.20324738e-01
-1.08777773e+00 1.61842313e-02 4.39930588e-01 5.06888330e-01
-7.23375678e-02 -3.87676037e-03 1.17543302e-01 -4.77073520e-01
5.28813899e-01 -4.48975235e-01 -6.89730406e-01 -1.89066511e-02
3.22040647e-01 -3.17932338e-01 8.14781904e-01 -4.08205926e-01
-8.29245031e-01 3.83885443e-01 -3.64228129e-01 -4.82857168e-01
-3.45028937e-01 9.22476277e-02 1.66359209e-02 5.16269766e-02
3.74796122e-01 2.48731256e-01 -3.18263590e-01 -4.80271488e-01
1.12303361e-01 5.53360879e-01 7.11396396e-01 -2.52140313e-01
7.80204773e-01 9.16964352e-01 1.78902775e-01 -1.04199052e+00
-5.32128215e-01 -4.43205059e-01 -7.55190134e-01 -4.46908861e-01
1.03770351e+00 -5.42487264e-01 -1.09038782e+00 9.05289233e-01
-1.77042067e+00 1.87350325e-02 1.61623344e-01 5.11334240e-01
-1.66188344e-01 5.70521474e-01 -7.91263402e-01 -1.17416918e+00
-5.33224106e-01 -1.02445781e+00 1.17466354e+00 1.42653286e-01
3.17028649e-02 -8.56928349e-01 1.61463797e-01 2.10236862e-01
2.57614672e-01 8.02351832e-01 3.27650845e-01 -5.52680314e-01
-1.52304471e-01 -4.00952637e-01 -2.57418990e-01 4.83632416e-01
1.04241244e-01 1.35777757e-01 -1.01060236e+00 -2.91894168e-01
1.94903746e-01 -1.36655971e-01 1.06921005e+00 3.04680258e-01
1.02785134e+00 -4.79773730e-01 -2.75778025e-01 5.25335908e-01
1.15479612e+00 2.59495109e-01 7.03403234e-01 2.64416218e-01
8.84793580e-01 7.45588481e-01 1.45915344e-01 6.94266737e-01
3.66489291e-01 7.86447942e-01 7.75688648e-01 -3.22373629e-01
1.96607158e-01 -6.78522214e-02 4.93437558e-01 4.44225311e-01
-5.42676330e-01 -1.79757044e-01 -1.07387149e+00 5.71219385e-01
-2.01615882e+00 -1.43050563e+00 -4.10756737e-01 2.19244170e+00
5.93995810e-01 1.08957015e-01 3.49745870e-01 4.19975340e-01
8.06623459e-01 1.84157312e-01 -2.59636670e-01 -1.64851114e-01
2.16388136e-01 2.05885485e-01 5.47331989e-01 3.30937415e-01
-1.61145318e+00 8.49877357e-01 5.13524055e+00 7.01203227e-01
-1.35454977e+00 2.99660087e-01 4.93784338e-01 -2.01947287e-01
3.59785140e-01 -5.28291583e-01 -5.81356168e-01 6.49404347e-01
7.15949893e-01 5.17961979e-01 2.53727287e-01 5.50347269e-01
6.32334769e-01 -1.03117891e-01 -8.87741923e-01 1.04862475e+00
9.84618440e-02 -9.82924223e-01 -2.57173866e-01 -1.70128625e-02
2.94984818e-01 -2.31098399e-01 -2.83661997e-04 -5.24980873e-02
-1.10175043e-01 -8.66015732e-01 8.72626364e-01 6.04793608e-01
5.01561046e-01 -8.25875938e-01 8.62128139e-01 5.34830868e-01
-1.23060930e+00 -2.26576626e-01 -1.56114757e-01 -3.31266820e-01
2.54532576e-01 5.64908743e-01 -5.89968503e-01 3.96688432e-01
8.20881903e-01 7.45327353e-01 -4.91741240e-01 1.01212215e+00
-5.21472633e-01 6.27064764e-01 -3.93002242e-01 6.63625970e-02
3.63097608e-01 -1.61557961e-02 5.78249812e-01 1.49231148e+00
3.16025436e-01 2.49028698e-01 -1.01967648e-01 6.96786344e-01
7.28804842e-02 -2.05730468e-01 -7.77490139e-01 3.22315902e-01
2.50355601e-01 1.20085382e+00 -6.60369754e-01 -1.59232721e-01
-4.33450758e-01 9.59052682e-01 2.58445531e-01 1.17116846e-01
-1.28758574e+00 -4.42159802e-01 4.95257705e-01 1.27662778e-01
3.24671388e-01 -4.18689698e-01 2.39471085e-02 -1.11204517e+00
4.26120400e-01 -8.24428499e-01 2.96465218e-01 -2.59064794e-01
-9.34271455e-01 4.87904221e-01 1.92895547e-01 -1.07833493e+00
-2.65478522e-01 -5.91675997e-01 -6.84266627e-01 3.02637607e-01
-1.33021569e+00 -1.26732934e+00 -3.29948276e-01 8.36401820e-01
4.56423670e-01 1.97828069e-01 4.49991494e-01 6.36659384e-01
-9.12271678e-01 5.18182755e-01 -2.42608190e-01 6.28058076e-01
4.16320413e-01 -7.03198731e-01 5.43035269e-01 1.26003468e+00
1.06153630e-01 4.12911206e-01 6.12549722e-01 -5.10938466e-01
-1.29397202e+00 -1.06003225e+00 6.71328127e-01 -2.68034816e-01
6.61086261e-01 -5.30424833e-01 -7.62871742e-01 5.81944108e-01
4.97115105e-02 -1.23242997e-02 2.37834468e-01 -4.04020369e-01
-2.36807644e-01 -1.04843840e-01 -1.19499350e+00 4.74221766e-01
1.01915121e+00 -3.09109807e-01 -5.30423284e-01 2.92709246e-02
3.76194209e-01 -3.18809599e-01 -2.88265765e-01 4.64820623e-01
6.94004655e-01 -1.35310733e+00 8.12061369e-01 -1.85219020e-01
5.53320646e-01 -1.36614636e-01 1.02332659e-01 -8.57072890e-01
-1.41033679e-01 -3.68316591e-01 -3.10032308e-01 1.05408359e+00
-1.43315289e-02 -7.66761184e-01 5.99488854e-01 4.21230584e-01
1.03874899e-01 -7.73795724e-01 -1.24123454e+00 -6.42373919e-01
-3.26921880e-01 -6.51219547e-01 2.64514625e-01 8.52147937e-01
-1.53015986e-01 4.58667576e-02 -8.06203008e-01 3.57248545e-01
6.46817803e-01 -5.70203364e-01 6.78280592e-01 -1.00685406e+00
-3.00144345e-01 -3.68413597e-01 -6.73174024e-01 -7.64588773e-01
1.85922403e-02 -2.48138517e-01 -2.07304716e-01 -1.38812494e+00
4.23913032e-01 1.77928671e-01 -1.14135832e-01 9.07054484e-01
-9.45347827e-03 5.52799821e-01 2.67377943e-01 3.73850651e-02
-6.41687274e-01 4.70080346e-01 8.54012132e-01 -9.68169123e-02
-9.27610472e-02 -4.93359119e-02 -1.82529449e-01 1.08458400e+00
8.16849291e-01 -5.16355038e-01 -9.97923836e-02 -6.60356462e-01
1.43207699e-01 -5.44675142e-02 8.91649306e-01 -1.25936544e+00
3.15004319e-01 -1.84205212e-02 3.98397416e-01 -3.95411342e-01
5.93805015e-01 -9.57128763e-01 -6.97718486e-02 7.82683790e-01
-4.61264998e-02 1.06677465e-01 2.37797156e-01 2.34702051e-01
-8.06974694e-02 -2.34883696e-01 8.39938641e-01 4.10341918e-02
-4.37239110e-01 2.55500525e-01 -4.95909691e-01 -2.67606497e-01
1.11604631e+00 -8.79182592e-02 -3.69349867e-01 -5.16689539e-01
-2.39822209e-01 -7.15691745e-02 1.97577640e-01 4.04713899e-01
7.70790875e-01 -1.03621042e+00 -7.97288358e-01 -4.70043533e-03
-3.23450714e-01 -3.13406140e-01 2.38628492e-01 9.99766052e-01
-6.34644985e-01 4.66532111e-01 -2.36748576e-01 -4.75444764e-01
-1.43994880e+00 3.60652924e-01 3.51753175e-01 -2.83435524e-01
-5.87753534e-01 6.23644352e-01 2.98778381e-04 -1.10164713e-02
2.90000945e-01 -2.89118528e-01 -3.94077152e-01 5.40501997e-02
8.03230822e-01 7.36916542e-01 1.08703308e-01 -7.64951229e-01
-5.59510291e-01 5.33408105e-01 1.69577360e-01 -1.06921524e-01
1.63174033e+00 2.23839492e-01 -7.52970129e-02 1.30449668e-01
1.14574993e+00 -2.19602585e-02 -1.35436285e+00 7.92741627e-02
-1.60229936e-01 -4.61913228e-01 5.30095361e-02 -3.06726754e-01
-1.27837050e+00 1.08244443e+00 6.86458468e-01 3.42487067e-01
1.31383586e+00 -1.84163347e-01 1.05669916e+00 3.47969621e-01
4.06729490e-01 -8.97123337e-01 1.05777182e-01 3.19450200e-01
8.27152610e-01 -1.21147728e+00 -9.20887813e-02 -1.41272172e-01
-3.18702072e-01 1.21832120e+00 4.16397631e-01 -1.81352347e-01
2.98727959e-01 2.80742109e-01 -1.49431393e-01 -1.66919842e-01
-3.41028780e-01 -2.67645746e-01 1.64597869e-01 4.19049829e-01
3.12807709e-01 2.16701883e-03 -5.00297606e-01 4.79754418e-01
1.24158628e-01 -1.47905335e-01 3.44853908e-01 9.37797129e-01
-3.86917502e-01 -8.72549772e-01 -6.26628339e-01 1.68010846e-01
-7.13737786e-01 1.16819972e-02 -5.10769725e-01 7.68578649e-01
3.95089567e-01 1.14046490e+00 -1.42734542e-01 -4.24570680e-01
4.77645218e-01 -3.28276098e-01 3.99825126e-01 -2.26882517e-01
-6.50907874e-01 -9.49358419e-02 -2.62179133e-02 -8.70201528e-01
-4.11714673e-01 -6.35567605e-01 -9.83923435e-01 -3.56034875e-01
-9.44775715e-02 -3.34999979e-01 6.01662993e-01 1.04940534e+00
1.13210522e-01 2.47396424e-01 5.95263302e-01 -1.15238523e+00
-4.55432922e-01 -9.01719093e-01 -2.48179182e-01 4.93536025e-01
5.65769970e-01 -7.19599009e-01 -1.33334771e-01 1.44288708e-02] | [7.9570512771606445, 0.7918953895568848] |
30941d1a-ee1f-4afb-8400-e7a268724b6c | robust-structured-declarative-classifiers-for | 2203.15245 | null | https://arxiv.org/abs/2203.15245v1 | https://arxiv.org/pdf/2203.15245v1.pdf | Robust Structured Declarative Classifiers for 3D Point Clouds: Defending Adversarial Attacks with Implicit Gradients | Deep neural networks for 3D point cloud classification, such as PointNet, have been demonstrated to be vulnerable to adversarial attacks. Current adversarial defenders often learn to denoise the (attacked) point clouds by reconstruction, and then feed them to the classifiers as input. In contrast to the literature, we propose a family of robust structured declarative classifiers for point cloud classification, where the internal constrained optimization mechanism can effectively defend adversarial attacks through implicit gradients. Such classifiers can be formulated using a bilevel optimization framework. We further propose an effective and efficient instantiation of our approach, namely, Lattice Point Classifier (LPC), based on structured sparse coding in the permutohedral lattice and 2D convolutional neural networks (CNNs) that is end-to-end trainable. We demonstrate state-of-the-art robust point cloud classification performance on ModelNet40 and ScanNet under seven different attackers. For instance, we achieve 89.51% and 83.16% test accuracy on each dataset under the recent JGBA attacker that outperforms DUP-Net and IF-Defense with PointNet by ~70%. Demo code is available at https://zhang-vislab.github.io. | ['Guanghui Wang', 'Cuncong Zhong', 'Ziming Zhang', 'Kaidong Li'] | 2022-03-29 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Li_Robust_Structured_Declarative_Classifiers_for_3D_Point_Clouds_Defending_Adversarial_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Li_Robust_Structured_Declarative_Classifiers_for_3D_Point_Clouds_Defending_Adversarial_CVPR_2022_paper.pdf | cvpr-2022-1 | ['point-cloud-classification'] | ['computer-vision'] | [-2.91121691e-01 -2.38067936e-02 6.48805723e-02 -2.60363847e-01
-9.19715345e-01 -1.23192918e+00 6.48197412e-01 -1.92208529e-01
-2.61910796e-01 2.27674380e-01 -3.12935829e-01 -6.09755278e-01
-2.21669860e-02 -9.38813448e-01 -1.39292872e+00 -6.19305193e-01
-5.73567688e-01 5.35634041e-01 1.07253045e-01 -3.72470438e-01
3.69615257e-01 1.31403553e+00 -9.51874971e-01 2.15504244e-01
7.08090842e-01 1.35293710e+00 -7.03238308e-01 6.60911262e-01
1.95932791e-01 6.68498635e-01 -8.16299856e-01 -7.66789913e-01
9.41987038e-01 7.34444320e-01 -5.89985251e-01 -4.53116566e-01
9.50035334e-01 -4.12679762e-01 -8.62900734e-01 1.35942757e+00
3.74666393e-01 -2.96527028e-01 6.06819332e-01 -1.49038994e+00
-8.16990137e-01 3.75631422e-01 -4.44594145e-01 8.40108022e-02
-4.85280119e-02 5.98313332e-01 7.34660745e-01 -8.97316515e-01
3.47695559e-01 1.48941827e+00 9.73932683e-01 6.21364176e-01
-1.11167526e+00 -1.34271455e+00 1.33842930e-01 4.20197919e-02
-1.51390135e+00 -8.93023163e-02 8.87400568e-01 -3.80059212e-01
8.91056418e-01 5.60184300e-01 3.84513378e-01 1.59570813e+00
2.98247844e-01 5.78283906e-01 6.98517203e-01 3.26629102e-01
4.12984103e-01 -4.25927490e-01 -1.46496043e-01 5.18741071e-01
3.75055999e-01 7.35664546e-01 -1.40119597e-01 -8.08049142e-01
6.09396458e-01 7.45565519e-02 -4.89542820e-02 -4.95211005e-01
-8.66485417e-01 1.19101954e+00 1.20400965e+00 -2.73171186e-01
-1.40883982e-01 5.07924497e-01 5.81952691e-01 3.69175464e-01
5.86441457e-01 5.78033566e-01 -3.41487825e-01 3.34582448e-01
-7.33721793e-01 6.23053014e-01 8.59663546e-01 9.25578475e-01
5.35731435e-01 4.99141157e-01 4.63021070e-01 3.58671457e-01
4.79493588e-01 8.74475121e-01 -3.52811337e-01 -8.14217269e-01
6.92617297e-01 2.16244563e-01 -1.00429788e-01 -1.32210374e+00
-8.47570524e-02 -5.49888611e-01 -1.21408510e+00 9.03492749e-01
1.90202668e-02 1.33239171e-02 -1.10148227e+00 1.39858890e+00
4.11147892e-01 8.72846723e-01 1.80188149e-01 9.89514053e-01
6.71028435e-01 4.71004874e-01 -2.44120825e-02 6.14279628e-01
8.20097804e-01 -5.11712968e-01 1.20079555e-01 -1.33594871e-01
2.38712698e-01 -5.09596229e-01 6.66674912e-01 4.58308399e-01
-8.14307868e-01 -2.12221399e-01 -1.38445747e+00 2.06265569e-01
-4.09462571e-01 -6.39686763e-01 5.52793801e-01 6.11081541e-01
-7.15216458e-01 7.98240721e-01 -1.16520023e+00 2.59757698e-01
1.16385615e+00 5.04697382e-01 -4.88448113e-01 1.30269870e-01
-1.22214842e+00 7.24303007e-01 5.54778986e-02 2.50206858e-01
-1.60114264e+00 -1.04503679e+00 -7.67357647e-01 -1.04917526e-01
1.12739932e-02 -5.79815805e-01 9.36765015e-01 -4.95334983e-01
-1.21211994e+00 9.10648286e-01 5.92996240e-01 -1.02337956e+00
7.53579557e-01 -5.37633896e-01 -2.62115598e-01 3.61903965e-01
1.32511765e-01 6.47482514e-01 1.30908012e+00 -1.55090070e+00
-1.86865538e-01 -4.70882505e-01 2.02654272e-01 -1.56579927e-01
1.01732120e-01 1.97638273e-01 4.75138761e-02 -7.96627760e-01
2.42722020e-01 -1.22026205e+00 -4.03462291e-01 5.06876528e-01
-8.32539558e-01 -2.47171596e-02 1.26533103e+00 -2.34361902e-01
2.85317808e-01 -2.17811942e+00 2.25386564e-02 5.91329515e-01
4.89319205e-01 5.05014598e-01 -1.47714242e-01 1.83191106e-01
-4.03359503e-01 4.82991040e-01 -5.05389154e-01 -4.56168383e-01
2.01923341e-01 3.15859318e-01 -1.25231230e+00 1.08834696e+00
3.47919554e-01 8.88507962e-01 -7.34657705e-01 1.61393642e-01
3.83604854e-01 4.92731214e-01 -6.91127300e-01 1.74672753e-01
-1.80307999e-01 4.37134296e-01 -5.29139578e-01 1.11657476e+00
1.36691070e+00 1.76623613e-01 -5.60348153e-01 1.71722646e-03
2.42204815e-01 9.96233746e-02 -8.95232975e-01 1.54579365e+00
-2.30347663e-01 4.00961518e-01 5.38891554e-01 -8.50130856e-01
1.03417921e+00 8.17235112e-02 4.47713703e-01 -1.09957717e-01
1.17594995e-01 3.32744658e-01 -3.55718255e-01 1.80620909e-01
5.52756153e-02 1.26432749e-02 -5.03483593e-01 1.74711496e-02
-1.50445402e-01 -5.20502448e-01 -7.82039523e-01 3.39338154e-01
1.46038544e+00 -2.05092832e-01 -3.00501794e-01 -1.98247686e-01
5.11151373e-01 7.99421668e-02 4.00034815e-01 9.83951151e-01
-1.45416990e-01 7.56011188e-01 4.40383434e-01 -9.76729393e-01
-1.00461388e+00 -1.57329130e+00 -3.31400156e-01 4.31001216e-01
2.82650203e-01 -2.65088141e-01 -4.67906386e-01 -1.03248215e+00
7.98330426e-01 6.64680660e-01 -4.79895830e-01 -3.90846997e-01
-7.70286679e-01 -3.21571589e-01 1.54205143e+00 3.52589667e-01
5.63349485e-01 -7.32957482e-01 -1.92419708e-01 -9.43942294e-02
4.08501089e-01 -1.06621015e+00 -6.34982884e-02 1.96352720e-01
-7.96887577e-01 -1.14607441e+00 1.52130460e-03 -3.06478769e-01
3.24509174e-01 1.46340698e-01 1.28493047e+00 3.12353700e-01
-8.37251637e-03 2.22435579e-01 -2.35266268e-01 -6.53695285e-01
-2.93166250e-01 1.62647977e-01 5.14434218e-01 -1.27099276e-01
1.80030301e-01 -1.06490111e+00 -5.63318849e-01 4.01591212e-01
-1.01980567e+00 -6.95310116e-01 2.52513289e-01 6.67023361e-01
6.89079344e-01 -3.33877295e-01 8.44248682e-02 -6.76710129e-01
2.97612786e-01 -9.00353968e-01 -7.92291582e-01 -2.89387226e-01
-7.62985721e-02 -3.00180793e-01 7.81325698e-01 -2.99783617e-01
-1.71962649e-01 5.39661311e-02 -7.23022342e-01 -1.60474300e+00
-4.40959424e-01 1.62528381e-01 -2.58836478e-01 -1.09770203e+00
1.02854395e+00 -1.87938139e-02 -6.36219233e-02 -3.81137133e-01
4.04129148e-01 3.70519519e-01 9.48458195e-01 -9.19645250e-01
1.92487669e+00 9.59456921e-01 2.87347764e-01 -6.54814124e-01
-6.50241435e-01 -6.56915654e-04 -4.84752804e-01 1.39998719e-02
7.74638176e-01 -1.12486446e+00 -9.73882437e-01 5.36376476e-01
-1.41871381e+00 -1.28755793e-01 -1.83196172e-01 3.59801576e-02
-4.87110287e-01 5.67660809e-01 -5.29995680e-01 -4.44258660e-01
-6.16106331e-01 -1.18041623e+00 1.29090488e+00 -4.90786850e-01
1.43081367e-01 -6.48802578e-01 9.81330127e-03 1.16154276e-01
4.27155197e-01 1.09458399e+00 6.42053366e-01 -9.22458351e-01
-9.91212785e-01 -6.57176852e-01 -3.02586257e-02 5.71964383e-01
-4.51249868e-01 1.14578113e-01 -1.07556105e+00 -6.34631217e-01
3.78629535e-01 -5.11700034e-01 8.59705448e-01 -4.79262583e-02
1.61302376e+00 -7.08845317e-01 -1.87754840e-01 1.70528352e+00
1.36232495e+00 -2.56830752e-01 8.90742064e-01 5.61146498e-01
1.08286381e+00 -1.00622155e-01 1.81582674e-01 4.80990373e-02
-1.75291039e-02 5.02100587e-01 1.49080670e+00 1.13533400e-01
3.40771616e-01 -2.56742567e-01 2.88831532e-01 2.73511171e-01
9.28287208e-02 -1.02661997e-01 -1.26837838e+00 2.70859450e-01
-1.69253457e+00 -9.78691041e-01 5.02446741e-02 1.62951803e+00
3.06939662e-01 5.03294289e-01 -1.55998975e-01 2.86468789e-02
5.03329277e-01 6.31080508e-01 -7.96565354e-01 -2.59165883e-01
-2.29275435e-01 4.92248863e-01 1.12215853e+00 4.69028860e-01
-1.62421906e+00 1.14612544e+00 5.52031183e+00 9.75459933e-01
-1.33211911e+00 1.37420893e-01 4.47864383e-01 -2.83996582e-01
-8.45137462e-02 -1.39870763e-01 -4.69325483e-01 5.67415535e-01
8.01304698e-01 1.21564575e-01 4.28390384e-01 1.40140355e+00
-3.37114304e-01 8.53583515e-01 -9.63794410e-01 1.00731432e+00
-1.37214363e-01 -1.92887497e+00 1.95697561e-01 2.43539676e-01
5.75196147e-01 8.80380094e-01 5.08161128e-01 2.46091336e-01
7.45103061e-01 -1.43568313e+00 1.03517234e+00 1.17855422e-01
8.85389805e-01 -1.06313384e+00 4.82865483e-01 3.79127234e-01
-1.03512764e+00 -1.03986233e-01 -7.71739602e-01 9.27101672e-02
-9.70690101e-02 2.58476347e-01 -5.15315950e-01 5.90697229e-01
1.13555336e+00 8.34377885e-01 -4.44538385e-01 9.30413842e-01
-5.39016366e-01 7.91716158e-01 -8.30395043e-01 4.17796940e-01
7.66712487e-01 8.64951015e-02 1.36390352e+00 9.45454478e-01
9.53286737e-02 1.04920529e-01 1.85154572e-01 1.11384845e+00
-2.19423056e-01 -6.30057514e-01 -1.04007447e+00 4.40742642e-01
6.73397541e-01 9.91111815e-01 -1.79653317e-01 2.42683560e-01
-3.17164771e-02 6.11578107e-01 4.01378423e-01 2.54696578e-01
-9.84600544e-01 -2.73801625e-01 1.62170970e+00 -5.65109476e-02
5.12989998e-01 -5.72441041e-01 -7.34255254e-01 -1.35190654e+00
4.21995483e-02 -1.05491304e+00 2.61980027e-01 -4.54516649e-01
-1.91222763e+00 8.70623171e-01 -2.66833324e-02 -1.62017107e+00
9.35996771e-02 -9.06604350e-01 -1.07917702e+00 6.60699666e-01
-1.31845725e+00 -1.54141819e+00 -2.01116741e-01 8.58595490e-01
8.03718194e-02 -6.90198481e-01 8.46211374e-01 1.57978699e-01
-2.34759644e-01 8.47821116e-01 2.46605501e-02 6.39956713e-01
2.25408062e-01 -1.16626883e+00 1.28997183e+00 1.01014566e+00
3.04433227e-01 5.71395755e-01 4.95720863e-01 -8.71395171e-01
-1.68621969e+00 -1.65143168e+00 -7.48504922e-02 -9.50530410e-01
9.47880208e-01 -8.30180049e-01 -1.12864053e+00 8.54505479e-01
-5.68679795e-02 5.54818332e-01 3.48342210e-01 -2.62671024e-01
-1.10073531e+00 4.66038706e-04 -1.54838479e+00 5.60788572e-01
1.22769523e+00 -7.85331607e-01 -5.95777571e-01 6.58108413e-01
1.04026091e+00 -9.33908582e-01 -8.67011905e-01 6.08090699e-01
5.80166839e-02 -7.46076643e-01 1.70469081e+00 -8.89971435e-01
2.86664486e-01 -5.95739305e-01 -4.89971936e-01 -1.21073389e+00
-4.48344529e-01 -8.96201789e-01 -4.03915286e-01 5.09172499e-01
1.49241522e-01 -9.30420399e-01 1.18431962e+00 2.70520419e-01
-4.54791725e-01 -8.02997530e-01 -1.64721727e+00 -9.68748450e-01
7.83464730e-01 -8.11787724e-01 9.72542763e-01 1.05910444e+00
-7.79686511e-01 -3.02455574e-01 -3.80123705e-01 1.20389283e+00
1.31528664e+00 -2.31045499e-01 1.03261697e+00 -1.14264178e+00
4.95851673e-02 -2.99880862e-01 -1.03159106e+00 -6.83014393e-01
7.48135507e-01 -1.20809662e+00 -3.77033830e-01 -7.57267833e-01
-7.94557750e-01 -7.75437117e-01 -1.69823498e-01 5.93613148e-01
1.90572515e-01 5.29414415e-01 4.55537379e-01 4.68137383e-01
-3.25209409e-01 5.80470860e-01 6.29333496e-01 -6.63965523e-01
3.37242037e-01 2.87388563e-01 -5.07887006e-01 7.65435398e-01
7.96820819e-01 -8.36237073e-01 7.37905875e-02 -8.63717020e-01
8.37473422e-02 -3.67021888e-01 1.12643516e+00 -1.28448498e+00
3.48375469e-01 -3.14651392e-02 4.95861053e-01 -7.69406319e-01
5.36676407e-01 -1.17754912e+00 3.54503840e-01 6.17700398e-01
3.82714868e-02 1.87769294e-01 4.95367199e-01 7.83401251e-01
-1.86648071e-01 1.37081742e-01 9.80035305e-01 -1.44217029e-01
-3.94287050e-01 9.10297692e-01 1.33657053e-01 1.54240364e-02
1.06650972e+00 -8.42748731e-02 -7.14165449e-01 -2.66787261e-01
-5.62593639e-01 3.51042211e-01 6.80271268e-01 4.68915462e-01
9.93170798e-01 -1.55059910e+00 -9.70092535e-01 3.86153311e-01
-1.23013094e-01 5.14519572e-01 9.29777995e-02 2.08059683e-01
-1.15429807e+00 1.87648572e-02 -2.56785959e-01 -1.01396835e+00
-9.39889193e-01 7.08108544e-01 6.55155659e-01 1.38202906e-01
-8.75688136e-01 1.17385221e+00 6.76918700e-02 -8.16147327e-01
3.34452868e-01 -4.90926057e-02 6.05073869e-01 -5.80268383e-01
3.24655503e-01 1.49955273e-01 1.68627709e-01 -6.66617692e-01
-7.57295310e-01 4.98494476e-01 -4.27179262e-02 3.06957096e-01
1.67472756e+00 7.01763213e-01 -2.08152920e-01 -3.45210701e-01
1.37377989e+00 2.69411430e-02 -1.31064725e+00 3.52327898e-02
-1.67804778e-01 -8.11615288e-01 -1.24894455e-01 -4.60072964e-01
-1.23604560e+00 7.59025276e-01 4.70304430e-01 2.48842910e-01
6.74034536e-01 -3.70428450e-02 8.28316510e-01 7.13239253e-01
6.19619906e-01 -1.68391779e-01 -8.99134651e-02 8.24992061e-01
1.28925526e+00 -9.67842519e-01 1.97063945e-02 -5.83214223e-01
-3.99346352e-01 1.02592075e+00 6.01294816e-01 -1.28397560e+00
9.83323932e-01 4.63441700e-01 1.56033292e-01 -4.19629306e-01
-6.12232447e-01 4.45589036e-01 1.37391299e-01 8.88981700e-01
-7.98443496e-01 4.75327298e-02 8.09774220e-01 4.19120878e-01
-6.07255220e-01 -7.04734504e-01 1.47197321e-01 8.83361936e-01
-1.20854340e-01 -6.46426380e-01 -6.24883890e-01 1.53040066e-01
-5.32170177e-01 6.71460330e-02 -6.19491816e-01 7.40500271e-01
7.36056194e-02 5.39658666e-01 1.00899838e-01 -8.18108916e-01
4.44172204e-01 -3.88262808e-01 -2.20908821e-02 -4.73557323e-01
-9.95811284e-01 -6.00508928e-01 -1.79285556e-01 -9.86409783e-01
2.28282139e-01 -5.03527761e-01 -7.42403686e-01 -7.28916764e-01
1.63473971e-02 1.30707890e-01 6.76846981e-01 4.54900652e-01
4.68127012e-01 8.62708688e-02 1.00169003e+00 -1.47500777e+00
-1.17206740e+00 -5.57154596e-01 -2.41891503e-01 4.14730877e-01
6.62464201e-01 -7.41882443e-01 -8.81354630e-01 -7.55630791e-01] | [7.691714286804199, -4.482879638671875] |
9568df1d-72e8-443c-a1c3-f5a085316396 | signal-level-deep-metric-learning-for | 2004.11085 | null | https://arxiv.org/abs/2004.11085v4 | https://arxiv.org/pdf/2004.11085v4.pdf | SL-DML: Signal Level Deep Metric Learning for Multimodal One-Shot Action Recognition | Recognizing an activity with a single reference sample using metric learning approaches is a promising research field. The majority of few-shot methods focus on object recognition or face-identification. We propose a metric learning approach to reduce the action recognition problem to a nearest neighbor search in embedding space. We encode signals into images and extract features using a deep residual CNN. Using triplet loss, we learn a feature embedding. The resulting encoder transforms features into an embedding space in which closer distances encode similar actions while higher distances encode different actions. Our approach is based on a signal level formulation and remains flexible across a variety of modalities. It further outperforms the baseline on the large scale NTU RGB+D 120 dataset for the One-Shot action recognition protocol by 5.6%. With just 60% of the training data, our approach still outperforms the baseline approach by 3.7%. With 40% of the training data, our approach performs comparably well to the second follow up. Further, we show that our approach generalizes well in experiments on the UTD-MHAD dataset for inertial, skeleton and fused data and the Simitate dataset for motion capturing data. Furthermore, our inter-joint and inter-sensor experiments suggest good capabilities on previously unseen setups. | ['Dietrich Paulus', 'Raphael Memmesheimer', 'Nick Theisen'] | 2020-04-23 | null | null | null | null | ['one-shot-3d-action-recognition'] | ['computer-vision'] | [ 6.08373702e-01 -2.03599438e-01 -5.04893005e-01 -6.34777606e-01
-1.34590876e+00 -1.24879442e-01 7.47524858e-01 -3.99295837e-01
-6.61263466e-01 4.76572782e-01 6.52853966e-01 4.30282533e-01
-2.51995981e-01 -5.27882516e-01 -6.82828844e-01 -6.00035906e-01
-2.76493132e-01 1.79324493e-01 8.61599743e-02 6.14146590e-02
1.60832241e-01 4.67359751e-01 -1.59208345e+00 3.59892279e-01
-6.45722747e-02 1.56982350e+00 -3.34012628e-01 8.04922342e-01
5.19997299e-01 1.06761730e+00 -3.99509311e-01 -1.27027379e-02
6.21753633e-01 -4.70237851e-01 -8.87537897e-01 2.38414228e-01
8.32794607e-01 -5.97904503e-01 -7.70410478e-01 5.39464891e-01
9.83613610e-01 2.77611613e-01 4.69096214e-01 -1.45338202e+00
-5.38183808e-01 6.29406422e-02 -3.93338770e-01 2.32850865e-01
8.55525553e-01 4.86334786e-02 1.01115501e+00 -9.47470009e-01
5.98208845e-01 1.17704022e+00 7.66710699e-01 6.98044598e-01
-1.26428843e+00 -3.44623685e-01 -3.32236022e-01 6.27544761e-01
-1.36839032e+00 -7.03980148e-01 6.42538309e-01 -3.56420130e-01
1.28779507e+00 1.94436252e-01 5.57427645e-01 1.42559588e+00
6.54588267e-02 9.36140418e-01 8.98317456e-01 -3.12578589e-01
3.18451911e-01 -4.44001764e-01 -9.82742533e-02 5.98013222e-01
-2.60005891e-01 1.92335874e-01 -1.05500674e+00 -6.80309162e-02
5.01432002e-01 3.93173397e-01 -3.18667024e-01 -7.66822517e-01
-1.52814484e+00 7.74362504e-01 2.93299198e-01 4.12070662e-01
-3.01908225e-01 6.08397484e-01 4.43719715e-01 4.89715755e-01
3.29386026e-01 1.30799696e-01 -3.52342755e-01 -7.97342539e-01
-9.50899899e-01 5.51705509e-02 6.40860438e-01 6.62140131e-01
5.54820240e-01 2.32426692e-02 -2.38913879e-01 7.53128648e-01
1.77833796e-01 5.60089409e-01 7.25470781e-01 -1.42353833e+00
5.97664416e-01 3.51793349e-01 -1.06323361e-02 -8.96825492e-01
-3.92245322e-01 5.58667034e-02 -4.69529092e-01 3.76456380e-01
3.36348146e-01 -2.50987187e-02 -7.86582112e-01 1.72091794e+00
2.47672215e-01 5.27490318e-01 8.25728253e-02 9.55439150e-01
3.82012874e-01 2.52859324e-01 -3.04150611e-01 2.96682213e-02
1.05808938e+00 -7.88026512e-01 -5.62899053e-01 -1.29900917e-01
7.68666089e-01 -3.93963456e-01 9.02904630e-01 4.28319007e-01
-8.15364361e-01 -6.18895710e-01 -1.34493756e+00 -1.48654565e-01
-2.83570617e-01 7.55824223e-02 3.55313241e-01 7.54504263e-01
-1.08643186e+00 1.00672841e+00 -1.12922847e+00 -7.65819967e-01
5.22590578e-01 4.90363419e-01 -8.63416493e-01 -3.77958924e-01
-9.51627076e-01 8.24212074e-01 1.34494722e-01 -4.03204143e-01
-9.28126454e-01 -5.23240864e-01 -1.05902386e+00 -3.28145415e-01
1.95156828e-01 -3.95993114e-01 1.18314862e+00 -6.20254278e-01
-1.69363832e+00 7.40023553e-01 2.52522677e-02 -6.83584809e-01
4.88544405e-01 -3.72735798e-01 -6.34929955e-01 3.28370363e-01
8.06814358e-02 7.18201697e-01 8.07454288e-01 -4.82578218e-01
-5.83443999e-01 -6.62751257e-01 4.27858196e-02 2.63793409e-01
-4.02656019e-01 -1.64320506e-02 -1.28245339e-01 -5.79022169e-01
1.82209253e-01 -1.15157223e+00 2.27047175e-01 4.13720638e-01
-1.42848184e-02 7.27197751e-02 9.80470955e-01 -4.60752755e-01
8.00683737e-01 -2.22621274e+00 4.71828401e-01 -5.61895706e-02
-7.30617642e-02 -7.83739761e-02 -2.22201496e-01 3.63574952e-01
-3.16495627e-01 -2.83561766e-01 -2.08341822e-01 -4.33626562e-01
3.49030882e-01 4.52276200e-01 4.58558165e-02 9.49238479e-01
5.71316369e-02 9.16908622e-01 -6.99713349e-01 -2.67973483e-01
4.41789895e-01 6.69272065e-01 -4.80437964e-01 1.56009972e-01
4.05804783e-01 3.92566621e-01 -1.88270614e-01 1.01216447e+00
1.69485644e-01 -3.01324334e-02 -1.41623363e-01 -5.32707095e-01
2.44309261e-01 7.76072890e-02 -1.35238504e+00 2.50689721e+00
-3.74777555e-01 8.73132467e-01 -1.96797922e-01 -1.37314451e+00
7.25929677e-01 2.76058018e-01 1.11573529e+00 -1.05238783e+00
1.18070310e-02 9.06101167e-02 -2.16260061e-01 -6.86576247e-01
2.64295429e-01 -4.96078506e-02 -2.83670813e-01 6.33617938e-01
3.69361222e-01 2.11829141e-01 -1.11511350e-01 1.54154962e-02
1.86386895e+00 4.29492414e-01 1.72662675e-01 2.11502776e-01
3.71687144e-01 -4.40213352e-01 4.63874042e-01 4.84889269e-01
-6.47955596e-01 8.53520930e-01 1.30135939e-01 -5.26573956e-01
-7.83007205e-01 -1.14946890e+00 -2.13539883e-01 1.11750162e+00
-1.05305150e-01 -5.20623803e-01 -5.40678144e-01 -7.50492215e-01
1.65992275e-01 4.04351860e-01 -9.27261114e-01 -3.77512515e-01
-5.08461833e-01 -4.31941152e-01 7.79085040e-01 8.73144209e-01
7.35489130e-01 -8.76304150e-01 -8.23880732e-01 1.62164018e-01
-1.75033286e-01 -1.20818877e+00 -4.30863261e-01 1.99697435e-01
-8.72915924e-01 -1.23644996e+00 -7.80407250e-01 -2.60685235e-01
8.67009833e-02 1.86050355e-01 8.04264784e-01 -5.68206429e-01
-4.47438091e-01 1.12337792e+00 -4.83549088e-01 -2.09180154e-02
1.23483174e-01 -2.91257262e-01 3.91451389e-01 4.20080364e-01
6.82429552e-01 -5.61960399e-01 -7.19092846e-01 4.38682735e-01
-6.05318129e-01 -6.41937256e-01 4.01479483e-01 6.90991461e-01
5.80049574e-01 -4.75713700e-01 1.88318491e-01 4.57807332e-02
2.81344324e-01 -4.19266164e-01 5.94280995e-02 1.44907057e-01
-4.41097587e-01 2.19089180e-01 -1.01653412e-02 -4.79088128e-01
-5.41792393e-01 3.82177621e-01 8.45681354e-02 -6.53501213e-01
-3.18314999e-01 -3.21424492e-02 -3.07844669e-01 -3.84458482e-01
7.98906028e-01 1.49867028e-01 1.13706224e-01 -6.34388268e-01
3.20292711e-01 7.97403395e-01 5.90289950e-01 -3.80254716e-01
4.80044812e-01 8.48536074e-01 1.32596478e-01 -8.20792198e-01
-5.27471304e-01 -4.83148307e-01 -8.58316720e-01 -3.03780496e-01
1.06956613e+00 -1.01654351e+00 -6.43965423e-01 4.76448536e-01
-8.18607748e-01 -3.23261082e-01 -7.10360944e-01 1.00760889e+00
-1.11594164e+00 3.93912256e-01 -4.17158037e-01 -6.25345886e-01
-8.89773294e-02 -1.13135660e+00 1.69665885e+00 -2.33520329e-01
-4.52355444e-01 -6.61294997e-01 5.05407929e-01 5.38876235e-01
2.55202502e-01 5.69867074e-01 1.19377010e-01 -5.93987584e-01
-3.32434833e-01 -3.84321302e-01 1.42887056e-01 5.22330046e-01
2.90348709e-01 -5.74840009e-01 -1.03483462e+00 -3.43390912e-01
6.52755946e-02 -7.99175322e-01 9.10799682e-01 1.30890578e-01
1.04635668e+00 -8.05486068e-02 -1.04631059e-01 7.06107736e-01
1.09343731e+00 1.86739907e-01 8.45350087e-01 4.56853896e-01
7.00155854e-01 2.24843591e-01 7.22301424e-01 6.70010984e-01
3.03186685e-01 1.01495898e+00 4.62385774e-01 2.39816263e-01
-1.67291462e-01 -3.13453078e-02 8.39449286e-01 4.26829606e-01
-8.77977386e-02 9.08468142e-02 -7.46478140e-01 4.70747858e-01
-1.89803040e+00 -1.33246565e+00 4.56414729e-01 2.19063497e+00
4.37635690e-01 -1.58689648e-01 3.54622245e-01 6.18855298e-01
3.82680953e-01 3.93525213e-01 -6.38650119e-01 -2.82624453e-01
-1.65845430e-03 4.48859602e-01 4.98347998e-01 2.90705234e-01
-1.30311680e+00 3.88836443e-01 6.28635836e+00 5.90954900e-01
-1.02683365e+00 3.12019646e-01 2.47834980e-01 -6.97006881e-01
4.60801005e-01 -3.62374812e-01 -4.83032167e-01 5.19483626e-01
1.28148258e+00 8.31181556e-02 4.41631585e-01 8.45889270e-01
7.09646940e-02 -2.53222317e-01 -1.68103695e+00 1.60613024e+00
6.74842656e-01 -1.06754100e+00 -4.91759986e-01 1.58251151e-01
4.00594741e-01 4.48656082e-01 -8.95946398e-02 2.78590739e-01
-6.64038956e-02 -1.16960824e+00 6.71653926e-01 7.69814253e-01
9.32515681e-01 -5.12189448e-01 4.97991353e-01 1.02152079e-01
-1.29385448e+00 -1.66733757e-01 -2.05141783e-01 -2.10191667e-01
1.48979828e-01 1.14605829e-01 -5.43576896e-01 4.27113414e-01
7.95025706e-01 1.23870730e+00 -6.31984413e-01 8.72475624e-01
1.62283123e-01 3.63607556e-01 -3.82450342e-01 2.89027810e-01
5.80584221e-02 7.19165206e-02 3.46321791e-01 9.32063043e-01
6.35063708e-01 6.64740801e-02 1.66851193e-01 2.25346461e-01
-5.01296334e-02 -2.48155445e-01 -9.69851553e-01 7.01246783e-02
2.35575456e-02 7.96576619e-01 -8.29840600e-02 -2.27031246e-01
-5.82040548e-01 1.49665129e+00 2.43021116e-01 1.94898471e-01
-9.60999370e-01 -4.06562746e-01 9.34633315e-01 -1.82656154e-01
5.16370475e-01 -3.01611453e-01 1.88655123e-01 -1.25801218e+00
3.01197350e-01 -8.44439507e-01 5.75089931e-01 -7.89639056e-01
-9.38959181e-01 2.81395197e-01 -6.60204515e-03 -1.75335741e+00
-6.50844336e-01 -7.02843964e-01 -1.98920608e-01 3.18019032e-01
-1.00681508e+00 -9.52788830e-01 -3.87786210e-01 9.20612514e-01
5.16449332e-01 -3.03227007e-01 9.97382641e-01 5.40745378e-01
-3.85203183e-01 5.55885851e-01 1.37944624e-01 2.88519591e-01
8.42032194e-01 -1.06681466e+00 1.50687903e-01 5.52541792e-01
6.80651963e-01 9.93524566e-02 3.42594653e-01 -1.97129056e-01
-1.71797001e+00 -9.52201247e-01 7.32375860e-01 -7.94586837e-01
6.52052343e-01 -3.83011580e-01 -5.24605751e-01 7.52694249e-01
4.21761796e-02 7.21181214e-01 1.03021204e+00 -2.25060821e-01
-5.13339341e-01 -4.34666485e-01 -1.36489999e+00 6.23964332e-02
1.53071797e+00 -8.90724421e-01 -7.26900935e-01 4.31831717e-01
3.11478525e-01 -3.43462288e-01 -1.34901643e+00 4.07259434e-01
9.57314253e-01 -1.07076275e+00 1.01450765e+00 -7.57314742e-01
6.29320666e-02 -2.86095142e-01 -9.49085176e-01 -1.12858248e+00
-2.38838941e-01 -5.63822448e-01 -4.98683691e-01 9.07325864e-01
8.67057778e-03 -3.01059991e-01 9.13182914e-01 5.66069126e-01
1.43579364e-01 -6.59788966e-01 -1.57866836e+00 -1.18641245e+00
-2.63596684e-01 -9.39594984e-01 4.30141032e-01 7.84913659e-01
1.44676507e-01 1.46874502e-01 -7.24037707e-01 -2.77208954e-01
7.59127140e-01 -5.68573959e-02 8.21012974e-01 -9.09166157e-01
-4.41130817e-01 -7.85376951e-02 -1.34014845e+00 -9.05171752e-01
2.30764583e-01 -8.29379618e-01 -1.97569340e-01 -1.22139239e+00
8.24679062e-02 2.34485552e-01 -5.61773360e-01 7.45790124e-01
4.70434576e-01 6.91994905e-01 2.26652697e-01 5.23551628e-02
-9.44651425e-01 7.74126709e-01 5.76553881e-01 -4.69778985e-01
1.91065282e-01 -2.61032641e-01 -1.95238352e-01 5.64952552e-01
7.38505661e-01 -4.79709923e-01 -3.33426148e-01 -4.43140298e-01
-2.95380235e-01 4.47849929e-02 6.50986075e-01 -1.57844555e+00
1.57792449e-01 2.38744747e-02 4.94657129e-01 -3.17870170e-01
8.08381915e-01 -1.02589583e+00 9.83617231e-02 4.57706451e-01
-3.84173542e-01 -7.63101280e-02 -4.22051139e-02 6.97934330e-01
-2.37237006e-01 1.72066763e-01 6.42679989e-01 2.43105739e-01
-1.06064618e+00 4.34717745e-01 -5.50145023e-02 2.56233186e-01
1.20316756e+00 -7.36031771e-01 1.19318344e-01 -4.65309381e-01
-9.63843107e-01 -4.76749241e-02 4.35440302e-01 6.97790802e-01
6.41475737e-01 -2.01152873e+00 -5.01132071e-01 2.67854273e-01
3.70170534e-01 -6.55671477e-01 6.07497059e-02 1.05517244e+00
-8.14074203e-02 4.97064859e-01 -4.84918058e-01 -8.47889662e-01
-1.26904476e+00 2.57017523e-01 4.49719191e-01 9.90850776e-02
-4.92157817e-01 5.88641822e-01 -4.80632693e-01 -3.58315855e-01
5.12173474e-01 -3.25064421e-01 1.18060179e-01 2.01146990e-01
7.76778221e-01 8.38192880e-01 2.06865042e-01 -9.96365070e-01
-7.35087931e-01 8.72490525e-01 2.90320188e-01 -4.17311043e-01
1.44973981e+00 -2.58927071e-03 5.26787817e-01 6.06095314e-01
1.94899786e+00 -4.65299636e-01 -1.55689418e+00 -2.69223779e-01
1.05481632e-01 -7.72833347e-01 6.36833087e-02 -5.52908838e-01
-1.08414471e+00 9.29067850e-01 1.24086106e+00 -1.78555772e-01
1.03713048e+00 1.69278309e-01 6.99865818e-01 7.19902515e-01
5.46120048e-01 -1.27725589e+00 5.78381658e-01 3.63445282e-01
1.00765204e+00 -1.54420590e+00 1.21871568e-01 3.13651413e-01
-6.05436802e-01 1.03986776e+00 3.29552472e-01 -1.42056957e-01
5.09769499e-01 2.14520350e-01 -1.72184780e-01 -1.91344276e-01
-4.27936226e-01 -3.73692602e-01 2.45742530e-01 8.38355064e-01
2.90322721e-01 -1.18861273e-01 -1.58144869e-02 2.50040442e-01
1.37349805e-02 2.54469961e-01 1.58152014e-01 1.27449870e+00
-2.49072820e-01 -1.15221262e+00 -2.59410143e-01 3.08948219e-01
-3.70986462e-01 3.62824887e-01 -4.96160835e-01 8.73487055e-01
1.50562897e-01 9.61068511e-01 2.06890255e-01 -7.75362909e-01
5.92057645e-01 3.55581522e-01 8.29732120e-01 -4.85564649e-01
-8.19572285e-02 -2.32522398e-01 2.47552291e-01 -1.55114675e+00
-9.01053429e-01 -1.16810203e+00 -1.15239620e+00 -7.21885040e-02
1.44963622e-01 -3.95604312e-01 6.51794314e-01 8.87761772e-01
6.04628563e-01 2.27070853e-01 7.59100676e-01 -1.03415275e+00
-7.94757485e-01 -8.58814299e-01 -6.71359360e-01 5.82460999e-01
3.63794416e-01 -9.68291104e-01 -3.14120829e-01 7.06805885e-02] | [7.925797462463379, 0.5926263332366943] |
62e3f441-6164-427f-aa54-5b07976cf174 | the-adaptive-multi-factor-model-and-the | 2107.14410 | null | https://arxiv.org/abs/2107.14410v2 | https://arxiv.org/pdf/2107.14410v2.pdf | The Adaptive Multi-Factor Model and the Financial Market | Modern evolvements of the technologies have been leading to a profound influence on the financial market. The introduction of constituents like Exchange-Traded Funds, and the wide-use of advanced technologies such as algorithmic trading, results in a boom of the data which provides more opportunities to reveal deeper insights. However, traditional statistical methods always suffer from the high-dimensional, high-correlation, and time-varying instinct of the financial data. In this dissertation, we focus on developing techniques to stress these difficulties. With the proposed methodologies, we can have more interpretable models, clearer explanations, and better predictions. | ['Liao Zhu'] | 2021-07-30 | null | null | null | null | ['algorithmic-trading'] | ['time-series'] | [-6.84432626e-01 -3.98723125e-01 -1.70066699e-01 -3.87823313e-01
8.95447806e-02 -7.09725738e-01 6.17792010e-01 -2.04914182e-01
3.75466049e-02 9.28589106e-01 2.76510179e-01 -5.15715480e-01
-2.69131660e-01 -8.58729541e-01 -1.11986116e-01 -4.43781406e-01
-3.42055410e-01 2.87640452e-01 1.04644842e-01 -2.89869219e-01
7.98326671e-01 4.34227049e-01 -1.09811115e+00 6.43419921e-02
3.56639326e-01 1.05298889e+00 -9.63925794e-02 1.89205065e-01
-4.81532872e-01 4.79957640e-01 -4.16348755e-01 -6.51208758e-01
5.95154703e-01 -2.80364484e-01 -2.67305821e-01 -1.04329340e-01
-4.60244447e-01 -1.97098091e-01 -2.41830543e-01 1.08667493e+00
-1.06320448e-01 -6.54970780e-02 3.59661996e-01 -1.00676394e+00
-6.53717637e-01 7.12498903e-01 -9.10453618e-01 3.98470610e-01
2.32527703e-01 -3.56916003e-02 1.10655081e+00 -8.33503783e-01
5.07096469e-01 1.49052012e+00 5.39679348e-01 9.07244161e-02
-9.58343625e-01 -8.19758952e-01 4.31955680e-02 6.17682263e-02
-7.01161206e-01 -1.25912046e-02 9.42432046e-01 -4.18122083e-01
5.54313064e-01 3.50462675e-01 8.73564303e-01 8.95542443e-01
3.07213515e-01 5.55381238e-01 1.50337350e+00 -2.77108103e-01
-1.30375037e-02 4.75959867e-01 2.66801625e-01 7.86845163e-02
6.66329861e-01 5.23383081e-01 -5.78049064e-01 -2.36975595e-01
7.99716055e-01 4.78408933e-01 1.86598636e-02 -3.30568403e-02
-7.44074643e-01 1.01925457e+00 4.57480289e-02 7.23275363e-01
-3.01452190e-01 -2.38526270e-01 2.98982263e-01 3.90424550e-01
2.90567130e-01 6.47176623e-01 -7.32930243e-01 -5.34515321e-01
-9.52100694e-01 3.55661213e-01 1.08080614e+00 2.96326041e-01
4.17359114e-01 9.30053741e-02 5.12635648e-01 1.79771245e-01
6.04956388e-01 3.47692579e-01 8.74024928e-01 -9.16730464e-01
5.67276299e-01 7.47104108e-01 1.21855721e-01 -1.40528512e+00
-3.26144427e-01 -8.06834817e-01 -7.10621178e-01 2.30572835e-01
5.17324507e-01 -1.62675455e-01 -5.35041511e-01 1.37309349e+00
2.16323182e-01 -2.31373563e-01 1.04921330e-02 8.28321934e-01
-1.26005962e-01 5.22960901e-01 -2.99591541e-01 -3.74009520e-01
1.00583148e+00 -3.59197468e-01 -9.11740422e-01 1.63881317e-01
1.62223712e-01 -8.35474432e-01 7.89519131e-01 6.66342795e-01
-7.91262567e-01 -4.00826812e-01 -9.43195224e-01 4.07929003e-01
-4.44719821e-01 -6.59306228e-01 1.01430237e+00 8.02552223e-01
-6.22542381e-01 7.94819057e-01 -8.74992609e-01 7.91291371e-02
1.78827390e-01 2.18851417e-01 -3.55700701e-02 3.66734505e-01
-1.16484785e+00 8.66367996e-01 2.74023592e-01 2.88207293e-01
-5.70647940e-02 -6.86629057e-01 -2.04512373e-01 -4.00164351e-02
3.37048054e-01 -2.31381983e-01 9.81885433e-01 -8.10458899e-01
-1.29821301e+00 4.19002861e-01 2.17660785e-01 -4.95583236e-01
8.35489392e-01 -4.55869853e-01 -6.15517974e-01 -2.50772625e-01
-1.81238249e-01 -3.17312211e-01 4.49845672e-01 -8.99600685e-01
-6.30098045e-01 -6.94040000e-01 -3.03591669e-01 -2.60473043e-01
-2.82561898e-01 1.44095212e-01 1.81215271e-01 -1.07853734e+00
5.14957845e-01 -6.55952036e-01 -3.55924457e-01 -5.15686274e-01
-2.29086190e-01 9.48707089e-02 9.51591909e-01 -6.79415584e-01
1.34362960e+00 -2.13961911e+00 -2.37920031e-01 3.45775723e-01
2.67731816e-01 8.04132149e-02 7.88806200e-01 6.16265178e-01
-2.63236582e-01 5.20382762e-01 1.08654857e-01 -2.89938413e-02
6.41438365e-02 2.34368458e-01 -5.94236970e-01 1.70195878e-01
-1.30723119e-01 6.86066091e-01 -6.86141729e-01 -2.49124467e-01
-3.34700830e-02 6.33219210e-03 -2.51873612e-01 -6.62637036e-03
-2.80533582e-02 4.65614051e-01 -9.06298518e-01 9.36489403e-01
5.46538591e-01 -3.96893263e-01 2.19302818e-01 1.62827983e-01
-5.78091681e-01 1.35455251e-01 -1.10532951e+00 1.16398752e+00
1.65169507e-01 6.65836990e-01 -3.43394130e-02 -1.11548674e+00
1.03547680e+00 2.24207610e-01 3.97429466e-01 -9.66776609e-01
1.12139009e-01 5.53867102e-01 1.41321644e-01 -4.76082027e-01
4.64066982e-01 -6.87526584e-01 1.16444536e-01 6.18068874e-01
-5.24099350e-01 2.41060257e-01 1.00814268e-01 -1.37752652e-01
6.62765265e-01 -4.74644266e-03 3.99392426e-01 -3.39428604e-01
1.27497353e-02 -1.65557221e-01 9.30424809e-01 2.61921704e-01
-7.18981922e-02 1.69961467e-01 6.65485978e-01 -7.18693554e-01
-9.13935006e-01 -8.82383287e-01 -2.96754569e-01 3.00157785e-01
-2.08584920e-01 2.95361783e-02 -3.19072008e-01 -2.28117540e-01
5.71444035e-01 4.31955189e-01 -4.77303982e-01 1.20767258e-01
-3.58686596e-01 -1.20220625e+00 1.01665311e-01 4.24509734e-01
5.88846862e-01 -7.27577329e-01 -7.56548762e-01 4.12483096e-01
3.80876601e-01 -8.63676965e-01 8.25839415e-02 2.84942221e-02
-1.52663374e+00 -7.94198930e-01 -5.41764438e-01 8.63393955e-03
2.23624721e-01 4.62084748e-02 1.02752173e+00 -1.35527641e-01
-2.07117587e-01 -2.66728163e-01 -3.00035506e-01 -6.84707761e-01
-1.26344070e-01 -2.45605096e-01 1.29602864e-01 1.00232385e-01
4.13954824e-01 -8.00255895e-01 -5.22054851e-01 1.64330050e-01
-7.47329712e-01 -1.16936550e-01 6.76745892e-01 6.13339841e-01
1.12381525e-01 3.48232239e-01 7.56841660e-01 -9.75877821e-01
7.25937366e-01 -6.90217078e-01 -8.84752154e-01 2.12468967e-01
-1.31578696e+00 3.97904128e-01 3.58101010e-01 -4.70730335e-01
-1.19629073e+00 -6.23458445e-01 5.89180112e-01 -1.55173823e-01
1.70202494e-01 1.09864330e+00 1.38636619e-01 2.55221754e-01
2.80519843e-01 -8.33504051e-02 1.42134400e-02 -9.82356369e-01
1.66235760e-01 6.00510716e-01 3.38613331e-01 -5.74133039e-01
9.32082236e-01 3.91345322e-01 1.69719577e-01 -6.98631942e-01
-5.67833364e-01 -1.58707947e-01 -4.72259283e-01 -1.50686577e-01
4.67505693e-01 -5.50488889e-01 -3.93453658e-01 5.92090011e-01
-8.34405005e-01 1.89614937e-01 -1.67356417e-01 8.15443277e-01
2.22337451e-02 2.23086193e-01 -6.51976585e-01 -1.27168524e+00
2.80963117e-03 -6.89322591e-01 1.39242098e-01 6.04933739e-01
-4.45379734e-01 -1.32906663e+00 4.48116958e-01 3.83414030e-01
5.62232912e-01 6.48344219e-01 9.23006237e-01 -9.28919256e-01
-8.56405735e-01 -4.32422608e-01 -1.97086185e-01 2.16428518e-01
3.68303359e-01 2.47317657e-01 -6.72119975e-01 7.35594258e-02
6.38938606e-01 1.53352126e-01 5.28415084e-01 1.07119367e-01
7.55092204e-01 -4.00916994e-01 -1.06683463e-01 5.07980227e-01
1.24823105e+00 7.27553844e-01 2.86948413e-01 6.24103010e-01
1.60530925e-01 8.18204939e-01 7.63833284e-01 8.22527230e-01
3.01963091e-01 1.50508389e-01 2.82800585e-01 1.41829610e-01
6.18284523e-01 -2.55551189e-01 8.06414559e-02 1.04482114e+00
-3.66846830e-01 2.08090842e-01 -1.04988801e+00 2.63343304e-01
-1.84565794e+00 -9.57964242e-01 -1.85887486e-01 2.02161121e+00
7.78222382e-01 4.02232468e-01 1.54392451e-01 2.47273594e-01
5.30426979e-01 6.90276250e-02 -5.75751543e-01 -3.13224792e-01
-1.36583060e-01 8.20306242e-02 5.58539271e-01 1.94178388e-01
-6.53880537e-01 3.62912953e-01 7.69380999e+00 2.10784674e-01
-1.45955110e+00 -3.14835966e-01 8.53516638e-01 -1.56714559e-01
-4.87736344e-01 2.32866287e-01 -7.41128922e-01 7.57798851e-01
9.86450374e-01 -5.75021505e-01 2.33130470e-01 6.22527897e-01
3.56989741e-01 -1.07720234e-01 -7.47477949e-01 8.68291616e-01
-4.25302356e-01 -1.16141486e+00 2.78774910e-02 6.07292175e-01
5.60404718e-01 -1.66059375e-01 3.88466477e-01 -7.36451941e-03
1.34646967e-01 -8.72152686e-01 7.30619192e-01 8.80135894e-01
-7.26072267e-02 -5.59838176e-01 8.92481744e-01 2.86695033e-01
-7.76307702e-01 -2.60080874e-01 -2.23659948e-01 -6.81676328e-01
7.60309473e-02 9.99196172e-01 -3.14073384e-01 7.50299454e-01
7.29128063e-01 7.65672207e-01 -2.13324502e-01 9.51431990e-01
1.55920506e-01 6.01921320e-01 -3.18808258e-01 -3.33189249e-01
9.01760608e-02 -7.54966795e-01 4.41322178e-01 6.48514211e-01
5.28824329e-01 3.05056512e-01 -1.85883656e-01 9.26744282e-01
2.47679248e-01 4.75814044e-02 -5.76954544e-01 -5.89414656e-01
5.81466332e-02 9.69787359e-01 -9.31008637e-01 -2.61607289e-01
-8.78889263e-01 2.84934312e-01 -1.49026364e-01 2.87112713e-01
-5.67489266e-01 1.09758370e-01 7.08170474e-01 3.71034801e-01
-6.71806708e-02 -6.51163101e-01 -6.96949601e-01 -1.21104717e+00
1.86399087e-01 -1.01776707e+00 4.08392936e-01 -2.19035402e-01
-1.48379242e+00 9.70933363e-02 6.86632516e-03 -9.20704842e-01
-3.37411016e-01 -7.60928035e-01 -6.75292075e-01 8.19293797e-01
-1.23727500e+00 -3.21110308e-01 2.51873374e-01 3.19101840e-01
2.24861935e-01 -4.65744376e-01 4.86163080e-01 2.39742100e-01
-5.11546671e-01 1.13968449e-02 4.51337427e-01 1.72857836e-01
4.62353081e-01 -1.19542909e+00 2.22060785e-01 7.31488705e-01
2.76909202e-01 8.16371500e-01 9.44142640e-01 -6.55371368e-01
-1.29180110e+00 -2.10594922e-01 7.32470632e-01 -4.23203140e-01
1.43823850e+00 -1.36082679e-01 -9.88204956e-01 5.17551720e-01
-1.32032353e-02 -3.84163260e-01 8.10852289e-01 4.04647142e-01
-3.16861004e-01 -2.68086314e-01 -8.92983317e-01 3.47864777e-01
3.42825413e-01 -4.20585215e-01 -1.05563581e+00 -2.16391489e-01
4.08598781e-01 7.66637549e-02 -6.80512071e-01 3.16628546e-01
8.97852242e-01 -1.24008322e+00 5.79174519e-01 -8.17542374e-01
4.59933877e-02 2.01243147e-01 -9.20288414e-02 -8.15613270e-01
-8.94251466e-02 -9.11467493e-01 -6.18218854e-02 1.17102420e+00
6.30321860e-01 -1.02240562e+00 9.33603168e-01 1.31018507e+00
4.85916823e-01 -7.26363659e-01 -7.81122863e-01 -7.79198349e-01
6.21869005e-02 -1.48055807e-01 6.84599578e-01 1.24748421e+00
5.11982977e-01 -1.31506324e-01 -3.04922819e-01 -3.34185034e-01
8.01846385e-01 7.81544507e-01 5.12520552e-01 -1.50177991e+00
-4.65324104e-01 -6.12361848e-01 -4.51787084e-01 -7.36885607e-01
-5.36965489e-01 -2.26753250e-01 -6.14950657e-01 -7.68863261e-01
2.31418908e-01 -4.23856914e-01 -6.96927011e-01 -2.55854651e-02
-2.38248557e-02 -1.64271116e-01 2.18239382e-01 4.42319691e-01
-1.03026062e-01 2.97054529e-01 1.06436503e+00 3.33431602e-01
-2.36783430e-01 2.49256253e-01 -6.64245427e-01 8.23156178e-01
9.48160708e-01 -4.66891229e-01 -2.00654015e-01 -6.91315606e-02
7.74347425e-01 2.87020475e-01 1.71642795e-01 -6.81231678e-01
9.16995332e-02 -4.68136996e-01 4.61325854e-01 -5.58690608e-01
7.56568462e-02 -7.38760471e-01 5.60557425e-01 7.69948721e-01
7.54010454e-02 5.36858797e-01 -5.88158704e-02 3.77980113e-01
-7.79862940e-01 -1.29276752e-01 4.01431203e-01 -4.28512365e-01
-2.46569663e-01 1.17753997e-01 -1.28175512e-01 2.29978599e-02
9.84463096e-01 -2.66712606e-01 -1.39338020e-02 -3.02493781e-01
-5.63604832e-01 2.07679689e-01 2.75166452e-01 5.70285320e-01
2.98737317e-01 -1.09869146e+00 -4.91018236e-01 1.89772934e-01
-6.84552312e-01 -2.67815739e-01 -1.04796037e-01 7.48300970e-01
-5.59754789e-01 7.46407866e-01 -3.48243177e-01 -7.38905556e-03
-8.87742400e-01 3.89173061e-01 1.22478195e-01 -5.22973537e-01
-4.90460008e-01 3.99371386e-01 2.82469451e-01 1.94255084e-01
-2.65133768e-01 -6.22503757e-01 -1.09777935e-01 5.27000725e-01
3.91582400e-01 6.11988366e-01 -4.72802252e-01 -1.49089411e-01
-1.46319315e-01 4.49359804e-01 -7.29519576e-02 -3.83473456e-01
1.60269570e+00 -7.74787664e-02 -2.48728931e-01 1.10333884e+00
6.81294024e-01 2.34161377e-01 -1.00856614e+00 -8.62470120e-02
7.24190950e-01 -7.69157827e-01 -1.27176031e-01 -5.78875840e-01
-1.13478935e+00 7.22871006e-01 2.45966867e-01 9.45618927e-01
7.66706228e-01 -2.33709216e-01 7.59042501e-01 1.54658243e-01
5.26449680e-01 -1.11991417e+00 -2.06687544e-02 2.54101783e-01
5.59852242e-01 -1.22443986e+00 2.08020344e-01 -1.88524891e-02
-2.88285822e-01 1.33344042e+00 3.92852984e-02 -1.49613470e-02
1.17053533e+00 3.91668290e-01 2.75189281e-01 -2.30641678e-01
-7.90791690e-01 4.15715218e-01 7.04134256e-02 1.25131533e-01
4.88852888e-01 1.71498001e-01 -5.41441023e-01 9.12317097e-01
-5.23884773e-01 -1.23331174e-02 3.79003108e-01 8.52231801e-01
-6.02436423e-01 -1.44285071e+00 -5.69418192e-01 4.13186669e-01
-1.10324299e+00 7.10605308e-02 -5.30079126e-01 1.20691562e+00
-3.00223887e-01 5.86154103e-01 -1.25731807e-02 -2.35922411e-01
2.25751847e-01 2.01048270e-01 -1.65153053e-02 -2.62739003e-01
-2.88600653e-01 1.57532915e-01 -2.16188788e-01 -5.31017125e-01
-4.45303321e-01 -1.26061440e+00 -1.19081652e+00 -6.32213354e-01
-4.17319000e-01 4.37005132e-01 7.89182723e-01 9.10814583e-01
1.64883345e-01 1.60924464e-01 8.25393081e-01 -2.60248214e-01
-1.00215971e+00 -1.06801105e+00 -1.00613582e+00 2.20599294e-01
1.59073770e-01 -7.29860008e-01 -6.40162528e-01 -2.13007703e-01] | [4.605595111846924, 4.1562323570251465] |
eeb7d4fe-8e11-40c0-94fc-73c53a0d75d3 | ftgan-a-fully-trained-generative-adversarial | 1904.05729 | null | http://arxiv.org/abs/1904.05729v1 | http://arxiv.org/pdf/1904.05729v1.pdf | FTGAN: A Fully-trained Generative Adversarial Networks for Text to Face Generation | As a sub-domain of text-to-image synthesis, text-to-face generation has huge
potentials in public safety domain. With lack of dataset, there are almost no
related research focusing on text-to-face synthesis. In this paper, we propose
a fully-trained Generative Adversarial Network (FTGAN) that trains the text
encoder and image decoder at the same time for fine-grained text-to-face
generation. With a novel fully-trained generative network, FTGAN can synthesize
higher-quality images and urge the outputs of the FTGAN are more relevant to
the input sentences. In addition, we build a dataset called SCU-Text2face for
text-to-face synthesis. Through extensive experiments, the FTGAN shows its
superiority in boosting both generated images' quality and similarity to the
input descriptions. The proposed FTGAN outperforms the previous state of the
art, boosting the best reported Inception Score to 4.63 on the CUB dataset. On
SCU-text2face, the face images generated by our proposed FTGAN just based on
the input descriptions is of average 59% similarity to the ground-truth, which
set a baseline for text-to-face synthesis. | ['Yining Xu', 'Lingbo Qing', 'Xiaohai He', 'Xiaodong Luo', 'Xiang Chen'] | 2019-04-11 | null | null | null | null | ['text-to-face-generation'] | ['computer-vision'] | [ 3.87074828e-01 4.74806279e-01 1.48771197e-01 -4.01156604e-01
-7.89966404e-01 -3.04314673e-01 9.16730940e-01 -1.11728489e+00
4.17355299e-01 8.51252854e-01 3.22081864e-01 -5.80468029e-02
6.56136692e-01 -1.06228900e+00 -1.11036038e+00 -7.52800643e-01
7.28193760e-01 3.17836076e-01 -3.01266134e-01 -3.01125914e-01
-1.58469692e-01 4.10843581e-01 -1.48569679e+00 7.85048664e-01
7.29061246e-01 1.20619023e+00 -5.77717461e-02 4.85305220e-01
-8.39880183e-02 7.69011736e-01 -8.57288301e-01 -1.12954497e+00
4.45782244e-01 -1.04291677e+00 -3.60432476e-01 2.34556094e-01
8.08829010e-01 -6.83377266e-01 -5.66015661e-01 9.65154171e-01
8.00084531e-01 -3.22732240e-01 8.39193046e-01 -1.56629324e+00
-1.17049766e+00 4.34918851e-01 -5.05365312e-01 -5.58534801e-01
3.68979335e-01 5.55327117e-01 4.85180080e-01 -1.10571110e+00
7.90887237e-01 1.86876380e+00 3.81627470e-01 1.17946684e+00
-8.60575914e-01 -1.20541966e+00 -4.46236312e-01 -1.21547922e-01
-1.27991021e+00 -9.54921961e-01 6.91437721e-01 -4.00390744e-01
6.16529524e-01 1.44228980e-01 2.07867473e-01 1.50988126e+00
4.08161730e-01 4.90975589e-01 1.09781432e+00 -4.05515909e-01
-1.81766480e-01 2.26715595e-01 -9.68458235e-01 8.01417530e-01
1.94611996e-02 4.65816915e-01 -5.39758205e-01 3.28771740e-01
8.21281254e-01 -1.56702667e-01 -5.48172072e-02 4.08221930e-01
-8.87209892e-01 8.53319585e-01 3.08300823e-01 4.14173678e-02
-2.65418202e-01 3.21470499e-01 3.49507689e-01 3.12455028e-01
6.19976461e-01 1.05508998e-01 2.09302261e-01 3.81799132e-01
-9.61983263e-01 4.49111640e-01 4.57824469e-01 1.39804018e+00
5.35544276e-01 6.95818841e-01 -8.11992943e-01 6.66115582e-01
3.40170771e-01 1.21636474e+00 3.62519503e-01 -7.11274087e-01
6.11889005e-01 4.53108579e-01 -2.08834529e-01 -9.48449075e-01
4.41788286e-01 -3.12488824e-02 -9.72718537e-01 2.76934355e-01
-4.56740819e-02 -5.24151266e-01 -1.15673113e+00 1.53457963e+00
2.09313720e-01 1.64868400e-01 1.74091905e-01 7.33289123e-01
1.35269368e+00 1.08108759e+00 -6.03869446e-02 -1.22676961e-01
1.16839397e+00 -1.08750904e+00 -9.84539568e-01 -2.38573954e-01
8.35591275e-03 -1.14124811e+00 8.09492648e-01 2.08138451e-02
-1.14287150e+00 -7.88536310e-01 -8.55753720e-01 8.29884633e-02
-1.47894502e-01 3.98936242e-01 1.74839541e-01 1.03028882e+00
-1.23536038e+00 2.51126289e-01 -5.45862131e-02 -2.23188341e-01
7.37525105e-01 1.16200067e-01 -6.07518375e-01 -2.26311877e-01
-1.13794649e+00 6.25123322e-01 2.85366714e-01 2.52397973e-02
-1.41631758e+00 -5.86842477e-01 -7.79998779e-01 -1.41607866e-01
2.44645879e-01 -7.93542087e-01 1.15450442e+00 -1.22331202e+00
-1.80051494e+00 9.12948430e-01 -2.34635875e-01 -2.56958693e-01
7.30260074e-01 1.67378187e-01 -6.21624768e-01 2.16507867e-01
2.89153844e-01 1.14991188e+00 1.50852561e+00 -1.34853590e+00
-4.52345580e-01 -4.03777063e-02 -2.36313060e-01 -1.00558944e-01
-1.65729567e-01 1.00380734e-01 -4.67496067e-01 -9.15003240e-01
-5.96897006e-01 -8.47360015e-01 3.40865701e-01 8.93722996e-02
-7.36486793e-01 -1.55007780e-01 1.27776158e+00 -7.72083104e-01
8.73524189e-01 -2.17287517e+00 -1.63884997e-01 -2.54757941e-01
-9.36055854e-02 4.81890023e-01 -4.42866653e-01 6.82581127e-01
-3.17157954e-01 3.07284623e-01 -9.69607383e-02 -4.19729948e-01
8.40839557e-03 -5.21049351e-02 -6.47648871e-01 1.64902389e-01
6.71066523e-01 1.30219686e+00 -5.49768746e-01 -6.37990832e-01
2.56698877e-01 5.78286469e-01 -5.68484366e-01 5.58466017e-01
-3.86607230e-01 6.16087615e-01 -4.35230553e-01 7.28038490e-01
8.47039044e-01 1.46858245e-01 -1.75343812e-01 -2.37732485e-01
3.19689989e-01 -4.04418647e-01 -5.21628082e-01 1.24141634e+00
-3.96005660e-01 7.49118805e-01 -2.44037390e-01 -5.62025428e-01
1.12736583e+00 6.60822690e-01 1.40435845e-01 -7.66522348e-01
3.18056792e-01 2.68521875e-01 -1.14218414e-01 -5.30657291e-01
1.74377963e-01 -5.72126508e-01 9.97932181e-02 2.99894780e-01
2.73227274e-01 -5.23866236e-01 1.35334611e-01 7.39714578e-02
6.66356921e-01 3.07083756e-01 -1.99800611e-01 1.16937561e-03
6.23794258e-01 -3.74223620e-01 1.40221402e-01 2.46358976e-01
2.44166866e-01 9.71936822e-01 4.63855028e-01 -2.84894317e-01
-1.42525721e+00 -8.46796334e-01 8.55043754e-02 6.67341232e-01
-2.92244047e-01 -1.87632561e-01 -1.38797355e+00 -9.65244055e-01
-1.17972746e-01 8.41089427e-01 -7.28116035e-01 -2.44050726e-01
-3.61339629e-01 -3.51025730e-01 9.69139218e-01 1.71666220e-01
9.74153519e-01 -1.49612403e+00 1.09575309e-01 -4.57475074e-02
-3.21520358e-01 -1.30948818e+00 -9.08166289e-01 -8.60659182e-01
-4.54785407e-01 -9.00948822e-01 -1.00724244e+00 -8.50355148e-01
9.16843772e-01 -7.31962696e-02 9.99663830e-01 8.02198425e-02
-2.66068667e-01 -1.07127354e-01 -4.43692416e-01 -7.28634477e-01
-1.04595029e+00 -3.72772872e-01 -1.99946553e-01 5.03983915e-01
-6.61838949e-02 -1.38256416e-01 -5.01323104e-01 4.95973080e-01
-1.10804427e+00 3.38109732e-01 6.79442644e-01 9.37462032e-01
3.27478498e-01 1.46810710e-01 9.22476232e-01 -8.62257302e-01
6.17669880e-01 -2.82595694e-01 -5.18657744e-01 2.17739284e-01
-4.79524553e-01 -2.09482703e-02 9.62811112e-01 -3.09873402e-01
-1.43237054e+00 -1.08732730e-01 -3.99093330e-01 -7.85975277e-01
-8.11807215e-02 -9.61706340e-02 -5.11351645e-01 -5.69163263e-02
6.19951785e-01 4.37375277e-01 1.42488301e-01 -1.24160107e-02
4.33291733e-01 9.12387371e-01 5.13929009e-01 -4.75800574e-01
1.11528945e+00 3.75974596e-01 1.66505277e-01 -6.88341022e-01
-5.98320544e-01 5.38865566e-01 -1.33565351e-01 -3.48682880e-01
1.03241765e+00 -1.03402221e+00 -6.61496103e-01 8.86900008e-01
-1.31099868e+00 -3.77130240e-01 -1.68456107e-01 -5.90196736e-02
-5.95222533e-01 6.60583377e-02 -3.88518125e-01 -7.33109534e-01
-7.06243455e-01 -1.35117400e+00 1.65696812e+00 1.57456279e-01
2.87733942e-01 -7.02543378e-01 -3.18705231e-01 6.21264577e-01
4.13348913e-01 6.16223037e-01 6.72815442e-01 -1.78130060e-01
-5.81204355e-01 -2.64639854e-01 -3.53224337e-01 7.19713569e-01
2.94111788e-01 1.40125930e-01 -1.15249598e+00 -3.19746315e-01
-3.44483964e-02 -4.38434541e-01 6.23820603e-01 2.16360837e-01
1.10826278e+00 -7.23955929e-01 -2.91246802e-01 6.83959961e-01
1.23716974e+00 4.05558228e-01 1.20483875e+00 -3.11325759e-01
7.32740402e-01 6.35160863e-01 5.63218355e-01 2.30657503e-01
1.76053438e-02 6.61825478e-01 4.38248158e-01 -2.00914681e-01
-8.70351613e-01 -7.56340086e-01 6.77626908e-01 3.41531366e-01
7.36877471e-02 -7.26207674e-01 -4.44882661e-01 3.04775417e-01
-1.28017581e+00 -1.24512994e+00 -5.05044572e-02 1.68895531e+00
8.61802697e-01 -1.66243792e-01 -1.24184258e-01 -6.45616874e-02
1.08938384e+00 1.09892160e-01 -4.52705830e-01 -4.42808092e-01
-1.07750826e-01 4.58954483e-01 2.21416920e-01 2.05008224e-01
-7.99521923e-01 1.20477760e+00 6.31686592e+00 1.28650784e+00
-1.37985682e+00 2.13447288e-01 1.12835681e+00 5.49750477e-02
-2.70389766e-01 -3.19221497e-01 -8.90453458e-01 7.19719946e-01
8.94523859e-01 -1.60737291e-01 4.70744133e-01 7.33920336e-01
2.40550756e-01 3.97983432e-01 -1.03698182e+00 1.05172729e+00
5.04707813e-01 -1.32801795e+00 7.27028906e-01 1.65064454e-01
1.25905490e+00 -6.66277945e-01 3.83513629e-01 2.54264265e-01
1.26231059e-01 -1.56707954e+00 8.65114093e-01 4.15340126e-01
1.76302171e+00 -9.48398888e-01 7.66510725e-01 1.27610505e-01
-1.10244250e+00 1.07177094e-01 -3.37081224e-01 4.04623121e-01
9.81213376e-02 4.42093700e-01 -1.11845303e+00 7.68736482e-01
3.61835688e-01 4.75554407e-01 -4.48634595e-01 2.28427932e-01
-3.90362591e-01 5.79944491e-01 2.44802728e-01 1.79635763e-01
1.12488233e-01 -2.03570239e-02 2.77236223e-01 1.00521243e+00
8.15917432e-01 1.23050861e-01 -1.89675242e-01 1.14945531e+00
-6.07504189e-01 9.66673866e-02 -1.05091417e+00 -4.38803852e-01
3.44862968e-01 1.21057355e+00 -2.40334302e-01 -5.54918408e-01
-3.19689065e-01 1.07633543e+00 -2.18682081e-01 3.06552470e-01
-1.08015597e+00 -5.09876430e-01 3.78319979e-01 3.93641621e-01
2.54438132e-01 4.38865125e-01 -2.88603697e-02 -9.01731074e-01
-1.27559558e-01 -1.30510378e+00 -9.29521546e-02 -1.16030085e+00
-1.23883796e+00 1.05291605e+00 -1.02323897e-01 -1.15499198e+00
-5.47532320e-01 -6.10796988e-01 -7.87947595e-01 1.13703716e+00
-1.30009937e+00 -1.78033340e+00 -4.17428225e-01 7.64986515e-01
7.56965160e-01 -7.77689159e-01 5.98067701e-01 3.04895699e-01
-4.43961799e-01 1.06280720e+00 -2.02357680e-01 3.71199161e-01
9.36564445e-01 -5.85992634e-01 8.14123750e-01 1.01356316e+00
-2.37283170e-01 2.89443433e-01 5.69833755e-01 -9.70341384e-01
-1.53703475e+00 -1.72905862e+00 5.93476057e-01 -3.78217816e-01
4.23559025e-02 -4.23209488e-01 -2.96119303e-01 6.44718230e-01
6.64528012e-01 -6.10440597e-02 1.85940638e-01 -9.89403367e-01
-3.19638669e-01 -2.66533017e-01 -1.67130518e+00 6.33267999e-01
1.04051554e+00 -4.70047504e-01 -2.08386764e-01 3.53426367e-01
8.62100542e-01 -4.68527168e-01 -6.73234403e-01 3.35783571e-01
5.23755431e-01 -9.56257045e-01 9.28397119e-01 -2.02201381e-01
1.23874903e+00 -1.52836367e-01 -1.32829800e-01 -1.26813972e+00
-1.02118561e-02 -8.00506115e-01 3.18522245e-01 1.55821514e+00
1.76768899e-01 -6.19850934e-01 6.73795760e-01 6.35302514e-02
-3.29238683e-01 -5.63136041e-01 -6.48975492e-01 -7.40613461e-01
1.35429099e-01 -1.91190746e-02 1.09186018e+00 6.91311061e-01
-6.59619153e-01 3.90098244e-01 -8.99322391e-01 -1.75474524e-01
6.65412664e-01 -1.26060620e-01 1.14884543e+00 -6.50237918e-01
-5.89685421e-03 -2.91236788e-01 -2.85314649e-01 -6.54674232e-01
4.08094645e-01 -1.15168738e+00 1.74412876e-01 -1.34181702e+00
1.84158176e-01 2.49344595e-02 5.60209453e-01 4.94969457e-01
-1.76803216e-01 7.32948244e-01 3.67407143e-01 -2.53448337e-01
1.96613178e-01 7.22995162e-01 2.06849217e+00 -3.09938133e-01
3.45202804e-01 -3.62831831e-01 -7.07079053e-01 1.97308928e-01
6.43596590e-01 -4.17950034e-01 -5.25161088e-01 -3.31804186e-01
-3.11857581e-01 2.50692725e-01 3.55064064e-01 -1.00754309e+00
-2.12478295e-01 -2.71042466e-01 6.14741921e-01 -4.38943595e-01
4.26943153e-01 -6.07126057e-01 6.77430987e-01 4.31894362e-01
-1.24916665e-01 -2.19732866e-01 2.19732210e-01 1.78266391e-01
-2.89442807e-01 -1.94686353e-01 1.01019406e+00 -1.86738014e-01
-2.91242033e-01 6.62019074e-01 1.61211248e-02 1.21964112e-01
1.24606931e+00 -1.26566052e-01 -5.00064969e-01 -8.09855878e-01
-1.43450737e-01 -1.86976567e-01 3.57306302e-01 6.48014188e-01
9.94019032e-01 -1.78137982e+00 -1.28755426e+00 5.80878854e-01
2.53361929e-02 -2.53927648e-01 2.96741873e-01 1.39102966e-01
-5.24306953e-01 7.14297771e-01 -4.47961330e-01 -4.19629425e-01
-1.21984124e+00 5.95812440e-01 3.23291808e-01 1.18824638e-01
-3.33476752e-01 8.06775570e-01 7.01358378e-01 -2.91727096e-01
-1.85161233e-01 1.91115975e-01 3.46356668e-02 -2.56372929e-01
6.20225549e-01 1.84098065e-01 -4.14059572e-02 -9.95870113e-01
2.91636232e-02 6.38227761e-01 2.43197918e-01 -1.90597713e-01
1.09510291e+00 3.37550163e-01 -1.92039430e-01 -3.94987226e-01
1.23577356e+00 6.19871616e-02 -1.21428692e+00 1.92356959e-01
-8.69673908e-01 -7.59568453e-01 -2.32319474e-01 -7.93768227e-01
-1.69315004e+00 9.96698558e-01 5.83691418e-01 -1.91578254e-01
1.30036867e+00 -2.00243533e-01 9.83828187e-01 4.27354947e-02
2.82349080e-01 -5.73829710e-01 5.08544266e-01 2.23983318e-01
1.47537136e+00 -1.13882577e+00 -2.76537031e-01 -5.17669797e-01
-8.57460916e-01 1.02683318e+00 8.50621760e-01 -1.40115976e-01
3.09334904e-01 2.35172942e-01 3.20113115e-02 -2.28275694e-02
-7.90757775e-01 1.72933757e-01 5.23503721e-01 7.97879457e-01
4.22372520e-01 -3.25570516e-02 -1.99443735e-02 3.24988067e-01
-5.02058685e-01 2.51305819e-01 3.35267872e-01 3.58872831e-01
-1.07394956e-01 -1.35483730e+00 -6.15254819e-01 5.00614226e-01
-6.64812446e-01 -2.74874181e-01 -4.89693314e-01 6.16882026e-01
2.73639828e-01 1.23820508e+00 -9.45307910e-02 -5.53925157e-01
3.05143684e-01 1.71279997e-01 5.61716259e-01 -5.75359881e-01
-5.75487375e-01 4.31232080e-02 1.42149508e-01 -4.71181631e-01
-1.88477337e-01 -2.62185097e-01 -9.14239287e-01 -6.79429770e-01
-2.76465237e-01 -1.19473912e-01 6.63439929e-01 7.03148842e-01
3.72236758e-01 6.29751861e-01 1.01025581e+00 -7.95707583e-01
-3.21378529e-01 -1.15163505e+00 -3.45920652e-01 5.70894718e-01
1.38073474e-01 -4.16948706e-01 -7.55933300e-02 5.27582347e-01] | [12.33445930480957, -0.16620200872421265] |
6e4a7293-4283-404e-9006-de23eebab6e9 | quantizable-transformers-removing-outliers-by | 2306.12929 | null | https://arxiv.org/abs/2306.12929v1 | https://arxiv.org/pdf/2306.12929v1.pdf | Quantizable Transformers: Removing Outliers by Helping Attention Heads Do Nothing | Transformer models have been widely adopted in various domains over the last years, and especially large language models have advanced the field of AI significantly. Due to their size, the capability of these networks has increased tremendously, but this has come at the cost of a significant increase in necessary compute. Quantization is one of the most effective ways to reduce the computational time and memory consumption of neural networks. Many studies have shown, however, that modern transformer models tend to learn strong outliers in their activations, making them difficult to quantize. To retain acceptable performance, the existence of these outliers requires activations to be in higher bitwidth or the use of different numeric formats, extra fine-tuning, or other workarounds. We show that strong outliers are related to very specific behavior of attention heads that try to learn a "no-op" or just a partial update of the residual. To achieve the exact zeros needed in the attention matrix for a no-update, the input to the softmax is pushed to be larger and larger during training, causing outliers in other parts of the network. Based on these observations, we propose two simple (independent) modifications to the attention mechanism - clipped softmax and gated attention. We empirically show that models pre-trained using our methods learn significantly smaller outliers while maintaining and sometimes even improving the floating-point task performance. This enables us to quantize transformers to full INT8 quantization of the activations without any additional effort. We demonstrate the effectiveness of our methods on both language models (BERT, OPT) and vision transformers. | ['Tijmen Blankevoort', 'Markus Nagel', 'Yelysei Bondarenko'] | 2023-06-22 | null | null | null | null | ['quantization'] | ['methodology'] | [ 4.11557630e-02 8.03866461e-02 1.08202726e-01 -4.90898907e-01
-6.74422681e-01 -2.00403169e-01 3.04783911e-01 1.91091418e-01
-6.86586320e-01 5.98118842e-01 -1.69673890e-01 -3.26368093e-01
1.06447026e-01 -6.38389707e-01 -9.82406259e-01 -6.75425589e-01
-2.83299237e-02 4.65178758e-01 3.87529433e-01 -2.49802917e-01
2.80057997e-01 4.47305173e-01 -1.51798451e+00 1.82727903e-01
8.40603054e-01 1.20470130e+00 2.85883248e-01 5.74725747e-01
-2.03675345e-01 7.76388884e-01 -8.33362758e-01 -4.49052364e-01
3.85490149e-01 -1.02750100e-01 -4.41891551e-01 -2.24635422e-01
5.78942180e-01 -3.06320786e-01 -2.94501811e-01 1.23410392e+00
5.19118845e-01 2.44985700e-01 3.96259665e-01 -1.09884894e+00
-6.51666462e-01 9.87551749e-01 -5.09754181e-01 1.89550683e-01
-2.19934687e-01 2.31291816e-01 9.62079823e-01 -9.75928187e-01
1.68624192e-01 1.30266154e+00 7.75460124e-01 2.47792408e-01
-1.19664502e+00 -7.97166169e-01 2.46229872e-01 1.81901738e-01
-1.60264683e+00 -6.40984714e-01 5.34688652e-01 -6.03680275e-02
1.40293157e+00 1.95387095e-01 5.02101719e-01 8.66365016e-01
2.45819807e-01 4.46016967e-01 5.88007271e-01 -3.45764339e-01
3.85162115e-01 2.71943331e-01 -1.36493385e-01 7.24773347e-01
2.97125727e-01 -3.10720354e-01 -4.89240438e-01 -6.83594421e-02
8.04219604e-01 8.97165462e-02 -2.47577384e-01 -5.79276495e-02
-7.88510799e-01 8.17082763e-01 5.91615081e-01 4.37186569e-01
-3.68674189e-01 5.61599314e-01 3.28608721e-01 3.99019808e-01
3.50170612e-01 6.19877875e-01 -6.15318596e-01 -3.73620331e-01
-1.10106242e+00 -3.62175144e-02 5.84333003e-01 6.47674501e-01
7.78103650e-01 3.27671319e-01 -1.03795484e-01 8.46252561e-01
-6.07707463e-02 3.90934139e-01 8.12920272e-01 -7.13759780e-01
7.21934497e-01 6.83999300e-01 -3.53055447e-02 -1.00948465e+00
-4.64649469e-01 -5.94035089e-01 -9.35358346e-01 3.37001741e-01
6.59712613e-01 -4.36667539e-02 -1.21893954e+00 1.78310204e+00
-1.56409889e-01 1.25258863e-01 -3.00771981e-01 8.16415787e-01
1.27381936e-01 7.05673933e-01 -1.48726860e-02 9.97686833e-02
9.59040880e-01 -6.48961961e-01 -4.66216743e-01 -4.82686073e-01
8.04891288e-01 -7.26437509e-01 1.49576414e+00 7.09710419e-01
-1.38780212e+00 -5.73641002e-01 -1.10776138e+00 -2.56214231e-01
-3.82624656e-01 1.19177081e-01 5.73025048e-01 5.68262160e-01
-9.39188004e-01 8.77648234e-01 -1.12995887e+00 4.06383500e-02
4.25306350e-01 7.79995739e-01 -2.44027540e-01 1.88883081e-01
-9.97702539e-01 9.75000203e-01 3.84238869e-01 3.36061388e-01
-4.01547194e-01 -5.64423859e-01 -7.16990650e-01 6.07774258e-01
4.15055394e-01 -2.96454579e-01 1.08853364e+00 -1.20910120e+00
-1.43069959e+00 4.07263815e-01 -3.05179566e-01 -6.87661290e-01
3.89403015e-01 -2.54300475e-01 -3.37796733e-02 -2.19058752e-01
-2.57610351e-01 7.23828733e-01 1.10585773e+00 -6.22019887e-01
-5.24470210e-01 -3.57574224e-01 3.21507044e-02 1.61913671e-02
-8.18179607e-01 -7.97490254e-02 -6.21922135e-01 -7.13527560e-01
1.67110220e-01 -9.22252774e-01 -2.53551036e-01 -2.07764655e-01
-3.42501372e-01 -2.25013241e-01 4.39048946e-01 -4.02633816e-01
1.38683724e+00 -2.33195186e+00 2.96850149e-02 4.70693022e-01
2.08915710e-01 2.79361427e-01 -9.84529257e-02 -4.02462147e-02
-1.13636225e-01 2.74352103e-01 -1.16285063e-01 -6.70726001e-01
-5.14409840e-02 4.75551337e-01 -4.94006634e-01 4.15917933e-01
2.52966583e-01 7.39575624e-01 -6.07620358e-01 -1.71038136e-01
-2.58322340e-02 5.23267269e-01 -9.08677280e-01 -1.80919692e-02
-1.00331962e-01 -5.66279776e-02 -4.11883509e-03 3.63424093e-01
4.39238518e-01 -2.71716297e-01 -1.25160754e-01 -1.16127282e-01
-8.17088559e-02 6.05107427e-01 -1.41343117e+00 1.38069224e+00
-4.85722274e-01 6.55308962e-01 -1.15706034e-01 -1.13206041e+00
5.98384917e-01 2.05633491e-01 1.60145715e-01 -8.39952886e-01
3.25597644e-01 2.68975407e-01 3.12168926e-01 -1.74785733e-01
5.54403007e-01 6.32049814e-02 6.66862652e-02 2.01349467e-01
-5.17763719e-02 -1.92596659e-01 3.26222509e-01 -3.58051038e-04
9.20634270e-01 -2.66853482e-01 -1.07911505e-01 -9.35940887e-04
2.98123568e-01 -3.58845651e-01 5.68103850e-01 9.15480852e-01
1.35157526e-01 5.15946686e-01 6.88436568e-01 -2.95398951e-01
-1.12668765e+00 -8.51965427e-01 4.70145047e-02 1.38627255e+00
-3.18259478e-01 -5.00343502e-01 -7.88833857e-01 -2.66545415e-01
4.13346849e-02 6.83438063e-01 -5.54271460e-01 -5.23039699e-01
-7.88675368e-01 -7.14096129e-01 7.66660154e-01 7.97763228e-01
4.00852501e-01 -1.00562382e+00 -6.61030591e-01 3.67386609e-01
1.55995116e-01 -9.47320700e-01 -4.39744085e-01 8.45107496e-01
-1.01172972e+00 -4.64273721e-01 -5.82823813e-01 -6.06988132e-01
8.74103725e-01 -1.07237406e-01 8.51853669e-01 2.90917128e-01
-1.39883876e-01 -2.29369015e-01 8.92755464e-02 -4.94447917e-01
-1.95361748e-01 3.18566829e-01 2.98427314e-01 -2.20077440e-01
2.12167963e-01 -5.99219918e-01 -2.98524290e-01 1.11328699e-01
-9.43412542e-01 -2.58306801e-01 6.82637215e-01 1.01635957e+00
4.15996104e-01 2.37611905e-01 2.46141002e-01 -5.58127642e-01
6.78803086e-01 -8.12348723e-02 -8.95259559e-01 -6.24689125e-02
-5.85192680e-01 6.29521489e-01 1.10356808e+00 -8.01464915e-01
-5.14128387e-01 -7.64195770e-02 -2.85598844e-01 -7.44816422e-01
1.92241296e-01 5.24354458e-01 4.76092994e-02 -1.07675426e-01
6.91712201e-01 -1.57415643e-02 -2.57704288e-01 -4.00908798e-01
2.33134329e-01 4.68127728e-01 4.95653540e-01 -3.49386632e-01
7.79371083e-01 2.61142731e-01 -2.99480021e-01 -8.05490792e-01
-6.62557721e-01 -1.25053793e-01 -3.22071195e-01 2.68720895e-01
4.82141465e-01 -7.25578606e-01 -7.66498506e-01 3.79402131e-01
-1.01870537e+00 -4.90983665e-01 -3.73548269e-01 3.35679859e-01
-1.41247183e-01 1.09575778e-01 -6.70991540e-01 -6.52310908e-01
-2.40662262e-01 -1.40323925e+00 7.77465403e-01 1.47554323e-01
-3.40811878e-01 -7.51456916e-01 -3.80086333e-01 -9.19024423e-02
6.88773096e-01 -4.36038584e-01 9.91456032e-01 -5.51095665e-01
-5.66477239e-01 -1.07998803e-01 -1.86323047e-01 5.88967502e-01
-5.79541996e-02 -9.96717997e-03 -1.05737138e+00 -2.69147277e-01
1.66700725e-02 -2.24080443e-01 9.27291453e-01 3.90748024e-01
1.44640076e+00 -3.83231074e-01 -1.19608253e-01 7.95849860e-01
1.13637364e+00 1.36289731e-01 6.08336270e-01 2.77939737e-01
5.45711935e-01 3.66666615e-02 1.33536294e-01 5.00795722e-01
8.60386901e-03 6.05819285e-01 5.27794182e-01 -1.44220531e-01
1.77559927e-01 -5.74848391e-02 5.40183723e-01 7.36442447e-01
-5.81206158e-02 -9.64714959e-02 -8.47847581e-01 5.26435316e-01
-1.69350004e+00 -8.79178047e-01 1.89703494e-01 2.56917429e+00
9.75369751e-01 8.47516000e-01 -2.37934425e-01 4.03948873e-01
2.27615118e-01 -1.42332897e-01 -5.75698555e-01 -8.04595172e-01
1.12495082e-03 6.48343861e-01 8.28233719e-01 6.63590789e-01
-7.62418509e-01 1.13061833e+00 6.31656504e+00 7.64996529e-01
-1.79548609e+00 -1.13062441e-01 6.08829439e-01 -4.47581261e-01
-1.36225522e-01 -1.64317355e-01 -8.48944843e-01 5.26494920e-01
1.09992313e+00 1.24576002e-01 7.92962968e-01 9.82999325e-01
2.14608803e-01 -1.84801579e-01 -1.05674267e+00 1.08545840e+00
-6.23564161e-02 -1.13722301e+00 7.01180249e-02 -4.01811749e-02
5.03872752e-01 3.26222748e-01 2.74097562e-01 4.96650338e-01
1.31898746e-01 -1.18629014e+00 8.65069509e-01 2.79212505e-01
6.19101882e-01 -8.62642825e-01 7.14537799e-01 3.26673090e-01
-9.16906595e-01 -3.32156658e-01 -8.02984893e-01 -2.88948268e-01
-1.41386911e-01 8.36046338e-01 -9.35054600e-01 -1.43430844e-01
8.62466395e-01 1.37666613e-01 -5.96774459e-01 8.87328029e-01
-2.51952231e-01 7.25086749e-01 -8.50968540e-01 -8.73721614e-02
4.20179963e-01 1.72749069e-02 1.70164838e-01 9.15043056e-01
4.04860348e-01 -2.85961539e-01 -3.35159153e-02 8.71002376e-01
-2.60707587e-01 -6.33465573e-02 -2.34506980e-01 -1.49616763e-01
4.68054861e-01 9.90064621e-01 -5.19285619e-01 -5.74218333e-01
-3.73095423e-01 1.03361464e+00 6.06629372e-01 2.98116416e-01
-1.10105932e+00 -5.74849486e-01 6.94033980e-01 2.37818897e-01
5.40404022e-01 -3.56314480e-01 -5.20292163e-01 -1.09910023e+00
1.89944237e-01 -8.85517597e-01 8.66217390e-02 -7.03822970e-01
-8.30119610e-01 5.16143501e-01 -3.55415970e-01 -6.86495006e-01
-4.38412279e-01 -5.72072566e-01 -3.73274863e-01 9.22699869e-01
-1.35704243e+00 -4.60230589e-01 -1.11219279e-01 6.44109964e-01
3.32807302e-01 1.02046028e-01 6.93453848e-01 5.93852937e-01
-6.50235176e-01 9.74646151e-01 1.27915647e-02 1.50499463e-01
8.58882904e-01 -1.13181627e+00 4.50919211e-01 8.41488242e-01
2.58333027e-01 9.49959338e-01 7.23631918e-01 -3.64423454e-01
-1.24204147e+00 -8.11989665e-01 9.99785125e-01 -2.40898237e-01
6.84270203e-01 -5.87686658e-01 -1.11953938e+00 9.15623903e-01
-1.15339838e-01 -1.62295625e-01 1.33603081e-01 1.65948883e-01
-3.50762367e-01 -3.38730693e-01 -7.71419704e-01 7.85001993e-01
6.52345240e-01 -5.01540422e-01 -5.22216141e-01 7.22088069e-02
5.02295375e-01 -4.85296160e-01 -5.10640979e-01 1.39778554e-01
4.32569504e-01 -1.04945433e+00 8.76913190e-01 -3.67982566e-01
1.28865167e-01 -1.72883481e-01 2.08603013e-02 -1.21124506e+00
-3.00006330e-01 -5.74966192e-01 1.83388554e-02 8.35569501e-01
5.50306797e-01 -7.69258261e-01 8.19585562e-01 7.24641919e-01
-2.02091813e-01 -7.60237396e-01 -1.07237852e+00 -6.32180035e-01
8.18572566e-02 -7.12714970e-01 3.70773315e-01 7.39090025e-01
-1.89956147e-02 2.64703423e-01 -2.68087894e-01 2.48295978e-01
1.02956571e-01 -3.20059001e-01 6.69083893e-01 -1.09668183e+00
-4.48972195e-01 -7.19123483e-01 -3.97622764e-01 -1.20788598e+00
-1.79779053e-01 -6.82282031e-01 1.12451091e-01 -1.06891370e+00
-2.30801597e-01 -5.92054665e-01 -3.74300927e-01 9.61594284e-01
-2.97253132e-01 4.49696481e-01 1.82096496e-01 4.33859229e-02
-2.82686234e-01 3.57152939e-01 6.22920692e-01 -1.41856566e-01
-4.10200626e-01 -1.22758627e-01 -5.74365079e-01 9.86801267e-01
7.48915553e-01 -6.06925428e-01 -3.64649922e-01 -6.90721214e-01
7.15721130e-01 -4.22496527e-01 2.73797750e-01 -1.30810797e+00
3.10348481e-01 1.34865344e-01 5.36660552e-01 -3.72698992e-01
5.39392710e-01 -9.59965825e-01 -2.09546596e-01 5.57564020e-01
-2.76216418e-01 2.98170209e-01 3.68803829e-01 2.38552894e-02
-1.12019330e-01 -3.22785616e-01 8.50435436e-01 -8.14908147e-02
-3.55842948e-01 1.56755999e-01 -4.02724862e-01 -1.45383449e-02
5.63244283e-01 -2.88270801e-01 8.08736011e-02 -4.01300102e-01
-5.44447541e-01 4.44496870e-02 4.11390901e-01 2.78657913e-01
3.30456406e-01 -8.92736495e-01 -2.64472574e-01 5.14547646e-01
-3.21562648e-01 1.77804112e-01 -9.18941200e-02 9.94301081e-01
-3.44435275e-01 4.37514126e-01 -8.22625384e-02 -5.74877262e-01
-1.04104698e+00 4.64890569e-01 3.68605345e-01 -3.88352484e-01
-5.44724882e-01 1.04633570e+00 -5.92166446e-02 -1.05707027e-01
6.74529433e-01 -1.12411046e+00 1.33887529e-01 8.85908827e-02
5.22798121e-01 1.35118440e-01 4.32876140e-01 -9.64258909e-02
-3.00487787e-01 4.22567576e-01 -2.07019478e-01 -9.23699513e-03
1.27518177e+00 1.72660857e-01 -2.65250616e-02 4.23913002e-01
9.92157459e-01 2.46072680e-01 -1.16714704e+00 -1.09171040e-01
-5.81599362e-02 -2.40418017e-01 1.70964554e-01 -5.12039244e-01
-1.11323965e+00 1.19576001e+00 4.90472496e-01 2.48108149e-01
1.10733593e+00 -3.33010733e-01 8.25829923e-01 7.06097543e-01
2.30702102e-01 -1.25767958e+00 -4.10852693e-02 7.79153228e-01
6.63958907e-01 -1.01938188e+00 -1.12076052e-01 7.49335364e-02
-3.84356797e-01 1.09194696e+00 6.11754954e-01 -1.77649423e-01
3.49563956e-01 7.43342221e-01 -1.31316464e-02 1.99605376e-01
-7.44320214e-01 1.42455623e-02 1.03378564e-01 1.08768448e-01
3.65797490e-01 -3.46222520e-01 8.47745314e-02 4.03300583e-01
-3.88852626e-01 -7.37516209e-02 3.35702449e-01 7.19004691e-01
-6.62988603e-01 -9.47267771e-01 -4.49046493e-01 4.80936229e-01
-6.10831559e-01 -5.39843261e-01 -6.73339441e-02 6.17129803e-01
1.40414760e-01 6.45446658e-01 3.87679964e-01 -9.60936323e-02
3.41254056e-01 3.12845856e-01 3.83828074e-01 -3.77507538e-01
-7.44465172e-01 6.33028522e-02 -2.74510503e-01 -7.30356812e-01
1.96756944e-01 -1.56893745e-01 -1.59328902e+00 -3.77329886e-01
-4.91820902e-01 2.03488525e-02 6.98566914e-01 1.03540456e+00
3.53842646e-01 6.46868348e-01 1.31032765e-01 -9.43628252e-01
-9.34773266e-01 -1.08068717e+00 -2.87165791e-01 2.96723157e-01
3.28395158e-01 -5.44699430e-01 -4.97585773e-01 -9.78173465e-02] | [8.568939208984375, 3.2344563007354736] |
ea8e7d60-3e5a-4f9c-9807-290d62d24bf6 | meta-analysis-of-transfer-learning-for | 2306.11714 | null | https://arxiv.org/abs/2306.11714v1 | https://arxiv.org/pdf/2306.11714v1.pdf | Meta-Analysis of Transfer Learning for Segmentation of Brain Lesions | A major challenge in stroke research and stroke recovery predictions is the determination of a stroke lesion's extent and its impact on relevant brain systems. Manual segmentation of stroke lesions from 3D magnetic resonance (MR) imaging volumes, the current gold standard, is not only very time-consuming, but its accuracy highly depends on the operator's experience. As a result, there is a need for a fully automated segmentation method that can efficiently and objectively measure lesion extent and the impact of each lesion to predict impairment and recovery potential which might be beneficial for clinical, translational, and research settings. We have implemented and tested a fully automatic method for stroke lesion segmentation which was developed using eight different 2D-model architectures trained via transfer learning (TL) and mixed data approaches. Additionally, the final prediction was made using a novel ensemble method involving stacking and agreement window. Our novel method was evaluated in a novel in-house dataset containing 22 T1w brain MR images, which were challenging in various perspectives, but mostly because they included T1w MR images from the subacute (which typically less well defined T1 lesions) and chronic stroke phase (which typically means well defined T1-lesions). Cross-validation results indicate that our new method can efficiently and automatically segment lesions fast and with high accuracy compared to ground truth. In addition to segmentation, we provide lesion volume and weighted lesion load of relevant brain systems based on the lesions' overlap with a canonical structural motor system that stretches from the cortical motor region to the lowest end of the brain stem. | ['Gottfried Schlaug', 'Sirisha Nouduri', 'Aleksei Rutkovskii', 'Anant Shinde', 'Advait Gosai', 'Sovesh Mohapatra'] | 2023-06-20 | null | null | null | null | ['lesion-segmentation'] | ['medical'] | [ 2.27448002e-01 -2.53992379e-01 -2.32258201e-01 -1.06952138e-01
-1.05764520e+00 -6.22005761e-01 5.34253061e-01 9.95582864e-02
-6.22041941e-01 7.22326219e-01 6.98816895e-01 -4.71168548e-01
-3.47287744e-01 -5.82223833e-01 -3.11597586e-01 -5.66388011e-01
-1.98418051e-01 9.32721794e-01 7.26028323e-01 -7.21545815e-02
4.13602978e-01 9.66994226e-01 -9.43928123e-01 5.29890239e-01
1.17177725e+00 5.39578438e-01 6.19694471e-01 5.27429342e-01
-7.60985836e-02 5.85852742e-01 -2.66944408e-01 8.45335722e-02
1.60497248e-01 -4.54978734e-01 -8.83419394e-01 -1.62021145e-01
1.87500462e-01 -5.73353708e-01 -2.86913842e-01 6.65936410e-01
9.93292034e-01 3.59323509e-02 1.05042708e+00 -5.27955472e-01
-1.61705762e-02 4.12798136e-01 -4.31191802e-01 6.66729867e-01
-2.10733518e-01 2.67382473e-01 2.83963859e-01 -7.21187651e-01
9.39191103e-01 6.42095625e-01 6.81341767e-01 3.68319273e-01
-1.08501554e+00 -5.01297414e-01 -8.43310729e-02 7.65357971e-01
-8.14800858e-01 -5.75600788e-02 3.96725804e-01 -1.10632908e+00
9.48030829e-01 1.95459083e-01 9.36817825e-01 8.97280395e-01
5.11772692e-01 6.81779623e-01 1.47273219e+00 -1.07690670e-01
2.45299071e-01 -4.24010932e-01 4.35533524e-01 1.54360145e-01
7.80910701e-02 5.64989708e-02 -3.80626065e-03 4.93291882e-04
6.93237424e-01 1.18668608e-01 -6.43184721e-01 -4.48984414e-01
-1.31522262e+00 7.06024587e-01 5.56345284e-01 5.90465367e-01
-7.44797647e-01 -3.39359850e-01 5.11518657e-01 -2.14999497e-01
2.78392971e-01 3.01186629e-02 -3.05732369e-01 -1.25949696e-01
-1.37316847e+00 -1.76023441e-05 3.08539033e-01 2.11424872e-01
1.44287214e-01 -2.69691825e-01 -6.16271973e-01 1.03539491e+00
-1.42210200e-01 2.43691787e-01 9.30049002e-01 -6.70668125e-01
4.63852555e-01 8.30037773e-01 -1.30750686e-01 -3.39849859e-01
-8.60921800e-01 -6.16635263e-01 -8.81974876e-01 7.76351810e-01
7.63158202e-01 -2.59781897e-01 -1.20496178e+00 1.38463259e+00
1.77045260e-02 3.03054880e-02 -3.10766071e-01 1.28526020e+00
6.18161142e-01 3.38717662e-02 3.03960174e-01 -1.22486152e-01
1.11050558e+00 -7.58869529e-01 -2.42430985e-01 -2.58152485e-01
9.94539201e-01 -4.49660420e-01 9.32892919e-01 1.05031312e-01
-1.07788837e+00 1.37249893e-02 -7.98078358e-01 7.43695721e-02
-2.78969049e-01 3.80709976e-01 1.31419241e-01 6.35405898e-01
-8.73031974e-01 6.21933877e-01 -1.06784785e+00 -4.93969381e-01
7.34420598e-01 2.56667167e-01 -5.89366078e-01 -3.64300981e-02
-9.23305333e-01 1.91674614e+00 2.73148060e-01 1.19259924e-01
-6.44153833e-01 -1.03488564e+00 -2.43540704e-01 -1.79236323e-01
-3.18675563e-02 -5.76167047e-01 7.51683533e-01 -4.30345625e-01
-1.27431715e+00 9.19275343e-01 -2.02395961e-01 -6.51139438e-01
9.93333519e-01 -1.32732287e-01 -1.49670362e-01 4.78947192e-01
4.00978327e-02 3.07844847e-01 2.36931264e-01 -1.06078684e+00
-3.89918864e-01 -8.80497277e-01 -5.19307852e-01 2.36243099e-01
2.56771505e-01 3.13352615e-01 3.09938133e-01 -5.87125301e-01
2.43492931e-01 -7.67787516e-01 -3.66026103e-01 -7.87185803e-02
-1.09585673e-02 1.20173305e-01 6.99030280e-01 -1.44948959e+00
8.51676702e-01 -1.59803045e+00 1.77063555e-01 3.41850638e-01
5.16864061e-01 6.54276788e-01 2.84247875e-01 1.13642827e-01
-3.09886187e-01 3.74309085e-02 -6.41945422e-01 3.94291788e-01
-3.23688060e-01 -1.79792553e-01 -1.71616554e-01 6.77636385e-01
-2.21413732e-01 1.17323387e+00 -5.13159156e-01 -4.66319054e-01
5.11453092e-01 4.62656587e-01 -3.41416597e-01 2.26665974e-01
4.22039360e-01 7.86864519e-01 -2.58760065e-01 1.56818733e-01
4.25198615e-01 2.68053412e-01 -4.96678166e-02 -2.72363156e-01
-3.58732849e-01 -1.26162067e-01 -7.75249779e-01 1.61540878e+00
-2.07374200e-01 5.96536040e-01 -4.27779295e-02 -1.46605933e+00
8.74531329e-01 3.80019069e-01 9.45776820e-01 -5.52853823e-01
3.83879304e-01 5.01329064e-01 4.76642430e-01 -7.08284736e-01
-4.43593591e-01 -3.03957283e-01 5.12920320e-01 6.47579014e-01
-1.56435847e-01 -1.19618014e-01 2.26015240e-01 -7.03174397e-02
1.20883036e+00 1.56154796e-01 1.29608586e-01 -4.80277121e-01
5.13920546e-01 3.00032735e-01 3.69182110e-01 6.86807394e-01
-6.53108478e-01 9.83268738e-01 5.09818196e-01 -2.22984508e-01
-9.13750648e-01 -1.28256238e+00 -3.99795204e-01 7.01518595e-01
-1.58457890e-01 2.08094552e-01 -1.06775999e+00 -4.87349927e-01
-2.43124917e-01 7.44503021e-01 -5.51080048e-01 -1.66484579e-01
-1.05545235e+00 -1.07204401e+00 4.07592654e-01 7.10049331e-01
5.03501236e-01 -1.09065807e+00 -9.34867442e-01 3.01807016e-01
-4.56507146e-01 -8.43946755e-01 -4.10016000e-01 2.73505580e-02
-1.05833602e+00 -1.39814818e+00 -1.45969558e+00 -8.30182731e-01
3.81585270e-01 9.55749080e-02 4.83539432e-01 -5.08311726e-02
-5.87966979e-01 1.09933443e-01 -3.69379312e-01 -2.51208961e-01
-2.94132710e-01 3.67534049e-02 -3.31528187e-01 -1.56143233e-01
1.03827007e-01 -7.19097853e-01 -1.00544941e+00 2.59281695e-01
-6.62423491e-01 1.54445902e-01 1.04846299e+00 6.08908236e-01
6.43132448e-01 -5.85487664e-01 8.73607755e-01 -4.97714370e-01
7.10158944e-01 -5.10343194e-01 6.22561527e-03 5.88918030e-01
-5.67343175e-01 -1.57694504e-01 2.51525998e-01 -4.44505215e-01
-9.11231458e-01 5.89689575e-02 -4.50761355e-02 4.00843397e-02
-4.43753600e-01 4.39596474e-01 -1.22091779e-02 -5.03888056e-02
8.96776915e-01 2.19893381e-01 5.39101183e-01 -4.02962953e-01
2.12938100e-01 7.71806598e-01 8.40523958e-01 -4.63112772e-01
2.39040017e-01 4.38305914e-01 1.01418674e-01 -7.37678826e-01
-2.49078199e-01 -6.55093193e-01 -1.37727463e+00 -5.64535260e-01
1.01008630e+00 -3.85890365e-01 5.37711009e-03 6.35442793e-01
-9.46956336e-01 -7.95405507e-01 -1.04020417e-01 1.00101399e+00
-7.13659465e-01 6.07637882e-01 -4.54033792e-01 -2.73529738e-01
-9.10771728e-01 -1.59082460e+00 5.44156492e-01 -1.24351971e-01
-2.30245709e-01 -7.87972867e-01 1.77313358e-01 4.88566607e-01
8.69554996e-01 6.25852764e-01 1.43113542e+00 -7.03612804e-01
-1.12468243e-01 -4.35151130e-01 -5.38909793e-01 1.86158419e-01
-1.19765652e-02 -4.37977493e-01 -1.66261852e-01 -8.12465474e-02
-1.03229709e-01 2.20771972e-02 7.93242812e-01 9.25005674e-01
7.18439698e-01 4.87521857e-01 -3.32561642e-01 2.38190144e-01
1.15560269e+00 2.19288796e-01 7.62017071e-01 4.99160409e-01
4.51405406e-01 7.51328349e-01 7.80114159e-03 -9.19535235e-02
2.62459546e-01 7.27939546e-01 9.07361433e-02 -1.61217123e-01
-9.28759217e-01 3.77450109e-01 1.80677682e-01 4.25068170e-01
-5.43972492e-01 3.50429147e-01 -1.16333473e+00 5.78417122e-01
-1.71138108e+00 -1.03340745e+00 -7.75201857e-01 2.31871772e+00
7.19836652e-01 4.32161056e-03 4.52292681e-01 1.49400413e-01
9.42268908e-01 -3.80854964e-01 -6.65291071e-01 -9.08052921e-02
-1.58274457e-01 4.53607887e-01 5.67430317e-01 5.81679940e-01
-7.16105402e-01 6.77856266e-01 6.68959141e+00 4.46624756e-01
-1.33432603e+00 5.29902339e-01 3.41518968e-01 -2.29355276e-01
2.68007219e-01 -1.23892598e-01 -2.46322677e-01 4.57866818e-01
7.89606810e-01 -2.25045770e-01 3.43399405e-01 3.17094713e-01
6.81680977e-01 -2.96614796e-01 -5.57119071e-01 4.65666443e-01
-2.05743298e-01 -1.20798242e+00 -2.38001689e-01 -1.70570072e-02
4.05714482e-01 5.96092880e-01 -5.46679556e-01 1.03952512e-01
-9.49777663e-02 -8.98205221e-01 5.88547409e-01 8.52931559e-01
9.00965035e-01 -5.05757451e-01 7.81260371e-01 5.01640677e-01
-8.76890600e-01 4.74641547e-02 9.88856032e-02 2.39453480e-01
5.23035109e-01 4.52066749e-01 -8.35931003e-01 1.00741901e-01
6.59345984e-01 4.15839791e-01 -5.44965506e-01 1.61478400e+00
-2.18968436e-01 6.42779529e-01 -2.03466892e-01 3.72190028e-01
4.98252399e-02 -3.51395011e-01 7.02852070e-01 1.25037766e+00
3.25221837e-01 3.52070451e-01 1.45967705e-02 8.51345420e-01
6.17944181e-01 4.15651858e-01 -1.67771041e-01 5.90022862e-01
6.95447996e-03 1.16849184e+00 -1.11506617e+00 -3.28778923e-01
-2.48660713e-01 5.41095197e-01 3.67090642e-01 4.60336417e-01
-5.09779155e-01 -2.23458335e-01 9.60139781e-02 5.55816948e-01
-1.56077757e-01 -2.56828249e-01 -8.24348092e-01 -1.04088318e+00
1.66899070e-01 -4.22168404e-01 3.41073275e-01 -6.80750728e-01
-9.72138822e-01 5.74601471e-01 1.65684059e-01 -9.96013105e-01
-2.86849797e-01 -6.49450839e-01 -9.38184321e-01 1.28228366e+00
-1.19880009e+00 -8.60667348e-01 -4.56443220e-01 5.83735228e-01
2.86712229e-01 -1.89009547e-01 7.78586030e-01 2.60493100e-01
-6.27358139e-01 1.76923141e-01 1.19138248e-01 1.37423083e-01
5.52481413e-01 -1.17033470e+00 -2.19849020e-01 1.03244233e+00
-4.75653350e-01 1.99998498e-01 4.35989529e-01 -8.94653916e-01
-6.72176063e-01 -1.04094994e+00 5.96832991e-01 -2.73816854e-01
6.30492926e-01 2.68579781e-01 -1.07726085e+00 4.96874750e-01
-1.82180434e-01 -8.97453278e-02 6.49497986e-01 -5.17567933e-01
3.24313901e-02 2.71862358e-01 -1.22952378e+00 5.90177059e-01
9.48235929e-01 -2.09412828e-01 -1.07968819e+00 4.32502955e-01
-2.79565662e-01 -3.85348260e-01 -1.05188501e+00 7.41795063e-01
7.38711178e-01 -9.69268680e-01 1.08956254e+00 -7.09076643e-01
2.76929855e-01 -4.42613475e-02 2.08042234e-01 -1.41975307e+00
-3.50601822e-01 8.45664293e-02 1.14345945e-01 5.67340136e-01
2.29907051e-01 -7.02798545e-01 8.35406899e-01 9.43064868e-01
-6.42332196e-01 -9.19683635e-01 -1.03104353e+00 -5.92581987e-01
7.83063710e-01 -3.50678712e-01 2.17669919e-01 4.03316289e-01
1.80751443e-01 -1.60142198e-01 2.50111282e-01 -6.01590760e-02
5.54328918e-01 1.61898509e-01 2.14111924e-01 -1.35151625e+00
2.73765773e-01 -1.23860216e+00 -4.35131997e-01 -3.96762460e-01
2.93749601e-01 -1.66052830e+00 -7.41710812e-02 -2.27951169e+00
3.85864556e-01 -2.65061468e-01 -2.48497874e-01 3.35886627e-01
-6.29641786e-02 1.19714513e-01 -3.96997891e-02 5.30191779e-01
3.28240752e-01 2.22921923e-01 1.41213965e+00 4.47498076e-02
-3.53702426e-01 1.21435501e-01 -2.87959665e-01 6.35066807e-01
1.07314253e+00 -2.73298562e-01 -2.44470820e-01 -3.05429399e-01
-7.19667375e-01 2.49076471e-01 6.33721352e-01 -9.94523704e-01
2.51141727e-01 -1.74538180e-01 4.90716487e-01 -6.38507187e-01
-3.30950432e-02 -5.85188389e-01 -1.29938573e-02 7.68917263e-01
-2.85196096e-01 -2.72090733e-01 -7.89434314e-02 -4.44429778e-02
1.57156691e-01 -3.79950702e-01 1.05338907e+00 -1.58018202e-01
-5.46670675e-01 4.54665601e-01 -9.51074421e-01 1.06152527e-01
1.16378260e+00 -3.83979052e-01 -3.49306494e-01 -9.30682570e-02
-1.20889556e+00 -4.01297957e-03 5.99361062e-02 4.05888706e-01
6.33178651e-01 -1.05502868e+00 -1.08339155e+00 -1.47721723e-01
-2.05464825e-01 -3.50639820e-01 4.79118764e-01 1.66759741e+00
-7.37879336e-01 5.24184763e-01 -7.65216351e-01 -5.84968805e-01
-1.01893973e+00 6.59753531e-02 7.67736137e-01 -3.86425972e-01
-1.33724141e+00 1.28934368e-01 -1.86298758e-01 -2.90205956e-01
1.50903150e-01 -2.74590880e-01 -6.73738658e-01 9.68266428e-02
6.09768987e-01 7.57002056e-01 3.87776554e-01 -1.07153988e+00
-2.90463477e-01 6.30352259e-01 5.43223433e-02 -2.04831243e-01
1.27470064e+00 -2.74633206e-02 -1.95636317e-01 -3.31565589e-02
1.07933748e+00 -5.53472102e-01 -1.07385957e+00 -1.15275159e-01
1.45699218e-01 -1.03423046e-02 5.40243387e-01 -1.26099002e+00
-1.22031474e+00 1.10278654e+00 1.19137096e+00 -3.84775639e-01
9.00774300e-01 -9.98863503e-02 1.05496573e+00 -6.91857338e-02
4.12298799e-01 -8.70979249e-01 -4.48750883e-01 1.18134290e-01
1.18854523e+00 -9.67162669e-01 -2.28842810e-01 -1.31977320e-01
-7.03349113e-01 1.30085266e+00 3.10855716e-01 -3.06637377e-01
7.93734610e-01 8.18747953e-02 2.84510434e-01 -1.93856910e-01
3.29214334e-02 -3.14371198e-01 5.39997876e-01 7.97010005e-01
3.70367080e-01 3.44348669e-01 -9.81449068e-01 9.69795763e-01
-2.32545465e-01 2.37768590e-01 3.93583745e-01 6.83526397e-01
-6.35491252e-01 -1.13665295e+00 -1.87358171e-01 1.13427758e+00
-3.07955593e-01 1.92869648e-01 -2.68809050e-01 6.48938954e-01
2.32153893e-01 7.16478169e-01 -4.80316550e-01 -8.79824013e-02
5.53885877e-01 3.19516540e-01 5.72974026e-01 -4.53581452e-01
-6.78261936e-01 -1.00755960e-01 -1.81836888e-01 -3.56445700e-01
-2.03854069e-01 -1.05078483e+00 -1.72385180e+00 1.44093931e-01
-1.19278029e-01 -9.82105210e-02 7.78219163e-01 1.48908710e+00
2.70498604e-01 3.77443820e-01 2.10384727e-01 -1.00435543e+00
-3.35339427e-01 -1.32712269e+00 -5.24337292e-01 3.57197434e-01
-1.56683177e-01 -9.27254558e-01 -1.78070039e-01 -3.43682989e-02] | [14.219627380371094, -2.0566253662109375] |
a70e7113-0f4c-48a4-9543-4b5821fd3571 | fair-multilingual-vandalism-detection-system | 2306.01650 | null | https://arxiv.org/abs/2306.01650v1 | https://arxiv.org/pdf/2306.01650v1.pdf | Fair multilingual vandalism detection system for Wikipedia | This paper presents a novel design of the system aimed at supporting the Wikipedia community in addressing vandalism on the platform. To achieve this, we collected a massive dataset of 47 languages, and applied advanced filtering and feature engineering techniques, including multilingual masked language modeling to build the training dataset from human-generated data. The performance of the system was evaluated through comparison with the one used in production in Wikipedia, known as ORES. Our research results in a significant increase in the number of languages covered, making Wikipedia patrolling more efficient to a wider range of communities. Furthermore, our model outperforms ORES, ensuring that the results provided are not only more accurate but also less biased against certain groups of contributors. | ['Diego Saez-Trumper', 'Ricardo Baeza-Yates', 'Ai-Jou Chou', 'Muniza Aslam', 'Mykola Trokhymovych'] | 2023-06-02 | null | null | null | null | ['feature-engineering'] | ['methodology'] | [-6.61947131e-01 7.36948550e-02 -3.42669152e-02 1.37185663e-01
-6.85126424e-01 -6.72756314e-01 6.76813006e-01 7.40835309e-01
-7.24720418e-01 7.54975557e-01 4.19624001e-01 -2.05747381e-01
1.29649565e-01 -9.93941844e-01 -6.10902190e-01 2.69242913e-01
1.19720429e-01 2.98494220e-01 2.84718871e-01 -6.84727967e-01
6.83901072e-01 1.90754548e-01 -1.60926449e+00 5.14062881e-01
1.68684554e+00 7.67519847e-02 2.58847177e-01 2.28442639e-01
-7.53604650e-01 9.55594599e-01 -7.62428701e-01 -8.20599973e-01
3.66324544e-01 1.34826943e-01 -7.51798868e-01 -2.30252236e-01
7.96515763e-01 1.73560306e-02 -1.30295902e-02 1.34859192e+00
4.48030263e-01 -2.96790719e-01 4.85072762e-01 -7.59133220e-01
-8.29356909e-01 9.42290306e-01 -5.12298405e-01 -1.82278231e-01
7.95668721e-01 -8.24609101e-02 7.61178851e-01 -1.10675335e+00
1.12110686e+00 1.23762619e+00 7.64682353e-01 2.20559746e-01
-9.12942886e-01 -7.04653442e-01 1.95808448e-02 -2.89176069e-02
-1.64646935e+00 -3.84349287e-01 3.77847582e-01 -9.81720209e-01
6.60592794e-01 2.64651895e-01 6.50648952e-01 6.42360866e-01
1.76996216e-02 5.09257019e-01 1.27336037e+00 -9.15316999e-01
-7.06735849e-02 1.09661412e+00 4.34800506e-01 8.87737274e-01
8.96918416e-01 -6.06085300e-01 -7.19326138e-01 -4.32484508e-01
1.25092119e-01 -4.97189164e-01 -3.35299373e-01 -4.10411000e-01
-8.83478880e-01 9.59504008e-01 9.40825045e-02 4.09085095e-01
-2.74027497e-01 -6.01893961e-01 5.50268114e-01 4.51896667e-01
6.46094859e-01 6.72841728e-01 -3.01619917e-01 8.86545703e-02
-1.09244645e+00 3.05881947e-01 1.18634868e+00 9.37957764e-01
8.82102370e-01 -3.17959845e-01 3.18169184e-02 7.22572505e-01
5.90883374e-01 6.11184537e-01 3.30312878e-01 -6.53599024e-01
9.42307174e-01 1.32043672e+00 3.66359293e-01 -1.17732882e+00
-1.52005062e-01 -4.28846955e-01 -2.95208871e-01 3.28485757e-01
5.56226015e-01 9.93524306e-03 -5.96560359e-01 1.32246351e+00
3.18600655e-01 -8.74386668e-01 -2.36376394e-02 4.66957629e-01
6.19709492e-01 4.71916556e-01 2.01906547e-01 -7.65029490e-02
1.20806313e+00 -7.42812276e-01 -9.02270198e-01 -1.08623356e-02
8.69254649e-01 -1.25288904e+00 1.17765093e+00 6.03684723e-01
-7.16080725e-01 -5.88067114e-01 -9.59532738e-01 -7.50439689e-02
-8.48845422e-01 3.25713426e-01 4.68369424e-01 1.03021371e+00
-9.87760961e-01 3.68303776e-01 -1.70798421e-01 -7.91220427e-01
1.08598009e-01 -3.40634696e-02 -5.50602376e-01 -8.31414238e-02
-1.29866648e+00 1.02172470e+00 4.11445320e-01 -2.24593639e-01
-4.64383274e-01 -6.53550148e-01 -6.57221198e-01 -3.77282023e-01
3.58767033e-01 -2.20958263e-01 7.26903021e-01 -5.71828187e-01
-8.56411636e-01 9.91466761e-01 1.43214567e-02 -3.07007849e-01
1.05816865e+00 -5.16316950e-01 -5.55245697e-01 -2.31104344e-01
7.65303493e-01 1.14791781e-01 3.23642999e-01 -1.23030102e+00
-8.60361099e-01 -3.81384194e-01 9.17661488e-02 7.50323012e-02
-8.78952384e-01 4.32441652e-01 -6.13938570e-01 -4.10795808e-01
-3.28702241e-01 -5.31701863e-01 -2.55975038e-01 -2.20679432e-01
-5.93760371e-01 -1.03375502e-01 5.25285363e-01 -1.48555028e+00
1.73092902e+00 -1.92996657e+00 -3.67516950e-02 5.25114596e-01
5.11717081e-01 5.32391429e-01 -1.96828563e-02 9.97912824e-01
5.65756261e-01 6.12261057e-01 1.68967411e-01 -2.04392985e-01
-1.05579942e-01 -2.54838526e-01 -1.78690981e-02 4.21510816e-01
-3.39376181e-01 2.48826340e-01 -7.57495165e-01 -7.45498717e-01
-8.03803727e-02 2.32472181e-01 -5.42256117e-01 -1.02122677e-02
-2.42744952e-01 6.85387328e-02 -3.39975566e-01 6.26592457e-01
8.75526667e-01 3.82746965e-01 2.53141612e-01 1.74790874e-01
-7.84681499e-01 1.10448770e-01 -1.61889613e+00 1.47094297e+00
-4.55314964e-01 4.65136170e-01 1.81142136e-01 1.61093771e-02
1.43716645e+00 -7.06209764e-02 2.11315081e-01 -7.77647316e-01
-2.28222743e-01 5.19330740e-01 -2.38380566e-01 -8.68309438e-01
1.00023818e+00 7.60535598e-01 -1.35548234e-01 4.49802816e-01
-8.58068988e-02 3.28419894e-01 9.63829875e-01 9.03964162e-01
5.15647888e-01 1.88099116e-01 1.66329771e-01 -5.55917084e-01
8.89803767e-01 5.17020822e-01 3.52503002e-01 8.61186445e-01
-2.15279646e-02 -3.69680412e-02 3.34991455e-01 -2.34733880e-01
-1.24353266e+00 -8.39794159e-01 3.83486561e-02 8.25784802e-01
-6.66420981e-02 -9.07821655e-01 -9.12274480e-01 -8.14789236e-01
2.49737099e-01 7.55069435e-01 -4.65793192e-01 2.13091061e-01
-3.04631919e-01 -5.11357009e-01 6.47218704e-01 -2.79889703e-01
5.70148051e-01 -7.78829038e-01 -1.28542423e-01 3.08686435e-01
-4.29230660e-01 -7.89217591e-01 -4.19896841e-01 -3.13794017e-01
-4.61830944e-01 -1.46519005e+00 -4.07049596e-01 -9.25583422e-01
6.24761879e-01 1.60232291e-01 1.03083873e+00 1.96252286e-01
-3.96301448e-01 1.80428803e-01 -4.60173011e-01 -7.70503461e-01
-1.01918685e+00 3.20952833e-01 6.86725900e-02 -2.11043760e-01
6.58842862e-01 1.41552046e-01 -4.02702540e-02 -1.08839475e-01
-7.13398278e-01 -2.27892056e-01 4.97216433e-01 3.26318055e-01
-2.61825100e-02 7.20247924e-02 5.48231840e-01 -1.41537619e+00
1.25401270e+00 -4.21223521e-01 -6.26528323e-01 5.30022442e-01
-9.60904241e-01 -5.24751805e-02 6.31615281e-01 -1.49879515e-01
-1.27119911e+00 -6.71255141e-02 7.77290538e-02 4.73355711e-01
1.28684655e-01 8.94704938e-01 -1.81427956e-01 -3.44528049e-01
1.04449606e+00 5.68126515e-02 2.27799177e-01 -1.01284397e+00
5.16575634e-01 1.31851053e+00 1.56860590e-01 -3.81131053e-01
8.81923556e-01 9.45520028e-02 -7.81957507e-01 -9.10452604e-01
-4.75222379e-01 -7.36345470e-01 -9.05100107e-01 -6.03254318e-01
3.46779078e-01 -1.18723869e+00 -8.64405483e-02 5.81405103e-01
-1.13816929e+00 2.36881390e-01 -1.25487834e-01 3.86280745e-01
3.50662649e-01 7.22440183e-01 -4.95352805e-01 -9.69328165e-01
-5.61359167e-01 -7.77756155e-01 4.03845161e-01 1.76945522e-01
-2.00829312e-01 -9.22440410e-01 5.01020551e-01 7.26112127e-01
4.79410917e-01 1.81812823e-01 8.51927519e-01 -7.54466176e-01
-2.96069801e-01 -5.13576329e-01 -2.04756454e-01 3.74906540e-01
-1.67786032e-01 5.87489128e-01 -4.53110695e-01 -2.89506525e-01
-4.50530410e-01 -2.05186948e-01 5.42572916e-01 -4.40378934e-01
2.82500029e-01 -1.90059558e-01 -1.59868851e-01 4.04875204e-02
1.68299806e+00 -3.41242790e-01 7.30989337e-01 1.11998367e+00
7.02356994e-01 9.40685391e-01 8.43995214e-01 4.31535691e-01
7.21642733e-01 5.44151902e-01 2.74509311e-01 -3.07871699e-02
-3.64913642e-01 -6.69076145e-01 4.09534097e-01 1.18509316e+00
-4.48617563e-02 6.67553842e-02 -1.12601304e+00 7.26552665e-01
-1.75061166e+00 -7.67598927e-01 -8.36431324e-01 2.32915115e+00
9.34522629e-01 1.23591453e-01 2.67433405e-01 -2.12105021e-01
8.89288187e-01 -2.91704297e-01 7.27073774e-02 -6.22683525e-01
-2.82523900e-01 -1.33252844e-01 6.32119715e-01 5.90413392e-01
-8.18521321e-01 9.96967196e-01 6.63567734e+00 5.95629454e-01
-7.88055599e-01 9.74126384e-02 -2.18466818e-01 2.17516065e-01
-4.45906848e-01 1.77570522e-01 -9.76542056e-01 5.34309089e-01
8.51256490e-01 -7.58341849e-01 2.14938775e-01 7.62507975e-01
4.85857487e-01 -2.01344326e-01 -2.74225086e-01 4.25086141e-01
3.72704536e-01 -1.36441016e+00 2.29684487e-01 3.13929975e-01
7.17176735e-01 2.18789905e-01 -4.24235076e-01 5.48024297e-01
5.42111099e-01 -5.17211080e-01 1.22007227e+00 6.87751532e-01
4.87877101e-01 -7.04320073e-01 7.42538273e-01 6.52682960e-01
-7.96819925e-01 -7.89286792e-02 -4.45543170e-01 -1.71018571e-01
-1.03639714e-01 6.93390548e-01 -8.40772748e-01 8.08504641e-01
7.54128098e-01 4.35191095e-01 -1.29063165e+00 1.40630484e+00
-3.12125534e-01 3.82801294e-01 -9.10855234e-02 -3.64024282e-01
-7.26358220e-02 -4.14215922e-01 4.16373610e-01 1.56530833e+00
1.66751325e-01 -9.28651154e-01 4.54926491e-01 5.26386738e-01
-2.41188526e-01 1.15480280e+00 -6.93422735e-01 5.33883497e-02
7.96514273e-01 1.29320765e+00 -1.63903847e-01 -2.92498946e-01
-5.36533296e-01 6.12146258e-01 7.04910696e-01 9.68754664e-02
-5.54811001e-01 -7.33690262e-01 1.38429001e-01 5.69357991e-01
-1.02108702e-01 -2.61289060e-01 -4.38011549e-02 -1.22783720e+00
1.91252440e-01 -1.11678886e+00 4.60274547e-01 -4.32918519e-01
-1.14349616e+00 6.53151035e-01 -2.47449711e-01 -1.04748070e+00
1.46904096e-01 -3.84635359e-01 -2.83699155e-01 1.27606320e+00
-1.28618193e+00 -1.51160848e+00 -2.49529526e-01 4.91289020e-01
1.48257628e-01 -5.08174658e-01 6.73503399e-01 7.52526224e-01
-6.29640877e-01 4.68442589e-01 3.57714385e-01 1.88224375e-01
1.20022333e+00 -1.24692547e+00 1.17471427e-01 1.29272819e+00
1.69283941e-01 1.14397287e+00 7.64403820e-01 -1.13712752e+00
-1.25289881e+00 -1.22722614e+00 1.36606777e+00 -5.87619722e-01
8.93307924e-01 -7.13855326e-01 -7.90811360e-01 3.56336057e-01
3.65041733e-01 -7.60981679e-01 6.78221941e-01 3.94374728e-01
-6.35416567e-01 7.28584379e-02 -1.30924094e+00 4.65234399e-01
7.26010084e-01 -6.05527163e-01 -7.05227137e-01 5.24909735e-01
2.31448337e-01 -6.43426031e-02 -1.03317070e+00 -2.12105125e-01
3.72913420e-01 -7.17815280e-01 4.57610488e-01 -5.52826583e-01
2.39543438e-01 -4.89987105e-01 4.45367992e-02 -1.24722958e+00
-3.49660784e-01 -3.21577311e-01 2.31935531e-02 1.79794741e+00
8.63282263e-01 -6.94413424e-01 6.12015069e-01 4.96898055e-01
1.35197461e-01 -5.36643118e-02 -6.93060338e-01 -7.10689306e-01
2.95943052e-01 -1.44517362e-01 1.60658076e-01 1.02379060e+00
1.51815832e-01 2.53684431e-01 -4.31506157e-01 1.01540960e-01
8.66954565e-01 -1.99561596e-01 1.28349984e+00 -1.50000870e+00
2.90177494e-01 -1.02099009e-01 -1.96407974e-01 -2.87477136e-01
-2.00824291e-01 -1.17173386e+00 -4.32139516e-01 -1.88032484e+00
5.38246393e-01 -3.62854242e-01 2.59473681e-01 2.73530245e-01
-2.28838176e-01 3.20598990e-01 2.45196879e-01 6.03707135e-01
-6.80005550e-01 -7.22898021e-02 8.04863214e-01 -1.05907016e-01
-4.17534083e-01 -3.41750503e-01 -1.17585480e+00 6.76914692e-01
6.82676554e-01 -5.57895362e-01 1.38507739e-01 -3.85648608e-01
6.25166357e-01 -7.89641500e-01 -8.45263451e-02 -1.08137703e+00
4.42599922e-01 1.88460588e-01 2.10402206e-01 -5.07959783e-01
-4.81985241e-01 -4.89512026e-01 1.22895926e-01 7.25818634e-01
-2.05961689e-01 -4.01274636e-02 1.00652538e-01 3.01505029e-01
-2.79552996e-01 -6.11783266e-01 4.16553617e-01 -2.16990024e-01
-1.14813089e+00 -1.32920951e-01 -6.26899004e-01 1.80188343e-01
1.02982557e+00 -1.09754205e-01 -6.41185522e-01 -1.15160465e-01
-6.04310453e-01 3.70768696e-01 7.74352312e-01 5.97492814e-01
4.47044522e-02 -9.34353173e-01 -1.16999149e+00 -4.19092923e-02
6.05852604e-01 -7.31870890e-01 4.51338068e-02 5.05324662e-01
-1.06008148e+00 3.03523540e-01 -3.09391469e-01 -1.06039606e-01
-1.34544265e+00 4.58628714e-01 1.34581374e-03 -5.23036003e-01
-5.28386354e-01 2.94602394e-01 -7.65914500e-01 -1.05029976e+00
2.38521516e-01 5.10119379e-01 -7.41751492e-01 3.10602307e-01
6.48657918e-01 5.47662973e-01 1.52911738e-01 -7.40642309e-01
-2.80510694e-01 3.24694902e-01 -4.61828083e-01 -5.09179644e-02
1.27890444e+00 -4.30219650e-01 -5.35203099e-01 3.11969399e-01
6.76962316e-01 1.27110338e+00 -5.12714684e-01 -2.00756311e-01
3.54654133e-01 -6.33046567e-01 -1.73939735e-01 -1.17628646e+00
-6.90802336e-01 3.40203643e-01 5.98101199e-01 3.84268731e-01
5.05182505e-01 -2.49711618e-01 2.66553253e-01 2.95231760e-01
8.29545319e-01 -1.58126879e+00 -3.36365491e-01 3.88014644e-01
8.46814871e-01 -1.21734047e+00 3.11961234e-01 -6.91460669e-01
-6.25063598e-01 1.22172976e+00 8.09599996e-01 1.86409175e-01
5.39342046e-01 2.00254306e-01 5.25945127e-01 -2.03116968e-01
-4.03343230e-01 -1.72165930e-01 5.12666628e-02 7.98687518e-01
8.05808604e-01 -6.05690889e-02 -1.07543266e+00 5.15283585e-01
-1.45297393e-01 8.74084756e-02 1.12620521e+00 9.02005911e-01
-7.03076363e-01 -1.36614442e+00 -6.72635853e-01 5.30286729e-01
-5.69847643e-01 -1.41909570e-01 -6.81250095e-01 1.24871027e+00
5.39478183e-01 9.72578287e-01 -4.68011588e-01 -3.02790314e-01
7.74909675e-01 -1.57278359e-01 8.84592999e-03 -7.95499921e-01
-9.81324315e-01 -1.77521080e-01 6.91787541e-01 -3.27550948e-01
-1.62460431e-01 -7.91997850e-01 -7.23271132e-01 -4.77313668e-01
-4.94521886e-01 6.51696682e-01 9.86096680e-01 4.44581300e-01
3.82170141e-01 1.12236619e-01 5.08694530e-01 -9.37561840e-02
-7.64642000e-01 -1.07716727e+00 -6.08221531e-01 7.42182195e-01
-3.44680160e-01 -2.32506439e-01 -2.59846747e-01 -4.80593294e-02] | [9.736517906188965, 9.806747436523438] |
acd5a114-ede7-43b4-8b1c-af03570c9256 | attention-over-parameters-for-dialogue | 2001.01871 | null | https://arxiv.org/abs/2001.01871v2 | https://arxiv.org/pdf/2001.01871v2.pdf | Attention over Parameters for Dialogue Systems | Dialogue systems require a great deal of different but complementary expertise to assist, inform, and entertain humans. For example, different domains (e.g., restaurant reservation, train ticket booking) of goal-oriented dialogue systems can be viewed as different skills, and so does ordinary chatting abilities of chit-chat dialogue systems. In this paper, we propose to learn a dialogue system that independently parameterizes different dialogue skills, and learns to select and combine each of them through Attention over Parameters (AoP). The experimental results show that this approach achieves competitive performance on a combined dataset of MultiWOZ, In-Car Assistant, and Persona-Chat. Finally, we demonstrate that each dialogue skill is effectively learned and can be combined with other skills to produce selective responses. | ['Chien-Sheng Wu', 'Pascale Fung', 'Jamin Shin', 'Andrea Madotto', 'Zhaojiang Lin'] | 2020-01-07 | null | null | null | null | ['goal-oriented-dialogue-systems'] | ['natural-language-processing'] | [ 3.49915959e-02 4.48810309e-01 1.58631012e-01 -5.85283160e-01
-7.68467963e-01 -8.12426925e-01 5.84662020e-01 -1.07174769e-01
-6.28952265e-01 1.05450833e+00 3.46327901e-01 -2.77221203e-01
-1.66960657e-02 -5.39155066e-01 5.06664217e-02 -4.99209881e-01
3.18566322e-01 1.26481962e+00 4.59530264e-01 -1.03748858e+00
2.96887666e-01 1.66463032e-01 -1.02882409e+00 3.40206563e-01
1.10154200e+00 3.80292833e-01 3.41761738e-01 1.05066788e+00
-2.92257428e-01 1.40072429e+00 -9.77478981e-01 -3.47112209e-01
-9.56892446e-02 -8.33947539e-01 -1.46304715e+00 1.40589550e-01
1.99145749e-01 -6.26990438e-01 -2.76645720e-01 6.11507237e-01
5.68791807e-01 7.72416770e-01 7.20811784e-01 -1.16515923e+00
-3.48792315e-01 7.56278694e-01 -1.57818664e-02 1.10692225e-01
6.19818807e-01 5.10518312e-01 1.06316495e+00 -5.14574587e-01
2.92085379e-01 1.55436552e+00 1.97716132e-01 1.16572011e+00
-1.24393690e+00 -3.16657692e-01 1.74420103e-01 -1.45232961e-01
-6.93036199e-01 -5.35648882e-01 4.10861582e-01 -1.76134333e-01
1.10203445e+00 2.87773758e-01 4.71058547e-01 1.00006354e+00
-2.55275100e-01 1.12596393e+00 1.08785498e+00 -2.90166050e-01
-4.17137258e-02 5.06290913e-01 3.45065206e-01 7.60705292e-01
-5.69755197e-01 -4.51343477e-01 -6.15154207e-01 -2.95726120e-01
7.31823564e-01 -2.72781432e-01 -2.41085619e-01 -4.47027124e-02
-1.06094432e+00 1.11607754e+00 -7.65368119e-02 2.21858740e-01
-1.38856515e-01 -1.60617158e-01 4.39691722e-01 9.16946948e-01
2.19571501e-01 1.10749125e+00 -5.26997447e-01 -4.46283281e-01
1.24913253e-01 5.08487761e-01 1.55101442e+00 1.22748983e+00
5.45329452e-01 -1.33705959e-01 -4.27031070e-01 1.32305074e+00
1.90105122e-02 3.79627496e-01 3.69091302e-01 -1.28675914e+00
6.30563617e-01 5.36688805e-01 4.27380979e-01 -3.97070616e-01
-5.97562730e-01 4.40623820e-01 -4.85683769e-01 -4.37505320e-02
8.40404391e-01 -7.78069794e-01 -3.82348329e-01 1.57918882e+00
2.77722031e-01 -1.44818529e-01 3.27456295e-01 8.41062844e-01
1.00339401e+00 5.92878580e-01 2.26883635e-01 -1.73621904e-02
1.43923962e+00 -1.37023950e+00 -6.59701943e-01 -5.49600124e-01
6.69889748e-01 -5.17697394e-01 1.26621521e+00 4.01851237e-01
-1.55485535e+00 -3.92958790e-01 -5.44168055e-01 -1.86726227e-01
-4.05203626e-02 -2.37186685e-01 5.32550931e-01 1.85655922e-01
-9.52180743e-01 3.84896934e-01 -3.23725045e-01 -4.32526052e-01
-1.87968403e-01 2.65349835e-01 -1.85682267e-01 6.22402653e-02
-1.56125438e+00 1.39653265e+00 2.69653559e-01 -3.63095313e-01
-8.75735164e-01 -3.07731599e-01 -7.71949112e-01 2.69037664e-01
6.72306776e-01 -5.92049122e-01 2.03031278e+00 -6.54991269e-01
-2.08952022e+00 7.60677218e-01 2.14658096e-01 -3.83823097e-01
4.92040187e-01 -3.40652794e-01 1.11816265e-01 3.93323839e-01
-1.48392403e-02 7.59094775e-01 4.42901611e-01 -9.63814139e-01
-1.17591417e+00 -8.47036541e-02 7.83816159e-01 9.67043459e-01
-2.45112389e-01 2.98513293e-01 -3.64415169e-01 2.84765996e-02
-3.51809114e-01 -7.37031281e-01 -1.83488145e-01 -4.72574860e-01
-2.58065671e-01 -8.01962554e-01 4.65348333e-01 -4.40217942e-01
1.10985780e+00 -1.95896173e+00 5.34686029e-01 -9.24068764e-02
3.04480225e-01 2.17316806e-01 -1.92080513e-01 6.36467159e-01
6.27313912e-01 -7.81775936e-02 -1.83291752e-02 -1.96004570e-01
1.38495326e-01 2.38039881e-01 1.63229123e-01 -8.51246119e-02
2.55025268e-01 7.34995604e-01 -1.18818319e+00 -5.41390479e-01
7.55541399e-02 -9.46290717e-02 -5.53600252e-01 6.92616522e-01
-4.97644722e-01 7.72804379e-01 -7.84959197e-01 2.18332827e-01
-4.11754735e-02 -2.84148127e-01 3.29840928e-01 4.38347995e-01
1.16847582e-01 7.51133800e-01 -7.09819376e-01 1.47644496e+00
-7.49278486e-01 5.24365067e-01 4.57249969e-01 -9.01591182e-01
9.47942853e-01 5.20351291e-01 1.33650526e-01 -5.89230180e-01
2.16293573e-01 -3.79927233e-02 4.39495027e-01 -6.18496656e-01
7.05934882e-01 -4.52169329e-02 -4.25584167e-01 9.11245286e-01
3.44376236e-01 -5.15257776e-01 2.21794218e-01 3.65680367e-01
9.28613842e-01 -4.22102362e-01 2.75256336e-01 -1.19582459e-01
7.56677687e-01 3.33367169e-01 1.98931992e-01 1.04892492e+00
-4.03782785e-01 1.15454428e-01 7.73240507e-01 -2.06271991e-01
-8.33449364e-01 -5.94692647e-01 2.82301962e-01 1.98108757e+00
1.70824468e-01 -5.77358082e-02 -8.36840570e-01 -1.00133824e+00
-6.47425652e-03 9.13449049e-01 -1.92859799e-01 -1.68033153e-01
-9.93832827e-01 -1.25464156e-01 7.21953213e-01 3.97227913e-01
7.76891112e-01 -1.34609723e+00 -6.71316803e-01 4.64938790e-01
-5.18116117e-01 -9.73010600e-01 -7.89423347e-01 3.94268125e-01
-3.69333088e-01 -1.09650803e+00 -6.08912885e-01 -9.60399508e-01
2.94935852e-01 4.86108482e-01 1.48010170e+00 4.15518880e-01
2.92711854e-01 8.03850353e-01 -4.42698032e-01 -2.24459708e-01
-8.90044987e-01 5.28728127e-01 -1.70613434e-02 -3.45120937e-01
5.90087414e-01 -2.89044559e-01 -3.34334016e-01 7.98616409e-01
-3.29311758e-01 1.66383326e-01 2.09588528e-01 1.31309664e+00
-3.21599841e-01 -2.14634031e-01 9.66598690e-01 -1.29773855e+00
1.55756736e+00 -5.11876404e-01 -1.27401844e-01 4.96119946e-01
-2.52705336e-01 -9.11148414e-02 6.97278082e-01 -6.61393166e-01
-1.39297283e+00 -2.16384888e-01 -1.53467447e-01 1.51322365e-01
-5.35740018e-01 3.96348774e-01 -1.43734887e-01 1.09026782e-01
8.37098479e-01 2.02342182e-01 2.99282759e-01 -4.17582393e-01
3.65780026e-01 1.02787936e+00 3.87215942e-01 -1.07547164e+00
3.79747272e-01 -1.88847125e-01 -7.59153247e-01 -7.54726648e-01
-6.31803751e-01 -5.68856776e-01 -2.95020789e-01 -2.65185982e-01
7.86515534e-01 -7.59488881e-01 -1.23191273e+00 5.74875712e-01
-1.15849519e+00 -1.06030536e+00 -3.82090844e-02 1.63412839e-01
-6.56334698e-01 9.93270874e-02 -9.68667030e-01 -1.02734208e+00
-2.13028893e-01 -1.09263003e+00 6.48427844e-01 6.90963268e-01
-5.39864242e-01 -1.07395148e+00 -1.88060269e-01 6.20484054e-01
5.67413270e-01 -5.75317502e-01 9.07202542e-01 -1.45658684e+00
-2.81008303e-01 4.42460440e-02 3.75247262e-02 2.23775536e-01
1.94004983e-01 -5.08177757e-01 -6.27027452e-01 -2.11064979e-01
-3.38514373e-02 -1.26944673e+00 3.75009477e-01 -3.14871706e-02
5.08119583e-01 -5.70204079e-01 -6.39806911e-02 -9.45357308e-02
3.32375348e-01 5.75477898e-01 2.09198594e-01 1.61798999e-01
4.08230454e-01 1.05754566e+00 1.01829815e+00 3.86851549e-01
6.83698773e-01 6.42434955e-01 -5.48032708e-02 2.01214617e-03
3.58777493e-01 -4.46794704e-02 2.39583760e-01 5.19888163e-01
-1.37194932e-01 -3.05062234e-01 -9.49210644e-01 5.29682815e-01
-2.07767987e+00 -9.36627030e-01 2.55214304e-01 1.70431256e+00
1.42449069e+00 7.51833096e-02 4.81541902e-01 -4.63216722e-01
5.85014522e-01 1.06042966e-01 -7.77195096e-01 -6.21253610e-01
1.48907870e-01 2.01562382e-02 1.48785055e-01 8.24014902e-01
-9.17384982e-01 1.32598341e+00 6.63259029e+00 5.80878675e-01
-6.76503003e-01 -3.38114495e-03 5.79105258e-01 2.81037465e-02
-1.53462425e-01 -2.69491017e-01 -8.47715855e-01 2.37279475e-01
8.01622748e-01 -4.02169317e-01 7.16854513e-01 9.01887417e-01
-1.37927681e-01 -2.64629126e-01 -1.26647067e+00 4.09719497e-01
3.67790163e-02 -7.92918265e-01 -2.88770705e-01 -3.57416689e-01
3.14050734e-01 -3.56564939e-01 -1.06868416e-01 8.53590250e-01
1.22657120e+00 -1.05242825e+00 1.40799776e-01 3.03021550e-01
4.09368783e-01 -7.30076253e-01 7.16643333e-01 9.95981693e-01
-6.10804558e-01 -1.89315632e-01 -1.72842115e-01 -8.69785622e-02
1.99546441e-01 -6.42933846e-01 -1.27822363e+00 1.73518494e-01
4.92704123e-01 6.26781583e-02 -2.76312027e-02 5.10002911e-01
-4.57838178e-01 3.88401896e-01 -1.10391535e-01 -6.60543442e-01
3.77917856e-01 -2.54018307e-01 4.26304758e-01 1.14020205e+00
-1.97485387e-01 8.80692363e-01 5.52889526e-01 5.86563647e-01
3.97467092e-02 -3.25267203e-02 -5.02761006e-01 -1.66616082e-01
8.04342866e-01 1.15511572e+00 -1.96149185e-01 -6.27010703e-01
-5.96260309e-01 9.51772809e-01 4.98323321e-01 3.45148444e-01
-5.66494107e-01 -4.44147646e-01 8.30398917e-01 -1.68719694e-01
-1.14534535e-01 -1.32007718e-01 1.02083325e-01 -1.06609631e+00
-6.17950261e-01 -1.54912817e+00 4.40897048e-01 -6.76488459e-01
-1.42585909e+00 5.92061996e-01 1.42639518e-01 -8.29341531e-01
-7.49360740e-01 -5.35719991e-01 -7.84127593e-01 1.13384724e+00
-1.19386506e+00 -8.40325058e-01 -1.58134490e-01 8.55636060e-01
1.11206317e+00 -5.89021444e-01 1.05941260e+00 -4.04624864e-02
-3.30001712e-01 5.79055071e-01 -2.71968424e-01 4.59705114e-01
1.11904216e+00 -1.64771068e+00 4.25763160e-01 8.44577402e-02
-3.91484767e-01 7.64831185e-01 7.44108856e-01 -3.55201811e-01
-1.20073903e+00 -5.54763198e-01 8.71388078e-01 -5.66936255e-01
6.57157063e-01 -1.73043728e-01 -1.06281042e+00 7.06248701e-01
7.25750566e-01 -8.38541508e-01 6.55259073e-01 4.96137559e-01
-2.99777724e-02 3.28991085e-01 -1.20112431e+00 8.98005545e-01
7.40038753e-01 -4.89735156e-01 -9.70766246e-01 5.25206268e-01
6.70929313e-01 -9.28918242e-01 -7.38396704e-01 -1.10295944e-01
3.33805799e-01 -8.94371688e-01 9.11578119e-01 -1.18965292e+00
3.93458128e-01 2.31407657e-01 3.81242037e-01 -1.65114605e+00
-1.86936289e-01 -7.53086805e-01 2.31040195e-02 1.08539927e+00
4.61857915e-01 -7.17336535e-01 5.81350029e-01 1.21395898e+00
-1.62794068e-01 -6.30500913e-01 -6.03876352e-01 -3.24114025e-01
2.51388252e-01 2.70507872e-01 6.56069934e-01 1.12736893e+00
7.37099588e-01 1.05460989e+00 -7.53304958e-01 -2.09807500e-01
4.37448919e-02 -4.31179293e-02 1.12923586e+00 -1.23508310e+00
-6.12028182e-01 -5.43894947e-01 4.24117208e-01 -1.79337716e+00
2.46318206e-01 -5.19454479e-01 5.10894775e-01 -1.33081305e+00
-1.27863482e-01 -7.14030683e-01 7.32290596e-02 6.19165957e-01
-5.03795028e-01 -4.10958707e-01 2.68342912e-01 -5.27530676e-03
-9.10754919e-01 3.69748116e-01 1.51233113e+00 -2.72198915e-01
-4.95270818e-01 4.35511649e-01 -9.82772827e-01 6.10607088e-01
9.93969202e-01 -1.18852310e-01 -7.34806955e-01 -3.65984410e-01
-1.43281519e-01 8.33835959e-01 -1.21267967e-01 -4.08142418e-01
6.01713836e-01 -5.30462503e-01 -1.58737466e-01 1.97288878e-02
4.96619135e-01 -5.34684360e-01 -7.54096091e-01 1.45595968e-01
-8.22582006e-01 -1.04848988e-01 9.61044338e-03 2.88811952e-01
-2.41880462e-01 -4.89691168e-01 7.45343566e-01 -6.51979208e-01
-6.24974966e-01 -1.13531295e-02 -9.33302224e-01 4.58115131e-01
7.64571548e-01 -5.64698130e-02 -6.67133868e-01 -1.08296406e+00
-5.84696472e-01 1.20841062e+00 1.17200583e-01 3.99006039e-01
4.63812560e-01 -8.45480084e-01 -9.13539648e-01 -8.93465355e-02
8.49222466e-02 3.04654181e-01 2.86462814e-01 5.66706836e-01
-1.92906693e-01 3.57706338e-01 -4.06029403e-01 -3.42944533e-01
-1.47582579e+00 6.60536764e-03 5.06163120e-01 -4.97813940e-01
-4.92985338e-01 1.12799394e+00 1.97195545e-01 -8.74180377e-01
6.07277930e-01 2.24540234e-02 -7.20852911e-01 9.74749401e-02
5.68043768e-01 1.07140422e-01 -4.31802243e-01 -2.13997364e-01
9.67697948e-02 -3.61926109e-02 -5.29265463e-01 -4.66033757e-01
9.98145223e-01 -2.39943802e-01 1.58946007e-01 4.66231257e-01
4.39837903e-01 -2.13249445e-01 -1.15801656e+00 -5.61211288e-01
5.64174429e-02 -2.39218563e-01 -6.80056810e-01 -1.03986168e+00
-4.94815677e-01 7.51274943e-01 -1.70779094e-01 7.54285336e-01
8.16880941e-01 1.12837724e-01 6.04070663e-01 1.16111243e+00
5.09579122e-01 -1.41259468e+00 4.26805854e-01 1.15645921e+00
9.07120705e-01 -1.39829111e+00 -4.09055084e-01 -1.70673162e-01
-1.47927701e+00 9.59573448e-01 1.37082982e+00 4.08696644e-02
1.16966979e-03 1.59010321e-01 3.06365490e-01 -1.43279567e-01
-1.29524732e+00 -2.63391763e-01 -8.03200901e-02 5.76602578e-01
6.14695609e-01 -6.76112920e-02 -3.08372200e-01 6.56951368e-01
-1.10279880e-01 -4.27966923e-01 5.03273785e-01 1.03171623e+00
-7.74189055e-01 -1.20845926e+00 -1.37976706e-01 3.76277804e-01
-1.60191849e-01 -8.65008775e-03 -8.49743307e-01 8.70004237e-01
-5.35880029e-01 1.31345809e+00 1.11640263e-02 -3.75375777e-01
6.88659012e-01 5.83468735e-01 2.90227592e-01 -1.01547885e+00
-1.34545016e+00 -7.10746273e-02 8.02320480e-01 -2.30323657e-01
-2.23029077e-01 -3.56417626e-01 -1.24365342e+00 -4.51211959e-01
-4.22033370e-01 7.00213492e-01 -5.69502227e-02 9.39273119e-01
-6.66400567e-02 5.03425300e-01 6.82596505e-01 -4.12020266e-01
-1.00565195e+00 -1.44527721e+00 -6.40398741e-01 4.35094327e-01
2.12068811e-01 -5.64920664e-01 -3.84715050e-01 -2.92576283e-01] | [12.851606369018555, 8.053685188293457] |
b59c7272-8102-4908-aee3-9e9b3ce4692a | multimodal-neural-databases | 2305.01447 | null | https://arxiv.org/abs/2305.01447v1 | https://arxiv.org/pdf/2305.01447v1.pdf | Multimodal Neural Databases | The rise in loosely-structured data available through text, images, and other modalities has called for new ways of querying them. Multimedia Information Retrieval has filled this gap and has witnessed exciting progress in recent years. Tasks such as search and retrieval of extensive multimedia archives have undergone massive performance improvements, driven to a large extent by recent developments in multimodal deep learning. However, methods in this field remain limited in the kinds of queries they support and, in particular, their inability to answer database-like queries. For this reason, inspired by recent work on neural databases, we propose a new framework, which we name Multimodal Neural Databases (MMNDBs). MMNDBs can answer complex database-like queries that involve reasoning over different input modalities, such as text and images, at scale. In this paper, we present the first architecture able to fulfill this set of requirements and test it with several baselines, showing the limitations of currently available models. The results show the potential of these new techniques to process unstructured data coming from different modalities, paving the way for future research in the area. Code to replicate the experiments will be released at https://github.com/GiovanniTRA/MultimodalNeuralDatabases | ['Fabrizio Silvestri', 'Alon Halevy', 'Emanuele Rodolà', 'Andrea Santilli', 'Giovanni Trappolini'] | 2023-05-02 | null | null | null | null | ['multimodal-deep-learning'] | ['natural-language-processing'] | [-1.38639733e-01 -2.11605072e-01 -9.68089104e-02 -4.27693248e-01
-9.99538004e-01 -7.66582012e-01 9.49449062e-01 3.48160774e-01
-7.96255887e-01 6.47172034e-01 3.72470021e-01 3.50708701e-02
-1.24978721e-01 -8.16862106e-01 -5.96007407e-01 -4.59835559e-01
9.07976851e-02 8.02098513e-01 4.93044764e-01 -4.48919207e-01
8.57820138e-02 6.37767851e-01 -2.01140857e+00 8.51643145e-01
1.60250023e-01 1.27995670e+00 2.66711980e-01 5.76927423e-01
-3.65749061e-01 8.17031801e-01 -2.83890188e-01 -4.62193698e-01
-1.36601344e-01 -4.74673472e-02 -1.03478134e+00 -1.31364852e-01
5.18813193e-01 -3.65677625e-01 -5.14824629e-01 7.21431136e-01
7.37962008e-01 2.43693158e-01 3.96830082e-01 -1.21230376e+00
-5.47782004e-01 3.57144594e-01 -7.83632398e-02 1.12478904e-01
6.46535456e-01 -1.51538104e-01 1.07598042e+00 -1.09681773e+00
8.76147151e-01 1.52760041e+00 3.03024441e-01 7.18177736e-01
-1.04063773e+00 -9.86686945e-02 3.73499072e-03 3.91588390e-01
-1.29613876e+00 -6.81014240e-01 3.93634647e-01 -1.01717643e-01
1.05279768e+00 4.89768296e-01 1.60545841e-01 1.30363739e+00
-3.92981887e-01 1.25645220e+00 9.47987795e-01 -4.54914480e-01
8.40405971e-02 2.58264631e-01 -6.94880122e-03 4.89011556e-01
-4.09928747e-02 -1.92260772e-01 -7.18088925e-01 -5.90634160e-02
4.40013230e-01 -1.14235438e-01 -2.06111893e-01 -4.29088473e-01
-1.41437769e+00 8.20198894e-01 3.52384508e-01 5.72725534e-01
-1.51815847e-01 5.76729178e-02 5.98353267e-01 4.54055667e-01
1.76023677e-01 2.19593614e-01 -3.05166841e-01 -1.64242178e-01
-8.56613696e-01 3.78753275e-01 1.10792804e+00 8.02213728e-01
7.12674916e-01 -4.50829744e-01 2.07486954e-02 1.17031622e+00
4.05205637e-01 4.55703825e-01 6.18916333e-01 -9.17416871e-01
6.17773831e-01 6.34917080e-01 1.15355335e-01 -1.08506262e+00
-5.25445461e-01 1.96425870e-01 -8.46705675e-01 -7.63517767e-02
4.28875744e-01 1.68105528e-01 -9.48168755e-01 1.64756370e+00
3.44748735e-01 -4.51590776e-01 7.67397583e-02 9.93169546e-01
1.08278918e+00 7.70923615e-01 1.94853172e-02 2.16917112e-01
1.40940869e+00 -7.07346976e-01 -5.96262872e-01 -1.17451608e-01
4.35846835e-01 -8.20849419e-01 9.95437026e-01 4.33267444e-01
-1.33496702e+00 -4.84010637e-01 -7.42794812e-01 -3.78043145e-01
-9.43104088e-01 -1.49863780e-01 5.04673183e-01 3.87070924e-01
-1.62161517e+00 2.20903620e-01 -8.57772827e-01 -8.30385268e-01
3.77241224e-01 4.69568551e-01 -7.20106900e-01 -2.69598544e-01
-1.42830539e+00 8.53291035e-01 5.51257014e-01 1.63616106e-01
-6.88752115e-01 -1.58748060e-01 -7.03441381e-01 2.26631705e-02
5.60019374e-01 -7.15666175e-01 1.39889705e+00 -7.10213065e-01
-1.05967391e+00 1.24237061e+00 -1.33380964e-01 -4.17088300e-01
5.38205206e-01 -3.05970982e-02 -3.94388348e-01 4.19088840e-01
-1.09559119e-01 1.17829096e+00 5.04244506e-01 -1.25504339e+00
-6.23453975e-01 -6.32007003e-01 3.49602312e-01 2.07814008e-01
-6.36196613e-01 2.43858486e-01 -9.91606236e-01 -4.22909319e-01
3.03591806e-02 -9.45420623e-01 -9.21932310e-02 8.40999484e-02
-3.26443285e-01 -4.72281963e-01 7.48560429e-01 -3.33807617e-01
1.04288578e+00 -2.13029170e+00 4.96520102e-01 8.89224038e-02
7.51701742e-02 2.24980980e-01 -3.26115549e-01 1.01134539e+00
2.16126412e-01 2.56427288e-01 -2.57014155e-01 -6.26532376e-01
2.69065201e-01 4.60964292e-01 -3.09351623e-01 3.42761055e-02
1.57253996e-01 9.54779685e-01 -5.42744637e-01 -6.76575840e-01
4.32935879e-02 6.54092193e-01 -1.36948794e-01 1.22793354e-01
-5.21982193e-01 1.65962785e-01 -4.31152523e-01 9.10612762e-01
3.64899635e-01 -3.03876936e-01 -6.32192120e-02 -3.00912440e-01
-6.08501956e-02 -6.09214231e-03 -1.25641215e+00 1.96087611e+00
-2.97948718e-01 6.68133736e-01 3.71073514e-01 -1.10746086e+00
5.33995271e-01 4.60319936e-01 5.23419082e-01 -1.04290915e+00
5.97062409e-02 3.22246581e-01 -3.92094940e-01 -7.50539422e-01
6.14821970e-01 -1.20086528e-01 -9.60085019e-02 4.47328776e-01
2.11278886e-01 -1.25480890e-01 5.97156167e-01 3.93630236e-01
8.84655178e-01 3.26455384e-02 -4.67874445e-02 3.06423634e-01
6.24498785e-01 2.79037077e-02 -1.97553933e-01 9.15034413e-01
-2.97259968e-02 7.66680360e-01 3.08445096e-01 -4.19960111e-01
-8.37974012e-01 -1.00206280e+00 -1.31949261e-01 1.38115728e+00
-3.37844305e-02 -3.45412046e-01 -5.17597258e-01 -4.25887883e-01
4.13585529e-02 1.40322834e-01 -4.87175137e-01 2.40646020e-01
-3.91600370e-01 -5.91992080e-01 6.72030985e-01 4.19401616e-01
4.21024680e-01 -1.38609195e+00 -5.71437955e-01 3.57046686e-02
-4.04175997e-01 -1.29252923e+00 6.95429593e-02 3.83702256e-02
-9.78846729e-01 -9.10470724e-01 -1.00167966e+00 -6.84487104e-01
2.46234328e-01 2.17193574e-01 1.35246086e+00 8.58103111e-03
-3.47276926e-01 1.05105209e+00 -4.41829830e-01 -4.43768591e-01
-3.70196998e-01 3.47890377e-01 -1.59579754e-01 1.44427285e-01
2.53048927e-01 -3.62369061e-01 -6.22094333e-01 2.80777276e-01
-1.79309571e+00 -1.56167150e-01 7.28450716e-01 5.46565771e-01
6.37596369e-01 -2.05066383e-01 6.19246900e-01 -7.63583839e-01
7.87378192e-01 -6.98774815e-01 -5.63170135e-01 3.22517365e-01
-2.31520221e-01 1.31453335e-01 1.59735620e-01 -2.10215941e-01
-9.67860699e-01 -4.51219939e-02 -3.34967524e-01 -2.72038996e-01
-6.10551775e-01 1.11220741e+00 -2.12273195e-01 1.12957843e-01
6.33856118e-01 8.76023397e-02 6.78372160e-02 -6.47464454e-01
4.61683154e-01 7.13522613e-01 4.43732381e-01 -5.26702523e-01
3.93464595e-01 7.63088703e-01 -4.59656529e-02 -8.81891429e-01
-5.72351694e-01 -6.29350185e-01 -4.95346069e-01 -2.35510722e-01
8.70751202e-01 -8.68182540e-01 -7.14491606e-01 5.60409844e-01
-1.11563981e+00 -1.17615283e-01 -1.45488754e-01 2.69553125e-01
-5.33756912e-01 3.82206947e-01 -7.03627706e-01 -6.15286827e-01
-8.12145993e-02 -1.01508868e+00 1.19160652e+00 1.08629137e-01
-7.89182037e-02 -1.10032082e+00 9.10594687e-02 6.22994006e-01
6.95748925e-01 6.93249851e-02 8.28533769e-01 -8.69176447e-01
-9.08927202e-01 -5.44335902e-01 -2.84146011e-01 3.90084654e-01
-1.39066264e-01 -3.04639131e-01 -1.08240759e+00 -2.33600304e-01
-1.18744157e-01 -9.50114667e-01 1.09140277e+00 -1.25559010e-02
1.04571557e+00 6.60625547e-02 -2.74182916e-01 1.65218860e-01
1.44138539e+00 -9.31502357e-02 6.28439069e-01 4.87136126e-01
4.78688687e-01 8.10518801e-01 3.31372619e-01 2.88799882e-01
6.14392221e-01 7.70547211e-01 7.70799398e-01 -1.06273949e-01
-8.32174048e-02 1.04595952e-01 1.23095497e-01 6.23091400e-01
-8.75428040e-03 -5.59761286e-01 -1.21548855e+00 6.69035614e-01
-1.97217727e+00 -1.02570510e+00 1.30134001e-01 2.10484767e+00
8.49799216e-01 -9.52625796e-02 1.34871006e-01 -1.49173871e-01
3.67997199e-01 2.37676665e-01 -5.03043115e-01 -2.35904366e-01
-4.16183710e-01 1.03132688e-01 -4.97996844e-02 3.55430007e-01
-1.18196738e+00 7.40427911e-01 5.66943121e+00 6.77274525e-01
-1.08813214e+00 -5.79097159e-02 5.39206982e-01 -2.15961531e-01
-3.92493933e-01 -3.01769435e-01 -8.26414466e-01 1.70446426e-01
1.08863103e+00 2.31854424e-01 4.51312333e-01 4.65458095e-01
-2.72971652e-02 -3.55401456e-01 -1.34249783e+00 1.22007859e+00
3.64694983e-01 -1.35700417e+00 3.51351261e-01 -6.71684742e-02
3.17248791e-01 5.76849401e-01 1.32267356e-01 3.43250424e-01
-9.42568630e-02 -1.00586665e+00 6.35406673e-01 7.43766427e-01
3.96635383e-01 -4.10494387e-01 6.92629278e-01 3.57337445e-01
-8.41958046e-01 -4.31312583e-02 -2.19848707e-01 2.83289969e-01
3.00123263e-02 2.09082276e-01 -5.93804061e-01 5.73279619e-01
8.63700807e-01 5.51801085e-01 -8.16092074e-01 1.11402524e+00
3.15700799e-01 1.96803719e-01 -5.08151054e-01 -2.11579464e-02
3.04369956e-01 5.97014353e-02 3.16320390e-01 1.31348217e+00
3.19954753e-01 -1.42863855e-01 4.43463475e-02 5.52600086e-01
-3.98044467e-01 1.99932486e-01 -7.13425219e-01 -3.24105263e-01
2.71406770e-01 1.26876891e+00 -6.71445608e-01 -1.91674352e-01
-8.02263260e-01 8.69310319e-01 4.34968024e-01 5.20191073e-01
-5.51786602e-01 -1.52732179e-01 2.33013794e-01 1.41371608e-01
1.47837967e-01 -2.74303883e-01 2.49892861e-01 -1.24526441e+00
2.40769267e-01 -1.10677505e+00 7.67951429e-01 -9.84288692e-01
-1.37745273e+00 6.77727759e-01 2.86342561e-01 -9.85087693e-01
-2.27901205e-01 -7.07492888e-01 1.74092233e-01 6.95423305e-01
-1.55757940e+00 -1.13488126e+00 -3.44021827e-01 9.30289686e-01
4.96131241e-01 -8.52698609e-02 1.00087535e+00 5.75496137e-01
-1.48384646e-01 2.72819430e-01 -9.73312929e-03 2.10040867e-01
1.00231111e+00 -1.03374875e+00 -1.00044468e-02 2.68871546e-01
5.66861272e-01 4.46520209e-01 4.23546046e-01 -1.32066950e-01
-1.74920619e+00 -5.14295936e-01 9.32735443e-01 -6.08130276e-01
7.92579412e-01 -3.95782471e-01 -9.38674212e-01 5.99918365e-01
5.92085004e-01 -6.86374679e-02 7.81080604e-01 5.80421612e-02
-5.84812284e-01 -2.20661163e-01 -8.49144757e-01 5.12888491e-01
6.55776620e-01 -7.45945752e-01 -4.43843871e-01 3.91261369e-01
4.80750352e-01 -4.22892481e-01 -9.12657261e-01 4.91520286e-01
4.55080658e-01 -1.15465772e+00 1.09491217e+00 -7.40303278e-01
3.80135417e-01 -2.91963071e-02 -5.00165820e-01 -9.05082285e-01
2.89036214e-01 -2.12442920e-01 -8.56059119e-02 1.13668621e+00
5.09680212e-01 -3.98894906e-01 7.22800493e-01 7.72588074e-01
-8.88347104e-02 -7.51828730e-01 -1.14651120e+00 -4.08547938e-01
-2.39909277e-03 -6.35109067e-01 3.37164164e-01 8.00694942e-01
-1.06085181e-01 3.96619439e-01 -1.79699317e-01 -7.16737807e-02
2.29786247e-01 8.89465399e-03 6.25883639e-01 -1.27338338e+00
-8.95669758e-02 -5.13124347e-01 -5.48603952e-01 -9.86105442e-01
3.94629240e-02 -8.67883265e-01 -1.99581668e-01 -1.85206258e+00
1.51180595e-01 -2.83560395e-01 -3.35717201e-01 6.75030828e-01
2.51429409e-01 4.97609347e-01 4.51315075e-01 3.34137380e-01
-1.08365428e+00 4.23105508e-01 9.89651859e-01 -3.35017234e-01
7.76969716e-02 -1.32084280e-01 -4.99280930e-01 5.53572714e-01
9.41879928e-01 -4.66331951e-02 -4.14945006e-01 -7.74030149e-01
7.98447311e-01 2.46260181e-01 5.49654305e-01 -1.10533845e+00
6.12726569e-01 1.28306150e-01 2.41622955e-01 -7.23236442e-01
7.48406887e-01 -1.03737926e+00 -1.86555106e-02 -3.85999940e-02
-5.91135800e-01 1.88851535e-01 4.48351294e-01 5.21441519e-01
-8.34152639e-01 -1.89702317e-01 3.90068144e-01 -3.85706306e-01
-9.16488349e-01 3.59808505e-01 -5.52724659e-01 8.66235495e-02
6.31653965e-01 -4.27822620e-02 -3.50192219e-01 -5.97006440e-01
-1.14023507e+00 3.28052878e-01 2.91128874e-01 8.11175883e-01
7.72850871e-01 -1.24085987e+00 -6.44189894e-01 -8.30304846e-02
3.90232891e-01 -1.22798525e-01 3.23336184e-01 8.26307535e-01
-3.92082363e-01 6.87738359e-01 -8.59270319e-02 -6.26961350e-01
-1.26488698e+00 5.94896257e-01 3.35835636e-01 -1.66611612e-01
-1.03301913e-01 6.59273684e-01 -1.00630932e-01 -4.89773303e-01
5.20021498e-01 -8.33275765e-02 -3.41195971e-01 5.67929506e-01
5.34765899e-01 1.44896299e-01 3.55034381e-01 -5.97552001e-01
-3.03152531e-01 2.51420379e-01 -1.27630070e-01 -3.88089001e-01
1.38271260e+00 -2.75382787e-01 -4.50298131e-01 6.20733380e-01
1.26323533e+00 -1.92198575e-01 -6.51627123e-01 -3.91267806e-01
3.21221590e-01 -1.58881947e-01 -1.90927804e-01 -9.78422105e-01
-9.70472515e-01 9.71865118e-01 6.92622542e-01 6.17089510e-01
1.31147158e+00 4.28761244e-01 6.73812032e-01 8.54431510e-01
3.31000686e-01 -1.10505986e+00 1.29415646e-01 6.41446352e-01
1.02584016e+00 -1.63192713e+00 -2.84039855e-01 3.33640948e-02
-4.60525811e-01 1.22120059e+00 3.56803060e-01 3.82813603e-01
6.20777547e-01 8.20587426e-02 2.09704354e-01 -3.22248518e-01
-9.17564034e-01 -4.96176839e-01 3.43917906e-01 4.28572744e-01
5.72126389e-01 -2.98817158e-01 -5.49099222e-03 1.74350634e-01
2.68675357e-01 1.48514345e-01 1.84702531e-01 1.04240787e+00
-4.30848688e-01 -1.51540470e+00 -5.49377024e-01 2.38926992e-01
-5.01554430e-01 -1.30711153e-01 -6.55502141e-01 9.74395216e-01
-2.18617946e-01 1.04586947e+00 -7.61266649e-02 -2.20706891e-02
5.31533957e-01 2.75964528e-01 4.27259028e-01 -4.35161948e-01
-4.32587951e-01 5.06578572e-02 1.87425897e-01 -7.26899445e-01
-7.78188944e-01 -6.29701614e-01 -1.06858563e+00 -3.49439792e-02
1.21114984e-01 9.37634259e-02 7.82472193e-01 7.42459714e-01
4.17492807e-01 5.57928905e-02 5.13465255e-02 -9.48975086e-01
-4.04550344e-01 -7.74873495e-01 -1.86765313e-01 5.92195570e-01
3.46102417e-01 -3.70936185e-01 -8.78004804e-02 1.61054283e-01] | [10.706006050109863, 1.4991449117660522] |
e7b16f0d-0875-4e03-b36b-79f02803f291 | research-on-discourse-parsing-from-the | null | null | https://aclanthology.org/2020.iwdp-1.1 | https://aclanthology.org/2020.iwdp-1.1.pdf | Research on Discourse Parsing: from the Dependency View | Discourse parsing aims to comprehensively acquire the logical structure of the whole text which may be helpful to some downstream applications such as summarization, reading comprehension, QA and so on. One important issue behind discourse parsing is the representation of discourse structure. Up to now, many discourse structures have been proposed, and the correponding parsing methods are designed, promoting the development of discourse research. In this paper, we mainly introduce our recent discourse research and its preliminary application from the dependency view. | ['Sujian Li'] | null | null | null | null | aacl-iwdp-2020-12 | ['discourse-parsing'] | ['natural-language-processing'] | [ 4.92599338e-01 8.73548269e-01 -4.13471460e-01 -4.18107539e-01
-5.63357949e-01 -5.65667033e-01 8.97437990e-01 7.16495812e-01
4.87593077e-02 1.11651421e+00 1.17962968e+00 -6.60362542e-01
7.60622099e-02 -8.49917948e-01 -1.56615108e-01 -2.93378919e-01
5.89888990e-02 1.38749629e-01 5.75212836e-01 -5.67647934e-01
7.27509618e-01 -1.66743785e-01 -1.38476574e+00 6.27888203e-01
1.20858192e+00 3.21681291e-01 4.55878377e-01 7.09288299e-01
-7.58410811e-01 1.51358032e+00 -1.07054102e+00 -4.49723512e-01
-8.41181576e-01 -1.12460792e+00 -1.46354938e+00 -7.16158226e-02
-2.11355075e-01 -1.75247699e-01 3.65785100e-02 1.15149868e+00
1.75891414e-01 7.54041672e-02 4.03027683e-01 -6.96036279e-01
-5.39574385e-01 1.23856282e+00 -2.28628799e-01 4.67592686e-01
9.29140389e-01 -4.54211533e-01 1.07745886e+00 -2.65574604e-01
8.19957078e-01 1.39528775e+00 5.50660156e-02 5.73905349e-01
-2.94034600e-01 1.32934749e-01 4.84972984e-01 6.02061749e-01
-4.89111632e-01 -4.58377600e-01 8.97868156e-01 -3.84207845e-01
9.14340079e-01 6.42505467e-01 5.77214956e-01 9.29241538e-01
1.99633047e-01 1.08226120e+00 9.96093154e-01 -8.16990733e-01
1.53005347e-01 -6.65254518e-02 8.25466096e-01 2.52651244e-01
1.48477525e-01 -6.06032073e-01 -4.31363374e-01 1.59527630e-01
3.74404281e-01 -9.59734678e-01 -5.59208333e-01 4.48068708e-01
-9.14866388e-01 8.07199657e-01 -3.21161486e-02 8.09483528e-01
-2.29874387e-01 -4.52060312e-01 7.31716275e-01 2.33984053e-01
4.09486681e-01 3.23893845e-01 -2.51942165e-02 -5.38357735e-01
-3.78295600e-01 3.78588796e-01 1.06597304e+00 7.47057676e-01
9.22083259e-02 -2.50542700e-01 -2.49752820e-01 4.28828448e-01
5.28381467e-01 6.47751093e-02 4.74635810e-01 -1.06656265e+00
9.39281285e-01 9.65348542e-01 -1.53359054e-02 -1.12339818e+00
-4.31830049e-01 2.87391424e-01 -6.75207436e-01 -4.83028531e-01
2.64136583e-01 -3.02956760e-01 -2.01544210e-01 1.41702735e+00
5.12740910e-01 -3.03652704e-01 4.83440697e-01 7.51520097e-01
1.50092757e+00 1.07950342e+00 3.36641520e-01 -7.86101282e-01
1.63067305e+00 -7.14054286e-01 -1.37810659e+00 -2.08939910e-01
8.84830356e-01 -8.50597501e-01 6.89431489e-01 7.08943829e-02
-1.26462209e+00 -3.37239712e-01 -9.93121386e-01 -5.46966851e-01
8.10216293e-02 -4.41787206e-02 4.76237744e-01 5.00846803e-01
-7.43418515e-01 2.61406183e-01 -7.23432064e-01 -1.70265123e-01
3.10971648e-01 -1.33137926e-01 3.01358346e-02 8.21311995e-02
-1.62189651e+00 1.26026750e+00 1.02172720e+00 2.82515913e-01
-1.22864597e-01 -8.96835551e-02 -9.09569502e-01 -7.31563643e-02
6.17587745e-01 -7.09594548e-01 1.53502464e+00 -8.15700948e-01
-1.90607655e+00 8.41410697e-01 -6.90211296e-01 -6.81761682e-01
3.90488356e-01 -4.54810828e-01 -4.01531875e-01 2.03474909e-01
1.44344956e-01 -7.58644491e-02 3.03283989e-01 -8.84193063e-01
-7.83485889e-01 -4.07947063e-01 6.09067976e-01 7.20650792e-01
4.44628559e-02 6.35665536e-01 -2.03747854e-01 -4.49232846e-01
1.50055796e-01 -1.82997659e-01 -6.72824532e-02 -7.54386961e-01
-6.84205770e-01 -7.31790662e-01 7.52572596e-01 -9.75289881e-01
1.86451733e+00 -1.71272814e+00 6.36482716e-01 -3.60210061e-01
2.16641709e-01 3.65761578e-01 3.79150778e-01 8.96808505e-01
-7.50480266e-03 3.15690786e-01 -3.74986500e-01 1.07935607e-01
-1.27829909e-01 1.51024163e-01 -5.87266982e-01 1.68364599e-01
3.26244771e-01 9.81383026e-01 -8.63915801e-01 -8.30960929e-01
3.10182095e-01 -3.61022912e-02 -1.32465750e-01 5.45495212e-01
-7.41400540e-01 7.33741701e-01 -9.41283584e-01 3.38988900e-01
3.51783097e-01 -1.90047041e-01 6.95321500e-01 7.25461915e-02
-5.67689896e-01 1.00382590e+00 -7.31255591e-01 1.61313617e+00
1.82834208e-01 8.83854806e-01 1.88042954e-01 -1.30617511e+00
6.52160168e-01 4.27898765e-01 4.71333191e-02 -7.55504906e-01
4.80221212e-01 -2.36809179e-02 2.70272136e-01 -1.11756825e+00
9.87449884e-01 -1.95998102e-01 -3.15336704e-01 4.82084781e-01
-4.38206822e-01 1.62816480e-01 5.67508996e-01 4.06605810e-01
6.27145767e-01 1.40125737e-01 9.32893515e-01 -2.17455894e-01
1.05735922e+00 6.79755867e-01 2.88251966e-01 8.79865438e-02
-5.47161624e-02 2.03312382e-01 9.79515374e-01 1.60971403e-01
-8.00448239e-01 -4.23288763e-01 -1.91432610e-01 1.01818717e+00
3.30082595e-01 -6.38962567e-01 -1.15782523e+00 -3.92432928e-01
-6.51915908e-01 1.06813085e+00 -3.58293563e-01 3.90288830e-01
-1.42231333e+00 -5.95437169e-01 5.52690923e-01 5.17921329e-01
7.59835422e-01 -1.28242791e+00 -8.20813596e-01 3.97096664e-01
-6.08228564e-01 -1.02821577e+00 4.07404244e-01 -3.54283571e-01
-8.40536416e-01 -1.36526263e+00 -4.27058995e-01 -9.14323926e-01
3.53546679e-01 2.74681807e-01 8.28366756e-01 4.27961588e-01
5.92395067e-01 -9.52703506e-02 -8.63634586e-01 -5.79195082e-01
-9.37023580e-01 3.52572739e-01 -5.33060491e-01 -7.70099521e-01
1.61660597e-01 -8.51110220e-02 -1.25808910e-01 -2.20546067e-01
-8.03081393e-01 4.44369078e-01 8.10568631e-02 5.68648338e-01
-1.91857424e-02 -1.76174156e-02 6.65044129e-01 -1.44696558e+00
1.34537303e+00 -5.73824823e-01 -2.76713401e-01 6.38566911e-01
-4.33260612e-02 -3.09782997e-02 5.45197785e-01 1.95015654e-01
-1.83828366e+00 -9.49088812e-01 -5.79230726e-01 9.54833746e-01
-1.18118770e-01 1.13559473e+00 -5.60310662e-01 6.91858292e-01
4.41561460e-01 -6.37108609e-02 1.06451519e-01 -3.37453485e-01
5.20771623e-01 6.96187675e-01 4.89620388e-01 -7.29069769e-01
1.87465355e-01 3.70462760e-02 -2.04916254e-01 -1.14593601e+00
-9.88239706e-01 -2.86377698e-01 -7.18564570e-01 -2.45247543e-01
1.02304721e+00 -5.23261845e-01 -6.34566307e-01 2.95022398e-01
-1.74508548e+00 -1.50525972e-01 -3.72555219e-02 2.36352086e-01
-3.54248613e-01 9.92714942e-01 -6.34389102e-01 -9.71716702e-01
-3.83805543e-01 -8.92172635e-01 7.00514674e-01 7.57032156e-01
-6.66076303e-01 -1.25480843e+00 1.65479735e-01 4.43571597e-01
4.50116321e-02 2.97529131e-01 1.32134783e+00 -5.59984148e-01
-3.17527354e-01 5.01091659e-01 -1.49082139e-01 6.26227930e-02
-1.64284557e-02 1.52546704e-01 -7.44919002e-01 2.43669719e-01
4.49179649e-01 -2.67059684e-01 4.64546680e-01 3.86844069e-01
8.51373315e-01 -4.43557769e-01 -3.29075336e-01 -1.16022281e-01
8.45988095e-01 4.37044382e-01 1.05923426e+00 5.20604789e-01
3.66379708e-01 1.30925095e+00 9.58189249e-01 4.09668908e-02
7.07636356e-01 2.81551659e-01 1.15435995e-01 3.71364415e-01
-1.73716500e-01 -3.22605610e-01 2.50327051e-01 1.51580167e+00
-2.77643651e-01 -5.93826652e-01 -9.16605294e-01 4.12017822e-01
-2.06235671e+00 -1.25161183e+00 -6.68447018e-01 1.43141484e+00
8.52717459e-01 2.40728974e-01 -2.98171520e-01 3.17137927e-01
9.22656357e-01 7.17835605e-01 -2.74731964e-01 -4.93868977e-01
-3.32836390e-01 9.93046910e-02 -3.60475749e-01 7.05292046e-01
-9.06821668e-01 8.61223519e-01 6.75845003e+00 3.84594083e-01
-8.63526881e-01 -1.83055401e-01 3.02734226e-01 6.21072948e-01
-4.46160823e-01 2.92326063e-01 -6.76426113e-01 2.62354612e-01
8.50822508e-01 -9.43019688e-01 -2.43017375e-01 5.85822880e-01
3.69759828e-01 -6.88352823e-01 -7.41924405e-01 4.66140926e-01
3.96646280e-03 -1.69911230e+00 9.14113671e-02 -2.26307049e-01
3.09098512e-01 -6.05191767e-01 -5.39031267e-01 2.46988967e-01
6.63364977e-02 -9.45645630e-01 5.00688016e-01 2.10090220e-01
3.78840238e-01 -4.20277596e-01 9.78096008e-01 9.33430195e-01
-9.72186625e-01 1.45055637e-01 -7.03277811e-02 -6.84120774e-01
8.96128237e-01 2.44546980e-01 -7.23720372e-01 1.25045240e+00
1.13316558e-01 6.72354937e-01 -2.40119725e-01 5.61972380e-01
-9.23046231e-01 8.08630407e-01 6.61930591e-02 -5.76478720e-01
7.77837709e-02 -3.87087315e-01 7.44560301e-01 1.06238091e+00
-1.18076177e-02 8.91881287e-01 3.08715692e-03 4.07683402e-01
6.58088773e-02 4.15318817e-01 -2.84795672e-01 -1.19336300e-01
6.98597014e-01 6.35027230e-01 -6.16578817e-01 -5.32146633e-01
-1.76549435e-01 4.01705533e-01 2.82023758e-01 -3.71792279e-02
-6.54679716e-01 -1.32664442e-01 1.84059907e-02 3.96979712e-02
-1.76632330e-01 -2.97965586e-01 -4.87176001e-01 -1.14878893e+00
2.41272047e-01 -1.00957990e+00 4.55975682e-01 -5.85428238e-01
-9.12192047e-01 6.06663048e-01 4.20474112e-01 -9.07771707e-01
-5.15411556e-01 -4.83067751e-01 -1.10439658e+00 8.72108996e-01
-1.50641811e+00 -6.89891636e-01 -9.84674692e-02 -7.66530260e-02
7.74460912e-01 1.58411935e-01 9.78948712e-01 -1.62670135e-01
-8.25085878e-01 -2.69896954e-01 -2.25194797e-01 3.54926378e-01
3.28374773e-01 -1.04681206e+00 2.44241118e-01 9.99669611e-01
-1.56164348e-01 8.18674564e-01 9.45235133e-01 -6.40203357e-01
-1.25596368e+00 -4.53155041e-01 1.34456909e+00 -2.56115705e-01
7.01774061e-01 1.98491722e-01 -1.31190419e+00 5.17912149e-01
8.88707161e-01 -9.33598995e-01 6.76712334e-01 1.87530652e-01
3.38050276e-01 5.58781028e-01 -8.68194640e-01 6.46115541e-01
8.58736038e-01 -4.11650687e-01 -1.72832656e+00 1.41038358e-01
8.35203946e-01 -9.82824564e-01 -8.66466820e-01 2.79604018e-01
2.17256919e-01 -1.01990306e+00 4.79882717e-01 -8.85424376e-01
8.89573693e-01 -3.22118521e-01 1.55489678e-02 -9.80249584e-01
1.07391186e-01 -6.91490889e-01 -3.70713353e-01 1.52776909e+00
2.94633567e-01 -4.71780866e-01 6.52617931e-01 7.53088534e-01
-4.80183750e-01 -4.62404460e-01 -9.06987965e-01 -1.05458535e-02
3.25570434e-01 -3.81291330e-01 4.59303886e-01 8.89361918e-01
9.68999326e-01 1.06019914e+00 6.17102236e-02 -2.50142574e-01
1.18295975e-01 3.69657427e-01 6.75750554e-01 -1.25895000e+00
1.85699351e-02 -6.59558058e-01 2.31205568e-01 -1.72928178e+00
1.33359134e-01 -5.19093096e-01 -1.18317038e-01 -2.14933372e+00
-5.53034768e-02 -7.67580569e-02 6.33872807e-01 -1.25366285e-01
-4.06801283e-01 -9.05397177e-01 8.48327726e-02 2.74309337e-01
-6.13167822e-01 5.58930516e-01 1.46934474e+00 -8.14274698e-02
-4.01441008e-01 -4.05817628e-02 -9.96649384e-01 9.91457641e-01
8.81329775e-01 -1.46833792e-01 -4.88456249e-01 -5.90535462e-01
2.38553002e-01 5.40554881e-01 -2.37008005e-01 -2.30577886e-01
4.66818184e-01 -4.93410289e-01 -2.55864263e-01 -1.01399601e+00
6.00520931e-02 -1.21289626e-01 -4.36038785e-02 2.94896245e-01
-4.90937203e-01 9.88148227e-02 3.29332910e-02 2.08949178e-01
-8.22923124e-01 -7.84575462e-01 2.25723535e-01 -2.46487275e-01
-9.69437420e-01 -4.96764839e-01 -5.09015381e-01 5.56043625e-01
1.19410753e+00 -1.48458496e-01 -9.24781859e-01 -1.64922997e-01
-3.88967961e-01 5.12465179e-01 2.41019472e-01 2.88151622e-01
5.20115018e-01 -8.26483190e-01 -7.83970833e-01 -4.24793422e-01
-3.17999959e-01 5.73928535e-01 3.29936683e-01 6.93409920e-01
-7.56561458e-01 7.71785259e-01 -1.46124467e-01 -3.85400951e-01
-1.52844119e+00 2.36496225e-01 -1.72666803e-01 -5.54418445e-01
-6.35548890e-01 3.74720484e-01 -4.78533693e-02 1.87316447e-01
6.18010946e-02 -3.70442390e-01 -1.19036293e+00 2.86476046e-01
1.01324522e+00 5.53097546e-01 -2.45835230e-01 -8.48414898e-01
-2.16068283e-01 3.82584155e-01 9.90194734e-03 -1.99149296e-01
1.06379628e+00 -5.01508057e-01 -6.50421143e-01 5.59669793e-01
6.37865365e-01 1.43650055e-01 -5.30994892e-01 5.96936457e-02
6.19884551e-01 -1.72626346e-01 -3.07593524e-01 -3.73536319e-01
-2.80858371e-02 1.00750589e+00 -5.43398798e-01 1.08514154e+00
9.55782950e-01 2.28367165e-01 6.39095962e-01 3.33438993e-01
4.01313394e-01 -1.26672626e+00 -4.11044061e-01 9.11948383e-01
9.89893854e-01 -9.14228976e-01 4.18570250e-01 -9.20558453e-01
-8.78712654e-01 1.42851055e+00 3.99763018e-01 1.22827269e-01
2.85769612e-01 2.50395507e-01 2.22654585e-02 -3.77104670e-01
-6.79398954e-01 -2.36544833e-01 5.90307191e-02 6.79146111e-01
9.37375605e-01 4.72663343e-02 -1.07571423e+00 6.71579003e-01
-4.42930907e-01 -1.35528296e-01 6.07019007e-01 1.01195180e+00
-9.33617949e-01 -1.51185942e+00 -6.29889607e-01 1.08635023e-01
-4.79914874e-01 2.05245122e-01 -7.41787553e-01 9.87340033e-01
-3.18142176e-01 1.26929307e+00 -8.70360956e-02 1.14821889e-01
4.62835908e-01 -7.35085234e-02 5.71988523e-01 -7.91250646e-01
-6.57025814e-01 -1.47734359e-01 1.01615846e+00 1.69666465e-02
-1.29754698e+00 -8.34393442e-01 -1.72186446e+00 -4.18116182e-01
-3.80003393e-01 5.49774349e-01 2.83358544e-01 1.29385471e+00
1.98931675e-02 6.71658278e-01 3.30576956e-01 -2.10313678e-01
-3.44887137e-01 -1.26872373e+00 -1.90743636e-02 1.23508260e-01
-8.71823262e-03 -4.05233413e-01 4.01724316e-02 1.36631683e-01] | [10.805436134338379, 9.485799789428711] |
758b13ef-12d6-484c-a54e-cf7701b1b0bb | s-page-a-speaker-and-position-aware-graph | 2112.12389 | null | https://arxiv.org/abs/2112.12389v1 | https://arxiv.org/pdf/2112.12389v1.pdf | S+PAGE: A Speaker and Position-Aware Graph Neural Network Model for Emotion Recognition in Conversation | Emotion recognition in conversation (ERC) has attracted much attention in recent years for its necessity in widespread applications. Existing ERC methods mostly model the self and inter-speaker context separately, posing a major issue for lacking enough interaction between them. In this paper, we propose a novel Speaker and Position-Aware Graph neural network model for ERC (S+PAGE), which contains three stages to combine the benefits of both Transformer and relational graph convolution network (R-GCN) for better contextual modeling. Firstly, a two-stream conversational Transformer is presented to extract the coarse self and inter-speaker contextual features for each utterance. Then, a speaker and position-aware conversation graph is constructed, and we propose an enhanced R-GCN model, called PAG, to refine the coarse features guided by a relative positional encoding. Finally, both of the features from the former two stages are input into a conditional random field layer to model the emotion transfer. | ['Yang Dong', 'Yongliang Wang', 'Juyang Huang', 'Jing Xu', 'Chong Yang', 'Chen Liang'] | 2021-12-23 | null | null | null | null | ['emotion-recognition-in-conversation'] | ['natural-language-processing'] | [ 2.53060043e-01 1.80671394e-01 2.63756990e-01 -7.54984856e-01
-5.83042324e-01 -1.38154492e-01 6.73826694e-01 1.21690616e-01
-2.11688995e-01 3.48367602e-01 6.42306089e-01 -1.88504755e-01
1.97118819e-01 -7.45014250e-01 -2.74636149e-01 -6.85486794e-01
-7.14032948e-02 2.95224134e-02 9.70808864e-02 -4.57337260e-01
4.09879722e-02 6.57721683e-02 -1.35917187e+00 5.96638143e-01
9.25607681e-01 1.39513707e+00 2.18776315e-01 7.82637298e-01
-7.07783341e-01 1.04356718e+00 -5.39226234e-01 -3.16826254e-01
-5.26042819e-01 -8.05829763e-01 -7.78880179e-01 7.39800632e-02
-3.89758915e-01 8.33443925e-02 -2.79833168e-01 9.29760158e-01
5.68046033e-01 5.71062386e-01 4.37899709e-01 -1.26817226e+00
-5.40178657e-01 8.83556724e-01 -4.09127295e-01 -1.15061812e-01
4.30437863e-01 -1.44586816e-01 1.01862514e+00 -7.59324849e-01
2.34668896e-01 1.44038236e+00 5.45334458e-01 5.53833425e-01
-8.07476461e-01 -6.47770226e-01 7.56583631e-01 4.11182851e-01
-1.15296829e+00 -2.87104368e-01 1.40289116e+00 -7.98627958e-02
1.08784616e+00 1.45528480e-01 9.61497843e-01 1.26486909e+00
9.49556977e-02 8.82408381e-01 1.00966191e+00 -2.92702913e-01
2.75353998e-01 2.15263352e-01 2.27751002e-01 4.90729958e-01
-7.22799540e-01 -1.35149390e-01 -6.92413986e-01 -5.10110743e-02
4.05151397e-01 2.44806111e-02 -3.68767560e-01 5.71416318e-02
-5.80221772e-01 7.68793643e-01 6.60975039e-01 5.41107714e-01
-4.75176990e-01 4.55337353e-02 5.48017144e-01 1.22895606e-01
7.76359320e-01 -6.69621751e-02 -2.22703487e-01 -4.48169351e-01
-5.22023737e-01 -5.73959313e-02 8.95394564e-01 9.56856966e-01
7.67862678e-01 -5.54729328e-02 -2.20658198e-01 1.03260791e+00
5.53919077e-01 1.32544473e-01 4.43387777e-01 -5.29035389e-01
5.27187169e-01 9.44802642e-01 -2.67855942e-01 -1.45970702e+00
-5.43033183e-01 -5.17907381e-01 -1.14273119e+00 -4.55507338e-01
-3.23367298e-01 -2.93477744e-01 -4.60105926e-01 1.88668001e+00
4.13468838e-01 4.79333937e-01 2.59869397e-01 8.14491272e-01
1.25248611e+00 1.02870834e+00 3.69918168e-01 -1.99754030e-01
1.39928567e+00 -1.12302458e+00 -9.99811411e-01 -4.46527421e-01
3.76281261e-01 -4.57967788e-01 8.93627524e-01 1.16645165e-01
-8.20282519e-01 -5.65264940e-01 -8.51946831e-01 -2.42164657e-01
-5.67216277e-01 2.42526866e-02 5.96857369e-01 2.78820157e-01
-1.07451570e+00 1.98653296e-01 -5.25506258e-01 -1.46885574e-01
8.07682499e-02 2.39419177e-01 -1.70751825e-01 4.83236127e-02
-1.74496675e+00 7.17415750e-01 1.96822256e-01 8.86244655e-01
-5.03724158e-01 -8.96049589e-02 -1.16048861e+00 4.35715258e-01
2.60834754e-01 -3.40462208e-01 1.10436571e+00 -1.21439886e+00
-2.00728607e+00 3.48854035e-01 -5.01794457e-01 -5.25043812e-03
1.02861539e-01 -7.98353925e-02 -7.72546530e-01 -5.89786738e-04
-4.12672609e-01 4.77766991e-01 8.52868676e-01 -1.34794915e+00
-6.54843748e-01 -4.59645897e-01 9.32313502e-02 6.42872930e-01
-2.00237781e-01 8.63434374e-02 -8.19599271e-01 -4.72981840e-01
1.69260614e-02 -5.92442572e-01 -2.68056780e-01 -7.07633078e-01
-4.33593124e-01 -6.93621755e-01 8.98793340e-01 -6.60903454e-01
1.47142613e+00 -2.27241755e+00 3.57886374e-01 1.85688958e-01
2.48160765e-01 1.24524243e-01 -4.41440463e-01 6.59834743e-01
-1.86313391e-01 -5.02988324e-02 -1.25691995e-01 -8.27736616e-01
5.61879948e-02 5.79220094e-02 -1.53079510e-01 8.13474581e-02
4.99941915e-01 9.71375406e-01 -8.52488875e-01 -4.61610943e-01
1.11256748e-01 9.16128099e-01 -4.25148338e-01 5.36429346e-01
-2.49613255e-01 4.54292208e-01 -5.95898986e-01 1.25629827e-01
7.87550569e-01 -2.37903476e-01 4.60497975e-01 -1.11278728e-01
-3.13971601e-02 4.28709447e-01 -9.85538423e-01 1.43793225e+00
-7.93885529e-01 4.05347139e-01 3.81890684e-01 -9.83094692e-01
1.21342731e+00 5.44025958e-01 2.32582912e-01 -7.18237400e-01
4.18169469e-01 -2.78519481e-01 -2.63707191e-01 -4.75589156e-01
7.32835591e-01 -2.46271729e-01 -3.07108015e-01 4.15894419e-01
1.33347586e-01 -2.93694381e-02 -3.49413455e-01 4.51538533e-01
8.09775412e-01 -1.20362127e-02 1.00539185e-01 3.47157046e-02
9.73592520e-01 -5.89714468e-01 7.84451962e-01 2.01405317e-01
-3.94205779e-01 3.13077539e-01 7.55133033e-01 -2.14765489e-01
-1.12280302e-01 -4.90550250e-01 5.01950324e-01 1.27142346e+00
3.90881896e-01 -6.41150236e-01 -8.32967997e-01 -8.86022687e-01
-3.65563929e-01 7.04298496e-01 -8.37754071e-01 -4.36185658e-01
-5.13496220e-01 -2.95723498e-01 1.56238765e-01 5.61105609e-01
6.30289435e-01 -1.51400280e+00 -1.22840472e-01 4.31124985e-01
-5.95195770e-01 -1.09748816e+00 -7.40855336e-01 3.05425555e-01
-4.14998710e-01 -7.55123794e-01 -4.85173136e-01 -8.55584681e-01
4.46248204e-01 1.50949240e-01 9.07677889e-01 1.30752429e-01
2.24362880e-01 3.49661916e-01 -6.38537586e-01 -1.97693110e-01
-1.35479748e-01 2.28428334e-01 -3.05123478e-01 6.08277559e-01
5.46137691e-01 -7.21134007e-01 -5.45450687e-01 1.66503862e-01
-6.85498416e-01 2.62286395e-01 4.27526146e-01 7.70667315e-01
2.55652785e-01 2.13328436e-01 9.16055083e-01 -9.77590501e-01
9.75880623e-01 -4.96682286e-01 -2.42544562e-01 2.54574388e-01
-2.76095480e-01 -1.75898626e-01 7.06096709e-01 -3.13344151e-01
-1.53995669e+00 -1.15322351e-01 -4.75606531e-01 -2.01171443e-01
-1.47391692e-01 8.39342952e-01 -7.29551971e-01 3.50338429e-01
4.08837907e-02 3.05866361e-01 -6.13265671e-02 -1.93674967e-01
5.11799634e-01 1.11340010e+00 3.58094007e-01 -4.94451940e-01
2.59166062e-01 -3.44100595e-02 -5.67606449e-01 -8.17216873e-01
-8.79914582e-01 -4.90625352e-01 -3.87568951e-01 -5.45465529e-01
1.07873058e+00 -8.25046003e-01 -9.20473158e-01 7.61715353e-01
-1.27790475e+00 -5.35153747e-01 4.40712646e-03 4.59283739e-01
-2.70168632e-01 3.07910591e-01 -7.88424313e-01 -1.19804955e+00
-3.98470938e-01 -1.18329847e+00 1.17825377e+00 5.35445869e-01
9.79974270e-02 -1.15092206e+00 -1.60218522e-01 2.23262072e-01
6.46717250e-01 -5.53946793e-02 8.18249583e-01 -7.21782863e-01
-3.23852211e-01 -2.58224402e-02 -4.80844021e-01 4.13001299e-01
8.55681449e-02 -2.40153477e-01 -1.20646369e+00 2.29772534e-02
1.07613988e-01 -2.83144832e-01 7.65851915e-01 1.79804772e-01
1.23855782e+00 -1.68372557e-01 -3.06702346e-01 4.30357665e-01
1.05779397e+00 3.95100862e-01 6.24260008e-01 -1.46995470e-01
7.37696230e-01 8.73040915e-01 3.55422080e-01 3.92019540e-01
8.99202943e-01 3.74389023e-01 3.09031785e-01 -1.40811652e-01
1.48431674e-01 -2.92034239e-01 4.19630378e-01 1.41188967e+00
-6.32376038e-03 -4.35055912e-01 -5.44439077e-01 3.31671923e-01
-1.95555329e+00 -6.88165069e-01 1.14352591e-01 1.59515989e+00
7.21165478e-01 1.67547800e-02 -1.64074019e-01 -1.49578691e-01
9.88967896e-01 5.72700739e-01 -4.30500895e-01 -5.37106216e-01
-6.56973124e-02 -4.73842472e-02 -3.39723021e-01 6.41787291e-01
-8.79742205e-01 9.73417222e-01 5.04075575e+00 7.72921979e-01
-1.14774382e+00 -1.74950391e-01 7.23650992e-01 3.68318081e-01
-6.22392416e-01 5.54507934e-02 -6.21842980e-01 3.83196890e-01
9.42049146e-01 1.68116465e-01 6.22575879e-01 8.10980439e-01
1.74449757e-01 9.40550566e-02 -8.03707242e-01 1.07316995e+00
2.69092292e-01 -9.21720862e-01 -1.77568421e-01 -1.26156524e-01
2.26070285e-01 -3.06039780e-01 -1.96438625e-01 8.40563595e-01
3.20428938e-01 -8.77530575e-01 4.68940765e-01 5.39122224e-01
6.18679464e-01 -1.05327845e+00 9.52934980e-01 2.89939642e-01
-1.51068556e+00 1.09091513e-01 -1.50733799e-01 3.69573683e-02
3.49844784e-01 4.77553606e-01 -4.39850837e-01 8.09305251e-01
5.84561348e-01 7.76712298e-01 -3.06534916e-01 3.17828417e-01
-3.59747320e-01 7.73602962e-01 -1.16171747e-01 -4.39638197e-01
3.91102731e-01 -3.41562480e-01 3.01157326e-01 1.49538648e+00
9.97622162e-02 3.65850061e-01 1.71693474e-01 6.12779081e-01
-1.65823534e-01 3.93278927e-01 -2.81284064e-01 7.06521980e-03
4.67894524e-01 1.56730354e+00 -4.96890873e-01 -3.38970304e-01
-5.50072908e-01 1.17020690e+00 6.81655169e-01 6.18180871e-01
-8.68191898e-01 -7.19611943e-01 3.87307554e-01 -5.03142834e-01
3.82313848e-01 5.45441248e-02 1.36748090e-01 -1.18626595e+00
-1.83318898e-01 -7.18080640e-01 3.76752049e-01 -6.62895620e-01
-1.40671468e+00 1.02786684e+00 -4.41172242e-01 -7.63141751e-01
-2.14332104e-01 -4.02767420e-01 -1.05377054e+00 1.25444531e+00
-1.63826740e+00 -1.36736584e+00 -4.79167610e-01 7.59680808e-01
3.75301868e-01 7.72641301e-02 9.08958495e-01 7.53620937e-02
-8.17668498e-01 6.06939197e-01 -4.68099296e-01 2.34355941e-01
5.57345450e-01 -1.20264637e+00 2.60495603e-01 5.71596444e-01
-2.83061236e-01 7.12953091e-01 2.49902904e-01 -5.88346004e-01
-1.26538026e+00 -1.27432251e+00 1.21965957e+00 8.77869725e-02
5.18634379e-01 -6.92038536e-01 -1.11135733e+00 6.32384479e-01
5.11092246e-01 -1.73722774e-01 6.63390458e-01 5.75816512e-01
-3.60501081e-01 -2.35103071e-01 -8.23044837e-01 5.39677978e-01
8.01599801e-01 -9.69626486e-01 -5.29956579e-01 -2.08665088e-01
1.11331379e+00 -2.01870322e-01 -8.17642510e-01 1.68838546e-01
3.08950961e-01 -9.12762642e-01 4.95283782e-01 -1.72276169e-01
3.78206104e-01 -6.84640482e-02 -1.20283261e-01 -1.54311252e+00
-2.44923145e-01 -7.20790565e-01 1.04609884e-01 1.69242167e+00
3.34246725e-01 -7.76542604e-01 5.43382168e-01 4.11964953e-01
-4.66196388e-01 -8.11968148e-01 -7.44204044e-01 -4.85322550e-02
-2.55636483e-01 -7.84686625e-01 8.60675514e-01 9.84677792e-01
4.91819799e-01 1.08454204e+00 -3.48772377e-01 2.83708293e-02
9.53485146e-02 3.08015972e-01 5.39053857e-01 -1.04101789e+00
-1.79474264e-01 -4.95526880e-01 2.76270360e-02 -1.36822164e+00
3.93039346e-01 -8.52583408e-01 5.12949824e-01 -1.70866251e+00
4.46270816e-02 -4.43255752e-01 -4.73041266e-01 1.99965209e-01
-6.12635672e-01 -3.10099125e-01 3.86111476e-02 -3.82170290e-01
-8.09666455e-01 1.14168489e+00 1.28901267e+00 3.82075273e-02
-4.68235970e-01 -5.08483648e-02 -8.11201692e-01 4.41223860e-01
7.02760220e-01 3.23375501e-02 -4.95888948e-01 -1.46617621e-01
1.77006990e-01 4.06603396e-01 2.00287089e-01 -6.66261315e-01
4.04823780e-01 -6.59059733e-03 2.16127381e-01 -8.35049331e-01
4.71044153e-01 -7.74169922e-01 -2.13812739e-01 -1.98719382e-01
-5.03460348e-01 -7.50759915e-02 5.81421405e-02 8.24970663e-01
-6.25334322e-01 1.88363045e-01 5.30581653e-01 1.44949295e-02
-5.78078151e-01 4.37607437e-01 -4.76577908e-01 -1.12324566e-01
6.38726652e-01 2.84779489e-01 -1.73575178e-01 -8.38166773e-01
-8.50375533e-01 5.57711005e-01 -1.68013424e-01 5.81158102e-01
6.75376475e-01 -1.28732407e+00 -4.91906643e-01 2.64146149e-01
2.84635514e-01 1.39308646e-01 9.00489092e-01 7.33438015e-01
9.98821780e-02 1.95002228e-01 3.39883775e-01 -3.89103681e-01
-1.06891513e+00 5.57993531e-01 3.72738332e-01 -5.20594180e-01
-5.97254634e-01 9.06167388e-01 3.55451882e-01 -7.71044254e-01
4.73644137e-01 -3.86738092e-01 -6.02044404e-01 2.07451835e-01
5.30032277e-01 -7.53955171e-02 -9.60686058e-02 -8.23252738e-01
-3.14656258e-01 2.71856248e-01 -2.95430166e-03 -2.77615935e-01
1.36167336e+00 -5.13406754e-01 -3.26113850e-01 5.43559253e-01
1.26952517e+00 -1.20463399e-02 -1.22161281e+00 -4.32014108e-01
-2.14604646e-01 1.44851848e-01 2.11269334e-01 -7.32203662e-01
-1.06645656e+00 1.19360030e+00 2.30772585e-01 4.25749093e-01
1.41693211e+00 8.12683553e-02 9.52456832e-01 8.27048495e-02
-5.36976941e-02 -1.07987320e+00 3.07588894e-02 8.48363340e-01
9.28064406e-01 -8.77182961e-01 -6.83951616e-01 -5.16745329e-01
-9.40519810e-01 1.07590187e+00 6.87677920e-01 7.50394091e-02
8.57974648e-01 2.79842705e-01 2.71540463e-01 -2.93446064e-01
-9.08333898e-01 -2.35114262e-01 2.25526318e-01 5.61952889e-01
5.99771619e-01 1.33824319e-01 -8.01324621e-02 1.21352661e+00
-1.96359932e-01 -1.51123151e-01 9.84697044e-03 6.37636423e-01
-2.26930872e-01 -1.05069447e+00 1.25184745e-01 1.63538486e-01
-1.53792500e-01 -2.32398361e-01 -4.24151301e-01 4.79913771e-01
3.23362611e-02 1.28072929e+00 1.74036771e-01 -8.27694476e-01
3.26079071e-01 2.45149389e-01 -7.48642385e-02 -4.54723626e-01
-9.20435667e-01 4.26811695e-01 3.03122848e-01 -6.54903471e-01
-5.98610103e-01 -4.13017750e-01 -1.37486863e+00 -1.37156472e-02
-4.47824806e-01 6.41020775e-01 6.31953061e-01 9.60546732e-01
5.09963512e-01 9.21911955e-01 9.54359412e-01 -8.06690037e-01
2.56440472e-02 -1.32267869e+00 -7.34999418e-01 3.56694162e-01
3.32922250e-01 -4.38239515e-01 -4.12411094e-01 -2.00584456e-01] | [13.004432678222656, 6.1357808113098145] |
335ff0b8-6c68-4692-a22b-cec73e2240e3 | seeing-deeply-and-bidirectionally-a-deep | null | null | http://openaccess.thecvf.com/content_ECCV_2018/html/Jie_Yang_Seeing_Deeply_and_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Jie_Yang_Seeing_Deeply_and_ECCV_2018_paper.pdf | Seeing Deeply and Bidirectionally: A Deep Learning Approach for Single Image Reflection Removal | Reflections often obstruct the desired scene when taking photos through glass panels. Removing unwanted reflection automatically from the photos is highly desirable. Traditional methods often impose certain priors or assumptions to target particular type(s) of reflection such as shifted double reflection, thus have difficulty to generalise to other types. Very recently a deep learning approach has been proposed. It learns a deep neural network that directly maps a reflection contaminated image to a background (target) image (ie reflection free image) in an end to end fashion, and outperforms the previous methods. We argue that, to remove reflection truly well, we should estimate the reflection and utilise it to estimate the background image. We propose a cascade deep neural network, which estimates both the background image and the reflection. This significantly improves reflection removal. In the cascade deep network, we use the estimated background image to estimate the reflection, and then use the estimated reflection to estimate the background image, facilitating our idea of seeing deeply and bidirectionally. | ['Dong Gong', 'Qinfeng Shi', 'Lingqiao Liu', 'Jie Yang'] | 2018-09-01 | null | null | null | eccv-2018-9 | ['reflection-removal'] | ['computer-vision'] | [ 8.81846488e-01 1.37615144e-01 5.69631398e-01 -3.20322722e-01
-5.70394158e-01 -3.77885610e-01 6.11173570e-01 -8.26806366e-01
-1.17787123e-01 4.77549523e-01 2.67051786e-01 -3.54105562e-01
6.34469569e-01 -9.01840389e-01 -9.22153354e-01 -1.13522398e+00
5.77476680e-01 -1.30461097e-01 2.53268749e-01 -2.21502990e-01
1.46365151e-01 5.47076941e-01 -1.51449966e+00 6.70632780e-01
4.08520460e-01 8.32948327e-01 1.48507327e-01 8.07002187e-01
-1.10207468e-01 1.17441857e+00 -8.08307707e-01 -1.93053856e-01
6.84834361e-01 -6.08561575e-01 -3.64133269e-01 3.04173619e-01
8.54430795e-01 -8.27861369e-01 -3.24014455e-01 9.85224307e-01
4.58654732e-01 1.46501034e-01 5.26098549e-01 -7.56909072e-01
-5.14632821e-01 2.53665537e-01 -9.86796498e-01 1.16188088e-02
3.36932063e-01 5.12785241e-02 3.05440962e-01 -8.87044966e-01
1.89327508e-01 1.11220944e+00 8.95202458e-01 6.12990856e-01
-9.71704185e-01 -6.56455278e-01 3.66701841e-01 -2.39830837e-01
-1.19600630e+00 -6.57247961e-01 9.80228662e-01 -1.88013494e-01
6.39943302e-01 4.42305654e-01 6.09260142e-01 1.37947476e+00
-1.73537377e-02 7.86134362e-01 1.25559735e+00 -6.11395419e-01
-2.57132143e-01 3.41343671e-01 -2.13807628e-01 5.55970907e-01
1.66908383e-01 1.74591139e-01 -2.17855111e-01 2.37164497e-01
7.85582781e-01 2.89972305e-01 -7.15241194e-01 1.31582081e-01
-8.56014192e-01 1.14640668e-01 4.61736530e-01 2.90606730e-02
-3.14934731e-01 1.93782598e-01 -2.09146678e-01 2.71245271e-01
6.33495331e-01 -8.22932348e-02 -2.46944144e-01 6.25943720e-01
-8.75785112e-01 -1.14177220e-01 8.90050292e-01 6.18067265e-01
9.68983710e-01 1.47187263e-01 -4.99698073e-02 9.31524277e-01
4.12055165e-01 7.47156620e-01 -1.14228942e-01 -9.04392362e-01
3.16997647e-01 4.34313029e-01 4.93251741e-01 -7.69959092e-01
-2.87228733e-01 -4.10572737e-01 -1.13935673e+00 7.39467740e-01
3.94430250e-01 -3.73647898e-01 -1.30466688e+00 1.36209321e+00
4.36486095e-01 4.24019724e-01 1.07281744e-01 1.13669670e+00
9.15565312e-01 7.05082476e-01 -5.94262600e-01 6.44890368e-02
8.62371147e-01 -1.10060203e+00 -4.70551401e-01 -5.26698709e-01
1.74710527e-01 -9.32336509e-01 9.60363686e-01 8.20448577e-01
-9.44780290e-01 -5.73108673e-01 -1.03222191e+00 -1.70175701e-01
-1.07975148e-01 4.62668419e-01 3.48577678e-01 8.33607316e-01
-1.14436150e+00 2.99277246e-01 -5.75964391e-01 -8.09348598e-02
3.13694388e-01 5.27583435e-02 -1.22841403e-01 -3.78489643e-01
-9.09039795e-01 1.00548506e+00 -1.75530717e-01 8.51727307e-01
-1.14820611e+00 -4.77673918e-01 -5.19599438e-01 -2.72669569e-02
4.21459883e-01 -7.11895168e-01 9.73174214e-01 -1.71424925e+00
-1.90246999e+00 9.79840100e-01 -1.89517722e-01 -1.16272107e-01
7.62216032e-01 -6.24483645e-01 -3.41869205e-01 2.17453167e-02
-4.51552719e-01 1.33877248e-01 1.41495168e+00 -1.88261890e+00
-4.99358743e-01 -6.71851709e-02 3.62990558e-01 4.45325196e-01
1.20920576e-01 -1.03172168e-01 -5.06583035e-01 -8.24159291e-03
4.77428772e-02 -6.85122311e-01 -2.13514924e-01 5.33922948e-02
-6.73698604e-01 5.45873642e-01 9.17631209e-01 -6.11257613e-01
5.45476794e-01 -2.09998012e+00 -3.30203384e-01 1.17115773e-01
3.62030596e-01 4.89787608e-01 -2.93952376e-01 3.69139344e-01
-1.46328598e-01 -3.62186760e-01 -2.26908460e-01 -2.87631273e-01
-4.66489345e-01 -1.74819633e-01 -4.35572594e-01 7.56900012e-01
2.58515745e-01 5.71803510e-01 -6.42825305e-01 1.50955737e-01
4.64505255e-01 1.09160280e+00 -3.53407353e-01 3.72178972e-01
-7.59873912e-02 6.03922725e-01 -8.48298818e-02 3.04766029e-01
1.29868305e+00 -1.51146390e-02 1.86109632e-01 -5.02716422e-01
-3.14311355e-01 8.13245699e-02 -1.06262338e+00 9.39620018e-01
-7.80787766e-01 9.88097072e-01 4.38548326e-01 -8.47572744e-01
1.17289674e+00 -8.71267635e-03 3.12647820e-01 -9.32639182e-01
6.44644126e-02 -6.20596968e-02 -1.66317388e-01 -6.17104292e-01
2.15482369e-01 -3.14933926e-01 4.74072218e-01 4.05166984e-01
-5.06492794e-01 -1.22030973e-01 -5.42389333e-01 -1.25803068e-01
1.00920200e+00 5.57916164e-01 -4.00630720e-02 1.49522901e-01
7.17540205e-01 -4.69836384e-01 3.09164882e-01 9.85560060e-01
3.05111885e-01 1.19989622e+00 7.13917837e-02 -7.83813179e-01
-8.30924571e-01 -1.28536916e+00 3.29661638e-01 8.56525242e-01
4.39556807e-01 2.31226340e-01 -6.78759277e-01 -4.11411136e-01
-3.25724959e-01 6.60155892e-01 -4.51306194e-01 -2.65557677e-01
-7.38391101e-01 -7.02107489e-01 2.77577847e-01 2.45807588e-01
9.20864820e-01 -1.02353382e+00 -6.18120909e-01 1.27629414e-02
-4.16109771e-01 -9.55778658e-01 -1.09885000e-01 1.32357761e-01
-2.71307081e-01 -1.15426576e+00 -8.29437435e-01 -4.67841446e-01
7.71983564e-01 9.91981864e-01 1.31884027e+00 4.05368388e-01
-4.57069427e-01 3.68595958e-01 -1.55726045e-01 -6.59052789e-01
-4.02325481e-01 -5.83087742e-01 -4.23868954e-01 4.61444050e-01
3.36729884e-01 -4.70851481e-01 -1.00964630e+00 3.62682015e-01
-1.05919576e+00 3.53160620e-01 6.92098141e-01 4.72370893e-01
8.55989382e-02 2.82764256e-01 -1.85878258e-02 -1.05910969e+00
2.43391663e-01 -8.45331699e-02 -8.40541959e-01 1.75489590e-01
9.07539874e-02 -3.66466165e-01 6.07850850e-01 -2.97684431e-01
-1.70726180e+00 -5.64628560e-03 -2.66835451e-01 -3.55418235e-01
-3.54090005e-01 -2.96616275e-03 -3.33132297e-01 -9.12058279e-02
6.31040752e-01 1.99186921e-01 -2.48821840e-01 -3.72147381e-01
4.08544749e-01 6.69620097e-01 4.68149632e-01 -2.14712381e-01
7.88517594e-01 1.16688406e+00 -5.80160767e-02 -1.30499721e+00
-1.52588940e+00 -5.16433775e-01 -5.68606436e-01 -4.89796817e-01
7.63391733e-01 -1.08744681e+00 -6.60649240e-01 7.22339630e-01
-1.15166342e+00 -8.64604056e-01 -2.08819341e-02 3.34675759e-01
-3.25925112e-01 3.21906865e-01 -4.24270242e-01 -1.31479025e+00
-2.06151500e-01 -8.28921854e-01 1.31520069e+00 2.27506816e-01
2.00725839e-01 -8.70324135e-01 4.10243347e-02 5.13286710e-01
6.13081098e-01 1.69178486e-01 4.56772387e-01 1.88005656e-01
-9.57764447e-01 -1.18738219e-01 -7.69380927e-01 8.25327218e-01
1.64416432e-01 2.59373933e-02 -1.66780579e+00 -1.08599104e-01
4.64081436e-01 -3.39878462e-02 1.40379465e+00 4.90481377e-01
1.09118664e+00 -2.35059708e-01 -1.95358947e-01 9.04794276e-01
1.69426703e+00 1.39404386e-02 1.34906125e+00 3.59204978e-01
1.04500735e+00 6.27701402e-01 2.97350377e-01 1.67730108e-01
4.43329513e-02 3.96435499e-01 8.11303914e-01 -8.75017345e-01
-5.25599658e-01 1.50260881e-01 6.57746315e-01 3.23714793e-01
-2.91820347e-01 -7.31507957e-01 -5.85155249e-01 2.38836572e-01
-1.50368381e+00 -1.09351254e+00 -4.06193495e-01 2.23847437e+00
4.86865848e-01 -2.02625655e-02 -2.87643522e-01 -1.03674293e-01
6.10674143e-01 2.27123126e-01 -4.39372659e-01 -2.71238178e-01
-2.46355489e-01 1.52270958e-01 5.76516449e-01 7.98926234e-01
-9.10282135e-01 8.52738142e-01 6.68615675e+00 1.91229150e-01
-1.51349282e+00 -1.71017796e-01 6.47073090e-01 -4.47703116e-02
-5.04991710e-01 -1.04772009e-01 -8.24396133e-01 1.13084085e-01
6.98865712e-01 6.47998035e-01 5.86636603e-01 3.36099625e-01
4.62584198e-01 -3.61400396e-01 -9.11018729e-01 9.60369110e-01
3.82601440e-01 -8.48730147e-01 -3.00662499e-02 -1.45753846e-01
8.11599553e-01 1.52092487e-01 9.40648019e-02 4.27725576e-02
4.69358444e-01 -1.13751364e+00 4.36381608e-01 1.02405453e+00
5.77449679e-01 -5.46743274e-01 6.39610648e-01 3.60064715e-01
-6.66034997e-01 6.97783306e-02 -5.89395225e-01 -1.41351700e-01
4.24042083e-02 9.55874979e-01 -7.47887492e-01 3.20245802e-01
8.46450269e-01 6.17350340e-01 -9.18260887e-02 8.77579749e-01
-4.83458549e-01 5.07808387e-01 -3.89538378e-01 5.12643754e-01
8.68665799e-02 -7.30114400e-01 4.85875309e-01 1.26375890e+00
3.15583348e-01 -1.08780220e-01 8.09746534e-02 1.01644576e+00
-6.16213456e-02 -4.79645401e-01 -7.83322871e-01 3.53157192e-01
-1.94304600e-01 1.42359853e+00 -6.21036708e-01 -2.97582895e-01
-7.40419090e-01 1.32357466e+00 1.03836590e-02 8.89032125e-01
-9.04802203e-01 -1.65647820e-01 5.91481447e-01 3.80777866e-01
3.89299542e-01 1.46186100e-02 9.42934901e-02 -1.35413587e+00
1.21236965e-01 -7.22755313e-01 -8.16133842e-02 -1.36148703e+00
-1.26161468e+00 4.86750424e-01 -3.52136135e-01 -1.20504022e+00
5.26067555e-01 -7.83555210e-01 -9.03816938e-01 1.06849885e+00
-1.97576451e+00 -1.13274074e+00 -8.49624574e-01 6.17634356e-01
3.37749720e-01 3.91192526e-01 4.60128665e-01 3.96452218e-01
-4.25612211e-01 1.29385591e-01 2.69618154e-01 1.20472141e-01
8.27316523e-01 -1.05590701e+00 1.64115205e-01 9.77421701e-01
8.85573849e-02 5.37473738e-01 7.08973229e-01 -3.70045424e-01
-1.30162644e+00 -1.20777726e+00 2.42191568e-01 -4.49168324e-01
2.30281010e-01 -4.84783411e-01 -8.27107072e-01 7.93568075e-01
4.45166886e-01 -3.62906158e-02 4.02476549e-01 -1.19439207e-01
-5.53977072e-01 -4.12465841e-01 -9.51308429e-01 9.24653649e-01
9.87349451e-01 -4.76277351e-01 -1.36869416e-01 4.31368172e-01
3.84636343e-01 -4.38019961e-01 -1.85651019e-01 1.63151264e-01
7.99100876e-01 -1.69132566e+00 1.31986725e+00 -4.71713841e-02
4.42920625e-01 -4.42601085e-01 -9.23186168e-02 -1.32816064e+00
-1.29038915e-01 -6.02584064e-01 2.13659942e-01 9.53542471e-01
3.36242795e-01 -8.04561377e-01 8.97786736e-01 4.20731127e-01
-1.68286383e-01 -1.21069051e-01 -2.60530561e-01 -4.52488482e-01
-2.70739675e-01 -3.24203879e-01 3.58006984e-01 7.26432443e-01
-9.95225370e-01 3.31795216e-01 -8.25666964e-01 5.36470652e-01
7.89987266e-01 4.32933599e-01 1.22024381e+00 -1.05664468e+00
-3.19852918e-01 -1.49373159e-01 5.72662540e-02 -1.33651900e+00
-2.21313592e-02 -4.30317998e-01 4.77021694e-01 -1.71524096e+00
1.31665945e-01 -1.39756709e-01 -2.37895578e-01 2.22235739e-01
-2.84398198e-01 7.43415475e-01 -1.40098408e-01 -7.64508620e-02
-4.40727592e-01 3.17460239e-01 1.39069223e+00 -5.49208000e-02
-1.97472647e-01 2.80222476e-01 -8.93350720e-01 1.07558095e+00
7.69230425e-01 -2.58221805e-01 -2.74330884e-01 -7.43053675e-01
4.89540666e-01 -2.46739879e-01 6.63818061e-01 -8.18182170e-01
-2.34180465e-01 -1.41930990e-02 8.74309957e-01 -6.54329419e-01
5.66482723e-01 -1.04482496e+00 1.45348683e-01 2.06091907e-02
-1.73877254e-01 -7.45751262e-01 1.03761390e-01 7.10389316e-01
-6.72340617e-02 -9.39793661e-02 8.61167967e-01 -5.17634332e-01
-4.05982286e-01 9.99775156e-02 -4.25764680e-01 -2.22837329e-01
5.40117800e-01 -4.92021233e-01 -4.37007844e-01 -7.18161643e-01
-5.13588011e-01 -2.65897721e-01 4.88643736e-01 7.53589645e-02
7.09927857e-01 -8.65020275e-01 -7.60640562e-01 3.05678368e-01
-2.28440180e-01 3.42902005e-01 3.48057538e-01 9.73532438e-01
-7.35051513e-01 -2.41975680e-01 6.47909120e-02 -5.64381480e-01
-1.35929501e+00 2.90157467e-01 7.42000222e-01 1.68297738e-02
-1.18850589e+00 9.53640759e-01 9.36160147e-01 -3.89744043e-01
1.09064624e-01 -2.46798947e-01 -1.16093367e-01 -3.32216054e-01
9.76898491e-01 1.91222966e-01 3.16387266e-02 -3.73111874e-01
-4.81280647e-02 6.22191548e-01 2.51089013e-03 1.29378125e-01
1.52495372e+00 -5.10128260e-01 -2.42317021e-01 3.40114623e-01
1.37631762e+00 4.44582313e-01 -1.64774644e+00 -2.69046962e-01
-5.23064613e-01 -6.98178053e-01 2.29241982e-01 -8.00785124e-01
-1.40259516e+00 1.09165204e+00 3.38552982e-01 2.50418156e-01
1.41508055e+00 -3.93632770e-01 5.61100841e-01 4.48806345e-01
-3.60957980e-02 -7.24258780e-01 1.81327581e-01 4.71096218e-01
8.49269211e-01 -1.35757971e+00 1.67117208e-01 -5.16478539e-01
-5.15524089e-01 1.50931978e+00 5.72080791e-01 -3.80890280e-01
5.65454781e-01 6.67067766e-01 7.64701903e-01 -3.01521599e-01
-5.03761590e-01 -3.59668016e-01 2.17803881e-01 9.08882082e-01
4.50558901e-01 -3.37364972e-01 5.98630130e-01 -1.70825496e-01
2.84793358e-02 -8.64623785e-02 8.12443793e-01 5.78349948e-01
-5.39238095e-01 -7.37860739e-01 -6.03035450e-01 1.98149189e-01
-4.19951886e-01 -2.84039408e-01 -7.02024639e-01 6.35367095e-01
4.22563665e-02 9.60726380e-01 2.96679214e-02 7.81828985e-02
1.41912058e-01 -2.67121434e-01 7.63985753e-01 -6.02215350e-01
-5.11054158e-01 3.84328842e-01 1.68281674e-01 -6.69996262e-01
-8.32994640e-01 -1.26198828e-01 -6.61870897e-01 -2.94446293e-02
-6.04994774e-01 -4.60938126e-01 6.16928160e-01 8.35098326e-01
-8.61172453e-02 6.23853683e-01 5.90725601e-01 -1.25550020e+00
-7.55858272e-02 -7.05510795e-01 -5.30194104e-01 4.11327064e-01
8.49350810e-01 -2.77988911e-01 -8.01091671e-01 8.36852118e-02] | [10.58234977722168, -2.7873902320861816] |
dd852ce6-b89c-4e55-ac08-86a6651f6dfb | fake-news-detection-as-natural-language | 1907.07347 | null | https://arxiv.org/abs/1907.07347v1 | https://arxiv.org/pdf/1907.07347v1.pdf | Fake News Detection as Natural Language Inference | This report describes the entry by the Intelligent Knowledge Management (IKM) Lab in the WSDM 2019 Fake News Classification challenge. We treat the task as natural language inference (NLI). We individually train a number of the strongest NLI models as well as BERT. We ensemble these results and retrain with noisy labels in two stages. We analyze transitivity relations in the train and test sets and determine a set of test cases that can be reliably classified on this basis. The remainder of test cases are classified by our ensemble. Our entry achieves test set accuracy of 88.063% for 3rd place in the competition. | ['Hung-Yu Kao', 'Timothy Niven', 'Kai-Chou Yang'] | 2019-07-17 | null | null | null | null | ['news-classification'] | ['natural-language-processing'] | [-7.84198567e-03 6.48677707e-01 -8.15315783e-01 -6.24738634e-01
-9.52652812e-01 -7.46806204e-01 1.08225441e+00 1.65493429e-01
-4.31434125e-01 1.24097359e+00 1.81041077e-01 -5.68459988e-01
-2.14626253e-01 -5.84298909e-01 -9.17697251e-01 -1.06971152e-01
-9.44753736e-03 1.09721696e+00 3.38378012e-01 -2.47037306e-01
2.83024788e-01 1.81675136e-01 -9.13011551e-01 1.05451691e+00
6.52662516e-01 1.05263627e+00 -7.63360143e-01 5.28299332e-01
2.15463266e-01 1.41460955e+00 -1.06028020e+00 -7.48779297e-01
-8.68879841e-04 -1.64849326e-01 -1.63242912e+00 -2.82524347e-01
3.04957390e-01 1.59839764e-02 -3.39832723e-01 9.15434718e-01
-5.51479794e-02 4.76711988e-02 7.72979259e-01 -1.44479358e+00
-5.26349485e-01 1.00975931e+00 8.13065190e-03 5.22123992e-01
5.25592864e-01 -3.28381240e-01 1.09370804e+00 -9.64285254e-01
1.12714815e+00 1.08598864e+00 5.65147936e-01 4.36435878e-01
-1.03457475e+00 -6.61662877e-01 1.14537373e-01 7.63642848e-01
-1.26615226e+00 -6.42395794e-01 2.67017722e-01 -2.87806958e-01
1.39598596e+00 2.55749345e-01 1.94589138e-01 1.38436115e+00
1.87476203e-01 9.51695979e-01 1.27126062e+00 -5.77218652e-01
1.46464437e-01 4.06284273e-01 8.06122720e-01 7.05533087e-01
3.59880537e-01 1.50549188e-01 -5.81024885e-01 -2.72854954e-01
-7.62417307e-03 -8.11086118e-01 -1.70167625e-01 5.99962234e-01
-1.13807034e+00 8.37099373e-01 4.37878311e-01 4.32882756e-01
-8.63810629e-02 -6.80263042e-02 2.36101419e-01 7.77249455e-01
9.10699010e-01 1.16855907e+00 -9.45657194e-01 2.02746671e-02
-8.49109769e-01 4.25869286e-01 1.44827104e+00 9.82298315e-01
2.28075087e-01 -4.83760655e-01 -1.28699988e-01 7.76834071e-01
1.59763753e-01 7.41582885e-02 3.72272521e-01 -8.57245684e-01
6.24031663e-01 3.53550792e-01 3.31209093e-01 -9.48942184e-01
-4.32499379e-01 -4.04073000e-01 -4.29180682e-01 -1.57208800e-01
5.30208886e-01 -1.40862793e-01 -9.77443278e-01 1.38284636e+00
2.53955759e-02 3.98341238e-01 2.45126903e-01 6.43494308e-01
1.02556133e+00 6.81220949e-01 3.64338011e-02 -1.82910100e-01
1.20730305e+00 -1.09282875e+00 -7.21066117e-01 -3.33358884e-01
1.08652413e+00 -7.97522724e-01 2.42684841e-01 7.17064381e-01
-8.50077093e-01 -1.10317893e-01 -1.28004432e+00 -3.31370421e-02
-7.50422597e-01 1.32805005e-01 5.97030163e-01 1.13064297e-01
-7.60742605e-01 7.12862074e-01 -5.06060779e-01 9.13210306e-03
4.31906402e-01 3.22983980e-01 -4.01543617e-01 -2.21410722e-01
-1.77809238e+00 1.66003907e+00 6.81591570e-01 1.29128009e-01
-7.38803685e-01 -4.34300065e-01 -6.63178086e-01 -3.20909768e-01
4.75162745e-01 -2.94008821e-01 1.37254643e+00 -5.76101124e-01
-1.15286064e+00 1.24774361e+00 -2.14551792e-01 -8.71048212e-01
7.03598917e-01 -1.09430619e-01 -9.53725517e-01 -1.44171953e-01
4.00516748e-01 2.28468195e-01 3.49479675e-01 -1.29658246e+00
-9.77900088e-01 -1.95923299e-02 -6.85678944e-02 -2.33170763e-01
3.06009382e-01 2.65560478e-01 -2.06557870e-01 -3.13524127e-01
3.54735434e-01 -8.55934262e-01 2.89681196e-01 -8.44538987e-01
-8.74775112e-01 -6.02005839e-01 8.02867115e-01 -7.93018878e-01
1.14573789e+00 -1.63460243e+00 -1.08201705e-01 4.95054930e-01
3.32828820e-01 3.06175947e-01 -2.03898251e-02 -2.61226650e-02
-1.74222127e-01 5.62197983e-01 3.27550650e-01 -2.35664755e-01
-3.63590717e-02 3.52665246e-01 -6.18372679e-01 2.79363066e-01
4.70239103e-01 9.27792728e-01 -1.00035822e+00 -4.33173388e-01
-2.25107506e-01 -2.66194850e-01 -3.52444977e-01 -6.61773980e-02
-4.38223958e-01 4.91719157e-01 -4.09814835e-01 7.57652760e-01
3.34569365e-01 -4.27380651e-01 2.40123272e-01 -1.92528531e-01
2.84066170e-01 1.08388448e+00 -7.46656775e-01 9.18002903e-01
-3.04786801e-01 8.59598935e-01 -4.75737780e-01 -1.20053828e+00
7.12537766e-01 4.86251354e-01 -6.57798350e-03 -4.66533303e-01
-5.81302196e-02 4.91663992e-01 1.36126682e-01 -7.32640564e-01
3.30942243e-01 -1.50179699e-01 -3.74046206e-01 3.14276129e-01
4.97073948e-01 -5.22978380e-02 1.10662617e-01 1.77529931e-01
1.43246841e+00 3.90466228e-02 3.22526217e-01 -3.27544153e-01
2.24594727e-01 5.67048311e-01 7.15356886e-01 1.20367241e+00
-1.22112684e-01 1.63379565e-01 7.08545029e-01 -8.31070781e-01
-1.03476942e+00 -7.98884690e-01 -4.53974664e-01 8.83410931e-01
4.00381759e-02 -5.28785646e-01 -1.41802326e-01 -1.24157107e+00
-5.43487854e-02 1.39161897e+00 -6.60436571e-01 -2.85882205e-01
-4.60998207e-01 -8.59492362e-01 9.07749712e-01 1.31819576e-01
5.43822527e-01 -1.10088360e+00 4.19611335e-01 1.91300318e-01
-7.11652637e-01 -1.46312535e+00 2.71975994e-01 3.47623378e-01
-3.03901553e-01 -1.10167217e+00 3.38585705e-01 -9.33984518e-01
4.02766764e-01 -4.06007946e-01 1.51482224e+00 2.14726016e-01
7.12282211e-02 -2.93822378e-01 -6.09470606e-01 -5.74476957e-01
-9.83694315e-01 2.77000427e-01 2.02772930e-01 -2.71528274e-01
7.46643662e-01 -1.01548649e-01 1.48071870e-01 5.65875769e-01
-4.16861892e-01 9.59910676e-02 2.63272852e-01 1.13898265e+00
3.84311736e-01 4.00350362e-01 8.89559805e-01 -1.34796357e+00
4.72740710e-01 -8.69805336e-01 -4.77497280e-01 6.59428895e-01
-6.67197466e-01 -1.05666891e-01 4.95004147e-01 -1.92040727e-01
-9.63878334e-01 -4.66105133e-01 3.04167196e-02 1.77727953e-01
-3.14588338e-01 9.58610058e-01 6.50412589e-02 8.25180635e-02
9.36052084e-01 -3.45160097e-01 -5.51078260e-01 -3.02395225e-01
2.19745964e-01 1.03196299e+00 5.24227560e-01 -6.77704275e-01
5.91250598e-01 7.94709101e-02 -2.82047808e-01 -5.57710648e-01
-1.83151841e+00 -3.76778305e-01 -4.88529325e-01 -9.55678299e-02
7.18234420e-01 -7.14052916e-01 -5.97530007e-01 2.50624537e-01
-1.46291173e+00 -3.09951723e-01 2.33366666e-03 6.72076225e-01
-2.60972112e-01 -2.24265859e-01 -8.29589605e-01 -6.51489913e-01
-1.07716106e-01 -8.48621666e-01 6.77195251e-01 -1.70897320e-01
-6.33857191e-01 -9.12713289e-01 -1.34303078e-01 8.73385847e-01
4.28889580e-02 1.69853061e-01 9.50434029e-01 -1.58117700e+00
-3.40284258e-01 -8.41542006e-01 -7.04564974e-02 4.17782813e-01
-3.07564318e-01 -5.78469671e-02 -1.03216600e+00 8.02651867e-02
-6.48411512e-02 -8.62273574e-01 1.00952244e+00 4.43517342e-02
9.56757963e-01 -5.70001900e-01 -5.92736602e-01 1.21501185e-01
1.00608218e+00 5.89104146e-02 6.34402335e-01 5.74681461e-01
4.61205900e-01 5.84447265e-01 6.45648718e-01 -9.69113111e-02
3.57165158e-01 5.74960947e-01 -4.17393818e-02 2.79733270e-01
2.05274552e-01 -2.72532463e-01 1.08554848e-01 7.01878607e-01
-1.66862309e-01 -5.28402328e-01 -1.29613566e+00 3.36635381e-01
-2.04247236e+00 -1.10740972e+00 -1.61085397e-01 1.76277792e+00
1.13007259e+00 7.43425906e-01 -3.19593102e-01 1.30700007e-01
5.53703010e-01 -2.07575813e-01 -1.67094562e-02 -4.27962214e-01
-3.39915693e-01 3.17758322e-01 6.11415088e-01 9.33864892e-01
-1.39927959e+00 1.34401107e+00 7.36967850e+00 1.05685210e+00
-4.87991899e-01 2.22714335e-01 8.73892665e-01 2.09390316e-02
-9.48329717e-02 1.56761557e-01 -1.08861387e+00 3.59055609e-01
1.39023387e+00 -3.43474224e-02 3.61721158e-01 5.70170581e-01
-1.05154254e-01 -2.73186445e-01 -1.26621401e+00 3.83985966e-01
2.42144801e-02 -1.71241832e+00 -8.60025585e-02 -7.41784051e-02
1.01804435e+00 4.05128121e-01 -4.01543885e-01 7.86783457e-01
7.01365888e-01 -1.42039454e+00 7.61050045e-01 7.04662979e-01
5.37048936e-01 -5.49772084e-01 1.32841587e+00 6.48174047e-01
-1.82276741e-01 -1.76011864e-02 -9.08287764e-02 -1.96900800e-01
-4.98641282e-03 7.31313944e-01 -1.21692920e+00 6.68790340e-01
3.82052690e-01 6.77100956e-01 -5.47886550e-01 8.30058455e-01
-6.49637461e-01 1.11668444e+00 -4.90858525e-01 -1.66156009e-01
1.52724013e-01 1.13396406e-01 5.65996110e-01 1.26677179e+00
-3.67244601e-01 3.98687690e-01 2.68031597e-01 7.41522908e-01
-4.66713130e-01 -4.97644395e-01 -5.78978777e-01 -1.72079921e-01
5.82605183e-01 8.09838653e-01 -3.51312220e-01 -9.03814495e-01
-2.22543269e-01 7.14754462e-01 5.32922745e-01 3.63866717e-01
-7.75648892e-01 -1.02690838e-01 2.01445445e-01 -2.74577320e-01
-1.69460163e-01 1.33185415e-02 -5.02414525e-01 -1.47412765e+00
2.44778432e-02 -8.27997148e-01 4.84102190e-01 -6.95850313e-01
-1.71714008e+00 7.05872893e-01 9.22911391e-02 -1.03230846e+00
-4.32864159e-01 -7.92620718e-01 -3.64855498e-01 7.33722627e-01
-1.36572838e+00 -1.10577047e+00 1.24934316e-02 4.74722497e-02
1.46887019e-01 -3.48267794e-01 1.17452490e+00 3.38177770e-01
-5.34419179e-01 6.86411858e-01 -1.52256833e-02 5.85117519e-01
8.06988120e-01 -1.12662756e+00 4.46882904e-01 4.81809318e-01
3.04459125e-01 4.36568439e-01 7.90503561e-01 -8.76259446e-01
-5.55560708e-01 -1.04200816e+00 1.76169288e+00 -1.10448146e+00
1.01617277e+00 -2.06261292e-01 -8.55395317e-01 1.21516168e+00
-1.59275636e-01 3.09572071e-02 6.20562851e-01 3.99351478e-01
-6.82891607e-01 3.77889842e-01 -1.09019053e+00 2.98711210e-01
8.20192635e-01 -5.51085353e-01 -1.11488628e+00 1.00250387e+00
6.10614002e-01 -6.31902695e-01 -8.71047735e-01 4.40929770e-01
3.50964874e-01 -3.22671443e-01 6.63795114e-01 -1.55892754e+00
7.61766553e-01 -2.75494754e-01 -1.48224786e-01 -1.26885259e+00
-1.69401020e-01 -3.32734346e-01 -5.65630160e-02 9.43644345e-01
1.30979538e+00 -7.30257571e-01 5.29087722e-01 7.49870539e-01
1.30614862e-01 -8.26113522e-01 -1.08219910e+00 -9.54312980e-01
1.77795500e-01 -7.17372239e-01 3.39587145e-02 1.40003741e+00
3.69467705e-01 6.35213435e-01 -3.14515412e-01 2.22935140e-01
5.72473645e-01 5.37497029e-02 4.17522550e-01 -1.19917178e+00
-3.17100316e-01 -3.56574297e-01 -5.82299650e-01 -6.93172276e-01
4.51380461e-01 -1.30624211e+00 -1.40580768e-02 -1.43389094e+00
1.36983797e-01 -3.47378373e-01 -1.68536767e-01 8.03466558e-01
7.79569373e-02 3.64965767e-01 -1.70051575e-01 4.67111349e-01
-8.49610329e-01 -2.98977178e-02 7.50006437e-01 -2.85939217e-01
4.77913022e-02 1.59115985e-01 -5.30892134e-01 7.33080924e-01
9.52529907e-01 -6.74286246e-01 1.42444149e-01 -3.43522519e-01
2.67701358e-01 -9.35511198e-03 4.46184784e-01 -6.10938072e-01
9.34320763e-02 -2.16683075e-01 5.54571390e-01 -3.63637358e-01
1.23923771e-01 -2.58982837e-01 -9.92434695e-02 1.45489126e-01
-8.30861390e-01 -4.92380589e-01 1.10037595e-01 3.43956828e-01
-2.79545426e-01 -3.88293982e-01 6.66659057e-01 7.32193440e-02
-6.09658122e-01 -1.20315097e-01 -3.32843840e-01 3.46535414e-01
7.70339787e-01 2.94244915e-01 -9.47518528e-01 -2.76911169e-01
-1.29784548e+00 3.83175910e-01 -2.35833421e-01 4.26805705e-01
2.58293450e-01 -1.28844714e+00 -1.11754715e+00 -1.65399119e-01
2.40564436e-01 -1.26904175e-01 -3.08842003e-01 9.16634023e-01
-3.79895210e-01 6.14142656e-01 3.29488844e-01 -1.16605379e-01
-7.79095769e-01 2.10735008e-01 4.21040535e-01 -6.79265440e-01
-2.78381646e-01 9.62961733e-01 -6.36947393e-01 -4.43823487e-01
4.79396805e-02 -5.56419864e-02 -1.87930793e-01 -4.46372889e-02
5.79217911e-01 1.59118921e-01 3.34978104e-01 -4.25832719e-01
-5.43957531e-01 -1.37395307e-01 -3.61217082e-01 -3.46804172e-01
1.37400532e+00 3.07643920e-01 -3.19139272e-01 6.05847418e-01
1.39871383e+00 -2.03803241e-01 -2.12345630e-01 -5.55346847e-01
6.83529794e-01 -1.85341313e-01 1.18582882e-01 -1.60168171e+00
-4.50257629e-01 -1.66186001e-02 -2.53507167e-01 3.87161046e-01
3.81472379e-01 3.70807737e-01 4.98514086e-01 7.75744259e-01
4.68106270e-01 -1.31682992e+00 -4.29227084e-01 9.36090469e-01
1.05700564e+00 -1.41015840e+00 6.81905076e-02 -4.92132246e-01
-4.45000112e-01 9.65241313e-01 5.02045393e-01 -4.40636218e-01
8.19852769e-01 3.16828400e-01 -6.62125647e-02 -4.88355786e-01
-1.12053716e+00 1.50806367e-01 5.51889241e-01 2.48999223e-01
3.62354368e-01 2.83949763e-01 -6.35012984e-01 7.84021795e-01
-4.15188879e-01 6.21916763e-02 4.30706799e-01 8.24883759e-01
-2.95686513e-01 -7.30012774e-01 -9.53748003e-02 8.81219923e-01
-4.02981907e-01 -1.96034715e-01 -7.32368469e-01 8.02006543e-01
1.47340715e-01 1.30020773e+00 -1.69875905e-01 -9.49869037e-01
3.87976825e-01 5.10371864e-01 2.78083920e-01 -6.26192153e-01
-5.05983233e-01 -4.81236130e-01 1.20247471e+00 -5.16472876e-01
-3.68482083e-01 -3.87786925e-01 -1.06509554e+00 -5.20279765e-01
-6.35903120e-01 4.04485315e-01 5.38083375e-01 1.64374971e+00
7.18924180e-02 4.05203670e-01 3.91902715e-01 -2.63889402e-01
-6.86463296e-01 -1.28439462e+00 -3.06302637e-01 5.13075233e-01
2.61894703e-01 -6.17831409e-01 -6.23622715e-01 -1.23097179e-02] | [9.637542724609375, 8.750138282775879] |
983868dd-054b-4594-8dda-9b4b7dc66b81 | efficient-compressed-ratio-estimation-using | 2211.04284 | null | https://arxiv.org/abs/2211.04284v3 | https://arxiv.org/pdf/2211.04284v3.pdf | Efficient Compressed Ratio Estimation Using Online Sequential Learning for Edge Computing | Owing to the widespread adoption of the Internet of Things, a vast amount of sensor information is being acquired in real time. Accordingly, the communication cost of data from edge devices is increasing. Compressed sensing (CS), a data compression method that can be used on edge devices, has been attracting attention as a method to reduce communication costs. In CS, estimating the appropriate compression ratio is important. There is a method to adaptively estimate the compression ratio for the acquired data using reinforcement learning (RL). However, the computational costs associated with existing RL methods that can be utilized on edges are often high. In this study, we developed an efficient RL method for edge devices, referred to as the actor--critic online sequential extreme learning machine (AC-OSELM), and a system to compress data by estimating an appropriate compression ratio on the edge using AC-OSELM. The performance of the proposed method in estimating the compression ratio is evaluated by comparing it with other RL methods for edge devices. The experimental results indicate that AC-OSELM demonstrated the same or better compression performance and faster compression ratio estimation than the existing methods. | ['Noboru Koshizuka', 'Hangli Ge', 'Hiroki Oikawa'] | 2022-11-08 | null | null | null | null | ['data-compression'] | ['time-series'] | [ 2.61811823e-01 -1.34694427e-01 -2.39207223e-01 -1.29666761e-01
-5.41782022e-01 2.44330466e-01 1.05539225e-02 1.84813917e-01
-3.83629173e-01 5.40652037e-01 8.28488022e-02 -8.78093578e-03
-2.17402935e-01 -7.95833290e-01 -5.83537340e-01 -6.82314515e-01
-3.36305462e-02 8.05795845e-03 -7.48189315e-02 1.08199030e-01
3.42632771e-01 9.02489647e-02 -1.35350013e+00 -8.09967071e-02
1.00985384e+00 1.77126575e+00 7.43349493e-01 5.39279103e-01
-1.72650427e-01 8.06605339e-01 -1.38001055e-01 -2.24723816e-01
3.39330226e-01 -7.30790854e-01 -2.62425393e-01 1.14481777e-01
-5.03254116e-01 -4.40076739e-01 -5.56790769e-01 1.14665043e+00
6.51237905e-01 7.53753930e-02 3.72903645e-01 -1.24460971e+00
-3.42148185e-01 7.87000775e-01 -8.20858479e-01 2.14554593e-01
2.90649444e-01 -1.34254739e-01 5.03291905e-01 -6.81231797e-01
2.14967281e-01 7.45245814e-01 7.29179442e-01 2.65164673e-01
-6.77456558e-01 -5.17922997e-01 -2.88210392e-01 6.23315215e-01
-1.54731095e+00 -5.26313066e-01 1.13057387e+00 2.86599211e-02
7.53657341e-01 2.54442263e-02 1.00863218e+00 6.30117953e-01
5.62906861e-01 9.29099381e-01 9.63628650e-01 -7.07561672e-01
8.58245492e-01 -5.79077043e-02 -4.56556559e-01 4.83637422e-01
2.41252899e-01 1.83255106e-01 -4.13485467e-01 -2.77559366e-02
5.30756593e-01 3.23500961e-01 -5.33873867e-03 1.89122651e-02
-6.86770797e-01 7.37970233e-01 3.83887917e-01 2.73398191e-01
-8.03688824e-01 2.67548352e-01 6.53522611e-01 7.16385990e-02
3.43879789e-01 1.53246976e-03 -2.45947912e-01 -5.00848472e-01
-1.00824058e+00 -2.62713611e-01 6.82485938e-01 1.00480950e+00
1.83403894e-01 1.96826890e-01 5.42785712e-02 7.14674234e-01
3.27377111e-01 5.05989432e-01 8.63193512e-01 -1.15397334e+00
5.07063508e-01 5.62013984e-01 3.28701064e-02 -1.13780272e+00
-2.72222102e-01 -3.42461526e-01 -1.44543755e+00 -1.46743596e-01
-2.76878685e-01 -4.00777668e-01 -1.51719928e-01 1.47265732e+00
1.67357996e-01 5.35312653e-01 1.92187682e-01 8.05936337e-01
2.40703970e-01 9.11139488e-01 1.17934801e-01 -7.60667801e-01
1.00296474e+00 -6.34719253e-01 -1.04693806e+00 -2.43370056e-01
3.33791584e-01 -4.88423824e-01 8.42491984e-01 6.19703352e-01
-1.17416501e+00 -6.02760136e-01 -1.26102912e+00 3.73798817e-01
-8.43970850e-03 2.16347754e-01 5.15983105e-01 5.34796953e-01
-7.86350965e-01 7.65475750e-01 -9.73940015e-01 -1.31229754e-03
5.82639217e-01 2.58204848e-01 2.24168763e-01 -1.59592256e-01
-1.23534572e+00 4.92779583e-01 6.48096025e-01 4.45482768e-02
-4.62093443e-01 -2.81929910e-01 -6.69933319e-01 4.83524114e-01
3.13929737e-01 -3.15026820e-01 1.00764894e+00 -1.00803983e+00
-1.80836296e+00 2.24880517e-01 2.05730602e-01 -8.18869114e-01
4.34156328e-01 -7.11765438e-02 -5.53707600e-01 4.63192940e-01
-3.17098647e-01 4.75384980e-01 9.96701598e-01 -7.15346038e-01
-5.30934513e-01 -4.21323955e-01 -4.74722624e-01 2.17662305e-01
-7.91357994e-01 -4.65076834e-01 -1.41292274e-01 -6.18995070e-01
1.88095093e-01 -8.04839730e-01 -2.79618174e-01 2.13460758e-01
2.38206424e-02 -1.75361171e-01 1.19743764e+00 -8.98752987e-01
1.55378056e+00 -2.08616161e+00 1.55897915e-01 5.96578121e-02
-2.84437527e-04 3.69387627e-01 3.20664644e-01 3.81269693e-01
3.93424451e-01 -1.10599265e-01 -3.55844080e-01 -2.76363760e-01
-3.23466092e-01 7.22810030e-02 4.23999429e-02 3.19995582e-01
-3.88047934e-01 7.07347512e-01 -7.84820557e-01 -7.64373720e-01
3.86506766e-01 6.14669561e-01 -4.53399718e-01 4.24234807e-01
-1.54643431e-01 3.84759218e-01 -6.20549142e-01 5.59244275e-01
5.63384831e-01 -5.38309276e-01 1.77281439e-01 -2.17146084e-01
1.51593268e-01 -7.16138899e-01 -1.33426845e+00 1.80018640e+00
-7.55394459e-01 5.78355968e-01 5.04348427e-02 -1.30643940e+00
1.22791541e+00 2.96791106e-01 9.94824111e-01 -9.20401454e-01
5.56969941e-01 2.54036576e-01 -3.88411909e-01 -1.02856779e+00
3.19970608e-01 -2.84301024e-02 6.91644624e-02 3.52307707e-01
-4.66720849e-01 -6.72085211e-02 -2.42011137e-02 3.03259753e-02
1.10065866e+00 -4.07057479e-02 5.15998960e-01 1.95259869e-04
5.55357933e-01 -5.24561703e-01 6.43330812e-01 3.20281416e-01
-1.58307344e-01 8.01217626e-04 -2.40217805e-01 -3.71840239e-01
-1.20247865e+00 -5.37191749e-01 -7.15092793e-02 3.51989359e-01
5.47190964e-01 -4.63669151e-01 -8.26373637e-01 -5.61842397e-02
-2.24272445e-01 7.56677866e-01 -5.49008744e-03 -6.32477045e-01
-1.33668408e-01 -5.12380004e-01 1.65800363e-01 5.98755121e-01
1.31300819e+00 -1.05672646e+00 -1.29360497e+00 3.49800199e-01
-2.64797598e-01 -1.16382313e+00 -4.02650625e-01 -1.66039839e-01
-1.21194220e+00 -6.25626028e-01 -5.58390021e-01 -6.29877567e-01
5.26963174e-01 1.10785961e-01 7.46950388e-01 -5.32106794e-02
-2.25962520e-01 2.90734112e-01 -6.62348747e-01 -4.59332049e-01
-2.80253172e-01 -5.75715564e-02 -3.94238299e-03 2.75779635e-01
3.20848882e-01 -6.59040034e-01 -9.32550371e-01 4.64233868e-02
-9.64627981e-01 2.38074988e-01 7.77660549e-01 6.72213316e-01
9.80617344e-01 7.60733068e-01 9.53598917e-01 -8.59636307e-01
7.06625700e-01 -7.00049579e-01 -7.56104290e-01 1.80973530e-01
-1.10558188e+00 2.12828264e-01 1.01263332e+00 -4.65434253e-01
-9.27394032e-01 1.33957297e-01 -1.21982060e-01 -6.14976585e-01
3.02518994e-01 5.86509228e-01 -2.19870254e-01 -5.37321866e-02
2.12770447e-01 2.21322820e-01 4.35859300e-02 -2.51371771e-01
-4.07079086e-02 1.37114024e+00 3.82124633e-01 -1.13130279e-01
1.82434861e-02 1.82823509e-01 2.06696227e-01 -9.07650292e-01
-4.85421896e-01 -2.17765585e-01 2.78151408e-02 -4.58972305e-01
6.76508069e-01 -8.17034721e-01 -9.30795252e-01 5.56027234e-01
-6.57535076e-01 -1.08184241e-01 -5.10824919e-01 7.92420030e-01
-8.59882891e-01 4.32705581e-01 -6.27172887e-01 -1.18430030e+00
-8.74938250e-01 -1.08492100e+00 7.68532753e-01 4.89680707e-01
1.34992778e-01 -7.38239050e-01 -3.54375362e-01 2.67181784e-01
7.57018566e-01 3.92356694e-01 6.55213714e-01 -1.17200524e-01
-4.55984980e-01 -5.94420612e-01 2.36468557e-02 2.94157296e-01
-3.47645320e-02 -4.53320295e-01 -4.87456948e-01 -5.37377179e-01
5.37051916e-01 -2.01294705e-01 2.81961501e-01 8.21323037e-01
1.86176014e+00 -5.13544917e-01 -2.98446923e-01 5.85670769e-01
1.78678644e+00 5.18430769e-01 7.58637965e-01 4.95086201e-02
2.38197699e-01 -7.75244534e-02 6.66994870e-01 1.04436839e+00
1.66492239e-01 3.23199838e-01 5.16060889e-01 2.73511201e-01
1.66592762e-01 -5.45607924e-01 1.52946055e-01 1.26420438e+00
1.48904637e-01 -2.69998819e-01 -3.30187231e-01 6.43581599e-02
-1.82039595e+00 -7.80559897e-01 4.08918291e-01 2.27201128e+00
5.33489227e-01 3.25852782e-01 -3.57756913e-02 7.94599235e-01
1.05550909e+00 -1.64740816e-01 -1.15883648e+00 -2.94266403e-01
2.82257974e-01 -2.02828422e-01 5.65549731e-01 1.57527402e-01
-6.16632462e-01 1.72769070e-01 5.75510120e+00 8.66123199e-01
-1.10469687e+00 -3.50088091e-03 9.76273835e-01 5.25119789e-02
-5.73958531e-02 -1.08935930e-01 -4.58694398e-01 1.09050226e+00
1.30381751e+00 -2.08209261e-01 9.13574576e-01 1.09825242e+00
5.00038326e-01 -3.33465248e-01 -7.86806047e-01 1.61983788e+00
2.03076308e-03 -1.15690076e+00 -3.14543545e-01 2.66487096e-02
7.08554089e-01 -2.75448024e-01 -3.00411016e-01 9.43794250e-02
-3.37180912e-01 -4.70316410e-01 3.91219586e-01 6.55082226e-01
1.14811957e+00 -8.77704144e-01 9.10810471e-01 7.03692734e-01
-1.36567152e+00 -6.25276685e-01 -5.36094844e-01 -9.00312662e-02
3.91738832e-01 9.46045816e-01 -2.63272882e-01 1.72819182e-01
6.14414930e-01 7.54164636e-01 -1.28459483e-01 1.07824254e+00
1.83367699e-01 5.77144146e-01 -3.21571529e-01 -2.87437439e-01
-2.93186426e-01 -4.40707177e-01 3.44605893e-01 6.20785773e-01
8.38038266e-01 3.46017510e-01 2.73699135e-01 4.16087985e-01
-3.47106725e-01 1.63823754e-01 -5.46970546e-01 -2.17837393e-01
7.33507872e-01 1.12067711e+00 -7.74419844e-01 -3.06171954e-01
-4.51191574e-01 9.69587326e-01 1.31091401e-02 -7.63505995e-02
-8.25307131e-01 -4.79562432e-01 -3.50492656e-01 2.60402173e-01
1.08543865e-01 -1.32633820e-01 -2.02853650e-01 -8.18069518e-01
-5.59140928e-02 -4.86754239e-01 4.98037159e-01 -9.70720172e-01
-1.13163483e+00 4.55144018e-01 -1.36162668e-01 -1.61628664e+00
-1.39329538e-01 -4.54718247e-02 -4.80174541e-01 2.64180809e-01
-1.34457648e+00 -4.85726178e-01 -5.85208595e-01 5.36262333e-01
8.62251639e-01 -3.07178587e-01 6.05363309e-01 3.82259667e-01
-5.16243041e-01 5.53908706e-01 3.62144709e-01 -3.53129566e-01
-5.83690666e-02 -7.05966175e-01 -3.71534735e-01 6.44665420e-01
-1.21180989e-01 -2.92675406e-01 6.97990716e-01 -6.75817430e-01
-1.89759433e+00 -7.56529450e-01 3.56906086e-01 8.07060003e-01
2.36723348e-01 1.94255725e-01 -6.89401031e-01 4.53831911e-01
2.08591625e-01 2.86445051e-01 4.62019116e-01 -7.60802567e-01
5.02658546e-01 -7.02037036e-01 -1.61792028e+00 2.07564741e-01
8.58379364e-01 -1.74351692e-01 -8.64866003e-02 1.33089349e-01
7.29734659e-01 -3.72472227e-01 -1.10963619e+00 3.51237655e-01
4.03347254e-01 -8.32286716e-01 6.74722612e-01 1.51138157e-01
5.36241412e-01 4.49348576e-02 -2.94248760e-01 -1.25322676e+00
-6.49372339e-02 -6.17220938e-01 -9.61728692e-01 1.07542324e+00
-1.92660749e-01 -4.45432395e-01 8.96227956e-01 8.05549383e-01
1.48994789e-01 -1.12145674e+00 -9.84969020e-01 -5.80869079e-01
-6.68955505e-01 -2.39828125e-01 5.68706393e-01 4.87856001e-01
3.12848181e-01 2.69375175e-01 -4.58659083e-01 -1.69520214e-01
7.85309017e-01 6.01924770e-02 1.39403448e-01 -1.15997767e+00
-6.12929285e-01 -1.23444915e-01 -5.73132098e-01 -9.09403861e-01
-1.66702226e-01 -8.63021672e-01 -1.68554425e-01 -1.36236298e+00
1.00555442e-01 -5.76958001e-01 -4.28524256e-01 -1.14690237e-01
8.20484292e-03 -1.60705656e-01 1.06192410e-01 3.80849093e-01
-7.68277824e-01 1.00613642e+00 1.12459183e+00 -1.19112834e-01
-3.42445284e-01 2.78637230e-01 -5.06381094e-01 6.27415061e-01
9.68398809e-01 -3.58630061e-01 -5.84239841e-01 -3.38350773e-01
3.75654459e-01 9.60447133e-01 -2.53052115e-01 -1.48614538e+00
5.88435173e-01 9.38652381e-02 5.53425431e-01 -4.82635289e-01
2.71901250e-01 -1.45395529e+00 4.04591650e-01 8.21806252e-01
-4.22752619e-01 1.50896326e-01 -1.74674302e-01 8.54294181e-01
-1.84168786e-01 -4.53307509e-01 7.49689758e-01 -1.07202925e-01
-6.26807749e-01 4.02671784e-01 -7.50493109e-02 -9.87916291e-02
1.30340850e+00 -2.03301430e-01 2.95900434e-01 -5.93267441e-01
-5.00742733e-01 2.59071976e-01 7.57836923e-02 1.19477466e-01
1.04394138e+00 -1.45095003e+00 -4.27696407e-01 3.14687461e-01
-1.67917266e-01 9.86971613e-03 3.88997585e-01 5.64234793e-01
-4.84503955e-01 -4.98731025e-02 -1.10261425e-01 -4.98137504e-01
-7.39340425e-01 8.99923801e-01 6.67416453e-02 -3.06119233e-01
-8.34319890e-01 1.36829644e-01 -8.90571535e-01 3.55227351e-01
3.23996574e-01 6.08214512e-02 -1.11718282e-01 -3.57764304e-01
6.26359403e-01 9.33595300e-01 -3.43346559e-02 -1.42197758e-01
1.03457160e-01 5.83067060e-01 4.38592046e-01 1.15917921e-01
1.53342438e+00 -6.75320685e-01 1.23722345e-01 3.89605105e-01
1.18179393e+00 -4.13431555e-01 -1.52725399e+00 -2.62777537e-01
-1.15002677e-01 -4.87192303e-01 6.31517947e-01 -3.84521008e-01
-1.43905520e+00 6.23906672e-01 1.19561768e+00 4.30866092e-01
1.73297381e+00 -3.05823952e-01 1.48397923e+00 2.27630779e-01
7.72035837e-01 -1.57446241e+00 1.02060750e-01 -3.98234010e-01
5.62012076e-01 -1.28176296e+00 1.87519416e-01 -7.69930929e-02
-4.65697855e-01 1.02284837e+00 4.52784628e-01 -1.79021582e-01
1.06755078e+00 5.31915307e-01 -5.50156355e-01 1.34429395e-01
-3.30783546e-01 5.44433475e-01 -4.33648795e-01 2.13584349e-01
6.60610497e-02 1.23671718e-01 -6.58711314e-01 6.29114568e-01
1.33042037e-01 6.53321922e-01 3.13666493e-01 1.07460034e+00
-4.38493639e-01 -8.01950097e-01 -3.32595199e-01 8.88199568e-01
-4.80250925e-01 2.12615520e-01 4.55995560e-01 -5.38560282e-03
-3.55685502e-02 1.10765719e+00 6.98477626e-02 -5.42692065e-01
3.47301899e-03 -2.62512296e-01 3.58724505e-01 1.98748499e-01
2.14850873e-01 -1.29596442e-01 -6.12322092e-01 -4.88969088e-01
-4.86534148e-01 -3.49019110e-01 -1.41891861e+00 -3.26123297e-01
-4.90800738e-01 2.38711700e-01 9.87084687e-01 9.45826232e-01
6.48225904e-01 5.24068415e-01 1.13291109e+00 -8.32563400e-01
-7.10552752e-01 -8.20460737e-01 -7.77405858e-01 3.82354856e-01
-1.12226129e-01 -4.21957195e-01 -2.44628713e-01 6.10980354e-02] | [11.288043022155762, -1.579903244972229] |
8dbfeac3-bf6c-4733-a23b-bfcafadf8f76 | fine-grained-visual-classification-of-plant | 2106.02141 | null | https://arxiv.org/abs/2106.02141v1 | https://arxiv.org/pdf/2106.02141v1.pdf | Fine-Grained Visual Classification of Plant Species In The Wild: Object Detection as A Reinforced Means of Attention | Plant species identification in the wild is a difficult problem in part due to the high variability of the input data, but also because of complications induced by the long-tail effects of the datasets distribution. Inspired by the most recent fine-grained visual classification approaches which are based on attention to mitigate the effects of data variability, we explore the idea of using object detection as a form of attention. We introduce a bottom-up approach based on detecting plant organs and fusing the predictions of a variable number of organ-based species classifiers. We also curate a new dataset with a long-tail distribution for evaluating plant organ detection and organ-based species identification, which is publicly available. | ['Gianfranco Doretto', 'Donald A. Adjeroh', 'Cole Henderson', 'Meghana Kovur', 'Ram J. Zaveri', 'Matthew R. Keaton'] | 2021-06-03 | null | null | null | null | ['organ-detection'] | ['medical'] | [ 8.77901092e-02 -4.61594582e-01 3.89051326e-02 -5.07277548e-02
-8.61027986e-02 -1.20198500e+00 5.38471162e-01 6.71166003e-01
-1.06307037e-01 4.01517689e-01 9.93686449e-03 -3.50801140e-01
-2.09893733e-01 -8.86569977e-01 -7.12093413e-01 -5.86515427e-01
-2.57118400e-02 3.95337313e-01 5.05573273e-01 -7.90121034e-02
3.84937942e-01 9.37172294e-01 -1.76230407e+00 3.54262471e-01
9.52072561e-01 1.06579220e+00 4.61231768e-01 7.13450670e-01
-6.85909569e-01 4.70507085e-01 -7.13446438e-01 -1.63689464e-01
4.79394011e-02 -1.49783537e-01 -7.01096356e-01 3.59611660e-01
4.46178347e-01 1.03814207e-01 2.49568611e-01 1.07753849e+00
3.56858194e-01 -1.92582503e-01 8.64351332e-01 -1.41249573e+00
-1.05110705e+00 6.00621939e-01 -6.75075769e-01 1.71123952e-01
4.20558602e-02 2.97827393e-01 9.51222062e-01 -7.53122985e-01
4.80122447e-01 1.49993539e+00 7.21011758e-01 8.01882297e-02
-1.74383032e+00 -2.20366985e-01 5.63241005e-01 2.92251348e-01
-1.56506038e+00 7.89527893e-02 6.13827169e-01 -1.00950778e+00
5.61593890e-01 4.87373948e-01 5.52686930e-01 8.35729063e-01
-9.87504944e-02 6.23676598e-01 8.90439689e-01 -3.65188390e-01
2.90528387e-01 1.27107695e-01 3.04723024e-01 4.53617215e-01
2.26213753e-01 7.19162449e-02 3.34013589e-02 -1.72095761e-01
5.50404191e-01 1.92679107e-01 -2.09796458e-01 -7.84439802e-01
-1.10206091e+00 8.04407120e-01 1.16249776e+00 4.37526673e-01
-5.15727818e-01 -1.40074044e-01 5.31647861e-01 -1.82395235e-01
2.50382513e-01 7.27913141e-01 -6.33470058e-01 3.11476111e-01
-8.06180954e-01 1.13023385e-01 7.83486545e-01 5.94511032e-01
5.73425710e-01 2.28173174e-02 -7.05974162e-01 8.53752494e-01
2.75479347e-01 2.13883623e-01 3.31843108e-01 -4.40431118e-01
-3.00035596e-01 9.54053104e-01 1.77369431e-01 -7.87650704e-01
-3.48199457e-01 -4.16051596e-01 -7.91866362e-01 5.48477530e-01
6.00575209e-01 2.76907772e-01 -1.29142988e+00 1.74452603e+00
3.66426885e-01 -1.83166817e-01 -4.90649670e-01 6.61244094e-01
9.86833215e-01 3.91397446e-01 5.56791663e-01 1.56869084e-01
1.41901720e+00 -8.69851828e-01 -3.54643136e-01 -1.54789118e-03
3.84106457e-01 -5.21883667e-01 1.01552153e+00 1.13484986e-01
-5.25420904e-01 -7.57627726e-01 -7.67604411e-01 5.57837263e-02
-1.12008262e+00 9.12059546e-02 5.33756077e-01 6.00833535e-01
-7.68285692e-01 6.14207268e-01 -4.58483547e-01 -6.24781907e-01
7.92163730e-01 1.21649757e-01 -4.87533897e-01 8.46087039e-02
-5.90719461e-01 9.61162627e-01 7.78079808e-01 1.89013466e-01
-8.16618323e-01 -9.57321405e-01 -6.03058398e-01 6.02776885e-01
4.13453907e-01 -4.15561050e-01 7.64367580e-01 -1.05141807e+00
-1.03155661e+00 9.84790444e-01 2.29705736e-01 -1.27701879e-01
3.68630946e-01 2.10838884e-01 -6.81797788e-02 -2.44374737e-01
7.86477253e-02 7.05596626e-01 7.35533595e-01 -1.38614714e+00
-4.84442949e-01 -5.84005058e-01 8.63614380e-02 -3.04373175e-01
-3.17563057e-01 -1.17054760e-01 8.54129624e-03 -5.60289383e-01
-1.78848028e-01 -1.09423852e+00 -3.99444193e-01 5.95113933e-01
-4.35968578e-01 -2.81645805e-02 8.75279248e-01 -5.27118564e-01
9.48104739e-01 -2.22918439e+00 3.79003346e-01 -8.47615600e-02
2.31790781e-01 5.28896034e-01 -4.79489982e-01 4.40738797e-01
-3.23725790e-01 5.56987524e-01 -2.22624883e-01 4.87054735e-02
-4.90742624e-02 1.42979160e-01 -2.29774475e-01 1.55028403e-01
6.33055091e-01 7.99063623e-01 -7.38160491e-01 -3.83106351e-01
5.87893605e-01 4.12111521e-01 -2.03509182e-01 2.67018050e-01
-3.28179002e-01 3.62731487e-01 -3.35400015e-01 9.68989491e-01
9.35522079e-01 -3.05173993e-01 -8.40856805e-02 -2.11579785e-01
-4.55122143e-01 -2.60314047e-01 -1.09077871e+00 1.13964927e+00
-3.43742877e-01 3.71910512e-01 2.67017543e-01 -8.21277857e-01
8.64187598e-01 1.75996833e-02 3.79786074e-01 -6.05832674e-02
1.19085699e-01 2.51079381e-01 2.24581316e-01 2.80316155e-02
2.64071077e-01 7.50955939e-02 -2.02514650e-03 -1.82728723e-01
3.02520692e-01 -3.34698975e-01 3.99179012e-01 1.66048650e-02
9.32611883e-01 1.77572191e-01 6.84331238e-01 -5.83437264e-01
4.80722517e-01 1.44433901e-01 5.64879239e-01 9.49100316e-01
-5.56818306e-01 7.18613565e-01 6.22165918e-01 -5.30830443e-01
-9.06782150e-01 -1.08388186e+00 -4.45432246e-01 1.34815896e+00
-7.36096948e-02 -3.11447859e-01 -3.60748470e-01 -8.22551429e-01
5.66022813e-01 5.08349836e-01 -1.13162363e+00 -2.94801295e-01
2.45019168e-01 -8.39991271e-01 3.00330639e-01 9.28698242e-01
2.61386465e-02 -1.39867961e+00 -6.01361454e-01 1.02989666e-01
7.75924996e-02 -1.05445373e+00 -1.12679996e-01 8.29312086e-01
-4.14666981e-01 -9.00664151e-01 -9.32342410e-01 -4.81705487e-01
4.80078876e-01 2.57320791e-01 1.40978527e+00 2.03447476e-01
-8.76139581e-01 2.52880976e-02 -4.01809126e-01 -7.94971824e-01
-4.34402347e-01 1.92076862e-01 -3.75926793e-01 1.37231156e-01
5.52904792e-02 -4.60888803e-01 -3.01438361e-01 8.49263072e-02
-8.41153145e-01 -1.29800647e-01 5.37349343e-01 9.68308628e-01
5.29540420e-01 -1.18064426e-01 3.53166431e-01 -7.08247125e-01
1.95145756e-01 -6.30516589e-01 -9.28019404e-01 5.87170780e-01
-8.57893378e-02 1.75207689e-01 6.41620338e-01 -6.69720590e-01
-6.73949361e-01 5.59183002e-01 7.04235807e-02 -5.03910244e-01
-5.37788212e-01 5.28077483e-01 -3.93015325e-01 -3.09897035e-01
7.74553061e-01 -1.67093500e-01 -1.80163458e-01 -5.25350690e-01
4.44456249e-01 6.03750229e-01 4.03184593e-01 -4.02107626e-01
7.37640619e-01 1.91548079e-01 2.84597099e-01 -9.47567284e-01
-9.61733401e-01 -4.79435027e-01 -1.07976997e+00 -2.40844056e-01
9.48339939e-01 -7.25669265e-01 -9.61197257e-01 6.14701748e-01
-1.16930771e+00 -1.71139032e-01 -4.91054744e-01 1.43209487e-01
-3.96479577e-01 4.07549828e-01 -2.25437388e-01 -8.11377347e-01
-2.48507783e-01 -1.04990160e+00 1.13850915e+00 3.39066207e-01
4.18803692e-02 -7.84020543e-01 -1.24373272e-01 -2.96018451e-01
5.58855474e-01 4.56143588e-01 1.13027310e+00 -5.87062359e-01
-4.68856424e-01 -4.08891916e-01 -7.80067325e-01 2.20933586e-01
3.50674778e-01 6.36000991e-01 -1.07786095e+00 -1.64691374e-01
-5.07387698e-01 -2.64867246e-01 9.42668736e-01 5.58559775e-01
1.43990648e+00 3.24786186e-01 -3.70477498e-01 6.06433034e-01
1.61207259e+00 -1.88991874e-01 1.64432779e-01 1.55903339e-01
1.01354635e+00 7.88028240e-01 3.80375385e-01 6.20044768e-01
8.55051503e-02 6.65182710e-01 1.03726268e+00 -3.88343751e-01
-9.61750150e-02 -9.17371139e-02 -3.39272499e-01 -6.45069703e-02
3.28265838e-02 -4.62061018e-01 -9.74835038e-01 6.60658538e-01
-1.90624428e+00 -9.56677437e-01 -4.06222254e-01 2.22431660e+00
4.79077876e-01 -1.80516347e-01 3.75349432e-01 1.37473777e-01
8.96682322e-01 -6.55590324e-03 -5.99911630e-01 -4.94753629e-01
-3.71857256e-01 -6.40854985e-02 6.22729599e-01 1.48256972e-01
-1.30267906e+00 9.08528566e-01 6.72706652e+00 7.13764429e-01
-1.35821199e+00 -4.11434829e-01 4.09847111e-01 4.33770180e-01
6.81738853e-02 8.41858760e-02 -6.25348389e-01 6.98406279e-01
4.90722090e-01 3.77913564e-02 2.56268322e-01 8.20270240e-01
-2.80021101e-01 -2.89710425e-02 -9.57937956e-01 7.22028732e-01
-2.17539325e-01 -1.16128504e+00 1.15651295e-01 1.00541197e-01
2.94874430e-01 -3.71518359e-02 -1.59982473e-01 2.77302116e-01
5.49738109e-01 -9.00139391e-01 8.23608756e-01 1.66637793e-01
4.85943168e-01 -3.40867102e-01 4.88891780e-01 3.98583531e-01
-1.44705629e+00 -5.46676755e-01 -4.26900387e-01 -3.25653739e-02
-4.62159067e-02 4.76258665e-01 -6.63289785e-01 3.67196679e-01
9.86682177e-01 5.62418640e-01 -1.18100202e+00 1.51285899e+00
4.42097895e-02 3.37969840e-01 -4.37906206e-01 -9.20444634e-03
5.32901399e-02 1.22440532e-01 3.61001551e-01 1.25226414e+00
3.20959508e-01 -5.09496391e-01 6.57547355e-01 1.20268798e+00
1.99835673e-01 2.11813003e-02 -5.76600075e-01 -3.34795982e-01
4.55020875e-01 1.73274124e+00 -9.56714153e-01 -1.11561216e-01
-3.22394997e-01 1.00215101e+00 5.07118344e-01 -2.39094514e-02
-5.05951345e-01 -1.30971164e-01 3.68377835e-01 1.16949551e-01
7.85809219e-01 -1.25368819e-01 -2.32927456e-01 -9.65751290e-01
-1.86862573e-01 -4.62822974e-01 5.71943343e-01 -8.81236255e-01
-1.77242160e+00 5.32927036e-01 -2.35563532e-01 -9.82457936e-01
1.59776121e-01 -9.79506552e-01 -5.61419964e-01 1.07829666e+00
-1.40498400e+00 -1.67792296e+00 -7.98011959e-01 3.56525362e-01
2.99877882e-01 1.93629310e-01 1.21231365e+00 2.39169240e-01
-4.77438450e-01 1.64537594e-01 1.19420394e-01 -7.21618310e-02
5.69898844e-01 -1.52016675e+00 3.00160289e-01 6.92579985e-01
2.10316449e-01 1.55281201e-01 6.28561974e-01 -3.91323417e-01
-1.00737226e+00 -1.13430762e+00 6.92344248e-01 -5.26151657e-01
8.40457261e-01 -5.55917561e-01 -1.18330383e+00 5.48551023e-01
-6.89246133e-02 5.63893020e-01 5.89715242e-01 2.82984555e-01
-5.11536956e-01 5.98972552e-02 -1.26881123e+00 3.14893961e-01
6.68100834e-01 -4.38442230e-01 -2.37614483e-01 2.68076688e-01
4.70234096e-01 1.69497579e-02 -8.00408423e-01 6.29543662e-01
6.05380177e-01 -5.55082142e-01 1.16801524e+00 -6.38574898e-01
2.22602084e-01 -4.80120003e-01 -3.67022723e-01 -1.57517731e+00
-1.11243761e+00 1.88884195e-02 2.26793036e-01 1.59703803e+00
1.41675949e-01 -1.54907838e-01 3.27544391e-01 2.71089673e-01
3.23986597e-02 -5.19590117e-02 -5.09978831e-01 -9.56520081e-01
1.48804396e-01 2.81894296e-01 6.42959535e-01 6.50828838e-01
-3.86803955e-01 3.58788580e-01 -1.34845063e-01 1.96460664e-01
3.84246796e-01 4.65982586e-01 5.44751048e-01 -1.90376198e+00
-1.72308996e-01 -7.24197030e-01 -7.15792358e-01 -3.02586645e-01
1.37809679e-01 -9.45926487e-01 4.99040186e-01 -1.24341857e+00
3.62437487e-01 -4.12284553e-01 -5.66345453e-01 4.71886903e-01
-4.37875807e-01 3.01437169e-01 6.28258109e-01 -1.04597555e-02
-3.69011253e-01 4.04633731e-01 9.74477768e-01 -2.09303156e-01
-1.39181420e-01 2.36596894e-02 -6.32285357e-01 6.43691778e-01
4.57142323e-01 -3.27178180e-01 3.62773091e-02 -2.84702629e-01
-2.94865936e-01 -1.85428560e-01 8.00093472e-01 -9.77332711e-01
-3.02309394e-01 -3.40007544e-01 7.16577828e-01 -7.78938890e-01
2.93101847e-01 -1.02361536e+00 -1.09832352e-02 6.35051250e-01
-3.44527364e-01 -2.89744325e-02 4.87069726e-01 4.64881003e-01
-1.05511464e-01 -3.04400623e-01 1.17285240e+00 -3.84826452e-01
-8.21576118e-01 4.21046078e-01 -3.91469479e-01 -2.64703840e-01
9.20091152e-01 -1.54287532e-01 -3.67345780e-01 2.00224131e-01
-7.29639113e-01 -1.99381225e-02 4.92524534e-01 7.76022434e-01
-1.20880075e-01 -1.10962403e+00 -7.85394967e-01 1.54355198e-01
7.16113985e-01 -3.89766783e-01 2.92730659e-01 5.07180691e-01
-3.95236313e-01 2.15180248e-01 -9.13353801e-01 -8.75109494e-01
-1.37485135e+00 1.23826134e+00 2.24911273e-01 -1.83635667e-01
-2.47832254e-01 8.46527994e-01 6.16144061e-01 -5.78299046e-01
6.48322776e-02 -1.88516170e-01 -4.25285727e-01 3.03076357e-01
2.10143626e-01 2.52840538e-02 8.28668177e-02 -6.62861049e-01
-7.20806837e-01 4.86646354e-01 1.51244223e-01 3.35484594e-01
1.28127885e+00 1.91279128e-01 -7.82515034e-02 6.53584957e-01
6.71475589e-01 -2.03191385e-01 -1.14256871e+00 -8.00795034e-02
2.24786565e-01 -6.57146394e-01 -2.35832692e-03 -1.21917462e+00
-8.95181954e-01 1.05592489e+00 9.55736697e-01 7.13571548e-01
9.68173683e-01 -3.86435539e-02 -1.03332907e-01 -1.25144377e-01
1.32599652e-01 -7.63740480e-01 -2.33575135e-01 4.42589611e-01
1.08157265e+00 -1.51177514e+00 -1.30560011e-01 -5.38063943e-01
-2.50800282e-01 9.73746717e-01 7.10669041e-01 -3.71339209e-02
8.14101398e-01 4.24326271e-01 1.38955703e-02 6.27375618e-02
-5.97380221e-01 -7.10673153e-01 4.06621784e-01 9.64849055e-01
5.80765545e-01 2.64941573e-01 -1.06795959e-01 4.48300362e-01
2.77007103e-01 4.08655182e-02 3.24825048e-01 8.96781802e-01
-4.42553371e-01 -1.12818134e+00 -4.47503537e-01 3.42363805e-01
-2.54681200e-01 -4.24014255e-02 -9.44763839e-01 4.78753805e-01
3.01153630e-01 6.30866885e-01 1.90396849e-02 -8.18012059e-02
4.25265670e-01 -8.40508714e-02 4.01739150e-01 -6.85386240e-01
-9.07796860e-01 -2.51022547e-01 -3.25032681e-01 -2.39601940e-01
1.16054085e-03 -6.42511487e-01 -7.38376558e-01 -1.64553523e-01
-5.84031701e-01 -4.82865930e-01 7.73052037e-01 5.72300315e-01
5.61367631e-01 7.96888947e-01 5.52526474e-01 -1.04860759e+00
-6.47399008e-01 -9.48092699e-01 -7.00750351e-01 7.96146095e-01
3.43921989e-01 -8.33668888e-01 -2.52272606e-01 -1.46815673e-01] | [9.583440780639648, 2.0053212642669678] |
33dc7e19-62c8-4c78-86d6-9364216d6008 | bayesian-optimisation-for-a-biologically | 2104.05989 | null | https://arxiv.org/abs/2104.05989v1 | https://arxiv.org/pdf/2104.05989v1.pdf | Bayesian Optimisation for a Biologically Inspired Population Neural Network | We have used Bayesian Optimisation (BO) to find hyper-parameters in an existing biologically plausible population neural network. The 8-dimensional optimal hyper-parameter combination should be such that the network dynamics simulate the resting state alpha rhythm (8 - 13 Hz rhythms in brain signals). Each combination of these eight hyper-parameters constitutes a 'datapoint' in the parameter space. The best combination of these parameters leads to the neural network's output power spectral peak being constraint within the alpha band. Further, constraints were introduced to the BO algorithm based on qualitative observation of the network output time series, so that high amplitude pseudo-periodic oscillations are removed. Upon successful implementation for alpha band, we further optimised the network to oscillate within the theta (4 - 8 Hz) and beta (13 - 30 Hz) bands. The changing rhythms in the model can now be studied using the identified optimal hyper-parameters for the respective frequency bands. We have previously tuned parameters in the existing neural network by the trial-and-error approach; however, due to time and computational constraints, we could not vary more than three parameters at once. The approach detailed here, allows an automatic hyper-parameter search, producing reliable parameter sets for the network. | ['Basabdatta Sen Bhattacharya', 'Elham Zareian', 'Jun Chen', 'Swapna Sasi', 'Mahak Kothari'] | 2021-04-13 | null | null | null | null | ['bayesian-optimisation'] | ['methodology'] | [ 3.15032214e-01 3.27971429e-01 1.83717459e-01 2.53976971e-01
-3.92400064e-02 -2.62623280e-01 4.95806187e-01 -5.14126793e-02
-7.39360631e-01 1.13532960e+00 5.97535484e-02 -4.90798429e-02
-7.24306881e-01 -4.45038319e-01 -2.96418965e-01 -1.10766041e+00
-6.76272571e-01 3.84864300e-01 5.63356817e-01 -2.49919727e-01
3.40294302e-01 4.78037477e-01 -1.72979748e+00 -8.38329867e-02
4.35393721e-01 5.07049024e-01 1.08157985e-01 6.26437366e-01
6.27995729e-01 -1.54229105e-01 -9.31689382e-01 2.47135848e-01
9.15112793e-02 -7.10416019e-01 -4.58187491e-01 -3.02795470e-01
-4.75309819e-01 2.65908718e-01 9.05875564e-02 8.62887204e-01
7.65158832e-01 3.12481552e-01 7.24638641e-01 -1.05192327e+00
1.67822003e-01 5.71304619e-01 -2.07056969e-01 5.50439358e-01
1.09063208e-01 2.59878933e-01 4.93315935e-01 -4.02566493e-01
5.63927948e-01 9.66675162e-01 8.32475185e-01 4.57318604e-01
-1.64840615e+00 -6.87900305e-01 -2.29150578e-01 2.47317310e-02
-1.75506628e+00 -5.53910196e-01 6.35230005e-01 -4.16740209e-01
1.32165778e+00 6.72219619e-02 1.38680089e+00 1.10615420e+00
8.04453254e-01 -3.09478521e-01 8.94287169e-01 -7.55830646e-01
5.80961585e-01 -6.25285581e-02 -2.24460945e-01 1.91613764e-01
2.58886755e-01 3.87223750e-01 -3.88372630e-01 -4.04247522e-01
9.52560067e-01 -8.15181851e-01 -4.28296059e-01 -2.44274195e-02
-1.35301888e+00 6.78096771e-01 1.11233696e-01 7.55089998e-01
-6.04929090e-01 3.37419182e-01 2.22322285e-01 1.56971425e-01
7.30027184e-02 8.32621515e-01 -5.70574999e-01 -1.94065630e-01
-7.97989786e-01 4.10382003e-01 8.58419895e-01 2.03247234e-01
4.82447118e-01 3.07574838e-01 2.09356129e-01 1.01800823e+00
5.99912226e-01 8.19782764e-02 7.37271726e-01 -1.12037361e+00
2.86600832e-02 1.57228187e-01 -8.79082903e-02 -9.80679035e-01
-8.77530038e-01 -5.50538421e-01 -8.36720407e-01 1.92734495e-01
7.07490861e-01 -5.18507123e-01 -7.38067925e-01 1.87975538e+00
2.45023742e-01 2.12569669e-01 8.26302320e-02 8.36336732e-01
1.95671394e-01 7.19439507e-01 8.37424845e-02 -8.78414512e-01
1.39427006e+00 1.58095449e-01 -8.66214514e-01 -1.78465143e-01
4.34947312e-01 -4.93340373e-01 6.98708296e-01 5.31863809e-01
-1.07741833e+00 -1.79462433e-01 -1.35960424e+00 8.74278605e-01
-2.64567941e-01 -2.36635923e-01 4.01627272e-01 6.68605804e-01
-1.09694684e+00 5.98665714e-01 -8.85685742e-01 -3.39950621e-01
-5.63781448e-02 8.23825777e-01 -1.55599952e-01 8.04870248e-01
-1.56218302e+00 1.16393089e+00 1.00761247e+00 5.00819087e-01
-3.75843704e-01 -4.23580736e-01 -3.21171582e-01 2.43564863e-02
-6.98347986e-02 -8.57126653e-01 8.32750022e-01 -9.94013906e-01
-1.84121060e+00 6.01657033e-01 8.71831924e-02 -3.11607301e-01
-1.50644043e-02 5.68929136e-01 -3.67497742e-01 3.08034390e-01
-6.43705785e-01 8.95842910e-01 6.14201546e-01 -1.28617799e+00
9.56642702e-02 -1.25470176e-01 -4.28787440e-01 2.55452633e-01
-2.25545391e-02 1.68065414e-01 -3.79128605e-02 -6.58144057e-01
4.35682297e-01 -1.10271823e+00 -8.87715444e-02 -4.73472565e-01
-2.85545867e-02 -8.81055743e-02 1.13864087e-01 -3.26298922e-01
1.28017652e+00 -2.03129292e+00 3.86686146e-01 8.02139580e-01
-7.58099109e-02 1.31333731e-02 1.54475912e-01 4.45296764e-01
-5.36118150e-01 -3.29592451e-02 -2.55795330e-01 4.55479950e-01
-2.56518424e-01 2.40597516e-01 1.13684803e-01 7.21219718e-01
-2.38623493e-03 3.76044631e-01 -5.72965205e-01 -3.44045132e-01
-1.88219771e-01 9.17687654e-01 -6.15658462e-01 -1.20779730e-01
-5.95580414e-02 1.96384728e-01 -1.04674816e-01 1.09712407e-01
1.66594937e-01 2.05829054e-01 1.96725026e-01 -1.09065264e-01
-2.13403076e-01 1.95474491e-01 -1.28016818e+00 1.01629341e+00
4.44495445e-03 8.24556828e-01 -7.19820932e-02 -9.62026417e-01
1.13050497e+00 7.00467050e-01 6.94408059e-01 -4.36808884e-01
4.02094215e-01 2.62669891e-01 9.58332360e-01 -3.64761978e-01
-5.46462424e-02 -3.18950981e-01 3.27526897e-01 3.40824097e-01
2.67456353e-01 -2.63037682e-01 2.66827226e-01 -3.09858859e-01
6.91455007e-01 8.15449655e-02 3.32700521e-01 -7.76767731e-01
3.22854966e-01 -2.81591386e-01 4.45533156e-01 5.17687976e-01
-1.65086105e-01 5.09030700e-01 8.72793674e-01 -9.87205952e-02
-1.04334939e+00 -5.33175230e-01 -6.49902344e-01 7.06824481e-01
-3.47605079e-01 -1.89199522e-01 -1.18241322e+00 3.98081273e-01
-5.19368351e-01 4.92036492e-01 -5.95865667e-01 -3.05383354e-01
-7.28438079e-01 -1.42564738e+00 5.03826618e-01 -2.65939802e-01
2.11151154e-03 -1.47181988e+00 -1.18267226e+00 6.42918289e-01
-9.32845287e-03 -4.97177690e-01 7.76703581e-02 9.02285218e-01
-1.12335396e+00 -8.45808744e-01 -6.13102794e-01 -6.35831118e-01
6.21873856e-01 -6.12788677e-01 7.20707297e-01 1.83307752e-01
-3.56798589e-01 -6.97910935e-02 5.14517762e-02 -4.59737450e-01
-3.46240431e-01 2.20074225e-03 2.66074300e-01 -3.92570704e-01
3.23674828e-02 -1.03892469e+00 -5.48621535e-01 3.28699797e-01
-7.93700576e-01 -1.45885855e-01 2.66113907e-01 8.10699522e-01
4.38031495e-01 4.08061385e-01 1.03890312e+00 -3.09761792e-01
8.85597646e-01 -4.48477566e-01 -6.66236162e-01 -1.48507968e-01
-4.84547794e-01 2.35239491e-01 2.94928938e-01 -1.06192529e+00
-6.91639483e-01 5.09087667e-02 -5.00944778e-02 -4.81121875e-02
-2.67567366e-01 5.60659945e-01 1.85042305e-03 -7.91243464e-02
1.06374252e+00 -2.38690511e-04 9.19904187e-02 3.95648666e-02
-1.86625287e-01 3.29099506e-01 5.12112021e-01 -4.97036368e-01
2.67382950e-01 -1.47004128e-01 1.09488383e-01 -9.41714883e-01
-7.77956322e-02 9.46510732e-02 -4.97039616e-01 -6.14025831e-01
6.91694975e-01 -6.79867625e-01 -8.70926321e-01 4.90532696e-01
-1.02137351e+00 -3.61565083e-01 4.89279293e-02 7.88034379e-01
-8.90921116e-01 -1.22418599e-02 -1.59767523e-01 -1.05517614e+00
-1.43495977e-01 -1.05356121e+00 3.76433343e-01 1.97165683e-01
-8.05021346e-01 -9.35850561e-01 3.94649178e-01 -3.63689572e-01
2.70862252e-01 4.14484262e-01 1.05540740e+00 -5.75095117e-01
-1.17735402e-03 -4.23601083e-02 4.74163145e-01 -3.10206592e-01
-3.06880027e-01 3.82734835e-01 -9.03299570e-01 -2.67117083e-01
3.91526222e-01 6.19095564e-02 4.64566708e-01 7.35192001e-01
7.40531981e-01 -4.27967012e-01 -3.13344628e-01 5.24305761e-01
1.28607368e+00 6.86753511e-01 7.92899549e-01 7.22985864e-01
-2.55431291e-02 8.16176176e-01 -1.40680894e-01 3.61351162e-01
-9.27732363e-02 7.72914410e-01 3.56688768e-01 3.67300391e-01
2.80990750e-01 2.89982468e-01 1.91532180e-01 6.56164825e-01
-3.47767889e-01 -1.23185135e-01 -1.01627088e+00 6.42672896e-01
-1.69161701e+00 -9.61369514e-01 -7.58175030e-02 2.32366824e+00
1.16334629e+00 6.46302998e-01 3.83735180e-01 4.97648716e-01
6.67341590e-01 -3.00624073e-01 -2.22544283e-01 -5.39448857e-01
-1.20258726e-01 1.16433725e-01 4.02189493e-01 5.38520932e-01
-5.13819635e-01 3.58301848e-01 7.46092987e+00 5.05548954e-01
-8.48550856e-01 -3.69941920e-01 2.77733415e-01 -4.37050909e-01
-7.85606131e-02 1.63796619e-02 -8.21751237e-01 6.80887461e-01
1.68097734e+00 -1.74446657e-01 8.55423272e-01 1.77881300e-01
6.32081032e-01 -7.26660788e-01 -7.88307607e-01 8.27581465e-01
-3.14817160e-01 -1.03905511e+00 -4.09440666e-01 3.75200093e-01
5.23838758e-01 -1.53520778e-01 -3.18692654e-01 -4.49384972e-02
-6.21223748e-02 -1.07426929e+00 6.12267673e-01 7.79083490e-01
2.94623315e-01 -1.13554466e+00 6.88073635e-01 5.35862029e-01
-7.26987660e-01 -1.56924739e-01 -4.08274770e-01 6.73906086e-03
4.73780669e-02 6.03212416e-01 -6.65151060e-01 -2.14399129e-01
6.08480811e-01 3.22555043e-02 -2.39161879e-01 1.36112273e+00
1.72557905e-01 5.72846591e-01 -9.59258616e-01 -2.07878083e-01
-3.50504257e-02 -3.22362572e-01 6.80051029e-01 1.16973603e+00
4.29226637e-01 3.39671195e-01 -5.23605824e-01 8.73539388e-01
5.12994051e-01 -2.26248100e-01 -4.48192686e-01 1.44427449e-01
7.75497139e-01 8.27572644e-01 -1.24844551e+00 1.45954579e-01
3.52199405e-01 8.07183906e-02 8.11104998e-02 5.51276863e-01
-5.75665712e-01 -6.80690646e-01 4.33453053e-01 -2.13121232e-02
1.80758625e-01 -1.55678168e-01 -3.46975505e-01 -5.83537579e-01
-2.49910280e-01 -9.07358825e-01 2.02864319e-01 -8.05244327e-01
-6.20964706e-01 6.41410649e-01 8.64672661e-01 -7.42419720e-01
-8.24223995e-01 -3.33566695e-01 -6.03941977e-01 1.06576180e+00
-6.77655160e-01 -2.75816888e-01 3.89496654e-01 4.59375799e-01
-1.04506858e-01 -1.38899967e-01 9.63113070e-01 -1.79569528e-01
-6.16677046e-01 3.02790850e-01 3.56245972e-02 -5.23590922e-01
2.71746039e-01 -8.96900356e-01 -1.54889449e-01 4.46872473e-01
-1.96128666e-01 8.89527857e-01 1.33784914e+00 -5.48277080e-01
-8.86480868e-01 -3.00756872e-01 9.74403918e-01 -4.13461924e-02
7.24746764e-01 -4.38832551e-01 -1.13758683e+00 1.54464722e-01
3.06323797e-01 -4.48594362e-01 5.91104448e-01 -9.41902250e-02
3.19612205e-01 1.57735541e-01 -1.00701177e+00 8.93541217e-01
5.84760666e-01 -9.88514721e-02 -6.89691007e-01 2.23591387e-01
1.98224783e-01 -2.46164039e-01 -1.16080725e+00 3.50251049e-01
6.34204388e-01 -8.80397201e-01 9.01069760e-01 -1.28883734e-01
-2.44871765e-01 -4.51632202e-01 3.50611329e-01 -1.63150263e+00
-3.95428926e-01 -1.01197803e+00 -1.27673745e-01 9.28388119e-01
3.90994430e-01 -8.64706874e-01 3.22652817e-01 2.99487889e-01
1.15049891e-01 -8.07182848e-01 -1.38637662e+00 -3.78849506e-01
-4.99696992e-02 -2.50298023e-01 2.42571980e-01 4.49568897e-01
5.67176163e-01 2.45976135e-01 4.15688828e-02 5.43284602e-02
3.77336591e-01 -6.82256341e-01 1.15109093e-01 -1.28369629e+00
-4.72283155e-01 -9.07395780e-01 -3.92214298e-01 -3.58236164e-01
-2.78485450e-03 -4.64511424e-01 2.94766456e-01 -1.01744509e+00
-8.00376460e-02 -2.52561748e-01 -1.44457564e-01 5.11010647e-01
7.80488923e-02 2.37030953e-01 -3.01836748e-02 2.36232743e-01
2.73165762e-01 2.79146075e-01 7.94438839e-01 4.83256340e-01
-6.60977542e-01 -8.72914642e-02 -3.98515999e-01 6.60920680e-01
9.49728012e-01 -7.27490783e-01 -4.98215377e-01 1.65899009e-01
6.49404287e-01 2.05316499e-01 1.84474453e-01 -1.03743947e+00
9.25685316e-02 -1.40304819e-01 6.63802207e-01 -1.89564541e-01
5.08527040e-01 -7.65525103e-01 8.35029721e-01 6.15976274e-01
-3.30914736e-01 1.47370966e-02 5.52059829e-01 2.89552003e-01
1.31602585e-01 -6.47854507e-01 9.98357296e-01 6.81166574e-02
1.12973027e-01 -5.18454611e-01 -1.18986106e+00 -2.68586189e-01
8.78954172e-01 -7.73696482e-01 -1.00978710e-01 -1.64898947e-01
-1.07215309e+00 2.95902696e-02 1.58967078e-01 6.68829232e-02
1.68146551e-01 -1.05019033e+00 -4.60791230e-01 4.13780391e-01
-3.10168177e-01 -1.84799388e-01 7.97571465e-02 8.95601690e-01
-5.29955864e-01 4.21015859e-01 -6.56237483e-01 -6.57013416e-01
-1.07648802e+00 8.84327441e-02 9.32100475e-01 7.50738978e-02
-5.08476257e-01 5.35932779e-01 -3.48763406e-01 -2.04711850e-03
3.65684301e-01 -3.24767381e-01 -7.89206862e-01 4.54254568e-01
2.32144773e-01 1.19707577e-01 1.04103670e-01 -5.94364524e-01
-3.16445291e-01 5.93848169e-01 4.45862114e-01 -6.75998092e-01
1.47232461e+00 -4.93506864e-02 -1.77741885e-01 6.57917500e-01
9.31914449e-01 -4.62196678e-01 -1.30077255e+00 4.60012376e-01
1.00356445e-01 5.55419810e-02 3.00824821e-01 -6.69429839e-01
-6.89227164e-01 3.47310990e-01 7.22938359e-01 4.85503942e-01
1.23542035e+00 -3.03893417e-01 -9.93313640e-02 3.35027069e-01
1.57671481e-01 -1.15694273e+00 -2.77555048e-01 6.08409882e-01
9.82591450e-01 -3.50378275e-01 1.95083693e-01 1.11039795e-01
-3.25082690e-01 1.41362822e+00 4.40742940e-01 -3.08873445e-01
7.61801481e-01 3.28360617e-01 -1.54718205e-01 -2.04121172e-01
-1.18791378e+00 2.83330023e-01 3.50826472e-01 7.56720603e-01
5.02561510e-01 -2.53389537e-01 -6.51883543e-01 2.04371989e-01
-6.45848751e-01 -1.04832552e-01 7.33830333e-01 5.80463350e-01
-6.14714801e-01 -7.12675154e-01 -7.80400991e-01 2.59881377e-01
-4.37062800e-01 8.91294777e-02 -2.05422059e-01 1.03162265e+00
6.94901422e-02 7.90248632e-01 3.49770188e-01 1.36924341e-01
3.01070809e-01 4.46843714e-01 6.62914217e-01 -1.33614317e-01
-8.92035425e-01 7.76194870e-01 1.39297143e-01 -2.47072294e-01
-4.29973572e-01 -6.96660817e-01 -1.43350971e+00 -4.95882146e-02
-4.68263119e-01 2.16019183e-01 8.82025659e-01 1.00363362e+00
-2.76101474e-02 7.13688314e-01 2.47807711e-01 -1.23489261e+00
-3.10535848e-01 -1.20264232e+00 -4.60090578e-01 -3.23741883e-01
2.38019213e-01 -7.71122456e-01 -7.26058245e-01 2.42464975e-01] | [8.02729606628418, 2.8985331058502197] |
5458c3da-121f-441b-9127-7e3da46599d4 | biologically-constrained-graphs-for-global | null | null | http://openaccess.thecvf.com/content_CVPR_2019/html/Matejek_Biologically-Constrained_Graphs_for_Global_Connectomics_Reconstruction_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Matejek_Biologically-Constrained_Graphs_for_Global_Connectomics_Reconstruction_CVPR_2019_paper.pdf | Biologically-Constrained Graphs for Global Connectomics Reconstruction | Most current state-of-the-art connectome reconstruction pipelines have two major steps: initial pixel-based segmentation with affinity prediction and watershed transform, and refined segmentation by merging over-segmented regions. These methods rely only on local context and are typically agnostic to the underlying biology. Since a few merge errors can lead to several incorrectly merged neuronal processes, these algorithms are currently tuned towards over-segmentation producing an overburden of costly proofreading. We propose a third step for connectomics reconstruction pipelines to refine an over-segmentation using both local and global context with an emphasis on adhering to the underlying biology. We first extract a graph from an input segmentation where nodes correspond to segment labels and edges indicate potential split errors in the over-segmentation. In order to increase throughput and allow for large-scale reconstruction, we employ biologically inspired geometric constraints based on neuron morphology to reduce the number of nodes and edges. Next, two neural networks learn these neuronal shapes to further aid the graph construction process. Lastly, we reformulate the region merging problem as a graph partitioning one to leverage global context. We demonstrate the performance of our approach on four real-world connectomics datasets with an average variation of information improvement of 21.3%.
| [' Hanspeter Pfister', ' Toufiq Parag', ' Donglai Wei', ' Haidong Zhu', ' Daniel Haehn', 'Brian Matejek'] | 2019-06-01 | null | null | null | cvpr-2019-6 | ['electron-microscopy-image-segmentation'] | ['computer-vision'] | [ 4.98421669e-01 3.47646236e-01 1.80974424e-01 -4.11911219e-01
-6.35787487e-01 -7.88442612e-01 3.19160938e-01 5.63863814e-01
-5.45613706e-01 8.40753496e-01 -2.74724007e-01 -2.15421230e-01
1.18899412e-01 -8.05956185e-01 -1.01025224e+00 -5.14921725e-01
-3.36636081e-02 6.55308604e-01 6.33212149e-01 2.33766779e-01
4.83437419e-01 6.51658714e-01 -9.00515437e-01 1.78971570e-02
1.13261390e+00 6.30957484e-01 2.72315860e-01 7.04406142e-01
-3.33516538e-01 -1.53921051e-02 -4.25815612e-01 -4.83194977e-01
4.96006787e-01 -4.29521799e-01 -1.00099540e+00 1.27169967e-01
3.66429180e-01 -3.91212553e-02 1.07727833e-01 1.26864505e+00
3.74676675e-01 -2.68893898e-01 6.10583067e-01 -9.96107817e-01
-2.36947417e-01 6.40942395e-01 -6.43032789e-01 3.97497952e-01
-6.80080280e-02 4.23152357e-01 8.22414458e-01 -4.92540091e-01
1.04976046e+00 8.44220221e-01 8.38867188e-01 3.35647821e-01
-1.82609904e+00 -5.01980484e-01 3.78133416e-01 -2.48835206e-01
-1.24729359e+00 -3.13497931e-01 5.85733235e-01 -7.40837395e-01
9.88521934e-01 -9.29331779e-03 1.25326300e+00 6.42319858e-01
2.90910333e-01 2.53105164e-01 9.69626963e-01 4.42113392e-02
4.88292247e-01 -4.81924921e-01 1.63798735e-01 8.23023438e-01
3.77264470e-01 -4.08215940e-01 -9.63169560e-02 -2.56904401e-02
1.07684672e+00 -3.11391279e-02 -1.93063751e-01 -3.47777903e-01
-1.39160192e+00 6.09184921e-01 6.53333962e-01 9.62788239e-02
-3.86385620e-01 1.99863404e-01 2.67334104e-01 -2.26517260e-01
3.50144178e-01 5.93923569e-01 -4.41316992e-01 2.45414853e-01
-1.26473951e+00 2.46940091e-01 5.80170393e-01 8.07688534e-01
1.08415270e+00 -2.04885811e-01 1.64806843e-01 6.16233468e-01
3.72031420e-01 -1.25035092e-01 1.32177338e-01 -1.24748969e+00
1.47467837e-01 8.84518564e-01 -3.49691659e-01 -8.66862178e-01
-7.07977414e-01 -4.21561122e-01 -7.32101083e-01 2.91009575e-01
8.44086051e-01 -9.61969569e-02 -1.36838043e+00 1.64391243e+00
5.28743088e-01 2.33368844e-01 -3.33927423e-01 7.62371659e-01
5.86176991e-01 8.78267139e-02 1.91953793e-01 -5.66829275e-03
1.27197683e+00 -7.60579169e-01 -2.83336580e-01 -2.18854800e-01
6.98697567e-01 -4.06619191e-01 6.94820464e-01 2.86655784e-01
-1.09474552e+00 -8.12226385e-02 -9.78534400e-01 -2.20038399e-01
-4.38334435e-01 -1.81984231e-01 4.92396355e-01 2.65935838e-01
-1.29234850e+00 8.24656129e-01 -1.06674945e+00 -7.05587149e-01
9.49582696e-01 5.80302238e-01 -3.80916983e-01 2.19829515e-01
-4.70323533e-01 6.52503192e-01 5.50578058e-01 -1.08195998e-01
-6.71841323e-01 -9.50880289e-01 -5.40587187e-01 4.23564352e-02
2.76996344e-01 -1.07626081e+00 7.20737815e-01 -6.99412704e-01
-1.21589351e+00 1.22135806e+00 -7.31873885e-02 -5.10616481e-01
5.59135556e-01 2.37836182e-01 1.19366199e-01 4.73468542e-01
2.57671148e-01 1.28470623e+00 3.75410467e-01 -1.21416032e+00
-5.00989139e-01 -6.39249742e-01 -2.20820710e-01 4.79663461e-02
3.73056650e-01 -3.04184556e-01 -7.53202438e-01 -6.08127534e-01
6.74926400e-01 -8.34767580e-01 -5.52091002e-01 4.58618462e-01
-6.97298586e-01 1.77428797e-01 5.25904357e-01 -6.43921494e-01
6.76541507e-01 -1.69576764e+00 2.45804489e-01 3.16982001e-01
6.45485580e-01 -1.50492832e-01 -1.51837945e-01 4.92947400e-02
4.17121239e-02 5.82860470e-01 -7.16704309e-01 -2.77051151e-01
-4.13059711e-01 1.26726761e-01 1.26154780e-01 6.94443226e-01
3.63228112e-01 1.07195425e+00 -9.16892946e-01 -6.42170787e-01
5.75463939e-03 3.31076801e-01 -6.71408057e-01 -6.32945299e-02
-3.17467511e-01 7.56390333e-01 -3.50541174e-01 7.30566263e-01
6.40281260e-01 -3.50110650e-01 2.44507417e-01 -1.96142852e-01
-1.48644090e-01 1.73986584e-01 -9.43247914e-01 1.96010768e+00
3.13624769e-01 3.54516059e-01 4.96385157e-01 -9.31360006e-01
8.14720094e-01 -1.18656188e-01 7.51169503e-01 -8.66740718e-02
2.16974854e-01 1.29720002e-01 1.47881299e-01 -3.50039527e-02
6.55574948e-02 -1.71270117e-01 3.76680970e-01 3.71795923e-01
1.80368125e-01 -3.80950570e-01 2.61053473e-01 4.56143916e-01
1.29934776e+00 4.45133597e-01 1.33586824e-01 -5.51373243e-01
6.44739792e-02 2.37306952e-01 9.71574962e-01 3.87355953e-01
-4.78186727e-01 9.83589053e-01 7.90132880e-01 -3.11304361e-01
-1.29354882e+00 -9.56670225e-01 -6.12454750e-02 6.17589891e-01
1.89906895e-01 -2.85515100e-01 -1.17202437e+00 -7.00140059e-01
-1.87060758e-01 3.56629461e-01 -5.46233535e-01 1.56984344e-01
-4.01752055e-01 -9.52735305e-01 8.71061206e-01 3.38222384e-01
3.24809432e-01 -9.14719224e-01 -7.86561906e-01 3.68610293e-01
8.52169171e-02 -1.12727666e+00 -2.98471987e-01 3.71674061e-01
-1.17078650e+00 -1.20060647e+00 -7.84062982e-01 -7.35906601e-01
1.10147774e+00 4.67784004e-03 9.43224728e-01 4.28119332e-01
-4.24869210e-01 7.82443881e-02 -2.66031921e-02 -1.61390066e-01
-1.02657564e-01 1.44040287e-01 -2.40193695e-01 -3.94368649e-01
-5.23628481e-02 -9.54847574e-01 -8.04433465e-01 1.24547988e-01
-7.93526173e-01 3.59890819e-01 5.41388333e-01 6.25752687e-01
1.24062467e+00 -1.82835400e-01 3.96427721e-01 -1.08543313e+00
3.71688724e-01 -5.00868797e-01 -7.52206028e-01 2.13968605e-01
-6.39603734e-01 2.65364889e-02 4.16819304e-01 -3.18684131e-01
-6.31625712e-01 4.71189648e-01 -1.39312878e-01 -3.37968022e-01
-3.21137428e-01 4.18745726e-01 -2.06707820e-01 -2.51588017e-01
3.99187684e-01 1.94626912e-01 4.91812974e-02 -2.76489496e-01
3.03730667e-01 -8.24855343e-02 6.20231986e-01 -6.16575241e-01
3.17365527e-01 5.96565425e-01 2.46240765e-01 -6.18567646e-01
-3.21651846e-01 -2.68731624e-01 -1.12961602e+00 -2.39367917e-01
1.11166978e+00 -4.58135456e-01 -6.57112300e-01 4.58843708e-01
-1.07764697e+00 -7.66283989e-01 -2.04129800e-01 2.00916171e-01
-4.61624175e-01 5.87970078e-01 -7.94592500e-01 -3.95155370e-01
-4.20296133e-01 -1.31178141e+00 1.04279852e+00 2.32932895e-01
-1.79628700e-01 -8.02086115e-01 9.80302617e-02 2.86462843e-01
8.27633217e-02 4.54138070e-01 1.08211708e+00 -7.68391848e-01
-7.36810148e-01 2.41456270e-01 -4.22036082e-01 -4.31386769e-01
-2.32683010e-02 3.38360965e-01 -7.28514671e-01 -1.50491446e-02
-5.05569100e-01 -9.31228176e-02 9.75841641e-01 5.18697262e-01
1.13709593e+00 -2.61053853e-02 -6.30011976e-01 9.54677522e-01
1.34029734e+00 -1.14368042e-02 5.53040087e-01 4.28705662e-01
9.03670430e-01 8.57851505e-01 2.21114904e-02 7.64951780e-02
4.29185241e-01 2.30100125e-01 5.37802875e-01 -2.59569556e-01
-2.14294001e-01 -1.07702650e-01 -1.55703187e-01 5.45217156e-01
5.78382565e-03 -9.88295749e-02 -1.10349429e+00 6.28073633e-01
-1.73186612e+00 -6.24907076e-01 -4.93074328e-01 2.01685810e+00
8.82326245e-01 2.98891693e-01 3.41304868e-01 -2.03934148e-01
9.35975790e-01 -2.52627790e-01 -1.07551169e+00 -8.93100947e-02
-1.39118984e-01 6.26544431e-02 7.53056943e-01 5.26172042e-01
-8.38426411e-01 1.33670712e+00 6.96228456e+00 4.45846975e-01
-1.01898110e+00 -1.63358271e-01 1.10632157e+00 -3.14351078e-03
-4.63803768e-01 3.05631578e-01 -5.58609247e-01 3.18699270e-01
4.92191792e-01 3.41167688e-01 6.26398563e-01 3.43977123e-01
1.29031226e-01 -4.09881473e-01 -9.91598487e-01 7.67603397e-01
-4.52833265e-01 -1.48011720e+00 2.26414092e-02 3.36426377e-01
6.10624909e-01 4.64902550e-01 -4.87217188e-01 -1.70314252e-01
6.27066255e-01 -9.51280177e-01 6.53253973e-01 6.28052354e-01
6.65915549e-01 -6.02627456e-01 3.43928099e-01 2.77828783e-01
-1.18948293e+00 5.13014019e-01 -3.82614732e-01 1.35132998e-01
3.19999039e-01 8.64384651e-01 -8.90392601e-01 3.09723496e-01
7.93692112e-01 5.06778598e-01 -5.50937712e-01 1.31845510e+00
-2.43238986e-01 4.25653428e-01 -6.27187610e-01 3.35224956e-01
-1.76988598e-02 -6.04525387e-01 6.88921034e-01 1.26088512e+00
1.03365913e-01 2.34453559e-01 2.71090508e-01 1.45901024e+00
-2.91839421e-01 -1.63711626e-02 -4.31168288e-01 -5.57454675e-02
5.42145848e-01 1.51216555e+00 -1.75501966e+00 -4.03214008e-01
-8.43881443e-02 1.02895188e+00 7.98956394e-01 4.69195992e-01
-5.70928335e-01 -1.83685869e-01 6.24614000e-01 1.47202283e-01
2.90536255e-01 -3.55938792e-01 -7.97516346e-01 -9.03353274e-01
-2.44737074e-01 -4.43089038e-01 2.70427674e-01 -7.13038504e-01
-1.15131688e+00 2.29524240e-01 -3.64473581e-01 -4.10027862e-01
3.64571959e-01 -3.41029912e-01 -7.21111834e-01 8.15461397e-01
-1.30055380e+00 -1.10629857e+00 -2.05791235e-01 1.09395295e-01
4.56650078e-01 3.79662365e-01 4.97833371e-01 9.54231694e-02
-8.59151423e-01 2.13406920e-01 -3.13157707e-01 1.01852991e-01
4.49478179e-01 -1.31644058e+00 6.62290990e-01 1.01420927e+00
-6.32961914e-02 8.65660965e-01 5.26422024e-01 -1.23610234e+00
-1.09210825e+00 -1.26538873e+00 4.53890711e-01 -1.42322928e-01
7.23003268e-01 -4.48446542e-01 -1.04763377e+00 8.72305453e-01
-7.21325800e-02 2.08978727e-01 5.05763352e-01 -1.38067886e-01
-1.24998018e-01 2.48505831e-01 -1.46598458e+00 7.05628932e-01
1.29643893e+00 -2.58385152e-01 -3.66790593e-01 8.19835439e-02
5.21710575e-01 -3.26370180e-01 -1.05886960e+00 2.82861799e-01
4.47028935e-01 -7.84348726e-01 9.16806281e-01 -4.81835157e-01
3.37784886e-01 -6.09049797e-01 2.26365298e-01 -1.11286700e+00
-4.95337218e-01 -5.34585059e-01 2.07780182e-01 1.33334935e+00
5.46703041e-01 -5.11595964e-01 1.03118289e+00 7.60751128e-01
-4.07343417e-01 -8.71578932e-01 -8.08558762e-01 -3.63183409e-01
2.02628210e-01 -2.55358368e-01 4.76505309e-01 9.42309439e-01
-4.14807349e-02 2.65222609e-01 2.82858431e-01 1.04010575e-01
7.60626137e-01 -6.70471117e-02 6.81353331e-01 -1.45736527e+00
-3.00199203e-02 -8.60574126e-01 -4.84823525e-01 -1.00814068e+00
-5.51644266e-02 -1.08796823e+00 2.65936881e-01 -1.75115156e+00
3.52294922e-01 -4.31502163e-01 -9.15108249e-02 6.15749836e-01
-2.90277034e-01 5.25048256e-01 -5.02591245e-02 3.62611324e-01
-6.26530528e-01 2.74163395e-01 1.31811213e+00 -2.25025881e-02
-2.94474661e-01 -5.60679734e-01 -6.70332789e-01 9.28438127e-01
9.30459440e-01 -5.86504400e-01 -1.46445930e-01 -4.42375481e-01
8.79379958e-02 -1.72569871e-01 4.26730752e-01 -9.26638067e-01
3.47220898e-01 -1.02177076e-01 5.19435644e-01 -6.33499980e-01
1.63415834e-01 -6.00856602e-01 2.44388983e-01 4.81414795e-01
-3.06258053e-01 -1.13417402e-01 1.97670549e-01 6.54718459e-01
3.67607415e-01 -1.69260446e-02 9.48919415e-01 -4.82639223e-01
-3.87455910e-01 4.01841581e-01 -5.29399812e-01 1.39499856e-02
1.12226009e+00 -6.82247519e-01 -3.43880683e-01 3.77919711e-02
-9.05745506e-01 2.84395605e-01 1.06963766e+00 -3.39869171e-01
4.54336792e-01 -7.02811062e-01 -4.67285901e-01 3.13668065e-02
-1.38055697e-01 4.78283852e-01 1.73651762e-02 1.09904885e+00
-9.06734109e-01 1.01831369e-01 -4.32700127e-01 -7.83496678e-01
-1.17184818e+00 4.29964215e-01 4.08079058e-01 -2.30611473e-01
-7.97051370e-01 9.24429834e-01 2.65907288e-01 -6.41498446e-01
-1.45575136e-01 -6.70923352e-01 -7.75914788e-02 -1.87358215e-01
-2.89027039e-02 3.98500621e-01 -1.78904342e-03 -6.97747767e-01
-4.16892350e-01 6.43038988e-01 -5.13051562e-02 -7.66323060e-02
1.41968405e+00 -3.91550958e-01 -3.26547205e-01 2.61628807e-01
8.38425398e-01 -1.81925818e-01 -1.62030780e+00 1.71422720e-01
1.93280056e-01 -6.01355657e-02 3.23306844e-02 -8.57890844e-01
-1.27118993e+00 8.08804512e-01 3.62066448e-01 5.96128032e-02
8.38811636e-01 9.09002200e-02 8.94150972e-01 1.19394653e-01
3.39736521e-01 -9.44873631e-01 -3.16879839e-01 4.23176646e-01
4.72743273e-01 -9.58085716e-01 1.70050025e-01 -7.30602801e-01
-2.02703714e-01 1.05766666e+00 6.56078458e-01 -2.33902529e-01
3.49365056e-01 3.32681507e-01 -9.40424874e-02 -5.30334890e-01
-3.50487858e-01 -1.74737245e-01 1.97151527e-01 6.69551075e-01
2.87020743e-01 -1.45121098e-01 -2.39241377e-01 5.43542981e-01
-1.04713120e-01 -9.74608734e-02 3.14230561e-01 8.60163212e-01
-6.14456177e-01 -9.66576993e-01 -2.20764846e-01 6.98984981e-01
-6.12843752e-01 -1.75565153e-01 -8.40747058e-01 5.20237327e-01
1.96532428e-01 5.69567978e-01 1.35003477e-01 -1.27931297e-01
-6.08096011e-02 -1.29052438e-02 5.14216065e-01 -7.67522454e-01
-5.56777656e-01 5.07488966e-01 -1.01431534e-01 -5.65222263e-01
-2.20779940e-01 -7.02957511e-01 -2.01284885e+00 -1.64905652e-01
-2.56114244e-01 -2.76667267e-01 6.95206702e-01 1.05085158e+00
6.91555560e-01 7.23034322e-01 -2.74393857e-01 -1.03327739e+00
3.22725266e-01 -7.47791648e-01 -4.94431734e-01 2.39631712e-01
-7.94177968e-03 -4.82083142e-01 -3.75038274e-02 3.30611885e-01] | [14.27199935913086, -3.127729892730713] |
1454e93b-6ae2-4e8e-b7e3-954b2a7e28b7 | 2305-14706 | 2305.14706 | null | https://arxiv.org/abs/2305.14706v1 | https://arxiv.org/pdf/2305.14706v1.pdf | PruMUX: Augmenting Data Multiplexing with Model Compression | As language models increase in size by the day, methods for efficient inference are critical to leveraging their capabilities for various applications. Prior work has investigated techniques like model pruning, knowledge distillation, and data multiplexing to increase model throughput without sacrificing accuracy. In this paper, we combine two such methods -- structured pruning and data multiplexing -- to compound the speedup gains obtained by either method. Our approach, PruMUX, obtains up to 7.5-29.5X throughput improvement over BERT-base model with accuracy threshold from 80% to 74%. We further study various combinations of parameters (such as sparsity and multiplexing factor) in the two techniques to provide a comprehensive analysis of the tradeoff between accuracy and throughput in the resulting models. We then propose Auto-PruMUX, a meta-level model that can predict the high-performance parameters for pruning and multiplexing given a desired accuracy loss budget, providing a practical method to leverage the combination effectively. | ['Kai Li', 'Karthik Narasimhan', 'Vishvak Murahari', 'Yushan Su'] | 2023-05-24 | null | null | null | null | ['model-compression'] | ['methodology'] | [-9.33425035e-03 3.25439535e-02 -6.55582190e-01 -3.70683402e-01
-1.07181323e+00 -3.21944177e-01 3.31110448e-01 2.93512821e-01
-4.33513224e-01 7.36049414e-01 2.08999515e-02 -1.00254154e+00
-1.38974532e-01 -7.54333079e-01 -5.06780624e-01 -2.61111185e-02
-2.37378120e-01 4.86144155e-01 2.94008285e-01 1.48032099e-01
3.37954164e-01 4.82528150e-01 -1.23165894e+00 5.33606648e-01
9.63538706e-01 9.32017803e-01 -1.44242197e-01 9.74012196e-01
-2.50123829e-01 9.59238350e-01 -5.53814769e-01 -4.51265395e-01
4.68830585e-01 -9.29645076e-02 -7.37381995e-01 -4.03777748e-01
2.47183636e-01 -9.01129007e-01 -4.68150914e-01 5.27625740e-01
2.96630949e-01 -1.24596074e-01 3.74366373e-01 -1.17610085e+00
6.54374510e-02 1.08619332e+00 -9.50462639e-01 5.67364514e-01
-1.63143240e-02 5.74575588e-02 1.08364427e+00 -6.45125806e-01
3.13718826e-01 1.34418583e+00 5.38229704e-01 1.76620603e-01
-1.45530307e+00 -9.77387547e-01 -6.83867605e-04 -7.61954486e-03
-1.61683381e+00 -1.01022470e+00 2.33926512e-02 -2.32971311e-01
1.40517247e+00 3.78041208e-01 4.24614310e-01 2.90907592e-01
9.80191380e-02 8.91644537e-01 9.98960316e-01 -7.57399797e-01
2.06139475e-01 2.56298095e-01 4.54054922e-01 8.12241316e-01
8.15176606e-01 1.16567172e-01 -9.62745965e-01 -5.40815949e-01
6.79781020e-01 -1.22075729e-01 1.71331748e-01 7.03131780e-02
-6.36382222e-01 6.78869307e-01 9.44405943e-02 -1.89703256e-01
5.29849110e-03 6.39251888e-01 4.51738238e-01 3.33924621e-01
4.33874279e-01 7.56003797e-01 -6.04486465e-01 -4.46964741e-01
-1.64190328e+00 4.43733633e-01 1.00798643e+00 1.29426527e+00
5.82229555e-01 7.15822577e-02 -5.30101657e-01 7.32227087e-01
2.77270854e-01 3.45623791e-01 -9.51941013e-02 -1.16200149e+00
7.37100661e-01 4.54477310e-01 1.60958573e-01 -5.69487035e-01
-3.83281261e-01 -5.56921303e-01 -3.48282605e-01 -3.88650894e-01
3.49423259e-01 -2.66586453e-01 -9.14038539e-01 1.65145981e+00
2.96742141e-01 1.73647791e-01 -3.92393023e-03 2.80849099e-01
1.64287046e-01 5.98938882e-01 1.12115383e-01 -1.78300187e-01
1.35953224e+00 -9.70675170e-01 -3.50016832e-01 -1.52029872e-01
1.05054331e+00 -8.75517488e-01 1.04953706e+00 3.84010315e-01
-1.40225732e+00 -4.77099232e-02 -1.18345439e+00 -2.32283652e-01
1.69156000e-01 -5.59605137e-02 8.92828226e-01 8.19615006e-01
-8.73639464e-01 5.42470872e-01 -1.36110747e+00 -2.96543717e-01
5.76921523e-01 4.38578516e-01 3.99364382e-01 -1.55777290e-01
-6.82790279e-01 9.05707598e-01 4.88806099e-01 -4.37637329e-01
-6.50431693e-01 -1.23295414e+00 -3.13979387e-01 3.79264563e-01
5.63660324e-01 -7.59341300e-01 1.61289597e+00 -1.17976494e-01
-1.16133845e+00 3.48126918e-01 -4.30055857e-01 -9.85318244e-01
3.30796748e-01 -5.58730662e-01 -3.33582789e-01 -2.73041651e-02
-3.20237160e-01 6.50036991e-01 2.93524802e-01 -1.05386305e+00
-9.00642931e-01 -2.10702673e-01 3.66203815e-01 1.78790614e-02
-5.34969509e-01 8.72911215e-02 -8.22358966e-01 -4.69389915e-01
-9.88999009e-02 -7.21917272e-01 -4.35209066e-01 -1.59421742e-01
-5.68721414e-01 3.27890217e-01 7.63956487e-01 -6.20756924e-01
1.87381971e+00 -1.90346408e+00 -5.22126317e-01 4.79369491e-01
4.16950941e-01 2.87226200e-01 5.46251535e-02 4.94842947e-01
8.04297328e-01 4.13191885e-01 9.13580656e-02 -4.73308206e-01
-3.06282759e-01 3.74087989e-01 -2.47979790e-01 -9.40353330e-03
1.34062454e-01 7.26650536e-01 -6.44358099e-01 -5.81924617e-01
-1.44347027e-01 2.31519029e-01 -1.00950480e+00 3.36511359e-02
-4.03752089e-01 -4.82012518e-02 -5.53547978e-01 7.85331070e-01
4.31546032e-01 -4.73360538e-01 4.59887862e-01 -4.54618372e-02
1.36596099e-01 6.11699522e-01 -9.96942461e-01 1.11269891e+00
-9.46403086e-01 5.30221343e-01 3.20371300e-01 -5.02027571e-01
6.30962491e-01 -9.88069773e-02 1.95724115e-01 -5.56508303e-01
1.32944033e-01 2.62409508e-01 1.67979881e-01 -1.40219003e-01
9.14339662e-01 1.18423581e-01 1.14157714e-01 4.51927722e-01
-3.14425111e-01 1.10851385e-01 4.14838046e-01 4.50122923e-01
1.35740125e+00 -3.78370076e-01 3.97661179e-01 -4.32946086e-01
-7.05258101e-02 6.26400337e-02 3.65647763e-01 1.19076800e+00
1.89944595e-01 5.80837484e-04 5.93415141e-01 -1.22006372e-01
-1.32787204e+00 -6.94237053e-01 -2.93403476e-01 1.09254515e+00
-1.66714147e-01 -8.71060729e-01 -6.33711457e-01 -2.72691220e-01
4.98070121e-01 9.91255522e-01 -5.71312830e-02 -1.50779992e-01
-3.88300508e-01 -8.73576939e-01 1.01173282e+00 7.65284479e-01
5.62272787e-01 -2.41814572e-02 -7.78620481e-01 2.40794316e-01
1.30300626e-01 -1.35480142e+00 -1.90201670e-01 2.26751655e-01
-1.18001437e+00 -7.01608360e-01 8.70356262e-02 -3.41875741e-04
3.73126060e-01 2.92248428e-01 1.27040052e+00 4.21559215e-01
-1.46112978e-01 -3.26224491e-02 -2.04448447e-01 -5.56370258e-01
-6.35092020e-01 4.75571394e-01 6.35859882e-03 -5.78617752e-01
3.08790445e-01 -7.49142528e-01 -5.24489641e-01 6.15314469e-02
-7.93305755e-01 3.17468494e-01 8.34128380e-01 7.73519218e-01
5.48927486e-01 -1.88866351e-02 5.29603183e-01 -1.26266611e+00
5.95902681e-01 -5.83621621e-01 -7.00334430e-01 4.02596503e-01
-1.19556189e+00 4.15417016e-01 5.51375806e-01 -1.35682330e-01
-8.02654386e-01 -3.27560753e-01 -3.91752049e-02 -3.53048742e-01
4.80039358e-01 8.46452892e-01 1.84999108e-01 -6.83844909e-02
6.45567656e-01 -1.86684772e-01 -4.38577402e-03 -4.56572831e-01
5.04412830e-01 7.15763509e-01 5.03399551e-01 -1.02998638e+00
2.96999484e-01 2.68482029e-01 8.02821070e-02 -6.71845198e-01
-9.27961528e-01 -4.69494164e-01 -1.27018183e-01 4.53216463e-01
5.62880337e-02 -1.10482073e+00 -5.17933071e-01 -5.81666604e-02
-8.23651612e-01 -4.80725795e-01 -2.13907138e-01 4.56260324e-01
-2.81824917e-01 1.32578507e-01 -8.58668685e-01 -1.14695907e+00
-6.40177011e-01 -9.96964335e-01 9.72765625e-01 -8.31662491e-02
-2.53400415e-01 -5.44198632e-01 -3.91723305e-01 3.77228796e-01
5.04215598e-01 -1.74055755e-01 1.16461420e+00 -8.52321744e-01
-8.62007439e-01 -2.17854813e-01 -6.27801418e-01 1.08219713e-01
-3.87209535e-01 1.29225001e-01 -8.53305042e-01 -4.75443065e-01
-4.44344699e-01 -2.96206832e-01 8.61067235e-01 3.80466104e-01
1.49257255e+00 -5.65492928e-01 -6.90060794e-01 6.40387774e-01
1.50591159e+00 2.09156480e-02 4.08358365e-01 -8.71548578e-02
5.13644040e-01 -1.01859316e-01 5.18634319e-01 7.65228033e-01
3.73016983e-01 8.51521671e-01 -4.08152230e-02 4.84722219e-02
-8.80269334e-02 -4.10403669e-01 1.18086338e-01 8.95204425e-01
9.37311053e-02 -3.41108024e-01 -1.14144337e+00 4.66430187e-01
-1.70776820e+00 -6.28506005e-01 2.05782294e-01 2.41520596e+00
1.17050505e+00 7.09905386e-01 2.15489000e-01 1.83865845e-01
1.95258334e-01 -2.48997107e-01 -7.12074399e-01 -6.48716748e-01
4.14460629e-01 3.48524839e-01 1.19094920e+00 6.79729700e-01
-4.91618842e-01 9.25818503e-01 7.39753485e+00 1.09981251e+00
-8.88934553e-01 8.68964046e-02 8.28858972e-01 -6.92409873e-01
-2.74501324e-01 3.48427981e-01 -1.50886977e+00 3.00835162e-01
1.61199951e+00 -4.74612594e-01 6.31974459e-01 8.90738606e-01
2.30164364e-01 -3.52634609e-01 -1.28008175e+00 5.43134630e-01
-3.03954452e-01 -1.60470092e+00 2.50669718e-01 2.20275670e-01
7.90848970e-01 5.34538776e-02 -2.15816736e-01 5.09001851e-01
7.01251626e-01 -9.23436701e-01 5.08209348e-01 3.65717024e-01
1.01502454e+00 -9.20486391e-01 5.49669683e-01 5.09070933e-01
-1.03544235e+00 -3.60667765e-01 -1.90060332e-01 -3.24041545e-01
2.61446297e-01 8.83207321e-01 -1.36497068e+00 4.90414858e-01
3.84064674e-01 -1.29684794e-03 -5.19798636e-01 9.78445351e-01
3.86203468e-01 1.16912878e+00 -9.69843328e-01 1.05283223e-02
-2.15996429e-02 4.87243026e-01 2.77724057e-01 1.43372715e+00
2.91134536e-01 -1.46860136e-02 3.64481211e-01 5.82020998e-01
-2.14002088e-01 -1.27894238e-01 -1.36223108e-01 -1.80004761e-01
1.51192081e+00 8.78155828e-01 -5.58547616e-01 -5.57282686e-01
-3.99790436e-01 1.95873901e-01 4.94223952e-01 1.48664594e-01
-7.65349865e-01 -2.38178238e-01 6.37477219e-01 4.56957996e-01
2.95448691e-01 -5.36681771e-01 -8.26257944e-01 -9.26715076e-01
-1.51515543e-01 -9.21190858e-01 3.76673311e-01 -3.12478393e-01
-8.35203052e-01 1.18051812e-01 5.14583230e-01 -4.52168226e-01
-1.93209410e-01 -3.77623230e-01 -2.23750934e-01 8.39284718e-01
-1.34581172e+00 -1.05358684e+00 2.59275306e-02 5.32199629e-02
3.93146604e-01 -7.44498819e-02 5.72804034e-01 5.71013272e-01
-7.31496334e-01 1.36440706e+00 3.94431174e-01 -2.70921528e-01
2.70262510e-01 -7.76445448e-01 5.28437793e-01 8.94908190e-01
1.24603562e-01 9.78375375e-01 7.10872233e-01 -5.11925459e-01
-1.45939696e+00 -1.03500831e+00 7.05157638e-01 -4.38394547e-01
5.15190601e-01 -3.90115827e-01 -6.79888606e-01 7.10642159e-01
-3.68185431e-01 -2.26859689e-01 6.39353752e-01 6.85675859e-01
-4.48463231e-01 -3.08462143e-01 -1.11533451e+00 7.92559206e-01
1.03908062e+00 -2.77147472e-01 3.56037058e-02 1.15393654e-01
1.00372195e+00 -6.90508306e-01 -1.09093475e+00 4.61340398e-01
8.39952648e-01 -7.70841479e-01 8.51638734e-01 -8.32372904e-01
5.11637211e-01 -4.83811693e-03 -4.52797562e-01 -7.48411298e-01
1.68083105e-02 -7.25248933e-01 -9.13798153e-01 1.13174689e+00
6.68083549e-01 -4.02362049e-01 9.37739611e-01 7.18790591e-01
5.00801243e-02 -1.28516567e+00 -6.84297204e-01 -8.57931376e-01
-3.43207717e-02 -6.53509974e-01 9.59180653e-01 4.93661195e-01
-9.26029757e-02 4.28333998e-01 -3.30894738e-01 1.35093898e-01
4.65051591e-01 -9.60754603e-02 1.05669248e+00 -7.79375136e-01
-6.62391543e-01 -4.89400655e-01 -1.26784265e-01 -1.48689353e+00
-4.13098276e-01 -9.23582911e-01 -4.02922571e-01 -1.15754282e+00
4.21860635e-01 -9.86395657e-01 -2.38673329e-01 6.18898094e-01
-1.80834398e-01 -2.00408354e-01 2.43377760e-01 3.52311432e-01
-5.46867430e-01 1.04438178e-01 8.33855033e-01 3.34809452e-01
-4.78917211e-01 -5.51298074e-02 -1.07449925e+00 5.11661291e-01
7.66886711e-01 -3.75882804e-01 -6.43317878e-01 -6.60675704e-01
1.76309094e-01 3.69905055e-01 1.21381164e-01 -1.07370329e+00
4.07347620e-01 -1.83470383e-01 1.79530919e-01 -5.31501770e-01
2.96907634e-01 -6.20495081e-01 3.37296142e-03 4.41935450e-01
-6.30393326e-01 1.58985376e-01 5.36281765e-01 6.24272287e-01
1.41853690e-01 2.90805921e-02 7.18553960e-01 2.21749902e-01
-3.98309529e-01 7.00763986e-02 -8.60682651e-02 2.99465656e-01
7.07677603e-01 -2.57186498e-02 -4.63449806e-01 -3.61644953e-01
-2.29063183e-01 6.37858689e-01 2.61173338e-01 2.45464612e-02
1.75387263e-01 -9.69727516e-01 -6.97715402e-01 7.13705420e-02
-1.49686886e-02 -9.82678961e-03 3.29195289e-03 9.13318515e-01
-6.13071680e-01 6.74585879e-01 2.02452049e-01 -5.12406409e-01
-1.32538855e+00 3.01401585e-01 1.28356263e-01 -9.08384979e-01
-2.46470213e-01 8.90572488e-01 -4.33254570e-01 -8.83430839e-02
3.25833887e-01 -5.45681119e-01 6.85086370e-01 -5.25775254e-01
4.65915769e-01 7.59013891e-01 2.56053448e-01 2.35542729e-01
-1.83037937e-01 -3.37920040e-01 -5.58908999e-01 -8.44699740e-02
9.32743967e-01 -7.42076710e-02 7.73735791e-02 2.62764305e-01
1.02804601e+00 1.52806833e-01 -9.38386261e-01 -4.78387892e-01
1.25976190e-01 -6.07150793e-01 3.73363405e-01 -9.83239293e-01
-8.66872609e-01 6.52208328e-01 1.81764781e-01 2.11755216e-01
1.01572812e+00 -1.84731081e-01 1.01927733e+00 5.45300841e-01
5.72917819e-01 -9.88736808e-01 -3.43661278e-01 3.74988884e-01
2.38040283e-01 -8.10280502e-01 6.94596529e-01 -5.28819799e-01
-2.45053783e-01 7.28923917e-01 7.11070776e-01 2.66882241e-01
6.88354909e-01 9.48718131e-01 -3.54600757e-01 -3.40223052e-02
-1.28323209e+00 9.95267034e-02 -2.88513359e-02 1.21079393e-01
5.10204792e-01 2.54167646e-01 -4.24075782e-01 6.46939337e-01
-3.93805981e-01 2.70213813e-01 2.76528209e-01 1.23202872e+00
-6.54751837e-01 -1.26160610e+00 -3.54206175e-01 1.31089449e+00
-6.49283826e-01 -3.82631361e-01 5.40440828e-02 8.83425534e-01
-2.81266034e-01 1.05001271e+00 2.00018227e-01 -6.16352081e-01
2.06064552e-01 -2.30709612e-02 4.97829586e-01 -6.55617058e-01
-4.96023804e-01 1.28112704e-01 6.48196161e-01 -4.21580344e-01
-1.20395916e-02 -3.95208597e-01 -1.21157587e+00 -1.11195207e+00
-7.14342117e-01 -5.43411449e-03 3.80554557e-01 9.01004553e-01
8.50590169e-01 5.27106106e-01 2.12874860e-01 -3.66890609e-01
-9.84779239e-01 -7.66641498e-01 -4.57571208e-01 -3.54445368e-01
1.07817329e-01 -6.46642506e-01 -4.72039878e-02 -3.25019181e-01] | [8.682157516479492, 3.458923816680908] |
3e3eff3f-8433-44a4-b4c3-8f339e0f9c5b | geo-defakehop-high-performance-geographic | 2110.09795 | null | https://arxiv.org/abs/2110.09795v1 | https://arxiv.org/pdf/2110.09795v1.pdf | Geo-DefakeHop: High-Performance Geographic Fake Image Detection | A robust fake satellite image detection method, called Geo-DefakeHop, is proposed in this work. Geo-DefakeHop is developed based on the parallel subspace learning (PSL) methodology. PSL maps the input image space into several feature subspaces using multiple filter banks. By exploring response differences of different channels between real and fake images for a filter bank, Geo-DefakeHop learns the most discriminant channels and uses their soft decision scores as features. Then, Geo-DefakeHop selects a few discriminant features from each filter bank and ensemble them to make a final binary decision. Geo-DefakeHop offers a light-weight high-performance solution to fake satellite images detection. Its model size is analyzed, which ranges from 0.8 to 62K parameters. Furthermore, it is shown by experimental results that it achieves an F1-score higher than 95\% under various common image manipulations such as resizing, compression and noise corruption. | ['C. -C. Jay Kuo', 'Suya You', 'Shuowen Hu', 'Kaitai Zhang', 'Hong-Shuo Chen'] | 2021-10-19 | null | null | null | null | ['fake-image-detection'] | ['computer-vision'] | [-8.69552977e-03 -6.65479541e-01 -7.78365210e-02 1.49098998e-02
-6.90876245e-01 -4.55404937e-01 4.44019467e-01 -3.35607022e-01
-2.60324150e-01 4.13867205e-01 7.75416521e-03 -2.18605176e-01
3.23159695e-02 -5.15652537e-01 -3.23157907e-01 -1.07090795e+00
-4.67637360e-01 -9.70214531e-02 3.24011356e-01 -2.42263898e-01
4.73417640e-01 7.47605324e-01 -1.29969692e+00 5.98568082e-01
9.01107490e-01 9.52875197e-01 -1.34819688e-03 6.45899117e-01
3.85695487e-01 2.33426392e-01 -5.09238124e-01 -4.34356518e-02
7.73478627e-01 -1.78318769e-01 -4.95810926e-01 1.21296659e-01
3.38814944e-01 -6.54773474e-01 -7.41768360e-01 1.44575691e+00
3.40411246e-01 -3.24285477e-01 5.80500603e-01 -1.39053309e+00
-2.37675592e-01 1.45200714e-01 -7.17130542e-01 5.52412689e-01
2.54238285e-02 2.35177994e-01 5.11292934e-01 -1.42692626e+00
4.32710409e-01 1.27051198e+00 6.29586101e-01 -2.33584475e-02
-1.47413731e+00 -1.02436864e+00 -5.48829079e-01 5.94172657e-01
-1.77809465e+00 -3.51164758e-01 7.95987248e-01 -5.31551957e-01
6.29058421e-01 6.54299617e-01 5.30257344e-01 5.76067686e-01
6.87585890e-01 7.58396685e-01 1.67936921e+00 -4.49552566e-01
9.95190889e-02 3.50681216e-01 1.50064394e-01 5.17833650e-01
3.08475137e-01 2.79985756e-01 -5.49907148e-01 -7.61111736e-01
7.16815889e-01 -5.79578504e-02 -6.13726616e-01 -1.92442298e-01
-1.26488030e+00 9.44885194e-01 3.84638131e-01 2.67843485e-01
-1.75622389e-01 -2.61440516e-01 3.11851501e-01 5.89482903e-01
2.25279793e-01 3.92311096e-01 -1.14073090e-01 4.81947064e-01
-1.13138759e+00 2.80042235e-02 5.24816096e-01 6.63218617e-01
8.21840703e-01 1.42133728e-01 8.50415751e-02 5.60594857e-01
2.06215963e-01 1.00838530e+00 7.74915576e-01 -6.45556390e-01
4.89030123e-01 3.52259040e-01 2.21345827e-01 -1.71785426e+00
-3.31409454e-01 -2.36569092e-01 -9.84175384e-01 2.35152736e-01
9.06083435e-02 7.38357827e-02 -5.53546906e-01 8.84134114e-01
2.79840678e-01 4.97754008e-01 2.32732028e-01 1.26952446e+00
4.28400844e-01 9.45880175e-01 -3.33308131e-01 -4.19568509e-01
1.18704581e+00 -6.25443459e-01 -2.60731459e-01 -2.24720314e-01
4.45575982e-01 -7.72470117e-01 7.74649799e-01 7.16834545e-01
-2.04912111e-01 -3.63627106e-01 -1.14557743e+00 5.83636999e-01
-8.95451903e-02 6.04235530e-01 4.77461517e-01 8.09111595e-01
-9.11665440e-01 5.74682236e-01 -6.18496835e-01 -1.43343017e-01
1.03294097e-01 3.80309939e-01 -6.23173237e-01 5.34759238e-02
-1.07926440e+00 6.24278486e-01 6.16799474e-01 7.84117579e-02
-7.59312928e-01 1.93501394e-02 -5.52398145e-01 -1.07168257e-01
-7.91584998e-02 -4.78863046e-02 5.06218016e-01 -1.08384931e+00
-1.27722359e+00 7.77148008e-01 8.32124352e-02 -4.60207403e-01
3.47464979e-01 1.53349228e-02 -8.66872668e-01 6.01569057e-01
6.26485646e-02 1.06169343e-01 1.61304390e+00 -1.19573367e+00
-4.49962318e-01 -5.47958851e-01 -7.18504131e-01 -2.45163292e-02
-6.86325669e-01 2.26791561e-01 -1.97922215e-01 -9.10480797e-01
9.71756041e-01 -1.10111547e+00 -1.41651794e-01 -1.10487312e-01
-5.44058383e-01 3.67638648e-01 1.40437579e+00 -7.80363142e-01
1.19215167e+00 -2.50956774e+00 -2.80535612e-02 6.37212873e-01
1.78039700e-01 5.00344813e-01 -1.20471820e-01 3.61454636e-01
-3.34111810e-01 -1.52607784e-01 -1.58687696e-01 3.64459366e-01
-4.53879297e-01 -2.22815812e-01 -6.52717054e-01 1.20805931e+00
-5.80396466e-02 2.69618124e-01 -7.76426613e-01 -4.52177256e-01
2.49998674e-01 7.58305117e-02 -2.59743094e-01 -5.35210446e-02
6.33473992e-01 1.54573068e-01 -4.62851584e-01 9.08128083e-01
1.19238734e+00 -2.44971141e-02 1.15500078e-01 -3.88306826e-01
-2.14539215e-01 -3.36430371e-01 -1.30790269e+00 8.04003358e-01
9.84778181e-02 9.25218999e-01 2.12505266e-01 -1.10015726e+00
1.20171213e+00 1.78760797e-01 3.92338157e-01 -4.47211295e-01
9.79260206e-02 5.42405665e-01 -4.36836243e-01 -6.66261792e-01
6.66135192e-01 -1.37090115e-02 -2.02881947e-01 2.92524666e-01
-3.67133156e-03 -9.22360122e-02 -3.04942816e-01 2.13150576e-01
9.39308941e-01 -3.44262660e-01 3.95647168e-01 -5.11490703e-01
8.00417602e-01 1.65883064e-01 5.89955628e-01 6.74419641e-01
-5.72001636e-01 2.05371499e-01 1.73102364e-01 -5.65487742e-01
-8.31224799e-01 -9.33654010e-01 -4.60232258e-01 6.15060866e-01
6.06514454e-01 -3.66719186e-01 -5.75493336e-01 -5.16299367e-01
3.94179374e-01 4.01089251e-01 -1.11925922e-01 -2.80109674e-01
-2.70138174e-01 -1.05157959e+00 7.72310078e-01 -2.12434128e-01
9.32572722e-01 -5.31621814e-01 -3.41842532e-01 1.32314429e-01
-1.79246023e-01 -9.42867339e-01 -2.53506601e-01 -1.10833555e-01
-1.18600249e+00 -9.32415068e-01 -4.87869889e-01 -6.23463631e-01
5.85673273e-01 9.97028768e-01 3.83571118e-01 -2.41375826e-02
-3.01752359e-01 8.44655931e-02 -4.70242471e-01 1.32973462e-01
-3.48318726e-01 -5.77201009e-01 4.83621627e-01 4.53397870e-01
4.94630069e-01 -3.18159997e-01 -4.55991775e-01 5.30749798e-01
-7.17687488e-01 -1.40821576e-01 7.54646301e-01 1.23857594e+00
5.18476427e-01 4.89093184e-01 1.70590635e-02 -3.01319987e-01
2.94653058e-01 -5.64473033e-01 -8.90377343e-01 6.35046735e-02
-6.03687704e-01 -1.97683826e-01 7.27707446e-01 -3.45783144e-01
-5.40867925e-01 3.04250062e-01 6.55308008e-01 -6.49933159e-01
-6.60313806e-03 3.59079063e-01 1.15847051e-01 -6.29194736e-01
9.46778059e-01 8.39083612e-01 7.79514909e-02 -2.88344026e-01
7.75286630e-02 1.29110003e+00 7.72646308e-01 -5.66775911e-02
1.39408791e+00 7.33214855e-01 -2.12715775e-01 -1.43377376e+00
-6.95395395e-02 -7.60474563e-01 -5.07355511e-01 -3.51290494e-01
2.11510643e-01 -1.25151789e+00 -5.12888193e-01 9.19847310e-01
-8.28706086e-01 2.17587680e-01 4.45499718e-01 7.15136886e-01
-1.59749329e-01 7.37630665e-01 -6.66721940e-01 -8.25606823e-01
-2.64661819e-01 -1.19481242e+00 8.95540357e-01 1.76143706e-01
2.93320239e-01 -4.83573675e-01 -3.02629888e-01 4.06322330e-01
5.10404855e-02 1.95259154e-01 4.13510382e-01 -4.54672128e-01
-7.69748688e-01 -4.66391981e-01 -3.30270529e-01 3.68249387e-01
-1.67487800e-01 -2.32419387e-01 -9.29993093e-01 -8.59549165e-01
2.96289027e-01 -1.51403740e-01 8.93767118e-01 8.08168352e-02
1.00040090e+00 -5.04601955e-01 -4.49382037e-01 8.78142357e-01
1.63792253e+00 1.25098433e-02 7.40867317e-01 6.68767214e-01
4.94111925e-01 2.91804254e-01 7.88083315e-01 4.99790281e-01
1.01538710e-01 7.01716959e-01 3.95619452e-01 3.80992168e-03
3.53917599e-01 1.65279955e-01 7.51737297e-01 7.43548155e-01
2.38729775e-01 1.38637498e-01 -8.20177555e-01 3.21691245e-01
-1.43672979e+00 -1.17208803e+00 -4.65715677e-01 2.28216958e+00
4.99301821e-01 -1.53085962e-01 -8.33054930e-02 3.21199417e-01
9.75686371e-01 2.37005472e-01 -3.64539951e-01 -4.63304408e-02
-4.29852456e-01 -3.05507809e-01 1.12426686e+00 2.33740434e-01
-1.58314919e+00 1.12490845e+00 6.08986378e+00 1.00520647e+00
-1.59705067e+00 -6.29386678e-02 4.46071327e-01 3.36397201e-01
1.27820268e-01 2.06678271e-01 -5.84624529e-01 6.93999052e-01
7.75025129e-01 -2.55727887e-01 4.77666050e-01 8.89199853e-01
5.34650087e-01 -3.87808114e-01 -2.10821420e-01 1.40663207e+00
2.98816651e-01 -1.10206139e+00 -5.55300480e-03 1.79504439e-01
6.51520193e-01 1.77289143e-01 8.99025518e-03 -1.02421060e-01
-3.30828242e-02 -7.69558489e-01 6.10332370e-01 4.04830784e-01
8.83765697e-01 -7.52642334e-01 7.69443035e-01 4.19248432e-01
-9.49179828e-01 -5.02931893e-01 -7.03726470e-01 1.06806695e-01
-3.77588481e-01 6.87348664e-01 -9.98508275e-01 6.40344739e-01
8.58912647e-01 6.92544878e-01 -6.60333693e-01 1.18311107e+00
-6.23024032e-02 9.04973090e-01 -3.73949766e-01 1.65768802e-01
9.78341848e-02 -3.87714982e-01 8.60251784e-01 1.26001430e+00
6.08605146e-01 4.85376149e-01 4.04834658e-01 3.61201733e-01
4.44848269e-01 2.02869177e-01 -6.24014854e-01 -3.40745486e-02
8.54373395e-01 1.07680309e+00 -6.80636048e-01 -3.64778787e-01
-7.11102337e-02 1.35904753e+00 -3.78223002e-01 2.54943371e-01
-5.27773857e-01 -5.34411907e-01 6.23094678e-01 -2.23672554e-01
2.87229806e-01 -3.17718744e-01 -2.95723587e-01 -1.46872389e+00
-1.34922966e-01 -1.18582928e+00 4.11528826e-01 -6.84991479e-01
-9.52205062e-01 5.55101335e-01 -1.84339628e-01 -2.02225661e+00
1.16379544e-01 -7.27331042e-01 -3.58379453e-01 8.76198649e-01
-1.33510482e+00 -9.63696599e-01 -3.59893143e-01 9.95164514e-01
3.12394530e-01 -6.50644064e-01 8.48673582e-01 -5.31360321e-02
-6.10650420e-01 5.51835537e-01 7.82406032e-01 9.45109949e-02
6.50169849e-01 -7.48837709e-01 1.68837443e-01 1.11743355e+00
1.20699085e-01 2.91540802e-01 8.76372278e-01 -6.47040725e-01
-1.71611178e+00 -8.90139759e-01 7.50204504e-01 4.69617695e-02
6.54860139e-01 -1.11291498e-01 -9.07579482e-01 1.91549703e-01
-3.07650149e-01 3.47580940e-01 5.54778099e-01 -6.84451461e-01
-6.44762754e-01 -2.80192614e-01 -1.35836852e+00 2.28822082e-01
2.77378410e-01 -7.92642415e-01 -4.39818174e-01 5.84550917e-01
-5.38679138e-02 -3.39003861e-01 -6.54295027e-01 -2.13939786e-01
5.63123107e-01 -1.23869228e+00 1.15470469e+00 -2.31105655e-01
-1.36438003e-02 -7.23138213e-01 -3.49741906e-01 -1.32518578e+00
-5.15904486e-01 -7.52521455e-01 1.53421953e-01 4.81306344e-01
-8.62907022e-02 -8.87920439e-01 6.52198613e-01 -1.69065118e-01
2.04719856e-01 -1.79935306e-01 -1.15415084e+00 -1.06800175e+00
-2.24955350e-01 -1.29686713e-01 3.43321115e-01 1.26732588e+00
1.72106221e-01 -3.34512740e-01 -7.42642641e-01 9.30594623e-01
1.08220780e+00 3.65339637e-01 8.64825010e-01 -1.20506144e+00
-2.96850234e-01 4.21327725e-02 -1.00491965e+00 -7.75738299e-01
9.15449113e-03 -7.55907178e-01 -1.10653952e-01 -5.10787785e-01
1.43035248e-01 -2.10824147e-01 -1.57608286e-01 4.81023818e-01
-9.05742347e-02 7.10551381e-01 1.82891712e-01 9.99913573e-01
3.22511271e-02 3.71159524e-01 8.22909892e-01 -4.49393511e-01
-2.25429922e-01 -2.09252015e-02 -2.16972113e-01 5.13847768e-01
7.98911095e-01 -5.43017268e-01 6.52672201e-02 -7.64323324e-02
-5.07647038e-01 3.76151055e-01 6.83798790e-01 -1.23895931e+00
5.89262322e-02 -2.97702700e-01 6.17304265e-01 -4.52856034e-01
2.90686667e-01 -8.25614989e-01 1.77195728e-01 9.74426866e-01
3.81362677e-01 -1.63912177e-01 7.51192793e-02 7.35858440e-01
-3.13908219e-01 -8.53663590e-03 9.30359483e-01 1.03255779e-01
-1.17438316e+00 -1.76983461e-01 -5.65626144e-01 -6.54003859e-01
1.12541473e+00 -6.06464326e-01 -3.38192791e-01 -2.72264004e-01
-4.25813794e-01 -8.33078474e-02 4.87687677e-01 1.35853663e-01
8.42984200e-01 -1.26540935e+00 -8.39346290e-01 7.11762011e-01
2.54863560e-01 -8.77496898e-01 3.03340763e-01 9.73289549e-01
-7.41967261e-01 3.78490031e-01 -3.23491514e-01 -9.01626885e-01
-1.58470523e+00 3.42583299e-01 1.89168304e-01 1.97314590e-01
-7.39872754e-01 7.24057853e-01 -3.47932667e-01 -6.91779032e-02
-4.31121796e-01 3.14358562e-01 -1.64441124e-01 6.87358156e-02
7.17981994e-01 5.34794986e-01 1.47087365e-01 -1.36181033e+00
-6.00608647e-01 5.04621029e-01 9.06948820e-02 -1.77336290e-01
1.20938826e+00 -1.48462042e-01 -3.99618119e-01 -5.93050346e-02
1.26297116e+00 6.76000565e-02 -1.10051048e+00 -2.57510126e-01
1.94579333e-01 -1.21147776e+00 5.16233087e-01 -3.85853052e-01
-8.98509443e-01 3.25018674e-01 1.06562793e+00 1.25708774e-01
1.44009769e+00 -3.91228318e-01 5.98542333e-01 6.11167431e-01
8.44956398e-01 -1.08937085e+00 3.27021703e-02 4.29407328e-01
9.18089569e-01 -1.23155606e+00 3.26965868e-01 -4.47338521e-01
-3.59239191e-01 1.28936017e+00 2.65935689e-01 -4.30156589e-01
4.85673666e-01 -1.53704286e-01 2.79555529e-01 -2.02891991e-01
-1.63718045e-01 2.32409462e-01 1.09806648e-02 4.20609415e-01
-2.28257105e-01 3.45827878e-01 -5.41467369e-01 1.67811234e-02
-1.71610266e-01 -2.58860767e-01 7.06010759e-01 8.66785645e-01
-7.12750971e-01 -7.50399590e-01 -1.43621576e+00 4.60033566e-01
-8.20064172e-02 5.45721166e-02 -3.02172244e-01 5.94981730e-01
-9.81603637e-02 9.84597325e-01 -1.01767249e-01 -1.11028099e+00
-2.06479654e-02 -1.74157396e-01 4.89306338e-02 -1.31442085e-01
-2.68448412e-01 2.39953920e-01 -3.08658272e-01 -7.90942371e-01
-1.91150233e-01 -7.48227596e-01 -8.15968513e-01 -3.65269899e-01
-6.61051095e-01 3.10787499e-01 7.42466390e-01 7.33018935e-01
3.10976058e-01 -3.92950416e-01 1.43024075e+00 -1.13525391e+00
-1.05568612e+00 -7.85526037e-01 -1.08845639e+00 1.48382887e-01
3.94974619e-01 -5.64753294e-01 -7.55366862e-01 3.01156696e-02] | [12.374902725219727, 0.7276913523674011] |
896e31a5-583c-42af-ab51-8df67773876e | psuedoprop-at-semeval-2020-task-11-propaganda | null | null | https://aclanthology.org/2020.semeval-1.233 | https://aclanthology.org/2020.semeval-1.233.pdf | PsuedoProp at SemEval-2020 Task 11: Propaganda Span Detection Using BERT-CRF and Ensemble Sentence Level Classifier | This paper explains our teams{'} submission to the Shared Task of Fine-Grained Propaganda Detection in which we propose a sequential BERT-CRF based Span Identification model where the fine-grained detection is carried out only on the articles that are flagged as containing propaganda by an ensemble SLC model. We propose this setup bearing in mind the practicality of this approach in identifying propaganda spans in the exponentially increasing content base where the fine-tuned analysis of the entire data repository may not be the optimal choice due to its massive computational resource requirements. We present our analysis on different voting ensembles for the SLC model. Our system ranks 14th on the test set and 22nd on the development set and with an F1 score of 0.41 and 0.39 respectively. | ['Harshita Diddee', 'Aniruddha Chauhan'] | 2020-12-01 | null | null | null | semeval-2020 | ['propaganda-detection'] | ['natural-language-processing'] | [-1.65922374e-01 -1.30752966e-01 4.81436327e-02 4.77472991e-02
-7.29518056e-01 -7.34910309e-01 1.01207411e+00 4.06076878e-01
-8.05821180e-01 9.47425246e-01 4.70748663e-01 -5.05806684e-01
-3.27973962e-01 -6.76410973e-01 -3.48980099e-01 -4.83499408e-01
3.96427885e-02 6.69197619e-01 2.86832482e-01 -1.61319301e-01
8.45962703e-01 3.96155924e-01 -9.45975959e-01 1.99484229e-01
1.21057320e+00 5.86036325e-01 2.52153501e-02 8.03036094e-01
-6.82319179e-02 1.16575396e+00 -1.06554687e+00 -4.87860858e-01
3.67663689e-02 -2.51344860e-01 -9.40780461e-01 -3.75896394e-01
4.99145120e-01 -1.81188300e-01 -1.42218620e-01 9.33146417e-01
1.85222819e-01 -2.37111539e-01 7.69497752e-01 -4.52681929e-01
-3.42269480e-01 1.20601428e+00 -7.43914247e-01 6.56350553e-01
3.46943527e-01 -4.12022203e-01 9.39934552e-01 -7.98131943e-01
1.02219307e+00 1.05808854e+00 7.57461369e-01 1.58278614e-01
-1.14718580e+00 -3.51124823e-01 -1.21242501e-01 1.12775415e-01
-1.31629205e+00 -2.84978896e-01 4.90531832e-01 -9.21712637e-01
7.75889993e-01 1.38599217e-01 4.28235680e-01 1.12240827e+00
3.07130754e-01 1.83655709e-01 1.59832537e+00 -8.09215963e-01
1.54056713e-01 1.05534475e-02 6.43330455e-01 7.40518332e-01
5.78115880e-01 -1.63420156e-01 -5.46151340e-01 -6.37579501e-01
4.08277392e-01 -5.40460587e-01 2.25078776e-01 7.35804975e-01
-1.00312603e+00 1.09631062e+00 2.27512382e-02 6.03739679e-01
-5.08763492e-01 4.82715890e-02 5.14739871e-01 2.71784335e-01
8.68085325e-01 5.02669811e-01 -2.01421306e-01 -1.56870872e-01
-1.56900048e+00 5.16121030e-01 7.95604944e-01 2.93317616e-01
2.93655246e-01 -2.23283485e-01 -5.68032682e-01 6.36454523e-01
1.16095856e-01 3.18728268e-01 8.44468623e-02 -3.33634228e-01
4.34273720e-01 4.69241560e-01 2.89560705e-01 -6.36415720e-01
-5.67176938e-01 -6.61438107e-01 -5.81564009e-01 6.89819902e-02
7.21514165e-01 -4.94611055e-01 -9.50223327e-01 1.52823770e+00
2.28944689e-01 -1.90675259e-01 -5.82269430e-01 5.75146854e-01
6.51271045e-01 3.58669817e-01 4.82169151e-01 -4.47069615e-01
1.56685197e+00 -5.70497513e-01 -5.57694554e-01 3.68461698e-01
4.29098815e-01 -1.12729025e+00 4.94298965e-01 4.01583016e-01
-6.02466941e-01 -2.51885891e-01 -9.04799044e-01 3.52532029e-01
-9.88820270e-02 1.95800766e-01 3.55535626e-01 6.37111783e-01
-6.55421555e-01 6.69351637e-01 -4.02186245e-01 -3.18063825e-01
3.27849299e-01 -2.75303900e-01 1.73602032e-03 2.65165210e-01
-1.29414237e+00 1.21217966e+00 4.25577044e-01 -1.04754627e-01
-8.37581992e-01 -4.13777024e-01 -2.42432848e-01 -1.12780243e-01
2.51439750e-01 -3.58918697e-01 7.91588366e-01 -2.78564900e-01
-7.99824715e-01 1.03784084e+00 1.92487046e-01 -5.96713006e-01
8.41559708e-01 -3.80648881e-01 -6.84107959e-01 8.10314044e-02
4.39341098e-01 -2.49366179e-01 9.83690858e-01 -7.92485416e-01
-5.21237910e-01 -1.42912462e-01 -2.18238942e-02 -3.39703470e-01
-6.15139864e-02 8.12528312e-01 2.16262668e-01 -7.34108925e-01
-3.20252568e-01 -8.76082003e-01 -1.87408835e-01 -1.04392481e+00
-7.73298323e-01 -6.63817585e-01 4.24983680e-01 -1.28280735e+00
1.85577703e+00 -1.56345737e+00 -1.82152316e-02 4.41707850e-01
5.23268640e-01 9.09351110e-02 1.79422602e-01 8.73850107e-01
2.66528130e-01 3.19146842e-01 4.15465444e-01 -4.72892188e-02
-5.24595566e-03 -4.20328587e-01 -5.86010635e-01 6.49092436e-01
2.78397631e-02 4.06805634e-01 -7.07852185e-01 -9.14297342e-01
-3.04595649e-01 1.29568474e-02 -1.21980645e-01 1.14374161e-01
-2.13146910e-01 3.28448266e-01 -5.54692924e-01 5.87081730e-01
5.94933510e-01 -3.65564913e-01 1.94281757e-01 2.12083548e-01
-6.58992410e-01 4.68444049e-01 -9.44661021e-01 1.24484921e+00
-3.52177918e-02 6.41687632e-01 5.70176467e-02 -4.33328658e-01
7.43572116e-01 1.93474278e-01 2.09502786e-01 -4.50269490e-01
1.21319383e-01 1.78593233e-01 1.56759486e-01 -1.71879083e-01
8.44654739e-01 -1.43507302e-01 -6.45479918e-01 6.73506260e-01
-2.58695409e-02 4.50846523e-01 7.04705954e-01 7.09504247e-01
1.53306484e+00 3.04413866e-02 4.58473295e-01 -6.22087717e-01
4.09918517e-01 4.48505163e-01 5.15121698e-01 1.16295326e+00
-1.43097103e-01 1.87729195e-01 5.95019519e-01 -3.80552083e-01
-1.15322042e+00 -9.14774954e-01 -5.62402725e-01 1.36158657e+00
-4.04364526e-01 -7.24226058e-01 -7.78799891e-01 -9.93013620e-01
-2.44049728e-01 5.75378001e-01 -8.33973944e-01 5.92476249e-01
-7.00481474e-01 -6.47440553e-01 9.47277308e-01 -1.50424942e-01
4.21832979e-01 -1.06971872e+00 -7.60686755e-01 3.47061336e-01
-2.28703290e-01 -7.06217468e-01 -1.52652279e-01 2.17532098e-01
-2.88003236e-01 -1.15884769e+00 -5.77064812e-01 -3.76532048e-01
2.82456964e-01 -3.21928829e-01 1.11899483e+00 2.39326939e-01
-3.77318323e-01 -2.67569155e-01 -6.98640049e-01 -3.02718788e-01
-7.65112281e-01 2.14369252e-01 2.11485833e-01 -4.77921486e-01
1.59051582e-01 -2.46517614e-01 -2.39915147e-01 -9.01693106e-02
-3.89756322e-01 -2.52419449e-02 2.91584104e-01 9.13514733e-01
-2.02191547e-02 3.75521407e-02 9.41261649e-01 -1.25139046e+00
1.06313884e+00 -5.58196843e-01 -5.67522287e-01 4.95428115e-01
-5.97492874e-01 -6.21928973e-03 5.84118545e-01 -1.20834492e-01
-1.15059888e+00 -4.87036169e-01 -1.02784470e-01 1.35862738e-01
-1.17174968e-01 6.52787507e-01 5.82091033e-01 1.18963510e-01
1.01618826e+00 -3.07258427e-01 -5.82309008e-01 -9.57837641e-01
1.95229694e-01 7.50690222e-01 2.84056902e-01 -5.49317002e-01
8.94194782e-01 -3.51796113e-03 -2.39679888e-01 -6.41743660e-01
-9.81707394e-01 -5.75859487e-01 -6.17600203e-01 -3.17514598e-01
7.84707844e-01 -8.97948444e-01 -4.19577479e-01 4.07707512e-01
-1.45745397e+00 9.12388787e-02 6.78392425e-02 3.32149804e-01
-1.78123936e-01 6.23776376e-01 -8.73376489e-01 -9.87448990e-01
-7.81544209e-01 -4.90692854e-01 7.31014609e-01 1.47762641e-01
-7.55524337e-01 -8.87327731e-01 6.25528574e-01 2.77000099e-01
2.06242427e-01 5.42035937e-01 1.01575136e+00 -8.12701285e-01
1.81753308e-01 -1.67293563e-01 -2.28226721e-01 -7.41381720e-02
-2.19721541e-01 1.55930623e-01 -8.66264343e-01 2.73627881e-03
-2.08315313e-01 -3.97071242e-01 1.18864882e+00 3.64435643e-01
7.23232448e-01 -1.87203959e-01 -3.07178319e-01 -2.53615737e-01
1.55806255e+00 -2.83362031e-01 4.54642922e-01 5.33756912e-01
4.81151730e-01 5.56647718e-01 7.22505033e-01 7.87894845e-01
-1.15779765e-01 6.72048211e-01 1.17108040e-01 2.67556310e-01
-2.99539566e-01 -4.07746047e-01 3.42833936e-01 6.77731693e-01
-5.93707144e-01 -2.41272941e-01 -9.62552249e-01 6.99751854e-01
-1.76002002e+00 -1.51821661e+00 -6.41706169e-01 1.96569037e+00
8.80878568e-01 2.80241609e-01 4.79493439e-01 1.60962716e-02
9.59410727e-01 2.17932537e-01 3.26616228e-01 -8.10802817e-01
-1.98701665e-01 4.20350134e-01 3.93847942e-01 4.84690696e-01
-1.32813895e+00 8.27531755e-01 6.69158077e+00 1.34221959e+00
-5.73532224e-01 4.54663813e-01 3.75007182e-01 -2.14015357e-02
-1.54042423e-01 1.62837401e-01 -1.07371926e+00 8.71401608e-01
1.08898592e+00 -4.84567732e-02 1.31629109e-01 5.16727686e-01
2.50391334e-01 -4.99467969e-01 -5.92418432e-01 4.81541932e-01
-1.14960365e-01 -1.77171481e+00 -2.96853662e-01 3.40903699e-01
8.70289743e-01 3.35251004e-01 -4.92173195e-01 2.91554958e-01
7.55188644e-01 -1.05909169e+00 1.06507158e+00 5.87555289e-01
8.26797843e-01 -5.77959955e-01 5.70195615e-01 7.78603315e-01
-8.26410413e-01 9.27125439e-02 -2.38987520e-01 -2.44327307e-01
4.23974246e-01 1.08918273e+00 -9.86362219e-01 5.87087989e-01
4.54292387e-01 -1.46946125e-02 -5.72358310e-01 1.07286274e+00
-3.58296096e-01 1.19302273e+00 -3.50195289e-01 -2.39432946e-01
3.45239818e-01 1.28042564e-01 6.81742787e-01 1.68419290e+00
2.59866744e-01 -3.03276241e-01 3.01244676e-01 5.20286560e-01
-9.16305650e-03 9.97976121e-03 -2.09246531e-01 5.88817112e-02
5.64143777e-01 1.46997571e+00 -7.02005863e-01 -2.86890447e-01
6.03112504e-02 4.41793263e-01 7.51533508e-01 -1.76172629e-01
-7.69403279e-01 -3.18276286e-01 -6.05153926e-02 2.69789696e-01
2.51924306e-01 -2.01271340e-01 -4.86715347e-01 -9.78914857e-01
-3.42407167e-01 -4.92135316e-01 8.04831207e-01 -9.77794454e-02
-1.49838209e+00 8.39509189e-01 7.57790282e-02 -8.42064202e-01
-4.31591183e-01 -4.48899865e-01 -8.76473427e-01 1.03515100e+00
-9.39356923e-01 -1.12717998e+00 4.71974909e-02 2.18908891e-01
4.11271393e-01 -2.20150217e-01 8.78316760e-01 9.80651900e-02
-3.66118133e-01 1.81750670e-01 1.10406511e-01 1.41758010e-01
6.65177107e-01 -1.18596470e+00 9.98026431e-02 1.04165375e+00
5.25818653e-02 6.72780156e-01 1.20768964e+00 -1.08928382e+00
-6.72341824e-01 -8.94770563e-01 1.37741125e+00 -6.95325315e-01
1.09291458e+00 -2.88739502e-01 -5.90234220e-01 2.70170420e-01
4.45559233e-01 -6.55885518e-01 6.70919657e-01 6.90062463e-01
-6.73397779e-01 5.16752541e-01 -1.12887812e+00 3.05666532e-02
7.78387249e-01 -4.20786709e-01 -8.86986852e-01 7.42983401e-01
1.44576997e-01 3.76383737e-02 -1.03016782e+00 2.03011604e-03
6.81184590e-01 -5.90502322e-01 3.24073315e-01 -5.94523370e-01
7.51401782e-01 -1.62713259e-01 -3.75897177e-02 -1.09800220e+00
-7.92603672e-01 -3.81061018e-01 -1.16281897e-01 1.36535895e+00
2.93755025e-01 -2.94825256e-01 4.35777009e-01 -1.18265696e-01
1.94168687e-01 -4.13781434e-01 -1.25729442e+00 -7.20510542e-01
2.21771508e-01 -2.46298119e-01 8.26729685e-02 8.19526315e-01
2.63661265e-01 6.64247990e-01 -8.19495082e-01 -7.57594928e-02
9.04210508e-01 5.02413034e-01 4.64595735e-01 -1.54674304e+00
-4.06741172e-01 -2.88060188e-01 -9.49187502e-02 -3.60520482e-01
8.17021355e-02 -6.84602976e-01 2.62207314e-02 -1.47869909e+00
7.40552306e-01 -6.55702591e-01 -2.12077036e-01 1.47137463e-01
3.63539234e-02 3.86181563e-01 1.46864116e-01 5.79014003e-01
-7.60138333e-01 -1.50252059e-01 6.80952549e-01 4.61846776e-02
9.68791842e-02 -9.57159474e-02 -6.42447174e-01 7.10184336e-01
8.08167934e-01 -6.93659306e-01 2.81982750e-01 -2.43635267e-01
5.54181337e-01 -2.96832528e-02 2.74716020e-01 -6.91942155e-01
1.02377735e-01 -3.85534823e-01 5.23031056e-01 -9.55865204e-01
-2.38903433e-01 -1.04581788e-01 3.12067717e-01 7.33471990e-01
-1.38504148e-01 -1.60186559e-01 -1.92771554e-01 5.54800153e-01
1.54386997e-01 -6.78704202e-01 6.50547564e-01 -2.32136831e-01
-5.05877137e-01 -1.18142903e-01 -5.96994638e-01 -8.16103369e-02
7.07544386e-01 2.00726852e-01 -9.74944770e-01 5.44432513e-02
-7.74502158e-01 -1.51067317e-01 3.41998756e-01 1.07374147e-01
1.09490603e-01 -9.28925216e-01 -1.39276040e+00 -3.53564203e-01
-1.49563774e-02 -7.99239337e-01 3.06681722e-01 9.94329691e-01
-5.82571924e-01 4.43522096e-01 -1.76001385e-01 -7.82788619e-02
-1.57851803e+00 2.60752946e-01 -3.01851183e-01 -1.23835635e+00
-4.27822202e-01 8.12165082e-01 -2.22755194e-01 1.46787181e-01
-3.72879386e-01 4.06605452e-01 -5.65596640e-01 2.22068161e-01
5.83409488e-01 9.18277979e-01 1.80680647e-01 -7.42889106e-01
-7.46374786e-01 1.42471224e-01 -4.86706406e-01 -7.19994083e-02
1.29572821e+00 1.27199903e-01 -3.70822281e-01 1.79716393e-01
4.31643218e-01 8.38289857e-01 -5.93679905e-01 -1.34258136e-01
2.22918734e-01 -9.53861251e-02 8.40113983e-02 -1.14548445e+00
-1.26807302e-01 3.88292879e-01 2.25686714e-01 7.24131644e-01
5.19077182e-01 1.99804708e-01 2.52224892e-01 4.85027358e-02
4.96855110e-01 -1.55388808e+00 -2.55839199e-01 9.59665239e-01
9.15080190e-01 -6.56687677e-01 6.66667044e-01 -3.86467904e-01
-2.59671599e-01 1.01501012e+00 1.63890749e-01 -5.15425086e-01
4.96433973e-01 1.55058190e-01 -4.21732366e-01 -5.76802313e-01
-7.66756594e-01 -2.30552197e-01 2.96821535e-01 1.94779858e-01
4.03675616e-01 4.94115889e-01 -1.35503566e+00 5.11061668e-01
-2.46859878e-01 -1.78800508e-01 4.61313754e-01 6.16491854e-01
-8.42873096e-01 -8.86973977e-01 -6.05391502e-01 8.83790135e-01
-1.07851589e+00 -2.42932066e-01 -6.30315125e-01 5.64670920e-01
3.56461674e-01 1.19928205e+00 -3.28888521e-02 -4.31947023e-01
-9.25528705e-02 2.08265573e-01 6.64056182e-01 -5.89978278e-01
-1.17136014e+00 1.81840986e-01 9.82761621e-01 9.58884433e-02
-3.60243559e-01 -7.52753556e-01 -9.19822812e-01 -7.09052026e-01
-4.67332870e-01 5.40206730e-01 5.93348444e-01 1.17967415e+00
-2.84608193e-02 3.09345037e-01 6.04982316e-01 -3.27828795e-01
-9.59580600e-01 -1.50317633e+00 -7.93909729e-01 2.38644421e-01
-4.88547422e-02 -6.32843614e-01 -2.51368761e-01 2.11886764e-02] | [8.48582935333252, 10.735417366027832] |
7d7deb6e-7d8e-4250-8efc-ba9e0f6c4879 | show-attend-and-read-a-simple-and-strong | 1811.00751 | null | http://arxiv.org/abs/1811.00751v2 | http://arxiv.org/pdf/1811.00751v2.pdf | Show, Attend and Read: A Simple and Strong Baseline for Irregular Text Recognition | Recognizing irregular text in natural scene images is challenging due to the
large variance in text appearance, such as curvature, orientation and
distortion. Most existing approaches rely heavily on sophisticated model
designs and/or extra fine-grained annotations, which, to some extent, increase
the difficulty in algorithm implementation and data collection. In this work,
we propose an easy-to-implement strong baseline for irregular scene text
recognition, using off-the-shelf neural network components and only word-level
annotations. It is composed of a $31$-layer ResNet, an LSTM-based
encoder-decoder framework and a 2-dimensional attention module. Despite its
simplicity, the proposed method is robust and achieves state-of-the-art
performance on both regular and irregular scene text recognition benchmarks.
Code is available at: https://tinyurl.com/ShowAttendRead | ['Peng Wang', 'Chunhua Shen', 'Hui Li', 'Guyu Zhang'] | 2018-11-02 | null | null | null | null | ['irregular-text-recognition'] | ['computer-vision'] | [ 2.91797161e-01 -3.71461272e-01 2.88062636e-02 -4.70178515e-01
-6.72715425e-01 -3.63179326e-01 6.80170119e-01 1.45178726e-02
-3.61177772e-01 6.26307502e-02 2.32744738e-01 -3.15396219e-01
5.16602159e-01 -4.57922429e-01 -7.96318233e-01 -5.58629274e-01
5.91446102e-01 3.07573527e-01 2.24704996e-01 -6.92301244e-02
4.69010144e-01 3.80231678e-01 -1.25151205e+00 3.70948106e-01
8.18214953e-01 1.00024283e+00 3.48643988e-01 9.24049199e-01
-4.31735128e-01 1.13408995e+00 -3.25685531e-01 -6.41060591e-01
-3.98618244e-02 -4.86724414e-02 -6.49950266e-01 3.24669808e-01
7.70848989e-01 -6.12924814e-01 -7.36397505e-01 8.07553470e-01
6.22248232e-01 3.13347043e-03 5.13912916e-01 -4.49196368e-01
-9.67263758e-01 4.74167466e-01 -7.18485415e-01 4.07496095e-03
2.57000744e-01 2.24673405e-01 9.82497811e-01 -1.14086103e+00
1.76133707e-01 9.32634294e-01 8.28286767e-01 4.77392465e-01
-1.00624466e+00 -4.83871341e-01 2.99201697e-01 7.67066702e-02
-1.43099856e+00 -7.52148628e-01 7.96010733e-01 -4.65113044e-01
1.31982851e+00 1.13065951e-01 3.43992025e-01 1.47662461e+00
1.69635564e-01 1.14108956e+00 8.26895654e-01 -3.59978795e-01
-1.03114359e-02 -5.08943684e-02 9.12571624e-02 8.44602406e-01
1.78841695e-01 -6.23941958e-01 -4.39444274e-01 2.25521222e-01
9.23969805e-01 1.06744595e-01 -2.63273209e-01 -2.24492610e-01
-1.01067662e+00 4.81555432e-01 3.57441247e-01 2.27726713e-01
-1.08964011e-01 2.96706498e-01 4.98128623e-01 -4.92141349e-03
5.67377150e-01 1.46290556e-01 -4.30451274e-01 -4.32673365e-01
-1.01936257e+00 -1.50628567e-01 6.12410486e-01 9.12665963e-01
5.18798769e-01 3.44174504e-01 -1.38615640e-02 1.17090404e+00
3.32346052e-01 5.45954466e-01 7.43702352e-01 -1.66081876e-01
6.96281016e-01 7.56930590e-01 -1.98861077e-01 -9.30369794e-01
-3.38309646e-01 -2.62619942e-01 -9.58738923e-01 -6.68872893e-02
2.23448232e-01 5.51513545e-02 -1.23557293e+00 1.11627090e+00
-3.92958410e-02 1.12156365e-02 -3.00607800e-01 8.75800073e-01
9.77079570e-01 4.22779262e-01 -1.84979782e-01 5.12898505e-01
1.04401433e+00 -1.38690770e+00 -5.32466829e-01 -5.20606577e-01
5.71120024e-01 -1.10849690e+00 1.66416407e+00 2.85961807e-01
-1.19454432e+00 -4.44420040e-01 -1.12006962e+00 -6.10507429e-01
-4.32381183e-01 4.54897583e-01 2.67504841e-01 6.28061652e-01
-1.00411212e+00 3.66981506e-01 -1.02322686e+00 -5.67953646e-01
5.72117209e-01 2.63390332e-01 -7.92660788e-02 -1.36619762e-01
-5.12117684e-01 6.87813580e-01 -6.61923960e-02 2.40964845e-01
-5.63765466e-01 -3.47560048e-01 -8.18181098e-01 5.10365106e-02
3.39249492e-01 -4.88677025e-01 1.43501556e+00 -1.01740718e+00
-1.84000254e+00 9.21893120e-01 -3.00805032e-01 -2.49389887e-01
6.72126234e-01 -3.37849259e-01 -2.51680493e-01 -5.18128425e-02
-2.41635069e-01 3.89507979e-01 1.20524096e+00 -9.81247723e-01
-2.10705087e-01 -3.14616412e-01 -2.50451267e-01 3.24970841e-01
-5.00670969e-01 1.75475165e-01 -8.27272594e-01 -9.80024159e-01
-7.07866028e-02 -7.62847364e-01 -2.82232761e-02 6.01456799e-02
-6.14706933e-01 -4.71919328e-02 1.02259970e+00 -7.67850161e-01
1.04999864e+00 -2.24452519e+00 -1.27288595e-01 -1.96995169e-01
1.47815555e-01 2.41958529e-01 -1.22423865e-01 2.82072574e-01
1.37398094e-01 1.96471885e-01 -1.11726671e-01 -7.65045166e-01
2.74221987e-01 -9.71050039e-02 -3.70200932e-01 5.67596614e-01
1.50428116e-01 1.02931511e+00 -2.68917680e-01 -3.76076400e-01
5.61148286e-01 7.02680826e-01 -2.94001073e-01 1.59550339e-01
-4.71359074e-01 4.61058393e-02 -4.35381114e-01 7.74161160e-01
6.45130396e-01 -6.26497626e-01 -1.85046718e-01 -1.75806746e-01
-2.01694787e-01 5.40236890e-01 -1.04788983e+00 1.84192193e+00
-4.48568523e-01 1.00694048e+00 5.51851131e-02 -7.07795441e-01
7.45426357e-01 1.35377765e-01 6.77428171e-02 -7.96713769e-01
4.03391749e-01 1.81310833e-01 -3.01065832e-01 -4.46288943e-01
8.31662595e-01 5.03415167e-01 1.99113801e-01 4.90649730e-01
-1.94937319e-01 -1.69106036e-01 -3.17722782e-02 2.91067250e-02
1.10695028e+00 1.42934382e-01 7.44237378e-02 -1.66716591e-01
4.04769719e-01 -3.82642299e-01 9.09471512e-02 7.25743234e-01
2.61373408e-02 9.46902871e-01 2.88801759e-01 -6.25699937e-01
-1.32577538e+00 -6.71172380e-01 -2.07183018e-01 1.31213665e+00
-2.24857740e-02 -4.84324813e-01 -7.53841043e-01 -5.28666854e-01
-1.59507886e-01 5.25829256e-01 -6.74395502e-01 2.42559180e-01
-5.42566359e-01 -6.08400345e-01 8.13720524e-01 6.77696884e-01
7.82749712e-01 -1.00436962e+00 -4.53567505e-01 -1.05761938e-01
3.12038660e-02 -1.36664855e+00 -8.94675791e-01 2.60578752e-01
-7.76624680e-01 -7.36455560e-01 -7.36960769e-01 -8.00799787e-01
8.15397978e-01 4.74065334e-01 9.96227980e-01 3.03179640e-02
-3.60030085e-01 2.94218928e-01 -4.10808593e-01 -3.05492401e-01
2.54516699e-03 1.80949509e-01 -3.83124858e-01 9.43039134e-02
3.97049546e-01 -1.56668469e-01 -6.47433996e-01 3.95356536e-01
-9.51419830e-01 4.74549085e-01 7.40197420e-01 7.97804415e-01
4.29277778e-01 -2.16127381e-01 6.64411113e-02 -6.73851728e-01
5.16236305e-01 9.79949012e-02 -6.48036182e-01 9.90888774e-02
-5.17520189e-01 -1.38406858e-01 8.49027872e-01 -4.26042378e-01
-1.06648076e+00 1.36438236e-01 -2.87320137e-01 -3.45112622e-01
-4.26176906e-01 4.45412070e-01 3.04524917e-02 -2.72948056e-01
6.01610899e-01 5.92671812e-01 -4.03612584e-01 -6.63721263e-01
1.52397439e-01 9.91837084e-01 2.43812382e-01 -4.03864413e-01
6.20384991e-01 4.41212386e-01 -5.38918257e-01 -1.24484527e+00
-8.83228540e-01 -2.36692384e-01 -7.26857901e-01 9.10637453e-02
7.75961936e-01 -1.01841366e+00 -4.43863094e-01 1.10777962e+00
-1.00447345e+00 -9.75448966e-01 -1.03636622e-01 1.28343433e-01
-3.61735851e-01 5.80715716e-01 -8.50053847e-01 -4.97097939e-01
-6.53309286e-01 -1.26615345e+00 1.43496335e+00 1.96035966e-01
8.05704221e-02 -9.79778051e-01 -2.72133708e-01 5.67542195e-01
7.24554479e-01 -2.39160255e-01 6.82852209e-01 -4.33834165e-01
-5.95977187e-01 -3.40307266e-01 -6.20549381e-01 3.09184670e-01
5.44245690e-02 2.37772599e-01 -1.14490020e+00 -3.82996291e-01
-1.90779254e-01 -6.37581289e-01 8.24215472e-01 5.34079373e-01
1.65834570e+00 -2.08254725e-01 -6.05012872e-04 8.38147342e-01
1.26217794e+00 -1.80587068e-01 7.38100290e-01 3.72384340e-01
1.26573372e+00 1.02854662e-01 -6.47584498e-02 6.61734939e-01
5.55727422e-01 6.66527331e-01 1.99205816e-01 -3.63291711e-01
-3.42188716e-01 -1.26313597e-01 3.40114802e-01 9.31123912e-01
8.15499723e-02 -4.32442486e-01 -1.15042734e+00 3.14885318e-01
-1.72731352e+00 -7.15068579e-01 -3.37000281e-01 2.05317783e+00
9.21356380e-01 1.61165550e-01 -4.10718173e-02 -9.59769189e-02
6.22345388e-01 4.70764667e-01 -7.38767564e-01 -6.24605894e-01
-3.09326053e-01 5.91745153e-02 6.44254744e-01 3.66809636e-01
-1.11008751e+00 1.26231110e+00 6.10734415e+00 9.38205004e-01
-1.60114670e+00 -8.77992809e-02 7.73649752e-01 -1.58047646e-01
-1.31601293e-03 -5.06104231e-01 -8.06781769e-01 3.73317242e-01
8.50051463e-01 5.66230305e-02 5.85905969e-01 7.86845803e-01
2.44930446e-01 1.78694829e-01 -9.28466439e-01 1.12752390e+00
5.54259539e-01 -1.32145870e+00 2.91490499e-02 -6.02348186e-02
6.97883427e-01 7.49145865e-01 2.14684620e-01 2.44846158e-02
2.45288625e-01 -1.19509208e+00 7.35728204e-01 2.64128476e-01
1.21221554e+00 -3.84962171e-01 5.87650836e-01 1.06492788e-01
-1.30020475e+00 1.19531967e-01 -3.31716806e-01 3.84913310e-02
-2.66721278e-01 4.74916726e-01 -5.85290670e-01 1.07485056e-01
8.93454194e-01 1.03112066e+00 -7.84305990e-01 8.81905556e-01
-1.27868950e-01 8.14357996e-01 -3.31141472e-01 -2.66285151e-01
3.35172743e-01 3.38689648e-02 1.16370983e-01 1.68075359e+00
1.83415025e-01 -2.54667789e-01 1.84934869e-01 6.56831682e-01
-5.59332609e-01 3.14130127e-01 -5.04699111e-01 -8.86995569e-02
9.65366066e-02 1.28251088e+00 -7.17099488e-01 -3.19777131e-01
-9.06777561e-01 1.24184203e+00 3.99061382e-01 3.96351725e-01
-9.11545515e-01 -6.06015325e-01 3.21748137e-01 1.41391799e-01
6.51492834e-01 -3.73749137e-01 -6.78280771e-01 -1.56323695e+00
3.43840450e-01 -1.11908793e+00 1.10208720e-01 -9.64507043e-01
-1.21846223e+00 6.68653250e-01 -7.10415244e-01 -8.98760021e-01
1.75161213e-01 -8.32975030e-01 -6.08631194e-01 8.22248220e-01
-1.59704208e+00 -1.41785288e+00 -7.48897910e-01 7.95335770e-01
1.21363139e+00 -1.01181395e-01 7.79002726e-01 2.50466317e-01
-9.78970587e-01 8.91508639e-01 3.92468542e-01 4.76322085e-01
8.08862090e-01 -1.21378398e+00 9.55360770e-01 8.11245859e-01
8.71175975e-02 2.82674611e-01 4.13696647e-01 -4.73857433e-01
-1.80176628e+00 -1.17210078e+00 7.07169771e-01 -4.48420346e-01
9.80495811e-01 -8.58460844e-01 -1.04934931e+00 8.47010195e-01
4.15836900e-01 1.87249556e-02 5.71915269e-01 1.71928748e-01
-6.16365433e-01 6.31276146e-02 -8.04990590e-01 9.95377898e-01
7.37322748e-01 -7.49012411e-01 -3.45132142e-01 3.61664802e-01
3.79472136e-01 -6.16681218e-01 -7.49079943e-01 1.54151976e-01
7.15754211e-01 -8.05446088e-01 6.05783880e-01 -1.59682021e-01
6.96714699e-01 2.12531071e-03 -1.41779706e-01 -9.12932634e-01
-4.42911595e-01 -6.50117397e-01 -1.12275332e-01 1.02730024e+00
4.65530396e-01 -7.16624796e-01 9.71927643e-01 6.71600938e-01
-3.62423062e-01 -8.63989472e-01 -7.08647907e-01 -4.71368611e-01
1.19432926e-01 -4.85321403e-01 2.79435962e-01 9.19932544e-01
-6.31688610e-02 5.08041799e-01 -4.40403551e-01 -8.06757016e-04
3.68160069e-01 2.38914900e-02 8.61783564e-01 -9.28983629e-01
-2.23402590e-01 -7.58096635e-01 -2.58245677e-01 -1.55299103e+00
5.36924042e-02 -5.09105444e-01 1.75597057e-01 -1.54036641e+00
3.26911479e-01 -2.77700156e-01 -6.40196726e-02 6.02184892e-01
-6.13146797e-02 3.88190061e-01 6.76661590e-03 2.55691558e-01
-8.87144446e-01 7.30788589e-01 1.04950988e+00 -3.29884261e-01
-6.70586154e-02 -3.41835052e-01 -6.27098024e-01 9.21818674e-01
1.10066414e+00 -7.00508058e-02 -8.59083831e-02 -1.22106016e+00
2.03288302e-01 -4.57886606e-01 2.96621233e-01 -8.87085676e-01
3.81273448e-01 1.79193150e-02 5.53228736e-01 -6.27514780e-01
4.29388076e-01 -6.56574428e-01 -4.18641567e-01 9.51899588e-02
-4.46462512e-01 1.40771851e-01 4.52637494e-01 3.70587856e-01
-4.47437726e-03 -2.11165771e-01 8.26023042e-01 6.60185739e-02
-3.48835886e-01 3.58566642e-01 -5.23427725e-01 1.44405477e-02
4.70636278e-01 -3.76777619e-01 -8.01177442e-01 -3.95281434e-01
-5.96829280e-02 -4.31122929e-02 9.47539091e-01 4.59159881e-01
5.90947032e-01 -8.47376585e-01 -7.92952538e-01 2.51443952e-01
1.21200368e-01 1.84767202e-01 2.90113956e-01 8.88310492e-01
-6.56665266e-01 5.26107609e-01 9.79583561e-02 -6.12170637e-01
-1.29543459e+00 1.98463395e-01 3.82896900e-01 -2.07373619e-01
-9.27923739e-01 7.77269900e-01 2.44849533e-01 -3.06329161e-01
5.22422194e-01 -4.71602082e-01 1.76456794e-01 -4.24614817e-01
4.81817752e-01 1.53908476e-01 2.36232415e-01 -6.34442270e-01
-1.98234975e-01 9.31399345e-01 -4.28718656e-01 2.34531701e-01
1.10576570e+00 -1.58166707e-01 1.26837984e-01 5.27122736e-01
1.10559070e+00 1.39098361e-01 -1.36200595e+00 -3.71279627e-01
-2.89438277e-01 -4.91452008e-01 3.03980798e-01 -8.76841784e-01
-9.62321997e-01 1.05272830e+00 4.02265817e-01 2.03517720e-01
1.10668111e+00 -1.81957662e-01 8.59256744e-01 8.34249079e-01
-1.32991686e-01 -1.25379837e+00 2.56577849e-01 8.75178576e-01
9.14125443e-01 -1.49315333e+00 4.27027009e-02 -1.59311518e-01
-7.14202344e-01 1.09243000e+00 6.35765851e-01 -7.28348792e-02
4.52270061e-01 6.43105626e-01 2.64977098e-01 -4.51670438e-02
-8.36016059e-01 -2.41602473e-02 2.65518934e-01 2.61021972e-01
1.03011239e+00 -2.12644488e-01 2.84508318e-01 2.40089834e-01
-6.39598891e-02 -1.79479852e-01 4.81525987e-01 7.45074213e-01
-2.86347091e-01 -8.26926649e-01 -9.80845764e-02 7.72059202e-01
-6.62897050e-01 -4.77783650e-01 -5.62676191e-01 5.54247797e-01
-4.28199768e-01 6.95858359e-01 1.93730786e-01 -3.41611087e-01
3.54235768e-01 -2.02756301e-02 3.78495485e-01 -3.93368542e-01
-6.07208490e-01 3.25957328e-01 5.82229048e-02 -4.69323546e-01
-2.78791841e-02 -5.98446012e-01 -1.08420646e+00 -5.64164877e-01
-4.15839940e-01 -4.41545665e-01 9.65004563e-01 7.86832094e-01
6.62639260e-01 5.94859302e-01 4.61428195e-01 -1.02571785e+00
-3.60742688e-01 -1.20823896e+00 -3.46597791e-01 2.88039148e-01
3.02181125e-01 -1.68743595e-01 -1.66489035e-01 3.68696988e-01] | [11.890118598937988, 2.20296049118042] |
93efa047-bf69-4cbd-9a49-18e522ea7020 | semantic-nearest-neighbor-fields-monocular | 1904.00738 | null | http://arxiv.org/abs/1904.00738v1 | http://arxiv.org/pdf/1904.00738v1.pdf | Semantic Nearest Neighbor Fields Monocular Edge Visual-Odometry | Recent advances in deep learning for edge detection and segmentation opens up
a new path for semantic-edge-based ego-motion estimation. In this work, we
propose a robust monocular visual odometry (VO) framework using category-aware
semantic edges. It can reconstruct large-scale semantic maps in challenging
outdoor environments. The core of our approach is a semantic nearest neighbor
field that facilitates a robust data association of edges across frames using
semantics. This significantly enlarges the convergence radius during tracking
phases. The proposed edge registration method can be easily integrated into
direct VO frameworks to estimate photometrically, geometrically, and
semantically consistent camera motions. Different types of edges are evaluated
and extensive experiments demonstrate that our proposed system outperforms
state-of-art indirect, direct, and semantic monocular VO systems. | ['Assia Benbihi', 'Cedric Pradalier', 'Antoine Richard', 'Xiaolong Wu'] | 2019-04-01 | null | null | null | null | ['monocular-visual-odometry'] | ['robots'] | [-4.85577613e-01 -3.93035114e-01 -1.09028399e-01 -3.92456651e-01
-1.93145126e-01 -5.61062753e-01 4.72191781e-01 -3.84458661e-01
-3.45153272e-01 6.16423011e-01 -4.00036313e-02 1.73278376e-01
1.74971133e-01 -8.87277424e-01 -9.28816319e-01 -3.85491788e-01
2.19031706e-01 4.34663385e-01 8.00229073e-01 -8.15560147e-02
2.92747289e-01 4.56502348e-01 -1.55647016e+00 -4.44526345e-01
8.72837722e-01 8.49678457e-01 5.68778634e-01 4.82990921e-01
-1.11573301e-01 5.75216234e-01 6.40196055e-02 -8.37835968e-02
3.12933564e-01 -9.02829468e-02 -6.24088585e-01 3.69251132e-01
9.90943491e-01 -4.29038465e-01 -6.66061461e-01 1.36793089e+00
5.44278204e-01 3.04240495e-01 4.16448265e-01 -1.36206794e+00
-6.65369332e-01 3.67025919e-02 -4.30229098e-01 1.04797808e-02
6.63158953e-01 4.93950322e-02 6.77644074e-01 -1.08656073e+00
1.17720735e+00 1.04205501e+00 9.95885551e-01 2.26394027e-01
-7.43730664e-01 -3.33041877e-01 2.69769996e-01 6.15365088e-01
-1.66243458e+00 -4.41600829e-01 9.78128433e-01 -3.29092354e-01
1.06037056e+00 -1.49616897e-01 9.66154456e-01 8.31403792e-01
1.73390985e-01 7.15072215e-01 7.58547723e-01 -2.39982024e-01
2.22400308e-01 -1.38844147e-01 -7.93444216e-02 9.53602374e-01
5.19034147e-01 2.15180263e-01 -5.94828725e-01 2.72806942e-01
1.07691169e+00 9.72485542e-02 -4.53856260e-01 -1.18247163e+00
-1.12614119e+00 4.86850411e-01 7.16715097e-01 1.17464744e-01
-2.03883827e-01 5.29207766e-01 1.45086020e-01 2.62454748e-02
4.96174127e-01 -1.80643320e-01 -3.21654141e-01 1.15754426e-01
-7.80871689e-01 1.74142141e-02 7.01296449e-01 1.57818246e+00
1.14434505e+00 3.20548415e-01 2.87997305e-01 6.27229154e-01
4.56128299e-01 6.97686672e-01 2.40299225e-01 -1.38641441e+00
1.55608058e-02 4.36061263e-01 3.99485826e-01 -1.25883138e+00
-5.92545688e-01 -3.39138150e-01 -4.79566544e-01 1.33839741e-01
3.64314109e-01 9.84788388e-02 -7.90042758e-01 1.49219370e+00
7.18408287e-01 8.46812963e-01 -1.23811372e-01 1.21327817e+00
8.25881660e-01 1.33140400e-01 -7.67770857e-02 3.17480952e-01
1.06413102e+00 -1.18601108e+00 -6.54380381e-01 -5.45704126e-01
3.58599633e-01 -7.63344347e-01 4.44491059e-01 -3.06006148e-02
-9.69673395e-01 -6.44865096e-01 -9.76905167e-01 -4.94039267e-01
-3.66055936e-01 5.73410839e-02 8.55869353e-01 5.23177445e-01
-1.56419683e+00 4.58755106e-01 -8.39326143e-01 -5.94816446e-01
2.29257181e-01 2.41836742e-01 -5.56518555e-01 -3.70995760e-01
-1.03339970e+00 6.63490951e-01 5.29103100e-01 -8.73068348e-03
-7.53986537e-01 -4.00986284e-01 -1.55165100e+00 -1.76897362e-01
3.18120569e-01 -1.43873990e+00 7.38053322e-01 -5.96576631e-01
-1.34319973e+00 1.18876886e+00 -5.37220180e-01 -3.70262057e-01
8.08702528e-01 -3.74148518e-01 -1.96606338e-01 2.37237722e-01
3.68420482e-01 1.02623105e+00 6.09228611e-01 -1.36883509e+00
-7.66141534e-01 -4.60424423e-01 -3.07740122e-01 5.72132409e-01
6.80359975e-02 -4.75539565e-01 -1.03868055e+00 -4.25220281e-01
7.05897510e-01 -1.07771671e+00 -2.39769787e-01 3.59582365e-01
-2.21310019e-01 6.76051080e-02 7.27092385e-01 -6.04139447e-01
6.54934704e-01 -1.78749931e+00 8.02636333e-03 -8.36954638e-02
1.66782245e-01 -9.16072503e-02 2.41858169e-01 -8.82698596e-02
3.87880594e-01 -5.59769571e-01 -1.00323083e-02 -6.68787301e-01
5.03135845e-02 2.15373635e-01 1.60523683e-01 8.90552282e-01
-3.74436885e-01 1.17301214e+00 -1.20708275e+00 -4.73388761e-01
8.41962278e-01 7.00008094e-01 -5.83643317e-01 -4.79800105e-02
9.61426869e-02 5.62338293e-01 -3.16516817e-01 8.42868507e-01
1.05476153e+00 -2.31206760e-01 -1.13737874e-01 -2.80386508e-01
-2.12488607e-01 1.16330065e-01 -1.65855026e+00 2.42902112e+00
-2.35318124e-01 8.82497013e-01 -2.72104330e-02 -7.85623550e-01
1.00487471e+00 -1.54775098e-01 6.04892969e-01 -4.64494348e-01
2.88351893e-01 1.99053660e-01 -9.47750211e-01 -1.76582158e-01
9.81735528e-01 3.38854879e-01 1.34924904e-01 -2.15715840e-01
2.15630010e-01 -2.58025795e-01 -4.68062386e-02 7.24879578e-02
4.97400284e-01 7.45013654e-01 3.70850235e-01 -3.43804508e-01
7.65165091e-01 2.78374285e-01 8.93619120e-01 5.60497224e-01
-5.77871382e-01 7.76317716e-01 -6.43051505e-01 -5.08441567e-01
-1.03599691e+00 -1.40428972e+00 -1.77769944e-01 3.77826482e-01
1.22487009e+00 -5.69274686e-02 -6.88089430e-01 -3.06812555e-01
1.79651424e-01 2.66428620e-01 -7.79889971e-02 7.68505409e-02
-4.10066605e-01 -3.19159091e-01 1.27277479e-01 7.04462826e-01
9.58014250e-01 -6.78407907e-01 -3.18361044e-01 3.23763460e-01
-6.41623616e-01 -1.79035628e+00 -7.84038305e-01 -3.34640503e-01
-9.07439590e-01 -1.09566486e+00 -6.40808225e-01 -1.14027488e+00
4.99236763e-01 1.00914538e+00 1.04051042e+00 -4.24736351e-01
-1.71954677e-01 7.07243145e-01 -4.41216417e-02 2.49814969e-02
-5.86468391e-02 -2.43524715e-01 2.72038639e-01 -1.30277187e-01
7.38209486e-01 -5.30395031e-01 -1.02333128e+00 4.67199534e-01
-2.96296149e-01 7.54950941e-02 7.20258206e-02 4.56482708e-01
7.87867725e-01 -3.60279828e-01 1.09568238e-01 -3.43861341e-01
-2.14057148e-01 -4.27304745e-01 -9.57220018e-01 1.20519117e-01
-6.10279799e-01 -2.86102351e-02 1.67137340e-01 -1.45995356e-02
-1.10546470e+00 4.04749036e-01 9.61873084e-02 -5.96264005e-01
-3.24318558e-01 -1.38054550e-01 -2.03406140e-01 -7.36914396e-01
3.97677302e-01 4.19077784e-01 -2.93874651e-01 -2.27763966e-01
7.87021458e-01 5.62025189e-01 1.12708724e+00 -1.93235084e-01
7.80777872e-01 1.24185288e+00 -1.86744202e-02 -8.41779590e-01
-4.45445061e-01 -1.09953547e+00 -9.19884562e-01 -4.63486403e-01
1.16012239e+00 -1.63800704e+00 -5.58557391e-01 8.71797621e-01
-1.29090273e+00 -2.44412228e-01 1.28245084e-02 7.51888573e-01
-9.13791358e-01 9.20005679e-01 -7.00811207e-01 -4.97850537e-01
-2.67190725e-01 -1.06831956e+00 1.41024280e+00 4.44354415e-01
1.04499636e-02 -1.39429092e+00 9.81178135e-02 3.91094655e-01
1.00770965e-01 3.00229698e-01 -2.86731213e-01 7.53738731e-02
-1.28637767e+00 5.52965775e-02 -3.98326874e-01 -6.71272050e-04
5.60822822e-02 -4.73973751e-01 -9.82262313e-01 -2.40508154e-01
-8.84518772e-02 2.16436103e-01 7.65294850e-01 7.51935542e-01
4.67721343e-01 3.63915086e-01 -6.23892367e-01 1.41311514e+00
1.73891449e+00 -4.75417711e-02 6.32888615e-01 6.81870103e-01
1.29763508e+00 2.38261715e-01 6.52979136e-01 3.86036813e-01
1.08746839e+00 8.47119093e-01 3.73236805e-01 -1.41299963e-01
-4.70085919e-01 -4.05808836e-01 3.30743402e-01 7.34574974e-01
9.48721729e-03 -3.24957222e-02 -6.94389939e-01 8.72003436e-01
-1.94039512e+00 -9.14672554e-01 -5.93818784e-01 2.09670019e+00
3.52737904e-01 -1.80103138e-01 -3.62493187e-01 -3.33588779e-01
9.89727259e-01 1.09645866e-01 -6.15575016e-01 1.41500011e-01
-3.42261791e-01 -1.98367119e-01 1.04403841e+00 8.24258208e-01
-1.36764145e+00 1.60802412e+00 6.00858784e+00 4.46241766e-01
-8.66705358e-01 3.04239333e-01 1.46714404e-01 4.69526738e-01
-1.55043975e-01 1.01405643e-01 -1.08592308e+00 2.44846702e-01
2.99239337e-01 -9.82106626e-02 3.43011051e-01 1.22287953e+00
1.63836390e-01 -3.07559520e-01 -8.19196165e-01 1.47445309e+00
2.39926144e-01 -1.57346082e+00 -1.61285773e-01 -8.76476169e-02
1.30437374e+00 6.12891734e-01 -4.16322351e-01 -2.66804785e-01
6.10518754e-01 -4.23697293e-01 9.12569642e-01 5.62080801e-01
5.46642900e-01 -5.41449845e-01 4.93943512e-01 1.29961595e-01
-1.82510829e+00 1.90982088e-01 -6.50433183e-01 -1.25146374e-01
6.30191863e-01 5.12086749e-01 -5.60559213e-01 9.17290151e-01
9.82292593e-01 1.36573493e+00 -4.49000120e-01 1.34266984e+00
-2.78342664e-01 -3.78396481e-01 -3.29507709e-01 3.66845638e-01
2.33388186e-01 -5.80291450e-01 7.18662441e-01 1.04256964e+00
4.62423831e-01 -1.29615039e-01 3.38514954e-01 9.88042772e-01
1.47073558e-02 -1.15215980e-01 -8.50854933e-01 6.99990094e-01
8.23849618e-01 1.17069554e+00 -9.07434762e-01 -5.33942163e-01
-5.89597285e-01 1.60945487e+00 2.89878666e-01 4.60641474e-01
-7.28177965e-01 -7.90103227e-02 9.04956698e-01 -1.35469958e-01
3.97026867e-01 -7.72990584e-01 -2.65237361e-01 -1.58419144e+00
5.96866719e-02 -3.32782753e-02 1.56097993e-01 -1.20870543e+00
-9.86439526e-01 2.02117354e-01 -2.81071991e-01 -1.32868719e+00
-1.27990305e-01 -4.52829778e-01 -3.79956812e-01 6.21537507e-01
-1.96252561e+00 -1.16345108e+00 -1.16557908e+00 8.98873985e-01
5.04981995e-01 1.68862939e-01 2.38029703e-01 4.06541795e-01
-2.68059447e-02 2.07879722e-01 3.64614457e-01 2.25205287e-01
7.19536483e-01 -1.21553373e+00 8.90957594e-01 1.27604485e+00
3.64858717e-01 3.51606816e-01 5.54686487e-01 -8.02479625e-01
-1.41096413e+00 -1.35117555e+00 7.29544401e-01 -6.62394702e-01
4.54707295e-01 -2.00352207e-01 -7.17800379e-01 8.23458731e-01
-1.14523374e-01 1.79793507e-01 -2.26055235e-02 -6.15764439e-01
-1.62099246e-02 1.07946217e-01 -1.01939523e+00 5.99367738e-01
1.82068086e+00 -5.72746336e-01 -4.12471652e-01 1.38733849e-01
7.58169472e-01 -8.16539705e-01 -7.70897985e-01 2.57138073e-01
4.23651934e-01 -1.06716633e+00 1.56904948e+00 1.78720742e-01
-3.27114224e-01 -6.13970816e-01 -1.33720979e-01 -1.10084081e+00
-1.69329941e-01 -6.35740101e-01 -1.73604935e-01 9.66066420e-01
-4.86570179e-01 -7.37284362e-01 9.35366094e-01 4.54803765e-01
-4.46160972e-01 9.20622870e-02 -8.91294241e-01 -9.86455083e-01
-3.93128693e-01 -5.28114974e-01 3.91661227e-01 1.27193844e+00
-3.62863779e-01 -1.56510293e-01 -3.65659654e-01 5.41450560e-01
1.22426450e+00 6.14519529e-02 1.03669441e+00 -1.26975930e+00
-5.83925843e-02 -2.83082128e-01 -1.16847646e+00 -1.62179255e+00
3.22339267e-01 -9.52121019e-01 1.50086684e-02 -1.82096791e+00
-7.93705210e-02 -4.22026664e-01 2.97232643e-02 -9.13782418e-02
-2.47129798e-01 5.31242430e-01 -8.30753669e-02 2.78793842e-01
-8.27004194e-01 7.45382309e-01 1.12709343e+00 1.22211212e-02
-2.57160246e-01 -4.41254139e-01 -1.01673797e-01 9.50455606e-01
5.44440925e-01 -3.13735873e-01 -3.04335892e-01 -5.54888368e-01
3.93692963e-02 -1.82882994e-01 6.77513361e-01 -1.20999110e+00
6.74032867e-01 -2.05587707e-02 4.50120479e-01 -6.67971253e-01
3.00626248e-01 -8.97081494e-01 4.00705069e-01 3.82420033e-01
5.09987772e-01 -9.70675703e-03 -1.95210904e-03 9.67803299e-01
-2.07057491e-01 1.07809857e-01 6.74799621e-01 -2.38288879e-01
-1.81326425e+00 5.64154565e-01 1.68381602e-01 2.24694833e-01
1.17178345e+00 -7.05242217e-01 -1.79550812e-01 -3.99651408e-01
-8.16195726e-01 3.01478982e-01 1.09134984e+00 6.94892168e-01
8.02576125e-01 -1.52097416e+00 -3.16064894e-01 1.91057503e-01
3.21560770e-01 1.04706682e-01 3.73579949e-01 8.57566595e-01
-1.03677547e+00 4.77212250e-01 -3.74346614e-01 -1.16539347e+00
-1.04729891e+00 4.82164860e-01 5.66260099e-01 4.79570925e-01
-1.03833640e+00 7.00320303e-01 2.62672424e-01 -4.88562316e-01
1.97722577e-02 -2.27815032e-01 8.33323747e-02 -3.96310240e-01
2.30306402e-01 7.18792558e-01 -9.39319730e-02 -1.15857697e+00
-5.62816918e-01 1.21097541e+00 5.30900359e-01 3.31305861e-02
1.08895695e+00 -1.00552762e+00 7.97109157e-02 1.71929255e-01
1.21249342e+00 -1.88936830e-01 -1.73379648e+00 -3.48352671e-01
-8.29828382e-02 -7.33930588e-01 3.78215134e-01 2.43790466e-02
-1.05957687e+00 5.69752872e-01 8.25552702e-01 -4.53784078e-01
8.45433891e-01 4.88110213e-03 9.77383196e-01 1.78915679e-01
1.05621898e+00 -1.15449357e+00 -2.21275494e-01 4.98070091e-01
1.56476572e-01 -1.57739902e+00 -4.77896966e-02 -8.47907484e-01
-4.40719455e-01 1.03634453e+00 6.34584248e-01 -3.87959540e-01
5.65210700e-01 -4.53947335e-02 1.16781734e-01 -2.50280380e-01
7.08507523e-02 -6.19574606e-01 4.44575608e-01 9.37681377e-01
6.37784749e-02 -6.43087775e-02 -1.34690285e-01 -1.81181505e-01
1.09146629e-02 9.43055153e-02 5.86762488e-01 7.10464001e-01
-6.21156871e-01 -8.52045238e-01 -4.84772563e-01 -5.56879580e-01
1.33029193e-01 -5.64964786e-02 -6.53632954e-02 9.30078983e-01
2.94482987e-02 9.67540979e-01 2.89674312e-01 -8.35117996e-02
1.02485716e-01 -2.22281702e-02 6.11796021e-01 -1.64773121e-01
1.27753809e-01 -1.35397958e-02 -7.77910696e-03 -1.12677097e+00
-6.53671861e-01 -6.34523451e-01 -1.43086231e+00 -3.67122054e-01
-3.79532546e-01 -2.93471992e-01 9.12328541e-01 9.38393474e-01
6.92690253e-01 1.91960737e-01 3.59978348e-01 -1.22263920e+00
5.33481762e-02 -3.88327539e-01 -4.75226671e-01 6.61193073e-01
1.17911272e-01 -9.73127604e-01 -2.67433226e-01 2.56147385e-01] | [8.020857810974121, -2.216367721557617] |
eb941546-73f2-4c59-a0f5-1ab83e3212be | boosting-graph-neural-networks-via-adaptive | 2210.05920 | null | https://arxiv.org/abs/2210.05920v2 | https://arxiv.org/pdf/2210.05920v2.pdf | Boosting Graph Neural Networks via Adaptive Knowledge Distillation | Graph neural networks (GNNs) have shown remarkable performance on diverse graph mining tasks. Although different GNNs can be unified as the same message passing framework, they learn complementary knowledge from the same graph. Knowledge distillation (KD) is developed to combine the diverse knowledge from multiple models. It transfers knowledge from high-capacity teachers to a lightweight student. However, to avoid oversmoothing, GNNs are often shallow, which deviates from the setting of KD. In this context, we revisit KD by separating its benefits from model compression and emphasizing its power of transferring knowledge. To this end, we need to tackle two challenges: how to transfer knowledge from compact teachers to a student with the same capacity; and, how to exploit student GNN's own strength to learn knowledge. In this paper, we propose a novel adaptive KD framework, called BGNN, which sequentially transfers knowledge from multiple GNNs into a student GNN. We also introduce an adaptive temperature module and a weight boosting module. These modules guide the student to the appropriate knowledge for effective learning. Extensive experiments have demonstrated the effectiveness of BGNN. In particular, we achieve up to 3.05% improvement for node classification and 6.35% improvement for graph classification over vanilla GNNs. | ['Nitesh Chawla', 'Chuxu Zhang', 'Yijun Tian', 'Yujie Fan', 'Chunhui Zhang', 'Zhichun Guo'] | 2022-10-12 | null | null | null | null | ['graph-mining'] | ['graphs'] | [ 5.28222881e-02 2.59414643e-01 -4.47385550e-01 -1.74642861e-01
-1.44931421e-01 -4.50095147e-01 2.92444369e-03 4.62553412e-01
-3.33260566e-01 6.38083518e-01 -3.95011842e-01 -5.79275727e-01
-3.55634093e-01 -1.26886380e+00 -8.62614989e-01 -7.90787697e-01
1.54380966e-02 2.73294657e-01 6.84468925e-01 -2.98067570e-01
-1.78299576e-01 5.77120185e-01 -1.52322698e+00 -7.37531856e-02
1.32839382e+00 7.23201036e-01 2.33913496e-01 7.06125319e-01
-2.72752464e-01 1.12342429e+00 -5.76628566e-01 -5.72090924e-01
-7.49752596e-02 -2.92404056e-01 -9.31551337e-01 -2.67036021e-01
4.07291412e-01 -2.33394012e-01 -6.76178455e-01 8.64017546e-01
5.05965710e-01 2.61763275e-01 3.28851074e-01 -1.35610640e+00
-6.46861017e-01 1.21647418e+00 -8.10757816e-01 1.89512417e-01
-1.20413363e-01 -1.71469688e-01 7.39850581e-01 -5.05557716e-01
2.56070226e-01 1.05031312e+00 6.16167486e-01 5.65057278e-01
-8.87826502e-01 -8.14636111e-01 3.54723036e-01 4.53884095e-01
-1.39994323e+00 2.02236287e-02 7.82994390e-01 -8.94101560e-02
5.67941248e-01 2.26855669e-02 9.91316378e-01 6.56376839e-01
-2.94061244e-01 1.11764693e+00 7.32190490e-01 -4.83291239e-01
-9.14335847e-02 1.21025115e-01 4.25554037e-01 1.17561126e+00
6.32706106e-01 -4.44106996e-01 -8.49279165e-01 1.09470235e-02
4.00170058e-01 9.26031247e-02 -4.06309605e-01 -6.35825336e-01
-5.79879999e-01 8.34686995e-01 7.96568871e-01 8.29661340e-02
-1.25728503e-01 3.73363107e-01 3.26222718e-01 5.18793404e-01
3.29745412e-01 -2.00167298e-01 -5.92424810e-01 -4.02486287e-02
-5.12220621e-01 -2.84133941e-01 1.09451950e+00 1.05662870e+00
9.27871048e-01 3.10759902e-01 -2.96957865e-02 7.59435058e-01
3.79634529e-01 4.33146030e-01 4.89800841e-01 -1.35199130e-01
3.19253832e-01 1.07125437e+00 -9.06308770e-01 -1.01939034e+00
-2.02038482e-01 -8.21628094e-01 -1.10521674e+00 -2.27348521e-01
1.50283992e-01 -3.32485616e-01 -1.13380635e+00 1.66478395e+00
8.78813684e-01 8.12074840e-01 7.04797879e-02 3.03385526e-01
1.30837309e+00 6.22568429e-01 7.24525452e-02 -1.65405422e-02
1.15005577e+00 -1.13701510e+00 -2.24991918e-01 -2.05657765e-01
9.35525596e-01 -3.27782154e-01 6.18606269e-01 6.25394642e-01
-1.19821346e+00 -6.09895468e-01 -1.13492632e+00 -7.04717860e-02
-4.58001822e-01 -3.84558469e-01 6.81264937e-01 8.20992827e-01
-1.29400349e+00 7.31756806e-01 -5.50637186e-01 -1.77778766e-01
5.84908843e-01 5.29892504e-01 -2.22488251e-02 -3.12833309e-01
-1.15383303e+00 5.47407091e-01 8.07349384e-01 -1.80812612e-01
-9.17679131e-01 -1.00771868e+00 -8.22014570e-01 4.37995702e-01
7.08229899e-01 -7.47189164e-01 1.26902139e+00 -5.23614168e-01
-1.60024059e+00 3.53751928e-01 2.31998160e-01 -4.51866895e-01
1.88265249e-01 -4.56066057e-02 -1.63634926e-01 1.13304481e-01
-5.25512516e-01 4.41524863e-01 7.67090321e-01 -1.06537390e+00
-7.62824297e-01 -2.95897007e-01 1.25326477e-02 2.34438077e-01
-9.28217888e-01 -3.77364337e-01 -9.43710029e-01 -5.41502714e-01
2.41681531e-01 -8.07427168e-01 -1.42713696e-01 -2.94310153e-01
-2.64684618e-01 -6.60443962e-01 9.91704285e-01 -3.06580782e-01
1.59490132e+00 -1.76473260e+00 5.38286269e-02 6.66500092e-01
9.95266080e-01 7.80597568e-01 -2.33970821e-01 1.83093175e-01
1.28931031e-01 -2.21385108e-03 8.82746428e-02 6.50617704e-02
-1.61196485e-01 5.40161490e-01 3.40597741e-02 1.27626181e-01
2.70964392e-02 1.13609695e+00 -1.03847969e+00 -5.88575304e-01
-2.86313146e-02 5.01732111e-01 -5.19721389e-01 1.71891630e-01
-1.20644726e-01 -2.70411000e-02 -5.28703392e-01 5.82891703e-01
7.96470642e-01 -6.38452947e-01 5.48219323e-01 1.06226072e-01
5.23274720e-01 2.12522298e-01 -1.24956000e+00 1.34087574e+00
-3.04443419e-01 3.19243252e-01 3.45575184e-01 -1.46340692e+00
1.19604099e+00 5.08541539e-02 2.54671365e-01 -5.03952801e-01
1.60541475e-01 -4.37695347e-02 1.06931828e-01 -1.59522310e-01
5.83979428e-01 1.84807420e-01 2.50856727e-01 6.63687110e-01
3.51485133e-01 -4.07439284e-02 2.03765929e-01 6.42584085e-01
1.17620325e+00 -2.64824808e-01 1.12771973e-01 -1.12445399e-01
5.06874382e-01 -3.23596239e-01 6.33536160e-01 9.59093809e-01
-1.01069793e-01 -1.97273701e-01 4.19845700e-01 -4.66274738e-01
-2.97793746e-01 -1.02447426e+00 4.76975322e-01 1.58659387e+00
1.71281368e-01 -7.50787556e-01 -5.91250896e-01 -1.13116193e+00
2.19971120e-01 2.36159891e-01 -3.90831262e-01 -6.01571798e-01
-6.92306042e-01 -6.79573238e-01 5.36575735e-01 6.09579265e-01
5.86135983e-01 -7.70187914e-01 -1.73025548e-01 3.60596865e-01
2.57626057e-01 -9.41175282e-01 -2.32274666e-01 3.02504897e-01
-9.63884830e-01 -1.14713073e+00 -4.27479088e-01 -1.15017688e+00
8.61706257e-01 8.05066764e-01 1.32757640e+00 7.38973439e-01
-1.47283286e-01 6.02020025e-01 -4.13211316e-01 -8.15647125e-01
-4.90906537e-01 6.20927095e-01 8.78453851e-02 -2.46290326e-01
4.97801900e-01 -8.65702152e-01 -2.76978374e-01 1.00212596e-01
-1.02713740e+00 3.01002115e-01 7.26841331e-01 6.41963482e-01
4.16362077e-01 3.66333902e-01 5.71819007e-01 -1.10912681e+00
6.67120755e-01 -6.74947441e-01 -5.91461301e-01 6.15812302e-01
-9.85557437e-01 2.65576601e-01 7.52467871e-01 -5.51020622e-01
-7.83594728e-01 -8.48653764e-02 1.08810710e-02 -4.03557390e-01
2.13574946e-01 8.00959706e-01 1.31096512e-01 -5.96023023e-01
4.61277008e-01 2.04032987e-01 2.63240486e-02 -4.19473916e-01
4.45772469e-01 4.03745532e-01 3.27137738e-01 -9.00720239e-01
1.12395978e+00 -2.70117018e-02 3.83089930e-02 -7.02537000e-01
-8.92196834e-01 -3.06917071e-01 -5.49381018e-01 -2.77826577e-01
2.07324937e-01 -9.99211788e-01 -1.22037804e+00 6.28749192e-01
-7.56680727e-01 -6.92454100e-01 -2.76571661e-01 3.78773540e-01
2.05920085e-01 4.07482982e-01 -7.79382885e-01 -4.19760406e-01
-5.97240150e-01 -9.45964873e-01 8.53291675e-02 8.12049210e-01
4.72386152e-01 -1.12835956e+00 -1.28051400e-01 5.39427638e-01
5.89981794e-01 -2.17828929e-01 8.88142705e-01 -9.94555891e-01
-6.48077071e-01 1.50293261e-01 -2.80337632e-01 3.05873364e-01
-1.15308724e-02 7.55807832e-02 -9.09458876e-01 -6.16060615e-01
-4.38077986e-01 -7.89527893e-01 1.18406749e+00 -1.15872035e-03
1.55290437e+00 -2.13954747e-01 -4.73650217e-01 6.03899896e-01
1.28084052e+00 -2.96279285e-02 7.42623881e-02 1.59238815e-01
1.19657743e+00 3.62681359e-01 1.48193752e-02 1.96612164e-01
7.58712173e-01 2.82396317e-01 4.05964494e-01 -6.25306293e-02
-2.87384301e-01 -5.01692891e-01 3.84267658e-01 1.75096941e+00
4.96790521e-02 -3.52543563e-01 -1.15498352e+00 5.27357399e-01
-1.68958688e+00 -3.25860441e-01 1.15461624e-03 1.88550019e+00
1.07198155e+00 1.45343438e-01 -7.52861053e-03 7.08516538e-02
7.42524147e-01 1.05987474e-01 -6.32584155e-01 -5.26040614e-01
2.41293103e-01 4.76720750e-01 5.10957241e-01 3.55016261e-01
-6.16561472e-01 9.69144702e-01 5.01964092e+00 1.40507877e+00
-1.07909286e+00 6.33999996e-04 4.22223628e-01 1.13392957e-01
-3.68824780e-01 -8.40629861e-02 -1.15054309e+00 1.20880432e-01
1.25326812e+00 -4.48393583e-01 3.94338936e-01 8.88837934e-01
-4.80362535e-01 1.19166851e-01 -7.03890741e-01 8.07922602e-01
6.93317056e-02 -1.27634144e+00 2.88723350e-01 -1.52949831e-02
9.66799498e-01 1.44496247e-01 -2.21754357e-01 8.83579731e-01
1.01917541e+00 -8.69696498e-01 1.09554775e-01 1.96032748e-01
3.48157555e-01 -1.12021792e+00 4.19314981e-01 4.85534370e-01
-1.54177797e+00 8.16370398e-02 -4.37582582e-01 4.10689190e-02
-4.61835027e-01 7.35882998e-01 -1.30279493e+00 1.05530214e+00
7.06704855e-01 5.47140062e-01 -8.33880603e-01 9.14122045e-01
-4.62788731e-01 9.51946676e-01 -4.45943922e-01 -3.93532962e-01
1.91153288e-01 -1.31879136e-01 6.08157031e-02 1.05361259e+00
2.96850532e-01 2.52637297e-01 4.73760664e-01 4.47060734e-01
-7.36858845e-01 7.45839253e-02 -5.98221123e-01 -4.64654602e-02
8.04663658e-01 1.36183655e+00 -6.73351765e-01 -4.28591639e-01
-5.84161758e-01 7.15408981e-01 1.03338397e+00 1.96900040e-01
-6.62664294e-01 -6.41651332e-01 3.50843281e-01 -1.70363039e-01
5.40590405e-01 -2.00335979e-02 1.86772272e-01 -9.13536906e-01
-9.33894888e-02 -7.55873382e-01 7.74074554e-01 -4.98148054e-01
-9.84798193e-01 2.88990676e-01 -4.86424565e-02 -6.98347986e-01
2.21610487e-01 -4.04994041e-01 -8.31736147e-01 5.21010935e-01
-1.81719613e+00 -1.03200877e+00 -6.38813853e-01 7.95989394e-01
4.47024852e-02 -1.23910971e-01 4.38483715e-01 2.58515209e-01
-9.61060286e-01 1.12018001e+00 1.63454250e-01 2.75582701e-01
5.10452688e-01 -1.40268505e+00 3.67506981e-01 7.69475460e-01
3.00562590e-01 3.82352114e-01 1.20606922e-01 -5.83247304e-01
-1.77209949e+00 -1.26917338e+00 6.06730163e-01 4.03667986e-02
5.90780497e-01 -9.96500477e-02 -1.31692076e+00 7.18515813e-01
6.53830320e-02 2.19586000e-01 7.01117039e-01 3.81125897e-01
-3.97585481e-01 -3.58774990e-01 -9.19914484e-01 6.54003382e-01
9.16885316e-01 -2.76063025e-01 -3.09144735e-01 1.77061498e-01
9.87767100e-01 -6.41134560e-01 -1.12067401e+00 4.25999016e-01
2.24047795e-01 -8.11186492e-01 9.91406739e-01 -6.40408278e-01
7.18063638e-02 -1.11425482e-01 4.74030823e-01 -1.53121960e+00
-2.64618635e-01 -5.99642396e-01 -7.28688896e-01 1.32941723e+00
2.27689669e-01 -5.98917603e-01 1.35251677e+00 3.16957474e-01
-8.94542933e-02 -1.16247296e+00 -4.81608242e-01 -7.56790996e-01
1.09810367e-01 -4.69670534e-01 8.41954947e-01 1.26268220e+00
-1.61910221e-01 5.89250743e-01 -1.60137132e-01 2.18234628e-01
6.45477831e-01 1.75891057e-01 1.11630070e+00 -1.56768942e+00
-2.49992937e-01 -4.54672903e-01 -3.43257904e-01 -1.36179101e+00
9.59348232e-02 -1.26780963e+00 -3.31560254e-01 -1.45212317e+00
3.49576861e-01 -7.35389709e-01 -5.28237581e-01 8.00334930e-01
-5.48634589e-01 -2.86699105e-02 1.84180856e-01 -4.90548640e-01
-9.16850865e-01 5.77842295e-01 1.47138441e+00 -2.06298321e-01
-3.05468738e-01 -7.83153984e-04 -9.76434469e-01 7.05868661e-01
1.11354315e+00 -6.40153527e-01 -8.67880881e-01 -4.29862767e-01
3.03103000e-01 -1.55884355e-01 7.52443746e-02 -9.84895110e-01
8.66455376e-01 -1.45345375e-01 3.40959013e-01 -5.28566718e-01
-1.19103774e-01 -6.57924533e-01 -1.93977788e-01 7.05408692e-01
8.15441366e-03 -1.60215601e-01 3.56180519e-01 6.92664385e-01
-4.73126434e-02 -2.38633454e-01 7.46467769e-01 -2.11816784e-02
-8.55458081e-01 6.09656215e-01 -1.69101264e-02 2.28819132e-01
1.05532610e+00 -8.09040368e-02 -6.07863843e-01 -3.03317726e-01
-4.23421234e-01 7.45067775e-01 1.18965004e-02 2.56534070e-01
7.42305517e-01 -1.17047727e+00 -6.13363564e-01 4.52384114e-01
-1.18292877e-02 5.91809094e-01 4.10411835e-01 7.83632994e-01
-4.42009121e-01 1.82889357e-01 8.04270729e-02 -4.43830132e-01
-1.54565632e+00 5.79681575e-01 9.70048085e-02 -5.23769796e-01
-5.32242775e-01 1.29354477e+00 -8.98453668e-02 -8.64163637e-01
5.72605550e-01 -3.70482683e-01 -1.55848101e-01 6.52975589e-02
5.57530999e-01 5.76005757e-01 2.59842396e-01 7.98738897e-02
-4.01071087e-03 4.01386619e-01 -5.26455939e-01 8.01008046e-01
1.30276120e+00 6.21500127e-02 -1.14458069e-01 -1.35839963e-03
9.43965912e-01 5.06441556e-02 -8.59480560e-01 -8.44590843e-01
-3.74054871e-02 -3.03448420e-02 1.22499220e-01 -6.01960838e-01
-1.61632431e+00 6.69058144e-01 1.37226418e-01 3.76746505e-01
1.28232574e+00 -4.78028245e-02 9.47462380e-01 7.12075472e-01
2.16691405e-01 -9.81658697e-01 1.11489601e-01 6.44838572e-01
6.67384341e-02 -1.18601596e+00 2.86635339e-01 -5.84624767e-01
-1.51529953e-01 1.23542452e+00 1.12282693e+00 3.68940890e-01
7.10040271e-01 9.16751325e-02 -2.24141002e-01 -4.09776449e-01
-1.00145435e+00 -2.45517716e-01 3.17996860e-01 5.07456005e-01
1.12473100e-01 1.58521205e-01 2.24093366e-02 6.34806573e-01
-1.51800498e-01 3.05875763e-02 5.72244525e-01 1.04893768e+00
-8.47652376e-01 -1.19972122e+00 -1.82182670e-01 6.81694150e-01
-1.70002356e-01 -1.05475448e-01 -3.45996559e-01 8.63560319e-01
-1.14050088e-02 9.19547379e-01 -2.92036593e-01 -7.77927220e-01
1.62171543e-01 9.89821181e-02 4.56731200e-01 -5.77877700e-01
-8.30662012e-01 -5.05397081e-01 -9.50837508e-02 -3.17573994e-01
-2.01392174e-01 1.93772227e-01 -1.26303411e+00 -9.37357247e-01
-4.87390637e-01 5.87229788e-01 3.95468801e-01 6.68785751e-01
3.50006491e-01 8.68632674e-01 4.36724186e-01 -1.59312755e-01
-5.34173727e-01 -6.51160836e-01 -7.01960862e-01 -1.66281939e-01
1.64478973e-01 -3.45879883e-01 -3.39172482e-01 -3.94116610e-01] | [9.496859550476074, 3.3726789951324463] |
f4d0916c-2d57-486b-b997-823a6ef7c860 | toch-spatio-temporal-object-correspondence-to | 2205.07982 | null | https://arxiv.org/abs/2205.07982v2 | https://arxiv.org/pdf/2205.07982v2.pdf | TOCH: Spatio-Temporal Object-to-Hand Correspondence for Motion Refinement | We present TOCH, a method for refining incorrect 3D hand-object interaction sequences using a data prior. Existing hand trackers, especially those that rely on very few cameras, often produce visually unrealistic results with hand-object intersection or missing contacts. Although correcting such errors requires reasoning about temporal aspects of interaction, most previous works focus on static grasps and contacts. The core of our method are TOCH fields, a novel spatio-temporal representation for modeling correspondences between hands and objects during interaction. TOCH fields are a point-wise, object-centric representation, which encode the hand position relative to the object. Leveraging this novel representation, we learn a latent manifold of plausible TOCH fields with a temporal denoising auto-encoder. Experiments demonstrate that TOCH outperforms state-of-the-art 3D hand-object interaction models, which are limited to static grasps and contacts. More importantly, our method produces smooth interactions even before and after contact. Using a single trained TOCH model, we quantitatively and qualitatively demonstrate its usefulness for correcting erroneous sequences from off-the-shelf RGB/RGB-D hand-object reconstruction methods and transferring grasps across objects. | ['Bharat Lal Bhatnagar', 'Gerard Pons-Moll', 'Jan Eric Lenssen', 'Keyang Zhou'] | 2022-05-16 | null | null | null | null | ['object-reconstruction'] | ['computer-vision'] | [ 2.55898871e-02 -2.40222275e-01 1.42653985e-02 -2.31941968e-01
-4.38501567e-01 -5.93612611e-01 3.52450341e-01 -1.97178096e-01
9.57925543e-02 1.59482121e-01 2.01488063e-01 2.43027866e-01
-8.79001990e-02 -2.86223948e-01 -1.17041171e+00 -3.96931648e-01
1.41227737e-01 9.12551582e-01 4.47220325e-01 -2.08761603e-01
3.43329906e-01 8.57922554e-01 -1.61868882e+00 4.10444170e-01
4.00756449e-01 7.36661255e-01 5.26953280e-01 9.38313246e-01
1.80891886e-01 6.33475780e-01 -5.53601742e-01 -2.98808992e-01
4.09129024e-01 -1.41261294e-02 -9.05708134e-01 1.78059444e-01
5.89385748e-01 -9.37695026e-01 -7.29958594e-01 8.22696745e-01
4.80071127e-01 4.37456891e-02 5.43682754e-01 -1.40560591e+00
-9.47849035e-01 3.65479946e-01 -5.86531460e-01 -3.34814459e-01
9.41826463e-01 7.42235601e-01 6.59198523e-01 -1.02894008e+00
1.02589166e+00 1.70186806e+00 8.76188278e-01 6.92212462e-01
-1.43329215e+00 -3.36810112e-01 2.69102544e-01 1.57733992e-01
-1.19957519e+00 -1.87316254e-01 7.55591452e-01 -7.70891368e-01
1.30669963e+00 2.54271895e-01 9.48900640e-01 1.52167273e+00
3.07446778e-01 1.16311336e+00 6.85700119e-01 -3.67480725e-01
-1.11726455e-01 -5.38092911e-01 1.32955804e-01 4.78077263e-01
2.15516333e-02 3.26985568e-01 -9.27333117e-01 2.99587902e-02
1.46998584e+00 3.94537508e-01 -4.58396614e-01 -8.42610657e-01
-1.62515640e+00 7.86317214e-02 4.14404958e-01 -3.11813354e-02
-6.58989012e-01 5.20009816e-01 2.76964046e-02 -1.46941766e-01
3.03093940e-01 -5.21916337e-02 -4.66070294e-01 -4.72081095e-01
-6.40758634e-01 5.97007096e-01 6.13450468e-01 1.72705412e+00
3.63939881e-01 -3.60923856e-01 -4.05021995e-01 1.72496408e-01
3.41078520e-01 6.26894295e-01 -3.85421067e-01 -1.44124866e+00
4.84164774e-01 3.69733393e-01 6.59110606e-01 -7.30114937e-01
-2.25330904e-01 1.84548482e-01 -4.89749700e-01 7.23978162e-01
5.01598418e-01 4.73336548e-01 -1.02398527e+00 1.32314563e+00
3.33339542e-01 6.42821938e-02 -3.61416966e-01 1.26345611e+00
4.98963058e-01 2.83652902e-01 -1.42170966e-01 9.98878106e-02
8.67116690e-01 -9.72014844e-01 -1.04151809e+00 1.40687093e-01
1.39676094e-01 -9.66981947e-01 1.26395082e+00 6.36353433e-01
-1.47607732e+00 -6.42308116e-01 -6.87157214e-01 -5.78422129e-01
-1.36920720e-01 1.50348321e-01 6.85592949e-01 -1.18683621e-01
-8.43586087e-01 1.04938960e+00 -1.44014275e+00 -3.40037078e-01
2.64715433e-01 4.74115491e-01 -4.06510055e-01 -8.22425038e-02
-2.58724779e-01 1.14980185e+00 6.51975200e-02 5.08955121e-01
-1.04529941e+00 -8.85988832e-01 -7.56264746e-01 -2.34950632e-01
4.40820456e-01 -7.65731990e-01 1.60440898e+00 -2.11138949e-01
-1.60867143e+00 7.00109422e-01 -3.29337686e-01 3.04625146e-02
8.54691565e-01 -9.06163990e-01 2.45703235e-01 7.55287632e-02
2.54307059e-03 7.38317847e-01 8.67860019e-01 -1.91542196e+00
-2.85337567e-01 -5.83250999e-01 -1.59817301e-02 6.62451703e-03
3.96427393e-01 -4.62152474e-02 -5.59225082e-01 -8.09048295e-01
3.86833280e-01 -9.72859383e-01 1.35550231e-01 6.40306771e-01
-7.09178150e-01 -4.35949653e-01 1.33077073e+00 -9.30914521e-01
5.73947966e-01 -1.77940309e+00 8.55905652e-01 9.95085239e-02
1.38199016e-01 4.54986915e-02 5.24039008e-02 3.41527551e-01
1.25253737e-01 -4.34983850e-01 2.28583664e-02 -8.36657703e-01
2.74156570e-01 2.40484074e-01 -5.45302272e-01 5.62021375e-01
9.37962756e-02 1.12872207e+00 -1.20850444e+00 -2.20302373e-01
9.09307599e-01 1.03558171e+00 -5.27534723e-01 6.93940997e-01
-6.96521103e-01 9.29833889e-01 -1.10112764e-01 9.69637454e-01
6.77494228e-01 -1.54932573e-01 -1.34397224e-01 -6.17762148e-01
-1.38579264e-01 1.94587380e-01 -1.26040876e+00 2.43104362e+00
-1.06080525e-01 3.51508677e-01 9.39304382e-02 -2.83624917e-01
5.69280446e-01 4.72330868e-01 7.90917635e-01 -7.95658454e-02
2.38562703e-01 1.06239498e-01 -5.15122056e-01 -4.71692890e-01
3.72646660e-01 3.29489201e-01 4.01731789e-01 4.50994343e-01
2.95310199e-01 -5.56553304e-01 -2.90527642e-01 1.88706592e-01
1.00986266e+00 1.15845752e+00 -2.47321531e-01 2.65460998e-01
-2.66024053e-01 1.02863036e-01 8.34026337e-02 6.34657502e-01
-8.92005488e-02 1.18703854e+00 -5.76672852e-02 -4.07261878e-01
-1.23054326e+00 -1.34892511e+00 1.46825314e-01 7.75890231e-01
6.55618668e-01 -3.06974143e-01 -6.40868127e-01 -4.63510126e-01
4.15137202e-01 5.55563748e-01 -6.15088999e-01 3.80418338e-02
-7.46865451e-01 2.78628141e-01 5.54041490e-02 1.15900779e+00
7.67253265e-02 -1.28513563e+00 -9.23805594e-01 2.66901195e-01
-1.30114153e-01 -1.06806743e+00 -8.10057282e-01 5.89207225e-02
-9.97053564e-01 -1.38569820e+00 -1.06794572e+00 -6.55911565e-01
7.44850755e-01 4.19584900e-01 1.12844241e+00 1.01001775e-02
-6.34101272e-01 1.12255144e+00 -4.01998073e-01 -2.15670049e-01
-3.34033370e-01 -4.42952842e-01 3.16962570e-01 -5.30624032e-01
3.38508099e-01 -5.67295611e-01 -6.41401708e-01 3.91094744e-01
-4.37529981e-01 2.45952874e-01 2.86892772e-01 5.73640704e-01
6.20082438e-01 -6.38520837e-01 -3.51730227e-01 -2.42831595e-02
2.81276196e-01 1.47166833e-01 -5.50933659e-01 4.25808042e-01
-3.38486470e-02 8.10948610e-02 -3.72581482e-02 -7.79554129e-01
-1.31930995e+00 6.28353596e-01 1.96111009e-01 -1.20749354e+00
-2.80338079e-01 -3.01417649e-01 -1.00235879e-01 -2.14392785e-02
5.29786170e-01 -7.70734027e-02 -5.52382786e-03 -9.25562978e-01
5.74732900e-01 3.72944176e-01 1.08550310e+00 -8.54416370e-01
6.36367500e-01 6.72569692e-01 -7.09018111e-02 -3.68623286e-01
-6.09904289e-01 -5.83310068e-01 -1.51556790e+00 -4.50618684e-01
9.29558218e-01 -5.42682111e-01 -1.49945188e+00 8.16557109e-01
-1.96905541e+00 -6.32411480e-01 -5.17331421e-01 5.13776422e-01
-1.06460726e+00 4.30232435e-01 -7.77873755e-01 -1.03005636e+00
1.85762253e-02 -1.25931227e+00 1.91341019e+00 -1.76162973e-01
-4.75650758e-01 -4.52119023e-01 -1.21476911e-01 5.74291013e-02
-2.45539784e-01 3.59279960e-01 4.71416205e-01 3.76171887e-01
-1.03701270e+00 -8.39784667e-02 -1.14214309e-01 1.30167916e-01
4.47358459e-01 1.84808418e-01 -8.33426237e-01 -4.95056123e-01
-2.48562217e-01 -2.82307327e-01 5.58910370e-01 7.05225170e-01
1.34337091e+00 -1.71859860e-01 -6.12404346e-01 4.69277203e-01
9.77964759e-01 4.86778505e-02 5.86131692e-01 -2.13989522e-02
1.23729467e+00 5.65397263e-01 7.63107121e-01 6.05062962e-01
2.98527390e-01 1.11795473e+00 8.26301694e-01 2.64622420e-01
-4.87922668e-01 -4.58541930e-01 3.39847475e-01 5.37068129e-01
-7.92388856e-01 -2.16665789e-02 -7.88748145e-01 5.13727248e-01
-2.04305267e+00 -7.42776871e-01 -2.57137567e-01 2.19245791e+00
9.88754094e-01 1.36927096e-02 1.54951796e-01 2.37342805e-01
5.98447740e-01 -3.23837101e-01 -8.10924292e-01 1.85585260e-01
5.74892648e-02 2.38340199e-01 3.71288061e-01 7.47596800e-01
-6.97285950e-01 1.09747696e+00 6.53378725e+00 2.83932406e-02
-6.75378978e-01 4.27656323e-02 -5.35112143e-01 -1.81163862e-01
3.21726575e-02 -8.15350935e-03 -7.10988641e-01 1.13049053e-01
-7.13044405e-02 3.60935360e-01 7.66069472e-01 7.19568074e-01
1.07868314e-01 -1.03835188e-01 -1.92908406e+00 1.25818896e+00
5.67008406e-02 -1.14021814e+00 8.12636316e-03 -1.50901183e-01
5.94826102e-01 -3.97734910e-01 -6.27328008e-02 -2.22835913e-01
5.60497582e-01 -9.85340238e-01 1.38581717e+00 9.71392512e-01
7.81881094e-01 -1.78729475e-01 2.69788474e-01 2.82934189e-01
-1.29801035e+00 2.09866449e-01 9.29770526e-03 1.31943775e-02
5.95196843e-01 1.94638655e-01 -5.05092978e-01 3.18168193e-01
1.27517712e+00 9.64630902e-01 -2.88757890e-01 8.24265480e-01
-2.28630453e-01 -1.18144289e-01 -4.79335934e-01 5.28794587e-01
-2.59237528e-01 2.01870903e-01 8.47913861e-01 9.18610573e-01
2.08169863e-01 4.59625304e-01 2.22519994e-01 1.26341259e+00
2.51099855e-01 -8.65226150e-01 -6.41873121e-01 9.47029516e-02
3.96706045e-01 5.30566931e-01 -4.00642157e-01 -2.98237979e-01
-1.02006741e-01 1.63196862e+00 1.22053012e-01 5.13820231e-01
-7.07887590e-01 -1.89401463e-01 7.35436857e-01 2.76694328e-01
3.26976538e-01 -8.90555680e-01 -5.50384164e-01 -1.17292809e+00
5.45428872e-01 -4.22787309e-01 -1.97290763e-01 -1.39068425e+00
-1.34862328e+00 1.18896224e-01 2.87297130e-01 -1.30393028e+00
-1.36367992e-01 -9.70685303e-01 -4.15099747e-02 8.92360270e-01
-8.81966889e-01 -1.64936233e+00 -8.22563350e-01 8.47954214e-01
7.70876467e-01 4.44714814e-01 7.99437821e-01 -1.90659434e-01
2.07661569e-01 2.51476318e-01 -3.82482976e-01 -7.06268102e-02
7.62564242e-01 -1.37213993e+00 6.03677154e-01 4.70775932e-01
9.66246054e-03 8.08439493e-01 8.19693446e-01 -1.10633254e+00
-1.96282029e+00 -6.27248168e-01 3.73827904e-01 -1.26046181e+00
7.82887861e-02 -4.77611363e-01 -9.10887122e-01 1.40545475e+00
-5.47420532e-02 1.81418702e-01 -1.87879741e-01 -3.41471247e-02
-4.05416518e-01 3.73533815e-01 -1.16360354e+00 6.94104731e-01
1.78574896e+00 -6.95852399e-01 -9.20631826e-01 5.88620961e-01
7.38742411e-01 -1.22906399e+00 -9.18428481e-01 4.95701700e-01
1.02987576e+00 -9.60797548e-01 1.41426682e+00 -6.88390434e-01
3.00475448e-01 -3.92012626e-01 -1.38859481e-01 -1.15369034e+00
-3.12673122e-01 -8.29520583e-01 -9.12440419e-01 7.59314597e-01
-5.31153977e-01 1.44550085e-01 8.18174481e-01 7.14350045e-01
-2.63716459e-01 -4.77204174e-01 -7.13443160e-01 -9.72118795e-01
-2.94038147e-01 -6.20557308e-01 5.96149385e-01 4.95332420e-01
-2.59138141e-02 -3.26523095e-01 -4.62519646e-01 4.11236405e-01
1.11863923e+00 5.30468822e-02 1.08786654e+00 -1.41274393e+00
-2.16875196e-01 -3.16652507e-01 -3.65497500e-01 -1.62041795e+00
1.70855448e-01 -3.79872710e-01 7.41579235e-01 -1.57973742e+00
1.24636836e-01 -3.60498786e-01 2.46798262e-01 6.85268760e-01
7.95887187e-02 8.95080268e-02 3.85508806e-01 5.02467632e-01
-4.01830107e-01 5.56813955e-01 1.62114501e+00 -2.42958456e-01
-4.17677760e-01 -2.86585763e-02 2.84723669e-01 8.08817267e-01
1.30222857e-01 -3.32534939e-01 2.01625116e-02 -9.00189579e-01
-2.72991925e-01 7.98468366e-02 1.24313164e+00 -7.91433990e-01
2.76419461e-01 -3.23414207e-01 5.92591286e-01 -1.16278255e+00
8.33230078e-01 -1.15370893e+00 4.46176797e-01 3.34329784e-01
-2.32473627e-01 9.78989080e-02 2.61700958e-01 6.45334244e-01
2.38956511e-01 1.56198278e-01 3.13005149e-01 -2.34734058e-01
-6.38835907e-01 4.99121159e-01 2.22767331e-02 -4.49588567e-01
1.00153422e+00 -4.15065706e-01 -7.71128163e-02 -3.02876085e-01
-1.08420777e+00 1.47752957e-02 7.10049272e-01 9.37756240e-01
9.20500398e-01 -1.45327711e+00 -3.33137393e-01 2.94125050e-01
3.14423814e-02 7.08640933e-01 1.75054476e-01 6.29724920e-01
-4.63080227e-01 4.10815597e-01 -2.56002426e-01 -1.30553854e+00
-1.39941275e+00 5.47169149e-01 8.71134698e-02 5.02482951e-01
-1.27985251e+00 8.88946772e-01 -7.06732199e-02 -5.18849671e-01
8.98001671e-01 -9.63814139e-01 5.97543716e-01 -6.56830370e-01
2.85346448e-01 6.73236191e-01 2.38710921e-02 -6.90280437e-01
-3.57136071e-01 9.95858788e-01 2.35095635e-01 -2.33519539e-01
1.31601405e+00 1.31467476e-01 -1.67173017e-02 5.93894839e-01
7.60386348e-01 -4.32156980e-01 -2.21979117e+00 -1.24939322e-01
-2.05589861e-01 -9.87596512e-01 -2.53354609e-01 -9.97372091e-01
-8.43716025e-01 1.18224585e+00 5.89688182e-01 -2.86598831e-01
6.92035019e-01 3.82403016e-01 8.57985377e-01 5.01538455e-01
9.48735774e-01 -9.23578858e-01 4.19945210e-01 5.27203083e-01
1.63808692e+00 -1.16619289e+00 -8.29628259e-02 -7.13223994e-01
-4.21572387e-01 1.18110800e+00 7.73026764e-01 -1.24590941e-01
5.50639629e-01 6.09285176e-01 -2.80129194e-01 -4.14687842e-01
-2.16245562e-01 9.34607089e-02 3.26074004e-01 9.53429937e-01
9.95867103e-02 2.40619872e-02 5.64204931e-01 3.51804256e-01
-4.51821126e-02 4.89300281e-01 6.67432770e-02 1.62802982e+00
9.77735594e-02 -1.04913342e+00 -5.62662780e-01 -1.60111785e-02
9.08993334e-02 2.84930140e-01 -3.44380707e-01 6.49967849e-01
7.19619319e-02 5.84013164e-01 1.68950170e-01 -4.87223446e-01
8.54756355e-01 -1.00688510e-01 1.53874576e+00 -6.91355407e-01
-4.54601198e-01 3.98052707e-02 -6.64306343e-01 -1.16394937e+00
-6.63986921e-01 -6.77285433e-01 -1.44076240e+00 -5.30255556e-01
-4.74307746e-01 -5.18737078e-01 5.53604960e-01 9.65688050e-01
4.93155807e-01 4.81892288e-01 -1.30731449e-01 -2.05940843e+00
-6.69802547e-01 -1.02892447e+00 -5.53896129e-01 7.16177225e-01
6.60008550e-01 -1.37781775e+00 -3.26816551e-02 6.17330313e-01] | [6.371459484100342, -1.0095680952072144] |
a2b0fe26-09dd-41ab-9be1-dbd81eb44e00 | leveraging-large-scale-uncurated-data-for | 1905.01278 | null | https://arxiv.org/abs/1905.01278v3 | https://arxiv.org/pdf/1905.01278v3.pdf | Unsupervised Pre-Training of Image Features on Non-Curated Data | Pre-training general-purpose visual features with convolutional neural networks without relying on annotations is a challenging and important task. Most recent efforts in unsupervised feature learning have focused on either small or highly curated datasets like ImageNet, whereas using uncurated raw datasets was found to decrease the feature quality when evaluated on a transfer task. Our goal is to bridge the performance gap between unsupervised methods trained on curated data, which are costly to obtain, and massive raw datasets that are easily available. To that effect, we propose a new unsupervised approach which leverages self-supervision and clustering to capture complementary statistics from large-scale data. We validate our approach on 96 million images from YFCC100M, achieving state-of-the-art results among unsupervised methods on standard benchmarks, which confirms the potential of unsupervised learning when only uncurated data are available. We also show that pre-training a supervised VGG-16 with our method achieves 74.9% top-1 classification accuracy on the validation set of ImageNet, which is an improvement of +0.8% over the same network trained from scratch. Our code is available at https://github.com/facebookresearch/DeeperCluster. | ['Julien Mairal', 'Armand Joulin', 'Piotr Bojanowski', 'Mathilde Caron'] | 2019-05-03 | unsupervised-pre-training-of-image-features | http://openaccess.thecvf.com/content_ICCV_2019/html/Caron_Unsupervised_Pre-Training_of_Image_Features_on_Non-Curated_Data_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Caron_Unsupervised_Pre-Training_of_Image_Features_on_Non-Curated_Data_ICCV_2019_paper.pdf | iccv-2019-10 | ['self-supervised-image-classification'] | ['computer-vision'] | [-2.14022491e-02 5.47442324e-02 -2.17708394e-01 -4.24657166e-01
-7.44679511e-01 -4.63110894e-01 6.17146134e-01 1.44082159e-01
-7.85815001e-01 6.01023138e-01 9.50076655e-02 1.95929427e-02
-4.54838946e-02 -6.69628799e-01 -8.71001720e-01 -6.09520435e-01
-1.55519158e-01 2.43144855e-01 1.39844447e-01 -2.63276584e-02
-8.02650023e-03 2.74987280e-01 -1.74160635e+00 2.66677678e-01
5.87261856e-01 1.12380052e+00 1.19943418e-01 5.95166445e-01
6.77824989e-02 6.01796031e-01 -3.04858744e-01 -1.03358671e-01
3.78541827e-01 -1.22138904e-02 -9.08416569e-01 3.11948389e-01
7.51149893e-01 -1.85952634e-01 -3.02987456e-01 9.29777622e-01
2.40028158e-01 -3.18523049e-02 7.29781449e-01 -1.35716975e+00
-6.53013647e-01 5.42995095e-01 -4.21270847e-01 3.97334397e-02
-4.37173694e-01 3.16043079e-01 1.08012760e+00 -7.70972371e-01
8.29556465e-01 6.56853318e-01 7.85345078e-01 4.47126389e-01
-1.34782934e+00 -6.84093356e-01 -1.68589294e-01 4.22836319e-02
-1.43278742e+00 -5.40811479e-01 6.29515827e-01 -6.96180105e-01
1.01911855e+00 -4.31434065e-02 4.72259074e-01 1.10221398e+00
-3.27663541e-01 5.58460951e-01 1.14960504e+00 -4.28512424e-01
1.22684091e-01 2.22512439e-01 2.05779046e-01 7.38635182e-01
3.76889497e-01 1.14090942e-01 -4.23044443e-01 2.24235594e-01
5.89171171e-01 2.48024106e-01 -1.31393686e-01 -5.97802043e-01
-1.20985270e+00 9.57198620e-01 8.97449970e-01 4.39598113e-01
-2.84959733e-01 3.41780901e-01 4.77561712e-01 4.39521760e-01
6.73304379e-01 6.16567254e-01 -8.16406548e-01 -1.94492787e-01
-1.08929777e+00 -2.49091238e-01 5.62489629e-01 9.45667088e-01
1.21881819e+00 -4.75269789e-03 1.41354337e-01 8.20695162e-01
1.34007841e-01 4.25733984e-01 5.00183821e-01 -8.91065955e-01
2.04901516e-01 8.51966381e-01 -2.47404471e-01 -8.42076719e-01
-4.96687502e-01 -6.53594494e-01 -9.30574954e-01 4.36617762e-01
7.22114444e-01 -2.21319154e-01 -1.31178212e+00 1.55573165e+00
6.40759766e-02 -1.31956428e-01 -7.87701011e-02 7.62176991e-01
1.06417394e+00 2.35427514e-01 -1.88570525e-02 2.38529652e-01
9.37316656e-01 -1.18211126e+00 -1.79639250e-01 5.68774678e-02
9.27637279e-01 -5.43746650e-01 1.16634357e+00 3.32707196e-01
-4.67930287e-01 -5.84900022e-01 -1.05421400e+00 -4.18580789e-03
-7.90695786e-01 3.35933656e-01 7.31022418e-01 4.13229108e-01
-1.29148293e+00 7.57734060e-01 -8.76099944e-01 -6.88799858e-01
1.09777057e+00 4.92446095e-01 -9.09113169e-01 -9.25448388e-02
-5.66871881e-01 5.36277473e-01 5.39690614e-01 -1.46226749e-01
-1.05170691e+00 -8.46565545e-01 -6.78898036e-01 3.72287864e-03
4.05781865e-01 -2.70720273e-01 9.02166128e-01 -1.22956586e+00
-1.08274627e+00 9.49076414e-01 2.99632460e-01 -5.97797275e-01
4.99783307e-01 -1.80686697e-01 -1.64313301e-01 3.04403156e-01
1.39966562e-01 1.12280095e+00 7.45744407e-01 -1.30452120e+00
-4.89654839e-01 -1.60321012e-01 -1.80172473e-02 -2.42838144e-01
-9.08140838e-01 -2.29536742e-01 -5.12521386e-01 -4.12589371e-01
-2.60390371e-01 -9.60427225e-01 -3.17616522e-01 6.84466958e-02
-6.06426120e-01 -1.41362801e-01 8.58808994e-01 -3.34584832e-01
7.41881847e-01 -2.17805886e+00 -1.30246550e-01 3.60769361e-01
5.28995395e-01 3.72351378e-01 -2.76305914e-01 3.01161021e-01
-2.54627198e-01 3.31363082e-01 -2.76712120e-01 -3.53288889e-01
-1.27971068e-01 4.93424535e-02 1.69867352e-01 4.72302943e-01
4.74106044e-01 1.16662991e+00 -8.21718156e-01 -4.45001960e-01
1.77600130e-01 4.92053658e-01 -5.18763065e-01 3.39823700e-02
-6.70862719e-02 5.29584348e-01 -1.97974905e-01 7.32590258e-01
3.62284034e-01 -8.44896555e-01 3.36846970e-02 -2.49406487e-01
-1.04731517e-02 -9.38404128e-02 -7.00819492e-01 1.93472672e+00
-3.32979888e-01 9.03669596e-01 -2.72677422e-01 -1.24533772e+00
7.75994241e-01 1.20663913e-02 5.16156673e-01 -6.32945836e-01
3.37163627e-01 1.72431022e-01 1.17765382e-01 -2.73667246e-01
3.09702396e-01 2.32934013e-01 2.02343017e-02 4.06420499e-01
8.03416014e-01 8.93223286e-02 4.27948833e-01 3.89736772e-01
1.37819016e+00 7.92120397e-03 4.22468260e-02 -4.74658906e-01
1.91776920e-02 2.72922784e-01 2.45507866e-01 7.99526811e-01
-1.40826508e-01 8.54553819e-01 5.53228199e-01 -5.10014176e-01
-1.24169552e+00 -7.31612563e-01 -2.77087510e-01 1.17675138e+00
-3.82460028e-01 -7.23268688e-01 -7.72253990e-01 -1.11779928e+00
1.74183428e-01 1.02161191e-01 -1.14032161e+00 5.15472814e-02
-5.74137224e-03 -7.15643644e-01 4.75493491e-01 6.93398356e-01
5.26570737e-01 -1.06861711e+00 -4.52376276e-01 -5.63648306e-02
2.82394618e-01 -1.15975237e+00 3.29387598e-02 5.91267347e-01
-7.21919537e-01 -1.33621037e+00 -6.58485532e-01 -7.11710393e-01
1.04860675e+00 2.35799447e-01 1.23963392e+00 4.52383488e-01
-5.31873822e-01 4.47312266e-01 -6.25798881e-01 -3.97237033e-01
-1.05295181e-02 5.57152867e-01 -4.24507223e-02 3.24568190e-02
4.21975702e-01 -7.06265330e-01 -6.19733512e-01 1.71139300e-01
-9.34242547e-01 3.99456918e-02 6.98732138e-01 9.50938165e-01
5.31164467e-01 -1.23145983e-01 5.82446039e-01 -1.06995034e+00
4.68102209e-02 -5.07005930e-01 -4.77543086e-01 -1.04027599e-01
-7.43076265e-01 4.87712249e-02 6.97945654e-01 -2.45641604e-01
-5.26535213e-01 3.52436811e-01 5.82065284e-02 -5.20187676e-01
-5.52269936e-01 5.01095712e-01 1.42949790e-01 -2.72578984e-01
9.12227094e-01 9.44263861e-03 2.54619628e-01 -5.21589160e-01
4.87982392e-01 6.94703937e-01 4.85073388e-01 -4.32612419e-01
1.01122010e+00 6.64656341e-01 -9.48083624e-02 -7.86705852e-01
-1.02103043e+00 -7.26234496e-01 -1.01151133e+00 -1.41730472e-01
8.37944865e-01 -9.72296596e-01 -3.97042722e-01 4.79703277e-01
-4.94891793e-01 -8.66182506e-01 -4.20281470e-01 3.52059424e-01
-4.37260747e-01 1.38356939e-01 -3.41230273e-01 -1.77215263e-01
-3.48046988e-01 -9.53476369e-01 8.45572710e-01 2.64326334e-01
-3.91403139e-02 -9.49157953e-01 -4.14969325e-02 2.83198982e-01
7.57690430e-01 4.81850117e-01 5.26207805e-01 -9.24443305e-01
-4.95253772e-01 -2.11445376e-01 -5.09450018e-01 6.77484751e-01
3.11923891e-01 1.30034536e-01 -1.09389865e+00 -4.97937620e-01
-7.42935240e-01 -1.02484322e+00 1.35708189e+00 4.72607315e-02
1.39355993e+00 3.82578485e-02 -1.84177354e-01 8.32370102e-01
1.57112944e+00 -5.06028712e-01 4.79694992e-01 5.92100084e-01
9.42409337e-01 3.52616221e-01 4.66654837e-01 3.10777664e-01
2.46332780e-01 3.37358385e-01 5.58957279e-01 -4.06858712e-01
-3.48952472e-01 -3.78941558e-02 -1.88718680e-02 7.89187372e-01
-3.49253654e-01 1.04480788e-01 -1.14680088e+00 9.16593254e-01
-1.90130520e+00 -6.01272106e-01 -1.64283246e-01 2.04595375e+00
9.27883625e-01 2.64970571e-01 1.70110479e-01 -6.35190681e-03
4.85145122e-01 -9.33453366e-02 -5.17693460e-01 1.68916583e-01
-1.71027139e-01 4.91518348e-01 7.60720611e-01 1.90657564e-02
-1.54557288e+00 1.15347350e+00 5.84216499e+00 8.01099002e-01
-1.31254029e+00 2.18314976e-01 8.51939559e-01 -2.68782735e-01
1.94474906e-01 6.59165485e-03 -5.83873272e-01 4.95704263e-01
1.08049464e+00 1.73065975e-01 2.21370593e-01 1.12545025e+00
-1.08180888e-01 -1.65767699e-01 -1.00534034e+00 9.78950799e-01
2.06000591e-03 -1.43889451e+00 -1.11430287e-01 2.78749049e-01
1.15591252e+00 6.91801846e-01 4.56886217e-02 3.40305150e-01
4.26498592e-01 -1.31462514e+00 3.66112053e-01 5.39737344e-01
1.07870615e+00 -8.08468997e-01 8.14164877e-01 1.00008599e-01
-9.85479414e-01 1.53837264e-01 -5.64335048e-01 5.42568751e-02
-3.99197489e-01 7.24180162e-01 -8.20533216e-01 3.90615940e-01
1.11880338e+00 1.10410774e+00 -1.24052799e+00 1.20963466e+00
-2.25981891e-01 8.92744780e-01 -4.25192595e-01 1.52746335e-01
4.07384098e-01 1.74533933e-01 4.30749841e-02 1.30141628e+00
8.78506899e-02 -5.17284632e-01 2.45066896e-01 6.33486331e-01
-4.93159175e-01 1.83295831e-01 -7.74406135e-01 -2.74283528e-01
5.40069789e-02 1.81660414e+00 -9.83491838e-01 -4.96506482e-01
-5.15059173e-01 7.17546463e-01 8.08457315e-01 2.81296700e-01
-6.54744148e-01 -5.69429278e-01 3.67486507e-01 5.05944751e-02
6.22926891e-01 -2.56048769e-01 -6.62871897e-02 -1.17816639e+00
-1.23112418e-01 -7.20379710e-01 3.03470761e-01 -5.59081495e-01
-1.42303824e+00 6.00605726e-01 -1.94610819e-01 -1.28987396e+00
-7.40043670e-02 -8.97695303e-01 -5.31946123e-01 3.62538129e-01
-1.65039062e+00 -1.29217231e+00 -7.93814421e-01 7.74279654e-01
2.74215460e-01 -4.02512044e-01 8.77288938e-01 2.31713146e-01
-4.60047930e-01 7.41902947e-01 4.49017018e-01 6.00955844e-01
9.04005647e-01 -1.34823489e+00 2.37777621e-01 6.77296281e-01
3.81576240e-01 5.32718301e-01 1.63340509e-01 -4.09044325e-01
-1.23533273e+00 -1.41483247e+00 4.15831417e-01 -4.97032672e-01
8.61877084e-01 -5.55839896e-01 -9.00863469e-01 6.74359918e-01
3.31590116e-01 6.59047365e-01 7.86112189e-01 2.94218510e-01
-6.18120372e-01 -1.54785648e-01 -9.41077352e-01 1.38528556e-01
1.08380520e+00 -4.61986929e-01 -2.49671072e-01 4.07348126e-01
6.08471215e-01 -3.55285630e-02 -1.14543343e+00 4.50635761e-01
4.11686629e-01 -8.32438767e-01 7.31413186e-01 -7.56625772e-01
6.92338228e-01 -2.91618496e-01 -1.97380409e-01 -1.41782045e+00
-2.76317954e-01 -3.70615393e-01 5.81317283e-02 1.34917533e+00
6.86478436e-01 -5.92236698e-01 9.43604529e-01 2.89947867e-01
-8.70745033e-02 -6.92500412e-01 -4.90292519e-01 -9.93742764e-01
1.58100322e-01 -4.19681370e-01 2.18339115e-01 1.19829082e+00
-1.93066001e-01 2.09579185e-01 -2.45413348e-01 -2.38178715e-01
6.51377976e-01 -8.56273621e-03 1.11156213e+00 -1.37445486e+00
-1.10779278e-01 -2.68650562e-01 -7.27184355e-01 -4.69042510e-01
1.86453477e-01 -1.23292553e+00 2.96089333e-03 -1.43176305e+00
5.33603013e-01 -5.74679375e-01 -6.36058629e-01 1.06008947e+00
5.31243421e-02 9.32296932e-01 1.29559293e-01 2.40340143e-01
-1.03485548e+00 5.22075415e-01 9.32989180e-01 -2.38337204e-01
-7.22424244e-04 -5.71479201e-01 -8.07085752e-01 7.85062611e-01
1.04305875e+00 -6.19917512e-01 -1.98662460e-01 -3.11897099e-01
-2.85197422e-02 -7.92958736e-01 5.72400093e-01 -1.34513140e+00
1.88333884e-01 1.80767059e-01 7.25811660e-01 -3.81350547e-01
1.81402918e-02 -7.71632016e-01 -1.97633997e-01 1.36940017e-01
-2.69179940e-01 -2.42701694e-01 2.75798529e-01 5.46738982e-01
-2.85806209e-01 -8.29787180e-02 6.88040078e-01 -2.28422120e-01
-9.47655320e-01 4.75757629e-01 -3.54075804e-02 1.19842194e-01
8.24172616e-01 6.13559075e-02 -6.73176408e-01 -2.58919001e-01
-7.09672749e-01 8.33720639e-02 7.71056831e-01 3.87656569e-01
3.91837060e-01 -1.26786256e+00 -6.54535830e-01 8.16993713e-02
4.75630909e-01 5.25207445e-02 -3.63446213e-02 9.30806220e-01
-5.29846907e-01 3.05018067e-01 -4.79551464e-01 -8.92337322e-01
-1.12575293e+00 4.17866856e-01 4.47649509e-02 -2.39105165e-01
-5.57443321e-01 7.60482609e-01 -2.39467584e-02 -5.83891094e-01
1.73887178e-01 -5.56565411e-02 -1.40582994e-01 1.86479419e-01
3.33310634e-01 -6.66061342e-02 2.34219328e-01 -6.61171138e-01
-2.97728688e-01 5.06814420e-01 -3.11525911e-01 2.44181111e-01
1.86577213e+00 2.24804685e-01 -7.68601596e-02 3.96578610e-01
1.72605026e+00 -2.12053046e-01 -1.47840047e+00 -3.58697027e-01
1.01513915e-01 -4.74010527e-01 1.95406005e-01 -7.51207292e-01
-1.50336385e+00 8.19702029e-01 6.59607053e-01 1.94387197e-01
9.58692551e-01 2.50079215e-01 4.21498299e-01 8.30203593e-01
3.42605710e-01 -1.20781779e+00 4.53270525e-01 5.60477197e-01
7.38837183e-01 -1.75380421e+00 5.34110852e-02 -1.55494571e-01
-4.76195186e-01 1.12802207e+00 7.50639677e-01 -4.67331380e-01
8.85253370e-01 7.78340101e-02 1.15605898e-01 -4.36627954e-01
-5.98177314e-01 -7.49125063e-01 3.51346523e-01 6.30740106e-01
3.51344734e-01 -3.38378698e-02 9.30658057e-02 4.13879842e-01
-2.86538154e-01 -2.86629666e-02 3.94500405e-01 1.08544278e+00
-3.11475426e-01 -9.89951909e-01 5.92095554e-02 8.30221057e-01
-5.23304939e-01 -3.01575929e-01 -4.55966741e-01 1.10664928e+00
7.08036944e-02 8.30191970e-01 1.56330556e-01 -5.57047188e-01
-3.60320657e-02 1.82463616e-01 2.77000189e-01 -7.35757291e-01
-4.70691293e-01 -1.08958133e-01 1.32538185e-01 -6.60622299e-01
-6.73414290e-01 -4.40698773e-01 -1.00660741e+00 -2.90421620e-02
-3.03551376e-01 -8.09905604e-02 7.40569055e-01 8.52607131e-01
5.12825727e-01 4.67829466e-01 5.90563834e-01 -9.62016582e-01
-1.98102266e-01 -1.15067410e+00 -4.38403010e-01 7.59146631e-01
2.21782759e-01 -6.67968571e-01 -4.57365006e-01 2.31923819e-01] | [9.527772903442383, 2.4386377334594727] |
4810301e-d63f-45c8-af51-d80e012f693f | wordalchemy-a-transformer-based-reverse | 2204.10181 | null | https://arxiv.org/abs/2204.10181v1 | https://arxiv.org/pdf/2204.10181v1.pdf | WordAlchemy: A transformer-based Reverse Dictionary | A reverse dictionary takes a target word's description as input and returns the words that fit the description. Reverse Dictionaries are useful for new language learners, anomia patients, and for solving common tip-of-the-tongue problems (lethologica). Currently, there does not exist any Reverse Dictionary provider with support for any Indian Language. We present a novel open-source cross-lingual reverse dictionary system with support for Indian languages. In this paper, we propose a transformer-based deep learning approach to tackle the limitations faced by the existing systems using the mT5 model. This architecture uses the Translation Language Modeling (TLM) technique, rather than the conventional BERT's Masked Language Modeling (MLM) technique. | ['Pranav Sadavarte', 'Kanhaiya Madaswar', 'Harshal Patil', 'Dr. Sunil B. Mane'] | 2022-04-16 | null | null | null | null | ['reverse-dictionary'] | ['natural-language-processing'] | [-3.74447525e-01 -1.35365427e-02 -7.44138360e-01 -2.42917418e-01
-7.64550507e-01 -6.05565906e-01 5.68836808e-01 -5.16538210e-02
-5.34832001e-01 5.51096678e-01 4.08732027e-01 -9.06165600e-01
3.33372742e-01 -5.69752455e-01 -3.54864836e-01 -2.15659067e-01
4.95791644e-01 9.44537461e-01 -8.75895172e-02 -7.84552455e-01
1.54901385e-01 1.26446515e-01 -9.40270126e-01 6.88932121e-01
1.05224931e+00 3.38157237e-01 6.29804492e-01 -3.73620242e-02
-7.49586284e-01 1.00659776e+00 -4.47174519e-01 -7.05479801e-01
1.13479398e-01 -4.27864283e-01 -8.25644314e-01 -4.03010368e-01
2.46836185e-01 -2.49096990e-01 -3.36800754e-01 1.10814464e+00
5.24007559e-01 -2.03607157e-01 1.40414044e-01 -7.93560565e-01
-1.53624487e+00 1.06798971e+00 -3.82686019e-01 4.54488873e-01
4.59187448e-01 -2.91967601e-01 7.35085785e-01 -1.06234479e+00
7.58817554e-01 1.25703275e+00 7.13834345e-01 7.62019098e-01
-9.33569789e-01 -9.67021585e-01 1.77542359e-01 4.69934553e-01
-1.50643981e+00 -5.11965871e-01 6.24596417e-01 -2.95297086e-01
1.54843068e+00 -1.00810125e-01 1.00995898e+00 1.42786765e+00
5.16188741e-01 8.55225325e-01 1.45788777e+00 -8.08224618e-01
-1.65638342e-01 3.97143930e-01 -1.32957995e-01 8.42028141e-01
-1.40383065e-01 6.64334893e-02 -9.53142643e-01 2.31064916e-01
6.18074894e-01 -3.66932184e-01 2.00457767e-01 8.82912353e-02
-1.08949101e+00 9.10993099e-01 8.53729993e-02 6.37064815e-01
-1.12564802e-01 -1.52177364e-01 4.25958574e-01 8.52788806e-01
5.24385154e-01 1.15559109e-01 -6.73610926e-01 -1.91979498e-01
-9.09088314e-01 -9.09039304e-02 4.90190536e-01 1.24890161e+00
2.13920027e-01 4.20664340e-01 3.76198173e-01 9.08203721e-01
8.03317606e-01 4.53123003e-01 1.38359320e+00 -3.98878723e-01
3.85458589e-01 5.70371985e-01 -5.64013481e-01 -4.12011951e-01
-3.13904196e-01 -6.72036171e-01 -3.03287566e-01 -2.37139627e-01
1.62872359e-01 2.33703360e-01 -1.05182219e+00 1.79146659e+00
1.07272081e-01 -2.27233201e-01 3.77392381e-01 5.91908276e-01
1.06601191e+00 5.35470963e-01 8.46033990e-02 -6.16472475e-02
1.48728907e+00 -1.12017930e+00 -8.94085765e-01 -6.01778209e-01
8.32405567e-01 -1.11025977e+00 1.31914222e+00 3.52530271e-01
-1.27641010e+00 -2.26986110e-01 -8.99719417e-01 -6.60855770e-01
-8.21756184e-01 3.88283581e-02 6.72468662e-01 8.62635791e-01
-1.37889194e+00 -7.07157180e-02 -7.98447073e-01 -6.55572057e-01
-7.35548884e-02 3.75290036e-01 -5.12377977e-01 -2.47461051e-01
-1.34473789e+00 1.66299617e+00 4.37976390e-01 -3.13666791e-01
-7.48561561e-01 -4.94553626e-01 -9.46514547e-01 -3.36128294e-01
-3.04430872e-01 -6.19318306e-01 1.30977750e+00 -1.13679600e+00
-1.64240253e+00 1.47618914e+00 -3.23832512e-01 -3.79408747e-01
2.76104063e-01 -2.07924340e-02 -8.76174808e-01 -4.07587945e-01
3.22781295e-01 5.46423078e-01 6.02471292e-01 -8.28116536e-01
-5.42521119e-01 -4.23499763e-01 1.46473676e-01 3.43648911e-01
-1.42034233e-01 4.77267027e-01 -4.46877033e-01 -9.77066040e-01
1.72394171e-01 -8.18322361e-01 3.29790115e-02 -2.38725349e-01
-1.71889421e-02 -8.54900256e-02 4.56647784e-01 -1.15969253e+00
1.28297281e+00 -1.98132491e+00 2.52799124e-01 -7.32398331e-02
3.27231884e-02 3.61831218e-01 -2.60883093e-01 6.86294734e-01
-1.41831577e-01 -1.76978111e-02 -4.67090458e-02 -5.58892548e-01
1.60632078e-02 5.61956406e-01 -3.73235255e-01 3.16110939e-01
-3.73222321e-01 9.38113332e-01 -8.06443870e-01 -4.47117537e-01
2.22177282e-01 6.47460759e-01 -6.80296838e-01 -7.69248679e-02
-9.71630141e-02 4.82353479e-01 1.76195264e-01 8.47109318e-01
3.64021033e-01 2.84851700e-01 4.85112190e-01 1.75483972e-02
-4.41362560e-01 9.39538956e-01 -6.16927028e-01 2.05243587e+00
-9.29832995e-01 4.98093426e-01 -1.98800508e-02 -6.86220825e-01
9.60027397e-01 6.21559560e-01 2.68112898e-01 -1.17560196e+00
1.30387962e-01 9.11772847e-01 2.25712031e-01 -3.85250926e-01
4.61451381e-01 -7.51439631e-01 6.95638731e-02 5.05820990e-01
3.76434058e-01 2.23202676e-01 8.61429721e-02 3.56727056e-02
7.37622380e-01 3.32581192e-01 5.98136127e-01 -5.13437390e-01
4.53417897e-01 2.20493272e-01 4.91320819e-01 1.45307004e-01
2.86453357e-03 1.29413784e-01 -2.21778467e-01 -5.74515104e-01
-8.59839320e-01 -1.02778637e+00 -8.80802795e-02 1.26157498e+00
-3.15146983e-01 -7.26522326e-01 -5.59218884e-01 -3.30404460e-01
-2.95558512e-01 9.37119961e-01 -2.24241480e-01 -3.22521806e-01
-6.48519814e-01 -4.90918100e-01 7.58584321e-01 4.00802881e-01
3.49399090e-01 -1.13372636e+00 -1.99265063e-01 4.20593470e-01
-4.61805612e-01 -1.07513487e+00 -7.51748979e-01 3.69440734e-01
-6.38386667e-01 -5.25447249e-01 -4.75543231e-01 -1.54742908e+00
4.86511290e-01 3.39666307e-02 1.05891526e+00 -5.20255007e-02
1.59685209e-01 6.84934035e-02 -3.34581226e-01 -1.38888538e-01
-7.78675854e-01 3.05381000e-01 4.65719938e-01 -4.46104616e-01
8.93144846e-01 -6.54663384e-01 1.08126581e-01 -2.03883156e-01
-6.74941480e-01 2.96925426e-01 5.64416766e-01 5.87274671e-01
5.10201454e-01 -2.38319710e-01 5.62893689e-01 -7.10923016e-01
8.85994732e-01 -4.49968368e-01 -4.34033245e-01 3.42292339e-01
-8.18509936e-01 5.87175339e-02 7.23199725e-01 -6.43949568e-01
-7.15923607e-01 -1.08489603e-01 -6.91819847e-01 -2.59063542e-01
3.16737026e-01 8.84039938e-01 -1.24358788e-01 -1.86829388e-01
4.23389405e-01 6.45743251e-01 -1.67614400e-01 -8.96900356e-01
4.16924804e-01 9.35308874e-01 5.32606423e-01 -5.20030558e-01
5.30912995e-01 6.23030365e-02 -5.82670093e-01 -6.31653011e-01
-6.87135160e-01 -1.89246103e-01 -8.02222490e-01 2.62502786e-02
7.02016413e-01 -1.28929329e+00 -3.26271147e-01 2.69304991e-01
-1.19598496e+00 -2.47416139e-01 -2.06661627e-01 4.92788285e-01
-4.96481150e-01 9.84414443e-02 -8.42278898e-01 -3.20305288e-01
-4.77726072e-01 -1.34495878e+00 6.79656804e-01 -2.09685594e-01
-3.11105818e-01 -1.33821917e+00 2.53020138e-01 4.84885186e-01
8.83538783e-01 -4.07230258e-01 1.30179799e+00 -6.89339221e-01
-7.74777979e-02 1.55195951e-01 1.97659448e-01 1.02821454e-01
2.59669632e-01 -5.43274641e-01 -6.49745286e-01 -4.08062071e-01
1.44259796e-01 -6.63253441e-02 2.99110025e-01 6.94877952e-02
4.68093097e-01 -5.19367397e-01 1.59832686e-02 8.74980927e-01
1.28421533e+00 3.54406476e-01 3.28716964e-01 7.46810019e-01
6.93346083e-01 3.03709090e-01 7.79399136e-03 -5.92432264e-03
1.15089154e+00 7.98793614e-01 -1.06164820e-01 -9.56200957e-02
-5.96015513e-01 -7.26843894e-01 1.05563462e+00 1.78978515e+00
4.61119801e-01 5.37915081e-02 -1.12360299e+00 8.52583528e-01
-1.34358037e+00 -6.38740540e-01 -1.02596544e-01 1.89956272e+00
1.27746546e+00 -4.62010056e-02 -1.45924360e-01 -1.46030366e-01
4.89430159e-01 -2.43728951e-01 -3.91858906e-01 -9.70306933e-01
-2.76199967e-01 6.50751591e-01 3.74634117e-01 9.39332902e-01
-5.47021568e-01 1.70955300e+00 7.03016949e+00 6.46514475e-01
-1.45707607e+00 8.17366004e-01 -2.17066690e-01 7.69760758e-02
-7.58129120e-01 2.70926982e-01 -7.49348044e-01 3.85478228e-01
1.22154641e+00 -3.72810274e-01 9.28326726e-01 6.51653469e-01
1.18064612e-01 4.78727698e-01 -8.91071320e-01 1.20950150e+00
3.40820819e-01 -1.13343859e+00 3.45199674e-01 1.22611158e-01
3.97000611e-01 5.68145633e-01 3.57272267e-01 6.14725173e-01
4.37924266e-01 -9.18477714e-01 1.15578282e+00 -8.85528885e-03
1.08478320e+00 -5.03630579e-01 2.48100072e-01 3.63353789e-01
-1.05556130e+00 1.16922468e-01 -2.78610736e-01 -4.95705903e-02
1.50273129e-01 2.12591246e-01 -7.01316297e-01 7.78914467e-02
5.28999150e-01 7.02445745e-01 -5.61166942e-01 4.57065552e-01
-3.66542757e-01 5.13272583e-01 -1.80227503e-01 3.19221139e-01
1.79006070e-01 -2.91171849e-01 4.31234837e-01 1.15831614e+00
5.33109605e-01 -1.79784656e-01 2.92784810e-01 6.35214984e-01
-1.47725001e-01 6.45930767e-01 -6.31425083e-01 -2.15489835e-01
4.08654392e-01 7.73808122e-01 -4.25765723e-01 -3.56433779e-01
-8.16511929e-01 1.23064017e+00 3.61401349e-01 1.85469780e-02
-5.33274412e-01 1.40593484e-01 5.98626435e-01 2.49231145e-01
1.23270983e-02 -4.89371687e-01 -3.60935628e-01 -1.56869543e+00
-9.03088227e-02 -1.51918018e+00 3.69723171e-01 -7.95187712e-01
-1.29232812e+00 8.54523897e-01 -2.64128834e-01 -1.08110952e+00
-3.00424933e-01 -7.27612793e-01 -4.31409515e-02 1.11551118e+00
-1.69238591e+00 -1.78343701e+00 4.76673216e-01 9.70673621e-01
8.10685933e-01 -6.65485859e-01 1.32811522e+00 7.69829869e-01
-3.21360201e-01 8.08767676e-01 1.01385579e-01 8.56220573e-02
7.77770877e-01 -1.13510644e+00 6.33371115e-01 8.13019335e-01
2.97775418e-01 1.16299188e+00 5.09504318e-01 -7.49089599e-01
-1.58616281e+00 -8.73424530e-01 1.70537901e+00 -4.52661663e-01
8.01334858e-01 -3.09290737e-01 -6.97880864e-01 1.30791807e+00
7.45724559e-01 -3.66205752e-01 1.16011286e+00 1.99245531e-02
-5.30282140e-01 3.29499282e-02 -1.23344100e+00 7.06727684e-01
1.01284051e+00 -9.11742985e-01 -9.06958997e-01 4.94707853e-01
7.00480402e-01 -5.49498677e-01 -9.24756944e-01 -1.16953194e-01
6.49115026e-01 -3.71766120e-01 6.08489454e-01 -6.88619137e-01
2.27310926e-01 -2.66959101e-01 -5.34183562e-01 -1.59781480e+00
-3.14458191e-01 -5.79547405e-01 -1.73230618e-01 1.01520193e+00
6.00005865e-01 -7.54457772e-01 2.76197761e-01 2.62457371e-01
-3.95782202e-01 -4.47613716e-01 -1.26279879e+00 -7.36897707e-01
6.04202688e-01 -5.71057081e-01 6.71519160e-01 1.54075611e+00
3.24078411e-01 6.07263982e-01 -4.00912076e-01 -1.04854517e-01
2.60854781e-01 -3.53647843e-02 1.94534674e-01 -8.69568408e-01
-3.89664710e-01 -4.16060150e-01 -1.99480265e-01 -8.23038459e-01
4.45182353e-01 -1.79380691e+00 -4.77995038e-01 -1.78534484e+00
-8.18603709e-02 -4.89021957e-01 -2.11900488e-01 8.77469063e-01
4.60712612e-01 2.74797797e-01 2.41456106e-01 2.72478551e-01
-7.48519925e-03 4.50504750e-01 8.88562620e-01 -1.59297436e-01
-9.94111821e-02 -4.21141982e-01 -1.08856094e+00 6.66787565e-01
9.68888640e-01 -6.91234529e-01 -4.47083443e-01 -1.34668636e+00
4.85537589e-01 -3.54910716e-02 -3.64910722e-01 -8.03052008e-01
3.22650135e-01 -6.73214942e-02 3.72826084e-02 -3.57629985e-01
2.35312238e-01 -7.83372223e-01 2.44010314e-01 6.23738110e-01
-2.04676732e-01 1.06936324e+00 4.04434472e-01 -4.31639761e-01
-1.54817805e-01 -9.04809460e-02 7.94286549e-01 -5.27838767e-01
-7.02730417e-01 3.74410115e-02 -8.79480124e-01 1.32541731e-01
7.15548217e-01 9.39806551e-02 -2.93061882e-01 -2.20531285e-01
-7.49907434e-01 -9.32173878e-02 5.39773464e-01 9.61479425e-01
7.12094605e-01 -1.51487768e+00 -7.29591668e-01 6.22086823e-01
2.54517406e-01 -7.36847103e-01 -3.37154925e-01 8.15615058e-01
-5.77594638e-01 5.99277198e-01 -4.27088439e-01 2.64218599e-02
-9.38212454e-01 6.82091951e-01 5.96754193e-01 -1.60129458e-01
-8.11136127e-01 6.85652852e-01 -2.42842808e-01 -1.08779001e+00
-8.42874777e-03 -2.90604293e-01 -1.45474941e-01 -2.75127357e-03
5.17008543e-01 -1.32535532e-01 3.70939016e-01 -1.21509755e+00
-5.72594047e-01 4.52202618e-01 -3.06767881e-01 -5.47025323e-01
1.24154484e+00 -2.76207536e-01 -4.74082381e-01 6.28578842e-01
9.84642088e-01 4.48496014e-01 5.01281358e-02 -4.91595477e-01
7.94682205e-02 -1.49828851e-01 3.54667425e-01 -1.20509839e+00
-1.14380980e+00 8.78073573e-01 8.47356439e-01 -4.87062603e-01
9.93003070e-01 -1.71960309e-01 1.11884630e+00 1.11901738e-01
7.28323638e-01 -1.03841662e+00 -5.38665593e-01 9.85612929e-01
6.36702240e-01 -8.76704097e-01 -3.80909145e-01 1.10834070e-01
-5.97707212e-01 9.85296965e-01 4.67570782e-01 2.30386257e-01
7.21669436e-01 5.61665654e-01 6.56203568e-01 -2.32353076e-01
-8.76803219e-01 -2.70554364e-01 1.04489729e-01 6.87319577e-01
1.04287612e+00 2.44565338e-01 -7.89278150e-01 6.35916293e-01
-8.15723062e-01 -2.50775572e-02 4.61292535e-01 1.04388380e+00
-1.84093550e-01 -1.77980030e+00 -3.59436959e-01 1.48405164e-01
-6.50503516e-01 -8.63111258e-01 -5.44182003e-01 6.07852757e-01
4.52627867e-01 7.60023415e-01 -8.86297524e-02 -5.22442579e-01
2.34321028e-01 4.13726002e-01 7.14105964e-01 -8.74281943e-01
-8.94109726e-01 -4.38376367e-02 -5.15606888e-02 -3.36241126e-01
-2.05604911e-01 -6.45294249e-01 -1.31516278e+00 -6.03647232e-01
-1.12445401e-02 1.51251629e-03 1.02897751e+00 1.06688762e+00
1.72842398e-01 1.98736683e-01 7.81164318e-02 -1.96235225e-01
-1.51468396e-01 -1.00440252e+00 -2.54708201e-01 -1.77669659e-01
1.57265201e-01 -5.57674289e-01 1.81149662e-01 -3.77550647e-02] | [11.147696495056152, 9.98476505279541] |
a3927d5d-6026-4838-a95b-2ccc23c54d6f | acoustic-identification-of-ae-aegypti | 2306.10091 | null | https://arxiv.org/abs/2306.10091v1 | https://arxiv.org/pdf/2306.10091v1.pdf | Acoustic Identification of Ae. aegypti Mosquitoes using Smartphone Apps and Residual Convolutional Neural Networks | In this paper, we advocate in favor of smartphone apps as low-cost, easy-to-deploy solution for raising awareness among the population on the proliferation of Aedes aegypti mosquitoes. Nevertheless, devising such a smartphone app is challenging, for many reasons, including the required maturity level of techniques for identifying mosquitoes based on features that can be captured using smartphone resources. In this paper, we identify a set of (non-exhaustive) requirements that smartphone apps must meet to become an effective tooling in the fight against Ae. aegypti, and advance the state-of-the-art with (i) a residual convolutional neural network for classifying Ae. aegypti mosquitoes from their wingbeat sound, (ii) a methodology for reducing the influence of background noise in the classification process, and (iii) a dataset for benchmarking solutions for detecting Ae. aegypti mosquitoes from wingbeat sound recordings. From the analysis of accuracy and recall, we provide evidence that convolutional neural networks have potential as a cornerstone for tracking mosquito apps for smartphones. | ['Weverton Cordeiro', 'Rodrigo Brandão Mansilha1', 'Mariana Recamonde-Mendoza', 'Ricardo Rohweder', 'Kayuã Oleques Paim'] | 2023-06-16 | null | null | null | null | ['benchmarking', 'benchmarking'] | ['miscellaneous', 'robots'] | [ 1.68039247e-01 -5.08415580e-01 1.15704119e-01 -3.49233568e-01
-3.05756062e-01 -7.99194753e-01 6.64729536e-01 1.37276563e-03
-4.29246783e-01 4.20681953e-01 9.01893601e-02 -6.15179121e-01
-1.59652740e-01 -7.95397818e-01 -4.61101145e-01 -6.15775943e-01
-6.02265537e-01 -1.02820478e-01 -2.31506959e-01 -3.04556400e-01
1.46751136e-01 6.79228723e-01 -1.73615265e+00 2.99805105e-01
4.28423285e-01 1.22021008e+00 -1.97863221e-01 8.60756934e-01
1.08330123e-01 3.53883117e-01 -1.11670578e+00 -3.94445211e-01
1.77029192e-01 -3.34713608e-02 -2.76660055e-01 -6.95855439e-01
4.97422338e-01 -7.24135101e-01 2.57426411e-01 7.65641510e-01
6.62193596e-01 -5.49478412e-01 4.55417126e-01 -1.11363542e+00
7.41187558e-02 1.05431609e-01 -3.68506551e-01 7.73597240e-01
6.29527748e-01 4.61152107e-01 2.06567451e-01 -5.26081502e-01
1.27787963e-01 8.45822394e-01 1.28008926e+00 7.37721562e-01
-8.32720816e-01 -1.04728889e+00 -2.95708984e-01 -1.76615536e-01
-1.28250098e+00 -9.17211473e-01 3.15230191e-01 -7.89493859e-01
1.15561020e+00 5.03041625e-01 9.36090946e-01 1.32830667e+00
3.90392035e-01 4.23399657e-02 8.61068964e-01 -9.60970297e-02
3.38538617e-01 1.45693406e-01 -1.19922450e-02 3.44791979e-01
6.29099011e-01 2.03157648e-01 -6.01441443e-01 -4.82930928e-01
2.87799060e-01 3.58654648e-01 -1.38547152e-01 2.30078727e-01
-4.82439667e-01 8.85723233e-01 1.36535347e-01 5.30268192e-01
-4.19855565e-01 -4.82673980e-02 4.84106243e-01 1.65151313e-01
8.13304901e-01 6.48392558e-01 -6.57308877e-01 -5.74590743e-01
-1.39208651e+00 1.73078761e-01 9.23112452e-01 2.54362151e-02
3.55756372e-01 1.43001661e-01 1.28471151e-01 3.67345601e-01
5.00647545e-01 9.88451242e-01 1.72061533e-01 -4.77930814e-01
2.30930358e-01 7.16485918e-01 2.13506982e-01 -1.11020267e+00
-5.24526358e-01 -2.81825304e-01 -6.49581194e-01 -8.19067284e-02
2.72215128e-01 -7.88462162e-01 -4.45026547e-01 1.48991179e+00
2.82512099e-01 2.40946516e-01 -1.47240624e-01 5.01622915e-01
9.04456675e-01 7.91526020e-01 2.18811959e-01 -7.33133852e-02
1.35733426e+00 -3.13394442e-02 -4.63746220e-01 -2.13287249e-01
3.83155406e-01 -5.65115631e-01 5.43371320e-01 4.17902321e-01
-7.05166221e-01 -4.64895308e-01 -1.10199463e+00 6.97465539e-01
-1.13792670e+00 2.65744269e-01 9.35228765e-01 1.38658988e+00
-1.23990798e+00 2.95587152e-01 -1.02209949e+00 -5.75679243e-01
5.46768665e-01 8.17116916e-01 -4.05289382e-01 4.03269321e-01
-8.88241649e-01 6.94258511e-01 -2.30978742e-01 2.42083490e-01
-7.01198220e-01 -9.46436584e-01 -7.18016446e-01 -6.95745135e-03
-2.08790004e-01 -5.36403954e-02 9.67820585e-01 -7.48275280e-01
-1.39611530e+00 6.76897824e-01 -7.66489655e-02 -7.06386268e-01
2.41174251e-02 -4.76950139e-01 -6.58599615e-01 2.35664397e-01
1.52674511e-01 5.64417422e-01 9.44688499e-01 -5.85538507e-01
-1.01963997e+00 -6.92716241e-01 5.64853884e-02 -3.25503200e-01
-8.09227645e-01 2.68209308e-01 1.62887037e-01 -4.24527168e-01
-8.36619198e-01 -7.32408583e-01 1.26352623e-01 -2.87404984e-01
1.44193813e-01 -1.59108266e-01 1.26006317e+00 -9.16373312e-01
1.33972192e+00 -1.95037019e+00 -4.51097995e-01 1.30852520e-01
1.56116739e-01 1.15375340e+00 6.65823836e-03 4.34879810e-01
2.44806856e-01 2.85860151e-01 4.17264998e-02 -1.26716837e-01
-3.91056091e-01 -2.61690259e-01 -5.86227119e-01 3.72496963e-01
4.34047371e-01 4.53345865e-01 -8.59753191e-01 8.49947333e-02
5.79549611e-01 1.41218960e+00 -3.06260496e-01 4.89686280e-01
2.32919216e-01 3.46157581e-01 -2.67480075e-01 7.89916575e-01
6.92421198e-01 2.08147243e-01 -1.69884652e-01 -1.01354368e-01
-4.66706425e-01 4.07849133e-01 -7.40251482e-01 9.38345790e-01
-5.80604315e-01 8.49438906e-01 3.52781892e-01 -5.81344485e-01
8.13003063e-01 4.10567462e-01 7.75479198e-01 -3.28279167e-01
2.10039988e-01 1.54699922e-01 -2.47908960e-04 -4.61980104e-01
2.45118752e-01 5.05216062e-01 1.40138149e-01 4.70159084e-01
4.09019679e-01 2.81332076e-01 4.36109036e-01 -2.67806172e-01
1.06322503e+00 -9.26232263e-02 1.04466349e-01 -6.08721554e-01
6.88838661e-01 -3.21751833e-01 8.89461711e-02 5.74394107e-01
-3.41920435e-01 3.71958539e-02 1.53359801e-01 -1.20952940e+00
-3.02908450e-01 -5.12151420e-01 -2.33144626e-01 1.16249788e+00
-5.49810171e-01 -5.12776434e-01 -1.37916052e+00 -4.67458785e-01
-3.02995533e-01 1.92936257e-01 -9.60785925e-01 1.15204588e-01
-3.65881473e-01 -1.18598509e+00 9.47684407e-01 2.14792892e-01
6.44008756e-01 -9.02268887e-01 -1.62652767e+00 1.64966300e-01
2.14456171e-01 -8.53933573e-01 -1.84530988e-01 3.68020386e-01
-5.04682243e-01 -1.35617912e+00 -6.32293701e-01 -2.70129919e-01
3.96154135e-01 5.58069885e-01 8.72622371e-01 1.14122689e-01
-3.84658039e-01 5.79143822e-01 -1.31451890e-01 -7.95720577e-01
-1.39242306e-01 1.11665659e-01 2.09313974e-01 -4.94282022e-02
1.20725691e+00 -2.11032882e-01 -8.00719023e-01 2.39379585e-01
-7.96434700e-01 -4.42311555e-01 2.19577968e-01 1.76272213e-01
-2.84733251e-02 1.05993688e-01 2.82266855e-01 -3.40022594e-01
7.67844856e-01 -7.50061154e-01 -8.15604746e-01 1.17489114e-01
-9.05298516e-02 -5.38233042e-01 6.94172561e-01 -2.86205262e-01
-6.59400940e-01 2.21771419e-01 -3.92558485e-01 2.96696633e-01
-5.79930484e-01 1.91794142e-01 1.96898565e-01 -4.28254008e-01
6.04194343e-01 2.84402352e-02 -2.64370125e-02 -1.38328910e-01
-1.33103073e-01 9.56732810e-01 2.76912630e-01 -5.01802601e-02
3.16802055e-01 5.34635842e-01 1.15454108e-01 -1.36055529e+00
-4.73142058e-01 -6.03223741e-01 -1.45262629e-01 -4.49526250e-01
8.38014841e-01 -9.02312756e-01 -9.84129012e-01 6.93368912e-01
-1.31776655e+00 -1.66550532e-01 4.01739746e-01 2.19626814e-01
2.82158524e-01 2.44789850e-02 -3.63547325e-01 -1.12783647e+00
-8.66599917e-01 -1.20385134e+00 1.25994051e+00 5.59994459e-01
-4.95011270e-01 -9.48889077e-01 5.59180915e-01 1.92446351e-01
1.10398853e+00 5.58157682e-01 -1.38797581e-01 -7.18408227e-01
2.15991676e-01 -5.46850026e-01 -1.34152234e-01 1.73059717e-01
4.69568193e-01 3.59603614e-01 -1.31970811e+00 -4.67316657e-01
2.23299950e-01 -1.27405077e-01 6.20263934e-01 6.04807973e-01
7.68779457e-01 -6.41896725e-01 -3.11254352e-01 8.28216434e-01
1.32908177e+00 5.24111927e-01 2.78284341e-01 2.15060696e-01
2.30664000e-01 8.56009185e-01 -7.04672933e-02 6.08848333e-01
1.76999837e-01 5.96525908e-01 2.90305197e-01 -8.43507797e-02
3.61317366e-01 9.31748655e-04 6.76717222e-01 1.37561902e-01
-4.26813394e-01 -1.92914829e-01 -1.11833119e+00 2.26322055e-01
-1.18118179e+00 -9.01926517e-01 1.99941218e-01 2.30428100e+00
3.70516241e-01 -2.94976652e-01 6.04570925e-01 1.06496684e-01
4.48220670e-01 -1.11648165e-01 -1.25366867e-01 -4.98938322e-01
2.62222677e-01 5.99841356e-01 5.54270923e-01 3.82176280e-01
-1.47033238e+00 6.43486857e-01 6.17966366e+00 3.57326716e-01
-1.64383149e+00 -5.35232686e-02 7.63687909e-01 8.94830152e-02
5.59528649e-01 -7.98238814e-01 -1.12797523e+00 9.52755511e-01
1.45723796e+00 4.53390539e-01 5.76346159e-01 7.94906735e-01
1.30312935e-01 -4.52948827e-03 -7.93084443e-01 1.12678230e+00
3.46299142e-01 -1.33330369e+00 -3.03574577e-02 3.08272481e-01
3.82321596e-01 5.08130752e-02 2.62290895e-01 -1.07066162e-01
-1.22784652e-01 -1.14466751e+00 1.24617200e-03 3.12707275e-01
7.40635753e-01 -8.56571317e-01 1.35240686e+00 -1.17902696e-01
-1.33888400e+00 -2.52211411e-02 -3.72994505e-02 -3.26332361e-01
-2.89608896e-01 1.83739170e-01 -6.94461703e-01 -9.08793733e-02
1.24880433e+00 5.71304083e-01 -8.21561277e-01 8.59073102e-01
1.19866177e-01 8.33603919e-01 -6.37944877e-01 -3.37245315e-01
2.59970337e-01 4.62157466e-02 1.34683311e-01 1.64483130e+00
5.78227699e-01 -1.19038582e-01 -3.99544954e-01 4.76864070e-01
4.18816537e-01 -2.38937557e-01 -1.04933178e+00 -3.79641503e-01
4.34138536e-01 1.28899252e+00 -8.46763372e-01 2.69296691e-02
-3.94109279e-01 4.23725158e-01 -3.85776550e-01 1.43456012e-01
-3.06939244e-01 -5.39558351e-01 8.65866065e-01 1.50780991e-01
2.11224839e-01 7.78923184e-02 4.33680005e-02 -7.39583313e-01
-1.94293424e-01 -9.97467101e-01 1.52081072e-01 -1.89955488e-01
-8.90361488e-01 8.48281324e-01 -9.75065529e-02 -9.04308975e-01
-7.15190649e-01 -9.44074154e-01 -8.04914713e-01 6.36476099e-01
-1.14986515e+00 -1.25095904e+00 -4.26652223e-01 3.31369966e-01
2.92834252e-01 -5.95535219e-01 1.18896735e+00 2.14445889e-01
-3.57187659e-01 2.78583556e-01 -2.67992795e-01 1.06358632e-01
2.13504750e-02 -8.80080700e-01 6.52311444e-01 8.09908152e-01
1.88132003e-01 1.09244359e+00 6.90119863e-01 -5.13736606e-01
-1.74283278e+00 -9.31321859e-01 7.21024036e-01 -8.01166177e-01
4.96971190e-01 -6.53436005e-01 -4.40903604e-01 2.87583649e-01
2.70279258e-01 -4.34250504e-01 1.28728795e+00 -4.48409170e-02
-2.83963922e-02 -3.30082506e-01 -1.39996004e+00 4.10259098e-01
2.47047201e-01 -3.58682007e-01 -3.56928021e-01 1.63912073e-01
8.08750093e-02 4.46300432e-02 -6.69584215e-01 3.11533481e-01
9.23115492e-01 -1.14250660e+00 9.97321367e-01 -2.67034471e-01
1.43652886e-01 -2.01643407e-01 -2.45063435e-02 -1.07195365e+00
2.89627671e-01 -1.02317047e+00 -2.41569012e-01 1.18097663e+00
2.35354695e-02 -6.35523617e-01 8.59490991e-01 3.07497233e-01
4.50990468e-01 -6.53692544e-01 -9.67028022e-01 -4.20148224e-01
-3.03017825e-01 -4.41889018e-01 7.38431633e-01 4.62992430e-01
-2.84229189e-01 2.61255950e-01 -6.09336615e-01 1.45829648e-01
3.19657564e-01 -6.48931742e-01 8.34917307e-01 -1.39551473e+00
1.65605426e-01 -3.63037348e-01 -5.54166734e-01 -3.33209187e-01
-3.79350513e-01 6.46031424e-02 -2.10057721e-01 -1.12546217e+00
-2.39496365e-01 1.51608497e-01 -1.63119733e-01 4.37215596e-01
2.50727534e-01 6.25652730e-01 -6.17348664e-02 -1.35314152e-01
-4.77234662e-01 -2.70792246e-01 -6.21477980e-03 -1.69487279e-02
-4.12366360e-01 2.60162532e-01 -6.86200261e-01 6.49141729e-01
9.95106578e-01 -3.73829871e-01 -2.41059855e-01 -4.53733295e-01
3.44760716e-01 -2.28852585e-01 4.22582388e-01 -1.42735624e+00
2.37457216e-01 2.85077304e-01 7.30578125e-01 -3.30741733e-01
4.91464883e-01 -1.02800548e+00 1.18944801e-01 9.15754259e-01
9.03573483e-02 3.53154212e-01 8.30433011e-01 9.82157663e-02
1.73200537e-02 1.38914093e-01 4.71669227e-01 3.07964049e-02
-2.63480097e-01 1.91125885e-01 -7.39097714e-01 -5.43228127e-02
7.16609955e-01 -4.87169325e-01 -5.25821328e-01 -1.32731780e-01
9.21620950e-02 -1.89439982e-01 1.97711244e-01 5.04352629e-01
4.87433225e-01 -6.17723227e-01 -5.23010314e-01 6.61067486e-01
-9.40208882e-02 -7.11765945e-01 8.36859867e-02 3.09599102e-01
-7.60936320e-01 9.46361303e-01 -7.76850402e-01 -6.45042121e-01
-1.49995792e+00 2.27318376e-01 2.88080484e-01 -3.87150720e-02
2.07400054e-01 9.19926524e-01 -1.67590499e-01 -3.23457509e-01
5.80958903e-01 -2.47725621e-01 -6.05629981e-01 1.60536379e-01
1.36311519e+00 4.61820543e-01 2.61187017e-01 -7.05165148e-01
-8.33682179e-01 4.91742909e-01 1.18815266e-01 1.52839646e-01
1.51085413e+00 5.12284875e-01 -1.31178364e-01 1.06003813e-01
9.49490249e-01 4.01022695e-02 -1.24728084e+00 9.14232552e-01
-7.83265457e-02 -2.45370895e-01 2.03601569e-01 -9.77938473e-01
-9.85765696e-01 1.10972452e+00 1.46853352e+00 7.10308492e-01
9.46942389e-01 -5.46333730e-01 5.95362186e-01 4.12211031e-01
4.22159314e-01 -7.91986585e-01 -2.53040195e-01 1.40155658e-01
5.32165229e-01 -1.17887318e+00 -1.01767525e-01 1.13619745e-01
-3.65412357e-04 1.30272388e+00 4.23052251e-01 4.43531610e-02
7.86991775e-01 6.09270632e-01 1.12986594e-01 -3.40226352e-01
-4.14457440e-01 7.25564510e-02 2.63198018e-01 1.21641648e+00
6.72398746e-01 1.57670617e-01 -6.18620440e-02 4.92328703e-01
-1.79452330e-01 1.81215674e-01 -5.74501976e-02 8.39227915e-01
-4.52374041e-01 -8.03692877e-01 -3.84182721e-01 7.31747925e-01
-8.87138486e-01 -9.55058485e-02 -8.59867454e-01 6.49630010e-01
6.22080743e-01 1.46646667e+00 1.57721132e-01 -6.39777720e-01
6.31633997e-02 -4.18813735e-01 7.51903877e-02 -3.53621125e-01
-1.25122154e+00 -3.47098708e-01 -1.99203953e-01 -3.51109356e-01
-6.29687726e-01 -3.02662969e-01 6.65905997e-02 -3.92243713e-01
6.55493140e-03 3.07900041e-01 1.13765156e+00 7.23137736e-01
6.03802443e-01 -1.01390118e-02 3.50566268e-01 -9.93817210e-01
-2.52994820e-02 -9.98777807e-01 4.91590276e-02 -8.18321630e-02
4.29767668e-01 -5.12996972e-01 -4.75819468e-01 -1.44815579e-01] | [13.311408996582031, 1.1460202932357788] |
0a6c8687-55f9-40f6-9cf1-e0269203b9f7 | masked-autoencoders-for-generic-event | 2206.08610 | null | https://arxiv.org/abs/2206.08610v1 | https://arxiv.org/pdf/2206.08610v1.pdf | Masked Autoencoders for Generic Event Boundary Detection CVPR'2022 Kinetics-GEBD Challenge | Generic Event Boundary Detection (GEBD) tasks aim at detecting generic, taxonomy-free event boundaries that segment a whole video into chunks. In this paper, we apply Masked Autoencoders to improve algorithm performance on the GEBD tasks. Our approach mainly adopted the ensemble of Masked Autoencoders fine-tuned on the GEBD task as a self-supervised learner with other base models. Moreover, we also use a semi-supervised pseudo-label method to take full advantage of the abundant unlabeled Kinetics-400 data while training. In addition, we propose a soft-label method to partially balance the positive and negative samples and alleviate the problem of ambiguous labeling in this task. Lastly, a tricky segmentation alignment policy is implemented to refine boundaries predicted by our models to more accurate locations. With our approach, we achieved 85.94% on the F1-score on the Kinetics-GEBD test set, which improved the F1-score by 2.31% compared to the winner of the 2021 Kinetics-GEBD Challenge. Our code is available at https://github.com/ContentAndMaterialPortrait/MAE-GEBD. | ['Jie Tang', 'Xu Cheng', 'Feng Hu', 'Zuwei Huang', 'Youzeng Li', 'Yuanxi Sun', 'Rui He'] | 2022-06-17 | null | null | null | null | ['boundary-detection'] | ['computer-vision'] | [-2.28306651e-01 1.53278381e-01 -4.95918579e-02 -3.15106064e-01
-9.98728395e-01 -4.66095507e-01 3.35552454e-01 2.59516425e-02
-7.65017271e-01 6.96992874e-01 1.42025769e-01 1.15171149e-01
4.49474156e-01 -6.36952639e-01 -1.07309306e+00 -6.74767971e-01
4.24157530e-02 5.30374110e-01 4.56422597e-01 1.06745206e-01
-2.17978090e-01 1.43381238e-01 -1.51528335e+00 5.95052361e-01
8.44793499e-01 1.06133640e+00 2.48181626e-01 7.11331487e-01
2.18506023e-01 7.04005361e-01 -7.29216218e-01 -3.30624670e-01
3.13927650e-01 -5.38970709e-01 -7.46066689e-01 5.62781692e-02
3.05101037e-01 -3.96271110e-01 -5.21361113e-01 8.50545347e-01
5.92774689e-01 1.76319808e-01 7.51723170e-01 -1.28444409e+00
-1.88218221e-01 6.19733989e-01 -7.35878885e-01 4.80111301e-01
-1.23008698e-01 1.72612816e-01 8.59735906e-01 -9.75232244e-01
6.57668710e-01 6.20096743e-01 5.81020415e-01 6.93583965e-01
-9.33841348e-01 -7.61213362e-01 2.66538650e-01 5.21391392e-01
-1.53169036e+00 -2.37266794e-01 4.79180962e-01 -6.66612208e-01
9.01241541e-01 7.39656342e-03 6.87179267e-01 1.38871455e+00
1.44339958e-02 9.99901295e-01 7.87251770e-01 -3.76264781e-01
2.19989702e-01 -3.05757113e-02 1.55677021e-01 6.92090929e-01
1.36498259e-02 1.36586562e-01 -5.46121836e-01 1.85634404e-01
6.94735348e-01 -3.25879045e-02 -2.68853635e-01 -5.23429848e-02
-1.15422070e+00 7.39688516e-01 3.12769204e-01 2.07388356e-01
-4.78300720e-01 -2.44682893e-01 6.03762150e-01 -9.80807021e-02
6.58040464e-01 2.06348807e-01 -6.31489277e-01 -9.64503214e-02
-9.69311059e-01 2.26055160e-01 6.18252039e-01 7.88541257e-01
6.97151661e-01 -1.40717134e-01 -4.04843658e-01 6.59580588e-01
3.07875443e-02 9.06774625e-02 6.20509744e-01 -1.06200457e+00
2.74325401e-01 4.70876127e-01 1.69236079e-01 -5.06318510e-01
-5.13325453e-01 -4.40924078e-01 -6.69664860e-01 1.35777891e-01
5.60355127e-01 -4.32304531e-01 -1.02369559e+00 1.76876354e+00
6.25879586e-01 4.96948749e-01 -1.51344880e-01 1.22789288e+00
9.21473920e-01 7.62629330e-01 2.89028704e-01 -6.72682375e-02
1.53046227e+00 -1.16637886e+00 -7.04715073e-01 -1.06235139e-01
6.96251869e-01 -5.46745121e-01 9.07701015e-01 4.48099315e-01
-9.07561004e-01 -7.51093626e-01 -1.09667349e+00 -6.04559900e-03
-1.96326137e-01 5.53618073e-01 3.72326761e-01 2.75299907e-01
-9.44992542e-01 5.09297729e-01 -1.18157649e+00 -1.02325775e-01
4.63935286e-01 2.28824124e-01 -2.03658372e-01 2.66276926e-01
-1.33471811e+00 4.49322730e-01 9.08051431e-01 8.12965408e-02
-1.21214056e+00 -5.48254371e-01 -6.69830382e-01 -2.08435338e-02
7.06104338e-01 -4.59639400e-01 1.44596076e+00 -9.99263704e-01
-1.29280937e+00 1.15242982e+00 -1.27000108e-01 -6.39628410e-01
7.56961524e-01 -4.60096210e-01 -3.84460777e-01 3.43650699e-01
1.79007322e-01 1.03678274e+00 6.66545868e-01 -9.43555772e-01
-8.99257481e-01 -1.09380921e-02 -2.15226114e-01 1.47795916e-01
-5.66524491e-02 3.75707969e-02 -6.28965616e-01 -7.40963221e-01
-1.27308950e-01 -9.69979286e-01 -6.50466830e-02 -4.13232207e-01
-1.99496061e-01 -4.37015444e-01 6.62627041e-01 -1.03985739e+00
1.14815509e+00 -2.31860089e+00 1.32558057e-02 -1.73612028e-01
3.96914929e-01 2.68120527e-01 5.27568683e-02 1.75212070e-01
-2.63517827e-01 2.52699456e-03 -1.42528117e-01 -4.54816848e-01
-1.98292166e-01 -3.59016843e-02 -1.03166915e-01 3.55587602e-01
3.84045810e-01 8.05553555e-01 -7.03012288e-01 -5.29381514e-01
-1.65709853e-02 3.34559947e-01 -3.77818108e-01 3.55544478e-01
-4.40317929e-01 4.90045607e-01 -8.02272186e-02 4.56506938e-01
4.35440421e-01 -4.30900991e-01 2.29568034e-02 -9.36866254e-02
-5.12153730e-02 2.78464526e-01 -1.18185890e+00 1.72437501e+00
-2.43249405e-02 5.88777959e-01 -4.68264567e-03 -9.19189751e-01
6.27200603e-01 4.71919805e-01 6.83095217e-01 -4.70521092e-01
3.56382728e-01 7.38919079e-02 -7.60259405e-02 -5.12428463e-01
3.03680718e-01 9.03867856e-02 3.23789679e-02 4.45770830e-01
2.86208272e-01 5.14999032e-01 5.16853034e-01 2.34679326e-01
1.07487094e+00 4.00468647e-01 2.34431863e-01 -8.77089500e-02
2.56590724e-01 1.93978876e-01 1.10091043e+00 6.76961780e-01
-5.08014739e-01 8.83049667e-01 3.92855376e-01 -4.56720829e-01
-9.67822909e-01 -8.91378820e-01 5.13865538e-02 1.09535980e+00
1.23304211e-01 -4.91358250e-01 -1.19334137e+00 -9.34833527e-01
-2.80120850e-01 5.78235805e-01 -6.47618413e-01 -1.41117498e-01
-5.19626141e-01 -9.32990551e-01 6.84017062e-01 8.26664388e-01
5.71796954e-01 -1.15390539e+00 -5.03729761e-01 3.96452010e-01
-6.40308738e-01 -1.18759894e+00 -6.23493850e-01 3.54998261e-01
-5.53262889e-01 -1.13897860e+00 -7.86676466e-01 -9.03084457e-01
3.53165239e-01 -5.83283678e-02 8.31761420e-01 -2.96337247e-01
-2.28190958e-01 -1.91353541e-02 -6.40735745e-01 -4.55334306e-01
-3.05264831e-01 1.22356333e-01 5.48275262e-02 4.84596007e-02
5.92832983e-01 -2.43905768e-01 -7.54047573e-01 5.27205884e-01
-7.31444776e-01 2.40358382e-01 3.88099670e-01 8.00611854e-01
8.84631217e-01 1.66574419e-02 6.72410071e-01 -5.67752123e-01
1.85623094e-01 -5.33776283e-01 -5.73682189e-01 -1.59117002e-02
-4.96413410e-01 -2.49318406e-01 2.17573896e-01 -6.76215827e-01
-1.20774710e+00 1.05573796e-01 -3.92312020e-01 -6.34432256e-01
-4.86506999e-01 2.20965847e-01 -2.80258000e-01 3.93908173e-01
7.57584989e-01 8.46939087e-02 -2.03563660e-01 -4.94443417e-01
1.16195202e-01 8.14455509e-01 7.28079081e-01 -5.30958176e-01
5.43160997e-02 4.05114412e-01 -5.84696591e-01 -3.50318104e-01
-9.85048711e-01 -5.32999754e-01 -5.11159062e-01 -3.03606451e-01
1.13425493e+00 -1.38281584e+00 -5.40867984e-01 8.07094634e-01
-1.05206251e+00 -8.95227969e-01 -2.89780855e-01 6.34974897e-01
-4.88090992e-01 4.03410226e-01 -1.11845922e+00 -4.55692053e-01
-3.28080118e-01 -9.76013720e-01 9.39338267e-01 4.60950434e-01
-2.72849679e-01 -4.40467954e-01 6.27856255e-02 4.67401743e-01
-1.08142465e-01 2.57426739e-01 3.55702132e-01 -1.03859162e+00
-3.57832909e-01 -3.28385271e-02 -8.73756856e-02 4.35372114e-01
-1.58736661e-01 3.90132628e-02 -1.10034633e+00 -1.05128936e-01
-1.75209697e-02 -3.67085934e-01 1.02443969e+00 5.73767185e-01
1.24374235e+00 -1.29029751e-01 -3.46581191e-01 5.01663089e-01
9.04373944e-01 3.34616303e-01 4.36738253e-01 4.43268925e-01
5.90242684e-01 5.08997798e-01 8.74411523e-01 6.95772588e-01
4.71909136e-01 6.50428891e-01 2.16026366e-01 -6.16974980e-02
-2.50730783e-01 -2.79762745e-01 3.97119910e-01 4.97840106e-01
8.22654217e-02 -6.77331209e-01 -1.00627005e+00 6.16064548e-01
-2.06278706e+00 -7.71533728e-01 -2.90436655e-01 1.87123013e+00
9.40469503e-01 3.14906359e-01 4.68048930e-01 3.61396819e-02
1.12342620e+00 -5.14905639e-02 -4.61538464e-01 -1.65855046e-02
-1.80239454e-01 1.72294229e-02 4.41447139e-01 2.49749452e-01
-1.55217767e+00 1.08264005e+00 5.14769268e+00 1.11933947e+00
-1.06268883e+00 5.29191077e-01 8.81446540e-01 -2.39127085e-01
2.99972445e-01 -2.92117804e-01 -9.78142083e-01 7.96947181e-01
1.11039150e+00 2.13241100e-01 2.07193360e-01 7.06475198e-01
3.21920991e-01 -2.02575013e-01 -9.26206768e-01 9.43798840e-01
-1.21525034e-01 -1.19875550e+00 -5.22543430e-01 -3.77540849e-03
8.65053833e-01 3.57006967e-01 -3.59716862e-01 5.58221757e-01
2.75945425e-01 -6.39373899e-01 1.02186871e+00 2.87420928e-01
7.05101132e-01 -5.91355443e-01 6.92424834e-01 5.11368632e-01
-9.90686715e-01 8.61406773e-02 -1.19162306e-01 -1.08704813e-01
3.32200974e-01 8.06860030e-01 -8.86371315e-01 4.28085983e-01
1.04450691e+00 4.79222596e-01 -3.60058606e-01 1.23194337e+00
-4.72075343e-01 9.79547024e-01 -4.95862097e-01 2.89414108e-01
1.10268600e-01 3.42781725e-03 3.79320532e-01 1.31300116e+00
1.09321199e-01 1.65156886e-01 2.62943923e-01 6.96827710e-01
-1.47407055e-01 1.18515063e-02 -1.12170339e-01 2.48852260e-02
3.54556292e-01 1.30839264e+00 -9.86744940e-01 -5.57697296e-01
-2.38915160e-01 1.13800263e+00 3.43317837e-01 4.20729518e-01
-1.29700220e+00 -2.13373452e-01 4.19840276e-01 1.43593833e-01
5.27350366e-01 -7.67229646e-02 -8.02720636e-02 -1.28188300e+00
-7.83413723e-02 -8.12854350e-01 7.39427567e-01 -7.00493157e-01
-1.16991615e+00 6.38306558e-01 -4.34867926e-02 -1.01832747e+00
-2.37203732e-01 -3.40829134e-01 -5.47958195e-01 5.40334046e-01
-9.98480558e-01 -9.41923678e-01 -4.43112910e-01 3.94947350e-01
6.75117612e-01 1.40150398e-01 4.85164940e-01 5.72746456e-01
-9.29308295e-01 5.81052899e-01 4.16867323e-02 4.63786483e-01
9.91973758e-01 -1.18530607e+00 4.37146217e-01 1.08715796e+00
1.44837588e-01 1.47312015e-01 4.85339969e-01 -8.91274631e-01
-3.31427842e-01 -1.21607828e+00 6.46792412e-01 -1.67679965e-01
5.01790822e-01 -4.73197907e-01 -1.10800803e+00 9.67263937e-01
9.72011313e-02 5.57178557e-02 7.08507299e-01 4.00084555e-02
-1.82533905e-01 1.46930516e-01 -9.15645659e-01 4.01344597e-01
1.18155026e+00 -3.07187915e-01 -5.10915339e-01 2.81623542e-01
8.07129562e-01 -7.29132414e-01 -9.49989200e-01 4.96976316e-01
2.44286299e-01 -9.22937393e-01 6.04018211e-01 -4.43847120e-01
3.68309885e-01 -5.25029719e-01 1.02374807e-01 -1.10138690e+00
-3.10331702e-01 -3.13505024e-01 -4.41558182e-01 1.33278346e+00
1.56237438e-01 -4.01120543e-01 9.41033840e-01 3.71496856e-01
-3.71492535e-01 -7.65126586e-01 -8.86361480e-01 -7.73522913e-01
-2.08714098e-01 -2.82617301e-01 3.16737950e-01 7.20802188e-01
-1.34232104e-01 1.92968398e-01 -2.78024316e-01 2.49266639e-01
4.04645056e-01 -6.35556579e-02 5.08501232e-01 -8.37232113e-01
-5.19949853e-01 -3.02421212e-01 -1.92654014e-01 -1.16812563e+00
2.18910366e-01 -8.15365434e-01 2.39949420e-01 -1.15462661e+00
3.19694996e-01 -2.19707727e-01 -5.30727088e-01 7.68727005e-01
-3.81539822e-01 4.53549474e-01 5.96470060e-03 1.81992173e-01
-9.00247455e-01 5.92151880e-01 8.63108277e-01 -1.05552878e-02
-3.50881368e-01 -1.10450670e-01 -5.36615431e-01 5.54480970e-01
1.19913471e+00 -7.00388551e-01 -2.20027298e-01 -2.34267175e-01
-1.00748383e-01 -1.11901022e-01 3.20219666e-01 -1.04195869e+00
6.47937134e-02 7.65390054e-04 5.04162371e-01 -6.78795874e-01
9.49434862e-02 -4.54019666e-01 3.52975845e-01 3.80189449e-01
-1.59924939e-01 -1.92835316e-01 4.00844127e-01 5.61122775e-01
-2.15818420e-01 -2.07896039e-01 5.13414323e-01 -9.37834159e-02
-9.91956472e-01 2.08501518e-01 -5.75751662e-01 1.40704662e-01
1.31291413e+00 -5.77533208e-02 -3.43380690e-01 -2.48272941e-01
-9.67214227e-01 4.29880083e-01 5.26879311e-01 3.05041105e-01
3.04158688e-01 -9.89152074e-01 -7.66245902e-01 2.88814101e-02
1.21698901e-01 3.70506853e-01 6.27471566e-01 1.02507222e+00
-5.31824946e-01 6.83086589e-02 -2.41905436e-01 -6.26347661e-01
-1.09046447e+00 5.80450475e-01 4.61460710e-01 -3.80253196e-01
-7.19334722e-01 1.02579713e+00 3.91060293e-01 -1.70180485e-01
2.87074924e-01 -5.26287183e-02 -9.17256027e-02 2.07958072e-01
5.73641121e-01 4.14638549e-01 1.97814390e-01 -4.34051007e-01
-4.00415421e-01 1.26708254e-01 -2.67361552e-01 -1.70679778e-01
1.21674716e+00 -1.27543643e-01 3.79151344e-01 2.90246427e-01
9.39388037e-01 -1.14250474e-01 -1.87796211e+00 -9.93022025e-02
-5.07370606e-02 1.05514161e-01 -1.12908725e-02 -9.88639176e-01
-1.03472114e+00 6.69948936e-01 6.29057467e-01 -5.73506430e-02
1.16348684e+00 1.40899658e-01 9.36147451e-01 -1.63477510e-02
5.03213406e-02 -1.07696116e+00 6.96777627e-02 3.73910248e-01
5.90800166e-01 -1.36110437e+00 -3.31944019e-01 -3.51674795e-01
-9.03271616e-01 8.25060606e-01 8.27121496e-01 -9.08268094e-02
4.35141623e-01 2.63755649e-01 2.05230936e-01 -6.22368753e-02
-7.01146781e-01 -3.03888559e-01 2.10715532e-01 3.83686572e-01
2.20284835e-01 9.97857526e-02 -4.71493751e-01 1.08535552e+00
8.11234340e-02 2.27484778e-01 4.04673845e-01 8.26484144e-01
-3.49845409e-01 -8.56723845e-01 -1.88792169e-01 3.94016713e-01
-6.85735822e-01 1.01632468e-01 -2.58131325e-01 7.30351031e-01
3.50253999e-01 8.42836559e-01 1.31338298e-01 -5.40775239e-01
1.00557938e-01 3.69155437e-01 2.92274058e-01 -5.12984574e-01
-5.65469146e-01 5.36253512e-01 8.17612410e-02 -6.07850671e-01
-2.85357833e-01 -7.69479513e-01 -1.57591641e+00 -9.03181061e-02
-2.73328602e-01 1.71436578e-01 3.92728657e-01 7.13135064e-01
6.47429824e-01 7.34491885e-01 2.11238265e-01 -7.99617589e-01
-2.32284769e-01 -1.03288031e+00 -5.70252478e-01 3.19286257e-01
7.80833140e-02 -7.23365188e-01 -2.12335646e-01 3.93337876e-01] | [8.704931259155273, 0.3719427287578583] |
5b61f5b3-8f5e-45d9-94a0-7dcf40c77d2d | robustness-and-risk-management-via | 2112.15430 | null | https://arxiv.org/abs/2112.15430v1 | https://arxiv.org/pdf/2112.15430v1.pdf | Robustness and risk management via distributional dynamic programming | In dynamic programming (DP) and reinforcement learning (RL), an agent learns to act optimally in terms of expected long-term return by sequentially interacting with its environment modeled by a Markov decision process (MDP). More generally in distributional reinforcement learning (DRL), the focus is on the whole distribution of the return, not just its expectation. Although DRL-based methods produced state-of-the-art performance in RL with function approximation, they involve additional quantities (compared to the non-distributional setting) that are still not well understood. As a first contribution, we introduce a new class of distributional operators, together with a practical DP algorithm for policy evaluation, that come with a robust MDP interpretation. Indeed, our approach reformulates through an augmented state space where each state is split into a worst-case substate and a best-case substate, whose values are maximized by safe and risky policies respectively. Finally, we derive distributional operators and DP algorithms solving a new control task: How to distinguish safe from risky optimal actions in order to break ties in the space of optimal policies? | ['Gergely Neu', 'Mastane Achab'] | 2021-12-28 | null | null | null | null | ['distributional-reinforcement-learning'] | ['methodology'] | [-2.29772162e-02 3.78494591e-01 -4.61763412e-01 -2.01128051e-01
-7.89183497e-01 -7.05626845e-01 8.07622075e-01 2.74691105e-01
-9.10991371e-01 1.20239663e+00 1.90198660e-01 -5.00286937e-01
-5.61222255e-01 -9.13789928e-01 -7.70984769e-01 -1.15868890e+00
-2.96991259e-01 8.53168309e-01 -1.71393052e-01 -9.52455215e-03
7.48810619e-02 4.93935287e-01 -1.41254020e+00 -1.84831724e-01
9.28299606e-01 9.85612214e-01 1.31272361e-01 5.82076550e-01
-1.43281251e-01 1.10504949e+00 -5.92022777e-01 -2.31465816e-01
2.34390840e-01 -4.07760352e-01 -8.74917626e-01 -1.82734936e-01
-2.90923059e-01 -5.38489044e-01 -9.07460451e-02 1.24744689e+00
5.47377110e-01 5.69949985e-01 8.97225678e-01 -1.40412343e+00
-3.40447694e-01 8.85514081e-01 -3.27341884e-01 -9.24962834e-02
3.45665306e-01 4.25511926e-01 1.19555318e+00 -2.22163647e-01
5.23047864e-01 1.48282230e+00 2.09169779e-02 5.68076730e-01
-1.61554682e+00 -3.97462063e-02 6.47521555e-01 8.02964717e-02
-5.56868553e-01 6.62234873e-02 3.17100108e-01 -4.34747696e-01
9.16607022e-01 -8.12751576e-02 7.65128911e-01 1.31563568e+00
2.74568409e-01 1.19805312e+00 1.49761295e+00 -3.92786294e-01
8.73594165e-01 1.51607499e-03 -5.95459752e-02 3.23801130e-01
2.38106009e-02 6.54593766e-01 -4.81748618e-02 -3.72876465e-01
5.51963866e-01 -1.12724878e-01 -9.16003734e-02 -8.19519460e-01
-1.03378665e+00 1.10543466e+00 -1.15882911e-01 6.05110452e-02
-9.22195852e-01 4.10654336e-01 4.43496019e-01 4.10769165e-01
1.83359072e-01 2.52379745e-01 -3.23073477e-01 -4.78972435e-01
-5.35496294e-01 8.42574537e-01 1.07948971e+00 6.49592936e-01
5.65290093e-01 1.43456191e-01 -7.41801679e-01 4.85848188e-01
2.68880099e-01 7.65361905e-01 7.00625330e-02 -1.13540244e+00
5.39175570e-01 -4.95006666e-02 7.36169398e-01 -4.10749704e-01
-4.00468975e-01 -4.96159345e-01 -4.36475277e-01 5.65127373e-01
7.79192626e-01 -5.97485363e-01 -4.59232360e-01 2.39339781e+00
2.54772514e-01 4.11086008e-02 2.91998535e-01 5.93639076e-01
-3.99361551e-01 7.32485771e-01 3.43543321e-01 -7.37607419e-01
7.42664039e-01 -6.98717415e-01 -6.21415555e-01 -4.89188023e-02
5.16783297e-01 -1.11440077e-01 1.16933167e+00 7.01963603e-01
-1.25987434e+00 -4.01369371e-02 -6.92967892e-01 3.96720171e-01
-8.72226283e-02 -2.48902440e-01 3.55347633e-01 5.11289060e-01
-1.10192335e+00 9.24010754e-01 -7.93260515e-01 8.33739787e-02
2.58203238e-01 1.59888104e-01 6.76395446e-02 2.05677301e-01
-1.26890028e+00 1.25706732e+00 5.15213907e-01 -1.05699420e-01
-1.57916665e+00 -5.89290619e-01 -7.49069095e-01 9.73307043e-02
1.08366787e+00 -6.99814677e-01 1.81235290e+00 -7.65492976e-01
-2.06067991e+00 5.13030410e-01 3.50443989e-01 -8.68679643e-01
9.58516657e-01 -2.40699679e-01 4.15473655e-02 6.35327101e-02
-1.14310227e-01 1.88458547e-01 9.28001344e-01 -1.11754608e+00
-8.47636759e-01 -4.23806250e-01 5.93861401e-01 4.05189991e-01
4.84517924e-02 -1.23278581e-01 3.39964807e-01 -5.12430727e-01
-5.67782402e-01 -9.07396793e-01 -5.24785399e-01 -4.18204039e-01
-2.59768456e-01 -5.17512798e-01 1.75677508e-01 -3.33290517e-01
1.31226158e+00 -1.88168585e+00 6.33901596e-01 1.90337285e-01
-2.39861816e-01 3.47271152e-02 -1.68934330e-01 5.77936292e-01
-5.15859053e-02 -3.07522295e-03 -4.15571511e-01 -3.20064873e-01
8.62779915e-01 5.94899654e-01 -6.92944467e-01 6.36818349e-01
-6.78935349e-02 7.02230990e-01 -1.23882949e+00 -2.03390747e-01
1.91042870e-01 -1.40257597e-01 -5.70168734e-01 2.96873480e-01
-8.16433072e-01 2.69378453e-01 -6.50027394e-01 2.47697040e-01
4.06642377e-01 3.20838422e-01 2.52410740e-01 5.83259583e-01
-3.64766747e-01 1.24931321e-01 -1.36003625e+00 1.43983650e+00
-5.70340991e-01 -1.11481741e-01 2.08689317e-01 -1.26300216e+00
5.89305758e-01 1.27891675e-01 7.59253919e-01 -5.44714332e-01
4.76630293e-02 1.17080331e-01 -1.64362639e-01 -4.84282345e-01
4.88710582e-01 -5.52634478e-01 -2.86090642e-01 9.15801048e-01
8.28479454e-02 -1.22478120e-01 3.20602685e-01 -2.42409445e-02
1.01824892e+00 5.09975493e-01 4.98425335e-01 -5.52818596e-01
4.85263824e-01 -2.24242911e-01 6.56295776e-01 9.92277920e-01
-3.27032357e-01 -2.28147253e-01 1.38774979e+00 -1.39296129e-01
-8.37348998e-01 -1.29110646e+00 8.52776095e-02 1.25931168e+00
-2.18870372e-01 -1.20161206e-03 -8.26046526e-01 -8.54434490e-01
3.57275158e-01 1.29857552e+00 -6.61545634e-01 -2.74017960e-01
-5.09344161e-01 -6.69513881e-01 3.55502844e-01 3.26189846e-01
2.24175051e-01 -1.32087982e+00 -1.02188337e+00 3.93124342e-01
2.12911382e-01 -5.51633656e-01 -3.86153460e-01 4.44858760e-01
-6.30860746e-01 -9.15542126e-01 -9.48824048e-01 -9.92130563e-02
3.05936754e-01 -5.50165832e-01 1.21572423e+00 -8.39389920e-01
2.30214447e-01 7.42016375e-01 -7.66585693e-02 -6.48607850e-01
-4.16162729e-01 -2.19470471e-01 3.62892598e-01 7.77683854e-02
3.30169685e-02 -4.95959818e-01 -4.89652902e-01 -9.95883346e-03
-9.15970862e-01 -4.44647074e-01 5.01976132e-01 8.39415431e-01
7.54224241e-01 1.98952723e-02 6.80324435e-01 -8.23411644e-01
9.68975067e-01 -5.42152941e-01 -1.01077151e+00 4.03092176e-01
-5.84134936e-01 7.25162625e-01 7.79876113e-01 -4.67178136e-01
-1.31872272e+00 -1.98821902e-01 -9.77291688e-02 -2.20165521e-01
-1.80195738e-02 4.38382953e-01 -2.97920257e-01 5.87621570e-01
4.34540212e-01 3.04633081e-01 1.94728494e-01 -3.57273817e-01
6.22418702e-01 2.78959900e-01 3.11046869e-01 -1.31341338e+00
3.80032152e-01 1.83795825e-01 2.17369318e-01 -5.33858478e-01
-7.97094285e-01 1.00281224e-01 -3.48337263e-01 -3.96680623e-01
7.33985245e-01 -3.80350053e-01 -1.21169353e+00 3.77997816e-01
-7.55965054e-01 -8.69854510e-01 -1.07311559e+00 5.36075234e-01
-1.49063551e+00 1.92056045e-01 -2.50933141e-01 -1.52037621e+00
1.13884933e-01 -1.21339798e+00 7.22440720e-01 7.63718486e-02
1.39992639e-01 -1.20741761e+00 4.45058823e-01 -3.81736994e-01
2.23377421e-01 3.94545734e-01 1.30503833e+00 -6.45576775e-01
-2.39431962e-01 3.27230364e-01 2.45277613e-01 6.30937696e-01
-3.93216878e-01 -1.92932382e-01 -5.64619482e-01 -3.72875005e-01
3.05287600e-01 -3.64035875e-01 7.45440781e-01 7.13476241e-01
1.09054554e+00 -5.58651090e-01 5.16512394e-02 2.61627883e-01
1.45301569e+00 4.20165539e-01 3.95190507e-01 4.17313904e-01
1.32199945e-02 8.30905855e-01 8.91802311e-01 9.87017334e-01
2.28590980e-01 4.59100962e-01 7.08427310e-01 6.23441815e-01
6.87602520e-01 -5.00478446e-01 9.01604950e-01 -2.81604230e-02
-7.63259158e-02 -2.14821860e-01 -7.27835357e-01 2.72735029e-01
-2.17318344e+00 -1.26200581e+00 4.44244534e-01 2.72165799e+00
8.79061162e-01 2.06095144e-01 6.50011837e-01 -7.48677850e-02
4.98610079e-01 2.08947524e-01 -1.03750932e+00 -6.89191818e-01
-1.17274271e-02 1.65694639e-01 7.00909913e-01 7.46041000e-01
-8.88081372e-01 5.29406488e-01 6.65734339e+00 9.54849541e-01
-8.65212381e-01 -3.66196148e-02 5.54420829e-01 -3.47942799e-01
-4.79950100e-01 -2.69898046e-02 -8.24452579e-01 4.46702719e-01
1.06819355e+00 -5.47689617e-01 6.43571675e-01 1.15082252e+00
4.48853433e-01 -3.86095941e-01 -1.36209297e+00 7.94364572e-01
-5.26228845e-01 -9.18492496e-01 -1.84697360e-01 3.51769686e-01
7.78787673e-01 -2.32444495e-01 2.90241241e-01 8.70379865e-01
9.69048023e-01 -1.01527321e+00 1.03640044e+00 9.17937160e-01
4.55804199e-01 -1.23625457e+00 5.45502663e-01 7.65544355e-01
-6.83620334e-01 -6.09924495e-01 -1.93243653e-01 -1.14848442e-01
1.45397022e-01 4.95088011e-01 -3.83750170e-01 5.78435361e-01
1.86232612e-01 5.86184680e-01 2.07577944e-01 6.48912728e-01
-4.32485759e-01 4.13747579e-01 -3.15282673e-01 -2.72600859e-01
6.93434298e-01 -6.46459162e-01 6.81559384e-01 9.27081466e-01
1.41656443e-01 -1.19107410e-01 5.07283270e-01 9.25284505e-01
1.54570937e-01 7.84210339e-02 -7.08857536e-01 -8.71483609e-02
1.54840499e-01 8.27852249e-01 -2.49013796e-01 -3.08813304e-01
-1.25501826e-01 4.77538258e-01 4.20661241e-01 5.31795681e-01
-9.14191544e-01 -1.13010816e-01 9.59517777e-01 -1.82545424e-01
1.05945706e-01 -2.18023077e-01 1.51514634e-01 -9.20954227e-01
8.90236497e-02 -8.87353063e-01 6.17419422e-01 -1.26178369e-01
-1.32109892e+00 1.37396589e-01 5.39331853e-01 -8.87516916e-01
-9.44635987e-01 -5.56288600e-01 -3.75550836e-01 9.16779518e-01
-1.58829558e+00 -4.33554083e-01 6.38400972e-01 6.44241035e-01
3.22961599e-01 2.38318946e-02 7.23002315e-01 -7.90830851e-02
-6.47418201e-01 2.04682067e-01 6.05605245e-01 -3.42164099e-01
2.98418075e-01 -1.85561574e+00 -2.36867324e-01 5.45600474e-01
-3.46060604e-01 1.71380803e-01 9.51718271e-01 -3.95575851e-01
-1.43142021e+00 -9.21625257e-01 2.73984581e-01 -1.70152709e-01
9.55144584e-01 -2.14184076e-02 -5.96135616e-01 6.83465421e-01
7.89469332e-02 -1.24154553e-01 1.54402837e-01 -2.64521893e-02
1.19141713e-01 -8.76861252e-03 -1.35141504e+00 6.92667663e-01
8.70958686e-01 -2.85763562e-01 -6.13131523e-01 2.01683283e-01
5.96856654e-01 -2.37609223e-01 -7.14883804e-01 1.57313615e-01
4.42999303e-01 -9.45324242e-01 7.37189472e-01 -9.11195755e-01
2.47556135e-01 -5.90228960e-02 -2.16438681e-01 -1.67533970e+00
1.72763057e-02 -1.07790744e+00 -4.24896270e-01 9.45062637e-01
1.39071107e-01 -8.34093869e-01 5.50215721e-01 4.89677161e-01
-1.03586139e-02 -1.08642519e+00 -1.11345685e+00 -1.17736220e+00
6.09309614e-01 -5.39376557e-01 5.31583667e-01 2.56669611e-01
6.36525676e-02 -1.92606002e-02 -4.31337774e-01 -1.44116729e-01
8.96087348e-01 1.96504697e-01 3.96032006e-01 -9.34655070e-01
-6.95168376e-01 -7.93377042e-01 1.59402728e-01 -1.12484622e+00
6.27699196e-01 -7.60460317e-01 2.74714142e-01 -1.31313956e+00
5.74835297e-03 -4.62193429e-01 -3.24184567e-01 2.10678026e-01
1.85484737e-01 -8.60233724e-01 3.81027132e-01 -2.11433545e-01
-7.48952329e-01 9.58956897e-01 1.24639785e+00 6.70784265e-02
-3.79819632e-01 5.01211941e-01 -4.69140798e-01 7.63007104e-01
7.59219825e-01 -4.71537381e-01 -6.66142881e-01 -8.57097879e-02
4.87814218e-01 7.02942550e-01 2.00249165e-01 -5.58470309e-01
4.15022373e-02 -8.85090172e-01 -2.37535924e-01 -5.21824002e-01
2.17595194e-02 -5.16310275e-01 -4.31655981e-02 6.97057724e-01
-8.42738450e-01 1.41970351e-01 -1.13706715e-01 8.78829181e-01
-1.97699610e-02 -5.76325059e-01 9.10535038e-01 -1.93228588e-01
-3.98376465e-01 4.58493233e-01 -6.62556529e-01 3.89073670e-01
1.44896472e+00 4.37112153e-01 4.08073841e-03 -4.92045552e-01
-1.03595698e+00 5.85823357e-01 1.97207004e-01 -6.35002926e-02
2.42379859e-01 -1.19482386e+00 -6.67793274e-01 -2.66872942e-01
-2.80396491e-01 7.18120858e-02 2.96690732e-01 7.98418403e-01
-1.29108846e-01 4.30059373e-01 -1.98428869e-01 -2.41311997e-01
-5.96791804e-01 8.44583094e-01 4.76456851e-01 -9.37191546e-01
-4.71444458e-01 4.89167780e-01 3.56220342e-02 -2.81264603e-01
6.36629403e-01 -5.83729148e-01 -1.21996179e-01 4.62050349e-01
5.87836921e-01 5.42419076e-01 -2.80545443e-01 -1.55173331e-01
2.58254707e-02 1.33389875e-01 2.12648585e-01 -6.84125245e-01
1.31939805e+00 6.96788263e-03 -1.93188787e-02 7.16719747e-01
7.30063498e-01 -3.05605501e-01 -1.73931956e+00 -3.17279547e-01
3.36415052e-01 -2.59245902e-01 -1.68323636e-01 -8.66900086e-01
-6.67924285e-01 7.72378504e-01 3.42509896e-01 4.93736237e-01
9.72235382e-01 -1.56905994e-01 3.01583439e-01 5.35770595e-01
5.90714157e-01 -1.57411003e+00 -2.83745490e-02 7.85778522e-01
8.96768093e-01 -8.38548183e-01 -2.38244936e-01 5.14814377e-01
-9.46163177e-01 9.18240845e-01 9.08953622e-02 -2.97500134e-01
6.34589791e-01 2.77601808e-01 -4.69216198e-01 2.78485537e-01
-1.03828359e+00 -5.31445682e-01 -2.92794287e-01 7.20227778e-01
-1.27162844e-01 3.97418261e-01 -3.93967837e-01 6.53760374e-01
-8.33225697e-02 -7.10055158e-02 4.45743412e-01 1.11489642e+00
-5.39424837e-01 -1.39501107e+00 -2.63328046e-01 4.19494838e-01
-5.56939960e-01 2.28295729e-01 1.32306010e-01 6.65429771e-01
-2.32909322e-01 7.26608813e-01 -3.43215582e-03 1.76166639e-01
5.05730331e-01 2.09164888e-01 6.39596403e-01 -6.49721086e-01
-4.15561050e-01 -1.51685491e-01 -9.57413763e-02 -6.99384034e-01
-3.06115627e-01 -1.00687277e+00 -1.12908554e+00 -3.81471694e-01
2.99932420e-01 2.35211208e-01 5.98890424e-01 1.07675958e+00
-4.89578210e-02 5.05164087e-01 8.20551932e-01 -6.84298992e-01
-1.80338037e+00 -7.20633924e-01 -1.00363660e+00 1.24502011e-01
4.70127851e-01 -8.16578448e-01 -3.51406574e-01 -5.07136405e-01] | [4.247817039489746, 2.556037187576294] |
54708cce-9103-4e17-a825-ee30745a429a | modeling-syntactic-semantic-dependency | null | null | https://aclanthology.org/2022.acl-long.548 | https://aclanthology.org/2022.acl-long.548.pdf | Modeling Syntactic-Semantic Dependency Correlations in Semantic Role Labeling Using Mixture Models | In this paper, we propose a mixture model-based end-to-end method to model the syntactic-semantic dependency correlation in Semantic Role Labeling (SRL). Semantic dependencies in SRL are modeled as a distribution over semantic dependency labels conditioned on a predicate and an argument word.The semantic label distribution varies depending on Shortest Syntactic Dependency Path (SSDP) hop patterns.We target the variation of semantic label distributions using a mixture model, separately estimating semantic label distributions for different hop patterns and probabilistically clustering hop patterns with similar semantic label distributions.Experiments show that the proposed method successfully learns a cluster assignment reflecting the variation of semantic label distributions.Modeling the variation improves performance in predicting short distance semantic dependencies, in addition to the improvement on long distance semantic dependencies that previous syntax-aware methods have achieved.The proposed method achieves a small but statistically significant improvement over baseline methods in English, German, and Spanish and obtains competitive performance with state-of-the-art methods in English. | ['Yusuke Miyao', 'Xiangheng He', 'Junjie Chen'] | null | null | null | null | acl-2022-5 | ['semantic-role-labeling'] | ['natural-language-processing'] | [-2.49706302e-03 3.45730245e-01 -6.80363834e-01 -1.01546836e+00
-1.03197026e+00 -8.21814060e-01 4.93180513e-01 3.38981360e-01
-5.62604666e-01 4.16939199e-01 7.46028543e-01 -8.36984292e-02
-1.33998841e-01 -2.98000574e-01 -3.75674903e-01 -6.07928991e-01
4.59059887e-02 1.10818219e+00 5.67277431e-01 -1.25292823e-01
2.62264878e-01 1.57171506e-02 -1.37782037e+00 6.87945843e-01
7.30927885e-01 4.59240973e-01 5.68637073e-01 4.00372535e-01
-5.50316334e-01 1.02022243e+00 -7.40496397e-01 -4.45747226e-02
-3.34502250e-01 -4.04025674e-01 -1.23500717e+00 -2.93672383e-01
1.70941949e-01 5.50282598e-01 2.19879504e-02 1.00538421e+00
2.88716137e-01 2.39183962e-01 9.42456067e-01 -1.12401009e+00
-5.26429296e-01 1.21507204e+00 -6.31733954e-01 7.26458877e-02
1.96753681e-01 -8.18593442e-01 1.64892614e+00 -7.65586555e-01
5.82991004e-01 1.78837609e+00 3.55640322e-01 7.38328934e-01
-1.33569741e+00 -5.36102116e-01 6.47934794e-01 2.38070816e-01
-1.27307928e+00 -8.63634869e-02 8.51062596e-01 -3.77921939e-01
1.16382599e+00 -2.43458048e-01 -4.39462960e-02 1.01484239e+00
-3.14386398e-01 1.00246572e+00 9.79995489e-01 -6.79353714e-01
4.06915963e-01 -1.03726462e-02 7.04297602e-01 3.64375234e-01
-2.19430313e-01 -5.65572917e-01 -5.80444872e-01 -3.13121945e-01
5.38424738e-02 -4.20345485e-01 1.29141748e-01 -2.77868450e-01
-9.51114535e-01 9.60706055e-01 2.21631050e-01 4.66932476e-01
1.86810777e-01 1.15522698e-01 4.51812208e-01 4.98051122e-02
6.66286409e-01 -8.90332181e-03 -1.04993260e+00 -1.89525466e-02
-6.14513695e-01 8.47523510e-02 8.98972869e-01 1.07648706e+00
5.61950028e-01 -5.10573328e-01 -6.53239116e-02 1.55047953e+00
7.27967918e-01 3.91669303e-01 4.03654486e-01 -1.15811276e+00
5.70762634e-01 4.90277141e-01 1.11181252e-01 -4.75906670e-01
-7.02396691e-01 1.36012971e-01 3.21641266e-02 -1.03435852e-01
5.66439927e-01 1.65706649e-01 -9.25769448e-01 2.12640238e+00
3.30768645e-01 -2.60055680e-02 2.08000228e-01 6.27034247e-01
6.82075918e-01 6.23851955e-01 9.62089598e-01 2.04815026e-02
1.68079650e+00 -1.20743799e+00 -6.09736741e-01 -5.06350398e-01
1.18544710e+00 -6.90338671e-01 1.29362297e+00 -3.35181504e-02
-6.92199230e-01 -5.64106107e-01 -6.74520135e-01 -2.44832769e-01
-2.92200595e-01 1.59964547e-01 5.62532306e-01 5.41455984e-01
-9.84922230e-01 4.63600338e-01 -9.36091959e-01 -4.13178444e-01
3.04813296e-01 2.61558175e-01 -2.65637368e-01 -2.66952544e-01
-1.34100771e+00 6.45030975e-01 7.41152525e-01 -5.52659214e-01
-6.40297532e-01 -7.08161175e-01 -1.15031707e+00 2.68868934e-02
2.78897822e-01 -2.66575813e-01 1.35250556e+00 -9.01624501e-01
-1.26169133e+00 1.28760779e+00 -6.36431217e-01 -9.44759771e-02
-1.21785946e-01 -2.15753973e-01 -4.77378428e-01 -5.80356419e-02
6.41240954e-01 9.99402404e-01 5.26895940e-01 -1.46427882e+00
-1.07275867e+00 -6.04718864e-01 -1.16313279e-01 5.74421167e-01
3.35729830e-02 6.14277899e-01 -4.69781786e-01 -5.83671510e-01
2.30974153e-01 -1.14447296e+00 -1.43650845e-01 -7.03955770e-01
-3.44523460e-01 -9.56152141e-01 7.45958745e-01 -6.23080313e-01
1.12481987e+00 -2.27830195e+00 2.55531788e-01 -7.75079355e-02
-3.06421876e-01 -1.05273068e-01 -2.02395409e-01 2.63601869e-01
-2.25537881e-01 2.17778701e-02 -5.88274717e-01 -6.92702532e-01
1.05393946e-01 6.20660007e-01 -1.22597590e-01 3.59932244e-01
-3.94321512e-03 6.10597193e-01 -1.27461183e+00 -4.92820442e-01
-9.81136858e-02 4.55745935e-01 -4.30532128e-01 1.52475178e-01
-6.74595118e-01 6.61323071e-01 -4.68116939e-01 5.60891032e-01
6.06058419e-01 -3.34455997e-01 9.20337200e-01 -3.08572222e-02
2.43021548e-01 8.43354106e-01 -9.30500448e-01 2.18564391e+00
-6.64945662e-01 2.26779267e-01 4.63837758e-02 -1.23738420e+00
9.83778417e-01 2.13752314e-01 3.98781329e-01 -3.27783495e-01
-1.58682223e-02 3.36837918e-01 7.27578923e-02 -1.03825815e-01
2.23583624e-01 -3.18736702e-01 -7.04182565e-01 4.52054650e-01
2.55541414e-01 1.03801817e-01 2.46402606e-01 2.24204361e-01
9.69255626e-01 3.80233228e-01 9.93850827e-02 -7.21183956e-01
5.82356274e-01 1.36429653e-01 5.57619870e-01 3.89550835e-01
-1.92709833e-01 4.57276344e-01 7.53297448e-01 5.97369745e-02
-6.35042191e-01 -1.25977767e+00 -3.33541185e-01 1.90958977e+00
4.28700417e-01 -5.54903030e-01 -7.45710552e-01 -1.48393941e+00
3.57998237e-02 1.21570599e+00 -5.20767093e-01 2.26248149e-02
-8.01507771e-01 -8.22805166e-01 6.58171892e-01 8.66783619e-01
3.02739173e-01 -1.28775096e+00 7.94998929e-02 2.66856045e-01
-3.70371938e-01 -1.58866298e+00 -2.59454131e-01 5.82157493e-01
-7.04943836e-01 -1.08386278e+00 -2.83798456e-01 -1.24952519e+00
5.63561559e-01 -3.13949957e-03 1.48566079e+00 -3.92740786e-01
3.08443550e-02 2.48922393e-01 -6.50411546e-01 -2.29201034e-01
-5.18701613e-01 1.62119895e-01 -1.96200177e-01 -3.86454135e-01
9.00445640e-01 -3.36999834e-01 -3.54173452e-01 4.34051424e-01
-5.72869837e-01 -4.39556211e-01 2.30383705e-02 6.65816963e-01
7.51808822e-01 2.05685034e-01 6.06651664e-01 -1.41809142e+00
2.18672365e-01 -8.12499523e-01 -2.73285925e-01 2.61695117e-01
-5.59853911e-01 3.74290198e-01 5.47638893e-01 -1.63191959e-01
-1.55455875e+00 1.03092447e-01 -1.93761334e-01 1.20369405e-01
-6.49547398e-01 2.16604576e-01 -6.22459888e-01 7.09298372e-01
5.10771990e-01 -3.37971270e-01 -2.61026621e-01 -8.15424860e-01
6.84541941e-01 5.79752445e-01 2.86556721e-01 -1.01029170e+00
1.61735713e-01 6.56039655e-01 -1.27687767e-01 -6.07017279e-01
-1.59470642e+00 -9.36319649e-01 -9.31746304e-01 3.45675200e-01
1.34334898e+00 -1.22050810e+00 -8.67821276e-02 3.73349160e-01
-1.23901772e+00 -5.67216754e-01 -1.17939152e-01 4.14885134e-01
-5.41233718e-01 2.64837712e-01 -8.30168903e-01 -2.94389278e-01
1.96282923e-01 -9.73809242e-01 1.56915545e+00 -5.34202345e-02
-5.13448417e-01 -1.59967113e+00 2.29473680e-01 6.43123984e-01
-1.46045029e-01 -1.93931505e-01 1.25090826e+00 -1.03643942e+00
-2.86591817e-02 3.54397923e-01 -2.50278145e-01 5.09638488e-01
2.74456173e-01 -4.55679804e-01 -8.68126094e-01 -9.42573249e-02
-2.80608326e-01 -2.48871803e-01 1.03574431e+00 6.69554353e-01
9.73048568e-01 2.62034386e-01 -7.12921143e-01 2.26243153e-01
1.29707325e+00 6.30929172e-02 1.99488193e-01 2.43135557e-01
1.04921103e+00 1.14016378e+00 1.05747509e+00 9.71670970e-02
7.09206223e-01 7.46381581e-01 1.24742873e-01 2.00062245e-01
-3.32740128e-01 -3.90583098e-01 5.23944438e-01 4.97846484e-01
4.25201029e-01 -2.22489819e-01 -1.03035212e+00 6.06872499e-01
-2.13077164e+00 -5.25776148e-01 -3.25664222e-01 1.80908859e+00
8.39251578e-01 6.04219660e-02 2.37482235e-01 -4.13405001e-02
1.03301215e+00 3.41444969e-01 -3.27142537e-01 -3.50902498e-01
-1.15312278e-01 2.28733003e-01 5.65614820e-01 8.30450237e-01
-1.16361308e+00 1.64889050e+00 5.94739246e+00 8.99660289e-01
-5.62328696e-01 5.55536866e-01 4.39945132e-01 2.53809243e-01
-4.43503141e-01 2.23767057e-01 -1.31454611e+00 5.66269040e-01
1.02912605e+00 2.38529131e-01 -9.79635641e-02 1.02106988e+00
-9.79376659e-02 -1.50365204e-01 -1.03361762e+00 6.71522558e-01
1.62695140e-01 -7.51734257e-01 -1.84767514e-01 -2.17300266e-01
7.36436665e-01 2.91817248e-01 -2.63049096e-01 3.33432734e-01
9.20549572e-01 -1.00250828e+00 7.12764502e-01 -1.93229735e-01
6.30050182e-01 -8.62694263e-01 6.39020920e-01 3.28605622e-01
-1.29967034e+00 -1.42789453e-01 -4.50403154e-01 6.58474192e-02
3.48080307e-01 5.95157266e-01 -6.11415863e-01 2.57844537e-01
7.90678084e-01 9.80429888e-01 -2.82026768e-01 6.52422234e-02
-1.01256347e+00 9.93309855e-01 -1.83229104e-01 4.65914384e-02
2.71355987e-01 -7.47332573e-02 3.61491174e-01 1.61615026e+00
-1.73394158e-01 -8.64448100e-02 4.46511835e-01 6.36334121e-01
7.05254376e-02 2.85023808e-01 -3.26205522e-01 7.33483955e-02
8.90356064e-01 1.02473736e+00 -1.24652386e+00 -2.62755305e-01
-5.76357782e-01 1.20939207e+00 5.90497613e-01 3.01089644e-01
-8.93594503e-01 1.00898363e-01 8.71793747e-01 -2.58784276e-02
3.38919014e-01 -1.67757452e-01 -3.92329186e-01 -8.61771345e-01
-1.02244548e-01 -4.13154483e-01 1.00895119e+00 -2.71709353e-01
-1.47765648e+00 3.62675190e-01 4.37050521e-01 -6.96815312e-01
-7.45274574e-02 -7.91079283e-01 -2.68730253e-01 8.87419283e-01
-1.55682874e+00 -1.36130428e+00 1.33223832e-01 5.03233314e-01
8.69279087e-01 -1.84004188e-01 1.01323688e+00 1.07421331e-01
-1.03676833e-01 4.43898380e-01 -7.76431290e-03 1.13184310e-01
9.06946301e-01 -1.67402148e+00 3.66765767e-01 3.95278811e-01
4.00113344e-01 3.30962509e-01 4.21746343e-01 -6.03635073e-01
-4.36750770e-01 -1.18342662e+00 1.34878957e+00 -8.51687074e-01
6.48016393e-01 -4.39629853e-01 -7.83026397e-01 8.07740629e-01
-9.29549336e-02 4.70194146e-02 1.00407457e+00 6.00191891e-01
-8.50008965e-01 3.26082826e-01 -1.06268084e+00 2.12386817e-01
1.37527692e+00 -6.38088346e-01 -8.49886358e-01 3.28859538e-01
9.22079146e-01 2.70389114e-02 -6.94315970e-01 4.72492963e-01
1.36445105e-01 -9.03090060e-01 9.21382904e-01 -5.37581325e-01
3.57833892e-01 -4.77328420e-01 -5.67987442e-01 -1.49637067e+00
-4.94500279e-01 9.58170816e-02 3.07571650e-01 1.24192929e+00
8.44704390e-01 -4.07884002e-01 9.37615275e-01 4.82426256e-01
-3.01714480e-01 -3.11590701e-01 -7.57841706e-01 -8.29744518e-01
3.78656387e-01 -5.12510657e-01 2.51320660e-01 1.17673552e+00
1.33889556e-01 8.93301666e-01 2.36930832e-01 4.80169535e-01
7.95275390e-01 1.53688833e-01 8.34380016e-02 -1.24857128e+00
-5.52431583e-01 -3.79058331e-01 -2.62879908e-01 -1.06026483e+00
1.36509979e+00 -1.39830148e+00 2.13466123e-01 -1.65968692e+00
2.95832425e-01 -8.84359062e-01 -3.93907040e-01 6.26127422e-01
-5.63152492e-01 -6.00711405e-02 -3.50090563e-02 2.50231862e-01
-9.76724088e-01 4.32407141e-01 7.14084208e-01 8.33949912e-03
-2.03936175e-01 -1.86171278e-01 -6.79540396e-01 1.02513194e+00
8.52021992e-01 -9.41851318e-01 -8.21272433e-01 -4.81624633e-01
8.82563218e-02 -2.02438861e-01 -2.02210233e-01 -6.40140891e-01
-2.02106297e-01 4.42497618e-02 -5.82147986e-02 -3.63990158e-01
1.54246494e-01 -6.99253500e-01 -3.78383934e-01 6.95309862e-02
-7.40564466e-01 -3.28429937e-01 -9.76332575e-02 7.13463187e-01
-4.42156434e-01 -3.77336919e-01 9.66941714e-01 -1.25939935e-01
-1.07202423e+00 -1.33289427e-01 -2.95988798e-01 5.46700180e-01
9.74254787e-01 2.88531423e-01 -9.17566940e-02 1.91781987e-02
-1.10653186e+00 3.56665045e-01 3.18625003e-01 8.44609737e-01
1.16764545e-01 -1.13551307e+00 -7.63596594e-01 -1.44799858e-01
4.06291038e-01 1.75550669e-01 8.52889046e-02 1.88973144e-01
2.83526499e-02 2.68595248e-01 3.88852388e-01 -6.91421568e-01
-1.39537168e+00 3.09968561e-01 9.27242637e-02 -5.58335483e-01
-4.19016033e-01 1.41038990e+00 4.32477534e-01 -1.12084115e+00
2.43058875e-01 3.01878583e-02 -5.20088553e-01 7.90534168e-02
2.56738663e-01 2.41033480e-01 -1.50773600e-01 -9.31358457e-01
-6.51988208e-01 6.50555909e-01 -1.53231740e-01 -2.27612257e-01
9.91518140e-01 -2.37831265e-01 -3.39923054e-01 7.19932854e-01
1.34943366e+00 2.19842985e-01 -1.09944546e+00 -2.61240423e-01
6.88919604e-01 -1.66503772e-01 -2.22116619e-01 -1.02077377e+00
-6.65853739e-01 6.01293325e-01 2.31120050e-01 -8.86550173e-02
6.10241950e-01 1.14065433e+00 6.04788780e-01 -7.38734305e-02
4.07242626e-01 -1.26141787e+00 1.66280136e-01 9.41327155e-01
2.83634841e-01 -1.18546629e+00 -5.78391016e-01 -9.74768341e-01
-9.99061227e-01 6.20947301e-01 5.79384327e-01 -1.68154508e-01
1.11580884e+00 4.27011639e-01 3.83822143e-01 -2.70200610e-01
-5.67506254e-01 -3.31801206e-01 1.58209741e-01 8.33032131e-01
9.83073175e-01 5.03683031e-01 -5.39212584e-01 6.34707868e-01
-9.65135098e-02 -7.32054353e-01 -6.65741935e-02 8.18107486e-01
-5.27685583e-01 -1.81905127e+00 -1.07750744e-01 -1.38115406e-01
-6.95959747e-01 -4.47936088e-01 -2.78523505e-01 6.26038730e-01
9.50807780e-02 1.23689985e+00 3.94468576e-01 6.89134672e-02
3.16963702e-01 5.63231587e-01 5.47620833e-01 -1.19778955e+00
-8.03755149e-02 1.09045751e-01 6.21292651e-01 -6.17167830e-01
-6.50760233e-01 -6.96989357e-01 -2.25303793e+00 4.79417413e-01
9.93831083e-02 4.80937630e-01 8.19330692e-01 1.13355315e+00
2.77240515e-01 3.53825957e-01 5.23006439e-01 -4.68596399e-01
-3.86967272e-01 -1.08636105e+00 -9.35997367e-01 8.03128719e-01
-2.72324026e-01 -8.30204606e-01 -4.94714350e-01 1.26218364e-01] | [10.400146484375, 9.455375671386719] |
f4a04052-f912-4206-b19f-b07e9ddbf5e0 | boosting-neural-networks-to-decompile | 2301.00969 | null | https://arxiv.org/abs/2301.00969v1 | https://arxiv.org/pdf/2301.00969v1.pdf | Boosting Neural Networks to Decompile Optimized Binaries | Decompilation aims to transform a low-level program language (LPL) (eg., binary file) into its functionally-equivalent high-level program language (HPL) (e.g., C/C++). It is a core technology in software security, especially in vulnerability discovery and malware analysis. In recent years, with the successful application of neural machine translation (NMT) models in natural language processing (NLP), researchers have tried to build neural decompilers by borrowing the idea of NMT. They formulate the decompilation process as a translation problem between LPL and HPL, aiming to reduce the human cost required to develop decompilation tools and improve their generalizability. However, state-of-the-art learning-based decompilers do not cope well with compiler-optimized binaries. Since real-world binaries are mostly compiler-optimized, decompilers that do not consider optimized binaries have limited practical significance. In this paper, we propose a novel learning-based approach named NeurDP, that targets compiler-optimized binaries. NeurDP uses a graph neural network (GNN) model to convert LPL to an intermediate representation (IR), which bridges the gap between source code and optimized binary. We also design an Optimized Translation Unit (OTU) to split functions into smaller code fragments for better translation performance. Evaluation results on datasets containing various types of statements show that NeurDP can decompile optimized binaries with 45.21% higher accuracy than state-of-the-art neural decompilation frameworks. | ['Peiwei Hu', 'Kai Chen', 'Ruigang Liang', 'Ying Cao'] | 2023-01-03 | null | null | null | null | ['nmt'] | ['computer-code'] | [ 1.62044272e-01 -1.70914501e-01 -7.35408664e-01 -1.61806002e-01
-4.75415826e-01 -6.76286221e-01 2.55594522e-01 3.18639696e-01
-7.89391175e-02 4.12276715e-01 -6.90865517e-02 -1.21341085e+00
5.11312008e-01 -1.22875428e+00 -1.18141484e+00 -9.30235609e-02
-4.95190434e-02 3.64067793e-01 1.44787595e-01 -3.96641850e-01
2.14297727e-01 4.10716444e-01 -1.19638598e+00 3.36045027e-01
1.20828581e+00 5.69638133e-01 1.55954525e-01 6.07826650e-01
-3.10100466e-01 5.65501451e-01 -6.40514314e-01 -6.45496547e-01
3.98737311e-01 -4.71209615e-01 -7.49472201e-01 -6.32221937e-01
9.06351432e-02 -3.05110253e-02 -4.70564902e-01 1.73692942e+00
4.66869920e-02 -3.62245739e-01 3.58707458e-01 -1.36215985e+00
-9.26663697e-01 1.33665431e+00 -3.08390260e-01 1.12510167e-01
8.93633068e-02 2.60504246e-01 9.95793939e-01 -9.36645269e-02
4.33336377e-01 1.18098772e+00 7.38739967e-01 5.10223031e-01
-1.41848934e+00 -6.52059376e-01 -2.94997364e-01 5.38012147e-01
-1.17473388e+00 1.13040470e-01 7.06915498e-01 -4.63916987e-01
1.61512244e+00 3.18230629e-01 6.22293532e-01 9.35734391e-01
9.26562488e-01 5.69970250e-01 9.68824208e-01 -3.35323393e-01
1.58548117e-01 8.98108445e-03 5.69616437e-01 8.56139302e-01
4.60937828e-01 5.07739127e-01 -5.50400242e-02 -5.54791391e-02
1.87230572e-01 -7.90994056e-03 -3.41624916e-01 6.81921141e-04
-1.00737095e+00 1.05434287e+00 9.00970817e-01 4.44001555e-01
-8.13516006e-02 1.15945162e-02 9.97731388e-01 5.00120163e-01
1.67317525e-01 6.53503060e-01 -2.38035321e-01 -2.22591087e-01
-1.08900034e+00 4.47363444e-02 1.12048793e+00 6.37536883e-01
7.71867037e-01 5.89464962e-01 1.92384690e-01 4.27403092e-01
2.29426458e-01 4.15932536e-01 8.48768055e-01 -2.94038542e-02
7.40684092e-01 9.58892703e-01 -1.01953816e+00 -1.09570289e+00
-1.96555063e-01 -3.68357182e-01 -1.02166390e+00 1.48574993e-01
-1.12584792e-01 2.41225347e-01 -8.13971221e-01 1.52393866e+00
-9.85117257e-02 -1.40801042e-01 2.69818693e-01 6.77661955e-01
6.57332182e-01 1.03673470e+00 -2.68606812e-01 3.31696775e-03
1.13205063e+00 -9.89316940e-01 -1.15745895e-01 -4.66943830e-01
7.62638152e-01 -5.17211437e-01 1.12857640e+00 3.50187272e-01
-6.18989110e-01 -5.98839700e-01 -1.45990419e+00 1.70001328e-01
-6.81896269e-01 -1.15263507e-01 6.23548210e-01 8.33004534e-01
-9.30949032e-01 8.01879704e-01 -9.43305731e-01 -1.60796344e-01
1.29380360e-01 5.06170094e-01 -4.06533301e-01 2.36406773e-01
-1.24144685e+00 8.14100802e-01 1.18470871e+00 -2.16597453e-01
-1.28563273e+00 -6.25899792e-01 -1.07964051e+00 2.68816113e-01
3.25826585e-01 -3.37726921e-01 9.28641617e-01 -9.26397979e-01
-1.39172924e+00 5.69680154e-01 2.21302718e-01 -9.73309219e-01
-2.63663560e-01 9.53970850e-02 -4.11828041e-01 -2.74284005e-01
-1.38964623e-01 1.95473358e-01 1.04739344e+00 -8.92243564e-01
-1.90304518e-01 -1.67191908e-01 2.77888477e-01 -5.16706765e-01
-4.27690208e-01 3.14818799e-01 -4.90389131e-02 -6.15514457e-01
-3.54775071e-01 -1.04518306e+00 1.29103707e-02 -7.23117650e-01
-9.00789022e-01 -8.20538923e-02 9.77019429e-01 -1.27399063e+00
1.53676093e+00 -1.88670921e+00 7.54637480e-01 1.41444057e-01
4.65612888e-01 8.06372941e-01 -2.04624265e-01 2.59224683e-01
-6.26595438e-01 3.72924387e-01 -5.29239714e-01 -7.73337707e-02
2.59014726e-01 2.79844850e-01 -6.37526393e-01 3.59386772e-01
1.28613338e-02 1.31007242e+00 -6.90787315e-01 -4.13599849e-01
-7.55010452e-03 1.54396057e-01 -8.74296784e-01 3.17244917e-01
-6.61218584e-01 -3.91240977e-02 -1.09290749e-01 1.04640043e+00
5.50281703e-01 9.05364379e-02 1.13518968e-01 -2.12643147e-01
-1.88847467e-01 4.95824635e-01 -2.27601975e-01 1.27692413e+00
-7.64502287e-01 8.91830027e-01 -1.72687978e-01 -1.33011580e+00
9.41563427e-01 -1.24515496e-01 -1.30778179e-01 -5.77728271e-01
3.46511573e-01 2.39294767e-01 4.24321383e-01 -2.41768047e-01
4.43585545e-01 -1.36253476e-01 -3.98017704e-01 6.42360687e-01
4.03821841e-02 -3.37855667e-01 3.76192003e-01 -5.16175367e-02
1.11790514e+00 2.61618625e-02 7.12083638e-01 -7.07887411e-02
7.60106564e-01 1.81751162e-01 5.16410589e-01 4.44802821e-01
6.47479743e-02 -1.95924472e-03 8.40940237e-01 -3.49858284e-01
-1.13254380e+00 -8.99035275e-01 1.86361596e-01 9.23312306e-01
-3.22858453e-01 -8.87869775e-01 -1.20522439e+00 -8.91569853e-01
-2.09684119e-01 1.05493402e+00 -3.02826375e-01 -8.20928514e-01
-8.79012465e-01 -9.42333996e-01 1.13939714e+00 3.64990085e-01
6.21491313e-01 -1.15314603e+00 -2.55925715e-01 1.81118369e-01
-1.73753396e-01 -7.32634544e-01 -4.26636010e-01 4.59772110e-01
-7.31701851e-01 -9.45026457e-01 -5.12824059e-02 -7.10893452e-01
6.85597897e-01 -2.24241540e-01 1.08136737e+00 1.08337440e-01
-1.49271458e-01 -4.46116567e-01 -5.99851727e-01 -6.92018867e-02
-1.43164873e+00 5.87784767e-01 2.10398465e-01 -3.10121298e-01
3.48011941e-01 -6.45786583e-01 4.12263274e-01 2.66741798e-03
-1.29735768e+00 1.67653710e-01 7.30933249e-01 9.45394635e-01
4.47372168e-01 4.43136215e-01 -6.08162098e-02 -5.99790812e-01
8.02747905e-01 -4.60604429e-01 -1.14463687e+00 3.83417547e-01
-8.11687827e-01 4.70186472e-01 1.50874996e+00 -6.10107839e-01
-5.32814026e-01 -2.85281897e-01 -4.53101486e-01 -5.58049858e-01
8.56035203e-02 1.03756726e+00 -4.69206631e-01 -4.86887902e-01
9.62093234e-01 5.95033884e-01 2.04040520e-02 -3.58369201e-01
1.21825188e-01 7.45397687e-01 6.93887472e-01 -7.45896757e-01
1.28083539e+00 -3.09034228e-01 -4.64102179e-02 -4.20317888e-01
-4.81761575e-01 2.92725176e-01 -5.65474987e-01 1.92355514e-01
8.14472556e-01 -2.76073039e-01 -6.44445479e-01 5.56956768e-01
-1.40985298e+00 -4.63795155e-01 -1.23389550e-02 6.80152327e-02
-3.87736708e-01 6.60461962e-01 -7.87001312e-01 -5.13886772e-02
-7.33018577e-01 -1.80492902e+00 6.96615756e-01 2.99037665e-01
-5.48857190e-02 -8.94137263e-01 2.63062954e-01 1.59282342e-01
5.12459815e-01 1.46355644e-01 1.75272429e+00 -8.55583489e-01
-6.37784421e-01 -4.44411010e-01 -3.75622898e-01 8.26256573e-01
3.54476739e-03 2.38572191e-02 -3.08508068e-01 -4.64284569e-01
1.00874960e-01 -1.82966471e-01 8.14917624e-01 -1.35307953e-01
1.19142425e+00 -6.22137189e-01 -2.37319708e-01 1.26641202e+00
1.66126823e+00 2.23362267e-01 6.84224665e-01 6.56960189e-01
9.44866955e-01 1.16498224e-01 2.77726918e-01 -7.52451122e-02
1.99623436e-01 5.58801830e-01 8.80720317e-01 4.32783246e-01
-1.43695027e-01 -6.30985022e-01 1.10741425e+00 1.14697826e+00
1.21142872e-01 -1.03332065e-01 -1.34852147e+00 1.38457671e-01
-1.39148343e+00 -6.47480547e-01 5.17183915e-02 2.12263870e+00
9.96298909e-01 4.34184283e-01 -1.71541259e-01 1.77511171e-01
7.21913397e-01 1.73678458e-01 -6.00200653e-01 -9.81473267e-01
4.31897417e-02 5.08941770e-01 6.78689301e-01 5.74554086e-01
-1.05271256e+00 1.27969301e+00 5.12882519e+00 1.14330316e+00
-1.66175485e+00 3.04639578e-01 3.03201586e-01 4.67765927e-01
-2.17587769e-01 3.36764812e-01 -1.06591463e+00 5.94816267e-01
1.54948866e+00 -5.95917702e-01 1.07345295e+00 1.16909838e+00
-2.69361556e-01 4.55796570e-01 -1.11650348e+00 6.53616965e-01
2.91624665e-01 -1.43534148e+00 1.45343974e-01 1.02340437e-01
4.19102877e-01 3.02297622e-01 -2.77230274e-02 9.18189228e-01
2.40308717e-01 -1.17870128e+00 6.73100412e-01 4.97430079e-02
9.17599916e-01 -8.39241266e-01 8.99613857e-01 3.85951132e-01
-1.31612635e+00 -2.62336522e-01 -5.00290990e-01 1.07298277e-01
-2.18757950e-02 4.67595696e-01 -9.93052125e-01 7.81128705e-01
5.87491691e-01 7.98940420e-01 -9.31503832e-01 6.76014423e-01
-6.57445133e-01 7.13780820e-01 -9.80086774e-02 -1.78406894e-01
8.69618952e-02 -1.25119418e-01 7.75637805e-01 1.10769081e+00
3.12279105e-01 -5.65098226e-01 1.15696013e-01 1.37178481e+00
-1.42862841e-01 -1.06114700e-01 -5.99138081e-01 -9.07570243e-01
1.21533908e-01 1.19552886e+00 -6.30070210e-01 -2.32936993e-01
-1.00650743e-01 8.27972889e-01 4.29772943e-01 -6.76733404e-02
-1.30107343e+00 -7.03007758e-01 7.29926586e-01 -1.99897587e-03
1.29465476e-01 -3.13106358e-01 -2.07403466e-01 -1.40089679e+00
-1.24997109e-01 -1.54109406e+00 3.73555645e-02 -3.96581292e-01
-1.02453208e+00 1.22612023e+00 1.14581414e-01 -1.09491646e+00
-5.68592250e-01 -7.82887518e-01 -9.47225809e-01 7.24697709e-01
-1.26496017e+00 -1.30559731e+00 6.59914836e-02 2.75467455e-01
4.26400393e-01 -6.08394742e-01 7.04916716e-01 2.16660693e-01
-9.84264195e-01 8.66854429e-01 -7.97706656e-03 2.81948626e-01
2.57213950e-01 -1.00560641e+00 8.37587774e-01 1.42020822e+00
-1.40205085e-01 8.40327919e-01 6.55591607e-01 -9.37335670e-01
-1.77329028e+00 -1.49183965e+00 7.93846905e-01 4.68573906e-02
1.11683631e+00 -5.92615843e-01 -1.00586903e+00 7.06831038e-01
4.86685485e-02 -3.44797283e-01 5.08868814e-01 -2.94450402e-01
-7.85967529e-01 -6.25811741e-02 -9.57733035e-01 5.72396696e-01
5.99996567e-01 -9.01183248e-01 -6.89061582e-01 3.50529581e-01
1.29814792e+00 -4.17577088e-01 -8.63509774e-01 3.99647981e-01
1.16738141e-01 -8.79862905e-01 1.03370440e+00 -6.30717695e-01
8.06221545e-01 -3.62546742e-01 -3.62446934e-01 -1.46603727e+00
-1.68639764e-01 -4.67622906e-01 -6.08085811e-01 1.12876809e+00
2.54118502e-01 -8.94935131e-01 5.51882565e-01 -5.58114275e-02
-4.80412602e-01 -7.79622614e-01 -9.35322285e-01 -1.05463946e+00
3.75085384e-01 -5.71624279e-01 8.85237217e-01 7.61145830e-01
1.42171234e-01 2.31870562e-01 -6.59905598e-02 2.15770200e-01
5.44371367e-01 4.61682141e-01 6.62361145e-01 -9.70174253e-01
-9.26006973e-01 -8.10913742e-01 -5.81047118e-01 -6.58972561e-01
1.08801639e+00 -1.59308016e+00 -7.24758431e-02 -9.88519073e-01
1.64672479e-01 -1.65462807e-01 -2.50721984e-02 1.01116347e+00
6.70577064e-02 2.45982766e-01 3.35532606e-01 1.90717533e-01
1.71736866e-01 3.79075080e-01 6.23900115e-01 -7.05492556e-01
-9.91801247e-02 -1.89243212e-01 -6.14121735e-01 5.53628147e-01
9.70793664e-01 -7.35507727e-01 -5.93045764e-02 -2.55109489e-01
2.67351061e-01 2.80618425e-02 3.23188424e-01 -1.10146558e+00
1.21832184e-01 -2.44201913e-01 -3.05384934e-01 -4.29390788e-01
-8.51927921e-02 -4.12250400e-01 3.12412858e-01 9.86242592e-01
-1.36589422e-03 3.66046011e-01 3.58289242e-01 -3.24655022e-03
-4.31609780e-01 -7.07068920e-01 9.05811667e-01 -7.15544075e-02
-6.35576546e-01 3.08747053e-01 -3.07564765e-01 -2.63482571e-01
1.06761158e+00 1.28124610e-01 -6.59737825e-01 1.37947440e-01
-3.17631930e-01 -2.81470984e-01 7.43702173e-01 5.27106285e-01
6.92254901e-01 -1.01923490e+00 -5.61914325e-01 6.87795222e-01
-5.36509827e-02 -4.35562789e-01 -1.46004200e-01 7.57148325e-01
-1.07769430e+00 9.49475944e-01 -5.16466975e-01 -3.19549173e-01
-1.48755693e+00 9.04790521e-01 3.58704209e-01 -8.48231375e-01
-3.16503376e-01 5.43603420e-01 -9.01775286e-02 -8.86084378e-01
-6.45718351e-02 -4.81679678e-01 4.78571504e-02 -4.32817489e-01
4.59204704e-01 3.62155437e-02 1.87315702e-01 -6.82498693e-01
-2.49651298e-01 2.34480798e-01 -1.73268110e-01 4.95991409e-01
1.33664608e+00 7.44419158e-01 -1.10345519e+00 -9.48530361e-02
1.48896921e+00 -2.20124036e-01 -3.51226151e-01 3.46150808e-02
-7.06168860e-02 -2.84090996e-01 1.10941596e-01 -4.85876232e-01
-1.11506319e+00 1.02122188e+00 3.12938273e-01 2.68787146e-01
1.19055951e+00 -1.58187281e-02 1.26175201e+00 5.92061460e-01
5.31804204e-01 -4.10431087e-01 -2.19221443e-01 9.72819746e-01
6.21653318e-01 -8.05868745e-01 -2.03466967e-01 -1.95779368e-01
-3.99432937e-03 1.42391169e+00 7.97355294e-01 -1.30747676e-01
3.55811566e-01 5.47551095e-01 -5.79106987e-01 2.65695035e-01
-4.28381920e-01 8.01575035e-02 2.08332196e-01 7.61546791e-01
6.28452823e-02 2.66404092e-01 -1.96633801e-01 7.76875377e-01
-7.80433416e-01 -2.97476172e-01 7.01456130e-01 8.10377121e-01
-2.60887265e-01 -1.67369592e+00 -6.83465660e-01 5.67855299e-01
-2.82335252e-01 -6.16185188e-01 -3.19019735e-01 6.28228426e-01
2.09115177e-01 4.95000601e-01 -1.92304268e-01 -1.13821650e+00
-7.33482931e-03 3.41414101e-02 3.99384022e-01 -7.43745983e-01
-9.86363590e-01 -6.20722175e-01 -1.78393573e-01 -5.15398145e-01
2.94021696e-01 -1.80011481e-01 -1.14420092e+00 -6.97788954e-01
-2.14177504e-01 1.34324998e-01 8.96329701e-01 8.64175737e-01
1.87781706e-01 8.08988869e-01 4.80458736e-01 -1.07371271e+00
-7.13672638e-01 -6.47418380e-01 -1.27188593e-01 -1.16995893e-01
2.13747248e-01 -2.58399457e-01 -3.47351640e-01 -4.74788472e-02] | [7.172488689422607, 7.836678504943848] |
bcbecec0-fb4b-4950-9451-1d1352b30b5a | fairness-in-streaming-submodular-maximization | 2010.07431 | null | https://arxiv.org/abs/2010.07431v2 | https://arxiv.org/pdf/2010.07431v2.pdf | Fairness in Streaming Submodular Maximization: Algorithms and Hardness | Submodular maximization has become established as the method of choice for the task of selecting representative and diverse summaries of data. However, if datapoints have sensitive attributes such as gender or age, such machine learning algorithms, left unchecked, are known to exhibit bias: under- or over-representation of particular groups. This has made the design of fair machine learning algorithms increasingly important. In this work we address the question: Is it possible to create fair summaries for massive datasets? To this end, we develop the first streaming approximation algorithms for submodular maximization under fairness constraints, for both monotone and non-monotone functions. We validate our findings empirically on exemplar-based clustering, movie recommendation, DPP-based summarization, and maximum coverage in social networks, showing that fairness constraints do not significantly impact utility. | ['Jakub Tarnawski', 'Jakab Tardos', 'Ashkan Norouzi-Fard', 'Slobodan Mitrović', 'Marwa El Halabi'] | 2020-10-14 | null | http://proceedings.neurips.cc/paper/2020/hash/9d752cb08ef466fc480fba981cfa44a1-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/9d752cb08ef466fc480fba981cfa44a1-Paper.pdf | neurips-2020-12 | ['movie-recommendation'] | ['miscellaneous'] | [ 2.00876176e-01 5.13923049e-01 -9.33801830e-01 -4.98136491e-01
-7.62803555e-01 -4.72814381e-01 3.00849766e-01 6.89773381e-01
-3.44804943e-01 1.16522670e+00 6.00349307e-01 1.19054061e-03
-4.85041022e-01 -7.18528330e-01 -5.87708533e-01 -5.99178135e-01
-3.70229721e-01 6.32349908e-01 -3.95883083e-01 2.00362783e-02
4.04697895e-01 1.46082580e-01 -1.32489192e+00 4.02217284e-02
1.21138024e+00 6.78826809e-01 -2.77116627e-01 4.82297599e-01
1.03012145e-01 6.86688960e-01 -5.19922316e-01 -8.37721646e-01
3.16206098e-01 -3.47646922e-01 -6.00143552e-01 1.51214018e-01
6.09822452e-01 -5.85811734e-01 -1.99672282e-01 1.00890410e+00
6.68000162e-01 3.56927693e-01 9.50313330e-01 -1.59620881e+00
-3.52088451e-01 1.28764558e+00 -8.46202850e-01 6.78677857e-02
3.05585414e-01 -3.13726425e-01 1.63995755e+00 -3.51952106e-01
7.44242370e-01 1.27277672e+00 3.87800485e-01 5.54477990e-01
-1.27718115e+00 -3.82388443e-01 2.41245568e-01 -1.55071225e-02
-9.45957363e-01 -6.48548305e-01 5.72394311e-01 -4.71457504e-02
2.24578395e-01 8.99125457e-01 5.50752223e-01 7.55757451e-01
-4.97006029e-02 1.09126222e+00 6.23225570e-01 -7.05714226e-02
5.91937006e-01 2.04239264e-01 8.74175280e-02 1.65173560e-01
1.03535783e+00 -3.82245123e-01 -7.78483450e-01 -8.03568721e-01
8.88879150e-02 9.81532112e-02 -2.36392304e-01 -5.03360331e-01
-8.67301643e-01 1.24012232e+00 2.76353639e-02 -2.87134707e-01
-5.61982155e-01 1.67826906e-01 3.59123498e-01 3.09882462e-01
7.69809723e-01 7.02332973e-01 -2.56551951e-01 -1.60671607e-01
-1.43610132e+00 7.78098166e-01 1.05172431e+00 9.06309724e-01
3.56998742e-01 -1.15279123e-01 -4.60092664e-01 8.78954887e-01
-6.21475391e-02 3.89984071e-01 1.13972671e-01 -1.52075052e+00
4.56870973e-01 4.48865354e-01 2.31270805e-01 -1.12473202e+00
-3.83061588e-01 -3.82918566e-01 -9.73789454e-01 -3.42696697e-01
5.57184339e-01 -5.90417445e-01 -1.81978121e-01 1.93293715e+00
4.57341969e-01 -3.53375852e-01 -1.29441619e-01 7.63251841e-01
5.69214344e-01 5.73887110e-01 -3.80035825e-02 -9.69275117e-01
7.50670850e-01 -4.50735658e-01 -6.80761456e-01 3.28946151e-02
5.19591510e-01 -2.67390847e-01 6.25142515e-01 5.25161088e-01
-1.50611174e+00 2.83508927e-01 -6.87971830e-01 -9.51434392e-03
3.09359550e-01 -4.98418361e-01 8.93411517e-01 7.87151158e-01
-7.79240072e-01 8.08385313e-01 -3.06306779e-01 -4.42973107e-01
9.58935559e-01 2.30696008e-01 1.17558902e-02 -2.55356103e-01
-8.60939264e-01 4.38002765e-01 -8.73751100e-03 -5.74553430e-01
-5.98331153e-01 -1.11466467e+00 -5.05692661e-01 4.03632045e-01
5.20790458e-01 -9.64675605e-01 9.06317949e-01 -9.54024315e-01
-9.32267904e-01 8.12241793e-01 -5.93232065e-02 -6.55289114e-01
8.79305780e-01 1.98307028e-03 1.32079706e-01 9.75090861e-02
1.34350151e-01 5.22419155e-01 7.69822657e-01 -1.21601760e+00
-7.92035103e-01 -6.60699069e-01 2.06403136e-01 4.84505206e-01
-8.11301291e-01 2.19376497e-02 1.34907112e-01 -4.95924234e-01
-2.46478081e-01 -6.27994716e-01 -6.07134998e-01 -1.77016765e-01
-6.79126441e-01 -2.45703399e-01 1.08921997e-01 -6.24951243e-01
1.54188621e+00 -1.80938649e+00 2.33592823e-01 2.92389899e-01
3.17085534e-01 -4.87223685e-01 4.41196226e-02 4.94235009e-01
6.01871550e-01 2.70526946e-01 -2.75169522e-01 -3.96573722e-01
2.75569469e-01 2.10339334e-02 -9.13190991e-02 7.86244690e-01
-3.98854703e-01 5.44268310e-01 -8.30377579e-01 -6.83943510e-01
-4.93689984e-01 -2.73480594e-01 -8.58295798e-01 -2.03050729e-02
-4.61715877e-01 -9.87430438e-02 -4.27226663e-01 5.85835159e-01
7.19618917e-01 -2.00560972e-01 4.15051699e-01 3.68570179e-01
1.04264081e-01 -9.65885702e-04 -1.12758291e+00 1.35934949e+00
-2.75814887e-02 4.72386956e-01 4.01268363e-01 -1.07858443e+00
6.55490100e-01 1.41180024e-01 8.47643971e-01 -2.06568286e-01
1.12078786e-01 1.11635610e-01 -1.34727806e-01 -4.11836207e-01
9.54533041e-01 -2.45706618e-01 -2.71159828e-01 9.10227835e-01
-4.35028344e-01 -1.29175022e-01 4.63041931e-01 7.03089118e-01
9.38887417e-01 -7.32841671e-01 3.67821842e-01 -5.91679156e-01
-8.58851299e-02 1.85427740e-01 7.17875659e-01 8.80261242e-01
1.47232534e-02 8.22365522e-01 9.63744819e-01 -4.61231358e-02
-1.11365354e+00 -9.25953805e-01 -2.52549559e-01 1.31669438e+00
7.07095787e-02 -2.02680185e-01 -7.34316468e-01 -5.22607684e-01
5.56485057e-01 9.30067420e-01 -5.14358401e-01 -1.42645061e-01
-1.08116962e-01 -1.10736656e+00 1.89621717e-01 1.86937861e-02
-7.37774372e-02 -4.80201781e-01 -5.71614683e-01 -3.86869200e-02
-2.29619697e-01 -8.17110837e-01 -8.50647628e-01 -3.39935631e-01
-1.12624824e+00 -9.20156240e-01 -9.98195231e-01 -2.50106245e-01
7.77526140e-01 1.71569183e-01 1.31072628e+00 -8.01752359e-02
-9.22781378e-02 7.07447469e-01 -1.39905870e-01 -5.83726287e-01
-1.36545703e-01 3.71442616e-01 1.02277733e-01 1.36995390e-01
6.29558712e-02 -7.85506845e-01 -8.05252731e-01 -5.86797558e-02
-9.31700349e-01 -2.27273703e-01 1.25623018e-01 4.10828710e-01
3.20361942e-01 -1.11359403e-01 1.22939610e+00 -1.30713797e+00
1.12156188e+00 -9.84702766e-01 -2.76887894e-01 2.38203034e-01
-7.54474998e-01 -1.68534815e-01 4.58204687e-01 -3.88172597e-01
-8.80483150e-01 -3.60672474e-01 4.41321820e-01 1.01792002e-02
3.50540608e-01 6.02356493e-01 -2.68751651e-01 4.53826547e-01
7.32960343e-01 -5.68298586e-02 1.32978812e-01 -2.04204097e-01
4.48919773e-01 8.63724470e-01 1.46529153e-01 -6.70121789e-01
3.35506082e-01 7.08313942e-01 2.69476146e-01 -9.99468982e-01
-8.12514782e-01 -3.34711522e-01 1.29056005e-02 -2.26221353e-01
4.19069827e-01 -8.70847046e-01 -6.35395348e-01 7.55197704e-02
-7.30315208e-01 -1.71738535e-01 -7.39589691e-01 2.62894303e-01
-5.87301254e-01 4.44697291e-01 -2.72260249e-01 -1.00473523e+00
-6.21606767e-01 -4.94725466e-01 5.14755845e-01 2.86137342e-01
-5.60884595e-01 -8.11694920e-01 8.94542411e-02 6.46033168e-01
3.52738321e-01 4.91843313e-01 7.36491442e-01 -9.25208807e-01
-5.97199857e-01 -2.81395823e-01 1.35758659e-02 1.47468925e-01
-1.55110359e-01 1.02830596e-01 -5.76313972e-01 -5.20127714e-01
-3.71843040e-01 -4.35593247e-01 8.73966813e-01 9.74398136e-01
1.23792267e+00 -7.58975804e-01 -1.73906505e-01 4.97717559e-01
1.19931173e+00 -4.19338435e-01 2.49447450e-01 -5.16384281e-03
3.35900605e-01 1.03679776e+00 6.84349477e-01 1.43731630e+00
7.27296054e-01 2.68930525e-01 4.40143198e-01 4.52264726e-01
4.88664806e-01 -2.14639023e-01 2.40174398e-01 3.33708376e-01
3.54677588e-02 -5.46662748e-01 -4.13374186e-01 9.00721490e-01
-2.03029513e+00 -1.17189205e+00 -5.04777096e-02 2.58105016e+00
8.70040357e-01 -1.85158268e-01 7.31175363e-01 8.94051641e-02
7.66365886e-01 3.28917593e-01 -7.10937858e-01 -6.40578091e-01
-3.44868273e-01 -2.88480133e-01 6.72329366e-01 3.13601524e-01
-7.52586663e-01 2.71218002e-01 5.83457899e+00 9.09101665e-01
-6.23220801e-01 -6.20257966e-02 1.26326156e+00 -1.06790316e+00
-1.07436407e+00 -1.90489545e-01 -4.74513888e-01 6.06292903e-01
7.30130613e-01 -9.05018032e-01 4.90242362e-01 7.60943532e-01
5.65450549e-01 -3.10180515e-01 -1.22989023e+00 8.93937826e-01
1.40931994e-01 -1.30854321e+00 -4.24496494e-02 2.97240824e-01
1.38934934e+00 -3.76753747e-01 1.60205498e-01 5.50609939e-02
4.74313915e-01 -1.06335378e+00 5.76232612e-01 3.08349520e-01
6.79612875e-01 -1.17092228e+00 2.25246251e-01 4.79936272e-01
-2.69478321e-01 -2.79339701e-01 -7.42403388e-01 -3.48123834e-02
3.39956194e-01 1.12683833e+00 -3.98535669e-01 2.55264193e-01
3.88819665e-01 4.86156583e-01 -5.70774414e-02 1.22390187e+00
2.16336504e-01 6.76309645e-01 -4.49758321e-01 -2.39347324e-01
-1.34582028e-01 -2.03225657e-01 8.21801186e-01 7.98570096e-01
4.77942556e-01 1.35991439e-01 -6.26095163e-04 7.56646752e-01
-7.12252915e-01 4.68636245e-01 -4.15467829e-01 -8.57654437e-02
7.31383801e-01 1.24118245e+00 -4.55288142e-01 -1.32195696e-01
-1.17813945e-01 5.40151238e-01 6.95749223e-02 3.15789878e-01
-5.39356828e-01 -7.32912496e-02 8.52174997e-01 5.18855035e-01
-4.79514711e-02 2.35408023e-01 -8.93948615e-01 -1.14238012e+00
-1.89905912e-01 -8.89722764e-01 9.12785590e-01 -7.33475089e-02
-1.26016235e+00 -3.87803376e-01 1.79574579e-01 -5.00310004e-01
-1.70699775e-01 2.99811661e-01 -8.61381233e-01 3.66090983e-01
-1.25456429e+00 -7.01341033e-01 2.59908615e-03 2.81877697e-01
3.13521445e-01 -1.07584335e-01 8.56652930e-02 1.91911697e-01
-4.94595975e-01 8.02916765e-01 6.14169896e-01 -5.78763187e-01
6.40048802e-01 -1.25667727e+00 -1.84877664e-01 6.09075487e-01
-2.07962781e-01 1.59278706e-01 1.07004547e+00 -6.15315914e-01
-1.41545248e+00 -9.09205973e-01 9.36663091e-01 -1.73266411e-01
1.24615684e-01 -1.32894650e-01 -4.64947045e-01 2.99867421e-01
1.27136514e-01 -4.06213850e-01 8.75853121e-01 3.39762390e-01
1.18862800e-01 -3.31795961e-01 -1.69541645e+00 6.54208004e-01
1.21908879e+00 1.87827632e-01 7.01137912e-03 4.98979241e-01
6.69926941e-01 -2.65467167e-03 -9.39027011e-01 2.34579414e-01
4.62591469e-01 -9.01298940e-01 9.36973989e-01 -8.06393921e-01
9.32279110e-01 5.57536840e-01 -2.69242615e-01 -1.34776533e+00
-1.74008206e-01 -1.02081847e+00 -3.03831190e-01 1.44881618e+00
5.72862685e-01 -3.99062753e-01 1.37047446e+00 1.08035076e+00
3.45170230e-01 -8.96985233e-01 -8.32434654e-01 -5.50776839e-01
3.64471078e-01 -6.02319203e-02 7.16068387e-01 1.00544226e+00
4.13575888e-01 1.56367451e-01 -6.52933776e-01 -2.72550225e-01
1.12218702e+00 4.61106092e-01 8.70424092e-01 -1.37853456e+00
-1.80642024e-01 -6.17222428e-01 7.23867938e-02 -8.62223148e-01
1.64749473e-01 -9.23939109e-01 -3.51767123e-01 -1.48725867e+00
7.35324979e-01 -4.72622693e-01 9.03998837e-02 -1.97146237e-01
-3.09822142e-01 -2.07691729e-01 2.27023825e-01 9.84011311e-03
-9.87249672e-01 6.47266626e-01 1.12675047e+00 -5.35967499e-02
-2.85456121e-01 5.20382524e-01 -1.34755874e+00 2.88912594e-01
8.86254489e-01 -4.50888813e-01 -4.98015791e-01 -2.73315366e-02
7.05238581e-01 2.06219137e-01 -9.54274163e-02 -4.18971390e-01
1.46992147e-01 -6.40192389e-01 -6.55262768e-02 -5.03925502e-01
1.08427316e-01 -5.89711964e-01 3.74072380e-02 2.88880825e-01
-8.90394568e-01 -2.83574343e-01 -4.24249560e-01 6.54410660e-01
8.94495547e-02 -4.00578201e-01 6.58422291e-01 -7.07249418e-02
1.60870627e-01 6.57274902e-01 -1.97343901e-01 9.01343167e-01
1.00946212e+00 -1.07726023e-01 -2.66133904e-01 -1.14771855e+00
-2.70366430e-01 7.63081014e-01 6.56337738e-01 -8.99298768e-03
5.62607646e-01 -1.16217482e+00 -1.30374646e+00 -5.11338472e-01
-1.16763771e-01 1.77255765e-01 4.50124711e-01 8.63100827e-01
-1.74412996e-01 1.25294998e-01 1.14038043e-01 -1.62126750e-01
-1.31184030e+00 3.92931759e-01 -1.04913041e-01 -2.02297002e-01
-6.91028684e-02 7.85154879e-01 4.06709872e-02 -2.86987811e-01
3.98494780e-01 3.33407633e-02 -1.92909062e-01 6.98641479e-01
2.94945061e-01 1.21895051e+00 -3.47276777e-01 -1.80955827e-01
1.33494381e-02 -2.01577246e-01 1.39563560e-01 -4.68941331e-02
1.61485314e+00 -3.71990770e-01 -3.59688967e-01 2.32636794e-01
9.83580232e-01 1.63754597e-01 -9.53810155e-01 -1.11258313e-01
-6.40693530e-02 -6.94090605e-01 -1.22504659e-01 -2.55950272e-01
-1.06299031e+00 3.62256587e-01 -1.99425384e-01 6.45366371e-01
8.78467441e-01 -1.02059999e-02 8.10246348e-01 1.60305083e-01
2.58807659e-01 -1.29612720e+00 -2.06407860e-01 -6.25061765e-02
7.09087789e-01 -1.25832784e+00 5.44986725e-01 -2.13710606e-01
-7.03848898e-01 6.40739858e-01 3.51888627e-01 -7.83455838e-03
6.79883897e-01 -7.09595755e-02 -5.60182691e-01 -8.11395571e-02
-1.12892902e+00 2.86748797e-01 1.24840707e-01 3.11657846e-01
2.30388433e-01 3.79343033e-01 -8.84878516e-01 8.23947072e-01
-2.81792104e-01 -1.61107779e-01 9.73170996e-01 6.20758772e-01
-7.07240701e-01 -7.71597624e-01 -4.49396759e-01 1.35895252e+00
-8.79327595e-01 7.10913911e-02 -3.99039477e-01 2.47934669e-01
-3.01793188e-01 1.03073263e+00 9.48576927e-02 1.27876580e-01
1.57471225e-02 -3.03124756e-01 3.71246487e-01 -4.22943205e-01
-5.04918933e-01 -3.23204607e-01 3.50961655e-01 -2.62642473e-01
-2.16149598e-01 -1.22710681e+00 -8.51134300e-01 -9.20191526e-01
-2.03129694e-01 2.43868902e-01 3.81941080e-01 4.96954441e-01
3.09710383e-01 3.84939201e-02 8.17723572e-01 -5.74696600e-01
-1.28140056e+00 -5.63088119e-01 -9.37818229e-01 6.26200855e-01
1.79842636e-01 -5.90647906e-02 -5.17904937e-01 -3.63960445e-01] | [6.616864204406738, 4.949122428894043] |
1087ad39-87a9-47a5-a179-abcc6370b1ac | cloud-detection-from-rgb-color-remote-sensing | 1801.08706 | null | http://arxiv.org/abs/1801.08706v1 | http://arxiv.org/pdf/1801.08706v1.pdf | Cloud Detection From RGB Color Remote Sensing Images With Deep Pyramid Networks | Cloud detection from remotely observed data is a critical pre-processing step
for various remote sensing applications. In particular, this problem becomes
even harder for RGB color images, since there is no distinct spectral pattern
for clouds, which is directly separable from the Earth surface. In this paper,
we adapt a deep pyramid network (DPN) to tackle this problem. For this purpose,
the network is enhanced with a pre-trained parameter model at the encoder
layer. Moreover, the method is able to obtain accurate pixel-level segmentation
and classification results from a set of noisy labeled RGB color images. In
order to demonstrate the superiority of the method, we collect and label data
with the corresponding cloud/non-cloudy masks acquired from low-orbit Gokturk-2
and RASAT satellites. The experimental results validates that the proposed
method outperforms several baselines even for hard cases (e.g. snowy mountains)
that are perceptually difficult to distinguish by human eyes. | ['Savas Ozkan', 'Mehmet Efendioglu', 'Caner Demirpolat'] | 2018-01-26 | null | null | null | null | ['cloud-detection'] | ['computer-vision'] | [ 3.72598380e-01 -4.93504077e-01 3.13506484e-01 -4.26177174e-01
-5.66671669e-01 -6.81138575e-01 2.59153008e-01 -9.77868289e-02
-5.70034027e-01 7.25070357e-01 -7.00633585e-01 -4.98829514e-01
4.12309319e-02 -9.45525169e-01 -4.56632078e-01 -1.06387687e+00
-7.71744475e-02 1.75978646e-01 5.01817325e-03 -1.37908161e-01
-5.20616956e-02 7.84869313e-01 -1.69570768e+00 1.74868077e-01
1.34000289e+00 1.18747795e+00 8.02373052e-01 3.38389635e-01
-6.68627322e-02 2.76567638e-01 -1.72286138e-01 1.17572695e-01
6.43187165e-01 -2.96674430e-01 -4.10422623e-01 4.44832385e-01
5.71190476e-01 -3.96645129e-01 1.57002881e-01 1.64802802e+00
2.11742401e-01 6.48858920e-02 6.21702731e-01 -9.43712175e-01
-3.18866700e-01 -9.20251459e-02 -7.20370412e-01 1.39907733e-01
-4.17042702e-01 2.83783108e-01 9.48459387e-01 -8.49434376e-01
1.67062104e-01 8.97845984e-01 4.27215487e-01 3.30010764e-02
-9.78439808e-01 -7.52110839e-01 2.13517264e-01 3.67602527e-01
-1.54057348e+00 -5.04440032e-02 6.37773037e-01 -4.68663573e-01
2.79372811e-01 4.75010931e-01 8.29566240e-01 4.37552780e-01
-9.15386081e-02 5.89661062e-01 1.82999837e+00 -2.50874043e-01
2.60121673e-01 6.70863092e-02 -6.40127361e-02 3.54102165e-01
5.98591149e-01 1.35800302e-01 1.67037115e-01 2.80950695e-01
6.03068113e-01 2.16946274e-01 -6.79252863e-01 -1.49287786e-02
-6.96958244e-01 6.47360265e-01 9.28209662e-01 3.00272286e-01
-5.96715152e-01 -1.58085704e-01 -2.70690233e-01 1.57977015e-01
4.43593949e-01 1.67557314e-01 -3.48372549e-01 4.93817300e-01
-1.36698902e+00 5.16364276e-02 3.05826157e-01 7.23749459e-01
1.01537418e+00 3.10498238e-01 6.82665110e-02 5.63452721e-01
5.37257552e-01 9.80346620e-01 9.00649801e-02 -5.72199047e-01
3.25354487e-01 4.99651849e-01 4.99927998e-01 -9.40638959e-01
-3.30979496e-01 -6.92380786e-01 -1.15818667e+00 7.05033541e-01
2.44646922e-01 3.38095985e-03 -1.28652334e+00 9.20565724e-01
2.82070041e-01 1.89864740e-01 2.07192883e-01 1.53472173e+00
7.13574231e-01 8.03265035e-01 -9.38754305e-02 -1.38303071e-01
1.50370729e+00 -8.27777088e-01 -5.18253744e-01 -5.76804519e-01
4.52834256e-02 -5.32829821e-01 1.18439567e+00 4.55113322e-01
-3.58420849e-01 -7.26976693e-01 -1.01253629e+00 2.37852544e-01
-4.84633893e-01 6.40263081e-01 6.76656604e-01 6.97793126e-01
-8.86039197e-01 6.65906429e-01 -8.68667901e-01 -1.15515769e-01
3.88204157e-01 8.52292627e-02 -8.00430775e-02 -2.52586395e-01
-1.17724085e+00 6.25336170e-01 6.59542263e-01 1.02663517e+00
-8.33769977e-01 -1.79940820e-01 -5.88400066e-01 2.83554822e-01
2.08708614e-01 -1.46714106e-01 7.77291834e-01 -1.50979555e+00
-1.15192378e+00 8.16727102e-01 -6.67142272e-02 -2.75819242e-01
5.56643903e-01 -1.63041711e-01 -4.29064214e-01 3.61427605e-01
-9.97914225e-02 3.86720121e-01 1.18835115e+00 -1.74874175e+00
-8.03877056e-01 -4.46412444e-01 2.15301365e-02 3.59363556e-01
-8.56250450e-02 -8.71712267e-02 -4.88183856e-01 -5.31213105e-01
3.99158567e-01 -9.74916101e-01 -3.06436449e-01 1.03616998e-01
-4.09802675e-01 2.51067698e-01 8.00677419e-01 -9.44868565e-01
5.91953635e-01 -2.30836582e+00 -3.76252621e-01 3.75832200e-01
-9.96791385e-03 5.72594881e-01 -1.10720560e-01 9.29492433e-03
-4.79618721e-02 1.75258487e-01 -7.26604223e-01 -7.07805976e-02
-2.28694543e-01 3.98397833e-01 -1.84892893e-01 6.77138746e-01
3.95095527e-01 5.55799961e-01 -6.41027927e-01 -4.07117367e-01
1.47138581e-01 4.64944243e-01 -7.42418841e-02 1.91822797e-01
-4.12117422e-01 8.22369218e-01 -3.58372271e-01 7.93950438e-01
1.41887248e+00 -2.02915967e-01 8.97399560e-02 7.06318347e-03
-3.91228706e-01 -9.16913077e-02 -1.34049952e+00 1.06801772e+00
-2.62319446e-01 7.70338893e-01 4.46092635e-01 -7.84395158e-01
8.71885180e-01 1.03660397e-01 3.40484940e-02 -7.90425062e-01
1.42490625e-01 3.07501256e-01 1.20451204e-01 -7.18384504e-01
5.61708093e-01 -4.29343402e-01 4.28728461e-01 -8.79530013e-02
-6.07335985e-01 -4.80734795e-01 -2.80370265e-02 -3.15181851e-01
2.80448914e-01 5.08845672e-02 -8.37802589e-02 -2.11609736e-01
6.31260633e-01 2.93447673e-01 8.65888178e-01 6.95163488e-01
-1.12283081e-01 8.42361212e-01 1.29401445e-01 -3.78046304e-01
-7.48849690e-01 -8.03824306e-01 -3.48657101e-01 5.81679046e-01
4.95997459e-01 3.80640119e-01 -5.49732625e-01 -3.08406740e-01
-1.41942408e-02 4.22297508e-01 -3.60108227e-01 3.69838476e-01
-2.03035921e-01 -1.05702102e+00 1.94003209e-01 2.85482019e-01
1.08443701e+00 -9.85404968e-01 -6.06854856e-01 8.51740241e-02
-3.33589554e-01 -1.36328781e+00 2.07694679e-01 1.46264598e-01
-1.00401974e+00 -1.00170457e+00 -5.60679317e-01 -7.47681618e-01
7.84285426e-01 6.80823326e-01 8.85363340e-01 2.79441297e-01
-1.99106857e-01 -3.21088672e-01 -5.49094915e-01 -4.17872459e-01
-1.68987755e-02 -2.62830436e-01 -2.84010231e-01 3.81450385e-01
2.88087249e-01 -3.30519468e-01 -7.66328692e-01 2.65244126e-01
-1.16934824e+00 2.73930281e-02 8.52860212e-01 7.27387249e-01
8.07507634e-01 7.23494172e-01 -6.84536695e-02 -6.94517553e-01
1.65107578e-01 -7.32753426e-02 -1.18681908e+00 1.96338460e-01
-3.96118075e-01 -3.66744787e-01 5.96067250e-01 1.55094668e-01
-1.15349054e+00 5.66653430e-01 -5.25223389e-02 -3.21567893e-01
-4.93021488e-01 7.20113814e-01 -3.20065171e-01 -2.95781612e-01
4.11772966e-01 4.27436322e-01 -2.64061004e-01 -6.05816722e-01
1.38124824e-01 1.16052365e+00 4.24408555e-01 -1.29472345e-01
1.16400921e+00 9.05052960e-01 -1.11821756e-01 -1.22782171e+00
-6.14632249e-01 -6.34995997e-01 -6.62329435e-01 -3.31067055e-01
1.02873015e+00 -1.26276743e+00 -4.39332157e-01 6.20957613e-01
-8.91803980e-01 -3.16409260e-01 1.60303518e-01 5.92017293e-01
-1.87980104e-02 5.65664589e-01 -4.48165148e-01 -1.11657107e+00
-4.18413043e-01 -1.08965111e+00 9.41949785e-01 4.17573690e-01
7.50145435e-01 -4.68026698e-01 -3.30095202e-01 4.55439359e-01
3.97579312e-01 4.08727914e-01 6.60635471e-01 -2.72466913e-02
-9.22741473e-01 -8.57476965e-02 -7.09166229e-01 8.54201436e-01
1.18941605e-01 2.33342186e-01 -1.33767200e+00 -3.79175901e-01
7.48929679e-02 -1.68300554e-01 1.15676975e+00 2.61111975e-01
1.27477300e+00 -1.29437093e-02 -6.33424753e-03 8.09499860e-01
1.89330864e+00 4.14424539e-02 7.03902006e-01 4.66124058e-01
6.60124004e-01 5.80841839e-01 8.36913109e-01 2.76920617e-01
1.99481770e-01 2.81999439e-01 1.01947701e+00 -5.91088116e-01
8.00676495e-02 3.95145774e-01 6.68806508e-02 3.89289767e-01
-4.63633984e-01 -9.32138041e-02 -1.00947881e+00 5.25609374e-01
-1.45534277e+00 -6.67897701e-01 -6.55447185e-01 2.16736150e+00
6.06394172e-01 -3.54628563e-02 -5.02152085e-01 1.88042521e-01
6.38301551e-01 1.91786975e-01 -4.16068405e-01 1.50602803e-01
-3.20085645e-01 3.67664725e-01 8.02713096e-01 4.00528342e-01
-1.49258518e+00 9.49268937e-01 4.98876762e+00 7.47390926e-01
-1.65845847e+00 4.05037031e-02 4.93609577e-01 3.76920104e-01
-1.14885256e-01 -3.65285277e-02 -4.47043657e-01 5.54283202e-01
3.91617835e-01 5.62401593e-01 4.42297727e-01 6.32054925e-01
5.96988499e-01 -5.24875820e-01 -4.53356832e-01 1.00976324e+00
-3.24114501e-01 -7.71256089e-01 -1.94879830e-01 -7.08049163e-02
7.28698134e-01 3.73142391e-01 5.61488420e-02 9.58093405e-02
9.10127014e-02 -1.00195837e+00 6.07933342e-01 4.51933682e-01
6.58533394e-01 -5.10055959e-01 8.95912468e-01 5.16605914e-01
-1.06890738e+00 -7.17833936e-02 -6.95038915e-01 -1.20102555e-01
-2.14846805e-01 9.33212757e-01 -5.92179358e-01 7.84644306e-01
8.93970728e-01 5.01786768e-01 -6.11560881e-01 1.46788633e+00
-6.62924528e-01 6.93439066e-01 -4.78466451e-01 2.83200264e-01
4.68065858e-01 -6.85380459e-01 1.79333806e-01 1.08572686e+00
4.35696185e-01 3.74334246e-01 1.42188624e-01 8.57883632e-01
4.01069969e-02 5.53257316e-02 -3.23359460e-01 -5.88729419e-02
1.39377251e-01 1.55558693e+00 -9.31751668e-01 -2.47562677e-01
-3.49172443e-01 1.04693425e+00 -1.93952665e-01 6.45076990e-01
-7.18500614e-01 -3.00419450e-01 6.09504044e-01 -1.22432962e-01
4.93351102e-01 -4.79934335e-01 -2.95700341e-01 -1.24650276e+00
2.20534891e-01 -9.42669392e-01 -8.22504014e-02 -1.07386899e+00
-1.04401016e+00 8.33869696e-01 -3.45754117e-01 -1.77117217e+00
3.36091727e-01 -8.81469250e-01 -5.64988673e-01 1.20969903e+00
-2.43823361e+00 -1.08260441e+00 -1.01842618e+00 4.74220663e-01
5.66659234e-02 4.14086014e-01 6.21993065e-01 3.50691319e-01
-5.33117533e-01 -9.15486887e-02 6.35857821e-01 2.36840114e-01
3.15976560e-01 -1.28118527e+00 1.26604706e-01 1.29113483e+00
-1.02312751e-02 5.07037267e-02 5.95051765e-01 -4.97583807e-01
-1.15112376e+00 -1.42975354e+00 4.11094517e-01 2.82067329e-01
2.70423293e-01 -2.82769769e-01 -1.02145922e+00 3.49521875e-01
1.24511838e-01 2.91627467e-01 3.94658536e-01 -3.96231353e-01
-3.40179317e-02 -4.65860963e-01 -1.02602696e+00 1.83268443e-01
3.98498386e-01 -4.08651471e-01 -2.93149740e-01 5.59266567e-01
1.86927676e-01 -3.46694410e-01 -3.87973934e-01 5.47085822e-01
2.80169547e-01 -1.13943696e+00 7.01540351e-01 -1.86363965e-01
3.11695844e-01 -8.63443971e-01 -2.77417392e-01 -1.33927095e+00
-2.58590698e-01 2.03263387e-02 6.34160101e-01 7.97271013e-01
3.02556515e-01 -5.82390845e-01 7.65690088e-01 2.53149956e-01
-3.02890595e-02 -1.12532161e-01 -8.11882496e-01 -9.44328487e-01
-9.47146714e-02 -3.89473826e-01 4.44295287e-01 1.14431369e+00
-8.08205903e-01 -1.29841581e-01 -3.14648479e-01 1.04169321e+00
7.28561401e-01 8.59506726e-01 6.22667491e-01 -1.41070771e+00
-2.26638049e-01 -1.61162749e-01 -2.87047297e-01 -8.95416319e-01
-1.70484781e-01 -6.50676310e-01 5.26420593e-01 -1.61897731e+00
-4.25224416e-02 -7.91049659e-01 -2.50026077e-01 4.74791825e-01
-3.87181878e-01 5.38992226e-01 3.08579654e-01 3.86604518e-01
-1.82444900e-01 7.51789629e-01 1.38520491e+00 -4.69629705e-01
-9.37237442e-02 1.27700448e-01 -1.78701848e-01 7.22030640e-01
1.08321369e+00 -4.94585931e-01 -4.72491458e-02 -7.45073915e-01
1.51614875e-01 -7.35411718e-02 5.97794890e-01 -1.22491050e+00
-2.18583830e-02 -2.70532638e-01 5.43015003e-01 -7.51420081e-01
2.70551533e-01 -1.26319456e+00 9.98337045e-02 4.74888831e-01
3.28369677e-01 -3.47288132e-01 1.94151878e-01 5.63976467e-01
-5.88650584e-01 -3.25924814e-01 1.09137893e+00 -2.37922519e-01
-9.77152586e-01 2.77029514e-01 -4.18458045e-01 -2.82600552e-01
9.33090806e-01 -2.19432056e-01 -1.51330546e-01 -2.36623302e-01
-6.71993196e-01 4.23020810e-01 5.34841120e-01 2.20508743e-02
6.48684382e-01 -8.85467231e-01 -8.01408410e-01 4.26284164e-01
1.99143916e-01 4.65016127e-01 4.33978021e-01 9.47362185e-01
-1.08860922e+00 6.80555403e-02 -3.05093110e-01 -7.33372152e-01
-1.20733356e+00 2.82547504e-01 6.74714208e-01 1.34448394e-01
-6.21790528e-01 6.97448730e-01 -5.51143941e-03 -4.05577451e-01
-5.65206185e-02 -8.43302667e-01 -3.02533239e-01 6.16112240e-02
4.07262087e-01 -7.50722215e-02 3.81754965e-01 -6.43612981e-01
-2.84299374e-01 4.94471788e-01 3.43515635e-01 9.22388732e-02
1.33410120e+00 -7.18637928e-02 -4.01583433e-01 1.82697728e-01
7.67062902e-01 -3.24578695e-02 -1.41886818e+00 -2.62046665e-01
-2.18268141e-01 -8.61466527e-01 4.52363938e-01 -6.07581377e-01
-1.55269384e+00 1.22629213e+00 1.01941359e+00 2.96996415e-01
1.43788612e+00 -5.41980028e-01 4.22827959e-01 4.93286371e-01
1.58620089e-01 -1.03086567e+00 -5.08493066e-01 4.01869953e-01
7.07972288e-01 -1.56534338e+00 9.48692635e-02 -5.14862537e-01
-5.58687687e-01 1.12596929e+00 4.32381868e-01 -5.15180118e-02
6.03788495e-01 -1.74241707e-01 5.32424212e-01 -1.76423699e-01
2.03519445e-02 -9.02884603e-01 1.70062587e-01 3.38282734e-01
6.43551955e-03 4.41843122e-01 -1.56018212e-01 2.28957951e-01
6.03482574e-02 -7.84818381e-02 5.91464996e-01 7.88015068e-01
-6.97139263e-01 -7.24152863e-01 -9.03030992e-01 4.74863648e-01
-3.07166904e-01 -3.81219834e-01 -1.01859421e-01 8.28721106e-01
3.74475271e-01 1.03844786e+00 9.83524397e-02 -2.11900890e-01
7.62683386e-03 -2.31728405e-01 1.63068280e-01 -4.95795280e-01
-3.75100672e-01 3.33752185e-01 -1.60954922e-01 -2.58977175e-01
-7.22350299e-01 -5.87187529e-01 -1.12517476e+00 5.43890446e-02
-5.67831755e-01 2.30904117e-01 8.54475379e-01 7.18696594e-01
1.54660353e-02 3.87433738e-01 7.85782337e-01 -1.06965041e+00
-2.96527326e-01 -9.32588160e-01 -1.30557764e+00 2.63853878e-01
6.14888430e-01 -4.91033912e-01 -3.26357096e-01 5.66600673e-02] | [9.794560432434082, -1.724785327911377] |
715bbcca-057e-42e7-a236-9abee83e9ce5 | ultra-high-definition-low-light-image | 2212.11548 | null | https://arxiv.org/abs/2212.11548v1 | https://arxiv.org/pdf/2212.11548v1.pdf | Ultra-High-Definition Low-Light Image Enhancement: A Benchmark and Transformer-Based Method | As the quality of optical sensors improves, there is a need for processing large-scale images. In particular, the ability of devices to capture ultra-high definition (UHD) images and video places new demands on the image processing pipeline. In this paper, we consider the task of low-light image enhancement (LLIE) and introduce a large-scale database consisting of images at 4K and 8K resolution. We conduct systematic benchmarking studies and provide a comparison of current LLIE algorithms. As a second contribution, we introduce LLFormer, a transformer-based low-light enhancement method. The core components of LLFormer are the axis-based multi-head self-attention and cross-layer attention fusion block, which significantly reduces the linear complexity. Extensive experiments on the new dataset and existing public datasets show that LLFormer outperforms state-of-the-art methods. We also show that employing existing LLIE methods trained on our benchmark as a pre-processing step significantly improves the performance of downstream tasks, e.g., face detection in low-light conditions. The source code and pre-trained models are available at https://github.com/TaoWangzj/LLFormer. | ['Tong Lu', 'Bjorn Stenger', 'Wenhan Luo', 'Tianrun Shen', 'Kaihao Zhang', 'Tao Wang'] | 2022-12-22 | null | null | null | null | ['face-detection', 'low-light-image-enhancement'] | ['computer-vision', 'computer-vision'] | [ 3.44675004e-01 -3.53358895e-01 2.52991915e-01 -4.60433334e-01
-1.00972438e+00 -1.14629515e-01 3.20159942e-01 -2.38032818e-01
-5.21345913e-01 3.09377372e-01 2.64347672e-01 -4.64693271e-03
3.36005628e-01 -6.58115685e-01 -7.78404653e-01 -7.00216711e-01
4.20285732e-01 -2.54743934e-01 1.39973521e-01 -7.14072376e-04
-1.60525460e-02 4.93854314e-01 -1.89607453e+00 5.14611006e-01
4.98053908e-01 1.28215218e+00 3.32639039e-01 8.13391387e-01
3.70620817e-01 5.13702512e-01 -1.53656811e-01 -6.60123944e-01
4.22221005e-01 -9.53999013e-02 -4.79582131e-01 7.57639483e-02
9.39924717e-01 -8.60611439e-01 -3.42642158e-01 1.03737664e+00
1.24370313e+00 -6.45426214e-02 2.44960934e-01 -1.00939620e+00
-7.73550689e-01 -1.42385632e-01 -8.00178230e-01 4.05012965e-01
2.91901559e-01 4.04258043e-01 6.05936170e-01 -1.38662684e+00
4.51514363e-01 1.08129239e+00 7.62468934e-01 7.35148132e-01
-1.11057746e+00 -1.01579082e+00 -3.04302424e-01 3.78425747e-01
-1.16095948e+00 -1.14706886e+00 5.61682343e-01 -2.20953599e-01
1.19043839e+00 4.91340049e-02 3.42246741e-01 8.61868203e-01
1.35808736e-01 6.99047327e-01 1.35602212e+00 -4.28174525e-01
-1.09771952e-01 -6.82556778e-02 -1.51069582e-01 6.65342391e-01
2.37561375e-01 2.10874096e-01 -7.93719471e-01 1.46937132e-01
5.24037659e-01 -6.43265992e-02 -3.40731919e-01 1.98193848e-01
-7.15861022e-01 4.23784524e-01 4.11398828e-01 1.37836009e-01
-3.49701613e-01 5.81724048e-02 2.45871499e-01 -6.88039660e-02
8.37797344e-01 8.44962224e-02 -3.82740945e-01 9.45162848e-02
-8.35490704e-01 -7.70226270e-02 9.38608646e-02 8.55468094e-01
7.24536777e-01 -3.01973075e-01 -3.81860822e-01 9.61374998e-01
4.04152066e-01 7.38196313e-01 7.91076422e-02 -1.01524520e+00
3.05586576e-01 2.59433597e-01 6.85721263e-02 -6.52104199e-01
-4.80599612e-01 -2.74970978e-01 -7.28497088e-01 2.54758000e-01
1.31712243e-01 -7.84490630e-02 -8.46701086e-01 1.59750807e+00
6.79238856e-01 4.52708215e-01 -1.62076995e-01 1.01022291e+00
1.16747749e+00 5.34439147e-01 2.09720712e-02 -4.64921474e-01
1.54376566e+00 -9.44166541e-01 -7.86841810e-01 -1.54001698e-01
5.38432598e-02 -1.14932537e+00 1.12438393e+00 3.60399276e-01
-1.40206850e+00 -6.48525357e-01 -7.62953401e-01 -5.93058288e-01
-2.34884918e-01 3.70361954e-01 5.96004188e-01 7.44604051e-01
-1.41076553e+00 1.31930113e-01 -6.87092185e-01 -5.17495811e-01
6.39631748e-01 3.38153601e-01 -3.30970794e-01 -4.09816623e-01
-8.30096960e-01 4.92835104e-01 -1.17347583e-01 2.04352647e-01
-8.03178430e-01 -9.35937047e-01 -7.60052621e-01 -2.73846567e-01
2.29573831e-01 -5.36016405e-01 1.48700595e+00 -2.68047094e-01
-1.53134441e+00 1.21161866e+00 -4.49605644e-01 2.01312751e-02
1.35677174e-01 -2.55840480e-01 -4.53430235e-01 4.21783119e-01
-4.70394455e-02 6.15149617e-01 1.04288697e+00 -1.04110527e+00
-6.20578587e-01 -4.65980560e-01 -8.18532929e-02 3.58287096e-02
-5.10500729e-01 7.56868422e-01 -1.06674182e+00 -3.47026706e-01
-2.92530924e-01 -6.27811849e-01 1.95594415e-01 1.38613492e-01
-2.20962852e-01 -2.04507038e-01 9.11446869e-01 -6.66445732e-01
1.06773388e+00 -2.21410179e+00 -4.32288229e-01 -4.16631281e-01
4.23385561e-01 5.16877711e-01 -2.54568726e-01 -1.16086684e-01
-5.56999594e-02 -2.28876948e-01 -1.49014607e-01 -8.73360515e-01
-3.45507152e-02 -2.24408671e-01 1.07168369e-01 6.77967191e-01
2.63172776e-01 8.37488055e-01 -7.50594676e-01 -5.24060905e-01
4.42038715e-01 9.78901744e-01 -6.88331902e-01 4.14487988e-01
1.48977190e-01 4.76160347e-01 -7.20026158e-03 9.67937350e-01
1.07576311e+00 -3.60263109e-01 -2.28028119e-01 -9.26088274e-01
-4.22521412e-01 4.23864163e-02 -9.01477814e-01 1.64632618e+00
-6.77245677e-01 7.14493454e-01 3.45047235e-01 -3.25104386e-01
4.33486253e-01 2.41834551e-01 5.04346371e-01 -1.15627503e+00
4.20031130e-01 -7.60691315e-02 -5.22781491e-01 -6.64596140e-01
4.31978792e-01 -2.21089885e-01 3.84124190e-01 5.18366098e-01
8.93310830e-02 6.10719807e-02 2.37527564e-01 1.78464517e-01
9.54417050e-01 -4.61917035e-02 9.31918249e-02 1.63379356e-01
4.27705139e-01 -7.85460293e-01 4.41262394e-01 4.62221682e-01
-4.09188479e-01 8.82551670e-01 -1.37253314e-01 -1.76320821e-01
-9.35916305e-01 -1.05859172e+00 -5.62488019e-01 1.26594079e+00
-3.40617076e-02 -4.88144487e-01 -8.42230737e-01 -3.19583267e-01
-2.43534058e-01 2.93811262e-01 -5.38270891e-01 1.11732870e-01
-2.96220690e-01 -1.25053835e+00 2.40894452e-01 3.77631187e-01
8.61295938e-01 -8.06205094e-01 -7.30286002e-01 -9.88677070e-02
-2.87374914e-01 -1.59084022e+00 -4.99393284e-01 -9.25128087e-02
-2.87115097e-01 -1.05485868e+00 -5.78690827e-01 -4.23068136e-01
5.33873916e-01 6.41889453e-01 1.07873189e+00 1.70495585e-02
-7.31432498e-01 7.09156752e-01 -2.16876313e-01 -5.11510968e-01
9.84944329e-02 -4.14633363e-01 1.51189491e-01 3.01255763e-01
4.94116277e-01 -2.92645812e-01 -1.08094013e+00 8.80494416e-02
-8.52823079e-01 1.13170601e-01 7.21276760e-01 5.92581093e-01
8.18722486e-01 3.75906341e-02 2.84414440e-01 -5.16699135e-01
4.17772204e-01 -1.34026706e-01 -7.40579188e-01 1.22023210e-01
-6.90531731e-01 -2.22442344e-01 2.62224495e-01 -6.26940206e-02
-1.53123653e+00 3.14790085e-02 -4.19895411e-01 -3.90513569e-01
-2.73310542e-01 -2.17470467e-01 -4.43527788e-01 -3.99382085e-01
3.68162423e-01 -2.02608593e-02 -2.88329899e-01 -7.05287099e-01
3.22503537e-01 9.59878325e-01 6.91726565e-01 -1.96522564e-01
5.71743906e-01 7.31284380e-01 9.28313509e-02 -1.05556214e+00
-9.10524011e-01 -6.12768650e-01 -4.93675292e-01 -5.51828265e-01
1.09101665e+00 -1.20800507e+00 -9.20979321e-01 8.75947177e-01
-1.17818069e+00 -6.03924274e-01 3.53264958e-02 2.80938029e-01
-1.69155538e-01 3.16169918e-01 -8.30041647e-01 -7.79265761e-01
-6.66218162e-01 -1.23533690e+00 1.62827694e+00 4.00794536e-01
4.63513136e-01 -4.30095762e-01 -8.76014382e-02 6.57852471e-01
5.29684365e-01 -6.46811873e-02 2.67539561e-01 2.60133386e-01
-7.35405385e-01 3.97658162e-02 -7.70874143e-01 5.48657954e-01
2.19975295e-03 -1.79193899e-01 -1.58033299e+00 -4.57917541e-01
2.76370794e-02 -5.30448139e-01 8.92022073e-01 6.25537336e-01
1.31255865e+00 1.56679004e-02 -7.51933381e-02 1.09354424e+00
1.46594465e+00 -1.30792335e-01 7.49608278e-01 2.30418101e-01
6.02532327e-01 4.30166513e-01 6.28373504e-01 6.07431293e-01
5.32271147e-01 8.89511049e-01 4.40102369e-01 -4.50509578e-01
-6.91534460e-01 4.26266715e-02 4.05358821e-01 5.99551558e-01
-1.54773101e-01 -1.11871585e-01 -7.13693917e-01 3.03230315e-01
-1.31526232e+00 -9.73441005e-01 -1.86458662e-01 2.07535958e+00
8.37275982e-01 -2.88871288e-01 9.33046862e-02 4.95211780e-02
6.40413225e-01 1.61098868e-01 -6.56091392e-01 -1.11008935e-01
-1.77110583e-01 4.91651326e-01 4.88876879e-01 4.34746951e-01
-1.23603785e+00 8.07243586e-01 6.06189489e+00 7.53947496e-01
-1.22619152e+00 7.49400318e-01 8.56257439e-01 -6.29846573e-01
2.84720719e-01 -6.05161846e-01 -1.07871652e+00 4.84696507e-01
9.19353604e-01 2.68447489e-01 4.89980578e-01 5.37466407e-01
4.46123987e-01 -3.04417223e-01 -8.25602889e-01 1.47414720e+00
4.75155085e-01 -9.81198430e-01 -3.89479578e-01 -4.65461463e-02
7.06352890e-01 5.06686687e-01 3.11781228e-01 -1.17425352e-01
-1.28297180e-01 -7.03553557e-01 4.45537060e-01 5.35067141e-01
1.44736624e+00 -6.44809008e-01 4.66462493e-01 -2.02630386e-01
-1.27319920e+00 -5.00465278e-03 -5.17145813e-01 2.27843881e-01
1.94219336e-01 8.51583242e-01 -4.25955236e-01 2.84509897e-01
1.28633058e+00 7.55818784e-01 -7.74797857e-01 1.01758063e+00
-3.52431267e-01 5.62587440e-01 -4.09201473e-01 4.62431729e-01
-3.80297065e-01 1.76429413e-02 3.20799530e-01 1.45558465e+00
3.18897635e-01 4.55553710e-01 -1.87832028e-01 5.40323317e-01
-3.79449904e-01 -1.00959048e-01 -3.50305527e-01 2.73377180e-01
1.00857086e-01 1.73708642e+00 -3.79547417e-01 -1.49245754e-01
-8.74320626e-01 1.08351755e+00 6.45333529e-02 2.87595063e-01
-9.29619908e-01 -4.25902247e-01 7.40640998e-01 2.82534897e-01
1.86697304e-01 -1.24187373e-01 -4.54041511e-02 -1.13306034e+00
1.38476983e-01 -7.50717461e-01 3.49849731e-01 -1.15469468e+00
-1.09165335e+00 6.86136663e-01 -3.26007724e-01 -8.21389973e-01
3.36339772e-01 -7.87079096e-01 -3.09563577e-01 7.24061787e-01
-2.17533278e+00 -1.36687160e+00 -7.65446424e-01 8.28212082e-01
3.96988153e-01 1.80129573e-01 3.87359440e-01 9.38362539e-01
-8.71464968e-01 7.97814667e-01 5.24741709e-02 1.19629651e-02
1.01183152e+00 -7.42049634e-01 2.62839764e-01 1.23596036e+00
-1.90626189e-01 3.27670395e-01 4.68559682e-01 -4.48696852e-01
-1.75356901e+00 -1.38021564e+00 7.76365578e-01 -4.76617903e-01
3.84274542e-01 -6.17772222e-01 -6.51571572e-01 5.41718900e-01
1.90269321e-01 5.11492431e-01 7.45351970e-01 -1.55053824e-01
-2.70832777e-01 -5.05536258e-01 -1.24824297e+00 2.76829571e-01
1.30596125e+00 -7.10294664e-01 -1.83830708e-02 4.62716460e-01
6.15172446e-01 -3.96736175e-01 -7.13744760e-01 5.49144268e-01
6.35042429e-01 -1.36856985e+00 1.24768496e+00 -1.17577044e-02
4.08589125e-01 -2.65395045e-01 -2.47361377e-01 -9.65686262e-01
-4.56900835e-01 -7.00317025e-01 -2.22272456e-01 1.40331280e+00
5.72432354e-02 -5.85412264e-01 5.30781150e-01 6.77929699e-01
-1.55471608e-01 -6.87255561e-01 -8.09679449e-01 -5.03260255e-01
-6.45076811e-01 -8.05985272e-01 5.18398881e-01 5.10825753e-01
-3.52679372e-01 2.96115726e-01 -5.48767090e-01 3.99396241e-01
1.11178207e+00 1.49438053e-01 6.83983922e-01 -7.40495980e-01
-3.39465380e-01 2.41401717e-02 -2.14511082e-01 -9.31422412e-01
1.20309172e-02 -7.43035555e-01 1.08380891e-01 -1.49073160e+00
6.09933913e-01 -9.45009440e-02 -3.35903943e-01 5.84136486e-01
-2.88031429e-01 1.00933611e+00 1.05903946e-01 -6.10886030e-02
-9.48643386e-01 5.83486378e-01 1.05542922e+00 1.00447245e-01
2.99564414e-02 -3.84315163e-01 -6.97644770e-01 6.73349023e-01
7.84059227e-01 -1.92465767e-01 -1.27942875e-01 -8.16433907e-01
2.08770350e-01 -3.71604174e-01 4.58157957e-01 -1.14029312e+00
4.16707724e-01 2.32459024e-01 5.98062813e-01 -5.75735450e-01
6.72956824e-01 -6.22845292e-01 -1.86169997e-01 2.06447840e-01
3.84905301e-02 -1.85982943e-01 3.39158207e-01 2.72368312e-01
9.20786932e-02 2.37289518e-01 1.29768169e+00 1.99205145e-01
-7.54553854e-01 6.81197941e-01 -2.74501052e-02 -6.60201982e-02
8.82472098e-01 3.67798135e-02 -6.52040064e-01 -1.15669124e-01
-2.81409204e-01 -5.70784025e-02 4.77753907e-01 2.52910465e-01
9.09897983e-01 -1.17777300e+00 -6.74154103e-01 4.75278646e-01
1.60416871e-01 -2.03666747e-01 5.66185951e-01 1.11632740e+00
-9.02671218e-02 3.71364534e-01 -2.13929400e-01 -3.92665476e-01
-1.79528558e+00 4.51967776e-01 2.53283501e-01 1.19089246e-01
-5.62732160e-01 1.13167429e+00 4.28099215e-01 -4.86102663e-02
1.10832185e-01 -2.69090533e-01 -3.69688645e-02 -1.12333305e-01
1.22039974e+00 4.73974138e-01 3.83780926e-01 -7.58003056e-01
-3.52474183e-01 7.08057523e-01 -1.94644518e-02 1.91761181e-01
1.73784482e+00 -4.91776705e-01 -2.84843326e-01 3.16650681e-02
1.30228651e+00 6.28442392e-02 -1.31383848e+00 -2.91960806e-01
-5.77893734e-01 -9.24427271e-01 5.70250571e-01 -6.95052266e-01
-1.36490309e+00 9.95328307e-01 1.23607767e+00 -2.56601483e-01
1.93828833e+00 2.07605451e-01 1.01662552e+00 1.88330691e-02
2.44661570e-01 -1.08383667e+00 3.07788461e-01 3.29755396e-01
8.08788955e-01 -1.60999382e+00 -2.24766992e-02 -5.41019619e-01
-1.66995212e-01 7.34171450e-01 7.03272581e-01 5.28507173e-01
6.28101766e-01 6.64602458e-01 1.11293991e-03 -3.54877383e-01
-6.26575530e-01 -5.27475357e-01 2.04112306e-01 6.25575364e-01
5.73509753e-01 -2.47874051e-01 5.96023947e-02 4.97028708e-01
-2.12521013e-02 2.44877502e-01 2.42265791e-01 7.36014366e-01
-4.02975649e-01 -8.93835187e-01 -6.03824437e-01 4.25753593e-01
-6.90498710e-01 -3.90188456e-01 -9.15393755e-02 2.14601919e-01
4.75868434e-01 1.20240283e+00 -4.89327684e-03 -3.76258194e-01
2.87643611e-01 -1.74047783e-01 7.37491012e-01 -4.34647918e-01
-2.83910453e-01 1.54162988e-01 -1.13079496e-01 -9.69368577e-01
-6.06939971e-01 -6.93708479e-01 -8.63444567e-01 -3.70608270e-01
-3.03575963e-01 -4.56858248e-01 7.41086483e-01 4.48519021e-01
7.56641746e-01 4.00008827e-01 5.27176976e-01 -1.12929714e+00
4.88812104e-02 -9.05495226e-01 -4.82548475e-01 4.97616738e-01
4.73260045e-01 -6.65592134e-01 -2.17481464e-01 1.31068185e-01] | [10.807023048400879, -2.381856918334961] |
829405d0-9e18-4097-b571-ec2ee994bafd | learning-to-segment-every-referring-object | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Qu_Learning_To_Segment_Every_Referring_Object_Point_by_Point_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Qu_Learning_To_Segment_Every_Referring_Object_Point_by_Point_CVPR_2023_paper.pdf | Learning To Segment Every Referring Object Point by Point | Referring Expression Segmentation (RES) can facilitate pixel-level semantic alignment between vision and language. Most of the existing RES approaches require massive pixel-level annotations, which are expensive and exhaustive. In this paper, we propose a new partially supervised training paradigm for RES, i.e., training using abundant referring bounding boxes and only a few (e.g., 1%) pixel-level referring masks. To maximize the transferability from the REC model, we construct our model based on the point-based sequence prediction model. We propose the co-content teacher-forcing to make the model explicitly associate the point coordinates (scale values) with the referred spatial features, which alleviates the exposure bias caused by the limited segmentation masks. To make the most of referring bounding box annotations, we further propose the resampling pseudo points strategy to select more accurate pseudo-points as supervision. Extensive experiments show that our model achieves 52.06% in terms of accuracy (versus 58.93% in fully supervised setting) on RefCOCO+@testA, when only using 1% of the mask annotations. | ['Yao Zhao', 'Xiaodan Liang', 'Wu Liu', 'Yunchao Wei', 'Yu Wu', 'Mengxue Qu'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['referring-expression', 'referring-expression-segmentation'] | ['computer-vision', 'computer-vision'] | [ 3.71453673e-01 1.71676472e-01 -3.75828207e-01 -5.07286191e-01
-8.80401134e-01 -5.19669592e-01 3.44511539e-01 -3.51237684e-01
-6.10615849e-01 5.40050209e-01 -1.63430318e-01 -1.58849359e-01
4.37610954e-01 -5.92135787e-01 -1.16593850e+00 -5.35769343e-01
5.14080763e-01 1.66895241e-01 5.41771054e-01 -1.18191965e-01
2.59319454e-01 3.82079691e-01 -1.52541077e+00 1.38003856e-01
1.31500816e+00 1.05150664e+00 5.85912168e-01 3.51480186e-01
-4.78985846e-01 6.31566882e-01 -6.74749494e-01 -3.71364474e-01
3.66985023e-01 -2.99417973e-01 -7.59089589e-01 4.06427920e-01
7.29767144e-01 -3.06842864e-01 -5.46427257e-03 1.23429561e+00
2.92575508e-01 1.70276061e-01 4.84338880e-01 -1.13045275e+00
-6.40784562e-01 3.79268229e-01 -1.16691649e+00 1.75659135e-01
1.30320460e-01 3.31314057e-01 9.55897033e-01 -9.48572278e-01
6.08897805e-01 1.01057279e+00 7.39888549e-01 5.28330803e-01
-1.14827514e+00 -7.52539098e-01 6.03792489e-01 2.24467203e-01
-1.68792927e+00 -4.85812694e-01 9.88561749e-01 -4.45492387e-01
6.77508831e-01 2.17044264e-01 5.22337198e-01 7.33731866e-01
-3.78320694e-01 8.28005135e-01 1.26429439e+00 -3.74798089e-01
6.71924353e-02 1.46908596e-01 3.19851637e-02 8.14712584e-01
-2.35803381e-01 -3.45700115e-01 -3.98553878e-01 1.35665625e-01
1.03452039e+00 -4.66994457e-02 -3.22702438e-01 -3.52670133e-01
-1.18462777e+00 5.52143574e-01 6.96222425e-01 1.85968727e-01
-1.89967304e-01 2.50227302e-01 1.69519722e-01 -2.51875162e-01
6.02814317e-01 2.11385369e-01 -6.31151915e-01 1.19514344e-02
-1.05874801e+00 -2.05698505e-01 9.32430029e-02 1.40635860e+00
9.77394700e-01 -6.46857768e-02 -2.33400762e-01 1.16233158e+00
1.32942230e-01 4.85436201e-01 3.66556406e-01 -1.15115368e+00
6.76741064e-01 6.44092858e-01 1.24313921e-01 -9.22738373e-01
-1.32229522e-01 -5.36628902e-01 -6.18194997e-01 -1.68451324e-01
4.91954505e-01 -1.26532856e-02 -1.14860249e+00 1.72771943e+00
4.50987518e-01 4.12565053e-01 -2.02338696e-01 1.10536528e+00
6.62145615e-01 6.12308085e-01 2.39127800e-01 -1.66712329e-01
1.31602228e+00 -1.24821997e+00 -4.74613726e-01 -2.96009898e-01
7.89987206e-01 -7.77094543e-01 1.68054020e+00 1.02342658e-01
-1.00741518e+00 -7.01315522e-01 -7.94006050e-01 -3.06745201e-01
-1.99170426e-01 5.11943042e-01 3.43083680e-01 3.78085881e-01
-9.68713582e-01 3.54186952e-01 -7.34951973e-01 -1.52662009e-01
7.36498654e-01 2.16850698e-01 -5.73103614e-02 -1.62034124e-01
-9.02606010e-01 4.87829983e-01 2.37302259e-01 -3.03225107e-02
-3.16129804e-01 -7.46093869e-01 -6.89776480e-01 -1.09394185e-01
5.23051620e-01 -4.98269081e-01 1.02906883e+00 -1.24184072e+00
-1.35787845e+00 1.12784207e+00 -4.58080262e-01 -3.09714109e-01
7.66670287e-01 -2.98961490e-01 1.50651634e-01 3.42591703e-01
3.70643109e-01 1.35617399e+00 8.09715092e-01 -1.40314639e+00
-7.16759384e-01 -2.92873770e-01 1.35051757e-01 3.11467588e-01
-2.79447198e-01 -3.23414318e-02 -8.47225189e-01 -7.51613736e-01
3.95105541e-01 -1.01924717e+00 -2.84502685e-01 3.28902245e-01
-5.75605750e-01 -3.65936041e-01 8.65402699e-01 -8.38238478e-01
8.84267986e-01 -2.08882737e+00 -1.65643960e-01 5.36675826e-02
7.63511751e-03 1.38717219e-01 -5.85326925e-02 -2.68358201e-01
-5.75250760e-02 1.51125982e-01 -5.06880164e-01 -4.17998642e-01
-3.02969486e-01 2.71727920e-01 -2.49608606e-01 3.78532320e-01
2.30915427e-01 9.19930339e-01 -8.73595417e-01 -9.26809967e-01
4.46877509e-01 2.87774593e-01 -4.71495658e-01 9.58573073e-02
-5.93361616e-01 6.99810445e-01 -6.69854105e-01 6.15491807e-01
9.14454281e-01 -4.95453984e-01 -2.11229429e-01 -5.96467137e-01
-1.95354074e-01 6.82554767e-02 -9.93881404e-01 1.90985441e+00
-5.20355642e-01 4.60401058e-01 -1.38466746e-01 -8.19638073e-01
9.09170806e-01 -2.13445947e-02 5.77098370e-01 -6.90477431e-01
5.17925210e-02 1.82543397e-01 -1.91026121e-01 -4.13322985e-01
4.31472868e-01 3.59366536e-01 1.53147534e-01 7.51349330e-02
-2.47812599e-01 -2.17595950e-01 -5.29341884e-02 1.73974223e-02
7.21332073e-01 5.79999924e-01 2.92964429e-01 -1.84630841e-01
6.43528402e-01 1.57294095e-01 7.64507532e-01 5.04209995e-01
-4.88898993e-01 8.98621798e-01 2.73286939e-01 -1.24086648e-01
-1.17903888e+00 -6.14680767e-01 -2.03856975e-01 1.14756596e+00
5.12418866e-01 -3.12099308e-01 -1.06616378e+00 -8.92578423e-01
-4.55506027e-01 7.09681392e-01 -3.61194283e-01 2.69713700e-01
-7.28215516e-01 -5.97101748e-01 4.44203734e-01 7.88469791e-01
9.27523494e-01 -8.58677685e-01 -5.63115835e-01 -2.16511101e-01
-3.29741240e-01 -1.44217253e+00 -7.76010752e-01 4.79524620e-02
-7.62865126e-01 -8.51337612e-01 -9.97404337e-01 -8.17484379e-01
1.00490308e+00 3.65189523e-01 8.61635089e-01 1.42674789e-01
1.01513907e-01 1.13650061e-01 -4.43332374e-01 9.00371745e-03
7.63640553e-02 8.14856887e-02 -3.08863491e-01 -9.92885306e-02
3.04585993e-01 -4.98801112e-01 -8.44237447e-01 5.40845394e-01
-6.97238505e-01 6.29387677e-01 5.86655736e-01 9.04171288e-01
1.19910645e+00 -4.19564515e-01 4.02353048e-01 -9.15001690e-01
-2.10578050e-02 -1.10576354e-01 -5.66569448e-01 2.51667380e-01
-4.01717424e-01 -1.08125493e-01 6.72305942e-01 -6.25300288e-01
-1.02298164e+00 4.79395181e-01 -1.74872771e-01 -8.02533507e-01
-3.54598939e-01 1.26139382e-02 -2.81213492e-01 -3.00887734e-01
4.76668000e-01 3.69122773e-01 -3.04675967e-01 -4.34256673e-01
7.73357034e-01 7.32829630e-01 6.36308789e-01 -7.24824071e-01
6.67240143e-01 4.90529299e-01 -2.66371340e-01 -6.91779435e-01
-1.14805400e+00 -5.27841449e-01 -8.10838103e-01 -1.29564241e-01
1.07786238e+00 -8.94872665e-01 -4.05216008e-01 2.46792376e-01
-1.26046765e+00 -4.18776155e-01 -3.25088680e-01 1.53851315e-01
-6.73953474e-01 4.53703880e-01 -4.31519419e-01 -6.98378921e-01
-2.58425117e-01 -1.21085215e+00 1.37292862e+00 4.89489168e-01
4.11085598e-03 -4.82880175e-01 -4.40440565e-01 5.87848544e-01
8.12006742e-02 1.48860127e-01 5.27344704e-01 -4.59078699e-01
-7.72197068e-01 2.72100925e-01 -7.21805155e-01 2.93641746e-01
1.37335598e-01 -1.43020779e-01 -1.17825007e+00 1.45983770e-01
-1.02698289e-01 -2.12399408e-01 6.35361791e-01 3.25496167e-01
1.96929574e+00 -1.25261515e-01 -3.67925912e-01 8.69642198e-01
1.13455153e+00 1.87871337e-01 4.09404397e-01 3.24694395e-01
1.23705447e+00 6.15025401e-01 9.84459817e-01 2.96663284e-01
4.84921515e-01 9.04613674e-01 4.13519233e-01 -2.81756103e-01
-3.30801308e-01 -3.78758520e-01 -3.41745317e-02 6.01143003e-01
-1.91024337e-02 -2.47102696e-02 -9.48790193e-01 4.98537689e-01
-1.73883736e+00 -5.67364037e-01 -1.35585532e-01 1.98958004e+00
1.14807653e+00 4.64133471e-02 -3.88462692e-02 -1.44593343e-01
9.46535468e-01 1.45418078e-01 -7.78784811e-01 1.40001863e-01
-2.13526979e-01 5.02886251e-02 8.66665661e-01 3.84436101e-01
-1.11503839e+00 1.34823406e+00 5.43677616e+00 1.23285854e+00
-1.08404982e+00 2.15051159e-01 1.07229066e+00 6.35088012e-02
-2.08477646e-01 -1.01576164e-01 -8.31116021e-01 6.87851071e-01
4.83669937e-01 2.90790915e-01 2.51790375e-01 9.65108812e-01
4.26940382e-01 -1.72086865e-01 -8.99394989e-01 1.28051102e+00
1.04469508e-01 -1.17527652e+00 1.61394030e-02 -6.53554872e-02
9.63611066e-01 -1.83164515e-02 -2.47146152e-02 1.41820787e-02
7.81126991e-02 -8.86003852e-01 8.33564937e-01 5.56615710e-01
1.05543685e+00 -5.87334931e-01 5.81374109e-01 4.48108554e-01
-1.19552028e+00 2.18230575e-01 -2.88440853e-01 3.48942846e-01
1.84416145e-01 4.55435514e-01 -5.73116720e-01 4.85142618e-01
8.41195405e-01 7.80867994e-01 -6.50791466e-01 8.00605237e-01
-2.08673060e-01 6.79232359e-01 -5.94404161e-01 1.58821791e-01
1.68810964e-01 -3.20133060e-01 3.16765815e-01 1.09649575e+00
1.48545504e-01 9.57103893e-02 4.49202716e-01 1.04640770e+00
-2.43106693e-01 2.72113323e-01 -1.69635147e-01 3.74389112e-01
7.53584504e-01 1.16656959e+00 -8.50834608e-01 -3.66954625e-01
-4.47813630e-01 1.26425159e+00 4.11869645e-01 4.74527061e-01
-1.07885087e+00 -1.31253973e-01 2.66656667e-01 1.92643046e-01
4.34334517e-01 -1.79932415e-01 -6.15411878e-01 -9.61610436e-01
2.72377521e-01 -5.94731212e-01 1.56100422e-01 -1.23054338e+00
-1.25320864e+00 4.95829284e-01 8.79315063e-02 -1.31563628e+00
-3.99747007e-02 -4.06322598e-01 -4.63378727e-01 1.00260663e+00
-1.66604352e+00 -1.39393568e+00 -5.98589182e-01 4.75556254e-01
7.61775970e-01 2.57112712e-01 3.36219341e-01 2.86313772e-01
-7.43634343e-01 7.48006523e-01 -2.17539489e-01 1.77678660e-01
7.77848303e-01 -1.16843891e+00 2.45763570e-01 7.70230830e-01
1.93853915e-01 6.79294586e-01 6.13651037e-01 -4.77521122e-01
-7.78109491e-01 -1.24324143e+00 6.55850530e-01 -2.79097974e-01
5.18540025e-01 -2.40680337e-01 -1.01808703e+00 5.88371098e-01
-1.65733248e-01 4.16341633e-01 2.86800057e-01 -2.27675229e-01
-1.71128452e-01 -6.99852854e-02 -1.23982251e+00 8.88174832e-01
1.37711573e+00 -5.21510303e-01 -4.53316450e-01 4.28760439e-01
1.05011857e+00 -6.30561471e-01 -6.97693527e-01 3.75352859e-01
1.72152042e-01 -6.99407458e-01 1.03485584e+00 -1.21969983e-01
4.15714294e-01 -7.27643907e-01 -2.24587306e-01 -8.98642182e-01
1.31083295e-01 -3.72995555e-01 -8.06555338e-03 1.52229667e+00
2.01693818e-01 -4.23576117e-01 9.88446474e-01 5.36155283e-01
-2.69984663e-01 -1.06957066e+00 -9.75819528e-01 -5.86709142e-01
-9.49494839e-02 -4.65407312e-01 5.53204536e-01 1.00885355e+00
-5.72481751e-01 2.10042194e-01 -1.67912796e-01 1.80541664e-01
3.90718162e-01 9.40255150e-02 7.99997568e-01 -8.02326560e-01
-2.03058347e-01 -3.71768296e-01 -2.24880919e-01 -1.72702289e+00
2.84930050e-01 -6.56741202e-01 2.51172334e-01 -1.30477130e+00
7.46234879e-02 -8.79817128e-01 -1.84278816e-01 5.85540771e-01
-4.34978783e-01 4.46259201e-01 1.11265317e-01 5.99611819e-01
-6.58308983e-01 5.89743733e-01 1.50807416e+00 -6.21171221e-02
-2.65822917e-01 -1.31839931e-01 -4.58488017e-01 1.06038249e+00
8.27877223e-01 -2.40391225e-01 -4.48882848e-01 -5.03024340e-01
-1.89395770e-01 -2.03024343e-01 5.35678148e-01 -8.33035827e-01
1.45826668e-01 -4.25945342e-01 4.43552166e-01 -9.35986042e-01
3.63328367e-01 -8.01532805e-01 -2.14067638e-01 -1.01998284e-01
-4.20346797e-01 -1.15998663e-01 8.36516842e-02 5.73679745e-01
-3.58672477e-02 -3.32239807e-01 8.91555011e-01 2.86459760e-03
-9.15817320e-01 3.41649771e-01 3.36834818e-01 2.32031737e-02
1.04898047e+00 -4.18677896e-01 -2.26231679e-01 -1.53229296e-01
-5.89405298e-01 3.20271999e-01 6.73828006e-01 1.26180157e-01
4.01640892e-01 -1.08235240e+00 -1.64810449e-01 -1.27741843e-01
2.39896446e-01 5.83906889e-01 2.99689323e-01 9.38747525e-01
-4.90012228e-01 3.88073623e-01 3.97470221e-03 -9.49082375e-01
-1.26280725e+00 5.59316039e-01 4.46266353e-01 3.90681885e-02
-8.89417231e-01 9.84515548e-01 4.55844343e-01 -2.93371499e-01
1.53138325e-01 -3.94105077e-01 -1.56839550e-01 -3.48868400e-01
1.90972507e-01 1.47282720e-01 -2.54364640e-01 -8.79443884e-01
-2.88157225e-01 9.60808575e-01 -1.25338107e-01 -1.71797857e-01
8.33072126e-01 -3.42949986e-01 1.09360263e-01 4.46019888e-01
1.00649130e+00 1.73517302e-01 -1.75644767e+00 -3.74112040e-01
-9.46948305e-02 -5.08531988e-01 9.37239826e-02 -7.07402110e-01
-1.14667058e+00 9.04615939e-01 6.25379443e-01 -1.35483414e-01
1.17813814e+00 1.03476606e-01 9.08593774e-01 1.10743053e-01
3.35809410e-01 -1.11195385e+00 1.42127899e-02 2.84925461e-01
6.05212927e-01 -1.23903227e+00 -1.05072193e-01 -1.07767844e+00
-6.47762716e-01 7.56678462e-01 9.89761829e-01 -2.07906097e-01
1.69089511e-01 -3.42357112e-03 1.41474664e-01 7.47556016e-02
-2.60280520e-01 -4.36482817e-01 3.04704189e-01 6.65501893e-01
5.22768140e-01 -8.43948722e-02 -4.41553444e-01 5.47138691e-01
-3.55000556e-01 -7.79588297e-02 2.00105697e-01 7.14759231e-01
-4.50751483e-01 -9.06616628e-01 -3.25717598e-01 3.92635971e-01
-2.75415748e-01 -2.97020882e-01 -2.17595369e-01 6.16630316e-01
3.29401404e-01 7.23132670e-01 2.32293487e-01 -3.18386793e-01
2.48182192e-01 1.95187498e-02 3.32555890e-01 -5.83876729e-01
-2.31139079e-01 2.55327851e-01 -3.54613364e-02 -5.67214549e-01
-6.56262398e-01 -5.10580480e-01 -1.61678898e+00 4.08057272e-02
-4.35860664e-01 -1.40778407e-01 4.96820420e-01 1.10348010e+00
3.87101054e-01 6.59954607e-01 4.96896774e-01 -7.81862140e-01
-3.90016347e-01 -9.53207135e-01 -1.90168113e-01 4.24626440e-01
7.59047493e-02 -7.55651593e-01 -2.11924836e-01 3.86476725e-01] | [9.73613452911377, 0.6888526678085327] |
64269269-52ba-4613-99d1-a8a33e0dad8b | recurrent-squeeze-and-excitation-context | 1807.05698 | null | http://arxiv.org/abs/1807.05698v2 | http://arxiv.org/pdf/1807.05698v2.pdf | Recurrent Squeeze-and-Excitation Context Aggregation Net for Single Image Deraining | Rain streaks can severely degrade the visibility, which causes many current
computer vision algorithms fail to work. So it is necessary to remove the rain
from images. We propose a novel deep network architecture based on deep
convolutional and recurrent neural networks for single image deraining. As
contextual information is very important for rain removal, we first adopt the
dilated convolutional neural network to acquire large receptive field. To
better fit the rain removal task, we also modify the network. In heavy rain,
rain streaks have various directions and shapes, which can be regarded as the
accumulation of multiple rain streak layers. We assign different alpha-values
to various rain streak layers according to the intensity and transparency by
incorporating the squeeze-and-excitation block. Since rain streak layers
overlap with each other, it is not easy to remove the rain in one stage. So we
further decompose the rain removal into multiple stages. Recurrent neural
network is incorporated to preserve the useful information in previous stages
and benefit the rain removal in later stages. We conduct extensive experiments
on both synthetic and real-world datasets. Our proposed method outperforms the
state-of-the-art approaches under all evaluation metrics. Codes and
supplementary material are available at our project webpage:
https://xialipku.github.io/RESCAN . | ['Hongbin Zha', 'Hong Liu', 'Zhouchen Lin', 'Xia Li', 'Jianlong Wu'] | 2018-07-16 | recurrent-squeeze-and-excitation-context-1 | http://openaccess.thecvf.com/content_ECCV_2018/html/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf | eccv-2018-9 | ['single-image-deraining'] | ['computer-vision'] | [-5.50030917e-02 -6.06310248e-01 3.84042472e-01 -5.15235484e-01
-7.61357620e-02 -3.54775995e-01 -2.07757577e-03 -4.44558740e-01
-3.46709371e-01 8.14538300e-01 1.29391700e-01 -3.42733115e-01
3.30875784e-01 -8.57145488e-01 -6.09810174e-01 -1.19950557e+00
1.26390412e-01 -2.75330633e-01 3.65238369e-01 -3.93401831e-01
6.27434775e-02 5.75301826e-01 -1.47009385e+00 1.35355696e-01
1.28421402e+00 4.53489572e-01 6.49632812e-01 6.33696735e-01
-1.41295865e-01 9.10017729e-01 -6.71976626e-01 1.40791044e-01
2.95159340e-01 -7.04395950e-01 5.18293446e-03 -7.45184571e-02
4.54122186e-01 -6.62795842e-01 -4.69448328e-01 1.28304303e+00
6.46211565e-01 1.38950661e-01 2.49215618e-01 -5.25634646e-01
-6.88333631e-01 2.25543469e-01 -8.76913607e-01 6.85844541e-01
-3.02161604e-01 2.22403064e-01 5.64114094e-01 -1.03450537e+00
1.48401007e-01 1.10135412e+00 4.49970096e-01 4.15709496e-01
-6.90001726e-01 -9.19182241e-01 3.49620104e-01 1.84234560e-01
-1.07872880e+00 -5.02953589e-01 6.76172018e-01 -1.37779877e-01
3.73684675e-01 4.96165276e-01 7.99409091e-01 7.16900289e-01
1.94669276e-01 7.34359920e-01 1.26563156e+00 -9.08933580e-02
-8.61957222e-02 -1.82586178e-01 3.42843503e-01 5.77909648e-01
7.82554030e-01 4.78450470e-02 1.44454300e-01 3.93446147e-01
9.18404281e-01 5.46940923e-01 -8.16065013e-01 1.85135737e-01
-8.15941453e-01 8.15294802e-01 9.69020128e-01 2.07573667e-01
-4.84592468e-01 -2.69985534e-02 3.82309519e-02 4.41811562e-01
5.80608547e-01 2.44503245e-01 -2.52334028e-01 4.95168239e-01
-1.00693142e+00 4.27543253e-01 4.41640884e-01 3.47901374e-01
9.51072514e-01 4.14932609e-01 -1.45774439e-01 1.01532996e+00
3.47093076e-01 1.12728775e+00 -1.13839796e-03 -6.77196383e-01
5.32391131e-01 9.27788019e-02 4.26068932e-01 -9.29952323e-01
-4.37001377e-01 -6.20798171e-01 -1.44140208e+00 5.21567047e-01
1.42825454e-01 -3.97401184e-01 -1.46690834e+00 1.18302190e+00
1.89280555e-01 5.42920589e-01 2.78992802e-01 1.58541369e+00
1.14117920e+00 1.12981296e+00 -5.44610694e-02 -6.66445613e-01
1.09931278e+00 -1.04342961e+00 -8.94887149e-01 -5.56784391e-01
1.90038130e-01 -9.38978791e-01 8.95966828e-01 3.04127812e-01
-9.07365561e-01 -5.24913192e-01 -1.10957956e+00 5.35807461e-02
-1.60088483e-02 2.39811271e-01 6.52505457e-01 1.81935474e-01
-7.77490318e-01 3.96138698e-01 -7.88796186e-01 -5.38609885e-02
4.38522339e-01 -1.37487113e-01 1.41490281e-01 -5.33738554e-01
-1.47780716e+00 6.75761521e-01 1.10879675e-01 1.27612102e+00
-8.76628697e-01 -2.49632806e-01 -6.62553966e-01 -1.42516777e-01
1.39523923e-01 -5.42131007e-01 8.20137024e-01 -1.08497787e+00
-1.12518406e+00 3.74080300e-01 -3.20766956e-01 -9.83387157e-02
2.83911407e-01 -6.23755395e-01 -4.89079922e-01 3.04115582e-02
-1.68852404e-01 1.09809466e-01 1.17559147e+00 -1.59327853e+00
-7.02280045e-01 -1.39012158e-01 1.00776725e-01 5.85477114e-01
1.74537152e-01 -3.70082334e-02 -4.96802032e-01 -7.14373708e-01
2.84244299e-01 -8.10001373e-01 -4.76047188e-01 -2.67566204e-01
-3.05147111e-01 4.25265133e-01 9.18602288e-01 -9.85567033e-01
1.30232966e+00 -2.10235381e+00 9.91771668e-02 -4.60881554e-02
4.51135486e-01 8.63538504e-01 -2.20328912e-01 -9.50008780e-02
-8.80654231e-02 -5.43065593e-02 -5.45011878e-01 -2.25669146e-01
-6.34821296e-01 3.67488265e-01 -5.45953631e-01 5.86287796e-01
6.48321584e-02 5.78759849e-01 -7.04406738e-01 -2.00131953e-01
3.31743598e-01 6.07683420e-01 -2.64272988e-02 4.93507773e-01
-1.58367872e-01 5.12269378e-01 -5.44311404e-01 7.10891068e-01
1.57928562e+00 1.64650127e-01 -2.41274774e-01 -1.35125145e-01
-2.89348751e-01 -5.36142476e-02 -1.13243937e+00 1.03042293e+00
-5.56149423e-01 5.71629405e-01 3.75648886e-01 -6.97198570e-01
1.00563657e+00 1.21416658e-01 -2.81384021e-01 -7.47484088e-01
4.07658592e-02 3.55698377e-01 6.49468303e-02 -8.76914561e-01
3.57238859e-01 -3.18000168e-01 4.95973885e-01 7.41654783e-02
-6.15203083e-01 1.04739636e-01 -1.19481221e-01 -8.01512506e-03
7.64307082e-01 -1.09919026e-01 -1.40323907e-01 1.75737459e-02
4.19697076e-01 -3.37042630e-01 9.30027604e-01 7.04105020e-01
-1.41198710e-01 9.01719332e-01 9.10616070e-02 -6.30948722e-01
-8.45163584e-01 -1.00607026e+00 -5.26343137e-02 8.79048824e-01
5.04024029e-01 2.54696816e-01 -4.59591657e-01 -4.22786117e-01
-9.66414213e-02 4.16782379e-01 -7.78667510e-01 4.33585979e-02
-8.28194499e-01 -1.41071141e+00 2.48726383e-01 2.67704844e-01
1.01028895e+00 -1.57012320e+00 -5.36604762e-01 1.28378868e-01
-3.21075052e-01 -8.31966519e-01 -1.53466642e-01 1.39292985e-01
-8.95991266e-01 -8.99613202e-01 -9.35149312e-01 -7.95148730e-01
5.45775354e-01 1.01481438e+00 1.01624620e+00 5.63219786e-01
-2.97094017e-01 -6.24483168e-01 -7.12839544e-01 -4.83735859e-01
8.66967738e-02 2.15005670e-02 -3.16877246e-01 -6.06649593e-02
1.38548359e-01 -7.19165266e-01 -1.03335369e+00 -6.37454465e-02
-1.11802459e+00 4.30364870e-02 8.37137222e-01 7.95205712e-01
3.70974243e-01 1.67090416e-01 1.98251441e-01 -9.42129493e-01
4.44982141e-01 -4.79725599e-01 -7.25778461e-01 -7.50826299e-02
-1.46291882e-01 -1.96621045e-01 5.65563738e-01 -2.38143820e-02
-1.39072382e+00 -1.70609280e-01 -2.10422114e-01 -6.35755718e-01
-2.50693619e-01 6.08037293e-01 -2.79150695e-01 -3.38703990e-02
3.45714152e-01 3.04061472e-01 -3.82459760e-01 -5.52034438e-01
1.70301527e-01 5.75226545e-01 3.79270613e-01 1.80885509e-01
1.11623251e+00 5.25839210e-01 -2.39589378e-01 -9.70004439e-01
-1.23969686e+00 -3.45717698e-01 -3.48229885e-01 -1.85931310e-01
8.17982674e-01 -1.11651278e+00 -3.70646358e-01 9.54223752e-01
-1.15978348e+00 -4.26603138e-01 1.86205730e-01 4.25377250e-01
4.38947737e-01 3.92698854e-01 -7.29475856e-01 -9.86564159e-01
-6.50007486e-01 -8.58590126e-01 6.23395443e-01 8.78006160e-01
6.60540044e-01 -6.73446834e-01 1.05495073e-01 5.08495644e-02
5.30246317e-01 1.17897391e-01 3.98004562e-01 2.19375983e-01
-7.78749168e-01 1.40015259e-01 -5.70590317e-01 4.88015711e-01
3.76177013e-01 2.06019878e-01 -1.10243654e+00 -4.28064018e-01
1.32578731e-01 -1.08973078e-01 1.78447938e+00 6.85251951e-01
8.90842199e-01 -2.39618883e-01 -8.26988816e-02 1.10520196e+00
1.62916470e+00 1.97917774e-01 1.15965855e+00 6.22008801e-01
1.09006679e+00 5.05394757e-01 8.02388489e-01 2.90663034e-01
8.11155811e-02 5.99367321e-02 7.16905653e-01 -6.23913050e-01
-1.83098271e-01 2.86228657e-01 2.51529425e-01 9.17280197e-01
-5.20832002e-01 -4.33259189e-01 -5.88170588e-01 5.68985403e-01
-1.80710828e+00 -1.22843099e+00 -4.41402644e-01 2.07220459e+00
6.51333332e-01 1.80295661e-01 -4.86543328e-01 -1.99609071e-01
7.12710977e-01 8.10573697e-01 -5.73999822e-01 -1.44624025e-01
-3.82967353e-01 3.27561706e-01 7.15506315e-01 7.95629978e-01
-1.27561486e+00 1.07839787e+00 4.91943836e+00 4.61026073e-01
-1.39465165e+00 1.53987454e-02 4.61722791e-01 -1.69150427e-01
-3.12145621e-01 -8.39488357e-02 -7.32678771e-01 7.31249094e-01
4.17821795e-01 4.44783360e-01 5.28260887e-01 2.25246832e-01
7.48167098e-01 -2.30862409e-01 -3.93515378e-02 7.98828781e-01
-1.17732264e-01 -9.30254459e-01 3.01023126e-02 -3.82373542e-01
6.73626781e-01 4.24523115e-01 -5.03178798e-02 2.96907604e-01
3.32525790e-01 -1.10089254e+00 1.41796857e-01 9.21716392e-01
5.81239820e-01 -7.92922437e-01 1.04558730e+00 6.31763116e-02
-1.16859066e+00 2.90718433e-02 -8.49441290e-01 -1.54544160e-01
-6.65261298e-02 1.28767359e+00 -1.92581207e-01 7.49263883e-01
1.03164887e+00 9.11530554e-01 -4.12597239e-01 1.28875315e+00
-7.11043417e-01 8.53763580e-01 -3.19513738e-01 3.53682309e-01
2.23622978e-01 -6.65702879e-01 5.49963772e-01 1.32976949e+00
2.94237494e-01 4.36232775e-01 1.56222982e-02 5.41179657e-01
-1.25515923e-01 -2.24473536e-01 -4.82959718e-01 4.67934728e-01
3.79345030e-01 1.30792534e+00 -5.40065527e-01 -3.64898741e-01
-3.28676701e-01 1.14524388e+00 -1.07282795e-01 8.27757299e-01
-8.60067308e-01 -8.93669009e-01 8.90301347e-01 -2.38878220e-01
6.03456795e-01 -1.78923085e-01 -1.50464952e-01 -1.43292391e+00
8.88682157e-02 -8.58645916e-01 3.88160646e-02 -9.27832007e-01
-1.05810702e+00 8.54076684e-01 -4.35122013e-01 -1.33300853e+00
3.97655696e-01 -1.99238643e-01 -9.97184038e-01 1.25674772e+00
-2.21952391e+00 -8.99888217e-01 -1.04793346e+00 3.59374374e-01
5.36155701e-01 2.83347636e-01 4.25885230e-01 4.15006936e-01
-9.33792949e-01 1.43838912e-01 4.94181693e-01 7.94902444e-02
9.74802792e-01 -9.76465702e-01 4.29184020e-01 1.36886477e+00
-2.78651804e-01 4.22557145e-01 1.01122153e+00 -7.57425129e-01
-9.19643283e-01 -1.48392296e+00 6.20176971e-01 1.68043479e-01
2.78964907e-01 -1.32560968e-01 -1.54517531e+00 5.25292695e-01
1.95621744e-01 3.95510048e-01 1.52117396e-02 -8.38124845e-03
-3.18679005e-01 -4.89092588e-01 -7.46922791e-01 5.40104151e-01
7.51632690e-01 -1.58407748e-01 -5.26340008e-01 2.92449266e-01
7.47257173e-01 -5.16432703e-01 -1.01599023e-01 6.32628262e-01
3.25515360e-01 -1.29178298e+00 7.61076868e-01 -6.43422380e-02
5.68774223e-01 -6.49966061e-01 -3.70123098e-03 -1.67365706e+00
-5.31344771e-01 -2.48061270e-01 -2.51183286e-02 9.85888004e-01
1.50756598e-01 -7.51670659e-01 5.84961534e-01 -1.83821112e-01
-2.60990381e-01 -6.25732124e-01 -4.62188452e-01 -2.12437928e-01
-1.21836327e-01 1.63907334e-01 4.10885036e-01 1.02967727e+00
-8.85060906e-01 4.15669024e-01 -6.94936335e-01 8.56035471e-01
7.32574463e-01 6.77321017e-01 5.63408315e-01 -1.17627978e+00
-7.26362616e-02 -1.42041057e-01 1.44945666e-01 -1.07915735e+00
-2.06776932e-01 -3.30954969e-01 5.01245320e-01 -1.81138420e+00
2.12695092e-01 -4.78941530e-01 -6.36873782e-01 6.11063182e-01
-8.04793954e-01 4.56211507e-01 1.39859572e-01 6.61220491e-01
-4.02808666e-01 8.90443385e-01 1.63686800e+00 -7.06900582e-02
-4.98839855e-01 1.32214680e-01 -5.27787030e-01 8.69757593e-01
1.16063893e+00 -3.78566891e-01 -1.71893284e-01 -9.50803697e-01
1.26828685e-01 1.92957576e-02 3.73116046e-01 -1.08667552e+00
-1.25292093e-01 -3.08252007e-01 6.17129207e-01 -4.90811884e-01
2.56060243e-01 -5.90770006e-01 -1.26527384e-01 4.65488404e-01
1.15696423e-01 -2.96654880e-01 3.50386471e-01 4.76551533e-01
-2.96035171e-01 -1.44685447e-01 1.02449203e+00 -2.12148324e-01
-7.04243243e-01 4.35610235e-01 -4.15452063e-01 -3.27955246e-01
5.63448846e-01 1.11167543e-01 -6.37117803e-01 -3.00259501e-01
-6.66995883e-01 4.81006682e-01 3.02839398e-01 2.01029405e-01
9.11836326e-01 -8.13044131e-01 -1.11450434e+00 2.56375164e-01
-2.28729188e-01 2.31211647e-01 7.09193885e-01 7.92248309e-01
-9.57808018e-01 -1.07943997e-01 -3.07871819e-01 -1.41435608e-01
-1.47381222e+00 2.39171371e-01 6.65841341e-01 -1.07736215e-01
-8.32378447e-01 9.28677738e-01 6.70142710e-01 -1.40642449e-01
-3.34254056e-02 -3.86874527e-01 -5.65685213e-01 -1.47137642e-02
7.08750546e-01 1.63670421e-01 -2.03144625e-02 -2.10515067e-01
-1.11513689e-01 5.07589698e-01 -4.15164024e-01 3.72060984e-01
1.45045507e+00 -3.88258636e-01 -3.82474750e-01 4.88116950e-01
7.76749253e-01 1.55568749e-01 -1.29997623e+00 -1.34618551e-01
-8.00534010e-01 -5.88900983e-01 4.23948854e-01 -7.11295605e-01
-1.69305348e+00 1.08244717e+00 8.43923032e-01 1.71108976e-01
1.54249966e+00 -4.73050117e-01 9.75224257e-01 4.55284745e-01
-1.68155551e-01 -5.02915859e-01 -2.71344960e-01 8.82301331e-01
7.78564513e-01 -1.18492520e+00 2.25863531e-01 -4.25268203e-01
-5.52953601e-01 9.71843064e-01 6.59975827e-01 -4.22368139e-01
5.93060553e-01 3.28676969e-01 7.04721332e-01 -2.25223809e-01
-3.58378291e-01 -3.72927964e-01 -3.30296010e-01 3.07785481e-01
3.17742556e-01 9.48998034e-02 -3.16119134e-01 3.69304538e-01
2.49007538e-01 9.96797308e-02 7.96432018e-01 8.36920202e-01
-1.02564108e+00 -6.37728751e-01 -6.68912590e-01 5.83661437e-01
-3.17111164e-01 -5.32172740e-01 1.92741275e-01 5.14239311e-01
3.19227099e-01 9.60954964e-01 4.01528217e-02 -3.12386930e-01
2.75613368e-01 -4.72317427e-01 2.14763731e-01 -4.61151421e-01
-2.49456316e-01 2.24337146e-01 -1.00820750e-01 -3.79662961e-01
-6.24874115e-01 -5.26617289e-01 -1.14988959e+00 -3.97329867e-01
-3.11541319e-01 3.36538732e-01 1.62988886e-01 7.67286479e-01
6.54203519e-02 8.18669200e-01 9.40161228e-01 -1.01362443e+00
6.30376190e-02 -1.21305168e+00 -1.01251245e+00 7.60281517e-04
9.89317596e-01 -5.57407200e-01 -6.87561095e-01 -5.86323142e-02] | [10.899057388305664, -3.265470266342163] |
20a03ed3-f4ff-415f-bf77-eda35c3e45d3 | dq-detr-dual-query-detection-transformer-for | 2211.15516 | null | https://arxiv.org/abs/2211.15516v2 | https://arxiv.org/pdf/2211.15516v2.pdf | DQ-DETR: Dual Query Detection Transformer for Phrase Extraction and Grounding | In this paper, we study the problem of visual grounding by considering both phrase extraction and grounding (PEG). In contrast to the previous phrase-known-at-test setting, PEG requires a model to extract phrases from text and locate objects from images simultaneously, which is a more practical setting in real applications. As phrase extraction can be regarded as a $1$D text segmentation problem, we formulate PEG as a dual detection problem and propose a novel DQ-DETR model, which introduces dual queries to probe different features from image and text for object prediction and phrase mask prediction. Each pair of dual queries is designed to have shared positional parts but different content parts. Such a design effectively alleviates the difficulty of modality alignment between image and text (in contrast to a single query design) and empowers Transformer decoder to leverage phrase mask-guided attention to improve performance. To evaluate the performance of PEG, we also propose a new metric CMAP (cross-modal average precision), analogous to the AP metric in object detection. The new metric overcomes the ambiguity of Recall@1 in many-box-to-one-phrase cases in phrase grounding. As a result, our PEG pre-trained DQ-DETR establishes new state-of-the-art results on all visual grounding benchmarks with a ResNet-101 backbone. For example, it achieves $91.04\%$ and $83.51\%$ in terms of recall rate on RefCOCO testA and testB with a ResNet-101 backbone. Code will be availabl at \url{https://github.com/IDEA-Research/DQ-DETR}. | ['Lei Zhang', 'Jun Zhu', 'Hang Su', 'Hao Zhang', 'Shijia Huang', 'Feng Li', 'Yaoyuan Liang', 'Shilong Liu'] | 2022-11-28 | null | null | null | null | ['phrase-extraction-and-grounding-peg', 'phrase-grounding'] | ['computer-vision', 'natural-language-processing'] | [ 3.42428803e-01 6.07424751e-02 -3.98414820e-01 -1.61062434e-01
-1.29470921e+00 -6.50588930e-01 3.33052516e-01 -4.05350253e-02
-5.59140086e-01 4.31952745e-01 -1.78057656e-01 -4.30590153e-01
1.56119645e-01 -7.35999346e-01 -1.03863764e+00 -6.19049847e-01
3.59433502e-01 4.28789437e-01 4.34212565e-01 -9.74527374e-02
1.09442391e-01 -2.76544709e-02 -1.26631975e+00 3.99303883e-01
6.44302070e-01 1.10449553e+00 7.29512036e-01 4.92026001e-01
-2.62108088e-01 2.02855930e-01 -5.43009102e-01 -7.85461247e-01
3.49694222e-01 -3.49137694e-01 -7.37169445e-01 3.76517743e-01
9.01713014e-01 -3.54997844e-01 -3.50329369e-01 1.25259125e+00
5.47240734e-01 -1.25157595e-01 3.94957513e-01 -1.13823175e+00
-7.33582437e-01 6.23386085e-01 -1.22393513e+00 3.41437310e-01
2.17038393e-01 3.07441682e-01 1.70340919e+00 -1.25978589e+00
3.63534987e-01 1.05735278e+00 4.58455861e-01 4.90654737e-01
-1.46711540e+00 -8.52473915e-01 4.62470889e-01 9.07203555e-02
-1.60899961e+00 -1.46186039e-01 3.40171933e-01 -4.14790124e-01
9.72292304e-01 2.40505636e-01 6.13265514e-01 1.00099564e+00
5.08404113e-02 1.13487244e+00 8.66692066e-01 -2.91224688e-01
-1.38152853e-01 -2.72141881e-02 1.92277282e-01 9.30013597e-01
3.14490050e-01 -2.27843583e-01 -9.50502455e-01 1.88254341e-01
7.07100809e-01 -1.14943683e-01 -4.78969395e-01 -1.13165848e-01
-1.31023812e+00 6.50914133e-01 6.06751025e-01 3.62804905e-02
-6.06921874e-02 4.76002008e-01 1.55538740e-02 -1.52648985e-01
3.38062853e-01 3.33717138e-01 -4.21094686e-01 1.14921838e-01
-1.12093425e+00 4.23703820e-01 4.22698617e-01 1.30449200e+00
7.83852398e-01 -3.40731919e-01 -6.85141504e-01 7.80996025e-01
5.08921266e-01 8.50771725e-01 7.91691020e-02 -4.33827251e-01
9.82199371e-01 6.14368200e-01 -5.94430864e-02 -9.20805335e-01
-2.90309966e-01 -7.68863797e-01 -7.02612460e-01 -3.11996520e-01
4.12582070e-01 1.54435843e-01 -1.17543066e+00 1.73459339e+00
9.27301720e-02 3.87628555e-01 -3.94167423e-01 1.01606131e+00
1.01595390e+00 7.44580507e-01 1.09865025e-01 1.23591321e-02
1.84127605e+00 -1.11795568e+00 -5.46094775e-01 -6.08905673e-01
5.36102951e-01 -8.13801050e-01 1.49484229e+00 2.52314717e-01
-9.90617573e-01 -4.94023114e-01 -1.09684002e+00 -2.53548354e-01
-1.27257109e-01 1.11747988e-01 2.65831828e-01 4.15768117e-01
-1.12015867e+00 1.57426238e-01 -6.88943088e-01 -2.64840186e-01
6.41826928e-01 5.74793458e-01 -3.25324200e-02 -1.88163936e-01
-9.84371066e-01 4.53383625e-01 3.86044323e-01 9.74870846e-02
-7.13977993e-01 -6.63031399e-01 -6.72697961e-01 1.44409359e-01
6.57979012e-01 -9.03443813e-01 1.27573121e+00 -5.29156029e-01
-7.95474529e-01 1.19793880e+00 -2.84873456e-01 -4.98596191e-01
2.54086405e-01 -3.46393228e-01 -1.24445714e-01 1.91192001e-01
4.41242129e-01 1.24916267e+00 8.83172691e-01 -1.17735851e+00
-1.01474214e+00 -2.68593580e-01 3.01381528e-01 3.62194180e-01
-2.59520441e-01 -1.09717436e-01 -1.29028392e+00 -8.97065222e-01
3.07077527e-01 -9.78763878e-01 -9.52027738e-03 -3.47400717e-02
-8.18233430e-01 -1.36747763e-01 5.60052335e-01 -5.41032791e-01
1.39744210e+00 -2.18055105e+00 1.34808868e-01 2.04300135e-02
5.22050500e-01 1.31665513e-01 -2.35259384e-01 2.19794270e-02
1.11859947e-01 4.99458015e-01 -1.90896645e-01 -6.70267224e-01
2.40975529e-01 9.21048224e-02 -5.03996909e-01 2.10556611e-01
4.57342327e-01 1.22925735e+00 -6.64883733e-01 -6.62335694e-01
-1.30049229e-01 2.09004015e-01 -6.44098639e-01 -6.14580996e-02
-5.75540662e-01 2.18983755e-01 -4.94205236e-01 8.17522764e-01
7.47941613e-01 -7.76643455e-01 -1.10179923e-01 -2.13783652e-01
1.83240175e-01 2.43147463e-01 -9.28283691e-01 1.82247663e+00
-1.93450928e-01 7.05650270e-01 -9.94032696e-02 -7.25961447e-01
5.97566247e-01 2.18717650e-01 3.02243531e-01 -9.48805749e-01
6.60844669e-02 1.41223907e-01 -1.54949650e-01 -2.90721089e-01
6.84113383e-01 1.05317952e-02 -1.92522913e-01 2.92341918e-01
1.27123684e-01 -2.90602185e-02 2.74999857e-01 2.31183693e-01
1.17153418e+00 1.67806412e-03 -3.63839231e-02 -2.47948050e-01
2.46110350e-01 4.62551266e-02 5.96044719e-01 1.01438236e+00
-9.71212313e-02 1.00428963e+00 5.52747190e-01 -1.18888028e-01
-8.75779927e-01 -1.10329306e+00 -2.01050103e-01 1.24904370e+00
6.38369441e-01 -6.45045102e-01 -6.87408507e-01 -8.96450818e-01
-7.00448006e-02 4.68269587e-01 -4.24976349e-01 1.28124073e-01
-6.13067985e-01 -8.44358206e-01 3.69750112e-01 4.62562263e-01
7.00809836e-01 -7.97678769e-01 -4.05870378e-01 2.90246215e-02
-4.95967954e-01 -1.50270009e+00 -8.70877683e-01 1.62041962e-01
-5.96690059e-01 -8.87725949e-01 -8.04886162e-01 -9.35610056e-01
4.94380385e-01 5.33399045e-01 1.42563057e+00 1.22979902e-01
-6.73439950e-02 2.20319435e-01 -4.59472120e-01 -3.39673072e-01
2.63346553e-01 3.57910365e-01 -3.55223328e-01 -8.00661296e-02
4.63996649e-01 -2.74042368e-01 -9.56779063e-01 5.36281288e-01
-8.65324140e-01 4.94213462e-01 7.41346776e-01 9.60467279e-01
1.08171606e+00 -1.39876634e-01 2.06699178e-01 -6.28113210e-01
3.82191867e-01 -3.02937090e-01 -7.94862986e-01 2.05151156e-01
-4.64368075e-01 -1.00114495e-01 1.62260652e-01 -2.60760754e-01
-3.84417802e-01 5.33280836e-04 -1.18377201e-01 -4.57623154e-01
1.44720018e-01 4.61811274e-01 -2.60532111e-01 1.32848546e-01
3.86849940e-01 3.47371817e-01 -5.07321298e-01 -3.55152071e-01
2.91848511e-01 4.41052169e-01 3.93416733e-01 -6.42022431e-01
8.61971974e-01 3.34510505e-01 -1.70119673e-01 -5.28208673e-01
-1.19268095e+00 -6.65404618e-01 -2.15003222e-01 5.78335635e-02
1.08834255e+00 -1.15196872e+00 -5.65551221e-01 3.12652946e-01
-1.38559902e+00 -3.95887434e-01 -2.17815280e-01 2.13961065e-01
-4.52132136e-01 1.39879882e-01 -3.95989597e-01 -3.14213693e-01
-4.49615568e-01 -1.44212866e+00 1.49155366e+00 2.28182748e-01
4.77621146e-02 -4.85681802e-01 -3.30607295e-01 5.48304856e-01
-3.78897320e-03 -1.62257895e-01 7.91080534e-01 -4.71042573e-01
-1.10537910e+00 1.02048300e-01 -7.27214754e-01 1.97262913e-01
-2.55200505e-01 -4.32328492e-01 -9.69792008e-01 -4.29654300e-01
-1.63541690e-01 -1.72925994e-01 1.14819551e+00 3.66063505e-01
1.27955031e+00 -2.08956450e-01 -3.50033939e-01 6.46754444e-01
1.47142172e+00 5.52568734e-02 5.06829381e-01 3.48507941e-01
1.02203298e+00 2.45999977e-01 6.98154151e-01 3.18478495e-01
4.84151661e-01 1.07423258e+00 5.79038262e-01 -2.64133215e-01
-3.90991032e-01 -2.88181990e-01 2.74618298e-01 4.12632644e-01
3.26651663e-01 -7.00058043e-01 -1.30080187e+00 6.01604819e-01
-1.90238738e+00 -5.73366046e-01 -1.11031510e-01 1.98602498e+00
9.22398567e-01 4.06582385e-01 3.70751768e-02 -6.63507953e-02
9.35858309e-01 2.41109848e-01 -4.17025208e-01 9.36437324e-02
-1.74733296e-01 4.23044920e-01 6.10271096e-01 4.14460659e-01
-1.22372949e+00 1.15916896e+00 4.93775940e+00 1.22210634e+00
-9.93317008e-01 3.39025021e-01 8.69086385e-01 -2.53275990e-01
-2.94542700e-01 -1.71555683e-01 -1.43932450e+00 5.23927391e-01
5.34339786e-01 7.92979300e-02 1.70204893e-01 4.80573803e-01
2.29213759e-02 -1.14001855e-01 -1.19186032e+00 1.26396275e+00
-4.72790822e-02 -1.39665461e+00 2.40795657e-01 2.08691597e-01
6.43999279e-01 1.85263678e-01 4.69904214e-01 2.00690255e-01
-8.02888498e-02 -1.10841417e+00 1.02945530e+00 -5.15600783e-04
9.86321568e-01 -4.55679029e-01 6.54342115e-01 1.34103462e-01
-1.42304134e+00 6.33549467e-02 -2.06355795e-01 3.33281696e-01
1.13116823e-01 6.28077507e-01 -7.99543023e-01 4.44141150e-01
9.02621269e-01 5.17746687e-01 -6.18417501e-01 1.12722933e+00
-3.96591723e-01 6.36315584e-01 -3.76272649e-01 9.30261053e-03
4.77543622e-01 1.73033789e-01 7.22646952e-01 1.26688957e+00
2.64907271e-01 2.89796460e-02 2.93716192e-01 1.10595536e+00
-4.20995265e-01 5.12904488e-02 -3.02676916e-01 -6.40296238e-03
4.04588819e-01 1.04779148e+00 -1.00553083e+00 -1.11427665e-01
-6.04361534e-01 9.01216924e-01 1.37482420e-01 4.42453951e-01
-1.20042276e+00 -1.79737613e-01 4.30333376e-01 3.67511123e-01
7.42789447e-01 -2.54013240e-01 -3.75369668e-01 -1.22360563e+00
3.93417805e-01 -8.22735310e-01 2.72074133e-01 -7.06197262e-01
-1.08944225e+00 6.00349426e-01 2.16320232e-02 -1.29618084e+00
2.85708129e-01 -7.16442764e-01 -1.98794186e-01 9.53917205e-01
-1.59585273e+00 -1.37764668e+00 -2.80601740e-01 4.20561582e-01
8.33070099e-01 2.19514281e-01 4.30570841e-01 4.02843535e-01
-7.89690197e-01 9.49381113e-01 -2.37199515e-01 3.29281181e-01
6.02655113e-01 -1.21146512e+00 6.68066502e-01 1.03697550e+00
5.41312337e-01 4.33855385e-01 5.13842881e-01 -3.88571024e-01
-1.34535623e+00 -1.23870039e+00 7.26160824e-01 -4.97897863e-01
6.24444246e-01 -6.97976947e-01 -8.47272396e-01 6.43262684e-01
9.58740562e-02 1.85505614e-01 5.66276848e-01 3.53301950e-02
-4.14442837e-01 -6.56104237e-02 -7.09634304e-01 7.76541889e-01
1.26794302e+00 -5.09413600e-01 -3.90504628e-01 4.60073531e-01
1.10614240e+00 -6.68705106e-01 -5.45656621e-01 5.21381915e-01
4.49764937e-01 -7.72426367e-01 1.01059258e+00 -1.78712681e-01
5.35497129e-01 -4.21677500e-01 -4.13479984e-01 -6.47804797e-01
-2.92454392e-01 -5.48854411e-01 -2.06999928e-01 1.28505123e+00
8.33881974e-01 -4.43108886e-01 1.08278048e+00 2.14718640e-01
-1.27156273e-01 -1.16814196e+00 -9.67523217e-01 -7.42549300e-01
-3.75999399e-02 -5.67506850e-01 4.88509059e-01 5.98497748e-01
-3.77721608e-01 5.54205120e-01 -6.15898930e-02 3.95539403e-01
4.93987143e-01 2.68877268e-01 5.09378195e-01 -7.82532454e-01
-6.09729052e-01 -5.53609252e-01 -3.87015343e-01 -1.72941697e+00
5.62382266e-02 -9.28710461e-01 2.90060729e-01 -1.50745583e+00
4.52248305e-01 -5.06361842e-01 -4.64310735e-01 6.16152704e-01
-4.32889104e-01 7.01643586e-01 5.12920141e-01 3.74096066e-01
-9.50081289e-01 5.06866157e-01 1.37694454e+00 -4.54288781e-01
-2.39879917e-02 -7.63074160e-02 -8.99318635e-01 5.86202025e-01
7.66202867e-01 -7.08834648e-01 -3.85209471e-01 -8.51049900e-01
5.71433961e-01 -5.82669601e-02 5.67362964e-01 -8.66147995e-01
2.54296660e-01 6.58048317e-02 1.31601259e-01 -9.24290478e-01
3.05839151e-01 -6.36807621e-01 -2.73251444e-01 1.73496410e-01
-1.05051123e-01 2.01386601e-01 4.79100347e-01 6.45383656e-01
-2.89473355e-01 -1.85845941e-01 5.42150855e-01 -3.30710746e-02
-8.98463249e-01 5.22895098e-01 1.17093027e-01 2.82822162e-01
8.30269992e-01 -2.21416041e-01 -6.87514842e-01 -6.96294531e-02
-6.05665982e-01 5.22037327e-01 3.57657611e-01 3.52031678e-01
6.41493082e-01 -1.12207627e+00 -7.45106578e-01 -1.20835993e-02
3.53170753e-01 4.55528647e-01 1.70387447e-01 1.03570700e+00
-4.37505007e-01 5.66690743e-01 1.74245968e-01 -1.11498582e+00
-1.19776165e+00 3.49926412e-01 2.91480005e-01 -4.86874849e-01
-5.64477384e-01 1.34900677e+00 8.42944205e-01 3.86488922e-02
3.80148619e-01 -3.17514569e-01 7.59005025e-02 -9.98568349e-03
3.96242172e-01 -1.55547664e-01 1.94101602e-01 -5.32827139e-01
-3.25676888e-01 8.17990482e-01 -4.13756698e-01 -2.95220047e-01
8.58932555e-01 -1.03326358e-01 7.33276410e-03 1.91190749e-01
1.06589293e+00 -1.67819977e-01 -1.26958692e+00 -4.59500074e-01
-7.31164292e-02 -4.27457333e-01 1.00286514e-01 -8.03956687e-01
-1.16341996e+00 9.79086399e-01 6.36665046e-01 6.55440688e-02
1.18348014e+00 3.38342160e-01 9.04029191e-01 2.73807406e-01
2.27054894e-01 -7.34052837e-01 4.47044849e-01 5.92092514e-01
9.17189181e-01 -1.26619577e+00 -5.07008620e-02 -6.27948344e-01
-3.68661493e-01 5.18892169e-01 7.69243062e-01 1.25592694e-01
4.02126729e-01 1.89271316e-01 -2.31469259e-01 -3.43635768e-01
-6.54351473e-01 -4.80897337e-01 6.19013667e-01 2.52879679e-01
3.22422087e-01 -7.00945454e-03 -1.96102023e-01 7.30492711e-01
-1.96697220e-01 -2.53728300e-01 5.31480610e-02 8.58053088e-01
-4.70982790e-01 -1.05412662e+00 -2.36650959e-01 5.09653032e-01
-6.44147515e-01 -5.74593663e-01 -2.24620923e-01 8.31206083e-01
3.47942978e-01 8.50997269e-01 1.72510713e-01 -5.62039554e-01
2.92048335e-01 -1.32521838e-01 4.60050613e-01 -9.96282220e-01
-4.82315958e-01 2.84900844e-01 -4.73420694e-02 -5.81000745e-01
-2.00992480e-01 -3.42030913e-01 -1.24548292e+00 -1.61780685e-01
-6.59236431e-01 -1.49508953e-01 4.62509096e-01 7.55064368e-01
5.03694117e-01 6.08806491e-01 2.41662934e-01 -7.29386270e-01
-2.72055447e-01 -7.82301366e-01 -4.48986769e-01 1.08544976e-01
2.09625468e-01 -6.54228508e-01 -8.47353786e-02 5.04135899e-02] | [10.032334327697754, 0.9985991716384888] |
2297052c-c30b-41b6-87ad-3489875f6a01 | generalized-bilinear-deep-convolutional | 1807.01298 | null | http://arxiv.org/abs/1807.01298v1 | http://arxiv.org/pdf/1807.01298v1.pdf | Generalized Bilinear Deep Convolutional Neural Networks for Multimodal Biometric Identification | In this paper, we propose to employ a bank of modality-dedicated
Convolutional Neural Networks (CNNs), fuse, train, and optimize them together
for person classification tasks. A modality-dedicated CNN is used for each
modality to extract modality-specific features. We demonstrate that, rather
than spatial fusion at the convolutional layers, the fusion can be performed on
the outputs of the fully-connected layers of the modality-specific CNNs without
any loss of performance and with significant reduction in the number of
parameters. We show that, using multiple CNNs with multimodal fusion at the
feature-level, we significantly outperform systems that use unimodal
representation. We study weighted feature, bilinear, and compact bilinear
feature-level fusion algorithms for multimodal biometric person identification.
Finally, We propose generalized compact bilinear fusion algorithm to deploy
both the weighted feature fusion and compact bilinear schemes. We provide the
results for the proposed algorithms on three challenging databases: CMU
Multi-PIE, BioCop, and BIOMDATA. | ['Sobhan Soleymani', 'Nasser M. Nasrabadi', 'Jeremy Dawson', 'Amirsina Torfi'] | 2018-07-03 | null | null | null | null | ['person-identification'] | ['computer-vision'] | [ 2.19111234e-01 -4.99511570e-01 2.93732196e-01 -6.28766119e-01
-9.86852467e-01 -6.51929915e-01 6.37736857e-01 2.44600564e-01
-6.70811892e-01 7.66897619e-01 2.36088187e-01 6.11171052e-02
-8.80771428e-02 -5.67501068e-01 -6.58825934e-01 -7.26161897e-01
1.30997673e-01 1.08427256e-01 -4.64458436e-01 -1.11079589e-01
-1.07570127e-01 6.70352340e-01 -1.44991958e+00 5.76637089e-01
6.37744009e-01 1.43724322e+00 -5.28280139e-01 9.50983346e-01
4.26008821e-01 2.50999421e-01 -2.57615238e-01 -8.83351088e-01
2.56763697e-01 -2.70037800e-02 -5.82381487e-01 -9.24354196e-02
9.77179527e-01 -2.42070779e-01 -6.88980281e-01 9.13576245e-01
8.54703963e-01 2.47220665e-01 6.62556350e-01 -1.43436730e+00
-7.28211224e-01 4.29256022e-01 -5.27420759e-01 -3.32917497e-02
5.72210073e-01 2.59319216e-01 7.37028480e-01 -1.08837295e+00
3.23441327e-01 1.39036369e+00 1.15699840e+00 5.17015159e-01
-1.14442480e+00 -4.61990058e-01 -2.46987626e-01 1.26850292e-01
-1.70373023e+00 -5.78270316e-01 6.59239948e-01 -2.54274398e-01
9.32639241e-01 3.25752139e-01 2.95798123e-01 1.16385639e+00
7.13858157e-02 7.09195077e-01 1.04009545e+00 -2.75237888e-01
-2.16790900e-01 -9.42944735e-02 4.10444021e-01 9.89164233e-01
1.32234007e-01 1.00941569e-01 -8.03826332e-01 -4.49252129e-01
5.47454238e-01 4.14136469e-01 -3.42717409e-01 3.99925606e-03
-1.34760725e+00 3.87913227e-01 5.95369875e-01 2.02358499e-01
-5.35825729e-01 2.99638242e-01 3.00075650e-01 2.03517988e-01
1.08223312e-01 -1.32251224e-02 -5.69408059e-01 1.47420675e-01
-1.13409090e+00 5.12696087e-01 5.70329130e-01 9.14245844e-01
7.77132213e-01 -3.43696833e-01 -6.91760600e-01 9.48587894e-01
5.37819386e-01 7.09766209e-01 2.26112321e-01 -6.34314656e-01
5.93012214e-01 8.37275803e-01 -9.14446414e-02 -8.94306362e-01
-7.07025230e-01 -2.01203749e-01 -1.32244229e+00 -3.45413029e-01
5.18098533e-01 -3.30866724e-01 -1.33583629e+00 1.98927164e+00
-3.11618578e-02 3.24903697e-01 1.85789049e-01 8.55802894e-01
1.21576190e+00 1.66228458e-01 3.48216385e-01 3.26812834e-01
1.71071720e+00 -7.14916825e-01 -6.48595035e-01 3.72800082e-01
3.94813299e-01 -7.41201460e-01 4.89117771e-01 4.58185747e-02
-1.39189994e+00 -8.63630533e-01 -1.11245179e+00 -4.25543189e-01
-1.06759465e+00 3.93729627e-01 5.81643641e-01 9.80648994e-01
-1.35129523e+00 3.47143203e-01 -4.31796193e-01 -6.25062287e-01
5.82590342e-01 8.20182741e-01 -1.01873720e+00 -3.15988779e-01
-1.19216621e+00 7.87748694e-01 5.23007929e-01 4.96953815e-01
-5.74535429e-01 -6.30686104e-01 -1.11525106e+00 1.08148977e-01
-3.39089662e-01 -1.08419168e+00 6.17385268e-01 -5.62779963e-01
-9.84397531e-01 7.43182242e-01 -3.58411610e-01 -2.29942143e-01
2.19621181e-01 1.22663677e-02 -4.41632360e-01 1.30102709e-01
-3.05787981e-01 1.05110133e+00 5.65021038e-01 -1.07882130e+00
-7.30916023e-01 -6.00832880e-01 -1.24084670e-02 8.27273652e-02
-5.27053893e-01 7.16485083e-02 -5.78661978e-01 -6.12344980e-01
-6.86064875e-03 -8.86455476e-01 1.20576005e-02 5.26780039e-02
-6.70765102e-01 -3.02502096e-01 6.83825016e-01 -1.01812327e+00
7.25483775e-01 -2.19789243e+00 3.56735200e-01 5.72073221e-01
3.03012878e-01 9.54940543e-03 -3.06665182e-01 2.00195044e-01
2.54566167e-02 -1.77675970e-02 -5.09729683e-01 -9.51936841e-01
1.76342592e-01 1.00289859e-01 2.60622293e-01 6.18656039e-01
3.27092797e-01 1.32332182e+00 -2.59862602e-01 -4.02330190e-01
2.26104349e-01 8.58698964e-01 -5.22475123e-01 7.27314129e-02
5.23209691e-01 2.20779344e-01 9.68007073e-02 1.25303090e+00
1.24028718e+00 -6.40893504e-02 1.85220558e-02 -8.16151619e-01
1.55812711e-01 -4.50308979e-01 -1.17206156e+00 1.98119962e+00
-1.22100830e-01 4.34561878e-01 6.38859048e-02 -1.01674843e+00
5.13924599e-01 4.91071671e-01 5.80987334e-01 -4.48329419e-01
3.91429931e-01 2.08485350e-01 -2.72724777e-01 -2.86401331e-01
6.38947129e-01 1.00330651e-01 -4.46225286e-01 1.80940539e-01
9.09568429e-01 7.62854517e-01 3.82344015e-02 3.49404626e-02
6.81848228e-01 -9.59590822e-02 -5.03452681e-02 -3.41343768e-02
8.95138979e-01 -5.69155216e-01 1.83462977e-01 9.44298565e-01
-3.66465181e-01 8.34042490e-01 1.00237526e-01 -4.50681984e-01
-8.57838035e-01 -1.17976224e+00 -1.49821252e-01 1.21254933e+00
-6.86475728e-03 -5.69256693e-02 -7.02444613e-01 -5.57073534e-01
3.83592635e-01 -2.41213396e-01 -8.57558250e-01 -1.11580223e-01
-4.77791220e-01 -1.00944817e+00 1.37363982e+00 7.43625939e-01
9.46681142e-01 -6.99986935e-01 -7.24042058e-02 -1.65493190e-01
-2.95831233e-01 -1.26175201e+00 -6.60574198e-01 6.42589554e-02
-4.62699056e-01 -1.18196869e+00 -1.23554957e+00 -7.80625939e-01
5.94286919e-01 -5.29729761e-02 7.01326847e-01 3.00309826e-02
-1.99479997e-01 8.50515783e-01 -1.00666486e-01 -2.23088875e-01
4.10011023e-01 2.75901765e-01 1.39569968e-01 6.67854846e-01
5.05018950e-01 -2.58607328e-01 -6.79299235e-01 -1.22004844e-01
-9.58877563e-01 -1.95275337e-01 5.06039441e-01 1.17735386e+00
2.01836437e-01 -4.46123928e-01 4.82234210e-01 -8.08258429e-02
7.32002258e-01 -2.76238918e-01 1.11724950e-01 7.38171041e-01
-1.89664755e-02 -5.54795042e-02 1.24007519e-02 -3.09572816e-01
-9.73247647e-01 3.81450295e-01 -1.22000687e-01 -2.98756868e-01
-4.95102167e-01 6.30425334e-01 -1.81052625e-01 -5.30767977e-01
4.65198785e-01 3.57712328e-01 -3.67195457e-02 -4.79289740e-01
6.84008181e-01 7.62780368e-01 1.05559504e+00 -7.68446505e-01
5.92126608e-01 4.41607863e-01 1.83976099e-01 -5.62018812e-01
-2.14463085e-01 -4.03972745e-01 -8.21865022e-01 -2.22136006e-01
1.06211114e+00 -1.12078416e+00 -1.19727862e+00 1.02975941e+00
-1.18390167e+00 4.25835043e-01 2.25886777e-01 3.21277350e-01
-2.94406444e-01 3.78096908e-01 -6.49027288e-01 -7.11897314e-01
-6.12056315e-01 -1.18551040e+00 1.24384642e+00 7.91759908e-01
2.09104136e-01 -1.11018777e+00 -2.86779910e-01 4.52229112e-01
5.70346534e-01 4.69452292e-01 5.62478721e-01 -7.92998791e-01
-3.02524835e-01 -5.42899072e-01 -6.77651703e-01 1.26855433e-01
2.29168847e-01 -3.01015377e-01 -1.46844184e+00 -5.31529605e-01
-7.43324041e-01 -4.26320702e-01 1.25275803e+00 5.35781503e-01
1.03635347e+00 -2.74986159e-02 -1.54995173e-01 9.62766528e-01
1.28560555e+00 -1.83324724e-01 5.29643655e-01 -1.44606680e-01
9.90079165e-01 5.07376552e-01 -1.60326406e-01 3.88911396e-01
7.18822122e-01 6.38044000e-01 2.44495645e-02 -3.44234169e-01
-1.70214549e-01 1.25134841e-01 5.30752689e-02 2.96209633e-01
-5.36027849e-01 -1.85196906e-01 -7.56497145e-01 5.75806141e-01
-2.06254029e+00 -7.72775590e-01 1.47996664e-01 1.94152248e+00
6.54733896e-01 -6.84459805e-01 3.90913099e-01 -8.33752099e-03
7.40843952e-01 -2.72488654e-01 -2.40067437e-01 -1.50931060e-01
-7.26462185e-01 4.29835767e-01 4.94991332e-01 4.51205581e-01
-1.48335338e+00 6.70577705e-01 6.50034142e+00 4.84679371e-01
-1.08320177e+00 1.88963577e-01 5.83299756e-01 -9.70348194e-02
1.19529262e-01 -5.89042962e-01 -6.58733189e-01 3.85606408e-01
8.57037306e-01 3.17258745e-01 4.86571878e-01 2.30320051e-01
-3.46579105e-01 9.48346332e-02 -1.27632344e+00 1.46894217e+00
3.41529369e-01 -1.25383186e+00 1.99187383e-01 6.80315914e-03
5.14305651e-01 -2.33792558e-01 5.33635616e-01 1.19164906e-01
-6.79854974e-02 -1.40216994e+00 4.51232731e-01 1.20267999e+00
6.74496114e-01 -9.05572891e-01 1.20636868e+00 -2.24672735e-01
-1.43187547e+00 -1.84476301e-01 -1.66161910e-01 4.46731180e-01
3.65546457e-02 1.13991283e-01 -3.41045558e-01 1.07985032e+00
7.43316412e-01 5.05146921e-01 -1.06107736e+00 1.09545302e+00
6.69097900e-01 -6.19496144e-02 -3.63463432e-01 3.09404761e-01
-2.85264244e-03 2.40070954e-01 1.83266550e-01 1.65227187e+00
5.30330777e-01 -7.48246834e-02 1.70976579e-01 5.88268697e-01
-3.01699013e-01 -5.93972392e-02 -3.43993515e-01 9.05799121e-02
1.88241214e-01 1.37483633e+00 -1.29406258e-01 -3.92858654e-01
-5.83359778e-01 1.08154869e+00 1.82855889e-01 5.90185583e-01
-7.05501258e-01 -3.41961652e-01 6.42618954e-01 -6.30638838e-01
1.86539963e-01 4.56334613e-02 -5.55347145e-01 -1.30909681e+00
-6.25031143e-02 -7.32113600e-01 7.90034413e-01 -5.14878273e-01
-1.75661445e+00 4.49135840e-01 6.17105216e-02 -8.53584826e-01
-7.47590661e-02 -1.04845369e+00 -2.31196582e-01 1.48107076e+00
-1.37353730e+00 -2.12743855e+00 -4.30830956e-01 1.31728220e+00
-1.52467921e-01 -6.77193820e-01 8.28905046e-01 7.56492853e-01
-6.32573605e-01 1.43047535e+00 -3.54900897e-01 4.87768620e-01
9.18375134e-01 -1.09544098e+00 -5.90552427e-02 6.69578075e-01
-2.54207224e-01 1.03567517e+00 1.72235399e-01 -3.94134343e-01
-1.61422932e+00 -1.06191611e+00 9.65579808e-01 -3.56980771e-01
1.10430650e-01 -2.16772258e-01 -5.34605980e-01 5.65799415e-01
5.77555418e-01 2.35055804e-01 1.10803103e+00 2.38182649e-01
-6.48241282e-01 -1.58608884e-01 -1.73103559e+00 3.48021269e-01
7.99963415e-01 -8.09564471e-01 -3.09747517e-01 5.60553707e-02
2.61355788e-01 -3.40658665e-01 -1.32655072e+00 5.87334454e-01
1.14112473e+00 -5.77424049e-01 1.35866785e+00 -8.73764515e-01
8.89791846e-02 -4.17342514e-01 -7.55545735e-01 -1.09346116e+00
-4.18932527e-01 -1.23142473e-01 -3.30938697e-01 1.18663859e+00
5.00536442e-01 -6.45172179e-01 3.91939670e-01 8.12889814e-01
3.34064942e-03 -5.00052571e-01 -1.21204209e+00 -2.56682605e-01
2.64444984e-02 -2.37457529e-01 8.02289963e-01 8.54845285e-01
-1.71785504e-01 -1.49469435e-01 -6.53425336e-01 4.38629687e-01
9.32122767e-01 -3.85533869e-01 4.84982580e-01 -1.15168202e+00
-4.12120372e-02 -5.26759803e-01 -7.20689714e-01 -4.73378956e-01
-1.11337639e-01 -8.99464309e-01 -1.10572331e-01 -1.12537181e+00
5.75726986e-01 -5.05603943e-03 -1.12011492e+00 8.28626335e-01
-3.46883684e-01 9.14153457e-01 4.68480259e-01 -1.50253668e-01
-5.64594984e-01 2.18850687e-01 8.48075688e-01 -5.36244154e-01
-1.28783584e-01 -4.32490557e-01 -8.55369866e-01 3.26298177e-01
5.58900654e-01 2.95093834e-01 5.06961383e-02 -2.89421052e-01
-1.09252274e-01 -1.86338663e-01 8.78544450e-01 -1.19943488e+00
6.55323386e-01 2.35133350e-01 1.32196271e+00 -8.11078072e-01
8.14238429e-01 -7.03103721e-01 1.15508340e-01 2.14962661e-01
-4.45689201e-01 1.52966678e-01 3.86187553e-01 2.59205639e-01
-1.96173504e-01 4.14148271e-01 7.66786754e-01 1.66564658e-01
-4.16191727e-01 5.36631286e-01 2.77711893e-04 -8.29713166e-01
7.06434608e-01 -2.04618394e-01 -5.00916541e-01 -1.76675469e-01
-1.10869706e+00 3.09313267e-01 -1.23226941e-02 4.31819111e-01
8.93059313e-01 -1.97446430e+00 -7.73819208e-01 3.54822069e-01
3.31837118e-01 -5.79989612e-01 7.60400891e-01 1.03208351e+00
-1.87463939e-01 6.65552914e-01 -6.88975573e-01 -6.26188040e-01
-1.41968334e+00 3.88565451e-01 6.95717037e-01 -8.23030025e-02
1.62004620e-01 8.69876206e-01 -1.02822535e-01 -9.23379004e-01
2.66359240e-01 -1.41028687e-01 -3.57278079e-01 2.36311346e-01
5.91010273e-01 2.74190068e-01 6.24841154e-02 -1.07653916e+00
-6.49627686e-01 6.24938965e-01 -7.60194585e-02 -3.48419487e-01
1.14428353e+00 4.16728593e-02 -2.82509089e-01 -5.01573645e-02
1.32225025e+00 -5.72309375e-01 -8.75791013e-01 -4.21211332e-01
-3.57492268e-01 -1.51220649e-01 -2.70134993e-02 -1.06221402e+00
-1.25212324e+00 8.31157625e-01 1.31343400e+00 1.10078529e-02
1.40615046e+00 -1.97612852e-01 7.38890111e-01 4.26236868e-01
3.75829265e-02 -9.52233195e-01 -4.15390372e-01 3.82624924e-01
7.95217395e-01 -1.26421762e+00 -2.39929050e-01 -1.98972419e-01
-3.20099294e-01 1.07179451e+00 5.25239348e-01 8.01123008e-02
9.40658867e-01 4.89392951e-02 -1.36083692e-01 -7.56411403e-02
-3.39264005e-01 -4.93709326e-01 9.95702863e-01 8.63408744e-01
4.75363106e-01 2.72085220e-01 -3.66312265e-02 7.74023890e-01
1.16460793e-01 1.06253095e-01 -1.95954576e-01 8.99134636e-01
-1.69320516e-02 -1.05343735e+00 -7.27861941e-01 5.55297971e-01
-3.58475953e-01 -1.17185108e-01 -4.40046132e-01 4.73110557e-01
7.03772366e-01 9.50976372e-01 3.59327272e-02 -7.53352761e-01
2.54439414e-01 4.75588977e-01 7.96696723e-01 1.34905010e-01
-9.15840328e-01 -1.26233399e-01 -7.50080645e-02 -3.24719965e-01
-7.41986990e-01 -7.14343905e-01 -5.80712318e-01 -6.72216117e-01
8.21551606e-02 -4.40225244e-01 5.36630213e-01 1.21599901e+00
5.40748894e-01 3.68050575e-01 2.90039361e-01 -9.72581625e-01
-1.33082867e-01 -1.30655754e+00 -3.23028684e-01 5.89033723e-01
6.21075153e-01 -7.81573892e-01 2.87211418e-01 1.63443014e-01] | [14.612428665161133, 0.9948211908340454] |
c44a7729-0c26-48e4-877a-aa721843e11c | inor-net-incremental-3d-object-recognition | 2302.09886 | null | https://arxiv.org/abs/2302.09886v1 | https://arxiv.org/pdf/2302.09886v1.pdf | InOR-Net: Incremental 3D Object Recognition Network for Point Cloud Representation | 3D object recognition has successfully become an appealing research topic in the real-world. However, most existing recognition models unreasonably assume that the categories of 3D objects cannot change over time in the real-world. This unrealistic assumption may result in significant performance degradation for them to learn new classes of 3D objects consecutively, due to the catastrophic forgetting on old learned classes. Moreover, they cannot explore which 3D geometric characteristics are essential to alleviate the catastrophic forgetting on old classes of 3D objects. To tackle the above challenges, we develop a novel Incremental 3D Object Recognition Network (i.e., InOR-Net), which could recognize new classes of 3D objects continuously via overcoming the catastrophic forgetting on old classes. Specifically, a category-guided geometric reasoning is proposed to reason local geometric structures with distinctive 3D characteristics of each class by leveraging intrinsic category information. We then propose a novel critic-induced geometric attention mechanism to distinguish which 3D geometric characteristics within each class are beneficial to overcome the catastrophic forgetting on old classes of 3D objects, while preventing the negative influence of useless 3D characteristics. In addition, a dual adaptive fairness compensations strategy is designed to overcome the forgetting brought by class imbalance, by compensating biased weights and predictions of the classifier. Comparison experiments verify the state-of-the-art performance of the proposed InOR-Net model on several public point cloud datasets. | ['Ender Konukoglu', 'Jun Li', 'Lingjuan Lyu', 'Lixu Wang', 'Gan Sun', 'Yang Cong', 'Jiahua Dong'] | 2023-02-20 | null | null | null | null | ['3d-object-recognition', 'object-recognition'] | ['computer-vision', 'computer-vision'] | [ 3.61616388e-02 3.79892215e-02 5.86940348e-02 -4.90945518e-01
-1.57449022e-01 -3.44103813e-01 3.50731224e-01 4.55565900e-02
-2.32958108e-01 6.74182296e-01 -2.09351122e-01 -6.07464202e-02
-2.35561252e-01 -8.62689912e-01 -7.46222377e-01 -7.70361483e-01
8.54513496e-02 5.35645008e-01 2.42354408e-01 3.09554227e-02
4.49273437e-01 1.06814039e+00 -2.12789297e+00 2.29480360e-02
1.16388679e+00 1.36145842e+00 1.30317375e-01 1.14571899e-01
-5.09128034e-01 4.40667212e-01 -6.74076259e-01 -1.76778778e-01
4.61168587e-01 6.44087419e-02 -4.31984603e-01 2.11586028e-01
6.60513103e-01 -3.65417510e-01 -4.11621779e-01 1.03680706e+00
4.84160334e-01 2.94404775e-01 6.22553170e-01 -1.42332518e+00
-9.99119520e-01 3.14177006e-01 -5.51322520e-01 3.10716152e-01
-8.98024812e-02 4.31453139e-01 6.08677506e-01 -1.09245551e+00
2.97917634e-01 1.43307805e+00 5.88038325e-01 6.52767539e-01
-9.72480893e-01 -8.75710905e-01 7.72756100e-01 3.73057097e-01
-1.58349299e+00 -1.95207208e-01 1.19360757e+00 -3.78334105e-01
9.00557697e-01 2.01543316e-01 7.69850135e-01 8.76346350e-01
2.95455068e-01 8.57507050e-01 8.87387931e-01 3.06322575e-02
3.41622472e-01 -6.33312482e-03 2.68069983e-01 4.49230611e-01
5.17808795e-01 2.07359254e-01 -4.67728555e-01 1.47706419e-01
5.47960639e-01 5.91094017e-01 -8.88174027e-02 -6.49444699e-01
-1.06563520e+00 3.91443878e-01 8.02652359e-01 4.82473820e-02
-3.28706264e-01 -6.58149719e-02 1.87745839e-01 3.93736839e-01
6.51621819e-01 1.36674449e-01 -6.55837655e-01 2.43278742e-01
-4.97160822e-01 1.30379111e-01 1.97810203e-01 1.11761332e+00
7.54532635e-01 2.95109779e-01 -2.93611258e-01 6.69964433e-01
2.36541584e-01 8.92648399e-01 4.91541922e-01 -5.87461710e-01
2.58976161e-01 1.18932974e+00 -7.64478669e-02 -1.05476010e+00
-4.40375268e-01 -9.59520698e-01 -1.22045696e+00 4.44487214e-01
1.58905700e-01 4.81739491e-01 -1.21952248e+00 1.77627325e+00
4.89746630e-01 9.05760005e-02 -4.69445856e-03 1.05851424e+00
7.86508441e-01 3.95983547e-01 1.08245231e-01 -2.31603771e-01
9.33117509e-01 -6.40796542e-01 -7.88388848e-02 -1.39861837e-01
2.16814458e-01 -4.16953653e-01 1.00237155e+00 2.16286451e-01
-7.45210409e-01 -8.70178878e-01 -9.92673039e-01 2.59011745e-01
-5.02407432e-01 -1.15119554e-01 5.84884405e-01 5.56911409e-01
-4.65308219e-01 5.90382457e-01 -5.18285036e-01 -1.11742996e-01
1.04492664e+00 4.79724795e-01 -1.54836878e-01 -2.39169985e-01
-9.71366525e-01 8.27442884e-01 5.72347879e-01 1.79531872e-01
-1.07222271e+00 -8.94711018e-01 -4.09845054e-01 1.58319920e-01
5.10974884e-01 -6.57158613e-01 9.85073864e-01 -1.02596414e+00
-1.05432868e+00 8.85635197e-01 2.14419708e-01 -3.09599489e-01
6.59111559e-01 -1.49814114e-01 -5.20764112e-01 -3.03963900e-01
-6.52288646e-03 6.98210478e-01 1.19239271e+00 -1.35673058e+00
-7.68190205e-01 -7.01243401e-01 6.46969825e-02 4.60075080e-01
-2.42286652e-01 -9.09719050e-01 -1.93033833e-02 -7.98416972e-01
6.41535580e-01 -8.38395298e-01 -5.80937639e-02 4.46095318e-01
-1.60126463e-01 -4.93403792e-01 9.03508127e-01 -7.89601803e-02
6.38407111e-01 -2.13838053e+00 -3.86470445e-02 -3.29725184e-02
3.52095127e-01 2.47361168e-01 -2.66049743e-01 -4.26262707e-01
-6.79084808e-02 8.90248939e-02 -1.96221054e-01 1.17905006e-01
-1.92635164e-01 2.84364730e-01 -6.67148471e-01 4.02704507e-01
4.17954624e-01 7.48207510e-01 -9.56502914e-01 -6.87207878e-02
3.11085582e-01 4.49290693e-01 -3.64698678e-01 1.81188390e-01
-3.37018460e-01 3.57723862e-01 -5.79723954e-01 9.64665174e-01
1.17838728e+00 -1.15653887e-01 -3.39579791e-01 -1.73253462e-01
3.10799256e-02 -2.16237560e-01 -1.03453028e+00 1.55448699e+00
-2.07066596e-01 -2.97801849e-02 -4.11850780e-01 -1.07273412e+00
1.37249458e+00 -2.32061327e-01 3.48755807e-01 -7.26321816e-01
1.67124480e-01 2.54237324e-01 -6.17712289e-02 -1.02677897e-01
3.24410975e-01 -4.01804805e-01 -1.35941952e-01 3.46167088e-01
-6.78831860e-02 -2.37697214e-01 -4.91328716e-01 -8.07272270e-02
9.79973972e-01 3.50185903e-04 3.92522402e-02 -2.05075026e-01
7.30839789e-01 -5.47791831e-02 8.56343627e-01 9.71000850e-01
-5.78831732e-01 5.40530860e-01 -9.81415957e-02 -1.10069096e+00
-9.47318375e-01 -1.40852714e+00 -1.02960942e-02 6.59985244e-01
6.82886422e-01 5.77484250e-01 -1.05085231e-01 -1.08883286e+00
5.25961637e-01 8.10889065e-01 -7.02858388e-01 -9.51857924e-01
-2.89591849e-01 -7.06423044e-01 2.15307832e-01 4.32596266e-01
6.38281047e-01 -1.07740724e+00 -9.33257341e-01 1.66986823e-01
1.08511165e-01 -6.37085795e-01 -3.58407110e-01 2.60151714e-01
-1.27381480e+00 -1.10863078e+00 -7.68448830e-01 -6.13011956e-01
9.46732223e-01 7.95100689e-01 9.76082623e-01 3.40708405e-01
-3.79239053e-01 3.47546965e-01 -4.46344227e-01 -5.31573534e-01
-1.09095179e-01 2.56843209e-01 4.84341323e-01 2.05959961e-01
4.79790509e-01 -7.31744409e-01 -6.59640074e-01 4.61199999e-01
-6.92866623e-01 1.93957418e-01 8.56051683e-01 8.35986972e-01
8.28167498e-01 3.53136599e-01 8.68147373e-01 -6.07130647e-01
7.57388324e-02 -5.40557027e-01 -5.32793581e-01 4.24883544e-01
-8.05360377e-01 1.39356285e-01 7.45850980e-01 -8.29284966e-01
-1.10249591e+00 -3.04630212e-02 1.94756597e-01 -9.70046759e-01
-1.78441495e-01 7.93807805e-02 -3.95106047e-01 -1.71110600e-01
3.53635848e-01 4.36334461e-01 -7.51145557e-02 -5.94866753e-01
2.35161856e-01 3.99294376e-01 5.46642900e-01 -6.55889869e-01
1.10787404e+00 4.90955353e-01 -4.46961168e-03 -3.88005316e-01
-1.16422653e+00 -1.59783900e-01 -6.94593549e-01 -4.78821069e-01
4.51969117e-01 -1.02302182e+00 -7.10852683e-01 8.83686066e-01
-1.04936588e+00 4.54618856e-02 -6.56947792e-01 2.40685552e-01
-3.86999965e-01 2.79278666e-01 -9.95739922e-02 -9.67315257e-01
-4.52685177e-01 -7.43277013e-01 7.67705023e-01 5.37450612e-01
3.21253091e-01 -2.66700923e-01 -4.10866916e-01 1.66136980e-01
5.16863346e-01 2.45127127e-01 1.09859025e+00 -5.14903545e-01
-1.00783241e+00 -2.79267192e-01 -5.22469580e-01 3.11176658e-01
2.71731377e-01 -2.86012799e-01 -8.71574700e-01 -4.51947808e-01
8.69994387e-02 -2.51459628e-01 9.36515152e-01 -1.97283458e-02
1.33516395e+00 -2.31476560e-01 -2.89292902e-01 5.32336175e-01
1.26507139e+00 4.21274900e-01 4.34160471e-01 4.36545163e-02
7.31450319e-01 3.11439753e-01 7.28216112e-01 7.18359232e-01
2.66595900e-01 3.39592963e-01 7.95476139e-01 3.42110276e-01
-2.25445092e-01 -3.77001524e-01 -3.40424888e-02 8.96760702e-01
2.25213114e-02 -1.85776964e-01 -8.61227632e-01 4.90967005e-01
-1.65563965e+00 -7.29788959e-01 2.75905758e-01 2.38571930e+00
5.10749280e-01 5.78578591e-01 -4.66197968e-01 2.91454997e-02
9.57378268e-01 8.74838606e-03 -1.48132145e+00 1.07321680e-01
-4.95951384e-01 -1.57962739e-01 3.35536748e-01 -6.44713342e-02
-9.42619085e-01 7.12981641e-01 4.86513472e+00 8.30965936e-01
-1.13336599e+00 -1.15757786e-01 8.24497104e-01 -1.09526366e-01
-3.15686941e-01 -7.00982809e-02 -7.50947833e-01 4.66323167e-01
2.67696530e-01 -2.55720288e-01 4.08074856e-01 1.07515824e+00
-2.48924837e-01 1.02027379e-01 -9.76124465e-01 1.22271860e+00
2.09828660e-01 -1.21572340e+00 6.51855707e-01 -2.24509537e-01
6.21746182e-01 -1.12806417e-01 2.17561454e-01 4.99715537e-01
1.08479574e-01 -5.92439950e-01 1.05810714e+00 9.83840168e-01
9.14000809e-01 -7.68029094e-01 6.63707852e-01 3.35967958e-01
-1.05026162e+00 -3.85912240e-01 -6.00224733e-01 -4.94859628e-02
-1.77812055e-01 8.44675303e-01 -3.94563079e-01 5.70900023e-01
1.06054378e+00 7.02634454e-01 -8.17617953e-01 1.18118274e+00
-4.67571393e-02 1.90131858e-01 -3.55622530e-01 6.14994355e-02
1.14151519e-02 2.47125417e-01 8.15094650e-01 3.40382695e-01
5.24518549e-01 4.44829553e-01 6.09587431e-02 9.78223443e-01
-2.51601160e-01 -2.03915238e-01 -4.91653800e-01 2.24983662e-01
5.72208464e-01 8.68276477e-01 -9.05069649e-01 -3.92062843e-01
-9.19609442e-02 8.26832056e-01 4.19197023e-01 2.07855310e-02
-6.46659553e-01 -2.02499017e-01 5.87101758e-01 -4.76832762e-02
4.81837332e-01 -6.55051544e-02 -5.08417904e-01 -1.36810529e+00
1.50275812e-01 -5.52919149e-01 5.98872960e-01 -9.49147940e-01
-1.77505624e+00 6.53762877e-01 -2.95302302e-01 -1.53357148e+00
7.37001479e-01 -4.60032523e-01 -4.99522626e-01 5.78847051e-01
-1.63086760e+00 -8.99408698e-01 -6.77451253e-01 5.44364035e-01
6.47790015e-01 -3.72196883e-01 4.82944041e-01 2.91742712e-01
-3.46656471e-01 7.05360353e-01 -6.08305074e-02 -3.68149847e-01
5.89131653e-01 -8.37278724e-01 3.29536855e-01 5.39787769e-01
-1.39235824e-01 2.00021699e-01 3.53135377e-01 -9.17548835e-01
-1.37808931e+00 -1.34186172e+00 6.02782905e-01 -4.31102008e-01
9.61448476e-02 -3.15206885e-01 -1.28257072e+00 1.81564391e-01
-6.64680600e-01 1.47640929e-01 3.36714029e-01 -8.85996968e-02
-7.74194062e-01 -5.70599020e-01 -1.37512374e+00 3.13833088e-01
1.55225194e+00 -1.46170586e-01 -8.52243304e-01 4.13341783e-02
1.03646636e+00 -2.47772396e-01 -5.72244644e-01 9.34100270e-01
5.24607182e-01 -7.75618851e-01 1.03973889e+00 -5.72460711e-01
1.63033530e-01 -6.36944234e-01 -1.82429209e-01 -1.08964360e+00
-4.88787979e-01 -6.61297962e-02 -3.22415054e-01 1.02645338e+00
3.50443833e-02 -6.93604290e-01 9.71916616e-01 6.39318347e-01
-4.28714931e-01 -5.77813625e-01 -1.44289565e+00 -1.08361125e+00
9.63136330e-02 -3.90298903e-01 9.82621968e-01 9.05338645e-01
-7.51848102e-01 2.26626918e-01 -2.34264523e-01 2.43398130e-01
8.60298753e-01 5.60272276e-01 5.34096181e-01 -1.75765276e+00
1.62679642e-01 -6.13507271e-01 -7.31707931e-01 -9.94381428e-01
2.71426290e-02 -1.09231126e+00 9.61297601e-02 -1.00753546e+00
2.66562551e-01 -1.15154505e+00 -8.71939301e-01 7.31263757e-01
-1.49272412e-01 1.45466760e-01 3.67820829e-01 5.62568247e-01
-8.01763892e-01 1.19093812e+00 1.40185237e+00 -7.04993665e-01
-1.86676934e-01 -1.23360576e-02 -7.66063690e-01 5.45388460e-01
6.21047497e-01 -5.52782536e-01 -3.81206095e-01 -5.36317527e-01
1.10550821e-01 -2.01242477e-01 4.90190655e-01 -1.34979391e+00
2.05250338e-01 -3.12768757e-01 7.46409774e-01 -1.19190001e+00
9.48363915e-02 -1.10141337e+00 1.59947842e-01 6.17472529e-01
-1.03894234e-01 -3.45792800e-01 3.85640204e-01 9.02402282e-01
1.58315882e-01 7.64561370e-02 8.86795938e-01 -1.44554928e-01
-8.68662298e-01 8.01709235e-01 -5.87361529e-02 -1.40321348e-02
1.20210600e+00 -4.46307510e-01 -2.62133360e-01 1.25560433e-01
-7.12746501e-01 3.38310212e-01 5.42445242e-01 8.97042930e-01
9.79194880e-01 -1.47259319e+00 -5.83032250e-01 5.20912588e-01
3.28728914e-01 3.16095501e-01 9.35173094e-01 2.03977197e-01
-1.92337483e-01 6.68985993e-02 -5.94974399e-01 -6.49755836e-01
-8.19756567e-01 1.13541996e+00 3.80401492e-01 5.74248135e-02
-6.43713236e-01 8.59505057e-01 3.83602500e-01 -5.36266983e-01
3.17367911e-01 -2.07769245e-01 4.01895940e-02 1.49221523e-02
3.39347363e-01 1.90309003e-01 7.68353194e-02 -5.10375559e-01
-5.82094848e-01 6.81565583e-01 -3.22871566e-01 7.73671031e-01
1.46999884e+00 -3.79199266e-01 2.08077565e-01 6.36268675e-01
7.94901013e-01 -6.42957807e-01 -1.65962744e+00 -3.69221598e-01
-2.42871001e-01 -7.00734258e-01 -1.48639783e-01 -1.07873344e+00
-1.27203417e+00 9.23037589e-01 1.04542530e+00 -1.71232328e-01
1.16347480e+00 -1.14892401e-01 7.78431475e-01 6.02587223e-01
7.48611450e-01 -1.03641880e+00 1.34118944e-01 5.97635269e-01
9.33143437e-01 -1.25500119e+00 4.47548740e-02 -1.75563037e-01
-3.30932140e-01 1.14373386e+00 9.86799002e-01 -2.38403603e-01
7.81184852e-01 -5.35126805e-01 -1.67311862e-01 -1.30438730e-01
-6.52629733e-01 2.73055732e-02 3.14548701e-01 6.30624175e-01
-3.92000705e-01 5.93282841e-02 2.01712221e-01 8.25356722e-01
1.62787035e-01 -9.87795666e-02 3.22396934e-01 8.92002821e-01
-5.23112714e-01 -5.45873821e-01 -3.20866078e-01 6.91662252e-01
2.18387574e-01 1.17450744e-01 -2.04200089e-01 4.75141585e-01
5.98382592e-01 4.58437681e-01 2.61406839e-01 -5.50970733e-01
5.27084708e-01 1.19017221e-01 4.12118345e-01 -4.51385289e-01
-1.25790522e-01 -4.62831736e-01 -7.01819658e-01 -1.10514939e-01
-3.22822154e-01 -5.11448383e-01 -1.09917819e+00 -3.28595072e-01
-4.32965249e-01 -9.18202624e-02 1.80516705e-01 7.45868385e-01
6.80396736e-01 5.32589436e-01 9.44017470e-01 -8.79943550e-01
-6.00413680e-01 -6.20683789e-01 -5.39361000e-01 5.06323755e-01
2.94458747e-01 -1.27494633e+00 -5.59303224e-01 -2.41072670e-01] | [7.853485584259033, -3.2083256244659424] |
1f9a70ee-aa88-4544-a597-9fd09c529027 | content-based-image-retrieval-speedup | 1911.11379 | null | https://arxiv.org/abs/1911.11379v2 | https://arxiv.org/pdf/1911.11379v2.pdf | Content-based image retrieval speedup | Content-based image retrieval (CBIR) is a task of retrieving images from their contents. Since retrieval process is a time-consuming task in large image databases, acceleration methods can be very useful. This paper presents a novel method to speed up CBIR systems. In the proposed method, first Zernike moments are extracted from query image and an interval is calculated for that query. Images in database which are out of the interval are ignored in retrieval process. Therefore, a database reduction occurs before retrieval which leads to speed up. It is shown that in reduced database, relevant images to query image are preserved and irrelevant images are throwed away. Therefore, the proposed method speed up retrieval process and preserve CBIR accuracy, simultaneously. | ['Sadegh Fadaei', 'Elyas Rashno', 'Abdolreza Rashno'] | 2019-11-26 | null | null | null | null | ['content-based-image-retrieval'] | ['computer-vision'] | [ 3.69971931e-01 -9.06560719e-01 -1.55699342e-01 -1.62959948e-01
-8.11343610e-01 -5.75912058e-01 1.87690750e-01 4.13273603e-01
-7.60592699e-01 4.38206851e-01 -1.44131452e-01 5.89230582e-02
-5.77821791e-01 -9.56358016e-01 -7.08514825e-02 -6.94534123e-01
3.03267032e-01 2.97448277e-01 6.52235508e-01 -1.86573341e-01
7.73582041e-01 1.01950073e+00 -1.82144725e+00 1.50543347e-01
3.81040484e-01 1.08959591e+00 4.42512244e-01 6.90894663e-01
-3.51296902e-01 6.95091069e-01 -9.69563544e-01 3.88831273e-02
4.83733624e-01 -4.56160992e-01 -8.61334026e-01 1.70762017e-01
-9.25004762e-03 -6.20094121e-01 -3.22181553e-01 1.20296097e+00
5.91873288e-01 5.26618361e-01 8.10555041e-01 -1.18416762e+00
-6.70463860e-01 -2.13237718e-01 -7.75677025e-01 4.51451480e-01
2.77373910e-01 -7.11849630e-01 7.60614574e-01 -8.95660520e-01
6.93042874e-01 1.09499621e+00 7.19549730e-02 8.98249820e-02
-8.56207371e-01 -5.23777187e-01 -2.69651741e-01 5.99178910e-01
-1.93950319e+00 -1.76047787e-01 8.79532456e-01 -1.95105262e-02
5.64087629e-01 6.89608276e-01 6.30755365e-01 -4.06000644e-01
3.74098897e-01 6.95803761e-01 7.67468035e-01 -8.16538811e-01
1.25642985e-01 1.79655343e-01 3.60440940e-01 2.68499583e-01
2.89736301e-01 -1.95862040e-01 -1.68977916e-01 -6.94935545e-02
7.06565440e-01 6.41616821e-01 -1.18074134e-01 -1.94842875e-01
-9.29146528e-01 7.85482168e-01 4.37640637e-01 5.97514212e-01
-5.38859189e-01 -1.79462001e-01 4.88588184e-01 4.76801485e-01
1.13616347e-01 6.43856898e-02 5.44158071e-02 3.30947161e-01
-9.48379934e-01 3.72866929e-01 4.42762017e-01 9.79925811e-01
8.77039015e-01 -4.83194828e-01 -8.58558267e-02 1.07586515e+00
-1.55666023e-02 7.45557845e-01 5.81925273e-01 -6.72293723e-01
-3.00257891e-01 6.98094070e-01 1.11615390e-01 -1.52751386e+00
-4.09836955e-02 1.05701610e-01 -6.85388446e-01 5.40904105e-02
-3.92132364e-02 6.54941142e-01 -8.64217520e-01 7.73421228e-01
2.83915818e-01 -4.62453544e-01 3.08231562e-02 9.60239947e-01
8.44892740e-01 1.02219045e+00 -1.33331418e-01 -5.28729320e-01
1.57348919e+00 -7.21721470e-01 -9.01087403e-01 2.92803168e-01
6.64028451e-02 -1.45283461e+00 8.35753143e-01 4.26911592e-01
-9.37347293e-01 -6.15369022e-01 -9.56518650e-01 -2.23514408e-01
-4.98934835e-01 1.22945115e-01 3.98909479e-01 1.94353178e-01
-9.39760149e-01 1.09689303e-01 -2.10613400e-01 -2.38355115e-01
-9.29444060e-02 3.84804308e-01 -3.89951885e-01 -4.48238373e-01
-9.88231122e-01 6.97648287e-01 6.11337066e-01 9.86321270e-02
-3.77313405e-01 -1.19054109e-01 -4.97748286e-01 7.06108883e-02
3.27116668e-01 -1.04884140e-01 1.03293753e+00 -9.03285265e-01
-9.42303240e-01 8.64032924e-01 -3.85088593e-01 -7.26482570e-02
1.60809368e-01 7.59747773e-02 -3.78931612e-01 7.04597652e-01
2.71940529e-01 5.13367176e-01 9.64259982e-01 -1.02729940e+00
-7.09734917e-01 -4.76233244e-01 -2.00998381e-01 3.99555981e-01
-2.30968371e-01 2.78494239e-01 -9.79039073e-01 -3.64080369e-01
5.70061326e-01 -8.39278400e-01 -5.32703139e-02 -1.43086361e-02
-1.98753756e-02 -3.43405426e-01 1.10541964e+00 -5.94294310e-01
1.32060504e+00 -2.28712106e+00 -4.75292683e-01 5.45734227e-01
-4.06137705e-02 3.07180613e-01 2.02032439e-02 6.58339560e-01
1.53124630e-01 1.41487882e-01 3.37892324e-01 4.49269623e-01
-5.58190644e-01 6.26566187e-02 -5.76636083e-02 2.94331074e-01
-1.99114844e-01 3.92346352e-01 -5.73536098e-01 -1.08599675e+00
4.65933412e-01 5.66361547e-01 -2.21494928e-01 -8.27307478e-02
3.91405970e-01 1.40054464e-01 -5.83592296e-01 8.52435052e-01
1.02275658e+00 -2.10864507e-02 -2.87031144e-01 -4.20896798e-01
-1.82396084e-01 -3.29511553e-01 -1.18976140e+00 1.10635614e+00
-4.08121556e-01 6.50613189e-01 -2.40579367e-01 -9.98833716e-01
1.20479608e+00 3.31732571e-01 7.31442809e-01 -1.10852706e+00
2.38502502e-01 1.79693773e-01 -1.31595358e-01 -4.02915984e-01
7.99069762e-01 -2.96822041e-02 3.41307729e-01 5.01438737e-01
-4.97025400e-01 -4.03498858e-01 5.31039298e-01 1.17737524e-01
5.52850246e-01 -4.87447798e-01 5.62862277e-01 -2.87888229e-01
9.89816606e-01 3.80284607e-01 2.62419224e-01 7.12364733e-01
-3.33328545e-01 3.99956107e-01 -1.95902154e-01 -6.41996145e-01
-1.16311765e+00 -9.05863762e-01 -2.59614944e-01 8.43320072e-01
7.71494567e-01 -1.75087258e-01 -4.21103716e-01 -7.86790997e-02
-1.49029866e-01 -4.08961177e-02 -1.91077277e-01 -1.05322182e-01
-4.09376621e-01 -3.33409280e-01 2.57423590e-03 1.29097506e-01
9.59794641e-01 -1.07316077e+00 -7.01593935e-01 1.73063040e-01
-2.53973484e-01 -6.27188504e-01 -5.45355022e-01 -6.38094187e-01
-1.12588906e+00 -1.10282946e+00 -1.08026564e+00 -1.13427997e+00
9.88865077e-01 1.19177687e+00 7.43753850e-01 6.55725956e-01
-7.13785946e-01 2.85653502e-01 -5.69212854e-01 -5.85605145e-01
-1.40765861e-01 -2.91199148e-01 -3.52441519e-01 -3.67344826e-01
6.67801321e-01 -2.54144091e-02 -9.81857300e-01 4.61705089e-01
-1.30529046e+00 -3.68648261e-01 5.91424584e-01 8.80380154e-01
1.09875786e+00 8.86324823e-01 4.57025588e-01 -3.58537972e-01
8.17477822e-01 1.64562359e-01 -1.03657699e+00 4.97977704e-01
-7.50331104e-01 -1.00224987e-01 4.48249400e-01 -4.91583705e-01
-8.87517273e-01 1.04790904e-01 1.26553506e-01 -3.17987144e-01
1.46878943e-01 5.05551636e-01 5.74630015e-02 -1.55793041e-01
4.48445439e-01 6.00104153e-01 2.43673712e-01 -4.27737772e-01
-1.36207342e-01 9.68665540e-01 3.28431159e-01 -2.23038644e-02
5.16171753e-01 5.18659651e-01 3.29360008e-01 -1.05112946e+00
-3.03730994e-01 -1.24239409e+00 -5.15090466e-01 -2.51966298e-01
5.65360725e-01 -6.97374582e-01 -7.80635834e-01 2.36014605e-01
-1.00746179e+00 7.02005446e-01 1.65938601e-01 5.75824082e-01
-5.56906089e-02 3.69537175e-01 -4.69346344e-01 -1.11771703e+00
-8.49518478e-01 -1.16237724e+00 8.44353735e-01 4.45918679e-01
2.34229326e-01 -5.73253572e-01 -1.22016646e-01 2.31773943e-01
3.56547445e-01 -1.57427087e-01 7.76882350e-01 -6.76360726e-01
-5.91222167e-01 -8.59326482e-01 -5.72852492e-01 1.99135646e-01
4.80081052e-01 2.85733670e-01 -5.58546722e-01 -3.27948689e-01
7.42719918e-02 2.26843372e-01 6.71459913e-01 3.58285546e-01
1.09501994e+00 -3.58896047e-01 -3.71079206e-01 2.77794246e-02
1.77398312e+00 1.03545356e+00 1.01058257e+00 5.80599010e-01
1.81058735e-01 4.48522300e-01 1.11529410e+00 3.21858108e-01
-1.22214392e-01 4.36098158e-01 -1.38453752e-01 -1.47341266e-01
8.41445774e-02 6.67641908e-02 -3.23735952e-01 7.27221429e-01
2.00948548e-02 -6.67163283e-02 -6.67227685e-01 6.79522634e-01
-1.49837577e+00 -9.74252641e-01 -2.03516945e-01 2.46199369e+00
6.96064770e-01 -2.27106869e-01 -7.90116042e-02 6.78561389e-01
9.05070305e-01 -1.75971657e-01 -4.46842253e-01 -2.90733248e-01
2.12276459e-01 2.40626082e-01 3.38159829e-01 5.88607132e-01
-9.82657790e-01 8.25134277e-01 6.30731583e+00 1.08992898e+00
-1.36661243e+00 -9.22355130e-02 5.26363015e-01 2.21291393e-01
5.12810834e-02 1.29986644e-01 -6.74154818e-01 1.81547314e-01
3.64431709e-01 -6.91744208e-01 2.30405852e-01 1.02132082e+00
1.87731072e-01 -8.53848815e-01 -5.75070083e-01 1.35142267e+00
1.55702814e-01 -9.57661986e-01 5.70636570e-01 -7.60206431e-02
5.49589217e-01 -7.13204741e-01 -1.88944638e-02 -1.98113382e-01
-4.69009191e-01 -5.75242639e-01 2.61658430e-01 4.85310942e-01
5.89949250e-01 -1.26140797e+00 9.93605137e-01 3.30507666e-01
-1.29326034e+00 1.86011001e-01 -9.75487590e-01 1.40721858e-01
-2.00043440e-01 6.48417771e-01 -8.17718863e-01 5.46169698e-01
7.27042556e-01 1.70710802e-01 -5.55379272e-01 1.39471292e+00
2.66259819e-01 -5.27593270e-02 -2.80038387e-01 -8.82737562e-02
-5.45487218e-02 -4.40063983e-01 2.89441913e-01 8.93875957e-01
4.60709244e-01 4.51402664e-01 1.07646994e-01 2.69394815e-01
2.36001387e-02 8.54168236e-01 -8.30273271e-01 1.42561391e-01
6.32091224e-01 1.12775028e+00 -1.21880758e+00 -5.97736239e-01
-7.72937760e-02 1.12949479e+00 -4.61393356e-01 1.74663723e-01
-3.64424974e-01 -1.00921094e+00 -3.15853506e-02 -6.80869892e-02
2.27059182e-02 6.47897497e-02 7.58237988e-02 -7.13429689e-01
-2.00271666e-01 -6.32345259e-01 5.28737545e-01 -8.48611772e-01
-7.29481280e-01 5.30060410e-01 1.62615985e-01 -1.55507767e+00
-1.97961062e-01 -2.39550769e-01 -1.99110627e-01 8.71010780e-01
-1.35165048e+00 -7.78272986e-01 -3.12097251e-01 9.70629096e-01
6.14767849e-01 1.41469106e-01 7.51944661e-01 5.68172812e-01
-1.55433983e-01 4.32173789e-01 4.47134495e-01 2.11660028e-01
7.26951778e-01 -5.83034396e-01 -5.49685240e-01 7.45478153e-01
1.87998161e-01 8.72518182e-01 5.51820815e-01 -6.00138426e-01
-1.28116250e+00 -4.89840418e-01 1.01241732e+00 3.61147225e-01
1.04928017e-01 2.55214423e-01 -7.91360676e-01 1.36663139e-01
-4.72221710e-02 -1.09433932e-02 5.53677797e-01 -4.75802034e-01
-7.95250610e-02 -6.73604369e-01 -1.36089325e+00 5.86946189e-01
2.24473432e-01 -6.99750125e-01 -3.37822646e-01 3.95205587e-01
3.51811856e-01 -1.10380530e-01 -7.53588438e-01 2.10968792e-01
9.13223982e-01 -8.29084158e-01 1.33713603e+00 1.30960509e-01
3.03062368e-02 -5.19759119e-01 -2.57605389e-02 -7.13705182e-01
-4.88691078e-03 -1.03410386e-01 7.56374180e-01 1.03605306e+00
-3.98450643e-02 -3.17515075e-01 4.33568180e-01 6.61667585e-01
5.62967181e-01 -2.57680535e-01 -6.88924849e-01 -9.27392304e-01
-2.99366295e-01 1.71116859e-01 6.17451310e-01 5.67748725e-01
-2.18373761e-01 2.52906531e-01 -2.34440833e-01 -6.31511658e-02
5.14359832e-01 4.19444680e-01 5.11359096e-01 -1.30016291e+00
2.19654411e-01 -2.99175918e-01 -5.96402407e-01 -7.87696660e-01
-4.62571025e-01 -3.49733353e-01 9.56008881e-02 -1.65852964e+00
5.35358608e-01 -3.78538996e-01 -4.99030024e-01 2.23594025e-01
-2.21036807e-01 6.65003300e-01 3.43499601e-01 9.36546862e-01
-5.34735322e-01 -8.65836814e-02 1.40448511e+00 -2.75693983e-01
-4.03170198e-01 -2.60017440e-02 -1.68968573e-01 4.77145672e-01
8.55516076e-01 -5.57116449e-01 -6.84564412e-01 -1.54354051e-01
-4.23760526e-02 3.82789642e-01 2.38605775e-02 -8.99377167e-01
3.58015656e-01 -1.86630324e-01 5.45908034e-01 -1.13548732e+00
3.29133898e-01 -1.09277940e+00 3.00156176e-01 6.70376778e-01
-1.59258559e-01 3.66579145e-01 1.30403966e-01 2.92485416e-01
-6.83831692e-01 -8.02009344e-01 6.50322735e-01 -2.78935522e-01
-8.90126705e-01 1.72145873e-01 -4.85030830e-01 -7.17903972e-01
9.79045451e-01 -4.69188780e-01 5.11566699e-02 -5.44323623e-01
-5.13144493e-01 -1.91546798e-01 4.04020101e-01 -6.17227033e-02
9.67656612e-01 -1.37922442e+00 -3.76062125e-01 4.92169671e-02
2.83058107e-01 -9.20901820e-02 3.19901347e-01 5.04257381e-01
-1.13563418e+00 5.44965565e-01 -2.22117603e-01 -5.51531017e-01
-1.89871013e+00 9.68409181e-01 -1.75335556e-01 -3.64006341e-01
-4.72251594e-01 4.68291819e-01 -6.84686974e-02 2.54546434e-01
-5.78764081e-02 4.56314310e-02 -4.96184230e-01 2.58057117e-01
1.00085139e+00 3.84580880e-01 2.49579296e-01 -7.98290014e-01
-3.49335313e-01 1.00828433e+00 -4.39752221e-01 -3.95136744e-01
8.35975647e-01 -3.62326652e-01 -6.09224916e-01 1.20029785e-01
1.60217440e+00 1.03599519e-01 -2.58735001e-01 -2.36108914e-01
8.46705958e-02 -1.01690388e+00 2.16986835e-01 -4.51856345e-01
-9.85821903e-01 7.07808554e-01 9.89237428e-01 2.45193064e-01
1.77194893e+00 -3.01311046e-01 6.92576110e-01 7.98556089e-01
5.46422482e-01 -1.28274727e+00 5.10679893e-02 1.39222771e-01
8.99946392e-01 -1.41723526e+00 4.87242311e-01 -3.80572140e-01
-3.64737809e-01 1.19715381e+00 2.07717925e-01 -2.66428292e-01
7.12983489e-01 -1.93527490e-01 8.60256404e-02 -1.16132326e-01
-2.21882328e-01 -3.18829060e-01 2.93540180e-01 3.47836614e-01
2.58477062e-01 -3.26864541e-01 -9.99070525e-01 -2.14650538e-02
7.94048682e-02 1.17522292e-01 3.99151057e-01 1.36978805e+00
-7.48115957e-01 -1.11648953e+00 -9.82274652e-01 2.98939168e-01
-7.34463573e-01 -1.83945224e-01 -2.63910592e-01 8.90444994e-01
-3.19686204e-01 1.29293787e+00 2.80407369e-01 -9.76523980e-02
1.81744993e-01 -9.92527977e-02 4.16529655e-01 -1.17165513e-01
-2.72010535e-01 5.80656528e-01 -4.35279071e-01 -2.63658226e-01
-5.03097653e-01 -1.58126965e-01 -1.43072701e+00 -4.06210154e-01
-7.81358480e-01 5.26494801e-01 8.98717642e-01 4.69133377e-01
1.96981430e-01 9.83493775e-02 7.55431831e-01 -3.90328079e-01
-1.37365848e-01 -6.65144920e-01 -7.07104146e-01 5.44754982e-01
2.31784120e-01 -5.01218617e-01 -1.88069969e-01 2.82823503e-01] | [10.768134117126465, 0.08648086339235306] |
3321bdee-4a41-4989-ae64-e856f9f66a0e | video-summarization-using-keyframe-extraction | 1910.04792 | null | https://arxiv.org/abs/1910.04792v2 | https://arxiv.org/pdf/1910.04792v2.pdf | Unsupervised video summarization framework using keyframe extraction and video skimming | Video is one of the robust sources of information and the consumption of online and offline videos has reached an unprecedented level in the last few years. A fundamental challenge of extracting information from videos is a viewer has to go through the complete video to understand the context, as opposed to an image where the viewer can extract information from a single frame. Apart from context understanding, it almost impossible to create a universal summarized video for everyone, as everyone has their own bias of keyframe, e.g; In a soccer game, a coach person might consider those frames which consist of information on player placement, techniques, etc; however, a person with less knowledge about a soccer game, will focus more on frames which consist of goals and score-board. Therefore, if we were to tackle problem video summarization through a supervised learning path, it will require extensive personalized labeling of data. In this paper, we attempt to solve video summarization through unsupervised learning by employing traditional vision-based algorithmic methodologies for accurate feature extraction from video frames. We have also proposed a deep learning-based feature extraction followed by multiple clustering methods to find an effective way of summarizing a video by interesting key-frame extraction. We have compared the performance of these approaches on the SumMe dataset and showcased that using deep learning-based feature extraction has been proven to perform better in case of dynamic viewpoint videos. | ['Shruti Jadon', 'Mahmood Jasim'] | 2019-10-10 | null | null | null | null | ['unsupervised-video-summarization'] | ['computer-vision'] | [ 2.75814980e-01 -1.30248843e-02 1.19117953e-01 -1.92340612e-01
-8.65725338e-01 -4.97062206e-01 3.11431497e-01 4.49462026e-01
-4.58885372e-01 6.31472707e-01 2.68060893e-01 6.13704957e-02
-3.07374418e-01 -5.75810194e-01 -7.48076320e-01 -5.39602339e-01
-1.41971350e-01 2.83538431e-01 3.62969756e-01 -2.05260664e-01
5.78409970e-01 2.90458351e-01 -2.05970383e+00 6.21211350e-01
2.47066751e-01 9.95666921e-01 5.19631207e-01 8.62896144e-01
-1.38530627e-01 1.10729635e+00 -7.78595984e-01 -2.50791639e-01
3.28162730e-01 -5.91172755e-01 -7.58848429e-01 6.05565548e-01
6.44722044e-01 -4.18466985e-01 -1.33786872e-01 8.54746938e-01
3.36329043e-01 4.12033111e-01 5.53504944e-01 -1.07126176e+00
2.00071961e-01 6.50155246e-01 -5.42487562e-01 3.68797570e-01
8.06491435e-01 7.86825418e-02 9.65634823e-01 -5.27219534e-01
1.01378715e+00 7.23492265e-01 4.60374236e-01 1.08936176e-01
-7.32422411e-01 -1.66414589e-01 1.19746439e-01 7.22626090e-01
-1.15420437e+00 -3.55457872e-01 9.66520786e-01 -6.33119226e-01
6.78903699e-01 3.07350248e-01 1.05308509e+00 9.30403173e-01
3.47425006e-02 6.62201047e-01 8.71326983e-01 -5.22407532e-01
3.54768097e-01 1.47866637e-01 1.16754323e-01 5.47292352e-01
1.21633619e-01 -2.48175338e-01 -6.38189197e-01 3.55995297e-01
4.36190367e-01 2.12990552e-01 -3.38345587e-01 -4.64157045e-01
-1.05656862e+00 5.70451379e-01 -4.47861627e-02 4.40988570e-01
-4.92958546e-01 -1.60380639e-02 6.58162177e-01 2.80849397e-01
3.40169482e-02 2.77569473e-01 -2.30003417e-01 -4.55665469e-01
-1.52293575e+00 5.03501117e-01 9.00840819e-01 7.81011760e-01
8.20323467e-01 -9.14358944e-02 5.22248074e-02 4.14460361e-01
4.78658220e-03 -1.37450546e-01 4.40959394e-01 -1.35013294e+00
4.50657189e-01 7.89247036e-01 1.24060877e-01 -1.56352711e+00
-3.71934444e-01 -1.36064053e-01 -5.67405581e-01 3.98683757e-01
5.82128704e-01 -1.15704745e-01 -6.36119604e-01 1.06157160e+00
1.86654449e-01 -1.00753002e-01 -1.20537102e-01 1.02315366e+00
9.07288671e-01 7.98648775e-01 -2.02232718e-01 -6.63053155e-01
1.44785690e+00 -8.25690031e-01 -7.29260206e-01 1.19741842e-01
2.57786453e-01 -7.98711181e-01 6.49171770e-01 9.80113685e-01
-1.08539975e+00 -7.48954892e-01 -1.13066709e+00 -6.87351376e-02
-3.32098186e-01 3.70334163e-02 2.67918617e-01 3.99017155e-01
-8.44924212e-01 7.33877957e-01 -7.20583737e-01 -6.60630226e-01
2.79691070e-01 3.48247379e-01 -6.88603103e-01 -1.43102646e-01
-7.34773457e-01 7.24574864e-01 7.05903769e-01 -2.01075058e-03
-1.04469478e+00 -4.38133806e-01 -6.89776599e-01 2.86352515e-01
8.61834288e-01 -4.66323733e-01 1.02483082e+00 -1.31299996e+00
-1.30434549e+00 5.88943541e-01 -9.51150209e-02 -5.87302983e-01
6.62689507e-01 -3.21464717e-01 -2.40493175e-02 6.36482656e-01
-9.77411941e-02 4.18272436e-01 1.02275503e+00 -1.29453838e+00
-1.02319241e+00 -2.48063549e-01 5.04535377e-01 2.47957915e-01
-2.98368961e-01 6.29153475e-02 -4.36373532e-01 -4.49951708e-01
-2.47658752e-02 -6.25442743e-01 -2.69795191e-02 -3.48973691e-01
-1.25772968e-01 4.41008247e-02 1.03925550e+00 -1.00959194e+00
1.35992765e+00 -2.04544854e+00 3.88677448e-01 7.29975700e-02
3.25489759e-01 3.00811380e-01 4.11003411e-01 7.12987840e-01
-1.33281186e-01 -1.72185600e-01 -2.25207154e-02 -1.82274595e-01
-1.79274559e-01 4.56100516e-02 -2.07481682e-02 3.96743834e-01
-9.43097770e-02 2.26205811e-01 -9.36689794e-01 -9.04259741e-01
4.83895421e-01 3.55011553e-01 -6.98584318e-01 1.29288644e-01
-1.18518539e-01 3.92616451e-01 -2.52262920e-01 5.60962439e-01
3.62376720e-01 1.91346835e-02 2.31527403e-01 -4.90425944e-01
-4.22866642e-01 -3.72293919e-01 -1.51422584e+00 1.88040280e+00
-1.96790874e-01 9.84162569e-01 -7.62886703e-02 -1.48828316e+00
6.70329273e-01 4.08298731e-01 8.25874686e-01 -3.87728721e-01
3.03635210e-01 -3.13460976e-02 -2.11843237e-01 -9.24479187e-01
9.21935260e-01 1.25090569e-01 1.73754580e-02 3.18103194e-01
2.19102651e-01 1.57411367e-01 7.18199015e-01 2.25460917e-01
1.15901721e+00 3.69409800e-01 5.33317924e-01 1.46507964e-01
6.43501759e-01 2.62667179e-01 4.85049605e-01 5.96945822e-01
-2.24272013e-01 8.67894053e-01 6.34565473e-01 -6.83880031e-01
-1.11705208e+00 -6.47498250e-01 3.99958789e-01 8.17490578e-01
4.89673950e-02 -7.38118052e-01 -8.41716528e-01 -4.06572551e-01
-5.44464707e-01 2.73171812e-01 -3.93117547e-01 1.45926356e-01
-7.19304800e-01 -2.39011288e-01 -1.46124795e-01 3.19042839e-02
5.74463904e-01 -9.17190075e-01 -1.26281869e+00 3.98586035e-01
-3.37917745e-01 -1.23143554e+00 -1.79874435e-01 6.33394718e-02
-6.74362302e-01 -1.25248361e+00 -5.88152051e-01 -6.85286403e-01
3.97607684e-01 3.97378027e-01 9.75891531e-01 2.03620549e-02
-5.27532518e-01 6.76451445e-01 -8.99169505e-01 -2.35761285e-01
-2.81800777e-01 9.12252888e-02 -2.08089322e-01 1.70370638e-01
2.15799496e-01 -6.25555873e-01 -7.07921624e-01 4.57701236e-02
-9.64550376e-01 9.45057645e-02 5.51472604e-01 3.61002833e-01
5.73939800e-01 4.66544092e-01 1.82528034e-01 -6.87038779e-01
2.76656598e-01 -4.25306797e-01 -3.57489765e-01 2.46629283e-01
6.67984486e-02 -2.27078125e-01 6.88452065e-01 -7.16924593e-02
-9.28126156e-01 3.84478360e-01 1.70872509e-02 -4.34011072e-01
-4.45176840e-01 3.82258683e-01 -1.35884285e-01 1.27623603e-01
4.69937354e-01 3.06682527e-01 -2.17605501e-01 -2.84950763e-01
2.31798947e-01 6.42803073e-01 4.87694621e-01 -3.36289436e-01
6.04871511e-01 4.35865670e-01 -4.24287468e-03 -1.20995879e+00
-5.91959059e-01 -7.60970891e-01 -8.35149944e-01 -8.84194016e-01
1.08498812e+00 -7.93941617e-01 -7.64643252e-01 2.80836046e-01
-1.19666195e+00 2.08174199e-01 -3.90336484e-01 4.54479009e-01
-8.50667953e-01 7.20836341e-01 -2.15261191e-01 -7.69625783e-01
-4.34372295e-03 -1.19104075e+00 8.17899227e-01 4.64264184e-01
-3.02415222e-01 -7.56682456e-01 -8.58188942e-02 7.58221805e-01
2.85262708e-03 6.71546459e-01 4.84816581e-01 -7.52454996e-01
-8.21006298e-01 -2.78959632e-01 7.55151510e-02 4.19094920e-01
1.41125977e-01 1.95033178e-01 -7.86657810e-01 -1.57739490e-01
1.40028834e-01 -8.41518566e-02 6.25707924e-01 4.94602323e-01
1.26416123e+00 -2.65011758e-01 -3.36292312e-02 3.88131052e-01
1.55644202e+00 3.08468014e-01 6.77807689e-01 4.52120870e-01
6.43083155e-01 7.34921038e-01 9.20746267e-01 6.07490540e-01
3.20220619e-01 6.94642425e-01 5.72101533e-01 2.23861411e-01
-3.11181881e-02 -4.63127904e-02 3.58272702e-01 7.72238016e-01
-6.22217894e-01 -2.19753191e-01 -5.83035290e-01 5.13769686e-01
-2.01231575e+00 -1.50659454e+00 7.87747875e-02 2.14257431e+00
2.17980564e-01 1.35695338e-01 4.96905804e-01 6.32894754e-01
7.06896484e-01 1.12091519e-01 -1.98321998e-01 -3.52555633e-01
1.25904724e-01 -4.07596976e-02 3.34441423e-01 2.02906072e-01
-1.20477796e+00 6.30652070e-01 5.48548079e+00 8.89487326e-01
-9.13152695e-01 1.01164661e-01 5.01471162e-01 -3.76600593e-01
1.30889863e-01 4.02244329e-02 -6.05320513e-01 5.90205908e-01
7.36162603e-01 -4.13732082e-02 2.65475214e-01 8.66022587e-01
6.44987285e-01 -7.08660841e-01 -1.03710163e+00 1.28540587e+00
4.48789358e-01 -1.37163126e+00 1.70758218e-01 5.93525879e-02
6.93760931e-01 -5.63771784e-01 -3.38648766e-01 1.12636879e-01
-2.63940364e-01 -8.31209481e-01 8.56278181e-01 6.83790445e-01
1.49223492e-01 -9.52218771e-01 7.77033567e-01 4.03299689e-01
-1.21407676e+00 -3.23453397e-01 -2.23482192e-01 -3.08495075e-01
3.54944944e-01 3.95984471e-01 -6.77388906e-01 9.15987492e-01
8.96409929e-01 6.84093952e-01 -5.19332528e-01 1.27444243e+00
2.18961269e-01 2.18898550e-01 -1.60947129e-01 8.98368135e-02
2.26021439e-01 -2.66438156e-01 6.89352512e-01 1.16404653e+00
5.66513419e-01 2.02224627e-01 3.13514501e-01 2.57677734e-01
1.92617446e-01 2.63468057e-01 -6.29917979e-01 -8.02484676e-02
5.48616908e-02 1.49330604e+00 -1.12465060e+00 -3.20942491e-01
-4.24588084e-01 9.22193408e-01 -9.35318395e-02 1.51808023e-01
-8.75687957e-01 -3.91536474e-01 2.91738212e-01 5.11990845e-01
5.88518500e-01 -4.44198012e-01 2.59586126e-01 -1.13842225e+00
1.13078035e-01 -1.09514117e+00 4.42202210e-01 -7.87387013e-01
-5.44906199e-01 5.36603510e-01 2.07334608e-01 -1.54889727e+00
-4.93480384e-01 -5.48313141e-01 -5.45760512e-01 1.55072287e-01
-1.08995676e+00 -9.47118819e-01 -5.70756018e-01 6.38985455e-01
1.13502240e+00 -3.08684766e-01 2.16281235e-01 4.40592498e-01
-3.43946427e-01 3.70496735e-02 -1.30954102e-01 -1.61873959e-02
5.03584325e-01 -1.13911772e+00 -4.40149605e-01 1.02721298e+00
5.00239313e-01 2.77328819e-01 1.18604779e+00 -4.28928167e-01
-1.59231055e+00 -6.05829537e-01 5.24863601e-01 -3.08535874e-01
4.31866437e-01 -1.03018982e-02 -6.43759012e-01 4.73404199e-01
4.26906019e-01 -3.70664924e-01 7.06693292e-01 -3.01708490e-01
4.54318404e-01 -4.08017069e-01 -8.08566570e-01 3.89817595e-01
8.57425630e-01 -2.79356062e-01 -8.76217246e-01 3.03573728e-01
3.23619038e-01 -3.36865395e-01 -6.87559366e-01 1.84554532e-01
5.76041162e-01 -1.46884096e+00 7.76827335e-01 -3.57169271e-01
5.62950015e-01 -5.82717836e-01 -1.60223246e-01 -1.08848834e+00
1.79827616e-01 -5.48009932e-01 -3.74184549e-02 1.38654065e+00
-1.12777300e-01 2.02682137e-01 9.75680470e-01 2.65715867e-01
-1.37790993e-01 -4.72425461e-01 -6.46308362e-01 -3.11343104e-01
-6.41262114e-01 -5.43993533e-01 7.79874399e-02 5.95974326e-01
4.15454693e-02 -2.69560963e-02 -5.81733108e-01 -4.20553274e-02
6.44845843e-01 1.78155407e-01 1.11588800e+00 -1.16045022e+00
-4.06884044e-01 -2.99148053e-01 -8.93444538e-01 -8.00319016e-01
-6.56155199e-02 -3.82452309e-01 -1.71210855e-01 -1.72320986e+00
3.19280088e-01 3.29406738e-01 5.23642031e-03 -5.46925515e-02
1.95486829e-01 2.64829397e-01 5.24383307e-01 4.00639810e-02
-1.05347037e+00 -1.28235314e-02 9.85617697e-01 -5.34227751e-02
-1.99303195e-01 -1.82966758e-02 -3.68515193e-01 9.03384686e-01
4.96899575e-01 -2.85837859e-01 -5.06914496e-01 -9.13578048e-02
3.75717103e-01 3.82433772e-01 3.53434861e-01 -1.45876086e+00
5.95520437e-01 -2.56539863e-02 4.23856825e-01 -8.03476632e-01
3.43757153e-01 -1.03692949e+00 4.85573441e-01 3.47031593e-01
-4.38021943e-02 5.36764599e-02 -1.18695140e-01 6.19478643e-01
-5.77121615e-01 -4.63910401e-01 4.26734865e-01 -6.26865566e-01
-1.31634057e+00 5.18968776e-02 -6.74753845e-01 -1.90154865e-01
1.35211623e+00 -8.28403711e-01 -2.25149207e-02 -5.55514097e-01
-1.02517998e+00 2.64372062e-02 4.99140024e-01 2.35265583e-01
6.46111906e-01 -8.64484906e-01 -4.79788840e-01 -1.54748663e-01
-1.08659789e-01 -1.74955744e-02 7.39618421e-01 7.11008251e-01
-1.15085530e+00 2.58583784e-01 -6.52337074e-01 -7.22496271e-01
-1.62547004e+00 6.44046664e-01 -2.33386941e-02 -1.57845780e-01
-6.33784235e-01 3.25084031e-01 -7.53124803e-02 3.64077747e-01
3.22755992e-01 -2.69701809e-01 -7.79824257e-01 7.11511672e-01
7.12910414e-01 5.45498192e-01 1.62755847e-01 -8.05848897e-01
-8.49952102e-02 8.00482869e-01 -3.25576179e-02 3.68492417e-02
1.50309432e+00 -3.85668427e-01 5.40657640e-02 4.83209163e-01
1.13384283e+00 -6.66047931e-02 -1.24833548e+00 1.84445322e-01
1.03989445e-01 -5.89014173e-01 1.12240892e-02 -3.33340824e-01
-1.04995739e+00 7.63265848e-01 5.74794114e-01 4.13721204e-01
1.38139713e+00 -1.79923743e-01 6.58656538e-01 4.08605129e-01
6.07210100e-01 -1.28231335e+00 2.63663501e-01 1.72643512e-01
6.16758525e-01 -1.31658161e+00 2.96820968e-01 -1.77653491e-01
-6.44518912e-01 1.51443803e+00 2.43900150e-01 -1.42600298e-01
3.72499049e-01 6.62595779e-02 -1.60153359e-01 -2.58161604e-01
-5.67746937e-01 -2.90715605e-01 1.93056002e-01 4.20504987e-01
1.96747229e-01 -2.01053798e-01 -4.17473316e-01 4.46797192e-01
-2.36105531e-01 2.07472086e-01 9.64917064e-01 1.01748133e+00
-8.55279207e-01 -1.05134606e+00 -5.02637327e-01 3.13577473e-01
-5.98894775e-01 4.26708370e-01 -2.58828640e-01 9.72757220e-01
5.38522363e-01 9.12338614e-01 3.30234915e-02 -3.79816025e-01
3.50451767e-01 8.18684623e-02 6.36939287e-01 -6.75389647e-01
-6.18697524e-01 8.56052190e-02 -7.75433378e-03 -6.22765481e-01
-9.54297245e-01 -9.37406123e-01 -9.63258803e-01 -2.50479639e-01
8.21779445e-02 2.55861014e-01 7.53544211e-01 1.14373577e+00
-1.16520435e-01 6.10001802e-01 4.24876392e-01 -1.31173360e+00
8.70584771e-02 -6.31731629e-01 -5.71913719e-01 4.60573912e-01
3.03760499e-01 -6.22054696e-01 -2.47323588e-01 6.87011421e-01] | [8.412813186645508, 0.11022111028432846] |
ebdefc0f-17f2-46fa-a9a0-8273cb813c9e | deep-ppg-large-scale-heart-rate-estimation | null | null | https://doi.org/10.3390/s19143079 | https://www.mdpi.com/1424-8220/19/14/3079/pdf | Deep PPG: Large-Scale Heart Rate Estimation with Convolutional Neural Networks | Photoplethysmography (PPG)-based continuous heart rate monitoring is essential in a number of domains, e.g., for healthcare or fitness applications. Recently, methods based on time-frequency spectra emerged to address the challenges of motion artefact compensation. However, existing approaches are highly parametrised and optimised for specific scenarios of small, public datasets. We address this fragmentation by contributing research into the robustness and generalisation capabilities of PPG-based heart rate estimation approaches. First, we introduce a novel large-scale dataset (called PPG-DaLiA), including a wide range of activities performed under close to real-life conditions. Second, we extend a state-of-the-art algorithm, significantly improving its performance on several datasets. Third, we introduce deep learning to this domain, and investigate various convolutional neural network architectures. Our end-to-end learning approach takes the time-frequency spectra of synchronised PPG- and accelerometer-signals as input, and provides the estimated heart rate as output. Finally, we compare the novel deep learning approach to classical methods, performing evaluation on four public datasets. We show that on large datasets the deep learning model significantly outperforms other methods: The mean absolute error could be reduced by 31% on the new dataset PPG-DaLiA, and by 21% on the dataset WESAD. | ['Ina Indlekofer', 'Attila Reiss', 'Philip Schmidt', 'Kristof Van Laerhoven'] | 2019-07-12 | null | null | null | sensors-2019-7 | ['photoplethysmography-ppg', 'heart-rate-estimation'] | ['medical', 'medical'] | [ 1.77248687e-01 -1.17173716e-02 -6.17819354e-02 -2.44773954e-01
-7.52758443e-01 -1.72540873e-01 1.66213065e-01 -7.51426220e-02
-4.13514018e-01 7.65208125e-01 3.45157951e-01 1.00334145e-01
1.38612716e-02 -4.57623690e-01 -4.45547312e-01 -5.28216898e-01
-4.73175883e-01 -4.77992110e-02 -2.93421179e-01 -2.06138566e-02
2.49723196e-02 3.23240995e-01 -1.09070051e+00 -3.14778578e-03
5.48274279e-01 1.23963916e+00 -6.00153923e-01 1.00472188e+00
7.10905552e-01 4.06650782e-01 -9.35480595e-01 -5.13560288e-02
4.96363416e-02 -6.18662000e-01 -5.22009134e-01 -3.14499468e-01
5.64207852e-01 -4.52207834e-01 -2.03952000e-01 3.29911143e-01
1.40695536e+00 -1.05126686e-01 5.77601269e-02 -1.07386339e+00
-2.51411229e-01 1.80547506e-01 -1.57569334e-01 4.01399374e-01
4.98354822e-01 4.09450859e-01 3.56844753e-01 -3.83426189e-01
2.95174628e-01 6.82286263e-01 1.43719852e+00 5.47533691e-01
-1.35156178e+00 -4.55944806e-01 -4.97109443e-01 6.98966980e-02
-1.26396728e+00 -4.89325464e-01 9.15087342e-01 -2.45658934e-01
1.13930261e+00 5.05684674e-01 1.03231478e+00 1.41787672e+00
3.28109801e-01 4.72863883e-01 1.25932717e+00 -1.90147489e-01
3.20510656e-01 -2.03075394e-01 -2.26242706e-01 3.09331715e-01
3.97018820e-01 1.63125888e-01 -8.07836533e-01 -1.07045226e-01
6.36367083e-01 -2.71663874e-01 -4.46019530e-01 -1.88449010e-01
-1.41479576e+00 4.64730918e-01 -4.71352227e-02 4.09193814e-01
-4.68286902e-01 5.12306094e-01 6.53487444e-01 1.56720191e-01
6.27794504e-01 4.92727339e-01 -7.32666433e-01 -8.76753986e-01
-1.09095263e+00 4.36712384e-01 1.10407674e+00 3.35204035e-01
1.39665633e-01 1.16985641e-01 -1.59596160e-01 6.92037106e-01
2.42992163e-01 5.06655514e-01 7.72227824e-01 -1.07336891e+00
2.71214664e-01 2.47122750e-01 1.82742774e-01 -8.84693265e-01
-1.13489151e+00 -4.54566866e-01 -9.82882857e-01 -1.42882422e-01
5.47174931e-01 -3.96366566e-01 -5.49758315e-01 1.77667594e+00
3.78222436e-01 5.56912184e-01 -1.39289573e-01 1.07532287e+00
9.00321424e-01 1.58156604e-01 8.03863406e-02 -4.36666429e-01
1.29068398e+00 -4.47256476e-01 -8.37298155e-01 -2.61217982e-01
6.41400397e-01 -3.11113626e-01 9.99240518e-01 6.94380105e-01
-1.08193445e+00 -8.09907556e-01 -1.23199081e+00 -7.65741765e-02
-2.97491074e-01 1.78225040e-01 4.10165846e-01 1.17916644e+00
-1.01316094e+00 1.23684144e+00 -9.63813901e-01 -4.12383020e-01
2.65584260e-01 3.87415260e-01 -2.56194919e-01 3.89389306e-01
-1.65320253e+00 8.10971797e-01 7.99727812e-02 4.21632648e-01
-5.06995142e-01 -8.55967343e-01 -8.38542640e-01 -9.87054184e-02
-6.77109212e-02 -6.56111002e-01 1.11690164e+00 -3.69185328e-01
-1.98314273e+00 8.32997501e-01 2.59144992e-01 -7.77286351e-01
9.60119963e-01 -6.72953665e-01 -7.05051482e-01 2.46433333e-01
-2.88723767e-01 3.25346768e-01 7.40784824e-01 -4.82707411e-01
-1.88961197e-02 -2.07088768e-01 -3.66419911e-01 3.22628729e-02
-2.38171548e-01 -2.37028256e-01 -2.48428345e-01 -4.33576673e-01
-2.02844292e-01 -1.08774936e+00 3.59397270e-02 -7.34966993e-02
-1.87620938e-01 8.47281143e-02 4.38741565e-01 -8.34986925e-01
1.28306735e+00 -2.09718871e+00 -4.52878661e-02 -8.68398920e-02
3.89520645e-01 5.68198264e-01 2.61913598e-01 2.79810727e-01
-1.83456525e-01 2.78692432e-02 -1.78933188e-01 -5.27625740e-01
7.10037202e-02 1.48411438e-01 1.24094263e-01 8.52088332e-01
-2.26422143e-03 9.44936872e-01 -7.16097414e-01 -1.85321495e-01
6.97558820e-01 7.62910903e-01 -2.48779461e-01 1.98192060e-01
3.61229360e-01 6.67492986e-01 1.15034640e-01 5.58589756e-01
4.51033443e-01 -1.92531690e-01 2.56343156e-01 -4.66860116e-01
1.57551393e-01 2.49900341e-01 -1.08046591e+00 2.02098632e+00
-6.40208066e-01 8.16100597e-01 -4.90248591e-01 -9.58115578e-01
1.10805905e+00 5.39122641e-01 9.83502328e-01 -7.88674235e-01
1.88870683e-01 3.38231444e-01 9.03776810e-02 -8.84259224e-01
1.52170897e-01 -3.33943605e-01 -2.88489789e-01 4.08610672e-01
4.31992151e-02 3.88569012e-02 -1.74464852e-01 -2.21604988e-01
1.05191231e+00 3.22106153e-01 4.67934459e-01 -2.44488940e-01
4.64475781e-01 -5.74523985e-01 4.57971931e-01 6.42144024e-01
-7.71663487e-01 1.04011106e+00 4.68151093e-01 -8.99026990e-01
-7.57315755e-01 -7.71309912e-01 -2.44370282e-01 5.53312898e-01
-1.55184731e-01 -4.93779629e-01 -7.64052927e-01 -3.08327585e-01
2.12575272e-01 4.04529959e-01 -9.40209508e-01 -4.17776287e-01
-8.06772351e-01 -1.20238483e+00 1.13591266e+00 8.17798138e-01
7.64172733e-01 -1.04816449e+00 -1.27492547e+00 4.94838357e-01
-4.52692091e-01 -1.30973721e+00 -2.09732920e-01 1.09459661e-01
-1.02117515e+00 -1.11476457e+00 -7.85766959e-01 1.45846412e-01
-3.44554454e-01 -4.07711416e-01 1.37161481e+00 -1.70937225e-01
-5.66122472e-01 5.73095679e-01 -1.16410412e-01 -4.10057843e-01
-1.30400375e-01 2.14817360e-01 3.19108933e-01 1.36310771e-01
5.39312720e-01 -7.47046232e-01 -1.09924817e+00 1.53413236e-01
-3.02199095e-01 -2.69614369e-01 2.20501393e-01 4.66848373e-01
3.91657591e-01 -7.25991070e-01 8.79410625e-01 -5.87286711e-01
8.13004732e-01 -2.66705364e-01 -3.37430239e-01 -3.04680705e-01
-7.43261039e-01 -3.18618506e-01 4.03188050e-01 -5.16567409e-01
-3.47012222e-01 6.19197376e-02 -3.25808942e-01 -4.16437984e-01
-1.83059797e-01 3.69598657e-01 -3.52681577e-02 5.02286153e-03
1.01137006e+00 3.40342075e-01 1.79699779e-01 -3.80468965e-01
1.60810441e-01 5.09934723e-01 1.08619714e+00 -4.72660571e-01
2.03143403e-01 4.44583327e-01 4.18256491e-01 -9.13790941e-01
-3.87388319e-01 -2.90723503e-01 -6.80919409e-01 -4.34148937e-01
9.22269285e-01 -1.19668138e+00 -9.61464822e-01 7.66235113e-01
-7.65118182e-01 -6.41564012e-01 -3.95385623e-01 7.45738387e-01
-9.31775570e-01 5.02678871e-01 -4.97044235e-01 -8.94070148e-01
-8.44806969e-01 -5.53176939e-01 1.25262046e+00 1.75275400e-01
-5.82143605e-01 -1.18436265e+00 5.16973317e-01 3.61906797e-01
7.84818590e-01 1.20229113e+00 5.73098212e-02 -5.00792325e-01
4.87348512e-02 -5.38570583e-01 2.54583418e-01 5.17290890e-01
1.37718588e-01 -4.42096502e-01 -1.49008358e+00 -4.38629210e-01
2.06681222e-01 -4.16142732e-01 6.07120752e-01 5.70905805e-01
1.09066868e+00 6.01490587e-02 -3.95958386e-02 9.54356134e-01
1.20379925e+00 -1.73115924e-01 1.14249694e+00 3.86234701e-01
6.57644093e-01 2.66122609e-01 2.74305761e-01 6.51764572e-01
3.35200638e-01 6.93351626e-01 2.45883375e-01 -3.76758873e-01
-1.44024715e-01 2.26207927e-01 3.14829111e-01 5.18050849e-01
-5.31021476e-01 8.11418146e-02 -6.09371066e-01 5.58342934e-01
-1.70085800e+00 -7.63222694e-01 -2.81304330e-01 2.40970325e+00
8.39498043e-01 -3.16599868e-02 6.02545381e-01 5.24088442e-01
4.42026287e-01 4.31133032e-01 -7.55229771e-01 -4.69252825e-01
3.90776508e-02 3.49854738e-01 4.79219645e-01 3.02008297e-02
-1.25354230e+00 1.20948710e-01 6.84335709e+00 2.76059061e-02
-1.44583488e+00 1.34420455e-01 6.58466458e-01 -4.22492683e-01
6.16423130e-01 -5.57140529e-01 -4.42213088e-01 7.37815201e-01
1.79404104e+00 -3.30168977e-02 3.02198172e-01 6.59757078e-01
5.76295376e-01 -4.17246930e-02 -1.25072193e+00 1.39416909e+00
2.98390865e-01 -1.07406914e+00 -1.05507588e+00 1.73101258e-02
2.67684817e-01 2.10641921e-01 -3.14532548e-01 2.70685524e-01
-5.16833246e-01 -9.85762477e-01 2.64051229e-01 6.69522166e-01
1.16171455e+00 -5.02405941e-01 8.24124753e-01 -3.59030999e-02
-1.06966150e+00 1.65713623e-01 -6.48588836e-02 -2.89000899e-01
1.48886383e-01 7.97395229e-01 -4.54860389e-01 6.54752851e-01
7.42738247e-01 8.30503702e-01 -5.72800398e-01 1.10654056e+00
-1.04828805e-01 7.66714752e-01 -5.95636368e-01 1.77069381e-01
-3.66010398e-01 3.51410151e-01 2.08094954e-01 1.38088310e+00
3.53250742e-01 -4.84469160e-03 -2.73841709e-01 7.69267976e-01
1.23886876e-02 -2.15693668e-01 -5.18398643e-01 2.17665121e-01
1.60228536e-01 1.37656868e+00 -3.81151736e-01 -4.32514757e-01
-3.52383047e-01 7.79815555e-01 -2.27869928e-01 3.03096492e-02
-1.13292933e+00 -6.60942614e-01 6.28927112e-01 1.78263053e-01
-8.47983882e-02 -4.36774008e-02 -4.11963254e-01 -1.22344732e+00
1.65816471e-01 -8.82368803e-01 4.61875021e-01 -7.38808393e-01
-9.36798334e-01 4.18517470e-01 9.07832161e-02 -1.37795258e+00
-6.36429965e-01 -3.89123023e-01 -6.14582598e-01 1.05720031e+00
-1.46396244e+00 -6.48324847e-01 -8.23058188e-01 4.52412546e-01
2.56967247e-01 2.24498704e-01 9.81252432e-01 7.71847486e-01
-5.97416699e-01 6.24074459e-01 -3.03741664e-01 8.60644057e-02
9.07508492e-01 -1.37613606e+00 8.15887451e-01 8.18888128e-01
-1.04268871e-01 3.35819483e-01 6.30497217e-01 -3.34195316e-01
-1.55422914e+00 -1.13355160e+00 9.05651033e-01 -6.87783182e-01
3.21955502e-01 -4.21528697e-01 -8.76820743e-01 3.58405322e-01
9.49463621e-02 6.49420202e-01 6.80552006e-01 -2.49033775e-02
-6.99621588e-02 -3.74775141e-01 -1.27943468e+00 3.77698392e-02
7.71500051e-01 -5.24419904e-01 -5.05599499e-01 2.93479651e-01
3.10523003e-01 -1.03044498e+00 -1.47018027e+00 4.61334914e-01
1.14751148e+00 -1.08416307e+00 1.03754973e+00 -3.28414530e-01
2.79611081e-01 -2.14270532e-01 1.83344349e-01 -1.34610820e+00
-1.07433483e-01 -1.10205340e+00 -6.65765464e-01 7.73273349e-01
-1.20784774e-01 -7.55174339e-01 7.87567317e-01 6.91483200e-01
-7.80465454e-02 -6.87662303e-01 -1.00817859e+00 -7.35572457e-01
-2.38529388e-02 -6.50320172e-01 4.53779399e-01 9.11536396e-01
1.36843786e-01 1.66804269e-01 -1.03313613e+00 -8.66529047e-02
5.56134939e-01 -8.69592428e-02 9.59981799e-01 -1.14516079e+00
-4.69777614e-01 8.72978196e-03 -5.69116414e-01 -5.38829923e-01
-5.24453342e-01 -3.62166733e-01 6.37947442e-03 -1.37120688e+00
-3.18283290e-01 9.52132642e-02 -6.14168167e-01 5.53479433e-01
-3.10942203e-01 9.04571652e-01 2.04376832e-01 -4.74693365e-02
-3.92126352e-01 1.85919002e-01 8.23997200e-01 9.42696482e-02
-5.75607181e-01 5.20677529e-02 -4.64798898e-01 5.53673387e-01
9.37848687e-01 -2.54113078e-01 -2.91921020e-01 4.48564105e-02
2.23247036e-01 2.80049771e-01 6.05881214e-01 -1.54651010e+00
-2.29380772e-01 3.26932818e-01 8.13796997e-01 -3.71682614e-01
3.95168185e-01 -5.07245541e-01 4.28114742e-01 7.81567991e-01
-2.96407431e-01 -1.92187876e-02 4.41874176e-01 3.00952107e-01
1.62616804e-01 4.50129241e-01 7.72771418e-01 -3.71702127e-02
-2.75931567e-01 3.74753624e-02 -1.30714580e-01 3.59857202e-01
4.49233443e-01 -3.43653202e-01 -4.28438276e-01 -4.29751098e-01
-7.94391215e-01 -1.10637292e-01 -3.41961309e-02 4.97550428e-01
4.37858433e-01 -1.40029323e+00 -7.33761549e-01 2.93835104e-01
2.47048095e-01 -1.83702007e-01 3.78269404e-01 1.41235220e+00
-6.52391255e-01 3.35872114e-01 -4.43371803e-01 -8.48988712e-01
-9.85549331e-01 3.06754827e-01 8.33900034e-01 -2.09616259e-01
-9.91493046e-01 3.16033393e-01 -4.18977320e-01 -1.02848865e-01
5.76992221e-02 -7.83050239e-01 -9.04244706e-02 -1.06524415e-02
6.02789223e-01 5.73291898e-01 4.79533464e-01 -3.21813405e-01
-6.43736422e-01 6.99545324e-01 4.62782294e-01 1.35202974e-01
1.16285121e+00 -1.35639355e-01 4.06208009e-01 5.02516985e-01
1.29798031e+00 -1.48750335e-01 -1.28858256e+00 1.46548957e-01
-1.54221728e-01 -3.12642545e-01 -3.99463661e-02 -1.15193915e+00
-1.24342513e+00 9.92071867e-01 1.28633833e+00 2.87340790e-01
1.46190047e+00 -6.05963409e-01 8.30061376e-01 1.84294686e-01
6.50344416e-02 -1.10918593e+00 -2.76738331e-02 -4.72091138e-03
5.71458399e-01 -1.09717095e+00 3.26845825e-01 5.47725111e-02
-5.61531782e-01 1.16833985e+00 3.03401440e-01 -1.56405926e-01
4.31930155e-01 2.02572629e-01 4.01846439e-01 -1.47504658e-01
-5.69514811e-01 6.47088736e-02 1.68059573e-01 8.15042675e-01
6.95574224e-01 8.30882695e-03 -6.19401217e-01 4.75868464e-01
-1.93173409e-01 7.93243110e-01 6.48710608e-01 8.00532699e-01
3.65854315e-02 -7.13277698e-01 -2.28182107e-01 3.15318942e-01
-8.16555142e-01 1.53114796e-01 -1.60588846e-01 1.01871312e+00
9.44702253e-02 9.14604127e-01 -1.12178892e-01 -4.26581770e-01
4.81181741e-01 3.25815260e-01 4.97892082e-01 -7.51945749e-02
-8.98085177e-01 1.30125791e-01 3.43109190e-01 -1.00649714e+00
-6.85529232e-01 -5.92333317e-01 -8.69982183e-01 -1.42260715e-01
1.60110265e-01 -2.43626818e-01 7.70026624e-01 8.37073207e-01
5.99100590e-01 7.40190685e-01 5.43840647e-01 -1.00571930e+00
-4.18065906e-01 -1.12283158e+00 -5.47782004e-01 5.63595772e-01
4.92116362e-01 -4.55722868e-01 -3.89485359e-01 1.44397125e-01] | [13.906702041625977, 2.9495062828063965] |
415b1386-6319-4c05-9220-0465c21f7a4f | bioem-gpu-accelerated-computing-of-bayesian | 1609.06634 | null | http://arxiv.org/abs/1609.06634v1 | http://arxiv.org/pdf/1609.06634v1.pdf | BioEM: GPU-accelerated computing of Bayesian inference of electron microscopy images | In cryo-electron microscopy (EM), molecular structures are determined from
large numbers of projection images of individual particles. To harness the full
power of this single-molecule information, we use the Bayesian inference of EM
(BioEM) formalism. By ranking structural models using posterior probabilities
calculated for individual images, BioEM in principle addresses the challenge of
working with highly dynamic or heterogeneous systems not easily handled in
traditional EM reconstruction. However, the calculation of these posteriors for
large numbers of particles and models is computationally demanding. Here we
present highly parallelized, GPU-accelerated computer software that performs
this task efficiently. Our flexible formulation employs CUDA, OpenMP, and MPI
parallelization combined with both CPU and GPU computing. The resulting BioEM
software scales nearly ideally both on pure CPU and on CPU+GPU architectures,
thus enabling Bayesian analysis of tens of thousands of images in a reasonable
time. The general mathematical framework and robust algorithms are not limited
to cryo-electron microscopy but can be generalized for electron tomography and
other imaging experiments. | ['David Rohr', 'Pilar Cossio', 'Volker Lindenstruth', 'Markus Rampp', 'Fabio Baruffa', 'Gerhard Hummer'] | 2016-09-21 | null | null | null | null | ['electron-tomography'] | ['medical'] | [ 2.99581856e-01 -5.26702285e-01 4.50986326e-01 -1.08681999e-01
-7.27125347e-01 -5.60492635e-01 7.03554451e-01 1.15249649e-01
-9.26249266e-01 9.71311390e-01 -2.46462598e-01 -4.83902901e-01
-5.74941970e-02 -5.75946212e-01 -3.80834997e-01 -1.08540368e+00
1.25250682e-01 1.28439867e+00 5.88822126e-01 4.09518331e-01
3.59764516e-01 8.31235170e-01 -1.47146523e+00 3.01652938e-01
3.11618775e-01 5.61713517e-01 7.59569943e-01 1.00841141e+00
1.47984654e-01 2.09596589e-01 -6.80013970e-02 -1.32615060e-01
-2.49216169e-01 -2.15931103e-01 -6.86836183e-01 -5.60189746e-02
5.04510924e-02 -3.83842528e-01 8.03093985e-02 7.96635747e-01
8.09178412e-01 1.77147433e-01 5.42734802e-01 -3.15779716e-01
9.44617763e-02 -2.82972127e-01 -4.03965682e-01 4.08612967e-01
5.14203846e-01 2.31186733e-01 6.21149063e-01 -9.88133729e-01
9.57319438e-01 1.15079999e+00 5.17095029e-01 2.37619653e-01
-1.90089881e+00 -2.14576647e-01 -3.50555152e-01 1.73398092e-01
-1.14869916e+00 -4.27718997e-01 1.54271394e-01 -6.49846375e-01
1.22633052e+00 2.16208041e-01 6.92661345e-01 9.47139084e-01
6.50304258e-01 1.87711135e-01 1.43848217e+00 -3.48222315e-01
4.33314621e-01 -8.00050795e-02 1.83590814e-01 4.97381061e-01
2.58347929e-01 1.62441265e-02 -4.67113137e-01 -8.98430586e-01
5.20097673e-01 2.26015002e-01 -7.88557902e-02 -3.31377238e-01
-1.40195251e+00 6.26773596e-01 -3.58891457e-01 -1.67778641e-01
-5.56978226e-01 -1.62744328e-01 4.37803954e-01 -2.35514164e-01
2.06147432e-01 1.93851918e-01 -5.62586963e-01 -3.23978335e-01
-1.13402593e+00 4.13450152e-01 7.72523761e-01 4.05763209e-01
7.14457452e-01 -4.30498362e-01 5.03887951e-01 4.85903412e-01
4.02213186e-01 4.12897468e-01 2.03298464e-01 -1.17673671e+00
-3.59661132e-01 -2.04044525e-02 3.97678137e-01 -5.27899563e-01
-5.01753569e-01 1.76830888e-01 -7.59189367e-01 4.82972533e-01
3.42033356e-01 5.56264594e-02 -7.24712908e-01 1.32674706e+00
8.65388453e-01 -7.85869081e-03 -3.12113017e-01 5.22023261e-01
5.73846459e-01 7.28418589e-01 1.64341405e-01 -6.64359927e-01
1.51973712e+00 -3.23258758e-01 -4.26570684e-01 1.82909116e-01
3.22788149e-01 -1.00606477e+00 3.75262231e-01 5.61445355e-01
-1.13691008e+00 -2.41180941e-01 -7.82736301e-01 -1.20553866e-01
-2.64200658e-01 -1.02159165e-01 6.24946177e-01 4.85231161e-01
-1.08509660e+00 1.01400137e+00 -1.53905714e+00 -4.89415288e-01
1.99549124e-01 6.69443369e-01 -4.48011041e-01 -9.86996368e-02
-4.57113206e-01 9.93103564e-01 5.37764728e-01 -1.23156801e-01
-7.15343654e-01 -6.42677963e-01 -3.76865029e-01 -2.71805912e-01
-1.41319200e-01 -1.12370872e+00 8.81720126e-01 -3.39538977e-02
-1.61153495e+00 9.18063045e-01 -6.70838296e-01 -1.40336812e-01
3.24466348e-01 2.53184408e-01 1.26909122e-01 4.90227461e-01
-1.90838695e-01 5.57791352e-01 4.75896239e-01 -1.05398333e+00
-1.23163439e-01 -6.15169048e-01 -5.55427253e-01 1.64167490e-02
3.52830678e-01 4.73757118e-01 -1.87744036e-01 2.45891601e-01
3.48456770e-01 -9.68022645e-01 -4.27887470e-01 -2.13801369e-01
-1.15910359e-01 1.87474921e-01 7.92899668e-01 -5.47792137e-01
5.81549227e-01 -1.68082750e+00 3.41938376e-01 1.25820056e-01
3.75115305e-01 1.88069567e-01 3.89119864e-01 7.91495621e-01
1.91419981e-02 -3.88166100e-01 -1.34759516e-01 -3.71371895e-01
-6.56463429e-02 3.21346164e-01 -1.04143783e-01 7.87504315e-01
-2.37866566e-01 4.65544343e-01 -7.40400672e-01 -6.45901144e-01
5.82544148e-01 8.05961668e-01 -6.22525036e-01 1.73798725e-01
-4.76524718e-02 9.70919430e-01 -4.14748251e-01 3.76981288e-01
9.32491899e-01 -7.26294100e-01 9.72525716e-01 -5.34714498e-02
-4.73815471e-01 3.94995391e-01 -1.11779308e+00 1.53500068e+00
-6.54218197e-02 2.26575643e-01 4.85345453e-01 -6.20769024e-01
5.77565551e-01 2.63148338e-01 6.00212753e-01 -1.01392746e-01
-6.39972985e-02 1.57584324e-01 1.12311557e-01 -1.43614426e-01
3.44304085e-01 -6.38163507e-01 4.93364841e-01 8.05189908e-01
1.79126382e-01 -2.66029686e-01 2.85937548e-01 2.22709075e-01
1.08997917e+00 4.37970340e-01 4.95446473e-01 -6.43737912e-01
2.89201647e-01 1.25380293e-01 3.88081402e-01 6.09693587e-01
-6.96059614e-02 3.86981696e-01 2.00987175e-01 -7.68731117e-01
-1.63329339e+00 -1.24767840e+00 -9.11669314e-01 1.00376523e+00
-8.59919414e-02 -7.73730218e-01 -6.95387363e-01 3.37415397e-01
-2.21477121e-01 1.05346084e-01 -1.62683830e-01 4.33716536e-01
-2.63152272e-01 -1.77534580e+00 1.13926902e-02 1.45881884e-02
-1.66997135e-01 -9.44977343e-01 -9.48883414e-01 5.82427621e-01
2.64515489e-01 -1.14614224e+00 2.35375658e-01 3.91959727e-01
-1.11094809e+00 -8.20939839e-01 -3.38110119e-01 -1.79763794e-01
5.25063634e-01 2.64714122e-01 1.04088414e+00 6.79963753e-02
-7.78162837e-01 2.31299609e-01 8.90747607e-02 -3.17498147e-02
-4.48183239e-01 -3.20815325e-01 4.84463364e-01 -3.49388450e-01
4.01758939e-01 -1.16368091e+00 -6.18116975e-01 2.40276501e-01
-7.63468802e-01 4.05819029e-01 3.56739253e-01 1.10592663e+00
1.05396700e+00 -1.02203943e-01 -5.96290678e-02 -8.59225392e-01
2.42816910e-01 -3.02043140e-01 -8.98015916e-01 9.16193500e-02
-4.48931962e-01 -1.91037372e-01 5.87328970e-01 -1.56668782e-01
-1.12251663e+00 1.98339716e-01 -5.13270855e-01 -5.35041615e-02
-4.36823934e-01 3.62206131e-01 1.18167163e-03 -1.33570582e-01
2.16255084e-01 4.28580195e-01 1.59051135e-01 -6.16735756e-01
-1.66396469e-01 4.23615009e-01 3.78981620e-01 -1.03106761e+00
3.93922478e-02 1.02244318e+00 4.58925307e-01 -1.15077770e+00
-4.00088698e-01 -4.59199101e-01 -9.60763752e-01 1.59827378e-02
9.22745109e-01 -6.52465522e-01 -1.30567479e+00 2.31707871e-01
-1.20669091e+00 -3.90273035e-02 2.83982724e-01 7.95780599e-01
-7.06974387e-01 8.32153022e-01 -9.27122831e-01 -9.00235951e-01
-2.15110555e-01 -1.58382833e+00 1.22796655e+00 -1.95335876e-03
-3.07365030e-01 -9.75926220e-01 5.76997995e-01 4.47912008e-01
1.84580281e-01 8.62725079e-02 8.97231042e-01 -2.01184139e-01
-6.75706863e-01 -2.03347921e-01 -2.16990769e-01 -2.98187882e-01
-3.41150641e-01 3.40036124e-01 -9.66402948e-01 -6.26875937e-01
3.29914719e-01 -3.68995398e-01 8.35312903e-01 5.87666810e-01
9.04691815e-01 2.14158848e-01 -6.58892214e-01 6.99752748e-01
1.50626552e+00 9.44231898e-02 8.10238898e-01 3.23296309e-01
5.28262019e-01 6.52797341e-01 1.89200059e-01 7.79909909e-01
1.13171041e-01 7.57173419e-01 2.55147427e-01 2.52013654e-01
4.38220024e-01 2.66475707e-01 2.22448170e-01 1.10732794e+00
-5.48069715e-01 1.68409571e-02 -1.00830626e+00 1.05680585e-01
-1.71724379e+00 -1.30213404e+00 -4.40372735e-01 2.35060525e+00
7.91902542e-01 -1.23835109e-01 7.96418712e-02 -2.91465342e-01
6.28257215e-01 -2.68894196e-01 -5.49277544e-01 -2.69923329e-01
-3.63706648e-02 4.81874019e-01 2.25905448e-01 5.44772506e-01
-8.24729502e-01 5.44174314e-01 7.74096203e+00 7.80776858e-01
-7.60915637e-01 3.92478466e-01 4.34892893e-01 -2.81322837e-01
4.31103483e-02 5.00788152e-01 -1.05887806e+00 6.15681946e-01
1.53500533e+00 1.25725791e-02 4.75976795e-01 6.11998320e-01
3.82830024e-01 -7.90497422e-01 -9.69030976e-01 1.01466596e+00
-3.86127234e-01 -1.55089498e+00 -5.51610813e-02 6.92311585e-01
4.59041536e-01 6.45633519e-01 -2.37789601e-01 -4.69498903e-01
5.29481232e-01 -8.70622277e-01 2.21941203e-01 4.32400227e-01
5.80705225e-01 -6.45747185e-01 6.38684213e-01 5.55933416e-01
-7.85322785e-01 4.87490356e-01 -8.82665575e-01 -2.37777784e-01
6.78075254e-01 8.78386915e-01 -6.55130088e-01 3.21912110e-01
8.01979840e-01 3.85153651e-01 -8.03898722e-02 7.03936398e-01
3.91892403e-01 2.71849841e-01 -7.88531780e-01 1.58356667e-01
-5.90229360e-03 -8.48616302e-01 3.91850621e-01 1.32201779e+00
2.72284210e-01 8.71726647e-02 2.26607025e-01 8.36410642e-01
4.39982265e-01 -2.90840626e-01 -3.00715834e-01 7.15718605e-04
3.49251270e-01 1.52649403e+00 -1.02962399e+00 -4.29562747e-01
-2.22844660e-01 6.95359826e-01 4.96966243e-01 -1.18206248e-01
-4.66726124e-01 1.14485897e-01 6.99517429e-01 1.23190336e-01
4.70056266e-01 -6.19098485e-01 1.24885030e-01 -1.12106943e+00
-4.15413201e-01 -6.64738774e-01 2.16836959e-01 -8.53526115e-01
-1.25215769e+00 4.83117104e-01 -2.63094045e-02 -4.85826284e-01
-2.05343291e-01 -1.03205037e+00 -5.14157534e-01 9.06908870e-01
-1.13846040e+00 -8.13097835e-01 2.13926703e-01 2.68764824e-01
-7.29874298e-02 2.74959326e-01 1.32050657e+00 -4.38259309e-03
-6.25333548e-01 -3.02456617e-01 9.33172286e-01 -5.59961140e-01
5.20661533e-01 -1.22153592e+00 2.65545875e-01 5.14645815e-01
-1.65835798e-01 1.13578546e+00 9.91475224e-01 -7.87883401e-01
-1.73369563e+00 -5.61300576e-01 6.32656097e-01 -7.19380796e-01
8.59774709e-01 -3.23863119e-01 -9.41504359e-01 5.47914445e-01
1.62321672e-01 1.34663463e-01 1.06061530e+00 1.15398630e-01
-1.26506105e-01 6.21657193e-01 -1.13918269e+00 2.80371070e-01
5.87065995e-01 -7.45947421e-01 -4.10901397e-01 6.64153099e-01
-5.83497286e-02 -3.11149657e-01 -1.34120142e+00 3.87981646e-02
8.50687325e-01 -1.28614402e+00 1.29961550e+00 -4.92492348e-01
4.36014794e-02 -5.07173240e-01 -3.45918238e-01 -7.20664144e-01
-3.50365251e-01 -8.16825688e-01 8.59193727e-02 6.03663206e-01
5.08697294e-02 -7.15218902e-01 6.48455501e-01 5.95459580e-01
-6.08771592e-02 -6.41775727e-01 -1.16505814e+00 -5.79993486e-01
4.18383069e-02 -2.88992226e-01 2.64097452e-01 5.92392385e-01
2.47906864e-01 3.13679576e-01 -2.68367440e-01 1.39124900e-01
1.19101000e+00 3.63893688e-01 6.64441228e-01 -1.41922355e+00
-7.69767702e-01 2.24912502e-02 -4.75646526e-01 -8.17651987e-01
8.16985965e-02 -7.02983201e-01 -4.60548326e-02 -1.08634579e+00
1.08953857e+00 6.19386844e-02 1.49236917e-01 -1.15357690e-01
1.24264983e-02 3.85326624e-01 -1.92730084e-01 5.92705727e-01
-7.95500398e-01 3.24590206e-01 1.03832161e+00 4.02946979e-01
3.50208610e-01 -4.69357699e-01 3.79747264e-02 8.98096919e-01
4.61147338e-01 -7.23465502e-01 1.57606244e-01 -2.11203158e-01
3.54259580e-01 1.13227397e-01 6.07746661e-01 -9.02033746e-01
4.17965114e-01 -7.04697818e-02 5.49304903e-01 -9.81936693e-01
7.76968360e-01 -4.49874580e-01 7.61072397e-01 4.64680195e-01
2.62603164e-01 1.59381658e-01 1.31331459e-01 6.12256467e-01
1.07758202e-01 -3.23391795e-01 1.26974678e+00 -7.03208506e-01
-1.53883085e-01 2.63888210e-01 -8.11759710e-01 -4.37549800e-01
7.66774237e-01 -2.86256075e-01 -2.75281578e-01 7.01479018e-02
-1.04537165e+00 -2.23930180e-01 1.14250433e+00 -8.24138284e-01
4.79907155e-01 -8.16324949e-01 -4.20209229e-01 -2.55450979e-02
-2.43159667e-01 -7.01415837e-02 5.99265516e-01 1.17075145e+00
-9.94515777e-01 5.50570428e-01 -4.90009129e-01 -9.82147932e-01
-1.36634970e+00 6.53454900e-01 1.49521649e-01 -4.43826914e-01
-7.48099267e-01 4.27623361e-01 3.05184126e-01 -4.85179543e-01
-5.47969759e-01 3.97907376e-01 1.73195511e-01 -3.03567708e-01
9.45578396e-01 4.36937422e-01 2.05359712e-01 -6.64320409e-01
-3.69932085e-01 5.89881957e-01 -4.59662318e-01 -1.97011694e-01
1.70578551e+00 -3.00598115e-01 -7.28188396e-01 3.82805675e-01
9.11409318e-01 -2.00446680e-01 -1.30138385e+00 1.62495315e-01
-1.37447909e-01 -2.29406103e-01 1.79043278e-01 -3.42838705e-01
-2.62622297e-01 1.02227843e+00 3.38912755e-01 -1.78769380e-01
5.85975885e-01 3.99950668e-02 6.65608168e-01 5.84265411e-01
7.74910152e-01 -9.28483784e-01 -4.65003908e-01 4.40443844e-01
1.62094593e-01 -9.74300504e-01 6.71480358e-01 -2.07561523e-01
3.78856622e-02 1.20145178e+00 6.26023440e-03 -8.98916274e-02
5.62461615e-01 6.73963726e-01 -5.11265695e-01 -4.69068289e-01
-1.14668679e+00 1.30158793e-02 -3.51888210e-01 5.56777298e-01
5.13416290e-01 2.84168404e-02 -2.16401398e-01 1.39754098e-02
1.12531818e-01 2.64022686e-02 5.85821629e-01 1.24291050e+00
-5.05174160e-01 -1.40421200e+00 -5.90322196e-01 4.64089811e-01
-6.79078937e-01 -1.95180625e-01 -7.01767802e-02 3.86304915e-01
-1.42904878e-01 6.26286924e-01 2.27850512e-01 -3.46508622e-02
-4.46065843e-01 1.98117778e-01 9.28444028e-01 -6.21773183e-01
-1.99962839e-01 3.35747838e-01 -3.60027291e-02 -4.15620863e-01
-6.32007122e-01 -1.04092383e+00 -1.10812831e+00 -8.31239700e-01
-3.75418991e-01 2.95132816e-01 9.38542366e-01 1.00333011e+00
5.97406805e-01 5.38097583e-02 5.91292456e-02 -1.62610161e+00
-3.68490189e-01 -8.29620004e-01 -9.34337854e-01 9.88307148e-02
-1.52078137e-01 -7.26502180e-01 -2.79932082e-01 4.59943675e-02] | [13.30691146850586, -3.0685536861419678] |
d3b917cc-9a6e-43d4-9cc8-fe72eec61bf8 | depth-guided-adaptive-meta-fusion-network-for | 2010.09982 | null | https://arxiv.org/abs/2010.09982v1 | https://arxiv.org/pdf/2010.09982v1.pdf | Depth Guided Adaptive Meta-Fusion Network for Few-shot Video Recognition | Humans can easily recognize actions with only a few examples given, while the existing video recognition models still heavily rely on the large-scale labeled data inputs. This observation has motivated an increasing interest in few-shot video action recognition, which aims at learning new actions with only very few labeled samples. In this paper, we propose a depth guided Adaptive Meta-Fusion Network for few-shot video recognition which is termed as AMeFu-Net. Concretely, we tackle the few-shot recognition problem from three aspects: firstly, we alleviate this extremely data-scarce problem by introducing depth information as a carrier of the scene, which will bring extra visual information to our model; secondly, we fuse the representation of original RGB clips with multiple non-strictly corresponding depth clips sampled by our temporal asynchronization augmentation mechanism, which synthesizes new instances at feature-level; thirdly, a novel Depth Guided Adaptive Instance Normalization (DGAdaIN) fusion module is proposed to fuse the two-stream modalities efficiently. Additionally, to better mimic the few-shot recognition process, our model is trained in the meta-learning way. Extensive experiments on several action recognition benchmarks demonstrate the effectiveness of our model. | ['Yu-Gang Jiang', 'Yanwei Fu', 'Junke Wang', 'Li Zhang', 'Yuqian Fu'] | 2020-10-20 | null | null | null | null | ['few-shot-action-recognition'] | ['computer-vision'] | [ 4.87188995e-01 -1.90533757e-01 -3.31820220e-01 -4.06277567e-01
-6.40588105e-01 6.89078718e-02 7.60638297e-01 -1.33792803e-01
-3.83433402e-01 4.14898396e-01 3.94414395e-01 3.95510972e-01
2.18742937e-01 -6.60768688e-01 -7.88675547e-01 -9.49564815e-01
3.91119063e-01 -1.25084355e-01 3.73125553e-01 -5.01175709e-02
8.14829767e-02 2.79866338e-01 -1.85321677e+00 5.54363728e-01
5.95222890e-01 1.22226274e+00 6.07400835e-02 5.06457984e-01
-1.43742010e-01 1.34666038e+00 -3.58911693e-01 -2.20621422e-01
3.38024169e-01 -7.38320470e-01 -4.89911556e-01 9.01781142e-01
5.06872952e-01 -7.35870600e-01 -7.19671428e-01 9.24129188e-01
2.98581541e-01 6.33159518e-01 3.21426332e-01 -1.40441513e+00
-3.13014448e-01 3.82929862e-01 -5.25454402e-01 3.70946050e-01
3.68516654e-01 5.47743499e-01 5.70466399e-01 -9.74115610e-01
6.49577737e-01 1.16636682e+00 1.82027116e-01 8.36567581e-01
-7.75000572e-01 -4.57494497e-01 5.42632043e-01 7.39404082e-01
-1.20954609e+00 -7.81972289e-01 1.04922700e+00 -2.35467464e-01
8.55862796e-01 -8.38689655e-02 9.59455431e-01 1.26407492e+00
-1.69196889e-01 1.17414784e+00 7.04072177e-01 -3.14635605e-01
3.93666893e-01 -2.97543436e-01 -1.41581371e-01 5.53350449e-01
-1.90506414e-01 6.64740056e-02 -8.17520082e-01 3.21141630e-01
8.56116414e-01 9.20866489e-01 -3.66167873e-01 -4.94345933e-01
-1.39089251e+00 4.18365538e-01 2.73265243e-01 4.53596085e-01
-5.11183321e-01 2.07431227e-01 6.27113819e-01 3.84896070e-01
4.80121583e-01 -1.32775649e-01 -3.60546440e-01 -3.52173686e-01
-7.56265581e-01 -1.22067258e-01 3.19085687e-01 9.58957732e-01
8.51257861e-01 3.18287581e-01 -3.34641933e-01 7.01032102e-01
2.15582512e-02 3.63271654e-01 8.17489624e-01 -1.08327949e+00
7.18081832e-01 7.03516006e-01 3.03946696e-02 -8.64977121e-01
-1.96598172e-01 7.07462057e-03 -1.00509238e+00 -7.81394988e-02
3.19892526e-01 9.19036046e-02 -9.18680429e-01 1.65084732e+00
4.30856228e-01 9.33980048e-01 1.82810679e-01 1.02763402e+00
7.72108614e-01 7.51395643e-01 3.78955761e-03 -6.53046548e-01
1.15984154e+00 -1.34096396e+00 -7.99979568e-01 -2.55091667e-01
7.18092799e-01 -2.13831812e-01 9.26874101e-01 3.70749533e-01
-1.01402128e+00 -8.04780722e-01 -1.03497386e+00 -1.40448026e-02
-2.16174051e-01 7.67302793e-03 5.67304432e-01 3.64327669e-01
-5.78517199e-01 5.49975097e-01 -1.02212095e+00 -2.66914338e-01
6.82217896e-01 9.52667817e-02 -5.47760367e-01 -5.75722992e-01
-8.58410299e-01 3.66313428e-01 5.30528903e-01 1.45912305e-01
-9.55327451e-01 -3.94078732e-01 -1.04562056e+00 -3.52180302e-02
9.91919816e-01 -5.99739134e-01 1.20313990e+00 -1.19158399e+00
-1.78012478e+00 4.41977143e-01 -3.04160327e-01 -3.30703825e-01
4.06297207e-01 -7.00164512e-02 -4.87175345e-01 5.12552440e-01
-9.85662714e-02 5.52139103e-01 1.16901541e+00 -9.10271645e-01
-9.42770660e-01 -5.92010140e-01 3.42666626e-01 4.02595609e-01
-5.89678407e-01 -1.94991142e-01 -5.64388931e-01 -8.14283788e-01
1.83531716e-01 -5.62427640e-01 -3.12256187e-01 7.38651603e-02
9.47261602e-02 -1.36703521e-01 7.41542041e-01 -3.31031054e-01
9.79928613e-01 -2.37997818e+00 4.50890571e-01 -2.63521463e-01
1.90625355e-01 2.89275587e-01 -2.24535495e-01 1.45701677e-01
-8.38056430e-02 -4.74062145e-01 -2.42810518e-01 -5.86996198e-01
-2.13532910e-01 3.99044394e-01 -3.98866892e-01 4.89472806e-01
2.31851846e-01 9.24744546e-01 -1.06695461e+00 -4.87380415e-01
6.67984962e-01 3.21669430e-01 -5.30970216e-01 5.08867085e-01
-2.69878119e-01 5.62932134e-01 -3.37834805e-01 8.75508428e-01
5.45564353e-01 -2.17379928e-01 -1.83218434e-01 -2.53440976e-01
-9.36485007e-02 -2.27897272e-01 -1.30773175e+00 2.31281710e+00
-2.80219615e-01 1.67507246e-01 -3.36303324e-01 -1.31281304e+00
6.68464601e-01 3.79962713e-01 7.92931855e-01 -8.21308017e-01
3.42658937e-01 4.71930243e-02 -3.01053941e-01 -6.25130415e-01
2.40532115e-01 -3.22444260e-01 5.22217900e-02 4.07739371e-01
3.57897818e-01 2.78477192e-01 3.11359853e-01 9.61497352e-02
1.16657984e+00 2.54090339e-01 3.44857186e-01 6.26099586e-01
7.59544492e-01 -2.61941403e-01 9.04265463e-01 4.77824748e-01
-5.69395483e-01 6.96712911e-01 2.39200532e-01 -5.27979314e-01
-7.41170347e-01 -6.22028828e-01 3.22518200e-01 1.11319482e+00
3.91547263e-01 -5.36633253e-01 -6.58374190e-01 -7.41779745e-01
-3.84735823e-01 3.00494879e-01 -7.34459937e-01 -4.38651890e-01
-5.44457316e-01 -4.00953531e-01 1.74716190e-01 9.01525199e-01
7.66617894e-01 -1.15335178e+00 -9.00324106e-01 2.98465878e-01
-2.13186756e-01 -1.40278649e+00 -5.18334866e-01 7.61063993e-02
-1.01947331e+00 -1.02810228e+00 -1.01746500e+00 -5.29563546e-01
5.98415077e-01 7.43923664e-01 4.64952737e-01 -6.46033734e-02
-2.69786954e-01 7.35266328e-01 -8.51868451e-01 -7.62890652e-02
8.25899467e-02 -3.34028482e-01 1.66113064e-01 7.33983278e-01
6.18062139e-01 -6.98351145e-01 -6.23908341e-01 1.88681260e-01
-1.14553404e+00 2.77386487e-01 6.69959366e-01 9.00021970e-01
6.38799727e-01 -1.40936807e-01 4.91011351e-01 -6.24953926e-01
-1.21442750e-01 -3.94348085e-01 -1.65557012e-01 2.59059280e-01
-9.18318257e-02 -1.04679152e-01 7.77846575e-01 -7.18384743e-01
-1.24136364e+00 3.88823926e-01 1.49867162e-01 -1.11433768e+00
-3.57423067e-01 2.84225643e-01 -5.67322016e-01 8.61925781e-02
3.20114642e-01 6.33232772e-01 -1.41858554e-03 -3.61180693e-01
6.39338195e-01 5.34298420e-01 6.70848370e-01 -2.85582751e-01
7.19233930e-01 7.83647060e-01 -9.50880721e-02 -7.78413951e-01
-8.64651680e-01 -6.61579430e-01 -9.66371298e-01 -5.03085792e-01
9.04486716e-01 -1.05115557e+00 -5.81475616e-01 8.10592175e-01
-9.79973853e-01 -2.60804176e-01 -6.63369358e-01 6.60161972e-01
-8.71796727e-01 5.46227813e-01 -4.25127059e-01 -8.09100211e-01
6.49007596e-03 -9.66439962e-01 1.14421093e+00 3.23927671e-01
3.08784634e-01 -6.65676236e-01 -4.42263531e-03 4.59776402e-01
4.60426733e-02 2.81337768e-01 3.23584735e-01 -4.89468455e-01
-7.36666501e-01 -1.40744239e-01 -1.96756542e-01 4.76155013e-01
2.95690417e-01 -3.50492388e-01 -1.02665746e+00 -2.44273558e-01
3.48785877e-01 -5.22787571e-01 1.05686522e+00 1.76140308e-01
1.22843528e+00 -2.01668873e-01 -7.60237277e-02 7.56202221e-01
1.23775077e+00 3.35364759e-01 5.97386122e-01 1.09271407e-01
1.02208364e+00 3.20504427e-01 8.95543396e-01 8.97115707e-01
2.91070491e-01 6.68026447e-01 4.71340954e-01 1.95238397e-01
-1.24788746e-01 -1.97537065e-01 5.00372946e-01 1.00058687e+00
-3.55589539e-01 -3.99553254e-02 -4.38283563e-01 3.00095767e-01
-2.22224712e+00 -1.33212602e+00 3.88434291e-01 1.98765457e+00
4.56587046e-01 -1.15887420e-02 2.28883088e-01 2.64483750e-01
6.23498976e-01 6.36352897e-01 -8.20382476e-01 2.08733901e-01
-1.15206741e-01 -2.18445212e-02 1.16651289e-01 -8.13138708e-02
-1.19376731e+00 8.01366031e-01 4.86326981e+00 7.83672333e-01
-1.11466086e+00 1.84563577e-01 4.11626965e-01 -3.92928749e-01
2.21204668e-01 -6.25328347e-02 -5.15731931e-01 5.77624619e-01
6.91147625e-01 -2.60658950e-01 4.42180127e-01 9.46475506e-01
1.45734280e-01 1.94029461e-04 -1.33994746e+00 1.44148397e+00
5.65275788e-01 -1.22267866e+00 2.58158654e-01 -7.91103318e-02
7.36237109e-01 -2.57736504e-01 -1.58002958e-01 4.84652996e-01
-2.03718230e-01 -5.00992954e-01 6.41649842e-01 9.11782861e-01
6.89413309e-01 -7.65760958e-01 5.55055022e-01 4.40754771e-01
-1.52128494e+00 -5.64391553e-01 -4.51029658e-01 -3.58051658e-01
4.14958209e-01 2.60700792e-01 -1.58905715e-01 7.35825658e-01
4.94848549e-01 1.45092940e+00 -5.40098608e-01 1.03592145e+00
-1.32793235e-03 2.32163936e-01 -2.60688495e-02 2.82062590e-01
2.49766827e-01 -6.92412034e-02 3.20766151e-01 7.17841148e-01
2.79724300e-01 6.93785012e-01 3.09848011e-01 4.18226093e-01
-3.74789126e-02 5.52435778e-02 -5.49491167e-01 -4.19576205e-02
2.12622046e-01 1.05420756e+00 -6.26686692e-01 -7.07416415e-01
-8.69764030e-01 1.39264929e+00 1.59862310e-01 3.05575073e-01
-8.20729613e-01 -1.67193159e-01 4.63612676e-01 -2.31331542e-01
4.61309195e-01 1.26192328e-02 2.11970001e-01 -1.86318457e+00
2.19762713e-01 -9.81964171e-01 4.77303058e-01 -7.95907378e-01
-1.13225603e+00 3.07133079e-01 -3.19720656e-02 -1.85462093e+00
-3.87245983e-01 -4.93897796e-01 -4.99823064e-01 1.80337802e-01
-1.43323624e+00 -1.17747521e+00 -7.26358950e-01 9.63241696e-01
1.09113264e+00 -2.78658181e-01 5.83419323e-01 5.90471089e-01
-7.74310172e-01 4.26488638e-01 -1.39108613e-01 1.05219968e-01
7.06810653e-01 -8.28754783e-01 1.67501435e-01 9.74956632e-01
2.71045655e-01 2.43393570e-01 2.38403395e-01 -4.18348730e-01
-1.69632387e+00 -1.24013412e+00 4.53985453e-01 -2.93605834e-01
4.65602368e-01 -6.55627847e-02 -1.01696908e+00 6.53237462e-01
-1.46571770e-01 5.44819951e-01 6.22170091e-01 -4.81195807e-01
-3.45254242e-01 -3.39744329e-01 -8.07761133e-01 5.18818974e-01
1.33829832e+00 -5.33886611e-01 -7.05160499e-01 2.62118280e-01
7.81083643e-01 -2.90244073e-01 -7.20696688e-01 4.71927911e-01
4.61616755e-01 -1.04242206e+00 8.58875275e-01 -6.90319359e-01
5.89109123e-01 -3.31741780e-01 -4.49884415e-01 -1.11896420e+00
-2.14437827e-01 -2.54752070e-01 -6.33010328e-01 1.23148572e+00
-3.54424208e-01 -3.58142436e-01 8.26891005e-01 6.12435758e-01
-1.91705927e-01 -7.44949162e-01 -1.03849113e+00 -7.91403353e-01
-4.43679959e-01 -5.99436879e-01 5.54466188e-01 9.28735137e-01
1.69459090e-01 1.43678054e-01 -8.36033940e-01 -1.26487389e-01
6.69036925e-01 1.12615094e-01 1.03336143e+00 -9.32048678e-01
-4.98245507e-01 -2.74375677e-01 -8.23740900e-01 -1.31927073e+00
1.33766636e-01 -5.78837991e-01 1.05094619e-01 -1.23038864e+00
2.73901820e-01 2.74302751e-01 -4.91146356e-01 5.14194131e-01
-1.30056068e-01 2.01536894e-01 3.90312940e-01 2.41597220e-01
-1.17847371e+00 1.16201246e+00 1.20975876e+00 -3.63473624e-01
-8.30553323e-02 -1.07866682e-01 -3.73917162e-01 8.12127471e-01
2.61074156e-01 -2.44169250e-01 -6.33618295e-01 -3.49431336e-01
-2.96815753e-01 3.45861763e-01 3.69834065e-01 -1.37108564e+00
4.70348060e-01 -4.83426839e-01 3.68413448e-01 -5.40629208e-01
6.05876148e-01 -1.04095793e+00 -1.74692109e-01 2.09545806e-01
-2.27598131e-01 -3.42531234e-01 -1.78752244e-01 9.29156065e-01
-4.54678684e-01 -4.67455536e-02 6.64430916e-01 -1.91179588e-01
-1.28776324e+00 7.54443049e-01 -2.13821590e-01 -1.61436960e-01
1.26643622e+00 -3.76041293e-01 -2.16448382e-01 -1.63779736e-01
-7.83472121e-01 2.11708143e-01 4.42979813e-01 5.90310991e-01
8.38304460e-01 -1.71358633e+00 -3.72454286e-01 3.44382167e-01
5.92618644e-01 1.20641440e-01 8.90722692e-01 9.81708407e-01
-2.38175821e-02 1.01980597e-01 -3.33077431e-01 -6.03064537e-01
-1.07597101e+00 9.27796543e-01 1.35271907e-01 1.26153782e-01
-9.28623259e-01 7.43628800e-01 2.03012615e-01 7.27873147e-02
4.33702230e-01 -3.46580803e-01 -2.60592848e-01 3.78845602e-01
1.07658803e+00 4.93484318e-01 -2.48869240e-01 -8.02509785e-01
-3.15078497e-01 5.73242247e-01 -8.69418457e-02 -6.93244636e-02
1.46210873e+00 -3.01842660e-01 2.02350557e-01 9.57962155e-01
1.17976475e+00 -6.42176569e-01 -1.77495384e+00 -5.99634111e-01
-4.31066185e-01 -8.77144456e-01 -1.34311706e-01 -2.03064308e-01
-1.28749692e+00 1.08536398e+00 5.24742007e-01 -1.72084332e-01
1.38984776e+00 -2.32824102e-01 9.94796693e-01 6.58077836e-01
4.81444806e-01 -1.19639552e+00 5.54327488e-01 4.24165815e-01
5.47952533e-01 -1.40869296e+00 -7.85611644e-02 -1.37572110e-01
-6.88046873e-01 1.17226493e+00 8.66968572e-01 -7.13981092e-02
5.28698444e-01 -1.51175126e-01 -8.49628150e-02 -1.97883267e-02
-9.10078585e-01 -3.35241854e-01 6.88752532e-03 5.45304060e-01
-1.45401061e-01 -2.86786199e-01 1.70091148e-02 8.06987762e-01
5.54391742e-01 5.19459546e-01 5.26530802e-01 1.12682414e+00
-4.77919489e-01 -8.68959069e-01 -1.16911538e-01 3.55772197e-01
4.61275652e-02 1.37840822e-01 -1.34073615e-01 4.26130086e-01
2.07852483e-01 7.90740371e-01 8.15060437e-02 -6.58999264e-01
4.54136312e-01 1.72599584e-01 5.54941535e-01 -7.21088409e-01
-1.14620999e-01 1.37839407e-01 -4.84896243e-01 -9.11902726e-01
-9.85744059e-01 -7.49265015e-01 -1.22243810e+00 1.32613722e-02
-1.83531374e-01 -1.39810026e-01 2.37678498e-01 1.32911003e+00
2.94983238e-01 6.47260368e-01 8.92687201e-01 -1.26495516e+00
-4.30280119e-01 -9.19217825e-01 -7.38821566e-01 6.07968748e-01
3.62905830e-01 -8.11433017e-01 -4.76906449e-01 4.13922459e-01] | [8.552201271057129, 0.7019497156143188] |
03db1423-a581-4091-a866-7d66c0fcef9a | a-dual-attention-learning-network-with-word | 2210.00220 | null | https://arxiv.org/abs/2210.00220v2 | https://arxiv.org/pdf/2210.00220v2.pdf | A Dual-Attention Learning Network with Word and Sentence Embedding for Medical Visual Question Answering | Research in medical visual question answering (MVQA) can contribute to the development of computeraided diagnosis. MVQA is a task that aims to predict accurate and convincing answers based on given medical images and associated natural language questions. This task requires extracting medical knowledge-rich feature content and making fine-grained understandings of them. Therefore, constructing an effective feature extraction and understanding scheme are keys to modeling. Existing MVQA question extraction schemes mainly focus on word information, ignoring medical information in the text. Meanwhile, some visual and textual feature understanding schemes cannot effectively capture the correlation between regions and keywords for reasonable visual reasoning. In this study, a dual-attention learning network with word and sentence embedding (WSDAN) is proposed. We design a module, transformer with sentence embedding (TSE), to extract a double embedding representation of questions containing keywords and medical information. A dualattention learning (DAL) module consisting of self-attention and guided attention is proposed to model intensive intramodal and intermodal interactions. With multiple DAL modules (DALs), learning visual and textual co-attention can increase the granularity of understanding and improve visual reasoning. Experimental results on the ImageCLEF 2019 VQA-MED (VQA-MED 2019) and VQA-RAD datasets demonstrate that our proposed method outperforms previous state-of-the-art methods. According to the ablation studies and Grad-CAM maps, WSDAN can extract rich textual information and has strong visual reasoning ability. | ['Hongfang Gong', 'Xiaofei Huang'] | 2022-10-01 | null | null | null | null | ['visual-reasoning', 'visual-reasoning'] | ['computer-vision', 'reasoning'] | [-2.10960526e-02 3.44210684e-01 -3.04194778e-01 -4.42166328e-01
-9.30255949e-01 -2.75590569e-01 4.09673542e-01 2.15384647e-01
-3.53920847e-01 3.12105656e-01 6.55307412e-01 -4.62103814e-01
-1.14001125e-01 -8.07027996e-01 -4.80735064e-01 -5.29701054e-01
3.35683465e-01 3.94473255e-01 1.67626873e-01 -3.33116025e-01
3.69822904e-02 2.11353377e-01 -1.06154168e+00 1.19729638e+00
1.14612591e+00 7.84439266e-01 6.49526179e-01 7.33537912e-01
-7.96756685e-01 1.36456561e+00 -4.70807284e-01 -5.60576975e-01
-3.03078353e-01 -8.95519793e-01 -1.18827045e+00 4.43057083e-02
2.74510473e-01 -3.93540144e-01 -5.57018995e-01 9.67440128e-01
5.11247754e-01 -2.76689976e-01 7.69926727e-01 -1.17498672e+00
-1.35253191e+00 3.96450222e-01 -6.36670589e-01 5.89859664e-01
3.75774831e-01 3.56549621e-01 9.24291193e-01 -9.18869734e-01
6.02735043e-01 1.36590755e+00 3.05864334e-01 7.11648643e-01
-7.17952967e-01 -3.79569441e-01 9.61054713e-02 8.45110714e-01
-1.03016078e+00 6.58851787e-02 8.59029055e-01 -3.76024783e-01
8.31240952e-01 4.48731333e-01 7.76551723e-01 9.12309289e-01
6.28125191e-01 1.34409964e+00 9.69765246e-01 -2.99098343e-01
-6.88000992e-02 3.76642883e-01 4.08428937e-01 1.14180732e+00
-5.96055016e-02 -1.95513427e-01 -3.75636607e-01 9.33964923e-02
6.11429155e-01 2.73742735e-01 -4.91009951e-01 -3.09742391e-01
-1.45406532e+00 1.11486030e+00 1.11285734e+00 5.76792359e-01
-5.46564281e-01 2.99229044e-02 4.47946906e-01 6.21152967e-02
1.33661866e-01 3.29488963e-01 -2.07080528e-01 5.56443393e-01
-7.10716307e-01 -1.79751933e-01 2.47742161e-01 5.69427788e-01
5.39492190e-01 -1.13785610e-01 -7.68666387e-01 4.87949252e-01
5.61064661e-01 8.27349782e-01 8.38268459e-01 -6.60804212e-01
4.71871793e-01 1.15877175e+00 -4.28970635e-01 -1.11532915e+00
-6.47605181e-01 -4.03283894e-01 -9.68524933e-01 3.47926132e-02
1.51965052e-01 1.55717298e-01 -1.31452811e+00 1.51376092e+00
3.48109871e-01 -2.39993557e-01 4.17942703e-01 1.06500840e+00
1.75181735e+00 6.41256452e-01 4.19290066e-01 -7.77528882e-02
2.00326967e+00 -1.15228474e+00 -1.26395762e+00 -3.67836893e-01
7.57765591e-01 -5.67630768e-01 1.35835135e+00 -6.34955615e-02
-9.78992224e-01 -6.62715733e-01 -9.91991520e-01 -6.66909039e-01
-5.97776294e-01 2.55786091e-01 6.35367930e-01 3.33886862e-01
-8.37458670e-01 -2.11892024e-01 -5.91905236e-01 -1.20026298e-01
8.43932688e-01 1.60118207e-01 -4.01728332e-01 -4.65187073e-01
-1.50611329e+00 1.15291584e+00 2.29919225e-01 2.38997385e-01
-9.78976071e-01 -8.55298460e-01 -1.00782204e+00 2.03911215e-01
3.21287870e-01 -1.26800060e+00 1.00775826e+00 -1.03402030e+00
-8.62339079e-01 9.28607523e-01 -2.11347565e-01 -2.21763358e-01
9.84821692e-02 1.33206761e-02 -4.94481325e-01 9.43919778e-01
2.64712095e-01 1.00937366e+00 7.65821159e-01 -1.28644776e+00
-3.38294089e-01 -6.35303676e-01 2.29272068e-01 4.03165102e-01
-3.96015853e-01 -4.11610186e-01 -4.96091694e-01 -5.64521432e-01
-1.24833453e-02 -2.83492833e-01 -1.41772225e-01 3.48831773e-01
-3.08048874e-01 -3.78425777e-01 6.78017437e-01 -1.12814057e+00
1.03409696e+00 -1.97998869e+00 2.61157244e-01 -4.68749888e-02
7.35770941e-01 2.77578890e-01 -3.36724699e-01 9.81516317e-02
-1.06309012e-01 -6.25339225e-02 -1.23468846e-01 2.57197350e-01
-1.62044317e-01 4.66167539e-01 -1.21843018e-01 2.66536266e-01
3.24075341e-01 1.71322930e+00 -9.46084738e-01 -1.12089550e+00
3.15315396e-01 4.99174684e-01 -5.24363577e-01 3.04074168e-01
-2.65058130e-01 2.42118806e-01 -7.10746586e-01 8.41032803e-01
5.46067834e-01 -8.78484130e-01 -5.81511930e-02 -8.96248817e-01
5.04921496e-01 -2.93591738e-01 -4.05491054e-01 1.79163909e+00
-5.80367565e-01 5.37938416e-01 -1.42118037e-01 -1.06324899e+00
5.89251399e-01 3.35869402e-01 3.29742134e-01 -1.26055658e+00
2.59939313e-01 -1.88373342e-01 2.72760075e-02 -1.18597138e+00
-1.01320021e-01 -2.31173143e-01 1.74460411e-01 1.75963670e-01
3.40995699e-01 6.49965778e-02 -1.36799430e-02 7.63977051e-01
9.09745395e-01 -4.04068649e-01 4.14052427e-01 -2.46141210e-01
9.01785433e-01 1.84650317e-01 2.36984864e-01 4.22912866e-01
-2.50895590e-01 3.64970624e-01 5.66011369e-01 -3.91154051e-01
-6.84485495e-01 -1.24001527e+00 3.80265266e-02 7.73349404e-01
4.93307441e-01 -3.19932967e-01 -6.67162836e-01 -1.02224767e+00
-6.16006739e-02 7.55342066e-01 -1.19645178e+00 -3.62022907e-01
-2.90767193e-01 -6.33096039e-01 2.62694031e-01 7.65396297e-01
5.86389124e-01 -1.50413501e+00 -6.37664795e-01 -1.67613849e-01
-5.83009958e-01 -9.32177126e-01 -6.38041556e-01 -2.95678467e-01
-7.12036312e-01 -1.46566188e+00 -9.74846125e-01 -1.05034840e+00
9.05828059e-01 3.11050177e-01 1.25138009e+00 2.16001779e-01
-6.85688794e-01 8.07692528e-01 -3.31058115e-01 -4.07175004e-01
-2.67507017e-01 -2.87790298e-01 -7.06860423e-01 -5.47846593e-02
4.91836101e-01 5.84668666e-02 -9.83159840e-01 -1.09003179e-01
-1.11110425e+00 5.21068633e-01 1.19340253e+00 1.18217218e+00
5.91381788e-01 -5.14117062e-01 5.76904833e-01 -9.05750751e-01
7.02592552e-01 -5.43706894e-01 1.08373687e-02 8.64670932e-01
-5.14881492e-01 2.51477599e-01 3.33774894e-01 -3.03331614e-01
-1.29640007e+00 -2.37829521e-01 -3.37765276e-01 -3.93383354e-01
-1.64080430e-02 4.68073040e-01 -2.12461770e-01 9.28481296e-02
4.95866418e-01 5.47625959e-01 1.80646092e-01 -4.47480381e-02
7.70441413e-01 4.91836756e-01 6.27258420e-01 -7.03337342e-02
3.89087856e-01 5.14976680e-01 -2.17957005e-01 -6.56720817e-01
-1.07199717e+00 -4.56312031e-01 -4.44324017e-01 -3.54509145e-01
1.58589268e+00 -8.55700016e-01 -9.94534373e-01 -1.32542402e-01
-1.32671297e+00 5.59148863e-02 -2.70571053e-01 3.93798351e-01
-3.62221062e-01 5.64120531e-01 -7.41745889e-01 -4.42354232e-01
-7.05179751e-01 -1.28045249e+00 1.05678618e+00 4.38289165e-01
1.42877102e-02 -1.15763474e+00 6.29889816e-02 9.39377129e-01
2.72729516e-01 1.40884802e-01 1.34160006e+00 -3.98436487e-01
-6.28903568e-01 1.10286556e-01 -7.26777017e-01 1.31165385e-02
1.35240108e-01 -5.91425419e-01 -8.27002585e-01 5.34428023e-02
6.55765310e-02 -4.87713605e-01 1.18585455e+00 5.79578578e-01
1.29234934e+00 -1.81018695e-01 -4.13543612e-01 4.92508560e-01
1.33237886e+00 2.42895424e-01 6.66122317e-01 6.90887049e-02
8.94221783e-01 6.33652866e-01 4.54005957e-01 -4.01409119e-02
9.36590016e-01 1.49142936e-01 7.23514557e-01 -7.78234065e-01
-5.11002183e-01 -1.50389239e-01 1.31115288e-01 9.31199849e-01
2.92286277e-01 -2.51522958e-02 -9.99606848e-01 7.53037393e-01
-1.59521425e+00 -8.58302951e-01 -2.40177020e-01 1.28423667e+00
8.49407434e-01 -2.18171775e-01 -3.34654301e-01 -7.33255073e-02
2.91253448e-01 2.67657042e-02 -6.62269115e-01 -2.50090003e-01
-9.81879756e-02 -6.19062036e-03 2.36456692e-02 4.91358399e-01
-7.80712366e-01 7.00847983e-01 5.38426447e+00 9.10567641e-01
-8.21163535e-01 4.24984246e-01 6.87113464e-01 2.94720262e-01
-9.50229943e-01 -5.64640343e-01 -4.21456695e-01 8.71191397e-02
5.21584034e-01 1.15115270e-02 -1.49418879e-02 6.07276380e-01
-9.87797752e-02 -7.96132460e-02 -8.87397230e-01 1.14671540e+00
6.21341527e-01 -1.71663058e+00 6.41158998e-01 -2.49343202e-01
4.07598495e-01 -3.69712830e-01 2.30986938e-01 4.26406771e-01
-2.87618255e-03 -1.26588416e+00 1.46919921e-01 9.80316579e-01
7.74851143e-01 -5.85162401e-01 1.11083245e+00 1.58009097e-01
-1.26135850e+00 -7.85761476e-02 -3.09423864e-01 6.33059800e-01
1.41233563e-01 2.36955091e-01 -7.43240237e-01 8.72080982e-01
7.72611380e-01 6.17102802e-01 -9.50999856e-01 7.10473895e-01
-3.52667361e-01 3.58344346e-01 4.10651743e-01 -2.21391872e-01
3.62435311e-01 2.77881444e-01 5.58324195e-02 1.15888274e+00
-1.34964123e-01 5.92016757e-01 -3.67327072e-02 1.03774881e+00
1.36826457e-02 4.61713552e-01 -2.97050983e-01 -1.49925858e-01
-2.54655123e-01 1.39038169e+00 -7.82227397e-01 -6.29890740e-01
-6.71076834e-01 1.01265407e+00 3.23438257e-01 4.49126899e-01
-8.18984091e-01 -2.53917605e-01 1.30178660e-01 -1.09865457e-01
1.19850464e-01 2.54633397e-01 -3.52776766e-01 -1.22432220e+00
-2.69419789e-01 -1.01900744e+00 8.51004958e-01 -1.18396902e+00
-1.40048397e+00 8.80522370e-01 -1.81592003e-01 -1.00954378e+00
1.88278973e-01 -8.33488941e-01 -5.01350939e-01 8.89657676e-01
-1.75845027e+00 -1.62959564e+00 -6.17425799e-01 1.08232141e+00
8.51582170e-01 -1.90793946e-01 8.84057999e-01 8.38162601e-02
-2.74837583e-01 5.57365954e-01 -4.33320284e-01 2.50146657e-01
6.05781078e-01 -1.33292484e+00 -4.11006123e-01 5.04171848e-01
3.56600463e-01 4.87332761e-01 3.38995308e-01 -5.21778286e-01
-1.42416441e+00 -8.95876706e-01 7.54780412e-01 -5.34165859e-01
3.26053172e-01 -4.10016850e-02 -1.23703909e+00 4.53814507e-01
7.32513905e-01 5.66935837e-02 9.17484581e-01 -2.45931327e-01
-2.96106637e-01 -5.29670902e-02 -1.00204790e+00 5.37286520e-01
5.16230762e-01 -7.17227459e-01 -1.17397237e+00 5.42849958e-01
1.01682818e+00 -2.77445912e-01 -9.03305590e-01 5.21131217e-01
4.00642186e-01 -6.93574727e-01 1.19690883e+00 -9.33626711e-01
6.43242598e-01 -2.80795544e-01 1.88013911e-02 -1.01426196e+00
-3.33589077e-01 2.94272959e-01 -2.34657750e-01 7.43426561e-01
3.78111660e-01 -1.34511471e-01 5.66125631e-01 1.94006890e-01
-9.14006606e-02 -1.03651810e+00 -5.61217785e-01 2.97934003e-02
9.48098078e-02 -1.98814169e-01 2.64528692e-01 1.12329078e+00
1.70631155e-01 7.51460314e-01 -2.32286826e-01 3.02958250e-01
4.62838799e-01 3.80576879e-01 2.98300505e-01 -8.14725935e-01
-1.85131252e-01 -3.86543125e-01 -2.28793353e-01 -8.56388211e-01
3.91773842e-02 -1.16685736e+00 -3.49020988e-01 -2.38553238e+00
7.17981279e-01 3.42201978e-01 -4.56889838e-01 6.24886930e-01
-5.64629078e-01 1.79616153e-01 1.52483895e-01 -1.78229749e-01
-8.36940706e-01 8.11684430e-01 1.97543871e+00 -6.43499017e-01
2.12181255e-01 -4.54427719e-01 -7.19017386e-01 6.22405171e-01
5.52886963e-01 -1.33200422e-01 -7.03216076e-01 -3.79002988e-01
2.19718426e-01 4.24361587e-01 6.30841732e-01 -5.51485896e-01
3.30609113e-01 -1.07351951e-02 8.72676790e-01 -1.03489017e+00
1.75541639e-01 -7.46091664e-01 -5.21600187e-01 9.05089200e-01
-4.95352566e-01 9.30715948e-02 3.79446983e-01 5.32820225e-01
-5.07171035e-01 4.83472422e-02 5.40493608e-01 -3.62882435e-01
-1.03033531e+00 3.48099023e-01 -2.93155640e-01 2.79473782e-01
1.01172566e+00 1.11787513e-01 -5.73134780e-01 -4.80018914e-01
-9.65555072e-01 8.38535190e-01 -3.11516136e-01 5.41183054e-01
1.39955044e+00 -1.47132778e+00 -8.72331858e-01 3.59034240e-02
4.30401325e-01 -3.08294624e-01 1.03078520e+00 1.03867233e+00
-7.22519934e-01 6.59604669e-01 -3.27189595e-01 -6.94001079e-01
-1.37545514e+00 9.58000660e-01 4.08955872e-01 -5.08483469e-01
-6.67031527e-01 8.26578140e-01 9.66038644e-01 -3.70550215e-01
3.27217802e-02 -3.81067216e-01 -8.32346022e-01 5.91220073e-02
8.11813891e-01 -7.06410408e-02 -2.30380148e-01 -5.79190671e-01
-4.38182294e-01 7.00236976e-01 -2.26203665e-01 2.16805935e-01
8.98355961e-01 -1.54846519e-01 -1.03465088e-01 1.80177689e-01
1.17462933e+00 -2.47312501e-01 -6.30710602e-01 -2.58472323e-01
-2.53116816e-01 -1.73268586e-01 1.83222115e-01 -1.04516971e+00
-1.24238157e+00 1.44324267e+00 9.98999596e-01 -1.05353512e-01
1.27664471e+00 5.32882512e-01 8.17648053e-01 1.66490659e-01
-3.46186459e-01 -6.43171191e-01 7.26530254e-01 -5.06619960e-02
1.17825723e+00 -1.50033462e+00 -1.26037627e-01 -3.96049947e-01
-1.06590199e+00 1.15994823e+00 9.53072369e-01 2.29178414e-01
4.98915344e-01 -1.22263111e-01 4.43089992e-01 -7.35591888e-01
-5.93585432e-01 -3.96493226e-01 1.06922102e+00 4.65633512e-01
2.77471393e-01 -1.26987308e-01 -2.95202821e-01 9.40009415e-01
2.24741071e-01 -2.26750627e-01 9.45080444e-02 5.40155113e-01
-5.47771752e-01 -6.38017654e-01 -3.41222942e-01 5.34142911e-01
-3.75431091e-01 -3.01467210e-01 -2.31072649e-01 8.60179722e-01
3.79108340e-01 8.40747178e-01 -1.00816093e-01 -2.43598327e-01
2.33916670e-01 2.53173523e-02 4.90080237e-01 -3.51522982e-01
-4.27328885e-01 -9.97405499e-02 -2.16704652e-01 -6.28539681e-01
-5.52610457e-01 -2.92220432e-02 -1.60802841e+00 2.09452450e-01
-1.66480586e-01 1.59511611e-01 3.05417269e-01 1.12434733e+00
1.72724634e-01 1.22645748e+00 2.25833610e-01 5.83213614e-03
-1.50794908e-01 -7.59742916e-01 -2.53705710e-01 5.23789525e-01
4.34698939e-01 -4.22500312e-01 -4.51929605e-04 1.09731227e-01] | [10.954687118530273, 1.697878360748291] |
803e383c-e834-4d2b-9021-3b27dd3c3ae0 | implicit-quantile-networks-for-distributional | 1806.06923 | null | http://arxiv.org/abs/1806.06923v1 | http://arxiv.org/pdf/1806.06923v1.pdf | Implicit Quantile Networks for Distributional Reinforcement Learning | In this work, we build on recent advances in distributional reinforcement
learning to give a generally applicable, flexible, and state-of-the-art
distributional variant of DQN. We achieve this by using quantile regression to
approximate the full quantile function for the state-action return
distribution. By reparameterizing a distribution over the sample space, this
yields an implicitly defined return distribution and gives rise to a large
class of risk-sensitive policies. We demonstrate improved performance on the 57
Atari 2600 games in the ALE, and use our algorithm's implicitly defined
distributions to study the effects of risk-sensitive policies in Atari games. | ['Rémi Munos', 'David Silver', 'Georg Ostrovski', 'Will Dabney'] | 2018-06-14 | implicit-quantile-networks-for-distributional-1 | https://icml.cc/Conferences/2018/Schedule?showEvent=2450 | http://proceedings.mlr.press/v80/dabney18a/dabney18a.pdf | icml-2018-7 | ['distributional-reinforcement-learning'] | ['methodology'] | [-4.89231020e-01 -2.21126571e-01 -3.86763871e-01 -3.18323880e-01
-1.26700878e+00 -5.84158003e-01 6.21949792e-01 -2.30348483e-01
-8.47597182e-01 1.36393678e+00 1.98093146e-01 -6.33391738e-01
-3.71258497e-01 -9.44179893e-01 -6.48671091e-01 -7.82318950e-01
-4.04539824e-01 9.55845654e-01 2.34934166e-01 -4.68203634e-01
2.90089101e-01 3.50318462e-01 -1.31189406e+00 -1.65769204e-01
8.77767503e-01 7.34740257e-01 -3.72832209e-01 9.29784834e-01
2.01642141e-01 8.30442965e-01 -1.04720080e+00 -4.03463602e-01
2.70868659e-01 -3.96034181e-01 -4.09700274e-01 -3.80638778e-01
8.76020491e-02 -8.60520065e-01 -4.23761189e-01 1.05724931e+00
5.53979218e-01 5.15781045e-01 1.12925935e+00 -1.21877611e+00
-5.04451871e-01 8.33328605e-01 -9.08102572e-01 4.93700892e-01
1.48617566e-01 1.69534460e-01 9.74362791e-01 -1.33566678e-01
3.23287666e-01 1.39208245e+00 3.47812772e-01 5.83937049e-01
-1.54865396e+00 -7.13792801e-01 -8.30562189e-02 -1.32729739e-01
-1.19607460e+00 -3.66534889e-02 3.51898849e-01 -2.64537066e-01
8.72959197e-01 -1.68374136e-01 6.35460973e-01 1.23991716e+00
5.47154427e-01 7.81462610e-01 1.27367091e+00 -3.91799212e-01
7.72837400e-01 -4.34672982e-01 -2.09512845e-01 1.78896800e-01
1.33412302e-01 9.56359029e-01 4.71940674e-02 -6.73523307e-01
9.25302148e-01 -7.17937946e-02 1.91742897e-01 -7.05604553e-01
-4.98627067e-01 1.31894004e+00 -6.39197091e-03 -3.26983750e-01
-4.44300920e-01 7.07606316e-01 4.96115565e-01 4.52753842e-01
5.36439180e-01 2.19910964e-01 -3.05233419e-01 -7.55344927e-01
-8.36837947e-01 1.01019764e+00 9.23865497e-01 7.32687354e-01
4.69841152e-01 5.44386506e-01 -5.47034860e-01 6.76247537e-01
1.46251038e-01 8.73582840e-01 1.70803189e-01 -1.40655422e+00
5.33365428e-01 -4.35161084e-01 6.96767986e-01 -4.59003448e-03
-1.70984700e-01 -2.16078594e-01 -1.90906465e-01 6.79369569e-01
7.40127325e-01 -6.60104036e-01 -9.37464654e-01 1.97540450e+00
1.04966819e-01 1.97623953e-01 1.92956492e-01 5.76854646e-01
-4.16619211e-01 5.76863706e-01 4.40429688e-01 -1.46295607e-01
5.19726872e-01 -2.69042581e-01 -3.34065139e-01 3.65327507e-01
3.67522210e-01 -4.29446846e-01 1.18641710e+00 5.95935583e-01
-1.02729309e+00 -1.46374837e-01 -8.65527451e-01 5.93563139e-01
6.34259507e-02 -5.27098358e-01 4.81502980e-01 9.04440224e-01
-1.03596485e+00 8.90145302e-01 -8.04565132e-01 6.62859827e-02
3.34734470e-01 2.33974278e-01 3.72521877e-01 6.24997579e-02
-1.34880006e+00 9.85342085e-01 3.97714764e-01 -6.17365420e-01
-1.61896622e+00 -7.06624925e-01 -6.91740870e-01 -9.69203040e-02
7.16028214e-01 -2.21638784e-01 1.89157712e+00 -4.56971705e-01
-1.90722847e+00 2.04715043e-01 5.91938555e-01 -8.54312539e-01
5.74325800e-01 -3.01484644e-01 -3.16774398e-01 1.07202873e-01
-1.97129771e-02 2.84481376e-01 8.08384657e-01 -9.04066443e-01
-7.38168538e-01 -1.56202123e-01 1.67262495e-01 3.19067419e-01
1.13137037e-01 -1.08482510e-01 3.59895706e-01 -5.94591737e-01
-1.01251304e+00 -8.80923629e-01 -2.72924334e-01 -6.29984379e-01
-3.27318162e-02 -4.13528383e-01 2.18043879e-01 -3.41713190e-01
1.22032070e+00 -2.04197049e+00 1.06819153e-01 3.63340914e-01
-1.05265461e-01 1.62199736e-02 -2.16149211e-01 6.16923332e-01
8.54828805e-02 2.00114157e-02 -7.45106637e-02 -1.32523581e-01
5.35898745e-01 6.16577923e-01 -6.93938494e-01 6.43373132e-01
-1.34809062e-01 5.58211148e-01 -9.14849937e-01 -1.75274894e-01
2.89795518e-01 -5.81927933e-02 -8.17638814e-01 3.10562968e-01
-6.07138455e-01 -2.14854162e-02 -5.78886092e-01 3.32579792e-01
4.24109936e-01 7.33860552e-01 5.60134165e-02 5.97225904e-01
-1.12182513e-01 4.27302392e-03 -7.47232914e-01 1.51046562e+00
-2.67966866e-01 -5.21978326e-02 -2.93285549e-02 -1.12899315e+00
8.99509907e-01 1.92870200e-02 6.23667538e-01 -6.71834052e-01
2.40647018e-01 1.90821946e-01 1.47165120e-01 -8.00958946e-02
6.34700477e-01 -7.61443496e-01 -6.07702434e-01 6.90922737e-01
1.39500082e-01 -4.47015285e-01 2.13762179e-01 1.10136934e-01
9.47680950e-01 7.38354743e-01 7.95075595e-02 -8.03984463e-01
-2.21079037e-01 -1.39442444e-01 3.39471579e-01 1.11794281e+00
-6.23993158e-01 1.67349204e-01 9.87777591e-01 -3.06459278e-01
-1.30180907e+00 -1.79994845e+00 -2.21505627e-01 1.35832286e+00
-2.62771279e-01 -1.47919357e-01 -7.86897659e-01 -7.25007117e-01
4.81570512e-01 1.36721873e+00 -8.96952212e-01 -4.27080154e-01
-3.09513956e-01 -6.60659909e-01 8.76965880e-01 6.79387867e-01
9.94644314e-02 -9.84080970e-01 -6.24079406e-01 3.68736982e-01
3.25338215e-01 -3.02035093e-01 -5.34179747e-01 5.04582942e-01
-5.48834085e-01 -8.72962952e-01 -8.13924730e-01 6.11285977e-02
-3.57410163e-01 -4.76122081e-01 1.39642704e+00 -7.44438350e-01
-6.88847387e-03 6.09851599e-01 -7.97122419e-02 -5.99137664e-01
-6.10078454e-01 -7.25181997e-02 2.67599165e-01 -6.31068766e-01
2.49276772e-01 -5.40843368e-01 -3.93117517e-01 1.02592476e-01
-9.05063927e-01 -9.40454066e-01 1.15450181e-01 9.33487654e-01
5.81325889e-01 3.77941243e-02 8.71275485e-01 -7.83215642e-01
1.14732993e+00 -6.39878035e-01 -1.01395345e+00 2.01623514e-01
-4.57557768e-01 5.29705763e-01 6.84382260e-01 -6.54977143e-01
-1.23532331e+00 -5.14145136e-01 -2.25855246e-01 -5.08385479e-01
-2.26770919e-02 2.82423764e-01 1.01934776e-01 4.54622418e-01
8.99606526e-01 -2.04490125e-02 2.75994122e-01 -1.81510836e-01
7.07302690e-01 5.28325737e-01 5.11478484e-01 -1.55754972e+00
5.42172074e-01 5.52329281e-03 -6.99374899e-02 -4.66405481e-01
-8.85652065e-01 -2.88531948e-02 1.86747327e-01 1.49697978e-02
6.71006322e-01 -8.06252658e-01 -9.28213418e-01 2.95074880e-01
-2.65618891e-01 -1.10263562e+00 -8.66518497e-01 4.78437424e-01
-1.58295119e+00 8.03731456e-02 -6.59999073e-01 -1.03852808e+00
-8.90905559e-02 -9.89249408e-01 8.09381366e-01 2.64617085e-01
6.47504926e-02 -1.03987253e+00 8.58303666e-01 -3.94054264e-01
5.44674039e-01 9.47507620e-02 9.73900735e-01 -6.33599639e-01
1.06487595e-01 3.83437514e-01 1.75458744e-01 4.58054543e-01
-7.38103837e-02 4.22618687e-02 -4.54077780e-01 -5.40511727e-01
-9.57817510e-02 -7.72515535e-01 8.81013691e-01 7.64658332e-01
1.15628159e+00 1.28880860e-02 2.33971998e-01 4.77292389e-01
1.42402709e+00 4.15994227e-01 8.10247481e-01 4.87429589e-01
2.08114330e-02 9.62693691e-02 9.23991382e-01 9.30181265e-01
2.57049680e-01 4.05170709e-01 4.40860957e-01 4.01073426e-01
5.77874720e-01 -5.90825140e-01 6.01063967e-01 -1.64543808e-01
-9.59592760e-02 -1.79116711e-01 -7.01558590e-01 5.04690647e-01
-1.79131091e+00 -1.24842501e+00 5.88017941e-01 2.43311644e+00
1.17899740e+00 2.88940996e-01 8.45412195e-01 -4.68876660e-01
6.09834373e-01 -1.78133056e-03 -8.25527668e-01 -8.18266034e-01
2.95152038e-01 8.53990078e-01 8.77433181e-01 6.65728271e-01
-9.54022169e-01 9.73092914e-01 8.12011909e+00 1.35113466e+00
-5.89164257e-01 -1.75239578e-01 6.50411308e-01 -3.39497536e-01
-4.18537199e-01 -1.47298232e-01 -7.39286244e-01 4.45454031e-01
1.38198471e+00 -5.86700141e-01 7.18385577e-01 1.01327777e+00
1.19826712e-01 -3.22223455e-01 -8.57932150e-01 4.89212155e-01
-4.02718931e-01 -9.24278080e-01 -4.02147621e-02 3.87549758e-01
7.18803346e-01 1.09214582e-01 4.23897564e-01 1.07980788e+00
1.39820111e+00 -1.16759741e+00 6.80250525e-01 4.79456753e-01
9.83200312e-01 -1.76824081e+00 5.98680198e-01 3.03415328e-01
-6.38250053e-01 -1.86088204e-01 -7.93024838e-01 -1.66464671e-01
-1.40497670e-01 1.66326150e-01 -5.13615489e-01 3.87158543e-01
5.66801667e-01 2.87882596e-01 9.74142104e-02 8.24709058e-01
-9.68067870e-02 9.75500405e-01 -4.80244905e-01 -1.22152358e-01
5.47591269e-01 -3.77406210e-01 4.00708646e-01 7.65273035e-01
2.44069263e-01 1.10038288e-01 5.55979013e-01 8.22451949e-01
1.43466383e-01 -2.47045904e-01 -7.16238379e-01 8.53685662e-02
3.97303730e-01 8.08142245e-01 -1.88086644e-01 -3.39617044e-01
-7.37121031e-02 3.61145318e-01 5.95408559e-01 6.04318321e-01
-1.20952702e+00 -4.09388274e-01 1.01371837e+00 -1.02061503e-01
5.25531650e-01 -2.50701845e-01 2.33828753e-01 -1.04354477e+00
-6.19484007e-01 -9.33001935e-01 6.83244109e-01 -3.87839735e-01
-1.64601994e+00 1.84880748e-01 6.52512670e-01 -7.27879703e-01
-8.91023397e-01 -6.39770269e-01 -5.92947721e-01 9.90533173e-01
-1.31116486e+00 -4.93410170e-01 7.38971353e-01 7.93250144e-01
2.23015517e-01 -4.31537896e-01 7.98174679e-01 -1.21005006e-01
-1.83408126e-01 6.24409378e-01 6.30116165e-01 -1.14497267e-01
8.78941834e-01 -1.80808139e+00 2.68555969e-01 3.39786679e-01
-3.53259221e-02 3.72038484e-01 9.90308523e-01 -6.74755394e-01
-1.06457412e+00 -7.92421699e-01 -4.78956074e-01 -4.99259055e-01
1.16417062e+00 -5.28818071e-02 -5.26696444e-01 7.27587700e-01
2.32120842e-01 -3.04855138e-01 6.78463101e-01 2.36046001e-01
-3.04076731e-01 -2.57264469e-02 -1.33340251e+00 6.41656578e-01
6.27127945e-01 -3.83162946e-01 -9.14396167e-01 -2.06908554e-01
5.17109573e-01 -4.09988105e-01 -8.73944402e-01 4.04279865e-02
4.07352358e-01 -9.75946844e-01 8.20182443e-01 -1.02951539e+00
2.91119725e-01 1.07389405e-01 -5.16661346e-01 -1.87757742e+00
-1.40767202e-01 -8.72604191e-01 -8.05208236e-02 9.16377544e-01
7.14354813e-02 -7.60620594e-01 7.37163782e-01 5.63497603e-01
-6.16022162e-02 -7.47830749e-01 -1.24448836e+00 -1.00421381e+00
1.13444769e+00 -4.37254548e-01 7.48291790e-01 2.48231396e-01
1.01590678e-01 -1.42084360e-01 -5.18419862e-01 -3.02840441e-01
1.09992516e+00 7.32141063e-02 5.04524171e-01 -7.20861614e-01
-7.20894277e-01 -3.81774843e-01 -3.52083296e-02 -9.62019145e-01
5.41758895e-01 -4.67471808e-01 1.59761861e-01 -9.03229177e-01
2.10252732e-01 -4.91176933e-01 -5.55613816e-01 2.82604426e-01
-2.88895853e-02 -1.22942058e-02 8.83348733e-02 -1.70955434e-01
-8.13155413e-01 1.01795423e+00 1.17246521e+00 1.75804913e-01
-1.31317645e-01 1.98126018e-01 -7.89024115e-01 4.79830742e-01
1.01771665e+00 -4.36882973e-01 -7.29972661e-01 6.56902120e-02
-6.25924021e-02 5.25381565e-01 9.97371897e-02 -6.35141551e-01
-4.52206701e-01 -8.10050309e-01 3.49847019e-01 -4.87762988e-01
1.30616739e-01 -3.14474702e-01 -2.15211958e-01 4.97358233e-01
-5.08645535e-01 1.51044160e-01 4.25853401e-01 6.95901811e-01
6.88503087e-02 -1.70529246e-01 9.08824980e-01 -8.27483454e-05
-4.33081150e-01 3.94899428e-01 -6.65239871e-01 7.88765609e-01
1.08355165e+00 3.05309594e-01 -2.22607613e-01 -7.48191535e-01
-6.90193892e-01 3.29961449e-01 5.61188102e-01 -4.36653681e-02
3.34197104e-01 -1.26708400e+00 -7.72476554e-01 -4.66982834e-02
-2.10617706e-01 -3.88118118e-01 1.78621907e-03 1.36712492e-01
-3.27161580e-01 -8.13023001e-02 -5.15528142e-01 -2.92481542e-01
-3.83921832e-01 5.53750575e-01 6.92301750e-01 -5.60146034e-01
-3.41273516e-01 3.65056425e-01 1.96147524e-02 -3.80405426e-01
-3.46513116e-03 -1.26763880e-01 1.76220328e-01 -1.77692413e-01
5.28649688e-01 4.55869526e-01 -6.08256638e-01 5.84060093e-03
-6.65083006e-02 1.83764145e-01 -9.56165045e-02 -7.51860082e-01
1.24896729e+00 1.23149544e-01 4.02722865e-01 4.18111175e-01
6.27933681e-01 -5.71735427e-02 -1.97394490e+00 1.51831552e-01
-1.65106341e-01 -4.77820754e-01 -8.10654014e-02 -8.08519483e-01
-7.54462659e-01 7.22492516e-01 4.43540573e-01 2.51688659e-01
7.23325193e-01 -1.69153944e-01 3.58510107e-01 1.13558114e-01
8.38576078e-01 -1.31222081e+00 5.31914383e-02 7.77715027e-01
5.29288232e-01 -6.68630898e-01 9.58438516e-02 4.46074605e-01
-9.35275435e-01 9.59760070e-01 5.36129713e-01 -8.18404496e-01
6.58262730e-01 6.89782441e-01 -9.80766118e-02 2.09968746e-01
-7.97347844e-01 -1.57613233e-01 -2.88733453e-01 1.09277630e+00
4.31181937e-02 5.36970139e-01 -2.41652429e-01 5.80246329e-01
-3.34118396e-01 -5.72852865e-02 7.84861445e-01 8.39098811e-01
-5.56671977e-01 -1.35591125e+00 -2.36874849e-01 5.20261824e-01
-6.99838221e-01 2.96085384e-02 2.54306436e-01 1.09187627e+00
-6.12484813e-01 7.87950993e-01 1.29949197e-01 -5.18398806e-02
3.24860156e-01 5.22211716e-02 8.97055209e-01 -5.47492564e-01
-2.08668262e-01 1.86305434e-01 1.50911197e-01 -5.66795766e-01
1.82808507e-02 -7.70825148e-01 -1.16681147e+00 -7.28663981e-01
2.72349566e-02 3.34347695e-01 2.57945001e-01 7.92264283e-01
-9.68047902e-02 3.05379301e-01 5.34198523e-01 -7.91583300e-01
-1.97288573e+00 -8.71004105e-01 -1.13461113e+00 3.83026779e-01
1.97296828e-01 -1.10310590e+00 -3.61470491e-01 -7.06663966e-01] | [4.080747604370117, 2.5523266792297363] |
6ac1e27c-341c-4577-a48c-4e6c67e549f8 | findings-of-the-third-shared-task-on | null | null | https://aclanthology.org/W18-6402 | https://aclanthology.org/W18-6402.pdf | Findings of the Third Shared Task on Multimodal Machine Translation | We present the results from the third shared task on multimodal machine translation. In this task a source sentence in English is supplemented by an image and participating systems are required to generate a translation for such a sentence into German, French or Czech. The image can be used in addition to (or instead of) the source sentence. This year the task was extended with a third target language (Czech) and a new test set. In addition, a variant of this task was introduced with its own test set where the source sentence is given in multiple languages: English, French and German, and participating systems are required to generate a translation in Czech. Seven teams submitted 45 different systems to the two variants of the task. Compared to last year, the performance of the multimodal submissions improved, but text-only systems remain competitive. | ['Chiraag Lala', 'Lo{\\"\\i}c Barrault', 'Lucia Specia', 'Stella Frank', 'Fethi Bougares', 'Desmond Elliott'] | 2018-10-01 | null | null | null | ws-2018-10 | ['multimodal-machine-translation'] | ['natural-language-processing'] | [ 4.64245617e-01 1.28008336e-01 2.65583992e-01 -3.90489012e-01
-1.49249172e+00 -1.10487592e+00 1.06321883e+00 -2.01088652e-01
-8.38226140e-01 1.18196285e+00 1.00229762e-01 -3.03110540e-01
8.64085317e-01 -2.52825975e-01 -6.54442608e-01 -4.81059402e-01
6.22676015e-01 9.48409498e-01 1.54206872e-01 -4.87751544e-01
-1.01678953e-01 1.23895817e-01 -9.21900332e-01 1.05597997e+00
5.48724651e-01 3.31429720e-01 5.32350004e-01 1.02547085e+00
-1.37926236e-01 1.62271783e-01 -6.95688486e-01 -1.01117516e+00
3.20901304e-01 -8.19037974e-01 -1.15308881e+00 3.50100607e-01
7.31767952e-01 -2.99587827e-02 8.62565488e-02 8.86240840e-01
8.13434780e-01 -1.42321274e-01 5.35770416e-01 -1.06724155e+00
-4.78663176e-01 5.37010372e-01 -3.69554698e-01 -2.83201337e-01
1.02843392e+00 2.47847334e-01 5.53100288e-01 -1.44518590e+00
1.11251771e+00 1.27012455e+00 2.25303695e-01 8.54258776e-01
-1.25728440e+00 -1.98449105e-01 -9.84056890e-02 -2.24026099e-01
-1.22136152e+00 -7.50490487e-01 1.07957467e-01 -2.96796083e-01
1.25776982e+00 3.30939174e-01 1.99641332e-01 1.55882490e+00
1.26836136e-01 6.81748569e-01 1.24036288e+00 -8.73483181e-01
-9.72392261e-02 9.05764401e-01 -4.31805789e-01 4.69940931e-01
-3.02151859e-01 -2.25840673e-01 -4.64564651e-01 -1.25840962e-01
1.22251056e-01 -7.50980377e-01 -3.65987629e-01 -4.32589278e-03
-1.88999653e+00 6.85432434e-01 2.64559668e-02 4.90895033e-01
-8.06349888e-02 -3.10607582e-01 6.03666067e-01 7.74813414e-01
3.70666862e-01 4.46050465e-01 -3.72486591e-01 -1.66261435e-01
-9.54776287e-01 3.20489436e-01 9.94293869e-01 9.84136403e-01
7.63034225e-01 -3.15756679e-01 -2.73851305e-01 9.19050753e-01
1.88609079e-01 8.40712965e-01 4.70163554e-01 -6.78845286e-01
1.19989610e+00 3.59464496e-01 4.12997156e-01 -3.01731497e-01
-3.03484410e-01 -4.89506591e-03 -6.25920773e-01 1.65814549e-01
5.41351616e-01 -6.51047707e-01 -8.78083408e-01 1.60429537e+00
5.43912202e-02 -6.49835348e-01 6.26600504e-01 1.12459028e+00
1.15699625e+00 9.74776208e-01 -2.89413512e-01 -2.90761590e-01
1.26489496e+00 -1.39300442e+00 -5.40511787e-01 -1.73791960e-01
6.71261668e-01 -1.56677759e+00 9.45783317e-01 3.00338328e-01
-1.51642787e+00 -6.46492541e-01 -7.47509658e-01 -1.11926720e-01
-5.08765459e-01 4.75430757e-01 -2.21639518e-02 4.12435621e-01
-1.71847844e+00 2.33565226e-01 -2.60877907e-01 -8.13460112e-01
-4.15095925e-01 4.94485706e-01 -9.56039608e-01 -1.40353158e-01
-1.37524021e+00 1.46993339e+00 3.09267551e-01 2.36181114e-02
-5.14431238e-01 2.63274647e-02 -1.05484748e+00 -5.28538585e-01
-1.29072756e-01 -1.12551582e+00 1.49995983e+00 -1.66853654e+00
-1.78813589e+00 1.59239244e+00 -4.32370961e-01 -2.29261532e-01
1.01401520e+00 1.44869357e-01 -3.63544673e-01 2.80503511e-01
4.87683326e-01 1.14175606e+00 7.39558518e-01 -1.19679904e+00
-6.50367200e-01 -1.21943973e-01 2.45849401e-01 6.75260186e-01
2.60797650e-01 6.48290277e-01 -7.30861604e-01 -2.46173516e-01
-3.41544539e-01 -1.20966756e+00 -4.30476032e-02 -6.80835128e-01
-5.28699696e-01 -5.61943799e-02 4.72524375e-01 -7.17023730e-01
6.58546209e-01 -1.94878471e+00 7.44130433e-01 -2.53112346e-01
-2.88753033e-01 3.69328866e-03 -6.58283710e-01 8.34403753e-01
-1.44705579e-01 3.62366289e-02 -5.56258440e-01 -7.89512336e-01
-4.04175371e-02 -7.72503912e-02 -3.21397126e-01 1.55118078e-01
4.66761321e-01 9.76705670e-01 -7.64205277e-01 -4.75582689e-01
1.75088253e-02 2.15351239e-01 2.16461182e-01 2.52417568e-02
-5.45475706e-02 6.74716771e-01 1.21874817e-01 3.48670483e-01
7.07154870e-01 2.19481811e-02 1.01947024e-01 6.23405762e-02
-2.86792785e-01 1.10043623e-01 -9.15952027e-01 1.99392760e+00
-5.68320453e-01 7.06307471e-01 4.32452679e-01 -4.68298614e-01
7.61708498e-01 9.38024879e-01 -1.81878954e-02 -4.96714145e-01
-1.64575338e-01 6.26972914e-01 1.25031903e-01 -4.09121990e-01
6.71261013e-01 -3.07006598e-01 -3.24846178e-01 6.59258187e-01
3.26459676e-01 -6.80981994e-01 7.11105764e-01 3.46627384e-01
7.02757657e-01 4.50728297e-01 -6.94181174e-02 9.25506838e-03
8.62036943e-01 3.31880867e-01 4.37900387e-02 5.25637925e-01
-6.63462579e-02 1.07191622e+00 3.67844254e-01 -3.41675043e-01
-1.28592443e+00 -9.80276048e-01 1.11924917e-01 9.91646767e-01
-9.04445425e-02 -4.76294786e-01 -9.12359834e-01 -9.32603478e-01
-5.81670284e-01 4.39168066e-01 -4.77849424e-01 3.44066143e-01
-6.56532705e-01 -4.19909775e-01 7.79043198e-01 6.14724718e-02
5.06628513e-01 -1.19346941e+00 -2.66154677e-01 6.22294284e-02
-9.21954513e-01 -1.39715683e+00 -7.25673735e-01 -2.17568502e-03
-5.73792756e-01 -7.78436661e-01 -1.44215655e+00 -1.19083107e+00
6.14883721e-01 4.40618070e-03 1.30419123e+00 -2.50300407e-01
2.36038536e-01 5.10977745e-01 -3.05043161e-01 -3.45420271e-01
-8.22873950e-01 2.55967259e-01 -1.10065565e-01 7.86869675e-02
2.96809115e-02 1.76268741e-01 5.48798684e-03 4.31589484e-01
-8.66358042e-01 3.99445683e-01 6.70014262e-01 9.09652054e-01
1.75672621e-01 -8.57659936e-01 5.08568287e-01 -5.66526234e-01
7.55446851e-01 -9.13811624e-02 -3.82589877e-01 3.28051656e-01
2.10474953e-01 -2.47349918e-01 4.96198297e-01 -3.51633698e-01
-9.86206770e-01 5.41505754e-01 -1.21259704e-01 -5.99704161e-02
-3.64041775e-01 5.10632992e-01 -2.46982187e-01 1.40343100e-01
6.35236204e-01 2.60983884e-01 8.26833323e-02 -1.31252259e-01
2.96857119e-01 8.36972535e-01 5.14851868e-01 -2.57441193e-01
7.89924681e-01 1.48573108e-02 -1.62848160e-01 -6.97934330e-01
-2.37955540e-01 -2.47051075e-01 -1.12574530e+00 -6.30160049e-02
1.03641760e+00 -1.17287457e+00 1.02237329e-01 5.15638053e-01
-1.92250705e+00 -3.68346006e-01 1.43753691e-03 6.31512821e-01
-6.56744778e-01 2.72945344e-01 -6.19734466e-01 -5.82225084e-01
-1.98374242e-01 -1.56254864e+00 1.40738833e+00 -8.82144719e-02
-3.02397490e-01 -1.14537799e+00 4.17681843e-01 5.66275597e-01
2.37579316e-01 9.81719941e-02 5.06905973e-01 -6.03594363e-01
-2.66489051e-02 -1.60993502e-01 -2.35063538e-01 3.07267845e-01
3.25192474e-02 7.97792524e-03 -7.66529083e-01 -3.69518906e-01
-2.17056274e-01 -6.64144337e-01 9.23468113e-01 -2.07364589e-01
-3.35590303e-01 8.11015368e-02 -1.53487861e-01 3.29956636e-02
1.05115509e+00 6.25463203e-02 6.45420611e-01 3.44481945e-01
2.07391471e-01 1.04890013e+00 4.35773253e-01 -2.26422071e-01
5.75621486e-01 1.08251309e+00 6.42866343e-02 -3.50275993e-01
-6.02902956e-02 1.77669391e-01 1.06384015e+00 9.70899343e-01
-1.94659144e-01 -4.77422416e-01 -8.74095559e-01 6.62425876e-01
-1.90697682e+00 -9.06760097e-01 -5.37288845e-01 2.28340149e+00
8.62465978e-01 -3.64927232e-01 3.16601574e-01 -3.50445360e-01
9.25874650e-01 -1.40824616e-01 6.86057983e-03 -8.79745305e-01
-6.75130010e-01 -6.21531904e-02 -1.39549021e-02 9.78406608e-01
-1.22494042e+00 1.14918840e+00 6.53399849e+00 4.38697904e-01
-1.29177451e+00 3.66659224e-01 4.74475145e-01 -6.51018694e-02
-2.72075385e-01 6.06565289e-02 -6.98876739e-01 2.91307032e-01
1.21068561e+00 -4.92579415e-02 2.95924842e-01 7.18343332e-02
1.66405976e-01 -3.82811427e-01 -1.29914296e+00 1.01830924e+00
5.15293002e-01 -9.62374687e-01 1.83399290e-01 -9.48838890e-02
9.57472563e-01 2.08173230e-01 1.21905012e-02 3.99176687e-01
-1.96114257e-02 -1.08648777e+00 6.16281748e-01 5.26366830e-01
1.11584699e+00 -6.01864278e-01 1.23234665e+00 4.92990941e-01
-9.03807282e-01 5.95958948e-01 -2.94679791e-01 -1.23441488e-01
4.50433075e-01 -7.64391795e-02 -1.01870644e+00 1.06467772e+00
4.38913256e-01 3.59937549e-01 -6.98355615e-01 6.70279205e-01
-3.48841727e-01 8.37393031e-02 -1.05936341e-01 7.75482599e-03
5.43752074e-01 -1.63946375e-01 7.35100389e-01 1.69420910e+00
4.24594671e-01 -4.68180567e-01 3.94669503e-01 5.53605497e-01
-2.51303613e-01 5.48260987e-01 -9.57890630e-01 3.91684175e-02
-3.07798296e-01 1.54186273e+00 -5.59247673e-01 -5.63001513e-01
-5.42865515e-01 1.84381938e+00 6.57868013e-02 6.12064242e-01
-6.44521713e-01 -5.50252318e-01 3.24047804e-02 -9.99310315e-02
-5.61997965e-02 -5.08623347e-02 1.30680799e-01 -1.34132135e+00
1.54200971e-01 -1.04505038e+00 6.41699210e-02 -1.17968011e+00
-1.08171666e+00 1.19479024e+00 -1.42404795e-01 -1.36441684e+00
-6.16824627e-01 -6.37509644e-01 -4.97304052e-01 1.63535917e+00
-1.13552487e+00 -1.35072601e+00 1.19498551e-01 7.98807025e-01
7.72613108e-01 -6.57023191e-01 1.19855142e+00 2.66680390e-01
-1.57839686e-01 4.00068581e-01 -1.55681312e-01 9.56834704e-02
1.66709542e+00 -1.34206450e+00 3.94636631e-01 6.82806611e-01
3.87448847e-01 3.66124004e-01 5.92701554e-01 -5.47802806e-01
-1.18178523e+00 -9.77654278e-01 1.73307288e+00 -7.51948774e-01
6.01579309e-01 -5.14182508e-01 -3.72052193e-01 7.09661841e-01
1.30793345e+00 -4.94165599e-01 4.95647997e-01 -4.30976540e-01
-5.14324531e-02 3.05864036e-01 -1.03411126e+00 4.43705887e-01
3.92411292e-01 -5.58669090e-01 -6.04506433e-01 6.53121412e-01
6.97709322e-01 -8.02980363e-01 -4.24933702e-01 1.85824469e-01
3.64031434e-01 -6.92826450e-01 3.90034914e-01 -5.64347684e-01
8.51912796e-01 -4.99495775e-01 -2.34801501e-01 -1.57173872e+00
2.23746791e-01 -8.04733455e-01 6.57444596e-01 9.70755994e-01
1.21752250e+00 -6.49395764e-01 1.86041951e-01 3.44959527e-01
-2.95463413e-01 -3.99358243e-01 -1.14043581e+00 -4.73793685e-01
4.50979233e-01 -1.68622002e-01 5.24121374e-02 6.70555472e-01
1.12468220e-01 1.05186939e+00 -4.45859224e-01 -1.93828478e-01
-2.06544306e-02 5.13406396e-02 1.14460981e+00 -7.81572759e-01
-1.51945606e-01 -3.83634686e-01 -1.96928054e-01 -9.34464693e-01
2.06217796e-01 -1.21165335e+00 5.50756156e-02 -1.93650639e+00
4.40722436e-01 3.55700821e-01 2.82340139e-01 4.14220393e-01
-1.07369050e-01 8.28163862e-01 7.63759553e-01 3.28504771e-01
-6.79114342e-01 1.82410166e-01 1.28687108e+00 -3.32438231e-01
-1.43192619e-01 1.30294293e-01 -4.27443862e-01 4.88694519e-01
5.90863526e-01 -7.16054142e-02 -4.06595916e-02 -6.49610579e-01
4.42881078e-01 3.38797688e-01 3.97858858e-01 -4.81797755e-01
2.03599542e-01 2.84735978e-01 4.35458928e-01 -6.60412610e-01
4.75587636e-01 -5.52025616e-01 1.40383080e-01 1.95206389e-01
-2.42451966e-01 7.18521416e-01 4.80819792e-01 -1.10382333e-01
-5.91196477e-01 -3.42636973e-01 6.83275819e-01 -3.68971407e-01
-3.55136782e-01 -3.49888384e-01 -6.20141327e-01 -2.16623902e-01
1.00785565e+00 -2.68693209e-01 -3.79464895e-01 -6.85742021e-01
-1.29865873e+00 2.82645524e-01 6.69939756e-01 5.55882752e-01
6.78851545e-01 -1.46586573e+00 -1.46556044e+00 -2.19592348e-01
1.55497447e-01 -4.57881868e-01 9.42785293e-02 1.40747488e+00
-5.20417631e-01 6.51193559e-01 -1.65512115e-01 -7.37273991e-01
-1.58222544e+00 3.08930010e-01 2.33912751e-01 -5.15515804e-01
-1.16135418e-01 6.36332214e-01 2.08996207e-01 -1.18352461e+00
-1.67088553e-01 3.20475250e-01 -1.29179224e-01 2.21121266e-01
5.95845282e-01 -1.26462489e-01 2.51278549e-01 -1.20891905e+00
-4.61750746e-01 6.60498798e-01 1.03927568e-01 -1.16705680e+00
8.33485544e-01 -4.44561213e-01 -5.03421903e-01 8.53671193e-01
1.11869419e+00 4.93282050e-01 -5.55728197e-01 -7.91396201e-02
-1.89128488e-01 7.53387883e-02 -7.27699101e-01 -1.47616374e+00
-5.92814565e-01 9.75215554e-01 5.58397412e-01 -1.90216392e-01
9.26371157e-01 1.65436387e-01 4.65487361e-01 3.92159581e-01
5.05656958e-01 -1.03387225e+00 -1.41884193e-01 1.02871943e+00
1.41534233e+00 -1.61772943e+00 -4.72761631e-01 -7.43935928e-02
-1.23127091e+00 1.52124953e+00 4.18486804e-01 2.63330996e-01
-1.97517350e-01 9.13665891e-02 7.24291503e-01 2.76115894e-01
-8.48153591e-01 -2.11368605e-01 5.68945229e-01 6.96092248e-01
9.76465940e-01 3.91927687e-03 -4.50961500e-01 3.36311907e-01
-3.17360789e-01 -4.53388691e-01 7.59199321e-01 7.16827571e-01
-6.73587024e-02 -1.63859046e+00 -6.48708165e-01 -2.14052811e-01
-4.49010640e-01 -4.06062007e-01 -1.28970683e+00 8.39805305e-01
1.81033790e-01 1.31687212e+00 -1.73998550e-01 -2.93470770e-01
4.21236157e-01 3.97989571e-01 6.67925954e-01 -9.60783899e-01
-1.14852273e+00 3.87682617e-01 5.34711897e-01 -2.58837610e-01
-6.38902247e-01 -7.00629592e-01 -9.45160091e-01 6.01281151e-02
-3.42761315e-02 4.50795293e-01 9.15483117e-01 9.41576600e-01
2.17729434e-01 8.14043283e-02 5.71482837e-01 -1.11473739e+00
-1.98213845e-01 -1.38840342e+00 -2.82007400e-02 2.11454570e-01
2.58458018e-01 3.96741182e-02 -3.14648896e-01 3.37821513e-01] | [11.495931625366211, 1.5251071453094482] |
ab52dfbc-3d61-4119-9350-59459c0e87d9 | hyperlink-induced-pre-training-for-passage-1 | 2203.06942 | null | https://arxiv.org/abs/2203.06942v2 | https://arxiv.org/pdf/2203.06942v2.pdf | Hyperlink-induced Pre-training for Passage Retrieval in Open-domain Question Answering | To alleviate the data scarcity problem in training question answering systems, recent works propose additional intermediate pre-training for dense passage retrieval (DPR). However, there still remains a large discrepancy between the provided upstream signals and the downstream question-passage relevance, which leads to less improvement. To bridge this gap, we propose the HyperLink-induced Pre-training (HLP), a method to pre-train the dense retriever with the text relevance induced by hyperlink-based topology within Web documents. We demonstrate that the hyperlink-based structures of dual-link and co-mention can provide effective relevance signals for large-scale pre-training that better facilitate downstream passage retrieval. We investigate the effectiveness of our approach across a wide range of open-domain QA datasets under zero-shot, few-shot, multi-hop, and out-of-domain scenarios. The experiments show our HLP outperforms the BM25 by up to 7 points as well as other pre-training methods by more than 10 points in terms of top-20 retrieval accuracy under the zero-shot scenario. Furthermore, HLP significantly outperforms other pre-training methods under the other scenarios. | ['Lei Chen', 'Qun Liu', 'Xin Jiang', 'Fan Yu', 'Zhao Cao', 'Hao Jiang', 'Xinyu Zhang', 'Enrui Hu', 'Ke Zhan', 'Lan Luo', 'Lifeng Shang', 'Xiaoguang Li', 'Jiawei Zhou'] | 2022-03-14 | null | https://aclanthology.org/2022.acl-long.493 | https://aclanthology.org/2022.acl-long.493.pdf | acl-2022-5 | ['passage-retrieval'] | ['natural-language-processing'] | [-1.44091919e-01 -9.98344272e-02 -3.01171035e-01 -4.52974401e-02
-1.66011596e+00 -6.41363561e-01 6.96966529e-01 4.82081920e-01
-4.23366010e-01 7.68039286e-01 6.50810122e-01 -4.09760386e-01
-7.29528487e-01 -9.33339715e-01 -6.85924411e-01 -2.49178365e-01
-3.09493691e-02 6.66273594e-01 9.52007711e-01 -9.57884073e-01
3.53205711e-01 -1.28270179e-01 -1.36749554e+00 5.22919953e-01
1.33976114e+00 6.06329560e-01 3.27221662e-01 6.42077863e-01
-4.04690474e-01 4.27466422e-01 -6.06711388e-01 -4.15895134e-01
1.20857447e-01 -3.14279586e-01 -1.18919289e+00 -5.63389659e-01
4.32689279e-01 -3.51284355e-01 -6.34462535e-01 7.56880105e-01
8.84229958e-01 5.62027752e-01 5.60883880e-01 -6.28522575e-01
-9.95933115e-01 6.92458451e-01 -3.71770233e-01 6.64891720e-01
7.93875396e-01 -1.35556012e-01 1.59704125e+00 -9.15484309e-01
7.71688700e-01 1.36774921e+00 4.69402701e-01 1.83916390e-01
-8.84579957e-01 -1.73697039e-01 -1.20636066e-02 6.11885965e-01
-1.25620151e+00 -1.69217840e-01 3.75847578e-01 8.49959999e-03
1.06853533e+00 3.12845647e-01 2.24680945e-01 9.80654061e-01
-1.43142179e-01 8.53740513e-01 6.21011913e-01 -7.36559331e-01
1.40619382e-01 -1.38505608e-01 6.27923369e-01 3.60238522e-01
1.26757175e-01 -2.30482563e-01 -4.32865053e-01 -5.44175625e-01
4.61992174e-01 -1.21387511e-01 -5.10028422e-01 2.07035422e-01
-8.95299613e-01 8.47151935e-01 6.17151320e-01 3.74583304e-01
-2.63918579e-01 -3.42965811e-01 3.45441729e-01 6.89232111e-01
3.68133277e-01 8.73024702e-01 -4.82488841e-01 7.77570456e-02
-7.73395717e-01 2.92898148e-01 9.40458357e-01 9.83795643e-01
7.44342327e-01 -7.13185668e-01 -9.62589204e-01 1.36039007e+00
3.36413413e-01 4.58991706e-01 6.13022149e-01 -7.29371130e-01
9.33006883e-01 4.88319606e-01 2.60459840e-01 -8.43434453e-01
-2.26679612e-02 -5.64833045e-01 -4.38102335e-01 -9.12125409e-01
2.89076477e-01 4.61668074e-02 -7.65898883e-01 1.44915795e+00
4.85887051e-01 9.96531770e-02 1.56568587e-01 1.11719525e+00
1.01389670e+00 1.05148435e+00 7.86867663e-02 -1.77504599e-01
1.45355487e+00 -1.45478332e+00 -6.92742288e-01 -2.99017113e-02
9.01364982e-01 -9.73963737e-01 1.54873681e+00 -8.96133259e-02
-9.53182995e-01 -4.37532157e-01 -8.32163930e-01 -4.21661496e-01
-3.28024060e-01 -2.89708287e-01 1.84488788e-01 1.62097052e-01
-9.29433584e-01 5.21328628e-01 -1.02113143e-01 -5.57739496e-01
-1.39815673e-01 -8.05184469e-02 -4.11894871e-03 -7.32971132e-01
-2.02491832e+00 8.66485178e-01 1.90025583e-01 -3.04019541e-01
-7.66143918e-01 -1.08252728e+00 -4.31180030e-01 4.20886874e-01
5.09886980e-01 -9.81831849e-01 1.21270049e+00 -1.91020131e-01
-1.23048198e+00 4.00971085e-01 -1.18427031e-01 -2.34615535e-01
1.57406226e-01 -4.59155411e-01 -4.02940840e-01 5.41439891e-01
6.67330846e-02 5.15442371e-01 4.10211056e-01 -1.19012094e+00
-5.40606022e-01 -2.48937860e-01 6.35941088e-01 7.60375142e-01
-4.70754147e-01 -6.37346655e-02 -1.02480531e+00 -4.65485096e-01
-9.04457793e-02 -5.51747262e-01 -1.02039322e-01 -4.38824147e-01
-2.42323756e-01 -7.87391007e-01 5.30065596e-01 -6.73823535e-01
1.61656857e+00 -1.76507866e+00 -1.11567900e-01 1.57648847e-01
-5.86229004e-02 5.31613350e-01 -8.45532417e-01 9.49547112e-01
4.37480718e-01 1.08612075e-01 1.92358911e-01 -3.02582104e-02
7.19364509e-02 2.61931270e-01 -5.19419193e-01 -1.66833609e-01
-7.89681301e-02 1.08325100e+00 -1.32745028e+00 -8.87048542e-01
-2.47020617e-01 2.16050267e-01 -5.12920439e-01 4.54148054e-01
-3.82245541e-01 4.63017039e-02 -7.90278554e-01 6.74675465e-01
2.61944741e-01 -5.19006252e-01 -6.79018795e-02 1.29727721e-02
3.97944301e-01 7.63562799e-01 -6.95262015e-01 1.95882165e+00
-5.01496434e-01 3.82014036e-01 -1.86131626e-01 -6.85729086e-01
7.40355015e-01 3.46204013e-01 2.88864821e-01 -1.27404547e+00
-1.90631747e-01 2.11298823e-01 -1.73303604e-01 -7.91756868e-01
9.78452206e-01 1.42350346e-01 1.20775484e-01 4.87968683e-01
3.82898636e-02 6.37224764e-02 5.40507376e-01 7.32097685e-01
1.52771080e+00 -2.66658485e-01 -6.20572381e-02 -2.01125965e-01
4.47570294e-01 -4.34147567e-02 1.55592352e-01 1.10996902e+00
-1.10181995e-01 6.63484633e-01 1.93726927e-01 4.05494630e-01
-9.30143476e-01 -1.05670023e+00 -8.79377350e-02 1.58567810e+00
5.15229702e-01 -4.57293093e-01 -5.48578680e-01 -6.93879604e-01
5.70462048e-02 6.42608047e-01 -2.50855565e-01 -3.90514463e-01
-4.87638175e-01 -5.89991748e-01 5.68614781e-01 2.58082986e-01
3.18122357e-01 -8.36905241e-01 1.71684161e-01 2.07764462e-01
-8.89707267e-01 -1.12092769e+00 -6.08348966e-01 -2.41898641e-01
-9.46355939e-01 -8.77894819e-01 -1.16054010e+00 -7.47395456e-01
4.65625048e-01 7.45554209e-01 1.35783494e+00 5.09110987e-01
-1.43994465e-01 5.62536597e-01 -1.11787307e+00 1.74358934e-01
-8.27874430e-03 4.86426532e-01 -3.54357541e-01 -5.96022129e-01
4.45810080e-01 -3.94268930e-01 -1.02374554e+00 5.71676493e-01
-1.09790063e+00 -4.75685954e-01 6.14414394e-01 9.60820854e-01
4.70391154e-01 -3.20156366e-01 1.16092312e+00 -5.83798170e-01
1.34518671e+00 -9.14653718e-01 -1.51574969e-01 1.00089288e+00
-7.72782683e-01 1.73033267e-01 4.05946821e-01 -4.23221141e-01
-1.14314210e+00 -9.81674790e-01 -3.14474523e-01 -1.44627392e-01
3.15980673e-01 8.75919402e-01 1.52044296e-01 1.26701847e-01
9.95012999e-01 -6.07731715e-02 -5.62065780e-01 -7.55168736e-01
7.35450506e-01 8.19397449e-01 1.17663212e-01 -8.96116853e-01
6.45067275e-01 8.03143233e-02 -4.15395051e-01 -6.80551827e-01
-1.14310861e+00 -1.23920226e+00 -3.28899384e-01 -2.78732833e-03
4.36311454e-01 -7.90318727e-01 -1.15134031e-01 -1.49594411e-01
-1.05727041e+00 -1.58879742e-01 -3.26031774e-01 3.74285758e-01
-3.13519947e-02 9.22140002e-01 -8.96567881e-01 -5.45004904e-01
-9.38475013e-01 -7.58803427e-01 1.08683336e+00 2.26618141e-01
1.23959437e-01 -1.01585102e+00 3.26386482e-01 7.41263032e-01
4.65397745e-01 -5.83297729e-01 1.18836927e+00 -9.75456357e-01
-6.69413269e-01 -3.07417065e-01 -4.48466808e-01 6.69396743e-02
-7.53831938e-02 -3.21431905e-01 -6.18891299e-01 -3.73247772e-01
-3.82299483e-01 -6.24280572e-01 9.39404190e-01 1.10115461e-01
7.78795004e-01 -2.81703293e-01 -2.69791245e-01 -9.46072564e-02
1.36520267e+00 -2.51440287e-01 8.32146704e-01 5.01856327e-01
2.12764889e-01 6.47044182e-01 1.06508577e+00 3.18277359e-01
5.53261042e-01 7.21533835e-01 2.00010940e-01 9.07778665e-02
-3.23148310e-01 -5.18972754e-01 1.16561860e-01 1.30389798e+00
2.22797722e-01 -5.35233796e-01 -7.16395736e-01 8.80221725e-01
-1.84916008e+00 -7.86522388e-01 -1.36028349e-01 2.29175162e+00
1.23515916e+00 -4.58455421e-02 -4.33323197e-02 -1.27315074e-01
6.87309921e-01 4.70460951e-02 -2.52829432e-01 1.87116284e-02
-4.84452508e-02 2.38743261e-01 1.75156906e-01 6.60154164e-01
-5.48345745e-01 9.58172441e-01 6.45900679e+00 1.29325163e+00
-4.47211057e-01 2.26358026e-01 2.11218849e-01 1.35538995e-01
-6.66338265e-01 1.26010761e-01 -1.04089880e+00 3.66662115e-01
9.61359262e-01 -4.81210023e-01 2.82981277e-01 5.96270919e-01
2.77612861e-02 -1.91054612e-01 -8.43845844e-01 4.84702289e-01
1.46549866e-01 -1.11540329e+00 4.28398579e-01 -2.96733856e-01
8.77133846e-01 2.07948118e-01 -1.47265002e-01 8.22381318e-01
4.20326740e-01 -5.58983505e-01 -3.62765579e-03 3.80524606e-01
5.50771534e-01 -4.04563963e-01 8.83108318e-01 5.98676264e-01
-1.01561773e+00 -8.89726058e-02 -8.34313452e-01 2.31445178e-01
3.85756612e-01 7.90759146e-01 -8.52671802e-01 1.08272851e+00
5.34678221e-01 2.16778338e-01 -5.25947392e-01 1.37877369e+00
-4.03927624e-01 8.61280620e-01 -4.85594630e-01 -1.99419320e-01
4.48474944e-01 -3.83775681e-02 5.50034523e-01 1.21880150e+00
4.36301917e-01 4.03206944e-01 3.68037894e-02 4.06239271e-01
-3.42696071e-01 4.58567411e-01 -2.06824899e-01 7.14951828e-02
8.53458643e-01 1.07148504e+00 -2.07874939e-01 -5.70404470e-01
-3.32161099e-01 7.24436939e-01 4.46506172e-01 6.91586137e-01
-6.07444704e-01 -7.49618709e-01 2.42980123e-01 1.07075728e-01
2.67109722e-01 -2.28450634e-02 2.16794580e-01 -1.22235739e+00
1.80454031e-01 -8.34820330e-01 8.70714664e-01 -8.64903986e-01
-1.58411229e+00 3.78680289e-01 1.21262208e-01 -1.08376205e+00
-1.96776196e-01 -1.32659152e-01 -5.05880535e-01 9.65895951e-01
-2.19027925e+00 -8.13453674e-01 -2.28344068e-01 5.98756492e-01
7.12708116e-01 1.86040878e-01 7.16255605e-01 7.39111543e-01
-2.31416762e-01 7.73767650e-01 5.50659001e-01 9.32975323e-04
1.13171470e+00 -9.92527306e-01 1.50356054e-01 5.99859178e-01
1.73336640e-01 9.83229399e-01 5.72553694e-01 -6.39018178e-01
-1.39642751e+00 -9.33299541e-01 1.00917220e+00 -4.54151541e-01
6.75699294e-01 1.95433378e-01 -1.28710473e+00 2.22266629e-01
4.62904662e-01 -2.25390539e-01 7.82696009e-01 5.51360071e-01
-5.37798524e-01 -2.00377062e-01 -8.95562649e-01 5.46761096e-01
9.11769032e-01 -6.19756162e-01 -1.31747925e+00 6.44120157e-01
1.27201986e+00 -2.20265374e-01 -1.13424110e+00 5.35496354e-01
1.79023713e-01 -4.98673558e-01 1.35011816e+00 -7.14344978e-01
6.19165063e-01 -9.02508050e-02 -1.95417047e-01 -1.36651909e+00
-3.03274184e-01 -3.84016961e-01 -3.56245250e-01 1.29158866e+00
6.73095405e-01 -4.56153423e-01 3.74202996e-01 3.10074121e-01
-2.54140943e-01 -8.31367075e-01 -8.35792959e-01 -7.95711815e-01
3.12497854e-01 -1.98154692e-02 5.32330573e-01 8.62767518e-01
4.70157564e-01 8.08876574e-01 -6.08595982e-02 -3.13334055e-02
2.69502223e-01 1.97271183e-01 5.26732981e-01 -9.36804175e-01
-4.84566897e-01 -2.23159134e-01 2.39783689e-01 -1.92004538e+00
-1.13727622e-01 -8.98842573e-01 1.87547803e-01 -2.12267876e+00
2.91541278e-01 -6.41885817e-01 -5.21383882e-01 1.91873431e-01
-7.94109464e-01 -6.80851415e-02 -9.55578610e-02 5.28169870e-01
-1.24267280e+00 7.60062754e-01 1.55886400e+00 -6.97822049e-02
-6.24212623e-02 -1.50804356e-01 -6.41228974e-01 8.17094892e-02
4.14878994e-01 -4.34423774e-01 -7.66603887e-01 -7.61256695e-01
4.31227982e-01 2.87687927e-01 -9.11336318e-02 -6.11784637e-01
4.49048638e-01 9.16042328e-02 -1.94366351e-01 -6.73689485e-01
2.77747601e-01 -5.83883047e-01 -6.34495199e-01 2.18125191e-02
-7.79960513e-01 -1.42407790e-01 8.68080929e-03 8.09917927e-01
-4.00178164e-01 -6.26928926e-01 3.14399958e-01 -1.41130701e-01
-5.92044532e-01 2.08784074e-01 1.42259635e-02 7.13990152e-01
4.24866706e-01 1.21080056e-01 -9.16986108e-01 -5.67123413e-01
-2.47422770e-01 8.63200545e-01 2.51576379e-02 5.71578205e-01
5.52746058e-01 -1.30089474e+00 -6.84300423e-01 -3.94088596e-01
3.92956764e-01 -7.99366832e-02 5.33119440e-01 8.44166338e-01
-1.69513583e-01 8.43226790e-01 2.56248713e-01 -4.03876215e-01
-1.15422761e+00 4.38793242e-01 -1.03747897e-01 -9.45412099e-01
-6.00823402e-01 8.70821834e-01 -2.18022354e-02 -4.62443680e-01
3.20298910e-01 -5.68419732e-02 -4.94827241e-01 7.38096908e-02
6.94472075e-01 4.62527245e-01 2.68999130e-01 -1.59236208e-01
2.25302517e-01 4.74494010e-01 -3.38585645e-01 -2.71118551e-01
9.42904592e-01 -4.91509616e-01 -5.53920232e-02 1.32199705e-01
1.32712519e+00 -1.90749019e-01 -6.79161966e-01 -7.80998170e-01
1.81080550e-01 -6.20210052e-01 -9.27076340e-02 -8.63065183e-01
-4.48877543e-01 9.39514816e-01 1.83058307e-01 3.63223761e-01
9.86805439e-01 1.00268669e-01 1.34876573e+00 8.58953595e-01
4.57434505e-01 -1.30684292e+00 4.93292570e-01 8.89199257e-01
8.46757472e-01 -1.04118359e+00 -5.57020605e-02 -3.30536783e-01
-4.36036706e-01 6.95955396e-01 6.03751719e-01 1.30248427e-01
4.61755067e-01 -5.56143045e-01 -3.62503268e-02 -2.83046275e-01
-1.01397848e+00 -5.75099349e-01 6.65354431e-01 3.02947670e-01
3.97053123e-01 -3.23101878e-01 -8.44324946e-01 4.40660506e-01
9.01351566e-04 -7.74621814e-02 8.98325592e-02 9.26329374e-01
-8.51463556e-01 -1.11147738e+00 -2.49586940e-01 5.25676310e-01
-4.56330121e-01 -4.22984809e-01 -2.70170063e-01 5.86198330e-01
-5.76127648e-01 1.22819567e+00 -2.62498945e-01 -2.48027489e-01
4.48508948e-01 1.83949798e-01 3.30450028e-01 -6.21452630e-01
-5.66128731e-01 -3.13009997e-03 3.17782909e-01 -3.07449430e-01
-1.86709911e-01 4.48855050e-02 -1.10365903e+00 -1.27955407e-01
-9.90200222e-01 7.52715528e-01 2.44687662e-01 9.74076867e-01
6.41749620e-01 4.28297967e-01 6.26627803e-01 -1.88771129e-01
-1.08818984e+00 -1.40870917e+00 -3.23757648e-01 7.16950953e-01
1.36042142e-03 -5.27487099e-01 -5.25546432e-01 -3.38536561e-01] | [11.418475151062012, 7.7341837882995605] |
c13f0efe-db65-439e-8ee7-039d4fea86d7 | task-aware-monocular-depth-estimation-for-3d | 1909.07701 | null | https://arxiv.org/abs/1909.07701v2 | https://arxiv.org/pdf/1909.07701v2.pdf | Task-Aware Monocular Depth Estimation for 3D Object Detection | Monocular depth estimation enables 3D perception from a single 2D image, thus attracting much research attention for years. Almost all methods treat foreground and background regions ("things and stuff") in an image equally. However, not all pixels are equal. Depth of foreground objects plays a crucial role in 3D object recognition and localization. To date how to boost the depth prediction accuracy of foreground objects is rarely discussed. In this paper, we first analyse the data distributions and interaction of foreground and background, then propose the foreground-background separated monocular depth estimation (ForeSeE) method, to estimate the foreground depth and background depth using separate optimization objectives and depth decoders. Our method significantly improves the depth estimation performance on foreground objects. Applying ForeSeE to 3D object detection, we achieve 7.5 AP gains and set new state-of-the-art results among other monocular methods. Code will be available at: https://github.com/WXinlong/ForeSeE. | ['Lei LI', 'Wei Yin', 'Chunhua Shen', 'Yuning Jiang', 'Xinlong Wang', 'Tao Kong'] | 2019-09-17 | null | null | null | null | ['3d-object-recognition'] | ['computer-vision'] | [ 4.67364609e-01 -1.18734725e-01 -2.26249143e-01 -3.13030005e-01
-4.92627114e-01 -5.46473086e-01 5.73283970e-01 -4.25607026e-01
-1.69707060e-01 5.24304628e-01 5.53807616e-02 -3.63929093e-01
6.88708663e-01 -6.49648845e-01 -6.31312370e-01 -9.64301884e-01
2.36491904e-01 1.99241340e-01 9.52454269e-01 5.86298525e-01
4.47335005e-01 6.52585268e-01 -1.58378828e+00 5.88664412e-01
6.26618922e-01 1.12501860e+00 7.27439761e-01 1.02771831e+00
-3.09636474e-01 1.11786425e+00 -5.53046942e-01 -1.62102893e-01
4.13194269e-01 -4.54553187e-01 -5.48740506e-01 3.83308291e-01
7.43094563e-01 -9.60999727e-01 -5.87448835e-01 1.01790667e+00
5.94631672e-01 -4.11310643e-01 5.19022405e-01 -1.13111472e+00
-3.96218807e-01 1.22791551e-01 -1.00754344e+00 5.81619740e-01
3.39160293e-01 1.02462508e-01 3.83656800e-01 -1.10593855e+00
4.95227545e-01 1.30355060e+00 6.60917684e-02 7.88479984e-01
-9.71854150e-01 -6.79122686e-01 5.69740713e-01 3.44296217e-01
-1.23173821e+00 -5.01366079e-01 6.63279176e-01 -4.48999316e-01
8.93138289e-01 2.01692790e-01 7.69007385e-01 8.66994321e-01
2.64445990e-01 1.38651776e+00 1.07021451e+00 -3.67381364e-01
1.01886719e-01 4.14584801e-02 -7.71672092e-03 4.84966934e-01
5.00452876e-01 1.22355059e-01 -6.93884969e-01 4.43207681e-01
9.09672558e-01 -6.50846884e-02 -4.34912860e-01 -3.78381729e-01
-1.19694304e+00 3.01299065e-01 1.42497629e-01 1.22547699e-02
5.30808270e-02 4.65583235e-01 -8.76957104e-02 -2.16105014e-01
7.38270223e-01 -2.02269629e-01 -4.03504372e-01 -1.69463873e-01
-7.63562441e-01 8.62879679e-02 3.98919404e-01 1.19410598e+00
6.14366055e-01 -6.67084828e-02 1.87503565e-02 5.66443622e-01
4.07623708e-01 9.15432513e-01 -8.36677700e-02 -1.25436151e+00
3.87233526e-01 7.14088917e-01 1.73528716e-01 -6.53735042e-01
-2.46764138e-01 -1.72461942e-01 -5.11021793e-01 4.79166985e-01
7.08231390e-01 -6.98099136e-02 -9.12495375e-01 1.09232700e+00
7.37482131e-01 2.78500766e-01 -2.43091747e-01 1.18398702e+00
1.05895805e+00 8.24003518e-01 -4.18718964e-01 -2.66529359e-02
1.20395947e+00 -9.15666997e-01 -5.49992979e-01 -8.19317043e-01
3.01400900e-01 -8.69613111e-01 5.55779874e-01 6.60923481e-01
-1.35182583e+00 -4.69266266e-01 -8.68506312e-01 -4.62339222e-01
-2.08681114e-02 1.56020284e-01 6.90339446e-01 1.01864314e+00
-9.50509131e-01 8.75436813e-02 -9.67829108e-01 -6.38606548e-02
6.40989482e-01 3.48798901e-01 1.40247587e-02 -4.76010799e-01
-6.42990172e-01 6.48034692e-01 2.96610773e-01 -1.58486713e-04
-1.10311210e+00 -6.31020129e-01 -7.01281726e-01 -4.64179993e-01
6.19437099e-01 -7.09757090e-01 1.40724194e+00 -7.12996244e-01
-1.44083071e+00 1.40874505e+00 -6.55093849e-01 -2.51937360e-01
5.68367422e-01 -4.15002674e-01 3.85791436e-02 3.61051112e-01
2.09117718e-02 8.74527395e-01 6.84272587e-01 -1.51386034e+00
-1.27575707e+00 -5.43128788e-01 2.07184535e-02 3.17370087e-01
3.41254473e-01 9.63254422e-02 -1.04406130e+00 -1.21935442e-01
6.16919756e-01 -6.27542138e-01 1.03194803e-01 5.43337286e-01
-4.09965038e-01 1.96936935e-01 7.75609910e-01 -5.49631834e-01
8.91837895e-01 -2.12918234e+00 4.15318832e-02 -4.55045968e-01
5.27772725e-01 2.45250612e-02 2.47817785e-01 -1.18525490e-01
3.59271169e-01 -2.99543977e-01 -8.50573778e-02 -3.43579262e-01
-1.60125881e-01 2.02245727e-01 -1.00855760e-01 8.26045215e-01
1.49616107e-01 8.74345720e-01 -9.07766342e-01 -5.84425330e-01
6.94790602e-01 5.55043757e-01 -5.94570935e-01 1.37310356e-01
-3.30383152e-01 4.75490212e-01 -2.74130225e-01 1.25711071e+00
1.31345630e+00 -3.67760867e-01 1.01121850e-01 -1.25859007e-01
-4.11297500e-01 1.90307438e-01 -1.24614131e+00 1.25192785e+00
6.88941032e-03 1.08302832e+00 2.20299989e-01 -3.97807151e-01
7.09608853e-01 -1.02760613e-01 4.54992682e-01 -8.26697290e-01
2.55337089e-01 2.68354207e-01 -6.54617473e-02 -2.32176885e-01
3.73925149e-01 7.13368505e-02 3.02044123e-01 4.60489799e-04
-2.88879812e-01 -4.11573976e-01 -4.58535086e-03 1.41378269e-01
9.84033644e-01 2.86734432e-01 2.05776677e-01 -4.57536755e-03
4.26323086e-01 -1.00514241e-01 5.82528591e-01 6.91290140e-01
-4.83224630e-01 8.71108174e-01 5.35411119e-01 -1.56303689e-01
-8.27333272e-01 -1.37951374e+00 -3.00580293e-01 6.96446717e-01
8.90192389e-01 1.71962276e-01 -5.95079839e-01 -3.23186368e-01
2.03225851e-01 5.40587425e-01 -5.23187399e-01 2.33407512e-01
-5.70016384e-01 -6.49172723e-01 1.51627883e-01 4.84939784e-01
6.88812137e-01 -5.83997428e-01 -9.40548182e-01 -7.48783723e-02
-3.19239050e-01 -1.60705101e+00 -3.30664277e-01 4.03055966e-01
-1.00999475e+00 -9.16808367e-01 -9.46430922e-01 -5.50604224e-01
4.93376076e-01 1.04019701e+00 1.18421412e+00 -2.21311674e-01
-4.45007980e-01 3.39706600e-01 -2.05276743e-01 -5.06931186e-01
-1.00350507e-01 -3.02394301e-01 -2.63275564e-01 -2.15418241e-03
5.77118874e-01 -3.50509793e-01 -1.02759075e+00 5.90406179e-01
-5.93572199e-01 6.15763068e-01 5.76882720e-01 2.68830359e-01
5.92178762e-01 -1.64340168e-01 -7.39840865e-02 -7.78171122e-01
-6.05859220e-01 -2.57636338e-01 -1.15047598e+00 -3.31658334e-01
-2.37857759e-01 -3.37435365e-01 -2.99555331e-01 -3.37969661e-01
-1.24782562e+00 2.09000081e-01 1.05299488e-01 -4.52260017e-01
-3.37095469e-01 -4.78843182e-01 -6.79452717e-01 -1.90936372e-01
3.77300978e-01 3.87919307e-01 -3.38905394e-01 -4.94811356e-01
2.15205163e-01 7.05629647e-01 4.50142622e-01 -2.84878492e-01
5.77853799e-01 1.19683886e+00 -1.43324798e-02 -1.08258319e+00
-1.01069558e+00 -8.24259818e-01 -7.19829559e-01 -5.73325038e-01
1.07758605e+00 -1.33511615e+00 -6.69227242e-01 8.48590612e-01
-1.35372460e+00 -7.23361433e-01 1.10881612e-01 5.24984419e-01
-3.22058141e-01 4.52371359e-01 -6.14243090e-01 -1.13299096e+00
1.24484524e-01 -1.05542481e+00 1.55678320e+00 2.01695397e-01
2.86143899e-01 -7.31217265e-01 -3.95151615e-01 5.83485425e-01
1.18700620e-02 1.90958813e-01 5.08145094e-01 2.87665904e-01
-1.63565969e+00 1.99126825e-01 -8.13569605e-01 2.33278304e-01
1.21691264e-01 -4.61007990e-02 -1.49118149e+00 1.17626630e-01
1.07961878e-01 2.89498538e-01 9.92069840e-01 9.18609321e-01
1.02173829e+00 1.61433354e-01 -6.49967372e-01 8.60015452e-01
1.43071306e+00 5.98746181e-01 8.67738664e-01 2.59558201e-01
9.84481692e-01 5.83962262e-01 6.91078424e-01 5.22353709e-01
2.56932437e-01 6.58515453e-01 6.75035596e-01 -5.28235957e-02
-5.87752402e-01 9.02523398e-02 4.15123105e-01 4.57277834e-01
7.08548799e-02 -5.10430336e-01 -1.02664495e+00 6.02170110e-01
-1.41687262e+00 -1.01534534e+00 -6.13400280e-01 2.02107477e+00
6.77450538e-01 3.68326843e-01 1.74247636e-03 6.97208792e-02
7.71850526e-01 1.78581849e-02 -8.21181297e-01 1.91695556e-01
-5.56271195e-01 -1.79687500e-01 9.05072391e-01 7.56059945e-01
-1.19780087e+00 9.39479113e-01 5.66114426e+00 7.25983918e-01
-9.64630544e-01 3.60795185e-02 8.55544746e-01 -5.67354798e-01
-1.86256886e-01 -1.14339992e-01 -1.49354243e+00 4.87903625e-01
3.54358464e-01 1.20081365e-01 1.59871176e-01 8.02016199e-01
2.85664290e-01 -8.96719813e-01 -1.13704860e+00 1.45725989e+00
1.13520108e-01 -1.13094831e+00 -1.81422293e-01 3.67587507e-01
1.07411504e+00 6.81406632e-02 -3.91064808e-02 -2.16527537e-01
-4.96240854e-02 -6.68963850e-01 9.14126337e-01 2.86879510e-01
6.68181062e-01 -4.87976432e-01 5.40260196e-01 3.56236577e-01
-1.09024179e+00 7.82366749e-03 -4.90762621e-01 -2.03577146e-01
3.35403621e-01 1.09169769e+00 -5.98125994e-01 1.51760355e-01
7.65239358e-01 8.79363596e-01 -5.35659075e-01 1.12519383e+00
-2.35705346e-01 3.79014611e-01 -2.01224446e-01 -8.34202915e-02
2.52509210e-02 -7.04813972e-02 4.64255959e-01 1.18189335e+00
9.90958586e-02 3.00633490e-01 -1.07640333e-01 9.35169280e-01
4.40001599e-02 -2.86837190e-01 -5.14434636e-01 1.29057705e-01
3.60316277e-01 8.64787459e-01 -1.01228011e+00 -2.75440007e-01
-7.32656121e-01 1.29094446e+00 -3.07826042e-01 2.33534396e-01
-9.37727630e-01 -4.12320271e-02 8.92677963e-01 4.73402560e-01
4.15795386e-01 -3.36174160e-01 -5.30796766e-01 -1.27184856e+00
-4.90391143e-02 -5.97652555e-01 -7.29316324e-02 -1.07345057e+00
-9.02837396e-01 1.39986305e-02 8.90083015e-02 -1.22956049e+00
3.06797594e-01 -1.13370085e+00 -5.48600554e-02 7.87508309e-01
-1.73365569e+00 -6.95072234e-01 -8.01978886e-01 3.12497258e-01
8.24114978e-01 4.15091604e-01 5.37622683e-02 6.43795848e-01
-3.95791352e-01 6.11734055e-02 9.16438773e-02 8.16749930e-02
5.94909251e-01 -1.22973156e+00 7.26364076e-01 1.00056541e+00
-2.93353423e-02 4.86357696e-02 4.63659286e-01 -7.68758357e-01
-1.67508304e+00 -8.71290863e-01 7.34582067e-01 -8.56157005e-01
2.57338315e-01 -6.85378492e-01 -7.09085464e-01 4.46847469e-01
-5.99775240e-02 2.38448493e-02 2.02958569e-01 -4.89521235e-01
8.42211843e-02 2.14507971e-02 -8.53093684e-01 7.63277650e-01
1.48338306e+00 -3.27072769e-01 -3.21420245e-02 2.56092548e-01
7.66025126e-01 -8.14301789e-01 -1.71965405e-01 4.13867563e-01
6.56504631e-01 -1.51258755e+00 1.28447449e+00 2.96097308e-01
4.82052982e-01 -6.63463593e-01 -6.00063026e-01 -4.53698575e-01
3.89363803e-02 -2.94273794e-01 -6.24537349e-01 9.07875180e-01
7.28729218e-02 -4.78318810e-01 1.30623710e+00 2.59056032e-01
-1.25120446e-01 -6.37406349e-01 -9.11171496e-01 -7.04399288e-01
-1.94190681e-01 -6.20299041e-01 1.32104516e-01 5.68517387e-01
-5.80562711e-01 2.65306860e-01 -2.37325504e-01 5.64620972e-01
9.99410033e-01 3.77675623e-01 1.00907075e+00 -1.01918936e+00
-3.93149704e-01 -5.15534163e-01 -5.85321963e-01 -1.94018757e+00
-3.58093858e-01 -5.64897835e-01 -1.88832451e-02 -1.68796968e+00
4.16270494e-01 -8.30624923e-02 2.21251443e-01 -8.58932659e-02
-3.19382071e-01 4.96374249e-01 1.00823022e-01 -4.50407788e-02
-6.24925137e-01 4.04577762e-01 1.51706100e+00 -2.99073923e-02
-2.20143632e-03 2.74681728e-02 -5.08586466e-01 8.54052424e-01
7.49008060e-01 -2.93718934e-01 -2.80776650e-01 -6.54499948e-01
-9.41662490e-02 8.37099925e-02 5.59076846e-01 -1.07044673e+00
9.27431583e-02 -2.55930245e-01 8.98829937e-01 -1.25352728e+00
6.77039742e-01 -7.12644398e-01 3.90261896e-02 5.62603891e-01
3.81003112e-01 -5.16176999e-01 3.95555973e-01 5.65567672e-01
8.96617491e-03 -1.13179855e-01 1.05963933e+00 -2.61693954e-01
-1.07112610e+00 2.70777196e-01 -4.56518173e-01 1.54461727e-01
1.04119515e+00 -7.88667917e-01 -5.49888074e-01 -2.39075139e-01
-3.90929967e-01 8.02612454e-02 6.27520740e-01 1.90548003e-01
8.27842176e-01 -1.07454324e+00 -6.86773598e-01 1.69277951e-01
6.16959967e-02 2.69624323e-01 3.33870500e-01 7.84136951e-01
-9.58817184e-01 5.49582779e-01 2.91717499e-02 -1.10293889e+00
-1.47137451e+00 4.90861684e-01 5.49228251e-01 4.09740984e-01
-5.86400270e-01 1.31589866e+00 1.04608417e+00 4.70898747e-02
3.75852197e-01 -6.75661564e-01 3.74030948e-01 -1.40106589e-01
7.75617719e-01 6.07126534e-01 -2.15706646e-01 -2.80826449e-01
-4.95007813e-01 8.17712903e-01 1.40546650e-01 -1.01004757e-01
8.91550422e-01 -5.32126844e-01 5.22129759e-02 7.90061116e-01
9.94094908e-01 1.30869120e-01 -1.80995739e+00 -2.31829241e-01
-2.71860808e-01 -1.26632226e+00 3.06941867e-01 -6.64405167e-01
-9.81364489e-01 1.24653971e+00 7.71725714e-01 -1.39375702e-01
1.23952150e+00 3.51679742e-01 5.48889399e-01 -4.38165329e-02
4.94225711e-01 -7.05897987e-01 2.90180326e-01 5.10710716e-01
4.06974018e-01 -1.55214417e+00 1.90495491e-01 -9.47359264e-01
-4.05097008e-01 8.50296080e-01 7.60219991e-01 1.47922590e-01
4.47867423e-01 6.14152908e-01 1.14306688e-01 -3.93881090e-02
-6.23621583e-01 -5.23213506e-01 2.82682151e-01 7.56040990e-01
4.04333502e-01 -1.92882076e-01 3.55489820e-01 4.59224433e-02
3.25911373e-01 -2.22328544e-01 4.33934361e-01 8.47731411e-01
-8.34500968e-01 -9.33960557e-01 -5.90313971e-01 1.86323673e-01
-4.19754773e-01 -2.14306980e-01 -4.81856197e-01 8.27996135e-01
3.64804298e-01 9.77765262e-01 1.41092315e-01 -7.38003775e-02
1.58447191e-01 -2.41167769e-01 9.85089779e-01 -7.26053655e-01
4.06414531e-02 4.19071019e-01 -4.02820110e-02 -6.76038802e-01
-3.50096256e-01 -9.06129837e-01 -1.12123132e+00 -5.84732473e-01
-4.60666776e-01 -6.82648718e-01 7.11469829e-01 5.07220805e-01
1.24415144e-01 4.81233835e-01 4.30992275e-01 -1.26484537e+00
2.59947419e-01 -4.94458705e-01 -7.66325176e-01 -1.76967993e-01
4.66905147e-01 -7.63062835e-01 -5.68656266e-01 1.88130483e-01] | [7.958476543426514, -2.509626865386963] |
0d15964d-0e22-4f1e-979a-8616edd670dd | logiformer-a-two-branch-graph-transformer | 2205.00731 | null | https://arxiv.org/abs/2205.00731v2 | https://arxiv.org/pdf/2205.00731v2.pdf | Logiformer: A Two-Branch Graph Transformer Network for Interpretable Logical Reasoning | Machine reading comprehension has aroused wide concerns, since it explores the potential of model for text understanding. To further equip the machine with the reasoning capability, the challenging task of logical reasoning is proposed. Previous works on logical reasoning have proposed some strategies to extract the logical units from different aspects. However, there still remains a challenge to model the long distance dependency among the logical units. Also, it is demanding to uncover the logical structures of the text and further fuse the discrete logic to the continuous text embedding. To tackle the above issues, we propose an end-to-end model Logiformer which utilizes a two-branch graph transformer network for logical reasoning of text. Firstly, we introduce different extraction strategies to split the text into two sets of logical units, and construct the logical graph and the syntax graph respectively. The logical graph models the causal relations for the logical branch while the syntax graph captures the co-occurrence relations for the syntax branch. Secondly, to model the long distance dependency, the node sequence from each graph is fed into the fully connected graph transformer structures. The two adjacent matrices are viewed as the attention biases for the graph transformer layers, which map the discrete logical structures to the continuous text embedding space. Thirdly, a dynamic gate mechanism and a question-aware self-attention module are introduced before the answer prediction to update the features. The reasoning process provides the interpretability by employing the logical units, which are consistent with human cognition. The experimental results show the superiority of our model, which outperforms the state-of-the-art single model on two logical reasoning benchmarks. | ['Jun Liu', 'Lingling Zhang', 'Yudai Pan', 'Qika Lin', 'Fangzhi Xu'] | 2022-05-02 | null | null | null | null | ['machine-reading-comprehension'] | ['natural-language-processing'] | [ 1.96596727e-01 6.13478065e-01 -3.35113355e-03 -4.14727926e-01
1.36052683e-01 -4.40612018e-01 5.20593941e-01 4.82428163e-01
-5.14215007e-02 2.25589290e-01 6.30479991e-01 -7.48330057e-01
-3.40905637e-01 -1.26198006e+00 -7.22809970e-01 -3.55078250e-01
4.78230834e-01 3.96584123e-01 4.20130789e-01 -5.46990693e-01
1.78651780e-01 -5.45603670e-02 -1.28803241e+00 5.88479459e-01
1.19469965e+00 1.00434959e+00 1.03347346e-01 5.87358952e-01
-8.24757516e-01 1.50000858e+00 -3.16664398e-01 -7.49211252e-01
-2.17058197e-01 -8.04949701e-01 -1.01297593e+00 -3.23670059e-01
2.66002595e-01 -3.04998398e-01 -6.58780932e-01 1.28557968e+00
8.75956789e-02 4.17247927e-03 5.01866460e-01 -1.27813971e+00
-1.16455042e+00 1.20840633e+00 -2.03884408e-01 2.67028987e-01
6.88143909e-01 1.50840595e-01 1.57434905e+00 -6.26208544e-01
3.39498520e-01 1.73590243e+00 4.75475878e-01 2.26930097e-01
-8.45210016e-01 -3.49972039e-01 7.24945068e-01 9.02800083e-01
-9.24225986e-01 3.69071364e-02 9.67489183e-01 -3.50574613e-01
1.09348834e+00 1.92636475e-01 8.96132946e-01 7.98233330e-01
6.29114330e-01 9.13694918e-01 9.85301256e-01 -4.45488125e-01
1.35233356e-02 3.78286652e-02 9.99880433e-01 1.26707923e+00
2.17220217e-01 -3.17488074e-01 -4.93374914e-01 2.52460659e-01
4.17882413e-01 1.53360665e-01 -3.67214113e-01 -1.94240540e-01
-9.68243301e-01 9.49611247e-01 8.57531071e-01 1.77098393e-01
-2.97322601e-01 1.15810566e-01 4.12094295e-01 4.51079637e-01
-6.85772598e-02 3.38172615e-01 -4.48517472e-01 2.41288289e-01
-2.53649116e-01 1.00986967e-02 1.02957702e+00 9.45004940e-01
5.84614873e-01 -3.07691455e-01 -4.14279878e-01 4.34709072e-01
8.63285065e-01 2.67858505e-01 3.37776035e-01 -3.53420258e-01
7.77542233e-01 1.48562336e+00 -6.83306694e-01 -1.44239628e+00
-5.42063594e-01 -4.12220448e-01 -9.46055472e-01 -3.05983633e-01
1.49369940e-01 2.10969463e-01 -5.28873861e-01 1.71089447e+00
2.34240472e-01 -1.78966969e-02 1.46120101e-01 7.12614119e-01
1.21675062e+00 6.37875438e-01 -1.06707998e-02 7.94293955e-02
1.92770791e+00 -1.26156497e+00 -1.05553341e+00 -3.56672496e-01
4.61742222e-01 -2.64959127e-01 1.33723080e+00 1.57906666e-01
-1.00681508e+00 -7.18653083e-01 -1.26676524e+00 -6.79913819e-01
-6.50078416e-01 -1.04757451e-01 5.44041753e-01 1.36524975e-01
-8.24992239e-01 1.57261834e-01 -5.77603996e-01 -3.39313537e-01
4.27089274e-01 5.70173189e-02 -8.82557556e-02 -3.43420386e-01
-1.90798664e+00 9.30558681e-01 8.06743026e-01 5.30789852e-01
-3.18942994e-01 -3.98059547e-01 -1.17368269e+00 6.98373020e-01
6.42856836e-01 -1.00510550e+00 7.16161311e-01 -5.69167316e-01
-1.33279979e+00 4.91574019e-01 -1.69748172e-01 -3.42786729e-01
2.16911271e-01 -1.86581850e-01 -3.09379369e-01 5.42876162e-02
-8.97329226e-02 2.55767226e-01 7.57925868e-01 -9.79333460e-01
-4.51830298e-01 -2.70814806e-01 5.89929581e-01 4.82130915e-01
-4.41451252e-01 -2.90536761e-01 -5.07422209e-01 -4.82363045e-01
3.40038866e-01 -3.64502758e-01 2.62612253e-01 -1.91515610e-01
-5.93721688e-01 -5.83122194e-01 5.78574300e-01 -7.89967120e-01
1.73960888e+00 -1.79973674e+00 5.86415350e-01 4.53525037e-02
8.88048470e-01 -9.21892822e-02 -3.06515731e-02 5.00798762e-01
-2.47927845e-01 2.10373834e-01 -8.36162418e-02 1.41981408e-01
3.64686459e-01 2.41009265e-01 -6.46603465e-01 -9.81022194e-02
3.66156965e-01 1.44250488e+00 -9.91535902e-01 -4.49974775e-01
6.60791621e-02 -5.08957431e-02 -5.90435624e-01 4.60639864e-01
-5.48602343e-01 -1.57625139e-01 -6.01661325e-01 2.67800778e-01
6.32695794e-01 -5.90979636e-01 3.22465390e-01 -5.77891529e-01
4.55277741e-01 7.05254078e-01 -9.36808765e-01 1.38415754e+00
-1.39880195e-01 4.19715554e-01 -3.31609011e-01 -1.14582431e+00
8.92058194e-01 -9.77061316e-02 -2.65811741e-01 -8.33005607e-01
3.12521428e-01 -2.43273556e-01 3.77279788e-01 -8.43453288e-01
5.83942235e-02 -3.29252005e-01 -1.68342099e-01 5.54658175e-01
1.60967216e-01 -7.51868039e-02 1.59848794e-01 5.28144598e-01
1.23367155e+00 2.06632525e-01 3.16879183e-01 -2.38062590e-01
1.08179271e+00 -1.94683760e-01 2.34542266e-01 4.87310737e-01
1.93796046e-02 -2.38281991e-02 1.24776256e+00 -5.03178656e-01
-4.17707801e-01 -1.23498213e+00 3.28031421e-01 9.44037020e-01
4.03621256e-01 -7.75114655e-01 -7.92025149e-01 -8.37259650e-01
-2.99198069e-02 9.55171883e-01 -8.35431576e-01 -8.21630597e-01
-6.04021668e-01 -4.95094925e-01 5.44808984e-01 6.81804895e-01
9.74259973e-01 -1.15482485e+00 -3.65463972e-01 1.51476964e-01
-3.79971415e-01 -1.12660635e+00 -3.08935165e-01 1.82628006e-01
-4.83280867e-01 -1.27833819e+00 4.69248563e-01 -8.74398887e-01
6.49894357e-01 -3.98720838e-02 1.19194806e+00 6.34801686e-01
1.70832962e-01 2.30499655e-01 -3.38354170e-01 -3.16351533e-01
-3.24953437e-01 1.19309694e-01 -3.67779940e-01 1.03401184e-01
6.65161371e-01 -3.72331113e-01 -2.07267836e-01 -3.79334874e-02
-9.11686361e-01 5.07728755e-01 5.69860518e-01 1.01570356e+00
3.62401068e-01 5.67215562e-01 3.92679036e-01 -9.46979821e-01
1.17529047e+00 -5.48020780e-01 -3.51970077e-01 6.83764935e-01
-6.82660937e-01 4.97540683e-01 1.06657171e+00 -1.65853664e-01
-1.16917145e+00 -4.55122501e-01 1.77754890e-02 -2.03898251e-01
2.71486670e-01 1.12387812e+00 -5.89706838e-01 5.09586811e-01
2.29600444e-01 3.33283812e-01 -8.09985399e-02 -8.45493376e-02
6.95350051e-01 2.78611004e-01 2.89275080e-01 -6.14838481e-01
9.13387299e-01 3.42152417e-02 2.50112601e-02 -4.58154798e-01
-1.16339362e+00 1.52924890e-02 -7.24834144e-01 8.08876231e-02
1.10775089e+00 -4.75039750e-01 -1.01077211e+00 4.02572215e-01
-1.38375878e+00 -1.90268219e-01 -2.77562708e-01 1.14706144e-01
-3.49854529e-01 4.57732588e-01 -8.40304494e-01 -4.91475463e-01
-3.31962436e-01 -1.18282497e+00 8.08490872e-01 4.19681847e-01
-1.76164225e-01 -1.47453547e+00 -2.51925379e-01 5.49458146e-01
1.22223981e-01 -8.26030374e-02 1.91129100e+00 -7.60362625e-01
-8.06008279e-01 -3.08465566e-02 -6.56989872e-01 2.63216794e-01
-4.50278400e-03 -1.86464921e-01 -7.75945485e-01 7.75180012e-02
3.04263860e-01 -2.98705757e-01 9.05981004e-01 -1.34996340e-01
1.19244444e+00 -3.83644670e-01 -1.33222163e-01 4.35889333e-01
1.13009226e+00 6.59733415e-02 7.25535631e-01 5.79761304e-02
1.13364303e+00 5.97544789e-01 1.55542642e-01 -1.08827591e-01
1.12843430e+00 5.55559956e-02 3.51730049e-01 -7.14838877e-03
-1.44825324e-01 -8.84634078e-01 3.04322869e-01 1.32368112e+00
2.95032442e-01 -5.08179069e-01 -9.90050018e-01 3.05388868e-03
-1.97038603e+00 -8.53041649e-01 -2.37734601e-01 1.66235852e+00
8.21260154e-01 5.85448921e-01 -5.35353780e-01 2.34918028e-01
3.23592395e-01 2.73500860e-01 -6.06516600e-01 -4.76292580e-01
-5.03787920e-02 1.18675135e-01 -9.11678746e-02 7.75573134e-01
-6.93901777e-01 1.06894684e+00 5.34489155e+00 5.56187749e-01
-7.32429147e-01 -2.06302807e-01 3.33441764e-01 3.99924070e-01
-6.37268066e-01 2.82255858e-01 -7.35720873e-01 2.74002612e-01
7.37220168e-01 -1.44575909e-01 5.62356532e-01 2.81845778e-01
-1.27156168e-01 -3.97521034e-02 -1.43564904e+00 6.42720580e-01
3.38321358e-01 -1.12651110e+00 6.04950607e-01 -2.77085155e-01
1.57662764e-01 -5.72631299e-01 -3.15509588e-01 8.05671573e-01
3.85875642e-01 -1.29009938e+00 7.10077226e-01 9.92536724e-01
4.51824248e-01 -4.30248052e-01 8.49155724e-01 5.90406001e-01
-1.65508306e+00 -3.38133097e-01 -4.24802661e-01 -4.37022179e-01
-2.49764994e-02 2.79393792e-01 -5.52988410e-01 8.57653141e-01
5.36447525e-01 9.46039081e-01 -1.10359132e+00 2.24334076e-01
-9.94971216e-01 5.54114282e-01 3.23058330e-02 -5.19702435e-01
2.03665912e-01 -4.17848736e-01 1.47416428e-01 8.71050775e-01
-2.27386132e-01 3.01627725e-01 8.08793828e-02 1.39040077e+00
-3.07580322e-01 4.14716005e-02 -4.23289567e-01 -3.52334499e-01
3.20890903e-01 1.15002859e+00 -5.54947555e-01 -5.17535925e-01
-5.54410100e-01 1.01733232e+00 8.19543779e-01 3.44685912e-01
-1.00812936e+00 -6.44031882e-01 2.15725407e-01 -1.58168860e-02
6.49559125e-02 -2.01654248e-02 -5.19691408e-01 -1.40908444e+00
3.19758207e-01 -1.02484345e+00 6.81125760e-01 -1.04835343e+00
-1.39007914e+00 5.66001356e-01 7.22310171e-02 -6.03951931e-01
1.68396190e-01 -8.22620809e-01 -8.80913019e-01 1.11656463e+00
-1.45798790e+00 -1.24240971e+00 -6.38266683e-01 4.80958581e-01
4.99393642e-01 -9.89533886e-02 6.60133660e-01 -1.16407499e-01
-7.50305593e-01 5.55516183e-01 -4.79173154e-01 3.59636545e-01
7.60337338e-02 -1.30381644e+00 4.14030105e-01 8.86051178e-01
3.30755413e-02 1.10297763e+00 2.96393007e-01 -6.11037970e-01
-1.74948311e+00 -7.71514595e-01 1.04148793e+00 -7.23163843e-01
1.09273040e+00 -7.00174272e-01 -1.36952603e+00 9.90744948e-01
4.13402438e-01 -3.59779030e-01 4.45511043e-01 1.66605040e-01
-5.79651713e-01 -8.49961787e-02 -6.25749767e-01 9.02353525e-01
1.13011217e+00 -7.97871888e-01 -1.46280730e+00 1.80932581e-01
1.20713317e+00 -3.39114696e-01 -6.56018317e-01 1.44237563e-01
3.85794640e-01 -8.11109841e-01 7.20276475e-01 -9.14695024e-01
8.64041686e-01 -4.49378580e-01 -4.68053529e-03 -1.25949335e+00
-6.35720313e-01 -1.55910209e-01 -3.55578929e-01 1.20013821e+00
4.59663153e-01 -8.09933186e-01 3.79352748e-01 5.37114739e-01
1.25060398e-02 -1.12580752e+00 -5.88784099e-01 -1.32945269e-01
1.52133867e-01 -3.54137748e-01 7.33669519e-01 9.17561829e-01
4.91347402e-01 1.35628438e+00 1.49552763e-01 8.86726007e-02
2.83858955e-01 2.82561570e-01 4.01901245e-01 -1.38937175e+00
-3.32227290e-01 -7.48901486e-01 -2.99876511e-01 -1.50819004e+00
3.94363552e-01 -1.31920874e+00 -1.93337560e-01 -2.14710307e+00
1.76587656e-01 4.73320298e-02 -3.82079244e-01 5.13403654e-01
-7.61577845e-01 -8.65974844e-01 1.11693785e-01 -2.34087005e-01
-5.73730350e-01 7.79972970e-01 1.50919223e+00 -4.84101295e-01
1.63086534e-01 -4.42927003e-01 -1.10049021e+00 7.49563932e-01
5.91597855e-01 -2.42049862e-02 -1.07201529e+00 -5.81757009e-01
8.26971829e-01 -2.32359022e-02 3.25930297e-01 -5.90344131e-01
5.71846366e-01 -1.75106257e-01 2.62115568e-01 -6.35368407e-01
2.82226056e-02 -1.01320791e+00 -3.66426587e-01 5.66901088e-01
-5.38437188e-01 2.52484262e-01 1.36679277e-01 6.75206661e-01
-1.85993344e-01 -1.45415515e-01 5.01322508e-01 -6.43137693e-02
-5.45340121e-01 1.34292215e-01 -9.12624449e-02 2.97852010e-01
6.67092085e-01 5.40977810e-03 -6.68062925e-01 -2.52394646e-01
-4.84820038e-01 7.63645530e-01 -3.41153406e-02 6.92023575e-01
8.37250233e-01 -1.33523619e+00 -4.91110474e-01 5.42698383e-01
1.02733351e-01 1.78316787e-01 2.01477334e-01 8.10428619e-01
-5.07583320e-01 4.14153278e-01 -1.21928521e-01 -2.61992991e-01
-9.86003280e-01 9.43078935e-01 5.92467189e-01 -6.74804986e-01
-5.71023643e-01 8.21106911e-01 4.78674352e-01 -5.93364716e-01
2.01612264e-01 -9.84300435e-01 -6.36745453e-01 1.31332442e-01
5.18884182e-01 1.60773650e-01 -1.67286277e-01 -2.16190964e-01
-3.06090504e-01 5.82071722e-01 -9.17314887e-02 3.08965594e-01
1.00766253e+00 -2.25112140e-01 -7.52829015e-01 6.80004597e-01
9.13148880e-01 -1.27458766e-01 -7.53537536e-01 -5.40179968e-01
8.78299847e-02 4.77186441e-02 -9.45490748e-02 -7.57871211e-01
-8.08671296e-01 1.32070148e+00 -5.02042063e-02 5.42546034e-01
1.06640697e+00 -9.58893225e-02 6.62726820e-01 6.43752396e-01
-5.76874949e-02 -7.55783141e-01 2.25668788e-01 1.13784719e+00
1.11357307e+00 -1.00132322e+00 2.22873054e-02 -7.17022777e-01
-4.85479116e-01 1.47568440e+00 9.39446390e-01 2.82436535e-02
7.13189781e-01 1.28895357e-01 -1.90706983e-01 -6.27727747e-01
-1.01846099e+00 -1.29393479e-02 5.04619777e-01 1.56547204e-01
3.61827254e-01 -2.32726727e-02 -2.91761965e-01 1.10219586e+00
-4.21036333e-01 -2.31796697e-01 3.45935225e-01 5.90498149e-01
-4.26616073e-01 -7.74716437e-01 -1.28562719e-01 5.48460066e-01
1.85100511e-01 -5.19458115e-01 -6.53967738e-01 6.38280928e-01
1.52349174e-01 9.37956333e-01 -1.35702685e-01 -5.63558877e-01
5.83853781e-01 4.61360872e-01 3.68087471e-01 -6.11957133e-01
-5.15843570e-01 -4.26336944e-01 -2.78392099e-02 -4.26740348e-01
1.28361553e-01 -2.50774711e-01 -1.77134383e+00 -3.23803246e-01
-2.83452243e-01 2.27563344e-02 -3.58571075e-02 1.22267711e+00
2.22698241e-01 1.27541316e+00 2.78759301e-01 1.94061518e-01
-6.25305831e-01 -9.49108005e-01 -3.16380560e-01 5.25625944e-01
1.84972733e-01 -6.17374837e-01 -3.51456165e-01 -3.47924866e-02] | [9.664746284484863, 7.643124103546143] |
258c0397-0fcb-4e61-97f2-afb3407b6c65 | on-the-robustness-of-self-supervised-1 | 2209.15483 | null | https://arxiv.org/abs/2209.15483v2 | https://arxiv.org/pdf/2209.15483v2.pdf | Augmentation Invariant Discrete Representation for Generative Spoken Language Modeling | Generative Spoken Language Modeling research focuses on optimizing speech Language Models (LMs) using raw audio recordings without accessing any textual supervision. Such speech LMs usually operate over discrete units obtained from quantizing internal representations of self-supervised models. Although such units show impressive modeling results, their robustness capabilities have not been extensively investigated. This work focuses on improving the robustness of discrete input representations for generative spoken language modeling. First, we formally define how to measure the robustness of such representations to various signal variations that do not alter the spoken information (e.g., time-stretch). Next, we empirically demonstrate how current state-of-the-art representation models lack robustness to such variations. To overcome this, we propose an effective and efficient method to learn robust discrete speech representation for generative spoken language modeling. The proposed approach is based on applying a set of signal transformations to the speech signal and optimizing the model using an iterative pseudo-labeling scheme. Our method significantly improves over the evaluated baselines when considering encoding and modeling metrics. We additionally evaluate our method on the speech-to-speech translation task, considering Spanish-English and French-English translations, and show the proposed approach outperforms the evaluated baselines. | ['Emmanuel Dupoux', 'Gabriel Synnaeve', 'Tu Anh Nguyen', 'Yossi Adi', 'Jade Copet', 'Ann Lee', 'Felix Kreuk', 'Itai Gat'] | 2022-09-30 | null | null | null | null | ['speech-to-speech-translation'] | ['speech'] | [ 4.41299319e-01 3.81303757e-01 7.17989430e-02 -7.55145073e-01
-1.49206412e+00 -5.60123146e-01 8.10589373e-01 -1.36298269e-01
-2.79405385e-01 5.15329182e-01 4.51980591e-01 -2.23156139e-01
3.09700072e-01 -4.75194097e-01 -9.41084325e-01 -6.38541341e-01
6.49328008e-02 3.21407586e-01 -1.02169029e-01 -2.12448090e-01
-1.20710609e-02 3.45579445e-01 -1.49446380e+00 2.84093559e-01
7.63971090e-01 6.02180481e-01 1.69958621e-01 8.58863711e-01
-2.11023241e-01 6.08835757e-01 -9.99577105e-01 1.25617325e-01
-6.89513609e-02 -8.44233632e-01 -6.74161375e-01 5.11422396e-01
3.06012332e-01 -1.31963998e-01 -2.27880538e-01 1.00236320e+00
6.64972186e-01 2.76992768e-01 7.76570499e-01 -9.46605802e-01
-7.50595391e-01 1.14199984e+00 1.59098133e-01 1.58590630e-01
3.96309346e-01 -3.64164077e-02 8.54185522e-01 -1.17480969e+00
4.12187278e-01 1.60399854e+00 2.15282217e-01 5.97171128e-01
-1.45142877e+00 -5.14798999e-01 3.16896111e-01 3.72032188e-02
-1.69963205e+00 -1.09282196e+00 7.64554620e-01 -1.66837111e-01
1.26468658e+00 2.78660804e-01 5.32427849e-03 1.18935454e+00
-3.03450711e-02 8.08329523e-01 9.23627853e-01 -7.80972183e-01
3.58024180e-01 4.39855635e-01 1.52638629e-01 5.16414940e-01
-4.17708516e-01 5.46940081e-02 -6.03540778e-01 -3.78338806e-02
3.00996631e-01 -5.56170642e-01 -3.72429669e-01 -4.63602208e-02
-1.10779345e+00 7.04504669e-01 7.44233951e-02 4.44030374e-01
-3.18616986e-01 2.23087087e-01 2.80914694e-01 4.75622058e-01
6.74200356e-01 1.67954892e-01 -2.54060775e-01 -2.96918035e-01
-1.25383544e+00 -1.58851773e-01 7.49664307e-01 1.16377461e+00
5.84882081e-01 7.05741644e-01 -1.11476310e-01 1.06327319e+00
5.42718768e-01 6.07665896e-01 7.93869674e-01 -5.79585135e-01
5.28808832e-01 1.08381018e-01 -1.19838312e-01 -5.95768750e-01
-1.95345387e-01 -4.37616199e-01 -7.13262618e-01 -1.84466064e-01
-9.39232484e-02 -2.08384708e-01 -1.02835381e+00 1.87313151e+00
1.39780091e-02 2.80437887e-01 4.81168002e-01 6.10871732e-01
7.87543595e-01 1.09488702e+00 -9.86993760e-02 -4.64851826e-01
8.37108970e-01 -1.03090036e+00 -1.04210615e+00 -2.52606273e-01
5.28615892e-01 -8.62545013e-01 1.33079302e+00 4.32092309e-01
-1.26739609e+00 -7.27840781e-01 -1.18982089e+00 1.21131338e-01
-2.58814365e-01 4.71327990e-01 -1.84722140e-01 9.33550537e-01
-1.24861526e+00 3.93278539e-01 -1.00647438e+00 -2.82926172e-01
-2.86043674e-01 3.36308062e-01 -1.30939662e-01 2.79283106e-01
-1.40708482e+00 7.66533434e-01 2.61916012e-01 8.56658146e-02
-1.27108455e+00 -3.88790816e-01 -1.03423119e+00 1.19731650e-01
2.47294247e-01 -2.87868440e-01 1.59371412e+00 -8.79623055e-01
-2.08997059e+00 4.77983743e-01 -4.97388005e-01 -7.07668602e-01
2.16274098e-01 -2.25632429e-01 -6.88180208e-01 1.04117142e-02
-2.66038805e-01 5.48555791e-01 1.21872556e+00 -1.44464242e+00
-2.67977983e-01 7.37059563e-02 -2.96791971e-01 1.15261599e-01
-4.42444712e-01 1.50131255e-01 -2.32591346e-01 -9.11343396e-01
2.16517165e-01 -8.91982436e-01 -7.59800673e-02 -5.99160254e-01
-4.62399453e-01 -8.93669128e-02 6.67620599e-01 -6.41837299e-01
1.57668817e+00 -2.12736416e+00 2.49319136e-01 1.41403764e-01
-5.46488225e-01 3.03254277e-01 -3.16139907e-01 7.52106130e-01
-8.90125930e-02 3.26718807e-01 -2.85835832e-01 -1.07438684e+00
2.79832721e-01 3.61571372e-01 -7.96061099e-01 2.80904651e-01
2.95593888e-01 6.97579443e-01 -5.57466209e-01 -3.82693321e-01
3.43532056e-01 8.82281423e-01 -3.40400070e-01 3.96827519e-01
-2.85352886e-01 3.76695693e-01 -7.17955921e-03 4.51741159e-01
2.10848048e-01 2.07106993e-01 1.92399826e-02 -1.30066359e-02
-1.30564913e-01 9.49702978e-01 -1.22583067e+00 1.64579463e+00
-8.28940570e-01 5.08731186e-01 -1.15675130e-03 -9.73637223e-01
1.25143993e+00 7.20714033e-01 3.51574831e-02 -3.74419332e-01
1.26381412e-01 1.94121867e-01 -1.96124882e-01 -2.06454530e-01
4.76131231e-01 -5.64859062e-02 -1.31229579e-01 6.73670352e-01
4.57090527e-01 -5.39211631e-01 -6.46810606e-02 6.46076277e-02
6.42870963e-01 -1.07655143e-02 3.62117589e-01 3.38013507e-02
6.22142792e-01 -4.79562730e-01 9.90850553e-02 7.36256599e-01
6.64236918e-02 7.98385322e-01 1.72635075e-02 2.60914117e-01
-9.05621946e-01 -1.13682127e+00 7.40600079e-02 1.27527320e+00
-3.00515234e-01 -5.41938066e-01 -1.16747665e+00 -2.55991638e-01
-4.91954476e-01 1.26853430e+00 -3.56775641e-01 -4.94786441e-01
-4.67737019e-01 -5.27865648e-01 9.81995165e-01 4.92104858e-01
1.33239791e-01 -9.46071625e-01 8.63152556e-03 4.95878100e-01
-2.20034167e-01 -1.12966514e+00 -7.37890422e-01 2.77631193e-01
-7.79919207e-01 -2.23531142e-01 -6.76544726e-01 -7.84325004e-01
4.43996489e-01 1.68566838e-01 9.62382674e-01 -3.89151126e-01
6.84051588e-02 6.00467563e-01 -4.51095968e-01 -3.87950003e-01
-1.10467458e+00 2.58952588e-01 3.10642809e-01 2.94598848e-01
1.34915620e-01 -5.63690305e-01 8.08211043e-02 3.40763867e-01
-1.07665873e+00 -2.03942835e-01 5.76958239e-01 8.27288449e-01
8.14816296e-01 -1.08074300e-01 8.50256383e-01 -5.66757202e-01
1.04407001e+00 -2.53871650e-01 -2.99982697e-01 4.51610595e-01
-6.32142663e-01 4.49828893e-01 7.65186727e-01 -6.67802393e-01
-9.98292565e-01 1.90223217e-01 -3.17048073e-01 -4.93582785e-01
-2.86972165e-01 5.99597871e-01 -4.78337020e-01 2.03357324e-01
7.08630204e-01 6.96994781e-01 -1.93421841e-01 -4.71531123e-01
7.35651672e-01 1.08668697e+00 5.26990712e-01 -4.78850633e-01
6.87252939e-01 2.99750939e-02 -5.26249349e-01 -1.30040586e+00
-4.56200927e-01 -3.12533200e-01 -6.41910911e-01 3.09310146e-02
5.71825564e-01 -8.85964334e-01 -1.47750631e-01 3.14437479e-01
-1.24892604e+00 -2.70637572e-01 -2.02112541e-01 5.51153302e-01
-8.60757411e-01 3.96605402e-01 -5.91176331e-01 -1.09659147e+00
-3.84615690e-01 -1.27923036e+00 1.37455964e+00 -2.95320243e-01
-4.08126503e-01 -1.03675759e+00 1.49406374e-01 1.30968317e-01
5.98711729e-01 -2.21228912e-01 7.51992106e-01 -8.43920648e-01
-3.79156619e-01 -5.88612370e-02 5.49419522e-01 8.05959523e-01
4.63997960e-01 1.36726350e-01 -1.33733130e+00 -4.08117652e-01
8.04500654e-02 -2.43553877e-01 7.36235738e-01 2.10722774e-01
7.89068460e-01 -5.26779354e-01 9.11514759e-02 4.56108779e-01
8.10567856e-01 3.27967584e-01 4.71718311e-01 -1.64204556e-02
3.89938533e-01 4.60514486e-01 5.05159736e-01 3.70656908e-01
2.13228449e-01 7.81383455e-01 1.16374746e-01 -3.35710272e-02
-1.79041058e-01 -3.78898978e-01 9.79861975e-01 1.41705561e+00
3.32805365e-01 -5.54660559e-01 -6.88116968e-01 5.42865634e-01
-1.57825172e+00 -7.32572317e-01 3.55571538e-01 2.20611429e+00
9.22528446e-01 1.91817373e-01 -2.60495190e-02 4.27800298e-01
5.59980512e-01 2.64083654e-01 -3.65259349e-01 -6.09846830e-01
-1.58719882e-01 3.83079886e-01 1.62318766e-01 9.32875931e-01
-1.00344682e+00 1.17566109e+00 6.69380379e+00 6.67447984e-01
-1.29529321e+00 1.64856493e-01 3.42173576e-01 -1.05475061e-01
-3.93099636e-01 -2.51955152e-01 -8.55282724e-01 1.29229993e-01
1.60987508e+00 -4.19306457e-01 5.54898858e-01 7.41467893e-01
7.09383130e-01 4.97937888e-01 -1.36038017e+00 8.75414550e-01
4.42259282e-01 -9.09223437e-01 3.55454564e-01 -2.85413712e-01
6.03412747e-01 -1.48035467e-01 2.40979955e-01 4.42749679e-01
1.03504010e-01 -1.22546494e+00 1.04542601e+00 5.23805261e-01
7.59248555e-01 -7.72385180e-01 5.30015469e-01 4.53903705e-01
-1.23865962e+00 3.40328127e-01 -2.70024598e-01 2.45827869e-01
3.98276508e-01 2.54760712e-01 -1.42618406e+00 3.61053467e-01
1.98407784e-01 3.92489374e-01 -4.32402194e-01 4.79572356e-01
-3.81004900e-01 1.26248920e+00 -3.27961147e-01 -9.48818102e-02
2.77686745e-01 6.75726915e-04 6.21515274e-01 1.62304688e+00
5.41851819e-01 -7.84954578e-02 7.99292400e-02 1.02374566e+00
-3.08058802e-02 2.70028174e-01 -5.34116447e-01 -4.99381036e-01
6.15206540e-01 6.67866051e-01 -4.10270512e-01 -4.62841153e-01
-2.70422548e-01 1.05490851e+00 6.41402882e-03 5.69985330e-01
-7.39372790e-01 -2.96263903e-01 6.47533298e-01 -1.45382341e-02
1.56484187e-01 -5.36325693e-01 2.02570129e-02 -1.01857567e+00
-3.69700901e-02 -1.05812120e+00 -1.13258973e-01 -7.30253875e-01
-9.85360563e-01 1.07896197e+00 5.92049323e-02 -1.21889842e+00
-1.07264912e+00 -1.14237852e-01 -3.69088739e-01 9.92127419e-01
-1.25312841e+00 -1.19358492e+00 -2.61185169e-02 4.99201477e-01
1.06096411e+00 -3.11633497e-01 1.17790532e+00 1.86433285e-01
-3.74942511e-01 9.09042001e-01 8.22230503e-02 -1.56911120e-01
7.40744591e-01 -1.16732764e+00 8.06229711e-01 1.01872396e+00
6.16597354e-01 8.02183926e-01 9.51336861e-01 -2.89539784e-01
-1.27105391e+00 -1.23958600e+00 1.06114614e+00 -2.60586709e-01
5.29255748e-01 -6.84356153e-01 -1.18288553e+00 9.24589336e-01
3.80202323e-01 -2.51264721e-01 7.28221476e-01 -2.38747418e-01
-2.48188004e-01 5.72743192e-02 -7.53802598e-01 4.80762422e-01
7.65391946e-01 -9.41938162e-01 -8.99752676e-01 1.86213061e-01
1.06197286e+00 -2.99909651e-01 -7.34810650e-01 1.09357543e-01
1.53234005e-01 -6.11728609e-01 9.18565869e-01 -5.83123505e-01
-2.32458040e-02 -3.80780160e-01 -5.35266757e-01 -1.70810115e+00
1.07458951e-02 -9.27277386e-01 -4.91227135e-02 1.49542689e+00
6.19141936e-01 -6.04426324e-01 3.02977115e-01 3.26936156e-01
-4.88283664e-01 -4.41278607e-01 -9.99283135e-01 -9.53805804e-01
9.62016210e-02 -5.76177239e-01 5.47836065e-01 6.50768459e-01
4.61962633e-02 4.99731392e-01 -4.93376672e-01 5.39001584e-01
4.18862015e-01 -2.16600999e-01 7.88643122e-01 -6.45010412e-01
-3.91701013e-01 -2.88803399e-01 -3.31890643e-01 -1.30237043e+00
4.02184606e-01 -8.90595376e-01 4.20382857e-01 -1.44018376e+00
-4.33091938e-01 1.62877340e-03 -1.88041404e-01 4.71200675e-01
2.19164062e-02 -1.29152641e-01 2.06865340e-01 3.26762795e-02
-3.71215343e-01 9.71361399e-01 7.05630004e-01 -2.91609585e-01
-5.10668933e-01 1.71649516e-01 -3.00234318e-01 4.36169177e-01
8.77745926e-01 -4.39377517e-01 -7.39917815e-01 -3.79906952e-01
-3.64705861e-01 8.21872130e-02 5.59374876e-02 -9.59287643e-01
3.07757169e-01 7.42648914e-02 -2.75097996e-01 -4.29836363e-01
6.03924572e-01 -5.56730688e-01 2.29449719e-01 1.37164354e-01
-7.35058188e-01 -2.99477167e-02 3.04291338e-01 3.57032180e-01
-7.49319196e-01 -2.57120341e-01 7.93535054e-01 1.56706467e-01
-3.82037938e-01 -1.26825005e-01 -6.88191772e-01 -1.49102971e-01
6.36005461e-01 -5.12839667e-02 1.21188108e-02 -8.32732797e-01
-1.05473435e+00 -1.75388724e-01 8.72353688e-02 7.40672648e-01
7.77289212e-01 -1.25497496e+00 -8.30343306e-01 4.96852428e-01
1.13039277e-01 -3.13128978e-01 -1.36031657e-01 4.75346386e-01
-1.95175204e-02 6.97363555e-01 3.62763226e-01 -6.43625915e-01
-1.38793457e+00 4.35575753e-01 4.85577792e-01 1.28640428e-01
-2.58559078e-01 7.34147310e-01 8.84126350e-02 -5.06412864e-01
6.33401811e-01 -1.00356901e+00 -3.58783081e-02 -1.02860518e-01
4.82643068e-01 2.12198108e-01 4.16428059e-01 -8.64677191e-01
-3.20611179e-01 3.19740742e-01 8.17690045e-02 -6.31846368e-01
1.12075841e+00 -4.81139839e-01 3.28791559e-01 1.02611089e+00
1.19229579e+00 -5.25563769e-02 -9.52596009e-01 -2.78829902e-01
2.85233445e-02 6.38898984e-02 1.12293430e-01 -6.51980221e-01
-6.14810884e-01 1.19922471e+00 6.13637030e-01 2.37155780e-01
9.93628561e-01 -6.13929667e-02 6.51219666e-01 5.24696708e-01
4.34631616e-01 -1.04973924e+00 6.98075294e-02 6.66245043e-01
1.19150841e+00 -1.04013407e+00 -5.13398945e-01 -3.98512661e-01
-6.75239027e-01 1.04682612e+00 1.21933244e-01 5.16060665e-02
5.89941382e-01 4.67042983e-01 4.10303384e-01 2.45407656e-01
-9.32786047e-01 -1.20023817e-01 4.50919122e-01 4.34782982e-01
5.80637872e-01 1.26278818e-01 -1.20432109e-01 7.31197059e-01
-5.23439169e-01 -2.45344952e-01 5.39548695e-01 7.03639209e-01
-5.89563251e-01 -1.21881974e+00 -5.94643474e-01 -9.90608558e-02
-2.56278217e-01 -2.55555630e-01 -6.83308542e-01 4.44757819e-01
-4.36676621e-01 1.30548346e+00 -1.22735001e-01 -4.98060048e-01
5.57521522e-01 4.90701884e-01 2.40467861e-01 -9.21819687e-01
-4.25204873e-01 4.57823753e-01 1.13274455e-01 -3.09891403e-01
-4.76234972e-01 -5.87409973e-01 -1.48813808e+00 2.56197870e-01
-3.15505147e-01 2.61576444e-01 9.11438763e-01 8.48483205e-01
2.68163025e-01 8.21649194e-01 7.70581603e-01 -9.49976265e-01
-1.09190094e+00 -1.30844879e+00 -5.34963012e-01 1.29192740e-01
5.12995005e-01 -4.57447410e-01 -6.52413964e-01 4.47281003e-01] | [14.698742866516113, 6.6308512687683105] |
a709ac4f-340e-469d-8b62-606f06a02336 | fine-grained-image-classification-by | 1512.02665 | null | http://arxiv.org/abs/1512.02665v2 | http://arxiv.org/pdf/1512.02665v2.pdf | Fine-grained Image Classification by Exploring Bipartite-Graph Labels | Given a food image, can a fine-grained object recognition engine tell "which
restaurant which dish" the food belongs to? Such ultra-fine grained image
recognition is the key for many applications like search by images, but it is
very challenging because it needs to discern subtle difference between classes
while dealing with the scarcity of training data. Fortunately, the ultra-fine
granularity naturally brings rich relationships among object classes. This
paper proposes a novel approach to exploit the rich relationships through
bipartite-graph labels (BGL). We show how to model BGL in an overall
convolutional neural networks and the resulting system can be optimized through
back-propagation. We also show that it is computationally efficient in
inference thanks to the bipartite structure. To facilitate the study, we
construct a new food benchmark dataset, which consists of 37,885 food images
collected from 6 restaurants and totally 975 menus. Experimental results on
this new food and three other datasets demonstrates BGL advances previous works
in fine-grained object recognition. An online demo is available at
http://www.f-zhou.com/fg_demo/. | ['Yuanqing Lin', 'Feng Zhou'] | 2015-12-08 | fine-grained-image-classification-by-1 | http://openaccess.thecvf.com/content_cvpr_2016/html/Zhou_Fine-Grained_Image_Classification_CVPR_2016_paper.html | http://openaccess.thecvf.com/content_cvpr_2016/papers/Zhou_Fine-Grained_Image_Classification_CVPR_2016_paper.pdf | cvpr-2016-6 | ['fine-grained-image-recognition'] | ['computer-vision'] | [ 1.91037625e-01 -2.04518288e-01 -3.72894019e-01 -7.78460681e-01
-5.41765630e-01 -5.51232457e-01 1.89486489e-01 6.08376086e-01
-1.93091586e-01 5.63330710e-01 2.67829746e-01 1.61765348e-02
-4.12173241e-01 -1.35277021e+00 -1.19229782e+00 -7.98341870e-01
-2.59667456e-01 4.14530665e-01 -2.29743466e-01 -9.46226418e-02
-8.53812620e-02 2.47328728e-01 -1.62074053e+00 7.71052361e-01
5.98920047e-01 1.27571559e+00 2.36804694e-01 6.12998664e-01
-6.75300732e-02 8.97798240e-01 -1.95538178e-01 -4.70072657e-01
4.63564485e-01 -2.13896677e-01 -7.26138890e-01 2.45965675e-01
7.00046957e-01 -3.49994987e-01 -2.52507180e-01 1.58758986e+00
2.89547116e-01 1.34710640e-01 5.58424413e-01 -1.00465786e+00
-1.36933219e+00 1.20269001e+00 -5.64807951e-01 2.36993685e-01
1.58310726e-01 1.24137238e-01 1.09192038e+00 -5.56532145e-01
1.23750508e-01 1.49068034e+00 6.74206674e-01 1.44683868e-01
-1.23119855e+00 -5.49491465e-01 2.74746150e-01 2.71773785e-01
-1.47694087e+00 -6.33172542e-02 6.22588456e-01 -4.41940635e-01
6.71279788e-01 2.95406550e-01 6.05482101e-01 7.92414069e-01
5.10141067e-02 9.09542263e-01 1.22216535e+00 -3.96402478e-01
5.16506750e-03 -1.35378122e-01 8.48224044e-01 1.13723361e+00
5.85050881e-01 2.83590794e-01 -4.49954867e-01 2.55673498e-01
7.24992216e-01 3.87121677e-01 -1.16876587e-01 -2.68007040e-01
-1.41853535e+00 1.05363524e+00 1.23065436e+00 2.86386132e-01
-4.12909836e-01 5.08162417e-02 1.83473006e-01 4.26423103e-01
5.09706080e-01 2.23756075e-01 -4.36689258e-01 6.01884902e-01
-3.85025680e-01 2.92898137e-02 9.79692757e-01 8.59227061e-01
9.47820783e-01 -1.50685608e-01 -9.21704099e-02 8.25892091e-01
3.48001391e-01 5.43266237e-01 3.31214547e-01 -5.67278624e-01
3.12562853e-01 7.23039627e-01 -3.51366065e-02 -1.39420390e+00
-7.22929120e-01 -3.40198606e-01 -1.29705524e+00 -1.60313681e-01
6.97461665e-01 1.52504295e-01 -1.07242870e+00 1.43128502e+00
3.51730227e-01 -2.54875392e-01 -2.67514050e-01 1.07706308e+00
1.46005023e+00 5.18173277e-01 1.26986191e-01 2.80592740e-01
1.87316597e+00 -1.16706955e+00 -5.22116184e-01 -2.16685936e-01
3.63264769e-01 -6.06385589e-01 9.30751443e-01 2.24415898e-01
-5.61037004e-01 -7.75357008e-01 -1.13993585e+00 -8.80567357e-02
-9.59190786e-01 1.18558742e-01 1.11984324e+00 6.09371424e-01
-7.75814056e-01 3.84752095e-01 -7.17955589e-01 -3.67390543e-01
5.00728846e-01 5.17155170e-01 -3.86580288e-01 -3.84013474e-01
-1.24626756e+00 5.69000244e-01 9.11047220e-01 2.93569028e-01
-5.73642135e-01 -5.56140304e-01 -1.16902518e+00 1.83301196e-01
3.97680730e-01 -7.75749326e-01 1.05018282e+00 -9.41483021e-01
-1.14341259e+00 8.46106708e-01 4.07753736e-01 -3.89888406e-01
6.17937930e-02 5.82869388e-02 -3.65458488e-01 -3.58168408e-02
2.00055197e-01 8.50204527e-01 4.84027088e-01 -8.10410738e-01
-8.93849313e-01 -7.18738854e-01 6.26276314e-01 -1.85570717e-01
-2.55389065e-02 -1.11038290e-01 -1.88162982e-01 -7.40672946e-01
5.45445271e-02 -9.17355299e-01 -2.83868730e-01 -1.40576184e-01
-5.88657856e-01 -3.08711767e-01 9.89421681e-02 -3.89612168e-01
8.63598168e-01 -2.09762931e+00 -3.71229559e-01 1.21445701e-01
1.70594648e-01 -3.68740745e-02 -2.12279975e-01 1.22760542e-01
1.39448158e-02 -1.35885864e-01 -5.63713238e-02 4.75957125e-01
5.11522174e-01 2.90446669e-01 2.09547415e-01 5.36472678e-01
8.28309208e-02 1.10587990e+00 -1.05728412e+00 -3.32391441e-01
4.33854043e-01 5.25007129e-01 -4.64871138e-01 6.12753667e-02
-3.79174292e-01 1.89826712e-01 -5.55246770e-01 1.03869295e+00
7.84206986e-01 -8.09817374e-01 1.78855911e-01 -8.59336495e-01
-4.68832292e-02 -1.31366238e-01 -1.34543049e+00 1.60806060e+00
3.18430364e-02 7.24480227e-02 -1.01664975e-01 -1.24427378e+00
8.97920191e-01 -2.28285223e-01 2.91820049e-01 -9.92411077e-01
2.95914292e-01 -7.88302813e-03 -2.95637269e-02 -3.93149167e-01
4.21107441e-01 -1.09312730e-02 -4.23699051e-01 2.74764538e-01
1.23921610e-01 3.54329199e-01 7.63616145e-01 -2.71239817e-01
6.22217536e-01 1.01097794e-02 5.33576488e-01 -7.02227354e-01
2.07393929e-01 2.19572127e-01 4.85539258e-01 1.04988456e+00
-3.36745918e-01 2.05935255e-01 -3.58344987e-02 -1.06659114e+00
-7.27422953e-01 -9.73806202e-01 -1.24987818e-01 1.49744713e+00
3.81079108e-01 -1.85653627e-01 -5.36429524e-01 -7.01603949e-01
2.73783535e-01 2.03214005e-01 -9.98636484e-01 -1.42065035e-02
-4.92807925e-01 -1.11284161e+00 4.06851113e-01 6.63761556e-01
8.96310925e-01 -1.04235435e+00 -3.50156844e-01 1.53671265e-01
-2.36996830e-01 -7.87843585e-01 -7.68484116e-01 5.47083616e-01
-4.71732140e-01 -1.31093276e+00 -5.80432355e-01 -1.14924467e+00
4.79662120e-01 5.53442180e-01 1.56732702e+00 -7.38623887e-02
-5.77412546e-01 -4.48015630e-02 -4.49310333e-01 -3.07798117e-01
-2.50132173e-01 6.83058947e-02 -1.33960128e-01 1.40183344e-01
1.00963712e+00 -2.04264745e-01 -9.35395300e-01 3.41805816e-01
-9.17390525e-01 1.22310288e-01 6.99113786e-01 9.85290527e-01
1.02250862e+00 3.61802280e-01 4.18520510e-01 -9.76571679e-01
1.37425631e-01 -5.50632834e-01 -8.28938663e-01 4.06685680e-01
-3.68713796e-01 3.17451656e-01 5.46360910e-01 -3.49702895e-01
-8.56758237e-01 3.64548057e-01 -2.43674114e-01 1.52463436e-01
-6.00778162e-01 6.22635365e-01 -1.59235954e-01 -2.43831193e-03
5.13930500e-01 -6.96434900e-02 -4.01984036e-01 -7.73615658e-01
7.04502106e-01 5.87764323e-01 4.31072891e-01 -4.46936280e-01
2.08025217e-01 2.78331071e-01 -1.49806067e-01 -7.08898783e-01
-1.36091721e+00 -6.50763035e-01 -7.26253748e-01 2.71862559e-02
9.77600455e-01 -9.34269428e-01 -1.24210238e+00 2.91743129e-01
-6.77676439e-01 -3.48235697e-01 -5.01654223e-02 7.68999636e-01
-4.06013131e-01 4.69390415e-02 -9.30990219e-01 -2.51743793e-01
-3.93232197e-01 -1.12585950e+00 9.36847746e-01 2.68555224e-01
2.42910802e-01 -8.45327139e-01 -5.39294444e-02 5.60684919e-01
1.97552606e-01 4.18542683e-01 9.76450682e-01 -4.14829522e-01
-6.94716036e-01 1.38023704e-01 -8.62639189e-01 2.23628078e-02
4.18658167e-01 -3.41687083e-01 -6.25900686e-01 -2.93760121e-01
-3.13053846e-01 -4.35851336e-01 1.18124795e+00 6.04483545e-01
1.07836771e+00 -4.28300470e-01 -1.26592726e-01 7.06441581e-01
1.72662115e+00 1.28045544e-01 9.41542014e-02 4.43014234e-01
1.07315886e+00 5.07548034e-01 5.36247313e-01 3.00103396e-01
6.94428742e-01 6.21290267e-01 5.25044501e-01 -2.09161833e-01
-5.52269578e-01 -1.84210539e-01 1.36254787e-01 9.24957335e-01
3.09789218e-02 -1.93255037e-01 -4.31982189e-01 4.06874210e-01
-1.91745961e+00 -8.61523807e-01 -1.43289968e-01 1.70117712e+00
8.63527298e-01 -1.88936830e-01 1.90121174e-01 -5.56365326e-02
8.49071026e-01 1.58269286e-01 -6.54851079e-01 -2.40944758e-01
-1.86249480e-01 9.57959965e-02 8.35483849e-01 2.98072606e-01
-1.60960817e+00 7.21084833e-01 5.61020660e+00 8.43410611e-01
-8.85604799e-01 -4.15825434e-02 7.47966230e-01 2.18530670e-01
1.79885700e-01 -6.75023198e-01 -9.23789740e-01 4.47482795e-01
7.44559288e-01 2.48163491e-01 4.82818842e-01 7.94317782e-01
-1.61171913e-01 1.54846027e-01 -1.19414449e+00 1.00710452e+00
6.41953573e-02 -1.29037809e+00 2.49121994e-01 -2.73068666e-01
6.97949886e-01 3.15991074e-01 -1.47977427e-01 3.12603444e-01
9.30394828e-01 -9.41967845e-01 7.04653680e-01 4.42678541e-01
4.62721974e-01 -7.36338317e-01 5.24847686e-01 3.04036319e-01
-1.58142436e+00 -3.49243842e-02 -8.02396417e-01 -6.11074157e-02
-3.10041189e-01 8.53289545e-01 -6.13742173e-01 6.96561515e-01
1.12493849e+00 6.72215700e-01 -5.50663829e-01 9.65550423e-01
-1.40975684e-01 4.05144244e-01 -3.91581148e-01 -2.90686280e-01
4.65566516e-01 -2.72434145e-01 -1.02473289e-01 1.50827730e+00
1.34449810e-01 1.74756438e-01 7.91600883e-01 7.92832494e-01
-2.59269476e-01 8.17813277e-02 -5.08470356e-01 -1.63030997e-01
-1.56173393e-01 1.28033710e+00 -1.21421123e+00 -2.56386578e-01
-5.23365021e-01 7.48818934e-01 5.14368176e-01 9.66760218e-02
-6.78975999e-01 -1.86543599e-01 4.69003320e-01 -4.11307722e-01
5.54863632e-01 -7.97852501e-02 1.79407433e-01 -1.29388750e+00
-3.12048316e-01 -9.72286224e-01 9.08535779e-01 -4.84612226e-01
-1.81899726e+00 5.57723999e-01 -2.08478764e-01 -7.05214620e-01
1.04841538e-01 -9.13291872e-01 2.70276755e-01 4.10588980e-01
-1.69520414e+00 -1.58969557e+00 -4.72507328e-01 8.08568776e-01
5.96719146e-01 3.12098831e-01 1.00328326e+00 7.60586977e-01
-4.98891175e-01 5.83449900e-01 2.58637905e-01 5.83804846e-01
4.25596237e-01 -1.52072525e+00 2.15438068e-01 4.37147021e-01
5.10982573e-01 3.99436653e-01 5.21594167e-01 -5.24333954e-01
-1.70507407e+00 -1.38060331e+00 7.82513976e-01 -1.90919027e-01
6.45490170e-01 -3.85182232e-01 -7.18792021e-01 6.63901746e-01
9.09518227e-02 2.65914559e-01 9.42518473e-01 2.15565175e-01
-6.36196256e-01 -4.29985464e-01 -1.05078101e+00 2.09749237e-01
1.08375466e+00 -3.08743417e-01 -5.83640754e-01 7.39734232e-01
8.35326195e-01 -2.32117355e-01 -1.15912652e+00 4.52351570e-01
7.57769406e-01 -9.26834762e-01 1.22571504e+00 -6.33530021e-01
1.49672180e-01 -4.60248142e-01 -6.08393669e-01 -1.44175088e+00
-9.16662335e-01 6.43476322e-02 8.09463561e-02 9.50730503e-01
2.46622264e-01 -6.30208671e-01 7.80907512e-01 2.75644928e-01
5.12831844e-02 -4.62782055e-01 -3.26484591e-01 -7.30925441e-01
-1.81978866e-01 -8.95707160e-02 1.12414181e+00 9.82287109e-01
-2.93234795e-01 3.03165376e-01 -3.94898355e-01 3.28465372e-01
9.12514269e-01 1.03367603e+00 4.10401195e-01 -1.13739169e+00
-4.85497802e-01 -3.97094756e-01 -3.48441958e-01 -1.17791390e+00
-1.60552651e-01 -1.19492543e+00 3.50400686e-01 -1.47973299e+00
4.96548951e-01 -4.97257054e-01 -8.22759867e-01 6.89502954e-01
-7.13211969e-02 8.33218336e-01 1.75636306e-01 4.21402603e-02
-8.19289029e-01 1.14633128e-01 1.41875660e+00 -8.42580140e-01
1.34990603e-01 4.85237762e-02 -8.91947746e-01 7.76686490e-01
8.60690534e-01 -3.38801712e-01 -1.14116110e-01 -5.71494937e-01
3.13295484e-01 -1.01574704e-01 3.37197900e-01 -6.22898936e-01
2.00549990e-01 -2.50280231e-01 5.63839078e-01 -6.63888097e-01
9.43155866e-03 -8.66344512e-01 3.45903367e-01 7.06059217e-01
-3.98175657e-01 -8.50541368e-02 1.18994818e-03 5.54384828e-01
-2.71636307e-01 -9.34342295e-03 5.95103741e-01 -5.90697169e-01
-1.03724623e+00 6.88042641e-01 8.67840052e-02 -2.02155724e-01
6.59347475e-01 1.71324924e-01 -5.48438489e-01 2.14284301e-01
-7.55282342e-01 1.64844111e-01 6.75052330e-02 4.38018501e-01
2.72355646e-01 -1.55449879e+00 -8.59062493e-01 4.40669328e-01
3.01291466e-01 -2.73440361e-01 3.08049649e-01 5.85249066e-01
-5.07043004e-01 7.34908581e-01 -5.10272622e-01 -5.89828312e-01
-1.11143577e+00 1.09452665e+00 7.15545416e-02 -3.45963329e-01
-7.01698959e-01 8.33005548e-01 6.61494553e-01 -5.24585605e-01
2.43245438e-01 -7.06257761e-01 -4.01249945e-01 9.70811918e-02
7.88501441e-01 2.32276648e-01 1.96827307e-01 -6.87914550e-01
-3.04671913e-01 5.57843268e-01 -1.16380900e-01 8.49807382e-01
1.44841599e+00 -3.38718951e-01 -1.14221677e-01 2.69588739e-01
1.22957551e+00 -4.91599798e-01 -1.14359725e+00 -3.77418458e-01
3.39756943e-02 -5.13099909e-01 -3.27044129e-02 -9.84666109e-01
-1.34740496e+00 6.18131042e-01 8.55147719e-01 6.55219913e-01
1.08931875e+00 1.19511597e-01 8.19918811e-01 6.92537665e-01
5.40640712e-01 -6.56386971e-01 -1.60014987e-01 4.73577917e-01
6.08278096e-01 -1.61526525e+00 -1.98573217e-01 -5.91115355e-01
9.08370689e-03 1.12134123e+00 2.42163882e-01 -3.63606781e-01
9.31626081e-01 2.27198154e-01 5.12340926e-02 -5.39133370e-01
-2.32172400e-01 -5.20525932e-01 4.67247516e-01 5.36913633e-01
4.09876049e-01 7.25400448e-01 -1.37128085e-01 6.59170449e-01
-2.11525619e-01 2.08751112e-01 2.97889840e-02 6.84201717e-01
-6.69257641e-01 -8.57787132e-01 -3.45775187e-01 5.98835528e-01
-5.19328356e-01 -3.35376501e-01 -1.48396092e-02 5.85772872e-01
5.61265171e-01 1.12825835e+00 -8.26846436e-02 -2.05033094e-01
2.09367886e-01 -3.42405915e-01 7.48771548e-01 -4.91696030e-01
-7.38518715e-01 -2.38113031e-02 1.71481356e-01 -6.51202261e-01
-8.65475118e-01 -2.06739590e-01 -9.84470963e-01 -5.98892450e-01
-3.39316368e-01 9.80779529e-02 4.71236646e-01 6.64716542e-01
1.59664437e-01 5.43733001e-01 4.12928104e-01 -7.01535761e-01
-5.22677600e-01 -7.61810064e-01 -9.19675231e-01 6.13536596e-01
3.34695846e-01 -7.07731187e-01 -9.19750147e-03 3.31889749e-01] | [11.55046558380127, 4.37186336517334] |
5d641f6f-6767-4ae8-ba49-f67c1275d734 | modeling-intra-relation-in-math-word-problems | null | null | https://aclanthology.org/P19-1619 | https://aclanthology.org/P19-1619.pdf | Modeling Intra-Relation in Math Word Problems with Different Functional Multi-Head Attentions | Several deep learning models have been proposed for solving math word problems (MWPs) automatically. Although these models have the ability to capture features without manual efforts, their approaches to capturing features are not specifically designed for MWPs. To utilize the merits of deep learning models with simultaneous consideration of MWPs{'} specific features, we propose a group attention mechanism to extract global features, quantity-related features, quantity-pair features and question-related features in MWPs respectively. The experimental results show that the proposed approach performs significantly better than previous state-of-the-art methods, and boost performance from 66.9{\%} to 69.5{\%} on Math23K with training-test split, from 65.8{\%} to 66.9{\%} on Math23K with 5-fold cross-validation and from 69.2{\%} to 76.1{\%} on MAWPS. | ['Lei Wang', 'Jipeng Zhang', 'Bing Tian Dai', 'Yan Wang', 'Jierui Li', 'Dongxiang Zhang'] | 2019-07-01 | null | null | null | acl-2019-7 | ['math-word-problem-solving', 'math-word-problem-solving', 'math-word-problem-solving'] | ['knowledge-base', 'reasoning', 'time-series'] | [-1.29025981e-01 -1.75424471e-01 -9.94059220e-02 -6.42493308e-01
-1.08164334e+00 -5.09015560e-01 2.09779233e-01 4.37025279e-01
-5.12226045e-01 8.17529023e-01 -1.16181597e-01 -3.80779266e-01
-7.30469346e-01 -1.16680670e+00 -6.74722612e-01 -3.62971783e-01
9.35695600e-03 4.05482471e-01 1.41386569e-01 -3.57593685e-01
7.10288227e-01 4.37774867e-01 -1.38202929e+00 5.20313680e-01
1.23081374e+00 1.16872203e+00 2.22635970e-01 8.15468431e-01
-5.08388281e-01 9.24978197e-01 -9.02795613e-01 -6.47531867e-01
1.33360267e-01 -1.17985278e-01 -9.56519783e-01 -2.32609928e-01
7.45318949e-01 -3.53082538e-01 -5.71065784e-01 9.00345147e-01
5.12721658e-01 5.82401574e-01 8.06753874e-01 -1.16018331e+00
-1.26341558e+00 5.80027282e-01 -6.55328393e-01 6.03505254e-01
2.94755161e-01 1.09616958e-01 1.38290048e+00 -8.42825413e-01
3.03954352e-02 1.00318003e+00 3.68027508e-01 3.19589913e-01
-7.84126520e-01 -8.86723101e-01 8.54831114e-02 4.60744977e-01
-1.28234744e+00 -3.72737423e-02 7.00321257e-01 -4.06743586e-01
1.29149306e+00 3.50558937e-01 1.77670881e-01 6.08722866e-01
1.20728493e-01 9.43834245e-01 1.26142597e+00 -4.18946743e-01
-1.18437730e-01 7.31720254e-02 8.64520848e-01 8.14421177e-01
6.29419088e-02 -4.65159595e-01 -5.18232167e-01 8.32075998e-02
7.01877177e-01 -1.34691805e-01 -2.39701420e-01 4.06166464e-01
-7.20197380e-01 1.07397783e+00 3.51050943e-01 3.19220603e-01
-2.12629005e-01 2.42759049e-01 9.39276908e-03 2.91585833e-01
4.32697892e-01 6.64018989e-01 -8.58292103e-01 -3.05867404e-01
-8.66727710e-01 6.80858552e-01 5.02443433e-01 1.18849337e+00
7.49160767e-01 5.73490635e-02 -4.23322946e-01 1.12150919e+00
7.87734315e-02 3.37060958e-01 6.16732240e-01 -8.24453533e-01
1.01593363e+00 8.78877938e-01 -3.48090008e-02 -1.03944397e+00
-5.16686082e-01 -6.04182720e-01 -6.64824605e-01 -3.44326794e-01
3.30132931e-01 -4.81604859e-02 -9.85753536e-01 1.59298551e+00
2.34827101e-02 -7.75899813e-02 6.21333495e-02 5.75956225e-01
1.48280478e+00 1.12680554e+00 2.15019926e-01 1.72032584e-02
1.25139177e+00 -1.06747019e+00 -5.16944408e-01 -2.71024238e-02
6.70946956e-01 -9.01761413e-01 1.35702825e+00 4.16339219e-01
-1.40092993e+00 -9.24248219e-01 -8.28637123e-01 -3.53978753e-01
-4.83212262e-01 2.28973746e-01 5.94906092e-01 7.52244294e-01
-8.69264185e-01 7.33629286e-01 -1.50450021e-01 8.70585740e-02
5.26836574e-01 6.29728913e-01 -1.05785906e-01 -2.90999591e-01
-1.31281292e+00 6.62378192e-01 2.77191132e-01 -2.13575318e-01
-6.69981062e-01 -1.00805843e+00 -8.82348835e-01 5.15874803e-01
2.27680594e-01 -2.51941592e-01 1.22385037e+00 -4.87737626e-01
-1.38207889e+00 5.63081384e-01 -8.13833177e-02 -8.72299746e-02
9.02268440e-02 -3.38125229e-01 -3.73277932e-01 1.07725322e-01
4.23834585e-02 4.68283683e-01 3.17625731e-01 -7.92678058e-01
-7.34723449e-01 -1.49126321e-01 3.37732166e-01 1.17203340e-01
-6.68107927e-01 9.57084298e-02 -3.63970280e-01 -7.16278493e-01
-2.40325630e-02 -4.98892367e-01 2.62747351e-02 -4.20661718e-01
-2.65417248e-01 -9.03389812e-01 4.01551098e-01 -7.93679357e-01
1.34956360e+00 -1.61709344e+00 -1.59875620e-02 1.71965808e-01
2.71815658e-01 6.95142448e-01 -5.58041096e-01 5.74600518e-01
-1.95759069e-02 8.72069970e-02 4.75803725e-02 -7.02185109e-02
3.23015064e-01 -1.01914033e-01 6.97708922e-03 4.40265238e-02
4.16187078e-01 9.75603819e-01 -6.32556856e-01 -2.82269239e-01
3.12412113e-01 6.11480400e-02 -7.26092577e-01 3.26447159e-01
-2.01371208e-01 -2.23104596e-01 -6.87296569e-01 5.91116250e-01
9.02401388e-01 -3.20031077e-01 -1.60743132e-01 1.81211680e-01
2.95383949e-02 6.33478761e-01 -1.20635211e+00 1.57949054e+00
-5.89991987e-01 5.26273608e-01 -3.25800657e-01 -1.21133363e+00
1.14080667e+00 1.14119478e-01 3.30820709e-01 -9.30113554e-01
1.05932288e-01 -1.69424284e-02 2.02087790e-01 -7.38949835e-01
6.45088494e-01 1.52918755e-03 -3.30595039e-02 2.61178106e-01
4.62718189e-01 -6.70698285e-02 4.13637131e-01 1.72946602e-01
1.23196054e+00 -2.06767485e-01 -8.41494277e-02 -3.93694282e-01
7.71491945e-01 -2.12173447e-01 5.04772127e-01 7.32157826e-01
-7.26306438e-02 6.22781038e-01 5.16789913e-01 -3.63017797e-01
-5.85899174e-01 -8.91435266e-01 -7.99978524e-02 1.37109935e+00
-1.02934912e-01 -6.71520293e-01 -7.04190016e-01 -6.96461260e-01
5.89861721e-02 8.75403047e-01 -4.20005262e-01 -2.53844440e-01
-5.70218205e-01 -8.77034426e-01 4.31618363e-01 7.07554877e-01
7.67156124e-01 -1.13792253e+00 -1.23242572e-01 2.30913401e-01
-3.36545646e-01 -9.48646843e-01 -1.19525254e-01 -6.28022756e-03
-5.73408186e-01 -1.03271139e+00 -8.00394356e-01 -7.72166014e-01
3.93433601e-01 1.64575428e-01 1.34118998e+00 2.66463101e-01
-3.85321170e-01 2.14070886e-01 -5.28055310e-01 -5.59697688e-01
2.12470204e-01 1.94628134e-01 -2.39786893e-01 -3.60788077e-01
5.88258266e-01 -4.03125525e-01 -4.89895523e-01 -1.69815626e-02
-7.47126341e-01 -1.00909479e-01 7.18226135e-01 6.29382014e-01
3.43753934e-01 2.87348032e-01 9.80309129e-01 -8.24062884e-01
6.89628899e-01 -5.70069134e-01 -5.00096977e-01 3.23711306e-01
-4.87242490e-01 -1.82073474e-01 7.78647602e-01 -3.39870155e-01
-8.67432892e-01 -5.60133278e-01 -3.65738630e-01 -1.46836355e-01
-1.79868013e-01 6.38454139e-01 -3.29764158e-01 2.77704950e-02
4.49650586e-01 2.76461631e-01 -6.62603796e-01 -6.84576392e-01
1.65232331e-01 7.32248425e-01 1.12962976e-01 -9.20940638e-01
6.99319959e-01 -4.22981501e-01 -1.74132705e-01 -7.35293388e-01
-1.08791900e+00 -3.23194116e-01 -2.44320303e-01 8.84058252e-02
7.22053349e-01 -7.76721895e-01 -1.02913761e+00 5.46869397e-01
-1.01072836e+00 -1.66185260e-01 -7.36636147e-02 5.97030044e-01
-2.67857164e-01 3.74152243e-01 -9.15011346e-01 -7.52006829e-01
-6.04664683e-01 -1.07434547e+00 7.79151917e-01 6.97094142e-01
-2.45492280e-01 -7.28897154e-01 -2.84156024e-01 9.75059688e-01
4.85167295e-01 -1.62972569e-01 1.52191162e+00 -9.20319676e-01
-6.74783349e-01 -4.49246109e-01 -6.55443609e-01 6.59070075e-01
9.06768516e-02 -5.38836680e-02 -8.76225829e-01 -1.30371181e-02
-2.62397557e-01 -5.37550926e-01 7.37962008e-01 3.93012524e-01
1.76883817e+00 -2.60287374e-01 1.02699861e-01 2.22471386e-01
1.42127347e+00 3.56968045e-01 6.54034257e-01 2.60667622e-01
5.20934105e-01 3.90387118e-01 5.35315514e-01 4.08302516e-01
6.62510872e-01 3.63364190e-01 2.74389684e-01 3.29264879e-01
-5.32871559e-02 7.04282150e-03 1.61890373e-01 1.01591885e+00
-1.29174784e-01 -5.34928620e-01 -9.80753839e-01 7.12966323e-01
-1.51577604e+00 -6.22616351e-01 -4.52345699e-01 1.90738440e+00
9.01717067e-01 1.65995732e-01 -1.73016876e-01 1.93883821e-01
4.67048019e-01 1.97281271e-01 -2.42409464e-02 -6.81381524e-01
1.93180904e-01 1.06477952e+00 1.80931911e-01 2.11111307e-01
-1.01191127e+00 1.03184164e+00 5.27206802e+00 1.28680074e+00
-6.31037951e-01 -1.60465240e-01 6.43923759e-01 -1.28377676e-01
-2.64115393e-01 -3.42969149e-01 -9.47757125e-01 5.08449376e-01
1.04764295e+00 -8.30869675e-02 1.85164049e-01 7.30668545e-01
-1.95020363e-01 -1.70048133e-01 -9.77712989e-01 1.04005826e+00
6.83901981e-02 -1.30782270e+00 3.65239456e-02 -1.72165826e-01
9.39286709e-01 -3.36026937e-01 3.05399746e-01 8.21780741e-01
3.12723517e-01 -1.47033632e+00 2.82933444e-01 6.01585925e-01
5.12493551e-01 -1.10189092e+00 9.09121633e-01 5.30893385e-01
-1.03561389e+00 -7.80254602e-04 -7.07756102e-01 -4.75889802e-01
-3.69847983e-01 7.06439912e-01 -4.19674903e-01 8.26411128e-01
9.11626101e-01 3.11090559e-01 -7.03571558e-01 9.09513950e-01
-2.50175834e-01 7.47027576e-01 -1.00470871e-01 -5.37549257e-01
4.55746740e-01 -9.39165652e-02 -2.03294232e-01 1.13877249e+00
4.31689620e-01 5.90583026e-01 9.97885987e-02 8.38824689e-01
-4.40976620e-01 4.29772735e-01 2.79237088e-02 -3.71517926e-01
3.50383550e-01 1.27886021e+00 -4.03742492e-01 -5.44464529e-01
-5.02937734e-01 7.23004639e-01 6.90842986e-01 2.62550265e-01
-1.08861661e+00 -1.03158414e+00 5.60561776e-01 -2.57365536e-02
3.87814760e-01 -1.78124979e-01 -2.58913755e-01 -1.00238407e+00
1.83212131e-01 -8.16877544e-01 5.75121522e-01 -8.19662988e-01
-1.63493848e+00 3.65986019e-01 6.29052073e-02 -6.33321583e-01
1.95037693e-01 -8.83200526e-01 -7.93482721e-01 1.32116258e+00
-1.50989389e+00 -1.07683539e+00 -4.23010528e-01 4.97594714e-01
7.14919269e-01 -3.82874846e-01 8.36717129e-01 4.05098587e-01
-5.37369609e-01 8.68103504e-01 1.83699146e-01 3.42275530e-01
4.74769890e-01 -1.23116767e+00 2.63361126e-01 4.81141388e-01
2.01765612e-01 6.22663856e-01 2.60726392e-01 -2.54273266e-01
-1.34149694e+00 -1.07678616e+00 1.34810722e+00 -3.11440259e-01
5.11904061e-01 -1.74513116e-01 -1.09545612e+00 5.79823256e-01
3.68383706e-01 -6.86176345e-02 1.01352501e+00 3.21955502e-01
-3.01416308e-01 -1.16526656e-01 -1.30494678e+00 3.40330929e-01
7.33239055e-01 -1.94996893e-01 -8.14570189e-01 5.54586351e-01
6.17718160e-01 -6.58926189e-01 -1.18635952e+00 4.92052883e-01
9.27329734e-02 -6.07522249e-01 1.02970839e+00 -1.01604748e+00
1.03663623e+00 1.79876760e-01 -2.73111999e-01 -1.13295567e+00
-7.05415070e-01 -1.54940486e-01 -1.54214799e-01 1.47935700e+00
3.66139650e-01 -5.99238515e-01 8.41560423e-01 7.46155262e-01
-3.54647577e-01 -1.11566758e+00 -8.18655074e-01 -5.49107671e-01
6.43381000e-01 -5.86409926e-01 6.30504191e-01 1.20940483e+00
-2.56740719e-01 2.86735684e-01 -2.07313642e-01 1.71456963e-01
1.18096411e-01 3.69982272e-01 4.71845210e-01 -1.01512444e+00
-3.03908318e-01 -5.10442913e-01 -4.15495694e-01 -9.00863051e-01
1.93614855e-01 -8.97742510e-01 -4.88596201e-01 -1.91131473e+00
3.37890863e-01 -5.00973523e-01 -6.90501928e-01 5.38149714e-01
-4.06761169e-01 2.00026885e-01 1.69900537e-01 -5.08575022e-01
-5.26426792e-01 6.70425117e-01 1.29168093e+00 -5.52244782e-02
1.00887053e-01 -3.19463387e-03 -9.23354089e-01 5.21440446e-01
1.09713805e+00 -3.34280640e-01 -4.77671176e-01 -6.48264885e-01
8.02207664e-02 -6.95636049e-02 1.88381746e-01 -9.33577538e-01
-7.54241552e-03 -4.93640244e-01 7.57131755e-01 -7.62115061e-01
4.05016214e-01 -2.99575001e-01 -4.98021483e-01 1.60252422e-01
-5.17869890e-01 1.27139360e-01 4.21021432e-01 1.86041310e-01
-1.65775478e-01 -6.66456521e-01 5.98078728e-01 -2.35090613e-01
-6.88105583e-01 3.47270370e-01 -2.89193362e-01 4.25660342e-01
9.74413216e-01 2.25994185e-01 -4.78257895e-01 -2.44647458e-01
-3.92014205e-01 4.71451521e-01 -3.76723140e-01 5.70861757e-01
6.48018062e-01 -1.42269754e+00 -7.91137516e-01 1.17845077e-03
-4.69470769e-02 2.00611129e-01 7.20173895e-01 3.49835753e-01
-6.29904807e-01 7.13622808e-01 -1.23915032e-01 1.38750514e-02
-1.15809619e+00 3.65101635e-01 5.09973988e-02 -5.46976447e-01
-2.69962460e-01 1.43512988e+00 -6.32271357e-03 -6.64589167e-01
3.04538846e-01 -5.98945141e-01 -3.31808507e-01 3.84891755e-03
5.49952328e-01 6.96155548e-01 3.30878109e-01 -2.92591989e-01
-1.88426733e-01 6.59991741e-01 -2.70403743e-01 2.98452944e-01
1.43276370e+00 3.03153217e-01 -1.36191756e-01 1.50752440e-01
1.36073601e+00 -5.57861924e-02 -6.89827442e-01 -3.02377492e-01
-2.16714710e-01 -6.14128947e-01 -3.23146693e-02 -1.19269800e+00
-1.25393403e+00 1.29479074e+00 3.50328863e-01 3.09705967e-03
1.01864243e+00 -2.49131262e-01 1.15247738e+00 5.07907152e-01
7.03602210e-02 -1.30778134e+00 3.29967618e-01 6.30338192e-01
7.20568299e-01 -1.23428047e+00 1.14511587e-01 -5.07109404e-01
-3.31727952e-01 1.18105865e+00 9.41241622e-01 -2.94777483e-01
4.06749249e-01 -2.61847824e-01 -3.26021850e-01 -1.98468849e-01
-7.84501076e-01 -1.10806137e-01 7.05524206e-01 2.63763636e-01
7.47927547e-01 6.91184923e-02 -4.11800623e-01 1.25596142e+00
-4.60071653e-01 -8.32074881e-02 4.19600785e-01 9.59313273e-01
-6.52893662e-01 -9.58622396e-01 -1.19308546e-01 7.62651265e-01
-6.66523814e-01 -3.99304986e-01 -1.82164714e-01 9.50879872e-01
2.84082860e-01 1.13467824e+00 -5.44747077e-02 -3.07068139e-01
3.35215092e-01 3.98639679e-01 6.76104665e-01 -6.80487633e-01
-8.26790452e-01 -2.29273349e-01 4.17149402e-02 -2.97035009e-01
-3.83330397e-02 -2.57919073e-01 -1.30951571e+00 -5.84707797e-01
-3.60453784e-01 2.09913805e-01 3.67560297e-01 1.05063164e+00
1.90516993e-01 9.95640874e-01 3.81775826e-01 -2.17916444e-01
-7.34040558e-01 -1.12964189e+00 -6.51494563e-01 3.41229647e-01
-7.70493448e-02 -4.90350008e-01 -2.22797394e-01 -3.81369174e-01] | [9.813572883605957, 7.50708532333374] |
a3a86349-28f8-4a20-b8e3-d687142feea2 | getting-the-most-out-of-amr-parsing | null | null | https://aclanthology.org/D17-1129 | https://aclanthology.org/D17-1129.pdf | Getting the Most out of AMR Parsing | This paper proposes to tackle the AMR parsing bottleneck by improving two components of an AMR parser: concept identification and alignment. We first build a Bidirectional LSTM based concept identifier that is able to incorporate richer contextual information to learn sparse AMR concept labels. We then extend an HMM-based word-to-concept alignment model with graph distance distortion and a rescoring method during decoding to incorporate the structural information in the AMR graph. We show integrating the two components into an existing AMR parser results in consistently better performance over the state of the art on various datasets. | ['Chuan Wang', 'Nianwen Xue'] | 2017-09-01 | null | null | null | emnlp-2017-9 | ['concept-alignment'] | ['computer-vision'] | [ 5.13813257e-01 5.66318274e-01 -1.81648687e-01 -5.77974916e-01
-1.11932397e+00 -6.83713853e-01 4.79752839e-01 5.97419024e-01
-5.12028098e-01 3.11809599e-01 4.32848871e-01 -6.98471844e-01
3.13987851e-01 -7.50846803e-01 -6.58331752e-01 -1.56497568e-01
3.13798971e-02 8.72924030e-01 9.96316150e-02 -1.32743925e-01
6.80858567e-02 3.84813339e-01 -9.99969482e-01 5.77172220e-01
4.32831764e-01 6.14238322e-01 4.32488173e-01 8.44791830e-01
-7.87087023e-01 1.01441717e+00 -5.22667468e-01 -4.39383686e-01
-1.33920871e-02 -3.94373596e-01 -9.97409821e-01 -8.01031664e-02
3.46449226e-01 7.78058469e-02 -1.99121013e-01 7.97574341e-01
1.97088242e-01 2.98660547e-01 2.75758386e-01 -4.22378927e-01
-6.76840425e-01 1.44341135e+00 -4.01527435e-01 2.99421906e-01
4.72894669e-01 -4.71603423e-01 1.45327866e+00 -7.73753226e-01
7.16534138e-01 1.43643010e+00 5.48934758e-01 7.97151387e-01
-1.28706205e+00 -5.31344235e-01 6.03534281e-01 1.57233924e-01
-1.20799232e+00 -4.93231148e-01 7.87159562e-01 -3.97214331e-02
1.95422626e+00 -4.43788804e-02 6.23776674e-01 1.13312197e+00
5.21529801e-02 6.72050118e-01 7.41837740e-01 -8.36678445e-01
3.46405685e-01 -2.74453372e-01 5.31615674e-01 8.03510725e-01
-9.12927613e-02 -2.41560712e-01 -5.67617893e-01 8.30105681e-04
6.21519268e-01 -3.07171196e-01 5.10277569e-01 -4.30728216e-03
-8.46991181e-01 1.12581336e+00 2.88716227e-01 5.05231440e-01
-1.51694104e-01 3.61760646e-01 4.12783027e-01 2.00942546e-01
5.41293025e-01 4.02193069e-01 -4.80963618e-01 -2.81821936e-01
-7.02502668e-01 -4.11036700e-01 8.72708499e-01 1.10177302e+00
6.65447295e-01 4.02373374e-01 -1.40187163e-02 1.13177264e+00
3.62211764e-01 1.30699664e-01 7.00354338e-01 -9.52698648e-01
7.85980642e-01 4.63478863e-01 -5.90188146e-01 -4.84281540e-01
-4.39849406e-01 -6.49495065e-01 -4.24266160e-01 -4.40481216e-01
-1.15565583e-01 -4.55124304e-03 -1.39632618e+00 1.70018148e+00
1.00316755e-01 2.25665584e-01 2.18574256e-01 3.42592686e-01
8.24605584e-01 7.11654603e-01 4.92563605e-01 -1.49378747e-01
1.20869064e+00 -1.07293713e+00 -9.44253147e-01 -5.87874532e-01
1.20137787e+00 -4.94433731e-01 6.64289057e-01 1.82878345e-01
-1.21256649e+00 -6.62322402e-01 -1.13223791e+00 -2.62725264e-01
-4.01715308e-01 -1.08826891e-01 6.26835227e-01 6.71416163e-01
-1.52979434e+00 4.24546123e-01 -1.00984502e+00 -3.28041285e-01
1.63874581e-01 4.49725538e-01 -5.09823263e-01 -3.62439334e-01
-1.21417701e+00 1.13923788e+00 7.81708658e-01 9.73356217e-02
-6.79681182e-01 -1.44859269e-01 -1.50812078e+00 9.29695461e-03
2.85376728e-01 -7.54687488e-01 1.40564048e+00 -8.71851683e-01
-1.76011586e+00 6.74454868e-01 -5.24167120e-01 -8.44570816e-01
-2.25265890e-01 -2.79285043e-01 -3.81833643e-01 1.28220648e-01
-1.26396433e-01 8.16603720e-01 6.03167772e-01 -1.08275735e+00
-4.86330956e-01 -3.24447095e-01 -5.56789190e-02 4.67284143e-01
-5.49758896e-02 1.29857883e-01 -4.26045388e-01 -7.73688853e-01
4.62753892e-01 -7.31307268e-01 -4.99916285e-01 -1.17434406e+00
-3.29761952e-01 -3.07474524e-01 3.49710524e-01 -9.53380108e-01
1.23617160e+00 -1.76499045e+00 3.44278067e-01 1.13190867e-01
1.62529320e-01 2.55876482e-01 -7.15764403e-01 6.50865734e-01
-2.73313433e-01 3.06321025e-01 -3.42287838e-01 -9.32142377e-01
-2.22150132e-01 5.32475889e-01 -3.33893955e-01 1.25110745e-01
4.80769753e-01 1.05584705e+00 -8.28685760e-01 -4.18346107e-01
2.87425607e-01 5.60991824e-01 -2.89504230e-01 2.69700259e-01
-3.89977127e-01 2.81101733e-01 -3.07229400e-01 5.72299719e-01
2.37717494e-01 -2.21252665e-02 6.27955317e-01 3.80087823e-01
1.29633635e-01 9.73097384e-01 -7.96460092e-01 2.24869704e+00
-7.09854007e-01 5.09350836e-01 -3.26060392e-02 -1.20987105e+00
1.22958171e+00 5.60199082e-01 1.99481010e-01 -7.27982461e-01
3.60245965e-02 1.49433926e-01 2.86007434e-01 2.30047032e-01
7.02285886e-01 -1.36411995e-01 -4.87278700e-02 3.10030341e-01
4.64195579e-01 -3.82924005e-02 9.59320366e-02 5.14480174e-01
1.48730576e+00 4.17761385e-01 4.82421219e-01 3.89772765e-02
6.29086971e-01 -7.30437366e-03 5.05940378e-01 9.44988132e-01
3.28530222e-01 4.56844836e-01 6.96110949e-02 -3.41122448e-01
-9.15024698e-01 -1.03295982e+00 8.47362820e-03 1.36967516e+00
-5.88990033e-01 -7.64516413e-01 -7.51589835e-01 -8.21504116e-01
-4.28055316e-01 1.25137115e+00 -4.60510045e-01 3.44013833e-02
-1.25421727e+00 -5.61381102e-01 5.01547098e-01 1.06587660e+00
3.70196439e-03 -1.11686301e+00 -8.77200589e-02 7.79574692e-01
-1.48154289e-01 -1.54882729e+00 -6.10729568e-02 6.84037447e-01
-1.20977640e+00 -4.34575766e-01 -7.03726560e-02 -9.99974370e-01
4.53513771e-01 -1.71244994e-01 1.43712378e+00 6.02845363e-02
-6.63493648e-02 3.68714690e-01 -8.55593085e-01 -1.65568396e-01
-8.81347835e-01 4.64511812e-01 -3.36395562e-01 -5.83547533e-01
6.99809790e-01 -6.00603759e-01 1.80186108e-01 -2.10398063e-01
-6.34796679e-01 1.57658145e-01 6.60230815e-01 6.45424247e-01
6.54065251e-01 -3.85140449e-01 5.34954965e-01 -1.21065867e+00
2.79648960e-01 -3.17523181e-01 -4.17837799e-01 9.67668965e-02
-4.85717386e-01 3.82104874e-01 6.30414367e-01 -3.14080149e-01
-9.94931042e-01 4.81634676e-01 -7.50386357e-01 -4.61046677e-03
-3.25188547e-01 6.22077703e-01 -1.97436154e-01 2.86220551e-01
2.16288082e-02 2.17876047e-01 -6.21801019e-01 -7.39683926e-01
9.71144736e-01 5.47910273e-01 4.58686799e-01 -6.44437611e-01
5.76402128e-01 4.08187285e-02 2.75829323e-02 -8.73564661e-01
-9.56626773e-01 -6.51307523e-01 -9.32362676e-01 2.72002667e-01
1.08237410e+00 -1.00922859e+00 -1.98560134e-01 -5.62664270e-02
-1.56901634e+00 -3.23267967e-01 -3.78382146e-01 4.52319771e-01
-4.79282707e-01 3.48031223e-01 -1.17259181e+00 -8.62166286e-01
-5.84302545e-01 -9.01415706e-01 9.62301075e-01 -2.10123673e-01
-1.57732397e-01 -1.14879489e+00 3.71459275e-01 3.10358822e-01
3.40514511e-01 -1.14977919e-01 1.15202379e+00 -1.41653073e+00
-4.07250375e-01 -1.39297545e-01 -1.44130345e-02 2.37283871e-01
-6.79097846e-02 -3.28171700e-01 -9.63253140e-01 -3.10826540e-01
9.73764136e-02 -4.13404331e-02 1.35459507e+00 4.07970637e-01
6.99784517e-01 -2.30824128e-01 -4.51649040e-01 4.54416603e-01
1.33012545e+00 4.89377230e-01 6.14951849e-01 3.14132422e-01
1.20215082e+00 6.16350412e-01 3.12851667e-01 -2.57060051e-01
6.60806298e-01 7.01716721e-01 2.40076989e-01 -4.55486849e-02
-2.06335336e-01 -4.46951896e-01 6.44101501e-01 1.77657211e+00
2.30046004e-01 -4.65765744e-01 -1.10790110e+00 5.73032379e-01
-1.83361208e+00 -5.11586964e-01 -4.34701592e-02 1.98587298e+00
6.95076108e-01 3.21655393e-01 -1.12441488e-01 -1.12436622e-01
7.45521009e-01 1.98146984e-01 -7.12384954e-02 -1.15744781e+00
3.87612060e-02 7.91911542e-01 6.65961981e-01 8.86175334e-01
-9.87384021e-01 1.65119135e+00 7.22245312e+00 5.87159991e-01
-4.30050343e-01 4.41553116e-01 2.97647893e-01 2.18489125e-01
-4.16628450e-01 3.02177668e-01 -1.11977351e+00 -1.33517519e-01
1.59573817e+00 3.97286773e-01 4.20641273e-01 7.89863050e-01
-1.68743908e-01 1.43844157e-01 -1.32491934e+00 5.90833306e-01
3.02692145e-01 -1.13182700e+00 4.14466083e-01 -2.07132846e-02
1.98364452e-01 3.15547615e-01 -3.25640619e-01 3.95497769e-01
9.66235280e-01 -1.22147655e+00 5.02223909e-01 1.44720033e-01
4.68754560e-01 -7.54842877e-01 8.39907527e-01 3.73818949e-02
-1.77354300e+00 -4.64941747e-02 -5.55897832e-01 -2.41064429e-01
4.10365611e-01 2.21949622e-01 -1.05738342e+00 7.96434402e-01
2.57948041e-01 8.26927722e-01 -7.09613860e-01 4.31351066e-01
-6.32795155e-01 8.55255604e-01 1.17543358e-02 1.41457304e-01
3.72390836e-01 -2.50369012e-01 7.39571750e-01 1.64482987e+00
1.79724656e-02 -9.66837853e-02 2.87474662e-01 6.35842144e-01
-1.17272757e-01 2.64528692e-01 -6.04479194e-01 -2.33727783e-01
6.87454283e-01 1.15116334e+00 -7.83819675e-01 -5.34746885e-01
-6.15976512e-01 9.24793124e-01 7.99084783e-01 1.92813843e-01
-3.87939990e-01 -6.04269542e-02 3.85816187e-01 -1.93377346e-01
5.57740450e-01 -6.90996766e-01 -1.41501933e-01 -1.05633092e+00
-1.98355839e-01 -7.53534317e-01 8.61119390e-01 -5.87755680e-01
-1.16369152e+00 9.05773282e-01 -4.62226123e-02 -4.39624995e-01
-8.47891986e-01 -6.20623887e-01 -4.59127128e-01 8.83270502e-01
-1.44040120e+00 -1.47656429e+00 4.01632488e-01 1.57431036e-01
9.36432540e-01 -1.23254932e-01 1.27874351e+00 -4.13998999e-02
-4.92887408e-01 3.23305547e-01 -2.50267595e-01 4.70164984e-01
1.85501099e-01 -1.73481929e+00 1.51070642e+00 1.21886528e+00
9.22963619e-01 8.72335374e-01 3.24651390e-01 -8.67447734e-01
-1.29561138e+00 -1.14237344e+00 9.65995133e-01 -6.92233384e-01
7.42426634e-01 -8.61733794e-01 -1.06267476e+00 1.38986683e+00
3.52741569e-01 -3.01999122e-01 7.63884187e-01 4.95834410e-01
-6.27330184e-01 2.00741082e-01 -6.69668913e-01 2.52789229e-01
1.20549488e+00 -6.61006927e-01 -1.24052405e+00 5.77301122e-02
1.41466916e+00 -3.73675048e-01 -8.26533854e-01 3.33069593e-01
2.07613349e-01 -3.90141994e-01 8.77226293e-01 -7.66018748e-01
1.24659881e-01 -3.69278900e-02 -3.89688253e-01 -1.28323364e+00
-4.24465925e-01 -5.64162493e-01 -3.36305082e-01 1.42244995e+00
7.51682520e-01 -5.97644925e-01 5.96693873e-01 3.32430482e-01
-4.90235597e-01 -4.01935846e-01 -1.16896749e+00 -7.83858359e-01
1.86673269e-01 -7.46869981e-01 2.07920805e-01 8.07854116e-01
5.49206845e-02 9.94687676e-01 -1.77096307e-01 9.83358398e-02
4.15951103e-01 -1.66845888e-01 1.56853110e-01 -1.14673519e+00
-4.81624812e-01 -1.47316352e-01 -4.82797354e-01 -1.02203000e+00
4.64299947e-01 -1.21386063e+00 1.32256985e-01 -2.13235879e+00
1.84764996e-01 -1.61796421e-01 -5.17226458e-01 7.40733981e-01
-4.25658897e-02 -1.64131336e-02 2.18277827e-01 -1.31653324e-01
-7.20802307e-01 2.86414176e-01 4.59210485e-01 -1.07432436e-02
-4.26758528e-01 -4.93526727e-01 -3.95486176e-01 6.99351668e-01
7.86289334e-01 -9.74944472e-01 -1.73913956e-01 -8.66944015e-01
2.44900584e-01 1.40217945e-01 -2.88951427e-01 -8.95102262e-01
2.26242185e-01 3.97596419e-01 3.15016508e-01 -6.53334141e-01
4.91894513e-01 -5.51093936e-01 -1.84116699e-02 3.37553293e-01
-6.35646284e-01 6.76803172e-01 3.53961647e-01 7.41324544e-01
-1.63305596e-01 -3.96241814e-01 5.18405139e-01 -4.22167808e-01
-5.51321268e-01 -9.77144986e-02 -6.83914125e-01 1.41838998e-01
3.98292422e-01 -1.39129281e-01 -3.32616091e-01 -3.22319627e-01
-8.53722632e-01 -1.44639788e-02 2.85369754e-01 5.99724889e-01
7.20348120e-01 -1.04340601e+00 -7.30180264e-01 1.16909496e-01
-3.66460979e-02 -6.58747628e-02 -2.45918617e-01 3.22375476e-01
-2.55793512e-01 6.56906128e-01 1.71857655e-01 -4.27548915e-01
-1.33319819e+00 6.49953187e-01 1.54211238e-01 -5.94637513e-01
-8.54951084e-01 7.67716944e-01 -7.51310214e-02 -4.56970930e-01
2.95352489e-01 -3.52280527e-01 -5.60592830e-01 -9.23435092e-02
2.78931320e-01 -1.66254619e-03 2.83306509e-01 -8.81523967e-01
-5.12611568e-01 5.11026025e-01 -3.45415503e-01 -5.03553212e-01
1.44274557e+00 -2.56650954e-01 -1.24122903e-01 6.29790604e-01
9.92604434e-01 -1.12728272e-02 -5.26211083e-01 -3.53199065e-01
6.24798477e-01 9.65789258e-02 1.16844259e-01 -7.47703671e-01
-7.40782619e-01 1.04665923e+00 3.72705191e-01 1.10701494e-01
8.44985247e-01 1.52325287e-01 9.38343108e-01 6.36840940e-01
1.42893493e-01 -1.16205812e+00 5.28910644e-02 9.09813583e-01
3.31029296e-01 -6.55854821e-01 -1.95133269e-01 -4.97986466e-01
-6.73941195e-01 1.22679436e+00 2.68273771e-01 -5.91147318e-02
2.85949916e-01 4.32356268e-01 7.29507729e-02 -7.28119463e-02
-9.71053660e-01 -4.83123749e-01 3.05438787e-01 7.08835542e-01
6.37491822e-01 -1.67262193e-03 -4.34997201e-01 5.12713194e-01
-1.61290333e-01 -6.40422881e-01 6.01932704e-01 8.70544851e-01
-4.35572267e-01 -1.71804965e+00 2.05237180e-01 2.94367343e-01
-7.79843390e-01 -7.62557924e-01 -4.98451233e-01 6.88560009e-01
-3.50703895e-01 1.18335831e+00 3.76556218e-01 -4.32863832e-01
1.81535468e-01 5.62611043e-01 4.16499823e-01 -1.37874079e+00
-7.30171740e-01 2.16999710e-01 6.60361171e-01 -6.34355187e-01
-5.08679807e-01 -4.96763438e-01 -1.39645934e+00 3.90472203e-01
-4.11147863e-01 1.76219940e-01 9.95131373e-01 1.29750872e+00
3.20695788e-01 7.33323693e-01 1.95016637e-01 -3.85216385e-01
-1.71111748e-01 -1.17319560e+00 -2.95778692e-01 -7.06230327e-02
6.31085411e-02 -2.71976560e-01 -1.68174013e-01 -7.60837421e-02] | [10.437275886535645, 9.21379566192627] |
26daa574-8cef-4797-a07e-2474c33f8527 | collect-and-distribute-transformer-for-3d | 2306.01257 | null | https://arxiv.org/abs/2306.01257v1 | https://arxiv.org/pdf/2306.01257v1.pdf | Collect-and-Distribute Transformer for 3D Point Cloud Analysis | Although remarkable advancements have been made recently in point cloud analysis through the exploration of transformer architecture, it remains challenging to effectively learn local and global structures within point clouds. In this paper, we propose a new transformer architecture equipped with a collect-and-distribute mechanism to communicate short- and long-range contexts of point clouds, which we refer to as CDFormer. Specifically, we first utilize self-attention to capture short-range interactions within each local patch, and the updated local features are then collected into a set of proxy reference points from which we can extract long-range contexts. Afterward, we distribute the learned long-range contexts back to local points via cross-attention. To address the position clues for short- and long-range contexts, we also introduce context-aware position encoding to facilitate position-aware communications between points. We perform experiments on four popular point cloud datasets, namely ModelNet40, ScanObjectNN, S3DIS, and ShapeNetPart, for classification and segmentation. Results show the effectiveness of the proposed CDFormer, delivering several new state-of-the-art performances on point cloud classification and segmentation tasks. The code is available at \url{https://github.com/haibo-qiu/CDFormer}. | ['DaCheng Tao', 'Baosheng Yu', 'Haibo Qiu'] | 2023-06-02 | null | null | null | null | ['point-cloud-classification'] | ['computer-vision'] | [-2.59491354e-01 -3.86873126e-01 3.04635037e-02 -6.20933890e-01
-8.21886420e-01 -5.98606706e-01 5.23080468e-01 2.45751992e-01
5.02346493e-02 4.15574908e-01 -2.18157724e-01 -1.51646465e-01
-3.42575043e-01 -9.82121766e-01 -1.07187939e+00 -6.37660384e-01
-2.03727573e-01 6.66898072e-01 5.34664750e-01 -3.13392989e-02
3.74745876e-01 1.01461196e+00 -1.72549677e+00 2.11374760e-01
8.43555272e-01 1.31638360e+00 5.16955853e-01 3.51750523e-01
-4.33851361e-01 1.56522378e-01 -4.74403113e-01 -8.32369272e-03
2.05464855e-01 3.52259666e-01 -7.17307925e-01 -2.31286753e-02
5.58366001e-01 -2.78384894e-01 -2.31256410e-01 8.08416426e-01
4.85889912e-01 4.25889157e-02 3.00779730e-01 -1.01516664e+00
-5.16860843e-01 2.79175699e-01 -7.57778525e-01 3.73325229e-01
1.22791894e-01 3.33646208e-01 1.05109012e+00 -1.16053152e+00
4.56947833e-01 1.13346100e+00 6.22213542e-01 1.30766213e-01
-8.24462056e-01 -1.02373457e+00 5.44121504e-01 1.79628253e-01
-1.47708464e+00 -1.40211016e-01 9.43098724e-01 -3.18345159e-01
9.15997505e-01 3.04951072e-01 8.04614246e-01 5.78738749e-01
-1.37114003e-02 7.52632260e-01 6.97695017e-01 1.30407095e-01
1.49346560e-01 -3.68580282e-01 1.66058779e-01 4.72791851e-01
-4.27868143e-02 1.36773884e-01 -3.95310283e-01 -3.40358675e-01
1.04039645e+00 4.77983981e-01 -2.51105636e-01 -3.16481948e-01
-1.22333598e+00 6.63870573e-01 1.10059929e+00 2.56387800e-01
-4.79817510e-01 3.13159138e-01 5.91981933e-02 -8.13796297e-02
6.44777834e-01 1.25033498e-01 -5.08249104e-01 4.35995273e-02
-7.68953621e-01 3.56338292e-01 3.61506402e-01 1.39989161e+00
1.31120896e+00 -4.18545723e-01 -6.93456605e-02 8.41039181e-01
3.98767978e-01 7.51133800e-01 4.79564928e-02 -7.58449316e-01
5.72969794e-01 7.14456141e-01 -7.61451647e-02 -1.04771733e+00
-2.64011115e-01 -6.91734493e-01 -5.97599566e-01 -6.63499087e-02
-9.02384520e-02 2.49747798e-01 -9.01861370e-01 1.21512854e+00
7.03300834e-01 8.56559753e-01 -4.47730452e-01 9.96322036e-01
8.79516006e-01 8.63396645e-01 -2.22603213e-02 3.39965940e-01
1.22238386e+00 -8.59337986e-01 9.19779111e-03 -1.93429694e-01
1.41932890e-01 -6.91792846e-01 1.10950720e+00 7.58814216e-02
-9.27858293e-01 -5.82342446e-01 -8.20585072e-01 -2.21408382e-01
-2.88022608e-01 -1.18710794e-01 5.61214685e-01 -9.70431268e-02
-8.30870628e-01 6.49686396e-01 -1.03422594e+00 -6.70771375e-02
8.84286225e-01 4.73874927e-01 -8.01933110e-02 -1.76880240e-01
-7.39480555e-01 -8.73088390e-02 6.52589649e-02 3.27731937e-01
-7.48979270e-01 -1.16297770e+00 -5.38613498e-01 3.30248512e-02
3.00347716e-01 -7.04558134e-01 1.15302718e+00 -4.63048071e-01
-1.00159335e+00 7.70738304e-01 -3.43818188e-01 1.95243005e-02
1.40534177e-01 -2.76091248e-01 -1.21428691e-01 1.76675811e-01
4.36791033e-01 7.56864786e-01 5.32609761e-01 -1.71111226e+00
-1.00558031e+00 -6.72379613e-01 6.00904524e-02 1.60759717e-01
4.59323823e-02 -1.70412421e-01 -1.03037632e+00 -5.26764274e-01
5.48603475e-01 -9.18165982e-01 -1.32967904e-01 1.59175634e-01
-5.04420996e-01 -4.44965839e-01 1.17124414e+00 -2.04015151e-01
1.03109491e+00 -2.28642559e+00 -1.05195843e-01 4.68279868e-01
4.02894050e-01 1.54713973e-01 -5.83328269e-02 2.72066236e-01
-4.28743660e-02 4.63394672e-02 -3.82115424e-01 -6.16458416e-01
-1.09374020e-02 4.68748540e-01 -5.29529929e-01 4.04804230e-01
1.37922317e-01 9.42122817e-01 -9.04632270e-01 -3.41275424e-01
5.26239216e-01 5.40211201e-01 -6.02096736e-01 2.09092706e-01
-4.36194390e-01 6.60895646e-01 -9.37161863e-01 1.08996391e+00
1.12676990e+00 -4.46815819e-01 -4.84902978e-01 -1.86814561e-01
-2.49569282e-01 2.17889875e-01 -8.41809690e-01 1.80533659e+00
-4.53161150e-01 3.54165077e-01 6.53006360e-02 -5.06446660e-01
9.32135284e-01 -1.61590785e-01 7.20902205e-01 -5.60425043e-01
-4.95100729e-02 2.54632771e-01 -4.89932030e-01 -1.86594665e-01
4.54463333e-01 1.24110363e-01 1.05415694e-01 9.43283886e-02
-2.98380315e-01 -2.61968911e-01 -2.94370145e-01 8.55632126e-02
9.98966098e-01 1.39088139e-01 -1.24145597e-01 -1.02099806e-01
5.13457000e-01 4.87625748e-02 4.90335703e-01 5.16075075e-01
5.70852384e-02 9.86543834e-01 1.53515577e-01 -4.59499896e-01
-7.46349752e-01 -1.13732874e+00 -3.83376896e-01 9.10605609e-01
6.20784581e-01 -5.31037271e-01 -3.47869992e-01 -6.70186043e-01
4.64899123e-01 3.54867369e-01 -3.90076607e-01 1.79719925e-01
-6.62634969e-01 -4.25194710e-01 1.54825717e-01 7.18941629e-01
5.09164870e-01 -9.90875721e-01 -5.00652611e-01 2.00592741e-01
-7.03708222e-03 -1.10118639e+00 -5.09090483e-01 -2.71785837e-02
-1.12967634e+00 -8.72971714e-01 -5.47327101e-01 -5.59447765e-01
6.28390908e-01 5.72059035e-01 1.25919294e+00 3.94159943e-01
1.55941248e-02 3.75857621e-01 -3.96403760e-01 -4.47828770e-01
5.51427543e-01 2.56643772e-01 -3.97691905e-01 -1.40613746e-02
3.58084768e-01 -9.48037684e-01 -9.26185727e-01 6.54018939e-01
-6.53524935e-01 -2.03276664e-01 5.45841396e-01 5.92107177e-01
1.33227396e+00 -2.21975639e-01 1.83546066e-01 -7.51408279e-01
1.20705560e-01 -7.84387767e-01 -7.29313195e-01 -8.04278776e-02
-2.15952203e-01 -3.66808921e-01 2.88715780e-01 3.22647355e-02
-5.89087069e-01 8.50940309e-03 -3.94948542e-01 -1.06357157e+00
-3.66069645e-01 3.53206128e-01 -4.61287171e-01 -3.18384409e-01
1.36363804e-01 1.97073564e-01 -4.81924444e-01 -7.69344151e-01
2.65482366e-01 6.94020748e-01 5.58185697e-01 -1.05414867e+00
8.63002181e-01 7.83099473e-01 -1.62493840e-01 -7.94621229e-01
-8.12703490e-01 -7.28697002e-01 -6.38632059e-01 -1.19947523e-01
7.89200306e-01 -9.29054976e-01 -7.08238602e-01 4.26768035e-01
-1.21966720e+00 -2.98853099e-01 -4.76869196e-01 5.13167977e-02
-3.52104425e-01 2.02678517e-01 -4.30864751e-01 -4.76825863e-01
-3.81403714e-01 -1.37058449e+00 1.75305724e+00 3.21188539e-01
3.41444224e-01 -8.36645901e-01 -8.90773162e-02 2.29626596e-01
2.20493063e-01 2.82137960e-01 6.79615378e-01 -3.98867577e-01
-1.43275952e+00 -5.24957329e-02 -4.97655809e-01 3.61350104e-02
-8.53572693e-03 4.36277352e-02 -1.00257027e+00 -3.02499026e-01
-1.87604159e-01 5.59882075e-02 6.52085364e-01 3.40634316e-01
1.85452640e+00 4.17460352e-02 -8.00684631e-01 1.23662972e+00
1.35220397e+00 6.26556501e-02 5.56007504e-01 2.69674599e-01
1.07341969e+00 2.87263215e-01 8.15292835e-01 5.62150836e-01
7.85022020e-01 6.97796941e-01 8.26381266e-01 -2.83056237e-02
-3.47190909e-02 -3.59095305e-01 -3.80503565e-01 8.69314492e-01
-2.49938965e-01 -1.94153160e-01 -1.16309655e+00 5.58194160e-01
-1.85597110e+00 -6.16370320e-01 -1.45453975e-01 1.95994616e+00
3.76671940e-01 5.57464175e-02 -1.71620622e-01 -3.79697829e-01
6.17320895e-01 3.80502194e-01 -7.52788365e-01 3.14276189e-01
3.35611850e-02 3.19112957e-01 5.65046370e-01 3.24866533e-01
-1.20242608e+00 9.87429321e-01 4.88311481e+00 9.87586617e-01
-1.44578946e+00 1.56475246e-01 5.88348210e-01 -2.26466045e-01
-4.43141609e-01 5.94393536e-02 -8.20350885e-01 7.39622712e-01
5.15713096e-01 1.95340902e-01 1.69780478e-01 9.97788608e-01
3.24551649e-02 1.05436064e-01 -8.89580369e-01 1.18032193e+00
-2.11424276e-01 -1.48505425e+00 -8.13282207e-02 2.19450891e-01
6.63902402e-01 7.10549474e-01 3.08473408e-02 1.00258209e-01
-1.72924455e-02 -6.38442874e-01 8.07569861e-01 5.20008206e-01
8.37836623e-01 -8.28681111e-01 5.37010550e-01 2.45484591e-01
-1.72035980e+00 4.18248251e-02 -6.36234164e-01 3.06351155e-01
4.08738926e-02 8.55190158e-01 -5.34658492e-01 9.73448336e-01
9.97818589e-01 1.10617781e+00 -5.38547158e-01 1.28216803e+00
-1.48830384e-01 4.51569885e-01 -7.25765824e-01 2.62608737e-01
3.46231759e-01 -2.55834103e-01 6.50304317e-01 9.25197601e-01
5.52957654e-01 2.97890365e-01 3.25904965e-01 1.01701713e+00
-3.03847436e-02 -1.75219193e-01 -4.22334731e-01 3.86900187e-01
9.17948663e-01 1.22123158e+00 -8.54275107e-01 -2.70656437e-01
-3.76673520e-01 5.82882524e-01 3.74155372e-01 3.05342644e-01
-9.09649730e-01 -2.89554298e-01 1.02126217e+00 4.46869075e-01
6.05567396e-01 -4.23352689e-01 -3.87824535e-01 -1.07243216e+00
2.46132597e-01 -4.61090446e-01 2.35569239e-01 -9.70335901e-01
-1.42633522e+00 7.25652337e-01 2.26582587e-01 -1.46652365e+00
3.72651368e-01 -3.66557896e-01 -8.41779888e-01 8.32046092e-01
-1.67668962e+00 -1.40658092e+00 -6.96418643e-01 6.55749500e-01
5.03892243e-01 1.11556359e-01 2.44389772e-01 5.89929640e-01
-3.54482353e-01 3.46397281e-01 7.91624114e-02 -2.10434068e-02
4.00716186e-01 -1.14288473e+00 6.34004414e-01 4.93104994e-01
1.57299593e-01 6.80499196e-01 2.29254022e-01 -7.90139616e-01
-1.43118119e+00 -1.32037973e+00 3.67272854e-01 -5.12724221e-01
4.22708333e-01 -5.26208103e-01 -1.16251433e+00 7.72173762e-01
-2.10495889e-01 5.25781929e-01 2.76941240e-01 4.61908840e-02
-1.78544894e-01 -3.47905636e-01 -1.01560307e+00 2.23295778e-01
1.43815196e+00 -5.79057157e-01 -4.34777856e-01 4.30353314e-01
9.55812514e-01 -7.51150966e-01 -7.94577539e-01 5.52141845e-01
2.04889789e-01 -9.96967196e-01 1.23674774e+00 -6.30990267e-02
3.45652729e-01 -5.52872598e-01 -3.46333176e-01 -1.14528883e+00
-3.70800704e-01 -2.99519837e-01 9.20256972e-02 1.29069316e+00
2.20445395e-01 -8.09962630e-01 1.04196131e+00 3.19249064e-01
-6.77773058e-01 -1.23801601e+00 -1.06941938e+00 -4.83081222e-01
1.33493301e-02 -6.79789245e-01 1.35675168e+00 8.71792197e-01
-6.30765975e-01 4.88565415e-02 1.88110262e-01 6.61200762e-01
6.85938418e-01 7.88155377e-01 9.14322138e-01 -1.17969859e+00
-9.96187236e-03 -3.96801591e-01 -5.51225841e-01 -1.64629447e+00
5.21085709e-02 -1.00447619e+00 -1.77416969e-02 -1.56894505e+00
-1.85875878e-01 -1.13258219e+00 -2.21618891e-01 4.20568347e-01
-1.20368719e-01 1.91411301e-01 2.79215038e-01 6.43365800e-01
-6.78682864e-01 7.65704870e-01 1.22341132e+00 -2.54769288e-02
-7.81265721e-02 2.27142155e-01 -4.44480568e-01 5.84075093e-01
6.06732488e-01 -4.41832513e-01 -2.86596358e-01 -8.54319990e-01
-6.03500009e-03 -8.84100050e-02 6.10771894e-01 -1.28898370e+00
4.46921378e-01 -1.05097823e-01 4.53168064e-01 -1.39165771e+00
5.65347612e-01 -1.05007088e+00 2.84219682e-01 -1.42715663e-01
1.48570567e-01 2.90792704e-01 2.23141566e-01 7.01977491e-01
-3.63866925e-01 1.51827365e-01 5.86780846e-01 -1.28498390e-01
-7.90456176e-01 9.74980891e-01 4.98432279e-01 -1.68423191e-01
1.16086090e+00 -2.16510981e-01 -3.80184561e-01 -1.81630533e-02
-6.67685688e-01 6.31911099e-01 7.00321555e-01 3.62675130e-01
8.21395755e-01 -1.31346512e+00 -3.88214588e-01 5.05592704e-01
3.34937006e-01 8.79288435e-01 6.36707425e-01 6.97331846e-01
-6.34145200e-01 2.34024256e-01 3.26998495e-02 -1.33193469e+00
-9.90487516e-01 4.12567139e-01 3.00732583e-01 1.67143822e-01
-9.62405622e-01 1.10866535e+00 4.77811873e-01 -5.95511258e-01
6.04399703e-02 -7.20802903e-01 1.65512264e-01 -1.26084983e-01
2.39380375e-01 1.57464266e-01 2.57282495e-01 -7.29053259e-01
-5.92922211e-01 1.16338253e+00 -9.92988981e-03 3.26008558e-01
1.44241107e+00 -1.68602139e-01 -2.42227331e-01 2.39845455e-01
1.24616289e+00 1.39698312e-01 -1.48108304e+00 -3.51296127e-01
-1.15556955e-01 -9.28059042e-01 1.29472390e-01 -4.94971067e-01
-1.53347886e+00 8.86991858e-01 5.74542403e-01 4.56028767e-02
1.11315155e+00 5.16875684e-01 1.00369859e+00 1.49099723e-01
6.79712474e-01 -4.59302247e-01 -2.66705900e-01 5.86747468e-01
9.32937920e-01 -1.13336837e+00 -8.26663524e-02 -6.51802480e-01
-3.09468836e-01 8.25327873e-01 6.53567791e-01 -3.36532891e-01
1.05455494e+00 3.13347220e-01 -5.81109449e-02 -6.84318304e-01
-6.58534646e-01 -1.58738807e-01 2.29574725e-01 4.66123223e-01
6.75846785e-02 1.70166597e-01 3.26993555e-01 5.50176561e-01
-3.02967012e-01 -2.24043429e-02 -1.22495435e-01 1.11172068e+00
-4.52334613e-01 -1.00870097e+00 -4.94090915e-01 5.59452355e-01
-3.85621563e-03 1.01184137e-01 -1.11650519e-01 6.37120962e-01
3.53059798e-01 5.45575082e-01 4.90815878e-01 -5.09835780e-01
6.51886940e-01 -3.29665482e-01 8.28602761e-02 -7.50129104e-01
-6.21585846e-01 2.79973358e-01 -3.26351732e-01 -9.08948004e-01
-3.06046039e-01 -6.68754637e-01 -1.47049272e+00 -4.12948579e-01
-1.33972496e-01 1.74558640e-01 6.96788549e-01 6.58607721e-01
8.90756786e-01 6.67360485e-01 6.57583296e-01 -1.38406980e+00
-1.80589408e-01 -6.16140366e-01 -4.38911200e-01 1.58528373e-01
4.92024451e-01 -7.84755349e-01 -2.88842738e-01 -3.94543767e-01] | [7.950525283813477, -3.403618812561035] |
f78aa6be-3d92-4533-a5aa-eb59f9a1d578 | learning-based-natural-geometric-matching | 1807.05119 | null | http://arxiv.org/abs/1807.05119v1 | http://arxiv.org/pdf/1807.05119v1.pdf | Learning-based Natural Geometric Matching with Homography Prior | Geometric matching is a key step in computer vision tasks. Previous
learning-based methods for geometric matching concentrate more on improving
alignment quality, while we argue the importance of naturalness issue
simultaneously. To deal with this, firstly, Pearson correlation is applied to
handle large intra-class variations of features in feature matching stage.
Then, we parametrize homography transformation with 9 parameters in full
connected layer of our network, to better characterize large viewpoint
variations compared with affine transformation. Furthermore, a novel loss
function with Gaussian weights guarantees the model accuracy and efficiency in
training procedure. Finally, we provide two choices for different purposes in
geometric matching. When compositing homography with affine transformation, the
alignment accuracy improves and all lines are preserved, which results in a
more natural transformed image. When compositing homography with non-rigid
thin-plate-spline transformation, the alignment accuracy further improves.
Experimental results on Proposal Flow dataset show that our method outperforms
state-of-the-art methods, both in terms of alignment accuracy and naturalness. | ['Tianli Liao', 'Yifang Xu', 'Jing Chen'] | 2018-07-13 | null | null | null | null | ['geometric-matching'] | ['computer-vision'] | [-8.75000283e-02 -3.35226029e-01 -2.43519798e-01 -4.58956093e-01
-4.82189029e-01 -4.51907098e-01 4.68215227e-01 -2.07473353e-01
-3.70072007e-01 4.95943904e-01 1.44617930e-01 1.51077345e-01
-2.89595753e-01 -7.60817349e-01 -6.07824624e-01 -6.18901312e-01
1.82327051e-02 3.49378556e-01 1.59592539e-01 -1.23390496e-01
4.74888772e-01 8.52664888e-01 -1.06456399e+00 -2.56922513e-01
9.83924627e-01 8.03518176e-01 -1.64561227e-01 2.46752694e-01
5.55542968e-02 3.76084894e-01 -2.15090156e-01 -5.36346197e-01
8.36096585e-01 -2.41435871e-01 -7.92886317e-01 1.45152137e-01
1.14435530e+00 -4.00560141e-01 -7.59603679e-01 1.20636439e+00
5.00800431e-01 4.06729400e-01 5.60052574e-01 -1.29498172e+00
-5.48932850e-01 3.77016097e-01 -7.95431435e-01 -1.47244930e-01
3.11413705e-01 1.35091051e-01 1.26752496e+00 -7.21201360e-01
7.51665592e-01 1.47304869e+00 7.75378942e-01 2.71774024e-01
-1.20645046e+00 -7.22051442e-01 -1.29337413e-02 3.51897240e-01
-1.27472937e+00 -2.88836151e-01 1.21141922e+00 -3.71277958e-01
6.45730019e-01 2.50920445e-01 7.12676585e-01 4.88525897e-01
1.84810802e-01 6.68601751e-01 7.42911160e-01 -3.00049037e-01
-3.59541208e-01 -2.31288135e-01 -4.10420671e-02 7.44287789e-01
2.16114074e-01 3.71374547e-01 -3.05899829e-01 1.14257738e-01
1.27690613e+00 2.42784292e-01 -4.53376919e-01 -9.06988800e-01
-1.42213368e+00 6.89269185e-01 8.63560200e-01 3.93341511e-01
-1.88632324e-01 1.46328792e-01 3.38744968e-01 3.77409488e-01
1.02527343e-01 5.46888530e-01 -1.29249290e-01 5.88149019e-02
-9.12510574e-01 2.45199218e-01 4.50086504e-01 1.20362830e+00
8.84627402e-01 9.15052742e-02 -1.08344756e-01 9.18089688e-01
2.97430843e-01 4.40467834e-01 4.29926157e-01 -1.04713190e+00
6.40712678e-01 6.16868734e-01 -1.63202971e-01 -1.41015363e+00
-5.73219538e-01 -5.02348065e-01 -1.32207823e+00 3.95920247e-01
7.27799594e-01 1.77503034e-01 -5.47570586e-01 1.57855427e+00
2.34002739e-01 3.88890952e-01 -1.04910493e-01 1.15927291e+00
6.81251168e-01 2.31644154e-01 -2.82528639e-01 1.57413661e-01
1.23712766e+00 -1.15733397e+00 -5.59011459e-01 1.61154382e-02
5.07379949e-01 -1.16468120e+00 9.72341001e-01 -3.47905904e-02
-1.11203551e+00 -1.06785965e+00 -1.05483115e+00 -1.57728985e-01
-8.74944311e-03 1.12911791e-01 5.96505642e-01 4.45475608e-01
-8.91065300e-01 1.01246655e+00 -7.32826173e-01 -2.64581144e-01
4.05363888e-01 3.89100671e-01 -6.21055484e-01 -1.28375655e-02
-1.04197693e+00 9.70810831e-01 2.02107877e-01 1.47326784e-02
-4.13818434e-02 -8.35333049e-01 -9.66788411e-01 1.38941213e-01
4.24368791e-02 -9.97287810e-01 7.16980398e-01 -7.03071833e-01
-1.51276934e+00 9.75595415e-01 -1.08724227e-02 -2.92762130e-01
1.04966164e+00 -2.84669995e-01 -1.92146271e-01 -3.35802436e-02
-1.51477098e-01 7.81856418e-01 8.68440628e-01 -1.11567318e+00
-5.23174107e-01 -2.99123883e-01 9.07285959e-02 2.58917391e-01
-5.07877022e-02 -2.57582515e-01 -3.85778636e-01 -7.27601290e-01
4.39631373e-01 -1.04177225e+00 -2.68714845e-01 4.15963382e-01
-1.87729210e-01 -1.01990104e-01 8.31385434e-01 -5.97372890e-01
1.05764079e+00 -2.11517572e+00 8.82246718e-02 4.35755759e-01
3.29279661e-01 3.04834545e-01 -1.95359394e-01 1.12049028e-01
-2.56477624e-01 -1.85786754e-01 -1.23466522e-01 -3.53232265e-01
2.88873371e-02 1.63640268e-02 -2.02005744e-01 7.87304163e-01
2.53968537e-01 1.00017285e+00 -7.62224197e-01 -6.43820286e-01
5.14292538e-01 4.84735370e-01 -7.52026916e-01 1.37163237e-01
4.90362585e-01 7.26400495e-01 -5.17824709e-01 3.57079059e-01
1.05388379e+00 -3.82782817e-02 -3.29217702e-01 -6.88163519e-01
-7.70334005e-02 8.29060562e-03 -1.35826898e+00 2.18210030e+00
-4.80506897e-01 6.49612963e-01 -3.20798039e-01 -1.03533137e+00
1.36562741e+00 6.88172355e-02 7.47343481e-01 -5.78474224e-01
3.16040635e-01 2.12307066e-01 2.15443090e-01 -4.29722190e-01
3.12335044e-01 3.06497067e-01 4.69506979e-01 1.70005277e-01
-7.07063600e-02 -3.99524570e-01 1.00461036e-01 -1.96989581e-01
6.15301251e-01 3.94443244e-01 2.90294319e-01 -4.30463850e-01
9.13840055e-01 -3.65313768e-01 6.31698072e-01 4.10870939e-01
-3.60864609e-01 8.58834267e-01 1.83584243e-01 -9.15273428e-01
-1.28360939e+00 -9.27362442e-01 -2.57029057e-01 2.88308769e-01
5.68356097e-01 -6.73667192e-02 -5.56985378e-01 -5.73397875e-01
1.42998978e-01 1.72377914e-01 -3.61534059e-01 -2.52749681e-01
-1.20280969e+00 -3.96601856e-01 2.83322901e-01 6.47019744e-01
1.12714255e+00 -8.84998083e-01 -5.17745733e-01 1.24677263e-01
-1.87453404e-01 -1.11832893e+00 -9.41259563e-01 -5.55949450e-01
-1.11143184e+00 -1.30012107e+00 -6.77831829e-01 -9.68391001e-01
8.58080387e-01 4.31064248e-01 1.02088094e+00 3.28792572e-01
-4.24560666e-01 -3.80414315e-02 -1.06303148e-01 2.41406605e-01
-7.82191232e-02 6.34924769e-02 -1.90963581e-01 8.12692493e-02
1.62100300e-01 -9.23437774e-01 -7.53843427e-01 5.36401749e-01
-6.13938034e-01 6.05063215e-02 5.59472144e-01 8.72143865e-01
3.72843295e-01 -2.62022644e-01 8.63575637e-02 -5.37285209e-01
2.71050870e-01 3.65179420e-01 -8.42710674e-01 3.01434129e-01
-5.38462579e-01 1.89381391e-01 6.71735942e-01 -5.71132421e-01
-7.32018530e-01 1.34680122e-01 -9.60328206e-02 -8.45737934e-01
-3.47333178e-02 3.80975241e-03 -1.36715516e-01 -7.32358932e-01
5.15312016e-01 9.59029794e-02 1.88869417e-01 -2.29104757e-01
4.19675738e-01 1.57233372e-01 7.98670352e-01 -6.32192492e-01
1.35356951e+00 5.26779056e-01 5.98353863e-01 -4.94231671e-01
-4.15212393e-01 -5.79348564e-01 -1.13348651e+00 -2.14882225e-01
7.36140430e-01 -6.26967132e-01 -9.79023159e-01 5.51241696e-01
-1.21234274e+00 7.28710815e-02 -9.14839581e-02 9.13357437e-01
-8.57185304e-01 5.82609773e-01 -4.39039797e-01 -3.04746658e-01
-4.31516260e-01 -1.27706492e+00 8.93790364e-01 2.69836396e-01
-9.83310044e-02 -1.31105471e+00 7.71966279e-02 1.86600253e-01
4.99840140e-01 4.53400284e-01 8.19194615e-01 -5.08268058e-01
-7.48016298e-01 -2.23090723e-01 -5.31411529e-01 2.17390269e-01
3.91876012e-01 4.51626368e-02 -7.12020159e-01 -4.47258502e-01
-6.36636838e-02 9.53208134e-02 7.32288599e-01 3.25696617e-01
1.06030440e+00 -1.89331889e-01 -1.46366328e-01 1.23378038e+00
1.37682152e+00 9.20162871e-02 7.46532619e-01 6.39548838e-01
1.07782090e+00 6.60366178e-01 5.98292768e-01 1.15488648e-01
2.92717189e-01 8.67947638e-01 3.26302916e-01 -4.34491932e-01
-3.62762809e-01 -2.00348735e-01 -1.91337213e-01 7.14656055e-01
-3.09188426e-01 1.88404024e-01 -6.86375022e-01 2.21240938e-01
-1.99012291e+00 -9.09724534e-01 -2.14673027e-01 2.33830428e+00
5.25032043e-01 -6.02447093e-02 3.77506949e-02 1.77583143e-01
7.50372887e-01 2.77720064e-01 -4.37818259e-01 -2.64041405e-02
-2.67096370e-01 -7.38091841e-02 5.63653231e-01 7.11383104e-01
-1.17090750e+00 1.04763377e+00 5.51141548e+00 7.32355058e-01
-1.24156833e+00 -3.81076187e-01 4.13883537e-01 3.73701304e-01
-9.58789960e-02 1.71075314e-01 -6.31319344e-01 2.12537333e-01
-9.61673111e-02 -9.51211452e-02 4.47094232e-01 6.47636473e-01
8.56968611e-02 3.26446265e-01 -1.28349531e+00 1.45943129e+00
1.71854466e-01 -1.35874426e+00 4.29741852e-02 -1.97835132e-01
9.15045738e-01 -3.40208769e-01 -6.27526864e-02 -8.04295987e-02
-5.38923480e-02 -8.27458382e-01 4.41501737e-01 6.79392099e-01
5.37993848e-01 -7.61496365e-01 8.04599643e-01 -2.92271785e-02
-1.38428032e+00 4.45284367e-01 -5.69511175e-01 2.26128757e-01
2.34212622e-01 2.49048620e-01 -3.71067226e-01 8.28720212e-01
5.48191130e-01 1.03030479e+00 -4.72400486e-01 1.50407040e+00
-1.63955286e-01 -1.08814470e-01 -2.33374730e-01 4.24766123e-01
2.19635695e-01 -7.77125359e-01 6.41236663e-01 1.07318544e+00
1.42730877e-01 -1.20378606e-01 3.29010487e-01 9.09242034e-01
-8.39452446e-02 3.04094255e-01 -5.73604763e-01 5.77870846e-01
4.09631163e-01 1.28487682e+00 -5.20352721e-01 -1.79091036e-01
-3.93396437e-01 8.36614668e-01 2.72291780e-01 3.53634655e-01
-7.02002466e-01 -5.07871091e-01 6.77527249e-01 2.52355319e-02
2.79180501e-02 -3.98414671e-01 -3.88567448e-01 -1.29737723e+00
2.70079464e-01 -7.14369118e-01 7.92845488e-02 -5.23037314e-01
-1.32003963e+00 5.77158272e-01 -5.33711612e-02 -1.73538744e+00
-1.97986603e-01 -4.90773022e-01 -9.11700249e-01 9.38016593e-01
-1.65772843e+00 -1.47646856e+00 -6.22631013e-01 6.98562562e-01
4.61817294e-01 -1.50979519e-01 4.60893959e-01 5.70558131e-01
-2.67374694e-01 9.00650918e-01 -1.84395850e-01 5.16079783e-01
1.13904035e+00 -9.80134666e-01 6.18860722e-01 8.31774533e-01
2.08502591e-01 6.63299918e-01 4.93491262e-01 -3.60448152e-01
-1.26685607e+00 -1.02239668e+00 9.08910751e-01 -1.29517972e-01
5.59104502e-01 1.07746804e-02 -1.23121297e+00 5.21336317e-01
5.33494800e-02 1.64530158e-01 1.24886766e-01 -2.14824930e-01
-5.24620295e-01 -2.57847935e-01 -1.16677606e+00 7.19097555e-01
1.23585796e+00 -4.47785974e-01 -5.38958848e-01 2.25454107e-01
4.85682636e-01 -7.05620825e-01 -1.12287307e+00 6.21610999e-01
7.19812393e-01 -9.71234679e-01 1.30641222e+00 -6.25938952e-01
2.38132447e-01 -5.17510593e-01 5.53259887e-02 -1.19284320e+00
-5.79311609e-01 -8.35423529e-01 4.43678558e-01 1.11912513e+00
1.01298384e-01 -7.55167663e-01 7.85687029e-01 5.34845769e-01
-1.56087190e-01 -6.67581081e-01 -8.07639778e-01 -9.44353282e-01
2.36043319e-01 9.14469883e-02 7.86059022e-01 1.18977869e+00
-1.74474612e-01 -2.61083208e-02 -5.48889816e-01 1.11299738e-01
7.64104307e-01 3.29235494e-01 1.12262285e+00 -1.15334547e+00
1.06165987e-02 -8.93500745e-01 -9.44156289e-01 -1.27266896e+00
3.35132122e-01 -8.14581573e-01 -2.39309624e-01 -1.07802141e+00
2.80086007e-02 -5.72322369e-01 9.12348181e-02 2.07967311e-01
-2.62981921e-01 2.92878717e-01 3.76766235e-01 4.27947193e-01
-1.34258792e-01 6.45050168e-01 1.74295175e+00 -6.20519966e-02
-2.86241710e-01 8.67767707e-02 5.64288301e-03 6.87753022e-01
8.10208201e-01 -1.26059145e-01 -1.41587615e-01 -6.44143283e-01
-3.08028311e-01 -6.30448684e-02 2.98968047e-01 -1.06935263e+00
5.20770192e-01 -2.34983653e-01 3.48010093e-01 -4.73378390e-01
8.22627395e-02 -1.00100744e+00 1.71505228e-01 5.00633597e-01
-3.24286193e-01 3.51556659e-01 3.23659293e-02 -6.92580128e-03
-4.76748258e-01 -8.14529881e-02 1.10186851e+00 2.50039071e-01
-4.73548681e-01 8.58673990e-01 4.61471915e-01 4.23712581e-02
7.55989730e-01 -5.12707472e-01 -1.47580475e-01 -2.72795618e-01
-4.35851634e-01 1.71636522e-01 5.30896068e-01 6.99133217e-01
5.93964219e-01 -1.72996998e+00 -7.96775460e-01 5.75470984e-01
5.37041342e-04 1.61103994e-01 1.12690628e-01 9.82111275e-01
-8.07785332e-01 3.14124167e-01 -5.91679335e-01 -7.22078323e-01
-1.20359230e+00 3.83776873e-01 4.98666406e-01 -2.83693314e-01
-7.08274782e-01 5.31892419e-01 3.73948336e-01 -4.52177107e-01
2.98803538e-01 -3.32130075e-01 -2.91790336e-01 -2.36606166e-01
2.39885271e-01 5.23556292e-01 -7.77976587e-02 -8.47900927e-01
-2.04174966e-01 1.52284467e+00 7.82172754e-02 2.21633732e-01
1.09826124e+00 -1.09343998e-01 -5.51043488e-02 -9.26622003e-02
1.51141846e+00 9.06632990e-02 -1.35050488e+00 -4.88989443e-01
-1.86247945e-01 -9.55748796e-01 -1.43523693e-01 -1.47876903e-01
-1.52430403e+00 1.00788367e+00 6.38565183e-01 -2.36336052e-01
9.20866430e-01 -5.82760215e-01 6.68156803e-01 3.51220667e-01
2.03928858e-01 -6.32098913e-01 3.77525091e-02 5.23620725e-01
1.07394326e+00 -1.47943056e+00 8.30491334e-02 -6.21983528e-01
-3.42365444e-01 1.51258242e+00 8.00549388e-01 -7.51611352e-01
6.47050738e-01 -1.56034932e-01 2.19266713e-01 1.10685542e-01
-3.67341861e-02 2.59147156e-02 8.96072268e-01 5.99465370e-01
4.46328402e-01 -1.69267118e-01 -3.12317640e-01 -2.49161974e-01
-6.00002944e-01 -4.50376838e-01 1.60013326e-02 5.83834112e-01
-2.84413278e-01 -1.26445794e+00 -3.56471509e-01 3.46936025e-02
-1.20646946e-01 3.50105427e-02 -1.43481076e-01 1.00823569e+00
-1.57978684e-01 6.15781903e-01 2.49018952e-01 -3.07566196e-01
6.03460193e-01 -4.36592609e-01 6.87555015e-01 4.73042466e-02
-6.34947240e-01 2.81740665e-01 -2.70468414e-01 -8.44002783e-01
-5.59436917e-01 -5.53057432e-01 -1.10087073e+00 -5.41765630e-01
-4.71952260e-01 -1.70128606e-02 4.91346449e-01 8.75079155e-01
1.11899920e-01 3.64962012e-01 8.57822597e-01 -9.46269870e-01
-6.88655794e-01 -7.86242723e-01 -2.43837044e-01 8.86467636e-01
4.05261993e-01 -6.79094911e-01 -4.03515965e-01 9.70907286e-02] | [8.611462593078613, -2.219709634780884] |
2096d23f-7f37-42c4-b106-3ce165f72e6f | towards-adversarial-robustness-of-bayesian | null | null | https://openreview.net/forum?id=Cue2ZEBf12 | https://openreview.net/pdf?id=Cue2ZEBf12 | Towards Adversarial Robustness of Bayesian Neural Network through Hierarchical Variational Inference | Recent works have applied Bayesian Neural Network (BNN) to adversarial training, and shown the improvement of adversarial robustness via the BNN's strength of stochastic gradient defense. However, we have found that in general, the BNN loses its stochasticity after its training with the posterior. As a result, the lack of the stochasticity leads to weak regularization effect to the BNN, which increases KL divergence in ELBO from variational inference. In this paper, we propose an enhanced Bayesian regularizer through hierarchical variational inference in order to boost adversarial robustness against gradient-based attack. Furthermore, we also prove that the proposed method allows the BNN's stochasticity to be elevated with the reduced KL divergence. Exhaustive experiment results demonstrate the effectiveness of the proposed method by showing the improvement of adversarial robustness for the BNN, compared with adversarial training (Madry et al., 2018) and adversarial-BNN (Liu et al., 2019) under PGD attack and EOT-PGD attack to the $L_{\infty}$ perturbation on CIFAR-10/100, STL-10, and Tiny-ImageNet. | ['Yong Man Ro', 'Youngjoon Yu', 'Byung-Kwan Lee'] | 2021-01-01 | null | null | null | null | ['probabilistic-deep-learning'] | ['computer-vision'] | [-3.3648911e-01 8.2967781e-02 2.9255390e-01 -1.7504673e-01
-6.1997163e-01 -7.0204914e-01 2.9274681e-01 -8.1284887e-01
-4.8372027e-01 9.6025741e-01 2.2508124e-02 -3.7315947e-01
-2.5545010e-01 -8.0930614e-01 -9.8704499e-01 -1.0727390e+00
-7.1039632e-02 -2.8301751e-02 2.9864624e-01 -2.6585037e-01
-1.2911531e-01 5.8099163e-01 -5.8222389e-01 -2.8386635e-01
7.2636223e-01 1.0188287e+00 -2.6976165e-01 4.7201073e-01
4.8460808e-01 8.6081022e-01 -9.1677696e-01 -7.6510781e-01
6.4966136e-01 -1.9988599e-01 -2.1629477e-01 -7.7412999e-01
5.9751755e-01 -7.8217024e-01 -1.1402309e+00 1.6610807e+00
8.3555907e-01 3.2043415e-01 7.4245131e-01 -1.2419220e+00
-6.0029566e-01 9.2445999e-01 -4.7896236e-01 1.7914851e-01
-3.8494191e-01 4.2865011e-01 6.6252649e-01 -4.8841673e-01
1.0081111e-01 1.7734269e+00 8.2025033e-01 1.0859619e+00
-1.1756954e+00 -1.3485857e+00 4.0343297e-01 -4.9794320e-02
-1.3851491e+00 -1.9520529e-01 8.2356924e-01 -1.2704210e-01
3.8021451e-01 1.6776577e-01 2.3134153e-01 1.7841690e+00
1.4658502e-01 7.3440337e-01 1.0524045e+00 1.6156249e-01
2.5234205e-01 6.4340308e-02 1.4162519e-02 5.5035222e-01
2.8251159e-01 7.3517954e-01 -1.9840628e-01 -2.7375212e-01
6.7569607e-01 8.3898023e-02 -3.2206169e-01 8.2252063e-02
-4.3023840e-01 9.7546470e-01 7.5230712e-01 -2.4486583e-01
-1.0148900e-01 6.0124373e-01 4.2351902e-01 5.5662405e-02
3.4481212e-01 1.3928691e-01 -5.0991404e-01 1.9054087e-01
-7.4805391e-01 2.9578400e-01 7.8554666e-01 8.6066133e-01
2.6012510e-01 7.0212388e-01 -4.8461133e-01 6.1437809e-01
8.3804321e-01 1.0508884e+00 1.8493420e-01 -9.8072803e-01
4.1964284e-01 -2.5943980e-01 -1.4822511e-01 -1.1650212e+00
-8.3297715e-02 -6.7749143e-01 -1.3245194e+00 5.2702785e-01
3.8678500e-01 -6.8764901e-01 -8.8215142e-01 2.4071174e+00
2.0870285e-01 5.8384073e-01 3.1158736e-01 7.7622879e-01
6.4444464e-01 6.5802574e-01 -2.3031257e-02 -3.3866514e-02
7.8338969e-01 -6.1908329e-01 -8.1289369e-01 -7.2388090e-02
-7.8293785e-02 -4.3245545e-01 9.0710038e-01 4.2595041e-01
-8.2800144e-01 -5.2608341e-01 -1.2540088e+00 5.2940404e-01
-1.0854210e-01 -4.7033876e-01 4.2286408e-01 1.2971281e+00
-6.1209190e-01 6.9689047e-01 -9.2333245e-01 2.3993501e-01
8.5349423e-01 3.1646746e-01 2.2189299e-02 -3.4976088e-02
-1.9102342e+00 7.8067166e-01 5.1685512e-01 3.6552221e-01
-1.6922866e+00 -9.3572783e-01 -4.6782255e-01 -7.7194400e-02
3.2522631e-01 -4.5570323e-01 8.5133821e-01 -3.9038435e-01
-1.8811821e+00 3.0334949e-02 4.2297369e-01 -8.2176995e-01
8.8225466e-01 -4.8778546e-01 -2.1876976e-01 8.6082391e-02
-3.9911696e-01 5.8012658e-01 1.1185476e+00 -1.2026774e+00
-1.6142222e-01 -4.0395561e-01 4.6652788e-01 -2.8964574e-02
-5.6135428e-01 -1.5756553e-01 -2.0233659e-01 -9.0578997e-01
-4.7183659e-02 -9.9560201e-01 -2.4861111e-01 -1.9655527e-01
-7.5648916e-01 4.7497310e-02 1.1826788e+00 -7.3410380e-01
9.9465823e-01 -2.3355753e+00 -1.3291704e-02 3.8174719e-01
1.7880636e-01 4.5711568e-01 1.0270958e-01 -1.1312510e-01
1.0229387e-01 5.8891028e-01 -3.2289669e-01 -1.8055859e-01
4.2735434e-01 3.3251864e-01 -8.5970783e-01 7.0239562e-01
-9.4453461e-02 7.5244135e-01 -7.1941179e-01 -3.4450588e-01
5.0061584e-02 7.3355126e-01 -8.4178203e-01 1.4955877e-01
-6.8655208e-02 5.5280030e-01 -4.9958321e-01 3.9786598e-01
1.1025254e+00 5.8551079e-01 -1.8542786e-01 -2.3319559e-01
6.4414179e-01 -2.9073498e-01 -1.2643331e+00 1.1798615e+00
-2.5418523e-01 5.6082928e-01 2.9680139e-01 -8.3214188e-01
6.9082713e-01 3.8036823e-01 2.3181763e-02 -2.2980317e-01
1.9302355e-01 -8.8831428e-03 1.2975119e-01 8.6485958e-03
-1.5769668e-01 -2.4928169e-01 -1.5288255e-01 4.5427278e-02
1.3774055e-01 -3.0239472e-01 -1.6688532e-01 4.0741235e-01
9.1359061e-01 4.8520222e-02 -5.1125503e-01 -2.8576314e-01
4.9466318e-01 -7.3070562e-01 8.9559937e-01 1.2888217e+00
-5.1047331e-01 2.5717127e-01 6.1595470e-01 -1.0002198e-02
-7.8128785e-01 -1.5864103e+00 -4.3041638e-01 8.2217538e-01
1.2190225e-02 6.9324583e-02 -1.0321668e+00 -9.7294539e-01
9.5768891e-02 9.6582919e-01 -5.3987056e-01 -6.9689804e-01
-3.5747960e-01 -1.1399925e+00 1.4341494e+00 2.6260722e-01
1.0542686e+00 -7.3677230e-01 -1.6105672e-02 -6.4090066e-02
1.0478332e-02 -1.0311993e+00 -3.5943502e-01 9.0151333e-04
-8.1972599e-01 -6.6843134e-01 -5.8620381e-01 -8.7337844e-02
5.0209737e-01 -4.7936934e-01 6.5197641e-01 -3.5328537e-01
-7.5296029e-03 1.7779998e-01 -5.0860833e-02 -6.8213379e-01
-4.6596992e-01 -3.0002722e-01 6.2152630e-01 -1.4704199e-01
-2.5924388e-01 -8.6098111e-01 -5.4149663e-01 4.4896039e-01
-1.0552042e+00 -7.3287857e-01 3.8109291e-01 1.0083427e+00
3.5767141e-01 6.9529063e-01 5.4943860e-01 -8.5105169e-01
6.0971296e-01 -3.6814424e-01 -9.2599040e-01 -5.7088215e-02
-4.5651627e-01 9.2047982e-02 9.0429813e-01 -7.2752166e-01
-1.2552861e+00 -5.3327787e-01 -5.4539794e-01 -9.5005673e-01
7.1619898e-02 3.1726894e-01 -6.3517976e-01 -1.4943105e-01
7.8511947e-01 2.7168414e-02 -3.0458659e-01 -3.8625520e-01
4.4020021e-01 2.4635315e-01 5.0860697e-01 -8.4077352e-01
1.4579297e+00 5.4570431e-01 2.7930948e-01 -4.1004601e-01
-1.1429617e+00 4.6843082e-01 -1.6473916e-01 -1.8438479e-01
7.8046227e-01 -7.9651290e-01 -9.8466313e-01 8.7756103e-01
-1.0394584e+00 -3.4292153e-01 -1.1205433e-01 7.2045124e-01
-2.5698873e-01 4.0808687e-01 -8.3742064e-01 -1.0007731e+00
-3.5365990e-01 -1.1610326e+00 2.2755140e-01 2.8758481e-01
3.9215297e-01 -1.0409306e+00 -1.2505738e-01 3.6427796e-01
5.6826097e-01 6.9740516e-01 7.7038080e-01 -8.0135739e-01
-3.5202986e-01 -3.3990973e-01 -7.0857130e-02 1.1481552e+00
-2.9400712e-01 1.8365377e-01 -1.1868308e+00 -6.0691124e-01
5.6254363e-01 -5.0125200e-01 1.0072498e+00 5.6749547e-01
1.2574481e+00 -6.7761832e-01 1.5623328e-01 1.0609484e+00
1.3213060e+00 1.1953675e-01 8.2702523e-01 3.1805921e-02
9.0731817e-01 1.6402289e-01 3.3459300e-01 1.6768180e-01
-4.0615493e-01 3.1806955e-01 9.9982136e-01 1.3362046e-01
2.9834650e-02 -3.5525736e-01 7.9600310e-01 5.2831024e-01
9.2219047e-02 -2.8188220e-01 -5.7910645e-01 -4.1785460e-02
-1.6382451e+00 -1.1715597e+00 7.7246502e-02 2.0754097e+00
1.1260982e+00 6.2197781e-01 -3.2828426e-01 -9.4988212e-02
7.1398824e-01 2.2659658e-01 -9.6676546e-01 -2.3743977e-01
-2.7685586e-01 7.4616522e-02 8.3908463e-01 6.6573906e-01
-1.1974696e+00 1.0152358e+00 5.8015471e+00 1.4091655e+00
-8.8940185e-01 2.2361943e-01 8.2343835e-01 -3.3053476e-01
-7.2217010e-02 -1.9767195e-01 -9.7836900e-01 8.2129848e-01
9.5945835e-01 1.7907774e-02 7.0926249e-01 8.8650364e-01
9.9195936e-04 3.3078888e-01 -8.1475776e-01 5.4370522e-01
-1.2350831e-01 -8.6771327e-01 2.8211821e-02 1.3357283e-01
8.4276485e-01 5.7487093e-02 6.6503823e-01 8.0220038e-01
8.9569843e-01 -1.2719938e+00 7.2738850e-01 6.4223886e-01
4.2147884e-01 -1.3085910e+00 9.5614016e-01 4.8307291e-01
-4.6906021e-01 -2.2559527e-01 -7.1224040e-01 3.4974563e-01
-5.4071844e-04 7.7648830e-01 -4.3300465e-01 4.4591424e-01
9.4349468e-01 2.6113117e-01 -2.5988197e-01 3.3703467e-01
-7.2565204e-01 1.1390114e+00 -4.7411638e-01 1.2617698e-01
4.2220289e-01 -2.2299212e-01 1.0835587e+00 1.0040040e+00
-4.7831358e-03 -5.0467651e-02 1.0268960e-01 1.0901645e+00
-3.7118092e-01 -3.2230219e-01 -5.6665885e-01 1.1503811e-01
5.1754284e-01 9.2256653e-01 -2.7933779e-01 -6.6790476e-02
1.9716628e-01 6.5921593e-01 8.0282658e-02 7.4235797e-01
-1.4675910e+00 -6.3962638e-01 8.1625843e-01 -5.0428927e-01
2.6609769e-01 -5.0142329e-02 -1.4303936e-01 -7.8871703e-01
-1.1992205e-01 -1.0233430e+00 2.2382602e-01 -4.9921972e-01
-1.6025391e+00 5.3640991e-01 1.8116099e-01 -7.3401129e-01
6.5675706e-02 -7.5313765e-01 -6.3431090e-01 1.0548813e+00
-1.1386811e+00 -8.9841169e-01 2.2025874e-01 8.8437426e-01
4.5741044e-02 -5.5112249e-01 5.1766360e-01 3.6216843e-01
-1.0343839e+00 1.2140526e+00 5.0462246e-01 6.8091816e-01
5.8233798e-01 -1.1104437e+00 9.3197599e-02 1.2331139e+00
-1.3899094e-01 8.0171013e-01 9.0388018e-01 -5.8567387e-01
-1.1160630e+00 -1.3754869e+00 -3.1086928e-01 -6.0590631e-01
8.0492276e-01 -4.5438945e-01 -7.5469387e-01 5.1370293e-01
1.5559047e-01 4.0184729e-02 4.3495864e-01 -1.9349872e-01
-7.2058374e-01 -4.0701428e-01 -1.6981477e+00 9.6547562e-01
7.8306466e-01 -5.5023843e-01 -4.6444607e-01 3.5801238e-01
1.1811717e+00 -3.9424643e-01 -8.6494118e-01 7.2806299e-01
4.7577947e-01 -8.8770640e-01 1.3144048e+00 -7.5273156e-01
2.2668734e-01 -2.3763278e-01 -5.8866793e-01 -1.3120270e+00
-7.1779929e-02 -8.7418175e-01 -3.3042571e-01 1.3779297e+00
3.1954020e-01 -8.3444750e-01 5.5260086e-01 3.7450370e-01
-6.7987703e-02 -6.0331315e-01 -1.2019399e+00 -1.1056935e+00
7.1371204e-01 -6.2786877e-01 2.9727986e-01 5.3679222e-01
-7.2633630e-01 -9.4472744e-02 -6.7455560e-01 7.6513374e-01
1.0132499e+00 -8.2204098e-01 7.4588841e-01 -8.7629837e-01
-4.3432063e-01 -4.1299540e-01 -3.5047486e-01 -8.9050967e-01
4.8972434e-01 -7.4721718e-01 1.0804570e-01 -8.8070101e-01
-1.2861800e-02 -6.6545531e-02 -8.6850262e-01 2.4800996e-01
-3.3961785e-01 2.7582958e-01 2.9273996e-01 -1.5668023e-02
-3.8047358e-01 6.8809789e-01 1.2156473e+00 -4.3034273e-01
1.6200478e-01 2.5674665e-01 -6.3463509e-01 9.1941094e-01
9.2816788e-01 -8.9881700e-01 -5.9634215e-01 -2.0599709e-01
5.8426872e-02 -3.6281008e-01 6.4170712e-01 -1.1663735e+00
6.3014485e-02 -8.6499013e-02 4.1691843e-01 -3.8160789e-01
3.0971879e-01 -8.7713850e-01 -1.1635720e-01 7.8147048e-01
-2.5576320e-01 -4.3375459e-01 3.6469445e-01 9.2487794e-01
5.4061282e-02 -3.1638813e-01 1.2992011e+00 8.9665227e-02
1.7217223e-02 6.3139933e-01 -3.2312095e-01 4.0892372e-01
7.8816450e-01 3.0374888e-01 -2.2540204e-01 -4.4656828e-01
-6.6223621e-01 2.2844343e-01 -1.1485770e-01 -4.3593541e-02
6.1662257e-01 -1.3935467e+00 -9.6771437e-01 3.5703924e-02
-6.7444265e-01 1.5445822e-01 4.9973863e-01 6.7700058e-01
-4.5913747e-01 1.4682760e-02 -1.3833325e-01 -4.4455305e-01
-8.8672274e-01 5.7642496e-01 6.0779470e-01 -3.4247896e-01
-3.1654626e-01 1.3431423e+00 3.2173282e-01 -5.1873314e-01
7.8289324e-01 1.6008475e-01 1.3551314e-01 -2.8556541e-01
6.4343467e-02 5.4590684e-01 -3.4283811e-01 -2.7444974e-01
-3.1489179e-01 1.4147831e-01 -9.5423743e-02 -5.0742161e-01
1.0339004e+00 1.0260606e-01 -3.3251788e-02 1.1157148e-01
1.1203377e+00 1.8114613e-01 -1.7288162e+00 -1.4253885e-01
-4.4576448e-01 -3.4828410e-01 1.7104994e-01 -8.0378634e-01
-1.4304726e+00 1.0603619e+00 8.8985854e-01 1.0965576e-01
9.4869840e-01 -4.3703717e-01 7.8102279e-01 7.6008022e-01
1.5140118e-01 -1.0407273e+00 1.8619254e-01 7.8569096e-01
8.7425256e-01 -1.0548329e+00 -2.4064099e-02 2.0050193e-01
-4.4995925e-01 6.4633524e-01 6.6459656e-01 -4.3801251e-01
1.1075177e+00 2.3961931e-01 1.2223153e-01 2.3306613e-01
-3.4777766e-01 4.4881076e-01 8.5976221e-02 5.8876264e-01
-2.2259191e-01 -9.7217895e-02 7.9243995e-02 7.6297402e-01
-3.3092993e-01 -6.8717629e-01 2.2982505e-01 5.8678985e-01
-1.5823928e-01 -6.7810035e-01 -5.1752412e-01 1.7414971e-01
-1.0127901e+00 -3.3453897e-01 -1.4282016e-01 8.5964113e-01
1.3697104e-01 1.0778357e+00 -4.4055119e-01 -4.0657443e-01
2.6744339e-01 4.1967105e-02 2.9329881e-01 -1.5383071e-01
-4.5402107e-01 -7.4559073e-03 -4.5176685e-01 -4.8717126e-01
1.8605158e-01 -4.9706879e-01 -9.2987418e-01 -3.7953362e-01
-3.3054793e-01 2.4143572e-01 6.8826729e-01 7.7044362e-01
-7.0965067e-02 8.4647763e-01 9.1709983e-01 -5.5514199e-01
-1.3837433e+00 -1.2343732e+00 -7.2798401e-01 4.3513963e-01
2.8167596e-01 -6.4164913e-01 -1.0455505e+00 -3.2347563e-01] | [5.613193035125732, 7.874966621398926] |
66ec24d2-68c9-42c7-bc84-5844c7ee7f03 | crosspyramid-neural-ordinary-differential | 2212.03560 | null | https://arxiv.org/abs/2212.03560v1 | https://arxiv.org/pdf/2212.03560v1.pdf | CrossPyramid: Neural Ordinary Differential Equations Architecture for Partially-observed Time-series | Ordinary Differential Equations (ODE)-based models have become popular foundation models to solve many time-series problems. Combining neural ODEs with traditional RNN models has provided the best representation for irregular time series. However, ODE-based models require the trajectory of hidden states to be defined based on the initial observed value or the last available observation. This fact raises questions about how long the generated hidden state is sufficient and whether it is effective when long sequences are used instead of the typically used shorter sequences. In this article, we introduce CrossPyramid, a novel ODE-based model that aims to enhance the generalizability of sequences representation. CrossPyramid does not rely only on the hidden state from the last observed value; it also considers ODE latent representations learned from other samples. The main idea of our proposed model is to define the hidden state for the unobserved values based on the non-linear correlation between samples. Accordingly, CrossPyramid is built with three distinctive parts: (1) ODE Auto-Encoder to learn the best data representation. (2) Pyramidal attention method to categorize the learned representations (hidden state) based on the relationship characteristics between samples. (3) Cross-level ODE-RNN to integrate the previously learned information and provide the final latent state for each sample. Through extensive experiments on partially-observed synthetic and real-world datasets, we show that the proposed architecture can effectively model the long gaps in intermittent series and outperforms state-of-the-art approaches. The results show an average improvement of 10\% on univariate and multivariate datasets for both forecasting and classification tasks. | ['Flora D. Salim', 'Yongli Ren', 'Hao Xue', 'Futoon M. Abushaqra'] | 2022-12-07 | null | null | null | null | ['irregular-time-series'] | ['time-series'] | [-1.43765181e-01 -5.06190300e-01 -3.96614403e-01 -2.05554098e-01
-3.11495394e-01 -3.93592805e-01 5.52840352e-01 -1.04738906e-01
2.57859752e-02 6.56420350e-01 2.66960025e-01 -3.03861916e-01
-2.67701685e-01 -6.03006363e-01 -5.47284305e-01 -9.15299952e-01
-4.29204553e-01 2.95265406e-01 -1.84131965e-01 -3.13685298e-01
-2.98641473e-02 5.19902766e-01 -1.43513906e+00 1.67350322e-01
9.66119051e-01 1.21188235e+00 2.56507337e-01 4.91341084e-01
-2.18489006e-01 1.13926971e+00 -6.98547363e-01 3.33975136e-01
2.20621988e-01 -6.98009431e-01 -2.59957343e-01 -8.20393190e-02
-4.54999298e-01 -3.69536072e-01 -6.80360019e-01 5.44053316e-01
3.25244546e-01 3.07861477e-01 8.47142994e-01 -1.33716917e+00
-8.10819328e-01 6.42141581e-01 -1.70220152e-01 5.85673749e-01
-6.24408722e-02 3.47240925e-01 7.49183416e-01 -4.58154976e-01
3.21957350e-01 1.03768015e+00 7.50140131e-01 4.71863091e-01
-1.13714969e+00 -6.45723343e-01 4.48113471e-01 2.72527784e-01
-1.21084332e+00 -1.16260894e-01 9.35433149e-01 -5.98301232e-01
1.14656937e+00 7.12207928e-02 7.45885253e-01 1.48283350e+00
4.54699516e-01 6.68127179e-01 8.74940515e-01 6.80886358e-02
3.11397612e-01 -9.02586579e-02 3.84693950e-01 2.82032162e-01
-1.03499368e-01 1.60187736e-01 -1.93651199e-01 -2.58867294e-01
9.00901318e-01 5.81051350e-01 -4.22212601e-01 -1.17750697e-01
-1.09107709e+00 7.15302110e-01 3.03997606e-01 4.74751025e-01
-9.17201340e-01 1.81116953e-01 4.62846518e-01 3.48726988e-01
4.72966373e-01 3.78775865e-01 -7.71132350e-01 -3.31686139e-01
-7.94810414e-01 2.06494797e-02 6.99442327e-01 6.59712970e-01
4.69835788e-01 7.77107775e-01 -2.54586011e-01 7.09920406e-01
1.35327548e-01 4.98428345e-01 1.00018311e+00 -7.47220933e-01
3.55653077e-01 5.53915024e-01 9.61671397e-02 -8.88909757e-01
-2.91453898e-01 -7.75410235e-01 -1.13972354e+00 -2.29726136e-01
1.36930659e-01 -4.06795979e-01 -9.96621192e-01 1.97853279e+00
-4.20703813e-02 7.21134126e-01 2.99315333e-01 4.84363109e-01
4.43565905e-01 1.33358228e+00 -2.55642861e-01 -5.91480613e-01
7.92955160e-01 -8.42836380e-01 -1.02289689e+00 2.11347342e-01
4.59493667e-01 -1.50620416e-01 6.47416830e-01 2.17352331e-01
-7.99846530e-01 -7.80447960e-01 -8.91249061e-01 2.72818536e-01
-4.95538861e-01 2.26446733e-01 3.64778876e-01 -3.16998996e-02
-8.15853715e-01 9.63535309e-01 -1.15875471e+00 -1.44879326e-01
-6.38846904e-02 2.67441094e-01 -1.05460575e-02 3.99059355e-01
-1.49934340e+00 7.35486746e-01 3.56830388e-01 4.67968047e-01
-1.17581868e+00 -8.27591300e-01 -7.99341321e-01 2.87660033e-01
1.15004592e-01 -4.66286451e-01 9.38420832e-01 -8.13301086e-01
-1.68672347e+00 2.18768772e-02 -3.32464725e-01 -8.14711332e-01
3.46824825e-01 -1.89142466e-01 -6.87085867e-01 5.15533388e-02
-2.23297924e-02 1.68696016e-01 9.17167485e-01 -9.41178679e-01
-1.86563894e-01 4.54124734e-02 -3.45669746e-01 -1.19730063e-01
-2.11244702e-01 -4.40811396e-01 -1.50330395e-01 -8.92290711e-01
1.23539660e-02 -1.11582410e+00 -2.58825719e-01 -2.74520010e-01
-1.03885449e-01 -3.94390613e-01 9.27228987e-01 -7.99354792e-01
1.64934886e+00 -2.06727457e+00 1.88596159e-01 -2.00738702e-02
-1.33848544e-02 4.52445298e-01 -5.04099950e-02 9.51960087e-01
-4.86264229e-01 -9.46015045e-02 -2.16649801e-01 -2.85426050e-01
-1.75849244e-01 4.95816141e-01 -9.47757721e-01 3.47380787e-01
2.66840279e-01 8.49605203e-01 -7.77284682e-01 4.78298925e-02
3.24509382e-01 7.73857355e-01 -1.74768329e-01 3.73138845e-01
-2.79476285e-01 6.81275010e-01 -4.02772814e-01 1.08584255e-01
3.18727642e-01 -4.53334242e-01 9.63386334e-03 -7.72044435e-02
-3.02871883e-01 3.29321921e-01 -1.07036304e+00 1.38561869e+00
-4.68564242e-01 6.19432747e-01 -5.90399802e-01 -1.19659865e+00
1.27299786e+00 5.87752104e-01 8.50692630e-01 -4.04561490e-01
2.02324837e-01 2.24423513e-01 8.36438537e-02 -6.11295164e-01
6.95345774e-02 7.13174269e-02 -8.85739997e-02 4.01108116e-01
4.26467285e-02 3.65929872e-01 2.39607155e-01 -5.86856641e-02
1.04465353e+00 2.98783630e-01 3.53431612e-01 -1.21138632e-01
6.29785001e-01 -1.03503518e-01 9.13091362e-01 5.51050603e-01
-3.82621437e-02 4.22892243e-01 7.31596351e-01 -6.69121504e-01
-1.03200841e+00 -9.07686114e-01 -6.57522082e-02 4.91030484e-01
-5.80204539e-02 -2.37789601e-01 -3.48430127e-01 -2.92854488e-01
-3.19394097e-02 8.11524391e-01 -8.94407272e-01 -4.09410030e-01
-6.14371061e-01 -6.88515842e-01 5.24903774e-01 7.62936115e-01
4.53630954e-01 -1.28293931e+00 -5.97159028e-01 5.59056759e-01
-1.87279359e-01 -8.98771942e-01 -4.10681605e-01 4.57404822e-01
-1.02902663e+00 -7.14833677e-01 -8.84831131e-01 -3.20884138e-01
5.22889912e-01 -3.36762075e-03 5.78522325e-01 -3.07561368e-01
1.09602503e-01 2.49512672e-01 -3.95589769e-01 -1.94381401e-01
-3.62727076e-01 1.43878192e-01 2.86712646e-01 4.22733665e-01
1.24912083e-01 -7.91124105e-01 -4.30714935e-01 1.71034604e-01
-8.64001095e-01 -1.20003887e-01 5.68282962e-01 9.31831717e-01
3.86942059e-01 6.87673613e-02 8.04224610e-01 -3.95036936e-01
8.23446214e-01 -9.00227845e-01 -4.41090852e-01 2.79969931e-01
-6.17276609e-01 5.76977670e-01 1.20733619e+00 -9.46373224e-01
-8.41061771e-01 -1.71306103e-01 -4.36507016e-02 -1.10763049e+00
1.90272957e-01 6.90225482e-01 3.47595751e-01 6.86669946e-01
3.22734118e-01 6.99816942e-01 2.11616024e-01 -5.33143699e-01
-9.56473053e-02 6.13876700e-01 5.06111383e-01 -5.04301190e-01
3.93954724e-01 3.74655277e-01 -1.21094109e-02 -8.22367370e-01
-6.89824402e-01 -3.41042221e-01 -5.94380677e-01 -5.46047240e-02
7.57899046e-01 -1.08563364e+00 -8.98615956e-01 5.84412932e-01
-1.13740444e+00 -5.01980960e-01 -4.26739335e-01 7.10409164e-01
-5.17843246e-01 1.62094384e-01 -8.55559468e-01 -1.11496079e+00
-2.49585688e-01 -9.83473361e-01 7.53301203e-01 1.08347647e-01
-2.71479487e-01 -1.23516464e+00 3.69983763e-01 -3.83279085e-01
4.29823846e-01 6.46752596e-01 8.80731225e-01 -6.59891307e-01
-3.01692158e-01 -3.02138388e-01 1.72591016e-01 6.14684403e-01
3.04667622e-01 9.95804891e-02 -7.13005185e-01 -1.96164206e-01
2.44331151e-01 -3.24915722e-03 7.85710394e-01 5.18365145e-01
1.12594378e+00 -3.86921883e-01 -2.53194541e-01 6.11234486e-01
1.35160685e+00 6.79050207e-01 6.32645965e-01 -2.31689196e-02
5.92449605e-01 3.94009262e-01 2.23215923e-01 7.01462984e-01
3.32240641e-01 3.85073990e-01 3.42169076e-01 1.30083874e-01
2.21524015e-01 -3.78297031e-01 7.91901708e-01 1.57660270e+00
-1.10482864e-01 -3.48072678e-01 -8.34481239e-01 6.78273320e-01
-2.07280612e+00 -1.30265093e+00 -1.95084885e-01 2.07864285e+00
5.93696594e-01 8.56258571e-02 8.94799903e-02 1.77607998e-01
6.08780563e-01 3.71511310e-01 -1.02944839e+00 -4.41372454e-01
-1.40247047e-01 7.16179460e-02 1.88509360e-01 1.82778671e-01
-8.69957745e-01 4.45383370e-01 6.13734531e+00 5.62990963e-01
-1.44331503e+00 -9.15835574e-02 3.40452760e-01 5.82030900e-02
-4.53437157e-02 -1.07360721e-01 -8.49527121e-01 9.92597938e-01
1.27157152e+00 -3.05337995e-01 4.12008494e-01 7.60853112e-01
4.99177545e-01 4.00884956e-01 -1.08145499e+00 9.34701145e-01
-5.41662574e-02 -1.06408155e+00 1.61091179e-01 1.12879962e-01
8.92795265e-01 2.20548902e-02 1.87038258e-01 7.02975810e-01
2.37041444e-01 -8.58312666e-01 5.75453162e-01 1.06115949e+00
4.87587869e-01 -4.94513571e-01 7.32860744e-01 6.54498816e-01
-1.54007876e+00 -4.09440935e-01 -2.35159740e-01 -2.68118709e-01
2.36823276e-01 2.73680747e-01 -5.33084452e-01 6.67394817e-01
5.48193693e-01 1.32924569e+00 -1.75703526e-01 6.82033658e-01
1.07570197e-02 8.55673552e-01 -2.97765106e-01 1.41305721e-03
5.06491184e-01 -3.35333884e-01 4.39393193e-01 9.10612583e-01
6.85585856e-01 1.51590370e-02 9.56446677e-02 9.70824063e-01
4.41949278e-01 -2.98497796e-01 -5.80834389e-01 -3.12134564e-01
4.24540102e-01 8.04693758e-01 -3.95918101e-01 -5.40517926e-01
-3.15607518e-01 8.91047776e-01 1.40953660e-01 6.31385386e-01
-1.13315666e+00 -2.97863960e-01 7.89876461e-01 -2.90560648e-02
6.47859454e-01 -4.44539398e-01 1.30431518e-01 -1.42587733e+00
-1.22480758e-01 -8.23075891e-01 3.41132939e-01 -8.67865443e-01
-1.32406080e+00 9.02949810e-01 9.20967683e-02 -1.72746372e+00
-8.92936468e-01 -4.19311941e-01 -7.33309507e-01 9.51239586e-01
-1.34576130e+00 -6.55385554e-01 -2.93354362e-01 5.49962699e-01
6.96932733e-01 -3.40785086e-02 7.65660346e-01 2.32928768e-01
-9.60692108e-01 1.65606052e-01 5.86693227e-01 3.05761755e-01
2.22872078e-01 -8.93601477e-01 4.27093506e-01 8.36523235e-01
-3.52773368e-02 7.87157297e-01 6.98051095e-01 -7.51383841e-01
-1.12265503e+00 -1.08078277e+00 6.57653809e-01 -2.41567969e-01
7.62573242e-01 -2.02856824e-01 -1.28856933e+00 8.08709264e-01
9.94597226e-02 1.41565770e-01 3.91945630e-01 -3.89011711e-01
-2.03702405e-01 -2.19565421e-01 -6.16445541e-01 3.84439737e-01
8.98955047e-01 -5.85502625e-01 -6.94862366e-01 -2.01099902e-01
7.38361657e-01 -2.47658685e-01 -1.03519261e+00 6.07868314e-01
5.58187306e-01 -7.71451354e-01 7.68411994e-01 -7.41960645e-01
4.27448988e-01 -3.97692144e-01 -3.68215069e-02 -1.50332165e+00
-4.46758151e-01 -7.08304405e-01 -7.15104222e-01 1.14969552e+00
2.02611759e-01 -9.35526550e-01 2.65316218e-01 4.41349626e-01
-2.34206006e-01 -1.18786597e+00 -5.81405282e-01 -1.19809866e+00
-5.16814888e-02 -4.15687531e-01 8.96443486e-01 1.03045619e+00
-2.35206559e-01 1.59746379e-01 -7.64137268e-01 2.87566990e-01
2.23765627e-01 2.50473171e-01 6.52317762e-01 -1.31610787e+00
-3.92074078e-01 -2.69288868e-01 -3.09585065e-01 -1.29100406e+00
2.82179683e-01 -6.85668588e-01 -1.61198452e-01 -1.39321136e+00
-1.56244576e-01 -3.18808675e-01 -6.77326262e-01 3.03539753e-01
-6.72848746e-02 -3.51964116e-01 2.17186064e-01 5.36278963e-01
-2.31185630e-01 9.42461908e-01 9.30354059e-01 9.43417773e-02
-5.15775859e-01 1.17986135e-01 -2.11996540e-01 2.70108908e-01
8.68168235e-01 -4.88604397e-01 -6.24738514e-01 -1.18179217e-01
-1.39204320e-02 4.72557992e-01 2.76948541e-01 -1.09944308e+00
1.55137688e-01 -1.15326926e-01 3.64804149e-01 -9.51066852e-01
5.99473059e-01 -8.15157950e-01 5.58071434e-01 7.33031690e-01
-4.53245372e-01 4.38641608e-01 1.65156171e-01 7.67098129e-01
-3.00337613e-01 -2.46593617e-02 4.71402168e-01 -5.70545495e-02
-5.45483291e-01 5.53761184e-01 -5.22529304e-01 1.37142884e-02
1.04900980e+00 -7.77655393e-02 -1.32235542e-01 -7.66152263e-01
-7.33351231e-01 2.97360003e-01 -7.31015578e-02 6.16274655e-01
3.93947899e-01 -1.46112549e+00 -4.61996198e-01 4.51335788e-01
-1.37101471e-01 -3.20692688e-01 3.43312383e-01 8.50599647e-01
-1.93633154e-01 7.12673604e-01 -1.74315482e-01 -6.76362395e-01
-6.63117766e-01 8.32702994e-01 4.62837815e-01 -5.49088657e-01
-6.31063044e-01 2.18383908e-01 1.47097215e-01 -3.05823028e-01
1.94143265e-01 -8.75751257e-01 -3.09276640e-01 1.80215612e-01
3.63057941e-01 5.73692203e-01 -4.39860076e-01 -8.06933641e-01
-1.59923881e-01 6.20029926e-01 1.26401141e-01 1.21924365e-02
1.54091465e+00 -5.47521524e-02 1.03448354e-01 1.28377295e+00
1.46679318e+00 -6.03015006e-01 -1.64160609e+00 -2.21762538e-01
-2.45160535e-01 1.19238324e-01 -1.81501433e-01 -3.88940692e-01
-1.04589045e+00 1.08519030e+00 4.74105656e-01 5.20028949e-01
1.08829045e+00 -4.03456539e-01 1.02730215e+00 2.33488947e-01
1.55502751e-01 -7.83732295e-01 1.08601496e-01 5.81752717e-01
8.59890401e-01 -8.74565542e-01 -2.84387171e-01 2.00819835e-01
-8.14923644e-01 1.25739884e+00 3.89665633e-01 -4.59976763e-01
8.84717166e-01 1.33581176e-01 -7.58910105e-02 -8.92792828e-03
-1.21524131e+00 -1.82536066e-01 2.74390519e-01 2.58818358e-01
1.91437796e-01 -1.34779498e-01 -7.90805183e-03 7.30041444e-01
-2.40517547e-04 9.05855447e-02 4.05753493e-01 6.79840624e-01
-4.97691892e-02 -7.68105149e-01 -3.15185517e-01 4.65530574e-01
-3.11799705e-01 2.09151462e-01 9.68787596e-02 7.77450383e-01
7.53871212e-03 7.39648819e-01 3.91955167e-01 -3.52690130e-01
2.26925224e-01 3.45620781e-01 -2.12424528e-02 -2.76790708e-01
-5.53661287e-01 6.21884614e-02 -4.52294141e-01 -3.54866445e-01
-2.71106929e-01 -8.09636593e-01 -1.28219259e+00 -1.85175762e-01
-2.54346907e-01 1.89346075e-01 4.96928811e-01 9.58588600e-01
7.31039703e-01 8.80731225e-01 9.99816239e-01 -8.09416115e-01
-7.92398334e-01 -1.06204677e+00 -4.58633184e-01 2.87588388e-01
7.88230956e-01 -6.87743843e-01 -7.09152222e-01 4.89764884e-02] | [7.001523494720459, 3.1326916217803955] |
bc0448ca-0785-4d8f-b64f-87018ec9c982 | importance-weighted-structure-learning-for | 2205.07017 | null | https://arxiv.org/abs/2205.07017v1 | https://arxiv.org/pdf/2205.07017v1.pdf | Importance Weighted Structure Learning for Scene Graph Generation | Scene graph generation is a structured prediction task aiming to explicitly model objects and their relationships via constructing a visually-grounded scene graph for an input image. Currently, the message passing neural network based mean field variational Bayesian methodology is the ubiquitous solution for such a task, in which the variational inference objective is often assumed to be the classical evidence lower bound. However, the variational approximation inferred from such loose objective generally underestimates the underlying posterior, which often leads to inferior generation performance. In this paper, we propose a novel importance weighted structure learning method aiming to approximate the underlying log-partition function with a tighter importance weighted lower bound, which is computed from multiple samples drawn from a reparameterizable Gumbel-Softmax sampler. A generic entropic mirror descent algorithm is applied to solve the resulting constrained variational inference task. The proposed method achieves the state-of-the-art performance on various popular scene graph generation benchmarks. | ['Josef Kittler', 'Miroslaw Bober', 'Daqi Liu'] | 2022-05-14 | null | null | null | null | ['scene-graph-generation'] | ['computer-vision'] | [ 5.54370224e-01 4.64421958e-01 -5.81991561e-02 -3.23752195e-01
-1.06363094e+00 -5.11377603e-02 9.11091745e-01 1.15764081e-01
-2.66811609e-01 9.54451263e-01 2.11330906e-01 -3.13475169e-02
-1.55666798e-01 -7.52832711e-01 -1.06987965e+00 -9.42515850e-01
3.47125113e-01 6.97462440e-01 -8.76656845e-02 3.30407977e-01
3.96129161e-01 1.24559544e-01 -1.47004449e+00 -1.39516182e-02
1.00773942e+00 9.30581331e-01 6.37102723e-01 6.50260627e-01
-8.61836821e-02 9.43260372e-01 -2.19120175e-01 -7.32653916e-01
-6.47024885e-02 -4.60017979e-01 -7.89512992e-01 2.99407125e-01
6.26377583e-01 -3.93373966e-01 -4.18728113e-01 1.45614409e+00
9.31146964e-02 5.82462013e-01 1.21576309e+00 -1.05647957e+00
-6.67421997e-01 6.92783415e-01 -7.61282623e-01 -1.52979046e-01
1.07034527e-01 8.44657868e-02 1.37652707e+00 -7.59631693e-01
7.17580259e-01 1.30553806e+00 1.42187417e-01 3.37235957e-01
-1.92550111e+00 -1.32989898e-01 3.65153462e-01 3.56940389e-01
-1.63726306e+00 -2.20139846e-01 1.07887292e+00 -6.34346128e-01
5.45602441e-01 3.92359346e-02 6.71701372e-01 1.31045949e+00
1.32498518e-01 1.02506149e+00 7.95441329e-01 -1.92476079e-01
5.58134794e-01 1.69044986e-01 -9.72008482e-02 8.72959912e-01
4.50251639e-01 -2.95065939e-01 -5.30549824e-01 -4.11995351e-01
6.28315747e-01 -9.69141126e-02 -3.19992691e-01 -6.44935906e-01
-9.63679135e-01 1.07565355e+00 5.57740092e-01 -3.59583437e-01
-5.03747582e-01 5.78776240e-01 1.37774616e-01 -5.24587691e-01
6.52121723e-01 -4.50321697e-02 3.25352140e-02 2.79182464e-01
-1.10710609e+00 6.47090793e-01 7.96592593e-01 9.52023208e-01
1.00070572e+00 2.00314909e-01 -6.67613268e-01 6.37676656e-01
1.04324329e+00 4.11656260e-01 -1.67705700e-01 -1.14672947e+00
2.03695878e-01 2.79146791e-01 2.52726793e-01 -9.50775981e-01
2.69669265e-01 -3.48682642e-01 -1.12610376e+00 2.86324412e-01
5.42276442e-01 -3.11250072e-02 -8.51226747e-01 1.96104503e+00
6.11235559e-01 4.94771719e-01 -2.10628405e-01 9.06058669e-01
5.69148481e-01 9.05982792e-01 1.44717410e-01 -3.12152952e-01
1.19057930e+00 -7.75187373e-01 -5.82334459e-01 -1.11131318e-01
-6.16952330e-02 -6.09608173e-01 8.48077059e-01 5.06639540e-01
-1.04545522e+00 -4.32562202e-01 -1.02618408e+00 -1.68748781e-01
7.75108337e-02 -1.05123974e-01 6.15474463e-01 5.16730070e-01
-8.52501869e-01 7.55464613e-01 -8.42030287e-01 -1.66987162e-02
5.67407966e-01 -5.03195934e-02 -5.32438755e-02 -1.05203003e-01
-7.35660434e-01 5.09069264e-01 6.12430036e-01 3.15527499e-01
-1.32718790e+00 -5.66790342e-01 -1.00436902e+00 1.12329863e-01
5.20522535e-01 -1.19941926e+00 1.06395423e+00 -4.54112291e-01
-1.63051164e+00 7.45872021e-01 -3.52138370e-01 -6.21827781e-01
6.64543211e-01 -3.49781252e-02 4.56511140e-01 -1.20638266e-01
-8.19519013e-02 7.14512646e-01 1.33796322e+00 -1.53995728e+00
-2.78404802e-01 -2.17332289e-01 -5.30226668e-03 2.65543193e-01
1.54681876e-01 -5.93873382e-01 -4.75147426e-01 -4.28149164e-01
-4.85528260e-02 -6.90744340e-01 -4.63359803e-01 8.51079524e-02
-7.92260408e-01 -3.56780827e-01 3.49399954e-01 -5.65614462e-01
9.17631090e-01 -1.65733802e+00 4.86492604e-01 1.51224762e-01
3.96639436e-01 -2.88985491e-01 1.75862253e-01 3.19306523e-01
2.93678582e-01 -1.45337641e-01 -6.40071034e-01 -7.83538759e-01
3.94672632e-01 1.27738401e-01 -3.55976820e-01 7.32728958e-01
2.21438892e-02 7.52773404e-01 -1.02924192e+00 -7.14581490e-01
4.21522975e-01 6.79877937e-01 -8.77687752e-01 4.32548732e-01
-7.57631361e-01 4.12883461e-01 -3.79919976e-01 1.04964800e-01
8.85131419e-01 -5.68190455e-01 2.24457771e-01 -1.56803265e-01
2.26247102e-01 -1.52032435e-01 -1.42740154e+00 2.01523423e+00
-8.49092528e-02 6.55592144e-01 -8.63528401e-02 -1.07215166e+00
7.78656662e-01 1.82768144e-02 3.61589015e-01 1.40820295e-01
1.27991363e-01 -2.35176563e-01 -3.99506032e-01 -2.92778641e-01
6.27352655e-01 -2.26016238e-01 2.06309587e-01 1.49412557e-01
1.07297286e-01 -4.45486695e-01 1.06435806e-01 4.76145655e-01
7.66457736e-01 6.06428564e-01 2.84802258e-01 -2.98360020e-01
5.49085140e-01 -3.01846236e-01 3.95712763e-01 1.03884017e+00
1.28955320e-01 7.84172535e-01 5.85129857e-01 -1.55432373e-01
-1.11551607e+00 -1.55163395e+00 -2.12807238e-01 7.96685874e-01
2.42704570e-01 -2.97845721e-01 -9.68801975e-01 -4.70669240e-01
-1.30794078e-01 1.06673002e+00 -5.16079783e-01 -1.50625676e-01
6.80915490e-02 -7.98224628e-01 1.41090602e-01 -1.17436029e-01
3.60600352e-01 -9.06891346e-01 -3.08655351e-01 3.82854193e-01
-4.34144318e-01 -1.06296217e+00 -5.86605310e-01 -2.51110822e-01
-7.00447559e-01 -9.42084908e-01 -1.02622354e+00 -4.17616606e-01
7.01181829e-01 -1.88459888e-01 1.10863626e+00 -3.57500136e-01
-2.87361503e-01 2.18390599e-01 9.13583115e-02 -1.12378985e-01
-3.19220990e-01 -1.96112692e-01 2.94300802e-02 6.12138987e-01
2.35936865e-02 -4.98542756e-01 -7.19292045e-01 -3.55541319e-01
-9.61248100e-01 5.13681829e-01 3.65304410e-01 9.10726011e-01
9.51777279e-01 -1.22075088e-01 8.06419253e-02 -8.40230227e-01
5.82207859e-01 -5.16960561e-01 -1.02283359e+00 2.59441435e-01
-5.26694536e-01 5.89092076e-01 4.03911442e-01 -4.04735416e-01
-1.50448954e+00 1.47193342e-01 -6.38888311e-03 -6.21606112e-01
-2.18673587e-01 5.98506093e-01 -2.44835261e-02 1.04012109e-01
4.62752223e-01 5.04106581e-01 -2.97944546e-01 -3.02275300e-01
6.54949844e-01 2.00272754e-01 6.68019891e-01 -1.01363134e+00
6.66620672e-01 7.04270482e-01 4.50066239e-01 -8.91119361e-01
-1.05168509e+00 -2.61865616e-01 -2.88688660e-01 -4.66170609e-01
1.05897927e+00 -6.98024929e-01 -1.02508950e+00 3.26154947e-01
-1.54580379e+00 -4.41861033e-01 -3.19413394e-01 5.76440573e-01
-9.20946598e-01 7.77148247e-01 -3.63835931e-01 -1.37562883e+00
-2.29757026e-01 -1.23982739e+00 1.30741072e+00 2.43636385e-01
1.22144856e-02 -9.42235649e-01 3.59932184e-01 4.05123651e-01
-8.74803066e-02 7.18774557e-01 8.82884741e-01 -7.29847848e-02
-1.06096625e+00 -1.45757310e-02 -4.35939312e-01 1.98861778e-01
-2.16389075e-01 3.30682665e-01 -1.14497471e+00 -1.32424220e-01
-4.64768112e-02 -2.80547798e-01 1.05694842e+00 9.90956664e-01
1.24520481e+00 -3.62315506e-01 -1.75803885e-01 6.97159469e-01
1.61127484e+00 -3.41928601e-01 5.46119213e-01 -4.28507298e-01
8.95328999e-01 6.62768722e-01 2.25800544e-01 7.29842782e-01
4.96524125e-01 5.31690359e-01 8.38287711e-01 5.80097973e-01
7.53286704e-02 -7.07820654e-01 2.69911617e-01 5.38001239e-01
-6.31958395e-02 -4.89774376e-01 -6.64861143e-01 5.21979511e-01
-2.18365622e+00 -9.93200600e-01 -3.65204096e-01 2.32867265e+00
8.39741886e-01 2.63861507e-01 -2.30545953e-01 -3.38853300e-01
8.15498054e-01 3.69278997e-01 -6.06908798e-01 3.59878503e-02
1.47555217e-01 -6.02651238e-02 2.35114262e-01 9.07406807e-01
-8.59489501e-01 9.19937074e-01 5.37901402e+00 1.25908184e+00
-5.38576365e-01 1.07736342e-01 6.82728589e-01 -2.27305386e-02
-4.58396822e-01 2.60485977e-01 -7.65482247e-01 4.97024626e-01
5.79496861e-01 -2.98420191e-01 6.33522511e-01 9.39769626e-01
3.21580946e-01 -3.66797030e-01 -1.17587113e+00 1.23904836e+00
-2.74646338e-02 -1.61778665e+00 3.29880327e-01 4.31716204e-01
1.04013491e+00 1.95533708e-02 2.55178139e-02 1.02987260e-01
5.63295126e-01 -8.53172123e-01 9.55488026e-01 9.94100213e-01
6.30840003e-01 -6.96521521e-01 3.16125482e-01 5.64772308e-01
-1.17390549e+00 4.59173858e-01 -6.04204178e-01 2.33983174e-02
5.68462074e-01 8.38473082e-01 -6.09998643e-01 3.30954760e-01
4.48853195e-01 5.03668129e-01 -5.27231880e-02 1.14161134e+00
-2.79728025e-01 5.79453528e-01 -4.18458968e-01 -2.77806014e-01
2.57139683e-01 -8.06158602e-01 9.21631217e-01 1.05151761e+00
1.71036795e-01 -2.45413557e-01 1.32457569e-01 1.62453628e+00
-3.87945294e-01 -5.31261228e-02 -7.17723787e-01 4.67239283e-02
6.40525063e-03 1.37102616e+00 -8.71053398e-01 -4.50898528e-01
-7.41717592e-02 1.01701736e+00 6.39629781e-01 5.81328511e-01
-9.90256965e-01 -1.10165803e-02 6.92020237e-01 -1.65158406e-01
3.60081166e-01 -7.61812180e-02 -1.32632583e-01 -1.32198012e+00
-1.84826910e-01 -3.70950907e-01 1.94161326e-01 -1.01557469e+00
-1.34923422e+00 2.28909254e-01 3.27373803e-01 -8.10352206e-01
-3.17530870e-01 -5.60909033e-01 -7.32001245e-01 1.02427065e+00
-1.33268321e+00 -9.47139323e-01 -3.07131350e-01 5.27847230e-01
6.10460103e-01 2.10517138e-01 4.99510527e-01 -8.73470381e-02
-5.51211655e-01 1.76632777e-01 2.39226237e-01 -2.37103954e-01
3.05678785e-01 -1.58846533e+00 1.27782598e-01 8.84762168e-01
3.71797800e-01 4.04252499e-01 1.12370658e+00 -7.74832964e-01
-1.32024801e+00 -1.07644212e+00 4.47972357e-01 -4.12968129e-01
6.51629210e-01 -5.26531816e-01 -9.57412124e-01 4.90978211e-01
4.21416253e-01 -9.42227021e-02 2.17776597e-01 -1.80133715e-01
-1.28190205e-01 1.20726950e-01 -1.05702245e+00 9.16292012e-01
8.48242164e-01 -5.64227521e-01 -3.10230970e-01 5.41101635e-01
6.63679659e-01 -3.45004261e-01 -7.44302511e-01 -4.81604412e-03
4.82181370e-01 -8.79469573e-01 9.57396150e-01 -5.45186639e-01
7.45981097e-01 -3.41164947e-01 -3.35116833e-01 -1.14927256e+00
-3.09675545e-01 -1.01237500e+00 -6.33063793e-01 1.05270743e+00
2.20964178e-01 -2.58875787e-01 1.10927451e+00 5.20646572e-01
1.73010111e-01 -6.69264674e-01 -8.80330861e-01 -3.32285851e-01
-2.83147454e-01 -8.17342997e-01 2.40102649e-01 4.79488343e-01
-4.32910651e-01 6.56516492e-01 -6.81925833e-01 2.49901712e-01
1.78172338e+00 4.96462658e-02 8.57684433e-01 -1.29526281e+00
-6.66926265e-01 -4.94338453e-01 -2.10967630e-01 -1.42053270e+00
5.02425671e-01 -9.93267000e-01 4.41501468e-01 -1.65668941e+00
5.43917298e-01 2.55133718e-01 -5.77652864e-02 -3.82566512e-01
-4.48321700e-01 4.71705367e-04 9.75734591e-02 -8.62417221e-02
-6.40420735e-01 1.13530254e+00 1.34182274e+00 -3.97781342e-01
6.55287802e-02 1.86879665e-01 -5.19592047e-01 9.54809427e-01
3.42463464e-01 -7.17812836e-01 -7.23365426e-01 -2.38336384e-01
4.67978001e-01 2.03311533e-01 6.64116621e-01 -5.91016591e-01
4.62129414e-01 -3.80861431e-01 1.72348693e-01 -8.09184313e-01
6.66022778e-01 -4.95868653e-01 3.10722440e-01 2.68437058e-01
-3.24156284e-01 -6.72816575e-01 -2.80249357e-01 1.25543785e+00
6.83670491e-02 -5.36452174e-01 8.71008217e-01 -1.38631538e-01
-3.44207942e-01 6.55765653e-01 -3.38714302e-01 1.54894024e-01
6.01692259e-01 -1.20955676e-01 1.93631351e-01 -6.22023165e-01
-7.45340288e-01 1.07024178e-01 1.78391218e-01 -1.53835982e-01
8.40695381e-01 -1.38737130e+00 -1.01866794e+00 -1.88579291e-01
1.39279827e-01 3.67573291e-01 4.20367569e-01 7.02119231e-01
-4.36518073e-01 5.48202246e-02 3.37817341e-01 -9.19387877e-01
-8.24681282e-01 4.83656257e-01 2.87291318e-01 -4.93428826e-01
-5.72980106e-01 1.03429675e+00 5.68343043e-01 -2.05323413e-01
2.60873049e-01 -2.80345142e-01 -5.19950129e-02 -2.23077927e-02
2.82206178e-01 4.96271580e-01 -4.48631644e-01 -6.19234383e-01
4.33151461e-02 2.46728167e-01 8.88238624e-02 -4.03409600e-01
1.13235211e+00 -2.87955403e-01 -2.90947556e-02 5.01882315e-01
1.15303493e+00 -3.70214850e-01 -1.86026573e+00 -4.41579938e-01
-1.34045333e-01 -3.53902459e-01 3.21903676e-01 8.77493694e-02
-8.00053298e-01 8.39080811e-01 2.83157051e-01 3.80643636e-01
4.88598347e-01 7.59001151e-02 4.80058730e-01 5.20527303e-01
3.31624329e-01 -9.89159107e-01 -5.10766841e-02 2.36576736e-01
9.22608078e-01 -1.37484920e+00 2.28410050e-01 -2.12900385e-01
-4.28680331e-01 8.92857790e-01 2.85330981e-01 -4.59542692e-01
7.64888704e-01 -1.65358126e-01 -7.69357383e-01 -2.12019756e-01
-6.67759180e-01 -1.94554359e-01 5.69248497e-01 5.51952243e-01
1.26248270e-01 1.05094187e-01 -6.87119439e-02 2.32801080e-01
-1.81620270e-02 -1.34231731e-01 2.55335629e-01 3.33324730e-01
-3.14874768e-01 -5.77007413e-01 -2.70570874e-01 5.41820288e-01
-4.98770565e-01 -3.02181184e-01 -5.64727150e-02 2.89673895e-01
-1.83380648e-01 8.44433904e-01 -1.22761101e-01 1.91344887e-01
-1.93613581e-02 -1.35892943e-01 7.21303582e-01 -5.24804294e-01
1.83539063e-01 3.32113802e-01 -3.11685175e-01 -4.98784333e-01
-3.71381432e-01 -7.26963699e-01 -8.05149615e-01 -3.59491467e-01
-3.10722440e-01 3.40945274e-02 5.71717203e-01 1.01394677e+00
-1.33362308e-01 5.43938160e-01 2.29261413e-01 -1.38700187e+00
-6.22191250e-01 -1.03278625e+00 -7.47455239e-01 4.49552178e-01
2.94367164e-01 -7.13992894e-01 -3.80023926e-01 3.99278969e-01] | [7.137632846832275, 3.6866416931152344] |
0ccc8f94-0e75-490f-bad9-a0073c48afad | an-instance-segmentation-dataset-of-yeast | 2304.07597 | null | https://arxiv.org/abs/2304.07597v2 | https://arxiv.org/pdf/2304.07597v2.pdf | An Instance Segmentation Dataset of Yeast Cells in Microstructures | Extracting single-cell information from microscopy data requires accurate instance-wise segmentations. Obtaining pixel-wise segmentations from microscopy imagery remains a challenging task, especially with the added complexity of microstructured environments. This paper presents a novel dataset for segmenting yeast cells in microstructures. We offer pixel-wise instance segmentation labels for both cells and trap microstructures. In total, we release 493 densely annotated microscopy images. To facilitate a unified comparison between novel segmentation algorithms, we propose a standardized evaluation strategy for our dataset. The aim of the dataset and evaluation strategy is to facilitate the development of new cell segmentation approaches. The dataset is publicly available at https://christophreich1996.github.io/yeast_in_microstructures_dataset/ . | ['Heinz Koeppl', 'André O. Françani', 'Tim Prangemeier', 'Christoph Reich'] | 2023-04-15 | null | null | null | null | ['panoptic-segmentation', 'cell-segmentation'] | ['computer-vision', 'medical'] | [ 4.36072826e-01 -2.91240066e-01 3.76962930e-01 -3.21170330e-01
-9.00962234e-01 -8.07608604e-01 2.04107776e-01 1.93326846e-01
-6.79838061e-01 1.10321772e+00 -6.79390371e-01 -4.87911329e-02
1.38880491e-01 -5.75314641e-01 -7.01241612e-01 -8.79903138e-01
1.90352678e-01 7.48545825e-01 3.41020823e-01 5.57339311e-01
6.19842649e-01 7.38556981e-01 -1.15792382e+00 2.19009399e-01
6.12263203e-01 8.19158554e-01 9.76356626e-01 1.13348305e+00
-7.82623366e-02 1.19913086e-01 -6.13894463e-01 6.33356422e-02
1.12984978e-01 -3.96162271e-01 -1.23318303e+00 3.79365832e-01
1.63610429e-01 -1.30841499e-02 2.44215012e-01 8.72812152e-01
3.57568324e-01 -1.37705714e-01 9.16965306e-01 -9.71588314e-01
-3.65213990e-01 1.77919552e-01 -4.97088909e-01 4.48785454e-01
4.08724956e-02 3.37047815e-01 6.51013255e-01 -8.14996481e-01
1.25906777e+00 7.00332999e-01 4.49964672e-01 5.97000122e-01
-1.44905734e+00 -9.30979550e-02 -9.29689035e-02 1.22632524e-02
-1.43743742e+00 -5.70936859e-01 5.61905563e-01 -5.95613956e-01
6.78110659e-01 2.11156547e-01 6.61785364e-01 6.21661484e-01
1.79827452e-01 6.87348545e-01 1.32886314e+00 -2.21030399e-01
4.27623272e-01 -2.08153084e-01 1.63372323e-01 6.65795326e-01
4.27579820e-01 -4.17321622e-01 -3.90788913e-01 1.86850473e-01
1.32456505e+00 1.93697941e-02 -2.28696182e-01 -2.26137459e-01
-1.65862310e+00 2.20842600e-01 -5.96793555e-03 5.52776158e-01
-4.06306572e-02 9.50642452e-02 1.16842173e-01 -1.06655501e-01
3.69209915e-01 5.37536800e-01 -6.09710038e-01 -6.57260269e-02
-1.21141720e+00 2.73319006e-01 5.14938354e-01 8.98160219e-01
1.07195437e+00 -3.75541985e-01 2.98213065e-01 6.28783643e-01
1.04307480e-01 3.24514061e-01 2.88244635e-01 -1.51981485e+00
-2.82803059e-01 6.33028746e-01 9.88530964e-02 -6.08470142e-01
-6.76225960e-01 6.93095252e-02 -7.18934596e-01 1.11067452e-01
1.12481821e+00 -7.45137408e-02 -8.85967135e-01 1.21663368e+00
4.78606492e-01 -1.27921671e-01 -2.92180270e-01 7.38296509e-01
8.02740157e-01 2.99873143e-01 -2.51244068e-01 -1.54055789e-01
1.37254620e+00 -6.80264413e-01 -6.59089684e-01 2.64960140e-01
5.27951956e-01 -7.95976281e-01 9.75349009e-01 3.33596438e-01
-1.02767539e+00 -4.06265587e-01 -7.75536001e-01 -4.22855765e-01
-8.05292368e-01 2.46012807e-01 5.46069264e-01 2.25280225e-01
-1.22009408e+00 4.97070789e-01 -1.13375056e+00 -6.60043120e-01
8.10581923e-01 5.08386433e-01 -5.71261287e-01 3.06516647e-01
-1.45275384e-01 5.25760055e-01 3.19083154e-01 -1.98529556e-01
-9.56575513e-01 -7.88137853e-01 -5.71186662e-01 -5.14573336e-01
1.38411552e-01 -8.33251834e-01 1.12303865e+00 -1.02473214e-01
-1.36594093e+00 1.68383420e+00 -5.79099715e-01 -2.35775679e-01
5.59812009e-01 2.75235295e-01 3.23909909e-01 4.65728372e-01
1.78679213e-01 1.05270875e+00 1.86289638e-01 -1.58466220e+00
-7.91126847e-01 -6.88338518e-01 -2.45390132e-01 -3.10852110e-01
2.00126782e-01 -3.15416694e-01 -6.03116453e-01 -3.89991671e-01
3.22635658e-02 -5.92977345e-01 -2.83003628e-01 6.45954683e-02
-9.03677940e-01 1.00895174e-01 9.71172035e-01 -6.07202291e-01
5.63516378e-01 -1.82195961e+00 3.69926691e-01 -2.28223905e-01
4.62983072e-01 -3.10550213e-01 1.05047956e-01 3.08892298e-02
4.70072091e-01 2.77899683e-01 -8.05210650e-01 -8.06153774e-01
-1.57976806e-01 1.47241518e-01 2.29129344e-01 6.87970877e-01
2.21898660e-01 8.79201770e-01 -7.82180488e-01 -9.78812754e-01
4.68375176e-01 4.78227168e-01 -3.87245417e-01 3.27159315e-01
-4.37906891e-01 1.30929971e+00 -1.77873760e-01 1.37886155e+00
7.11045921e-01 -4.92331445e-01 7.65230581e-02 -1.90049738e-01
-5.34207463e-01 -3.24147403e-01 -8.42708170e-01 1.80982888e+00
5.67087717e-02 5.82626045e-01 5.13855696e-01 -8.72082055e-01
7.16613650e-01 6.63679466e-02 8.56134772e-01 -2.14745641e-01
2.82696068e-01 5.29105484e-01 -4.28257912e-01 -2.52206057e-01
3.89990181e-01 4.83428920e-03 -2.05742382e-02 1.36524677e-01
2.49623761e-01 -7.16705322e-01 9.99783456e-01 -1.27253801e-01
1.00976694e+00 4.52555627e-01 2.17180818e-01 -6.38564408e-01
5.89269757e-01 4.29777682e-01 6.33604765e-01 4.90175039e-01
-5.53758979e-01 1.06948745e+00 4.73858267e-01 -5.15720665e-01
-1.28291142e+00 -1.05382323e+00 -5.81836402e-01 5.24970710e-01
3.50875765e-01 4.18489091e-02 -1.14749610e+00 -3.30724627e-01
-6.50322586e-02 -8.01249370e-02 -6.32041156e-01 7.28661895e-01
-4.43700582e-01 -1.14717174e+00 5.28482854e-01 7.08345026e-02
6.18210375e-01 -1.01561201e+00 -7.45502293e-01 5.09168133e-02
-2.82074898e-01 -1.27529657e+00 -1.87493026e-01 5.93603253e-01
-7.48926997e-01 -1.44120562e+00 -1.03843594e+00 -1.08894682e+00
1.00483263e+00 9.85712186e-02 1.06951499e+00 1.80390507e-01
-7.57694900e-01 2.68164814e-01 9.30998400e-02 -2.96872377e-01
4.53074910e-02 3.81642342e-01 -2.24169008e-02 -5.15823781e-01
1.46938488e-01 -6.00258052e-01 -7.71593213e-01 3.64352703e-01
-9.77562845e-01 8.79583880e-02 2.15727687e-01 7.17312038e-01
1.67595208e+00 2.93766558e-02 2.51305401e-01 -1.23365521e+00
2.11443022e-01 -2.52528131e-01 -9.03510273e-01 7.88129568e-02
-4.24408652e-02 -3.81465793e-01 7.71403849e-01 -8.76959860e-02
-7.36046255e-01 4.04584646e-01 -1.80653647e-01 5.38678169e-02
-9.65098262e-01 1.46242768e-01 -2.39727899e-01 -2.88469702e-01
1.66593120e-01 3.03557873e-01 6.02579191e-02 -5.47274053e-01
8.79332498e-02 2.90612012e-01 8.17181289e-01 -7.88353145e-01
3.80064458e-01 1.02084303e+00 2.18239903e-01 -1.17108405e+00
-3.70833337e-01 -7.96262145e-01 -1.16715908e+00 -2.63935238e-01
1.09457970e+00 -3.45901430e-01 -8.01203907e-01 8.23074341e-01
-1.04840815e+00 -1.01204598e+00 -4.43048418e-01 -6.62523210e-02
-8.99960220e-01 3.52495641e-01 -1.01670468e+00 -5.50433040e-01
1.13500571e-02 -1.49681735e+00 1.41460454e+00 4.49276060e-01
-2.98984826e-01 -1.29986405e+00 2.03445137e-01 4.94618893e-01
2.41540298e-02 5.43724835e-01 7.44787097e-01 -3.53119522e-01
-9.01692331e-01 1.12770505e-01 -1.19274333e-01 -1.56513020e-01
4.26563054e-01 6.81544065e-01 -1.08636224e+00 7.01010926e-03
-1.86633423e-01 -2.47161135e-01 1.05911446e+00 7.85250068e-01
1.41067255e+00 3.21752280e-01 -6.40779555e-01 8.57809007e-01
1.53448522e+00 2.16304138e-01 3.58355165e-01 4.71211076e-01
6.45299971e-01 6.24168694e-01 3.59512836e-01 3.62292349e-01
2.72902369e-01 4.27927613e-01 2.80941218e-01 -2.77494878e-01
-1.92176983e-01 4.64818418e-01 -7.57076442e-02 6.24244273e-01
-1.41273052e-01 -2.63993025e-01 -1.03124392e+00 6.77306533e-01
-1.35093594e+00 -5.77009857e-01 -3.06292206e-01 1.81982613e+00
9.01721001e-01 -2.44398147e-01 3.48477870e-01 1.96660012e-01
7.00424433e-01 -3.46651971e-01 -6.64329112e-01 9.49116498e-02
-5.06306648e-01 1.00026436e-01 3.60178739e-01 7.94617116e-01
-1.26333249e+00 9.59720790e-01 6.46073961e+00 8.57970774e-01
-9.14407849e-01 -1.21603698e-01 1.19886529e+00 5.45217423e-04
-2.29781047e-02 -2.08450153e-01 -1.07535899e+00 6.62445664e-01
6.74584627e-01 5.22519685e-02 3.26791316e-01 2.22777516e-01
4.16119903e-01 -6.26126170e-01 -1.05804920e+00 7.55766332e-01
-4.48251337e-01 -1.56022525e+00 -3.82472798e-02 5.19522250e-01
7.66948581e-01 -1.81323010e-02 1.46431327e-01 -4.74003315e-01
1.66195869e-01 -9.15466726e-01 6.07702911e-01 6.20664716e-01
8.94933462e-01 -4.58097696e-01 8.01807821e-01 9.43090618e-02
-1.15912807e+00 5.31195343e-01 -4.79494631e-01 4.66275997e-02
3.10145706e-01 1.05239427e+00 -7.84957647e-01 4.16696608e-01
7.09487736e-01 7.26920605e-01 -7.75619984e-01 1.05113411e+00
3.77965838e-01 3.12189043e-01 -5.31533360e-01 2.83700615e-01
-1.67484120e-01 -6.70133293e-01 3.14293534e-01 1.57077014e+00
3.06308210e-01 -1.81377381e-01 1.95423648e-01 1.21830237e+00
-2.34299734e-01 -1.43868625e-01 -6.60129249e-01 -2.82772958e-01
4.50075895e-01 1.69778252e+00 -1.77428162e+00 -2.35899210e-01
1.27873391e-01 9.21602607e-01 5.05850911e-01 3.70829612e-01
-5.12495100e-01 -3.05474311e-01 7.14239657e-01 1.62585333e-01
1.07846722e-01 -5.50180852e-01 -8.03671002e-01 -7.98952460e-01
-3.21180880e-01 -2.83331841e-01 3.07460129e-01 -5.55107415e-01
-1.21186185e+00 1.35145634e-01 -1.95976883e-01 -5.44449985e-01
2.17996106e-01 -1.03444004e+00 -3.54083270e-01 3.73852253e-01
-1.38375950e+00 -1.16663265e+00 -4.12418067e-01 2.85923213e-01
5.28306007e-01 2.81360298e-01 9.23332095e-01 1.78832173e-01
-9.66209590e-01 -1.72328338e-01 3.16237271e-01 1.06977648e-03
5.24228811e-01 -1.75021207e+00 1.86030552e-01 6.21052444e-01
-4.59774621e-02 7.91399777e-01 6.84552729e-01 -5.04939914e-01
-1.20785224e+00 -1.07366121e+00 6.17380202e-01 -7.13973999e-01
5.34573257e-01 -2.69375324e-01 -7.28768885e-01 5.76475859e-01
1.89754874e-01 2.45464556e-02 9.30625319e-01 -5.19606054e-01
3.49457473e-01 1.78966358e-01 -1.53895509e+00 5.91640353e-01
9.36013758e-01 -4.67584759e-01 7.55787045e-02 5.06411016e-01
3.11388165e-01 -5.01759887e-01 -1.25361454e+00 3.55895132e-01
3.49170327e-01 -9.84739959e-01 8.84721339e-01 1.31605461e-01
3.98565620e-01 -6.65900290e-01 -2.39409674e-02 -8.53249788e-01
-8.72669816e-02 -4.55335230e-01 -1.25582695e-01 1.21786714e+00
4.09958780e-01 -4.14506167e-01 1.06960714e+00 2.90153205e-01
-3.68012816e-01 -8.17261994e-01 -1.00164437e+00 -6.88922763e-01
3.73029619e-01 4.58351038e-02 4.36148822e-01 6.92041695e-01
2.77004633e-02 -2.23489851e-02 3.33851635e-01 -1.32071450e-01
1.02807391e+00 2.35165566e-01 7.84817338e-01 -1.08586955e+00
2.05116913e-01 -6.91065848e-01 -2.27623224e-01 -1.01376188e+00
6.39776513e-02 -7.25068033e-01 1.70728520e-01 -1.70635200e+00
4.59883898e-01 -1.15493588e-01 -1.13066211e-01 2.29949370e-01
-7.17689842e-02 8.49559665e-01 -1.74651831e-01 3.61627668e-01
-9.50433791e-01 -6.43661851e-03 1.56066465e+00 -2.02122033e-01
5.75285703e-02 -4.01969284e-01 -2.05779225e-01 7.22936690e-01
1.33600342e+00 -7.69056305e-02 2.38889363e-02 -3.66846681e-01
-3.69186819e-01 -3.41802627e-01 4.66822267e-01 -1.26603234e+00
3.37876678e-01 -2.88062572e-01 6.19143367e-01 -8.01448345e-01
4.52654630e-01 -5.67644536e-01 3.34049284e-01 4.35871184e-01
-8.20093751e-02 -1.45578384e-01 1.29064173e-01 3.26544195e-01
-1.67351574e-01 -7.57573247e-02 1.08321738e+00 -7.24430501e-01
-6.25902712e-01 3.09482366e-01 -8.09733868e-01 1.10945001e-01
1.07542408e+00 -6.22454047e-01 -3.97648662e-01 3.44280392e-01
-8.03315878e-01 5.81100918e-02 1.42834592e+00 -5.77917635e-01
5.11463583e-01 -7.95915246e-01 -3.00605923e-01 -1.23661570e-02
-2.19509304e-02 5.92035830e-01 1.97119638e-01 9.19677258e-01
-1.23952270e+00 5.42597830e-01 -4.52641606e-01 -8.43617380e-01
-1.15351415e+00 2.98935473e-01 3.37588102e-01 -3.05517256e-01
-2.09462240e-01 1.03354001e+00 2.62973338e-01 -5.80939949e-01
-1.54555067e-01 -6.63171351e-01 -8.42935592e-02 -2.37462267e-01
3.49189579e-01 5.39717197e-01 -1.15206838e-01 -5.58300912e-01
-3.99027765e-01 7.13757455e-01 1.09893531e-01 2.26182733e-02
1.20919132e+00 -3.76439840e-01 -6.28973186e-01 8.99750412e-01
9.42217231e-01 -1.94311783e-01 -1.38684261e+00 4.15428251e-01
3.15888375e-01 -3.38043302e-01 -3.30369741e-01 -4.27685618e-01
-1.08040893e+00 7.95190096e-01 2.70799965e-01 2.63760775e-01
7.76256680e-01 7.78000429e-02 8.47069085e-01 3.14296722e-01
6.49734616e-01 -1.05781829e+00 -6.15127459e-02 5.49820364e-01
3.11016649e-01 -1.31094778e+00 7.56523572e-04 -6.59882009e-01
5.16705774e-03 1.00638258e+00 6.01220667e-01 9.95013565e-02
4.75781173e-01 7.31868446e-01 1.79939270e-01 -3.63925904e-01
-3.97331715e-01 -2.36085087e-01 -4.73438591e-01 9.93324697e-01
7.67810941e-01 3.37295011e-02 -3.19466777e-02 4.90798056e-01
-1.51952595e-01 1.07544400e-01 5.51956594e-01 1.00428438e+00
-5.68365812e-01 -1.19502771e+00 -4.03968066e-01 4.86773431e-01
-5.83617508e-01 2.84598470e-01 -6.90675974e-01 6.53259158e-01
3.90686989e-01 8.98874342e-01 2.02103823e-01 6.01605698e-02
-6.30752295e-02 -4.31102365e-02 6.45146608e-01 -6.11043870e-01
-1.87983066e-01 2.37258643e-01 -3.42737943e-01 -3.04848224e-01
-7.65513182e-01 -7.69454956e-01 -1.81679499e+00 -4.95606929e-01
1.07684806e-01 1.01608671e-01 4.14354265e-01 6.69241667e-01
3.56515825e-01 6.08326137e-01 9.58879888e-02 -1.42914963e+00
5.00225902e-01 -6.98658347e-01 -9.00731504e-01 3.85009021e-01
2.10304141e-01 -5.72089672e-01 -3.15117568e-01 1.01233113e+00] | [14.445114135742188, -3.1599833965301514] |
d6000174-cded-4a10-a209-95031103c292 | a-holistic-approach-to-polyphonic-music | 1910.12086 | null | https://arxiv.org/abs/1910.12086v1 | https://arxiv.org/pdf/1910.12086v1.pdf | A holistic approach to polyphonic music transcription with neural networks | We present a framework based on neural networks to extract music scores directly from polyphonic audio in an end-to-end fashion. Most previous Automatic Music Transcription (AMT) methods seek a piano-roll representation of the pitches, that can be further transformed into a score by incorporating tempo estimation, beat tracking, key estimation or rhythm quantization. Unlike these methods, our approach generates music notation directly from the input audio in a single stage. For this, we use a Convolutional Recurrent Neural Network (CRNN) with Connectionist Temporal Classification (CTC) loss function which does not require annotated alignments of audio frames with the score rhythmic information. We trained our model using as input Haydn, Mozart, and Beethoven string quartets and Bach chorales synthesized with different tempos and expressive performances. The output is a textual representation of four-voice music scores based on **kern format. Although the proposed approach is evaluated in a simplified scenario, results show that this model can learn to transcribe scores directly from audio signals, opening a promising avenue towards complete AMT. | ['Jorge Calvo-Zaragoza', 'Miguel A. Román', 'Antonio Pertusa'] | 2019-10-26 | null | null | null | null | ['music-transcription'] | ['music'] | [ 5.48338711e-01 -9.13705006e-02 8.61951411e-02 -1.62872165e-01
-1.36833036e+00 -8.72013807e-01 3.88917774e-01 -3.53360698e-02
-2.42258862e-01 4.72999543e-01 4.67330039e-01 8.36267918e-02
-3.19825917e-01 -4.07580823e-01 -5.14911354e-01 -6.29721344e-01
-1.15184888e-01 3.70831668e-01 -1.74562424e-01 -2.03590795e-01
2.34246612e-01 2.12063715e-01 -1.56087089e+00 7.25802481e-01
2.78969109e-01 1.25919652e+00 -7.20062852e-02 1.51313591e+00
1.93108544e-01 1.14159822e+00 -1.10373056e+00 -4.35195893e-01
4.00609076e-01 -9.54395294e-01 -7.41625071e-01 -3.05219442e-01
5.11667728e-01 -1.15792945e-01 -1.83600098e-01 6.82014167e-01
7.18737721e-01 8.98771882e-02 6.48895442e-01 -7.32158840e-01
-1.02679782e-01 1.24632013e+00 1.17621012e-02 -1.49597138e-01
4.67888594e-01 8.94442573e-03 1.51148999e+00 -8.14376831e-01
4.19395000e-01 8.78270745e-01 1.00663638e+00 1.25378594e-01
-1.45695198e+00 -6.41687095e-01 -4.83003795e-01 3.95290524e-01
-1.32355011e+00 -4.55654293e-01 1.01626503e+00 -3.15179944e-01
8.28717113e-01 6.68164134e-01 1.15388060e+00 1.13166320e+00
-1.38193250e-01 8.81970704e-01 7.69210041e-01 -6.79733276e-01
1.93064347e-01 -5.23801208e-01 -4.70204145e-01 2.61543691e-01
-8.30379963e-01 1.24986306e-01 -1.00101376e+00 -7.89035410e-02
6.73709631e-01 -3.77976090e-01 -2.45191410e-01 -2.67735254e-02
-1.75614190e+00 6.19461536e-01 3.66814673e-01 3.61735821e-01
-6.04979515e-01 5.90908825e-01 8.90022993e-01 5.91539502e-01
1.63403019e-01 5.41766644e-01 -2.81369537e-01 -5.58418512e-01
-1.73325527e+00 6.11660063e-01 8.00069571e-01 4.14670855e-01
2.85552502e-01 6.31007552e-01 -4.50497627e-01 8.66762161e-01
-2.23601624e-01 3.48764479e-01 7.91464984e-01 -1.14431870e+00
5.95046401e-01 -9.30291414e-02 1.46687984e-01 -8.93281221e-01
-2.77330428e-01 -5.76968968e-01 -6.44167960e-01 2.65558213e-01
5.39743125e-01 -1.77096188e-01 -6.38011873e-01 1.67244422e+00
-1.14910498e-01 3.56599420e-01 2.87286099e-02 9.97257352e-01
6.69852018e-01 9.51039195e-01 -5.28746068e-01 -2.97025025e-01
1.18013489e+00 -1.06339705e+00 -9.45282936e-01 5.10598481e-01
4.60880548e-01 -1.14748693e+00 1.21236646e+00 1.11487234e+00
-1.53321302e+00 -9.19276476e-01 -1.18810582e+00 -2.36285791e-01
1.44144207e-01 6.94545984e-01 1.63048729e-01 4.46680278e-01
-1.04585767e+00 1.12783813e+00 -6.79330945e-01 9.32336673e-02
-9.27399099e-02 4.30205464e-01 -3.21293175e-02 7.45221317e-01
-1.07691288e+00 4.94431585e-01 6.07076526e-01 1.17804296e-01
-9.27681327e-01 -6.73851728e-01 -7.07794368e-01 2.99591154e-01
-4.83486103e-03 -3.34522128e-01 1.80070138e+00 -1.28337955e+00
-2.34886479e+00 6.22045994e-01 6.23120740e-02 -9.17322516e-01
5.77511549e-01 -4.65386152e-01 -3.41059834e-01 3.22154522e-01
-3.53803933e-01 6.09651327e-01 9.46061194e-01 -6.29450858e-01
-5.15842557e-01 1.63039610e-01 -2.12038234e-01 2.83126980e-01
-2.17422828e-01 2.80740798e-01 -7.57899210e-02 -1.21293283e+00
4.28971909e-02 -1.02138495e+00 1.33355319e-01 -4.43004906e-01
-5.38043201e-01 8.70546512e-03 3.23514253e-01 -9.17346537e-01
1.60926127e+00 -2.07639813e+00 6.36822879e-01 1.71585366e-01
-3.71937603e-01 1.79241762e-01 -2.81151533e-01 6.25283480e-01
-1.40182093e-01 -3.79303843e-01 -1.66110232e-01 -3.85297686e-01
3.23699534e-01 -2.23626927e-01 -7.27267325e-01 1.34196997e-01
2.68698394e-01 8.80724728e-01 -7.49888301e-01 -2.20001459e-01
2.04341486e-01 6.32443249e-01 -6.71017945e-01 2.80231804e-01
-4.09235448e-01 6.93809748e-01 1.51624396e-01 2.70865440e-01
-5.38733974e-02 3.67511928e-01 1.84967056e-01 -1.79163933e-01
-3.80426675e-01 1.09699380e+00 -1.29305828e+00 2.16353703e+00
-7.55801857e-01 8.43709290e-01 -2.16347709e-01 -7.29604483e-01
1.29699695e+00 9.34975028e-01 5.54847717e-01 -2.88055420e-01
2.16365561e-01 3.77086163e-01 1.96265727e-01 -1.95294380e-01
6.89838231e-01 -1.82983190e-01 -3.35227817e-01 4.78552282e-01
3.34506422e-01 -5.51734447e-01 3.60851318e-01 -3.80669951e-01
1.14480674e+00 4.27407891e-01 1.96256712e-01 3.54043931e-01
4.20433670e-01 1.63896177e-02 3.62030238e-01 4.04017270e-01
3.20898682e-01 1.15051341e+00 5.12015820e-01 -6.26182854e-01
-1.32866538e+00 -1.17474210e+00 2.68261760e-01 1.32473087e+00
-7.50643313e-01 -9.00418162e-01 -8.68549585e-01 -9.31537002e-02
-6.15044653e-01 5.37197351e-01 -3.55584443e-01 1.37460455e-01
-1.06255710e+00 -2.32552439e-01 1.07178783e+00 4.15558219e-01
1.22655474e-01 -1.52040756e+00 -6.88281178e-01 6.59112990e-01
-4.06028777e-01 -6.51671469e-01 -7.17184186e-01 5.62858701e-01
-7.23437011e-01 -7.24439502e-01 -1.11050868e+00 -8.54109287e-01
-3.46556544e-01 -4.66645658e-01 1.16472805e+00 -5.32606840e-01
-1.34799838e-01 -6.73336238e-02 -3.49761903e-01 -4.89210874e-01
-5.12816429e-01 4.37264562e-01 4.98514548e-02 4.24755394e-01
-3.84114124e-02 -1.03979290e+00 -6.19078696e-01 9.98118334e-03
-6.79105580e-01 2.03449652e-01 5.03026307e-01 8.01767707e-01
7.17258871e-01 -3.09660196e-01 7.51384735e-01 -4.18940902e-01
7.51669347e-01 1.32173494e-01 -4.39532548e-01 -2.94240832e-01
-4.77716252e-02 -1.62467644e-01 1.06558287e+00 -7.84169078e-01
-6.65547311e-01 4.36196238e-01 -3.55663866e-01 -8.34365666e-01
-7.91992918e-02 4.58063871e-01 1.57631755e-01 6.55852556e-01
9.31870520e-01 9.18848217e-02 -4.69785720e-01 -6.28315806e-01
4.56014872e-01 7.31418133e-01 1.15272439e+00 -5.79552531e-01
7.04476416e-01 1.78326949e-01 -1.46405309e-01 -5.53618968e-01
-9.40574706e-01 -3.87843698e-01 -8.24543834e-01 -2.66330093e-01
8.75007987e-01 -8.71061563e-01 -8.59002769e-01 5.41067161e-02
-1.29662836e+00 -4.06154066e-01 -7.25985646e-01 8.79056156e-01
-1.28849292e+00 -8.27495679e-02 -1.08381927e+00 -8.98031771e-01
-6.93736494e-01 -8.78291011e-01 1.14719987e+00 -2.15845153e-01
-1.02234209e+00 -4.77743924e-01 4.91866499e-01 1.95815861e-01
2.08999261e-01 5.68943620e-01 8.57394814e-01 -3.84426951e-01
-3.98804963e-01 -2.57172406e-01 3.55955988e-01 5.46556294e-01
-3.72792222e-02 5.94509281e-02 -1.31395721e+00 -5.08850589e-02
-1.13064311e-01 -6.02502406e-01 8.16377759e-01 3.45783770e-01
1.09692645e+00 -6.17996693e-01 6.62516117e-01 7.53813922e-01
1.02987611e+00 2.72755086e-01 5.68948686e-01 8.80303904e-02
5.58384180e-01 4.32390541e-01 4.10304397e-01 5.90350926e-01
-1.56920835e-01 1.12499964e+00 2.02143550e-01 7.52214640e-02
-4.25639540e-01 -3.90486836e-01 9.60656881e-01 1.65903497e+00
-4.69023854e-01 2.03734055e-01 -7.23504126e-01 7.03424573e-01
-1.82799411e+00 -1.41008997e+00 -1.24421135e-01 2.11786103e+00
1.15335500e+00 1.71492592e-01 5.68100393e-01 1.03299117e+00
5.35930812e-01 2.04112455e-01 -3.96502316e-01 -8.91527295e-01
1.54943261e-02 9.66046691e-01 -1.10451607e-02 1.74760118e-01
-9.77622747e-01 7.31806695e-01 6.13630676e+00 9.87387955e-01
-1.40236723e+00 3.52188898e-03 1.44167334e-01 -6.70897007e-01
5.59982732e-02 -1.54566407e-01 -1.60156205e-01 2.41141796e-01
1.44086111e+00 1.20821454e-01 7.78190792e-01 4.81096834e-01
4.17938501e-01 6.19857252e-01 -1.24365819e+00 1.26821482e+00
-1.02726921e-01 -1.34222269e+00 6.98422734e-03 -3.56073290e-01
7.83314466e-01 -1.25501424e-01 3.05345893e-01 3.47770214e-01
1.42115265e-01 -1.11568069e+00 1.37199068e+00 6.81535423e-01
1.09150195e+00 -1.13425541e+00 5.07990479e-01 7.65625611e-02
-1.37851107e+00 -6.60081208e-02 -5.57996258e-02 -3.75340194e-01
2.24486947e-01 3.15666616e-01 -1.20042205e+00 5.75085521e-01
4.70190257e-01 7.39195466e-01 -2.15898320e-01 1.00718200e+00
-3.57220531e-01 1.13480413e+00 -2.48560950e-01 1.94175065e-01
2.24412084e-01 -1.00411788e-01 5.27701139e-01 1.51595581e+00
8.12018216e-01 -3.95077735e-01 9.19953883e-02 8.08224618e-01
-9.90567654e-02 3.09120327e-01 -2.48507082e-01 -2.60403037e-01
2.62081772e-01 1.16192043e+00 -6.65103436e-01 -2.74845511e-01
5.45733310e-02 9.35257554e-01 7.43935704e-02 1.53472319e-01
-7.60317266e-01 -8.52646172e-01 3.42876583e-01 -4.11715545e-02
3.88000965e-01 -2.58992910e-01 -3.16028655e-01 -9.59386051e-01
-2.86881416e-03 -1.05704927e+00 2.33331040e-01 -1.06121099e+00
-8.33800077e-01 7.76629746e-01 -4.03427780e-01 -1.72378564e+00
-1.03454959e+00 -1.90490276e-01 -8.10627162e-01 7.38770485e-01
-1.03134656e+00 -9.37499225e-01 1.50142759e-01 3.95459443e-01
7.96868920e-01 -3.20586085e-01 1.20711827e+00 3.31146419e-01
7.01426789e-02 4.52636927e-01 2.48330273e-02 4.08047706e-01
8.14788043e-01 -1.46387434e+00 5.05220711e-01 3.34233582e-01
1.04886818e+00 1.60327718e-01 7.33747542e-01 -4.33102474e-02
-8.59879673e-01 -1.01910913e+00 9.16579187e-01 -3.35049242e-01
6.55721605e-01 -5.02905488e-01 -6.46052718e-01 4.28256243e-01
4.28100318e-01 -3.38118643e-01 6.90127313e-01 1.89846009e-02
-3.21517289e-01 -2.45884344e-01 -3.40518326e-01 6.80753589e-01
7.60617197e-01 -7.90989101e-01 -7.24058270e-01 -9.97652113e-03
7.13925242e-01 -5.11284232e-01 -9.02915418e-01 3.73371750e-01
8.06103349e-01 -1.06548798e+00 8.17299008e-01 -3.20977539e-01
6.23916566e-01 -6.30408764e-01 -1.46763474e-01 -1.36973727e+00
-9.02599376e-03 -1.23895597e+00 -1.91625327e-01 1.04667127e+00
3.44469309e-01 4.11772162e-01 6.33749425e-01 -4.75535393e-01
-4.04366761e-01 -4.64341611e-01 -1.01679707e+00 -6.73188627e-01
-1.72877148e-01 -7.75843918e-01 6.06018782e-01 7.63389885e-01
4.84634228e-02 5.11181295e-01 -8.08336079e-01 -1.78166628e-01
3.23856086e-01 3.04478765e-01 8.76813352e-01 -1.26055944e+00
-7.21387327e-01 -6.22793198e-01 -4.35184509e-01 -6.46496713e-01
2.23642047e-02 -1.16970980e+00 4.65016961e-02 -1.05925548e+00
-2.99186677e-01 2.16832533e-02 -6.50295317e-01 4.96933609e-01
4.59442049e-01 7.92691648e-01 6.33459985e-01 2.77994394e-01
-4.56225306e-01 5.51313877e-01 9.71346736e-01 -3.42838377e-01
-5.25306761e-01 1.91904709e-01 1.85993612e-01 8.52395475e-01
8.72790575e-01 -6.85691178e-01 -2.11382285e-01 -1.50557011e-01
5.70606053e-01 6.17821753e-01 2.17413992e-01 -1.54434609e+00
1.52498037e-01 2.84776777e-01 4.29003477e-01 -8.17161739e-01
7.58725524e-01 -3.62385929e-01 4.63366419e-01 3.92382771e-01
-9.29067731e-01 2.60432690e-01 3.55810672e-02 1.68575063e-01
-7.15432405e-01 -2.79070616e-01 5.78512669e-01 -1.92348808e-02
1.42227203e-01 -1.91029698e-01 -3.87813419e-01 -2.98909277e-01
2.51405776e-01 -2.03322291e-01 4.30134147e-01 -5.65435767e-01
-1.26592720e+00 -5.93873084e-01 -8.55372697e-02 3.41207176e-01
4.34969246e-01 -1.82666790e+00 -9.86583591e-01 1.47477508e-01
-7.79802799e-02 -1.18634976e-01 1.01364210e-01 7.32036889e-01
-6.85502470e-01 4.54451144e-01 -3.55517417e-01 -6.53093040e-01
-1.31975889e+00 2.46849269e-01 1.88450664e-01 -4.62097108e-01
-7.84706473e-01 4.35278982e-01 -1.17204003e-01 -4.20761108e-01
6.20071054e-01 -7.08152950e-01 -2.74153680e-01 3.72807354e-01
6.02553368e-01 3.23066324e-01 3.34769130e-01 -4.97474223e-01
2.28147566e-01 5.43587029e-01 3.21660727e-01 -8.65263462e-01
1.39684427e+00 2.44688928e-01 1.23787723e-01 1.12135351e+00
1.02251029e+00 3.90516877e-01 -1.25972116e+00 -1.33076869e-02
3.44413817e-01 2.40475982e-02 -1.19449474e-01 -8.34673703e-01
-8.01988423e-01 1.02577937e+00 4.17411089e-01 1.88095465e-01
1.20209646e+00 -5.24935842e-01 1.00650394e+00 5.83926380e-01
1.86976746e-01 -1.18481910e+00 4.94287819e-01 8.06237102e-01
1.16874027e+00 -3.44767392e-01 -5.41529536e-01 4.03597087e-01
-5.54780960e-01 1.48686683e+00 -1.51211560e-01 -3.67432058e-01
1.84065863e-01 1.83968872e-01 3.86793315e-01 1.86320588e-01
-9.38998520e-01 -3.05372387e-01 4.81796741e-01 1.82014570e-01
7.16943800e-01 5.64674623e-02 -7.80520886e-02 7.31311083e-01
-1.25056207e+00 -1.09730355e-01 6.53707743e-01 3.16890746e-01
-2.38420472e-01 -1.09694719e+00 -7.26715922e-01 -1.06958650e-01
-8.12147856e-01 -2.47743189e-01 -5.60638070e-01 2.89752066e-01
2.15962827e-01 7.64266193e-01 1.84947737e-02 -5.38543284e-01
3.81431073e-01 6.86924577e-01 5.93283951e-01 -6.18943632e-01
-1.26125312e+00 5.67406952e-01 6.78111017e-02 -3.22327405e-01
-3.49082768e-01 -5.49845159e-01 -1.07907856e+00 1.70827672e-01
-2.45716944e-02 3.01684141e-01 8.18449140e-01 5.27878582e-01
-1.39271803e-02 9.42220628e-01 6.50787354e-01 -1.38815975e+00
-4.29223150e-01 -1.24389040e+00 -5.56597531e-01 4.60305721e-01
5.88093698e-01 9.27894115e-02 -5.53529039e-02 5.24980485e-01] | [15.852195739746094, 5.430830001831055] |
99273368-0ec8-49ae-be17-bdab606c37ef | efficient-multi-task-and-transfer | 2306.01839 | null | https://arxiv.org/abs/2306.01839v1 | https://arxiv.org/pdf/2306.01839v1.pdf | Efficient Multi-Task and Transfer Reinforcement Learning with Parameter-Compositional Framework | In this work, we investigate the potential of improving multi-task training and also leveraging it for transferring in the reinforcement learning setting. We identify several challenges towards this goal and propose a transferring approach with a parameter-compositional formulation. We investigate ways to improve the training of multi-task reinforcement learning which serves as the foundation for transferring. Then we conduct a number of transferring experiments on various manipulation tasks. Experimental results demonstrate that the proposed approach can have improved performance in the multi-task training stage, and further show effective transferring in terms of both sample efficiency and performance. | ['Masayoshi Tomizuka', 'Wei Xu', 'Haichao Zhang', 'Lingfeng Sun'] | 2023-06-02 | null | null | null | null | ['transfer-reinforcement-learning'] | ['methodology'] | [ 3.99907261e-01 -2.68640310e-01 -2.50253022e-01 -1.57097787e-01
-1.08097816e+00 -4.82101917e-01 3.69833708e-01 -4.38240111e-01
-6.90477610e-01 8.70790958e-01 -1.18047766e-01 -2.53176451e-01
-2.51601130e-01 -7.13334918e-01 -1.09776759e+00 -7.32853711e-01
-1.55334219e-01 3.88831705e-01 2.87480503e-01 -3.82477403e-01
1.83542594e-01 1.66064337e-01 -1.33626151e+00 4.61775213e-01
7.11126626e-01 8.12645137e-01 4.02177840e-01 6.76759183e-01
3.28356326e-01 8.86330426e-01 -6.25115633e-01 -1.92156523e-01
3.64560932e-01 -1.98049888e-01 -1.21274710e+00 7.99894109e-02
2.78301060e-01 -7.49927163e-01 -1.63532689e-01 6.81401134e-01
8.00569892e-01 4.95152473e-01 4.61461604e-01 -1.50588071e+00
-6.76986635e-01 6.40452623e-01 -5.15694797e-01 1.95825920e-01
1.62249073e-01 1.85120165e-01 8.68742049e-01 -8.46404314e-01
8.02795738e-02 1.42906988e+00 4.89017993e-01 7.75858283e-01
-9.09674466e-01 -6.81050956e-01 2.71070212e-01 3.15621316e-01
-7.42327631e-01 -3.35340023e-01 7.69676685e-01 -1.47772014e-01
9.77514088e-01 -2.98549563e-01 3.16775739e-01 1.35967076e+00
1.55943096e-01 1.43828154e+00 1.44031262e+00 -5.44387460e-01
-1.94622006e-03 -4.26534899e-02 -1.57528162e-01 8.89240384e-01
-1.91461012e-01 3.81231248e-01 -8.69486809e-01 -6.25424534e-02
9.71290946e-01 -1.29112035e-01 5.29783498e-03 -3.94134313e-01
-1.42345393e+00 9.66231823e-01 2.72815138e-01 3.05918008e-01
-2.57178634e-01 7.16371357e-01 7.63458431e-01 7.88663268e-01
6.82587981e-01 3.63575578e-01 -6.67296350e-01 -3.07960451e-01
-5.61053872e-01 3.53264511e-01 5.42772830e-01 1.21524787e+00
8.54636014e-01 1.54974073e-01 -4.14491713e-01 1.04808307e+00
7.80346105e-04 4.87393975e-01 8.38541567e-01 -1.07841337e+00
7.10774839e-01 9.15302262e-02 7.46878758e-02 -3.51244092e-01
-4.97036099e-01 -4.50822674e-02 -5.05253315e-01 1.46055132e-01
3.07784826e-01 -4.32346642e-01 -6.04627013e-01 1.81623518e+00
5.01454651e-01 3.52884471e-01 3.13608944e-02 6.05492890e-01
3.36226821e-01 3.75235558e-01 2.74067432e-01 -1.63753163e-02
1.25673139e+00 -1.58131397e+00 -6.73102796e-01 -1.99494854e-01
9.85907793e-01 -6.46597266e-01 1.35889637e+00 3.81763995e-01
-1.15830827e+00 -9.82853591e-01 -9.18613493e-01 1.33897915e-01
-3.53953838e-01 2.37695351e-01 9.33379948e-01 5.60136080e-01
-9.67094839e-01 9.58675683e-01 -8.26094329e-01 -4.52361740e-02
6.28155768e-01 3.24614853e-01 -1.90092564e-01 -1.42976612e-01
-1.24547970e+00 1.16181278e+00 7.84610391e-01 -1.00981951e-01
-1.31587839e+00 -7.14716792e-01 -8.45278084e-01 9.07680839e-02
6.18484497e-01 -7.77949631e-01 1.60824239e+00 -5.78986406e-01
-1.94968045e+00 4.59613711e-01 1.06544726e-01 -2.91336328e-01
4.96531010e-01 -5.16706645e-01 -3.82089354e-02 2.40573481e-01
4.06831540e-02 8.44785154e-01 1.16510785e+00 -1.09075880e+00
-8.51908207e-01 -3.49578381e-01 4.15190786e-01 4.31834877e-01
-7.45550871e-01 -3.28120068e-02 2.49940753e-02 -8.25705469e-01
-6.48295939e-01 -1.36969507e+00 -1.12829708e-01 -2.12594986e-01
6.99774474e-02 -6.56426132e-01 8.42927039e-01 -3.80552351e-01
9.69214380e-01 -2.07949972e+00 3.57579768e-01 -2.16891885e-01
7.73972422e-02 3.74854833e-01 -5.35613060e-01 5.85561335e-01
5.33889979e-02 1.47978868e-02 -1.85778081e-01 -4.80179518e-01
1.56734660e-01 4.16951448e-01 -4.27151680e-01 2.43276507e-01
4.14902717e-01 1.31956375e+00 -9.73110139e-01 -4.75815296e-01
1.45710915e-01 9.04508904e-02 -6.33310199e-01 5.10501683e-01
-2.32636884e-01 5.95953643e-01 -8.79680574e-01 5.45769513e-01
5.13025284e-01 -1.99162722e-01 5.38503528e-02 -3.84620065e-03
3.46765667e-01 1.80441618e-01 -1.00710547e+00 2.08955979e+00
-8.06204200e-01 8.97303298e-02 -4.89837080e-02 -1.66604733e+00
6.34193063e-01 2.95479685e-01 7.30305791e-01 -6.73010111e-01
-1.20562330e-01 5.19821569e-02 1.32109836e-01 -6.54913604e-01
7.48909354e-01 -4.54002440e-01 -2.44252592e-01 7.17753470e-01
5.65236211e-01 -2.98716664e-01 1.35313213e-01 -9.71654132e-02
8.83738399e-01 8.24794471e-01 2.42168307e-01 -3.43051106e-01
3.59437019e-01 -7.16558620e-02 1.14903040e-01 6.73877180e-01
-4.55156446e-01 -1.95125550e-01 1.20267719e-01 -2.14787528e-01
-8.42355072e-01 -7.77324140e-01 4.66671698e-02 1.96206760e+00
-6.18195981e-02 -1.82442248e-01 -5.20291448e-01 -1.01775444e+00
3.23736131e-01 5.07210374e-01 -7.95294583e-01 -5.29354930e-01
-8.62957418e-01 -7.70898283e-01 7.73851871e-01 9.01465237e-01
5.88457167e-01 -1.28606737e+00 -8.23737442e-01 4.20329690e-01
-2.97356904e-01 -1.26284921e+00 -4.50286984e-01 4.67729449e-01
-1.05473375e+00 -9.52450514e-01 -9.16757464e-01 -1.01686621e+00
3.15665811e-01 5.72327793e-01 8.99437368e-01 1.15867414e-01
5.90173937e-02 6.57846749e-01 -6.20102465e-01 -5.62150478e-01
-3.27513695e-01 3.38871717e-01 8.44135601e-03 -4.03141111e-01
-6.62521496e-02 -6.03975892e-01 -3.65577787e-01 5.28358519e-01
-1.09486496e+00 -1.74628347e-01 7.47317135e-01 1.10633647e+00
2.34137133e-01 -1.56345516e-01 1.23209357e+00 -8.31476152e-01
9.72657204e-01 -3.89199644e-01 -1.88231036e-01 3.96638542e-01
-4.93563771e-01 1.60179839e-01 7.70818055e-01 -6.87400699e-01
-1.08618009e+00 -8.22044685e-02 -5.14608137e-02 -4.55970973e-01
5.94014709e-04 4.18858349e-01 4.26029742e-01 -4.58733290e-01
5.44145763e-01 3.43567431e-01 1.53055906e-01 -3.36468339e-01
6.18631124e-01 5.05527854e-01 5.00457827e-03 -1.34598386e+00
6.61282837e-01 2.58062005e-01 2.30948757e-02 -3.60778600e-01
-9.50484097e-01 -4.90324676e-01 -6.16901696e-01 -3.44091386e-01
4.45360452e-01 -9.28798258e-01 -1.03215373e+00 5.62826693e-01
-8.43335032e-01 -9.71720159e-01 -1.17512263e-01 4.74104047e-01
-1.27070808e+00 3.97499502e-01 -8.59540999e-01 -4.38938558e-01
-2.33115450e-01 -1.49478745e+00 1.29610729e+00 -1.41689494e-01
3.19943964e-01 -1.24990654e+00 1.11128867e-01 3.46927017e-01
6.31671011e-01 -1.92991257e-01 9.24521029e-01 -6.09762132e-01
-2.64932722e-01 4.05212134e-01 -9.21644866e-02 3.91948909e-01
1.39836520e-01 -6.39758289e-01 -9.94976521e-01 -7.21287191e-01
-2.44411007e-02 -1.33289003e+00 9.47751164e-01 1.60668984e-01
1.47368526e+00 1.12823583e-01 -2.00333729e-01 4.32808876e-01
1.31379771e+00 8.47952589e-02 5.17954171e-01 5.83004892e-01
6.15119278e-01 5.22788167e-01 1.08349538e+00 5.63643754e-01
4.47635621e-01 8.18364680e-01 4.10959929e-01 -1.66134033e-02
-1.24690421e-01 -1.56082645e-01 6.62919402e-01 9.92912829e-01
-1.59906358e-01 1.21211030e-01 -5.60053349e-01 5.67366362e-01
-2.04966187e+00 -8.42233121e-01 4.40330744e-01 1.73137522e+00
1.16751349e+00 -1.18821137e-01 4.18135256e-01 -2.72981226e-02
5.34292936e-01 -5.82580967e-03 -6.70353353e-01 -5.13277411e-01
3.53316039e-01 6.26959562e-01 3.55698615e-01 2.65043706e-01
-1.27084076e+00 1.36071873e+00 7.59416962e+00 1.07910120e+00
-8.63891721e-01 2.15163946e-01 2.59286165e-01 1.36183605e-01
2.25110024e-01 -3.32791656e-01 -9.66464162e-01 1.14647478e-01
8.56483817e-01 -3.00679147e-01 7.84648478e-01 1.02881455e+00
-1.25646457e-01 -1.08907307e-02 -1.24002075e+00 5.44673979e-01
4.12586667e-02 -8.21966648e-01 1.41488492e-01 -8.76741409e-02
6.80338860e-01 -1.55103475e-01 1.31953254e-01 1.07836890e+00
4.76653934e-01 -8.06672454e-01 5.94077468e-01 -3.72956991e-02
8.12670887e-01 -6.48044348e-01 3.95880669e-01 5.87579906e-01
-1.19447601e+00 -1.77294046e-01 -5.27494729e-01 -1.68917716e-01
-1.35730639e-01 -1.26748115e-01 -9.50337768e-01 9.21490490e-01
6.02522194e-01 8.36575866e-01 -3.13128650e-01 7.28997648e-01
-1.99475840e-01 5.33062041e-01 -3.64083826e-04 -4.25251611e-02
4.02273595e-01 -1.02572486e-01 -4.79382984e-02 1.22784495e+00
1.36324272e-01 -2.20302150e-01 7.82961905e-01 4.76503104e-01
-1.73323557e-01 1.56978797e-02 -6.70719504e-01 1.16927445e-01
4.37180817e-01 1.31886303e+00 -4.48633760e-01 -3.48087817e-01
-5.21313131e-01 1.06015372e+00 8.75497043e-01 2.79903382e-01
-1.28946054e+00 -1.78106949e-01 3.04010451e-01 -6.71031237e-01
6.13723159e-01 -3.91716987e-01 5.38624562e-02 -1.02386403e+00
5.61985373e-02 -1.06593466e+00 3.98279041e-01 -5.53101718e-01
-1.34958839e+00 1.74939394e-01 3.65949810e-01 -1.30018818e+00
-3.24788302e-01 -9.23251450e-01 -3.33735496e-01 6.05725646e-01
-2.30503774e+00 -1.44403398e+00 -8.49121064e-02 9.93686736e-01
8.21925998e-01 -2.23194063e-01 9.29002464e-01 3.24207604e-01
-4.83820617e-01 8.04517329e-01 4.12038155e-02 -1.27807528e-01
1.04778314e+00 -1.25899053e+00 1.53430685e-01 2.83468693e-01
1.37604322e-04 4.51339722e-01 1.00896589e-01 -2.53180325e-01
-1.63140249e+00 -1.16230524e+00 -1.61915971e-03 -2.34557137e-01
8.92600536e-01 -1.88591570e-01 -6.82270348e-01 1.14556062e+00
3.14433485e-01 -1.17853247e-01 7.68448949e-01 2.48598278e-01
-1.48542106e-01 2.34393001e-01 -1.01323128e+00 3.54250908e-01
1.16975212e+00 -5.79355419e-01 -8.55384707e-01 5.44116139e-01
8.35870743e-01 -5.62777638e-01 -1.47157240e+00 3.98545772e-01
5.09632468e-01 -4.68059957e-01 1.13732541e+00 -9.78161752e-01
7.53984690e-01 2.88579911e-01 -3.27007356e-03 -1.98716462e+00
-4.27054435e-01 -5.89035809e-01 -1.09252796e-01 9.50974703e-01
1.97424084e-01 -6.24891579e-01 5.56918383e-01 1.09286867e-01
-4.50999707e-01 -8.95348072e-01 -8.97951365e-01 -1.16667974e+00
6.85614824e-01 -3.77016455e-01 2.95319676e-01 9.41813290e-01
1.98176250e-01 4.23155397e-01 -7.86033154e-01 -2.79470295e-01
4.22005534e-01 3.14755946e-01 8.95936906e-01 -6.74869478e-01
-5.83685994e-01 -2.61600882e-01 2.95936257e-01 -1.48384726e+00
4.80361164e-01 -9.98004556e-01 1.53908134e-01 -1.13795459e+00
2.71296233e-01 -5.64425290e-01 -5.59526145e-01 8.56248915e-01
-6.21821225e-01 -1.62816226e-01 5.15869141e-01 1.86228931e-01
-9.66220379e-01 9.84800339e-01 1.89268851e+00 5.64334691e-02
9.66271535e-02 5.61715364e-02 -6.59280956e-01 4.33843732e-01
1.08378601e+00 -6.06619477e-01 -5.05733967e-01 -7.20532119e-01
-1.43458456e-01 8.62933621e-02 1.93932995e-01 -8.36114883e-01
-6.72397837e-02 -2.67585844e-01 2.29708105e-01 -2.40165681e-01
3.48528028e-01 -8.51508737e-01 -7.41243839e-01 5.06480873e-01
-6.11156225e-01 2.13048682e-01 6.46264493e-01 6.59766555e-01
-8.24153498e-02 -2.54695237e-01 7.48856544e-01 -2.00972468e-01
-8.61164987e-01 2.85252094e-01 -1.96695372e-01 2.11988494e-01
1.21927512e+00 1.81396604e-01 -5.06831825e-01 -1.55382335e-01
-5.94224989e-01 6.23862803e-01 -8.68205503e-02 5.42112350e-01
7.16075838e-01 -1.59934378e+00 -6.82812870e-01 3.00850645e-02
1.43943518e-01 -3.41511071e-01 1.37246624e-01 8.20871949e-01
1.71772674e-01 4.35610831e-01 -6.93995893e-01 -4.52671856e-01
-1.05636919e+00 7.53857732e-01 1.46303728e-01 -7.42834210e-01
-4.00138199e-01 6.54904783e-01 8.59622732e-02 -7.94024587e-01
2.41526216e-01 -4.66954559e-01 -1.35880783e-01 -9.36836228e-02
3.80776703e-01 4.59744930e-01 -4.44736220e-02 -9.26879197e-02
-6.49840906e-02 5.31652749e-01 -2.74634272e-01 -4.17373478e-02
1.50377727e+00 1.46042421e-01 3.23870122e-01 3.78742456e-01
1.24644399e+00 -5.08204579e-01 -1.45671785e+00 -4.60385054e-01
-1.21743754e-02 -4.32574332e-01 -4.57545072e-02 -7.35249639e-01
-9.57312346e-01 8.87180030e-01 3.50809008e-01 2.08056539e-01
1.06148255e+00 -1.32558212e-01 8.83802712e-01 9.38371956e-01
6.70839190e-01 -1.42334688e+00 7.62198031e-01 6.85071588e-01
1.02088451e+00 -1.36798990e+00 -3.81103717e-02 -2.98156261e-01
-8.29375803e-01 1.24590194e+00 9.14874792e-01 -4.04469490e-01
4.66328353e-01 1.80927739e-01 -4.10162687e-01 -9.58018899e-02
-8.74279261e-01 -4.90103096e-01 -2.22627260e-02 7.03343451e-01
4.59642261e-01 -7.59638697e-02 -1.84445843e-01 2.44212896e-01
1.03709385e-01 3.13306391e-01 2.34249666e-01 1.29155326e+00
-5.36144018e-01 -1.47123599e+00 -2.08124146e-01 3.03587407e-01
-3.77101272e-01 -5.87340482e-02 9.20775160e-02 9.49136674e-01
-1.80555090e-01 8.78806531e-01 -4.64420527e-01 -4.16105717e-01
5.05606472e-01 2.68614829e-01 1.13810992e+00 -6.54747069e-01
-8.37111235e-01 -6.22887090e-02 2.23838568e-01 -6.31183565e-01
-9.28384542e-01 -4.50160265e-01 -1.00636256e+00 -1.17979102e-01
-4.41411793e-01 -5.62866963e-02 4.52434450e-01 1.09754789e+00
2.23376036e-01 8.68596196e-01 9.51661825e-01 -9.41699028e-01
-1.45460188e+00 -1.20178020e+00 -5.31250834e-01 6.02410972e-01
1.66076332e-01 -1.12821496e+00 1.08469175e-02 9.51172337e-02] | [3.9742431640625, 1.7958205938339233] |
06daed84-34ea-44fd-80cc-3b2fd0e852ff | estan-enhanced-small-tumor-aware-network-for | 2009.12894 | null | https://arxiv.org/abs/2009.12894v1 | https://arxiv.org/pdf/2009.12894v1.pdf | ESTAN: Enhanced Small Tumor-Aware Network for Breast Ultrasound Image Segmentation | Breast tumor segmentation is a critical task in computer-aided diagnosis (CAD) systems for breast cancer detection because accurate tumor size, shape and location are important for further tumor quantification and classification. However, segmenting small tumors in ultrasound images is challenging, due to the speckle noise, varying tumor shapes and sizes among patients, and the existence of tumor-like image regions. Recently, deep learning-based approaches have achieved great success for biomedical image analysis, but current state-of-the-art approaches achieve poor performance for segmenting small breast tumors. In this paper, we propose a novel deep neural network architecture, namely Enhanced Small Tumor-Aware Network (ESTAN), to accurately and robustly segment breast tumors. ESTAN introduces two encoders to extract and fuse image context information at different scales and utilizes row-column-wise kernels in the encoder to adapt to breast anatomy. We validate the proposed approach and compare it to nine state-of-the-art approaches on three public breast ultrasound datasets using seven quantitative metrics. The results demonstrate that the proposed approach achieves the best overall performance and outperforms all other approaches on small tumor segmentation. | ['Alex Vakanski', 'Phoebe E. Freer', 'Min Xian', 'Bryar Shareef'] | 2020-09-27 | null | null | null | null | ['breast-cancer-detection', 'breast-cancer-detection'] | ['knowledge-base', 'medical'] | [ 4.09079939e-01 1.54765353e-01 -4.07808125e-01 -5.99441051e-01
-1.16785419e+00 -4.65747118e-02 5.01207681e-03 4.92098212e-01
-3.74583751e-01 3.00160199e-01 4.53271829e-02 -5.65416455e-01
7.06702173e-02 -5.90663970e-01 -4.97571290e-01 -8.58991146e-01
-1.70673132e-01 5.75882137e-01 4.46034133e-01 1.13524109e-01
-1.37900561e-01 6.41890407e-01 -7.60689557e-01 5.41802347e-01
9.27556515e-01 1.39516294e+00 2.82562375e-01 7.56089807e-01
-2.61981457e-01 9.27719057e-01 -2.79544413e-01 -3.03381711e-01
-4.47440967e-02 -3.91499579e-01 -7.52065182e-01 -4.38086763e-02
4.96147782e-01 -3.97428215e-01 -3.29203755e-01 1.19140649e+00
5.81950963e-01 -6.16945684e-01 6.33676767e-01 -5.90616286e-01
-4.25015390e-01 5.68143368e-01 -7.32349098e-01 3.46848071e-01
-4.78740305e-01 -1.29197344e-01 4.31135207e-01 -5.15843868e-01
5.55081666e-01 6.29082680e-01 9.37378585e-01 6.41440332e-01
-9.93631482e-01 -7.45677412e-01 -3.57187748e-01 3.04245472e-01
-1.15867782e+00 -3.91793400e-01 5.44941306e-01 -4.95574951e-01
3.79394919e-01 3.95056218e-01 7.35617697e-01 7.43870378e-01
6.59629405e-01 9.54954565e-01 8.18869054e-01 -3.08700532e-01
1.26717523e-01 1.04639843e-01 1.59111887e-01 1.09620321e+00
3.90752405e-01 -1.30539179e-01 2.61835493e-02 -3.84349823e-02
7.50038326e-01 5.24722524e-02 -2.92238235e-01 -4.69378084e-01
-1.14019287e+00 7.19462276e-01 7.71899104e-01 6.74361765e-01
-4.75862980e-01 3.01004142e-01 7.15887964e-01 -3.56601238e-01
4.87983793e-01 -3.80313061e-02 -1.13245532e-01 4.85155545e-02
-1.06288540e+00 -9.27868485e-02 6.09308779e-01 4.14098471e-01
2.09705457e-01 -2.01741427e-01 -5.24446309e-01 8.50979745e-01
2.14529529e-01 3.15726608e-01 9.38343048e-01 -4.69923884e-01
1.11978874e-01 5.56292236e-01 -2.83550769e-01 -8.39805484e-01
-8.63602936e-01 -8.31694245e-01 -1.22844148e+00 -1.87998980e-01
3.86654794e-01 4.54197489e-02 -1.33878386e+00 1.29476333e+00
4.70128834e-01 2.16998547e-01 4.68753688e-02 8.59072149e-01
1.31267905e+00 1.05273329e-01 3.29918683e-01 -1.73115730e-01
1.43733764e+00 -1.01675069e+00 -7.78246641e-01 -3.35404128e-01
1.19753993e+00 -5.88395834e-01 4.58093762e-01 -7.80074298e-02
-9.46294129e-01 -2.98751593e-01 -8.74285221e-01 1.54175954e-02
-1.66295961e-01 7.65514135e-01 9.73288000e-01 8.50448072e-01
-8.77591968e-01 2.70098805e-01 -1.43598342e+00 -4.36665058e-01
1.12636411e+00 4.51786160e-01 -3.72785479e-01 -2.36252666e-01
-7.92349756e-01 8.21569026e-01 3.67812574e-01 3.90531987e-01
-9.06635582e-01 -9.89994287e-01 -8.35958660e-01 -6.21336214e-02
2.97172368e-01 -4.09055024e-01 1.42399049e+00 -9.58759606e-01
-1.17687774e+00 1.05522072e+00 -2.24095300e-01 -6.86364591e-01
5.15455604e-01 2.55650371e-01 -2.06174642e-01 4.28603292e-01
3.22189108e-02 6.14013314e-01 4.50510710e-01 -1.10285473e+00
-6.12062335e-01 -4.96878177e-01 -4.50834662e-01 -1.44879401e-01
-4.72689569e-01 -1.10695951e-01 -5.98829627e-01 -5.73086321e-01
4.05565619e-01 -8.48681450e-01 -5.78125238e-01 4.38344955e-01
-3.78750086e-01 8.53642076e-02 8.94506872e-01 -9.59151268e-01
1.25241363e+00 -2.07071543e+00 -2.82670353e-02 1.94328338e-01
5.17698705e-01 3.74692768e-01 6.96510077e-02 -4.50716317e-01
-3.59091908e-02 2.83776615e-02 -2.59487003e-01 -1.52184486e-01
-2.91994065e-01 2.50512064e-01 6.12317085e-01 6.82380378e-01
-7.95264076e-03 1.27433538e+00 -7.35109687e-01 -9.97116685e-01
3.07557762e-01 4.14724350e-01 -3.93476337e-01 6.65990785e-02
6.88947812e-02 3.90270531e-01 -5.64368725e-01 1.11152625e+00
6.92148209e-01 -4.57702696e-01 1.92305833e-01 -7.39835501e-01
2.68850863e-01 -3.39592278e-01 -6.31212711e-01 1.58567286e+00
-4.89926666e-01 7.37241924e-01 4.69676256e-01 -1.35851562e+00
5.03941774e-01 3.30250382e-01 1.03117049e+00 -5.24153769e-01
5.01077473e-01 6.31644905e-01 4.59650129e-01 -9.00444984e-01
8.59895870e-02 -9.30685848e-02 2.92908400e-01 4.40244190e-02
-1.16827749e-01 -1.31874695e-01 2.39352703e-01 6.70785755e-02
1.39216971e+00 -4.06410038e-01 4.74509269e-01 -1.96997315e-01
5.90250671e-01 6.36345297e-02 5.81737697e-01 5.07015884e-01
-6.73344851e-01 6.08613372e-01 5.93593836e-01 -5.09456098e-01
-5.04642308e-01 -7.86567986e-01 -4.84795183e-01 5.90422273e-01
-8.06607539e-04 2.88100522e-02 -7.89743483e-01 -8.87830973e-01
2.64843609e-02 2.55079091e-01 -1.12574482e+00 -6.84671476e-02
-7.26305187e-01 -1.04101825e+00 5.42977095e-01 8.05657268e-01
5.64736724e-01 -7.59488106e-01 -8.15561295e-01 3.08091372e-01
-3.13883483e-01 -1.24767935e+00 -3.22545856e-01 4.33243930e-01
-9.41337764e-01 -1.25049829e+00 -1.10764074e+00 -1.03082097e+00
1.08960426e+00 -3.71759571e-02 9.12387252e-01 1.21447975e-02
-1.02694619e+00 9.08648297e-02 -2.94677615e-01 -5.79580069e-01
-8.40669215e-01 2.04479396e-01 -5.88027239e-01 -2.53740586e-02
4.08830732e-01 6.81477860e-02 -6.49953842e-01 1.00786485e-01
-9.37968969e-01 2.81666905e-01 1.13582921e+00 1.15598357e+00
7.98902512e-01 9.03047156e-03 2.14915127e-01 -1.02501118e+00
3.12908649e-01 -2.44580761e-01 -3.30910742e-01 3.09036642e-01
-1.08721189e-01 -2.87625909e-01 3.40710193e-01 -2.04574972e-01
-8.43700588e-01 2.53985852e-01 -3.21115345e-01 -2.72125363e-01
-2.34825686e-01 7.25513697e-01 2.65691072e-01 -6.26042724e-01
6.25436187e-01 2.04020634e-01 2.62354940e-01 -4.35250886e-02
-2.38876343e-01 5.97795367e-01 4.98046309e-01 -1.35408014e-01
3.09405923e-01 6.49509251e-01 2.85900861e-01 -6.18566632e-01
-9.31870580e-01 -5.65491438e-01 -6.43851399e-01 -2.56251633e-01
9.27083492e-01 -6.05420232e-01 -4.49674308e-01 6.56612039e-01
-8.39851856e-01 -3.62036079e-01 -3.03642899e-01 4.66217220e-01
-2.70712733e-01 1.75858229e-01 -9.06932175e-01 -2.90847957e-01
-5.96350133e-01 -1.60881054e+00 1.24533808e+00 3.94711256e-01
4.91831414e-02 -1.00922012e+00 -3.32607150e-01 3.77104878e-01
8.62235665e-01 4.61356819e-01 9.17432904e-01 -7.56423712e-01
-2.98628837e-01 -6.27803862e-01 -5.80045581e-01 1.97657153e-01
4.80585188e-01 -7.27483407e-02 -7.48962879e-01 -2.96064287e-01
-3.02343637e-01 -3.20073903e-01 1.12693262e+00 1.16675210e+00
1.71280503e+00 9.99586582e-02 -1.01726270e+00 9.08908844e-01
1.32426465e+00 2.23013505e-01 3.24261576e-01 1.50459155e-01
7.26271927e-01 3.39310765e-01 4.06901062e-01 2.55418152e-01
2.37565085e-01 2.44301364e-01 5.95560253e-01 -6.46823823e-01
-2.98677802e-01 4.06173110e-01 -3.15398395e-01 5.45298755e-01
7.92165771e-02 3.11805494e-03 -1.18401897e+00 6.43543780e-01
-1.59881616e+00 -6.25468075e-01 8.96960795e-02 1.74701178e+00
9.49505389e-01 9.04755518e-02 -4.97604191e-01 -8.73536021e-02
4.24920201e-01 6.43167868e-02 -7.13567197e-01 -3.72464322e-02
1.62508801e-01 3.99923056e-01 8.00922096e-01 1.24024272e-01
-1.44070697e+00 6.31222546e-01 6.36153793e+00 1.06160557e+00
-1.51062191e+00 3.20657849e-01 1.30238533e+00 1.33618340e-01
1.39560595e-01 -5.60340285e-01 -6.12618804e-01 1.03604026e-01
5.93818367e-01 2.55608112e-01 -4.32009995e-01 8.10868204e-01
6.33541644e-02 -3.60418260e-01 -1.03302610e+00 9.11374748e-01
1.29997417e-01 -1.61801028e+00 -3.02044660e-01 -4.48289812e-02
7.40910709e-01 3.22672501e-02 1.47889610e-02 8.24035481e-02
-9.53654274e-02 -1.13788652e+00 1.86613679e-01 4.32397962e-01
9.23186839e-01 -5.35135508e-01 1.40494752e+00 -2.25284118e-02
-1.19508910e+00 -3.46545428e-02 -2.11450636e-01 8.03373396e-01
-1.50650129e-01 6.73613429e-01 -1.16836751e+00 3.30256969e-01
6.27852261e-01 7.27770686e-01 -7.70272613e-01 1.12355995e+00
4.42579389e-01 8.28607678e-01 -3.16439152e-01 -5.18169180e-02
2.10044578e-01 1.47344530e-01 1.45309255e-01 1.39495587e+00
4.93175000e-01 1.66323632e-01 2.91053712e-01 6.52075231e-01
-1.74936019e-02 2.63961941e-01 4.32165563e-02 -2.53180325e-01
1.33988395e-01 1.60010588e+00 -1.08787501e+00 -4.66280073e-01
-4.90381151e-01 7.43066609e-01 4.88905273e-02 1.16589025e-01
-9.30156350e-01 -2.39994228e-01 2.05910295e-01 1.87386453e-01
2.70359278e-01 1.09226234e-01 -3.12689632e-01 -7.10352063e-01
-1.68508083e-01 -8.07357728e-01 3.83903474e-01 -2.60759890e-01
-9.28556740e-01 5.50265729e-01 -3.08591038e-01 -1.05387723e+00
2.40616240e-02 -7.60008037e-01 -4.36034292e-01 4.67577338e-01
-1.64841032e+00 -1.57802665e+00 -8.17420721e-01 1.67233124e-01
5.67515254e-01 -1.07525632e-01 9.03888822e-01 3.83675188e-01
-6.13745689e-01 9.39820111e-01 1.81963891e-01 4.76833999e-01
7.57058144e-01 -1.14206970e+00 -2.36713573e-01 5.20656526e-01
-4.58019465e-01 3.64230797e-02 3.25272739e-01 -4.64287132e-01
-1.41840184e+00 -1.31277728e+00 3.71989340e-01 2.23598659e-01
6.03625417e-01 3.50222439e-02 -7.49505103e-01 5.66235244e-01
6.54754341e-02 8.43235970e-01 8.86783123e-01 -4.59402859e-01
2.96181411e-01 -4.07781541e-01 -1.37627459e+00 2.60413259e-01
5.27616858e-01 3.79240625e-02 2.74001420e-01 4.88458872e-01
3.06764036e-01 -1.10084808e+00 -1.02433479e+00 9.66950715e-01
6.68372691e-01 -9.53537524e-01 8.49950373e-01 -2.81699777e-01
6.06338024e-01 7.08919019e-02 -8.34702551e-02 -1.21930432e+00
-3.57348382e-01 6.52206391e-02 -3.82211581e-02 6.18786216e-01
3.57236475e-01 -5.53549588e-01 1.32656658e+00 4.30736125e-01
-4.48116124e-01 -1.37722874e+00 -1.01548016e+00 -1.36555314e-01
2.13370383e-01 -2.61938334e-01 2.67407328e-01 6.85889244e-01
-3.22602719e-01 -4.96906877e-01 2.81064242e-01 1.57638878e-01
6.22221589e-01 6.12962283e-02 3.30133259e-01 -9.04148340e-01
-6.94355965e-02 -7.78366446e-01 -8.00648391e-01 -7.51807094e-01
-4.53434996e-02 -9.16364849e-01 1.62852824e-01 -1.78656936e+00
5.41298509e-01 -5.97559810e-01 -4.71903980e-01 6.17471933e-01
-2.82735109e-01 5.61477721e-01 -2.32682109e-01 -1.35675013e-01
-4.12281483e-01 2.24153131e-01 1.67026711e+00 -7.13158607e-01
6.24698140e-02 6.66141137e-02 -5.09766400e-01 8.12187910e-01
5.85553586e-01 -3.29810590e-01 8.23635012e-02 -2.76363343e-01
-4.75305736e-01 2.07406163e-01 2.72164285e-01 -1.37639463e+00
2.86794215e-01 -1.06548794e-01 8.18544507e-01 -8.25176001e-01
1.64855734e-01 -9.17766452e-01 -2.04272911e-01 9.94203031e-01
-3.62711132e-01 -5.73595464e-01 4.50935990e-01 2.83414811e-01
-4.65444714e-01 -1.79988146e-01 1.07386410e+00 -1.59766003e-01
-5.26139617e-01 6.29941344e-01 -4.52079803e-01 -2.05827877e-01
1.34428561e+00 -2.91662306e-01 -6.92427307e-02 -1.70805864e-02
-7.15843618e-01 1.76258609e-01 -3.95566896e-02 1.00142770e-01
6.80092037e-01 -1.24059069e+00 -1.01075864e+00 2.14493454e-01
2.47718439e-01 3.06741506e-01 4.68333274e-01 1.53018439e+00
-9.64139819e-01 4.82832164e-01 3.07472758e-02 -9.96766508e-01
-1.57636786e+00 9.08865966e-03 9.27733421e-01 -6.70071006e-01
-4.75980669e-01 1.29069459e+00 2.98901618e-01 -2.74846971e-01
4.19619232e-01 -8.46191764e-01 -2.24082246e-01 -2.33071327e-01
4.28754926e-01 -3.97548079e-02 2.77017891e-01 -6.33860946e-01
-2.85517067e-01 5.83549559e-01 -5.63386142e-01 6.40392959e-01
1.09348571e+00 2.59021699e-01 -1.64502308e-01 1.51583720e-02
1.27023947e+00 -3.45912725e-01 -7.43886828e-01 -3.73669744e-01
-2.91978955e-01 -3.25931638e-01 5.46702445e-01 -8.44126582e-01
-1.63148749e+00 9.23105538e-01 1.29065311e+00 -6.58235475e-02
1.42016327e+00 8.87469947e-02 1.04412174e+00 2.29215890e-01
-5.01126386e-02 -7.83680916e-01 -8.00324678e-02 3.65153104e-02
6.04225814e-01 -1.82617652e+00 1.54377118e-01 -6.68920457e-01
-4.86617297e-01 1.47441185e+00 7.97084093e-01 2.09667221e-01
9.35046852e-01 6.97237551e-01 3.69394213e-01 -1.67382255e-01
-3.73492032e-01 -4.30313982e-02 3.25787306e-01 3.81786704e-01
7.95010030e-01 2.71222144e-01 -2.01023683e-01 7.12472975e-01
1.71433687e-01 2.68616766e-01 1.42280579e-01 9.75395381e-01
-5.35026252e-01 -7.67253697e-01 -3.33368897e-01 1.06869125e+00
-7.53546178e-01 1.13214470e-01 -1.33160099e-01 8.84745598e-01
2.57037908e-01 3.89880598e-01 1.03037222e-03 -1.58389378e-02
1.68773029e-02 -4.13110822e-01 5.88985920e-01 -4.46345568e-01
-4.85993743e-01 2.74490863e-01 -7.66382143e-02 -5.16943157e-01
-3.93216968e-01 -5.17872632e-01 -1.44138265e+00 1.77153632e-01
-5.08276820e-01 -3.55797857e-02 8.22214365e-01 8.09178889e-01
1.34956449e-01 1.16581655e+00 3.56178731e-01 -7.57287323e-01
-5.43967485e-01 -1.08923233e+00 -4.84508336e-01 1.61168680e-01
4.56628054e-01 -4.54164982e-01 1.11697204e-01 1.39804006e-01] | [15.06545639038086, -2.544410467147827] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.